2023-04-16 11:31:55,722 INFO [train.py:963] (0/4) Training started 2023-04-16 11:31:55,724 INFO [train.py:973] (0/4) Device: cuda:0 2023-04-16 11:31:55,727 INFO [train.py:982] (0/4) {'frame_shift_ms': 10.0, 'allowed_excess_duration_ratio': 0.1, 'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'warm_step': 2000, 'env_info': {'k2-version': '1.23.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '266eaf80fe659f113b4392b20dc1e35812f677cd', 'k2-git-date': 'Tue Feb 14 11:19:02 2023', 'lhotse-version': '1.14.0.dev+git.b55313a.clean', 'torch-version': '1.13.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.1', 'icefall-git-branch': 'master', 'icefall-git-sha1': 'ba3c923-dirty', 'icefall-git-date': 'Sun Apr 16 10:42:45 2023', 'icefall-path': '/k2-dev/yangyifan/icefall-master', 'k2-path': '/k2-dev/yangyifan/anaconda3/envs/icefall/lib/python3.10/site-packages/k2-1.23.4.dev20230214+cuda11.6.torch1.13.1-py3.10-linux-x86_64.egg/k2/__init__.py', 'lhotse-path': '/k2-dev/yangyifan/anaconda3/envs/icefall/lib/python3.10/site-packages/lhotse-1.14.0.dev0+git.b55313a.clean-py3.10.egg/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-6-1219221738-65dd59bbf8-2ghmr', 'IP address': '10.177.28.85'}, 'world_size': 4, 'master_port': 12354, 'tensorboard': True, 'num_epochs': 30, 'start_epoch': 1, 'start_batch': 0, 'exp_dir': PosixPath('pruned_transducer_stateless7/exp'), 'bpe_model': 'data/en/lang_bpe_500/bpe.model', 'base_lr': 0.05, 'lr_batches': 5000, 'lr_epochs': 3.5, 'context_size': 2, 'prune_range': 5, 'lm_scale': 0.25, 'am_scale': 0.0, 'simple_loss_scale': 0.5, 'seed': 42, 'print_diagnostics': False, 'inf_check': False, 'save_every_n': 2000, 'keep_last_k': 30, 'average_period': 200, 'use_fp16': True, 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'language': 'en', 'cv_manifest_dir': PosixPath('data/en/fbank'), 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 550, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'blank_id': 0, 'vocab_size': 500} 2023-04-16 11:31:55,728 INFO [train.py:984] (0/4) About to create model 2023-04-16 11:31:56,656 INFO [zipformer.py:178] (0/4) At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8. 2023-04-16 11:31:56,680 INFO [train.py:988] (0/4) Number of model parameters: 70369391 2023-04-16 11:32:00,059 INFO [train.py:1003] (0/4) Using DDP 2023-04-16 11:32:00,366 INFO [asr_datamodule.py:404] (0/4) About to get train cuts 2023-04-16 11:32:00,368 INFO [asr_datamodule.py:230] (0/4) Enable MUSAN 2023-04-16 11:32:00,368 INFO [asr_datamodule.py:231] (0/4) About to get Musan cuts 2023-04-16 11:32:03,429 INFO [asr_datamodule.py:255] (0/4) Enable SpecAugment 2023-04-16 11:32:03,430 INFO [asr_datamodule.py:256] (0/4) Time warp factor: 80 2023-04-16 11:32:03,430 INFO [asr_datamodule.py:266] (0/4) Num frame mask: 10 2023-04-16 11:32:03,430 INFO [asr_datamodule.py:279] (0/4) About to create train dataset 2023-04-16 11:32:03,430 INFO [asr_datamodule.py:306] (0/4) Using DynamicBucketingSampler. 2023-04-16 11:32:07,703 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 11:32:09,370 INFO [asr_datamodule.py:321] (0/4) About to create train dataloader 2023-04-16 11:32:09,371 INFO [asr_datamodule.py:411] (0/4) About to get dev cuts 2023-04-16 11:32:09,373 INFO [asr_datamodule.py:352] (0/4) About to create dev dataset 2023-04-16 11:32:10,337 INFO [asr_datamodule.py:369] (0/4) About to create dev dataloader 2023-04-16 11:32:10,337 INFO [train.py:1199] (0/4) Sanity check -- see if any of the batches in epoch 1 would cause OOM. 2023-04-16 11:32:14,857 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 11:32:21,018 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 11:32:32,850 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 11:32:32,850 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 11:32:32,850 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 11:32:32,857 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 11:32:32,873 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 11:32:32,894 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 11:32:32,904 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 11:33:20,490 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 11:34:15,652 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 11:34:51,494 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 11:34:52,881 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 11:36:07,142 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 11:36:25,451 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 11:37:09,459 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 11:37:33,171 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 11:37:43,922 INFO [train.py:1227] (0/4) Maximum memory allocated so far is 11469MB 2023-04-16 11:37:45,091 INFO [train.py:1227] (0/4) Maximum memory allocated so far is 11954MB 2023-04-16 11:37:46,441 INFO [train.py:1227] (0/4) Maximum memory allocated so far is 11954MB 2023-04-16 11:37:47,791 INFO [train.py:1227] (0/4) Maximum memory allocated so far is 11954MB 2023-04-16 11:37:49,472 INFO [train.py:1227] (0/4) Maximum memory allocated so far is 11954MB 2023-04-16 11:37:51,090 INFO [train.py:1227] (0/4) Maximum memory allocated so far is 11954MB 2023-04-16 11:38:07,883 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 11:38:11,927 INFO [train.py:893] (0/4) Epoch 1, batch 0, loss[loss=7.163, simple_loss=6.495, pruned_loss=6.671, over 13394.00 frames. ], tot_loss[loss=7.163, simple_loss=6.495, pruned_loss=6.671, over 13394.00 frames. ], batch size: 62, lr: 2.50e-02, grad_scale: 2.0 2023-04-16 11:38:11,928 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 11:38:34,892 INFO [train.py:927] (0/4) Epoch 1, validation: loss=6.498, simple_loss=5.86, pruned_loss=6.372, over 2446609.00 frames. 2023-04-16 11:38:34,892 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 11:38:35,130 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9931, 4.9931, 4.9931, 4.9930, 4.9929, 4.9931, 4.9931, 4.9930], device='cuda:0'), covar=tensor([0.0039, 0.0028, 0.0027, 0.0016, 0.0033, 0.0033, 0.0027, 0.0022], device='cuda:0'), in_proj_covar=tensor([0.0009, 0.0008, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:0'), out_proj_covar=tensor([8.9688e-06, 8.7191e-06, 8.7382e-06, 8.6879e-06, 8.7628e-06, 8.6855e-06, 8.6903e-06, 8.8188e-06], device='cuda:0') 2023-04-16 11:38:37,618 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 11:38:49,409 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:38:49,629 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=5.75 vs. limit=2.0 2023-04-16 11:39:02,765 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=170.17 vs. limit=5.0 2023-04-16 11:39:04,015 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=5.45 vs. limit=2.0 2023-04-16 11:39:08,540 INFO [train.py:893] (0/4) Epoch 1, batch 50, loss[loss=1.049, simple_loss=0.9302, pruned_loss=1.06, over 13462.00 frames. ], tot_loss[loss=1.891, simple_loss=1.72, pruned_loss=1.648, over 604341.63 frames. ], batch size: 79, lr: 2.75e-02, grad_scale: 2.0 2023-04-16 11:39:28,038 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 11:39:28,039 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 11:39:28,039 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 11:39:28,046 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 11:39:28,061 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 11:39:28,082 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 11:39:28,092 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 11:39:31,275 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:39:44,035 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 4.680e+01 1.295e+02 2.313e+02 7.962e+02 1.225e+04, threshold=4.625e+02, percent-clipped=0.0 2023-04-16 11:39:44,060 INFO [train.py:893] (0/4) Epoch 1, batch 100, loss[loss=0.8887, simple_loss=0.7682, pruned_loss=0.9617, over 13537.00 frames. ], tot_loss[loss=1.372, simple_loss=1.229, pruned_loss=1.295, over 1058847.31 frames. ], batch size: 98, lr: 3.00e-02, grad_scale: 2.0 2023-04-16 11:39:55,196 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2328, 4.2498, 4.2530, 4.2528, 4.2351, 4.2477, 4.2469, 4.2479], device='cuda:0'), covar=tensor([0.0028, 0.0022, 0.0015, 0.0023, 0.0028, 0.0038, 0.0027, 0.0028], device='cuda:0'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:0'), out_proj_covar=tensor([9.1209e-06, 9.2508e-06, 9.1537e-06, 9.1408e-06, 9.5108e-06, 9.1873e-06, 9.2229e-06, 9.3572e-06], device='cuda:0') 2023-04-16 11:39:59,258 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=44.38 vs. limit=5.0 2023-04-16 11:40:13,302 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2023-04-16 11:40:14,369 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=144.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-16 11:40:18,444 INFO [train.py:893] (0/4) Epoch 1, batch 150, loss[loss=0.8428, simple_loss=0.7212, pruned_loss=0.8863, over 13470.00 frames. ], tot_loss[loss=1.155, simple_loss=1.022, pruned_loss=1.134, over 1402581.63 frames. ], batch size: 81, lr: 3.25e-02, grad_scale: 2.0 2023-04-16 11:40:27,024 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=13.53 vs. limit=5.0 2023-04-16 11:40:55,349 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 5.673e+01 1.416e+02 2.268e+02 3.806e+02 1.185e+03, threshold=4.536e+02, percent-clipped=12.0 2023-04-16 11:40:55,375 INFO [train.py:893] (0/4) Epoch 1, batch 200, loss[loss=0.8272, simple_loss=0.7038, pruned_loss=0.8333, over 13506.00 frames. ], tot_loss[loss=1.033, simple_loss=0.9061, pruned_loss=1.024, over 1669918.06 frames. ], batch size: 85, lr: 3.50e-02, grad_scale: 2.0 2023-04-16 11:41:14,758 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-16 11:41:29,423 INFO [train.py:893] (0/4) Epoch 1, batch 250, loss[loss=0.8075, simple_loss=0.6867, pruned_loss=0.7693, over 13529.00 frames. ], tot_loss[loss=0.9525, simple_loss=0.8294, pruned_loss=0.9389, over 1882078.03 frames. ], batch size: 91, lr: 3.75e-02, grad_scale: 2.0 2023-04-16 11:42:01,879 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:42:04,254 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-16 11:42:04,599 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.029e+01 1.011e+02 1.415e+02 2.248e+02 8.541e+02, threshold=2.829e+02, percent-clipped=4.0 2023-04-16 11:42:04,625 INFO [train.py:893] (0/4) Epoch 1, batch 300, loss[loss=0.7757, simple_loss=0.6547, pruned_loss=0.7233, over 13443.00 frames. ], tot_loss[loss=0.8968, simple_loss=0.7757, pruned_loss=0.8744, over 2051701.30 frames. ], batch size: 103, lr: 4.00e-02, grad_scale: 2.0 2023-04-16 11:42:38,588 INFO [train.py:893] (0/4) Epoch 1, batch 350, loss[loss=0.7956, simple_loss=0.6675, pruned_loss=0.7236, over 13478.00 frames. ], tot_loss[loss=0.8557, simple_loss=0.7349, pruned_loss=0.825, over 2191907.07 frames. ], batch size: 100, lr: 4.25e-02, grad_scale: 2.0 2023-04-16 11:42:43,064 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=357.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-16 11:43:03,947 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=387.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 11:43:13,441 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 6.345e+01 9.312e+01 1.191e+02 1.814e+02 3.998e+02, threshold=2.382e+02, percent-clipped=6.0 2023-04-16 11:43:13,466 INFO [train.py:893] (0/4) Epoch 1, batch 400, loss[loss=0.7435, simple_loss=0.6183, pruned_loss=0.6674, over 13454.00 frames. ], tot_loss[loss=0.8286, simple_loss=0.7069, pruned_loss=0.7879, over 2292386.74 frames. ], batch size: 106, lr: 4.50e-02, grad_scale: 4.0 2023-04-16 11:43:37,625 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1416, 5.0239, 4.8057, 5.1405, 5.0897, 4.3012, 5.0698, 4.5682], device='cuda:0'), covar=tensor([0.0044, 0.0112, 0.0341, 0.0041, 0.0081, 0.1202, 0.0050, 0.0793], device='cuda:0'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0010, 0.0009, 0.0009], device='cuda:0'), out_proj_covar=tensor([9.1965e-06, 9.5556e-06, 9.3325e-06, 8.9579e-06, 9.0792e-06, 9.8104e-06, 9.0815e-06, 9.4695e-06], device='cuda:0') 2023-04-16 11:43:39,280 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=439.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-16 11:43:41,677 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6604, 4.6304, 4.0911, 4.7037, 4.6243, 4.5431, 4.4830, 4.6177], device='cuda:0'), covar=tensor([0.0066, 0.0084, 0.0514, 0.0049, 0.0078, 0.0114, 0.0089, 0.0075], device='cuda:0'), in_proj_covar=tensor([0.0008, 0.0008, 0.0009, 0.0008, 0.0008, 0.0008, 0.0008, 0.0008], device='cuda:0'), out_proj_covar=tensor([8.5244e-06, 8.7170e-06, 8.5521e-06, 8.2265e-06, 8.4153e-06, 8.4228e-06, 8.1616e-06, 8.3375e-06], device='cuda:0') 2023-04-16 11:43:46,676 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=448.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-16 11:43:48,214 INFO [train.py:893] (0/4) Epoch 1, batch 450, loss[loss=0.7608, simple_loss=0.6312, pruned_loss=0.6628, over 13205.00 frames. ], tot_loss[loss=0.8134, simple_loss=0.6892, pruned_loss=0.7614, over 2378247.52 frames. ], batch size: 132, lr: 4.75e-02, grad_scale: 4.0 2023-04-16 11:44:05,709 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 11:44:21,851 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.092e+01 1.034e+02 1.217e+02 1.498e+02 2.544e+02, threshold=2.433e+02, percent-clipped=2.0 2023-04-16 11:44:21,876 INFO [train.py:893] (0/4) Epoch 1, batch 500, loss[loss=0.731, simple_loss=0.5991, pruned_loss=0.6358, over 13555.00 frames. ], tot_loss[loss=0.7987, simple_loss=0.6726, pruned_loss=0.7358, over 2439378.06 frames. ], batch size: 78, lr: 4.99e-02, grad_scale: 4.0 2023-04-16 11:44:43,021 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6261, 4.2189, 4.2354, 4.3006, 4.2917, 4.6518, 4.2328, 4.4095], device='cuda:0'), covar=tensor([0.0675, 0.1610, 0.1448, 0.1149, 0.1044, 0.0374, 0.0924, 0.0801], device='cuda:0'), in_proj_covar=tensor([0.0013, 0.0014, 0.0016, 0.0015, 0.0014, 0.0013, 0.0015, 0.0015], device='cuda:0'), out_proj_covar=tensor([1.1905e-05, 1.2451e-05, 1.3597e-05, 1.3898e-05, 1.3351e-05, 1.2222e-05, 1.3511e-05, 1.3185e-05], device='cuda:0') 2023-04-16 11:44:45,607 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.67 vs. limit=2.0 2023-04-16 11:44:46,775 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=5.33 vs. limit=2.0 2023-04-16 11:44:57,642 INFO [train.py:893] (0/4) Epoch 1, batch 550, loss[loss=0.7047, simple_loss=0.5801, pruned_loss=0.5876, over 11629.00 frames. ], tot_loss[loss=0.7831, simple_loss=0.6563, pruned_loss=0.7083, over 2487750.70 frames. ], batch size: 157, lr: 4.98e-02, grad_scale: 4.0 2023-04-16 11:45:02,168 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.9557, 5.8955, 5.4335, 5.0758, 5.2241, 5.6178, 5.8540, 5.8411], device='cuda:0'), covar=tensor([0.0093, 0.0258, 0.2687, 0.6203, 0.2384, 0.1241, 0.0540, 0.0374], device='cuda:0'), in_proj_covar=tensor([0.0008, 0.0008, 0.0009, 0.0009, 0.0008, 0.0008, 0.0008, 0.0008], device='cuda:0'), out_proj_covar=tensor([8.0046e-06, 8.4579e-06, 8.1343e-06, 9.1111e-06, 8.7788e-06, 8.1516e-06, 8.5512e-06, 8.2623e-06], device='cuda:0') 2023-04-16 11:45:04,602 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=562.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 11:45:08,288 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=568.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:45:24,228 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=590.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:45:25,682 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-16 11:45:30,975 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=600.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-16 11:45:31,317 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.423e+01 1.230e+02 1.562e+02 1.964e+02 3.659e+02, threshold=3.124e+02, percent-clipped=9.0 2023-04-16 11:45:31,352 INFO [train.py:893] (0/4) Epoch 1, batch 600, loss[loss=0.7011, simple_loss=0.5825, pruned_loss=0.5562, over 13488.00 frames. ], tot_loss[loss=0.7665, simple_loss=0.6405, pruned_loss=0.6779, over 2525352.28 frames. ], batch size: 81, lr: 4.98e-02, grad_scale: 4.0 2023-04-16 11:45:47,885 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=623.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-16 11:45:51,436 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=629.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 11:46:05,777 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=648.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:46:07,487 INFO [train.py:893] (0/4) Epoch 1, batch 650, loss[loss=0.7055, simple_loss=0.5894, pruned_loss=0.5394, over 13366.00 frames. ], tot_loss[loss=0.7478, simple_loss=0.6246, pruned_loss=0.6445, over 2551975.07 frames. ], batch size: 109, lr: 4.98e-02, grad_scale: 4.0 2023-04-16 11:46:07,783 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=651.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 11:46:08,331 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=652.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-16 11:46:28,472 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=4.06 vs. limit=2.0 2023-04-16 11:46:28,973 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.82 vs. limit=5.0 2023-04-16 11:46:33,639 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=9.50 vs. limit=5.0 2023-04-16 11:46:42,079 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.101e+02 2.199e+02 2.780e+02 3.736e+02 1.020e+03, threshold=5.560e+02, percent-clipped=40.0 2023-04-16 11:46:42,104 INFO [train.py:893] (0/4) Epoch 1, batch 700, loss[loss=0.6514, simple_loss=0.5496, pruned_loss=0.4769, over 13532.00 frames. ], tot_loss[loss=0.7266, simple_loss=0.6077, pruned_loss=0.6087, over 2576422.83 frames. ], batch size: 83, lr: 4.98e-02, grad_scale: 4.0 2023-04-16 11:46:55,880 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=3.00 vs. limit=2.0 2023-04-16 11:47:09,366 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=739.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 11:47:11,701 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=743.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-16 11:47:16,218 INFO [train.py:893] (0/4) Epoch 1, batch 750, loss[loss=0.6684, simple_loss=0.5662, pruned_loss=0.4761, over 13439.00 frames. ], tot_loss[loss=0.7083, simple_loss=0.5934, pruned_loss=0.577, over 2596413.52 frames. ], batch size: 95, lr: 4.97e-02, grad_scale: 4.0 2023-04-16 11:47:41,179 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=787.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:47:51,428 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.745e+02 2.893e+02 3.595e+02 5.208e+02 1.075e+03, threshold=7.189e+02, percent-clipped=20.0 2023-04-16 11:47:51,456 INFO [train.py:893] (0/4) Epoch 1, batch 800, loss[loss=0.6426, simple_loss=0.548, pruned_loss=0.4438, over 13475.00 frames. ], tot_loss[loss=0.6903, simple_loss=0.5799, pruned_loss=0.5464, over 2610047.25 frames. ], batch size: 79, lr: 4.97e-02, grad_scale: 8.0 2023-04-16 11:48:23,023 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=847.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:48:25,229 INFO [train.py:893] (0/4) Epoch 1, batch 850, loss[loss=0.5814, simple_loss=0.4906, pruned_loss=0.4042, over 13515.00 frames. ], tot_loss[loss=0.6729, simple_loss=0.5673, pruned_loss=0.5176, over 2622787.40 frames. ], batch size: 70, lr: 4.96e-02, grad_scale: 8.0 2023-04-16 11:48:25,719 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.35 vs. limit=2.0 2023-04-16 11:48:37,381 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9899, 1.8874, 2.0002, 1.8911, 1.6729, 2.0011, 1.9292, 1.9829], device='cuda:0'), covar=tensor([0.3165, 0.3797, 0.3123, 0.3870, 0.3376, 0.3651, 0.4889, 0.3099], device='cuda:0'), in_proj_covar=tensor([0.0044, 0.0046, 0.0044, 0.0044, 0.0043, 0.0044, 0.0046, 0.0037], device='cuda:0'), out_proj_covar=tensor([3.9461e-05, 4.5109e-05, 3.9482e-05, 4.2121e-05, 4.1829e-05, 4.3487e-05, 4.3493e-05, 3.4511e-05], device='cuda:0') 2023-04-16 11:48:49,484 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7940, 1.7445, 1.8857, 1.6432, 1.4676, 1.8570, 1.7891, 1.7367], device='cuda:0'), covar=tensor([0.4322, 0.4995, 0.4150, 0.5629, 0.5048, 0.4423, 0.6524, 0.4150], device='cuda:0'), in_proj_covar=tensor([0.0046, 0.0047, 0.0046, 0.0046, 0.0045, 0.0045, 0.0048, 0.0038], device='cuda:0'), out_proj_covar=tensor([4.1198e-05, 4.6215e-05, 4.1257e-05, 4.3780e-05, 4.4060e-05, 4.4032e-05, 4.4724e-05, 3.4904e-05], device='cuda:0') 2023-04-16 11:49:00,833 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.621e+02 2.908e+02 3.677e+02 4.771e+02 1.382e+03, threshold=7.354e+02, percent-clipped=2.0 2023-04-16 11:49:00,858 INFO [train.py:893] (0/4) Epoch 1, batch 900, loss[loss=0.6078, simple_loss=0.5187, pruned_loss=0.4075, over 13386.00 frames. ], tot_loss[loss=0.6553, simple_loss=0.5541, pruned_loss=0.4909, over 2634764.47 frames. ], batch size: 109, lr: 4.96e-02, grad_scale: 8.0 2023-04-16 11:49:05,311 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=908.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-16 11:49:11,504 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=918.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-16 11:49:15,786 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=924.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-16 11:49:23,721 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 11:49:28,961 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2023-04-16 11:49:31,101 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=946.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 11:49:34,004 INFO [train.py:893] (0/4) Epoch 1, batch 950, loss[loss=0.5853, simple_loss=0.4996, pruned_loss=0.3872, over 13368.00 frames. ], tot_loss[loss=0.6351, simple_loss=0.5387, pruned_loss=0.4643, over 2641262.72 frames. ], batch size: 113, lr: 4.96e-02, grad_scale: 8.0 2023-04-16 11:49:34,780 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=952.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 11:50:06,563 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6832, 4.8200, 4.6632, 4.4681, 4.9197, 4.6584, 4.9057, 4.8832], device='cuda:0'), covar=tensor([0.0781, 0.0673, 0.0982, 0.0513, 0.0559, 0.0879, 0.0596, 0.0639], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0034, 0.0036, 0.0035, 0.0037, 0.0034, 0.0025, 0.0032], device='cuda:0'), out_proj_covar=tensor([3.1776e-05, 3.1297e-05, 3.5659e-05, 3.1387e-05, 3.3794e-05, 3.1717e-05, 2.6673e-05, 3.0320e-05], device='cuda:0') 2023-04-16 11:50:09,407 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1000.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 11:50:09,806 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.606e+02 3.300e+02 3.832e+02 5.280e+02 1.324e+03, threshold=7.664e+02, percent-clipped=10.0 2023-04-16 11:50:09,835 INFO [train.py:893] (0/4) Epoch 1, batch 1000, loss[loss=0.5355, simple_loss=0.4579, pruned_loss=0.3493, over 13390.00 frames. ], tot_loss[loss=0.6178, simple_loss=0.5256, pruned_loss=0.4412, over 2646422.87 frames. ], batch size: 62, lr: 4.95e-02, grad_scale: 8.0 2023-04-16 11:50:30,003 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.91 vs. limit=5.0 2023-04-16 11:50:39,213 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1043.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 11:50:44,832 INFO [train.py:893] (0/4) Epoch 1, batch 1050, loss[loss=0.5983, simple_loss=0.5204, pruned_loss=0.3747, over 13388.00 frames. ], tot_loss[loss=0.601, simple_loss=0.5133, pruned_loss=0.4193, over 2644168.77 frames. ], batch size: 113, lr: 4.95e-02, grad_scale: 8.0 2023-04-16 11:51:14,987 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1091.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:51:21,210 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 3.452e+02 4.416e+02 5.723e+02 1.913e+03, threshold=8.832e+02, percent-clipped=10.0 2023-04-16 11:51:21,234 INFO [train.py:893] (0/4) Epoch 1, batch 1100, loss[loss=0.5245, simple_loss=0.462, pruned_loss=0.3186, over 13333.00 frames. ], tot_loss[loss=0.589, simple_loss=0.5052, pruned_loss=0.4016, over 2651187.06 frames. ], batch size: 73, lr: 4.94e-02, grad_scale: 8.0 2023-04-16 11:51:57,241 INFO [train.py:893] (0/4) Epoch 1, batch 1150, loss[loss=0.5651, simple_loss=0.497, pruned_loss=0.3415, over 13246.00 frames. ], tot_loss[loss=0.5767, simple_loss=0.497, pruned_loss=0.3845, over 2653457.52 frames. ], batch size: 124, lr: 4.94e-02, grad_scale: 8.0 2023-04-16 11:51:58,799 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1153.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:52:32,379 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.264e+02 3.530e+02 4.274e+02 5.391e+02 1.250e+03, threshold=8.549e+02, percent-clipped=2.0 2023-04-16 11:52:32,403 INFO [train.py:893] (0/4) Epoch 1, batch 1200, loss[loss=0.5093, simple_loss=0.4483, pruned_loss=0.3054, over 13526.00 frames. ], tot_loss[loss=0.5682, simple_loss=0.4912, pruned_loss=0.3718, over 2654356.73 frames. ], batch size: 83, lr: 4.93e-02, grad_scale: 8.0 2023-04-16 11:52:33,907 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1203.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-16 11:52:42,065 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1214.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-16 11:52:45,889 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1218.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 11:52:49,474 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1224.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 11:52:53,006 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 11:53:01,413 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 11:53:04,702 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1246.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 11:53:09,103 INFO [train.py:893] (0/4) Epoch 1, batch 1250, loss[loss=0.5007, simple_loss=0.4421, pruned_loss=0.2969, over 13381.00 frames. ], tot_loss[loss=0.5586, simple_loss=0.4842, pruned_loss=0.3595, over 2658480.28 frames. ], batch size: 67, lr: 4.92e-02, grad_scale: 8.0 2023-04-16 11:53:18,605 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1266.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:53:22,794 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1272.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 11:53:31,382 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6805, 2.8468, 2.9168, 2.4755, 2.5305, 2.8798, 2.9051, 2.8151], device='cuda:0'), covar=tensor([0.1387, 0.1115, 0.1280, 0.1418, 0.1383, 0.2328, 0.1993, 0.1645], device='cuda:0'), in_proj_covar=tensor([0.0052, 0.0051, 0.0054, 0.0052, 0.0049, 0.0049, 0.0053, 0.0047], device='cuda:0'), out_proj_covar=tensor([4.6152e-05, 4.7392e-05, 4.8093e-05, 4.6546e-05, 4.5542e-05, 5.0474e-05, 4.8590e-05, 4.3812e-05], device='cuda:0') 2023-04-16 11:53:39,154 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1294.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 11:53:44,034 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.345e+02 3.938e+02 4.877e+02 6.128e+02 1.226e+03, threshold=9.754e+02, percent-clipped=3.0 2023-04-16 11:53:44,065 INFO [train.py:893] (0/4) Epoch 1, batch 1300, loss[loss=0.5562, simple_loss=0.4884, pruned_loss=0.3306, over 13406.00 frames. ], tot_loss[loss=0.5507, simple_loss=0.4787, pruned_loss=0.3488, over 2651974.97 frames. ], batch size: 113, lr: 4.92e-02, grad_scale: 8.0 2023-04-16 11:53:51,570 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.11 vs. limit=5.0 2023-04-16 11:54:21,042 INFO [train.py:893] (0/4) Epoch 1, batch 1350, loss[loss=0.4626, simple_loss=0.4151, pruned_loss=0.2651, over 13538.00 frames. ], tot_loss[loss=0.5434, simple_loss=0.4738, pruned_loss=0.3391, over 2653433.25 frames. ], batch size: 76, lr: 4.91e-02, grad_scale: 8.0 2023-04-16 11:54:42,576 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2023-04-16 11:54:46,581 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.85 vs. limit=5.0 2023-04-16 11:54:57,881 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.254e+02 3.467e+02 4.556e+02 5.828e+02 1.072e+03, threshold=9.111e+02, percent-clipped=1.0 2023-04-16 11:54:57,906 INFO [train.py:893] (0/4) Epoch 1, batch 1400, loss[loss=0.5118, simple_loss=0.4605, pruned_loss=0.2911, over 13384.00 frames. ], tot_loss[loss=0.5335, simple_loss=0.4668, pruned_loss=0.3282, over 2657998.34 frames. ], batch size: 113, lr: 4.91e-02, grad_scale: 8.0 2023-04-16 11:55:22,690 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1434.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 11:55:35,007 INFO [train.py:893] (0/4) Epoch 1, batch 1450, loss[loss=0.4871, simple_loss=0.44, pruned_loss=0.2745, over 13521.00 frames. ], tot_loss[loss=0.5254, simple_loss=0.4612, pruned_loss=0.3191, over 2658399.64 frames. ], batch size: 91, lr: 4.90e-02, grad_scale: 8.0 2023-04-16 11:55:40,587 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3541, 4.5341, 4.3046, 4.2553, 4.5982, 4.2101, 4.6855, 4.5029], device='cuda:0'), covar=tensor([0.0393, 0.0331, 0.0457, 0.0298, 0.0344, 0.0464, 0.0244, 0.0310], device='cuda:0'), in_proj_covar=tensor([0.0050, 0.0048, 0.0048, 0.0047, 0.0047, 0.0048, 0.0036, 0.0043], device='cuda:0'), out_proj_covar=tensor([4.4364e-05, 4.4303e-05, 4.5571e-05, 4.0002e-05, 4.3873e-05, 4.4290e-05, 3.8214e-05, 3.8560e-05], device='cuda:0') 2023-04-16 11:55:52,769 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2023-04-16 11:56:08,341 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1495.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 11:56:12,847 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.702e+02 4.580e+02 5.694e+02 7.552e+02 1.475e+03, threshold=1.139e+03, percent-clipped=15.0 2023-04-16 11:56:12,874 INFO [train.py:893] (0/4) Epoch 1, batch 1500, loss[loss=0.4992, simple_loss=0.4515, pruned_loss=0.28, over 13534.00 frames. ], tot_loss[loss=0.5181, simple_loss=0.4563, pruned_loss=0.3108, over 2653655.92 frames. ], batch size: 87, lr: 4.89e-02, grad_scale: 8.0 2023-04-16 11:56:14,496 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1503.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:56:19,285 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1509.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-16 11:56:47,580 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7502, 3.7291, 3.6171, 3.6812, 3.5256, 3.6031, 3.5078, 3.6526], device='cuda:0'), covar=tensor([0.0516, 0.0883, 0.0759, 0.0865, 0.0992, 0.0793, 0.0978, 0.0629], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0023, 0.0030, 0.0022, 0.0026, 0.0028, 0.0026, 0.0026], device='cuda:0'), out_proj_covar=tensor([1.7254e-05, 1.8127e-05, 2.3625e-05, 1.8106e-05, 2.0768e-05, 2.1317e-05, 2.1105e-05, 2.0982e-05], device='cuda:0') 2023-04-16 11:56:49,936 INFO [train.py:893] (0/4) Epoch 1, batch 1550, loss[loss=0.5194, simple_loss=0.4634, pruned_loss=0.2953, over 13372.00 frames. ], tot_loss[loss=0.5124, simple_loss=0.4527, pruned_loss=0.3039, over 2656312.88 frames. ], batch size: 109, lr: 4.89e-02, grad_scale: 8.0 2023-04-16 11:56:50,118 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1551.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 11:57:02,062 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1566.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 11:57:29,613 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.538e+02 4.492e+02 5.897e+02 7.825e+02 1.707e+03, threshold=1.179e+03, percent-clipped=8.0 2023-04-16 11:57:29,649 INFO [train.py:893] (0/4) Epoch 1, batch 1600, loss[loss=0.5105, simple_loss=0.4543, pruned_loss=0.2901, over 13514.00 frames. ], tot_loss[loss=0.5073, simple_loss=0.4499, pruned_loss=0.2975, over 2655290.05 frames. ], batch size: 98, lr: 4.88e-02, grad_scale: 8.0 2023-04-16 11:57:36,717 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1611.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 11:57:49,443 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1627.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-16 11:58:06,421 INFO [train.py:893] (0/4) Epoch 1, batch 1650, loss[loss=0.4887, simple_loss=0.4402, pruned_loss=0.2731, over 13369.00 frames. ], tot_loss[loss=0.504, simple_loss=0.4482, pruned_loss=0.2929, over 2650927.03 frames. ], batch size: 109, lr: 4.87e-02, grad_scale: 8.0 2023-04-16 11:58:23,330 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1672.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-16 11:58:44,062 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2023-04-16 11:58:45,443 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.07 vs. limit=5.0 2023-04-16 11:58:45,611 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.181e+02 4.485e+02 5.687e+02 7.651e+02 1.834e+03, threshold=1.137e+03, percent-clipped=3.0 2023-04-16 11:58:45,637 INFO [train.py:893] (0/4) Epoch 1, batch 1700, loss[loss=0.4235, simple_loss=0.3804, pruned_loss=0.2367, over 12754.00 frames. ], tot_loss[loss=0.4966, simple_loss=0.4438, pruned_loss=0.2855, over 2647994.91 frames. ], batch size: 52, lr: 4.86e-02, grad_scale: 8.0 2023-04-16 11:59:21,369 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.15 vs. limit=5.0 2023-04-16 11:59:24,350 INFO [train.py:893] (0/4) Epoch 1, batch 1750, loss[loss=0.4638, simple_loss=0.426, pruned_loss=0.2526, over 13458.00 frames. ], tot_loss[loss=0.4868, simple_loss=0.4375, pruned_loss=0.277, over 2652724.01 frames. ], batch size: 103, lr: 4.86e-02, grad_scale: 8.0 2023-04-16 11:59:54,215 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2023-04-16 11:59:54,509 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1790.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 12:00:03,038 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.329e+02 4.138e+02 5.716e+02 7.236e+02 4.256e+03, threshold=1.143e+03, percent-clipped=8.0 2023-04-16 12:00:03,062 INFO [train.py:893] (0/4) Epoch 1, batch 1800, loss[loss=0.4199, simple_loss=0.3979, pruned_loss=0.221, over 13537.00 frames. ], tot_loss[loss=0.4817, simple_loss=0.4343, pruned_loss=0.2718, over 2657805.96 frames. ], batch size: 98, lr: 4.85e-02, grad_scale: 8.0 2023-04-16 12:00:09,369 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1809.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 12:00:17,036 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6469, 3.0291, 2.7806, 2.8017, 2.0170, 2.6640, 2.8586, 2.5678], device='cuda:0'), covar=tensor([0.0239, 0.0276, 0.0529, 0.0248, 0.0657, 0.0173, 0.0298, 0.0470], device='cuda:0'), in_proj_covar=tensor([0.0037, 0.0036, 0.0041, 0.0035, 0.0031, 0.0034, 0.0041, 0.0043], device='cuda:0'), out_proj_covar=tensor([3.1402e-05, 3.2600e-05, 3.7486e-05, 2.9526e-05, 3.0240e-05, 2.9350e-05, 3.7311e-05, 3.7910e-05], device='cuda:0') 2023-04-16 12:00:24,592 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5091, 4.5043, 4.5587, 4.3729, 4.9006, 4.2090, 4.8972, 4.7490], device='cuda:0'), covar=tensor([0.0356, 0.0461, 0.0461, 0.0342, 0.0323, 0.0522, 0.0260, 0.0268], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0066, 0.0061, 0.0059, 0.0065, 0.0066, 0.0043, 0.0059], device='cuda:0'), out_proj_covar=tensor([5.9644e-05, 6.3288e-05, 6.1691e-05, 5.4145e-05, 6.5493e-05, 6.5098e-05, 4.8800e-05, 5.4545e-05], device='cuda:0') 2023-04-16 12:00:32,079 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1838.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 12:00:42,435 INFO [train.py:893] (0/4) Epoch 1, batch 1850, loss[loss=0.4824, simple_loss=0.4397, pruned_loss=0.2639, over 13375.00 frames. ], tot_loss[loss=0.4759, simple_loss=0.4304, pruned_loss=0.2667, over 2657123.38 frames. ], batch size: 84, lr: 4.84e-02, grad_scale: 8.0 2023-04-16 12:00:43,958 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 12:00:47,685 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1857.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:00:47,758 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1857.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 12:00:57,444 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1870.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 12:01:10,376 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4455, 3.8932, 3.6033, 3.5314, 3.6813, 3.8105, 3.6061, 3.5072], device='cuda:0'), covar=tensor([0.0307, 0.0195, 0.0141, 0.0350, 0.0181, 0.0175, 0.0262, 0.0254], device='cuda:0'), in_proj_covar=tensor([0.0031, 0.0029, 0.0029, 0.0031, 0.0031, 0.0034, 0.0034, 0.0028], device='cuda:0'), out_proj_covar=tensor([2.5997e-05, 2.2692e-05, 2.2835e-05, 2.5812e-05, 2.4767e-05, 2.7320e-05, 2.8510e-05, 2.1273e-05], device='cuda:0') 2023-04-16 12:01:20,474 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1899.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-16 12:01:21,552 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.333e+02 4.117e+02 5.326e+02 6.729e+02 1.222e+03, threshold=1.065e+03, percent-clipped=1.0 2023-04-16 12:01:21,577 INFO [train.py:893] (0/4) Epoch 1, batch 1900, loss[loss=0.4774, simple_loss=0.4279, pruned_loss=0.2647, over 13268.00 frames. ], tot_loss[loss=0.4688, simple_loss=0.4255, pruned_loss=0.2609, over 2656396.37 frames. ], batch size: 124, lr: 4.83e-02, grad_scale: 8.0 2023-04-16 12:01:35,007 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1918.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 12:01:38,473 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1922.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 12:01:45,461 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1931.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-16 12:01:57,168 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5365, 3.8447, 4.3628, 4.0322, 3.9288, 4.3062, 4.0789, 2.7257], device='cuda:0'), covar=tensor([0.0079, 0.0680, 0.0139, 0.0117, 0.0284, 0.0149, 0.0169, 0.0878], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0021, 0.0019, 0.0014, 0.0024, 0.0023, 0.0016, 0.0030], device='cuda:0'), out_proj_covar=tensor([1.5343e-05, 1.8442e-05, 1.4671e-05, 1.2864e-05, 2.0073e-05, 1.9528e-05, 1.4610e-05, 2.7601e-05], device='cuda:0') 2023-04-16 12:02:00,340 INFO [train.py:893] (0/4) Epoch 1, batch 1950, loss[loss=0.4994, simple_loss=0.4312, pruned_loss=0.2848, over 12055.00 frames. ], tot_loss[loss=0.4642, simple_loss=0.4224, pruned_loss=0.2568, over 2660382.08 frames. ], batch size: 158, lr: 4.83e-02, grad_scale: 8.0 2023-04-16 12:02:06,986 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1146, 3.0488, 3.6336, 2.8907, 3.4609, 2.6370, 3.8280, 2.3037], device='cuda:0'), covar=tensor([0.4270, 0.5096, 0.1779, 0.4144, 0.2019, 1.1235, 0.1113, 1.0988], device='cuda:0'), in_proj_covar=tensor([0.0040, 0.0033, 0.0032, 0.0041, 0.0034, 0.0072, 0.0022, 0.0055], device='cuda:0'), out_proj_covar=tensor([3.7012e-05, 3.4791e-05, 2.8336e-05, 3.9910e-05, 2.9262e-05, 7.1490e-05, 2.2203e-05, 5.4890e-05], device='cuda:0') 2023-04-16 12:02:13,035 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1967.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:02:39,332 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-2000.pt 2023-04-16 12:02:43,812 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.097e+02 4.764e+02 5.634e+02 7.381e+02 1.662e+03, threshold=1.127e+03, percent-clipped=9.0 2023-04-16 12:02:43,848 INFO [train.py:893] (0/4) Epoch 1, batch 2000, loss[loss=0.463, simple_loss=0.434, pruned_loss=0.246, over 13368.00 frames. ], tot_loss[loss=0.4652, simple_loss=0.424, pruned_loss=0.2562, over 2660484.95 frames. ], batch size: 109, lr: 4.82e-02, grad_scale: 16.0 2023-04-16 12:02:50,084 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 12:03:17,493 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.40 vs. limit=5.0 2023-04-16 12:03:26,603 INFO [train.py:893] (0/4) Epoch 1, batch 2050, loss[loss=0.4516, simple_loss=0.4265, pruned_loss=0.2384, over 13494.00 frames. ], tot_loss[loss=0.4591, simple_loss=0.4212, pruned_loss=0.2508, over 2662819.24 frames. ], batch size: 81, lr: 4.81e-02, grad_scale: 16.0 2023-04-16 12:04:00,147 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2090.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 12:04:09,339 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.349e+02 3.615e+02 4.558e+02 5.619e+02 1.123e+03, threshold=9.116e+02, percent-clipped=0.0 2023-04-16 12:04:09,364 INFO [train.py:893] (0/4) Epoch 1, batch 2100, loss[loss=0.45, simple_loss=0.4268, pruned_loss=0.2366, over 13455.00 frames. ], tot_loss[loss=0.4504, simple_loss=0.417, pruned_loss=0.2437, over 2667028.53 frames. ], batch size: 103, lr: 4.80e-02, grad_scale: 16.0 2023-04-16 12:04:37,565 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6783, 3.8273, 3.3756, 3.4406, 3.6176, 3.8050, 3.1845, 2.9023], device='cuda:0'), covar=tensor([0.0140, 0.0322, 0.0489, 0.0196, 0.0276, 0.0199, 0.0489, 0.0645], device='cuda:0'), in_proj_covar=tensor([0.0017, 0.0023, 0.0033, 0.0019, 0.0025, 0.0025, 0.0028, 0.0033], device='cuda:0'), out_proj_covar=tensor([1.5216e-05, 1.8721e-05, 2.8688e-05, 1.6507e-05, 2.0853e-05, 2.0071e-05, 2.4373e-05, 2.8936e-05], device='cuda:0') 2023-04-16 12:04:39,537 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2138.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:04:50,549 INFO [train.py:893] (0/4) Epoch 1, batch 2150, loss[loss=0.4286, simple_loss=0.4105, pruned_loss=0.2234, over 13275.00 frames. ], tot_loss[loss=0.4455, simple_loss=0.4143, pruned_loss=0.2398, over 2664854.45 frames. ], batch size: 124, lr: 4.79e-02, grad_scale: 16.0 2023-04-16 12:05:26,251 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2194.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:05:32,314 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.396e+02 4.119e+02 5.010e+02 7.199e+02 1.601e+03, threshold=1.002e+03, percent-clipped=13.0 2023-04-16 12:05:32,338 INFO [train.py:893] (0/4) Epoch 1, batch 2200, loss[loss=0.4082, simple_loss=0.3827, pruned_loss=0.2169, over 13514.00 frames. ], tot_loss[loss=0.4387, simple_loss=0.4108, pruned_loss=0.2344, over 2663653.07 frames. ], batch size: 70, lr: 4.78e-02, grad_scale: 16.0 2023-04-16 12:05:42,932 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2213.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:05:50,568 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2222.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 12:05:53,476 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2226.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:05:54,495 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.33 vs. limit=5.0 2023-04-16 12:06:15,141 INFO [train.py:893] (0/4) Epoch 1, batch 2250, loss[loss=0.4068, simple_loss=0.3878, pruned_loss=0.2129, over 13534.00 frames. ], tot_loss[loss=0.4296, simple_loss=0.4047, pruned_loss=0.2281, over 2660789.18 frames. ], batch size: 85, lr: 4.77e-02, grad_scale: 16.0 2023-04-16 12:06:28,320 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2267.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 12:06:31,139 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2270.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:06:37,313 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2023-04-16 12:06:51,365 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0863, 4.1129, 4.0032, 3.4297, 2.4356, 3.1406, 3.5874, 4.2288], device='cuda:0'), covar=tensor([0.0068, 0.0176, 0.0143, 0.0500, 0.0840, 0.0542, 0.0185, 0.0061], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0023, 0.0023, 0.0034, 0.0038, 0.0036, 0.0019, 0.0016], device='cuda:0'), out_proj_covar=tensor([1.7669e-05, 1.8169e-05, 1.9040e-05, 2.9930e-05, 3.1901e-05, 3.0955e-05, 1.5797e-05, 1.2566e-05], device='cuda:0') 2023-04-16 12:06:56,372 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.310e+02 3.867e+02 4.930e+02 5.960e+02 1.008e+03, threshold=9.860e+02, percent-clipped=1.0 2023-04-16 12:06:56,397 INFO [train.py:893] (0/4) Epoch 1, batch 2300, loss[loss=0.4149, simple_loss=0.4035, pruned_loss=0.2131, over 13500.00 frames. ], tot_loss[loss=0.4245, simple_loss=0.4017, pruned_loss=0.2243, over 2658553.94 frames. ], batch size: 93, lr: 4.77e-02, grad_scale: 16.0 2023-04-16 12:07:09,526 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2315.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:07:31,041 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2340.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 12:07:39,996 INFO [train.py:893] (0/4) Epoch 1, batch 2350, loss[loss=0.4247, simple_loss=0.4066, pruned_loss=0.2214, over 13379.00 frames. ], tot_loss[loss=0.4188, simple_loss=0.3983, pruned_loss=0.2202, over 2659783.64 frames. ], batch size: 113, lr: 4.76e-02, grad_scale: 16.0 2023-04-16 12:07:51,583 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8141, 3.0759, 2.7271, 3.0717, 3.1025, 3.0111, 3.0127, 2.8885], device='cuda:0'), covar=tensor([0.0325, 0.0283, 0.0302, 0.0207, 0.0237, 0.0353, 0.0297, 0.0493], device='cuda:0'), in_proj_covar=tensor([0.0036, 0.0032, 0.0034, 0.0033, 0.0030, 0.0031, 0.0035, 0.0035], device='cuda:0'), out_proj_covar=tensor([2.9900e-05, 2.7154e-05, 2.7221e-05, 2.8522e-05, 2.5470e-05, 2.6249e-05, 3.0017e-05, 2.8899e-05], device='cuda:0') 2023-04-16 12:08:00,412 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 12:08:22,532 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.956e+02 4.306e+02 5.227e+02 6.724e+02 1.071e+03, threshold=1.045e+03, percent-clipped=3.0 2023-04-16 12:08:22,556 INFO [train.py:893] (0/4) Epoch 1, batch 2400, loss[loss=0.3542, simple_loss=0.3383, pruned_loss=0.185, over 11751.00 frames. ], tot_loss[loss=0.4148, simple_loss=0.3956, pruned_loss=0.2174, over 2658322.07 frames. ], batch size: 48, lr: 4.75e-02, grad_scale: 16.0 2023-04-16 12:08:22,831 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2401.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-16 12:08:32,829 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2023-04-16 12:09:04,547 INFO [train.py:893] (0/4) Epoch 1, batch 2450, loss[loss=0.4141, simple_loss=0.4043, pruned_loss=0.2119, over 13372.00 frames. ], tot_loss[loss=0.4122, simple_loss=0.3945, pruned_loss=0.2153, over 2659284.16 frames. ], batch size: 84, lr: 4.74e-02, grad_scale: 16.0 2023-04-16 12:09:20,691 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0390, 1.7690, 2.4872, 2.3622, 1.7887, 1.5051, 2.1701, 2.3454], device='cuda:0'), covar=tensor([0.0267, 0.0940, 0.0458, 0.0188, 0.0215, 0.0598, 0.0869, 0.0555], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0043, 0.0044, 0.0038, 0.0036, 0.0034, 0.0039, 0.0041], device='cuda:0'), out_proj_covar=tensor([3.3751e-05, 4.0207e-05, 3.9063e-05, 3.4208e-05, 3.1970e-05, 3.9467e-05, 3.6130e-05, 3.5052e-05], device='cuda:0') 2023-04-16 12:09:41,205 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2494.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 12:09:41,294 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8218, 1.9188, 3.0650, 2.8834, 2.9954, 2.2180, 2.5332, 2.9523], device='cuda:0'), covar=tensor([0.0200, 0.1058, 0.0239, 0.0260, 0.0138, 0.0337, 0.0178, 0.0279], device='cuda:0'), in_proj_covar=tensor([0.0026, 0.0038, 0.0027, 0.0024, 0.0024, 0.0028, 0.0023, 0.0025], device='cuda:0'), out_proj_covar=tensor([2.0040e-05, 3.4680e-05, 2.1348e-05, 1.8450e-05, 1.7771e-05, 2.1899e-05, 1.6835e-05, 1.9547e-05], device='cuda:0') 2023-04-16 12:09:47,713 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.314e+02 3.928e+02 4.819e+02 6.017e+02 8.631e+02, threshold=9.637e+02, percent-clipped=0.0 2023-04-16 12:09:47,745 INFO [train.py:893] (0/4) Epoch 1, batch 2500, loss[loss=0.4203, simple_loss=0.3954, pruned_loss=0.2226, over 13402.00 frames. ], tot_loss[loss=0.4086, simple_loss=0.3928, pruned_loss=0.2124, over 2664426.84 frames. ], batch size: 65, lr: 4.73e-02, grad_scale: 16.0 2023-04-16 12:09:57,642 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2513.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:10:08,442 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2526.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:10:23,018 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2542.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:10:30,390 INFO [train.py:893] (0/4) Epoch 1, batch 2550, loss[loss=0.3721, simple_loss=0.3705, pruned_loss=0.1868, over 13539.00 frames. ], tot_loss[loss=0.4072, simple_loss=0.3919, pruned_loss=0.2115, over 2664459.62 frames. ], batch size: 78, lr: 4.72e-02, grad_scale: 16.0 2023-04-16 12:10:39,736 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2561.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:10:50,249 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2574.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:10:52,528 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 12:11:13,921 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.061e+02 4.539e+02 5.406e+02 6.708e+02 1.265e+03, threshold=1.081e+03, percent-clipped=3.0 2023-04-16 12:11:13,946 INFO [train.py:893] (0/4) Epoch 1, batch 2600, loss[loss=0.3772, simple_loss=0.3712, pruned_loss=0.1916, over 13348.00 frames. ], tot_loss[loss=0.4049, simple_loss=0.3908, pruned_loss=0.2096, over 2666338.80 frames. ], batch size: 73, lr: 4.71e-02, grad_scale: 16.0 2023-04-16 12:11:18,180 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8206, 3.1336, 3.3067, 3.5776, 4.1456, 3.2099, 3.8131, 4.0249], device='cuda:0'), covar=tensor([0.0320, 0.0226, 0.0270, 0.0426, 0.0129, 0.0289, 0.0161, 0.0161], device='cuda:0'), in_proj_covar=tensor([0.0037, 0.0025, 0.0031, 0.0037, 0.0032, 0.0033, 0.0031, 0.0027], device='cuda:0'), out_proj_covar=tensor([4.2369e-05, 3.1445e-05, 3.4934e-05, 4.1947e-05, 3.6202e-05, 3.6497e-05, 3.4760e-05, 3.0126e-05], device='cuda:0') 2023-04-16 12:11:18,260 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8681, 3.7353, 2.8994, 3.4934, 3.5240, 3.5204, 3.0410, 2.5139], device='cuda:0'), covar=tensor([0.0163, 0.0414, 0.0696, 0.0155, 0.0391, 0.0286, 0.0783, 0.1461], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0029, 0.0042, 0.0021, 0.0031, 0.0031, 0.0036, 0.0057], device='cuda:0'), out_proj_covar=tensor([2.0215e-05, 2.6405e-05, 4.0368e-05, 2.0249e-05, 2.8504e-05, 2.7805e-05, 3.4039e-05, 5.1835e-05], device='cuda:0') 2023-04-16 12:11:53,081 INFO [train.py:893] (0/4) Epoch 1, batch 2650, loss[loss=0.3693, simple_loss=0.3649, pruned_loss=0.1868, over 13364.00 frames. ], tot_loss[loss=0.4036, simple_loss=0.3898, pruned_loss=0.2088, over 2662345.47 frames. ], batch size: 73, lr: 4.70e-02, grad_scale: 16.0 2023-04-16 12:12:03,123 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1212, 2.9167, 3.0392, 2.8197, 1.6387, 2.7112, 3.3867, 3.4360], device='cuda:0'), covar=tensor([0.0458, 0.1327, 0.1208, 0.2274, 0.3536, 0.1731, 0.0596, 0.0279], device='cuda:0'), in_proj_covar=tensor([0.0028, 0.0027, 0.0028, 0.0047, 0.0052, 0.0045, 0.0024, 0.0020], device='cuda:0'), out_proj_covar=tensor([2.1870e-05, 2.3500e-05, 2.4593e-05, 4.3014e-05, 4.5364e-05, 3.9460e-05, 2.1100e-05, 1.6313e-05], device='cuda:0') 2023-04-16 12:12:08,642 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9461, 4.9106, 5.1519, 4.9813, 5.4667, 4.7742, 5.4403, 5.2973], device='cuda:0'), covar=tensor([0.0373, 0.0321, 0.0352, 0.0439, 0.0323, 0.0431, 0.0281, 0.0284], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0091, 0.0087, 0.0076, 0.0105, 0.0095, 0.0067, 0.0076], device='cuda:0'), out_proj_covar=tensor([9.3919e-05, 1.0150e-04, 9.4389e-05, 7.9818e-05, 1.2025e-04, 1.0934e-04, 7.8254e-05, 7.9597e-05], device='cuda:0') 2023-04-16 12:12:17,950 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2685.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 12:12:25,448 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2696.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 12:12:28,998 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-1.pt 2023-04-16 12:12:53,583 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 12:12:56,984 INFO [train.py:893] (0/4) Epoch 2, batch 0, loss[loss=0.4232, simple_loss=0.4008, pruned_loss=0.2228, over 13374.00 frames. ], tot_loss[loss=0.4232, simple_loss=0.4008, pruned_loss=0.2228, over 13374.00 frames. ], batch size: 118, lr: 4.60e-02, grad_scale: 16.0 2023-04-16 12:12:56,985 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 12:13:16,699 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.5120, 1.7773, 2.1045, 1.9511, 1.7345, 1.8046, 1.8315, 1.9670], device='cuda:0'), covar=tensor([0.0537, 0.1009, 0.0574, 0.0611, 0.0490, 0.1216, 0.1302, 0.1087], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0044, 0.0045, 0.0038, 0.0037, 0.0032, 0.0038, 0.0040], device='cuda:0'), out_proj_covar=tensor([3.3607e-05, 4.0980e-05, 4.1313e-05, 3.3994e-05, 3.4313e-05, 3.8522e-05, 3.5881e-05, 3.5684e-05], device='cuda:0') 2023-04-16 12:13:19,793 INFO [train.py:927] (0/4) Epoch 2, validation: loss=0.3243, simple_loss=0.342, pruned_loss=0.1533, over 2446609.00 frames. 2023-04-16 12:13:19,793 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 12:13:20,438 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.595e+02 4.408e+02 5.707e+02 7.369e+02 1.306e+03, threshold=1.141e+03, percent-clipped=2.0 2023-04-16 12:14:00,168 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2746.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-16 12:14:02,998 INFO [train.py:893] (0/4) Epoch 2, batch 50, loss[loss=0.4149, simple_loss=0.4037, pruned_loss=0.2131, over 13445.00 frames. ], tot_loss[loss=0.3767, simple_loss=0.3692, pruned_loss=0.192, over 603335.56 frames. ], batch size: 79, lr: 4.59e-02, grad_scale: 16.0 2023-04-16 12:14:27,315 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 12:14:27,315 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 12:14:27,315 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 12:14:27,322 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 12:14:27,337 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 12:14:28,013 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 12:14:28,023 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 12:14:47,613 INFO [train.py:893] (0/4) Epoch 2, batch 100, loss[loss=0.4136, simple_loss=0.3905, pruned_loss=0.2183, over 13511.00 frames. ], tot_loss[loss=0.3908, simple_loss=0.3786, pruned_loss=0.2015, over 1057813.96 frames. ], batch size: 91, lr: 4.58e-02, grad_scale: 16.0 2023-04-16 12:14:48,354 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.351e+02 4.141e+02 5.422e+02 6.663e+02 1.362e+03, threshold=1.084e+03, percent-clipped=1.0 2023-04-16 12:14:52,169 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2023-04-16 12:15:08,950 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2824.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 12:15:23,996 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.79 vs. limit=5.0 2023-04-16 12:15:31,829 INFO [train.py:893] (0/4) Epoch 2, batch 150, loss[loss=0.3782, simple_loss=0.3825, pruned_loss=0.187, over 13484.00 frames. ], tot_loss[loss=0.3939, simple_loss=0.3813, pruned_loss=0.2032, over 1401141.43 frames. ], batch size: 93, lr: 4.57e-02, grad_scale: 16.0 2023-04-16 12:15:46,573 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8945, 3.0332, 2.6316, 3.4017, 3.5687, 3.1251, 3.0187, 2.6460], device='cuda:0'), covar=tensor([0.0297, 0.0298, 0.0289, 0.0120, 0.0125, 0.0262, 0.0263, 0.0407], device='cuda:0'), in_proj_covar=tensor([0.0031, 0.0031, 0.0029, 0.0028, 0.0027, 0.0025, 0.0033, 0.0035], device='cuda:0'), out_proj_covar=tensor([3.0165e-05, 2.8859e-05, 2.6602e-05, 2.5094e-05, 2.5548e-05, 2.4255e-05, 3.0499e-05, 3.1388e-05], device='cuda:0') 2023-04-16 12:15:51,141 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2037, 4.0673, 4.3985, 4.7528, 4.9998, 4.1842, 4.7346, 4.5056], device='cuda:0'), covar=tensor([0.0175, 0.0187, 0.0125, 0.0221, 0.0130, 0.0119, 0.0090, 0.0225], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0026, 0.0033, 0.0038, 0.0033, 0.0034, 0.0033, 0.0029], device='cuda:0'), out_proj_covar=tensor([4.9342e-05, 3.5372e-05, 4.0836e-05, 4.7589e-05, 4.1037e-05, 3.9598e-05, 3.8367e-05, 3.5968e-05], device='cuda:0') 2023-04-16 12:16:02,144 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2885.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-16 12:16:14,817 INFO [train.py:893] (0/4) Epoch 2, batch 200, loss[loss=0.4081, simple_loss=0.4006, pruned_loss=0.2078, over 13291.00 frames. ], tot_loss[loss=0.3955, simple_loss=0.3833, pruned_loss=0.2038, over 1678662.98 frames. ], batch size: 124, lr: 4.56e-02, grad_scale: 16.0 2023-04-16 12:16:15,565 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.311e+02 4.527e+02 5.410e+02 6.609e+02 1.468e+03, threshold=1.082e+03, percent-clipped=2.0 2023-04-16 12:16:44,750 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2023-04-16 12:16:47,586 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2937.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 12:16:58,409 INFO [train.py:893] (0/4) Epoch 2, batch 250, loss[loss=0.365, simple_loss=0.3757, pruned_loss=0.1772, over 13520.00 frames. ], tot_loss[loss=0.392, simple_loss=0.3811, pruned_loss=0.2015, over 1893784.08 frames. ], batch size: 76, lr: 4.55e-02, grad_scale: 16.0 2023-04-16 12:17:01,960 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2023-04-16 12:17:38,665 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2996.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 12:17:41,004 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2998.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-16 12:17:42,573 INFO [train.py:893] (0/4) Epoch 2, batch 300, loss[loss=0.411, simple_loss=0.402, pruned_loss=0.21, over 13534.00 frames. ], tot_loss[loss=0.39, simple_loss=0.3807, pruned_loss=0.1996, over 2058146.28 frames. ], batch size: 78, lr: 4.54e-02, grad_scale: 16.0 2023-04-16 12:17:43,328 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.732e+02 4.238e+02 5.185e+02 6.684e+02 1.179e+03, threshold=1.037e+03, percent-clipped=2.0 2023-04-16 12:17:43,953 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-16 12:17:59,125 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3019.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 12:18:17,778 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3041.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:18:20,930 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3044.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 12:18:25,554 INFO [train.py:893] (0/4) Epoch 2, batch 350, loss[loss=0.3825, simple_loss=0.3895, pruned_loss=0.1878, over 13453.00 frames. ], tot_loss[loss=0.3912, simple_loss=0.3815, pruned_loss=0.2005, over 2185428.36 frames. ], batch size: 100, lr: 4.53e-02, grad_scale: 16.0 2023-04-16 12:18:43,056 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0734, 4.0691, 3.5887, 3.4788, 3.6631, 3.5936, 4.0951, 4.2777], device='cuda:0'), covar=tensor([0.0257, 0.0338, 0.0578, 0.0531, 0.0662, 0.0436, 0.0348, 0.0219], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0073, 0.0059, 0.0080, 0.0060, 0.0068, 0.0062, 0.0056], device='cuda:0'), out_proj_covar=tensor([7.5960e-05, 8.0880e-05, 6.9644e-05, 1.0202e-04, 7.7048e-05, 8.0244e-05, 7.2926e-05, 6.5581e-05], device='cuda:0') 2023-04-16 12:18:47,049 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.04 vs. limit=5.0 2023-04-16 12:18:52,596 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3080.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-16 12:19:10,159 INFO [train.py:893] (0/4) Epoch 2, batch 400, loss[loss=0.369, simple_loss=0.3628, pruned_loss=0.1876, over 13427.00 frames. ], tot_loss[loss=0.3888, simple_loss=0.3806, pruned_loss=0.1986, over 2288874.77 frames. ], batch size: 65, lr: 4.52e-02, grad_scale: 16.0 2023-04-16 12:19:10,821 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.297e+02 4.068e+02 5.091e+02 6.517e+02 1.067e+03, threshold=1.018e+03, percent-clipped=1.0 2023-04-16 12:19:54,902 INFO [train.py:893] (0/4) Epoch 2, batch 450, loss[loss=0.3933, simple_loss=0.3671, pruned_loss=0.2097, over 13173.00 frames. ], tot_loss[loss=0.3902, simple_loss=0.3822, pruned_loss=0.1991, over 2371759.38 frames. ], batch size: 58, lr: 4.51e-02, grad_scale: 16.0 2023-04-16 12:20:18,781 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 12:20:21,147 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3180.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:20:38,551 INFO [train.py:893] (0/4) Epoch 2, batch 500, loss[loss=0.4618, simple_loss=0.4401, pruned_loss=0.2417, over 13478.00 frames. ], tot_loss[loss=0.3904, simple_loss=0.3828, pruned_loss=0.199, over 2437777.78 frames. ], batch size: 106, lr: 4.50e-02, grad_scale: 16.0 2023-04-16 12:20:39,291 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.435e+02 4.634e+02 5.605e+02 7.324e+02 2.141e+03, threshold=1.121e+03, percent-clipped=6.0 2023-04-16 12:21:22,354 INFO [train.py:893] (0/4) Epoch 2, batch 550, loss[loss=0.3586, simple_loss=0.3417, pruned_loss=0.1877, over 12848.00 frames. ], tot_loss[loss=0.3872, simple_loss=0.3811, pruned_loss=0.1966, over 2487558.05 frames. ], batch size: 52, lr: 4.49e-02, grad_scale: 16.0 2023-04-16 12:21:48,039 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.92 vs. limit=5.0 2023-04-16 12:21:55,512 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3288.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 12:22:00,126 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3293.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:22:05,325 INFO [train.py:893] (0/4) Epoch 2, batch 600, loss[loss=0.3701, simple_loss=0.3473, pruned_loss=0.1964, over 11742.00 frames. ], tot_loss[loss=0.3839, simple_loss=0.3782, pruned_loss=0.1948, over 2519261.85 frames. ], batch size: 48, lr: 4.48e-02, grad_scale: 16.0 2023-04-16 12:22:06,085 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.737e+02 4.280e+02 5.504e+02 6.489e+02 1.365e+03, threshold=1.101e+03, percent-clipped=1.0 2023-04-16 12:22:42,632 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3341.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:22:49,848 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3349.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-16 12:22:50,297 INFO [train.py:893] (0/4) Epoch 2, batch 650, loss[loss=0.3639, simple_loss=0.3705, pruned_loss=0.1787, over 13442.00 frames. ], tot_loss[loss=0.3807, simple_loss=0.3762, pruned_loss=0.1926, over 2551767.78 frames. ], batch size: 106, lr: 4.47e-02, grad_scale: 16.0 2023-04-16 12:23:12,818 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3375.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:23:24,528 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3389.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:23:27,367 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-04-16 12:23:34,477 INFO [train.py:893] (0/4) Epoch 2, batch 700, loss[loss=0.3523, simple_loss=0.3572, pruned_loss=0.1737, over 13339.00 frames. ], tot_loss[loss=0.3771, simple_loss=0.374, pruned_loss=0.1901, over 2577371.35 frames. ], batch size: 118, lr: 4.46e-02, grad_scale: 16.0 2023-04-16 12:23:35,209 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.882e+02 4.256e+02 4.913e+02 6.838e+02 1.348e+03, threshold=9.827e+02, percent-clipped=2.0 2023-04-16 12:24:18,227 INFO [train.py:893] (0/4) Epoch 2, batch 750, loss[loss=0.4171, simple_loss=0.3938, pruned_loss=0.2202, over 13377.00 frames. ], tot_loss[loss=0.3778, simple_loss=0.3743, pruned_loss=0.1907, over 2600603.19 frames. ], batch size: 73, lr: 4.45e-02, grad_scale: 16.0 2023-04-16 12:24:44,559 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3480.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:24:52,068 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5529, 3.8163, 3.8202, 4.0162, 4.4375, 3.6682, 4.0463, 4.3151], device='cuda:0'), covar=tensor([0.0285, 0.0140, 0.0205, 0.0375, 0.0130, 0.0219, 0.0191, 0.0218], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0023, 0.0031, 0.0040, 0.0034, 0.0033, 0.0034, 0.0029], device='cuda:0'), out_proj_covar=tensor([5.7577e-05, 3.9903e-05, 4.8206e-05, 6.0150e-05, 5.1459e-05, 4.8233e-05, 4.8542e-05, 4.4855e-05], device='cuda:0') 2023-04-16 12:24:55,125 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9273, 2.2139, 1.7964, 1.8767, 1.7119, 1.5955, 1.6503, 2.0435], device='cuda:0'), covar=tensor([0.0246, 0.0670, 0.1245, 0.0331, 0.0534, 0.0115, 0.0624, 0.0433], device='cuda:0'), in_proj_covar=tensor([0.0036, 0.0036, 0.0035, 0.0035, 0.0027, 0.0031, 0.0042, 0.0039], device='cuda:0'), out_proj_covar=tensor([3.2230e-05, 3.5692e-05, 3.5652e-05, 2.9220e-05, 3.1451e-05, 2.4333e-05, 3.7124e-05, 3.3412e-05], device='cuda:0') 2023-04-16 12:25:01,845 INFO [train.py:893] (0/4) Epoch 2, batch 800, loss[loss=0.3902, simple_loss=0.3975, pruned_loss=0.1914, over 13528.00 frames. ], tot_loss[loss=0.3802, simple_loss=0.3768, pruned_loss=0.1918, over 2614541.55 frames. ], batch size: 87, lr: 4.44e-02, grad_scale: 16.0 2023-04-16 12:25:02,610 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.861e+02 5.123e+02 6.281e+02 8.038e+02 1.655e+03, threshold=1.256e+03, percent-clipped=13.0 2023-04-16 12:25:27,273 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3528.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:25:43,008 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3546.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 12:25:45,909 INFO [train.py:893] (0/4) Epoch 2, batch 850, loss[loss=0.3783, simple_loss=0.3507, pruned_loss=0.203, over 12528.00 frames. ], tot_loss[loss=0.3821, simple_loss=0.3782, pruned_loss=0.1931, over 2622277.29 frames. ], batch size: 51, lr: 4.43e-02, grad_scale: 16.0 2023-04-16 12:26:13,360 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.21 vs. limit=2.0 2023-04-16 12:26:25,107 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3593.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:26:31,495 INFO [train.py:893] (0/4) Epoch 2, batch 900, loss[loss=0.3668, simple_loss=0.3639, pruned_loss=0.1849, over 13510.00 frames. ], tot_loss[loss=0.3794, simple_loss=0.3764, pruned_loss=0.1912, over 2630871.17 frames. ], batch size: 70, lr: 4.42e-02, grad_scale: 16.0 2023-04-16 12:26:32,229 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.963e+02 5.281e+02 6.987e+02 9.717e+02 1.667e+03, threshold=1.397e+03, percent-clipped=8.0 2023-04-16 12:26:38,185 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3607.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-16 12:27:01,646 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 12:27:08,107 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3641.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:27:10,657 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3644.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 12:27:15,900 INFO [train.py:893] (0/4) Epoch 2, batch 950, loss[loss=0.3633, simple_loss=0.3651, pruned_loss=0.1807, over 13534.00 frames. ], tot_loss[loss=0.377, simple_loss=0.3735, pruned_loss=0.1903, over 2631692.87 frames. ], batch size: 91, lr: 4.41e-02, grad_scale: 16.0 2023-04-16 12:27:36,733 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2023-04-16 12:27:38,963 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3675.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:27:39,030 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0912, 4.1366, 3.3596, 4.1161, 3.9493, 3.7213, 3.6042, 2.7017], device='cuda:0'), covar=tensor([0.0157, 0.0295, 0.0502, 0.0107, 0.0197, 0.0342, 0.0503, 0.1674], device='cuda:0'), in_proj_covar=tensor([0.0034, 0.0049, 0.0074, 0.0034, 0.0049, 0.0059, 0.0062, 0.0113], device='cuda:0'), out_proj_covar=tensor([4.0109e-05, 5.3914e-05, 7.7947e-05, 3.9438e-05, 5.3338e-05, 6.5134e-05, 6.7577e-05, 1.1019e-04], device='cuda:0') 2023-04-16 12:27:40,422 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3676.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 12:28:02,034 INFO [train.py:893] (0/4) Epoch 2, batch 1000, loss[loss=0.3472, simple_loss=0.3553, pruned_loss=0.1696, over 13444.00 frames. ], tot_loss[loss=0.3709, simple_loss=0.3687, pruned_loss=0.1865, over 2635542.05 frames. ], batch size: 106, lr: 4.40e-02, grad_scale: 16.0 2023-04-16 12:28:02,797 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.113e+02 4.777e+02 5.956e+02 7.807e+02 2.216e+03, threshold=1.191e+03, percent-clipped=1.0 2023-04-16 12:28:23,451 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3723.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:28:35,896 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3737.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-16 12:28:47,098 INFO [train.py:893] (0/4) Epoch 2, batch 1050, loss[loss=0.3726, simple_loss=0.3824, pruned_loss=0.1814, over 13084.00 frames. ], tot_loss[loss=0.3652, simple_loss=0.3647, pruned_loss=0.1829, over 2641109.42 frames. ], batch size: 142, lr: 4.39e-02, grad_scale: 16.0 2023-04-16 12:29:25,842 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-04-16 12:29:32,976 INFO [train.py:893] (0/4) Epoch 2, batch 1100, loss[loss=0.338, simple_loss=0.3536, pruned_loss=0.1612, over 13352.00 frames. ], tot_loss[loss=0.3642, simple_loss=0.3647, pruned_loss=0.1818, over 2646987.10 frames. ], batch size: 73, lr: 4.37e-02, grad_scale: 16.0 2023-04-16 12:29:33,668 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.108e+02 5.042e+02 6.518e+02 8.032e+02 1.563e+03, threshold=1.304e+03, percent-clipped=1.0 2023-04-16 12:29:58,756 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9709, 3.5831, 3.7154, 2.9862, 3.3137, 2.0975, 3.9629, 2.6202], device='cuda:0'), covar=tensor([0.1382, 0.0686, 0.0428, 0.0978, 0.0508, 0.2787, 0.0262, 0.2239], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0049, 0.0052, 0.0067, 0.0049, 0.0081, 0.0038, 0.0076], device='cuda:0'), out_proj_covar=tensor([8.5007e-05, 5.7023e-05, 5.4465e-05, 7.3419e-05, 5.1868e-05, 8.8214e-05, 4.2619e-05, 8.6612e-05], device='cuda:0') 2023-04-16 12:30:18,716 INFO [train.py:893] (0/4) Epoch 2, batch 1150, loss[loss=0.3213, simple_loss=0.3402, pruned_loss=0.1512, over 13369.00 frames. ], tot_loss[loss=0.3616, simple_loss=0.364, pruned_loss=0.1796, over 2650524.65 frames. ], batch size: 73, lr: 4.36e-02, grad_scale: 16.0 2023-04-16 12:30:55,491 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6898, 3.9117, 3.7059, 3.8774, 3.6743, 4.3975, 3.9502, 4.5375], device='cuda:0'), covar=tensor([0.0368, 0.0269, 0.0258, 0.0382, 0.0349, 0.0425, 0.0301, 0.0123], device='cuda:0'), in_proj_covar=tensor([0.0053, 0.0044, 0.0045, 0.0058, 0.0045, 0.0053, 0.0043, 0.0046], device='cuda:0'), out_proj_covar=tensor([8.4093e-05, 6.9588e-05, 7.5040e-05, 9.3748e-05, 8.1272e-05, 8.6881e-05, 6.7992e-05, 7.2084e-05], device='cuda:0') 2023-04-16 12:30:55,788 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2023-04-16 12:31:04,125 INFO [train.py:893] (0/4) Epoch 2, batch 1200, loss[loss=0.3537, simple_loss=0.3674, pruned_loss=0.17, over 13231.00 frames. ], tot_loss[loss=0.3608, simple_loss=0.3641, pruned_loss=0.1788, over 2646625.04 frames. ], batch size: 117, lr: 4.35e-02, grad_scale: 16.0 2023-04-16 12:31:04,948 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.864e+02 4.882e+02 5.871e+02 7.581e+02 1.368e+03, threshold=1.174e+03, percent-clipped=2.0 2023-04-16 12:31:05,873 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3902.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 12:31:27,888 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3926.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:31:31,000 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 12:31:43,956 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 12:31:45,029 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3944.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 12:31:50,589 INFO [train.py:893] (0/4) Epoch 2, batch 1250, loss[loss=0.304, simple_loss=0.3053, pruned_loss=0.1513, over 12585.00 frames. ], tot_loss[loss=0.3623, simple_loss=0.3651, pruned_loss=0.1798, over 2650183.62 frames. ], batch size: 51, lr: 4.34e-02, grad_scale: 16.0 2023-04-16 12:32:01,490 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9707, 4.2368, 3.7338, 4.1697, 4.1437, 4.6189, 4.2060, 4.7197], device='cuda:0'), covar=tensor([0.0316, 0.0210, 0.0319, 0.0281, 0.0251, 0.0286, 0.0237, 0.0117], device='cuda:0'), in_proj_covar=tensor([0.0052, 0.0044, 0.0045, 0.0058, 0.0044, 0.0054, 0.0043, 0.0046], device='cuda:0'), out_proj_covar=tensor([8.5249e-05, 7.0690e-05, 7.7548e-05, 9.5141e-05, 8.0422e-05, 8.8839e-05, 6.7795e-05, 7.2228e-05], device='cuda:0') 2023-04-16 12:32:24,499 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3987.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:32:29,291 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3992.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 12:32:36,010 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-4000.pt 2023-04-16 12:32:39,808 INFO [train.py:893] (0/4) Epoch 2, batch 1300, loss[loss=0.3566, simple_loss=0.3653, pruned_loss=0.174, over 13434.00 frames. ], tot_loss[loss=0.364, simple_loss=0.3666, pruned_loss=0.1807, over 2654047.33 frames. ], batch size: 88, lr: 4.33e-02, grad_scale: 32.0 2023-04-16 12:32:40,631 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.644e+02 4.834e+02 5.973e+02 8.685e+02 1.398e+03, threshold=1.195e+03, percent-clipped=5.0 2023-04-16 12:33:10,696 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4032.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 12:33:27,001 INFO [train.py:893] (0/4) Epoch 2, batch 1350, loss[loss=0.3557, simple_loss=0.3655, pruned_loss=0.1729, over 13384.00 frames. ], tot_loss[loss=0.3635, simple_loss=0.3666, pruned_loss=0.1802, over 2658258.82 frames. ], batch size: 118, lr: 4.32e-02, grad_scale: 32.0 2023-04-16 12:33:34,307 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-16 12:33:36,582 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.17 vs. limit=2.0 2023-04-16 12:33:45,564 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6360, 2.1762, 1.8646, 2.0057, 1.3308, 1.7437, 1.6458, 1.9671], device='cuda:0'), covar=tensor([0.0416, 0.0473, 0.0920, 0.0252, 0.0271, 0.0159, 0.0488, 0.0320], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0033, 0.0036, 0.0034, 0.0030, 0.0031, 0.0047, 0.0039], device='cuda:0'), out_proj_covar=tensor([3.7193e-05, 3.4504e-05, 3.8106e-05, 2.9060e-05, 3.4808e-05, 2.5330e-05, 4.4975e-05, 3.4479e-05], device='cuda:0') 2023-04-16 12:34:00,017 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4037, 4.0532, 3.5059, 3.8757, 3.9085, 4.3642, 3.9183, 4.4101], device='cuda:0'), covar=tensor([0.0369, 0.0168, 0.0261, 0.0355, 0.0161, 0.0158, 0.0232, 0.0112], device='cuda:0'), in_proj_covar=tensor([0.0053, 0.0045, 0.0045, 0.0060, 0.0043, 0.0053, 0.0042, 0.0046], device='cuda:0'), out_proj_covar=tensor([8.8033e-05, 7.4546e-05, 7.8687e-05, 9.9686e-05, 7.9482e-05, 9.0240e-05, 6.7961e-05, 7.4581e-05], device='cuda:0') 2023-04-16 12:34:13,919 INFO [train.py:893] (0/4) Epoch 2, batch 1400, loss[loss=0.3512, simple_loss=0.3593, pruned_loss=0.1715, over 13432.00 frames. ], tot_loss[loss=0.3606, simple_loss=0.3646, pruned_loss=0.1783, over 2661569.30 frames. ], batch size: 95, lr: 4.31e-02, grad_scale: 32.0 2023-04-16 12:34:14,700 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.422e+02 4.905e+02 6.254e+02 7.280e+02 1.526e+03, threshold=1.251e+03, percent-clipped=2.0 2023-04-16 12:34:50,557 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.07 vs. limit=2.0 2023-04-16 12:34:51,883 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=4141.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:34:59,849 INFO [train.py:893] (0/4) Epoch 2, batch 1450, loss[loss=0.3772, simple_loss=0.3841, pruned_loss=0.1852, over 13532.00 frames. ], tot_loss[loss=0.3606, simple_loss=0.3643, pruned_loss=0.1784, over 2660142.05 frames. ], batch size: 98, lr: 4.30e-02, grad_scale: 32.0 2023-04-16 12:35:19,731 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.0330, 4.8827, 5.1459, 5.0511, 5.3853, 4.8414, 5.3764, 5.3201], device='cuda:0'), covar=tensor([0.0227, 0.0315, 0.0290, 0.0242, 0.0378, 0.0375, 0.0280, 0.0270], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0113, 0.0105, 0.0088, 0.0138, 0.0119, 0.0082, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 12:35:47,356 INFO [train.py:893] (0/4) Epoch 2, batch 1500, loss[loss=0.3379, simple_loss=0.3583, pruned_loss=0.1587, over 13363.00 frames. ], tot_loss[loss=0.3592, simple_loss=0.3633, pruned_loss=0.1775, over 2656482.98 frames. ], batch size: 118, lr: 4.29e-02, grad_scale: 32.0 2023-04-16 12:35:48,168 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.727e+02 4.901e+02 5.855e+02 6.848e+02 1.405e+03, threshold=1.171e+03, percent-clipped=2.0 2023-04-16 12:35:49,215 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4202.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 12:35:49,277 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=4202.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:36:07,953 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3985, 4.7544, 4.8462, 4.7122, 4.7977, 4.8222, 5.1496, 4.7949], device='cuda:0'), covar=tensor([0.0835, 0.0815, 0.1936, 0.2479, 0.0617, 0.1006, 0.1023, 0.0744], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0144, 0.0183, 0.0189, 0.0110, 0.0170, 0.0192, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001], device='cuda:0') 2023-04-16 12:36:09,785 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4435, 2.8717, 3.3634, 2.8247, 3.7984, 3.2192, 3.6324, 3.3191], device='cuda:0'), covar=tensor([0.0099, 0.0192, 0.0138, 0.0108, 0.0085, 0.0153, 0.0133, 0.0095], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0026, 0.0023, 0.0025, 0.0019, 0.0024, 0.0018, 0.0021], device='cuda:0'), out_proj_covar=tensor([4.0967e-05, 4.5788e-05, 3.6944e-05, 4.4069e-05, 3.1843e-05, 4.1360e-05, 3.0606e-05, 3.5801e-05], device='cuda:0') 2023-04-16 12:36:32,960 INFO [train.py:893] (0/4) Epoch 2, batch 1550, loss[loss=0.3569, simple_loss=0.3675, pruned_loss=0.1731, over 13233.00 frames. ], tot_loss[loss=0.3578, simple_loss=0.3623, pruned_loss=0.1766, over 2659305.98 frames. ], batch size: 124, lr: 4.28e-02, grad_scale: 32.0 2023-04-16 12:36:33,147 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4250.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 12:37:02,836 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4282.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:37:09,372 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=4288.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:37:16,983 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8484, 3.8522, 2.9045, 3.7206, 3.7987, 2.9476, 3.3056, 2.6949], device='cuda:0'), covar=tensor([0.0171, 0.0296, 0.0895, 0.0128, 0.0201, 0.0762, 0.0662, 0.1815], device='cuda:0'), in_proj_covar=tensor([0.0044, 0.0056, 0.0093, 0.0041, 0.0059, 0.0076, 0.0077, 0.0131], device='cuda:0'), out_proj_covar=tensor([5.9918e-05, 7.0432e-05, 1.0787e-04, 5.2248e-05, 7.1871e-05, 9.0500e-05, 9.3271e-05, 1.3882e-04], device='cuda:0') 2023-04-16 12:37:19,994 INFO [train.py:893] (0/4) Epoch 2, batch 1600, loss[loss=0.3609, simple_loss=0.372, pruned_loss=0.175, over 13502.00 frames. ], tot_loss[loss=0.3556, simple_loss=0.3611, pruned_loss=0.1751, over 2652087.10 frames. ], batch size: 91, lr: 4.27e-02, grad_scale: 32.0 2023-04-16 12:37:20,816 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.949e+02 4.864e+02 6.312e+02 7.658e+02 1.439e+03, threshold=1.262e+03, percent-clipped=4.0 2023-04-16 12:37:41,098 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9188, 2.3596, 2.6894, 2.9942, 2.7641, 3.1023, 2.4916, 1.8577], device='cuda:0'), covar=tensor([0.0235, 0.0556, 0.0254, 0.0099, 0.0418, 0.0181, 0.0310, 0.0947], device='cuda:0'), in_proj_covar=tensor([0.0037, 0.0058, 0.0044, 0.0037, 0.0033, 0.0034, 0.0042, 0.0064], device='cuda:0'), out_proj_covar=tensor([4.6739e-05, 6.4129e-05, 5.3200e-05, 4.1203e-05, 4.2234e-05, 4.4614e-05, 4.9145e-05, 6.9530e-05], device='cuda:0') 2023-04-16 12:37:50,362 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4332.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 12:37:51,251 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9681, 3.7211, 3.6442, 3.0392, 3.3567, 2.2411, 3.7827, 2.8276], device='cuda:0'), covar=tensor([0.1422, 0.0500, 0.0373, 0.0993, 0.0579, 0.2779, 0.0323, 0.2839], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0052, 0.0059, 0.0078, 0.0063, 0.0097, 0.0045, 0.0107], device='cuda:0'), out_proj_covar=tensor([1.0218e-04, 5.9849e-05, 6.5036e-05, 8.8904e-05, 6.7843e-05, 1.0786e-04, 5.1482e-05, 1.3018e-04], device='cuda:0') 2023-04-16 12:38:06,063 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=4349.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:38:06,565 INFO [train.py:893] (0/4) Epoch 2, batch 1650, loss[loss=0.3013, simple_loss=0.3243, pruned_loss=0.1391, over 13376.00 frames. ], tot_loss[loss=0.3576, simple_loss=0.3633, pruned_loss=0.1759, over 2652010.76 frames. ], batch size: 73, lr: 4.26e-02, grad_scale: 32.0 2023-04-16 12:38:34,802 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4380.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 12:38:52,834 INFO [train.py:893] (0/4) Epoch 2, batch 1700, loss[loss=0.3391, simple_loss=0.3486, pruned_loss=0.1649, over 13193.00 frames. ], tot_loss[loss=0.3576, simple_loss=0.3637, pruned_loss=0.1758, over 2652764.49 frames. ], batch size: 132, lr: 4.25e-02, grad_scale: 16.0 2023-04-16 12:38:53,189 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7898, 3.8486, 2.9378, 3.7726, 3.8498, 2.9001, 3.2647, 2.6850], device='cuda:0'), covar=tensor([0.0165, 0.0393, 0.0852, 0.0117, 0.0263, 0.0697, 0.0783, 0.1743], device='cuda:0'), in_proj_covar=tensor([0.0044, 0.0058, 0.0096, 0.0042, 0.0061, 0.0078, 0.0077, 0.0133], device='cuda:0'), out_proj_covar=tensor([6.0610e-05, 7.3778e-05, 1.1243e-04, 5.4297e-05, 7.4692e-05, 9.3880e-05, 9.4876e-05, 1.4217e-04], device='cuda:0') 2023-04-16 12:38:54,507 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.588e+02 4.984e+02 6.280e+02 7.975e+02 1.860e+03, threshold=1.256e+03, percent-clipped=4.0 2023-04-16 12:39:38,759 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5793, 4.2181, 3.5951, 4.3301, 3.9848, 4.7275, 4.2793, 4.8065], device='cuda:0'), covar=tensor([0.0434, 0.0213, 0.0278, 0.0288, 0.0275, 0.0250, 0.0197, 0.0133], device='cuda:0'), in_proj_covar=tensor([0.0054, 0.0047, 0.0047, 0.0062, 0.0045, 0.0056, 0.0044, 0.0049], device='cuda:0'), out_proj_covar=tensor([9.4274e-05, 8.3367e-05, 8.9416e-05, 1.0834e-04, 8.8156e-05, 1.0371e-04, 7.7795e-05, 8.6173e-05], device='cuda:0') 2023-04-16 12:39:39,337 INFO [train.py:893] (0/4) Epoch 2, batch 1750, loss[loss=0.2919, simple_loss=0.305, pruned_loss=0.1394, over 13157.00 frames. ], tot_loss[loss=0.3528, simple_loss=0.36, pruned_loss=0.1728, over 2656310.04 frames. ], batch size: 58, lr: 4.24e-02, grad_scale: 16.0 2023-04-16 12:39:43,274 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-16 12:40:22,247 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4497.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:40:23,221 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9247, 3.7825, 3.5861, 3.7826, 2.2864, 3.6757, 3.7915, 1.9377], device='cuda:0'), covar=tensor([0.0080, 0.0485, 0.0197, 0.0184, 0.2031, 0.0344, 0.0497, 0.2966], device='cuda:0'), in_proj_covar=tensor([0.0040, 0.0054, 0.0052, 0.0038, 0.0103, 0.0068, 0.0059, 0.0114], device='cuda:0'), out_proj_covar=tensor([4.1705e-05, 6.6496e-05, 5.7154e-05, 4.4554e-05, 1.1029e-04, 7.1235e-05, 6.9601e-05, 1.1772e-04], device='cuda:0') 2023-04-16 12:40:24,554 INFO [train.py:893] (0/4) Epoch 2, batch 1800, loss[loss=0.3113, simple_loss=0.3308, pruned_loss=0.1459, over 13522.00 frames. ], tot_loss[loss=0.3493, simple_loss=0.3575, pruned_loss=0.1705, over 2659364.60 frames. ], batch size: 76, lr: 4.23e-02, grad_scale: 16.0 2023-04-16 12:40:26,850 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.000e+02 4.797e+02 5.759e+02 6.677e+02 1.553e+03, threshold=1.152e+03, percent-clipped=1.0 2023-04-16 12:40:41,295 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8625, 3.6929, 3.5264, 3.7913, 2.1197, 3.3418, 3.7158, 1.6587], device='cuda:0'), covar=tensor([0.0124, 0.0444, 0.0181, 0.0130, 0.1899, 0.0394, 0.0358, 0.2836], device='cuda:0'), in_proj_covar=tensor([0.0041, 0.0054, 0.0052, 0.0038, 0.0104, 0.0069, 0.0060, 0.0116], device='cuda:0'), out_proj_covar=tensor([4.2778e-05, 6.7204e-05, 5.8065e-05, 4.4843e-05, 1.1162e-04, 7.2217e-05, 7.0756e-05, 1.1967e-04], device='cuda:0') 2023-04-16 12:40:42,310 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2023-04-16 12:40:51,159 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.5682, 2.0223, 3.6810, 3.1927, 3.2390, 3.4308, 2.8173, 2.5269], device='cuda:0'), covar=tensor([0.1328, 0.1765, 0.0217, 0.0321, 0.0143, 0.0180, 0.0194, 0.0708], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0106, 0.0051, 0.0048, 0.0047, 0.0052, 0.0040, 0.0065], device='cuda:0'), out_proj_covar=tensor([7.4468e-05, 1.0775e-04, 4.5556e-05, 4.3964e-05, 4.0828e-05, 4.8397e-05, 3.7157e-05, 6.1630e-05], device='cuda:0') 2023-04-16 12:41:07,504 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8459, 4.8620, 5.0662, 4.9498, 5.3004, 4.7510, 5.2754, 5.2207], device='cuda:0'), covar=tensor([0.0323, 0.0347, 0.0388, 0.0358, 0.0470, 0.0431, 0.0396, 0.0358], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0120, 0.0109, 0.0090, 0.0145, 0.0126, 0.0085, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 12:41:10,033 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0159, 2.5927, 2.1944, 3.3038, 3.4569, 3.4195, 2.3848, 1.9443], device='cuda:0'), covar=tensor([0.0229, 0.0595, 0.0444, 0.0123, 0.0193, 0.0168, 0.0390, 0.0895], device='cuda:0'), in_proj_covar=tensor([0.0040, 0.0063, 0.0049, 0.0040, 0.0036, 0.0036, 0.0047, 0.0070], device='cuda:0'), out_proj_covar=tensor([5.2435e-05, 7.1546e-05, 6.0638e-05, 4.6965e-05, 4.7101e-05, 4.7109e-05, 5.6235e-05, 7.7415e-05], device='cuda:0') 2023-04-16 12:41:11,402 INFO [train.py:893] (0/4) Epoch 2, batch 1850, loss[loss=0.3456, simple_loss=0.3489, pruned_loss=0.1711, over 13405.00 frames. ], tot_loss[loss=0.3461, simple_loss=0.3551, pruned_loss=0.1686, over 2659886.63 frames. ], batch size: 65, lr: 4.22e-02, grad_scale: 16.0 2023-04-16 12:41:12,052 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2023-04-16 12:41:16,378 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 12:41:21,824 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-04-16 12:41:41,019 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4582.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:41:58,133 INFO [train.py:893] (0/4) Epoch 2, batch 1900, loss[loss=0.3274, simple_loss=0.3399, pruned_loss=0.1575, over 13539.00 frames. ], tot_loss[loss=0.3457, simple_loss=0.3543, pruned_loss=0.1685, over 2662549.87 frames. ], batch size: 87, lr: 4.21e-02, grad_scale: 16.0 2023-04-16 12:41:59,808 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.107e+02 4.525e+02 5.790e+02 7.364e+02 1.699e+03, threshold=1.158e+03, percent-clipped=5.0 2023-04-16 12:42:10,922 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0377, 1.9399, 1.9532, 1.9201, 1.5543, 1.4117, 1.5845, 1.8476], device='cuda:0'), covar=tensor([0.0209, 0.0420, 0.0606, 0.0185, 0.0160, 0.0302, 0.0589, 0.0272], device='cuda:0'), in_proj_covar=tensor([0.0030, 0.0030, 0.0030, 0.0028, 0.0027, 0.0029, 0.0031, 0.0027], device='cuda:0'), out_proj_covar=tensor([3.1617e-05, 3.4532e-05, 3.2788e-05, 2.6421e-05, 2.6363e-05, 3.5179e-05, 3.2062e-05, 2.7456e-05], device='cuda:0') 2023-04-16 12:42:25,426 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4630.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:42:38,614 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4644.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:42:44,061 INFO [train.py:893] (0/4) Epoch 2, batch 1950, loss[loss=0.3428, simple_loss=0.3562, pruned_loss=0.1647, over 13478.00 frames. ], tot_loss[loss=0.3439, simple_loss=0.3527, pruned_loss=0.1676, over 2664458.62 frames. ], batch size: 93, lr: 4.20e-02, grad_scale: 16.0 2023-04-16 12:43:07,650 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.30 vs. limit=2.0 2023-04-16 12:43:28,628 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5805, 4.0137, 3.9776, 3.4704, 3.5679, 2.6460, 4.1812, 3.1496], device='cuda:0'), covar=tensor([0.0987, 0.0348, 0.0244, 0.0555, 0.0473, 0.2152, 0.0172, 0.2057], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0069, 0.0073, 0.0098, 0.0079, 0.0120, 0.0053, 0.0136], device='cuda:0'), out_proj_covar=tensor([1.3287e-04, 8.1434e-05, 8.2625e-05, 1.1352e-04, 8.7217e-05, 1.3745e-04, 6.4365e-05, 1.6883e-04], device='cuda:0') 2023-04-16 12:43:30,751 INFO [train.py:893] (0/4) Epoch 2, batch 2000, loss[loss=0.4364, simple_loss=0.4272, pruned_loss=0.2228, over 13249.00 frames. ], tot_loss[loss=0.3457, simple_loss=0.3542, pruned_loss=0.1685, over 2663930.52 frames. ], batch size: 124, lr: 4.19e-02, grad_scale: 16.0 2023-04-16 12:43:32,435 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.843e+02 4.530e+02 5.357e+02 6.358e+02 1.307e+03, threshold=1.071e+03, percent-clipped=1.0 2023-04-16 12:43:38,383 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 12:43:56,029 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6166, 3.4134, 3.6342, 3.1587, 4.1144, 3.5037, 3.9315, 3.5714], device='cuda:0'), covar=tensor([0.0160, 0.0166, 0.0202, 0.0151, 0.0134, 0.0160, 0.0225, 0.0115], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0026, 0.0025, 0.0025, 0.0019, 0.0025, 0.0019, 0.0020], device='cuda:0'), out_proj_covar=tensor([4.5000e-05, 5.3500e-05, 4.6728e-05, 4.9402e-05, 3.6948e-05, 4.8631e-05, 3.5051e-05, 3.9471e-05], device='cuda:0') 2023-04-16 12:44:17,032 INFO [train.py:893] (0/4) Epoch 2, batch 2050, loss[loss=0.3427, simple_loss=0.3564, pruned_loss=0.1645, over 13537.00 frames. ], tot_loss[loss=0.348, simple_loss=0.3566, pruned_loss=0.1697, over 2663960.32 frames. ], batch size: 85, lr: 4.17e-02, grad_scale: 16.0 2023-04-16 12:45:01,017 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4797.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:45:03,199 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.10 vs. limit=2.0 2023-04-16 12:45:03,714 INFO [train.py:893] (0/4) Epoch 2, batch 2100, loss[loss=0.3635, simple_loss=0.3746, pruned_loss=0.1763, over 13490.00 frames. ], tot_loss[loss=0.3456, simple_loss=0.3545, pruned_loss=0.1684, over 2656104.16 frames. ], batch size: 93, lr: 4.16e-02, grad_scale: 16.0 2023-04-16 12:45:05,358 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.512e+02 4.812e+02 5.608e+02 7.076e+02 1.162e+03, threshold=1.122e+03, percent-clipped=2.0 2023-04-16 12:45:45,644 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.96 vs. limit=5.0 2023-04-16 12:45:45,936 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4845.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:45:50,056 INFO [train.py:893] (0/4) Epoch 2, batch 2150, loss[loss=0.3242, simple_loss=0.348, pruned_loss=0.1502, over 13461.00 frames. ], tot_loss[loss=0.3421, simple_loss=0.3527, pruned_loss=0.1658, over 2662505.42 frames. ], batch size: 79, lr: 4.15e-02, grad_scale: 16.0 2023-04-16 12:46:24,765 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.05 vs. limit=5.0 2023-04-16 12:46:36,481 INFO [train.py:893] (0/4) Epoch 2, batch 2200, loss[loss=0.3048, simple_loss=0.3226, pruned_loss=0.1435, over 13561.00 frames. ], tot_loss[loss=0.3403, simple_loss=0.3514, pruned_loss=0.1646, over 2659361.94 frames. ], batch size: 76, lr: 4.14e-02, grad_scale: 16.0 2023-04-16 12:46:38,214 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.919e+02 3.986e+02 4.975e+02 7.024e+02 1.362e+03, threshold=9.950e+02, percent-clipped=2.0 2023-04-16 12:47:15,493 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-04-16 12:47:16,970 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4944.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:47:23,264 INFO [train.py:893] (0/4) Epoch 2, batch 2250, loss[loss=0.3233, simple_loss=0.339, pruned_loss=0.1538, over 13462.00 frames. ], tot_loss[loss=0.3373, simple_loss=0.3487, pruned_loss=0.1629, over 2662318.29 frames. ], batch size: 100, lr: 4.13e-02, grad_scale: 16.0 2023-04-16 12:47:36,716 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.3627, 2.0957, 1.6884, 1.6589, 1.2467, 1.6224, 1.4864, 1.9239], device='cuda:0'), covar=tensor([0.0540, 0.0608, 0.1047, 0.0422, 0.0360, 0.0234, 0.0695, 0.0430], device='cuda:0'), in_proj_covar=tensor([0.0045, 0.0039, 0.0039, 0.0044, 0.0036, 0.0039, 0.0055, 0.0045], device='cuda:0'), out_proj_covar=tensor([4.4729e-05, 3.9570e-05, 4.0885e-05, 3.7789e-05, 4.1429e-05, 3.2810e-05, 5.3555e-05, 4.2721e-05], device='cuda:0') 2023-04-16 12:48:01,015 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4992.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 12:48:09,163 INFO [train.py:893] (0/4) Epoch 2, batch 2300, loss[loss=0.3318, simple_loss=0.3369, pruned_loss=0.1633, over 13499.00 frames. ], tot_loss[loss=0.3366, simple_loss=0.3484, pruned_loss=0.1625, over 2663997.62 frames. ], batch size: 70, lr: 4.12e-02, grad_scale: 16.0 2023-04-16 12:48:10,810 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.990e+02 4.129e+02 4.819e+02 6.882e+02 1.653e+03, threshold=9.638e+02, percent-clipped=3.0 2023-04-16 12:48:11,943 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5803, 4.5694, 4.8052, 4.6915, 4.9910, 4.4600, 5.0152, 4.9054], device='cuda:0'), covar=tensor([0.0352, 0.0306, 0.0353, 0.0248, 0.0362, 0.0420, 0.0327, 0.0334], device='cuda:0'), in_proj_covar=tensor([0.0100, 0.0116, 0.0107, 0.0083, 0.0140, 0.0123, 0.0085, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 12:48:40,201 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-16 12:48:55,292 INFO [train.py:893] (0/4) Epoch 2, batch 2350, loss[loss=0.2868, simple_loss=0.3192, pruned_loss=0.1272, over 13471.00 frames. ], tot_loss[loss=0.3357, simple_loss=0.3475, pruned_loss=0.1619, over 2662019.74 frames. ], batch size: 79, lr: 4.11e-02, grad_scale: 16.0 2023-04-16 12:49:20,464 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 12:49:41,167 INFO [train.py:893] (0/4) Epoch 2, batch 2400, loss[loss=0.3669, simple_loss=0.3705, pruned_loss=0.1816, over 13183.00 frames. ], tot_loss[loss=0.3356, simple_loss=0.3474, pruned_loss=0.162, over 2665038.88 frames. ], batch size: 132, lr: 4.10e-02, grad_scale: 16.0 2023-04-16 12:49:43,551 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.916e+02 4.826e+02 5.829e+02 7.201e+02 1.538e+03, threshold=1.166e+03, percent-clipped=7.0 2023-04-16 12:50:27,574 INFO [train.py:893] (0/4) Epoch 2, batch 2450, loss[loss=0.3086, simple_loss=0.3348, pruned_loss=0.1412, over 13504.00 frames. ], tot_loss[loss=0.3362, simple_loss=0.3475, pruned_loss=0.1625, over 2658649.68 frames. ], batch size: 93, lr: 4.09e-02, grad_scale: 16.0 2023-04-16 12:51:13,486 INFO [train.py:893] (0/4) Epoch 2, batch 2500, loss[loss=0.3749, simple_loss=0.3744, pruned_loss=0.1876, over 13264.00 frames. ], tot_loss[loss=0.3357, simple_loss=0.3475, pruned_loss=0.1619, over 2658023.83 frames. ], batch size: 124, lr: 4.08e-02, grad_scale: 16.0 2023-04-16 12:51:15,139 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.258e+02 4.389e+02 5.492e+02 6.426e+02 1.391e+03, threshold=1.098e+03, percent-clipped=1.0 2023-04-16 12:51:59,486 INFO [train.py:893] (0/4) Epoch 2, batch 2550, loss[loss=0.3346, simple_loss=0.3463, pruned_loss=0.1615, over 13474.00 frames. ], tot_loss[loss=0.335, simple_loss=0.347, pruned_loss=0.1615, over 2658210.90 frames. ], batch size: 79, lr: 4.07e-02, grad_scale: 16.0 2023-04-16 12:52:22,239 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.63 vs. limit=5.0 2023-04-16 12:52:24,136 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 12:52:46,265 INFO [train.py:893] (0/4) Epoch 2, batch 2600, loss[loss=0.3438, simple_loss=0.3517, pruned_loss=0.168, over 13532.00 frames. ], tot_loss[loss=0.3359, simple_loss=0.3473, pruned_loss=0.1622, over 2652486.77 frames. ], batch size: 83, lr: 4.06e-02, grad_scale: 16.0 2023-04-16 12:52:47,919 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.769e+02 4.568e+02 5.360e+02 6.678e+02 1.569e+03, threshold=1.072e+03, percent-clipped=3.0 2023-04-16 12:52:55,017 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-16 12:53:23,382 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2061, 4.2975, 4.6366, 4.0388, 4.4916, 2.7079, 4.9027, 3.3929], device='cuda:0'), covar=tensor([0.0697, 0.0313, 0.0171, 0.0460, 0.0205, 0.2040, 0.0102, 0.2188], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0084, 0.0094, 0.0118, 0.0099, 0.0138, 0.0067, 0.0169], device='cuda:0'), out_proj_covar=tensor([1.6721e-04, 1.0251e-04, 1.0989e-04, 1.4017e-04, 1.1624e-04, 1.6346e-04, 8.3052e-05, 2.1951e-04], device='cuda:0') 2023-04-16 12:53:28,393 INFO [train.py:893] (0/4) Epoch 2, batch 2650, loss[loss=0.3245, simple_loss=0.3421, pruned_loss=0.1535, over 13352.00 frames. ], tot_loss[loss=0.3367, simple_loss=0.3476, pruned_loss=0.1629, over 2652760.74 frames. ], batch size: 109, lr: 4.05e-02, grad_scale: 16.0 2023-04-16 12:53:59,707 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.25 vs. limit=5.0 2023-04-16 12:54:07,076 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-2.pt 2023-04-16 12:54:31,641 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 12:54:35,459 INFO [train.py:893] (0/4) Epoch 3, batch 0, loss[loss=0.2726, simple_loss=0.2996, pruned_loss=0.1228, over 13355.00 frames. ], tot_loss[loss=0.2726, simple_loss=0.2996, pruned_loss=0.1228, over 13355.00 frames. ], batch size: 62, lr: 3.84e-02, grad_scale: 16.0 2023-04-16 12:54:35,460 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 12:54:57,766 INFO [train.py:927] (0/4) Epoch 3, validation: loss=0.2472, simple_loss=0.2811, pruned_loss=0.1067, over 2446609.00 frames. 2023-04-16 12:54:57,767 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 12:54:58,034 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7622, 2.1660, 2.3610, 3.3518, 3.1655, 3.3185, 2.4052, 1.8752], device='cuda:0'), covar=tensor([0.0328, 0.0920, 0.0581, 0.0066, 0.0148, 0.0110, 0.0429, 0.1074], device='cuda:0'), in_proj_covar=tensor([0.0046, 0.0081, 0.0066, 0.0042, 0.0036, 0.0041, 0.0059, 0.0078], device='cuda:0'), out_proj_covar=tensor([6.6326e-05, 1.0018e-04, 8.4349e-05, 5.1878e-05, 5.3061e-05, 5.6477e-05, 7.5600e-05, 9.4443e-05], device='cuda:0') 2023-04-16 12:55:00,478 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.744e+02 4.537e+02 5.513e+02 6.512e+02 1.233e+03, threshold=1.103e+03, percent-clipped=2.0 2023-04-16 12:55:44,991 INFO [train.py:893] (0/4) Epoch 3, batch 50, loss[loss=0.3716, simple_loss=0.3699, pruned_loss=0.1867, over 11810.00 frames. ], tot_loss[loss=0.3206, simple_loss=0.3329, pruned_loss=0.1542, over 599639.94 frames. ], batch size: 157, lr: 3.83e-02, grad_scale: 16.0 2023-04-16 12:56:07,362 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2023-04-16 12:56:09,252 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 12:56:09,253 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 12:56:09,253 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 12:56:09,260 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 12:56:09,277 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 12:56:09,299 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 12:56:09,309 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 12:56:31,861 INFO [train.py:893] (0/4) Epoch 3, batch 100, loss[loss=0.3397, simple_loss=0.3474, pruned_loss=0.166, over 13458.00 frames. ], tot_loss[loss=0.3277, simple_loss=0.3385, pruned_loss=0.1585, over 1056597.62 frames. ], batch size: 79, lr: 3.82e-02, grad_scale: 16.0 2023-04-16 12:56:34,459 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.338e+02 4.385e+02 5.122e+02 6.357e+02 1.277e+03, threshold=1.024e+03, percent-clipped=2.0 2023-04-16 12:57:18,801 INFO [train.py:893] (0/4) Epoch 3, batch 150, loss[loss=0.3294, simple_loss=0.3309, pruned_loss=0.1639, over 13380.00 frames. ], tot_loss[loss=0.3344, simple_loss=0.3431, pruned_loss=0.1628, over 1409704.43 frames. ], batch size: 62, lr: 3.81e-02, grad_scale: 16.0 2023-04-16 12:57:37,401 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5478, 4.1115, 3.7860, 4.0748, 3.7440, 4.4881, 4.1281, 4.3486], device='cuda:0'), covar=tensor([0.0379, 0.0236, 0.0197, 0.0365, 0.0254, 0.0192, 0.0210, 0.0242], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0051, 0.0049, 0.0068, 0.0048, 0.0059, 0.0046, 0.0059], device='cuda:0'), out_proj_covar=tensor([1.1219e-04, 1.0921e-04, 1.1116e-04, 1.3766e-04, 1.1210e-04, 1.3063e-04, 9.7607e-05, 1.2499e-04], device='cuda:0') 2023-04-16 12:58:06,408 INFO [train.py:893] (0/4) Epoch 3, batch 200, loss[loss=0.3935, simple_loss=0.3891, pruned_loss=0.1989, over 13528.00 frames. ], tot_loss[loss=0.3357, simple_loss=0.3458, pruned_loss=0.1628, over 1689671.38 frames. ], batch size: 85, lr: 3.80e-02, grad_scale: 16.0 2023-04-16 12:58:09,202 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.881e+02 4.242e+02 5.295e+02 6.523e+02 1.110e+03, threshold=1.059e+03, percent-clipped=3.0 2023-04-16 12:58:52,262 INFO [train.py:893] (0/4) Epoch 3, batch 250, loss[loss=0.3335, simple_loss=0.3491, pruned_loss=0.159, over 13223.00 frames. ], tot_loss[loss=0.3368, simple_loss=0.3468, pruned_loss=0.1634, over 1905723.63 frames. ], batch size: 132, lr: 3.79e-02, grad_scale: 16.0 2023-04-16 12:58:53,503 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5650.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 12:59:04,315 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9691, 3.6272, 3.5923, 3.7734, 1.8774, 3.1003, 3.4104, 1.8475], device='cuda:0'), covar=tensor([0.0080, 0.0459, 0.0226, 0.0140, 0.2014, 0.0529, 0.0534, 0.2990], device='cuda:0'), in_proj_covar=tensor([0.0052, 0.0071, 0.0073, 0.0050, 0.0126, 0.0091, 0.0086, 0.0140], device='cuda:0'), out_proj_covar=tensor([6.2275e-05, 9.4939e-05, 8.9815e-05, 6.5205e-05, 1.4669e-04, 1.0836e-04, 1.1066e-04, 1.5803e-04], device='cuda:0') 2023-04-16 12:59:40,029 INFO [train.py:893] (0/4) Epoch 3, batch 300, loss[loss=0.3406, simple_loss=0.3383, pruned_loss=0.1715, over 13493.00 frames. ], tot_loss[loss=0.3341, simple_loss=0.3453, pruned_loss=0.1614, over 2062192.61 frames. ], batch size: 70, lr: 3.78e-02, grad_scale: 16.0 2023-04-16 12:59:42,577 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.716e+02 4.500e+02 5.480e+02 6.361e+02 1.148e+03, threshold=1.096e+03, percent-clipped=2.0 2023-04-16 12:59:51,154 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5711.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 12:59:53,074 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2023-04-16 12:59:55,992 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3964, 3.1659, 3.5461, 3.6430, 4.0189, 3.4480, 3.6838, 4.0681], device='cuda:0'), covar=tensor([0.0227, 0.0209, 0.0167, 0.0370, 0.0122, 0.0170, 0.0167, 0.0118], device='cuda:0'), in_proj_covar=tensor([0.0040, 0.0023, 0.0035, 0.0049, 0.0038, 0.0034, 0.0037, 0.0029], device='cuda:0'), out_proj_covar=tensor([9.6524e-05, 6.8431e-05, 8.8499e-05, 1.1778e-04, 9.7901e-05, 8.5765e-05, 8.9335e-05, 7.4837e-05], device='cuda:0') 2023-04-16 13:00:11,054 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3068, 4.1639, 4.0630, 3.3648, 3.8768, 2.2547, 4.3931, 3.1239], device='cuda:0'), covar=tensor([0.1015, 0.0247, 0.0233, 0.0664, 0.0349, 0.2453, 0.0131, 0.2111], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0095, 0.0103, 0.0128, 0.0109, 0.0151, 0.0075, 0.0183], device='cuda:0'), out_proj_covar=tensor([1.8596e-04, 1.1671e-04, 1.2262e-04, 1.5237e-04, 1.2862e-04, 1.8112e-04, 9.2839e-05, 2.3972e-04], device='cuda:0') 2023-04-16 13:00:27,089 INFO [train.py:893] (0/4) Epoch 3, batch 350, loss[loss=0.3643, simple_loss=0.3605, pruned_loss=0.184, over 13483.00 frames. ], tot_loss[loss=0.3361, simple_loss=0.3468, pruned_loss=0.1627, over 2199791.47 frames. ], batch size: 93, lr: 3.77e-02, grad_scale: 16.0 2023-04-16 13:00:32,472 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5755.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:00:55,003 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-16 13:01:02,313 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2581, 3.9881, 3.8949, 3.2730, 3.5118, 2.3806, 4.2326, 2.8786], device='cuda:0'), covar=tensor([0.1079, 0.0318, 0.0285, 0.0693, 0.0512, 0.2379, 0.0163, 0.2789], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0095, 0.0105, 0.0130, 0.0110, 0.0153, 0.0076, 0.0184], device='cuda:0'), out_proj_covar=tensor([1.8894e-04, 1.1689e-04, 1.2478e-04, 1.5638e-04, 1.2964e-04, 1.8404e-04, 9.4509e-05, 2.4126e-04], device='cuda:0') 2023-04-16 13:01:13,830 INFO [train.py:893] (0/4) Epoch 3, batch 400, loss[loss=0.2972, simple_loss=0.3326, pruned_loss=0.1309, over 13529.00 frames. ], tot_loss[loss=0.3337, simple_loss=0.3453, pruned_loss=0.161, over 2303328.51 frames. ], batch size: 87, lr: 3.76e-02, grad_scale: 16.0 2023-04-16 13:01:16,661 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.858e+02 4.189e+02 5.044e+02 6.148e+02 1.450e+03, threshold=1.009e+03, percent-clipped=1.0 2023-04-16 13:01:30,397 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.1557, 1.7604, 1.4601, 1.1321, 1.4021, 1.2951, 1.2427, 1.7742], device='cuda:0'), covar=tensor([0.0393, 0.0374, 0.0847, 0.0541, 0.0220, 0.0179, 0.0464, 0.0429], device='cuda:0'), in_proj_covar=tensor([0.0046, 0.0043, 0.0043, 0.0049, 0.0036, 0.0041, 0.0055, 0.0045], device='cuda:0'), out_proj_covar=tensor([4.4719e-05, 4.0995e-05, 4.4424e-05, 4.2574e-05, 4.1273e-05, 3.5180e-05, 5.3801e-05, 4.1382e-05], device='cuda:0') 2023-04-16 13:01:30,412 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5816.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:02:00,368 INFO [train.py:893] (0/4) Epoch 3, batch 450, loss[loss=0.3315, simple_loss=0.3527, pruned_loss=0.1551, over 13528.00 frames. ], tot_loss[loss=0.3354, simple_loss=0.3469, pruned_loss=0.162, over 2381264.72 frames. ], batch size: 83, lr: 3.76e-02, grad_scale: 16.0 2023-04-16 13:02:08,872 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5858.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:02:25,928 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 13:02:46,670 INFO [train.py:893] (0/4) Epoch 3, batch 500, loss[loss=0.3069, simple_loss=0.3291, pruned_loss=0.1423, over 13221.00 frames. ], tot_loss[loss=0.3356, simple_loss=0.3478, pruned_loss=0.1617, over 2443787.36 frames. ], batch size: 132, lr: 3.75e-02, grad_scale: 16.0 2023-04-16 13:02:49,922 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.506e+02 4.151e+02 5.175e+02 6.712e+02 1.423e+03, threshold=1.035e+03, percent-clipped=8.0 2023-04-16 13:03:05,843 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5919.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:03:11,669 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6466, 3.9783, 3.6140, 3.9740, 4.0447, 4.5506, 3.9865, 4.2719], device='cuda:0'), covar=tensor([0.0346, 0.0298, 0.0285, 0.0448, 0.0183, 0.0183, 0.0245, 0.0217], device='cuda:0'), in_proj_covar=tensor([0.0059, 0.0054, 0.0054, 0.0076, 0.0051, 0.0062, 0.0050, 0.0061], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-16 13:03:27,192 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.0207, 4.8085, 5.1627, 5.0150, 5.4107, 4.7584, 5.4150, 5.3554], device='cuda:0'), covar=tensor([0.0295, 0.0448, 0.0460, 0.0369, 0.0532, 0.0566, 0.0416, 0.0428], device='cuda:0'), in_proj_covar=tensor([0.0106, 0.0130, 0.0119, 0.0094, 0.0160, 0.0137, 0.0092, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 13:03:33,424 INFO [train.py:893] (0/4) Epoch 3, batch 550, loss[loss=0.3121, simple_loss=0.3316, pruned_loss=0.1463, over 13477.00 frames. ], tot_loss[loss=0.3307, simple_loss=0.3444, pruned_loss=0.1585, over 2491959.02 frames. ], batch size: 93, lr: 3.74e-02, grad_scale: 16.0 2023-04-16 13:03:41,253 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0973, 2.9242, 2.7541, 2.5280, 3.0062, 2.2872, 3.1559, 1.9208], device='cuda:0'), covar=tensor([0.0184, 0.0347, 0.0234, 0.0271, 0.0314, 0.0279, 0.0353, 0.0777], device='cuda:0'), in_proj_covar=tensor([0.0042, 0.0036, 0.0043, 0.0043, 0.0055, 0.0041, 0.0034, 0.0059], device='cuda:0'), out_proj_covar=tensor([5.6223e-05, 5.3775e-05, 5.4235e-05, 5.4672e-05, 7.5142e-05, 5.2556e-05, 5.0278e-05, 7.8113e-05], device='cuda:0') 2023-04-16 13:04:11,525 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5990.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:04:19,894 INFO [train.py:893] (0/4) Epoch 3, batch 600, loss[loss=0.3377, simple_loss=0.3462, pruned_loss=0.1646, over 13331.00 frames. ], tot_loss[loss=0.331, simple_loss=0.344, pruned_loss=0.1589, over 2529728.48 frames. ], batch size: 118, lr: 3.73e-02, grad_scale: 16.0 2023-04-16 13:04:21,088 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-6000.pt 2023-04-16 13:04:26,351 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.607e+02 4.651e+02 5.653e+02 7.517e+02 1.178e+03, threshold=1.131e+03, percent-clipped=4.0 2023-04-16 13:04:29,950 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6006.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 13:04:54,457 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-04-16 13:04:55,001 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.2608, 1.9228, 2.0282, 1.4849, 2.1175, 2.0436, 2.2334, 1.4862], device='cuda:0'), covar=tensor([0.0291, 0.0954, 0.0307, 0.0473, 0.0559, 0.0314, 0.0677, 0.0893], device='cuda:0'), in_proj_covar=tensor([0.0044, 0.0037, 0.0042, 0.0043, 0.0056, 0.0042, 0.0035, 0.0061], device='cuda:0'), out_proj_covar=tensor([5.8988e-05, 5.5406e-05, 5.3806e-05, 5.5260e-05, 7.6542e-05, 5.4971e-05, 5.2032e-05, 8.0295e-05], device='cuda:0') 2023-04-16 13:05:03,906 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6043.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:05:05,717 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-04-16 13:05:08,424 INFO [train.py:893] (0/4) Epoch 3, batch 650, loss[loss=0.3174, simple_loss=0.3387, pruned_loss=0.148, over 13557.00 frames. ], tot_loss[loss=0.3286, simple_loss=0.3418, pruned_loss=0.1577, over 2558867.49 frames. ], batch size: 78, lr: 3.72e-02, grad_scale: 16.0 2023-04-16 13:05:11,130 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6051.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:05:13,605 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8529, 1.8498, 3.5356, 3.3421, 3.3080, 2.9958, 2.9283, 1.9924], device='cuda:0'), covar=tensor([0.1626, 0.2162, 0.0122, 0.0254, 0.0219, 0.0297, 0.0164, 0.1308], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0138, 0.0065, 0.0061, 0.0063, 0.0071, 0.0058, 0.0100], device='cuda:0'), out_proj_covar=tensor([1.1728e-04, 1.3882e-04, 6.3017e-05, 6.4514e-05, 6.4637e-05, 6.8437e-05, 6.0604e-05, 1.0054e-04], device='cuda:0') 2023-04-16 13:05:24,566 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.11 vs. limit=5.0 2023-04-16 13:05:33,134 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0150, 2.7585, 2.9788, 2.2852, 2.6589, 2.5044, 2.5423, 1.4416], device='cuda:0'), covar=tensor([0.0284, 0.0804, 0.0228, 0.0349, 0.0640, 0.0304, 0.1410, 0.1154], device='cuda:0'), in_proj_covar=tensor([0.0046, 0.0038, 0.0043, 0.0044, 0.0058, 0.0044, 0.0035, 0.0064], device='cuda:0'), out_proj_covar=tensor([6.1621e-05, 5.6549e-05, 5.5530e-05, 5.7054e-05, 8.0291e-05, 5.7613e-05, 5.3073e-05, 8.4267e-05], device='cuda:0') 2023-04-16 13:05:34,786 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3583, 4.0947, 3.3388, 3.9403, 3.9819, 4.3848, 4.1191, 4.1884], device='cuda:0'), covar=tensor([0.0491, 0.0199, 0.0412, 0.0479, 0.0211, 0.0221, 0.0195, 0.0254], device='cuda:0'), in_proj_covar=tensor([0.0059, 0.0054, 0.0053, 0.0077, 0.0052, 0.0065, 0.0050, 0.0062], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 13:05:55,428 INFO [train.py:893] (0/4) Epoch 3, batch 700, loss[loss=0.2878, simple_loss=0.3116, pruned_loss=0.1321, over 13542.00 frames. ], tot_loss[loss=0.3253, simple_loss=0.3393, pruned_loss=0.1557, over 2582126.62 frames. ], batch size: 72, lr: 3.71e-02, grad_scale: 16.0 2023-04-16 13:05:58,628 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.252e+02 4.009e+02 5.106e+02 6.344e+02 2.444e+03, threshold=1.021e+03, percent-clipped=2.0 2023-04-16 13:06:00,689 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6104.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:06:06,326 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6111.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:06:29,425 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6135.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:06:35,930 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2023-04-16 13:06:41,881 INFO [train.py:893] (0/4) Epoch 3, batch 750, loss[loss=0.3154, simple_loss=0.3381, pruned_loss=0.1464, over 13494.00 frames. ], tot_loss[loss=0.3258, simple_loss=0.3398, pruned_loss=0.1559, over 2596931.11 frames. ], batch size: 93, lr: 3.70e-02, grad_scale: 16.0 2023-04-16 13:06:55,532 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0770, 1.9764, 1.5469, 1.4167, 1.5177, 1.5666, 2.1710, 2.1093], device='cuda:0'), covar=tensor([0.0268, 0.0440, 0.0682, 0.0341, 0.0259, 0.0272, 0.0399, 0.0270], device='cuda:0'), in_proj_covar=tensor([0.0031, 0.0030, 0.0034, 0.0027, 0.0032, 0.0029, 0.0031, 0.0032], device='cuda:0'), out_proj_covar=tensor([3.5654e-05, 3.5978e-05, 4.0543e-05, 3.0843e-05, 3.4493e-05, 3.6164e-05, 3.3341e-05, 3.4883e-05], device='cuda:0') 2023-04-16 13:07:02,276 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6170.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:07:26,856 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6196.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:07:29,129 INFO [train.py:893] (0/4) Epoch 3, batch 800, loss[loss=0.3298, simple_loss=0.3376, pruned_loss=0.161, over 13355.00 frames. ], tot_loss[loss=0.3283, simple_loss=0.3418, pruned_loss=0.1574, over 2610028.12 frames. ], batch size: 73, lr: 3.69e-02, grad_scale: 16.0 2023-04-16 13:07:31,840 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.704e+02 5.071e+02 5.835e+02 7.161e+02 1.086e+03, threshold=1.167e+03, percent-clipped=4.0 2023-04-16 13:07:32,963 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6203.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:07:40,332 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6211.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:07:42,854 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6214.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:07:58,995 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6231.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:07:59,850 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5422, 3.3608, 3.6615, 3.4799, 4.1974, 3.7616, 3.7894, 4.0434], device='cuda:0'), covar=tensor([0.0210, 0.0167, 0.0152, 0.0455, 0.0132, 0.0137, 0.0156, 0.0164], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0022, 0.0035, 0.0047, 0.0037, 0.0033, 0.0036, 0.0028], device='cuda:0'), out_proj_covar=tensor([9.9261e-05, 7.0883e-05, 9.5511e-05, 1.2047e-04, 1.0488e-04, 9.0313e-05, 9.3947e-05, 7.7446e-05], device='cuda:0') 2023-04-16 13:08:12,571 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.5306, 1.4926, 1.3180, 1.1857, 1.2121, 0.9078, 1.5418, 1.4367], device='cuda:0'), covar=tensor([0.0208, 0.0324, 0.0224, 0.0320, 0.0280, 0.0287, 0.0362, 0.0295], device='cuda:0'), in_proj_covar=tensor([0.0031, 0.0030, 0.0033, 0.0026, 0.0031, 0.0028, 0.0029, 0.0032], device='cuda:0'), out_proj_covar=tensor([3.5515e-05, 3.5937e-05, 3.9767e-05, 2.9979e-05, 3.3131e-05, 3.4861e-05, 3.2088e-05, 3.4409e-05], device='cuda:0') 2023-04-16 13:08:15,449 INFO [train.py:893] (0/4) Epoch 3, batch 850, loss[loss=0.3229, simple_loss=0.337, pruned_loss=0.1544, over 13574.00 frames. ], tot_loss[loss=0.3292, simple_loss=0.3429, pruned_loss=0.1577, over 2617607.45 frames. ], batch size: 78, lr: 3.68e-02, grad_scale: 16.0 2023-04-16 13:08:29,058 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6264.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:08:30,640 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6266.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:08:35,659 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6272.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:09:00,770 INFO [train.py:893] (0/4) Epoch 3, batch 900, loss[loss=0.3198, simple_loss=0.316, pruned_loss=0.1618, over 12835.00 frames. ], tot_loss[loss=0.3304, simple_loss=0.3437, pruned_loss=0.1585, over 2631151.37 frames. ], batch size: 52, lr: 3.67e-02, grad_scale: 16.0 2023-04-16 13:09:03,272 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.335e+02 4.386e+02 5.429e+02 6.427e+02 2.158e+03, threshold=1.086e+03, percent-clipped=2.0 2023-04-16 13:09:07,395 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6306.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 13:09:26,976 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6327.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:09:30,130 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-16 13:09:31,120 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 13:09:44,913 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6346.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:09:47,207 INFO [train.py:893] (0/4) Epoch 3, batch 950, loss[loss=0.3073, simple_loss=0.3229, pruned_loss=0.1458, over 13178.00 frames. ], tot_loss[loss=0.3298, simple_loss=0.3426, pruned_loss=0.1585, over 2640086.18 frames. ], batch size: 132, lr: 3.66e-02, grad_scale: 16.0 2023-04-16 13:09:52,157 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6354.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 13:10:19,854 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1255, 4.7041, 4.6362, 4.4595, 4.1858, 4.6134, 5.0191, 4.5246], device='cuda:0'), covar=tensor([0.0762, 0.0757, 0.1937, 0.2610, 0.0851, 0.1107, 0.0835, 0.0915], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0170, 0.0234, 0.0237, 0.0125, 0.0198, 0.0212, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 13:10:32,248 INFO [train.py:893] (0/4) Epoch 3, batch 1000, loss[loss=0.3118, simple_loss=0.3244, pruned_loss=0.1495, over 13548.00 frames. ], tot_loss[loss=0.3262, simple_loss=0.3397, pruned_loss=0.1563, over 2648532.65 frames. ], batch size: 89, lr: 3.66e-02, grad_scale: 32.0 2023-04-16 13:10:32,424 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6399.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:10:35,113 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.399e+02 4.240e+02 5.473e+02 6.639e+02 1.475e+03, threshold=1.095e+03, percent-clipped=3.0 2023-04-16 13:10:45,010 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6411.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:11:20,000 INFO [train.py:893] (0/4) Epoch 3, batch 1050, loss[loss=0.3347, simple_loss=0.3512, pruned_loss=0.1591, over 13529.00 frames. ], tot_loss[loss=0.3216, simple_loss=0.3367, pruned_loss=0.1533, over 2651656.75 frames. ], batch size: 98, lr: 3.65e-02, grad_scale: 16.0 2023-04-16 13:11:28,609 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6459.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:11:31,217 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6462.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:11:35,875 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6467.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:11:39,040 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5057, 2.1788, 1.9125, 2.1547, 2.3985, 1.8929, 2.5358, 1.1541], device='cuda:0'), covar=tensor([0.0332, 0.0783, 0.0421, 0.0370, 0.0580, 0.0490, 0.0434, 0.1074], device='cuda:0'), in_proj_covar=tensor([0.0047, 0.0038, 0.0045, 0.0042, 0.0058, 0.0048, 0.0033, 0.0062], device='cuda:0'), out_proj_covar=tensor([6.2977e-05, 5.8576e-05, 5.9506e-05, 5.6022e-05, 8.2354e-05, 6.4801e-05, 5.1004e-05, 8.3524e-05], device='cuda:0') 2023-04-16 13:11:57,138 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.5034, 2.0302, 1.5724, 1.2270, 1.1618, 1.3212, 1.4014, 1.9885], device='cuda:0'), covar=tensor([0.0437, 0.0538, 0.0900, 0.0787, 0.0356, 0.0103, 0.0566, 0.0400], device='cuda:0'), in_proj_covar=tensor([0.0049, 0.0043, 0.0041, 0.0054, 0.0038, 0.0041, 0.0054, 0.0045], device='cuda:0'), out_proj_covar=tensor([4.6639e-05, 3.9671e-05, 4.3134e-05, 4.8574e-05, 4.4695e-05, 3.4490e-05, 5.1762e-05, 4.0598e-05], device='cuda:0') 2023-04-16 13:11:57,829 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6491.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:12:05,643 INFO [train.py:893] (0/4) Epoch 3, batch 1100, loss[loss=0.3476, simple_loss=0.3498, pruned_loss=0.1727, over 11844.00 frames. ], tot_loss[loss=0.3222, simple_loss=0.3378, pruned_loss=0.1533, over 2652903.46 frames. ], batch size: 157, lr: 3.64e-02, grad_scale: 16.0 2023-04-16 13:12:09,690 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.609e+02 4.664e+02 5.670e+02 6.887e+02 1.017e+03, threshold=1.134e+03, percent-clipped=0.0 2023-04-16 13:12:19,723 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6514.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:12:26,416 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3255, 3.4676, 3.6321, 3.4637, 4.0878, 3.5312, 3.6935, 4.0911], device='cuda:0'), covar=tensor([0.0271, 0.0151, 0.0144, 0.0472, 0.0112, 0.0157, 0.0177, 0.0109], device='cuda:0'), in_proj_covar=tensor([0.0040, 0.0022, 0.0035, 0.0047, 0.0038, 0.0034, 0.0037, 0.0029], device='cuda:0'), out_proj_covar=tensor([1.0502e-04, 7.1494e-05, 9.6657e-05, 1.2585e-04, 1.1042e-04, 9.5741e-05, 9.7987e-05, 8.1970e-05], device='cuda:0') 2023-04-16 13:12:27,260 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6523.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 13:12:30,257 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6526.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:12:31,980 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6528.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:12:50,191 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3773, 2.0798, 2.1341, 2.6602, 2.1260, 1.7954, 2.3914, 1.1410], device='cuda:0'), covar=tensor([0.0363, 0.0974, 0.0351, 0.0279, 0.0632, 0.0488, 0.0922, 0.1151], device='cuda:0'), in_proj_covar=tensor([0.0046, 0.0038, 0.0044, 0.0042, 0.0058, 0.0049, 0.0034, 0.0062], device='cuda:0'), out_proj_covar=tensor([6.2389e-05, 5.8644e-05, 5.7847e-05, 5.6093e-05, 8.2393e-05, 6.5771e-05, 5.2954e-05, 8.3693e-05], device='cuda:0') 2023-04-16 13:12:50,280 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3326, 4.0993, 3.9215, 3.3772, 3.7493, 2.4287, 4.2770, 2.8827], device='cuda:0'), covar=tensor([0.0882, 0.0203, 0.0262, 0.0647, 0.0288, 0.2159, 0.0123, 0.2600], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0106, 0.0122, 0.0146, 0.0125, 0.0170, 0.0089, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0003], device='cuda:0') 2023-04-16 13:12:51,517 INFO [train.py:893] (0/4) Epoch 3, batch 1150, loss[loss=0.3066, simple_loss=0.3318, pruned_loss=0.1407, over 13552.00 frames. ], tot_loss[loss=0.3208, simple_loss=0.3372, pruned_loss=0.1522, over 2652788.75 frames. ], batch size: 78, lr: 3.63e-02, grad_scale: 16.0 2023-04-16 13:12:53,722 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.07 vs. limit=2.0 2023-04-16 13:13:00,769 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6559.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:13:04,069 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6562.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:13:08,974 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6567.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:13:38,330 INFO [train.py:893] (0/4) Epoch 3, batch 1200, loss[loss=0.2963, simple_loss=0.3101, pruned_loss=0.1412, over 13386.00 frames. ], tot_loss[loss=0.319, simple_loss=0.3361, pruned_loss=0.151, over 2651436.76 frames. ], batch size: 62, lr: 3.62e-02, grad_scale: 16.0 2023-04-16 13:13:42,314 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.523e+02 4.328e+02 5.221e+02 6.462e+02 1.421e+03, threshold=1.044e+03, percent-clipped=2.0 2023-04-16 13:13:54,386 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.4704, 2.5792, 1.9861, 2.4209, 1.9540, 1.7386, 2.5753, 1.5908], device='cuda:0'), covar=tensor([0.0346, 0.0545, 0.0438, 0.0357, 0.0921, 0.0729, 0.0673, 0.1026], device='cuda:0'), in_proj_covar=tensor([0.0046, 0.0039, 0.0044, 0.0043, 0.0058, 0.0051, 0.0035, 0.0064], device='cuda:0'), out_proj_covar=tensor([6.2865e-05, 5.9884e-05, 5.8329e-05, 5.8056e-05, 8.2655e-05, 6.9049e-05, 5.4012e-05, 8.6192e-05], device='cuda:0') 2023-04-16 13:13:54,426 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0240, 4.1492, 2.7613, 4.2968, 4.0205, 2.2398, 3.4298, 2.4652], device='cuda:0'), covar=tensor([0.0230, 0.0343, 0.1559, 0.0073, 0.0214, 0.1649, 0.0984, 0.2482], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0078, 0.0144, 0.0058, 0.0083, 0.0126, 0.0108, 0.0159], device='cuda:0'), out_proj_covar=tensor([1.2043e-04, 1.2535e-04, 2.0420e-04, 9.5616e-05, 1.2783e-04, 1.7748e-04, 1.6656e-04, 2.1690e-04], device='cuda:0') 2023-04-16 13:13:59,143 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6622.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:14:05,243 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 13:14:06,340 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6629.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:14:16,281 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 13:14:16,564 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5429, 4.0228, 3.8095, 2.7381, 1.8190, 2.7403, 4.0103, 4.0826], device='cuda:0'), covar=tensor([0.0438, 0.0337, 0.0276, 0.1351, 0.2069, 0.0988, 0.0123, 0.0049], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0070, 0.0076, 0.0128, 0.0133, 0.0106, 0.0059, 0.0051], device='cuda:0'), out_proj_covar=tensor([1.1874e-04, 9.8155e-05, 9.9933e-05, 1.6048e-04, 1.5732e-04, 1.3422e-04, 7.8677e-05, 6.8403e-05], device='cuda:0') 2023-04-16 13:14:21,652 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6646.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:14:23,886 INFO [train.py:893] (0/4) Epoch 3, batch 1250, loss[loss=0.3251, simple_loss=0.3413, pruned_loss=0.1544, over 13536.00 frames. ], tot_loss[loss=0.3215, simple_loss=0.3377, pruned_loss=0.1526, over 2651444.76 frames. ], batch size: 85, lr: 3.61e-02, grad_scale: 16.0 2023-04-16 13:14:45,962 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5404, 4.0428, 4.0741, 4.3069, 2.2984, 3.9405, 4.0307, 1.9600], device='cuda:0'), covar=tensor([0.0057, 0.0373, 0.0250, 0.0117, 0.1882, 0.0284, 0.0458, 0.2620], device='cuda:0'), in_proj_covar=tensor([0.0057, 0.0084, 0.0089, 0.0054, 0.0135, 0.0102, 0.0102, 0.0149], device='cuda:0'), out_proj_covar=tensor([7.5630e-05, 1.1745e-04, 1.1258e-04, 7.7068e-05, 1.6431e-04, 1.2788e-04, 1.3607e-04, 1.7919e-04], device='cuda:0') 2023-04-16 13:15:02,889 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6690.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:15:06,837 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6694.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:15:10,892 INFO [train.py:893] (0/4) Epoch 3, batch 1300, loss[loss=0.3263, simple_loss=0.3457, pruned_loss=0.1535, over 13370.00 frames. ], tot_loss[loss=0.3235, simple_loss=0.3391, pruned_loss=0.1539, over 2649398.56 frames. ], batch size: 109, lr: 3.60e-02, grad_scale: 16.0 2023-04-16 13:15:11,147 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6699.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:15:14,244 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.037e+02 4.584e+02 5.620e+02 7.352e+02 1.514e+03, threshold=1.124e+03, percent-clipped=3.0 2023-04-16 13:15:27,654 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7418, 4.0576, 3.7315, 3.6494, 3.7205, 3.5618, 4.0905, 4.1237], device='cuda:0'), covar=tensor([0.0335, 0.0283, 0.0357, 0.0451, 0.0474, 0.0546, 0.0381, 0.0304], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0103, 0.0076, 0.0105, 0.0077, 0.0098, 0.0080, 0.0084], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 13:15:38,210 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9468, 3.9742, 2.4769, 4.1003, 3.7545, 1.9646, 3.3580, 2.5282], device='cuda:0'), covar=tensor([0.0203, 0.0377, 0.1650, 0.0132, 0.0219, 0.1732, 0.0801, 0.2208], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0078, 0.0137, 0.0057, 0.0079, 0.0122, 0.0104, 0.0154], device='cuda:0'), out_proj_covar=tensor([1.1825e-04, 1.2574e-04, 1.9703e-04, 9.4418e-05, 1.2341e-04, 1.7382e-04, 1.6137e-04, 2.1204e-04], device='cuda:0') 2023-04-16 13:15:54,307 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6747.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:15:55,742 INFO [train.py:893] (0/4) Epoch 3, batch 1350, loss[loss=0.3476, simple_loss=0.357, pruned_loss=0.1691, over 13461.00 frames. ], tot_loss[loss=0.325, simple_loss=0.3407, pruned_loss=0.1546, over 2653513.14 frames. ], batch size: 103, lr: 3.59e-02, grad_scale: 16.0 2023-04-16 13:16:34,982 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6791.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:16:36,901 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2023-04-16 13:16:41,386 INFO [train.py:893] (0/4) Epoch 3, batch 1400, loss[loss=0.3106, simple_loss=0.3246, pruned_loss=0.1483, over 13045.00 frames. ], tot_loss[loss=0.3232, simple_loss=0.3392, pruned_loss=0.1535, over 2656547.79 frames. ], batch size: 142, lr: 3.59e-02, grad_scale: 16.0 2023-04-16 13:16:45,013 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.054e+02 4.522e+02 5.361e+02 7.195e+02 2.528e+03, threshold=1.072e+03, percent-clipped=2.0 2023-04-16 13:16:59,960 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6818.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 13:17:04,000 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0276, 4.2751, 2.4132, 4.2214, 4.1304, 2.1640, 3.5483, 2.5759], device='cuda:0'), covar=tensor([0.0234, 0.0295, 0.1610, 0.0066, 0.0205, 0.1595, 0.0688, 0.2367], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0077, 0.0135, 0.0055, 0.0079, 0.0120, 0.0101, 0.0150], device='cuda:0'), out_proj_covar=tensor([1.1655e-04, 1.2619e-04, 1.9453e-04, 9.1599e-05, 1.2405e-04, 1.7119e-04, 1.5827e-04, 2.0831e-04], device='cuda:0') 2023-04-16 13:17:04,637 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6823.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:17:07,175 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6826.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:17:18,838 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6839.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:17:20,792 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6463, 4.1324, 3.8625, 2.9727, 2.0718, 2.8982, 3.9622, 4.2374], device='cuda:0'), covar=tensor([0.0380, 0.0339, 0.0323, 0.1230, 0.1907, 0.0979, 0.0182, 0.0083], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0076, 0.0080, 0.0136, 0.0144, 0.0114, 0.0064, 0.0053], device='cuda:0'), out_proj_covar=tensor([1.3325e-04, 1.0858e-04, 1.0560e-04, 1.7347e-04, 1.7202e-04, 1.4626e-04, 8.6685e-05, 7.2260e-05], device='cuda:0') 2023-04-16 13:17:27,667 INFO [train.py:893] (0/4) Epoch 3, batch 1450, loss[loss=0.3543, simple_loss=0.3587, pruned_loss=0.1749, over 13489.00 frames. ], tot_loss[loss=0.322, simple_loss=0.3382, pruned_loss=0.153, over 2656628.71 frames. ], batch size: 93, lr: 3.58e-02, grad_scale: 16.0 2023-04-16 13:17:37,054 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6859.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:17:43,894 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6867.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:17:50,331 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6874.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:17:53,835 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6878.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 13:17:59,626 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-16 13:18:12,396 INFO [train.py:893] (0/4) Epoch 3, batch 1500, loss[loss=0.3485, simple_loss=0.3594, pruned_loss=0.1688, over 13457.00 frames. ], tot_loss[loss=0.3204, simple_loss=0.3375, pruned_loss=0.1516, over 2659796.15 frames. ], batch size: 106, lr: 3.57e-02, grad_scale: 16.0 2023-04-16 13:18:16,571 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.769e+02 4.038e+02 4.963e+02 5.872e+02 1.075e+03, threshold=9.927e+02, percent-clipped=1.0 2023-04-16 13:18:20,062 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6907.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:18:21,924 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6909.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:18:28,299 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6915.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:18:34,147 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6922.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:18:49,017 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6939.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 13:18:58,210 INFO [train.py:893] (0/4) Epoch 3, batch 1550, loss[loss=0.3165, simple_loss=0.3288, pruned_loss=0.1521, over 13517.00 frames. ], tot_loss[loss=0.3202, simple_loss=0.3379, pruned_loss=0.1512, over 2662867.99 frames. ], batch size: 76, lr: 3.56e-02, grad_scale: 16.0 2023-04-16 13:18:58,554 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.4958, 2.0843, 1.5179, 1.2343, 1.0628, 1.3490, 1.4485, 2.1691], device='cuda:0'), covar=tensor([0.0445, 0.0239, 0.1025, 0.0650, 0.0275, 0.0200, 0.0456, 0.0318], device='cuda:0'), in_proj_covar=tensor([0.0053, 0.0046, 0.0044, 0.0059, 0.0043, 0.0046, 0.0059, 0.0049], device='cuda:0'), out_proj_covar=tensor([5.0581e-05, 4.1619e-05, 4.6358e-05, 5.3039e-05, 5.0590e-05, 3.9055e-05, 5.5209e-05, 4.2563e-05], device='cuda:0') 2023-04-16 13:19:16,937 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6970.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:19:17,102 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6970.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:19:21,359 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2023-04-16 13:19:31,021 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6985.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:19:43,997 INFO [train.py:893] (0/4) Epoch 3, batch 1600, loss[loss=0.3256, simple_loss=0.3428, pruned_loss=0.1542, over 13259.00 frames. ], tot_loss[loss=0.3184, simple_loss=0.3368, pruned_loss=0.15, over 2653612.45 frames. ], batch size: 124, lr: 3.55e-02, grad_scale: 16.0 2023-04-16 13:19:47,561 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.095e+02 4.094e+02 4.857e+02 5.932e+02 1.050e+03, threshold=9.715e+02, percent-clipped=2.0 2023-04-16 13:20:29,970 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7048.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 13:20:30,410 INFO [train.py:893] (0/4) Epoch 3, batch 1650, loss[loss=0.335, simple_loss=0.331, pruned_loss=0.1695, over 13199.00 frames. ], tot_loss[loss=0.3182, simple_loss=0.3371, pruned_loss=0.1496, over 2656688.48 frames. ], batch size: 58, lr: 3.54e-02, grad_scale: 16.0 2023-04-16 13:20:58,960 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6975, 3.9365, 2.2981, 4.1068, 3.7285, 2.0574, 3.2328, 2.4797], device='cuda:0'), covar=tensor([0.0298, 0.0275, 0.1855, 0.0057, 0.0252, 0.1692, 0.1044, 0.2119], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0079, 0.0140, 0.0055, 0.0082, 0.0125, 0.0108, 0.0153], device='cuda:0'), out_proj_covar=tensor([1.2429e-04, 1.3310e-04, 2.0478e-04, 9.6291e-05, 1.3019e-04, 1.8037e-04, 1.7255e-04, 2.1733e-04], device='cuda:0') 2023-04-16 13:21:18,188 INFO [train.py:893] (0/4) Epoch 3, batch 1700, loss[loss=0.267, simple_loss=0.297, pruned_loss=0.1185, over 13379.00 frames. ], tot_loss[loss=0.3168, simple_loss=0.3362, pruned_loss=0.1486, over 2657871.45 frames. ], batch size: 73, lr: 3.54e-02, grad_scale: 16.0 2023-04-16 13:21:21,487 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.842e+02 4.361e+02 5.760e+02 7.858e+02 1.557e+03, threshold=1.152e+03, percent-clipped=12.0 2023-04-16 13:21:26,749 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7109.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 13:21:29,183 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9092, 4.4281, 4.2807, 4.1921, 4.1900, 4.2900, 4.8211, 4.2983], device='cuda:0'), covar=tensor([0.0904, 0.0951, 0.2598, 0.3537, 0.0782, 0.1482, 0.0976, 0.0958], device='cuda:0'), in_proj_covar=tensor([0.0185, 0.0187, 0.0257, 0.0257, 0.0133, 0.0214, 0.0231, 0.0162], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 13:21:34,774 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7118.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 13:21:38,786 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7123.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:22:03,251 INFO [train.py:893] (0/4) Epoch 3, batch 1750, loss[loss=0.2992, simple_loss=0.3297, pruned_loss=0.1343, over 13211.00 frames. ], tot_loss[loss=0.3138, simple_loss=0.3339, pruned_loss=0.1469, over 2659940.74 frames. ], batch size: 117, lr: 3.53e-02, grad_scale: 16.0 2023-04-16 13:22:17,223 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7163.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:22:19,602 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7166.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:22:23,610 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7171.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:22:29,119 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2023-04-16 13:22:49,674 INFO [train.py:893] (0/4) Epoch 3, batch 1800, loss[loss=0.3323, simple_loss=0.3587, pruned_loss=0.153, over 13462.00 frames. ], tot_loss[loss=0.3112, simple_loss=0.3318, pruned_loss=0.1453, over 2660428.81 frames. ], batch size: 103, lr: 3.52e-02, grad_scale: 16.0 2023-04-16 13:22:53,263 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.362e+02 3.709e+02 4.674e+02 5.794e+02 1.455e+03, threshold=9.349e+02, percent-clipped=1.0 2023-04-16 13:23:13,978 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7224.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 13:23:22,104 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7234.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 13:23:30,298 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9548, 3.6456, 3.9236, 3.8283, 4.5947, 3.9303, 4.0152, 4.4521], device='cuda:0'), covar=tensor([0.0155, 0.0098, 0.0119, 0.0374, 0.0071, 0.0120, 0.0115, 0.0079], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0022, 0.0036, 0.0050, 0.0039, 0.0034, 0.0036, 0.0027], device='cuda:0'), out_proj_covar=tensor([1.1117e-04, 7.4104e-05, 1.0597e-04, 1.4403e-04, 1.2829e-04, 1.0307e-04, 1.0432e-04, 8.5844e-05], device='cuda:0') 2023-04-16 13:23:34,801 INFO [train.py:893] (0/4) Epoch 3, batch 1850, loss[loss=0.3084, simple_loss=0.3336, pruned_loss=0.1416, over 13431.00 frames. ], tot_loss[loss=0.3122, simple_loss=0.3323, pruned_loss=0.1461, over 2660147.90 frames. ], batch size: 88, lr: 3.51e-02, grad_scale: 16.0 2023-04-16 13:23:38,092 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 13:23:49,718 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7265.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:23:52,471 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9907, 1.7763, 1.6943, 1.4437, 1.4804, 1.0577, 1.9445, 2.0494], device='cuda:0'), covar=tensor([0.0113, 0.0261, 0.0476, 0.0260, 0.0239, 0.0308, 0.0302, 0.0226], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0030, 0.0031, 0.0026, 0.0031, 0.0029, 0.0030, 0.0032], device='cuda:0'), out_proj_covar=tensor([3.3042e-05, 3.8454e-05, 3.8371e-05, 2.9329e-05, 3.4928e-05, 3.5916e-05, 3.5042e-05, 3.9108e-05], device='cuda:0') 2023-04-16 13:23:58,963 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7275.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:24:09,348 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7285.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:24:15,436 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.07 vs. limit=2.0 2023-04-16 13:24:20,800 INFO [train.py:893] (0/4) Epoch 3, batch 1900, loss[loss=0.2941, simple_loss=0.3234, pruned_loss=0.1324, over 13450.00 frames. ], tot_loss[loss=0.311, simple_loss=0.3312, pruned_loss=0.1454, over 2659927.67 frames. ], batch size: 103, lr: 3.50e-02, grad_scale: 16.0 2023-04-16 13:24:24,138 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.183e+02 4.383e+02 5.266e+02 6.271e+02 1.325e+03, threshold=1.053e+03, percent-clipped=5.0 2023-04-16 13:24:33,015 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-16 13:24:35,181 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0277, 4.0439, 2.5489, 4.1536, 3.8090, 2.2416, 3.5381, 2.6052], device='cuda:0'), covar=tensor([0.0171, 0.0298, 0.1433, 0.0076, 0.0181, 0.1453, 0.0631, 0.1747], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0078, 0.0138, 0.0055, 0.0081, 0.0129, 0.0107, 0.0152], device='cuda:0'), out_proj_covar=tensor([1.2399e-04, 1.3211e-04, 2.0504e-04, 9.6725e-05, 1.3324e-04, 1.8920e-04, 1.7358e-04, 2.2041e-04], device='cuda:0') 2023-04-16 13:24:52,252 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7333.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:24:55,545 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7336.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:25:08,205 INFO [train.py:893] (0/4) Epoch 3, batch 1950, loss[loss=0.3063, simple_loss=0.336, pruned_loss=0.1383, over 13378.00 frames. ], tot_loss[loss=0.3096, simple_loss=0.3301, pruned_loss=0.1446, over 2662644.61 frames. ], batch size: 109, lr: 3.49e-02, grad_scale: 16.0 2023-04-16 13:25:20,340 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6540, 4.1988, 3.8167, 3.9844, 4.1700, 4.5341, 4.1380, 4.1530], device='cuda:0'), covar=tensor([0.0409, 0.0226, 0.0251, 0.0766, 0.0189, 0.0225, 0.0270, 0.0267], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0061, 0.0058, 0.0093, 0.0058, 0.0071, 0.0057, 0.0063], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 13:25:40,634 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6817, 1.8696, 3.5740, 3.3803, 3.3595, 2.9823, 3.1349, 2.2095], device='cuda:0'), covar=tensor([0.2215, 0.2327, 0.0095, 0.0214, 0.0171, 0.0365, 0.0155, 0.1131], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0157, 0.0067, 0.0070, 0.0072, 0.0094, 0.0065, 0.0116], device='cuda:0'), out_proj_covar=tensor([1.5106e-04, 1.6085e-04, 7.0407e-05, 8.1774e-05, 8.2159e-05, 9.8302e-05, 7.3261e-05, 1.1978e-04], device='cuda:0') 2023-04-16 13:25:53,552 INFO [train.py:893] (0/4) Epoch 3, batch 2000, loss[loss=0.3071, simple_loss=0.3286, pruned_loss=0.1428, over 13494.00 frames. ], tot_loss[loss=0.3138, simple_loss=0.3335, pruned_loss=0.147, over 2662125.77 frames. ], batch size: 81, lr: 3.49e-02, grad_scale: 16.0 2023-04-16 13:25:57,274 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.912e+02 4.643e+02 5.664e+02 6.636e+02 1.338e+03, threshold=1.133e+03, percent-clipped=3.0 2023-04-16 13:25:58,201 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 13:25:58,318 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7404.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 13:26:00,821 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7407.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:26:21,135 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5538, 2.6221, 2.0764, 2.3267, 2.1194, 1.2805, 2.4762, 1.1630], device='cuda:0'), covar=tensor([0.0330, 0.0599, 0.0388, 0.0368, 0.0732, 0.0944, 0.1159, 0.1208], device='cuda:0'), in_proj_covar=tensor([0.0054, 0.0045, 0.0052, 0.0049, 0.0066, 0.0063, 0.0042, 0.0070], device='cuda:0'), out_proj_covar=tensor([7.6588e-05, 7.3473e-05, 7.3063e-05, 6.8880e-05, 9.7572e-05, 8.8399e-05, 6.8186e-05, 9.7757e-05], device='cuda:0') 2023-04-16 13:26:40,427 INFO [train.py:893] (0/4) Epoch 3, batch 2050, loss[loss=0.3266, simple_loss=0.3446, pruned_loss=0.1543, over 13376.00 frames. ], tot_loss[loss=0.3159, simple_loss=0.3351, pruned_loss=0.1484, over 2658791.70 frames. ], batch size: 109, lr: 3.48e-02, grad_scale: 16.0 2023-04-16 13:26:57,093 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7468.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:27:24,612 INFO [train.py:893] (0/4) Epoch 3, batch 2100, loss[loss=0.277, simple_loss=0.3076, pruned_loss=0.1232, over 13417.00 frames. ], tot_loss[loss=0.3148, simple_loss=0.3344, pruned_loss=0.1476, over 2658918.95 frames. ], batch size: 65, lr: 3.47e-02, grad_scale: 16.0 2023-04-16 13:27:29,170 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.529e+02 4.232e+02 5.146e+02 5.882e+02 1.040e+03, threshold=1.029e+03, percent-clipped=0.0 2023-04-16 13:27:39,335 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8280, 3.3266, 2.8034, 3.4202, 4.3093, 3.0509, 4.2231, 3.6161], device='cuda:0'), covar=tensor([0.0117, 0.0219, 0.0458, 0.0120, 0.0090, 0.0304, 0.0089, 0.0108], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0032, 0.0037, 0.0031, 0.0021, 0.0033, 0.0022, 0.0026], device='cuda:0'), out_proj_covar=tensor([9.0818e-05, 1.0515e-04, 1.1228e-04, 9.5772e-05, 6.7521e-05, 1.0502e-04, 6.7088e-05, 8.2645e-05], device='cuda:0') 2023-04-16 13:27:43,600 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7519.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 13:27:52,665 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.2914, 1.5666, 1.2103, 1.2109, 1.1954, 1.0854, 1.4893, 1.6732], device='cuda:0'), covar=tensor([0.0595, 0.0472, 0.1210, 0.1037, 0.0188, 0.0173, 0.0649, 0.0346], device='cuda:0'), in_proj_covar=tensor([0.0053, 0.0049, 0.0047, 0.0067, 0.0041, 0.0044, 0.0060, 0.0049], device='cuda:0'), out_proj_covar=tensor([4.9471e-05, 4.3977e-05, 4.8145e-05, 6.0441e-05, 4.7383e-05, 3.7916e-05, 5.5815e-05, 4.2026e-05], device='cuda:0') 2023-04-16 13:27:58,417 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7534.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 13:28:07,035 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-16 13:28:10,586 INFO [train.py:893] (0/4) Epoch 3, batch 2150, loss[loss=0.3336, simple_loss=0.3604, pruned_loss=0.1534, over 13467.00 frames. ], tot_loss[loss=0.3157, simple_loss=0.3356, pruned_loss=0.1479, over 2658664.39 frames. ], batch size: 103, lr: 3.46e-02, grad_scale: 16.0 2023-04-16 13:28:26,869 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7565.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:28:33,464 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9719, 3.4940, 3.5695, 3.9743, 1.9575, 3.2346, 3.3651, 1.6721], device='cuda:0'), covar=tensor([0.0096, 0.0442, 0.0333, 0.0134, 0.1900, 0.0456, 0.0801, 0.3078], device='cuda:0'), in_proj_covar=tensor([0.0059, 0.0091, 0.0102, 0.0059, 0.0143, 0.0112, 0.0110, 0.0150], device='cuda:0'), out_proj_covar=tensor([8.0497e-05, 1.3129e-04, 1.3466e-04, 8.8215e-05, 1.7884e-04, 1.4721e-04, 1.5118e-04, 1.8963e-04], device='cuda:0') 2023-04-16 13:28:40,561 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7582.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 13:28:40,937 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2023-04-16 13:28:45,574 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.71 vs. limit=5.0 2023-04-16 13:28:57,096 INFO [train.py:893] (0/4) Epoch 3, batch 2200, loss[loss=0.3194, simple_loss=0.3408, pruned_loss=0.149, over 13442.00 frames. ], tot_loss[loss=0.3138, simple_loss=0.3339, pruned_loss=0.1468, over 2657927.08 frames. ], batch size: 103, lr: 3.45e-02, grad_scale: 16.0 2023-04-16 13:29:00,654 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.779e+02 4.630e+02 5.506e+02 6.810e+02 1.281e+03, threshold=1.101e+03, percent-clipped=2.0 2023-04-16 13:29:02,877 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.27 vs. limit=5.0 2023-04-16 13:29:04,360 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7019, 1.7424, 1.4922, 1.5339, 0.9101, 1.0261, 1.6113, 1.8492], device='cuda:0'), covar=tensor([0.0225, 0.0281, 0.0281, 0.0287, 0.0204, 0.0256, 0.0468, 0.0281], device='cuda:0'), in_proj_covar=tensor([0.0031, 0.0031, 0.0033, 0.0027, 0.0030, 0.0029, 0.0031, 0.0033], device='cuda:0'), out_proj_covar=tensor([3.4899e-05, 3.9156e-05, 4.1201e-05, 3.1050e-05, 3.4444e-05, 3.6811e-05, 3.7259e-05, 3.9812e-05], device='cuda:0') 2023-04-16 13:29:07,972 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-16 13:29:09,247 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7613.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:29:19,194 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1159, 2.6611, 2.1642, 2.6222, 2.3882, 1.6669, 2.7303, 1.6831], device='cuda:0'), covar=tensor([0.0273, 0.0724, 0.0401, 0.0279, 0.0755, 0.0694, 0.0739, 0.0940], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0045, 0.0054, 0.0049, 0.0068, 0.0064, 0.0041, 0.0069], device='cuda:0'), out_proj_covar=tensor([7.9026e-05, 7.4286e-05, 7.6646e-05, 7.0685e-05, 1.0095e-04, 9.0754e-05, 6.7313e-05, 9.7780e-05], device='cuda:0') 2023-04-16 13:29:19,199 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7624.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:29:26,925 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7631.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:29:38,705 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0591, 1.9296, 1.6496, 1.3515, 1.3889, 1.0390, 1.8258, 2.3603], device='cuda:0'), covar=tensor([0.0212, 0.0300, 0.0365, 0.0335, 0.0144, 0.0238, 0.0460, 0.0194], device='cuda:0'), in_proj_covar=tensor([0.0030, 0.0030, 0.0033, 0.0026, 0.0030, 0.0028, 0.0030, 0.0032], device='cuda:0'), out_proj_covar=tensor([3.3540e-05, 3.8433e-05, 4.0397e-05, 3.0499e-05, 3.4147e-05, 3.5708e-05, 3.5304e-05, 3.8789e-05], device='cuda:0') 2023-04-16 13:29:40,440 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4275, 4.1460, 3.9724, 3.6411, 3.5187, 2.5279, 4.3235, 3.0539], device='cuda:0'), covar=tensor([0.0686, 0.0182, 0.0225, 0.0472, 0.0402, 0.1838, 0.0129, 0.1957], device='cuda:0'), in_proj_covar=tensor([0.0187, 0.0130, 0.0142, 0.0174, 0.0141, 0.0184, 0.0102, 0.0233], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0003], device='cuda:0') 2023-04-16 13:29:42,337 INFO [train.py:893] (0/4) Epoch 3, batch 2250, loss[loss=0.2775, simple_loss=0.312, pruned_loss=0.1215, over 13568.00 frames. ], tot_loss[loss=0.3106, simple_loss=0.3313, pruned_loss=0.145, over 2655448.94 frames. ], batch size: 78, lr: 3.45e-02, grad_scale: 16.0 2023-04-16 13:30:15,340 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7685.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:30:28,783 INFO [train.py:893] (0/4) Epoch 3, batch 2300, loss[loss=0.3222, simple_loss=0.3356, pruned_loss=0.1544, over 13541.00 frames. ], tot_loss[loss=0.3094, simple_loss=0.3306, pruned_loss=0.1441, over 2655297.49 frames. ], batch size: 87, lr: 3.44e-02, grad_scale: 16.0 2023-04-16 13:30:32,068 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.377e+02 3.990e+02 4.633e+02 6.005e+02 9.986e+02, threshold=9.265e+02, percent-clipped=0.0 2023-04-16 13:30:33,087 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7704.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 13:30:42,977 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8664, 4.8708, 5.1470, 4.9009, 5.3149, 4.8189, 5.3069, 5.2785], device='cuda:0'), covar=tensor([0.0357, 0.0395, 0.0488, 0.0463, 0.0529, 0.0621, 0.0352, 0.0419], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0149, 0.0137, 0.0104, 0.0188, 0.0162, 0.0112, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 13:31:09,226 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7743.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:31:10,421 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-04-16 13:31:10,924 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0693, 4.2066, 2.5226, 4.1240, 4.0595, 2.3697, 3.5093, 2.4307], device='cuda:0'), covar=tensor([0.0180, 0.0241, 0.1806, 0.0101, 0.0275, 0.1530, 0.0811, 0.2630], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0083, 0.0147, 0.0060, 0.0085, 0.0133, 0.0112, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 13:31:13,939 INFO [train.py:893] (0/4) Epoch 3, batch 2350, loss[loss=0.3517, simple_loss=0.363, pruned_loss=0.1701, over 13550.00 frames. ], tot_loss[loss=0.3072, simple_loss=0.3292, pruned_loss=0.1426, over 2661892.99 frames. ], batch size: 87, lr: 3.43e-02, grad_scale: 16.0 2023-04-16 13:31:17,274 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7752.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 13:31:27,670 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7763.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:31:34,187 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 13:31:49,987 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7494, 4.3895, 4.4962, 3.2757, 2.5868, 3.2679, 4.5933, 4.6831], device='cuda:0'), covar=tensor([0.0505, 0.0363, 0.0296, 0.1137, 0.1647, 0.0742, 0.0112, 0.0110], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0082, 0.0093, 0.0146, 0.0153, 0.0123, 0.0069, 0.0061], device='cuda:0'), out_proj_covar=tensor([1.5279e-04, 1.2120e-04, 1.2657e-04, 1.9204e-04, 1.9078e-04, 1.6548e-04, 9.4405e-05, 8.5602e-05], device='cuda:0') 2023-04-16 13:31:54,897 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-16 13:32:00,158 INFO [train.py:893] (0/4) Epoch 3, batch 2400, loss[loss=0.3438, simple_loss=0.3503, pruned_loss=0.1687, over 13539.00 frames. ], tot_loss[loss=0.307, simple_loss=0.3288, pruned_loss=0.1426, over 2662186.12 frames. ], batch size: 78, lr: 3.42e-02, grad_scale: 16.0 2023-04-16 13:32:03,326 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7802.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:32:03,857 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.317e+02 4.190e+02 5.036e+02 6.396e+02 1.178e+03, threshold=1.007e+03, percent-clipped=5.0 2023-04-16 13:32:05,034 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7804.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:32:10,250 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.07 vs. limit=5.0 2023-04-16 13:32:15,877 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-04-16 13:32:20,263 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7819.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 13:32:46,645 INFO [train.py:893] (0/4) Epoch 3, batch 2450, loss[loss=0.3191, simple_loss=0.3424, pruned_loss=0.1479, over 13454.00 frames. ], tot_loss[loss=0.3084, simple_loss=0.3296, pruned_loss=0.1436, over 2662567.33 frames. ], batch size: 100, lr: 3.41e-02, grad_scale: 16.0 2023-04-16 13:32:59,402 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7863.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:33:02,562 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7867.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:33:02,860 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-16 13:33:31,673 INFO [train.py:893] (0/4) Epoch 3, batch 2500, loss[loss=0.2726, simple_loss=0.2884, pruned_loss=0.1284, over 12391.00 frames. ], tot_loss[loss=0.3071, simple_loss=0.3285, pruned_loss=0.1429, over 2654073.54 frames. ], batch size: 50, lr: 3.41e-02, grad_scale: 16.0 2023-04-16 13:33:35,660 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.517e+02 4.233e+02 5.484e+02 6.792e+02 1.086e+03, threshold=1.097e+03, percent-clipped=1.0 2023-04-16 13:34:00,682 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7931.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:34:06,627 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-16 13:34:17,472 INFO [train.py:893] (0/4) Epoch 3, batch 2550, loss[loss=0.3163, simple_loss=0.3387, pruned_loss=0.147, over 13535.00 frames. ], tot_loss[loss=0.3074, simple_loss=0.3288, pruned_loss=0.143, over 2656378.19 frames. ], batch size: 85, lr: 3.40e-02, grad_scale: 16.0 2023-04-16 13:34:38,367 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 13:34:44,033 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2023-04-16 13:34:44,991 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7979.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:34:45,838 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7980.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:34:46,614 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8255, 4.2763, 4.2872, 4.1117, 3.8091, 4.1460, 4.7014, 4.1462], device='cuda:0'), covar=tensor([0.0760, 0.0906, 0.2383, 0.3146, 0.0925, 0.1423, 0.1035, 0.0986], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0182, 0.0252, 0.0253, 0.0126, 0.0207, 0.0227, 0.0162], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 13:35:01,884 INFO [train.py:893] (0/4) Epoch 3, batch 2600, loss[loss=0.284, simple_loss=0.3106, pruned_loss=0.1287, over 13541.00 frames. ], tot_loss[loss=0.3077, simple_loss=0.3286, pruned_loss=0.1434, over 2654902.17 frames. ], batch size: 87, lr: 3.39e-02, grad_scale: 16.0 2023-04-16 13:35:03,089 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-8000.pt 2023-04-16 13:35:09,081 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.046e+02 4.385e+02 5.353e+02 6.768e+02 2.035e+03, threshold=1.071e+03, percent-clipped=3.0 2023-04-16 13:35:46,771 INFO [train.py:893] (0/4) Epoch 3, batch 2650, loss[loss=0.3129, simple_loss=0.3348, pruned_loss=0.1455, over 13475.00 frames. ], tot_loss[loss=0.3092, simple_loss=0.3294, pruned_loss=0.1445, over 2657418.44 frames. ], batch size: 79, lr: 3.38e-02, grad_scale: 16.0 2023-04-16 13:35:57,822 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8063.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:36:24,664 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-3.pt 2023-04-16 13:36:49,430 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 13:36:53,366 INFO [train.py:893] (0/4) Epoch 4, batch 0, loss[loss=0.3152, simple_loss=0.3328, pruned_loss=0.1488, over 13421.00 frames. ], tot_loss[loss=0.3152, simple_loss=0.3328, pruned_loss=0.1488, over 13421.00 frames. ], batch size: 106, lr: 3.16e-02, grad_scale: 16.0 2023-04-16 13:36:53,366 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 13:37:13,674 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6307, 1.4661, 1.4102, 1.3134, 1.2035, 1.3314, 1.6193, 1.9216], device='cuda:0'), covar=tensor([0.0242, 0.0374, 0.0323, 0.0397, 0.0182, 0.0258, 0.0503, 0.0308], device='cuda:0'), in_proj_covar=tensor([0.0033, 0.0029, 0.0035, 0.0029, 0.0032, 0.0029, 0.0033, 0.0036], device='cuda:0'), out_proj_covar=tensor([3.8529e-05, 3.7272e-05, 4.3199e-05, 3.3824e-05, 3.6622e-05, 3.5343e-05, 3.9286e-05, 4.3801e-05], device='cuda:0') 2023-04-16 13:37:16,200 INFO [train.py:927] (0/4) Epoch 4, validation: loss=0.2213, simple_loss=0.2628, pruned_loss=0.08994, over 2446609.00 frames. 2023-04-16 13:37:16,201 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 13:37:17,255 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8099.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:37:20,320 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.901e+02 4.145e+02 5.208e+02 6.723e+02 1.537e+03, threshold=1.042e+03, percent-clipped=4.0 2023-04-16 13:37:27,225 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8111.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:38:02,420 INFO [train.py:893] (0/4) Epoch 4, batch 50, loss[loss=0.3065, simple_loss=0.3347, pruned_loss=0.1391, over 13528.00 frames. ], tot_loss[loss=0.3014, simple_loss=0.3211, pruned_loss=0.1408, over 599670.80 frames. ], batch size: 98, lr: 3.15e-02, grad_scale: 16.0 2023-04-16 13:38:12,399 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8158.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:38:26,305 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 13:38:26,306 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 13:38:26,306 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 13:38:26,321 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 13:38:27,065 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 13:38:27,080 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 13:38:27,099 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 13:38:37,131 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6447, 2.1696, 1.7373, 1.3937, 1.3862, 1.9641, 1.8253, 2.0547], device='cuda:0'), covar=tensor([0.0446, 0.0406, 0.1058, 0.1157, 0.0356, 0.0406, 0.0550, 0.0303], device='cuda:0'), in_proj_covar=tensor([0.0059, 0.0050, 0.0053, 0.0075, 0.0046, 0.0051, 0.0065, 0.0050], device='cuda:0'), out_proj_covar=tensor([5.2925e-05, 4.2489e-05, 5.2481e-05, 6.8702e-05, 5.1400e-05, 4.4459e-05, 6.0154e-05, 4.0812e-05], device='cuda:0') 2023-04-16 13:38:48,446 INFO [train.py:893] (0/4) Epoch 4, batch 100, loss[loss=0.3378, simple_loss=0.3493, pruned_loss=0.1631, over 13539.00 frames. ], tot_loss[loss=0.3052, simple_loss=0.324, pruned_loss=0.1432, over 1058783.31 frames. ], batch size: 87, lr: 3.14e-02, grad_scale: 16.0 2023-04-16 13:38:48,828 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8123, 3.3080, 3.3354, 3.7060, 2.0703, 2.9551, 3.1712, 1.8196], device='cuda:0'), covar=tensor([0.0112, 0.0386, 0.0391, 0.0167, 0.1691, 0.0536, 0.0661, 0.2557], device='cuda:0'), in_proj_covar=tensor([0.0064, 0.0103, 0.0113, 0.0064, 0.0151, 0.0120, 0.0119, 0.0156], device='cuda:0'), out_proj_covar=tensor([9.0686e-05, 1.4822e-04, 1.5390e-04, 9.6740e-05, 1.9375e-04, 1.6232e-04, 1.6660e-04, 2.0201e-04], device='cuda:0') 2023-04-16 13:38:53,074 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.887e+02 4.151e+02 4.936e+02 5.923e+02 1.224e+03, threshold=9.871e+02, percent-clipped=1.0 2023-04-16 13:39:30,973 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.12 vs. limit=2.0 2023-04-16 13:39:35,388 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6226, 3.9709, 3.6794, 3.9900, 4.0359, 4.5528, 4.0558, 4.2085], device='cuda:0'), covar=tensor([0.0442, 0.0311, 0.0325, 0.0928, 0.0245, 0.0222, 0.0276, 0.0287], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0068, 0.0065, 0.0106, 0.0064, 0.0079, 0.0061, 0.0068], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 13:39:35,966 INFO [train.py:893] (0/4) Epoch 4, batch 150, loss[loss=0.3001, simple_loss=0.3227, pruned_loss=0.1388, over 13553.00 frames. ], tot_loss[loss=0.3086, simple_loss=0.3264, pruned_loss=0.1454, over 1406354.48 frames. ], batch size: 85, lr: 3.14e-02, grad_scale: 8.0 2023-04-16 13:39:37,145 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.1602, 1.7854, 3.8758, 3.6135, 3.9016, 3.4176, 3.7412, 2.8110], device='cuda:0'), covar=tensor([0.1835, 0.2014, 0.0071, 0.0264, 0.0121, 0.0267, 0.0147, 0.0801], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0150, 0.0067, 0.0075, 0.0073, 0.0098, 0.0068, 0.0115], device='cuda:0'), out_proj_covar=tensor([1.5227e-04, 1.5733e-04, 7.1938e-05, 8.9454e-05, 8.4280e-05, 1.0490e-04, 7.9170e-05, 1.2264e-04], device='cuda:0') 2023-04-16 13:39:51,424 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8265.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:40:06,050 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8280.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:40:21,456 INFO [train.py:893] (0/4) Epoch 4, batch 200, loss[loss=0.3074, simple_loss=0.3277, pruned_loss=0.1435, over 13521.00 frames. ], tot_loss[loss=0.3103, simple_loss=0.3286, pruned_loss=0.146, over 1685656.88 frames. ], batch size: 72, lr: 3.13e-02, grad_scale: 8.0 2023-04-16 13:40:26,450 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.066e+02 4.274e+02 5.241e+02 6.308e+02 1.197e+03, threshold=1.048e+03, percent-clipped=2.0 2023-04-16 13:40:47,730 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8326.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:40:49,217 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8328.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:40:53,506 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4491, 4.0986, 3.9677, 2.8582, 2.1303, 2.9326, 4.1601, 4.3256], device='cuda:0'), covar=tensor([0.0518, 0.0267, 0.0247, 0.1208, 0.1673, 0.0796, 0.0112, 0.0072], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0089, 0.0096, 0.0153, 0.0158, 0.0131, 0.0074, 0.0064], device='cuda:0'), out_proj_covar=tensor([1.7256e-04, 1.3231e-04, 1.3370e-04, 2.0338e-04, 2.0014e-04, 1.7818e-04, 1.0440e-04, 9.1478e-05], device='cuda:0') 2023-04-16 13:41:08,871 INFO [train.py:893] (0/4) Epoch 4, batch 250, loss[loss=0.3217, simple_loss=0.3346, pruned_loss=0.1544, over 13575.00 frames. ], tot_loss[loss=0.3105, simple_loss=0.3289, pruned_loss=0.146, over 1896605.36 frames. ], batch size: 89, lr: 3.12e-02, grad_scale: 8.0 2023-04-16 13:41:19,532 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2023-04-16 13:41:27,993 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2023-04-16 13:41:54,042 INFO [train.py:893] (0/4) Epoch 4, batch 300, loss[loss=0.2631, simple_loss=0.2923, pruned_loss=0.1169, over 13360.00 frames. ], tot_loss[loss=0.3096, simple_loss=0.3287, pruned_loss=0.1453, over 2059221.94 frames. ], batch size: 67, lr: 3.12e-02, grad_scale: 8.0 2023-04-16 13:41:55,794 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8399.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:42:01,462 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.173e+02 4.573e+02 5.288e+02 6.511e+02 1.234e+03, threshold=1.058e+03, percent-clipped=2.0 2023-04-16 13:42:41,070 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8447.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:42:41,756 INFO [train.py:893] (0/4) Epoch 4, batch 350, loss[loss=0.244, simple_loss=0.2731, pruned_loss=0.1074, over 13364.00 frames. ], tot_loss[loss=0.3106, simple_loss=0.3297, pruned_loss=0.1458, over 2192209.09 frames. ], batch size: 67, lr: 3.11e-02, grad_scale: 8.0 2023-04-16 13:42:50,422 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8458.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:43:03,599 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3139, 3.7757, 3.7742, 4.3606, 1.8424, 3.1980, 3.7275, 1.9042], device='cuda:0'), covar=tensor([0.0073, 0.0502, 0.0403, 0.0092, 0.2176, 0.0638, 0.0580, 0.2880], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0105, 0.0117, 0.0064, 0.0154, 0.0123, 0.0123, 0.0159], device='cuda:0'), out_proj_covar=tensor([9.2801e-05, 1.5277e-04, 1.6148e-04, 9.8159e-05, 1.9886e-04, 1.6650e-04, 1.7216e-04, 2.0741e-04], device='cuda:0') 2023-04-16 13:43:06,090 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2023-04-16 13:43:27,838 INFO [train.py:893] (0/4) Epoch 4, batch 400, loss[loss=0.2878, simple_loss=0.3226, pruned_loss=0.1265, over 13495.00 frames. ], tot_loss[loss=0.3096, simple_loss=0.3298, pruned_loss=0.1446, over 2298691.16 frames. ], batch size: 81, lr: 3.10e-02, grad_scale: 8.0 2023-04-16 13:43:32,889 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.623e+02 4.353e+02 5.245e+02 6.268e+02 1.002e+03, threshold=1.049e+03, percent-clipped=0.0 2023-04-16 13:43:35,404 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8506.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:44:12,225 INFO [train.py:893] (0/4) Epoch 4, batch 450, loss[loss=0.3063, simple_loss=0.3269, pruned_loss=0.1429, over 13553.00 frames. ], tot_loss[loss=0.3108, simple_loss=0.3311, pruned_loss=0.1453, over 2378801.86 frames. ], batch size: 87, lr: 3.10e-02, grad_scale: 8.0 2023-04-16 13:44:37,101 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 13:44:50,045 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.99 vs. limit=5.0 2023-04-16 13:44:58,355 INFO [train.py:893] (0/4) Epoch 4, batch 500, loss[loss=0.3035, simple_loss=0.3327, pruned_loss=0.1372, over 13387.00 frames. ], tot_loss[loss=0.3086, simple_loss=0.3302, pruned_loss=0.1435, over 2447253.78 frames. ], batch size: 77, lr: 3.09e-02, grad_scale: 8.0 2023-04-16 13:45:04,383 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.634e+02 4.142e+02 5.081e+02 6.274e+02 1.184e+03, threshold=1.016e+03, percent-clipped=3.0 2023-04-16 13:45:04,819 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7933, 4.4592, 4.2646, 3.4253, 2.5322, 3.2808, 4.5615, 4.5008], device='cuda:0'), covar=tensor([0.0460, 0.0257, 0.0314, 0.0969, 0.1466, 0.0840, 0.0095, 0.0124], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0091, 0.0100, 0.0157, 0.0159, 0.0131, 0.0075, 0.0063], device='cuda:0'), out_proj_covar=tensor([1.7618e-04, 1.3590e-04, 1.3826e-04, 2.1011e-04, 2.0367e-04, 1.8108e-04, 1.0684e-04, 9.1823e-05], device='cuda:0') 2023-04-16 13:45:20,080 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8621.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:45:43,361 INFO [train.py:893] (0/4) Epoch 4, batch 550, loss[loss=0.444, simple_loss=0.4159, pruned_loss=0.236, over 11837.00 frames. ], tot_loss[loss=0.3067, simple_loss=0.3288, pruned_loss=0.1423, over 2495634.64 frames. ], batch size: 157, lr: 3.08e-02, grad_scale: 8.0 2023-04-16 13:45:44,540 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7038, 2.6987, 2.3164, 2.4835, 2.4343, 1.3270, 2.8381, 1.4641], device='cuda:0'), covar=tensor([0.0369, 0.0661, 0.0350, 0.0490, 0.0739, 0.1174, 0.0758, 0.1477], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0050, 0.0061, 0.0057, 0.0080, 0.0083, 0.0049, 0.0082], device='cuda:0'), out_proj_covar=tensor([9.8792e-05, 8.2661e-05, 8.9369e-05, 8.4675e-05, 1.2398e-04, 1.2000e-04, 8.3429e-05, 1.2003e-04], device='cuda:0') 2023-04-16 13:46:29,272 INFO [train.py:893] (0/4) Epoch 4, batch 600, loss[loss=0.3053, simple_loss=0.3293, pruned_loss=0.1407, over 13530.00 frames. ], tot_loss[loss=0.3065, simple_loss=0.3284, pruned_loss=0.1423, over 2532241.20 frames. ], batch size: 85, lr: 3.08e-02, grad_scale: 8.0 2023-04-16 13:46:34,198 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.491e+02 4.071e+02 5.189e+02 6.724e+02 1.349e+03, threshold=1.038e+03, percent-clipped=5.0 2023-04-16 13:47:14,842 INFO [train.py:893] (0/4) Epoch 4, batch 650, loss[loss=0.3323, simple_loss=0.3479, pruned_loss=0.1584, over 13457.00 frames. ], tot_loss[loss=0.3043, simple_loss=0.3267, pruned_loss=0.1409, over 2562235.97 frames. ], batch size: 106, lr: 3.07e-02, grad_scale: 8.0 2023-04-16 13:47:25,873 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8760.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:47:27,591 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-04-16 13:48:00,254 INFO [train.py:893] (0/4) Epoch 4, batch 700, loss[loss=0.3073, simple_loss=0.3254, pruned_loss=0.1446, over 13262.00 frames. ], tot_loss[loss=0.3007, simple_loss=0.3238, pruned_loss=0.1388, over 2580255.55 frames. ], batch size: 124, lr: 3.06e-02, grad_scale: 8.0 2023-04-16 13:48:05,596 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.391e+02 3.986e+02 4.492e+02 5.620e+02 1.146e+03, threshold=8.983e+02, percent-clipped=1.0 2023-04-16 13:48:21,780 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8821.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:48:21,818 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2620, 3.5949, 3.3945, 4.1540, 1.8243, 3.1277, 3.7378, 1.9457], device='cuda:0'), covar=tensor([0.0059, 0.0360, 0.0445, 0.0108, 0.1750, 0.0517, 0.0399, 0.2249], device='cuda:0'), in_proj_covar=tensor([0.0061, 0.0104, 0.0116, 0.0066, 0.0150, 0.0122, 0.0116, 0.0152], device='cuda:0'), out_proj_covar=tensor([8.8446e-05, 1.5266e-04, 1.6017e-04, 1.0114e-04, 1.9654e-04, 1.6687e-04, 1.6402e-04, 2.0182e-04], device='cuda:0') 2023-04-16 13:48:25,167 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.63 vs. limit=5.0 2023-04-16 13:48:46,629 INFO [train.py:893] (0/4) Epoch 4, batch 750, loss[loss=0.3016, simple_loss=0.3188, pruned_loss=0.1422, over 13455.00 frames. ], tot_loss[loss=0.3011, simple_loss=0.3237, pruned_loss=0.1392, over 2590245.01 frames. ], batch size: 65, lr: 3.06e-02, grad_scale: 8.0 2023-04-16 13:49:30,922 INFO [train.py:893] (0/4) Epoch 4, batch 800, loss[loss=0.3052, simple_loss=0.3353, pruned_loss=0.1376, over 13530.00 frames. ], tot_loss[loss=0.3026, simple_loss=0.325, pruned_loss=0.1401, over 2599386.96 frames. ], batch size: 91, lr: 3.05e-02, grad_scale: 8.0 2023-04-16 13:49:36,354 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.977e+02 4.254e+02 5.779e+02 7.290e+02 1.699e+03, threshold=1.156e+03, percent-clipped=10.0 2023-04-16 13:49:52,819 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8921.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:49:58,001 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8927.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:50:17,241 INFO [train.py:893] (0/4) Epoch 4, batch 850, loss[loss=0.2981, simple_loss=0.3264, pruned_loss=0.1349, over 13455.00 frames. ], tot_loss[loss=0.3023, simple_loss=0.3256, pruned_loss=0.1395, over 2614891.94 frames. ], batch size: 103, lr: 3.04e-02, grad_scale: 8.0 2023-04-16 13:50:37,336 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8969.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:50:39,285 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-16 13:50:52,189 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8986.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:50:53,966 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8988.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:51:03,291 INFO [train.py:893] (0/4) Epoch 4, batch 900, loss[loss=0.3176, simple_loss=0.3325, pruned_loss=0.1514, over 13528.00 frames. ], tot_loss[loss=0.3029, simple_loss=0.3255, pruned_loss=0.1402, over 2624294.43 frames. ], batch size: 85, lr: 3.04e-02, grad_scale: 8.0 2023-04-16 13:51:09,133 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.569e+02 4.214e+02 4.860e+02 6.080e+02 1.245e+03, threshold=9.720e+02, percent-clipped=1.0 2023-04-16 13:51:16,777 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9013.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:51:22,004 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2023-04-16 13:51:34,928 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 13:51:36,081 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6854, 2.7484, 2.6193, 2.6789, 2.8159, 1.4789, 2.9145, 1.3928], device='cuda:0'), covar=tensor([0.0457, 0.0796, 0.0387, 0.0339, 0.0736, 0.1232, 0.0923, 0.1299], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0050, 0.0063, 0.0057, 0.0080, 0.0085, 0.0049, 0.0079], device='cuda:0'), out_proj_covar=tensor([1.0261e-04, 8.4099e-05, 9.3979e-05, 8.6742e-05, 1.2500e-04, 1.2526e-04, 8.3339e-05, 1.1619e-04], device='cuda:0') 2023-04-16 13:51:47,575 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9047.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:51:48,083 INFO [train.py:893] (0/4) Epoch 4, batch 950, loss[loss=0.2842, simple_loss=0.3116, pruned_loss=0.1284, over 13501.00 frames. ], tot_loss[loss=0.3015, simple_loss=0.3234, pruned_loss=0.1398, over 2633527.04 frames. ], batch size: 70, lr: 3.03e-02, grad_scale: 8.0 2023-04-16 13:52:06,965 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7695, 2.0107, 3.9666, 3.6460, 3.9261, 3.7077, 3.8250, 2.5888], device='cuda:0'), covar=tensor([0.2167, 0.1998, 0.0103, 0.0234, 0.0198, 0.0262, 0.0154, 0.1130], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0160, 0.0074, 0.0080, 0.0082, 0.0108, 0.0079, 0.0127], device='cuda:0'), out_proj_covar=tensor([1.6991e-04, 1.7104e-04, 8.1613e-05, 9.7351e-05, 9.7940e-05, 1.1921e-04, 9.3726e-05, 1.3729e-04], device='cuda:0') 2023-04-16 13:52:09,712 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2023-04-16 13:52:12,723 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9074.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:52:25,571 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1016, 4.6277, 4.2707, 4.1327, 4.2978, 4.1057, 4.5577, 4.6445], device='cuda:0'), covar=tensor([0.0300, 0.0199, 0.0291, 0.0365, 0.0257, 0.0345, 0.0342, 0.0206], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0115, 0.0082, 0.0113, 0.0080, 0.0106, 0.0086, 0.0086], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 13:52:34,816 INFO [train.py:893] (0/4) Epoch 4, batch 1000, loss[loss=0.2646, simple_loss=0.3007, pruned_loss=0.1143, over 13554.00 frames. ], tot_loss[loss=0.2995, simple_loss=0.3213, pruned_loss=0.1388, over 2639785.20 frames. ], batch size: 72, lr: 3.02e-02, grad_scale: 8.0 2023-04-16 13:52:39,600 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.434e+02 3.856e+02 4.710e+02 6.550e+02 2.342e+03, threshold=9.421e+02, percent-clipped=6.0 2023-04-16 13:52:50,583 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9116.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:53:19,574 INFO [train.py:893] (0/4) Epoch 4, batch 1050, loss[loss=0.2658, simple_loss=0.3024, pruned_loss=0.1146, over 13524.00 frames. ], tot_loss[loss=0.2964, simple_loss=0.3194, pruned_loss=0.1367, over 2646732.22 frames. ], batch size: 85, lr: 3.02e-02, grad_scale: 8.0 2023-04-16 13:54:04,170 INFO [train.py:893] (0/4) Epoch 4, batch 1100, loss[loss=0.2997, simple_loss=0.3231, pruned_loss=0.1382, over 13485.00 frames. ], tot_loss[loss=0.2969, simple_loss=0.3205, pruned_loss=0.1367, over 2647630.24 frames. ], batch size: 81, lr: 3.01e-02, grad_scale: 8.0 2023-04-16 13:54:09,399 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.293e+02 4.489e+02 5.675e+02 6.902e+02 1.290e+03, threshold=1.135e+03, percent-clipped=12.0 2023-04-16 13:54:34,166 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.22 vs. limit=5.0 2023-04-16 13:54:49,880 INFO [train.py:893] (0/4) Epoch 4, batch 1150, loss[loss=0.2581, simple_loss=0.3003, pruned_loss=0.108, over 13477.00 frames. ], tot_loss[loss=0.2943, simple_loss=0.3188, pruned_loss=0.1349, over 2637012.42 frames. ], batch size: 81, lr: 3.00e-02, grad_scale: 8.0 2023-04-16 13:55:14,057 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.23 vs. limit=2.0 2023-04-16 13:55:22,807 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9283.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:55:35,798 INFO [train.py:893] (0/4) Epoch 4, batch 1200, loss[loss=0.3008, simple_loss=0.3293, pruned_loss=0.1362, over 13041.00 frames. ], tot_loss[loss=0.2939, simple_loss=0.3189, pruned_loss=0.1344, over 2639906.97 frames. ], batch size: 142, lr: 3.00e-02, grad_scale: 8.0 2023-04-16 13:55:41,984 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.867e+02 4.129e+02 4.847e+02 6.110e+02 1.320e+03, threshold=9.693e+02, percent-clipped=2.0 2023-04-16 13:56:02,689 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 13:56:14,805 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 13:56:16,571 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9342.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:56:22,749 INFO [train.py:893] (0/4) Epoch 4, batch 1250, loss[loss=0.3242, simple_loss=0.3472, pruned_loss=0.1506, over 13270.00 frames. ], tot_loss[loss=0.2959, simple_loss=0.3207, pruned_loss=0.1355, over 2645718.45 frames. ], batch size: 124, lr: 2.99e-02, grad_scale: 8.0 2023-04-16 13:56:41,607 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9369.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:56:52,968 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6476, 4.0494, 3.8183, 3.6925, 3.8364, 3.6925, 4.0971, 4.1415], device='cuda:0'), covar=tensor([0.0331, 0.0347, 0.0299, 0.0406, 0.0414, 0.0380, 0.0377, 0.0297], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0122, 0.0084, 0.0115, 0.0086, 0.0111, 0.0091, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 13:57:04,587 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8145, 3.9418, 3.9347, 4.6226, 2.1158, 3.5250, 4.0839, 2.1278], device='cuda:0'), covar=tensor([0.0049, 0.0384, 0.0430, 0.0094, 0.1956, 0.0559, 0.0431, 0.2325], device='cuda:0'), in_proj_covar=tensor([0.0069, 0.0112, 0.0128, 0.0070, 0.0159, 0.0134, 0.0128, 0.0163], device='cuda:0'), out_proj_covar=tensor([9.8912e-05, 1.6785e-04, 1.7950e-04, 1.0880e-04, 2.1124e-04, 1.8450e-04, 1.8222e-04, 2.2045e-04], device='cuda:0') 2023-04-16 13:57:08,259 INFO [train.py:893] (0/4) Epoch 4, batch 1300, loss[loss=0.2916, simple_loss=0.3165, pruned_loss=0.1333, over 13550.00 frames. ], tot_loss[loss=0.2981, simple_loss=0.3227, pruned_loss=0.1367, over 2651687.96 frames. ], batch size: 76, lr: 2.99e-02, grad_scale: 8.0 2023-04-16 13:57:14,241 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.514e+02 4.181e+02 5.160e+02 6.106e+02 1.309e+03, threshold=1.032e+03, percent-clipped=5.0 2023-04-16 13:57:15,380 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5147, 2.2215, 2.2617, 3.8771, 3.3503, 3.9056, 3.0319, 2.2721], device='cuda:0'), covar=tensor([0.0203, 0.1287, 0.0978, 0.0067, 0.0276, 0.0061, 0.0448, 0.1248], device='cuda:0'), in_proj_covar=tensor([0.0061, 0.0117, 0.0106, 0.0062, 0.0060, 0.0051, 0.0098, 0.0108], device='cuda:0'), out_proj_covar=tensor([1.0146e-04, 1.6865e-04, 1.5358e-04, 9.2290e-05, 1.0072e-04, 8.0921e-05, 1.4147e-04, 1.5378e-04], device='cuda:0') 2023-04-16 13:57:23,500 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9414.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:57:25,104 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9416.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:57:54,160 INFO [train.py:893] (0/4) Epoch 4, batch 1350, loss[loss=0.2432, simple_loss=0.2673, pruned_loss=0.1095, over 12753.00 frames. ], tot_loss[loss=0.2986, simple_loss=0.3231, pruned_loss=0.1371, over 2650026.79 frames. ], batch size: 52, lr: 2.98e-02, grad_scale: 8.0 2023-04-16 13:58:08,933 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9464.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:58:12,396 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1292, 3.6975, 3.4988, 4.1215, 2.1415, 3.0563, 3.5433, 1.8668], device='cuda:0'), covar=tensor([0.0101, 0.0299, 0.0399, 0.0104, 0.1476, 0.0548, 0.0496, 0.2345], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0107, 0.0127, 0.0071, 0.0155, 0.0131, 0.0123, 0.0161], device='cuda:0'), out_proj_covar=tensor([9.7623e-05, 1.6117e-04, 1.7770e-04, 1.0956e-04, 2.0612e-04, 1.8170e-04, 1.7677e-04, 2.1691e-04], device='cuda:0') 2023-04-16 13:58:18,068 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-16 13:58:18,770 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9475.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:58:39,245 INFO [train.py:893] (0/4) Epoch 4, batch 1400, loss[loss=0.3229, simple_loss=0.3398, pruned_loss=0.153, over 13230.00 frames. ], tot_loss[loss=0.2977, simple_loss=0.3222, pruned_loss=0.1366, over 2656375.76 frames. ], batch size: 132, lr: 2.97e-02, grad_scale: 8.0 2023-04-16 13:58:44,925 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.732e+02 4.170e+02 5.056e+02 6.250e+02 1.519e+03, threshold=1.011e+03, percent-clipped=3.0 2023-04-16 13:59:22,677 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5522, 4.3411, 4.1508, 3.6843, 3.8703, 2.5155, 4.4482, 2.9213], device='cuda:0'), covar=tensor([0.0788, 0.0158, 0.0232, 0.0499, 0.0268, 0.1811, 0.0122, 0.1935], device='cuda:0'), in_proj_covar=tensor([0.0216, 0.0161, 0.0174, 0.0202, 0.0160, 0.0211, 0.0122, 0.0259], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 13:59:23,884 INFO [train.py:893] (0/4) Epoch 4, batch 1450, loss[loss=0.2911, simple_loss=0.3181, pruned_loss=0.1321, over 13488.00 frames. ], tot_loss[loss=0.2963, simple_loss=0.321, pruned_loss=0.1358, over 2656571.36 frames. ], batch size: 100, lr: 2.97e-02, grad_scale: 8.0 2023-04-16 13:59:40,808 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4427, 4.1431, 3.9667, 2.8195, 2.2570, 2.7165, 4.0975, 4.2872], device='cuda:0'), covar=tensor([0.0629, 0.0367, 0.0328, 0.1399, 0.1695, 0.1061, 0.0168, 0.0075], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0100, 0.0111, 0.0168, 0.0168, 0.0139, 0.0083, 0.0071], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 13:59:56,217 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9583.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 13:59:57,104 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9584.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:00:01,444 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7159, 1.5100, 1.2077, 1.9320, 1.0791, 1.2068, 1.6676, 1.8483], device='cuda:0'), covar=tensor([0.0207, 0.0366, 0.0257, 0.0189, 0.0192, 0.0184, 0.0402, 0.0290], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0033, 0.0035, 0.0029, 0.0035, 0.0029, 0.0036, 0.0037], device='cuda:0'), out_proj_covar=tensor([4.1764e-05, 4.1660e-05, 4.3745e-05, 3.6638e-05, 4.1582e-05, 3.4414e-05, 4.4801e-05, 4.6115e-05], device='cuda:0') 2023-04-16 14:00:10,766 INFO [train.py:893] (0/4) Epoch 4, batch 1500, loss[loss=0.3044, simple_loss=0.3065, pruned_loss=0.1512, over 12613.00 frames. ], tot_loss[loss=0.2949, simple_loss=0.3204, pruned_loss=0.1347, over 2658147.12 frames. ], batch size: 51, lr: 2.96e-02, grad_scale: 8.0 2023-04-16 14:00:16,018 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.528e+02 4.187e+02 4.847e+02 6.018e+02 1.274e+03, threshold=9.694e+02, percent-clipped=5.0 2023-04-16 14:00:39,072 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3929, 4.6332, 2.9947, 4.7165, 4.4804, 2.4584, 3.8099, 2.9000], device='cuda:0'), covar=tensor([0.0204, 0.0319, 0.1527, 0.0084, 0.0289, 0.1896, 0.0878, 0.2342], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0095, 0.0155, 0.0067, 0.0095, 0.0141, 0.0116, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-04-16 14:00:41,434 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9631.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:00:50,634 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9642.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:00:53,972 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9645.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:00:56,117 INFO [train.py:893] (0/4) Epoch 4, batch 1550, loss[loss=0.2972, simple_loss=0.3284, pruned_loss=0.133, over 13272.00 frames. ], tot_loss[loss=0.2935, simple_loss=0.3193, pruned_loss=0.1339, over 2653955.85 frames. ], batch size: 124, lr: 2.96e-02, grad_scale: 8.0 2023-04-16 14:01:06,367 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9658.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:01:15,431 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9669.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:01:35,753 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9690.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:01:42,231 INFO [train.py:893] (0/4) Epoch 4, batch 1600, loss[loss=0.2936, simple_loss=0.324, pruned_loss=0.1315, over 13258.00 frames. ], tot_loss[loss=0.2941, simple_loss=0.32, pruned_loss=0.1341, over 2653474.48 frames. ], batch size: 124, lr: 2.95e-02, grad_scale: 8.0 2023-04-16 14:01:46,953 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.855e+02 4.347e+02 5.620e+02 6.749e+02 1.079e+03, threshold=1.124e+03, percent-clipped=2.0 2023-04-16 14:01:59,698 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9717.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:02:02,280 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9719.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:02:12,198 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7145, 4.1450, 3.5797, 3.7213, 3.9411, 4.3691, 4.0680, 4.0348], device='cuda:0'), covar=tensor([0.0325, 0.0260, 0.0348, 0.1147, 0.0256, 0.0251, 0.0285, 0.0334], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0077, 0.0076, 0.0130, 0.0076, 0.0086, 0.0072, 0.0071], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 14:02:23,445 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9743.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:02:27,978 INFO [train.py:893] (0/4) Epoch 4, batch 1650, loss[loss=0.2984, simple_loss=0.3286, pruned_loss=0.1341, over 13577.00 frames. ], tot_loss[loss=0.2945, simple_loss=0.321, pruned_loss=0.134, over 2656128.88 frames. ], batch size: 89, lr: 2.94e-02, grad_scale: 8.0 2023-04-16 14:02:47,825 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9770.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:03:12,544 INFO [train.py:893] (0/4) Epoch 4, batch 1700, loss[loss=0.2864, simple_loss=0.3117, pruned_loss=0.1305, over 13530.00 frames. ], tot_loss[loss=0.2948, simple_loss=0.322, pruned_loss=0.1339, over 2659207.28 frames. ], batch size: 85, lr: 2.94e-02, grad_scale: 8.0 2023-04-16 14:03:12,862 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3232, 2.3320, 2.1029, 2.2447, 2.3648, 1.2259, 2.5636, 1.3031], device='cuda:0'), covar=tensor([0.0471, 0.0808, 0.0456, 0.0348, 0.0656, 0.1282, 0.0771, 0.1155], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0053, 0.0067, 0.0058, 0.0080, 0.0094, 0.0054, 0.0082], device='cuda:0'), out_proj_covar=tensor([1.0879e-04, 8.9307e-05, 1.0277e-04, 8.8358e-05, 1.2491e-04, 1.4075e-04, 9.1050e-05, 1.2338e-04], device='cuda:0') 2023-04-16 14:03:18,391 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.356e+02 4.078e+02 4.799e+02 5.634e+02 1.078e+03, threshold=9.597e+02, percent-clipped=0.0 2023-04-16 14:03:18,725 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9804.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:03:58,750 INFO [train.py:893] (0/4) Epoch 4, batch 1750, loss[loss=0.2365, simple_loss=0.2745, pruned_loss=0.09926, over 13441.00 frames. ], tot_loss[loss=0.2931, simple_loss=0.3204, pruned_loss=0.1329, over 2662277.01 frames. ], batch size: 65, lr: 2.93e-02, grad_scale: 8.0 2023-04-16 14:04:00,841 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.45 vs. limit=5.0 2023-04-16 14:04:24,649 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6551, 4.0846, 2.3437, 4.2134, 3.8665, 1.9752, 3.1026, 2.5043], device='cuda:0'), covar=tensor([0.0269, 0.0242, 0.1632, 0.0123, 0.0260, 0.1616, 0.0916, 0.1889], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0093, 0.0155, 0.0067, 0.0093, 0.0140, 0.0117, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-04-16 14:04:43,280 INFO [train.py:893] (0/4) Epoch 4, batch 1800, loss[loss=0.3002, simple_loss=0.3194, pruned_loss=0.1405, over 13532.00 frames. ], tot_loss[loss=0.2903, simple_loss=0.3181, pruned_loss=0.1312, over 2665823.68 frames. ], batch size: 83, lr: 2.93e-02, grad_scale: 8.0 2023-04-16 14:04:49,600 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.550e+02 3.859e+02 4.667e+02 6.022e+02 1.586e+03, threshold=9.334e+02, percent-clipped=8.0 2023-04-16 14:05:18,903 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.5323, 2.4798, 1.8073, 1.6049, 1.3515, 1.4661, 1.8259, 2.4224], device='cuda:0'), covar=tensor([0.0654, 0.0269, 0.0828, 0.1040, 0.0713, 0.0324, 0.0578, 0.0347], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0055, 0.0055, 0.0087, 0.0053, 0.0059, 0.0071, 0.0057], device='cuda:0'), out_proj_covar=tensor([5.7906e-05, 4.6550e-05, 5.1793e-05, 8.0028e-05, 5.7528e-05, 4.8111e-05, 6.3417e-05, 4.5564e-05], device='cuda:0') 2023-04-16 14:05:22,830 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9940.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:05:29,076 INFO [train.py:893] (0/4) Epoch 4, batch 1850, loss[loss=0.315, simple_loss=0.3344, pruned_loss=0.1478, over 13367.00 frames. ], tot_loss[loss=0.29, simple_loss=0.318, pruned_loss=0.131, over 2667474.85 frames. ], batch size: 109, lr: 2.92e-02, grad_scale: 8.0 2023-04-16 14:05:31,578 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 14:05:42,284 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6276, 2.4489, 2.8192, 4.1467, 3.5237, 3.9880, 3.1638, 2.5242], device='cuda:0'), covar=tensor([0.0159, 0.1258, 0.0835, 0.0066, 0.0246, 0.0057, 0.0461, 0.1001], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0118, 0.0110, 0.0062, 0.0060, 0.0054, 0.0100, 0.0111], device='cuda:0'), out_proj_covar=tensor([1.0997e-04, 1.7281e-04, 1.5964e-04, 9.4495e-05, 1.0182e-04, 8.5685e-05, 1.4667e-04, 1.6301e-04], device='cuda:0') 2023-04-16 14:05:44,850 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0352, 2.1317, 3.6823, 3.5686, 3.6110, 3.4954, 3.1069, 2.8701], device='cuda:0'), covar=tensor([0.2197, 0.2085, 0.0101, 0.0262, 0.0204, 0.0314, 0.0285, 0.0890], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0166, 0.0075, 0.0082, 0.0082, 0.0114, 0.0084, 0.0132], device='cuda:0'), out_proj_covar=tensor([1.7267e-04, 1.7872e-04, 8.4675e-05, 9.9416e-05, 1.0049e-04, 1.2824e-04, 1.0224e-04, 1.4577e-04], device='cuda:0') 2023-04-16 14:06:14,475 INFO [train.py:893] (0/4) Epoch 4, batch 1900, loss[loss=0.3181, simple_loss=0.3413, pruned_loss=0.1474, over 13359.00 frames. ], tot_loss[loss=0.2909, simple_loss=0.3181, pruned_loss=0.1319, over 2664766.67 frames. ], batch size: 109, lr: 2.91e-02, grad_scale: 8.0 2023-04-16 14:06:16,417 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-10000.pt 2023-04-16 14:06:22,803 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0397, 4.3359, 2.6966, 4.4827, 4.1421, 2.2637, 3.5896, 2.6934], device='cuda:0'), covar=tensor([0.0215, 0.0244, 0.1391, 0.0063, 0.0246, 0.1489, 0.0618, 0.1727], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0095, 0.0156, 0.0067, 0.0093, 0.0140, 0.0117, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-04-16 14:06:23,226 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.631e+02 4.099e+02 5.413e+02 6.648e+02 1.210e+03, threshold=1.083e+03, percent-clipped=7.0 2023-04-16 14:06:31,771 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10014.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:06:54,495 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10039.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:07:01,576 INFO [train.py:893] (0/4) Epoch 4, batch 1950, loss[loss=0.3122, simple_loss=0.3301, pruned_loss=0.1471, over 13533.00 frames. ], tot_loss[loss=0.2889, simple_loss=0.3164, pruned_loss=0.1307, over 2660223.62 frames. ], batch size: 85, lr: 2.91e-02, grad_scale: 8.0 2023-04-16 14:07:14,748 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10062.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 14:07:21,956 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10070.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:07:47,210 INFO [train.py:893] (0/4) Epoch 4, batch 2000, loss[loss=0.3537, simple_loss=0.3739, pruned_loss=0.1667, over 13393.00 frames. ], tot_loss[loss=0.2907, simple_loss=0.318, pruned_loss=0.1316, over 2661385.46 frames. ], batch size: 113, lr: 2.90e-02, grad_scale: 8.0 2023-04-16 14:07:48,264 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10099.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:07:49,239 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10100.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 14:07:52,259 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.683e+02 4.046e+02 5.081e+02 6.604e+02 1.044e+03, threshold=1.016e+03, percent-clipped=0.0 2023-04-16 14:07:53,094 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 14:08:04,522 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10118.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:08:10,039 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10123.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 14:08:31,934 INFO [train.py:893] (0/4) Epoch 4, batch 2050, loss[loss=0.2655, simple_loss=0.2997, pruned_loss=0.1157, over 13525.00 frames. ], tot_loss[loss=0.2921, simple_loss=0.3191, pruned_loss=0.1326, over 2661297.00 frames. ], batch size: 76, lr: 2.90e-02, grad_scale: 8.0 2023-04-16 14:09:18,324 INFO [train.py:893] (0/4) Epoch 4, batch 2100, loss[loss=0.341, simple_loss=0.3372, pruned_loss=0.1724, over 11723.00 frames. ], tot_loss[loss=0.292, simple_loss=0.319, pruned_loss=0.1324, over 2662872.22 frames. ], batch size: 157, lr: 2.89e-02, grad_scale: 8.0 2023-04-16 14:09:23,521 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.173e+02 4.294e+02 5.277e+02 6.464e+02 1.434e+03, threshold=1.055e+03, percent-clipped=4.0 2023-04-16 14:09:41,837 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2023-04-16 14:09:57,756 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10240.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:10:05,655 INFO [train.py:893] (0/4) Epoch 4, batch 2150, loss[loss=0.2981, simple_loss=0.3309, pruned_loss=0.1326, over 13447.00 frames. ], tot_loss[loss=0.2898, simple_loss=0.3176, pruned_loss=0.1309, over 2662398.35 frames. ], batch size: 95, lr: 2.88e-02, grad_scale: 16.0 2023-04-16 14:10:41,506 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10288.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:10:49,437 INFO [train.py:893] (0/4) Epoch 4, batch 2200, loss[loss=0.275, simple_loss=0.2974, pruned_loss=0.1263, over 13353.00 frames. ], tot_loss[loss=0.2877, simple_loss=0.3159, pruned_loss=0.1298, over 2662543.77 frames. ], batch size: 67, lr: 2.88e-02, grad_scale: 16.0 2023-04-16 14:10:49,832 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6362, 4.6408, 4.2880, 3.7420, 4.1582, 2.4814, 4.8123, 2.7783], device='cuda:0'), covar=tensor([0.0975, 0.0223, 0.0334, 0.0755, 0.0289, 0.2344, 0.0135, 0.2898], device='cuda:0'), in_proj_covar=tensor([0.0224, 0.0172, 0.0182, 0.0212, 0.0167, 0.0217, 0.0131, 0.0266], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 14:10:55,071 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.608e+02 3.593e+02 4.262e+02 5.500e+02 1.224e+03, threshold=8.524e+02, percent-clipped=2.0 2023-04-16 14:11:05,673 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10314.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:11:33,112 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6811, 4.5338, 4.5745, 4.4989, 4.8953, 4.2578, 4.8442, 4.8910], device='cuda:0'), covar=tensor([0.0473, 0.0824, 0.0899, 0.0610, 0.0987, 0.1234, 0.0713, 0.0777], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0159, 0.0146, 0.0110, 0.0203, 0.0176, 0.0125, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 14:11:36,205 INFO [train.py:893] (0/4) Epoch 4, batch 2250, loss[loss=0.2811, simple_loss=0.3051, pruned_loss=0.1286, over 13538.00 frames. ], tot_loss[loss=0.2854, simple_loss=0.3133, pruned_loss=0.1287, over 2658441.35 frames. ], batch size: 83, lr: 2.87e-02, grad_scale: 16.0 2023-04-16 14:11:47,603 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2023-04-16 14:11:48,029 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10362.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:12:03,369 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7791, 2.7440, 2.0711, 3.5082, 4.2105, 3.1467, 4.1359, 3.7377], device='cuda:0'), covar=tensor([0.0111, 0.0448, 0.0817, 0.0113, 0.0092, 0.0300, 0.0096, 0.0076], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0045, 0.0057, 0.0041, 0.0028, 0.0043, 0.0026, 0.0033], device='cuda:0'), out_proj_covar=tensor([1.3601e-04, 1.7190e-04, 2.0525e-04, 1.5917e-04, 1.1053e-04, 1.6728e-04, 9.7966e-05, 1.2724e-04], device='cuda:0') 2023-04-16 14:12:18,719 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10395.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 14:12:21,695 INFO [train.py:893] (0/4) Epoch 4, batch 2300, loss[loss=0.2911, simple_loss=0.3209, pruned_loss=0.1307, over 13266.00 frames. ], tot_loss[loss=0.2848, simple_loss=0.3134, pruned_loss=0.1281, over 2663048.65 frames. ], batch size: 124, lr: 2.87e-02, grad_scale: 16.0 2023-04-16 14:12:22,858 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10399.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:12:24,192 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1289, 4.1208, 3.7245, 3.4980, 3.0611, 2.2778, 4.3087, 2.5532], device='cuda:0'), covar=tensor([0.1058, 0.0248, 0.0365, 0.0579, 0.0543, 0.2197, 0.0151, 0.2397], device='cuda:0'), in_proj_covar=tensor([0.0230, 0.0174, 0.0184, 0.0215, 0.0169, 0.0223, 0.0133, 0.0272], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 14:12:27,576 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.018e+02 3.904e+02 5.068e+02 6.089e+02 1.155e+03, threshold=1.014e+03, percent-clipped=7.0 2023-04-16 14:12:32,767 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10409.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:12:40,241 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10418.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 14:13:00,441 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0652, 2.2713, 1.3467, 1.8591, 1.3268, 1.5795, 1.9115, 2.0586], device='cuda:0'), covar=tensor([0.0131, 0.0143, 0.0315, 0.0181, 0.0165, 0.0145, 0.0331, 0.0170], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0034, 0.0037, 0.0031, 0.0037, 0.0032, 0.0038, 0.0040], device='cuda:0'), out_proj_covar=tensor([4.2740e-05, 4.1728e-05, 4.6285e-05, 3.9020e-05, 4.3978e-05, 3.8083e-05, 4.7060e-05, 5.1135e-05], device='cuda:0') 2023-04-16 14:13:06,932 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10447.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:13:07,578 INFO [train.py:893] (0/4) Epoch 4, batch 2350, loss[loss=0.2857, simple_loss=0.3147, pruned_loss=0.1284, over 13417.00 frames. ], tot_loss[loss=0.284, simple_loss=0.313, pruned_loss=0.1275, over 2665160.07 frames. ], batch size: 88, lr: 2.86e-02, grad_scale: 16.0 2023-04-16 14:13:12,926 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10454.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:13:23,272 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.3809, 2.2929, 1.8048, 1.1863, 1.0412, 1.4246, 1.4517, 2.2730], device='cuda:0'), covar=tensor([0.0756, 0.0354, 0.1092, 0.1349, 0.0345, 0.0212, 0.0746, 0.0321], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0062, 0.0063, 0.0094, 0.0059, 0.0062, 0.0080, 0.0063], device='cuda:0'), out_proj_covar=tensor([6.5302e-05, 5.1461e-05, 5.8105e-05, 8.7133e-05, 6.2038e-05, 5.1439e-05, 6.9517e-05, 4.9432e-05], device='cuda:0') 2023-04-16 14:13:28,026 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10470.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:13:29,462 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 14:13:52,776 INFO [train.py:893] (0/4) Epoch 4, batch 2400, loss[loss=0.2923, simple_loss=0.3193, pruned_loss=0.1326, over 13401.00 frames. ], tot_loss[loss=0.2832, simple_loss=0.312, pruned_loss=0.1273, over 2665140.63 frames. ], batch size: 113, lr: 2.86e-02, grad_scale: 16.0 2023-04-16 14:13:58,433 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.647e+02 3.691e+02 4.650e+02 5.820e+02 1.816e+03, threshold=9.300e+02, percent-clipped=3.0 2023-04-16 14:14:02,067 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8944, 3.8275, 3.9239, 3.1690, 4.4430, 3.9234, 4.0191, 4.3538], device='cuda:0'), covar=tensor([0.0192, 0.0089, 0.0119, 0.0620, 0.0092, 0.0122, 0.0130, 0.0084], device='cuda:0'), in_proj_covar=tensor([0.0046, 0.0025, 0.0040, 0.0060, 0.0048, 0.0039, 0.0042, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:0') 2023-04-16 14:14:08,861 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10515.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 14:14:38,408 INFO [train.py:893] (0/4) Epoch 4, batch 2450, loss[loss=0.2609, simple_loss=0.2978, pruned_loss=0.112, over 13358.00 frames. ], tot_loss[loss=0.2829, simple_loss=0.3119, pruned_loss=0.1269, over 2665765.05 frames. ], batch size: 62, lr: 2.85e-02, grad_scale: 16.0 2023-04-16 14:15:24,373 INFO [train.py:893] (0/4) Epoch 4, batch 2500, loss[loss=0.2547, simple_loss=0.289, pruned_loss=0.1102, over 13530.00 frames. ], tot_loss[loss=0.2814, simple_loss=0.311, pruned_loss=0.1259, over 2659109.56 frames. ], batch size: 72, lr: 2.85e-02, grad_scale: 16.0 2023-04-16 14:15:30,373 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.433e+02 3.816e+02 4.383e+02 5.440e+02 1.104e+03, threshold=8.765e+02, percent-clipped=1.0 2023-04-16 14:15:54,393 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6556, 2.6837, 2.1453, 3.3525, 4.1056, 3.0943, 4.0700, 3.8203], device='cuda:0'), covar=tensor([0.0114, 0.0412, 0.0709, 0.0112, 0.0068, 0.0284, 0.0081, 0.0065], device='cuda:0'), in_proj_covar=tensor([0.0036, 0.0046, 0.0057, 0.0041, 0.0028, 0.0044, 0.0026, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 14:16:10,382 INFO [train.py:893] (0/4) Epoch 4, batch 2550, loss[loss=0.273, simple_loss=0.3072, pruned_loss=0.1193, over 13531.00 frames. ], tot_loss[loss=0.2811, simple_loss=0.3108, pruned_loss=0.1257, over 2657559.34 frames. ], batch size: 98, lr: 2.84e-02, grad_scale: 16.0 2023-04-16 14:16:34,260 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 14:16:48,998 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2023-04-16 14:16:52,053 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0435, 3.0677, 3.0582, 3.0132, 2.9025, 1.8006, 3.2304, 1.8691], device='cuda:0'), covar=tensor([0.0386, 0.0498, 0.0305, 0.0366, 0.0614, 0.1248, 0.0878, 0.1165], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0062, 0.0072, 0.0064, 0.0086, 0.0105, 0.0060, 0.0090], device='cuda:0'), out_proj_covar=tensor([1.2127e-04, 1.0358e-04, 1.1209e-04, 9.8027e-05, 1.3713e-04, 1.5947e-04, 1.0394e-04, 1.3599e-04], device='cuda:0') 2023-04-16 14:16:53,625 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10695.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:16:53,735 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6707, 2.3356, 1.6356, 1.3943, 1.1412, 1.7946, 1.5389, 2.3311], device='cuda:0'), covar=tensor([0.0705, 0.0558, 0.1292, 0.1309, 0.0398, 0.0289, 0.0764, 0.0488], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0065, 0.0062, 0.0096, 0.0060, 0.0065, 0.0081, 0.0066], device='cuda:0'), out_proj_covar=tensor([6.6658e-05, 5.3826e-05, 5.7661e-05, 8.9269e-05, 6.0934e-05, 5.4731e-05, 7.0770e-05, 5.3208e-05], device='cuda:0') 2023-04-16 14:16:55,854 INFO [train.py:893] (0/4) Epoch 4, batch 2600, loss[loss=0.2556, simple_loss=0.2839, pruned_loss=0.1137, over 13365.00 frames. ], tot_loss[loss=0.2812, simple_loss=0.3102, pruned_loss=0.1261, over 2653998.63 frames. ], batch size: 67, lr: 2.83e-02, grad_scale: 16.0 2023-04-16 14:17:01,529 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.544e+02 3.996e+02 5.390e+02 7.018e+02 1.415e+03, threshold=1.078e+03, percent-clipped=12.0 2023-04-16 14:17:04,414 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.7943, 4.2622, 4.1050, 4.6536, 5.1435, 4.1752, 4.9961, 4.7824], device='cuda:0'), covar=tensor([0.0059, 0.0176, 0.0284, 0.0057, 0.0043, 0.0144, 0.0085, 0.0046], device='cuda:0'), in_proj_covar=tensor([0.0036, 0.0046, 0.0058, 0.0042, 0.0028, 0.0044, 0.0027, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 14:17:06,224 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0712, 1.5088, 1.2454, 1.7487, 1.1189, 1.2318, 1.6574, 1.9709], device='cuda:0'), covar=tensor([0.0114, 0.0243, 0.0213, 0.0137, 0.0183, 0.0193, 0.0286, 0.0226], device='cuda:0'), in_proj_covar=tensor([0.0037, 0.0038, 0.0039, 0.0032, 0.0039, 0.0034, 0.0039, 0.0042], device='cuda:0'), out_proj_covar=tensor([4.5291e-05, 4.6163e-05, 4.8980e-05, 3.9923e-05, 4.7319e-05, 4.0031e-05, 4.9616e-05, 5.3052e-05], device='cuda:0') 2023-04-16 14:17:06,429 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2023-04-16 14:17:14,898 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10718.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 14:17:30,468 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.78 vs. limit=5.0 2023-04-16 14:17:34,590 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10743.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:17:35,544 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10744.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:17:38,437 INFO [train.py:893] (0/4) Epoch 4, batch 2650, loss[loss=0.2761, simple_loss=0.3118, pruned_loss=0.1202, over 13528.00 frames. ], tot_loss[loss=0.2849, simple_loss=0.3127, pruned_loss=0.1286, over 2652369.46 frames. ], batch size: 76, lr: 2.83e-02, grad_scale: 16.0 2023-04-16 14:17:46,698 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2932, 4.1170, 4.4084, 4.2599, 4.4757, 4.0429, 4.5478, 4.5568], device='cuda:0'), covar=tensor([0.0309, 0.0505, 0.0435, 0.0331, 0.0650, 0.0713, 0.0417, 0.0390], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0166, 0.0149, 0.0109, 0.0210, 0.0184, 0.0127, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 14:17:51,216 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10765.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:17:51,939 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10766.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 14:17:58,103 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10774.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:18:05,806 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-16 14:18:16,192 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-4.pt 2023-04-16 14:18:40,367 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 14:18:44,291 INFO [train.py:893] (0/4) Epoch 5, batch 0, loss[loss=0.3226, simple_loss=0.338, pruned_loss=0.1536, over 13424.00 frames. ], tot_loss[loss=0.3226, simple_loss=0.338, pruned_loss=0.1536, over 13424.00 frames. ], batch size: 106, lr: 2.63e-02, grad_scale: 16.0 2023-04-16 14:18:44,292 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 14:19:06,509 INFO [train.py:927] (0/4) Epoch 5, validation: loss=0.2077, simple_loss=0.2523, pruned_loss=0.08155, over 2446609.00 frames. 2023-04-16 14:19:06,510 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 14:19:13,546 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.868e+02 4.381e+02 5.128e+02 6.177e+02 2.040e+03, threshold=1.026e+03, percent-clipped=5.0 2023-04-16 14:19:14,757 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10805.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:19:19,604 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10810.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 14:19:20,054 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-16 14:19:43,006 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10835.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:19:53,469 INFO [train.py:893] (0/4) Epoch 5, batch 50, loss[loss=0.285, simple_loss=0.3163, pruned_loss=0.1268, over 13408.00 frames. ], tot_loss[loss=0.2808, simple_loss=0.3064, pruned_loss=0.1276, over 600424.21 frames. ], batch size: 95, lr: 2.63e-02, grad_scale: 16.0 2023-04-16 14:20:13,453 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8553, 4.3077, 3.7141, 3.9902, 3.8657, 4.5643, 4.1613, 4.2227], device='cuda:0'), covar=tensor([0.0365, 0.0289, 0.0347, 0.1093, 0.0323, 0.0235, 0.0271, 0.0231], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0078, 0.0079, 0.0142, 0.0081, 0.0091, 0.0077, 0.0074], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 14:20:18,224 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 14:20:18,224 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 14:20:18,224 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 14:20:18,237 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 14:20:18,245 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 14:20:19,203 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 14:20:19,222 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 14:20:25,279 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9194, 4.7148, 5.1651, 4.8354, 5.2907, 4.6517, 5.3018, 5.2752], device='cuda:0'), covar=tensor([0.0314, 0.0522, 0.0461, 0.0452, 0.0572, 0.0838, 0.0433, 0.0411], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0175, 0.0158, 0.0116, 0.0222, 0.0193, 0.0136, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 14:20:28,740 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10885.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 14:20:40,858 INFO [train.py:893] (0/4) Epoch 5, batch 100, loss[loss=0.3006, simple_loss=0.3286, pruned_loss=0.1363, over 13533.00 frames. ], tot_loss[loss=0.2824, simple_loss=0.3086, pruned_loss=0.1281, over 1062145.45 frames. ], batch size: 87, lr: 2.62e-02, grad_scale: 16.0 2023-04-16 14:20:46,854 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.725e+02 3.963e+02 4.715e+02 5.825e+02 1.168e+03, threshold=9.430e+02, percent-clipped=1.0 2023-04-16 14:21:17,511 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-16 14:21:17,970 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7893, 4.0432, 2.6248, 4.2560, 3.8736, 2.0994, 3.2640, 2.4535], device='cuda:0'), covar=tensor([0.0278, 0.0285, 0.1514, 0.0058, 0.0262, 0.1632, 0.0702, 0.2026], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0102, 0.0161, 0.0066, 0.0098, 0.0145, 0.0125, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-04-16 14:21:27,210 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10946.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 14:21:27,774 INFO [train.py:893] (0/4) Epoch 5, batch 150, loss[loss=0.2851, simple_loss=0.3229, pruned_loss=0.1236, over 13528.00 frames. ], tot_loss[loss=0.2857, simple_loss=0.3113, pruned_loss=0.1301, over 1404022.72 frames. ], batch size: 98, lr: 2.62e-02, grad_scale: 16.0 2023-04-16 14:21:41,346 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-16 14:22:15,272 INFO [train.py:893] (0/4) Epoch 5, batch 200, loss[loss=0.3353, simple_loss=0.346, pruned_loss=0.1623, over 13521.00 frames. ], tot_loss[loss=0.2878, simple_loss=0.3137, pruned_loss=0.1309, over 1682397.11 frames. ], batch size: 70, lr: 2.61e-02, grad_scale: 16.0 2023-04-16 14:22:22,138 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.004e+02 4.283e+02 5.211e+02 6.279e+02 9.649e+02, threshold=1.042e+03, percent-clipped=2.0 2023-04-16 14:22:25,101 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7816, 4.0775, 2.6635, 4.2586, 3.7891, 2.1818, 3.4098, 2.6287], device='cuda:0'), covar=tensor([0.0301, 0.0310, 0.1446, 0.0106, 0.0326, 0.1627, 0.0805, 0.1863], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0100, 0.0159, 0.0065, 0.0096, 0.0143, 0.0123, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-04-16 14:23:02,406 INFO [train.py:893] (0/4) Epoch 5, batch 250, loss[loss=0.2999, simple_loss=0.3339, pruned_loss=0.133, over 13522.00 frames. ], tot_loss[loss=0.2881, simple_loss=0.3144, pruned_loss=0.1309, over 1896229.52 frames. ], batch size: 91, lr: 2.61e-02, grad_scale: 16.0 2023-04-16 14:23:20,302 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11065.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:23:22,821 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6116, 2.6548, 1.8465, 3.3963, 3.9068, 3.0052, 3.9016, 3.5429], device='cuda:0'), covar=tensor([0.0071, 0.0404, 0.0726, 0.0089, 0.0051, 0.0247, 0.0052, 0.0071], device='cuda:0'), in_proj_covar=tensor([0.0037, 0.0048, 0.0061, 0.0044, 0.0030, 0.0046, 0.0028, 0.0035], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 14:23:50,387 INFO [train.py:893] (0/4) Epoch 5, batch 300, loss[loss=0.3101, simple_loss=0.3369, pruned_loss=0.1417, over 13381.00 frames. ], tot_loss[loss=0.2864, simple_loss=0.3135, pruned_loss=0.1296, over 2063178.44 frames. ], batch size: 113, lr: 2.60e-02, grad_scale: 16.0 2023-04-16 14:23:53,154 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11100.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:23:56,164 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.454e+02 3.650e+02 4.425e+02 5.854e+02 1.368e+03, threshold=8.851e+02, percent-clipped=2.0 2023-04-16 14:24:02,072 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11110.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 14:24:04,579 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11113.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:24:20,976 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11130.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:24:23,542 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1463, 4.5388, 4.0193, 4.1417, 4.3454, 4.7546, 4.4288, 4.3755], device='cuda:0'), covar=tensor([0.0253, 0.0232, 0.0327, 0.1132, 0.0183, 0.0220, 0.0237, 0.0202], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0081, 0.0083, 0.0149, 0.0083, 0.0096, 0.0081, 0.0075], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 14:24:36,580 INFO [train.py:893] (0/4) Epoch 5, batch 350, loss[loss=0.2907, simple_loss=0.32, pruned_loss=0.1307, over 13379.00 frames. ], tot_loss[loss=0.2869, simple_loss=0.314, pruned_loss=0.1299, over 2195316.79 frames. ], batch size: 109, lr: 2.60e-02, grad_scale: 16.0 2023-04-16 14:24:46,750 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11158.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:25:01,567 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11174.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:25:12,878 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6770, 2.3553, 1.6990, 1.5537, 1.0222, 1.8515, 1.7275, 2.3991], device='cuda:0'), covar=tensor([0.0602, 0.0395, 0.1050, 0.1204, 0.0311, 0.0276, 0.0684, 0.0404], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0065, 0.0059, 0.0096, 0.0060, 0.0061, 0.0077, 0.0059], device='cuda:0'), out_proj_covar=tensor([6.1877e-05, 5.2385e-05, 5.4170e-05, 8.8337e-05, 6.0003e-05, 4.9514e-05, 6.5571e-05, 4.6836e-05], device='cuda:0') 2023-04-16 14:25:22,622 INFO [train.py:893] (0/4) Epoch 5, batch 400, loss[loss=0.3339, simple_loss=0.3524, pruned_loss=0.1578, over 11400.00 frames. ], tot_loss[loss=0.2874, simple_loss=0.3151, pruned_loss=0.1299, over 2300820.48 frames. ], batch size: 157, lr: 2.59e-02, grad_scale: 16.0 2023-04-16 14:25:29,549 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.497e+02 3.829e+02 4.791e+02 5.945e+02 1.287e+03, threshold=9.581e+02, percent-clipped=2.0 2023-04-16 14:25:58,561 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11235.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:26:04,320 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11241.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 14:26:09,825 INFO [train.py:893] (0/4) Epoch 5, batch 450, loss[loss=0.2903, simple_loss=0.3184, pruned_loss=0.1311, over 13529.00 frames. ], tot_loss[loss=0.2877, simple_loss=0.3159, pruned_loss=0.1298, over 2382663.87 frames. ], batch size: 76, lr: 2.59e-02, grad_scale: 16.0 2023-04-16 14:26:11,939 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0485, 3.9655, 3.6363, 3.2260, 3.2404, 2.1576, 4.1312, 2.6039], device='cuda:0'), covar=tensor([0.0836, 0.0225, 0.0304, 0.0642, 0.0369, 0.2150, 0.0115, 0.2137], device='cuda:0'), in_proj_covar=tensor([0.0237, 0.0183, 0.0190, 0.0218, 0.0171, 0.0226, 0.0134, 0.0277], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 14:26:27,226 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4198, 4.3645, 4.6782, 4.5250, 4.8142, 4.3396, 4.7942, 4.7712], device='cuda:0'), covar=tensor([0.0367, 0.0490, 0.0475, 0.0434, 0.0542, 0.0702, 0.0521, 0.0432], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0176, 0.0163, 0.0122, 0.0227, 0.0197, 0.0142, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0004, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 14:26:38,543 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 14:26:42,024 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1130, 4.6563, 4.6147, 4.4793, 4.2457, 4.5050, 5.0568, 4.5336], device='cuda:0'), covar=tensor([0.0780, 0.1001, 0.2452, 0.2906, 0.0874, 0.1437, 0.0909, 0.1072], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0215, 0.0292, 0.0298, 0.0157, 0.0233, 0.0268, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 14:26:57,074 INFO [train.py:893] (0/4) Epoch 5, batch 500, loss[loss=0.2593, simple_loss=0.2998, pruned_loss=0.1094, over 13534.00 frames. ], tot_loss[loss=0.2857, simple_loss=0.3148, pruned_loss=0.1283, over 2447941.20 frames. ], batch size: 87, lr: 2.58e-02, grad_scale: 16.0 2023-04-16 14:27:03,859 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.346e+02 3.644e+02 4.498e+02 5.635e+02 7.796e+02, threshold=8.995e+02, percent-clipped=0.0 2023-04-16 14:27:19,650 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3939, 3.5613, 2.5117, 4.0360, 4.8306, 3.5286, 4.7922, 4.3850], device='cuda:0'), covar=tensor([0.0077, 0.0284, 0.0723, 0.0093, 0.0068, 0.0260, 0.0066, 0.0063], device='cuda:0'), in_proj_covar=tensor([0.0037, 0.0048, 0.0061, 0.0044, 0.0029, 0.0047, 0.0027, 0.0036], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 14:27:44,999 INFO [train.py:893] (0/4) Epoch 5, batch 550, loss[loss=0.2448, simple_loss=0.2826, pruned_loss=0.1035, over 13532.00 frames. ], tot_loss[loss=0.2842, simple_loss=0.314, pruned_loss=0.1272, over 2497210.77 frames. ], batch size: 76, lr: 2.58e-02, grad_scale: 16.0 2023-04-16 14:28:22,386 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-16 14:28:31,863 INFO [train.py:893] (0/4) Epoch 5, batch 600, loss[loss=0.2749, simple_loss=0.3084, pruned_loss=0.1206, over 13531.00 frames. ], tot_loss[loss=0.2828, simple_loss=0.3123, pruned_loss=0.1266, over 2529646.98 frames. ], batch size: 91, lr: 2.57e-02, grad_scale: 16.0 2023-04-16 14:28:35,659 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11400.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:28:38,565 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.783e+02 4.017e+02 4.794e+02 6.254e+02 1.645e+03, threshold=9.587e+02, percent-clipped=3.0 2023-04-16 14:28:42,195 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.4345, 2.1902, 1.9432, 1.0997, 0.8372, 1.1923, 1.4483, 2.1010], device='cuda:0'), covar=tensor([0.0685, 0.0286, 0.0748, 0.1587, 0.0260, 0.0139, 0.0698, 0.0212], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0066, 0.0060, 0.0097, 0.0059, 0.0062, 0.0077, 0.0061], device='cuda:0'), out_proj_covar=tensor([6.3140e-05, 5.3332e-05, 5.5928e-05, 8.9085e-05, 5.8660e-05, 5.0167e-05, 6.6021e-05, 4.8652e-05], device='cuda:0') 2023-04-16 14:29:03,641 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11430.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:29:09,562 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9088, 3.0878, 2.7046, 3.1883, 2.5349, 1.3648, 3.0648, 1.4698], device='cuda:0'), covar=tensor([0.0408, 0.0574, 0.0327, 0.0286, 0.0751, 0.1631, 0.0858, 0.1449], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0063, 0.0076, 0.0068, 0.0091, 0.0109, 0.0064, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 14:29:13,831 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9909, 3.0273, 2.0762, 3.6774, 4.4165, 3.1475, 4.3783, 3.9038], device='cuda:0'), covar=tensor([0.0082, 0.0329, 0.0828, 0.0123, 0.0061, 0.0298, 0.0064, 0.0070], device='cuda:0'), in_proj_covar=tensor([0.0037, 0.0047, 0.0061, 0.0044, 0.0029, 0.0046, 0.0027, 0.0036], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 14:29:19,324 INFO [train.py:893] (0/4) Epoch 5, batch 650, loss[loss=0.2575, simple_loss=0.2952, pruned_loss=0.1099, over 13478.00 frames. ], tot_loss[loss=0.2811, simple_loss=0.3109, pruned_loss=0.1256, over 2557725.07 frames. ], batch size: 81, lr: 2.57e-02, grad_scale: 16.0 2023-04-16 14:29:20,303 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11448.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:29:48,617 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11478.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:30:05,798 INFO [train.py:893] (0/4) Epoch 5, batch 700, loss[loss=0.2881, simple_loss=0.3165, pruned_loss=0.1299, over 13237.00 frames. ], tot_loss[loss=0.2786, simple_loss=0.3089, pruned_loss=0.1242, over 2580388.50 frames. ], batch size: 117, lr: 2.56e-02, grad_scale: 16.0 2023-04-16 14:30:12,401 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.679e+02 3.843e+02 4.625e+02 6.070e+02 1.081e+03, threshold=9.250e+02, percent-clipped=1.0 2023-04-16 14:30:36,440 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11530.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:30:39,051 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11532.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:30:47,086 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11541.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 14:30:51,863 INFO [train.py:893] (0/4) Epoch 5, batch 750, loss[loss=0.2955, simple_loss=0.3218, pruned_loss=0.1346, over 13237.00 frames. ], tot_loss[loss=0.2806, simple_loss=0.3102, pruned_loss=0.1254, over 2601898.77 frames. ], batch size: 124, lr: 2.56e-02, grad_scale: 16.0 2023-04-16 14:30:53,937 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.17 vs. limit=5.0 2023-04-16 14:31:31,760 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11589.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 14:31:35,131 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11593.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:31:39,418 INFO [train.py:893] (0/4) Epoch 5, batch 800, loss[loss=0.2791, simple_loss=0.3108, pruned_loss=0.1237, over 13529.00 frames. ], tot_loss[loss=0.2816, simple_loss=0.3114, pruned_loss=0.1259, over 2615510.76 frames. ], batch size: 76, lr: 2.55e-02, grad_scale: 16.0 2023-04-16 14:31:46,338 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.973e+02 3.812e+02 4.919e+02 6.723e+02 1.204e+03, threshold=9.839e+02, percent-clipped=3.0 2023-04-16 14:32:15,354 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11635.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:32:26,062 INFO [train.py:893] (0/4) Epoch 5, batch 850, loss[loss=0.3075, simple_loss=0.336, pruned_loss=0.1395, over 13533.00 frames. ], tot_loss[loss=0.2825, simple_loss=0.3125, pruned_loss=0.1262, over 2628581.63 frames. ], batch size: 98, lr: 2.55e-02, grad_scale: 16.0 2023-04-16 14:32:46,559 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8357, 1.7955, 3.8841, 3.7038, 3.7805, 3.2105, 3.7320, 2.5681], device='cuda:0'), covar=tensor([0.2421, 0.2380, 0.0092, 0.0284, 0.0131, 0.0499, 0.0112, 0.1126], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0175, 0.0074, 0.0089, 0.0091, 0.0133, 0.0082, 0.0149], device='cuda:0'), out_proj_covar=tensor([1.9733e-04, 1.9438e-04, 8.7879e-05, 1.1047e-04, 1.1277e-04, 1.5366e-04, 1.0097e-04, 1.6945e-04], device='cuda:0') 2023-04-16 14:33:12,594 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11696.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:33:13,140 INFO [train.py:893] (0/4) Epoch 5, batch 900, loss[loss=0.3214, simple_loss=0.3359, pruned_loss=0.1534, over 13543.00 frames. ], tot_loss[loss=0.2815, simple_loss=0.3114, pruned_loss=0.1258, over 2638514.64 frames. ], batch size: 76, lr: 2.54e-02, grad_scale: 16.0 2023-04-16 14:33:15,729 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11699.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:33:19,659 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.389e+02 3.741e+02 4.900e+02 6.173e+02 9.470e+02, threshold=9.800e+02, percent-clipped=0.0 2023-04-16 14:33:45,852 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 14:33:59,690 INFO [train.py:893] (0/4) Epoch 5, batch 950, loss[loss=0.2761, simple_loss=0.2865, pruned_loss=0.1328, over 12782.00 frames. ], tot_loss[loss=0.2815, simple_loss=0.3107, pruned_loss=0.1261, over 2646012.31 frames. ], batch size: 52, lr: 2.54e-02, grad_scale: 16.0 2023-04-16 14:34:11,892 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11760.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:34:16,379 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-16 14:34:18,744 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-16 14:34:37,436 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0774, 4.0951, 3.5515, 3.1494, 3.1797, 2.2432, 4.0954, 2.4453], device='cuda:0'), covar=tensor([0.0838, 0.0198, 0.0325, 0.0644, 0.0387, 0.1928, 0.0142, 0.2256], device='cuda:0'), in_proj_covar=tensor([0.0234, 0.0188, 0.0193, 0.0218, 0.0170, 0.0223, 0.0139, 0.0275], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 14:34:46,140 INFO [train.py:893] (0/4) Epoch 5, batch 1000, loss[loss=0.2372, simple_loss=0.2816, pruned_loss=0.09641, over 13333.00 frames. ], tot_loss[loss=0.2781, simple_loss=0.3076, pruned_loss=0.1243, over 2650793.14 frames. ], batch size: 73, lr: 2.53e-02, grad_scale: 16.0 2023-04-16 14:34:53,073 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.519e+02 3.813e+02 4.790e+02 5.853e+02 1.056e+03, threshold=9.580e+02, percent-clipped=1.0 2023-04-16 14:35:05,205 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11816.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:35:08,116 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-16 14:35:17,697 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11830.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:35:33,459 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11846.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:35:33,989 INFO [train.py:893] (0/4) Epoch 5, batch 1050, loss[loss=0.2554, simple_loss=0.2975, pruned_loss=0.1067, over 13365.00 frames. ], tot_loss[loss=0.2753, simple_loss=0.3058, pruned_loss=0.1223, over 2655107.03 frames. ], batch size: 113, lr: 2.53e-02, grad_scale: 16.0 2023-04-16 14:36:02,938 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11877.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:36:03,629 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11878.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:36:12,634 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11888.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:36:20,694 INFO [train.py:893] (0/4) Epoch 5, batch 1100, loss[loss=0.2349, simple_loss=0.2696, pruned_loss=0.1001, over 13382.00 frames. ], tot_loss[loss=0.2748, simple_loss=0.3062, pruned_loss=0.1217, over 2657431.78 frames. ], batch size: 62, lr: 2.53e-02, grad_scale: 16.0 2023-04-16 14:36:27,224 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.277e+02 3.881e+02 4.569e+02 5.813e+02 1.527e+03, threshold=9.139e+02, percent-clipped=3.0 2023-04-16 14:36:30,919 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11907.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:36:42,662 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11920.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:37:07,307 INFO [train.py:893] (0/4) Epoch 5, batch 1150, loss[loss=0.2411, simple_loss=0.2842, pruned_loss=0.099, over 13530.00 frames. ], tot_loss[loss=0.2735, simple_loss=0.3059, pruned_loss=0.1206, over 2663574.25 frames. ], batch size: 72, lr: 2.52e-02, grad_scale: 16.0 2023-04-16 14:37:08,501 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11948.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:37:23,937 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5733, 4.5184, 3.9814, 3.4554, 3.4731, 2.6250, 4.6972, 2.9498], device='cuda:0'), covar=tensor([0.1043, 0.0199, 0.0446, 0.0864, 0.0536, 0.2272, 0.0145, 0.2530], device='cuda:0'), in_proj_covar=tensor([0.0235, 0.0190, 0.0193, 0.0218, 0.0171, 0.0222, 0.0136, 0.0275], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 14:37:38,851 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11981.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:37:41,934 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11984.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:37:48,210 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11991.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:37:53,631 INFO [train.py:893] (0/4) Epoch 5, batch 1200, loss[loss=0.2857, simple_loss=0.319, pruned_loss=0.1262, over 13460.00 frames. ], tot_loss[loss=0.2734, simple_loss=0.3059, pruned_loss=0.1205, over 2660365.05 frames. ], batch size: 106, lr: 2.52e-02, grad_scale: 16.0 2023-04-16 14:37:56,434 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-12000.pt 2023-04-16 14:38:03,291 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.645e+02 3.919e+02 5.054e+02 6.149e+02 1.256e+03, threshold=1.011e+03, percent-clipped=6.0 2023-04-16 14:38:07,822 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12009.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:38:24,783 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 14:38:37,243 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 14:38:40,766 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12045.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:38:42,801 INFO [train.py:893] (0/4) Epoch 5, batch 1250, loss[loss=0.3076, simple_loss=0.3308, pruned_loss=0.1422, over 13374.00 frames. ], tot_loss[loss=0.2754, simple_loss=0.3074, pruned_loss=0.1217, over 2665396.38 frames. ], batch size: 113, lr: 2.51e-02, grad_scale: 16.0 2023-04-16 14:38:50,519 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12055.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:39:09,569 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12076.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:39:29,492 INFO [train.py:893] (0/4) Epoch 5, batch 1300, loss[loss=0.2856, simple_loss=0.3198, pruned_loss=0.1257, over 13463.00 frames. ], tot_loss[loss=0.2767, simple_loss=0.3086, pruned_loss=0.1224, over 2665790.55 frames. ], batch size: 103, lr: 2.51e-02, grad_scale: 16.0 2023-04-16 14:39:35,346 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.667e+02 3.833e+02 4.783e+02 5.972e+02 1.404e+03, threshold=9.567e+02, percent-clipped=5.0 2023-04-16 14:39:59,088 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.4635, 1.9948, 1.5358, 2.3453, 1.6571, 1.7272, 2.1855, 2.3075], device='cuda:0'), covar=tensor([0.0085, 0.0197, 0.0232, 0.0141, 0.0143, 0.0118, 0.0243, 0.0155], device='cuda:0'), in_proj_covar=tensor([0.0042, 0.0042, 0.0046, 0.0036, 0.0044, 0.0039, 0.0046, 0.0045], device='cuda:0'), out_proj_covar=tensor([4.9766e-05, 5.1650e-05, 5.8672e-05, 4.5438e-05, 5.3309e-05, 4.6168e-05, 5.8476e-05, 5.5635e-05], device='cuda:0') 2023-04-16 14:40:02,366 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2023-04-16 14:40:06,437 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12137.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:40:15,966 INFO [train.py:893] (0/4) Epoch 5, batch 1350, loss[loss=0.2364, simple_loss=0.2694, pruned_loss=0.1017, over 13426.00 frames. ], tot_loss[loss=0.2773, simple_loss=0.3092, pruned_loss=0.1227, over 2668010.52 frames. ], batch size: 65, lr: 2.50e-02, grad_scale: 16.0 2023-04-16 14:40:36,974 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.22 vs. limit=5.0 2023-04-16 14:40:39,055 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12172.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:40:48,688 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2023-04-16 14:40:54,621 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12188.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:41:01,904 INFO [train.py:893] (0/4) Epoch 5, batch 1400, loss[loss=0.2703, simple_loss=0.2885, pruned_loss=0.1261, over 13415.00 frames. ], tot_loss[loss=0.2759, simple_loss=0.3076, pruned_loss=0.1221, over 2670246.32 frames. ], batch size: 65, lr: 2.50e-02, grad_scale: 16.0 2023-04-16 14:41:06,592 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12202.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:41:08,833 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.680e+02 3.684e+02 4.695e+02 6.104e+02 1.016e+03, threshold=9.390e+02, percent-clipped=1.0 2023-04-16 14:41:37,554 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-16 14:41:39,433 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12236.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:41:49,229 INFO [train.py:893] (0/4) Epoch 5, batch 1450, loss[loss=0.2807, simple_loss=0.3098, pruned_loss=0.1258, over 13544.00 frames. ], tot_loss[loss=0.2762, simple_loss=0.3077, pruned_loss=0.1224, over 2670454.35 frames. ], batch size: 87, lr: 2.49e-02, grad_scale: 32.0 2023-04-16 14:42:16,353 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12276.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:42:30,913 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12291.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:42:36,229 INFO [train.py:893] (0/4) Epoch 5, batch 1500, loss[loss=0.2549, simple_loss=0.2952, pruned_loss=0.1073, over 13429.00 frames. ], tot_loss[loss=0.2736, simple_loss=0.3058, pruned_loss=0.1207, over 2668509.75 frames. ], batch size: 88, lr: 2.49e-02, grad_scale: 32.0 2023-04-16 14:42:42,833 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12304.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:42:43,471 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.442e+02 3.651e+02 4.300e+02 5.550e+02 1.141e+03, threshold=8.601e+02, percent-clipped=1.0 2023-04-16 14:42:53,902 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0151, 1.7789, 1.2985, 1.8245, 1.2464, 1.4825, 1.7715, 1.8978], device='cuda:0'), covar=tensor([0.0133, 0.0213, 0.0213, 0.0177, 0.0162, 0.0191, 0.0246, 0.0127], device='cuda:0'), in_proj_covar=tensor([0.0043, 0.0042, 0.0046, 0.0038, 0.0046, 0.0039, 0.0045, 0.0043], device='cuda:0'), out_proj_covar=tensor([5.1701e-05, 5.1090e-05, 5.8663e-05, 4.7642e-05, 5.5773e-05, 4.6849e-05, 5.7674e-05, 5.4244e-05], device='cuda:0') 2023-04-16 14:42:54,141 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-16 14:42:59,486 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9471, 4.8191, 5.1548, 4.8562, 5.2755, 4.8298, 5.2905, 5.2875], device='cuda:0'), covar=tensor([0.0272, 0.0429, 0.0403, 0.0365, 0.0484, 0.0602, 0.0376, 0.0368], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0179, 0.0163, 0.0122, 0.0232, 0.0196, 0.0144, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0004, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 14:43:15,458 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12339.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:43:16,965 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12340.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:43:23,425 INFO [train.py:893] (0/4) Epoch 5, batch 1550, loss[loss=0.2834, simple_loss=0.2996, pruned_loss=0.1336, over 11858.00 frames. ], tot_loss[loss=0.2737, simple_loss=0.306, pruned_loss=0.1207, over 2667093.97 frames. ], batch size: 157, lr: 2.49e-02, grad_scale: 16.0 2023-04-16 14:43:30,243 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12355.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:44:02,811 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.21 vs. limit=5.0 2023-04-16 14:44:09,638 INFO [train.py:893] (0/4) Epoch 5, batch 1600, loss[loss=0.3046, simple_loss=0.3258, pruned_loss=0.1417, over 11870.00 frames. ], tot_loss[loss=0.2731, simple_loss=0.3055, pruned_loss=0.1203, over 2655626.57 frames. ], batch size: 157, lr: 2.48e-02, grad_scale: 16.0 2023-04-16 14:44:15,393 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12403.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:44:17,654 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.771e+02 3.931e+02 5.417e+02 6.159e+02 1.328e+03, threshold=1.083e+03, percent-clipped=6.0 2023-04-16 14:44:43,297 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12432.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:44:57,068 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12446.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:44:57,542 INFO [train.py:893] (0/4) Epoch 5, batch 1650, loss[loss=0.2882, simple_loss=0.3184, pruned_loss=0.1289, over 13529.00 frames. ], tot_loss[loss=0.2732, simple_loss=0.3064, pruned_loss=0.12, over 2650766.81 frames. ], batch size: 87, lr: 2.48e-02, grad_scale: 16.0 2023-04-16 14:44:57,985 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.66 vs. limit=5.0 2023-04-16 14:45:21,481 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12472.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:45:44,635 INFO [train.py:893] (0/4) Epoch 5, batch 1700, loss[loss=0.2734, simple_loss=0.3103, pruned_loss=0.1183, over 13325.00 frames. ], tot_loss[loss=0.2746, simple_loss=0.3076, pruned_loss=0.1208, over 2650536.62 frames. ], batch size: 118, lr: 2.47e-02, grad_scale: 16.0 2023-04-16 14:45:49,669 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12502.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:45:51,910 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.574e+02 4.089e+02 5.133e+02 6.937e+02 1.880e+03, threshold=1.027e+03, percent-clipped=5.0 2023-04-16 14:45:53,856 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12507.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:46:06,221 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12520.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:46:31,419 INFO [train.py:893] (0/4) Epoch 5, batch 1750, loss[loss=0.273, simple_loss=0.3032, pruned_loss=0.1214, over 13533.00 frames. ], tot_loss[loss=0.2726, simple_loss=0.3061, pruned_loss=0.1196, over 2655832.93 frames. ], batch size: 83, lr: 2.47e-02, grad_scale: 16.0 2023-04-16 14:46:34,156 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12550.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:46:58,547 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12576.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:47:12,718 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-04-16 14:47:18,892 INFO [train.py:893] (0/4) Epoch 5, batch 1800, loss[loss=0.2705, simple_loss=0.3076, pruned_loss=0.1167, over 13451.00 frames. ], tot_loss[loss=0.2705, simple_loss=0.3047, pruned_loss=0.1182, over 2659645.96 frames. ], batch size: 106, lr: 2.46e-02, grad_scale: 16.0 2023-04-16 14:47:22,032 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7484, 3.4344, 3.6972, 2.5861, 4.1137, 3.8708, 3.8018, 4.2474], device='cuda:0'), covar=tensor([0.0196, 0.0161, 0.0137, 0.0810, 0.0148, 0.0135, 0.0138, 0.0085], device='cuda:0'), in_proj_covar=tensor([0.0052, 0.0028, 0.0045, 0.0069, 0.0055, 0.0047, 0.0046, 0.0036], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:0') 2023-04-16 14:47:25,305 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12604.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:47:25,957 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.081e+02 3.697e+02 4.449e+02 5.454e+02 1.218e+03, threshold=8.898e+02, percent-clipped=2.0 2023-04-16 14:47:44,590 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12624.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:47:58,721 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12640.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:48:05,088 INFO [train.py:893] (0/4) Epoch 5, batch 1850, loss[loss=0.2346, simple_loss=0.2789, pruned_loss=0.09513, over 13533.00 frames. ], tot_loss[loss=0.2697, simple_loss=0.3041, pruned_loss=0.1177, over 2660474.26 frames. ], batch size: 70, lr: 2.46e-02, grad_scale: 16.0 2023-04-16 14:48:09,220 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 14:48:11,157 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12652.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:48:18,830 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9834, 4.7518, 5.0727, 4.8312, 5.3132, 4.6729, 5.3613, 5.3378], device='cuda:0'), covar=tensor([0.0303, 0.0445, 0.0482, 0.0433, 0.0508, 0.0703, 0.0377, 0.0366], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0177, 0.0163, 0.0120, 0.0230, 0.0194, 0.0141, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 14:48:44,543 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12688.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:48:52,018 INFO [train.py:893] (0/4) Epoch 5, batch 1900, loss[loss=0.2722, simple_loss=0.3014, pruned_loss=0.1215, over 13419.00 frames. ], tot_loss[loss=0.2687, simple_loss=0.3029, pruned_loss=0.1173, over 2662926.07 frames. ], batch size: 95, lr: 2.46e-02, grad_scale: 16.0 2023-04-16 14:48:59,625 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.759e+02 3.839e+02 4.372e+02 5.320e+02 1.307e+03, threshold=8.743e+02, percent-clipped=2.0 2023-04-16 14:49:24,676 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12732.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:49:26,166 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.5317, 5.0593, 5.0017, 4.8449, 4.5980, 4.8271, 5.4659, 4.8285], device='cuda:0'), covar=tensor([0.0709, 0.0780, 0.2121, 0.2857, 0.0901, 0.1276, 0.0939, 0.1046], device='cuda:0'), in_proj_covar=tensor([0.0217, 0.0236, 0.0315, 0.0316, 0.0168, 0.0243, 0.0284, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 14:49:38,019 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5068, 4.4108, 3.9809, 3.4489, 3.7443, 2.4820, 4.7201, 2.7636], device='cuda:0'), covar=tensor([0.0819, 0.0187, 0.0330, 0.0632, 0.0302, 0.1771, 0.0096, 0.2180], device='cuda:0'), in_proj_covar=tensor([0.0239, 0.0198, 0.0203, 0.0223, 0.0176, 0.0228, 0.0143, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 14:49:40,115 INFO [train.py:893] (0/4) Epoch 5, batch 1950, loss[loss=0.2171, simple_loss=0.2538, pruned_loss=0.09015, over 12822.00 frames. ], tot_loss[loss=0.268, simple_loss=0.3024, pruned_loss=0.1168, over 2668044.48 frames. ], batch size: 52, lr: 2.45e-02, grad_scale: 16.0 2023-04-16 14:49:46,381 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6236, 4.3608, 4.3016, 3.3728, 2.6928, 3.3299, 4.4972, 4.5852], device='cuda:0'), covar=tensor([0.0841, 0.0329, 0.0279, 0.1087, 0.1450, 0.0916, 0.0127, 0.0061], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0126, 0.0129, 0.0185, 0.0183, 0.0153, 0.0104, 0.0087], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-04-16 14:49:53,785 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12763.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:50:10,897 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12780.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:50:25,727 INFO [train.py:893] (0/4) Epoch 5, batch 2000, loss[loss=0.2862, simple_loss=0.3228, pruned_loss=0.1248, over 13383.00 frames. ], tot_loss[loss=0.2717, simple_loss=0.3054, pruned_loss=0.119, over 2666149.99 frames. ], batch size: 109, lr: 2.45e-02, grad_scale: 16.0 2023-04-16 14:50:30,269 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12802.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:50:32,509 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 14:50:33,264 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.676e+02 3.883e+02 4.614e+02 5.841e+02 1.269e+03, threshold=9.228e+02, percent-clipped=6.0 2023-04-16 14:50:51,116 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12824.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 14:51:12,436 INFO [train.py:893] (0/4) Epoch 5, batch 2050, loss[loss=0.2578, simple_loss=0.302, pruned_loss=0.1067, over 13542.00 frames. ], tot_loss[loss=0.2746, simple_loss=0.308, pruned_loss=0.1206, over 2664811.42 frames. ], batch size: 83, lr: 2.44e-02, grad_scale: 16.0 2023-04-16 14:51:57,020 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2023-04-16 14:51:57,937 INFO [train.py:893] (0/4) Epoch 5, batch 2100, loss[loss=0.2726, simple_loss=0.3013, pruned_loss=0.1219, over 13381.00 frames. ], tot_loss[loss=0.2736, simple_loss=0.3071, pruned_loss=0.1201, over 2665482.60 frames. ], batch size: 118, lr: 2.44e-02, grad_scale: 16.0 2023-04-16 14:52:05,132 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2023-04-16 14:52:05,974 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 3.926e+02 4.592e+02 5.677e+02 1.574e+03, threshold=9.184e+02, percent-clipped=7.0 2023-04-16 14:52:18,866 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2148, 4.3433, 2.9379, 4.5878, 4.0884, 2.4310, 3.6946, 2.7455], device='cuda:0'), covar=tensor([0.0232, 0.0306, 0.1318, 0.0061, 0.0233, 0.1499, 0.0577, 0.1777], device='cuda:0'), in_proj_covar=tensor([0.0109, 0.0107, 0.0166, 0.0075, 0.0100, 0.0152, 0.0134, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 14:52:42,742 INFO [train.py:893] (0/4) Epoch 5, batch 2150, loss[loss=0.2616, simple_loss=0.2958, pruned_loss=0.1137, over 13072.00 frames. ], tot_loss[loss=0.2726, simple_loss=0.3067, pruned_loss=0.1193, over 2662453.20 frames. ], batch size: 142, lr: 2.44e-02, grad_scale: 16.0 2023-04-16 14:53:01,115 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7822, 4.0305, 3.8499, 3.7341, 3.7639, 3.6378, 4.0986, 4.1049], device='cuda:0'), covar=tensor([0.0209, 0.0250, 0.0224, 0.0319, 0.0311, 0.0301, 0.0239, 0.0235], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0134, 0.0099, 0.0127, 0.0090, 0.0126, 0.0093, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 14:53:22,249 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-16 14:53:29,483 INFO [train.py:893] (0/4) Epoch 5, batch 2200, loss[loss=0.2175, simple_loss=0.2574, pruned_loss=0.08878, over 12719.00 frames. ], tot_loss[loss=0.27, simple_loss=0.3046, pruned_loss=0.1177, over 2657535.38 frames. ], batch size: 52, lr: 2.43e-02, grad_scale: 16.0 2023-04-16 14:53:36,948 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.620e+02 3.386e+02 4.268e+02 5.257e+02 8.997e+02, threshold=8.536e+02, percent-clipped=0.0 2023-04-16 14:54:15,611 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.29 vs. limit=5.0 2023-04-16 14:54:15,908 INFO [train.py:893] (0/4) Epoch 5, batch 2250, loss[loss=0.2305, simple_loss=0.2671, pruned_loss=0.09693, over 13178.00 frames. ], tot_loss[loss=0.2681, simple_loss=0.3027, pruned_loss=0.1168, over 2655054.07 frames. ], batch size: 58, lr: 2.43e-02, grad_scale: 16.0 2023-04-16 14:55:03,687 INFO [train.py:893] (0/4) Epoch 5, batch 2300, loss[loss=0.2721, simple_loss=0.3025, pruned_loss=0.1209, over 13453.00 frames. ], tot_loss[loss=0.2671, simple_loss=0.3019, pruned_loss=0.1161, over 2659841.73 frames. ], batch size: 106, lr: 2.42e-02, grad_scale: 16.0 2023-04-16 14:55:08,177 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=13102.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:55:11,125 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.210e+02 3.671e+02 4.489e+02 5.410e+02 8.963e+02, threshold=8.978e+02, percent-clipped=2.0 2023-04-16 14:55:19,894 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7608, 1.7115, 3.9328, 3.6916, 3.6421, 2.8163, 3.5843, 2.6642], device='cuda:0'), covar=tensor([0.2296, 0.2663, 0.0069, 0.0156, 0.0219, 0.0707, 0.0138, 0.1068], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0178, 0.0079, 0.0091, 0.0092, 0.0136, 0.0090, 0.0147], device='cuda:0'), out_proj_covar=tensor([1.9905e-04, 2.0283e-04, 9.4702e-05, 1.1224e-04, 1.1538e-04, 1.6067e-04, 1.1256e-04, 1.7105e-04], device='cuda:0') 2023-04-16 14:55:24,531 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=13119.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 14:55:30,421 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8775, 4.3234, 2.5543, 4.3703, 4.0713, 1.6804, 3.4255, 2.7470], device='cuda:0'), covar=tensor([0.0334, 0.0304, 0.1833, 0.0102, 0.0244, 0.2211, 0.0731, 0.1985], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0107, 0.0164, 0.0076, 0.0100, 0.0152, 0.0134, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 14:55:48,796 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6268, 4.1645, 4.2658, 3.2324, 2.5210, 3.1642, 4.3750, 4.4732], device='cuda:0'), covar=tensor([0.0720, 0.0440, 0.0232, 0.1011, 0.1591, 0.0964, 0.0130, 0.0080], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0129, 0.0132, 0.0182, 0.0186, 0.0152, 0.0105, 0.0088], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:0') 2023-04-16 14:55:49,229 INFO [train.py:893] (0/4) Epoch 5, batch 2350, loss[loss=0.3023, simple_loss=0.3231, pruned_loss=0.1408, over 13539.00 frames. ], tot_loss[loss=0.2659, simple_loss=0.3009, pruned_loss=0.1154, over 2659701.20 frames. ], batch size: 85, lr: 2.42e-02, grad_scale: 16.0 2023-04-16 14:55:52,646 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=13150.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 14:56:13,246 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 14:56:18,615 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.69 vs. limit=5.0 2023-04-16 14:56:36,313 INFO [train.py:893] (0/4) Epoch 5, batch 2400, loss[loss=0.2585, simple_loss=0.2929, pruned_loss=0.1121, over 13432.00 frames. ], tot_loss[loss=0.2657, simple_loss=0.3005, pruned_loss=0.1154, over 2661060.92 frames. ], batch size: 65, lr: 2.42e-02, grad_scale: 16.0 2023-04-16 14:56:43,920 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.422e+02 3.773e+02 4.446e+02 5.529e+02 1.176e+03, threshold=8.892e+02, percent-clipped=3.0 2023-04-16 14:57:23,069 INFO [train.py:893] (0/4) Epoch 5, batch 2450, loss[loss=0.2671, simple_loss=0.2994, pruned_loss=0.1174, over 13531.00 frames. ], tot_loss[loss=0.266, simple_loss=0.3007, pruned_loss=0.1157, over 2663221.74 frames. ], batch size: 72, lr: 2.41e-02, grad_scale: 16.0 2023-04-16 14:58:09,766 INFO [train.py:893] (0/4) Epoch 5, batch 2500, loss[loss=0.2974, simple_loss=0.3263, pruned_loss=0.1343, over 13534.00 frames. ], tot_loss[loss=0.2664, simple_loss=0.3009, pruned_loss=0.1159, over 2665381.17 frames. ], batch size: 83, lr: 2.41e-02, grad_scale: 16.0 2023-04-16 14:58:17,221 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.475e+02 3.839e+02 4.684e+02 5.527e+02 8.923e+02, threshold=9.368e+02, percent-clipped=1.0 2023-04-16 14:58:57,640 INFO [train.py:893] (0/4) Epoch 5, batch 2550, loss[loss=0.2312, simple_loss=0.2671, pruned_loss=0.09764, over 13418.00 frames. ], tot_loss[loss=0.267, simple_loss=0.3019, pruned_loss=0.1161, over 2667757.89 frames. ], batch size: 65, lr: 2.40e-02, grad_scale: 16.0 2023-04-16 14:59:22,415 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 14:59:43,136 INFO [train.py:893] (0/4) Epoch 5, batch 2600, loss[loss=0.2759, simple_loss=0.3108, pruned_loss=0.1205, over 13429.00 frames. ], tot_loss[loss=0.2682, simple_loss=0.3026, pruned_loss=0.1169, over 2670597.53 frames. ], batch size: 95, lr: 2.40e-02, grad_scale: 16.0 2023-04-16 14:59:50,061 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.342e+02 4.063e+02 4.946e+02 6.470e+02 1.543e+03, threshold=9.893e+02, percent-clipped=8.0 2023-04-16 15:00:02,546 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5646, 2.1442, 2.6372, 3.8620, 3.5028, 3.9002, 2.9079, 2.0995], device='cuda:0'), covar=tensor([0.0303, 0.1469, 0.0984, 0.0067, 0.0291, 0.0053, 0.0800, 0.1521], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0147, 0.0136, 0.0075, 0.0078, 0.0066, 0.0129, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:00:04,208 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=13419.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:00:25,854 INFO [train.py:893] (0/4) Epoch 5, batch 2650, loss[loss=0.3345, simple_loss=0.3486, pruned_loss=0.1602, over 13087.00 frames. ], tot_loss[loss=0.2694, simple_loss=0.3033, pruned_loss=0.1177, over 2666689.87 frames. ], batch size: 142, lr: 2.40e-02, grad_scale: 16.0 2023-04-16 15:00:28,585 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-16 15:00:40,993 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=13467.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:01:04,066 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-5.pt 2023-04-16 15:01:28,496 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 15:01:32,173 INFO [train.py:893] (0/4) Epoch 6, batch 0, loss[loss=0.2591, simple_loss=0.2977, pruned_loss=0.1103, over 13469.00 frames. ], tot_loss[loss=0.2591, simple_loss=0.2977, pruned_loss=0.1103, over 13469.00 frames. ], batch size: 79, lr: 2.23e-02, grad_scale: 16.0 2023-04-16 15:01:32,174 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 15:01:54,542 INFO [train.py:927] (0/4) Epoch 6, validation: loss=0.195, simple_loss=0.2424, pruned_loss=0.07384, over 2446609.00 frames. 2023-04-16 15:01:54,542 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 15:02:01,934 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-16 15:02:02,879 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.472e+02 3.959e+02 4.690e+02 5.704e+02 9.254e+02, threshold=9.380e+02, percent-clipped=0.0 2023-04-16 15:02:05,080 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.58 vs. limit=5.0 2023-04-16 15:02:21,059 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9202, 4.7333, 5.0850, 4.7378, 5.2619, 4.6132, 5.2920, 5.2131], device='cuda:0'), covar=tensor([0.0301, 0.0469, 0.0475, 0.0502, 0.0484, 0.0640, 0.0422, 0.0424], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0189, 0.0171, 0.0131, 0.0245, 0.0209, 0.0148, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0004, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:02:24,025 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2023-04-16 15:02:34,285 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9988, 1.9046, 1.2364, 1.9991, 1.3110, 1.7356, 1.7384, 1.9680], device='cuda:0'), covar=tensor([0.0065, 0.0143, 0.0149, 0.0097, 0.0141, 0.0127, 0.0217, 0.0120], device='cuda:0'), in_proj_covar=tensor([0.0045, 0.0045, 0.0050, 0.0038, 0.0051, 0.0040, 0.0049, 0.0045], device='cuda:0'), out_proj_covar=tensor([5.3830e-05, 5.6514e-05, 6.4026e-05, 4.7954e-05, 6.1618e-05, 4.7756e-05, 6.2917e-05, 5.6301e-05], device='cuda:0') 2023-04-16 15:02:41,153 INFO [train.py:893] (0/4) Epoch 6, batch 50, loss[loss=0.2588, simple_loss=0.2941, pruned_loss=0.1118, over 13339.00 frames. ], tot_loss[loss=0.2638, simple_loss=0.2955, pruned_loss=0.116, over 596381.72 frames. ], batch size: 118, lr: 2.23e-02, grad_scale: 16.0 2023-04-16 15:03:05,485 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 15:03:05,485 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 15:03:05,485 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 15:03:05,500 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 15:03:05,508 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 15:03:05,528 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 15:03:05,537 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 15:03:14,746 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7789, 2.2853, 2.0023, 1.4613, 1.0733, 1.9241, 1.6909, 2.5450], device='cuda:0'), covar=tensor([0.0640, 0.0324, 0.0991, 0.1306, 0.0264, 0.0264, 0.0629, 0.0230], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0084, 0.0073, 0.0117, 0.0066, 0.0077, 0.0092, 0.0071], device='cuda:0'), out_proj_covar=tensor([7.6118e-05, 6.5750e-05, 6.4695e-05, 1.0324e-04, 6.3217e-05, 6.1166e-05, 7.5712e-05, 5.4580e-05], device='cuda:0') 2023-04-16 15:03:26,048 INFO [train.py:893] (0/4) Epoch 6, batch 100, loss[loss=0.2439, simple_loss=0.2802, pruned_loss=0.1038, over 13362.00 frames. ], tot_loss[loss=0.2669, simple_loss=0.2982, pruned_loss=0.1178, over 1054576.52 frames. ], batch size: 73, lr: 2.22e-02, grad_scale: 16.0 2023-04-16 15:03:35,310 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.691e+02 3.972e+02 4.871e+02 6.380e+02 1.409e+03, threshold=9.742e+02, percent-clipped=4.0 2023-04-16 15:04:12,844 INFO [train.py:893] (0/4) Epoch 6, batch 150, loss[loss=0.288, simple_loss=0.3153, pruned_loss=0.1303, over 13527.00 frames. ], tot_loss[loss=0.2704, simple_loss=0.301, pruned_loss=0.1199, over 1407022.45 frames. ], batch size: 83, lr: 2.22e-02, grad_scale: 16.0 2023-04-16 15:04:47,950 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1479, 4.6632, 4.4242, 4.4767, 4.4335, 4.5137, 5.0612, 4.5890], device='cuda:0'), covar=tensor([0.0669, 0.0828, 0.2144, 0.2584, 0.0620, 0.1269, 0.0909, 0.0758], device='cuda:0'), in_proj_covar=tensor([0.0216, 0.0235, 0.0314, 0.0315, 0.0166, 0.0241, 0.0281, 0.0209], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 15:04:54,710 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5443, 3.3092, 3.6039, 2.6247, 3.7337, 3.4785, 3.5119, 3.8584], device='cuda:0'), covar=tensor([0.0157, 0.0131, 0.0102, 0.0736, 0.0124, 0.0142, 0.0126, 0.0064], device='cuda:0'), in_proj_covar=tensor([0.0056, 0.0031, 0.0048, 0.0074, 0.0061, 0.0051, 0.0049, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:04:58,316 INFO [train.py:893] (0/4) Epoch 6, batch 200, loss[loss=0.262, simple_loss=0.2947, pruned_loss=0.1146, over 13532.00 frames. ], tot_loss[loss=0.2716, simple_loss=0.3023, pruned_loss=0.1204, over 1687124.98 frames. ], batch size: 85, lr: 2.22e-02, grad_scale: 16.0 2023-04-16 15:05:06,627 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.332e+02 3.880e+02 4.687e+02 5.534e+02 1.302e+03, threshold=9.374e+02, percent-clipped=1.0 2023-04-16 15:05:11,870 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7426, 3.4698, 3.7751, 2.6085, 4.1801, 3.6893, 3.7145, 4.2038], device='cuda:0'), covar=tensor([0.0199, 0.0169, 0.0147, 0.0944, 0.0135, 0.0172, 0.0164, 0.0072], device='cuda:0'), in_proj_covar=tensor([0.0057, 0.0032, 0.0049, 0.0075, 0.0062, 0.0051, 0.0050, 0.0039], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:05:36,653 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=13737.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:05:41,518 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5551, 3.7515, 2.5733, 3.7962, 3.5335, 2.0223, 3.0251, 2.4987], device='cuda:0'), covar=tensor([0.0259, 0.0254, 0.1181, 0.0107, 0.0265, 0.1317, 0.0680, 0.1521], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0111, 0.0166, 0.0077, 0.0108, 0.0152, 0.0136, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 15:05:44,470 INFO [train.py:893] (0/4) Epoch 6, batch 250, loss[loss=0.2572, simple_loss=0.2878, pruned_loss=0.1133, over 13356.00 frames. ], tot_loss[loss=0.2718, simple_loss=0.3026, pruned_loss=0.1205, over 1900579.22 frames. ], batch size: 73, lr: 2.21e-02, grad_scale: 16.0 2023-04-16 15:06:05,858 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3386, 2.1557, 1.5206, 2.4085, 1.5669, 2.2081, 2.3459, 2.2340], device='cuda:0'), covar=tensor([0.0097, 0.0189, 0.0169, 0.0085, 0.0149, 0.0082, 0.0173, 0.0139], device='cuda:0'), in_proj_covar=tensor([0.0044, 0.0045, 0.0050, 0.0037, 0.0050, 0.0040, 0.0047, 0.0045], device='cuda:0'), out_proj_covar=tensor([5.3478e-05, 5.5381e-05, 6.3465e-05, 4.6104e-05, 6.0143e-05, 4.7326e-05, 6.0407e-05, 5.6524e-05], device='cuda:0') 2023-04-16 15:06:08,260 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1303, 3.5104, 3.6492, 2.7535, 2.3334, 2.8611, 3.7787, 3.8846], device='cuda:0'), covar=tensor([0.0622, 0.0508, 0.0264, 0.0922, 0.1372, 0.0845, 0.0161, 0.0086], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0133, 0.0139, 0.0185, 0.0189, 0.0154, 0.0109, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0001], device='cuda:0') 2023-04-16 15:06:09,495 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2023-04-16 15:06:29,314 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5901, 2.2968, 2.7243, 3.8142, 3.5779, 3.7515, 2.9839, 1.9841], device='cuda:0'), covar=tensor([0.0139, 0.1250, 0.0738, 0.0063, 0.0193, 0.0065, 0.0602, 0.1443], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0143, 0.0132, 0.0074, 0.0077, 0.0066, 0.0127, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:06:30,652 INFO [train.py:893] (0/4) Epoch 6, batch 300, loss[loss=0.2597, simple_loss=0.2872, pruned_loss=0.1161, over 13420.00 frames. ], tot_loss[loss=0.2703, simple_loss=0.3021, pruned_loss=0.1192, over 2067684.36 frames. ], batch size: 65, lr: 2.21e-02, grad_scale: 16.0 2023-04-16 15:06:32,535 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=13798.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 15:06:38,900 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.402e+02 3.920e+02 4.915e+02 6.095e+02 1.429e+03, threshold=9.831e+02, percent-clipped=4.0 2023-04-16 15:07:16,567 INFO [train.py:893] (0/4) Epoch 6, batch 350, loss[loss=0.2698, simple_loss=0.3043, pruned_loss=0.1177, over 13526.00 frames. ], tot_loss[loss=0.2699, simple_loss=0.3022, pruned_loss=0.1188, over 2196625.80 frames. ], batch size: 85, lr: 2.21e-02, grad_scale: 16.0 2023-04-16 15:08:02,696 INFO [train.py:893] (0/4) Epoch 6, batch 400, loss[loss=0.3039, simple_loss=0.3354, pruned_loss=0.1362, over 13440.00 frames. ], tot_loss[loss=0.2691, simple_loss=0.3019, pruned_loss=0.1182, over 2298745.91 frames. ], batch size: 95, lr: 2.20e-02, grad_scale: 16.0 2023-04-16 15:08:10,846 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.241e+02 3.793e+02 4.578e+02 5.807e+02 9.391e+02, threshold=9.156e+02, percent-clipped=0.0 2023-04-16 15:08:49,262 INFO [train.py:893] (0/4) Epoch 6, batch 450, loss[loss=0.2983, simple_loss=0.327, pruned_loss=0.1348, over 13457.00 frames. ], tot_loss[loss=0.2693, simple_loss=0.3023, pruned_loss=0.1181, over 2379039.59 frames. ], batch size: 100, lr: 2.20e-02, grad_scale: 16.0 2023-04-16 15:09:12,312 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 15:09:16,567 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7700, 2.0295, 1.7878, 1.2128, 1.2495, 1.9570, 1.4275, 2.3122], device='cuda:0'), covar=tensor([0.0475, 0.0233, 0.0688, 0.1196, 0.0165, 0.0301, 0.0593, 0.0175], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0079, 0.0070, 0.0115, 0.0065, 0.0079, 0.0091, 0.0068], device='cuda:0'), out_proj_covar=tensor([7.3847e-05, 6.2366e-05, 6.1859e-05, 1.0253e-04, 6.2709e-05, 6.3225e-05, 7.4696e-05, 5.3019e-05], device='cuda:0') 2023-04-16 15:09:34,391 INFO [train.py:893] (0/4) Epoch 6, batch 500, loss[loss=0.3044, simple_loss=0.3259, pruned_loss=0.1415, over 13090.00 frames. ], tot_loss[loss=0.269, simple_loss=0.3022, pruned_loss=0.1179, over 2436329.32 frames. ], batch size: 142, lr: 2.20e-02, grad_scale: 16.0 2023-04-16 15:09:38,839 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-14000.pt 2023-04-16 15:09:42,853 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-16 15:09:46,560 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.786e+02 3.676e+02 4.489e+02 5.724e+02 1.174e+03, threshold=8.978e+02, percent-clipped=2.0 2023-04-16 15:10:14,491 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5973, 1.9930, 1.8780, 2.3973, 1.5384, 2.2444, 2.2022, 2.5895], device='cuda:0'), covar=tensor([0.0101, 0.0211, 0.0164, 0.0171, 0.0193, 0.0114, 0.0300, 0.0140], device='cuda:0'), in_proj_covar=tensor([0.0045, 0.0046, 0.0051, 0.0039, 0.0052, 0.0041, 0.0050, 0.0049], device='cuda:0'), out_proj_covar=tensor([5.4541e-05, 5.6615e-05, 6.5516e-05, 4.8333e-05, 6.3624e-05, 4.9215e-05, 6.3414e-05, 6.0401e-05], device='cuda:0') 2023-04-16 15:10:25,582 INFO [train.py:893] (0/4) Epoch 6, batch 550, loss[loss=0.2462, simple_loss=0.2845, pruned_loss=0.104, over 13187.00 frames. ], tot_loss[loss=0.2672, simple_loss=0.301, pruned_loss=0.1167, over 2486719.55 frames. ], batch size: 132, lr: 2.19e-02, grad_scale: 16.0 2023-04-16 15:10:32,418 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8277, 3.4585, 3.8431, 2.7023, 4.2947, 3.8452, 3.8811, 4.2755], device='cuda:0'), covar=tensor([0.0200, 0.0165, 0.0121, 0.0816, 0.0114, 0.0123, 0.0119, 0.0088], device='cuda:0'), in_proj_covar=tensor([0.0056, 0.0030, 0.0048, 0.0074, 0.0060, 0.0050, 0.0049, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:10:36,968 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0011, 2.4455, 1.9890, 1.4439, 1.3001, 2.1223, 1.6470, 2.5637], device='cuda:0'), covar=tensor([0.0634, 0.0343, 0.1005, 0.1319, 0.0436, 0.0345, 0.0700, 0.0298], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0078, 0.0071, 0.0114, 0.0066, 0.0079, 0.0090, 0.0070], device='cuda:0'), out_proj_covar=tensor([7.3738e-05, 6.1053e-05, 6.2825e-05, 1.0107e-04, 6.3106e-05, 6.3696e-05, 7.3751e-05, 5.3740e-05], device='cuda:0') 2023-04-16 15:10:50,799 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14072.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:10:57,529 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7659, 3.5306, 3.7574, 2.6697, 4.2929, 3.8790, 3.8044, 4.2224], device='cuda:0'), covar=tensor([0.0198, 0.0125, 0.0129, 0.0806, 0.0103, 0.0120, 0.0117, 0.0069], device='cuda:0'), in_proj_covar=tensor([0.0057, 0.0030, 0.0048, 0.0073, 0.0060, 0.0050, 0.0048, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:11:09,786 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14093.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 15:11:12,003 INFO [train.py:893] (0/4) Epoch 6, batch 600, loss[loss=0.3016, simple_loss=0.3165, pruned_loss=0.1434, over 13267.00 frames. ], tot_loss[loss=0.2663, simple_loss=0.2999, pruned_loss=0.1163, over 2525226.35 frames. ], batch size: 124, lr: 2.19e-02, grad_scale: 16.0 2023-04-16 15:11:18,764 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14103.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:11:21,646 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.367e+02 3.788e+02 4.827e+02 5.908e+02 1.295e+03, threshold=9.655e+02, percent-clipped=4.0 2023-04-16 15:11:34,462 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14120.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 15:11:42,533 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7999, 3.4072, 3.1481, 3.4562, 3.0609, 1.5417, 3.4659, 1.9381], device='cuda:0'), covar=tensor([0.0597, 0.0572, 0.0313, 0.0270, 0.0726, 0.1727, 0.0884, 0.1322], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0078, 0.0087, 0.0080, 0.0107, 0.0128, 0.0083, 0.0107], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 15:11:42,684 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-16 15:11:46,678 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14133.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:11:57,818 INFO [train.py:893] (0/4) Epoch 6, batch 650, loss[loss=0.2503, simple_loss=0.2941, pruned_loss=0.1032, over 13458.00 frames. ], tot_loss[loss=0.2658, simple_loss=0.2993, pruned_loss=0.1161, over 2547714.17 frames. ], batch size: 79, lr: 2.19e-02, grad_scale: 8.0 2023-04-16 15:12:14,565 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14164.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:12:23,836 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2023-04-16 15:12:30,268 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14181.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 15:12:33,680 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-16 15:12:44,464 INFO [train.py:893] (0/4) Epoch 6, batch 700, loss[loss=0.2506, simple_loss=0.2965, pruned_loss=0.1023, over 13501.00 frames. ], tot_loss[loss=0.2634, simple_loss=0.298, pruned_loss=0.1144, over 2574795.69 frames. ], batch size: 93, lr: 2.18e-02, grad_scale: 8.0 2023-04-16 15:12:53,793 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.573e+02 3.430e+02 4.164e+02 5.021e+02 1.126e+03, threshold=8.328e+02, percent-clipped=1.0 2023-04-16 15:13:13,671 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2936, 4.5694, 2.8276, 4.7234, 4.3107, 2.1732, 3.7080, 2.7620], device='cuda:0'), covar=tensor([0.0331, 0.0313, 0.1452, 0.0065, 0.0258, 0.1692, 0.0646, 0.1849], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0115, 0.0169, 0.0079, 0.0105, 0.0153, 0.0134, 0.0163], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 15:13:30,108 INFO [train.py:893] (0/4) Epoch 6, batch 750, loss[loss=0.255, simple_loss=0.283, pruned_loss=0.1136, over 13540.00 frames. ], tot_loss[loss=0.264, simple_loss=0.2983, pruned_loss=0.1148, over 2596518.76 frames. ], batch size: 72, lr: 2.18e-02, grad_scale: 8.0 2023-04-16 15:13:43,386 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2164, 4.7051, 4.5910, 4.5064, 4.3293, 4.4774, 5.1079, 4.6724], device='cuda:0'), covar=tensor([0.0804, 0.0884, 0.2517, 0.3333, 0.0888, 0.1540, 0.1019, 0.0923], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0242, 0.0317, 0.0321, 0.0171, 0.0244, 0.0287, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 15:14:17,551 INFO [train.py:893] (0/4) Epoch 6, batch 800, loss[loss=0.2818, simple_loss=0.3185, pruned_loss=0.1225, over 13534.00 frames. ], tot_loss[loss=0.2662, simple_loss=0.3001, pruned_loss=0.1162, over 2614065.52 frames. ], batch size: 85, lr: 2.18e-02, grad_scale: 8.0 2023-04-16 15:14:21,956 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5992, 4.1106, 4.1066, 4.1186, 3.9612, 3.9246, 4.5379, 4.1014], device='cuda:0'), covar=tensor([0.0703, 0.0942, 0.1868, 0.2652, 0.0849, 0.1385, 0.0912, 0.0959], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0246, 0.0318, 0.0324, 0.0171, 0.0246, 0.0285, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 15:14:25,822 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.150e+02 4.040e+02 4.852e+02 5.703e+02 9.161e+02, threshold=9.703e+02, percent-clipped=4.0 2023-04-16 15:15:01,592 INFO [train.py:893] (0/4) Epoch 6, batch 850, loss[loss=0.2845, simple_loss=0.3085, pruned_loss=0.1302, over 13546.00 frames. ], tot_loss[loss=0.2664, simple_loss=0.301, pruned_loss=0.1159, over 2625617.90 frames. ], batch size: 78, lr: 2.17e-02, grad_scale: 8.0 2023-04-16 15:15:19,543 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.6070, 4.9763, 5.1494, 4.9782, 4.7281, 4.9385, 5.5003, 4.9756], device='cuda:0'), covar=tensor([0.0586, 0.1083, 0.1748, 0.3041, 0.0674, 0.1266, 0.1164, 0.0908], device='cuda:0'), in_proj_covar=tensor([0.0224, 0.0249, 0.0325, 0.0329, 0.0172, 0.0251, 0.0291, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 15:15:44,472 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14393.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:15:46,742 INFO [train.py:893] (0/4) Epoch 6, batch 900, loss[loss=0.2718, simple_loss=0.298, pruned_loss=0.1228, over 13526.00 frames. ], tot_loss[loss=0.2676, simple_loss=0.3017, pruned_loss=0.1167, over 2636166.73 frames. ], batch size: 72, lr: 2.17e-02, grad_scale: 8.0 2023-04-16 15:15:48,919 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-16 15:15:51,162 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2023-04-16 15:15:56,734 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.530e+02 4.113e+02 4.965e+02 5.706e+02 1.317e+03, threshold=9.930e+02, percent-clipped=5.0 2023-04-16 15:16:02,895 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14413.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:16:16,999 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14428.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:16:17,786 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 15:16:29,139 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14441.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:16:30,945 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4008, 4.6849, 2.8022, 4.7281, 4.4269, 2.4223, 3.8681, 2.8648], device='cuda:0'), covar=tensor([0.0252, 0.0285, 0.1632, 0.0090, 0.0191, 0.1556, 0.0546, 0.1877], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0115, 0.0165, 0.0079, 0.0106, 0.0152, 0.0134, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 15:16:33,031 INFO [train.py:893] (0/4) Epoch 6, batch 950, loss[loss=0.275, simple_loss=0.3093, pruned_loss=0.1203, over 13486.00 frames. ], tot_loss[loss=0.2673, simple_loss=0.3011, pruned_loss=0.1168, over 2641970.44 frames. ], batch size: 100, lr: 2.17e-02, grad_scale: 8.0 2023-04-16 15:16:35,004 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0976, 2.4549, 2.1475, 1.4226, 1.4585, 2.0467, 1.7071, 2.6499], device='cuda:0'), covar=tensor([0.0561, 0.0309, 0.0934, 0.1398, 0.0465, 0.0295, 0.0741, 0.0264], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0076, 0.0070, 0.0114, 0.0063, 0.0077, 0.0090, 0.0070], device='cuda:0'), out_proj_covar=tensor([7.2502e-05, 5.9842e-05, 6.0607e-05, 9.9951e-05, 5.9366e-05, 6.1196e-05, 7.3498e-05, 5.3584e-05], device='cuda:0') 2023-04-16 15:16:45,383 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14459.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:16:54,377 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7069, 2.5034, 1.9330, 3.5825, 4.2023, 3.2520, 4.1382, 3.7152], device='cuda:0'), covar=tensor([0.0100, 0.0581, 0.0897, 0.0113, 0.0066, 0.0316, 0.0097, 0.0076], device='cuda:0'), in_proj_covar=tensor([0.0044, 0.0061, 0.0072, 0.0051, 0.0035, 0.0056, 0.0033, 0.0043], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 15:16:59,140 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14474.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:17:00,598 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14476.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 15:17:11,102 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14487.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:17:18,271 INFO [train.py:893] (0/4) Epoch 6, batch 1000, loss[loss=0.2387, simple_loss=0.26, pruned_loss=0.1087, over 12859.00 frames. ], tot_loss[loss=0.2643, simple_loss=0.2982, pruned_loss=0.1153, over 2646800.91 frames. ], batch size: 52, lr: 2.16e-02, grad_scale: 8.0 2023-04-16 15:17:28,003 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.605e+02 3.533e+02 4.135e+02 5.114e+02 9.887e+02, threshold=8.269e+02, percent-clipped=0.0 2023-04-16 15:18:05,130 INFO [train.py:893] (0/4) Epoch 6, batch 1050, loss[loss=0.2461, simple_loss=0.2829, pruned_loss=0.1046, over 13387.00 frames. ], tot_loss[loss=0.2618, simple_loss=0.2961, pruned_loss=0.1138, over 2648147.01 frames. ], batch size: 113, lr: 2.16e-02, grad_scale: 8.0 2023-04-16 15:18:07,014 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14548.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:18:14,871 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2023-04-16 15:18:50,787 INFO [train.py:893] (0/4) Epoch 6, batch 1100, loss[loss=0.2657, simple_loss=0.2952, pruned_loss=0.1181, over 13520.00 frames. ], tot_loss[loss=0.2607, simple_loss=0.2962, pruned_loss=0.1127, over 2654121.87 frames. ], batch size: 70, lr: 2.16e-02, grad_scale: 8.0 2023-04-16 15:19:00,920 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.517e+02 3.733e+02 4.712e+02 5.511e+02 1.792e+03, threshold=9.424e+02, percent-clipped=5.0 2023-04-16 15:19:02,325 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-16 15:19:03,178 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.31 vs. limit=5.0 2023-04-16 15:19:37,950 INFO [train.py:893] (0/4) Epoch 6, batch 1150, loss[loss=0.2661, simple_loss=0.3065, pruned_loss=0.1129, over 13549.00 frames. ], tot_loss[loss=0.259, simple_loss=0.2955, pruned_loss=0.1113, over 2654391.21 frames. ], batch size: 87, lr: 2.15e-02, grad_scale: 8.0 2023-04-16 15:20:23,487 INFO [train.py:893] (0/4) Epoch 6, batch 1200, loss[loss=0.2101, simple_loss=0.2604, pruned_loss=0.07989, over 13547.00 frames. ], tot_loss[loss=0.2584, simple_loss=0.2951, pruned_loss=0.1109, over 2653977.61 frames. ], batch size: 89, lr: 2.15e-02, grad_scale: 8.0 2023-04-16 15:20:32,929 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.652e+02 3.608e+02 4.164e+02 5.469e+02 1.026e+03, threshold=8.328e+02, percent-clipped=1.0 2023-04-16 15:20:50,212 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 15:20:52,840 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14728.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:21:03,258 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 15:21:04,317 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14740.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:21:09,003 INFO [train.py:893] (0/4) Epoch 6, batch 1250, loss[loss=0.2918, simple_loss=0.3236, pruned_loss=0.13, over 13484.00 frames. ], tot_loss[loss=0.2603, simple_loss=0.2965, pruned_loss=0.1121, over 2654987.54 frames. ], batch size: 93, lr: 2.15e-02, grad_scale: 8.0 2023-04-16 15:21:21,596 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14759.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:21:31,333 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14769.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:21:37,134 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14776.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:21:37,193 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14776.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 15:21:56,345 INFO [train.py:893] (0/4) Epoch 6, batch 1300, loss[loss=0.2594, simple_loss=0.2995, pruned_loss=0.1097, over 13557.00 frames. ], tot_loss[loss=0.262, simple_loss=0.2983, pruned_loss=0.1128, over 2656823.07 frames. ], batch size: 76, lr: 2.14e-02, grad_scale: 8.0 2023-04-16 15:21:56,593 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8926, 4.2422, 4.0370, 3.9908, 3.8202, 3.7664, 4.2671, 4.3028], device='cuda:0'), covar=tensor([0.0218, 0.0250, 0.0220, 0.0327, 0.0351, 0.0371, 0.0294, 0.0223], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0140, 0.0101, 0.0130, 0.0094, 0.0134, 0.0097, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:22:01,091 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14801.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:22:05,026 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.718e+02 3.545e+02 4.251e+02 5.698e+02 1.080e+03, threshold=8.501e+02, percent-clipped=3.0 2023-04-16 15:22:06,046 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14807.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:22:22,109 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14824.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 15:22:35,601 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.84 vs. limit=5.0 2023-04-16 15:22:38,573 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5787, 4.8835, 3.4118, 4.9654, 4.6130, 2.7198, 3.8661, 3.2621], device='cuda:0'), covar=tensor([0.0241, 0.0233, 0.1121, 0.0125, 0.0197, 0.1328, 0.0609, 0.1509], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0119, 0.0171, 0.0084, 0.0107, 0.0156, 0.0139, 0.0163], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 15:22:39,252 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14843.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:22:41,604 INFO [train.py:893] (0/4) Epoch 6, batch 1350, loss[loss=0.2389, simple_loss=0.2706, pruned_loss=0.1036, over 13415.00 frames. ], tot_loss[loss=0.2628, simple_loss=0.299, pruned_loss=0.1133, over 2657961.51 frames. ], batch size: 65, lr: 2.14e-02, grad_scale: 8.0 2023-04-16 15:22:57,188 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7263, 2.9153, 2.8017, 3.0378, 2.5787, 1.3446, 2.9861, 1.6278], device='cuda:0'), covar=tensor([0.0584, 0.1047, 0.0392, 0.0319, 0.0766, 0.1976, 0.0945, 0.1626], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0080, 0.0089, 0.0077, 0.0110, 0.0133, 0.0087, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 15:23:27,736 INFO [train.py:893] (0/4) Epoch 6, batch 1400, loss[loss=0.23, simple_loss=0.2738, pruned_loss=0.09312, over 13492.00 frames. ], tot_loss[loss=0.2625, simple_loss=0.2989, pruned_loss=0.1131, over 2662304.57 frames. ], batch size: 70, lr: 2.14e-02, grad_scale: 8.0 2023-04-16 15:23:36,719 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.557e+02 3.909e+02 4.450e+02 5.303e+02 8.590e+02, threshold=8.900e+02, percent-clipped=1.0 2023-04-16 15:24:01,985 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-16 15:24:12,713 INFO [train.py:893] (0/4) Epoch 6, batch 1450, loss[loss=0.2376, simple_loss=0.2854, pruned_loss=0.09486, over 13409.00 frames. ], tot_loss[loss=0.2617, simple_loss=0.2978, pruned_loss=0.1127, over 2663456.19 frames. ], batch size: 113, lr: 2.13e-02, grad_scale: 8.0 2023-04-16 15:24:38,719 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1289, 5.0101, 5.3424, 4.9138, 5.4518, 4.9777, 5.4687, 5.5130], device='cuda:0'), covar=tensor([0.0280, 0.0462, 0.0414, 0.0406, 0.0485, 0.0651, 0.0452, 0.0404], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0199, 0.0176, 0.0140, 0.0262, 0.0219, 0.0159, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0004, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:24:57,563 INFO [train.py:893] (0/4) Epoch 6, batch 1500, loss[loss=0.2474, simple_loss=0.2794, pruned_loss=0.1077, over 13507.00 frames. ], tot_loss[loss=0.2609, simple_loss=0.2974, pruned_loss=0.1122, over 2661524.52 frames. ], batch size: 70, lr: 2.13e-02, grad_scale: 8.0 2023-04-16 15:24:59,491 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7833, 2.0433, 3.8813, 3.6659, 3.8524, 3.1359, 3.6233, 2.7449], device='cuda:0'), covar=tensor([0.2315, 0.1866, 0.0070, 0.0190, 0.0088, 0.0609, 0.0163, 0.0923], device='cuda:0'), in_proj_covar=tensor([0.0186, 0.0183, 0.0090, 0.0097, 0.0097, 0.0145, 0.0098, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 15:25:06,396 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.363e+02 3.329e+02 4.229e+02 4.892e+02 9.506e+02, threshold=8.458e+02, percent-clipped=2.0 2023-04-16 15:25:13,897 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-16 15:25:31,454 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-16 15:25:43,716 INFO [train.py:893] (0/4) Epoch 6, batch 1550, loss[loss=0.2537, simple_loss=0.2972, pruned_loss=0.1051, over 13394.00 frames. ], tot_loss[loss=0.2599, simple_loss=0.2969, pruned_loss=0.1115, over 2660993.04 frames. ], batch size: 113, lr: 2.13e-02, grad_scale: 8.0 2023-04-16 15:26:04,596 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15069.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:26:29,860 INFO [train.py:893] (0/4) Epoch 6, batch 1600, loss[loss=0.2764, simple_loss=0.3117, pruned_loss=0.1205, over 13460.00 frames. ], tot_loss[loss=0.2601, simple_loss=0.2973, pruned_loss=0.1115, over 2659060.18 frames. ], batch size: 79, lr: 2.12e-02, grad_scale: 8.0 2023-04-16 15:26:30,022 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=15096.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:26:32,567 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3136, 2.9507, 2.5316, 4.1299, 4.7536, 3.4744, 4.4706, 4.2214], device='cuda:0'), covar=tensor([0.0076, 0.0523, 0.0750, 0.0095, 0.0060, 0.0323, 0.0094, 0.0081], device='cuda:0'), in_proj_covar=tensor([0.0046, 0.0064, 0.0074, 0.0053, 0.0036, 0.0057, 0.0034, 0.0045], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 15:26:39,280 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.239e+02 3.868e+02 4.961e+02 5.958e+02 8.730e+02, threshold=9.921e+02, percent-clipped=2.0 2023-04-16 15:26:49,405 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15117.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:27:13,654 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15143.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:27:15,902 INFO [train.py:893] (0/4) Epoch 6, batch 1650, loss[loss=0.3058, simple_loss=0.3306, pruned_loss=0.1405, over 13579.00 frames. ], tot_loss[loss=0.2602, simple_loss=0.2979, pruned_loss=0.1112, over 2662380.16 frames. ], batch size: 89, lr: 2.12e-02, grad_scale: 8.0 2023-04-16 15:27:43,347 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9740, 4.3321, 3.9461, 4.0095, 4.1977, 4.5021, 4.2770, 4.2423], device='cuda:0'), covar=tensor([0.0227, 0.0173, 0.0230, 0.1059, 0.0199, 0.0197, 0.0232, 0.0257], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0096, 0.0099, 0.0183, 0.0102, 0.0115, 0.0100, 0.0087], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 15:27:55,552 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1379, 4.6370, 4.3086, 4.4177, 4.2367, 4.1575, 4.6911, 4.7333], device='cuda:0'), covar=tensor([0.0206, 0.0196, 0.0206, 0.0235, 0.0307, 0.0260, 0.0257, 0.0164], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0144, 0.0104, 0.0131, 0.0096, 0.0134, 0.0099, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:27:56,319 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15191.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:27:59,802 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=15195.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:28:00,329 INFO [train.py:893] (0/4) Epoch 6, batch 1700, loss[loss=0.2284, simple_loss=0.2695, pruned_loss=0.09369, over 13216.00 frames. ], tot_loss[loss=0.2596, simple_loss=0.2977, pruned_loss=0.1108, over 2664739.96 frames. ], batch size: 58, lr: 2.12e-02, grad_scale: 8.0 2023-04-16 15:28:10,737 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.413e+02 3.567e+02 4.505e+02 5.770e+02 1.056e+03, threshold=9.010e+02, percent-clipped=1.0 2023-04-16 15:28:46,143 INFO [train.py:893] (0/4) Epoch 6, batch 1750, loss[loss=0.2258, simple_loss=0.2725, pruned_loss=0.08955, over 13558.00 frames. ], tot_loss[loss=0.257, simple_loss=0.2958, pruned_loss=0.1091, over 2667341.68 frames. ], batch size: 78, lr: 2.11e-02, grad_scale: 8.0 2023-04-16 15:28:55,538 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=15256.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 15:29:02,101 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2023-04-16 15:29:32,562 INFO [train.py:893] (0/4) Epoch 6, batch 1800, loss[loss=0.2239, simple_loss=0.2693, pruned_loss=0.08925, over 13456.00 frames. ], tot_loss[loss=0.2562, simple_loss=0.2951, pruned_loss=0.1087, over 2666179.33 frames. ], batch size: 79, lr: 2.11e-02, grad_scale: 8.0 2023-04-16 15:29:40,590 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.268e+02 3.529e+02 4.378e+02 5.414e+02 8.128e+02, threshold=8.756e+02, percent-clipped=0.0 2023-04-16 15:29:52,261 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8959, 4.4414, 4.4365, 4.2932, 4.1478, 4.2045, 4.8211, 4.4102], device='cuda:0'), covar=tensor([0.0763, 0.1019, 0.2207, 0.2756, 0.0761, 0.1419, 0.0954, 0.0934], device='cuda:0'), in_proj_covar=tensor([0.0229, 0.0252, 0.0325, 0.0334, 0.0176, 0.0251, 0.0297, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 15:30:01,789 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2292, 3.5338, 3.8359, 2.7778, 2.3359, 3.0104, 3.9056, 4.0231], device='cuda:0'), covar=tensor([0.0749, 0.0714, 0.0296, 0.1096, 0.1496, 0.0894, 0.0177, 0.0096], device='cuda:0'), in_proj_covar=tensor([0.0190, 0.0153, 0.0141, 0.0193, 0.0194, 0.0154, 0.0117, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0001], device='cuda:0') 2023-04-16 15:30:17,100 INFO [train.py:893] (0/4) Epoch 6, batch 1850, loss[loss=0.2616, simple_loss=0.2991, pruned_loss=0.112, over 13435.00 frames. ], tot_loss[loss=0.2553, simple_loss=0.2942, pruned_loss=0.1082, over 2660200.67 frames. ], batch size: 106, lr: 2.11e-02, grad_scale: 8.0 2023-04-16 15:30:18,611 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2023-04-16 15:30:18,840 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 15:30:28,607 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3142, 2.8877, 2.4608, 4.1044, 4.7366, 3.6675, 4.6616, 4.2801], device='cuda:0'), covar=tensor([0.0073, 0.0501, 0.0756, 0.0084, 0.0048, 0.0231, 0.0052, 0.0060], device='cuda:0'), in_proj_covar=tensor([0.0046, 0.0063, 0.0073, 0.0053, 0.0037, 0.0056, 0.0033, 0.0044], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 15:31:03,342 INFO [train.py:893] (0/4) Epoch 6, batch 1900, loss[loss=0.2715, simple_loss=0.3131, pruned_loss=0.1149, over 13377.00 frames. ], tot_loss[loss=0.2547, simple_loss=0.2936, pruned_loss=0.1079, over 2662446.48 frames. ], batch size: 109, lr: 2.10e-02, grad_scale: 8.0 2023-04-16 15:31:03,571 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15396.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:31:12,695 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.431e+02 3.722e+02 4.317e+02 5.433e+02 1.645e+03, threshold=8.635e+02, percent-clipped=4.0 2023-04-16 15:31:46,537 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-16 15:31:46,994 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15444.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:31:48,466 INFO [train.py:893] (0/4) Epoch 6, batch 1950, loss[loss=0.2484, simple_loss=0.2864, pruned_loss=0.1052, over 13526.00 frames. ], tot_loss[loss=0.2546, simple_loss=0.2932, pruned_loss=0.108, over 2664993.11 frames. ], batch size: 87, lr: 2.10e-02, grad_scale: 8.0 2023-04-16 15:32:04,661 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0339, 3.6095, 4.0063, 2.7698, 4.4369, 4.0398, 4.0787, 4.4688], device='cuda:0'), covar=tensor([0.0178, 0.0162, 0.0148, 0.0889, 0.0109, 0.0152, 0.0127, 0.0085], device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0031, 0.0052, 0.0077, 0.0064, 0.0054, 0.0052, 0.0043], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:32:34,554 INFO [train.py:893] (0/4) Epoch 6, batch 2000, loss[loss=0.225, simple_loss=0.2651, pruned_loss=0.09244, over 13540.00 frames. ], tot_loss[loss=0.2575, simple_loss=0.2956, pruned_loss=0.1097, over 2667217.90 frames. ], batch size: 72, lr: 2.10e-02, grad_scale: 8.0 2023-04-16 15:32:40,171 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 15:32:43,467 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.298e+02 3.534e+02 4.151e+02 5.068e+02 1.089e+03, threshold=8.302e+02, percent-clipped=2.0 2023-04-16 15:33:02,482 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2416, 4.8043, 4.6503, 4.6607, 4.5977, 4.6519, 5.1673, 4.6630], device='cuda:0'), covar=tensor([0.0550, 0.0701, 0.1878, 0.2682, 0.0591, 0.1164, 0.0838, 0.0842], device='cuda:0'), in_proj_covar=tensor([0.0226, 0.0254, 0.0328, 0.0339, 0.0177, 0.0251, 0.0305, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 15:33:03,692 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2023-04-16 15:33:20,625 INFO [train.py:893] (0/4) Epoch 6, batch 2050, loss[loss=0.2634, simple_loss=0.2967, pruned_loss=0.1151, over 13360.00 frames. ], tot_loss[loss=0.2593, simple_loss=0.2971, pruned_loss=0.1108, over 2661960.07 frames. ], batch size: 73, lr: 2.10e-02, grad_scale: 8.0 2023-04-16 15:33:24,876 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=15551.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 15:34:05,168 INFO [train.py:893] (0/4) Epoch 6, batch 2100, loss[loss=0.2567, simple_loss=0.2946, pruned_loss=0.1094, over 13444.00 frames. ], tot_loss[loss=0.2586, simple_loss=0.2965, pruned_loss=0.1103, over 2663400.78 frames. ], batch size: 106, lr: 2.09e-02, grad_scale: 8.0 2023-04-16 15:34:07,221 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2830, 3.7478, 3.3656, 4.1355, 1.9389, 2.7336, 3.5908, 1.8745], device='cuda:0'), covar=tensor([0.0072, 0.0328, 0.0535, 0.0285, 0.1492, 0.0879, 0.0503, 0.2115], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0145, 0.0161, 0.0129, 0.0171, 0.0168, 0.0145, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-04-16 15:34:15,328 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.728e+02 3.686e+02 4.329e+02 5.255e+02 8.476e+02, threshold=8.658e+02, percent-clipped=1.0 2023-04-16 15:34:51,927 INFO [train.py:893] (0/4) Epoch 6, batch 2150, loss[loss=0.2728, simple_loss=0.3071, pruned_loss=0.1193, over 13533.00 frames. ], tot_loss[loss=0.2577, simple_loss=0.2961, pruned_loss=0.1096, over 2663344.30 frames. ], batch size: 87, lr: 2.09e-02, grad_scale: 8.0 2023-04-16 15:35:36,596 INFO [train.py:893] (0/4) Epoch 6, batch 2200, loss[loss=0.2803, simple_loss=0.3183, pruned_loss=0.1212, over 13523.00 frames. ], tot_loss[loss=0.2553, simple_loss=0.2942, pruned_loss=0.1082, over 2661016.13 frames. ], batch size: 91, lr: 2.09e-02, grad_scale: 8.0 2023-04-16 15:35:39,601 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-16 15:35:46,065 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.020e+02 3.661e+02 4.082e+02 5.276e+02 1.095e+03, threshold=8.164e+02, percent-clipped=2.0 2023-04-16 15:36:21,887 INFO [train.py:893] (0/4) Epoch 6, batch 2250, loss[loss=0.2866, simple_loss=0.3203, pruned_loss=0.1264, over 13353.00 frames. ], tot_loss[loss=0.2535, simple_loss=0.2927, pruned_loss=0.1072, over 2660780.15 frames. ], batch size: 109, lr: 2.08e-02, grad_scale: 8.0 2023-04-16 15:36:43,978 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4871, 3.7567, 3.5953, 3.3670, 3.6674, 3.9021, 3.8436, 3.5695], device='cuda:0'), covar=tensor([0.0283, 0.0261, 0.0274, 0.1155, 0.0264, 0.0238, 0.0205, 0.0836], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0102, 0.0104, 0.0190, 0.0106, 0.0120, 0.0102, 0.0086], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 15:37:08,043 INFO [train.py:893] (0/4) Epoch 6, batch 2300, loss[loss=0.2549, simple_loss=0.2954, pruned_loss=0.1072, over 13243.00 frames. ], tot_loss[loss=0.253, simple_loss=0.2921, pruned_loss=0.107, over 2662153.56 frames. ], batch size: 124, lr: 2.08e-02, grad_scale: 8.0 2023-04-16 15:37:08,441 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5585, 3.9690, 3.7810, 4.3803, 2.2622, 2.8208, 3.8828, 2.1767], device='cuda:0'), covar=tensor([0.0090, 0.0445, 0.0617, 0.0327, 0.1779, 0.1087, 0.0615, 0.2158], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0144, 0.0158, 0.0126, 0.0171, 0.0169, 0.0145, 0.0167], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:37:16,721 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.314e+02 3.478e+02 4.218e+02 5.479e+02 1.103e+03, threshold=8.435e+02, percent-clipped=4.0 2023-04-16 15:37:50,567 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.79 vs. limit=5.0 2023-04-16 15:37:52,548 INFO [train.py:893] (0/4) Epoch 6, batch 2350, loss[loss=0.2622, simple_loss=0.3028, pruned_loss=0.1108, over 13195.00 frames. ], tot_loss[loss=0.2526, simple_loss=0.2918, pruned_loss=0.1067, over 2661785.91 frames. ], batch size: 132, lr: 2.08e-02, grad_scale: 8.0 2023-04-16 15:37:57,490 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15851.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:38:16,823 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 15:38:26,894 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=15883.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:38:27,988 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-16 15:38:39,143 INFO [train.py:893] (0/4) Epoch 6, batch 2400, loss[loss=0.2809, simple_loss=0.3104, pruned_loss=0.1258, over 13524.00 frames. ], tot_loss[loss=0.2532, simple_loss=0.292, pruned_loss=0.1072, over 2664276.90 frames. ], batch size: 91, lr: 2.07e-02, grad_scale: 8.0 2023-04-16 15:38:41,817 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15899.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:38:47,397 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.949e+02 3.560e+02 4.458e+02 5.993e+02 1.349e+03, threshold=8.916e+02, percent-clipped=5.0 2023-04-16 15:39:22,646 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=15944.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:39:24,046 INFO [train.py:893] (0/4) Epoch 6, batch 2450, loss[loss=0.2438, simple_loss=0.2909, pruned_loss=0.09836, over 13380.00 frames. ], tot_loss[loss=0.2522, simple_loss=0.2912, pruned_loss=0.1066, over 2662353.50 frames. ], batch size: 113, lr: 2.07e-02, grad_scale: 8.0 2023-04-16 15:39:27,049 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3445, 4.2359, 3.7484, 3.2036, 2.9922, 2.2702, 4.4676, 2.5443], device='cuda:0'), covar=tensor([0.0956, 0.0192, 0.0372, 0.0892, 0.0522, 0.2274, 0.0109, 0.2562], device='cuda:0'), in_proj_covar=tensor([0.0259, 0.0219, 0.0219, 0.0248, 0.0191, 0.0243, 0.0150, 0.0298], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 15:40:01,043 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2755, 3.3785, 3.7900, 2.8868, 2.3431, 2.8350, 3.9533, 4.0159], device='cuda:0'), covar=tensor([0.0756, 0.0835, 0.0276, 0.1060, 0.1372, 0.0910, 0.0155, 0.0086], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0161, 0.0145, 0.0193, 0.0191, 0.0153, 0.0124, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0001], device='cuda:0') 2023-04-16 15:40:09,829 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8379, 4.2160, 2.5187, 4.3077, 3.9825, 2.3426, 3.4867, 2.6498], device='cuda:0'), covar=tensor([0.0257, 0.0286, 0.1377, 0.0111, 0.0249, 0.1374, 0.0615, 0.1677], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0118, 0.0167, 0.0086, 0.0110, 0.0152, 0.0139, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 15:40:10,318 INFO [train.py:893] (0/4) Epoch 6, batch 2500, loss[loss=0.2619, simple_loss=0.3032, pruned_loss=0.1103, over 13448.00 frames. ], tot_loss[loss=0.2507, simple_loss=0.2899, pruned_loss=0.1058, over 2657341.73 frames. ], batch size: 103, lr: 2.07e-02, grad_scale: 8.0 2023-04-16 15:40:13,952 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-16000.pt 2023-04-16 15:40:18,622 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16001.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:40:23,169 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.821e+02 3.155e+02 3.869e+02 4.967e+02 1.311e+03, threshold=7.737e+02, percent-clipped=1.0 2023-04-16 15:40:59,435 INFO [train.py:893] (0/4) Epoch 6, batch 2550, loss[loss=0.2597, simple_loss=0.2999, pruned_loss=0.1098, over 13363.00 frames. ], tot_loss[loss=0.2525, simple_loss=0.2912, pruned_loss=0.1069, over 2657616.29 frames. ], batch size: 118, lr: 2.07e-02, grad_scale: 8.0 2023-04-16 15:41:14,427 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16062.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:41:23,970 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 15:41:36,227 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16085.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:41:44,944 INFO [train.py:893] (0/4) Epoch 6, batch 2600, loss[loss=0.2658, simple_loss=0.3009, pruned_loss=0.1153, over 13429.00 frames. ], tot_loss[loss=0.253, simple_loss=0.2912, pruned_loss=0.1073, over 2656260.09 frames. ], batch size: 95, lr: 2.06e-02, grad_scale: 8.0 2023-04-16 15:41:53,825 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.362e+02 4.100e+02 4.621e+02 5.666e+02 1.546e+03, threshold=9.242e+02, percent-clipped=11.0 2023-04-16 15:42:05,468 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-16 15:42:24,675 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-04-16 15:42:26,559 INFO [train.py:893] (0/4) Epoch 6, batch 2650, loss[loss=0.2594, simple_loss=0.2942, pruned_loss=0.1123, over 13531.00 frames. ], tot_loss[loss=0.2534, simple_loss=0.2917, pruned_loss=0.1075, over 2656106.52 frames. ], batch size: 87, lr: 2.06e-02, grad_scale: 16.0 2023-04-16 15:42:26,826 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16146.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:42:29,013 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16149.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:43:04,408 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-6.pt 2023-04-16 15:43:29,533 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 15:43:33,345 INFO [train.py:893] (0/4) Epoch 7, batch 0, loss[loss=0.2217, simple_loss=0.2596, pruned_loss=0.09195, over 13334.00 frames. ], tot_loss[loss=0.2217, simple_loss=0.2596, pruned_loss=0.09195, over 13334.00 frames. ], batch size: 67, lr: 1.93e-02, grad_scale: 16.0 2023-04-16 15:43:33,346 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 15:43:55,951 INFO [train.py:927] (0/4) Epoch 7, validation: loss=0.1848, simple_loss=0.2348, pruned_loss=0.06743, over 2446609.00 frames. 2023-04-16 15:43:55,952 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 15:44:05,324 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.167e+02 3.559e+02 4.435e+02 5.202e+02 1.050e+03, threshold=8.871e+02, percent-clipped=1.0 2023-04-16 15:44:08,861 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16210.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:44:35,545 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16239.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:44:40,951 INFO [train.py:893] (0/4) Epoch 7, batch 50, loss[loss=0.2583, simple_loss=0.2958, pruned_loss=0.1105, over 13526.00 frames. ], tot_loss[loss=0.2474, simple_loss=0.2842, pruned_loss=0.1054, over 599848.33 frames. ], batch size: 91, lr: 1.92e-02, grad_scale: 16.0 2023-04-16 15:44:58,483 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9885, 4.3015, 4.1461, 3.9690, 4.0936, 4.5484, 4.4519, 4.1999], device='cuda:0'), covar=tensor([0.0258, 0.0257, 0.0209, 0.1200, 0.0251, 0.0220, 0.0189, 0.0302], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0105, 0.0106, 0.0196, 0.0108, 0.0124, 0.0104, 0.0087], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-16 15:45:05,834 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 15:45:05,834 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 15:45:05,834 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 15:45:05,841 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 15:45:05,857 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 15:45:05,878 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 15:45:06,627 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 15:45:27,304 INFO [train.py:893] (0/4) Epoch 7, batch 100, loss[loss=0.2246, simple_loss=0.2711, pruned_loss=0.08908, over 13488.00 frames. ], tot_loss[loss=0.2488, simple_loss=0.286, pruned_loss=0.1059, over 1056619.88 frames. ], batch size: 93, lr: 1.92e-02, grad_scale: 16.0 2023-04-16 15:45:36,609 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.237e+02 3.339e+02 4.034e+02 5.230e+02 9.968e+02, threshold=8.069e+02, percent-clipped=1.0 2023-04-16 15:46:12,563 INFO [train.py:893] (0/4) Epoch 7, batch 150, loss[loss=0.2771, simple_loss=0.2904, pruned_loss=0.1319, over 13215.00 frames. ], tot_loss[loss=0.2535, simple_loss=0.2893, pruned_loss=0.1089, over 1403506.43 frames. ], batch size: 58, lr: 1.92e-02, grad_scale: 16.0 2023-04-16 15:46:24,675 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16357.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:46:27,631 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-16 15:46:30,482 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6547, 3.4439, 3.5873, 2.7043, 4.0102, 3.7824, 3.7644, 4.0088], device='cuda:0'), covar=tensor([0.0193, 0.0114, 0.0147, 0.0719, 0.0119, 0.0156, 0.0137, 0.0084], device='cuda:0'), in_proj_covar=tensor([0.0057, 0.0031, 0.0051, 0.0075, 0.0064, 0.0054, 0.0052, 0.0042], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:46:36,866 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16371.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:46:58,254 INFO [train.py:893] (0/4) Epoch 7, batch 200, loss[loss=0.2625, simple_loss=0.2994, pruned_loss=0.1128, over 13527.00 frames. ], tot_loss[loss=0.2573, simple_loss=0.2929, pruned_loss=0.1108, over 1677485.68 frames. ], batch size: 76, lr: 1.92e-02, grad_scale: 16.0 2023-04-16 15:47:08,043 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.396e+02 3.451e+02 4.179e+02 5.669e+02 1.465e+03, threshold=8.357e+02, percent-clipped=6.0 2023-04-16 15:47:16,114 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3304, 4.7503, 4.4471, 4.4875, 4.3774, 4.2315, 4.8165, 4.7533], device='cuda:0'), covar=tensor([0.0174, 0.0175, 0.0181, 0.0223, 0.0304, 0.0247, 0.0219, 0.0180], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0142, 0.0106, 0.0131, 0.0099, 0.0135, 0.0099, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:47:31,772 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16432.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:47:40,286 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16441.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:47:44,102 INFO [train.py:893] (0/4) Epoch 7, batch 250, loss[loss=0.2422, simple_loss=0.2913, pruned_loss=0.09655, over 13425.00 frames. ], tot_loss[loss=0.2587, simple_loss=0.2943, pruned_loss=0.1115, over 1886353.17 frames. ], batch size: 95, lr: 1.91e-02, grad_scale: 16.0 2023-04-16 15:48:00,682 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4013, 4.3696, 3.8491, 3.1182, 3.2400, 2.3455, 4.5919, 2.5910], device='cuda:0'), covar=tensor([0.0936, 0.0204, 0.0386, 0.0885, 0.0437, 0.2201, 0.0103, 0.2338], device='cuda:0'), in_proj_covar=tensor([0.0264, 0.0223, 0.0226, 0.0249, 0.0194, 0.0246, 0.0152, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 15:48:28,652 INFO [train.py:893] (0/4) Epoch 7, batch 300, loss[loss=0.292, simple_loss=0.3267, pruned_loss=0.1286, over 11760.00 frames. ], tot_loss[loss=0.2576, simple_loss=0.2942, pruned_loss=0.1105, over 2055505.58 frames. ], batch size: 157, lr: 1.91e-02, grad_scale: 16.0 2023-04-16 15:48:38,410 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16505.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:48:39,008 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.699e+02 3.423e+02 4.208e+02 4.999e+02 1.353e+03, threshold=8.417e+02, percent-clipped=5.0 2023-04-16 15:49:08,844 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16539.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:49:14,159 INFO [train.py:893] (0/4) Epoch 7, batch 350, loss[loss=0.2795, simple_loss=0.3176, pruned_loss=0.1207, over 13232.00 frames. ], tot_loss[loss=0.2577, simple_loss=0.2943, pruned_loss=0.1106, over 2188272.45 frames. ], batch size: 132, lr: 1.91e-02, grad_scale: 16.0 2023-04-16 15:49:51,230 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16587.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:49:57,803 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6614, 4.0516, 3.7378, 4.3882, 2.4225, 3.2494, 4.0425, 2.0583], device='cuda:0'), covar=tensor([0.0110, 0.0410, 0.0657, 0.0392, 0.1560, 0.0832, 0.0485, 0.2278], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0151, 0.0165, 0.0136, 0.0170, 0.0172, 0.0152, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 15:49:58,976 INFO [train.py:893] (0/4) Epoch 7, batch 400, loss[loss=0.2134, simple_loss=0.263, pruned_loss=0.08188, over 13542.00 frames. ], tot_loss[loss=0.258, simple_loss=0.2948, pruned_loss=0.1106, over 2292364.54 frames. ], batch size: 72, lr: 1.91e-02, grad_scale: 16.0 2023-04-16 15:50:00,546 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2023-04-16 15:50:09,784 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.481e+02 3.841e+02 4.424e+02 5.759e+02 1.128e+03, threshold=8.848e+02, percent-clipped=5.0 2023-04-16 15:50:45,083 INFO [train.py:893] (0/4) Epoch 7, batch 450, loss[loss=0.2434, simple_loss=0.2855, pruned_loss=0.1006, over 13345.00 frames. ], tot_loss[loss=0.2587, simple_loss=0.2961, pruned_loss=0.1107, over 2369703.22 frames. ], batch size: 73, lr: 1.90e-02, grad_scale: 16.0 2023-04-16 15:50:56,128 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16657.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:51:08,820 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 15:51:30,841 INFO [train.py:893] (0/4) Epoch 7, batch 500, loss[loss=0.2704, simple_loss=0.3054, pruned_loss=0.1177, over 13281.00 frames. ], tot_loss[loss=0.2563, simple_loss=0.2942, pruned_loss=0.1092, over 2435339.30 frames. ], batch size: 124, lr: 1.90e-02, grad_scale: 16.0 2023-04-16 15:51:40,738 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16705.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:51:41,376 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.243e+02 3.362e+02 3.949e+02 4.886e+02 8.592e+02, threshold=7.898e+02, percent-clipped=0.0 2023-04-16 15:51:54,057 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-16 15:52:00,107 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16727.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:52:13,158 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16741.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:52:16,924 INFO [train.py:893] (0/4) Epoch 7, batch 550, loss[loss=0.2379, simple_loss=0.2845, pruned_loss=0.0957, over 13357.00 frames. ], tot_loss[loss=0.2557, simple_loss=0.2941, pruned_loss=0.1086, over 2486135.57 frames. ], batch size: 67, lr: 1.90e-02, grad_scale: 16.0 2023-04-16 15:52:36,380 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5020, 2.9759, 2.6141, 2.7644, 2.7470, 1.5184, 2.9829, 1.7440], device='cuda:0'), covar=tensor([0.0622, 0.0702, 0.0453, 0.0408, 0.0610, 0.1857, 0.0837, 0.1345], device='cuda:0'), in_proj_covar=tensor([0.0105, 0.0085, 0.0099, 0.0082, 0.0116, 0.0140, 0.0097, 0.0115], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:52:56,388 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16789.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:53:01,814 INFO [train.py:893] (0/4) Epoch 7, batch 600, loss[loss=0.2172, simple_loss=0.2595, pruned_loss=0.08751, over 13499.00 frames. ], tot_loss[loss=0.2533, simple_loss=0.2916, pruned_loss=0.1075, over 2527785.32 frames. ], batch size: 70, lr: 1.90e-02, grad_scale: 16.0 2023-04-16 15:53:11,167 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16805.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:53:11,751 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.461e+02 3.645e+02 4.474e+02 6.050e+02 1.289e+03, threshold=8.949e+02, percent-clipped=7.0 2023-04-16 15:53:34,538 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2831, 2.5821, 2.0012, 4.1136, 4.7132, 3.3856, 4.5538, 4.1784], device='cuda:0'), covar=tensor([0.0075, 0.0609, 0.0859, 0.0081, 0.0042, 0.0333, 0.0075, 0.0069], device='cuda:0'), in_proj_covar=tensor([0.0048, 0.0064, 0.0075, 0.0054, 0.0038, 0.0059, 0.0036, 0.0046], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:53:47,773 INFO [train.py:893] (0/4) Epoch 7, batch 650, loss[loss=0.2414, simple_loss=0.2876, pruned_loss=0.09756, over 13517.00 frames. ], tot_loss[loss=0.2536, simple_loss=0.292, pruned_loss=0.1076, over 2556032.78 frames. ], batch size: 91, lr: 1.89e-02, grad_scale: 16.0 2023-04-16 15:53:54,706 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16853.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:54:08,670 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16868.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 15:54:13,529 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2693, 4.2145, 3.6830, 3.2473, 3.3344, 2.5002, 4.4081, 2.6882], device='cuda:0'), covar=tensor([0.0963, 0.0213, 0.0413, 0.0781, 0.0404, 0.2170, 0.0129, 0.2340], device='cuda:0'), in_proj_covar=tensor([0.0263, 0.0224, 0.0223, 0.0246, 0.0196, 0.0246, 0.0155, 0.0299], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 15:54:17,667 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.33 vs. limit=5.0 2023-04-16 15:54:21,306 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0429, 4.4806, 4.1835, 4.1336, 4.2376, 3.9978, 4.4949, 4.4978], device='cuda:0'), covar=tensor([0.0218, 0.0185, 0.0212, 0.0327, 0.0234, 0.0274, 0.0249, 0.0204], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0143, 0.0107, 0.0133, 0.0099, 0.0136, 0.0100, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:54:32,821 INFO [train.py:893] (0/4) Epoch 7, batch 700, loss[loss=0.2489, simple_loss=0.2916, pruned_loss=0.1031, over 13210.00 frames. ], tot_loss[loss=0.2522, simple_loss=0.2907, pruned_loss=0.1068, over 2580716.45 frames. ], batch size: 124, lr: 1.89e-02, grad_scale: 16.0 2023-04-16 15:54:42,799 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.357e+02 3.467e+02 4.354e+02 5.050e+02 1.182e+03, threshold=8.708e+02, percent-clipped=3.0 2023-04-16 15:54:43,103 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16906.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:55:04,075 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16929.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 15:55:18,184 INFO [train.py:893] (0/4) Epoch 7, batch 750, loss[loss=0.2151, simple_loss=0.2659, pruned_loss=0.08218, over 13513.00 frames. ], tot_loss[loss=0.2522, simple_loss=0.2905, pruned_loss=0.1069, over 2599767.63 frames. ], batch size: 81, lr: 1.89e-02, grad_scale: 16.0 2023-04-16 15:55:24,205 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8157, 2.2044, 4.0413, 3.7674, 3.9894, 3.2834, 3.6669, 2.6698], device='cuda:0'), covar=tensor([0.2292, 0.1803, 0.0067, 0.0248, 0.0140, 0.0566, 0.0221, 0.1049], device='cuda:0'), in_proj_covar=tensor([0.0197, 0.0186, 0.0093, 0.0104, 0.0107, 0.0156, 0.0107, 0.0162], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 15:55:38,965 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16967.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 15:55:51,842 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3007, 3.4894, 3.8339, 2.8145, 2.5834, 2.9648, 4.0981, 4.1132], device='cuda:0'), covar=tensor([0.0766, 0.0743, 0.0299, 0.1159, 0.1290, 0.0829, 0.0168, 0.0093], device='cuda:0'), in_proj_covar=tensor([0.0199, 0.0169, 0.0152, 0.0196, 0.0195, 0.0158, 0.0129, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0001], device='cuda:0') 2023-04-16 15:56:04,367 INFO [train.py:893] (0/4) Epoch 7, batch 800, loss[loss=0.2732, simple_loss=0.315, pruned_loss=0.1157, over 13481.00 frames. ], tot_loss[loss=0.253, simple_loss=0.2914, pruned_loss=0.1074, over 2609985.40 frames. ], batch size: 100, lr: 1.89e-02, grad_scale: 16.0 2023-04-16 15:56:13,896 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.339e+02 3.609e+02 4.209e+02 5.245e+02 1.076e+03, threshold=8.417e+02, percent-clipped=1.0 2023-04-16 15:56:32,715 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17027.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:56:49,413 INFO [train.py:893] (0/4) Epoch 7, batch 850, loss[loss=0.2434, simple_loss=0.2875, pruned_loss=0.09967, over 13526.00 frames. ], tot_loss[loss=0.254, simple_loss=0.2926, pruned_loss=0.1077, over 2623502.32 frames. ], batch size: 91, lr: 1.88e-02, grad_scale: 16.0 2023-04-16 15:57:16,805 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17075.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:57:35,250 INFO [train.py:893] (0/4) Epoch 7, batch 900, loss[loss=0.2824, simple_loss=0.3109, pruned_loss=0.127, over 13349.00 frames. ], tot_loss[loss=0.2549, simple_loss=0.2932, pruned_loss=0.1083, over 2635905.69 frames. ], batch size: 118, lr: 1.88e-02, grad_scale: 16.0 2023-04-16 15:57:44,758 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.319e+02 3.791e+02 4.573e+02 5.198e+02 1.092e+03, threshold=9.145e+02, percent-clipped=2.0 2023-04-16 15:57:59,060 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-16 15:58:03,493 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 15:58:19,320 INFO [train.py:893] (0/4) Epoch 7, batch 950, loss[loss=0.2795, simple_loss=0.3113, pruned_loss=0.1239, over 13318.00 frames. ], tot_loss[loss=0.2556, simple_loss=0.2929, pruned_loss=0.1091, over 2642029.28 frames. ], batch size: 118, lr: 1.88e-02, grad_scale: 16.0 2023-04-16 15:58:57,260 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4013, 2.2265, 2.6993, 3.8676, 3.4174, 3.8376, 3.0403, 2.3192], device='cuda:0'), covar=tensor([0.0339, 0.1152, 0.0727, 0.0053, 0.0300, 0.0042, 0.0624, 0.1109], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0144, 0.0142, 0.0074, 0.0089, 0.0066, 0.0140, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:58:59,597 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17189.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 15:59:04,947 INFO [train.py:893] (0/4) Epoch 7, batch 1000, loss[loss=0.2302, simple_loss=0.2706, pruned_loss=0.0949, over 13372.00 frames. ], tot_loss[loss=0.2534, simple_loss=0.2906, pruned_loss=0.1081, over 2644986.51 frames. ], batch size: 109, lr: 1.88e-02, grad_scale: 16.0 2023-04-16 15:59:14,952 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.147e+02 3.772e+02 4.502e+02 5.538e+02 9.283e+02, threshold=9.005e+02, percent-clipped=1.0 2023-04-16 15:59:22,263 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8404, 4.1378, 2.5348, 4.3165, 3.8834, 2.1459, 3.5502, 2.4777], device='cuda:0'), covar=tensor([0.0364, 0.0365, 0.1781, 0.0137, 0.0423, 0.1777, 0.0761, 0.1951], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0126, 0.0169, 0.0094, 0.0113, 0.0157, 0.0142, 0.0163], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 15:59:30,710 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-16 15:59:31,894 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17224.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 15:59:39,969 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6731, 2.5012, 1.9405, 1.4344, 1.1458, 2.0302, 1.7659, 2.5533], device='cuda:0'), covar=tensor([0.0797, 0.0325, 0.0783, 0.1383, 0.0158, 0.0214, 0.0612, 0.0253], device='cuda:0'), in_proj_covar=tensor([0.0100, 0.0086, 0.0078, 0.0133, 0.0072, 0.0087, 0.0100, 0.0077], device='cuda:0'), out_proj_covar=tensor([7.9651e-05, 6.7662e-05, 6.5738e-05, 1.1169e-04, 6.4573e-05, 6.7383e-05, 8.1830e-05, 5.6999e-05], device='cuda:0') 2023-04-16 15:59:43,200 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6863, 2.5567, 2.9804, 4.1791, 3.6227, 4.2120, 3.1665, 2.6131], device='cuda:0'), covar=tensor([0.0357, 0.1066, 0.0747, 0.0049, 0.0261, 0.0039, 0.0675, 0.0939], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0146, 0.0144, 0.0075, 0.0091, 0.0067, 0.0143, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-04-16 15:59:50,060 INFO [train.py:893] (0/4) Epoch 7, batch 1050, loss[loss=0.2852, simple_loss=0.3143, pruned_loss=0.1281, over 11845.00 frames. ], tot_loss[loss=0.2507, simple_loss=0.2887, pruned_loss=0.1064, over 2647453.68 frames. ], batch size: 157, lr: 1.87e-02, grad_scale: 16.0 2023-04-16 15:59:55,414 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17250.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:00:06,307 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17262.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 16:00:36,194 INFO [train.py:893] (0/4) Epoch 7, batch 1100, loss[loss=0.2349, simple_loss=0.2799, pruned_loss=0.09495, over 13531.00 frames. ], tot_loss[loss=0.2499, simple_loss=0.2885, pruned_loss=0.1056, over 2637830.61 frames. ], batch size: 72, lr: 1.87e-02, grad_scale: 16.0 2023-04-16 16:00:46,001 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.333e+02 3.290e+02 4.098e+02 5.268e+02 1.127e+03, threshold=8.196e+02, percent-clipped=3.0 2023-04-16 16:00:49,130 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-16 16:01:02,819 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1013, 4.1341, 3.4124, 2.9145, 3.1854, 2.2667, 4.2620, 2.4932], device='cuda:0'), covar=tensor([0.0977, 0.0208, 0.0448, 0.0986, 0.0412, 0.2117, 0.0134, 0.2408], device='cuda:0'), in_proj_covar=tensor([0.0263, 0.0226, 0.0227, 0.0251, 0.0198, 0.0247, 0.0155, 0.0302], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:01:21,993 INFO [train.py:893] (0/4) Epoch 7, batch 1150, loss[loss=0.2565, simple_loss=0.2896, pruned_loss=0.1117, over 13514.00 frames. ], tot_loss[loss=0.2485, simple_loss=0.2883, pruned_loss=0.1043, over 2643393.01 frames. ], batch size: 70, lr: 1.87e-02, grad_scale: 16.0 2023-04-16 16:01:37,332 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17362.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:01:56,666 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17383.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 16:02:06,941 INFO [train.py:893] (0/4) Epoch 7, batch 1200, loss[loss=0.2565, simple_loss=0.2988, pruned_loss=0.1071, over 13104.00 frames. ], tot_loss[loss=0.2478, simple_loss=0.2879, pruned_loss=0.1038, over 2641612.54 frames. ], batch size: 142, lr: 1.87e-02, grad_scale: 16.0 2023-04-16 16:02:17,747 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.320e+02 3.565e+02 4.089e+02 4.949e+02 1.181e+03, threshold=8.177e+02, percent-clipped=3.0 2023-04-16 16:02:31,485 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 16:02:33,459 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17423.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:02:43,524 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 16:02:52,779 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17444.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 16:02:53,243 INFO [train.py:893] (0/4) Epoch 7, batch 1250, loss[loss=0.2765, simple_loss=0.3004, pruned_loss=0.1263, over 13175.00 frames. ], tot_loss[loss=0.2491, simple_loss=0.2888, pruned_loss=0.1047, over 2648240.15 frames. ], batch size: 58, lr: 1.86e-02, grad_scale: 16.0 2023-04-16 16:03:05,191 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.62 vs. limit=5.0 2023-04-16 16:03:38,908 INFO [train.py:893] (0/4) Epoch 7, batch 1300, loss[loss=0.2641, simple_loss=0.3066, pruned_loss=0.1108, over 13493.00 frames. ], tot_loss[loss=0.2508, simple_loss=0.2902, pruned_loss=0.1056, over 2649239.78 frames. ], batch size: 93, lr: 1.86e-02, grad_scale: 16.0 2023-04-16 16:03:46,195 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.26 vs. limit=5.0 2023-04-16 16:03:48,782 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.402e+02 3.839e+02 4.451e+02 5.527e+02 8.705e+02, threshold=8.903e+02, percent-clipped=2.0 2023-04-16 16:03:55,384 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6161, 4.4745, 4.6979, 4.5398, 4.8653, 4.3796, 4.8424, 4.8525], device='cuda:0'), covar=tensor([0.0273, 0.0396, 0.0513, 0.0366, 0.0485, 0.0700, 0.0466, 0.0418], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0212, 0.0197, 0.0151, 0.0278, 0.0238, 0.0174, 0.0186], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:04:03,800 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6370, 4.5678, 4.7700, 4.5587, 4.9342, 4.4346, 4.9321, 4.9408], device='cuda:0'), covar=tensor([0.0304, 0.0418, 0.0521, 0.0417, 0.0504, 0.0687, 0.0460, 0.0418], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0212, 0.0197, 0.0152, 0.0279, 0.0239, 0.0174, 0.0186], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:04:05,355 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17524.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 16:04:24,322 INFO [train.py:893] (0/4) Epoch 7, batch 1350, loss[loss=0.2488, simple_loss=0.2863, pruned_loss=0.1057, over 13251.00 frames. ], tot_loss[loss=0.2521, simple_loss=0.2916, pruned_loss=0.1063, over 2650459.00 frames. ], batch size: 124, lr: 1.86e-02, grad_scale: 16.0 2023-04-16 16:04:24,510 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17545.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:04:38,582 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7399, 1.9611, 4.1393, 3.7663, 4.0715, 3.4060, 3.6753, 2.7055], device='cuda:0'), covar=tensor([0.2186, 0.1782, 0.0062, 0.0204, 0.0162, 0.0358, 0.0172, 0.0919], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0184, 0.0095, 0.0104, 0.0106, 0.0153, 0.0106, 0.0163], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 16:04:40,142 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17562.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 16:04:49,086 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17572.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 16:04:58,010 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8056, 4.3441, 4.0696, 4.0682, 4.0673, 3.8926, 4.3887, 4.3958], device='cuda:0'), covar=tensor([0.0226, 0.0194, 0.0190, 0.0291, 0.0259, 0.0253, 0.0256, 0.0208], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0145, 0.0108, 0.0134, 0.0099, 0.0137, 0.0101, 0.0115], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 16:05:09,770 INFO [train.py:893] (0/4) Epoch 7, batch 1400, loss[loss=0.2438, simple_loss=0.2867, pruned_loss=0.1004, over 13304.00 frames. ], tot_loss[loss=0.2503, simple_loss=0.2901, pruned_loss=0.1052, over 2655109.98 frames. ], batch size: 124, lr: 1.86e-02, grad_scale: 16.0 2023-04-16 16:05:20,572 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.063e+02 3.603e+02 4.503e+02 5.400e+02 1.169e+03, threshold=9.005e+02, percent-clipped=1.0 2023-04-16 16:05:24,077 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17610.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:05:37,149 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.4397, 2.0472, 3.8465, 3.6421, 3.7413, 2.8861, 3.3754, 2.8211], device='cuda:0'), covar=tensor([0.2627, 0.1614, 0.0074, 0.0209, 0.0185, 0.0717, 0.0289, 0.0798], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0183, 0.0096, 0.0103, 0.0106, 0.0154, 0.0107, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 16:05:48,188 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-16 16:05:55,693 INFO [train.py:893] (0/4) Epoch 7, batch 1450, loss[loss=0.2535, simple_loss=0.3008, pruned_loss=0.1031, over 13441.00 frames. ], tot_loss[loss=0.249, simple_loss=0.2888, pruned_loss=0.1046, over 2657725.94 frames. ], batch size: 106, lr: 1.85e-02, grad_scale: 16.0 2023-04-16 16:06:40,975 INFO [train.py:893] (0/4) Epoch 7, batch 1500, loss[loss=0.2403, simple_loss=0.284, pruned_loss=0.09835, over 13528.00 frames. ], tot_loss[loss=0.247, simple_loss=0.2874, pruned_loss=0.1033, over 2660086.94 frames. ], batch size: 76, lr: 1.85e-02, grad_scale: 16.0 2023-04-16 16:06:45,757 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17700.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:06:50,423 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.535e+02 3.584e+02 4.081e+02 5.020e+02 1.019e+03, threshold=8.163e+02, percent-clipped=1.0 2023-04-16 16:06:54,955 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4024, 4.4694, 3.9068, 3.2092, 3.5597, 2.5685, 4.6992, 2.7698], device='cuda:0'), covar=tensor([0.1123, 0.0205, 0.0432, 0.1075, 0.0419, 0.2183, 0.0133, 0.2461], device='cuda:0'), in_proj_covar=tensor([0.0272, 0.0234, 0.0235, 0.0257, 0.0205, 0.0255, 0.0164, 0.0311], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:07:01,867 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17718.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:07:13,138 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17730.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:07:14,313 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.96 vs. limit=5.0 2023-04-16 16:07:20,411 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17739.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 16:07:25,688 INFO [train.py:893] (0/4) Epoch 7, batch 1550, loss[loss=0.2606, simple_loss=0.3022, pruned_loss=0.1095, over 13353.00 frames. ], tot_loss[loss=0.2474, simple_loss=0.288, pruned_loss=0.1034, over 2662409.07 frames. ], batch size: 73, lr: 1.85e-02, grad_scale: 16.0 2023-04-16 16:07:40,536 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17761.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:08:07,949 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17791.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:08:11,688 INFO [train.py:893] (0/4) Epoch 7, batch 1600, loss[loss=0.2491, simple_loss=0.2962, pruned_loss=0.101, over 13527.00 frames. ], tot_loss[loss=0.2467, simple_loss=0.2875, pruned_loss=0.1029, over 2665111.73 frames. ], batch size: 87, lr: 1.85e-02, grad_scale: 16.0 2023-04-16 16:08:21,303 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.592e+02 3.759e+02 4.311e+02 5.092e+02 1.331e+03, threshold=8.621e+02, percent-clipped=7.0 2023-04-16 16:08:57,467 INFO [train.py:893] (0/4) Epoch 7, batch 1650, loss[loss=0.2608, simple_loss=0.3018, pruned_loss=0.1099, over 13531.00 frames. ], tot_loss[loss=0.2459, simple_loss=0.2873, pruned_loss=0.1022, over 2662852.58 frames. ], batch size: 91, lr: 1.84e-02, grad_scale: 16.0 2023-04-16 16:08:57,738 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17845.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:09:41,153 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17893.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:09:42,612 INFO [train.py:893] (0/4) Epoch 7, batch 1700, loss[loss=0.2359, simple_loss=0.283, pruned_loss=0.0944, over 11913.00 frames. ], tot_loss[loss=0.2464, simple_loss=0.2883, pruned_loss=0.1022, over 2662869.78 frames. ], batch size: 158, lr: 1.84e-02, grad_scale: 16.0 2023-04-16 16:09:52,382 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.676e+02 3.617e+02 4.238e+02 5.324e+02 1.365e+03, threshold=8.475e+02, percent-clipped=1.0 2023-04-16 16:10:28,920 INFO [train.py:893] (0/4) Epoch 7, batch 1750, loss[loss=0.2351, simple_loss=0.2748, pruned_loss=0.0977, over 12025.00 frames. ], tot_loss[loss=0.2456, simple_loss=0.2874, pruned_loss=0.1019, over 2660641.67 frames. ], batch size: 157, lr: 1.84e-02, grad_scale: 16.0 2023-04-16 16:10:42,279 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17960.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:11:12,780 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4749, 4.8076, 3.2818, 4.9261, 4.6424, 2.8111, 4.0594, 3.0311], device='cuda:0'), covar=tensor([0.0281, 0.0279, 0.1198, 0.0125, 0.0206, 0.1326, 0.0402, 0.1817], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0126, 0.0168, 0.0097, 0.0110, 0.0156, 0.0143, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 16:11:13,895 INFO [train.py:893] (0/4) Epoch 7, batch 1800, loss[loss=0.2188, simple_loss=0.2497, pruned_loss=0.09396, over 12888.00 frames. ], tot_loss[loss=0.2441, simple_loss=0.2864, pruned_loss=0.1009, over 2665818.10 frames. ], batch size: 52, lr: 1.84e-02, grad_scale: 16.0 2023-04-16 16:11:15,998 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-16 16:11:18,997 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-18000.pt 2023-04-16 16:11:27,080 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9390, 4.3228, 2.7152, 4.4304, 4.1160, 2.2099, 3.4351, 2.7830], device='cuda:0'), covar=tensor([0.0288, 0.0273, 0.1208, 0.0145, 0.0265, 0.1374, 0.0620, 0.1398], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0126, 0.0169, 0.0098, 0.0110, 0.0156, 0.0143, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 16:11:28,369 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.327e+02 3.248e+02 3.972e+02 4.740e+02 1.184e+03, threshold=7.944e+02, percent-clipped=3.0 2023-04-16 16:11:39,192 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18018.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:11:41,639 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18021.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:11:58,682 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18039.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 16:12:03,912 INFO [train.py:893] (0/4) Epoch 7, batch 1850, loss[loss=0.2317, simple_loss=0.2791, pruned_loss=0.09213, over 13219.00 frames. ], tot_loss[loss=0.243, simple_loss=0.2852, pruned_loss=0.1004, over 2665650.66 frames. ], batch size: 132, lr: 1.83e-02, grad_scale: 16.0 2023-04-16 16:12:06,397 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 16:12:13,051 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18056.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:12:22,678 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18066.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:12:40,530 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18086.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:12:41,377 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18087.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 16:12:43,768 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8267, 2.2945, 1.9979, 2.5552, 1.6976, 2.5405, 2.5667, 2.3591], device='cuda:0'), covar=tensor([0.0067, 0.0116, 0.0114, 0.0127, 0.0160, 0.0104, 0.0166, 0.0123], device='cuda:0'), in_proj_covar=tensor([0.0053, 0.0055, 0.0062, 0.0053, 0.0067, 0.0052, 0.0060, 0.0056], device='cuda:0'), out_proj_covar=tensor([6.1933e-05, 6.5190e-05, 7.7263e-05, 6.4591e-05, 8.0853e-05, 6.1769e-05, 7.4824e-05, 6.7543e-05], device='cuda:0') 2023-04-16 16:12:48,930 INFO [train.py:893] (0/4) Epoch 7, batch 1900, loss[loss=0.2454, simple_loss=0.2862, pruned_loss=0.1024, over 13542.00 frames. ], tot_loss[loss=0.2433, simple_loss=0.285, pruned_loss=0.1008, over 2665208.11 frames. ], batch size: 87, lr: 1.83e-02, grad_scale: 16.0 2023-04-16 16:12:58,583 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.339e+02 3.460e+02 4.332e+02 5.125e+02 1.072e+03, threshold=8.663e+02, percent-clipped=3.0 2023-04-16 16:13:20,167 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6671, 3.8668, 4.4084, 3.3582, 2.8949, 3.0424, 4.6209, 4.7062], device='cuda:0'), covar=tensor([0.0842, 0.0919, 0.0288, 0.1138, 0.1279, 0.1026, 0.0172, 0.0100], device='cuda:0'), in_proj_covar=tensor([0.0198, 0.0175, 0.0149, 0.0193, 0.0193, 0.0157, 0.0133, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0001], device='cuda:0') 2023-04-16 16:13:34,749 INFO [train.py:893] (0/4) Epoch 7, batch 1950, loss[loss=0.2286, simple_loss=0.2741, pruned_loss=0.09156, over 13542.00 frames. ], tot_loss[loss=0.2421, simple_loss=0.2838, pruned_loss=0.1002, over 2660817.58 frames. ], batch size: 72, lr: 1.83e-02, grad_scale: 32.0 2023-04-16 16:13:48,161 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5933, 3.9616, 4.2721, 3.2410, 2.8037, 3.2570, 4.5330, 4.5722], device='cuda:0'), covar=tensor([0.0868, 0.0727, 0.0282, 0.1149, 0.1435, 0.0983, 0.0163, 0.0095], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0174, 0.0148, 0.0191, 0.0191, 0.0155, 0.0131, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0001], device='cuda:0') 2023-04-16 16:14:01,856 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18174.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:14:02,683 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18175.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:14:20,800 INFO [train.py:893] (0/4) Epoch 7, batch 2000, loss[loss=0.2665, simple_loss=0.3066, pruned_loss=0.1131, over 13465.00 frames. ], tot_loss[loss=0.2444, simple_loss=0.2859, pruned_loss=0.1015, over 2661668.56 frames. ], batch size: 100, lr: 1.83e-02, grad_scale: 32.0 2023-04-16 16:14:26,665 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 16:14:27,020 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5333, 4.4171, 3.9750, 3.2733, 3.3691, 2.4889, 4.6625, 2.7490], device='cuda:0'), covar=tensor([0.1008, 0.0210, 0.0386, 0.0944, 0.0448, 0.2450, 0.0121, 0.2364], device='cuda:0'), in_proj_covar=tensor([0.0270, 0.0232, 0.0238, 0.0256, 0.0202, 0.0256, 0.0165, 0.0306], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:14:30,677 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.042e+02 3.662e+02 4.315e+02 5.072e+02 8.289e+02, threshold=8.631e+02, percent-clipped=0.0 2023-04-16 16:14:57,912 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18235.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:14:58,737 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18236.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:15:06,525 INFO [train.py:893] (0/4) Epoch 7, batch 2050, loss[loss=0.2951, simple_loss=0.3275, pruned_loss=0.1313, over 11815.00 frames. ], tot_loss[loss=0.2454, simple_loss=0.2869, pruned_loss=0.1019, over 2662109.73 frames. ], batch size: 158, lr: 1.82e-02, grad_scale: 32.0 2023-04-16 16:15:12,479 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2267, 3.5558, 3.2346, 3.8316, 1.9073, 2.7868, 3.6013, 1.8308], device='cuda:0'), covar=tensor([0.0094, 0.0445, 0.0650, 0.0345, 0.1577, 0.0895, 0.0523, 0.2174], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0153, 0.0170, 0.0150, 0.0175, 0.0178, 0.0156, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:15:33,916 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9569, 4.3299, 4.1721, 4.0225, 4.0264, 3.7939, 4.4476, 4.4340], device='cuda:0'), covar=tensor([0.0224, 0.0246, 0.0180, 0.0301, 0.0331, 0.0300, 0.0245, 0.0223], device='cuda:0'), in_proj_covar=tensor([0.0105, 0.0153, 0.0113, 0.0139, 0.0106, 0.0144, 0.0102, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:15:52,005 INFO [train.py:893] (0/4) Epoch 7, batch 2100, loss[loss=0.2381, simple_loss=0.2771, pruned_loss=0.09952, over 13327.00 frames. ], tot_loss[loss=0.2446, simple_loss=0.2863, pruned_loss=0.1015, over 2666496.44 frames. ], batch size: 67, lr: 1.82e-02, grad_scale: 32.0 2023-04-16 16:16:01,878 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.613e+02 3.375e+02 4.098e+02 5.034e+02 9.532e+02, threshold=8.196e+02, percent-clipped=1.0 2023-04-16 16:16:11,140 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18316.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:16:26,862 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0122, 4.3956, 4.0244, 4.0768, 4.1634, 3.8867, 4.5016, 4.4755], device='cuda:0'), covar=tensor([0.0232, 0.0241, 0.0271, 0.0338, 0.0304, 0.0307, 0.0247, 0.0239], device='cuda:0'), in_proj_covar=tensor([0.0104, 0.0148, 0.0112, 0.0137, 0.0103, 0.0140, 0.0100, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 16:16:36,241 INFO [train.py:893] (0/4) Epoch 7, batch 2150, loss[loss=0.2326, simple_loss=0.2733, pruned_loss=0.09592, over 13381.00 frames. ], tot_loss[loss=0.2444, simple_loss=0.2863, pruned_loss=0.1012, over 2663577.83 frames. ], batch size: 73, lr: 1.82e-02, grad_scale: 32.0 2023-04-16 16:16:47,378 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18356.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:17:13,945 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18386.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:17:21,648 INFO [train.py:893] (0/4) Epoch 7, batch 2200, loss[loss=0.2387, simple_loss=0.2851, pruned_loss=0.09617, over 13431.00 frames. ], tot_loss[loss=0.2428, simple_loss=0.2853, pruned_loss=0.1001, over 2667266.25 frames. ], batch size: 95, lr: 1.82e-02, grad_scale: 32.0 2023-04-16 16:17:30,265 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18404.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:17:31,681 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.091e+02 3.363e+02 3.835e+02 4.605e+02 1.288e+03, threshold=7.670e+02, percent-clipped=1.0 2023-04-16 16:17:57,906 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18434.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:18:08,163 INFO [train.py:893] (0/4) Epoch 7, batch 2250, loss[loss=0.2245, simple_loss=0.2786, pruned_loss=0.08521, over 13470.00 frames. ], tot_loss[loss=0.2416, simple_loss=0.2844, pruned_loss=0.09939, over 2667684.96 frames. ], batch size: 100, lr: 1.82e-02, grad_scale: 32.0 2023-04-16 16:18:49,098 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18490.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:18:53,482 INFO [train.py:893] (0/4) Epoch 7, batch 2300, loss[loss=0.2548, simple_loss=0.2986, pruned_loss=0.1055, over 13522.00 frames. ], tot_loss[loss=0.2416, simple_loss=0.2844, pruned_loss=0.09938, over 2666878.27 frames. ], batch size: 91, lr: 1.81e-02, grad_scale: 32.0 2023-04-16 16:19:03,359 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.168e+02 3.469e+02 4.011e+02 5.110e+02 9.671e+02, threshold=8.022e+02, percent-clipped=1.0 2023-04-16 16:19:05,304 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18508.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 16:19:13,325 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.5836, 2.5868, 2.1121, 1.2910, 1.1586, 1.9352, 1.6687, 2.6064], device='cuda:0'), covar=tensor([0.0914, 0.0283, 0.0843, 0.2097, 0.0318, 0.0465, 0.0867, 0.0248], device='cuda:0'), in_proj_covar=tensor([0.0105, 0.0088, 0.0081, 0.0141, 0.0075, 0.0091, 0.0106, 0.0086], device='cuda:0'), out_proj_covar=tensor([8.2767e-05, 6.8840e-05, 6.8260e-05, 1.1709e-04, 6.6254e-05, 7.0532e-05, 8.5238e-05, 6.4721e-05], device='cuda:0') 2023-04-16 16:19:25,253 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18530.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:19:26,085 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18531.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:19:37,544 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-16 16:19:38,755 INFO [train.py:893] (0/4) Epoch 7, batch 2350, loss[loss=0.2193, simple_loss=0.2616, pruned_loss=0.08851, over 13429.00 frames. ], tot_loss[loss=0.2406, simple_loss=0.2836, pruned_loss=0.09884, over 2666918.62 frames. ], batch size: 65, lr: 1.81e-02, grad_scale: 32.0 2023-04-16 16:19:44,390 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18551.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:19:55,866 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5289, 3.9045, 3.4671, 4.1554, 2.1112, 2.9633, 3.8294, 2.0467], device='cuda:0'), covar=tensor([0.0129, 0.0490, 0.0717, 0.0470, 0.1769, 0.0984, 0.0672, 0.2289], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0153, 0.0168, 0.0151, 0.0171, 0.0176, 0.0154, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:20:00,602 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18569.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 16:20:01,170 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 16:20:02,113 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9660, 4.8522, 5.1882, 4.8530, 5.3430, 4.8582, 5.3693, 5.3421], device='cuda:0'), covar=tensor([0.0321, 0.0506, 0.0460, 0.0465, 0.0490, 0.0670, 0.0414, 0.0397], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0207, 0.0191, 0.0147, 0.0282, 0.0236, 0.0171, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:20:23,723 INFO [train.py:893] (0/4) Epoch 7, batch 2400, loss[loss=0.2208, simple_loss=0.2671, pruned_loss=0.08732, over 13374.00 frames. ], tot_loss[loss=0.2404, simple_loss=0.2827, pruned_loss=0.09901, over 2662780.50 frames. ], batch size: 67, lr: 1.81e-02, grad_scale: 32.0 2023-04-16 16:20:34,557 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.416e+02 3.464e+02 4.191e+02 4.925e+02 8.297e+02, threshold=8.381e+02, percent-clipped=2.0 2023-04-16 16:20:43,600 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18616.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:21:05,306 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-16 16:21:09,659 INFO [train.py:893] (0/4) Epoch 7, batch 2450, loss[loss=0.2782, simple_loss=0.3158, pruned_loss=0.1203, over 13442.00 frames. ], tot_loss[loss=0.2401, simple_loss=0.2826, pruned_loss=0.09879, over 2661476.75 frames. ], batch size: 106, lr: 1.81e-02, grad_scale: 32.0 2023-04-16 16:21:26,028 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18664.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:21:36,313 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18674.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:21:53,928 INFO [train.py:893] (0/4) Epoch 7, batch 2500, loss[loss=0.2517, simple_loss=0.289, pruned_loss=0.1072, over 13467.00 frames. ], tot_loss[loss=0.2401, simple_loss=0.2823, pruned_loss=0.0989, over 2658530.91 frames. ], batch size: 100, lr: 1.80e-02, grad_scale: 32.0 2023-04-16 16:22:04,434 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.081e+02 3.550e+02 4.127e+02 4.846e+02 9.992e+02, threshold=8.253e+02, percent-clipped=2.0 2023-04-16 16:22:30,414 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18735.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:22:39,103 INFO [train.py:893] (0/4) Epoch 7, batch 2550, loss[loss=0.2707, simple_loss=0.2988, pruned_loss=0.1213, over 11976.00 frames. ], tot_loss[loss=0.2404, simple_loss=0.2828, pruned_loss=0.09899, over 2656808.45 frames. ], batch size: 157, lr: 1.80e-02, grad_scale: 32.0 2023-04-16 16:23:04,396 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 16:23:25,395 INFO [train.py:893] (0/4) Epoch 7, batch 2600, loss[loss=0.2321, simple_loss=0.277, pruned_loss=0.09365, over 13541.00 frames. ], tot_loss[loss=0.241, simple_loss=0.2831, pruned_loss=0.09948, over 2659081.53 frames. ], batch size: 85, lr: 1.80e-02, grad_scale: 32.0 2023-04-16 16:23:26,046 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-16 16:23:35,304 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.202e+02 3.564e+02 4.195e+02 5.330e+02 9.508e+02, threshold=8.389e+02, percent-clipped=2.0 2023-04-16 16:23:44,359 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2392, 4.6816, 4.6630, 4.6110, 4.4593, 4.5214, 5.2004, 4.6952], device='cuda:0'), covar=tensor([0.0774, 0.0932, 0.2141, 0.2681, 0.0867, 0.1256, 0.0874, 0.0901], device='cuda:0'), in_proj_covar=tensor([0.0234, 0.0261, 0.0333, 0.0343, 0.0186, 0.0254, 0.0306, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 16:23:54,832 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18830.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:23:55,533 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18831.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:24:05,695 INFO [train.py:893] (0/4) Epoch 7, batch 2650, loss[loss=0.2404, simple_loss=0.2854, pruned_loss=0.09774, over 13206.00 frames. ], tot_loss[loss=0.2422, simple_loss=0.2839, pruned_loss=0.1002, over 2655962.13 frames. ], batch size: 132, lr: 1.80e-02, grad_scale: 32.0 2023-04-16 16:24:06,631 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18846.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:24:21,325 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18864.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 16:24:27,712 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0271, 3.8545, 3.3998, 2.8112, 2.8271, 2.2377, 4.0011, 2.2873], device='cuda:0'), covar=tensor([0.0975, 0.0284, 0.0451, 0.1012, 0.0537, 0.2115, 0.0152, 0.2642], device='cuda:0'), in_proj_covar=tensor([0.0269, 0.0233, 0.0236, 0.0258, 0.0203, 0.0256, 0.0167, 0.0307], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:24:31,896 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18878.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:24:32,554 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18879.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:24:45,573 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-7.pt 2023-04-16 16:25:09,624 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 16:25:13,580 INFO [train.py:893] (0/4) Epoch 8, batch 0, loss[loss=0.2488, simple_loss=0.2863, pruned_loss=0.1057, over 13525.00 frames. ], tot_loss[loss=0.2488, simple_loss=0.2863, pruned_loss=0.1057, over 13525.00 frames. ], batch size: 98, lr: 1.69e-02, grad_scale: 16.0 2023-04-16 16:25:13,580 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 16:25:34,773 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5566, 2.3043, 2.9552, 4.0203, 3.8310, 4.0332, 3.4064, 2.4698], device='cuda:0'), covar=tensor([0.0231, 0.1174, 0.0721, 0.0047, 0.0182, 0.0054, 0.0504, 0.0987], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0141, 0.0143, 0.0075, 0.0092, 0.0070, 0.0140, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-04-16 16:25:34,783 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8315, 1.8776, 3.8731, 3.8653, 3.7598, 3.3215, 3.8178, 2.7800], device='cuda:0'), covar=tensor([0.2067, 0.1545, 0.0096, 0.0199, 0.0241, 0.0555, 0.0141, 0.0808], device='cuda:0'), in_proj_covar=tensor([0.0201, 0.0189, 0.0098, 0.0108, 0.0112, 0.0154, 0.0108, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 16:25:35,920 INFO [train.py:927] (0/4) Epoch 8, validation: loss=0.1764, simple_loss=0.2286, pruned_loss=0.06213, over 2446609.00 frames. 2023-04-16 16:25:35,921 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 16:25:47,140 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.595e+02 3.840e+02 4.521e+02 5.320e+02 9.570e+02, threshold=9.042e+02, percent-clipped=2.0 2023-04-16 16:25:52,412 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-16 16:26:07,715 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18930.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:26:21,958 INFO [train.py:893] (0/4) Epoch 8, batch 50, loss[loss=0.2345, simple_loss=0.268, pruned_loss=0.1005, over 13526.00 frames. ], tot_loss[loss=0.2365, simple_loss=0.2755, pruned_loss=0.09874, over 601493.97 frames. ], batch size: 76, lr: 1.69e-02, grad_scale: 16.0 2023-04-16 16:26:46,542 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 16:26:46,542 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 16:26:46,542 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 16:26:46,556 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 16:26:47,286 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 16:26:47,311 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 16:26:48,045 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 16:27:04,783 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18991.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:27:07,781 INFO [train.py:893] (0/4) Epoch 8, batch 100, loss[loss=0.2323, simple_loss=0.2788, pruned_loss=0.09297, over 13457.00 frames. ], tot_loss[loss=0.2399, simple_loss=0.2789, pruned_loss=0.1005, over 1049769.12 frames. ], batch size: 106, lr: 1.69e-02, grad_scale: 16.0 2023-04-16 16:27:19,878 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.147e+02 3.465e+02 4.037e+02 4.694e+02 8.484e+02, threshold=8.075e+02, percent-clipped=0.0 2023-04-16 16:27:40,533 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19030.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:27:54,667 INFO [train.py:893] (0/4) Epoch 8, batch 150, loss[loss=0.24, simple_loss=0.2791, pruned_loss=0.1004, over 13063.00 frames. ], tot_loss[loss=0.2436, simple_loss=0.2825, pruned_loss=0.1024, over 1394712.44 frames. ], batch size: 142, lr: 1.68e-02, grad_scale: 16.0 2023-04-16 16:28:30,755 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6222, 4.1708, 4.1302, 4.1726, 3.8481, 4.0419, 4.6028, 4.0522], device='cuda:0'), covar=tensor([0.0892, 0.1067, 0.2199, 0.2746, 0.0970, 0.1371, 0.0969, 0.1107], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0277, 0.0352, 0.0357, 0.0196, 0.0267, 0.0324, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 16:28:41,429 INFO [train.py:893] (0/4) Epoch 8, batch 200, loss[loss=0.247, simple_loss=0.2862, pruned_loss=0.1039, over 13519.00 frames. ], tot_loss[loss=0.2452, simple_loss=0.2845, pruned_loss=0.1029, over 1667149.76 frames. ], batch size: 85, lr: 1.68e-02, grad_scale: 16.0 2023-04-16 16:28:53,339 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.148e+02 3.539e+02 3.896e+02 4.763e+02 1.039e+03, threshold=7.792e+02, percent-clipped=3.0 2023-04-16 16:29:07,764 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6173, 2.6994, 2.2226, 1.6882, 1.6082, 2.3729, 2.2022, 2.9042], device='cuda:0'), covar=tensor([0.0833, 0.0333, 0.0917, 0.1381, 0.0581, 0.0436, 0.0632, 0.0308], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0095, 0.0086, 0.0150, 0.0080, 0.0097, 0.0108, 0.0090], device='cuda:0'), out_proj_covar=tensor([8.5548e-05, 7.3123e-05, 7.1955e-05, 1.2338e-04, 7.1180e-05, 7.4325e-05, 8.6402e-05, 6.7004e-05], device='cuda:0') 2023-04-16 16:29:12,760 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2023-04-16 16:29:14,173 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19129.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 16:29:28,592 INFO [train.py:893] (0/4) Epoch 8, batch 250, loss[loss=0.2416, simple_loss=0.2838, pruned_loss=0.09971, over 13407.00 frames. ], tot_loss[loss=0.2451, simple_loss=0.2849, pruned_loss=0.1026, over 1883092.27 frames. ], batch size: 113, lr: 1.68e-02, grad_scale: 16.0 2023-04-16 16:29:29,762 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19146.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:29:33,241 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1555, 2.3735, 1.9263, 3.8943, 4.5325, 3.3425, 4.4577, 4.1559], device='cuda:0'), covar=tensor([0.0095, 0.0785, 0.1066, 0.0103, 0.0071, 0.0404, 0.0067, 0.0072], device='cuda:0'), in_proj_covar=tensor([0.0050, 0.0069, 0.0078, 0.0057, 0.0041, 0.0063, 0.0038, 0.0049], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 16:29:46,590 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19164.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 16:29:59,100 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5419, 2.2328, 2.8123, 4.1767, 3.8141, 4.1887, 3.2960, 2.3583], device='cuda:0'), covar=tensor([0.0286, 0.1315, 0.0776, 0.0041, 0.0185, 0.0044, 0.0607, 0.1151], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0143, 0.0145, 0.0077, 0.0094, 0.0070, 0.0142, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-04-16 16:30:10,828 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19190.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 16:30:12,060 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-16 16:30:14,660 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19194.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:30:15,322 INFO [train.py:893] (0/4) Epoch 8, batch 300, loss[loss=0.2244, simple_loss=0.271, pruned_loss=0.08886, over 13512.00 frames. ], tot_loss[loss=0.2451, simple_loss=0.2851, pruned_loss=0.1025, over 2053903.73 frames. ], batch size: 70, lr: 1.68e-02, grad_scale: 16.0 2023-04-16 16:30:27,117 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.082e+02 3.374e+02 4.034e+02 4.942e+02 9.430e+02, threshold=8.068e+02, percent-clipped=2.0 2023-04-16 16:30:31,503 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19212.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 16:31:01,899 INFO [train.py:893] (0/4) Epoch 8, batch 350, loss[loss=0.2736, simple_loss=0.3054, pruned_loss=0.1209, over 13095.00 frames. ], tot_loss[loss=0.2451, simple_loss=0.2847, pruned_loss=0.1027, over 2185087.38 frames. ], batch size: 142, lr: 1.68e-02, grad_scale: 16.0 2023-04-16 16:31:39,951 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19286.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:31:48,569 INFO [train.py:893] (0/4) Epoch 8, batch 400, loss[loss=0.2546, simple_loss=0.2935, pruned_loss=0.1079, over 13524.00 frames. ], tot_loss[loss=0.2447, simple_loss=0.2848, pruned_loss=0.1023, over 2288289.65 frames. ], batch size: 85, lr: 1.67e-02, grad_scale: 16.0 2023-04-16 16:31:59,043 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.962e+02 3.520e+02 4.019e+02 4.754e+02 8.353e+02, threshold=8.037e+02, percent-clipped=1.0 2023-04-16 16:32:20,897 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19330.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:32:30,988 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-04-16 16:32:33,825 INFO [train.py:893] (0/4) Epoch 8, batch 450, loss[loss=0.2589, simple_loss=0.2957, pruned_loss=0.1111, over 13530.00 frames. ], tot_loss[loss=0.2473, simple_loss=0.2872, pruned_loss=0.1037, over 2364832.78 frames. ], batch size: 76, lr: 1.67e-02, grad_scale: 16.0 2023-04-16 16:32:57,110 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9596, 4.3808, 4.0331, 4.6536, 2.5565, 3.2148, 4.3066, 2.3513], device='cuda:0'), covar=tensor([0.0093, 0.0421, 0.0650, 0.0457, 0.1507, 0.0993, 0.0539, 0.2161], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0162, 0.0172, 0.0156, 0.0175, 0.0179, 0.0159, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:32:59,487 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 16:33:05,417 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19378.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:33:18,154 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.62 vs. limit=5.0 2023-04-16 16:33:21,085 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19394.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:33:21,690 INFO [train.py:893] (0/4) Epoch 8, batch 500, loss[loss=0.2353, simple_loss=0.2801, pruned_loss=0.09525, over 13490.00 frames. ], tot_loss[loss=0.246, simple_loss=0.2867, pruned_loss=0.1027, over 2429061.22 frames. ], batch size: 93, lr: 1.67e-02, grad_scale: 16.0 2023-04-16 16:33:31,896 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.189e+02 3.388e+02 3.878e+02 4.820e+02 9.115e+02, threshold=7.756e+02, percent-clipped=3.0 2023-04-16 16:33:37,193 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7155, 1.4832, 3.7558, 3.5626, 3.6568, 2.6027, 3.5189, 2.5646], device='cuda:0'), covar=tensor([0.2059, 0.2083, 0.0063, 0.0287, 0.0130, 0.0772, 0.0147, 0.0857], device='cuda:0'), in_proj_covar=tensor([0.0193, 0.0183, 0.0093, 0.0104, 0.0107, 0.0151, 0.0104, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 16:33:44,798 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2023-04-16 16:33:58,047 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-16 16:34:07,899 INFO [train.py:893] (0/4) Epoch 8, batch 550, loss[loss=0.238, simple_loss=0.2972, pruned_loss=0.08938, over 13337.00 frames. ], tot_loss[loss=0.2439, simple_loss=0.2852, pruned_loss=0.1013, over 2483252.95 frames. ], batch size: 118, lr: 1.67e-02, grad_scale: 16.0 2023-04-16 16:34:18,206 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19455.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:34:44,339 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.35 vs. limit=5.0 2023-04-16 16:34:45,342 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2023-04-16 16:34:46,510 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19485.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 16:34:55,324 INFO [train.py:893] (0/4) Epoch 8, batch 600, loss[loss=0.2686, simple_loss=0.3024, pruned_loss=0.1174, over 13437.00 frames. ], tot_loss[loss=0.2439, simple_loss=0.285, pruned_loss=0.1014, over 2523164.32 frames. ], batch size: 95, lr: 1.67e-02, grad_scale: 16.0 2023-04-16 16:35:01,352 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9175, 2.7950, 2.3700, 3.0861, 2.3265, 2.9010, 2.8808, 2.8202], device='cuda:0'), covar=tensor([0.0114, 0.0117, 0.0135, 0.0095, 0.0144, 0.0079, 0.0198, 0.0125], device='cuda:0'), in_proj_covar=tensor([0.0056, 0.0058, 0.0065, 0.0053, 0.0070, 0.0053, 0.0060, 0.0059], device='cuda:0'), out_proj_covar=tensor([6.5292e-05, 6.8048e-05, 7.9523e-05, 6.4168e-05, 8.4854e-05, 6.1885e-05, 7.3063e-05, 7.0161e-05], device='cuda:0') 2023-04-16 16:35:05,980 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.294e+02 3.228e+02 3.690e+02 4.422e+02 7.814e+02, threshold=7.380e+02, percent-clipped=1.0 2023-04-16 16:35:13,654 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5832, 3.9755, 3.6582, 4.3697, 2.3065, 3.0237, 4.0197, 2.0992], device='cuda:0'), covar=tensor([0.0139, 0.0474, 0.0724, 0.0352, 0.1802, 0.1004, 0.0511, 0.2425], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0161, 0.0172, 0.0157, 0.0175, 0.0178, 0.0159, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:35:31,070 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8300, 1.9195, 4.1914, 3.7014, 4.0606, 3.2379, 3.8909, 2.8758], device='cuda:0'), covar=tensor([0.2201, 0.1813, 0.0062, 0.0191, 0.0137, 0.0514, 0.0148, 0.0824], device='cuda:0'), in_proj_covar=tensor([0.0200, 0.0188, 0.0094, 0.0105, 0.0110, 0.0155, 0.0107, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 16:35:41,872 INFO [train.py:893] (0/4) Epoch 8, batch 650, loss[loss=0.229, simple_loss=0.2803, pruned_loss=0.08887, over 13445.00 frames. ], tot_loss[loss=0.243, simple_loss=0.284, pruned_loss=0.1011, over 2547318.37 frames. ], batch size: 103, lr: 1.66e-02, grad_scale: 16.0 2023-04-16 16:36:20,089 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19586.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:36:27,513 INFO [train.py:893] (0/4) Epoch 8, batch 700, loss[loss=0.2551, simple_loss=0.2895, pruned_loss=0.1103, over 13024.00 frames. ], tot_loss[loss=0.2411, simple_loss=0.2826, pruned_loss=0.09978, over 2571514.43 frames. ], batch size: 142, lr: 1.66e-02, grad_scale: 16.0 2023-04-16 16:36:40,421 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.363e+02 3.372e+02 3.978e+02 4.795e+02 1.054e+03, threshold=7.956e+02, percent-clipped=5.0 2023-04-16 16:36:50,595 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19618.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:37:05,793 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19634.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:37:06,683 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8769, 4.4249, 4.3299, 4.3379, 4.0376, 4.2630, 4.8238, 4.3112], device='cuda:0'), covar=tensor([0.0894, 0.0991, 0.2378, 0.2887, 0.0961, 0.1637, 0.1034, 0.0965], device='cuda:0'), in_proj_covar=tensor([0.0237, 0.0270, 0.0339, 0.0346, 0.0193, 0.0259, 0.0312, 0.0231], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 16:37:16,241 INFO [train.py:893] (0/4) Epoch 8, batch 750, loss[loss=0.2306, simple_loss=0.2681, pruned_loss=0.09651, over 11838.00 frames. ], tot_loss[loss=0.2408, simple_loss=0.2825, pruned_loss=0.09956, over 2591187.85 frames. ], batch size: 157, lr: 1.66e-02, grad_scale: 16.0 2023-04-16 16:37:41,331 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9636, 2.5536, 2.2249, 3.0920, 2.4561, 2.8451, 2.7043, 2.7397], device='cuda:0'), covar=tensor([0.0089, 0.0118, 0.0135, 0.0092, 0.0117, 0.0079, 0.0190, 0.0111], device='cuda:0'), in_proj_covar=tensor([0.0056, 0.0058, 0.0065, 0.0053, 0.0070, 0.0052, 0.0060, 0.0057], device='cuda:0'), out_proj_covar=tensor([6.5409e-05, 6.7700e-05, 8.0468e-05, 6.4317e-05, 8.4463e-05, 6.1620e-05, 7.2134e-05, 6.8023e-05], device='cuda:0') 2023-04-16 16:37:46,165 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6228, 1.4500, 3.8558, 3.6816, 3.7142, 2.7567, 3.5445, 2.7101], device='cuda:0'), covar=tensor([0.2537, 0.2494, 0.0074, 0.0153, 0.0256, 0.0806, 0.0211, 0.0960], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0185, 0.0093, 0.0105, 0.0107, 0.0150, 0.0106, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 16:37:47,841 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19679.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:38:03,036 INFO [train.py:893] (0/4) Epoch 8, batch 800, loss[loss=0.2317, simple_loss=0.2779, pruned_loss=0.09275, over 13199.00 frames. ], tot_loss[loss=0.2415, simple_loss=0.2831, pruned_loss=0.09989, over 2605586.73 frames. ], batch size: 132, lr: 1.66e-02, grad_scale: 16.0 2023-04-16 16:38:14,376 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.290e+02 3.416e+02 4.034e+02 4.840e+02 9.110e+02, threshold=8.068e+02, percent-clipped=3.0 2023-04-16 16:38:24,168 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0599, 4.3640, 2.6723, 4.4937, 4.1098, 2.4494, 3.4912, 2.6356], device='cuda:0'), covar=tensor([0.0314, 0.0244, 0.1411, 0.0096, 0.0246, 0.1322, 0.0589, 0.1664], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0127, 0.0169, 0.0101, 0.0109, 0.0153, 0.0144, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 16:38:49,580 INFO [train.py:893] (0/4) Epoch 8, batch 850, loss[loss=0.2703, simple_loss=0.3011, pruned_loss=0.1197, over 11770.00 frames. ], tot_loss[loss=0.2423, simple_loss=0.284, pruned_loss=0.1003, over 2618008.19 frames. ], batch size: 157, lr: 1.66e-02, grad_scale: 16.0 2023-04-16 16:38:53,931 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19750.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:39:23,250 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2023-04-16 16:39:26,304 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19785.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 16:39:35,730 INFO [train.py:893] (0/4) Epoch 8, batch 900, loss[loss=0.2498, simple_loss=0.2933, pruned_loss=0.1032, over 13519.00 frames. ], tot_loss[loss=0.2433, simple_loss=0.2842, pruned_loss=0.1012, over 2626508.23 frames. ], batch size: 98, lr: 1.65e-02, grad_scale: 16.0 2023-04-16 16:39:43,536 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6120, 3.4659, 2.7418, 3.1682, 2.8765, 1.7088, 3.3210, 1.8194], device='cuda:0'), covar=tensor([0.0678, 0.0453, 0.0468, 0.0352, 0.0855, 0.2096, 0.0850, 0.1561], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0100, 0.0106, 0.0090, 0.0128, 0.0156, 0.0106, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 16:39:46,759 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0610, 2.4405, 2.0355, 3.6562, 4.3120, 3.1702, 4.2911, 4.0255], device='cuda:0'), covar=tensor([0.0132, 0.0953, 0.1291, 0.0169, 0.0177, 0.0585, 0.0151, 0.0119], device='cuda:0'), in_proj_covar=tensor([0.0053, 0.0070, 0.0080, 0.0059, 0.0042, 0.0064, 0.0040, 0.0051], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 16:39:47,274 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.192e+02 3.363e+02 4.160e+02 5.148e+02 9.333e+02, threshold=8.319e+02, percent-clipped=2.0 2023-04-16 16:40:06,860 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 16:40:12,407 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19833.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 16:40:22,074 INFO [train.py:893] (0/4) Epoch 8, batch 950, loss[loss=0.2252, simple_loss=0.2693, pruned_loss=0.09056, over 13477.00 frames. ], tot_loss[loss=0.2441, simple_loss=0.2838, pruned_loss=0.1022, over 2636746.43 frames. ], batch size: 79, lr: 1.65e-02, grad_scale: 16.0 2023-04-16 16:41:08,910 INFO [train.py:893] (0/4) Epoch 8, batch 1000, loss[loss=0.1903, simple_loss=0.2428, pruned_loss=0.06894, over 13379.00 frames. ], tot_loss[loss=0.2415, simple_loss=0.2815, pruned_loss=0.1007, over 2639628.40 frames. ], batch size: 73, lr: 1.65e-02, grad_scale: 16.0 2023-04-16 16:41:19,649 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.219e+02 3.520e+02 4.323e+02 5.176e+02 1.444e+03, threshold=8.646e+02, percent-clipped=3.0 2023-04-16 16:41:54,452 INFO [train.py:893] (0/4) Epoch 8, batch 1050, loss[loss=0.2428, simple_loss=0.28, pruned_loss=0.1028, over 13518.00 frames. ], tot_loss[loss=0.2384, simple_loss=0.2792, pruned_loss=0.09883, over 2645105.18 frames. ], batch size: 83, lr: 1.65e-02, grad_scale: 16.0 2023-04-16 16:42:14,526 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.52 vs. limit=5.0 2023-04-16 16:42:21,716 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19974.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:42:39,622 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5571, 2.1849, 1.9316, 2.6698, 1.6489, 2.5059, 2.5146, 2.3372], device='cuda:0'), covar=tensor([0.0102, 0.0156, 0.0170, 0.0109, 0.0218, 0.0106, 0.0192, 0.0162], device='cuda:0'), in_proj_covar=tensor([0.0059, 0.0061, 0.0068, 0.0056, 0.0073, 0.0054, 0.0061, 0.0059], device='cuda:0'), out_proj_covar=tensor([6.8877e-05, 7.1765e-05, 8.3911e-05, 6.8272e-05, 8.8753e-05, 6.3723e-05, 7.3632e-05, 7.0045e-05], device='cuda:0') 2023-04-16 16:42:41,851 INFO [train.py:893] (0/4) Epoch 8, batch 1100, loss[loss=0.2248, simple_loss=0.2773, pruned_loss=0.08617, over 13524.00 frames. ], tot_loss[loss=0.2376, simple_loss=0.2795, pruned_loss=0.09786, over 2649704.21 frames. ], batch size: 91, lr: 1.65e-02, grad_scale: 16.0 2023-04-16 16:42:46,225 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-20000.pt 2023-04-16 16:42:55,814 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.721e+02 3.510e+02 4.200e+02 4.928e+02 1.765e+03, threshold=8.400e+02, percent-clipped=2.0 2023-04-16 16:43:09,140 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20020.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:43:31,185 INFO [train.py:893] (0/4) Epoch 8, batch 1150, loss[loss=0.2224, simple_loss=0.2605, pruned_loss=0.09212, over 13335.00 frames. ], tot_loss[loss=0.2363, simple_loss=0.2792, pruned_loss=0.0967, over 2646266.76 frames. ], batch size: 62, lr: 1.64e-02, grad_scale: 16.0 2023-04-16 16:43:36,350 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20050.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:43:40,157 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-16 16:43:44,094 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.50 vs. limit=5.0 2023-04-16 16:43:59,285 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.11 vs. limit=5.0 2023-04-16 16:44:03,277 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20081.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:44:11,167 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20089.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:44:15,942 INFO [train.py:893] (0/4) Epoch 8, batch 1200, loss[loss=0.2345, simple_loss=0.2841, pruned_loss=0.09248, over 13440.00 frames. ], tot_loss[loss=0.2363, simple_loss=0.2798, pruned_loss=0.09635, over 2648945.58 frames. ], batch size: 103, lr: 1.64e-02, grad_scale: 16.0 2023-04-16 16:44:18,556 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20098.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:44:19,386 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2992, 4.9135, 4.7097, 4.6709, 4.5757, 4.6636, 5.2099, 4.8152], device='cuda:0'), covar=tensor([0.0673, 0.0929, 0.2066, 0.2935, 0.0853, 0.1372, 0.0947, 0.0989], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0282, 0.0353, 0.0366, 0.0200, 0.0271, 0.0324, 0.0241], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 16:44:26,836 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.337e+02 3.392e+02 3.993e+02 4.511e+02 8.412e+02, threshold=7.987e+02, percent-clipped=1.0 2023-04-16 16:44:31,288 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20112.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:44:42,160 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 16:44:54,487 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 16:45:00,990 INFO [train.py:893] (0/4) Epoch 8, batch 1250, loss[loss=0.225, simple_loss=0.2696, pruned_loss=0.09017, over 13475.00 frames. ], tot_loss[loss=0.2371, simple_loss=0.2805, pruned_loss=0.09688, over 2649863.94 frames. ], batch size: 100, lr: 1.64e-02, grad_scale: 16.0 2023-04-16 16:45:06,222 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20150.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:45:27,100 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20173.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:45:47,253 INFO [train.py:893] (0/4) Epoch 8, batch 1300, loss[loss=0.2735, simple_loss=0.3081, pruned_loss=0.1194, over 13444.00 frames. ], tot_loss[loss=0.2387, simple_loss=0.2824, pruned_loss=0.09753, over 2655754.08 frames. ], batch size: 106, lr: 1.64e-02, grad_scale: 16.0 2023-04-16 16:45:55,851 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20204.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:45:59,489 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.442e+02 3.397e+02 3.954e+02 4.859e+02 9.243e+02, threshold=7.909e+02, percent-clipped=2.0 2023-04-16 16:46:33,718 INFO [train.py:893] (0/4) Epoch 8, batch 1350, loss[loss=0.2409, simple_loss=0.2843, pruned_loss=0.09875, over 13519.00 frames. ], tot_loss[loss=0.2383, simple_loss=0.2822, pruned_loss=0.09724, over 2659822.58 frames. ], batch size: 85, lr: 1.64e-02, grad_scale: 16.0 2023-04-16 16:46:52,190 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20265.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:47:00,978 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20274.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:47:19,766 INFO [train.py:893] (0/4) Epoch 8, batch 1400, loss[loss=0.2419, simple_loss=0.2801, pruned_loss=0.1018, over 13572.00 frames. ], tot_loss[loss=0.2373, simple_loss=0.2811, pruned_loss=0.09678, over 2659091.49 frames. ], batch size: 89, lr: 1.64e-02, grad_scale: 16.0 2023-04-16 16:47:31,198 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.220e+02 3.226e+02 3.715e+02 4.443e+02 7.884e+02, threshold=7.431e+02, percent-clipped=0.0 2023-04-16 16:47:44,600 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20322.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:48:00,097 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8440, 4.1504, 3.8439, 3.8969, 3.9510, 3.7256, 4.1576, 4.2065], device='cuda:0'), covar=tensor([0.0202, 0.0273, 0.0243, 0.0321, 0.0265, 0.0306, 0.0321, 0.0273], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0148, 0.0111, 0.0139, 0.0102, 0.0142, 0.0101, 0.0119], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:48:05,624 INFO [train.py:893] (0/4) Epoch 8, batch 1450, loss[loss=0.2135, simple_loss=0.269, pruned_loss=0.07899, over 13549.00 frames. ], tot_loss[loss=0.2374, simple_loss=0.2809, pruned_loss=0.09689, over 2661977.08 frames. ], batch size: 72, lr: 1.63e-02, grad_scale: 16.0 2023-04-16 16:48:33,533 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20376.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:48:44,983 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20388.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:48:51,088 INFO [train.py:893] (0/4) Epoch 8, batch 1500, loss[loss=0.1931, simple_loss=0.243, pruned_loss=0.07158, over 13387.00 frames. ], tot_loss[loss=0.2372, simple_loss=0.2808, pruned_loss=0.09684, over 2660277.69 frames. ], batch size: 65, lr: 1.63e-02, grad_scale: 16.0 2023-04-16 16:49:02,076 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.502e+02 3.447e+02 3.995e+02 5.100e+02 1.034e+03, threshold=7.990e+02, percent-clipped=5.0 2023-04-16 16:49:12,874 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20418.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:49:15,373 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4970, 2.5848, 2.8383, 4.1730, 3.7514, 4.1945, 3.3421, 2.5931], device='cuda:0'), covar=tensor([0.0337, 0.1120, 0.0843, 0.0042, 0.0216, 0.0040, 0.0615, 0.1033], device='cuda:0'), in_proj_covar=tensor([0.0100, 0.0148, 0.0154, 0.0079, 0.0099, 0.0074, 0.0148, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-04-16 16:49:37,033 INFO [train.py:893] (0/4) Epoch 8, batch 1550, loss[loss=0.2396, simple_loss=0.2897, pruned_loss=0.0948, over 13480.00 frames. ], tot_loss[loss=0.2371, simple_loss=0.2806, pruned_loss=0.09683, over 2657091.95 frames. ], batch size: 93, lr: 1.63e-02, grad_scale: 16.0 2023-04-16 16:49:37,293 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20445.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:49:40,795 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20449.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:49:41,068 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-16 16:49:57,709 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20468.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:50:08,357 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20479.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:50:10,993 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-16 16:50:22,568 INFO [train.py:893] (0/4) Epoch 8, batch 1600, loss[loss=0.2104, simple_loss=0.2609, pruned_loss=0.07991, over 13370.00 frames. ], tot_loss[loss=0.2365, simple_loss=0.2804, pruned_loss=0.09635, over 2663803.91 frames. ], batch size: 73, lr: 1.63e-02, grad_scale: 16.0 2023-04-16 16:50:33,978 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.804e+02 3.330e+02 4.045e+02 4.918e+02 9.492e+02, threshold=8.090e+02, percent-clipped=1.0 2023-04-16 16:50:43,072 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-16 16:51:08,737 INFO [train.py:893] (0/4) Epoch 8, batch 1650, loss[loss=0.2173, simple_loss=0.278, pruned_loss=0.07837, over 13446.00 frames. ], tot_loss[loss=0.2366, simple_loss=0.2812, pruned_loss=0.09599, over 2658829.77 frames. ], batch size: 106, lr: 1.63e-02, grad_scale: 16.0 2023-04-16 16:51:22,929 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20560.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:51:49,161 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20588.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:51:54,563 INFO [train.py:893] (0/4) Epoch 8, batch 1700, loss[loss=0.2347, simple_loss=0.2828, pruned_loss=0.0933, over 13562.00 frames. ], tot_loss[loss=0.2358, simple_loss=0.2809, pruned_loss=0.09533, over 2657216.90 frames. ], batch size: 78, lr: 1.62e-02, grad_scale: 16.0 2023-04-16 16:52:06,199 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.424e+02 3.366e+02 4.046e+02 4.775e+02 1.003e+03, threshold=8.092e+02, percent-clipped=1.0 2023-04-16 16:52:40,926 INFO [train.py:893] (0/4) Epoch 8, batch 1750, loss[loss=0.2718, simple_loss=0.3078, pruned_loss=0.1179, over 13250.00 frames. ], tot_loss[loss=0.2339, simple_loss=0.2793, pruned_loss=0.09428, over 2663930.12 frames. ], batch size: 124, lr: 1.62e-02, grad_scale: 16.0 2023-04-16 16:52:45,268 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20649.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:53:09,319 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20676.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:53:26,202 INFO [train.py:893] (0/4) Epoch 8, batch 1800, loss[loss=0.2189, simple_loss=0.273, pruned_loss=0.08237, over 13484.00 frames. ], tot_loss[loss=0.2331, simple_loss=0.2788, pruned_loss=0.09366, over 2664125.41 frames. ], batch size: 93, lr: 1.62e-02, grad_scale: 16.0 2023-04-16 16:53:34,738 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9351, 4.3975, 4.0995, 3.7447, 4.1237, 4.4857, 4.3262, 4.1697], device='cuda:0'), covar=tensor([0.0365, 0.0294, 0.0354, 0.1714, 0.0292, 0.0366, 0.0318, 0.0344], device='cuda:0'), in_proj_covar=tensor([0.0108, 0.0113, 0.0121, 0.0217, 0.0120, 0.0140, 0.0119, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-16 16:53:37,552 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.294e+02 3.249e+02 3.823e+02 4.877e+02 1.148e+03, threshold=7.645e+02, percent-clipped=3.0 2023-04-16 16:53:40,701 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.09 vs. limit=5.0 2023-04-16 16:53:53,241 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20724.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:53:55,806 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7727, 4.2885, 3.8811, 3.9761, 3.9719, 4.4745, 4.2279, 4.0232], device='cuda:0'), covar=tensor([0.0453, 0.0321, 0.0360, 0.1300, 0.0337, 0.0262, 0.0330, 0.0373], device='cuda:0'), in_proj_covar=tensor([0.0108, 0.0114, 0.0122, 0.0219, 0.0122, 0.0140, 0.0119, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-16 16:54:11,928 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20744.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:54:12,574 INFO [train.py:893] (0/4) Epoch 8, batch 1850, loss[loss=0.204, simple_loss=0.2491, pruned_loss=0.07946, over 13414.00 frames. ], tot_loss[loss=0.2327, simple_loss=0.2785, pruned_loss=0.09343, over 2661649.24 frames. ], batch size: 65, lr: 1.62e-02, grad_scale: 16.0 2023-04-16 16:54:12,853 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20745.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:54:15,136 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 16:54:33,361 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20768.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:54:39,669 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20774.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:54:56,662 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20793.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:54:58,137 INFO [train.py:893] (0/4) Epoch 8, batch 1900, loss[loss=0.1752, simple_loss=0.2226, pruned_loss=0.06392, over 13191.00 frames. ], tot_loss[loss=0.2325, simple_loss=0.278, pruned_loss=0.09353, over 2663639.23 frames. ], batch size: 58, lr: 1.62e-02, grad_scale: 16.0 2023-04-16 16:55:09,577 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.241e+02 3.078e+02 3.600e+02 4.474e+02 8.582e+02, threshold=7.200e+02, percent-clipped=1.0 2023-04-16 16:55:17,442 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20816.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:55:29,316 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-16 16:55:43,899 INFO [train.py:893] (0/4) Epoch 8, batch 1950, loss[loss=0.2564, simple_loss=0.2962, pruned_loss=0.1083, over 13459.00 frames. ], tot_loss[loss=0.2326, simple_loss=0.2775, pruned_loss=0.09389, over 2662923.80 frames. ], batch size: 103, lr: 1.61e-02, grad_scale: 16.0 2023-04-16 16:55:46,747 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6463, 1.6551, 3.4418, 3.2331, 3.3464, 2.4687, 3.3597, 2.5495], device='cuda:0'), covar=tensor([0.2148, 0.1812, 0.0112, 0.0236, 0.0214, 0.0809, 0.0177, 0.0899], device='cuda:0'), in_proj_covar=tensor([0.0198, 0.0187, 0.0099, 0.0108, 0.0109, 0.0152, 0.0111, 0.0162], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 16:55:57,904 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20860.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:56:29,847 INFO [train.py:893] (0/4) Epoch 8, batch 2000, loss[loss=0.2768, simple_loss=0.3098, pruned_loss=0.1219, over 13451.00 frames. ], tot_loss[loss=0.2352, simple_loss=0.2794, pruned_loss=0.09549, over 2662624.17 frames. ], batch size: 106, lr: 1.61e-02, grad_scale: 32.0 2023-04-16 16:56:34,829 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1887, 4.3000, 3.6669, 3.0352, 3.0953, 2.4893, 4.4952, 2.6732], device='cuda:0'), covar=tensor([0.1206, 0.0216, 0.0540, 0.1089, 0.0608, 0.2354, 0.0112, 0.2719], device='cuda:0'), in_proj_covar=tensor([0.0272, 0.0246, 0.0249, 0.0263, 0.0211, 0.0263, 0.0169, 0.0312], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 16:56:36,110 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 16:56:40,199 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.374e+02 3.423e+02 4.035e+02 4.744e+02 7.852e+02, threshold=8.070e+02, percent-clipped=3.0 2023-04-16 16:56:41,229 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20908.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:57:09,087 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6955, 3.6932, 2.9349, 3.6397, 3.0376, 1.7038, 3.6078, 1.7787], device='cuda:0'), covar=tensor([0.0782, 0.0491, 0.0496, 0.0219, 0.0791, 0.2135, 0.0793, 0.1659], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0098, 0.0102, 0.0085, 0.0122, 0.0151, 0.0107, 0.0121], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 16:57:13,978 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20944.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:57:15,186 INFO [train.py:893] (0/4) Epoch 8, batch 2050, loss[loss=0.1958, simple_loss=0.2423, pruned_loss=0.07464, over 13178.00 frames. ], tot_loss[loss=0.2362, simple_loss=0.2806, pruned_loss=0.09588, over 2664988.92 frames. ], batch size: 58, lr: 1.61e-02, grad_scale: 32.0 2023-04-16 16:57:22,667 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6931, 3.5067, 3.6413, 2.4218, 4.1101, 3.7703, 3.9023, 4.0728], device='cuda:0'), covar=tensor([0.0288, 0.0184, 0.0204, 0.1308, 0.0232, 0.0291, 0.0158, 0.0138], device='cuda:0'), in_proj_covar=tensor([0.0063, 0.0036, 0.0057, 0.0083, 0.0071, 0.0065, 0.0057, 0.0049], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 16:57:32,306 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-16 16:57:46,803 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7548, 3.9753, 2.6644, 4.1240, 3.8338, 2.0712, 3.3391, 2.5948], device='cuda:0'), covar=tensor([0.0229, 0.0300, 0.1022, 0.0109, 0.0221, 0.1383, 0.0533, 0.1399], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0128, 0.0167, 0.0106, 0.0111, 0.0153, 0.0146, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 16:58:00,254 INFO [train.py:893] (0/4) Epoch 8, batch 2100, loss[loss=0.2167, simple_loss=0.2609, pruned_loss=0.08628, over 13363.00 frames. ], tot_loss[loss=0.2358, simple_loss=0.2804, pruned_loss=0.09553, over 2665295.44 frames. ], batch size: 67, lr: 1.61e-02, grad_scale: 32.0 2023-04-16 16:58:12,041 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.172e+02 3.303e+02 3.918e+02 4.795e+02 7.015e+02, threshold=7.837e+02, percent-clipped=0.0 2023-04-16 16:58:12,446 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2361, 3.4310, 3.8907, 2.8141, 2.4925, 2.7592, 4.0515, 4.1652], device='cuda:0'), covar=tensor([0.0900, 0.0980, 0.0331, 0.1297, 0.1502, 0.1167, 0.0252, 0.0113], device='cuda:0'), in_proj_covar=tensor([0.0212, 0.0203, 0.0163, 0.0200, 0.0200, 0.0163, 0.0147, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 16:58:46,054 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21044.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:58:46,592 INFO [train.py:893] (0/4) Epoch 8, batch 2150, loss[loss=0.2356, simple_loss=0.2896, pruned_loss=0.09079, over 13392.00 frames. ], tot_loss[loss=0.2358, simple_loss=0.2806, pruned_loss=0.09554, over 2662253.00 frames. ], batch size: 73, lr: 1.61e-02, grad_scale: 32.0 2023-04-16 16:58:53,282 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21052.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:59:13,001 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21074.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:59:28,931 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21092.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:59:31,166 INFO [train.py:893] (0/4) Epoch 8, batch 2200, loss[loss=0.2259, simple_loss=0.2745, pruned_loss=0.0887, over 13539.00 frames. ], tot_loss[loss=0.2351, simple_loss=0.2802, pruned_loss=0.09505, over 2655007.65 frames. ], batch size: 98, lr: 1.61e-02, grad_scale: 32.0 2023-04-16 16:59:37,749 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4718, 3.0068, 2.8330, 4.3499, 4.9560, 3.7571, 4.9051, 4.4666], device='cuda:0'), covar=tensor([0.0065, 0.0575, 0.0625, 0.0088, 0.0052, 0.0302, 0.0054, 0.0062], device='cuda:0'), in_proj_covar=tensor([0.0054, 0.0074, 0.0082, 0.0062, 0.0044, 0.0066, 0.0040, 0.0053], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 16:59:42,195 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.587e+02 3.381e+02 4.062e+02 4.939e+02 9.669e+02, threshold=8.123e+02, percent-clipped=2.0 2023-04-16 16:59:47,549 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21113.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:59:52,450 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21118.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 16:59:55,590 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21122.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:00:17,221 INFO [train.py:893] (0/4) Epoch 8, batch 2250, loss[loss=0.2339, simple_loss=0.2738, pruned_loss=0.09703, over 13543.00 frames. ], tot_loss[loss=0.2338, simple_loss=0.2786, pruned_loss=0.09447, over 2657589.82 frames. ], batch size: 85, lr: 1.60e-02, grad_scale: 32.0 2023-04-16 17:00:42,030 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3784, 2.9885, 2.4552, 4.3055, 4.9422, 3.6865, 4.8370, 4.4143], device='cuda:0'), covar=tensor([0.0083, 0.0559, 0.0716, 0.0088, 0.0037, 0.0324, 0.0047, 0.0061], device='cuda:0'), in_proj_covar=tensor([0.0053, 0.0073, 0.0081, 0.0061, 0.0043, 0.0065, 0.0040, 0.0052], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:00:48,427 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21179.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:01:02,558 INFO [train.py:893] (0/4) Epoch 8, batch 2300, loss[loss=0.2272, simple_loss=0.2736, pruned_loss=0.09042, over 13553.00 frames. ], tot_loss[loss=0.2327, simple_loss=0.2776, pruned_loss=0.09386, over 2657904.71 frames. ], batch size: 89, lr: 1.60e-02, grad_scale: 32.0 2023-04-16 17:01:14,069 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.369e+02 3.357e+02 4.098e+02 4.938e+02 8.843e+02, threshold=8.196e+02, percent-clipped=1.0 2023-04-16 17:01:47,494 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21244.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:01:48,041 INFO [train.py:893] (0/4) Epoch 8, batch 2350, loss[loss=0.2491, simple_loss=0.2774, pruned_loss=0.1103, over 13352.00 frames. ], tot_loss[loss=0.2319, simple_loss=0.2769, pruned_loss=0.09348, over 2660346.66 frames. ], batch size: 62, lr: 1.60e-02, grad_scale: 32.0 2023-04-16 17:02:12,351 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 17:02:15,425 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.55 vs. limit=5.0 2023-04-16 17:02:27,503 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9626, 2.5490, 2.1784, 3.8121, 4.3974, 3.1712, 4.3466, 4.0386], device='cuda:0'), covar=tensor([0.0072, 0.0709, 0.0843, 0.0102, 0.0047, 0.0413, 0.0067, 0.0064], device='cuda:0'), in_proj_covar=tensor([0.0054, 0.0073, 0.0081, 0.0061, 0.0043, 0.0066, 0.0040, 0.0053], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:02:31,271 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21292.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:02:34,363 INFO [train.py:893] (0/4) Epoch 8, batch 2400, loss[loss=0.2398, simple_loss=0.2816, pruned_loss=0.09893, over 13482.00 frames. ], tot_loss[loss=0.2304, simple_loss=0.2756, pruned_loss=0.09259, over 2659481.04 frames. ], batch size: 81, lr: 1.60e-02, grad_scale: 32.0 2023-04-16 17:02:44,935 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.442e+02 3.489e+02 4.000e+02 4.780e+02 1.257e+03, threshold=8.000e+02, percent-clipped=2.0 2023-04-16 17:02:45,302 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.5291, 2.5308, 2.2315, 1.3970, 1.4268, 2.2637, 1.9327, 2.6336], device='cuda:0'), covar=tensor([0.0975, 0.0393, 0.0646, 0.1769, 0.0384, 0.0354, 0.0782, 0.0293], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0096, 0.0088, 0.0153, 0.0083, 0.0101, 0.0111, 0.0088], device='cuda:0'), out_proj_covar=tensor([8.8238e-05, 7.3602e-05, 7.2663e-05, 1.2491e-04, 7.2122e-05, 7.7123e-05, 8.7685e-05, 6.6168e-05], device='cuda:0') 2023-04-16 17:02:49,298 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21312.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:03:04,415 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21328.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:03:12,001 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2023-04-16 17:03:18,899 INFO [train.py:893] (0/4) Epoch 8, batch 2450, loss[loss=0.2342, simple_loss=0.2809, pruned_loss=0.0938, over 13515.00 frames. ], tot_loss[loss=0.2308, simple_loss=0.2758, pruned_loss=0.09288, over 2663168.17 frames. ], batch size: 91, lr: 1.60e-02, grad_scale: 32.0 2023-04-16 17:03:44,779 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21373.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:03:59,374 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21389.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:04:04,622 INFO [train.py:893] (0/4) Epoch 8, batch 2500, loss[loss=0.2002, simple_loss=0.2462, pruned_loss=0.07707, over 13360.00 frames. ], tot_loss[loss=0.231, simple_loss=0.2758, pruned_loss=0.09308, over 2660342.15 frames. ], batch size: 67, lr: 1.60e-02, grad_scale: 32.0 2023-04-16 17:04:05,123 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-16 17:04:15,350 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.088e+02 3.094e+02 3.679e+02 4.725e+02 7.841e+02, threshold=7.359e+02, percent-clipped=0.0 2023-04-16 17:04:16,441 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21408.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:04:49,334 INFO [train.py:893] (0/4) Epoch 8, batch 2550, loss[loss=0.247, simple_loss=0.2854, pruned_loss=0.1043, over 13188.00 frames. ], tot_loss[loss=0.2306, simple_loss=0.2753, pruned_loss=0.09292, over 2657738.68 frames. ], batch size: 132, lr: 1.59e-02, grad_scale: 32.0 2023-04-16 17:05:15,274 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 17:05:16,182 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21474.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:05:36,111 INFO [train.py:893] (0/4) Epoch 8, batch 2600, loss[loss=0.2522, simple_loss=0.2846, pruned_loss=0.1099, over 11861.00 frames. ], tot_loss[loss=0.2315, simple_loss=0.2756, pruned_loss=0.09367, over 2653559.98 frames. ], batch size: 157, lr: 1.59e-02, grad_scale: 32.0 2023-04-16 17:05:45,870 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.283e+02 3.241e+02 4.304e+02 5.606e+02 8.528e+02, threshold=8.608e+02, percent-clipped=3.0 2023-04-16 17:05:59,306 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2645, 4.7494, 4.6453, 4.6789, 4.6512, 4.6172, 5.2044, 4.7363], device='cuda:0'), covar=tensor([0.0718, 0.1042, 0.1981, 0.2386, 0.0671, 0.1316, 0.0826, 0.0965], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0289, 0.0358, 0.0371, 0.0202, 0.0280, 0.0335, 0.0251], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 17:06:07,214 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.33 vs. limit=5.0 2023-04-16 17:06:17,821 INFO [train.py:893] (0/4) Epoch 8, batch 2650, loss[loss=0.2613, simple_loss=0.2992, pruned_loss=0.1117, over 13444.00 frames. ], tot_loss[loss=0.2335, simple_loss=0.2775, pruned_loss=0.09473, over 2654008.83 frames. ], batch size: 95, lr: 1.59e-02, grad_scale: 16.0 2023-04-16 17:06:55,187 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-8.pt 2023-04-16 17:07:19,822 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 17:07:23,773 INFO [train.py:893] (0/4) Epoch 9, batch 0, loss[loss=0.2696, simple_loss=0.2991, pruned_loss=0.1201, over 13430.00 frames. ], tot_loss[loss=0.2696, simple_loss=0.2991, pruned_loss=0.1201, over 13430.00 frames. ], batch size: 106, lr: 1.50e-02, grad_scale: 16.0 2023-04-16 17:07:23,774 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 17:07:39,023 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8443, 5.0751, 5.0652, 4.9266, 5.0122, 4.9369, 5.1798, 5.1793], device='cuda:0'), covar=tensor([0.0143, 0.0166, 0.0116, 0.0200, 0.0170, 0.0172, 0.0147, 0.0129], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0149, 0.0114, 0.0140, 0.0109, 0.0148, 0.0102, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 17:07:46,617 INFO [train.py:927] (0/4) Epoch 9, validation: loss=0.1695, simple_loss=0.223, pruned_loss=0.05802, over 2446609.00 frames. 2023-04-16 17:07:46,618 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 17:07:48,575 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21595.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:08:00,060 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.409e+02 3.326e+02 3.891e+02 4.627e+02 8.563e+02, threshold=7.782e+02, percent-clipped=0.0 2023-04-16 17:08:04,385 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21613.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:08:26,648 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2023-04-16 17:08:32,558 INFO [train.py:893] (0/4) Epoch 9, batch 50, loss[loss=0.2334, simple_loss=0.2696, pruned_loss=0.09862, over 13339.00 frames. ], tot_loss[loss=0.2232, simple_loss=0.2673, pruned_loss=0.08957, over 603543.64 frames. ], batch size: 73, lr: 1.50e-02, grad_scale: 16.0 2023-04-16 17:08:43,500 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21656.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:08:55,511 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21668.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:08:57,828 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 17:08:57,828 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 17:08:57,828 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 17:08:57,840 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 17:08:57,857 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 17:08:57,877 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 17:08:57,887 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 17:08:59,343 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-16 17:09:00,573 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21674.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:09:09,266 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21684.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:09:18,937 INFO [train.py:893] (0/4) Epoch 9, batch 100, loss[loss=0.2557, simple_loss=0.2869, pruned_loss=0.1123, over 13535.00 frames. ], tot_loss[loss=0.2277, simple_loss=0.2701, pruned_loss=0.09268, over 1052092.02 frames. ], batch size: 87, lr: 1.50e-02, grad_scale: 16.0 2023-04-16 17:09:31,066 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.350e+02 3.037e+02 3.556e+02 4.306e+02 7.126e+02, threshold=7.112e+02, percent-clipped=0.0 2023-04-16 17:09:31,299 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21708.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:09:39,477 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2469, 4.7625, 4.6484, 4.6534, 4.3851, 4.6099, 5.1484, 4.7336], device='cuda:0'), covar=tensor([0.0685, 0.0960, 0.2124, 0.2684, 0.0830, 0.1351, 0.0890, 0.1009], device='cuda:0'), in_proj_covar=tensor([0.0262, 0.0292, 0.0371, 0.0382, 0.0207, 0.0284, 0.0340, 0.0256], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 17:10:03,136 INFO [train.py:893] (0/4) Epoch 9, batch 150, loss[loss=0.2422, simple_loss=0.2771, pruned_loss=0.1037, over 13529.00 frames. ], tot_loss[loss=0.2329, simple_loss=0.2743, pruned_loss=0.09579, over 1401268.23 frames. ], batch size: 83, lr: 1.50e-02, grad_scale: 16.0 2023-04-16 17:10:06,951 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2528, 4.4828, 2.8777, 4.4569, 4.2268, 2.3875, 3.5566, 2.5954], device='cuda:0'), covar=tensor([0.0231, 0.0248, 0.1358, 0.0144, 0.0187, 0.1588, 0.0566, 0.1913], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0128, 0.0166, 0.0110, 0.0111, 0.0154, 0.0145, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 17:10:14,141 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21756.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:10:22,341 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21763.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:10:31,548 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21774.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:10:50,059 INFO [train.py:893] (0/4) Epoch 9, batch 200, loss[loss=0.295, simple_loss=0.3293, pruned_loss=0.1303, over 13586.00 frames. ], tot_loss[loss=0.2359, simple_loss=0.2779, pruned_loss=0.09692, over 1679474.81 frames. ], batch size: 89, lr: 1.50e-02, grad_scale: 16.0 2023-04-16 17:11:02,744 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.061e+02 3.486e+02 4.224e+02 5.229e+02 1.059e+03, threshold=8.447e+02, percent-clipped=3.0 2023-04-16 17:11:16,561 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21822.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:11:18,406 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21824.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:11:35,554 INFO [train.py:893] (0/4) Epoch 9, batch 250, loss[loss=0.2282, simple_loss=0.2775, pruned_loss=0.08945, over 13454.00 frames. ], tot_loss[loss=0.2354, simple_loss=0.2782, pruned_loss=0.09627, over 1897983.55 frames. ], batch size: 103, lr: 1.50e-02, grad_scale: 16.0 2023-04-16 17:11:41,550 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21850.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:12:22,910 INFO [train.py:893] (0/4) Epoch 9, batch 300, loss[loss=0.1935, simple_loss=0.2363, pruned_loss=0.07533, over 12845.00 frames. ], tot_loss[loss=0.2353, simple_loss=0.2787, pruned_loss=0.09592, over 2065613.33 frames. ], batch size: 52, lr: 1.49e-02, grad_scale: 8.0 2023-04-16 17:12:35,096 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.144e+02 3.113e+02 3.883e+02 4.592e+02 8.864e+02, threshold=7.767e+02, percent-clipped=2.0 2023-04-16 17:12:37,774 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21911.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:13:06,479 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7562, 3.5455, 2.8687, 3.4229, 2.8810, 1.8593, 3.4786, 1.7670], device='cuda:0'), covar=tensor([0.0631, 0.0517, 0.0482, 0.0268, 0.0738, 0.1844, 0.0960, 0.1598], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0101, 0.0107, 0.0092, 0.0128, 0.0156, 0.0115, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:13:07,847 INFO [train.py:893] (0/4) Epoch 9, batch 350, loss[loss=0.2121, simple_loss=0.2557, pruned_loss=0.08426, over 13527.00 frames. ], tot_loss[loss=0.236, simple_loss=0.279, pruned_loss=0.09645, over 2198755.24 frames. ], batch size: 76, lr: 1.49e-02, grad_scale: 8.0 2023-04-16 17:13:15,296 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21951.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:13:29,426 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21968.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:13:30,129 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21969.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:13:31,060 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21970.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:13:45,492 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21984.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:13:53,623 INFO [train.py:893] (0/4) Epoch 9, batch 400, loss[loss=0.2212, simple_loss=0.2779, pruned_loss=0.08228, over 13521.00 frames. ], tot_loss[loss=0.235, simple_loss=0.2789, pruned_loss=0.09551, over 2306989.92 frames. ], batch size: 85, lr: 1.49e-02, grad_scale: 8.0 2023-04-16 17:13:59,156 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-22000.pt 2023-04-16 17:14:12,157 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.421e+02 3.487e+02 4.119e+02 4.911e+02 7.347e+02, threshold=8.238e+02, percent-clipped=0.0 2023-04-16 17:14:18,898 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22016.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:14:31,269 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22031.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:14:31,996 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22032.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:14:44,306 INFO [train.py:893] (0/4) Epoch 9, batch 450, loss[loss=0.2037, simple_loss=0.2485, pruned_loss=0.07945, over 13417.00 frames. ], tot_loss[loss=0.2354, simple_loss=0.2796, pruned_loss=0.09565, over 2388840.06 frames. ], batch size: 65, lr: 1.49e-02, grad_scale: 8.0 2023-04-16 17:15:08,182 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 17:15:27,921 INFO [train.py:893] (0/4) Epoch 9, batch 500, loss[loss=0.2207, simple_loss=0.2739, pruned_loss=0.08379, over 13582.00 frames. ], tot_loss[loss=0.2343, simple_loss=0.2793, pruned_loss=0.09461, over 2454982.08 frames. ], batch size: 89, lr: 1.49e-02, grad_scale: 8.0 2023-04-16 17:15:42,973 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.552e+02 3.311e+02 3.830e+02 4.684e+02 7.371e+02, threshold=7.659e+02, percent-clipped=0.0 2023-04-16 17:15:51,407 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22119.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:16:00,645 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6430, 3.7886, 4.3600, 3.2702, 2.6898, 3.1954, 4.5000, 4.5913], device='cuda:0'), covar=tensor([0.1016, 0.0968, 0.0274, 0.1242, 0.1501, 0.0896, 0.0193, 0.0118], device='cuda:0'), in_proj_covar=tensor([0.0207, 0.0204, 0.0162, 0.0198, 0.0196, 0.0162, 0.0147, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:16:01,962 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22130.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:16:14,796 INFO [train.py:893] (0/4) Epoch 9, batch 550, loss[loss=0.228, simple_loss=0.2736, pruned_loss=0.09123, over 13524.00 frames. ], tot_loss[loss=0.2348, simple_loss=0.2798, pruned_loss=0.09485, over 2499239.10 frames. ], batch size: 85, lr: 1.49e-02, grad_scale: 8.0 2023-04-16 17:16:27,086 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.78 vs. limit=5.0 2023-04-16 17:16:44,323 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22176.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:16:56,934 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22191.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:16:59,767 INFO [train.py:893] (0/4) Epoch 9, batch 600, loss[loss=0.2398, simple_loss=0.2831, pruned_loss=0.09825, over 13385.00 frames. ], tot_loss[loss=0.2327, simple_loss=0.2776, pruned_loss=0.0939, over 2539630.91 frames. ], batch size: 113, lr: 1.48e-02, grad_scale: 8.0 2023-04-16 17:17:11,934 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22206.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:17:14,236 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.989e+02 3.192e+02 3.802e+02 4.959e+02 1.314e+03, threshold=7.604e+02, percent-clipped=2.0 2023-04-16 17:17:15,376 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2086, 2.5459, 2.0182, 4.0583, 4.6111, 3.4181, 4.4950, 4.2395], device='cuda:0'), covar=tensor([0.0135, 0.0903, 0.1084, 0.0133, 0.0140, 0.0468, 0.0159, 0.0097], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0075, 0.0082, 0.0062, 0.0044, 0.0066, 0.0041, 0.0055], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:17:19,438 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22215.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:17:20,289 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7891, 3.5784, 2.8857, 3.4734, 3.1668, 1.7865, 3.6070, 1.9427], device='cuda:0'), covar=tensor([0.0628, 0.0442, 0.0455, 0.0278, 0.0606, 0.1918, 0.0556, 0.1389], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0101, 0.0104, 0.0089, 0.0125, 0.0153, 0.0114, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:17:39,839 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22237.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:17:39,872 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22237.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:17:45,624 INFO [train.py:893] (0/4) Epoch 9, batch 650, loss[loss=0.2188, simple_loss=0.2448, pruned_loss=0.09645, over 7381.00 frames. ], tot_loss[loss=0.2319, simple_loss=0.2765, pruned_loss=0.09369, over 2549553.33 frames. ], batch size: 29, lr: 1.48e-02, grad_scale: 8.0 2023-04-16 17:17:45,803 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.5011, 5.0988, 4.8232, 4.8810, 4.8591, 4.9490, 5.5350, 5.0488], device='cuda:0'), covar=tensor([0.0736, 0.1030, 0.2465, 0.2754, 0.0731, 0.1390, 0.0870, 0.1029], device='cuda:0'), in_proj_covar=tensor([0.0248, 0.0283, 0.0358, 0.0367, 0.0204, 0.0276, 0.0332, 0.0248], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 17:17:51,412 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22251.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:18:05,490 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-16 17:18:08,580 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22269.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:18:14,454 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22276.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:18:29,998 INFO [train.py:893] (0/4) Epoch 9, batch 700, loss[loss=0.2417, simple_loss=0.2897, pruned_loss=0.09685, over 13483.00 frames. ], tot_loss[loss=0.2305, simple_loss=0.2754, pruned_loss=0.09276, over 2575759.77 frames. ], batch size: 93, lr: 1.48e-02, grad_scale: 8.0 2023-04-16 17:18:34,182 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22298.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:18:34,887 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22299.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:18:42,934 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.092e+02 3.305e+02 4.012e+02 4.954e+02 1.106e+03, threshold=8.025e+02, percent-clipped=3.0 2023-04-16 17:18:49,080 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6409, 3.4023, 2.7757, 2.9235, 2.9405, 1.8819, 3.2657, 1.8448], device='cuda:0'), covar=tensor([0.0620, 0.0503, 0.0418, 0.0402, 0.0596, 0.1725, 0.0868, 0.1412], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0101, 0.0104, 0.0089, 0.0124, 0.0152, 0.0114, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:18:49,735 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22317.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:18:59,342 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22326.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:19:14,747 INFO [train.py:893] (0/4) Epoch 9, batch 750, loss[loss=0.2438, simple_loss=0.2883, pruned_loss=0.09963, over 13360.00 frames. ], tot_loss[loss=0.2308, simple_loss=0.2753, pruned_loss=0.09317, over 2596301.56 frames. ], batch size: 118, lr: 1.48e-02, grad_scale: 8.0 2023-04-16 17:19:53,847 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22385.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:20:01,662 INFO [train.py:893] (0/4) Epoch 9, batch 800, loss[loss=0.2378, simple_loss=0.2893, pruned_loss=0.09311, over 13487.00 frames. ], tot_loss[loss=0.2328, simple_loss=0.2771, pruned_loss=0.09423, over 2611522.22 frames. ], batch size: 93, lr: 1.48e-02, grad_scale: 8.0 2023-04-16 17:20:14,430 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.259e+02 3.347e+02 3.901e+02 4.564e+02 8.768e+02, threshold=7.802e+02, percent-clipped=1.0 2023-04-16 17:20:15,597 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8869, 3.7695, 3.9044, 2.4560, 4.3464, 4.0273, 3.9908, 4.4266], device='cuda:0'), covar=tensor([0.0276, 0.0181, 0.0164, 0.1239, 0.0210, 0.0228, 0.0206, 0.0165], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0037, 0.0057, 0.0081, 0.0071, 0.0068, 0.0058, 0.0050], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:20:24,311 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22419.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:20:45,105 INFO [train.py:893] (0/4) Epoch 9, batch 850, loss[loss=0.2348, simple_loss=0.2765, pruned_loss=0.09652, over 13443.00 frames. ], tot_loss[loss=0.2333, simple_loss=0.2781, pruned_loss=0.09427, over 2623307.71 frames. ], batch size: 95, lr: 1.48e-02, grad_scale: 8.0 2023-04-16 17:20:47,823 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22446.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:21:06,435 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22467.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:21:24,534 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22486.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:21:31,014 INFO [train.py:893] (0/4) Epoch 9, batch 900, loss[loss=0.2225, simple_loss=0.2661, pruned_loss=0.08948, over 13540.00 frames. ], tot_loss[loss=0.2339, simple_loss=0.2779, pruned_loss=0.09491, over 2629554.26 frames. ], batch size: 83, lr: 1.47e-02, grad_scale: 8.0 2023-04-16 17:21:40,967 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22506.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:21:43,897 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.166e+02 3.283e+02 3.837e+02 4.867e+02 8.567e+02, threshold=7.675e+02, percent-clipped=1.0 2023-04-16 17:21:54,774 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-04-16 17:22:00,018 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 17:22:04,285 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22532.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:22:15,461 INFO [train.py:893] (0/4) Epoch 9, batch 950, loss[loss=0.2207, simple_loss=0.2649, pruned_loss=0.08826, over 13040.00 frames. ], tot_loss[loss=0.2334, simple_loss=0.277, pruned_loss=0.0949, over 2633987.42 frames. ], batch size: 142, lr: 1.47e-02, grad_scale: 8.0 2023-04-16 17:22:24,614 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22554.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:22:39,551 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22571.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:22:59,554 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22593.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:23:00,240 INFO [train.py:893] (0/4) Epoch 9, batch 1000, loss[loss=0.2308, simple_loss=0.275, pruned_loss=0.09333, over 13543.00 frames. ], tot_loss[loss=0.2303, simple_loss=0.2742, pruned_loss=0.09322, over 2642588.48 frames. ], batch size: 85, lr: 1.47e-02, grad_scale: 8.0 2023-04-16 17:23:15,071 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.089e+02 2.969e+02 3.521e+02 4.471e+02 1.035e+03, threshold=7.042e+02, percent-clipped=1.0 2023-04-16 17:23:20,334 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2540, 3.4237, 3.3008, 3.8327, 1.9024, 2.4910, 3.4446, 1.9337], device='cuda:0'), covar=tensor([0.0099, 0.0602, 0.0761, 0.0553, 0.1732, 0.1129, 0.0719, 0.2233], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0160, 0.0176, 0.0174, 0.0173, 0.0180, 0.0164, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 17:23:30,018 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22626.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:23:42,433 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22639.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:23:46,957 INFO [train.py:893] (0/4) Epoch 9, batch 1050, loss[loss=0.2418, simple_loss=0.2928, pruned_loss=0.09536, over 13400.00 frames. ], tot_loss[loss=0.2294, simple_loss=0.2736, pruned_loss=0.09256, over 2637879.98 frames. ], batch size: 113, lr: 1.47e-02, grad_scale: 8.0 2023-04-16 17:24:13,848 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22674.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:24:30,787 INFO [train.py:893] (0/4) Epoch 9, batch 1100, loss[loss=0.2212, simple_loss=0.2739, pruned_loss=0.08425, over 13530.00 frames. ], tot_loss[loss=0.2285, simple_loss=0.2734, pruned_loss=0.09177, over 2640496.37 frames. ], batch size: 91, lr: 1.47e-02, grad_scale: 8.0 2023-04-16 17:24:38,029 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22700.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 17:24:45,841 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.277e+02 3.274e+02 3.908e+02 4.877e+02 1.233e+03, threshold=7.817e+02, percent-clipped=3.0 2023-04-16 17:25:15,067 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22741.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:25:17,338 INFO [train.py:893] (0/4) Epoch 9, batch 1150, loss[loss=0.2402, simple_loss=0.2878, pruned_loss=0.09635, over 13519.00 frames. ], tot_loss[loss=0.2267, simple_loss=0.2724, pruned_loss=0.09051, over 2644788.12 frames. ], batch size: 85, lr: 1.47e-02, grad_scale: 8.0 2023-04-16 17:25:47,632 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8309, 3.5552, 3.8239, 2.3948, 4.0884, 3.8334, 3.8649, 4.0594], device='cuda:0'), covar=tensor([0.0189, 0.0129, 0.0120, 0.0970, 0.0137, 0.0179, 0.0128, 0.0090], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0037, 0.0057, 0.0081, 0.0071, 0.0069, 0.0058, 0.0051], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:25:55,004 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22786.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:25:59,749 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2967, 2.0477, 2.4295, 3.6309, 3.2620, 3.7436, 2.9391, 1.9779], device='cuda:0'), covar=tensor([0.0235, 0.1093, 0.0800, 0.0057, 0.0275, 0.0033, 0.0528, 0.1150], device='cuda:0'), in_proj_covar=tensor([0.0104, 0.0148, 0.0153, 0.0083, 0.0103, 0.0071, 0.0150, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:26:02,611 INFO [train.py:893] (0/4) Epoch 9, batch 1200, loss[loss=0.2013, simple_loss=0.2509, pruned_loss=0.07586, over 12579.00 frames. ], tot_loss[loss=0.2275, simple_loss=0.2735, pruned_loss=0.09077, over 2646716.51 frames. ], batch size: 51, lr: 1.47e-02, grad_scale: 8.0 2023-04-16 17:26:02,922 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22794.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:26:16,683 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-04-16 17:26:16,761 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.433e+02 3.622e+02 4.040e+02 4.982e+02 8.406e+02, threshold=8.080e+02, percent-clipped=3.0 2023-04-16 17:26:28,054 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 17:26:31,481 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22825.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:26:38,646 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22832.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:26:40,124 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 17:26:40,208 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22834.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:26:47,182 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.77 vs. limit=5.0 2023-04-16 17:26:48,394 INFO [train.py:893] (0/4) Epoch 9, batch 1250, loss[loss=0.2471, simple_loss=0.2986, pruned_loss=0.09778, over 13483.00 frames. ], tot_loss[loss=0.2286, simple_loss=0.2743, pruned_loss=0.09144, over 2649110.63 frames. ], batch size: 93, lr: 1.46e-02, grad_scale: 8.0 2023-04-16 17:26:59,442 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22855.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:27:14,026 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22871.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:27:21,386 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22880.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:27:27,976 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22886.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:27:35,086 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22893.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:27:35,627 INFO [train.py:893] (0/4) Epoch 9, batch 1300, loss[loss=0.2501, simple_loss=0.2793, pruned_loss=0.1104, over 12138.00 frames. ], tot_loss[loss=0.2299, simple_loss=0.2757, pruned_loss=0.09208, over 2652570.40 frames. ], batch size: 157, lr: 1.46e-02, grad_scale: 8.0 2023-04-16 17:27:48,046 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.165e+02 3.359e+02 3.921e+02 4.678e+02 8.201e+02, threshold=7.842e+02, percent-clipped=1.0 2023-04-16 17:27:57,735 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22919.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:28:16,881 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22941.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:28:19,294 INFO [train.py:893] (0/4) Epoch 9, batch 1350, loss[loss=0.2217, simple_loss=0.2698, pruned_loss=0.08681, over 13554.00 frames. ], tot_loss[loss=0.2302, simple_loss=0.2762, pruned_loss=0.09209, over 2655741.66 frames. ], batch size: 76, lr: 1.46e-02, grad_scale: 8.0 2023-04-16 17:28:25,940 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4736, 3.7399, 3.6372, 4.1912, 1.8588, 3.0376, 3.7413, 2.1192], device='cuda:0'), covar=tensor([0.0089, 0.0465, 0.0579, 0.0353, 0.1690, 0.0826, 0.0585, 0.1943], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0165, 0.0177, 0.0176, 0.0173, 0.0182, 0.0165, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 17:29:04,924 INFO [train.py:893] (0/4) Epoch 9, batch 1400, loss[loss=0.2209, simple_loss=0.2588, pruned_loss=0.09152, over 13387.00 frames. ], tot_loss[loss=0.2294, simple_loss=0.2755, pruned_loss=0.0916, over 2655564.89 frames. ], batch size: 62, lr: 1.46e-02, grad_scale: 8.0 2023-04-16 17:29:05,938 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22995.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 17:29:13,954 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.05 vs. limit=5.0 2023-04-16 17:29:18,241 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.543e+02 3.220e+02 3.733e+02 4.427e+02 7.256e+02, threshold=7.465e+02, percent-clipped=0.0 2023-04-16 17:29:30,966 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-16 17:29:33,133 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4432, 3.5980, 4.0435, 2.9760, 2.7086, 2.9905, 4.2762, 4.3220], device='cuda:0'), covar=tensor([0.0971, 0.1003, 0.0298, 0.1256, 0.1309, 0.1003, 0.0214, 0.0124], device='cuda:0'), in_proj_covar=tensor([0.0212, 0.0209, 0.0165, 0.0202, 0.0199, 0.0167, 0.0153, 0.0115], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:29:47,725 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23041.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:29:49,842 INFO [train.py:893] (0/4) Epoch 9, batch 1450, loss[loss=0.2333, simple_loss=0.2835, pruned_loss=0.09156, over 13525.00 frames. ], tot_loss[loss=0.2294, simple_loss=0.2752, pruned_loss=0.09177, over 2657721.39 frames. ], batch size: 72, lr: 1.46e-02, grad_scale: 8.0 2023-04-16 17:30:04,342 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2023-04-16 17:30:12,139 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6734, 3.5089, 3.0333, 3.4683, 2.8979, 1.7249, 3.4335, 1.9363], device='cuda:0'), covar=tensor([0.0739, 0.0550, 0.0399, 0.0323, 0.0892, 0.2086, 0.1146, 0.1485], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0104, 0.0106, 0.0090, 0.0129, 0.0157, 0.0116, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:30:15,374 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5501, 3.7479, 4.2402, 3.1657, 2.8378, 3.0166, 4.3805, 4.4207], device='cuda:0'), covar=tensor([0.0830, 0.0869, 0.0263, 0.1147, 0.1230, 0.1060, 0.0172, 0.0108], device='cuda:0'), in_proj_covar=tensor([0.0211, 0.0208, 0.0164, 0.0202, 0.0198, 0.0167, 0.0152, 0.0115], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:30:29,733 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23089.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:30:33,729 INFO [train.py:893] (0/4) Epoch 9, batch 1500, loss[loss=0.2383, simple_loss=0.2896, pruned_loss=0.0935, over 13448.00 frames. ], tot_loss[loss=0.2296, simple_loss=0.2754, pruned_loss=0.09193, over 2661716.47 frames. ], batch size: 106, lr: 1.46e-02, grad_scale: 8.0 2023-04-16 17:30:47,117 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23108.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:30:47,595 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.350e+02 3.219e+02 3.782e+02 4.835e+02 1.139e+03, threshold=7.564e+02, percent-clipped=5.0 2023-04-16 17:31:14,875 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.5751, 2.5672, 1.6356, 1.3871, 1.3211, 2.0668, 1.9095, 2.6591], device='cuda:0'), covar=tensor([0.0926, 0.0393, 0.1283, 0.1732, 0.0219, 0.0404, 0.0790, 0.0213], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0100, 0.0090, 0.0163, 0.0084, 0.0110, 0.0119, 0.0093], device='cuda:0'), out_proj_covar=tensor([9.3623e-05, 7.7156e-05, 7.4235e-05, 1.3112e-04, 7.0724e-05, 8.3573e-05, 9.3913e-05, 6.9219e-05], device='cuda:0') 2023-04-16 17:31:20,073 INFO [train.py:893] (0/4) Epoch 9, batch 1550, loss[loss=0.2129, simple_loss=0.2556, pruned_loss=0.08511, over 13555.00 frames. ], tot_loss[loss=0.2293, simple_loss=0.2751, pruned_loss=0.09176, over 2656683.80 frames. ], batch size: 85, lr: 1.45e-02, grad_scale: 8.0 2023-04-16 17:31:25,192 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23150.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:31:42,409 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23169.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:31:48,940 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-16 17:31:53,626 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23181.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:32:04,386 INFO [train.py:893] (0/4) Epoch 9, batch 1600, loss[loss=0.2433, simple_loss=0.2941, pruned_loss=0.09621, over 13363.00 frames. ], tot_loss[loss=0.2291, simple_loss=0.275, pruned_loss=0.09155, over 2657735.01 frames. ], batch size: 109, lr: 1.45e-02, grad_scale: 8.0 2023-04-16 17:32:19,475 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.289e+02 3.375e+02 3.850e+02 4.710e+02 9.191e+02, threshold=7.699e+02, percent-clipped=3.0 2023-04-16 17:32:26,452 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23217.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:32:37,669 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23229.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:32:50,309 INFO [train.py:893] (0/4) Epoch 9, batch 1650, loss[loss=0.2317, simple_loss=0.2762, pruned_loss=0.09357, over 13367.00 frames. ], tot_loss[loss=0.2279, simple_loss=0.2744, pruned_loss=0.09067, over 2653199.70 frames. ], batch size: 109, lr: 1.45e-02, grad_scale: 8.0 2023-04-16 17:33:16,708 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.1897, 2.0367, 4.1781, 3.8602, 4.0277, 3.2300, 3.7408, 3.0316], device='cuda:0'), covar=tensor([0.1888, 0.1713, 0.0068, 0.0201, 0.0142, 0.0571, 0.0199, 0.0758], device='cuda:0'), in_proj_covar=tensor([0.0201, 0.0185, 0.0100, 0.0112, 0.0112, 0.0158, 0.0117, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:33:20,710 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23278.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:33:31,049 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23290.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:33:34,770 INFO [train.py:893] (0/4) Epoch 9, batch 1700, loss[loss=0.2225, simple_loss=0.2714, pruned_loss=0.08677, over 13520.00 frames. ], tot_loss[loss=0.2277, simple_loss=0.2745, pruned_loss=0.09047, over 2655326.60 frames. ], batch size: 98, lr: 1.45e-02, grad_scale: 8.0 2023-04-16 17:33:35,823 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23295.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:33:43,897 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.68 vs. limit=5.0 2023-04-16 17:33:48,288 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.390e+02 3.407e+02 3.875e+02 4.640e+02 9.292e+02, threshold=7.749e+02, percent-clipped=4.0 2023-04-16 17:34:13,671 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2023-04-16 17:34:19,232 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23343.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:34:19,886 INFO [train.py:893] (0/4) Epoch 9, batch 1750, loss[loss=0.2411, simple_loss=0.2857, pruned_loss=0.09823, over 13240.00 frames. ], tot_loss[loss=0.227, simple_loss=0.2738, pruned_loss=0.09013, over 2655347.28 frames. ], batch size: 132, lr: 1.45e-02, grad_scale: 8.0 2023-04-16 17:35:05,503 INFO [train.py:893] (0/4) Epoch 9, batch 1800, loss[loss=0.2224, simple_loss=0.2785, pruned_loss=0.08319, over 13383.00 frames. ], tot_loss[loss=0.2251, simple_loss=0.2723, pruned_loss=0.08896, over 2659636.95 frames. ], batch size: 113, lr: 1.45e-02, grad_scale: 8.0 2023-04-16 17:35:15,816 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0000, 4.3988, 4.0922, 4.0684, 4.1200, 4.0317, 4.3619, 4.3755], device='cuda:0'), covar=tensor([0.0177, 0.0196, 0.0207, 0.0288, 0.0298, 0.0233, 0.0268, 0.0221], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0153, 0.0119, 0.0142, 0.0110, 0.0147, 0.0103, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 17:35:18,934 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.368e+02 3.089e+02 3.721e+02 4.545e+02 8.418e+02, threshold=7.442e+02, percent-clipped=1.0 2023-04-16 17:35:39,034 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-16 17:35:49,971 INFO [train.py:893] (0/4) Epoch 9, batch 1850, loss[loss=0.2148, simple_loss=0.262, pruned_loss=0.08386, over 13527.00 frames. ], tot_loss[loss=0.2243, simple_loss=0.2717, pruned_loss=0.08842, over 2660456.97 frames. ], batch size: 85, lr: 1.45e-02, grad_scale: 8.0 2023-04-16 17:35:51,498 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 17:35:55,675 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23450.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:36:08,523 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23464.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:36:23,935 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23481.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:36:31,048 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23488.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:36:35,754 INFO [train.py:893] (0/4) Epoch 9, batch 1900, loss[loss=0.2216, simple_loss=0.2697, pruned_loss=0.0867, over 13412.00 frames. ], tot_loss[loss=0.2239, simple_loss=0.271, pruned_loss=0.08843, over 2658046.08 frames. ], batch size: 95, lr: 1.44e-02, grad_scale: 8.0 2023-04-16 17:36:39,366 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23498.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:36:49,682 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.068e+02 3.397e+02 4.025e+02 4.847e+02 8.504e+02, threshold=8.050e+02, percent-clipped=4.0 2023-04-16 17:37:07,613 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23529.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:37:21,170 INFO [train.py:893] (0/4) Epoch 9, batch 1950, loss[loss=0.2414, simple_loss=0.2855, pruned_loss=0.09868, over 13443.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.2703, pruned_loss=0.0883, over 2655977.29 frames. ], batch size: 106, lr: 1.44e-02, grad_scale: 8.0 2023-04-16 17:37:26,193 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23549.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 17:37:47,807 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23573.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:37:58,607 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-04-16 17:37:59,154 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23585.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:38:06,244 INFO [train.py:893] (0/4) Epoch 9, batch 2000, loss[loss=0.1853, simple_loss=0.2435, pruned_loss=0.06358, over 13542.00 frames. ], tot_loss[loss=0.2267, simple_loss=0.2733, pruned_loss=0.09005, over 2656732.48 frames. ], batch size: 76, lr: 1.44e-02, grad_scale: 8.0 2023-04-16 17:38:09,768 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 17:38:21,023 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.996e+02 3.137e+02 3.681e+02 4.329e+02 7.461e+02, threshold=7.363e+02, percent-clipped=0.0 2023-04-16 17:38:52,214 INFO [train.py:893] (0/4) Epoch 9, batch 2050, loss[loss=0.2471, simple_loss=0.2908, pruned_loss=0.1017, over 13467.00 frames. ], tot_loss[loss=0.2291, simple_loss=0.2752, pruned_loss=0.09154, over 2660215.81 frames. ], batch size: 79, lr: 1.44e-02, grad_scale: 8.0 2023-04-16 17:39:38,269 INFO [train.py:893] (0/4) Epoch 9, batch 2100, loss[loss=0.2037, simple_loss=0.2588, pruned_loss=0.07433, over 13478.00 frames. ], tot_loss[loss=0.2272, simple_loss=0.2735, pruned_loss=0.0904, over 2660707.79 frames. ], batch size: 79, lr: 1.44e-02, grad_scale: 8.0 2023-04-16 17:39:51,092 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.062e+02 3.176e+02 3.858e+02 4.612e+02 8.885e+02, threshold=7.716e+02, percent-clipped=1.0 2023-04-16 17:40:23,388 INFO [train.py:893] (0/4) Epoch 9, batch 2150, loss[loss=0.2291, simple_loss=0.2747, pruned_loss=0.0918, over 13064.00 frames. ], tot_loss[loss=0.2254, simple_loss=0.2724, pruned_loss=0.0892, over 2663228.24 frames. ], batch size: 142, lr: 1.44e-02, grad_scale: 8.0 2023-04-16 17:40:41,956 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23764.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:40:44,666 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-16 17:41:09,293 INFO [train.py:893] (0/4) Epoch 9, batch 2200, loss[loss=0.2477, simple_loss=0.2968, pruned_loss=0.09937, over 13496.00 frames. ], tot_loss[loss=0.2256, simple_loss=0.273, pruned_loss=0.08912, over 2667273.99 frames. ], batch size: 93, lr: 1.44e-02, grad_scale: 8.0 2023-04-16 17:41:22,565 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.574e+02 3.265e+02 4.064e+02 4.864e+02 9.339e+02, threshold=8.128e+02, percent-clipped=4.0 2023-04-16 17:41:25,120 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23812.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:41:54,097 INFO [train.py:893] (0/4) Epoch 9, batch 2250, loss[loss=0.2014, simple_loss=0.2547, pruned_loss=0.07407, over 13556.00 frames. ], tot_loss[loss=0.2242, simple_loss=0.2713, pruned_loss=0.08855, over 2665002.48 frames. ], batch size: 78, lr: 1.43e-02, grad_scale: 8.0 2023-04-16 17:41:54,288 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23844.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 17:42:21,250 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23873.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:42:32,573 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23885.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:42:33,329 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9887, 2.6620, 2.0186, 3.7913, 4.5716, 3.4510, 4.4375, 4.1100], device='cuda:0'), covar=tensor([0.0101, 0.0741, 0.0963, 0.0107, 0.0052, 0.0381, 0.0070, 0.0074], device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0076, 0.0085, 0.0063, 0.0047, 0.0068, 0.0042, 0.0056], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:42:40,219 INFO [train.py:893] (0/4) Epoch 9, batch 2300, loss[loss=0.2211, simple_loss=0.2723, pruned_loss=0.08497, over 13528.00 frames. ], tot_loss[loss=0.2243, simple_loss=0.2719, pruned_loss=0.08831, over 2666970.15 frames. ], batch size: 98, lr: 1.43e-02, grad_scale: 16.0 2023-04-16 17:42:53,342 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.220e+02 3.044e+02 3.754e+02 4.701e+02 8.519e+02, threshold=7.507e+02, percent-clipped=1.0 2023-04-16 17:43:04,668 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23921.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:43:15,945 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23933.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:43:25,985 INFO [train.py:893] (0/4) Epoch 9, batch 2350, loss[loss=0.2375, simple_loss=0.2792, pruned_loss=0.09789, over 13374.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.2713, pruned_loss=0.08779, over 2667893.93 frames. ], batch size: 109, lr: 1.43e-02, grad_scale: 16.0 2023-04-16 17:43:45,713 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 17:44:10,690 INFO [train.py:893] (0/4) Epoch 9, batch 2400, loss[loss=0.2209, simple_loss=0.2676, pruned_loss=0.08712, over 13266.00 frames. ], tot_loss[loss=0.2226, simple_loss=0.2703, pruned_loss=0.08746, over 2664991.33 frames. ], batch size: 124, lr: 1.43e-02, grad_scale: 16.0 2023-04-16 17:44:16,869 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-24000.pt 2023-04-16 17:44:29,305 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.691e+02 3.069e+02 3.860e+02 4.518e+02 6.769e+02, threshold=7.720e+02, percent-clipped=0.0 2023-04-16 17:44:54,829 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24037.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:45:01,087 INFO [train.py:893] (0/4) Epoch 9, batch 2450, loss[loss=0.2357, simple_loss=0.279, pruned_loss=0.09625, over 11717.00 frames. ], tot_loss[loss=0.2226, simple_loss=0.2704, pruned_loss=0.0874, over 2668368.09 frames. ], batch size: 157, lr: 1.43e-02, grad_scale: 16.0 2023-04-16 17:45:42,591 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6168, 2.9724, 2.5923, 2.8251, 2.8081, 1.8451, 2.9890, 1.8192], device='cuda:0'), covar=tensor([0.0558, 0.0833, 0.0411, 0.0435, 0.0630, 0.1872, 0.0889, 0.1415], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0106, 0.0107, 0.0093, 0.0127, 0.0157, 0.0116, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:45:45,262 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-16 17:45:46,335 INFO [train.py:893] (0/4) Epoch 9, batch 2500, loss[loss=0.2519, simple_loss=0.2963, pruned_loss=0.1037, over 13536.00 frames. ], tot_loss[loss=0.2219, simple_loss=0.27, pruned_loss=0.08691, over 2669907.20 frames. ], batch size: 98, lr: 1.43e-02, grad_scale: 16.0 2023-04-16 17:45:49,999 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24098.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:46:00,013 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.936e+02 3.188e+02 3.790e+02 4.783e+02 9.771e+02, threshold=7.581e+02, percent-clipped=3.0 2023-04-16 17:46:31,825 INFO [train.py:893] (0/4) Epoch 9, batch 2550, loss[loss=0.211, simple_loss=0.2675, pruned_loss=0.07724, over 13392.00 frames. ], tot_loss[loss=0.2219, simple_loss=0.27, pruned_loss=0.0869, over 2669730.08 frames. ], batch size: 113, lr: 1.43e-02, grad_scale: 16.0 2023-04-16 17:46:32,043 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=24144.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 17:46:48,521 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2024, 4.6295, 4.4088, 3.9823, 4.3749, 4.7338, 4.6320, 4.4497], device='cuda:0'), covar=tensor([0.0321, 0.0251, 0.0308, 0.1362, 0.0281, 0.0272, 0.0299, 0.0281], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0118, 0.0128, 0.0225, 0.0127, 0.0143, 0.0128, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-16 17:46:52,934 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 17:46:59,025 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0281, 4.5267, 4.3491, 4.2687, 4.2443, 4.1555, 4.5313, 4.4720], device='cuda:0'), covar=tensor([0.0206, 0.0199, 0.0167, 0.0281, 0.0305, 0.0220, 0.0318, 0.0239], device='cuda:0'), in_proj_covar=tensor([0.0106, 0.0156, 0.0123, 0.0144, 0.0112, 0.0151, 0.0105, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 17:47:14,707 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-04-16 17:47:15,067 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=24192.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:47:15,970 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2275, 2.6011, 1.9180, 4.0745, 4.6640, 3.4654, 4.5593, 4.2571], device='cuda:0'), covar=tensor([0.0090, 0.0760, 0.1042, 0.0102, 0.0057, 0.0388, 0.0075, 0.0067], device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0077, 0.0086, 0.0063, 0.0047, 0.0069, 0.0043, 0.0057], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:47:16,501 INFO [train.py:893] (0/4) Epoch 9, batch 2600, loss[loss=0.2294, simple_loss=0.2775, pruned_loss=0.09069, over 13543.00 frames. ], tot_loss[loss=0.2226, simple_loss=0.2703, pruned_loss=0.08752, over 2669887.73 frames. ], batch size: 87, lr: 1.42e-02, grad_scale: 16.0 2023-04-16 17:47:21,592 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.7584, 4.5747, 4.7653, 4.6677, 5.0242, 4.4956, 5.0500, 5.0272], device='cuda:0'), covar=tensor([0.0315, 0.0558, 0.0619, 0.0472, 0.0537, 0.0746, 0.0508, 0.0484], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0232, 0.0219, 0.0169, 0.0317, 0.0263, 0.0198, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 17:47:30,176 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.381e+02 3.194e+02 3.780e+02 4.738e+02 1.181e+03, threshold=7.561e+02, percent-clipped=2.0 2023-04-16 17:47:58,330 INFO [train.py:893] (0/4) Epoch 9, batch 2650, loss[loss=0.2322, simple_loss=0.2824, pruned_loss=0.09104, over 13380.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.2712, pruned_loss=0.08848, over 2664858.16 frames. ], batch size: 109, lr: 1.42e-02, grad_scale: 16.0 2023-04-16 17:48:37,234 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-9.pt 2023-04-16 17:49:01,151 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 17:49:04,932 INFO [train.py:893] (0/4) Epoch 10, batch 0, loss[loss=0.2216, simple_loss=0.2677, pruned_loss=0.08773, over 13574.00 frames. ], tot_loss[loss=0.2216, simple_loss=0.2677, pruned_loss=0.08773, over 13574.00 frames. ], batch size: 89, lr: 1.35e-02, grad_scale: 16.0 2023-04-16 17:49:04,932 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 17:49:27,008 INFO [train.py:927] (0/4) Epoch 10, validation: loss=0.1639, simple_loss=0.2187, pruned_loss=0.05457, over 2446609.00 frames. 2023-04-16 17:49:27,009 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 17:49:41,914 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.104e+02 3.192e+02 3.835e+02 4.467e+02 7.040e+02, threshold=7.670e+02, percent-clipped=0.0 2023-04-16 17:50:12,914 INFO [train.py:893] (0/4) Epoch 10, batch 50, loss[loss=0.2352, simple_loss=0.2896, pruned_loss=0.09035, over 13513.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.2637, pruned_loss=0.08808, over 595889.55 frames. ], batch size: 91, lr: 1.35e-02, grad_scale: 16.0 2023-04-16 17:50:36,716 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 17:50:36,717 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 17:50:36,717 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 17:50:36,723 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 17:50:37,435 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 17:50:37,452 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 17:50:37,462 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 17:50:58,022 INFO [train.py:893] (0/4) Epoch 10, batch 100, loss[loss=0.2317, simple_loss=0.2708, pruned_loss=0.09624, over 13527.00 frames. ], tot_loss[loss=0.2239, simple_loss=0.267, pruned_loss=0.09041, over 1053995.81 frames. ], batch size: 83, lr: 1.35e-02, grad_scale: 16.0 2023-04-16 17:50:58,229 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24393.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:51:13,610 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.019e+02 3.298e+02 3.891e+02 4.543e+02 7.563e+02, threshold=7.783e+02, percent-clipped=0.0 2023-04-16 17:51:13,912 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.4978, 2.1116, 2.0200, 2.5967, 1.6604, 2.5462, 2.4772, 2.2609], device='cuda:0'), covar=tensor([0.0067, 0.0170, 0.0167, 0.0105, 0.0208, 0.0098, 0.0191, 0.0118], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0071, 0.0078, 0.0068, 0.0087, 0.0064, 0.0070, 0.0066], device='cuda:0'), out_proj_covar=tensor([7.4921e-05, 8.2682e-05, 9.2525e-05, 7.9707e-05, 1.0289e-04, 7.2997e-05, 8.2796e-05, 7.5435e-05], device='cuda:0') 2023-04-16 17:51:44,485 INFO [train.py:893] (0/4) Epoch 10, batch 150, loss[loss=0.2319, simple_loss=0.2839, pruned_loss=0.08994, over 13381.00 frames. ], tot_loss[loss=0.2271, simple_loss=0.2704, pruned_loss=0.0919, over 1400135.10 frames. ], batch size: 118, lr: 1.35e-02, grad_scale: 16.0 2023-04-16 17:52:23,678 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1456, 5.0208, 5.1407, 5.0409, 5.4932, 4.8781, 5.4744, 5.4817], device='cuda:0'), covar=tensor([0.0350, 0.0531, 0.0658, 0.0499, 0.0534, 0.0861, 0.0557, 0.0474], device='cuda:0'), in_proj_covar=tensor([0.0213, 0.0237, 0.0223, 0.0171, 0.0322, 0.0268, 0.0202, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 17:52:30,652 INFO [train.py:893] (0/4) Epoch 10, batch 200, loss[loss=0.2186, simple_loss=0.2755, pruned_loss=0.08084, over 13434.00 frames. ], tot_loss[loss=0.2277, simple_loss=0.2718, pruned_loss=0.0918, over 1675288.80 frames. ], batch size: 106, lr: 1.35e-02, grad_scale: 16.0 2023-04-16 17:52:44,380 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.196e+02 3.067e+02 3.720e+02 4.944e+02 1.446e+03, threshold=7.439e+02, percent-clipped=2.0 2023-04-16 17:53:01,430 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8201, 2.6408, 2.4151, 2.8432, 2.2674, 2.9850, 2.5249, 2.7460], device='cuda:0'), covar=tensor([0.0066, 0.0146, 0.0161, 0.0151, 0.0172, 0.0099, 0.0270, 0.0146], device='cuda:0'), in_proj_covar=tensor([0.0068, 0.0073, 0.0080, 0.0070, 0.0087, 0.0066, 0.0072, 0.0068], device='cuda:0'), out_proj_covar=tensor([7.6208e-05, 8.4692e-05, 9.4502e-05, 8.1699e-05, 1.0353e-04, 7.5024e-05, 8.5238e-05, 7.7414e-05], device='cuda:0') 2023-04-16 17:53:09,723 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9334, 4.1339, 3.3258, 2.7040, 2.8083, 2.3012, 4.2430, 2.3587], device='cuda:0'), covar=tensor([0.1334, 0.0264, 0.0719, 0.1544, 0.0739, 0.2877, 0.0199, 0.3161], device='cuda:0'), in_proj_covar=tensor([0.0282, 0.0253, 0.0259, 0.0273, 0.0215, 0.0274, 0.0180, 0.0324], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 17:53:16,043 INFO [train.py:893] (0/4) Epoch 10, batch 250, loss[loss=0.2098, simple_loss=0.2604, pruned_loss=0.0796, over 13195.00 frames. ], tot_loss[loss=0.2274, simple_loss=0.2718, pruned_loss=0.09156, over 1892032.35 frames. ], batch size: 132, lr: 1.34e-02, grad_scale: 16.0 2023-04-16 17:53:19,769 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24547.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:53:23,833 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.7523, 4.0291, 3.6053, 4.5742, 2.3679, 2.8190, 4.0649, 2.2993], device='cuda:0'), covar=tensor([0.0075, 0.0464, 0.0754, 0.0341, 0.1589, 0.1202, 0.0641, 0.1982], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0164, 0.0178, 0.0181, 0.0171, 0.0181, 0.0163, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 17:53:41,241 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24569.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:54:03,217 INFO [train.py:893] (0/4) Epoch 10, batch 300, loss[loss=0.2225, simple_loss=0.2753, pruned_loss=0.08488, over 13446.00 frames. ], tot_loss[loss=0.2267, simple_loss=0.2716, pruned_loss=0.09083, over 2054208.45 frames. ], batch size: 103, lr: 1.34e-02, grad_scale: 16.0 2023-04-16 17:54:17,124 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24608.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:54:17,541 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.057e+02 2.986e+02 3.516e+02 4.571e+02 9.026e+02, threshold=7.032e+02, percent-clipped=2.0 2023-04-16 17:54:38,202 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24630.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:54:48,634 INFO [train.py:893] (0/4) Epoch 10, batch 350, loss[loss=0.2302, simple_loss=0.2797, pruned_loss=0.09032, over 13449.00 frames. ], tot_loss[loss=0.2275, simple_loss=0.2727, pruned_loss=0.09118, over 2185922.57 frames. ], batch size: 106, lr: 1.34e-02, grad_scale: 16.0 2023-04-16 17:55:11,772 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1514, 4.2365, 2.6471, 4.2265, 4.0737, 2.2157, 3.5611, 2.6278], device='cuda:0'), covar=tensor([0.0227, 0.0218, 0.1260, 0.0169, 0.0171, 0.1514, 0.0557, 0.1505], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0135, 0.0173, 0.0117, 0.0113, 0.0155, 0.0149, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 17:55:15,310 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-16 17:55:34,162 INFO [train.py:893] (0/4) Epoch 10, batch 400, loss[loss=0.2516, simple_loss=0.2946, pruned_loss=0.1043, over 13386.00 frames. ], tot_loss[loss=0.2284, simple_loss=0.2737, pruned_loss=0.09152, over 2283342.22 frames. ], batch size: 113, lr: 1.34e-02, grad_scale: 16.0 2023-04-16 17:55:34,360 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=24693.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:55:48,032 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.056e+02 3.256e+02 3.949e+02 5.125e+02 8.205e+02, threshold=7.898e+02, percent-clipped=4.0 2023-04-16 17:55:54,943 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-16 17:56:17,965 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=24741.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:56:19,376 INFO [train.py:893] (0/4) Epoch 10, batch 450, loss[loss=0.2199, simple_loss=0.259, pruned_loss=0.09041, over 13419.00 frames. ], tot_loss[loss=0.2287, simple_loss=0.2745, pruned_loss=0.09144, over 2371561.35 frames. ], batch size: 65, lr: 1.34e-02, grad_scale: 16.0 2023-04-16 17:56:39,224 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24765.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:56:41,667 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5060, 3.0335, 2.6207, 4.4509, 5.0120, 3.8257, 4.8764, 4.5207], device='cuda:0'), covar=tensor([0.0079, 0.0597, 0.0757, 0.0078, 0.0048, 0.0309, 0.0066, 0.0064], device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0077, 0.0084, 0.0063, 0.0047, 0.0069, 0.0043, 0.0057], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 17:56:43,009 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 17:57:05,060 INFO [train.py:893] (0/4) Epoch 10, batch 500, loss[loss=0.2187, simple_loss=0.2714, pruned_loss=0.08301, over 13504.00 frames. ], tot_loss[loss=0.2275, simple_loss=0.2739, pruned_loss=0.09056, over 2434381.71 frames. ], batch size: 93, lr: 1.34e-02, grad_scale: 16.0 2023-04-16 17:57:06,077 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.0817, 4.9036, 5.1865, 4.9812, 5.4161, 4.8737, 5.4593, 5.4300], device='cuda:0'), covar=tensor([0.0305, 0.0527, 0.0547, 0.0496, 0.0483, 0.0826, 0.0398, 0.0391], device='cuda:0'), in_proj_covar=tensor([0.0209, 0.0234, 0.0220, 0.0170, 0.0318, 0.0267, 0.0199, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 17:57:10,438 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-16 17:57:20,448 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.148e+02 3.204e+02 3.731e+02 4.118e+02 8.079e+02, threshold=7.462e+02, percent-clipped=1.0 2023-04-16 17:57:35,641 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24826.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:57:38,067 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7422, 1.8262, 3.9923, 3.7683, 3.9029, 3.0394, 3.6439, 2.6675], device='cuda:0'), covar=tensor([0.2384, 0.1821, 0.0082, 0.0250, 0.0193, 0.0584, 0.0229, 0.0947], device='cuda:0'), in_proj_covar=tensor([0.0199, 0.0182, 0.0101, 0.0110, 0.0109, 0.0153, 0.0114, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-04-16 17:57:51,347 INFO [train.py:893] (0/4) Epoch 10, batch 550, loss[loss=0.229, simple_loss=0.2813, pruned_loss=0.0883, over 13270.00 frames. ], tot_loss[loss=0.2262, simple_loss=0.273, pruned_loss=0.08974, over 2483296.50 frames. ], batch size: 124, lr: 1.34e-02, grad_scale: 16.0 2023-04-16 17:57:56,366 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24848.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:58:35,980 INFO [train.py:893] (0/4) Epoch 10, batch 600, loss[loss=0.2414, simple_loss=0.2855, pruned_loss=0.0986, over 13464.00 frames. ], tot_loss[loss=0.2246, simple_loss=0.2713, pruned_loss=0.08893, over 2515718.94 frames. ], batch size: 103, lr: 1.34e-02, grad_scale: 16.0 2023-04-16 17:58:45,611 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24903.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:58:50,926 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.078e+02 3.175e+02 3.637e+02 4.549e+02 1.249e+03, threshold=7.275e+02, percent-clipped=2.0 2023-04-16 17:58:51,274 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24909.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:58:58,563 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7927, 2.6405, 2.2951, 2.8342, 2.2990, 2.9894, 2.7293, 2.4756], device='cuda:0'), covar=tensor([0.0071, 0.0135, 0.0134, 0.0145, 0.0160, 0.0095, 0.0209, 0.0134], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0072, 0.0080, 0.0070, 0.0087, 0.0065, 0.0071, 0.0067], device='cuda:0'), out_proj_covar=tensor([7.5315e-05, 8.3789e-05, 9.4193e-05, 8.1499e-05, 1.0267e-04, 7.3561e-05, 8.3318e-05, 7.5858e-05], device='cuda:0') 2023-04-16 17:59:04,931 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24925.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 17:59:21,850 INFO [train.py:893] (0/4) Epoch 10, batch 650, loss[loss=0.2098, simple_loss=0.2666, pruned_loss=0.07653, over 13527.00 frames. ], tot_loss[loss=0.2238, simple_loss=0.2705, pruned_loss=0.08855, over 2546646.49 frames. ], batch size: 76, lr: 1.33e-02, grad_scale: 16.0 2023-04-16 18:00:03,605 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6237, 2.2563, 2.7790, 3.8250, 3.6325, 3.8421, 3.0699, 2.2281], device='cuda:0'), covar=tensor([0.0207, 0.1020, 0.0735, 0.0067, 0.0179, 0.0035, 0.0596, 0.1064], device='cuda:0'), in_proj_covar=tensor([0.0105, 0.0148, 0.0151, 0.0083, 0.0103, 0.0073, 0.0153, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0002], device='cuda:0') 2023-04-16 18:00:05,337 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3158, 3.4982, 3.3338, 3.9097, 2.0655, 2.8717, 3.6088, 2.1923], device='cuda:0'), covar=tensor([0.0099, 0.0503, 0.0664, 0.0399, 0.1548, 0.0920, 0.0507, 0.1778], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0162, 0.0179, 0.0183, 0.0172, 0.0183, 0.0165, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:00:06,632 INFO [train.py:893] (0/4) Epoch 10, batch 700, loss[loss=0.2486, simple_loss=0.2911, pruned_loss=0.103, over 13526.00 frames. ], tot_loss[loss=0.2232, simple_loss=0.2705, pruned_loss=0.08797, over 2573242.40 frames. ], batch size: 83, lr: 1.33e-02, grad_scale: 16.0 2023-04-16 18:00:09,245 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24995.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:00:22,692 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.375e+02 3.121e+02 3.580e+02 4.230e+02 8.275e+02, threshold=7.159e+02, percent-clipped=1.0 2023-04-16 18:00:35,598 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6302, 4.8163, 3.1289, 4.7305, 4.6100, 2.8272, 4.0103, 3.2945], device='cuda:0'), covar=tensor([0.0201, 0.0178, 0.1060, 0.0170, 0.0173, 0.1074, 0.0347, 0.1226], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0134, 0.0173, 0.0120, 0.0115, 0.0153, 0.0148, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:00:53,434 INFO [train.py:893] (0/4) Epoch 10, batch 750, loss[loss=0.2489, simple_loss=0.2883, pruned_loss=0.1047, over 13485.00 frames. ], tot_loss[loss=0.2239, simple_loss=0.2705, pruned_loss=0.08869, over 2591130.38 frames. ], batch size: 100, lr: 1.33e-02, grad_scale: 16.0 2023-04-16 18:01:04,559 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25056.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 18:01:39,401 INFO [train.py:893] (0/4) Epoch 10, batch 800, loss[loss=0.2408, simple_loss=0.2861, pruned_loss=0.09772, over 13379.00 frames. ], tot_loss[loss=0.2251, simple_loss=0.2718, pruned_loss=0.08919, over 2605807.98 frames. ], batch size: 113, lr: 1.33e-02, grad_scale: 16.0 2023-04-16 18:01:53,421 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.389e+02 3.119e+02 3.527e+02 4.411e+02 6.666e+02, threshold=7.054e+02, percent-clipped=0.0 2023-04-16 18:02:04,141 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25121.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:02:14,205 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-16 18:02:24,252 INFO [train.py:893] (0/4) Epoch 10, batch 850, loss[loss=0.2284, simple_loss=0.2725, pruned_loss=0.0922, over 13492.00 frames. ], tot_loss[loss=0.2255, simple_loss=0.2724, pruned_loss=0.08927, over 2620957.97 frames. ], batch size: 70, lr: 1.33e-02, grad_scale: 16.0 2023-04-16 18:03:10,130 INFO [train.py:893] (0/4) Epoch 10, batch 900, loss[loss=0.2512, simple_loss=0.2884, pruned_loss=0.107, over 13347.00 frames. ], tot_loss[loss=0.2258, simple_loss=0.2722, pruned_loss=0.08969, over 2628891.55 frames. ], batch size: 73, lr: 1.33e-02, grad_scale: 16.0 2023-04-16 18:03:19,477 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25203.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:03:20,185 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25204.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:03:24,266 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.167e+02 3.308e+02 3.841e+02 4.546e+02 7.372e+02, threshold=7.682e+02, percent-clipped=2.0 2023-04-16 18:03:39,625 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 18:03:39,792 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25225.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:03:51,335 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0135, 4.4734, 4.2667, 4.1362, 4.2679, 4.0278, 4.5461, 4.4766], device='cuda:0'), covar=tensor([0.0245, 0.0235, 0.0204, 0.0356, 0.0247, 0.0321, 0.0255, 0.0281], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0164, 0.0126, 0.0152, 0.0118, 0.0160, 0.0108, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 18:03:55,701 INFO [train.py:893] (0/4) Epoch 10, batch 950, loss[loss=0.235, simple_loss=0.2801, pruned_loss=0.09489, over 13533.00 frames. ], tot_loss[loss=0.2263, simple_loss=0.2719, pruned_loss=0.09034, over 2638419.24 frames. ], batch size: 85, lr: 1.33e-02, grad_scale: 16.0 2023-04-16 18:04:03,684 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25251.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:04:22,521 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25273.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:04:40,950 INFO [train.py:893] (0/4) Epoch 10, batch 1000, loss[loss=0.2062, simple_loss=0.2545, pruned_loss=0.079, over 13568.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.2689, pruned_loss=0.08891, over 2640824.35 frames. ], batch size: 89, lr: 1.33e-02, grad_scale: 16.0 2023-04-16 18:04:46,904 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3992, 2.2959, 1.9833, 2.5305, 1.8732, 2.6248, 2.2881, 2.2740], device='cuda:0'), covar=tensor([0.0073, 0.0139, 0.0134, 0.0109, 0.0174, 0.0082, 0.0222, 0.0137], device='cuda:0'), in_proj_covar=tensor([0.0069, 0.0074, 0.0081, 0.0071, 0.0089, 0.0065, 0.0073, 0.0070], device='cuda:0'), out_proj_covar=tensor([7.7278e-05, 8.5490e-05, 9.5943e-05, 8.3020e-05, 1.0453e-04, 7.3377e-05, 8.4514e-05, 7.8661e-05], device='cuda:0') 2023-04-16 18:04:55,571 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.932e+02 3.162e+02 3.915e+02 4.659e+02 6.893e+02, threshold=7.829e+02, percent-clipped=0.0 2023-04-16 18:05:26,399 INFO [train.py:893] (0/4) Epoch 10, batch 1050, loss[loss=0.2048, simple_loss=0.2544, pruned_loss=0.07754, over 13529.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.2677, pruned_loss=0.08754, over 2647892.19 frames. ], batch size: 83, lr: 1.32e-02, grad_scale: 16.0 2023-04-16 18:05:33,865 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25351.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 18:05:55,762 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8019, 4.7538, 4.9417, 4.7240, 5.1960, 4.6511, 5.1846, 5.2082], device='cuda:0'), covar=tensor([0.0411, 0.0470, 0.0638, 0.0514, 0.0489, 0.0754, 0.0478, 0.0414], device='cuda:0'), in_proj_covar=tensor([0.0213, 0.0235, 0.0223, 0.0170, 0.0325, 0.0271, 0.0203, 0.0226], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:06:11,654 INFO [train.py:893] (0/4) Epoch 10, batch 1100, loss[loss=0.2154, simple_loss=0.2598, pruned_loss=0.08548, over 13538.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.268, pruned_loss=0.08691, over 2654731.92 frames. ], batch size: 72, lr: 1.32e-02, grad_scale: 16.0 2023-04-16 18:06:26,836 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.017e+02 3.163e+02 3.560e+02 4.411e+02 7.950e+02, threshold=7.120e+02, percent-clipped=1.0 2023-04-16 18:06:38,144 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25421.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:06:44,090 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=25428.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 18:06:58,304 INFO [train.py:893] (0/4) Epoch 10, batch 1150, loss[loss=0.2352, simple_loss=0.2836, pruned_loss=0.0934, over 13359.00 frames. ], tot_loss[loss=0.2205, simple_loss=0.268, pruned_loss=0.08645, over 2654540.31 frames. ], batch size: 84, lr: 1.32e-02, grad_scale: 16.0 2023-04-16 18:07:07,634 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2023-04-16 18:07:21,781 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25469.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:07:22,712 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6896, 1.6629, 3.5736, 3.3837, 3.4294, 2.7732, 3.3563, 2.4083], device='cuda:0'), covar=tensor([0.2058, 0.1646, 0.0107, 0.0202, 0.0170, 0.0686, 0.0169, 0.0957], device='cuda:0'), in_proj_covar=tensor([0.0206, 0.0188, 0.0104, 0.0115, 0.0115, 0.0158, 0.0117, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:07:39,610 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25489.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 18:07:42,562 INFO [train.py:893] (0/4) Epoch 10, batch 1200, loss[loss=0.2641, simple_loss=0.3045, pruned_loss=0.1118, over 13522.00 frames. ], tot_loss[loss=0.2208, simple_loss=0.2686, pruned_loss=0.08651, over 2653602.02 frames. ], batch size: 85, lr: 1.32e-02, grad_scale: 16.0 2023-04-16 18:07:53,652 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25504.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:07:57,438 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.109e+02 3.224e+02 3.678e+02 4.471e+02 9.246e+02, threshold=7.356e+02, percent-clipped=2.0 2023-04-16 18:08:07,748 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2023-04-16 18:08:09,681 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 18:08:22,493 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 18:08:28,385 INFO [train.py:893] (0/4) Epoch 10, batch 1250, loss[loss=0.2192, simple_loss=0.2701, pruned_loss=0.08421, over 13409.00 frames. ], tot_loss[loss=0.2221, simple_loss=0.2697, pruned_loss=0.08719, over 2654545.97 frames. ], batch size: 113, lr: 1.32e-02, grad_scale: 16.0 2023-04-16 18:08:32,671 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.14 vs. limit=5.0 2023-04-16 18:08:36,616 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25552.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:08:36,739 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2057, 2.6026, 2.2874, 4.1257, 4.6797, 3.6018, 4.5497, 4.3497], device='cuda:0'), covar=tensor([0.0092, 0.0753, 0.0882, 0.0082, 0.0058, 0.0386, 0.0086, 0.0065], device='cuda:0'), in_proj_covar=tensor([0.0059, 0.0077, 0.0084, 0.0064, 0.0048, 0.0069, 0.0043, 0.0057], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:09:13,503 INFO [train.py:893] (0/4) Epoch 10, batch 1300, loss[loss=0.2022, simple_loss=0.2515, pruned_loss=0.07644, over 13361.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.2716, pruned_loss=0.08832, over 2656029.69 frames. ], batch size: 67, lr: 1.32e-02, grad_scale: 16.0 2023-04-16 18:09:28,836 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.392e+02 3.558e+02 4.001e+02 4.769e+02 9.079e+02, threshold=8.002e+02, percent-clipped=3.0 2023-04-16 18:09:59,314 INFO [train.py:893] (0/4) Epoch 10, batch 1350, loss[loss=0.2097, simple_loss=0.2554, pruned_loss=0.08204, over 13367.00 frames. ], tot_loss[loss=0.2249, simple_loss=0.2722, pruned_loss=0.08885, over 2659263.95 frames. ], batch size: 62, lr: 1.32e-02, grad_scale: 16.0 2023-04-16 18:10:06,130 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25651.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 18:10:33,745 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5382, 3.4912, 2.8285, 2.8211, 2.7896, 1.8159, 3.3701, 1.7088], device='cuda:0'), covar=tensor([0.0717, 0.0410, 0.0444, 0.0522, 0.0747, 0.2024, 0.0955, 0.1674], device='cuda:0'), in_proj_covar=tensor([0.0122, 0.0108, 0.0112, 0.0096, 0.0132, 0.0164, 0.0121, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:10:45,425 INFO [train.py:893] (0/4) Epoch 10, batch 1400, loss[loss=0.2144, simple_loss=0.249, pruned_loss=0.08987, over 12721.00 frames. ], tot_loss[loss=0.2221, simple_loss=0.2695, pruned_loss=0.08737, over 2654731.47 frames. ], batch size: 52, lr: 1.32e-02, grad_scale: 16.0 2023-04-16 18:10:46,563 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2345, 4.1625, 4.3713, 4.3831, 4.5599, 4.2382, 4.5673, 4.5777], device='cuda:0'), covar=tensor([0.0410, 0.0565, 0.0604, 0.0427, 0.0593, 0.0871, 0.0507, 0.0463], device='cuda:0'), in_proj_covar=tensor([0.0213, 0.0235, 0.0223, 0.0169, 0.0326, 0.0270, 0.0204, 0.0226], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:10:50,745 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25699.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:10:58,842 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.4285, 4.8942, 4.9376, 4.8659, 4.6097, 4.7918, 5.3401, 4.8975], device='cuda:0'), covar=tensor([0.0774, 0.1104, 0.2067, 0.2386, 0.0891, 0.1410, 0.0813, 0.1007], device='cuda:0'), in_proj_covar=tensor([0.0265, 0.0296, 0.0377, 0.0382, 0.0216, 0.0291, 0.0343, 0.0261], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 18:10:59,446 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.903e+02 3.130e+02 3.640e+02 4.432e+02 7.364e+02, threshold=7.280e+02, percent-clipped=0.0 2023-04-16 18:11:24,438 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3386, 3.3620, 3.9225, 2.8142, 2.4745, 2.7163, 4.0889, 4.2205], device='cuda:0'), covar=tensor([0.1027, 0.1279, 0.0356, 0.1429, 0.1561, 0.1204, 0.0242, 0.0145], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0221, 0.0167, 0.0205, 0.0203, 0.0168, 0.0162, 0.0128], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:11:30,587 INFO [train.py:893] (0/4) Epoch 10, batch 1450, loss[loss=0.2278, simple_loss=0.2753, pruned_loss=0.09016, over 13506.00 frames. ], tot_loss[loss=0.2222, simple_loss=0.2695, pruned_loss=0.08748, over 2654439.75 frames. ], batch size: 91, lr: 1.31e-02, grad_scale: 16.0 2023-04-16 18:12:09,690 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25784.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 18:12:16,780 INFO [train.py:893] (0/4) Epoch 10, batch 1500, loss[loss=0.2135, simple_loss=0.2704, pruned_loss=0.07828, over 13459.00 frames. ], tot_loss[loss=0.2217, simple_loss=0.2694, pruned_loss=0.08701, over 2658913.57 frames. ], batch size: 100, lr: 1.31e-02, grad_scale: 16.0 2023-04-16 18:12:31,672 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.085e+02 3.068e+02 3.732e+02 4.562e+02 1.216e+03, threshold=7.464e+02, percent-clipped=1.0 2023-04-16 18:13:02,775 INFO [train.py:893] (0/4) Epoch 10, batch 1550, loss[loss=0.2169, simple_loss=0.2654, pruned_loss=0.0842, over 13454.00 frames. ], tot_loss[loss=0.2217, simple_loss=0.2693, pruned_loss=0.08707, over 2648650.00 frames. ], batch size: 106, lr: 1.31e-02, grad_scale: 16.0 2023-04-16 18:13:16,728 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=25858.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 18:13:20,118 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-16 18:13:23,870 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9009, 3.6003, 3.0699, 3.4361, 3.1000, 2.0645, 3.8009, 1.9922], device='cuda:0'), covar=tensor([0.0673, 0.0547, 0.0474, 0.0299, 0.0617, 0.1858, 0.0664, 0.1442], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0110, 0.0112, 0.0095, 0.0131, 0.0164, 0.0120, 0.0127], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:13:48,265 INFO [train.py:893] (0/4) Epoch 10, batch 1600, loss[loss=0.2003, simple_loss=0.2596, pruned_loss=0.07049, over 13437.00 frames. ], tot_loss[loss=0.2215, simple_loss=0.2696, pruned_loss=0.0867, over 2646061.98 frames. ], batch size: 106, lr: 1.31e-02, grad_scale: 32.0 2023-04-16 18:14:04,226 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.197e+02 3.045e+02 3.726e+02 4.313e+02 1.035e+03, threshold=7.453e+02, percent-clipped=2.0 2023-04-16 18:14:13,113 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25919.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 18:14:21,688 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-16 18:14:35,513 INFO [train.py:893] (0/4) Epoch 10, batch 1650, loss[loss=0.2203, simple_loss=0.2608, pruned_loss=0.08987, over 13411.00 frames. ], tot_loss[loss=0.222, simple_loss=0.2707, pruned_loss=0.08664, over 2647240.19 frames. ], batch size: 65, lr: 1.31e-02, grad_scale: 32.0 2023-04-16 18:15:20,316 INFO [train.py:893] (0/4) Epoch 10, batch 1700, loss[loss=0.2276, simple_loss=0.2784, pruned_loss=0.08837, over 13494.00 frames. ], tot_loss[loss=0.2215, simple_loss=0.2707, pruned_loss=0.08611, over 2652366.20 frames. ], batch size: 81, lr: 1.31e-02, grad_scale: 32.0 2023-04-16 18:15:28,417 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-26000.pt 2023-04-16 18:15:39,374 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.206e+02 3.199e+02 3.766e+02 4.421e+02 9.906e+02, threshold=7.531e+02, percent-clipped=2.0 2023-04-16 18:15:47,310 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7406, 3.7730, 3.2712, 2.7105, 2.6636, 2.2061, 3.9138, 2.2302], device='cuda:0'), covar=tensor([0.1151, 0.0276, 0.0592, 0.1268, 0.0635, 0.2701, 0.0183, 0.3325], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0256, 0.0258, 0.0274, 0.0219, 0.0277, 0.0180, 0.0326], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 18:15:52,393 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-16 18:15:56,207 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8250, 4.7720, 4.9798, 4.8524, 5.1979, 4.7830, 5.2711, 5.2230], device='cuda:0'), covar=tensor([0.0389, 0.0486, 0.0640, 0.0461, 0.0599, 0.0804, 0.0446, 0.0466], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0242, 0.0229, 0.0173, 0.0336, 0.0277, 0.0208, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:16:10,317 INFO [train.py:893] (0/4) Epoch 10, batch 1750, loss[loss=0.1883, simple_loss=0.2398, pruned_loss=0.06837, over 13397.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.2695, pruned_loss=0.08557, over 2654612.73 frames. ], batch size: 65, lr: 1.31e-02, grad_scale: 32.0 2023-04-16 18:16:12,645 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2023-04-16 18:16:44,659 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9962, 4.9674, 5.1609, 4.9395, 5.4042, 4.8695, 5.4011, 5.3884], device='cuda:0'), covar=tensor([0.0370, 0.0473, 0.0607, 0.0461, 0.0515, 0.0791, 0.0422, 0.0426], device='cuda:0'), in_proj_covar=tensor([0.0216, 0.0239, 0.0226, 0.0171, 0.0331, 0.0272, 0.0204, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:16:47,958 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26084.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 18:16:55,981 INFO [train.py:893] (0/4) Epoch 10, batch 1800, loss[loss=0.1942, simple_loss=0.2556, pruned_loss=0.06638, over 13533.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.2684, pruned_loss=0.08447, over 2657820.72 frames. ], batch size: 76, lr: 1.31e-02, grad_scale: 32.0 2023-04-16 18:17:04,903 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26102.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:17:06,979 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-16 18:17:10,493 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.103e+02 2.999e+02 3.760e+02 4.430e+02 9.683e+02, threshold=7.520e+02, percent-clipped=2.0 2023-04-16 18:17:11,533 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9019, 3.6357, 3.8722, 2.2517, 4.2287, 3.9761, 3.9931, 4.0529], device='cuda:0'), covar=tensor([0.0192, 0.0116, 0.0122, 0.1199, 0.0114, 0.0168, 0.0113, 0.0111], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0040, 0.0061, 0.0086, 0.0077, 0.0076, 0.0062, 0.0054], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:17:14,097 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-16 18:17:31,660 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26132.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 18:17:35,069 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7862, 2.7059, 2.3848, 2.8975, 2.3244, 3.0063, 2.8658, 2.5301], device='cuda:0'), covar=tensor([0.0059, 0.0111, 0.0127, 0.0122, 0.0148, 0.0081, 0.0183, 0.0161], device='cuda:0'), in_proj_covar=tensor([0.0069, 0.0073, 0.0079, 0.0071, 0.0087, 0.0065, 0.0071, 0.0070], device='cuda:0'), out_proj_covar=tensor([7.6692e-05, 8.3851e-05, 9.2373e-05, 8.2975e-05, 1.0145e-04, 7.3114e-05, 8.1949e-05, 7.8683e-05], device='cuda:0') 2023-04-16 18:17:41,231 INFO [train.py:893] (0/4) Epoch 10, batch 1850, loss[loss=0.2013, simple_loss=0.2562, pruned_loss=0.0732, over 13365.00 frames. ], tot_loss[loss=0.218, simple_loss=0.2675, pruned_loss=0.08427, over 2655951.71 frames. ], batch size: 73, lr: 1.30e-02, grad_scale: 32.0 2023-04-16 18:17:43,767 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 18:17:59,302 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26163.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:18:03,372 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26168.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:18:07,587 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1101, 4.1192, 3.5050, 2.9395, 2.9019, 2.4042, 4.3746, 2.4780], device='cuda:0'), covar=tensor([0.1384, 0.0340, 0.0695, 0.1511, 0.0714, 0.2953, 0.0172, 0.3339], device='cuda:0'), in_proj_covar=tensor([0.0286, 0.0258, 0.0262, 0.0279, 0.0220, 0.0280, 0.0183, 0.0330], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 18:18:25,589 INFO [train.py:893] (0/4) Epoch 10, batch 1900, loss[loss=0.2247, simple_loss=0.2743, pruned_loss=0.08759, over 13421.00 frames. ], tot_loss[loss=0.2167, simple_loss=0.2659, pruned_loss=0.08373, over 2655148.50 frames. ], batch size: 95, lr: 1.30e-02, grad_scale: 32.0 2023-04-16 18:18:30,216 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-16 18:18:39,848 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.726e+02 3.263e+02 3.745e+02 4.273e+02 6.858e+02, threshold=7.490e+02, percent-clipped=0.0 2023-04-16 18:18:44,163 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26214.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 18:18:57,949 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26229.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:19:09,751 INFO [train.py:893] (0/4) Epoch 10, batch 1950, loss[loss=0.2177, simple_loss=0.2608, pruned_loss=0.08733, over 13255.00 frames. ], tot_loss[loss=0.2172, simple_loss=0.2661, pruned_loss=0.08412, over 2656874.26 frames. ], batch size: 124, lr: 1.30e-02, grad_scale: 32.0 2023-04-16 18:19:12,674 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.18 vs. limit=5.0 2023-04-16 18:19:15,721 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.4142, 2.0736, 1.9930, 2.5157, 1.7826, 2.4926, 2.4634, 2.1171], device='cuda:0'), covar=tensor([0.0065, 0.0140, 0.0125, 0.0086, 0.0153, 0.0085, 0.0135, 0.0131], device='cuda:0'), in_proj_covar=tensor([0.0069, 0.0073, 0.0079, 0.0071, 0.0087, 0.0064, 0.0070, 0.0070], device='cuda:0'), out_proj_covar=tensor([7.6570e-05, 8.3595e-05, 9.2284e-05, 8.2162e-05, 1.0111e-04, 7.2271e-05, 8.0798e-05, 7.8340e-05], device='cuda:0') 2023-04-16 18:19:16,534 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26250.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:19:23,798 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9089, 4.1961, 2.5296, 4.2407, 4.1105, 2.6245, 3.6208, 2.6505], device='cuda:0'), covar=tensor([0.0353, 0.0499, 0.1412, 0.0295, 0.0277, 0.1265, 0.0578, 0.1515], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0137, 0.0169, 0.0124, 0.0114, 0.0151, 0.0145, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:19:44,165 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2727, 3.3882, 3.9818, 2.8426, 2.5278, 2.7245, 4.1251, 4.3118], device='cuda:0'), covar=tensor([0.0990, 0.1216, 0.0291, 0.1244, 0.1382, 0.1174, 0.0196, 0.0126], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0221, 0.0168, 0.0207, 0.0203, 0.0169, 0.0164, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:19:55,263 INFO [train.py:893] (0/4) Epoch 10, batch 2000, loss[loss=0.2144, simple_loss=0.2607, pruned_loss=0.08406, over 13351.00 frames. ], tot_loss[loss=0.2196, simple_loss=0.2682, pruned_loss=0.08551, over 2658325.00 frames. ], batch size: 73, lr: 1.30e-02, grad_scale: 32.0 2023-04-16 18:20:02,519 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 18:20:09,140 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.941e+02 3.261e+02 3.830e+02 4.478e+02 8.145e+02, threshold=7.659e+02, percent-clipped=2.0 2023-04-16 18:20:11,800 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26311.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:20:23,361 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8908, 1.7652, 3.5440, 3.4136, 3.2424, 2.6373, 3.2103, 2.3885], device='cuda:0'), covar=tensor([0.2010, 0.1639, 0.0078, 0.0193, 0.0305, 0.0764, 0.0271, 0.1039], device='cuda:0'), in_proj_covar=tensor([0.0201, 0.0184, 0.0101, 0.0111, 0.0115, 0.0156, 0.0118, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:20:28,434 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6092, 3.8953, 2.5053, 3.8128, 3.7026, 2.1787, 3.2070, 2.3185], device='cuda:0'), covar=tensor([0.0299, 0.0280, 0.1211, 0.0279, 0.0290, 0.1359, 0.0600, 0.1593], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0135, 0.0165, 0.0122, 0.0112, 0.0149, 0.0142, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:20:40,242 INFO [train.py:893] (0/4) Epoch 10, batch 2050, loss[loss=0.256, simple_loss=0.2948, pruned_loss=0.1086, over 11982.00 frames. ], tot_loss[loss=0.2216, simple_loss=0.2699, pruned_loss=0.08664, over 2654590.33 frames. ], batch size: 157, lr: 1.30e-02, grad_scale: 32.0 2023-04-16 18:21:01,817 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4061, 3.4420, 4.1235, 3.0745, 2.8349, 2.8116, 4.2905, 4.4350], device='cuda:0'), covar=tensor([0.0859, 0.1190, 0.0297, 0.1138, 0.1225, 0.1209, 0.0173, 0.0120], device='cuda:0'), in_proj_covar=tensor([0.0218, 0.0219, 0.0167, 0.0206, 0.0202, 0.0168, 0.0163, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:21:01,979 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-16 18:21:24,491 INFO [train.py:893] (0/4) Epoch 10, batch 2100, loss[loss=0.2121, simple_loss=0.2652, pruned_loss=0.0795, over 13541.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.2695, pruned_loss=0.08615, over 2655424.28 frames. ], batch size: 83, lr: 1.30e-02, grad_scale: 32.0 2023-04-16 18:21:40,090 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.897e+02 2.995e+02 3.490e+02 4.198e+02 1.045e+03, threshold=6.980e+02, percent-clipped=1.0 2023-04-16 18:21:44,439 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26414.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:22:07,259 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.7913, 4.6770, 4.8406, 4.6959, 5.1288, 4.5223, 5.1397, 5.1215], device='cuda:0'), covar=tensor([0.0371, 0.0561, 0.0656, 0.0498, 0.0552, 0.0893, 0.0452, 0.0460], device='cuda:0'), in_proj_covar=tensor([0.0218, 0.0243, 0.0230, 0.0174, 0.0332, 0.0278, 0.0207, 0.0226], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:22:11,098 INFO [train.py:893] (0/4) Epoch 10, batch 2150, loss[loss=0.1875, simple_loss=0.2329, pruned_loss=0.07109, over 13145.00 frames. ], tot_loss[loss=0.2194, simple_loss=0.2688, pruned_loss=0.08506, over 2657411.51 frames. ], batch size: 58, lr: 1.30e-02, grad_scale: 32.0 2023-04-16 18:22:18,745 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2665, 2.1337, 2.3614, 3.6738, 3.3419, 3.6452, 2.8998, 2.1940], device='cuda:0'), covar=tensor([0.0322, 0.0991, 0.0851, 0.0056, 0.0298, 0.0051, 0.0658, 0.1039], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0151, 0.0159, 0.0087, 0.0108, 0.0080, 0.0161, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0002], device='cuda:0') 2023-04-16 18:22:24,404 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26458.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:22:40,907 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26475.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:22:55,980 INFO [train.py:893] (0/4) Epoch 10, batch 2200, loss[loss=0.2213, simple_loss=0.2748, pruned_loss=0.08387, over 13429.00 frames. ], tot_loss[loss=0.2191, simple_loss=0.2683, pruned_loss=0.085, over 2652926.55 frames. ], batch size: 95, lr: 1.30e-02, grad_scale: 32.0 2023-04-16 18:23:10,976 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.188e+02 3.062e+02 3.589e+02 4.255e+02 9.664e+02, threshold=7.177e+02, percent-clipped=3.0 2023-04-16 18:23:15,484 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26514.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 18:23:24,136 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26524.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:23:41,894 INFO [train.py:893] (0/4) Epoch 10, batch 2250, loss[loss=0.2179, simple_loss=0.2681, pruned_loss=0.0838, over 13492.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.2665, pruned_loss=0.08412, over 2653642.64 frames. ], batch size: 93, lr: 1.30e-02, grad_scale: 32.0 2023-04-16 18:23:49,096 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-16 18:23:58,587 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26562.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 18:24:03,173 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5373, 2.3972, 2.1896, 2.5833, 1.7305, 2.5663, 2.4565, 2.2934], device='cuda:0'), covar=tensor([0.0077, 0.0137, 0.0118, 0.0097, 0.0179, 0.0101, 0.0160, 0.0130], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0077, 0.0081, 0.0073, 0.0089, 0.0066, 0.0073, 0.0072], device='cuda:0'), out_proj_covar=tensor([7.9243e-05, 8.7951e-05, 9.5384e-05, 8.5238e-05, 1.0435e-04, 7.4472e-05, 8.5095e-05, 8.0673e-05], device='cuda:0') 2023-04-16 18:24:26,373 INFO [train.py:893] (0/4) Epoch 10, batch 2300, loss[loss=0.2226, simple_loss=0.2737, pruned_loss=0.08576, over 13561.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2658, pruned_loss=0.08371, over 2657220.36 frames. ], batch size: 89, lr: 1.29e-02, grad_scale: 32.0 2023-04-16 18:24:38,814 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26606.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:24:41,108 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.014e+02 2.982e+02 3.568e+02 4.617e+02 1.178e+03, threshold=7.136e+02, percent-clipped=4.0 2023-04-16 18:24:45,616 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9528, 2.1779, 4.2365, 3.8147, 3.9501, 3.1793, 3.8124, 2.9628], device='cuda:0'), covar=tensor([0.2056, 0.1685, 0.0045, 0.0210, 0.0143, 0.0684, 0.0220, 0.0852], device='cuda:0'), in_proj_covar=tensor([0.0209, 0.0194, 0.0104, 0.0115, 0.0119, 0.0166, 0.0124, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:25:04,059 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-16 18:25:05,797 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-16 18:25:11,874 INFO [train.py:893] (0/4) Epoch 10, batch 2350, loss[loss=0.2009, simple_loss=0.2533, pruned_loss=0.07426, over 13512.00 frames. ], tot_loss[loss=0.2163, simple_loss=0.2657, pruned_loss=0.08349, over 2662563.34 frames. ], batch size: 76, lr: 1.29e-02, grad_scale: 16.0 2023-04-16 18:25:36,291 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 18:25:57,728 INFO [train.py:893] (0/4) Epoch 10, batch 2400, loss[loss=0.2518, simple_loss=0.2972, pruned_loss=0.1032, over 13049.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2659, pruned_loss=0.08363, over 2663930.71 frames. ], batch size: 142, lr: 1.29e-02, grad_scale: 16.0 2023-04-16 18:26:12,464 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.295e+02 3.122e+02 3.581e+02 4.550e+02 8.173e+02, threshold=7.162e+02, percent-clipped=1.0 2023-04-16 18:26:42,373 INFO [train.py:893] (0/4) Epoch 10, batch 2450, loss[loss=0.2454, simple_loss=0.2856, pruned_loss=0.1027, over 13071.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.2655, pruned_loss=0.08363, over 2657447.22 frames. ], batch size: 142, lr: 1.29e-02, grad_scale: 16.0 2023-04-16 18:26:47,475 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.5583, 1.7600, 3.6294, 3.3843, 3.3591, 2.7036, 3.2382, 2.4731], device='cuda:0'), covar=tensor([0.2429, 0.1726, 0.0124, 0.0199, 0.0250, 0.0795, 0.0254, 0.0984], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0188, 0.0102, 0.0113, 0.0118, 0.0161, 0.0123, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:26:54,748 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0437, 4.0628, 3.4828, 2.9048, 3.0120, 2.4052, 4.2279, 2.4588], device='cuda:0'), covar=tensor([0.1149, 0.0259, 0.0537, 0.1284, 0.0560, 0.2574, 0.0173, 0.3057], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0260, 0.0261, 0.0280, 0.0221, 0.0279, 0.0182, 0.0329], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 18:26:57,723 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26758.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:27:07,620 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26770.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:27:20,597 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6484, 2.6010, 2.2128, 1.3976, 1.4654, 1.9498, 1.9189, 2.7704], device='cuda:0'), covar=tensor([0.1028, 0.0332, 0.0757, 0.2052, 0.0451, 0.0486, 0.0866, 0.0258], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0103, 0.0094, 0.0173, 0.0090, 0.0121, 0.0125, 0.0099], device='cuda:0'), out_proj_covar=tensor([1.0111e-04, 7.8791e-05, 7.6544e-05, 1.3632e-04, 7.2100e-05, 9.2533e-05, 9.7920e-05, 7.3457e-05], device='cuda:0') 2023-04-16 18:27:28,695 INFO [train.py:893] (0/4) Epoch 10, batch 2500, loss[loss=0.2153, simple_loss=0.2607, pruned_loss=0.08496, over 13523.00 frames. ], tot_loss[loss=0.2171, simple_loss=0.2661, pruned_loss=0.08408, over 2658259.65 frames. ], batch size: 85, lr: 1.29e-02, grad_scale: 16.0 2023-04-16 18:27:40,104 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26806.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:27:43,918 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.938e+02 3.049e+02 3.538e+02 4.460e+02 1.180e+03, threshold=7.075e+02, percent-clipped=3.0 2023-04-16 18:27:57,973 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26824.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:28:14,260 INFO [train.py:893] (0/4) Epoch 10, batch 2550, loss[loss=0.21, simple_loss=0.2674, pruned_loss=0.07629, over 13465.00 frames. ], tot_loss[loss=0.2169, simple_loss=0.2661, pruned_loss=0.08384, over 2659212.35 frames. ], batch size: 79, lr: 1.29e-02, grad_scale: 16.0 2023-04-16 18:28:38,984 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 18:28:41,398 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26872.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:29:00,094 INFO [train.py:893] (0/4) Epoch 10, batch 2600, loss[loss=0.1958, simple_loss=0.2519, pruned_loss=0.06988, over 13419.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.2662, pruned_loss=0.08419, over 2656832.92 frames. ], batch size: 95, lr: 1.29e-02, grad_scale: 16.0 2023-04-16 18:29:03,737 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4731, 3.7237, 3.0730, 4.2809, 1.7486, 2.5262, 3.6983, 1.9653], device='cuda:0'), covar=tensor([0.0082, 0.0497, 0.0873, 0.0440, 0.1714, 0.1364, 0.0724, 0.2098], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0163, 0.0185, 0.0186, 0.0171, 0.0182, 0.0166, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:29:10,024 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26904.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 18:29:11,605 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26906.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:29:14,775 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.365e+02 3.149e+02 3.762e+02 4.546e+02 1.526e+03, threshold=7.525e+02, percent-clipped=2.0 2023-04-16 18:29:41,132 INFO [train.py:893] (0/4) Epoch 10, batch 2650, loss[loss=0.2528, simple_loss=0.2911, pruned_loss=0.1072, over 13281.00 frames. ], tot_loss[loss=0.2193, simple_loss=0.2676, pruned_loss=0.08554, over 2651635.50 frames. ], batch size: 124, lr: 1.29e-02, grad_scale: 16.0 2023-04-16 18:29:46,445 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26950.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:29:49,389 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26954.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:29:57,772 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26965.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 18:30:18,340 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-10.pt 2023-04-16 18:30:42,582 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 18:30:46,268 INFO [train.py:893] (0/4) Epoch 11, batch 0, loss[loss=0.2301, simple_loss=0.2749, pruned_loss=0.09262, over 13493.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.2749, pruned_loss=0.09262, over 13493.00 frames. ], batch size: 81, lr: 1.23e-02, grad_scale: 16.0 2023-04-16 18:30:46,269 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 18:30:56,343 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4103, 4.7568, 3.2320, 4.6207, 4.4815, 2.9983, 4.0768, 3.1119], device='cuda:0'), covar=tensor([0.0209, 0.0152, 0.0896, 0.0258, 0.0183, 0.1000, 0.0415, 0.1471], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0138, 0.0171, 0.0127, 0.0116, 0.0152, 0.0145, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:31:08,657 INFO [train.py:927] (0/4) Epoch 11, validation: loss=0.1601, simple_loss=0.2155, pruned_loss=0.05235, over 2446609.00 frames. 2023-04-16 18:31:08,658 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 18:31:10,691 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6482, 3.7306, 4.2922, 3.1510, 2.7810, 3.0524, 4.4745, 4.5887], device='cuda:0'), covar=tensor([0.0993, 0.1280, 0.0307, 0.1391, 0.1482, 0.1164, 0.0217, 0.0119], device='cuda:0'), in_proj_covar=tensor([0.0217, 0.0223, 0.0167, 0.0205, 0.0202, 0.0169, 0.0166, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:31:18,370 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4057, 4.1873, 4.3366, 2.7420, 4.7179, 4.4797, 4.4315, 4.6435], device='cuda:0'), covar=tensor([0.0186, 0.0084, 0.0125, 0.1078, 0.0128, 0.0196, 0.0110, 0.0093], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0038, 0.0060, 0.0084, 0.0075, 0.0073, 0.0060, 0.0053], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:31:24,696 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.130e+02 3.114e+02 4.055e+02 4.999e+02 1.494e+03, threshold=8.109e+02, percent-clipped=6.0 2023-04-16 18:31:26,620 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27011.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:31:53,718 INFO [train.py:893] (0/4) Epoch 11, batch 50, loss[loss=0.2266, simple_loss=0.2665, pruned_loss=0.09333, over 13432.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.2616, pruned_loss=0.08439, over 604147.73 frames. ], batch size: 65, lr: 1.23e-02, grad_scale: 16.0 2023-04-16 18:32:19,820 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 18:32:19,820 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 18:32:19,821 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 18:32:19,830 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 18:32:19,847 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 18:32:19,867 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 18:32:19,877 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 18:32:20,079 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27070.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:32:39,611 INFO [train.py:893] (0/4) Epoch 11, batch 100, loss[loss=0.2142, simple_loss=0.2542, pruned_loss=0.08712, over 13420.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.2607, pruned_loss=0.08412, over 1057848.55 frames. ], batch size: 65, lr: 1.22e-02, grad_scale: 16.0 2023-04-16 18:32:47,511 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7511, 3.7417, 3.2301, 2.5966, 2.6065, 2.1678, 3.8848, 2.0840], device='cuda:0'), covar=tensor([0.1243, 0.0307, 0.0624, 0.1486, 0.0705, 0.2842, 0.0223, 0.3630], device='cuda:0'), in_proj_covar=tensor([0.0284, 0.0261, 0.0261, 0.0278, 0.0221, 0.0280, 0.0182, 0.0329], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 18:32:52,649 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-16 18:32:56,091 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.133e+02 2.776e+02 3.223e+02 3.818e+02 7.236e+02, threshold=6.446e+02, percent-clipped=0.0 2023-04-16 18:33:03,055 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27118.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:33:25,356 INFO [train.py:893] (0/4) Epoch 11, batch 150, loss[loss=0.2171, simple_loss=0.2681, pruned_loss=0.08301, over 13386.00 frames. ], tot_loss[loss=0.2183, simple_loss=0.2642, pruned_loss=0.0862, over 1411161.08 frames. ], batch size: 113, lr: 1.22e-02, grad_scale: 16.0 2023-04-16 18:33:30,773 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=27148.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:34:11,539 INFO [train.py:893] (0/4) Epoch 11, batch 200, loss[loss=0.2444, simple_loss=0.2915, pruned_loss=0.09859, over 13528.00 frames. ], tot_loss[loss=0.221, simple_loss=0.2675, pruned_loss=0.08725, over 1686901.63 frames. ], batch size: 83, lr: 1.22e-02, grad_scale: 16.0 2023-04-16 18:34:27,104 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27209.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:34:27,622 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.283e+02 3.174e+02 3.920e+02 4.819e+02 9.986e+02, threshold=7.840e+02, percent-clipped=7.0 2023-04-16 18:34:40,942 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.56 vs. limit=5.0 2023-04-16 18:34:57,072 INFO [train.py:893] (0/4) Epoch 11, batch 250, loss[loss=0.1783, simple_loss=0.2289, pruned_loss=0.06388, over 13395.00 frames. ], tot_loss[loss=0.2196, simple_loss=0.2664, pruned_loss=0.08637, over 1894103.04 frames. ], batch size: 62, lr: 1.22e-02, grad_scale: 16.0 2023-04-16 18:35:13,568 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27260.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 18:35:33,697 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.5676, 5.1171, 4.9368, 4.9804, 4.7735, 4.9660, 5.5726, 5.0795], device='cuda:0'), covar=tensor([0.0799, 0.1100, 0.2362, 0.2928, 0.0882, 0.1681, 0.0925, 0.0984], device='cuda:0'), in_proj_covar=tensor([0.0274, 0.0314, 0.0386, 0.0400, 0.0226, 0.0297, 0.0354, 0.0275], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 18:35:41,630 INFO [train.py:893] (0/4) Epoch 11, batch 300, loss[loss=0.2153, simple_loss=0.2614, pruned_loss=0.08465, over 13550.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.2673, pruned_loss=0.08624, over 2068403.31 frames. ], batch size: 72, lr: 1.22e-02, grad_scale: 16.0 2023-04-16 18:35:53,842 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27306.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:35:56,885 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.999e+02 3.048e+02 3.431e+02 4.475e+02 6.700e+02, threshold=6.863e+02, percent-clipped=0.0 2023-04-16 18:36:19,600 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.7348, 4.0528, 3.8395, 4.4911, 2.4615, 3.2928, 4.2818, 2.3977], device='cuda:0'), covar=tensor([0.0122, 0.0405, 0.0688, 0.0596, 0.1419, 0.0795, 0.0371, 0.1819], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0168, 0.0189, 0.0194, 0.0177, 0.0185, 0.0170, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:36:26,317 INFO [train.py:893] (0/4) Epoch 11, batch 350, loss[loss=0.2606, simple_loss=0.306, pruned_loss=0.1076, over 13422.00 frames. ], tot_loss[loss=0.2212, simple_loss=0.2685, pruned_loss=0.08692, over 2204212.99 frames. ], batch size: 95, lr: 1.22e-02, grad_scale: 16.0 2023-04-16 18:37:02,691 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-16 18:37:11,383 INFO [train.py:893] (0/4) Epoch 11, batch 400, loss[loss=0.2339, simple_loss=0.2844, pruned_loss=0.09167, over 13387.00 frames. ], tot_loss[loss=0.2222, simple_loss=0.27, pruned_loss=0.08717, over 2304486.58 frames. ], batch size: 113, lr: 1.22e-02, grad_scale: 16.0 2023-04-16 18:37:29,239 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.964e+02 3.046e+02 3.470e+02 4.359e+02 1.054e+03, threshold=6.940e+02, percent-clipped=5.0 2023-04-16 18:37:58,216 INFO [train.py:893] (0/4) Epoch 11, batch 450, loss[loss=0.2335, simple_loss=0.2794, pruned_loss=0.09383, over 13492.00 frames. ], tot_loss[loss=0.224, simple_loss=0.2719, pruned_loss=0.08805, over 2382049.45 frames. ], batch size: 93, lr: 1.22e-02, grad_scale: 16.0 2023-04-16 18:38:11,211 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.53 vs. limit=5.0 2023-04-16 18:38:23,445 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 18:38:42,132 INFO [train.py:893] (0/4) Epoch 11, batch 500, loss[loss=0.2229, simple_loss=0.2707, pruned_loss=0.08761, over 13577.00 frames. ], tot_loss[loss=0.2227, simple_loss=0.2708, pruned_loss=0.08734, over 2446610.49 frames. ], batch size: 89, lr: 1.22e-02, grad_scale: 16.0 2023-04-16 18:38:54,945 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27504.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:38:59,643 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.020e+02 3.104e+02 3.811e+02 4.555e+02 9.896e+02, threshold=7.623e+02, percent-clipped=2.0 2023-04-16 18:39:04,965 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-16 18:39:28,705 INFO [train.py:893] (0/4) Epoch 11, batch 550, loss[loss=0.2391, simple_loss=0.2791, pruned_loss=0.09957, over 13472.00 frames. ], tot_loss[loss=0.2219, simple_loss=0.2703, pruned_loss=0.08675, over 2492006.11 frames. ], batch size: 100, lr: 1.21e-02, grad_scale: 16.0 2023-04-16 18:39:44,485 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27560.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 18:40:13,418 INFO [train.py:893] (0/4) Epoch 11, batch 600, loss[loss=0.2432, simple_loss=0.2881, pruned_loss=0.09916, over 13556.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.2675, pruned_loss=0.08542, over 2531039.82 frames. ], batch size: 78, lr: 1.21e-02, grad_scale: 16.0 2023-04-16 18:40:27,680 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27606.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:40:29,316 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27608.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 18:40:30,678 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.019e+02 2.977e+02 3.539e+02 4.534e+02 8.649e+02, threshold=7.078e+02, percent-clipped=2.0 2023-04-16 18:40:58,232 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5522, 2.5880, 2.7151, 4.0066, 3.7346, 4.0165, 3.1793, 2.3784], device='cuda:0'), covar=tensor([0.0305, 0.0835, 0.0807, 0.0042, 0.0178, 0.0038, 0.0597, 0.1019], device='cuda:0'), in_proj_covar=tensor([0.0105, 0.0148, 0.0155, 0.0084, 0.0102, 0.0076, 0.0156, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0002], device='cuda:0') 2023-04-16 18:40:59,638 INFO [train.py:893] (0/4) Epoch 11, batch 650, loss[loss=0.1986, simple_loss=0.2362, pruned_loss=0.08054, over 12783.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.2672, pruned_loss=0.08521, over 2560512.41 frames. ], batch size: 52, lr: 1.21e-02, grad_scale: 16.0 2023-04-16 18:41:10,240 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27654.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:41:42,722 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0023, 4.0743, 3.4195, 2.8412, 3.1270, 2.4723, 4.2358, 2.4264], device='cuda:0'), covar=tensor([0.1236, 0.0272, 0.0607, 0.1381, 0.0536, 0.2603, 0.0157, 0.3168], device='cuda:0'), in_proj_covar=tensor([0.0288, 0.0264, 0.0264, 0.0279, 0.0222, 0.0282, 0.0184, 0.0329], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 18:41:44,846 INFO [train.py:893] (0/4) Epoch 11, batch 700, loss[loss=0.2007, simple_loss=0.2554, pruned_loss=0.07302, over 13526.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.2662, pruned_loss=0.08414, over 2585263.75 frames. ], batch size: 76, lr: 1.21e-02, grad_scale: 16.0 2023-04-16 18:42:00,292 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.022e+02 3.180e+02 3.711e+02 4.284e+02 7.518e+02, threshold=7.422e+02, percent-clipped=1.0 2023-04-16 18:42:03,819 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=27714.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 18:42:11,563 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9799, 4.4829, 4.3857, 4.4660, 4.2299, 4.3421, 4.9449, 4.5337], device='cuda:0'), covar=tensor([0.0809, 0.1178, 0.2285, 0.2674, 0.0975, 0.1573, 0.0861, 0.0934], device='cuda:0'), in_proj_covar=tensor([0.0274, 0.0312, 0.0388, 0.0403, 0.0229, 0.0302, 0.0353, 0.0273], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 18:42:29,604 INFO [train.py:893] (0/4) Epoch 11, batch 750, loss[loss=0.2019, simple_loss=0.2531, pruned_loss=0.07532, over 13382.00 frames. ], tot_loss[loss=0.2172, simple_loss=0.2659, pruned_loss=0.08427, over 2600915.75 frames. ], batch size: 77, lr: 1.21e-02, grad_scale: 16.0 2023-04-16 18:42:49,909 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7250, 2.7380, 3.0083, 4.2808, 3.9616, 4.3073, 3.4310, 2.7879], device='cuda:0'), covar=tensor([0.0295, 0.0885, 0.0704, 0.0044, 0.0172, 0.0035, 0.0585, 0.0871], device='cuda:0'), in_proj_covar=tensor([0.0106, 0.0148, 0.0155, 0.0084, 0.0102, 0.0077, 0.0155, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0002], device='cuda:0') 2023-04-16 18:42:59,883 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27775.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 18:43:15,445 INFO [train.py:893] (0/4) Epoch 11, batch 800, loss[loss=0.2255, simple_loss=0.2516, pruned_loss=0.09965, over 12491.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.2675, pruned_loss=0.08499, over 2614499.36 frames. ], batch size: 51, lr: 1.21e-02, grad_scale: 16.0 2023-04-16 18:43:20,927 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9686, 2.4443, 1.8850, 3.8147, 4.3647, 3.2101, 4.1986, 4.0279], device='cuda:0'), covar=tensor([0.0097, 0.0826, 0.1064, 0.0100, 0.0062, 0.0490, 0.0092, 0.0077], device='cuda:0'), in_proj_covar=tensor([0.0062, 0.0081, 0.0089, 0.0068, 0.0051, 0.0073, 0.0045, 0.0061], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:43:23,810 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4790, 2.2152, 2.6868, 3.8626, 3.6652, 3.8946, 3.1302, 2.2741], device='cuda:0'), covar=tensor([0.0245, 0.1043, 0.0762, 0.0059, 0.0188, 0.0049, 0.0570, 0.0990], device='cuda:0'), in_proj_covar=tensor([0.0105, 0.0148, 0.0154, 0.0084, 0.0101, 0.0077, 0.0155, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0002], device='cuda:0') 2023-04-16 18:43:26,210 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27804.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:43:30,861 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.353e+02 3.099e+02 3.629e+02 4.867e+02 1.066e+03, threshold=7.258e+02, percent-clipped=2.0 2023-04-16 18:43:50,795 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2023-04-16 18:43:58,174 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-04-16 18:43:59,878 INFO [train.py:893] (0/4) Epoch 11, batch 850, loss[loss=0.216, simple_loss=0.2585, pruned_loss=0.08673, over 13506.00 frames. ], tot_loss[loss=0.2202, simple_loss=0.2689, pruned_loss=0.08569, over 2628549.04 frames. ], batch size: 70, lr: 1.21e-02, grad_scale: 16.0 2023-04-16 18:44:10,240 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27852.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:44:14,076 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-16 18:44:30,143 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.0997, 4.9439, 5.0901, 4.9004, 5.4263, 4.8684, 5.4369, 5.3964], device='cuda:0'), covar=tensor([0.0358, 0.0496, 0.0607, 0.0503, 0.0528, 0.0820, 0.0481, 0.0585], device='cuda:0'), in_proj_covar=tensor([0.0227, 0.0250, 0.0237, 0.0179, 0.0343, 0.0285, 0.0213, 0.0241], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:44:45,320 INFO [train.py:893] (0/4) Epoch 11, batch 900, loss[loss=0.2753, simple_loss=0.3141, pruned_loss=0.1183, over 13438.00 frames. ], tot_loss[loss=0.2201, simple_loss=0.2687, pruned_loss=0.08574, over 2640582.64 frames. ], batch size: 100, lr: 1.21e-02, grad_scale: 16.0 2023-04-16 18:45:01,523 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.325e+02 3.181e+02 3.611e+02 4.274e+02 7.628e+02, threshold=7.222e+02, percent-clipped=1.0 2023-04-16 18:45:05,351 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-16 18:45:05,834 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3201, 4.1057, 4.3116, 2.8688, 4.6902, 4.4233, 4.3514, 4.6263], device='cuda:0'), covar=tensor([0.0207, 0.0116, 0.0131, 0.0894, 0.0128, 0.0218, 0.0143, 0.0119], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0040, 0.0062, 0.0087, 0.0077, 0.0076, 0.0062, 0.0054], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:45:15,493 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 18:45:30,661 INFO [train.py:893] (0/4) Epoch 11, batch 950, loss[loss=0.2384, simple_loss=0.2815, pruned_loss=0.09765, over 13588.00 frames. ], tot_loss[loss=0.22, simple_loss=0.268, pruned_loss=0.08604, over 2648875.10 frames. ], batch size: 89, lr: 1.21e-02, grad_scale: 16.0 2023-04-16 18:46:12,295 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0138, 4.4454, 4.2191, 4.1183, 4.1795, 4.0674, 4.5185, 4.4870], device='cuda:0'), covar=tensor([0.0242, 0.0203, 0.0190, 0.0318, 0.0313, 0.0264, 0.0226, 0.0236], device='cuda:0'), in_proj_covar=tensor([0.0109, 0.0162, 0.0126, 0.0147, 0.0117, 0.0157, 0.0108, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 18:46:13,709 INFO [train.py:893] (0/4) Epoch 11, batch 1000, loss[loss=0.195, simple_loss=0.2264, pruned_loss=0.08183, over 9878.00 frames. ], tot_loss[loss=0.2171, simple_loss=0.265, pruned_loss=0.08454, over 2649602.80 frames. ], batch size: 40, lr: 1.21e-02, grad_scale: 16.0 2023-04-16 18:46:17,410 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7608, 2.5118, 2.2047, 1.5725, 1.3864, 2.2302, 1.9896, 2.7415], device='cuda:0'), covar=tensor([0.1059, 0.0411, 0.0768, 0.1881, 0.0259, 0.0405, 0.1006, 0.0255], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0105, 0.0095, 0.0175, 0.0091, 0.0117, 0.0130, 0.0102], device='cuda:0'), out_proj_covar=tensor([1.0202e-04, 8.0775e-05, 7.6497e-05, 1.3651e-04, 7.2911e-05, 8.9342e-05, 1.0097e-04, 7.5809e-05], device='cuda:0') 2023-04-16 18:46:20,603 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-28000.pt 2023-04-16 18:46:34,491 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.038e+02 3.069e+02 3.661e+02 4.441e+02 9.390e+02, threshold=7.322e+02, percent-clipped=1.0 2023-04-16 18:47:03,890 INFO [train.py:893] (0/4) Epoch 11, batch 1050, loss[loss=0.2105, simple_loss=0.2669, pruned_loss=0.07703, over 13540.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.2639, pruned_loss=0.08358, over 2649560.74 frames. ], batch size: 83, lr: 1.20e-02, grad_scale: 16.0 2023-04-16 18:47:06,828 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-16 18:47:23,838 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2770, 4.8059, 4.6237, 4.4258, 4.5484, 4.3834, 4.8661, 4.8218], device='cuda:0'), covar=tensor([0.0205, 0.0190, 0.0151, 0.0347, 0.0207, 0.0249, 0.0252, 0.0208], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0165, 0.0128, 0.0150, 0.0119, 0.0160, 0.0110, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 18:47:29,424 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28070.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 18:47:47,431 INFO [train.py:893] (0/4) Epoch 11, batch 1100, loss[loss=0.2728, simple_loss=0.3015, pruned_loss=0.1221, over 11973.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.2638, pruned_loss=0.08263, over 2653926.53 frames. ], batch size: 157, lr: 1.20e-02, grad_scale: 16.0 2023-04-16 18:47:57,107 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3798, 4.7088, 4.4879, 3.8340, 4.4703, 4.7511, 4.7286, 4.6492], device='cuda:0'), covar=tensor([0.0294, 0.0258, 0.0292, 0.1606, 0.0281, 0.0368, 0.0264, 0.0200], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0127, 0.0138, 0.0232, 0.0138, 0.0154, 0.0137, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2023-04-16 18:48:04,936 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.261e+02 3.035e+02 3.453e+02 4.146e+02 7.146e+02, threshold=6.906e+02, percent-clipped=0.0 2023-04-16 18:48:16,269 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1749, 4.2508, 3.4225, 2.8402, 3.0299, 2.5072, 4.4949, 2.5343], device='cuda:0'), covar=tensor([0.1672, 0.0377, 0.0930, 0.1752, 0.0767, 0.3075, 0.0175, 0.3644], device='cuda:0'), in_proj_covar=tensor([0.0293, 0.0268, 0.0267, 0.0282, 0.0224, 0.0288, 0.0185, 0.0334], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 18:48:34,321 INFO [train.py:893] (0/4) Epoch 11, batch 1150, loss[loss=0.2232, simple_loss=0.2777, pruned_loss=0.08439, over 13536.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.263, pruned_loss=0.08153, over 2656554.77 frames. ], batch size: 87, lr: 1.20e-02, grad_scale: 16.0 2023-04-16 18:49:18,352 INFO [train.py:893] (0/4) Epoch 11, batch 1200, loss[loss=0.1991, simple_loss=0.2459, pruned_loss=0.07613, over 13354.00 frames. ], tot_loss[loss=0.2132, simple_loss=0.2632, pruned_loss=0.08155, over 2653244.15 frames. ], batch size: 73, lr: 1.20e-02, grad_scale: 16.0 2023-04-16 18:49:34,902 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.076e+02 2.988e+02 3.619e+02 4.401e+02 8.717e+02, threshold=7.238e+02, percent-clipped=6.0 2023-04-16 18:49:44,800 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 18:49:53,675 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28229.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:49:53,752 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6712, 4.0204, 2.4029, 3.8333, 3.8132, 2.3422, 3.2657, 2.5465], device='cuda:0'), covar=tensor([0.0367, 0.0295, 0.1452, 0.0350, 0.0319, 0.1321, 0.0666, 0.1571], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0141, 0.0171, 0.0135, 0.0119, 0.0153, 0.0147, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 18:49:57,728 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 18:50:04,364 INFO [train.py:893] (0/4) Epoch 11, batch 1250, loss[loss=0.2129, simple_loss=0.2642, pruned_loss=0.08078, over 13416.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2642, pruned_loss=0.08202, over 2654661.96 frames. ], batch size: 88, lr: 1.20e-02, grad_scale: 16.0 2023-04-16 18:50:16,776 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28255.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 18:50:45,837 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28287.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:50:48,932 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28290.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:50:50,186 INFO [train.py:893] (0/4) Epoch 11, batch 1300, loss[loss=0.2055, simple_loss=0.2557, pruned_loss=0.07765, over 13351.00 frames. ], tot_loss[loss=0.2162, simple_loss=0.2661, pruned_loss=0.08312, over 2658910.02 frames. ], batch size: 67, lr: 1.20e-02, grad_scale: 16.0 2023-04-16 18:51:05,261 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.287e+02 3.361e+02 3.950e+02 4.861e+02 7.192e+02, threshold=7.900e+02, percent-clipped=0.0 2023-04-16 18:51:11,882 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28316.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 18:51:34,394 INFO [train.py:893] (0/4) Epoch 11, batch 1350, loss[loss=0.2254, simple_loss=0.2815, pruned_loss=0.08464, over 13476.00 frames. ], tot_loss[loss=0.2171, simple_loss=0.2669, pruned_loss=0.0837, over 2659431.12 frames. ], batch size: 100, lr: 1.20e-02, grad_scale: 16.0 2023-04-16 18:51:40,412 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28348.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:51:55,587 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6187, 2.6433, 2.3742, 1.5033, 1.5349, 2.2821, 2.1301, 2.8478], device='cuda:0'), covar=tensor([0.0921, 0.0296, 0.0753, 0.1772, 0.0436, 0.0493, 0.0759, 0.0248], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0104, 0.0095, 0.0174, 0.0091, 0.0119, 0.0128, 0.0101], device='cuda:0'), out_proj_covar=tensor([1.0066e-04, 7.9747e-05, 7.6943e-05, 1.3606e-04, 7.3231e-05, 9.0789e-05, 9.9451e-05, 7.5412e-05], device='cuda:0') 2023-04-16 18:51:59,717 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28370.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 18:52:20,673 INFO [train.py:893] (0/4) Epoch 11, batch 1400, loss[loss=0.1826, simple_loss=0.2345, pruned_loss=0.06539, over 13489.00 frames. ], tot_loss[loss=0.2163, simple_loss=0.266, pruned_loss=0.08333, over 2661657.38 frames. ], batch size: 70, lr: 1.20e-02, grad_scale: 16.0 2023-04-16 18:52:36,463 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.913e+02 3.138e+02 3.548e+02 4.144e+02 7.337e+02, threshold=7.097e+02, percent-clipped=0.0 2023-04-16 18:52:45,465 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28418.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 18:53:06,671 INFO [train.py:893] (0/4) Epoch 11, batch 1450, loss[loss=0.1984, simple_loss=0.2522, pruned_loss=0.07235, over 13541.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.2657, pruned_loss=0.08354, over 2660091.83 frames. ], batch size: 72, lr: 1.20e-02, grad_scale: 16.0 2023-04-16 18:53:47,165 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28486.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:53:51,731 INFO [train.py:893] (0/4) Epoch 11, batch 1500, loss[loss=0.2183, simple_loss=0.2685, pruned_loss=0.08407, over 13497.00 frames. ], tot_loss[loss=0.2157, simple_loss=0.2654, pruned_loss=0.08303, over 2662268.22 frames. ], batch size: 93, lr: 1.19e-02, grad_scale: 16.0 2023-04-16 18:53:51,991 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28492.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:54:09,338 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.247e+02 3.197e+02 3.873e+02 4.387e+02 6.380e+02, threshold=7.745e+02, percent-clipped=0.0 2023-04-16 18:54:20,921 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8339, 4.3920, 4.2608, 4.2744, 4.1968, 4.0797, 4.8164, 4.4626], device='cuda:0'), covar=tensor([0.0927, 0.1241, 0.2523, 0.3077, 0.0991, 0.1756, 0.1041, 0.1017], device='cuda:0'), in_proj_covar=tensor([0.0277, 0.0313, 0.0398, 0.0404, 0.0230, 0.0301, 0.0360, 0.0277], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 18:54:38,281 INFO [train.py:893] (0/4) Epoch 11, batch 1550, loss[loss=0.1701, simple_loss=0.2248, pruned_loss=0.05774, over 13443.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2642, pruned_loss=0.08215, over 2663016.14 frames. ], batch size: 65, lr: 1.19e-02, grad_scale: 16.0 2023-04-16 18:54:42,602 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28547.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:54:47,524 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28553.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:55:15,946 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28585.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:55:20,115 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7460, 1.9318, 3.4845, 3.4603, 3.4245, 2.7550, 3.3587, 2.3694], device='cuda:0'), covar=tensor([0.2247, 0.1383, 0.0110, 0.0170, 0.0204, 0.0698, 0.0170, 0.0994], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0180, 0.0104, 0.0109, 0.0115, 0.0156, 0.0116, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:55:21,451 INFO [train.py:893] (0/4) Epoch 11, batch 1600, loss[loss=0.229, simple_loss=0.2818, pruned_loss=0.08804, over 13366.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.2648, pruned_loss=0.08217, over 2660431.26 frames. ], batch size: 109, lr: 1.19e-02, grad_scale: 16.0 2023-04-16 18:55:39,671 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.973e+02 3.051e+02 3.528e+02 4.292e+02 7.736e+02, threshold=7.056e+02, percent-clipped=0.0 2023-04-16 18:55:40,659 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28611.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 18:56:08,787 INFO [train.py:893] (0/4) Epoch 11, batch 1650, loss[loss=0.236, simple_loss=0.2815, pruned_loss=0.0953, over 13526.00 frames. ], tot_loss[loss=0.2146, simple_loss=0.2654, pruned_loss=0.08195, over 2656932.53 frames. ], batch size: 91, lr: 1.19e-02, grad_scale: 32.0 2023-04-16 18:56:09,933 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28643.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:56:54,198 INFO [train.py:893] (0/4) Epoch 11, batch 1700, loss[loss=0.1844, simple_loss=0.2419, pruned_loss=0.06348, over 13115.00 frames. ], tot_loss[loss=0.2146, simple_loss=0.2656, pruned_loss=0.08182, over 2650468.41 frames. ], batch size: 142, lr: 1.19e-02, grad_scale: 32.0 2023-04-16 18:56:56,124 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8318, 2.5205, 2.4273, 2.9980, 2.1278, 3.1312, 2.8162, 2.5763], device='cuda:0'), covar=tensor([0.0080, 0.0167, 0.0147, 0.0138, 0.0204, 0.0086, 0.0196, 0.0151], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0079, 0.0085, 0.0077, 0.0092, 0.0071, 0.0074, 0.0074], device='cuda:0'), out_proj_covar=tensor([8.0220e-05, 8.9841e-05, 9.8902e-05, 8.8750e-05, 1.0633e-04, 7.9019e-05, 8.4333e-05, 8.2314e-05], device='cuda:0') 2023-04-16 18:57:10,236 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.256e+02 2.920e+02 3.491e+02 4.240e+02 8.801e+02, threshold=6.981e+02, percent-clipped=2.0 2023-04-16 18:57:19,411 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9830, 2.4857, 2.0832, 3.8232, 4.3781, 3.2814, 4.2547, 4.0012], device='cuda:0'), covar=tensor([0.0088, 0.0821, 0.1001, 0.0105, 0.0070, 0.0447, 0.0090, 0.0076], device='cuda:0'), in_proj_covar=tensor([0.0063, 0.0081, 0.0088, 0.0069, 0.0052, 0.0072, 0.0045, 0.0062], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 18:57:34,973 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.82 vs. limit=5.0 2023-04-16 18:57:39,430 INFO [train.py:893] (0/4) Epoch 11, batch 1750, loss[loss=0.215, simple_loss=0.2662, pruned_loss=0.08189, over 13390.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.2645, pruned_loss=0.08127, over 2655210.80 frames. ], batch size: 109, lr: 1.19e-02, grad_scale: 32.0 2023-04-16 18:57:57,856 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0694, 4.5594, 4.3899, 4.2522, 4.2381, 4.1414, 4.6101, 4.6187], device='cuda:0'), covar=tensor([0.0238, 0.0182, 0.0169, 0.0284, 0.0286, 0.0237, 0.0269, 0.0218], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0169, 0.0129, 0.0150, 0.0120, 0.0166, 0.0113, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 18:58:25,773 INFO [train.py:893] (0/4) Epoch 11, batch 1800, loss[loss=0.183, simple_loss=0.248, pruned_loss=0.05902, over 13457.00 frames. ], tot_loss[loss=0.2126, simple_loss=0.2638, pruned_loss=0.08071, over 2655525.88 frames. ], batch size: 79, lr: 1.19e-02, grad_scale: 32.0 2023-04-16 18:58:38,594 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0966, 4.4207, 4.1747, 4.2509, 4.2437, 4.6404, 4.3510, 4.2694], device='cuda:0'), covar=tensor([0.0294, 0.0283, 0.0300, 0.0873, 0.0277, 0.0199, 0.0305, 0.0282], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0127, 0.0139, 0.0234, 0.0141, 0.0154, 0.0138, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2023-04-16 18:58:40,787 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.909e+02 3.056e+02 3.632e+02 4.305e+02 5.743e+02, threshold=7.265e+02, percent-clipped=0.0 2023-04-16 18:59:09,927 INFO [train.py:893] (0/4) Epoch 11, batch 1850, loss[loss=0.2167, simple_loss=0.2689, pruned_loss=0.0823, over 13530.00 frames. ], tot_loss[loss=0.2119, simple_loss=0.2627, pruned_loss=0.08055, over 2649171.96 frames. ], batch size: 85, lr: 1.19e-02, grad_scale: 32.0 2023-04-16 18:59:10,113 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28842.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:59:13,904 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 18:59:16,275 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28848.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:59:26,489 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2023-04-16 18:59:28,563 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28861.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:59:36,345 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-16 18:59:40,992 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2515, 4.1592, 3.6362, 2.9992, 3.1327, 2.4918, 4.4547, 2.5555], device='cuda:0'), covar=tensor([0.1223, 0.0298, 0.0627, 0.1539, 0.0617, 0.2700, 0.0174, 0.3114], device='cuda:0'), in_proj_covar=tensor([0.0294, 0.0270, 0.0270, 0.0284, 0.0225, 0.0288, 0.0188, 0.0336], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 18:59:50,163 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28885.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 18:59:56,440 INFO [train.py:893] (0/4) Epoch 11, batch 1900, loss[loss=0.2283, simple_loss=0.2754, pruned_loss=0.09066, over 13048.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2619, pruned_loss=0.08035, over 2652665.13 frames. ], batch size: 142, lr: 1.19e-02, grad_scale: 32.0 2023-04-16 19:00:12,396 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.034e+02 3.088e+02 3.593e+02 4.419e+02 8.610e+02, threshold=7.186e+02, percent-clipped=4.0 2023-04-16 19:00:13,565 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28911.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 19:00:24,076 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28922.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:00:33,232 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28933.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:00:41,287 INFO [train.py:893] (0/4) Epoch 11, batch 1950, loss[loss=0.2322, simple_loss=0.284, pruned_loss=0.09018, over 13405.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2614, pruned_loss=0.08046, over 2653042.22 frames. ], batch size: 113, lr: 1.19e-02, grad_scale: 32.0 2023-04-16 19:00:42,387 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28943.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:00:56,908 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28959.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 19:01:09,542 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-16 19:01:25,269 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28991.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:01:25,901 INFO [train.py:893] (0/4) Epoch 11, batch 2000, loss[loss=0.2422, simple_loss=0.2942, pruned_loss=0.09511, over 13374.00 frames. ], tot_loss[loss=0.213, simple_loss=0.2633, pruned_loss=0.08134, over 2651680.00 frames. ], batch size: 113, lr: 1.18e-02, grad_scale: 32.0 2023-04-16 19:01:33,008 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 19:01:35,544 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29002.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:01:43,241 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.955e+02 3.039e+02 3.558e+02 4.462e+02 6.294e+02, threshold=7.116e+02, percent-clipped=0.0 2023-04-16 19:02:12,959 INFO [train.py:893] (0/4) Epoch 11, batch 2050, loss[loss=0.2135, simple_loss=0.2652, pruned_loss=0.08086, over 13559.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.2653, pruned_loss=0.08183, over 2655644.70 frames. ], batch size: 89, lr: 1.18e-02, grad_scale: 32.0 2023-04-16 19:02:30,619 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29063.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:02:57,166 INFO [train.py:893] (0/4) Epoch 11, batch 2100, loss[loss=0.1942, simple_loss=0.251, pruned_loss=0.06866, over 13549.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.2645, pruned_loss=0.08124, over 2658621.62 frames. ], batch size: 78, lr: 1.18e-02, grad_scale: 32.0 2023-04-16 19:03:14,862 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.887e+02 3.135e+02 3.626e+02 4.199e+02 7.888e+02, threshold=7.251e+02, percent-clipped=1.0 2023-04-16 19:03:43,825 INFO [train.py:893] (0/4) Epoch 11, batch 2150, loss[loss=0.1904, simple_loss=0.2493, pruned_loss=0.06578, over 13475.00 frames. ], tot_loss[loss=0.2132, simple_loss=0.2647, pruned_loss=0.0808, over 2658644.01 frames. ], batch size: 81, lr: 1.18e-02, grad_scale: 32.0 2023-04-16 19:03:44,051 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29142.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:03:49,008 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29148.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:03:51,812 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-04-16 19:04:11,041 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3775, 2.9250, 2.4027, 2.7581, 2.6507, 1.7417, 3.0065, 1.9096], device='cuda:0'), covar=tensor([0.0632, 0.0809, 0.0436, 0.0408, 0.0623, 0.1882, 0.0944, 0.1230], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0113, 0.0116, 0.0100, 0.0133, 0.0170, 0.0130, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:04:27,831 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29190.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:04:29,345 INFO [train.py:893] (0/4) Epoch 11, batch 2200, loss[loss=0.2425, simple_loss=0.2759, pruned_loss=0.1045, over 11752.00 frames. ], tot_loss[loss=0.2127, simple_loss=0.2645, pruned_loss=0.08049, over 2657582.22 frames. ], batch size: 157, lr: 1.18e-02, grad_scale: 32.0 2023-04-16 19:04:34,229 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29196.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:04:45,928 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.987e+02 3.146e+02 3.718e+02 4.404e+02 7.530e+02, threshold=7.436e+02, percent-clipped=2.0 2023-04-16 19:04:51,871 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29217.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:04:53,698 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7562, 3.8079, 4.7153, 3.1676, 2.8610, 3.1495, 4.7706, 4.9455], device='cuda:0'), covar=tensor([0.1026, 0.1339, 0.0264, 0.1650, 0.1590, 0.1287, 0.0175, 0.0150], device='cuda:0'), in_proj_covar=tensor([0.0218, 0.0229, 0.0172, 0.0206, 0.0202, 0.0166, 0.0170, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:05:10,395 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29236.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:05:15,029 INFO [train.py:893] (0/4) Epoch 11, batch 2250, loss[loss=0.1898, simple_loss=0.2415, pruned_loss=0.06909, over 13530.00 frames. ], tot_loss[loss=0.2105, simple_loss=0.2618, pruned_loss=0.07957, over 2659591.10 frames. ], batch size: 76, lr: 1.18e-02, grad_scale: 32.0 2023-04-16 19:06:00,790 INFO [train.py:893] (0/4) Epoch 11, batch 2300, loss[loss=0.2204, simple_loss=0.2655, pruned_loss=0.08764, over 13491.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2608, pruned_loss=0.07875, over 2659851.82 frames. ], batch size: 81, lr: 1.18e-02, grad_scale: 32.0 2023-04-16 19:06:06,060 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29297.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:06:16,522 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.915e+02 2.601e+02 3.338e+02 4.081e+02 5.584e+02, threshold=6.676e+02, percent-clipped=0.0 2023-04-16 19:06:45,620 INFO [train.py:893] (0/4) Epoch 11, batch 2350, loss[loss=0.1923, simple_loss=0.2473, pruned_loss=0.06867, over 13482.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2601, pruned_loss=0.07825, over 2657782.81 frames. ], batch size: 81, lr: 1.18e-02, grad_scale: 32.0 2023-04-16 19:07:01,844 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29358.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:07:09,934 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 19:07:32,340 INFO [train.py:893] (0/4) Epoch 11, batch 2400, loss[loss=0.2075, simple_loss=0.2548, pruned_loss=0.08008, over 13050.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2599, pruned_loss=0.07879, over 2657978.02 frames. ], batch size: 142, lr: 1.18e-02, grad_scale: 32.0 2023-04-16 19:07:48,293 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.129e+02 3.051e+02 3.687e+02 4.571e+02 9.088e+02, threshold=7.373e+02, percent-clipped=3.0 2023-04-16 19:08:17,321 INFO [train.py:893] (0/4) Epoch 11, batch 2450, loss[loss=0.2238, simple_loss=0.2733, pruned_loss=0.0871, over 13400.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.2601, pruned_loss=0.07876, over 2658130.07 frames. ], batch size: 113, lr: 1.18e-02, grad_scale: 32.0 2023-04-16 19:09:03,792 INFO [train.py:893] (0/4) Epoch 11, batch 2500, loss[loss=0.2145, simple_loss=0.2722, pruned_loss=0.07837, over 13529.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2592, pruned_loss=0.07763, over 2659070.08 frames. ], batch size: 87, lr: 1.17e-02, grad_scale: 32.0 2023-04-16 19:09:11,536 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0517, 4.4621, 4.1719, 4.1170, 4.2229, 4.0852, 4.4745, 4.5065], device='cuda:0'), covar=tensor([0.0197, 0.0201, 0.0212, 0.0317, 0.0281, 0.0260, 0.0238, 0.0205], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0168, 0.0129, 0.0151, 0.0118, 0.0167, 0.0112, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 19:09:20,207 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.679e+02 3.058e+02 3.500e+02 4.137e+02 9.490e+02, threshold=7.000e+02, percent-clipped=2.0 2023-04-16 19:09:26,761 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29517.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:09:35,736 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9657, 3.7747, 3.9275, 2.2434, 4.3017, 4.0964, 4.0420, 4.2463], device='cuda:0'), covar=tensor([0.0216, 0.0118, 0.0124, 0.1203, 0.0131, 0.0202, 0.0128, 0.0090], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0042, 0.0064, 0.0091, 0.0081, 0.0081, 0.0065, 0.0057], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:09:48,370 INFO [train.py:893] (0/4) Epoch 11, batch 2550, loss[loss=0.2302, simple_loss=0.2752, pruned_loss=0.09261, over 13359.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2602, pruned_loss=0.0786, over 2662065.01 frames. ], batch size: 84, lr: 1.17e-02, grad_scale: 32.0 2023-04-16 19:09:56,079 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.4759, 2.1011, 1.9076, 2.4745, 1.6800, 2.5011, 2.3142, 2.1334], device='cuda:0'), covar=tensor([0.0085, 0.0193, 0.0187, 0.0175, 0.0207, 0.0117, 0.0215, 0.0171], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0080, 0.0088, 0.0079, 0.0094, 0.0072, 0.0077, 0.0076], device='cuda:0'), out_proj_covar=tensor([8.0555e-05, 9.0712e-05, 1.0155e-04, 8.9288e-05, 1.0740e-04, 7.9875e-05, 8.6746e-05, 8.4238e-05], device='cuda:0') 2023-04-16 19:09:56,882 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29551.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:10:08,228 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29565.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:10:12,105 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 19:10:28,869 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-04-16 19:10:32,324 INFO [train.py:893] (0/4) Epoch 11, batch 2600, loss[loss=0.2364, simple_loss=0.2841, pruned_loss=0.09439, over 13552.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2602, pruned_loss=0.07904, over 2648959.35 frames. ], batch size: 87, lr: 1.17e-02, grad_scale: 32.0 2023-04-16 19:10:32,501 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29592.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:10:35,070 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5603, 3.2661, 2.5586, 2.9873, 2.7022, 1.9734, 3.3207, 1.9615], device='cuda:0'), covar=tensor([0.0722, 0.0645, 0.0491, 0.0443, 0.0818, 0.1941, 0.0942, 0.1426], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0114, 0.0116, 0.0100, 0.0133, 0.0171, 0.0132, 0.0128], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:10:38,404 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9935, 4.0981, 3.3714, 2.8246, 2.9392, 2.3986, 4.2558, 2.3877], device='cuda:0'), covar=tensor([0.1271, 0.0269, 0.0731, 0.1411, 0.0640, 0.2622, 0.0198, 0.3168], device='cuda:0'), in_proj_covar=tensor([0.0294, 0.0266, 0.0270, 0.0286, 0.0225, 0.0288, 0.0187, 0.0337], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 19:10:43,959 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2023-04-16 19:10:48,305 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.167e+02 3.181e+02 3.982e+02 4.759e+02 1.056e+03, threshold=7.963e+02, percent-clipped=3.0 2023-04-16 19:10:50,335 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29612.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:11:02,473 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2023-04-16 19:11:13,230 INFO [train.py:893] (0/4) Epoch 11, batch 2650, loss[loss=0.2359, simple_loss=0.2854, pruned_loss=0.0932, over 13416.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2623, pruned_loss=0.08057, over 2652076.80 frames. ], batch size: 95, lr: 1.17e-02, grad_scale: 32.0 2023-04-16 19:11:25,466 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29658.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:11:28,388 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29662.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:11:50,970 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-11.pt 2023-04-16 19:12:15,999 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 19:12:19,421 INFO [train.py:893] (0/4) Epoch 12, batch 0, loss[loss=0.2229, simple_loss=0.2678, pruned_loss=0.08903, over 13570.00 frames. ], tot_loss[loss=0.2229, simple_loss=0.2678, pruned_loss=0.08903, over 13570.00 frames. ], batch size: 89, lr: 1.12e-02, grad_scale: 32.0 2023-04-16 19:12:19,422 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 19:12:26,197 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9454, 4.1528, 3.7427, 4.8390, 5.3139, 4.3886, 5.2189, 4.9378], device='cuda:0'), covar=tensor([0.0074, 0.0347, 0.0445, 0.0072, 0.0040, 0.0202, 0.0043, 0.0055], device='cuda:0'), in_proj_covar=tensor([0.0062, 0.0080, 0.0088, 0.0069, 0.0052, 0.0072, 0.0045, 0.0061], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:12:29,354 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.4234, 5.2965, 5.3502, 5.1520, 5.6668, 5.2801, 5.6528, 5.6519], device='cuda:0'), covar=tensor([0.0249, 0.0421, 0.0545, 0.0382, 0.0388, 0.0604, 0.0356, 0.0347], device='cuda:0'), in_proj_covar=tensor([0.0225, 0.0242, 0.0240, 0.0181, 0.0344, 0.0282, 0.0209, 0.0237], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 19:12:42,244 INFO [train.py:927] (0/4) Epoch 12, validation: loss=0.1552, simple_loss=0.2116, pruned_loss=0.04941, over 2446609.00 frames. 2023-04-16 19:12:42,245 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 19:12:44,269 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3387, 2.0212, 2.3603, 3.7033, 3.3246, 3.6903, 2.7811, 2.2307], device='cuda:0'), covar=tensor([0.0270, 0.1288, 0.1043, 0.0063, 0.0295, 0.0068, 0.0875, 0.1106], device='cuda:0'), in_proj_covar=tensor([0.0108, 0.0153, 0.0160, 0.0083, 0.0109, 0.0080, 0.0160, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0002], device='cuda:0') 2023-04-16 19:12:56,721 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29706.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:12:59,579 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-16 19:12:59,728 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.306e+02 3.231e+02 3.816e+02 4.221e+02 8.534e+02, threshold=7.632e+02, percent-clipped=1.0 2023-04-16 19:13:12,113 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29723.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 19:13:28,119 INFO [train.py:893] (0/4) Epoch 12, batch 50, loss[loss=0.2151, simple_loss=0.2604, pruned_loss=0.08494, over 13349.00 frames. ], tot_loss[loss=0.209, simple_loss=0.2578, pruned_loss=0.08009, over 598360.00 frames. ], batch size: 67, lr: 1.12e-02, grad_scale: 32.0 2023-04-16 19:13:33,381 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.75 vs. limit=5.0 2023-04-16 19:13:39,211 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-16 19:13:49,481 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1706, 4.4963, 3.0482, 4.5433, 4.4156, 2.6389, 3.7227, 2.9362], device='cuda:0'), covar=tensor([0.0313, 0.0319, 0.1100, 0.0267, 0.0265, 0.1171, 0.0605, 0.1348], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0141, 0.0170, 0.0139, 0.0118, 0.0154, 0.0147, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 19:13:50,032 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 19:13:50,032 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 19:13:50,032 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 19:13:50,041 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 19:13:50,056 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 19:13:50,069 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 19:13:50,078 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 19:13:54,616 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2023-04-16 19:14:13,986 INFO [train.py:893] (0/4) Epoch 12, batch 100, loss[loss=0.2251, simple_loss=0.2537, pruned_loss=0.09827, over 13234.00 frames. ], tot_loss[loss=0.2102, simple_loss=0.2582, pruned_loss=0.08107, over 1051596.49 frames. ], batch size: 58, lr: 1.12e-02, grad_scale: 32.0 2023-04-16 19:14:25,222 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29802.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:14:32,110 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.956e+02 3.079e+02 3.520e+02 4.186e+02 7.081e+02, threshold=7.041e+02, percent-clipped=0.0 2023-04-16 19:14:44,464 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6260, 3.9361, 3.7226, 4.3551, 2.2311, 3.1469, 4.1029, 2.3612], device='cuda:0'), covar=tensor([0.0204, 0.0560, 0.0778, 0.0560, 0.1826, 0.1113, 0.0515, 0.1965], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0167, 0.0188, 0.0196, 0.0176, 0.0185, 0.0170, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 19:15:01,256 INFO [train.py:893] (0/4) Epoch 12, batch 150, loss[loss=0.1968, simple_loss=0.2557, pruned_loss=0.06898, over 13521.00 frames. ], tot_loss[loss=0.2134, simple_loss=0.2607, pruned_loss=0.08303, over 1396869.18 frames. ], batch size: 91, lr: 1.12e-02, grad_scale: 32.0 2023-04-16 19:15:21,686 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29863.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:15:34,048 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.1606, 2.7965, 2.4616, 1.9784, 1.8115, 2.5543, 2.3808, 3.1100], device='cuda:0'), covar=tensor([0.0764, 0.0370, 0.0761, 0.1300, 0.0557, 0.0536, 0.0711, 0.0258], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0107, 0.0096, 0.0178, 0.0092, 0.0125, 0.0132, 0.0106], device='cuda:0'), out_proj_covar=tensor([1.0179e-04, 8.1926e-05, 7.7697e-05, 1.3833e-04, 7.3538e-05, 9.4661e-05, 1.0165e-04, 7.8852e-05], device='cuda:0') 2023-04-16 19:15:47,062 INFO [train.py:893] (0/4) Epoch 12, batch 200, loss[loss=0.2132, simple_loss=0.2642, pruned_loss=0.08106, over 13539.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.263, pruned_loss=0.08396, over 1672160.28 frames. ], batch size: 78, lr: 1.12e-02, grad_scale: 16.0 2023-04-16 19:15:48,003 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29892.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:16:02,669 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29907.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:16:05,677 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.117e+02 3.029e+02 3.563e+02 3.950e+02 7.099e+02, threshold=7.126e+02, percent-clipped=1.0 2023-04-16 19:16:33,330 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29940.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:16:34,026 INFO [train.py:893] (0/4) Epoch 12, batch 250, loss[loss=0.2113, simple_loss=0.2565, pruned_loss=0.08303, over 13537.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.2643, pruned_loss=0.08393, over 1896455.12 frames. ], batch size: 70, lr: 1.12e-02, grad_scale: 16.0 2023-04-16 19:17:20,730 INFO [train.py:893] (0/4) Epoch 12, batch 300, loss[loss=0.2204, simple_loss=0.2744, pruned_loss=0.08323, over 13517.00 frames. ], tot_loss[loss=0.2153, simple_loss=0.2645, pruned_loss=0.0831, over 2068992.72 frames. ], batch size: 91, lr: 1.12e-02, grad_scale: 16.0 2023-04-16 19:17:22,754 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4672, 3.4462, 4.0042, 2.8446, 2.5307, 2.5616, 4.2310, 4.3371], device='cuda:0'), covar=tensor([0.1009, 0.1341, 0.0372, 0.1542, 0.1593, 0.1596, 0.0212, 0.0165], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0231, 0.0176, 0.0211, 0.0205, 0.0169, 0.0171, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:17:30,019 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-30000.pt 2023-04-16 19:17:43,028 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.413e+02 3.134e+02 3.533e+02 4.035e+02 9.509e+02, threshold=7.065e+02, percent-clipped=1.0 2023-04-16 19:17:49,996 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30018.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 19:18:11,225 INFO [train.py:893] (0/4) Epoch 12, batch 350, loss[loss=0.2181, simple_loss=0.2726, pruned_loss=0.08182, over 13429.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.2648, pruned_loss=0.08297, over 2205150.82 frames. ], batch size: 106, lr: 1.12e-02, grad_scale: 16.0 2023-04-16 19:18:58,005 INFO [train.py:893] (0/4) Epoch 12, batch 400, loss[loss=0.2149, simple_loss=0.2715, pruned_loss=0.07916, over 13458.00 frames. ], tot_loss[loss=0.2144, simple_loss=0.2643, pruned_loss=0.08219, over 2305856.25 frames. ], batch size: 79, lr: 1.11e-02, grad_scale: 16.0 2023-04-16 19:19:16,168 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.193e+02 3.172e+02 3.511e+02 4.299e+02 8.379e+02, threshold=7.022e+02, percent-clipped=2.0 2023-04-16 19:19:30,526 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30125.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:19:44,286 INFO [train.py:893] (0/4) Epoch 12, batch 450, loss[loss=0.2361, simple_loss=0.2763, pruned_loss=0.09793, over 13528.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.2658, pruned_loss=0.08319, over 2386940.18 frames. ], batch size: 83, lr: 1.11e-02, grad_scale: 16.0 2023-04-16 19:19:44,850 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-16 19:19:51,572 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-16 19:19:53,721 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8816, 3.9561, 2.8995, 3.7018, 3.0017, 1.8962, 3.9388, 2.0224], device='cuda:0'), covar=tensor([0.0747, 0.0386, 0.0606, 0.0226, 0.0694, 0.2326, 0.0743, 0.1493], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0115, 0.0116, 0.0100, 0.0133, 0.0171, 0.0133, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:19:58,108 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.16 vs. limit=5.0 2023-04-16 19:20:00,065 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30158.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:20:08,999 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 19:20:26,252 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30186.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:20:30,076 INFO [train.py:893] (0/4) Epoch 12, batch 500, loss[loss=0.2435, simple_loss=0.2884, pruned_loss=0.09926, over 13528.00 frames. ], tot_loss[loss=0.2163, simple_loss=0.2663, pruned_loss=0.08314, over 2447972.13 frames. ], batch size: 85, lr: 1.11e-02, grad_scale: 16.0 2023-04-16 19:20:44,679 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30207.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:20:45,652 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5905, 3.5385, 3.0466, 2.5476, 2.5892, 2.1749, 3.6970, 2.1480], device='cuda:0'), covar=tensor([0.1291, 0.0338, 0.0718, 0.1515, 0.0656, 0.2842, 0.0264, 0.3142], device='cuda:0'), in_proj_covar=tensor([0.0292, 0.0267, 0.0271, 0.0284, 0.0223, 0.0288, 0.0184, 0.0334], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 19:20:46,446 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30209.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:20:48,534 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.264e+02 3.071e+02 3.696e+02 4.723e+02 7.948e+02, threshold=7.391e+02, percent-clipped=1.0 2023-04-16 19:21:15,815 INFO [train.py:893] (0/4) Epoch 12, batch 550, loss[loss=0.2278, simple_loss=0.2606, pruned_loss=0.09753, over 13358.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.2663, pruned_loss=0.08294, over 2495261.12 frames. ], batch size: 62, lr: 1.11e-02, grad_scale: 16.0 2023-04-16 19:21:21,712 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6996, 2.5166, 2.3213, 1.5141, 1.3988, 2.1251, 2.0329, 2.7019], device='cuda:0'), covar=tensor([0.0867, 0.0271, 0.0528, 0.1581, 0.0275, 0.0386, 0.0656, 0.0234], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0108, 0.0099, 0.0181, 0.0093, 0.0127, 0.0135, 0.0108], device='cuda:0'), out_proj_covar=tensor([1.0572e-04, 8.2362e-05, 7.9604e-05, 1.4042e-04, 7.4126e-05, 9.6155e-05, 1.0395e-04, 8.0489e-05], device='cuda:0') 2023-04-16 19:21:29,040 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30255.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:21:42,527 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30270.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:22:01,137 INFO [train.py:893] (0/4) Epoch 12, batch 600, loss[loss=0.2676, simple_loss=0.3014, pruned_loss=0.1169, over 11865.00 frames. ], tot_loss[loss=0.214, simple_loss=0.2639, pruned_loss=0.08204, over 2533812.88 frames. ], batch size: 157, lr: 1.11e-02, grad_scale: 16.0 2023-04-16 19:22:14,379 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.5516, 5.1595, 4.9845, 4.8878, 4.7371, 4.9013, 5.5488, 5.1746], device='cuda:0'), covar=tensor([0.0868, 0.1063, 0.2419, 0.3131, 0.1034, 0.1685, 0.1266, 0.1050], device='cuda:0'), in_proj_covar=tensor([0.0281, 0.0316, 0.0401, 0.0411, 0.0239, 0.0306, 0.0365, 0.0282], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 19:22:20,552 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.048e+02 3.005e+02 3.534e+02 4.580e+02 1.320e+03, threshold=7.068e+02, percent-clipped=5.0 2023-04-16 19:22:26,744 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30318.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 19:22:31,948 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-16 19:22:48,457 INFO [train.py:893] (0/4) Epoch 12, batch 650, loss[loss=0.2101, simple_loss=0.2632, pruned_loss=0.07852, over 13485.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2626, pruned_loss=0.0811, over 2564548.53 frames. ], batch size: 93, lr: 1.11e-02, grad_scale: 16.0 2023-04-16 19:23:04,802 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30360.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:23:10,429 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30366.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:23:14,612 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3349, 4.8552, 4.7234, 4.8243, 4.5814, 4.7016, 5.3336, 4.8750], device='cuda:0'), covar=tensor([0.0731, 0.0900, 0.2474, 0.2383, 0.0897, 0.1327, 0.0843, 0.1003], device='cuda:0'), in_proj_covar=tensor([0.0276, 0.0313, 0.0397, 0.0405, 0.0237, 0.0302, 0.0361, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 19:23:32,569 INFO [train.py:893] (0/4) Epoch 12, batch 700, loss[loss=0.2485, simple_loss=0.2845, pruned_loss=0.1063, over 13501.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2614, pruned_loss=0.08016, over 2589164.01 frames. ], batch size: 85, lr: 1.11e-02, grad_scale: 16.0 2023-04-16 19:23:52,289 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.015e+02 2.964e+02 3.447e+02 4.569e+02 1.039e+03, threshold=6.894e+02, percent-clipped=4.0 2023-04-16 19:24:00,868 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30421.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:24:20,322 INFO [train.py:893] (0/4) Epoch 12, batch 750, loss[loss=0.2317, simple_loss=0.2769, pruned_loss=0.09331, over 13243.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2616, pruned_loss=0.08023, over 2608843.75 frames. ], batch size: 117, lr: 1.11e-02, grad_scale: 16.0 2023-04-16 19:24:34,819 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30458.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:24:52,747 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8470, 2.5003, 2.3111, 1.5412, 1.4109, 2.2219, 2.0948, 2.8094], device='cuda:0'), covar=tensor([0.0933, 0.0405, 0.0827, 0.1946, 0.0402, 0.0431, 0.0832, 0.0236], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0108, 0.0098, 0.0182, 0.0091, 0.0128, 0.0135, 0.0107], device='cuda:0'), out_proj_covar=tensor([1.0579e-04, 8.2636e-05, 7.9428e-05, 1.4067e-04, 7.3247e-05, 9.7152e-05, 1.0381e-04, 7.9933e-05], device='cuda:0') 2023-04-16 19:24:56,698 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30481.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:25:04,822 INFO [train.py:893] (0/4) Epoch 12, batch 800, loss[loss=0.2204, simple_loss=0.272, pruned_loss=0.08444, over 13218.00 frames. ], tot_loss[loss=0.2125, simple_loss=0.2627, pruned_loss=0.08114, over 2618936.70 frames. ], batch size: 132, lr: 1.11e-02, grad_scale: 16.0 2023-04-16 19:25:20,056 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30506.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:25:23,615 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9881, 2.5054, 1.8149, 3.8545, 4.3124, 3.2585, 4.2416, 3.9851], device='cuda:0'), covar=tensor([0.0099, 0.0854, 0.1093, 0.0085, 0.0059, 0.0465, 0.0074, 0.0081], device='cuda:0'), in_proj_covar=tensor([0.0064, 0.0082, 0.0090, 0.0070, 0.0054, 0.0074, 0.0046, 0.0063], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:25:24,124 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.034e+02 3.051e+02 3.650e+02 4.433e+02 6.512e+02, threshold=7.301e+02, percent-clipped=0.0 2023-04-16 19:25:51,401 INFO [train.py:893] (0/4) Epoch 12, batch 850, loss[loss=0.2319, simple_loss=0.2785, pruned_loss=0.09268, over 13533.00 frames. ], tot_loss[loss=0.2139, simple_loss=0.2642, pruned_loss=0.08181, over 2627332.13 frames. ], batch size: 85, lr: 1.11e-02, grad_scale: 16.0 2023-04-16 19:26:13,921 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30565.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:26:37,262 INFO [train.py:893] (0/4) Epoch 12, batch 900, loss[loss=0.2093, simple_loss=0.2532, pruned_loss=0.08266, over 13550.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.2648, pruned_loss=0.08284, over 2635729.90 frames. ], batch size: 72, lr: 1.11e-02, grad_scale: 16.0 2023-04-16 19:26:54,220 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.806e+02 3.028e+02 3.474e+02 4.092e+02 7.864e+02, threshold=6.948e+02, percent-clipped=2.0 2023-04-16 19:26:58,836 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-16 19:27:07,884 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 19:27:11,343 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9609, 4.3529, 4.1448, 4.1005, 4.1361, 4.5592, 4.3638, 4.1042], device='cuda:0'), covar=tensor([0.0347, 0.0304, 0.0304, 0.1030, 0.0284, 0.0242, 0.0265, 0.0315], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0130, 0.0141, 0.0236, 0.0142, 0.0157, 0.0141, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2023-04-16 19:27:21,867 INFO [train.py:893] (0/4) Epoch 12, batch 950, loss[loss=0.2226, simple_loss=0.2704, pruned_loss=0.08744, over 13285.00 frames. ], tot_loss[loss=0.2149, simple_loss=0.2638, pruned_loss=0.08298, over 2642673.33 frames. ], batch size: 124, lr: 1.10e-02, grad_scale: 16.0 2023-04-16 19:27:34,255 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2079, 4.6082, 4.3246, 4.3255, 4.3242, 4.7792, 4.5683, 4.3870], device='cuda:0'), covar=tensor([0.0372, 0.0281, 0.0294, 0.1001, 0.0279, 0.0221, 0.0257, 0.0248], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0131, 0.0141, 0.0237, 0.0143, 0.0157, 0.0141, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2023-04-16 19:27:36,006 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30655.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:27:37,719 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-16 19:28:08,986 INFO [train.py:893] (0/4) Epoch 12, batch 1000, loss[loss=0.2196, simple_loss=0.2614, pruned_loss=0.08894, over 13526.00 frames. ], tot_loss[loss=0.213, simple_loss=0.2616, pruned_loss=0.08221, over 2645505.83 frames. ], batch size: 83, lr: 1.10e-02, grad_scale: 16.0 2023-04-16 19:28:25,725 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.256e+02 3.039e+02 3.506e+02 4.139e+02 8.473e+02, threshold=7.013e+02, percent-clipped=1.0 2023-04-16 19:28:30,839 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30716.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:28:30,995 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30716.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 19:28:52,241 INFO [train.py:893] (0/4) Epoch 12, batch 1050, loss[loss=0.1848, simple_loss=0.2412, pruned_loss=0.06418, over 13524.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2591, pruned_loss=0.08019, over 2651403.62 frames. ], batch size: 83, lr: 1.10e-02, grad_scale: 16.0 2023-04-16 19:28:54,193 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8653, 1.7611, 3.7023, 3.5318, 3.5162, 2.7079, 3.3910, 2.7196], device='cuda:0'), covar=tensor([0.2098, 0.1744, 0.0096, 0.0180, 0.0218, 0.0731, 0.0228, 0.0826], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0189, 0.0108, 0.0114, 0.0121, 0.0164, 0.0119, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:29:30,081 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30781.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:29:38,070 INFO [train.py:893] (0/4) Epoch 12, batch 1100, loss[loss=0.2072, simple_loss=0.266, pruned_loss=0.07423, over 13486.00 frames. ], tot_loss[loss=0.2095, simple_loss=0.2593, pruned_loss=0.07987, over 2652269.60 frames. ], batch size: 81, lr: 1.10e-02, grad_scale: 16.0 2023-04-16 19:29:38,720 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2023-04-16 19:29:56,587 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.861e+02 2.954e+02 3.511e+02 4.090e+02 1.224e+03, threshold=7.022e+02, percent-clipped=4.0 2023-04-16 19:30:12,550 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30829.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:30:23,882 INFO [train.py:893] (0/4) Epoch 12, batch 1150, loss[loss=0.2002, simple_loss=0.261, pruned_loss=0.06967, over 13483.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2596, pruned_loss=0.07945, over 2654625.28 frames. ], batch size: 81, lr: 1.10e-02, grad_scale: 16.0 2023-04-16 19:30:28,031 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2857, 4.8310, 4.6355, 4.7356, 4.3664, 4.6295, 5.2426, 4.7310], device='cuda:0'), covar=tensor([0.0640, 0.0933, 0.2108, 0.2441, 0.0978, 0.1366, 0.0923, 0.1101], device='cuda:0'), in_proj_covar=tensor([0.0282, 0.0316, 0.0401, 0.0414, 0.0238, 0.0304, 0.0366, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 19:30:45,814 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30865.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:31:10,377 INFO [train.py:893] (0/4) Epoch 12, batch 1200, loss[loss=0.1856, simple_loss=0.2305, pruned_loss=0.07035, over 13390.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2599, pruned_loss=0.07878, over 2658827.37 frames. ], batch size: 62, lr: 1.10e-02, grad_scale: 16.0 2023-04-16 19:31:22,174 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30903.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:31:30,205 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.083e+02 2.813e+02 3.265e+02 3.803e+02 5.981e+02, threshold=6.529e+02, percent-clipped=0.0 2023-04-16 19:31:32,066 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30913.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:31:39,316 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 19:31:42,249 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30925.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:31:50,859 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 19:31:51,107 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30934.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:31:58,130 INFO [train.py:893] (0/4) Epoch 12, batch 1250, loss[loss=0.2125, simple_loss=0.2695, pruned_loss=0.07782, over 13221.00 frames. ], tot_loss[loss=0.2091, simple_loss=0.2602, pruned_loss=0.07897, over 2656285.21 frames. ], batch size: 124, lr: 1.10e-02, grad_scale: 16.0 2023-04-16 19:32:06,085 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-16 19:32:17,677 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9746, 4.1090, 2.6321, 4.0124, 3.9384, 2.3321, 3.5129, 2.8151], device='cuda:0'), covar=tensor([0.0281, 0.0242, 0.1199, 0.0262, 0.0266, 0.1456, 0.0551, 0.1255], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0145, 0.0171, 0.0144, 0.0121, 0.0157, 0.0149, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 19:32:18,430 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30964.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:32:28,081 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5961, 3.3623, 2.7918, 2.9511, 2.9138, 2.0034, 3.3718, 1.8103], device='cuda:0'), covar=tensor([0.0722, 0.0530, 0.0446, 0.0439, 0.0754, 0.1762, 0.1012, 0.1581], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0117, 0.0118, 0.0101, 0.0139, 0.0174, 0.0136, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:32:31,474 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8036, 3.9884, 4.6162, 3.3232, 2.9912, 3.1367, 4.8821, 4.8792], device='cuda:0'), covar=tensor([0.1152, 0.1189, 0.0320, 0.1537, 0.1549, 0.1247, 0.0197, 0.0327], device='cuda:0'), in_proj_covar=tensor([0.0225, 0.0234, 0.0176, 0.0210, 0.0207, 0.0170, 0.0176, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:32:38,063 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30986.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:32:41,985 INFO [train.py:893] (0/4) Epoch 12, batch 1300, loss[loss=0.2144, simple_loss=0.2681, pruned_loss=0.08036, over 13426.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2605, pruned_loss=0.0789, over 2659230.12 frames. ], batch size: 65, lr: 1.10e-02, grad_scale: 16.0 2023-04-16 19:32:46,903 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30995.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:33:01,086 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.190e+02 3.401e+02 4.070e+02 4.749e+02 7.610e+02, threshold=8.140e+02, percent-clipped=5.0 2023-04-16 19:33:01,281 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31011.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 19:33:01,402 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31011.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 19:33:05,369 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31016.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:33:28,365 INFO [train.py:893] (0/4) Epoch 12, batch 1350, loss[loss=0.2413, simple_loss=0.2783, pruned_loss=0.1022, over 13471.00 frames. ], tot_loss[loss=0.2101, simple_loss=0.2613, pruned_loss=0.07942, over 2661666.83 frames. ], batch size: 100, lr: 1.10e-02, grad_scale: 16.0 2023-04-16 19:33:39,656 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.4712, 4.9770, 4.8947, 4.8516, 4.7667, 4.7196, 5.4598, 5.0257], device='cuda:0'), covar=tensor([0.0749, 0.1247, 0.2387, 0.3212, 0.1058, 0.1695, 0.1063, 0.1171], device='cuda:0'), in_proj_covar=tensor([0.0287, 0.0323, 0.0411, 0.0424, 0.0243, 0.0311, 0.0375, 0.0289], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 19:33:50,716 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31064.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:33:57,686 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31072.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 19:34:01,045 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6913, 3.4883, 2.8355, 3.0296, 2.9550, 1.9649, 3.4309, 1.9516], device='cuda:0'), covar=tensor([0.0639, 0.0473, 0.0417, 0.0405, 0.0727, 0.1876, 0.0828, 0.1456], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0116, 0.0117, 0.0101, 0.0138, 0.0173, 0.0135, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:34:02,734 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7956, 2.1766, 3.9949, 3.8079, 3.8659, 3.0693, 3.6517, 2.9886], device='cuda:0'), covar=tensor([0.2278, 0.1566, 0.0108, 0.0158, 0.0228, 0.0711, 0.0231, 0.0777], device='cuda:0'), in_proj_covar=tensor([0.0209, 0.0188, 0.0107, 0.0114, 0.0122, 0.0165, 0.0122, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:34:14,694 INFO [train.py:893] (0/4) Epoch 12, batch 1400, loss[loss=0.193, simple_loss=0.2467, pruned_loss=0.06959, over 13502.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2608, pruned_loss=0.07934, over 2660162.37 frames. ], batch size: 93, lr: 1.10e-02, grad_scale: 16.0 2023-04-16 19:34:32,861 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.986e+02 2.961e+02 3.516e+02 3.951e+02 9.779e+02, threshold=7.031e+02, percent-clipped=1.0 2023-04-16 19:35:00,589 INFO [train.py:893] (0/4) Epoch 12, batch 1450, loss[loss=0.2165, simple_loss=0.2746, pruned_loss=0.07921, over 13399.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.261, pruned_loss=0.07977, over 2659143.13 frames. ], batch size: 113, lr: 1.10e-02, grad_scale: 16.0 2023-04-16 19:35:13,394 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2950, 2.1383, 2.3719, 3.6775, 3.3435, 3.7618, 3.0813, 2.1278], device='cuda:0'), covar=tensor([0.0193, 0.1061, 0.0822, 0.0058, 0.0242, 0.0038, 0.0516, 0.0969], device='cuda:0'), in_proj_covar=tensor([0.0108, 0.0149, 0.0160, 0.0083, 0.0108, 0.0076, 0.0158, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0002], device='cuda:0') 2023-04-16 19:35:48,318 INFO [train.py:893] (0/4) Epoch 12, batch 1500, loss[loss=0.2116, simple_loss=0.2653, pruned_loss=0.07892, over 13542.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.2606, pruned_loss=0.07928, over 2659797.48 frames. ], batch size: 87, lr: 1.10e-02, grad_scale: 16.0 2023-04-16 19:36:05,515 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 3.015e+02 3.444e+02 4.195e+02 7.088e+02, threshold=6.887e+02, percent-clipped=1.0 2023-04-16 19:36:05,833 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0670, 2.0179, 2.2823, 3.2753, 3.0779, 3.3529, 2.8358, 2.1458], device='cuda:0'), covar=tensor([0.0173, 0.0942, 0.0720, 0.0069, 0.0229, 0.0046, 0.0483, 0.0837], device='cuda:0'), in_proj_covar=tensor([0.0108, 0.0150, 0.0160, 0.0084, 0.0108, 0.0077, 0.0159, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0002], device='cuda:0') 2023-04-16 19:36:34,158 INFO [train.py:893] (0/4) Epoch 12, batch 1550, loss[loss=0.1976, simple_loss=0.2491, pruned_loss=0.073, over 13465.00 frames. ], tot_loss[loss=0.209, simple_loss=0.2601, pruned_loss=0.07897, over 2654839.31 frames. ], batch size: 79, lr: 1.09e-02, grad_scale: 16.0 2023-04-16 19:36:51,526 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31259.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:37:10,908 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31281.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:37:12,057 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-16 19:37:19,868 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31290.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:37:20,594 INFO [train.py:893] (0/4) Epoch 12, batch 1600, loss[loss=0.2057, simple_loss=0.2655, pruned_loss=0.07294, over 13493.00 frames. ], tot_loss[loss=0.2091, simple_loss=0.2607, pruned_loss=0.07874, over 2659208.73 frames. ], batch size: 81, lr: 1.09e-02, grad_scale: 16.0 2023-04-16 19:37:21,676 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7314, 3.6792, 2.6262, 3.5591, 3.5941, 2.2446, 3.2416, 2.5091], device='cuda:0'), covar=tensor([0.0263, 0.0222, 0.1180, 0.0325, 0.0253, 0.1309, 0.0617, 0.1476], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0146, 0.0172, 0.0146, 0.0122, 0.0157, 0.0151, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 19:37:38,545 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.949e+02 2.989e+02 3.671e+02 4.365e+02 8.305e+02, threshold=7.342e+02, percent-clipped=2.0 2023-04-16 19:37:38,770 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31311.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:38:06,494 INFO [train.py:893] (0/4) Epoch 12, batch 1650, loss[loss=0.2311, simple_loss=0.279, pruned_loss=0.09161, over 13518.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2614, pruned_loss=0.07859, over 2655136.60 frames. ], batch size: 91, lr: 1.09e-02, grad_scale: 16.0 2023-04-16 19:38:22,888 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31359.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:38:29,437 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31367.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 19:38:45,172 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-04-16 19:38:51,402 INFO [train.py:893] (0/4) Epoch 12, batch 1700, loss[loss=0.2048, simple_loss=0.2563, pruned_loss=0.07668, over 13530.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2614, pruned_loss=0.07823, over 2659085.10 frames. ], batch size: 83, lr: 1.09e-02, grad_scale: 16.0 2023-04-16 19:39:06,183 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3322, 4.5314, 2.9524, 4.3484, 4.2814, 2.6810, 3.7682, 2.9631], device='cuda:0'), covar=tensor([0.0231, 0.0172, 0.1111, 0.0310, 0.0177, 0.1218, 0.0457, 0.1385], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0147, 0.0174, 0.0147, 0.0122, 0.0157, 0.0151, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 19:39:09,988 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31409.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:39:11,407 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.968e+02 3.094e+02 3.544e+02 4.109e+02 6.529e+02, threshold=7.088e+02, percent-clipped=0.0 2023-04-16 19:39:15,375 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.09 vs. limit=5.0 2023-04-16 19:39:39,661 INFO [train.py:893] (0/4) Epoch 12, batch 1750, loss[loss=0.1986, simple_loss=0.2576, pruned_loss=0.06979, over 13532.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.26, pruned_loss=0.07738, over 2657435.58 frames. ], batch size: 91, lr: 1.09e-02, grad_scale: 16.0 2023-04-16 19:39:46,881 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2023-04-16 19:40:06,853 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31470.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:40:17,029 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7127, 3.4197, 3.6349, 2.2805, 3.8260, 3.7586, 3.6805, 3.7028], device='cuda:0'), covar=tensor([0.0188, 0.0111, 0.0127, 0.1070, 0.0113, 0.0151, 0.0095, 0.0087], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0041, 0.0066, 0.0091, 0.0083, 0.0083, 0.0066, 0.0057], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:40:24,955 INFO [train.py:893] (0/4) Epoch 12, batch 1800, loss[loss=0.188, simple_loss=0.2402, pruned_loss=0.06783, over 13378.00 frames. ], tot_loss[loss=0.2067, simple_loss=0.2593, pruned_loss=0.07706, over 2654608.69 frames. ], batch size: 109, lr: 1.09e-02, grad_scale: 16.0 2023-04-16 19:40:25,211 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31491.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 19:40:43,242 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.162e+02 2.911e+02 3.544e+02 4.163e+02 7.056e+02, threshold=7.087e+02, percent-clipped=0.0 2023-04-16 19:41:11,533 INFO [train.py:893] (0/4) Epoch 12, batch 1850, loss[loss=0.2184, simple_loss=0.2496, pruned_loss=0.09361, over 9362.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2582, pruned_loss=0.07659, over 2652257.17 frames. ], batch size: 37, lr: 1.09e-02, grad_scale: 16.0 2023-04-16 19:41:14,108 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 19:41:21,109 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31552.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 19:41:27,640 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31559.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:41:47,514 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31581.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:41:55,740 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31590.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:41:56,305 INFO [train.py:893] (0/4) Epoch 12, batch 1900, loss[loss=0.2208, simple_loss=0.2717, pruned_loss=0.08492, over 13370.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2576, pruned_loss=0.07662, over 2653219.33 frames. ], batch size: 109, lr: 1.09e-02, grad_scale: 16.0 2023-04-16 19:42:11,509 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31607.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:42:14,659 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.793e+02 3.446e+02 4.085e+02 7.140e+02, threshold=6.892e+02, percent-clipped=1.0 2023-04-16 19:42:31,970 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31629.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:42:39,396 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31638.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:42:41,732 INFO [train.py:893] (0/4) Epoch 12, batch 1950, loss[loss=0.1835, simple_loss=0.2329, pruned_loss=0.06707, over 13441.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2575, pruned_loss=0.07688, over 2651340.19 frames. ], batch size: 65, lr: 1.09e-02, grad_scale: 16.0 2023-04-16 19:42:47,798 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8575, 3.6087, 3.8380, 2.4337, 4.1512, 3.9350, 3.8840, 3.9982], device='cuda:0'), covar=tensor([0.0230, 0.0155, 0.0137, 0.1003, 0.0132, 0.0211, 0.0146, 0.0106], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0043, 0.0068, 0.0093, 0.0085, 0.0085, 0.0067, 0.0060], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 19:42:54,243 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7579, 2.6699, 2.2902, 1.6475, 1.4456, 2.4092, 2.1600, 2.7822], device='cuda:0'), covar=tensor([0.0927, 0.0251, 0.0834, 0.1600, 0.0241, 0.0380, 0.0746, 0.0217], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0107, 0.0100, 0.0181, 0.0091, 0.0129, 0.0137, 0.0107], device='cuda:0'), out_proj_covar=tensor([1.0393e-04, 8.2059e-05, 8.0520e-05, 1.3989e-04, 7.1882e-05, 9.7852e-05, 1.0588e-04, 7.9364e-05], device='cuda:0') 2023-04-16 19:43:05,470 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31667.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 19:43:22,199 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6731, 1.7186, 3.5361, 3.4198, 3.2458, 2.6884, 3.2544, 2.4588], device='cuda:0'), covar=tensor([0.2228, 0.1795, 0.0156, 0.0228, 0.0289, 0.0761, 0.0241, 0.0947], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0186, 0.0106, 0.0115, 0.0120, 0.0164, 0.0121, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:43:28,475 INFO [train.py:893] (0/4) Epoch 12, batch 2000, loss[loss=0.224, simple_loss=0.2769, pruned_loss=0.08559, over 13524.00 frames. ], tot_loss[loss=0.2091, simple_loss=0.2607, pruned_loss=0.07868, over 2656642.99 frames. ], batch size: 91, lr: 1.09e-02, grad_scale: 16.0 2023-04-16 19:43:33,283 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 19:43:39,303 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31704.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:43:45,686 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.068e+02 3.036e+02 3.598e+02 4.473e+02 7.037e+02, threshold=7.196e+02, percent-clipped=1.0 2023-04-16 19:43:49,147 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31715.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 19:44:13,467 INFO [train.py:893] (0/4) Epoch 12, batch 2050, loss[loss=0.1947, simple_loss=0.2515, pruned_loss=0.06901, over 13459.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2618, pruned_loss=0.07952, over 2658532.90 frames. ], batch size: 100, lr: 1.09e-02, grad_scale: 16.0 2023-04-16 19:44:35,331 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31765.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:44:35,428 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31765.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:44:57,184 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31788.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:44:59,366 INFO [train.py:893] (0/4) Epoch 12, batch 2100, loss[loss=0.1686, simple_loss=0.2297, pruned_loss=0.05374, over 13535.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.2613, pruned_loss=0.079, over 2657259.36 frames. ], batch size: 76, lr: 1.09e-02, grad_scale: 16.0 2023-04-16 19:45:18,211 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.630e+02 2.912e+02 3.561e+02 4.179e+02 8.521e+02, threshold=7.122e+02, percent-clipped=3.0 2023-04-16 19:45:29,035 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31823.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:45:45,575 INFO [train.py:893] (0/4) Epoch 12, batch 2150, loss[loss=0.2152, simple_loss=0.2695, pruned_loss=0.08042, over 13277.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2607, pruned_loss=0.07818, over 2659830.90 frames. ], batch size: 124, lr: 1.08e-02, grad_scale: 16.0 2023-04-16 19:45:51,511 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31847.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 19:45:53,368 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31849.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:46:21,038 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31878.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:46:26,174 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31884.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:46:31,452 INFO [train.py:893] (0/4) Epoch 12, batch 2200, loss[loss=0.1776, simple_loss=0.2391, pruned_loss=0.05803, over 13563.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2614, pruned_loss=0.07846, over 2645653.91 frames. ], batch size: 78, lr: 1.08e-02, grad_scale: 32.0 2023-04-16 19:46:50,882 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.988e+02 2.906e+02 3.487e+02 4.056e+02 8.174e+02, threshold=6.975e+02, percent-clipped=2.0 2023-04-16 19:46:54,587 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31915.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:47:03,818 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9119, 1.8751, 3.8224, 3.6283, 3.5955, 2.8055, 3.3231, 2.6688], device='cuda:0'), covar=tensor([0.2079, 0.1772, 0.0094, 0.0182, 0.0172, 0.0730, 0.0266, 0.0873], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0186, 0.0106, 0.0114, 0.0120, 0.0164, 0.0122, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:47:16,807 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31939.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:47:18,211 INFO [train.py:893] (0/4) Epoch 12, batch 2250, loss[loss=0.1691, simple_loss=0.2247, pruned_loss=0.05677, over 13534.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2605, pruned_loss=0.07804, over 2649228.15 frames. ], batch size: 72, lr: 1.08e-02, grad_scale: 32.0 2023-04-16 19:47:24,496 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.67 vs. limit=5.0 2023-04-16 19:47:26,122 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-16 19:47:26,879 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31950.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:47:51,294 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31976.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:48:04,679 INFO [train.py:893] (0/4) Epoch 12, batch 2300, loss[loss=0.2161, simple_loss=0.267, pruned_loss=0.08263, over 13518.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2593, pruned_loss=0.07713, over 2653624.19 frames. ], batch size: 91, lr: 1.08e-02, grad_scale: 32.0 2023-04-16 19:48:10,899 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5266, 2.2127, 2.6677, 4.0630, 3.6115, 4.0868, 3.2316, 2.4566], device='cuda:0'), covar=tensor([0.0280, 0.1133, 0.0786, 0.0041, 0.0214, 0.0045, 0.0573, 0.0908], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0149, 0.0159, 0.0085, 0.0111, 0.0079, 0.0160, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0002], device='cuda:0') 2023-04-16 19:48:13,354 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-32000.pt 2023-04-16 19:48:27,151 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.818e+02 2.839e+02 3.553e+02 4.109e+02 7.664e+02, threshold=7.105e+02, percent-clipped=2.0 2023-04-16 19:48:27,490 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=32011.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:48:54,939 INFO [train.py:893] (0/4) Epoch 12, batch 2350, loss[loss=0.239, simple_loss=0.2877, pruned_loss=0.09513, over 13467.00 frames. ], tot_loss[loss=0.205, simple_loss=0.2579, pruned_loss=0.07606, over 2659233.22 frames. ], batch size: 100, lr: 1.08e-02, grad_scale: 32.0 2023-04-16 19:49:13,264 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32060.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:49:15,709 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 19:49:17,443 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32065.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:49:41,338 INFO [train.py:893] (0/4) Epoch 12, batch 2400, loss[loss=0.1915, simple_loss=0.2514, pruned_loss=0.06576, over 13543.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2561, pruned_loss=0.07534, over 2657636.14 frames. ], batch size: 83, lr: 1.08e-02, grad_scale: 32.0 2023-04-16 19:49:51,535 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0611, 3.7864, 3.2502, 3.5246, 3.1781, 2.1155, 3.7816, 2.2324], device='cuda:0'), covar=tensor([0.0531, 0.0464, 0.0322, 0.0321, 0.0699, 0.1930, 0.0849, 0.1311], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0120, 0.0121, 0.0101, 0.0141, 0.0176, 0.0140, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:49:59,434 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.098e+02 2.793e+02 3.359e+02 3.984e+02 7.428e+02, threshold=6.718e+02, percent-clipped=1.0 2023-04-16 19:50:01,217 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32113.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:50:27,473 INFO [train.py:893] (0/4) Epoch 12, batch 2450, loss[loss=0.1781, simple_loss=0.232, pruned_loss=0.0621, over 13512.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2561, pruned_loss=0.07515, over 2657155.03 frames. ], batch size: 70, lr: 1.08e-02, grad_scale: 32.0 2023-04-16 19:50:30,171 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32144.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:50:33,464 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32147.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 19:51:03,287 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32179.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:51:14,866 INFO [train.py:893] (0/4) Epoch 12, batch 2500, loss[loss=0.2188, simple_loss=0.2786, pruned_loss=0.07948, over 13446.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2557, pruned_loss=0.07502, over 2659251.17 frames. ], batch size: 103, lr: 1.08e-02, grad_scale: 32.0 2023-04-16 19:51:17,802 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4420, 3.4640, 4.1253, 3.0104, 2.7343, 2.8632, 4.3582, 4.4944], device='cuda:0'), covar=tensor([0.1010, 0.1453, 0.0313, 0.1384, 0.1391, 0.1202, 0.0207, 0.0164], device='cuda:0'), in_proj_covar=tensor([0.0220, 0.0232, 0.0171, 0.0204, 0.0200, 0.0165, 0.0169, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:51:18,508 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32195.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 19:51:33,612 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.938e+02 2.861e+02 3.286e+02 4.044e+02 6.156e+02, threshold=6.571e+02, percent-clipped=0.0 2023-04-16 19:51:55,431 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32234.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:52:01,066 INFO [train.py:893] (0/4) Epoch 12, batch 2550, loss[loss=0.2279, simple_loss=0.2771, pruned_loss=0.08939, over 13540.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2555, pruned_loss=0.07445, over 2661208.58 frames. ], batch size: 98, lr: 1.08e-02, grad_scale: 32.0 2023-04-16 19:52:14,767 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0530, 4.2354, 3.5657, 2.9465, 3.0121, 2.4288, 4.3997, 2.4465], device='cuda:0'), covar=tensor([0.1417, 0.0261, 0.0702, 0.1472, 0.0690, 0.3038, 0.0162, 0.3495], device='cuda:0'), in_proj_covar=tensor([0.0295, 0.0269, 0.0276, 0.0291, 0.0230, 0.0295, 0.0185, 0.0340], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 19:52:26,034 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 19:52:29,378 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32271.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:52:48,282 INFO [train.py:893] (0/4) Epoch 12, batch 2600, loss[loss=0.1848, simple_loss=0.2448, pruned_loss=0.06239, over 13271.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2562, pruned_loss=0.0748, over 2664142.94 frames. ], batch size: 124, lr: 1.08e-02, grad_scale: 32.0 2023-04-16 19:53:02,236 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32306.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:53:07,052 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.200e+02 2.795e+02 3.210e+02 3.908e+02 6.210e+02, threshold=6.421e+02, percent-clipped=0.0 2023-04-16 19:53:30,364 INFO [train.py:893] (0/4) Epoch 12, batch 2650, loss[loss=0.2482, simple_loss=0.2874, pruned_loss=0.1045, over 11963.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2579, pruned_loss=0.07657, over 2663022.61 frames. ], batch size: 157, lr: 1.08e-02, grad_scale: 32.0 2023-04-16 19:53:44,863 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32360.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:54:08,873 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-12.pt 2023-04-16 19:54:34,006 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 19:54:37,920 INFO [train.py:893] (0/4) Epoch 13, batch 0, loss[loss=0.2044, simple_loss=0.257, pruned_loss=0.07588, over 13535.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.257, pruned_loss=0.07588, over 13535.00 frames. ], batch size: 87, lr: 1.03e-02, grad_scale: 16.0 2023-04-16 19:54:37,921 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 19:55:00,634 INFO [train.py:927] (0/4) Epoch 13, validation: loss=0.1517, simple_loss=0.209, pruned_loss=0.04718, over 2446609.00 frames. 2023-04-16 19:55:00,635 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 19:55:13,728 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0218, 2.6177, 2.4263, 1.9611, 1.5875, 2.4473, 2.3456, 2.9586], device='cuda:0'), covar=tensor([0.0928, 0.0359, 0.0818, 0.1561, 0.0378, 0.0408, 0.0724, 0.0293], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0111, 0.0103, 0.0182, 0.0094, 0.0133, 0.0138, 0.0110], device='cuda:0'), out_proj_covar=tensor([1.0553e-04, 8.4610e-05, 8.2241e-05, 1.4053e-04, 7.4119e-05, 1.0028e-04, 1.0660e-04, 8.1965e-05], device='cuda:0') 2023-04-16 19:55:17,715 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32408.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:55:20,770 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.013e+02 3.015e+02 3.552e+02 4.254e+02 7.750e+02, threshold=7.105e+02, percent-clipped=5.0 2023-04-16 19:55:21,975 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1020, 4.9169, 5.1520, 4.9147, 5.4553, 4.9121, 5.4665, 5.4485], device='cuda:0'), covar=tensor([0.0362, 0.0541, 0.0670, 0.0562, 0.0495, 0.0801, 0.0381, 0.0433], device='cuda:0'), in_proj_covar=tensor([0.0234, 0.0252, 0.0250, 0.0186, 0.0352, 0.0288, 0.0218, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 19:55:32,844 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7055, 2.5379, 2.9768, 4.2085, 3.6582, 4.2488, 3.2434, 2.6068], device='cuda:0'), covar=tensor([0.0268, 0.1069, 0.0710, 0.0048, 0.0280, 0.0034, 0.0708, 0.0970], device='cuda:0'), in_proj_covar=tensor([0.0108, 0.0148, 0.0154, 0.0083, 0.0107, 0.0076, 0.0157, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0002], device='cuda:0') 2023-04-16 19:55:47,343 INFO [train.py:893] (0/4) Epoch 13, batch 50, loss[loss=0.2119, simple_loss=0.2657, pruned_loss=0.07908, over 13541.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2496, pruned_loss=0.07494, over 597377.40 frames. ], batch size: 87, lr: 1.03e-02, grad_scale: 16.0 2023-04-16 19:55:49,307 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9659, 4.8157, 5.0688, 4.8122, 5.3415, 4.8045, 5.3664, 5.3341], device='cuda:0'), covar=tensor([0.0355, 0.0592, 0.0685, 0.0649, 0.0556, 0.0853, 0.0418, 0.0483], device='cuda:0'), in_proj_covar=tensor([0.0236, 0.0255, 0.0251, 0.0187, 0.0355, 0.0290, 0.0219, 0.0248], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 19:55:50,946 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32444.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:56:05,859 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-16 19:56:10,480 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 19:56:10,481 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 19:56:10,481 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 19:56:10,488 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 19:56:10,496 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 19:56:11,264 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 19:56:12,031 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 19:56:23,196 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32479.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:56:33,364 INFO [train.py:893] (0/4) Epoch 13, batch 100, loss[loss=0.1954, simple_loss=0.244, pruned_loss=0.07336, over 13362.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2538, pruned_loss=0.07769, over 1053745.87 frames. ], batch size: 118, lr: 1.03e-02, grad_scale: 16.0 2023-04-16 19:56:35,912 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32492.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:56:53,988 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.249e+02 2.922e+02 3.454e+02 3.991e+02 1.041e+03, threshold=6.908e+02, percent-clipped=2.0 2023-04-16 19:57:07,968 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32527.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:57:14,731 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32534.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:57:19,384 INFO [train.py:893] (0/4) Epoch 13, batch 150, loss[loss=0.2128, simple_loss=0.2557, pruned_loss=0.08493, over 13453.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2577, pruned_loss=0.07969, over 1411840.03 frames. ], batch size: 79, lr: 1.03e-02, grad_scale: 16.0 2023-04-16 19:57:48,414 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32571.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:57:58,989 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32582.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:58:06,372 INFO [train.py:893] (0/4) Epoch 13, batch 200, loss[loss=0.2365, simple_loss=0.2788, pruned_loss=0.09709, over 13429.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2608, pruned_loss=0.08139, over 1689524.88 frames. ], batch size: 95, lr: 1.03e-02, grad_scale: 16.0 2023-04-16 19:58:21,024 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32606.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:58:27,221 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.025e+02 3.162e+02 3.651e+02 4.238e+02 1.178e+03, threshold=7.301e+02, percent-clipped=2.0 2023-04-16 19:58:33,897 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32619.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:58:38,138 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.4324, 4.9020, 4.8855, 4.8835, 4.6454, 4.7408, 5.4179, 4.9595], device='cuda:0'), covar=tensor([0.0735, 0.1056, 0.2034, 0.2595, 0.0798, 0.1742, 0.0777, 0.0940], device='cuda:0'), in_proj_covar=tensor([0.0287, 0.0319, 0.0411, 0.0420, 0.0242, 0.0314, 0.0368, 0.0292], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 19:58:41,490 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4957, 3.0175, 2.4539, 4.3288, 4.9245, 3.5136, 4.8161, 4.4952], device='cuda:0'), covar=tensor([0.0089, 0.0619, 0.0808, 0.0088, 0.0050, 0.0397, 0.0064, 0.0074], device='cuda:0'), in_proj_covar=tensor([0.0064, 0.0082, 0.0088, 0.0070, 0.0054, 0.0073, 0.0046, 0.0064], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 19:58:52,785 INFO [train.py:893] (0/4) Epoch 13, batch 250, loss[loss=0.2002, simple_loss=0.2524, pruned_loss=0.07406, over 13521.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2611, pruned_loss=0.08111, over 1903296.77 frames. ], batch size: 76, lr: 1.03e-02, grad_scale: 16.0 2023-04-16 19:59:06,380 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32654.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 19:59:32,480 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-16 19:59:40,133 INFO [train.py:893] (0/4) Epoch 13, batch 300, loss[loss=0.2165, simple_loss=0.2724, pruned_loss=0.08033, over 13527.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2611, pruned_loss=0.08041, over 2068677.03 frames. ], batch size: 98, lr: 1.03e-02, grad_scale: 16.0 2023-04-16 19:59:57,248 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2023-04-16 19:59:59,897 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.826e+02 2.901e+02 3.465e+02 4.067e+02 1.042e+03, threshold=6.930e+02, percent-clipped=2.0 2023-04-16 20:00:26,610 INFO [train.py:893] (0/4) Epoch 13, batch 350, loss[loss=0.1954, simple_loss=0.255, pruned_loss=0.06787, over 13534.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2621, pruned_loss=0.08113, over 2195198.51 frames. ], batch size: 76, lr: 1.03e-02, grad_scale: 16.0 2023-04-16 20:01:03,312 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9180, 3.7266, 3.9913, 2.2945, 4.2723, 4.0419, 4.0056, 4.2270], device='cuda:0'), covar=tensor([0.0261, 0.0160, 0.0133, 0.1205, 0.0148, 0.0226, 0.0151, 0.0089], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0043, 0.0066, 0.0092, 0.0082, 0.0084, 0.0066, 0.0059], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:01:12,943 INFO [train.py:893] (0/4) Epoch 13, batch 400, loss[loss=0.1878, simple_loss=0.2458, pruned_loss=0.06489, over 13353.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2613, pruned_loss=0.08003, over 2301112.37 frames. ], batch size: 73, lr: 1.03e-02, grad_scale: 16.0 2023-04-16 20:01:33,957 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.169e+02 2.983e+02 3.508e+02 4.207e+02 1.024e+03, threshold=7.015e+02, percent-clipped=2.0 2023-04-16 20:01:59,757 INFO [train.py:893] (0/4) Epoch 13, batch 450, loss[loss=0.2358, simple_loss=0.2824, pruned_loss=0.09458, over 13045.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2637, pruned_loss=0.08125, over 2382360.29 frames. ], batch size: 142, lr: 1.03e-02, grad_scale: 16.0 2023-04-16 20:02:26,239 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 20:02:46,452 INFO [train.py:893] (0/4) Epoch 13, batch 500, loss[loss=0.1871, simple_loss=0.2466, pruned_loss=0.06386, over 13539.00 frames. ], tot_loss[loss=0.2121, simple_loss=0.2636, pruned_loss=0.08034, over 2444677.00 frames. ], batch size: 76, lr: 1.03e-02, grad_scale: 16.0 2023-04-16 20:03:07,864 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.049e+02 2.979e+02 3.537e+02 4.144e+02 6.731e+02, threshold=7.074e+02, percent-clipped=0.0 2023-04-16 20:03:33,887 INFO [train.py:893] (0/4) Epoch 13, batch 550, loss[loss=0.1927, simple_loss=0.2362, pruned_loss=0.0746, over 13151.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2629, pruned_loss=0.07975, over 2492805.42 frames. ], batch size: 58, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:03:58,265 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=32966.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:04:20,143 INFO [train.py:893] (0/4) Epoch 13, batch 600, loss[loss=0.2209, simple_loss=0.2678, pruned_loss=0.08703, over 13438.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2617, pruned_loss=0.07917, over 2524412.71 frames. ], batch size: 103, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:04:41,168 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.223e+02 2.895e+02 3.369e+02 4.031e+02 6.824e+02, threshold=6.738e+02, percent-clipped=0.0 2023-04-16 20:04:41,847 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-16 20:04:55,524 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33027.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:05:07,991 INFO [train.py:893] (0/4) Epoch 13, batch 650, loss[loss=0.2103, simple_loss=0.2739, pruned_loss=0.07332, over 13383.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2601, pruned_loss=0.07845, over 2557487.17 frames. ], batch size: 109, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:05:15,837 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4906, 4.2839, 4.5395, 4.5284, 4.7729, 4.3600, 4.8279, 4.8062], device='cuda:0'), covar=tensor([0.0426, 0.0617, 0.0645, 0.0469, 0.0574, 0.0788, 0.0399, 0.0417], device='cuda:0'), in_proj_covar=tensor([0.0238, 0.0254, 0.0254, 0.0186, 0.0362, 0.0293, 0.0222, 0.0249], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:05:26,589 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-16 20:05:34,814 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33069.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:05:42,284 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1143, 4.4147, 2.8932, 4.1371, 4.3010, 2.8686, 3.6713, 3.0129], device='cuda:0'), covar=tensor([0.0292, 0.0240, 0.1111, 0.0342, 0.0195, 0.1102, 0.0565, 0.1189], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0147, 0.0173, 0.0152, 0.0123, 0.0155, 0.0149, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:05:47,016 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2773, 2.5489, 2.3226, 4.1873, 4.7931, 3.4367, 4.7143, 4.3385], device='cuda:0'), covar=tensor([0.0117, 0.0916, 0.1006, 0.0127, 0.0068, 0.0507, 0.0079, 0.0097], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0085, 0.0092, 0.0072, 0.0056, 0.0076, 0.0048, 0.0066], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:05:54,199 INFO [train.py:893] (0/4) Epoch 13, batch 700, loss[loss=0.2027, simple_loss=0.247, pruned_loss=0.07924, over 13507.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2583, pruned_loss=0.07739, over 2581814.12 frames. ], batch size: 70, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:06:14,877 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.165e+02 2.977e+02 3.487e+02 4.255e+02 9.376e+02, threshold=6.973e+02, percent-clipped=2.0 2023-04-16 20:06:31,807 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33130.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:06:40,700 INFO [train.py:893] (0/4) Epoch 13, batch 750, loss[loss=0.1842, simple_loss=0.2294, pruned_loss=0.06949, over 13189.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.258, pruned_loss=0.07739, over 2599729.22 frames. ], batch size: 58, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:06:54,902 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33155.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:07:27,556 INFO [train.py:893] (0/4) Epoch 13, batch 800, loss[loss=0.2052, simple_loss=0.255, pruned_loss=0.07773, over 12035.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2591, pruned_loss=0.07754, over 2615164.12 frames. ], batch size: 157, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:07:28,690 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8635, 1.9361, 3.9350, 3.8007, 3.7224, 2.9990, 3.4757, 2.9073], device='cuda:0'), covar=tensor([0.2142, 0.1590, 0.0079, 0.0165, 0.0205, 0.0673, 0.0214, 0.0666], device='cuda:0'), in_proj_covar=tensor([0.0212, 0.0186, 0.0110, 0.0118, 0.0123, 0.0170, 0.0125, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:07:48,342 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.106e+02 3.031e+02 3.657e+02 4.273e+02 9.126e+02, threshold=7.314e+02, percent-clipped=2.0 2023-04-16 20:07:52,067 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33216.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:08:11,954 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0969, 2.7452, 2.7404, 1.9772, 1.8053, 2.6601, 2.4395, 3.0405], device='cuda:0'), covar=tensor([0.0975, 0.0292, 0.0564, 0.1606, 0.0557, 0.0474, 0.0837, 0.0255], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0111, 0.0102, 0.0185, 0.0095, 0.0131, 0.0140, 0.0111], device='cuda:0'), out_proj_covar=tensor([1.0737e-04, 8.4683e-05, 8.2062e-05, 1.4232e-04, 7.4543e-05, 9.9270e-05, 1.0749e-04, 8.2749e-05], device='cuda:0') 2023-04-16 20:08:14,171 INFO [train.py:893] (0/4) Epoch 13, batch 850, loss[loss=0.214, simple_loss=0.2643, pruned_loss=0.08187, over 13569.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2596, pruned_loss=0.07781, over 2628988.75 frames. ], batch size: 78, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:08:58,071 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-16 20:09:00,999 INFO [train.py:893] (0/4) Epoch 13, batch 900, loss[loss=0.2024, simple_loss=0.2564, pruned_loss=0.07421, over 13347.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2595, pruned_loss=0.07784, over 2638525.42 frames. ], batch size: 73, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:09:08,640 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5557, 3.2504, 4.5817, 4.3230, 4.4899, 3.9008, 4.2664, 3.8150], device='cuda:0'), covar=tensor([0.0964, 0.0868, 0.0061, 0.0155, 0.0118, 0.0371, 0.0138, 0.0437], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0183, 0.0109, 0.0115, 0.0121, 0.0168, 0.0124, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:09:18,834 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-16 20:09:21,683 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.055e+02 2.966e+02 3.370e+02 4.159e+02 7.301e+02, threshold=6.740e+02, percent-clipped=0.0 2023-04-16 20:09:31,074 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33322.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:09:32,711 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 20:09:48,323 INFO [train.py:893] (0/4) Epoch 13, batch 950, loss[loss=0.2117, simple_loss=0.256, pruned_loss=0.08373, over 13566.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2585, pruned_loss=0.07802, over 2642572.93 frames. ], batch size: 78, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:10:12,607 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8763, 2.3807, 2.2889, 2.9896, 2.3025, 2.9436, 2.5796, 2.5112], device='cuda:0'), covar=tensor([0.0062, 0.0218, 0.0120, 0.0098, 0.0151, 0.0090, 0.0195, 0.0130], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0085, 0.0093, 0.0084, 0.0098, 0.0078, 0.0082, 0.0080], device='cuda:0'), out_proj_covar=tensor([8.3043e-05, 9.4985e-05, 1.0520e-04, 9.3570e-05, 1.1011e-04, 8.4894e-05, 9.1666e-05, 8.7072e-05], device='cuda:0') 2023-04-16 20:10:26,765 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0378, 4.5634, 4.2010, 4.3404, 4.2432, 4.2106, 4.5927, 4.6296], device='cuda:0'), covar=tensor([0.0249, 0.0214, 0.0251, 0.0289, 0.0308, 0.0255, 0.0259, 0.0221], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0172, 0.0133, 0.0154, 0.0122, 0.0169, 0.0116, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 20:10:31,227 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.28 vs. limit=5.0 2023-04-16 20:10:34,633 INFO [train.py:893] (0/4) Epoch 13, batch 1000, loss[loss=0.1804, simple_loss=0.2311, pruned_loss=0.06487, over 13432.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.257, pruned_loss=0.07771, over 2644467.26 frames. ], batch size: 106, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:10:55,540 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.940e+02 2.886e+02 3.416e+02 4.096e+02 6.022e+02, threshold=6.831e+02, percent-clipped=0.0 2023-04-16 20:11:07,511 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33425.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:11:21,171 INFO [train.py:893] (0/4) Epoch 13, batch 1050, loss[loss=0.1826, simple_loss=0.2399, pruned_loss=0.06262, over 13539.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2556, pruned_loss=0.07664, over 2646819.57 frames. ], batch size: 83, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:11:52,178 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33472.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:11:55,499 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9437, 4.0605, 3.3277, 2.7900, 2.9061, 2.4298, 4.1918, 2.3779], device='cuda:0'), covar=tensor([0.1376, 0.0337, 0.0734, 0.1582, 0.0672, 0.2699, 0.0197, 0.3553], device='cuda:0'), in_proj_covar=tensor([0.0299, 0.0274, 0.0281, 0.0296, 0.0235, 0.0300, 0.0190, 0.0342], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 20:12:08,477 INFO [train.py:893] (0/4) Epoch 13, batch 1100, loss[loss=0.1784, simple_loss=0.2286, pruned_loss=0.06408, over 13341.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.2561, pruned_loss=0.07599, over 2653090.62 frames. ], batch size: 62, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:12:15,588 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3510, 4.0402, 3.9729, 4.2521, 4.5245, 3.8940, 4.5007, 4.5741], device='cuda:0'), covar=tensor([0.0966, 0.1648, 0.1963, 0.1051, 0.1663, 0.2107, 0.0884, 0.0989], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0257, 0.0254, 0.0187, 0.0365, 0.0296, 0.0223, 0.0253], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:12:28,312 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33511.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:12:28,913 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.702e+02 2.943e+02 3.547e+02 4.473e+02 8.261e+02, threshold=7.094e+02, percent-clipped=4.0 2023-04-16 20:12:48,784 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33533.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:12:53,812 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8808, 4.3846, 4.3105, 4.4097, 4.1474, 4.1706, 4.8717, 4.4563], device='cuda:0'), covar=tensor([0.0795, 0.1303, 0.2423, 0.2702, 0.1008, 0.1672, 0.0937, 0.1033], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0315, 0.0403, 0.0411, 0.0245, 0.0311, 0.0369, 0.0287], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 20:12:55,321 INFO [train.py:893] (0/4) Epoch 13, batch 1150, loss[loss=0.2021, simple_loss=0.2535, pruned_loss=0.07534, over 11846.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.2567, pruned_loss=0.07581, over 2649459.23 frames. ], batch size: 157, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:13:15,158 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3496, 3.2988, 3.8409, 2.7442, 2.3564, 2.5749, 4.1095, 4.2024], device='cuda:0'), covar=tensor([0.1146, 0.1453, 0.0345, 0.1494, 0.1649, 0.1442, 0.0273, 0.0337], device='cuda:0'), in_proj_covar=tensor([0.0226, 0.0238, 0.0175, 0.0210, 0.0207, 0.0171, 0.0179, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:13:43,281 INFO [train.py:893] (0/4) Epoch 13, batch 1200, loss[loss=0.1961, simple_loss=0.2489, pruned_loss=0.07166, over 13531.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2569, pruned_loss=0.07527, over 2653562.15 frames. ], batch size: 76, lr: 1.02e-02, grad_scale: 16.0 2023-04-16 20:14:03,746 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.965e+02 2.832e+02 3.223e+02 3.875e+02 8.978e+02, threshold=6.447e+02, percent-clipped=1.0 2023-04-16 20:14:09,590 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 20:14:13,898 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33622.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:14:22,933 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 20:14:29,586 INFO [train.py:893] (0/4) Epoch 13, batch 1250, loss[loss=0.1913, simple_loss=0.2443, pruned_loss=0.06917, over 13538.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2569, pruned_loss=0.07548, over 2655102.59 frames. ], batch size: 72, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:14:57,796 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=33670.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:15:04,620 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33677.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:15:16,450 INFO [train.py:893] (0/4) Epoch 13, batch 1300, loss[loss=0.1852, simple_loss=0.2458, pruned_loss=0.06226, over 13431.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2586, pruned_loss=0.07627, over 2658547.17 frames. ], batch size: 103, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:15:32,633 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8548, 2.8261, 3.2396, 4.4505, 3.9091, 4.4444, 3.6893, 2.8500], device='cuda:0'), covar=tensor([0.0260, 0.0985, 0.0702, 0.0037, 0.0232, 0.0038, 0.0558, 0.0914], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0152, 0.0158, 0.0084, 0.0113, 0.0080, 0.0162, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:15:37,996 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.940e+02 3.082e+02 3.569e+02 4.063e+02 6.502e+02, threshold=7.138e+02, percent-clipped=1.0 2023-04-16 20:15:44,473 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.57 vs. limit=5.0 2023-04-16 20:15:49,979 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33725.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:15:53,392 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3154, 2.1254, 2.5979, 3.7698, 3.3878, 3.7569, 3.1404, 2.0803], device='cuda:0'), covar=tensor([0.0298, 0.1063, 0.0789, 0.0051, 0.0243, 0.0047, 0.0546, 0.1072], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0152, 0.0158, 0.0084, 0.0113, 0.0080, 0.0161, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:16:01,591 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33738.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 20:16:03,654 INFO [train.py:893] (0/4) Epoch 13, batch 1350, loss[loss=0.1767, simple_loss=0.2383, pruned_loss=0.05757, over 13553.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2591, pruned_loss=0.07654, over 2663713.92 frames. ], batch size: 78, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:16:31,494 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7119, 3.9612, 2.5276, 3.7406, 3.7592, 2.2330, 3.4141, 2.6769], device='cuda:0'), covar=tensor([0.0294, 0.0294, 0.1348, 0.0350, 0.0301, 0.1433, 0.0562, 0.1379], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0151, 0.0174, 0.0155, 0.0123, 0.0157, 0.0152, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:16:34,615 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=33773.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:16:37,302 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.4896, 3.4896, 2.6738, 3.2404, 2.8902, 2.0838, 3.4641, 1.8148], device='cuda:0'), covar=tensor([0.0794, 0.0545, 0.0550, 0.0396, 0.0709, 0.1931, 0.1029, 0.1478], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0121, 0.0121, 0.0101, 0.0139, 0.0176, 0.0142, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:16:46,629 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4125, 3.7221, 3.5160, 4.1036, 2.2714, 2.9292, 3.7965, 2.2181], device='cuda:0'), covar=tensor([0.0127, 0.0496, 0.0736, 0.0523, 0.1523, 0.0982, 0.0552, 0.1918], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0172, 0.0192, 0.0209, 0.0177, 0.0189, 0.0172, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:16:50,505 INFO [train.py:893] (0/4) Epoch 13, batch 1400, loss[loss=0.218, simple_loss=0.2694, pruned_loss=0.08333, over 13553.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2581, pruned_loss=0.07605, over 2665100.71 frames. ], batch size: 83, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:17:10,927 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33811.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:17:11,545 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.842e+02 3.141e+02 3.812e+02 4.224e+02 6.150e+02, threshold=7.624e+02, percent-clipped=0.0 2023-04-16 20:17:26,070 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33828.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:17:38,111 INFO [train.py:893] (0/4) Epoch 13, batch 1450, loss[loss=0.2182, simple_loss=0.2778, pruned_loss=0.07927, over 13510.00 frames. ], tot_loss[loss=0.205, simple_loss=0.2576, pruned_loss=0.0762, over 2659311.59 frames. ], batch size: 98, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:17:55,441 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=33859.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:18:24,063 INFO [train.py:893] (0/4) Epoch 13, batch 1500, loss[loss=0.206, simple_loss=0.2433, pruned_loss=0.08438, over 13380.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.258, pruned_loss=0.07647, over 2655943.14 frames. ], batch size: 62, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:18:45,822 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.120e+02 3.008e+02 3.356e+02 4.011e+02 6.551e+02, threshold=6.712e+02, percent-clipped=0.0 2023-04-16 20:19:11,236 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33939.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:19:12,633 INFO [train.py:893] (0/4) Epoch 13, batch 1550, loss[loss=0.205, simple_loss=0.2561, pruned_loss=0.077, over 13532.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2581, pruned_loss=0.07669, over 2655756.51 frames. ], batch size: 83, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:19:31,793 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4762, 4.2700, 4.5517, 4.4621, 4.7480, 4.2934, 4.7649, 4.7723], device='cuda:0'), covar=tensor([0.0431, 0.0579, 0.0682, 0.0553, 0.0610, 0.0824, 0.0521, 0.0473], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0260, 0.0259, 0.0190, 0.0371, 0.0298, 0.0229, 0.0254], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:19:58,104 INFO [train.py:893] (0/4) Epoch 13, batch 1600, loss[loss=0.1973, simple_loss=0.2548, pruned_loss=0.06988, over 13538.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.259, pruned_loss=0.07701, over 2661666.82 frames. ], batch size: 87, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:20:08,262 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-34000.pt 2023-04-16 20:20:12,447 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34000.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:20:17,455 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5176, 2.2612, 4.4533, 3.9814, 4.1940, 3.5812, 4.0871, 3.2405], device='cuda:0'), covar=tensor([0.1506, 0.1318, 0.0060, 0.0265, 0.0252, 0.0410, 0.0145, 0.0584], device='cuda:0'), in_proj_covar=tensor([0.0210, 0.0185, 0.0109, 0.0118, 0.0123, 0.0171, 0.0126, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:20:22,388 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8230, 2.4835, 1.8087, 3.6247, 4.1395, 3.2019, 4.1312, 3.8492], device='cuda:0'), covar=tensor([0.0079, 0.0911, 0.1087, 0.0096, 0.0075, 0.0459, 0.0064, 0.0081], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0083, 0.0089, 0.0070, 0.0055, 0.0075, 0.0048, 0.0065], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:20:22,931 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.987e+02 3.172e+02 3.831e+02 4.874e+02 1.162e+03, threshold=7.661e+02, percent-clipped=7.0 2023-04-16 20:20:24,109 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4156, 3.6255, 3.4520, 4.0114, 2.0440, 2.9190, 3.6186, 2.1073], device='cuda:0'), covar=tensor([0.0095, 0.0537, 0.0670, 0.0565, 0.1567, 0.0941, 0.0728, 0.1842], device='cuda:0'), in_proj_covar=tensor([0.0104, 0.0174, 0.0192, 0.0212, 0.0177, 0.0189, 0.0173, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:20:42,904 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34033.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 20:20:49,347 INFO [train.py:893] (0/4) Epoch 13, batch 1650, loss[loss=0.2016, simple_loss=0.2651, pruned_loss=0.06904, over 13461.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2587, pruned_loss=0.0765, over 2664256.30 frames. ], batch size: 103, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:20:52,960 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34044.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:20:58,218 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6236, 3.6179, 4.2331, 2.9790, 2.7592, 2.8924, 4.4961, 4.7015], device='cuda:0'), covar=tensor([0.1058, 0.1385, 0.0358, 0.1590, 0.1586, 0.1488, 0.0243, 0.0137], device='cuda:0'), in_proj_covar=tensor([0.0227, 0.0240, 0.0176, 0.0213, 0.0208, 0.0172, 0.0181, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:21:36,473 INFO [train.py:893] (0/4) Epoch 13, batch 1700, loss[loss=0.1919, simple_loss=0.2491, pruned_loss=0.06734, over 13529.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2588, pruned_loss=0.07588, over 2665510.34 frames. ], batch size: 87, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:21:40,238 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-16 20:21:50,120 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34105.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:21:55,755 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.016e+02 2.745e+02 3.237e+02 3.831e+02 8.000e+02, threshold=6.474e+02, percent-clipped=1.0 2023-04-16 20:22:12,376 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34128.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:22:22,243 INFO [train.py:893] (0/4) Epoch 13, batch 1750, loss[loss=0.2041, simple_loss=0.2538, pruned_loss=0.07713, over 13540.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2574, pruned_loss=0.07513, over 2664737.03 frames. ], batch size: 83, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:22:26,685 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34145.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:22:34,089 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-04-16 20:22:39,767 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1838, 4.7151, 4.6587, 4.6088, 4.3913, 4.5312, 5.1137, 4.5889], device='cuda:0'), covar=tensor([0.0751, 0.0995, 0.2496, 0.2498, 0.1024, 0.1517, 0.0890, 0.1212], device='cuda:0'), in_proj_covar=tensor([0.0296, 0.0323, 0.0413, 0.0422, 0.0252, 0.0318, 0.0375, 0.0296], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 20:22:52,179 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3911, 5.2124, 5.4238, 5.1029, 5.7365, 5.2045, 5.7372, 5.7230], device='cuda:0'), covar=tensor([0.0390, 0.0531, 0.0622, 0.0591, 0.0521, 0.0695, 0.0444, 0.0463], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0258, 0.0254, 0.0189, 0.0367, 0.0296, 0.0227, 0.0254], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:22:52,208 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5707, 5.0201, 4.7667, 4.7373, 4.7894, 4.6393, 5.0795, 5.0449], device='cuda:0'), covar=tensor([0.0163, 0.0162, 0.0174, 0.0274, 0.0238, 0.0211, 0.0243, 0.0198], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0173, 0.0132, 0.0156, 0.0123, 0.0170, 0.0115, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 20:22:55,619 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34176.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:23:09,232 INFO [train.py:893] (0/4) Epoch 13, batch 1800, loss[loss=0.172, simple_loss=0.2336, pruned_loss=0.05518, over 13373.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2561, pruned_loss=0.07427, over 2662624.43 frames. ], batch size: 73, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:23:17,362 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-16 20:23:24,311 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34206.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:23:30,363 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.991e+02 2.794e+02 3.346e+02 4.036e+02 7.861e+02, threshold=6.691e+02, percent-clipped=3.0 2023-04-16 20:23:55,754 INFO [train.py:893] (0/4) Epoch 13, batch 1850, loss[loss=0.2148, simple_loss=0.2644, pruned_loss=0.08256, over 13542.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2554, pruned_loss=0.07411, over 2660789.42 frames. ], batch size: 98, lr: 1.01e-02, grad_scale: 16.0 2023-04-16 20:24:00,683 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 20:24:41,997 INFO [train.py:893] (0/4) Epoch 13, batch 1900, loss[loss=0.1831, simple_loss=0.2354, pruned_loss=0.06538, over 13530.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.2552, pruned_loss=0.07431, over 2662942.59 frames. ], batch size: 72, lr: 1.00e-02, grad_scale: 16.0 2023-04-16 20:24:46,433 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34295.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:25:01,820 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.076e+02 2.981e+02 3.408e+02 4.099e+02 6.680e+02, threshold=6.816e+02, percent-clipped=0.0 2023-04-16 20:25:13,143 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2903, 3.2255, 3.8424, 2.8300, 2.5006, 2.7206, 4.0639, 4.1981], device='cuda:0'), covar=tensor([0.1141, 0.1612, 0.0369, 0.1585, 0.1640, 0.1456, 0.0249, 0.0186], device='cuda:0'), in_proj_covar=tensor([0.0227, 0.0242, 0.0176, 0.0213, 0.0208, 0.0172, 0.0180, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:25:20,519 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34333.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:25:28,205 INFO [train.py:893] (0/4) Epoch 13, batch 1950, loss[loss=0.2445, simple_loss=0.2982, pruned_loss=0.09537, over 13456.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2557, pruned_loss=0.07457, over 2664753.74 frames. ], batch size: 103, lr: 1.00e-02, grad_scale: 16.0 2023-04-16 20:25:47,187 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8490, 4.1378, 2.5634, 3.9377, 4.0064, 2.5342, 3.6401, 2.6091], device='cuda:0'), covar=tensor([0.0273, 0.0244, 0.1215, 0.0316, 0.0214, 0.1235, 0.0489, 0.1456], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0153, 0.0174, 0.0158, 0.0125, 0.0159, 0.0153, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:25:57,941 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6966, 2.3925, 2.2500, 2.7614, 2.0534, 2.8569, 2.8849, 2.3036], device='cuda:0'), covar=tensor([0.0084, 0.0155, 0.0135, 0.0100, 0.0182, 0.0085, 0.0117, 0.0158], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0088, 0.0097, 0.0088, 0.0103, 0.0080, 0.0085, 0.0083], device='cuda:0'), out_proj_covar=tensor([8.7459e-05, 9.7171e-05, 1.1064e-04, 9.8024e-05, 1.1521e-04, 8.6403e-05, 9.4398e-05, 9.0910e-05], device='cuda:0') 2023-04-16 20:26:06,016 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34381.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:26:11,071 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5641, 3.0636, 2.6738, 4.3899, 5.0134, 3.9338, 4.8315, 4.5792], device='cuda:0'), covar=tensor([0.0121, 0.0737, 0.0815, 0.0101, 0.0055, 0.0297, 0.0088, 0.0079], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0082, 0.0089, 0.0070, 0.0055, 0.0074, 0.0048, 0.0065], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:26:13,277 INFO [train.py:893] (0/4) Epoch 13, batch 2000, loss[loss=0.2386, simple_loss=0.2737, pruned_loss=0.1017, over 11981.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2577, pruned_loss=0.07572, over 2663730.45 frames. ], batch size: 157, lr: 1.00e-02, grad_scale: 32.0 2023-04-16 20:26:20,700 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 20:26:24,532 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34400.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:26:35,285 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.214e+02 3.106e+02 3.552e+02 4.151e+02 6.678e+02, threshold=7.105e+02, percent-clipped=0.0 2023-04-16 20:27:02,003 INFO [train.py:893] (0/4) Epoch 13, batch 2050, loss[loss=0.1925, simple_loss=0.2524, pruned_loss=0.06633, over 13365.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2592, pruned_loss=0.07652, over 2661822.34 frames. ], batch size: 73, lr: 1.00e-02, grad_scale: 32.0 2023-04-16 20:27:47,841 INFO [train.py:893] (0/4) Epoch 13, batch 2100, loss[loss=0.1824, simple_loss=0.245, pruned_loss=0.0599, over 13560.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2574, pruned_loss=0.07556, over 2661044.03 frames. ], batch size: 76, lr: 1.00e-02, grad_scale: 32.0 2023-04-16 20:27:59,434 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34501.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:28:08,470 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.948e+02 2.768e+02 3.313e+02 3.820e+02 1.138e+03, threshold=6.626e+02, percent-clipped=1.0 2023-04-16 20:28:35,140 INFO [train.py:893] (0/4) Epoch 13, batch 2150, loss[loss=0.2129, simple_loss=0.2645, pruned_loss=0.08067, over 13535.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2567, pruned_loss=0.07488, over 2662055.34 frames. ], batch size: 76, lr: 1.00e-02, grad_scale: 32.0 2023-04-16 20:29:00,579 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.86 vs. limit=5.0 2023-04-16 20:29:22,364 INFO [train.py:893] (0/4) Epoch 13, batch 2200, loss[loss=0.2277, simple_loss=0.2747, pruned_loss=0.09039, over 13526.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2562, pruned_loss=0.0743, over 2662057.32 frames. ], batch size: 85, lr: 1.00e-02, grad_scale: 32.0 2023-04-16 20:29:25,173 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34593.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:29:26,869 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34595.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:29:29,592 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-16 20:29:41,971 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.890e+02 2.826e+02 3.177e+02 3.694e+02 6.051e+02, threshold=6.354e+02, percent-clipped=0.0 2023-04-16 20:29:52,922 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34622.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:29:56,263 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34626.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:30:08,580 INFO [train.py:893] (0/4) Epoch 13, batch 2250, loss[loss=0.1766, simple_loss=0.2254, pruned_loss=0.06392, over 13230.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2542, pruned_loss=0.07377, over 2661413.88 frames. ], batch size: 58, lr: 1.00e-02, grad_scale: 32.0 2023-04-16 20:30:11,253 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34643.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:30:23,110 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34654.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:30:36,557 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9590, 4.4232, 4.1991, 4.1955, 4.2466, 4.0488, 4.4466, 4.4929], device='cuda:0'), covar=tensor([0.0254, 0.0248, 0.0263, 0.0393, 0.0320, 0.0329, 0.0322, 0.0226], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0172, 0.0132, 0.0155, 0.0123, 0.0168, 0.0114, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 20:30:49,915 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34683.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 20:30:54,070 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34687.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:30:56,312 INFO [train.py:893] (0/4) Epoch 13, batch 2300, loss[loss=0.1992, simple_loss=0.2595, pruned_loss=0.06949, over 13395.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2538, pruned_loss=0.07324, over 2660393.06 frames. ], batch size: 109, lr: 9.99e-03, grad_scale: 16.0 2023-04-16 20:31:05,743 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34700.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:31:14,709 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34710.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 20:31:17,616 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.020e+02 2.763e+02 3.206e+02 3.873e+02 6.964e+02, threshold=6.413e+02, percent-clipped=1.0 2023-04-16 20:31:25,560 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-16 20:31:42,628 INFO [train.py:893] (0/4) Epoch 13, batch 2350, loss[loss=0.1542, simple_loss=0.2134, pruned_loss=0.04748, over 13523.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2532, pruned_loss=0.07269, over 2657616.08 frames. ], batch size: 76, lr: 9.99e-03, grad_scale: 16.0 2023-04-16 20:31:50,106 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34748.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:32:03,525 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6180, 3.9690, 3.8189, 4.2919, 2.4424, 3.2320, 4.1497, 2.1599], device='cuda:0'), covar=tensor([0.0094, 0.0470, 0.0679, 0.0504, 0.1444, 0.0888, 0.0435, 0.1920], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0173, 0.0192, 0.0210, 0.0176, 0.0189, 0.0172, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:32:06,575 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 20:32:11,852 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34771.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 20:32:15,822 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1886, 2.6767, 2.0272, 4.1561, 4.6829, 3.4470, 4.4973, 4.2732], device='cuda:0'), covar=tensor([0.0111, 0.0839, 0.1055, 0.0098, 0.0061, 0.0459, 0.0095, 0.0085], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0083, 0.0090, 0.0071, 0.0056, 0.0075, 0.0049, 0.0066], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:32:29,622 INFO [train.py:893] (0/4) Epoch 13, batch 2400, loss[loss=0.1818, simple_loss=0.2383, pruned_loss=0.06264, over 13520.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2531, pruned_loss=0.07292, over 2660923.78 frames. ], batch size: 70, lr: 9.98e-03, grad_scale: 16.0 2023-04-16 20:32:39,428 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34801.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:32:51,404 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.128e+02 2.703e+02 3.191e+02 3.711e+02 8.271e+02, threshold=6.381e+02, percent-clipped=2.0 2023-04-16 20:33:08,708 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.72 vs. limit=5.0 2023-04-16 20:33:10,129 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9543, 2.5366, 1.9770, 3.7674, 4.3028, 3.1929, 4.1373, 3.9352], device='cuda:0'), covar=tensor([0.0090, 0.0804, 0.0959, 0.0096, 0.0067, 0.0446, 0.0109, 0.0082], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0083, 0.0090, 0.0071, 0.0056, 0.0075, 0.0049, 0.0066], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:33:16,439 INFO [train.py:893] (0/4) Epoch 13, batch 2450, loss[loss=0.1874, simple_loss=0.2433, pruned_loss=0.06576, over 13512.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2536, pruned_loss=0.07314, over 2662399.01 frames. ], batch size: 70, lr: 9.97e-03, grad_scale: 16.0 2023-04-16 20:33:24,814 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34849.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:33:36,562 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-16 20:34:02,366 INFO [train.py:893] (0/4) Epoch 13, batch 2500, loss[loss=0.1993, simple_loss=0.2532, pruned_loss=0.07265, over 13015.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2538, pruned_loss=0.07298, over 2663741.41 frames. ], batch size: 142, lr: 9.96e-03, grad_scale: 16.0 2023-04-16 20:34:24,102 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.999e+02 2.919e+02 3.279e+02 3.793e+02 5.690e+02, threshold=6.558e+02, percent-clipped=0.0 2023-04-16 20:34:37,704 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8632, 4.3290, 4.0901, 4.1424, 4.1572, 3.9948, 4.4033, 4.4176], device='cuda:0'), covar=tensor([0.0219, 0.0213, 0.0206, 0.0317, 0.0280, 0.0298, 0.0248, 0.0254], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0177, 0.0137, 0.0161, 0.0127, 0.0175, 0.0118, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 20:34:45,250 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3871, 4.5142, 3.0907, 4.3361, 4.3953, 2.7789, 4.0449, 3.0636], device='cuda:0'), covar=tensor([0.0229, 0.0226, 0.1103, 0.0379, 0.0169, 0.1213, 0.0398, 0.1508], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0153, 0.0171, 0.0157, 0.0125, 0.0157, 0.0151, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:34:49,063 INFO [train.py:893] (0/4) Epoch 13, batch 2550, loss[loss=0.2119, simple_loss=0.2633, pruned_loss=0.08026, over 13525.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2543, pruned_loss=0.07353, over 2664220.46 frames. ], batch size: 91, lr: 9.96e-03, grad_scale: 16.0 2023-04-16 20:34:58,763 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34949.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:35:15,573 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-16 20:35:15,870 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 20:35:25,376 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5772, 3.6577, 2.9318, 3.4280, 2.9365, 1.9150, 3.6254, 1.8653], device='cuda:0'), covar=tensor([0.0735, 0.0414, 0.0409, 0.0322, 0.0684, 0.1951, 0.0817, 0.1347], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0122, 0.0121, 0.0102, 0.0138, 0.0176, 0.0144, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:35:26,051 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34978.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 20:35:29,339 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34982.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:35:35,931 INFO [train.py:893] (0/4) Epoch 13, batch 2600, loss[loss=0.2072, simple_loss=0.25, pruned_loss=0.08217, over 13177.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2544, pruned_loss=0.07419, over 2663747.66 frames. ], batch size: 58, lr: 9.95e-03, grad_scale: 16.0 2023-04-16 20:35:43,726 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6585, 3.7257, 4.3711, 3.1780, 2.8889, 3.1213, 4.5024, 4.7069], device='cuda:0'), covar=tensor([0.1026, 0.1260, 0.0319, 0.1358, 0.1342, 0.1156, 0.0206, 0.0164], device='cuda:0'), in_proj_covar=tensor([0.0223, 0.0241, 0.0174, 0.0209, 0.0204, 0.0170, 0.0176, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:35:58,856 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.271e+02 3.103e+02 3.615e+02 4.155e+02 9.868e+02, threshold=7.230e+02, percent-clipped=6.0 2023-04-16 20:36:19,373 INFO [train.py:893] (0/4) Epoch 13, batch 2650, loss[loss=0.1942, simple_loss=0.2438, pruned_loss=0.07228, over 13422.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2545, pruned_loss=0.07473, over 2656487.34 frames. ], batch size: 88, lr: 9.94e-03, grad_scale: 16.0 2023-04-16 20:36:40,112 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35066.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 20:36:58,973 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-13.pt 2023-04-16 20:37:24,656 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 20:37:28,635 INFO [train.py:893] (0/4) Epoch 14, batch 0, loss[loss=0.2197, simple_loss=0.2673, pruned_loss=0.08604, over 13466.00 frames. ], tot_loss[loss=0.2197, simple_loss=0.2673, pruned_loss=0.08604, over 13466.00 frames. ], batch size: 103, lr: 9.58e-03, grad_scale: 16.0 2023-04-16 20:37:28,635 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 20:37:37,100 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9176, 3.7156, 3.9516, 2.6346, 4.2114, 3.9306, 3.8963, 4.1166], device='cuda:0'), covar=tensor([0.0246, 0.0147, 0.0148, 0.0934, 0.0115, 0.0221, 0.0132, 0.0103], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0046, 0.0069, 0.0094, 0.0085, 0.0088, 0.0069, 0.0062], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 20:37:38,138 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.6091, 5.0997, 5.1612, 5.1594, 5.0073, 5.1459, 5.6068, 5.1627], device='cuda:0'), covar=tensor([0.0810, 0.1190, 0.2255, 0.2577, 0.0928, 0.1406, 0.0930, 0.1059], device='cuda:0'), in_proj_covar=tensor([0.0297, 0.0330, 0.0418, 0.0427, 0.0253, 0.0319, 0.0378, 0.0303], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 20:37:39,155 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.7846, 5.2556, 5.4147, 5.4048, 5.2427, 5.2698, 5.7880, 5.3288], device='cuda:0'), covar=tensor([0.0695, 0.1008, 0.1812, 0.2098, 0.0714, 0.1402, 0.0778, 0.0884], device='cuda:0'), in_proj_covar=tensor([0.0297, 0.0330, 0.0418, 0.0427, 0.0253, 0.0319, 0.0378, 0.0303], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 20:37:50,926 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6867, 4.4506, 4.6564, 3.3796, 5.0760, 4.6920, 4.8234, 4.9233], device='cuda:0'), covar=tensor([0.0193, 0.0161, 0.0096, 0.0609, 0.0075, 0.0175, 0.0073, 0.0085], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0046, 0.0069, 0.0094, 0.0085, 0.0088, 0.0069, 0.0062], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 20:37:51,600 INFO [train.py:927] (0/4) Epoch 14, validation: loss=0.1497, simple_loss=0.207, pruned_loss=0.04622, over 2446609.00 frames. 2023-04-16 20:37:51,601 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 20:37:54,486 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35092.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:38:13,109 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.840e+02 2.859e+02 3.403e+02 4.075e+02 8.833e+02, threshold=6.807e+02, percent-clipped=1.0 2023-04-16 20:38:37,369 INFO [train.py:893] (0/4) Epoch 14, batch 50, loss[loss=0.2178, simple_loss=0.2621, pruned_loss=0.08673, over 11793.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2439, pruned_loss=0.0706, over 602976.57 frames. ], batch size: 157, lr: 9.57e-03, grad_scale: 16.0 2023-04-16 20:38:50,829 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35153.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:39:01,990 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 20:39:01,990 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 20:39:01,990 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 20:39:01,998 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 20:39:02,014 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 20:39:02,720 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 20:39:02,742 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 20:39:24,428 INFO [train.py:893] (0/4) Epoch 14, batch 100, loss[loss=0.1982, simple_loss=0.2441, pruned_loss=0.07618, over 13514.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2496, pruned_loss=0.07399, over 1061461.94 frames. ], batch size: 70, lr: 9.56e-03, grad_scale: 16.0 2023-04-16 20:39:27,224 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9283, 2.3230, 1.7744, 3.7566, 4.1717, 3.1148, 4.0853, 3.8958], device='cuda:0'), covar=tensor([0.0069, 0.0902, 0.0992, 0.0077, 0.0045, 0.0415, 0.0079, 0.0063], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0084, 0.0090, 0.0072, 0.0056, 0.0075, 0.0050, 0.0066], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:39:46,296 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.998e+02 3.037e+02 3.469e+02 4.318e+02 9.058e+02, threshold=6.938e+02, percent-clipped=1.0 2023-04-16 20:40:11,166 INFO [train.py:893] (0/4) Epoch 14, batch 150, loss[loss=0.2122, simple_loss=0.2648, pruned_loss=0.07986, over 13519.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2529, pruned_loss=0.07559, over 1403915.63 frames. ], batch size: 98, lr: 9.56e-03, grad_scale: 16.0 2023-04-16 20:40:21,460 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35249.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:40:29,607 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9325, 4.0535, 2.7834, 3.8396, 3.9072, 2.5500, 3.5038, 2.6374], device='cuda:0'), covar=tensor([0.0235, 0.0211, 0.1120, 0.0261, 0.0177, 0.1162, 0.0575, 0.1361], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0152, 0.0173, 0.0158, 0.0126, 0.0157, 0.0153, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:40:47,750 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35278.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 20:40:51,076 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35282.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:40:51,280 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-16 20:40:58,491 INFO [train.py:893] (0/4) Epoch 14, batch 200, loss[loss=0.2192, simple_loss=0.276, pruned_loss=0.08116, over 13433.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2558, pruned_loss=0.07701, over 1674935.95 frames. ], batch size: 106, lr: 9.55e-03, grad_scale: 16.0 2023-04-16 20:41:05,500 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35297.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:41:20,379 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.003e+02 2.883e+02 3.259e+02 4.058e+02 6.702e+02, threshold=6.518e+02, percent-clipped=0.0 2023-04-16 20:41:21,540 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.5183, 5.0400, 4.8794, 4.9608, 4.8925, 4.8876, 5.5042, 5.0235], device='cuda:0'), covar=tensor([0.0767, 0.0985, 0.2263, 0.2997, 0.0830, 0.1582, 0.0894, 0.1024], device='cuda:0'), in_proj_covar=tensor([0.0295, 0.0330, 0.0421, 0.0429, 0.0251, 0.0321, 0.0379, 0.0303], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 20:41:33,089 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35326.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:41:36,467 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35330.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:41:44,610 INFO [train.py:893] (0/4) Epoch 14, batch 250, loss[loss=0.2158, simple_loss=0.2705, pruned_loss=0.08059, over 13367.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2559, pruned_loss=0.07677, over 1895119.40 frames. ], batch size: 109, lr: 9.54e-03, grad_scale: 16.0 2023-04-16 20:41:51,469 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35345.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 20:42:10,625 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35366.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 20:42:18,089 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35374.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:42:31,805 INFO [train.py:893] (0/4) Epoch 14, batch 300, loss[loss=0.245, simple_loss=0.2906, pruned_loss=0.09968, over 13542.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2573, pruned_loss=0.07681, over 2065959.22 frames. ], batch size: 72, lr: 9.54e-03, grad_scale: 16.0 2023-04-16 20:42:46,020 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1058, 3.0054, 3.6052, 2.6756, 2.3946, 2.5519, 3.8661, 4.0271], device='cuda:0'), covar=tensor([0.1143, 0.1622, 0.0369, 0.1402, 0.1600, 0.1361, 0.0295, 0.0151], device='cuda:0'), in_proj_covar=tensor([0.0228, 0.0247, 0.0179, 0.0214, 0.0209, 0.0173, 0.0179, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:42:48,421 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35406.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 20:42:54,699 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.158e+02 3.002e+02 3.392e+02 4.204e+02 7.144e+02, threshold=6.784e+02, percent-clipped=2.0 2023-04-16 20:42:56,406 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35414.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 20:43:05,866 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9821, 4.0205, 3.0698, 3.6369, 3.1249, 2.1021, 3.8674, 1.9507], device='cuda:0'), covar=tensor([0.0793, 0.0278, 0.0547, 0.0304, 0.0739, 0.2093, 0.1023, 0.1496], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0122, 0.0121, 0.0102, 0.0138, 0.0174, 0.0144, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:43:15,097 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35435.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:43:19,559 INFO [train.py:893] (0/4) Epoch 14, batch 350, loss[loss=0.1871, simple_loss=0.2296, pruned_loss=0.07227, over 13379.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2583, pruned_loss=0.07733, over 2197696.46 frames. ], batch size: 62, lr: 9.53e-03, grad_scale: 16.0 2023-04-16 20:43:27,869 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35448.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:43:32,245 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1822, 4.3283, 3.4727, 2.9212, 3.1174, 2.5379, 4.4826, 2.5384], device='cuda:0'), covar=tensor([0.1463, 0.0298, 0.0857, 0.1728, 0.0668, 0.2934, 0.0170, 0.3301], device='cuda:0'), in_proj_covar=tensor([0.0299, 0.0278, 0.0282, 0.0299, 0.0236, 0.0301, 0.0195, 0.0346], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 20:44:06,003 INFO [train.py:893] (0/4) Epoch 14, batch 400, loss[loss=0.1772, simple_loss=0.2312, pruned_loss=0.06162, over 13369.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2593, pruned_loss=0.07747, over 2300917.46 frames. ], batch size: 67, lr: 9.52e-03, grad_scale: 16.0 2023-04-16 20:44:28,070 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.910e+02 3.027e+02 3.448e+02 3.927e+02 6.855e+02, threshold=6.896e+02, percent-clipped=1.0 2023-04-16 20:44:41,799 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-16 20:44:43,218 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35530.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:44:52,039 INFO [train.py:893] (0/4) Epoch 14, batch 450, loss[loss=0.1662, simple_loss=0.2152, pruned_loss=0.05859, over 10412.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2607, pruned_loss=0.07827, over 2381473.76 frames. ], batch size: 42, lr: 9.52e-03, grad_scale: 16.0 2023-04-16 20:45:04,022 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8754, 2.4011, 2.4137, 2.8534, 2.2455, 2.9908, 2.8143, 2.4493], device='cuda:0'), covar=tensor([0.0069, 0.0182, 0.0145, 0.0127, 0.0197, 0.0094, 0.0189, 0.0172], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0091, 0.0099, 0.0092, 0.0107, 0.0082, 0.0088, 0.0086], device='cuda:0'), out_proj_covar=tensor([9.0107e-05, 1.0088e-04, 1.1216e-04, 1.0209e-04, 1.1964e-04, 8.9036e-05, 9.8305e-05, 9.3274e-05], device='cuda:0') 2023-04-16 20:45:09,028 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0165, 1.8412, 2.1115, 3.3243, 3.0945, 3.3471, 2.6541, 2.2151], device='cuda:0'), covar=tensor([0.0223, 0.1042, 0.0821, 0.0060, 0.0238, 0.0062, 0.0584, 0.0822], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0147, 0.0156, 0.0082, 0.0109, 0.0080, 0.0159, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:45:13,281 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-04-16 20:45:16,064 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 20:45:19,623 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4720, 3.3996, 4.0503, 2.8385, 2.7821, 2.8224, 4.3079, 4.4948], device='cuda:0'), covar=tensor([0.1122, 0.1537, 0.0308, 0.1617, 0.1435, 0.1252, 0.0243, 0.0225], device='cuda:0'), in_proj_covar=tensor([0.0227, 0.0245, 0.0177, 0.0213, 0.0209, 0.0171, 0.0179, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:45:38,276 INFO [train.py:893] (0/4) Epoch 14, batch 500, loss[loss=0.2007, simple_loss=0.249, pruned_loss=0.07621, over 13373.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2597, pruned_loss=0.07715, over 2445385.62 frames. ], batch size: 62, lr: 9.51e-03, grad_scale: 16.0 2023-04-16 20:45:40,163 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35591.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:45:41,282 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.64 vs. limit=5.0 2023-04-16 20:46:01,679 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.031e+02 2.935e+02 3.425e+02 4.284e+02 7.521e+02, threshold=6.851e+02, percent-clipped=2.0 2023-04-16 20:46:26,135 INFO [train.py:893] (0/4) Epoch 14, batch 550, loss[loss=0.1993, simple_loss=0.2497, pruned_loss=0.07443, over 13352.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2598, pruned_loss=0.07716, over 2489883.26 frames. ], batch size: 118, lr: 9.50e-03, grad_scale: 16.0 2023-04-16 20:46:33,410 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2023-04-16 20:47:12,174 INFO [train.py:893] (0/4) Epoch 14, batch 600, loss[loss=0.188, simple_loss=0.2465, pruned_loss=0.06474, over 13552.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2585, pruned_loss=0.07647, over 2534173.70 frames. ], batch size: 89, lr: 9.50e-03, grad_scale: 16.0 2023-04-16 20:47:24,544 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35701.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 20:47:32,515 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-16 20:47:35,222 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.988e+02 2.727e+02 3.202e+02 3.710e+02 6.936e+02, threshold=6.404e+02, percent-clipped=1.0 2023-04-16 20:47:51,290 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35730.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:48:00,086 INFO [train.py:893] (0/4) Epoch 14, batch 650, loss[loss=0.1758, simple_loss=0.2328, pruned_loss=0.05938, over 13513.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2569, pruned_loss=0.07584, over 2562038.08 frames. ], batch size: 76, lr: 9.49e-03, grad_scale: 16.0 2023-04-16 20:48:06,232 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8828, 2.4193, 2.4619, 2.8305, 2.1822, 3.0459, 2.7720, 2.5060], device='cuda:0'), covar=tensor([0.0069, 0.0158, 0.0146, 0.0143, 0.0207, 0.0087, 0.0168, 0.0144], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0089, 0.0096, 0.0090, 0.0104, 0.0080, 0.0086, 0.0084], device='cuda:0'), out_proj_covar=tensor([8.6486e-05, 9.8591e-05, 1.0868e-04, 9.8998e-05, 1.1662e-04, 8.6896e-05, 9.5294e-05, 9.1497e-05], device='cuda:0') 2023-04-16 20:48:07,880 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35748.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:48:30,033 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35771.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:48:46,710 INFO [train.py:893] (0/4) Epoch 14, batch 700, loss[loss=0.1968, simple_loss=0.257, pruned_loss=0.06831, over 13458.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2552, pruned_loss=0.07441, over 2580443.77 frames. ], batch size: 100, lr: 9.48e-03, grad_scale: 16.0 2023-04-16 20:48:53,366 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35796.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:48:58,008 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4881, 4.8635, 4.5865, 4.5328, 4.6202, 4.9557, 4.8211, 4.6548], device='cuda:0'), covar=tensor([0.0306, 0.0235, 0.0316, 0.0946, 0.0239, 0.0225, 0.0237, 0.0178], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0136, 0.0151, 0.0247, 0.0151, 0.0166, 0.0148, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2023-04-16 20:49:07,406 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35811.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:49:08,663 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.910e+02 3.025e+02 3.522e+02 4.201e+02 9.258e+02, threshold=7.044e+02, percent-clipped=2.0 2023-04-16 20:49:27,478 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35832.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:49:33,923 INFO [train.py:893] (0/4) Epoch 14, batch 750, loss[loss=0.204, simple_loss=0.2633, pruned_loss=0.07231, over 13450.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2555, pruned_loss=0.07483, over 2598897.46 frames. ], batch size: 106, lr: 9.48e-03, grad_scale: 16.0 2023-04-16 20:50:04,830 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35872.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:50:05,670 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4862, 3.5235, 4.0453, 2.7362, 2.5247, 2.6807, 4.3080, 4.4703], device='cuda:0'), covar=tensor([0.1074, 0.1439, 0.0336, 0.1663, 0.1696, 0.1472, 0.0220, 0.0199], device='cuda:0'), in_proj_covar=tensor([0.0229, 0.0245, 0.0179, 0.0212, 0.0209, 0.0174, 0.0181, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 20:50:17,044 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35886.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:50:20,144 INFO [train.py:893] (0/4) Epoch 14, batch 800, loss[loss=0.1866, simple_loss=0.2458, pruned_loss=0.0637, over 13509.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2566, pruned_loss=0.0755, over 2610649.28 frames. ], batch size: 81, lr: 9.47e-03, grad_scale: 16.0 2023-04-16 20:50:37,859 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2023-04-16 20:50:41,654 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.031e+02 2.971e+02 3.594e+02 4.355e+02 9.749e+02, threshold=7.188e+02, percent-clipped=3.0 2023-04-16 20:51:04,917 INFO [train.py:893] (0/4) Epoch 14, batch 850, loss[loss=0.2149, simple_loss=0.267, pruned_loss=0.08142, over 13488.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2575, pruned_loss=0.07606, over 2622600.20 frames. ], batch size: 93, lr: 9.46e-03, grad_scale: 16.0 2023-04-16 20:51:46,023 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.7375, 4.9239, 3.7777, 4.6344, 4.7186, 3.7975, 4.3216, 3.5851], device='cuda:0'), covar=tensor([0.0192, 0.0185, 0.0707, 0.0427, 0.0118, 0.0711, 0.0249, 0.1001], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0154, 0.0170, 0.0161, 0.0127, 0.0157, 0.0150, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 20:51:51,613 INFO [train.py:893] (0/4) Epoch 14, batch 900, loss[loss=0.201, simple_loss=0.2525, pruned_loss=0.07477, over 13488.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2572, pruned_loss=0.07583, over 2630793.58 frames. ], batch size: 70, lr: 9.46e-03, grad_scale: 8.0 2023-04-16 20:52:01,959 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-36000.pt 2023-04-16 20:52:06,879 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36001.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 20:52:19,158 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.226e+02 2.907e+02 3.358e+02 4.155e+02 9.448e+02, threshold=6.715e+02, percent-clipped=1.0 2023-04-16 20:52:25,858 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 20:52:33,440 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36030.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:52:42,262 INFO [train.py:893] (0/4) Epoch 14, batch 950, loss[loss=0.194, simple_loss=0.2446, pruned_loss=0.07176, over 13476.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2567, pruned_loss=0.07589, over 2638988.24 frames. ], batch size: 70, lr: 9.45e-03, grad_scale: 8.0 2023-04-16 20:52:51,398 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36049.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 20:53:18,985 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36078.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:53:28,947 INFO [train.py:893] (0/4) Epoch 14, batch 1000, loss[loss=0.2056, simple_loss=0.2416, pruned_loss=0.08485, over 12572.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2545, pruned_loss=0.07489, over 2644110.70 frames. ], batch size: 51, lr: 9.44e-03, grad_scale: 8.0 2023-04-16 20:53:52,249 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.888e+02 2.953e+02 3.397e+02 4.421e+02 7.418e+02, threshold=6.794e+02, percent-clipped=2.0 2023-04-16 20:54:04,125 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36127.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:54:15,574 INFO [train.py:893] (0/4) Epoch 14, batch 1050, loss[loss=0.1704, simple_loss=0.2368, pruned_loss=0.05198, over 13044.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2526, pruned_loss=0.07339, over 2647097.68 frames. ], batch size: 142, lr: 9.44e-03, grad_scale: 8.0 2023-04-16 20:54:42,527 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36167.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:54:55,730 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36181.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:54:59,892 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36186.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:55:02,042 INFO [train.py:893] (0/4) Epoch 14, batch 1100, loss[loss=0.1951, simple_loss=0.2382, pruned_loss=0.07604, over 13187.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2533, pruned_loss=0.07318, over 2649458.75 frames. ], batch size: 58, lr: 9.43e-03, grad_scale: 8.0 2023-04-16 20:55:25,259 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-04-16 20:55:26,408 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.646e+02 2.611e+02 3.323e+02 3.965e+02 5.525e+02, threshold=6.646e+02, percent-clipped=0.0 2023-04-16 20:55:45,658 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36234.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:55:49,717 INFO [train.py:893] (0/4) Epoch 14, batch 1150, loss[loss=0.1957, simple_loss=0.2461, pruned_loss=0.07266, over 13398.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2531, pruned_loss=0.07262, over 2653966.45 frames. ], batch size: 65, lr: 9.43e-03, grad_scale: 8.0 2023-04-16 20:55:53,429 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36242.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:56:36,375 INFO [train.py:893] (0/4) Epoch 14, batch 1200, loss[loss=0.2043, simple_loss=0.2559, pruned_loss=0.07633, over 13524.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2532, pruned_loss=0.07204, over 2655223.54 frames. ], batch size: 85, lr: 9.42e-03, grad_scale: 8.0 2023-04-16 20:56:59,636 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.943e+02 3.077e+02 3.534e+02 4.288e+02 1.109e+03, threshold=7.069e+02, percent-clipped=3.0 2023-04-16 20:57:01,425 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 20:57:14,248 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 20:57:23,281 INFO [train.py:893] (0/4) Epoch 14, batch 1250, loss[loss=0.1948, simple_loss=0.251, pruned_loss=0.06924, over 13502.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2543, pruned_loss=0.07297, over 2657921.11 frames. ], batch size: 81, lr: 9.41e-03, grad_scale: 8.0 2023-04-16 20:57:59,834 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-16 20:58:10,727 INFO [train.py:893] (0/4) Epoch 14, batch 1300, loss[loss=0.2162, simple_loss=0.2659, pruned_loss=0.08328, over 13529.00 frames. ], tot_loss[loss=0.2013, simple_loss=0.2555, pruned_loss=0.07355, over 2658822.14 frames. ], batch size: 83, lr: 9.41e-03, grad_scale: 8.0 2023-04-16 20:58:34,165 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.990e+02 3.063e+02 3.435e+02 4.051e+02 8.324e+02, threshold=6.871e+02, percent-clipped=1.0 2023-04-16 20:58:47,724 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36427.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:58:57,485 INFO [train.py:893] (0/4) Epoch 14, batch 1350, loss[loss=0.1742, simple_loss=0.2306, pruned_loss=0.05895, over 11957.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2551, pruned_loss=0.07314, over 2658681.76 frames. ], batch size: 157, lr: 9.40e-03, grad_scale: 8.0 2023-04-16 20:59:02,315 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-16 20:59:11,319 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-04-16 20:59:24,286 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36467.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:59:30,916 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36475.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 20:59:45,705 INFO [train.py:893] (0/4) Epoch 14, batch 1400, loss[loss=0.2127, simple_loss=0.263, pruned_loss=0.08118, over 13517.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2543, pruned_loss=0.07301, over 2662484.67 frames. ], batch size: 98, lr: 9.39e-03, grad_scale: 8.0 2023-04-16 21:00:09,052 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.745e+02 3.192e+02 3.647e+02 6.550e+02, threshold=6.384e+02, percent-clipped=0.0 2023-04-16 21:00:10,035 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36515.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:00:29,139 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36537.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:00:30,575 INFO [train.py:893] (0/4) Epoch 14, batch 1450, loss[loss=0.1897, simple_loss=0.2469, pruned_loss=0.06631, over 13348.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2546, pruned_loss=0.07342, over 2665052.23 frames. ], batch size: 73, lr: 9.39e-03, grad_scale: 8.0 2023-04-16 21:01:17,371 INFO [train.py:893] (0/4) Epoch 14, batch 1500, loss[loss=0.1842, simple_loss=0.234, pruned_loss=0.06719, over 13424.00 frames. ], tot_loss[loss=0.2, simple_loss=0.254, pruned_loss=0.07299, over 2660526.52 frames. ], batch size: 65, lr: 9.38e-03, grad_scale: 8.0 2023-04-16 21:01:32,414 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36605.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:01:35,827 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36608.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:01:39,137 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36612.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:01:40,470 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.986e+02 2.902e+02 3.279e+02 4.029e+02 7.502e+02, threshold=6.558e+02, percent-clipped=1.0 2023-04-16 21:02:04,663 INFO [train.py:893] (0/4) Epoch 14, batch 1550, loss[loss=0.1623, simple_loss=0.223, pruned_loss=0.05082, over 13369.00 frames. ], tot_loss[loss=0.2013, simple_loss=0.255, pruned_loss=0.07382, over 2656115.12 frames. ], batch size: 73, lr: 9.37e-03, grad_scale: 8.0 2023-04-16 21:02:28,506 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36666.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:02:33,404 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36669.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:02:36,837 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36673.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:02:39,395 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-04-16 21:02:47,934 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2023-04-16 21:02:50,546 INFO [train.py:893] (0/4) Epoch 14, batch 1600, loss[loss=0.2082, simple_loss=0.2626, pruned_loss=0.07691, over 13442.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2543, pruned_loss=0.07361, over 2646290.65 frames. ], batch size: 95, lr: 9.37e-03, grad_scale: 8.0 2023-04-16 21:03:01,623 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36700.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 21:03:14,049 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 2.981e+02 3.490e+02 4.331e+02 9.127e+02, threshold=6.980e+02, percent-clipped=1.0 2023-04-16 21:03:37,810 INFO [train.py:893] (0/4) Epoch 14, batch 1650, loss[loss=0.1994, simple_loss=0.2585, pruned_loss=0.07012, over 13459.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2552, pruned_loss=0.07352, over 2645335.58 frames. ], batch size: 106, lr: 9.36e-03, grad_scale: 8.0 2023-04-16 21:03:38,990 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5483, 2.6384, 2.8133, 4.3495, 3.9427, 4.3493, 3.3362, 2.5984], device='cuda:0'), covar=tensor([0.0314, 0.0941, 0.0800, 0.0035, 0.0173, 0.0045, 0.0629, 0.0968], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0145, 0.0157, 0.0083, 0.0110, 0.0079, 0.0162, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:03:40,698 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4459, 4.1191, 4.2465, 2.7774, 4.7707, 4.4729, 4.4642, 4.7103], device='cuda:0'), covar=tensor([0.0213, 0.0119, 0.0153, 0.1020, 0.0140, 0.0208, 0.0126, 0.0093], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0045, 0.0070, 0.0094, 0.0087, 0.0089, 0.0069, 0.0062], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 21:03:44,742 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1462, 4.6730, 4.5091, 4.5755, 4.3287, 4.4649, 5.0886, 4.6947], device='cuda:0'), covar=tensor([0.0722, 0.1178, 0.2307, 0.2671, 0.0993, 0.1717, 0.0953, 0.1081], device='cuda:0'), in_proj_covar=tensor([0.0296, 0.0335, 0.0429, 0.0434, 0.0256, 0.0319, 0.0382, 0.0305], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 21:03:56,858 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0403, 2.9349, 3.4881, 2.6732, 2.4561, 2.5440, 3.7262, 3.7711], device='cuda:0'), covar=tensor([0.1163, 0.1651, 0.0386, 0.1314, 0.1460, 0.1336, 0.0295, 0.0257], device='cuda:0'), in_proj_covar=tensor([0.0230, 0.0248, 0.0181, 0.0214, 0.0211, 0.0175, 0.0183, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:03:57,591 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36761.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 21:04:22,443 INFO [train.py:893] (0/4) Epoch 14, batch 1700, loss[loss=0.2035, simple_loss=0.2574, pruned_loss=0.07483, over 13459.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2548, pruned_loss=0.07255, over 2650080.98 frames. ], batch size: 106, lr: 9.36e-03, grad_scale: 8.0 2023-04-16 21:04:38,584 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36805.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 21:04:45,891 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.849e+02 2.903e+02 3.363e+02 3.975e+02 5.959e+02, threshold=6.726e+02, percent-clipped=0.0 2023-04-16 21:04:50,521 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2023-04-16 21:05:08,703 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36837.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:05:10,207 INFO [train.py:893] (0/4) Epoch 14, batch 1750, loss[loss=0.1921, simple_loss=0.2485, pruned_loss=0.06785, over 13269.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2533, pruned_loss=0.07142, over 2654909.73 frames. ], batch size: 124, lr: 9.35e-03, grad_scale: 8.0 2023-04-16 21:05:36,567 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36866.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 21:05:53,878 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36885.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:05:57,804 INFO [train.py:893] (0/4) Epoch 14, batch 1800, loss[loss=0.1869, simple_loss=0.2461, pruned_loss=0.06384, over 13463.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2533, pruned_loss=0.07172, over 2652480.49 frames. ], batch size: 106, lr: 9.34e-03, grad_scale: 8.0 2023-04-16 21:06:13,610 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-16 21:06:14,200 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3027, 3.6087, 3.5313, 3.9983, 2.0911, 3.0455, 3.6758, 2.0573], device='cuda:0'), covar=tensor([0.0121, 0.0516, 0.0663, 0.0544, 0.1560, 0.0885, 0.0573, 0.1810], device='cuda:0'), in_proj_covar=tensor([0.0106, 0.0175, 0.0192, 0.0215, 0.0175, 0.0190, 0.0168, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:06:17,385 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36911.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:06:19,632 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.892e+02 2.727e+02 3.240e+02 3.998e+02 7.863e+02, threshold=6.481e+02, percent-clipped=1.0 2023-04-16 21:06:43,938 INFO [train.py:893] (0/4) Epoch 14, batch 1850, loss[loss=0.1893, simple_loss=0.247, pruned_loss=0.06576, over 13452.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2524, pruned_loss=0.07115, over 2650661.26 frames. ], batch size: 103, lr: 9.34e-03, grad_scale: 8.0 2023-04-16 21:06:44,915 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 21:06:56,163 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1278, 4.2078, 3.3701, 2.8468, 2.9579, 2.5368, 4.4119, 2.5028], device='cuda:0'), covar=tensor([0.1429, 0.0328, 0.0907, 0.1720, 0.0726, 0.2838, 0.0180, 0.3479], device='cuda:0'), in_proj_covar=tensor([0.0299, 0.0275, 0.0284, 0.0300, 0.0236, 0.0303, 0.0194, 0.0347], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 21:07:02,623 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36958.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:07:05,097 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36961.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:07:07,620 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36964.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:07:08,610 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2115, 4.3716, 2.9995, 4.2285, 4.1655, 2.7633, 3.8440, 2.8911], device='cuda:0'), covar=tensor([0.0256, 0.0268, 0.1109, 0.0319, 0.0243, 0.1151, 0.0465, 0.1383], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0155, 0.0172, 0.0166, 0.0125, 0.0157, 0.0151, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:07:10,840 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36968.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:07:14,382 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36972.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:07:31,586 INFO [train.py:893] (0/4) Epoch 14, batch 1900, loss[loss=0.1885, simple_loss=0.2481, pruned_loss=0.06443, over 13533.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2515, pruned_loss=0.07095, over 2656745.96 frames. ], batch size: 91, lr: 9.33e-03, grad_scale: 8.0 2023-04-16 21:07:44,573 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37004.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:07:53,574 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.879e+02 2.818e+02 3.261e+02 3.804e+02 6.289e+02, threshold=6.522e+02, percent-clipped=0.0 2023-04-16 21:07:58,741 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37019.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:08:13,291 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-16 21:08:16,144 INFO [train.py:893] (0/4) Epoch 14, batch 1950, loss[loss=0.223, simple_loss=0.2642, pruned_loss=0.09091, over 13525.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2512, pruned_loss=0.07107, over 2658340.34 frames. ], batch size: 85, lr: 9.32e-03, grad_scale: 8.0 2023-04-16 21:08:33,441 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-16 21:08:33,799 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37056.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 21:08:41,339 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37065.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:09:03,559 INFO [train.py:893] (0/4) Epoch 14, batch 2000, loss[loss=0.1795, simple_loss=0.235, pruned_loss=0.06198, over 13346.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2532, pruned_loss=0.07233, over 2654802.33 frames. ], batch size: 67, lr: 9.32e-03, grad_scale: 8.0 2023-04-16 21:09:05,466 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1530, 4.6699, 4.5175, 4.4090, 4.4610, 4.2875, 4.6640, 4.7042], device='cuda:0'), covar=tensor([0.0219, 0.0190, 0.0167, 0.0321, 0.0226, 0.0251, 0.0273, 0.0209], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0180, 0.0142, 0.0165, 0.0128, 0.0180, 0.0121, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 21:09:07,706 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 21:09:26,942 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.742e+02 2.839e+02 3.422e+02 4.168e+02 1.026e+03, threshold=6.843e+02, percent-clipped=3.0 2023-04-16 21:09:36,628 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8989, 2.7808, 2.4340, 1.7533, 1.7019, 2.4381, 2.3675, 2.9069], device='cuda:0'), covar=tensor([0.1032, 0.0303, 0.0833, 0.1707, 0.0396, 0.0484, 0.0806, 0.0305], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0118, 0.0105, 0.0191, 0.0098, 0.0138, 0.0146, 0.0115], device='cuda:0'), out_proj_covar=tensor([1.1193e-04, 8.8514e-05, 8.3916e-05, 1.4544e-04, 7.6339e-05, 1.0410e-04, 1.1262e-04, 8.5522e-05], device='cuda:0') 2023-04-16 21:09:49,573 INFO [train.py:893] (0/4) Epoch 14, batch 2050, loss[loss=0.1859, simple_loss=0.244, pruned_loss=0.06393, over 13553.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2543, pruned_loss=0.07298, over 2653478.28 frames. ], batch size: 78, lr: 9.31e-03, grad_scale: 8.0 2023-04-16 21:10:09,981 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37161.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 21:10:36,601 INFO [train.py:893] (0/4) Epoch 14, batch 2100, loss[loss=0.1899, simple_loss=0.2438, pruned_loss=0.06794, over 13583.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2534, pruned_loss=0.07244, over 2655529.57 frames. ], batch size: 89, lr: 9.31e-03, grad_scale: 8.0 2023-04-16 21:11:00,506 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.834e+02 2.540e+02 2.961e+02 3.544e+02 6.483e+02, threshold=5.922e+02, percent-clipped=0.0 2023-04-16 21:11:05,080 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1739, 3.9884, 3.1755, 3.8755, 3.3786, 2.0943, 3.9468, 2.1349], device='cuda:0'), covar=tensor([0.0665, 0.0452, 0.0479, 0.0201, 0.0784, 0.2085, 0.0821, 0.1419], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0127, 0.0126, 0.0105, 0.0144, 0.0179, 0.0151, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:11:14,034 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-16 21:11:19,329 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4765, 3.5365, 4.1905, 2.8510, 2.8385, 2.8569, 4.4249, 4.5652], device='cuda:0'), covar=tensor([0.1198, 0.1494, 0.0368, 0.1708, 0.1519, 0.1423, 0.0245, 0.0204], device='cuda:0'), in_proj_covar=tensor([0.0227, 0.0246, 0.0178, 0.0212, 0.0205, 0.0173, 0.0182, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:11:23,871 INFO [train.py:893] (0/4) Epoch 14, batch 2150, loss[loss=0.1784, simple_loss=0.243, pruned_loss=0.05688, over 13371.00 frames. ], tot_loss[loss=0.198, simple_loss=0.253, pruned_loss=0.0715, over 2657935.43 frames. ], batch size: 73, lr: 9.30e-03, grad_scale: 8.0 2023-04-16 21:11:41,501 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0321, 4.5248, 4.3176, 4.2215, 4.2724, 4.0816, 4.5348, 4.5539], device='cuda:0'), covar=tensor([0.0201, 0.0196, 0.0206, 0.0348, 0.0290, 0.0283, 0.0291, 0.0227], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0180, 0.0143, 0.0163, 0.0128, 0.0179, 0.0120, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 21:11:44,075 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37261.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:11:46,714 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37264.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:11:49,147 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37267.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:11:50,050 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37268.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:11:50,082 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37268.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:12:08,913 INFO [train.py:893] (0/4) Epoch 14, batch 2200, loss[loss=0.1904, simple_loss=0.2431, pruned_loss=0.06886, over 13550.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2534, pruned_loss=0.07183, over 2653487.45 frames. ], batch size: 76, lr: 9.29e-03, grad_scale: 8.0 2023-04-16 21:12:28,754 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37309.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:12:29,761 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7808, 2.2598, 2.3334, 2.8019, 2.1134, 2.7557, 2.8226, 2.2478], device='cuda:0'), covar=tensor([0.0073, 0.0184, 0.0128, 0.0114, 0.0195, 0.0121, 0.0134, 0.0163], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0091, 0.0099, 0.0092, 0.0106, 0.0084, 0.0086, 0.0085], device='cuda:0'), out_proj_covar=tensor([8.8992e-05, 9.9970e-05, 1.1064e-04, 1.0136e-04, 1.1885e-04, 9.1319e-05, 9.4989e-05, 9.2184e-05], device='cuda:0') 2023-04-16 21:12:30,718 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1010, 4.1615, 3.4407, 2.8758, 2.9454, 2.5220, 4.3702, 2.4187], device='cuda:0'), covar=tensor([0.1445, 0.0282, 0.0834, 0.1721, 0.0752, 0.2962, 0.0182, 0.3683], device='cuda:0'), in_proj_covar=tensor([0.0298, 0.0277, 0.0285, 0.0301, 0.0238, 0.0304, 0.0195, 0.0346], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 21:12:31,314 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37312.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:12:32,934 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.839e+02 2.752e+02 3.188e+02 3.717e+02 5.607e+02, threshold=6.377e+02, percent-clipped=0.0 2023-04-16 21:12:33,134 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37314.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:12:34,805 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37316.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:12:36,052 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-16 21:12:47,421 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37329.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:12:56,961 INFO [train.py:893] (0/4) Epoch 14, batch 2250, loss[loss=0.2251, simple_loss=0.2787, pruned_loss=0.0858, over 13340.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2518, pruned_loss=0.0714, over 2650155.54 frames. ], batch size: 118, lr: 9.29e-03, grad_scale: 8.0 2023-04-16 21:13:13,036 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37356.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 21:13:16,442 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37360.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:13:20,596 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37364.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:13:43,687 INFO [train.py:893] (0/4) Epoch 14, batch 2300, loss[loss=0.1942, simple_loss=0.2529, pruned_loss=0.0677, over 13190.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2513, pruned_loss=0.07101, over 2651519.86 frames. ], batch size: 132, lr: 9.28e-03, grad_scale: 8.0 2023-04-16 21:13:58,158 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37404.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 21:14:06,333 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.211e+02 2.889e+02 3.254e+02 3.813e+02 6.240e+02, threshold=6.509e+02, percent-clipped=0.0 2023-04-16 21:14:18,273 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37425.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:14:30,526 INFO [train.py:893] (0/4) Epoch 14, batch 2350, loss[loss=0.2013, simple_loss=0.2606, pruned_loss=0.07098, over 13525.00 frames. ], tot_loss[loss=0.1969, simple_loss=0.2514, pruned_loss=0.07119, over 2653550.73 frames. ], batch size: 70, lr: 9.28e-03, grad_scale: 8.0 2023-04-16 21:14:49,841 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0176, 4.1063, 2.8033, 3.8628, 3.9360, 2.4186, 3.6473, 2.7854], device='cuda:0'), covar=tensor([0.0308, 0.0247, 0.1099, 0.0370, 0.0270, 0.1325, 0.0528, 0.1219], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0155, 0.0173, 0.0168, 0.0127, 0.0158, 0.0151, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:14:51,991 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 21:14:52,213 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37461.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 21:15:17,099 INFO [train.py:893] (0/4) Epoch 14, batch 2400, loss[loss=0.1703, simple_loss=0.2279, pruned_loss=0.05633, over 13500.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2512, pruned_loss=0.07109, over 2659627.20 frames. ], batch size: 70, lr: 9.27e-03, grad_scale: 8.0 2023-04-16 21:15:36,042 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37509.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 21:15:39,009 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-16 21:15:40,915 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.026e+02 2.653e+02 3.203e+02 3.692e+02 8.487e+02, threshold=6.405e+02, percent-clipped=2.0 2023-04-16 21:16:03,455 INFO [train.py:893] (0/4) Epoch 14, batch 2450, loss[loss=0.18, simple_loss=0.2246, pruned_loss=0.06771, over 13205.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.2504, pruned_loss=0.07057, over 2656325.83 frames. ], batch size: 58, lr: 9.26e-03, grad_scale: 8.0 2023-04-16 21:16:30,134 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37567.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:16:51,645 INFO [train.py:893] (0/4) Epoch 14, batch 2500, loss[loss=0.1997, simple_loss=0.2586, pruned_loss=0.0704, over 13402.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2508, pruned_loss=0.07053, over 2656489.46 frames. ], batch size: 113, lr: 9.26e-03, grad_scale: 8.0 2023-04-16 21:17:00,618 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5120, 4.5784, 3.1683, 4.3960, 4.3886, 3.0906, 4.0792, 3.0235], device='cuda:0'), covar=tensor([0.0257, 0.0228, 0.1080, 0.0384, 0.0238, 0.1095, 0.0415, 0.1449], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0156, 0.0174, 0.0170, 0.0127, 0.0159, 0.0153, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:17:11,059 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-16 21:17:14,574 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.069e+02 2.773e+02 3.225e+02 3.758e+02 6.643e+02, threshold=6.450e+02, percent-clipped=1.0 2023-04-16 21:17:14,891 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37614.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:17:15,624 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37615.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:17:24,953 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37624.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:17:38,667 INFO [train.py:893] (0/4) Epoch 14, batch 2550, loss[loss=0.189, simple_loss=0.2482, pruned_loss=0.06489, over 13456.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2507, pruned_loss=0.07051, over 2655404.00 frames. ], batch size: 103, lr: 9.25e-03, grad_scale: 8.0 2023-04-16 21:17:58,129 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37660.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:17:59,629 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37662.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:18:01,212 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 21:18:22,197 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2033, 4.4597, 4.2216, 4.2192, 4.3166, 4.6787, 4.4454, 4.3411], device='cuda:0'), covar=tensor([0.0266, 0.0283, 0.0319, 0.0941, 0.0258, 0.0215, 0.0293, 0.0269], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0137, 0.0151, 0.0250, 0.0155, 0.0170, 0.0150, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2023-04-16 21:18:25,177 INFO [train.py:893] (0/4) Epoch 14, batch 2600, loss[loss=0.1852, simple_loss=0.2312, pruned_loss=0.06964, over 13378.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2502, pruned_loss=0.0708, over 2658226.06 frames. ], batch size: 65, lr: 9.25e-03, grad_scale: 8.0 2023-04-16 21:18:26,349 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.7576, 5.2939, 5.2738, 5.2250, 5.1273, 5.1327, 5.7272, 5.2417], device='cuda:0'), covar=tensor([0.0551, 0.0987, 0.1732, 0.2288, 0.0676, 0.1256, 0.0710, 0.1082], device='cuda:0'), in_proj_covar=tensor([0.0295, 0.0338, 0.0427, 0.0440, 0.0256, 0.0321, 0.0385, 0.0311], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 21:18:30,960 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-16 21:18:31,532 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4175, 2.6662, 2.3952, 4.2656, 4.8502, 3.5455, 4.7235, 4.4017], device='cuda:0'), covar=tensor([0.0113, 0.0890, 0.1028, 0.0109, 0.0102, 0.0449, 0.0096, 0.0105], device='cuda:0'), in_proj_covar=tensor([0.0068, 0.0085, 0.0092, 0.0074, 0.0059, 0.0076, 0.0051, 0.0068], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:18:42,897 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37708.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:18:47,446 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.401e+02 3.085e+02 3.718e+02 4.390e+02 1.242e+03, threshold=7.436e+02, percent-clipped=3.0 2023-04-16 21:18:52,004 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37720.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:19:07,751 INFO [train.py:893] (0/4) Epoch 14, batch 2650, loss[loss=0.1922, simple_loss=0.246, pruned_loss=0.06918, over 13531.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2516, pruned_loss=0.07224, over 2650384.46 frames. ], batch size: 83, lr: 9.24e-03, grad_scale: 8.0 2023-04-16 21:19:45,560 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-14.pt 2023-04-16 21:20:10,977 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 21:20:15,081 INFO [train.py:893] (0/4) Epoch 15, batch 0, loss[loss=0.1865, simple_loss=0.2436, pruned_loss=0.06466, over 13244.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2436, pruned_loss=0.06466, over 13244.00 frames. ], batch size: 124, lr: 8.92e-03, grad_scale: 8.0 2023-04-16 21:20:15,082 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 21:20:21,929 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8908, 4.2000, 3.8722, 4.2464, 3.6928, 3.0974, 4.0654, 2.9693], device='cuda:0'), covar=tensor([0.0365, 0.0368, 0.0259, 0.0148, 0.0468, 0.1186, 0.0951, 0.0882], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0129, 0.0127, 0.0106, 0.0144, 0.0180, 0.0154, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 21:20:37,763 INFO [train.py:927] (0/4) Epoch 15, validation: loss=0.1473, simple_loss=0.2051, pruned_loss=0.04473, over 2446609.00 frames. 2023-04-16 21:20:37,764 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 21:21:03,135 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.991e+02 2.725e+02 3.297e+02 3.929e+02 8.342e+02, threshold=6.594e+02, percent-clipped=2.0 2023-04-16 21:21:22,617 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-16 21:21:24,612 INFO [train.py:893] (0/4) Epoch 15, batch 50, loss[loss=0.2072, simple_loss=0.2582, pruned_loss=0.0781, over 13453.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2454, pruned_loss=0.06996, over 601565.50 frames. ], batch size: 106, lr: 8.91e-03, grad_scale: 8.0 2023-04-16 21:21:48,722 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 21:21:48,722 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 21:21:48,723 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 21:21:48,729 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 21:21:48,737 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 21:21:48,761 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 21:21:48,770 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 21:22:04,140 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-16 21:22:11,253 INFO [train.py:893] (0/4) Epoch 15, batch 100, loss[loss=0.1971, simple_loss=0.237, pruned_loss=0.07863, over 13366.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.2492, pruned_loss=0.07319, over 1055013.89 frames. ], batch size: 62, lr: 8.91e-03, grad_scale: 8.0 2023-04-16 21:22:14,074 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3140, 3.4736, 3.4017, 4.0305, 1.9763, 2.7953, 3.7554, 2.0090], device='cuda:0'), covar=tensor([0.0152, 0.0649, 0.0843, 0.0572, 0.1760, 0.1118, 0.0667, 0.1987], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0176, 0.0197, 0.0219, 0.0179, 0.0192, 0.0171, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:22:20,754 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7452, 3.5987, 3.7081, 2.1258, 4.1475, 3.9335, 3.8977, 4.2182], device='cuda:0'), covar=tensor([0.0319, 0.0204, 0.0195, 0.1499, 0.0257, 0.0289, 0.0199, 0.0106], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0047, 0.0074, 0.0098, 0.0092, 0.0094, 0.0073, 0.0065], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 21:22:36,142 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 2.904e+02 3.473e+02 4.227e+02 6.603e+02, threshold=6.947e+02, percent-clipped=1.0 2023-04-16 21:22:46,269 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37924.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:22:58,547 INFO [train.py:893] (0/4) Epoch 15, batch 150, loss[loss=0.2075, simple_loss=0.2579, pruned_loss=0.07851, over 13467.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2507, pruned_loss=0.07366, over 1407382.59 frames. ], batch size: 100, lr: 8.90e-03, grad_scale: 8.0 2023-04-16 21:23:24,863 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7683, 3.6266, 3.7639, 2.1353, 3.9938, 3.8190, 3.7878, 3.9373], device='cuda:0'), covar=tensor([0.0218, 0.0132, 0.0138, 0.1178, 0.0120, 0.0213, 0.0132, 0.0103], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0047, 0.0072, 0.0096, 0.0090, 0.0092, 0.0072, 0.0064], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 21:23:31,302 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37972.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:23:37,373 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37979.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:23:46,321 INFO [train.py:893] (0/4) Epoch 15, batch 200, loss[loss=0.1849, simple_loss=0.2167, pruned_loss=0.07656, over 8092.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2531, pruned_loss=0.07464, over 1672721.21 frames. ], batch size: 32, lr: 8.90e-03, grad_scale: 16.0 2023-04-16 21:23:57,253 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-38000.pt 2023-04-16 21:24:13,370 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.082e+02 2.887e+02 3.344e+02 4.175e+02 8.023e+02, threshold=6.688e+02, percent-clipped=1.0 2023-04-16 21:24:20,116 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=38020.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:24:36,574 INFO [train.py:893] (0/4) Epoch 15, batch 250, loss[loss=0.2068, simple_loss=0.2624, pruned_loss=0.0756, over 13498.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2548, pruned_loss=0.07575, over 1886598.62 frames. ], batch size: 93, lr: 8.89e-03, grad_scale: 16.0 2023-04-16 21:24:38,573 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38040.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:25:04,608 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=38068.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:25:22,619 INFO [train.py:893] (0/4) Epoch 15, batch 300, loss[loss=0.2189, simple_loss=0.2864, pruned_loss=0.07568, over 13473.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2563, pruned_loss=0.07634, over 2056839.58 frames. ], batch size: 100, lr: 8.88e-03, grad_scale: 16.0 2023-04-16 21:25:31,821 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5266, 3.2349, 2.5420, 2.8476, 2.7825, 1.8163, 3.3300, 1.8026], device='cuda:0'), covar=tensor([0.0666, 0.0690, 0.0561, 0.0510, 0.0745, 0.2211, 0.0852, 0.1492], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0130, 0.0128, 0.0108, 0.0146, 0.0182, 0.0155, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 21:25:32,016 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-16 21:25:40,871 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8348, 2.3299, 1.7892, 3.6953, 4.2071, 3.0932, 4.0606, 3.8401], device='cuda:0'), covar=tensor([0.0097, 0.0998, 0.1148, 0.0109, 0.0059, 0.0510, 0.0091, 0.0094], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0084, 0.0092, 0.0073, 0.0059, 0.0076, 0.0050, 0.0068], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:25:47,314 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.869e+02 2.752e+02 3.316e+02 4.021e+02 5.948e+02, threshold=6.631e+02, percent-clipped=0.0 2023-04-16 21:26:08,994 INFO [train.py:893] (0/4) Epoch 15, batch 350, loss[loss=0.177, simple_loss=0.2257, pruned_loss=0.06411, over 13361.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2561, pruned_loss=0.0759, over 2182719.15 frames. ], batch size: 67, lr: 8.88e-03, grad_scale: 16.0 2023-04-16 21:26:56,434 INFO [train.py:893] (0/4) Epoch 15, batch 400, loss[loss=0.214, simple_loss=0.2725, pruned_loss=0.07778, over 13452.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2566, pruned_loss=0.07557, over 2283465.79 frames. ], batch size: 106, lr: 8.87e-03, grad_scale: 16.0 2023-04-16 21:27:21,336 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.975e+02 2.951e+02 3.498e+02 4.195e+02 7.332e+02, threshold=6.996e+02, percent-clipped=1.0 2023-04-16 21:27:28,253 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4246, 2.3226, 2.6298, 4.0107, 3.6406, 4.0395, 3.0248, 2.3305], device='cuda:0'), covar=tensor([0.0267, 0.1022, 0.0818, 0.0046, 0.0212, 0.0042, 0.0726, 0.1008], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0148, 0.0159, 0.0086, 0.0111, 0.0083, 0.0164, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0001, 0.0002, 0.0001, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:27:44,052 INFO [train.py:893] (0/4) Epoch 15, batch 450, loss[loss=0.2898, simple_loss=0.3183, pruned_loss=0.1307, over 11827.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2575, pruned_loss=0.07592, over 2360211.66 frames. ], batch size: 157, lr: 8.87e-03, grad_scale: 16.0 2023-04-16 21:27:54,926 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3809, 4.8314, 4.7644, 4.9064, 4.5691, 4.7441, 5.3716, 4.8417], device='cuda:0'), covar=tensor([0.0697, 0.1074, 0.1977, 0.2227, 0.1072, 0.1504, 0.0774, 0.1122], device='cuda:0'), in_proj_covar=tensor([0.0295, 0.0337, 0.0426, 0.0440, 0.0260, 0.0324, 0.0384, 0.0311], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 21:28:07,474 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 21:28:31,550 INFO [train.py:893] (0/4) Epoch 15, batch 500, loss[loss=0.2846, simple_loss=0.3205, pruned_loss=0.1243, over 11877.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.257, pruned_loss=0.07514, over 2427254.48 frames. ], batch size: 157, lr: 8.86e-03, grad_scale: 16.0 2023-04-16 21:28:55,560 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.883e+02 2.767e+02 3.304e+02 3.926e+02 2.162e+03, threshold=6.608e+02, percent-clipped=2.0 2023-04-16 21:29:15,796 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=38335.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:29:18,102 INFO [train.py:893] (0/4) Epoch 15, batch 550, loss[loss=0.1715, simple_loss=0.229, pruned_loss=0.05697, over 13371.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.256, pruned_loss=0.07414, over 2479883.91 frames. ], batch size: 67, lr: 8.86e-03, grad_scale: 16.0 2023-04-16 21:29:38,302 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=38360.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:29:44,040 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7777, 3.8846, 3.0596, 2.5700, 2.6604, 2.2672, 3.9545, 2.1684], device='cuda:0'), covar=tensor([0.1513, 0.0326, 0.0944, 0.1920, 0.0792, 0.2984, 0.0231, 0.3624], device='cuda:0'), in_proj_covar=tensor([0.0306, 0.0283, 0.0295, 0.0310, 0.0244, 0.0312, 0.0199, 0.0355], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 21:30:04,324 INFO [train.py:893] (0/4) Epoch 15, batch 600, loss[loss=0.2317, simple_loss=0.2793, pruned_loss=0.09205, over 11856.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.254, pruned_loss=0.07337, over 2522692.89 frames. ], batch size: 157, lr: 8.85e-03, grad_scale: 16.0 2023-04-16 21:30:29,601 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.966e+02 2.757e+02 3.172e+02 3.903e+02 7.089e+02, threshold=6.344e+02, percent-clipped=2.0 2023-04-16 21:30:31,015 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.44 vs. limit=5.0 2023-04-16 21:30:35,966 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38421.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:30:51,597 INFO [train.py:893] (0/4) Epoch 15, batch 650, loss[loss=0.2202, simple_loss=0.269, pruned_loss=0.08568, over 13494.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2531, pruned_loss=0.07258, over 2555366.52 frames. ], batch size: 93, lr: 8.84e-03, grad_scale: 16.0 2023-04-16 21:30:58,522 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-16 21:31:16,785 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6914, 4.0234, 3.8334, 4.4153, 2.6403, 3.3053, 4.1680, 2.3297], device='cuda:0'), covar=tensor([0.0115, 0.0496, 0.0698, 0.0664, 0.1443, 0.1013, 0.0464, 0.1949], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0177, 0.0196, 0.0219, 0.0178, 0.0192, 0.0172, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:31:38,116 INFO [train.py:893] (0/4) Epoch 15, batch 700, loss[loss=0.1995, simple_loss=0.2562, pruned_loss=0.07143, over 13519.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2519, pruned_loss=0.07169, over 2576953.66 frames. ], batch size: 91, lr: 8.84e-03, grad_scale: 16.0 2023-04-16 21:31:43,624 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-16 21:32:02,934 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.840e+02 2.763e+02 3.318e+02 3.949e+02 6.088e+02, threshold=6.635e+02, percent-clipped=0.0 2023-04-16 21:32:08,344 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8943, 3.8492, 3.8736, 2.2484, 4.1895, 3.9864, 4.0070, 4.1450], device='cuda:0'), covar=tensor([0.0213, 0.0128, 0.0128, 0.1147, 0.0132, 0.0227, 0.0113, 0.0087], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0048, 0.0073, 0.0097, 0.0091, 0.0093, 0.0072, 0.0065], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 21:32:25,727 INFO [train.py:893] (0/4) Epoch 15, batch 750, loss[loss=0.2045, simple_loss=0.2567, pruned_loss=0.07618, over 13120.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2526, pruned_loss=0.07214, over 2595331.54 frames. ], batch size: 142, lr: 8.83e-03, grad_scale: 16.0 2023-04-16 21:33:12,675 INFO [train.py:893] (0/4) Epoch 15, batch 800, loss[loss=0.1953, simple_loss=0.2565, pruned_loss=0.06702, over 13437.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2543, pruned_loss=0.07314, over 2608015.85 frames. ], batch size: 103, lr: 8.83e-03, grad_scale: 16.0 2023-04-16 21:33:36,766 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.167e+02 2.924e+02 3.517e+02 4.186e+02 6.673e+02, threshold=7.034e+02, percent-clipped=1.0 2023-04-16 21:33:58,005 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=38635.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:34:00,217 INFO [train.py:893] (0/4) Epoch 15, batch 850, loss[loss=0.1959, simple_loss=0.2355, pruned_loss=0.0781, over 13378.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2548, pruned_loss=0.07343, over 2617776.21 frames. ], batch size: 62, lr: 8.82e-03, grad_scale: 16.0 2023-04-16 21:34:09,816 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.89 vs. limit=5.0 2023-04-16 21:34:13,694 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6919, 5.0294, 4.7730, 4.7392, 4.8196, 4.6309, 5.0236, 5.0380], device='cuda:0'), covar=tensor([0.0145, 0.0170, 0.0215, 0.0291, 0.0180, 0.0215, 0.0240, 0.0173], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0181, 0.0140, 0.0162, 0.0126, 0.0175, 0.0118, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 21:34:42,323 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=38683.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:34:46,947 INFO [train.py:893] (0/4) Epoch 15, batch 900, loss[loss=0.2134, simple_loss=0.2584, pruned_loss=0.08418, over 13057.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2545, pruned_loss=0.07391, over 2625433.68 frames. ], batch size: 142, lr: 8.82e-03, grad_scale: 16.0 2023-04-16 21:34:53,931 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-16 21:35:04,707 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6062, 3.3866, 2.6726, 3.1322, 2.7893, 1.9773, 3.3884, 1.7778], device='cuda:0'), covar=tensor([0.0740, 0.0660, 0.0582, 0.0393, 0.0814, 0.2146, 0.1174, 0.1488], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0130, 0.0126, 0.0107, 0.0144, 0.0181, 0.0155, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 21:35:10,823 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.917e+02 2.822e+02 3.453e+02 4.319e+02 8.057e+02, threshold=6.905e+02, percent-clipped=3.0 2023-04-16 21:35:12,748 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=38716.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:35:18,437 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 21:35:33,151 INFO [train.py:893] (0/4) Epoch 15, batch 950, loss[loss=0.2264, simple_loss=0.2689, pruned_loss=0.09199, over 13414.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2536, pruned_loss=0.07389, over 2637591.14 frames. ], batch size: 95, lr: 8.81e-03, grad_scale: 16.0 2023-04-16 21:35:48,728 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8927, 2.4003, 1.9204, 3.7424, 4.2854, 3.2675, 4.2444, 3.9492], device='cuda:0'), covar=tensor([0.0136, 0.1048, 0.1165, 0.0136, 0.0106, 0.0441, 0.0103, 0.0105], device='cuda:0'), in_proj_covar=tensor([0.0068, 0.0086, 0.0093, 0.0074, 0.0059, 0.0077, 0.0051, 0.0068], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:36:08,497 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9132, 4.7945, 5.0180, 4.8362, 5.2689, 4.7563, 5.2984, 5.2719], device='cuda:0'), covar=tensor([0.0353, 0.0519, 0.0595, 0.0531, 0.0474, 0.0802, 0.0411, 0.0423], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0271, 0.0270, 0.0199, 0.0386, 0.0311, 0.0242, 0.0265], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:36:20,150 INFO [train.py:893] (0/4) Epoch 15, batch 1000, loss[loss=0.1994, simple_loss=0.2556, pruned_loss=0.07163, over 13445.00 frames. ], tot_loss[loss=0.199, simple_loss=0.252, pruned_loss=0.07294, over 2646042.78 frames. ], batch size: 103, lr: 8.80e-03, grad_scale: 16.0 2023-04-16 21:36:20,470 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.7940, 4.0068, 3.6361, 4.3977, 2.5978, 3.2295, 4.1273, 2.3594], device='cuda:0'), covar=tensor([0.0105, 0.0526, 0.0760, 0.0681, 0.1348, 0.0915, 0.0534, 0.1605], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0178, 0.0198, 0.0221, 0.0180, 0.0193, 0.0174, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:36:44,532 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 2.946e+02 3.426e+02 3.965e+02 6.022e+02, threshold=6.852e+02, percent-clipped=0.0 2023-04-16 21:36:53,943 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=38823.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:37:06,382 INFO [train.py:893] (0/4) Epoch 15, batch 1050, loss[loss=0.1975, simple_loss=0.2626, pruned_loss=0.06625, over 13521.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2507, pruned_loss=0.07181, over 2650281.35 frames. ], batch size: 91, lr: 8.80e-03, grad_scale: 16.0 2023-04-16 21:37:49,202 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38884.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:37:52,988 INFO [train.py:893] (0/4) Epoch 15, batch 1100, loss[loss=0.2126, simple_loss=0.2649, pruned_loss=0.08017, over 13525.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2513, pruned_loss=0.07168, over 2653036.15 frames. ], batch size: 91, lr: 8.79e-03, grad_scale: 16.0 2023-04-16 21:38:16,578 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.006e+02 2.873e+02 3.467e+02 4.296e+02 7.811e+02, threshold=6.933e+02, percent-clipped=2.0 2023-04-16 21:38:22,000 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.0264, 4.8375, 5.1358, 4.9415, 5.3811, 4.8921, 5.3793, 5.3861], device='cuda:0'), covar=tensor([0.0345, 0.0611, 0.0595, 0.0560, 0.0541, 0.0874, 0.0453, 0.0459], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0271, 0.0269, 0.0203, 0.0387, 0.0312, 0.0243, 0.0267], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:38:26,196 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6621, 2.3747, 2.1482, 2.7437, 2.1055, 2.7179, 2.7006, 2.2818], device='cuda:0'), covar=tensor([0.0110, 0.0183, 0.0161, 0.0124, 0.0199, 0.0134, 0.0165, 0.0189], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0095, 0.0102, 0.0097, 0.0110, 0.0086, 0.0088, 0.0089], device='cuda:0'), out_proj_covar=tensor([9.3093e-05, 1.0387e-04, 1.1354e-04, 1.0641e-04, 1.2275e-04, 9.2742e-05, 9.6808e-05, 9.6354e-05], device='cuda:0') 2023-04-16 21:38:38,250 INFO [train.py:893] (0/4) Epoch 15, batch 1150, loss[loss=0.2048, simple_loss=0.2661, pruned_loss=0.07174, over 13517.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2508, pruned_loss=0.07071, over 2655199.82 frames. ], batch size: 83, lr: 8.79e-03, grad_scale: 16.0 2023-04-16 21:39:26,105 INFO [train.py:893] (0/4) Epoch 15, batch 1200, loss[loss=0.1919, simple_loss=0.2513, pruned_loss=0.0662, over 13441.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.2505, pruned_loss=0.07015, over 2656103.72 frames. ], batch size: 106, lr: 8.78e-03, grad_scale: 16.0 2023-04-16 21:39:50,724 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.045e+02 2.812e+02 3.202e+02 4.092e+02 7.055e+02, threshold=6.404e+02, percent-clipped=2.0 2023-04-16 21:39:52,719 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39016.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:39:55,049 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 21:40:05,964 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 21:40:12,609 INFO [train.py:893] (0/4) Epoch 15, batch 1250, loss[loss=0.1937, simple_loss=0.2523, pruned_loss=0.06757, over 13353.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2513, pruned_loss=0.07088, over 2658179.02 frames. ], batch size: 73, lr: 8.78e-03, grad_scale: 16.0 2023-04-16 21:40:21,222 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5576, 3.6169, 2.9464, 3.2725, 2.8150, 2.0185, 3.5656, 1.9867], device='cuda:0'), covar=tensor([0.0705, 0.0481, 0.0389, 0.0374, 0.0665, 0.2094, 0.0696, 0.1373], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0129, 0.0125, 0.0108, 0.0144, 0.0180, 0.0153, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:40:37,449 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39064.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:40:50,561 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9759, 4.4242, 4.4333, 4.5314, 4.1127, 4.2978, 4.9581, 4.4217], device='cuda:0'), covar=tensor([0.0855, 0.1274, 0.2217, 0.2707, 0.1129, 0.1709, 0.0846, 0.1100], device='cuda:0'), in_proj_covar=tensor([0.0299, 0.0339, 0.0430, 0.0441, 0.0263, 0.0330, 0.0390, 0.0314], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 21:40:59,510 INFO [train.py:893] (0/4) Epoch 15, batch 1300, loss[loss=0.1923, simple_loss=0.2563, pruned_loss=0.06409, over 13536.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2524, pruned_loss=0.07123, over 2658509.43 frames. ], batch size: 76, lr: 8.77e-03, grad_scale: 16.0 2023-04-16 21:41:23,781 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.038e+02 2.996e+02 3.347e+02 3.864e+02 5.865e+02, threshold=6.694e+02, percent-clipped=0.0 2023-04-16 21:41:25,843 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9667, 3.9951, 3.0337, 3.6857, 3.0491, 2.3012, 3.9393, 2.1289], device='cuda:0'), covar=tensor([0.0591, 0.0324, 0.0469, 0.0234, 0.0710, 0.1706, 0.0593, 0.1276], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0128, 0.0124, 0.0107, 0.0143, 0.0179, 0.0152, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:41:44,162 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.88 vs. limit=5.0 2023-04-16 21:41:46,889 INFO [train.py:893] (0/4) Epoch 15, batch 1350, loss[loss=0.1855, simple_loss=0.2243, pruned_loss=0.07335, over 10729.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.2525, pruned_loss=0.07127, over 2658883.85 frames. ], batch size: 43, lr: 8.77e-03, grad_scale: 16.0 2023-04-16 21:42:11,609 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0275, 2.6651, 2.4589, 1.7456, 1.6306, 2.3304, 2.4492, 2.9597], device='cuda:0'), covar=tensor([0.0882, 0.0273, 0.0540, 0.1519, 0.0287, 0.0415, 0.0702, 0.0191], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0120, 0.0106, 0.0194, 0.0098, 0.0143, 0.0150, 0.0116], device='cuda:0'), out_proj_covar=tensor([1.1137e-04, 9.0559e-05, 8.3895e-05, 1.4700e-04, 7.5334e-05, 1.0796e-04, 1.1560e-04, 8.6252e-05], device='cuda:0') 2023-04-16 21:42:24,301 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39179.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:42:29,309 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39185.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:42:32,182 INFO [train.py:893] (0/4) Epoch 15, batch 1400, loss[loss=0.2141, simple_loss=0.2643, pruned_loss=0.08193, over 13492.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2512, pruned_loss=0.07091, over 2661714.26 frames. ], batch size: 93, lr: 8.76e-03, grad_scale: 16.0 2023-04-16 21:42:42,371 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0570, 1.7561, 3.8035, 3.7402, 3.7644, 2.9348, 3.4178, 2.8132], device='cuda:0'), covar=tensor([0.2057, 0.1684, 0.0132, 0.0162, 0.0193, 0.0756, 0.0270, 0.0842], device='cuda:0'), in_proj_covar=tensor([0.0211, 0.0183, 0.0113, 0.0120, 0.0124, 0.0172, 0.0132, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:42:42,412 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7881, 4.0296, 2.6325, 3.9469, 4.0494, 2.6410, 3.5497, 3.0102], device='cuda:0'), covar=tensor([0.0369, 0.0429, 0.1204, 0.0369, 0.0248, 0.1156, 0.0544, 0.1150], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0155, 0.0170, 0.0170, 0.0125, 0.0155, 0.0151, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:42:45,603 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-16 21:42:55,638 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 2.883e+02 3.401e+02 4.315e+02 6.663e+02, threshold=6.801e+02, percent-clipped=0.0 2023-04-16 21:43:17,758 INFO [train.py:893] (0/4) Epoch 15, batch 1450, loss[loss=0.2057, simple_loss=0.265, pruned_loss=0.07323, over 13232.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2514, pruned_loss=0.07094, over 2666228.26 frames. ], batch size: 124, lr: 8.75e-03, grad_scale: 16.0 2023-04-16 21:43:24,848 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39246.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:43:59,223 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39284.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:44:03,095 INFO [train.py:893] (0/4) Epoch 15, batch 1500, loss[loss=0.2071, simple_loss=0.266, pruned_loss=0.07411, over 13247.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2515, pruned_loss=0.071, over 2665355.17 frames. ], batch size: 124, lr: 8.75e-03, grad_scale: 16.0 2023-04-16 21:44:08,321 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1308, 4.5872, 4.4053, 4.3041, 4.3968, 4.1753, 4.6563, 4.6795], device='cuda:0'), covar=tensor([0.0214, 0.0218, 0.0168, 0.0315, 0.0255, 0.0271, 0.0245, 0.0204], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0183, 0.0141, 0.0165, 0.0129, 0.0179, 0.0121, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 21:44:26,764 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.2942, 2.0278, 2.0302, 2.3674, 1.7307, 2.3701, 2.3165, 1.9847], device='cuda:0'), covar=tensor([0.0090, 0.0234, 0.0140, 0.0155, 0.0212, 0.0133, 0.0191, 0.0183], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0097, 0.0103, 0.0100, 0.0113, 0.0089, 0.0092, 0.0092], device='cuda:0'), out_proj_covar=tensor([9.4619e-05, 1.0669e-04, 1.1406e-04, 1.0991e-04, 1.2598e-04, 9.5678e-05, 1.0107e-04, 9.8380e-05], device='cuda:0') 2023-04-16 21:44:27,200 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.904e+02 2.935e+02 3.502e+02 4.230e+02 9.051e+02, threshold=7.004e+02, percent-clipped=3.0 2023-04-16 21:44:49,927 INFO [train.py:893] (0/4) Epoch 15, batch 1550, loss[loss=0.2313, simple_loss=0.2801, pruned_loss=0.09127, over 13217.00 frames. ], tot_loss[loss=0.1969, simple_loss=0.2517, pruned_loss=0.071, over 2665326.01 frames. ], batch size: 132, lr: 8.74e-03, grad_scale: 16.0 2023-04-16 21:44:56,937 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39345.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:45:04,196 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3356, 2.4526, 1.9804, 3.9420, 4.6598, 3.3796, 4.5550, 4.3110], device='cuda:0'), covar=tensor([0.0103, 0.1019, 0.1168, 0.0148, 0.0087, 0.0496, 0.0083, 0.0100], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0087, 0.0094, 0.0076, 0.0061, 0.0078, 0.0051, 0.0070], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:45:37,386 INFO [train.py:893] (0/4) Epoch 15, batch 1600, loss[loss=0.2001, simple_loss=0.2631, pruned_loss=0.06858, over 13448.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2511, pruned_loss=0.07005, over 2665393.80 frames. ], batch size: 100, lr: 8.74e-03, grad_scale: 16.0 2023-04-16 21:45:42,788 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8464, 1.7667, 3.5779, 3.6178, 3.5686, 2.6231, 3.3594, 2.6116], device='cuda:0'), covar=tensor([0.2154, 0.1632, 0.0149, 0.0173, 0.0254, 0.0875, 0.0220, 0.0880], device='cuda:0'), in_proj_covar=tensor([0.0210, 0.0182, 0.0113, 0.0119, 0.0124, 0.0170, 0.0132, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:45:55,390 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-16 21:45:57,677 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.37 vs. limit=5.0 2023-04-16 21:46:01,160 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.914e+02 2.757e+02 3.152e+02 3.828e+02 7.575e+02, threshold=6.304e+02, percent-clipped=2.0 2023-04-16 21:46:17,810 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-16 21:46:23,029 INFO [train.py:893] (0/4) Epoch 15, batch 1650, loss[loss=0.2071, simple_loss=0.2707, pruned_loss=0.07182, over 13390.00 frames. ], tot_loss[loss=0.1962, simple_loss=0.2524, pruned_loss=0.06996, over 2665682.72 frames. ], batch size: 113, lr: 8.73e-03, grad_scale: 16.0 2023-04-16 21:46:39,906 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5886, 3.6525, 2.6961, 3.3799, 2.7784, 2.0504, 3.5753, 1.8778], device='cuda:0'), covar=tensor([0.0826, 0.0487, 0.0534, 0.0298, 0.0861, 0.1814, 0.0931, 0.1502], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0129, 0.0124, 0.0107, 0.0145, 0.0180, 0.0154, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:46:45,841 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39463.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:46:50,884 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-16 21:47:01,493 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39479.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:47:08,577 INFO [train.py:893] (0/4) Epoch 15, batch 1700, loss[loss=0.2102, simple_loss=0.2639, pruned_loss=0.07822, over 13432.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2524, pruned_loss=0.06982, over 2667902.22 frames. ], batch size: 106, lr: 8.73e-03, grad_scale: 16.0 2023-04-16 21:47:20,596 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4818, 3.4342, 4.0186, 3.0586, 2.7681, 2.8135, 4.2569, 4.3809], device='cuda:0'), covar=tensor([0.0998, 0.1450, 0.0314, 0.1453, 0.1377, 0.1242, 0.0322, 0.0192], device='cuda:0'), in_proj_covar=tensor([0.0228, 0.0245, 0.0178, 0.0212, 0.0205, 0.0174, 0.0185, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:47:24,793 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4504, 4.0861, 4.3443, 2.8619, 4.8185, 4.5102, 4.3992, 4.7195], device='cuda:0'), covar=tensor([0.0199, 0.0129, 0.0124, 0.0990, 0.0127, 0.0232, 0.0144, 0.0095], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0049, 0.0074, 0.0099, 0.0091, 0.0096, 0.0073, 0.0066], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 21:47:32,709 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.821e+02 2.928e+02 3.437e+02 4.043e+02 6.312e+02, threshold=6.874e+02, percent-clipped=1.0 2023-04-16 21:47:41,639 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39524.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:47:44,012 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39527.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:47:54,044 INFO [train.py:893] (0/4) Epoch 15, batch 1750, loss[loss=0.1907, simple_loss=0.2424, pruned_loss=0.06953, over 13396.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2508, pruned_loss=0.06895, over 2664941.81 frames. ], batch size: 109, lr: 8.72e-03, grad_scale: 16.0 2023-04-16 21:47:57,601 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39541.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:48:35,449 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9828, 4.1725, 2.8240, 3.8845, 4.0120, 2.5079, 3.5034, 2.8269], device='cuda:0'), covar=tensor([0.0333, 0.0287, 0.1082, 0.0381, 0.0259, 0.1420, 0.0618, 0.1315], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0159, 0.0174, 0.0175, 0.0129, 0.0159, 0.0154, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:48:40,251 INFO [train.py:893] (0/4) Epoch 15, batch 1800, loss[loss=0.181, simple_loss=0.2422, pruned_loss=0.05992, over 13471.00 frames. ], tot_loss[loss=0.1946, simple_loss=0.251, pruned_loss=0.06907, over 2663487.59 frames. ], batch size: 100, lr: 8.72e-03, grad_scale: 16.0 2023-04-16 21:49:01,671 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9043, 3.8870, 2.8199, 3.6072, 2.9549, 2.2618, 3.8801, 2.1464], device='cuda:0'), covar=tensor([0.0663, 0.0325, 0.0548, 0.0265, 0.0686, 0.1761, 0.0647, 0.1365], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0129, 0.0124, 0.0108, 0.0144, 0.0179, 0.0154, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:49:04,641 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.825e+02 2.614e+02 3.171e+02 3.838e+02 5.993e+02, threshold=6.341e+02, percent-clipped=0.0 2023-04-16 21:49:10,025 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0562, 4.5899, 4.3755, 4.3121, 4.3673, 4.1421, 4.6492, 4.6598], device='cuda:0'), covar=tensor([0.0189, 0.0200, 0.0191, 0.0291, 0.0256, 0.0281, 0.0248, 0.0229], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0181, 0.0139, 0.0163, 0.0128, 0.0178, 0.0119, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 21:49:28,272 INFO [train.py:893] (0/4) Epoch 15, batch 1850, loss[loss=0.2107, simple_loss=0.2564, pruned_loss=0.08245, over 12066.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.2499, pruned_loss=0.06862, over 2656741.54 frames. ], batch size: 157, lr: 8.71e-03, grad_scale: 16.0 2023-04-16 21:49:29,413 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39639.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 21:49:30,128 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39640.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:49:30,880 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 21:49:38,695 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2023-04-16 21:50:00,780 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9271, 1.6891, 3.3826, 3.4003, 3.3823, 2.4963, 3.1596, 2.6418], device='cuda:0'), covar=tensor([0.2178, 0.1894, 0.0149, 0.0149, 0.0323, 0.0913, 0.0257, 0.0832], device='cuda:0'), in_proj_covar=tensor([0.0207, 0.0180, 0.0112, 0.0120, 0.0121, 0.0169, 0.0131, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:50:14,103 INFO [train.py:893] (0/4) Epoch 15, batch 1900, loss[loss=0.1637, simple_loss=0.2196, pruned_loss=0.05385, over 13353.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2505, pruned_loss=0.06951, over 2654672.13 frames. ], batch size: 67, lr: 8.71e-03, grad_scale: 16.0 2023-04-16 21:50:27,053 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39700.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 21:50:38,874 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.021e+02 2.868e+02 3.320e+02 3.903e+02 8.982e+02, threshold=6.640e+02, percent-clipped=2.0 2023-04-16 21:51:01,186 INFO [train.py:893] (0/4) Epoch 15, batch 1950, loss[loss=0.1842, simple_loss=0.2398, pruned_loss=0.06433, over 13378.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2495, pruned_loss=0.06896, over 2658196.58 frames. ], batch size: 118, lr: 8.70e-03, grad_scale: 16.0 2023-04-16 21:51:29,365 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39767.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:51:36,447 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9856, 2.5079, 2.4590, 2.9670, 2.2719, 3.0085, 2.9960, 2.5771], device='cuda:0'), covar=tensor([0.0092, 0.0171, 0.0144, 0.0126, 0.0180, 0.0119, 0.0147, 0.0133], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0096, 0.0103, 0.0099, 0.0111, 0.0088, 0.0091, 0.0091], device='cuda:0'), out_proj_covar=tensor([9.4133e-05, 1.0514e-04, 1.1424e-04, 1.0902e-04, 1.2314e-04, 9.4711e-05, 9.9976e-05, 9.7552e-05], device='cuda:0') 2023-04-16 21:51:49,298 INFO [train.py:893] (0/4) Epoch 15, batch 2000, loss[loss=0.1852, simple_loss=0.2342, pruned_loss=0.06813, over 13204.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2498, pruned_loss=0.06877, over 2659542.11 frames. ], batch size: 58, lr: 8.69e-03, grad_scale: 16.0 2023-04-16 21:51:55,277 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 21:52:12,887 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.198e+02 2.857e+02 3.261e+02 3.989e+02 6.795e+02, threshold=6.522e+02, percent-clipped=2.0 2023-04-16 21:52:14,824 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39815.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:52:19,771 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39819.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:52:27,323 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39828.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:52:35,369 INFO [train.py:893] (0/4) Epoch 15, batch 2050, loss[loss=0.2067, simple_loss=0.264, pruned_loss=0.07472, over 13519.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2524, pruned_loss=0.07043, over 2658800.80 frames. ], batch size: 85, lr: 8.69e-03, grad_scale: 16.0 2023-04-16 21:52:38,864 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39841.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:53:11,071 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39876.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:53:14,137 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2088, 5.0352, 5.2342, 5.0014, 5.5145, 5.0514, 5.5291, 5.4803], device='cuda:0'), covar=tensor([0.0303, 0.0527, 0.0627, 0.0527, 0.0491, 0.0849, 0.0442, 0.0470], device='cuda:0'), in_proj_covar=tensor([0.0249, 0.0265, 0.0264, 0.0203, 0.0381, 0.0305, 0.0241, 0.0262], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:53:23,137 INFO [train.py:893] (0/4) Epoch 15, batch 2100, loss[loss=0.2204, simple_loss=0.2694, pruned_loss=0.08569, over 11903.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.2514, pruned_loss=0.06972, over 2660114.69 frames. ], batch size: 157, lr: 8.68e-03, grad_scale: 16.0 2023-04-16 21:53:23,492 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3512, 2.0550, 2.0297, 2.4379, 1.7900, 2.3969, 2.3668, 2.0186], device='cuda:0'), covar=tensor([0.0107, 0.0211, 0.0140, 0.0140, 0.0225, 0.0133, 0.0170, 0.0174], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0096, 0.0103, 0.0099, 0.0111, 0.0088, 0.0090, 0.0090], device='cuda:0'), out_proj_covar=tensor([9.3692e-05, 1.0432e-04, 1.1438e-04, 1.0828e-04, 1.2312e-04, 9.4606e-05, 9.8975e-05, 9.6372e-05], device='cuda:0') 2023-04-16 21:53:24,224 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39889.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:53:35,976 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5840, 2.2960, 2.1391, 2.7137, 1.8988, 2.6558, 2.6624, 2.1815], device='cuda:0'), covar=tensor([0.0095, 0.0164, 0.0146, 0.0112, 0.0207, 0.0127, 0.0152, 0.0181], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0095, 0.0103, 0.0098, 0.0110, 0.0088, 0.0090, 0.0090], device='cuda:0'), out_proj_covar=tensor([9.3526e-05, 1.0406e-04, 1.1396e-04, 1.0793e-04, 1.2272e-04, 9.4561e-05, 9.8732e-05, 9.6667e-05], device='cuda:0') 2023-04-16 21:53:48,113 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.102e+02 2.822e+02 3.258e+02 3.975e+02 7.473e+02, threshold=6.516e+02, percent-clipped=2.0 2023-04-16 21:53:51,261 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-16 21:54:09,167 INFO [train.py:893] (0/4) Epoch 15, batch 2150, loss[loss=0.1988, simple_loss=0.2558, pruned_loss=0.0709, over 13547.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.2512, pruned_loss=0.0693, over 2661204.06 frames. ], batch size: 89, lr: 8.68e-03, grad_scale: 16.0 2023-04-16 21:54:11,008 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39940.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:54:54,913 INFO [train.py:893] (0/4) Epoch 15, batch 2200, loss[loss=0.181, simple_loss=0.2383, pruned_loss=0.06185, over 13260.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2502, pruned_loss=0.06876, over 2662655.90 frames. ], batch size: 124, lr: 8.67e-03, grad_scale: 32.0 2023-04-16 21:54:55,128 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39988.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:55:01,843 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39995.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 21:55:06,385 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-40000.pt 2023-04-16 21:55:24,251 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.868e+02 2.683e+02 3.225e+02 3.898e+02 7.029e+02, threshold=6.450e+02, percent-clipped=1.0 2023-04-16 21:55:46,742 INFO [train.py:893] (0/4) Epoch 15, batch 2250, loss[loss=0.1768, simple_loss=0.2308, pruned_loss=0.06141, over 13478.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2497, pruned_loss=0.06889, over 2662338.72 frames. ], batch size: 100, lr: 8.67e-03, grad_scale: 32.0 2023-04-16 21:56:06,429 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-16 21:56:16,996 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0334, 4.1983, 2.7311, 3.9725, 3.9746, 2.6644, 3.6381, 2.6619], device='cuda:0'), covar=tensor([0.0282, 0.0226, 0.1140, 0.0321, 0.0285, 0.1083, 0.0471, 0.1374], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0156, 0.0172, 0.0173, 0.0128, 0.0156, 0.0152, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 21:56:32,582 INFO [train.py:893] (0/4) Epoch 15, batch 2300, loss[loss=0.1993, simple_loss=0.2543, pruned_loss=0.07214, over 13455.00 frames. ], tot_loss[loss=0.1924, simple_loss=0.2482, pruned_loss=0.06826, over 2662656.76 frames. ], batch size: 100, lr: 8.66e-03, grad_scale: 32.0 2023-04-16 21:56:56,944 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.950e+02 2.485e+02 3.043e+02 3.886e+02 6.974e+02, threshold=6.085e+02, percent-clipped=1.0 2023-04-16 21:57:01,358 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40119.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:57:05,550 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40123.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:57:19,157 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4351, 3.4253, 4.0963, 2.9605, 2.6420, 2.7600, 4.3124, 4.4277], device='cuda:0'), covar=tensor([0.1140, 0.1574, 0.0340, 0.1566, 0.1604, 0.1412, 0.0277, 0.0185], device='cuda:0'), in_proj_covar=tensor([0.0232, 0.0250, 0.0181, 0.0216, 0.0209, 0.0176, 0.0188, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 21:57:20,492 INFO [train.py:893] (0/4) Epoch 15, batch 2350, loss[loss=0.1717, simple_loss=0.2357, pruned_loss=0.05385, over 13530.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2482, pruned_loss=0.06811, over 2661983.63 frames. ], batch size: 76, lr: 8.66e-03, grad_scale: 32.0 2023-04-16 21:57:30,223 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.18 vs. limit=5.0 2023-04-16 21:57:43,077 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 21:57:48,249 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40167.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:57:51,600 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40171.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:58:06,584 INFO [train.py:893] (0/4) Epoch 15, batch 2400, loss[loss=0.1734, simple_loss=0.2393, pruned_loss=0.05376, over 13539.00 frames. ], tot_loss[loss=0.1915, simple_loss=0.2472, pruned_loss=0.0679, over 2661163.05 frames. ], batch size: 72, lr: 8.65e-03, grad_scale: 32.0 2023-04-16 21:58:13,694 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40195.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:58:30,547 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.938e+02 2.754e+02 3.113e+02 3.687e+02 7.218e+02, threshold=6.226e+02, percent-clipped=3.0 2023-04-16 21:58:35,163 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40218.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:58:54,052 INFO [train.py:893] (0/4) Epoch 15, batch 2450, loss[loss=0.1985, simple_loss=0.2551, pruned_loss=0.07096, over 13231.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.2466, pruned_loss=0.06757, over 2662661.41 frames. ], batch size: 124, lr: 8.65e-03, grad_scale: 32.0 2023-04-16 21:59:11,062 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40256.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:59:26,274 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40272.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:59:32,109 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40279.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 21:59:36,152 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9725, 4.3174, 4.0429, 4.1398, 4.1482, 4.5042, 4.2619, 4.1982], device='cuda:0'), covar=tensor([0.0345, 0.0262, 0.0370, 0.0889, 0.0273, 0.0239, 0.0325, 0.0297], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0140, 0.0156, 0.0249, 0.0158, 0.0173, 0.0156, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2023-04-16 21:59:38,716 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2148, 2.6180, 2.1030, 4.1244, 4.6477, 3.3961, 4.4778, 4.2939], device='cuda:0'), covar=tensor([0.0107, 0.0870, 0.1058, 0.0102, 0.0056, 0.0444, 0.0089, 0.0081], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0086, 0.0095, 0.0076, 0.0061, 0.0078, 0.0052, 0.0071], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 21:59:40,943 INFO [train.py:893] (0/4) Epoch 15, batch 2500, loss[loss=0.212, simple_loss=0.2683, pruned_loss=0.07787, over 13482.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.2464, pruned_loss=0.06733, over 2664794.04 frames. ], batch size: 100, lr: 8.64e-03, grad_scale: 32.0 2023-04-16 21:59:47,749 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40295.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 21:59:49,826 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-16 21:59:50,637 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.23 vs. limit=5.0 2023-04-16 21:59:53,885 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0651, 3.4771, 3.2923, 3.7637, 2.0312, 2.9400, 3.5716, 2.0424], device='cuda:0'), covar=tensor([0.0245, 0.0469, 0.0657, 0.0365, 0.1529, 0.0848, 0.0557, 0.1767], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0177, 0.0194, 0.0221, 0.0178, 0.0192, 0.0173, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 22:00:05,316 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.732e+02 2.683e+02 3.082e+02 3.857e+02 6.124e+02, threshold=6.165e+02, percent-clipped=0.0 2023-04-16 22:00:05,786 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-16 22:00:21,432 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40330.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:00:24,002 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40333.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:00:27,912 INFO [train.py:893] (0/4) Epoch 15, batch 2550, loss[loss=0.1767, simple_loss=0.2472, pruned_loss=0.05305, over 13374.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.247, pruned_loss=0.06743, over 2661278.09 frames. ], batch size: 113, lr: 8.64e-03, grad_scale: 32.0 2023-04-16 22:00:33,040 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40343.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 22:00:53,362 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 22:01:03,726 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6580, 3.7344, 4.4743, 3.1517, 3.0678, 2.9906, 4.6754, 4.7906], device='cuda:0'), covar=tensor([0.1034, 0.1290, 0.0266, 0.1387, 0.1186, 0.1451, 0.0253, 0.0175], device='cuda:0'), in_proj_covar=tensor([0.0231, 0.0251, 0.0181, 0.0215, 0.0209, 0.0177, 0.0187, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:01:15,765 INFO [train.py:893] (0/4) Epoch 15, batch 2600, loss[loss=0.1951, simple_loss=0.2394, pruned_loss=0.07542, over 12219.00 frames. ], tot_loss[loss=0.1906, simple_loss=0.2462, pruned_loss=0.0675, over 2654634.92 frames. ], batch size: 49, lr: 8.63e-03, grad_scale: 32.0 2023-04-16 22:01:18,758 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40391.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:01:39,452 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.061e+02 2.820e+02 3.308e+02 3.713e+02 1.179e+03, threshold=6.616e+02, percent-clipped=5.0 2023-04-16 22:01:46,538 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40423.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:01:58,692 INFO [train.py:893] (0/4) Epoch 15, batch 2650, loss[loss=0.2039, simple_loss=0.2393, pruned_loss=0.08424, over 10910.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2475, pruned_loss=0.06883, over 2653155.27 frames. ], batch size: 44, lr: 8.63e-03, grad_scale: 32.0 2023-04-16 22:02:24,408 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40471.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:02:24,474 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40471.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:02:38,073 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-15.pt 2023-04-16 22:03:03,177 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 22:03:07,420 INFO [train.py:893] (0/4) Epoch 16, batch 0, loss[loss=0.2104, simple_loss=0.2552, pruned_loss=0.08284, over 13371.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2552, pruned_loss=0.08284, over 13371.00 frames. ], batch size: 109, lr: 8.34e-03, grad_scale: 32.0 2023-04-16 22:03:07,421 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 22:03:25,038 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6639, 2.3685, 2.3001, 2.7842, 2.0740, 2.8165, 2.7580, 2.2076], device='cuda:0'), covar=tensor([0.0114, 0.0195, 0.0152, 0.0168, 0.0226, 0.0119, 0.0195, 0.0204], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0099, 0.0106, 0.0100, 0.0114, 0.0089, 0.0092, 0.0091], device='cuda:0'), out_proj_covar=tensor([9.5390e-05, 1.0749e-04, 1.1806e-04, 1.0956e-04, 1.2664e-04, 9.5101e-05, 1.0094e-04, 9.7093e-05], device='cuda:0') 2023-04-16 22:03:27,267 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8358, 2.4842, 2.4090, 2.9475, 2.2801, 3.0107, 2.8693, 2.4337], device='cuda:0'), covar=tensor([0.0157, 0.0262, 0.0181, 0.0141, 0.0239, 0.0142, 0.0200, 0.0205], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0099, 0.0106, 0.0100, 0.0114, 0.0089, 0.0092, 0.0091], device='cuda:0'), out_proj_covar=tensor([9.5390e-05, 1.0749e-04, 1.1806e-04, 1.0956e-04, 1.2664e-04, 9.5101e-05, 1.0094e-04, 9.7093e-05], device='cuda:0') 2023-04-16 22:03:30,246 INFO [train.py:927] (0/4) Epoch 16, validation: loss=0.1448, simple_loss=0.2029, pruned_loss=0.04338, over 2446609.00 frames. 2023-04-16 22:03:30,247 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 22:03:35,675 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40492.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:03:55,233 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.945e+02 2.953e+02 3.389e+02 4.043e+02 6.293e+02, threshold=6.778e+02, percent-clipped=0.0 2023-04-16 22:04:00,503 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40519.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:04:18,044 INFO [train.py:893] (0/4) Epoch 16, batch 50, loss[loss=0.1756, simple_loss=0.2366, pruned_loss=0.05733, over 13548.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2428, pruned_loss=0.06946, over 599095.26 frames. ], batch size: 83, lr: 8.34e-03, grad_scale: 32.0 2023-04-16 22:04:30,704 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40551.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:04:32,476 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40553.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:04:42,914 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 22:04:42,914 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 22:04:42,914 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 22:04:42,921 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 22:04:42,937 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 22:04:42,957 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 22:04:42,966 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 22:04:52,254 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40574.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:05:03,727 INFO [train.py:893] (0/4) Epoch 16, batch 100, loss[loss=0.1859, simple_loss=0.2408, pruned_loss=0.06544, over 13481.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2454, pruned_loss=0.07119, over 1049186.54 frames. ], batch size: 81, lr: 8.33e-03, grad_scale: 32.0 2023-04-16 22:05:11,603 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7830, 3.4062, 2.8493, 3.2565, 2.8310, 1.9782, 3.5268, 2.0345], device='cuda:0'), covar=tensor([0.0645, 0.0787, 0.0426, 0.0417, 0.0672, 0.1991, 0.1043, 0.1378], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0130, 0.0125, 0.0110, 0.0145, 0.0182, 0.0155, 0.0137], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 22:05:28,494 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.136e+02 2.861e+02 3.473e+02 4.004e+02 9.328e+02, threshold=6.946e+02, percent-clipped=3.0 2023-04-16 22:05:41,993 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40628.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:05:50,829 INFO [train.py:893] (0/4) Epoch 16, batch 150, loss[loss=0.2027, simple_loss=0.2548, pruned_loss=0.07534, over 13362.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2484, pruned_loss=0.0719, over 1404393.56 frames. ], batch size: 118, lr: 8.33e-03, grad_scale: 32.0 2023-04-16 22:06:03,576 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40651.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:06:24,603 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-16 22:06:37,182 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40686.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:06:37,819 INFO [train.py:893] (0/4) Epoch 16, batch 200, loss[loss=0.1904, simple_loss=0.2497, pruned_loss=0.06554, over 13560.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2505, pruned_loss=0.0723, over 1680442.37 frames. ], batch size: 78, lr: 8.32e-03, grad_scale: 32.0 2023-04-16 22:06:47,177 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5870, 4.3971, 4.6332, 4.5683, 4.8629, 4.4114, 4.9006, 4.8862], device='cuda:0'), covar=tensor([0.0412, 0.0568, 0.0637, 0.0577, 0.0531, 0.0858, 0.0438, 0.0439], device='cuda:0'), in_proj_covar=tensor([0.0259, 0.0274, 0.0275, 0.0207, 0.0392, 0.0318, 0.0248, 0.0270], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 22:07:01,380 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40712.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:07:02,786 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 2.805e+02 3.442e+02 4.281e+02 7.612e+02, threshold=6.884e+02, percent-clipped=2.0 2023-04-16 22:07:25,033 INFO [train.py:893] (0/4) Epoch 16, batch 250, loss[loss=0.1923, simple_loss=0.2463, pruned_loss=0.06913, over 13502.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2515, pruned_loss=0.07252, over 1896944.71 frames. ], batch size: 93, lr: 8.32e-03, grad_scale: 32.0 2023-04-16 22:07:54,654 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9788, 4.0765, 3.1596, 2.7780, 2.9675, 2.4796, 4.2615, 2.5071], device='cuda:0'), covar=tensor([0.1530, 0.0349, 0.1068, 0.1879, 0.0790, 0.3172, 0.0203, 0.3841], device='cuda:0'), in_proj_covar=tensor([0.0302, 0.0279, 0.0292, 0.0308, 0.0241, 0.0308, 0.0197, 0.0350], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 22:08:08,911 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7339, 2.4291, 2.4198, 2.8054, 2.0694, 2.9025, 2.7672, 2.3550], device='cuda:0'), covar=tensor([0.0090, 0.0216, 0.0169, 0.0163, 0.0232, 0.0114, 0.0165, 0.0169], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0098, 0.0105, 0.0099, 0.0111, 0.0089, 0.0091, 0.0090], device='cuda:0'), out_proj_covar=tensor([9.5484e-05, 1.0662e-04, 1.1626e-04, 1.0853e-04, 1.2325e-04, 9.5489e-05, 9.8939e-05, 9.6407e-05], device='cuda:0') 2023-04-16 22:08:12,651 INFO [train.py:893] (0/4) Epoch 16, batch 300, loss[loss=0.225, simple_loss=0.2711, pruned_loss=0.0894, over 13232.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2526, pruned_loss=0.0726, over 2068761.44 frames. ], batch size: 124, lr: 8.31e-03, grad_scale: 32.0 2023-04-16 22:08:38,222 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.958e+02 2.972e+02 3.495e+02 4.194e+02 5.977e+02, threshold=6.991e+02, percent-clipped=0.0 2023-04-16 22:08:59,500 INFO [train.py:893] (0/4) Epoch 16, batch 350, loss[loss=0.2015, simple_loss=0.2589, pruned_loss=0.07204, over 13375.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2522, pruned_loss=0.07251, over 2197031.56 frames. ], batch size: 118, lr: 8.31e-03, grad_scale: 32.0 2023-04-16 22:09:01,467 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6196, 4.0580, 4.1211, 4.1782, 3.8664, 3.9048, 4.5552, 4.0538], device='cuda:0'), covar=tensor([0.0804, 0.1403, 0.2338, 0.2471, 0.1106, 0.1701, 0.1083, 0.1350], device='cuda:0'), in_proj_covar=tensor([0.0307, 0.0354, 0.0443, 0.0456, 0.0265, 0.0336, 0.0404, 0.0324], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 22:09:10,608 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40848.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:09:13,169 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40851.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:09:25,848 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40865.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:09:34,223 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40874.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:09:47,328 INFO [train.py:893] (0/4) Epoch 16, batch 400, loss[loss=0.1893, simple_loss=0.2523, pruned_loss=0.06314, over 13481.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2526, pruned_loss=0.07191, over 2303359.70 frames. ], batch size: 93, lr: 8.30e-03, grad_scale: 32.0 2023-04-16 22:09:58,473 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40899.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:09:59,391 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40900.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:10:12,108 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.876e+02 2.703e+02 3.218e+02 3.995e+02 8.871e+02, threshold=6.435e+02, percent-clipped=1.0 2023-04-16 22:10:19,932 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40922.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:10:23,569 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40926.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:10:25,258 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40928.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:10:34,480 INFO [train.py:893] (0/4) Epoch 16, batch 450, loss[loss=0.1714, simple_loss=0.2372, pruned_loss=0.05284, over 13443.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2539, pruned_loss=0.07221, over 2381262.95 frames. ], batch size: 100, lr: 8.30e-03, grad_scale: 32.0 2023-04-16 22:10:53,869 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4237, 2.5173, 2.8442, 4.0780, 3.6275, 4.1146, 3.1778, 2.2970], device='cuda:0'), covar=tensor([0.0265, 0.0941, 0.0760, 0.0041, 0.0253, 0.0040, 0.0663, 0.1031], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0146, 0.0158, 0.0087, 0.0112, 0.0084, 0.0163, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0001, 0.0003, 0.0003], device='cuda:0') 2023-04-16 22:10:56,439 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40961.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 22:10:59,393 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 22:11:10,913 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40976.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:11:20,165 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40986.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:11:20,759 INFO [train.py:893] (0/4) Epoch 16, batch 500, loss[loss=0.1645, simple_loss=0.211, pruned_loss=0.05895, over 13194.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2541, pruned_loss=0.07219, over 2441995.80 frames. ], batch size: 58, lr: 8.29e-03, grad_scale: 32.0 2023-04-16 22:11:30,053 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40997.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:11:40,284 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41007.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:11:46,651 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.017e+02 2.721e+02 3.359e+02 3.870e+02 6.391e+02, threshold=6.718e+02, percent-clipped=0.0 2023-04-16 22:12:01,445 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-16 22:12:05,234 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41034.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:12:08,067 INFO [train.py:893] (0/4) Epoch 16, batch 550, loss[loss=0.2497, simple_loss=0.2986, pruned_loss=0.1004, over 13367.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2539, pruned_loss=0.07223, over 2490353.06 frames. ], batch size: 118, lr: 8.29e-03, grad_scale: 32.0 2023-04-16 22:12:15,859 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41045.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:12:27,811 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41058.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:12:39,417 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-16 22:12:54,708 INFO [train.py:893] (0/4) Epoch 16, batch 600, loss[loss=0.1747, simple_loss=0.2338, pruned_loss=0.05783, over 13336.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2516, pruned_loss=0.07089, over 2533462.97 frames. ], batch size: 67, lr: 8.28e-03, grad_scale: 32.0 2023-04-16 22:13:09,392 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7268, 3.5160, 2.7231, 3.1041, 2.9108, 1.9434, 3.4458, 2.1004], device='cuda:0'), covar=tensor([0.0680, 0.0510, 0.0498, 0.0456, 0.0678, 0.2040, 0.0996, 0.1339], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0128, 0.0124, 0.0108, 0.0143, 0.0178, 0.0154, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:13:09,397 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41102.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:13:12,610 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41106.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:13:12,682 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5065, 3.8033, 3.6472, 4.2047, 2.2677, 3.2429, 3.9461, 2.2089], device='cuda:0'), covar=tensor([0.0114, 0.0503, 0.0731, 0.0596, 0.1531, 0.0846, 0.0549, 0.1868], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0178, 0.0196, 0.0224, 0.0178, 0.0193, 0.0173, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 22:13:17,752 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1537, 2.6973, 2.1160, 4.0740, 4.6164, 3.3413, 4.4408, 4.2327], device='cuda:0'), covar=tensor([0.0098, 0.0802, 0.0971, 0.0093, 0.0052, 0.0463, 0.0075, 0.0078], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0085, 0.0094, 0.0075, 0.0060, 0.0076, 0.0051, 0.0071], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 22:13:19,902 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.056e+02 2.815e+02 3.188e+02 4.197e+02 8.444e+02, threshold=6.376e+02, percent-clipped=3.0 2023-04-16 22:13:31,142 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8890, 4.4032, 4.3095, 4.3961, 4.1252, 4.1548, 4.8606, 4.4195], device='cuda:0'), covar=tensor([0.0754, 0.1139, 0.2119, 0.2762, 0.1094, 0.1758, 0.0989, 0.1209], device='cuda:0'), in_proj_covar=tensor([0.0309, 0.0352, 0.0446, 0.0455, 0.0268, 0.0340, 0.0404, 0.0322], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 22:13:41,397 INFO [train.py:893] (0/4) Epoch 16, batch 650, loss[loss=0.1756, simple_loss=0.2407, pruned_loss=0.05522, over 13449.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2505, pruned_loss=0.07029, over 2565360.50 frames. ], batch size: 106, lr: 8.28e-03, grad_scale: 32.0 2023-04-16 22:13:51,582 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41148.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:14:05,864 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41163.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:14:28,918 INFO [train.py:893] (0/4) Epoch 16, batch 700, loss[loss=0.1766, simple_loss=0.2373, pruned_loss=0.05795, over 13242.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2495, pruned_loss=0.06952, over 2586644.91 frames. ], batch size: 132, lr: 8.27e-03, grad_scale: 32.0 2023-04-16 22:14:37,421 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41196.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:14:54,222 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.098e+02 2.785e+02 3.219e+02 3.700e+02 6.141e+02, threshold=6.437e+02, percent-clipped=0.0 2023-04-16 22:15:01,289 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41221.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:15:16,662 INFO [train.py:893] (0/4) Epoch 16, batch 750, loss[loss=0.1826, simple_loss=0.2357, pruned_loss=0.0648, over 13413.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.2498, pruned_loss=0.07029, over 2601692.04 frames. ], batch size: 65, lr: 8.27e-03, grad_scale: 32.0 2023-04-16 22:15:34,485 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41256.0, num_to_drop=1, layers_to_drop={3} 2023-04-16 22:16:03,269 INFO [train.py:893] (0/4) Epoch 16, batch 800, loss[loss=0.2143, simple_loss=0.2706, pruned_loss=0.07898, over 13567.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2513, pruned_loss=0.07089, over 2617971.79 frames. ], batch size: 89, lr: 8.26e-03, grad_scale: 32.0 2023-04-16 22:16:22,761 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41307.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:16:29,240 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.237e+02 2.904e+02 3.409e+02 3.941e+02 8.201e+02, threshold=6.817e+02, percent-clipped=2.0 2023-04-16 22:16:51,226 INFO [train.py:893] (0/4) Epoch 16, batch 850, loss[loss=0.1885, simple_loss=0.2466, pruned_loss=0.06518, over 13527.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2516, pruned_loss=0.07073, over 2628641.52 frames. ], batch size: 98, lr: 8.26e-03, grad_scale: 32.0 2023-04-16 22:17:06,366 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41353.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:17:08,121 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41355.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:17:36,898 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.77 vs. limit=5.0 2023-04-16 22:17:38,108 INFO [train.py:893] (0/4) Epoch 16, batch 900, loss[loss=0.1993, simple_loss=0.2534, pruned_loss=0.07257, over 13498.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.252, pruned_loss=0.07143, over 2636602.22 frames. ], batch size: 93, lr: 8.25e-03, grad_scale: 32.0 2023-04-16 22:17:38,419 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0295, 1.8987, 3.7370, 3.6202, 3.5862, 2.8227, 3.4401, 2.7132], device='cuda:0'), covar=tensor([0.1897, 0.1536, 0.0115, 0.0147, 0.0188, 0.0711, 0.0234, 0.0817], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0181, 0.0111, 0.0120, 0.0122, 0.0170, 0.0133, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:17:52,143 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41401.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:18:03,454 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.089e+02 3.038e+02 3.486e+02 4.012e+02 7.844e+02, threshold=6.973e+02, percent-clipped=1.0 2023-04-16 22:18:09,453 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5768, 3.2939, 2.7301, 3.0385, 2.7520, 1.9670, 3.3526, 1.8047], device='cuda:0'), covar=tensor([0.0702, 0.0671, 0.0496, 0.0444, 0.0693, 0.1908, 0.0887, 0.1410], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0128, 0.0123, 0.0106, 0.0142, 0.0177, 0.0154, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:18:09,954 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 22:18:24,931 INFO [train.py:893] (0/4) Epoch 16, batch 950, loss[loss=0.2061, simple_loss=0.2633, pruned_loss=0.07448, over 13366.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2507, pruned_loss=0.07132, over 2641019.21 frames. ], batch size: 109, lr: 8.25e-03, grad_scale: 32.0 2023-04-16 22:18:31,554 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-04-16 22:18:45,287 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41458.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:19:11,655 INFO [train.py:893] (0/4) Epoch 16, batch 1000, loss[loss=0.1768, simple_loss=0.2379, pruned_loss=0.05782, over 13550.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.2489, pruned_loss=0.07059, over 2639914.00 frames. ], batch size: 78, lr: 8.24e-03, grad_scale: 32.0 2023-04-16 22:19:37,200 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.105e+02 2.855e+02 3.387e+02 3.992e+02 7.617e+02, threshold=6.774e+02, percent-clipped=2.0 2023-04-16 22:19:44,058 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41521.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:19:59,066 INFO [train.py:893] (0/4) Epoch 16, batch 1050, loss[loss=0.1886, simple_loss=0.241, pruned_loss=0.06815, over 13362.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2478, pruned_loss=0.06974, over 2646839.68 frames. ], batch size: 73, lr: 8.24e-03, grad_scale: 32.0 2023-04-16 22:20:16,853 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41556.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 22:20:29,124 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41569.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:20:38,407 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.12 vs. limit=5.0 2023-04-16 22:20:45,605 INFO [train.py:893] (0/4) Epoch 16, batch 1100, loss[loss=0.2023, simple_loss=0.2566, pruned_loss=0.07396, over 13367.00 frames. ], tot_loss[loss=0.1934, simple_loss=0.2481, pruned_loss=0.06934, over 2651140.88 frames. ], batch size: 113, lr: 8.23e-03, grad_scale: 32.0 2023-04-16 22:21:02,268 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41604.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:21:11,238 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.019e+02 2.712e+02 3.141e+02 3.905e+02 7.780e+02, threshold=6.282e+02, percent-clipped=1.0 2023-04-16 22:21:34,295 INFO [train.py:893] (0/4) Epoch 16, batch 1150, loss[loss=0.203, simple_loss=0.2683, pruned_loss=0.06891, over 13385.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2483, pruned_loss=0.06883, over 2651224.70 frames. ], batch size: 113, lr: 8.23e-03, grad_scale: 32.0 2023-04-16 22:21:48,780 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41653.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:21:49,648 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41654.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:21:55,681 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9134, 4.0100, 3.2189, 2.8401, 2.8285, 2.4964, 4.2332, 2.4026], device='cuda:0'), covar=tensor([0.1463, 0.0342, 0.0967, 0.1827, 0.0810, 0.2925, 0.0208, 0.3700], device='cuda:0'), in_proj_covar=tensor([0.0305, 0.0281, 0.0297, 0.0310, 0.0245, 0.0311, 0.0202, 0.0354], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 22:22:03,050 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6452, 2.3132, 2.2821, 2.6676, 1.9007, 2.6542, 2.5858, 2.2476], device='cuda:0'), covar=tensor([0.0079, 0.0165, 0.0163, 0.0139, 0.0217, 0.0133, 0.0181, 0.0176], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0098, 0.0108, 0.0102, 0.0113, 0.0092, 0.0092, 0.0092], device='cuda:0'), out_proj_covar=tensor([9.7392e-05, 1.0651e-04, 1.1987e-04, 1.1061e-04, 1.2533e-04, 9.9050e-05, 1.0016e-04, 9.8652e-05], device='cuda:0') 2023-04-16 22:22:20,420 INFO [train.py:893] (0/4) Epoch 16, batch 1200, loss[loss=0.195, simple_loss=0.2418, pruned_loss=0.07409, over 13529.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2485, pruned_loss=0.06862, over 2657202.36 frames. ], batch size: 83, lr: 8.23e-03, grad_scale: 16.0 2023-04-16 22:22:34,022 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41701.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:22:34,077 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41701.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:22:46,472 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.687e+02 2.840e+02 3.425e+02 4.080e+02 1.396e+03, threshold=6.851e+02, percent-clipped=2.0 2023-04-16 22:22:46,828 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41715.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 22:22:47,457 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 22:22:59,789 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 22:23:06,408 INFO [train.py:893] (0/4) Epoch 16, batch 1250, loss[loss=0.1979, simple_loss=0.2525, pruned_loss=0.07162, over 13558.00 frames. ], tot_loss[loss=0.1934, simple_loss=0.2488, pruned_loss=0.06896, over 2657534.12 frames. ], batch size: 87, lr: 8.22e-03, grad_scale: 16.0 2023-04-16 22:23:17,052 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.84 vs. limit=5.0 2023-04-16 22:23:17,621 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41749.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:23:26,899 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41758.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:23:30,261 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0244, 2.0690, 3.7793, 3.6887, 3.6902, 2.7906, 3.5290, 2.8282], device='cuda:0'), covar=tensor([0.1860, 0.1361, 0.0131, 0.0130, 0.0153, 0.0799, 0.0216, 0.0729], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0182, 0.0112, 0.0119, 0.0124, 0.0170, 0.0133, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:23:44,564 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-04-16 22:23:52,546 INFO [train.py:893] (0/4) Epoch 16, batch 1300, loss[loss=0.185, simple_loss=0.2502, pruned_loss=0.05987, over 13492.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.25, pruned_loss=0.06948, over 2660980.86 frames. ], batch size: 81, lr: 8.22e-03, grad_scale: 16.0 2023-04-16 22:24:10,613 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.4150, 4.8838, 4.7074, 4.8855, 4.9031, 4.6932, 5.3990, 4.9308], device='cuda:0'), covar=tensor([0.0659, 0.1158, 0.2345, 0.2431, 0.0702, 0.1492, 0.0810, 0.1030], device='cuda:0'), in_proj_covar=tensor([0.0302, 0.0346, 0.0436, 0.0444, 0.0266, 0.0331, 0.0399, 0.0318], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 22:24:11,501 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41806.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:24:18,909 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.763e+02 2.850e+02 3.302e+02 3.897e+02 7.866e+02, threshold=6.605e+02, percent-clipped=2.0 2023-04-16 22:24:40,586 INFO [train.py:893] (0/4) Epoch 16, batch 1350, loss[loss=0.1761, simple_loss=0.2376, pruned_loss=0.05726, over 13531.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.2506, pruned_loss=0.06954, over 2664692.88 frames. ], batch size: 76, lr: 8.21e-03, grad_scale: 16.0 2023-04-16 22:25:27,115 INFO [train.py:893] (0/4) Epoch 16, batch 1400, loss[loss=0.2113, simple_loss=0.2616, pruned_loss=0.0805, over 13430.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.25, pruned_loss=0.0694, over 2660769.22 frames. ], batch size: 95, lr: 8.21e-03, grad_scale: 16.0 2023-04-16 22:25:45,089 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41906.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:25:53,693 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.042e+02 2.927e+02 3.297e+02 3.876e+02 1.002e+03, threshold=6.593e+02, percent-clipped=4.0 2023-04-16 22:26:13,623 INFO [train.py:893] (0/4) Epoch 16, batch 1450, loss[loss=0.1856, simple_loss=0.2466, pruned_loss=0.06226, over 13523.00 frames. ], tot_loss[loss=0.1942, simple_loss=0.2495, pruned_loss=0.06943, over 2663766.67 frames. ], batch size: 98, lr: 8.20e-03, grad_scale: 16.0 2023-04-16 22:26:24,768 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41948.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 22:26:33,326 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3662, 3.5530, 3.3948, 3.9820, 2.1781, 2.8820, 3.7252, 2.1713], device='cuda:0'), covar=tensor([0.0120, 0.0472, 0.0849, 0.0558, 0.1602, 0.1029, 0.0616, 0.1867], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0178, 0.0196, 0.0224, 0.0178, 0.0192, 0.0174, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 22:26:42,371 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41967.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:27:00,187 INFO [train.py:893] (0/4) Epoch 16, batch 1500, loss[loss=0.2067, simple_loss=0.2588, pruned_loss=0.07733, over 13523.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2482, pruned_loss=0.0688, over 2660759.74 frames. ], batch size: 91, lr: 8.20e-03, grad_scale: 16.0 2023-04-16 22:27:13,117 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-42000.pt 2023-04-16 22:27:23,264 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0875, 2.8878, 2.5779, 3.1230, 2.5231, 3.2505, 3.1244, 2.7634], device='cuda:0'), covar=tensor([0.0069, 0.0128, 0.0125, 0.0131, 0.0149, 0.0104, 0.0140, 0.0138], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0097, 0.0107, 0.0102, 0.0113, 0.0091, 0.0093, 0.0092], device='cuda:0'), out_proj_covar=tensor([9.7574e-05, 1.0541e-04, 1.1880e-04, 1.1132e-04, 1.2464e-04, 9.7670e-05, 1.0078e-04, 9.8038e-05], device='cuda:0') 2023-04-16 22:27:26,463 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42009.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 22:27:27,135 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42010.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 22:27:31,080 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.141e+02 2.694e+02 3.135e+02 3.781e+02 7.515e+02, threshold=6.271e+02, percent-clipped=2.0 2023-04-16 22:27:52,309 INFO [train.py:893] (0/4) Epoch 16, batch 1550, loss[loss=0.2001, simple_loss=0.26, pruned_loss=0.07006, over 13379.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2485, pruned_loss=0.06838, over 2663895.09 frames. ], batch size: 109, lr: 8.19e-03, grad_scale: 16.0 2023-04-16 22:28:32,339 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-16 22:28:39,478 INFO [train.py:893] (0/4) Epoch 16, batch 1600, loss[loss=0.1874, simple_loss=0.2458, pruned_loss=0.06452, over 13529.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2488, pruned_loss=0.06823, over 2660278.27 frames. ], batch size: 83, lr: 8.19e-03, grad_scale: 16.0 2023-04-16 22:28:56,701 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9751, 1.9003, 3.9403, 3.7944, 3.7957, 3.0881, 3.6586, 2.9320], device='cuda:0'), covar=tensor([0.2104, 0.1583, 0.0104, 0.0195, 0.0215, 0.0623, 0.0216, 0.0776], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0180, 0.0111, 0.0118, 0.0123, 0.0167, 0.0131, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:29:05,744 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.862e+02 2.671e+02 3.172e+02 3.980e+02 8.718e+02, threshold=6.343e+02, percent-clipped=3.0 2023-04-16 22:29:09,703 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2023-04-16 22:29:25,414 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.1955, 2.1166, 4.2110, 3.8553, 4.1030, 3.2198, 3.8591, 3.0660], device='cuda:0'), covar=tensor([0.1853, 0.1555, 0.0099, 0.0186, 0.0172, 0.0576, 0.0192, 0.0754], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0183, 0.0112, 0.0119, 0.0124, 0.0170, 0.0133, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:29:27,689 INFO [train.py:893] (0/4) Epoch 16, batch 1650, loss[loss=0.1542, simple_loss=0.2114, pruned_loss=0.04851, over 13524.00 frames. ], tot_loss[loss=0.1916, simple_loss=0.2485, pruned_loss=0.06736, over 2658966.30 frames. ], batch size: 70, lr: 8.18e-03, grad_scale: 16.0 2023-04-16 22:30:14,883 INFO [train.py:893] (0/4) Epoch 16, batch 1700, loss[loss=0.1933, simple_loss=0.2498, pruned_loss=0.06842, over 13537.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2484, pruned_loss=0.06701, over 2661103.14 frames. ], batch size: 87, lr: 8.18e-03, grad_scale: 16.0 2023-04-16 22:30:41,189 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.067e+02 2.979e+02 3.408e+02 4.299e+02 7.415e+02, threshold=6.815e+02, percent-clipped=3.0 2023-04-16 22:30:47,799 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-04-16 22:31:02,236 INFO [train.py:893] (0/4) Epoch 16, batch 1750, loss[loss=0.2017, simple_loss=0.2563, pruned_loss=0.07353, over 13444.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2466, pruned_loss=0.0658, over 2664990.73 frames. ], batch size: 103, lr: 8.17e-03, grad_scale: 16.0 2023-04-16 22:31:05,958 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.0270, 4.8636, 5.1298, 4.9329, 5.3465, 4.8396, 5.3736, 5.3590], device='cuda:0'), covar=tensor([0.0376, 0.0563, 0.0545, 0.0534, 0.0505, 0.0857, 0.0445, 0.0468], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0267, 0.0272, 0.0205, 0.0388, 0.0313, 0.0238, 0.0265], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 22:31:26,452 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42262.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:31:49,973 INFO [train.py:893] (0/4) Epoch 16, batch 1800, loss[loss=0.1932, simple_loss=0.2512, pruned_loss=0.06764, over 13530.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2463, pruned_loss=0.06549, over 2664378.95 frames. ], batch size: 98, lr: 8.17e-03, grad_scale: 16.0 2023-04-16 22:31:53,660 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7689, 4.1956, 4.0134, 3.9889, 4.0155, 3.8523, 4.2491, 4.2840], device='cuda:0'), covar=tensor([0.0261, 0.0227, 0.0212, 0.0340, 0.0290, 0.0292, 0.0264, 0.0277], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0185, 0.0145, 0.0168, 0.0132, 0.0183, 0.0123, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 22:31:59,796 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7224, 2.4667, 1.9875, 1.2916, 1.6337, 1.8968, 2.1014, 2.5224], device='cuda:0'), covar=tensor([0.0825, 0.0285, 0.0801, 0.1615, 0.0167, 0.0529, 0.0720, 0.0268], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0127, 0.0109, 0.0196, 0.0102, 0.0148, 0.0157, 0.0121], device='cuda:0'), out_proj_covar=tensor([1.1582e-04, 9.5451e-05, 8.6846e-05, 1.4843e-04, 7.6891e-05, 1.1218e-04, 1.1983e-04, 8.9957e-05], device='cuda:0') 2023-04-16 22:32:04,642 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42304.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 22:32:10,457 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42310.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 22:32:15,941 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.816e+02 2.666e+02 3.144e+02 3.768e+02 7.198e+02, threshold=6.287e+02, percent-clipped=1.0 2023-04-16 22:32:32,202 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.4303, 2.0156, 4.0419, 3.7971, 3.9362, 3.2031, 3.7623, 2.9258], device='cuda:0'), covar=tensor([0.1439, 0.1472, 0.0098, 0.0179, 0.0164, 0.0537, 0.0167, 0.0727], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0176, 0.0110, 0.0117, 0.0120, 0.0165, 0.0130, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:32:36,100 INFO [train.py:893] (0/4) Epoch 16, batch 1850, loss[loss=0.2162, simple_loss=0.2712, pruned_loss=0.08057, over 13449.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2469, pruned_loss=0.06581, over 2663544.77 frames. ], batch size: 100, lr: 8.16e-03, grad_scale: 16.0 2023-04-16 22:32:39,355 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 22:32:56,283 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=42358.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:33:16,495 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42379.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:33:21,583 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0837, 3.8606, 3.1199, 3.6766, 3.0735, 2.1473, 3.9012, 2.0599], device='cuda:0'), covar=tensor([0.0714, 0.0606, 0.0507, 0.0346, 0.0755, 0.2052, 0.0910, 0.1513], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0131, 0.0127, 0.0108, 0.0145, 0.0180, 0.0157, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 22:33:23,835 INFO [train.py:893] (0/4) Epoch 16, batch 1900, loss[loss=0.1983, simple_loss=0.2507, pruned_loss=0.07289, over 13575.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2468, pruned_loss=0.06639, over 2664998.53 frames. ], batch size: 89, lr: 8.16e-03, grad_scale: 16.0 2023-04-16 22:33:29,840 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0805, 3.8842, 3.0903, 3.7416, 3.1423, 2.1233, 3.8459, 2.1210], device='cuda:0'), covar=tensor([0.0685, 0.0437, 0.0502, 0.0289, 0.0724, 0.2134, 0.1072, 0.1444], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0130, 0.0126, 0.0107, 0.0144, 0.0179, 0.0156, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 22:33:50,546 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.816e+02 3.160e+02 3.935e+02 6.934e+02, threshold=6.320e+02, percent-clipped=1.0 2023-04-16 22:34:09,991 INFO [train.py:893] (0/4) Epoch 16, batch 1950, loss[loss=0.1705, simple_loss=0.228, pruned_loss=0.05647, over 13362.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2455, pruned_loss=0.06614, over 2660763.65 frames. ], batch size: 67, lr: 8.15e-03, grad_scale: 16.0 2023-04-16 22:34:13,712 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42440.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:34:57,436 INFO [train.py:893] (0/4) Epoch 16, batch 2000, loss[loss=0.1644, simple_loss=0.2108, pruned_loss=0.05893, over 12817.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.248, pruned_loss=0.06713, over 2663419.52 frames. ], batch size: 52, lr: 8.15e-03, grad_scale: 16.0 2023-04-16 22:35:05,270 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 22:35:17,096 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42506.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:35:24,432 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.996e+02 2.838e+02 3.154e+02 3.694e+02 1.052e+03, threshold=6.307e+02, percent-clipped=1.0 2023-04-16 22:35:46,085 INFO [train.py:893] (0/4) Epoch 16, batch 2050, loss[loss=0.2193, simple_loss=0.2654, pruned_loss=0.0866, over 13524.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2502, pruned_loss=0.06809, over 2661398.95 frames. ], batch size: 91, lr: 8.14e-03, grad_scale: 16.0 2023-04-16 22:36:08,882 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42562.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:36:14,152 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42567.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:36:31,595 INFO [train.py:893] (0/4) Epoch 16, batch 2100, loss[loss=0.273, simple_loss=0.3169, pruned_loss=0.1145, over 11795.00 frames. ], tot_loss[loss=0.1924, simple_loss=0.2495, pruned_loss=0.06769, over 2660290.25 frames. ], batch size: 157, lr: 8.14e-03, grad_scale: 16.0 2023-04-16 22:36:48,886 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42604.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 22:36:53,924 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=42610.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:36:57,949 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.948e+02 2.754e+02 3.328e+02 3.868e+02 6.469e+02, threshold=6.656e+02, percent-clipped=2.0 2023-04-16 22:37:19,345 INFO [train.py:893] (0/4) Epoch 16, batch 2150, loss[loss=0.2197, simple_loss=0.2783, pruned_loss=0.08059, over 13081.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2499, pruned_loss=0.06767, over 2657083.72 frames. ], batch size: 142, lr: 8.13e-03, grad_scale: 16.0 2023-04-16 22:37:29,664 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42649.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:37:32,909 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=42652.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 22:37:41,374 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3967, 2.5065, 2.7253, 3.9336, 3.5619, 3.9762, 2.9207, 2.2424], device='cuda:0'), covar=tensor([0.0303, 0.0746, 0.0663, 0.0039, 0.0200, 0.0044, 0.0638, 0.0952], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0147, 0.0162, 0.0090, 0.0115, 0.0087, 0.0166, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-16 22:38:05,985 INFO [train.py:893] (0/4) Epoch 16, batch 2200, loss[loss=0.2119, simple_loss=0.2592, pruned_loss=0.08234, over 13501.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2501, pruned_loss=0.06777, over 2658063.44 frames. ], batch size: 93, lr: 8.13e-03, grad_scale: 16.0 2023-04-16 22:38:27,200 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42710.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:38:31,883 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.739e+02 2.651e+02 3.047e+02 3.675e+02 6.847e+02, threshold=6.094e+02, percent-clipped=1.0 2023-04-16 22:38:45,932 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0253, 4.1252, 2.7311, 3.8882, 4.0473, 2.5369, 3.6172, 2.6616], device='cuda:0'), covar=tensor([0.0319, 0.0326, 0.1222, 0.0356, 0.0238, 0.1283, 0.0552, 0.1433], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0161, 0.0175, 0.0179, 0.0130, 0.0158, 0.0157, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 22:38:50,775 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42735.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:38:52,283 INFO [train.py:893] (0/4) Epoch 16, batch 2250, loss[loss=0.187, simple_loss=0.2461, pruned_loss=0.06397, over 13441.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2479, pruned_loss=0.06684, over 2660875.92 frames. ], batch size: 106, lr: 8.12e-03, grad_scale: 16.0 2023-04-16 22:39:39,971 INFO [train.py:893] (0/4) Epoch 16, batch 2300, loss[loss=0.1908, simple_loss=0.2488, pruned_loss=0.06637, over 13524.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2475, pruned_loss=0.06653, over 2661233.31 frames. ], batch size: 87, lr: 8.12e-03, grad_scale: 16.0 2023-04-16 22:40:05,866 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.686e+02 2.521e+02 3.073e+02 3.579e+02 5.794e+02, threshold=6.147e+02, percent-clipped=0.0 2023-04-16 22:40:13,936 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0954, 2.5318, 2.0608, 4.0031, 4.4294, 3.3482, 4.3476, 4.0637], device='cuda:0'), covar=tensor([0.0082, 0.0871, 0.1002, 0.0079, 0.0060, 0.0466, 0.0075, 0.0092], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0085, 0.0093, 0.0074, 0.0061, 0.0077, 0.0050, 0.0071], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:40:25,374 INFO [train.py:893] (0/4) Epoch 16, batch 2350, loss[loss=0.1958, simple_loss=0.2511, pruned_loss=0.07026, over 13345.00 frames. ], tot_loss[loss=0.1895, simple_loss=0.247, pruned_loss=0.06599, over 2665661.99 frames. ], batch size: 118, lr: 8.12e-03, grad_scale: 16.0 2023-04-16 22:40:47,071 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42859.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:40:49,440 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 22:40:49,568 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42862.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:40:49,953 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-16 22:41:06,647 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42879.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:41:13,720 INFO [train.py:893] (0/4) Epoch 16, batch 2400, loss[loss=0.1487, simple_loss=0.2136, pruned_loss=0.04187, over 13343.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2462, pruned_loss=0.06601, over 2658286.80 frames. ], batch size: 73, lr: 8.11e-03, grad_scale: 16.0 2023-04-16 22:41:39,547 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.683e+02 2.661e+02 3.119e+02 3.858e+02 1.033e+03, threshold=6.237e+02, percent-clipped=2.0 2023-04-16 22:41:44,076 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42920.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:41:58,914 INFO [train.py:893] (0/4) Epoch 16, batch 2450, loss[loss=0.183, simple_loss=0.245, pruned_loss=0.06051, over 13456.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2459, pruned_loss=0.06596, over 2659430.99 frames. ], batch size: 100, lr: 8.11e-03, grad_scale: 16.0 2023-04-16 22:42:01,667 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42940.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:42:20,133 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1799, 4.6253, 4.4936, 4.3816, 4.4013, 4.2691, 4.7017, 4.7259], device='cuda:0'), covar=tensor([0.0222, 0.0209, 0.0173, 0.0313, 0.0253, 0.0248, 0.0253, 0.0216], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0187, 0.0145, 0.0168, 0.0132, 0.0183, 0.0122, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 22:42:27,557 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0991, 2.0578, 4.0709, 3.8036, 3.8506, 3.0206, 3.6887, 3.0241], device='cuda:0'), covar=tensor([0.2047, 0.1569, 0.0099, 0.0215, 0.0235, 0.0747, 0.0248, 0.0794], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0182, 0.0113, 0.0120, 0.0124, 0.0168, 0.0134, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:42:46,443 INFO [train.py:893] (0/4) Epoch 16, batch 2500, loss[loss=0.1904, simple_loss=0.2484, pruned_loss=0.06615, over 13395.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2451, pruned_loss=0.06541, over 2663482.66 frames. ], batch size: 109, lr: 8.10e-03, grad_scale: 16.0 2023-04-16 22:43:03,971 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43005.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:43:13,892 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.663e+02 2.605e+02 2.966e+02 3.580e+02 6.970e+02, threshold=5.932e+02, percent-clipped=2.0 2023-04-16 22:43:17,853 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.83 vs. limit=5.0 2023-04-16 22:43:31,781 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43035.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:43:33,993 INFO [train.py:893] (0/4) Epoch 16, batch 2550, loss[loss=0.1687, simple_loss=0.2254, pruned_loss=0.05607, over 13445.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2458, pruned_loss=0.06593, over 2661741.59 frames. ], batch size: 65, lr: 8.10e-03, grad_scale: 16.0 2023-04-16 22:43:49,316 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3720, 2.1193, 2.0667, 2.3598, 1.8213, 2.4837, 2.2634, 2.0098], device='cuda:0'), covar=tensor([0.0081, 0.0204, 0.0135, 0.0151, 0.0188, 0.0116, 0.0210, 0.0167], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0101, 0.0110, 0.0103, 0.0114, 0.0093, 0.0095, 0.0094], device='cuda:0'), out_proj_covar=tensor([9.8782e-05, 1.0895e-04, 1.2157e-04, 1.1156e-04, 1.2616e-04, 9.9966e-05, 1.0288e-04, 1.0011e-04], device='cuda:0') 2023-04-16 22:43:58,312 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 22:44:08,092 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-16 22:44:09,672 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0839, 2.3673, 4.3209, 3.9457, 4.0998, 3.1634, 3.9003, 3.0701], device='cuda:0'), covar=tensor([0.1929, 0.1258, 0.0068, 0.0203, 0.0186, 0.0591, 0.0212, 0.0721], device='cuda:0'), in_proj_covar=tensor([0.0210, 0.0181, 0.0113, 0.0120, 0.0124, 0.0169, 0.0134, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:44:16,824 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43083.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:44:17,174 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2023-04-16 22:44:20,089 INFO [train.py:893] (0/4) Epoch 16, batch 2600, loss[loss=0.1713, simple_loss=0.2246, pruned_loss=0.05903, over 13405.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2459, pruned_loss=0.06635, over 2653604.58 frames. ], batch size: 62, lr: 8.09e-03, grad_scale: 16.0 2023-04-16 22:44:42,843 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3820, 4.1323, 4.3352, 4.3601, 4.5729, 4.2059, 4.6237, 4.5868], device='cuda:0'), covar=tensor([0.0446, 0.0599, 0.0667, 0.0535, 0.0636, 0.0805, 0.0492, 0.0541], device='cuda:0'), in_proj_covar=tensor([0.0264, 0.0274, 0.0278, 0.0209, 0.0395, 0.0319, 0.0245, 0.0273], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 22:44:46,499 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.090e+02 2.759e+02 3.409e+02 4.306e+02 1.371e+03, threshold=6.818e+02, percent-clipped=5.0 2023-04-16 22:45:03,096 INFO [train.py:893] (0/4) Epoch 16, batch 2650, loss[loss=0.1947, simple_loss=0.255, pruned_loss=0.06724, over 13260.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2468, pruned_loss=0.06705, over 2652825.44 frames. ], batch size: 124, lr: 8.09e-03, grad_scale: 16.0 2023-04-16 22:45:22,108 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43162.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:45:38,490 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-16 22:45:41,167 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-16.pt 2023-04-16 22:46:06,716 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 22:46:10,687 INFO [train.py:893] (0/4) Epoch 17, batch 0, loss[loss=0.2271, simple_loss=0.2679, pruned_loss=0.09316, over 13445.00 frames. ], tot_loss[loss=0.2271, simple_loss=0.2679, pruned_loss=0.09316, over 13445.00 frames. ], batch size: 95, lr: 7.84e-03, grad_scale: 16.0 2023-04-16 22:46:10,687 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 22:46:33,607 INFO [train.py:927] (0/4) Epoch 17, validation: loss=0.1429, simple_loss=0.2017, pruned_loss=0.04209, over 2446609.00 frames. 2023-04-16 22:46:33,608 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 22:46:33,916 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0626, 2.0301, 3.8633, 3.7506, 3.7093, 2.9013, 3.5238, 2.9177], device='cuda:0'), covar=tensor([0.1960, 0.1552, 0.0095, 0.0184, 0.0204, 0.0727, 0.0222, 0.0704], device='cuda:0'), in_proj_covar=tensor([0.0211, 0.0181, 0.0113, 0.0122, 0.0125, 0.0169, 0.0134, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:46:55,940 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43210.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:47:01,610 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.048e+02 2.768e+02 3.169e+02 4.178e+02 7.323e+02, threshold=6.338e+02, percent-clipped=1.0 2023-04-16 22:47:01,818 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43215.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:47:14,509 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.1096, 2.2788, 4.0583, 3.8381, 3.8896, 3.0543, 3.6436, 2.9663], device='cuda:0'), covar=tensor([0.2205, 0.1466, 0.0100, 0.0183, 0.0249, 0.0735, 0.0217, 0.0822], device='cuda:0'), in_proj_covar=tensor([0.0212, 0.0182, 0.0114, 0.0122, 0.0126, 0.0170, 0.0135, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:47:20,227 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43235.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:47:20,822 INFO [train.py:893] (0/4) Epoch 17, batch 50, loss[loss=0.2294, simple_loss=0.2688, pruned_loss=0.09498, over 13537.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2423, pruned_loss=0.06678, over 600687.12 frames. ], batch size: 85, lr: 7.83e-03, grad_scale: 16.0 2023-04-16 22:47:41,143 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4065, 4.1635, 3.3900, 4.0118, 3.2389, 2.3844, 4.2100, 2.4059], device='cuda:0'), covar=tensor([0.0619, 0.0532, 0.0438, 0.0187, 0.0647, 0.1721, 0.0696, 0.1221], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0133, 0.0128, 0.0112, 0.0145, 0.0181, 0.0159, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 22:47:45,128 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 22:47:45,129 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 22:47:45,129 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 22:47:45,141 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 22:47:45,149 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 22:47:45,175 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 22:47:45,193 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 22:48:06,636 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-04-16 22:48:06,954 INFO [train.py:893] (0/4) Epoch 17, batch 100, loss[loss=0.1762, simple_loss=0.234, pruned_loss=0.05925, over 13525.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2432, pruned_loss=0.06749, over 1055823.15 frames. ], batch size: 85, lr: 7.83e-03, grad_scale: 16.0 2023-04-16 22:48:09,181 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-16 22:48:24,961 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43305.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:48:34,474 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.195e+02 2.957e+02 3.496e+02 4.133e+02 8.444e+02, threshold=6.991e+02, percent-clipped=1.0 2023-04-16 22:48:50,105 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.03 vs. limit=5.0 2023-04-16 22:48:53,795 INFO [train.py:893] (0/4) Epoch 17, batch 150, loss[loss=0.2138, simple_loss=0.2671, pruned_loss=0.08022, over 13217.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2469, pruned_loss=0.06942, over 1406260.21 frames. ], batch size: 124, lr: 7.83e-03, grad_scale: 16.0 2023-04-16 22:49:09,911 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43353.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:49:22,960 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7742, 2.4257, 2.3568, 2.8291, 2.0415, 2.9606, 2.7961, 2.3784], device='cuda:0'), covar=tensor([0.0077, 0.0161, 0.0152, 0.0126, 0.0193, 0.0090, 0.0148, 0.0151], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0101, 0.0109, 0.0103, 0.0115, 0.0093, 0.0095, 0.0094], device='cuda:0'), out_proj_covar=tensor([9.8749e-05, 1.0898e-04, 1.2020e-04, 1.1133e-04, 1.2683e-04, 9.9373e-05, 1.0293e-04, 1.0025e-04], device='cuda:0') 2023-04-16 22:49:41,152 INFO [train.py:893] (0/4) Epoch 17, batch 200, loss[loss=0.2124, simple_loss=0.2647, pruned_loss=0.08008, over 13534.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2488, pruned_loss=0.07014, over 1679454.76 frames. ], batch size: 76, lr: 7.82e-03, grad_scale: 16.0 2023-04-16 22:50:09,167 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 2.754e+02 3.114e+02 3.666e+02 8.146e+02, threshold=6.228e+02, percent-clipped=1.0 2023-04-16 22:50:29,319 INFO [train.py:893] (0/4) Epoch 17, batch 250, loss[loss=0.1651, simple_loss=0.2214, pruned_loss=0.05438, over 13373.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2482, pruned_loss=0.06978, over 1895825.50 frames. ], batch size: 73, lr: 7.82e-03, grad_scale: 16.0 2023-04-16 22:51:17,452 INFO [train.py:893] (0/4) Epoch 17, batch 300, loss[loss=0.1542, simple_loss=0.2165, pruned_loss=0.04589, over 13364.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2475, pruned_loss=0.06884, over 2064606.98 frames. ], batch size: 67, lr: 7.81e-03, grad_scale: 16.0 2023-04-16 22:51:22,801 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5705, 4.6142, 3.2459, 4.4923, 4.5306, 3.0793, 4.1144, 3.2623], device='cuda:0'), covar=tensor([0.0249, 0.0206, 0.0929, 0.0424, 0.0190, 0.1053, 0.0365, 0.1181], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0161, 0.0174, 0.0180, 0.0130, 0.0157, 0.0156, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 22:51:33,764 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-16 22:51:35,916 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3264, 4.6234, 4.2916, 4.3531, 4.3791, 4.7649, 4.5468, 4.4680], device='cuda:0'), covar=tensor([0.0343, 0.0284, 0.0371, 0.1012, 0.0331, 0.0272, 0.0306, 0.0246], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0144, 0.0160, 0.0256, 0.0164, 0.0180, 0.0159, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2023-04-16 22:51:43,637 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43514.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:51:44,181 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.065e+02 2.747e+02 3.096e+02 3.753e+02 1.093e+03, threshold=6.191e+02, percent-clipped=3.0 2023-04-16 22:51:44,452 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43515.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:51:52,659 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.00 vs. limit=5.0 2023-04-16 22:52:03,576 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43535.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:52:04,114 INFO [train.py:893] (0/4) Epoch 17, batch 350, loss[loss=0.2271, simple_loss=0.2794, pruned_loss=0.08743, over 13568.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2487, pruned_loss=0.06933, over 2192320.80 frames. ], batch size: 89, lr: 7.81e-03, grad_scale: 16.0 2023-04-16 22:52:29,650 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43563.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:52:40,607 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43575.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:52:48,956 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43583.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:52:51,300 INFO [train.py:893] (0/4) Epoch 17, batch 400, loss[loss=0.2031, simple_loss=0.2585, pruned_loss=0.0738, over 13446.00 frames. ], tot_loss[loss=0.1934, simple_loss=0.2488, pruned_loss=0.069, over 2298493.38 frames. ], batch size: 95, lr: 7.80e-03, grad_scale: 16.0 2023-04-16 22:53:19,103 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.981e+02 2.711e+02 3.272e+02 3.876e+02 8.692e+02, threshold=6.544e+02, percent-clipped=2.0 2023-04-16 22:53:38,416 INFO [train.py:893] (0/4) Epoch 17, batch 450, loss[loss=0.1766, simple_loss=0.228, pruned_loss=0.06257, over 13500.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2502, pruned_loss=0.06975, over 2382718.78 frames. ], batch size: 70, lr: 7.80e-03, grad_scale: 16.0 2023-04-16 22:53:40,383 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1892, 2.6325, 2.1179, 4.0531, 4.6567, 3.4574, 4.5053, 4.2644], device='cuda:0'), covar=tensor([0.0102, 0.0915, 0.1073, 0.0115, 0.0084, 0.0481, 0.0081, 0.0090], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0087, 0.0095, 0.0077, 0.0062, 0.0079, 0.0052, 0.0073], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 22:53:52,393 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-16 22:54:01,425 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9986, 3.9826, 2.6798, 3.7380, 3.9175, 2.5276, 3.5286, 2.7320], device='cuda:0'), covar=tensor([0.0311, 0.0224, 0.1116, 0.0320, 0.0257, 0.1221, 0.0492, 0.1263], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0162, 0.0173, 0.0182, 0.0131, 0.0157, 0.0156, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 22:54:05,349 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 22:54:26,047 INFO [train.py:893] (0/4) Epoch 17, batch 500, loss[loss=0.2004, simple_loss=0.2636, pruned_loss=0.06865, over 13376.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2498, pruned_loss=0.06894, over 2442585.83 frames. ], batch size: 113, lr: 7.79e-03, grad_scale: 32.0 2023-04-16 22:54:48,925 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0534, 1.8780, 3.8238, 3.7103, 3.7243, 3.0004, 3.5817, 2.9026], device='cuda:0'), covar=tensor([0.2289, 0.1789, 0.0134, 0.0185, 0.0201, 0.0767, 0.0235, 0.0904], device='cuda:0'), in_proj_covar=tensor([0.0216, 0.0185, 0.0117, 0.0125, 0.0128, 0.0172, 0.0138, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 22:54:52,829 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 2.764e+02 3.201e+02 3.711e+02 6.287e+02, threshold=6.402e+02, percent-clipped=0.0 2023-04-16 22:55:01,314 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9993, 4.8437, 4.9927, 4.8544, 5.2832, 4.7690, 5.2762, 5.2737], device='cuda:0'), covar=tensor([0.0348, 0.0500, 0.0600, 0.0574, 0.0493, 0.0865, 0.0463, 0.0489], device='cuda:0'), in_proj_covar=tensor([0.0260, 0.0272, 0.0275, 0.0209, 0.0394, 0.0317, 0.0245, 0.0272], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 22:55:13,752 INFO [train.py:893] (0/4) Epoch 17, batch 550, loss[loss=0.1626, simple_loss=0.2228, pruned_loss=0.0512, over 13373.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.25, pruned_loss=0.06866, over 2491127.72 frames. ], batch size: 67, lr: 7.79e-03, grad_scale: 32.0 2023-04-16 22:55:15,689 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7364, 4.2642, 4.0171, 4.0619, 4.0417, 3.9525, 4.3087, 4.3809], device='cuda:0'), covar=tensor([0.0265, 0.0236, 0.0236, 0.0322, 0.0301, 0.0297, 0.0290, 0.0243], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0183, 0.0144, 0.0166, 0.0131, 0.0182, 0.0121, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 22:55:30,570 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2212, 2.9850, 3.6136, 2.7327, 2.4158, 2.5695, 3.8996, 4.0605], device='cuda:0'), covar=tensor([0.1224, 0.1850, 0.0447, 0.1581, 0.1549, 0.1523, 0.0346, 0.0248], device='cuda:0'), in_proj_covar=tensor([0.0235, 0.0253, 0.0185, 0.0217, 0.0211, 0.0178, 0.0196, 0.0162], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 22:55:35,698 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-16 22:55:59,539 INFO [train.py:893] (0/4) Epoch 17, batch 600, loss[loss=0.2028, simple_loss=0.2551, pruned_loss=0.07527, over 13548.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2492, pruned_loss=0.06831, over 2532666.35 frames. ], batch size: 87, lr: 7.79e-03, grad_scale: 32.0 2023-04-16 22:56:00,614 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43787.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:56:15,402 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9821, 3.9876, 3.1628, 3.6120, 3.1675, 2.1249, 3.9631, 2.2889], device='cuda:0'), covar=tensor([0.0717, 0.0333, 0.0500, 0.0310, 0.0703, 0.2102, 0.0716, 0.1329], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0131, 0.0127, 0.0111, 0.0145, 0.0180, 0.0160, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 22:56:27,698 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.795e+02 2.647e+02 3.088e+02 3.705e+02 5.223e+02, threshold=6.176e+02, percent-clipped=0.0 2023-04-16 22:56:47,896 INFO [train.py:893] (0/4) Epoch 17, batch 650, loss[loss=0.1823, simple_loss=0.2344, pruned_loss=0.0651, over 12002.00 frames. ], tot_loss[loss=0.1925, simple_loss=0.2486, pruned_loss=0.06826, over 2555116.88 frames. ], batch size: 157, lr: 7.78e-03, grad_scale: 32.0 2023-04-16 22:56:58,918 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43848.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:57:04,717 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1607, 4.0455, 4.2143, 2.5228, 4.5569, 4.2652, 4.2375, 4.5761], device='cuda:0'), covar=tensor([0.0229, 0.0109, 0.0114, 0.1119, 0.0132, 0.0208, 0.0114, 0.0078], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0048, 0.0074, 0.0098, 0.0092, 0.0096, 0.0073, 0.0067], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 22:57:12,304 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8516, 3.7137, 3.8723, 2.2727, 4.1161, 3.9246, 3.9077, 4.1392], device='cuda:0'), covar=tensor([0.0229, 0.0133, 0.0122, 0.1180, 0.0134, 0.0216, 0.0123, 0.0091], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0048, 0.0075, 0.0098, 0.0092, 0.0096, 0.0074, 0.0068], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 22:57:19,630 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43870.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:57:34,549 INFO [train.py:893] (0/4) Epoch 17, batch 700, loss[loss=0.1874, simple_loss=0.2475, pruned_loss=0.06372, over 13255.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.247, pruned_loss=0.06665, over 2581462.73 frames. ], batch size: 124, lr: 7.78e-03, grad_scale: 32.0 2023-04-16 22:57:56,577 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43909.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:58:01,413 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.703e+02 2.597e+02 2.999e+02 3.684e+02 5.228e+02, threshold=5.998e+02, percent-clipped=0.0 2023-04-16 22:58:21,226 INFO [train.py:893] (0/4) Epoch 17, batch 750, loss[loss=0.1794, simple_loss=0.2391, pruned_loss=0.05989, over 13330.00 frames. ], tot_loss[loss=0.1897, simple_loss=0.2463, pruned_loss=0.06659, over 2595801.39 frames. ], batch size: 73, lr: 7.77e-03, grad_scale: 32.0 2023-04-16 22:58:34,868 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-16 22:58:43,818 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43959.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:58:53,179 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43970.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:59:07,816 INFO [train.py:893] (0/4) Epoch 17, batch 800, loss[loss=0.1936, simple_loss=0.2492, pruned_loss=0.06902, over 13034.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.2472, pruned_loss=0.06709, over 2614851.66 frames. ], batch size: 142, lr: 7.77e-03, grad_scale: 32.0 2023-04-16 22:59:20,856 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-44000.pt 2023-04-16 22:59:38,396 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.890e+02 2.754e+02 3.341e+02 3.760e+02 1.385e+03, threshold=6.683e+02, percent-clipped=4.0 2023-04-16 22:59:44,386 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44020.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 22:59:58,196 INFO [train.py:893] (0/4) Epoch 17, batch 850, loss[loss=0.2045, simple_loss=0.255, pruned_loss=0.07701, over 11803.00 frames. ], tot_loss[loss=0.1925, simple_loss=0.2488, pruned_loss=0.06805, over 2625459.51 frames. ], batch size: 157, lr: 7.76e-03, grad_scale: 32.0 2023-04-16 23:00:39,637 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-16 23:00:44,723 INFO [train.py:893] (0/4) Epoch 17, batch 900, loss[loss=0.1751, simple_loss=0.2389, pruned_loss=0.05563, over 13442.00 frames. ], tot_loss[loss=0.1931, simple_loss=0.2491, pruned_loss=0.06851, over 2635115.10 frames. ], batch size: 95, lr: 7.76e-03, grad_scale: 32.0 2023-04-16 23:00:51,109 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-16 23:01:10,800 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.76 vs. limit=5.0 2023-04-16 23:01:11,105 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.969e+02 2.735e+02 3.180e+02 3.676e+02 7.064e+02, threshold=6.360e+02, percent-clipped=1.0 2023-04-16 23:01:14,922 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.2637, 3.0162, 2.4791, 1.8442, 2.1003, 2.6569, 2.6721, 3.2461], device='cuda:0'), covar=tensor([0.0828, 0.0294, 0.0583, 0.1590, 0.0407, 0.0394, 0.0614, 0.0198], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0123, 0.0107, 0.0198, 0.0103, 0.0148, 0.0158, 0.0120], device='cuda:0'), out_proj_covar=tensor([1.1478e-04, 9.2139e-05, 8.4567e-05, 1.4898e-04, 7.7646e-05, 1.1222e-04, 1.2047e-04, 8.9076e-05], device='cuda:0') 2023-04-16 23:01:15,506 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 23:01:28,120 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8165, 4.6018, 4.8319, 4.7777, 5.0868, 4.5967, 5.0951, 5.0873], device='cuda:0'), covar=tensor([0.0339, 0.0515, 0.0609, 0.0475, 0.0452, 0.0830, 0.0426, 0.0422], device='cuda:0'), in_proj_covar=tensor([0.0263, 0.0274, 0.0279, 0.0209, 0.0396, 0.0321, 0.0248, 0.0272], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:01:29,521 INFO [train.py:893] (0/4) Epoch 17, batch 950, loss[loss=0.1822, simple_loss=0.2416, pruned_loss=0.06143, over 13376.00 frames. ], tot_loss[loss=0.1925, simple_loss=0.2482, pruned_loss=0.06838, over 2636131.93 frames. ], batch size: 73, lr: 7.75e-03, grad_scale: 32.0 2023-04-16 23:01:36,550 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44143.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:01:45,710 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0789, 4.4340, 4.0379, 4.1744, 4.1742, 4.5718, 4.3569, 4.1963], device='cuda:0'), covar=tensor([0.0288, 0.0271, 0.0381, 0.0957, 0.0299, 0.0230, 0.0280, 0.0279], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0143, 0.0161, 0.0256, 0.0163, 0.0177, 0.0158, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2023-04-16 23:02:01,357 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44170.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:02:14,140 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1584, 4.2674, 2.8572, 4.1028, 4.1286, 2.5902, 3.7939, 2.8753], device='cuda:0'), covar=tensor([0.0262, 0.0174, 0.1013, 0.0339, 0.0224, 0.1194, 0.0451, 0.1282], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0162, 0.0171, 0.0182, 0.0130, 0.0156, 0.0154, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:02:17,148 INFO [train.py:893] (0/4) Epoch 17, batch 1000, loss[loss=0.198, simple_loss=0.2422, pruned_loss=0.07688, over 11898.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2464, pruned_loss=0.06798, over 2640676.45 frames. ], batch size: 157, lr: 7.75e-03, grad_scale: 32.0 2023-04-16 23:02:44,691 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.003e+02 2.659e+02 3.160e+02 3.851e+02 7.237e+02, threshold=6.320e+02, percent-clipped=1.0 2023-04-16 23:02:47,441 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44218.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:03:02,977 INFO [train.py:893] (0/4) Epoch 17, batch 1050, loss[loss=0.1694, simple_loss=0.2344, pruned_loss=0.05219, over 13490.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2446, pruned_loss=0.06636, over 2644386.20 frames. ], batch size: 93, lr: 7.75e-03, grad_scale: 32.0 2023-04-16 23:03:29,655 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.5445, 5.0391, 4.9361, 5.0450, 4.6766, 4.8804, 5.4841, 5.0271], device='cuda:0'), covar=tensor([0.0759, 0.1117, 0.2071, 0.2433, 0.0931, 0.1592, 0.0914, 0.1075], device='cuda:0'), in_proj_covar=tensor([0.0306, 0.0348, 0.0440, 0.0449, 0.0270, 0.0333, 0.0403, 0.0325], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:03:30,458 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44265.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:03:50,140 INFO [train.py:893] (0/4) Epoch 17, batch 1100, loss[loss=0.178, simple_loss=0.2403, pruned_loss=0.05786, over 13429.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2453, pruned_loss=0.06599, over 2652884.50 frames. ], batch size: 95, lr: 7.74e-03, grad_scale: 32.0 2023-04-16 23:04:16,535 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.690e+02 3.160e+02 3.946e+02 7.228e+02, threshold=6.319e+02, percent-clipped=1.0 2023-04-16 23:04:16,808 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44315.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:04:36,902 INFO [train.py:893] (0/4) Epoch 17, batch 1150, loss[loss=0.195, simple_loss=0.2567, pruned_loss=0.06667, over 13489.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2453, pruned_loss=0.06555, over 2649193.66 frames. ], batch size: 93, lr: 7.74e-03, grad_scale: 32.0 2023-04-16 23:05:22,752 INFO [train.py:893] (0/4) Epoch 17, batch 1200, loss[loss=0.177, simple_loss=0.2409, pruned_loss=0.05653, over 13471.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2463, pruned_loss=0.06567, over 2650827.65 frames. ], batch size: 79, lr: 7.73e-03, grad_scale: 32.0 2023-04-16 23:05:42,963 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-16 23:05:50,372 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.921e+02 2.713e+02 3.116e+02 3.508e+02 5.884e+02, threshold=6.232e+02, percent-clipped=0.0 2023-04-16 23:05:51,327 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 23:06:03,786 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 23:06:09,561 INFO [train.py:893] (0/4) Epoch 17, batch 1250, loss[loss=0.198, simple_loss=0.2624, pruned_loss=0.06676, over 13477.00 frames. ], tot_loss[loss=0.1897, simple_loss=0.2469, pruned_loss=0.06627, over 2654211.70 frames. ], batch size: 81, lr: 7.73e-03, grad_scale: 32.0 2023-04-16 23:06:16,210 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44443.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:06:38,857 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.31 vs. limit=5.0 2023-04-16 23:06:54,446 INFO [train.py:893] (0/4) Epoch 17, batch 1300, loss[loss=0.1923, simple_loss=0.2438, pruned_loss=0.07039, over 13545.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2479, pruned_loss=0.06686, over 2657531.64 frames. ], batch size: 85, lr: 7.72e-03, grad_scale: 32.0 2023-04-16 23:07:00,366 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44491.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:07:18,083 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-16 23:07:21,527 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.930e+02 2.691e+02 3.241e+02 3.696e+02 7.455e+02, threshold=6.482e+02, percent-clipped=2.0 2023-04-16 23:07:34,638 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=44527.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:07:41,945 INFO [train.py:893] (0/4) Epoch 17, batch 1350, loss[loss=0.1905, simple_loss=0.2468, pruned_loss=0.06712, over 13464.00 frames. ], tot_loss[loss=0.1913, simple_loss=0.2483, pruned_loss=0.06716, over 2656940.03 frames. ], batch size: 100, lr: 7.72e-03, grad_scale: 32.0 2023-04-16 23:07:55,263 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=44550.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:08:09,360 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44565.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:08:28,912 INFO [train.py:893] (0/4) Epoch 17, batch 1400, loss[loss=0.1705, simple_loss=0.2295, pruned_loss=0.05577, over 13444.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2478, pruned_loss=0.06732, over 2658216.85 frames. ], batch size: 106, lr: 7.72e-03, grad_scale: 32.0 2023-04-16 23:08:30,893 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44588.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:08:51,590 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44611.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:08:53,043 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44613.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:08:55,324 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.769e+02 2.732e+02 3.117e+02 3.575e+02 6.095e+02, threshold=6.234e+02, percent-clipped=0.0 2023-04-16 23:08:55,605 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44615.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:09:14,430 INFO [train.py:893] (0/4) Epoch 17, batch 1450, loss[loss=0.2075, simple_loss=0.2611, pruned_loss=0.07696, over 13522.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2482, pruned_loss=0.06808, over 2660701.52 frames. ], batch size: 76, lr: 7.71e-03, grad_scale: 32.0 2023-04-16 23:09:27,931 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0684, 4.2320, 2.7631, 3.9730, 4.0824, 2.6433, 3.6741, 2.9569], device='cuda:0'), covar=tensor([0.0298, 0.0213, 0.1161, 0.0397, 0.0208, 0.1224, 0.0463, 0.1327], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0162, 0.0171, 0.0184, 0.0129, 0.0157, 0.0155, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:09:36,414 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6863, 2.5702, 2.0386, 1.5789, 1.6276, 2.0908, 2.1719, 2.6743], device='cuda:0'), covar=tensor([0.0805, 0.0274, 0.0737, 0.1388, 0.0156, 0.0425, 0.0635, 0.0198], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0124, 0.0107, 0.0195, 0.0100, 0.0148, 0.0158, 0.0119], device='cuda:0'), out_proj_covar=tensor([1.1310e-04, 9.2340e-05, 8.4609e-05, 1.4688e-04, 7.5206e-05, 1.1184e-04, 1.2031e-04, 8.8675e-05], device='cuda:0') 2023-04-16 23:09:39,612 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44663.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:10:01,798 INFO [train.py:893] (0/4) Epoch 17, batch 1500, loss[loss=0.2288, simple_loss=0.2744, pruned_loss=0.09161, over 13360.00 frames. ], tot_loss[loss=0.1916, simple_loss=0.2479, pruned_loss=0.06763, over 2663828.12 frames. ], batch size: 118, lr: 7.71e-03, grad_scale: 32.0 2023-04-16 23:10:29,107 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.915e+02 2.738e+02 3.164e+02 3.676e+02 8.107e+02, threshold=6.328e+02, percent-clipped=2.0 2023-04-16 23:10:48,315 INFO [train.py:893] (0/4) Epoch 17, batch 1550, loss[loss=0.1566, simple_loss=0.2102, pruned_loss=0.05149, over 13329.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.2473, pruned_loss=0.06685, over 2663747.28 frames. ], batch size: 67, lr: 7.70e-03, grad_scale: 32.0 2023-04-16 23:10:55,369 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0330, 1.9896, 2.3762, 3.2959, 3.0612, 3.3524, 2.8001, 2.2674], device='cuda:0'), covar=tensor([0.0209, 0.0887, 0.0620, 0.0068, 0.0249, 0.0061, 0.0478, 0.0769], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0149, 0.0162, 0.0091, 0.0114, 0.0088, 0.0166, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:11:26,093 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1628, 3.9142, 3.2269, 3.8609, 3.2522, 2.4666, 4.0098, 2.3049], device='cuda:0'), covar=tensor([0.0629, 0.0554, 0.0498, 0.0272, 0.0666, 0.2006, 0.1094, 0.1515], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0133, 0.0130, 0.0111, 0.0147, 0.0183, 0.0162, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:11:34,993 INFO [train.py:893] (0/4) Epoch 17, batch 1600, loss[loss=0.196, simple_loss=0.2603, pruned_loss=0.06589, over 13459.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2471, pruned_loss=0.06657, over 2654734.21 frames. ], batch size: 103, lr: 7.70e-03, grad_scale: 32.0 2023-04-16 23:11:38,918 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-04-16 23:12:02,554 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.839e+02 2.671e+02 3.065e+02 4.037e+02 6.217e+02, threshold=6.130e+02, percent-clipped=0.0 2023-04-16 23:12:02,822 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0968, 4.3306, 4.0300, 4.1166, 4.2522, 4.5044, 4.3014, 4.0965], device='cuda:0'), covar=tensor([0.0258, 0.0275, 0.0346, 0.0845, 0.0248, 0.0209, 0.0272, 0.0333], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0146, 0.0163, 0.0256, 0.0165, 0.0181, 0.0160, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2023-04-16 23:12:23,133 INFO [train.py:893] (0/4) Epoch 17, batch 1650, loss[loss=0.1871, simple_loss=0.2445, pruned_loss=0.06489, over 13430.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2471, pruned_loss=0.06621, over 2653632.99 frames. ], batch size: 95, lr: 7.69e-03, grad_scale: 32.0 2023-04-16 23:12:24,319 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8986, 4.1005, 2.6509, 3.6530, 3.9038, 2.5371, 3.4967, 2.7851], device='cuda:0'), covar=tensor([0.0323, 0.0175, 0.1232, 0.0383, 0.0254, 0.1305, 0.0484, 0.1453], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0163, 0.0173, 0.0185, 0.0130, 0.0157, 0.0154, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:13:05,417 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3387, 2.2365, 2.5212, 3.7551, 3.4440, 3.8344, 3.0794, 2.2164], device='cuda:0'), covar=tensor([0.0240, 0.0927, 0.0798, 0.0062, 0.0240, 0.0048, 0.0542, 0.0983], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0152, 0.0166, 0.0093, 0.0117, 0.0090, 0.0169, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:13:06,089 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44883.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:13:08,326 INFO [train.py:893] (0/4) Epoch 17, batch 1700, loss[loss=0.2073, simple_loss=0.2607, pruned_loss=0.07689, over 13516.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2473, pruned_loss=0.0659, over 2656693.69 frames. ], batch size: 85, lr: 7.69e-03, grad_scale: 32.0 2023-04-16 23:13:25,323 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9445, 4.0666, 3.2067, 2.8072, 2.9076, 2.4857, 4.2087, 2.4108], device='cuda:0'), covar=tensor([0.1522, 0.0296, 0.1072, 0.1968, 0.0805, 0.3226, 0.0226, 0.3873], device='cuda:0'), in_proj_covar=tensor([0.0308, 0.0282, 0.0303, 0.0315, 0.0249, 0.0317, 0.0203, 0.0357], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 23:13:27,702 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44906.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:13:35,564 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.839e+02 2.756e+02 3.136e+02 3.621e+02 5.395e+02, threshold=6.271e+02, percent-clipped=0.0 2023-04-16 23:13:55,254 INFO [train.py:893] (0/4) Epoch 17, batch 1750, loss[loss=0.1951, simple_loss=0.2539, pruned_loss=0.06811, over 13264.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2468, pruned_loss=0.06565, over 2659110.40 frames. ], batch size: 124, lr: 7.69e-03, grad_scale: 32.0 2023-04-16 23:14:40,131 INFO [train.py:893] (0/4) Epoch 17, batch 1800, loss[loss=0.2072, simple_loss=0.2634, pruned_loss=0.0755, over 13285.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2461, pruned_loss=0.06508, over 2658843.57 frames. ], batch size: 124, lr: 7.68e-03, grad_scale: 16.0 2023-04-16 23:14:43,846 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7659, 2.3724, 2.4282, 2.8488, 2.0626, 2.9260, 2.8731, 2.3827], device='cuda:0'), covar=tensor([0.0114, 0.0211, 0.0177, 0.0169, 0.0224, 0.0139, 0.0183, 0.0185], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0102, 0.0110, 0.0106, 0.0118, 0.0095, 0.0098, 0.0096], device='cuda:0'), out_proj_covar=tensor([9.8430e-05, 1.0950e-04, 1.2118e-04, 1.1466e-04, 1.2978e-04, 1.0115e-04, 1.0639e-04, 1.0191e-04], device='cuda:0') 2023-04-16 23:15:08,028 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.790e+02 2.583e+02 3.065e+02 3.779e+02 6.192e+02, threshold=6.130e+02, percent-clipped=0.0 2023-04-16 23:15:28,116 INFO [train.py:893] (0/4) Epoch 17, batch 1850, loss[loss=0.1747, simple_loss=0.2362, pruned_loss=0.05654, over 13489.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2451, pruned_loss=0.06431, over 2662615.43 frames. ], batch size: 70, lr: 7.68e-03, grad_scale: 16.0 2023-04-16 23:15:31,431 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 23:15:46,536 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9133, 2.9829, 2.4341, 1.8551, 1.8638, 2.5236, 2.6865, 3.1499], device='cuda:0'), covar=tensor([0.1146, 0.0324, 0.0755, 0.1709, 0.0494, 0.0534, 0.0675, 0.0261], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0126, 0.0110, 0.0199, 0.0104, 0.0151, 0.0161, 0.0123], device='cuda:0'), out_proj_covar=tensor([1.1391e-04, 9.4499e-05, 8.7125e-05, 1.4983e-04, 7.8002e-05, 1.1384e-04, 1.2235e-04, 9.1377e-05], device='cuda:0') 2023-04-16 23:15:47,359 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45056.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:16:09,833 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45081.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:16:14,547 INFO [train.py:893] (0/4) Epoch 17, batch 1900, loss[loss=0.1891, simple_loss=0.2433, pruned_loss=0.0674, over 11891.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2445, pruned_loss=0.06453, over 2660833.49 frames. ], batch size: 157, lr: 7.67e-03, grad_scale: 16.0 2023-04-16 23:16:42,599 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.971e+02 2.721e+02 3.190e+02 3.626e+02 8.319e+02, threshold=6.379e+02, percent-clipped=2.0 2023-04-16 23:16:43,788 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45117.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 23:16:59,861 INFO [train.py:893] (0/4) Epoch 17, batch 1950, loss[loss=0.2179, simple_loss=0.2592, pruned_loss=0.08836, over 11774.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2444, pruned_loss=0.06483, over 2663334.42 frames. ], batch size: 157, lr: 7.67e-03, grad_scale: 16.0 2023-04-16 23:17:05,151 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45142.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:17:44,090 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45183.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:17:46,414 INFO [train.py:893] (0/4) Epoch 17, batch 2000, loss[loss=0.1714, simple_loss=0.2271, pruned_loss=0.05786, over 13466.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.2457, pruned_loss=0.06553, over 2657877.23 frames. ], batch size: 65, lr: 7.67e-03, grad_scale: 16.0 2023-04-16 23:17:53,076 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-16 23:18:02,093 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4992, 3.8803, 3.6485, 4.2311, 2.3202, 3.1696, 3.9459, 2.3720], device='cuda:0'), covar=tensor([0.0116, 0.0423, 0.0741, 0.0490, 0.1489, 0.0901, 0.0442, 0.1684], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0178, 0.0197, 0.0229, 0.0178, 0.0193, 0.0174, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:18:06,153 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45206.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:18:14,877 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.975e+02 2.748e+02 3.264e+02 3.939e+02 8.760e+02, threshold=6.528e+02, percent-clipped=3.0 2023-04-16 23:18:28,245 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45231.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:18:32,270 INFO [train.py:893] (0/4) Epoch 17, batch 2050, loss[loss=0.1937, simple_loss=0.2498, pruned_loss=0.0688, over 13534.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2467, pruned_loss=0.06596, over 2658620.20 frames. ], batch size: 83, lr: 7.66e-03, grad_scale: 16.0 2023-04-16 23:18:50,614 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45254.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:19:09,808 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2059, 4.6666, 4.4738, 4.3128, 4.5273, 4.2703, 4.7124, 4.7029], device='cuda:0'), covar=tensor([0.0255, 0.0306, 0.0279, 0.0435, 0.0270, 0.0339, 0.0375, 0.0294], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0193, 0.0152, 0.0176, 0.0137, 0.0190, 0.0127, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 23:19:20,264 INFO [train.py:893] (0/4) Epoch 17, batch 2100, loss[loss=0.1617, simple_loss=0.2313, pruned_loss=0.04607, over 13505.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.2458, pruned_loss=0.06535, over 2658128.72 frames. ], batch size: 81, lr: 7.66e-03, grad_scale: 16.0 2023-04-16 23:19:48,314 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.819e+02 2.694e+02 3.157e+02 3.891e+02 9.018e+02, threshold=6.314e+02, percent-clipped=1.0 2023-04-16 23:19:55,357 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0041, 4.2608, 4.1093, 3.8531, 4.1593, 4.4020, 4.2781, 4.1220], device='cuda:0'), covar=tensor([0.0348, 0.0304, 0.0340, 0.1254, 0.0315, 0.0320, 0.0301, 0.0387], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0146, 0.0165, 0.0261, 0.0167, 0.0183, 0.0161, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-16 23:19:59,904 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-04-16 23:20:07,002 INFO [train.py:893] (0/4) Epoch 17, batch 2150, loss[loss=0.1942, simple_loss=0.2495, pruned_loss=0.06944, over 13243.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2456, pruned_loss=0.06503, over 2660009.71 frames. ], batch size: 124, lr: 7.65e-03, grad_scale: 16.0 2023-04-16 23:20:18,041 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45348.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 23:20:52,809 INFO [train.py:893] (0/4) Epoch 17, batch 2200, loss[loss=0.1867, simple_loss=0.2498, pruned_loss=0.06177, over 13369.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2452, pruned_loss=0.06462, over 2663798.68 frames. ], batch size: 118, lr: 7.65e-03, grad_scale: 16.0 2023-04-16 23:21:15,903 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45409.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 23:21:18,257 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45412.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 23:21:21,302 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.722e+02 2.681e+02 3.203e+02 3.685e+02 6.103e+02, threshold=6.406e+02, percent-clipped=0.0 2023-04-16 23:21:40,218 INFO [train.py:893] (0/4) Epoch 17, batch 2250, loss[loss=0.1644, simple_loss=0.2269, pruned_loss=0.05097, over 13460.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.244, pruned_loss=0.06437, over 2664749.21 frames. ], batch size: 79, lr: 7.64e-03, grad_scale: 16.0 2023-04-16 23:21:40,401 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2084, 4.6961, 4.6363, 4.7316, 4.4217, 4.5192, 5.1577, 4.6370], device='cuda:0'), covar=tensor([0.0607, 0.1174, 0.1981, 0.2252, 0.0841, 0.1417, 0.0725, 0.1250], device='cuda:0'), in_proj_covar=tensor([0.0300, 0.0346, 0.0438, 0.0446, 0.0265, 0.0329, 0.0399, 0.0322], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:21:41,302 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45437.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:21:53,840 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.71 vs. limit=5.0 2023-04-16 23:22:24,771 INFO [train.py:893] (0/4) Epoch 17, batch 2300, loss[loss=0.1451, simple_loss=0.2038, pruned_loss=0.04326, over 13138.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2445, pruned_loss=0.06457, over 2660470.07 frames. ], batch size: 58, lr: 7.64e-03, grad_scale: 16.0 2023-04-16 23:22:51,732 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.737e+02 2.559e+02 2.932e+02 3.587e+02 6.349e+02, threshold=5.863e+02, percent-clipped=0.0 2023-04-16 23:23:11,060 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4920, 3.3696, 3.9851, 3.0078, 2.5383, 2.7621, 4.2828, 4.4128], device='cuda:0'), covar=tensor([0.1139, 0.1555, 0.0326, 0.1480, 0.1572, 0.1446, 0.0257, 0.0187], device='cuda:0'), in_proj_covar=tensor([0.0239, 0.0259, 0.0187, 0.0219, 0.0213, 0.0179, 0.0197, 0.0167], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:23:11,528 INFO [train.py:893] (0/4) Epoch 17, batch 2350, loss[loss=0.1835, simple_loss=0.2414, pruned_loss=0.06277, over 13518.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2441, pruned_loss=0.06456, over 2660141.82 frames. ], batch size: 91, lr: 7.64e-03, grad_scale: 16.0 2023-04-16 23:23:34,121 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-16 23:23:58,212 INFO [train.py:893] (0/4) Epoch 17, batch 2400, loss[loss=0.1908, simple_loss=0.2463, pruned_loss=0.06761, over 13361.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2431, pruned_loss=0.06414, over 2662959.86 frames. ], batch size: 62, lr: 7.63e-03, grad_scale: 16.0 2023-04-16 23:24:25,372 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8544, 3.6055, 2.9900, 3.3985, 3.0488, 2.1611, 3.7177, 1.9921], device='cuda:0'), covar=tensor([0.0682, 0.0747, 0.0428, 0.0357, 0.0613, 0.1827, 0.0841, 0.1361], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0135, 0.0130, 0.0112, 0.0147, 0.0184, 0.0163, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:24:26,799 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.045e+02 2.728e+02 3.241e+02 3.826e+02 5.379e+02, threshold=6.482e+02, percent-clipped=0.0 2023-04-16 23:24:40,388 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0926, 3.4005, 3.2910, 3.7029, 2.0954, 2.8471, 3.5067, 2.1254], device='cuda:0'), covar=tensor([0.0132, 0.0693, 0.0847, 0.0663, 0.1616, 0.1045, 0.0694, 0.1813], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0179, 0.0199, 0.0230, 0.0180, 0.0195, 0.0176, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:24:45,232 INFO [train.py:893] (0/4) Epoch 17, batch 2450, loss[loss=0.1746, simple_loss=0.2336, pruned_loss=0.05782, over 13472.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2434, pruned_loss=0.06446, over 2663922.95 frames. ], batch size: 79, lr: 7.63e-03, grad_scale: 16.0 2023-04-16 23:25:09,577 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1063, 4.5711, 4.3955, 4.3189, 4.2938, 4.2302, 4.6054, 4.6384], device='cuda:0'), covar=tensor([0.0236, 0.0213, 0.0180, 0.0309, 0.0324, 0.0267, 0.0261, 0.0228], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0192, 0.0152, 0.0175, 0.0138, 0.0190, 0.0127, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 23:25:32,478 INFO [train.py:893] (0/4) Epoch 17, batch 2500, loss[loss=0.1836, simple_loss=0.2404, pruned_loss=0.06335, over 13513.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2435, pruned_loss=0.0645, over 2662159.23 frames. ], batch size: 85, lr: 7.62e-03, grad_scale: 16.0 2023-04-16 23:25:35,292 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8020, 4.0308, 2.9288, 3.6389, 3.1461, 1.9982, 3.9926, 2.0641], device='cuda:0'), covar=tensor([0.0844, 0.0565, 0.0659, 0.0346, 0.0682, 0.2517, 0.0893, 0.1712], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0135, 0.0130, 0.0112, 0.0147, 0.0184, 0.0163, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:25:46,789 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45702.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:25:48,437 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45704.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 23:25:56,810 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45712.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 23:26:00,703 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.951e+02 2.760e+02 3.151e+02 3.629e+02 9.778e+02, threshold=6.301e+02, percent-clipped=0.0 2023-04-16 23:26:01,991 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.35 vs. limit=5.0 2023-04-16 23:26:09,470 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-16 23:26:17,926 INFO [train.py:893] (0/4) Epoch 17, batch 2550, loss[loss=0.1663, simple_loss=0.2279, pruned_loss=0.05233, over 13365.00 frames. ], tot_loss[loss=0.186, simple_loss=0.2435, pruned_loss=0.0643, over 2663528.32 frames. ], batch size: 73, lr: 7.62e-03, grad_scale: 16.0 2023-04-16 23:26:18,955 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45737.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:26:40,107 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-16 23:26:41,034 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45760.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:26:43,659 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45763.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:26:54,113 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45774.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:27:04,765 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45785.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:27:05,401 INFO [train.py:893] (0/4) Epoch 17, batch 2600, loss[loss=0.1873, simple_loss=0.2426, pruned_loss=0.06595, over 13451.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2443, pruned_loss=0.06513, over 2661207.93 frames. ], batch size: 79, lr: 7.62e-03, grad_scale: 16.0 2023-04-16 23:27:31,662 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.842e+02 2.635e+02 3.152e+02 4.000e+02 8.090e+02, threshold=6.305e+02, percent-clipped=5.0 2023-04-16 23:27:44,694 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0748, 3.9221, 4.0998, 2.4748, 4.4654, 4.1256, 4.1148, 4.3736], device='cuda:0'), covar=tensor([0.0252, 0.0148, 0.0139, 0.1151, 0.0142, 0.0238, 0.0150, 0.0121], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0050, 0.0076, 0.0100, 0.0094, 0.0098, 0.0075, 0.0069], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:27:46,967 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45835.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:27:47,566 INFO [train.py:893] (0/4) Epoch 17, batch 2650, loss[loss=0.1897, simple_loss=0.2465, pruned_loss=0.06649, over 13373.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2442, pruned_loss=0.06527, over 2661797.97 frames. ], batch size: 84, lr: 7.61e-03, grad_scale: 16.0 2023-04-16 23:27:56,726 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6198, 3.9035, 3.5620, 4.1951, 2.3693, 3.2082, 3.9731, 2.3565], device='cuda:0'), covar=tensor([0.0121, 0.0445, 0.0929, 0.0731, 0.1506, 0.0914, 0.0533, 0.1672], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0179, 0.0200, 0.0231, 0.0179, 0.0196, 0.0175, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:28:05,551 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45859.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:28:26,427 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-17.pt 2023-04-16 23:28:52,453 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-16 23:28:55,851 INFO [train.py:893] (0/4) Epoch 18, batch 0, loss[loss=0.2196, simple_loss=0.2704, pruned_loss=0.08439, over 13534.00 frames. ], tot_loss[loss=0.2196, simple_loss=0.2704, pruned_loss=0.08439, over 13534.00 frames. ], batch size: 85, lr: 7.39e-03, grad_scale: 16.0 2023-04-16 23:28:55,852 INFO [train.py:918] (0/4) Computing validation loss 2023-04-16 23:29:18,849 INFO [train.py:927] (0/4) Epoch 18, validation: loss=0.1416, simple_loss=0.2005, pruned_loss=0.04137, over 2446609.00 frames. 2023-04-16 23:29:18,850 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-16 23:29:35,647 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6772, 2.7678, 2.9678, 4.1774, 3.7987, 4.2266, 3.4798, 2.5682], device='cuda:0'), covar=tensor([0.0233, 0.0847, 0.0794, 0.0051, 0.0199, 0.0045, 0.0589, 0.0951], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0147, 0.0162, 0.0090, 0.0113, 0.0088, 0.0164, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:29:46,276 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.926e+02 2.882e+02 3.236e+02 3.835e+02 6.257e+02, threshold=6.473e+02, percent-clipped=0.0 2023-04-16 23:29:51,603 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45920.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:30:02,475 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9482, 4.5476, 3.8542, 3.5360, 3.9396, 3.2542, 4.7936, 2.9793], device='cuda:0'), covar=tensor([0.1144, 0.0322, 0.1004, 0.1550, 0.0547, 0.2428, 0.0188, 0.3735], device='cuda:0'), in_proj_covar=tensor([0.0306, 0.0280, 0.0302, 0.0317, 0.0248, 0.0318, 0.0202, 0.0356], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 23:30:05,446 INFO [train.py:893] (0/4) Epoch 18, batch 50, loss[loss=0.1874, simple_loss=0.2309, pruned_loss=0.07192, over 13418.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2356, pruned_loss=0.06205, over 595225.73 frames. ], batch size: 65, lr: 7.39e-03, grad_scale: 16.0 2023-04-16 23:30:17,418 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0891, 2.7695, 2.6918, 3.1231, 2.6663, 3.2740, 3.2524, 2.7980], device='cuda:0'), covar=tensor([0.0065, 0.0162, 0.0125, 0.0143, 0.0159, 0.0098, 0.0133, 0.0128], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0104, 0.0111, 0.0107, 0.0121, 0.0097, 0.0099, 0.0096], device='cuda:0'), out_proj_covar=tensor([9.9644e-05, 1.1240e-04, 1.2116e-04, 1.1530e-04, 1.3272e-04, 1.0350e-04, 1.0759e-04, 1.0184e-04], device='cuda:0') 2023-04-16 23:30:30,540 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-16 23:30:30,540 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-16 23:30:30,540 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-16 23:30:30,547 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-16 23:30:31,298 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-16 23:30:31,320 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-16 23:30:31,330 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-16 23:30:51,216 INFO [train.py:893] (0/4) Epoch 18, batch 100, loss[loss=0.177, simple_loss=0.2325, pruned_loss=0.0608, over 13532.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2399, pruned_loss=0.0653, over 1050659.25 frames. ], batch size: 87, lr: 7.38e-03, grad_scale: 16.0 2023-04-16 23:31:05,600 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-46000.pt 2023-04-16 23:31:12,843 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46004.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 23:31:23,505 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.126e+02 2.834e+02 3.248e+02 3.832e+02 7.333e+02, threshold=6.497e+02, percent-clipped=1.0 2023-04-16 23:31:40,952 INFO [train.py:893] (0/4) Epoch 18, batch 150, loss[loss=0.1694, simple_loss=0.237, pruned_loss=0.05089, over 13247.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.243, pruned_loss=0.0667, over 1409786.26 frames. ], batch size: 132, lr: 7.38e-03, grad_scale: 16.0 2023-04-16 23:31:47,305 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4189, 3.2388, 3.9722, 2.8418, 2.5526, 2.8097, 4.1946, 4.3292], device='cuda:0'), covar=tensor([0.1134, 0.1635, 0.0405, 0.1625, 0.1521, 0.1369, 0.0269, 0.0164], device='cuda:0'), in_proj_covar=tensor([0.0238, 0.0260, 0.0187, 0.0217, 0.0213, 0.0177, 0.0197, 0.0166], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:31:58,075 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46052.0, num_to_drop=1, layers_to_drop={1} 2023-04-16 23:32:03,722 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46058.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:32:27,723 INFO [train.py:893] (0/4) Epoch 18, batch 200, loss[loss=0.1782, simple_loss=0.2418, pruned_loss=0.05727, over 13192.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2445, pruned_loss=0.06738, over 1680469.37 frames. ], batch size: 132, lr: 7.37e-03, grad_scale: 16.0 2023-04-16 23:32:56,066 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.742e+02 2.872e+02 3.344e+02 4.213e+02 8.067e+02, threshold=6.687e+02, percent-clipped=4.0 2023-04-16 23:33:08,555 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46130.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:33:12,529 INFO [train.py:893] (0/4) Epoch 18, batch 250, loss[loss=0.1632, simple_loss=0.2258, pruned_loss=0.05035, over 13211.00 frames. ], tot_loss[loss=0.1906, simple_loss=0.2456, pruned_loss=0.06776, over 1900554.91 frames. ], batch size: 132, lr: 7.37e-03, grad_scale: 16.0 2023-04-16 23:34:00,314 INFO [train.py:893] (0/4) Epoch 18, batch 300, loss[loss=0.2033, simple_loss=0.2562, pruned_loss=0.07519, over 11715.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.2466, pruned_loss=0.06783, over 2065700.96 frames. ], batch size: 157, lr: 7.37e-03, grad_scale: 16.0 2023-04-16 23:34:26,037 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3301, 4.8316, 4.8060, 4.7734, 4.5997, 4.5898, 5.3166, 4.8526], device='cuda:0'), covar=tensor([0.0684, 0.1314, 0.1986, 0.2626, 0.0873, 0.1591, 0.0827, 0.1164], device='cuda:0'), in_proj_covar=tensor([0.0304, 0.0355, 0.0445, 0.0451, 0.0268, 0.0329, 0.0407, 0.0324], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:34:29,238 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46215.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:34:29,875 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.905e+02 2.634e+02 3.212e+02 4.032e+02 9.593e+02, threshold=6.425e+02, percent-clipped=2.0 2023-04-16 23:34:46,266 INFO [train.py:893] (0/4) Epoch 18, batch 350, loss[loss=0.1794, simple_loss=0.2399, pruned_loss=0.05944, over 13565.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2473, pruned_loss=0.06829, over 2199025.83 frames. ], batch size: 78, lr: 7.36e-03, grad_scale: 16.0 2023-04-16 23:35:05,109 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7313, 2.4761, 2.5196, 2.8400, 2.1567, 2.9359, 2.9575, 2.4059], device='cuda:0'), covar=tensor([0.0088, 0.0203, 0.0150, 0.0143, 0.0221, 0.0135, 0.0155, 0.0182], device='cuda:0'), in_proj_covar=tensor([0.0095, 0.0104, 0.0110, 0.0106, 0.0120, 0.0097, 0.0098, 0.0097], device='cuda:0'), out_proj_covar=tensor([9.8889e-05, 1.1202e-04, 1.2123e-04, 1.1412e-04, 1.3238e-04, 1.0400e-04, 1.0634e-04, 1.0339e-04], device='cuda:0') 2023-04-16 23:35:26,182 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46276.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:35:33,393 INFO [train.py:893] (0/4) Epoch 18, batch 400, loss[loss=0.1937, simple_loss=0.2491, pruned_loss=0.06914, over 13524.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2475, pruned_loss=0.06801, over 2304483.21 frames. ], batch size: 87, lr: 7.36e-03, grad_scale: 16.0 2023-04-16 23:36:02,175 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.971e+02 2.592e+02 3.217e+02 3.805e+02 6.157e+02, threshold=6.435e+02, percent-clipped=0.0 2023-04-16 23:36:07,459 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0761, 4.3618, 4.0586, 4.1566, 4.1509, 4.5292, 4.3116, 4.1824], device='cuda:0'), covar=tensor([0.0303, 0.0290, 0.0350, 0.0872, 0.0302, 0.0244, 0.0308, 0.0321], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0144, 0.0163, 0.0254, 0.0166, 0.0182, 0.0159, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2023-04-16 23:36:18,731 INFO [train.py:893] (0/4) Epoch 18, batch 450, loss[loss=0.2145, simple_loss=0.2694, pruned_loss=0.07976, over 13395.00 frames. ], tot_loss[loss=0.1923, simple_loss=0.2483, pruned_loss=0.06815, over 2386466.81 frames. ], batch size: 113, lr: 7.35e-03, grad_scale: 16.0 2023-04-16 23:36:21,597 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46337.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:36:39,136 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46358.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:36:43,039 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-16 23:37:01,920 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-16 23:37:04,567 INFO [train.py:893] (0/4) Epoch 18, batch 500, loss[loss=0.2528, simple_loss=0.2885, pruned_loss=0.1085, over 11917.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2484, pruned_loss=0.06778, over 2449464.17 frames. ], batch size: 157, lr: 7.35e-03, grad_scale: 16.0 2023-04-16 23:37:24,792 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46406.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:37:33,006 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.893e+02 2.887e+02 3.261e+02 4.060e+02 6.137e+02, threshold=6.521e+02, percent-clipped=0.0 2023-04-16 23:37:47,272 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46430.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:37:52,105 INFO [train.py:893] (0/4) Epoch 18, batch 550, loss[loss=0.1986, simple_loss=0.2477, pruned_loss=0.0748, over 13189.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.248, pruned_loss=0.06734, over 2495458.15 frames. ], batch size: 58, lr: 7.35e-03, grad_scale: 16.0 2023-04-16 23:38:26,343 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3061, 2.9019, 2.4183, 4.2068, 4.7215, 3.6662, 4.6145, 4.4105], device='cuda:0'), covar=tensor([0.0100, 0.0694, 0.0836, 0.0093, 0.0063, 0.0373, 0.0072, 0.0076], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0085, 0.0092, 0.0075, 0.0062, 0.0077, 0.0052, 0.0072], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-04-16 23:38:31,276 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46478.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:38:36,988 INFO [train.py:893] (0/4) Epoch 18, batch 600, loss[loss=0.2033, simple_loss=0.2483, pruned_loss=0.07908, over 11925.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2469, pruned_loss=0.06684, over 2535825.70 frames. ], batch size: 157, lr: 7.34e-03, grad_scale: 16.0 2023-04-16 23:38:45,700 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-16 23:39:05,380 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46515.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:39:05,948 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.019e+02 2.668e+02 3.118e+02 3.710e+02 6.528e+02, threshold=6.235e+02, percent-clipped=1.0 2023-04-16 23:39:24,376 INFO [train.py:893] (0/4) Epoch 18, batch 650, loss[loss=0.1873, simple_loss=0.2486, pruned_loss=0.06298, over 13217.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2455, pruned_loss=0.06615, over 2565833.06 frames. ], batch size: 132, lr: 7.34e-03, grad_scale: 16.0 2023-04-16 23:39:26,536 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-16 23:39:50,466 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46563.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:39:55,567 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4487, 3.8750, 3.5893, 4.2134, 2.3152, 3.1992, 3.9646, 2.3190], device='cuda:0'), covar=tensor([0.0132, 0.0444, 0.0814, 0.0447, 0.1580, 0.0919, 0.0520, 0.1794], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0180, 0.0201, 0.0234, 0.0181, 0.0196, 0.0176, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:40:09,750 INFO [train.py:893] (0/4) Epoch 18, batch 700, loss[loss=0.1693, simple_loss=0.2342, pruned_loss=0.05224, over 13538.00 frames. ], tot_loss[loss=0.1874, simple_loss=0.2445, pruned_loss=0.06513, over 2589973.69 frames. ], batch size: 98, lr: 7.33e-03, grad_scale: 16.0 2023-04-16 23:40:19,897 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46595.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:40:26,069 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46602.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:40:37,674 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1034, 4.3953, 4.1790, 4.2245, 4.2911, 4.6161, 4.3847, 4.2104], device='cuda:0'), covar=tensor([0.0266, 0.0275, 0.0330, 0.0852, 0.0260, 0.0227, 0.0281, 0.0319], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0144, 0.0163, 0.0253, 0.0165, 0.0181, 0.0159, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2023-04-16 23:40:38,197 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.838e+02 2.538e+02 2.982e+02 3.442e+02 5.381e+02, threshold=5.963e+02, percent-clipped=0.0 2023-04-16 23:40:44,976 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0614, 4.0574, 3.1705, 3.9155, 3.3448, 2.3708, 4.0326, 2.2575], device='cuda:0'), covar=tensor([0.0604, 0.0337, 0.0478, 0.0206, 0.0539, 0.1593, 0.0629, 0.1241], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0137, 0.0133, 0.0113, 0.0150, 0.0187, 0.0166, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:40:53,058 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46632.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:40:55,405 INFO [train.py:893] (0/4) Epoch 18, batch 750, loss[loss=0.1952, simple_loss=0.2518, pruned_loss=0.06929, over 13226.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2444, pruned_loss=0.06534, over 2605420.60 frames. ], batch size: 132, lr: 7.33e-03, grad_scale: 16.0 2023-04-16 23:41:16,463 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1458, 3.9935, 3.3058, 3.8775, 3.2683, 2.4354, 3.9632, 2.3623], device='cuda:0'), covar=tensor([0.0589, 0.0565, 0.0410, 0.0252, 0.0634, 0.1661, 0.1024, 0.1311], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0136, 0.0133, 0.0113, 0.0149, 0.0185, 0.0165, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:41:16,486 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46656.0, num_to_drop=1, layers_to_drop={2} 2023-04-16 23:41:22,208 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46663.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:41:43,679 INFO [train.py:893] (0/4) Epoch 18, batch 800, loss[loss=0.189, simple_loss=0.252, pruned_loss=0.06304, over 13453.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2453, pruned_loss=0.06587, over 2614388.66 frames. ], batch size: 95, lr: 7.33e-03, grad_scale: 16.0 2023-04-16 23:41:49,432 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-04-16 23:42:11,127 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.624e+02 2.676e+02 3.155e+02 3.704e+02 5.990e+02, threshold=6.311e+02, percent-clipped=1.0 2023-04-16 23:42:27,927 INFO [train.py:893] (0/4) Epoch 18, batch 850, loss[loss=0.2044, simple_loss=0.2633, pruned_loss=0.07274, over 13525.00 frames. ], tot_loss[loss=0.1899, simple_loss=0.2467, pruned_loss=0.06654, over 2625807.93 frames. ], batch size: 87, lr: 7.32e-03, grad_scale: 16.0 2023-04-16 23:43:04,200 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.90 vs. limit=5.0 2023-04-16 23:43:14,138 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-16 23:43:15,261 INFO [train.py:893] (0/4) Epoch 18, batch 900, loss[loss=0.1841, simple_loss=0.2382, pruned_loss=0.06505, over 13075.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.247, pruned_loss=0.06695, over 2636426.57 frames. ], batch size: 142, lr: 7.32e-03, grad_scale: 16.0 2023-04-16 23:43:16,654 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-16 23:43:39,074 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46810.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:43:44,621 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.126e+02 2.808e+02 3.282e+02 4.134e+02 5.929e+02, threshold=6.564e+02, percent-clipped=0.0 2023-04-16 23:43:45,567 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-16 23:44:00,550 INFO [train.py:893] (0/4) Epoch 18, batch 950, loss[loss=0.1639, simple_loss=0.2161, pruned_loss=0.0558, over 13528.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2456, pruned_loss=0.06703, over 2639105.11 frames. ], batch size: 70, lr: 7.32e-03, grad_scale: 16.0 2023-04-16 23:44:35,137 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46871.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:44:47,222 INFO [train.py:893] (0/4) Epoch 18, batch 1000, loss[loss=0.1719, simple_loss=0.2339, pruned_loss=0.05496, over 13463.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2437, pruned_loss=0.06615, over 2646313.46 frames. ], batch size: 79, lr: 7.31e-03, grad_scale: 16.0 2023-04-16 23:45:15,983 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.800e+02 2.599e+02 2.993e+02 3.540e+02 6.302e+02, threshold=5.986e+02, percent-clipped=0.0 2023-04-16 23:45:31,001 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46932.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:45:33,832 INFO [train.py:893] (0/4) Epoch 18, batch 1050, loss[loss=0.1982, simple_loss=0.2541, pruned_loss=0.07119, over 13525.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2421, pruned_loss=0.06505, over 2652166.58 frames. ], batch size: 91, lr: 7.31e-03, grad_scale: 16.0 2023-04-16 23:45:39,450 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-16 23:45:47,994 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46951.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 23:45:53,831 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46958.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:46:15,055 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46980.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:46:19,127 INFO [train.py:893] (0/4) Epoch 18, batch 1100, loss[loss=0.1801, simple_loss=0.2386, pruned_loss=0.06081, over 13444.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2417, pruned_loss=0.06408, over 2656857.32 frames. ], batch size: 65, lr: 7.30e-03, grad_scale: 32.0 2023-04-16 23:46:39,529 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9497, 2.0642, 2.3914, 3.3053, 3.0189, 3.3632, 2.7809, 2.2463], device='cuda:0'), covar=tensor([0.0306, 0.0917, 0.0701, 0.0071, 0.0259, 0.0056, 0.0513, 0.0803], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0149, 0.0164, 0.0091, 0.0115, 0.0089, 0.0166, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:46:48,439 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.808e+02 2.562e+02 3.010e+02 3.540e+02 7.955e+02, threshold=6.021e+02, percent-clipped=2.0 2023-04-16 23:47:06,595 INFO [train.py:893] (0/4) Epoch 18, batch 1150, loss[loss=0.1657, simple_loss=0.2346, pruned_loss=0.04845, over 13456.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2409, pruned_loss=0.06309, over 2659792.66 frames. ], batch size: 100, lr: 7.30e-03, grad_scale: 32.0 2023-04-16 23:47:52,593 INFO [train.py:893] (0/4) Epoch 18, batch 1200, loss[loss=0.1768, simple_loss=0.23, pruned_loss=0.06177, over 13564.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2411, pruned_loss=0.06278, over 2660618.96 frames. ], batch size: 78, lr: 7.30e-03, grad_scale: 32.0 2023-04-16 23:48:21,773 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.858e+02 2.660e+02 3.050e+02 3.598e+02 7.489e+02, threshold=6.100e+02, percent-clipped=1.0 2023-04-16 23:48:21,874 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-16 23:48:34,968 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-16 23:48:40,792 INFO [train.py:893] (0/4) Epoch 18, batch 1250, loss[loss=0.1751, simple_loss=0.2406, pruned_loss=0.05475, over 13488.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2426, pruned_loss=0.06384, over 2663668.88 frames. ], batch size: 81, lr: 7.29e-03, grad_scale: 32.0 2023-04-16 23:49:09,911 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47166.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:49:28,076 INFO [train.py:893] (0/4) Epoch 18, batch 1300, loss[loss=0.1787, simple_loss=0.238, pruned_loss=0.05968, over 13542.00 frames. ], tot_loss[loss=0.1869, simple_loss=0.2445, pruned_loss=0.06471, over 2661011.08 frames. ], batch size: 72, lr: 7.29e-03, grad_scale: 32.0 2023-04-16 23:49:52,805 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5994, 2.2844, 2.3295, 2.6966, 1.9668, 2.6640, 2.6397, 2.1112], device='cuda:0'), covar=tensor([0.0077, 0.0184, 0.0198, 0.0115, 0.0217, 0.0129, 0.0188, 0.0197], device='cuda:0'), in_proj_covar=tensor([0.0095, 0.0105, 0.0111, 0.0108, 0.0121, 0.0099, 0.0098, 0.0098], device='cuda:0'), out_proj_covar=tensor([9.8878e-05, 1.1280e-04, 1.2170e-04, 1.1602e-04, 1.3342e-04, 1.0571e-04, 1.0620e-04, 1.0476e-04], device='cuda:0') 2023-04-16 23:49:56,997 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.759e+02 2.746e+02 3.213e+02 3.709e+02 7.669e+02, threshold=6.426e+02, percent-clipped=2.0 2023-04-16 23:50:10,191 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-16 23:50:13,565 INFO [train.py:893] (0/4) Epoch 18, batch 1350, loss[loss=0.156, simple_loss=0.2213, pruned_loss=0.0454, over 13485.00 frames. ], tot_loss[loss=0.1874, simple_loss=0.2451, pruned_loss=0.06484, over 2662706.39 frames. ], batch size: 81, lr: 7.28e-03, grad_scale: 32.0 2023-04-16 23:50:18,496 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-16 23:50:19,825 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47241.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:50:29,521 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47251.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:50:32,004 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7138, 2.3235, 2.4185, 2.8039, 1.9879, 2.7517, 2.7771, 2.2333], device='cuda:0'), covar=tensor([0.0067, 0.0184, 0.0158, 0.0120, 0.0219, 0.0131, 0.0154, 0.0192], device='cuda:0'), in_proj_covar=tensor([0.0095, 0.0106, 0.0111, 0.0108, 0.0122, 0.0099, 0.0098, 0.0098], device='cuda:0'), out_proj_covar=tensor([9.8497e-05, 1.1352e-04, 1.2217e-04, 1.1616e-04, 1.3399e-04, 1.0602e-04, 1.0631e-04, 1.0423e-04], device='cuda:0') 2023-04-16 23:50:35,360 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47258.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:51:00,621 INFO [train.py:893] (0/4) Epoch 18, batch 1400, loss[loss=0.1872, simple_loss=0.2348, pruned_loss=0.06981, over 13438.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2442, pruned_loss=0.06471, over 2660451.27 frames. ], batch size: 65, lr: 7.28e-03, grad_scale: 32.0 2023-04-16 23:51:13,494 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47299.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:51:16,159 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47302.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:51:20,295 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47306.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:51:26,964 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8976, 4.0313, 3.1184, 2.7593, 2.7872, 2.5406, 4.2100, 2.3668], device='cuda:0'), covar=tensor([0.1706, 0.0365, 0.1193, 0.2059, 0.0887, 0.3132, 0.0253, 0.3957], device='cuda:0'), in_proj_covar=tensor([0.0308, 0.0282, 0.0304, 0.0319, 0.0250, 0.0319, 0.0204, 0.0357], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 23:51:29,927 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.066e+02 2.681e+02 3.277e+02 3.724e+02 6.638e+02, threshold=6.553e+02, percent-clipped=2.0 2023-04-16 23:51:46,915 INFO [train.py:893] (0/4) Epoch 18, batch 1450, loss[loss=0.1824, simple_loss=0.2504, pruned_loss=0.05726, over 13486.00 frames. ], tot_loss[loss=0.187, simple_loss=0.2439, pruned_loss=0.06504, over 2657471.17 frames. ], batch size: 81, lr: 7.28e-03, grad_scale: 32.0 2023-04-16 23:52:07,752 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0798, 3.9428, 3.0624, 3.8073, 3.2659, 2.1768, 3.9568, 2.2522], device='cuda:0'), covar=tensor([0.0710, 0.0454, 0.0554, 0.0241, 0.0670, 0.2025, 0.0871, 0.1362], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0135, 0.0131, 0.0112, 0.0147, 0.0185, 0.0164, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:52:29,362 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47379.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:52:34,026 INFO [train.py:893] (0/4) Epoch 18, batch 1500, loss[loss=0.2226, simple_loss=0.2754, pruned_loss=0.08492, over 13527.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.244, pruned_loss=0.06456, over 2658389.19 frames. ], batch size: 91, lr: 7.27e-03, grad_scale: 32.0 2023-04-16 23:52:43,595 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9530, 2.3724, 2.0669, 3.8616, 4.3100, 3.2476, 4.2069, 4.0243], device='cuda:0'), covar=tensor([0.0095, 0.0966, 0.1016, 0.0094, 0.0066, 0.0499, 0.0073, 0.0084], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0087, 0.0094, 0.0078, 0.0063, 0.0079, 0.0053, 0.0074], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 23:53:03,657 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.719e+02 2.645e+02 2.973e+02 3.718e+02 8.279e+02, threshold=5.945e+02, percent-clipped=1.0 2023-04-16 23:53:17,398 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.0809, 4.8690, 5.1175, 4.9214, 5.3893, 4.9093, 5.4155, 5.3212], device='cuda:0'), covar=tensor([0.0372, 0.0520, 0.0655, 0.0556, 0.0553, 0.0860, 0.0472, 0.0554], device='cuda:0'), in_proj_covar=tensor([0.0269, 0.0280, 0.0285, 0.0212, 0.0404, 0.0322, 0.0259, 0.0282], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:53:21,948 INFO [train.py:893] (0/4) Epoch 18, batch 1550, loss[loss=0.2172, simple_loss=0.2589, pruned_loss=0.08775, over 12127.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2431, pruned_loss=0.0641, over 2654393.67 frames. ], batch size: 157, lr: 7.27e-03, grad_scale: 32.0 2023-04-16 23:53:27,182 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47440.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:53:51,361 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47466.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:54:08,484 INFO [train.py:893] (0/4) Epoch 18, batch 1600, loss[loss=0.181, simple_loss=0.2401, pruned_loss=0.061, over 13387.00 frames. ], tot_loss[loss=0.1859, simple_loss=0.2435, pruned_loss=0.06416, over 2653320.78 frames. ], batch size: 62, lr: 7.27e-03, grad_scale: 32.0 2023-04-16 23:54:21,840 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6482, 2.1855, 2.2617, 2.7213, 1.9499, 2.7463, 2.7001, 2.1290], device='cuda:0'), covar=tensor([0.0087, 0.0245, 0.0176, 0.0162, 0.0252, 0.0123, 0.0175, 0.0209], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0110, 0.0116, 0.0113, 0.0126, 0.0103, 0.0102, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-16 23:54:35,009 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47514.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:54:36,507 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.837e+02 2.658e+02 3.121e+02 4.001e+02 6.335e+02, threshold=6.242e+02, percent-clipped=2.0 2023-04-16 23:54:54,974 INFO [train.py:893] (0/4) Epoch 18, batch 1650, loss[loss=0.2061, simple_loss=0.2623, pruned_loss=0.07496, over 13355.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2443, pruned_loss=0.06396, over 2656213.40 frames. ], batch size: 109, lr: 7.26e-03, grad_scale: 32.0 2023-04-16 23:55:21,390 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1833, 4.7224, 4.4549, 4.4542, 4.4995, 4.3155, 4.7413, 4.7634], device='cuda:0'), covar=tensor([0.0213, 0.0190, 0.0198, 0.0278, 0.0248, 0.0244, 0.0244, 0.0202], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0192, 0.0154, 0.0175, 0.0140, 0.0187, 0.0128, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 23:55:32,358 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.4456, 5.2562, 5.4773, 5.2016, 5.7508, 5.2950, 5.7812, 5.7458], device='cuda:0'), covar=tensor([0.0304, 0.0482, 0.0526, 0.0541, 0.0459, 0.0668, 0.0351, 0.0382], device='cuda:0'), in_proj_covar=tensor([0.0272, 0.0283, 0.0288, 0.0213, 0.0407, 0.0325, 0.0260, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-16 23:55:41,260 INFO [train.py:893] (0/4) Epoch 18, batch 1700, loss[loss=0.1689, simple_loss=0.2194, pruned_loss=0.0592, over 13368.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2448, pruned_loss=0.06407, over 2657317.46 frames. ], batch size: 62, lr: 7.26e-03, grad_scale: 32.0 2023-04-16 23:55:47,402 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4667, 2.8007, 2.7924, 4.4174, 4.9090, 3.6575, 4.7734, 4.5710], device='cuda:0'), covar=tensor([0.0072, 0.0797, 0.0759, 0.0087, 0.0047, 0.0417, 0.0070, 0.0071], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0090, 0.0096, 0.0079, 0.0064, 0.0081, 0.0054, 0.0074], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 23:55:52,277 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47597.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:56:09,620 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3688, 3.0536, 3.8254, 2.8423, 2.4450, 2.6254, 4.0763, 4.1685], device='cuda:0'), covar=tensor([0.1171, 0.1737, 0.0357, 0.1531, 0.1511, 0.1471, 0.0293, 0.0247], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0260, 0.0189, 0.0219, 0.0215, 0.0178, 0.0199, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-16 23:56:10,032 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.925e+02 2.763e+02 3.195e+02 3.652e+02 5.835e+02, threshold=6.391e+02, percent-clipped=0.0 2023-04-16 23:56:27,289 INFO [train.py:893] (0/4) Epoch 18, batch 1750, loss[loss=0.1741, simple_loss=0.2379, pruned_loss=0.05515, over 13456.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2442, pruned_loss=0.06356, over 2659366.92 frames. ], batch size: 103, lr: 7.25e-03, grad_scale: 32.0 2023-04-16 23:57:14,854 INFO [train.py:893] (0/4) Epoch 18, batch 1800, loss[loss=0.1801, simple_loss=0.2291, pruned_loss=0.06556, over 13214.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2435, pruned_loss=0.06315, over 2660677.70 frames. ], batch size: 58, lr: 7.25e-03, grad_scale: 32.0 2023-04-16 23:57:43,293 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.712e+02 2.641e+02 3.042e+02 3.514e+02 5.198e+02, threshold=6.083e+02, percent-clipped=0.0 2023-04-16 23:57:59,987 INFO [train.py:893] (0/4) Epoch 18, batch 1850, loss[loss=0.2037, simple_loss=0.2539, pruned_loss=0.07675, over 13206.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2429, pruned_loss=0.06272, over 2659642.15 frames. ], batch size: 132, lr: 7.25e-03, grad_scale: 32.0 2023-04-16 23:58:01,005 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47735.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:58:04,697 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47738.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:58:05,309 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-16 23:58:48,063 INFO [train.py:893] (0/4) Epoch 18, batch 1900, loss[loss=0.1874, simple_loss=0.245, pruned_loss=0.06488, over 13532.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2428, pruned_loss=0.06287, over 2664358.29 frames. ], batch size: 76, lr: 7.24e-03, grad_scale: 32.0 2023-04-16 23:58:49,126 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0784, 2.3782, 2.2026, 3.8172, 4.3353, 3.2271, 4.2397, 4.0953], device='cuda:0'), covar=tensor([0.0087, 0.0956, 0.0996, 0.0107, 0.0088, 0.0485, 0.0100, 0.0082], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0089, 0.0096, 0.0079, 0.0064, 0.0080, 0.0054, 0.0074], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-16 23:58:50,828 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47788.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:59:00,830 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47799.0, num_to_drop=0, layers_to_drop=set() 2023-04-16 23:59:10,478 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47809.0, num_to_drop=1, layers_to_drop={0} 2023-04-16 23:59:16,789 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.811e+02 2.747e+02 3.190e+02 3.655e+02 6.040e+02, threshold=6.381e+02, percent-clipped=0.0 2023-04-16 23:59:34,993 INFO [train.py:893] (0/4) Epoch 18, batch 1950, loss[loss=0.2058, simple_loss=0.2585, pruned_loss=0.07654, over 13364.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.242, pruned_loss=0.06254, over 2664240.37 frames. ], batch size: 73, lr: 7.24e-03, grad_scale: 32.0 2023-04-16 23:59:48,044 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0674, 2.0632, 3.7474, 3.6559, 3.6587, 2.9453, 3.4970, 2.7940], device='cuda:0'), covar=tensor([0.1880, 0.1359, 0.0162, 0.0169, 0.0233, 0.0701, 0.0231, 0.0830], device='cuda:0'), in_proj_covar=tensor([0.0212, 0.0182, 0.0116, 0.0121, 0.0129, 0.0171, 0.0136, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-16 23:59:48,074 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47849.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:00:07,784 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47870.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 00:00:20,510 INFO [train.py:893] (0/4) Epoch 18, batch 2000, loss[loss=0.2064, simple_loss=0.2656, pruned_loss=0.07355, over 13424.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2436, pruned_loss=0.06343, over 2661667.20 frames. ], batch size: 95, lr: 7.24e-03, grad_scale: 32.0 2023-04-17 00:00:26,474 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 00:00:32,467 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47897.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:00:37,449 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8782, 4.7104, 4.9171, 4.8341, 5.1839, 4.7217, 5.1805, 5.1476], device='cuda:0'), covar=tensor([0.0378, 0.0503, 0.0689, 0.0488, 0.0556, 0.0818, 0.0459, 0.0505], device='cuda:0'), in_proj_covar=tensor([0.0274, 0.0283, 0.0290, 0.0216, 0.0409, 0.0327, 0.0260, 0.0286], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:00:49,669 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.108e+02 2.833e+02 3.370e+02 4.001e+02 7.543e+02, threshold=6.741e+02, percent-clipped=2.0 2023-04-17 00:00:51,013 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8781, 2.5162, 2.5681, 3.0680, 2.3629, 3.0297, 2.9332, 2.4796], device='cuda:0'), covar=tensor([0.0094, 0.0173, 0.0145, 0.0133, 0.0182, 0.0111, 0.0177, 0.0165], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0106, 0.0113, 0.0110, 0.0121, 0.0099, 0.0099, 0.0097], device='cuda:0'), out_proj_covar=tensor([9.9211e-05, 1.1324e-04, 1.2349e-04, 1.1851e-04, 1.3282e-04, 1.0551e-04, 1.0639e-04, 1.0337e-04], device='cuda:0') 2023-04-17 00:00:59,555 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0469, 4.2952, 3.4226, 2.8952, 3.0727, 2.5539, 4.5313, 2.4796], device='cuda:0'), covar=tensor([0.1656, 0.0336, 0.1017, 0.1975, 0.0791, 0.3181, 0.0200, 0.3942], device='cuda:0'), in_proj_covar=tensor([0.0306, 0.0281, 0.0303, 0.0317, 0.0250, 0.0318, 0.0205, 0.0356], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:01:03,556 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6628, 4.4687, 4.7015, 4.6311, 4.9606, 4.4994, 4.9409, 4.9253], device='cuda:0'), covar=tensor([0.0393, 0.0538, 0.0617, 0.0530, 0.0510, 0.0806, 0.0403, 0.0439], device='cuda:0'), in_proj_covar=tensor([0.0272, 0.0282, 0.0288, 0.0215, 0.0407, 0.0326, 0.0259, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:01:08,132 INFO [train.py:893] (0/4) Epoch 18, batch 2050, loss[loss=0.1852, simple_loss=0.2476, pruned_loss=0.0614, over 13426.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2449, pruned_loss=0.06426, over 2656490.06 frames. ], batch size: 106, lr: 7.23e-03, grad_scale: 32.0 2023-04-17 00:01:10,818 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0106, 4.5303, 4.3184, 4.2546, 4.2939, 4.1503, 4.5575, 4.5687], device='cuda:0'), covar=tensor([0.0226, 0.0221, 0.0193, 0.0346, 0.0254, 0.0269, 0.0267, 0.0244], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0195, 0.0154, 0.0176, 0.0139, 0.0189, 0.0128, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:01:11,876 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-17 00:01:16,685 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47945.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:01:36,629 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4145, 2.1279, 2.6638, 3.8166, 3.5206, 3.8881, 3.0671, 2.1794], device='cuda:0'), covar=tensor([0.0251, 0.1070, 0.0759, 0.0060, 0.0197, 0.0058, 0.0615, 0.1052], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0153, 0.0167, 0.0093, 0.0118, 0.0092, 0.0172, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:01:54,813 INFO [train.py:893] (0/4) Epoch 18, batch 2100, loss[loss=0.1649, simple_loss=0.2332, pruned_loss=0.0483, over 13492.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2444, pruned_loss=0.06406, over 2653760.80 frames. ], batch size: 93, lr: 7.23e-03, grad_scale: 32.0 2023-04-17 00:02:09,138 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-48000.pt 2023-04-17 00:02:27,054 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.728e+02 2.550e+02 3.091e+02 3.728e+02 7.028e+02, threshold=6.181e+02, percent-clipped=2.0 2023-04-17 00:02:35,738 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2409, 4.2966, 2.9172, 4.0535, 4.1681, 2.7737, 3.8881, 2.7916], device='cuda:0'), covar=tensor([0.0288, 0.0253, 0.1135, 0.0374, 0.0262, 0.1159, 0.0423, 0.1380], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0169, 0.0177, 0.0196, 0.0135, 0.0160, 0.0159, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:02:45,204 INFO [train.py:893] (0/4) Epoch 18, batch 2150, loss[loss=0.1715, simple_loss=0.2272, pruned_loss=0.05793, over 13340.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2427, pruned_loss=0.06296, over 2656961.85 frames. ], batch size: 67, lr: 7.22e-03, grad_scale: 32.0 2023-04-17 00:02:45,524 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48035.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:03:20,033 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48072.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:03:30,776 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48083.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:03:32,349 INFO [train.py:893] (0/4) Epoch 18, batch 2200, loss[loss=0.1987, simple_loss=0.2507, pruned_loss=0.07335, over 13520.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2411, pruned_loss=0.06177, over 2659742.71 frames. ], batch size: 91, lr: 7.22e-03, grad_scale: 32.0 2023-04-17 00:03:40,740 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48094.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:04:00,791 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.580e+02 2.476e+02 3.037e+02 3.602e+02 5.277e+02, threshold=6.074e+02, percent-clipped=0.0 2023-04-17 00:04:05,073 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9198, 3.7537, 3.0725, 3.5477, 3.1567, 2.1262, 3.7844, 2.1301], device='cuda:0'), covar=tensor([0.0666, 0.0505, 0.0480, 0.0345, 0.0591, 0.2070, 0.0972, 0.1381], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0135, 0.0131, 0.0112, 0.0148, 0.0187, 0.0165, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:04:15,849 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48133.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:04:17,172 INFO [train.py:893] (0/4) Epoch 18, batch 2250, loss[loss=0.1766, simple_loss=0.2379, pruned_loss=0.05761, over 13515.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2398, pruned_loss=0.06136, over 2657266.17 frames. ], batch size: 98, lr: 7.22e-03, grad_scale: 32.0 2023-04-17 00:04:26,581 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48144.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:04:45,128 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48165.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 00:04:57,965 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7691, 2.3345, 2.5138, 2.8884, 2.2263, 2.9425, 2.9883, 2.3640], device='cuda:0'), covar=tensor([0.0080, 0.0220, 0.0131, 0.0147, 0.0196, 0.0119, 0.0129, 0.0172], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0108, 0.0116, 0.0113, 0.0124, 0.0102, 0.0102, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 00:05:03,553 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6467, 3.5333, 2.8666, 3.3540, 2.9747, 2.0690, 3.5993, 1.9791], device='cuda:0'), covar=tensor([0.0845, 0.0627, 0.0554, 0.0377, 0.0714, 0.2201, 0.1095, 0.1534], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0135, 0.0131, 0.0111, 0.0147, 0.0186, 0.0164, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:05:04,047 INFO [train.py:893] (0/4) Epoch 18, batch 2300, loss[loss=0.1768, simple_loss=0.2366, pruned_loss=0.05855, over 13486.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2397, pruned_loss=0.06129, over 2662482.90 frames. ], batch size: 100, lr: 7.21e-03, grad_scale: 32.0 2023-04-17 00:05:06,268 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-17 00:05:30,038 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48212.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 00:05:33,716 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.863e+02 2.589e+02 2.983e+02 3.618e+02 6.211e+02, threshold=5.966e+02, percent-clipped=2.0 2023-04-17 00:05:49,820 INFO [train.py:893] (0/4) Epoch 18, batch 2350, loss[loss=0.2274, simple_loss=0.2667, pruned_loss=0.0941, over 11845.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.2402, pruned_loss=0.06138, over 2666185.83 frames. ], batch size: 157, lr: 7.21e-03, grad_scale: 32.0 2023-04-17 00:05:50,919 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6075, 2.2705, 2.3434, 2.6583, 2.1021, 2.6829, 2.6586, 2.2830], device='cuda:0'), covar=tensor([0.0082, 0.0206, 0.0151, 0.0144, 0.0184, 0.0144, 0.0196, 0.0173], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0108, 0.0115, 0.0113, 0.0124, 0.0102, 0.0102, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 00:06:12,419 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 00:06:26,050 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48273.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 00:06:36,452 INFO [train.py:893] (0/4) Epoch 18, batch 2400, loss[loss=0.1973, simple_loss=0.2532, pruned_loss=0.07072, over 13394.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.2403, pruned_loss=0.06172, over 2663851.28 frames. ], batch size: 113, lr: 7.21e-03, grad_scale: 32.0 2023-04-17 00:07:05,451 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.915e+02 2.726e+02 3.153e+02 3.503e+02 4.945e+02, threshold=6.306e+02, percent-clipped=0.0 2023-04-17 00:07:12,440 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48324.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:07:22,287 INFO [train.py:893] (0/4) Epoch 18, batch 2450, loss[loss=0.2156, simple_loss=0.2715, pruned_loss=0.07979, over 13464.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2406, pruned_loss=0.0618, over 2663326.99 frames. ], batch size: 100, lr: 7.20e-03, grad_scale: 32.0 2023-04-17 00:07:27,608 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3937, 3.5901, 3.5662, 4.0011, 2.2122, 3.0202, 3.8145, 2.1505], device='cuda:0'), covar=tensor([0.0140, 0.0496, 0.0731, 0.0556, 0.1546, 0.0962, 0.0487, 0.1825], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0181, 0.0200, 0.0233, 0.0181, 0.0196, 0.0177, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:07:55,046 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1032, 3.8271, 4.0572, 2.6267, 4.3752, 4.0903, 4.0934, 4.3247], device='cuda:0'), covar=tensor([0.0244, 0.0161, 0.0142, 0.1091, 0.0141, 0.0239, 0.0141, 0.0118], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0051, 0.0078, 0.0100, 0.0095, 0.0100, 0.0076, 0.0071], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:07:56,091 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2023-04-17 00:08:09,077 INFO [train.py:893] (0/4) Epoch 18, batch 2500, loss[loss=0.1887, simple_loss=0.243, pruned_loss=0.06721, over 13249.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2408, pruned_loss=0.0619, over 2664651.10 frames. ], batch size: 124, lr: 7.20e-03, grad_scale: 32.0 2023-04-17 00:08:09,390 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48385.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:08:16,877 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48394.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:08:21,133 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48397.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:08:38,951 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.890e+02 2.803e+02 3.213e+02 3.680e+02 8.016e+02, threshold=6.426e+02, percent-clipped=1.0 2023-04-17 00:08:45,214 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48423.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:08:50,111 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48428.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:08:57,167 INFO [train.py:893] (0/4) Epoch 18, batch 2550, loss[loss=0.1918, simple_loss=0.2476, pruned_loss=0.06806, over 13480.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2409, pruned_loss=0.06222, over 2664243.64 frames. ], batch size: 70, lr: 7.19e-03, grad_scale: 32.0 2023-04-17 00:09:03,262 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48442.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:09:04,972 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48444.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:09:17,668 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 00:09:17,877 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48458.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:09:26,030 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48465.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 00:09:41,859 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48484.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:09:42,391 INFO [train.py:893] (0/4) Epoch 18, batch 2600, loss[loss=0.184, simple_loss=0.2426, pruned_loss=0.06273, over 13529.00 frames. ], tot_loss[loss=0.1831, simple_loss=0.2412, pruned_loss=0.06247, over 2666820.05 frames. ], batch size: 76, lr: 7.19e-03, grad_scale: 32.0 2023-04-17 00:09:43,708 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.53 vs. limit=5.0 2023-04-17 00:09:50,698 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48492.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:10:08,213 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48513.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 00:10:08,506 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.45 vs. limit=5.0 2023-04-17 00:10:10,307 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.856e+02 2.747e+02 3.259e+02 4.082e+02 6.442e+02, threshold=6.518e+02, percent-clipped=1.0 2023-04-17 00:10:24,720 INFO [train.py:893] (0/4) Epoch 18, batch 2650, loss[loss=0.183, simple_loss=0.2481, pruned_loss=0.059, over 13582.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2422, pruned_loss=0.06309, over 2656893.81 frames. ], batch size: 89, lr: 7.19e-03, grad_scale: 32.0 2023-04-17 00:10:39,825 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2023-04-17 00:10:43,890 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4772, 4.2158, 4.5272, 4.5248, 4.7780, 4.2903, 4.8239, 4.7231], device='cuda:0'), covar=tensor([0.0472, 0.0629, 0.0653, 0.0518, 0.0556, 0.0884, 0.0458, 0.0527], device='cuda:0'), in_proj_covar=tensor([0.0275, 0.0286, 0.0291, 0.0216, 0.0412, 0.0331, 0.0264, 0.0287], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:10:49,939 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48568.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 00:11:02,923 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-18.pt 2023-04-17 00:11:28,249 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-17 00:11:32,188 INFO [train.py:893] (0/4) Epoch 19, batch 0, loss[loss=0.2092, simple_loss=0.2591, pruned_loss=0.07962, over 13399.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2591, pruned_loss=0.07962, over 13399.00 frames. ], batch size: 113, lr: 6.99e-03, grad_scale: 32.0 2023-04-17 00:11:32,189 INFO [train.py:918] (0/4) Computing validation loss 2023-04-17 00:11:50,999 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2391, 3.9810, 4.2654, 2.8366, 4.5867, 4.2569, 4.3537, 4.4463], device='cuda:0'), covar=tensor([0.0240, 0.0158, 0.0143, 0.0992, 0.0135, 0.0236, 0.0124, 0.0122], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0050, 0.0076, 0.0097, 0.0093, 0.0097, 0.0074, 0.0069], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:11:53,919 INFO [train.py:927] (0/4) Epoch 19, validation: loss=0.1405, simple_loss=0.1995, pruned_loss=0.04077, over 2446609.00 frames. 2023-04-17 00:11:53,919 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-17 00:12:23,917 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.860e+02 2.769e+02 3.155e+02 3.700e+02 8.510e+02, threshold=6.310e+02, percent-clipped=1.0 2023-04-17 00:12:31,745 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3552, 4.1541, 4.2607, 2.9624, 4.7020, 4.3508, 4.4448, 4.6162], device='cuda:0'), covar=tensor([0.0243, 0.0130, 0.0142, 0.0937, 0.0142, 0.0250, 0.0134, 0.0098], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0051, 0.0076, 0.0098, 0.0094, 0.0098, 0.0074, 0.0069], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:12:41,455 INFO [train.py:893] (0/4) Epoch 19, batch 50, loss[loss=0.1916, simple_loss=0.2413, pruned_loss=0.07089, over 13440.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2338, pruned_loss=0.06176, over 600505.08 frames. ], batch size: 65, lr: 6.99e-03, grad_scale: 32.0 2023-04-17 00:12:47,680 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.2180, 2.0719, 3.7411, 3.6305, 3.6940, 2.8910, 3.4604, 2.7217], device='cuda:0'), covar=tensor([0.1648, 0.1362, 0.0122, 0.0153, 0.0185, 0.0676, 0.0225, 0.0797], device='cuda:0'), in_proj_covar=tensor([0.0214, 0.0182, 0.0115, 0.0122, 0.0128, 0.0170, 0.0137, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 00:13:06,404 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-17 00:13:06,405 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-17 00:13:06,405 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-17 00:13:06,412 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-17 00:13:07,185 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-17 00:13:07,204 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-17 00:13:07,228 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-17 00:13:23,761 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48680.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:13:26,844 INFO [train.py:893] (0/4) Epoch 19, batch 100, loss[loss=0.2163, simple_loss=0.2727, pruned_loss=0.07999, over 13527.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.237, pruned_loss=0.06272, over 1062283.96 frames. ], batch size: 98, lr: 6.98e-03, grad_scale: 32.0 2023-04-17 00:13:53,117 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0533, 4.5425, 4.3232, 4.3176, 4.3495, 4.1491, 4.6076, 4.6350], device='cuda:0'), covar=tensor([0.0357, 0.0355, 0.0326, 0.0496, 0.0404, 0.0430, 0.0323, 0.0298], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0194, 0.0153, 0.0174, 0.0139, 0.0190, 0.0127, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:13:56,997 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.794e+02 2.831e+02 3.394e+02 4.079e+02 8.667e+02, threshold=6.787e+02, percent-clipped=3.0 2023-04-17 00:14:08,907 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48728.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:14:14,723 INFO [train.py:893] (0/4) Epoch 19, batch 150, loss[loss=0.1939, simple_loss=0.2479, pruned_loss=0.06995, over 13567.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2412, pruned_loss=0.0648, over 1409364.33 frames. ], batch size: 89, lr: 6.98e-03, grad_scale: 32.0 2023-04-17 00:14:15,104 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7636, 3.8134, 2.7549, 3.5605, 3.7253, 2.3915, 3.3739, 2.5641], device='cuda:0'), covar=tensor([0.0318, 0.0298, 0.1088, 0.0369, 0.0272, 0.1259, 0.0549, 0.1404], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0170, 0.0177, 0.0199, 0.0135, 0.0160, 0.0159, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:14:15,191 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-17 00:14:31,378 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48753.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:14:49,130 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7941, 3.8267, 2.9593, 2.5929, 2.7849, 2.3550, 3.8903, 2.2225], device='cuda:0'), covar=tensor([0.1572, 0.0370, 0.1183, 0.2118, 0.0839, 0.3230, 0.0284, 0.3856], device='cuda:0'), in_proj_covar=tensor([0.0311, 0.0285, 0.0309, 0.0324, 0.0253, 0.0324, 0.0208, 0.0363], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:14:53,795 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48776.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:14:56,384 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48779.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:15:00,328 INFO [train.py:893] (0/4) Epoch 19, batch 200, loss[loss=0.1949, simple_loss=0.2452, pruned_loss=0.07229, over 13183.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.2438, pruned_loss=0.06575, over 1683939.69 frames. ], batch size: 132, lr: 6.98e-03, grad_scale: 32.0 2023-04-17 00:15:17,170 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-17 00:15:30,381 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.935e+02 2.775e+02 3.304e+02 3.931e+02 5.890e+02, threshold=6.608e+02, percent-clipped=0.0 2023-04-17 00:15:35,683 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48821.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:15:47,951 INFO [train.py:893] (0/4) Epoch 19, batch 250, loss[loss=0.1788, simple_loss=0.2302, pruned_loss=0.06363, over 13424.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2439, pruned_loss=0.06528, over 1893750.60 frames. ], batch size: 65, lr: 6.97e-03, grad_scale: 32.0 2023-04-17 00:15:52,202 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48838.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:16:02,275 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.7726, 4.6048, 4.5258, 4.5477, 5.0702, 4.1397, 4.9597, 5.0056], device='cuda:0'), covar=tensor([0.0769, 0.1101, 0.1508, 0.1044, 0.1119, 0.1975, 0.0998, 0.0998], device='cuda:0'), in_proj_covar=tensor([0.0272, 0.0282, 0.0288, 0.0214, 0.0406, 0.0325, 0.0262, 0.0283], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:16:20,186 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48868.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 00:16:31,851 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48882.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:16:34,314 INFO [train.py:893] (0/4) Epoch 19, batch 300, loss[loss=0.1823, simple_loss=0.2435, pruned_loss=0.06051, over 13542.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2439, pruned_loss=0.06519, over 2064962.69 frames. ], batch size: 78, lr: 6.97e-03, grad_scale: 32.0 2023-04-17 00:16:49,280 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48899.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:17:01,299 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-17 00:17:04,063 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.909e+02 2.666e+02 3.189e+02 3.769e+02 7.720e+02, threshold=6.378e+02, percent-clipped=3.0 2023-04-17 00:17:04,289 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48916.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 00:17:21,370 INFO [train.py:893] (0/4) Epoch 19, batch 350, loss[loss=0.1817, simple_loss=0.2306, pruned_loss=0.06639, over 11514.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2444, pruned_loss=0.06531, over 2197941.90 frames. ], batch size: 47, lr: 6.96e-03, grad_scale: 32.0 2023-04-17 00:17:38,192 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5567, 4.3692, 4.4409, 2.6996, 4.7897, 4.6336, 4.5763, 4.8575], device='cuda:0'), covar=tensor([0.0227, 0.0115, 0.0134, 0.1151, 0.0136, 0.0197, 0.0123, 0.0076], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0051, 0.0077, 0.0099, 0.0094, 0.0099, 0.0075, 0.0071], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:18:04,421 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48980.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:18:08,493 INFO [train.py:893] (0/4) Epoch 19, batch 400, loss[loss=0.1874, simple_loss=0.2441, pruned_loss=0.06536, over 13532.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2448, pruned_loss=0.06537, over 2300290.61 frames. ], batch size: 76, lr: 6.96e-03, grad_scale: 64.0 2023-04-17 00:18:38,844 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 2.600e+02 3.085e+02 3.830e+02 7.566e+02, threshold=6.170e+02, percent-clipped=2.0 2023-04-17 00:18:49,720 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49028.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:18:54,459 INFO [train.py:893] (0/4) Epoch 19, batch 450, loss[loss=0.1597, simple_loss=0.2215, pruned_loss=0.04894, over 13359.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2464, pruned_loss=0.06607, over 2377036.08 frames. ], batch size: 73, lr: 6.96e-03, grad_scale: 64.0 2023-04-17 00:18:56,644 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-17 00:19:12,836 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49053.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:19:21,054 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-17 00:19:38,113 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49079.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:19:41,980 INFO [train.py:893] (0/4) Epoch 19, batch 500, loss[loss=0.1594, simple_loss=0.2217, pruned_loss=0.04853, over 13521.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2457, pruned_loss=0.06524, over 2438340.50 frames. ], batch size: 70, lr: 6.95e-03, grad_scale: 64.0 2023-04-17 00:19:56,939 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49101.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:20:12,368 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.858e+02 2.668e+02 2.993e+02 3.591e+02 6.396e+02, threshold=5.987e+02, percent-clipped=1.0 2023-04-17 00:20:19,396 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1273, 4.9164, 5.1705, 4.9444, 5.4531, 4.8979, 5.4770, 5.3934], device='cuda:0'), covar=tensor([0.0351, 0.0587, 0.0621, 0.0635, 0.0547, 0.0907, 0.0456, 0.0491], device='cuda:0'), in_proj_covar=tensor([0.0273, 0.0283, 0.0288, 0.0214, 0.0407, 0.0326, 0.0262, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:20:21,842 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49127.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:20:28,417 INFO [train.py:893] (0/4) Epoch 19, batch 550, loss[loss=0.1832, simple_loss=0.2321, pruned_loss=0.06717, over 12578.00 frames. ], tot_loss[loss=0.1876, simple_loss=0.2456, pruned_loss=0.06483, over 2488631.39 frames. ], batch size: 51, lr: 6.95e-03, grad_scale: 32.0 2023-04-17 00:21:09,539 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49177.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:21:16,001 INFO [train.py:893] (0/4) Epoch 19, batch 600, loss[loss=0.1818, simple_loss=0.2441, pruned_loss=0.0598, over 13460.00 frames. ], tot_loss[loss=0.186, simple_loss=0.2437, pruned_loss=0.06419, over 2527659.59 frames. ], batch size: 100, lr: 6.95e-03, grad_scale: 32.0 2023-04-17 00:21:22,056 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2238, 4.6911, 4.6323, 4.6804, 4.4717, 4.5003, 5.1987, 4.8083], device='cuda:0'), covar=tensor([0.0629, 0.1230, 0.1994, 0.2399, 0.1115, 0.1459, 0.0823, 0.0992], device='cuda:0'), in_proj_covar=tensor([0.0314, 0.0363, 0.0446, 0.0454, 0.0280, 0.0333, 0.0414, 0.0331], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:21:24,550 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49194.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:21:47,110 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.916e+02 2.549e+02 2.912e+02 3.543e+02 6.651e+02, threshold=5.825e+02, percent-clipped=1.0 2023-04-17 00:22:02,905 INFO [train.py:893] (0/4) Epoch 19, batch 650, loss[loss=0.1958, simple_loss=0.2515, pruned_loss=0.07009, over 13534.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.243, pruned_loss=0.06388, over 2559044.88 frames. ], batch size: 91, lr: 6.94e-03, grad_scale: 32.0 2023-04-17 00:22:04,812 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49236.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:22:49,392 INFO [train.py:893] (0/4) Epoch 19, batch 700, loss[loss=0.1727, simple_loss=0.2319, pruned_loss=0.05672, over 13561.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2413, pruned_loss=0.06283, over 2585206.19 frames. ], batch size: 89, lr: 6.94e-03, grad_scale: 32.0 2023-04-17 00:23:02,190 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49297.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:23:20,166 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.748e+02 2.893e+02 3.409e+02 4.077e+02 6.619e+02, threshold=6.818e+02, percent-clipped=5.0 2023-04-17 00:23:24,100 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-17 00:23:26,633 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0804, 4.2190, 3.2872, 2.9580, 3.0657, 2.5256, 4.3334, 2.5048], device='cuda:0'), covar=tensor([0.1612, 0.0359, 0.1160, 0.1855, 0.0823, 0.3220, 0.0233, 0.3908], device='cuda:0'), in_proj_covar=tensor([0.0310, 0.0283, 0.0308, 0.0321, 0.0252, 0.0322, 0.0208, 0.0361], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:23:36,840 INFO [train.py:893] (0/4) Epoch 19, batch 750, loss[loss=0.1797, simple_loss=0.2398, pruned_loss=0.05978, over 13530.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2413, pruned_loss=0.0634, over 2600890.94 frames. ], batch size: 87, lr: 6.94e-03, grad_scale: 32.0 2023-04-17 00:24:23,051 INFO [train.py:893] (0/4) Epoch 19, batch 800, loss[loss=0.1987, simple_loss=0.2601, pruned_loss=0.06862, over 13518.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2432, pruned_loss=0.06444, over 2614499.06 frames. ], batch size: 91, lr: 6.93e-03, grad_scale: 32.0 2023-04-17 00:24:23,428 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9101, 4.2782, 3.9944, 4.6455, 2.9973, 3.5868, 4.3341, 2.7358], device='cuda:0'), covar=tensor([0.0130, 0.0370, 0.0615, 0.0573, 0.1168, 0.0760, 0.0364, 0.1441], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0178, 0.0197, 0.0228, 0.0177, 0.0192, 0.0172, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:24:55,575 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.943e+02 2.825e+02 3.338e+02 3.802e+02 7.720e+02, threshold=6.675e+02, percent-clipped=1.0 2023-04-17 00:24:59,991 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49422.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:25:09,066 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49432.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 00:25:10,428 INFO [train.py:893] (0/4) Epoch 19, batch 850, loss[loss=0.1847, simple_loss=0.2487, pruned_loss=0.06039, over 13397.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2448, pruned_loss=0.06488, over 2625739.35 frames. ], batch size: 113, lr: 6.93e-03, grad_scale: 32.0 2023-04-17 00:25:52,385 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49477.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:25:57,519 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49483.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:25:57,989 INFO [train.py:893] (0/4) Epoch 19, batch 900, loss[loss=0.2209, simple_loss=0.2541, pruned_loss=0.09382, over 11975.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2454, pruned_loss=0.06583, over 2634019.45 frames. ], batch size: 157, lr: 6.93e-03, grad_scale: 32.0 2023-04-17 00:26:06,724 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49493.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 00:26:07,389 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49494.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:26:28,268 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.933e+02 2.628e+02 3.112e+02 3.839e+02 6.691e+02, threshold=6.223e+02, percent-clipped=1.0 2023-04-17 00:26:28,338 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-17 00:26:36,310 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49525.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:26:44,228 INFO [train.py:893] (0/4) Epoch 19, batch 950, loss[loss=0.2108, simple_loss=0.2548, pruned_loss=0.0834, over 13209.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.245, pruned_loss=0.06594, over 2642364.33 frames. ], batch size: 132, lr: 6.92e-03, grad_scale: 32.0 2023-04-17 00:26:52,493 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49542.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:27:12,973 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-17 00:27:30,890 INFO [train.py:893] (0/4) Epoch 19, batch 1000, loss[loss=0.161, simple_loss=0.2222, pruned_loss=0.04994, over 13350.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2427, pruned_loss=0.06495, over 2644719.58 frames. ], batch size: 73, lr: 6.92e-03, grad_scale: 32.0 2023-04-17 00:27:39,090 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49592.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:27:47,985 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5842, 2.3100, 2.3189, 2.6717, 2.0078, 2.7269, 2.7272, 2.2373], device='cuda:0'), covar=tensor([0.0080, 0.0211, 0.0160, 0.0146, 0.0235, 0.0123, 0.0160, 0.0177], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0107, 0.0116, 0.0113, 0.0123, 0.0101, 0.0101, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 00:28:00,487 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9034, 4.2362, 4.0459, 4.0469, 4.0521, 3.9254, 4.2794, 4.3414], device='cuda:0'), covar=tensor([0.0213, 0.0254, 0.0216, 0.0349, 0.0277, 0.0281, 0.0252, 0.0231], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0195, 0.0156, 0.0176, 0.0142, 0.0191, 0.0128, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:28:01,914 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.985e+02 2.718e+02 3.260e+02 3.856e+02 5.927e+02, threshold=6.520e+02, percent-clipped=0.0 2023-04-17 00:28:17,894 INFO [train.py:893] (0/4) Epoch 19, batch 1050, loss[loss=0.1742, simple_loss=0.2389, pruned_loss=0.05477, over 13419.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.2412, pruned_loss=0.06374, over 2647947.61 frames. ], batch size: 95, lr: 6.92e-03, grad_scale: 32.0 2023-04-17 00:29:04,577 INFO [train.py:893] (0/4) Epoch 19, batch 1100, loss[loss=0.1636, simple_loss=0.2206, pruned_loss=0.05329, over 13359.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2414, pruned_loss=0.06303, over 2646978.88 frames. ], batch size: 67, lr: 6.91e-03, grad_scale: 32.0 2023-04-17 00:29:25,650 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-17 00:29:25,748 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.59 vs. limit=5.0 2023-04-17 00:29:31,240 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7426, 3.9163, 2.9278, 2.5477, 2.7448, 2.3717, 3.9236, 2.2110], device='cuda:0'), covar=tensor([0.1793, 0.0377, 0.1223, 0.2222, 0.0882, 0.3333, 0.0281, 0.4151], device='cuda:0'), in_proj_covar=tensor([0.0312, 0.0286, 0.0309, 0.0324, 0.0254, 0.0323, 0.0208, 0.0362], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:29:35,044 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.663e+02 2.716e+02 3.149e+02 3.915e+02 7.191e+02, threshold=6.299e+02, percent-clipped=1.0 2023-04-17 00:29:52,604 INFO [train.py:893] (0/4) Epoch 19, batch 1150, loss[loss=0.1945, simple_loss=0.2541, pruned_loss=0.06747, over 13520.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2407, pruned_loss=0.06231, over 2648179.45 frames. ], batch size: 91, lr: 6.91e-03, grad_scale: 32.0 2023-04-17 00:30:16,465 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.30 vs. limit=5.0 2023-04-17 00:30:22,728 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49766.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:30:32,617 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49778.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:30:39,127 INFO [train.py:893] (0/4) Epoch 19, batch 1200, loss[loss=0.1987, simple_loss=0.2625, pruned_loss=0.06743, over 13353.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.241, pruned_loss=0.06218, over 2649064.72 frames. ], batch size: 118, lr: 6.91e-03, grad_scale: 32.0 2023-04-17 00:30:42,757 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49788.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 00:31:07,011 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-17 00:31:10,212 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.639e+02 2.701e+02 3.177e+02 3.768e+02 6.623e+02, threshold=6.354e+02, percent-clipped=1.0 2023-04-17 00:31:19,519 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-17 00:31:20,640 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49827.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:31:26,193 INFO [train.py:893] (0/4) Epoch 19, batch 1250, loss[loss=0.1836, simple_loss=0.24, pruned_loss=0.0636, over 13424.00 frames. ], tot_loss[loss=0.1831, simple_loss=0.2414, pruned_loss=0.0624, over 2652626.31 frames. ], batch size: 95, lr: 6.90e-03, grad_scale: 32.0 2023-04-17 00:31:47,368 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9502, 4.2351, 3.2178, 2.8873, 3.0042, 2.5245, 4.3667, 2.4099], device='cuda:0'), covar=tensor([0.1768, 0.0340, 0.1225, 0.2045, 0.0887, 0.3439, 0.0213, 0.4195], device='cuda:0'), in_proj_covar=tensor([0.0307, 0.0283, 0.0306, 0.0320, 0.0251, 0.0320, 0.0205, 0.0359], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:32:13,456 INFO [train.py:893] (0/4) Epoch 19, batch 1300, loss[loss=0.1645, simple_loss=0.2232, pruned_loss=0.05288, over 13333.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2426, pruned_loss=0.06289, over 2657034.31 frames. ], batch size: 67, lr: 6.90e-03, grad_scale: 32.0 2023-04-17 00:32:21,121 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49892.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:32:44,299 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.904e+02 2.679e+02 3.163e+02 3.782e+02 7.559e+02, threshold=6.326e+02, percent-clipped=2.0 2023-04-17 00:32:59,142 INFO [train.py:893] (0/4) Epoch 19, batch 1350, loss[loss=0.1989, simple_loss=0.2632, pruned_loss=0.06725, over 13467.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2432, pruned_loss=0.06329, over 2656804.60 frames. ], batch size: 100, lr: 6.90e-03, grad_scale: 32.0 2023-04-17 00:33:05,808 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49940.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:33:17,526 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2053, 2.6267, 2.1765, 4.1631, 4.6833, 3.4788, 4.5872, 4.3691], device='cuda:0'), covar=tensor([0.0105, 0.0837, 0.0984, 0.0096, 0.0056, 0.0452, 0.0068, 0.0072], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0088, 0.0096, 0.0079, 0.0064, 0.0079, 0.0054, 0.0075], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:33:36,426 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-17 00:33:38,549 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5638, 3.7046, 4.1423, 2.8648, 2.7500, 2.8954, 4.4761, 4.5710], device='cuda:0'), covar=tensor([0.1166, 0.1382, 0.0386, 0.1743, 0.1650, 0.1427, 0.0295, 0.0311], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0261, 0.0191, 0.0220, 0.0216, 0.0181, 0.0204, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:33:41,697 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0961, 3.9308, 3.1295, 3.8311, 3.1622, 2.2643, 3.9803, 2.1848], device='cuda:0'), covar=tensor([0.0629, 0.0569, 0.0496, 0.0243, 0.0746, 0.1940, 0.0840, 0.1391], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0137, 0.0133, 0.0113, 0.0147, 0.0187, 0.0168, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:33:47,104 INFO [train.py:893] (0/4) Epoch 19, batch 1400, loss[loss=0.2084, simple_loss=0.2594, pruned_loss=0.0787, over 13575.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2427, pruned_loss=0.06322, over 2661884.50 frames. ], batch size: 89, lr: 6.89e-03, grad_scale: 32.0 2023-04-17 00:34:01,453 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-50000.pt 2023-04-17 00:34:21,728 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.018e+02 2.575e+02 3.265e+02 4.093e+02 1.236e+03, threshold=6.530e+02, percent-clipped=2.0 2023-04-17 00:34:36,623 INFO [train.py:893] (0/4) Epoch 19, batch 1450, loss[loss=0.1776, simple_loss=0.2451, pruned_loss=0.0551, over 13438.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2428, pruned_loss=0.06383, over 2661975.60 frames. ], batch size: 103, lr: 6.89e-03, grad_scale: 32.0 2023-04-17 00:35:19,172 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50078.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:35:23,958 INFO [train.py:893] (0/4) Epoch 19, batch 1500, loss[loss=0.1858, simple_loss=0.2478, pruned_loss=0.06186, over 13425.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2428, pruned_loss=0.06347, over 2662924.78 frames. ], batch size: 103, lr: 6.88e-03, grad_scale: 32.0 2023-04-17 00:35:27,440 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50088.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 00:35:43,474 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1013, 4.3262, 3.3875, 2.9261, 3.1598, 2.5989, 4.5184, 2.5645], device='cuda:0'), covar=tensor([0.1710, 0.0308, 0.1089, 0.2188, 0.0770, 0.3336, 0.0205, 0.3827], device='cuda:0'), in_proj_covar=tensor([0.0309, 0.0285, 0.0307, 0.0323, 0.0252, 0.0321, 0.0207, 0.0360], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:35:48,306 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8117, 3.8906, 2.9469, 2.5868, 2.6910, 2.3707, 3.9943, 2.2054], device='cuda:0'), covar=tensor([0.1631, 0.0395, 0.1212, 0.2251, 0.0853, 0.3227, 0.0286, 0.4214], device='cuda:0'), in_proj_covar=tensor([0.0309, 0.0285, 0.0307, 0.0323, 0.0253, 0.0322, 0.0207, 0.0361], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:35:54,509 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.061e+02 2.681e+02 3.044e+02 3.631e+02 7.393e+02, threshold=6.088e+02, percent-clipped=1.0 2023-04-17 00:35:58,870 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50122.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:36:02,804 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50126.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:36:04,679 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50128.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:36:08,871 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1687, 4.4456, 4.1820, 4.2348, 4.3075, 4.6407, 4.4342, 4.2807], device='cuda:0'), covar=tensor([0.0370, 0.0335, 0.0401, 0.0938, 0.0347, 0.0260, 0.0370, 0.0354], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0149, 0.0170, 0.0262, 0.0171, 0.0187, 0.0168, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 00:36:11,251 INFO [train.py:893] (0/4) Epoch 19, batch 1550, loss[loss=0.1765, simple_loss=0.2367, pruned_loss=0.0582, over 13522.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.242, pruned_loss=0.06253, over 2663469.10 frames. ], batch size: 87, lr: 6.88e-03, grad_scale: 32.0 2023-04-17 00:36:13,794 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50136.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 00:36:55,678 INFO [train.py:893] (0/4) Epoch 19, batch 1600, loss[loss=0.1965, simple_loss=0.2467, pruned_loss=0.07319, over 13533.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2425, pruned_loss=0.0626, over 2662094.28 frames. ], batch size: 83, lr: 6.88e-03, grad_scale: 32.0 2023-04-17 00:37:00,805 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50189.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:37:09,044 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50197.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 00:37:26,429 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.888e+02 2.851e+02 3.229e+02 3.845e+02 6.840e+02, threshold=6.459e+02, percent-clipped=1.0 2023-04-17 00:37:43,059 INFO [train.py:893] (0/4) Epoch 19, batch 1650, loss[loss=0.1673, simple_loss=0.2311, pruned_loss=0.05175, over 13495.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.2437, pruned_loss=0.06284, over 2662423.66 frames. ], batch size: 93, lr: 6.87e-03, grad_scale: 32.0 2023-04-17 00:38:04,823 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50258.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 00:38:28,023 INFO [train.py:893] (0/4) Epoch 19, batch 1700, loss[loss=0.2077, simple_loss=0.2669, pruned_loss=0.07423, over 13454.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2443, pruned_loss=0.06276, over 2656520.55 frames. ], batch size: 106, lr: 6.87e-03, grad_scale: 32.0 2023-04-17 00:38:39,953 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9648, 1.9659, 2.3360, 3.2982, 3.0076, 3.3118, 2.5640, 2.0345], device='cuda:0'), covar=tensor([0.0283, 0.0967, 0.0725, 0.0077, 0.0293, 0.0079, 0.0610, 0.0972], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0152, 0.0167, 0.0093, 0.0118, 0.0093, 0.0169, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:38:57,982 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.960e+02 2.607e+02 3.098e+02 3.565e+02 8.055e+02, threshold=6.196e+02, percent-clipped=3.0 2023-04-17 00:39:14,610 INFO [train.py:893] (0/4) Epoch 19, batch 1750, loss[loss=0.1765, simple_loss=0.226, pruned_loss=0.0635, over 13381.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2439, pruned_loss=0.06269, over 2660336.98 frames. ], batch size: 62, lr: 6.87e-03, grad_scale: 32.0 2023-04-17 00:39:26,880 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-04-17 00:39:51,347 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6992, 3.3561, 4.2326, 3.0465, 2.7755, 2.8464, 4.4896, 4.5393], device='cuda:0'), covar=tensor([0.1081, 0.1709, 0.0334, 0.1615, 0.1609, 0.1624, 0.0215, 0.0246], device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0259, 0.0189, 0.0219, 0.0214, 0.0178, 0.0202, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:40:01,627 INFO [train.py:893] (0/4) Epoch 19, batch 1800, loss[loss=0.1617, simple_loss=0.2249, pruned_loss=0.04927, over 13346.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2429, pruned_loss=0.0618, over 2661886.63 frames. ], batch size: 73, lr: 6.86e-03, grad_scale: 32.0 2023-04-17 00:40:33,679 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.914e+02 2.700e+02 3.166e+02 3.667e+02 8.040e+02, threshold=6.333e+02, percent-clipped=2.0 2023-04-17 00:40:34,071 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2509, 2.0971, 2.5523, 3.6735, 3.3168, 3.6981, 2.8521, 2.1167], device='cuda:0'), covar=tensor([0.0280, 0.0939, 0.0760, 0.0061, 0.0254, 0.0064, 0.0654, 0.0993], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0151, 0.0165, 0.0092, 0.0116, 0.0093, 0.0167, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:40:38,123 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50422.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:40:44,863 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0545, 4.3665, 4.1035, 4.1472, 4.1965, 4.5136, 4.3311, 4.0761], device='cuda:0'), covar=tensor([0.0282, 0.0260, 0.0293, 0.0792, 0.0247, 0.0194, 0.0261, 0.0463], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0149, 0.0169, 0.0259, 0.0170, 0.0186, 0.0166, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 00:40:48,087 INFO [train.py:893] (0/4) Epoch 19, batch 1850, loss[loss=0.1715, simple_loss=0.2335, pruned_loss=0.05475, over 13468.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2421, pruned_loss=0.0616, over 2662349.40 frames. ], batch size: 79, lr: 6.86e-03, grad_scale: 32.0 2023-04-17 00:40:48,617 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-17 00:40:52,341 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-17 00:40:53,258 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3690, 4.9156, 4.7615, 4.8426, 4.6540, 4.7140, 5.3214, 4.9298], device='cuda:0'), covar=tensor([0.0704, 0.1054, 0.2096, 0.2582, 0.1007, 0.1439, 0.0894, 0.1071], device='cuda:0'), in_proj_covar=tensor([0.0314, 0.0363, 0.0451, 0.0463, 0.0279, 0.0337, 0.0418, 0.0334], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:41:22,271 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50470.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:41:27,740 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-04-17 00:41:36,350 INFO [train.py:893] (0/4) Epoch 19, batch 1900, loss[loss=0.1858, simple_loss=0.2575, pruned_loss=0.0571, over 13231.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2418, pruned_loss=0.06181, over 2654853.59 frames. ], batch size: 132, lr: 6.86e-03, grad_scale: 32.0 2023-04-17 00:41:36,517 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50484.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:42:03,660 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2808, 4.7914, 4.6351, 4.7560, 4.5714, 4.6390, 5.2290, 4.8327], device='cuda:0'), covar=tensor([0.0706, 0.1087, 0.2043, 0.2448, 0.0855, 0.1589, 0.0859, 0.1096], device='cuda:0'), in_proj_covar=tensor([0.0315, 0.0363, 0.0449, 0.0462, 0.0278, 0.0335, 0.0417, 0.0334], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:42:05,952 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.964e+02 2.672e+02 3.086e+02 3.593e+02 6.893e+02, threshold=6.171e+02, percent-clipped=1.0 2023-04-17 00:42:16,077 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.4497, 4.8734, 4.8160, 4.9632, 4.6840, 4.7463, 5.4084, 4.9511], device='cuda:0'), covar=tensor([0.0658, 0.1194, 0.2239, 0.2503, 0.0965, 0.1495, 0.0859, 0.1255], device='cuda:0'), in_proj_covar=tensor([0.0315, 0.0364, 0.0449, 0.0463, 0.0278, 0.0336, 0.0417, 0.0335], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:42:20,692 INFO [train.py:893] (0/4) Epoch 19, batch 1950, loss[loss=0.164, simple_loss=0.2247, pruned_loss=0.05171, over 13503.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2413, pruned_loss=0.06141, over 2658766.68 frames. ], batch size: 70, lr: 6.85e-03, grad_scale: 32.0 2023-04-17 00:42:32,862 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50546.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:42:34,822 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2023-04-17 00:42:38,663 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50553.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 00:43:04,015 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-17 00:43:07,592 INFO [train.py:893] (0/4) Epoch 19, batch 2000, loss[loss=0.1887, simple_loss=0.2375, pruned_loss=0.06998, over 13399.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2431, pruned_loss=0.06241, over 2660049.66 frames. ], batch size: 62, lr: 6.85e-03, grad_scale: 32.0 2023-04-17 00:43:12,683 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 00:43:30,439 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50607.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:43:38,612 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.951e+02 2.994e+02 3.373e+02 4.024e+02 7.060e+02, threshold=6.745e+02, percent-clipped=2.0 2023-04-17 00:43:56,143 INFO [train.py:893] (0/4) Epoch 19, batch 2050, loss[loss=0.2336, simple_loss=0.2828, pruned_loss=0.09224, over 13547.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2448, pruned_loss=0.06329, over 2659327.41 frames. ], batch size: 91, lr: 6.85e-03, grad_scale: 32.0 2023-04-17 00:44:41,411 INFO [train.py:893] (0/4) Epoch 19, batch 2100, loss[loss=0.174, simple_loss=0.2335, pruned_loss=0.05719, over 13495.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2428, pruned_loss=0.06211, over 2661796.17 frames. ], batch size: 81, lr: 6.84e-03, grad_scale: 32.0 2023-04-17 00:45:12,558 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.790e+02 2.504e+02 2.896e+02 3.497e+02 6.589e+02, threshold=5.792e+02, percent-clipped=0.0 2023-04-17 00:45:29,436 INFO [train.py:893] (0/4) Epoch 19, batch 2150, loss[loss=0.1939, simple_loss=0.2501, pruned_loss=0.06882, over 13374.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2428, pruned_loss=0.0618, over 2663465.30 frames. ], batch size: 73, lr: 6.84e-03, grad_scale: 32.0 2023-04-17 00:45:37,129 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4758, 2.5161, 2.9491, 3.9913, 3.5770, 4.0851, 3.2166, 2.3218], device='cuda:0'), covar=tensor([0.0283, 0.0906, 0.0645, 0.0063, 0.0230, 0.0044, 0.0607, 0.0936], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0152, 0.0166, 0.0094, 0.0117, 0.0094, 0.0168, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:45:54,524 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.96 vs. limit=5.0 2023-04-17 00:46:05,274 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7224, 3.7687, 2.6637, 3.4899, 3.7296, 2.3199, 3.2421, 2.6162], device='cuda:0'), covar=tensor([0.0305, 0.0309, 0.1159, 0.0401, 0.0290, 0.1324, 0.0619, 0.1317], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0170, 0.0175, 0.0201, 0.0135, 0.0159, 0.0158, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:46:13,447 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50782.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:46:14,794 INFO [train.py:893] (0/4) Epoch 19, batch 2200, loss[loss=0.1963, simple_loss=0.2485, pruned_loss=0.07202, over 13505.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2422, pruned_loss=0.06127, over 2662447.64 frames. ], batch size: 93, lr: 6.84e-03, grad_scale: 16.0 2023-04-17 00:46:15,008 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50784.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:46:37,813 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50808.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 00:46:46,611 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.993e+02 2.746e+02 3.205e+02 4.052e+02 6.409e+02, threshold=6.411e+02, percent-clipped=1.0 2023-04-17 00:47:00,752 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50832.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:47:02,336 INFO [train.py:893] (0/4) Epoch 19, batch 2250, loss[loss=0.1659, simple_loss=0.222, pruned_loss=0.05486, over 13516.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2424, pruned_loss=0.06152, over 2664834.33 frames. ], batch size: 76, lr: 6.83e-03, grad_scale: 16.0 2023-04-17 00:47:02,745 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9640, 4.1319, 3.1683, 2.8191, 2.9584, 2.5589, 4.2370, 2.3715], device='cuda:0'), covar=tensor([0.1638, 0.0358, 0.1143, 0.2141, 0.0821, 0.3083, 0.0260, 0.4149], device='cuda:0'), in_proj_covar=tensor([0.0309, 0.0285, 0.0309, 0.0324, 0.0252, 0.0321, 0.0209, 0.0360], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:47:10,810 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50843.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:47:19,640 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50853.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 00:47:33,939 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50869.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 00:47:47,566 INFO [train.py:893] (0/4) Epoch 19, batch 2300, loss[loss=0.1725, simple_loss=0.2233, pruned_loss=0.06088, over 12775.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2416, pruned_loss=0.06096, over 2662141.60 frames. ], batch size: 52, lr: 6.83e-03, grad_scale: 16.0 2023-04-17 00:48:03,645 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50901.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 00:48:04,479 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50902.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:48:20,403 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.979e+02 2.847e+02 3.302e+02 3.831e+02 7.686e+02, threshold=6.603e+02, percent-clipped=1.0 2023-04-17 00:48:33,729 INFO [train.py:893] (0/4) Epoch 19, batch 2350, loss[loss=0.1821, simple_loss=0.2444, pruned_loss=0.05995, over 13465.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2404, pruned_loss=0.06043, over 2663932.14 frames. ], batch size: 103, lr: 6.83e-03, grad_scale: 16.0 2023-04-17 00:48:44,772 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50946.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:48:48,975 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4378, 2.8145, 2.3866, 4.3243, 4.8937, 3.6066, 4.8678, 4.4661], device='cuda:0'), covar=tensor([0.0084, 0.0891, 0.0993, 0.0119, 0.0062, 0.0492, 0.0075, 0.0101], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0089, 0.0096, 0.0079, 0.0065, 0.0081, 0.0054, 0.0075], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:48:57,037 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 00:49:11,548 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.2145, 2.2202, 4.2325, 3.9838, 4.1532, 3.3532, 3.9233, 3.2453], device='cuda:0'), covar=tensor([0.2209, 0.1617, 0.0108, 0.0210, 0.0188, 0.0594, 0.0211, 0.0822], device='cuda:0'), in_proj_covar=tensor([0.0215, 0.0185, 0.0118, 0.0124, 0.0129, 0.0172, 0.0139, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 00:49:21,056 INFO [train.py:893] (0/4) Epoch 19, batch 2400, loss[loss=0.177, simple_loss=0.237, pruned_loss=0.05845, over 13540.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2395, pruned_loss=0.06032, over 2665799.96 frames. ], batch size: 78, lr: 6.82e-03, grad_scale: 16.0 2023-04-17 00:49:41,816 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51007.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:49:53,371 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.675e+02 2.579e+02 3.050e+02 3.500e+02 7.370e+02, threshold=6.100e+02, percent-clipped=1.0 2023-04-17 00:50:07,320 INFO [train.py:893] (0/4) Epoch 19, batch 2450, loss[loss=0.1718, simple_loss=0.2356, pruned_loss=0.05402, over 13336.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.239, pruned_loss=0.06025, over 2663320.77 frames. ], batch size: 67, lr: 6.82e-03, grad_scale: 16.0 2023-04-17 00:50:33,200 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51062.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:50:54,312 INFO [train.py:893] (0/4) Epoch 19, batch 2500, loss[loss=0.1883, simple_loss=0.2443, pruned_loss=0.06614, over 13531.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.239, pruned_loss=0.06034, over 2663151.78 frames. ], batch size: 91, lr: 6.82e-03, grad_scale: 16.0 2023-04-17 00:51:19,423 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6023, 3.3217, 4.1513, 2.9980, 2.7255, 2.9106, 4.4167, 4.5200], device='cuda:0'), covar=tensor([0.1221, 0.1655, 0.0314, 0.1627, 0.1504, 0.1355, 0.0215, 0.0215], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0266, 0.0191, 0.0222, 0.0218, 0.0180, 0.0205, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:51:25,492 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.099e+02 2.551e+02 2.924e+02 3.342e+02 5.751e+02, threshold=5.847e+02, percent-clipped=0.0 2023-04-17 00:51:30,032 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51123.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:51:39,524 INFO [train.py:893] (0/4) Epoch 19, batch 2550, loss[loss=0.1969, simple_loss=0.2589, pruned_loss=0.06744, over 13520.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.239, pruned_loss=0.06036, over 2662391.70 frames. ], batch size: 91, lr: 6.81e-03, grad_scale: 16.0 2023-04-17 00:51:43,814 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51138.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:51:55,644 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2615, 4.7664, 4.6303, 4.7484, 4.6522, 4.6549, 5.2582, 4.8327], device='cuda:0'), covar=tensor([0.0730, 0.1217, 0.2327, 0.2722, 0.0909, 0.1630, 0.0904, 0.1159], device='cuda:0'), in_proj_covar=tensor([0.0322, 0.0367, 0.0458, 0.0469, 0.0285, 0.0343, 0.0424, 0.0338], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:52:03,726 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 00:52:08,007 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51164.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 00:52:17,286 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51172.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 00:52:27,012 INFO [train.py:893] (0/4) Epoch 19, batch 2600, loss[loss=0.2003, simple_loss=0.2579, pruned_loss=0.07134, over 13537.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.2398, pruned_loss=0.06086, over 2656019.83 frames. ], batch size: 91, lr: 6.81e-03, grad_scale: 16.0 2023-04-17 00:52:45,121 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51202.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:52:57,646 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.905e+02 2.632e+02 3.230e+02 4.137e+02 6.424e+02, threshold=6.459e+02, percent-clipped=3.0 2023-04-17 00:53:09,503 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51233.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 00:53:09,975 INFO [train.py:893] (0/4) Epoch 19, batch 2650, loss[loss=0.1879, simple_loss=0.2455, pruned_loss=0.06516, over 13544.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2394, pruned_loss=0.06103, over 2651544.42 frames. ], batch size: 87, lr: 6.81e-03, grad_scale: 16.0 2023-04-17 00:53:10,278 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4660, 3.2581, 3.9734, 2.9078, 2.7574, 2.7297, 4.2954, 4.3850], device='cuda:0'), covar=tensor([0.1168, 0.1863, 0.0405, 0.1699, 0.1500, 0.1513, 0.0272, 0.0239], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0265, 0.0191, 0.0223, 0.0217, 0.0181, 0.0205, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:53:21,990 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51250.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:53:27,382 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6911, 2.3394, 2.2446, 2.8365, 2.1192, 2.7784, 2.5247, 2.1730], device='cuda:0'), covar=tensor([0.0148, 0.0243, 0.0283, 0.0240, 0.0347, 0.0175, 0.0337, 0.0258], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0109, 0.0120, 0.0115, 0.0127, 0.0105, 0.0104, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 00:53:48,965 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-19.pt 2023-04-17 00:54:14,862 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-17 00:54:18,258 INFO [train.py:893] (0/4) Epoch 20, batch 0, loss[loss=0.1824, simple_loss=0.238, pruned_loss=0.06338, over 13516.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.238, pruned_loss=0.06338, over 13516.00 frames. ], batch size: 76, lr: 6.63e-03, grad_scale: 16.0 2023-04-17 00:54:18,259 INFO [train.py:918] (0/4) Computing validation loss 2023-04-17 00:54:26,035 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.0264, 4.5178, 4.5673, 4.6317, 4.4989, 4.5619, 5.0046, 4.6093], device='cuda:0'), covar=tensor([0.0725, 0.1284, 0.2010, 0.2227, 0.0926, 0.1443, 0.0977, 0.1128], device='cuda:0'), in_proj_covar=tensor([0.0317, 0.0360, 0.0449, 0.0460, 0.0280, 0.0337, 0.0417, 0.0329], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 00:54:28,966 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4568, 2.6866, 2.9990, 4.0386, 3.6606, 4.0911, 3.3232, 2.5765], device='cuda:0'), covar=tensor([0.0322, 0.0777, 0.0685, 0.0063, 0.0221, 0.0051, 0.0573, 0.0855], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0151, 0.0167, 0.0095, 0.0117, 0.0094, 0.0167, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 00:54:33,050 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9932, 2.5751, 2.5840, 3.0221, 2.4563, 3.0452, 3.1147, 2.5463], device='cuda:0'), covar=tensor([0.0055, 0.0192, 0.0148, 0.0153, 0.0187, 0.0131, 0.0135, 0.0149], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0109, 0.0119, 0.0115, 0.0126, 0.0104, 0.0103, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 00:54:41,316 INFO [train.py:927] (0/4) Epoch 20, validation: loss=0.1383, simple_loss=0.1976, pruned_loss=0.03948, over 2446609.00 frames. 2023-04-17 00:54:41,317 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-17 00:54:54,971 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7628, 3.9001, 3.0833, 2.6343, 2.7587, 2.3896, 3.9231, 2.2875], device='cuda:0'), covar=tensor([0.1764, 0.0359, 0.1110, 0.2241, 0.0867, 0.3261, 0.0306, 0.3760], device='cuda:0'), in_proj_covar=tensor([0.0308, 0.0283, 0.0309, 0.0326, 0.0253, 0.0322, 0.0207, 0.0359], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 00:54:58,940 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51302.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:55:13,677 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 2.582e+02 3.103e+02 3.598e+02 6.188e+02, threshold=6.206e+02, percent-clipped=0.0 2023-04-17 00:55:27,791 INFO [train.py:893] (0/4) Epoch 20, batch 50, loss[loss=0.1512, simple_loss=0.2204, pruned_loss=0.04104, over 13454.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2364, pruned_loss=0.06025, over 601746.82 frames. ], batch size: 79, lr: 6.63e-03, grad_scale: 16.0 2023-04-17 00:55:52,749 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-17 00:55:52,750 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-17 00:55:52,750 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-17 00:55:52,757 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-17 00:55:52,774 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-17 00:55:52,787 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-17 00:55:52,805 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-17 00:56:13,447 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51382.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:56:14,090 INFO [train.py:893] (0/4) Epoch 20, batch 100, loss[loss=0.2115, simple_loss=0.2625, pruned_loss=0.08024, over 13235.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2388, pruned_loss=0.06267, over 1058952.79 frames. ], batch size: 124, lr: 6.62e-03, grad_scale: 16.0 2023-04-17 00:56:23,396 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51392.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:56:47,317 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 2.853e+02 3.337e+02 4.244e+02 6.093e+02, threshold=6.673e+02, percent-clipped=0.0 2023-04-17 00:56:47,534 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51418.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:57:00,400 INFO [train.py:893] (0/4) Epoch 20, batch 150, loss[loss=0.1889, simple_loss=0.253, pruned_loss=0.06245, over 13493.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2423, pruned_loss=0.06431, over 1416735.45 frames. ], batch size: 93, lr: 6.62e-03, grad_scale: 16.0 2023-04-17 00:57:05,518 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51438.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:57:10,533 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51443.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:57:20,424 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51453.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:57:29,309 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51464.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 00:57:38,562 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-17 00:57:47,195 INFO [train.py:893] (0/4) Epoch 20, batch 200, loss[loss=0.2103, simple_loss=0.264, pruned_loss=0.07836, over 13509.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2435, pruned_loss=0.06478, over 1690147.57 frames. ], batch size: 93, lr: 6.62e-03, grad_scale: 16.0 2023-04-17 00:57:49,741 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51486.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:57:58,345 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51495.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:58:13,999 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51512.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 00:58:20,425 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.908e+02 2.666e+02 3.383e+02 3.892e+02 7.780e+02, threshold=6.766e+02, percent-clipped=1.0 2023-04-17 00:58:28,939 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51528.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 00:58:32,955 INFO [train.py:893] (0/4) Epoch 20, batch 250, loss[loss=0.1795, simple_loss=0.2445, pruned_loss=0.05718, over 13562.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2425, pruned_loss=0.06438, over 1903826.94 frames. ], batch size: 89, lr: 6.61e-03, grad_scale: 16.0 2023-04-17 00:58:55,560 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51556.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:59:20,632 INFO [train.py:893] (0/4) Epoch 20, batch 300, loss[loss=0.1584, simple_loss=0.2236, pruned_loss=0.04663, over 13352.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2429, pruned_loss=0.06406, over 2074589.16 frames. ], batch size: 73, lr: 6.61e-03, grad_scale: 16.0 2023-04-17 00:59:36,240 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.5914, 2.4587, 2.1038, 1.5489, 1.5942, 2.0416, 2.2048, 2.6503], device='cuda:0'), covar=tensor([0.0860, 0.0268, 0.0641, 0.1437, 0.0154, 0.0488, 0.0720, 0.0209], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0135, 0.0114, 0.0204, 0.0107, 0.0154, 0.0165, 0.0125], device='cuda:0'), out_proj_covar=tensor([1.1778e-04, 1.0126e-04, 8.9878e-05, 1.5268e-04, 7.8752e-05, 1.1657e-04, 1.2460e-04, 9.2605e-05], device='cuda:0') 2023-04-17 00:59:37,798 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51602.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 00:59:53,136 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.976e+02 2.888e+02 3.248e+02 3.960e+02 8.740e+02, threshold=6.497e+02, percent-clipped=3.0 2023-04-17 00:59:54,225 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0763, 4.3733, 4.1361, 4.1696, 4.1743, 4.5107, 4.3201, 4.1222], device='cuda:0'), covar=tensor([0.0314, 0.0275, 0.0358, 0.0812, 0.0340, 0.0228, 0.0360, 0.0383], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0150, 0.0170, 0.0260, 0.0170, 0.0187, 0.0167, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 01:00:06,405 INFO [train.py:893] (0/4) Epoch 20, batch 350, loss[loss=0.2088, simple_loss=0.2695, pruned_loss=0.074, over 13473.00 frames. ], tot_loss[loss=0.187, simple_loss=0.2442, pruned_loss=0.0649, over 2205611.36 frames. ], batch size: 103, lr: 6.61e-03, grad_scale: 16.0 2023-04-17 01:00:23,157 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51650.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:00:30,916 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8867, 2.8657, 2.3860, 1.9490, 1.9641, 2.5098, 2.5918, 3.1103], device='cuda:0'), covar=tensor([0.1092, 0.0394, 0.0821, 0.1629, 0.0428, 0.0535, 0.0792, 0.0231], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0135, 0.0115, 0.0203, 0.0107, 0.0155, 0.0165, 0.0125], device='cuda:0'), out_proj_covar=tensor([1.1854e-04, 1.0117e-04, 9.0060e-05, 1.5219e-04, 7.8442e-05, 1.1660e-04, 1.2488e-04, 9.2424e-05], device='cuda:0') 2023-04-17 01:00:54,203 INFO [train.py:893] (0/4) Epoch 20, batch 400, loss[loss=0.1607, simple_loss=0.2261, pruned_loss=0.04767, over 13449.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2434, pruned_loss=0.06395, over 2310842.63 frames. ], batch size: 79, lr: 6.60e-03, grad_scale: 16.0 2023-04-17 01:01:26,015 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.927e+02 2.630e+02 3.030e+02 3.596e+02 5.249e+02, threshold=6.061e+02, percent-clipped=0.0 2023-04-17 01:01:26,261 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51718.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:01:27,468 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-17 01:01:31,956 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9036, 2.4684, 2.5119, 2.8515, 2.3926, 2.8790, 2.8902, 2.4111], device='cuda:0'), covar=tensor([0.0074, 0.0172, 0.0129, 0.0151, 0.0175, 0.0108, 0.0149, 0.0153], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0109, 0.0118, 0.0116, 0.0126, 0.0105, 0.0104, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 01:01:39,782 INFO [train.py:893] (0/4) Epoch 20, batch 450, loss[loss=0.1878, simple_loss=0.2461, pruned_loss=0.06477, over 13531.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2448, pruned_loss=0.0648, over 2384758.91 frames. ], batch size: 98, lr: 6.60e-03, grad_scale: 16.0 2023-04-17 01:01:43,963 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51738.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:01:52,731 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51747.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:01:53,458 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51748.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:02:05,663 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-17 01:02:09,906 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51766.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:02:25,778 INFO [train.py:893] (0/4) Epoch 20, batch 500, loss[loss=0.1606, simple_loss=0.2215, pruned_loss=0.0498, over 13237.00 frames. ], tot_loss[loss=0.1859, simple_loss=0.2442, pruned_loss=0.06384, over 2449895.99 frames. ], batch size: 117, lr: 6.60e-03, grad_scale: 16.0 2023-04-17 01:02:49,849 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51808.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:02:58,570 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.164e+02 2.958e+02 3.503e+02 4.173e+02 8.375e+02, threshold=7.006e+02, percent-clipped=2.0 2023-04-17 01:03:08,038 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51828.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 01:03:10,456 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1149, 4.0299, 3.2499, 3.8999, 3.2808, 2.2830, 3.9999, 2.1095], device='cuda:0'), covar=tensor([0.0616, 0.0397, 0.0516, 0.0194, 0.0616, 0.1789, 0.0714, 0.1237], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0136, 0.0133, 0.0113, 0.0146, 0.0187, 0.0170, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 01:03:12,423 INFO [train.py:893] (0/4) Epoch 20, batch 550, loss[loss=0.1926, simple_loss=0.2513, pruned_loss=0.06695, over 13277.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2439, pruned_loss=0.06331, over 2501155.11 frames. ], batch size: 124, lr: 6.59e-03, grad_scale: 16.0 2023-04-17 01:03:29,448 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51851.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:03:32,506 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-17 01:03:52,755 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51876.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 01:03:59,078 INFO [train.py:893] (0/4) Epoch 20, batch 600, loss[loss=0.1706, simple_loss=0.2337, pruned_loss=0.05373, over 13427.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2426, pruned_loss=0.06262, over 2534680.13 frames. ], batch size: 88, lr: 6.59e-03, grad_scale: 16.0 2023-04-17 01:04:31,320 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.718e+02 2.517e+02 2.993e+02 3.459e+02 6.026e+02, threshold=5.986e+02, percent-clipped=0.0 2023-04-17 01:04:46,680 INFO [train.py:893] (0/4) Epoch 20, batch 650, loss[loss=0.1953, simple_loss=0.254, pruned_loss=0.06828, over 13484.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2421, pruned_loss=0.06243, over 2563274.21 frames. ], batch size: 93, lr: 6.59e-03, grad_scale: 16.0 2023-04-17 01:05:15,856 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5850, 2.5163, 2.8067, 4.1243, 3.6171, 4.1747, 3.2399, 2.4815], device='cuda:0'), covar=tensor([0.0273, 0.0830, 0.0773, 0.0044, 0.0240, 0.0041, 0.0598, 0.0845], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0149, 0.0165, 0.0094, 0.0117, 0.0093, 0.0166, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:05:23,623 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2301, 3.0765, 3.7183, 2.6908, 2.4832, 2.5495, 4.0227, 4.1379], device='cuda:0'), covar=tensor([0.1304, 0.1762, 0.0392, 0.1683, 0.1612, 0.1552, 0.0268, 0.0191], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0263, 0.0191, 0.0223, 0.0217, 0.0179, 0.0204, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 01:05:32,329 INFO [train.py:893] (0/4) Epoch 20, batch 700, loss[loss=0.1835, simple_loss=0.238, pruned_loss=0.06451, over 13522.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2415, pruned_loss=0.06193, over 2587222.87 frames. ], batch size: 70, lr: 6.59e-03, grad_scale: 16.0 2023-04-17 01:05:49,294 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-52000.pt 2023-04-17 01:06:10,156 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.613e+02 2.564e+02 3.049e+02 3.631e+02 6.863e+02, threshold=6.097e+02, percent-clipped=2.0 2023-04-17 01:06:24,287 INFO [train.py:893] (0/4) Epoch 20, batch 750, loss[loss=0.1716, simple_loss=0.2128, pruned_loss=0.0652, over 12356.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2411, pruned_loss=0.06213, over 2603994.53 frames. ], batch size: 50, lr: 6.58e-03, grad_scale: 16.0 2023-04-17 01:06:28,652 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52038.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:06:30,406 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52040.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:06:38,305 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52048.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:07:10,865 INFO [train.py:893] (0/4) Epoch 20, batch 800, loss[loss=0.1932, simple_loss=0.2436, pruned_loss=0.07137, over 11683.00 frames. ], tot_loss[loss=0.1831, simple_loss=0.2418, pruned_loss=0.06221, over 2615423.26 frames. ], batch size: 157, lr: 6.58e-03, grad_scale: 16.0 2023-04-17 01:07:13,477 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52086.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:07:22,595 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52096.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:07:26,991 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52101.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:07:29,251 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52103.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:07:42,865 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.162e+02 2.907e+02 3.374e+02 4.122e+02 8.456e+02, threshold=6.748e+02, percent-clipped=6.0 2023-04-17 01:07:57,072 INFO [train.py:893] (0/4) Epoch 20, batch 850, loss[loss=0.1664, simple_loss=0.2277, pruned_loss=0.05256, over 13518.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.242, pruned_loss=0.06254, over 2623709.38 frames. ], batch size: 85, lr: 6.58e-03, grad_scale: 16.0 2023-04-17 01:08:14,730 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52151.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:08:43,836 INFO [train.py:893] (0/4) Epoch 20, batch 900, loss[loss=0.1901, simple_loss=0.2516, pruned_loss=0.06433, over 13221.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2417, pruned_loss=0.06281, over 2631931.10 frames. ], batch size: 132, lr: 6.57e-03, grad_scale: 16.0 2023-04-17 01:08:52,390 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52193.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:08:57,983 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52199.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:09:00,507 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-17 01:09:15,363 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-17 01:09:15,526 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-17 01:09:16,317 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.090e+02 2.805e+02 3.291e+02 3.728e+02 9.830e+02, threshold=6.583e+02, percent-clipped=1.0 2023-04-17 01:09:29,520 INFO [train.py:893] (0/4) Epoch 20, batch 950, loss[loss=0.1727, simple_loss=0.2263, pruned_loss=0.05952, over 13496.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2411, pruned_loss=0.06297, over 2638741.39 frames. ], batch size: 81, lr: 6.57e-03, grad_scale: 16.0 2023-04-17 01:09:36,218 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8207, 4.6766, 4.8852, 4.7865, 5.1167, 4.7234, 5.0628, 5.0588], device='cuda:0'), covar=tensor([0.0419, 0.0544, 0.0619, 0.0511, 0.0517, 0.0774, 0.0507, 0.0544], device='cuda:0'), in_proj_covar=tensor([0.0279, 0.0288, 0.0295, 0.0219, 0.0422, 0.0335, 0.0270, 0.0294], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:09:46,450 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0225, 4.1685, 3.3611, 2.9067, 3.0855, 2.5804, 4.4387, 2.4380], device='cuda:0'), covar=tensor([0.1726, 0.0339, 0.1081, 0.2126, 0.0823, 0.3338, 0.0226, 0.4281], device='cuda:0'), in_proj_covar=tensor([0.0310, 0.0285, 0.0311, 0.0325, 0.0256, 0.0323, 0.0207, 0.0361], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 01:09:47,227 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.4924, 2.0805, 2.0979, 2.5308, 1.8310, 2.4933, 2.2824, 2.0282], device='cuda:0'), covar=tensor([0.0132, 0.0276, 0.0219, 0.0157, 0.0313, 0.0162, 0.0272, 0.0263], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0109, 0.0119, 0.0115, 0.0126, 0.0104, 0.0104, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 01:09:49,630 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52254.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:10:16,226 INFO [train.py:893] (0/4) Epoch 20, batch 1000, loss[loss=0.1802, simple_loss=0.2443, pruned_loss=0.058, over 13534.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.2398, pruned_loss=0.06266, over 2642763.41 frames. ], batch size: 76, lr: 6.57e-03, grad_scale: 16.0 2023-04-17 01:10:43,658 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52312.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:10:48,273 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.733e+02 2.626e+02 3.000e+02 3.496e+02 6.136e+02, threshold=6.000e+02, percent-clipped=0.0 2023-04-17 01:11:02,262 INFO [train.py:893] (0/4) Epoch 20, batch 1050, loss[loss=0.1824, simple_loss=0.248, pruned_loss=0.05838, over 13460.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2392, pruned_loss=0.06169, over 2648578.32 frames. ], batch size: 106, lr: 6.56e-03, grad_scale: 16.0 2023-04-17 01:11:10,891 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8118, 4.0548, 2.8007, 3.8730, 3.9302, 2.6386, 3.3056, 3.0098], device='cuda:0'), covar=tensor([0.0349, 0.0404, 0.0990, 0.0310, 0.0303, 0.1031, 0.0672, 0.0991], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0172, 0.0177, 0.0204, 0.0135, 0.0158, 0.0159, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:11:40,607 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52373.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:11:48,669 INFO [train.py:893] (0/4) Epoch 20, batch 1100, loss[loss=0.1754, simple_loss=0.2352, pruned_loss=0.05776, over 13188.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2389, pruned_loss=0.06108, over 2649732.35 frames. ], batch size: 132, lr: 6.56e-03, grad_scale: 16.0 2023-04-17 01:12:00,958 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52396.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:12:02,628 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3278, 2.5225, 2.0039, 4.1329, 4.6148, 3.4290, 4.5830, 4.3241], device='cuda:0'), covar=tensor([0.0084, 0.0921, 0.1078, 0.0087, 0.0062, 0.0448, 0.0062, 0.0071], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0090, 0.0097, 0.0081, 0.0067, 0.0081, 0.0055, 0.0077], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 01:12:08,614 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52403.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:12:17,993 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.83 vs. limit=5.0 2023-04-17 01:12:21,582 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 2.765e+02 3.180e+02 3.679e+02 6.525e+02, threshold=6.360e+02, percent-clipped=1.0 2023-04-17 01:12:36,197 INFO [train.py:893] (0/4) Epoch 20, batch 1150, loss[loss=0.1877, simple_loss=0.2499, pruned_loss=0.06275, over 13197.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2391, pruned_loss=0.06069, over 2652985.26 frames. ], batch size: 132, lr: 6.56e-03, grad_scale: 16.0 2023-04-17 01:12:39,893 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9171, 2.8192, 2.6165, 2.0319, 2.0171, 2.4649, 2.5593, 3.1211], device='cuda:0'), covar=tensor([0.0979, 0.0371, 0.0616, 0.1385, 0.0483, 0.0539, 0.0772, 0.0229], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0136, 0.0116, 0.0204, 0.0109, 0.0156, 0.0168, 0.0126], device='cuda:0'), out_proj_covar=tensor([1.1918e-04, 1.0217e-04, 9.0868e-05, 1.5339e-04, 7.9868e-05, 1.1819e-04, 1.2686e-04, 9.3679e-05], device='cuda:0') 2023-04-17 01:12:46,677 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5913, 4.8276, 4.6026, 4.5295, 4.6168, 5.0019, 4.7911, 4.6032], device='cuda:0'), covar=tensor([0.0209, 0.0247, 0.0275, 0.0883, 0.0256, 0.0192, 0.0284, 0.0208], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0150, 0.0170, 0.0260, 0.0170, 0.0186, 0.0167, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 01:12:53,216 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52451.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:13:22,949 INFO [train.py:893] (0/4) Epoch 20, batch 1200, loss[loss=0.1609, simple_loss=0.2274, pruned_loss=0.04719, over 13525.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.239, pruned_loss=0.06035, over 2648713.10 frames. ], batch size: 76, lr: 6.55e-03, grad_scale: 16.0 2023-04-17 01:13:49,587 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-17 01:13:56,084 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.988e+02 2.539e+02 3.132e+02 3.497e+02 5.997e+02, threshold=6.265e+02, percent-clipped=0.0 2023-04-17 01:14:02,634 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-17 01:14:10,243 INFO [train.py:893] (0/4) Epoch 20, batch 1250, loss[loss=0.1623, simple_loss=0.2295, pruned_loss=0.04758, over 13375.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2402, pruned_loss=0.06091, over 2650810.34 frames. ], batch size: 113, lr: 6.55e-03, grad_scale: 16.0 2023-04-17 01:14:25,427 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52549.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:14:31,050 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52555.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 01:14:55,429 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-04-17 01:14:56,400 INFO [train.py:893] (0/4) Epoch 20, batch 1300, loss[loss=0.17, simple_loss=0.2316, pruned_loss=0.05419, over 13535.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2414, pruned_loss=0.06168, over 2649850.02 frames. ], batch size: 87, lr: 6.55e-03, grad_scale: 16.0 2023-04-17 01:15:26,888 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9430, 4.3226, 4.1900, 4.1033, 4.2077, 4.0035, 4.3722, 4.4222], device='cuda:0'), covar=tensor([0.0216, 0.0233, 0.0197, 0.0319, 0.0236, 0.0288, 0.0278, 0.0258], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0193, 0.0156, 0.0175, 0.0141, 0.0190, 0.0130, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 01:15:28,442 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52616.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 01:15:29,625 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.989e+02 2.703e+02 3.302e+02 3.792e+02 7.848e+02, threshold=6.603e+02, percent-clipped=1.0 2023-04-17 01:15:43,076 INFO [train.py:893] (0/4) Epoch 20, batch 1350, loss[loss=0.1614, simple_loss=0.2292, pruned_loss=0.0468, over 13349.00 frames. ], tot_loss[loss=0.1831, simple_loss=0.2423, pruned_loss=0.06198, over 2654895.75 frames. ], batch size: 109, lr: 6.54e-03, grad_scale: 16.0 2023-04-17 01:16:15,629 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52668.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:16:30,177 INFO [train.py:893] (0/4) Epoch 20, batch 1400, loss[loss=0.1954, simple_loss=0.24, pruned_loss=0.07544, over 13410.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.241, pruned_loss=0.06159, over 2653522.53 frames. ], batch size: 65, lr: 6.54e-03, grad_scale: 16.0 2023-04-17 01:16:41,866 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52696.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:16:48,683 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-17 01:17:02,873 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.002e+02 2.642e+02 3.013e+02 3.621e+02 7.035e+02, threshold=6.026e+02, percent-clipped=2.0 2023-04-17 01:17:08,251 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.2036, 1.9353, 3.8017, 3.6080, 3.6894, 2.8294, 3.3878, 2.8405], device='cuda:0'), covar=tensor([0.1986, 0.1575, 0.0137, 0.0197, 0.0204, 0.0703, 0.0287, 0.0849], device='cuda:0'), in_proj_covar=tensor([0.0214, 0.0186, 0.0119, 0.0125, 0.0128, 0.0173, 0.0140, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 01:17:16,197 INFO [train.py:893] (0/4) Epoch 20, batch 1450, loss[loss=0.1935, simple_loss=0.2516, pruned_loss=0.06773, over 13525.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2406, pruned_loss=0.06173, over 2657024.60 frames. ], batch size: 76, lr: 6.54e-03, grad_scale: 16.0 2023-04-17 01:17:26,927 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52744.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:17:33,680 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8246, 2.9056, 3.2307, 4.3388, 3.9068, 4.4106, 3.6595, 2.7159], device='cuda:0'), covar=tensor([0.0238, 0.0778, 0.0609, 0.0047, 0.0221, 0.0036, 0.0501, 0.0849], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0142, 0.0160, 0.0092, 0.0114, 0.0090, 0.0160, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:17:34,580 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52752.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:17:43,374 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-17 01:17:44,687 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3256, 4.6107, 4.3321, 4.4144, 4.4196, 4.7744, 4.5669, 4.4220], device='cuda:0'), covar=tensor([0.0321, 0.0336, 0.0367, 0.0888, 0.0334, 0.0268, 0.0333, 0.0256], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0152, 0.0171, 0.0262, 0.0171, 0.0189, 0.0168, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 01:18:03,215 INFO [train.py:893] (0/4) Epoch 20, batch 1500, loss[loss=0.1908, simple_loss=0.2466, pruned_loss=0.06747, over 13045.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2401, pruned_loss=0.06115, over 2650158.40 frames. ], batch size: 142, lr: 6.54e-03, grad_scale: 32.0 2023-04-17 01:18:04,518 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5723, 3.3384, 4.0587, 2.8238, 2.6680, 2.7803, 4.3437, 4.4355], device='cuda:0'), covar=tensor([0.1184, 0.1775, 0.0358, 0.1696, 0.1552, 0.1427, 0.0286, 0.0219], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0261, 0.0190, 0.0221, 0.0217, 0.0180, 0.0205, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 01:18:31,449 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52813.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:18:36,243 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.621e+02 2.570e+02 2.979e+02 3.357e+02 4.933e+02, threshold=5.957e+02, percent-clipped=0.0 2023-04-17 01:18:45,044 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.87 vs. limit=5.0 2023-04-17 01:18:50,228 INFO [train.py:893] (0/4) Epoch 20, batch 1550, loss[loss=0.1802, simple_loss=0.2437, pruned_loss=0.05839, over 13497.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.239, pruned_loss=0.06026, over 2655272.10 frames. ], batch size: 81, lr: 6.53e-03, grad_scale: 32.0 2023-04-17 01:19:05,573 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52849.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:19:37,149 INFO [train.py:893] (0/4) Epoch 20, batch 1600, loss[loss=0.1718, simple_loss=0.2362, pruned_loss=0.05376, over 13480.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2396, pruned_loss=0.06041, over 2658084.69 frames. ], batch size: 79, lr: 6.53e-03, grad_scale: 32.0 2023-04-17 01:19:50,407 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52897.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:20:03,643 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52911.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 01:20:06,237 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4759, 2.2403, 2.3743, 3.7809, 3.4968, 3.8419, 2.8611, 2.1741], device='cuda:0'), covar=tensor([0.0236, 0.0962, 0.0992, 0.0063, 0.0205, 0.0055, 0.0748, 0.1080], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0146, 0.0164, 0.0094, 0.0116, 0.0091, 0.0163, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:20:09,268 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.827e+02 2.611e+02 3.148e+02 3.639e+02 5.620e+02, threshold=6.296e+02, percent-clipped=0.0 2023-04-17 01:20:23,719 INFO [train.py:893] (0/4) Epoch 20, batch 1650, loss[loss=0.1952, simple_loss=0.2471, pruned_loss=0.07165, over 13530.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.2408, pruned_loss=0.06034, over 2662587.69 frames. ], batch size: 83, lr: 6.53e-03, grad_scale: 32.0 2023-04-17 01:20:43,682 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1543, 2.8367, 2.7134, 3.1951, 2.3757, 3.2056, 3.1295, 2.6536], device='cuda:0'), covar=tensor([0.0069, 0.0148, 0.0122, 0.0128, 0.0194, 0.0095, 0.0130, 0.0129], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0112, 0.0119, 0.0118, 0.0127, 0.0105, 0.0106, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 01:20:55,754 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52968.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:21:09,741 INFO [train.py:893] (0/4) Epoch 20, batch 1700, loss[loss=0.1485, simple_loss=0.2102, pruned_loss=0.04344, over 13523.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2407, pruned_loss=0.05995, over 2665932.92 frames. ], batch size: 70, lr: 6.52e-03, grad_scale: 32.0 2023-04-17 01:21:40,924 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53016.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:21:43,048 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.046e+02 2.784e+02 3.291e+02 3.982e+02 7.323e+02, threshold=6.582e+02, percent-clipped=2.0 2023-04-17 01:21:55,744 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5033, 4.6790, 4.5090, 4.5228, 4.6345, 4.9143, 4.7158, 4.5627], device='cuda:0'), covar=tensor([0.0242, 0.0274, 0.0283, 0.0842, 0.0234, 0.0207, 0.0239, 0.0206], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0153, 0.0171, 0.0263, 0.0172, 0.0189, 0.0169, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 01:21:56,319 INFO [train.py:893] (0/4) Epoch 20, batch 1750, loss[loss=0.1554, simple_loss=0.212, pruned_loss=0.04939, over 13350.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2394, pruned_loss=0.05925, over 2662691.09 frames. ], batch size: 62, lr: 6.52e-03, grad_scale: 32.0 2023-04-17 01:22:01,817 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8979, 4.1187, 3.1533, 2.7822, 2.8479, 2.4838, 4.2789, 2.3790], device='cuda:0'), covar=tensor([0.1656, 0.0312, 0.1158, 0.2134, 0.0856, 0.3249, 0.0231, 0.3947], device='cuda:0'), in_proj_covar=tensor([0.0309, 0.0284, 0.0310, 0.0326, 0.0255, 0.0322, 0.0206, 0.0361], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 01:22:18,595 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53056.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:22:43,005 INFO [train.py:893] (0/4) Epoch 20, batch 1800, loss[loss=0.2143, simple_loss=0.2544, pruned_loss=0.08708, over 11804.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2394, pruned_loss=0.05968, over 2658263.42 frames. ], batch size: 157, lr: 6.52e-03, grad_scale: 32.0 2023-04-17 01:23:06,739 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53108.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:23:15,296 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53117.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:23:15,879 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.144e+02 2.701e+02 3.241e+02 3.784e+02 6.824e+02, threshold=6.483e+02, percent-clipped=1.0 2023-04-17 01:23:30,155 INFO [train.py:893] (0/4) Epoch 20, batch 1850, loss[loss=0.1721, simple_loss=0.2286, pruned_loss=0.05781, over 13346.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2383, pruned_loss=0.05913, over 2658543.55 frames. ], batch size: 73, lr: 6.51e-03, grad_scale: 32.0 2023-04-17 01:23:33,581 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-17 01:23:56,632 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6595, 2.4065, 2.9440, 4.1466, 3.6949, 4.2170, 3.2386, 2.4564], device='cuda:0'), covar=tensor([0.0243, 0.0943, 0.0731, 0.0047, 0.0219, 0.0045, 0.0624, 0.0893], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0146, 0.0165, 0.0095, 0.0117, 0.0092, 0.0163, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:24:16,927 INFO [train.py:893] (0/4) Epoch 20, batch 1900, loss[loss=0.2079, simple_loss=0.2662, pruned_loss=0.07476, over 13564.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2371, pruned_loss=0.05889, over 2662620.46 frames. ], batch size: 89, lr: 6.51e-03, grad_scale: 32.0 2023-04-17 01:24:42,994 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53211.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 01:24:49,387 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.873e+02 2.617e+02 3.070e+02 3.654e+02 7.055e+02, threshold=6.140e+02, percent-clipped=1.0 2023-04-17 01:25:03,680 INFO [train.py:893] (0/4) Epoch 20, batch 1950, loss[loss=0.1792, simple_loss=0.2459, pruned_loss=0.05628, over 13445.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2371, pruned_loss=0.05874, over 2662180.82 frames. ], batch size: 106, lr: 6.51e-03, grad_scale: 32.0 2023-04-17 01:25:11,378 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53241.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:25:27,805 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53259.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 01:25:36,441 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-17 01:25:50,121 INFO [train.py:893] (0/4) Epoch 20, batch 2000, loss[loss=0.1522, simple_loss=0.1988, pruned_loss=0.05279, over 11516.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2396, pruned_loss=0.05991, over 2663653.86 frames. ], batch size: 47, lr: 6.50e-03, grad_scale: 32.0 2023-04-17 01:25:56,684 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 01:26:08,442 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53302.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:26:10,137 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53304.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:26:20,363 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53315.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:26:23,461 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.898e+02 2.652e+02 3.059e+02 3.675e+02 5.199e+02, threshold=6.119e+02, percent-clipped=0.0 2023-04-17 01:26:37,650 INFO [train.py:893] (0/4) Epoch 20, batch 2050, loss[loss=0.1636, simple_loss=0.229, pruned_loss=0.04912, over 13483.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.2411, pruned_loss=0.0608, over 2664393.70 frames. ], batch size: 81, lr: 6.50e-03, grad_scale: 32.0 2023-04-17 01:27:02,967 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0477, 4.5323, 4.3343, 4.3142, 4.3307, 4.1330, 4.6169, 4.6694], device='cuda:0'), covar=tensor([0.0254, 0.0234, 0.0217, 0.0344, 0.0263, 0.0287, 0.0294, 0.0217], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0196, 0.0157, 0.0178, 0.0144, 0.0192, 0.0132, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 01:27:08,009 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53365.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:27:10,408 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53367.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:27:18,182 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53376.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:27:24,731 INFO [train.py:893] (0/4) Epoch 20, batch 2100, loss[loss=0.1987, simple_loss=0.2559, pruned_loss=0.07072, over 13446.00 frames. ], tot_loss[loss=0.1809, simple_loss=0.2408, pruned_loss=0.06055, over 2665433.37 frames. ], batch size: 106, lr: 6.50e-03, grad_scale: 32.0 2023-04-17 01:27:49,315 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53408.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:27:52,579 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53412.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:27:57,472 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.055e+02 2.630e+02 3.030e+02 3.715e+02 5.974e+02, threshold=6.059e+02, percent-clipped=0.0 2023-04-17 01:28:08,129 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53428.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 01:28:09,809 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8956, 2.5006, 2.6007, 2.9112, 2.2831, 2.9876, 2.9783, 2.4204], device='cuda:0'), covar=tensor([0.0110, 0.0193, 0.0163, 0.0167, 0.0195, 0.0121, 0.0154, 0.0178], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0113, 0.0120, 0.0118, 0.0127, 0.0105, 0.0105, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 01:28:12,180 INFO [train.py:893] (0/4) Epoch 20, batch 2150, loss[loss=0.1637, simple_loss=0.2274, pruned_loss=0.04996, over 13496.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.2417, pruned_loss=0.06066, over 2660073.15 frames. ], batch size: 70, lr: 6.50e-03, grad_scale: 32.0 2023-04-17 01:28:19,389 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0806, 1.9786, 3.9815, 3.8296, 3.8785, 3.1381, 3.7277, 2.9318], device='cuda:0'), covar=tensor([0.2119, 0.1823, 0.0122, 0.0184, 0.0225, 0.0714, 0.0219, 0.0850], device='cuda:0'), in_proj_covar=tensor([0.0214, 0.0186, 0.0121, 0.0126, 0.0131, 0.0171, 0.0139, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 01:28:34,424 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53456.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:28:52,055 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.68 vs. limit=5.0 2023-04-17 01:28:59,756 INFO [train.py:893] (0/4) Epoch 20, batch 2200, loss[loss=0.1765, simple_loss=0.2265, pruned_loss=0.06321, over 13346.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2408, pruned_loss=0.06005, over 2662942.93 frames. ], batch size: 67, lr: 6.49e-03, grad_scale: 32.0 2023-04-17 01:29:22,868 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4221, 3.3000, 4.0019, 2.8944, 2.6997, 2.6535, 4.3326, 4.3878], device='cuda:0'), covar=tensor([0.1297, 0.1855, 0.0418, 0.1836, 0.1633, 0.1913, 0.0254, 0.0261], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0264, 0.0191, 0.0222, 0.0217, 0.0181, 0.0204, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 01:29:26,236 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8933, 2.0985, 3.6538, 3.5737, 3.4949, 2.7959, 3.3827, 2.7242], device='cuda:0'), covar=tensor([0.2146, 0.1502, 0.0144, 0.0190, 0.0329, 0.0791, 0.0266, 0.0845], device='cuda:0'), in_proj_covar=tensor([0.0214, 0.0185, 0.0121, 0.0126, 0.0131, 0.0172, 0.0140, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 01:29:32,363 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.983e+02 2.529e+02 2.934e+02 3.600e+02 1.155e+03, threshold=5.867e+02, percent-clipped=3.0 2023-04-17 01:29:45,638 INFO [train.py:893] (0/4) Epoch 20, batch 2250, loss[loss=0.1652, simple_loss=0.2346, pruned_loss=0.04789, over 13463.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2392, pruned_loss=0.05955, over 2652136.17 frames. ], batch size: 106, lr: 6.49e-03, grad_scale: 32.0 2023-04-17 01:30:22,285 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1316, 4.2978, 3.0186, 3.8914, 4.1291, 2.7594, 3.7089, 2.8798], device='cuda:0'), covar=tensor([0.0263, 0.0208, 0.0961, 0.0391, 0.0240, 0.1100, 0.0462, 0.1182], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0173, 0.0178, 0.0207, 0.0136, 0.0159, 0.0160, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:30:32,426 INFO [train.py:893] (0/4) Epoch 20, batch 2300, loss[loss=0.1595, simple_loss=0.2064, pruned_loss=0.05624, over 12063.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2387, pruned_loss=0.05942, over 2653020.03 frames. ], batch size: 49, lr: 6.49e-03, grad_scale: 32.0 2023-04-17 01:30:44,837 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53597.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:31:05,028 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 2.476e+02 2.872e+02 3.310e+02 5.586e+02, threshold=5.744e+02, percent-clipped=0.0 2023-04-17 01:31:19,014 INFO [train.py:893] (0/4) Epoch 20, batch 2350, loss[loss=0.1815, simple_loss=0.2376, pruned_loss=0.06265, over 13488.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2383, pruned_loss=0.05914, over 2650291.47 frames. ], batch size: 81, lr: 6.48e-03, grad_scale: 32.0 2023-04-17 01:31:42,139 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 01:31:43,975 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53660.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:31:54,523 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53671.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:32:05,814 INFO [train.py:893] (0/4) Epoch 20, batch 2400, loss[loss=0.1878, simple_loss=0.2457, pruned_loss=0.06494, over 13438.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2369, pruned_loss=0.05897, over 2653719.72 frames. ], batch size: 100, lr: 6.48e-03, grad_scale: 32.0 2023-04-17 01:32:11,693 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-17 01:32:32,877 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53712.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:32:38,394 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.897e+02 2.663e+02 3.122e+02 3.605e+02 1.210e+03, threshold=6.243e+02, percent-clipped=2.0 2023-04-17 01:32:42,843 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53723.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 01:32:51,772 INFO [train.py:893] (0/4) Epoch 20, batch 2450, loss[loss=0.1846, simple_loss=0.2443, pruned_loss=0.06251, over 13391.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2369, pruned_loss=0.05878, over 2655669.98 frames. ], batch size: 109, lr: 6.48e-03, grad_scale: 32.0 2023-04-17 01:33:01,121 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6254, 2.3145, 1.9836, 1.4244, 1.8818, 1.8310, 2.0576, 2.4200], device='cuda:0'), covar=tensor([0.0747, 0.0243, 0.0638, 0.1192, 0.0176, 0.0459, 0.0561, 0.0239], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0139, 0.0119, 0.0207, 0.0109, 0.0158, 0.0171, 0.0129], device='cuda:0'), out_proj_covar=tensor([1.2329e-04, 1.0443e-04, 9.3300e-05, 1.5506e-04, 7.9953e-05, 1.1964e-04, 1.2924e-04, 9.5716e-05], device='cuda:0') 2023-04-17 01:33:17,553 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53760.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:33:38,966 INFO [train.py:893] (0/4) Epoch 20, batch 2500, loss[loss=0.1639, simple_loss=0.2153, pruned_loss=0.05627, over 13426.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2366, pruned_loss=0.05876, over 2647259.64 frames. ], batch size: 62, lr: 6.47e-03, grad_scale: 32.0 2023-04-17 01:34:11,278 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.985e+02 2.616e+02 2.994e+02 3.505e+02 5.181e+02, threshold=5.987e+02, percent-clipped=0.0 2023-04-17 01:34:26,072 INFO [train.py:893] (0/4) Epoch 20, batch 2550, loss[loss=0.1961, simple_loss=0.2528, pruned_loss=0.0697, over 13364.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2364, pruned_loss=0.05854, over 2649148.45 frames. ], batch size: 109, lr: 6.47e-03, grad_scale: 32.0 2023-04-17 01:34:48,590 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 01:35:11,914 INFO [train.py:893] (0/4) Epoch 20, batch 2600, loss[loss=0.1642, simple_loss=0.2302, pruned_loss=0.04912, over 13469.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2363, pruned_loss=0.05848, over 2653761.07 frames. ], batch size: 103, lr: 6.47e-03, grad_scale: 32.0 2023-04-17 01:35:19,892 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7818, 3.6105, 3.7937, 2.2991, 4.0000, 3.8184, 3.8595, 4.0157], device='cuda:0'), covar=tensor([0.0233, 0.0140, 0.0146, 0.1171, 0.0142, 0.0238, 0.0121, 0.0105], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0055, 0.0081, 0.0103, 0.0098, 0.0108, 0.0079, 0.0075], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:35:26,236 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53897.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:35:28,786 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7686, 2.4384, 2.4089, 2.8591, 2.0826, 2.8783, 2.6876, 2.2401], device='cuda:0'), covar=tensor([0.0115, 0.0193, 0.0152, 0.0143, 0.0218, 0.0119, 0.0263, 0.0207], device='cuda:0'), in_proj_covar=tensor([0.0104, 0.0114, 0.0121, 0.0118, 0.0129, 0.0107, 0.0106, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 01:35:43,023 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.831e+02 2.669e+02 3.176e+02 3.708e+02 8.423e+02, threshold=6.351e+02, percent-clipped=2.0 2023-04-17 01:35:54,208 INFO [train.py:893] (0/4) Epoch 20, batch 2650, loss[loss=0.1879, simple_loss=0.2429, pruned_loss=0.06641, over 13537.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2381, pruned_loss=0.05991, over 2655505.64 frames. ], batch size: 85, lr: 6.47e-03, grad_scale: 32.0 2023-04-17 01:36:03,756 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53945.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:36:15,741 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53960.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:36:24,043 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53971.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:36:33,087 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-20.pt 2023-04-17 01:36:58,589 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-17 01:37:02,396 INFO [train.py:893] (0/4) Epoch 21, batch 0, loss[loss=0.185, simple_loss=0.2305, pruned_loss=0.06977, over 13168.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2305, pruned_loss=0.06977, over 13168.00 frames. ], batch size: 58, lr: 6.30e-03, grad_scale: 32.0 2023-04-17 01:37:02,396 INFO [train.py:918] (0/4) Computing validation loss 2023-04-17 01:37:21,102 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5096, 4.5393, 4.5633, 4.5443, 4.8107, 4.5102, 4.7434, 4.7662], device='cuda:0'), covar=tensor([0.0399, 0.0453, 0.0603, 0.0456, 0.0501, 0.0660, 0.0471, 0.0437], device='cuda:0'), in_proj_covar=tensor([0.0281, 0.0291, 0.0299, 0.0223, 0.0422, 0.0335, 0.0271, 0.0296], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:37:25,154 INFO [train.py:927] (0/4) Epoch 21, validation: loss=0.1376, simple_loss=0.1969, pruned_loss=0.03912, over 2446609.00 frames. 2023-04-17 01:37:25,155 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-17 01:37:43,039 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-54000.pt 2023-04-17 01:37:54,557 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=54008.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:37:54,649 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0387, 4.2682, 4.0347, 4.1110, 4.2018, 4.4801, 4.3156, 4.0865], device='cuda:0'), covar=tensor([0.0464, 0.0342, 0.0520, 0.0907, 0.0398, 0.0298, 0.0459, 0.0366], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0154, 0.0173, 0.0262, 0.0174, 0.0189, 0.0170, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 01:38:03,427 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.635e+02 2.723e+02 3.218e+02 3.819e+02 6.582e+02, threshold=6.437e+02, percent-clipped=1.0 2023-04-17 01:38:04,497 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=54019.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:38:08,770 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=54023.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 01:38:16,637 INFO [train.py:893] (0/4) Epoch 21, batch 50, loss[loss=0.176, simple_loss=0.2375, pruned_loss=0.05728, over 13448.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2349, pruned_loss=0.06048, over 597919.93 frames. ], batch size: 103, lr: 6.30e-03, grad_scale: 32.0 2023-04-17 01:38:24,595 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9790, 2.4202, 1.9767, 3.8705, 4.2235, 3.1917, 4.1979, 3.9671], device='cuda:0'), covar=tensor([0.0088, 0.0933, 0.0988, 0.0081, 0.0060, 0.0469, 0.0075, 0.0073], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0091, 0.0096, 0.0081, 0.0066, 0.0080, 0.0055, 0.0076], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 01:38:39,951 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-17 01:38:39,951 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-17 01:38:39,951 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-17 01:38:39,958 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-17 01:38:39,976 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-17 01:38:39,989 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-17 01:38:40,006 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-17 01:38:53,038 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=54071.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:38:56,804 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-17 01:39:02,796 INFO [train.py:893] (0/4) Epoch 21, batch 100, loss[loss=0.1796, simple_loss=0.2409, pruned_loss=0.05911, over 13368.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2366, pruned_loss=0.06082, over 1053707.38 frames. ], batch size: 109, lr: 6.30e-03, grad_scale: 32.0 2023-04-17 01:39:20,520 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.0572, 4.9138, 5.1405, 4.9195, 5.4516, 4.9414, 5.4418, 5.4133], device='cuda:0'), covar=tensor([0.0411, 0.0628, 0.0650, 0.0673, 0.0510, 0.0883, 0.0431, 0.0466], device='cuda:0'), in_proj_covar=tensor([0.0280, 0.0289, 0.0298, 0.0221, 0.0420, 0.0335, 0.0268, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:39:24,039 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.50 vs. limit=5.0 2023-04-17 01:39:36,885 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.038e+02 2.853e+02 3.270e+02 3.963e+02 8.282e+02, threshold=6.541e+02, percent-clipped=1.0 2023-04-17 01:39:41,408 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6221, 3.3587, 2.6796, 3.0016, 2.8099, 1.8569, 3.4578, 2.1039], device='cuda:0'), covar=tensor([0.0717, 0.0523, 0.0522, 0.0460, 0.0736, 0.2194, 0.0853, 0.1209], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0137, 0.0133, 0.0115, 0.0147, 0.0188, 0.0170, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 01:39:49,418 INFO [train.py:893] (0/4) Epoch 21, batch 150, loss[loss=0.1748, simple_loss=0.2298, pruned_loss=0.0599, over 13544.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2386, pruned_loss=0.06196, over 1410803.59 frames. ], batch size: 74, lr: 6.30e-03, grad_scale: 32.0 2023-04-17 01:39:55,749 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-17 01:40:14,208 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8235, 3.2184, 3.1393, 3.4616, 2.1830, 2.7756, 3.2865, 1.9861], device='cuda:0'), covar=tensor([0.0155, 0.0607, 0.0659, 0.0410, 0.1495, 0.0961, 0.0596, 0.1872], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0183, 0.0206, 0.0241, 0.0184, 0.0199, 0.0178, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:40:32,975 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1762, 4.3805, 3.3063, 2.9698, 3.1604, 2.5979, 4.4354, 2.5296], device='cuda:0'), covar=tensor([0.1538, 0.0322, 0.1040, 0.2001, 0.0786, 0.3169, 0.0232, 0.3685], device='cuda:0'), in_proj_covar=tensor([0.0312, 0.0287, 0.0312, 0.0327, 0.0257, 0.0324, 0.0208, 0.0363], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 01:40:36,878 INFO [train.py:893] (0/4) Epoch 21, batch 200, loss[loss=0.1931, simple_loss=0.2488, pruned_loss=0.06869, over 13536.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2402, pruned_loss=0.06327, over 1671273.09 frames. ], batch size: 83, lr: 6.29e-03, grad_scale: 32.0 2023-04-17 01:41:10,110 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 2.703e+02 3.137e+02 3.628e+02 8.630e+02, threshold=6.273e+02, percent-clipped=1.0 2023-04-17 01:41:23,015 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-17 01:41:23,278 INFO [train.py:893] (0/4) Epoch 21, batch 250, loss[loss=0.1656, simple_loss=0.2208, pruned_loss=0.0552, over 13395.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2404, pruned_loss=0.06307, over 1885356.98 frames. ], batch size: 62, lr: 6.29e-03, grad_scale: 32.0 2023-04-17 01:42:09,742 INFO [train.py:893] (0/4) Epoch 21, batch 300, loss[loss=0.1977, simple_loss=0.2546, pruned_loss=0.07045, over 13481.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2403, pruned_loss=0.06263, over 2060775.08 frames. ], batch size: 93, lr: 6.29e-03, grad_scale: 32.0 2023-04-17 01:42:11,752 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6624, 2.4048, 2.2736, 2.7957, 1.9587, 2.7956, 2.6112, 2.1226], device='cuda:0'), covar=tensor([0.0132, 0.0221, 0.0228, 0.0195, 0.0328, 0.0154, 0.0260, 0.0269], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0112, 0.0120, 0.0118, 0.0127, 0.0107, 0.0105, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 01:42:43,127 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.688e+02 2.678e+02 3.033e+02 3.503e+02 6.854e+02, threshold=6.067e+02, percent-clipped=2.0 2023-04-17 01:42:55,536 INFO [train.py:893] (0/4) Epoch 21, batch 350, loss[loss=0.1984, simple_loss=0.2578, pruned_loss=0.06948, over 13453.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2393, pruned_loss=0.06233, over 2195784.89 frames. ], batch size: 103, lr: 6.28e-03, grad_scale: 32.0 2023-04-17 01:43:01,820 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.4231, 2.1510, 2.2329, 2.4157, 1.8828, 2.4348, 2.3863, 1.9977], device='cuda:0'), covar=tensor([0.0083, 0.0215, 0.0131, 0.0135, 0.0214, 0.0158, 0.0175, 0.0191], device='cuda:0'), in_proj_covar=tensor([0.0104, 0.0113, 0.0121, 0.0119, 0.0128, 0.0108, 0.0106, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 01:43:25,825 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-17 01:43:42,905 INFO [train.py:893] (0/4) Epoch 21, batch 400, loss[loss=0.1928, simple_loss=0.2532, pruned_loss=0.06616, over 13402.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2407, pruned_loss=0.06261, over 2299976.04 frames. ], batch size: 109, lr: 6.28e-03, grad_scale: 32.0 2023-04-17 01:43:44,695 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1969, 4.6509, 4.4318, 4.4193, 4.4109, 4.2863, 4.6661, 4.7009], device='cuda:0'), covar=tensor([0.0238, 0.0231, 0.0265, 0.0362, 0.0298, 0.0273, 0.0262, 0.0226], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0196, 0.0159, 0.0177, 0.0144, 0.0194, 0.0133, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 01:44:04,608 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-17 01:44:17,086 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.854e+02 2.635e+02 3.018e+02 3.787e+02 7.372e+02, threshold=6.036e+02, percent-clipped=3.0 2023-04-17 01:44:25,087 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.37 vs. limit=5.0 2023-04-17 01:44:29,300 INFO [train.py:893] (0/4) Epoch 21, batch 450, loss[loss=0.1859, simple_loss=0.2536, pruned_loss=0.0591, over 13477.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2423, pruned_loss=0.06318, over 2379725.45 frames. ], batch size: 81, lr: 6.28e-03, grad_scale: 32.0 2023-04-17 01:44:53,173 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-17 01:45:05,062 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.52 vs. limit=5.0 2023-04-17 01:45:16,436 INFO [train.py:893] (0/4) Epoch 21, batch 500, loss[loss=0.1834, simple_loss=0.2524, pruned_loss=0.05722, over 13442.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2421, pruned_loss=0.06261, over 2442642.04 frames. ], batch size: 95, lr: 6.28e-03, grad_scale: 32.0 2023-04-17 01:45:49,140 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.865e+02 2.575e+02 3.052e+02 3.653e+02 5.536e+02, threshold=6.104e+02, percent-clipped=0.0 2023-04-17 01:46:02,458 INFO [train.py:893] (0/4) Epoch 21, batch 550, loss[loss=0.1563, simple_loss=0.2254, pruned_loss=0.04365, over 13500.00 frames. ], tot_loss[loss=0.1831, simple_loss=0.2421, pruned_loss=0.06203, over 2489451.29 frames. ], batch size: 81, lr: 6.27e-03, grad_scale: 32.0 2023-04-17 01:46:41,943 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4371, 4.7143, 4.4825, 4.3536, 4.5152, 4.8212, 4.6805, 4.4729], device='cuda:0'), covar=tensor([0.0319, 0.0270, 0.0279, 0.0929, 0.0277, 0.0222, 0.0304, 0.0252], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0156, 0.0174, 0.0263, 0.0174, 0.0190, 0.0170, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 01:46:49,284 INFO [train.py:893] (0/4) Epoch 21, batch 600, loss[loss=0.1854, simple_loss=0.2476, pruned_loss=0.06157, over 13534.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2409, pruned_loss=0.06175, over 2524575.28 frames. ], batch size: 98, lr: 6.27e-03, grad_scale: 32.0 2023-04-17 01:47:23,766 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.860e+02 2.541e+02 2.955e+02 3.597e+02 5.812e+02, threshold=5.910e+02, percent-clipped=0.0 2023-04-17 01:47:37,106 INFO [train.py:893] (0/4) Epoch 21, batch 650, loss[loss=0.1769, simple_loss=0.2405, pruned_loss=0.05662, over 13359.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2404, pruned_loss=0.06144, over 2555603.83 frames. ], batch size: 73, lr: 6.27e-03, grad_scale: 32.0 2023-04-17 01:48:07,449 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=54664.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 01:48:13,349 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-17 01:48:23,888 INFO [train.py:893] (0/4) Epoch 21, batch 700, loss[loss=0.1814, simple_loss=0.2364, pruned_loss=0.06314, over 13551.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2387, pruned_loss=0.06062, over 2577828.73 frames. ], batch size: 87, lr: 6.26e-03, grad_scale: 32.0 2023-04-17 01:48:44,127 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2548, 4.5454, 4.2827, 4.1437, 4.4015, 4.6821, 4.4631, 4.3050], device='cuda:0'), covar=tensor([0.0315, 0.0286, 0.0345, 0.1145, 0.0265, 0.0270, 0.0300, 0.0347], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0155, 0.0174, 0.0263, 0.0173, 0.0190, 0.0171, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 01:48:45,178 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.53 vs. limit=5.0 2023-04-17 01:48:57,156 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.970e+02 2.709e+02 2.989e+02 3.684e+02 5.126e+02, threshold=5.978e+02, percent-clipped=0.0 2023-04-17 01:48:57,382 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.4432, 4.8975, 4.7687, 5.0148, 4.8102, 4.7998, 5.4551, 5.0103], device='cuda:0'), covar=tensor([0.0660, 0.1146, 0.2134, 0.2053, 0.0854, 0.1510, 0.0756, 0.0949], device='cuda:0'), in_proj_covar=tensor([0.0326, 0.0371, 0.0459, 0.0463, 0.0289, 0.0345, 0.0425, 0.0340], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 01:49:04,953 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=54725.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 01:49:11,224 INFO [train.py:893] (0/4) Epoch 21, batch 750, loss[loss=0.1929, simple_loss=0.2564, pruned_loss=0.06472, over 13556.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2392, pruned_loss=0.06112, over 2594462.00 frames. ], batch size: 89, lr: 6.26e-03, grad_scale: 32.0 2023-04-17 01:49:42,474 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2672, 4.7707, 4.6781, 4.8411, 4.6505, 4.7218, 5.2764, 4.8697], device='cuda:0'), covar=tensor([0.0753, 0.1275, 0.2233, 0.2360, 0.0863, 0.1321, 0.0845, 0.1131], device='cuda:0'), in_proj_covar=tensor([0.0329, 0.0374, 0.0463, 0.0468, 0.0291, 0.0348, 0.0428, 0.0343], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 01:49:56,453 INFO [train.py:893] (0/4) Epoch 21, batch 800, loss[loss=0.1745, simple_loss=0.2409, pruned_loss=0.05403, over 13542.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.24, pruned_loss=0.06141, over 2615483.37 frames. ], batch size: 78, lr: 6.26e-03, grad_scale: 64.0 2023-04-17 01:50:21,669 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.4536, 2.0879, 2.1371, 2.3753, 1.8736, 2.4213, 2.2651, 1.9754], device='cuda:0'), covar=tensor([0.0072, 0.0170, 0.0114, 0.0106, 0.0168, 0.0108, 0.0167, 0.0150], device='cuda:0'), in_proj_covar=tensor([0.0104, 0.0114, 0.0121, 0.0120, 0.0130, 0.0107, 0.0106, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 01:50:31,164 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.910e+02 2.667e+02 3.203e+02 3.678e+02 7.837e+02, threshold=6.406e+02, percent-clipped=2.0 2023-04-17 01:50:43,692 INFO [train.py:893] (0/4) Epoch 21, batch 850, loss[loss=0.1715, simple_loss=0.2352, pruned_loss=0.05385, over 13517.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2411, pruned_loss=0.06152, over 2628181.03 frames. ], batch size: 76, lr: 6.26e-03, grad_scale: 64.0 2023-04-17 01:51:11,904 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=54862.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:51:30,866 INFO [train.py:893] (0/4) Epoch 21, batch 900, loss[loss=0.1889, simple_loss=0.2428, pruned_loss=0.06752, over 13539.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2414, pruned_loss=0.06223, over 2636976.73 frames. ], batch size: 87, lr: 6.25e-03, grad_scale: 32.0 2023-04-17 01:52:00,543 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-17 01:52:05,573 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.031e+02 2.828e+02 3.269e+02 3.911e+02 9.881e+02, threshold=6.537e+02, percent-clipped=2.0 2023-04-17 01:52:09,241 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=54923.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 01:52:13,372 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4656, 2.3125, 2.8795, 3.8946, 3.5366, 3.9969, 3.0664, 2.3102], device='cuda:0'), covar=tensor([0.0276, 0.0860, 0.0630, 0.0065, 0.0234, 0.0054, 0.0584, 0.0886], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0149, 0.0166, 0.0098, 0.0119, 0.0095, 0.0166, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:52:17,033 INFO [train.py:893] (0/4) Epoch 21, batch 950, loss[loss=0.1702, simple_loss=0.234, pruned_loss=0.05325, over 13369.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2403, pruned_loss=0.06202, over 2645186.82 frames. ], batch size: 109, lr: 6.25e-03, grad_scale: 32.0 2023-04-17 01:53:03,583 INFO [train.py:893] (0/4) Epoch 21, batch 1000, loss[loss=0.1782, simple_loss=0.2319, pruned_loss=0.06221, over 13534.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2384, pruned_loss=0.06144, over 2643781.73 frames. ], batch size: 76, lr: 6.25e-03, grad_scale: 32.0 2023-04-17 01:53:20,683 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8964, 4.2980, 4.1421, 4.0981, 4.1667, 4.0082, 4.3717, 4.3980], device='cuda:0'), covar=tensor([0.0262, 0.0269, 0.0211, 0.0378, 0.0298, 0.0283, 0.0287, 0.0263], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0200, 0.0161, 0.0181, 0.0148, 0.0197, 0.0135, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:53:38,538 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.568e+02 2.603e+02 3.088e+02 3.704e+02 6.747e+02, threshold=6.176e+02, percent-clipped=2.0 2023-04-17 01:53:39,628 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55020.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 01:53:50,172 INFO [train.py:893] (0/4) Epoch 21, batch 1050, loss[loss=0.1526, simple_loss=0.2166, pruned_loss=0.0443, over 13228.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2364, pruned_loss=0.05996, over 2649558.81 frames. ], batch size: 58, lr: 6.24e-03, grad_scale: 32.0 2023-04-17 01:54:36,881 INFO [train.py:893] (0/4) Epoch 21, batch 1100, loss[loss=0.1792, simple_loss=0.2323, pruned_loss=0.06301, over 13350.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2375, pruned_loss=0.05984, over 2655591.26 frames. ], batch size: 67, lr: 6.24e-03, grad_scale: 32.0 2023-04-17 01:55:11,818 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.920e+02 2.491e+02 2.960e+02 3.530e+02 7.455e+02, threshold=5.920e+02, percent-clipped=4.0 2023-04-17 01:55:22,741 INFO [train.py:893] (0/4) Epoch 21, batch 1150, loss[loss=0.1489, simple_loss=0.2158, pruned_loss=0.04097, over 13370.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.2369, pruned_loss=0.05919, over 2659278.46 frames. ], batch size: 73, lr: 6.24e-03, grad_scale: 32.0 2023-04-17 01:55:35,488 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.7996, 4.5248, 4.8336, 4.7343, 5.0711, 4.6297, 5.0890, 5.0251], device='cuda:0'), covar=tensor([0.0377, 0.0614, 0.0633, 0.0494, 0.0510, 0.0768, 0.0439, 0.0440], device='cuda:0'), in_proj_covar=tensor([0.0284, 0.0293, 0.0302, 0.0224, 0.0426, 0.0338, 0.0276, 0.0295], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:55:39,443 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55148.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:56:10,687 INFO [train.py:893] (0/4) Epoch 21, batch 1200, loss[loss=0.1622, simple_loss=0.2215, pruned_loss=0.05144, over 13365.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2379, pruned_loss=0.0592, over 2660761.76 frames. ], batch size: 67, lr: 6.24e-03, grad_scale: 16.0 2023-04-17 01:56:28,727 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6454, 4.0382, 3.8594, 3.8656, 3.9644, 3.7911, 4.1323, 4.1465], device='cuda:0'), covar=tensor([0.0257, 0.0290, 0.0239, 0.0310, 0.0237, 0.0288, 0.0279, 0.0304], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0198, 0.0159, 0.0178, 0.0147, 0.0195, 0.0133, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 01:56:35,505 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2023-04-17 01:56:36,192 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55209.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:56:36,793 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-17 01:56:37,912 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55211.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:56:42,753 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4039, 1.9748, 2.6316, 3.7962, 3.4771, 3.8804, 3.0966, 2.1620], device='cuda:0'), covar=tensor([0.0301, 0.1387, 0.0887, 0.0086, 0.0273, 0.0082, 0.0751, 0.1152], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0148, 0.0165, 0.0097, 0.0118, 0.0094, 0.0165, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:56:44,313 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55218.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 01:56:45,832 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.729e+02 2.461e+02 2.911e+02 3.858e+02 6.055e+02, threshold=5.823e+02, percent-clipped=2.0 2023-04-17 01:56:47,663 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-17 01:56:57,272 INFO [train.py:893] (0/4) Epoch 21, batch 1250, loss[loss=0.1916, simple_loss=0.2527, pruned_loss=0.06522, over 13458.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2391, pruned_loss=0.06007, over 2657442.71 frames. ], batch size: 100, lr: 6.23e-03, grad_scale: 16.0 2023-04-17 01:56:58,769 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-17 01:57:08,554 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2023-04-17 01:57:34,393 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55272.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 01:57:43,313 INFO [train.py:893] (0/4) Epoch 21, batch 1300, loss[loss=0.1968, simple_loss=0.2592, pruned_loss=0.06715, over 13485.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.2401, pruned_loss=0.06077, over 2657284.28 frames. ], batch size: 100, lr: 6.23e-03, grad_scale: 16.0 2023-04-17 01:58:02,855 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-17 01:58:17,914 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.936e+02 2.608e+02 3.169e+02 3.646e+02 7.659e+02, threshold=6.337e+02, percent-clipped=2.0 2023-04-17 01:58:18,178 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55320.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 01:58:30,164 INFO [train.py:893] (0/4) Epoch 21, batch 1350, loss[loss=0.2, simple_loss=0.2573, pruned_loss=0.07137, over 13367.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2404, pruned_loss=0.06102, over 2661131.03 frames. ], batch size: 118, lr: 6.23e-03, grad_scale: 16.0 2023-04-17 01:58:32,155 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2591, 4.2970, 4.2973, 2.7935, 4.7219, 4.3910, 4.3787, 4.6572], device='cuda:0'), covar=tensor([0.0285, 0.0137, 0.0162, 0.1218, 0.0191, 0.0265, 0.0201, 0.0132], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0055, 0.0082, 0.0102, 0.0097, 0.0107, 0.0079, 0.0074], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 01:59:04,320 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55368.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 01:59:16,583 INFO [train.py:893] (0/4) Epoch 21, batch 1400, loss[loss=0.1562, simple_loss=0.2201, pruned_loss=0.04616, over 13375.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2383, pruned_loss=0.05985, over 2663457.04 frames. ], batch size: 73, lr: 6.22e-03, grad_scale: 16.0 2023-04-17 01:59:44,792 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2707, 4.6744, 4.5149, 4.4490, 4.5107, 4.3076, 4.7889, 4.7481], device='cuda:0'), covar=tensor([0.0236, 0.0238, 0.0226, 0.0394, 0.0283, 0.0277, 0.0247, 0.0251], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0200, 0.0161, 0.0180, 0.0148, 0.0196, 0.0134, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 01:59:52,764 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.807e+02 2.623e+02 2.979e+02 3.447e+02 8.566e+02, threshold=5.958e+02, percent-clipped=5.0 2023-04-17 01:59:53,041 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7811, 4.0286, 3.7749, 3.8317, 3.9367, 4.1619, 4.0074, 3.7261], device='cuda:0'), covar=tensor([0.0264, 0.0250, 0.0337, 0.0741, 0.0244, 0.0215, 0.0255, 0.0610], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0155, 0.0173, 0.0261, 0.0173, 0.0189, 0.0169, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 02:00:03,880 INFO [train.py:893] (0/4) Epoch 21, batch 1450, loss[loss=0.1693, simple_loss=0.2189, pruned_loss=0.05984, over 11930.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2382, pruned_loss=0.06025, over 2662694.30 frames. ], batch size: 157, lr: 6.22e-03, grad_scale: 16.0 2023-04-17 02:00:27,264 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4497, 4.5477, 3.2270, 4.1227, 4.4237, 2.8473, 3.9647, 3.0659], device='cuda:0'), covar=tensor([0.0240, 0.0179, 0.0988, 0.0480, 0.0184, 0.1112, 0.0415, 0.1267], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0175, 0.0178, 0.0215, 0.0137, 0.0160, 0.0162, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:00:30,985 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3021, 4.6106, 4.2608, 4.3974, 4.4500, 4.7573, 4.5316, 4.4665], device='cuda:0'), covar=tensor([0.0298, 0.0254, 0.0307, 0.0735, 0.0256, 0.0214, 0.0288, 0.0218], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0155, 0.0173, 0.0261, 0.0173, 0.0189, 0.0169, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 02:00:47,873 INFO [train.py:893] (0/4) Epoch 21, batch 1500, loss[loss=0.1554, simple_loss=0.204, pruned_loss=0.05337, over 12732.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.238, pruned_loss=0.0601, over 2662216.26 frames. ], batch size: 52, lr: 6.22e-03, grad_scale: 16.0 2023-04-17 02:01:08,548 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55504.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:01:22,000 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55518.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 02:01:23,374 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.154e+02 2.733e+02 3.047e+02 3.591e+02 6.078e+02, threshold=6.094e+02, percent-clipped=2.0 2023-04-17 02:01:34,161 INFO [train.py:893] (0/4) Epoch 21, batch 1550, loss[loss=0.166, simple_loss=0.2228, pruned_loss=0.05455, over 13365.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2385, pruned_loss=0.06005, over 2661986.99 frames. ], batch size: 67, lr: 6.22e-03, grad_scale: 16.0 2023-04-17 02:02:05,321 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55566.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:02:06,219 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55567.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:02:20,733 INFO [train.py:893] (0/4) Epoch 21, batch 1600, loss[loss=0.181, simple_loss=0.2426, pruned_loss=0.0597, over 11759.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2389, pruned_loss=0.05994, over 2659549.70 frames. ], batch size: 157, lr: 6.21e-03, grad_scale: 16.0 2023-04-17 02:02:30,743 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6721, 3.5357, 2.7083, 3.1090, 2.8125, 2.0645, 3.5253, 2.0128], device='cuda:0'), covar=tensor([0.0685, 0.0493, 0.0525, 0.0474, 0.0738, 0.1909, 0.0996, 0.1401], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0140, 0.0135, 0.0116, 0.0146, 0.0186, 0.0173, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 02:02:54,996 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4524, 4.1564, 3.4721, 3.9773, 3.3964, 2.5922, 4.1631, 2.5802], device='cuda:0'), covar=tensor([0.0498, 0.0419, 0.0397, 0.0226, 0.0585, 0.1604, 0.0806, 0.1057], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0139, 0.0134, 0.0116, 0.0145, 0.0186, 0.0172, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 02:02:55,243 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2023-04-17 02:02:56,301 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.917e+02 2.614e+02 2.838e+02 3.865e+02 7.437e+02, threshold=5.675e+02, percent-clipped=3.0 2023-04-17 02:03:07,350 INFO [train.py:893] (0/4) Epoch 21, batch 1650, loss[loss=0.1624, simple_loss=0.2164, pruned_loss=0.05426, over 13508.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2391, pruned_loss=0.05988, over 2653013.58 frames. ], batch size: 70, lr: 6.21e-03, grad_scale: 16.0 2023-04-17 02:03:54,315 INFO [train.py:893] (0/4) Epoch 21, batch 1700, loss[loss=0.1525, simple_loss=0.2213, pruned_loss=0.04183, over 13534.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2394, pruned_loss=0.05909, over 2655716.66 frames. ], batch size: 76, lr: 6.21e-03, grad_scale: 16.0 2023-04-17 02:04:29,423 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.606e+02 2.727e+02 3.146e+02 3.947e+02 9.582e+02, threshold=6.292e+02, percent-clipped=5.0 2023-04-17 02:04:40,170 INFO [train.py:893] (0/4) Epoch 21, batch 1750, loss[loss=0.1542, simple_loss=0.2157, pruned_loss=0.04633, over 13426.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2378, pruned_loss=0.05808, over 2657475.36 frames. ], batch size: 65, lr: 6.21e-03, grad_scale: 16.0 2023-04-17 02:04:46,231 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.4836, 2.4295, 2.0663, 1.4283, 1.6780, 2.1141, 2.0448, 2.6361], device='cuda:0'), covar=tensor([0.1065, 0.0301, 0.0875, 0.1602, 0.0187, 0.0440, 0.0715, 0.0249], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0143, 0.0123, 0.0210, 0.0113, 0.0162, 0.0173, 0.0130], device='cuda:0'), out_proj_covar=tensor([1.2330e-04, 1.0721e-04, 9.5895e-05, 1.5722e-04, 8.3020e-05, 1.2235e-04, 1.3113e-04, 9.6418e-05], device='cuda:0') 2023-04-17 02:04:55,985 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55748.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:05:12,549 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-17 02:05:25,977 INFO [train.py:893] (0/4) Epoch 21, batch 1800, loss[loss=0.1617, simple_loss=0.2288, pruned_loss=0.0473, over 13447.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2377, pruned_loss=0.05785, over 2654284.90 frames. ], batch size: 106, lr: 6.20e-03, grad_scale: 16.0 2023-04-17 02:05:46,252 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55804.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:05:50,476 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55809.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:06:00,122 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5113, 4.4323, 4.5398, 2.6928, 4.8287, 4.5854, 4.5995, 4.8089], device='cuda:0'), covar=tensor([0.0263, 0.0107, 0.0162, 0.1267, 0.0150, 0.0286, 0.0132, 0.0129], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0054, 0.0081, 0.0102, 0.0097, 0.0107, 0.0079, 0.0074], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:06:00,654 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.879e+02 2.483e+02 2.815e+02 3.302e+02 6.184e+02, threshold=5.630e+02, percent-clipped=0.0 2023-04-17 02:06:12,442 INFO [train.py:893] (0/4) Epoch 21, batch 1850, loss[loss=0.1749, simple_loss=0.2508, pruned_loss=0.04944, over 13441.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2377, pruned_loss=0.05778, over 2655447.49 frames. ], batch size: 95, lr: 6.20e-03, grad_scale: 16.0 2023-04-17 02:06:15,589 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-17 02:06:15,731 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-17 02:06:26,715 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55848.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:06:30,482 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55852.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:06:42,278 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9839, 4.3284, 4.0742, 4.0769, 4.1533, 4.4227, 4.2325, 4.0219], device='cuda:0'), covar=tensor([0.0299, 0.0235, 0.0288, 0.0787, 0.0257, 0.0227, 0.0314, 0.0460], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0155, 0.0175, 0.0264, 0.0174, 0.0190, 0.0170, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 02:06:44,812 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55867.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:06:58,009 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1354, 2.2926, 1.9914, 3.8922, 4.3517, 3.2820, 4.3067, 4.1451], device='cuda:0'), covar=tensor([0.0146, 0.1285, 0.1437, 0.0175, 0.0199, 0.0717, 0.0148, 0.0130], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0090, 0.0097, 0.0081, 0.0067, 0.0081, 0.0056, 0.0077], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 02:06:58,574 INFO [train.py:893] (0/4) Epoch 21, batch 1900, loss[loss=0.1957, simple_loss=0.2485, pruned_loss=0.07146, over 13479.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2377, pruned_loss=0.05828, over 2655315.21 frames. ], batch size: 93, lr: 6.20e-03, grad_scale: 16.0 2023-04-17 02:07:23,870 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55909.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:07:29,534 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55915.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:07:33,637 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.957e+02 2.684e+02 2.952e+02 3.656e+02 5.024e+02, threshold=5.905e+02, percent-clipped=0.0 2023-04-17 02:07:41,740 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-17 02:07:45,376 INFO [train.py:893] (0/4) Epoch 21, batch 1950, loss[loss=0.158, simple_loss=0.2183, pruned_loss=0.04889, over 13352.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2364, pruned_loss=0.05784, over 2657883.49 frames. ], batch size: 73, lr: 6.19e-03, grad_scale: 16.0 2023-04-17 02:08:31,990 INFO [train.py:893] (0/4) Epoch 21, batch 2000, loss[loss=0.1388, simple_loss=0.1988, pruned_loss=0.03943, over 13413.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2373, pruned_loss=0.05802, over 2661843.85 frames. ], batch size: 62, lr: 6.19e-03, grad_scale: 16.0 2023-04-17 02:08:39,736 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 02:08:50,042 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-56000.pt 2023-04-17 02:09:13,049 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 2.480e+02 2.958e+02 3.514e+02 5.155e+02, threshold=5.917e+02, percent-clipped=0.0 2023-04-17 02:09:23,724 INFO [train.py:893] (0/4) Epoch 21, batch 2050, loss[loss=0.1861, simple_loss=0.2434, pruned_loss=0.06442, over 13039.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2381, pruned_loss=0.05831, over 2660763.11 frames. ], batch size: 142, lr: 6.19e-03, grad_scale: 16.0 2023-04-17 02:10:09,182 INFO [train.py:893] (0/4) Epoch 21, batch 2100, loss[loss=0.1774, simple_loss=0.2416, pruned_loss=0.05656, over 13469.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2379, pruned_loss=0.05816, over 2662338.30 frames. ], batch size: 79, lr: 6.19e-03, grad_scale: 16.0 2023-04-17 02:10:29,004 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56104.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:10:38,526 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0770, 2.4832, 1.9805, 4.0071, 4.4657, 3.2753, 4.3291, 4.1596], device='cuda:0'), covar=tensor([0.0096, 0.0980, 0.1082, 0.0096, 0.0064, 0.0463, 0.0089, 0.0088], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0091, 0.0097, 0.0081, 0.0068, 0.0082, 0.0057, 0.0078], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 02:10:43,841 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.062e+02 2.607e+02 3.023e+02 3.462e+02 7.013e+02, threshold=6.046e+02, percent-clipped=1.0 2023-04-17 02:10:54,794 INFO [train.py:893] (0/4) Epoch 21, batch 2150, loss[loss=0.1661, simple_loss=0.2248, pruned_loss=0.05372, over 13471.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2375, pruned_loss=0.05778, over 2662419.97 frames. ], batch size: 79, lr: 6.18e-03, grad_scale: 16.0 2023-04-17 02:11:06,863 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56145.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:11:12,291 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-17 02:11:39,015 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-17 02:11:40,728 INFO [train.py:893] (0/4) Epoch 21, batch 2200, loss[loss=0.2177, simple_loss=0.2695, pruned_loss=0.08292, over 11513.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2369, pruned_loss=0.05751, over 2662125.58 frames. ], batch size: 157, lr: 6.18e-03, grad_scale: 16.0 2023-04-17 02:11:46,066 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3445, 2.0535, 3.8781, 3.7921, 3.8207, 3.0238, 3.5580, 2.9890], device='cuda:0'), covar=tensor([0.1775, 0.1529, 0.0137, 0.0183, 0.0210, 0.0666, 0.0260, 0.0720], device='cuda:0'), in_proj_covar=tensor([0.0217, 0.0186, 0.0122, 0.0128, 0.0132, 0.0174, 0.0144, 0.0184], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 02:12:01,649 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56204.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:12:03,495 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56206.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:12:15,680 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.840e+02 2.445e+02 3.046e+02 3.576e+02 5.421e+02, threshold=6.093e+02, percent-clipped=0.0 2023-04-17 02:12:26,667 INFO [train.py:893] (0/4) Epoch 21, batch 2250, loss[loss=0.1396, simple_loss=0.1945, pruned_loss=0.04232, over 13435.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2358, pruned_loss=0.05709, over 2656748.37 frames. ], batch size: 65, lr: 6.18e-03, grad_scale: 16.0 2023-04-17 02:12:53,119 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-17 02:13:12,795 INFO [train.py:893] (0/4) Epoch 21, batch 2300, loss[loss=0.1811, simple_loss=0.2435, pruned_loss=0.05936, over 13533.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2346, pruned_loss=0.05653, over 2657045.90 frames. ], batch size: 98, lr: 6.18e-03, grad_scale: 16.0 2023-04-17 02:13:48,868 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.697e+02 2.453e+02 2.876e+02 3.418e+02 5.342e+02, threshold=5.751e+02, percent-clipped=0.0 2023-04-17 02:13:59,955 INFO [train.py:893] (0/4) Epoch 21, batch 2350, loss[loss=0.1682, simple_loss=0.2309, pruned_loss=0.05274, over 13515.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2347, pruned_loss=0.05672, over 2659007.61 frames. ], batch size: 76, lr: 6.17e-03, grad_scale: 16.0 2023-04-17 02:14:09,568 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-17 02:14:21,858 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 02:14:42,749 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8946, 1.8380, 3.4725, 3.4554, 3.3723, 2.7023, 3.1887, 2.6025], device='cuda:0'), covar=tensor([0.1974, 0.1462, 0.0185, 0.0193, 0.0232, 0.0696, 0.0271, 0.0849], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0187, 0.0124, 0.0130, 0.0133, 0.0175, 0.0146, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 02:14:45,690 INFO [train.py:893] (0/4) Epoch 21, batch 2400, loss[loss=0.1624, simple_loss=0.2294, pruned_loss=0.04769, over 13515.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2347, pruned_loss=0.05737, over 2656496.13 frames. ], batch size: 76, lr: 6.17e-03, grad_scale: 16.0 2023-04-17 02:15:06,218 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56404.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:15:19,923 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.778e+02 2.549e+02 3.033e+02 3.703e+02 6.440e+02, threshold=6.067e+02, percent-clipped=2.0 2023-04-17 02:15:31,146 INFO [train.py:893] (0/4) Epoch 21, batch 2450, loss[loss=0.1703, simple_loss=0.2329, pruned_loss=0.05379, over 13522.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.2362, pruned_loss=0.05825, over 2658685.53 frames. ], batch size: 72, lr: 6.17e-03, grad_scale: 16.0 2023-04-17 02:15:49,580 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=56452.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:16:18,253 INFO [train.py:893] (0/4) Epoch 21, batch 2500, loss[loss=0.1556, simple_loss=0.2231, pruned_loss=0.04406, over 13501.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2359, pruned_loss=0.05815, over 2661477.42 frames. ], batch size: 81, lr: 6.16e-03, grad_scale: 16.0 2023-04-17 02:16:36,509 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56501.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:16:39,096 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56504.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:16:54,753 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.811e+02 2.337e+02 2.785e+02 3.508e+02 6.216e+02, threshold=5.571e+02, percent-clipped=2.0 2023-04-17 02:16:57,529 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56523.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 02:17:05,613 INFO [train.py:893] (0/4) Epoch 21, batch 2550, loss[loss=0.1642, simple_loss=0.2301, pruned_loss=0.04916, over 13357.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2362, pruned_loss=0.05799, over 2661240.44 frames. ], batch size: 73, lr: 6.16e-03, grad_scale: 16.0 2023-04-17 02:17:24,982 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=56552.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:17:28,216 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 02:17:40,915 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9610, 4.0608, 2.9664, 3.5902, 3.9277, 2.7881, 3.5524, 2.8409], device='cuda:0'), covar=tensor([0.0312, 0.0316, 0.0939, 0.0330, 0.0295, 0.1152, 0.0584, 0.1166], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0175, 0.0176, 0.0215, 0.0138, 0.0159, 0.0162, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:17:52,119 INFO [train.py:893] (0/4) Epoch 21, batch 2600, loss[loss=0.1697, simple_loss=0.2257, pruned_loss=0.05682, over 13519.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2354, pruned_loss=0.05762, over 2660465.48 frames. ], batch size: 70, lr: 6.16e-03, grad_scale: 16.0 2023-04-17 02:17:54,878 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56584.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 02:18:25,356 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.056e+02 2.625e+02 3.073e+02 3.881e+02 8.631e+02, threshold=6.146e+02, percent-clipped=4.0 2023-04-17 02:18:32,680 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-04-17 02:18:34,427 INFO [train.py:893] (0/4) Epoch 21, batch 2650, loss[loss=0.1637, simple_loss=0.2271, pruned_loss=0.05022, over 13375.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2365, pruned_loss=0.05816, over 2666530.77 frames. ], batch size: 62, lr: 6.16e-03, grad_scale: 16.0 2023-04-17 02:18:39,956 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.0371, 4.3363, 3.8337, 4.8391, 2.6149, 3.3553, 4.4592, 2.7811], device='cuda:0'), covar=tensor([0.0156, 0.0415, 0.0777, 0.0475, 0.1433, 0.1015, 0.0449, 0.1666], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0185, 0.0209, 0.0247, 0.0183, 0.0201, 0.0182, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:19:07,915 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1499, 4.4174, 4.1410, 4.1954, 4.3027, 4.5542, 4.4017, 4.2293], device='cuda:0'), covar=tensor([0.0290, 0.0285, 0.0346, 0.0851, 0.0290, 0.0244, 0.0309, 0.0289], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0159, 0.0177, 0.0266, 0.0176, 0.0194, 0.0173, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 02:19:12,997 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-21.pt 2023-04-17 02:19:37,772 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-17 02:19:41,991 INFO [train.py:893] (0/4) Epoch 22, batch 0, loss[loss=0.1839, simple_loss=0.2366, pruned_loss=0.06562, over 13384.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2366, pruned_loss=0.06562, over 13384.00 frames. ], batch size: 113, lr: 6.01e-03, grad_scale: 16.0 2023-04-17 02:19:41,992 INFO [train.py:918] (0/4) Computing validation loss 2023-04-17 02:20:04,056 INFO [train.py:927] (0/4) Epoch 22, validation: loss=0.136, simple_loss=0.1958, pruned_loss=0.03815, over 2446609.00 frames. 2023-04-17 02:20:04,057 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-17 02:20:24,525 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3229, 4.8366, 4.7568, 4.8458, 4.6673, 4.6877, 5.3040, 4.8400], device='cuda:0'), covar=tensor([0.0670, 0.1147, 0.1990, 0.2556, 0.0970, 0.1574, 0.0882, 0.1283], device='cuda:0'), in_proj_covar=tensor([0.0336, 0.0376, 0.0469, 0.0476, 0.0293, 0.0349, 0.0435, 0.0345], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 02:20:40,794 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.028e+02 2.714e+02 3.125e+02 3.595e+02 7.185e+02, threshold=6.251e+02, percent-clipped=1.0 2023-04-17 02:20:52,104 INFO [train.py:893] (0/4) Epoch 22, batch 50, loss[loss=0.177, simple_loss=0.2437, pruned_loss=0.05515, over 13480.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2309, pruned_loss=0.05681, over 605320.47 frames. ], batch size: 106, lr: 6.01e-03, grad_scale: 16.0 2023-04-17 02:21:14,861 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3346, 3.0098, 3.7321, 2.7736, 2.4615, 2.6624, 4.0076, 4.0619], device='cuda:0'), covar=tensor([0.1106, 0.1849, 0.0365, 0.1543, 0.1546, 0.1360, 0.0264, 0.0195], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0265, 0.0194, 0.0223, 0.0216, 0.0181, 0.0208, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 02:21:15,347 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-17 02:21:15,347 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-17 02:21:15,348 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-17 02:21:15,354 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-17 02:21:15,365 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-17 02:21:15,388 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-17 02:21:15,407 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-17 02:21:21,662 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-17 02:21:37,884 INFO [train.py:893] (0/4) Epoch 22, batch 100, loss[loss=0.1775, simple_loss=0.2416, pruned_loss=0.05671, over 13517.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2323, pruned_loss=0.05832, over 1057311.90 frames. ], batch size: 91, lr: 6.00e-03, grad_scale: 16.0 2023-04-17 02:21:58,571 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56801.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:22:00,295 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56803.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:22:15,153 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.959e+02 2.858e+02 3.303e+02 3.890e+02 1.630e+03, threshold=6.606e+02, percent-clipped=4.0 2023-04-17 02:22:26,757 INFO [train.py:893] (0/4) Epoch 22, batch 150, loss[loss=0.1847, simple_loss=0.2468, pruned_loss=0.06129, over 13388.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.236, pruned_loss=0.0603, over 1407682.68 frames. ], batch size: 109, lr: 6.00e-03, grad_scale: 16.0 2023-04-17 02:22:40,573 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9127, 3.6345, 4.5022, 3.4001, 3.1647, 3.0920, 4.7645, 4.8302], device='cuda:0'), covar=tensor([0.1080, 0.1668, 0.0310, 0.1409, 0.1414, 0.1422, 0.0211, 0.0186], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0264, 0.0193, 0.0222, 0.0214, 0.0180, 0.0207, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 02:22:42,780 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=56849.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:22:45,749 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.80 vs. limit=5.0 2023-04-17 02:22:58,002 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56864.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:23:09,053 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4063, 3.2648, 3.9317, 2.8589, 2.5572, 2.7087, 4.2090, 4.2765], device='cuda:0'), covar=tensor([0.1191, 0.1700, 0.0384, 0.1656, 0.1570, 0.1616, 0.0261, 0.0245], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0265, 0.0193, 0.0223, 0.0214, 0.0181, 0.0208, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 02:23:10,595 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56879.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 02:23:12,705 INFO [train.py:893] (0/4) Epoch 22, batch 200, loss[loss=0.1785, simple_loss=0.241, pruned_loss=0.05796, over 13472.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2383, pruned_loss=0.06085, over 1682235.45 frames. ], batch size: 79, lr: 6.00e-03, grad_scale: 16.0 2023-04-17 02:23:38,597 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56908.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:23:42,903 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.35 vs. limit=5.0 2023-04-17 02:23:47,846 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.2282, 2.1338, 3.9725, 3.8277, 3.8592, 3.2106, 3.5885, 2.9247], device='cuda:0'), covar=tensor([0.1940, 0.1471, 0.0116, 0.0191, 0.0197, 0.0638, 0.0292, 0.0850], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0184, 0.0124, 0.0128, 0.0133, 0.0173, 0.0145, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 02:23:49,168 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.056e+02 2.667e+02 3.208e+02 3.785e+02 6.289e+02, threshold=6.417e+02, percent-clipped=0.0 2023-04-17 02:24:00,159 INFO [train.py:893] (0/4) Epoch 22, batch 250, loss[loss=0.1845, simple_loss=0.2431, pruned_loss=0.06296, over 13496.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2388, pruned_loss=0.06137, over 1892243.17 frames. ], batch size: 93, lr: 6.00e-03, grad_scale: 16.0 2023-04-17 02:24:34,467 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56969.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:24:44,801 INFO [train.py:893] (0/4) Epoch 22, batch 300, loss[loss=0.1862, simple_loss=0.2495, pruned_loss=0.06148, over 13521.00 frames. ], tot_loss[loss=0.1809, simple_loss=0.2393, pruned_loss=0.06121, over 2059219.46 frames. ], batch size: 91, lr: 5.99e-03, grad_scale: 16.0 2023-04-17 02:24:56,537 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56993.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:24:58,237 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56995.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:25:13,408 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7469, 2.3904, 2.4484, 2.7888, 2.0683, 2.7690, 2.7149, 2.2256], device='cuda:0'), covar=tensor([0.0072, 0.0182, 0.0147, 0.0135, 0.0228, 0.0116, 0.0166, 0.0173], device='cuda:0'), in_proj_covar=tensor([0.0107, 0.0117, 0.0124, 0.0122, 0.0133, 0.0110, 0.0108, 0.0107], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 02:25:22,194 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.865e+02 2.638e+02 3.070e+02 3.794e+02 6.300e+02, threshold=6.141e+02, percent-clipped=0.0 2023-04-17 02:25:31,433 INFO [train.py:893] (0/4) Epoch 22, batch 350, loss[loss=0.1926, simple_loss=0.2523, pruned_loss=0.06641, over 13426.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2399, pruned_loss=0.06169, over 2195967.24 frames. ], batch size: 95, lr: 5.99e-03, grad_scale: 16.0 2023-04-17 02:25:53,938 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57054.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:25:55,657 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57056.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:26:18,064 INFO [train.py:893] (0/4) Epoch 22, batch 400, loss[loss=0.1947, simple_loss=0.257, pruned_loss=0.0662, over 13235.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2403, pruned_loss=0.0614, over 2300446.18 frames. ], batch size: 132, lr: 5.99e-03, grad_scale: 16.0 2023-04-17 02:26:31,079 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.4454, 4.9470, 4.8349, 4.9407, 4.7169, 4.8470, 5.4165, 5.0137], device='cuda:0'), covar=tensor([0.0684, 0.1164, 0.1971, 0.2481, 0.1000, 0.1370, 0.0870, 0.1196], device='cuda:0'), in_proj_covar=tensor([0.0341, 0.0385, 0.0476, 0.0483, 0.0300, 0.0354, 0.0442, 0.0353], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 02:26:39,502 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-17 02:26:56,969 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.094e+02 2.625e+02 3.015e+02 3.405e+02 5.837e+02, threshold=6.030e+02, percent-clipped=0.0 2023-04-17 02:26:57,660 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.53 vs. limit=5.0 2023-04-17 02:27:06,202 INFO [train.py:893] (0/4) Epoch 22, batch 450, loss[loss=0.1807, simple_loss=0.2425, pruned_loss=0.05945, over 13574.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2402, pruned_loss=0.0615, over 2380476.65 frames. ], batch size: 89, lr: 5.99e-03, grad_scale: 16.0 2023-04-17 02:27:33,697 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57159.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:27:34,349 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-17 02:27:54,364 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57179.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 02:27:55,697 INFO [train.py:893] (0/4) Epoch 22, batch 500, loss[loss=0.1853, simple_loss=0.253, pruned_loss=0.05876, over 13516.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2407, pruned_loss=0.06128, over 2443405.08 frames. ], batch size: 93, lr: 5.98e-03, grad_scale: 32.0 2023-04-17 02:28:00,267 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57186.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:28:08,753 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8349, 4.6570, 4.8593, 4.7557, 5.0919, 4.6571, 5.1200, 5.0629], device='cuda:0'), covar=tensor([0.0355, 0.0497, 0.0554, 0.0483, 0.0484, 0.0715, 0.0390, 0.0457], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0294, 0.0300, 0.0223, 0.0427, 0.0337, 0.0272, 0.0295], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:28:10,446 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57197.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:28:32,505 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.660e+02 2.678e+02 3.189e+02 3.924e+02 6.085e+02, threshold=6.377e+02, percent-clipped=1.0 2023-04-17 02:28:39,550 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57227.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 02:28:41,916 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3117, 2.0596, 2.4858, 3.6027, 3.2666, 3.6558, 2.8052, 2.3089], device='cuda:0'), covar=tensor([0.0250, 0.0951, 0.0834, 0.0077, 0.0291, 0.0082, 0.0683, 0.0927], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0151, 0.0169, 0.0101, 0.0121, 0.0098, 0.0170, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:28:42,452 INFO [train.py:893] (0/4) Epoch 22, batch 550, loss[loss=0.189, simple_loss=0.2461, pruned_loss=0.06591, over 13255.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2401, pruned_loss=0.06065, over 2494391.91 frames. ], batch size: 124, lr: 5.98e-03, grad_scale: 32.0 2023-04-17 02:28:56,657 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57244.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:28:59,144 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57247.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:28:59,986 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1856, 2.7734, 2.7523, 3.2148, 2.4808, 3.1756, 3.1879, 2.6327], device='cuda:0'), covar=tensor([0.0081, 0.0196, 0.0148, 0.0154, 0.0230, 0.0116, 0.0168, 0.0170], device='cuda:0'), in_proj_covar=tensor([0.0105, 0.0116, 0.0123, 0.0122, 0.0132, 0.0109, 0.0107, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 02:29:09,605 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57258.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 02:29:14,622 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57264.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:29:30,568 INFO [train.py:893] (0/4) Epoch 22, batch 600, loss[loss=0.1599, simple_loss=0.2229, pruned_loss=0.04846, over 13442.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2381, pruned_loss=0.05955, over 2531195.46 frames. ], batch size: 106, lr: 5.98e-03, grad_scale: 32.0 2023-04-17 02:29:54,922 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57305.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:30:01,034 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-17 02:30:07,436 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.837e+02 2.577e+02 3.083e+02 3.651e+02 6.901e+02, threshold=6.166e+02, percent-clipped=2.0 2023-04-17 02:30:18,843 INFO [train.py:893] (0/4) Epoch 22, batch 650, loss[loss=0.1778, simple_loss=0.2419, pruned_loss=0.05686, over 13524.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2377, pruned_loss=0.05947, over 2560133.55 frames. ], batch size: 85, lr: 5.98e-03, grad_scale: 32.0 2023-04-17 02:30:35,142 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57349.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:30:36,876 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57351.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:31:05,296 INFO [train.py:893] (0/4) Epoch 22, batch 700, loss[loss=0.2044, simple_loss=0.2475, pruned_loss=0.08062, over 12060.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2377, pruned_loss=0.0594, over 2585576.01 frames. ], batch size: 157, lr: 5.97e-03, grad_scale: 32.0 2023-04-17 02:31:22,380 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57398.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:31:43,049 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.065e+02 2.787e+02 3.172e+02 3.577e+02 6.833e+02, threshold=6.345e+02, percent-clipped=1.0 2023-04-17 02:31:46,679 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1932, 4.6381, 4.3997, 4.3781, 4.4948, 4.2494, 4.6822, 4.7376], device='cuda:0'), covar=tensor([0.0296, 0.0211, 0.0255, 0.0311, 0.0276, 0.0266, 0.0281, 0.0212], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0195, 0.0159, 0.0176, 0.0148, 0.0192, 0.0133, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 02:31:53,818 INFO [train.py:893] (0/4) Epoch 22, batch 750, loss[loss=0.1895, simple_loss=0.2395, pruned_loss=0.06975, over 13365.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2373, pruned_loss=0.05939, over 2602731.82 frames. ], batch size: 73, lr: 5.97e-03, grad_scale: 32.0 2023-04-17 02:32:21,041 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57459.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:32:21,111 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57459.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:32:23,668 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57462.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:32:40,665 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.0751, 4.4820, 4.4504, 4.5389, 4.4502, 4.3630, 5.0395, 4.6034], device='cuda:0'), covar=tensor([0.0746, 0.1283, 0.2259, 0.2735, 0.0975, 0.1550, 0.0933, 0.1214], device='cuda:0'), in_proj_covar=tensor([0.0336, 0.0381, 0.0476, 0.0479, 0.0297, 0.0351, 0.0437, 0.0350], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 02:32:40,790 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8797, 1.8548, 3.6787, 3.5447, 3.5291, 2.8910, 3.2767, 2.7199], device='cuda:0'), covar=tensor([0.1976, 0.1412, 0.0138, 0.0168, 0.0204, 0.0633, 0.0309, 0.0804], device='cuda:0'), in_proj_covar=tensor([0.0217, 0.0186, 0.0124, 0.0130, 0.0134, 0.0174, 0.0147, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 02:32:41,327 INFO [train.py:893] (0/4) Epoch 22, batch 800, loss[loss=0.1753, simple_loss=0.2436, pruned_loss=0.05352, over 13357.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2386, pruned_loss=0.05987, over 2611031.88 frames. ], batch size: 109, lr: 5.97e-03, grad_scale: 32.0 2023-04-17 02:32:54,737 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57495.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:33:05,893 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57507.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:33:13,528 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0548, 3.9224, 3.9942, 2.5066, 4.2955, 4.0813, 4.1004, 4.3052], device='cuda:0'), covar=tensor([0.0231, 0.0141, 0.0147, 0.1083, 0.0132, 0.0220, 0.0119, 0.0098], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0056, 0.0084, 0.0103, 0.0099, 0.0109, 0.0080, 0.0076], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:33:18,174 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.840e+02 2.750e+02 3.203e+02 3.790e+02 7.291e+02, threshold=6.405e+02, percent-clipped=0.0 2023-04-17 02:33:21,121 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57523.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:33:22,769 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1695, 2.0041, 2.3989, 3.5110, 3.2693, 3.5488, 2.7283, 2.3251], device='cuda:0'), covar=tensor([0.0314, 0.1172, 0.0888, 0.0088, 0.0260, 0.0097, 0.0704, 0.0946], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0152, 0.0169, 0.0101, 0.0122, 0.0098, 0.0171, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:33:28,397 INFO [train.py:893] (0/4) Epoch 22, batch 850, loss[loss=0.1895, simple_loss=0.2453, pruned_loss=0.0668, over 13520.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.24, pruned_loss=0.06051, over 2625831.71 frames. ], batch size: 91, lr: 5.97e-03, grad_scale: 32.0 2023-04-17 02:33:38,790 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57542.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:33:49,704 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57553.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 02:33:52,232 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57556.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:33:55,019 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-17 02:33:59,910 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57564.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:34:06,632 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8235, 2.5143, 1.9961, 3.8009, 4.1641, 3.0917, 4.1063, 3.9777], device='cuda:0'), covar=tensor([0.0106, 0.0905, 0.0999, 0.0089, 0.0069, 0.0517, 0.0076, 0.0077], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0090, 0.0097, 0.0081, 0.0068, 0.0081, 0.0056, 0.0077], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 02:34:15,655 INFO [train.py:893] (0/4) Epoch 22, batch 900, loss[loss=0.1813, simple_loss=0.225, pruned_loss=0.06876, over 13409.00 frames. ], tot_loss[loss=0.1809, simple_loss=0.2397, pruned_loss=0.06108, over 2630985.34 frames. ], batch size: 62, lr: 5.96e-03, grad_scale: 32.0 2023-04-17 02:34:33,470 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57600.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:34:45,144 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57612.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:34:48,268 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-17 02:34:52,402 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.880e+02 2.743e+02 3.191e+02 3.873e+02 7.153e+02, threshold=6.382e+02, percent-clipped=2.0 2023-04-17 02:35:02,822 INFO [train.py:893] (0/4) Epoch 22, batch 950, loss[loss=0.1672, simple_loss=0.2267, pruned_loss=0.0539, over 13280.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2381, pruned_loss=0.06029, over 2640181.66 frames. ], batch size: 124, lr: 5.96e-03, grad_scale: 32.0 2023-04-17 02:35:14,269 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.49 vs. limit=5.0 2023-04-17 02:35:20,439 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57649.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:35:22,056 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57651.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:35:50,159 INFO [train.py:893] (0/4) Epoch 22, batch 1000, loss[loss=0.1664, simple_loss=0.2236, pruned_loss=0.05456, over 13548.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2366, pruned_loss=0.06, over 2645259.36 frames. ], batch size: 76, lr: 5.96e-03, grad_scale: 32.0 2023-04-17 02:36:04,754 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57697.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:36:07,228 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57699.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:36:19,581 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9103, 3.9857, 2.7873, 3.5754, 3.8584, 2.6486, 3.4834, 2.7453], device='cuda:0'), covar=tensor([0.0274, 0.0272, 0.1011, 0.0375, 0.0305, 0.1153, 0.0561, 0.1250], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0174, 0.0176, 0.0216, 0.0138, 0.0158, 0.0161, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:36:25,802 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.668e+02 2.772e+02 3.188e+02 3.825e+02 8.358e+02, threshold=6.376e+02, percent-clipped=3.0 2023-04-17 02:36:28,737 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6269, 3.3287, 2.6983, 2.9597, 2.8371, 2.0360, 3.4619, 1.9523], device='cuda:0'), covar=tensor([0.0690, 0.0782, 0.0585, 0.0506, 0.0724, 0.1979, 0.1026, 0.1408], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0143, 0.0137, 0.0118, 0.0150, 0.0190, 0.0178, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 02:36:36,842 INFO [train.py:893] (0/4) Epoch 22, batch 1050, loss[loss=0.1909, simple_loss=0.2473, pruned_loss=0.06718, over 13520.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2354, pruned_loss=0.05915, over 2641263.27 frames. ], batch size: 83, lr: 5.96e-03, grad_scale: 32.0 2023-04-17 02:36:38,947 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57733.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:36:58,845 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57754.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:37:01,692 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.51 vs. limit=5.0 2023-04-17 02:37:18,297 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6107, 2.8279, 3.0157, 4.2311, 3.8159, 4.2909, 3.4056, 2.7486], device='cuda:0'), covar=tensor([0.0322, 0.0751, 0.0704, 0.0058, 0.0232, 0.0048, 0.0538, 0.0813], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0151, 0.0166, 0.0100, 0.0120, 0.0097, 0.0168, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:37:23,699 INFO [train.py:893] (0/4) Epoch 22, batch 1100, loss[loss=0.1779, simple_loss=0.2433, pruned_loss=0.05627, over 13460.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2354, pruned_loss=0.05835, over 2645715.83 frames. ], batch size: 103, lr: 5.95e-03, grad_scale: 32.0 2023-04-17 02:37:36,356 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57794.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:37:58,284 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57818.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:38:00,596 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.815e+02 2.427e+02 2.790e+02 3.332e+02 6.783e+02, threshold=5.579e+02, percent-clipped=1.0 2023-04-17 02:38:10,817 INFO [train.py:893] (0/4) Epoch 22, batch 1150, loss[loss=0.1591, simple_loss=0.2217, pruned_loss=0.04825, over 13530.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2352, pruned_loss=0.05767, over 2653911.22 frames. ], batch size: 72, lr: 5.95e-03, grad_scale: 32.0 2023-04-17 02:38:21,781 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57842.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:38:30,183 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57851.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:38:31,989 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57853.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 02:38:50,131 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3589, 1.8755, 4.0663, 3.8395, 3.9474, 3.1653, 3.6471, 2.9883], device='cuda:0'), covar=tensor([0.1914, 0.1851, 0.0112, 0.0249, 0.0211, 0.0664, 0.0291, 0.0778], device='cuda:0'), in_proj_covar=tensor([0.0218, 0.0187, 0.0125, 0.0130, 0.0135, 0.0174, 0.0147, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 02:38:52,725 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9437, 3.7183, 3.9105, 2.3081, 4.1292, 3.9777, 3.9825, 4.1877], device='cuda:0'), covar=tensor([0.0235, 0.0166, 0.0143, 0.1208, 0.0163, 0.0252, 0.0141, 0.0107], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0056, 0.0084, 0.0103, 0.0099, 0.0109, 0.0080, 0.0076], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:38:57,361 INFO [train.py:893] (0/4) Epoch 22, batch 1200, loss[loss=0.1768, simple_loss=0.2429, pruned_loss=0.0553, over 13328.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2367, pruned_loss=0.05807, over 2656058.26 frames. ], batch size: 118, lr: 5.95e-03, grad_scale: 32.0 2023-04-17 02:39:06,940 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57890.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:39:17,304 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57900.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:39:18,049 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57901.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:39:27,344 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-17 02:39:35,637 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.743e+02 2.547e+02 2.982e+02 3.467e+02 6.572e+02, threshold=5.963e+02, percent-clipped=2.0 2023-04-17 02:39:39,051 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-17 02:39:46,613 INFO [train.py:893] (0/4) Epoch 22, batch 1250, loss[loss=0.1781, simple_loss=0.2469, pruned_loss=0.05459, over 13515.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2376, pruned_loss=0.05857, over 2660322.65 frames. ], batch size: 91, lr: 5.94e-03, grad_scale: 32.0 2023-04-17 02:40:01,706 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57948.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:40:32,337 INFO [train.py:893] (0/4) Epoch 22, batch 1300, loss[loss=0.1909, simple_loss=0.2476, pruned_loss=0.06714, over 13541.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.238, pruned_loss=0.05821, over 2659243.64 frames. ], batch size: 76, lr: 5.94e-03, grad_scale: 32.0 2023-04-17 02:40:44,309 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.79 vs. limit=5.0 2023-04-17 02:40:50,919 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-58000.pt 2023-04-17 02:41:09,598 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-17 02:41:14,238 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.879e+02 2.746e+02 3.180e+02 3.759e+02 5.713e+02, threshold=6.361e+02, percent-clipped=0.0 2023-04-17 02:41:24,144 INFO [train.py:893] (0/4) Epoch 22, batch 1350, loss[loss=0.192, simple_loss=0.2538, pruned_loss=0.06511, over 13484.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2391, pruned_loss=0.05878, over 2653276.89 frames. ], batch size: 81, lr: 5.94e-03, grad_scale: 32.0 2023-04-17 02:41:47,121 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58054.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:42:03,846 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-17 02:42:09,777 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-04-17 02:42:12,625 INFO [train.py:893] (0/4) Epoch 22, batch 1400, loss[loss=0.1503, simple_loss=0.2082, pruned_loss=0.04616, over 13361.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2379, pruned_loss=0.05839, over 2655882.30 frames. ], batch size: 67, lr: 5.94e-03, grad_scale: 32.0 2023-04-17 02:42:12,882 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2819, 5.1527, 5.3301, 5.1260, 5.6555, 5.1463, 5.6640, 5.6222], device='cuda:0'), covar=tensor([0.0375, 0.0527, 0.0660, 0.0567, 0.0533, 0.0837, 0.0477, 0.0462], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0296, 0.0300, 0.0223, 0.0428, 0.0339, 0.0277, 0.0294], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:42:16,519 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-17 02:42:20,212 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58089.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:42:32,154 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58102.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:42:48,248 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58118.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:42:49,591 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.993e+02 2.607e+02 2.970e+02 3.475e+02 7.566e+02, threshold=5.940e+02, percent-clipped=1.0 2023-04-17 02:42:58,876 INFO [train.py:893] (0/4) Epoch 22, batch 1450, loss[loss=0.1909, simple_loss=0.251, pruned_loss=0.06545, over 13363.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2377, pruned_loss=0.05845, over 2660182.92 frames. ], batch size: 109, lr: 5.93e-03, grad_scale: 32.0 2023-04-17 02:43:10,120 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58141.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:43:12,740 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58144.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:43:19,039 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58151.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:43:32,643 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58166.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:43:47,414 INFO [train.py:893] (0/4) Epoch 22, batch 1500, loss[loss=0.1581, simple_loss=0.2198, pruned_loss=0.04815, over 13373.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2379, pruned_loss=0.05859, over 2659619.94 frames. ], batch size: 73, lr: 5.93e-03, grad_scale: 32.0 2023-04-17 02:44:04,462 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58199.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:44:08,253 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58202.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:44:10,972 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58205.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:44:12,281 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-17 02:44:23,979 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.863e+02 2.687e+02 3.148e+02 3.819e+02 8.002e+02, threshold=6.297e+02, percent-clipped=6.0 2023-04-17 02:44:27,269 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-17 02:44:34,890 INFO [train.py:893] (0/4) Epoch 22, batch 1550, loss[loss=0.1904, simple_loss=0.2473, pruned_loss=0.06678, over 13521.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2377, pruned_loss=0.05846, over 2661201.82 frames. ], batch size: 91, lr: 5.93e-03, grad_scale: 32.0 2023-04-17 02:45:12,914 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0588, 4.2133, 3.2185, 2.8358, 2.9117, 2.5507, 4.3428, 2.4774], device='cuda:0'), covar=tensor([0.1674, 0.0364, 0.1252, 0.2304, 0.0921, 0.3459, 0.0249, 0.4327], device='cuda:0'), in_proj_covar=tensor([0.0316, 0.0291, 0.0316, 0.0333, 0.0260, 0.0330, 0.0213, 0.0367], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 02:45:21,746 INFO [train.py:893] (0/4) Epoch 22, batch 1600, loss[loss=0.1487, simple_loss=0.2202, pruned_loss=0.03863, over 13344.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2375, pruned_loss=0.05811, over 2662269.92 frames. ], batch size: 73, lr: 5.93e-03, grad_scale: 32.0 2023-04-17 02:45:45,247 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-17 02:45:57,051 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.013e+02 2.646e+02 2.984e+02 3.538e+02 5.266e+02, threshold=5.968e+02, percent-clipped=0.0 2023-04-17 02:46:08,724 INFO [train.py:893] (0/4) Epoch 22, batch 1650, loss[loss=0.1973, simple_loss=0.2612, pruned_loss=0.06673, over 13534.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2386, pruned_loss=0.05828, over 2665021.85 frames. ], batch size: 72, lr: 5.92e-03, grad_scale: 32.0 2023-04-17 02:46:55,371 INFO [train.py:893] (0/4) Epoch 22, batch 1700, loss[loss=0.1725, simple_loss=0.2293, pruned_loss=0.05787, over 13369.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2382, pruned_loss=0.05797, over 2657207.25 frames. ], batch size: 67, lr: 5.92e-03, grad_scale: 32.0 2023-04-17 02:46:56,405 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1736, 4.6916, 4.6160, 4.6145, 4.4961, 4.5509, 5.1440, 4.7474], device='cuda:0'), covar=tensor([0.0709, 0.1329, 0.2076, 0.2671, 0.0831, 0.1619, 0.0893, 0.1209], device='cuda:0'), in_proj_covar=tensor([0.0330, 0.0380, 0.0468, 0.0476, 0.0292, 0.0350, 0.0434, 0.0346], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 02:47:02,313 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.66 vs. limit=5.0 2023-04-17 02:47:04,852 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58389.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:47:34,540 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.954e+02 2.578e+02 3.029e+02 3.680e+02 6.898e+02, threshold=6.059e+02, percent-clipped=3.0 2023-04-17 02:47:44,713 INFO [train.py:893] (0/4) Epoch 22, batch 1750, loss[loss=0.1432, simple_loss=0.2118, pruned_loss=0.03736, over 13466.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2371, pruned_loss=0.05687, over 2660409.01 frames. ], batch size: 79, lr: 5.92e-03, grad_scale: 32.0 2023-04-17 02:47:49,793 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58437.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:48:10,684 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6256, 3.1068, 2.5925, 2.7472, 2.7219, 2.0200, 3.2041, 2.0204], device='cuda:0'), covar=tensor([0.0622, 0.0641, 0.0442, 0.0446, 0.0619, 0.1786, 0.0824, 0.1073], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0143, 0.0137, 0.0118, 0.0149, 0.0190, 0.0177, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 02:48:33,553 INFO [train.py:893] (0/4) Epoch 22, batch 1800, loss[loss=0.1775, simple_loss=0.2443, pruned_loss=0.05534, over 13466.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2361, pruned_loss=0.05634, over 2662990.57 frames. ], batch size: 100, lr: 5.92e-03, grad_scale: 32.0 2023-04-17 02:48:47,796 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58497.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:48:50,281 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58500.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:49:08,653 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.958e+02 2.574e+02 2.968e+02 3.492e+02 5.400e+02, threshold=5.936e+02, percent-clipped=0.0 2023-04-17 02:49:18,720 INFO [train.py:893] (0/4) Epoch 22, batch 1850, loss[loss=0.1547, simple_loss=0.2143, pruned_loss=0.04753, over 13350.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2357, pruned_loss=0.05635, over 2663091.32 frames. ], batch size: 67, lr: 5.91e-03, grad_scale: 32.0 2023-04-17 02:49:21,282 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-17 02:49:40,183 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2002, 2.7387, 2.1628, 4.1821, 4.6177, 3.4197, 4.5095, 4.3653], device='cuda:0'), covar=tensor([0.0101, 0.0841, 0.1008, 0.0088, 0.0069, 0.0442, 0.0082, 0.0082], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0090, 0.0098, 0.0081, 0.0069, 0.0082, 0.0057, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 02:49:58,480 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6908, 2.6722, 3.0537, 4.2403, 3.8067, 4.2388, 3.3069, 2.7749], device='cuda:0'), covar=tensor([0.0278, 0.0938, 0.0787, 0.0059, 0.0229, 0.0068, 0.0658, 0.0867], device='cuda:0'), in_proj_covar=tensor([0.0122, 0.0154, 0.0170, 0.0102, 0.0122, 0.0099, 0.0172, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:50:05,943 INFO [train.py:893] (0/4) Epoch 22, batch 1900, loss[loss=0.1596, simple_loss=0.2156, pruned_loss=0.05183, over 13421.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2354, pruned_loss=0.05644, over 2662326.76 frames. ], batch size: 65, lr: 5.91e-03, grad_scale: 32.0 2023-04-17 02:50:40,429 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-17 02:50:42,962 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.656e+02 2.493e+02 2.844e+02 3.457e+02 5.617e+02, threshold=5.688e+02, percent-clipped=0.0 2023-04-17 02:50:52,240 INFO [train.py:893] (0/4) Epoch 22, batch 1950, loss[loss=0.1853, simple_loss=0.2386, pruned_loss=0.06604, over 13545.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2347, pruned_loss=0.05644, over 2662623.61 frames. ], batch size: 70, lr: 5.91e-03, grad_scale: 32.0 2023-04-17 02:51:38,483 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3326, 3.7314, 3.4948, 4.0783, 2.1862, 2.9142, 3.8429, 2.1302], device='cuda:0'), covar=tensor([0.0177, 0.0409, 0.0816, 0.0446, 0.1712, 0.1061, 0.0461, 0.1891], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0185, 0.0208, 0.0247, 0.0184, 0.0200, 0.0180, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:51:40,611 INFO [train.py:893] (0/4) Epoch 22, batch 2000, loss[loss=0.1981, simple_loss=0.2622, pruned_loss=0.06698, over 13048.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2372, pruned_loss=0.05811, over 2654843.16 frames. ], batch size: 142, lr: 5.91e-03, grad_scale: 32.0 2023-04-17 02:51:40,875 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58681.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:51:46,249 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 02:51:53,461 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-17 02:52:17,355 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.815e+02 2.571e+02 3.060e+02 3.640e+02 6.569e+02, threshold=6.120e+02, percent-clipped=2.0 2023-04-17 02:52:28,166 INFO [train.py:893] (0/4) Epoch 22, batch 2050, loss[loss=0.1574, simple_loss=0.2215, pruned_loss=0.04661, over 13530.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2391, pruned_loss=0.05893, over 2661141.14 frames. ], batch size: 83, lr: 5.90e-03, grad_scale: 32.0 2023-04-17 02:52:39,455 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58742.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:53:06,236 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9037, 3.9246, 3.0793, 2.7897, 2.7341, 2.4753, 4.0758, 2.3408], device='cuda:0'), covar=tensor([0.1814, 0.0417, 0.1329, 0.2252, 0.0947, 0.3336, 0.0302, 0.3916], device='cuda:0'), in_proj_covar=tensor([0.0315, 0.0291, 0.0317, 0.0333, 0.0261, 0.0330, 0.0213, 0.0368], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 02:53:10,214 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58775.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:53:14,975 INFO [train.py:893] (0/4) Epoch 22, batch 2100, loss[loss=0.1917, simple_loss=0.2503, pruned_loss=0.06659, over 13586.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2383, pruned_loss=0.05835, over 2663418.09 frames. ], batch size: 89, lr: 5.90e-03, grad_scale: 32.0 2023-04-17 02:53:15,470 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-17 02:53:20,554 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-17 02:53:31,574 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58797.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:53:34,487 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58800.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:53:51,733 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.861e+02 2.599e+02 3.041e+02 3.850e+02 9.204e+02, threshold=6.081e+02, percent-clipped=2.0 2023-04-17 02:54:03,795 INFO [train.py:893] (0/4) Epoch 22, batch 2150, loss[loss=0.1669, simple_loss=0.2386, pruned_loss=0.04765, over 13348.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2371, pruned_loss=0.0572, over 2664723.80 frames. ], batch size: 67, lr: 5.90e-03, grad_scale: 32.0 2023-04-17 02:54:08,905 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58836.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:54:16,270 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58845.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:54:18,774 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58848.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:54:28,488 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3481, 4.5930, 4.3367, 4.3428, 4.4072, 4.7463, 4.5171, 4.4890], device='cuda:0'), covar=tensor([0.0307, 0.0261, 0.0323, 0.0839, 0.0280, 0.0218, 0.0316, 0.0236], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0160, 0.0181, 0.0269, 0.0178, 0.0194, 0.0176, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 02:54:49,088 INFO [train.py:893] (0/4) Epoch 22, batch 2200, loss[loss=0.1821, simple_loss=0.2458, pruned_loss=0.05922, over 13376.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2372, pruned_loss=0.05704, over 2667223.76 frames. ], batch size: 113, lr: 5.90e-03, grad_scale: 32.0 2023-04-17 02:55:13,831 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58906.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:55:23,011 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7494, 3.6399, 4.3847, 3.0041, 2.9198, 2.9618, 4.7053, 4.7136], device='cuda:0'), covar=tensor([0.1213, 0.1528, 0.0342, 0.1740, 0.1472, 0.1545, 0.0228, 0.0205], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0270, 0.0196, 0.0226, 0.0219, 0.0182, 0.0212, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 02:55:25,995 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.683e+02 2.473e+02 2.857e+02 3.422e+02 4.907e+02, threshold=5.714e+02, percent-clipped=0.0 2023-04-17 02:55:37,274 INFO [train.py:893] (0/4) Epoch 22, batch 2250, loss[loss=0.1637, simple_loss=0.2184, pruned_loss=0.0545, over 13411.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2357, pruned_loss=0.05676, over 2658368.97 frames. ], batch size: 62, lr: 5.89e-03, grad_scale: 32.0 2023-04-17 02:56:12,326 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58967.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:56:23,054 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5126, 2.6077, 2.1948, 4.3264, 4.8423, 3.5390, 4.6854, 4.5158], device='cuda:0'), covar=tensor([0.0075, 0.1002, 0.1083, 0.0095, 0.0065, 0.0498, 0.0082, 0.0082], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0092, 0.0098, 0.0082, 0.0070, 0.0082, 0.0057, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 02:56:24,445 INFO [train.py:893] (0/4) Epoch 22, batch 2300, loss[loss=0.1573, simple_loss=0.225, pruned_loss=0.0448, over 13543.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2352, pruned_loss=0.05629, over 2661393.46 frames. ], batch size: 76, lr: 5.89e-03, grad_scale: 32.0 2023-04-17 02:57:02,029 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.824e+02 2.500e+02 2.983e+02 3.435e+02 5.453e+02, threshold=5.967e+02, percent-clipped=0.0 2023-04-17 02:57:12,924 INFO [train.py:893] (0/4) Epoch 22, batch 2350, loss[loss=0.1811, simple_loss=0.2307, pruned_loss=0.06573, over 13431.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2343, pruned_loss=0.05592, over 2659246.74 frames. ], batch size: 65, lr: 5.89e-03, grad_scale: 32.0 2023-04-17 02:57:18,298 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59037.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:57:27,917 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-17 02:57:30,988 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-17 02:57:36,448 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 02:57:59,831 INFO [train.py:893] (0/4) Epoch 22, batch 2400, loss[loss=0.1503, simple_loss=0.2201, pruned_loss=0.04029, over 13528.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2332, pruned_loss=0.05557, over 2657140.48 frames. ], batch size: 76, lr: 5.89e-03, grad_scale: 32.0 2023-04-17 02:58:36,862 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.705e+02 2.612e+02 2.930e+02 3.503e+02 4.997e+02, threshold=5.859e+02, percent-clipped=0.0 2023-04-17 02:58:43,853 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0921, 4.1895, 2.9001, 3.8646, 4.0623, 2.5376, 3.7792, 2.7773], device='cuda:0'), covar=tensor([0.0289, 0.0197, 0.1021, 0.0363, 0.0226, 0.1204, 0.0410, 0.1186], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0177, 0.0177, 0.0219, 0.0140, 0.0159, 0.0160, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 02:58:46,967 INFO [train.py:893] (0/4) Epoch 22, batch 2450, loss[loss=0.1771, simple_loss=0.2369, pruned_loss=0.05864, over 13532.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2326, pruned_loss=0.05533, over 2655140.38 frames. ], batch size: 91, lr: 5.88e-03, grad_scale: 32.0 2023-04-17 02:58:47,155 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59131.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:59:05,651 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8951, 2.5863, 2.5260, 2.9486, 2.3334, 2.9716, 2.9407, 2.5020], device='cuda:0'), covar=tensor([0.0076, 0.0202, 0.0170, 0.0150, 0.0205, 0.0121, 0.0172, 0.0180], device='cuda:0'), in_proj_covar=tensor([0.0105, 0.0117, 0.0125, 0.0123, 0.0134, 0.0111, 0.0109, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 02:59:18,215 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59164.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 02:59:34,872 INFO [train.py:893] (0/4) Epoch 22, batch 2500, loss[loss=0.1685, simple_loss=0.234, pruned_loss=0.05148, over 13568.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2328, pruned_loss=0.05569, over 2651194.39 frames. ], batch size: 89, lr: 5.88e-03, grad_scale: 64.0 2023-04-17 03:00:12,739 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.906e+02 2.615e+02 2.930e+02 3.639e+02 8.531e+02, threshold=5.860e+02, percent-clipped=2.0 2023-04-17 03:00:16,490 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59225.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:00:19,931 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8294, 3.9187, 2.9953, 2.5966, 2.7160, 2.4388, 3.9918, 2.2395], device='cuda:0'), covar=tensor([0.1727, 0.0368, 0.1287, 0.2293, 0.0971, 0.3325, 0.0284, 0.4152], device='cuda:0'), in_proj_covar=tensor([0.0314, 0.0291, 0.0319, 0.0334, 0.0261, 0.0330, 0.0214, 0.0368], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 03:00:22,151 INFO [train.py:893] (0/4) Epoch 22, batch 2550, loss[loss=0.1804, simple_loss=0.2406, pruned_loss=0.0601, over 13129.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2328, pruned_loss=0.05598, over 2646043.06 frames. ], batch size: 142, lr: 5.88e-03, grad_scale: 16.0 2023-04-17 03:00:47,174 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 03:00:51,302 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59262.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:01:08,891 INFO [train.py:893] (0/4) Epoch 22, batch 2600, loss[loss=0.168, simple_loss=0.2277, pruned_loss=0.05422, over 13450.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2332, pruned_loss=0.05626, over 2651768.50 frames. ], batch size: 103, lr: 5.88e-03, grad_scale: 16.0 2023-04-17 03:01:44,368 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.931e+02 2.687e+02 3.147e+02 3.617e+02 7.653e+02, threshold=6.293e+02, percent-clipped=2.0 2023-04-17 03:01:51,717 INFO [train.py:893] (0/4) Epoch 22, batch 2650, loss[loss=0.2227, simple_loss=0.2669, pruned_loss=0.08919, over 13036.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2341, pruned_loss=0.0573, over 2647206.83 frames. ], batch size: 142, lr: 5.87e-03, grad_scale: 16.0 2023-04-17 03:01:56,426 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59337.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:02:13,384 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5631, 2.8670, 2.5191, 4.4835, 4.9983, 3.7012, 4.8815, 4.6885], device='cuda:0'), covar=tensor([0.0099, 0.0816, 0.0927, 0.0106, 0.0107, 0.0433, 0.0087, 0.0088], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0091, 0.0098, 0.0082, 0.0069, 0.0082, 0.0058, 0.0078], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 03:02:30,923 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-22.pt 2023-04-17 03:02:55,882 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-17 03:03:00,009 INFO [train.py:893] (0/4) Epoch 23, batch 0, loss[loss=0.1573, simple_loss=0.2081, pruned_loss=0.05326, over 12401.00 frames. ], tot_loss[loss=0.1573, simple_loss=0.2081, pruned_loss=0.05326, over 12401.00 frames. ], batch size: 51, lr: 5.74e-03, grad_scale: 16.0 2023-04-17 03:03:00,010 INFO [train.py:918] (0/4) Computing validation loss 2023-04-17 03:03:15,665 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8611, 4.0032, 3.9299, 3.9251, 3.9805, 3.8931, 4.0531, 4.1116], device='cuda:0'), covar=tensor([0.0211, 0.0279, 0.0216, 0.0293, 0.0266, 0.0274, 0.0290, 0.0260], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0203, 0.0169, 0.0186, 0.0155, 0.0202, 0.0140, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:03:18,394 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4400, 2.4158, 2.9614, 3.8149, 3.4718, 3.8386, 3.1498, 2.5864], device='cuda:0'), covar=tensor([0.0254, 0.0805, 0.0561, 0.0072, 0.0233, 0.0067, 0.0565, 0.0787], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0151, 0.0166, 0.0100, 0.0120, 0.0097, 0.0168, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:03:23,127 INFO [train.py:927] (0/4) Epoch 23, validation: loss=0.135, simple_loss=0.1947, pruned_loss=0.03761, over 2446609.00 frames. 2023-04-17 03:03:23,128 INFO [train.py:928] (0/4) Maximum memory allocated so far is 12905MB 2023-04-17 03:03:26,594 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59385.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:03:54,095 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-04-17 03:04:01,981 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.652e+02 2.605e+02 3.123e+02 3.576e+02 6.770e+02, threshold=6.246e+02, percent-clipped=1.0 2023-04-17 03:04:05,922 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9348, 1.9496, 3.5628, 3.4649, 3.4115, 2.7495, 3.2578, 2.6391], device='cuda:0'), covar=tensor([0.2098, 0.1511, 0.0175, 0.0181, 0.0309, 0.0759, 0.0269, 0.0839], device='cuda:0'), in_proj_covar=tensor([0.0220, 0.0188, 0.0127, 0.0130, 0.0138, 0.0177, 0.0146, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 03:04:10,609 INFO [train.py:893] (0/4) Epoch 23, batch 50, loss[loss=0.1889, simple_loss=0.243, pruned_loss=0.0674, over 13474.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2277, pruned_loss=0.05579, over 604425.07 frames. ], batch size: 106, lr: 5.74e-03, grad_scale: 16.0 2023-04-17 03:04:10,887 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59431.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:04:35,476 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-17 03:04:35,476 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-17 03:04:35,476 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-17 03:04:35,483 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-17 03:04:35,491 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-17 03:04:35,512 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-17 03:04:36,190 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-17 03:04:55,723 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59479.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:04:57,222 INFO [train.py:893] (0/4) Epoch 23, batch 100, loss[loss=0.1728, simple_loss=0.2369, pruned_loss=0.0544, over 13540.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.231, pruned_loss=0.05709, over 1053194.29 frames. ], batch size: 91, lr: 5.74e-03, grad_scale: 16.0 2023-04-17 03:05:12,395 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8693, 1.9461, 3.5836, 3.4419, 3.4042, 2.7451, 3.2441, 2.6328], device='cuda:0'), covar=tensor([0.2167, 0.1478, 0.0154, 0.0171, 0.0322, 0.0721, 0.0266, 0.0820], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0188, 0.0127, 0.0130, 0.0139, 0.0178, 0.0145, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 03:05:17,326 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59502.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:05:34,908 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59520.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:05:36,336 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.879e+02 2.548e+02 3.039e+02 3.718e+02 1.416e+03, threshold=6.077e+02, percent-clipped=2.0 2023-04-17 03:05:44,660 INFO [train.py:893] (0/4) Epoch 23, batch 150, loss[loss=0.1913, simple_loss=0.2523, pruned_loss=0.06513, over 13245.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.2346, pruned_loss=0.05898, over 1411956.84 frames. ], batch size: 124, lr: 5.73e-03, grad_scale: 16.0 2023-04-17 03:05:47,413 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.4799, 5.0401, 4.9155, 4.9844, 4.7007, 4.8887, 5.4683, 5.0968], device='cuda:0'), covar=tensor([0.0717, 0.1246, 0.2083, 0.2303, 0.0969, 0.1450, 0.0773, 0.1117], device='cuda:0'), in_proj_covar=tensor([0.0338, 0.0385, 0.0474, 0.0479, 0.0299, 0.0356, 0.0442, 0.0351], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 03:06:14,292 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59562.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:06:15,152 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59563.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:06:31,614 INFO [train.py:893] (0/4) Epoch 23, batch 200, loss[loss=0.1346, simple_loss=0.1742, pruned_loss=0.04753, over 7969.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2359, pruned_loss=0.05955, over 1676760.46 frames. ], batch size: 29, lr: 5.73e-03, grad_scale: 16.0 2023-04-17 03:06:59,078 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59610.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:07:11,435 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.051e+02 2.601e+02 3.074e+02 3.611e+02 6.828e+02, threshold=6.148e+02, percent-clipped=4.0 2023-04-17 03:07:18,929 INFO [train.py:893] (0/4) Epoch 23, batch 250, loss[loss=0.2083, simple_loss=0.2587, pruned_loss=0.07889, over 13567.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2375, pruned_loss=0.06014, over 1899016.19 frames. ], batch size: 89, lr: 5.73e-03, grad_scale: 16.0 2023-04-17 03:07:54,789 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0232, 3.3751, 3.3497, 3.6762, 2.1285, 2.8536, 3.6027, 2.1512], device='cuda:0'), covar=tensor([0.0184, 0.0536, 0.0754, 0.0600, 0.1731, 0.0983, 0.0553, 0.1828], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0186, 0.0208, 0.0248, 0.0183, 0.0199, 0.0183, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:08:06,538 INFO [train.py:893] (0/4) Epoch 23, batch 300, loss[loss=0.1711, simple_loss=0.2322, pruned_loss=0.05501, over 13546.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2374, pruned_loss=0.05989, over 2065799.29 frames. ], batch size: 83, lr: 5.73e-03, grad_scale: 16.0 2023-04-17 03:08:26,312 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9384, 3.7289, 3.1820, 3.4335, 3.0781, 2.2602, 3.8691, 2.1056], device='cuda:0'), covar=tensor([0.0744, 0.0670, 0.0503, 0.0402, 0.0782, 0.2140, 0.0888, 0.1502], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0147, 0.0138, 0.0119, 0.0151, 0.0193, 0.0180, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 03:08:45,247 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.995e+02 2.544e+02 2.931e+02 3.487e+02 6.573e+02, threshold=5.861e+02, percent-clipped=2.0 2023-04-17 03:08:52,901 INFO [train.py:893] (0/4) Epoch 23, batch 350, loss[loss=0.1808, simple_loss=0.2435, pruned_loss=0.05908, over 13525.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2376, pruned_loss=0.0601, over 2193232.29 frames. ], batch size: 91, lr: 5.72e-03, grad_scale: 16.0 2023-04-17 03:09:40,693 INFO [train.py:893] (0/4) Epoch 23, batch 400, loss[loss=0.1857, simple_loss=0.2534, pruned_loss=0.05899, over 13268.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2376, pruned_loss=0.05985, over 2296151.32 frames. ], batch size: 124, lr: 5.72e-03, grad_scale: 16.0 2023-04-17 03:10:14,521 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.65 vs. limit=5.0 2023-04-17 03:10:16,679 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59820.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:10:18,079 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.855e+02 2.576e+02 2.992e+02 3.488e+02 5.440e+02, threshold=5.985e+02, percent-clipped=0.0 2023-04-17 03:10:26,414 INFO [train.py:893] (0/4) Epoch 23, batch 450, loss[loss=0.1841, simple_loss=0.2334, pruned_loss=0.06745, over 13393.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2392, pruned_loss=0.06023, over 2376069.08 frames. ], batch size: 62, lr: 5.72e-03, grad_scale: 16.0 2023-04-17 03:10:43,380 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59848.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:10:52,195 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-17 03:10:52,364 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59858.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:11:01,367 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59868.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:11:07,617 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9343, 2.3719, 1.8848, 3.8207, 4.2563, 3.1900, 4.1927, 4.0094], device='cuda:0'), covar=tensor([0.0083, 0.0967, 0.1064, 0.0089, 0.0072, 0.0474, 0.0071, 0.0085], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0090, 0.0096, 0.0080, 0.0068, 0.0080, 0.0057, 0.0077], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 03:11:14,013 INFO [train.py:893] (0/4) Epoch 23, batch 500, loss[loss=0.1583, simple_loss=0.2245, pruned_loss=0.046, over 13535.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2386, pruned_loss=0.05962, over 2437168.06 frames. ], batch size: 91, lr: 5.72e-03, grad_scale: 16.0 2023-04-17 03:11:21,055 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59889.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:11:40,937 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59909.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:11:49,382 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-04-17 03:11:52,063 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.757e+02 2.557e+02 2.976e+02 3.530e+02 5.641e+02, threshold=5.953e+02, percent-clipped=0.0 2023-04-17 03:11:54,953 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7137, 3.5649, 4.2050, 2.9118, 2.6137, 2.8181, 4.5092, 4.6490], device='cuda:0'), covar=tensor([0.1096, 0.1594, 0.0324, 0.1724, 0.1719, 0.1440, 0.0245, 0.0173], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0266, 0.0194, 0.0223, 0.0217, 0.0179, 0.0210, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-04-17 03:12:00,580 INFO [train.py:893] (0/4) Epoch 23, batch 550, loss[loss=0.1956, simple_loss=0.2496, pruned_loss=0.07082, over 11749.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2384, pruned_loss=0.05915, over 2487599.80 frames. ], batch size: 157, lr: 5.71e-03, grad_scale: 16.0 2023-04-17 03:12:18,413 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59950.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:12:39,426 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-17 03:12:47,121 INFO [train.py:893] (0/4) Epoch 23, batch 600, loss[loss=0.174, simple_loss=0.2364, pruned_loss=0.05574, over 13441.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2366, pruned_loss=0.05828, over 2527619.16 frames. ], batch size: 95, lr: 5.71e-03, grad_scale: 16.0 2023-04-17 03:13:06,229 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-60000.pt 2023-04-17 03:13:13,589 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9505, 2.9556, 2.6376, 2.0197, 2.0948, 2.5785, 2.6406, 3.2265], device='cuda:0'), covar=tensor([0.1016, 0.0343, 0.0584, 0.1506, 0.0492, 0.0483, 0.0744, 0.0317], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0146, 0.0123, 0.0210, 0.0114, 0.0165, 0.0174, 0.0131], device='cuda:0'), out_proj_covar=tensor([1.2521e-04, 1.0929e-04, 9.6127e-05, 1.5626e-04, 8.2841e-05, 1.2494e-04, 1.3162e-04, 9.7000e-05], device='cuda:0') 2023-04-17 03:13:29,648 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.782e+02 2.485e+02 2.931e+02 3.430e+02 4.559e+02, threshold=5.862e+02, percent-clipped=0.0 2023-04-17 03:13:38,792 INFO [train.py:893] (0/4) Epoch 23, batch 650, loss[loss=0.1861, simple_loss=0.2456, pruned_loss=0.0633, over 13494.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2361, pruned_loss=0.05785, over 2559171.32 frames. ], batch size: 93, lr: 5.71e-03, grad_scale: 16.0 2023-04-17 03:13:46,468 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60040.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:13:56,605 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5788, 3.4528, 2.7016, 3.0309, 2.9125, 2.0930, 3.5900, 1.9298], device='cuda:0'), covar=tensor([0.0760, 0.0677, 0.0624, 0.0520, 0.0725, 0.2167, 0.1105, 0.1473], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0146, 0.0137, 0.0118, 0.0150, 0.0192, 0.0179, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 03:13:58,209 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60052.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:14:02,575 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2023-04-17 03:14:10,423 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60065.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:14:21,071 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60078.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:14:22,767 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1282, 4.2791, 3.3674, 2.9968, 3.1408, 2.6594, 4.4895, 2.5908], device='cuda:0'), covar=tensor([0.1669, 0.0343, 0.1136, 0.2123, 0.0834, 0.3321, 0.0239, 0.3891], device='cuda:0'), in_proj_covar=tensor([0.0317, 0.0293, 0.0320, 0.0337, 0.0263, 0.0331, 0.0218, 0.0369], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 03:14:24,064 INFO [train.py:893] (0/4) Epoch 23, batch 700, loss[loss=0.1697, simple_loss=0.2293, pruned_loss=0.05506, over 13538.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2348, pruned_loss=0.05732, over 2573671.56 frames. ], batch size: 87, lr: 5.71e-03, grad_scale: 16.0 2023-04-17 03:14:43,025 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60101.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 03:14:53,750 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60113.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 03:15:01,791 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.826e+02 2.678e+02 3.015e+02 3.584e+02 9.072e+02, threshold=6.030e+02, percent-clipped=5.0 2023-04-17 03:15:06,919 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60126.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:15:10,902 INFO [train.py:893] (0/4) Epoch 23, batch 750, loss[loss=0.1744, simple_loss=0.2389, pruned_loss=0.05491, over 13516.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2353, pruned_loss=0.05803, over 2594490.08 frames. ], batch size: 85, lr: 5.71e-03, grad_scale: 16.0 2023-04-17 03:15:17,024 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6835, 2.8160, 2.5327, 1.8779, 1.8250, 2.4605, 2.5237, 3.0496], device='cuda:0'), covar=tensor([0.1171, 0.0290, 0.0501, 0.1438, 0.0261, 0.0432, 0.0675, 0.0210], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0146, 0.0125, 0.0214, 0.0116, 0.0169, 0.0176, 0.0134], device='cuda:0'), out_proj_covar=tensor([1.2726e-04, 1.0999e-04, 9.7489e-05, 1.5903e-04, 8.4682e-05, 1.2748e-04, 1.3331e-04, 9.9018e-05], device='cuda:0') 2023-04-17 03:15:17,944 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60139.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:15:36,549 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60158.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:15:57,191 INFO [train.py:893] (0/4) Epoch 23, batch 800, loss[loss=0.1811, simple_loss=0.2356, pruned_loss=0.06327, over 12058.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.237, pruned_loss=0.0587, over 2608497.50 frames. ], batch size: 157, lr: 5.70e-03, grad_scale: 16.0 2023-04-17 03:16:13,187 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60198.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:16:19,195 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60204.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:16:20,889 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60206.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:16:36,054 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.678e+02 2.576e+02 3.130e+02 3.550e+02 5.821e+02, threshold=6.259e+02, percent-clipped=0.0 2023-04-17 03:16:43,541 INFO [train.py:893] (0/4) Epoch 23, batch 850, loss[loss=0.1698, simple_loss=0.2371, pruned_loss=0.05129, over 13403.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2381, pruned_loss=0.05853, over 2622269.69 frames. ], batch size: 118, lr: 5.70e-03, grad_scale: 16.0 2023-04-17 03:16:57,107 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60245.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:17:10,283 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60259.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:17:29,953 INFO [train.py:893] (0/4) Epoch 23, batch 900, loss[loss=0.1703, simple_loss=0.2299, pruned_loss=0.05534, over 13217.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.238, pruned_loss=0.0588, over 2629682.91 frames. ], batch size: 132, lr: 5.70e-03, grad_scale: 16.0 2023-04-17 03:18:02,176 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-17 03:18:07,922 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.882e+02 2.736e+02 3.142e+02 3.652e+02 7.263e+02, threshold=6.284e+02, percent-clipped=3.0 2023-04-17 03:18:15,955 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-17 03:18:16,151 INFO [train.py:893] (0/4) Epoch 23, batch 950, loss[loss=0.1541, simple_loss=0.216, pruned_loss=0.04611, over 13422.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2371, pruned_loss=0.05929, over 2635010.75 frames. ], batch size: 65, lr: 5.70e-03, grad_scale: 16.0 2023-04-17 03:19:02,848 INFO [train.py:893] (0/4) Epoch 23, batch 1000, loss[loss=0.1706, simple_loss=0.233, pruned_loss=0.05412, over 13537.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2349, pruned_loss=0.05836, over 2643212.54 frames. ], batch size: 83, lr: 5.69e-03, grad_scale: 16.0 2023-04-17 03:19:16,276 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60396.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 03:19:27,024 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60408.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 03:19:39,632 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60421.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:19:40,247 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.841e+02 2.528e+02 2.847e+02 3.554e+02 6.904e+02, threshold=5.694e+02, percent-clipped=4.0 2023-04-17 03:19:48,609 INFO [train.py:893] (0/4) Epoch 23, batch 1050, loss[loss=0.156, simple_loss=0.2226, pruned_loss=0.04466, over 13485.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2335, pruned_loss=0.05742, over 2649160.22 frames. ], batch size: 81, lr: 5.69e-03, grad_scale: 16.0 2023-04-17 03:19:51,911 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60434.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:20:35,611 INFO [train.py:893] (0/4) Epoch 23, batch 1100, loss[loss=0.1735, simple_loss=0.2458, pruned_loss=0.05064, over 13519.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2345, pruned_loss=0.05725, over 2649880.33 frames. ], batch size: 91, lr: 5.69e-03, grad_scale: 16.0 2023-04-17 03:20:47,529 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4215, 3.2480, 3.9812, 2.7144, 2.6605, 2.7586, 4.2664, 4.3721], device='cuda:0'), covar=tensor([0.1261, 0.1886, 0.0400, 0.1920, 0.1629, 0.1585, 0.0263, 0.0243], device='cuda:0'), in_proj_covar=tensor([0.0248, 0.0270, 0.0197, 0.0225, 0.0220, 0.0182, 0.0214, 0.0184], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 03:20:52,267 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8289, 4.1511, 3.9901, 3.9457, 4.1108, 3.8090, 4.2139, 4.3118], device='cuda:0'), covar=tensor([0.0347, 0.0387, 0.0310, 0.0467, 0.0311, 0.0408, 0.0379, 0.0328], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0197, 0.0162, 0.0179, 0.0151, 0.0196, 0.0133, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 03:20:57,196 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60504.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:21:12,732 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.950e+02 2.578e+02 3.034e+02 3.780e+02 7.669e+02, threshold=6.069e+02, percent-clipped=4.0 2023-04-17 03:21:21,265 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7019, 2.6963, 2.4504, 1.6454, 1.6313, 2.3428, 2.3110, 2.9389], device='cuda:0'), covar=tensor([0.1100, 0.0355, 0.0654, 0.1626, 0.0215, 0.0416, 0.0732, 0.0228], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0147, 0.0125, 0.0215, 0.0117, 0.0169, 0.0177, 0.0133], device='cuda:0'), out_proj_covar=tensor([1.2752e-04, 1.1060e-04, 9.7529e-05, 1.5991e-04, 8.5111e-05, 1.2776e-04, 1.3406e-04, 9.8179e-05], device='cuda:0') 2023-04-17 03:21:21,697 INFO [train.py:893] (0/4) Epoch 23, batch 1150, loss[loss=0.2006, simple_loss=0.2589, pruned_loss=0.07121, over 13451.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2344, pruned_loss=0.05693, over 2655960.23 frames. ], batch size: 103, lr: 5.69e-03, grad_scale: 16.0 2023-04-17 03:21:35,655 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60545.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:21:41,490 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60552.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:21:43,922 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60554.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:22:08,294 INFO [train.py:893] (0/4) Epoch 23, batch 1200, loss[loss=0.1939, simple_loss=0.251, pruned_loss=0.06839, over 13492.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2354, pruned_loss=0.05722, over 2656355.71 frames. ], batch size: 93, lr: 5.68e-03, grad_scale: 16.0 2023-04-17 03:22:20,125 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60593.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:22:30,340 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1667, 4.4490, 4.1650, 4.2677, 4.2723, 4.6057, 4.3882, 4.2758], device='cuda:0'), covar=tensor([0.0320, 0.0308, 0.0353, 0.0819, 0.0320, 0.0230, 0.0327, 0.0297], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0158, 0.0180, 0.0266, 0.0177, 0.0192, 0.0173, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 03:22:37,672 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-17 03:22:47,543 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.088e+02 2.864e+02 3.217e+02 3.727e+02 5.611e+02, threshold=6.435e+02, percent-clipped=0.0 2023-04-17 03:22:50,059 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-17 03:22:55,875 INFO [train.py:893] (0/4) Epoch 23, batch 1250, loss[loss=0.1765, simple_loss=0.2412, pruned_loss=0.05591, over 13543.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2349, pruned_loss=0.05691, over 2654598.74 frames. ], batch size: 83, lr: 5.68e-03, grad_scale: 16.0 2023-04-17 03:23:16,368 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3873, 4.6341, 4.3542, 4.4224, 4.4282, 4.7955, 4.5442, 4.4791], device='cuda:0'), covar=tensor([0.0301, 0.0307, 0.0358, 0.0874, 0.0320, 0.0247, 0.0358, 0.0254], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0157, 0.0179, 0.0265, 0.0176, 0.0191, 0.0173, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 03:23:43,265 INFO [train.py:893] (0/4) Epoch 23, batch 1300, loss[loss=0.1852, simple_loss=0.247, pruned_loss=0.06173, over 13446.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2357, pruned_loss=0.05721, over 2654815.93 frames. ], batch size: 106, lr: 5.68e-03, grad_scale: 16.0 2023-04-17 03:23:46,091 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6490, 4.6492, 3.3994, 4.3937, 4.6074, 3.1869, 4.1075, 3.3603], device='cuda:0'), covar=tensor([0.0244, 0.0213, 0.0896, 0.0485, 0.0132, 0.1048, 0.0379, 0.1187], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0181, 0.0179, 0.0223, 0.0142, 0.0163, 0.0164, 0.0162], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:23:57,580 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60696.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:24:08,204 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60708.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:24:16,667 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8779, 2.6444, 2.3104, 1.6165, 1.5183, 2.2216, 2.3431, 2.8882], device='cuda:0'), covar=tensor([0.0960, 0.0435, 0.0612, 0.1714, 0.0244, 0.0603, 0.0821, 0.0231], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0148, 0.0125, 0.0216, 0.0117, 0.0170, 0.0178, 0.0133], device='cuda:0'), out_proj_covar=tensor([1.2815e-04, 1.1123e-04, 9.7956e-05, 1.6028e-04, 8.5030e-05, 1.2861e-04, 1.3444e-04, 9.8336e-05], device='cuda:0') 2023-04-17 03:24:17,417 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60718.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:24:20,763 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60721.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:24:21,334 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.809e+02 2.793e+02 3.175e+02 3.621e+02 6.568e+02, threshold=6.350e+02, percent-clipped=2.0 2023-04-17 03:24:29,570 INFO [train.py:893] (0/4) Epoch 23, batch 1350, loss[loss=0.173, simple_loss=0.2376, pruned_loss=0.05414, over 13460.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2367, pruned_loss=0.05754, over 2651875.52 frames. ], batch size: 106, lr: 5.68e-03, grad_scale: 16.0 2023-04-17 03:24:32,357 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60734.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:24:42,226 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60744.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:24:53,035 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60756.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:25:04,566 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60769.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:25:14,292 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60779.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:25:16,398 INFO [train.py:893] (0/4) Epoch 23, batch 1400, loss[loss=0.1986, simple_loss=0.2632, pruned_loss=0.06706, over 13434.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2368, pruned_loss=0.05784, over 2651571.75 frames. ], batch size: 95, lr: 5.67e-03, grad_scale: 16.0 2023-04-17 03:25:17,325 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60782.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:25:52,986 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.093e+02 2.602e+02 2.908e+02 3.577e+02 9.144e+02, threshold=5.816e+02, percent-clipped=1.0 2023-04-17 03:26:01,604 INFO [train.py:893] (0/4) Epoch 23, batch 1450, loss[loss=0.1802, simple_loss=0.2321, pruned_loss=0.06415, over 11706.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.2364, pruned_loss=0.05804, over 2652359.10 frames. ], batch size: 157, lr: 5.67e-03, grad_scale: 16.0 2023-04-17 03:26:24,293 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60854.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:26:49,316 INFO [train.py:893] (0/4) Epoch 23, batch 1500, loss[loss=0.1781, simple_loss=0.2417, pruned_loss=0.05727, over 13529.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2354, pruned_loss=0.05719, over 2647485.82 frames. ], batch size: 87, lr: 5.67e-03, grad_scale: 16.0 2023-04-17 03:26:52,141 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9782, 2.8734, 2.5196, 1.7657, 1.8487, 2.4083, 2.4791, 3.0944], device='cuda:0'), covar=tensor([0.0937, 0.0332, 0.0636, 0.1605, 0.0304, 0.0525, 0.0782, 0.0214], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0145, 0.0123, 0.0211, 0.0115, 0.0166, 0.0174, 0.0130], device='cuda:0'), out_proj_covar=tensor([1.2513e-04, 1.0862e-04, 9.6394e-05, 1.5652e-04, 8.3055e-05, 1.2521e-04, 1.3151e-04, 9.6239e-05], device='cuda:0') 2023-04-17 03:27:05,416 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9980, 1.9222, 3.9695, 3.8760, 3.8097, 3.0816, 3.6178, 2.9681], device='cuda:0'), covar=tensor([0.1959, 0.1589, 0.0132, 0.0154, 0.0194, 0.0679, 0.0258, 0.0693], device='cuda:0'), in_proj_covar=tensor([0.0211, 0.0182, 0.0124, 0.0128, 0.0133, 0.0173, 0.0142, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 03:27:09,258 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60902.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:27:27,731 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.986e+02 2.779e+02 3.136e+02 3.771e+02 6.785e+02, threshold=6.272e+02, percent-clipped=2.0 2023-04-17 03:27:36,803 INFO [train.py:893] (0/4) Epoch 23, batch 1550, loss[loss=0.154, simple_loss=0.2128, pruned_loss=0.04754, over 13561.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2357, pruned_loss=0.05723, over 2650606.99 frames. ], batch size: 78, lr: 5.67e-03, grad_scale: 16.0 2023-04-17 03:28:23,235 INFO [train.py:893] (0/4) Epoch 23, batch 1600, loss[loss=0.1751, simple_loss=0.2352, pruned_loss=0.05748, over 13543.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2363, pruned_loss=0.05716, over 2645159.75 frames. ], batch size: 85, lr: 5.67e-03, grad_scale: 16.0 2023-04-17 03:29:01,733 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.731e+02 2.549e+02 3.025e+02 3.497e+02 7.515e+02, threshold=6.050e+02, percent-clipped=2.0 2023-04-17 03:29:10,278 INFO [train.py:893] (0/4) Epoch 23, batch 1650, loss[loss=0.1842, simple_loss=0.2513, pruned_loss=0.05854, over 13440.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2366, pruned_loss=0.05677, over 2645802.97 frames. ], batch size: 103, lr: 5.66e-03, grad_scale: 16.0 2023-04-17 03:29:35,051 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-17 03:29:39,001 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61061.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:29:48,969 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2878, 4.7784, 4.5505, 4.5880, 4.5616, 4.4067, 4.8508, 4.8844], device='cuda:0'), covar=tensor([0.0278, 0.0229, 0.0227, 0.0391, 0.0322, 0.0259, 0.0240, 0.0243], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0203, 0.0168, 0.0185, 0.0156, 0.0202, 0.0136, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:29:51,333 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=61074.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:29:53,902 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.2947, 2.0955, 4.1093, 3.9173, 3.9092, 3.2271, 3.7481, 2.9318], device='cuda:0'), covar=tensor([0.1994, 0.1568, 0.0110, 0.0285, 0.0208, 0.0656, 0.0215, 0.0854], device='cuda:0'), in_proj_covar=tensor([0.0213, 0.0183, 0.0125, 0.0129, 0.0134, 0.0175, 0.0142, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 03:29:56,942 INFO [train.py:893] (0/4) Epoch 23, batch 1700, loss[loss=0.1631, simple_loss=0.2254, pruned_loss=0.05036, over 13369.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2369, pruned_loss=0.05665, over 2648257.64 frames. ], batch size: 73, lr: 5.66e-03, grad_scale: 16.0 2023-04-17 03:30:21,595 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-17 03:30:35,663 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.757e+02 2.631e+02 2.974e+02 3.481e+02 8.450e+02, threshold=5.948e+02, percent-clipped=1.0 2023-04-17 03:30:36,020 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61122.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:30:43,999 INFO [train.py:893] (0/4) Epoch 23, batch 1750, loss[loss=0.1803, simple_loss=0.2363, pruned_loss=0.06213, over 13538.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2359, pruned_loss=0.05625, over 2650521.73 frames. ], batch size: 72, lr: 5.66e-03, grad_scale: 16.0 2023-04-17 03:30:45,979 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.4245, 2.0469, 2.1619, 2.4684, 1.8376, 2.5060, 2.4405, 1.9686], device='cuda:0'), covar=tensor([0.0086, 0.0232, 0.0158, 0.0128, 0.0236, 0.0128, 0.0166, 0.0197], device='cuda:0'), in_proj_covar=tensor([0.0107, 0.0120, 0.0126, 0.0127, 0.0136, 0.0113, 0.0110, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 03:31:31,096 INFO [train.py:893] (0/4) Epoch 23, batch 1800, loss[loss=0.1564, simple_loss=0.2227, pruned_loss=0.04505, over 13484.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2349, pruned_loss=0.056, over 2653383.61 frames. ], batch size: 81, lr: 5.66e-03, grad_scale: 16.0 2023-04-17 03:31:33,957 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2882, 3.0937, 3.7984, 2.6762, 2.5095, 2.6164, 4.0393, 4.1829], device='cuda:0'), covar=tensor([0.1223, 0.1872, 0.0392, 0.1929, 0.1700, 0.1661, 0.0310, 0.0208], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0273, 0.0198, 0.0226, 0.0221, 0.0184, 0.0214, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 03:31:51,927 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.57 vs. limit=5.0 2023-04-17 03:32:09,373 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.629e+02 2.492e+02 2.908e+02 3.443e+02 5.707e+02, threshold=5.815e+02, percent-clipped=0.0 2023-04-17 03:32:17,477 INFO [train.py:893] (0/4) Epoch 23, batch 1850, loss[loss=0.1519, simple_loss=0.2082, pruned_loss=0.04779, over 13331.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2347, pruned_loss=0.05593, over 2658120.03 frames. ], batch size: 67, lr: 5.65e-03, grad_scale: 32.0 2023-04-17 03:32:19,218 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-17 03:33:03,920 INFO [train.py:893] (0/4) Epoch 23, batch 1900, loss[loss=0.1578, simple_loss=0.2208, pruned_loss=0.04739, over 13529.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2347, pruned_loss=0.05602, over 2662152.37 frames. ], batch size: 72, lr: 5.65e-03, grad_scale: 32.0 2023-04-17 03:33:05,342 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-17 03:33:41,138 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.059e+02 2.633e+02 3.066e+02 3.485e+02 7.604e+02, threshold=6.132e+02, percent-clipped=3.0 2023-04-17 03:33:49,773 INFO [train.py:893] (0/4) Epoch 23, batch 1950, loss[loss=0.1428, simple_loss=0.2022, pruned_loss=0.04177, over 13241.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.233, pruned_loss=0.05511, over 2665232.91 frames. ], batch size: 58, lr: 5.65e-03, grad_scale: 32.0 2023-04-17 03:33:51,797 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8308, 2.6114, 2.5023, 3.0055, 2.3150, 3.0233, 2.7829, 2.4460], device='cuda:0'), covar=tensor([0.0128, 0.0181, 0.0189, 0.0154, 0.0219, 0.0129, 0.0214, 0.0205], device='cuda:0'), in_proj_covar=tensor([0.0107, 0.0120, 0.0125, 0.0125, 0.0134, 0.0112, 0.0109, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 03:34:29,768 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=61374.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:34:35,484 INFO [train.py:893] (0/4) Epoch 23, batch 2000, loss[loss=0.1774, simple_loss=0.2436, pruned_loss=0.05556, over 13499.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2336, pruned_loss=0.05561, over 2655216.60 frames. ], batch size: 81, lr: 5.65e-03, grad_scale: 32.0 2023-04-17 03:34:37,685 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-17 03:34:40,485 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 03:34:53,298 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-17 03:35:01,601 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-17 03:35:08,666 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=61417.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:35:09,135 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2023-04-17 03:35:13,267 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.858e+02 2.773e+02 3.121e+02 3.784e+02 7.293e+02, threshold=6.243e+02, percent-clipped=1.0 2023-04-17 03:35:13,506 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=61422.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:35:22,152 INFO [train.py:893] (0/4) Epoch 23, batch 2050, loss[loss=0.1584, simple_loss=0.2177, pruned_loss=0.04962, over 13431.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2357, pruned_loss=0.0567, over 2658860.62 frames. ], batch size: 65, lr: 5.64e-03, grad_scale: 32.0 2023-04-17 03:35:56,504 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-17 03:36:06,620 INFO [train.py:893] (0/4) Epoch 23, batch 2100, loss[loss=0.1447, simple_loss=0.2108, pruned_loss=0.03926, over 13537.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2357, pruned_loss=0.05691, over 2656463.19 frames. ], batch size: 76, lr: 5.64e-03, grad_scale: 32.0 2023-04-17 03:36:45,708 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.854e+02 2.633e+02 3.023e+02 3.535e+02 5.995e+02, threshold=6.045e+02, percent-clipped=0.0 2023-04-17 03:36:53,713 INFO [train.py:893] (0/4) Epoch 23, batch 2150, loss[loss=0.1664, simple_loss=0.2277, pruned_loss=0.05254, over 13048.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2353, pruned_loss=0.05654, over 2650319.52 frames. ], batch size: 142, lr: 5.64e-03, grad_scale: 32.0 2023-04-17 03:37:39,564 INFO [train.py:893] (0/4) Epoch 23, batch 2200, loss[loss=0.1705, simple_loss=0.2349, pruned_loss=0.05307, over 13439.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2343, pruned_loss=0.05546, over 2653528.31 frames. ], batch size: 106, lr: 5.64e-03, grad_scale: 32.0 2023-04-17 03:38:19,059 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.670e+02 2.488e+02 2.952e+02 3.468e+02 6.307e+02, threshold=5.905e+02, percent-clipped=2.0 2023-04-17 03:38:26,948 INFO [train.py:893] (0/4) Epoch 23, batch 2250, loss[loss=0.1902, simple_loss=0.2511, pruned_loss=0.06468, over 13050.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2329, pruned_loss=0.05486, over 2656031.57 frames. ], batch size: 142, lr: 5.64e-03, grad_scale: 32.0 2023-04-17 03:38:30,539 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6854, 2.4904, 2.9752, 4.0873, 3.6534, 4.1523, 3.3068, 2.7256], device='cuda:0'), covar=tensor([0.0238, 0.0987, 0.0760, 0.0061, 0.0233, 0.0051, 0.0648, 0.0865], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0150, 0.0168, 0.0100, 0.0120, 0.0097, 0.0168, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:38:58,741 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-17 03:39:13,934 INFO [train.py:893] (0/4) Epoch 23, batch 2300, loss[loss=0.1781, simple_loss=0.2382, pruned_loss=0.05899, over 13533.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2331, pruned_loss=0.05493, over 2661995.61 frames. ], batch size: 85, lr: 5.63e-03, grad_scale: 32.0 2023-04-17 03:39:21,584 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-17 03:39:47,480 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=61717.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:39:51,393 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.815e+02 2.435e+02 2.861e+02 3.476e+02 7.164e+02, threshold=5.722e+02, percent-clipped=2.0 2023-04-17 03:39:58,987 INFO [train.py:893] (0/4) Epoch 23, batch 2350, loss[loss=0.1865, simple_loss=0.2498, pruned_loss=0.06161, over 13211.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2329, pruned_loss=0.05475, over 2664953.98 frames. ], batch size: 132, lr: 5.63e-03, grad_scale: 32.0 2023-04-17 03:40:00,149 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61732.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:40:11,617 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61743.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:40:22,171 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 03:40:30,529 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=61765.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:40:46,470 INFO [train.py:893] (0/4) Epoch 23, batch 2400, loss[loss=0.1675, simple_loss=0.234, pruned_loss=0.05054, over 13526.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2316, pruned_loss=0.05435, over 2662305.67 frames. ], batch size: 98, lr: 5.63e-03, grad_scale: 32.0 2023-04-17 03:40:50,762 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4564, 2.3576, 2.7544, 3.8586, 3.4363, 3.9412, 3.1156, 2.4394], device='cuda:0'), covar=tensor([0.0295, 0.0888, 0.0775, 0.0076, 0.0285, 0.0068, 0.0640, 0.0933], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0149, 0.0168, 0.0101, 0.0121, 0.0097, 0.0168, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:40:56,458 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61793.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:41:06,513 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61804.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:41:16,539 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6717, 3.8680, 3.7625, 3.7598, 3.8052, 3.6825, 3.9232, 4.0176], device='cuda:0'), covar=tensor([0.0242, 0.0340, 0.0288, 0.0364, 0.0300, 0.0313, 0.0303, 0.0295], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0204, 0.0168, 0.0188, 0.0157, 0.0206, 0.0139, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:41:22,686 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.841e+02 2.567e+02 2.922e+02 3.454e+02 5.739e+02, threshold=5.845e+02, percent-clipped=1.0 2023-04-17 03:41:30,301 INFO [train.py:893] (0/4) Epoch 23, batch 2450, loss[loss=0.184, simple_loss=0.2429, pruned_loss=0.06256, over 13529.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2317, pruned_loss=0.05422, over 2659714.61 frames. ], batch size: 83, lr: 5.63e-03, grad_scale: 32.0 2023-04-17 03:41:37,650 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9990, 4.7946, 5.0104, 4.8967, 5.2751, 4.8659, 5.2930, 5.2500], device='cuda:0'), covar=tensor([0.0404, 0.0556, 0.0679, 0.0528, 0.0552, 0.0803, 0.0423, 0.0472], device='cuda:0'), in_proj_covar=tensor([0.0295, 0.0301, 0.0307, 0.0231, 0.0443, 0.0349, 0.0284, 0.0305], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:42:15,703 INFO [train.py:893] (0/4) Epoch 23, batch 2500, loss[loss=0.1644, simple_loss=0.2303, pruned_loss=0.04928, over 13386.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2312, pruned_loss=0.05399, over 2657723.59 frames. ], batch size: 62, lr: 5.62e-03, grad_scale: 32.0 2023-04-17 03:42:16,110 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0132, 4.2792, 3.2422, 2.9648, 3.2047, 2.6141, 4.4035, 2.4919], device='cuda:0'), covar=tensor([0.1807, 0.0364, 0.1292, 0.2109, 0.0837, 0.3561, 0.0276, 0.3981], device='cuda:0'), in_proj_covar=tensor([0.0316, 0.0292, 0.0319, 0.0339, 0.0264, 0.0333, 0.0216, 0.0370], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 03:42:36,926 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3605, 2.5990, 2.0308, 4.2587, 4.7504, 3.4136, 4.6329, 4.4153], device='cuda:0'), covar=tensor([0.0097, 0.0886, 0.1121, 0.0090, 0.0060, 0.0478, 0.0078, 0.0086], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0092, 0.0100, 0.0083, 0.0070, 0.0083, 0.0058, 0.0080], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 03:42:45,111 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61914.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:42:51,234 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.719e+02 2.535e+02 2.894e+02 3.374e+02 4.864e+02, threshold=5.788e+02, percent-clipped=0.0 2023-04-17 03:42:59,187 INFO [train.py:893] (0/4) Epoch 23, batch 2550, loss[loss=0.1798, simple_loss=0.2435, pruned_loss=0.05805, over 13456.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2308, pruned_loss=0.05385, over 2653329.33 frames. ], batch size: 103, lr: 5.62e-03, grad_scale: 32.0 2023-04-17 03:43:18,423 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-17 03:43:23,758 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 03:43:27,202 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7522, 3.8542, 2.7279, 3.5310, 3.8638, 2.5837, 3.4962, 2.7968], device='cuda:0'), covar=tensor([0.0331, 0.0240, 0.1067, 0.0342, 0.0244, 0.1176, 0.0544, 0.1142], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0180, 0.0180, 0.0222, 0.0140, 0.0161, 0.0162, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:43:33,649 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2238, 2.3573, 2.0749, 4.0899, 4.5875, 3.3584, 4.4854, 4.2337], device='cuda:0'), covar=tensor([0.0090, 0.1014, 0.1072, 0.0088, 0.0076, 0.0458, 0.0083, 0.0086], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0092, 0.0100, 0.0082, 0.0069, 0.0083, 0.0058, 0.0080], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 03:43:34,521 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6839, 3.3722, 2.8099, 2.8989, 2.9163, 2.0590, 3.4439, 1.9474], device='cuda:0'), covar=tensor([0.0678, 0.0640, 0.0494, 0.0552, 0.0668, 0.2018, 0.1134, 0.1337], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0145, 0.0136, 0.0118, 0.0148, 0.0190, 0.0180, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 03:43:40,252 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61975.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:43:44,877 INFO [train.py:893] (0/4) Epoch 23, batch 2600, loss[loss=0.1764, simple_loss=0.2363, pruned_loss=0.05825, over 13524.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2306, pruned_loss=0.05416, over 2641707.36 frames. ], batch size: 70, lr: 5.62e-03, grad_scale: 32.0 2023-04-17 03:43:57,417 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0969, 2.5964, 2.0400, 3.9831, 4.4275, 3.2983, 4.3326, 4.1144], device='cuda:0'), covar=tensor([0.0094, 0.0840, 0.1016, 0.0090, 0.0079, 0.0417, 0.0088, 0.0091], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0092, 0.0100, 0.0082, 0.0070, 0.0083, 0.0058, 0.0080], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 03:44:02,302 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-62000.pt 2023-04-17 03:44:23,302 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.883e+02 2.551e+02 3.036e+02 3.753e+02 6.121e+02, threshold=6.072e+02, percent-clipped=2.0 2023-04-17 03:44:29,950 INFO [train.py:893] (0/4) Epoch 23, batch 2650, loss[loss=0.1611, simple_loss=0.2284, pruned_loss=0.04694, over 13516.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2319, pruned_loss=0.05497, over 2643321.93 frames. ], batch size: 98, lr: 5.62e-03, grad_scale: 32.0 2023-04-17 03:45:07,151 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-23.pt 2023-04-17 03:45:32,129 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-17 03:45:35,579 INFO [train.py:893] (0/4) Epoch 24, batch 0, loss[loss=0.1471, simple_loss=0.2025, pruned_loss=0.04589, over 12761.00 frames. ], tot_loss[loss=0.1471, simple_loss=0.2025, pruned_loss=0.04589, over 12761.00 frames. ], batch size: 52, lr: 5.49e-03, grad_scale: 32.0 2023-04-17 03:45:35,580 INFO [train.py:918] (0/4) Computing validation loss 2023-04-17 03:45:42,567 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0164, 4.7265, 3.7239, 3.3422, 3.9411, 3.1891, 4.8427, 2.7736], device='cuda:0'), covar=tensor([0.1007, 0.0298, 0.0992, 0.1872, 0.0595, 0.2874, 0.0228, 0.3947], device='cuda:0'), in_proj_covar=tensor([0.0314, 0.0289, 0.0316, 0.0337, 0.0262, 0.0331, 0.0214, 0.0367], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 03:45:58,006 INFO [train.py:927] (0/4) Epoch 24, validation: loss=0.1345, simple_loss=0.1942, pruned_loss=0.03737, over 2446609.00 frames. 2023-04-17 03:45:58,007 INFO [train.py:928] (0/4) Maximum memory allocated so far is 13044MB 2023-04-17 03:46:03,990 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9541, 4.0127, 3.9958, 2.4474, 4.3394, 4.1891, 4.1574, 4.3855], device='cuda:0'), covar=tensor([0.0316, 0.0162, 0.0172, 0.1388, 0.0212, 0.0276, 0.0167, 0.0124], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0057, 0.0085, 0.0103, 0.0100, 0.0111, 0.0083, 0.0078], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:46:05,562 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62088.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:46:14,736 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62099.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:46:19,644 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9044, 3.9271, 2.7334, 3.6127, 3.8935, 2.5162, 3.5211, 2.6066], device='cuda:0'), covar=tensor([0.0335, 0.0291, 0.1143, 0.0426, 0.0255, 0.1293, 0.0547, 0.1407], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0181, 0.0180, 0.0223, 0.0140, 0.0162, 0.0163, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:46:24,971 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-04-17 03:46:36,056 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62122.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:46:36,524 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.935e+02 2.622e+02 3.024e+02 3.638e+02 5.971e+02, threshold=6.048e+02, percent-clipped=0.0 2023-04-17 03:46:42,609 INFO [train.py:893] (0/4) Epoch 24, batch 50, loss[loss=0.1699, simple_loss=0.22, pruned_loss=0.05988, over 12856.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2282, pruned_loss=0.05486, over 601117.18 frames. ], batch size: 52, lr: 5.49e-03, grad_scale: 16.0 2023-04-17 03:47:06,731 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-17 03:47:06,732 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-17 03:47:06,732 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-17 03:47:06,739 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-17 03:47:06,747 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-17 03:47:06,768 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-17 03:47:07,525 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-17 03:47:29,860 INFO [train.py:893] (0/4) Epoch 24, batch 100, loss[loss=0.1797, simple_loss=0.2355, pruned_loss=0.06189, over 13083.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2311, pruned_loss=0.05603, over 1061197.18 frames. ], batch size: 142, lr: 5.49e-03, grad_scale: 16.0 2023-04-17 03:47:32,641 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62183.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:47:35,882 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0447, 4.0401, 2.8284, 3.7615, 3.9725, 2.6005, 3.6285, 2.7338], device='cuda:0'), covar=tensor([0.0271, 0.0221, 0.1021, 0.0472, 0.0252, 0.1252, 0.0455, 0.1277], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0181, 0.0180, 0.0223, 0.0141, 0.0162, 0.0163, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:48:08,097 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 2.677e+02 3.136e+02 3.979e+02 7.437e+02, threshold=6.271e+02, percent-clipped=3.0 2023-04-17 03:48:14,819 INFO [train.py:893] (0/4) Epoch 24, batch 150, loss[loss=0.1889, simple_loss=0.2267, pruned_loss=0.07558, over 7831.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2345, pruned_loss=0.05835, over 1410394.64 frames. ], batch size: 31, lr: 5.49e-03, grad_scale: 16.0 2023-04-17 03:48:19,387 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8233, 4.0694, 3.0562, 2.7554, 2.9035, 2.5061, 4.1483, 2.3348], device='cuda:0'), covar=tensor([0.1983, 0.0428, 0.1376, 0.2210, 0.0995, 0.3670, 0.0303, 0.4357], device='cuda:0'), in_proj_covar=tensor([0.0315, 0.0291, 0.0319, 0.0338, 0.0263, 0.0333, 0.0216, 0.0370], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 03:48:51,388 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62270.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:48:54,802 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62273.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:49:02,023 INFO [train.py:893] (0/4) Epoch 24, batch 200, loss[loss=0.187, simple_loss=0.2495, pruned_loss=0.06228, over 13499.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2358, pruned_loss=0.05879, over 1678715.97 frames. ], batch size: 93, lr: 5.49e-03, grad_scale: 16.0 2023-04-17 03:49:29,247 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7707, 3.5637, 3.7259, 2.1935, 3.9193, 3.8583, 3.8151, 3.9750], device='cuda:0'), covar=tensor([0.0239, 0.0197, 0.0152, 0.1252, 0.0171, 0.0246, 0.0136, 0.0120], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0058, 0.0086, 0.0104, 0.0101, 0.0112, 0.0083, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:49:32,549 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.7799, 4.7937, 4.6896, 3.6425, 4.9683, 4.8689, 4.8725, 4.9572], device='cuda:0'), covar=tensor([0.0233, 0.0127, 0.0149, 0.0847, 0.0171, 0.0263, 0.0131, 0.0137], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0058, 0.0086, 0.0104, 0.0101, 0.0112, 0.0083, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:49:33,510 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4153, 4.2800, 4.3492, 2.7457, 4.6684, 4.4883, 4.4785, 4.6734], device='cuda:0'), covar=tensor([0.0234, 0.0125, 0.0152, 0.1135, 0.0153, 0.0253, 0.0142, 0.0098], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0058, 0.0086, 0.0103, 0.0101, 0.0112, 0.0083, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:49:39,395 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5558, 4.7563, 4.5992, 4.5760, 4.5931, 4.9420, 4.6929, 4.7040], device='cuda:0'), covar=tensor([0.0279, 0.0262, 0.0248, 0.0817, 0.0296, 0.0251, 0.0248, 0.0194], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0158, 0.0179, 0.0266, 0.0179, 0.0195, 0.0176, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 03:49:40,837 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.804e+02 2.766e+02 3.047e+02 3.449e+02 5.084e+02, threshold=6.094e+02, percent-clipped=0.0 2023-04-17 03:49:44,850 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-17 03:49:47,383 INFO [train.py:893] (0/4) Epoch 24, batch 250, loss[loss=0.1545, simple_loss=0.2217, pruned_loss=0.0437, over 13514.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2367, pruned_loss=0.05951, over 1893712.70 frames. ], batch size: 76, lr: 5.48e-03, grad_scale: 16.0 2023-04-17 03:49:50,882 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62334.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:50:04,841 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-17 03:50:32,463 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7269, 3.6989, 2.8721, 3.2805, 3.0075, 2.2433, 3.7073, 2.0243], device='cuda:0'), covar=tensor([0.0798, 0.0617, 0.0574, 0.0451, 0.0738, 0.2005, 0.1049, 0.1462], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0147, 0.0136, 0.0119, 0.0150, 0.0192, 0.0182, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 03:50:33,910 INFO [train.py:893] (0/4) Epoch 24, batch 300, loss[loss=0.1772, simple_loss=0.2293, pruned_loss=0.06254, over 11719.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2377, pruned_loss=0.05931, over 2067043.93 frames. ], batch size: 157, lr: 5.48e-03, grad_scale: 16.0 2023-04-17 03:50:41,569 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62388.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:50:44,037 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62391.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:50:52,480 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62399.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:51:14,251 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.637e+02 2.996e+02 3.547e+02 6.141e+02, threshold=5.992e+02, percent-clipped=2.0 2023-04-17 03:51:21,371 INFO [train.py:893] (0/4) Epoch 24, batch 350, loss[loss=0.1577, simple_loss=0.2099, pruned_loss=0.05278, over 13426.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2385, pruned_loss=0.05996, over 2190399.74 frames. ], batch size: 65, lr: 5.48e-03, grad_scale: 16.0 2023-04-17 03:51:27,229 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62436.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:51:36,336 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62447.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:51:41,259 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62452.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:52:05,105 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62478.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:52:07,200 INFO [train.py:893] (0/4) Epoch 24, batch 400, loss[loss=0.187, simple_loss=0.2517, pruned_loss=0.06114, over 13360.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2377, pruned_loss=0.05902, over 2298301.90 frames. ], batch size: 118, lr: 5.48e-03, grad_scale: 16.0 2023-04-17 03:52:13,522 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.73 vs. limit=5.0 2023-04-17 03:52:39,668 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1925, 4.6669, 4.6326, 4.6761, 4.4756, 4.5363, 5.1518, 4.7001], device='cuda:0'), covar=tensor([0.0759, 0.1260, 0.2213, 0.2540, 0.0983, 0.1764, 0.0852, 0.1227], device='cuda:0'), in_proj_covar=tensor([0.0341, 0.0393, 0.0481, 0.0484, 0.0307, 0.0362, 0.0448, 0.0360], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 03:52:39,924 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-17 03:52:46,750 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.724e+02 2.666e+02 3.083e+02 3.648e+02 7.210e+02, threshold=6.167e+02, percent-clipped=2.0 2023-04-17 03:52:54,082 INFO [train.py:893] (0/4) Epoch 24, batch 450, loss[loss=0.174, simple_loss=0.2366, pruned_loss=0.05575, over 13522.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.239, pruned_loss=0.05969, over 2378860.66 frames. ], batch size: 98, lr: 5.48e-03, grad_scale: 16.0 2023-04-17 03:52:57,038 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.12 vs. limit=5.0 2023-04-17 03:53:07,492 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62545.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:53:18,456 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-17 03:53:30,056 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62570.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:53:38,742 INFO [train.py:893] (0/4) Epoch 24, batch 500, loss[loss=0.17, simple_loss=0.2337, pruned_loss=0.05316, over 13209.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2391, pruned_loss=0.05915, over 2436837.11 frames. ], batch size: 132, lr: 5.47e-03, grad_scale: 16.0 2023-04-17 03:53:43,741 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62585.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:54:03,835 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62606.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:54:05,849 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-17 03:54:09,979 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-17 03:54:14,538 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62618.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:54:19,282 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.673e+02 2.688e+02 3.153e+02 3.603e+02 8.252e+02, threshold=6.305e+02, percent-clipped=3.0 2023-04-17 03:54:25,247 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62629.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:54:25,853 INFO [train.py:893] (0/4) Epoch 24, batch 550, loss[loss=0.1764, simple_loss=0.2388, pruned_loss=0.05698, over 13364.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2385, pruned_loss=0.05842, over 2488362.61 frames. ], batch size: 118, lr: 5.47e-03, grad_scale: 16.0 2023-04-17 03:54:36,954 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6474, 2.6856, 2.9217, 4.2289, 3.8584, 4.3179, 3.4510, 2.6755], device='cuda:0'), covar=tensor([0.0317, 0.0873, 0.0866, 0.0065, 0.0200, 0.0053, 0.0536, 0.0890], device='cuda:0'), in_proj_covar=tensor([0.0122, 0.0148, 0.0168, 0.0102, 0.0121, 0.0098, 0.0169, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 03:54:40,300 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62646.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:54:59,632 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62666.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:55:11,628 INFO [train.py:893] (0/4) Epoch 24, batch 600, loss[loss=0.1773, simple_loss=0.2331, pruned_loss=0.06075, over 13172.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2369, pruned_loss=0.05772, over 2530723.72 frames. ], batch size: 58, lr: 5.47e-03, grad_scale: 16.0 2023-04-17 03:55:13,892 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.74 vs. limit=5.0 2023-04-17 03:55:14,611 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-17 03:55:23,003 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-17 03:55:51,672 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.553e+02 3.005e+02 3.508e+02 5.989e+02, threshold=6.009e+02, percent-clipped=0.0 2023-04-17 03:55:55,489 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62727.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:55:58,432 INFO [train.py:893] (0/4) Epoch 24, batch 650, loss[loss=0.1589, simple_loss=0.2192, pruned_loss=0.04927, over 13355.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2362, pruned_loss=0.05759, over 2553150.74 frames. ], batch size: 67, lr: 5.47e-03, grad_scale: 16.0 2023-04-17 03:56:13,648 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62747.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:56:37,383 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62773.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:56:42,091 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62778.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:56:44,232 INFO [train.py:893] (0/4) Epoch 24, batch 700, loss[loss=0.1853, simple_loss=0.247, pruned_loss=0.06176, over 13525.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2351, pruned_loss=0.05682, over 2579299.73 frames. ], batch size: 76, lr: 5.46e-03, grad_scale: 16.0 2023-04-17 03:57:24,531 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.679e+02 2.589e+02 3.020e+02 3.370e+02 5.327e+02, threshold=6.040e+02, percent-clipped=0.0 2023-04-17 03:57:27,337 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62826.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:57:30,677 INFO [train.py:893] (0/4) Epoch 24, batch 750, loss[loss=0.1629, simple_loss=0.2287, pruned_loss=0.04855, over 13267.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2347, pruned_loss=0.05676, over 2599129.03 frames. ], batch size: 124, lr: 5.46e-03, grad_scale: 16.0 2023-04-17 03:57:34,401 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62834.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:58:11,919 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.2316, 2.2499, 4.0609, 3.8444, 3.8596, 3.2891, 3.6429, 3.0744], device='cuda:0'), covar=tensor([0.2118, 0.1644, 0.0172, 0.0256, 0.0310, 0.0661, 0.0309, 0.0826], device='cuda:0'), in_proj_covar=tensor([0.0216, 0.0185, 0.0128, 0.0131, 0.0136, 0.0177, 0.0147, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 03:58:12,871 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7585, 3.6517, 4.3440, 3.0823, 2.8644, 2.9073, 4.6482, 4.6804], device='cuda:0'), covar=tensor([0.1132, 0.1528, 0.0329, 0.1705, 0.1532, 0.1580, 0.0234, 0.0308], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0274, 0.0200, 0.0226, 0.0221, 0.0183, 0.0216, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 03:58:17,449 INFO [train.py:893] (0/4) Epoch 24, batch 800, loss[loss=0.2141, simple_loss=0.2663, pruned_loss=0.08088, over 13458.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2357, pruned_loss=0.05728, over 2613015.17 frames. ], batch size: 100, lr: 5.46e-03, grad_scale: 16.0 2023-04-17 03:58:36,566 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62901.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:58:56,776 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.925e+02 2.701e+02 3.020e+02 3.657e+02 6.500e+02, threshold=6.040e+02, percent-clipped=1.0 2023-04-17 03:59:02,155 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62929.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:59:02,714 INFO [train.py:893] (0/4) Epoch 24, batch 850, loss[loss=0.1924, simple_loss=0.2508, pruned_loss=0.06695, over 13423.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.2367, pruned_loss=0.05792, over 2626116.83 frames. ], batch size: 95, lr: 5.46e-03, grad_scale: 16.0 2023-04-17 03:59:14,277 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62941.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:59:26,935 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-17 03:59:48,072 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62977.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 03:59:50,403 INFO [train.py:893] (0/4) Epoch 24, batch 900, loss[loss=0.1842, simple_loss=0.2426, pruned_loss=0.0629, over 13389.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2366, pruned_loss=0.05829, over 2635209.88 frames. ], batch size: 113, lr: 5.46e-03, grad_scale: 16.0 2023-04-17 04:00:21,156 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-17 04:00:28,791 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63022.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:00:29,472 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.868e+02 2.672e+02 3.167e+02 4.016e+02 1.001e+03, threshold=6.335e+02, percent-clipped=2.0 2023-04-17 04:00:36,753 INFO [train.py:893] (0/4) Epoch 24, batch 950, loss[loss=0.181, simple_loss=0.2485, pruned_loss=0.0567, over 13544.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2356, pruned_loss=0.0584, over 2640164.61 frames. ], batch size: 87, lr: 5.45e-03, grad_scale: 16.0 2023-04-17 04:00:52,693 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63047.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:01:13,338 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.1258, 1.7982, 3.8034, 3.6805, 3.5455, 2.9524, 3.4270, 2.8099], device='cuda:0'), covar=tensor([0.2030, 0.1857, 0.0151, 0.0182, 0.0297, 0.0816, 0.0278, 0.0831], device='cuda:0'), in_proj_covar=tensor([0.0215, 0.0184, 0.0127, 0.0132, 0.0136, 0.0175, 0.0147, 0.0184], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 04:01:22,942 INFO [train.py:893] (0/4) Epoch 24, batch 1000, loss[loss=0.1606, simple_loss=0.2147, pruned_loss=0.0533, over 13507.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2334, pruned_loss=0.05741, over 2648362.42 frames. ], batch size: 70, lr: 5.45e-03, grad_scale: 16.0 2023-04-17 04:01:36,836 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63095.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:01:49,147 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3184, 2.5002, 2.2367, 4.1640, 4.6859, 3.4970, 4.4989, 4.3809], device='cuda:0'), covar=tensor([0.0105, 0.0983, 0.1098, 0.0112, 0.0082, 0.0452, 0.0097, 0.0093], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0094, 0.0102, 0.0084, 0.0071, 0.0084, 0.0059, 0.0081], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 04:02:02,200 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.910e+02 2.653e+02 3.155e+02 3.689e+02 7.228e+02, threshold=6.310e+02, percent-clipped=3.0 2023-04-17 04:02:08,130 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63129.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:02:08,726 INFO [train.py:893] (0/4) Epoch 24, batch 1050, loss[loss=0.2023, simple_loss=0.2553, pruned_loss=0.07468, over 13531.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2327, pruned_loss=0.05646, over 2655477.69 frames. ], batch size: 91, lr: 5.45e-03, grad_scale: 16.0 2023-04-17 04:02:55,249 INFO [train.py:893] (0/4) Epoch 24, batch 1100, loss[loss=0.1752, simple_loss=0.2379, pruned_loss=0.05624, over 13019.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2341, pruned_loss=0.05666, over 2658285.09 frames. ], batch size: 142, lr: 5.45e-03, grad_scale: 16.0 2023-04-17 04:03:03,680 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5249, 4.5865, 3.2518, 4.1943, 4.4678, 3.1288, 3.9613, 3.3791], device='cuda:0'), covar=tensor([0.0244, 0.0205, 0.1016, 0.0413, 0.0190, 0.1042, 0.0442, 0.1095], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0181, 0.0180, 0.0226, 0.0141, 0.0162, 0.0163, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:03:16,466 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63201.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:03:33,921 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63221.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:03:35,169 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.822e+02 2.654e+02 2.983e+02 3.494e+02 4.938e+02, threshold=5.966e+02, percent-clipped=0.0 2023-04-17 04:03:42,633 INFO [train.py:893] (0/4) Epoch 24, batch 1150, loss[loss=0.1776, simple_loss=0.2359, pruned_loss=0.05961, over 13545.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2338, pruned_loss=0.05614, over 2659560.10 frames. ], batch size: 78, lr: 5.45e-03, grad_scale: 16.0 2023-04-17 04:03:52,022 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63241.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:03:59,620 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63249.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:04:27,369 INFO [train.py:893] (0/4) Epoch 24, batch 1200, loss[loss=0.1921, simple_loss=0.2486, pruned_loss=0.06781, over 13224.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2343, pruned_loss=0.05641, over 2660116.46 frames. ], batch size: 124, lr: 5.44e-03, grad_scale: 16.0 2023-04-17 04:04:29,396 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63282.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:04:35,112 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63289.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:04:54,563 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-17 04:05:06,070 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-17 04:05:06,342 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63322.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:05:06,915 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.834e+02 2.724e+02 3.082e+02 3.852e+02 6.773e+02, threshold=6.164e+02, percent-clipped=2.0 2023-04-17 04:05:10,596 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3391, 2.3219, 4.1783, 3.9793, 4.0379, 3.2237, 3.7209, 3.1215], device='cuda:0'), covar=tensor([0.1783, 0.1373, 0.0115, 0.0185, 0.0152, 0.0614, 0.0263, 0.0686], device='cuda:0'), in_proj_covar=tensor([0.0214, 0.0181, 0.0127, 0.0130, 0.0134, 0.0173, 0.0146, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 04:05:13,753 INFO [train.py:893] (0/4) Epoch 24, batch 1250, loss[loss=0.1895, simple_loss=0.2486, pruned_loss=0.06524, over 13565.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2343, pruned_loss=0.05641, over 2659535.27 frames. ], batch size: 89, lr: 5.44e-03, grad_scale: 16.0 2023-04-17 04:05:49,768 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63370.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:05:59,085 INFO [train.py:893] (0/4) Epoch 24, batch 1300, loss[loss=0.1651, simple_loss=0.2243, pruned_loss=0.05299, over 13069.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2355, pruned_loss=0.0568, over 2659278.81 frames. ], batch size: 142, lr: 5.44e-03, grad_scale: 16.0 2023-04-17 04:06:09,164 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2366, 2.4824, 2.1919, 4.1708, 4.6606, 3.4174, 4.5243, 4.3341], device='cuda:0'), covar=tensor([0.0105, 0.0961, 0.1003, 0.0098, 0.0072, 0.0488, 0.0078, 0.0092], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0094, 0.0101, 0.0083, 0.0070, 0.0083, 0.0058, 0.0081], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 04:06:15,201 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8218, 4.1247, 3.8907, 3.9831, 3.9872, 4.2943, 4.0824, 3.9739], device='cuda:0'), covar=tensor([0.0341, 0.0299, 0.0320, 0.0724, 0.0298, 0.0240, 0.0335, 0.0438], device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0161, 0.0182, 0.0271, 0.0181, 0.0198, 0.0178, 0.0114], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 04:06:36,002 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7211, 3.4981, 3.6878, 2.2594, 3.7697, 3.7487, 3.7033, 3.8694], device='cuda:0'), covar=tensor([0.0220, 0.0160, 0.0149, 0.1189, 0.0140, 0.0226, 0.0099, 0.0122], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0057, 0.0085, 0.0102, 0.0099, 0.0111, 0.0082, 0.0078], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:06:37,558 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63420.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:06:39,246 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63422.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:06:39,717 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 2.696e+02 3.109e+02 3.925e+02 6.868e+02, threshold=6.218e+02, percent-clipped=1.0 2023-04-17 04:06:42,874 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-17 04:06:44,990 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63429.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:06:45,538 INFO [train.py:893] (0/4) Epoch 24, batch 1350, loss[loss=0.1646, simple_loss=0.2152, pruned_loss=0.05701, over 13227.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2362, pruned_loss=0.05735, over 2658960.18 frames. ], batch size: 58, lr: 5.44e-03, grad_scale: 16.0 2023-04-17 04:06:50,719 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1257, 4.5927, 4.3667, 4.3882, 4.4025, 4.1593, 4.6659, 4.7053], device='cuda:0'), covar=tensor([0.0246, 0.0235, 0.0218, 0.0340, 0.0294, 0.0296, 0.0259, 0.0212], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0207, 0.0170, 0.0188, 0.0159, 0.0207, 0.0137, 0.0186], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:06:56,884 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8103, 4.0936, 3.9854, 3.5671, 3.8786, 4.2141, 4.1081, 4.0097], device='cuda:0'), covar=tensor([0.0437, 0.0367, 0.0413, 0.1320, 0.0453, 0.0398, 0.0377, 0.0616], device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0161, 0.0182, 0.0270, 0.0182, 0.0198, 0.0179, 0.0114], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 04:07:16,560 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7308, 3.6321, 2.9718, 3.2124, 2.8659, 2.2730, 3.6412, 2.1872], device='cuda:0'), covar=tensor([0.0747, 0.0633, 0.0509, 0.0523, 0.0790, 0.1989, 0.0985, 0.1346], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0146, 0.0135, 0.0119, 0.0149, 0.0192, 0.0181, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:07:20,149 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.22 vs. limit=5.0 2023-04-17 04:07:30,472 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63477.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:07:32,652 INFO [train.py:893] (0/4) Epoch 24, batch 1400, loss[loss=0.1802, simple_loss=0.2383, pruned_loss=0.06107, over 13361.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2359, pruned_loss=0.05675, over 2661841.77 frames. ], batch size: 67, lr: 5.43e-03, grad_scale: 16.0 2023-04-17 04:07:34,550 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63481.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:07:36,202 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63483.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:07:42,499 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-17 04:08:00,863 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.6611, 5.2136, 5.1166, 5.1991, 5.0819, 5.0118, 5.6454, 5.2150], device='cuda:0'), covar=tensor([0.0736, 0.1341, 0.2331, 0.2545, 0.1193, 0.1723, 0.0916, 0.1254], device='cuda:0'), in_proj_covar=tensor([0.0347, 0.0395, 0.0486, 0.0490, 0.0313, 0.0366, 0.0454, 0.0362], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:08:02,303 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-17 04:08:12,177 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.694e+02 3.164e+02 3.726e+02 7.551e+02, threshold=6.327e+02, percent-clipped=1.0 2023-04-17 04:08:14,386 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-17 04:08:14,846 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3412, 4.8409, 4.6806, 4.8329, 4.7102, 4.6797, 5.3224, 4.9217], device='cuda:0'), covar=tensor([0.0654, 0.1212, 0.2132, 0.2624, 0.0967, 0.1574, 0.0849, 0.1135], device='cuda:0'), in_proj_covar=tensor([0.0348, 0.0397, 0.0487, 0.0491, 0.0314, 0.0367, 0.0456, 0.0363], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:08:18,628 INFO [train.py:893] (0/4) Epoch 24, batch 1450, loss[loss=0.1534, simple_loss=0.216, pruned_loss=0.04541, over 13427.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2349, pruned_loss=0.05665, over 2664398.62 frames. ], batch size: 62, lr: 5.43e-03, grad_scale: 16.0 2023-04-17 04:08:55,243 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63569.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:08:55,453 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-17 04:09:02,479 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63577.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:09:05,533 INFO [train.py:893] (0/4) Epoch 24, batch 1500, loss[loss=0.1723, simple_loss=0.2309, pruned_loss=0.05689, over 13536.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2349, pruned_loss=0.05639, over 2664919.21 frames. ], batch size: 72, lr: 5.43e-03, grad_scale: 16.0 2023-04-17 04:09:07,833 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-17 04:09:10,741 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63586.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:09:26,830 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0847, 4.2518, 3.2467, 2.8797, 2.9371, 2.6126, 4.3127, 2.5337], device='cuda:0'), covar=tensor([0.1597, 0.0349, 0.1279, 0.2322, 0.0873, 0.3245, 0.0253, 0.3944], device='cuda:0'), in_proj_covar=tensor([0.0313, 0.0292, 0.0320, 0.0340, 0.0263, 0.0332, 0.0217, 0.0368], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 04:09:44,964 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.034e+02 2.667e+02 3.072e+02 3.554e+02 8.313e+02, threshold=6.144e+02, percent-clipped=1.0 2023-04-17 04:09:52,380 INFO [train.py:893] (0/4) Epoch 24, batch 1550, loss[loss=0.2061, simple_loss=0.2609, pruned_loss=0.07558, over 13040.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2349, pruned_loss=0.05628, over 2659209.80 frames. ], batch size: 142, lr: 5.43e-03, grad_scale: 16.0 2023-04-17 04:09:52,668 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63630.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:10:08,252 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63647.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:10:38,788 INFO [train.py:893] (0/4) Epoch 24, batch 1600, loss[loss=0.1618, simple_loss=0.2318, pruned_loss=0.04594, over 13545.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2355, pruned_loss=0.0563, over 2660729.92 frames. ], batch size: 72, lr: 5.43e-03, grad_scale: 16.0 2023-04-17 04:11:09,600 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5464, 2.8498, 2.3988, 4.4095, 4.9851, 3.7690, 4.8613, 4.5940], device='cuda:0'), covar=tensor([0.0081, 0.0794, 0.0954, 0.0095, 0.0060, 0.0421, 0.0062, 0.0082], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0093, 0.0100, 0.0083, 0.0070, 0.0083, 0.0058, 0.0080], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 04:11:19,071 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 2.516e+02 2.981e+02 3.582e+02 5.645e+02, threshold=5.962e+02, percent-clipped=0.0 2023-04-17 04:11:24,943 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5553, 2.6396, 3.0942, 4.1162, 3.7132, 4.1994, 3.2478, 2.6426], device='cuda:0'), covar=tensor([0.0339, 0.0799, 0.0704, 0.0066, 0.0218, 0.0058, 0.0622, 0.0861], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0147, 0.0166, 0.0101, 0.0121, 0.0099, 0.0169, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:11:25,482 INFO [train.py:893] (0/4) Epoch 24, batch 1650, loss[loss=0.184, simple_loss=0.2439, pruned_loss=0.06204, over 13516.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2352, pruned_loss=0.05585, over 2652297.94 frames. ], batch size: 70, lr: 5.42e-03, grad_scale: 16.0 2023-04-17 04:12:04,894 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9106, 2.6073, 2.3718, 1.5463, 1.5900, 2.2375, 2.2737, 2.8116], device='cuda:0'), covar=tensor([0.0874, 0.0331, 0.0531, 0.1609, 0.0187, 0.0568, 0.0700, 0.0221], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0147, 0.0124, 0.0211, 0.0114, 0.0166, 0.0178, 0.0136], device='cuda:0'), out_proj_covar=tensor([1.2596e-04, 1.0988e-04, 9.6539e-05, 1.5717e-04, 8.2373e-05, 1.2554e-04, 1.3413e-04, 1.0028e-04], device='cuda:0') 2023-04-17 04:12:08,801 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63776.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:12:10,503 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63778.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:12:12,628 INFO [train.py:893] (0/4) Epoch 24, batch 1700, loss[loss=0.2269, simple_loss=0.2711, pruned_loss=0.09135, over 11834.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.236, pruned_loss=0.05569, over 2652438.17 frames. ], batch size: 157, lr: 5.42e-03, grad_scale: 16.0 2023-04-17 04:12:41,149 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63810.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:12:53,078 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.875e+02 2.594e+02 3.016e+02 3.621e+02 1.016e+03, threshold=6.032e+02, percent-clipped=3.0 2023-04-17 04:12:59,063 INFO [train.py:893] (0/4) Epoch 24, batch 1750, loss[loss=0.1841, simple_loss=0.2411, pruned_loss=0.06359, over 13457.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2343, pruned_loss=0.05496, over 2658160.66 frames. ], batch size: 106, lr: 5.42e-03, grad_scale: 16.0 2023-04-17 04:13:13,890 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63845.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:13:33,370 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.73 vs. limit=5.0 2023-04-17 04:13:37,266 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63871.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:13:43,751 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63877.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:13:46,035 INFO [train.py:893] (0/4) Epoch 24, batch 1800, loss[loss=0.1678, simple_loss=0.2334, pruned_loss=0.05108, over 13541.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2339, pruned_loss=0.0545, over 2661126.97 frames. ], batch size: 76, lr: 5.42e-03, grad_scale: 16.0 2023-04-17 04:14:00,195 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0076, 3.7623, 3.9906, 2.3519, 4.2184, 4.0822, 4.0415, 4.2299], device='cuda:0'), covar=tensor([0.0240, 0.0165, 0.0131, 0.1142, 0.0133, 0.0204, 0.0122, 0.0102], device='cuda:0'), in_proj_covar=tensor([0.0095, 0.0058, 0.0086, 0.0104, 0.0100, 0.0113, 0.0084, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:14:10,096 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63906.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:14:19,835 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63916.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:14:20,703 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0691, 3.7443, 3.1247, 3.5180, 3.0472, 2.2858, 3.7985, 2.2724], device='cuda:0'), covar=tensor([0.0569, 0.0493, 0.0471, 0.0344, 0.0662, 0.1831, 0.0887, 0.1188], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0147, 0.0138, 0.0120, 0.0150, 0.0195, 0.0183, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:14:22,367 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.2447, 2.2099, 3.9795, 3.8470, 3.8265, 3.1840, 3.5776, 2.9549], device='cuda:0'), covar=tensor([0.1902, 0.1280, 0.0122, 0.0178, 0.0205, 0.0607, 0.0262, 0.0793], device='cuda:0'), in_proj_covar=tensor([0.0215, 0.0181, 0.0127, 0.0130, 0.0132, 0.0173, 0.0146, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 04:14:25,506 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.929e+02 2.462e+02 2.949e+02 3.735e+02 7.345e+02, threshold=5.897e+02, percent-clipped=3.0 2023-04-17 04:14:28,144 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63925.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:14:28,164 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63925.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:14:32,097 INFO [train.py:893] (0/4) Epoch 24, batch 1850, loss[loss=0.1948, simple_loss=0.2548, pruned_loss=0.06742, over 13436.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2337, pruned_loss=0.05438, over 2663625.62 frames. ], batch size: 106, lr: 5.42e-03, grad_scale: 16.0 2023-04-17 04:14:37,125 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-17 04:14:39,472 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-17 04:14:43,827 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63942.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:14:48,072 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0261, 1.9656, 2.3520, 3.2693, 3.0010, 3.3201, 2.6287, 2.3309], device='cuda:0'), covar=tensor([0.0299, 0.0924, 0.0823, 0.0101, 0.0297, 0.0100, 0.0633, 0.0791], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0147, 0.0166, 0.0100, 0.0121, 0.0099, 0.0168, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:15:17,199 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63977.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:15:19,403 INFO [train.py:893] (0/4) Epoch 24, batch 1900, loss[loss=0.1922, simple_loss=0.2456, pruned_loss=0.06945, over 13369.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2326, pruned_loss=0.05416, over 2662408.31 frames. ], batch size: 109, lr: 5.41e-03, grad_scale: 16.0 2023-04-17 04:15:28,864 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5335, 3.1500, 3.9740, 2.8129, 2.7043, 2.7395, 4.2922, 4.3525], device='cuda:0'), covar=tensor([0.1207, 0.2058, 0.0345, 0.1811, 0.1658, 0.1673, 0.0255, 0.0297], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0276, 0.0201, 0.0226, 0.0222, 0.0185, 0.0216, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:15:38,074 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-64000.pt 2023-04-17 04:16:03,221 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.832e+02 2.373e+02 2.755e+02 3.277e+02 5.989e+02, threshold=5.509e+02, percent-clipped=1.0 2023-04-17 04:16:03,584 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64023.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:16:09,824 INFO [train.py:893] (0/4) Epoch 24, batch 1950, loss[loss=0.1825, simple_loss=0.2435, pruned_loss=0.06073, over 13448.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2321, pruned_loss=0.05403, over 2665681.56 frames. ], batch size: 95, lr: 5.41e-03, grad_scale: 16.0 2023-04-17 04:16:45,281 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1770, 3.0091, 3.5013, 2.6141, 2.3664, 2.5133, 3.9216, 4.0178], device='cuda:0'), covar=tensor([0.1207, 0.1761, 0.0473, 0.1749, 0.1681, 0.1561, 0.0329, 0.0227], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0277, 0.0202, 0.0226, 0.0223, 0.0186, 0.0217, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:16:52,598 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64076.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:16:54,215 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64078.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:16:55,636 INFO [train.py:893] (0/4) Epoch 24, batch 2000, loss[loss=0.192, simple_loss=0.2586, pruned_loss=0.06273, over 13196.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2336, pruned_loss=0.0548, over 2664306.14 frames. ], batch size: 132, lr: 5.41e-03, grad_scale: 16.0 2023-04-17 04:16:59,927 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64084.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:17:02,305 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 04:17:35,974 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.856e+02 2.644e+02 3.150e+02 3.695e+02 7.820e+02, threshold=6.300e+02, percent-clipped=3.0 2023-04-17 04:17:37,641 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64124.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:17:39,274 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64126.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:17:43,179 INFO [train.py:893] (0/4) Epoch 24, batch 2050, loss[loss=0.1922, simple_loss=0.2531, pruned_loss=0.06562, over 13558.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2347, pruned_loss=0.05543, over 2661184.39 frames. ], batch size: 87, lr: 5.41e-03, grad_scale: 32.0 2023-04-17 04:17:47,726 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.1661, 2.0113, 3.8236, 3.7098, 3.6943, 3.0634, 3.4530, 2.8130], device='cuda:0'), covar=tensor([0.2020, 0.1499, 0.0139, 0.0150, 0.0209, 0.0667, 0.0252, 0.0804], device='cuda:0'), in_proj_covar=tensor([0.0216, 0.0182, 0.0128, 0.0131, 0.0133, 0.0174, 0.0146, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 04:18:16,664 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64166.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:18:29,319 INFO [train.py:893] (0/4) Epoch 24, batch 2100, loss[loss=0.1487, simple_loss=0.2203, pruned_loss=0.03861, over 13495.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2349, pruned_loss=0.05553, over 2667070.00 frames. ], batch size: 93, lr: 5.40e-03, grad_scale: 32.0 2023-04-17 04:18:30,474 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.2689, 1.9938, 3.7945, 3.6537, 3.7371, 3.0369, 3.4412, 2.8514], device='cuda:0'), covar=tensor([0.1795, 0.1451, 0.0162, 0.0299, 0.0216, 0.0682, 0.0266, 0.0781], device='cuda:0'), in_proj_covar=tensor([0.0214, 0.0180, 0.0127, 0.0130, 0.0133, 0.0172, 0.0145, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 04:18:46,622 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-17 04:18:49,014 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64201.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:18:50,828 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9295, 4.1593, 3.9445, 4.0413, 4.0685, 4.3754, 4.1524, 3.9823], device='cuda:0'), covar=tensor([0.0281, 0.0305, 0.0355, 0.0746, 0.0302, 0.0234, 0.0305, 0.0452], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0160, 0.0182, 0.0271, 0.0181, 0.0197, 0.0180, 0.0114], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 04:19:09,884 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.961e+02 2.501e+02 2.846e+02 3.361e+02 6.447e+02, threshold=5.692e+02, percent-clipped=1.0 2023-04-17 04:19:12,577 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64225.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:19:16,491 INFO [train.py:893] (0/4) Epoch 24, batch 2150, loss[loss=0.1603, simple_loss=0.2264, pruned_loss=0.04716, over 13033.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2342, pruned_loss=0.0549, over 2665424.57 frames. ], batch size: 142, lr: 5.40e-03, grad_scale: 32.0 2023-04-17 04:19:18,446 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64232.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:19:27,650 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64242.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:19:41,070 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9567, 2.8795, 2.5133, 1.7907, 1.8986, 2.4411, 2.5460, 3.0390], device='cuda:0'), covar=tensor([0.1324, 0.0372, 0.0747, 0.1860, 0.0424, 0.0567, 0.0869, 0.0257], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0150, 0.0128, 0.0214, 0.0117, 0.0169, 0.0182, 0.0139], device='cuda:0'), out_proj_covar=tensor([1.2882e-04, 1.1255e-04, 9.9074e-05, 1.5877e-04, 8.4528e-05, 1.2819e-04, 1.3688e-04, 1.0298e-04], device='cuda:0') 2023-04-17 04:19:54,991 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64272.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:19:55,844 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64273.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:20:02,258 INFO [train.py:893] (0/4) Epoch 24, batch 2200, loss[loss=0.163, simple_loss=0.2189, pruned_loss=0.05353, over 13188.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.234, pruned_loss=0.0546, over 2664087.61 frames. ], batch size: 58, lr: 5.40e-03, grad_scale: 32.0 2023-04-17 04:20:12,105 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64290.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:20:14,762 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64293.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:20:40,190 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-17 04:20:42,709 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.722e+02 2.378e+02 2.714e+02 3.263e+02 6.451e+02, threshold=5.428e+02, percent-clipped=1.0 2023-04-17 04:20:48,357 INFO [train.py:893] (0/4) Epoch 24, batch 2250, loss[loss=0.1687, simple_loss=0.2268, pruned_loss=0.05533, over 13524.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2332, pruned_loss=0.05462, over 2665935.44 frames. ], batch size: 85, lr: 5.40e-03, grad_scale: 32.0 2023-04-17 04:21:02,870 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64345.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:21:34,749 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64379.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:21:35,367 INFO [train.py:893] (0/4) Epoch 24, batch 2300, loss[loss=0.1668, simple_loss=0.2386, pruned_loss=0.04753, over 13518.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2333, pruned_loss=0.05468, over 2664497.17 frames. ], batch size: 91, lr: 5.40e-03, grad_scale: 32.0 2023-04-17 04:22:00,099 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64406.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:22:15,077 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.764e+02 2.581e+02 2.900e+02 3.410e+02 5.772e+02, threshold=5.800e+02, percent-clipped=1.0 2023-04-17 04:22:15,702 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-17 04:22:20,663 INFO [train.py:893] (0/4) Epoch 24, batch 2350, loss[loss=0.1594, simple_loss=0.2178, pruned_loss=0.05052, over 13486.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2337, pruned_loss=0.05485, over 2664298.97 frames. ], batch size: 93, lr: 5.39e-03, grad_scale: 32.0 2023-04-17 04:22:46,557 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 04:22:55,401 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64465.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:22:56,178 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64466.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:22:57,023 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3237, 2.2439, 3.9498, 3.8769, 3.8675, 3.2938, 3.5397, 3.0828], device='cuda:0'), covar=tensor([0.2177, 0.1505, 0.0186, 0.0219, 0.0204, 0.0603, 0.0314, 0.0848], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0185, 0.0130, 0.0133, 0.0136, 0.0177, 0.0149, 0.0186], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 04:23:08,872 INFO [train.py:893] (0/4) Epoch 24, batch 2400, loss[loss=0.161, simple_loss=0.2238, pruned_loss=0.04904, over 13247.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.233, pruned_loss=0.05484, over 2661660.16 frames. ], batch size: 124, lr: 5.39e-03, grad_scale: 32.0 2023-04-17 04:23:28,388 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64501.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:23:34,172 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7078, 3.5164, 4.2662, 3.0846, 2.8187, 2.8654, 4.6116, 4.7097], device='cuda:0'), covar=tensor([0.1300, 0.1824, 0.0368, 0.1719, 0.1685, 0.1677, 0.0285, 0.0251], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0277, 0.0202, 0.0226, 0.0223, 0.0186, 0.0216, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:23:40,517 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64514.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:23:47,926 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.856e+02 2.611e+02 3.047e+02 3.549e+02 5.859e+02, threshold=6.095e+02, percent-clipped=1.0 2023-04-17 04:23:51,651 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64526.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:23:55,415 INFO [train.py:893] (0/4) Epoch 24, batch 2450, loss[loss=0.1962, simple_loss=0.2546, pruned_loss=0.0689, over 13407.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2324, pruned_loss=0.0544, over 2659768.71 frames. ], batch size: 95, lr: 5.39e-03, grad_scale: 32.0 2023-04-17 04:24:01,717 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.37 vs. limit=5.0 2023-04-17 04:24:10,380 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-17 04:24:12,576 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64549.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:24:34,761 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64572.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:24:41,294 INFO [train.py:893] (0/4) Epoch 24, batch 2500, loss[loss=0.1705, simple_loss=0.2292, pruned_loss=0.0559, over 13274.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.233, pruned_loss=0.05486, over 2658017.95 frames. ], batch size: 124, lr: 5.39e-03, grad_scale: 32.0 2023-04-17 04:24:48,955 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64588.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:24:49,919 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64589.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:25:18,426 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64620.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:25:21,447 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.754e+02 2.530e+02 2.853e+02 3.517e+02 5.646e+02, threshold=5.707e+02, percent-clipped=0.0 2023-04-17 04:25:23,297 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3256, 4.8246, 4.6735, 4.8269, 4.6016, 4.6627, 5.2679, 4.8185], device='cuda:0'), covar=tensor([0.0619, 0.1141, 0.2086, 0.2150, 0.0985, 0.1614, 0.0833, 0.1099], device='cuda:0'), in_proj_covar=tensor([0.0348, 0.0396, 0.0492, 0.0492, 0.0311, 0.0364, 0.0457, 0.0363], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:25:27,370 INFO [train.py:893] (0/4) Epoch 24, batch 2550, loss[loss=0.1679, simple_loss=0.2299, pruned_loss=0.05295, over 13201.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2335, pruned_loss=0.0549, over 2662614.85 frames. ], batch size: 132, lr: 5.39e-03, grad_scale: 32.0 2023-04-17 04:25:43,935 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64648.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:25:45,668 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64650.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:25:51,134 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 04:26:07,827 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5056, 2.5360, 2.7940, 3.9154, 3.5104, 3.9979, 3.0837, 2.5413], device='cuda:0'), covar=tensor([0.0290, 0.0859, 0.0791, 0.0091, 0.0267, 0.0067, 0.0731, 0.0891], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0149, 0.0166, 0.0101, 0.0121, 0.0099, 0.0170, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:26:07,886 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0898, 4.1029, 2.8904, 3.7958, 3.9691, 2.6449, 3.5917, 2.8088], device='cuda:0'), covar=tensor([0.0332, 0.0240, 0.1114, 0.0406, 0.0274, 0.1334, 0.0579, 0.1166], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0180, 0.0178, 0.0226, 0.0141, 0.0162, 0.0163, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:26:12,605 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64679.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:26:13,198 INFO [train.py:893] (0/4) Epoch 24, batch 2600, loss[loss=0.1565, simple_loss=0.2125, pruned_loss=0.05027, over 13446.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2323, pruned_loss=0.05429, over 2668000.41 frames. ], batch size: 65, lr: 5.38e-03, grad_scale: 32.0 2023-04-17 04:26:18,507 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4032, 3.1806, 3.9216, 2.8264, 2.5605, 2.7054, 4.2111, 4.3047], device='cuda:0'), covar=tensor([0.1308, 0.1968, 0.0401, 0.1789, 0.1684, 0.1650, 0.0293, 0.0228], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0274, 0.0199, 0.0224, 0.0221, 0.0183, 0.0214, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:26:33,517 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9377, 4.4689, 4.3323, 4.5036, 4.1800, 4.2713, 4.9197, 4.4475], device='cuda:0'), covar=tensor([0.0839, 0.1136, 0.2079, 0.2523, 0.1201, 0.1724, 0.0936, 0.1213], device='cuda:0'), in_proj_covar=tensor([0.0348, 0.0396, 0.0490, 0.0491, 0.0311, 0.0363, 0.0456, 0.0363], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:26:33,532 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64701.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:26:35,092 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64703.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:26:39,766 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64709.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:26:50,189 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.964e+02 2.522e+02 3.008e+02 3.561e+02 6.412e+02, threshold=6.016e+02, percent-clipped=1.0 2023-04-17 04:26:53,415 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64727.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:26:55,606 INFO [train.py:893] (0/4) Epoch 24, batch 2650, loss[loss=0.1583, simple_loss=0.2173, pruned_loss=0.04963, over 13523.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2333, pruned_loss=0.05532, over 2656338.65 frames. ], batch size: 72, lr: 5.38e-03, grad_scale: 32.0 2023-04-17 04:27:07,160 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64745.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:27:12,599 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64752.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:27:22,192 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64764.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:27:34,651 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-24.pt 2023-04-17 04:27:56,670 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-17 04:28:00,593 INFO [train.py:893] (0/4) Epoch 25, batch 0, loss[loss=0.1484, simple_loss=0.213, pruned_loss=0.04188, over 13335.00 frames. ], tot_loss[loss=0.1484, simple_loss=0.213, pruned_loss=0.04188, over 13335.00 frames. ], batch size: 73, lr: 5.27e-03, grad_scale: 32.0 2023-04-17 04:28:00,594 INFO [train.py:918] (0/4) Computing validation loss 2023-04-17 04:28:09,088 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.6114, 5.1773, 5.1324, 5.2384, 5.0137, 5.1319, 5.6115, 5.1251], device='cuda:0'), covar=tensor([0.0749, 0.1185, 0.1860, 0.1966, 0.0924, 0.1351, 0.0885, 0.1155], device='cuda:0'), in_proj_covar=tensor([0.0347, 0.0394, 0.0488, 0.0488, 0.0308, 0.0362, 0.0454, 0.0361], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:28:14,115 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5514, 3.2785, 4.0647, 2.9418, 2.7731, 2.8625, 4.3970, 4.4403], device='cuda:0'), covar=tensor([0.1329, 0.2201, 0.0407, 0.1929, 0.1835, 0.1676, 0.0262, 0.0307], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0275, 0.0200, 0.0225, 0.0221, 0.0184, 0.0215, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:28:22,770 INFO [train.py:927] (0/4) Epoch 25, validation: loss=0.134, simple_loss=0.1935, pruned_loss=0.03718, over 2446609.00 frames. 2023-04-17 04:28:22,771 INFO [train.py:928] (0/4) Maximum memory allocated so far is 13044MB 2023-04-17 04:28:46,862 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64806.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:28:54,076 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64813.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:28:54,893 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4194, 2.6952, 2.8133, 3.9006, 3.5525, 3.9573, 3.0740, 2.6463], device='cuda:0'), covar=tensor([0.0310, 0.0751, 0.0770, 0.0071, 0.0253, 0.0066, 0.0674, 0.0812], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0147, 0.0164, 0.0100, 0.0120, 0.0099, 0.0168, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:29:00,517 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64821.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:29:02,047 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.683e+02 2.688e+02 3.039e+02 3.680e+02 5.833e+02, threshold=6.079e+02, percent-clipped=0.0 2023-04-17 04:29:07,894 INFO [train.py:893] (0/4) Epoch 25, batch 50, loss[loss=0.149, simple_loss=0.2062, pruned_loss=0.04585, over 13408.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2253, pruned_loss=0.05326, over 602992.68 frames. ], batch size: 62, lr: 5.27e-03, grad_scale: 32.0 2023-04-17 04:29:10,627 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1947, 2.5188, 2.3053, 4.1046, 4.6544, 3.4845, 4.5275, 4.3024], device='cuda:0'), covar=tensor([0.0129, 0.1021, 0.1086, 0.0133, 0.0101, 0.0539, 0.0121, 0.0114], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0092, 0.0099, 0.0083, 0.0070, 0.0082, 0.0058, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 04:29:30,887 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-17 04:29:30,887 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-17 04:29:30,888 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-17 04:29:31,602 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-17 04:29:31,610 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-17 04:29:31,632 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-17 04:29:32,383 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-17 04:29:52,473 INFO [train.py:893] (0/4) Epoch 25, batch 100, loss[loss=0.1781, simple_loss=0.2413, pruned_loss=0.05743, over 13376.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2292, pruned_loss=0.05541, over 1061169.53 frames. ], batch size: 73, lr: 5.27e-03, grad_scale: 32.0 2023-04-17 04:30:00,744 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64888.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:30:32,602 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.977e+02 2.699e+02 3.087e+02 3.746e+02 6.680e+02, threshold=6.173e+02, percent-clipped=1.0 2023-04-17 04:30:37,710 INFO [train.py:893] (0/4) Epoch 25, batch 150, loss[loss=0.1832, simple_loss=0.2474, pruned_loss=0.05945, over 13506.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2329, pruned_loss=0.0572, over 1417695.70 frames. ], batch size: 93, lr: 5.26e-03, grad_scale: 32.0 2023-04-17 04:30:38,921 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3662, 3.2135, 4.0189, 2.8422, 2.5637, 2.7548, 4.2293, 4.2881], device='cuda:0'), covar=tensor([0.1252, 0.1903, 0.0360, 0.1755, 0.1682, 0.1578, 0.0260, 0.0223], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0273, 0.0198, 0.0224, 0.0219, 0.0182, 0.0213, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:30:43,778 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64936.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:30:52,689 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64945.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:31:04,847 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7421, 3.8889, 2.8398, 2.6205, 2.5834, 2.3772, 3.9656, 2.2140], device='cuda:0'), covar=tensor([0.2147, 0.0451, 0.1647, 0.2594, 0.1165, 0.3752, 0.0353, 0.4965], device='cuda:0'), in_proj_covar=tensor([0.0316, 0.0294, 0.0324, 0.0343, 0.0265, 0.0336, 0.0220, 0.0373], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 04:31:18,112 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2023-04-17 04:31:20,238 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2914, 4.7918, 4.6604, 4.7998, 4.4861, 4.6619, 5.2468, 4.6962], device='cuda:0'), covar=tensor([0.0634, 0.1219, 0.2001, 0.2308, 0.1007, 0.1561, 0.0846, 0.1227], device='cuda:0'), in_proj_covar=tensor([0.0348, 0.0396, 0.0490, 0.0488, 0.0309, 0.0363, 0.0455, 0.0361], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:31:21,110 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64976.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:31:23,224 INFO [train.py:893] (0/4) Epoch 25, batch 200, loss[loss=0.1762, simple_loss=0.2334, pruned_loss=0.05949, over 13463.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2356, pruned_loss=0.05817, over 1684417.55 frames. ], batch size: 106, lr: 5.26e-03, grad_scale: 32.0 2023-04-17 04:31:43,552 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65001.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:31:45,980 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65004.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:31:48,608 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8274, 4.1561, 3.8744, 3.6899, 3.9579, 4.2334, 4.0077, 3.8402], device='cuda:0'), covar=tensor([0.0287, 0.0290, 0.0356, 0.1173, 0.0315, 0.0287, 0.0335, 0.0539], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0161, 0.0182, 0.0271, 0.0182, 0.0199, 0.0180, 0.0114], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 04:32:02,059 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.794e+02 2.476e+02 2.997e+02 3.532e+02 5.574e+02, threshold=5.994e+02, percent-clipped=0.0 2023-04-17 04:32:06,848 INFO [train.py:893] (0/4) Epoch 25, batch 250, loss[loss=0.1876, simple_loss=0.2389, pruned_loss=0.06817, over 13204.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2357, pruned_loss=0.05824, over 1902345.74 frames. ], batch size: 132, lr: 5.26e-03, grad_scale: 32.0 2023-04-17 04:32:15,137 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65037.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 04:32:26,097 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65049.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:32:31,178 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65055.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:32:34,324 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65059.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:32:52,653 INFO [train.py:893] (0/4) Epoch 25, batch 300, loss[loss=0.1679, simple_loss=0.2374, pruned_loss=0.0492, over 13533.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2359, pruned_loss=0.0576, over 2073089.46 frames. ], batch size: 87, lr: 5.26e-03, grad_scale: 32.0 2023-04-17 04:33:12,417 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65101.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:33:13,338 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7866, 2.7257, 2.3091, 1.7028, 1.7446, 2.2023, 2.3574, 2.9108], device='cuda:0'), covar=tensor([0.0956, 0.0307, 0.0666, 0.1497, 0.0242, 0.0548, 0.0651, 0.0202], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0147, 0.0125, 0.0210, 0.0115, 0.0167, 0.0178, 0.0137], device='cuda:0'), out_proj_covar=tensor([1.2819e-04, 1.1007e-04, 9.7623e-05, 1.5606e-04, 8.3373e-05, 1.2668e-04, 1.3414e-04, 1.0099e-04], device='cuda:0') 2023-04-17 04:33:18,022 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65108.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:33:25,599 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65116.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:33:29,570 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65121.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:33:30,964 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.916e+02 2.607e+02 3.049e+02 3.512e+02 7.816e+02, threshold=6.098e+02, percent-clipped=2.0 2023-04-17 04:33:37,306 INFO [train.py:893] (0/4) Epoch 25, batch 350, loss[loss=0.1837, simple_loss=0.2374, pruned_loss=0.06495, over 13515.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2357, pruned_loss=0.05752, over 2206643.53 frames. ], batch size: 85, lr: 5.26e-03, grad_scale: 32.0 2023-04-17 04:34:13,295 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65169.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:34:22,077 INFO [train.py:893] (0/4) Epoch 25, batch 400, loss[loss=0.1969, simple_loss=0.2522, pruned_loss=0.07074, over 13532.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2357, pruned_loss=0.05725, over 2307015.26 frames. ], batch size: 72, lr: 5.25e-03, grad_scale: 32.0 2023-04-17 04:34:39,114 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0549, 3.7977, 3.9387, 2.3989, 4.3338, 4.1218, 3.9880, 4.3029], device='cuda:0'), covar=tensor([0.0248, 0.0163, 0.0163, 0.1125, 0.0142, 0.0251, 0.0162, 0.0106], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0059, 0.0087, 0.0106, 0.0101, 0.0113, 0.0085, 0.0080], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:35:01,571 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.808e+02 2.550e+02 3.090e+02 3.761e+02 8.454e+02, threshold=6.180e+02, percent-clipped=2.0 2023-04-17 04:35:06,640 INFO [train.py:893] (0/4) Epoch 25, batch 450, loss[loss=0.1827, simple_loss=0.241, pruned_loss=0.06217, over 13543.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2368, pruned_loss=0.05756, over 2381883.63 frames. ], batch size: 87, lr: 5.25e-03, grad_scale: 32.0 2023-04-17 04:35:21,474 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65245.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:35:32,538 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-17 04:35:36,097 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2848, 3.0290, 3.7350, 2.7796, 2.4682, 2.6359, 4.0311, 4.1494], device='cuda:0'), covar=tensor([0.1222, 0.2059, 0.0438, 0.1746, 0.1736, 0.1468, 0.0293, 0.0242], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0275, 0.0201, 0.0226, 0.0221, 0.0184, 0.0215, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:35:49,067 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6797, 3.7522, 2.5798, 3.1840, 2.9030, 2.0289, 3.8671, 2.0803], device='cuda:0'), covar=tensor([0.0904, 0.0445, 0.0879, 0.0517, 0.0797, 0.2437, 0.0637, 0.1659], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0147, 0.0136, 0.0119, 0.0149, 0.0191, 0.0183, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:35:52,073 INFO [train.py:893] (0/4) Epoch 25, batch 500, loss[loss=0.1535, simple_loss=0.2193, pruned_loss=0.0439, over 13329.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2366, pruned_loss=0.05699, over 2442884.59 frames. ], batch size: 73, lr: 5.25e-03, grad_scale: 32.0 2023-04-17 04:35:58,628 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65285.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:36:05,057 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65293.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:36:10,728 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1728, 4.4384, 4.1579, 4.2873, 4.2687, 4.6007, 4.3534, 4.2450], device='cuda:0'), covar=tensor([0.0291, 0.0269, 0.0335, 0.0800, 0.0274, 0.0224, 0.0331, 0.0318], device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0162, 0.0183, 0.0271, 0.0182, 0.0200, 0.0181, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 04:36:15,579 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65304.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:36:32,418 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.759e+02 2.424e+02 2.748e+02 3.322e+02 7.565e+02, threshold=5.496e+02, percent-clipped=1.0 2023-04-17 04:36:37,397 INFO [train.py:893] (0/4) Epoch 25, batch 550, loss[loss=0.1692, simple_loss=0.2379, pruned_loss=0.05026, over 13534.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2362, pruned_loss=0.0566, over 2491941.18 frames. ], batch size: 76, lr: 5.25e-03, grad_scale: 32.0 2023-04-17 04:36:38,561 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2078, 4.0033, 4.2278, 2.6578, 4.5213, 4.2613, 4.1984, 4.4505], device='cuda:0'), covar=tensor([0.0250, 0.0151, 0.0155, 0.1097, 0.0157, 0.0275, 0.0163, 0.0122], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0058, 0.0086, 0.0105, 0.0101, 0.0113, 0.0084, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:36:41,529 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65332.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 04:36:42,519 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1761, 4.4267, 3.3640, 3.0759, 3.2976, 2.6629, 4.5080, 2.5877], device='cuda:0'), covar=tensor([0.1702, 0.0335, 0.1244, 0.2248, 0.0815, 0.3248, 0.0252, 0.3747], device='cuda:0'), in_proj_covar=tensor([0.0316, 0.0293, 0.0323, 0.0344, 0.0264, 0.0334, 0.0219, 0.0371], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 04:36:53,793 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65346.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:36:59,110 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65352.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:37:04,875 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65359.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:37:22,607 INFO [train.py:893] (0/4) Epoch 25, batch 600, loss[loss=0.1479, simple_loss=0.2148, pruned_loss=0.04054, over 13533.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2349, pruned_loss=0.05622, over 2531240.96 frames. ], batch size: 70, lr: 5.25e-03, grad_scale: 32.0 2023-04-17 04:37:42,709 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65401.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:37:47,777 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65407.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:37:48,683 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65408.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:37:51,921 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65411.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:38:02,392 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.737e+02 2.398e+02 2.864e+02 3.237e+02 5.045e+02, threshold=5.727e+02, percent-clipped=0.0 2023-04-17 04:38:08,684 INFO [train.py:893] (0/4) Epoch 25, batch 650, loss[loss=0.1803, simple_loss=0.2453, pruned_loss=0.05761, over 13232.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2339, pruned_loss=0.05545, over 2563511.57 frames. ], batch size: 132, lr: 5.24e-03, grad_scale: 32.0 2023-04-17 04:38:26,807 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65449.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:38:32,744 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65456.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:38:34,494 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0882, 4.4183, 4.1010, 4.1935, 4.2268, 4.5323, 4.3236, 4.1918], device='cuda:0'), covar=tensor([0.0300, 0.0255, 0.0315, 0.0763, 0.0240, 0.0232, 0.0264, 0.0307], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0160, 0.0181, 0.0268, 0.0180, 0.0198, 0.0179, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 04:38:54,409 INFO [train.py:893] (0/4) Epoch 25, batch 700, loss[loss=0.1666, simple_loss=0.2247, pruned_loss=0.05422, over 13194.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2327, pruned_loss=0.05489, over 2579941.96 frames. ], batch size: 132, lr: 5.24e-03, grad_scale: 32.0 2023-04-17 04:39:23,979 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-17 04:39:27,713 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9157, 1.8254, 3.6384, 3.5856, 3.5364, 2.8031, 3.2389, 2.7593], device='cuda:0'), covar=tensor([0.2024, 0.1497, 0.0185, 0.0208, 0.0256, 0.0732, 0.0332, 0.0833], device='cuda:0'), in_proj_covar=tensor([0.0220, 0.0185, 0.0130, 0.0135, 0.0137, 0.0177, 0.0149, 0.0186], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 04:39:33,903 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.923e+02 2.683e+02 3.062e+02 3.577e+02 6.124e+02, threshold=6.124e+02, percent-clipped=1.0 2023-04-17 04:39:38,797 INFO [train.py:893] (0/4) Epoch 25, batch 750, loss[loss=0.1552, simple_loss=0.2235, pruned_loss=0.04343, over 13428.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2327, pruned_loss=0.05554, over 2597016.65 frames. ], batch size: 106, lr: 5.24e-03, grad_scale: 32.0 2023-04-17 04:39:41,545 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65532.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 04:40:21,466 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-17 04:40:23,421 INFO [train.py:893] (0/4) Epoch 25, batch 800, loss[loss=0.1732, simple_loss=0.2251, pruned_loss=0.06063, over 13384.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.234, pruned_loss=0.05614, over 2611417.93 frames. ], batch size: 62, lr: 5.24e-03, grad_scale: 32.0 2023-04-17 04:40:32,609 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1357, 4.6591, 4.4762, 4.5911, 4.3617, 4.4908, 5.0523, 4.6809], device='cuda:0'), covar=tensor([0.0720, 0.1202, 0.2488, 0.2537, 0.1005, 0.1540, 0.0913, 0.1268], device='cuda:0'), in_proj_covar=tensor([0.0350, 0.0398, 0.0493, 0.0494, 0.0311, 0.0368, 0.0457, 0.0364], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:40:36,128 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65593.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 04:41:03,597 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.878e+02 2.542e+02 2.840e+02 3.512e+02 5.061e+02, threshold=5.680e+02, percent-clipped=0.0 2023-04-17 04:41:09,865 INFO [train.py:893] (0/4) Epoch 25, batch 850, loss[loss=0.1795, simple_loss=0.2403, pruned_loss=0.05937, over 13463.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2351, pruned_loss=0.05625, over 2627989.64 frames. ], batch size: 100, lr: 5.24e-03, grad_scale: 32.0 2023-04-17 04:41:12,727 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65632.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:41:20,072 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65641.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:41:25,757 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5301, 2.6610, 2.9234, 4.1045, 3.7264, 4.1867, 3.1989, 2.7538], device='cuda:0'), covar=tensor([0.0336, 0.0785, 0.0779, 0.0063, 0.0217, 0.0052, 0.0691, 0.0807], device='cuda:0'), in_proj_covar=tensor([0.0122, 0.0148, 0.0165, 0.0101, 0.0119, 0.0098, 0.0168, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:41:55,391 INFO [train.py:893] (0/4) Epoch 25, batch 900, loss[loss=0.1644, simple_loss=0.212, pruned_loss=0.05842, over 13162.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2351, pruned_loss=0.05637, over 2634047.25 frames. ], batch size: 58, lr: 5.23e-03, grad_scale: 32.0 2023-04-17 04:41:56,380 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65680.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:42:15,156 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4827, 2.6560, 2.8264, 4.0206, 3.6445, 4.1038, 3.1933, 2.6979], device='cuda:0'), covar=tensor([0.0332, 0.0801, 0.0825, 0.0067, 0.0229, 0.0063, 0.0678, 0.0840], device='cuda:0'), in_proj_covar=tensor([0.0122, 0.0148, 0.0164, 0.0101, 0.0120, 0.0098, 0.0167, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:42:24,630 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65711.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:42:26,072 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-17 04:42:35,418 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.039e+02 2.809e+02 3.107e+02 3.712e+02 7.214e+02, threshold=6.214e+02, percent-clipped=4.0 2023-04-17 04:42:40,576 INFO [train.py:893] (0/4) Epoch 25, batch 950, loss[loss=0.1765, simple_loss=0.23, pruned_loss=0.06149, over 12014.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2351, pruned_loss=0.05694, over 2638958.85 frames. ], batch size: 158, lr: 5.23e-03, grad_scale: 32.0 2023-04-17 04:42:41,703 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0201, 1.8632, 3.7885, 3.7213, 3.7160, 2.9045, 3.4469, 2.8958], device='cuda:0'), covar=tensor([0.2193, 0.1698, 0.0170, 0.0236, 0.0243, 0.0831, 0.0287, 0.0851], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0186, 0.0130, 0.0136, 0.0137, 0.0177, 0.0148, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 04:42:59,675 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9249, 4.7924, 5.0272, 4.8724, 5.2716, 4.8071, 5.2453, 5.2335], device='cuda:0'), covar=tensor([0.0447, 0.0579, 0.0649, 0.0630, 0.0536, 0.0821, 0.0513, 0.0495], device='cuda:0'), in_proj_covar=tensor([0.0301, 0.0305, 0.0312, 0.0232, 0.0450, 0.0356, 0.0290, 0.0308], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:43:08,410 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65759.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:43:26,045 INFO [train.py:893] (0/4) Epoch 25, batch 1000, loss[loss=0.1754, simple_loss=0.2411, pruned_loss=0.05479, over 13423.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2336, pruned_loss=0.05628, over 2643239.02 frames. ], batch size: 95, lr: 5.23e-03, grad_scale: 32.0 2023-04-17 04:43:28,025 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7316, 3.4261, 4.2957, 3.0599, 3.0004, 2.9747, 4.5969, 4.6980], device='cuda:0'), covar=tensor([0.1166, 0.1815, 0.0358, 0.1766, 0.1451, 0.1573, 0.0274, 0.0231], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0279, 0.0203, 0.0228, 0.0223, 0.0185, 0.0218, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:43:34,503 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-17 04:44:00,582 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8097, 3.6295, 3.0018, 3.3959, 2.9676, 2.2034, 3.6846, 2.2266], device='cuda:0'), covar=tensor([0.0678, 0.0557, 0.0491, 0.0374, 0.0653, 0.1934, 0.1066, 0.1283], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0147, 0.0135, 0.0119, 0.0149, 0.0191, 0.0183, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:44:02,871 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3079, 4.7714, 4.5370, 4.5576, 4.5753, 4.3513, 4.8390, 4.8342], device='cuda:0'), covar=tensor([0.0233, 0.0217, 0.0198, 0.0358, 0.0262, 0.0271, 0.0277, 0.0213], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0206, 0.0169, 0.0186, 0.0157, 0.0206, 0.0137, 0.0186], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:44:06,782 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.851e+02 2.700e+02 3.165e+02 3.648e+02 7.276e+02, threshold=6.330e+02, percent-clipped=2.0 2023-04-17 04:44:06,991 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.0059, 4.5132, 4.3540, 4.4795, 4.2723, 4.3306, 4.9386, 4.5312], device='cuda:0'), covar=tensor([0.0701, 0.1119, 0.2215, 0.2439, 0.0914, 0.1481, 0.0897, 0.1096], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0402, 0.0498, 0.0498, 0.0315, 0.0370, 0.0460, 0.0367], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:44:11,775 INFO [train.py:893] (0/4) Epoch 25, batch 1050, loss[loss=0.1679, simple_loss=0.2297, pruned_loss=0.05306, over 13027.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2321, pruned_loss=0.05555, over 2644411.43 frames. ], batch size: 142, lr: 5.23e-03, grad_scale: 32.0 2023-04-17 04:44:20,883 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65838.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 04:44:42,162 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-17 04:44:52,157 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6188, 2.3315, 2.1271, 1.3748, 1.8620, 1.7298, 2.0453, 2.4607], device='cuda:0'), covar=tensor([0.0877, 0.0291, 0.0653, 0.1598, 0.0219, 0.0617, 0.0803, 0.0262], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0149, 0.0126, 0.0212, 0.0117, 0.0170, 0.0180, 0.0137], device='cuda:0'), out_proj_covar=tensor([1.2823e-04, 1.1113e-04, 9.8425e-05, 1.5736e-04, 8.4847e-05, 1.2873e-04, 1.3555e-04, 1.0158e-04], device='cuda:0') 2023-04-17 04:44:57,411 INFO [train.py:893] (0/4) Epoch 25, batch 1100, loss[loss=0.171, simple_loss=0.2317, pruned_loss=0.05513, over 13466.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2323, pruned_loss=0.05522, over 2648381.24 frames. ], batch size: 100, lr: 5.23e-03, grad_scale: 32.0 2023-04-17 04:45:05,536 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65888.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 04:45:06,044 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-17 04:45:15,504 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65899.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 04:45:37,547 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.961e+02 2.539e+02 2.910e+02 3.534e+02 6.326e+02, threshold=5.820e+02, percent-clipped=0.0 2023-04-17 04:45:43,068 INFO [train.py:893] (0/4) Epoch 25, batch 1150, loss[loss=0.1602, simple_loss=0.2194, pruned_loss=0.05047, over 13563.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2323, pruned_loss=0.05441, over 2656595.89 frames. ], batch size: 78, lr: 5.22e-03, grad_scale: 32.0 2023-04-17 04:45:54,865 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65941.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:46:17,066 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.47 vs. limit=5.0 2023-04-17 04:46:28,757 INFO [train.py:893] (0/4) Epoch 25, batch 1200, loss[loss=0.1542, simple_loss=0.2134, pruned_loss=0.0475, over 13529.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2322, pruned_loss=0.05405, over 2651614.80 frames. ], batch size: 74, lr: 5.22e-03, grad_scale: 32.0 2023-04-17 04:46:37,318 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65989.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:46:47,817 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-66000.pt 2023-04-17 04:46:58,614 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-17 04:47:04,538 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.6923, 5.1462, 5.1422, 5.2218, 5.0479, 5.0358, 5.6866, 5.2783], device='cuda:0'), covar=tensor([0.0747, 0.1205, 0.2199, 0.2420, 0.0789, 0.1633, 0.0798, 0.1234], device='cuda:0'), in_proj_covar=tensor([0.0349, 0.0399, 0.0492, 0.0495, 0.0311, 0.0367, 0.0457, 0.0366], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:47:09,391 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-17 04:47:11,542 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.650e+02 2.513e+02 2.947e+02 3.399e+02 7.667e+02, threshold=5.893e+02, percent-clipped=1.0 2023-04-17 04:47:17,291 INFO [train.py:893] (0/4) Epoch 25, batch 1250, loss[loss=0.1824, simple_loss=0.2426, pruned_loss=0.06108, over 13117.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2328, pruned_loss=0.05485, over 2653039.89 frames. ], batch size: 142, lr: 5.22e-03, grad_scale: 32.0 2023-04-17 04:48:02,015 INFO [train.py:893] (0/4) Epoch 25, batch 1300, loss[loss=0.1757, simple_loss=0.2322, pruned_loss=0.05961, over 13094.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2346, pruned_loss=0.05601, over 2654985.84 frames. ], batch size: 142, lr: 5.22e-03, grad_scale: 32.0 2023-04-17 04:48:41,326 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.956e+02 2.611e+02 3.159e+02 3.839e+02 8.053e+02, threshold=6.319e+02, percent-clipped=2.0 2023-04-17 04:48:43,193 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1246, 4.6218, 4.5082, 4.6382, 4.4003, 4.4765, 5.0729, 4.5936], device='cuda:0'), covar=tensor([0.0697, 0.1203, 0.2026, 0.2284, 0.0985, 0.1531, 0.0851, 0.1198], device='cuda:0'), in_proj_covar=tensor([0.0349, 0.0401, 0.0490, 0.0495, 0.0312, 0.0368, 0.0460, 0.0367], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:48:47,607 INFO [train.py:893] (0/4) Epoch 25, batch 1350, loss[loss=0.1812, simple_loss=0.2414, pruned_loss=0.06049, over 13272.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2352, pruned_loss=0.05611, over 2660518.44 frames. ], batch size: 124, lr: 5.22e-03, grad_scale: 64.0 2023-04-17 04:49:00,374 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66144.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:49:16,158 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66160.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 04:49:31,718 INFO [train.py:893] (0/4) Epoch 25, batch 1400, loss[loss=0.191, simple_loss=0.2645, pruned_loss=0.05873, over 13481.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2342, pruned_loss=0.05562, over 2661134.38 frames. ], batch size: 100, lr: 5.21e-03, grad_scale: 64.0 2023-04-17 04:49:37,782 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-17 04:49:40,657 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66188.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 04:49:46,793 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66194.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 04:49:56,193 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66205.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 04:50:03,331 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0644, 4.4826, 4.3024, 4.3108, 4.3907, 4.1074, 4.5473, 4.5553], device='cuda:0'), covar=tensor([0.0218, 0.0222, 0.0183, 0.0318, 0.0200, 0.0257, 0.0259, 0.0231], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0205, 0.0169, 0.0185, 0.0156, 0.0205, 0.0136, 0.0184], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003], device='cuda:0') 2023-04-17 04:50:11,200 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66221.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 04:50:13,202 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.900e+02 2.492e+02 2.974e+02 3.412e+02 5.608e+02, threshold=5.949e+02, percent-clipped=0.0 2023-04-17 04:50:17,977 INFO [train.py:893] (0/4) Epoch 25, batch 1450, loss[loss=0.1775, simple_loss=0.2464, pruned_loss=0.05429, over 13393.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2346, pruned_loss=0.05594, over 2663692.53 frames. ], batch size: 113, lr: 5.21e-03, grad_scale: 64.0 2023-04-17 04:50:23,929 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66236.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 04:50:44,692 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66258.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:50:48,016 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8657, 4.1438, 3.0672, 2.8079, 3.0109, 2.5295, 4.2741, 2.4428], device='cuda:0'), covar=tensor([0.1830, 0.0409, 0.1383, 0.2307, 0.0892, 0.3363, 0.0265, 0.4193], device='cuda:0'), in_proj_covar=tensor([0.0317, 0.0296, 0.0326, 0.0346, 0.0266, 0.0337, 0.0221, 0.0373], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 04:51:02,131 INFO [train.py:893] (0/4) Epoch 25, batch 1500, loss[loss=0.1928, simple_loss=0.2545, pruned_loss=0.06558, over 13352.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2342, pruned_loss=0.05544, over 2662309.15 frames. ], batch size: 118, lr: 5.21e-03, grad_scale: 64.0 2023-04-17 04:51:13,616 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6226, 3.5739, 2.6365, 3.1975, 3.6019, 2.3416, 3.3374, 2.5039], device='cuda:0'), covar=tensor([0.0292, 0.0220, 0.1108, 0.0462, 0.0263, 0.1318, 0.0511, 0.1461], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0182, 0.0179, 0.0229, 0.0141, 0.0163, 0.0163, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:51:40,330 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66319.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:51:43,269 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.769e+02 2.622e+02 3.018e+02 3.858e+02 6.130e+02, threshold=6.036e+02, percent-clipped=2.0 2023-04-17 04:51:48,104 INFO [train.py:893] (0/4) Epoch 25, batch 1550, loss[loss=0.1714, simple_loss=0.237, pruned_loss=0.05288, over 13457.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2344, pruned_loss=0.05548, over 2659929.32 frames. ], batch size: 103, lr: 5.21e-03, grad_scale: 64.0 2023-04-17 04:52:25,339 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.88 vs. limit=5.0 2023-04-17 04:52:33,397 INFO [train.py:893] (0/4) Epoch 25, batch 1600, loss[loss=0.1822, simple_loss=0.2464, pruned_loss=0.05904, over 13288.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2353, pruned_loss=0.05584, over 2657584.71 frames. ], batch size: 124, lr: 5.21e-03, grad_scale: 64.0 2023-04-17 04:52:50,093 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4314, 3.2060, 3.8943, 2.7546, 2.6261, 2.6738, 4.2039, 4.2913], device='cuda:0'), covar=tensor([0.1331, 0.1894, 0.0425, 0.1871, 0.1620, 0.1563, 0.0306, 0.0246], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0276, 0.0202, 0.0227, 0.0222, 0.0185, 0.0217, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:53:07,265 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-17 04:53:13,885 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.786e+02 2.679e+02 3.209e+02 3.685e+02 7.595e+02, threshold=6.418e+02, percent-clipped=1.0 2023-04-17 04:53:15,746 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.4261, 5.3632, 5.5471, 5.3200, 5.8002, 5.3793, 5.7932, 5.7895], device='cuda:0'), covar=tensor([0.0392, 0.0558, 0.0596, 0.0516, 0.0488, 0.0713, 0.0407, 0.0374], device='cuda:0'), in_proj_covar=tensor([0.0304, 0.0310, 0.0316, 0.0236, 0.0456, 0.0361, 0.0295, 0.0314], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 04:53:18,084 INFO [train.py:893] (0/4) Epoch 25, batch 1650, loss[loss=0.1718, simple_loss=0.239, pruned_loss=0.05232, over 13374.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2352, pruned_loss=0.0554, over 2657543.58 frames. ], batch size: 118, lr: 5.20e-03, grad_scale: 32.0 2023-04-17 04:53:34,138 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9597, 3.7949, 3.0405, 3.4826, 2.9549, 2.2773, 3.8388, 2.1784], device='cuda:0'), covar=tensor([0.0642, 0.0460, 0.0513, 0.0347, 0.0697, 0.1779, 0.0881, 0.1284], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0147, 0.0137, 0.0121, 0.0150, 0.0193, 0.0184, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:53:40,917 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-17 04:54:04,121 INFO [train.py:893] (0/4) Epoch 25, batch 1700, loss[loss=0.1621, simple_loss=0.2182, pruned_loss=0.05299, over 13348.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.235, pruned_loss=0.05529, over 2656236.57 frames. ], batch size: 62, lr: 5.20e-03, grad_scale: 32.0 2023-04-17 04:54:17,270 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66494.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 04:54:19,719 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0407, 3.9824, 2.9173, 3.7013, 3.1044, 2.2090, 4.0722, 2.2501], device='cuda:0'), covar=tensor([0.0669, 0.0357, 0.0619, 0.0286, 0.0638, 0.1864, 0.0658, 0.1267], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0148, 0.0137, 0.0121, 0.0151, 0.0193, 0.0184, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 04:54:22,909 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66500.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 04:54:37,378 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66516.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 04:54:41,427 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66521.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 04:54:44,295 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.534e+02 2.889e+02 3.612e+02 5.828e+02, threshold=5.778e+02, percent-clipped=0.0 2023-04-17 04:54:48,322 INFO [train.py:893] (0/4) Epoch 25, batch 1750, loss[loss=0.184, simple_loss=0.2449, pruned_loss=0.06153, over 13469.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2339, pruned_loss=0.05446, over 2660957.93 frames. ], batch size: 100, lr: 5.20e-03, grad_scale: 32.0 2023-04-17 04:55:01,241 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66542.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 04:55:33,743 INFO [train.py:893] (0/4) Epoch 25, batch 1800, loss[loss=0.1383, simple_loss=0.2018, pruned_loss=0.03746, over 13448.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2327, pruned_loss=0.05396, over 2660201.57 frames. ], batch size: 65, lr: 5.20e-03, grad_scale: 32.0 2023-04-17 04:55:36,532 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66582.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 04:56:06,293 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66614.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:56:15,765 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.962e+02 2.657e+02 3.005e+02 3.637e+02 5.123e+02, threshold=6.010e+02, percent-clipped=0.0 2023-04-17 04:56:21,314 INFO [train.py:893] (0/4) Epoch 25, batch 1850, loss[loss=0.1635, simple_loss=0.2078, pruned_loss=0.05963, over 12663.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2312, pruned_loss=0.05342, over 2653234.42 frames. ], batch size: 52, lr: 5.20e-03, grad_scale: 32.0 2023-04-17 04:56:23,007 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-17 04:57:05,933 INFO [train.py:893] (0/4) Epoch 25, batch 1900, loss[loss=0.1567, simple_loss=0.2194, pruned_loss=0.04704, over 13518.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.231, pruned_loss=0.05381, over 2651923.02 frames. ], batch size: 70, lr: 5.19e-03, grad_scale: 32.0 2023-04-17 04:57:20,821 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2023-04-17 04:57:38,496 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8412, 3.9019, 2.9367, 2.7760, 2.8496, 2.4933, 3.9951, 2.3043], device='cuda:0'), covar=tensor([0.1753, 0.0405, 0.1397, 0.2176, 0.0940, 0.3293, 0.0333, 0.4138], device='cuda:0'), in_proj_covar=tensor([0.0319, 0.0296, 0.0326, 0.0347, 0.0267, 0.0338, 0.0222, 0.0374], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 04:57:47,649 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.644e+02 2.632e+02 3.055e+02 3.564e+02 5.579e+02, threshold=6.111e+02, percent-clipped=0.0 2023-04-17 04:57:51,040 INFO [train.py:893] (0/4) Epoch 25, batch 1950, loss[loss=0.1592, simple_loss=0.2214, pruned_loss=0.04855, over 13482.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2305, pruned_loss=0.05326, over 2657912.52 frames. ], batch size: 81, lr: 5.19e-03, grad_scale: 16.0 2023-04-17 04:57:52,952 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66731.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 04:58:34,711 INFO [train.py:893] (0/4) Epoch 25, batch 2000, loss[loss=0.1712, simple_loss=0.229, pruned_loss=0.05672, over 13505.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2324, pruned_loss=0.05425, over 2660290.71 frames. ], batch size: 72, lr: 5.19e-03, grad_scale: 16.0 2023-04-17 04:58:39,711 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 04:58:47,180 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66792.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 04:58:53,988 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66800.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 04:59:09,375 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66816.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 04:59:17,059 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.867e+02 2.596e+02 3.110e+02 3.946e+02 7.233e+02, threshold=6.221e+02, percent-clipped=2.0 2023-04-17 04:59:20,408 INFO [train.py:893] (0/4) Epoch 25, batch 2050, loss[loss=0.1702, simple_loss=0.2362, pruned_loss=0.05206, over 13446.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2345, pruned_loss=0.05513, over 2664027.25 frames. ], batch size: 95, lr: 5.19e-03, grad_scale: 16.0 2023-04-17 04:59:37,917 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66848.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 04:59:52,369 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66864.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 05:00:03,527 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66877.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 05:00:04,897 INFO [train.py:893] (0/4) Epoch 25, batch 2100, loss[loss=0.1364, simple_loss=0.2032, pruned_loss=0.03483, over 13489.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2346, pruned_loss=0.05501, over 2656900.05 frames. ], batch size: 70, lr: 5.19e-03, grad_scale: 16.0 2023-04-17 05:00:37,498 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66914.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:00:46,285 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.756e+02 2.554e+02 2.938e+02 3.443e+02 6.083e+02, threshold=5.876e+02, percent-clipped=0.0 2023-04-17 05:00:50,278 INFO [train.py:893] (0/4) Epoch 25, batch 2150, loss[loss=0.1656, simple_loss=0.225, pruned_loss=0.05306, over 13517.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2344, pruned_loss=0.05472, over 2658542.79 frames. ], batch size: 70, lr: 5.18e-03, grad_scale: 16.0 2023-04-17 05:01:15,356 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66957.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:01:20,012 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66962.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:01:35,977 INFO [train.py:893] (0/4) Epoch 25, batch 2200, loss[loss=0.1526, simple_loss=0.2092, pruned_loss=0.04797, over 13374.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2327, pruned_loss=0.05366, over 2661500.04 frames. ], batch size: 67, lr: 5.18e-03, grad_scale: 16.0 2023-04-17 05:02:10,810 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67018.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 05:02:16,888 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.719e+02 2.330e+02 2.750e+02 3.208e+02 5.338e+02, threshold=5.499e+02, percent-clipped=0.0 2023-04-17 05:02:20,291 INFO [train.py:893] (0/4) Epoch 25, batch 2250, loss[loss=0.1556, simple_loss=0.2223, pruned_loss=0.04442, over 13245.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2314, pruned_loss=0.05335, over 2661569.07 frames. ], batch size: 124, lr: 5.18e-03, grad_scale: 16.0 2023-04-17 05:02:47,707 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8317, 2.7657, 2.4613, 1.7596, 1.8072, 2.2713, 2.4449, 3.0336], device='cuda:0'), covar=tensor([0.1070, 0.0479, 0.0745, 0.1806, 0.0379, 0.0755, 0.0857, 0.0229], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0150, 0.0125, 0.0210, 0.0117, 0.0170, 0.0179, 0.0137], device='cuda:0'), out_proj_covar=tensor([1.2790e-04, 1.1227e-04, 9.7338e-05, 1.5585e-04, 8.4265e-05, 1.2844e-04, 1.3523e-04, 1.0080e-04], device='cuda:0') 2023-04-17 05:02:48,585 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0675, 2.8882, 2.6297, 1.8509, 2.0201, 2.4292, 2.6679, 3.2015], device='cuda:0'), covar=tensor([0.0962, 0.0517, 0.0650, 0.1649, 0.0495, 0.0620, 0.0742, 0.0257], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0150, 0.0125, 0.0210, 0.0117, 0.0170, 0.0179, 0.0137], device='cuda:0'), out_proj_covar=tensor([1.2798e-04, 1.1233e-04, 9.7408e-05, 1.5597e-04, 8.4306e-05, 1.2852e-04, 1.3531e-04, 1.0087e-04], device='cuda:0') 2023-04-17 05:02:57,834 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1515, 4.6957, 4.4238, 4.4235, 4.5139, 4.3045, 4.6957, 4.7504], device='cuda:0'), covar=tensor([0.0245, 0.0228, 0.0219, 0.0357, 0.0281, 0.0270, 0.0291, 0.0218], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0210, 0.0173, 0.0187, 0.0158, 0.0208, 0.0138, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:03:05,924 INFO [train.py:893] (0/4) Epoch 25, batch 2300, loss[loss=0.1758, simple_loss=0.2413, pruned_loss=0.05521, over 13286.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.231, pruned_loss=0.05295, over 2663655.35 frames. ], batch size: 124, lr: 5.18e-03, grad_scale: 16.0 2023-04-17 05:03:10,374 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67083.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:03:13,625 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67087.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 05:03:47,499 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.297e+02 2.747e+02 3.196e+02 6.009e+02, threshold=5.494e+02, percent-clipped=2.0 2023-04-17 05:03:49,642 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-17 05:03:52,299 INFO [train.py:893] (0/4) Epoch 25, batch 2350, loss[loss=0.1534, simple_loss=0.2165, pruned_loss=0.04514, over 13539.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2304, pruned_loss=0.05264, over 2666768.62 frames. ], batch size: 72, lr: 5.18e-03, grad_scale: 16.0 2023-04-17 05:04:05,394 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67144.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:04:05,749 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-04-17 05:04:10,769 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 05:04:34,555 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67177.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 05:04:35,980 INFO [train.py:893] (0/4) Epoch 25, batch 2400, loss[loss=0.1488, simple_loss=0.2178, pruned_loss=0.03987, over 13474.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2303, pruned_loss=0.05267, over 2669404.96 frames. ], batch size: 79, lr: 5.17e-03, grad_scale: 16.0 2023-04-17 05:05:02,943 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-17 05:05:18,055 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.660e+02 2.350e+02 2.674e+02 3.216e+02 6.257e+02, threshold=5.348e+02, percent-clipped=1.0 2023-04-17 05:05:18,989 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67225.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 05:05:22,107 INFO [train.py:893] (0/4) Epoch 25, batch 2450, loss[loss=0.1676, simple_loss=0.2284, pruned_loss=0.05341, over 13369.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2308, pruned_loss=0.05304, over 2661487.81 frames. ], batch size: 62, lr: 5.17e-03, grad_scale: 16.0 2023-04-17 05:06:06,347 INFO [train.py:893] (0/4) Epoch 25, batch 2500, loss[loss=0.1314, simple_loss=0.1989, pruned_loss=0.0319, over 13513.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2307, pruned_loss=0.05306, over 2665185.03 frames. ], batch size: 70, lr: 5.17e-03, grad_scale: 16.0 2023-04-17 05:06:37,121 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67313.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 05:06:48,970 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.493e+02 2.922e+02 3.566e+02 6.370e+02, threshold=5.844e+02, percent-clipped=1.0 2023-04-17 05:06:52,190 INFO [train.py:893] (0/4) Epoch 25, batch 2550, loss[loss=0.1518, simple_loss=0.2138, pruned_loss=0.04492, over 11859.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2311, pruned_loss=0.05342, over 2662474.41 frames. ], batch size: 158, lr: 5.17e-03, grad_scale: 16.0 2023-04-17 05:07:13,763 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 05:07:37,528 INFO [train.py:893] (0/4) Epoch 25, batch 2600, loss[loss=0.154, simple_loss=0.2147, pruned_loss=0.04664, over 13346.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2306, pruned_loss=0.05349, over 2665856.48 frames. ], batch size: 62, lr: 5.17e-03, grad_scale: 16.0 2023-04-17 05:07:45,759 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67387.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 05:07:57,343 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-17 05:08:09,608 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0139, 2.9271, 2.6416, 1.8859, 2.1081, 2.5394, 2.7127, 3.1930], device='cuda:0'), covar=tensor([0.0972, 0.0396, 0.0757, 0.1460, 0.0400, 0.0484, 0.0687, 0.0238], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0151, 0.0126, 0.0212, 0.0118, 0.0172, 0.0181, 0.0139], device='cuda:0'), out_proj_covar=tensor([1.2886e-04, 1.1306e-04, 9.8065e-05, 1.5721e-04, 8.5046e-05, 1.2996e-04, 1.3632e-04, 1.0198e-04], device='cuda:0') 2023-04-17 05:08:15,385 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.933e+02 2.504e+02 2.907e+02 3.489e+02 6.172e+02, threshold=5.813e+02, percent-clipped=2.0 2023-04-17 05:08:18,517 INFO [train.py:893] (0/4) Epoch 25, batch 2650, loss[loss=0.1727, simple_loss=0.2305, pruned_loss=0.05746, over 13062.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2312, pruned_loss=0.05381, over 2658828.11 frames. ], batch size: 142, lr: 5.17e-03, grad_scale: 16.0 2023-04-17 05:08:23,096 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67435.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 05:08:26,009 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67439.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:08:30,898 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-17 05:08:56,474 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-25.pt 2023-04-17 05:09:20,969 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-17 05:09:24,735 INFO [train.py:893] (0/4) Epoch 26, batch 0, loss[loss=0.1794, simple_loss=0.2431, pruned_loss=0.05789, over 13481.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2431, pruned_loss=0.05789, over 13481.00 frames. ], batch size: 93, lr: 5.06e-03, grad_scale: 16.0 2023-04-17 05:09:24,736 INFO [train.py:918] (0/4) Computing validation loss 2023-04-17 05:09:40,950 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7369, 2.4389, 2.4288, 2.8589, 2.0756, 2.8337, 2.7906, 2.4268], device='cuda:0'), covar=tensor([0.0091, 0.0218, 0.0145, 0.0137, 0.0248, 0.0162, 0.0180, 0.0183], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0125, 0.0130, 0.0130, 0.0141, 0.0117, 0.0114, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 05:09:46,368 INFO [train.py:927] (0/4) Epoch 26, validation: loss=0.1333, simple_loss=0.1928, pruned_loss=0.03689, over 2446609.00 frames. 2023-04-17 05:09:46,368 INFO [train.py:928] (0/4) Maximum memory allocated so far is 13044MB 2023-04-17 05:10:19,072 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67513.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:10:30,052 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.826e+02 2.528e+02 2.949e+02 3.569e+02 5.747e+02, threshold=5.898e+02, percent-clipped=0.0 2023-04-17 05:10:33,221 INFO [train.py:893] (0/4) Epoch 26, batch 50, loss[loss=0.157, simple_loss=0.2164, pruned_loss=0.04878, over 13375.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2264, pruned_loss=0.05306, over 601225.08 frames. ], batch size: 73, lr: 5.06e-03, grad_scale: 16.0 2023-04-17 05:10:41,776 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5093, 2.0913, 3.9321, 3.7711, 3.8152, 3.0418, 3.5347, 3.0657], device='cuda:0'), covar=tensor([0.1688, 0.1479, 0.0156, 0.0223, 0.0207, 0.0717, 0.0276, 0.0700], device='cuda:0'), in_proj_covar=tensor([0.0218, 0.0184, 0.0130, 0.0135, 0.0137, 0.0177, 0.0149, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 05:10:55,619 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-17 05:10:55,620 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-17 05:10:55,620 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-17 05:10:55,636 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-17 05:10:55,645 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-17 05:10:55,667 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-17 05:10:55,676 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-17 05:11:14,459 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67574.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 05:11:18,287 INFO [train.py:893] (0/4) Epoch 26, batch 100, loss[loss=0.1816, simple_loss=0.2423, pruned_loss=0.06046, over 13464.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2296, pruned_loss=0.05581, over 1051077.18 frames. ], batch size: 103, lr: 5.06e-03, grad_scale: 16.0 2023-04-17 05:11:40,600 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3156, 4.8288, 4.7025, 4.7859, 4.5846, 4.6486, 5.2420, 4.8696], device='cuda:0'), covar=tensor([0.0639, 0.1152, 0.2051, 0.2209, 0.0987, 0.1605, 0.0897, 0.1071], device='cuda:0'), in_proj_covar=tensor([0.0345, 0.0398, 0.0483, 0.0487, 0.0308, 0.0364, 0.0451, 0.0362], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 05:11:50,610 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67613.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:12:02,529 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.938e+02 2.463e+02 2.907e+02 3.549e+02 6.282e+02, threshold=5.815e+02, percent-clipped=1.0 2023-04-17 05:12:05,235 INFO [train.py:893] (0/4) Epoch 26, batch 150, loss[loss=0.171, simple_loss=0.2215, pruned_loss=0.06029, over 13205.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2323, pruned_loss=0.05747, over 1397438.01 frames. ], batch size: 58, lr: 5.06e-03, grad_scale: 16.0 2023-04-17 05:12:35,577 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67661.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:12:47,130 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0471, 4.2009, 2.9205, 3.8384, 4.0683, 2.7441, 3.7059, 2.8290], device='cuda:0'), covar=tensor([0.0304, 0.0220, 0.0947, 0.0395, 0.0309, 0.1215, 0.0519, 0.1153], device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0184, 0.0181, 0.0230, 0.0143, 0.0163, 0.0166, 0.0163], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:12:50,116 INFO [train.py:893] (0/4) Epoch 26, batch 200, loss[loss=0.206, simple_loss=0.2586, pruned_loss=0.07673, over 13428.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2337, pruned_loss=0.0575, over 1670372.60 frames. ], batch size: 95, lr: 5.05e-03, grad_scale: 16.0 2023-04-17 05:12:52,101 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67680.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:12:55,294 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-17 05:13:03,897 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67692.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:13:33,557 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.787e+02 2.465e+02 2.845e+02 3.399e+02 5.580e+02, threshold=5.689e+02, percent-clipped=0.0 2023-04-17 05:13:36,052 INFO [train.py:893] (0/4) Epoch 26, batch 250, loss[loss=0.1726, simple_loss=0.2313, pruned_loss=0.05688, over 13350.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2346, pruned_loss=0.05719, over 1891649.97 frames. ], batch size: 73, lr: 5.05e-03, grad_scale: 16.0 2023-04-17 05:13:46,187 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67739.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:13:47,932 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67741.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:13:59,453 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67753.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 05:14:20,962 INFO [train.py:893] (0/4) Epoch 26, batch 300, loss[loss=0.1581, simple_loss=0.2117, pruned_loss=0.05224, over 13157.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2356, pruned_loss=0.05729, over 2065990.40 frames. ], batch size: 58, lr: 5.05e-03, grad_scale: 16.0 2023-04-17 05:14:29,765 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67787.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:14:31,117 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-17 05:15:00,129 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7671, 2.4859, 2.2033, 1.6364, 1.6872, 2.0313, 2.0877, 2.7023], device='cuda:0'), covar=tensor([0.0879, 0.0307, 0.0558, 0.1524, 0.0179, 0.0579, 0.0905, 0.0240], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0153, 0.0127, 0.0215, 0.0118, 0.0174, 0.0184, 0.0139], device='cuda:0'), out_proj_covar=tensor([1.2990e-04, 1.1403e-04, 9.8706e-05, 1.5949e-04, 8.5319e-05, 1.3143e-04, 1.3827e-04, 1.0218e-04], device='cuda:0') 2023-04-17 05:15:03,936 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.916e+02 2.664e+02 3.111e+02 3.733e+02 5.868e+02, threshold=6.222e+02, percent-clipped=2.0 2023-04-17 05:15:04,458 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-17 05:15:06,380 INFO [train.py:893] (0/4) Epoch 26, batch 350, loss[loss=0.1894, simple_loss=0.2487, pruned_loss=0.06509, over 13256.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2354, pruned_loss=0.05747, over 2197755.03 frames. ], batch size: 132, lr: 5.05e-03, grad_scale: 16.0 2023-04-17 05:15:39,178 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9753, 2.9527, 2.5997, 1.8959, 1.9296, 2.4544, 2.5143, 3.1561], device='cuda:0'), covar=tensor([0.0996, 0.0304, 0.0561, 0.1515, 0.0404, 0.0695, 0.0768, 0.0244], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0153, 0.0128, 0.0216, 0.0119, 0.0174, 0.0184, 0.0139], device='cuda:0'), out_proj_covar=tensor([1.3044e-04, 1.1462e-04, 9.9036e-05, 1.5972e-04, 8.5857e-05, 1.3171e-04, 1.3854e-04, 1.0240e-04], device='cuda:0') 2023-04-17 05:15:45,642 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67869.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 05:15:53,449 INFO [train.py:893] (0/4) Epoch 26, batch 400, loss[loss=0.1673, simple_loss=0.2295, pruned_loss=0.0526, over 13521.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2355, pruned_loss=0.05698, over 2301119.50 frames. ], batch size: 98, lr: 5.05e-03, grad_scale: 16.0 2023-04-17 05:16:36,538 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.742e+02 2.634e+02 2.859e+02 3.427e+02 7.730e+02, threshold=5.718e+02, percent-clipped=1.0 2023-04-17 05:16:38,925 INFO [train.py:893] (0/4) Epoch 26, batch 450, loss[loss=0.1713, simple_loss=0.2324, pruned_loss=0.05511, over 13463.00 frames. ], tot_loss[loss=0.175, simple_loss=0.236, pruned_loss=0.05699, over 2380438.77 frames. ], batch size: 100, lr: 5.04e-03, grad_scale: 16.0 2023-04-17 05:17:04,161 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-17 05:17:25,147 INFO [train.py:893] (0/4) Epoch 26, batch 500, loss[loss=0.165, simple_loss=0.2365, pruned_loss=0.0468, over 13438.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2351, pruned_loss=0.0562, over 2443575.87 frames. ], batch size: 106, lr: 5.04e-03, grad_scale: 16.0 2023-04-17 05:17:27,940 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67981.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:17:42,087 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-17 05:17:45,713 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-68000.pt 2023-04-17 05:18:12,049 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.734e+02 2.466e+02 2.871e+02 3.533e+02 1.025e+03, threshold=5.741e+02, percent-clipped=2.0 2023-04-17 05:18:14,574 INFO [train.py:893] (0/4) Epoch 26, batch 550, loss[loss=0.1934, simple_loss=0.247, pruned_loss=0.06991, over 11659.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2347, pruned_loss=0.05551, over 2493973.41 frames. ], batch size: 157, lr: 5.04e-03, grad_scale: 16.0 2023-04-17 05:18:22,728 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68036.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:18:27,795 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68042.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:18:32,565 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68048.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 05:18:38,099 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2960, 5.0586, 5.3812, 5.1018, 5.5981, 5.1083, 5.6233, 5.5888], device='cuda:0'), covar=tensor([0.0416, 0.0655, 0.0576, 0.0568, 0.0536, 0.0806, 0.0485, 0.0508], device='cuda:0'), in_proj_covar=tensor([0.0306, 0.0311, 0.0316, 0.0236, 0.0458, 0.0358, 0.0295, 0.0314], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:18:59,723 INFO [train.py:893] (0/4) Epoch 26, batch 600, loss[loss=0.168, simple_loss=0.2318, pruned_loss=0.05209, over 13527.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2334, pruned_loss=0.05529, over 2523210.27 frames. ], batch size: 91, lr: 5.04e-03, grad_scale: 16.0 2023-04-17 05:19:04,673 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9730, 4.3769, 4.1569, 4.2051, 4.1682, 4.0342, 4.4221, 4.4372], device='cuda:0'), covar=tensor([0.0243, 0.0230, 0.0253, 0.0311, 0.0310, 0.0277, 0.0249, 0.0250], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0208, 0.0172, 0.0185, 0.0158, 0.0208, 0.0139, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:19:07,983 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68087.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 05:19:15,217 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4823, 3.2991, 4.0351, 2.8603, 2.7110, 2.8394, 4.2672, 4.3981], device='cuda:0'), covar=tensor([0.1255, 0.1844, 0.0346, 0.1767, 0.1555, 0.1397, 0.0306, 0.0201], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0279, 0.0202, 0.0228, 0.0225, 0.0186, 0.0219, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 05:19:43,075 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.786e+02 2.523e+02 2.959e+02 3.513e+02 6.901e+02, threshold=5.918e+02, percent-clipped=1.0 2023-04-17 05:19:45,591 INFO [train.py:893] (0/4) Epoch 26, batch 650, loss[loss=0.1726, simple_loss=0.2354, pruned_loss=0.05488, over 13532.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2324, pruned_loss=0.05464, over 2554030.53 frames. ], batch size: 83, lr: 5.04e-03, grad_scale: 16.0 2023-04-17 05:20:02,188 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-17 05:20:03,539 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68148.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 05:20:12,282 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3069, 3.7009, 3.5247, 4.0481, 2.3208, 3.1182, 3.8285, 2.2549], device='cuda:0'), covar=tensor([0.0178, 0.0399, 0.0735, 0.0575, 0.1519, 0.0873, 0.0444, 0.1641], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0191, 0.0209, 0.0255, 0.0186, 0.0203, 0.0183, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:20:20,286 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0604, 2.7077, 2.6860, 3.0976, 2.4809, 3.1967, 3.0480, 2.6961], device='cuda:0'), covar=tensor([0.0084, 0.0193, 0.0158, 0.0168, 0.0244, 0.0131, 0.0185, 0.0180], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0125, 0.0130, 0.0131, 0.0141, 0.0118, 0.0115, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 05:20:22,622 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68169.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:20:30,599 INFO [train.py:893] (0/4) Epoch 26, batch 700, loss[loss=0.1806, simple_loss=0.2435, pruned_loss=0.05888, over 13410.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.231, pruned_loss=0.05376, over 2577062.00 frames. ], batch size: 113, lr: 5.04e-03, grad_scale: 16.0 2023-04-17 05:20:51,939 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.1548, 1.8809, 3.7867, 3.6537, 3.5702, 2.9505, 3.4781, 2.8696], device='cuda:0'), covar=tensor([0.1979, 0.1518, 0.0151, 0.0253, 0.0272, 0.0744, 0.0266, 0.0783], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0187, 0.0132, 0.0137, 0.0138, 0.0179, 0.0150, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 05:21:02,072 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.16 vs. limit=5.0 2023-04-17 05:21:06,346 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68217.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:21:13,389 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.625e+02 2.575e+02 3.034e+02 3.634e+02 6.033e+02, threshold=6.069e+02, percent-clipped=2.0 2023-04-17 05:21:15,950 INFO [train.py:893] (0/4) Epoch 26, batch 750, loss[loss=0.1807, simple_loss=0.2299, pruned_loss=0.0658, over 13455.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.231, pruned_loss=0.0544, over 2596808.50 frames. ], batch size: 65, lr: 5.03e-03, grad_scale: 16.0 2023-04-17 05:22:01,805 INFO [train.py:893] (0/4) Epoch 26, batch 800, loss[loss=0.1821, simple_loss=0.2451, pruned_loss=0.05953, over 13377.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.232, pruned_loss=0.05477, over 2608676.00 frames. ], batch size: 109, lr: 5.03e-03, grad_scale: 16.0 2023-04-17 05:22:40,220 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8222, 4.0869, 3.8268, 3.9099, 3.9880, 4.1826, 4.0491, 3.7261], device='cuda:0'), covar=tensor([0.0262, 0.0244, 0.0320, 0.0644, 0.0239, 0.0220, 0.0255, 0.0567], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0166, 0.0187, 0.0274, 0.0184, 0.0204, 0.0183, 0.0115], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 05:22:45,506 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.980e+02 2.741e+02 3.112e+02 3.676e+02 5.585e+02, threshold=6.224e+02, percent-clipped=0.0 2023-04-17 05:22:47,974 INFO [train.py:893] (0/4) Epoch 26, batch 850, loss[loss=0.17, simple_loss=0.2347, pruned_loss=0.05267, over 13545.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2331, pruned_loss=0.05537, over 2616250.07 frames. ], batch size: 87, lr: 5.03e-03, grad_scale: 16.0 2023-04-17 05:22:55,818 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68336.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:22:57,323 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68337.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:23:07,076 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68348.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:23:30,463 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3350, 3.6548, 3.5000, 4.0168, 2.3094, 3.1527, 3.8232, 2.2981], device='cuda:0'), covar=tensor([0.0140, 0.0445, 0.0776, 0.0622, 0.1538, 0.0873, 0.0468, 0.1648], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0192, 0.0209, 0.0255, 0.0186, 0.0203, 0.0183, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:23:34,141 INFO [train.py:893] (0/4) Epoch 26, batch 900, loss[loss=0.1655, simple_loss=0.2147, pruned_loss=0.05816, over 13380.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2338, pruned_loss=0.05589, over 2627438.76 frames. ], batch size: 65, lr: 5.03e-03, grad_scale: 16.0 2023-04-17 05:23:39,854 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68384.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:23:50,505 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68396.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:23:53,683 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68399.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:24:03,592 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-17 05:24:16,635 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.847e+02 2.678e+02 3.031e+02 3.663e+02 7.979e+02, threshold=6.063e+02, percent-clipped=2.0 2023-04-17 05:24:19,108 INFO [train.py:893] (0/4) Epoch 26, batch 950, loss[loss=0.1531, simple_loss=0.2114, pruned_loss=0.04745, over 13362.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2328, pruned_loss=0.05598, over 2633308.66 frames. ], batch size: 73, lr: 5.03e-03, grad_scale: 16.0 2023-04-17 05:24:33,743 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68443.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 05:24:48,772 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68460.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:25:05,288 INFO [train.py:893] (0/4) Epoch 26, batch 1000, loss[loss=0.1767, simple_loss=0.2296, pruned_loss=0.06188, over 13116.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2315, pruned_loss=0.05553, over 2641737.14 frames. ], batch size: 142, lr: 5.02e-03, grad_scale: 16.0 2023-04-17 05:25:48,339 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.804e+02 2.639e+02 2.912e+02 3.425e+02 6.549e+02, threshold=5.823e+02, percent-clipped=1.0 2023-04-17 05:25:51,458 INFO [train.py:893] (0/4) Epoch 26, batch 1050, loss[loss=0.1716, simple_loss=0.2327, pruned_loss=0.05523, over 13452.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2296, pruned_loss=0.05419, over 2647949.27 frames. ], batch size: 106, lr: 5.02e-03, grad_scale: 16.0 2023-04-17 05:25:59,803 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2085, 2.4244, 2.1789, 4.0772, 4.5276, 3.2738, 4.4477, 4.3084], device='cuda:0'), covar=tensor([0.0163, 0.1255, 0.1402, 0.0171, 0.0226, 0.0761, 0.0172, 0.0137], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0094, 0.0102, 0.0085, 0.0073, 0.0083, 0.0059, 0.0082], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 05:26:07,991 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0585, 3.6546, 3.1414, 3.4307, 3.0166, 2.3841, 3.8097, 2.2345], device='cuda:0'), covar=tensor([0.0585, 0.0727, 0.0422, 0.0371, 0.0638, 0.1809, 0.1053, 0.1268], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0149, 0.0137, 0.0122, 0.0153, 0.0194, 0.0187, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 05:26:35,626 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1570, 2.4664, 2.1350, 4.1073, 4.5585, 3.4524, 4.4263, 4.2772], device='cuda:0'), covar=tensor([0.0092, 0.0936, 0.1019, 0.0099, 0.0066, 0.0395, 0.0082, 0.0085], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0093, 0.0102, 0.0085, 0.0072, 0.0083, 0.0059, 0.0082], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 05:26:36,936 INFO [train.py:893] (0/4) Epoch 26, batch 1100, loss[loss=0.1829, simple_loss=0.24, pruned_loss=0.06287, over 13516.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2292, pruned_loss=0.05336, over 2654472.81 frames. ], batch size: 85, lr: 5.02e-03, grad_scale: 16.0 2023-04-17 05:26:37,235 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68578.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:26:43,326 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.05 vs. limit=5.0 2023-04-17 05:26:56,170 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-17 05:26:59,536 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-17 05:27:01,845 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8073, 4.6636, 4.8431, 4.8303, 5.1335, 4.6146, 5.1515, 5.0864], device='cuda:0'), covar=tensor([0.0408, 0.0619, 0.0640, 0.0485, 0.0498, 0.0735, 0.0418, 0.0428], device='cuda:0'), in_proj_covar=tensor([0.0307, 0.0309, 0.0316, 0.0234, 0.0453, 0.0358, 0.0295, 0.0312], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:27:20,083 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.731e+02 2.635e+02 3.013e+02 3.644e+02 5.346e+02, threshold=6.026e+02, percent-clipped=0.0 2023-04-17 05:27:22,552 INFO [train.py:893] (0/4) Epoch 26, batch 1150, loss[loss=0.1705, simple_loss=0.231, pruned_loss=0.05501, over 13507.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2298, pruned_loss=0.05297, over 2657613.68 frames. ], batch size: 91, lr: 5.02e-03, grad_scale: 16.0 2023-04-17 05:27:30,196 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68637.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:27:32,602 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68639.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:27:33,023 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-17 05:28:07,332 INFO [train.py:893] (0/4) Epoch 26, batch 1200, loss[loss=0.1474, simple_loss=0.2114, pruned_loss=0.04169, over 13559.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2293, pruned_loss=0.05243, over 2659113.79 frames. ], batch size: 76, lr: 5.02e-03, grad_scale: 16.0 2023-04-17 05:28:13,921 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68685.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:28:14,002 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68685.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:28:20,027 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-17 05:28:31,961 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-17 05:28:33,860 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3859, 2.2090, 2.6849, 3.8888, 3.4654, 3.8963, 2.8533, 2.2253], device='cuda:0'), covar=tensor([0.0313, 0.0982, 0.0819, 0.0077, 0.0246, 0.0083, 0.0829, 0.1061], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0150, 0.0167, 0.0104, 0.0121, 0.0100, 0.0171, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:28:44,139 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-17 05:28:50,386 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.795e+02 2.463e+02 2.772e+02 3.545e+02 7.103e+02, threshold=5.544e+02, percent-clipped=1.0 2023-04-17 05:28:52,809 INFO [train.py:893] (0/4) Epoch 26, batch 1250, loss[loss=0.1586, simple_loss=0.222, pruned_loss=0.0476, over 13354.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2303, pruned_loss=0.05294, over 2662292.75 frames. ], batch size: 67, lr: 5.02e-03, grad_scale: 32.0 2023-04-17 05:29:06,839 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68743.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 05:29:10,100 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68746.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:29:18,084 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68755.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:29:39,306 INFO [train.py:893] (0/4) Epoch 26, batch 1300, loss[loss=0.1743, simple_loss=0.2367, pruned_loss=0.05591, over 13541.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2318, pruned_loss=0.0537, over 2664601.52 frames. ], batch size: 83, lr: 5.01e-03, grad_scale: 32.0 2023-04-17 05:29:50,700 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68791.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 05:30:07,866 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9466, 2.0327, 2.4045, 3.3351, 3.0306, 3.3200, 2.6532, 2.1980], device='cuda:0'), covar=tensor([0.0309, 0.0834, 0.0695, 0.0080, 0.0267, 0.0091, 0.0650, 0.0829], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0148, 0.0165, 0.0103, 0.0120, 0.0099, 0.0171, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:30:22,442 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.892e+02 2.771e+02 3.228e+02 3.857e+02 7.338e+02, threshold=6.456e+02, percent-clipped=3.0 2023-04-17 05:30:24,977 INFO [train.py:893] (0/4) Epoch 26, batch 1350, loss[loss=0.1539, simple_loss=0.2193, pruned_loss=0.04429, over 13540.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2331, pruned_loss=0.05426, over 2667153.06 frames. ], batch size: 78, lr: 5.01e-03, grad_scale: 32.0 2023-04-17 05:31:11,700 INFO [train.py:893] (0/4) Epoch 26, batch 1400, loss[loss=0.1858, simple_loss=0.244, pruned_loss=0.06378, over 11702.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2325, pruned_loss=0.05438, over 2665975.89 frames. ], batch size: 157, lr: 5.01e-03, grad_scale: 32.0 2023-04-17 05:31:25,117 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.4180, 4.9359, 4.8695, 4.8988, 4.7994, 4.7611, 5.3621, 4.9150], device='cuda:0'), covar=tensor([0.0671, 0.1287, 0.2073, 0.2268, 0.1017, 0.1476, 0.0849, 0.1220], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0405, 0.0495, 0.0498, 0.0316, 0.0371, 0.0458, 0.0369], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 05:31:29,852 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8889, 4.1844, 3.9179, 4.0335, 4.0720, 4.3065, 4.1739, 3.8182], device='cuda:0'), covar=tensor([0.0319, 0.0247, 0.0330, 0.0701, 0.0258, 0.0230, 0.0284, 0.0506], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0164, 0.0185, 0.0271, 0.0184, 0.0202, 0.0182, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 05:31:54,581 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.787e+02 2.459e+02 2.883e+02 3.453e+02 7.124e+02, threshold=5.766e+02, percent-clipped=1.0 2023-04-17 05:31:57,800 INFO [train.py:893] (0/4) Epoch 26, batch 1450, loss[loss=0.1588, simple_loss=0.222, pruned_loss=0.04781, over 12074.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.232, pruned_loss=0.0541, over 2664041.14 frames. ], batch size: 157, lr: 5.01e-03, grad_scale: 32.0 2023-04-17 05:31:58,279 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-17 05:32:04,112 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68934.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:32:43,396 INFO [train.py:893] (0/4) Epoch 26, batch 1500, loss[loss=0.1483, simple_loss=0.2139, pruned_loss=0.04135, over 13530.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2316, pruned_loss=0.05333, over 2666292.68 frames. ], batch size: 72, lr: 5.01e-03, grad_scale: 32.0 2023-04-17 05:32:56,828 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8841, 2.3293, 1.9147, 3.8083, 4.1581, 3.1578, 4.1565, 3.9889], device='cuda:0'), covar=tensor([0.0084, 0.1015, 0.1043, 0.0084, 0.0076, 0.0481, 0.0062, 0.0073], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0093, 0.0102, 0.0085, 0.0072, 0.0083, 0.0059, 0.0082], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 05:33:25,079 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.553e+02 2.658e+02 3.045e+02 3.572e+02 6.026e+02, threshold=6.090e+02, percent-clipped=3.0 2023-04-17 05:33:28,347 INFO [train.py:893] (0/4) Epoch 26, batch 1550, loss[loss=0.1805, simple_loss=0.2333, pruned_loss=0.06385, over 13537.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2316, pruned_loss=0.05335, over 2666290.03 frames. ], batch size: 85, lr: 5.00e-03, grad_scale: 32.0 2023-04-17 05:33:39,925 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69041.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:33:41,677 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69043.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:33:52,275 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69055.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:34:12,711 INFO [train.py:893] (0/4) Epoch 26, batch 1600, loss[loss=0.1644, simple_loss=0.2349, pruned_loss=0.04695, over 13518.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2318, pruned_loss=0.05334, over 2667637.92 frames. ], batch size: 85, lr: 5.00e-03, grad_scale: 32.0 2023-04-17 05:34:34,855 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0036, 4.2281, 3.1522, 2.8362, 3.0503, 2.6154, 4.3334, 2.4440], device='cuda:0'), covar=tensor([0.1803, 0.0369, 0.1362, 0.2359, 0.0917, 0.3332, 0.0296, 0.4300], device='cuda:0'), in_proj_covar=tensor([0.0322, 0.0299, 0.0330, 0.0351, 0.0270, 0.0341, 0.0225, 0.0377], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 05:34:36,260 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69103.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:34:37,269 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69104.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:34:39,215 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-04-17 05:34:54,006 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1911, 4.5233, 4.1714, 4.2928, 4.2488, 4.6484, 4.4227, 4.2658], device='cuda:0'), covar=tensor([0.0281, 0.0251, 0.0319, 0.0811, 0.0272, 0.0227, 0.0279, 0.0331], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0165, 0.0186, 0.0272, 0.0185, 0.0202, 0.0183, 0.0114], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 05:34:56,246 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.041e+02 2.558e+02 2.916e+02 3.500e+02 7.035e+02, threshold=5.832e+02, percent-clipped=3.0 2023-04-17 05:34:58,827 INFO [train.py:893] (0/4) Epoch 26, batch 1650, loss[loss=0.1535, simple_loss=0.213, pruned_loss=0.04699, over 13555.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.232, pruned_loss=0.05257, over 2669659.67 frames. ], batch size: 72, lr: 5.00e-03, grad_scale: 32.0 2023-04-17 05:35:22,620 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9741, 4.2009, 3.1568, 2.8754, 3.0158, 2.6048, 4.3096, 2.4584], device='cuda:0'), covar=tensor([0.1836, 0.0338, 0.1268, 0.2290, 0.0904, 0.3300, 0.0263, 0.4018], device='cuda:0'), in_proj_covar=tensor([0.0321, 0.0297, 0.0328, 0.0349, 0.0268, 0.0339, 0.0223, 0.0375], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 05:35:30,528 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69162.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:35:35,519 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69168.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:35:43,443 INFO [train.py:893] (0/4) Epoch 26, batch 1700, loss[loss=0.1598, simple_loss=0.2335, pruned_loss=0.0431, over 13481.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2326, pruned_loss=0.05277, over 2670480.64 frames. ], batch size: 93, lr: 5.00e-03, grad_scale: 32.0 2023-04-17 05:36:02,276 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7624, 2.7655, 2.4534, 1.7527, 1.8045, 2.3554, 2.3721, 2.9672], device='cuda:0'), covar=tensor([0.1182, 0.0370, 0.0699, 0.1763, 0.0403, 0.0548, 0.0827, 0.0259], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0150, 0.0127, 0.0213, 0.0117, 0.0170, 0.0184, 0.0139], device='cuda:0'), out_proj_covar=tensor([1.3027e-04, 1.1172e-04, 9.8693e-05, 1.5779e-04, 8.4470e-05, 1.2878e-04, 1.3814e-04, 1.0179e-04], device='cuda:0') 2023-04-17 05:36:16,111 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3328, 4.8911, 4.7422, 4.8609, 4.7473, 4.7173, 5.2872, 4.8714], device='cuda:0'), covar=tensor([0.0702, 0.1086, 0.1997, 0.2425, 0.0768, 0.1441, 0.0835, 0.0996], device='cuda:0'), in_proj_covar=tensor([0.0350, 0.0401, 0.0490, 0.0493, 0.0313, 0.0367, 0.0456, 0.0367], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 05:36:26,055 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69223.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:36:27,384 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.818e+02 2.684e+02 3.102e+02 3.690e+02 9.152e+02, threshold=6.204e+02, percent-clipped=1.0 2023-04-17 05:36:27,735 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69225.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:36:29,896 INFO [train.py:893] (0/4) Epoch 26, batch 1750, loss[loss=0.1836, simple_loss=0.2397, pruned_loss=0.06378, over 13527.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2326, pruned_loss=0.05271, over 2667858.03 frames. ], batch size: 85, lr: 5.00e-03, grad_scale: 32.0 2023-04-17 05:36:31,066 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69229.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:36:34,926 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69234.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:36:45,518 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69245.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 05:37:10,397 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69273.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:37:15,601 INFO [train.py:893] (0/4) Epoch 26, batch 1800, loss[loss=0.1652, simple_loss=0.2337, pruned_loss=0.04839, over 13479.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2321, pruned_loss=0.05253, over 2668405.20 frames. ], batch size: 100, lr: 5.00e-03, grad_scale: 32.0 2023-04-17 05:37:18,955 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69282.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:37:23,085 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69286.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:37:41,060 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69306.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 05:37:53,207 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69320.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:37:56,844 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.794e+02 2.500e+02 2.995e+02 3.542e+02 6.193e+02, threshold=5.991e+02, percent-clipped=0.0 2023-04-17 05:37:59,396 INFO [train.py:893] (0/4) Epoch 26, batch 1850, loss[loss=0.1855, simple_loss=0.2438, pruned_loss=0.06362, over 13330.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2315, pruned_loss=0.05233, over 2667872.23 frames. ], batch size: 118, lr: 4.99e-03, grad_scale: 32.0 2023-04-17 05:38:02,698 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-17 05:38:04,507 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69334.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 05:38:10,825 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69341.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:38:35,938 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69368.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:38:44,612 INFO [train.py:893] (0/4) Epoch 26, batch 1900, loss[loss=0.1599, simple_loss=0.22, pruned_loss=0.04987, over 13525.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2305, pruned_loss=0.05241, over 2665251.20 frames. ], batch size: 91, lr: 4.99e-03, grad_scale: 32.0 2023-04-17 05:38:47,371 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69381.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:38:53,900 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69389.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:38:58,078 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69394.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:39:02,662 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69399.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:39:04,672 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8167, 4.6359, 4.8519, 4.7933, 5.1198, 4.5829, 5.1165, 5.0712], device='cuda:0'), covar=tensor([0.0408, 0.0576, 0.0640, 0.0482, 0.0525, 0.0838, 0.0510, 0.0522], device='cuda:0'), in_proj_covar=tensor([0.0312, 0.0315, 0.0323, 0.0241, 0.0464, 0.0365, 0.0304, 0.0320], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:39:13,283 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.58 vs. limit=5.0 2023-04-17 05:39:26,382 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.879e+02 2.463e+02 2.828e+02 3.585e+02 5.409e+02, threshold=5.655e+02, percent-clipped=0.0 2023-04-17 05:39:29,047 INFO [train.py:893] (0/4) Epoch 26, batch 1950, loss[loss=0.1866, simple_loss=0.2428, pruned_loss=0.06519, over 13531.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.23, pruned_loss=0.05256, over 2664387.58 frames. ], batch size: 89, lr: 4.99e-03, grad_scale: 32.0 2023-04-17 05:39:30,262 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69429.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:39:53,769 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69455.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:40:15,394 INFO [train.py:893] (0/4) Epoch 26, batch 2000, loss[loss=0.2167, simple_loss=0.2742, pruned_loss=0.07962, over 13014.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2322, pruned_loss=0.05363, over 2666152.31 frames. ], batch size: 142, lr: 4.99e-03, grad_scale: 32.0 2023-04-17 05:40:21,115 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 05:40:51,442 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69518.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:40:57,042 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69524.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:40:57,627 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 2.941e+02 3.266e+02 3.787e+02 7.734e+02, threshold=6.532e+02, percent-clipped=5.0 2023-04-17 05:41:00,881 INFO [train.py:893] (0/4) Epoch 26, batch 2050, loss[loss=0.1741, simple_loss=0.2431, pruned_loss=0.05252, over 13251.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.233, pruned_loss=0.05414, over 2668395.20 frames. ], batch size: 124, lr: 4.99e-03, grad_scale: 32.0 2023-04-17 05:41:07,627 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7359, 3.4752, 3.6702, 2.2757, 3.8129, 3.7287, 3.6784, 3.8984], device='cuda:0'), covar=tensor([0.0198, 0.0183, 0.0145, 0.1218, 0.0131, 0.0192, 0.0118, 0.0113], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0061, 0.0088, 0.0106, 0.0104, 0.0116, 0.0087, 0.0082], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:41:46,302 INFO [train.py:893] (0/4) Epoch 26, batch 2100, loss[loss=0.1594, simple_loss=0.2287, pruned_loss=0.04499, over 13433.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2327, pruned_loss=0.05383, over 2663909.76 frames. ], batch size: 106, lr: 4.98e-03, grad_scale: 32.0 2023-04-17 05:41:48,970 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69581.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:42:07,620 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69601.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 05:42:29,063 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 2.433e+02 2.898e+02 3.474e+02 5.673e+02, threshold=5.796e+02, percent-clipped=0.0 2023-04-17 05:42:32,239 INFO [train.py:893] (0/4) Epoch 26, batch 2150, loss[loss=0.1401, simple_loss=0.2119, pruned_loss=0.03414, over 13521.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.233, pruned_loss=0.0534, over 2663326.45 frames. ], batch size: 76, lr: 4.98e-03, grad_scale: 32.0 2023-04-17 05:42:33,157 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69629.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 05:43:12,832 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1107, 2.5339, 2.2717, 4.0381, 4.6088, 3.3283, 4.5260, 4.2441], device='cuda:0'), covar=tensor([0.0129, 0.1058, 0.1112, 0.0132, 0.0092, 0.0586, 0.0112, 0.0119], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0094, 0.0102, 0.0086, 0.0073, 0.0083, 0.0060, 0.0082], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 05:43:12,878 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69674.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 05:43:15,015 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69676.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:43:17,084 INFO [train.py:893] (0/4) Epoch 26, batch 2200, loss[loss=0.1606, simple_loss=0.2275, pruned_loss=0.04682, over 13367.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.232, pruned_loss=0.05278, over 2667374.90 frames. ], batch size: 73, lr: 4.98e-03, grad_scale: 32.0 2023-04-17 05:43:35,829 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69699.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:43:52,997 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.7596, 4.0692, 3.8105, 4.5431, 2.4950, 3.5495, 4.3286, 2.5946], device='cuda:0'), covar=tensor([0.0129, 0.0511, 0.0924, 0.0533, 0.1702, 0.0932, 0.0433, 0.1704], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0193, 0.0210, 0.0255, 0.0187, 0.0204, 0.0184, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:43:58,448 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69724.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:43:59,054 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.856e+02 2.542e+02 2.911e+02 3.538e+02 6.573e+02, threshold=5.821e+02, percent-clipped=1.0 2023-04-17 05:44:01,569 INFO [train.py:893] (0/4) Epoch 26, batch 2250, loss[loss=0.1752, simple_loss=0.2291, pruned_loss=0.06067, over 13528.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2309, pruned_loss=0.05277, over 2666457.30 frames. ], batch size: 87, lr: 4.98e-03, grad_scale: 32.0 2023-04-17 05:44:07,632 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69735.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 05:44:13,407 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.8158, 2.7568, 2.5616, 1.8675, 1.8335, 2.4310, 2.4950, 3.0566], device='cuda:0'), covar=tensor([0.1156, 0.0394, 0.0638, 0.1555, 0.0353, 0.0601, 0.0773, 0.0250], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0153, 0.0129, 0.0215, 0.0117, 0.0172, 0.0184, 0.0140], device='cuda:0'), out_proj_covar=tensor([1.3113e-04, 1.1407e-04, 9.9900e-05, 1.5894e-04, 8.4715e-05, 1.2981e-04, 1.3827e-04, 1.0256e-04], device='cuda:0') 2023-04-17 05:44:19,094 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69747.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:44:21,572 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69750.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:44:24,169 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3351, 4.1534, 4.2620, 2.7039, 4.7025, 4.3207, 4.3579, 4.5638], device='cuda:0'), covar=tensor([0.0252, 0.0139, 0.0149, 0.1067, 0.0139, 0.0271, 0.0177, 0.0131], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0061, 0.0088, 0.0107, 0.0104, 0.0116, 0.0087, 0.0082], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:44:46,023 INFO [train.py:893] (0/4) Epoch 26, batch 2300, loss[loss=0.1667, simple_loss=0.2295, pruned_loss=0.05197, over 13474.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2305, pruned_loss=0.05231, over 2660999.07 frames. ], batch size: 79, lr: 4.98e-03, grad_scale: 32.0 2023-04-17 05:45:23,344 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69818.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:45:29,011 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69824.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:45:29,587 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 2.373e+02 2.733e+02 3.115e+02 4.841e+02, threshold=5.467e+02, percent-clipped=0.0 2023-04-17 05:45:32,211 INFO [train.py:893] (0/4) Epoch 26, batch 2350, loss[loss=0.1648, simple_loss=0.2328, pruned_loss=0.04835, over 13491.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2302, pruned_loss=0.05214, over 2660599.11 frames. ], batch size: 93, lr: 4.98e-03, grad_scale: 32.0 2023-04-17 05:45:55,781 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 05:46:04,421 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9375, 2.7052, 2.4233, 1.8031, 1.7172, 2.3798, 2.4487, 2.9951], device='cuda:0'), covar=tensor([0.0899, 0.0431, 0.0663, 0.1603, 0.0259, 0.0669, 0.0793, 0.0267], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0154, 0.0130, 0.0215, 0.0117, 0.0172, 0.0185, 0.0141], device='cuda:0'), out_proj_covar=tensor([1.3110e-04, 1.1460e-04, 1.0040e-04, 1.5931e-04, 8.5005e-05, 1.3000e-04, 1.3882e-04, 1.0363e-04], device='cuda:0') 2023-04-17 05:46:07,466 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69866.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:46:12,977 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69872.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:46:18,618 INFO [train.py:893] (0/4) Epoch 26, batch 2400, loss[loss=0.1785, simple_loss=0.2377, pruned_loss=0.05962, over 13578.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2295, pruned_loss=0.05203, over 2660860.81 frames. ], batch size: 89, lr: 4.97e-03, grad_scale: 32.0 2023-04-17 05:46:21,971 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69881.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:46:38,677 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69901.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 05:46:56,676 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3626, 2.2196, 2.7134, 3.7839, 3.3931, 3.7936, 2.9871, 2.3192], device='cuda:0'), covar=tensor([0.0271, 0.0801, 0.0627, 0.0067, 0.0254, 0.0064, 0.0582, 0.0867], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0150, 0.0168, 0.0104, 0.0122, 0.0101, 0.0171, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:46:59,676 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.919e+02 2.426e+02 2.838e+02 3.396e+02 4.576e+02, threshold=5.675e+02, percent-clipped=0.0 2023-04-17 05:47:02,274 INFO [train.py:893] (0/4) Epoch 26, batch 2450, loss[loss=0.1591, simple_loss=0.2234, pruned_loss=0.04736, over 13365.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2304, pruned_loss=0.05248, over 2661408.80 frames. ], batch size: 118, lr: 4.97e-03, grad_scale: 32.0 2023-04-17 05:47:03,900 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69929.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:47:03,957 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69929.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 05:47:22,572 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69949.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 05:47:46,934 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69976.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:47:48,343 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69977.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:47:49,009 INFO [train.py:893] (0/4) Epoch 26, batch 2500, loss[loss=0.1643, simple_loss=0.2315, pruned_loss=0.04851, over 13544.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.23, pruned_loss=0.05217, over 2654889.24 frames. ], batch size: 89, lr: 4.97e-03, grad_scale: 32.0 2023-04-17 05:48:09,700 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-70000.pt 2023-04-17 05:48:35,301 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70024.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:48:35,415 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70024.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:48:35,949 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.846e+02 2.455e+02 2.831e+02 3.406e+02 8.435e+02, threshold=5.661e+02, percent-clipped=3.0 2023-04-17 05:48:38,454 INFO [train.py:893] (0/4) Epoch 26, batch 2550, loss[loss=0.1524, simple_loss=0.2179, pruned_loss=0.04349, over 13350.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2303, pruned_loss=0.05252, over 2657127.86 frames. ], batch size: 73, lr: 4.97e-03, grad_scale: 32.0 2023-04-17 05:48:40,996 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70030.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 05:48:58,024 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70050.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:48:58,708 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 05:49:18,617 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70072.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:49:22,319 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-17 05:49:23,309 INFO [train.py:893] (0/4) Epoch 26, batch 2600, loss[loss=0.1614, simple_loss=0.226, pruned_loss=0.04837, over 13530.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2304, pruned_loss=0.05243, over 2661301.00 frames. ], batch size: 76, lr: 4.97e-03, grad_scale: 32.0 2023-04-17 05:49:41,773 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70098.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:49:51,385 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70110.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:49:51,468 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8995, 4.0774, 3.1504, 2.8147, 2.8806, 2.5407, 4.1167, 2.4095], device='cuda:0'), covar=tensor([0.1849, 0.0377, 0.1359, 0.2352, 0.0933, 0.3401, 0.0299, 0.4174], device='cuda:0'), in_proj_covar=tensor([0.0320, 0.0297, 0.0328, 0.0350, 0.0268, 0.0339, 0.0223, 0.0376], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 05:50:02,224 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.576e+02 3.040e+02 3.615e+02 6.720e+02, threshold=6.080e+02, percent-clipped=2.0 2023-04-17 05:50:04,352 INFO [train.py:893] (0/4) Epoch 26, batch 2650, loss[loss=0.1663, simple_loss=0.2327, pruned_loss=0.05001, over 13520.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2308, pruned_loss=0.05298, over 2652034.92 frames. ], batch size: 98, lr: 4.96e-03, grad_scale: 32.0 2023-04-17 05:50:37,238 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=70171.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:50:42,297 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-26.pt 2023-04-17 05:51:06,872 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-17 05:51:10,716 INFO [train.py:893] (0/4) Epoch 27, batch 0, loss[loss=0.1583, simple_loss=0.2195, pruned_loss=0.04857, over 13524.00 frames. ], tot_loss[loss=0.1583, simple_loss=0.2195, pruned_loss=0.04857, over 13524.00 frames. ], batch size: 72, lr: 4.87e-03, grad_scale: 32.0 2023-04-17 05:51:10,717 INFO [train.py:918] (0/4) Computing validation loss 2023-04-17 05:51:17,054 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7324, 2.5958, 4.2269, 3.9293, 4.1129, 3.4293, 3.7999, 3.2752], device='cuda:0'), covar=tensor([0.1555, 0.1122, 0.0121, 0.0231, 0.0205, 0.0520, 0.0258, 0.0629], device='cuda:0'), in_proj_covar=tensor([0.0220, 0.0182, 0.0131, 0.0136, 0.0138, 0.0177, 0.0151, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 05:51:18,027 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0749, 4.1779, 3.9778, 4.1062, 4.1926, 4.4059, 4.2005, 4.4120], device='cuda:0'), covar=tensor([0.0250, 0.0283, 0.0325, 0.0673, 0.0271, 0.0221, 0.0278, 0.0108], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0167, 0.0187, 0.0273, 0.0187, 0.0203, 0.0184, 0.0114], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 05:51:25,659 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4755, 4.8528, 4.6183, 4.6504, 4.7525, 4.5916, 4.9020, 4.9344], device='cuda:0'), covar=tensor([0.0232, 0.0192, 0.0206, 0.0305, 0.0265, 0.0260, 0.0221, 0.0199], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0219, 0.0182, 0.0196, 0.0167, 0.0219, 0.0145, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 05:51:30,391 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8998, 3.3515, 3.3822, 3.4664, 2.3815, 3.0741, 3.4598, 2.1385], device='cuda:0'), covar=tensor([0.0137, 0.0574, 0.0590, 0.0362, 0.1369, 0.0834, 0.0572, 0.1693], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0194, 0.0211, 0.0257, 0.0188, 0.0205, 0.0184, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:51:33,435 INFO [train.py:927] (0/4) Epoch 27, validation: loss=0.1326, simple_loss=0.1923, pruned_loss=0.03639, over 2446609.00 frames. 2023-04-17 05:51:33,436 INFO [train.py:928] (0/4) Maximum memory allocated so far is 13044MB 2023-04-17 05:51:36,127 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2301, 2.4645, 2.1151, 4.1271, 4.5390, 3.3420, 4.4871, 4.2864], device='cuda:0'), covar=tensor([0.0143, 0.1256, 0.1500, 0.0167, 0.0204, 0.0682, 0.0164, 0.0131], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0094, 0.0103, 0.0086, 0.0073, 0.0084, 0.0060, 0.0083], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 05:52:14,333 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1034, 4.3573, 4.1351, 4.1535, 4.2042, 4.5373, 4.3472, 4.1217], device='cuda:0'), covar=tensor([0.0280, 0.0285, 0.0306, 0.0848, 0.0281, 0.0224, 0.0276, 0.0401], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0165, 0.0185, 0.0270, 0.0184, 0.0201, 0.0182, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 05:52:16,508 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.830e+02 2.471e+02 2.835e+02 3.361e+02 5.660e+02, threshold=5.669e+02, percent-clipped=0.0 2023-04-17 05:52:18,126 INFO [train.py:893] (0/4) Epoch 27, batch 50, loss[loss=0.168, simple_loss=0.2312, pruned_loss=0.05239, over 13384.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2234, pruned_loss=0.05151, over 604709.51 frames. ], batch size: 113, lr: 4.87e-03, grad_scale: 32.0 2023-04-17 05:52:41,784 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-17 05:52:41,784 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-17 05:52:41,785 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-17 05:52:41,799 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-17 05:52:41,810 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-17 05:52:42,542 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-17 05:52:42,561 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-17 05:52:49,302 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70260.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:52:54,223 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7777, 2.5155, 2.3251, 1.5895, 1.6870, 2.0937, 2.1571, 2.7580], device='cuda:0'), covar=tensor([0.0943, 0.0360, 0.0539, 0.1475, 0.0195, 0.0540, 0.0787, 0.0219], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0153, 0.0128, 0.0213, 0.0116, 0.0171, 0.0183, 0.0140], device='cuda:0'), out_proj_covar=tensor([1.3005e-04, 1.1437e-04, 9.9110e-05, 1.5755e-04, 8.3575e-05, 1.2933e-04, 1.3790e-04, 1.0296e-04], device='cuda:0') 2023-04-17 05:53:05,185 INFO [train.py:893] (0/4) Epoch 27, batch 100, loss[loss=0.1726, simple_loss=0.2392, pruned_loss=0.05301, over 13459.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2242, pruned_loss=0.05197, over 1056194.19 frames. ], batch size: 103, lr: 4.87e-03, grad_scale: 32.0 2023-04-17 05:53:44,637 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=70321.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:53:48,115 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.920e+02 2.593e+02 3.180e+02 3.805e+02 8.227e+02, threshold=6.361e+02, percent-clipped=5.0 2023-04-17 05:53:50,470 INFO [train.py:893] (0/4) Epoch 27, batch 150, loss[loss=0.185, simple_loss=0.247, pruned_loss=0.06157, over 13536.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2281, pruned_loss=0.05374, over 1416385.51 frames. ], batch size: 91, lr: 4.86e-03, grad_scale: 32.0 2023-04-17 05:53:53,196 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70330.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 05:54:07,332 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2023-04-17 05:54:25,055 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.7479, 2.4191, 2.1953, 1.4751, 1.7860, 1.9441, 2.0409, 2.5929], device='cuda:0'), covar=tensor([0.0798, 0.0308, 0.0546, 0.1490, 0.0170, 0.0548, 0.0717, 0.0220], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0154, 0.0130, 0.0216, 0.0117, 0.0173, 0.0185, 0.0141], device='cuda:0'), out_proj_covar=tensor([1.3144e-04, 1.1460e-04, 1.0029e-04, 1.5955e-04, 8.4308e-05, 1.3042e-04, 1.3904e-04, 1.0349e-04], device='cuda:0') 2023-04-17 05:54:36,130 INFO [train.py:893] (0/4) Epoch 27, batch 200, loss[loss=0.1818, simple_loss=0.2432, pruned_loss=0.06018, over 13182.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2312, pruned_loss=0.05504, over 1672009.78 frames. ], batch size: 132, lr: 4.86e-03, grad_scale: 32.0 2023-04-17 05:54:37,147 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70378.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 05:55:01,038 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-17 05:55:15,891 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1547, 2.7760, 3.4108, 2.6778, 2.4410, 2.5359, 3.7064, 3.7562], device='cuda:0'), covar=tensor([0.1193, 0.1999, 0.0389, 0.1592, 0.1573, 0.1513, 0.0333, 0.0183], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0283, 0.0204, 0.0230, 0.0225, 0.0188, 0.0221, 0.0194], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 05:55:19,534 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.767e+02 2.575e+02 3.029e+02 3.545e+02 5.639e+02, threshold=6.057e+02, percent-clipped=0.0 2023-04-17 05:55:21,193 INFO [train.py:893] (0/4) Epoch 27, batch 250, loss[loss=0.179, simple_loss=0.2426, pruned_loss=0.05771, over 13430.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2321, pruned_loss=0.05564, over 1881444.80 frames. ], batch size: 95, lr: 4.86e-03, grad_scale: 32.0 2023-04-17 05:55:26,794 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-17 05:55:57,162 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70466.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:56:06,778 INFO [train.py:893] (0/4) Epoch 27, batch 300, loss[loss=0.1783, simple_loss=0.2447, pruned_loss=0.05592, over 13489.00 frames. ], tot_loss[loss=0.171, simple_loss=0.232, pruned_loss=0.05499, over 2056726.38 frames. ], batch size: 85, lr: 4.86e-03, grad_scale: 32.0 2023-04-17 05:56:50,516 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.718e+02 2.582e+02 3.015e+02 3.678e+02 5.260e+02, threshold=6.030e+02, percent-clipped=0.0 2023-04-17 05:56:52,137 INFO [train.py:893] (0/4) Epoch 27, batch 350, loss[loss=0.1798, simple_loss=0.2256, pruned_loss=0.06699, over 13420.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2327, pruned_loss=0.05509, over 2190153.20 frames. ], batch size: 65, lr: 4.86e-03, grad_scale: 32.0 2023-04-17 05:57:01,426 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9058, 2.7095, 2.3423, 1.6773, 1.6768, 2.3073, 2.3631, 2.9413], device='cuda:0'), covar=tensor([0.0969, 0.0396, 0.0739, 0.1701, 0.0268, 0.0616, 0.0802, 0.0282], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0152, 0.0129, 0.0213, 0.0116, 0.0171, 0.0183, 0.0140], device='cuda:0'), out_proj_covar=tensor([1.3102e-04, 1.1366e-04, 9.9650e-05, 1.5740e-04, 8.3379e-05, 1.2951e-04, 1.3789e-04, 1.0294e-04], device='cuda:0') 2023-04-17 05:57:38,643 INFO [train.py:893] (0/4) Epoch 27, batch 400, loss[loss=0.1673, simple_loss=0.2324, pruned_loss=0.05112, over 13538.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2323, pruned_loss=0.05435, over 2296860.95 frames. ], batch size: 76, lr: 4.85e-03, grad_scale: 32.0 2023-04-17 05:58:15,013 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70616.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 05:58:22,822 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.774e+02 2.559e+02 3.057e+02 3.646e+02 7.469e+02, threshold=6.114e+02, percent-clipped=3.0 2023-04-17 05:58:24,514 INFO [train.py:893] (0/4) Epoch 27, batch 450, loss[loss=0.1406, simple_loss=0.2107, pruned_loss=0.03531, over 13541.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2341, pruned_loss=0.05491, over 2376884.57 frames. ], batch size: 78, lr: 4.85e-03, grad_scale: 32.0 2023-04-17 05:58:50,666 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-17 05:59:09,771 INFO [train.py:893] (0/4) Epoch 27, batch 500, loss[loss=0.2065, simple_loss=0.2641, pruned_loss=0.07441, over 13229.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2344, pruned_loss=0.05487, over 2442973.35 frames. ], batch size: 124, lr: 4.85e-03, grad_scale: 16.0 2023-04-17 05:59:11,159 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-04-17 05:59:14,342 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2023-04-17 05:59:27,576 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.5943, 2.4307, 2.3722, 2.7753, 1.9406, 2.8132, 2.7467, 2.1792], device='cuda:0'), covar=tensor([0.0101, 0.0229, 0.0185, 0.0154, 0.0284, 0.0149, 0.0168, 0.0250], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0125, 0.0131, 0.0130, 0.0140, 0.0118, 0.0115, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 05:59:37,275 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-17 05:59:52,936 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.766e+02 2.540e+02 3.001e+02 3.649e+02 1.046e+03, threshold=6.001e+02, percent-clipped=2.0 2023-04-17 05:59:53,806 INFO [train.py:893] (0/4) Epoch 27, batch 550, loss[loss=0.1671, simple_loss=0.2317, pruned_loss=0.05131, over 13454.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.235, pruned_loss=0.05512, over 2491065.86 frames. ], batch size: 103, lr: 4.85e-03, grad_scale: 16.0 2023-04-17 06:00:05,320 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-17 06:00:28,942 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70766.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:00:38,863 INFO [train.py:893] (0/4) Epoch 27, batch 600, loss[loss=0.1621, simple_loss=0.222, pruned_loss=0.05113, over 13540.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2346, pruned_loss=0.05565, over 2514343.20 frames. ], batch size: 72, lr: 4.85e-03, grad_scale: 16.0 2023-04-17 06:00:39,140 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8189, 2.2997, 1.8526, 3.7625, 4.1076, 3.1201, 4.1022, 3.9050], device='cuda:0'), covar=tensor([0.0100, 0.1038, 0.1190, 0.0099, 0.0076, 0.0498, 0.0084, 0.0099], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0094, 0.0102, 0.0086, 0.0073, 0.0083, 0.0059, 0.0082], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 06:00:47,108 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9829, 3.9416, 2.9749, 3.7012, 3.9573, 2.7925, 3.6507, 2.7527], device='cuda:0'), covar=tensor([0.0324, 0.0238, 0.1002, 0.0459, 0.0252, 0.1151, 0.0462, 0.1221], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0186, 0.0182, 0.0236, 0.0144, 0.0166, 0.0167, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:01:13,295 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70814.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:01:23,613 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.781e+02 2.508e+02 2.864e+02 3.369e+02 6.579e+02, threshold=5.729e+02, percent-clipped=1.0 2023-04-17 06:01:24,466 INFO [train.py:893] (0/4) Epoch 27, batch 650, loss[loss=0.1696, simple_loss=0.2329, pruned_loss=0.0532, over 13385.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2341, pruned_loss=0.05521, over 2543863.93 frames. ], batch size: 109, lr: 4.85e-03, grad_scale: 16.0 2023-04-17 06:01:43,441 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2103, 5.0526, 5.2832, 5.0480, 5.5566, 5.0813, 5.5380, 5.5045], device='cuda:0'), covar=tensor([0.0469, 0.0629, 0.0567, 0.0571, 0.0526, 0.0893, 0.0445, 0.0439], device='cuda:0'), in_proj_covar=tensor([0.0312, 0.0313, 0.0321, 0.0240, 0.0460, 0.0361, 0.0301, 0.0317], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:02:08,170 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-17 06:02:10,094 INFO [train.py:893] (0/4) Epoch 27, batch 700, loss[loss=0.1511, simple_loss=0.2087, pruned_loss=0.04671, over 13421.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2326, pruned_loss=0.0545, over 2567743.41 frames. ], batch size: 65, lr: 4.84e-03, grad_scale: 16.0 2023-04-17 06:02:38,523 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.1056, 2.0271, 3.5673, 3.4761, 3.4980, 2.8452, 3.3558, 2.6883], device='cuda:0'), covar=tensor([0.1850, 0.1430, 0.0268, 0.0247, 0.0251, 0.0723, 0.0288, 0.0825], device='cuda:0'), in_proj_covar=tensor([0.0223, 0.0186, 0.0134, 0.0139, 0.0140, 0.0181, 0.0153, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 06:02:45,343 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.95 vs. limit=5.0 2023-04-17 06:02:45,844 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70916.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:02:55,324 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.829e+02 2.619e+02 3.031e+02 3.545e+02 7.685e+02, threshold=6.063e+02, percent-clipped=3.0 2023-04-17 06:02:55,349 INFO [train.py:893] (0/4) Epoch 27, batch 750, loss[loss=0.1981, simple_loss=0.2457, pruned_loss=0.07525, over 11984.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2323, pruned_loss=0.05456, over 2584445.80 frames. ], batch size: 158, lr: 4.84e-03, grad_scale: 8.0 2023-04-17 06:03:03,683 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3002, 4.5773, 3.5183, 3.0899, 3.3915, 2.7825, 4.6459, 2.7263], device='cuda:0'), covar=tensor([0.1562, 0.0317, 0.1143, 0.2321, 0.0797, 0.3239, 0.0246, 0.3757], device='cuda:0'), in_proj_covar=tensor([0.0320, 0.0296, 0.0330, 0.0350, 0.0268, 0.0339, 0.0223, 0.0376], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 06:03:17,044 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-17 06:03:20,814 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8403, 4.0755, 3.0314, 2.7347, 2.8882, 2.4661, 4.1450, 2.3077], device='cuda:0'), covar=tensor([0.1815, 0.0391, 0.1450, 0.2461, 0.0978, 0.3593, 0.0301, 0.4729], device='cuda:0'), in_proj_covar=tensor([0.0320, 0.0297, 0.0331, 0.0350, 0.0269, 0.0339, 0.0223, 0.0377], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 06:03:27,946 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70964.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:03:39,868 INFO [train.py:893] (0/4) Epoch 27, batch 800, loss[loss=0.1706, simple_loss=0.2317, pruned_loss=0.0548, over 13364.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2329, pruned_loss=0.05442, over 2601414.27 frames. ], batch size: 73, lr: 4.84e-03, grad_scale: 8.0 2023-04-17 06:03:41,837 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4837, 3.7658, 3.6046, 4.2347, 2.3570, 3.2128, 4.0495, 2.2964], device='cuda:0'), covar=tensor([0.0179, 0.0526, 0.0817, 0.0624, 0.1657, 0.0889, 0.0434, 0.1678], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0194, 0.0210, 0.0257, 0.0189, 0.0204, 0.0183, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:03:53,128 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70992.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:03:57,065 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.5377, 5.0634, 4.9404, 5.0699, 4.9527, 4.9095, 5.4898, 5.0553], device='cuda:0'), covar=tensor([0.0581, 0.1237, 0.2085, 0.2175, 0.0887, 0.1516, 0.0759, 0.1139], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0404, 0.0493, 0.0491, 0.0317, 0.0371, 0.0461, 0.0367], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 06:04:24,210 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 2.677e+02 3.139e+02 3.587e+02 7.209e+02, threshold=6.278e+02, percent-clipped=2.0 2023-04-17 06:04:24,235 INFO [train.py:893] (0/4) Epoch 27, batch 850, loss[loss=0.1744, simple_loss=0.2378, pruned_loss=0.05547, over 13527.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2343, pruned_loss=0.05531, over 2612362.82 frames. ], batch size: 87, lr: 4.84e-03, grad_scale: 8.0 2023-04-17 06:04:27,738 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71030.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:04:48,526 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71053.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:05:09,351 INFO [train.py:893] (0/4) Epoch 27, batch 900, loss[loss=0.1723, simple_loss=0.2335, pruned_loss=0.05551, over 13354.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2342, pruned_loss=0.05561, over 2625153.22 frames. ], batch size: 109, lr: 4.84e-03, grad_scale: 8.0 2023-04-17 06:05:22,971 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71091.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:05:38,417 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-17 06:05:55,246 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.070e+02 2.531e+02 3.075e+02 3.695e+02 7.717e+02, threshold=6.149e+02, percent-clipped=1.0 2023-04-17 06:05:55,271 INFO [train.py:893] (0/4) Epoch 27, batch 950, loss[loss=0.1701, simple_loss=0.227, pruned_loss=0.05658, over 13131.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2332, pruned_loss=0.05557, over 2635653.95 frames. ], batch size: 142, lr: 4.84e-03, grad_scale: 8.0 2023-04-17 06:06:12,089 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.98 vs. limit=5.0 2023-04-17 06:06:41,762 INFO [train.py:893] (0/4) Epoch 27, batch 1000, loss[loss=0.1582, simple_loss=0.2205, pruned_loss=0.04789, over 13440.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2313, pruned_loss=0.05465, over 2644017.32 frames. ], batch size: 106, lr: 4.83e-03, grad_scale: 8.0 2023-04-17 06:07:26,798 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.962e+02 2.549e+02 2.975e+02 3.461e+02 6.038e+02, threshold=5.950e+02, percent-clipped=0.0 2023-04-17 06:07:26,823 INFO [train.py:893] (0/4) Epoch 27, batch 1050, loss[loss=0.1785, simple_loss=0.2446, pruned_loss=0.05622, over 13531.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2299, pruned_loss=0.05347, over 2650944.09 frames. ], batch size: 83, lr: 4.83e-03, grad_scale: 8.0 2023-04-17 06:08:12,367 INFO [train.py:893] (0/4) Epoch 27, batch 1100, loss[loss=0.1634, simple_loss=0.2259, pruned_loss=0.05044, over 13209.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2292, pruned_loss=0.05282, over 2652640.64 frames. ], batch size: 124, lr: 4.83e-03, grad_scale: 8.0 2023-04-17 06:08:18,276 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8005, 3.9611, 2.7710, 3.5045, 3.8611, 2.5023, 3.4355, 2.8005], device='cuda:0'), covar=tensor([0.0334, 0.0326, 0.1060, 0.0521, 0.0298, 0.1348, 0.0563, 0.1123], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0184, 0.0179, 0.0232, 0.0142, 0.0163, 0.0163, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:08:57,474 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.923e+02 2.567e+02 2.971e+02 3.412e+02 6.882e+02, threshold=5.943e+02, percent-clipped=1.0 2023-04-17 06:08:57,499 INFO [train.py:893] (0/4) Epoch 27, batch 1150, loss[loss=0.1959, simple_loss=0.2503, pruned_loss=0.07069, over 13438.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2296, pruned_loss=0.05241, over 2656221.86 frames. ], batch size: 95, lr: 4.83e-03, grad_scale: 8.0 2023-04-17 06:09:16,409 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71348.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:09:42,215 INFO [train.py:893] (0/4) Epoch 27, batch 1200, loss[loss=0.1569, simple_loss=0.2237, pruned_loss=0.04507, over 13567.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2299, pruned_loss=0.05218, over 2657848.79 frames. ], batch size: 78, lr: 4.83e-03, grad_scale: 8.0 2023-04-17 06:09:50,449 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71386.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:10:09,320 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-17 06:10:20,560 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-17 06:10:28,585 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.042e+02 2.662e+02 2.989e+02 3.476e+02 6.704e+02, threshold=5.979e+02, percent-clipped=2.0 2023-04-17 06:10:28,612 INFO [train.py:893] (0/4) Epoch 27, batch 1250, loss[loss=0.1397, simple_loss=0.2065, pruned_loss=0.03646, over 13506.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2305, pruned_loss=0.05244, over 2657110.82 frames. ], batch size: 70, lr: 4.83e-03, grad_scale: 8.0 2023-04-17 06:10:37,983 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4840, 3.2613, 3.9388, 2.8448, 2.6544, 2.7307, 4.2559, 4.3325], device='cuda:0'), covar=tensor([0.1253, 0.1841, 0.0382, 0.1801, 0.1659, 0.1702, 0.0311, 0.0288], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0283, 0.0205, 0.0230, 0.0226, 0.0191, 0.0223, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:10:40,336 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2074, 4.5550, 4.2330, 4.3438, 4.3005, 4.6888, 4.4400, 4.4283], device='cuda:0'), covar=tensor([0.0287, 0.0275, 0.0342, 0.0802, 0.0310, 0.0212, 0.0326, 0.0224], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0167, 0.0188, 0.0273, 0.0186, 0.0204, 0.0185, 0.0115], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 06:10:59,939 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6671, 3.3638, 3.2045, 4.7497, 5.1578, 4.0052, 5.0458, 4.7951], device='cuda:0'), covar=tensor([0.0099, 0.0625, 0.0681, 0.0083, 0.0067, 0.0321, 0.0065, 0.0084], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0094, 0.0102, 0.0086, 0.0073, 0.0084, 0.0060, 0.0083], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 06:11:12,989 INFO [train.py:893] (0/4) Epoch 27, batch 1300, loss[loss=0.2005, simple_loss=0.2561, pruned_loss=0.0724, over 13528.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2313, pruned_loss=0.0527, over 2653719.62 frames. ], batch size: 83, lr: 4.82e-03, grad_scale: 8.0 2023-04-17 06:11:19,422 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-04-17 06:11:58,684 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.824e+02 2.626e+02 3.134e+02 3.874e+02 8.782e+02, threshold=6.268e+02, percent-clipped=3.0 2023-04-17 06:11:58,709 INFO [train.py:893] (0/4) Epoch 27, batch 1350, loss[loss=0.16, simple_loss=0.2223, pruned_loss=0.04885, over 13361.00 frames. ], tot_loss[loss=0.169, simple_loss=0.232, pruned_loss=0.05304, over 2654263.54 frames. ], batch size: 67, lr: 4.82e-03, grad_scale: 8.0 2023-04-17 06:12:25,810 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3530, 4.8380, 4.5445, 4.6274, 4.6182, 4.4351, 4.9333, 4.9340], device='cuda:0'), covar=tensor([0.0228, 0.0212, 0.0234, 0.0312, 0.0257, 0.0274, 0.0232, 0.0195], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0217, 0.0181, 0.0195, 0.0165, 0.0218, 0.0143, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 06:12:33,242 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9024, 4.0527, 3.1164, 2.8310, 2.9188, 2.5540, 4.1531, 2.3591], device='cuda:0'), covar=tensor([0.1873, 0.0390, 0.1347, 0.2338, 0.0873, 0.3251, 0.0301, 0.4175], device='cuda:0'), in_proj_covar=tensor([0.0322, 0.0298, 0.0333, 0.0354, 0.0269, 0.0342, 0.0225, 0.0379], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 06:12:34,939 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-17 06:12:43,422 INFO [train.py:893] (0/4) Epoch 27, batch 1400, loss[loss=0.1614, simple_loss=0.2228, pruned_loss=0.04998, over 13504.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2317, pruned_loss=0.05278, over 2661072.87 frames. ], batch size: 81, lr: 4.82e-03, grad_scale: 8.0 2023-04-17 06:12:49,310 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6202, 3.4036, 3.6238, 2.3085, 3.7799, 3.7176, 3.7035, 3.9008], device='cuda:0'), covar=tensor([0.0260, 0.0206, 0.0159, 0.1254, 0.0172, 0.0224, 0.0137, 0.0122], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0061, 0.0088, 0.0106, 0.0104, 0.0116, 0.0087, 0.0083], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:13:20,711 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71616.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:13:30,174 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.758e+02 2.491e+02 2.947e+02 3.481e+02 7.529e+02, threshold=5.893e+02, percent-clipped=1.0 2023-04-17 06:13:30,198 INFO [train.py:893] (0/4) Epoch 27, batch 1450, loss[loss=0.1701, simple_loss=0.2324, pruned_loss=0.05391, over 13527.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2312, pruned_loss=0.05293, over 2660779.38 frames. ], batch size: 91, lr: 4.82e-03, grad_scale: 8.0 2023-04-17 06:13:49,941 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=71648.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:14:14,954 INFO [train.py:893] (0/4) Epoch 27, batch 1500, loss[loss=0.1988, simple_loss=0.2519, pruned_loss=0.07285, over 13405.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2316, pruned_loss=0.05309, over 2657863.93 frames. ], batch size: 113, lr: 4.82e-03, grad_scale: 8.0 2023-04-17 06:14:15,296 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71677.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:14:23,948 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=71686.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:14:32,754 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=71696.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:14:50,542 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0007, 3.7603, 3.9358, 2.2943, 4.2116, 4.0418, 4.0124, 4.2223], device='cuda:0'), covar=tensor([0.0232, 0.0166, 0.0145, 0.1209, 0.0135, 0.0214, 0.0130, 0.0091], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0061, 0.0088, 0.0106, 0.0104, 0.0116, 0.0087, 0.0083], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:15:00,684 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.808e+02 2.588e+02 3.144e+02 3.766e+02 5.385e+02, threshold=6.289e+02, percent-clipped=0.0 2023-04-17 06:15:00,708 INFO [train.py:893] (0/4) Epoch 27, batch 1550, loss[loss=0.133, simple_loss=0.1958, pruned_loss=0.03512, over 13425.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2305, pruned_loss=0.0528, over 2651688.83 frames. ], batch size: 65, lr: 4.82e-03, grad_scale: 8.0 2023-04-17 06:15:06,435 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=71734.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:15:43,196 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1370, 2.8108, 3.3320, 2.6947, 2.4211, 2.4752, 3.7183, 3.7437], device='cuda:0'), covar=tensor([0.1239, 0.2118, 0.0403, 0.1647, 0.1583, 0.1614, 0.0323, 0.0290], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0281, 0.0204, 0.0228, 0.0225, 0.0189, 0.0221, 0.0194], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:15:46,401 INFO [train.py:893] (0/4) Epoch 27, batch 1600, loss[loss=0.1742, simple_loss=0.2432, pruned_loss=0.05262, over 13450.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.231, pruned_loss=0.05305, over 2646802.15 frames. ], batch size: 106, lr: 4.81e-03, grad_scale: 8.0 2023-04-17 06:16:07,350 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1586, 4.6689, 4.4608, 4.6354, 4.4036, 4.4993, 5.1019, 4.7554], device='cuda:0'), covar=tensor([0.0739, 0.1294, 0.2394, 0.2682, 0.1012, 0.1657, 0.0952, 0.1283], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0404, 0.0495, 0.0494, 0.0316, 0.0370, 0.0463, 0.0370], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 06:16:31,536 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.806e+02 2.588e+02 3.045e+02 3.762e+02 5.857e+02, threshold=6.090e+02, percent-clipped=0.0 2023-04-17 06:16:31,560 INFO [train.py:893] (0/4) Epoch 27, batch 1650, loss[loss=0.1694, simple_loss=0.2413, pruned_loss=0.04875, over 13512.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2312, pruned_loss=0.05252, over 2650714.39 frames. ], batch size: 85, lr: 4.81e-03, grad_scale: 8.0 2023-04-17 06:17:04,947 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.77 vs. limit=5.0 2023-04-17 06:17:16,696 INFO [train.py:893] (0/4) Epoch 27, batch 1700, loss[loss=0.1726, simple_loss=0.236, pruned_loss=0.0546, over 13476.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.232, pruned_loss=0.05247, over 2653697.67 frames. ], batch size: 100, lr: 4.81e-03, grad_scale: 8.0 2023-04-17 06:17:54,147 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71919.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:18:02,449 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.742e+02 2.543e+02 2.961e+02 3.560e+02 7.883e+02, threshold=5.923e+02, percent-clipped=1.0 2023-04-17 06:18:02,473 INFO [train.py:893] (0/4) Epoch 27, batch 1750, loss[loss=0.1546, simple_loss=0.2161, pruned_loss=0.04655, over 13540.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2305, pruned_loss=0.0517, over 2658213.46 frames. ], batch size: 87, lr: 4.81e-03, grad_scale: 8.0 2023-04-17 06:18:43,473 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71972.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:18:47,443 INFO [train.py:893] (0/4) Epoch 27, batch 1800, loss[loss=0.1572, simple_loss=0.2126, pruned_loss=0.05089, over 13342.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2299, pruned_loss=0.05134, over 2660697.51 frames. ], batch size: 62, lr: 4.81e-03, grad_scale: 8.0 2023-04-17 06:18:50,237 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71980.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:19:08,780 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-72000.pt 2023-04-17 06:19:28,445 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3354, 2.6044, 2.2603, 4.2768, 4.7154, 3.5605, 4.6072, 4.4220], device='cuda:0'), covar=tensor([0.0103, 0.0914, 0.1004, 0.0102, 0.0078, 0.0431, 0.0091, 0.0089], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0093, 0.0101, 0.0085, 0.0072, 0.0082, 0.0059, 0.0082], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 06:19:37,156 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.833e+02 2.380e+02 2.800e+02 3.278e+02 8.080e+02, threshold=5.600e+02, percent-clipped=1.0 2023-04-17 06:19:37,180 INFO [train.py:893] (0/4) Epoch 27, batch 1850, loss[loss=0.164, simple_loss=0.2254, pruned_loss=0.05125, over 13548.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2294, pruned_loss=0.05108, over 2660645.33 frames. ], batch size: 87, lr: 4.81e-03, grad_scale: 8.0 2023-04-17 06:19:40,588 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-17 06:19:43,211 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1789, 4.0285, 4.0616, 2.5554, 4.4297, 4.2679, 4.1672, 4.4111], device='cuda:0'), covar=tensor([0.0239, 0.0156, 0.0146, 0.1060, 0.0145, 0.0220, 0.0148, 0.0098], device='cuda:0'), in_proj_covar=tensor([0.0100, 0.0062, 0.0089, 0.0106, 0.0104, 0.0117, 0.0088, 0.0083], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:20:21,889 INFO [train.py:893] (0/4) Epoch 27, batch 1900, loss[loss=0.1619, simple_loss=0.2258, pruned_loss=0.04906, over 13534.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2287, pruned_loss=0.05116, over 2663320.98 frames. ], batch size: 83, lr: 4.80e-03, grad_scale: 8.0 2023-04-17 06:21:07,738 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.840e+02 2.498e+02 3.028e+02 3.455e+02 6.047e+02, threshold=6.055e+02, percent-clipped=2.0 2023-04-17 06:21:07,763 INFO [train.py:893] (0/4) Epoch 27, batch 1950, loss[loss=0.1823, simple_loss=0.2463, pruned_loss=0.05917, over 13446.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2281, pruned_loss=0.05084, over 2662888.80 frames. ], batch size: 103, lr: 4.80e-03, grad_scale: 8.0 2023-04-17 06:21:10,403 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72129.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:21:53,870 INFO [train.py:893] (0/4) Epoch 27, batch 2000, loss[loss=0.1985, simple_loss=0.2544, pruned_loss=0.07133, over 13049.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2302, pruned_loss=0.05188, over 2659228.29 frames. ], batch size: 142, lr: 4.80e-03, grad_scale: 8.0 2023-04-17 06:22:00,308 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 06:22:05,601 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72190.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:22:39,173 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.946e+02 2.675e+02 3.139e+02 3.825e+02 7.364e+02, threshold=6.278e+02, percent-clipped=1.0 2023-04-17 06:22:39,198 INFO [train.py:893] (0/4) Epoch 27, batch 2050, loss[loss=0.1907, simple_loss=0.2542, pruned_loss=0.06364, over 13344.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2318, pruned_loss=0.05284, over 2657155.69 frames. ], batch size: 118, lr: 4.80e-03, grad_scale: 8.0 2023-04-17 06:23:19,851 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72272.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:23:22,923 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72275.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:23:24,347 INFO [train.py:893] (0/4) Epoch 27, batch 2100, loss[loss=0.1641, simple_loss=0.2335, pruned_loss=0.04739, over 13533.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2313, pruned_loss=0.05267, over 2658266.38 frames. ], batch size: 98, lr: 4.80e-03, grad_scale: 8.0 2023-04-17 06:24:03,250 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=72320.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:24:08,326 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72326.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:24:08,538 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.69 vs. limit=5.0 2023-04-17 06:24:08,861 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.635e+02 2.231e+02 2.747e+02 3.271e+02 6.169e+02, threshold=5.495e+02, percent-clipped=0.0 2023-04-17 06:24:08,885 INFO [train.py:893] (0/4) Epoch 27, batch 2150, loss[loss=0.1631, simple_loss=0.2309, pruned_loss=0.04763, over 13395.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2309, pruned_loss=0.05186, over 2660556.33 frames. ], batch size: 113, lr: 4.80e-03, grad_scale: 8.0 2023-04-17 06:24:12,327 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4385, 4.8559, 4.6618, 4.6514, 4.7272, 4.4469, 4.9399, 4.9601], device='cuda:0'), covar=tensor([0.0205, 0.0222, 0.0220, 0.0297, 0.0251, 0.0239, 0.0250, 0.0191], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0216, 0.0180, 0.0193, 0.0164, 0.0216, 0.0144, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 06:24:54,754 INFO [train.py:893] (0/4) Epoch 27, batch 2200, loss[loss=0.1869, simple_loss=0.2478, pruned_loss=0.06299, over 13533.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2306, pruned_loss=0.0518, over 2651968.37 frames. ], batch size: 91, lr: 4.79e-03, grad_scale: 8.0 2023-04-17 06:25:01,208 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.80 vs. limit=5.0 2023-04-17 06:25:03,434 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72387.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:25:11,333 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0438, 2.0217, 2.3631, 3.2717, 3.0023, 3.3123, 2.6399, 2.2320], device='cuda:0'), covar=tensor([0.0281, 0.0840, 0.0733, 0.0099, 0.0306, 0.0088, 0.0648, 0.0844], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0152, 0.0171, 0.0106, 0.0126, 0.0103, 0.0173, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:25:36,921 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8533, 3.8940, 2.6812, 3.4679, 3.8286, 2.3817, 3.4964, 2.5806], device='cuda:0'), covar=tensor([0.0303, 0.0267, 0.1174, 0.0426, 0.0276, 0.1407, 0.0515, 0.1368], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0186, 0.0180, 0.0235, 0.0143, 0.0164, 0.0165, 0.0163], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:25:39,703 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.840e+02 2.441e+02 2.891e+02 3.478e+02 5.512e+02, threshold=5.782e+02, percent-clipped=2.0 2023-04-17 06:25:39,728 INFO [train.py:893] (0/4) Epoch 27, batch 2250, loss[loss=0.1683, simple_loss=0.2315, pruned_loss=0.05252, over 13365.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.229, pruned_loss=0.05131, over 2650117.90 frames. ], batch size: 109, lr: 4.79e-03, grad_scale: 8.0 2023-04-17 06:25:46,552 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.38 vs. limit=5.0 2023-04-17 06:26:25,345 INFO [train.py:893] (0/4) Epoch 27, batch 2300, loss[loss=0.1694, simple_loss=0.2408, pruned_loss=0.04898, over 13355.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2284, pruned_loss=0.05114, over 2651872.25 frames. ], batch size: 118, lr: 4.79e-03, grad_scale: 8.0 2023-04-17 06:26:32,143 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72485.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:26:42,636 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0360, 2.7069, 2.6675, 3.1628, 2.3805, 3.2260, 3.1674, 2.6365], device='cuda:0'), covar=tensor([0.0087, 0.0265, 0.0154, 0.0160, 0.0262, 0.0130, 0.0145, 0.0193], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0128, 0.0133, 0.0134, 0.0143, 0.0121, 0.0115, 0.0119], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 06:27:11,482 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.738e+02 2.357e+02 2.755e+02 3.330e+02 4.557e+02, threshold=5.510e+02, percent-clipped=0.0 2023-04-17 06:27:11,506 INFO [train.py:893] (0/4) Epoch 27, batch 2350, loss[loss=0.1667, simple_loss=0.2236, pruned_loss=0.05485, over 11939.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2285, pruned_loss=0.05081, over 2657798.79 frames. ], batch size: 157, lr: 4.79e-03, grad_scale: 8.0 2023-04-17 06:27:11,834 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7120, 3.8709, 2.7943, 3.6072, 3.7718, 2.5490, 3.3195, 2.8790], device='cuda:0'), covar=tensor([0.0405, 0.0407, 0.0944, 0.0388, 0.0323, 0.1163, 0.0693, 0.0956], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0187, 0.0181, 0.0236, 0.0144, 0.0165, 0.0166, 0.0163], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:27:14,150 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3311, 4.8375, 4.7271, 4.8529, 4.6596, 4.7281, 5.3247, 4.8790], device='cuda:0'), covar=tensor([0.0673, 0.1217, 0.2162, 0.2396, 0.0907, 0.1633, 0.0755, 0.1012], device='cuda:0'), in_proj_covar=tensor([0.0353, 0.0405, 0.0494, 0.0496, 0.0317, 0.0372, 0.0461, 0.0371], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 06:27:16,772 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5061, 4.2314, 3.4562, 3.9803, 3.4260, 2.8103, 4.1918, 2.7126], device='cuda:0'), covar=tensor([0.0494, 0.0319, 0.0483, 0.0224, 0.0647, 0.1505, 0.0806, 0.1137], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0149, 0.0140, 0.0120, 0.0151, 0.0193, 0.0188, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 06:27:27,891 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-17 06:27:32,833 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 06:27:37,341 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-17 06:27:49,129 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3144, 3.0490, 3.6846, 2.7516, 2.5389, 2.6146, 4.0838, 4.1327], device='cuda:0'), covar=tensor([0.1231, 0.2112, 0.0396, 0.1788, 0.1672, 0.1612, 0.0323, 0.0255], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0283, 0.0206, 0.0229, 0.0227, 0.0190, 0.0221, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:27:54,824 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72575.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:27:56,315 INFO [train.py:893] (0/4) Epoch 27, batch 2400, loss[loss=0.1472, simple_loss=0.2096, pruned_loss=0.04245, over 13346.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2279, pruned_loss=0.05056, over 2661356.68 frames. ], batch size: 73, lr: 4.79e-03, grad_scale: 8.0 2023-04-17 06:27:57,550 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4547, 3.2629, 3.8706, 2.8466, 2.6127, 2.7507, 4.2348, 4.2781], device='cuda:0'), covar=tensor([0.1233, 0.1784, 0.0406, 0.1816, 0.1710, 0.1581, 0.0290, 0.0259], device='cuda:0'), in_proj_covar=tensor([0.0257, 0.0283, 0.0207, 0.0230, 0.0227, 0.0190, 0.0222, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:28:16,914 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4066, 3.1292, 3.7403, 2.8488, 2.5549, 2.6963, 4.1444, 4.2533], device='cuda:0'), covar=tensor([0.1233, 0.1896, 0.0461, 0.1708, 0.1634, 0.1619, 0.0333, 0.0260], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0283, 0.0206, 0.0229, 0.0227, 0.0189, 0.0221, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:28:39,772 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=72623.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:28:42,936 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.028e+02 2.421e+02 2.989e+02 3.536e+02 6.414e+02, threshold=5.978e+02, percent-clipped=1.0 2023-04-17 06:28:42,972 INFO [train.py:893] (0/4) Epoch 27, batch 2450, loss[loss=0.187, simple_loss=0.2533, pruned_loss=0.06035, over 13380.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2281, pruned_loss=0.0508, over 2661041.23 frames. ], batch size: 118, lr: 4.79e-03, grad_scale: 8.0 2023-04-17 06:28:49,137 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-17 06:29:27,627 INFO [train.py:893] (0/4) Epoch 27, batch 2500, loss[loss=0.1575, simple_loss=0.2238, pruned_loss=0.04557, over 13378.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2281, pruned_loss=0.0508, over 2652951.93 frames. ], batch size: 67, lr: 4.78e-03, grad_scale: 8.0 2023-04-17 06:29:33,420 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72682.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:29:34,363 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7191, 3.4963, 2.7845, 3.1151, 2.9316, 2.2586, 3.5567, 2.0292], device='cuda:0'), covar=tensor([0.0692, 0.0540, 0.0547, 0.0471, 0.0673, 0.1897, 0.1068, 0.1363], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0150, 0.0140, 0.0121, 0.0152, 0.0194, 0.0189, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 06:29:59,246 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-17 06:30:15,077 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.688e+02 2.583e+02 2.896e+02 3.437e+02 6.200e+02, threshold=5.793e+02, percent-clipped=1.0 2023-04-17 06:30:15,101 INFO [train.py:893] (0/4) Epoch 27, batch 2550, loss[loss=0.17, simple_loss=0.2402, pruned_loss=0.04986, over 13414.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2278, pruned_loss=0.05084, over 2654793.81 frames. ], batch size: 95, lr: 4.78e-03, grad_scale: 8.0 2023-04-17 06:30:16,326 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0505, 2.4170, 1.8865, 4.0097, 4.4128, 3.2437, 4.3505, 4.0902], device='cuda:0'), covar=tensor([0.0103, 0.1031, 0.1155, 0.0104, 0.0080, 0.0515, 0.0079, 0.0098], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0093, 0.0101, 0.0086, 0.0073, 0.0083, 0.0060, 0.0083], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 06:30:20,579 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6213, 3.9810, 3.7823, 4.3626, 2.5731, 3.3463, 4.1955, 2.4428], device='cuda:0'), covar=tensor([0.0152, 0.0498, 0.0769, 0.0558, 0.1550, 0.0970, 0.0379, 0.1788], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0191, 0.0210, 0.0256, 0.0187, 0.0206, 0.0183, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:30:38,922 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 06:30:53,396 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-17 06:31:00,071 INFO [train.py:893] (0/4) Epoch 27, batch 2600, loss[loss=0.163, simple_loss=0.217, pruned_loss=0.05453, over 12726.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2278, pruned_loss=0.05075, over 2655715.13 frames. ], batch size: 52, lr: 4.78e-03, grad_scale: 8.0 2023-04-17 06:31:07,129 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72785.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:31:14,533 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6283, 3.9592, 3.7881, 4.3730, 2.4693, 3.3185, 4.1978, 2.4410], device='cuda:0'), covar=tensor([0.0164, 0.0490, 0.0723, 0.0513, 0.1585, 0.0906, 0.0378, 0.1588], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0191, 0.0209, 0.0255, 0.0187, 0.0205, 0.0182, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:31:40,335 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.031e+02 2.584e+02 3.071e+02 3.664e+02 5.210e+02, threshold=6.141e+02, percent-clipped=0.0 2023-04-17 06:31:40,360 INFO [train.py:893] (0/4) Epoch 27, batch 2650, loss[loss=0.1525, simple_loss=0.2138, pruned_loss=0.04564, over 13333.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2294, pruned_loss=0.05214, over 2651907.38 frames. ], batch size: 67, lr: 4.78e-03, grad_scale: 8.0 2023-04-17 06:31:45,037 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0355, 4.3856, 3.2981, 2.9511, 3.1276, 2.6184, 4.4790, 2.6015], device='cuda:0'), covar=tensor([0.1833, 0.0350, 0.1387, 0.2330, 0.0878, 0.3559, 0.0256, 0.4066], device='cuda:0'), in_proj_covar=tensor([0.0319, 0.0298, 0.0330, 0.0353, 0.0269, 0.0341, 0.0223, 0.0377], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 06:31:45,642 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=72833.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:32:19,233 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-27.pt 2023-04-17 06:32:43,523 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-17 06:32:47,347 INFO [train.py:893] (0/4) Epoch 28, batch 0, loss[loss=0.1423, simple_loss=0.2082, pruned_loss=0.03824, over 13381.00 frames. ], tot_loss[loss=0.1423, simple_loss=0.2082, pruned_loss=0.03824, over 13381.00 frames. ], batch size: 77, lr: 4.69e-03, grad_scale: 8.0 2023-04-17 06:32:47,348 INFO [train.py:918] (0/4) Computing validation loss 2023-04-17 06:32:58,855 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2426, 5.1881, 5.2518, 5.0879, 5.5665, 5.1193, 5.5416, 5.5530], device='cuda:0'), covar=tensor([0.0348, 0.0516, 0.0625, 0.0557, 0.0449, 0.0747, 0.0418, 0.0389], device='cuda:0'), in_proj_covar=tensor([0.0313, 0.0311, 0.0323, 0.0242, 0.0462, 0.0363, 0.0303, 0.0319], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:33:09,318 INFO [train.py:927] (0/4) Epoch 28, validation: loss=0.1317, simple_loss=0.1914, pruned_loss=0.03593, over 2446609.00 frames. 2023-04-17 06:33:09,319 INFO [train.py:928] (0/4) Maximum memory allocated so far is 13044MB 2023-04-17 06:33:15,543 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.31 vs. limit=5.0 2023-04-17 06:33:27,822 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72894.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:33:56,892 INFO [train.py:893] (0/4) Epoch 28, batch 50, loss[loss=0.1701, simple_loss=0.2358, pruned_loss=0.05217, over 13559.00 frames. ], tot_loss[loss=0.1583, simple_loss=0.2199, pruned_loss=0.04838, over 596900.77 frames. ], batch size: 78, lr: 4.69e-03, grad_scale: 16.0 2023-04-17 06:33:57,691 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.34 vs. limit=2.0 2023-04-17 06:33:57,747 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.022e+02 2.571e+02 2.970e+02 3.456e+02 5.243e+02, threshold=5.941e+02, percent-clipped=0.0 2023-04-17 06:34:19,633 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-17 06:34:19,634 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-17 06:34:19,634 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-17 06:34:19,653 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-17 06:34:19,664 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-17 06:34:20,420 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-17 06:34:20,430 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-17 06:34:22,276 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72955.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:34:30,504 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72964.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 06:34:39,457 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1533, 3.8843, 4.0215, 2.5342, 4.3673, 4.1866, 4.1528, 4.3372], device='cuda:0'), covar=tensor([0.0243, 0.0173, 0.0148, 0.1120, 0.0142, 0.0235, 0.0136, 0.0103], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0062, 0.0089, 0.0107, 0.0105, 0.0117, 0.0087, 0.0084], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:34:39,529 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7941, 2.4521, 2.4936, 2.9257, 2.1139, 2.9051, 2.8068, 2.2624], device='cuda:0'), covar=tensor([0.0088, 0.0227, 0.0173, 0.0153, 0.0267, 0.0145, 0.0181, 0.0228], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0129, 0.0134, 0.0134, 0.0144, 0.0121, 0.0117, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 06:34:40,020 INFO [train.py:893] (0/4) Epoch 28, batch 100, loss[loss=0.1681, simple_loss=0.2318, pruned_loss=0.05216, over 13419.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2249, pruned_loss=0.05158, over 1051013.14 frames. ], batch size: 95, lr: 4.69e-03, grad_scale: 16.0 2023-04-17 06:34:46,101 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72982.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:35:27,387 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73025.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 06:35:27,879 INFO [train.py:893] (0/4) Epoch 28, batch 150, loss[loss=0.171, simple_loss=0.2333, pruned_loss=0.05434, over 13378.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2275, pruned_loss=0.05283, over 1410507.35 frames. ], batch size: 73, lr: 4.69e-03, grad_scale: 16.0 2023-04-17 06:35:28,682 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.685e+02 2.652e+02 3.088e+02 3.819e+02 5.892e+02, threshold=6.175e+02, percent-clipped=0.0 2023-04-17 06:35:31,400 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73030.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:36:12,589 INFO [train.py:893] (0/4) Epoch 28, batch 200, loss[loss=0.1524, simple_loss=0.2143, pruned_loss=0.0453, over 13525.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2298, pruned_loss=0.05398, over 1685163.20 frames. ], batch size: 76, lr: 4.68e-03, grad_scale: 16.0 2023-04-17 06:36:19,371 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3443, 2.1438, 2.6056, 3.6792, 3.3678, 3.7590, 2.9235, 2.2320], device='cuda:0'), covar=tensor([0.0290, 0.0824, 0.0852, 0.0088, 0.0266, 0.0070, 0.0651, 0.0959], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0152, 0.0172, 0.0108, 0.0127, 0.0104, 0.0175, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:36:34,864 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2208, 2.1749, 2.6253, 3.5179, 3.2079, 3.6055, 2.9158, 2.2921], device='cuda:0'), covar=tensor([0.0327, 0.0835, 0.0769, 0.0093, 0.0306, 0.0078, 0.0610, 0.0894], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0152, 0.0172, 0.0108, 0.0127, 0.0104, 0.0175, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:36:46,613 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3547, 3.1226, 3.7749, 2.6237, 2.3799, 2.5800, 4.1122, 4.1953], device='cuda:0'), covar=tensor([0.1191, 0.1669, 0.0396, 0.1923, 0.1743, 0.1653, 0.0269, 0.0187], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0283, 0.0206, 0.0230, 0.0226, 0.0189, 0.0220, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:36:59,293 INFO [train.py:893] (0/4) Epoch 28, batch 250, loss[loss=0.1743, simple_loss=0.2314, pruned_loss=0.05864, over 13437.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.231, pruned_loss=0.05503, over 1900625.19 frames. ], batch size: 106, lr: 4.68e-03, grad_scale: 16.0 2023-04-17 06:37:00,081 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.971e+02 2.642e+02 3.255e+02 4.009e+02 7.837e+02, threshold=6.511e+02, percent-clipped=3.0 2023-04-17 06:37:15,709 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-17 06:37:43,143 INFO [train.py:893] (0/4) Epoch 28, batch 300, loss[loss=0.1857, simple_loss=0.2437, pruned_loss=0.06379, over 13319.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2307, pruned_loss=0.05433, over 2069012.40 frames. ], batch size: 118, lr: 4.68e-03, grad_scale: 16.0 2023-04-17 06:38:16,227 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9059, 2.7974, 2.3755, 1.7942, 1.7972, 2.3429, 2.5094, 3.0197], device='cuda:0'), covar=tensor([0.1128, 0.0389, 0.0746, 0.1706, 0.0384, 0.0613, 0.0760, 0.0297], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0156, 0.0134, 0.0218, 0.0120, 0.0175, 0.0185, 0.0143], device='cuda:0'), out_proj_covar=tensor([1.3375e-04, 1.1649e-04, 1.0294e-04, 1.6155e-04, 8.6271e-05, 1.3224e-04, 1.3948e-04, 1.0520e-04], device='cuda:0') 2023-04-17 06:38:29,132 INFO [train.py:893] (0/4) Epoch 28, batch 350, loss[loss=0.1527, simple_loss=0.2213, pruned_loss=0.04199, over 13532.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2313, pruned_loss=0.05446, over 2202063.31 frames. ], batch size: 76, lr: 4.68e-03, grad_scale: 16.0 2023-04-17 06:38:29,971 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.701e+02 2.601e+02 2.960e+02 3.409e+02 5.967e+02, threshold=5.920e+02, percent-clipped=0.0 2023-04-17 06:38:51,594 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73250.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:39:15,988 INFO [train.py:893] (0/4) Epoch 28, batch 400, loss[loss=0.1828, simple_loss=0.2471, pruned_loss=0.05925, over 13407.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2327, pruned_loss=0.05442, over 2306124.02 frames. ], batch size: 113, lr: 4.68e-03, grad_scale: 16.0 2023-04-17 06:39:30,163 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73293.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:39:36,063 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-17 06:39:49,595 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.1295, 2.0423, 3.7459, 3.6878, 3.5731, 2.9150, 3.4552, 2.7639], device='cuda:0'), covar=tensor([0.1980, 0.1432, 0.0161, 0.0214, 0.0258, 0.0707, 0.0248, 0.0822], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0184, 0.0134, 0.0136, 0.0140, 0.0175, 0.0151, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 06:39:55,157 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73320.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 06:39:59,894 INFO [train.py:893] (0/4) Epoch 28, batch 450, loss[loss=0.1703, simple_loss=0.2288, pruned_loss=0.05592, over 13436.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2329, pruned_loss=0.05448, over 2385834.61 frames. ], batch size: 65, lr: 4.68e-03, grad_scale: 16.0 2023-04-17 06:40:00,716 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.805e+02 2.473e+02 2.892e+02 3.329e+02 8.217e+02, threshold=5.784e+02, percent-clipped=1.0 2023-04-17 06:40:16,525 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-17 06:40:26,123 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73354.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:40:26,641 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-17 06:40:43,526 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-17 06:40:46,165 INFO [train.py:893] (0/4) Epoch 28, batch 500, loss[loss=0.1422, simple_loss=0.1907, pruned_loss=0.04682, over 12499.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2328, pruned_loss=0.05423, over 2446861.46 frames. ], batch size: 51, lr: 4.67e-03, grad_scale: 16.0 2023-04-17 06:41:25,413 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-17 06:41:30,428 INFO [train.py:893] (0/4) Epoch 28, batch 550, loss[loss=0.1913, simple_loss=0.2519, pruned_loss=0.06535, over 13571.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2329, pruned_loss=0.05402, over 2489214.83 frames. ], batch size: 89, lr: 4.67e-03, grad_scale: 16.0 2023-04-17 06:41:31,240 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.685e+02 2.577e+02 2.942e+02 3.367e+02 6.254e+02, threshold=5.883e+02, percent-clipped=2.0 2023-04-17 06:42:15,787 INFO [train.py:893] (0/4) Epoch 28, batch 600, loss[loss=0.1907, simple_loss=0.2415, pruned_loss=0.06996, over 11750.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2317, pruned_loss=0.0536, over 2525124.93 frames. ], batch size: 157, lr: 4.67e-03, grad_scale: 16.0 2023-04-17 06:42:36,957 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2767, 4.3401, 3.0559, 3.8494, 4.1450, 2.7234, 3.8696, 2.8975], device='cuda:0'), covar=tensor([0.0262, 0.0236, 0.0959, 0.0349, 0.0228, 0.1237, 0.0411, 0.1112], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0186, 0.0180, 0.0236, 0.0143, 0.0163, 0.0164, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:43:01,675 INFO [train.py:893] (0/4) Epoch 28, batch 650, loss[loss=0.1554, simple_loss=0.2174, pruned_loss=0.04669, over 13437.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2304, pruned_loss=0.05291, over 2553996.21 frames. ], batch size: 65, lr: 4.67e-03, grad_scale: 16.0 2023-04-17 06:43:02,442 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.755e+02 2.525e+02 2.817e+02 3.484e+02 5.990e+02, threshold=5.634e+02, percent-clipped=2.0 2023-04-17 06:43:21,517 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73550.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:43:45,699 INFO [train.py:893] (0/4) Epoch 28, batch 700, loss[loss=0.1902, simple_loss=0.2547, pruned_loss=0.06284, over 13523.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2299, pruned_loss=0.05235, over 2581665.39 frames. ], batch size: 83, lr: 4.67e-03, grad_scale: 16.0 2023-04-17 06:44:00,300 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73590.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:44:06,747 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73598.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:44:27,184 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73620.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 06:44:32,626 INFO [train.py:893] (0/4) Epoch 28, batch 750, loss[loss=0.165, simple_loss=0.2266, pruned_loss=0.0517, over 13512.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2304, pruned_loss=0.05272, over 2603257.82 frames. ], batch size: 70, lr: 4.67e-03, grad_scale: 16.0 2023-04-17 06:44:33,387 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.905e+02 2.661e+02 3.020e+02 3.513e+02 5.839e+02, threshold=6.040e+02, percent-clipped=2.0 2023-04-17 06:44:53,321 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73649.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:44:55,108 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73651.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:45:10,589 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73668.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 06:45:16,860 INFO [train.py:893] (0/4) Epoch 28, batch 800, loss[loss=0.1906, simple_loss=0.2476, pruned_loss=0.06681, over 13495.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2321, pruned_loss=0.05361, over 2617004.91 frames. ], batch size: 93, lr: 4.66e-03, grad_scale: 16.0 2023-04-17 06:45:44,080 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8541, 3.9339, 2.8564, 3.5698, 3.8131, 2.5773, 3.4942, 2.5839], device='cuda:0'), covar=tensor([0.0298, 0.0238, 0.0938, 0.0462, 0.0263, 0.1214, 0.0472, 0.1203], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0185, 0.0179, 0.0234, 0.0142, 0.0162, 0.0164, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:45:51,442 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2023-04-17 06:46:03,562 INFO [train.py:893] (0/4) Epoch 28, batch 850, loss[loss=0.178, simple_loss=0.2417, pruned_loss=0.05715, over 13530.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.233, pruned_loss=0.05395, over 2627342.62 frames. ], batch size: 85, lr: 4.66e-03, grad_scale: 16.0 2023-04-17 06:46:04,401 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.682e+02 2.586e+02 3.000e+02 3.524e+02 7.256e+02, threshold=5.999e+02, percent-clipped=2.0 2023-04-17 06:46:08,909 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73732.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:46:48,914 INFO [train.py:893] (0/4) Epoch 28, batch 900, loss[loss=0.1468, simple_loss=0.2086, pruned_loss=0.04248, over 13534.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2327, pruned_loss=0.05398, over 2635665.02 frames. ], batch size: 76, lr: 4.66e-03, grad_scale: 16.0 2023-04-17 06:47:04,420 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73793.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:47:18,673 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-17 06:47:34,602 INFO [train.py:893] (0/4) Epoch 28, batch 950, loss[loss=0.1637, simple_loss=0.2305, pruned_loss=0.04845, over 13476.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2315, pruned_loss=0.05404, over 2640564.65 frames. ], batch size: 100, lr: 4.66e-03, grad_scale: 16.0 2023-04-17 06:47:35,322 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.005e+02 2.551e+02 2.954e+02 3.450e+02 1.093e+03, threshold=5.908e+02, percent-clipped=4.0 2023-04-17 06:48:20,951 INFO [train.py:893] (0/4) Epoch 28, batch 1000, loss[loss=0.1412, simple_loss=0.2102, pruned_loss=0.03611, over 13482.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2294, pruned_loss=0.05314, over 2646743.22 frames. ], batch size: 79, lr: 4.66e-03, grad_scale: 16.0 2023-04-17 06:49:04,431 INFO [train.py:893] (0/4) Epoch 28, batch 1050, loss[loss=0.1497, simple_loss=0.2212, pruned_loss=0.0391, over 13256.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2282, pruned_loss=0.05244, over 2646785.12 frames. ], batch size: 124, lr: 4.66e-03, grad_scale: 16.0 2023-04-17 06:49:05,887 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.818e+02 2.540e+02 2.971e+02 3.541e+02 8.453e+02, threshold=5.943e+02, percent-clipped=1.0 2023-04-17 06:49:23,863 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73946.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:49:26,445 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73949.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:49:43,798 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.4485, 2.2341, 2.1318, 2.6988, 1.7985, 2.6595, 2.3231, 2.0587], device='cuda:0'), covar=tensor([0.0163, 0.0311, 0.0281, 0.0213, 0.0423, 0.0176, 0.0366, 0.0329], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0130, 0.0135, 0.0136, 0.0145, 0.0123, 0.0118, 0.0121], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 06:49:51,404 INFO [train.py:893] (0/4) Epoch 28, batch 1100, loss[loss=0.1615, simple_loss=0.2216, pruned_loss=0.05073, over 13083.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2293, pruned_loss=0.05228, over 2648945.49 frames. ], batch size: 142, lr: 4.66e-03, grad_scale: 16.0 2023-04-17 06:50:10,156 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73997.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:50:10,398 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8406, 4.0656, 3.1472, 2.8827, 2.8391, 2.5701, 4.1937, 2.4454], device='cuda:0'), covar=tensor([0.1920, 0.0391, 0.1319, 0.2363, 0.0952, 0.3468, 0.0282, 0.4173], device='cuda:0'), in_proj_covar=tensor([0.0321, 0.0300, 0.0334, 0.0355, 0.0271, 0.0343, 0.0226, 0.0379], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 06:50:13,541 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-74000.pt 2023-04-17 06:50:23,966 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1027, 4.4104, 4.1717, 4.2287, 4.2592, 4.5824, 4.3468, 4.2710], device='cuda:0'), covar=tensor([0.0331, 0.0285, 0.0359, 0.0816, 0.0255, 0.0225, 0.0370, 0.0348], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0168, 0.0192, 0.0276, 0.0191, 0.0208, 0.0188, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 06:50:38,566 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0886, 3.7546, 4.0064, 2.5300, 4.3482, 4.1535, 4.1248, 4.2764], device='cuda:0'), covar=tensor([0.0240, 0.0188, 0.0135, 0.1068, 0.0144, 0.0233, 0.0143, 0.0102], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0061, 0.0089, 0.0106, 0.0105, 0.0118, 0.0087, 0.0083], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:50:40,629 INFO [train.py:893] (0/4) Epoch 28, batch 1150, loss[loss=0.1473, simple_loss=0.2184, pruned_loss=0.0381, over 13031.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2288, pruned_loss=0.05169, over 2644165.45 frames. ], batch size: 142, lr: 4.65e-03, grad_scale: 16.0 2023-04-17 06:50:41,406 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.778e+02 2.470e+02 2.846e+02 3.350e+02 5.021e+02, threshold=5.693e+02, percent-clipped=0.0 2023-04-17 06:51:15,787 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74064.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:51:25,528 INFO [train.py:893] (0/4) Epoch 28, batch 1200, loss[loss=0.1274, simple_loss=0.1836, pruned_loss=0.0356, over 12782.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2289, pruned_loss=0.0518, over 2649915.61 frames. ], batch size: 52, lr: 4.65e-03, grad_scale: 16.0 2023-04-17 06:51:36,791 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74088.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:51:51,956 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-17 06:51:53,858 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2255, 4.5239, 3.2032, 4.3867, 4.4631, 3.2023, 3.7857, 3.3519], device='cuda:0'), covar=tensor([0.0349, 0.0344, 0.0849, 0.0385, 0.0219, 0.0891, 0.0593, 0.0860], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0189, 0.0183, 0.0239, 0.0144, 0.0166, 0.0167, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:52:04,039 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-17 06:52:08,419 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3851, 3.7620, 3.5673, 4.1235, 2.2517, 3.1790, 3.9886, 2.2428], device='cuda:0'), covar=tensor([0.0145, 0.0451, 0.0823, 0.0515, 0.1717, 0.0967, 0.0414, 0.1801], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0192, 0.0211, 0.0256, 0.0187, 0.0206, 0.0183, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:52:10,855 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74125.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:52:11,349 INFO [train.py:893] (0/4) Epoch 28, batch 1250, loss[loss=0.1746, simple_loss=0.2419, pruned_loss=0.05366, over 13558.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2294, pruned_loss=0.05205, over 2651257.31 frames. ], batch size: 89, lr: 4.65e-03, grad_scale: 16.0 2023-04-17 06:52:12,122 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.727e+02 2.579e+02 2.969e+02 3.696e+02 6.007e+02, threshold=5.937e+02, percent-clipped=2.0 2023-04-17 06:52:30,036 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-17 06:52:56,465 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3450, 4.5826, 4.3403, 4.4225, 4.4279, 4.7655, 4.5602, 4.4510], device='cuda:0'), covar=tensor([0.0322, 0.0282, 0.0342, 0.0776, 0.0266, 0.0218, 0.0302, 0.0271], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0168, 0.0192, 0.0275, 0.0191, 0.0207, 0.0187, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 06:52:57,764 INFO [train.py:893] (0/4) Epoch 28, batch 1300, loss[loss=0.1635, simple_loss=0.2297, pruned_loss=0.04867, over 13548.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2306, pruned_loss=0.0524, over 2653709.68 frames. ], batch size: 78, lr: 4.65e-03, grad_scale: 16.0 2023-04-17 06:53:17,927 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6950, 2.4571, 2.3979, 2.8313, 2.0294, 2.8170, 2.7039, 2.3604], device='cuda:0'), covar=tensor([0.0086, 0.0203, 0.0150, 0.0129, 0.0246, 0.0134, 0.0182, 0.0184], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0129, 0.0133, 0.0135, 0.0144, 0.0123, 0.0117, 0.0121], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 06:53:21,849 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-17 06:53:28,620 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2768, 2.6052, 2.0238, 4.2225, 4.7086, 3.4714, 4.6029, 4.4025], device='cuda:0'), covar=tensor([0.0106, 0.0941, 0.1056, 0.0101, 0.0060, 0.0466, 0.0064, 0.0087], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0094, 0.0101, 0.0086, 0.0073, 0.0083, 0.0060, 0.0083], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 06:53:29,621 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-17 06:53:43,807 INFO [train.py:893] (0/4) Epoch 28, batch 1350, loss[loss=0.1702, simple_loss=0.2358, pruned_loss=0.05235, over 13420.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.231, pruned_loss=0.05262, over 2654037.06 frames. ], batch size: 113, lr: 4.65e-03, grad_scale: 16.0 2023-04-17 06:53:44,620 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.857e+02 2.634e+02 3.033e+02 3.647e+02 6.795e+02, threshold=6.067e+02, percent-clipped=2.0 2023-04-17 06:53:57,424 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74240.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:54:02,963 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74246.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:54:30,464 INFO [train.py:893] (0/4) Epoch 28, batch 1400, loss[loss=0.1647, simple_loss=0.2277, pruned_loss=0.05081, over 13517.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2295, pruned_loss=0.05165, over 2656319.45 frames. ], batch size: 76, lr: 4.65e-03, grad_scale: 16.0 2023-04-17 06:54:46,477 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74294.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:54:53,973 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74301.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:55:15,846 INFO [train.py:893] (0/4) Epoch 28, batch 1450, loss[loss=0.1477, simple_loss=0.2157, pruned_loss=0.03983, over 13338.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2293, pruned_loss=0.05177, over 2654770.35 frames. ], batch size: 118, lr: 4.64e-03, grad_scale: 16.0 2023-04-17 06:55:16,670 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.844e+02 2.544e+02 3.026e+02 3.432e+02 5.527e+02, threshold=6.053e+02, percent-clipped=0.0 2023-04-17 06:56:02,057 INFO [train.py:893] (0/4) Epoch 28, batch 1500, loss[loss=0.1723, simple_loss=0.2384, pruned_loss=0.05311, over 13361.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2291, pruned_loss=0.05146, over 2654927.35 frames. ], batch size: 118, lr: 4.64e-03, grad_scale: 16.0 2023-04-17 06:56:12,852 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74388.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:56:42,328 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74420.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:56:47,108 INFO [train.py:893] (0/4) Epoch 28, batch 1550, loss[loss=0.1677, simple_loss=0.2384, pruned_loss=0.04849, over 13481.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2295, pruned_loss=0.0513, over 2656350.01 frames. ], batch size: 100, lr: 4.64e-03, grad_scale: 16.0 2023-04-17 06:56:48,572 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.854e+02 2.451e+02 2.902e+02 3.422e+02 5.266e+02, threshold=5.804e+02, percent-clipped=0.0 2023-04-17 06:56:57,506 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74436.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:57:33,952 INFO [train.py:893] (0/4) Epoch 28, batch 1600, loss[loss=0.1734, simple_loss=0.2421, pruned_loss=0.05236, over 13445.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2302, pruned_loss=0.05158, over 2654527.75 frames. ], batch size: 103, lr: 4.64e-03, grad_scale: 16.0 2023-04-17 06:57:36,699 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0989, 2.8636, 3.3491, 2.6718, 2.4324, 2.5455, 3.6473, 3.7201], device='cuda:0'), covar=tensor([0.1204, 0.1795, 0.0407, 0.1485, 0.1522, 0.1522, 0.0311, 0.0240], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0283, 0.0205, 0.0229, 0.0224, 0.0190, 0.0220, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 06:58:19,750 INFO [train.py:893] (0/4) Epoch 28, batch 1650, loss[loss=0.1667, simple_loss=0.2373, pruned_loss=0.04803, over 13352.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2305, pruned_loss=0.05131, over 2646281.56 frames. ], batch size: 73, lr: 4.64e-03, grad_scale: 16.0 2023-04-17 06:58:20,548 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.816e+02 2.550e+02 3.022e+02 3.774e+02 7.035e+02, threshold=6.044e+02, percent-clipped=3.0 2023-04-17 06:58:32,371 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-17 06:58:48,308 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.8406, 4.2035, 3.0009, 2.8150, 2.9242, 2.5329, 4.2370, 2.3596], device='cuda:0'), covar=tensor([0.2038, 0.0325, 0.1599, 0.2521, 0.0904, 0.3409, 0.0293, 0.4354], device='cuda:0'), in_proj_covar=tensor([0.0321, 0.0299, 0.0334, 0.0355, 0.0271, 0.0343, 0.0226, 0.0379], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 06:59:04,959 INFO [train.py:893] (0/4) Epoch 28, batch 1700, loss[loss=0.1579, simple_loss=0.2201, pruned_loss=0.04782, over 13415.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2315, pruned_loss=0.05159, over 2649459.34 frames. ], batch size: 65, lr: 4.64e-03, grad_scale: 16.0 2023-04-17 06:59:23,155 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74595.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:59:23,941 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74596.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 06:59:51,393 INFO [train.py:893] (0/4) Epoch 28, batch 1750, loss[loss=0.1578, simple_loss=0.2122, pruned_loss=0.05168, over 13197.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2297, pruned_loss=0.0509, over 2647687.43 frames. ], batch size: 58, lr: 4.64e-03, grad_scale: 16.0 2023-04-17 06:59:52,264 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.683e+02 2.479e+02 3.029e+02 3.465e+02 5.673e+02, threshold=6.058e+02, percent-clipped=0.0 2023-04-17 07:00:01,673 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1818, 2.0319, 2.4698, 3.4290, 3.1787, 3.5131, 2.7980, 2.1593], device='cuda:0'), covar=tensor([0.0250, 0.0893, 0.0801, 0.0100, 0.0268, 0.0074, 0.0696, 0.0983], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0152, 0.0171, 0.0108, 0.0126, 0.0104, 0.0175, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:00:19,692 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74656.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:00:38,099 INFO [train.py:893] (0/4) Epoch 28, batch 1800, loss[loss=0.1789, simple_loss=0.235, pruned_loss=0.06142, over 11864.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2288, pruned_loss=0.05017, over 2653964.18 frames. ], batch size: 157, lr: 4.63e-03, grad_scale: 16.0 2023-04-17 07:01:18,024 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74720.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:01:23,279 INFO [train.py:893] (0/4) Epoch 28, batch 1850, loss[loss=0.1388, simple_loss=0.2092, pruned_loss=0.03424, over 13508.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2284, pruned_loss=0.05004, over 2653098.26 frames. ], batch size: 76, lr: 4.63e-03, grad_scale: 16.0 2023-04-17 07:01:24,076 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.611e+02 2.423e+02 2.814e+02 3.269e+02 7.266e+02, threshold=5.629e+02, percent-clipped=1.0 2023-04-17 07:01:24,165 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-17 07:01:40,427 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-17 07:02:01,235 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74768.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:02:09,152 INFO [train.py:893] (0/4) Epoch 28, batch 1900, loss[loss=0.1872, simple_loss=0.2422, pruned_loss=0.06614, over 11878.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2279, pruned_loss=0.0501, over 2655741.90 frames. ], batch size: 157, lr: 4.63e-03, grad_scale: 16.0 2023-04-17 07:02:16,681 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1426, 4.0704, 3.0623, 3.8179, 4.0919, 2.7931, 3.7738, 2.7518], device='cuda:0'), covar=tensor([0.0289, 0.0284, 0.1003, 0.0409, 0.0249, 0.1190, 0.0468, 0.1227], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0190, 0.0182, 0.0240, 0.0145, 0.0166, 0.0167, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:02:54,448 INFO [train.py:893] (0/4) Epoch 28, batch 1950, loss[loss=0.1675, simple_loss=0.2333, pruned_loss=0.05085, over 13441.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2273, pruned_loss=0.05011, over 2658351.32 frames. ], batch size: 106, lr: 4.63e-03, grad_scale: 16.0 2023-04-17 07:02:55,262 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.498e+02 2.666e+02 3.093e+02 3.546e+02 8.280e+02, threshold=6.185e+02, percent-clipped=1.0 2023-04-17 07:03:40,123 INFO [train.py:893] (0/4) Epoch 28, batch 2000, loss[loss=0.2013, simple_loss=0.2697, pruned_loss=0.06651, over 13479.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2297, pruned_loss=0.05142, over 2660328.83 frames. ], batch size: 93, lr: 4.63e-03, grad_scale: 16.0 2023-04-17 07:03:42,006 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9830, 4.8024, 5.0379, 4.8888, 5.2899, 4.8189, 5.2933, 5.2210], device='cuda:0'), covar=tensor([0.0405, 0.0499, 0.0591, 0.0530, 0.0482, 0.0828, 0.0422, 0.0475], device='cuda:0'), in_proj_covar=tensor([0.0314, 0.0314, 0.0324, 0.0240, 0.0462, 0.0363, 0.0306, 0.0321], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:03:44,947 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 07:03:59,091 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74896.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:04:10,740 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74909.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:04:26,422 INFO [train.py:893] (0/4) Epoch 28, batch 2050, loss[loss=0.2223, simple_loss=0.2634, pruned_loss=0.09062, over 11500.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2311, pruned_loss=0.05227, over 2657596.95 frames. ], batch size: 157, lr: 4.63e-03, grad_scale: 32.0 2023-04-17 07:04:27,211 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.636e+02 2.577e+02 2.966e+02 3.569e+02 6.675e+02, threshold=5.932e+02, percent-clipped=3.0 2023-04-17 07:04:33,234 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-17 07:04:42,667 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74944.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:04:48,465 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74951.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:05:01,347 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74964.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:05:06,350 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74970.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:05:11,613 INFO [train.py:893] (0/4) Epoch 28, batch 2100, loss[loss=0.1762, simple_loss=0.2355, pruned_loss=0.05844, over 13424.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2301, pruned_loss=0.05165, over 2660955.84 frames. ], batch size: 103, lr: 4.62e-03, grad_scale: 32.0 2023-04-17 07:05:28,406 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-17 07:05:56,443 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75025.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:05:56,923 INFO [train.py:893] (0/4) Epoch 28, batch 2150, loss[loss=0.1728, simple_loss=0.2394, pruned_loss=0.05308, over 13472.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2304, pruned_loss=0.05139, over 2665902.02 frames. ], batch size: 106, lr: 4.62e-03, grad_scale: 32.0 2023-04-17 07:05:58,384 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.932e+02 2.614e+02 2.974e+02 3.546e+02 5.707e+02, threshold=5.948e+02, percent-clipped=0.0 2023-04-17 07:06:02,980 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7302, 3.8492, 2.9401, 2.7038, 2.7568, 2.4487, 3.9397, 2.2381], device='cuda:0'), covar=tensor([0.1916, 0.0413, 0.1414, 0.2373, 0.0927, 0.3485, 0.0325, 0.4467], device='cuda:0'), in_proj_covar=tensor([0.0324, 0.0302, 0.0336, 0.0357, 0.0273, 0.0346, 0.0228, 0.0382], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 07:06:42,117 INFO [train.py:893] (0/4) Epoch 28, batch 2200, loss[loss=0.1691, simple_loss=0.24, pruned_loss=0.04915, over 13442.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2296, pruned_loss=0.05105, over 2656913.30 frames. ], batch size: 106, lr: 4.62e-03, grad_scale: 16.0 2023-04-17 07:07:01,708 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0140, 4.1941, 4.1903, 3.9287, 4.2196, 3.7845, 4.2891, 4.3612], device='cuda:0'), covar=tensor([0.0267, 0.0401, 0.0319, 0.0546, 0.0328, 0.0431, 0.0415, 0.0366], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0213, 0.0178, 0.0193, 0.0164, 0.0213, 0.0141, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 07:07:21,137 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.6059, 5.1234, 4.9497, 5.0179, 4.9361, 4.8880, 5.5404, 5.1601], device='cuda:0'), covar=tensor([0.0593, 0.1078, 0.2202, 0.2339, 0.0896, 0.1527, 0.0865, 0.1221], device='cuda:0'), in_proj_covar=tensor([0.0355, 0.0402, 0.0495, 0.0497, 0.0317, 0.0371, 0.0458, 0.0372], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:07:28,228 INFO [train.py:893] (0/4) Epoch 28, batch 2250, loss[loss=0.1652, simple_loss=0.23, pruned_loss=0.05021, over 13443.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2283, pruned_loss=0.05052, over 2657792.63 frames. ], batch size: 95, lr: 4.62e-03, grad_scale: 16.0 2023-04-17 07:07:29,797 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.746e+02 2.493e+02 2.845e+02 3.262e+02 6.199e+02, threshold=5.691e+02, percent-clipped=1.0 2023-04-17 07:07:58,483 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2867, 3.1270, 3.6041, 2.6502, 2.3704, 2.5597, 3.9831, 4.0783], device='cuda:0'), covar=tensor([0.1159, 0.1930, 0.0436, 0.1965, 0.1893, 0.1803, 0.0301, 0.0182], device='cuda:0'), in_proj_covar=tensor([0.0259, 0.0286, 0.0207, 0.0232, 0.0226, 0.0192, 0.0223, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:08:13,256 INFO [train.py:893] (0/4) Epoch 28, batch 2300, loss[loss=0.1507, simple_loss=0.2152, pruned_loss=0.04306, over 13530.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2279, pruned_loss=0.05015, over 2656590.40 frames. ], batch size: 72, lr: 4.62e-03, grad_scale: 16.0 2023-04-17 07:08:45,591 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3855, 2.4324, 4.2007, 3.9159, 4.0847, 3.3875, 3.8030, 3.2572], device='cuda:0'), covar=tensor([0.1827, 0.1374, 0.0101, 0.0250, 0.0167, 0.0533, 0.0245, 0.0624], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0188, 0.0135, 0.0140, 0.0144, 0.0178, 0.0154, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 07:08:58,909 INFO [train.py:893] (0/4) Epoch 28, batch 2350, loss[loss=0.1636, simple_loss=0.2212, pruned_loss=0.05301, over 13470.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2278, pruned_loss=0.05016, over 2657165.43 frames. ], batch size: 106, lr: 4.62e-03, grad_scale: 16.0 2023-04-17 07:09:00,523 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.654e+02 2.368e+02 2.701e+02 3.226e+02 5.813e+02, threshold=5.403e+02, percent-clipped=1.0 2023-04-17 07:09:00,822 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=75228.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:09:06,441 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6783, 3.9557, 3.7950, 3.7986, 3.8451, 4.0179, 3.9070, 3.5729], device='cuda:0'), covar=tensor([0.0248, 0.0253, 0.0265, 0.0548, 0.0211, 0.0169, 0.0245, 0.0739], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0168, 0.0191, 0.0274, 0.0190, 0.0206, 0.0186, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 07:09:22,415 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 07:09:22,609 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75251.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:09:34,969 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75265.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:09:39,134 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=75270.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:09:44,434 INFO [train.py:893] (0/4) Epoch 28, batch 2400, loss[loss=0.1847, simple_loss=0.2387, pruned_loss=0.06537, over 13494.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2273, pruned_loss=0.05013, over 2655471.43 frames. ], batch size: 93, lr: 4.62e-03, grad_scale: 16.0 2023-04-17 07:09:47,309 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-17 07:09:50,537 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0919, 4.3550, 3.2077, 2.9440, 3.0664, 2.6093, 4.4621, 2.5752], device='cuda:0'), covar=tensor([0.1923, 0.0380, 0.1625, 0.2690, 0.0976, 0.3969, 0.0289, 0.4471], device='cuda:0'), in_proj_covar=tensor([0.0323, 0.0302, 0.0336, 0.0357, 0.0273, 0.0346, 0.0228, 0.0383], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 07:09:51,187 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.8423, 4.6664, 4.8874, 4.7797, 5.1187, 4.6526, 5.1161, 5.0707], device='cuda:0'), covar=tensor([0.0416, 0.0537, 0.0652, 0.0536, 0.0578, 0.0820, 0.0500, 0.0455], device='cuda:0'), in_proj_covar=tensor([0.0313, 0.0314, 0.0325, 0.0241, 0.0462, 0.0362, 0.0304, 0.0322], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:09:52,129 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2388, 4.2024, 3.1821, 3.9172, 4.1491, 2.6425, 3.8913, 2.8137], device='cuda:0'), covar=tensor([0.0265, 0.0205, 0.0897, 0.0405, 0.0260, 0.1251, 0.0410, 0.1217], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0189, 0.0181, 0.0240, 0.0145, 0.0164, 0.0167, 0.0162], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:09:57,094 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75289.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:10:05,937 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75299.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:10:25,069 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75320.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:10:29,674 INFO [train.py:893] (0/4) Epoch 28, batch 2450, loss[loss=0.1681, simple_loss=0.2324, pruned_loss=0.05186, over 13568.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2266, pruned_loss=0.0501, over 2654322.45 frames. ], batch size: 89, lr: 4.61e-03, grad_scale: 16.0 2023-04-17 07:10:31,328 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.808e+02 2.390e+02 2.893e+02 3.505e+02 5.299e+02, threshold=5.785e+02, percent-clipped=0.0 2023-04-17 07:10:34,842 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75331.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 07:10:51,932 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0828, 2.1261, 3.9688, 3.8209, 3.8007, 3.2821, 3.5315, 3.0463], device='cuda:0'), covar=tensor([0.2354, 0.1588, 0.0156, 0.0204, 0.0216, 0.0610, 0.0365, 0.0817], device='cuda:0'), in_proj_covar=tensor([0.0223, 0.0189, 0.0136, 0.0141, 0.0144, 0.0179, 0.0155, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 07:11:16,476 INFO [train.py:893] (0/4) Epoch 28, batch 2500, loss[loss=0.1421, simple_loss=0.1986, pruned_loss=0.04285, over 13201.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.2266, pruned_loss=0.05037, over 2651972.62 frames. ], batch size: 58, lr: 4.61e-03, grad_scale: 16.0 2023-04-17 07:11:45,375 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0056, 4.4592, 4.2692, 4.2041, 4.2940, 4.0741, 4.5201, 4.5262], device='cuda:0'), covar=tensor([0.0238, 0.0208, 0.0213, 0.0347, 0.0279, 0.0257, 0.0229, 0.0234], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0212, 0.0178, 0.0194, 0.0165, 0.0213, 0.0141, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 07:12:01,178 INFO [train.py:893] (0/4) Epoch 28, batch 2550, loss[loss=0.1429, simple_loss=0.1993, pruned_loss=0.04326, over 13196.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2269, pruned_loss=0.05031, over 2655154.82 frames. ], batch size: 58, lr: 4.61e-03, grad_scale: 16.0 2023-04-17 07:12:03,485 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.759e+02 2.463e+02 2.650e+02 3.401e+02 6.679e+02, threshold=5.300e+02, percent-clipped=1.0 2023-04-17 07:12:18,325 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9620, 3.7054, 3.8766, 2.3284, 4.1982, 4.0285, 3.9484, 4.1577], device='cuda:0'), covar=tensor([0.0240, 0.0154, 0.0154, 0.1166, 0.0151, 0.0236, 0.0147, 0.0143], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0063, 0.0092, 0.0109, 0.0108, 0.0121, 0.0091, 0.0086], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:12:25,408 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 07:12:26,515 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4616, 4.9199, 4.7165, 4.6859, 4.7181, 4.5009, 4.9553, 4.9626], device='cuda:0'), covar=tensor([0.0205, 0.0167, 0.0165, 0.0305, 0.0247, 0.0224, 0.0233, 0.0191], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0212, 0.0179, 0.0194, 0.0165, 0.0213, 0.0141, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 07:12:47,077 INFO [train.py:893] (0/4) Epoch 28, batch 2600, loss[loss=0.1596, simple_loss=0.2203, pruned_loss=0.04946, over 13575.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2265, pruned_loss=0.05015, over 2660648.85 frames. ], batch size: 72, lr: 4.61e-03, grad_scale: 16.0 2023-04-17 07:12:49,883 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9154, 1.9208, 3.4475, 3.4449, 3.3194, 2.7195, 3.1884, 2.6760], device='cuda:0'), covar=tensor([0.2094, 0.1549, 0.0241, 0.0218, 0.0274, 0.0841, 0.0351, 0.0834], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0189, 0.0136, 0.0140, 0.0143, 0.0179, 0.0154, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 07:13:06,918 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7138, 3.9419, 3.7892, 3.7957, 3.8520, 3.9796, 3.9002, 3.5300], device='cuda:0'), covar=tensor([0.0272, 0.0295, 0.0284, 0.0548, 0.0241, 0.0208, 0.0302, 0.0839], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0167, 0.0189, 0.0271, 0.0188, 0.0204, 0.0184, 0.0114], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 07:13:28,709 INFO [train.py:893] (0/4) Epoch 28, batch 2650, loss[loss=0.1717, simple_loss=0.2372, pruned_loss=0.05309, over 13498.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2276, pruned_loss=0.05069, over 2659317.53 frames. ], batch size: 93, lr: 4.61e-03, grad_scale: 16.0 2023-04-17 07:13:30,203 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.967e+02 2.635e+02 3.052e+02 3.899e+02 8.488e+02, threshold=6.104e+02, percent-clipped=5.0 2023-04-17 07:13:58,451 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75565.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:14:06,246 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-28.pt 2023-04-17 07:14:31,263 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-17 07:14:35,391 INFO [train.py:893] (0/4) Epoch 29, batch 0, loss[loss=0.1504, simple_loss=0.2132, pruned_loss=0.04383, over 13229.00 frames. ], tot_loss[loss=0.1504, simple_loss=0.2132, pruned_loss=0.04383, over 13229.00 frames. ], batch size: 124, lr: 4.52e-03, grad_scale: 16.0 2023-04-17 07:14:35,392 INFO [train.py:918] (0/4) Computing validation loss 2023-04-17 07:14:56,943 INFO [train.py:927] (0/4) Epoch 29, validation: loss=0.1316, simple_loss=0.1914, pruned_loss=0.03588, over 2446609.00 frames. 2023-04-17 07:14:56,944 INFO [train.py:928] (0/4) Maximum memory allocated so far is 13044MB 2023-04-17 07:15:06,281 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75584.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:15:31,948 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75613.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:15:38,494 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75620.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:15:42,453 INFO [train.py:893] (0/4) Epoch 29, batch 50, loss[loss=0.1746, simple_loss=0.2351, pruned_loss=0.05699, over 13445.00 frames. ], tot_loss[loss=0.162, simple_loss=0.2229, pruned_loss=0.05049, over 600413.28 frames. ], batch size: 95, lr: 4.52e-03, grad_scale: 16.0 2023-04-17 07:15:43,492 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75626.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 07:15:44,926 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.808e+02 2.542e+02 2.969e+02 3.744e+02 7.692e+02, threshold=5.938e+02, percent-clipped=2.0 2023-04-17 07:15:52,160 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-17 07:15:58,602 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2701, 4.1203, 4.1350, 2.6355, 4.5481, 4.3189, 4.2712, 4.4329], device='cuda:0'), covar=tensor([0.0253, 0.0130, 0.0152, 0.1119, 0.0153, 0.0251, 0.0142, 0.0121], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0063, 0.0092, 0.0108, 0.0108, 0.0121, 0.0090, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:16:06,946 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-17 07:16:06,946 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-17 07:16:06,946 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-17 07:16:06,954 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-17 07:16:06,971 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-17 07:16:06,985 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-17 07:16:07,002 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-17 07:16:21,732 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75668.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:16:27,289 INFO [train.py:893] (0/4) Epoch 29, batch 100, loss[loss=0.1735, simple_loss=0.2288, pruned_loss=0.05914, over 13516.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2248, pruned_loss=0.05207, over 1058302.28 frames. ], batch size: 85, lr: 4.52e-03, grad_scale: 16.0 2023-04-17 07:17:13,315 INFO [train.py:893] (0/4) Epoch 29, batch 150, loss[loss=0.1817, simple_loss=0.2467, pruned_loss=0.05839, over 13450.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.227, pruned_loss=0.05316, over 1399210.79 frames. ], batch size: 103, lr: 4.52e-03, grad_scale: 16.0 2023-04-17 07:17:16,374 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.802e+02 2.595e+02 2.945e+02 3.538e+02 7.026e+02, threshold=5.889e+02, percent-clipped=1.0 2023-04-17 07:17:59,197 INFO [train.py:893] (0/4) Epoch 29, batch 200, loss[loss=0.1696, simple_loss=0.236, pruned_loss=0.05161, over 13467.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2289, pruned_loss=0.05336, over 1666107.40 frames. ], batch size: 79, lr: 4.52e-03, grad_scale: 16.0 2023-04-17 07:18:45,527 INFO [train.py:893] (0/4) Epoch 29, batch 250, loss[loss=0.1478, simple_loss=0.2189, pruned_loss=0.03832, over 13498.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2293, pruned_loss=0.05341, over 1880101.47 frames. ], batch size: 81, lr: 4.52e-03, grad_scale: 16.0 2023-04-17 07:18:48,009 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.006e+02 2.662e+02 3.045e+02 3.757e+02 6.774e+02, threshold=6.090e+02, percent-clipped=3.0 2023-04-17 07:19:06,316 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-17 07:19:30,081 INFO [train.py:893] (0/4) Epoch 29, batch 300, loss[loss=0.166, simple_loss=0.233, pruned_loss=0.0495, over 13524.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2292, pruned_loss=0.05321, over 2043945.33 frames. ], batch size: 98, lr: 4.52e-03, grad_scale: 16.0 2023-04-17 07:19:38,377 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75884.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:19:42,056 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-17 07:19:47,059 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-17 07:19:48,350 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7541, 3.7675, 2.9957, 3.2943, 3.0749, 2.3187, 3.8014, 2.1232], device='cuda:0'), covar=tensor([0.0777, 0.0447, 0.0506, 0.0424, 0.0667, 0.1939, 0.1055, 0.1410], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0151, 0.0141, 0.0122, 0.0153, 0.0194, 0.0192, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 07:20:15,667 INFO [train.py:893] (0/4) Epoch 29, batch 350, loss[loss=0.1772, simple_loss=0.2357, pruned_loss=0.05932, over 13529.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2299, pruned_loss=0.05318, over 2183474.68 frames. ], batch size: 83, lr: 4.51e-03, grad_scale: 16.0 2023-04-17 07:20:16,767 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75926.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 07:20:18,050 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.812e+02 2.506e+02 3.011e+02 3.603e+02 7.167e+02, threshold=6.022e+02, percent-clipped=1.0 2023-04-17 07:20:22,825 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75932.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:20:59,768 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75974.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:21:00,427 INFO [train.py:893] (0/4) Epoch 29, batch 400, loss[loss=0.1726, simple_loss=0.2312, pruned_loss=0.05699, over 13555.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2296, pruned_loss=0.05268, over 2291148.47 frames. ], batch size: 83, lr: 4.51e-03, grad_scale: 16.0 2023-04-17 07:21:23,474 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-76000.pt 2023-04-17 07:21:49,780 INFO [train.py:893] (0/4) Epoch 29, batch 450, loss[loss=0.1684, simple_loss=0.2336, pruned_loss=0.05165, over 13572.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2314, pruned_loss=0.0532, over 2373352.59 frames. ], batch size: 89, lr: 4.51e-03, grad_scale: 16.0 2023-04-17 07:21:52,273 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.887e+02 2.551e+02 2.867e+02 3.275e+02 5.566e+02, threshold=5.734e+02, percent-clipped=0.0 2023-04-17 07:22:14,359 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-17 07:22:21,203 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6969, 3.5339, 4.3596, 3.1253, 2.9191, 2.9105, 4.6044, 4.7145], device='cuda:0'), covar=tensor([0.1177, 0.1688, 0.0330, 0.1716, 0.1553, 0.1529, 0.0254, 0.0289], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0283, 0.0205, 0.0228, 0.0223, 0.0189, 0.0221, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:22:33,987 INFO [train.py:893] (0/4) Epoch 29, batch 500, loss[loss=0.237, simple_loss=0.2829, pruned_loss=0.09552, over 11792.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2316, pruned_loss=0.05298, over 2434722.64 frames. ], batch size: 157, lr: 4.51e-03, grad_scale: 16.0 2023-04-17 07:22:39,227 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4218, 3.7028, 3.6212, 4.2264, 2.3286, 3.2456, 3.9382, 2.3305], device='cuda:0'), covar=tensor([0.0150, 0.0525, 0.0762, 0.0548, 0.1584, 0.0884, 0.0469, 0.1645], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0193, 0.0210, 0.0257, 0.0187, 0.0206, 0.0184, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:23:00,580 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6254, 3.4926, 2.7946, 3.2096, 3.5782, 2.3641, 3.3089, 2.5258], device='cuda:0'), covar=tensor([0.0267, 0.0210, 0.0909, 0.0354, 0.0289, 0.1192, 0.0496, 0.1162], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0187, 0.0179, 0.0237, 0.0144, 0.0163, 0.0165, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:23:19,784 INFO [train.py:893] (0/4) Epoch 29, batch 550, loss[loss=0.1806, simple_loss=0.2506, pruned_loss=0.05534, over 13453.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.232, pruned_loss=0.05285, over 2483892.67 frames. ], batch size: 100, lr: 4.51e-03, grad_scale: 16.0 2023-04-17 07:23:22,217 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.913e+02 2.461e+02 2.951e+02 3.383e+02 6.216e+02, threshold=5.902e+02, percent-clipped=2.0 2023-04-17 07:23:40,109 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5065, 3.8692, 3.5575, 4.4591, 2.1567, 2.9708, 3.9413, 2.3674], device='cuda:0'), covar=tensor([0.0138, 0.0487, 0.0824, 0.0405, 0.1826, 0.1156, 0.0542, 0.1738], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0192, 0.0210, 0.0257, 0.0187, 0.0205, 0.0183, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:24:03,471 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3998, 4.1849, 3.5881, 3.9243, 3.3740, 2.6883, 4.1711, 2.5435], device='cuda:0'), covar=tensor([0.0577, 0.0380, 0.0450, 0.0255, 0.0663, 0.1735, 0.0917, 0.1284], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0151, 0.0140, 0.0121, 0.0153, 0.0192, 0.0190, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 07:24:04,826 INFO [train.py:893] (0/4) Epoch 29, batch 600, loss[loss=0.1555, simple_loss=0.2149, pruned_loss=0.04804, over 11858.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2306, pruned_loss=0.05258, over 2513569.81 frames. ], batch size: 157, lr: 4.51e-03, grad_scale: 16.0 2023-04-17 07:24:15,404 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.1594, 2.0451, 3.8129, 3.6869, 3.6407, 2.8446, 3.4577, 2.8438], device='cuda:0'), covar=tensor([0.1901, 0.1485, 0.0192, 0.0247, 0.0259, 0.0792, 0.0290, 0.0843], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0189, 0.0135, 0.0140, 0.0144, 0.0181, 0.0153, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 07:24:50,355 INFO [train.py:893] (0/4) Epoch 29, batch 650, loss[loss=0.1713, simple_loss=0.2347, pruned_loss=0.05392, over 13393.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2299, pruned_loss=0.05214, over 2547575.21 frames. ], batch size: 62, lr: 4.51e-03, grad_scale: 16.0 2023-04-17 07:24:53,530 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.828e+02 2.554e+02 2.954e+02 3.493e+02 1.019e+03, threshold=5.908e+02, percent-clipped=2.0 2023-04-17 07:25:08,697 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-17 07:25:31,149 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76269.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 07:25:37,231 INFO [train.py:893] (0/4) Epoch 29, batch 700, loss[loss=0.1465, simple_loss=0.2119, pruned_loss=0.04053, over 13549.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2287, pruned_loss=0.05129, over 2574898.09 frames. ], batch size: 72, lr: 4.50e-03, grad_scale: 16.0 2023-04-17 07:25:50,590 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3995, 4.7955, 4.5495, 4.5610, 4.5989, 4.4087, 4.8381, 4.8673], device='cuda:0'), covar=tensor([0.0216, 0.0178, 0.0224, 0.0297, 0.0214, 0.0266, 0.0241, 0.0187], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0214, 0.0181, 0.0195, 0.0167, 0.0216, 0.0144, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 07:25:53,812 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76294.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:26:05,260 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-17 07:26:14,974 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-17 07:26:20,373 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76323.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 07:26:21,752 INFO [train.py:893] (0/4) Epoch 29, batch 750, loss[loss=0.1722, simple_loss=0.2371, pruned_loss=0.05366, over 13402.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2283, pruned_loss=0.05133, over 2595354.09 frames. ], batch size: 113, lr: 4.50e-03, grad_scale: 16.0 2023-04-17 07:26:24,124 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.818e+02 2.465e+02 2.925e+02 3.423e+02 5.602e+02, threshold=5.850e+02, percent-clipped=0.0 2023-04-17 07:26:26,885 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76330.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 07:26:49,914 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76355.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:27:03,409 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0878, 2.1389, 4.0214, 3.8289, 3.7870, 3.0453, 3.5554, 2.9615], device='cuda:0'), covar=tensor([0.2183, 0.1477, 0.0139, 0.0203, 0.0244, 0.0741, 0.0281, 0.0832], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0188, 0.0135, 0.0140, 0.0143, 0.0180, 0.0153, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 07:27:08,056 INFO [train.py:893] (0/4) Epoch 29, batch 800, loss[loss=0.1634, simple_loss=0.2242, pruned_loss=0.05127, over 13533.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2297, pruned_loss=0.05159, over 2612753.35 frames. ], batch size: 85, lr: 4.50e-03, grad_scale: 16.0 2023-04-17 07:27:16,399 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76384.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 07:27:53,025 INFO [train.py:893] (0/4) Epoch 29, batch 850, loss[loss=0.1726, simple_loss=0.2293, pruned_loss=0.058, over 13387.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2314, pruned_loss=0.05231, over 2626846.98 frames. ], batch size: 109, lr: 4.50e-03, grad_scale: 16.0 2023-04-17 07:27:56,141 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.913e+02 2.473e+02 2.862e+02 3.521e+02 5.957e+02, threshold=5.724e+02, percent-clipped=2.0 2023-04-17 07:28:10,323 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4671, 4.5791, 3.2141, 4.2366, 4.3986, 2.9733, 3.9812, 3.0162], device='cuda:0'), covar=tensor([0.0247, 0.0224, 0.0955, 0.0460, 0.0190, 0.1105, 0.0401, 0.1176], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0188, 0.0180, 0.0239, 0.0145, 0.0165, 0.0166, 0.0162], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:28:38,683 INFO [train.py:893] (0/4) Epoch 29, batch 900, loss[loss=0.1497, simple_loss=0.2153, pruned_loss=0.04205, over 13550.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2312, pruned_loss=0.0529, over 2636454.44 frames. ], batch size: 72, lr: 4.50e-03, grad_scale: 16.0 2023-04-17 07:29:10,031 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-17 07:29:24,505 INFO [train.py:893] (0/4) Epoch 29, batch 950, loss[loss=0.153, simple_loss=0.2152, pruned_loss=0.04542, over 13488.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2306, pruned_loss=0.05341, over 2643446.56 frames. ], batch size: 93, lr: 4.50e-03, grad_scale: 16.0 2023-04-17 07:29:27,604 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.840e+02 2.700e+02 3.084e+02 3.655e+02 6.677e+02, threshold=6.169e+02, percent-clipped=1.0 2023-04-17 07:30:09,170 INFO [train.py:893] (0/4) Epoch 29, batch 1000, loss[loss=0.1764, simple_loss=0.2362, pruned_loss=0.05834, over 13279.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2281, pruned_loss=0.05223, over 2644685.77 frames. ], batch size: 124, lr: 4.49e-03, grad_scale: 16.0 2023-04-17 07:30:09,351 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.7755, 5.2827, 5.1257, 5.2841, 5.1769, 5.0503, 5.7225, 5.2975], device='cuda:0'), covar=tensor([0.0659, 0.1180, 0.1965, 0.2071, 0.0862, 0.1673, 0.0745, 0.1077], device='cuda:0'), in_proj_covar=tensor([0.0356, 0.0406, 0.0500, 0.0502, 0.0321, 0.0375, 0.0464, 0.0378], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:30:11,087 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6845, 5.1006, 4.8836, 4.8833, 4.8896, 4.7077, 5.1274, 5.1449], device='cuda:0'), covar=tensor([0.0220, 0.0204, 0.0242, 0.0279, 0.0272, 0.0257, 0.0252, 0.0203], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0213, 0.0180, 0.0194, 0.0166, 0.0215, 0.0143, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 07:30:11,997 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1565, 4.4541, 4.1462, 4.2364, 4.3070, 4.6105, 4.4102, 4.2324], device='cuda:0'), covar=tensor([0.0308, 0.0318, 0.0388, 0.0873, 0.0263, 0.0239, 0.0328, 0.0337], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0169, 0.0194, 0.0278, 0.0191, 0.0209, 0.0187, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 07:30:28,473 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.25 vs. limit=5.0 2023-04-17 07:30:34,699 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.50 vs. limit=5.0 2023-04-17 07:30:55,835 INFO [train.py:893] (0/4) Epoch 29, batch 1050, loss[loss=0.1642, simple_loss=0.2345, pruned_loss=0.04697, over 13325.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2265, pruned_loss=0.05105, over 2648072.57 frames. ], batch size: 118, lr: 4.49e-03, grad_scale: 16.0 2023-04-17 07:30:56,025 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76625.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 07:30:58,263 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.799e+02 2.390e+02 2.885e+02 3.502e+02 6.067e+02, threshold=5.770e+02, percent-clipped=0.0 2023-04-17 07:31:17,259 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76650.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:31:40,180 INFO [train.py:893] (0/4) Epoch 29, batch 1100, loss[loss=0.1526, simple_loss=0.2215, pruned_loss=0.04181, over 13390.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2271, pruned_loss=0.05079, over 2652358.66 frames. ], batch size: 113, lr: 4.49e-03, grad_scale: 16.0 2023-04-17 07:31:43,801 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76679.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 07:31:56,579 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1839, 4.2992, 3.1950, 3.9331, 4.1625, 2.7597, 3.8565, 2.9030], device='cuda:0'), covar=tensor([0.0295, 0.0241, 0.0865, 0.0313, 0.0235, 0.1154, 0.0421, 0.1127], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0188, 0.0179, 0.0238, 0.0144, 0.0164, 0.0166, 0.0162], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:31:57,325 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1991, 4.0261, 4.0600, 2.4933, 4.4199, 4.2336, 4.1228, 4.3507], device='cuda:0'), covar=tensor([0.0244, 0.0170, 0.0168, 0.1130, 0.0160, 0.0280, 0.0180, 0.0115], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0063, 0.0091, 0.0108, 0.0107, 0.0120, 0.0089, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:32:26,134 INFO [train.py:893] (0/4) Epoch 29, batch 1150, loss[loss=0.1671, simple_loss=0.2399, pruned_loss=0.0471, over 13380.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.2269, pruned_loss=0.0502, over 2651489.17 frames. ], batch size: 109, lr: 4.49e-03, grad_scale: 16.0 2023-04-17 07:32:28,622 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.716e+02 2.373e+02 2.855e+02 3.407e+02 7.423e+02, threshold=5.711e+02, percent-clipped=5.0 2023-04-17 07:33:11,208 INFO [train.py:893] (0/4) Epoch 29, batch 1200, loss[loss=0.1629, simple_loss=0.2206, pruned_loss=0.05257, over 13382.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2271, pruned_loss=0.04977, over 2656393.30 frames. ], batch size: 67, lr: 4.49e-03, grad_scale: 16.0 2023-04-17 07:33:39,321 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-17 07:33:50,784 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-17 07:33:56,391 INFO [train.py:893] (0/4) Epoch 29, batch 1250, loss[loss=0.1739, simple_loss=0.2414, pruned_loss=0.05324, over 13259.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.227, pruned_loss=0.04988, over 2649488.19 frames. ], batch size: 124, lr: 4.49e-03, grad_scale: 16.0 2023-04-17 07:33:58,837 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.935e+02 2.494e+02 2.806e+02 3.320e+02 5.743e+02, threshold=5.612e+02, percent-clipped=1.0 2023-04-17 07:34:08,788 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76839.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:34:36,746 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-17 07:34:41,824 INFO [train.py:893] (0/4) Epoch 29, batch 1300, loss[loss=0.1479, simple_loss=0.2148, pruned_loss=0.04054, over 13519.00 frames. ], tot_loss[loss=0.164, simple_loss=0.228, pruned_loss=0.05004, over 2655386.17 frames. ], batch size: 70, lr: 4.49e-03, grad_scale: 16.0 2023-04-17 07:35:04,197 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76900.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:35:26,453 INFO [train.py:893] (0/4) Epoch 29, batch 1350, loss[loss=0.1614, simple_loss=0.2285, pruned_loss=0.04712, over 13396.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2288, pruned_loss=0.05033, over 2656893.18 frames. ], batch size: 113, lr: 4.48e-03, grad_scale: 16.0 2023-04-17 07:35:26,672 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76925.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 07:35:29,569 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.832e+02 2.606e+02 2.968e+02 3.524e+02 4.932e+02, threshold=5.935e+02, percent-clipped=0.0 2023-04-17 07:35:50,024 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76950.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:35:50,813 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3155, 5.1296, 5.3368, 5.0914, 5.6164, 5.1241, 5.6304, 5.5448], device='cuda:0'), covar=tensor([0.0398, 0.0537, 0.0598, 0.0543, 0.0468, 0.0863, 0.0373, 0.0468], device='cuda:0'), in_proj_covar=tensor([0.0319, 0.0316, 0.0330, 0.0246, 0.0469, 0.0368, 0.0308, 0.0326], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 07:36:10,265 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=76973.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 07:36:11,673 INFO [train.py:893] (0/4) Epoch 29, batch 1400, loss[loss=0.174, simple_loss=0.2382, pruned_loss=0.05487, over 13238.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2291, pruned_loss=0.05101, over 2654255.90 frames. ], batch size: 124, lr: 4.48e-03, grad_scale: 16.0 2023-04-17 07:36:15,993 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76979.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 07:36:33,345 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=76998.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:36:50,325 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.76 vs. limit=5.0 2023-04-17 07:36:57,269 INFO [train.py:893] (0/4) Epoch 29, batch 1450, loss[loss=0.1715, simple_loss=0.2408, pruned_loss=0.05104, over 13496.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2284, pruned_loss=0.05088, over 2658945.60 frames. ], batch size: 93, lr: 4.48e-03, grad_scale: 16.0 2023-04-17 07:36:59,086 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77027.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 07:37:00,374 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.817e+02 2.615e+02 3.065e+02 3.635e+02 5.318e+02, threshold=6.130e+02, percent-clipped=0.0 2023-04-17 07:37:17,162 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0392, 2.4206, 2.0011, 3.9010, 4.2953, 3.1823, 4.2254, 4.1123], device='cuda:0'), covar=tensor([0.0085, 0.1042, 0.1070, 0.0093, 0.0070, 0.0491, 0.0087, 0.0086], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0095, 0.0102, 0.0087, 0.0074, 0.0085, 0.0061, 0.0084], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 07:37:32,582 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.5211, 3.8826, 3.4469, 4.4371, 2.1470, 2.8307, 3.8738, 2.3520], device='cuda:0'), covar=tensor([0.0144, 0.0519, 0.0885, 0.0442, 0.1839, 0.1293, 0.0587, 0.1749], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0192, 0.0212, 0.0258, 0.0187, 0.0205, 0.0183, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:37:34,339 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.51 vs. limit=5.0 2023-04-17 07:37:43,764 INFO [train.py:893] (0/4) Epoch 29, batch 1500, loss[loss=0.1604, simple_loss=0.2268, pruned_loss=0.04696, over 13557.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2285, pruned_loss=0.05092, over 2661573.73 frames. ], batch size: 83, lr: 4.48e-03, grad_scale: 32.0 2023-04-17 07:38:02,571 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77095.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:38:29,887 INFO [train.py:893] (0/4) Epoch 29, batch 1550, loss[loss=0.1873, simple_loss=0.2541, pruned_loss=0.06021, over 13462.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2287, pruned_loss=0.05085, over 2663221.48 frames. ], batch size: 106, lr: 4.48e-03, grad_scale: 32.0 2023-04-17 07:38:32,483 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.779e+02 2.614e+02 2.982e+02 3.394e+02 6.130e+02, threshold=5.963e+02, percent-clipped=1.0 2023-04-17 07:38:58,510 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77156.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 07:39:14,434 INFO [train.py:893] (0/4) Epoch 29, batch 1600, loss[loss=0.1912, simple_loss=0.2637, pruned_loss=0.05939, over 13411.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2291, pruned_loss=0.05096, over 2664866.79 frames. ], batch size: 113, lr: 4.48e-03, grad_scale: 32.0 2023-04-17 07:39:17,206 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6325, 4.4444, 4.6658, 4.6278, 4.9164, 4.4454, 4.9274, 4.8995], device='cuda:0'), covar=tensor([0.0463, 0.0703, 0.0706, 0.0634, 0.0593, 0.0948, 0.0544, 0.0490], device='cuda:0'), in_proj_covar=tensor([0.0315, 0.0315, 0.0330, 0.0245, 0.0466, 0.0366, 0.0307, 0.0321], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:39:30,850 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9391, 3.8115, 3.0760, 3.4685, 3.0906, 2.4438, 3.8888, 2.3014], device='cuda:0'), covar=tensor([0.0650, 0.0420, 0.0563, 0.0360, 0.0679, 0.1678, 0.0706, 0.1219], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0152, 0.0142, 0.0123, 0.0154, 0.0194, 0.0191, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 07:39:33,770 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77195.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:39:36,494 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.0261, 3.9184, 3.2319, 3.5443, 3.1614, 2.5027, 3.9787, 2.3302], device='cuda:0'), covar=tensor([0.0681, 0.0491, 0.0515, 0.0419, 0.0698, 0.1949, 0.1015, 0.1401], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0151, 0.0142, 0.0123, 0.0154, 0.0194, 0.0191, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 07:39:44,323 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1973, 4.3339, 4.2188, 2.6180, 4.6384, 4.3924, 4.3415, 4.5886], device='cuda:0'), covar=tensor([0.0413, 0.0155, 0.0198, 0.1387, 0.0223, 0.0376, 0.0217, 0.0181], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0062, 0.0091, 0.0107, 0.0107, 0.0119, 0.0089, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:39:48,458 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77211.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:40:01,916 INFO [train.py:893] (0/4) Epoch 29, batch 1650, loss[loss=0.1522, simple_loss=0.2227, pruned_loss=0.04089, over 13255.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2295, pruned_loss=0.05069, over 2656268.98 frames. ], batch size: 124, lr: 4.48e-03, grad_scale: 32.0 2023-04-17 07:40:04,411 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.838e+02 2.449e+02 2.816e+02 3.387e+02 9.099e+02, threshold=5.633e+02, percent-clipped=2.0 2023-04-17 07:40:11,374 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.7949, 3.5021, 3.2281, 4.7174, 5.1626, 4.0131, 5.0161, 4.8722], device='cuda:0'), covar=tensor([0.0087, 0.0638, 0.0662, 0.0085, 0.0060, 0.0307, 0.0056, 0.0067], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0095, 0.0103, 0.0088, 0.0075, 0.0085, 0.0062, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 07:40:43,886 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77272.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:40:44,612 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77273.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:40:46,624 INFO [train.py:893] (0/4) Epoch 29, batch 1700, loss[loss=0.1486, simple_loss=0.217, pruned_loss=0.04005, over 13375.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2295, pruned_loss=0.0504, over 2655046.08 frames. ], batch size: 73, lr: 4.47e-03, grad_scale: 32.0 2023-04-17 07:40:49,108 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77277.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:41:05,366 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.9233, 3.8868, 2.9660, 3.4734, 3.1822, 2.4022, 3.9752, 2.1884], device='cuda:0'), covar=tensor([0.0763, 0.0454, 0.0623, 0.0425, 0.0670, 0.2023, 0.0824, 0.1570], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0152, 0.0142, 0.0123, 0.0154, 0.0195, 0.0191, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-04-17 07:41:25,718 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-17 07:41:32,661 INFO [train.py:893] (0/4) Epoch 29, batch 1750, loss[loss=0.1774, simple_loss=0.2406, pruned_loss=0.05714, over 13469.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2287, pruned_loss=0.05019, over 2659276.17 frames. ], batch size: 100, lr: 4.47e-03, grad_scale: 32.0 2023-04-17 07:41:34,725 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-17 07:41:35,060 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.926e+02 2.374e+02 2.741e+02 3.275e+02 6.260e+02, threshold=5.482e+02, percent-clipped=1.0 2023-04-17 07:41:40,275 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77334.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:41:44,170 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77338.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:42:10,699 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3981, 3.0550, 3.7783, 2.7607, 2.5455, 2.6826, 4.0307, 4.1387], device='cuda:0'), covar=tensor([0.1244, 0.2037, 0.0427, 0.1872, 0.1705, 0.1612, 0.0313, 0.0248], device='cuda:0'), in_proj_covar=tensor([0.0260, 0.0289, 0.0209, 0.0233, 0.0228, 0.0193, 0.0226, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:42:17,495 INFO [train.py:893] (0/4) Epoch 29, batch 1800, loss[loss=0.16, simple_loss=0.2182, pruned_loss=0.05092, over 13017.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2282, pruned_loss=0.04983, over 2663836.03 frames. ], batch size: 142, lr: 4.47e-03, grad_scale: 32.0 2023-04-17 07:42:21,688 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0568, 1.8436, 3.6210, 3.5620, 3.4707, 2.8449, 3.3499, 2.7453], device='cuda:0'), covar=tensor([0.2113, 0.1648, 0.0220, 0.0255, 0.0324, 0.0822, 0.0311, 0.0893], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0189, 0.0136, 0.0140, 0.0142, 0.0181, 0.0152, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 07:43:02,753 INFO [train.py:893] (0/4) Epoch 29, batch 1850, loss[loss=0.152, simple_loss=0.2228, pruned_loss=0.04063, over 13535.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2282, pruned_loss=0.05013, over 2656312.65 frames. ], batch size: 87, lr: 4.47e-03, grad_scale: 32.0 2023-04-17 07:43:05,116 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.962e+02 2.432e+02 2.941e+02 3.505e+02 5.866e+02, threshold=5.882e+02, percent-clipped=2.0 2023-04-17 07:43:06,712 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-17 07:43:26,596 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77451.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 07:43:48,648 INFO [train.py:893] (0/4) Epoch 29, batch 1900, loss[loss=0.1585, simple_loss=0.2273, pruned_loss=0.04482, over 13524.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2279, pruned_loss=0.05003, over 2656631.67 frames. ], batch size: 98, lr: 4.47e-03, grad_scale: 32.0 2023-04-17 07:43:51,411 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1569, 4.9782, 5.2405, 4.9638, 5.4719, 4.9830, 5.5151, 5.4417], device='cuda:0'), covar=tensor([0.0481, 0.0674, 0.0664, 0.0693, 0.0555, 0.0941, 0.0478, 0.0504], device='cuda:0'), in_proj_covar=tensor([0.0319, 0.0320, 0.0333, 0.0247, 0.0474, 0.0372, 0.0310, 0.0326], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 07:44:06,115 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77495.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:44:21,963 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-17 07:44:33,587 INFO [train.py:893] (0/4) Epoch 29, batch 1950, loss[loss=0.1384, simple_loss=0.1998, pruned_loss=0.03853, over 13377.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.227, pruned_loss=0.04971, over 2660076.97 frames. ], batch size: 62, lr: 4.47e-03, grad_scale: 32.0 2023-04-17 07:44:36,673 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.998e+02 2.561e+02 3.002e+02 3.542e+02 6.432e+02, threshold=6.003e+02, percent-clipped=1.0 2023-04-17 07:44:50,770 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77543.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:45:12,490 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77567.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:45:18,915 INFO [train.py:893] (0/4) Epoch 29, batch 2000, loss[loss=0.204, simple_loss=0.2673, pruned_loss=0.07034, over 13218.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2281, pruned_loss=0.05035, over 2658613.17 frames. ], batch size: 124, lr: 4.47e-03, grad_scale: 32.0 2023-04-17 07:45:23,985 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 07:45:47,414 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.2752, 2.9362, 3.5844, 2.6281, 2.3898, 2.5016, 3.9261, 3.9610], device='cuda:0'), covar=tensor([0.1189, 0.2046, 0.0413, 0.1936, 0.1794, 0.1816, 0.0329, 0.0218], device='cuda:0'), in_proj_covar=tensor([0.0259, 0.0287, 0.0209, 0.0231, 0.0227, 0.0192, 0.0223, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:45:49,773 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1669, 2.5688, 2.1498, 4.0595, 4.5027, 3.2907, 4.4444, 4.2664], device='cuda:0'), covar=tensor([0.0084, 0.0930, 0.1013, 0.0091, 0.0068, 0.0465, 0.0078, 0.0076], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0094, 0.0102, 0.0087, 0.0074, 0.0084, 0.0061, 0.0084], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 07:45:55,253 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77616.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:46:03,103 INFO [train.py:893] (0/4) Epoch 29, batch 2050, loss[loss=0.1847, simple_loss=0.2477, pruned_loss=0.06087, over 13540.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2297, pruned_loss=0.05125, over 2658188.95 frames. ], batch size: 85, lr: 4.46e-03, grad_scale: 32.0 2023-04-17 07:46:06,320 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.846e+02 2.483e+02 2.949e+02 3.550e+02 5.640e+02, threshold=5.898e+02, percent-clipped=0.0 2023-04-17 07:46:07,366 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77629.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:46:10,678 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77633.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:46:31,263 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6420, 3.7339, 2.5977, 3.3362, 3.5921, 2.3096, 3.2792, 2.4327], device='cuda:0'), covar=tensor([0.0382, 0.0297, 0.1192, 0.0411, 0.0289, 0.1390, 0.0583, 0.1507], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0190, 0.0181, 0.0240, 0.0146, 0.0164, 0.0166, 0.0163], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:46:47,899 INFO [train.py:893] (0/4) Epoch 29, batch 2100, loss[loss=0.1697, simple_loss=0.2319, pruned_loss=0.0537, over 13248.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.229, pruned_loss=0.05074, over 2659099.30 frames. ], batch size: 124, lr: 4.46e-03, grad_scale: 32.0 2023-04-17 07:46:49,831 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77677.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:46:49,851 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77677.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:47:34,415 INFO [train.py:893] (0/4) Epoch 29, batch 2150, loss[loss=0.1605, simple_loss=0.2293, pruned_loss=0.04581, over 13255.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2292, pruned_loss=0.05047, over 2662326.18 frames. ], batch size: 124, lr: 4.46e-03, grad_scale: 32.0 2023-04-17 07:47:36,702 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.954e+02 2.489e+02 2.950e+02 3.593e+02 6.767e+02, threshold=5.899e+02, percent-clipped=1.0 2023-04-17 07:47:45,453 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77738.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:47:48,842 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2023-04-17 07:47:57,295 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77751.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 07:48:19,626 INFO [train.py:893] (0/4) Epoch 29, batch 2200, loss[loss=0.1668, simple_loss=0.2374, pruned_loss=0.04809, over 13503.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2284, pruned_loss=0.05024, over 2654916.78 frames. ], batch size: 93, lr: 4.46e-03, grad_scale: 32.0 2023-04-17 07:48:41,186 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77799.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:49:05,602 INFO [train.py:893] (0/4) Epoch 29, batch 2250, loss[loss=0.1587, simple_loss=0.2143, pruned_loss=0.05153, over 13379.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2269, pruned_loss=0.04957, over 2655230.21 frames. ], batch size: 67, lr: 4.46e-03, grad_scale: 32.0 2023-04-17 07:49:08,042 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.824e+02 2.331e+02 2.696e+02 3.278e+02 5.521e+02, threshold=5.391e+02, percent-clipped=0.0 2023-04-17 07:49:38,560 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1467, 4.4203, 4.2240, 4.0487, 4.3168, 4.5292, 4.3624, 4.2896], device='cuda:0'), covar=tensor([0.0320, 0.0353, 0.0435, 0.1051, 0.0314, 0.0346, 0.0379, 0.0333], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0168, 0.0192, 0.0274, 0.0190, 0.0208, 0.0186, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 07:49:43,230 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77867.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:49:50,170 INFO [train.py:893] (0/4) Epoch 29, batch 2300, loss[loss=0.1834, simple_loss=0.2496, pruned_loss=0.05863, over 13480.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2266, pruned_loss=0.04914, over 2661666.53 frames. ], batch size: 93, lr: 4.46e-03, grad_scale: 32.0 2023-04-17 07:50:24,569 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-17 07:50:27,413 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77915.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:50:35,385 INFO [train.py:893] (0/4) Epoch 29, batch 2350, loss[loss=0.1522, simple_loss=0.208, pruned_loss=0.04817, over 13449.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2262, pruned_loss=0.04919, over 2661923.00 frames. ], batch size: 65, lr: 4.46e-03, grad_scale: 32.0 2023-04-17 07:50:38,564 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.953e+02 2.391e+02 2.757e+02 3.275e+02 5.484e+02, threshold=5.514e+02, percent-clipped=1.0 2023-04-17 07:50:39,674 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77929.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:50:42,801 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77933.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:50:51,016 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-17 07:50:59,347 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 07:51:11,861 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-17 07:51:18,148 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77972.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:51:20,411 INFO [train.py:893] (0/4) Epoch 29, batch 2400, loss[loss=0.1527, simple_loss=0.2228, pruned_loss=0.04125, over 13560.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2254, pruned_loss=0.04895, over 2660333.03 frames. ], batch size: 78, lr: 4.45e-03, grad_scale: 32.0 2023-04-17 07:51:23,585 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77977.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:51:26,991 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77981.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:51:44,336 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-78000.pt 2023-04-17 07:51:55,945 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7837, 2.5680, 2.5043, 2.8787, 2.1370, 2.8955, 2.8485, 2.3903], device='cuda:0'), covar=tensor([0.0081, 0.0202, 0.0175, 0.0176, 0.0282, 0.0140, 0.0197, 0.0218], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0132, 0.0136, 0.0138, 0.0146, 0.0125, 0.0118, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 07:52:10,233 INFO [train.py:893] (0/4) Epoch 29, batch 2450, loss[loss=0.1588, simple_loss=0.225, pruned_loss=0.04632, over 13458.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2254, pruned_loss=0.04887, over 2658134.01 frames. ], batch size: 103, lr: 4.45e-03, grad_scale: 32.0 2023-04-17 07:52:12,746 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.799e+02 2.344e+02 2.892e+02 3.560e+02 6.451e+02, threshold=5.784e+02, percent-clipped=2.0 2023-04-17 07:52:17,993 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78033.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:52:56,050 INFO [train.py:893] (0/4) Epoch 29, batch 2500, loss[loss=0.167, simple_loss=0.2338, pruned_loss=0.05015, over 13532.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2256, pruned_loss=0.04882, over 2660600.06 frames. ], batch size: 91, lr: 4.45e-03, grad_scale: 32.0 2023-04-17 07:53:10,150 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3710, 2.1824, 2.1446, 2.4507, 1.8800, 2.4609, 2.4209, 2.0546], device='cuda:0'), covar=tensor([0.0130, 0.0270, 0.0192, 0.0163, 0.0258, 0.0188, 0.0231, 0.0259], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0132, 0.0136, 0.0138, 0.0146, 0.0126, 0.0118, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 07:53:41,530 INFO [train.py:893] (0/4) Epoch 29, batch 2550, loss[loss=0.1419, simple_loss=0.2076, pruned_loss=0.03808, over 13351.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2256, pruned_loss=0.04888, over 2659132.09 frames. ], batch size: 67, lr: 4.45e-03, grad_scale: 32.0 2023-04-17 07:53:43,950 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.828e+02 2.449e+02 2.838e+02 3.334e+02 6.416e+02, threshold=5.677e+02, percent-clipped=2.0 2023-04-17 07:54:05,010 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 07:54:26,958 INFO [train.py:893] (0/4) Epoch 29, batch 2600, loss[loss=0.1761, simple_loss=0.2292, pruned_loss=0.06154, over 11766.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.2249, pruned_loss=0.04909, over 2654892.40 frames. ], batch size: 157, lr: 4.45e-03, grad_scale: 32.0 2023-04-17 07:54:28,988 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7433, 4.0038, 3.0775, 2.7822, 2.7861, 2.4735, 4.1083, 2.2931], device='cuda:0'), covar=tensor([0.2113, 0.0385, 0.1426, 0.2542, 0.1062, 0.3678, 0.0330, 0.4395], device='cuda:0'), in_proj_covar=tensor([0.0322, 0.0303, 0.0337, 0.0359, 0.0273, 0.0344, 0.0232, 0.0382], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 07:54:36,104 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1405, 3.9498, 4.0786, 2.5203, 4.4110, 4.1902, 4.1841, 4.3796], device='cuda:0'), covar=tensor([0.0253, 0.0166, 0.0146, 0.1073, 0.0149, 0.0257, 0.0134, 0.0117], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0063, 0.0092, 0.0109, 0.0108, 0.0121, 0.0090, 0.0086], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:54:58,373 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5957, 3.7213, 2.6616, 3.2492, 3.6397, 2.4727, 3.2663, 2.5545], device='cuda:0'), covar=tensor([0.0337, 0.0346, 0.1043, 0.0395, 0.0332, 0.1267, 0.0578, 0.1209], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0189, 0.0181, 0.0239, 0.0146, 0.0164, 0.0166, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:55:08,654 INFO [train.py:893] (0/4) Epoch 29, batch 2650, loss[loss=0.1512, simple_loss=0.2195, pruned_loss=0.04147, over 13192.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2262, pruned_loss=0.04991, over 2650280.48 frames. ], batch size: 132, lr: 4.45e-03, grad_scale: 32.0 2023-04-17 07:55:10,857 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.961e+02 2.802e+02 3.099e+02 3.581e+02 5.823e+02, threshold=6.198e+02, percent-clipped=1.0 2023-04-17 07:55:13,860 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78232.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:55:41,897 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0735, 4.3618, 4.0921, 4.1696, 4.1982, 4.4972, 4.3195, 4.1427], device='cuda:0'), covar=tensor([0.0290, 0.0302, 0.0386, 0.0776, 0.0325, 0.0240, 0.0374, 0.0366], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0168, 0.0191, 0.0272, 0.0190, 0.0205, 0.0186, 0.0115], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 07:55:44,005 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78272.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:55:46,235 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-29.pt 2023-04-17 07:56:11,353 WARNING [train.py:1054] (0/4) Exclude cut with ID db3105debd557279a777dee8a84ecb323b3b63f83e1d939111fe3f232d72bf04767da13a8383344d428a5a9ad767c044bce3837e3623d98f38591cac317d2932-common_voice_en_21838523 from training. Duration: 21.6 2023-04-17 07:56:14,651 INFO [train.py:893] (0/4) Epoch 30, batch 0, loss[loss=0.1646, simple_loss=0.2318, pruned_loss=0.04872, over 13531.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2318, pruned_loss=0.04872, over 13531.00 frames. ], batch size: 83, lr: 4.37e-03, grad_scale: 32.0 2023-04-17 07:56:14,652 INFO [train.py:918] (0/4) Computing validation loss 2023-04-17 07:56:20,178 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.1380, 4.4331, 3.2155, 2.9237, 3.0931, 2.6150, 4.5929, 2.4608], device='cuda:0'), covar=tensor([0.2018, 0.0351, 0.1661, 0.2729, 0.1029, 0.3936, 0.0259, 0.4953], device='cuda:0'), in_proj_covar=tensor([0.0324, 0.0305, 0.0340, 0.0362, 0.0275, 0.0347, 0.0233, 0.0384], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 07:56:35,052 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.9871, 3.2203, 4.7685, 5.1270, 3.4728, 3.3494, 5.3125, 5.4489], device='cuda:0'), covar=tensor([0.0337, 0.3176, 0.0370, 0.0349, 0.0941, 0.0961, 0.0113, 0.0153], device='cuda:0'), in_proj_covar=tensor([0.0259, 0.0286, 0.0209, 0.0231, 0.0227, 0.0192, 0.0224, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:56:36,838 INFO [train.py:927] (0/4) Epoch 30, validation: loss=0.1309, simple_loss=0.1907, pruned_loss=0.03552, over 2446609.00 frames. 2023-04-17 07:56:36,838 INFO [train.py:928] (0/4) Maximum memory allocated so far is 13044MB 2023-04-17 07:56:54,944 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78293.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 07:56:56,488 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1653, 4.6186, 4.4229, 4.4200, 4.4554, 4.2270, 4.6889, 4.6824], device='cuda:0'), covar=tensor([0.0236, 0.0217, 0.0273, 0.0352, 0.0309, 0.0279, 0.0231, 0.0248], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0218, 0.0183, 0.0200, 0.0169, 0.0220, 0.0148, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 07:57:19,892 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=78320.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:57:23,064 INFO [train.py:893] (0/4) Epoch 30, batch 50, loss[loss=0.182, simple_loss=0.2386, pruned_loss=0.06275, over 11691.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2233, pruned_loss=0.05091, over 602190.40 frames. ], batch size: 157, lr: 4.37e-03, grad_scale: 32.0 2023-04-17 07:57:23,416 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.7785, 3.8096, 2.7931, 3.5054, 3.7437, 2.6050, 3.4692, 2.5552], device='cuda:0'), covar=tensor([0.0293, 0.0221, 0.0960, 0.0345, 0.0310, 0.1095, 0.0553, 0.1224], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0189, 0.0182, 0.0239, 0.0146, 0.0164, 0.0166, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:57:26,331 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.743e+02 2.510e+02 3.075e+02 3.612e+02 7.972e+02, threshold=6.151e+02, percent-clipped=2.0 2023-04-17 07:57:31,304 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78333.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:57:47,341 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233310 from training. Duration: 33.876 2023-04-17 07:57:47,341 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233451 from training. Duration: 20.556 2023-04-17 07:57:47,341 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27233710 from training. Duration: 49.932 2023-04-17 07:57:47,348 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27289387 from training. Duration: 27.792 2023-04-17 07:57:47,363 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27312193 from training. Duration: 23.22 2023-04-17 07:57:47,383 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27367777 from training. Duration: 27.0 2023-04-17 07:57:48,121 WARNING [train.py:1054] (0/4) Exclude cut with ID 416dc03bdc0085ce8ea41524449ef5421eed43264a1c42a4fc14300111a73fbe13396d018c6972fa5ec859c678bb15e077675dc399e12f637812c624d6e993dc-common_voice_en_27371830 from training. Duration: 23.688 2023-04-17 07:58:08,162 INFO [train.py:893] (0/4) Epoch 30, batch 100, loss[loss=0.1514, simple_loss=0.2167, pruned_loss=0.0431, over 13459.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2253, pruned_loss=0.05096, over 1055669.04 frames. ], batch size: 100, lr: 4.37e-03, grad_scale: 32.0 2023-04-17 07:58:11,481 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0349, 3.8250, 3.9288, 2.2973, 4.2001, 4.0413, 4.0618, 4.1898], device='cuda:0'), covar=tensor([0.0215, 0.0144, 0.0144, 0.1166, 0.0130, 0.0255, 0.0122, 0.0113], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0063, 0.0091, 0.0108, 0.0108, 0.0121, 0.0090, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 07:58:15,425 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=78381.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:58:52,347 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-17 07:58:54,012 INFO [train.py:893] (0/4) Epoch 30, batch 150, loss[loss=0.1782, simple_loss=0.2261, pruned_loss=0.06518, over 10812.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.227, pruned_loss=0.05182, over 1403312.47 frames. ], batch size: 44, lr: 4.37e-03, grad_scale: 32.0 2023-04-17 07:58:58,105 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.972e+02 2.511e+02 2.997e+02 3.677e+02 5.041e+02, threshold=5.994e+02, percent-clipped=0.0 2023-04-17 07:59:27,617 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78460.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 07:59:40,913 INFO [train.py:893] (0/4) Epoch 30, batch 200, loss[loss=0.1757, simple_loss=0.2414, pruned_loss=0.05498, over 13478.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2282, pruned_loss=0.05225, over 1669834.64 frames. ], batch size: 81, lr: 4.36e-03, grad_scale: 32.0 2023-04-17 08:00:23,024 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78521.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:00:25,838 INFO [train.py:893] (0/4) Epoch 30, batch 250, loss[loss=0.1897, simple_loss=0.2462, pruned_loss=0.06665, over 13435.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2293, pruned_loss=0.05249, over 1890713.46 frames. ], batch size: 106, lr: 4.36e-03, grad_scale: 32.0 2023-04-17 08:00:29,096 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.795e+02 2.445e+02 2.983e+02 3.638e+02 6.878e+02, threshold=5.966e+02, percent-clipped=2.0 2023-04-17 08:00:41,815 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0965, 3.9020, 3.9997, 2.4809, 4.3502, 4.1342, 4.1475, 4.3339], device='cuda:0'), covar=tensor([0.0246, 0.0153, 0.0152, 0.1116, 0.0135, 0.0253, 0.0142, 0.0104], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0063, 0.0092, 0.0108, 0.0108, 0.0122, 0.0090, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:00:52,392 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.8551, 3.8930, 2.8992, 3.5903, 3.8725, 2.5702, 3.4952, 2.6841], device='cuda:0'), covar=tensor([0.0291, 0.0212, 0.0944, 0.0378, 0.0247, 0.1156, 0.0558, 0.1201], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0189, 0.0181, 0.0240, 0.0146, 0.0164, 0.0167, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:00:54,678 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78556.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:01:11,324 INFO [train.py:893] (0/4) Epoch 30, batch 300, loss[loss=0.1755, simple_loss=0.241, pruned_loss=0.05501, over 13529.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2293, pruned_loss=0.05177, over 2067143.82 frames. ], batch size: 83, lr: 4.36e-03, grad_scale: 32.0 2023-04-17 08:01:19,288 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.6785, 3.4298, 4.1026, 3.0412, 2.8087, 2.9081, 4.4383, 4.5470], device='cuda:0'), covar=tensor([0.1174, 0.1796, 0.0375, 0.1683, 0.1607, 0.1578, 0.0267, 0.0282], device='cuda:0'), in_proj_covar=tensor([0.0259, 0.0287, 0.0209, 0.0232, 0.0227, 0.0192, 0.0224, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:01:23,814 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78588.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 08:01:50,824 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78617.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:01:56,441 INFO [train.py:893] (0/4) Epoch 30, batch 350, loss[loss=0.1512, simple_loss=0.2029, pruned_loss=0.04977, over 12601.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2301, pruned_loss=0.05242, over 2197684.30 frames. ], batch size: 51, lr: 4.36e-03, grad_scale: 32.0 2023-04-17 08:02:00,469 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.006e+02 2.600e+02 2.996e+02 3.438e+02 5.374e+02, threshold=5.992e+02, percent-clipped=0.0 2023-04-17 08:02:38,501 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.1700, 2.1720, 3.8924, 3.8210, 3.7409, 3.0746, 3.6115, 3.0970], device='cuda:0'), covar=tensor([0.2021, 0.1479, 0.0169, 0.0175, 0.0233, 0.0746, 0.0260, 0.0707], device='cuda:0'), in_proj_covar=tensor([0.0216, 0.0185, 0.0133, 0.0138, 0.0140, 0.0178, 0.0149, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 08:02:42,442 INFO [train.py:893] (0/4) Epoch 30, batch 400, loss[loss=0.1723, simple_loss=0.2284, pruned_loss=0.05814, over 13345.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2311, pruned_loss=0.05263, over 2301279.29 frames. ], batch size: 73, lr: 4.36e-03, grad_scale: 32.0 2023-04-17 08:03:04,340 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.2542, 4.8192, 4.6493, 4.7272, 4.5774, 4.5906, 5.2024, 4.8183], device='cuda:0'), covar=tensor([0.0706, 0.1226, 0.2325, 0.2598, 0.1114, 0.1734, 0.0903, 0.1181], device='cuda:0'), in_proj_covar=tensor([0.0361, 0.0410, 0.0504, 0.0506, 0.0325, 0.0379, 0.0468, 0.0382], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:03:27,772 INFO [train.py:893] (0/4) Epoch 30, batch 450, loss[loss=0.2564, simple_loss=0.3169, pruned_loss=0.09795, over 11900.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2327, pruned_loss=0.05318, over 2382481.36 frames. ], batch size: 158, lr: 4.36e-03, grad_scale: 32.0 2023-04-17 08:03:32,443 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.010e+02 2.627e+02 2.942e+02 3.377e+02 5.601e+02, threshold=5.884e+02, percent-clipped=0.0 2023-04-17 08:03:53,003 WARNING [train.py:1054] (0/4) Exclude cut with ID c3c070732f2c104d269cc54d36a8684bdb15120f468e145af8bc6bcd0d942d561341b6d9e6c6f1cdd7f1a56d60c8536c3afea58ec24eaf7eb9629a8f96995f2d-common_voice_en_619791 from training. Duration: 20.496 2023-04-17 08:03:53,311 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.3663, 2.9078, 2.9252, 3.2560, 2.5721, 3.3723, 3.2996, 2.8257], device='cuda:0'), covar=tensor([0.0080, 0.0220, 0.0148, 0.0248, 0.0225, 0.0123, 0.0161, 0.0156], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0132, 0.0136, 0.0139, 0.0147, 0.0126, 0.0119, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 08:03:55,052 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-17 08:04:13,297 INFO [train.py:893] (0/4) Epoch 30, batch 500, loss[loss=0.1473, simple_loss=0.2064, pruned_loss=0.04412, over 13392.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2322, pruned_loss=0.05308, over 2443951.48 frames. ], batch size: 62, lr: 4.36e-03, grad_scale: 32.0 2023-04-17 08:04:36,165 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.3399, 4.8423, 4.7475, 4.8781, 4.6302, 4.6829, 5.2928, 4.8785], device='cuda:0'), covar=tensor([0.0725, 0.1272, 0.2190, 0.2275, 0.1111, 0.1536, 0.0876, 0.1276], device='cuda:0'), in_proj_covar=tensor([0.0359, 0.0408, 0.0502, 0.0502, 0.0323, 0.0377, 0.0465, 0.0380], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:04:51,920 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78816.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:04:57,630 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.6726, 2.7940, 2.3623, 1.7667, 1.7301, 2.4017, 2.4070, 2.9982], device='cuda:0'), covar=tensor([0.1187, 0.0362, 0.0757, 0.1466, 0.0268, 0.0468, 0.0802, 0.0218], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0158, 0.0133, 0.0215, 0.0119, 0.0174, 0.0186, 0.0143], device='cuda:0'), out_proj_covar=tensor([1.3311e-04, 1.1805e-04, 1.0190e-04, 1.5915e-04, 8.5285e-05, 1.3113e-04, 1.3984e-04, 1.0536e-04], device='cuda:0') 2023-04-17 08:05:00,451 INFO [train.py:893] (0/4) Epoch 30, batch 550, loss[loss=0.142, simple_loss=0.2042, pruned_loss=0.03989, over 13412.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2316, pruned_loss=0.05252, over 2492451.93 frames. ], batch size: 65, lr: 4.35e-03, grad_scale: 32.0 2023-04-17 08:05:03,824 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.765e+02 2.525e+02 2.925e+02 3.340e+02 5.483e+02, threshold=5.851e+02, percent-clipped=0.0 2023-04-17 08:05:19,697 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.5432, 5.0866, 4.9498, 5.1044, 4.8279, 4.9226, 5.4829, 5.0421], device='cuda:0'), covar=tensor([0.0718, 0.1244, 0.2036, 0.2142, 0.1071, 0.1536, 0.0908, 0.1234], device='cuda:0'), in_proj_covar=tensor([0.0360, 0.0409, 0.0503, 0.0503, 0.0325, 0.0378, 0.0467, 0.0381], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:05:45,186 INFO [train.py:893] (0/4) Epoch 30, batch 600, loss[loss=0.1616, simple_loss=0.2254, pruned_loss=0.04893, over 13471.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2299, pruned_loss=0.05206, over 2530638.31 frames. ], batch size: 100, lr: 4.35e-03, grad_scale: 32.0 2023-04-17 08:05:58,886 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78888.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:06:19,264 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78912.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:06:31,204 INFO [train.py:893] (0/4) Epoch 30, batch 650, loss[loss=0.1599, simple_loss=0.2259, pruned_loss=0.04694, over 13443.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.23, pruned_loss=0.05213, over 2560872.56 frames. ], batch size: 106, lr: 4.35e-03, grad_scale: 32.0 2023-04-17 08:06:34,572 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.693e+02 2.526e+02 3.052e+02 3.686e+02 4.958e+02, threshold=6.105e+02, percent-clipped=0.0 2023-04-17 08:06:42,049 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=78936.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:07:08,150 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78965.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:07:15,826 INFO [train.py:893] (0/4) Epoch 30, batch 700, loss[loss=0.1682, simple_loss=0.2296, pruned_loss=0.05343, over 13553.00 frames. ], tot_loss[loss=0.166, simple_loss=0.229, pruned_loss=0.0515, over 2582307.77 frames. ], batch size: 87, lr: 4.35e-03, grad_scale: 32.0 2023-04-17 08:07:45,711 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6528, 4.9550, 4.6572, 4.5575, 4.7654, 4.9933, 4.8823, 4.7466], device='cuda:0'), covar=tensor([0.0305, 0.0285, 0.0392, 0.1056, 0.0294, 0.0281, 0.0334, 0.0210], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0168, 0.0192, 0.0273, 0.0190, 0.0206, 0.0186, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 08:08:02,129 INFO [train.py:893] (0/4) Epoch 30, batch 750, loss[loss=0.1747, simple_loss=0.2306, pruned_loss=0.05936, over 13496.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2288, pruned_loss=0.05146, over 2604822.21 frames. ], batch size: 70, lr: 4.35e-03, grad_scale: 32.0 2023-04-17 08:08:04,071 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79026.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:08:05,377 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.768e+02 2.504e+02 2.968e+02 3.500e+02 7.955e+02, threshold=5.936e+02, percent-clipped=1.0 2023-04-17 08:08:28,507 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1802, 4.0108, 4.1200, 2.4905, 4.5055, 4.2909, 4.2282, 4.4618], device='cuda:0'), covar=tensor([0.0265, 0.0159, 0.0159, 0.1090, 0.0138, 0.0274, 0.0151, 0.0119], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0063, 0.0091, 0.0108, 0.0107, 0.0121, 0.0090, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:08:48,614 INFO [train.py:893] (0/4) Epoch 30, batch 800, loss[loss=0.167, simple_loss=0.2336, pruned_loss=0.05015, over 13506.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2297, pruned_loss=0.05183, over 2615661.00 frames. ], batch size: 81, lr: 4.35e-03, grad_scale: 64.0 2023-04-17 08:09:19,737 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.0958, 4.4123, 4.1682, 4.2210, 4.2929, 4.5359, 4.3943, 4.2219], device='cuda:0'), covar=tensor([0.0291, 0.0253, 0.0327, 0.0774, 0.0240, 0.0217, 0.0248, 0.0323], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0168, 0.0193, 0.0274, 0.0191, 0.0207, 0.0187, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 08:09:26,318 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79116.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:09:29,992 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-17 08:09:33,371 INFO [train.py:893] (0/4) Epoch 30, batch 850, loss[loss=0.1595, simple_loss=0.2253, pruned_loss=0.0469, over 13529.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2301, pruned_loss=0.0519, over 2629206.69 frames. ], batch size: 91, lr: 4.35e-03, grad_scale: 32.0 2023-04-17 08:09:37,515 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.639e+02 2.521e+02 3.034e+02 3.558e+02 4.951e+02, threshold=6.069e+02, percent-clipped=0.0 2023-04-17 08:10:09,930 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79164.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:10:13,549 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-17 08:10:18,665 INFO [train.py:893] (0/4) Epoch 30, batch 900, loss[loss=0.1711, simple_loss=0.2357, pruned_loss=0.05326, over 13377.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2308, pruned_loss=0.0524, over 2638724.04 frames. ], batch size: 118, lr: 4.35e-03, grad_scale: 32.0 2023-04-17 08:10:50,346 WARNING [train.py:1054] (0/4) Exclude cut with ID 04b63653596f689ef7216c2603c0f6523a707a0b3a695f6f7adb7314161502892f2f2d7a577d62fd2ea21a9b72d688e7f4f12ff3c26432edad22541afe263dbb-common_voice_en_530541 from training. Duration: 24.672 2023-04-17 08:10:52,960 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79212.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:11:03,401 INFO [train.py:893] (0/4) Epoch 30, batch 950, loss[loss=0.1669, simple_loss=0.233, pruned_loss=0.05039, over 13437.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2297, pruned_loss=0.05234, over 2644428.99 frames. ], batch size: 106, lr: 4.34e-03, grad_scale: 32.0 2023-04-17 08:11:08,147 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.773e+02 2.620e+02 3.020e+02 3.590e+02 5.764e+02, threshold=6.040e+02, percent-clipped=0.0 2023-04-17 08:11:13,331 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2229, 3.5642, 3.5026, 3.9369, 2.2468, 3.0797, 3.7891, 2.2644], device='cuda:0'), covar=tensor([0.0168, 0.0588, 0.0812, 0.0579, 0.1618, 0.0901, 0.0562, 0.1576], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0196, 0.0213, 0.0261, 0.0190, 0.0207, 0.0185, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:11:36,628 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79260.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:11:48,942 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9826, 4.0004, 2.7850, 3.6303, 3.9643, 2.6494, 3.5971, 2.7007], device='cuda:0'), covar=tensor([0.0286, 0.0279, 0.1018, 0.0387, 0.0262, 0.1187, 0.0536, 0.1222], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0189, 0.0180, 0.0241, 0.0146, 0.0164, 0.0167, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:11:49,414 INFO [train.py:893] (0/4) Epoch 30, batch 1000, loss[loss=0.1586, simple_loss=0.2235, pruned_loss=0.0469, over 13523.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2281, pruned_loss=0.05175, over 2648083.63 frames. ], batch size: 98, lr: 4.34e-03, grad_scale: 32.0 2023-04-17 08:11:58,641 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9290, 3.9830, 2.7672, 3.5112, 3.9112, 2.5925, 3.5966, 2.7235], device='cuda:0'), covar=tensor([0.0291, 0.0239, 0.1017, 0.0434, 0.0316, 0.1264, 0.0540, 0.1277], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0188, 0.0179, 0.0240, 0.0145, 0.0163, 0.0167, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:11:59,338 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79285.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 08:12:03,599 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2023-04-17 08:12:32,599 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79321.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:12:34,800 INFO [train.py:893] (0/4) Epoch 30, batch 1050, loss[loss=0.1647, simple_loss=0.2284, pruned_loss=0.05044, over 13532.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2268, pruned_loss=0.05095, over 2645616.25 frames. ], batch size: 78, lr: 4.34e-03, grad_scale: 32.0 2023-04-17 08:12:40,142 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.832e+02 2.591e+02 3.017e+02 3.726e+02 6.684e+02, threshold=6.035e+02, percent-clipped=2.0 2023-04-17 08:12:43,786 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79333.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:12:47,039 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.1543, 4.0907, 4.1519, 2.4849, 4.4786, 4.3007, 4.2416, 4.4816], device='cuda:0'), covar=tensor([0.0264, 0.0132, 0.0136, 0.1100, 0.0149, 0.0221, 0.0157, 0.0094], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0062, 0.0091, 0.0108, 0.0108, 0.0121, 0.0090, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:12:55,111 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79346.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 08:13:12,157 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79364.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:13:20,290 INFO [train.py:893] (0/4) Epoch 30, batch 1100, loss[loss=0.1509, simple_loss=0.2124, pruned_loss=0.04474, over 13402.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2267, pruned_loss=0.05061, over 2649801.51 frames. ], batch size: 84, lr: 4.34e-03, grad_scale: 32.0 2023-04-17 08:13:21,456 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79375.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 08:13:39,865 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79394.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:14:07,258 INFO [train.py:893] (0/4) Epoch 30, batch 1150, loss[loss=0.1693, simple_loss=0.2334, pruned_loss=0.05259, over 13432.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2269, pruned_loss=0.05014, over 2648668.35 frames. ], batch size: 103, lr: 4.34e-03, grad_scale: 32.0 2023-04-17 08:14:08,358 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79425.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:14:11,340 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.798e+02 2.546e+02 3.063e+02 3.476e+02 8.677e+02, threshold=6.127e+02, percent-clipped=1.0 2023-04-17 08:14:17,412 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79436.0, num_to_drop=1, layers_to_drop={3} 2023-04-17 08:14:51,387 INFO [train.py:893] (0/4) Epoch 30, batch 1200, loss[loss=0.1589, simple_loss=0.2244, pruned_loss=0.04668, over 13476.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.2275, pruned_loss=0.04998, over 2649545.98 frames. ], batch size: 79, lr: 4.34e-03, grad_scale: 32.0 2023-04-17 08:15:19,165 WARNING [train.py:1054] (0/4) Exclude cut with ID 48528338442073e08f7886e133be1f7f773562a5c0257c0f3c33e19edf5de29dcefe46b9a0e497820295177ae021e53c9518185178f98612c169580ff3a07084-common_voice_en_19393114 from training. Duration: 20.016 2023-04-17 08:15:31,933 WARNING [train.py:1054] (0/4) Exclude cut with ID c79cca7d4c2127ae9a701febbb70a6a67975417cf081587b35693f0d90245c00a68de59daa3af7798bfd7df858f7e1b43acc3ca7a6f2419bdfa14ae128150d0b-common_voice_en_170142 from training. Duration: 20.16 2023-04-17 08:15:37,639 INFO [train.py:893] (0/4) Epoch 30, batch 1250, loss[loss=0.1586, simple_loss=0.2284, pruned_loss=0.04441, over 13465.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2277, pruned_loss=0.05008, over 2650410.86 frames. ], batch size: 100, lr: 4.34e-03, grad_scale: 32.0 2023-04-17 08:15:38,629 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4334, 4.1912, 4.4179, 4.4140, 4.6192, 4.1834, 4.6516, 4.5794], device='cuda:0'), covar=tensor([0.0484, 0.0638, 0.0628, 0.0519, 0.0654, 0.0832, 0.0498, 0.0475], device='cuda:0'), in_proj_covar=tensor([0.0318, 0.0319, 0.0329, 0.0244, 0.0474, 0.0369, 0.0309, 0.0328], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 08:15:41,734 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.762e+02 2.699e+02 3.086e+02 3.611e+02 7.494e+02, threshold=6.172e+02, percent-clipped=2.0 2023-04-17 08:15:58,373 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.1356, 4.8807, 5.1414, 5.0256, 5.3895, 4.9286, 5.4147, 5.3513], device='cuda:0'), covar=tensor([0.0384, 0.0551, 0.0607, 0.0490, 0.0488, 0.0809, 0.0403, 0.0416], device='cuda:0'), in_proj_covar=tensor([0.0317, 0.0319, 0.0329, 0.0244, 0.0474, 0.0369, 0.0310, 0.0328], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-04-17 08:16:23,292 INFO [train.py:893] (0/4) Epoch 30, batch 1300, loss[loss=0.177, simple_loss=0.2258, pruned_loss=0.06409, over 10882.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2287, pruned_loss=0.05065, over 2653919.32 frames. ], batch size: 44, lr: 4.33e-03, grad_scale: 32.0 2023-04-17 08:16:34,058 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79585.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:17:07,015 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79621.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:17:09,206 INFO [train.py:893] (0/4) Epoch 30, batch 1350, loss[loss=0.1676, simple_loss=0.2335, pruned_loss=0.05088, over 13466.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2298, pruned_loss=0.05085, over 2659042.15 frames. ], batch size: 100, lr: 4.33e-03, grad_scale: 32.0 2023-04-17 08:17:13,986 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.034e+02 2.559e+02 2.996e+02 3.614e+02 6.449e+02, threshold=5.992e+02, percent-clipped=1.0 2023-04-17 08:17:25,403 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79641.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 08:17:30,314 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79646.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:17:39,153 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79657.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:17:50,404 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79669.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:17:55,582 INFO [train.py:893] (0/4) Epoch 30, batch 1400, loss[loss=0.1747, simple_loss=0.2387, pruned_loss=0.05541, over 11653.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.23, pruned_loss=0.05108, over 2657618.71 frames. ], batch size: 157, lr: 4.33e-03, grad_scale: 32.0 2023-04-17 08:18:02,497 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-17 08:18:08,740 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79689.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:18:34,654 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79718.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:18:36,862 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79720.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:18:40,081 INFO [train.py:893] (0/4) Epoch 30, batch 1450, loss[loss=0.1621, simple_loss=0.2241, pruned_loss=0.05005, over 13536.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2291, pruned_loss=0.05107, over 2659674.33 frames. ], batch size: 87, lr: 4.33e-03, grad_scale: 32.0 2023-04-17 08:18:44,693 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 2.494e+02 2.889e+02 3.354e+02 4.257e+02, threshold=5.778e+02, percent-clipped=0.0 2023-04-17 08:18:46,622 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79731.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 08:19:14,840 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79762.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:19:25,899 INFO [train.py:893] (0/4) Epoch 30, batch 1500, loss[loss=0.1651, simple_loss=0.2284, pruned_loss=0.05092, over 13538.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2292, pruned_loss=0.05094, over 2658591.76 frames. ], batch size: 98, lr: 4.33e-03, grad_scale: 32.0 2023-04-17 08:19:45,114 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-17 08:20:07,391 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([1.9398, 1.7775, 3.6865, 3.5971, 3.5542, 2.8860, 3.3153, 2.9011], device='cuda:0'), covar=tensor([0.2074, 0.1671, 0.0184, 0.0197, 0.0231, 0.0757, 0.0290, 0.0753], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0189, 0.0136, 0.0140, 0.0141, 0.0181, 0.0152, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 08:20:11,425 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79823.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:20:11,886 INFO [train.py:893] (0/4) Epoch 30, batch 1550, loss[loss=0.1945, simple_loss=0.2501, pruned_loss=0.06946, over 11564.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2279, pruned_loss=0.05034, over 2657425.93 frames. ], batch size: 158, lr: 4.33e-03, grad_scale: 32.0 2023-04-17 08:20:16,715 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.723e+02 2.643e+02 2.905e+02 3.656e+02 6.306e+02, threshold=5.811e+02, percent-clipped=2.0 2023-04-17 08:20:57,763 INFO [train.py:893] (0/4) Epoch 30, batch 1600, loss[loss=0.14, simple_loss=0.2042, pruned_loss=0.03787, over 13535.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2284, pruned_loss=0.05026, over 2656417.80 frames. ], batch size: 72, lr: 4.33e-03, grad_scale: 32.0 2023-04-17 08:21:02,902 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.5564, 2.3987, 2.8683, 3.9750, 3.6083, 4.0557, 3.1254, 2.5473], device='cuda:0'), covar=tensor([0.0258, 0.0952, 0.0855, 0.0070, 0.0236, 0.0061, 0.0675, 0.0871], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0152, 0.0172, 0.0109, 0.0129, 0.0105, 0.0175, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:21:27,856 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.3200, 2.4263, 4.1319, 3.9713, 4.0750, 3.2608, 3.6973, 3.1621], device='cuda:0'), covar=tensor([0.1857, 0.1342, 0.0141, 0.0193, 0.0151, 0.0670, 0.0271, 0.0705], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0187, 0.0135, 0.0139, 0.0139, 0.0180, 0.0151, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-04-17 08:21:43,809 INFO [train.py:893] (0/4) Epoch 30, batch 1650, loss[loss=0.1614, simple_loss=0.2252, pruned_loss=0.04883, over 13462.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2277, pruned_loss=0.04951, over 2657538.59 frames. ], batch size: 79, lr: 4.32e-03, grad_scale: 32.0 2023-04-17 08:21:48,675 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.912e+02 2.404e+02 2.671e+02 3.103e+02 7.046e+02, threshold=5.343e+02, percent-clipped=2.0 2023-04-17 08:21:49,859 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.6219, 2.2971, 2.2792, 2.6833, 2.0630, 2.7431, 2.6643, 2.1611], device='cuda:0'), covar=tensor([0.0091, 0.0235, 0.0187, 0.0166, 0.0256, 0.0138, 0.0172, 0.0233], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0132, 0.0135, 0.0139, 0.0146, 0.0125, 0.0117, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-04-17 08:21:59,460 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79941.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:21:59,496 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79941.0, num_to_drop=1, layers_to_drop={2} 2023-04-17 08:22:30,244 INFO [train.py:893] (0/4) Epoch 30, batch 1700, loss[loss=0.15, simple_loss=0.2172, pruned_loss=0.04146, over 13488.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.228, pruned_loss=0.04926, over 2656779.79 frames. ], batch size: 81, lr: 4.32e-03, grad_scale: 32.0 2023-04-17 08:22:31,391 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.6145, 3.9933, 3.8066, 4.3821, 2.5388, 3.3829, 4.1810, 2.4889], device='cuda:0'), covar=tensor([0.0173, 0.0419, 0.0726, 0.0562, 0.1506, 0.0867, 0.0408, 0.1576], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0195, 0.0213, 0.0260, 0.0190, 0.0207, 0.0185, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:22:43,779 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79989.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 08:22:43,816 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79989.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:22:47,838 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4199, 4.6257, 4.3419, 4.5103, 4.5220, 4.8701, 4.5486, 4.6118], device='cuda:0'), covar=tensor([0.0293, 0.0269, 0.0367, 0.0804, 0.0326, 0.0228, 0.0318, 0.0234], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0169, 0.0194, 0.0274, 0.0193, 0.0208, 0.0188, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 08:22:54,372 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/checkpoint-80000.pt 2023-04-17 08:23:04,103 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0624, 3.0796, 2.7577, 2.0545, 2.1434, 2.6515, 2.7635, 3.2174], device='cuda:0'), covar=tensor([0.1014, 0.0351, 0.0565, 0.1392, 0.0420, 0.0526, 0.0632, 0.0226], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0159, 0.0135, 0.0215, 0.0120, 0.0175, 0.0186, 0.0143], device='cuda:0'), out_proj_covar=tensor([1.3431e-04, 1.1866e-04, 1.0299e-04, 1.5904e-04, 8.5631e-05, 1.3198e-04, 1.3979e-04, 1.0533e-04], device='cuda:0') 2023-04-17 08:23:09,711 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80013.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:23:16,315 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80020.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:23:20,049 INFO [train.py:893] (0/4) Epoch 30, batch 1750, loss[loss=0.1447, simple_loss=0.2106, pruned_loss=0.0394, over 13345.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2268, pruned_loss=0.04874, over 2661592.71 frames. ], batch size: 67, lr: 4.32e-03, grad_scale: 32.0 2023-04-17 08:23:24,292 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.625e+02 2.541e+02 2.943e+02 3.454e+02 5.416e+02, threshold=5.886e+02, percent-clipped=2.0 2023-04-17 08:23:26,158 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80031.0, num_to_drop=1, layers_to_drop={0} 2023-04-17 08:23:32,289 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80037.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:23:39,315 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.58 vs. limit=5.0 2023-04-17 08:23:55,084 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.3462, 3.8003, 3.5579, 4.1403, 2.3907, 3.1325, 3.9212, 2.3178], device='cuda:0'), covar=tensor([0.0182, 0.0432, 0.0791, 0.0541, 0.1633, 0.1013, 0.0478, 0.1755], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0195, 0.0212, 0.0260, 0.0190, 0.0207, 0.0185, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:24:00,563 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80068.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:24:05,336 INFO [train.py:893] (0/4) Epoch 30, batch 1800, loss[loss=0.1864, simple_loss=0.2513, pruned_loss=0.06077, over 13221.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2264, pruned_loss=0.04845, over 2662724.40 frames. ], batch size: 124, lr: 4.32e-03, grad_scale: 32.0 2023-04-17 08:24:10,281 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80079.0, num_to_drop=1, layers_to_drop={1} 2023-04-17 08:24:45,231 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80118.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:24:50,705 INFO [train.py:893] (0/4) Epoch 30, batch 1850, loss[loss=0.1574, simple_loss=0.2147, pruned_loss=0.05004, over 13408.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2262, pruned_loss=0.0486, over 2659969.06 frames. ], batch size: 65, lr: 4.32e-03, grad_scale: 32.0 2023-04-17 08:24:52,418 WARNING [train.py:1054] (0/4) Exclude cut with ID d2dfd1e3b98dcf1e35fa846f29120a93095efe5069904116be6b42831927a53f115bfc3446ded9aa455d24bd9d4fc7289e2c122d3e44590ac9a27e0b7f43c89e-common_voice_en_517302 from training. Duration: 21.72 2023-04-17 08:24:55,472 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.690e+02 2.423e+02 2.778e+02 3.234e+02 9.030e+02, threshold=5.555e+02, percent-clipped=2.0 2023-04-17 08:25:04,709 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([5.7081, 5.2295, 5.0806, 5.1828, 5.1388, 5.0746, 5.6766, 5.2798], device='cuda:0'), covar=tensor([0.0675, 0.1154, 0.2194, 0.2383, 0.0817, 0.1567, 0.0756, 0.0974], device='cuda:0'), in_proj_covar=tensor([0.0362, 0.0407, 0.0501, 0.0502, 0.0323, 0.0378, 0.0469, 0.0383], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:25:36,161 INFO [train.py:893] (0/4) Epoch 30, batch 1900, loss[loss=0.1353, simple_loss=0.2, pruned_loss=0.03527, over 13366.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.226, pruned_loss=0.04852, over 2664492.66 frames. ], batch size: 62, lr: 4.32e-03, grad_scale: 32.0 2023-04-17 08:26:03,337 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.9926, 2.5662, 1.9869, 3.9388, 4.3849, 3.2855, 4.2904, 4.1527], device='cuda:0'), covar=tensor([0.0106, 0.0972, 0.1118, 0.0096, 0.0069, 0.0494, 0.0080, 0.0081], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0095, 0.0103, 0.0089, 0.0076, 0.0086, 0.0063, 0.0086], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 08:26:19,052 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-17 08:26:22,631 INFO [train.py:893] (0/4) Epoch 30, batch 1950, loss[loss=0.1734, simple_loss=0.2313, pruned_loss=0.0578, over 13514.00 frames. ], tot_loss[loss=0.1604, simple_loss=0.2249, pruned_loss=0.04793, over 2665468.80 frames. ], batch size: 85, lr: 4.32e-03, grad_scale: 32.0 2023-04-17 08:26:26,825 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.736e+02 2.552e+02 2.954e+02 3.319e+02 5.622e+02, threshold=5.908e+02, percent-clipped=2.0 2023-04-17 08:26:37,573 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80241.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:27:07,357 INFO [train.py:893] (0/4) Epoch 30, batch 2000, loss[loss=0.1487, simple_loss=0.2171, pruned_loss=0.04008, over 13346.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2265, pruned_loss=0.04845, over 2666247.89 frames. ], batch size: 73, lr: 4.32e-03, grad_scale: 32.0 2023-04-17 08:27:13,775 WARNING [train.py:1054] (0/4) Exclude cut with ID 66b27ad6bd4f9cb9a012d8e5cb8391b2c5494a180128e0910c3fa9b3566bbf033a3ef43178d36693ba7c41a5e1373d8fa4699e5c91228afc6a64c29f3e583b7a-common_voice_en_32639323 from training. Duration: 0.9 2023-04-17 08:27:16,485 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80283.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:27:22,107 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80289.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:27:44,019 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80313.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:27:46,334 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.52 vs. limit=5.0 2023-04-17 08:27:53,418 INFO [train.py:893] (0/4) Epoch 30, batch 2050, loss[loss=0.1589, simple_loss=0.2228, pruned_loss=0.04753, over 13440.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2286, pruned_loss=0.04922, over 2667241.05 frames. ], batch size: 103, lr: 4.31e-03, grad_scale: 32.0 2023-04-17 08:27:58,268 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.934e+02 2.617e+02 2.996e+02 3.704e+02 6.259e+02, threshold=5.993e+02, percent-clipped=2.0 2023-04-17 08:28:12,328 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80344.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:28:27,568 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80361.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:28:39,456 INFO [train.py:893] (0/4) Epoch 30, batch 2100, loss[loss=0.166, simple_loss=0.2367, pruned_loss=0.04766, over 13521.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2281, pruned_loss=0.04904, over 2668185.56 frames. ], batch size: 91, lr: 4.31e-03, grad_scale: 32.0 2023-04-17 08:28:56,856 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80393.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:29:19,770 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80418.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:29:25,184 INFO [train.py:893] (0/4) Epoch 30, batch 2150, loss[loss=0.1697, simple_loss=0.2348, pruned_loss=0.05232, over 13229.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2278, pruned_loss=0.04834, over 2670464.70 frames. ], batch size: 124, lr: 4.31e-03, grad_scale: 32.0 2023-04-17 08:29:29,313 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.734e+02 2.403e+02 2.859e+02 3.405e+02 5.659e+02, threshold=5.718e+02, percent-clipped=0.0 2023-04-17 08:29:52,257 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80454.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:29:59,497 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80463.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:30:02,975 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80466.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:30:10,031 INFO [train.py:893] (0/4) Epoch 30, batch 2200, loss[loss=0.1657, simple_loss=0.235, pruned_loss=0.04813, over 13400.00 frames. ], tot_loss[loss=0.1621, simple_loss=0.2274, pruned_loss=0.04845, over 2667264.41 frames. ], batch size: 109, lr: 4.31e-03, grad_scale: 32.0 2023-04-17 08:30:27,501 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80494.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:30:53,963 INFO [train.py:893] (0/4) Epoch 30, batch 2250, loss[loss=0.1583, simple_loss=0.2229, pruned_loss=0.04683, over 13028.00 frames. ], tot_loss[loss=0.1603, simple_loss=0.2253, pruned_loss=0.04764, over 2661723.85 frames. ], batch size: 142, lr: 4.31e-03, grad_scale: 32.0 2023-04-17 08:30:54,265 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80524.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:30:54,443 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.32 vs. limit=5.0 2023-04-17 08:30:55,182 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.7250, 3.8672, 2.9449, 2.6848, 2.7334, 2.4053, 3.9772, 2.2832], device='cuda:0'), covar=tensor([0.1947, 0.0403, 0.1454, 0.2512, 0.0995, 0.3556, 0.0324, 0.4190], device='cuda:0'), in_proj_covar=tensor([0.0324, 0.0303, 0.0339, 0.0360, 0.0275, 0.0346, 0.0233, 0.0383], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-04-17 08:30:58,858 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.751e+02 2.489e+02 2.862e+02 3.477e+02 6.119e+02, threshold=5.724e+02, percent-clipped=2.0 2023-04-17 08:31:23,265 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80555.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:31:41,412 INFO [train.py:893] (0/4) Epoch 30, batch 2300, loss[loss=0.148, simple_loss=0.218, pruned_loss=0.03903, over 13107.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.2247, pruned_loss=0.0473, over 2662762.30 frames. ], batch size: 142, lr: 4.31e-03, grad_scale: 32.0 2023-04-17 08:32:00,812 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-17 08:32:15,596 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4879, 3.7821, 3.6558, 4.2498, 2.4162, 3.2066, 4.0002, 2.3339], device='cuda:0'), covar=tensor([0.0179, 0.0488, 0.0795, 0.0594, 0.1614, 0.0936, 0.0464, 0.1685], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0196, 0.0213, 0.0262, 0.0191, 0.0207, 0.0186, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:32:25,296 INFO [train.py:893] (0/4) Epoch 30, batch 2350, loss[loss=0.1517, simple_loss=0.221, pruned_loss=0.04115, over 13549.00 frames. ], tot_loss[loss=0.1596, simple_loss=0.2245, pruned_loss=0.04731, over 2660787.44 frames. ], batch size: 87, lr: 4.31e-03, grad_scale: 32.0 2023-04-17 08:32:30,853 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.921e+02 2.384e+02 2.743e+02 3.266e+02 4.484e+02, threshold=5.485e+02, percent-clipped=0.0 2023-04-17 08:32:32,369 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2023-04-17 08:32:40,114 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80639.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:32:48,956 WARNING [train.py:1054] (0/4) Exclude cut with ID 7963691c43c8cc498c58f117527522bf772c76c38530570bc55ef04834f67fb7a9227bd0fa1f13e64e8de1cde6594f3501e172ab86559697c08726cac26f4c6f-common_voice_en_21692090 from training. Duration: 20.16 2023-04-17 08:33:12,432 INFO [train.py:893] (0/4) Epoch 30, batch 2400, loss[loss=0.1508, simple_loss=0.2214, pruned_loss=0.04012, over 13400.00 frames. ], tot_loss[loss=0.1584, simple_loss=0.2234, pruned_loss=0.04676, over 2662842.30 frames. ], batch size: 109, lr: 4.30e-03, grad_scale: 32.0 2023-04-17 08:33:57,816 INFO [train.py:893] (0/4) Epoch 30, batch 2450, loss[loss=0.1648, simple_loss=0.2291, pruned_loss=0.05025, over 13513.00 frames. ], tot_loss[loss=0.1591, simple_loss=0.224, pruned_loss=0.04711, over 2667833.73 frames. ], batch size: 76, lr: 4.30e-03, grad_scale: 32.0 2023-04-17 08:34:01,992 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.276e+02 2.683e+02 3.157e+02 8.436e+02, threshold=5.367e+02, percent-clipped=3.0 2023-04-17 08:34:02,279 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.4260, 4.2583, 4.3790, 2.7939, 4.7643, 4.4693, 4.4265, 4.6955], device='cuda:0'), covar=tensor([0.0246, 0.0135, 0.0145, 0.1025, 0.0128, 0.0265, 0.0149, 0.0101], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0061, 0.0089, 0.0106, 0.0106, 0.0120, 0.0089, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:34:09,722 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80736.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:34:20,286 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80749.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:34:42,748 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80773.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:34:43,342 INFO [train.py:893] (0/4) Epoch 30, batch 2500, loss[loss=0.172, simple_loss=0.2333, pruned_loss=0.05535, over 13458.00 frames. ], tot_loss[loss=0.1592, simple_loss=0.224, pruned_loss=0.04721, over 2664794.97 frames. ], batch size: 103, lr: 4.30e-03, grad_scale: 32.0 2023-04-17 08:35:04,666 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80797.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:35:24,611 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80819.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:35:24,762 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([2.0324, 3.0575, 2.6436, 2.0703, 2.0160, 2.5744, 2.7603, 3.1583], device='cuda:0'), covar=tensor([0.1029, 0.0345, 0.0610, 0.1394, 0.0447, 0.0635, 0.0744, 0.0278], device='cuda:0'), in_proj_covar=tensor([0.0184, 0.0161, 0.0135, 0.0219, 0.0122, 0.0176, 0.0190, 0.0143], device='cuda:0'), out_proj_covar=tensor([1.3560e-04, 1.2018e-04, 1.0386e-04, 1.6183e-04, 8.7158e-05, 1.3310e-04, 1.4222e-04, 1.0541e-04], device='cuda:0') 2023-04-17 08:35:28,607 INFO [train.py:893] (0/4) Epoch 30, batch 2550, loss[loss=0.1656, simple_loss=0.2331, pruned_loss=0.04905, over 13238.00 frames. ], tot_loss[loss=0.1602, simple_loss=0.2248, pruned_loss=0.04778, over 2667830.85 frames. ], batch size: 132, lr: 4.30e-03, grad_scale: 32.0 2023-04-17 08:35:33,326 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.688e+02 2.409e+02 2.835e+02 3.365e+02 6.222e+02, threshold=5.670e+02, percent-clipped=1.0 2023-04-17 08:35:38,504 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([4.2509, 4.4841, 4.2355, 4.2935, 4.3367, 4.6120, 4.4506, 4.4088], device='cuda:0'), covar=tensor([0.0245, 0.0264, 0.0307, 0.0804, 0.0283, 0.0235, 0.0275, 0.0245], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0167, 0.0192, 0.0273, 0.0192, 0.0207, 0.0187, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0003], device='cuda:0') 2023-04-17 08:35:38,591 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80834.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:35:52,115 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80850.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:35:52,415 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-17 08:35:54,362 WARNING [train.py:1054] (0/4) Exclude cut with ID 3c1e147a7e233e57b4a58ba4484e77c5a4ea7f61b0e09b142dd61df911d700cb91f4f6d097065ac73a6cdf3eba8864c9aeea6825a69566bbeb12797de4ca20b9-common_voice_en_505097 from training. Duration: 30.264 2023-04-17 08:36:13,728 INFO [train.py:893] (0/4) Epoch 30, batch 2600, loss[loss=0.174, simple_loss=0.2336, pruned_loss=0.05715, over 13526.00 frames. ], tot_loss[loss=0.1606, simple_loss=0.2249, pruned_loss=0.04815, over 2658820.21 frames. ], batch size: 85, lr: 4.30e-03, grad_scale: 32.0 2023-04-17 08:36:14,104 INFO [zipformer.py:1454] (0/4) attn_weights_entropy = tensor([3.4292, 3.2672, 3.9195, 2.7629, 2.5874, 2.6950, 4.1693, 4.2889], device='cuda:0'), covar=tensor([0.1499, 0.1925, 0.0437, 0.2100, 0.1831, 0.1743, 0.0373, 0.0329], device='cuda:0'), in_proj_covar=tensor([0.0260, 0.0289, 0.0209, 0.0232, 0.0227, 0.0192, 0.0226, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-04-17 08:36:14,336 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.55 vs. limit=5.0 2023-04-17 08:36:55,471 INFO [train.py:893] (0/4) Epoch 30, batch 2650, loss[loss=0.143, simple_loss=0.1954, pruned_loss=0.04528, over 12565.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2262, pruned_loss=0.04936, over 2655325.57 frames. ], batch size: 51, lr: 4.30e-03, grad_scale: 32.0 2023-04-17 08:36:59,340 INFO [optim.py:368] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.862e+02 2.620e+02 3.122e+02 3.929e+02 7.636e+02, threshold=6.244e+02, percent-clipped=3.0 2023-04-17 08:37:06,987 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80939.0, num_to_drop=0, layers_to_drop=set() 2023-04-17 08:37:33,618 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/epoch-30.pt 2023-04-17 08:37:42,764 INFO [train.py:1151] (0/4) Done!