add model

config_s2a.json

@@ -0,0 +1,69 @@
+{
+  "saving_path": "/home/ubuntu/experiments/a2s_mls",
+  "resume_checkpoint": null,
+  "vocoder_type": "SPEECHTOKENIZER",
+  "vocoder_config_path": null,
+  "vocoder_ckpt_path": null,
+  "metapath": [
+    "/var/data_mls/train.json"
+  ],
+  "val_metapath": [
+    "/var/data_mls/test.json"
+  ],
+  "pretrained_path": null,
+  "speaker_embedding_dir": null,
+  "sampledir": "/home/ubuntu/experiments/a2s_mls",
+  "lr": 0.0005,
+  "batch_size": 100.0,
+  "train_bucket_size": 8192,
+  "training_step": 800000,
+  "optim_flat_percent": 0.0,
+  "warmup_step": 10000,
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.98,
+  "ffd_size": 1024,
+  "hidden_size": 1024,
+  "enc_nlayers": 8,
+  "dec_nlayers": 6,
+  "nheads": 8,
+  "dropout": 0.1,
+  "depthwise_conv_kernel_size": 5,
+  "aligner_softmax_temp": 1.0,
+  "layer_norm_eps": 1e-05,
+  "use_sem_tokens": true,
+  "use_spkr_emb": false,
+  "use_text_emb": false,
+  "fairseq": false,
+  "only_inference": false,
+  "speaker_embed_dropout": 0.05,
+  "label_smoothing": 0.0,
+  "val_check_interval": 1,
+  "max_dataset_samples": -1,
+  "check_val_every_n_epoch": 1,
+  "precision": "bf16",
+  "nworkers": 12,
+  "distributed": true,
+  "accelerator": "gpu",
+  "version": null,
+  "accumulate_grad_batches": 1,
+  "sagemaker": false,
+  "use_repetition_token": false,
+  "use_repetition_gating": false,
+  "repetition_penalty": 1.0,
+  "sampling_temperature": 1.0,
+  "top_k": -1,
+  "min_top_k": 3,
+  "top_p": 0.8,
+  "sample_num": 4,
+  "length_penalty_max_length": 150,
+  "length_penalty_max_prob": 0.95,
+  "max_input_length": 2048,
+  "max_output_length": 2000,
+  "phone_context_window": 3,
+  "sample_rate": 16000,
+  "n_codes": 1024,
+  "n_cluster_groups": 7,
+  "first_n_lvls": 7,
+  "use_pretrained_ckpt_cfg": false,
+  "n_semantic_codes": 1024
+}

config_t2s.json

@@ -0,0 +1,29 @@
+{
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "classifier_dropout": 0.0,
+  "d_ff": 2048,
+  "d_kv": 64,
+  "d_model": 512,
+  "decoder_start_token_id": 0,
+  "dense_act_fn": "relu",
+  "dropout_rate": 0.1,
+  "eos_token_id": 2,
+  "feed_forward_proj": "relu",
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "is_gated_act": false,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "num_decoder_layers": 14,
+  "num_heads": 8,
+  "num_layers": 14,
+  "pad_token_id": 0,
+  "relative_attention_max_distance": 128,
+  "relative_attention_num_buckets": 32,
+  "torch_dtype": "float32",
+  "transformers_version": "4.34.1",
+  "use_cache": true,
+  "vocab_size": 1119
+}

s2a.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a753f373e9dbe936e719c5357180b825882366710df48c8d6f08ca2165c48624
+size 2609559838

t2s.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23637aca68f6c470f7a0c11162634c1990685d0710edf703cbc4123f22d18cc0
+size 413586297