hubert-base-ser
This model is a fine-tuned version of facebook/hubert-base-ls960 on the Crema dataset. It achieves the following results on the evaluation set:
- Loss: 1.0105
- Accuracy: 0.6313
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 1.0
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy |
---|---|---|---|---|
1.8106 | 0.01 | 10 | 1.7616 | 0.1974 |
1.7268 | 0.03 | 20 | 1.7187 | 0.2525 |
1.7269 | 0.04 | 30 | 1.6442 | 0.3096 |
1.7086 | 0.05 | 40 | 1.5834 | 0.3338 |
1.6983 | 0.07 | 50 | 1.6195 | 0.3600 |
1.5845 | 0.08 | 60 | 1.5753 | 0.3418 |
1.5744 | 0.09 | 70 | 1.5669 | 0.3707 |
1.5915 | 0.11 | 80 | 1.5412 | 0.3754 |
1.5105 | 0.12 | 90 | 2.0037 | 0.2612 |
1.4689 | 0.13 | 100 | 1.5440 | 0.3627 |
1.527 | 0.15 | 110 | 1.5400 | 0.3862 |
1.6481 | 0.16 | 120 | 1.6678 | 0.3298 |
1.7504 | 0.17 | 130 | 1.6078 | 0.2995 |
1.3748 | 0.19 | 140 | 1.5750 | 0.3251 |
1.6417 | 0.2 | 150 | 1.7034 | 0.2599 |
1.6146 | 0.21 | 160 | 1.6162 | 0.3519 |
1.4896 | 0.23 | 170 | 1.5245 | 0.3741 |
1.4278 | 0.24 | 180 | 1.7537 | 0.2424 |
1.4475 | 0.26 | 190 | 1.4769 | 0.3882 |
1.5416 | 0.27 | 200 | 1.4772 | 0.3949 |
1.5997 | 0.28 | 210 | 1.4428 | 0.4278 |
1.4337 | 0.3 | 220 | 1.4352 | 0.4124 |
1.415 | 0.31 | 230 | 1.4405 | 0.4157 |
1.5196 | 0.32 | 240 | 1.4197 | 0.4043 |
1.3866 | 0.34 | 250 | 1.5241 | 0.3734 |
1.3041 | 0.35 | 260 | 1.5703 | 0.4043 |
1.3618 | 0.36 | 270 | 1.3963 | 0.4285 |
1.3293 | 0.38 | 280 | 1.3478 | 0.4506 |
1.2215 | 0.39 | 290 | 1.5994 | 0.3842 |
1.6618 | 0.4 | 300 | 1.7751 | 0.2277 |
1.5349 | 0.42 | 310 | 1.6091 | 0.4036 |
1.4037 | 0.43 | 320 | 1.4741 | 0.4446 |
1.4844 | 0.44 | 330 | 1.4170 | 0.4399 |
1.2806 | 0.46 | 340 | 1.2887 | 0.5050 |
1.3818 | 0.47 | 350 | 1.2668 | 0.5017 |
1.3491 | 0.48 | 360 | 1.4721 | 0.4594 |
1.2347 | 0.5 | 370 | 1.2188 | 0.5245 |
1.2182 | 0.51 | 380 | 1.3813 | 0.4567 |
1.2513 | 0.52 | 390 | 1.2111 | 0.5205 |
1.2447 | 0.54 | 400 | 1.2231 | 0.5460 |
1.038 | 0.55 | 410 | 1.2563 | 0.5373 |
1.2409 | 0.56 | 420 | 1.3448 | 0.4936 |
1.2279 | 0.58 | 430 | 1.1972 | 0.5487 |
1.3256 | 0.59 | 440 | 1.1706 | 0.5742 |
1.2866 | 0.6 | 450 | 1.3091 | 0.5003 |
1.0574 | 0.62 | 460 | 1.2075 | 0.5500 |
1.2744 | 0.63 | 470 | 1.2831 | 0.5171 |
1.0836 | 0.64 | 480 | 1.1768 | 0.5608 |
1.135 | 0.66 | 490 | 1.1408 | 0.5776 |
1.1303 | 0.67 | 500 | 1.2320 | 0.5541 |
1.2068 | 0.69 | 510 | 1.1379 | 0.5796 |
1.1347 | 0.7 | 520 | 1.1124 | 0.5897 |
1.1846 | 0.71 | 530 | 1.1338 | 0.5803 |
1.2409 | 0.73 | 540 | 1.1259 | 0.5789 |
1.0664 | 0.74 | 550 | 1.0653 | 0.6038 |
1.1637 | 0.75 | 560 | 1.0550 | 0.5977 |
1.0707 | 0.77 | 570 | 1.0996 | 0.5715 |
1.2258 | 0.78 | 580 | 1.0804 | 0.5977 |
0.9256 | 0.79 | 590 | 1.1501 | 0.5809 |
1.1542 | 0.81 | 600 | 1.1089 | 0.5957 |
1.3931 | 0.82 | 610 | 1.1381 | 0.5856 |
1.1117 | 0.83 | 620 | 1.0933 | 0.6031 |
1.1433 | 0.85 | 630 | 1.0175 | 0.6219 |
1.0325 | 0.86 | 640 | 0.9885 | 0.6239 |
1.111 | 0.87 | 650 | 1.0048 | 0.6259 |
0.8125 | 0.89 | 660 | 1.0176 | 0.6165 |
1.0414 | 0.9 | 670 | 1.0290 | 0.6185 |
1.0037 | 0.91 | 680 | 1.0269 | 0.6253 |
0.9406 | 0.93 | 690 | 1.0301 | 0.6273 |
1.0129 | 0.94 | 700 | 1.0238 | 0.6326 |
1.2213 | 0.95 | 710 | 1.0181 | 0.6273 |
1.2519 | 0.97 | 720 | 1.0161 | 0.6266 |
0.9932 | 0.98 | 730 | 1.0112 | 0.6279 |
1.0135 | 0.99 | 740 | 1.0105 | 0.6313 |
Framework versions
- Transformers 4.18.0.dev0
- Pytorch 1.10.0+cu111
- Datasets 1.18.5.dev0
- Tokenizers 0.11.6
- Downloads last month
- 7