<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
torgo-TestedSpeakerM01-finetuned
This model is a fine-tuned version of yongjian/wav2vec2-large-a on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.3416
- Wer: 0.7714
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: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 30
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 28.3192 | 0.84 | 500 | 1.5650 | 1.1343 |
| 0.4741 | 1.69 | 1000 | 0.3845 | 1.2286 |
| 0.4324 | 2.53 | 1500 | 0.3041 | 1.2849 |
| 0.3128 | 3.37 | 2000 | 0.3093 | 1.3337 |
| 0.2875 | 4.22 | 2500 | 0.2870 | 1.1972 |
| 0.2763 | 5.06 | 3000 | 0.3064 | 1.0553 |
| 0.2719 | 5.9 | 3500 | 0.2424 | 0.9534 |
| 0.2436 | 6.75 | 4000 | 0.2772 | 1.0163 |
| 0.2262 | 7.59 | 4500 | 0.2532 | 0.9989 |
| 0.2248 | 8.43 | 5000 | 0.2495 | 0.9133 |
| 0.2267 | 9.27 | 5500 | 0.2771 | 0.9155 |
| 0.1923 | 10.12 | 6000 | 0.2918 | 0.8862 |
| 0.1905 | 10.96 | 6500 | 0.2306 | 0.8700 |
| 0.184 | 11.8 | 7000 | 0.2470 | 0.8949 |
| 0.1747 | 12.65 | 7500 | 0.2411 | 0.8256 |
| 0.1641 | 13.49 | 8000 | 0.3096 | 0.8949 |
| 0.1783 | 14.33 | 8500 | 0.2523 | 0.8776 |
| 0.1575 | 15.18 | 9000 | 0.2606 | 0.8223 |
| 0.1658 | 16.02 | 9500 | 0.2713 | 0.8364 |
| 0.141 | 16.86 | 10000 | 0.2423 | 0.8061 |
| 0.1302 | 17.71 | 10500 | 0.3030 | 0.7952 |
| 0.1285 | 18.55 | 11000 | 0.2719 | 0.7811 |
| 0.203 | 19.39 | 11500 | 0.3074 | 0.8061 |
| 0.1198 | 20.24 | 12000 | 0.3035 | 0.8017 |
| 0.109 | 21.08 | 12500 | 0.3218 | 0.8082 |
| 0.109 | 21.92 | 13000 | 0.3668 | 0.8202 |
| 0.1114 | 22.77 | 13500 | 0.2767 | 0.8007 |
| 0.0949 | 23.61 | 14000 | 0.2965 | 0.7844 |
| 0.0955 | 24.45 | 14500 | 0.3319 | 0.7692 |
| 0.0913 | 25.3 | 15000 | 0.3034 | 0.7833 |
| 0.0912 | 26.14 | 15500 | 0.2944 | 0.7736 |
| 0.0794 | 26.98 | 16000 | 0.3479 | 0.7736 |
| 0.0818 | 27.82 | 16500 | 0.3524 | 0.7790 |
| 0.0836 | 28.67 | 17000 | 0.3467 | 0.7714 |
| 0.0878 | 29.51 | 17500 | 0.3406 | 0.7746 |
Framework versions
- Transformers 4.23.1
- Pytorch 1.12.1+cu113
- Datasets 1.18.3
- Tokenizers 0.13.2
