<!-- 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. -->
base-on-torgo0003
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: 0.6579
- Wer: 0.7547
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 |
|---|---|---|---|---|
| 28.1611 | 0.46 | 500 | 3.4550 | 1.0163 |
| 3.2238 | 0.92 | 1000 | 2.8781 | 1.0411 |
| 2.8617 | 1.39 | 1500 | 2.9896 | 1.0028 |
| 2.5841 | 1.85 | 2000 | 2.3744 | 1.2896 |
| 2.2029 | 2.31 | 2500 | 1.8598 | 1.2722 |
| 1.9976 | 2.77 | 3000 | 1.6505 | 1.2513 |
| 1.7817 | 3.23 | 3500 | 1.5291 | 1.2294 |
| 1.6484 | 3.69 | 4000 | 1.4635 | 1.2106 |
| 1.56 | 4.16 | 4500 | 1.4251 | 1.1989 |
| 1.417 | 4.62 | 5000 | 1.4040 | 1.1904 |
| 1.2884 | 5.08 | 5500 | 1.2734 | 1.1568 |
| 1.2788 | 5.54 | 6000 | 1.2242 | 1.1384 |
| 1.2159 | 6.0 | 6500 | 1.0561 | 1.1349 |
| 1.1125 | 6.46 | 7000 | 1.1001 | 1.1175 |
| 1.1495 | 6.93 | 7500 | 1.0409 | 1.1112 |
| 1.0222 | 7.39 | 8000 | 1.0525 | 1.0952 |
| 1.0104 | 7.85 | 8500 | 1.0184 | 1.1048 |
| 0.9956 | 8.31 | 9000 | 1.0438 | 1.1196 |
| 0.8747 | 8.77 | 9500 | 1.0736 | 1.1005 |
| 0.8437 | 9.23 | 10000 | 1.0041 | 1.0768 |
| 0.861 | 9.7 | 10500 | 0.9407 | 1.0496 |
| 0.8238 | 10.16 | 11000 | 0.9237 | 1.0697 |
| 0.7806 | 10.62 | 11500 | 0.8706 | 1.0343 |
| 0.7475 | 11.08 | 12000 | 0.9576 | 1.0407 |
| 0.6963 | 11.54 | 12500 | 0.9195 | 1.0159 |
| 0.7624 | 12.0 | 13000 | 0.8102 | 1.0060 |
| 0.6311 | 12.47 | 13500 | 0.8208 | 0.9897 |
| 0.6649 | 12.93 | 14000 | 0.7699 | 0.9968 |
| 0.6025 | 13.39 | 14500 | 0.7834 | 0.9547 |
| 0.5691 | 13.85 | 15000 | 0.7414 | 0.9632 |
| 0.532 | 14.31 | 15500 | 0.7056 | 0.9473 |
| 0.5572 | 14.77 | 16000 | 0.8136 | 0.9929 |
| 0.5455 | 15.24 | 16500 | 0.7355 | 0.9264 |
| 0.5369 | 15.7 | 17000 | 0.7531 | 0.9352 |
| 0.4771 | 16.16 | 17500 | 0.7527 | 0.9228 |
| 0.4778 | 16.62 | 18000 | 0.7312 | 0.9218 |
| 0.4384 | 17.08 | 18500 | 0.6774 | 0.8913 |
| 0.4619 | 17.54 | 19000 | 0.6888 | 0.8896 |
| 0.4341 | 18.01 | 19500 | 0.7068 | 0.9030 |
| 0.4164 | 18.47 | 20000 | 0.6484 | 0.8754 |
| 0.3883 | 18.93 | 20500 | 0.6388 | 0.8676 |
| 0.4135 | 19.39 | 21000 | 0.6732 | 0.8683 |
| 0.4121 | 19.85 | 21500 | 0.6354 | 0.8591 |
| 0.3694 | 20.31 | 22000 | 0.6751 | 0.8581 |
| 0.367 | 20.78 | 22500 | 0.6487 | 0.8411 |
| 0.3798 | 21.24 | 23000 | 0.5955 | 0.8312 |
| 0.3249 | 21.7 | 23500 | 0.6209 | 0.8230 |
| 0.3182 | 22.16 | 24000 | 0.7341 | 0.8212 |
| 0.3196 | 22.62 | 24500 | 0.6533 | 0.8106 |
| 0.297 | 23.08 | 25000 | 0.7163 | 0.8177 |
| 0.3021 | 23.55 | 25500 | 0.7209 | 0.8149 |
| 0.3248 | 24.01 | 26000 | 0.6268 | 0.8018 |
| 0.3013 | 24.47 | 26500 | 0.7014 | 0.7915 |
| 0.2986 | 24.93 | 27000 | 0.7306 | 0.8028 |
| 0.2913 | 25.39 | 27500 | 0.6866 | 0.7912 |
| 0.2706 | 25.85 | 28000 | 0.6860 | 0.7851 |
| 0.2572 | 26.32 | 28500 | 0.6478 | 0.7752 |
| 0.2794 | 26.78 | 29000 | 0.6308 | 0.7703 |
| 0.2796 | 27.24 | 29500 | 0.6302 | 0.7653 |
| 0.2604 | 27.7 | 30000 | 0.6638 | 0.7621 |
| 0.2367 | 28.16 | 30500 | 0.6492 | 0.7593 |
| 0.2383 | 28.62 | 31000 | 0.6560 | 0.7614 |
| 0.2495 | 29.09 | 31500 | 0.6577 | 0.7593 |
| 0.2513 | 29.55 | 32000 | 0.6579 | 0.7547 |
Framework versions
- Transformers 4.23.1
- Pytorch 1.12.1+cu113
- Datasets 1.18.3
- Tokenizers 0.13.2
