<!-- 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. -->
wav2vec2-base-asc-demo-google-colab
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.9473
- Wer: 0.8943
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: 8
- 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: 100
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 10.9515 | 1.06 | 100 | 4.0400 | 1.0 |
| 3.026 | 2.13 | 200 | 3.7680 | 1.0 |
| 2.9349 | 3.19 | 300 | 3.8252 | 1.0 |
| 2.7955 | 4.26 | 400 | 3.4536 | 1.0 |
| 2.5002 | 5.32 | 500 | 3.1430 | 1.0 |
| 1.8767 | 6.38 | 600 | 2.5167 | 1.0 |
| 1.179 | 7.45 | 700 | 1.8392 | 0.9879 |
| 0.7846 | 8.51 | 800 | 1.3472 | 0.9293 |
| 0.4949 | 9.57 | 900 | 1.4321 | 0.94 |
| 0.3398 | 10.64 | 1000 | 1.2133 | 0.9214 |
| 0.3259 | 11.7 | 1100 | 1.6726 | 0.9207 |
| 0.2736 | 12.77 | 1200 | 1.6282 | 0.9114 |
| 0.224 | 13.83 | 1300 | 1.6113 | 0.9 |
| 0.1962 | 14.89 | 1400 | 1.6028 | 0.8986 |
| 0.1734 | 15.96 | 1500 | 1.7467 | 0.8979 |
| 0.1761 | 17.02 | 1600 | 1.8196 | 0.8993 |
| 0.1574 | 18.09 | 1700 | 1.8836 | 0.895 |
| 0.1397 | 19.15 | 1800 | 1.7974 | 0.8921 |
| 0.145 | 20.21 | 1900 | 1.7876 | 0.8971 |
| 0.1311 | 21.28 | 2000 | 1.8534 | 0.9029 |
| 0.1329 | 22.34 | 2100 | 1.8521 | 0.8914 |
| 0.1141 | 23.4 | 2200 | 1.8889 | 0.8914 |
| 0.1307 | 24.47 | 2300 | 1.9190 | 0.8957 |
| 0.1094 | 25.53 | 2400 | 1.9745 | 0.8943 |
| 0.1133 | 26.6 | 2500 | 1.9405 | 0.8936 |
| 0.0967 | 27.66 | 2600 | 1.9461 | 0.8914 |
| 0.1088 | 28.72 | 2700 | 1.9458 | 0.8943 |
| 0.1066 | 29.79 | 2800 | 1.9473 | 0.8943 |
Framework versions
- Transformers 4.24.0
- Pytorch 1.13.0+cu116
- Datasets 2.8.0
- Tokenizers 0.13.2
