metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
- generated_from_trainer
model-index:
- name: k2e-20s_asr-scr_w2v2-base_003
results: []
k2e-20s_asr-scr_w2v2-base_003
This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.6527
- Per: 0.1654
- Pcc: 0.5394
- Ctc Loss: 0.5408
- Mse Loss: 1.0678
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: 1e-05
- train_batch_size: 16
- eval_batch_size: 1
- seed: 3333
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 2235
- training_steps: 22350
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Per | Pcc | Ctc Loss | Mse Loss |
|---|---|---|---|---|---|---|---|
| 19.3751 | 3.0 | 2235 | 4.7433 | 0.9890 | 0.5024 | 3.8252 | 0.9851 |
| 4.4284 | 6.01 | 4470 | 4.4301 | 0.9890 | 0.5830 | 3.7430 | 0.8578 |
| 3.9191 | 9.01 | 6705 | 4.2301 | 0.9630 | 0.5507 | 3.4950 | 0.9591 |
| 2.958 | 12.02 | 8940 | 2.8191 | 0.5448 | 0.5582 | 1.7723 | 1.1213 |
| 1.4445 | 15.02 | 11175 | 2.0895 | 0.2288 | 0.5320 | 0.8429 | 1.2066 |
| 0.9277 | 18.02 | 13410 | 1.7961 | 0.1901 | 0.5394 | 0.6581 | 1.0903 |
| 0.7339 | 21.03 | 15645 | 1.8166 | 0.1779 | 0.5342 | 0.5943 | 1.1613 |
| 0.6237 | 24.03 | 17880 | 1.7384 | 0.1702 | 0.5290 | 0.5628 | 1.1210 |
| 0.5593 | 27.04 | 20115 | 1.6879 | 0.1667 | 0.5392 | 0.5459 | 1.0930 |
| 0.5212 | 30.04 | 22350 | 1.6527 | 0.1654 | 0.5394 | 0.5408 | 1.0678 |
Framework versions
- Transformers 4.38.1
- Pytorch 2.0.1
- Datasets 2.18.0
- Tokenizers 0.15.2