<!-- 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. -->
timit-distil-kl-alpha-0.25-T-1-take-3
This model is a fine-tuned version of on the None dataset. It achieves the following results on the evaluation set:
- Loss: 362.6431
- Wer: 0.8022
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: 56
- eval_batch_size: 56
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 112
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.2
- num_epochs: 50
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
836.3694 | 2.43 | 100 | 652.1216 | 1.0487 |
438.9098 | 4.87 | 200 | 438.7045 | 0.8833 |
320.2348 | 7.31 | 300 | 396.8615 | 0.8490 |
267.4869 | 9.75 | 400 | 381.3956 | 0.8325 |
243.05 | 12.19 | 500 | 374.6377 | 0.8292 |
226.4688 | 14.63 | 600 | 372.4966 | 0.8197 |
220.0781 | 17.07 | 700 | 368.0202 | 0.8213 |
206.6639 | 19.51 | 800 | 366.3605 | 0.8112 |
199.0381 | 21.94 | 900 | 366.9292 | 0.8271 |
198.3046 | 24.39 | 1000 | 365.8394 | 0.8088 |
188.066 | 26.82 | 1100 | 364.1574 | 0.8057 |
188.2653 | 29.27 | 1200 | 364.2211 | 0.8025 |
181.248 | 31.7 | 1300 | 363.9985 | 0.8071 |
182.5918 | 34.14 | 1400 | 363.5379 | 0.8042 |
177.1421 | 36.58 | 1500 | 363.5888 | 0.8032 |
179.904 | 39.02 | 1600 | 362.6931 | 0.8038 |
174.7976 | 41.46 | 1700 | 362.9397 | 0.8053 |
173.5596 | 43.89 | 1800 | 362.9175 | 0.8011 |
176.6717 | 46.34 | 1900 | 363.0562 | 0.8013 |
173.9602 | 48.77 | 2000 | 362.6431 | 0.8022 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.12.1
- Datasets 2.8.0
- Tokenizers 0.13.2