whisper-small-sw-ndizi-248
This model is a fine-tuned version of smutuvi/whisper-small-sw-common-voice on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.4545
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: 8
- seed: 42
- gradient_accumulation_steps: 22
- total_train_batch_size: 352
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- num_epochs: 100
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| No log | 1.0 | 1 | 1.5588 |
| No log | 2.0 | 3 | 1.5582 |
| No log | 3.0 | 4 | 1.5604 |
| No log | 4.0 | 6 | 1.5600 |
| No log | 5.0 | 8 | 1.5584 |
| No log | 6.0 | 9 | 1.5564 |
| No log | 7.0 | 11 | 1.5573 |
| No log | 8.0 | 13 | 1.5578 |
| No log | 9.0 | 14 | 1.5555 |
| No log | 10.0 | 16 | 1.5526 |
| No log | 11.0 | 17 | 1.5512 |
| No log | 12.0 | 18 | 1.5543 |
| No log | 13.0 | 20 | 1.5531 |
| No log | 14.0 | 21 | 1.5537 |
| No log | 15.0 | 23 | 1.5485 |
| 0.7121 | 16.0 | 25 | 1.5470 |
| 0.7121 | 17.0 | 26 | 1.5472 |
| 0.7121 | 18.0 | 28 | 1.5438 |
| 0.7121 | 19.0 | 30 | 1.5405 |
| 0.7121 | 20.0 | 31 | 1.5388 |
| 0.7121 | 21.0 | 33 | 1.5370 |
| 0.7121 | 22.0 | 34 | 1.5358 |
| 0.7121 | 23.0 | 35 | 1.5341 |
| 0.7121 | 24.0 | 37 | 1.5292 |
| 0.7121 | 25.0 | 38 | 1.5314 |
| 0.7121 | 26.0 | 40 | 1.5298 |
| 0.7121 | 27.0 | 42 | 1.5236 |
| 0.7121 | 28.0 | 43 | 1.5198 |
| 0.7121 | 29.0 | 45 | 1.5203 |
| 0.7121 | 30.0 | 47 | 1.5123 |
| 0.7121 | 31.0 | 48 | 1.5138 |
| 0.6959 | 32.0 | 50 | 1.5080 |
| 0.6959 | 33.0 | 51 | 1.5100 |
| 0.6959 | 34.0 | 52 | 1.5065 |
| 0.6959 | 35.0 | 54 | 1.5011 |
| 0.6959 | 36.0 | 55 | 1.4980 |
| 0.6959 | 37.0 | 57 | 1.4947 |
| 0.6959 | 38.0 | 59 | 1.4924 |
| 0.6959 | 39.0 | 60 | 1.4910 |
| 0.6959 | 40.0 | 62 | 1.4885 |
| 0.6959 | 41.0 | 64 | 1.4863 |
| 0.6959 | 42.0 | 65 | 1.4854 |
| 0.6959 | 43.0 | 67 | 1.4784 |
| 0.6959 | 44.0 | 68 | 1.4784 |
| 0.6959 | 45.0 | 69 | 1.4794 |
| 0.6959 | 46.0 | 71 | 1.4727 |
| 0.6959 | 47.0 | 72 | 1.4713 |
| 0.6959 | 48.0 | 74 | 1.4678 |
| 0.7021 | 49.0 | 76 | 1.4735 |
| 0.7021 | 50.0 | 77 | 1.4666 |
| 0.7021 | 51.0 | 79 | 1.4670 |
| 0.7021 | 52.0 | 81 | 1.4634 |
| 0.7021 | 53.0 | 82 | 1.4646 |
| 0.7021 | 54.0 | 84 | 1.4603 |
| 0.7021 | 55.0 | 85 | 1.4617 |
| 0.7021 | 56.0 | 86 | 1.4626 |
| 0.7021 | 57.0 | 88 | 1.4637 |
| 0.7021 | 58.0 | 89 | 1.4614 |
| 0.7021 | 59.0 | 91 | 1.4589 |
| 0.7021 | 60.0 | 93 | 1.4584 |
| 0.7021 | 61.0 | 94 | 1.4623 |
| 0.7021 | 62.0 | 96 | 1.4581 |
| 0.7021 | 63.0 | 98 | 1.4566 |
| 0.7021 | 64.0 | 99 | 1.4574 |
| 0.6465 | 64.43 | 100 | 1.4545 |
Framework versions
- PEFT 0.7.2.dev0
- Transformers 4.38.0.dev0
- Pytorch 2.0.0
- Datasets 2.16.1
- Tokenizers 0.15.0