metadata
license: apache-2.0
tags:
- generated_from_trainer
model-index:
name: Waynehills-STT-doogie-server
Waynehills-STT-doogie-server
This model is a fine-tuned version of Doogie/Waynehills-STT-doogie on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 2.8554
- Wer: 0.9881
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: 1
- 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: 20
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 4.6882 | 1.2 | 1000 | 4.8098 | 0.9970 |
| 2.4927 | 2.4 | 2000 | 2.9959 | 1.0416 |
| 1.6066 | 3.59 | 3000 | 2.5789 | 0.9442 |
| 1.2902 | 4.79 | 4000 | 2.6152 | 1.0439 |
| 1.0486 | 5.99 | 5000 | 2.5047 | 1.0861 |
| 0.8648 | 7.19 | 6000 | 2.4007 | 0.9347 |
| 0.7659 | 8.38 | 7000 | 2.5087 | 0.9602 |
| 0.6357 | 9.58 | 8000 | 2.5803 | 1.0018 |
| 0.5684 | 10.78 | 9000 | 2.6190 | 0.9656 |
| 0.5647 | 11.98 | 10000 | 2.6129 | 0.9436 |
| 0.5048 | 13.17 | 11000 | 2.7174 | 0.9816 |
| 0.4405 | 14.37 | 12000 | 2.7107 | 0.9519 |
| 0.4273 | 15.57 | 13000 | 2.7672 | 0.9673 |
| 0.4049 | 16.77 | 14000 | 2.8277 | 0.9869 |
| 0.3731 | 17.96 | 15000 | 2.8238 | 0.9786 |
| 0.3761 | 19.16 | 16000 | 2.8554 | 0.9881 |
Framework versions
- Transformers 4.12.5
- Pytorch 1.10.0+cu113
- Datasets 1.16.1
- Tokenizers 0.10.3