metadata
language:
- dv
license: apache-2.0
base_model: openai/whisper-large-v3
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: Whisper Large v3 DV - Alakxender
results: []
pipeline_tag: automatic-speech-recognition
Whisper Large v3 DV - Alakxender
This model is a fine-tuned version of openai/whisper-large-v3 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4610
- Wer: 71.0345
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: 36
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.4644 | 0.9772 | 300 | 1.0654 | 203.9655 |
| 0.2384 | 1.9544 | 600 | 0.3342 | 84.8276 |
| 0.1481 | 2.9316 | 900 | 0.2715 | 78.7931 |
| 0.0975 | 2.9772 | 1200 | 0.2635 | 76.0345 |
| 0.0616 | 3.9544 | 1500 | 0.2841 | 73.1034 |
| 0.0399 | 4.9772 | 1800 | 0.3215 | 72.2414 |
| 0.0218 | 5.9772 | 2100 | 0.3881 | 73.7931 |
| 0.046 | 6.9772 | 2400 | 0.2772 | 74.1379 |
| 0.018 | 7.9544 | 2700 | 0.3344 | 71.3793 |
| 0.0067 | 8.9316 | 3000 | 0.3947 | 71.7241 |
| 0.0023 | 9.9088 | 3300 | 0.4246 | 72.5862 |
| 0.0008 | 10.8860 | 3600 | 0.4503 | 71.7241 |
| 0.0003 | 11.8632 | 3900 | 0.4610 | 71.0345 |
Framework versions
- Transformers 4.41.0.dev0
- Pytorch 2.3.0+cu121
- Datasets 2.19.0
- Tokenizers 0.19.1