metadata
library_name: transformers
license: apache-2.0
base_model: openai/whisper-medium
tags:
- generated_from_trainer
datasets:
- swagen
metrics:
- wer
model-index:
- name: whisper-medium-swagen-combined-20hrs-model
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: swagen
type: swagen
metrics:
- name: Wer
type: wer
value: 0.22899756493506493
whisper-medium-swagen-combined-20hrs-model
This model is a fine-tuned version of openai/whisper-medium on the swagen dataset. It achieves the following results on the evaluation set:
- Loss: 0.3817
- Wer: 0.2290
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: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 8
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 30.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 2.7478 | 0.1238 | 200 | 0.7976 | 0.4823 |
| 1.6879 | 0.2477 | 400 | 0.6310 | 0.3895 |
| 1.7394 | 0.3715 | 600 | 0.5531 | 0.3318 |
| 1.4248 | 0.4954 | 800 | 0.5066 | 0.2981 |
| 1.2897 | 0.6192 | 1000 | 0.4650 | 0.2689 |
| 1.357 | 0.7430 | 1200 | 0.4458 | 0.2738 |
| 1.3443 | 0.8669 | 1400 | 0.4393 | 0.2709 |
| 1.2284 | 0.9907 | 1600 | 0.4151 | 0.2745 |
| 0.6748 | 1.1146 | 1800 | 0.4113 | 0.2520 |
| 0.7242 | 1.2384 | 2000 | 0.4233 | 0.2670 |
| 0.6472 | 1.3622 | 2200 | 0.4206 | 0.2592 |
| 0.6917 | 1.4861 | 2400 | 0.3990 | 0.2790 |
| 0.7172 | 1.6099 | 2600 | 0.3972 | 0.3005 |
| 0.6105 | 1.7337 | 2800 | 0.3926 | 0.2313 |
| 0.7442 | 1.8576 | 3000 | 0.3817 | 0.2290 |
| 0.7074 | 1.9814 | 3200 | 0.3849 | 0.2263 |
| 0.2659 | 2.1053 | 3400 | 0.4002 | 0.2486 |
| 0.3132 | 2.2291 | 3600 | 0.3958 | 0.2228 |
Framework versions
- Transformers 4.47.1
- Pytorch 2.5.1+cu124
- Datasets 3.2.0
- Tokenizers 0.21.0