metadata
library_name: transformers
license: apache-2.0
base_model: openai/whisper-medium
tags:
- generated_from_trainer
datasets:
- generator
metrics:
- wer
model-index:
- name: whisper-medium_ro-80mel
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: generator
type: generator
config: default
split: None
args: default
metrics:
- name: Wer
type: wer
value: 0.6935
whisper-medium_ro-80mel
This model is a fine-tuned version of openai/whisper-medium on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 0.0839
- Wer: 0.6935
- Cer: 1.202
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: 1
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 64
- total_train_batch_size: 64
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 500
- num_epochs: 8
Training results
| Training Loss | Epoch | Step | Cer | Validation Loss | Wer |
|---|---|---|---|---|---|
| 0.1665 | 0.7723 | 2500 | 0.9148 | 0.0839 | 0.7855 |
| 0.108 | 1.5443 | 5000 | 2.035 | 0.0722 | 2.1169 |
| 0.0805 | 2.3163 | 7500 | 2.5247 | 0.0718 | 2.4453 |
| 0.0589 | 3.0883 | 10000 | 2.8635 | 0.0583 | 2.8751 |
| 0.0516 | 3.8606 | 12500 | 3.1281 | 0.0511 | 3.2984 |
| 0.0415 | 4.6326 | 15000 | 3.2897 | 0.0540 | 3.365 |
| 0.0368 | 5.4047 | 17500 | 0.0543 | 3.4251 | 3.298 |
| 0.0315 | 6.1767 | 20000 | 0.0559 | 3.2634 | 3.2452 |
| 0.0316 | 6.9490 | 22500 | 0.0541 | 3.6173 | 3.4674 |
| 0.0267 | 7.7210 | 25000 | 0.0547 | 3.3599 | 3.2115 |
"eval_runtime": 113429.374, "eval_samples": 27174, "eval_samples_per_second": 0.24, "eval_steps_per_second": 0.24, "test_samples": 12987, "train_samples": 207181
Framework versions
- Transformers 4.57.0
- Pytorch 2.9.1+cu128
- Datasets 4.4.1
- Tokenizers 0.22.1