metadata
library_name: transformers
license: apache-2.0
base_model: openai/whisper-medium
tags:
- generated_from_trainer
datasets:
- bemgen
metrics:
- wer
model-index:
- name: whisper-medium-bemgen-male-model
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: bemgen
type: bemgen
metrics:
- name: Wer
type: wer
value: 0.4208404074702886
whisper-medium-bemgen-male-model
This model is a fine-tuned version of openai/whisper-medium on the bemgen dataset. It achieves the following results on the evaluation set:
- Loss: 0.5523
- Wer: 0.4208
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: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Use OptimizerNames.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
- training_steps: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.935 | 0.3960 | 200 | 0.9819 | 0.7042 |
| 1.5573 | 0.7921 | 400 | 0.7304 | 0.5454 |
| 1.0381 | 1.1881 | 600 | 0.6502 | 0.5091 |
| 0.9454 | 1.5842 | 800 | 0.5922 | 0.4584 |
| 0.8737 | 1.9802 | 1000 | 0.5523 | 0.4208 |
| 0.4803 | 2.3762 | 1200 | 0.5768 | 0.4037 |
| 0.4081 | 2.7723 | 1400 | 0.5654 | 0.4026 |
| 0.1932 | 3.1683 | 1600 | 0.5846 | 0.3852 |
Framework versions
- Transformers 4.47.1
- Pytorch 2.5.1+cu124
- Datasets 3.2.0
- Tokenizers 0.21.0