metadata
language:
- en
license: apache-2.0
base_model: openai/whisper-base.en
tags:
- generated_from_trainer
datasets:
- dailytalk
metrics:
- wer
model-index:
- name: whisper-base.en-mmb
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: dailytalk
type: dailytalk
args: 'config: en, split: test'
metrics:
- name: Wer
type: wer
value: 10.43014661335272
whisper-base.en-mmb
This model is a fine-tuned version of openai/whisper-base.en on the dailytalk dataset. It achieves the following results on the evaluation set:
- Loss: 0.1725
- Wer: 10.4301
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: 8
- 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 |
|---|---|---|---|---|
| 0.1555 | 0.93 | 1000 | 0.1675 | 10.6676 |
| 0.1267 | 1.87 | 2000 | 0.1613 | 10.3356 |
| 0.0876 | 2.8 | 3000 | 0.1662 | 10.4398 |
| 0.0544 | 3.74 | 4000 | 0.1725 | 10.4301 |
Framework versions
- Transformers 4.37.2
- Pytorch 2.2.1+cu121
- Datasets 2.20.0
- Tokenizers 0.15.2