metadata
library_name: transformers
license: apache-2.0
base_model: openai/whisper-large-v3
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: whisper-large-v3-multids-v3
results: []
whisper-large-v3-multids-v3
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: 0.0675
- Wer: 1.7195
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-06
- train_batch_size: 32
- eval_batch_size: 16
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- total_train_batch_size: 128
- total_eval_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 250
- training_steps: 2000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.3186 | 3.0215 | 250 | 0.1316 | 3.0916 |
| 0.1075 | 7.0085 | 500 | 0.0966 | 2.3375 |
| 0.0834 | 10.03 | 750 | 0.0832 | 2.0758 |
| 0.0774 | 14.017 | 1000 | 0.0762 | 1.8596 |
| 0.0693 | 18.004 | 1250 | 0.0721 | 1.7943 |
| 0.065 | 21.0255 | 1500 | 0.0696 | 1.7406 |
| 0.0634 | 25.0125 | 1750 | 0.0681 | 1.7324 |
| 0.0612 | 28.034 | 2000 | 0.0675 | 1.7195 |
Framework versions
- Transformers 4.45.0.dev0
- Pytorch 2.4.1+cu121
- Datasets 3.0.1.dev0
- Tokenizers 0.19.1