README.md

metadata

license: apache-2.0
base_model: openai/whisper-large-v3
tags:
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_16_0
metrics:
  - wer
model-index:
  - name: whisper-large-v3-pt-cv16-cuda
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: mozilla-foundation/common_voice_16_0 pt
          type: mozilla-foundation/common_voice_16_0
          split: None
          args: pt
        metrics:
          - name: Wer
            type: wer
            value: 0.9998545572074984

whisper-large-v3-pt-cv16-cuda

This model is a fine-tuned version of openai/whisper-large-v3 on the mozilla-foundation/common_voice_16_0 pt dataset. It achieves the following results on the evaluation set:

• Loss: 0.1325
• Wer: 0.9999

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: 8
• 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: 2000
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.199	0.26	1000	0.1563	0.1124
0.1654	0.52	2000	0.1500	0.1052
0.1794	0.77	3000	0.1379	0.0997
0.0821	1.03	4000	0.1321	1.0007
0.1292	1.29	5000	0.1325	0.9999

Framework versions

• Transformers 4.37.0.dev0
• Pytorch 2.2.0.dev20231212
• Datasets 2.15.1.dev0
• Tokenizers 0.15.0