Hugging Face's logo

fsicoli
/
whisper-medium-pt-1000h

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
whisper
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
whisper-medium-pt-1000h / README.md
fsicoli's picture
fsicoli
End of training
1cb5bd4 verified
preview code
|
raw
history blame contribute delete
2.98 kB
metadata
license: apache-2.0
base_model: openai/whisper-medium
tags:
  - generated_from_trainer
datasets:
  - >-
    fsicoli/cv17-fleurs-coraa-mls-ted-alcaim-cf-cdc-lapsbm-lapsmail-sydney-lingualibre-voxforge-tatoeba
metrics:
  - wer
model-index:
  - name: whisper-medium-pt-1000h
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: >-
            fsicoli/cv17-fleurs-coraa-mls-ted-alcaim-cf-cdc-lapsbm-lapsmail-sydney-lingualibre-voxforge-tatoeba
            default
          type: >-
            fsicoli/cv17-fleurs-coraa-mls-ted-alcaim-cf-cdc-lapsbm-lapsmail-sydney-lingualibre-voxforge-tatoeba
          args: default
        metrics:
          - name: Wer
            type: wer
            value: 0.11473958668640959

whisper-medium-pt-1000h

This model is a fine-tuned version of openai/whisper-medium on the fsicoli/cv17-fleurs-coraa-mls-ted-alcaim-cf-cdc-lapsbm-lapsmail-sydney-lingualibre-voxforge-tatoeba default dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6491
  • Wer: 0.1147

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: 5e-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: 10000
  • training_steps: 300000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.4574 0.2 20000 0.5339 0.1631
0.4124 0.39 40000 0.4512 0.1517
0.481 0.59 60000 0.4628 0.1466
0.3452 0.79 80000 0.4677 0.1392
0.4086 0.98 100000 0.4551 0.1364
0.1565 1.18 120000 0.5060 0.1316
0.1513 1.38 140000 0.5330 0.1286
0.1496 1.57 160000 0.5519 0.1263
0.1533 1.77 180000 0.5528 0.1234
0.1525 1.97 200000 0.4857 0.1194
0.1918 2.16 220000 0.5915 0.1189
0.1175 2.36 240000 0.6099 0.1174
0.0959 2.56 260000 0.6183 0.1157
0.0988 2.75 280000 0.6423 0.1152
0.0913 2.95 300000 0.6491 0.1147

Framework versions

  • Transformers 4.39.0.dev0
  • Pytorch 2.2.1+cu121
  • Datasets 2.18.1.dev0
  • Tokenizers 0.15.0