edmundchan70's picture
Model save
38cc941 verified
|
raw
history blame
2.03 kB
metadata
license: apache-2.0
base_model: edmundchan70/Cantonese_finetune_whisper
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: model_output
    results: []

model_output

This model is a fine-tuned version of edmundchan70/Cantonese_finetune_whisper on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1941
  • Wer: 30.7448

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: 1.25e-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_ratio: 0.1
  • num_epochs: 10
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
No log 1.0 156 0.1469 22.1870
No log 2.0 312 0.1667 27.2583
No log 3.0 468 0.1770 27.7338
0.0778 4.0 624 0.1847 29.7940
0.0778 5.0 780 0.1970 30.4279
0.0778 6.0 936 0.1845 29.7940
0.0083 7.0 1092 0.1922 30.2694
0.0083 8.0 1248 0.1926 29.6355
0.0083 9.0 1404 0.1932 30.1109
0.0009 10.0 1560 0.1941 30.7448

Framework versions

  • Transformers 4.44.0
  • Pytorch 2.3.1+cu121
  • Datasets 2.21.0
  • Tokenizers 0.19.1