sqrk's picture
Model save
37ad178 verified
|
raw
history blame
2.65 kB
metadata
license: apache-2.0
base_model: openai/whisper-large-v3
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: Sep26-Mixat-whisper-lg-3-transcript
    results: []

Sep26-Mixat-whisper-lg-3-transcript

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

  • Loss: 0.7722
  • Wer: 40.9156

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: 8
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 100
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.7871 0.4292 100 0.4161 34.9906
0.4974 0.8584 200 0.4319 33.8051
0.3796 1.2876 300 0.4348 39.1582
0.3711 1.7167 400 0.4504 38.4566
0.3527 2.1459 500 0.4676 41.7834
0.2699 2.5751 600 0.4600 37.4643
0.275 3.0043 700 0.4449 37.8945
0.1566 3.4335 800 0.5121 40.7568
0.1658 3.8627 900 0.5067 40.8252
0.1125 4.2918 1000 0.5469 41.1772
0.0913 4.7210 1100 0.5818 40.3803
0.0806 5.1502 1200 0.6051 41.2041
0.0519 5.5794 1300 0.5997 40.6908
0.0545 6.0086 1400 0.6158 41.0574
0.0323 6.4378 1500 0.6482 40.7617
0.034 6.8670 1600 0.6761 39.0140
0.0259 7.2961 1700 0.7324 42.2796
0.0249 7.7253 1800 0.7128 41.8616
0.021 8.1545 1900 0.7722 40.9156

Framework versions

  • Transformers 4.43.4
  • Pytorch 2.4.1
  • Datasets 3.0.0
  • Tokenizers 0.19.1