Edit model card

whisper-small-ft-common-language-id

This model is a fine-tuned version of openai/whisper-small on the common_language dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6409
  • Accuracy: 0.8860

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: 16
  • eval_batch_size: 16
  • seed: 0
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 32
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 10.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Accuracy
1.1767 1.0 694 1.1063 0.7514
0.582 2.0 1388 0.6595 0.8327
0.3172 3.0 2082 0.5887 0.8529
0.196 4.0 2776 0.5332 0.8701
0.0858 5.0 3470 0.5705 0.8733
0.0477 6.0 4164 0.6311 0.8779
0.0353 7.0 4858 0.6011 0.8825
0.0033 8.0 5552 0.6186 0.8843
0.0071 9.0 6246 0.6409 0.8860
0.0074 10.0 6940 0.6334 0.8860

Framework versions

  • Transformers 4.27.0.dev0
  • Pytorch 1.13.1
  • Datasets 2.9.0
  • Tokenizers 0.13.2
Downloads last month
297
Safetensors
Model size
88.4M params
Tensor type
F32
·

Finetuned from

Dataset used to train sanchit-gandhi/whisper-small-ft-common-language-id