whisper-small-mn-2

This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.7259
  • Wer: 40.8783
  • Cer: 13.9617

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: 32
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • training_steps: 15000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer Cer
0.0839 4.26 1000 0.4647 45.7286 16.0020
0.0093 8.51 2000 0.5434 43.9753 15.2446
0.0044 12.77 3000 0.6009 43.6257 15.1717
0.0029 17.02 4000 0.6166 43.0031 14.7578
0.002 21.28 5000 0.6390 42.6098 14.7286
0.001 25.53 6000 0.6558 41.7468 14.3516
0.0021 29.79 7000 0.6714 42.3039 14.4589
0.0003 34.04 8000 0.6791 41.0586 13.9506
0.0001 38.3 9000 0.6949 41.3808 14.1670
0.0013 42.55 10000 0.6875 41.4682 14.2983
0.0001 46.81 11000 0.6937 40.9165 13.9549
0.0001 51.06 12000 0.7092 40.9275 13.9549
0.0 55.32 13000 0.7190 40.9657 13.9703
0.0 59.57 14000 0.7259 40.8783 13.9617
0.0 63.83 15000 0.7292 40.8838 13.9274

Framework versions

  • Transformers 4.26.0.dev0
  • Pytorch 1.13.0+cu117
  • Datasets 2.7.1.dev0
  • Tokenizers 0.13.2
Downloads last month
17
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Datasets used to train bayartsogt/whisper-small-mn-2

Evaluation results