CoolFace
Modelpublic

lmejias/whisper-small-atc

sourceHugging Faceapache-2.0updated 10mo agoView on Hugging Face
0likes16downloads
Model Card

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

WhisperATCSmall

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

  • —Loss: 8.2932
  • —Wer Ortho: 76.8786
  • —Wer: 77.1429

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: 0.0001
  • —trainbatchsize: 32
  • —evalbatchsize: 32
  • —seed: 42
  • —optimizer: Use OptimizerNames.ADAMWTORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
  • —lrschedulertype: constantwithwarmup
  • —lrschedulerwarmup_steps: 50
  • —training_steps: 500
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWer OrthoWer
No log10.0102.4187111.5607105.7143
No log20.0202.442771.098371.4286
1.324630.0302.933382.659082.8571
1.324640.0402.341379.768879.4286
0.053450.0502.7293194.7977193.1429
0.053460.0603.901989.017389.1429
0.053470.0705.518379.768880.0
0.011980.0806.4220323.6994321.7143
0.011990.0905.2189654.9133646.8571
0.0614100.01007.4941435.2601430.2857
0.0614110.01107.098082.080982.2857
0.0614120.01206.3404204.0462202.8571
0.0130.01307.307798.843998.8571
0.0140.01407.185983.237083.4286
0.0150.01507.150885.549185.7143
0.0160.01606.952679.190879.4286
0.0170.01706.895684.971185.1429
0.0180.01806.871579.768880.0
0.0190.01907.171183.237083.4286
0.0200.02007.213483.237083.4286
0.0210.02107.245084.971185.1429
0.0220.02207.222883.237083.4286
0.0230.02307.216980.346880.5714
0.0240.02407.304378.034778.2857
0.0250.02507.449780.924981.1429
0.0260.02607.627579.768880.0
0.0270.02707.459080.346880.5714
0.0280.02807.376580.346880.5714
0.0290.02907.167576.878677.1429
0.0300.03007.170878.612778.8571
0.0310.03107.207879.768880.0
0.0320.03207.178178.612778.8571
0.0330.03307.372982.659082.8571
0.0340.03407.944383.815084.0
0.0350.03507.730583.815084.0
0.0360.03607.746084.393184.5714
0.0370.03707.439286.127286.2857
0.0380.03807.440881.502981.7143
0.0390.03907.276584.393184.5714
0.0400.04007.756182.080982.2857
0.0410.04107.196882.080982.2857
0.0420.04207.147380.924981.1429
0.0430.04307.219981.502981.7143
0.0440.04407.303979.190879.4286
0.0450.04507.370875.144575.4286
0.0460.04608.078579.190879.4286
0.0470.04708.361274.566574.8571
0.0480.04808.410277.456677.7143
0.0490.04908.302674.566574.8571
0.0500.05008.293276.878677.1429

Framework versions

  • —Transformers 4.51.3
  • —Pytorch 2.9.1+cu128
  • —Datasets 4.4.1
  • —Tokenizers 0.21.0