CoolFace
Modelpublic

bayartsogt/whisper-small-mn-8

sourceHugging Faceapache-2.0updated 4y agoView on Hugging Face
0likes997downloads
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. -->

whisper-small-mn-8

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: 0.2421
  • —Wer: 26.5185
  • —Cer: 9.4681

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
  • —trainbatchsize: 32
  • —evalbatchsize: 32
  • —seed: 42
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 500
  • —training_steps: 15000
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWerCer
0.37170.3510000.400446.957616.9664
0.2860.6920000.312937.393513.5504
0.22871.0430000.276833.193111.7806
0.22571.3940000.259030.724311.0232
0.20291.7350000.242829.200310.4144
0.16912.0860000.240828.435710.0306
0.16262.4370000.236928.058810.0486
0.15882.7780000.232127.23409.6819
0.12713.1290000.234926.84079.5574
0.12633.47100000.235627.16309.6519
0.13143.81110000.234026.55679.4278
0.10624.16120000.239026.63329.5162
0.10814.5130000.239826.58409.5085
0.10334.85140000.240226.70969.4801
0.0975.2150000.242126.51859.4681

Framework versions

  • —Transformers 4.26.0.dev0
  • —Pytorch 1.13.0+cu117
  • —Datasets 2.7.1.dev0
  • —Tokenizers 0.13.2