CoolFace
Modelpublic

alikanakar/whisper-synthesized-turkish-4-hour-hlr

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face
0likes5downloads
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-synthesized-turkish-4-hour-hlr

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: 0.4388
  • —Wer: 15.4240

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: 16
  • —evalbatchsize: 8
  • —seed: 42
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 500
  • —training_steps: 2000
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWer
0.73431.041000.258016.7757
0.15682.082000.271415.3129
0.0913.123000.309916.2573
0.08614.174000.394622.9910
0.09675.215000.488423.2132
0.07756.256000.426319.9852
0.06927.297000.442820.0099
0.0528.338000.440723.6761
0.04589.389000.476019.7445
0.032610.4210000.484718.6520
0.028111.4611000.493620.2074
0.022112.512000.465519.3495
0.012313.5413000.465717.5781
0.010514.5814000.449316.2264
0.004215.6215000.439615.5660
0.002916.6716000.441215.7882
0.001117.7117000.440015.8190
0.000518.7518000.440015.4672
0.000319.7919000.438915.4117
0.000220.8320000.438815.4240

Framework versions

  • —Transformers 4.28.0
  • —Pytorch 2.0.0+cu118
  • —Datasets 2.11.0
  • —Tokenizers 0.13.3