CoolFace
Modelpublic

lorenzoncina/whisper-medium-ru

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face
7likes28downloads
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 Medium Russian

This model is a fine-tuned version of openai/whisper-medium on the mozilla-foundation/commonvoice11_0 ru dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.2253
  • —Wer: 7.5624

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: 16
  • —evalbatchsize: 16
  • —seed: 42
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 500
  • —training_steps: 10000
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWer
0.15780.110000.16628.8290
0.0451.0820000.17488.9148
0.01762.0630000.18898.7848
0.01043.0440000.19228.4354
0.00514.0250000.20348.1865
0.00474.1260000.20128.0455
0.00185.170000.21177.6237
0.00046.0880000.21777.6078
0.00037.0690000.22447.6262
0.00028.04100000.22537.5624

Framework versions

  • —Transformers 4.28.0.dev0
  • —Pytorch 2.0.0+cu117
  • —Datasets 2.11.1.dev0
  • —Tokenizers 0.13.2