Team Ai
Modelpublic

Sania67/Urdu_Based_ASR_Model_FYP

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face
0likes7downloads
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. -->

UrduBasedASRModelFYP

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: 0.0030
  • —Wer: 93.0849

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: 8
  • —seed: 42
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 100
  • —training_steps: 700
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWer
1.24792.271001.0445103.8776
0.35294.552000.296670.4653
0.1486.823000.100654.3300
0.0479.094000.027452.7574
0.010111.365000.009965.1659
0.004113.646000.003984.2309
0.003315.917000.003093.0849

Framework versions

  • —Transformers 4.30.0.dev0
  • —Pytorch 2.0.1+cu118
  • —Datasets 2.12.0
  • —Tokenizers 0.13.3