arampacha/whisper-large-hy-2

<!-- 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-base-hy

This model is a fine-tuned version of openai/whisper-large-v2 on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• —eval_loss: 0.1806
• —eval_wer: 19.9869
• —eval_runtime: 1358.6954
• —evalsamplesper_second: 0.292
• —evalstepsper_second: 0.074
• —epoch: 13.33
• —step: 3000

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: 4
• —evalbatchsize: 4
• —seed: 42
• —gradientaccumulationsteps: 16
• —totaltrainbatch_size: 64
• —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• —lrschedulertype: linear
• —lrschedulerwarmup_steps: 400
• —training_steps: 4000
• —mixedprecisiontraining: Native AMP

Framework versions

• —Transformers 4.26.0.dev0
• —Pytorch 1.13.1+cu117
• —Datasets 2.7.1.dev0
• —Tokenizers 0.13.2