Scrya/whisper-large-v2-cantonese

<!-- 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 Large V2 - Cantonese - Augmented

This model is a fine-tuned version of openai/whisper-large-v2 on the mozilla-foundation/commonvoice11_0 dataset. It achieves the following results on the evaluation set:

• —Loss: 0.1828
• —Cer: 6.2133

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

Training:

• —mozilla-foundation/common_voice_11_0 (train+validation)

Evaluation:

• —mozilla-foundation/common_voice_11_0 (test)

Training procedure

Datasets were augmented on-the-fly using audiomentations via PitchShift and TimeStretch transformations at p=0.3.

Training hyperparameters

The following hyperparameters were used during training:

• —learning_rate: 1e-05
• —trainbatchsize: 8
• —evalbatchsize: 4
• —seed: 42
• —gradientaccumulationsteps: 4
• —totaltrainbatch_size: 32
• —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• —lrschedulertype: linear
• —lrschedulerwarmup_steps: 100
• —training_steps: 1000
• —mixedprecisiontraining: Native AMP

Training results

Framework versions

• —Transformers 4.26.0.dev0
• —Pytorch 1.13.1+cu117
• —Datasets 2.8.1.dev0
• —Tokenizers 0.13.2