whisper-large-v2-japanese

This model is a fine-tuned version of openai/whisper-large-v2 on the mozilla-foundation/common_voice_16_1 and google/fleurs datasets for Japanese.

Model description

5k steps on 2 datasets and includes ctranslate2 version. Drop into subtitle edit.

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
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Framework versions

• PEFT 0.10.1.dev0
• Transformers 4.37.2
• Pytorch 2.2.1+cu118
• Datasets 2.18.0
• Tokenizers 0.15.2