whisper-l3

This model is a fine-tuned version of openai/whisper-large-v3 on the common_voice_11_0 dataset. It achieves the following results on the evaluation set:

• Loss: 1.8569
• Wer Ortho: 66.8317
• Wer: 69.5276

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: 0.0002
• train_batch_size: 4
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 2
• training_steps: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
1.418	0.064	1	2.0816	66.7327	70.0
1.2547	0.128	2	2.0508	66.8317	70.1575
1.3085	0.192	3	1.9875	66.5347	70.0787
1.2471	0.256	4	1.9218	66.7327	70.2362
1.1331	0.32	5	1.8569	66.8317	69.5276

Framework versions

• PEFT 0.11.2.dev0
• Transformers 4.42.0.dev0
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1