whisper-large-v3-DI

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

• Loss: 0.5567
• Wer: 23.4828

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: 2e-05
• train_batch_size: 32
• eval_batch_size: 16
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 150
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.0	51	0.4317	28.0344
No log	2.0	102	0.4368	28.2382
No log	3.0	153	0.4949	30.1857
No log	4.0	204	0.5470	26.0870
No log	5.0	255	0.5556	34.5109
No log	6.0	306	0.5484	27.8306
No log	7.0	357	0.5615	26.2228
No log	8.0	408	0.5464	24.6377
No log	9.0	459	0.5568	23.9130
0.2046	10.0	510	0.5567	23.4828

Framework versions

• Transformers 4.46.2
• Pytorch 2.4.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.0