whisper-small-v4

This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.3193
• Wer: 100

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: 5e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• 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: 5
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
6.0361	0.3125	5	5.3765	100
4.4117	0.625	10	3.4696	100
3.0052	0.9375	15	2.8653	100
2.4717	1.25	20	2.5730	100
2.2188	1.5625	25	2.3918	100
2.085	1.875	30	2.3193	100

Framework versions

• Transformers 4.47.0.dev0
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1