whisper-medium-bsbigcgen-female-model

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

• Loss: 0.6368
• Wer: 0.4772

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: 1e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Use 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: 500
• num_epochs: 30.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.47	0.6163	200	0.9274	0.6675
2.569	1.2311	400	0.7236	0.5511
2.5305	1.8475	600	0.6368	0.4772
1.4563	2.4622	800	0.6518	0.4937
0.6413	3.0770	1000	0.6892	0.4782
0.581	3.6934	1200	0.6935	0.4751

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0