whisper-medium-bigcgen-baseline-model

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

• Loss: 0.6934
• Wer: 0.5197

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
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.3168	0.6102	200	0.9403	0.6505
2.7179	1.2227	400	0.7717	0.5571
2.4784	1.8330	600	0.7120	0.5119
1.6451	2.4455	800	0.6934	0.5197
0.9938	3.0580	1000	0.7158	0.4881
0.9703	3.6682	1200	0.7220	0.5018
0.5808	4.2807	1400	0.7521	0.4930

Framework versions

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