whisper-medium-bemgen-baseline-model

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

• Loss: 0.5039
• Wer: 0.4070

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
3.7384	0.3960	200	0.9033	0.6956
3.0032	0.7921	400	0.6777	0.5601
1.9783	1.1881	600	0.5935	0.4839
1.7382	1.5842	800	0.5463	0.4397
1.7106	1.9802	1000	0.5039	0.4070
0.8331	2.3762	1200	0.5411	0.4137
0.8928	2.7723	1400	0.5324	0.4012
0.3736	3.1683	1600	0.5273	0.3802

Framework versions

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