whisper-20hrs-class-seed333

This model is a fine-tuned version of openai/whisper-large-v2 on the JASMIN-CGN dataset. It achieves the following results on the evaluation set:

• Loss: 0.3263
• Wer: 14.9495

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: 0.0001
• train_batch_size: 48
• eval_batch_size: 32
• 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: 53
• num_epochs: 3.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.5994	0.2978	53	0.5447	27.0272
0.419	0.5955	106	0.3864	18.1736
0.3475	0.8933	159	0.3626	17.2610
0.3098	1.1910	212	0.3501	16.2646
0.331	1.4888	265	0.3400	16.6706
0.3332	1.7865	318	0.3336	15.7379
0.3129	2.0843	371	0.3309	15.1407
0.2813	2.3820	424	0.3291	14.8925
0.2912	2.6798	477	0.3278	15.0770
0.2725	2.9775	530	0.3263	14.9495

Framework versions

• PEFT 0.16.0
• Transformers 4.57.3
• Pytorch 2.7.1+cu126
• Datasets 3.6.0
• Tokenizers 0.22.2