whisper-20hrs-class-seed42

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: 0.3261
• Wer: 15.0502

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
0.5879	0.2994	53	0.5379	25.9838
0.4106	0.5989	106	0.3838	18.0193
0.3296	0.8983	159	0.3624	17.4556
0.3091	1.1977	212	0.3492	16.2814
0.321	1.4972	265	0.3424	16.5062
0.3195	1.7966	318	0.3339	15.3689
0.2962	2.0960	371	0.3314	15.3991
0.3024	2.3955	424	0.3305	15.2347
0.2938	2.6949	477	0.3277	14.8858
0.2834	2.9944	530	0.3261	15.0502

Framework versions

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