whisper-10hrs-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: 0.3603
• Wer: 16.9658

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: 26
• num_epochs: 3.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.976	0.2955	26	0.9290	35.2904
0.4909	0.5909	52	0.4665	20.5925
0.3898	0.8864	78	0.4072	19.2035
0.3478	1.1818	104	0.3891	18.1669
0.3719	1.4773	130	0.3781	17.7173
0.3604	1.7727	156	0.3689	17.5328
0.3652	2.0682	182	0.3648	16.9054
0.3593	2.3636	208	0.3633	16.8517
0.3107	2.6591	234	0.3611	16.8383
0.3073	2.9545	260	0.3603	16.9658

Framework versions

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