whisper-5hrs-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: 0.4007
• Wer: 18.9117

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: 13
• num_epochs: 3.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.1281	0.2955	13	1.1115	36.2264
0.8275	0.5909	26	0.7577	36.8504
0.5351	0.8864	39	0.5345	24.3567
0.5227	1.1818	52	0.4585	19.7135
0.3926	1.4773	65	0.4312	20.4650
0.3919	1.7727	78	0.4174	19.3176
0.3878	2.0682	91	0.4096	19.3545
0.3876	2.3636	104	0.4045	18.9653
0.3981	2.6591	117	0.4019	18.9922
0.3535	2.9545	130	0.4007	18.9117

Framework versions

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