whisper-10hrs-meta-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: 0.3612
• Wer: 15.7715

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: 24
• num_epochs: 3.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.8772	0.2963	24	0.9649	34.4248
0.4905	0.5926	48	0.4846	20.3040
0.4376	0.8889	72	0.4124	18.7372
0.3835	1.1852	96	0.3908	18.4520
0.3649	1.4815	120	0.3779	17.2443
0.365	1.7778	144	0.3717	16.9994
0.3455	2.0741	168	0.3660	16.6840
0.3926	2.3704	192	0.3647	16.8585
0.2964	2.6667	216	0.3619	15.8889
0.2999	2.9630	240	0.3612	15.7715

Framework versions

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