whisper-20hrs-random-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: 0.3445
• Wer: 15.8520

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: 44
• num_epochs: 3.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.7408	0.3014	44	0.6527	30.3452
0.4431	0.6027	88	0.4096	20.3476
0.3708	0.9041	132	0.3815	18.3950
0.3926	1.2055	176	0.3669	17.7542
0.371	1.5068	220	0.3594	16.5464
0.3694	1.8082	264	0.3523	16.6538
0.3133	2.1096	308	0.3479	15.8822
0.3322	2.4110	352	0.3469	16.0835
0.3521	2.7123	396	0.3445	15.8520

Framework versions

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