whisper-large-ur1

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3983
• Wer: 19.9234

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: 1e-06
• train_batch_size: 16
• eval_batch_size: 8
• 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: reduce_lr_on_plateau
• lr_scheduler_warmup_steps: 500
• training_steps: 8000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.005	1.0121	1000	0.3213	20.2006
0.0016	2.0243	2000	0.3359	19.9098
0.0006	3.0364	3000	0.3501	19.8986
0.0005	4.0486	4000	0.3612	19.8310
0.0004	5.0607	5000	0.3717	20.3110
0.0002	6.0729	6000	0.3816	20.0609
0.0002	7.0850	7000	0.3916	20.0654
0.0013	8.0972	8000	0.3983	19.9234

Framework versions

• Transformers 4.53.1
• Pytorch 2.5.1+cu124
• Datasets 3.5.1
• Tokenizers 0.21.2