whisper-medium-swagen-baseline-62

This model is a fine-tuned version of openai/whisper-medium on the swagen dataset. It achieves the following results on the evaluation set:

• Loss: 0.5485
• Wer: 0.3236

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-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 62
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.7072	0.4759	200	0.8145	0.4739
0.5449	0.9518	400	0.6483	0.4019
0.3228	1.4259	600	0.5974	0.3901
0.2807	1.9018	800	0.5491	0.3579
0.1053	2.3760	1000	0.5588	0.4012
0.1212	2.8519	1200	0.5485	0.3236
0.044	3.3260	1400	0.5706	0.3248
0.0578	3.8019	1600	0.5690	0.3924
0.0227	4.2760	1800	0.5783	0.3152
0.0201	4.7519	2000	0.5821	0.2834

Framework versions

• Transformers 4.53.0.dev0
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.0