whisper-medium-nyagen-baseline-42

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

• Loss: 0.4527
• Wer: 0.2946

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: 42
• 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
0.4878	0.9412	200	0.5977	0.4706
0.2552	1.88	400	0.4580	0.3165
0.1619	2.8188	600	0.4590	0.2765
0.0768	3.7576	800	0.4527	0.2946
0.0366	4.6965	1000	0.4769	0.2515
0.0243	5.6353	1200	0.4699	0.2461
0.0193	6.5741	1400	0.4911	0.2365
0.0076	7.5129	1600	0.4978	0.2792

Framework versions

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