whisper-medium-nyagen-combined-62

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: 0.4085
• Wer: 0.2932

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
• num_epochs: 30.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4785	0.4941	200	0.6877	0.5468
0.3598	0.9883	400	0.5085	0.4158
0.1951	1.4818	600	0.4591	0.3303
0.1769	1.9759	800	0.4085	0.2932
0.0995	2.4694	1000	0.4247	0.2799
0.0893	2.9636	1200	0.4095	0.3000
0.0401	3.4571	1400	0.4242	0.2517

Framework versions

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