whisper-medium-toigen-combined-model

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

• Loss: 0.6425
• Wer: 0.4498

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 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
3.9586	0.9467	200	0.8733	0.5994
2.4999	1.8899	400	0.6726	0.4648
1.7047	2.8331	600	0.6523	0.4585
0.9573	3.7763	800	0.6425	0.4498
0.4029	4.7195	1000	0.6657	0.4043
0.2311	5.6627	1200	0.6910	0.4187
0.1545	6.6059	1400	0.7208	0.3864

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0