kevintran1/test-1

Whisper Large v3

This model is a fine-tuned version of openai/whisper-large-v3 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0346

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: 0.001
• train_batch_size: 8
• 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: linear
• lr_scheduler_warmup_steps: 50
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.0923	0.2151	20	0.0633
0.045	0.4301	40	0.0610
0.0664	0.6452	60	0.0720
0.0622	0.8602	80	0.0552
0.0589	1.0753	100	0.0781
0.0382	1.2903	120	0.0480
0.03	1.5054	140	0.0445
0.022	1.7204	160	0.0404
0.0201	1.9355	180	0.0387
0.0131	2.1505	200	0.0412
0.0111	2.3656	220	0.0366
0.0111	2.5806	240	0.0357
0.0142	2.7957	260	0.0346

Framework versions

• PEFT 0.15.1
• Transformers 4.51.1
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1