whisper-large-v3-sandi-train-dev-1-pure-transcript-32

This model is a fine-tuned version of openai/whisper-large-v3 on the ntnu-smil/sandi2025-ds dataset. It achieves the following results on the evaluation set:

• Loss: 0.5260
• Wer: 52.2396
• Cer: 122.2059
• Decode Runtime: 215.9746
• Wer Runtime: 0.1832
• Cer Runtime: 0.3633

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: 7e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.98) and epsilon=1e-06 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• training_steps: 896

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer	Decode Runtime	Wer Runtime	Cer Runtime
0.5059	1.0223	224	0.6325	20.2210	44.1000	190.1468	0.1659	0.3130
0.5576	2.0446	448	0.5739	37.3895	102.2570	210.1091	0.1855	0.3559
0.4728	3.0670	672	0.5400	36.2490	70.8408	199.7892	0.1760	0.3378
0.813	4.0893	896	0.5260	52.2396	122.2059	215.9746	0.1832	0.3633

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.4.1+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1