whisper-large-v3-sandi-train-dev-1

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: 1.0265
• Wer: 80.7774
• Cer: 205.4415
• Decode Runtime: 296.9575
• Wer Runtime: 0.2339
• Cer Runtime: 0.5476

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
• gradient_accumulation_steps: 32
• total_train_batch_size: 1024
• optimizer: Use 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: 28

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer	Decode Runtime	Wer Runtime	Cer Runtime
1.9021	1.0357	7	1.3669	70.9647	206.0000	293.2787	0.2383	0.5705
1.248	2.0714	14	1.1785	90.1350	223.9722	301.9501	0.2377	0.5710
1.0696	3.1071	21	1.0601	84.5443	211.8357	295.8525	0.2329	0.5515
1.0339	4.1429	28	1.0265	80.7774	205.4415	296.9575	0.2339	0.5476

Framework versions

• PEFT 0.15.1
• Transformers 4.50.3
• Pytorch 2.1.0+cu118
• Datasets 3.5.0
• Tokenizers 0.21.1