dhaval0108/whisper-large-ce-only

ap-GFar2A5U2x3BasPO59aj8H

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

• Loss: 0.5782
• Model Preparation Time: 0.0113
• Wer: 0.1452

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 64
• 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: 400
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Wer
1.034	1.0	59	0.4106	0.0113	0.1457
0.3451	2.0	118	0.3813	0.0113	0.1515
0.2836	3.0	177	0.4086	0.0113	0.1441
0.2443	4.0	236	0.4156	0.0113	0.1723
0.204	5.0	295	0.3826	0.0113	0.1654
0.1922	6.0	354	0.4312	0.0113	0.1779
0.186	7.0	413	0.4824	0.0113	0.2273
0.1789	8.0	472	0.4474	0.0113	0.1921
0.143	9.0	531	0.4736	0.0113	0.1903
0.1246	10.0	590	0.4682	0.0113	0.1457
0.1015	11.0	649	0.5285	0.0113	0.1854
0.1041	12.0	708	0.4769	0.0113	0.1795
0.0844	13.0	767	0.5024	0.0113	0.1567
0.0748	14.0	826	0.5189	0.0113	0.1468
0.0696	15.0	885	0.5037	0.0113	0.1678
0.0508	16.0	944	0.5499	0.0113	0.1402
0.0427	17.0	1003	0.5307	0.0113	0.1538
0.0268	18.0	1062	0.5700	0.0113	0.1430
0.0178	19.0	1121	0.5722	0.0113	0.1515
0.011	19.6638	1160	0.5782	0.0113	0.1452

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.2