erisadhami/bitgenius-large-whsiper-v3-sq

bitgenius-large-commonvoice-v1

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

• Loss: 0.3804
• Wer: 15.0543

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: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 3500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2294	2.9762	500	0.2576	18.0158
0.0471	5.9524	1000	0.2719	16.9628
0.0068	8.9286	1500	0.3211	16.3705
0.0023	11.9048	2000	0.3455	15.3340
0.0006	14.8810	2500	0.3618	15.0049
0.0003	17.8571	3000	0.3760	14.9720
0.0003	20.8333	3500	0.3804	15.0543

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.4.1+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1