shreyasdesaisuperU/whisper-large-attempt1-with-silence-text-eleven-labs

Whisper Large with Silence text Eleven Labs SSD superU

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

• Loss: 0.0061
• Wer: 0.7018

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: 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: 500
• training_steps: 2000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0384	0.4032	100	0.0259	7.0175
0.0199	0.8065	200	0.0056	2.9825
0.0107	1.2097	300	0.0122	1.7544
0.0093	1.6129	400	0.0043	0.7018
0.0094	2.0161	500	0.0077	0.8772
0.0039	2.4194	600	0.0153	1.2281
0.0062	2.8226	700	0.0105	1.0526
0.0054	3.2258	800	0.0142	0.8772
0.0031	3.6290	900	0.0088	0.7018
0.007	4.0323	1000	0.0111	0.8772
0.0023	4.4355	1100	0.0046	1.0526
0.0023	4.8387	1200	0.0090	1.0526
0.0037	5.2419	1300	0.0053	0.7018
0.0034	5.6452	1400	0.0070	0.8772
0.0033	6.0484	1500	0.0073	1.0526
0.0011	6.4516	1600	0.0096	1.0526
0.0029	6.8548	1700	0.0085	0.8772
0.0002	7.2581	1800	0.0083	1.0526
0.0002	7.6613	1900	0.0074	1.0526
0.0004	8.0645	2000	0.0061	0.7018

Framework versions

• Transformers 4.46.2
• Pytorch 2.2.2+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3