You can deploy this model and start earning money today!

Badri0510/
whisper-tanglish-5e5-8000-experiment

Name: whisper-tanglish-5e5-8000-experiment
Availability: InStock

Views

No views yet

whisper-tanglish-5e5-8000-experiment

This model is a fine-tuned version of Badri0510/whisper-tanglish-v11-final-polish on the None dataset. It achieves the following results on the evaluation set:

More information needed

More information needed

More information needed

The following hyperparameters were used during training:

learning_rate: 5e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 300
training_steps: 8000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
0.1893	0.0625	500	0.0822	0.3074
0.2715	0.125	1000	0.1012	0.3229
0.1450	1.0462	1500	0.1071	0.3382
0.2623	1.1087	2000	0.1088	0.3366
0.0930	2.0307	2500	0.0853	0.2896
0.1142	2.0932	3000	0.0662	0.2751
0.1966	3.0151	3500	0.0744	0.2755
0.1235	3.0776	4000	0.0629	0.2807
0.0643	3.1401	4500	0.0497	0.2513
0.0398	4.0625	5000	0.0557	0.2565
0.0248	4.125	5500	0.0180	0.2312
0.1053	5.0477	6000	0.0217	0.2336
0.0138	5.1102	6500	0.0196	0.2325
0.0852	6.0323	7000	0.0179	0.2299
0.0254	6.0948	7500	0.0129	0.2196
0.0195	7.018	8000	0.0129	0.2190