nurzhanit/whisper-omg-2

Whisper Small Hi - Sanchit Gandhi

This model is a fine-tuned version of nurzhanit/whisper-enhanced-ml on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0001
• Wer: 0.0

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: 50
• training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0365	0.8621	50	0.0179	2.0516
0.0151	1.7241	100	0.0088	0.8943
0.0086	2.5862	150	0.0046	0.7365
0.0031	3.4483	200	0.0035	0.7891
0.0017	4.3103	250	0.0022	0.3945
0.0033	5.1724	300	0.0011	0.2104
0.0015	6.0345	350	0.0004	0.2630
0.0003	6.8966	400	0.0005	0.1841
0.0012	7.7586	450	0.0004	0.2104
0.0003	8.6207	500	0.0002	0.0
0.0002	9.4828	550	0.0002	0.0
0.0001	10.3448	600	0.0001	0.0
0.0001	11.2069	650	0.0001	0.0
0.0001	12.0690	700	0.0001	0.0
0.0001	12.9310	750	0.0001	0.0
0.0001	13.7931	800	0.0001	0.0
0.0001	14.6552	850	0.0001	0.0
0.0001	15.5172	900	0.0001	0.0
0.0001	16.3793	950	0.0001	0.0
0.0001	17.2414	1000	0.0001	0.0

Framework versions

• Transformers 4.40.0
• Pytorch 2.5.0+cu124
• Datasets 3.0.2
• Tokenizers 0.19.1