Whisper Small UZ - Bahriddin Muminov

This model is a fine-tuned version of openai/whisper-small on the Common Voice 16.1 dataset. It achieves the following results on the evaluation set:

Model description

More information needed

More information needed

More information needed

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: 1000
training_steps: 57000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
0.357	0.0352	2000	0.4996	42.0801
0.2917	0.0704	4000	0.4227	36.4010
0.2222	0.1056	6000	0.3806	36.8330
0.2127	0.1408	8000	0.3559	31.9044
0.2131	0.1760	10000	0.3392	32.2440
0.2283	0.2112	12000	0.3387	30.3111
0.2056	0.2464	14000	0.3301	29.3033
0.1956	0.2816	16000	0.3195	30.3610
0.1819	0.3168	18000	0.3076	30.7056
0.1969	0.3520	20000	0.3033	29.4395
0.156	0.3872	22000	0.3137	28.3081
0.1521	0.4224	24000	0.2946	28.2145
0.1736	0.4576	26000	0.2952	27.6800
0.1647	0.4928	28000	0.2889	26.7835
0.1596	0.5280	30000	0.2923	26.6998
0.1586	0.5632	32000	0.2821	26.6561
0.1299	0.5984	34000	0.2775	26.9783
0.1564	0.6336	36000	0.2811	26.4600
0.1525	0.6688	38000	0.2699	26.7485
0.1469	0.7041	40000	0.2699	26.2765
0.1362	0.7393	42000	0.2666	25.4761
0.1268	0.7745	44000	0.2590	26.6236
0.1389	0.8097	46000	0.2617	25.5485
0.1277	0.8449	48000	0.2600	24.7443
0.1312	0.8801	50000	0.2633	24.9579
0.1431	0.9153	52000	0.2604	24.8180
0.1366	0.9505	54000	0.2601	24.4384
0.1363	0.9857	56000	0.2593	24.4634