Whisper Small ko

This model is a fine-tuned version of openai/whisper-large-v3-turbo on the custom 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: 0.0001
train_batch_size: 64
eval_batch_size: 256
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: 50
training_steps: 500
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss
0.8642	0.1449	10	1.5475
0.8061	0.2899	20	1.4627
0.6409	0.4348	30	1.2193
0.3296	0.5797	40	0.8200
0.178	0.7246	50	0.7032
0.1336	0.8696	60	0.6138
0.1064	1.0145	70	0.5193
0.0846	1.1594	80	0.4691
0.0728	1.3043	90	0.4360
0.0681	1.4493	100	0.4071
0.0566	1.5942	110	0.3891
0.0613	1.7391	120	0.3695
0.0441	1.8841	130	0.3587
0.0469	2.0290	140	0.3461
0.0426	2.1739	150	0.3373
0.0383	2.3188	160	0.3249
0.037	2.4638	170	0.3150
0.0516	2.6087	180	0.2967
0.0403	2.7536	190	0.2888
0.045	2.8986	200	0.2782
0.0365	3.0435	210	0.2668
0.0309	3.1884	220	0.2598
0.0341	3.3333	230	0.2542
0.0289	3.4783	240	0.2487
0.0364	3.6232	250	0.2417
0.0353	3.7681	260	0.2372
0.0312	3.9130	270	0.2293
0.0317	4.0580	280	0.2263
0.029	4.2029	290	0.2254
0.0348	4.3478	300	0.2168
0.0299	4.4928	310	0.2101
0.0327	4.6377	320	0.2085
0.0252	4.7826	330	0.2071
0.0246	4.9275	340	0.2020
0.0219	5.0725	350	0.1990
0.0234	5.2174	360	0.1997
0.0269	5.3623	370	0.1969
0.0262	5.5072	380	0.1961
0.0293	5.6522	390	0.1920
0.0247	5.7971	400	0.1891
0.0273	5.9420	410	0.1869
0.0205	6.0870	420	0.1866
0.0168	6.2319	430	0.1860
0.0261	6.3768	440	0.1851
0.0254	6.5217	450	0.1839
0.0258	6.6667	460	0.1830
0.0242	6.8116	470	0.1825
0.0259	6.9565	480	0.1818
0.0234	7.1014	490	0.1813
0.0194	7.2464	500	0.1812