whisper-large-sorani-v1

This model is a fine-tuned version of openai/whisper-large-v3 on the None 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: 3e-05
train_batch_size: 16
eval_batch_size: 16
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: 1250
num_epochs: 15
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
0.2589	0.3365	500	0.2832	40.7338
0.227	0.6729	1000	0.2558	38.9114
0.2096	1.0094	1500	0.2204	34.3194
0.1681	1.3459	2000	0.2135	34.3436
0.1577	1.6824	2500	0.1895	29.9739
0.091	2.0188	3000	0.1869	28.1516
0.1088	2.3553	3500	0.1755	27.0302
0.1004	2.6918	4000	0.1702	26.5178
0.066	3.0283	4500	0.1733	24.9807
0.06	3.3647	5000	0.1719	24.7438
0.0649	3.7012	5500	0.1676	24.0526
0.0373	4.0377	6000	0.1743	22.6266
0.0383	4.3742	6500	0.1766	23.8496
0.0402	4.7106	7000	0.1747	23.1342
0.021	5.0471	7500	0.1886	22.9070
0.0256	5.3836	8000	0.1931	22.2641
0.0293	5.7201	8500	0.1820	21.4085
0.0175	6.0565	9000	0.1881	21.8677
0.0158	6.3930	9500	0.1936	21.8484
0.0195	6.7295	10000	0.1916	21.7952
0.0122	7.0659	10500	0.1994	21.2297
0.0103	7.4024	11000	0.1960	20.1131
0.0134	7.7389	11500	0.2023	20.9397
0.0055	8.0754	12000	0.2099	20.5143
0.0062	8.4118	12500	0.2074	20.4225
0.0079	8.7483	13000	0.2001	20.0503
0.0074	9.0848	13500	0.2001	19.5379
0.0061	9.4213	14000	0.1991	19.3929
0.0048	9.7577	14500	0.2006	19.1319
0.0023	10.0942	15000	0.2114	19.4896
0.0027	10.4307	15500	0.2041	18.6050
0.0034	10.7672	16000	0.2042	18.6195
0.0015	11.1036	16500	0.2010	18.3440
0.0016	11.4401	17000	0.2049	18.1506
0.0034	11.7766	17500	0.2119	18.5663
0.001	12.1131	18000	0.2096	17.5609
0.0014	12.4495	18500	0.2138	18.0878
0.0011	12.7860	19000	0.2068	17.5802
0.0003	13.1225	19500	0.2102	17.5319
0.0002	13.4590	20000	0.2146	17.4159
0.0002	13.7954	20500	0.2169	17.4352
0.0003	14.1319	21000	0.2196	17.0775
0.0	14.4684	21500	0.2206	17.0534
0.0	14.8048	22000	0.2215	17.0630