learning_rate: 1e-05
train_batch_size: 2
eval_batch_size: 2
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 4
optimizer: Use 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: 500
training_steps: 4000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
1.6772	0.0850	100	1.2145	243.8293
1.0201	0.1700	200	0.7053	346.5092
0.4475	0.2550	300	0.2848	111.7772
0.2613	0.3400	400	0.2360	77.3625
0.221	0.4250	500	0.2208	77.6798
0.193	0.5100	600	0.1740	71.6150
0.1901	0.5950	700	0.1513	69.6403
0.151	0.6800	800	0.1385	68.4415
0.1392	0.7650	900	0.1196	66.6784
0.1282	0.8500	1000	0.1186	65.4090
0.1176	0.9350	1100	0.1104	65.8322
0.1105	1.0195	1200	0.1103	69.9929
0.0829	1.1045	1300	0.1029	65.8674
0.0744	1.1895	1400	0.0943	62.9055
0.0812	1.2745	1500	0.0913	69.8166
0.0816	1.3595	1600	0.0864	59.7320
0.0723	1.4445	1700	0.0806	59.4499
0.071	1.5295	1800	0.0801	62.3766
0.0686	1.6145	1900	0.0780	60.1904
0.0658	1.6995	2000	0.0736	67.3836
0.0626	1.7845	2100	0.0686	58.7447
0.0647	1.8695	2200	0.0653	54.6544
0.0534	1.9545	2300	0.0646	57.3343
0.0522	2.0391	2400	0.0646	53.7024
0.0342	2.1241	2500	0.0626	55.0776
0.0335	2.2091	2600	0.0598	54.0550
0.0384	2.2941	2700	0.0604	54.3018
0.0379	2.3791	2800	0.0541	51.5867
0.0344	2.4641	2900	0.0536	59.0621
0.0248	2.5491	3000	0.0543	51.1989
0.0313	2.6341	3100	0.0530	50.5289
0.0269	2.7191	3200	0.0521	50.0
0.0285	2.8041	3300	0.0487	50.3526
0.0244	2.8891	3400	0.0480	50.5642
0.0207	2.9741	3500	0.0463	49.5416
0.0248	3.0586	3600	0.0472	49.1890
0.0141	3.1436	3700	0.0465	49.1185
0.0142	3.2286	3800	0.0458	48.2370
0.0127	3.3136	3900	0.0457	47.5317
0.0112	3.3986	4000	0.0454	47.9901

tranha1412/whisper-finetuned-v3_4kstep_augment_v2

whisper-finetuned-v3_4kstep_augment_v2