learning_rate: 1e-05
train_batch_size: 2
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 4
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 200
training_steps: 3000

Training Loss	Epoch	Step	Validation Loss	Wer
0.9712	0.2567	100	0.4179	18.1096
0.7342	0.5135	200	0.3519	16.3781
0.6759	0.7702	300	0.3459	14.2812
0.5592	1.0257	400	0.3234	14.1541
0.2861	1.2824	500	0.3121	13.5663
0.2964	1.5392	600	0.3143	13.4551
0.3229	1.7959	700	0.3115	13.6934
0.1074	2.0513	800	0.3033	13.5504
0.1300	2.3081	900	0.3079	13.1215
0.1056	2.5648	1000	0.3109	12.3749
0.0995	2.8216	1100	0.3163	13.5504
0.0359	3.0770	1200	0.3185	13.2963
0.0557	3.3338	1300	0.3249	13.2010
0.0492	3.5905	1400	0.3260	13.1374
0.0341	3.8472	1500	0.3214	12.9309
0.0149	4.1027	1600	0.3310	12.9309
0.0079	4.3594	1700	0.3334	13.4075
0.0150	4.6162	1800	0.3349	13.4075
0.0167	4.8729	1900	0.3390	13.7411
0.0063	5.1284	2000	0.3424	13.6616
0.0076	5.3851	2100	0.3454	13.6299
0.0064	5.6418	2200	0.3459	13.5187
0.0048	5.8986	2300	0.3474	13.8681
0.0042	6.1540	2400	0.3499	13.7887
0.0038	6.4108	2500	0.3498	13.7569
0.0029	6.6675	2600	0.3524	13.7887
0.0030	6.9243	2700	0.3521	13.9317
0.0027	7.1797	2800	0.3534	14.0111
0.0035	7.4365	2900	0.3536	14.0270
0.0027	7.6932	3000	0.3539	14.0270

AkshitSaxena1/whisper-tiny_to_chinese_accent_v2

Whisper tiny Chinese