learning_rate: 5e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 64
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_ratio: 0.1
num_epochs: 30

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	6	0.6841	0.5862
0.7295	2.0	12	0.6780	0.5517
0.7295	3.0	18	0.7819	0.7471
0.6413	4.0	24	0.5217	0.8506
0.5347	5.0	30	0.5265	0.7701
0.5347	6.0	36	0.4109	0.9425
0.4285	7.0	42	0.3706	0.9080
0.4285	8.0	48	0.3199	0.9195
0.3538	9.0	54	0.3152	0.9310
0.3156	10.0	60	0.2475	0.9770
0.3156	11.0	66	0.2624	0.9770
0.2622	12.0	72	0.2020	0.9770
0.2622	13.0	78	0.1916	0.9425
0.2311	14.0	84	0.2467	0.9080
0.2	15.0	90	0.1844	0.9770
0.2	16.0	96	0.1601	0.9770
0.1905	17.0	102	0.1457	0.9655
0.1905	18.0	108	0.4501	0.9425
0.1827	19.0	114	0.1523	0.9770
0.1541	20.0	120	0.2209	0.9195
0.1541	21.0	126	0.1444	0.9885
0.1449	22.0	132	0.1214	0.9885
0.1449	23.0	138	0.1605	0.9425
0.1333	24.0	144	0.1316	0.9540
0.1375	25.0	150	0.1509	0.9540
0.1375	26.0	156	0.1925	0.9425
0.1459	27.0	162	0.1918	0.9540
0.1459	28.0	168	0.0862	0.9770
0.1241	29.0	174	0.1545	0.9655
0.168	30.0	180	0.1845	0.9540

haongn/business_license_clf-2

business_license_clf-2