learning_rate: 0.001
train_batch_size: 32
eval_batch_size: 64
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 256
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: 32000
num_epochs: 20.0
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Accuracy
6.0432	0.9994	1506	4.4507	0.2897
4.0466	1.9997	3013	3.9699	0.3279
3.8106	2.9993	4519	3.7055	0.3498
3.5237	3.9995	6026	3.5475	0.3649
3.4169	4.9998	7533	3.4481	0.3747
3.2979	5.9993	9039	3.3910	0.3801
3.2311	6.9996	10546	3.3506	0.3844
3.1783	7.9998	12053	3.3274	0.3870
3.1263	8.9994	13559	3.3091	0.3888
3.1062	9.9997	15066	3.2951	0.3908
3.0612	10.9993	16572	3.2888	0.3917
3.0553	11.9995	18079	3.2795	0.3930
3.015	12.9998	19586	3.2736	0.3935
3.0208	13.9993	21092	3.2689	0.3942
2.9833	14.9996	22599	3.2651	0.3948
2.9977	15.9998	24106	3.2643	0.3951
2.9609	16.9994	25612	3.2610	0.3957
2.9794	17.9997	27119	3.2588	0.3958
2.9463	18.9993	28625	3.2590	0.3958
2.9699	19.9915	30120	3.2614	0.3960

qing-yao/short_first_small_seed-42_1e-3

short_first_small_seed-42_1e-3