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
6.2642	1.0	1731	4.5519	0.2722
4.3675	2.0	3462	4.0802	0.3075
3.9677	3.0	5193	3.8310	0.3286
3.832	4.0	6924	3.6886	0.3424
3.6544	5.0	8655	3.6002	0.3507
3.5548	6.0	10386	3.5555	0.3547
3.4849	7.0	12117	3.5217	0.3581
3.4337	8.0	13848	3.4956	0.3609
3.3926	9.0	15579	3.4885	0.3620
3.37	10.0	17310	3.4666	0.3640
3.3518	11.0	19041	3.4662	0.3643
3.3042	12.0	20772	3.4562	0.3655
3.2997	13.0	22503	3.4451	0.3669
3.2979	14.0	24234	3.4461	0.3671
3.2648	15.0	25965	3.4455	0.3669
3.2516	16.0	27696	3.4412	0.3674
3.264	17.0	29427	3.4405	0.3675
3.271	18.0	31158	3.4360	0.3676
3.2386	19.0	32889	3.4024	0.3719
3.1078	20.0	34620	3.3644	0.3778

qing-yao/random_first_nodatives_seed-42_1e-3

random_first_datives_seed-42_1e-3