learning_rate: 0.0001
train_batch_size: 256
eval_batch_size: 256
seed: 51
optimizer: Use 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: 500
num_epochs: 20
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss
4.4691	1.0	495	3.2805
3.0377	2.0	990	2.9782
2.8107	3.0	1485	2.8653
2.6911	4.0	1980	2.8056
2.6078	5.0	2475	2.7674
2.5419	6.0	2970	2.7416
2.4854	7.0	3465	2.7232
2.4346	8.0	3960	2.7114
2.3875	9.0	4455	2.7084
2.3426	10.0	4950	2.7085
2.2995	11.0	5445	2.7069
2.2582	12.0	5940	2.7145
2.219	13.0	6435	2.7212
2.1812	14.0	6930	2.7267
2.1452	15.0	7425	2.7415
2.1129	16.0	7920	2.7492
2.0848	17.0	8415	2.7625
2.0606	18.0	8910	2.7692
2.0402	19.0	9405	2.7763
2.0243	20.0	9900	2.7795

fpadovani/cds_o_new_51

cds_o_new_51