learning_rate: 0.0001
train_batch_size: 256
eval_batch_size: 256
seed: 42
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
4.585	1.0	496	3.5160
3.3235	2.0	992	3.3005
3.1692	3.0	1488	3.2236
3.0793	4.0	1984	3.1785
3.0144	5.0	2480	3.1462
2.9616	6.0	2976	3.1218
2.9137	7.0	3472	3.1084
2.8691	8.0	3968	3.0996
2.8269	9.0	4464	3.0946
2.7864	10.0	4960	3.0964
2.7482	11.0	5456	3.1013
2.7091	12.0	5952	3.1038
2.6712	13.0	6448	3.1147
2.637	14.0	6944	3.1215
2.6037	15.0	7440	3.1335
2.571	16.0	7936	3.1451
2.5432	17.0	8432	3.1547
2.5189	18.0	8928	3.1651
2.4994	19.0	9424	3.1719
2.4839	20.0	9920	3.1754

fpadovani/cds_w_42

cds_w_42