learning_rate: 0.0001
train_batch_size: 256
eval_batch_size: 256
seed: 30
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
6.8393	1.0	391	5.7447
5.3976	2.0	782	5.1813
5.0032	3.0	1173	4.9295
4.7792	4.0	1564	4.7712
4.6136	5.0	1955	4.6575
4.4794	6.0	2346	4.5687
4.3645	7.0	2737	4.5038
4.2636	8.0	3128	4.4510
4.173	9.0	3519	4.4146
4.09	10.0	3910	4.3859
4.016	11.0	4301	4.3664
3.9493	12.0	4692	4.3526
3.8894	13.0	5083	4.3425
3.8358	14.0	5474	4.3372
3.7876	15.0	5865	4.3352
3.7442	16.0	6256	4.3358
3.7063	17.0	6647	4.3362
3.6736	18.0	7038	4.3378
3.6464	19.0	7429	4.3405
3.626	20.0	7820	4.3416

fpadovani/bnc_o_30

bnc_o_30