learning_rate: 0.0001
train_batch_size: 256
eval_batch_size: 256
seed: 67
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
7.1819	1.0	392	6.5003
6.3074	2.0	784	6.2637
6.1389	3.0	1176	6.1570
6.0364	4.0	1568	6.0859
5.9555	5.0	1960	6.0310
5.886	6.0	2352	5.9908
5.8225	7.0	2744	5.9578
5.7626	8.0	3136	5.9318
5.7039	9.0	3528	5.9122
5.6469	10.0	3920	5.8963
5.5913	11.0	4312	5.8878
5.5373	12.0	4704	5.8858
5.4848	13.0	5096	5.8851
5.4351	14.0	5488	5.8900
5.3887	15.0	5880	5.8942
5.3465	16.0	6272	5.9013
5.3085	17.0	6664	5.9086
5.277	18.0	7056	5.9148
5.2522	19.0	7448	5.9202
5.234	20.0	7840	5.9234

fpadovani/bnc_sh1_67

bnc_sh1_67