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
6.8365	1.0	391	5.7459
5.3979	2.0	782	5.1817
5.0061	3.0	1173	4.9305
4.7815	4.0	1564	4.7744
4.6165	5.0	1955	4.6589
4.4815	6.0	2346	4.5716
4.3662	7.0	2737	4.5057
4.2647	8.0	3128	4.4560
4.1728	9.0	3519	4.4148
4.0896	10.0	3910	4.3853
4.0153	11.0	4301	4.3670
3.9487	12.0	4692	4.3534
3.8886	13.0	5083	4.3445
3.8345	14.0	5474	4.3405
3.786	15.0	5865	4.3356
3.7426	16.0	6256	4.3361
3.7044	17.0	6647	4.3377
3.6715	18.0	7038	4.3393
3.6449	19.0	7429	4.3405
3.624	20.0	7820	4.3424

fpadovani/bnc_o_67

bnc_o_67