learning_rate: 0.0001
train_batch_size: 256
eval_batch_size: 256
seed: 13
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
6.8358	1.0	391	5.7443
5.4018	2.0	782	5.1832
5.0084	3.0	1173	4.9334
4.784	4.0	1564	4.7744
4.6185	5.0	1955	4.6605
4.484	6.0	2346	4.5739
4.3687	7.0	2737	4.5070
4.2672	8.0	3128	4.4554
4.1753	9.0	3519	4.4167
4.0922	10.0	3910	4.3882
4.0179	11.0	4301	4.3677
3.9516	12.0	4692	4.3529
3.8915	13.0	5083	4.3446
3.8371	14.0	5474	4.3404
3.7888	15.0	5865	4.3370
3.7454	16.0	6256	4.3368
3.7072	17.0	6647	4.3382
3.6747	18.0	7038	4.3401
3.6475	19.0	7429	4.3413
3.627	20.0	7820	4.3431

fpadovani/bnc_o_13

bnc_o_13