learning_rate: 2e-05
train_batch_size: 16
eval_batch_size: 32
seed: 42
optimizer: Use OptimizerNames.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_ratio: 0.1
num_epochs: 20
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	F1 Macro	Precision	Recall	Accuracy
No log	1.0	114	2.4005	0.0156	0.0086	0.0909	0.0941
No log	2.0	228	2.3264	0.1611	0.1567	0.2007	0.3663
No log	3.0	342	2.2406	0.1608	0.1706	0.2238	0.2079
No log	4.0	456	2.2328	0.1905	0.2013	0.2166	0.3317
2.2426	5.0	570	2.2308	0.2492	0.2648	0.2990	0.3069
2.2426	6.0	684	2.3385	0.2694	0.3077	0.3001	0.3515
2.2426	7.0	798	2.4807	0.2756	0.3171	0.2876	0.3960
2.2426	8.0	912	2.5591	0.2712	0.2808	0.3050	0.3020
0.9769	9.0	1026	2.8100	0.2577	0.2667	0.2705	0.3663

rendchevi/longformer-base-4096-tqacd

longformer-base-4096-tqacd