learning_rate: 0.0001
train_batch_size: 64
eval_batch_size: 8
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_steps: 10000
num_epochs: 603
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss
6.7876	4.0161	2000	5.7572
5.6570	8.0321	4000	5.6031
5.4869	12.0482	6000	5.0994
4.7571	16.0643	8000	4.4091
4.2246	20.0803	10000	3.9246
3.6582	24.0964	12000	3.1099
2.6372	28.1124	14000	2.0094
1.8545	32.1285	16000	1.5949
1.4873	36.1446	18000	1.3468
1.2508	40.1606	20000	1.1876
1.0889	44.1767	22000	1.0385
0.9580	48.1928	24000	0.9333
0.8578	52.2088	26000	0.8454
0.7791	56.2249	28000	0.7766
0.7070	60.2410	30000	0.7423
0.6498	64.2570	32000	0.6968
0.6035	68.2731	34000	0.6754
0.5591	72.2892	36000	0.6412
0.5222	76.3052	38000	0.5862
0.4908	80.3213	40000	0.5847
0.4632	84.3373	42000	0.5529
0.4383	88.3534	44000	0.5341
0.4163	92.3695	46000	0.5323
0.3944	96.3855	48000	0.5072
0.3762	100.4016	50000	0.4913
0.3596	104.4177	52000	0.4944
0.3429	108.4337	54000	0.4898
0.3278	112.4498	56000	0.4626
0.3164	116.4659	58000	0.4516
0.3056	120.4819	60000	0.4548
0.2926	124.4980	62000	0.4538
0.2825	128.5141	64000	0.4484
0.2747	132.5301	66000	0.4448
0.2643	136.5462	68000	0.4394
0.2554	140.5622	70000	0.4172
0.2484	144.5783	72000	0.4299
0.2422	148.5944	74000	0.4298
0.2345	152.6104	76000	0.4291

wenda2025/QBert

QBert