loss	grad_norm	learning_rate	epoch	step	eval_loss	eval_f1_micro	eval_f1_macro	eval_precision	eval_recall	eval_runtime	eval_samples_per_second	eval_steps_per_second	train_runtime	train_samples_per_second	train_steps_per_second	total_flos	train_loss
0.0316	0.0846693217754364	5.1690821256038647e-05	0.4830917874396135	500													
0.0076	0.034613098949193954	3.3816425120772947e-06	0.966183574879227	1000													
			1.0	1035	0.027310708537697792	0.9825396825396825	0.9808089925902543	0.9880287310454908	0.9771112865035517	8.4322	107.089	6.76					
			1.0	1035									95.2688	86.912	10.864	7716522048675840.0	0.019097913535320817