loss,grad_norm,learning_rate,epoch,step,train_runtime,train_samples_per_second,train_steps_per_second,total_flos,train_loss
0.626,0.1765463650226593,0.00019,0.050505050505050504,20,,,,,
0.5359,0.08173302561044693,0.00019874255503213152,0.10101010101010101,40,,,,,
0.517,0.0902973935008049,0.0001947376831715892,0.15151515151515152,60,,,,,
0.5015,0.09258504956960678,0.00018809346758274012,0.20202020202020202,80,,,,,
0.496,0.09724155068397522,0.0001789950127367606,0.25252525252525254,100,,,,,
0.4962,0.0965561643242836,0.00016769579696382597,0.30303030303030304,120,,,,,
0.4908,0.09882321208715439,0.000154510610675594,0.35353535353535354,140,,,,,
0.4884,0.09999062120914459,0.0001398067864642431,0.40404040404040403,160,,,,,
0.489,0.10293831676244736,0.00012399396540305205,0.45454545454545453,180,,,,,
0.486,0.1035684272646904,0.00010751268465418785,0.5050505050505051,200,,,,,
0.4832,0.10418346524238586,9.082210432715197e-05,0.5555555555555556,220,,,,,
0.4772,0.10416751354932785,7.438721551137367e-05,0.6060606060606061,240,,,,,
0.4827,0.10619295388460159,5.866588586066481e-05,0.6565656565656566,260,,,,,
0.4766,0.11086689680814743,4.409610363297211e-05,0.7070707070707071,280,,,,,
0.4789,0.10573258250951767,3.108377556000037e-05,0.7575757575757576,300,,,,,
0.4782,0.10744805634021759,1.999141849185938e-05,0.8080808080808081,320,,,,,
0.4836,0.11020056158304214,1.11280598617714e-05,0.8585858585858586,340,,,,,
0.4807,0.10882183909416199,4.740628338761255e-06,0.9090909090909091,360,,,,,
0.4777,0.10317696630954742,1.0070745203721532e-06,0.9595959595959596,380,,,,,
,,,1.0,396,751.2952,67.313,0.527,9.263508544617984e+16,0.49640953540802