iteration,train_loss,val_loss,aux_loss,load_balancing,z_loss,entropy_loss,learning_rate,elapsed_time,gpu_memory_gb 0,10.912100,10.916830,-0.303276,0.005775,0.860769,-7.902559,0.000000e+00,24.33,5.31 5000,5.315395,5.326223,-0.213804,0.007529,1.399097,-7.224845,7.504294e-04,847.04,5.31 10000,5.109942,5.108413,-0.239121,0.004190,1.268296,-7.402808,4.942498e-04,1667.54,5.31 15000,4.976137,4.956175,-0.241881,0.005522,1.243843,-7.435752,1.921686e-04,2486.64,5.31 20000,4.860514,4.909583,-0.250848,0.002153,1.211053,-7.482130,6.000000e-05,3304.58,5.31 25000,4.854052,4.900092,-0.253470,0.002297,1.192996,-7.501315,6.000000e-05,4120.74,5.31 30000,4.962946,4.879098,-0.251730,0.001949,1.205974,-7.485529,6.000000e-05,4938.61,5.31 35000,4.941273,4.862639,-0.249057,0.005916,1.198534,-7.496536,6.000000e-05,5755.52,5.31