OpenR1-Math-Raw-all-correct-5k / trainer_state.json
neginr's picture
End of training
abdead8 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 6.955414012738854,
"eval_steps": 500,
"global_step": 364,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01910828025477707,
"grad_norm": 6.241959257018338,
"learning_rate": 5.405405405405406e-07,
"loss": 0.8081,
"step": 1
},
{
"epoch": 0.03821656050955414,
"grad_norm": 5.93951610288608,
"learning_rate": 1.0810810810810812e-06,
"loss": 0.7932,
"step": 2
},
{
"epoch": 0.05732484076433121,
"grad_norm": 6.117331699970184,
"learning_rate": 1.6216216216216219e-06,
"loss": 0.8046,
"step": 3
},
{
"epoch": 0.07643312101910828,
"grad_norm": 6.056401286764686,
"learning_rate": 2.1621621621621623e-06,
"loss": 0.8089,
"step": 4
},
{
"epoch": 0.09554140127388536,
"grad_norm": 5.670547996563616,
"learning_rate": 2.702702702702703e-06,
"loss": 0.7742,
"step": 5
},
{
"epoch": 0.11464968152866242,
"grad_norm": 4.557860256689445,
"learning_rate": 3.2432432432432437e-06,
"loss": 0.7241,
"step": 6
},
{
"epoch": 0.1337579617834395,
"grad_norm": 2.6659027981397676,
"learning_rate": 3.7837837837837844e-06,
"loss": 0.671,
"step": 7
},
{
"epoch": 0.15286624203821655,
"grad_norm": 2.4849109480390954,
"learning_rate": 4.324324324324325e-06,
"loss": 0.6766,
"step": 8
},
{
"epoch": 0.17197452229299362,
"grad_norm": 2.1624316975859803,
"learning_rate": 4.864864864864866e-06,
"loss": 0.6856,
"step": 9
},
{
"epoch": 0.1910828025477707,
"grad_norm": 4.148298157311749,
"learning_rate": 5.405405405405406e-06,
"loss": 0.6905,
"step": 10
},
{
"epoch": 0.21019108280254778,
"grad_norm": 4.310486230532461,
"learning_rate": 5.945945945945947e-06,
"loss": 0.6828,
"step": 11
},
{
"epoch": 0.22929936305732485,
"grad_norm": 4.221373334551036,
"learning_rate": 6.486486486486487e-06,
"loss": 0.6499,
"step": 12
},
{
"epoch": 0.2484076433121019,
"grad_norm": 4.159497021896487,
"learning_rate": 7.027027027027028e-06,
"loss": 0.605,
"step": 13
},
{
"epoch": 0.267515923566879,
"grad_norm": 3.9649148085408172,
"learning_rate": 7.567567567567569e-06,
"loss": 0.5955,
"step": 14
},
{
"epoch": 0.28662420382165604,
"grad_norm": 3.1244164325925077,
"learning_rate": 8.108108108108109e-06,
"loss": 0.5976,
"step": 15
},
{
"epoch": 0.3057324840764331,
"grad_norm": 2.2839183725846848,
"learning_rate": 8.64864864864865e-06,
"loss": 0.5812,
"step": 16
},
{
"epoch": 0.3248407643312102,
"grad_norm": 1.7777516271275047,
"learning_rate": 9.189189189189191e-06,
"loss": 0.5687,
"step": 17
},
{
"epoch": 0.34394904458598724,
"grad_norm": 2.509152685540221,
"learning_rate": 9.729729729729732e-06,
"loss": 0.5883,
"step": 18
},
{
"epoch": 0.3630573248407643,
"grad_norm": 2.4432996504137403,
"learning_rate": 1.027027027027027e-05,
"loss": 0.5751,
"step": 19
},
{
"epoch": 0.3821656050955414,
"grad_norm": 1.8422353813581844,
"learning_rate": 1.0810810810810812e-05,
"loss": 0.5499,
"step": 20
},
{
"epoch": 0.4012738853503185,
"grad_norm": 1.1057909974783728,
"learning_rate": 1.1351351351351352e-05,
"loss": 0.5235,
"step": 21
},
{
"epoch": 0.42038216560509556,
"grad_norm": 0.9697190839788165,
"learning_rate": 1.1891891891891894e-05,
"loss": 0.5215,
"step": 22
},
{
"epoch": 0.4394904458598726,
"grad_norm": 1.2122502380652562,
"learning_rate": 1.2432432432432433e-05,
"loss": 0.505,
"step": 23
},
{
"epoch": 0.4585987261146497,
"grad_norm": 1.0682411101830946,
"learning_rate": 1.2972972972972975e-05,
"loss": 0.4965,
"step": 24
},
{
"epoch": 0.47770700636942676,
"grad_norm": 0.6931141004599464,
"learning_rate": 1.3513513513513515e-05,
"loss": 0.5033,
"step": 25
},
{
"epoch": 0.4968152866242038,
"grad_norm": 0.8120655295262994,
"learning_rate": 1.4054054054054055e-05,
"loss": 0.4907,
"step": 26
},
{
"epoch": 0.5159235668789809,
"grad_norm": 0.8605412945791352,
"learning_rate": 1.4594594594594596e-05,
"loss": 0.5108,
"step": 27
},
{
"epoch": 0.535031847133758,
"grad_norm": 0.6950283976570016,
"learning_rate": 1.5135135135135138e-05,
"loss": 0.4904,
"step": 28
},
{
"epoch": 0.554140127388535,
"grad_norm": 0.8008602321042414,
"learning_rate": 1.5675675675675676e-05,
"loss": 0.4985,
"step": 29
},
{
"epoch": 0.5732484076433121,
"grad_norm": 0.6788741344833235,
"learning_rate": 1.6216216216216218e-05,
"loss": 0.4864,
"step": 30
},
{
"epoch": 0.5923566878980892,
"grad_norm": 0.5847327715971911,
"learning_rate": 1.6756756756756757e-05,
"loss": 0.4787,
"step": 31
},
{
"epoch": 0.6114649681528662,
"grad_norm": 0.6295835464148027,
"learning_rate": 1.72972972972973e-05,
"loss": 0.4756,
"step": 32
},
{
"epoch": 0.6305732484076433,
"grad_norm": 0.6953578678731234,
"learning_rate": 1.783783783783784e-05,
"loss": 0.4892,
"step": 33
},
{
"epoch": 0.6496815286624203,
"grad_norm": 0.6299509679567121,
"learning_rate": 1.8378378378378383e-05,
"loss": 0.4598,
"step": 34
},
{
"epoch": 0.6687898089171974,
"grad_norm": 0.6874759428693075,
"learning_rate": 1.891891891891892e-05,
"loss": 0.4892,
"step": 35
},
{
"epoch": 0.6878980891719745,
"grad_norm": 0.8115656896008068,
"learning_rate": 1.9459459459459463e-05,
"loss": 0.4838,
"step": 36
},
{
"epoch": 0.7070063694267515,
"grad_norm": 0.6244827147022581,
"learning_rate": 2e-05,
"loss": 0.4664,
"step": 37
},
{
"epoch": 0.7261146496815286,
"grad_norm": 0.8961283547814942,
"learning_rate": 1.9999538500851633e-05,
"loss": 0.4845,
"step": 38
},
{
"epoch": 0.7452229299363057,
"grad_norm": 0.5525724792788775,
"learning_rate": 1.9998154046002822e-05,
"loss": 0.4465,
"step": 39
},
{
"epoch": 0.7643312101910829,
"grad_norm": 0.6555379352374421,
"learning_rate": 1.9995846763238514e-05,
"loss": 0.4697,
"step": 40
},
{
"epoch": 0.7834394904458599,
"grad_norm": 0.5631135206229095,
"learning_rate": 1.9992616865520515e-05,
"loss": 0.4504,
"step": 41
},
{
"epoch": 0.802547770700637,
"grad_norm": 0.6789028072692714,
"learning_rate": 1.9988464650967834e-05,
"loss": 0.4749,
"step": 42
},
{
"epoch": 0.821656050955414,
"grad_norm": 0.672643203759416,
"learning_rate": 1.9983390502829168e-05,
"loss": 0.4657,
"step": 43
},
{
"epoch": 0.8407643312101911,
"grad_norm": 0.5704503082200754,
"learning_rate": 1.9977394889447526e-05,
"loss": 0.4796,
"step": 44
},
{
"epoch": 0.8598726114649682,
"grad_norm": 0.6325073555425712,
"learning_rate": 1.9970478364217e-05,
"loss": 0.4399,
"step": 45
},
{
"epoch": 0.8789808917197452,
"grad_norm": 0.6277184529351723,
"learning_rate": 1.9962641565531694e-05,
"loss": 0.4766,
"step": 46
},
{
"epoch": 0.8980891719745223,
"grad_norm": 0.5321974027084573,
"learning_rate": 1.9953885216726788e-05,
"loss": 0.4462,
"step": 47
},
{
"epoch": 0.9171974522292994,
"grad_norm": 0.5351714897610934,
"learning_rate": 1.994421012601179e-05,
"loss": 0.4463,
"step": 48
},
{
"epoch": 0.9363057324840764,
"grad_norm": 0.510188109232125,
"learning_rate": 1.9933617186395917e-05,
"loss": 0.4444,
"step": 49
},
{
"epoch": 0.9554140127388535,
"grad_norm": 0.6379736712943488,
"learning_rate": 1.99221073756057e-05,
"loss": 0.4609,
"step": 50
},
{
"epoch": 0.9745222929936306,
"grad_norm": 0.6073838438518463,
"learning_rate": 1.990968175599471e-05,
"loss": 0.4661,
"step": 51
},
{
"epoch": 0.9936305732484076,
"grad_norm": 0.6290637854162586,
"learning_rate": 1.9896341474445526e-05,
"loss": 0.4453,
"step": 52
},
{
"epoch": 1.0127388535031847,
"grad_norm": 0.5062948453792918,
"learning_rate": 1.9882087762263857e-05,
"loss": 0.4048,
"step": 53
},
{
"epoch": 1.0318471337579618,
"grad_norm": 0.5838989311580737,
"learning_rate": 1.9866921935064907e-05,
"loss": 0.4051,
"step": 54
},
{
"epoch": 1.0509554140127388,
"grad_norm": 0.5738805783291178,
"learning_rate": 1.985084539265195e-05,
"loss": 0.4097,
"step": 55
},
{
"epoch": 1.070063694267516,
"grad_norm": 0.47723217247191047,
"learning_rate": 1.983385961888711e-05,
"loss": 0.3788,
"step": 56
},
{
"epoch": 1.089171974522293,
"grad_norm": 0.5874935556940952,
"learning_rate": 1.9815966181554412e-05,
"loss": 0.3961,
"step": 57
},
{
"epoch": 1.10828025477707,
"grad_norm": 0.527376184226625,
"learning_rate": 1.9797166732215078e-05,
"loss": 0.3954,
"step": 58
},
{
"epoch": 1.127388535031847,
"grad_norm": 0.593272660193629,
"learning_rate": 1.977746300605507e-05,
"loss": 0.3968,
"step": 59
},
{
"epoch": 1.1464968152866242,
"grad_norm": 0.5980471015615794,
"learning_rate": 1.975685682172497e-05,
"loss": 0.3913,
"step": 60
},
{
"epoch": 1.1656050955414012,
"grad_norm": 0.5235527151254838,
"learning_rate": 1.973535008117207e-05,
"loss": 0.3751,
"step": 61
},
{
"epoch": 1.1847133757961783,
"grad_norm": 0.5958941715264482,
"learning_rate": 1.9712944769464864e-05,
"loss": 0.3815,
"step": 62
},
{
"epoch": 1.2038216560509554,
"grad_norm": 0.5761097172892824,
"learning_rate": 1.9689642954609808e-05,
"loss": 0.3786,
"step": 63
},
{
"epoch": 1.2229299363057324,
"grad_norm": 0.5458840041937888,
"learning_rate": 1.9665446787360444e-05,
"loss": 0.3793,
"step": 64
},
{
"epoch": 1.2420382165605095,
"grad_norm": 0.4971007114783199,
"learning_rate": 1.9640358501018885e-05,
"loss": 0.3612,
"step": 65
},
{
"epoch": 1.2611464968152866,
"grad_norm": 0.6718347539466439,
"learning_rate": 1.9614380411229693e-05,
"loss": 0.3624,
"step": 66
},
{
"epoch": 1.2802547770700636,
"grad_norm": 0.5064878166036954,
"learning_rate": 1.9587514915766124e-05,
"loss": 0.376,
"step": 67
},
{
"epoch": 1.2993630573248407,
"grad_norm": 0.6246815034620667,
"learning_rate": 1.9559764494308838e-05,
"loss": 0.3596,
"step": 68
},
{
"epoch": 1.3184713375796178,
"grad_norm": 0.5067838357996096,
"learning_rate": 1.9531131708217005e-05,
"loss": 0.3809,
"step": 69
},
{
"epoch": 1.3375796178343948,
"grad_norm": 0.5903894364762551,
"learning_rate": 1.950161920029191e-05,
"loss": 0.3876,
"step": 70
},
{
"epoch": 1.356687898089172,
"grad_norm": 0.5222173668941626,
"learning_rate": 1.9471229694533003e-05,
"loss": 0.3922,
"step": 71
},
{
"epoch": 1.3757961783439492,
"grad_norm": 0.5456976626348087,
"learning_rate": 1.943996599588649e-05,
"loss": 0.3866,
"step": 72
},
{
"epoch": 1.394904458598726,
"grad_norm": 0.5952350371600565,
"learning_rate": 1.940783098998643e-05,
"loss": 0.3744,
"step": 73
},
{
"epoch": 1.4140127388535033,
"grad_norm": 0.5368407199291275,
"learning_rate": 1.93748276428884e-05,
"loss": 0.3843,
"step": 74
},
{
"epoch": 1.4331210191082802,
"grad_norm": 0.48047359832300196,
"learning_rate": 1.9340959000795707e-05,
"loss": 0.3716,
"step": 75
},
{
"epoch": 1.4522292993630574,
"grad_norm": 0.5476639491643883,
"learning_rate": 1.9306228189778255e-05,
"loss": 0.3879,
"step": 76
},
{
"epoch": 1.4713375796178343,
"grad_norm": 0.5521373454836106,
"learning_rate": 1.927063841548398e-05,
"loss": 0.3979,
"step": 77
},
{
"epoch": 1.4904458598726116,
"grad_norm": 0.5163630198734224,
"learning_rate": 1.9234192962842996e-05,
"loss": 0.3727,
"step": 78
},
{
"epoch": 1.5095541401273884,
"grad_norm": 0.5377802325038572,
"learning_rate": 1.9196895195764363e-05,
"loss": 0.3772,
"step": 79
},
{
"epoch": 1.5286624203821657,
"grad_norm": 0.48305304783219255,
"learning_rate": 1.9158748556825637e-05,
"loss": 0.3761,
"step": 80
},
{
"epoch": 1.5477707006369426,
"grad_norm": 0.5657771713112786,
"learning_rate": 1.9119756566955092e-05,
"loss": 0.3869,
"step": 81
},
{
"epoch": 1.5668789808917198,
"grad_norm": 0.5112601627783994,
"learning_rate": 1.907992282510675e-05,
"loss": 0.3843,
"step": 82
},
{
"epoch": 1.5859872611464967,
"grad_norm": 0.5424671583111911,
"learning_rate": 1.90392510079282e-05,
"loss": 0.3822,
"step": 83
},
{
"epoch": 1.605095541401274,
"grad_norm": 0.4742501669637294,
"learning_rate": 1.8997744869421248e-05,
"loss": 0.346,
"step": 84
},
{
"epoch": 1.6242038216560508,
"grad_norm": 0.5249105358247995,
"learning_rate": 1.8955408240595396e-05,
"loss": 0.3649,
"step": 85
},
{
"epoch": 1.643312101910828,
"grad_norm": 0.5097036204556638,
"learning_rate": 1.891224502911428e-05,
"loss": 0.3746,
"step": 86
},
{
"epoch": 1.662420382165605,
"grad_norm": 0.5662117113906048,
"learning_rate": 1.886825921893497e-05,
"loss": 0.3756,
"step": 87
},
{
"epoch": 1.6815286624203822,
"grad_norm": 0.6328283864951902,
"learning_rate": 1.8823454869940243e-05,
"loss": 0.3614,
"step": 88
},
{
"epoch": 1.700636942675159,
"grad_norm": 0.6269449890382989,
"learning_rate": 1.8777836117563894e-05,
"loss": 0.4006,
"step": 89
},
{
"epoch": 1.7197452229299364,
"grad_norm": 0.5331460508011052,
"learning_rate": 1.873140717240899e-05,
"loss": 0.3749,
"step": 90
},
{
"epoch": 1.7388535031847132,
"grad_norm": 0.5786939187599284,
"learning_rate": 1.8684172319859258e-05,
"loss": 0.3845,
"step": 91
},
{
"epoch": 1.7579617834394905,
"grad_norm": 0.4907634197389636,
"learning_rate": 1.863613591968355e-05,
"loss": 0.3818,
"step": 92
},
{
"epoch": 1.7770700636942676,
"grad_norm": 0.5202257762170518,
"learning_rate": 1.858730240563342e-05,
"loss": 0.3595,
"step": 93
},
{
"epoch": 1.7961783439490446,
"grad_norm": 0.48160580323969776,
"learning_rate": 1.8537676285033886e-05,
"loss": 0.389,
"step": 94
},
{
"epoch": 1.8152866242038217,
"grad_norm": 0.5302833798878662,
"learning_rate": 1.848726213836744e-05,
"loss": 0.3748,
"step": 95
},
{
"epoch": 1.8343949044585988,
"grad_norm": 0.5140528411189847,
"learning_rate": 1.8436064618851225e-05,
"loss": 0.3711,
"step": 96
},
{
"epoch": 1.8535031847133758,
"grad_norm": 0.5064031163316228,
"learning_rate": 1.838408845200758e-05,
"loss": 0.3644,
"step": 97
},
{
"epoch": 1.872611464968153,
"grad_norm": 0.503838093734225,
"learning_rate": 1.8331338435227838e-05,
"loss": 0.3793,
"step": 98
},
{
"epoch": 1.89171974522293,
"grad_norm": 0.5178504358391139,
"learning_rate": 1.8277819437329577e-05,
"loss": 0.3927,
"step": 99
},
{
"epoch": 1.910828025477707,
"grad_norm": 0.5616485588822877,
"learning_rate": 1.8223536398107177e-05,
"loss": 0.3639,
"step": 100
},
{
"epoch": 1.929936305732484,
"grad_norm": 0.6369498890683533,
"learning_rate": 1.8168494327875918e-05,
"loss": 0.367,
"step": 101
},
{
"epoch": 1.9490445859872612,
"grad_norm": 0.5150930748541498,
"learning_rate": 1.8112698307009506e-05,
"loss": 0.3693,
"step": 102
},
{
"epoch": 1.9681528662420382,
"grad_norm": 0.5520960002034425,
"learning_rate": 1.8056153485471167e-05,
"loss": 0.3968,
"step": 103
},
{
"epoch": 1.9872611464968153,
"grad_norm": 0.5384367311815907,
"learning_rate": 1.799886508233829e-05,
"loss": 0.3792,
"step": 104
},
{
"epoch": 2.0063694267515926,
"grad_norm": 0.4833737140964603,
"learning_rate": 1.7940838385320732e-05,
"loss": 0.3335,
"step": 105
},
{
"epoch": 2.0254777070063694,
"grad_norm": 0.5260927691278876,
"learning_rate": 1.788207875027274e-05,
"loss": 0.2883,
"step": 106
},
{
"epoch": 2.0445859872611467,
"grad_norm": 0.8384414335598387,
"learning_rate": 1.7822591600698632e-05,
"loss": 0.2775,
"step": 107
},
{
"epoch": 2.0636942675159236,
"grad_norm": 0.5555880744650614,
"learning_rate": 1.776238242725217e-05,
"loss": 0.2912,
"step": 108
},
{
"epoch": 2.082802547770701,
"grad_norm": 0.6412087578595141,
"learning_rate": 1.7701456787229805e-05,
"loss": 0.2967,
"step": 109
},
{
"epoch": 2.1019108280254777,
"grad_norm": 0.5316627462328931,
"learning_rate": 1.7639820304057745e-05,
"loss": 0.3031,
"step": 110
},
{
"epoch": 2.121019108280255,
"grad_norm": 0.560050876705896,
"learning_rate": 1.7577478666772886e-05,
"loss": 0.2799,
"step": 111
},
{
"epoch": 2.140127388535032,
"grad_norm": 0.5206350732304817,
"learning_rate": 1.751443762949772e-05,
"loss": 0.2815,
"step": 112
},
{
"epoch": 2.159235668789809,
"grad_norm": 0.5161770774443827,
"learning_rate": 1.7450703010909263e-05,
"loss": 0.2821,
"step": 113
},
{
"epoch": 2.178343949044586,
"grad_norm": 0.5357328170213232,
"learning_rate": 1.738628069370195e-05,
"loss": 0.2898,
"step": 114
},
{
"epoch": 2.1974522292993632,
"grad_norm": 0.5451898462288003,
"learning_rate": 1.732117662404469e-05,
"loss": 0.2709,
"step": 115
},
{
"epoch": 2.21656050955414,
"grad_norm": 0.5554145421963202,
"learning_rate": 1.7255396811032014e-05,
"loss": 0.2781,
"step": 116
},
{
"epoch": 2.2356687898089174,
"grad_norm": 0.5324778549304677,
"learning_rate": 1.718894732612947e-05,
"loss": 0.2703,
"step": 117
},
{
"epoch": 2.254777070063694,
"grad_norm": 0.6007868123281157,
"learning_rate": 1.712183430261319e-05,
"loss": 0.2974,
"step": 118
},
{
"epoch": 2.2738853503184715,
"grad_norm": 0.5360814915069948,
"learning_rate": 1.7054063935003813e-05,
"loss": 0.2691,
"step": 119
},
{
"epoch": 2.2929936305732483,
"grad_norm": 0.48460906848899765,
"learning_rate": 1.698564247849473e-05,
"loss": 0.2773,
"step": 120
},
{
"epoch": 2.3121019108280256,
"grad_norm": 0.5239909963106646,
"learning_rate": 1.691657624837472e-05,
"loss": 0.2837,
"step": 121
},
{
"epoch": 2.3312101910828025,
"grad_norm": 0.5390366512991357,
"learning_rate": 1.684687161944506e-05,
"loss": 0.2899,
"step": 122
},
{
"epoch": 2.3503184713375798,
"grad_norm": 0.5279708871631977,
"learning_rate": 1.677653502543113e-05,
"loss": 0.2752,
"step": 123
},
{
"epoch": 2.3694267515923566,
"grad_norm": 0.4794831780046385,
"learning_rate": 1.6705572958388576e-05,
"loss": 0.2909,
"step": 124
},
{
"epoch": 2.388535031847134,
"grad_norm": 0.4938209392160801,
"learning_rate": 1.6633991968104095e-05,
"loss": 0.265,
"step": 125
},
{
"epoch": 2.4076433121019107,
"grad_norm": 0.5120347098421275,
"learning_rate": 1.6561798661490904e-05,
"loss": 0.2716,
"step": 126
},
{
"epoch": 2.426751592356688,
"grad_norm": 0.5345557237257497,
"learning_rate": 1.6488999701978905e-05,
"loss": 0.2731,
"step": 127
},
{
"epoch": 2.445859872611465,
"grad_norm": 0.5058403480580227,
"learning_rate": 1.6415601808899658e-05,
"loss": 0.2875,
"step": 128
},
{
"epoch": 2.464968152866242,
"grad_norm": 0.42703379280046516,
"learning_rate": 1.63416117568662e-05,
"loss": 0.279,
"step": 129
},
{
"epoch": 2.484076433121019,
"grad_norm": 0.5099199849885778,
"learning_rate": 1.6267036375147728e-05,
"loss": 0.2756,
"step": 130
},
{
"epoch": 2.5031847133757963,
"grad_norm": 0.557742919039989,
"learning_rate": 1.619188254703927e-05,
"loss": 0.2655,
"step": 131
},
{
"epoch": 2.522292993630573,
"grad_norm": 0.422062647710303,
"learning_rate": 1.6116157209226356e-05,
"loss": 0.3068,
"step": 132
},
{
"epoch": 2.5414012738853504,
"grad_norm": 0.567120633664578,
"learning_rate": 1.6039867351144778e-05,
"loss": 0.2914,
"step": 133
},
{
"epoch": 2.5605095541401273,
"grad_norm": 0.47571035654985677,
"learning_rate": 1.5963020014335437e-05,
"loss": 0.2758,
"step": 134
},
{
"epoch": 2.5796178343949046,
"grad_norm": 0.5298643684310551,
"learning_rate": 1.588562229179443e-05,
"loss": 0.2585,
"step": 135
},
{
"epoch": 2.5987261146496814,
"grad_norm": 0.5495367887893668,
"learning_rate": 1.5807681327318372e-05,
"loss": 0.2903,
"step": 136
},
{
"epoch": 2.6178343949044587,
"grad_norm": 0.5538129209735884,
"learning_rate": 1.5729204314845002e-05,
"loss": 0.2873,
"step": 137
},
{
"epoch": 2.6369426751592355,
"grad_norm": 0.47126203961555707,
"learning_rate": 1.56501984977892e-05,
"loss": 0.2897,
"step": 138
},
{
"epoch": 2.656050955414013,
"grad_norm": 0.6176273411501721,
"learning_rate": 1.557067116837444e-05,
"loss": 0.305,
"step": 139
},
{
"epoch": 2.6751592356687897,
"grad_norm": 0.4362824175670899,
"learning_rate": 1.5490629666959668e-05,
"loss": 0.3054,
"step": 140
},
{
"epoch": 2.694267515923567,
"grad_norm": 0.5344146206915595,
"learning_rate": 1.541008138136183e-05,
"loss": 0.2896,
"step": 141
},
{
"epoch": 2.713375796178344,
"grad_norm": 0.5100185255417189,
"learning_rate": 1.5329033746173975e-05,
"loss": 0.2777,
"step": 142
},
{
"epoch": 2.732484076433121,
"grad_norm": 0.49480941860523336,
"learning_rate": 1.5247494242079024e-05,
"loss": 0.2862,
"step": 143
},
{
"epoch": 2.7515923566878984,
"grad_norm": 0.5116390897620283,
"learning_rate": 1.5165470395159314e-05,
"loss": 0.2824,
"step": 144
},
{
"epoch": 2.770700636942675,
"grad_norm": 0.46550848518221105,
"learning_rate": 1.5082969776201948e-05,
"loss": 0.2729,
"step": 145
},
{
"epoch": 2.789808917197452,
"grad_norm": 0.4914325516706035,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.2844,
"step": 146
},
{
"epoch": 2.8089171974522293,
"grad_norm": 0.4327218630755152,
"learning_rate": 1.4916568724649688e-05,
"loss": 0.2726,
"step": 147
},
{
"epoch": 2.8280254777070066,
"grad_norm": 0.49931128864471586,
"learning_rate": 1.483268365084351e-05,
"loss": 0.2885,
"step": 148
},
{
"epoch": 2.8471337579617835,
"grad_norm": 0.42462659634268274,
"learning_rate": 1.4748352521159492e-05,
"loss": 0.2813,
"step": 149
},
{
"epoch": 2.8662420382165603,
"grad_norm": 0.4782092808735657,
"learning_rate": 1.466358311934654e-05,
"loss": 0.2558,
"step": 150
},
{
"epoch": 2.8853503184713376,
"grad_norm": 0.4470053446466154,
"learning_rate": 1.4578383269606004e-05,
"loss": 0.2945,
"step": 151
},
{
"epoch": 2.904458598726115,
"grad_norm": 0.43751701108009355,
"learning_rate": 1.4492760835869504e-05,
"loss": 0.2778,
"step": 152
},
{
"epoch": 2.9235668789808917,
"grad_norm": 0.41191558026004366,
"learning_rate": 1.4406723721073088e-05,
"loss": 0.2616,
"step": 153
},
{
"epoch": 2.9426751592356686,
"grad_norm": 0.42475125593412477,
"learning_rate": 1.4320279866427798e-05,
"loss": 0.2762,
"step": 154
},
{
"epoch": 2.961783439490446,
"grad_norm": 0.38908877064673336,
"learning_rate": 1.4233437250686695e-05,
"loss": 0.2829,
"step": 155
},
{
"epoch": 2.980891719745223,
"grad_norm": 0.3895202209738042,
"learning_rate": 1.4146203889408418e-05,
"loss": 0.2721,
"step": 156
},
{
"epoch": 3.0,
"grad_norm": 0.38091884693076805,
"learning_rate": 1.4058587834217356e-05,
"loss": 0.282,
"step": 157
},
{
"epoch": 3.0191082802547773,
"grad_norm": 0.5370560673123628,
"learning_rate": 1.3970597172060482e-05,
"loss": 0.222,
"step": 158
},
{
"epoch": 3.038216560509554,
"grad_norm": 0.4604622536048861,
"learning_rate": 1.3882240024460928e-05,
"loss": 0.2157,
"step": 159
},
{
"epoch": 3.0573248407643314,
"grad_norm": 0.7519886728759317,
"learning_rate": 1.3793524546768358e-05,
"loss": 0.2192,
"step": 160
},
{
"epoch": 3.0764331210191083,
"grad_norm": 0.468609599880104,
"learning_rate": 1.3704458927406261e-05,
"loss": 0.2072,
"step": 161
},
{
"epoch": 3.0955414012738856,
"grad_norm": 0.6309065386782458,
"learning_rate": 1.3615051387116131e-05,
"loss": 0.208,
"step": 162
},
{
"epoch": 3.1146496815286624,
"grad_norm": 0.5042966529981862,
"learning_rate": 1.3525310178198707e-05,
"loss": 0.2073,
"step": 163
},
{
"epoch": 3.1337579617834397,
"grad_norm": 0.4466868927297319,
"learning_rate": 1.3435243583752294e-05,
"loss": 0.1997,
"step": 164
},
{
"epoch": 3.1528662420382165,
"grad_norm": 0.49741697756563574,
"learning_rate": 1.3344859916908206e-05,
"loss": 0.2155,
"step": 165
},
{
"epoch": 3.171974522292994,
"grad_norm": 0.43472511201057934,
"learning_rate": 1.325416752006351e-05,
"loss": 0.1915,
"step": 166
},
{
"epoch": 3.1910828025477707,
"grad_norm": 0.4136961215523553,
"learning_rate": 1.3163174764110985e-05,
"loss": 0.2017,
"step": 167
},
{
"epoch": 3.210191082802548,
"grad_norm": 0.4404527179714845,
"learning_rate": 1.3071890047666498e-05,
"loss": 0.2061,
"step": 168
},
{
"epoch": 3.229299363057325,
"grad_norm": 0.47117791652033,
"learning_rate": 1.2980321796293838e-05,
"loss": 0.1904,
"step": 169
},
{
"epoch": 3.248407643312102,
"grad_norm": 0.4321419948688652,
"learning_rate": 1.288847846172701e-05,
"loss": 0.1853,
"step": 170
},
{
"epoch": 3.267515923566879,
"grad_norm": 0.4387904369550531,
"learning_rate": 1.2796368521090143e-05,
"loss": 0.1885,
"step": 171
},
{
"epoch": 3.286624203821656,
"grad_norm": 0.4653426006236831,
"learning_rate": 1.2704000476115079e-05,
"loss": 0.1992,
"step": 172
},
{
"epoch": 3.305732484076433,
"grad_norm": 0.42756041671139555,
"learning_rate": 1.2611382852356632e-05,
"loss": 0.1911,
"step": 173
},
{
"epoch": 3.3248407643312103,
"grad_norm": 0.45948983790837367,
"learning_rate": 1.2518524198405699e-05,
"loss": 0.2257,
"step": 174
},
{
"epoch": 3.343949044585987,
"grad_norm": 0.4364016264594612,
"learning_rate": 1.2425433085100224e-05,
"loss": 0.1903,
"step": 175
},
{
"epoch": 3.3630573248407645,
"grad_norm": 0.4365606694023847,
"learning_rate": 1.233211810473411e-05,
"loss": 0.2292,
"step": 176
},
{
"epoch": 3.3821656050955413,
"grad_norm": 0.43593421746602484,
"learning_rate": 1.2238587870264152e-05,
"loss": 0.2094,
"step": 177
},
{
"epoch": 3.4012738853503186,
"grad_norm": 0.4151965807893547,
"learning_rate": 1.2144851014515055e-05,
"loss": 0.2023,
"step": 178
},
{
"epoch": 3.4203821656050954,
"grad_norm": 0.4030152435778716,
"learning_rate": 1.2050916189382646e-05,
"loss": 0.2208,
"step": 179
},
{
"epoch": 3.4394904458598727,
"grad_norm": 0.4306112191522616,
"learning_rate": 1.1956792065035281e-05,
"loss": 0.2016,
"step": 180
},
{
"epoch": 3.4585987261146496,
"grad_norm": 0.4099650888297824,
"learning_rate": 1.1862487329113606e-05,
"loss": 0.2111,
"step": 181
},
{
"epoch": 3.477707006369427,
"grad_norm": 0.3907743339766982,
"learning_rate": 1.1768010685928686e-05,
"loss": 0.2151,
"step": 182
},
{
"epoch": 3.4968152866242037,
"grad_norm": 0.43020409607211063,
"learning_rate": 1.1673370855658592e-05,
"loss": 0.1998,
"step": 183
},
{
"epoch": 3.515923566878981,
"grad_norm": 0.4145002027117699,
"learning_rate": 1.1578576573543541e-05,
"loss": 0.204,
"step": 184
},
{
"epoch": 3.535031847133758,
"grad_norm": 0.39042318392064207,
"learning_rate": 1.1483636589079627e-05,
"loss": 0.2112,
"step": 185
},
{
"epoch": 3.554140127388535,
"grad_norm": 0.4075119844028756,
"learning_rate": 1.1388559665211241e-05,
"loss": 0.2126,
"step": 186
},
{
"epoch": 3.573248407643312,
"grad_norm": 0.3694705875362188,
"learning_rate": 1.1293354577522264e-05,
"loss": 0.2285,
"step": 187
},
{
"epoch": 3.5923566878980893,
"grad_norm": 0.43290257069774696,
"learning_rate": 1.1198030113426074e-05,
"loss": 0.2254,
"step": 188
},
{
"epoch": 3.611464968152866,
"grad_norm": 0.3950725832153689,
"learning_rate": 1.1102595071354471e-05,
"loss": 0.2183,
"step": 189
},
{
"epoch": 3.6305732484076434,
"grad_norm": 0.38704211113320974,
"learning_rate": 1.1007058259945584e-05,
"loss": 0.2063,
"step": 190
},
{
"epoch": 3.6496815286624202,
"grad_norm": 0.38273949780301664,
"learning_rate": 1.0911428497230834e-05,
"loss": 0.182,
"step": 191
},
{
"epoch": 3.6687898089171975,
"grad_norm": 0.3779345899671541,
"learning_rate": 1.0815714609821027e-05,
"loss": 0.2138,
"step": 192
},
{
"epoch": 3.6878980891719744,
"grad_norm": 0.37294722391486634,
"learning_rate": 1.0719925432091671e-05,
"loss": 0.1806,
"step": 193
},
{
"epoch": 3.7070063694267517,
"grad_norm": 0.40144804061495,
"learning_rate": 1.0624069805367558e-05,
"loss": 0.2042,
"step": 194
},
{
"epoch": 3.7261146496815285,
"grad_norm": 0.39956890984693877,
"learning_rate": 1.0528156577106703e-05,
"loss": 0.1855,
"step": 195
},
{
"epoch": 3.745222929936306,
"grad_norm": 0.39564241051576293,
"learning_rate": 1.043219460008374e-05,
"loss": 0.1959,
"step": 196
},
{
"epoch": 3.7643312101910826,
"grad_norm": 0.39044664569066817,
"learning_rate": 1.0336192731572805e-05,
"loss": 0.215,
"step": 197
},
{
"epoch": 3.78343949044586,
"grad_norm": 0.4017135641708565,
"learning_rate": 1.0240159832530007e-05,
"loss": 0.2121,
"step": 198
},
{
"epoch": 3.802547770700637,
"grad_norm": 0.41839004869218027,
"learning_rate": 1.0144104766775574e-05,
"loss": 0.1952,
"step": 199
},
{
"epoch": 3.821656050955414,
"grad_norm": 0.3974790127466975,
"learning_rate": 1.004803640017571e-05,
"loss": 0.1757,
"step": 200
},
{
"epoch": 3.840764331210191,
"grad_norm": 0.3681106324213222,
"learning_rate": 9.951963599824294e-06,
"loss": 0.1904,
"step": 201
},
{
"epoch": 3.859872611464968,
"grad_norm": 0.37757424131580586,
"learning_rate": 9.855895233224431e-06,
"loss": 0.2227,
"step": 202
},
{
"epoch": 3.8789808917197455,
"grad_norm": 0.39922116671908653,
"learning_rate": 9.759840167469995e-06,
"loss": 0.2154,
"step": 203
},
{
"epoch": 3.8980891719745223,
"grad_norm": 0.38552047164802783,
"learning_rate": 9.663807268427197e-06,
"loss": 0.1957,
"step": 204
},
{
"epoch": 3.917197452229299,
"grad_norm": 0.39636309115859425,
"learning_rate": 9.56780539991626e-06,
"loss": 0.2055,
"step": 205
},
{
"epoch": 3.9363057324840764,
"grad_norm": 0.4225597306351739,
"learning_rate": 9.471843422893299e-06,
"loss": 0.1971,
"step": 206
},
{
"epoch": 3.9554140127388537,
"grad_norm": 0.37443235054357454,
"learning_rate": 9.375930194632447e-06,
"loss": 0.2067,
"step": 207
},
{
"epoch": 3.9745222929936306,
"grad_norm": 0.4123012604836475,
"learning_rate": 9.28007456790833e-06,
"loss": 0.2152,
"step": 208
},
{
"epoch": 3.9936305732484074,
"grad_norm": 0.35798672433726103,
"learning_rate": 9.184285390178978e-06,
"loss": 0.1916,
"step": 209
},
{
"epoch": 4.012738853503185,
"grad_norm": 0.5089269182534356,
"learning_rate": 9.08857150276917e-06,
"loss": 0.1594,
"step": 210
},
{
"epoch": 4.031847133757962,
"grad_norm": 0.38640431577476686,
"learning_rate": 8.992941740054418e-06,
"loss": 0.1514,
"step": 211
},
{
"epoch": 4.050955414012739,
"grad_norm": 0.4366038511197679,
"learning_rate": 8.897404928645529e-06,
"loss": 0.152,
"step": 212
},
{
"epoch": 4.070063694267516,
"grad_norm": 0.6412545546535389,
"learning_rate": 8.80196988657393e-06,
"loss": 0.1779,
"step": 213
},
{
"epoch": 4.089171974522293,
"grad_norm": 0.39134537733029795,
"learning_rate": 8.706645422477739e-06,
"loss": 0.1705,
"step": 214
},
{
"epoch": 4.10828025477707,
"grad_norm": 0.44888777460322055,
"learning_rate": 8.611440334788762e-06,
"loss": 0.1766,
"step": 215
},
{
"epoch": 4.127388535031847,
"grad_norm": 0.4780850065483028,
"learning_rate": 8.516363410920376e-06,
"loss": 0.1355,
"step": 216
},
{
"epoch": 4.146496815286624,
"grad_norm": 0.43542927863174596,
"learning_rate": 8.42142342645646e-06,
"loss": 0.1631,
"step": 217
},
{
"epoch": 4.165605095541402,
"grad_norm": 0.40456188607567944,
"learning_rate": 8.326629144341408e-06,
"loss": 0.1489,
"step": 218
},
{
"epoch": 4.1847133757961785,
"grad_norm": 0.4605743384702378,
"learning_rate": 8.231989314071318e-06,
"loss": 0.1341,
"step": 219
},
{
"epoch": 4.203821656050955,
"grad_norm": 0.4478751556499832,
"learning_rate": 8.137512670886397e-06,
"loss": 0.1596,
"step": 220
},
{
"epoch": 4.222929936305732,
"grad_norm": 0.3530285323454552,
"learning_rate": 8.043207934964722e-06,
"loss": 0.1564,
"step": 221
},
{
"epoch": 4.24203821656051,
"grad_norm": 0.39569797878932367,
"learning_rate": 7.949083810617358e-06,
"loss": 0.1685,
"step": 222
},
{
"epoch": 4.261146496815287,
"grad_norm": 0.40099025584324766,
"learning_rate": 7.855148985484946e-06,
"loss": 0.1352,
"step": 223
},
{
"epoch": 4.280254777070064,
"grad_norm": 0.3690317074202254,
"learning_rate": 7.761412129735853e-06,
"loss": 0.1567,
"step": 224
},
{
"epoch": 4.2993630573248405,
"grad_norm": 0.3796001608092216,
"learning_rate": 7.667881895265895e-06,
"loss": 0.1354,
"step": 225
},
{
"epoch": 4.318471337579618,
"grad_norm": 0.3924929007417576,
"learning_rate": 7.574566914899779e-06,
"loss": 0.1432,
"step": 226
},
{
"epoch": 4.337579617834395,
"grad_norm": 0.3752952191912711,
"learning_rate": 7.481475801594302e-06,
"loss": 0.1518,
"step": 227
},
{
"epoch": 4.356687898089172,
"grad_norm": 0.34942276392410576,
"learning_rate": 7.388617147643371e-06,
"loss": 0.1455,
"step": 228
},
{
"epoch": 4.375796178343949,
"grad_norm": 0.3628175480795826,
"learning_rate": 7.295999523884921e-06,
"loss": 0.1487,
"step": 229
},
{
"epoch": 4.3949044585987265,
"grad_norm": 0.36433060410729295,
"learning_rate": 7.203631478909857e-06,
"loss": 0.1308,
"step": 230
},
{
"epoch": 4.414012738853503,
"grad_norm": 0.3531548535067812,
"learning_rate": 7.111521538272997e-06,
"loss": 0.1664,
"step": 231
},
{
"epoch": 4.43312101910828,
"grad_norm": 0.342738436892091,
"learning_rate": 7.019678203706164e-06,
"loss": 0.153,
"step": 232
},
{
"epoch": 4.452229299363057,
"grad_norm": 0.3902682078320996,
"learning_rate": 6.928109952333506e-06,
"loss": 0.1473,
"step": 233
},
{
"epoch": 4.471337579617835,
"grad_norm": 0.3649331356666892,
"learning_rate": 6.83682523588902e-06,
"loss": 0.1348,
"step": 234
},
{
"epoch": 4.490445859872612,
"grad_norm": 0.3540668474902925,
"learning_rate": 6.745832479936492e-06,
"loss": 0.1271,
"step": 235
},
{
"epoch": 4.509554140127388,
"grad_norm": 0.34532452896118054,
"learning_rate": 6.655140083091794e-06,
"loss": 0.134,
"step": 236
},
{
"epoch": 4.528662420382165,
"grad_norm": 0.37276122067279593,
"learning_rate": 6.564756416247712e-06,
"loss": 0.1415,
"step": 237
},
{
"epoch": 4.547770700636943,
"grad_norm": 0.36931433740955977,
"learning_rate": 6.474689821801295e-06,
"loss": 0.1333,
"step": 238
},
{
"epoch": 4.56687898089172,
"grad_norm": 0.3620981835063491,
"learning_rate": 6.384948612883872e-06,
"loss": 0.1607,
"step": 239
},
{
"epoch": 4.585987261146497,
"grad_norm": 0.36069482327928065,
"learning_rate": 6.2955410725937405e-06,
"loss": 0.142,
"step": 240
},
{
"epoch": 4.6050955414012735,
"grad_norm": 0.34850337967047446,
"learning_rate": 6.206475453231644e-06,
"loss": 0.1469,
"step": 241
},
{
"epoch": 4.624203821656051,
"grad_norm": 0.32486884027373397,
"learning_rate": 6.117759975539075e-06,
"loss": 0.1387,
"step": 242
},
{
"epoch": 4.643312101910828,
"grad_norm": 0.33958399302988723,
"learning_rate": 6.029402827939519e-06,
"loss": 0.136,
"step": 243
},
{
"epoch": 4.662420382165605,
"grad_norm": 0.36710935618687734,
"learning_rate": 5.941412165782645e-06,
"loss": 0.1411,
"step": 244
},
{
"epoch": 4.681528662420382,
"grad_norm": 0.3379620671461664,
"learning_rate": 5.853796110591583e-06,
"loss": 0.1418,
"step": 245
},
{
"epoch": 4.7006369426751595,
"grad_norm": 0.3303258074589593,
"learning_rate": 5.766562749313309e-06,
"loss": 0.1625,
"step": 246
},
{
"epoch": 4.719745222929936,
"grad_norm": 0.36912220390558065,
"learning_rate": 5.6797201335722064e-06,
"loss": 0.1337,
"step": 247
},
{
"epoch": 4.738853503184713,
"grad_norm": 0.34728625863844315,
"learning_rate": 5.593276278926912e-06,
"loss": 0.1304,
"step": 248
},
{
"epoch": 4.757961783439491,
"grad_norm": 0.3571181460211357,
"learning_rate": 5.507239164130501e-06,
"loss": 0.1429,
"step": 249
},
{
"epoch": 4.777070063694268,
"grad_norm": 0.3274899215773813,
"learning_rate": 5.421616730394e-06,
"loss": 0.1389,
"step": 250
},
{
"epoch": 4.796178343949045,
"grad_norm": 0.33880400939000826,
"learning_rate": 5.336416880653461e-06,
"loss": 0.1415,
"step": 251
},
{
"epoch": 4.8152866242038215,
"grad_norm": 0.34485393087606486,
"learning_rate": 5.251647478840511e-06,
"loss": 0.1469,
"step": 252
},
{
"epoch": 4.834394904458598,
"grad_norm": 0.36447860390850273,
"learning_rate": 5.167316349156495e-06,
"loss": 0.1295,
"step": 253
},
{
"epoch": 4.853503184713376,
"grad_norm": 0.33040578233991347,
"learning_rate": 5.083431275350312e-06,
"loss": 0.1669,
"step": 254
},
{
"epoch": 4.872611464968153,
"grad_norm": 0.34643695414677406,
"learning_rate": 5.000000000000003e-06,
"loss": 0.1427,
"step": 255
},
{
"epoch": 4.89171974522293,
"grad_norm": 0.34008836835894396,
"learning_rate": 4.917030223798057e-06,
"loss": 0.1409,
"step": 256
},
{
"epoch": 4.9108280254777075,
"grad_norm": 0.3307392075924563,
"learning_rate": 4.834529604840686e-06,
"loss": 0.1334,
"step": 257
},
{
"epoch": 4.929936305732484,
"grad_norm": 0.33463545632612857,
"learning_rate": 4.7525057579209775e-06,
"loss": 0.1327,
"step": 258
},
{
"epoch": 4.949044585987261,
"grad_norm": 0.33818547242322916,
"learning_rate": 4.670966253826027e-06,
"loss": 0.1509,
"step": 259
},
{
"epoch": 4.968152866242038,
"grad_norm": 0.3418854248518739,
"learning_rate": 4.589918618638173e-06,
"loss": 0.1567,
"step": 260
},
{
"epoch": 4.987261146496815,
"grad_norm": 0.3426014792444678,
"learning_rate": 4.5093703330403385e-06,
"loss": 0.1379,
"step": 261
},
{
"epoch": 5.006369426751593,
"grad_norm": 0.3665647363760616,
"learning_rate": 4.429328831625565e-06,
"loss": 0.1262,
"step": 262
},
{
"epoch": 5.025477707006369,
"grad_norm": 0.4397089507549027,
"learning_rate": 4.349801502210801e-06,
"loss": 0.1224,
"step": 263
},
{
"epoch": 5.044585987261146,
"grad_norm": 0.28862486095001927,
"learning_rate": 4.270795685155001e-06,
"loss": 0.1307,
"step": 264
},
{
"epoch": 5.063694267515924,
"grad_norm": 0.30544658345336123,
"learning_rate": 4.192318672681631e-06,
"loss": 0.1266,
"step": 265
},
{
"epoch": 5.082802547770701,
"grad_norm": 0.43383345818949426,
"learning_rate": 4.1143777082055715e-06,
"loss": 0.1213,
"step": 266
},
{
"epoch": 5.101910828025478,
"grad_norm": 0.4046391402055215,
"learning_rate": 4.036979985664566e-06,
"loss": 0.104,
"step": 267
},
{
"epoch": 5.1210191082802545,
"grad_norm": 0.3855607546334183,
"learning_rate": 3.960132648855226e-06,
"loss": 0.1099,
"step": 268
},
{
"epoch": 5.140127388535032,
"grad_norm": 0.29227874906516005,
"learning_rate": 3.883842790773647e-06,
"loss": 0.117,
"step": 269
},
{
"epoch": 5.159235668789809,
"grad_norm": 0.33909673264219736,
"learning_rate": 3.8081174529607346e-06,
"loss": 0.1032,
"step": 270
},
{
"epoch": 5.178343949044586,
"grad_norm": 0.34277940217869574,
"learning_rate": 3.732963624852275e-06,
"loss": 0.1184,
"step": 271
},
{
"epoch": 5.197452229299363,
"grad_norm": 0.35200594923051187,
"learning_rate": 3.6583882431338047e-06,
"loss": 0.117,
"step": 272
},
{
"epoch": 5.2165605095541405,
"grad_norm": 0.2867865548295543,
"learning_rate": 3.584398191100341e-06,
"loss": 0.1053,
"step": 273
},
{
"epoch": 5.235668789808917,
"grad_norm": 0.29023273904345026,
"learning_rate": 3.511000298021098e-06,
"loss": 0.095,
"step": 274
},
{
"epoch": 5.254777070063694,
"grad_norm": 0.31535284793689156,
"learning_rate": 3.4382013385090985e-06,
"loss": 0.0964,
"step": 275
},
{
"epoch": 5.273885350318471,
"grad_norm": 0.32775581968342726,
"learning_rate": 3.3660080318959043e-06,
"loss": 0.1052,
"step": 276
},
{
"epoch": 5.292993630573249,
"grad_norm": 0.3056482156990124,
"learning_rate": 3.2944270416114256e-06,
"loss": 0.098,
"step": 277
},
{
"epoch": 5.312101910828026,
"grad_norm": 0.3333608740856702,
"learning_rate": 3.223464974568874e-06,
"loss": 0.1009,
"step": 278
},
{
"epoch": 5.3312101910828025,
"grad_norm": 0.3040604842929987,
"learning_rate": 3.153128380554941e-06,
"loss": 0.1214,
"step": 279
},
{
"epoch": 5.350318471337579,
"grad_norm": 0.31800720642326713,
"learning_rate": 3.0834237516252817e-06,
"loss": 0.1013,
"step": 280
},
{
"epoch": 5.369426751592357,
"grad_norm": 0.29474795325853187,
"learning_rate": 3.0143575215052732e-06,
"loss": 0.1235,
"step": 281
},
{
"epoch": 5.388535031847134,
"grad_norm": 0.28282687070971513,
"learning_rate": 2.94593606499619e-06,
"loss": 0.0827,
"step": 282
},
{
"epoch": 5.407643312101911,
"grad_norm": 0.29409620900166117,
"learning_rate": 2.878165697386812e-06,
"loss": 0.0953,
"step": 283
},
{
"epoch": 5.426751592356688,
"grad_norm": 0.2784092576911946,
"learning_rate": 2.8110526738705345e-06,
"loss": 0.0878,
"step": 284
},
{
"epoch": 5.445859872611465,
"grad_norm": 0.29514758395130525,
"learning_rate": 2.7446031889679893e-06,
"loss": 0.0917,
"step": 285
},
{
"epoch": 5.464968152866242,
"grad_norm": 0.3143110523026061,
"learning_rate": 2.678823375955314e-06,
"loss": 0.1167,
"step": 286
},
{
"epoch": 5.484076433121019,
"grad_norm": 0.315694056109303,
"learning_rate": 2.6137193062980506e-06,
"loss": 0.0945,
"step": 287
},
{
"epoch": 5.503184713375796,
"grad_norm": 0.3261544809692315,
"learning_rate": 2.5492969890907383e-06,
"loss": 0.1339,
"step": 288
},
{
"epoch": 5.522292993630574,
"grad_norm": 0.3097820809684143,
"learning_rate": 2.485562370502279e-06,
"loss": 0.1137,
"step": 289
},
{
"epoch": 5.54140127388535,
"grad_norm": 0.29291564117318547,
"learning_rate": 2.4225213332271203e-06,
"loss": 0.118,
"step": 290
},
{
"epoch": 5.560509554140127,
"grad_norm": 0.28959010531691,
"learning_rate": 2.3601796959422585e-06,
"loss": 0.0979,
"step": 291
},
{
"epoch": 5.579617834394904,
"grad_norm": 0.27726943288956224,
"learning_rate": 2.2985432127701945e-06,
"loss": 0.1123,
"step": 292
},
{
"epoch": 5.598726114649682,
"grad_norm": 0.286947092747882,
"learning_rate": 2.2376175727478346e-06,
"loss": 0.109,
"step": 293
},
{
"epoch": 5.617834394904459,
"grad_norm": 0.27206755359891305,
"learning_rate": 2.1774083993013715e-06,
"loss": 0.1085,
"step": 294
},
{
"epoch": 5.6369426751592355,
"grad_norm": 0.2863090639175942,
"learning_rate": 2.1179212497272582e-06,
"loss": 0.102,
"step": 295
},
{
"epoch": 5.656050955414012,
"grad_norm": 0.2950645368927506,
"learning_rate": 2.0591616146792705e-06,
"loss": 0.1244,
"step": 296
},
{
"epoch": 5.67515923566879,
"grad_norm": 0.28340986980994437,
"learning_rate": 2.0011349176617133e-06,
"loss": 0.1257,
"step": 297
},
{
"epoch": 5.694267515923567,
"grad_norm": 0.28462815485654264,
"learning_rate": 1.9438465145288377e-06,
"loss": 0.1455,
"step": 298
},
{
"epoch": 5.713375796178344,
"grad_norm": 0.27472334188632586,
"learning_rate": 1.8873016929904942e-06,
"loss": 0.1035,
"step": 299
},
{
"epoch": 5.732484076433121,
"grad_norm": 0.26476697429227475,
"learning_rate": 1.8315056721240831e-06,
"loss": 0.1136,
"step": 300
},
{
"epoch": 5.751592356687898,
"grad_norm": 0.26455211518462884,
"learning_rate": 1.7764636018928249e-06,
"loss": 0.0915,
"step": 301
},
{
"epoch": 5.770700636942675,
"grad_norm": 0.2716186264959562,
"learning_rate": 1.722180562670428e-06,
"loss": 0.1222,
"step": 302
},
{
"epoch": 5.789808917197452,
"grad_norm": 0.27953494999772105,
"learning_rate": 1.6686615647721638e-06,
"loss": 0.0971,
"step": 303
},
{
"epoch": 5.80891719745223,
"grad_norm": 0.2756093225078449,
"learning_rate": 1.6159115479924259e-06,
"loss": 0.1031,
"step": 304
},
{
"epoch": 5.828025477707007,
"grad_norm": 0.2832117737285963,
"learning_rate": 1.5639353811487744e-06,
"loss": 0.1139,
"step": 305
},
{
"epoch": 5.8471337579617835,
"grad_norm": 0.3052706448026399,
"learning_rate": 1.5127378616325606e-06,
"loss": 0.113,
"step": 306
},
{
"epoch": 5.86624203821656,
"grad_norm": 0.2999140385337404,
"learning_rate": 1.462323714966114e-06,
"loss": 0.1177,
"step": 307
},
{
"epoch": 5.885350318471337,
"grad_norm": 0.2604054726344459,
"learning_rate": 1.4126975943665844e-06,
"loss": 0.0987,
"step": 308
},
{
"epoch": 5.904458598726115,
"grad_norm": 0.29275468674804284,
"learning_rate": 1.3638640803164516e-06,
"loss": 0.1301,
"step": 309
},
{
"epoch": 5.923566878980892,
"grad_norm": 0.29057510203503306,
"learning_rate": 1.3158276801407432e-06,
"loss": 0.1113,
"step": 310
},
{
"epoch": 5.942675159235669,
"grad_norm": 0.2823117671362716,
"learning_rate": 1.2685928275910142e-06,
"loss": 0.0923,
"step": 311
},
{
"epoch": 5.961783439490446,
"grad_norm": 0.28815918803199797,
"learning_rate": 1.222163882436107e-06,
"loss": 0.0924,
"step": 312
},
{
"epoch": 5.980891719745223,
"grad_norm": 0.28371249325519327,
"learning_rate": 1.1765451300597574e-06,
"loss": 0.104,
"step": 313
},
{
"epoch": 6.0,
"grad_norm": 0.2915984533087662,
"learning_rate": 1.1317407810650372e-06,
"loss": 0.117,
"step": 314
},
{
"epoch": 6.019108280254777,
"grad_norm": 0.33831219981090144,
"learning_rate": 1.0877549708857228e-06,
"loss": 0.093,
"step": 315
},
{
"epoch": 6.038216560509555,
"grad_norm": 0.3216688552275663,
"learning_rate": 1.0445917594046073e-06,
"loss": 0.1018,
"step": 316
},
{
"epoch": 6.057324840764331,
"grad_norm": 0.2686281332031815,
"learning_rate": 1.0022551305787564e-06,
"loss": 0.0933,
"step": 317
},
{
"epoch": 6.076433121019108,
"grad_norm": 0.24771264303054355,
"learning_rate": 9.607489920717983e-07,
"loss": 0.101,
"step": 318
},
{
"epoch": 6.095541401273885,
"grad_norm": 0.21708682495154552,
"learning_rate": 9.200771748932513e-07,
"loss": 0.0907,
"step": 319
},
{
"epoch": 6.114649681528663,
"grad_norm": 0.22569521424515723,
"learning_rate": 8.802434330449128e-07,
"loss": 0.08,
"step": 320
},
{
"epoch": 6.13375796178344,
"grad_norm": 0.24475648361426353,
"learning_rate": 8.412514431743657e-07,
"loss": 0.0937,
"step": 321
},
{
"epoch": 6.1528662420382165,
"grad_norm": 0.2691118169652071,
"learning_rate": 8.031048042356393e-07,
"loss": 0.085,
"step": 322
},
{
"epoch": 6.171974522292993,
"grad_norm": 0.29927997351893876,
"learning_rate": 7.65807037157007e-07,
"loss": 0.1088,
"step": 323
},
{
"epoch": 6.191082802547771,
"grad_norm": 0.2726091543278292,
"learning_rate": 7.293615845160196e-07,
"loss": 0.0925,
"step": 324
},
{
"epoch": 6.210191082802548,
"grad_norm": 0.26930688658991936,
"learning_rate": 6.937718102217461e-07,
"loss": 0.0988,
"step": 325
},
{
"epoch": 6.229299363057325,
"grad_norm": 0.2643257486986743,
"learning_rate": 6.590409992042957e-07,
"loss": 0.0998,
"step": 326
},
{
"epoch": 6.248407643312102,
"grad_norm": 0.24443302467612754,
"learning_rate": 6.251723571116031e-07,
"loss": 0.0958,
"step": 327
},
{
"epoch": 6.267515923566879,
"grad_norm": 0.23090275620250872,
"learning_rate": 5.921690100135713e-07,
"loss": 0.1111,
"step": 328
},
{
"epoch": 6.286624203821656,
"grad_norm": 0.22366775503115832,
"learning_rate": 5.600340041135133e-07,
"loss": 0.0822,
"step": 329
},
{
"epoch": 6.305732484076433,
"grad_norm": 0.22095216357534475,
"learning_rate": 5.287703054670012e-07,
"loss": 0.0808,
"step": 330
},
{
"epoch": 6.32484076433121,
"grad_norm": 0.23474442436705795,
"learning_rate": 4.983807997080925e-07,
"loss": 0.0995,
"step": 331
},
{
"epoch": 6.343949044585988,
"grad_norm": 0.24620687930140603,
"learning_rate": 4.6886829178299676e-07,
"loss": 0.109,
"step": 332
},
{
"epoch": 6.3630573248407645,
"grad_norm": 0.2282273470889745,
"learning_rate": 4.402355056911656e-07,
"loss": 0.0975,
"step": 333
},
{
"epoch": 6.382165605095541,
"grad_norm": 0.24130930441355783,
"learning_rate": 4.124850842338779e-07,
"loss": 0.1049,
"step": 334
},
{
"epoch": 6.401273885350318,
"grad_norm": 0.2369972272750921,
"learning_rate": 3.8561958877030957e-07,
"loss": 0.0836,
"step": 335
},
{
"epoch": 6.420382165605096,
"grad_norm": 0.23437243777443048,
"learning_rate": 3.5964149898111587e-07,
"loss": 0.0869,
"step": 336
},
{
"epoch": 6.439490445859873,
"grad_norm": 0.23702374263733794,
"learning_rate": 3.345532126395579e-07,
"loss": 0.0944,
"step": 337
},
{
"epoch": 6.45859872611465,
"grad_norm": 0.24397448805984406,
"learning_rate": 3.1035704539019384e-07,
"loss": 0.0928,
"step": 338
},
{
"epoch": 6.477707006369426,
"grad_norm": 0.22333004219316266,
"learning_rate": 2.870552305351382e-07,
"loss": 0.086,
"step": 339
},
{
"epoch": 6.496815286624204,
"grad_norm": 0.24155868621147708,
"learning_rate": 2.646499188279328e-07,
"loss": 0.0979,
"step": 340
},
{
"epoch": 6.515923566878981,
"grad_norm": 0.22675346332252252,
"learning_rate": 2.4314317827503375e-07,
"loss": 0.0959,
"step": 341
},
{
"epoch": 6.535031847133758,
"grad_norm": 0.2241863416494601,
"learning_rate": 2.2253699394493066e-07,
"loss": 0.0928,
"step": 342
},
{
"epoch": 6.554140127388535,
"grad_norm": 0.22498106512620902,
"learning_rate": 2.028332677849254e-07,
"loss": 0.0761,
"step": 343
},
{
"epoch": 6.573248407643312,
"grad_norm": 0.23146588879507815,
"learning_rate": 1.840338184455881e-07,
"loss": 0.1002,
"step": 344
},
{
"epoch": 6.592356687898089,
"grad_norm": 0.2366378457488816,
"learning_rate": 1.6614038111289034e-07,
"loss": 0.1007,
"step": 345
},
{
"epoch": 6.611464968152866,
"grad_norm": 0.2339299622295652,
"learning_rate": 1.49154607348051e-07,
"loss": 0.087,
"step": 346
},
{
"epoch": 6.630573248407643,
"grad_norm": 0.23232639138795694,
"learning_rate": 1.330780649350938e-07,
"loss": 0.0914,
"step": 347
},
{
"epoch": 6.649681528662421,
"grad_norm": 0.23544385192540226,
"learning_rate": 1.1791223773614635e-07,
"loss": 0.0868,
"step": 348
},
{
"epoch": 6.6687898089171975,
"grad_norm": 0.2364879774194191,
"learning_rate": 1.0365852555447642e-07,
"loss": 0.1041,
"step": 349
},
{
"epoch": 6.687898089171974,
"grad_norm": 0.23715728901897062,
"learning_rate": 9.031824400528854e-08,
"loss": 0.0956,
"step": 350
},
{
"epoch": 6.707006369426751,
"grad_norm": 0.24227616721179004,
"learning_rate": 7.789262439430012e-08,
"loss": 0.0927,
"step": 351
},
{
"epoch": 6.726114649681529,
"grad_norm": 0.22929420696904648,
"learning_rate": 6.638281360408339e-08,
"loss": 0.1037,
"step": 352
},
{
"epoch": 6.745222929936306,
"grad_norm": 0.2258254277459285,
"learning_rate": 5.578987398821345e-08,
"loss": 0.0873,
"step": 353
},
{
"epoch": 6.764331210191083,
"grad_norm": 0.23883346214149054,
"learning_rate": 4.6114783273213395e-08,
"loss": 0.0879,
"step": 354
},
{
"epoch": 6.7834394904458595,
"grad_norm": 0.25654835378278024,
"learning_rate": 3.735843446830867e-08,
"loss": 0.1099,
"step": 355
},
{
"epoch": 6.802547770700637,
"grad_norm": 0.2212532184458245,
"learning_rate": 2.9521635783001932e-08,
"loss": 0.081,
"step": 356
},
{
"epoch": 6.821656050955414,
"grad_norm": 0.22913700957627273,
"learning_rate": 2.2605110552477162e-08,
"loss": 0.1044,
"step": 357
},
{
"epoch": 6.840764331210191,
"grad_norm": 0.23052894110297553,
"learning_rate": 1.6609497170834154e-08,
"loss": 0.088,
"step": 358
},
{
"epoch": 6.859872611464969,
"grad_norm": 0.2193065992766503,
"learning_rate": 1.1535349032167908e-08,
"loss": 0.0921,
"step": 359
},
{
"epoch": 6.8789808917197455,
"grad_norm": 0.2313614102842318,
"learning_rate": 7.3831344794872415e-09,
"loss": 0.1063,
"step": 360
},
{
"epoch": 6.898089171974522,
"grad_norm": 0.2326220261825367,
"learning_rate": 4.153236761488266e-09,
"loss": 0.1038,
"step": 361
},
{
"epoch": 6.917197452229299,
"grad_norm": 0.22680244676310998,
"learning_rate": 1.8459539971804608e-09,
"loss": 0.0997,
"step": 362
},
{
"epoch": 6.936305732484076,
"grad_norm": 0.23789154446440378,
"learning_rate": 4.614991483686826e-10,
"loss": 0.0928,
"step": 363
},
{
"epoch": 6.955414012738854,
"grad_norm": 0.22259185828565997,
"learning_rate": 0.0,
"loss": 0.0835,
"step": 364
},
{
"epoch": 6.955414012738854,
"step": 364,
"total_flos": 3.947982283988664e+17,
"train_loss": 0.0,
"train_runtime": 2.6671,
"train_samples_per_second": 13122.755,
"train_steps_per_second": 136.477
}
],
"logging_steps": 1,
"max_steps": 364,
"num_input_tokens_seen": 0,
"num_train_epochs": 7,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 3.947982283988664e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}