aleegis10's picture
Training in progress, step 200, checkpoint
4141c81 verified
{
"best_metric": 1.3995150327682495,
"best_model_checkpoint": "miner_id_24/checkpoint-200",
"epoch": 0.0658165364047717,
"eval_steps": 50,
"global_step": 200,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.00032908268202385847,
"grad_norm": 2.5831387042999268,
"learning_rate": 1e-05,
"loss": 5.3014,
"step": 1
},
{
"epoch": 0.00032908268202385847,
"eval_loss": 1.6017260551452637,
"eval_runtime": 108.1261,
"eval_samples_per_second": 47.334,
"eval_steps_per_second": 11.838,
"step": 1
},
{
"epoch": 0.0006581653640477169,
"grad_norm": 2.6134402751922607,
"learning_rate": 2e-05,
"loss": 5.0754,
"step": 2
},
{
"epoch": 0.0009872480460715754,
"grad_norm": 2.155811071395874,
"learning_rate": 3e-05,
"loss": 5.2135,
"step": 3
},
{
"epoch": 0.0013163307280954339,
"grad_norm": 2.289625883102417,
"learning_rate": 4e-05,
"loss": 5.3694,
"step": 4
},
{
"epoch": 0.0016454134101192926,
"grad_norm": 1.8600523471832275,
"learning_rate": 5e-05,
"loss": 5.0685,
"step": 5
},
{
"epoch": 0.001974496092143151,
"grad_norm": 1.8842722177505493,
"learning_rate": 6e-05,
"loss": 5.2425,
"step": 6
},
{
"epoch": 0.0023035787741670093,
"grad_norm": 2.1337532997131348,
"learning_rate": 7e-05,
"loss": 5.0973,
"step": 7
},
{
"epoch": 0.0026326614561908677,
"grad_norm": 2.2842800617218018,
"learning_rate": 8e-05,
"loss": 5.6026,
"step": 8
},
{
"epoch": 0.0029617441382147267,
"grad_norm": 2.154679775238037,
"learning_rate": 9e-05,
"loss": 5.5742,
"step": 9
},
{
"epoch": 0.003290826820238585,
"grad_norm": 2.090522289276123,
"learning_rate": 0.0001,
"loss": 5.3858,
"step": 10
},
{
"epoch": 0.0036199095022624436,
"grad_norm": 2.0786349773406982,
"learning_rate": 9.999316524962345e-05,
"loss": 5.5018,
"step": 11
},
{
"epoch": 0.003948992184286302,
"grad_norm": 2.3556385040283203,
"learning_rate": 9.997266286704631e-05,
"loss": 5.6,
"step": 12
},
{
"epoch": 0.0042780748663101605,
"grad_norm": 2.1240100860595703,
"learning_rate": 9.993849845741524e-05,
"loss": 5.3077,
"step": 13
},
{
"epoch": 0.0046071575483340186,
"grad_norm": 1.9147992134094238,
"learning_rate": 9.989068136093873e-05,
"loss": 5.2992,
"step": 14
},
{
"epoch": 0.0049362402303578775,
"grad_norm": 2.1758053302764893,
"learning_rate": 9.98292246503335e-05,
"loss": 5.617,
"step": 15
},
{
"epoch": 0.0052653229123817355,
"grad_norm": 2.277902841567993,
"learning_rate": 9.975414512725057e-05,
"loss": 5.5635,
"step": 16
},
{
"epoch": 0.005594405594405594,
"grad_norm": 2.2570464611053467,
"learning_rate": 9.966546331768191e-05,
"loss": 5.4873,
"step": 17
},
{
"epoch": 0.005923488276429453,
"grad_norm": 2.402843952178955,
"learning_rate": 9.956320346634876e-05,
"loss": 5.6702,
"step": 18
},
{
"epoch": 0.006252570958453311,
"grad_norm": 2.279578924179077,
"learning_rate": 9.944739353007344e-05,
"loss": 5.2949,
"step": 19
},
{
"epoch": 0.00658165364047717,
"grad_norm": 2.406120777130127,
"learning_rate": 9.931806517013612e-05,
"loss": 5.5832,
"step": 20
},
{
"epoch": 0.006910736322501028,
"grad_norm": 2.2947046756744385,
"learning_rate": 9.917525374361912e-05,
"loss": 5.5558,
"step": 21
},
{
"epoch": 0.007239819004524887,
"grad_norm": 2.15261173248291,
"learning_rate": 9.901899829374047e-05,
"loss": 5.2593,
"step": 22
},
{
"epoch": 0.007568901686548745,
"grad_norm": 2.397843599319458,
"learning_rate": 9.884934153917997e-05,
"loss": 5.792,
"step": 23
},
{
"epoch": 0.007897984368572603,
"grad_norm": 2.3961777687072754,
"learning_rate": 9.86663298624003e-05,
"loss": 5.6876,
"step": 24
},
{
"epoch": 0.008227067050596462,
"grad_norm": 2.3227005004882812,
"learning_rate": 9.847001329696653e-05,
"loss": 5.6791,
"step": 25
},
{
"epoch": 0.008556149732620321,
"grad_norm": 2.6047770977020264,
"learning_rate": 9.826044551386744e-05,
"loss": 5.9215,
"step": 26
},
{
"epoch": 0.00888523241464418,
"grad_norm": 2.5942320823669434,
"learning_rate": 9.803768380684242e-05,
"loss": 5.8868,
"step": 27
},
{
"epoch": 0.009214315096668037,
"grad_norm": 2.6597602367401123,
"learning_rate": 9.780178907671789e-05,
"loss": 5.9174,
"step": 28
},
{
"epoch": 0.009543397778691896,
"grad_norm": 2.781270980834961,
"learning_rate": 9.755282581475769e-05,
"loss": 5.7107,
"step": 29
},
{
"epoch": 0.009872480460715755,
"grad_norm": 2.9205641746520996,
"learning_rate": 9.729086208503174e-05,
"loss": 5.4605,
"step": 30
},
{
"epoch": 0.010201563142739614,
"grad_norm": 2.8525655269622803,
"learning_rate": 9.701596950580806e-05,
"loss": 5.9747,
"step": 31
},
{
"epoch": 0.010530645824763471,
"grad_norm": 3.2996468544006348,
"learning_rate": 9.672822322997305e-05,
"loss": 6.4142,
"step": 32
},
{
"epoch": 0.01085972850678733,
"grad_norm": 3.177000045776367,
"learning_rate": 9.642770192448536e-05,
"loss": 6.2026,
"step": 33
},
{
"epoch": 0.011188811188811189,
"grad_norm": 3.165365695953369,
"learning_rate": 9.611448774886924e-05,
"loss": 5.7878,
"step": 34
},
{
"epoch": 0.011517893870835048,
"grad_norm": 3.517965793609619,
"learning_rate": 9.578866633275288e-05,
"loss": 6.4767,
"step": 35
},
{
"epoch": 0.011846976552858907,
"grad_norm": 3.2690157890319824,
"learning_rate": 9.545032675245813e-05,
"loss": 5.7876,
"step": 36
},
{
"epoch": 0.012176059234882764,
"grad_norm": 3.745285987854004,
"learning_rate": 9.509956150664796e-05,
"loss": 6.3668,
"step": 37
},
{
"epoch": 0.012505141916906623,
"grad_norm": 3.6998376846313477,
"learning_rate": 9.473646649103818e-05,
"loss": 6.4457,
"step": 38
},
{
"epoch": 0.012834224598930482,
"grad_norm": 3.857137441635132,
"learning_rate": 9.43611409721806e-05,
"loss": 6.1547,
"step": 39
},
{
"epoch": 0.01316330728095434,
"grad_norm": 3.786033868789673,
"learning_rate": 9.397368756032445e-05,
"loss": 6.1211,
"step": 40
},
{
"epoch": 0.013492389962978198,
"grad_norm": 4.1755051612854,
"learning_rate": 9.357421218136386e-05,
"loss": 6.3914,
"step": 41
},
{
"epoch": 0.013821472645002057,
"grad_norm": 4.517828464508057,
"learning_rate": 9.316282404787871e-05,
"loss": 6.7313,
"step": 42
},
{
"epoch": 0.014150555327025915,
"grad_norm": 4.888702392578125,
"learning_rate": 9.273963562927695e-05,
"loss": 6.2315,
"step": 43
},
{
"epoch": 0.014479638009049774,
"grad_norm": 5.499008655548096,
"learning_rate": 9.230476262104677e-05,
"loss": 7.4386,
"step": 44
},
{
"epoch": 0.014808720691073632,
"grad_norm": 5.009950637817383,
"learning_rate": 9.185832391312644e-05,
"loss": 6.3468,
"step": 45
},
{
"epoch": 0.01513780337309749,
"grad_norm": 5.937233924865723,
"learning_rate": 9.140044155740101e-05,
"loss": 7.0928,
"step": 46
},
{
"epoch": 0.01546688605512135,
"grad_norm": 6.439074993133545,
"learning_rate": 9.093124073433463e-05,
"loss": 7.5133,
"step": 47
},
{
"epoch": 0.015795968737145206,
"grad_norm": 8.385202407836914,
"learning_rate": 9.045084971874738e-05,
"loss": 8.2098,
"step": 48
},
{
"epoch": 0.016125051419169065,
"grad_norm": 11.009381294250488,
"learning_rate": 8.995939984474624e-05,
"loss": 8.5627,
"step": 49
},
{
"epoch": 0.016454134101192924,
"grad_norm": 14.385894775390625,
"learning_rate": 8.945702546981969e-05,
"loss": 9.6938,
"step": 50
},
{
"epoch": 0.016454134101192924,
"eval_loss": 1.592988133430481,
"eval_runtime": 108.8174,
"eval_samples_per_second": 47.033,
"eval_steps_per_second": 11.763,
"step": 50
},
{
"epoch": 0.016783216783216783,
"grad_norm": 14.585724830627441,
"learning_rate": 8.894386393810563e-05,
"loss": 6.0914,
"step": 51
},
{
"epoch": 0.017112299465240642,
"grad_norm": 12.33920669555664,
"learning_rate": 8.842005554284296e-05,
"loss": 5.8927,
"step": 52
},
{
"epoch": 0.0174413821472645,
"grad_norm": 8.384288787841797,
"learning_rate": 8.788574348801675e-05,
"loss": 5.7106,
"step": 53
},
{
"epoch": 0.01777046482928836,
"grad_norm": 5.3903937339782715,
"learning_rate": 8.73410738492077e-05,
"loss": 5.7199,
"step": 54
},
{
"epoch": 0.01809954751131222,
"grad_norm": 3.546074628829956,
"learning_rate": 8.678619553365659e-05,
"loss": 5.2362,
"step": 55
},
{
"epoch": 0.018428630193336074,
"grad_norm": 2.585486888885498,
"learning_rate": 8.622126023955446e-05,
"loss": 5.4502,
"step": 56
},
{
"epoch": 0.018757712875359933,
"grad_norm": 2.2425572872161865,
"learning_rate": 8.564642241456986e-05,
"loss": 5.1769,
"step": 57
},
{
"epoch": 0.019086795557383792,
"grad_norm": 1.9347532987594604,
"learning_rate": 8.506183921362443e-05,
"loss": 5.1616,
"step": 58
},
{
"epoch": 0.01941587823940765,
"grad_norm": 1.8669167757034302,
"learning_rate": 8.44676704559283e-05,
"loss": 5.3023,
"step": 59
},
{
"epoch": 0.01974496092143151,
"grad_norm": 1.8059626817703247,
"learning_rate": 8.386407858128706e-05,
"loss": 5.2465,
"step": 60
},
{
"epoch": 0.02007404360345537,
"grad_norm": 1.601872205734253,
"learning_rate": 8.32512286056924e-05,
"loss": 5.3677,
"step": 61
},
{
"epoch": 0.020403126285479228,
"grad_norm": 1.5859284400939941,
"learning_rate": 8.262928807620843e-05,
"loss": 5.4263,
"step": 62
},
{
"epoch": 0.020732208967503087,
"grad_norm": 1.4872530698776245,
"learning_rate": 8.199842702516583e-05,
"loss": 5.0603,
"step": 63
},
{
"epoch": 0.021061291649526942,
"grad_norm": 1.5247877836227417,
"learning_rate": 8.135881792367686e-05,
"loss": 5.5638,
"step": 64
},
{
"epoch": 0.0213903743315508,
"grad_norm": 1.6158437728881836,
"learning_rate": 8.07106356344834e-05,
"loss": 5.4161,
"step": 65
},
{
"epoch": 0.02171945701357466,
"grad_norm": 1.5754039287567139,
"learning_rate": 8.005405736415126e-05,
"loss": 5.4194,
"step": 66
},
{
"epoch": 0.02204853969559852,
"grad_norm": 1.5747605562210083,
"learning_rate": 7.938926261462366e-05,
"loss": 5.1095,
"step": 67
},
{
"epoch": 0.022377622377622378,
"grad_norm": 1.6026984453201294,
"learning_rate": 7.871643313414718e-05,
"loss": 5.4065,
"step": 68
},
{
"epoch": 0.022706705059646237,
"grad_norm": 1.6130657196044922,
"learning_rate": 7.803575286758364e-05,
"loss": 5.4167,
"step": 69
},
{
"epoch": 0.023035787741670095,
"grad_norm": 1.6912821531295776,
"learning_rate": 7.734740790612136e-05,
"loss": 5.6519,
"step": 70
},
{
"epoch": 0.023364870423693954,
"grad_norm": 1.742060899734497,
"learning_rate": 7.66515864363997e-05,
"loss": 5.8117,
"step": 71
},
{
"epoch": 0.023693953105717813,
"grad_norm": 1.6967644691467285,
"learning_rate": 7.594847868906076e-05,
"loss": 5.4258,
"step": 72
},
{
"epoch": 0.02402303578774167,
"grad_norm": 1.7872289419174194,
"learning_rate": 7.52382768867422e-05,
"loss": 5.6749,
"step": 73
},
{
"epoch": 0.024352118469765528,
"grad_norm": 1.8935869932174683,
"learning_rate": 7.452117519152542e-05,
"loss": 5.6662,
"step": 74
},
{
"epoch": 0.024681201151789386,
"grad_norm": 1.9136574268341064,
"learning_rate": 7.379736965185368e-05,
"loss": 5.2773,
"step": 75
},
{
"epoch": 0.025010283833813245,
"grad_norm": 2.138458728790283,
"learning_rate": 7.30670581489344e-05,
"loss": 5.712,
"step": 76
},
{
"epoch": 0.025339366515837104,
"grad_norm": 2.2096593379974365,
"learning_rate": 7.233044034264034e-05,
"loss": 5.6899,
"step": 77
},
{
"epoch": 0.025668449197860963,
"grad_norm": 2.2397191524505615,
"learning_rate": 7.158771761692464e-05,
"loss": 5.6085,
"step": 78
},
{
"epoch": 0.025997531879884822,
"grad_norm": 2.2335917949676514,
"learning_rate": 7.083909302476453e-05,
"loss": 5.5117,
"step": 79
},
{
"epoch": 0.02632661456190868,
"grad_norm": 2.3367552757263184,
"learning_rate": 7.008477123264848e-05,
"loss": 5.7426,
"step": 80
},
{
"epoch": 0.026655697243932536,
"grad_norm": 2.4400556087493896,
"learning_rate": 6.932495846462261e-05,
"loss": 6.1936,
"step": 81
},
{
"epoch": 0.026984779925956395,
"grad_norm": 2.3639447689056396,
"learning_rate": 6.855986244591104e-05,
"loss": 5.5684,
"step": 82
},
{
"epoch": 0.027313862607980254,
"grad_norm": 2.6021528244018555,
"learning_rate": 6.778969234612584e-05,
"loss": 6.0803,
"step": 83
},
{
"epoch": 0.027642945290004113,
"grad_norm": 2.728454351425171,
"learning_rate": 6.701465872208216e-05,
"loss": 6.0576,
"step": 84
},
{
"epoch": 0.027972027972027972,
"grad_norm": 2.7983932495117188,
"learning_rate": 6.623497346023418e-05,
"loss": 5.929,
"step": 85
},
{
"epoch": 0.02830111065405183,
"grad_norm": 2.7756054401397705,
"learning_rate": 6.545084971874738e-05,
"loss": 6.0336,
"step": 86
},
{
"epoch": 0.02863019333607569,
"grad_norm": 2.88670015335083,
"learning_rate": 6.466250186922325e-05,
"loss": 5.9091,
"step": 87
},
{
"epoch": 0.02895927601809955,
"grad_norm": 3.0520219802856445,
"learning_rate": 6.387014543809223e-05,
"loss": 6.4724,
"step": 88
},
{
"epoch": 0.029288358700123408,
"grad_norm": 3.214491367340088,
"learning_rate": 6.307399704769099e-05,
"loss": 6.3031,
"step": 89
},
{
"epoch": 0.029617441382147263,
"grad_norm": 3.2701969146728516,
"learning_rate": 6.227427435703997e-05,
"loss": 6.4869,
"step": 90
},
{
"epoch": 0.029946524064171122,
"grad_norm": 3.7406764030456543,
"learning_rate": 6.147119600233758e-05,
"loss": 6.277,
"step": 91
},
{
"epoch": 0.03027560674619498,
"grad_norm": 4.176053524017334,
"learning_rate": 6.066498153718735e-05,
"loss": 6.7109,
"step": 92
},
{
"epoch": 0.03060468942821884,
"grad_norm": 4.0273637771606445,
"learning_rate": 5.985585137257401e-05,
"loss": 6.8581,
"step": 93
},
{
"epoch": 0.0309337721102427,
"grad_norm": 5.2661566734313965,
"learning_rate": 5.90440267166055e-05,
"loss": 7.2504,
"step": 94
},
{
"epoch": 0.03126285479226656,
"grad_norm": 4.821865558624268,
"learning_rate": 5.8229729514036705e-05,
"loss": 6.854,
"step": 95
},
{
"epoch": 0.03159193747429041,
"grad_norm": 5.117144584655762,
"learning_rate": 5.74131823855921e-05,
"loss": 6.8295,
"step": 96
},
{
"epoch": 0.031921020156314275,
"grad_norm": 4.751927375793457,
"learning_rate": 5.6594608567103456e-05,
"loss": 7.0724,
"step": 97
},
{
"epoch": 0.03225010283833813,
"grad_norm": 5.606512069702148,
"learning_rate": 5.577423184847932e-05,
"loss": 7.2144,
"step": 98
},
{
"epoch": 0.03257918552036199,
"grad_norm": 6.709188938140869,
"learning_rate": 5.495227651252315e-05,
"loss": 7.9249,
"step": 99
},
{
"epoch": 0.03290826820238585,
"grad_norm": 8.786857604980469,
"learning_rate": 5.4128967273616625e-05,
"loss": 7.9591,
"step": 100
},
{
"epoch": 0.03290826820238585,
"eval_loss": 1.4470807313919067,
"eval_runtime": 108.7233,
"eval_samples_per_second": 47.074,
"eval_steps_per_second": 11.773,
"step": 100
},
{
"epoch": 0.03323735088440971,
"grad_norm": 3.913679599761963,
"learning_rate": 5.330452921628497e-05,
"loss": 5.1615,
"step": 101
},
{
"epoch": 0.033566433566433566,
"grad_norm": 4.0496602058410645,
"learning_rate": 5.247918773366112e-05,
"loss": 5.2302,
"step": 102
},
{
"epoch": 0.03389551624845742,
"grad_norm": 3.0982666015625,
"learning_rate": 5.165316846586541e-05,
"loss": 5.1526,
"step": 103
},
{
"epoch": 0.034224598930481284,
"grad_norm": 2.213831663131714,
"learning_rate": 5.0826697238317935e-05,
"loss": 4.9608,
"step": 104
},
{
"epoch": 0.03455368161250514,
"grad_norm": 1.6929481029510498,
"learning_rate": 5e-05,
"loss": 5.3703,
"step": 105
},
{
"epoch": 0.034882764294529,
"grad_norm": 1.5023082494735718,
"learning_rate": 4.917330276168208e-05,
"loss": 4.9951,
"step": 106
},
{
"epoch": 0.03521184697655286,
"grad_norm": 1.387826681137085,
"learning_rate": 4.834683153413459e-05,
"loss": 5.0006,
"step": 107
},
{
"epoch": 0.03554092965857672,
"grad_norm": 1.4069514274597168,
"learning_rate": 4.7520812266338885e-05,
"loss": 4.985,
"step": 108
},
{
"epoch": 0.035870012340600575,
"grad_norm": 1.5118708610534668,
"learning_rate": 4.669547078371504e-05,
"loss": 5.1854,
"step": 109
},
{
"epoch": 0.03619909502262444,
"grad_norm": 1.3567426204681396,
"learning_rate": 4.5871032726383386e-05,
"loss": 4.9487,
"step": 110
},
{
"epoch": 0.03652817770464829,
"grad_norm": 1.417423963546753,
"learning_rate": 4.504772348747687e-05,
"loss": 5.1187,
"step": 111
},
{
"epoch": 0.03685726038667215,
"grad_norm": 1.4244048595428467,
"learning_rate": 4.4225768151520694e-05,
"loss": 5.2293,
"step": 112
},
{
"epoch": 0.03718634306869601,
"grad_norm": 1.4993596076965332,
"learning_rate": 4.3405391432896555e-05,
"loss": 5.3436,
"step": 113
},
{
"epoch": 0.037515425750719866,
"grad_norm": 1.5554583072662354,
"learning_rate": 4.2586817614407895e-05,
"loss": 5.508,
"step": 114
},
{
"epoch": 0.03784450843274373,
"grad_norm": 1.4484570026397705,
"learning_rate": 4.17702704859633e-05,
"loss": 5.2137,
"step": 115
},
{
"epoch": 0.038173591114767584,
"grad_norm": 1.5424613952636719,
"learning_rate": 4.095597328339452e-05,
"loss": 5.4582,
"step": 116
},
{
"epoch": 0.038502673796791446,
"grad_norm": 1.4834437370300293,
"learning_rate": 4.0144148627425993e-05,
"loss": 5.2233,
"step": 117
},
{
"epoch": 0.0388317564788153,
"grad_norm": 1.5142015218734741,
"learning_rate": 3.933501846281267e-05,
"loss": 5.5604,
"step": 118
},
{
"epoch": 0.039160839160839164,
"grad_norm": 1.7506635189056396,
"learning_rate": 3.852880399766243e-05,
"loss": 5.532,
"step": 119
},
{
"epoch": 0.03948992184286302,
"grad_norm": 1.6103599071502686,
"learning_rate": 3.772572564296005e-05,
"loss": 5.495,
"step": 120
},
{
"epoch": 0.039819004524886875,
"grad_norm": 1.6691735982894897,
"learning_rate": 3.6926002952309016e-05,
"loss": 5.7139,
"step": 121
},
{
"epoch": 0.04014808720691074,
"grad_norm": 1.6904277801513672,
"learning_rate": 3.612985456190778e-05,
"loss": 5.1067,
"step": 122
},
{
"epoch": 0.04047716988893459,
"grad_norm": 1.7125405073165894,
"learning_rate": 3.533749813077677e-05,
"loss": 5.241,
"step": 123
},
{
"epoch": 0.040806252570958455,
"grad_norm": 1.8208637237548828,
"learning_rate": 3.4549150281252636e-05,
"loss": 5.7307,
"step": 124
},
{
"epoch": 0.04113533525298231,
"grad_norm": 1.96183180809021,
"learning_rate": 3.3765026539765834e-05,
"loss": 5.5029,
"step": 125
},
{
"epoch": 0.04146441793500617,
"grad_norm": 2.0086212158203125,
"learning_rate": 3.298534127791785e-05,
"loss": 5.8919,
"step": 126
},
{
"epoch": 0.04179350061703003,
"grad_norm": 1.984423279762268,
"learning_rate": 3.221030765387417e-05,
"loss": 5.5399,
"step": 127
},
{
"epoch": 0.042122583299053884,
"grad_norm": 2.2246546745300293,
"learning_rate": 3.144013755408895e-05,
"loss": 6.0394,
"step": 128
},
{
"epoch": 0.042451665981077746,
"grad_norm": 2.192348003387451,
"learning_rate": 3.0675041535377405e-05,
"loss": 5.7023,
"step": 129
},
{
"epoch": 0.0427807486631016,
"grad_norm": 2.2298166751861572,
"learning_rate": 2.991522876735154e-05,
"loss": 5.8551,
"step": 130
},
{
"epoch": 0.043109831345125464,
"grad_norm": 2.232309103012085,
"learning_rate": 2.916090697523549e-05,
"loss": 6.0707,
"step": 131
},
{
"epoch": 0.04343891402714932,
"grad_norm": 2.463205575942993,
"learning_rate": 2.8412282383075363e-05,
"loss": 5.8736,
"step": 132
},
{
"epoch": 0.04376799670917318,
"grad_norm": 2.293931484222412,
"learning_rate": 2.766955965735968e-05,
"loss": 5.3597,
"step": 133
},
{
"epoch": 0.04409707939119704,
"grad_norm": 2.5523781776428223,
"learning_rate": 2.693294185106562e-05,
"loss": 5.9258,
"step": 134
},
{
"epoch": 0.0444261620732209,
"grad_norm": 2.6863903999328613,
"learning_rate": 2.6202630348146324e-05,
"loss": 6.0132,
"step": 135
},
{
"epoch": 0.044755244755244755,
"grad_norm": 2.6302249431610107,
"learning_rate": 2.547882480847461e-05,
"loss": 6.0885,
"step": 136
},
{
"epoch": 0.04508432743726861,
"grad_norm": 2.7621395587921143,
"learning_rate": 2.476172311325783e-05,
"loss": 5.7707,
"step": 137
},
{
"epoch": 0.04541341011929247,
"grad_norm": 2.7906949520111084,
"learning_rate": 2.405152131093926e-05,
"loss": 6.1995,
"step": 138
},
{
"epoch": 0.04574249280131633,
"grad_norm": 3.112114191055298,
"learning_rate": 2.3348413563600325e-05,
"loss": 6.262,
"step": 139
},
{
"epoch": 0.04607157548334019,
"grad_norm": 3.2254395484924316,
"learning_rate": 2.2652592093878666e-05,
"loss": 6.3114,
"step": 140
},
{
"epoch": 0.046400658165364046,
"grad_norm": 3.1416678428649902,
"learning_rate": 2.196424713241637e-05,
"loss": 5.8329,
"step": 141
},
{
"epoch": 0.04672974084738791,
"grad_norm": 3.423264980316162,
"learning_rate": 2.128356686585282e-05,
"loss": 6.3435,
"step": 142
},
{
"epoch": 0.047058823529411764,
"grad_norm": 4.281876564025879,
"learning_rate": 2.061073738537635e-05,
"loss": 7.4538,
"step": 143
},
{
"epoch": 0.047387906211435626,
"grad_norm": 3.966642141342163,
"learning_rate": 1.9945942635848748e-05,
"loss": 6.7706,
"step": 144
},
{
"epoch": 0.04771698889345948,
"grad_norm": 4.2417449951171875,
"learning_rate": 1.928936436551661e-05,
"loss": 6.6406,
"step": 145
},
{
"epoch": 0.04804607157548334,
"grad_norm": 4.886592388153076,
"learning_rate": 1.8641182076323148e-05,
"loss": 6.8164,
"step": 146
},
{
"epoch": 0.0483751542575072,
"grad_norm": 5.308992862701416,
"learning_rate": 1.800157297483417e-05,
"loss": 7.6286,
"step": 147
},
{
"epoch": 0.048704236939531055,
"grad_norm": 5.5149149894714355,
"learning_rate": 1.7370711923791567e-05,
"loss": 6.7779,
"step": 148
},
{
"epoch": 0.04903331962155492,
"grad_norm": 6.1263203620910645,
"learning_rate": 1.6748771394307585e-05,
"loss": 7.2376,
"step": 149
},
{
"epoch": 0.04936240230357877,
"grad_norm": 7.847179889678955,
"learning_rate": 1.6135921418712956e-05,
"loss": 8.0329,
"step": 150
},
{
"epoch": 0.04936240230357877,
"eval_loss": 1.4012147188186646,
"eval_runtime": 108.8622,
"eval_samples_per_second": 47.014,
"eval_steps_per_second": 11.758,
"step": 150
},
{
"epoch": 0.049691484985602635,
"grad_norm": 1.3999724388122559,
"learning_rate": 1.553232954407171e-05,
"loss": 5.0001,
"step": 151
},
{
"epoch": 0.05002056766762649,
"grad_norm": 1.3710346221923828,
"learning_rate": 1.4938160786375572e-05,
"loss": 5.012,
"step": 152
},
{
"epoch": 0.05034965034965035,
"grad_norm": 1.3101725578308105,
"learning_rate": 1.435357758543015e-05,
"loss": 4.9152,
"step": 153
},
{
"epoch": 0.05067873303167421,
"grad_norm": 1.3432875871658325,
"learning_rate": 1.3778739760445552e-05,
"loss": 4.9053,
"step": 154
},
{
"epoch": 0.051007815713698064,
"grad_norm": 1.325979232788086,
"learning_rate": 1.3213804466343421e-05,
"loss": 4.9404,
"step": 155
},
{
"epoch": 0.051336898395721926,
"grad_norm": 1.313812494277954,
"learning_rate": 1.2658926150792322e-05,
"loss": 5.1097,
"step": 156
},
{
"epoch": 0.05166598107774578,
"grad_norm": 1.1831108331680298,
"learning_rate": 1.2114256511983274e-05,
"loss": 5.1606,
"step": 157
},
{
"epoch": 0.051995063759769644,
"grad_norm": 1.1860599517822266,
"learning_rate": 1.157994445715706e-05,
"loss": 5.0808,
"step": 158
},
{
"epoch": 0.0523241464417935,
"grad_norm": 1.2168653011322021,
"learning_rate": 1.1056136061894384e-05,
"loss": 5.0018,
"step": 159
},
{
"epoch": 0.05265322912381736,
"grad_norm": 1.2589752674102783,
"learning_rate": 1.0542974530180327e-05,
"loss": 5.3186,
"step": 160
},
{
"epoch": 0.05298231180584122,
"grad_norm": 1.3015327453613281,
"learning_rate": 1.0040600155253765e-05,
"loss": 5.0597,
"step": 161
},
{
"epoch": 0.05331139448786507,
"grad_norm": 1.2600263357162476,
"learning_rate": 9.549150281252633e-06,
"loss": 5.1801,
"step": 162
},
{
"epoch": 0.053640477169888935,
"grad_norm": 1.3149746656417847,
"learning_rate": 9.068759265665384e-06,
"loss": 5.307,
"step": 163
},
{
"epoch": 0.05396955985191279,
"grad_norm": 1.3375885486602783,
"learning_rate": 8.599558442598998e-06,
"loss": 5.3212,
"step": 164
},
{
"epoch": 0.05429864253393665,
"grad_norm": 1.4320242404937744,
"learning_rate": 8.141676086873572e-06,
"loss": 5.1572,
"step": 165
},
{
"epoch": 0.05462772521596051,
"grad_norm": 1.502285122871399,
"learning_rate": 7.695237378953223e-06,
"loss": 5.8479,
"step": 166
},
{
"epoch": 0.05495680789798437,
"grad_norm": 1.559400200843811,
"learning_rate": 7.260364370723044e-06,
"loss": 5.4897,
"step": 167
},
{
"epoch": 0.055285890580008226,
"grad_norm": 1.56315016746521,
"learning_rate": 6.837175952121306e-06,
"loss": 5.3452,
"step": 168
},
{
"epoch": 0.05561497326203209,
"grad_norm": 1.6930738687515259,
"learning_rate": 6.425787818636131e-06,
"loss": 5.5566,
"step": 169
},
{
"epoch": 0.055944055944055944,
"grad_norm": 1.5765074491500854,
"learning_rate": 6.026312439675552e-06,
"loss": 5.1601,
"step": 170
},
{
"epoch": 0.0562731386260798,
"grad_norm": 1.706904649734497,
"learning_rate": 5.6388590278194096e-06,
"loss": 5.401,
"step": 171
},
{
"epoch": 0.05660222130810366,
"grad_norm": 1.8030728101730347,
"learning_rate": 5.263533508961827e-06,
"loss": 5.756,
"step": 172
},
{
"epoch": 0.05693130399012752,
"grad_norm": 1.733009696006775,
"learning_rate": 4.900438493352055e-06,
"loss": 5.2383,
"step": 173
},
{
"epoch": 0.05726038667215138,
"grad_norm": 1.960353136062622,
"learning_rate": 4.549673247541875e-06,
"loss": 5.5716,
"step": 174
},
{
"epoch": 0.057589469354175235,
"grad_norm": 2.042295217514038,
"learning_rate": 4.2113336672471245e-06,
"loss": 5.6513,
"step": 175
},
{
"epoch": 0.0579185520361991,
"grad_norm": 1.9490305185317993,
"learning_rate": 3.885512251130763e-06,
"loss": 5.5778,
"step": 176
},
{
"epoch": 0.05824763471822295,
"grad_norm": 2.119715690612793,
"learning_rate": 3.5722980755146517e-06,
"loss": 5.5832,
"step": 177
},
{
"epoch": 0.058576717400246815,
"grad_norm": 2.2425053119659424,
"learning_rate": 3.271776770026963e-06,
"loss": 6.1047,
"step": 178
},
{
"epoch": 0.05890580008227067,
"grad_norm": 2.295201063156128,
"learning_rate": 2.9840304941919415e-06,
"loss": 6.001,
"step": 179
},
{
"epoch": 0.059234882764294526,
"grad_norm": 2.3784329891204834,
"learning_rate": 2.7091379149682685e-06,
"loss": 5.9826,
"step": 180
},
{
"epoch": 0.05956396544631839,
"grad_norm": 2.2095887660980225,
"learning_rate": 2.4471741852423237e-06,
"loss": 5.6293,
"step": 181
},
{
"epoch": 0.059893048128342244,
"grad_norm": 2.322864532470703,
"learning_rate": 2.1982109232821178e-06,
"loss": 5.4277,
"step": 182
},
{
"epoch": 0.060222130810366106,
"grad_norm": 2.7789011001586914,
"learning_rate": 1.962316193157593e-06,
"loss": 6.1969,
"step": 183
},
{
"epoch": 0.06055121349238996,
"grad_norm": 2.7475714683532715,
"learning_rate": 1.7395544861325718e-06,
"loss": 6.066,
"step": 184
},
{
"epoch": 0.060880296174413824,
"grad_norm": 2.9355430603027344,
"learning_rate": 1.5299867030334814e-06,
"loss": 6.1455,
"step": 185
},
{
"epoch": 0.06120937885643768,
"grad_norm": 2.8145554065704346,
"learning_rate": 1.333670137599713e-06,
"loss": 5.8417,
"step": 186
},
{
"epoch": 0.06153846153846154,
"grad_norm": 2.8764495849609375,
"learning_rate": 1.1506584608200367e-06,
"loss": 6.1211,
"step": 187
},
{
"epoch": 0.0618675442204854,
"grad_norm": 2.8763437271118164,
"learning_rate": 9.810017062595322e-07,
"loss": 5.8987,
"step": 188
},
{
"epoch": 0.06219662690250925,
"grad_norm": 3.3513901233673096,
"learning_rate": 8.247462563808817e-07,
"loss": 6.1156,
"step": 189
},
{
"epoch": 0.06252570958453312,
"grad_norm": 3.12103271484375,
"learning_rate": 6.819348298638839e-07,
"loss": 5.9653,
"step": 190
},
{
"epoch": 0.06285479226655698,
"grad_norm": 3.6623032093048096,
"learning_rate": 5.526064699265753e-07,
"loss": 6.6214,
"step": 191
},
{
"epoch": 0.06318387494858083,
"grad_norm": 3.9379560947418213,
"learning_rate": 4.367965336512403e-07,
"loss": 6.751,
"step": 192
},
{
"epoch": 0.06351295763060469,
"grad_norm": 3.7529475688934326,
"learning_rate": 3.3453668231809286e-07,
"loss": 6.2796,
"step": 193
},
{
"epoch": 0.06384204031262855,
"grad_norm": 4.024684429168701,
"learning_rate": 2.458548727494292e-07,
"loss": 6.3471,
"step": 194
},
{
"epoch": 0.06417112299465241,
"grad_norm": 4.757890701293945,
"learning_rate": 1.7077534966650766e-07,
"loss": 6.3977,
"step": 195
},
{
"epoch": 0.06450020567667626,
"grad_norm": 4.87769889831543,
"learning_rate": 1.0931863906127327e-07,
"loss": 7.1139,
"step": 196
},
{
"epoch": 0.06482928835870012,
"grad_norm": 4.875145435333252,
"learning_rate": 6.150154258476315e-08,
"loss": 6.4804,
"step": 197
},
{
"epoch": 0.06515837104072399,
"grad_norm": 6.488837242126465,
"learning_rate": 2.7337132953697554e-08,
"loss": 8.4814,
"step": 198
},
{
"epoch": 0.06548745372274783,
"grad_norm": 7.969512939453125,
"learning_rate": 6.834750376549792e-09,
"loss": 8.2439,
"step": 199
},
{
"epoch": 0.0658165364047717,
"grad_norm": 10.126773834228516,
"learning_rate": 0.0,
"loss": 7.8099,
"step": 200
},
{
"epoch": 0.0658165364047717,
"eval_loss": 1.3995150327682495,
"eval_runtime": 108.8732,
"eval_samples_per_second": 47.009,
"eval_steps_per_second": 11.757,
"step": 200
}
],
"logging_steps": 1,
"max_steps": 200,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 50,
"stateful_callbacks": {
"EarlyStoppingCallback": {
"args": {
"early_stopping_patience": 5,
"early_stopping_threshold": 0.0
},
"attributes": {
"early_stopping_patience_counter": 0
}
},
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 5.354419086006682e+16,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}