autoprogrammer's picture
Add checkpoint-614
805091f verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 500,
"global_step": 614,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0163265306122449,
"grad_norm": 64.88935089111328,
"learning_rate": 1.4285714285714286e-06,
"loss": 14.3746,
"step": 5
},
{
"epoch": 0.0326530612244898,
"grad_norm": 45.59659957885742,
"learning_rate": 3.2142857142857147e-06,
"loss": 13.7736,
"step": 10
},
{
"epoch": 0.04897959183673469,
"grad_norm": 34.90564727783203,
"learning_rate": 5e-06,
"loss": 11.5266,
"step": 15
},
{
"epoch": 0.0653061224489796,
"grad_norm": 26.288707733154297,
"learning_rate": 6.785714285714287e-06,
"loss": 8.9275,
"step": 20
},
{
"epoch": 0.08163265306122448,
"grad_norm": 20.026254653930664,
"learning_rate": 8.571428571428571e-06,
"loss": 7.6264,
"step": 25
},
{
"epoch": 0.09795918367346938,
"grad_norm": 16.850696563720703,
"learning_rate": 9.99996905885448e-06,
"loss": 6.2223,
"step": 30
},
{
"epoch": 0.11428571428571428,
"grad_norm": 13.888376235961914,
"learning_rate": 9.9988861589696e-06,
"loss": 5.7476,
"step": 35
},
{
"epoch": 0.1306122448979592,
"grad_norm": 14.347899436950684,
"learning_rate": 9.996256584729157e-06,
"loss": 5.0648,
"step": 40
},
{
"epoch": 0.1469387755102041,
"grad_norm": 12.692639350891113,
"learning_rate": 9.992081149733404e-06,
"loss": 4.9457,
"step": 45
},
{
"epoch": 0.16326530612244897,
"grad_norm": 13.268131256103516,
"learning_rate": 9.986361145877783e-06,
"loss": 4.7859,
"step": 50
},
{
"epoch": 0.17959183673469387,
"grad_norm": 12.126714706420898,
"learning_rate": 9.979098342953198e-06,
"loss": 4.3966,
"step": 55
},
{
"epoch": 0.19591836734693877,
"grad_norm": 11.363078117370605,
"learning_rate": 9.970294988098452e-06,
"loss": 4.5139,
"step": 60
},
{
"epoch": 0.21224489795918366,
"grad_norm": 12.150185585021973,
"learning_rate": 9.959953805104953e-06,
"loss": 4.1404,
"step": 65
},
{
"epoch": 0.22857142857142856,
"grad_norm": 12.580989837646484,
"learning_rate": 9.948077993573983e-06,
"loss": 4.2476,
"step": 70
},
{
"epoch": 0.24489795918367346,
"grad_norm": 13.421459197998047,
"learning_rate": 9.934671227926714e-06,
"loss": 4.3114,
"step": 75
},
{
"epoch": 0.2612244897959184,
"grad_norm": 12.11119270324707,
"learning_rate": 9.919737656267335e-06,
"loss": 4.1571,
"step": 80
},
{
"epoch": 0.27755102040816326,
"grad_norm": 11.114103317260742,
"learning_rate": 9.90328189909961e-06,
"loss": 4.3387,
"step": 85
},
{
"epoch": 0.2938775510204082,
"grad_norm": 10.965707778930664,
"learning_rate": 9.885309047897285e-06,
"loss": 4.129,
"step": 90
},
{
"epoch": 0.31020408163265306,
"grad_norm": 10.521084785461426,
"learning_rate": 9.86582466352875e-06,
"loss": 3.7176,
"step": 95
},
{
"epoch": 0.32653061224489793,
"grad_norm": 11.604248046875,
"learning_rate": 9.844834774536503e-06,
"loss": 4.0923,
"step": 100
},
{
"epoch": 0.34285714285714286,
"grad_norm": 10.684212684631348,
"learning_rate": 9.822345875271884e-06,
"loss": 3.8132,
"step": 105
},
{
"epoch": 0.35918367346938773,
"grad_norm": 10.826234817504883,
"learning_rate": 9.798364923885703e-06,
"loss": 3.743,
"step": 110
},
{
"epoch": 0.37551020408163266,
"grad_norm": 10.982087135314941,
"learning_rate": 9.772899340175362e-06,
"loss": 3.7181,
"step": 115
},
{
"epoch": 0.39183673469387753,
"grad_norm": 11.453109741210938,
"learning_rate": 9.745957003289138e-06,
"loss": 3.6383,
"step": 120
},
{
"epoch": 0.40816326530612246,
"grad_norm": 10.43824577331543,
"learning_rate": 9.71754624928835e-06,
"loss": 3.7914,
"step": 125
},
{
"epoch": 0.42448979591836733,
"grad_norm": 10.69090461730957,
"learning_rate": 9.687675868568145e-06,
"loss": 3.6724,
"step": 130
},
{
"epoch": 0.44081632653061226,
"grad_norm": 11.608959197998047,
"learning_rate": 9.65635510313772e-06,
"loss": 3.6466,
"step": 135
},
{
"epoch": 0.45714285714285713,
"grad_norm": 11.002665519714355,
"learning_rate": 9.623593643760805e-06,
"loss": 3.6917,
"step": 140
},
{
"epoch": 0.47346938775510206,
"grad_norm": 11.66286849975586,
"learning_rate": 9.589401626957309e-06,
"loss": 3.6477,
"step": 145
},
{
"epoch": 0.4897959183673469,
"grad_norm": 10.273452758789062,
"learning_rate": 9.553789631867039e-06,
"loss": 3.8276,
"step": 150
},
{
"epoch": 0.5061224489795918,
"grad_norm": 9.553406715393066,
"learning_rate": 9.516768676976476e-06,
"loss": 3.5288,
"step": 155
},
{
"epoch": 0.5224489795918368,
"grad_norm": 10.327857971191406,
"learning_rate": 9.478350216709609e-06,
"loss": 3.4926,
"step": 160
},
{
"epoch": 0.5387755102040817,
"grad_norm": 9.799028396606445,
"learning_rate": 9.438546137883898e-06,
"loss": 3.8356,
"step": 165
},
{
"epoch": 0.5551020408163265,
"grad_norm": 10.007837295532227,
"learning_rate": 9.397368756032445e-06,
"loss": 3.5522,
"step": 170
},
{
"epoch": 0.5714285714285714,
"grad_norm": 11.845935821533203,
"learning_rate": 9.354830811593527e-06,
"loss": 3.5413,
"step": 175
},
{
"epoch": 0.5877551020408164,
"grad_norm": 9.952783584594727,
"learning_rate": 9.31094546596864e-06,
"loss": 3.6146,
"step": 180
},
{
"epoch": 0.6040816326530613,
"grad_norm": 11.10181999206543,
"learning_rate": 9.265726297450332e-06,
"loss": 3.4828,
"step": 185
},
{
"epoch": 0.6204081632653061,
"grad_norm": 10.211321830749512,
"learning_rate": 9.219187297021015e-06,
"loss": 3.6493,
"step": 190
},
{
"epoch": 0.636734693877551,
"grad_norm": 9.923507690429688,
"learning_rate": 9.171342864024103e-06,
"loss": 3.2479,
"step": 195
},
{
"epoch": 0.6530612244897959,
"grad_norm": 10.2088623046875,
"learning_rate": 9.122207801708802e-06,
"loss": 3.4984,
"step": 200
},
{
"epoch": 0.6693877551020408,
"grad_norm": 10.336106300354004,
"learning_rate": 9.071797312649934e-06,
"loss": 3.5449,
"step": 205
},
{
"epoch": 0.6857142857142857,
"grad_norm": 10.09914493560791,
"learning_rate": 9.020126994044194e-06,
"loss": 3.369,
"step": 210
},
{
"epoch": 0.7020408163265306,
"grad_norm": 9.624906539916992,
"learning_rate": 8.96721283288432e-06,
"loss": 3.2544,
"step": 215
},
{
"epoch": 0.7183673469387755,
"grad_norm": 10.576995849609375,
"learning_rate": 8.91307120101266e-06,
"loss": 3.4857,
"step": 220
},
{
"epoch": 0.7346938775510204,
"grad_norm": 9.93189811706543,
"learning_rate": 8.857718850055663e-06,
"loss": 3.4717,
"step": 225
},
{
"epoch": 0.7510204081632653,
"grad_norm": 11.030794143676758,
"learning_rate": 8.801172906240857e-06,
"loss": 3.3982,
"step": 230
},
{
"epoch": 0.7673469387755102,
"grad_norm": 9.748960494995117,
"learning_rate": 8.743450865097929e-06,
"loss": 3.4867,
"step": 235
},
{
"epoch": 0.7836734693877551,
"grad_norm": 9.28258228302002,
"learning_rate": 8.68457058604555e-06,
"loss": 3.4942,
"step": 240
},
{
"epoch": 0.8,
"grad_norm": 9.684868812561035,
"learning_rate": 8.624550286865592e-06,
"loss": 3.5008,
"step": 245
},
{
"epoch": 0.8163265306122449,
"grad_norm": 9.635165214538574,
"learning_rate": 8.563408538066486e-06,
"loss": 3.5459,
"step": 250
},
{
"epoch": 0.8326530612244898,
"grad_norm": 9.419096946716309,
"learning_rate": 8.501164257137431e-06,
"loss": 3.4455,
"step": 255
},
{
"epoch": 0.8489795918367347,
"grad_norm": 9.14213752746582,
"learning_rate": 8.437836702695253e-06,
"loss": 3.2428,
"step": 260
},
{
"epoch": 0.8653061224489796,
"grad_norm": 9.956143379211426,
"learning_rate": 8.373445468525719e-06,
"loss": 3.4352,
"step": 265
},
{
"epoch": 0.8816326530612245,
"grad_norm": 9.325640678405762,
"learning_rate": 8.30801047752113e-06,
"loss": 3.4512,
"step": 270
},
{
"epoch": 0.8979591836734694,
"grad_norm": 10.310145378112793,
"learning_rate": 8.241551975516133e-06,
"loss": 3.654,
"step": 275
},
{
"epoch": 0.9142857142857143,
"grad_norm": 10.872224807739258,
"learning_rate": 8.17409052502355e-06,
"loss": 3.4794,
"step": 280
},
{
"epoch": 0.9306122448979591,
"grad_norm": 9.897497177124023,
"learning_rate": 8.105646998872275e-06,
"loss": 3.4442,
"step": 285
},
{
"epoch": 0.9469387755102041,
"grad_norm": 9.942341804504395,
"learning_rate": 8.036242573749142e-06,
"loss": 3.2163,
"step": 290
},
{
"epoch": 0.963265306122449,
"grad_norm": 10.519649505615234,
"learning_rate": 7.965898723646777e-06,
"loss": 3.6575,
"step": 295
},
{
"epoch": 0.9795918367346939,
"grad_norm": 9.339863777160645,
"learning_rate": 7.894637213219454e-06,
"loss": 3.1936,
"step": 300
},
{
"epoch": 0.9959183673469387,
"grad_norm": 10.674720764160156,
"learning_rate": 7.82248009104904e-06,
"loss": 3.5039,
"step": 305
},
{
"epoch": 1.0097959183673468,
"grad_norm": 8.822035789489746,
"learning_rate": 7.749449682823077e-06,
"loss": 2.4509,
"step": 310
},
{
"epoch": 1.026122448979592,
"grad_norm": 9.485854148864746,
"learning_rate": 7.675568584427125e-06,
"loss": 2.6552,
"step": 315
},
{
"epoch": 1.0424489795918368,
"grad_norm": 10.655379295349121,
"learning_rate": 7.600859654953513e-06,
"loss": 2.4981,
"step": 320
},
{
"epoch": 1.0587755102040817,
"grad_norm": 10.05766487121582,
"learning_rate": 7.525346009628647e-06,
"loss": 2.6852,
"step": 325
},
{
"epoch": 1.0751020408163265,
"grad_norm": 9.784882545471191,
"learning_rate": 7.449051012661073e-06,
"loss": 2.5807,
"step": 330
},
{
"epoch": 1.0914285714285714,
"grad_norm": 11.17435359954834,
"learning_rate": 7.371998270012504e-06,
"loss": 2.7083,
"step": 335
},
{
"epoch": 1.1077551020408163,
"grad_norm": 11.014193534851074,
"learning_rate": 7.2942116220940406e-06,
"loss": 2.587,
"step": 340
},
{
"epoch": 1.1240816326530612,
"grad_norm": 10.567625999450684,
"learning_rate": 7.215715136389862e-06,
"loss": 2.6983,
"step": 345
},
{
"epoch": 1.140408163265306,
"grad_norm": 11.436513900756836,
"learning_rate": 7.136533100010654e-06,
"loss": 2.6842,
"step": 350
},
{
"epoch": 1.1567346938775511,
"grad_norm": 9.689658164978027,
"learning_rate": 7.0566900121790775e-06,
"loss": 2.7408,
"step": 355
},
{
"epoch": 1.173061224489796,
"grad_norm": 10.072765350341797,
"learning_rate": 6.9762105766496315e-06,
"loss": 2.4803,
"step": 360
},
{
"epoch": 1.1893877551020409,
"grad_norm": 9.877284049987793,
"learning_rate": 6.8951196940652045e-06,
"loss": 2.5766,
"step": 365
},
{
"epoch": 1.2057142857142857,
"grad_norm": 10.164962768554688,
"learning_rate": 6.8134424542527215e-06,
"loss": 2.3245,
"step": 370
},
{
"epoch": 1.2220408163265306,
"grad_norm": 9.651641845703125,
"learning_rate": 6.731204128460265e-06,
"loss": 2.4775,
"step": 375
},
{
"epoch": 1.2383673469387755,
"grad_norm": 9.717443466186523,
"learning_rate": 6.648430161538047e-06,
"loss": 2.5835,
"step": 380
},
{
"epoch": 1.2546938775510204,
"grad_norm": 9.949031829833984,
"learning_rate": 6.5651461640656825e-06,
"loss": 2.6598,
"step": 385
},
{
"epoch": 1.2710204081632652,
"grad_norm": 10.311724662780762,
"learning_rate": 6.481377904428171e-06,
"loss": 2.567,
"step": 390
},
{
"epoch": 1.28734693877551,
"grad_norm": 10.227352142333984,
"learning_rate": 6.397151300843065e-06,
"loss": 2.652,
"step": 395
},
{
"epoch": 1.3036734693877552,
"grad_norm": 9.446685791015625,
"learning_rate": 6.312492413341274e-06,
"loss": 2.5414,
"step": 400
},
{
"epoch": 1.32,
"grad_norm": 10.463958740234375,
"learning_rate": 6.227427435703997e-06,
"loss": 2.6942,
"step": 405
},
{
"epoch": 1.336326530612245,
"grad_norm": 10.328426361083984,
"learning_rate": 6.141982687358255e-06,
"loss": 2.6176,
"step": 410
},
{
"epoch": 1.3526530612244898,
"grad_norm": 10.598652839660645,
"learning_rate": 6.056184605233576e-06,
"loss": 2.5755,
"step": 415
},
{
"epoch": 1.3689795918367347,
"grad_norm": 10.119470596313477,
"learning_rate": 5.970059735582295e-06,
"loss": 2.4044,
"step": 420
},
{
"epoch": 1.3853061224489795,
"grad_norm": 9.856766700744629,
"learning_rate": 5.8836347257660485e-06,
"loss": 2.6117,
"step": 425
},
{
"epoch": 1.4016326530612244,
"grad_norm": 11.175994873046875,
"learning_rate": 5.796936316010984e-06,
"loss": 2.5551,
"step": 430
},
{
"epoch": 1.4179591836734695,
"grad_norm": 10.019680976867676,
"learning_rate": 5.709991331134224e-06,
"loss": 2.7055,
"step": 435
},
{
"epoch": 1.4342857142857142,
"grad_norm": 9.860037803649902,
"learning_rate": 5.622826672244169e-06,
"loss": 2.5439,
"step": 440
},
{
"epoch": 1.4506122448979593,
"grad_norm": 10.403874397277832,
"learning_rate": 5.535469308417198e-06,
"loss": 2.4987,
"step": 445
},
{
"epoch": 1.4669387755102041,
"grad_norm": 11.318153381347656,
"learning_rate": 5.447946268353324e-06,
"loss": 2.414,
"step": 450
},
{
"epoch": 1.483265306122449,
"grad_norm": 11.142492294311523,
"learning_rate": 5.3602846320134216e-06,
"loss": 2.6493,
"step": 455
},
{
"epoch": 1.4995918367346939,
"grad_norm": 10.691166877746582,
"learning_rate": 5.272511522240574e-06,
"loss": 2.4802,
"step": 460
},
{
"epoch": 1.5159183673469387,
"grad_norm": 10.497188568115234,
"learning_rate": 5.184654096368172e-06,
"loss": 2.451,
"step": 465
},
{
"epoch": 1.5322448979591838,
"grad_norm": 10.306930541992188,
"learning_rate": 5.096739537817324e-06,
"loss": 2.4855,
"step": 470
},
{
"epoch": 1.5485714285714285,
"grad_norm": 10.184440612792969,
"learning_rate": 5.00879504768621e-06,
"loss": 2.6157,
"step": 475
},
{
"epoch": 1.5648979591836736,
"grad_norm": 10.472289085388184,
"learning_rate": 4.92084783633395e-06,
"loss": 2.6851,
"step": 480
},
{
"epoch": 1.5812244897959182,
"grad_norm": 11.001713752746582,
"learning_rate": 4.832925114961629e-06,
"loss": 2.4499,
"step": 485
},
{
"epoch": 1.5975510204081633,
"grad_norm": 9.671372413635254,
"learning_rate": 4.745054087193025e-06,
"loss": 2.7094,
"step": 490
},
{
"epoch": 1.6138775510204082,
"grad_norm": 10.427942276000977,
"learning_rate": 4.657261940657732e-06,
"loss": 2.6824,
"step": 495
},
{
"epoch": 1.630204081632653,
"grad_norm": 10.743348121643066,
"learning_rate": 4.569575838579184e-06,
"loss": 2.4526,
"step": 500
},
{
"epoch": 1.646530612244898,
"grad_norm": 10.53161334991455,
"learning_rate": 4.48202291137026e-06,
"loss": 2.4655,
"step": 505
},
{
"epoch": 1.6628571428571428,
"grad_norm": 9.792738914489746,
"learning_rate": 4.394630248239029e-06,
"loss": 2.5056,
"step": 510
},
{
"epoch": 1.679183673469388,
"grad_norm": 9.901719093322754,
"learning_rate": 4.307424888807242e-06,
"loss": 2.4083,
"step": 515
},
{
"epoch": 1.6955102040816326,
"grad_norm": 9.59782600402832,
"learning_rate": 4.220433814744179e-06,
"loss": 2.4498,
"step": 520
},
{
"epoch": 1.7118367346938776,
"grad_norm": 10.38283634185791,
"learning_rate": 4.133683941418411e-06,
"loss": 2.4888,
"step": 525
},
{
"epoch": 1.7281632653061223,
"grad_norm": 9.695813179016113,
"learning_rate": 4.047202109570086e-06,
"loss": 2.592,
"step": 530
},
{
"epoch": 1.7444897959183674,
"grad_norm": 10.314712524414062,
"learning_rate": 3.961015077006301e-06,
"loss": 2.4875,
"step": 535
},
{
"epoch": 1.7608163265306123,
"grad_norm": 10.28763198852539,
"learning_rate": 3.875149510322137e-06,
"loss": 2.4304,
"step": 540
},
{
"epoch": 1.7771428571428571,
"grad_norm": 9.332132339477539,
"learning_rate": 3.7896319766499073e-06,
"loss": 2.4338,
"step": 545
},
{
"epoch": 1.793469387755102,
"grad_norm": 10.027193069458008,
"learning_rate": 3.7044889354391934e-06,
"loss": 2.4814,
"step": 550
},
{
"epoch": 1.8097959183673469,
"grad_norm": 9.520379066467285,
"learning_rate": 3.619746730270185e-06,
"loss": 2.3753,
"step": 555
},
{
"epoch": 1.826122448979592,
"grad_norm": 10.98616886138916,
"learning_rate": 3.5354315807028826e-06,
"loss": 2.5019,
"step": 560
},
{
"epoch": 1.8424489795918366,
"grad_norm": 11.574503898620605,
"learning_rate": 3.451569574164658e-06,
"loss": 2.499,
"step": 565
},
{
"epoch": 1.8587755102040817,
"grad_norm": 9.708284378051758,
"learning_rate": 3.3681866578787124e-06,
"loss": 2.3545,
"step": 570
},
{
"epoch": 1.8751020408163264,
"grad_norm": 10.516324043273926,
"learning_rate": 3.285308630835903e-06,
"loss": 2.4484,
"step": 575
},
{
"epoch": 1.8914285714285715,
"grad_norm": 10.152556419372559,
"learning_rate": 3.202961135812437e-06,
"loss": 2.3661,
"step": 580
},
{
"epoch": 1.9077551020408163,
"grad_norm": 10.867803573608398,
"learning_rate": 3.121169651435903e-06,
"loss": 2.4558,
"step": 585
},
{
"epoch": 1.9240816326530612,
"grad_norm": 10.548734664916992,
"learning_rate": 3.039959484302083e-06,
"loss": 2.5513,
"step": 590
},
{
"epoch": 1.940408163265306,
"grad_norm": 10.610418319702148,
"learning_rate": 2.959355761145001e-06,
"loss": 2.6669,
"step": 595
},
{
"epoch": 1.956734693877551,
"grad_norm": 10.734642028808594,
"learning_rate": 2.8793834210626036e-06,
"loss": 2.5634,
"step": 600
},
{
"epoch": 1.973061224489796,
"grad_norm": 10.054177284240723,
"learning_rate": 2.8000672078005277e-06,
"loss": 2.4683,
"step": 605
},
{
"epoch": 1.9893877551020407,
"grad_norm": 10.605957984924316,
"learning_rate": 2.7214316620962727e-06,
"loss": 2.5544,
"step": 610
}
],
"logging_steps": 5,
"max_steps": 921,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 1.9375810322702008e+18,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}