sirui6011's picture
add checkpoints/codi-single-3b
8a4bc09 verified
Raw
History Blame Contribute Delete
28.8 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.14417531718569782,
"eval_steps": 500,
"global_step": 500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0014417531718569781,
"grad_norm": 3808.0,
"kd_loss": 0.4765625,
"learning_rate": 1.3333333333333334e-06,
"loss": 2.4865,
"step": 5,
"student_loss": 1.2782293558120728,
"teacher_loss": 0.0020202866289764643
},
{
"epoch": 0.0028835063437139563,
"grad_norm": 496.0,
"kd_loss": 0.453125,
"learning_rate": 3e-06,
"loss": 2.0957,
"step": 10,
"student_loss": 1.0292338132858276,
"teacher_loss": 0.005245466250926256
},
{
"epoch": 0.004325259515570935,
"grad_norm": 238.0,
"kd_loss": 0.4453125,
"learning_rate": 4.666666666666667e-06,
"loss": 1.9295,
"step": 15,
"student_loss": 0.631219208240509,
"teacher_loss": 0.0013347615022212267
},
{
"epoch": 0.0057670126874279125,
"grad_norm": 133.0,
"kd_loss": 0.44140625,
"learning_rate": 6.333333333333333e-06,
"loss": 1.8503,
"step": 20,
"student_loss": 1.4391331672668457,
"teacher_loss": 0.0005473981145769358
},
{
"epoch": 0.00720876585928489,
"grad_norm": 75.5,
"kd_loss": 0.3984375,
"learning_rate": 8.000000000000001e-06,
"loss": 1.0972,
"step": 25,
"student_loss": 0.18605084717273712,
"teacher_loss": 0.0011649713851511478
},
{
"epoch": 0.00865051903114187,
"grad_norm": 24.25,
"kd_loss": 0.376953125,
"learning_rate": 9.666666666666667e-06,
"loss": 0.5883,
"step": 30,
"student_loss": 0.05895603448152542,
"teacher_loss": 0.030797353014349937
},
{
"epoch": 0.010092272202998846,
"grad_norm": 9.1875,
"kd_loss": 0.318359375,
"learning_rate": 9.99958042442916e-06,
"loss": 0.4657,
"step": 35,
"student_loss": 0.00801500491797924,
"teacher_loss": 0.04946871101856232
},
{
"epoch": 0.011534025374855825,
"grad_norm": 7.09375,
"kd_loss": 0.296875,
"learning_rate": 9.997876019358083e-06,
"loss": 0.402,
"step": 40,
"student_loss": 0.27407729625701904,
"teacher_loss": 0.04900167137384415
},
{
"epoch": 0.012975778546712802,
"grad_norm": 10.6875,
"kd_loss": 0.265625,
"learning_rate": 9.99486100792044e-06,
"loss": 0.3281,
"step": 45,
"student_loss": 0.3285456597805023,
"teacher_loss": 0.003808467648923397
},
{
"epoch": 0.01441753171856978,
"grad_norm": 21.875,
"kd_loss": 0.2578125,
"learning_rate": 9.990536180750724e-06,
"loss": 0.351,
"step": 50,
"student_loss": 0.03528054431080818,
"teacher_loss": 0.04187293350696564
},
{
"epoch": 0.015859284890426758,
"grad_norm": 228.0,
"kd_loss": 0.2421875,
"learning_rate": 9.984902671959911e-06,
"loss": 0.3368,
"step": 55,
"student_loss": 0.004587164148688316,
"teacher_loss": 0.0026238495483994484
},
{
"epoch": 0.01730103806228374,
"grad_norm": 6.4375,
"kd_loss": 0.23046875,
"learning_rate": 9.97796195883804e-06,
"loss": 0.3291,
"step": 60,
"student_loss": 0.048888176679611206,
"teacher_loss": 0.0033420324325561523
},
{
"epoch": 0.018742791234140715,
"grad_norm": 6.375,
"kd_loss": 0.2099609375,
"learning_rate": 9.969715861466839e-06,
"loss": 0.3147,
"step": 65,
"student_loss": 0.11189773678779602,
"teacher_loss": 0.0413309670984745
},
{
"epoch": 0.020184544405997693,
"grad_norm": 4.09375,
"kd_loss": 0.208984375,
"learning_rate": 9.96016654224243e-06,
"loss": 0.3096,
"step": 70,
"student_loss": 0.026729928329586983,
"teacher_loss": 0.0019827873911708593
},
{
"epoch": 0.02162629757785467,
"grad_norm": 4.25,
"kd_loss": 0.1748046875,
"learning_rate": 9.94931650530827e-06,
"loss": 0.2729,
"step": 75,
"student_loss": 0.006473238579928875,
"teacher_loss": 0.005465318448841572
},
{
"epoch": 0.02306805074971165,
"grad_norm": 3.859375,
"kd_loss": 0.16796875,
"learning_rate": 9.93716859589851e-06,
"loss": 0.2662,
"step": 80,
"student_loss": 0.00662571657449007,
"teacher_loss": 0.004144964274019003
},
{
"epoch": 0.024509803921568627,
"grad_norm": 2.78125,
"kd_loss": 0.1640625,
"learning_rate": 9.923725999591846e-06,
"loss": 0.2261,
"step": 85,
"student_loss": 0.004623747896403074,
"teacher_loss": 0.002723712706938386
},
{
"epoch": 0.025951557093425604,
"grad_norm": 5.5625,
"kd_loss": 0.1943359375,
"learning_rate": 9.908992241476189e-06,
"loss": 0.2543,
"step": 90,
"student_loss": 0.11273087561130524,
"teacher_loss": 0.0015502618625760078
},
{
"epoch": 0.027393310265282585,
"grad_norm": 3.4375,
"kd_loss": 0.189453125,
"learning_rate": 9.892971185224244e-06,
"loss": 0.2267,
"step": 95,
"student_loss": 0.006197327747941017,
"teacher_loss": 0.008993545547127724
},
{
"epoch": 0.02883506343713956,
"grad_norm": 6.34375,
"kd_loss": 0.134765625,
"learning_rate": 9.875667032080354e-06,
"loss": 0.2274,
"step": 100,
"student_loss": 0.0032730416860431433,
"teacher_loss": 0.0036007578019052744
},
{
"epoch": 0.03027681660899654,
"grad_norm": 4.0,
"kd_loss": 0.146484375,
"learning_rate": 9.857084319758772e-06,
"loss": 0.2421,
"step": 105,
"student_loss": 0.04058241471648216,
"teacher_loss": 0.0012296679196879268
},
{
"epoch": 0.031718569780853516,
"grad_norm": 4.6875,
"kd_loss": 0.1787109375,
"learning_rate": 9.837227921253747e-06,
"loss": 0.2273,
"step": 110,
"student_loss": 0.004547884222120047,
"teacher_loss": 0.023880567401647568
},
{
"epoch": 0.03316032295271049,
"grad_norm": 2.625,
"kd_loss": 0.14453125,
"learning_rate": 9.816103043561648e-06,
"loss": 0.2142,
"step": 115,
"student_loss": 0.001855566632002592,
"teacher_loss": 0.0016716865357011557
},
{
"epoch": 0.03460207612456748,
"grad_norm": 3.078125,
"kd_loss": 0.1357421875,
"learning_rate": 9.79371522631553e-06,
"loss": 0.2149,
"step": 120,
"student_loss": 0.019737211987376213,
"teacher_loss": 0.0027425403241068125
},
{
"epoch": 0.036043829296424454,
"grad_norm": 3.390625,
"kd_loss": 0.146484375,
"learning_rate": 9.770070340332457e-06,
"loss": 0.1956,
"step": 125,
"student_loss": 0.10938042402267456,
"teacher_loss": 0.001064821844920516
},
{
"epoch": 0.03748558246828143,
"grad_norm": 3.734375,
"kd_loss": 0.146484375,
"learning_rate": 9.745174586073982e-06,
"loss": 0.2099,
"step": 130,
"student_loss": 0.0035836249589920044,
"teacher_loss": 0.002439548959955573
},
{
"epoch": 0.03892733564013841,
"grad_norm": 3.0,
"kd_loss": 0.1337890625,
"learning_rate": 9.719034492020183e-06,
"loss": 0.202,
"step": 135,
"student_loss": 0.003862213110551238,
"teacher_loss": 0.0010834896238520741
},
{
"epoch": 0.040369088811995385,
"grad_norm": 7.1875,
"kd_loss": 0.1357421875,
"learning_rate": 9.691656912957686e-06,
"loss": 0.218,
"step": 140,
"student_loss": 0.0022195407655090094,
"teacher_loss": 0.0014228483196347952
},
{
"epoch": 0.04181084198385236,
"grad_norm": 3.703125,
"kd_loss": 0.150390625,
"learning_rate": 9.663049028182112e-06,
"loss": 0.2077,
"step": 145,
"student_loss": 0.11333022266626358,
"teacher_loss": 0.00682886503636837
},
{
"epoch": 0.04325259515570934,
"grad_norm": 2.859375,
"kd_loss": 0.14453125,
"learning_rate": 9.633218339615433e-06,
"loss": 0.1935,
"step": 150,
"student_loss": 0.0032606760505586863,
"teacher_loss": 0.0031273479107767344
},
{
"epoch": 0.04469434832756632,
"grad_norm": 3.578125,
"kd_loss": 0.12451171875,
"learning_rate": 9.602172669838721e-06,
"loss": 0.2199,
"step": 155,
"student_loss": 0.0088576041162014,
"teacher_loss": 0.0016166985733434558
},
{
"epoch": 0.0461361014994233,
"grad_norm": 4.09375,
"kd_loss": 0.1708984375,
"learning_rate": 9.569920160040815e-06,
"loss": 0.2018,
"step": 160,
"student_loss": 0.13294154405593872,
"teacher_loss": 0.03791189566254616
},
{
"epoch": 0.04757785467128028,
"grad_norm": 3.921875,
"kd_loss": 0.138671875,
"learning_rate": 9.536469267883432e-06,
"loss": 0.208,
"step": 165,
"student_loss": 0.002772042527794838,
"teacher_loss": 0.005522818770259619
},
{
"epoch": 0.049019607843137254,
"grad_norm": 5.03125,
"kd_loss": 0.126953125,
"learning_rate": 9.501828765283295e-06,
"loss": 0.1962,
"step": 170,
"student_loss": 0.003656906308606267,
"teacher_loss": 0.0018494409741833806
},
{
"epoch": 0.05046136101499423,
"grad_norm": 3.515625,
"kd_loss": 0.12451171875,
"learning_rate": 9.466007736111846e-06,
"loss": 0.1935,
"step": 175,
"student_loss": 0.017079656943678856,
"teacher_loss": 0.0010717228287830949
},
{
"epoch": 0.05190311418685121,
"grad_norm": 5.03125,
"kd_loss": 0.11669921875,
"learning_rate": 9.429015573813163e-06,
"loss": 0.1861,
"step": 180,
"student_loss": 0.003456867765635252,
"teacher_loss": 0.0010596277425065637
},
{
"epoch": 0.05334486735870819,
"grad_norm": 4.0625,
"kd_loss": 0.1455078125,
"learning_rate": 9.390861978940687e-06,
"loss": 0.1921,
"step": 185,
"student_loss": 0.31187787652015686,
"teacher_loss": 0.0008243238553404808
},
{
"epoch": 0.05478662053056517,
"grad_norm": 5.09375,
"kd_loss": 0.1416015625,
"learning_rate": 9.351556956613423e-06,
"loss": 0.2044,
"step": 190,
"student_loss": 0.011734717525541782,
"teacher_loss": 0.0015390698099508882
},
{
"epoch": 0.056228373702422146,
"grad_norm": 4.5,
"kd_loss": 0.1337890625,
"learning_rate": 9.31111081389227e-06,
"loss": 0.1778,
"step": 195,
"student_loss": 0.05741060897707939,
"teacher_loss": 0.0007789382943883538
},
{
"epoch": 0.05767012687427912,
"grad_norm": 2.28125,
"kd_loss": 0.12890625,
"learning_rate": 9.269534157077177e-06,
"loss": 0.1743,
"step": 200,
"student_loss": 0.0014264394994825125,
"teacher_loss": 0.0006533891428261995
},
{
"epoch": 0.0591118800461361,
"grad_norm": 4.25,
"kd_loss": 0.1376953125,
"learning_rate": 9.226837888925813e-06,
"loss": 0.1969,
"step": 205,
"student_loss": 0.0015782959526404738,
"teacher_loss": 0.0368424728512764
},
{
"epoch": 0.06055363321799308,
"grad_norm": 7.5625,
"kd_loss": 0.1123046875,
"learning_rate": 9.183033205794525e-06,
"loss": 0.1836,
"step": 210,
"student_loss": 0.01342203002423048,
"teacher_loss": 0.0011842605890706182
},
{
"epoch": 0.061995386389850055,
"grad_norm": 3.0,
"kd_loss": 0.126953125,
"learning_rate": 9.13813159470227e-06,
"loss": 0.1824,
"step": 215,
"student_loss": 0.0014404732501134276,
"teacher_loss": 0.0007589462329633534
},
{
"epoch": 0.06343713956170703,
"grad_norm": 7.125,
"kd_loss": 0.1328125,
"learning_rate": 9.092144830318357e-06,
"loss": 0.21,
"step": 220,
"student_loss": 0.2732444703578949,
"teacher_loss": 0.00960276648402214
},
{
"epoch": 0.06487889273356401,
"grad_norm": 6.15625,
"kd_loss": 0.1083984375,
"learning_rate": 9.045084971874738e-06,
"loss": 0.1941,
"step": 225,
"student_loss": 0.2691424489021301,
"teacher_loss": 0.003315337933599949
},
{
"epoch": 0.06632064590542099,
"grad_norm": 2.46875,
"kd_loss": 0.10986328125,
"learning_rate": 8.99696436000368e-06,
"loss": 0.1702,
"step": 230,
"student_loss": 0.08001423627138138,
"teacher_loss": 0.009347192943096161
},
{
"epoch": 0.06776239907727798,
"grad_norm": 3.734375,
"kd_loss": 0.1337890625,
"learning_rate": 8.947795613501658e-06,
"loss": 0.1778,
"step": 235,
"student_loss": 0.003426821669563651,
"teacher_loss": 0.0008243515621870756
},
{
"epoch": 0.06920415224913495,
"grad_norm": 3.421875,
"kd_loss": 0.10205078125,
"learning_rate": 8.897591626020284e-06,
"loss": 0.1928,
"step": 240,
"student_loss": 0.004231320694088936,
"teacher_loss": 0.0010777115821838379
},
{
"epoch": 0.07064590542099193,
"grad_norm": 3.640625,
"kd_loss": 0.11181640625,
"learning_rate": 8.846365562685178e-06,
"loss": 0.1721,
"step": 245,
"student_loss": 0.003189836163073778,
"teacher_loss": 0.0029278292786329985
},
{
"epoch": 0.07208765859284891,
"grad_norm": 3.28125,
"kd_loss": 0.107421875,
"learning_rate": 8.794130856643635e-06,
"loss": 0.1624,
"step": 250,
"student_loss": 0.0030161093454807997,
"teacher_loss": 0.0015030049253255129
},
{
"epoch": 0.07352941176470588,
"grad_norm": 2.671875,
"kd_loss": 0.1171875,
"learning_rate": 8.74090120554202e-06,
"loss": 0.181,
"step": 255,
"student_loss": 0.0010447532404214144,
"teacher_loss": 0.0014572968939319253
},
{
"epoch": 0.07497116493656286,
"grad_norm": 3.03125,
"kd_loss": 0.125,
"learning_rate": 8.686690567933803e-06,
"loss": 0.18,
"step": 260,
"student_loss": 0.002235305029898882,
"teacher_loss": 0.03705403953790665
},
{
"epoch": 0.07641291810841984,
"grad_norm": 3.84375,
"kd_loss": 0.10107421875,
"learning_rate": 8.63151315961915e-06,
"loss": 0.1751,
"step": 265,
"student_loss": 0.0019888102542608976,
"teacher_loss": 0.0012628042604774237
},
{
"epoch": 0.07785467128027682,
"grad_norm": 4.03125,
"kd_loss": 0.103515625,
"learning_rate": 8.575383449917103e-06,
"loss": 0.1698,
"step": 270,
"student_loss": 0.009670126251876354,
"teacher_loss": 0.0018007074249908328
},
{
"epoch": 0.07929642445213379,
"grad_norm": 5.0625,
"kd_loss": 0.11669921875,
"learning_rate": 8.518316157871232e-06,
"loss": 0.1792,
"step": 275,
"student_loss": 0.0027291348669677973,
"teacher_loss": 0.03865275904536247
},
{
"epoch": 0.08073817762399077,
"grad_norm": 6.40625,
"kd_loss": 0.12451171875,
"learning_rate": 8.460326248389825e-06,
"loss": 0.1868,
"step": 280,
"student_loss": 0.0005779159837402403,
"teacher_loss": 0.0004988706787116826
},
{
"epoch": 0.08217993079584775,
"grad_norm": 5.03125,
"kd_loss": 0.1083984375,
"learning_rate": 8.401428928321607e-06,
"loss": 0.1777,
"step": 285,
"student_loss": 0.00322159961797297,
"teacher_loss": 0.0016653644852340221
},
{
"epoch": 0.08362168396770472,
"grad_norm": 4.15625,
"kd_loss": 0.119140625,
"learning_rate": 8.341639642468002e-06,
"loss": 0.2245,
"step": 290,
"student_loss": 0.025423452258110046,
"teacher_loss": 0.006107364781200886
},
{
"epoch": 0.0850634371395617,
"grad_norm": 2.21875,
"kd_loss": 0.10498046875,
"learning_rate": 8.280974069532999e-06,
"loss": 0.1742,
"step": 295,
"student_loss": 0.0032805479131639004,
"teacher_loss": 0.002079744590446353
},
{
"epoch": 0.08650519031141868,
"grad_norm": 4.875,
"kd_loss": 0.1220703125,
"learning_rate": 8.219448118011687e-06,
"loss": 0.1698,
"step": 300,
"student_loss": 0.05386965721845627,
"teacher_loss": 0.0015291464515030384
},
{
"epoch": 0.08794694348327567,
"grad_norm": 2.703125,
"kd_loss": 0.09375,
"learning_rate": 8.157077922018537e-06,
"loss": 0.1735,
"step": 305,
"student_loss": 0.007909238338470459,
"teacher_loss": 0.0032228778582066298
},
{
"epoch": 0.08938869665513265,
"grad_norm": 6.375,
"kd_loss": 0.091796875,
"learning_rate": 8.093879837056486e-06,
"loss": 0.1662,
"step": 310,
"student_loss": 0.0014559343690052629,
"teacher_loss": 0.0014570873463526368
},
{
"epoch": 0.09083044982698962,
"grad_norm": 6.09375,
"kd_loss": 0.09716796875,
"learning_rate": 8.029870435728018e-06,
"loss": 0.1905,
"step": 315,
"student_loss": 0.13904070854187012,
"teacher_loss": 0.00045576939010061324
},
{
"epoch": 0.0922722029988466,
"grad_norm": 8.75,
"kd_loss": 0.091796875,
"learning_rate": 7.965066503389264e-06,
"loss": 0.1801,
"step": 320,
"student_loss": 0.0017298327293246984,
"teacher_loss": 0.001036101020872593
},
{
"epoch": 0.09371395617070358,
"grad_norm": 5.53125,
"kd_loss": 0.15625,
"learning_rate": 7.89948503374835e-06,
"loss": 0.1636,
"step": 325,
"student_loss": 0.0033407427836209536,
"teacher_loss": 0.02077825367450714
},
{
"epoch": 0.09515570934256055,
"grad_norm": 5.09375,
"kd_loss": 0.10693359375,
"learning_rate": 7.833143224409076e-06,
"loss": 0.1884,
"step": 330,
"student_loss": 0.006418874487280846,
"teacher_loss": 0.0011637036222964525
},
{
"epoch": 0.09659746251441753,
"grad_norm": 4.71875,
"kd_loss": 0.09716796875,
"learning_rate": 7.766058472361154e-06,
"loss": 0.1577,
"step": 335,
"student_loss": 0.0016794800758361816,
"teacher_loss": 0.0023754944559186697
},
{
"epoch": 0.09803921568627451,
"grad_norm": 5.09375,
"kd_loss": 0.09326171875,
"learning_rate": 7.698248369418146e-06,
"loss": 0.1589,
"step": 340,
"student_loss": 0.044694170355796814,
"teacher_loss": 0.007826481945812702
},
{
"epoch": 0.09948096885813149,
"grad_norm": 5.375,
"kd_loss": 0.0966796875,
"learning_rate": 7.629730697604314e-06,
"loss": 0.1807,
"step": 345,
"student_loss": 0.09194417297840118,
"teacher_loss": 0.0007945778197608888
},
{
"epoch": 0.10092272202998846,
"grad_norm": 5.875,
"kd_loss": 0.11865234375,
"learning_rate": 7.560523424491595e-06,
"loss": 0.1526,
"step": 350,
"student_loss": 0.005946993827819824,
"teacher_loss": 0.0006145219667814672
},
{
"epoch": 0.10236447520184544,
"grad_norm": 9.1875,
"kd_loss": 0.10302734375,
"learning_rate": 7.490644698487909e-06,
"loss": 0.1627,
"step": 355,
"student_loss": 0.0015843416331335902,
"teacher_loss": 0.0014968032483011484
},
{
"epoch": 0.10380622837370242,
"grad_norm": 12.625,
"kd_loss": 0.10791015625,
"learning_rate": 7.420112844078066e-06,
"loss": 0.1682,
"step": 360,
"student_loss": 0.01987134851515293,
"teacher_loss": 0.001595525536686182
},
{
"epoch": 0.1052479815455594,
"grad_norm": 6.5625,
"kd_loss": 0.103515625,
"learning_rate": 7.348946357018479e-06,
"loss": 0.1509,
"step": 365,
"student_loss": 0.006010106764733791,
"teacher_loss": 0.032394833862781525
},
{
"epoch": 0.10668973471741638,
"grad_norm": 6.6875,
"kd_loss": 0.1181640625,
"learning_rate": 7.277163899486975e-06,
"loss": 0.1623,
"step": 370,
"student_loss": 0.15845070779323578,
"teacher_loss": 0.0004756299313157797
},
{
"epoch": 0.10813148788927336,
"grad_norm": 4.625,
"kd_loss": 0.1240234375,
"learning_rate": 7.204784295188959e-06,
"loss": 0.1506,
"step": 375,
"student_loss": 0.10649572312831879,
"teacher_loss": 0.02242193929851055
},
{
"epoch": 0.10957324106113034,
"grad_norm": 3.4375,
"kd_loss": 0.1025390625,
"learning_rate": 7.1318265244212305e-06,
"loss": 0.1752,
"step": 380,
"student_loss": 0.00281524658203125,
"teacher_loss": 0.0015117195434868336
},
{
"epoch": 0.11101499423298732,
"grad_norm": 2.359375,
"kd_loss": 0.10205078125,
"learning_rate": 7.05830971909472e-06,
"loss": 0.1547,
"step": 385,
"student_loss": 0.0016335069667547941,
"teacher_loss": 0.0012311713071539998
},
{
"epoch": 0.11245674740484429,
"grad_norm": 4.125,
"kd_loss": 0.10205078125,
"learning_rate": 6.9842531577174865e-06,
"loss": 0.1538,
"step": 390,
"student_loss": 0.0012884392635896802,
"teacher_loss": 0.001418368425220251
},
{
"epoch": 0.11389850057670127,
"grad_norm": 5.1875,
"kd_loss": 0.1025390625,
"learning_rate": 6.9096762603392595e-06,
"loss": 0.1698,
"step": 395,
"student_loss": 0.0018499374855309725,
"teacher_loss": 0.0013690440682694316
},
{
"epoch": 0.11534025374855825,
"grad_norm": 6.53125,
"kd_loss": 0.10546875,
"learning_rate": 6.834598583458862e-06,
"loss": 0.16,
"step": 400,
"student_loss": 0.0014036521315574646,
"teacher_loss": 0.00040830764919519424
},
{
"epoch": 0.11678200692041522,
"grad_norm": 4.71875,
"kd_loss": 0.0888671875,
"learning_rate": 6.7590398148958625e-06,
"loss": 0.1718,
"step": 405,
"student_loss": 0.10261467099189758,
"teacher_loss": 0.0006754493806511164
},
{
"epoch": 0.1182237600922722,
"grad_norm": 4.3125,
"kd_loss": 0.171875,
"learning_rate": 6.6830197686277945e-06,
"loss": 0.1878,
"step": 410,
"student_loss": 0.4882833659648895,
"teacher_loss": 0.00981883890926838
},
{
"epoch": 0.11966551326412918,
"grad_norm": 2.8125,
"kd_loss": 0.1083984375,
"learning_rate": 6.6065583795942625e-06,
"loss": 0.182,
"step": 415,
"student_loss": 0.03729023039340973,
"teacher_loss": 0.0042837257497012615
},
{
"epoch": 0.12110726643598616,
"grad_norm": 4.65625,
"kd_loss": 0.1064453125,
"learning_rate": 6.52967569846937e-06,
"loss": 0.1607,
"step": 420,
"student_loss": 0.05881139263510704,
"teacher_loss": 0.024456653743982315
},
{
"epoch": 0.12254901960784313,
"grad_norm": 4.25,
"kd_loss": 0.10009765625,
"learning_rate": 6.452391886403767e-06,
"loss": 0.1674,
"step": 425,
"student_loss": 0.05037780851125717,
"teacher_loss": 0.0040146904066205025
},
{
"epoch": 0.12399077277970011,
"grad_norm": 4.75,
"kd_loss": 0.1748046875,
"learning_rate": 6.374727209737743e-06,
"loss": 0.1766,
"step": 430,
"student_loss": 0.00238221138715744,
"teacher_loss": 0.06439146399497986
},
{
"epoch": 0.1254325259515571,
"grad_norm": 5.53125,
"kd_loss": 0.1015625,
"learning_rate": 6.296702034686726e-06,
"loss": 0.1714,
"step": 435,
"student_loss": 0.002659996272996068,
"teacher_loss": 0.0022907655220478773
},
{
"epoch": 0.12687427912341406,
"grad_norm": 3.078125,
"kd_loss": 0.1845703125,
"learning_rate": 6.218336822000598e-06,
"loss": 0.1775,
"step": 440,
"student_loss": 0.46329638361930847,
"teacher_loss": 0.008188321255147457
},
{
"epoch": 0.12831603229527105,
"grad_norm": 3.953125,
"kd_loss": 0.1259765625,
"learning_rate": 6.139652121598219e-06,
"loss": 0.1769,
"step": 445,
"student_loss": 0.0006292742909863591,
"teacher_loss": 0.02016839198768139
},
{
"epoch": 0.12975778546712802,
"grad_norm": 3.53125,
"kd_loss": 0.10400390625,
"learning_rate": 6.060668567178561e-06,
"loss": 0.1663,
"step": 450,
"student_loss": 0.002717025112360716,
"teacher_loss": 0.0016874197172001004
},
{
"epoch": 0.131199538638985,
"grad_norm": 2.671875,
"kd_loss": 0.087890625,
"learning_rate": 5.981406870809889e-06,
"loss": 0.1748,
"step": 455,
"student_loss": 0.012300008907914162,
"teacher_loss": 0.0016890015685930848
},
{
"epoch": 0.13264129181084197,
"grad_norm": 5.1875,
"kd_loss": 0.10791015625,
"learning_rate": 5.9018878174983674e-06,
"loss": 0.17,
"step": 460,
"student_loss": 0.03240777552127838,
"teacher_loss": 0.0010722745209932327
},
{
"epoch": 0.13408304498269896,
"grad_norm": 2.765625,
"kd_loss": 0.1328125,
"learning_rate": 5.822132259737565e-06,
"loss": 0.1858,
"step": 465,
"student_loss": 0.0023128872271627188,
"teacher_loss": 0.0006816239329054952
},
{
"epoch": 0.13552479815455595,
"grad_norm": 5.3125,
"kd_loss": 0.09521484375,
"learning_rate": 5.742161112040237e-06,
"loss": 0.1887,
"step": 470,
"student_loss": 0.0013243159046396613,
"teacher_loss": 0.0008191668312065303
},
{
"epoch": 0.13696655132641292,
"grad_norm": 4.34375,
"kd_loss": 0.12890625,
"learning_rate": 5.661995345453867e-06,
"loss": 0.1479,
"step": 475,
"student_loss": 0.0022922754287719727,
"teacher_loss": 0.0007053640438243747
},
{
"epoch": 0.1384083044982699,
"grad_norm": 4.1875,
"kd_loss": 0.10888671875,
"learning_rate": 5.581655982061367e-06,
"loss": 0.2052,
"step": 480,
"student_loss": 0.016067378222942352,
"teacher_loss": 0.0467948317527771
},
{
"epoch": 0.13985005767012687,
"grad_norm": 4.625,
"kd_loss": 0.09521484375,
"learning_rate": 5.501164089468406e-06,
"loss": 0.1535,
"step": 485,
"student_loss": 0.001838831347413361,
"teacher_loss": 0.0018535954877734184
},
{
"epoch": 0.14129181084198386,
"grad_norm": 4.3125,
"kd_loss": 0.099609375,
"learning_rate": 5.4205407752787884e-06,
"loss": 0.1702,
"step": 490,
"student_loss": 0.0021060549188405275,
"teacher_loss": 0.0013811348471790552
},
{
"epoch": 0.14273356401384082,
"grad_norm": 5.0,
"kd_loss": 0.12109375,
"learning_rate": 5.339807181559359e-06,
"loss": 0.1698,
"step": 495,
"student_loss": 0.004670781549066305,
"teacher_loss": 0.0008499641553498805
},
{
"epoch": 0.14417531718569782,
"grad_norm": 7.28125,
"kd_loss": 0.1005859375,
"learning_rate": 5.258984479295853e-06,
"loss": 0.1663,
"step": 500,
"student_loss": 0.0009078571456484497,
"teacher_loss": 0.0008732817368581891
}
],
"logging_steps": 5,
"max_steps": 1000,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"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": 0.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}