trapoom555's picture
Upload checkpoints
8753a12
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.18140918656120747,
"eval_steps": 500,
"global_step": 2500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0007256367462448298,
"grad_norm": 45.75,
"learning_rate": 5e-06,
"loss": 1.9641,
"step": 10
},
{
"epoch": 0.0014512734924896596,
"grad_norm": 46.5,
"learning_rate": 1e-05,
"loss": 1.7641,
"step": 20
},
{
"epoch": 0.0021769102387344894,
"grad_norm": 35.75,
"learning_rate": 1.5e-05,
"loss": 1.4723,
"step": 30
},
{
"epoch": 0.0029025469849793192,
"grad_norm": 31.0,
"learning_rate": 2e-05,
"loss": 1.0985,
"step": 40
},
{
"epoch": 0.003628183731224149,
"grad_norm": 28.875,
"learning_rate": 2.5e-05,
"loss": 1.0381,
"step": 50
},
{
"epoch": 0.004353820477468979,
"grad_norm": 26.625,
"learning_rate": 3e-05,
"loss": 0.6977,
"step": 60
},
{
"epoch": 0.005079457223713809,
"grad_norm": 23.75,
"learning_rate": 3.5e-05,
"loss": 0.6392,
"step": 70
},
{
"epoch": 0.0058050939699586385,
"grad_norm": 26.5,
"learning_rate": 4e-05,
"loss": 0.7459,
"step": 80
},
{
"epoch": 0.006530730716203468,
"grad_norm": 29.875,
"learning_rate": 4.5e-05,
"loss": 0.6155,
"step": 90
},
{
"epoch": 0.007256367462448298,
"grad_norm": 21.5,
"learning_rate": 5e-05,
"loss": 0.3941,
"step": 100
},
{
"epoch": 0.007982004208693128,
"grad_norm": 17.0,
"learning_rate": 4.9999934086574596e-05,
"loss": 0.4513,
"step": 110
},
{
"epoch": 0.008707640954937958,
"grad_norm": 23.875,
"learning_rate": 4.9999736346645943e-05,
"loss": 0.5292,
"step": 120
},
{
"epoch": 0.009433277701182788,
"grad_norm": 17.25,
"learning_rate": 4.999940678125673e-05,
"loss": 0.3869,
"step": 130
},
{
"epoch": 0.010158914447427617,
"grad_norm": 24.75,
"learning_rate": 4.9998945392144796e-05,
"loss": 0.4121,
"step": 140
},
{
"epoch": 0.010884551193672447,
"grad_norm": 16.625,
"learning_rate": 4.999835218174307e-05,
"loss": 0.4199,
"step": 150
},
{
"epoch": 0.011610187939917277,
"grad_norm": 11.8125,
"learning_rate": 4.99976271531796e-05,
"loss": 0.4694,
"step": 160
},
{
"epoch": 0.012335824686162107,
"grad_norm": 15.9375,
"learning_rate": 4.9996770310277506e-05,
"loss": 0.3736,
"step": 170
},
{
"epoch": 0.013061461432406937,
"grad_norm": 18.0,
"learning_rate": 4.9995781657555e-05,
"loss": 0.4224,
"step": 180
},
{
"epoch": 0.013787098178651766,
"grad_norm": 23.0,
"learning_rate": 4.99946612002253e-05,
"loss": 0.4307,
"step": 190
},
{
"epoch": 0.014512734924896596,
"grad_norm": 19.375,
"learning_rate": 4.9993408944196676e-05,
"loss": 0.3449,
"step": 200
},
{
"epoch": 0.015238371671141426,
"grad_norm": 11.0,
"learning_rate": 4.9992024896072364e-05,
"loss": 0.3243,
"step": 210
},
{
"epoch": 0.015964008417386256,
"grad_norm": 17.875,
"learning_rate": 4.999050906315055e-05,
"loss": 0.3002,
"step": 220
},
{
"epoch": 0.016689645163631087,
"grad_norm": 6.875,
"learning_rate": 4.998886145342434e-05,
"loss": 0.3882,
"step": 230
},
{
"epoch": 0.017415281909875915,
"grad_norm": 12.125,
"learning_rate": 4.9987082075581684e-05,
"loss": 0.4211,
"step": 240
},
{
"epoch": 0.018140918656120747,
"grad_norm": 16.875,
"learning_rate": 4.9985170939005386e-05,
"loss": 0.2967,
"step": 250
},
{
"epoch": 0.018866555402365575,
"grad_norm": 26.125,
"learning_rate": 4.998312805377302e-05,
"loss": 0.3283,
"step": 260
},
{
"epoch": 0.019592192148610407,
"grad_norm": 18.0,
"learning_rate": 4.998095343065685e-05,
"loss": 0.3455,
"step": 270
},
{
"epoch": 0.020317828894855235,
"grad_norm": 15.4375,
"learning_rate": 4.997864708112384e-05,
"loss": 0.2891,
"step": 280
},
{
"epoch": 0.021043465641100066,
"grad_norm": 30.25,
"learning_rate": 4.997620901733554e-05,
"loss": 0.3645,
"step": 290
},
{
"epoch": 0.021769102387344894,
"grad_norm": 9.875,
"learning_rate": 4.997363925214803e-05,
"loss": 0.3047,
"step": 300
},
{
"epoch": 0.022494739133589726,
"grad_norm": 18.625,
"learning_rate": 4.9970937799111896e-05,
"loss": 0.2943,
"step": 310
},
{
"epoch": 0.023220375879834554,
"grad_norm": 30.625,
"learning_rate": 4.996810467247207e-05,
"loss": 0.3315,
"step": 320
},
{
"epoch": 0.023946012626079385,
"grad_norm": 14.3125,
"learning_rate": 4.9965139887167856e-05,
"loss": 0.2473,
"step": 330
},
{
"epoch": 0.024671649372324213,
"grad_norm": 21.5,
"learning_rate": 4.996204345883278e-05,
"loss": 0.3173,
"step": 340
},
{
"epoch": 0.025397286118569045,
"grad_norm": 16.125,
"learning_rate": 4.9958815403794546e-05,
"loss": 0.3813,
"step": 350
},
{
"epoch": 0.026122922864813873,
"grad_norm": 15.0625,
"learning_rate": 4.995545573907492e-05,
"loss": 0.3366,
"step": 360
},
{
"epoch": 0.026848559611058705,
"grad_norm": 19.5,
"learning_rate": 4.995196448238966e-05,
"loss": 0.3188,
"step": 370
},
{
"epoch": 0.027574196357303533,
"grad_norm": 20.125,
"learning_rate": 4.9948341652148436e-05,
"loss": 0.2776,
"step": 380
},
{
"epoch": 0.028299833103548364,
"grad_norm": 17.25,
"learning_rate": 4.994458726745468e-05,
"loss": 0.3574,
"step": 390
},
{
"epoch": 0.029025469849793192,
"grad_norm": 5.375,
"learning_rate": 4.9940701348105554e-05,
"loss": 0.2566,
"step": 400
},
{
"epoch": 0.029751106596038024,
"grad_norm": 31.0,
"learning_rate": 4.99366839145918e-05,
"loss": 0.275,
"step": 410
},
{
"epoch": 0.030476743342282852,
"grad_norm": 21.75,
"learning_rate": 4.993253498809762e-05,
"loss": 0.2928,
"step": 420
},
{
"epoch": 0.031202380088527683,
"grad_norm": 21.25,
"learning_rate": 4.9928254590500646e-05,
"loss": 0.3466,
"step": 430
},
{
"epoch": 0.03192801683477251,
"grad_norm": 15.25,
"learning_rate": 4.9923842744371707e-05,
"loss": 0.2911,
"step": 440
},
{
"epoch": 0.03265365358101734,
"grad_norm": 8.9375,
"learning_rate": 4.99192994729748e-05,
"loss": 0.2153,
"step": 450
},
{
"epoch": 0.033379290327262175,
"grad_norm": 20.5,
"learning_rate": 4.991462480026693e-05,
"loss": 0.383,
"step": 460
},
{
"epoch": 0.034104927073507,
"grad_norm": 6.46875,
"learning_rate": 4.9909818750898e-05,
"loss": 0.3099,
"step": 470
},
{
"epoch": 0.03483056381975183,
"grad_norm": 34.0,
"learning_rate": 4.990488135021065e-05,
"loss": 0.2897,
"step": 480
},
{
"epoch": 0.03555620056599666,
"grad_norm": 7.78125,
"learning_rate": 4.989981262424017e-05,
"loss": 0.3098,
"step": 490
},
{
"epoch": 0.036281837312241494,
"grad_norm": 41.0,
"learning_rate": 4.989461259971432e-05,
"loss": 0.2428,
"step": 500
},
{
"epoch": 0.03700747405848632,
"grad_norm": 17.375,
"learning_rate": 4.988928130405323e-05,
"loss": 0.2485,
"step": 510
},
{
"epoch": 0.03773311080473115,
"grad_norm": 15.625,
"learning_rate": 4.9883818765369194e-05,
"loss": 0.2694,
"step": 520
},
{
"epoch": 0.03845874755097598,
"grad_norm": 17.75,
"learning_rate": 4.98782250124666e-05,
"loss": 0.2752,
"step": 530
},
{
"epoch": 0.03918438429722081,
"grad_norm": 9.8125,
"learning_rate": 4.987250007484172e-05,
"loss": 0.2687,
"step": 540
},
{
"epoch": 0.03991002104346564,
"grad_norm": 3.6875,
"learning_rate": 4.986664398268256e-05,
"loss": 0.268,
"step": 550
},
{
"epoch": 0.04063565778971047,
"grad_norm": 24.125,
"learning_rate": 4.986065676686874e-05,
"loss": 0.2602,
"step": 560
},
{
"epoch": 0.0413612945359553,
"grad_norm": 23.125,
"learning_rate": 4.98545384589713e-05,
"loss": 0.2731,
"step": 570
},
{
"epoch": 0.04208693128220013,
"grad_norm": 13.5625,
"learning_rate": 4.984828909125251e-05,
"loss": 0.2537,
"step": 580
},
{
"epoch": 0.042812568028444964,
"grad_norm": 6.15625,
"learning_rate": 4.9841908696665764e-05,
"loss": 0.1723,
"step": 590
},
{
"epoch": 0.04353820477468979,
"grad_norm": 5.03125,
"learning_rate": 4.9835397308855344e-05,
"loss": 0.2414,
"step": 600
},
{
"epoch": 0.04426384152093462,
"grad_norm": 24.0,
"learning_rate": 4.9828754962156286e-05,
"loss": 0.2811,
"step": 610
},
{
"epoch": 0.04498947826717945,
"grad_norm": 25.875,
"learning_rate": 4.9821981691594175e-05,
"loss": 0.2949,
"step": 620
},
{
"epoch": 0.04571511501342428,
"grad_norm": 12.875,
"learning_rate": 4.981507753288497e-05,
"loss": 0.2662,
"step": 630
},
{
"epoch": 0.04644075175966911,
"grad_norm": 19.375,
"learning_rate": 4.9808042522434814e-05,
"loss": 0.2659,
"step": 640
},
{
"epoch": 0.04716638850591394,
"grad_norm": 22.25,
"learning_rate": 4.9800876697339824e-05,
"loss": 0.3333,
"step": 650
},
{
"epoch": 0.04789202525215877,
"grad_norm": 18.875,
"learning_rate": 4.979358009538594e-05,
"loss": 0.3197,
"step": 660
},
{
"epoch": 0.0486176619984036,
"grad_norm": 24.0,
"learning_rate": 4.978615275504869e-05,
"loss": 0.2385,
"step": 670
},
{
"epoch": 0.04934329874464843,
"grad_norm": 19.5,
"learning_rate": 4.977859471549297e-05,
"loss": 0.2673,
"step": 680
},
{
"epoch": 0.05006893549089326,
"grad_norm": 11.1875,
"learning_rate": 4.977090601657289e-05,
"loss": 0.2107,
"step": 690
},
{
"epoch": 0.05079457223713809,
"grad_norm": 9.8125,
"learning_rate": 4.976308669883153e-05,
"loss": 0.2825,
"step": 700
},
{
"epoch": 0.05152020898338292,
"grad_norm": 12.6875,
"learning_rate": 4.975513680350073e-05,
"loss": 0.2604,
"step": 710
},
{
"epoch": 0.052245845729627746,
"grad_norm": 19.375,
"learning_rate": 4.974705637250089e-05,
"loss": 0.209,
"step": 720
},
{
"epoch": 0.05297148247587258,
"grad_norm": 20.625,
"learning_rate": 4.97388454484407e-05,
"loss": 0.255,
"step": 730
},
{
"epoch": 0.05369711922211741,
"grad_norm": 17.25,
"learning_rate": 4.973050407461698e-05,
"loss": 0.2526,
"step": 740
},
{
"epoch": 0.05442275596836224,
"grad_norm": 3.453125,
"learning_rate": 4.972203229501441e-05,
"loss": 0.2028,
"step": 750
},
{
"epoch": 0.055148392714607065,
"grad_norm": 18.625,
"learning_rate": 4.971343015430532e-05,
"loss": 0.2395,
"step": 760
},
{
"epoch": 0.0558740294608519,
"grad_norm": 27.5,
"learning_rate": 4.970469769784941e-05,
"loss": 0.218,
"step": 770
},
{
"epoch": 0.05659966620709673,
"grad_norm": 24.375,
"learning_rate": 4.96958349716936e-05,
"loss": 0.2264,
"step": 780
},
{
"epoch": 0.05732530295334156,
"grad_norm": 27.125,
"learning_rate": 4.968684202257169e-05,
"loss": 0.3467,
"step": 790
},
{
"epoch": 0.058050939699586385,
"grad_norm": 18.0,
"learning_rate": 4.967771889790416e-05,
"loss": 0.2894,
"step": 800
},
{
"epoch": 0.058776576445831216,
"grad_norm": 8.375,
"learning_rate": 4.966846564579792e-05,
"loss": 0.1818,
"step": 810
},
{
"epoch": 0.05950221319207605,
"grad_norm": 13.75,
"learning_rate": 4.965908231504607e-05,
"loss": 0.3376,
"step": 820
},
{
"epoch": 0.06022784993832088,
"grad_norm": 53.5,
"learning_rate": 4.964956895512759e-05,
"loss": 0.2843,
"step": 830
},
{
"epoch": 0.060953486684565704,
"grad_norm": 12.3125,
"learning_rate": 4.963992561620714e-05,
"loss": 0.2396,
"step": 840
},
{
"epoch": 0.061679123430810535,
"grad_norm": 16.0,
"learning_rate": 4.963015234913475e-05,
"loss": 0.255,
"step": 850
},
{
"epoch": 0.06240476017705537,
"grad_norm": 6.8125,
"learning_rate": 4.96202492054456e-05,
"loss": 0.2316,
"step": 860
},
{
"epoch": 0.06313039692330019,
"grad_norm": 22.875,
"learning_rate": 4.9610216237359684e-05,
"loss": 0.3816,
"step": 870
},
{
"epoch": 0.06385603366954502,
"grad_norm": 23.875,
"learning_rate": 4.960005349778159e-05,
"loss": 0.2203,
"step": 880
},
{
"epoch": 0.06458167041578985,
"grad_norm": 38.0,
"learning_rate": 4.95897610403002e-05,
"loss": 0.3093,
"step": 890
},
{
"epoch": 0.06530730716203469,
"grad_norm": 19.75,
"learning_rate": 4.95793389191884e-05,
"loss": 0.2255,
"step": 900
},
{
"epoch": 0.06603294390827952,
"grad_norm": 16.5,
"learning_rate": 4.95687871894028e-05,
"loss": 0.1937,
"step": 910
},
{
"epoch": 0.06675858065452435,
"grad_norm": 10.0,
"learning_rate": 4.9558105906583466e-05,
"loss": 0.2242,
"step": 920
},
{
"epoch": 0.06748421740076918,
"grad_norm": 21.0,
"learning_rate": 4.9547295127053586e-05,
"loss": 0.3581,
"step": 930
},
{
"epoch": 0.068209854147014,
"grad_norm": 17.125,
"learning_rate": 4.95363549078192e-05,
"loss": 0.2132,
"step": 940
},
{
"epoch": 0.06893549089325883,
"grad_norm": 17.25,
"learning_rate": 4.952528530656889e-05,
"loss": 0.2659,
"step": 950
},
{
"epoch": 0.06966112763950366,
"grad_norm": 19.5,
"learning_rate": 4.9514086381673496e-05,
"loss": 0.2169,
"step": 960
},
{
"epoch": 0.0703867643857485,
"grad_norm": 20.625,
"learning_rate": 4.9502758192185774e-05,
"loss": 0.3276,
"step": 970
},
{
"epoch": 0.07111240113199332,
"grad_norm": 23.125,
"learning_rate": 4.949130079784009e-05,
"loss": 0.2512,
"step": 980
},
{
"epoch": 0.07183803787823816,
"grad_norm": 18.125,
"learning_rate": 4.9479714259052143e-05,
"loss": 0.2827,
"step": 990
},
{
"epoch": 0.07256367462448299,
"grad_norm": 18.375,
"learning_rate": 4.946799863691862e-05,
"loss": 0.206,
"step": 1000
},
{
"epoch": 0.07328931137072782,
"grad_norm": 11.3125,
"learning_rate": 4.9456153993216855e-05,
"loss": 0.309,
"step": 1010
},
{
"epoch": 0.07401494811697264,
"grad_norm": 6.21875,
"learning_rate": 4.944418039040453e-05,
"loss": 0.2311,
"step": 1020
},
{
"epoch": 0.07474058486321747,
"grad_norm": 12.5,
"learning_rate": 4.943207789161934e-05,
"loss": 0.191,
"step": 1030
},
{
"epoch": 0.0754662216094623,
"grad_norm": 9.75,
"learning_rate": 4.9419846560678657e-05,
"loss": 0.2881,
"step": 1040
},
{
"epoch": 0.07619185835570713,
"grad_norm": 22.25,
"learning_rate": 4.940748646207919e-05,
"loss": 0.3118,
"step": 1050
},
{
"epoch": 0.07691749510195196,
"grad_norm": 19.5,
"learning_rate": 4.9394997660996664e-05,
"loss": 0.2437,
"step": 1060
},
{
"epoch": 0.0776431318481968,
"grad_norm": 12.5625,
"learning_rate": 4.938238022328545e-05,
"loss": 0.2487,
"step": 1070
},
{
"epoch": 0.07836876859444163,
"grad_norm": 33.5,
"learning_rate": 4.9369634215478225e-05,
"loss": 0.2295,
"step": 1080
},
{
"epoch": 0.07909440534068646,
"grad_norm": 12.75,
"learning_rate": 4.935675970478563e-05,
"loss": 0.1983,
"step": 1090
},
{
"epoch": 0.07982004208693128,
"grad_norm": 17.75,
"learning_rate": 4.9343756759095924e-05,
"loss": 0.2072,
"step": 1100
},
{
"epoch": 0.08054567883317611,
"grad_norm": 23.875,
"learning_rate": 4.933062544697459e-05,
"loss": 0.2555,
"step": 1110
},
{
"epoch": 0.08127131557942094,
"grad_norm": 14.375,
"learning_rate": 4.931736583766402e-05,
"loss": 0.3138,
"step": 1120
},
{
"epoch": 0.08199695232566577,
"grad_norm": 17.875,
"learning_rate": 4.9303978001083105e-05,
"loss": 0.2075,
"step": 1130
},
{
"epoch": 0.0827225890719106,
"grad_norm": 26.625,
"learning_rate": 4.929046200782691e-05,
"loss": 0.2282,
"step": 1140
},
{
"epoch": 0.08344822581815543,
"grad_norm": 18.625,
"learning_rate": 4.927681792916626e-05,
"loss": 0.395,
"step": 1150
},
{
"epoch": 0.08417386256440026,
"grad_norm": 8.1875,
"learning_rate": 4.926304583704739e-05,
"loss": 0.2529,
"step": 1160
},
{
"epoch": 0.0848994993106451,
"grad_norm": 20.125,
"learning_rate": 4.924914580409157e-05,
"loss": 0.3177,
"step": 1170
},
{
"epoch": 0.08562513605688993,
"grad_norm": 12.0625,
"learning_rate": 4.9235117903594696e-05,
"loss": 0.2536,
"step": 1180
},
{
"epoch": 0.08635077280313475,
"grad_norm": 25.125,
"learning_rate": 4.922096220952692e-05,
"loss": 0.2746,
"step": 1190
},
{
"epoch": 0.08707640954937958,
"grad_norm": 25.25,
"learning_rate": 4.9206678796532285e-05,
"loss": 0.2342,
"step": 1200
},
{
"epoch": 0.08780204629562441,
"grad_norm": 18.375,
"learning_rate": 4.919226773992826e-05,
"loss": 0.2721,
"step": 1210
},
{
"epoch": 0.08852768304186924,
"grad_norm": 18.75,
"learning_rate": 4.917772911570544e-05,
"loss": 0.1886,
"step": 1220
},
{
"epoch": 0.08925331978811407,
"grad_norm": 9.875,
"learning_rate": 4.916306300052704e-05,
"loss": 0.2213,
"step": 1230
},
{
"epoch": 0.0899789565343589,
"grad_norm": 9.0,
"learning_rate": 4.91482694717286e-05,
"loss": 0.3408,
"step": 1240
},
{
"epoch": 0.09070459328060373,
"grad_norm": 15.625,
"learning_rate": 4.9133348607317464e-05,
"loss": 0.2389,
"step": 1250
},
{
"epoch": 0.09143023002684857,
"grad_norm": 14.125,
"learning_rate": 4.911830048597247e-05,
"loss": 0.235,
"step": 1260
},
{
"epoch": 0.09215586677309338,
"grad_norm": 11.5,
"learning_rate": 4.910312518704348e-05,
"loss": 0.2105,
"step": 1270
},
{
"epoch": 0.09288150351933822,
"grad_norm": 20.5,
"learning_rate": 4.908782279055095e-05,
"loss": 0.2124,
"step": 1280
},
{
"epoch": 0.09360714026558305,
"grad_norm": 23.125,
"learning_rate": 4.907239337718557e-05,
"loss": 0.2684,
"step": 1290
},
{
"epoch": 0.09433277701182788,
"grad_norm": 15.0625,
"learning_rate": 4.9056837028307755e-05,
"loss": 0.2318,
"step": 1300
},
{
"epoch": 0.09505841375807271,
"grad_norm": 11.6875,
"learning_rate": 4.9041153825947314e-05,
"loss": 0.1496,
"step": 1310
},
{
"epoch": 0.09578405050431754,
"grad_norm": 17.875,
"learning_rate": 4.902534385280292e-05,
"loss": 0.2575,
"step": 1320
},
{
"epoch": 0.09650968725056237,
"grad_norm": 26.25,
"learning_rate": 4.900940719224173e-05,
"loss": 0.3376,
"step": 1330
},
{
"epoch": 0.0972353239968072,
"grad_norm": 32.5,
"learning_rate": 4.899334392829893e-05,
"loss": 0.3725,
"step": 1340
},
{
"epoch": 0.09796096074305202,
"grad_norm": 29.75,
"learning_rate": 4.897715414567732e-05,
"loss": 0.3137,
"step": 1350
},
{
"epoch": 0.09868659748929685,
"grad_norm": 26.125,
"learning_rate": 4.8960837929746794e-05,
"loss": 0.2413,
"step": 1360
},
{
"epoch": 0.09941223423554169,
"grad_norm": 6.03125,
"learning_rate": 4.894439536654399e-05,
"loss": 0.2918,
"step": 1370
},
{
"epoch": 0.10013787098178652,
"grad_norm": 24.125,
"learning_rate": 4.892782654277175e-05,
"loss": 0.2599,
"step": 1380
},
{
"epoch": 0.10086350772803135,
"grad_norm": 8.4375,
"learning_rate": 4.8911131545798715e-05,
"loss": 0.2697,
"step": 1390
},
{
"epoch": 0.10158914447427618,
"grad_norm": 14.5,
"learning_rate": 4.889431046365883e-05,
"loss": 0.1779,
"step": 1400
},
{
"epoch": 0.10231478122052101,
"grad_norm": 18.25,
"learning_rate": 4.887736338505092e-05,
"loss": 0.244,
"step": 1410
},
{
"epoch": 0.10304041796676584,
"grad_norm": 8.625,
"learning_rate": 4.886029039933817e-05,
"loss": 0.2245,
"step": 1420
},
{
"epoch": 0.10376605471301066,
"grad_norm": 7.0,
"learning_rate": 4.884309159654771e-05,
"loss": 0.291,
"step": 1430
},
{
"epoch": 0.10449169145925549,
"grad_norm": 20.375,
"learning_rate": 4.8825767067370086e-05,
"loss": 0.2257,
"step": 1440
},
{
"epoch": 0.10521732820550032,
"grad_norm": 16.125,
"learning_rate": 4.880831690315884e-05,
"loss": 0.1979,
"step": 1450
},
{
"epoch": 0.10594296495174516,
"grad_norm": 21.0,
"learning_rate": 4.879074119592998e-05,
"loss": 0.3273,
"step": 1460
},
{
"epoch": 0.10666860169798999,
"grad_norm": 23.875,
"learning_rate": 4.877304003836149e-05,
"loss": 0.2113,
"step": 1470
},
{
"epoch": 0.10739423844423482,
"grad_norm": 7.125,
"learning_rate": 4.87552135237929e-05,
"loss": 0.2496,
"step": 1480
},
{
"epoch": 0.10811987519047965,
"grad_norm": 17.375,
"learning_rate": 4.873726174622475e-05,
"loss": 0.3294,
"step": 1490
},
{
"epoch": 0.10884551193672448,
"grad_norm": 6.6875,
"learning_rate": 4.8719184800318066e-05,
"loss": 0.2847,
"step": 1500
},
{
"epoch": 0.1095711486829693,
"grad_norm": 19.0,
"learning_rate": 4.870098278139394e-05,
"loss": 0.2599,
"step": 1510
},
{
"epoch": 0.11029678542921413,
"grad_norm": 25.625,
"learning_rate": 4.8682655785432975e-05,
"loss": 0.2724,
"step": 1520
},
{
"epoch": 0.11102242217545896,
"grad_norm": 8.75,
"learning_rate": 4.866420390907476e-05,
"loss": 0.2099,
"step": 1530
},
{
"epoch": 0.1117480589217038,
"grad_norm": 9.1875,
"learning_rate": 4.8645627249617406e-05,
"loss": 0.2092,
"step": 1540
},
{
"epoch": 0.11247369566794863,
"grad_norm": 16.625,
"learning_rate": 4.8626925905017017e-05,
"loss": 0.149,
"step": 1550
},
{
"epoch": 0.11319933241419346,
"grad_norm": 24.375,
"learning_rate": 4.8608099973887167e-05,
"loss": 0.2192,
"step": 1560
},
{
"epoch": 0.11392496916043829,
"grad_norm": 10.25,
"learning_rate": 4.858914955549839e-05,
"loss": 0.1901,
"step": 1570
},
{
"epoch": 0.11465060590668312,
"grad_norm": 25.0,
"learning_rate": 4.857007474977764e-05,
"loss": 0.3054,
"step": 1580
},
{
"epoch": 0.11537624265292794,
"grad_norm": 6.21875,
"learning_rate": 4.855087565730778e-05,
"loss": 0.2321,
"step": 1590
},
{
"epoch": 0.11610187939917277,
"grad_norm": 18.25,
"learning_rate": 4.853155237932705e-05,
"loss": 0.2466,
"step": 1600
},
{
"epoch": 0.1168275161454176,
"grad_norm": 15.5,
"learning_rate": 4.8512105017728514e-05,
"loss": 0.2657,
"step": 1610
},
{
"epoch": 0.11755315289166243,
"grad_norm": 15.0625,
"learning_rate": 4.8492533675059564e-05,
"loss": 0.2376,
"step": 1620
},
{
"epoch": 0.11827878963790726,
"grad_norm": 26.75,
"learning_rate": 4.847283845452133e-05,
"loss": 0.2644,
"step": 1630
},
{
"epoch": 0.1190044263841521,
"grad_norm": 25.375,
"learning_rate": 4.8453019459968174e-05,
"loss": 0.2335,
"step": 1640
},
{
"epoch": 0.11973006313039693,
"grad_norm": 23.75,
"learning_rate": 4.843307679590711e-05,
"loss": 0.2546,
"step": 1650
},
{
"epoch": 0.12045569987664176,
"grad_norm": 5.84375,
"learning_rate": 4.8413010567497295e-05,
"loss": 0.16,
"step": 1660
},
{
"epoch": 0.12118133662288658,
"grad_norm": 6.09375,
"learning_rate": 4.8392820880549434e-05,
"loss": 0.1713,
"step": 1670
},
{
"epoch": 0.12190697336913141,
"grad_norm": 11.5625,
"learning_rate": 4.837250784152524e-05,
"loss": 0.3301,
"step": 1680
},
{
"epoch": 0.12263261011537624,
"grad_norm": 13.0,
"learning_rate": 4.835207155753686e-05,
"loss": 0.2185,
"step": 1690
},
{
"epoch": 0.12335824686162107,
"grad_norm": 12.5625,
"learning_rate": 4.833151213634635e-05,
"loss": 0.1979,
"step": 1700
},
{
"epoch": 0.1240838836078659,
"grad_norm": 10.5,
"learning_rate": 4.831082968636505e-05,
"loss": 0.1427,
"step": 1710
},
{
"epoch": 0.12480952035411073,
"grad_norm": 4.25,
"learning_rate": 4.8290024316653055e-05,
"loss": 0.1427,
"step": 1720
},
{
"epoch": 0.12553515710035557,
"grad_norm": 15.25,
"learning_rate": 4.826909613691861e-05,
"loss": 0.1467,
"step": 1730
},
{
"epoch": 0.12626079384660038,
"grad_norm": 15.875,
"learning_rate": 4.824804525751757e-05,
"loss": 0.2006,
"step": 1740
},
{
"epoch": 0.12698643059284523,
"grad_norm": 13.1875,
"learning_rate": 4.822687178945278e-05,
"loss": 0.2795,
"step": 1750
},
{
"epoch": 0.12771206733909005,
"grad_norm": 18.625,
"learning_rate": 4.820557584437349e-05,
"loss": 0.2728,
"step": 1760
},
{
"epoch": 0.1284377040853349,
"grad_norm": 28.875,
"learning_rate": 4.8184157534574806e-05,
"loss": 0.277,
"step": 1770
},
{
"epoch": 0.1291633408315797,
"grad_norm": 15.875,
"learning_rate": 4.816261697299706e-05,
"loss": 0.2371,
"step": 1780
},
{
"epoch": 0.12988897757782455,
"grad_norm": 17.125,
"learning_rate": 4.814095427322522e-05,
"loss": 0.2783,
"step": 1790
},
{
"epoch": 0.13061461432406937,
"grad_norm": 14.0,
"learning_rate": 4.811916954948832e-05,
"loss": 0.2762,
"step": 1800
},
{
"epoch": 0.1313402510703142,
"grad_norm": 15.8125,
"learning_rate": 4.809726291665881e-05,
"loss": 0.2073,
"step": 1810
},
{
"epoch": 0.13206588781655904,
"grad_norm": 26.375,
"learning_rate": 4.8075234490251994e-05,
"loss": 0.1943,
"step": 1820
},
{
"epoch": 0.13279152456280385,
"grad_norm": 12.1875,
"learning_rate": 4.805308438642539e-05,
"loss": 0.25,
"step": 1830
},
{
"epoch": 0.1335171613090487,
"grad_norm": 9.0625,
"learning_rate": 4.8030812721978135e-05,
"loss": 0.2351,
"step": 1840
},
{
"epoch": 0.13424279805529352,
"grad_norm": 30.25,
"learning_rate": 4.800841961435037e-05,
"loss": 0.2059,
"step": 1850
},
{
"epoch": 0.13496843480153836,
"grad_norm": 30.75,
"learning_rate": 4.7985905181622605e-05,
"loss": 0.1975,
"step": 1860
},
{
"epoch": 0.13569407154778318,
"grad_norm": 5.0,
"learning_rate": 4.7963269542515106e-05,
"loss": 0.1739,
"step": 1870
},
{
"epoch": 0.136419708294028,
"grad_norm": 12.4375,
"learning_rate": 4.794051281638729e-05,
"loss": 0.234,
"step": 1880
},
{
"epoch": 0.13714534504027284,
"grad_norm": 13.625,
"learning_rate": 4.7917635123237025e-05,
"loss": 0.2803,
"step": 1890
},
{
"epoch": 0.13787098178651766,
"grad_norm": 8.375,
"learning_rate": 4.789463658370013e-05,
"loss": 0.1942,
"step": 1900
},
{
"epoch": 0.1385966185327625,
"grad_norm": 20.5,
"learning_rate": 4.7871517319049566e-05,
"loss": 0.2595,
"step": 1910
},
{
"epoch": 0.13932225527900732,
"grad_norm": 16.375,
"learning_rate": 4.784827745119493e-05,
"loss": 0.2778,
"step": 1920
},
{
"epoch": 0.14004789202525217,
"grad_norm": 2.875,
"learning_rate": 4.7824917102681796e-05,
"loss": 0.2168,
"step": 1930
},
{
"epoch": 0.140773528771497,
"grad_norm": 16.5,
"learning_rate": 4.780143639669098e-05,
"loss": 0.249,
"step": 1940
},
{
"epoch": 0.14149916551774183,
"grad_norm": 14.8125,
"learning_rate": 4.7777835457038e-05,
"loss": 0.2158,
"step": 1950
},
{
"epoch": 0.14222480226398665,
"grad_norm": 15.0625,
"learning_rate": 4.775411440817236e-05,
"loss": 0.2566,
"step": 1960
},
{
"epoch": 0.14295043901023147,
"grad_norm": 14.6875,
"learning_rate": 4.7730273375176885e-05,
"loss": 0.2448,
"step": 1970
},
{
"epoch": 0.1436760757564763,
"grad_norm": 20.5,
"learning_rate": 4.770631248376713e-05,
"loss": 0.2312,
"step": 1980
},
{
"epoch": 0.14440171250272113,
"grad_norm": 21.0,
"learning_rate": 4.768223186029064e-05,
"loss": 0.3233,
"step": 1990
},
{
"epoch": 0.14512734924896598,
"grad_norm": 28.875,
"learning_rate": 4.765803163172632e-05,
"loss": 0.3363,
"step": 2000
},
{
"epoch": 0.1458529859952108,
"grad_norm": 14.3125,
"learning_rate": 4.7633711925683786e-05,
"loss": 0.1564,
"step": 2010
},
{
"epoch": 0.14657862274145564,
"grad_norm": 18.5,
"learning_rate": 4.760927287040262e-05,
"loss": 0.2202,
"step": 2020
},
{
"epoch": 0.14730425948770046,
"grad_norm": 12.1875,
"learning_rate": 4.75847145947518e-05,
"loss": 0.1876,
"step": 2030
},
{
"epoch": 0.14802989623394527,
"grad_norm": 19.75,
"learning_rate": 4.756003722822891e-05,
"loss": 0.2376,
"step": 2040
},
{
"epoch": 0.14875553298019012,
"grad_norm": 16.375,
"learning_rate": 4.753524090095953e-05,
"loss": 0.2729,
"step": 2050
},
{
"epoch": 0.14948116972643494,
"grad_norm": 15.0625,
"learning_rate": 4.751032574369655e-05,
"loss": 0.2593,
"step": 2060
},
{
"epoch": 0.15020680647267978,
"grad_norm": 5.5625,
"learning_rate": 4.748529188781942e-05,
"loss": 0.2139,
"step": 2070
},
{
"epoch": 0.1509324432189246,
"grad_norm": 7.53125,
"learning_rate": 4.746013946533352e-05,
"loss": 0.1892,
"step": 2080
},
{
"epoch": 0.15165807996516945,
"grad_norm": 16.5,
"learning_rate": 4.7434868608869435e-05,
"loss": 0.2272,
"step": 2090
},
{
"epoch": 0.15238371671141426,
"grad_norm": 6.5,
"learning_rate": 4.740947945168227e-05,
"loss": 0.199,
"step": 2100
},
{
"epoch": 0.1531093534576591,
"grad_norm": 18.875,
"learning_rate": 4.7383972127650924e-05,
"loss": 0.2047,
"step": 2110
},
{
"epoch": 0.15383499020390393,
"grad_norm": 12.125,
"learning_rate": 4.73583467712774e-05,
"loss": 0.1846,
"step": 2120
},
{
"epoch": 0.15456062695014874,
"grad_norm": 17.75,
"learning_rate": 4.733260351768611e-05,
"loss": 0.2217,
"step": 2130
},
{
"epoch": 0.1552862636963936,
"grad_norm": 12.0625,
"learning_rate": 4.7306742502623124e-05,
"loss": 0.2027,
"step": 2140
},
{
"epoch": 0.1560119004426384,
"grad_norm": 13.375,
"learning_rate": 4.72807638624555e-05,
"loss": 0.1647,
"step": 2150
},
{
"epoch": 0.15673753718888325,
"grad_norm": 5.40625,
"learning_rate": 4.7254667734170526e-05,
"loss": 0.2009,
"step": 2160
},
{
"epoch": 0.15746317393512807,
"grad_norm": 8.3125,
"learning_rate": 4.7228454255375024e-05,
"loss": 0.3072,
"step": 2170
},
{
"epoch": 0.15818881068137292,
"grad_norm": 28.0,
"learning_rate": 4.72021235642946e-05,
"loss": 0.2587,
"step": 2180
},
{
"epoch": 0.15891444742761773,
"grad_norm": 10.8125,
"learning_rate": 4.7175675799772935e-05,
"loss": 0.1726,
"step": 2190
},
{
"epoch": 0.15964008417386255,
"grad_norm": 18.375,
"learning_rate": 4.714911110127106e-05,
"loss": 0.1847,
"step": 2200
},
{
"epoch": 0.1603657209201074,
"grad_norm": 5.09375,
"learning_rate": 4.712242960886659e-05,
"loss": 0.2591,
"step": 2210
},
{
"epoch": 0.16109135766635221,
"grad_norm": 3.640625,
"learning_rate": 4.709563146325301e-05,
"loss": 0.147,
"step": 2220
},
{
"epoch": 0.16181699441259706,
"grad_norm": 17.25,
"learning_rate": 4.7068716805738935e-05,
"loss": 0.2291,
"step": 2230
},
{
"epoch": 0.16254263115884188,
"grad_norm": 14.375,
"learning_rate": 4.704168577824733e-05,
"loss": 0.2156,
"step": 2240
},
{
"epoch": 0.16326826790508672,
"grad_norm": 17.875,
"learning_rate": 4.701453852331482e-05,
"loss": 0.2056,
"step": 2250
},
{
"epoch": 0.16399390465133154,
"grad_norm": 18.125,
"learning_rate": 4.6987275184090876e-05,
"loss": 0.2112,
"step": 2260
},
{
"epoch": 0.16471954139757639,
"grad_norm": 23.25,
"learning_rate": 4.6959895904337115e-05,
"loss": 0.1982,
"step": 2270
},
{
"epoch": 0.1654451781438212,
"grad_norm": 13.5625,
"learning_rate": 4.6932400828426505e-05,
"loss": 0.1638,
"step": 2280
},
{
"epoch": 0.16617081489006602,
"grad_norm": 7.15625,
"learning_rate": 4.690479010134261e-05,
"loss": 0.2096,
"step": 2290
},
{
"epoch": 0.16689645163631087,
"grad_norm": 15.0,
"learning_rate": 4.687706386867885e-05,
"loss": 0.2676,
"step": 2300
},
{
"epoch": 0.16762208838255568,
"grad_norm": 9.5,
"learning_rate": 4.6849222276637686e-05,
"loss": 0.2291,
"step": 2310
},
{
"epoch": 0.16834772512880053,
"grad_norm": 34.75,
"learning_rate": 4.682126547202991e-05,
"loss": 0.2573,
"step": 2320
},
{
"epoch": 0.16907336187504535,
"grad_norm": 23.75,
"learning_rate": 4.679319360227381e-05,
"loss": 0.1805,
"step": 2330
},
{
"epoch": 0.1697989986212902,
"grad_norm": 3.6875,
"learning_rate": 4.676500681539444e-05,
"loss": 0.1618,
"step": 2340
},
{
"epoch": 0.170524635367535,
"grad_norm": 10.4375,
"learning_rate": 4.673670526002281e-05,
"loss": 0.3,
"step": 2350
},
{
"epoch": 0.17125027211377986,
"grad_norm": 5.28125,
"learning_rate": 4.6708289085395115e-05,
"loss": 0.2412,
"step": 2360
},
{
"epoch": 0.17197590886002467,
"grad_norm": 25.375,
"learning_rate": 4.6679758441351964e-05,
"loss": 0.1359,
"step": 2370
},
{
"epoch": 0.1727015456062695,
"grad_norm": 17.0,
"learning_rate": 4.665111347833754e-05,
"loss": 0.2926,
"step": 2380
},
{
"epoch": 0.17342718235251434,
"grad_norm": 10.375,
"learning_rate": 4.662235434739885e-05,
"loss": 0.1706,
"step": 2390
},
{
"epoch": 0.17415281909875915,
"grad_norm": 13.0,
"learning_rate": 4.659348120018493e-05,
"loss": 0.2141,
"step": 2400
},
{
"epoch": 0.174878455845004,
"grad_norm": 3.25,
"learning_rate": 4.6564494188946015e-05,
"loss": 0.2116,
"step": 2410
},
{
"epoch": 0.17560409259124882,
"grad_norm": 33.0,
"learning_rate": 4.653539346653278e-05,
"loss": 0.1661,
"step": 2420
},
{
"epoch": 0.17632972933749366,
"grad_norm": 28.125,
"learning_rate": 4.650617918639546e-05,
"loss": 0.2478,
"step": 2430
},
{
"epoch": 0.17705536608373848,
"grad_norm": 7.375,
"learning_rate": 4.647685150258314e-05,
"loss": 0.2043,
"step": 2440
},
{
"epoch": 0.1777810028299833,
"grad_norm": 22.125,
"learning_rate": 4.644741056974287e-05,
"loss": 0.2602,
"step": 2450
},
{
"epoch": 0.17850663957622814,
"grad_norm": 7.78125,
"learning_rate": 4.641785654311884e-05,
"loss": 0.1906,
"step": 2460
},
{
"epoch": 0.17923227632247296,
"grad_norm": 20.25,
"learning_rate": 4.6388189578551646e-05,
"loss": 0.2865,
"step": 2470
},
{
"epoch": 0.1799579130687178,
"grad_norm": 8.9375,
"learning_rate": 4.635840983247738e-05,
"loss": 0.2336,
"step": 2480
},
{
"epoch": 0.18068354981496262,
"grad_norm": 17.875,
"learning_rate": 4.632851746192685e-05,
"loss": 0.2388,
"step": 2490
},
{
"epoch": 0.18140918656120747,
"grad_norm": 28.0,
"learning_rate": 4.6298512624524734e-05,
"loss": 0.1824,
"step": 2500
}
],
"logging_steps": 10,
"max_steps": 13781,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"total_flos": 0.0,
"train_batch_size": 5,
"trial_name": null,
"trial_params": null
}