{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.998264893001735,
"eval_steps": 500,
"global_step": 1944,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01542317331791016,
"grad_norm": 0.06286442279815674,
"learning_rate": 5e-06,
"loss": 2.7232,
"step": 10
},
{
"epoch": 0.03084634663582032,
"grad_norm": 0.06427137553691864,
"learning_rate": 1e-05,
"loss": 2.8025,
"step": 20
},
{
"epoch": 0.04626951995373048,
"grad_norm": 0.17838376760482788,
"learning_rate": 1.5e-05,
"loss": 2.7466,
"step": 30
},
{
"epoch": 0.06169269327164064,
"grad_norm": 0.07409191876649857,
"learning_rate": 2e-05,
"loss": 2.7862,
"step": 40
},
{
"epoch": 0.0771158665895508,
"grad_norm": 0.08752360939979553,
"learning_rate": 2.5e-05,
"loss": 2.7836,
"step": 50
},
{
"epoch": 0.09253903990746096,
"grad_norm": 0.11069338023662567,
"learning_rate": 3e-05,
"loss": 2.7271,
"step": 60
},
{
"epoch": 0.10796221322537113,
"grad_norm": 0.1447472721338272,
"learning_rate": 3.5e-05,
"loss": 2.7783,
"step": 70
},
{
"epoch": 0.12338538654328128,
"grad_norm": 0.12579452991485596,
"learning_rate": 4e-05,
"loss": 2.7324,
"step": 80
},
{
"epoch": 0.13880855986119145,
"grad_norm": 0.16387595236301422,
"learning_rate": 4.5e-05,
"loss": 2.7053,
"step": 90
},
{
"epoch": 0.1542317331791016,
"grad_norm": 0.19016346335411072,
"learning_rate": 5e-05,
"loss": 2.5806,
"step": 100
},
{
"epoch": 0.16965490649701176,
"grad_norm": 0.21213597059249878,
"learning_rate": 5.500000000000001e-05,
"loss": 2.7242,
"step": 110
},
{
"epoch": 0.18507807981492191,
"grad_norm": 0.1880221664905548,
"learning_rate": 6e-05,
"loss": 2.6331,
"step": 120
},
{
"epoch": 0.20050125313283207,
"grad_norm": 0.7994162440299988,
"learning_rate": 6.500000000000001e-05,
"loss": 2.508,
"step": 130
},
{
"epoch": 0.21592442645074225,
"grad_norm": 0.48757001757621765,
"learning_rate": 7e-05,
"loss": 2.4189,
"step": 140
},
{
"epoch": 0.2313475997686524,
"grad_norm": 0.6417515277862549,
"learning_rate": 7.500000000000001e-05,
"loss": 2.351,
"step": 150
},
{
"epoch": 0.24677077308656256,
"grad_norm": 0.45560503005981445,
"learning_rate": 8e-05,
"loss": 2.2376,
"step": 160
},
{
"epoch": 0.26219394640447274,
"grad_norm": 0.43604227900505066,
"learning_rate": 8.5e-05,
"loss": 2.2678,
"step": 170
},
{
"epoch": 0.2776171197223829,
"grad_norm": 0.3299252986907959,
"learning_rate": 9e-05,
"loss": 2.2033,
"step": 180
},
{
"epoch": 0.29304029304029305,
"grad_norm": 0.3385520279407501,
"learning_rate": 9.5e-05,
"loss": 2.126,
"step": 190
},
{
"epoch": 0.3084634663582032,
"grad_norm": 0.380649209022522,
"learning_rate": 0.0001,
"loss": 2.4127,
"step": 200
},
{
"epoch": 0.32388663967611336,
"grad_norm": 0.4573318064212799,
"learning_rate": 9.999188786725007e-05,
"loss": 2.0718,
"step": 210
},
{
"epoch": 0.3393098129940235,
"grad_norm": 0.5013952255249023,
"learning_rate": 9.996755410126815e-05,
"loss": 2.2014,
"step": 220
},
{
"epoch": 0.3547329863119337,
"grad_norm": 0.3462372422218323,
"learning_rate": 9.992700659800388e-05,
"loss": 2.0259,
"step": 230
},
{
"epoch": 0.37015615962984383,
"grad_norm": 0.3076546788215637,
"learning_rate": 9.987025851452639e-05,
"loss": 2.3585,
"step": 240
},
{
"epoch": 0.385579332947754,
"grad_norm": 0.36271095275878906,
"learning_rate": 9.979732826475515e-05,
"loss": 2.2215,
"step": 250
},
{
"epoch": 0.40100250626566414,
"grad_norm": 0.3243629038333893,
"learning_rate": 9.970823951348487e-05,
"loss": 2.3239,
"step": 260
},
{
"epoch": 0.4164256795835743,
"grad_norm": 0.3210390508174896,
"learning_rate": 9.960302116870661e-05,
"loss": 2.2053,
"step": 270
},
{
"epoch": 0.4318488529014845,
"grad_norm": 0.5260313749313354,
"learning_rate": 9.948170737222762e-05,
"loss": 2.2688,
"step": 280
},
{
"epoch": 0.44727202621939466,
"grad_norm": 0.3700290024280548,
"learning_rate": 9.934433748859274e-05,
"loss": 2.2533,
"step": 290
},
{
"epoch": 0.4626951995373048,
"grad_norm": 0.4328364431858063,
"learning_rate": 9.919095609231126e-05,
"loss": 2.1101,
"step": 300
},
{
"epoch": 0.47811837285521497,
"grad_norm": 0.3829161524772644,
"learning_rate": 9.902161295339307e-05,
"loss": 2.1293,
"step": 310
},
{
"epoch": 0.4935415461731251,
"grad_norm": 0.43613630533218384,
"learning_rate": 9.883636302119912e-05,
"loss": 2.0417,
"step": 320
},
{
"epoch": 0.5089647194910353,
"grad_norm": 0.3980802595615387,
"learning_rate": 9.863526640661107e-05,
"loss": 2.1096,
"step": 330
},
{
"epoch": 0.5243878928089455,
"grad_norm": 0.44878003001213074,
"learning_rate": 9.841838836252627e-05,
"loss": 2.2787,
"step": 340
},
{
"epoch": 0.5398110661268556,
"grad_norm": 0.3995887339115143,
"learning_rate": 9.818579926268405e-05,
"loss": 2.3788,
"step": 350
},
{
"epoch": 0.5552342394447658,
"grad_norm": 0.3169827461242676,
"learning_rate": 9.793757457883062e-05,
"loss": 2.1264,
"step": 360
},
{
"epoch": 0.5706574127626759,
"grad_norm": 0.28045836091041565,
"learning_rate": 9.767379485622943e-05,
"loss": 2.2351,
"step": 370
},
{
"epoch": 0.5860805860805861,
"grad_norm": 0.3078824281692505,
"learning_rate": 9.739454568752556e-05,
"loss": 2.1219,
"step": 380
},
{
"epoch": 0.6015037593984962,
"grad_norm": 2.457259178161621,
"learning_rate": 9.709991768497208e-05,
"loss": 2.0565,
"step": 390
},
{
"epoch": 0.6169269327164064,
"grad_norm": 0.3353936970233917,
"learning_rate": 9.679000645102771e-05,
"loss": 2.2406,
"step": 400
},
{
"epoch": 0.6323501060343165,
"grad_norm": 0.4184532165527344,
"learning_rate": 9.646491254733532e-05,
"loss": 2.154,
"step": 410
},
{
"epoch": 0.6477732793522267,
"grad_norm": 0.4148164987564087,
"learning_rate": 9.612474146209096e-05,
"loss": 2.1421,
"step": 420
},
{
"epoch": 0.6631964526701368,
"grad_norm": 0.3182518780231476,
"learning_rate": 9.576960357581475e-05,
"loss": 2.1537,
"step": 430
},
{
"epoch": 0.678619625988047,
"grad_norm": 0.3275531828403473,
"learning_rate": 9.539961412553375e-05,
"loss": 2.1743,
"step": 440
},
{
"epoch": 0.6940427993059572,
"grad_norm": 0.31316521763801575,
"learning_rate": 9.501489316738945e-05,
"loss": 2.0946,
"step": 450
},
{
"epoch": 0.7094659726238673,
"grad_norm": 0.3164524734020233,
"learning_rate": 9.461556553768123e-05,
"loss": 2.1048,
"step": 460
},
{
"epoch": 0.7248891459417776,
"grad_norm": 0.29031479358673096,
"learning_rate": 9.420176081235881e-05,
"loss": 2.1311,
"step": 470
},
{
"epoch": 0.7403123192596877,
"grad_norm": 0.3423845171928406,
"learning_rate": 9.377361326497674e-05,
"loss": 2.3041,
"step": 480
},
{
"epoch": 0.7557354925775979,
"grad_norm": 0.4182330071926117,
"learning_rate": 9.333126182312465e-05,
"loss": 2.0653,
"step": 490
},
{
"epoch": 0.771158665895508,
"grad_norm": 0.3934147357940674,
"learning_rate": 9.287485002334733e-05,
"loss": 1.9974,
"step": 500
},
{
"epoch": 0.771158665895508,
"eval_loss": 2.0961015224456787,
"eval_runtime": 154.5123,
"eval_samples_per_second": 3.734,
"eval_steps_per_second": 3.734,
"step": 500
},
{
"epoch": 0.7865818392134182,
"grad_norm": 0.37575840950012207,
"learning_rate": 9.240452596456905e-05,
"loss": 1.7869,
"step": 510
},
{
"epoch": 0.8020050125313283,
"grad_norm": 0.5916942358016968,
"learning_rate": 9.192044226003789e-05,
"loss": 2.1779,
"step": 520
},
{
"epoch": 0.8174281858492385,
"grad_norm": 0.3484373092651367,
"learning_rate": 9.142275598780473e-05,
"loss": 2.32,
"step": 530
},
{
"epoch": 0.8328513591671486,
"grad_norm": 0.39120087027549744,
"learning_rate": 9.091162863975388e-05,
"loss": 1.9688,
"step": 540
},
{
"epoch": 0.8482745324850588,
"grad_norm": 0.6257333755493164,
"learning_rate": 9.03872260692014e-05,
"loss": 2.231,
"step": 550
},
{
"epoch": 0.863697705802969,
"grad_norm": 0.4078841209411621,
"learning_rate": 8.984971843707788e-05,
"loss": 2.1618,
"step": 560
},
{
"epoch": 0.8791208791208791,
"grad_norm": 0.3387174606323242,
"learning_rate": 8.9299280156714e-05,
"loss": 2.2796,
"step": 570
},
{
"epoch": 0.8945440524387893,
"grad_norm": 0.3817114233970642,
"learning_rate": 8.87360898372458e-05,
"loss": 1.9383,
"step": 580
},
{
"epoch": 0.9099672257566994,
"grad_norm": 0.3814316391944885,
"learning_rate": 8.816033022565864e-05,
"loss": 2.1884,
"step": 590
},
{
"epoch": 0.9253903990746096,
"grad_norm": 0.3003975450992584,
"learning_rate": 8.75721881474886e-05,
"loss": 2.1796,
"step": 600
},
{
"epoch": 0.9408135723925197,
"grad_norm": 0.4703477919101715,
"learning_rate": 8.697185444620024e-05,
"loss": 2.1362,
"step": 610
},
{
"epoch": 0.9562367457104299,
"grad_norm": 0.37337568402290344,
"learning_rate": 8.635952392126072e-05,
"loss": 2.239,
"step": 620
},
{
"epoch": 0.97165991902834,
"grad_norm": 0.3282889127731323,
"learning_rate": 8.573539526493023e-05,
"loss": 2.15,
"step": 630
},
{
"epoch": 0.9870830923462502,
"grad_norm": 0.3719846308231354,
"learning_rate": 8.509967099778933e-05,
"loss": 2.0602,
"step": 640
},
{
"epoch": 1.0025062656641603,
"grad_norm": 0.360990434885025,
"learning_rate": 8.445255740302391e-05,
"loss": 2.228,
"step": 650
},
{
"epoch": 1.0179294389820706,
"grad_norm": 0.3433365225791931,
"learning_rate": 8.379426445948932e-05,
"loss": 2.1467,
"step": 660
},
{
"epoch": 1.0333526122999808,
"grad_norm": 0.33261704444885254,
"learning_rate": 8.312500577357547e-05,
"loss": 2.2814,
"step": 670
},
{
"epoch": 1.048775785617891,
"grad_norm": 0.41563576459884644,
"learning_rate": 8.244499850989452e-05,
"loss": 2.046,
"step": 680
},
{
"epoch": 1.064198958935801,
"grad_norm": 0.44597044587135315,
"learning_rate": 8.17544633208142e-05,
"loss": 1.9755,
"step": 690
},
{
"epoch": 1.0796221322537112,
"grad_norm": 0.37877991795539856,
"learning_rate": 8.105362427485942e-05,
"loss": 2.0912,
"step": 700
},
{
"epoch": 1.0950453055716214,
"grad_norm": 0.392952561378479,
"learning_rate": 8.034270878400528e-05,
"loss": 2.2183,
"step": 710
},
{
"epoch": 1.1104684788895316,
"grad_norm": 0.3665911853313446,
"learning_rate": 7.962194752988519e-05,
"loss": 2.1854,
"step": 720
},
{
"epoch": 1.1258916522074416,
"grad_norm": 0.3779922127723694,
"learning_rate": 7.889157438893812e-05,
"loss": 2.1174,
"step": 730
},
{
"epoch": 1.1413148255253518,
"grad_norm": 0.4270557463169098,
"learning_rate": 7.815182635651913e-05,
"loss": 2.0819,
"step": 740
},
{
"epoch": 1.156737998843262,
"grad_norm": 0.4602227807044983,
"learning_rate": 7.740294346999786e-05,
"loss": 2.1205,
"step": 750
},
{
"epoch": 1.1721611721611722,
"grad_norm": 0.4194934666156769,
"learning_rate": 7.664516873086987e-05,
"loss": 2.1829,
"step": 760
},
{
"epoch": 1.1875843454790824,
"grad_norm": 0.3483639359474182,
"learning_rate": 7.58787480259063e-05,
"loss": 2.3507,
"step": 770
},
{
"epoch": 1.2030075187969924,
"grad_norm": 0.4419120252132416,
"learning_rate": 7.510393004736722e-05,
"loss": 2.1003,
"step": 780
},
{
"epoch": 1.2184306921149026,
"grad_norm": 0.4768407344818115,
"learning_rate": 7.432096621230455e-05,
"loss": 2.0724,
"step": 790
},
{
"epoch": 1.2338538654328128,
"grad_norm": 0.40922412276268005,
"learning_rate": 7.353011058098104e-05,
"loss": 2.0352,
"step": 800
},
{
"epoch": 1.249277038750723,
"grad_norm": 0.38617295026779175,
"learning_rate": 7.273161977443137e-05,
"loss": 2.1549,
"step": 810
},
{
"epoch": 1.2647002120686333,
"grad_norm": 0.3819389343261719,
"learning_rate": 7.192575289119245e-05,
"loss": 2.0827,
"step": 820
},
{
"epoch": 1.2801233853865432,
"grad_norm": 0.4457474946975708,
"learning_rate": 7.11127714232297e-05,
"loss": 1.9096,
"step": 830
},
{
"epoch": 1.2955465587044535,
"grad_norm": 0.47230038046836853,
"learning_rate": 7.029293917108678e-05,
"loss": 2.0398,
"step": 840
},
{
"epoch": 1.3109697320223637,
"grad_norm": 0.43403419852256775,
"learning_rate": 6.946652215828618e-05,
"loss": 2.1837,
"step": 850
},
{
"epoch": 1.3263929053402737,
"grad_norm": 0.3847104012966156,
"learning_rate": 6.863378854500846e-05,
"loss": 2.1997,
"step": 860
},
{
"epoch": 1.3418160786581839,
"grad_norm": 0.40409407019615173,
"learning_rate": 6.779500854107827e-05,
"loss": 1.8728,
"step": 870
},
{
"epoch": 1.357239251976094,
"grad_norm": 0.3734535574913025,
"learning_rate": 6.695045431828524e-05,
"loss": 2.0045,
"step": 880
},
{
"epoch": 1.3726624252940043,
"grad_norm": 0.40710312128067017,
"learning_rate": 6.610039992206814e-05,
"loss": 2.1053,
"step": 890
},
{
"epoch": 1.3880855986119145,
"grad_norm": 0.3717002868652344,
"learning_rate": 6.524512118259121e-05,
"loss": 2.1408,
"step": 900
},
{
"epoch": 1.4035087719298245,
"grad_norm": 0.3799400329589844,
"learning_rate": 6.438489562524142e-05,
"loss": 2.242,
"step": 910
},
{
"epoch": 1.4189319452477347,
"grad_norm": 0.3500377833843231,
"learning_rate": 6.35200023805754e-05,
"loss": 2.1352,
"step": 920
},
{
"epoch": 1.434355118565645,
"grad_norm": 0.4466979503631592,
"learning_rate": 6.265072209374574e-05,
"loss": 2.1492,
"step": 930
},
{
"epoch": 1.4497782918835551,
"grad_norm": 0.42561647295951843,
"learning_rate": 6.177733683343578e-05,
"loss": 2.1381,
"step": 940
},
{
"epoch": 1.4652014652014653,
"grad_norm": 0.3749467134475708,
"learning_rate": 6.090013000033251e-05,
"loss": 2.2012,
"step": 950
},
{
"epoch": 1.4806246385193753,
"grad_norm": 0.4494505822658539,
"learning_rate": 6.0019386235167055e-05,
"loss": 1.82,
"step": 960
},
{
"epoch": 1.4960478118372855,
"grad_norm": 0.3980197608470917,
"learning_rate": 5.9135391326353085e-05,
"loss": 2.2977,
"step": 970
},
{
"epoch": 1.5114709851551957,
"grad_norm": 0.4605059027671814,
"learning_rate": 5.824843211725265e-05,
"loss": 2.1393,
"step": 980
},
{
"epoch": 1.5268941584731057,
"grad_norm": 0.4184533655643463,
"learning_rate": 5.7358796413099635e-05,
"loss": 2.0266,
"step": 990
},
{
"epoch": 1.5423173317910162,
"grad_norm": 0.4229032099246979,
"learning_rate": 5.646677288761132e-05,
"loss": 2.2371,
"step": 1000
},
{
"epoch": 1.5423173317910162,
"eval_loss": 2.0582103729248047,
"eval_runtime": 154.9137,
"eval_samples_per_second": 3.725,
"eval_steps_per_second": 3.725,
"step": 1000
},
{
"epoch": 1.5577405051089261,
"grad_norm": 0.565112829208374,
"learning_rate": 5.5572650989317874e-05,
"loss": 2.1034,
"step": 1010
},
{
"epoch": 1.5731636784268364,
"grad_norm": 0.5104465484619141,
"learning_rate": 5.467672084764066e-05,
"loss": 2.1003,
"step": 1020
},
{
"epoch": 1.5885868517447466,
"grad_norm": 0.49492645263671875,
"learning_rate": 5.3779273178749425e-05,
"loss": 1.9111,
"step": 1030
},
{
"epoch": 1.6040100250626566,
"grad_norm": 0.3711722493171692,
"learning_rate": 5.288059919122922e-05,
"loss": 1.8673,
"step": 1040
},
{
"epoch": 1.6194331983805668,
"grad_norm": 0.44108933210372925,
"learning_rate": 5.1980990491587465e-05,
"loss": 2.1027,
"step": 1050
},
{
"epoch": 1.634856371698477,
"grad_norm": 0.6337707042694092,
"learning_rate": 5.108073898963194e-05,
"loss": 2.1683,
"step": 1060
},
{
"epoch": 1.650279545016387,
"grad_norm": 0.4429640471935272,
"learning_rate": 5.018013680375035e-05,
"loss": 1.9772,
"step": 1070
},
{
"epoch": 1.6657027183342974,
"grad_norm": 0.4720529019832611,
"learning_rate": 4.927947616612215e-05,
"loss": 2.008,
"step": 1080
},
{
"epoch": 1.6811258916522074,
"grad_norm": 0.44429996609687805,
"learning_rate": 4.8379049327893535e-05,
"loss": 2.2032,
"step": 1090
},
{
"epoch": 1.6965490649701176,
"grad_norm": 0.39796188473701477,
"learning_rate": 4.7479148464346276e-05,
"loss": 2.0932,
"step": 1100
},
{
"epoch": 1.7119722382880278,
"grad_norm": 0.3974200189113617,
"learning_rate": 4.658006558009099e-05,
"loss": 2.0747,
"step": 1110
},
{
"epoch": 1.7273954116059378,
"grad_norm": 0.37543272972106934,
"learning_rate": 4.568209241431614e-05,
"loss": 2.3118,
"step": 1120
},
{
"epoch": 1.7428185849238482,
"grad_norm": 0.5304645299911499,
"learning_rate": 4.478552034612277e-05,
"loss": 2.047,
"step": 1130
},
{
"epoch": 1.7582417582417582,
"grad_norm": 0.43397507071495056,
"learning_rate": 4.3890640299976345e-05,
"loss": 2.1676,
"step": 1140
},
{
"epoch": 1.7736649315596684,
"grad_norm": 0.4973481297492981,
"learning_rate": 4.299774265130609e-05,
"loss": 2.023,
"step": 1150
},
{
"epoch": 1.7890881048775786,
"grad_norm": 0.4363488554954529,
"learning_rate": 4.21071171322823e-05,
"loss": 2.0945,
"step": 1160
},
{
"epoch": 1.8045112781954886,
"grad_norm": 0.4055328369140625,
"learning_rate": 4.121905273780263e-05,
"loss": 2.2178,
"step": 1170
},
{
"epoch": 1.8199344515133988,
"grad_norm": 0.49545660614967346,
"learning_rate": 4.033383763171738e-05,
"loss": 1.9995,
"step": 1180
},
{
"epoch": 1.835357624831309,
"grad_norm": 0.37422749400138855,
"learning_rate": 3.9451759053324724e-05,
"loss": 2.1181,
"step": 1190
},
{
"epoch": 1.8507807981492193,
"grad_norm": 0.37200838327407837,
"learning_rate": 3.8573103224165555e-05,
"loss": 1.917,
"step": 1200
},
{
"epoch": 1.8662039714671295,
"grad_norm": 0.4319084584712982,
"learning_rate": 3.7698155255149005e-05,
"loss": 2.165,
"step": 1210
},
{
"epoch": 1.8816271447850395,
"grad_norm": 0.3685438930988312,
"learning_rate": 3.682719905403804e-05,
"loss": 2.0359,
"step": 1220
},
{
"epoch": 1.8970503181029497,
"grad_norm": 0.6587750315666199,
"learning_rate": 3.596051723332554e-05,
"loss": 2.0902,
"step": 1230
},
{
"epoch": 1.9124734914208599,
"grad_norm": 0.5964400768280029,
"learning_rate": 3.5098391018530816e-05,
"loss": 1.9265,
"step": 1240
},
{
"epoch": 1.9278966647387699,
"grad_norm": 0.5480875372886658,
"learning_rate": 3.42411001569459e-05,
"loss": 1.9755,
"step": 1250
},
{
"epoch": 1.9433198380566803,
"grad_norm": 0.4315269887447357,
"learning_rate": 3.338892282686179e-05,
"loss": 2.0685,
"step": 1260
},
{
"epoch": 1.9587430113745903,
"grad_norm": 0.56096351146698,
"learning_rate": 3.254213554730357e-05,
"loss": 2.1827,
"step": 1270
},
{
"epoch": 1.9741661846925005,
"grad_norm": 0.5271403193473816,
"learning_rate": 3.170101308830421e-05,
"loss": 2.1641,
"step": 1280
},
{
"epoch": 1.9895893580104107,
"grad_norm": 0.4906688332557678,
"learning_rate": 3.086582838174551e-05,
"loss": 2.085,
"step": 1290
},
{
"epoch": 2.0050125313283207,
"grad_norm": 0.38613101840019226,
"learning_rate": 3.0036852432795925e-05,
"loss": 2.0016,
"step": 1300
},
{
"epoch": 2.020435704646231,
"grad_norm": 0.5777701735496521,
"learning_rate": 2.9214354231973207e-05,
"loss": 1.7941,
"step": 1310
},
{
"epoch": 2.035858877964141,
"grad_norm": 0.48649054765701294,
"learning_rate": 2.839860066786103e-05,
"loss": 2.1192,
"step": 1320
},
{
"epoch": 2.051282051282051,
"grad_norm": 0.4648803472518921,
"learning_rate": 2.7589856440507522e-05,
"loss": 2.0756,
"step": 1330
},
{
"epoch": 2.0667052245999615,
"grad_norm": 0.5168564319610596,
"learning_rate": 2.6788383975533994e-05,
"loss": 2.2201,
"step": 1340
},
{
"epoch": 2.0821283979178715,
"grad_norm": 0.377386212348938,
"learning_rate": 2.5994443338981732e-05,
"loss": 2.0903,
"step": 1350
},
{
"epoch": 2.097551571235782,
"grad_norm": 0.5663664937019348,
"learning_rate": 2.5208292152924262e-05,
"loss": 2.0757,
"step": 1360
},
{
"epoch": 2.112974744553692,
"grad_norm": 0.5003238320350647,
"learning_rate": 2.4430185511872944e-05,
"loss": 2.1352,
"step": 1370
},
{
"epoch": 2.128397917871602,
"grad_norm": 0.5036555528640747,
"learning_rate": 2.3660375900002362e-05,
"loss": 1.9787,
"step": 1380
},
{
"epoch": 2.1438210911895124,
"grad_norm": 0.4774503707885742,
"learning_rate": 2.2899113109223114e-05,
"loss": 1.9539,
"step": 1390
},
{
"epoch": 2.1592442645074224,
"grad_norm": 0.48530662059783936,
"learning_rate": 2.2146644158127827e-05,
"loss": 2.179,
"step": 1400
},
{
"epoch": 2.174667437825333,
"grad_norm": 0.5162932276725769,
"learning_rate": 2.140321321183734e-05,
"loss": 2.278,
"step": 1410
},
{
"epoch": 2.1900906111432428,
"grad_norm": 0.4148849546909332,
"learning_rate": 2.0669061502772776e-05,
"loss": 2.1949,
"step": 1420
},
{
"epoch": 2.2055137844611528,
"grad_norm": 0.4029206931591034,
"learning_rate": 1.9944427252378978e-05,
"loss": 2.1155,
"step": 1430
},
{
"epoch": 2.220936957779063,
"grad_norm": 0.5142106413841248,
"learning_rate": 1.9229545593825366e-05,
"loss": 2.0562,
"step": 1440
},
{
"epoch": 2.236360131096973,
"grad_norm": 0.4345252513885498,
"learning_rate": 1.8524648495708512e-05,
"loss": 2.0335,
"step": 1450
},
{
"epoch": 2.251783304414883,
"grad_norm": 0.42935454845428467,
"learning_rate": 1.7829964686781793e-05,
"loss": 2.2724,
"step": 1460
},
{
"epoch": 2.2672064777327936,
"grad_norm": 0.5543069243431091,
"learning_rate": 1.7145719581736336e-05,
"loss": 2.0677,
"step": 1470
},
{
"epoch": 2.2826296510507036,
"grad_norm": 0.4770039916038513,
"learning_rate": 1.6472135208057126e-05,
"loss": 2.3684,
"step": 1480
},
{
"epoch": 2.298052824368614,
"grad_norm": 0.5183836221694946,
"learning_rate": 1.5809430133978508e-05,
"loss": 2.0769,
"step": 1490
},
{
"epoch": 2.313475997686524,
"grad_norm": 0.4000667929649353,
"learning_rate": 1.5157819397561862e-05,
"loss": 2.0174,
"step": 1500
},
{
"epoch": 2.313475997686524,
"eval_loss": 2.042402505874634,
"eval_runtime": 154.606,
"eval_samples_per_second": 3.732,
"eval_steps_per_second": 3.732,
"step": 1500
},
{
"epoch": 2.328899171004434,
"grad_norm": 0.5184345841407776,
"learning_rate": 1.4517514436918994e-05,
"loss": 2.041,
"step": 1510
},
{
"epoch": 2.3443223443223444,
"grad_norm": 0.5698699355125427,
"learning_rate": 1.3888723021603529e-05,
"loss": 2.026,
"step": 1520
},
{
"epoch": 2.3597455176402544,
"grad_norm": 0.5794140100479126,
"learning_rate": 1.327164918519282e-05,
"loss": 2.1522,
"step": 1530
},
{
"epoch": 2.375168690958165,
"grad_norm": 0.3988121747970581,
"learning_rate": 1.2666493159081943e-05,
"loss": 2.1803,
"step": 1540
},
{
"epoch": 2.390591864276075,
"grad_norm": 0.565253496170044,
"learning_rate": 1.207345130751164e-05,
"loss": 2.0834,
"step": 1550
},
{
"epoch": 2.406015037593985,
"grad_norm": 0.5080721378326416,
"learning_rate": 1.1492716063850973e-05,
"loss": 2.014,
"step": 1560
},
{
"epoch": 2.4214382109118953,
"grad_norm": 0.40610289573669434,
"learning_rate": 1.0924475868155493e-05,
"loss": 2.0482,
"step": 1570
},
{
"epoch": 2.4368613842298052,
"grad_norm": 0.48359382152557373,
"learning_rate": 1.0368915106021255e-05,
"loss": 1.9128,
"step": 1580
},
{
"epoch": 2.4522845575477152,
"grad_norm": 0.6242773532867432,
"learning_rate": 9.826214048754367e-06,
"loss": 1.9792,
"step": 1590
},
{
"epoch": 2.4677077308656257,
"grad_norm": 0.5390336513519287,
"learning_rate": 9.29654879487566e-06,
"loss": 1.9065,
"step": 1600
},
{
"epoch": 2.4831309041835357,
"grad_norm": 0.49583902955055237,
"learning_rate": 8.780091212979207e-06,
"loss": 2.1249,
"step": 1610
},
{
"epoch": 2.498554077501446,
"grad_norm": 0.5197187662124634,
"learning_rate": 8.277008885963594e-06,
"loss": 1.8338,
"step": 1620
},
{
"epoch": 2.513977250819356,
"grad_norm": 0.5529533624649048,
"learning_rate": 7.787465056653653e-06,
"loss": 1.9179,
"step": 1630
},
{
"epoch": 2.5294004241372665,
"grad_norm": 0.6007400155067444,
"learning_rate": 7.31161857483057e-06,
"loss": 2.0279,
"step": 1640
},
{
"epoch": 2.5448235974551765,
"grad_norm": 0.5567540526390076,
"learning_rate": 6.849623845687547e-06,
"loss": 2.0689,
"step": 1650
},
{
"epoch": 2.5602467707730865,
"grad_norm": 0.614508330821991,
"learning_rate": 6.401630779727452e-06,
"loss": 2.1043,
"step": 1660
},
{
"epoch": 2.575669944090997,
"grad_norm": 0.4546593427658081,
"learning_rate": 5.967784744119203e-06,
"loss": 2.0116,
"step": 1670
},
{
"epoch": 2.591093117408907,
"grad_norm": 0.35855767130851746,
"learning_rate": 5.548226515528132e-06,
"loss": 1.8218,
"step": 1680
},
{
"epoch": 2.606516290726817,
"grad_norm": 0.5323745012283325,
"learning_rate": 5.143092234436126e-06,
"loss": 2.0448,
"step": 1690
},
{
"epoch": 2.6219394640447273,
"grad_norm": 0.5990585684776306,
"learning_rate": 4.752513360965949e-06,
"loss": 1.9742,
"step": 1700
},
{
"epoch": 2.6373626373626373,
"grad_norm": 0.5095666646957397,
"learning_rate": 4.376616632224451e-06,
"loss": 2.1038,
"step": 1710
},
{
"epoch": 2.6527858106805473,
"grad_norm": 0.4896791875362396,
"learning_rate": 4.015524021178196e-06,
"loss": 2.0926,
"step": 1720
},
{
"epoch": 2.6682089839984577,
"grad_norm": 0.5555339455604553,
"learning_rate": 3.6693526970749957e-06,
"loss": 2.1485,
"step": 1730
},
{
"epoch": 2.6836321573163677,
"grad_norm": 0.5582926273345947,
"learning_rate": 3.3382149874242817e-06,
"loss": 2.1413,
"step": 1740
},
{
"epoch": 2.699055330634278,
"grad_norm": 0.48462751507759094,
"learning_rate": 3.0222183415484216e-06,
"loss": 2.0053,
"step": 1750
},
{
"epoch": 2.714478503952188,
"grad_norm": 0.5062531232833862,
"learning_rate": 2.721465295716996e-06,
"loss": 2.1571,
"step": 1760
},
{
"epoch": 2.7299016772700986,
"grad_norm": 0.464996337890625,
"learning_rate": 2.436053439875319e-06,
"loss": 2.2241,
"step": 1770
},
{
"epoch": 2.7453248505880086,
"grad_norm": 0.41364607214927673,
"learning_rate": 2.1660753859779227e-06,
"loss": 1.8589,
"step": 1780
},
{
"epoch": 2.7607480239059186,
"grad_norm": 0.5476650595664978,
"learning_rate": 1.911618737937304e-06,
"loss": 2.2024,
"step": 1790
},
{
"epoch": 2.776171197223829,
"grad_norm": 0.4496407210826874,
"learning_rate": 1.6727660631977893e-06,
"loss": 2.219,
"step": 1800
},
{
"epoch": 2.791594370541739,
"grad_norm": 0.46782755851745605,
"learning_rate": 1.4495948659435932e-06,
"loss": 1.9196,
"step": 1810
},
{
"epoch": 2.807017543859649,
"grad_norm": 0.5227280259132385,
"learning_rate": 1.24217756194982e-06,
"loss": 1.9901,
"step": 1820
},
{
"epoch": 2.8224407171775594,
"grad_norm": 0.4377809464931488,
"learning_rate": 1.0505814550846704e-06,
"loss": 2.1759,
"step": 1830
},
{
"epoch": 2.8378638904954694,
"grad_norm": 0.5370849370956421,
"learning_rate": 8.748687154702673e-07,
"loss": 2.0904,
"step": 1840
},
{
"epoch": 2.8532870638133794,
"grad_norm": 0.5060635805130005,
"learning_rate": 7.150963593094029e-07,
"loss": 1.9761,
"step": 1850
},
{
"epoch": 2.86871023713129,
"grad_norm": 0.49891987442970276,
"learning_rate": 5.713162303845887e-07,
"loss": 2.0288,
"step": 1860
},
{
"epoch": 2.8841334104492,
"grad_norm": 0.5800537467002869,
"learning_rate": 4.4357498323553405e-07,
"loss": 1.9074,
"step": 1870
},
{
"epoch": 2.8995565837671102,
"grad_norm": 0.5344066023826599,
"learning_rate": 3.3191406802041693e-07,
"loss": 1.8958,
"step": 1880
},
{
"epoch": 2.91497975708502,
"grad_norm": 0.45593753457069397,
"learning_rate": 2.3636971706592627e-07,
"loss": 2.0741,
"step": 1890
},
{
"epoch": 2.9304029304029307,
"grad_norm": 0.5734761953353882,
"learning_rate": 1.5697293311039974e-07,
"loss": 2.0763,
"step": 1900
},
{
"epoch": 2.9458261037208406,
"grad_norm": 0.5509295463562012,
"learning_rate": 9.374947924388755e-08,
"loss": 1.9926,
"step": 1910
},
{
"epoch": 2.9612492770387506,
"grad_norm": 0.4733286499977112,
"learning_rate": 4.671987054842841e-08,
"loss": 2.3555,
"step": 1920
},
{
"epoch": 2.976672450356661,
"grad_norm": 0.6256054043769836,
"learning_rate": 1.589936744116893e-08,
"loss": 1.9345,
"step": 1930
},
{
"epoch": 2.992095623674571,
"grad_norm": 0.5568147897720337,
"learning_rate": 1.2979707226135062e-09,
"loss": 2.0782,
"step": 1940
},
{
"epoch": 2.998264893001735,
"step": 1944,
"total_flos": 1.2799386304118784e+18,
"train_loss": 2.14731617155389,
"train_runtime": 14043.9874,
"train_samples_per_second": 1.108,
"train_steps_per_second": 0.138
}
],
"logging_steps": 10,
"max_steps": 1944,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"total_flos": 1.2799386304118784e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}