{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 5268,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0056947608200455585,
      "grad_norm": 12.728320121765137,
      "learning_rate": 4.990508731966591e-05,
      "loss": 1.816,
      "step": 10
    },
    {
      "epoch": 0.011389521640091117,
      "grad_norm": 7.951778888702393,
      "learning_rate": 4.981017463933182e-05,
      "loss": 0.9591,
      "step": 20
    },
    {
      "epoch": 0.017084282460136675,
      "grad_norm": 4.346314907073975,
      "learning_rate": 4.971526195899772e-05,
      "loss": 0.5211,
      "step": 30
    },
    {
      "epoch": 0.022779043280182234,
      "grad_norm": 2.759699821472168,
      "learning_rate": 4.962034927866363e-05,
      "loss": 0.5977,
      "step": 40
    },
    {
      "epoch": 0.02847380410022779,
      "grad_norm": 4.309758186340332,
      "learning_rate": 4.9525436598329536e-05,
      "loss": 0.5117,
      "step": 50
    },
    {
      "epoch": 0.03416856492027335,
      "grad_norm": 3.065225601196289,
      "learning_rate": 4.9430523917995447e-05,
      "loss": 0.4366,
      "step": 60
    },
    {
      "epoch": 0.03986332574031891,
      "grad_norm": 4.769909381866455,
      "learning_rate": 4.933561123766136e-05,
      "loss": 0.4092,
      "step": 70
    },
    {
      "epoch": 0.04555808656036447,
      "grad_norm": 1.6601901054382324,
      "learning_rate": 4.924069855732726e-05,
      "loss": 0.3484,
      "step": 80
    },
    {
      "epoch": 0.05125284738041002,
      "grad_norm": 4.936558723449707,
      "learning_rate": 4.9145785876993165e-05,
      "loss": 0.4445,
      "step": 90
    },
    {
      "epoch": 0.05694760820045558,
      "grad_norm": 1.2995065450668335,
      "learning_rate": 4.9050873196659076e-05,
      "loss": 0.3302,
      "step": 100
    },
    {
      "epoch": 0.06264236902050115,
      "grad_norm": 4.602411270141602,
      "learning_rate": 4.895596051632499e-05,
      "loss": 0.3969,
      "step": 110
    },
    {
      "epoch": 0.0683371298405467,
      "grad_norm": 4.256335258483887,
      "learning_rate": 4.886104783599089e-05,
      "loss": 0.3385,
      "step": 120
    },
    {
      "epoch": 0.07403189066059225,
      "grad_norm": 4.157656669616699,
      "learning_rate": 4.87661351556568e-05,
      "loss": 0.2872,
      "step": 130
    },
    {
      "epoch": 0.07972665148063782,
      "grad_norm": 3.6777021884918213,
      "learning_rate": 4.8671222475322705e-05,
      "loss": 0.2591,
      "step": 140
    },
    {
      "epoch": 0.08542141230068337,
      "grad_norm": 2.272169589996338,
      "learning_rate": 4.857630979498861e-05,
      "loss": 0.275,
      "step": 150
    },
    {
      "epoch": 0.09111617312072894,
      "grad_norm": 0.8441994786262512,
      "learning_rate": 4.848139711465452e-05,
      "loss": 0.2029,
      "step": 160
    },
    {
      "epoch": 0.09681093394077449,
      "grad_norm": 3.8441519737243652,
      "learning_rate": 4.838648443432043e-05,
      "loss": 0.3218,
      "step": 170
    },
    {
      "epoch": 0.10250569476082004,
      "grad_norm": 1.3639609813690186,
      "learning_rate": 4.8291571753986335e-05,
      "loss": 0.2317,
      "step": 180
    },
    {
      "epoch": 0.1082004555808656,
      "grad_norm": 2.0190043449401855,
      "learning_rate": 4.8196659073652246e-05,
      "loss": 0.227,
      "step": 190
    },
    {
      "epoch": 0.11389521640091116,
      "grad_norm": 2.7274367809295654,
      "learning_rate": 4.810174639331815e-05,
      "loss": 0.1802,
      "step": 200
    },
    {
      "epoch": 0.11958997722095673,
      "grad_norm": 4.253875732421875,
      "learning_rate": 4.8006833712984054e-05,
      "loss": 0.1921,
      "step": 210
    },
    {
      "epoch": 0.1252847380410023,
      "grad_norm": 2.3438820838928223,
      "learning_rate": 4.7911921032649964e-05,
      "loss": 0.235,
      "step": 220
    },
    {
      "epoch": 0.13097949886104784,
      "grad_norm": 2.578488349914551,
      "learning_rate": 4.7817008352315875e-05,
      "loss": 0.2131,
      "step": 230
    },
    {
      "epoch": 0.1366742596810934,
      "grad_norm": 1.772674560546875,
      "learning_rate": 4.772209567198178e-05,
      "loss": 0.2041,
      "step": 240
    },
    {
      "epoch": 0.14236902050113895,
      "grad_norm": 1.3497803211212158,
      "learning_rate": 4.762718299164768e-05,
      "loss": 0.2185,
      "step": 250
    },
    {
      "epoch": 0.1480637813211845,
      "grad_norm": 2.353133201599121,
      "learning_rate": 4.7532270311313594e-05,
      "loss": 0.2591,
      "step": 260
    },
    {
      "epoch": 0.15375854214123008,
      "grad_norm": 1.3427927494049072,
      "learning_rate": 4.74373576309795e-05,
      "loss": 0.2021,
      "step": 270
    },
    {
      "epoch": 0.15945330296127563,
      "grad_norm": 1.5266870260238647,
      "learning_rate": 4.734244495064541e-05,
      "loss": 0.2301,
      "step": 280
    },
    {
      "epoch": 0.16514806378132119,
      "grad_norm": 1.570088505744934,
      "learning_rate": 4.724753227031132e-05,
      "loss": 0.2059,
      "step": 290
    },
    {
      "epoch": 0.17084282460136674,
      "grad_norm": 1.912508487701416,
      "learning_rate": 4.715261958997722e-05,
      "loss": 0.2137,
      "step": 300
    },
    {
      "epoch": 0.1765375854214123,
      "grad_norm": 2.4157466888427734,
      "learning_rate": 4.705770690964313e-05,
      "loss": 0.218,
      "step": 310
    },
    {
      "epoch": 0.18223234624145787,
      "grad_norm": 4.4945831298828125,
      "learning_rate": 4.696279422930904e-05,
      "loss": 0.3602,
      "step": 320
    },
    {
      "epoch": 0.18792710706150342,
      "grad_norm": 1.9692164659500122,
      "learning_rate": 4.686788154897495e-05,
      "loss": 0.2795,
      "step": 330
    },
    {
      "epoch": 0.19362186788154898,
      "grad_norm": 3.558931350708008,
      "learning_rate": 4.677296886864085e-05,
      "loss": 0.1841,
      "step": 340
    },
    {
      "epoch": 0.19931662870159453,
      "grad_norm": 1.7558096647262573,
      "learning_rate": 4.6678056188306763e-05,
      "loss": 0.2018,
      "step": 350
    },
    {
      "epoch": 0.20501138952164008,
      "grad_norm": 1.1252487897872925,
      "learning_rate": 4.658314350797267e-05,
      "loss": 0.178,
      "step": 360
    },
    {
      "epoch": 0.21070615034168566,
      "grad_norm": 0.8949424028396606,
      "learning_rate": 4.648823082763857e-05,
      "loss": 0.1891,
      "step": 370
    },
    {
      "epoch": 0.2164009111617312,
      "grad_norm": 1.9878759384155273,
      "learning_rate": 4.639331814730448e-05,
      "loss": 0.1868,
      "step": 380
    },
    {
      "epoch": 0.22209567198177677,
      "grad_norm": 2.801661491394043,
      "learning_rate": 4.629840546697039e-05,
      "loss": 0.2122,
      "step": 390
    },
    {
      "epoch": 0.22779043280182232,
      "grad_norm": 1.7705647945404053,
      "learning_rate": 4.62034927866363e-05,
      "loss": 0.2144,
      "step": 400
    },
    {
      "epoch": 0.23348519362186787,
      "grad_norm": 3.3794710636138916,
      "learning_rate": 4.61085801063022e-05,
      "loss": 0.1505,
      "step": 410
    },
    {
      "epoch": 0.23917995444191345,
      "grad_norm": 2.587388277053833,
      "learning_rate": 4.601366742596811e-05,
      "loss": 0.1637,
      "step": 420
    },
    {
      "epoch": 0.244874715261959,
      "grad_norm": 1.953997254371643,
      "learning_rate": 4.5918754745634016e-05,
      "loss": 0.1736,
      "step": 430
    },
    {
      "epoch": 0.2505694760820046,
      "grad_norm": 1.6005550622940063,
      "learning_rate": 4.5823842065299926e-05,
      "loss": 0.1899,
      "step": 440
    },
    {
      "epoch": 0.25626423690205014,
      "grad_norm": 2.8049798011779785,
      "learning_rate": 4.572892938496584e-05,
      "loss": 0.2553,
      "step": 450
    },
    {
      "epoch": 0.2619589977220957,
      "grad_norm": 2.1441080570220947,
      "learning_rate": 4.563401670463174e-05,
      "loss": 0.1774,
      "step": 460
    },
    {
      "epoch": 0.26765375854214124,
      "grad_norm": 1.1879520416259766,
      "learning_rate": 4.5539104024297645e-05,
      "loss": 0.1974,
      "step": 470
    },
    {
      "epoch": 0.2733485193621868,
      "grad_norm": 1.3628582954406738,
      "learning_rate": 4.5444191343963556e-05,
      "loss": 0.154,
      "step": 480
    },
    {
      "epoch": 0.27904328018223234,
      "grad_norm": 1.7644487619400024,
      "learning_rate": 4.5349278663629466e-05,
      "loss": 0.171,
      "step": 490
    },
    {
      "epoch": 0.2847380410022779,
      "grad_norm": 1.8280380964279175,
      "learning_rate": 4.525436598329537e-05,
      "loss": 0.1542,
      "step": 500
    },
    {
      "epoch": 0.29043280182232345,
      "grad_norm": 1.6000663042068481,
      "learning_rate": 4.515945330296128e-05,
      "loss": 0.1963,
      "step": 510
    },
    {
      "epoch": 0.296127562642369,
      "grad_norm": 1.8194619417190552,
      "learning_rate": 4.5064540622627185e-05,
      "loss": 0.1298,
      "step": 520
    },
    {
      "epoch": 0.30182232346241455,
      "grad_norm": 1.9171711206436157,
      "learning_rate": 4.496962794229309e-05,
      "loss": 0.1936,
      "step": 530
    },
    {
      "epoch": 0.30751708428246016,
      "grad_norm": 1.4599436521530151,
      "learning_rate": 4.4874715261959e-05,
      "loss": 0.1442,
      "step": 540
    },
    {
      "epoch": 0.3132118451025057,
      "grad_norm": 0.9252722263336182,
      "learning_rate": 4.477980258162491e-05,
      "loss": 0.2002,
      "step": 550
    },
    {
      "epoch": 0.31890660592255127,
      "grad_norm": 1.1192878484725952,
      "learning_rate": 4.4684889901290815e-05,
      "loss": 0.1121,
      "step": 560
    },
    {
      "epoch": 0.3246013667425968,
      "grad_norm": 1.3718199729919434,
      "learning_rate": 4.4589977220956725e-05,
      "loss": 0.1876,
      "step": 570
    },
    {
      "epoch": 0.33029612756264237,
      "grad_norm": 1.7225291728973389,
      "learning_rate": 4.449506454062263e-05,
      "loss": 0.1476,
      "step": 580
    },
    {
      "epoch": 0.3359908883826879,
      "grad_norm": 1.7924920320510864,
      "learning_rate": 4.440015186028853e-05,
      "loss": 0.1476,
      "step": 590
    },
    {
      "epoch": 0.3416856492027335,
      "grad_norm": 2.5110108852386475,
      "learning_rate": 4.4305239179954444e-05,
      "loss": 0.1659,
      "step": 600
    },
    {
      "epoch": 0.34738041002277903,
      "grad_norm": 1.2690166234970093,
      "learning_rate": 4.4210326499620355e-05,
      "loss": 0.1385,
      "step": 610
    },
    {
      "epoch": 0.3530751708428246,
      "grad_norm": 1.8079414367675781,
      "learning_rate": 4.411541381928626e-05,
      "loss": 0.1213,
      "step": 620
    },
    {
      "epoch": 0.35876993166287013,
      "grad_norm": 2.604128360748291,
      "learning_rate": 4.402050113895216e-05,
      "loss": 0.115,
      "step": 630
    },
    {
      "epoch": 0.36446469248291574,
      "grad_norm": 1.001185655593872,
      "learning_rate": 4.3925588458618073e-05,
      "loss": 0.1608,
      "step": 640
    },
    {
      "epoch": 0.3701594533029613,
      "grad_norm": 1.0883430242538452,
      "learning_rate": 4.383067577828398e-05,
      "loss": 0.1177,
      "step": 650
    },
    {
      "epoch": 0.37585421412300685,
      "grad_norm": 2.1729352474212646,
      "learning_rate": 4.373576309794989e-05,
      "loss": 0.1514,
      "step": 660
    },
    {
      "epoch": 0.3815489749430524,
      "grad_norm": 5.217422962188721,
      "learning_rate": 4.36408504176158e-05,
      "loss": 0.1534,
      "step": 670
    },
    {
      "epoch": 0.38724373576309795,
      "grad_norm": 4.588407039642334,
      "learning_rate": 4.35459377372817e-05,
      "loss": 0.1499,
      "step": 680
    },
    {
      "epoch": 0.3929384965831435,
      "grad_norm": 1.5867393016815186,
      "learning_rate": 4.345102505694761e-05,
      "loss": 0.192,
      "step": 690
    },
    {
      "epoch": 0.39863325740318906,
      "grad_norm": 0.7483230829238892,
      "learning_rate": 4.335611237661352e-05,
      "loss": 0.1254,
      "step": 700
    },
    {
      "epoch": 0.4043280182232346,
      "grad_norm": 1.8365514278411865,
      "learning_rate": 4.326119969627943e-05,
      "loss": 0.1735,
      "step": 710
    },
    {
      "epoch": 0.41002277904328016,
      "grad_norm": 4.194180488586426,
      "learning_rate": 4.316628701594533e-05,
      "loss": 0.2097,
      "step": 720
    },
    {
      "epoch": 0.4157175398633257,
      "grad_norm": 3.2393548488616943,
      "learning_rate": 4.307137433561124e-05,
      "loss": 0.116,
      "step": 730
    },
    {
      "epoch": 0.4214123006833713,
      "grad_norm": 1.270488977432251,
      "learning_rate": 4.297646165527715e-05,
      "loss": 0.19,
      "step": 740
    },
    {
      "epoch": 0.4271070615034169,
      "grad_norm": 1.7453019618988037,
      "learning_rate": 4.288154897494305e-05,
      "loss": 0.1397,
      "step": 750
    },
    {
      "epoch": 0.4328018223234624,
      "grad_norm": 1.9998120069503784,
      "learning_rate": 4.278663629460896e-05,
      "loss": 0.1014,
      "step": 760
    },
    {
      "epoch": 0.438496583143508,
      "grad_norm": 1.3646454811096191,
      "learning_rate": 4.269172361427487e-05,
      "loss": 0.1727,
      "step": 770
    },
    {
      "epoch": 0.44419134396355353,
      "grad_norm": 0.5129444003105164,
      "learning_rate": 4.2596810933940777e-05,
      "loss": 0.1545,
      "step": 780
    },
    {
      "epoch": 0.4498861047835991,
      "grad_norm": 4.377679347991943,
      "learning_rate": 4.250189825360668e-05,
      "loss": 0.1725,
      "step": 790
    },
    {
      "epoch": 0.45558086560364464,
      "grad_norm": 1.2630000114440918,
      "learning_rate": 4.240698557327259e-05,
      "loss": 0.1531,
      "step": 800
    },
    {
      "epoch": 0.4612756264236902,
      "grad_norm": 2.136129856109619,
      "learning_rate": 4.2312072892938495e-05,
      "loss": 0.1341,
      "step": 810
    },
    {
      "epoch": 0.46697038724373574,
      "grad_norm": 1.4710100889205933,
      "learning_rate": 4.2217160212604406e-05,
      "loss": 0.1716,
      "step": 820
    },
    {
      "epoch": 0.47266514806378135,
      "grad_norm": 1.6353185176849365,
      "learning_rate": 4.212224753227032e-05,
      "loss": 0.143,
      "step": 830
    },
    {
      "epoch": 0.4783599088838269,
      "grad_norm": 9.000621795654297,
      "learning_rate": 4.202733485193622e-05,
      "loss": 0.1185,
      "step": 840
    },
    {
      "epoch": 0.48405466970387245,
      "grad_norm": 1.606629729270935,
      "learning_rate": 4.1932422171602125e-05,
      "loss": 0.1555,
      "step": 850
    },
    {
      "epoch": 0.489749430523918,
      "grad_norm": 1.9729490280151367,
      "learning_rate": 4.1837509491268035e-05,
      "loss": 0.1648,
      "step": 860
    },
    {
      "epoch": 0.49544419134396356,
      "grad_norm": 9.148164749145508,
      "learning_rate": 4.174259681093394e-05,
      "loss": 0.1562,
      "step": 870
    },
    {
      "epoch": 0.5011389521640092,
      "grad_norm": 2.9721126556396484,
      "learning_rate": 4.164768413059985e-05,
      "loss": 0.1489,
      "step": 880
    },
    {
      "epoch": 0.5068337129840547,
      "grad_norm": 1.9185500144958496,
      "learning_rate": 4.155277145026576e-05,
      "loss": 0.1663,
      "step": 890
    },
    {
      "epoch": 0.5125284738041003,
      "grad_norm": 1.8967223167419434,
      "learning_rate": 4.1457858769931665e-05,
      "loss": 0.138,
      "step": 900
    },
    {
      "epoch": 0.5182232346241458,
      "grad_norm": 1.2305383682250977,
      "learning_rate": 4.136294608959757e-05,
      "loss": 0.1595,
      "step": 910
    },
    {
      "epoch": 0.5239179954441914,
      "grad_norm": 1.6164170503616333,
      "learning_rate": 4.126803340926348e-05,
      "loss": 0.147,
      "step": 920
    },
    {
      "epoch": 0.5296127562642369,
      "grad_norm": 3.8432424068450928,
      "learning_rate": 4.117312072892939e-05,
      "loss": 0.2108,
      "step": 930
    },
    {
      "epoch": 0.5353075170842825,
      "grad_norm": 3.2506699562072754,
      "learning_rate": 4.1078208048595294e-05,
      "loss": 0.175,
      "step": 940
    },
    {
      "epoch": 0.541002277904328,
      "grad_norm": 3.8444082736968994,
      "learning_rate": 4.0983295368261205e-05,
      "loss": 0.1942,
      "step": 950
    },
    {
      "epoch": 0.5466970387243736,
      "grad_norm": 1.7612577676773071,
      "learning_rate": 4.088838268792711e-05,
      "loss": 0.1325,
      "step": 960
    },
    {
      "epoch": 0.5523917995444191,
      "grad_norm": 1.2005575895309448,
      "learning_rate": 4.079347000759301e-05,
      "loss": 0.1129,
      "step": 970
    },
    {
      "epoch": 0.5580865603644647,
      "grad_norm": 0.9540462493896484,
      "learning_rate": 4.0698557327258924e-05,
      "loss": 0.1258,
      "step": 980
    },
    {
      "epoch": 0.5637813211845103,
      "grad_norm": 2.20613431930542,
      "learning_rate": 4.0603644646924834e-05,
      "loss": 0.179,
      "step": 990
    },
    {
      "epoch": 0.5694760820045558,
      "grad_norm": 1.600826382637024,
      "learning_rate": 4.050873196659074e-05,
      "loss": 0.1809,
      "step": 1000
    },
    {
      "epoch": 0.5751708428246014,
      "grad_norm": 1.0167893171310425,
      "learning_rate": 4.041381928625664e-05,
      "loss": 0.0826,
      "step": 1010
    },
    {
      "epoch": 0.5808656036446469,
      "grad_norm": 1.0656858682632446,
      "learning_rate": 4.031890660592255e-05,
      "loss": 0.1323,
      "step": 1020
    },
    {
      "epoch": 0.5865603644646925,
      "grad_norm": 1.8568804264068604,
      "learning_rate": 4.022399392558846e-05,
      "loss": 0.1184,
      "step": 1030
    },
    {
      "epoch": 0.592255125284738,
      "grad_norm": 1.2088297605514526,
      "learning_rate": 4.012908124525437e-05,
      "loss": 0.115,
      "step": 1040
    },
    {
      "epoch": 0.5979498861047836,
      "grad_norm": 4.174439907073975,
      "learning_rate": 4.003416856492028e-05,
      "loss": 0.1541,
      "step": 1050
    },
    {
      "epoch": 0.6036446469248291,
      "grad_norm": 1.0429656505584717,
      "learning_rate": 3.993925588458618e-05,
      "loss": 0.1218,
      "step": 1060
    },
    {
      "epoch": 0.6093394077448747,
      "grad_norm": 4.0013651847839355,
      "learning_rate": 3.9844343204252087e-05,
      "loss": 0.1466,
      "step": 1070
    },
    {
      "epoch": 0.6150341685649203,
      "grad_norm": 1.9183622598648071,
      "learning_rate": 3.9749430523918e-05,
      "loss": 0.14,
      "step": 1080
    },
    {
      "epoch": 0.6207289293849658,
      "grad_norm": 2.64052152633667,
      "learning_rate": 3.965451784358391e-05,
      "loss": 0.1499,
      "step": 1090
    },
    {
      "epoch": 0.6264236902050114,
      "grad_norm": 2.6967151165008545,
      "learning_rate": 3.955960516324981e-05,
      "loss": 0.106,
      "step": 1100
    },
    {
      "epoch": 0.6321184510250569,
      "grad_norm": 1.5822768211364746,
      "learning_rate": 3.946469248291572e-05,
      "loss": 0.0858,
      "step": 1110
    },
    {
      "epoch": 0.6378132118451025,
      "grad_norm": 1.4362579584121704,
      "learning_rate": 3.936977980258163e-05,
      "loss": 0.1675,
      "step": 1120
    },
    {
      "epoch": 0.643507972665148,
      "grad_norm": 4.198774337768555,
      "learning_rate": 3.927486712224753e-05,
      "loss": 0.1887,
      "step": 1130
    },
    {
      "epoch": 0.6492027334851936,
      "grad_norm": 1.0296118259429932,
      "learning_rate": 3.917995444191344e-05,
      "loss": 0.174,
      "step": 1140
    },
    {
      "epoch": 0.6548974943052391,
      "grad_norm": 2.4037227630615234,
      "learning_rate": 3.908504176157935e-05,
      "loss": 0.1053,
      "step": 1150
    },
    {
      "epoch": 0.6605922551252847,
      "grad_norm": 1.8589047193527222,
      "learning_rate": 3.8990129081245256e-05,
      "loss": 0.1674,
      "step": 1160
    },
    {
      "epoch": 0.6662870159453302,
      "grad_norm": 0.8272297978401184,
      "learning_rate": 3.889521640091117e-05,
      "loss": 0.1646,
      "step": 1170
    },
    {
      "epoch": 0.6719817767653758,
      "grad_norm": 2.23038911819458,
      "learning_rate": 3.880030372057707e-05,
      "loss": 0.1499,
      "step": 1180
    },
    {
      "epoch": 0.6776765375854215,
      "grad_norm": 2.3637850284576416,
      "learning_rate": 3.8705391040242975e-05,
      "loss": 0.1577,
      "step": 1190
    },
    {
      "epoch": 0.683371298405467,
      "grad_norm": 3.252579927444458,
      "learning_rate": 3.8610478359908886e-05,
      "loss": 0.1223,
      "step": 1200
    },
    {
      "epoch": 0.6890660592255126,
      "grad_norm": 1.8192085027694702,
      "learning_rate": 3.8515565679574796e-05,
      "loss": 0.1594,
      "step": 1210
    },
    {
      "epoch": 0.6947608200455581,
      "grad_norm": 1.6105213165283203,
      "learning_rate": 3.84206529992407e-05,
      "loss": 0.1347,
      "step": 1220
    },
    {
      "epoch": 0.7004555808656037,
      "grad_norm": 1.618357539176941,
      "learning_rate": 3.8325740318906604e-05,
      "loss": 0.1309,
      "step": 1230
    },
    {
      "epoch": 0.7061503416856492,
      "grad_norm": 1.0436160564422607,
      "learning_rate": 3.8230827638572515e-05,
      "loss": 0.1221,
      "step": 1240
    },
    {
      "epoch": 0.7118451025056948,
      "grad_norm": 1.0755667686462402,
      "learning_rate": 3.813591495823842e-05,
      "loss": 0.138,
      "step": 1250
    },
    {
      "epoch": 0.7175398633257403,
      "grad_norm": 3.4023258686065674,
      "learning_rate": 3.804100227790433e-05,
      "loss": 0.2051,
      "step": 1260
    },
    {
      "epoch": 0.7232346241457859,
      "grad_norm": 1.2122399806976318,
      "learning_rate": 3.794608959757024e-05,
      "loss": 0.1616,
      "step": 1270
    },
    {
      "epoch": 0.7289293849658315,
      "grad_norm": 3.2242696285247803,
      "learning_rate": 3.7851176917236145e-05,
      "loss": 0.1678,
      "step": 1280
    },
    {
      "epoch": 0.734624145785877,
      "grad_norm": 0.8978271484375,
      "learning_rate": 3.775626423690205e-05,
      "loss": 0.1504,
      "step": 1290
    },
    {
      "epoch": 0.7403189066059226,
      "grad_norm": 2.9367473125457764,
      "learning_rate": 3.766135155656796e-05,
      "loss": 0.1859,
      "step": 1300
    },
    {
      "epoch": 0.7460136674259681,
      "grad_norm": 1.9458057880401611,
      "learning_rate": 3.756643887623387e-05,
      "loss": 0.1306,
      "step": 1310
    },
    {
      "epoch": 0.7517084282460137,
      "grad_norm": 1.1347579956054688,
      "learning_rate": 3.7471526195899774e-05,
      "loss": 0.1102,
      "step": 1320
    },
    {
      "epoch": 0.7574031890660592,
      "grad_norm": 1.1623060703277588,
      "learning_rate": 3.7376613515565685e-05,
      "loss": 0.126,
      "step": 1330
    },
    {
      "epoch": 0.7630979498861048,
      "grad_norm": 1.9671999216079712,
      "learning_rate": 3.728170083523159e-05,
      "loss": 0.1033,
      "step": 1340
    },
    {
      "epoch": 0.7687927107061503,
      "grad_norm": 0.981979489326477,
      "learning_rate": 3.718678815489749e-05,
      "loss": 0.1642,
      "step": 1350
    },
    {
      "epoch": 0.7744874715261959,
      "grad_norm": 4.021563529968262,
      "learning_rate": 3.7091875474563403e-05,
      "loss": 0.14,
      "step": 1360
    },
    {
      "epoch": 0.7801822323462415,
      "grad_norm": 0.7121520638465881,
      "learning_rate": 3.6996962794229314e-05,
      "loss": 0.1542,
      "step": 1370
    },
    {
      "epoch": 0.785876993166287,
      "grad_norm": 2.763511896133423,
      "learning_rate": 3.690205011389522e-05,
      "loss": 0.1408,
      "step": 1380
    },
    {
      "epoch": 0.7915717539863326,
      "grad_norm": 2.2274112701416016,
      "learning_rate": 3.680713743356112e-05,
      "loss": 0.131,
      "step": 1390
    },
    {
      "epoch": 0.7972665148063781,
      "grad_norm": 0.9881575703620911,
      "learning_rate": 3.671222475322703e-05,
      "loss": 0.1189,
      "step": 1400
    },
    {
      "epoch": 0.8029612756264237,
      "grad_norm": 1.166239619255066,
      "learning_rate": 3.661731207289294e-05,
      "loss": 0.1033,
      "step": 1410
    },
    {
      "epoch": 0.8086560364464692,
      "grad_norm": 0.5504360795021057,
      "learning_rate": 3.652239939255885e-05,
      "loss": 0.1182,
      "step": 1420
    },
    {
      "epoch": 0.8143507972665148,
      "grad_norm": 2.770014762878418,
      "learning_rate": 3.642748671222476e-05,
      "loss": 0.1144,
      "step": 1430
    },
    {
      "epoch": 0.8200455580865603,
      "grad_norm": 3.9895546436309814,
      "learning_rate": 3.633257403189066e-05,
      "loss": 0.1217,
      "step": 1440
    },
    {
      "epoch": 0.8257403189066059,
      "grad_norm": 3.492967128753662,
      "learning_rate": 3.6237661351556566e-05,
      "loss": 0.1508,
      "step": 1450
    },
    {
      "epoch": 0.8314350797266514,
      "grad_norm": 1.7888444662094116,
      "learning_rate": 3.614274867122248e-05,
      "loss": 0.1201,
      "step": 1460
    },
    {
      "epoch": 0.837129840546697,
      "grad_norm": 1.955121397972107,
      "learning_rate": 3.604783599088839e-05,
      "loss": 0.1618,
      "step": 1470
    },
    {
      "epoch": 0.8428246013667426,
      "grad_norm": 1.1534236669540405,
      "learning_rate": 3.595292331055429e-05,
      "loss": 0.1441,
      "step": 1480
    },
    {
      "epoch": 0.8485193621867881,
      "grad_norm": 1.145704746246338,
      "learning_rate": 3.58580106302202e-05,
      "loss": 0.1603,
      "step": 1490
    },
    {
      "epoch": 0.8542141230068337,
      "grad_norm": 1.3455891609191895,
      "learning_rate": 3.5763097949886106e-05,
      "loss": 0.1348,
      "step": 1500
    },
    {
      "epoch": 0.8599088838268792,
      "grad_norm": 1.2209105491638184,
      "learning_rate": 3.566818526955201e-05,
      "loss": 0.1008,
      "step": 1510
    },
    {
      "epoch": 0.8656036446469249,
      "grad_norm": 1.7053004503250122,
      "learning_rate": 3.557327258921792e-05,
      "loss": 0.0978,
      "step": 1520
    },
    {
      "epoch": 0.8712984054669703,
      "grad_norm": 2.105926036834717,
      "learning_rate": 3.547835990888383e-05,
      "loss": 0.1482,
      "step": 1530
    },
    {
      "epoch": 0.876993166287016,
      "grad_norm": 2.1003365516662598,
      "learning_rate": 3.5383447228549736e-05,
      "loss": 0.1308,
      "step": 1540
    },
    {
      "epoch": 0.8826879271070615,
      "grad_norm": 1.7915514707565308,
      "learning_rate": 3.528853454821565e-05,
      "loss": 0.1432,
      "step": 1550
    },
    {
      "epoch": 0.8883826879271071,
      "grad_norm": 3.394533395767212,
      "learning_rate": 3.519362186788155e-05,
      "loss": 0.1982,
      "step": 1560
    },
    {
      "epoch": 0.8940774487471527,
      "grad_norm": 3.049562931060791,
      "learning_rate": 3.5098709187547455e-05,
      "loss": 0.1317,
      "step": 1570
    },
    {
      "epoch": 0.8997722095671982,
      "grad_norm": 0.992368757724762,
      "learning_rate": 3.5003796507213365e-05,
      "loss": 0.1017,
      "step": 1580
    },
    {
      "epoch": 0.9054669703872438,
      "grad_norm": 1.088188886642456,
      "learning_rate": 3.4908883826879276e-05,
      "loss": 0.0877,
      "step": 1590
    },
    {
      "epoch": 0.9111617312072893,
      "grad_norm": 0.7528336048126221,
      "learning_rate": 3.481397114654518e-05,
      "loss": 0.1053,
      "step": 1600
    },
    {
      "epoch": 0.9168564920273349,
      "grad_norm": 5.400979042053223,
      "learning_rate": 3.4719058466211084e-05,
      "loss": 0.1763,
      "step": 1610
    },
    {
      "epoch": 0.9225512528473804,
      "grad_norm": 0.7936645150184631,
      "learning_rate": 3.4624145785876995e-05,
      "loss": 0.1084,
      "step": 1620
    },
    {
      "epoch": 0.928246013667426,
      "grad_norm": 1.9876333475112915,
      "learning_rate": 3.45292331055429e-05,
      "loss": 0.1491,
      "step": 1630
    },
    {
      "epoch": 0.9339407744874715,
      "grad_norm": 1.7372020483016968,
      "learning_rate": 3.443432042520881e-05,
      "loss": 0.0863,
      "step": 1640
    },
    {
      "epoch": 0.9396355353075171,
      "grad_norm": 0.9634379148483276,
      "learning_rate": 3.433940774487472e-05,
      "loss": 0.0897,
      "step": 1650
    },
    {
      "epoch": 0.9453302961275627,
      "grad_norm": 0.9360265135765076,
      "learning_rate": 3.4244495064540624e-05,
      "loss": 0.1238,
      "step": 1660
    },
    {
      "epoch": 0.9510250569476082,
      "grad_norm": 1.3839788436889648,
      "learning_rate": 3.414958238420653e-05,
      "loss": 0.1259,
      "step": 1670
    },
    {
      "epoch": 0.9567198177676538,
      "grad_norm": 2.2220139503479004,
      "learning_rate": 3.405466970387244e-05,
      "loss": 0.1176,
      "step": 1680
    },
    {
      "epoch": 0.9624145785876993,
      "grad_norm": 1.8980664014816284,
      "learning_rate": 3.395975702353835e-05,
      "loss": 0.1412,
      "step": 1690
    },
    {
      "epoch": 0.9681093394077449,
      "grad_norm": 7.803280830383301,
      "learning_rate": 3.3864844343204254e-05,
      "loss": 0.1334,
      "step": 1700
    },
    {
      "epoch": 0.9738041002277904,
      "grad_norm": 0.679198682308197,
      "learning_rate": 3.3769931662870164e-05,
      "loss": 0.0845,
      "step": 1710
    },
    {
      "epoch": 0.979498861047836,
      "grad_norm": 2.330993175506592,
      "learning_rate": 3.367501898253607e-05,
      "loss": 0.1032,
      "step": 1720
    },
    {
      "epoch": 0.9851936218678815,
      "grad_norm": 2.838163375854492,
      "learning_rate": 3.358010630220197e-05,
      "loss": 0.1171,
      "step": 1730
    },
    {
      "epoch": 0.9908883826879271,
      "grad_norm": 1.7064838409423828,
      "learning_rate": 3.348519362186788e-05,
      "loss": 0.0948,
      "step": 1740
    },
    {
      "epoch": 0.9965831435079726,
      "grad_norm": 4.830978870391846,
      "learning_rate": 3.3390280941533794e-05,
      "loss": 0.1238,
      "step": 1750
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.968639214616853,
      "eval_f1": 0.7834769305097028,
      "eval_loss": 0.1241590604186058,
      "eval_precision": 0.7507716049382716,
      "eval_recall": 0.8191614749957905,
      "eval_runtime": 25.3499,
      "eval_samples_per_second": 128.206,
      "eval_steps_per_second": 16.055,
      "step": 1756
    },
    {
      "epoch": 1.0022779043280183,
      "grad_norm": 0.97957444190979,
      "learning_rate": 3.32953682611997e-05,
      "loss": 0.1341,
      "step": 1760
    },
    {
      "epoch": 1.0079726651480638,
      "grad_norm": 1.6682242155075073,
      "learning_rate": 3.32004555808656e-05,
      "loss": 0.0814,
      "step": 1770
    },
    {
      "epoch": 1.0136674259681093,
      "grad_norm": 1.047098159790039,
      "learning_rate": 3.310554290053151e-05,
      "loss": 0.1028,
      "step": 1780
    },
    {
      "epoch": 1.0193621867881548,
      "grad_norm": 2.5938913822174072,
      "learning_rate": 3.3010630220197417e-05,
      "loss": 0.1297,
      "step": 1790
    },
    {
      "epoch": 1.0250569476082005,
      "grad_norm": 2.022613525390625,
      "learning_rate": 3.291571753986333e-05,
      "loss": 0.0786,
      "step": 1800
    },
    {
      "epoch": 1.030751708428246,
      "grad_norm": 1.2228267192840576,
      "learning_rate": 3.282080485952924e-05,
      "loss": 0.1056,
      "step": 1810
    },
    {
      "epoch": 1.0364464692482915,
      "grad_norm": 0.591299295425415,
      "learning_rate": 3.272589217919514e-05,
      "loss": 0.0958,
      "step": 1820
    },
    {
      "epoch": 1.042141230068337,
      "grad_norm": 1.737338900566101,
      "learning_rate": 3.2630979498861046e-05,
      "loss": 0.0887,
      "step": 1830
    },
    {
      "epoch": 1.0478359908883828,
      "grad_norm": 1.6468360424041748,
      "learning_rate": 3.253606681852696e-05,
      "loss": 0.0749,
      "step": 1840
    },
    {
      "epoch": 1.0535307517084282,
      "grad_norm": 0.9322206377983093,
      "learning_rate": 3.244115413819286e-05,
      "loss": 0.0978,
      "step": 1850
    },
    {
      "epoch": 1.0592255125284737,
      "grad_norm": 0.8681425452232361,
      "learning_rate": 3.234624145785877e-05,
      "loss": 0.0736,
      "step": 1860
    },
    {
      "epoch": 1.0649202733485195,
      "grad_norm": 1.229645013809204,
      "learning_rate": 3.225132877752468e-05,
      "loss": 0.0875,
      "step": 1870
    },
    {
      "epoch": 1.070615034168565,
      "grad_norm": 0.8657091856002808,
      "learning_rate": 3.2156416097190586e-05,
      "loss": 0.0984,
      "step": 1880
    },
    {
      "epoch": 1.0763097949886105,
      "grad_norm": 1.5904412269592285,
      "learning_rate": 3.206150341685649e-05,
      "loss": 0.1398,
      "step": 1890
    },
    {
      "epoch": 1.082004555808656,
      "grad_norm": 1.811844825744629,
      "learning_rate": 3.19665907365224e-05,
      "loss": 0.0844,
      "step": 1900
    },
    {
      "epoch": 1.0876993166287017,
      "grad_norm": 3.4504454135894775,
      "learning_rate": 3.187167805618831e-05,
      "loss": 0.1053,
      "step": 1910
    },
    {
      "epoch": 1.0933940774487472,
      "grad_norm": 2.874379873275757,
      "learning_rate": 3.1776765375854216e-05,
      "loss": 0.0868,
      "step": 1920
    },
    {
      "epoch": 1.0990888382687927,
      "grad_norm": 1.921471357345581,
      "learning_rate": 3.1681852695520126e-05,
      "loss": 0.0906,
      "step": 1930
    },
    {
      "epoch": 1.1047835990888384,
      "grad_norm": 1.5753288269042969,
      "learning_rate": 3.158694001518603e-05,
      "loss": 0.0923,
      "step": 1940
    },
    {
      "epoch": 1.1104783599088839,
      "grad_norm": 2.416849136352539,
      "learning_rate": 3.1492027334851934e-05,
      "loss": 0.0872,
      "step": 1950
    },
    {
      "epoch": 1.1161731207289294,
      "grad_norm": 1.3285380601882935,
      "learning_rate": 3.1397114654517845e-05,
      "loss": 0.1078,
      "step": 1960
    },
    {
      "epoch": 1.1218678815489749,
      "grad_norm": 2.1748809814453125,
      "learning_rate": 3.1302201974183756e-05,
      "loss": 0.0991,
      "step": 1970
    },
    {
      "epoch": 1.1275626423690206,
      "grad_norm": 1.5692195892333984,
      "learning_rate": 3.120728929384966e-05,
      "loss": 0.0852,
      "step": 1980
    },
    {
      "epoch": 1.133257403189066,
      "grad_norm": 1.2222596406936646,
      "learning_rate": 3.1112376613515564e-05,
      "loss": 0.1153,
      "step": 1990
    },
    {
      "epoch": 1.1389521640091116,
      "grad_norm": 2.329113245010376,
      "learning_rate": 3.1017463933181475e-05,
      "loss": 0.1013,
      "step": 2000
    },
    {
      "epoch": 1.144646924829157,
      "grad_norm": 0.9628048539161682,
      "learning_rate": 3.092255125284738e-05,
      "loss": 0.0983,
      "step": 2010
    },
    {
      "epoch": 1.1503416856492028,
      "grad_norm": 0.4654466211795807,
      "learning_rate": 3.082763857251329e-05,
      "loss": 0.0957,
      "step": 2020
    },
    {
      "epoch": 1.1560364464692483,
      "grad_norm": 0.722862184047699,
      "learning_rate": 3.07327258921792e-05,
      "loss": 0.0974,
      "step": 2030
    },
    {
      "epoch": 1.1617312072892938,
      "grad_norm": 4.287933349609375,
      "learning_rate": 3.0637813211845104e-05,
      "loss": 0.152,
      "step": 2040
    },
    {
      "epoch": 1.1674259681093395,
      "grad_norm": 1.48385488986969,
      "learning_rate": 3.054290053151101e-05,
      "loss": 0.0812,
      "step": 2050
    },
    {
      "epoch": 1.173120728929385,
      "grad_norm": 0.6806052327156067,
      "learning_rate": 3.044798785117692e-05,
      "loss": 0.082,
      "step": 2060
    },
    {
      "epoch": 1.1788154897494305,
      "grad_norm": 1.8856145143508911,
      "learning_rate": 3.0353075170842826e-05,
      "loss": 0.0913,
      "step": 2070
    },
    {
      "epoch": 1.184510250569476,
      "grad_norm": 1.6212036609649658,
      "learning_rate": 3.0258162490508733e-05,
      "loss": 0.0893,
      "step": 2080
    },
    {
      "epoch": 1.1902050113895217,
      "grad_norm": 2.615706205368042,
      "learning_rate": 3.0163249810174644e-05,
      "loss": 0.0886,
      "step": 2090
    },
    {
      "epoch": 1.1958997722095672,
      "grad_norm": 1.3023104667663574,
      "learning_rate": 3.0068337129840545e-05,
      "loss": 0.0953,
      "step": 2100
    },
    {
      "epoch": 1.2015945330296127,
      "grad_norm": 1.685548186302185,
      "learning_rate": 2.9973424449506455e-05,
      "loss": 0.147,
      "step": 2110
    },
    {
      "epoch": 1.2072892938496582,
      "grad_norm": 1.6766068935394287,
      "learning_rate": 2.9878511769172363e-05,
      "loss": 0.0698,
      "step": 2120
    },
    {
      "epoch": 1.212984054669704,
      "grad_norm": 1.5873198509216309,
      "learning_rate": 2.978359908883827e-05,
      "loss": 0.0766,
      "step": 2130
    },
    {
      "epoch": 1.2186788154897494,
      "grad_norm": 3.058396339416504,
      "learning_rate": 2.968868640850418e-05,
      "loss": 0.1245,
      "step": 2140
    },
    {
      "epoch": 1.224373576309795,
      "grad_norm": 0.7530670762062073,
      "learning_rate": 2.959377372817008e-05,
      "loss": 0.1088,
      "step": 2150
    },
    {
      "epoch": 1.2300683371298406,
      "grad_norm": 0.7259340882301331,
      "learning_rate": 2.949886104783599e-05,
      "loss": 0.0962,
      "step": 2160
    },
    {
      "epoch": 1.2357630979498861,
      "grad_norm": 0.7594104409217834,
      "learning_rate": 2.94039483675019e-05,
      "loss": 0.098,
      "step": 2170
    },
    {
      "epoch": 1.2414578587699316,
      "grad_norm": 0.6500957012176514,
      "learning_rate": 2.9309035687167807e-05,
      "loss": 0.0781,
      "step": 2180
    },
    {
      "epoch": 1.2471526195899771,
      "grad_norm": 1.0029600858688354,
      "learning_rate": 2.9214123006833714e-05,
      "loss": 0.0986,
      "step": 2190
    },
    {
      "epoch": 1.2528473804100229,
      "grad_norm": 2.1104345321655273,
      "learning_rate": 2.9119210326499625e-05,
      "loss": 0.111,
      "step": 2200
    },
    {
      "epoch": 1.2585421412300684,
      "grad_norm": 0.5959046483039856,
      "learning_rate": 2.9024297646165526e-05,
      "loss": 0.0705,
      "step": 2210
    },
    {
      "epoch": 1.2642369020501139,
      "grad_norm": 0.35329675674438477,
      "learning_rate": 2.8929384965831436e-05,
      "loss": 0.0849,
      "step": 2220
    },
    {
      "epoch": 1.2699316628701594,
      "grad_norm": 1.3002897500991821,
      "learning_rate": 2.8834472285497344e-05,
      "loss": 0.0823,
      "step": 2230
    },
    {
      "epoch": 1.275626423690205,
      "grad_norm": 1.5088927745819092,
      "learning_rate": 2.873955960516325e-05,
      "loss": 0.0931,
      "step": 2240
    },
    {
      "epoch": 1.2813211845102506,
      "grad_norm": 0.8356180191040039,
      "learning_rate": 2.8644646924829162e-05,
      "loss": 0.0908,
      "step": 2250
    },
    {
      "epoch": 1.287015945330296,
      "grad_norm": 1.3802344799041748,
      "learning_rate": 2.8549734244495063e-05,
      "loss": 0.1129,
      "step": 2260
    },
    {
      "epoch": 1.2927107061503418,
      "grad_norm": 3.4083027839660645,
      "learning_rate": 2.8454821564160973e-05,
      "loss": 0.0976,
      "step": 2270
    },
    {
      "epoch": 1.2984054669703873,
      "grad_norm": 2.6577627658843994,
      "learning_rate": 2.835990888382688e-05,
      "loss": 0.1106,
      "step": 2280
    },
    {
      "epoch": 1.3041002277904328,
      "grad_norm": 1.930678129196167,
      "learning_rate": 2.8264996203492788e-05,
      "loss": 0.1213,
      "step": 2290
    },
    {
      "epoch": 1.3097949886104785,
      "grad_norm": 0.6453108787536621,
      "learning_rate": 2.8170083523158695e-05,
      "loss": 0.0705,
      "step": 2300
    },
    {
      "epoch": 1.315489749430524,
      "grad_norm": 1.228203535079956,
      "learning_rate": 2.8075170842824606e-05,
      "loss": 0.1067,
      "step": 2310
    },
    {
      "epoch": 1.3211845102505695,
      "grad_norm": 2.278114080429077,
      "learning_rate": 2.7980258162490507e-05,
      "loss": 0.096,
      "step": 2320
    },
    {
      "epoch": 1.326879271070615,
      "grad_norm": 0.5799562335014343,
      "learning_rate": 2.7885345482156417e-05,
      "loss": 0.0707,
      "step": 2330
    },
    {
      "epoch": 1.3325740318906605,
      "grad_norm": 3.1779398918151855,
      "learning_rate": 2.7790432801822325e-05,
      "loss": 0.0929,
      "step": 2340
    },
    {
      "epoch": 1.3382687927107062,
      "grad_norm": 2.468745470046997,
      "learning_rate": 2.7695520121488232e-05,
      "loss": 0.0967,
      "step": 2350
    },
    {
      "epoch": 1.3439635535307517,
      "grad_norm": 1.107359528541565,
      "learning_rate": 2.7600607441154143e-05,
      "loss": 0.0798,
      "step": 2360
    },
    {
      "epoch": 1.3496583143507972,
      "grad_norm": 1.7985248565673828,
      "learning_rate": 2.7505694760820043e-05,
      "loss": 0.0876,
      "step": 2370
    },
    {
      "epoch": 1.355353075170843,
      "grad_norm": 1.1344681978225708,
      "learning_rate": 2.7410782080485954e-05,
      "loss": 0.0963,
      "step": 2380
    },
    {
      "epoch": 1.3610478359908884,
      "grad_norm": 1.5203508138656616,
      "learning_rate": 2.731586940015186e-05,
      "loss": 0.0715,
      "step": 2390
    },
    {
      "epoch": 1.366742596810934,
      "grad_norm": 2.8507308959960938,
      "learning_rate": 2.722095671981777e-05,
      "loss": 0.1341,
      "step": 2400
    },
    {
      "epoch": 1.3724373576309796,
      "grad_norm": 2.638719081878662,
      "learning_rate": 2.7126044039483676e-05,
      "loss": 0.1249,
      "step": 2410
    },
    {
      "epoch": 1.3781321184510251,
      "grad_norm": 0.7994898557662964,
      "learning_rate": 2.7031131359149587e-05,
      "loss": 0.1074,
      "step": 2420
    },
    {
      "epoch": 1.3838268792710706,
      "grad_norm": 1.2879323959350586,
      "learning_rate": 2.6936218678815488e-05,
      "loss": 0.0682,
      "step": 2430
    },
    {
      "epoch": 1.3895216400911161,
      "grad_norm": 0.7568653225898743,
      "learning_rate": 2.68413059984814e-05,
      "loss": 0.0929,
      "step": 2440
    },
    {
      "epoch": 1.3952164009111616,
      "grad_norm": 1.0090585947036743,
      "learning_rate": 2.6746393318147306e-05,
      "loss": 0.0945,
      "step": 2450
    },
    {
      "epoch": 1.4009111617312073,
      "grad_norm": 2.003541946411133,
      "learning_rate": 2.6651480637813213e-05,
      "loss": 0.0874,
      "step": 2460
    },
    {
      "epoch": 1.4066059225512528,
      "grad_norm": 2.603950023651123,
      "learning_rate": 2.6556567957479124e-05,
      "loss": 0.0616,
      "step": 2470
    },
    {
      "epoch": 1.4123006833712983,
      "grad_norm": 1.5461757183074951,
      "learning_rate": 2.6461655277145024e-05,
      "loss": 0.0893,
      "step": 2480
    },
    {
      "epoch": 1.417995444191344,
      "grad_norm": 1.2687543630599976,
      "learning_rate": 2.6366742596810935e-05,
      "loss": 0.0989,
      "step": 2490
    },
    {
      "epoch": 1.4236902050113895,
      "grad_norm": 1.3670613765716553,
      "learning_rate": 2.6271829916476843e-05,
      "loss": 0.1227,
      "step": 2500
    },
    {
      "epoch": 1.429384965831435,
      "grad_norm": 3.420806884765625,
      "learning_rate": 2.617691723614275e-05,
      "loss": 0.1326,
      "step": 2510
    },
    {
      "epoch": 1.4350797266514808,
      "grad_norm": 2.4182112216949463,
      "learning_rate": 2.608200455580866e-05,
      "loss": 0.1245,
      "step": 2520
    },
    {
      "epoch": 1.4407744874715263,
      "grad_norm": 0.6075257658958435,
      "learning_rate": 2.5987091875474568e-05,
      "loss": 0.0473,
      "step": 2530
    },
    {
      "epoch": 1.4464692482915718,
      "grad_norm": 2.5870461463928223,
      "learning_rate": 2.589217919514047e-05,
      "loss": 0.1004,
      "step": 2540
    },
    {
      "epoch": 1.4521640091116172,
      "grad_norm": 1.4278345108032227,
      "learning_rate": 2.579726651480638e-05,
      "loss": 0.1374,
      "step": 2550
    },
    {
      "epoch": 1.4578587699316627,
      "grad_norm": 2.7487120628356934,
      "learning_rate": 2.5702353834472287e-05,
      "loss": 0.0825,
      "step": 2560
    },
    {
      "epoch": 1.4635535307517085,
      "grad_norm": 0.9678652286529541,
      "learning_rate": 2.5607441154138194e-05,
      "loss": 0.0854,
      "step": 2570
    },
    {
      "epoch": 1.469248291571754,
      "grad_norm": 2.752776861190796,
      "learning_rate": 2.5512528473804105e-05,
      "loss": 0.0891,
      "step": 2580
    },
    {
      "epoch": 1.4749430523917995,
      "grad_norm": 2.2584118843078613,
      "learning_rate": 2.5417615793470005e-05,
      "loss": 0.0905,
      "step": 2590
    },
    {
      "epoch": 1.4806378132118452,
      "grad_norm": 0.9626738429069519,
      "learning_rate": 2.5322703113135916e-05,
      "loss": 0.0904,
      "step": 2600
    },
    {
      "epoch": 1.4863325740318907,
      "grad_norm": 2.8413443565368652,
      "learning_rate": 2.5227790432801824e-05,
      "loss": 0.0863,
      "step": 2610
    },
    {
      "epoch": 1.4920273348519362,
      "grad_norm": 2.619218349456787,
      "learning_rate": 2.513287775246773e-05,
      "loss": 0.1038,
      "step": 2620
    },
    {
      "epoch": 1.497722095671982,
      "grad_norm": 0.9081758260726929,
      "learning_rate": 2.503796507213364e-05,
      "loss": 0.08,
      "step": 2630
    },
    {
      "epoch": 1.5034168564920274,
      "grad_norm": 0.5638437867164612,
      "learning_rate": 2.4943052391799546e-05,
      "loss": 0.067,
      "step": 2640
    },
    {
      "epoch": 1.5091116173120729,
      "grad_norm": 0.5109384655952454,
      "learning_rate": 2.4848139711465453e-05,
      "loss": 0.09,
      "step": 2650
    },
    {
      "epoch": 1.5148063781321186,
      "grad_norm": 1.9207391738891602,
      "learning_rate": 2.475322703113136e-05,
      "loss": 0.1144,
      "step": 2660
    },
    {
      "epoch": 1.5205011389521639,
      "grad_norm": 8.998539924621582,
      "learning_rate": 2.4658314350797268e-05,
      "loss": 0.0964,
      "step": 2670
    },
    {
      "epoch": 1.5261958997722096,
      "grad_norm": 0.7903996109962463,
      "learning_rate": 2.4563401670463175e-05,
      "loss": 0.0715,
      "step": 2680
    },
    {
      "epoch": 1.531890660592255,
      "grad_norm": 0.3726605772972107,
      "learning_rate": 2.4468488990129082e-05,
      "loss": 0.1521,
      "step": 2690
    },
    {
      "epoch": 1.5375854214123006,
      "grad_norm": 0.42937004566192627,
      "learning_rate": 2.437357630979499e-05,
      "loss": 0.0708,
      "step": 2700
    },
    {
      "epoch": 1.5432801822323463,
      "grad_norm": 1.1452207565307617,
      "learning_rate": 2.4278663629460897e-05,
      "loss": 0.1002,
      "step": 2710
    },
    {
      "epoch": 1.5489749430523918,
      "grad_norm": 1.046943187713623,
      "learning_rate": 2.4183750949126804e-05,
      "loss": 0.1268,
      "step": 2720
    },
    {
      "epoch": 1.5546697038724373,
      "grad_norm": 3.5698766708374023,
      "learning_rate": 2.4088838268792712e-05,
      "loss": 0.0923,
      "step": 2730
    },
    {
      "epoch": 1.560364464692483,
      "grad_norm": 2.4647140502929688,
      "learning_rate": 2.399392558845862e-05,
      "loss": 0.0645,
      "step": 2740
    },
    {
      "epoch": 1.5660592255125285,
      "grad_norm": 2.386082887649536,
      "learning_rate": 2.3899012908124527e-05,
      "loss": 0.1273,
      "step": 2750
    },
    {
      "epoch": 1.571753986332574,
      "grad_norm": 0.756199061870575,
      "learning_rate": 2.3804100227790434e-05,
      "loss": 0.1246,
      "step": 2760
    },
    {
      "epoch": 1.5774487471526197,
      "grad_norm": 2.8555424213409424,
      "learning_rate": 2.370918754745634e-05,
      "loss": 0.0928,
      "step": 2770
    },
    {
      "epoch": 1.583143507972665,
      "grad_norm": 0.9831840991973877,
      "learning_rate": 2.361427486712225e-05,
      "loss": 0.0749,
      "step": 2780
    },
    {
      "epoch": 1.5888382687927107,
      "grad_norm": 1.1987966299057007,
      "learning_rate": 2.3519362186788156e-05,
      "loss": 0.1,
      "step": 2790
    },
    {
      "epoch": 1.5945330296127562,
      "grad_norm": 1.14650559425354,
      "learning_rate": 2.3424449506454063e-05,
      "loss": 0.0891,
      "step": 2800
    },
    {
      "epoch": 1.6002277904328017,
      "grad_norm": 2.077380657196045,
      "learning_rate": 2.332953682611997e-05,
      "loss": 0.0838,
      "step": 2810
    },
    {
      "epoch": 1.6059225512528474,
      "grad_norm": 0.967408299446106,
      "learning_rate": 2.3234624145785878e-05,
      "loss": 0.0725,
      "step": 2820
    },
    {
      "epoch": 1.611617312072893,
      "grad_norm": 0.815949022769928,
      "learning_rate": 2.3139711465451785e-05,
      "loss": 0.1292,
      "step": 2830
    },
    {
      "epoch": 1.6173120728929384,
      "grad_norm": 3.764101028442383,
      "learning_rate": 2.3044798785117693e-05,
      "loss": 0.1105,
      "step": 2840
    },
    {
      "epoch": 1.6230068337129842,
      "grad_norm": 1.8564128875732422,
      "learning_rate": 2.29498861047836e-05,
      "loss": 0.1112,
      "step": 2850
    },
    {
      "epoch": 1.6287015945330297,
      "grad_norm": 0.6224936246871948,
      "learning_rate": 2.2854973424449508e-05,
      "loss": 0.0828,
      "step": 2860
    },
    {
      "epoch": 1.6343963553530751,
      "grad_norm": 0.6306477785110474,
      "learning_rate": 2.2760060744115415e-05,
      "loss": 0.0787,
      "step": 2870
    },
    {
      "epoch": 1.6400911161731209,
      "grad_norm": 1.4436335563659668,
      "learning_rate": 2.2665148063781322e-05,
      "loss": 0.0877,
      "step": 2880
    },
    {
      "epoch": 1.6457858769931661,
      "grad_norm": 2.165911912918091,
      "learning_rate": 2.257023538344723e-05,
      "loss": 0.0771,
      "step": 2890
    },
    {
      "epoch": 1.6514806378132119,
      "grad_norm": 2.1198298931121826,
      "learning_rate": 2.2475322703113137e-05,
      "loss": 0.0801,
      "step": 2900
    },
    {
      "epoch": 1.6571753986332574,
      "grad_norm": 0.6920590400695801,
      "learning_rate": 2.2380410022779044e-05,
      "loss": 0.1039,
      "step": 2910
    },
    {
      "epoch": 1.6628701594533029,
      "grad_norm": 1.152017593383789,
      "learning_rate": 2.2285497342444952e-05,
      "loss": 0.1118,
      "step": 2920
    },
    {
      "epoch": 1.6685649202733486,
      "grad_norm": 1.1572786569595337,
      "learning_rate": 2.219058466211086e-05,
      "loss": 0.0853,
      "step": 2930
    },
    {
      "epoch": 1.674259681093394,
      "grad_norm": 1.9907824993133545,
      "learning_rate": 2.2095671981776766e-05,
      "loss": 0.1256,
      "step": 2940
    },
    {
      "epoch": 1.6799544419134396,
      "grad_norm": 0.39809224009513855,
      "learning_rate": 2.2000759301442674e-05,
      "loss": 0.0824,
      "step": 2950
    },
    {
      "epoch": 1.6856492027334853,
      "grad_norm": 1.5977418422698975,
      "learning_rate": 2.190584662110858e-05,
      "loss": 0.0604,
      "step": 2960
    },
    {
      "epoch": 1.6913439635535308,
      "grad_norm": 0.5906954407691956,
      "learning_rate": 2.181093394077449e-05,
      "loss": 0.0892,
      "step": 2970
    },
    {
      "epoch": 1.6970387243735763,
      "grad_norm": 1.2924199104309082,
      "learning_rate": 2.1716021260440396e-05,
      "loss": 0.0665,
      "step": 2980
    },
    {
      "epoch": 1.702733485193622,
      "grad_norm": 1.207828402519226,
      "learning_rate": 2.1621108580106303e-05,
      "loss": 0.1109,
      "step": 2990
    },
    {
      "epoch": 1.7084282460136673,
      "grad_norm": 2.1666877269744873,
      "learning_rate": 2.152619589977221e-05,
      "loss": 0.0754,
      "step": 3000
    },
    {
      "epoch": 1.714123006833713,
      "grad_norm": 2.3492608070373535,
      "learning_rate": 2.1431283219438118e-05,
      "loss": 0.1017,
      "step": 3010
    },
    {
      "epoch": 1.7198177676537585,
      "grad_norm": 1.4836989641189575,
      "learning_rate": 2.1336370539104025e-05,
      "loss": 0.0811,
      "step": 3020
    },
    {
      "epoch": 1.725512528473804,
      "grad_norm": 2.9044885635375977,
      "learning_rate": 2.1241457858769933e-05,
      "loss": 0.0541,
      "step": 3030
    },
    {
      "epoch": 1.7312072892938497,
      "grad_norm": 2.6181652545928955,
      "learning_rate": 2.114654517843584e-05,
      "loss": 0.0839,
      "step": 3040
    },
    {
      "epoch": 1.7369020501138952,
      "grad_norm": 0.5739513039588928,
      "learning_rate": 2.1051632498101747e-05,
      "loss": 0.0795,
      "step": 3050
    },
    {
      "epoch": 1.7425968109339407,
      "grad_norm": 1.8764702081680298,
      "learning_rate": 2.0956719817767655e-05,
      "loss": 0.0945,
      "step": 3060
    },
    {
      "epoch": 1.7482915717539864,
      "grad_norm": 1.5999131202697754,
      "learning_rate": 2.0861807137433562e-05,
      "loss": 0.0646,
      "step": 3070
    },
    {
      "epoch": 1.753986332574032,
      "grad_norm": 1.1158217191696167,
      "learning_rate": 2.076689445709947e-05,
      "loss": 0.0908,
      "step": 3080
    },
    {
      "epoch": 1.7596810933940774,
      "grad_norm": 0.7986214756965637,
      "learning_rate": 2.0671981776765377e-05,
      "loss": 0.153,
      "step": 3090
    },
    {
      "epoch": 1.7653758542141231,
      "grad_norm": 2.1894545555114746,
      "learning_rate": 2.0577069096431284e-05,
      "loss": 0.1073,
      "step": 3100
    },
    {
      "epoch": 1.7710706150341684,
      "grad_norm": 0.9665547013282776,
      "learning_rate": 2.048215641609719e-05,
      "loss": 0.0888,
      "step": 3110
    },
    {
      "epoch": 1.7767653758542141,
      "grad_norm": 0.6699535846710205,
      "learning_rate": 2.03872437357631e-05,
      "loss": 0.0872,
      "step": 3120
    },
    {
      "epoch": 1.7824601366742598,
      "grad_norm": 2.1171014308929443,
      "learning_rate": 2.0292331055429006e-05,
      "loss": 0.115,
      "step": 3130
    },
    {
      "epoch": 1.7881548974943051,
      "grad_norm": 1.2174530029296875,
      "learning_rate": 2.0197418375094914e-05,
      "loss": 0.0852,
      "step": 3140
    },
    {
      "epoch": 1.7938496583143508,
      "grad_norm": 2.183448553085327,
      "learning_rate": 2.010250569476082e-05,
      "loss": 0.0955,
      "step": 3150
    },
    {
      "epoch": 1.7995444191343963,
      "grad_norm": 0.9667510390281677,
      "learning_rate": 2.000759301442673e-05,
      "loss": 0.0902,
      "step": 3160
    },
    {
      "epoch": 1.8052391799544418,
      "grad_norm": 3.9229986667633057,
      "learning_rate": 1.9912680334092636e-05,
      "loss": 0.0941,
      "step": 3170
    },
    {
      "epoch": 1.8109339407744875,
      "grad_norm": 0.8753280639648438,
      "learning_rate": 1.9817767653758543e-05,
      "loss": 0.0776,
      "step": 3180
    },
    {
      "epoch": 1.816628701594533,
      "grad_norm": 1.5324796438217163,
      "learning_rate": 1.972285497342445e-05,
      "loss": 0.1014,
      "step": 3190
    },
    {
      "epoch": 1.8223234624145785,
      "grad_norm": 0.4302336275577545,
      "learning_rate": 1.9627942293090358e-05,
      "loss": 0.1232,
      "step": 3200
    },
    {
      "epoch": 1.8280182232346243,
      "grad_norm": 1.9868009090423584,
      "learning_rate": 1.9533029612756265e-05,
      "loss": 0.1117,
      "step": 3210
    },
    {
      "epoch": 1.8337129840546698,
      "grad_norm": 0.7633128762245178,
      "learning_rate": 1.9438116932422173e-05,
      "loss": 0.0646,
      "step": 3220
    },
    {
      "epoch": 1.8394077448747153,
      "grad_norm": 1.8619776964187622,
      "learning_rate": 1.934320425208808e-05,
      "loss": 0.1116,
      "step": 3230
    },
    {
      "epoch": 1.845102505694761,
      "grad_norm": 2.578927993774414,
      "learning_rate": 1.9248291571753987e-05,
      "loss": 0.0889,
      "step": 3240
    },
    {
      "epoch": 1.8507972665148062,
      "grad_norm": 2.5693464279174805,
      "learning_rate": 1.9153378891419895e-05,
      "loss": 0.0842,
      "step": 3250
    },
    {
      "epoch": 1.856492027334852,
      "grad_norm": 1.0885404348373413,
      "learning_rate": 1.9058466211085802e-05,
      "loss": 0.0909,
      "step": 3260
    },
    {
      "epoch": 1.8621867881548975,
      "grad_norm": 2.0621140003204346,
      "learning_rate": 1.896355353075171e-05,
      "loss": 0.0771,
      "step": 3270
    },
    {
      "epoch": 1.867881548974943,
      "grad_norm": 1.275061011314392,
      "learning_rate": 1.8868640850417617e-05,
      "loss": 0.1135,
      "step": 3280
    },
    {
      "epoch": 1.8735763097949887,
      "grad_norm": 1.1576536893844604,
      "learning_rate": 1.8773728170083524e-05,
      "loss": 0.0948,
      "step": 3290
    },
    {
      "epoch": 1.8792710706150342,
      "grad_norm": 0.5448009967803955,
      "learning_rate": 1.867881548974943e-05,
      "loss": 0.0908,
      "step": 3300
    },
    {
      "epoch": 1.8849658314350797,
      "grad_norm": 1.3317946195602417,
      "learning_rate": 1.858390280941534e-05,
      "loss": 0.1246,
      "step": 3310
    },
    {
      "epoch": 1.8906605922551254,
      "grad_norm": 1.0352773666381836,
      "learning_rate": 1.8488990129081246e-05,
      "loss": 0.074,
      "step": 3320
    },
    {
      "epoch": 1.896355353075171,
      "grad_norm": 1.7326701879501343,
      "learning_rate": 1.8394077448747154e-05,
      "loss": 0.0773,
      "step": 3330
    },
    {
      "epoch": 1.9020501138952164,
      "grad_norm": 1.2691752910614014,
      "learning_rate": 1.829916476841306e-05,
      "loss": 0.1251,
      "step": 3340
    },
    {
      "epoch": 1.907744874715262,
      "grad_norm": 3.039222240447998,
      "learning_rate": 1.8204252088078968e-05,
      "loss": 0.0819,
      "step": 3350
    },
    {
      "epoch": 1.9134396355353074,
      "grad_norm": 3.1683478355407715,
      "learning_rate": 1.8109339407744876e-05,
      "loss": 0.0673,
      "step": 3360
    },
    {
      "epoch": 1.919134396355353,
      "grad_norm": 1.2576136589050293,
      "learning_rate": 1.8014426727410783e-05,
      "loss": 0.0793,
      "step": 3370
    },
    {
      "epoch": 1.9248291571753986,
      "grad_norm": 1.9472663402557373,
      "learning_rate": 1.791951404707669e-05,
      "loss": 0.0968,
      "step": 3380
    },
    {
      "epoch": 1.930523917995444,
      "grad_norm": 2.7382168769836426,
      "learning_rate": 1.7824601366742598e-05,
      "loss": 0.0716,
      "step": 3390
    },
    {
      "epoch": 1.9362186788154898,
      "grad_norm": 0.67622971534729,
      "learning_rate": 1.7729688686408505e-05,
      "loss": 0.0688,
      "step": 3400
    },
    {
      "epoch": 1.9419134396355353,
      "grad_norm": 0.3040902614593506,
      "learning_rate": 1.7634776006074412e-05,
      "loss": 0.0733,
      "step": 3410
    },
    {
      "epoch": 1.9476082004555808,
      "grad_norm": 1.6363804340362549,
      "learning_rate": 1.753986332574032e-05,
      "loss": 0.1081,
      "step": 3420
    },
    {
      "epoch": 1.9533029612756265,
      "grad_norm": 0.7806687355041504,
      "learning_rate": 1.7444950645406227e-05,
      "loss": 0.0959,
      "step": 3430
    },
    {
      "epoch": 1.958997722095672,
      "grad_norm": 0.37240535020828247,
      "learning_rate": 1.7350037965072134e-05,
      "loss": 0.0776,
      "step": 3440
    },
    {
      "epoch": 1.9646924829157175,
      "grad_norm": 0.8642539978027344,
      "learning_rate": 1.7255125284738042e-05,
      "loss": 0.0689,
      "step": 3450
    },
    {
      "epoch": 1.9703872437357632,
      "grad_norm": 0.2137574553489685,
      "learning_rate": 1.716021260440395e-05,
      "loss": 0.0428,
      "step": 3460
    },
    {
      "epoch": 1.9760820045558085,
      "grad_norm": 1.5309349298477173,
      "learning_rate": 1.7065299924069857e-05,
      "loss": 0.1032,
      "step": 3470
    },
    {
      "epoch": 1.9817767653758542,
      "grad_norm": 0.8801184296607971,
      "learning_rate": 1.6970387243735764e-05,
      "loss": 0.0796,
      "step": 3480
    },
    {
      "epoch": 1.9874715261958997,
      "grad_norm": 0.5306097865104675,
      "learning_rate": 1.687547456340167e-05,
      "loss": 0.0421,
      "step": 3490
    },
    {
      "epoch": 1.9931662870159452,
      "grad_norm": 1.5307530164718628,
      "learning_rate": 1.678056188306758e-05,
      "loss": 0.0844,
      "step": 3500
    },
    {
      "epoch": 1.998861047835991,
      "grad_norm": 0.7120900750160217,
      "learning_rate": 1.6685649202733486e-05,
      "loss": 0.0752,
      "step": 3510
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9732167205578713,
      "eval_f1": 0.8090976270629773,
      "eval_loss": 0.10903619229793549,
      "eval_precision": 0.7895833333333333,
      "eval_recall": 0.8296009429196834,
      "eval_runtime": 25.6763,
      "eval_samples_per_second": 126.576,
      "eval_steps_per_second": 15.851,
      "step": 3512
    },
    {
      "epoch": 2.0045558086560367,
      "grad_norm": 1.5597575902938843,
      "learning_rate": 1.6590736522399393e-05,
      "loss": 0.0975,
      "step": 3520
    },
    {
      "epoch": 2.010250569476082,
      "grad_norm": 1.3607456684112549,
      "learning_rate": 1.64958238420653e-05,
      "loss": 0.0771,
      "step": 3530
    },
    {
      "epoch": 2.0159453302961277,
      "grad_norm": 1.3556320667266846,
      "learning_rate": 1.6400911161731208e-05,
      "loss": 0.0853,
      "step": 3540
    },
    {
      "epoch": 2.021640091116173,
      "grad_norm": 1.3193871974945068,
      "learning_rate": 1.6305998481397115e-05,
      "loss": 0.0966,
      "step": 3550
    },
    {
      "epoch": 2.0273348519362187,
      "grad_norm": 4.825347423553467,
      "learning_rate": 1.6211085801063023e-05,
      "loss": 0.0916,
      "step": 3560
    },
    {
      "epoch": 2.0330296127562644,
      "grad_norm": 0.9464587569236755,
      "learning_rate": 1.611617312072893e-05,
      "loss": 0.0686,
      "step": 3570
    },
    {
      "epoch": 2.0387243735763096,
      "grad_norm": 1.426204800605774,
      "learning_rate": 1.6021260440394838e-05,
      "loss": 0.0741,
      "step": 3580
    },
    {
      "epoch": 2.0444191343963554,
      "grad_norm": 0.7591248154640198,
      "learning_rate": 1.5926347760060745e-05,
      "loss": 0.0636,
      "step": 3590
    },
    {
      "epoch": 2.050113895216401,
      "grad_norm": 0.7133740186691284,
      "learning_rate": 1.5831435079726652e-05,
      "loss": 0.0371,
      "step": 3600
    },
    {
      "epoch": 2.0558086560364464,
      "grad_norm": 1.1163349151611328,
      "learning_rate": 1.573652239939256e-05,
      "loss": 0.0799,
      "step": 3610
    },
    {
      "epoch": 2.061503416856492,
      "grad_norm": 2.288480281829834,
      "learning_rate": 1.5641609719058467e-05,
      "loss": 0.0739,
      "step": 3620
    },
    {
      "epoch": 2.067198177676538,
      "grad_norm": 0.6606568694114685,
      "learning_rate": 1.5546697038724374e-05,
      "loss": 0.0633,
      "step": 3630
    },
    {
      "epoch": 2.072892938496583,
      "grad_norm": 0.63154137134552,
      "learning_rate": 1.545178435839028e-05,
      "loss": 0.064,
      "step": 3640
    },
    {
      "epoch": 2.078587699316629,
      "grad_norm": 1.6103167533874512,
      "learning_rate": 1.535687167805619e-05,
      "loss": 0.0416,
      "step": 3650
    },
    {
      "epoch": 2.084282460136674,
      "grad_norm": 0.6107991337776184,
      "learning_rate": 1.5261958997722096e-05,
      "loss": 0.0405,
      "step": 3660
    },
    {
      "epoch": 2.08997722095672,
      "grad_norm": 0.8715682029724121,
      "learning_rate": 1.5167046317388004e-05,
      "loss": 0.0814,
      "step": 3670
    },
    {
      "epoch": 2.0956719817767655,
      "grad_norm": 0.7718030214309692,
      "learning_rate": 1.5072133637053911e-05,
      "loss": 0.061,
      "step": 3680
    },
    {
      "epoch": 2.1013667425968108,
      "grad_norm": 0.5676106214523315,
      "learning_rate": 1.4977220956719817e-05,
      "loss": 0.0462,
      "step": 3690
    },
    {
      "epoch": 2.1070615034168565,
      "grad_norm": 0.8370518088340759,
      "learning_rate": 1.4882308276385726e-05,
      "loss": 0.0686,
      "step": 3700
    },
    {
      "epoch": 2.112756264236902,
      "grad_norm": 1.26499605178833,
      "learning_rate": 1.4787395596051635e-05,
      "loss": 0.0604,
      "step": 3710
    },
    {
      "epoch": 2.1184510250569475,
      "grad_norm": 2.1694042682647705,
      "learning_rate": 1.469248291571754e-05,
      "loss": 0.0658,
      "step": 3720
    },
    {
      "epoch": 2.124145785876993,
      "grad_norm": 1.244402527809143,
      "learning_rate": 1.4597570235383448e-05,
      "loss": 0.0474,
      "step": 3730
    },
    {
      "epoch": 2.129840546697039,
      "grad_norm": 0.6539400219917297,
      "learning_rate": 1.4502657555049357e-05,
      "loss": 0.06,
      "step": 3740
    },
    {
      "epoch": 2.135535307517084,
      "grad_norm": 1.2413984537124634,
      "learning_rate": 1.4407744874715263e-05,
      "loss": 0.0497,
      "step": 3750
    },
    {
      "epoch": 2.14123006833713,
      "grad_norm": 0.9623197317123413,
      "learning_rate": 1.431283219438117e-05,
      "loss": 0.053,
      "step": 3760
    },
    {
      "epoch": 2.146924829157175,
      "grad_norm": 6.023757457733154,
      "learning_rate": 1.4217919514047076e-05,
      "loss": 0.0844,
      "step": 3770
    },
    {
      "epoch": 2.152619589977221,
      "grad_norm": 0.9312296509742737,
      "learning_rate": 1.4123006833712985e-05,
      "loss": 0.0863,
      "step": 3780
    },
    {
      "epoch": 2.1583143507972666,
      "grad_norm": 1.2199676036834717,
      "learning_rate": 1.4028094153378892e-05,
      "loss": 0.0692,
      "step": 3790
    },
    {
      "epoch": 2.164009111617312,
      "grad_norm": 1.3029935359954834,
      "learning_rate": 1.3933181473044798e-05,
      "loss": 0.078,
      "step": 3800
    },
    {
      "epoch": 2.1697038724373576,
      "grad_norm": 0.484778493642807,
      "learning_rate": 1.3838268792710707e-05,
      "loss": 0.0595,
      "step": 3810
    },
    {
      "epoch": 2.1753986332574033,
      "grad_norm": 0.8675511479377747,
      "learning_rate": 1.3743356112376616e-05,
      "loss": 0.0575,
      "step": 3820
    },
    {
      "epoch": 2.1810933940774486,
      "grad_norm": 1.9901808500289917,
      "learning_rate": 1.3648443432042522e-05,
      "loss": 0.0949,
      "step": 3830
    },
    {
      "epoch": 2.1867881548974943,
      "grad_norm": 2.4126904010772705,
      "learning_rate": 1.3553530751708429e-05,
      "loss": 0.0813,
      "step": 3840
    },
    {
      "epoch": 2.19248291571754,
      "grad_norm": 1.9077173471450806,
      "learning_rate": 1.3458618071374338e-05,
      "loss": 0.1219,
      "step": 3850
    },
    {
      "epoch": 2.1981776765375853,
      "grad_norm": 2.687405586242676,
      "learning_rate": 1.3363705391040244e-05,
      "loss": 0.0884,
      "step": 3860
    },
    {
      "epoch": 2.203872437357631,
      "grad_norm": 1.8448173999786377,
      "learning_rate": 1.3268792710706151e-05,
      "loss": 0.08,
      "step": 3870
    },
    {
      "epoch": 2.2095671981776768,
      "grad_norm": 0.3753576874732971,
      "learning_rate": 1.3173880030372057e-05,
      "loss": 0.0749,
      "step": 3880
    },
    {
      "epoch": 2.215261958997722,
      "grad_norm": 2.6273550987243652,
      "learning_rate": 1.3078967350037966e-05,
      "loss": 0.0774,
      "step": 3890
    },
    {
      "epoch": 2.2209567198177678,
      "grad_norm": 2.107893228530884,
      "learning_rate": 1.2984054669703875e-05,
      "loss": 0.0708,
      "step": 3900
    },
    {
      "epoch": 2.226651480637813,
      "grad_norm": 0.2233879417181015,
      "learning_rate": 1.2889141989369779e-05,
      "loss": 0.054,
      "step": 3910
    },
    {
      "epoch": 2.2323462414578588,
      "grad_norm": 1.054631233215332,
      "learning_rate": 1.2794229309035688e-05,
      "loss": 0.0654,
      "step": 3920
    },
    {
      "epoch": 2.2380410022779045,
      "grad_norm": 1.4175485372543335,
      "learning_rate": 1.2699316628701597e-05,
      "loss": 0.0553,
      "step": 3930
    },
    {
      "epoch": 2.2437357630979498,
      "grad_norm": 0.12185627222061157,
      "learning_rate": 1.2604403948367503e-05,
      "loss": 0.0938,
      "step": 3940
    },
    {
      "epoch": 2.2494305239179955,
      "grad_norm": 1.3942961692810059,
      "learning_rate": 1.250949126803341e-05,
      "loss": 0.0696,
      "step": 3950
    },
    {
      "epoch": 2.255125284738041,
      "grad_norm": 0.9566187262535095,
      "learning_rate": 1.2414578587699317e-05,
      "loss": 0.0599,
      "step": 3960
    },
    {
      "epoch": 2.2608200455580865,
      "grad_norm": 0.304811030626297,
      "learning_rate": 1.2319665907365225e-05,
      "loss": 0.0907,
      "step": 3970
    },
    {
      "epoch": 2.266514806378132,
      "grad_norm": 1.0834438800811768,
      "learning_rate": 1.2224753227031132e-05,
      "loss": 0.0573,
      "step": 3980
    },
    {
      "epoch": 2.2722095671981775,
      "grad_norm": 1.1209036111831665,
      "learning_rate": 1.212984054669704e-05,
      "loss": 0.0789,
      "step": 3990
    },
    {
      "epoch": 2.277904328018223,
      "grad_norm": 2.8387298583984375,
      "learning_rate": 1.2034927866362947e-05,
      "loss": 0.0982,
      "step": 4000
    },
    {
      "epoch": 2.283599088838269,
      "grad_norm": 1.0303692817687988,
      "learning_rate": 1.1940015186028854e-05,
      "loss": 0.0724,
      "step": 4010
    },
    {
      "epoch": 2.289293849658314,
      "grad_norm": 0.5699959397315979,
      "learning_rate": 1.1845102505694761e-05,
      "loss": 0.068,
      "step": 4020
    },
    {
      "epoch": 2.29498861047836,
      "grad_norm": 0.30835121870040894,
      "learning_rate": 1.1750189825360669e-05,
      "loss": 0.0452,
      "step": 4030
    },
    {
      "epoch": 2.3006833712984056,
      "grad_norm": 1.6180905103683472,
      "learning_rate": 1.1655277145026576e-05,
      "loss": 0.0887,
      "step": 4040
    },
    {
      "epoch": 2.306378132118451,
      "grad_norm": 0.6430849432945251,
      "learning_rate": 1.1560364464692483e-05,
      "loss": 0.0593,
      "step": 4050
    },
    {
      "epoch": 2.3120728929384966,
      "grad_norm": 1.6492300033569336,
      "learning_rate": 1.146545178435839e-05,
      "loss": 0.0537,
      "step": 4060
    },
    {
      "epoch": 2.3177676537585423,
      "grad_norm": 3.9676427841186523,
      "learning_rate": 1.1370539104024298e-05,
      "loss": 0.0745,
      "step": 4070
    },
    {
      "epoch": 2.3234624145785876,
      "grad_norm": 1.95708167552948,
      "learning_rate": 1.1275626423690206e-05,
      "loss": 0.0638,
      "step": 4080
    },
    {
      "epoch": 2.3291571753986333,
      "grad_norm": 1.8522908687591553,
      "learning_rate": 1.1180713743356113e-05,
      "loss": 0.0745,
      "step": 4090
    },
    {
      "epoch": 2.334851936218679,
      "grad_norm": 1.5721604824066162,
      "learning_rate": 1.108580106302202e-05,
      "loss": 0.0624,
      "step": 4100
    },
    {
      "epoch": 2.3405466970387243,
      "grad_norm": 0.46538046002388,
      "learning_rate": 1.0990888382687928e-05,
      "loss": 0.0669,
      "step": 4110
    },
    {
      "epoch": 2.34624145785877,
      "grad_norm": 1.6386337280273438,
      "learning_rate": 1.0895975702353835e-05,
      "loss": 0.0804,
      "step": 4120
    },
    {
      "epoch": 2.3519362186788153,
      "grad_norm": 0.9034251570701599,
      "learning_rate": 1.0801063022019742e-05,
      "loss": 0.0562,
      "step": 4130
    },
    {
      "epoch": 2.357630979498861,
      "grad_norm": 7.529191970825195,
      "learning_rate": 1.070615034168565e-05,
      "loss": 0.104,
      "step": 4140
    },
    {
      "epoch": 2.3633257403189067,
      "grad_norm": 1.01326584815979,
      "learning_rate": 1.0611237661351557e-05,
      "loss": 0.073,
      "step": 4150
    },
    {
      "epoch": 2.369020501138952,
      "grad_norm": 1.8131357431411743,
      "learning_rate": 1.0516324981017464e-05,
      "loss": 0.0612,
      "step": 4160
    },
    {
      "epoch": 2.3747152619589977,
      "grad_norm": 0.7966225147247314,
      "learning_rate": 1.0421412300683372e-05,
      "loss": 0.0672,
      "step": 4170
    },
    {
      "epoch": 2.3804100227790435,
      "grad_norm": 0.7099034190177917,
      "learning_rate": 1.032649962034928e-05,
      "loss": 0.053,
      "step": 4180
    },
    {
      "epoch": 2.3861047835990887,
      "grad_norm": 1.4396919012069702,
      "learning_rate": 1.0231586940015187e-05,
      "loss": 0.0477,
      "step": 4190
    },
    {
      "epoch": 2.3917995444191344,
      "grad_norm": 1.2298674583435059,
      "learning_rate": 1.0136674259681094e-05,
      "loss": 0.0548,
      "step": 4200
    },
    {
      "epoch": 2.3974943052391797,
      "grad_norm": 0.5935720801353455,
      "learning_rate": 1.0041761579347001e-05,
      "loss": 0.0592,
      "step": 4210
    },
    {
      "epoch": 2.4031890660592254,
      "grad_norm": 1.7547837495803833,
      "learning_rate": 9.946848899012909e-06,
      "loss": 0.0619,
      "step": 4220
    },
    {
      "epoch": 2.408883826879271,
      "grad_norm": 0.3924281597137451,
      "learning_rate": 9.851936218678816e-06,
      "loss": 0.0316,
      "step": 4230
    },
    {
      "epoch": 2.4145785876993164,
      "grad_norm": 1.2887072563171387,
      "learning_rate": 9.757023538344723e-06,
      "loss": 0.0616,
      "step": 4240
    },
    {
      "epoch": 2.420273348519362,
      "grad_norm": 0.9915182590484619,
      "learning_rate": 9.66211085801063e-06,
      "loss": 0.0673,
      "step": 4250
    },
    {
      "epoch": 2.425968109339408,
      "grad_norm": 4.4653167724609375,
      "learning_rate": 9.567198177676538e-06,
      "loss": 0.0862,
      "step": 4260
    },
    {
      "epoch": 2.431662870159453,
      "grad_norm": 1.6697877645492554,
      "learning_rate": 9.472285497342445e-06,
      "loss": 0.048,
      "step": 4270
    },
    {
      "epoch": 2.437357630979499,
      "grad_norm": 1.1500027179718018,
      "learning_rate": 9.377372817008353e-06,
      "loss": 0.0733,
      "step": 4280
    },
    {
      "epoch": 2.4430523917995446,
      "grad_norm": 3.383201837539673,
      "learning_rate": 9.28246013667426e-06,
      "loss": 0.0666,
      "step": 4290
    },
    {
      "epoch": 2.44874715261959,
      "grad_norm": 0.4077356159687042,
      "learning_rate": 9.187547456340167e-06,
      "loss": 0.0694,
      "step": 4300
    },
    {
      "epoch": 2.4544419134396356,
      "grad_norm": 0.7718432545661926,
      "learning_rate": 9.092634776006075e-06,
      "loss": 0.0515,
      "step": 4310
    },
    {
      "epoch": 2.4601366742596813,
      "grad_norm": 0.4951815605163574,
      "learning_rate": 8.997722095671982e-06,
      "loss": 0.0511,
      "step": 4320
    },
    {
      "epoch": 2.4658314350797266,
      "grad_norm": 1.9625097513198853,
      "learning_rate": 8.90280941533789e-06,
      "loss": 0.0576,
      "step": 4330
    },
    {
      "epoch": 2.4715261958997723,
      "grad_norm": 2.4783849716186523,
      "learning_rate": 8.807896735003797e-06,
      "loss": 0.0924,
      "step": 4340
    },
    {
      "epoch": 2.477220956719818,
      "grad_norm": 0.669106125831604,
      "learning_rate": 8.712984054669704e-06,
      "loss": 0.0698,
      "step": 4350
    },
    {
      "epoch": 2.4829157175398633,
      "grad_norm": 1.0594432353973389,
      "learning_rate": 8.618071374335612e-06,
      "loss": 0.0562,
      "step": 4360
    },
    {
      "epoch": 2.488610478359909,
      "grad_norm": 3.4004504680633545,
      "learning_rate": 8.523158694001519e-06,
      "loss": 0.102,
      "step": 4370
    },
    {
      "epoch": 2.4943052391799543,
      "grad_norm": 1.7753770351409912,
      "learning_rate": 8.428246013667426e-06,
      "loss": 0.087,
      "step": 4380
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.4696385860443115,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.0491,
      "step": 4390
    },
    {
      "epoch": 2.5056947608200457,
      "grad_norm": 0.3402756452560425,
      "learning_rate": 8.238420652999241e-06,
      "loss": 0.0633,
      "step": 4400
    },
    {
      "epoch": 2.511389521640091,
      "grad_norm": 1.155333399772644,
      "learning_rate": 8.143507972665148e-06,
      "loss": 0.0814,
      "step": 4410
    },
    {
      "epoch": 2.5170842824601367,
      "grad_norm": 1.0245141983032227,
      "learning_rate": 8.048595292331056e-06,
      "loss": 0.0796,
      "step": 4420
    },
    {
      "epoch": 2.522779043280182,
      "grad_norm": 1.1780219078063965,
      "learning_rate": 7.953682611996963e-06,
      "loss": 0.0565,
      "step": 4430
    },
    {
      "epoch": 2.5284738041002277,
      "grad_norm": 1.4757074117660522,
      "learning_rate": 7.85876993166287e-06,
      "loss": 0.0689,
      "step": 4440
    },
    {
      "epoch": 2.5341685649202734,
      "grad_norm": 0.504363477230072,
      "learning_rate": 7.763857251328778e-06,
      "loss": 0.0483,
      "step": 4450
    },
    {
      "epoch": 2.5398633257403187,
      "grad_norm": 1.0538045167922974,
      "learning_rate": 7.668944570994685e-06,
      "loss": 0.0443,
      "step": 4460
    },
    {
      "epoch": 2.5455580865603644,
      "grad_norm": 0.31798219680786133,
      "learning_rate": 7.574031890660592e-06,
      "loss": 0.0616,
      "step": 4470
    },
    {
      "epoch": 2.55125284738041,
      "grad_norm": 2.8450510501861572,
      "learning_rate": 7.4791192103265e-06,
      "loss": 0.0416,
      "step": 4480
    },
    {
      "epoch": 2.5569476082004554,
      "grad_norm": 1.6050668954849243,
      "learning_rate": 7.384206529992407e-06,
      "loss": 0.0623,
      "step": 4490
    },
    {
      "epoch": 2.562642369020501,
      "grad_norm": 1.804625391960144,
      "learning_rate": 7.289293849658315e-06,
      "loss": 0.0684,
      "step": 4500
    },
    {
      "epoch": 2.568337129840547,
      "grad_norm": 0.7985472083091736,
      "learning_rate": 7.194381169324223e-06,
      "loss": 0.0778,
      "step": 4510
    },
    {
      "epoch": 2.574031890660592,
      "grad_norm": 0.7117947340011597,
      "learning_rate": 7.099468488990129e-06,
      "loss": 0.0729,
      "step": 4520
    },
    {
      "epoch": 2.579726651480638,
      "grad_norm": 4.265881061553955,
      "learning_rate": 7.004555808656037e-06,
      "loss": 0.0676,
      "step": 4530
    },
    {
      "epoch": 2.5854214123006836,
      "grad_norm": 1.5959819555282593,
      "learning_rate": 6.909643128321943e-06,
      "loss": 0.0794,
      "step": 4540
    },
    {
      "epoch": 2.591116173120729,
      "grad_norm": 0.9075917601585388,
      "learning_rate": 6.814730447987852e-06,
      "loss": 0.089,
      "step": 4550
    },
    {
      "epoch": 2.5968109339407746,
      "grad_norm": 0.2236226499080658,
      "learning_rate": 6.719817767653759e-06,
      "loss": 0.0651,
      "step": 4560
    },
    {
      "epoch": 2.6025056947608203,
      "grad_norm": 0.5465313792228699,
      "learning_rate": 6.624905087319666e-06,
      "loss": 0.0507,
      "step": 4570
    },
    {
      "epoch": 2.6082004555808656,
      "grad_norm": 1.364637017250061,
      "learning_rate": 6.529992406985573e-06,
      "loss": 0.0732,
      "step": 4580
    },
    {
      "epoch": 2.6138952164009113,
      "grad_norm": 0.6693851947784424,
      "learning_rate": 6.435079726651482e-06,
      "loss": 0.0568,
      "step": 4590
    },
    {
      "epoch": 2.619589977220957,
      "grad_norm": 0.8072832822799683,
      "learning_rate": 6.340167046317388e-06,
      "loss": 0.049,
      "step": 4600
    },
    {
      "epoch": 2.6252847380410023,
      "grad_norm": 2.479691505432129,
      "learning_rate": 6.245254365983296e-06,
      "loss": 0.051,
      "step": 4610
    },
    {
      "epoch": 2.630979498861048,
      "grad_norm": 1.4793975353240967,
      "learning_rate": 6.150341685649203e-06,
      "loss": 0.0959,
      "step": 4620
    },
    {
      "epoch": 2.6366742596810933,
      "grad_norm": 0.482715368270874,
      "learning_rate": 6.05542900531511e-06,
      "loss": 0.0661,
      "step": 4630
    },
    {
      "epoch": 2.642369020501139,
      "grad_norm": 1.2736594676971436,
      "learning_rate": 5.960516324981018e-06,
      "loss": 0.0703,
      "step": 4640
    },
    {
      "epoch": 2.6480637813211843,
      "grad_norm": 2.8748881816864014,
      "learning_rate": 5.865603644646925e-06,
      "loss": 0.0706,
      "step": 4650
    },
    {
      "epoch": 2.65375854214123,
      "grad_norm": 1.8054091930389404,
      "learning_rate": 5.7706909643128325e-06,
      "loss": 0.0602,
      "step": 4660
    },
    {
      "epoch": 2.6594533029612757,
      "grad_norm": 1.6916309595108032,
      "learning_rate": 5.67577828397874e-06,
      "loss": 0.0747,
      "step": 4670
    },
    {
      "epoch": 2.665148063781321,
      "grad_norm": 0.40390637516975403,
      "learning_rate": 5.580865603644647e-06,
      "loss": 0.0501,
      "step": 4680
    },
    {
      "epoch": 2.6708428246013667,
      "grad_norm": 0.5195274949073792,
      "learning_rate": 5.4859529233105546e-06,
      "loss": 0.0453,
      "step": 4690
    },
    {
      "epoch": 2.6765375854214124,
      "grad_norm": 1.2867252826690674,
      "learning_rate": 5.391040242976462e-06,
      "loss": 0.0704,
      "step": 4700
    },
    {
      "epoch": 2.6822323462414577,
      "grad_norm": 0.6296011209487915,
      "learning_rate": 5.296127562642369e-06,
      "loss": 0.0673,
      "step": 4710
    },
    {
      "epoch": 2.6879271070615034,
      "grad_norm": 2.337752342224121,
      "learning_rate": 5.201214882308277e-06,
      "loss": 0.079,
      "step": 4720
    },
    {
      "epoch": 2.693621867881549,
      "grad_norm": 0.4544166624546051,
      "learning_rate": 5.106302201974184e-06,
      "loss": 0.0703,
      "step": 4730
    },
    {
      "epoch": 2.6993166287015944,
      "grad_norm": 0.8933178782463074,
      "learning_rate": 5.011389521640091e-06,
      "loss": 0.0808,
      "step": 4740
    },
    {
      "epoch": 2.70501138952164,
      "grad_norm": 0.5265564322471619,
      "learning_rate": 4.916476841305999e-06,
      "loss": 0.0755,
      "step": 4750
    },
    {
      "epoch": 2.710706150341686,
      "grad_norm": 1.0185718536376953,
      "learning_rate": 4.821564160971906e-06,
      "loss": 0.0748,
      "step": 4760
    },
    {
      "epoch": 2.716400911161731,
      "grad_norm": 0.5462549924850464,
      "learning_rate": 4.7266514806378134e-06,
      "loss": 0.0675,
      "step": 4770
    },
    {
      "epoch": 2.722095671981777,
      "grad_norm": 0.5356141328811646,
      "learning_rate": 4.631738800303721e-06,
      "loss": 0.0601,
      "step": 4780
    },
    {
      "epoch": 2.7277904328018225,
      "grad_norm": 0.587726891040802,
      "learning_rate": 4.536826119969628e-06,
      "loss": 0.0542,
      "step": 4790
    },
    {
      "epoch": 2.733485193621868,
      "grad_norm": 0.5419397950172424,
      "learning_rate": 4.4419134396355355e-06,
      "loss": 0.0814,
      "step": 4800
    },
    {
      "epoch": 2.7391799544419135,
      "grad_norm": 0.7181763052940369,
      "learning_rate": 4.347000759301443e-06,
      "loss": 0.0674,
      "step": 4810
    },
    {
      "epoch": 2.7448747152619593,
      "grad_norm": 2.263903856277466,
      "learning_rate": 4.25208807896735e-06,
      "loss": 0.0572,
      "step": 4820
    },
    {
      "epoch": 2.7505694760820045,
      "grad_norm": 1.2922956943511963,
      "learning_rate": 4.157175398633258e-06,
      "loss": 0.0745,
      "step": 4830
    },
    {
      "epoch": 2.7562642369020502,
      "grad_norm": 0.6281763315200806,
      "learning_rate": 4.062262718299165e-06,
      "loss": 0.0686,
      "step": 4840
    },
    {
      "epoch": 2.7619589977220955,
      "grad_norm": 0.612315833568573,
      "learning_rate": 3.967350037965072e-06,
      "loss": 0.0805,
      "step": 4850
    },
    {
      "epoch": 2.7676537585421412,
      "grad_norm": 0.9715772271156311,
      "learning_rate": 3.87243735763098e-06,
      "loss": 0.0503,
      "step": 4860
    },
    {
      "epoch": 2.7733485193621865,
      "grad_norm": 2.821282148361206,
      "learning_rate": 3.777524677296887e-06,
      "loss": 0.0831,
      "step": 4870
    },
    {
      "epoch": 2.7790432801822322,
      "grad_norm": 1.8253675699234009,
      "learning_rate": 3.682611996962795e-06,
      "loss": 0.054,
      "step": 4880
    },
    {
      "epoch": 2.784738041002278,
      "grad_norm": 1.2902501821517944,
      "learning_rate": 3.5876993166287018e-06,
      "loss": 0.0664,
      "step": 4890
    },
    {
      "epoch": 2.7904328018223232,
      "grad_norm": 1.3351030349731445,
      "learning_rate": 3.492786636294609e-06,
      "loss": 0.0365,
      "step": 4900
    },
    {
      "epoch": 2.796127562642369,
      "grad_norm": 1.3298659324645996,
      "learning_rate": 3.3978739559605165e-06,
      "loss": 0.1199,
      "step": 4910
    },
    {
      "epoch": 2.8018223234624147,
      "grad_norm": 1.367462158203125,
      "learning_rate": 3.302961275626424e-06,
      "loss": 0.0567,
      "step": 4920
    },
    {
      "epoch": 2.80751708428246,
      "grad_norm": 0.7492626309394836,
      "learning_rate": 3.208048595292331e-06,
      "loss": 0.07,
      "step": 4930
    },
    {
      "epoch": 2.8132118451025057,
      "grad_norm": 1.3312326669692993,
      "learning_rate": 3.1131359149582386e-06,
      "loss": 0.0738,
      "step": 4940
    },
    {
      "epoch": 2.8189066059225514,
      "grad_norm": 1.0007520914077759,
      "learning_rate": 3.018223234624146e-06,
      "loss": 0.0574,
      "step": 4950
    },
    {
      "epoch": 2.8246013667425967,
      "grad_norm": 0.8825734257698059,
      "learning_rate": 2.9233105542900533e-06,
      "loss": 0.0805,
      "step": 4960
    },
    {
      "epoch": 2.8302961275626424,
      "grad_norm": 2.035719633102417,
      "learning_rate": 2.8283978739559607e-06,
      "loss": 0.1008,
      "step": 4970
    },
    {
      "epoch": 2.835990888382688,
      "grad_norm": 1.2549408674240112,
      "learning_rate": 2.733485193621868e-06,
      "loss": 0.093,
      "step": 4980
    },
    {
      "epoch": 2.8416856492027334,
      "grad_norm": 0.5822533965110779,
      "learning_rate": 2.6385725132877754e-06,
      "loss": 0.0503,
      "step": 4990
    },
    {
      "epoch": 2.847380410022779,
      "grad_norm": 1.7420156002044678,
      "learning_rate": 2.5436598329536827e-06,
      "loss": 0.0792,
      "step": 5000
    },
    {
      "epoch": 2.853075170842825,
      "grad_norm": 0.5257258415222168,
      "learning_rate": 2.44874715261959e-06,
      "loss": 0.0494,
      "step": 5010
    },
    {
      "epoch": 2.85876993166287,
      "grad_norm": 1.4526097774505615,
      "learning_rate": 2.3538344722854975e-06,
      "loss": 0.0743,
      "step": 5020
    },
    {
      "epoch": 2.864464692482916,
      "grad_norm": 1.7381209135055542,
      "learning_rate": 2.258921791951405e-06,
      "loss": 0.0697,
      "step": 5030
    },
    {
      "epoch": 2.8701594533029615,
      "grad_norm": 0.9051744341850281,
      "learning_rate": 2.164009111617312e-06,
      "loss": 0.0439,
      "step": 5040
    },
    {
      "epoch": 2.875854214123007,
      "grad_norm": 1.1579861640930176,
      "learning_rate": 2.0690964312832195e-06,
      "loss": 0.0514,
      "step": 5050
    },
    {
      "epoch": 2.8815489749430525,
      "grad_norm": 0.17134447395801544,
      "learning_rate": 1.974183750949127e-06,
      "loss": 0.0587,
      "step": 5060
    },
    {
      "epoch": 2.887243735763098,
      "grad_norm": 2.1867082118988037,
      "learning_rate": 1.879271070615034e-06,
      "loss": 0.0806,
      "step": 5070
    },
    {
      "epoch": 2.8929384965831435,
      "grad_norm": 1.2143350839614868,
      "learning_rate": 1.7843583902809414e-06,
      "loss": 0.0817,
      "step": 5080
    },
    {
      "epoch": 2.8986332574031892,
      "grad_norm": 1.7585662603378296,
      "learning_rate": 1.6894457099468492e-06,
      "loss": 0.0561,
      "step": 5090
    },
    {
      "epoch": 2.9043280182232345,
      "grad_norm": 1.874373435974121,
      "learning_rate": 1.5945330296127566e-06,
      "loss": 0.0553,
      "step": 5100
    },
    {
      "epoch": 2.91002277904328,
      "grad_norm": 5.039700031280518,
      "learning_rate": 1.4996203492786637e-06,
      "loss": 0.0845,
      "step": 5110
    },
    {
      "epoch": 2.9157175398633255,
      "grad_norm": 1.0349217653274536,
      "learning_rate": 1.404707668944571e-06,
      "loss": 0.0774,
      "step": 5120
    },
    {
      "epoch": 2.921412300683371,
      "grad_norm": 1.7359356880187988,
      "learning_rate": 1.3097949886104784e-06,
      "loss": 0.0483,
      "step": 5130
    },
    {
      "epoch": 2.927107061503417,
      "grad_norm": 3.1591551303863525,
      "learning_rate": 1.2148823082763858e-06,
      "loss": 0.0682,
      "step": 5140
    },
    {
      "epoch": 2.932801822323462,
      "grad_norm": 2.8375813961029053,
      "learning_rate": 1.1199696279422931e-06,
      "loss": 0.0608,
      "step": 5150
    },
    {
      "epoch": 2.938496583143508,
      "grad_norm": 0.6938211917877197,
      "learning_rate": 1.0250569476082005e-06,
      "loss": 0.0691,
      "step": 5160
    },
    {
      "epoch": 2.9441913439635536,
      "grad_norm": 0.46077239513397217,
      "learning_rate": 9.301442672741079e-07,
      "loss": 0.068,
      "step": 5170
    },
    {
      "epoch": 2.949886104783599,
      "grad_norm": 1.4797725677490234,
      "learning_rate": 8.352315869400152e-07,
      "loss": 0.0947,
      "step": 5180
    },
    {
      "epoch": 2.9555808656036446,
      "grad_norm": 1.3453457355499268,
      "learning_rate": 7.403189066059226e-07,
      "loss": 0.082,
      "step": 5190
    },
    {
      "epoch": 2.9612756264236904,
      "grad_norm": 0.7412590384483337,
      "learning_rate": 6.4540622627183e-07,
      "loss": 0.0533,
      "step": 5200
    },
    {
      "epoch": 2.9669703872437356,
      "grad_norm": 2.559391975402832,
      "learning_rate": 5.504935459377373e-07,
      "loss": 0.0693,
      "step": 5210
    },
    {
      "epoch": 2.9726651480637813,
      "grad_norm": 0.8116249442100525,
      "learning_rate": 4.5558086560364467e-07,
      "loss": 0.088,
      "step": 5220
    },
    {
      "epoch": 2.978359908883827,
      "grad_norm": 0.675849437713623,
      "learning_rate": 3.6066818526955203e-07,
      "loss": 0.072,
      "step": 5230
    },
    {
      "epoch": 2.9840546697038723,
      "grad_norm": 1.1731585264205933,
      "learning_rate": 2.657555049354594e-07,
      "loss": 0.0511,
      "step": 5240
    },
    {
      "epoch": 2.989749430523918,
      "grad_norm": 0.8083361983299255,
      "learning_rate": 1.7084282460136675e-07,
      "loss": 0.0528,
      "step": 5250
    },
    {
      "epoch": 2.995444191343964,
      "grad_norm": 0.12277799099683762,
      "learning_rate": 7.593014426727411e-08,
      "loss": 0.0482,
      "step": 5260
    }
  ],
  "logging_steps": 10,
  "max_steps": 5268,
  "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": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2751824963545344.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}