{
  "best_metric": 0.4036685526371002,
  "best_model_checkpoint": "comprehend-it-dnd/checkpoint-3341",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 3341,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0074827895839568994,
      "grad_norm": 6.5380635261535645,
      "learning_rate": 3.7313432835820893e-06,
      "loss": 0.7273,
      "step": 25
    },
    {
      "epoch": 0.014965579167913799,
      "grad_norm": 10.18101692199707,
      "learning_rate": 7.4626865671641785e-06,
      "loss": 0.6718,
      "step": 50
    },
    {
      "epoch": 0.022448368751870697,
      "grad_norm": 5.7114458084106445,
      "learning_rate": 1.119402985074627e-05,
      "loss": 0.6273,
      "step": 75
    },
    {
      "epoch": 0.029931158335827598,
      "grad_norm": 6.610714912414551,
      "learning_rate": 1.4925373134328357e-05,
      "loss": 0.6331,
      "step": 100
    },
    {
      "epoch": 0.037413947919784495,
      "grad_norm": 5.677740573883057,
      "learning_rate": 1.865671641791045e-05,
      "loss": 0.6242,
      "step": 125
    },
    {
      "epoch": 0.044896737503741395,
      "grad_norm": 7.09804630279541,
      "learning_rate": 2.238805970149254e-05,
      "loss": 0.5644,
      "step": 150
    },
    {
      "epoch": 0.052379527087698295,
      "grad_norm": 7.025632381439209,
      "learning_rate": 2.6119402985074626e-05,
      "loss": 0.5184,
      "step": 175
    },
    {
      "epoch": 0.059862316671655195,
      "grad_norm": 10.766270637512207,
      "learning_rate": 2.9850746268656714e-05,
      "loss": 0.5403,
      "step": 200
    },
    {
      "epoch": 0.06734510625561209,
      "grad_norm": 11.382627487182617,
      "learning_rate": 3.358208955223881e-05,
      "loss": 0.6532,
      "step": 225
    },
    {
      "epoch": 0.07482789583956899,
      "grad_norm": 9.035429000854492,
      "learning_rate": 3.73134328358209e-05,
      "loss": 0.5806,
      "step": 250
    },
    {
      "epoch": 0.08231068542352589,
      "grad_norm": 3.988738775253296,
      "learning_rate": 4.104477611940299e-05,
      "loss": 0.5588,
      "step": 275
    },
    {
      "epoch": 0.08979347500748279,
      "grad_norm": 10.277702331542969,
      "learning_rate": 4.477611940298508e-05,
      "loss": 0.5251,
      "step": 300
    },
    {
      "epoch": 0.09727626459143969,
      "grad_norm": 12.550909996032715,
      "learning_rate": 4.850746268656717e-05,
      "loss": 0.5079,
      "step": 325
    },
    {
      "epoch": 0.10475905417539659,
      "grad_norm": 14.78004264831543,
      "learning_rate": 4.975049900199601e-05,
      "loss": 0.4729,
      "step": 350
    },
    {
      "epoch": 0.11224184375935349,
      "grad_norm": 5.339699745178223,
      "learning_rate": 4.933466400532269e-05,
      "loss": 0.5178,
      "step": 375
    },
    {
      "epoch": 0.11972463334331039,
      "grad_norm": 7.226634502410889,
      "learning_rate": 4.891882900864937e-05,
      "loss": 0.5508,
      "step": 400
    },
    {
      "epoch": 0.12720742292726728,
      "grad_norm": 5.301985740661621,
      "learning_rate": 4.8502994011976046e-05,
      "loss": 0.4357,
      "step": 425
    },
    {
      "epoch": 0.13469021251122418,
      "grad_norm": 4.2193827629089355,
      "learning_rate": 4.808715901530273e-05,
      "loss": 0.5456,
      "step": 450
    },
    {
      "epoch": 0.14217300209518108,
      "grad_norm": 6.087996959686279,
      "learning_rate": 4.767132401862941e-05,
      "loss": 0.5997,
      "step": 475
    },
    {
      "epoch": 0.14965579167913798,
      "grad_norm": 6.838490962982178,
      "learning_rate": 4.725548902195609e-05,
      "loss": 0.4628,
      "step": 500
    },
    {
      "epoch": 0.15713858126309488,
      "grad_norm": 3.5523083209991455,
      "learning_rate": 4.683965402528277e-05,
      "loss": 0.5327,
      "step": 525
    },
    {
      "epoch": 0.16462137084705178,
      "grad_norm": 1.994860291481018,
      "learning_rate": 4.642381902860945e-05,
      "loss": 0.471,
      "step": 550
    },
    {
      "epoch": 0.17210416043100868,
      "grad_norm": 4.96771240234375,
      "learning_rate": 4.600798403193613e-05,
      "loss": 0.5739,
      "step": 575
    },
    {
      "epoch": 0.17958695001496558,
      "grad_norm": 3.6930949687957764,
      "learning_rate": 4.559214903526281e-05,
      "loss": 0.616,
      "step": 600
    },
    {
      "epoch": 0.18706973959892248,
      "grad_norm": 3.547224998474121,
      "learning_rate": 4.517631403858949e-05,
      "loss": 0.517,
      "step": 625
    },
    {
      "epoch": 0.19455252918287938,
      "grad_norm": 3.5869407653808594,
      "learning_rate": 4.476047904191617e-05,
      "loss": 0.6888,
      "step": 650
    },
    {
      "epoch": 0.20203531876683628,
      "grad_norm": 5.559492111206055,
      "learning_rate": 4.434464404524285e-05,
      "loss": 0.4852,
      "step": 675
    },
    {
      "epoch": 0.20951810835079318,
      "grad_norm": 11.106466293334961,
      "learning_rate": 4.392880904856953e-05,
      "loss": 0.3709,
      "step": 700
    },
    {
      "epoch": 0.21700089793475008,
      "grad_norm": 5.652654647827148,
      "learning_rate": 4.351297405189621e-05,
      "loss": 0.5913,
      "step": 725
    },
    {
      "epoch": 0.22448368751870698,
      "grad_norm": 5.713591575622559,
      "learning_rate": 4.3097139055222893e-05,
      "loss": 0.4729,
      "step": 750
    },
    {
      "epoch": 0.23196647710266388,
      "grad_norm": 8.043790817260742,
      "learning_rate": 4.268130405854957e-05,
      "loss": 0.4218,
      "step": 775
    },
    {
      "epoch": 0.23944926668662078,
      "grad_norm": 4.482822418212891,
      "learning_rate": 4.226546906187625e-05,
      "loss": 0.4395,
      "step": 800
    },
    {
      "epoch": 0.24693205627057768,
      "grad_norm": 4.601754188537598,
      "learning_rate": 4.184963406520293e-05,
      "loss": 0.3905,
      "step": 825
    },
    {
      "epoch": 0.25441484585453455,
      "grad_norm": 4.083902835845947,
      "learning_rate": 4.143379906852961e-05,
      "loss": 0.5315,
      "step": 850
    },
    {
      "epoch": 0.2618976354384915,
      "grad_norm": 2.112675666809082,
      "learning_rate": 4.101796407185629e-05,
      "loss": 0.4555,
      "step": 875
    },
    {
      "epoch": 0.26938042502244836,
      "grad_norm": 3.542994499206543,
      "learning_rate": 4.060212907518297e-05,
      "loss": 0.534,
      "step": 900
    },
    {
      "epoch": 0.2768632146064053,
      "grad_norm": 2.592478036880493,
      "learning_rate": 4.018629407850965e-05,
      "loss": 0.3814,
      "step": 925
    },
    {
      "epoch": 0.28434600419036216,
      "grad_norm": 4.014191627502441,
      "learning_rate": 3.977045908183633e-05,
      "loss": 0.4372,
      "step": 950
    },
    {
      "epoch": 0.2918287937743191,
      "grad_norm": 1.8578141927719116,
      "learning_rate": 3.935462408516301e-05,
      "loss": 0.4915,
      "step": 975
    },
    {
      "epoch": 0.29931158335827596,
      "grad_norm": 2.7152152061462402,
      "learning_rate": 3.893878908848969e-05,
      "loss": 0.4555,
      "step": 1000
    },
    {
      "epoch": 0.3067943729422329,
      "grad_norm": 6.794373035430908,
      "learning_rate": 3.852295409181637e-05,
      "loss": 0.476,
      "step": 1025
    },
    {
      "epoch": 0.31427716252618976,
      "grad_norm": 6.25506591796875,
      "learning_rate": 3.810711909514305e-05,
      "loss": 0.4724,
      "step": 1050
    },
    {
      "epoch": 0.3217599521101467,
      "grad_norm": 6.050549030303955,
      "learning_rate": 3.7691284098469734e-05,
      "loss": 0.5136,
      "step": 1075
    },
    {
      "epoch": 0.32924274169410356,
      "grad_norm": 3.307494878768921,
      "learning_rate": 3.727544910179641e-05,
      "loss": 0.458,
      "step": 1100
    },
    {
      "epoch": 0.3367255312780605,
      "grad_norm": 3.186598777770996,
      "learning_rate": 3.685961410512309e-05,
      "loss": 0.4234,
      "step": 1125
    },
    {
      "epoch": 0.34420832086201736,
      "grad_norm": 3.677882432937622,
      "learning_rate": 3.644377910844977e-05,
      "loss": 0.4784,
      "step": 1150
    },
    {
      "epoch": 0.3516911104459743,
      "grad_norm": 4.9095330238342285,
      "learning_rate": 3.6027944111776455e-05,
      "loss": 0.4095,
      "step": 1175
    },
    {
      "epoch": 0.35917390002993116,
      "grad_norm": 1.6195601224899292,
      "learning_rate": 3.5612109115103126e-05,
      "loss": 0.4862,
      "step": 1200
    },
    {
      "epoch": 0.36665668961388803,
      "grad_norm": 6.7247633934021,
      "learning_rate": 3.5196274118429805e-05,
      "loss": 0.3781,
      "step": 1225
    },
    {
      "epoch": 0.37413947919784496,
      "grad_norm": 14.867138862609863,
      "learning_rate": 3.478043912175648e-05,
      "loss": 0.5926,
      "step": 1250
    },
    {
      "epoch": 0.38162226878180183,
      "grad_norm": 13.891983985900879,
      "learning_rate": 3.436460412508317e-05,
      "loss": 0.4065,
      "step": 1275
    },
    {
      "epoch": 0.38910505836575876,
      "grad_norm": 2.9691779613494873,
      "learning_rate": 3.394876912840985e-05,
      "loss": 0.5314,
      "step": 1300
    },
    {
      "epoch": 0.39658784794971563,
      "grad_norm": 8.80683708190918,
      "learning_rate": 3.3532934131736525e-05,
      "loss": 0.4489,
      "step": 1325
    },
    {
      "epoch": 0.40407063753367256,
      "grad_norm": 3.2144343852996826,
      "learning_rate": 3.3117099135063204e-05,
      "loss": 0.4072,
      "step": 1350
    },
    {
      "epoch": 0.41155342711762943,
      "grad_norm": 1.2881354093551636,
      "learning_rate": 3.270126413838989e-05,
      "loss": 0.5381,
      "step": 1375
    },
    {
      "epoch": 0.41903621670158636,
      "grad_norm": 3.566432237625122,
      "learning_rate": 3.228542914171657e-05,
      "loss": 0.3924,
      "step": 1400
    },
    {
      "epoch": 0.42651900628554323,
      "grad_norm": 13.332468032836914,
      "learning_rate": 3.1869594145043246e-05,
      "loss": 0.4926,
      "step": 1425
    },
    {
      "epoch": 0.43400179586950016,
      "grad_norm": 2.200331687927246,
      "learning_rate": 3.1453759148369925e-05,
      "loss": 0.4952,
      "step": 1450
    },
    {
      "epoch": 0.44148458545345703,
      "grad_norm": 4.319591045379639,
      "learning_rate": 3.103792415169661e-05,
      "loss": 0.4534,
      "step": 1475
    },
    {
      "epoch": 0.44896737503741396,
      "grad_norm": 2.805899143218994,
      "learning_rate": 3.062208915502329e-05,
      "loss": 0.4999,
      "step": 1500
    },
    {
      "epoch": 0.45645016462137084,
      "grad_norm": 5.09625768661499,
      "learning_rate": 3.0206254158349967e-05,
      "loss": 0.4208,
      "step": 1525
    },
    {
      "epoch": 0.46393295420532776,
      "grad_norm": 2.7175803184509277,
      "learning_rate": 2.979041916167665e-05,
      "loss": 0.4694,
      "step": 1550
    },
    {
      "epoch": 0.47141574378928464,
      "grad_norm": 2.1803653240203857,
      "learning_rate": 2.9374584165003327e-05,
      "loss": 0.3932,
      "step": 1575
    },
    {
      "epoch": 0.47889853337324156,
      "grad_norm": 7.961592197418213,
      "learning_rate": 2.8958749168330006e-05,
      "loss": 0.4643,
      "step": 1600
    },
    {
      "epoch": 0.48638132295719844,
      "grad_norm": 3.3433873653411865,
      "learning_rate": 2.8542914171656687e-05,
      "loss": 0.3753,
      "step": 1625
    },
    {
      "epoch": 0.49386411254115536,
      "grad_norm": 3.7999348640441895,
      "learning_rate": 2.8127079174983366e-05,
      "loss": 0.3702,
      "step": 1650
    },
    {
      "epoch": 0.5013469021251122,
      "grad_norm": 5.084557056427002,
      "learning_rate": 2.7711244178310048e-05,
      "loss": 0.4402,
      "step": 1675
    },
    {
      "epoch": 0.5088296917090691,
      "grad_norm": 4.610659599304199,
      "learning_rate": 2.7295409181636726e-05,
      "loss": 0.4458,
      "step": 1700
    },
    {
      "epoch": 0.5163124812930261,
      "grad_norm": 15.546037673950195,
      "learning_rate": 2.6879574184963408e-05,
      "loss": 0.4366,
      "step": 1725
    },
    {
      "epoch": 0.523795270876983,
      "grad_norm": 3.105518341064453,
      "learning_rate": 2.6463739188290087e-05,
      "loss": 0.3496,
      "step": 1750
    },
    {
      "epoch": 0.5312780604609398,
      "grad_norm": 3.7555079460144043,
      "learning_rate": 2.604790419161677e-05,
      "loss": 0.5785,
      "step": 1775
    },
    {
      "epoch": 0.5387608500448967,
      "grad_norm": 2.50890851020813,
      "learning_rate": 2.5632069194943447e-05,
      "loss": 0.4319,
      "step": 1800
    },
    {
      "epoch": 0.5462436396288536,
      "grad_norm": 10.880901336669922,
      "learning_rate": 2.521623419827013e-05,
      "loss": 0.4058,
      "step": 1825
    },
    {
      "epoch": 0.5537264292128106,
      "grad_norm": 3.4000051021575928,
      "learning_rate": 2.4800399201596807e-05,
      "loss": 0.3231,
      "step": 1850
    },
    {
      "epoch": 0.5612092187967674,
      "grad_norm": 5.48628044128418,
      "learning_rate": 2.438456420492349e-05,
      "loss": 0.4956,
      "step": 1875
    },
    {
      "epoch": 0.5686920083807243,
      "grad_norm": 2.575294017791748,
      "learning_rate": 2.3968729208250168e-05,
      "loss": 0.4567,
      "step": 1900
    },
    {
      "epoch": 0.5761747979646812,
      "grad_norm": 2.3242228031158447,
      "learning_rate": 2.355289421157685e-05,
      "loss": 0.5118,
      "step": 1925
    },
    {
      "epoch": 0.5836575875486382,
      "grad_norm": 2.4815163612365723,
      "learning_rate": 2.3137059214903528e-05,
      "loss": 0.414,
      "step": 1950
    },
    {
      "epoch": 0.591140377132595,
      "grad_norm": 4.164279937744141,
      "learning_rate": 2.272122421823021e-05,
      "loss": 0.4025,
      "step": 1975
    },
    {
      "epoch": 0.5986231667165519,
      "grad_norm": 3.876215696334839,
      "learning_rate": 2.2305389221556888e-05,
      "loss": 0.4084,
      "step": 2000
    },
    {
      "epoch": 0.6061059563005088,
      "grad_norm": 5.335385322570801,
      "learning_rate": 2.1889554224883567e-05,
      "loss": 0.5159,
      "step": 2025
    },
    {
      "epoch": 0.6135887458844658,
      "grad_norm": 1.9354826211929321,
      "learning_rate": 2.147371922821025e-05,
      "loss": 0.4208,
      "step": 2050
    },
    {
      "epoch": 0.6210715354684226,
      "grad_norm": 2.973052978515625,
      "learning_rate": 2.1057884231536927e-05,
      "loss": 0.3974,
      "step": 2075
    },
    {
      "epoch": 0.6285543250523795,
      "grad_norm": 5.183703422546387,
      "learning_rate": 2.0642049234863606e-05,
      "loss": 0.4069,
      "step": 2100
    },
    {
      "epoch": 0.6360371146363364,
      "grad_norm": 5.665794372558594,
      "learning_rate": 2.0226214238190287e-05,
      "loss": 0.5142,
      "step": 2125
    },
    {
      "epoch": 0.6435199042202934,
      "grad_norm": 1.856615662574768,
      "learning_rate": 1.9810379241516966e-05,
      "loss": 0.4481,
      "step": 2150
    },
    {
      "epoch": 0.6510026938042502,
      "grad_norm": 4.736162185668945,
      "learning_rate": 1.9394544244843644e-05,
      "loss": 0.4197,
      "step": 2175
    },
    {
      "epoch": 0.6584854833882071,
      "grad_norm": 2.6821250915527344,
      "learning_rate": 1.8978709248170326e-05,
      "loss": 0.3709,
      "step": 2200
    },
    {
      "epoch": 0.665968272972164,
      "grad_norm": 4.792450904846191,
      "learning_rate": 1.8562874251497005e-05,
      "loss": 0.3753,
      "step": 2225
    },
    {
      "epoch": 0.673451062556121,
      "grad_norm": 3.6531944274902344,
      "learning_rate": 1.8147039254823687e-05,
      "loss": 0.4898,
      "step": 2250
    },
    {
      "epoch": 0.6809338521400778,
      "grad_norm": 3.014845132827759,
      "learning_rate": 1.7731204258150365e-05,
      "loss": 0.4225,
      "step": 2275
    },
    {
      "epoch": 0.6884166417240347,
      "grad_norm": 13.898189544677734,
      "learning_rate": 1.7315369261477047e-05,
      "loss": 0.4691,
      "step": 2300
    },
    {
      "epoch": 0.6958994313079916,
      "grad_norm": 2.389976978302002,
      "learning_rate": 1.6899534264803725e-05,
      "loss": 0.4701,
      "step": 2325
    },
    {
      "epoch": 0.7033822208919486,
      "grad_norm": 3.1257760524749756,
      "learning_rate": 1.6483699268130407e-05,
      "loss": 0.4253,
      "step": 2350
    },
    {
      "epoch": 0.7108650104759054,
      "grad_norm": 19.56914520263672,
      "learning_rate": 1.6067864271457086e-05,
      "loss": 0.358,
      "step": 2375
    },
    {
      "epoch": 0.7183478000598623,
      "grad_norm": 2.580307722091675,
      "learning_rate": 1.5652029274783768e-05,
      "loss": 0.4389,
      "step": 2400
    },
    {
      "epoch": 0.7258305896438192,
      "grad_norm": 5.392817497253418,
      "learning_rate": 1.5236194278110446e-05,
      "loss": 0.4549,
      "step": 2425
    },
    {
      "epoch": 0.7333133792277761,
      "grad_norm": 4.282964706420898,
      "learning_rate": 1.4820359281437126e-05,
      "loss": 0.4553,
      "step": 2450
    },
    {
      "epoch": 0.740796168811733,
      "grad_norm": 2.4465701580047607,
      "learning_rate": 1.4404524284763806e-05,
      "loss": 0.3517,
      "step": 2475
    },
    {
      "epoch": 0.7482789583956899,
      "grad_norm": 4.104219436645508,
      "learning_rate": 1.3988689288090487e-05,
      "loss": 0.4565,
      "step": 2500
    },
    {
      "epoch": 0.7557617479796468,
      "grad_norm": 2.6114227771759033,
      "learning_rate": 1.3572854291417167e-05,
      "loss": 0.3864,
      "step": 2525
    },
    {
      "epoch": 0.7632445375636037,
      "grad_norm": 2.788442373275757,
      "learning_rate": 1.3157019294743847e-05,
      "loss": 0.5104,
      "step": 2550
    },
    {
      "epoch": 0.7707273271475606,
      "grad_norm": 1.952867865562439,
      "learning_rate": 1.2741184298070527e-05,
      "loss": 0.4367,
      "step": 2575
    },
    {
      "epoch": 0.7782101167315175,
      "grad_norm": 5.853802680969238,
      "learning_rate": 1.2325349301397205e-05,
      "loss": 0.337,
      "step": 2600
    },
    {
      "epoch": 0.7856929063154744,
      "grad_norm": 3.4243264198303223,
      "learning_rate": 1.1909514304723886e-05,
      "loss": 0.423,
      "step": 2625
    },
    {
      "epoch": 0.7931756958994313,
      "grad_norm": 3.185460090637207,
      "learning_rate": 1.1493679308050566e-05,
      "loss": 0.4628,
      "step": 2650
    },
    {
      "epoch": 0.8006584854833882,
      "grad_norm": 3.206066608428955,
      "learning_rate": 1.1077844311377246e-05,
      "loss": 0.3611,
      "step": 2675
    },
    {
      "epoch": 0.8081412750673451,
      "grad_norm": 2.893115520477295,
      "learning_rate": 1.0662009314703926e-05,
      "loss": 0.4015,
      "step": 2700
    },
    {
      "epoch": 0.815624064651302,
      "grad_norm": 2.7097058296203613,
      "learning_rate": 1.0246174318030606e-05,
      "loss": 0.3969,
      "step": 2725
    },
    {
      "epoch": 0.8231068542352589,
      "grad_norm": 3.2964673042297363,
      "learning_rate": 9.830339321357286e-06,
      "loss": 0.374,
      "step": 2750
    },
    {
      "epoch": 0.8305896438192159,
      "grad_norm": 1.6408296823501587,
      "learning_rate": 9.414504324683967e-06,
      "loss": 0.3544,
      "step": 2775
    },
    {
      "epoch": 0.8380724334031727,
      "grad_norm": 2.8518118858337402,
      "learning_rate": 8.998669328010647e-06,
      "loss": 0.3976,
      "step": 2800
    },
    {
      "epoch": 0.8455552229871296,
      "grad_norm": 2.862133502960205,
      "learning_rate": 8.582834331337327e-06,
      "loss": 0.4011,
      "step": 2825
    },
    {
      "epoch": 0.8530380125710865,
      "grad_norm": 3.1799404621124268,
      "learning_rate": 8.166999334664007e-06,
      "loss": 0.3611,
      "step": 2850
    },
    {
      "epoch": 0.8605208021550435,
      "grad_norm": 2.606300115585327,
      "learning_rate": 7.751164337990686e-06,
      "loss": 0.3529,
      "step": 2875
    },
    {
      "epoch": 0.8680035917390003,
      "grad_norm": 17.52878189086914,
      "learning_rate": 7.335329341317365e-06,
      "loss": 0.4559,
      "step": 2900
    },
    {
      "epoch": 0.8754863813229572,
      "grad_norm": 2.021336555480957,
      "learning_rate": 6.919494344644045e-06,
      "loss": 0.3203,
      "step": 2925
    },
    {
      "epoch": 0.8829691709069141,
      "grad_norm": 6.868563652038574,
      "learning_rate": 6.503659347970725e-06,
      "loss": 0.4525,
      "step": 2950
    },
    {
      "epoch": 0.8904519604908709,
      "grad_norm": 2.3738279342651367,
      "learning_rate": 6.0878243512974054e-06,
      "loss": 0.3157,
      "step": 2975
    },
    {
      "epoch": 0.8979347500748279,
      "grad_norm": 10.115850448608398,
      "learning_rate": 5.671989354624086e-06,
      "loss": 0.3339,
      "step": 3000
    },
    {
      "epoch": 0.9054175396587848,
      "grad_norm": 0.5855059027671814,
      "learning_rate": 5.256154357950766e-06,
      "loss": 0.4237,
      "step": 3025
    },
    {
      "epoch": 0.9129003292427417,
      "grad_norm": 4.681796550750732,
      "learning_rate": 4.840319361277446e-06,
      "loss": 0.3796,
      "step": 3050
    },
    {
      "epoch": 0.9203831188266985,
      "grad_norm": 1.9072390794754028,
      "learning_rate": 4.424484364604125e-06,
      "loss": 0.4479,
      "step": 3075
    },
    {
      "epoch": 0.9278659084106555,
      "grad_norm": 5.226965427398682,
      "learning_rate": 4.008649367930805e-06,
      "loss": 0.3302,
      "step": 3100
    },
    {
      "epoch": 0.9353486979946124,
      "grad_norm": 8.390512466430664,
      "learning_rate": 3.592814371257485e-06,
      "loss": 0.4117,
      "step": 3125
    },
    {
      "epoch": 0.9428314875785693,
      "grad_norm": 4.233555316925049,
      "learning_rate": 3.1769793745841653e-06,
      "loss": 0.353,
      "step": 3150
    },
    {
      "epoch": 0.9503142771625261,
      "grad_norm": 9.861014366149902,
      "learning_rate": 2.761144377910845e-06,
      "loss": 0.3437,
      "step": 3175
    },
    {
      "epoch": 0.9577970667464831,
      "grad_norm": 11.64777660369873,
      "learning_rate": 2.345309381237525e-06,
      "loss": 0.4715,
      "step": 3200
    },
    {
      "epoch": 0.96527985633044,
      "grad_norm": 9.067724227905273,
      "learning_rate": 1.929474384564205e-06,
      "loss": 0.331,
      "step": 3225
    },
    {
      "epoch": 0.9727626459143969,
      "grad_norm": 2.9110236167907715,
      "learning_rate": 1.513639387890885e-06,
      "loss": 0.3992,
      "step": 3250
    },
    {
      "epoch": 0.9802454354983537,
      "grad_norm": 4.988080978393555,
      "learning_rate": 1.097804391217565e-06,
      "loss": 0.3636,
      "step": 3275
    },
    {
      "epoch": 0.9877282250823107,
      "grad_norm": 4.391211032867432,
      "learning_rate": 6.819693945442449e-07,
      "loss": 0.4535,
      "step": 3300
    },
    {
      "epoch": 0.9952110146662676,
      "grad_norm": 9.230216979980469,
      "learning_rate": 2.661343978709248e-07,
      "loss": 0.3611,
      "step": 3325
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.824577159107918,
      "eval_auc": 0.9074155635351444,
      "eval_f1": 0.8555226824457594,
      "eval_loss": 0.4036685526371002,
      "eval_precision": 0.8582735592381895,
      "eval_recall": 0.8527893831408209,
      "eval_runtime": 1643.0247,
      "eval_samples_per_second": 4.066,
      "eval_steps_per_second": 0.254,
      "step": 3341
    }
  ],
  "logging_steps": 25,
  "max_steps": 3341,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.01
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1757876485773312.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}