{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.20287157670832226,
  "eval_steps": 500,
  "global_step": 6104,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001030311087476735,
      "grad_norm": 60.25547409057617,
      "learning_rate": 1.0157273918741808e-06,
      "loss": 8.8455,
      "step": 31
    },
    {
      "epoch": 0.00206062217495347,
      "grad_norm": 15.669363975524902,
      "learning_rate": 2.0314547837483616e-06,
      "loss": 7.1553,
      "step": 62
    },
    {
      "epoch": 0.003090933262430205,
      "grad_norm": 15.366345405578613,
      "learning_rate": 3.0471821756225426e-06,
      "loss": 5.8784,
      "step": 93
    },
    {
      "epoch": 0.00412124434990694,
      "grad_norm": 36.30561828613281,
      "learning_rate": 4.062909567496723e-06,
      "loss": 4.7708,
      "step": 124
    },
    {
      "epoch": 0.005151555437383675,
      "grad_norm": 27.202678680419922,
      "learning_rate": 5.078636959370905e-06,
      "loss": 4.1629,
      "step": 155
    },
    {
      "epoch": 0.00618186652486041,
      "grad_norm": 24.30484962463379,
      "learning_rate": 6.094364351245085e-06,
      "loss": 3.867,
      "step": 186
    },
    {
      "epoch": 0.007212177612337145,
      "grad_norm": 19.916366577148438,
      "learning_rate": 7.110091743119267e-06,
      "loss": 3.6131,
      "step": 217
    },
    {
      "epoch": 0.00824248869981388,
      "grad_norm": 17.577274322509766,
      "learning_rate": 8.125819134993446e-06,
      "loss": 3.4772,
      "step": 248
    },
    {
      "epoch": 0.009272799787290615,
      "grad_norm": 12.133190155029297,
      "learning_rate": 9.141546526867629e-06,
      "loss": 3.3218,
      "step": 279
    },
    {
      "epoch": 0.01030311087476735,
      "grad_norm": 19.79263687133789,
      "learning_rate": 1.015727391874181e-05,
      "loss": 3.2055,
      "step": 310
    },
    {
      "epoch": 0.011333421962244085,
      "grad_norm": 16.38133430480957,
      "learning_rate": 1.117300131061599e-05,
      "loss": 3.1062,
      "step": 341
    },
    {
      "epoch": 0.01236373304972082,
      "grad_norm": 12.638299942016602,
      "learning_rate": 1.218872870249017e-05,
      "loss": 3.0106,
      "step": 372
    },
    {
      "epoch": 0.013394044137197554,
      "grad_norm": 9.46596908569336,
      "learning_rate": 1.3204456094364351e-05,
      "loss": 2.924,
      "step": 403
    },
    {
      "epoch": 0.01442435522467429,
      "grad_norm": 10.945392608642578,
      "learning_rate": 1.4220183486238533e-05,
      "loss": 2.844,
      "step": 434
    },
    {
      "epoch": 0.015454666312151024,
      "grad_norm": 8.474015235900879,
      "learning_rate": 1.5235910878112714e-05,
      "loss": 2.7892,
      "step": 465
    },
    {
      "epoch": 0.01648497739962776,
      "grad_norm": 9.370804786682129,
      "learning_rate": 1.6251638269986893e-05,
      "loss": 2.7509,
      "step": 496
    },
    {
      "epoch": 0.017515288487104493,
      "grad_norm": 11.63398551940918,
      "learning_rate": 1.7267365661861077e-05,
      "loss": 2.6999,
      "step": 527
    },
    {
      "epoch": 0.01854559957458123,
      "grad_norm": 9.17713451385498,
      "learning_rate": 1.8283093053735257e-05,
      "loss": 2.6459,
      "step": 558
    },
    {
      "epoch": 0.019575910662057962,
      "grad_norm": 7.119054794311523,
      "learning_rate": 1.9298820445609438e-05,
      "loss": 2.603,
      "step": 589
    },
    {
      "epoch": 0.0206062217495347,
      "grad_norm": 6.653646945953369,
      "learning_rate": 2.031454783748362e-05,
      "loss": 2.5588,
      "step": 620
    },
    {
      "epoch": 0.021636532837011432,
      "grad_norm": 8.332653045654297,
      "learning_rate": 2.13302752293578e-05,
      "loss": 2.5357,
      "step": 651
    },
    {
      "epoch": 0.02266684392448817,
      "grad_norm": 6.4949116706848145,
      "learning_rate": 2.234600262123198e-05,
      "loss": 2.4967,
      "step": 682
    },
    {
      "epoch": 0.023697155011964902,
      "grad_norm": 9.41009521484375,
      "learning_rate": 2.336173001310616e-05,
      "loss": 2.4563,
      "step": 713
    },
    {
      "epoch": 0.02472746609944164,
      "grad_norm": 7.840345859527588,
      "learning_rate": 2.437745740498034e-05,
      "loss": 2.4383,
      "step": 744
    },
    {
      "epoch": 0.025757777186918372,
      "grad_norm": 6.116458415985107,
      "learning_rate": 2.5393184796854525e-05,
      "loss": 2.3817,
      "step": 775
    },
    {
      "epoch": 0.02678808827439511,
      "grad_norm": 5.938300609588623,
      "learning_rate": 2.6408912188728702e-05,
      "loss": 2.3508,
      "step": 806
    },
    {
      "epoch": 0.027818399361871842,
      "grad_norm": 5.4408345222473145,
      "learning_rate": 2.7424639580602886e-05,
      "loss": 2.3325,
      "step": 837
    },
    {
      "epoch": 0.02884871044934858,
      "grad_norm": 5.375136375427246,
      "learning_rate": 2.8440366972477066e-05,
      "loss": 2.3101,
      "step": 868
    },
    {
      "epoch": 0.029879021536825312,
      "grad_norm": 5.149726867675781,
      "learning_rate": 2.9456094364351244e-05,
      "loss": 2.282,
      "step": 899
    },
    {
      "epoch": 0.03090933262430205,
      "grad_norm": 4.591221332550049,
      "learning_rate": 3.0471821756225428e-05,
      "loss": 2.2427,
      "step": 930
    },
    {
      "epoch": 0.031939643711778785,
      "grad_norm": 4.977034091949463,
      "learning_rate": 3.148754914809961e-05,
      "loss": 2.2218,
      "step": 961
    },
    {
      "epoch": 0.03296995479925552,
      "grad_norm": 5.038781642913818,
      "learning_rate": 3.2503276539973785e-05,
      "loss": 2.2044,
      "step": 992
    },
    {
      "epoch": 0.03400026588673225,
      "grad_norm": 4.872281551361084,
      "learning_rate": 3.351900393184797e-05,
      "loss": 2.1657,
      "step": 1023
    },
    {
      "epoch": 0.035030576974208985,
      "grad_norm": 4.370841979980469,
      "learning_rate": 3.453473132372215e-05,
      "loss": 2.1365,
      "step": 1054
    },
    {
      "epoch": 0.036060888061685725,
      "grad_norm": 4.087072849273682,
      "learning_rate": 3.555045871559633e-05,
      "loss": 2.1253,
      "step": 1085
    },
    {
      "epoch": 0.03709119914916246,
      "grad_norm": 4.113957405090332,
      "learning_rate": 3.6566186107470514e-05,
      "loss": 2.0973,
      "step": 1116
    },
    {
      "epoch": 0.03812151023663919,
      "grad_norm": 4.0119733810424805,
      "learning_rate": 3.7581913499344695e-05,
      "loss": 2.1024,
      "step": 1147
    },
    {
      "epoch": 0.039151821324115925,
      "grad_norm": 4.247573375701904,
      "learning_rate": 3.8597640891218876e-05,
      "loss": 2.0722,
      "step": 1178
    },
    {
      "epoch": 0.04018213241159266,
      "grad_norm": 3.5575129985809326,
      "learning_rate": 3.9613368283093056e-05,
      "loss": 2.056,
      "step": 1209
    },
    {
      "epoch": 0.0412124434990694,
      "grad_norm": 3.8885862827301025,
      "learning_rate": 4.062909567496724e-05,
      "loss": 2.0389,
      "step": 1240
    },
    {
      "epoch": 0.04224275458654613,
      "grad_norm": 3.680628538131714,
      "learning_rate": 4.164482306684142e-05,
      "loss": 2.0385,
      "step": 1271
    },
    {
      "epoch": 0.043273065674022865,
      "grad_norm": 3.780876874923706,
      "learning_rate": 4.26605504587156e-05,
      "loss": 2.0097,
      "step": 1302
    },
    {
      "epoch": 0.0443033767614996,
      "grad_norm": 4.235328674316406,
      "learning_rate": 4.367627785058978e-05,
      "loss": 2.0024,
      "step": 1333
    },
    {
      "epoch": 0.04533368784897634,
      "grad_norm": 3.326941967010498,
      "learning_rate": 4.469200524246396e-05,
      "loss": 1.9953,
      "step": 1364
    },
    {
      "epoch": 0.04636399893645307,
      "grad_norm": 3.28456449508667,
      "learning_rate": 4.570773263433814e-05,
      "loss": 1.9579,
      "step": 1395
    },
    {
      "epoch": 0.047394310023929805,
      "grad_norm": 16.107433319091797,
      "learning_rate": 4.672346002621232e-05,
      "loss": 1.9701,
      "step": 1426
    },
    {
      "epoch": 0.04842462111140654,
      "grad_norm": 3.5708224773406982,
      "learning_rate": 4.77391874180865e-05,
      "loss": 1.9621,
      "step": 1457
    },
    {
      "epoch": 0.04945493219888328,
      "grad_norm": 2.9053499698638916,
      "learning_rate": 4.875491480996068e-05,
      "loss": 1.9458,
      "step": 1488
    },
    {
      "epoch": 0.05048524328636001,
      "grad_norm": 3.0863258838653564,
      "learning_rate": 4.977064220183487e-05,
      "loss": 1.9483,
      "step": 1519
    },
    {
      "epoch": 0.051515554373836744,
      "grad_norm": 2.9012269973754883,
      "learning_rate": 4.9999915451558777e-05,
      "loss": 1.928,
      "step": 1550
    },
    {
      "epoch": 0.05254586546131348,
      "grad_norm": 3.0949041843414307,
      "learning_rate": 4.999955597496219e-05,
      "loss": 1.9229,
      "step": 1581
    },
    {
      "epoch": 0.05357617654879022,
      "grad_norm": 2.8687901496887207,
      "learning_rate": 4.9998914381774255e-05,
      "loss": 1.915,
      "step": 1612
    },
    {
      "epoch": 0.05460648763626695,
      "grad_norm": 3.2136878967285156,
      "learning_rate": 4.999799067923527e-05,
      "loss": 1.9197,
      "step": 1643
    },
    {
      "epoch": 0.055636798723743684,
      "grad_norm": 2.590843677520752,
      "learning_rate": 4.999678487776908e-05,
      "loss": 1.8756,
      "step": 1674
    },
    {
      "epoch": 0.05666710981122042,
      "grad_norm": 2.64634108543396,
      "learning_rate": 4.9995296990983006e-05,
      "loss": 1.9033,
      "step": 1705
    },
    {
      "epoch": 0.05769742089869716,
      "grad_norm": 3.0151331424713135,
      "learning_rate": 4.999352703566763e-05,
      "loss": 1.8883,
      "step": 1736
    },
    {
      "epoch": 0.05872773198617389,
      "grad_norm": 2.526806354522705,
      "learning_rate": 4.999147503179668e-05,
      "loss": 1.8666,
      "step": 1767
    },
    {
      "epoch": 0.059758043073650624,
      "grad_norm": 2.510300397872925,
      "learning_rate": 4.998914100252672e-05,
      "loss": 1.854,
      "step": 1798
    },
    {
      "epoch": 0.06078835416112736,
      "grad_norm": 2.4867682456970215,
      "learning_rate": 4.998652497419696e-05,
      "loss": 1.8548,
      "step": 1829
    },
    {
      "epoch": 0.0618186652486041,
      "grad_norm": 2.3920586109161377,
      "learning_rate": 4.9983626976328927e-05,
      "loss": 1.8495,
      "step": 1860
    },
    {
      "epoch": 0.06284897633608083,
      "grad_norm": 2.714177370071411,
      "learning_rate": 4.998044704162613e-05,
      "loss": 1.8433,
      "step": 1891
    },
    {
      "epoch": 0.06387928742355757,
      "grad_norm": 2.3094465732574463,
      "learning_rate": 4.9976985205973705e-05,
      "loss": 1.8382,
      "step": 1922
    },
    {
      "epoch": 0.0649095985110343,
      "grad_norm": 2.47184419631958,
      "learning_rate": 4.997324150843799e-05,
      "loss": 1.8464,
      "step": 1953
    },
    {
      "epoch": 0.06593990959851104,
      "grad_norm": 2.391841411590576,
      "learning_rate": 4.99692159912661e-05,
      "loss": 1.8179,
      "step": 1984
    },
    {
      "epoch": 0.06697022068598776,
      "grad_norm": 2.2471864223480225,
      "learning_rate": 4.996490869988546e-05,
      "loss": 1.8149,
      "step": 2015
    },
    {
      "epoch": 0.0680005317734645,
      "grad_norm": 2.5497376918792725,
      "learning_rate": 4.996031968290326e-05,
      "loss": 1.8099,
      "step": 2046
    },
    {
      "epoch": 0.06903084286094124,
      "grad_norm": 2.330463409423828,
      "learning_rate": 4.995544899210594e-05,
      "loss": 1.8267,
      "step": 2077
    },
    {
      "epoch": 0.07006115394841797,
      "grad_norm": 2.3259341716766357,
      "learning_rate": 4.9950296682458583e-05,
      "loss": 1.7801,
      "step": 2108
    },
    {
      "epoch": 0.07109146503589471,
      "grad_norm": 2.1711952686309814,
      "learning_rate": 4.994486281210429e-05,
      "loss": 1.7961,
      "step": 2139
    },
    {
      "epoch": 0.07212177612337145,
      "grad_norm": 2.1808884143829346,
      "learning_rate": 4.9939147442363566e-05,
      "loss": 1.8109,
      "step": 2170
    },
    {
      "epoch": 0.07315208721084818,
      "grad_norm": 2.089256525039673,
      "learning_rate": 4.9933150637733574e-05,
      "loss": 1.8026,
      "step": 2201
    },
    {
      "epoch": 0.07418239829832492,
      "grad_norm": 2.0864951610565186,
      "learning_rate": 4.992687246588743e-05,
      "loss": 1.7753,
      "step": 2232
    },
    {
      "epoch": 0.07521270938580164,
      "grad_norm": 2.36157488822937,
      "learning_rate": 4.992031299767347e-05,
      "loss": 1.7746,
      "step": 2263
    },
    {
      "epoch": 0.07624302047327838,
      "grad_norm": 2.5334439277648926,
      "learning_rate": 4.9913472307114386e-05,
      "loss": 1.7927,
      "step": 2294
    },
    {
      "epoch": 0.07727333156075512,
      "grad_norm": 2.2565715312957764,
      "learning_rate": 4.9906350471406446e-05,
      "loss": 1.7668,
      "step": 2325
    },
    {
      "epoch": 0.07830364264823185,
      "grad_norm": 2.1043128967285156,
      "learning_rate": 4.989894757091861e-05,
      "loss": 1.7771,
      "step": 2356
    },
    {
      "epoch": 0.07933395373570859,
      "grad_norm": 1.9659819602966309,
      "learning_rate": 4.989126368919158e-05,
      "loss": 1.7666,
      "step": 2387
    },
    {
      "epoch": 0.08036426482318532,
      "grad_norm": 2.0778403282165527,
      "learning_rate": 4.988329891293693e-05,
      "loss": 1.7405,
      "step": 2418
    },
    {
      "epoch": 0.08139457591066206,
      "grad_norm": 2.1767923831939697,
      "learning_rate": 4.987505333203608e-05,
      "loss": 1.7495,
      "step": 2449
    },
    {
      "epoch": 0.0824248869981388,
      "grad_norm": 2.260143280029297,
      "learning_rate": 4.9866527039539276e-05,
      "loss": 1.7504,
      "step": 2480
    },
    {
      "epoch": 0.08345519808561552,
      "grad_norm": 2.18271803855896,
      "learning_rate": 4.9857720131664594e-05,
      "loss": 1.7456,
      "step": 2511
    },
    {
      "epoch": 0.08448550917309226,
      "grad_norm": 2.209594964981079,
      "learning_rate": 4.9848632707796773e-05,
      "loss": 1.7528,
      "step": 2542
    },
    {
      "epoch": 0.085515820260569,
      "grad_norm": 2.0666229724884033,
      "learning_rate": 4.9839264870486155e-05,
      "loss": 1.7517,
      "step": 2573
    },
    {
      "epoch": 0.08654613134804573,
      "grad_norm": 2.1070454120635986,
      "learning_rate": 4.9829616725447526e-05,
      "loss": 1.7474,
      "step": 2604
    },
    {
      "epoch": 0.08757644243552247,
      "grad_norm": 1.9430303573608398,
      "learning_rate": 4.981968838155888e-05,
      "loss": 1.7348,
      "step": 2635
    },
    {
      "epoch": 0.0886067535229992,
      "grad_norm": 1.9638925790786743,
      "learning_rate": 4.980947995086024e-05,
      "loss": 1.7202,
      "step": 2666
    },
    {
      "epoch": 0.08963706461047594,
      "grad_norm": 1.8845652341842651,
      "learning_rate": 4.979899154855234e-05,
      "loss": 1.7375,
      "step": 2697
    },
    {
      "epoch": 0.09066737569795268,
      "grad_norm": 5.712058067321777,
      "learning_rate": 4.9788223292995386e-05,
      "loss": 1.7379,
      "step": 2728
    },
    {
      "epoch": 0.0916976867854294,
      "grad_norm": 1.9520670175552368,
      "learning_rate": 4.977717530570768e-05,
      "loss": 1.7302,
      "step": 2759
    },
    {
      "epoch": 0.09272799787290614,
      "grad_norm": 1.8802224397659302,
      "learning_rate": 4.976584771136425e-05,
      "loss": 1.74,
      "step": 2790
    },
    {
      "epoch": 0.09375830896038288,
      "grad_norm": 2.1098153591156006,
      "learning_rate": 4.975424063779547e-05,
      "loss": 1.7024,
      "step": 2821
    },
    {
      "epoch": 0.09478862004785961,
      "grad_norm": 2.1568291187286377,
      "learning_rate": 4.974235421598557e-05,
      "loss": 1.7131,
      "step": 2852
    },
    {
      "epoch": 0.09581893113533635,
      "grad_norm": 1.8769980669021606,
      "learning_rate": 4.973018858007122e-05,
      "loss": 1.7008,
      "step": 2883
    },
    {
      "epoch": 0.09684924222281308,
      "grad_norm": 1.8325533866882324,
      "learning_rate": 4.9717743867339963e-05,
      "loss": 1.7058,
      "step": 2914
    },
    {
      "epoch": 0.09787955331028982,
      "grad_norm": 2.086416721343994,
      "learning_rate": 4.9705020218228695e-05,
      "loss": 1.711,
      "step": 2945
    },
    {
      "epoch": 0.09890986439776656,
      "grad_norm": 1.8294793367385864,
      "learning_rate": 4.969201777632205e-05,
      "loss": 1.6998,
      "step": 2976
    },
    {
      "epoch": 0.09994017548524328,
      "grad_norm": 2.0608153343200684,
      "learning_rate": 4.9678736688350846e-05,
      "loss": 1.6948,
      "step": 3007
    },
    {
      "epoch": 0.10097048657272002,
      "grad_norm": 3.2166008949279785,
      "learning_rate": 4.966517710419033e-05,
      "loss": 1.6788,
      "step": 3038
    },
    {
      "epoch": 0.10200079766019676,
      "grad_norm": 1.9431313276290894,
      "learning_rate": 4.965133917685858e-05,
      "loss": 1.7115,
      "step": 3069
    },
    {
      "epoch": 0.10303110874767349,
      "grad_norm": 1.967512845993042,
      "learning_rate": 4.9637223062514714e-05,
      "loss": 1.7033,
      "step": 3100
    },
    {
      "epoch": 0.10406141983515023,
      "grad_norm": 1.9253389835357666,
      "learning_rate": 4.962282892045718e-05,
      "loss": 1.6856,
      "step": 3131
    },
    {
      "epoch": 0.10509173092262696,
      "grad_norm": 1.986840009689331,
      "learning_rate": 4.9608156913121904e-05,
      "loss": 1.723,
      "step": 3162
    },
    {
      "epoch": 0.1061220420101037,
      "grad_norm": 1.83523690700531,
      "learning_rate": 4.959320720608049e-05,
      "loss": 1.6912,
      "step": 3193
    },
    {
      "epoch": 0.10715235309758044,
      "grad_norm": 2.1271955966949463,
      "learning_rate": 4.9577979968038354e-05,
      "loss": 1.7032,
      "step": 3224
    },
    {
      "epoch": 0.10818266418505716,
      "grad_norm": 1.8383768796920776,
      "learning_rate": 4.956247537083282e-05,
      "loss": 1.6726,
      "step": 3255
    },
    {
      "epoch": 0.1092129752725339,
      "grad_norm": 1.8806651830673218,
      "learning_rate": 4.9546693589431145e-05,
      "loss": 1.6817,
      "step": 3286
    },
    {
      "epoch": 0.11024328636001064,
      "grad_norm": 1.7535260915756226,
      "learning_rate": 4.9530634801928595e-05,
      "loss": 1.6875,
      "step": 3317
    },
    {
      "epoch": 0.11127359744748737,
      "grad_norm": 1.765906810760498,
      "learning_rate": 4.9514299189546395e-05,
      "loss": 1.6859,
      "step": 3348
    },
    {
      "epoch": 0.11230390853496411,
      "grad_norm": 1.869828462600708,
      "learning_rate": 4.949768693662973e-05,
      "loss": 1.6915,
      "step": 3379
    },
    {
      "epoch": 0.11333421962244083,
      "grad_norm": 1.8347504138946533,
      "learning_rate": 4.948079823064559e-05,
      "loss": 1.6859,
      "step": 3410
    },
    {
      "epoch": 0.11436453070991758,
      "grad_norm": 1.7692474126815796,
      "learning_rate": 4.946363326218074e-05,
      "loss": 1.6565,
      "step": 3441
    },
    {
      "epoch": 0.11539484179739432,
      "grad_norm": 1.8231885433197021,
      "learning_rate": 4.9446192224939525e-05,
      "loss": 1.686,
      "step": 3472
    },
    {
      "epoch": 0.11642515288487104,
      "grad_norm": 1.7155958414077759,
      "learning_rate": 4.942847531574167e-05,
      "loss": 1.6538,
      "step": 3503
    },
    {
      "epoch": 0.11745546397234778,
      "grad_norm": 1.787183403968811,
      "learning_rate": 4.941048273452008e-05,
      "loss": 1.6776,
      "step": 3534
    },
    {
      "epoch": 0.11848577505982451,
      "grad_norm": 1.741213083267212,
      "learning_rate": 4.9392214684318605e-05,
      "loss": 1.6784,
      "step": 3565
    },
    {
      "epoch": 0.11951608614730125,
      "grad_norm": 1.7836824655532837,
      "learning_rate": 4.93736713712897e-05,
      "loss": 1.6557,
      "step": 3596
    },
    {
      "epoch": 0.12054639723477799,
      "grad_norm": 1.7103859186172485,
      "learning_rate": 4.9354853004692124e-05,
      "loss": 1.6606,
      "step": 3627
    },
    {
      "epoch": 0.12157670832225471,
      "grad_norm": 1.7865506410598755,
      "learning_rate": 4.93357597968886e-05,
      "loss": 1.6409,
      "step": 3658
    },
    {
      "epoch": 0.12260701940973145,
      "grad_norm": 1.7770143747329712,
      "learning_rate": 4.931639196334338e-05,
      "loss": 1.6574,
      "step": 3689
    },
    {
      "epoch": 0.1236373304972082,
      "grad_norm": 1.857575535774231,
      "learning_rate": 4.9296749722619826e-05,
      "loss": 1.6724,
      "step": 3720
    },
    {
      "epoch": 0.12466764158468492,
      "grad_norm": 1.8742581605911255,
      "learning_rate": 4.9276833296377966e-05,
      "loss": 1.6506,
      "step": 3751
    },
    {
      "epoch": 0.12569795267216166,
      "grad_norm": 1.827668309211731,
      "learning_rate": 4.925664290937196e-05,
      "loss": 1.6523,
      "step": 3782
    },
    {
      "epoch": 0.1267282637596384,
      "grad_norm": 1.7517486810684204,
      "learning_rate": 4.9236178789447576e-05,
      "loss": 1.6459,
      "step": 3813
    },
    {
      "epoch": 0.12775857484711514,
      "grad_norm": 1.8109570741653442,
      "learning_rate": 4.921544116753962e-05,
      "loss": 1.6614,
      "step": 3844
    },
    {
      "epoch": 0.12878888593459187,
      "grad_norm": 1.692597508430481,
      "learning_rate": 4.919443027766935e-05,
      "loss": 1.6431,
      "step": 3875
    },
    {
      "epoch": 0.1298191970220686,
      "grad_norm": 1.8650025129318237,
      "learning_rate": 4.91731463569418e-05,
      "loss": 1.6466,
      "step": 3906
    },
    {
      "epoch": 0.13084950810954532,
      "grad_norm": 1.6794081926345825,
      "learning_rate": 4.915158964554312e-05,
      "loss": 1.6504,
      "step": 3937
    },
    {
      "epoch": 0.13187981919702207,
      "grad_norm": 1.7685374021530151,
      "learning_rate": 4.912976038673786e-05,
      "loss": 1.6446,
      "step": 3968
    },
    {
      "epoch": 0.1329101302844988,
      "grad_norm": 1.7601110935211182,
      "learning_rate": 4.9107658826866254e-05,
      "loss": 1.631,
      "step": 3999
    },
    {
      "epoch": 0.13394044137197553,
      "grad_norm": 2.0616064071655273,
      "learning_rate": 4.908528521534139e-05,
      "loss": 1.6476,
      "step": 4030
    },
    {
      "epoch": 0.13497075245945228,
      "grad_norm": 1.8973504304885864,
      "learning_rate": 4.906263980464644e-05,
      "loss": 1.6582,
      "step": 4061
    },
    {
      "epoch": 0.136001063546929,
      "grad_norm": 1.7768895626068115,
      "learning_rate": 4.903972285033178e-05,
      "loss": 1.6159,
      "step": 4092
    },
    {
      "epoch": 0.13703137463440573,
      "grad_norm": 1.8264424800872803,
      "learning_rate": 4.901653461101213e-05,
      "loss": 1.6289,
      "step": 4123
    },
    {
      "epoch": 0.1380616857218825,
      "grad_norm": 1.7140119075775146,
      "learning_rate": 4.8993075348363626e-05,
      "loss": 1.6357,
      "step": 4154
    },
    {
      "epoch": 0.13909199680935921,
      "grad_norm": 1.6964486837387085,
      "learning_rate": 4.896934532712084e-05,
      "loss": 1.6233,
      "step": 4185
    },
    {
      "epoch": 0.14012230789683594,
      "grad_norm": 1.8008025884628296,
      "learning_rate": 4.8945344815073846e-05,
      "loss": 1.637,
      "step": 4216
    },
    {
      "epoch": 0.1411526189843127,
      "grad_norm": 1.562730073928833,
      "learning_rate": 4.892107408306516e-05,
      "loss": 1.6379,
      "step": 4247
    },
    {
      "epoch": 0.14218293007178942,
      "grad_norm": 1.8273371458053589,
      "learning_rate": 4.889653340498669e-05,
      "loss": 1.6246,
      "step": 4278
    },
    {
      "epoch": 0.14321324115926615,
      "grad_norm": 56.33716583251953,
      "learning_rate": 4.8871723057776664e-05,
      "loss": 1.6457,
      "step": 4309
    },
    {
      "epoch": 0.1442435522467429,
      "grad_norm": 1.746523380279541,
      "learning_rate": 4.8846643321416476e-05,
      "loss": 1.6343,
      "step": 4340
    },
    {
      "epoch": 0.14527386333421963,
      "grad_norm": 1.7737531661987305,
      "learning_rate": 4.882129447892753e-05,
      "loss": 1.6447,
      "step": 4371
    },
    {
      "epoch": 0.14630417442169635,
      "grad_norm": 1.660485863685608,
      "learning_rate": 4.8795676816368076e-05,
      "loss": 1.6192,
      "step": 4402
    },
    {
      "epoch": 0.14733448550917308,
      "grad_norm": 1.6823406219482422,
      "learning_rate": 4.876979062282995e-05,
      "loss": 1.6253,
      "step": 4433
    },
    {
      "epoch": 0.14836479659664983,
      "grad_norm": 7.78139066696167,
      "learning_rate": 4.8743636190435325e-05,
      "loss": 1.6234,
      "step": 4464
    },
    {
      "epoch": 0.14939510768412656,
      "grad_norm": 1.7426058053970337,
      "learning_rate": 4.871721381433344e-05,
      "loss": 1.6337,
      "step": 4495
    },
    {
      "epoch": 0.1504254187716033,
      "grad_norm": 1.6294783353805542,
      "learning_rate": 4.869052379269719e-05,
      "loss": 1.6217,
      "step": 4526
    },
    {
      "epoch": 0.15145572985908004,
      "grad_norm": 1.6523306369781494,
      "learning_rate": 4.866356642671985e-05,
      "loss": 1.605,
      "step": 4557
    },
    {
      "epoch": 0.15248604094655677,
      "grad_norm": 1.8571300506591797,
      "learning_rate": 4.8636342020611634e-05,
      "loss": 1.6218,
      "step": 4588
    },
    {
      "epoch": 0.1535163520340335,
      "grad_norm": 1.7754936218261719,
      "learning_rate": 4.860885088159626e-05,
      "loss": 1.6171,
      "step": 4619
    },
    {
      "epoch": 0.15454666312151025,
      "grad_norm": 1.91987943649292,
      "learning_rate": 4.858109331990751e-05,
      "loss": 1.6167,
      "step": 4650
    },
    {
      "epoch": 0.15557697420898697,
      "grad_norm": 1.5994452238082886,
      "learning_rate": 4.855306964878567e-05,
      "loss": 1.5951,
      "step": 4681
    },
    {
      "epoch": 0.1566072852964637,
      "grad_norm": 1.6490916013717651,
      "learning_rate": 4.8524780184474084e-05,
      "loss": 1.616,
      "step": 4712
    },
    {
      "epoch": 0.15763759638394045,
      "grad_norm": 1.5921640396118164,
      "learning_rate": 4.8496225246215496e-05,
      "loss": 1.6346,
      "step": 4743
    },
    {
      "epoch": 0.15866790747141718,
      "grad_norm": 1.6729261875152588,
      "learning_rate": 4.8467405156248505e-05,
      "loss": 1.6165,
      "step": 4774
    },
    {
      "epoch": 0.1596982185588939,
      "grad_norm": 1.628113031387329,
      "learning_rate": 4.843832023980392e-05,
      "loss": 1.6119,
      "step": 4805
    },
    {
      "epoch": 0.16072852964637063,
      "grad_norm": 1.651647925376892,
      "learning_rate": 4.840897082510106e-05,
      "loss": 1.5997,
      "step": 4836
    },
    {
      "epoch": 0.1617588407338474,
      "grad_norm": 1.5297720432281494,
      "learning_rate": 4.8379357243344084e-05,
      "loss": 1.6242,
      "step": 4867
    },
    {
      "epoch": 0.1627891518213241,
      "grad_norm": 1.5779869556427002,
      "learning_rate": 4.8349479828718236e-05,
      "loss": 1.6149,
      "step": 4898
    },
    {
      "epoch": 0.16381946290880084,
      "grad_norm": 1.5843939781188965,
      "learning_rate": 4.8319338918386075e-05,
      "loss": 1.5926,
      "step": 4929
    },
    {
      "epoch": 0.1648497739962776,
      "grad_norm": 2.3762106895446777,
      "learning_rate": 4.828893485248369e-05,
      "loss": 1.6108,
      "step": 4960
    },
    {
      "epoch": 0.16588008508375432,
      "grad_norm": 1.5871953964233398,
      "learning_rate": 4.825826797411682e-05,
      "loss": 1.6103,
      "step": 4991
    },
    {
      "epoch": 0.16691039617123105,
      "grad_norm": 1.5934125185012817,
      "learning_rate": 4.822733862935702e-05,
      "loss": 1.6091,
      "step": 5022
    },
    {
      "epoch": 0.1679407072587078,
      "grad_norm": 1.6997628211975098,
      "learning_rate": 4.819614716723775e-05,
      "loss": 1.6098,
      "step": 5053
    },
    {
      "epoch": 0.16897101834618453,
      "grad_norm": 1.682849645614624,
      "learning_rate": 4.8164693939750425e-05,
      "loss": 1.599,
      "step": 5084
    },
    {
      "epoch": 0.17000132943366125,
      "grad_norm": 1.709743857383728,
      "learning_rate": 4.813297930184042e-05,
      "loss": 1.6194,
      "step": 5115
    },
    {
      "epoch": 0.171031640521138,
      "grad_norm": 1.725879430770874,
      "learning_rate": 4.810100361140314e-05,
      "loss": 1.6115,
      "step": 5146
    },
    {
      "epoch": 0.17206195160861473,
      "grad_norm": 1.6710290908813477,
      "learning_rate": 4.8068767229279885e-05,
      "loss": 1.6032,
      "step": 5177
    },
    {
      "epoch": 0.17309226269609146,
      "grad_norm": 1.6156634092330933,
      "learning_rate": 4.8036270519253854e-05,
      "loss": 1.5973,
      "step": 5208
    },
    {
      "epoch": 0.1741225737835682,
      "grad_norm": 1.5654059648513794,
      "learning_rate": 4.8003513848046e-05,
      "loss": 1.5817,
      "step": 5239
    },
    {
      "epoch": 0.17515288487104494,
      "grad_norm": 1.5789822340011597,
      "learning_rate": 4.79704975853109e-05,
      "loss": 1.6138,
      "step": 5270
    },
    {
      "epoch": 0.17618319595852167,
      "grad_norm": 1.6022037267684937,
      "learning_rate": 4.793722210363262e-05,
      "loss": 1.5998,
      "step": 5301
    },
    {
      "epoch": 0.1772135070459984,
      "grad_norm": 1.5142741203308105,
      "learning_rate": 4.7903687778520414e-05,
      "loss": 1.6061,
      "step": 5332
    },
    {
      "epoch": 0.17824381813347515,
      "grad_norm": 1.6454212665557861,
      "learning_rate": 4.7869894988404593e-05,
      "loss": 1.6063,
      "step": 5363
    },
    {
      "epoch": 0.17927412922095187,
      "grad_norm": 1.5250823497772217,
      "learning_rate": 4.783584411463221e-05,
      "loss": 1.6038,
      "step": 5394
    },
    {
      "epoch": 0.1803044403084286,
      "grad_norm": 1.5829335451126099,
      "learning_rate": 4.780153554146274e-05,
      "loss": 1.5949,
      "step": 5425
    },
    {
      "epoch": 0.18133475139590535,
      "grad_norm": 1.5342432260513306,
      "learning_rate": 4.7766969656063766e-05,
      "loss": 1.5913,
      "step": 5456
    },
    {
      "epoch": 0.18236506248338208,
      "grad_norm": 1.6397250890731812,
      "learning_rate": 4.773214684850662e-05,
      "loss": 1.6102,
      "step": 5487
    },
    {
      "epoch": 0.1833953735708588,
      "grad_norm": 1.5228471755981445,
      "learning_rate": 4.769706751176193e-05,
      "loss": 1.5885,
      "step": 5518
    },
    {
      "epoch": 0.18442568465833556,
      "grad_norm": 1.6186103820800781,
      "learning_rate": 4.7661732041695264e-05,
      "loss": 1.6086,
      "step": 5549
    },
    {
      "epoch": 0.18545599574581229,
      "grad_norm": 1.6024582386016846,
      "learning_rate": 4.762614083706258e-05,
      "loss": 1.6004,
      "step": 5580
    },
    {
      "epoch": 0.186486306833289,
      "grad_norm": 1.5443711280822754,
      "learning_rate": 4.759029429950581e-05,
      "loss": 1.6048,
      "step": 5611
    },
    {
      "epoch": 0.18751661792076577,
      "grad_norm": 1.4831629991531372,
      "learning_rate": 4.7554192833548235e-05,
      "loss": 1.5841,
      "step": 5642
    },
    {
      "epoch": 0.1885469290082425,
      "grad_norm": 1.6426068544387817,
      "learning_rate": 4.751783684659e-05,
      "loss": 1.587,
      "step": 5673
    },
    {
      "epoch": 0.18957724009571922,
      "grad_norm": 1.4609078168869019,
      "learning_rate": 4.748122674890348e-05,
      "loss": 1.5945,
      "step": 5704
    },
    {
      "epoch": 0.19060755118319597,
      "grad_norm": 1.5365614891052246,
      "learning_rate": 4.7444362953628654e-05,
      "loss": 1.5737,
      "step": 5735
    },
    {
      "epoch": 0.1916378622706727,
      "grad_norm": 1.5755670070648193,
      "learning_rate": 4.7407245876768424e-05,
      "loss": 1.5862,
      "step": 5766
    },
    {
      "epoch": 0.19266817335814942,
      "grad_norm": 1.6469846963882446,
      "learning_rate": 4.736987593718397e-05,
      "loss": 1.5663,
      "step": 5797
    },
    {
      "epoch": 0.19369848444562615,
      "grad_norm": 1.5927278995513916,
      "learning_rate": 4.733225355658999e-05,
      "loss": 1.5776,
      "step": 5828
    },
    {
      "epoch": 0.1947287955331029,
      "grad_norm": 1.5593287944793701,
      "learning_rate": 4.7294379159549926e-05,
      "loss": 1.579,
      "step": 5859
    },
    {
      "epoch": 0.19575910662057963,
      "grad_norm": 1.534055233001709,
      "learning_rate": 4.725625317347119e-05,
      "loss": 1.6017,
      "step": 5890
    },
    {
      "epoch": 0.19678941770805636,
      "grad_norm": 1.5846387147903442,
      "learning_rate": 4.7217876028600374e-05,
      "loss": 1.5739,
      "step": 5921
    },
    {
      "epoch": 0.1978197287955331,
      "grad_norm": 1.5377682447433472,
      "learning_rate": 4.717924815801832e-05,
      "loss": 1.57,
      "step": 5952
    },
    {
      "epoch": 0.19885003988300984,
      "grad_norm": 1.467956781387329,
      "learning_rate": 4.714036999763532e-05,
      "loss": 1.5736,
      "step": 5983
    },
    {
      "epoch": 0.19988035097048656,
      "grad_norm": 1.601070523262024,
      "learning_rate": 4.7101241986186116e-05,
      "loss": 1.5861,
      "step": 6014
    },
    {
      "epoch": 0.20091066205796332,
      "grad_norm": 1.5051921606063843,
      "learning_rate": 4.7061864565225e-05,
      "loss": 1.5735,
      "step": 6045
    },
    {
      "epoch": 0.20194097314544004,
      "grad_norm": 1.462843418121338,
      "learning_rate": 4.702223817912081e-05,
      "loss": 1.582,
      "step": 6076
    }
  ],
  "logging_steps": 31,
  "max_steps": 30517,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 3052,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.527890108044542e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}