{
  "best_metric": 0.47253116965293884,
  "best_model_checkpoint": "Action_all_10_class/checkpoint-1500",
  "epoch": 10.0,
  "eval_steps": 100,
  "global_step": 2790,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "grad_norm": 1.6752883195877075,
      "learning_rate": 9.96415770609319e-05,
      "loss": 2.247,
      "step": 10
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8152681589126587,
      "learning_rate": 9.928315412186381e-05,
      "loss": 2.1602,
      "step": 20
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.164091110229492,
      "learning_rate": 9.892473118279571e-05,
      "loss": 2.0409,
      "step": 30
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6417195796966553,
      "learning_rate": 9.85663082437276e-05,
      "loss": 1.8783,
      "step": 40
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.199918746948242,
      "learning_rate": 9.820788530465951e-05,
      "loss": 1.7363,
      "step": 50
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.9530537128448486,
      "learning_rate": 9.78494623655914e-05,
      "loss": 1.5894,
      "step": 60
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.777188301086426,
      "learning_rate": 9.74910394265233e-05,
      "loss": 1.4617,
      "step": 70
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2515618801116943,
      "learning_rate": 9.713261648745519e-05,
      "loss": 1.4804,
      "step": 80
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.89544939994812,
      "learning_rate": 9.681003584229391e-05,
      "loss": 1.2876,
      "step": 90
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.372511625289917,
      "learning_rate": 9.645161290322581e-05,
      "loss": 1.2411,
      "step": 100
    },
    {
      "epoch": 0.36,
      "eval_accuracy": 0.754601226993865,
      "eval_loss": 1.1517137289047241,
      "eval_runtime": 18.3534,
      "eval_samples_per_second": 53.287,
      "eval_steps_per_second": 6.702,
      "step": 100
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.142375946044922,
      "learning_rate": 9.609318996415772e-05,
      "loss": 1.2389,
      "step": 110
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.8880653381347656,
      "learning_rate": 9.573476702508962e-05,
      "loss": 1.179,
      "step": 120
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.925424575805664,
      "learning_rate": 9.541218637992831e-05,
      "loss": 1.1092,
      "step": 130
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.03208589553833,
      "learning_rate": 9.505376344086023e-05,
      "loss": 1.1389,
      "step": 140
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.35840106010437,
      "learning_rate": 9.469534050179211e-05,
      "loss": 1.0824,
      "step": 150
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.0924174785614014,
      "learning_rate": 9.433691756272402e-05,
      "loss": 0.957,
      "step": 160
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.47157096862793,
      "learning_rate": 9.397849462365592e-05,
      "loss": 1.0797,
      "step": 170
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.9026050567626953,
      "learning_rate": 9.362007168458782e-05,
      "loss": 1.0961,
      "step": 180
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.673151969909668,
      "learning_rate": 9.326164874551971e-05,
      "loss": 1.0421,
      "step": 190
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5587886571884155,
      "learning_rate": 9.290322580645162e-05,
      "loss": 0.8932,
      "step": 200
    },
    {
      "epoch": 0.72,
      "eval_accuracy": 0.7975460122699386,
      "eval_loss": 0.785646915435791,
      "eval_runtime": 14.5537,
      "eval_samples_per_second": 67.199,
      "eval_steps_per_second": 8.451,
      "step": 200
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.5420029163360596,
      "learning_rate": 9.254480286738351e-05,
      "loss": 0.9145,
      "step": 210
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.8418989181518555,
      "learning_rate": 9.218637992831541e-05,
      "loss": 0.8473,
      "step": 220
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.9392898082733154,
      "learning_rate": 9.182795698924731e-05,
      "loss": 0.7867,
      "step": 230
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.856507778167725,
      "learning_rate": 9.146953405017922e-05,
      "loss": 0.9951,
      "step": 240
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.318254232406616,
      "learning_rate": 9.111111111111112e-05,
      "loss": 0.7852,
      "step": 250
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2948904037475586,
      "learning_rate": 9.0752688172043e-05,
      "loss": 0.9565,
      "step": 260
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.516943454742432,
      "learning_rate": 9.039426523297492e-05,
      "loss": 0.7034,
      "step": 270
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9273302555084229,
      "learning_rate": 9.003584229390681e-05,
      "loss": 0.8603,
      "step": 280
    },
    {
      "epoch": 1.04,
      "grad_norm": 9.084336280822754,
      "learning_rate": 8.967741935483871e-05,
      "loss": 0.6914,
      "step": 290
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.446329116821289,
      "learning_rate": 8.931899641577061e-05,
      "loss": 0.6907,
      "step": 300
    },
    {
      "epoch": 1.08,
      "eval_accuracy": 0.8220858895705522,
      "eval_loss": 0.6636486649513245,
      "eval_runtime": 14.1733,
      "eval_samples_per_second": 69.003,
      "eval_steps_per_second": 8.678,
      "step": 300
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.061271667480469,
      "learning_rate": 8.896057347670252e-05,
      "loss": 0.7286,
      "step": 310
    },
    {
      "epoch": 1.15,
      "grad_norm": 3.7218878269195557,
      "learning_rate": 8.86021505376344e-05,
      "loss": 0.7241,
      "step": 320
    },
    {
      "epoch": 1.18,
      "grad_norm": 3.3088059425354004,
      "learning_rate": 8.824372759856632e-05,
      "loss": 0.7051,
      "step": 330
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.993207931518555,
      "learning_rate": 8.788530465949821e-05,
      "loss": 0.6652,
      "step": 340
    },
    {
      "epoch": 1.25,
      "grad_norm": 3.7464661598205566,
      "learning_rate": 8.752688172043011e-05,
      "loss": 0.6898,
      "step": 350
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.751115798950195,
      "learning_rate": 8.716845878136201e-05,
      "loss": 0.8783,
      "step": 360
    },
    {
      "epoch": 1.33,
      "grad_norm": 3.408888578414917,
      "learning_rate": 8.681003584229391e-05,
      "loss": 0.6584,
      "step": 370
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.679331302642822,
      "learning_rate": 8.645161290322581e-05,
      "loss": 0.7819,
      "step": 380
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.112971305847168,
      "learning_rate": 8.60931899641577e-05,
      "loss": 0.5751,
      "step": 390
    },
    {
      "epoch": 1.43,
      "grad_norm": 3.214542865753174,
      "learning_rate": 8.573476702508962e-05,
      "loss": 0.5841,
      "step": 400
    },
    {
      "epoch": 1.43,
      "eval_accuracy": 0.8159509202453987,
      "eval_loss": 0.6388102173805237,
      "eval_runtime": 14.2283,
      "eval_samples_per_second": 68.736,
      "eval_steps_per_second": 8.645,
      "step": 400
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.963754653930664,
      "learning_rate": 8.53763440860215e-05,
      "loss": 0.7805,
      "step": 410
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.3538126945495605,
      "learning_rate": 8.501792114695341e-05,
      "loss": 0.6053,
      "step": 420
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.444653034210205,
      "learning_rate": 8.465949820788531e-05,
      "loss": 0.764,
      "step": 430
    },
    {
      "epoch": 1.58,
      "grad_norm": 3.9525132179260254,
      "learning_rate": 8.430107526881721e-05,
      "loss": 0.684,
      "step": 440
    },
    {
      "epoch": 1.61,
      "grad_norm": 3.4026594161987305,
      "learning_rate": 8.39426523297491e-05,
      "loss": 0.5835,
      "step": 450
    },
    {
      "epoch": 1.65,
      "grad_norm": 3.843327522277832,
      "learning_rate": 8.358422939068102e-05,
      "loss": 0.5352,
      "step": 460
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.771996021270752,
      "learning_rate": 8.32258064516129e-05,
      "loss": 0.631,
      "step": 470
    },
    {
      "epoch": 1.72,
      "grad_norm": 2.9874844551086426,
      "learning_rate": 8.28673835125448e-05,
      "loss": 0.611,
      "step": 480
    },
    {
      "epoch": 1.76,
      "grad_norm": 2.693260431289673,
      "learning_rate": 8.250896057347671e-05,
      "loss": 0.6031,
      "step": 490
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.329994201660156,
      "learning_rate": 8.215053763440861e-05,
      "loss": 0.5425,
      "step": 500
    },
    {
      "epoch": 1.79,
      "eval_accuracy": 0.843558282208589,
      "eval_loss": 0.5870967507362366,
      "eval_runtime": 14.0184,
      "eval_samples_per_second": 69.765,
      "eval_steps_per_second": 8.774,
      "step": 500
    },
    {
      "epoch": 1.83,
      "grad_norm": 3.7939178943634033,
      "learning_rate": 8.179211469534051e-05,
      "loss": 0.5645,
      "step": 510
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.04762077331543,
      "learning_rate": 8.14336917562724e-05,
      "loss": 0.5252,
      "step": 520
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.459852695465088,
      "learning_rate": 8.107526881720431e-05,
      "loss": 0.7369,
      "step": 530
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.254125118255615,
      "learning_rate": 8.07168458781362e-05,
      "loss": 0.6754,
      "step": 540
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.496582984924316,
      "learning_rate": 8.03584229390681e-05,
      "loss": 0.5504,
      "step": 550
    },
    {
      "epoch": 2.01,
      "grad_norm": 7.471122741699219,
      "learning_rate": 8e-05,
      "loss": 0.643,
      "step": 560
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.464521408081055,
      "learning_rate": 7.964157706093191e-05,
      "loss": 0.575,
      "step": 570
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7192500829696655,
      "learning_rate": 7.92831541218638e-05,
      "loss": 0.4709,
      "step": 580
    },
    {
      "epoch": 2.11,
      "grad_norm": 6.228574275970459,
      "learning_rate": 7.892473118279571e-05,
      "loss": 0.5173,
      "step": 590
    },
    {
      "epoch": 2.15,
      "grad_norm": 6.098475933074951,
      "learning_rate": 7.85663082437276e-05,
      "loss": 0.5929,
      "step": 600
    },
    {
      "epoch": 2.15,
      "eval_accuracy": 0.8210633946830266,
      "eval_loss": 0.5646098852157593,
      "eval_runtime": 14.2986,
      "eval_samples_per_second": 68.398,
      "eval_steps_per_second": 8.602,
      "step": 600
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.2586510181427,
      "learning_rate": 7.82078853046595e-05,
      "loss": 0.5392,
      "step": 610
    },
    {
      "epoch": 2.22,
      "grad_norm": 6.211545944213867,
      "learning_rate": 7.784946236559139e-05,
      "loss": 0.5032,
      "step": 620
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.691046953201294,
      "learning_rate": 7.74910394265233e-05,
      "loss": 0.5918,
      "step": 630
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.3668406009674072,
      "learning_rate": 7.71326164874552e-05,
      "loss": 0.6054,
      "step": 640
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.9461565017700195,
      "learning_rate": 7.67741935483871e-05,
      "loss": 0.4755,
      "step": 650
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.586925506591797,
      "learning_rate": 7.6415770609319e-05,
      "loss": 0.4968,
      "step": 660
    },
    {
      "epoch": 2.4,
      "grad_norm": 5.659130573272705,
      "learning_rate": 7.60573476702509e-05,
      "loss": 0.4894,
      "step": 670
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.458271503448486,
      "learning_rate": 7.56989247311828e-05,
      "loss": 0.4056,
      "step": 680
    },
    {
      "epoch": 2.47,
      "grad_norm": 5.034984588623047,
      "learning_rate": 7.53405017921147e-05,
      "loss": 0.6026,
      "step": 690
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.018673419952393,
      "learning_rate": 7.49820788530466e-05,
      "loss": 0.4406,
      "step": 700
    },
    {
      "epoch": 2.51,
      "eval_accuracy": 0.8404907975460123,
      "eval_loss": 0.543875515460968,
      "eval_runtime": 14.0957,
      "eval_samples_per_second": 69.383,
      "eval_steps_per_second": 8.726,
      "step": 700
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.96732759475708,
      "learning_rate": 7.462365591397849e-05,
      "loss": 0.6851,
      "step": 710
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.6570851802825928,
      "learning_rate": 7.42652329749104e-05,
      "loss": 0.4738,
      "step": 720
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.4345834255218506,
      "learning_rate": 7.39068100358423e-05,
      "loss": 0.5739,
      "step": 730
    },
    {
      "epoch": 2.65,
      "grad_norm": 5.19066858291626,
      "learning_rate": 7.35483870967742e-05,
      "loss": 0.4721,
      "step": 740
    },
    {
      "epoch": 2.69,
      "grad_norm": 5.423111915588379,
      "learning_rate": 7.318996415770609e-05,
      "loss": 0.5271,
      "step": 750
    },
    {
      "epoch": 2.72,
      "grad_norm": 5.806092262268066,
      "learning_rate": 7.2831541218638e-05,
      "loss": 0.5163,
      "step": 760
    },
    {
      "epoch": 2.76,
      "grad_norm": 6.047975540161133,
      "learning_rate": 7.247311827956989e-05,
      "loss": 0.5061,
      "step": 770
    },
    {
      "epoch": 2.8,
      "grad_norm": 7.956634044647217,
      "learning_rate": 7.211469534050179e-05,
      "loss": 0.5639,
      "step": 780
    },
    {
      "epoch": 2.83,
      "grad_norm": 5.515818119049072,
      "learning_rate": 7.17562724014337e-05,
      "loss": 0.5214,
      "step": 790
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.654768943786621,
      "learning_rate": 7.13978494623656e-05,
      "loss": 0.4541,
      "step": 800
    },
    {
      "epoch": 2.87,
      "eval_accuracy": 0.8415132924335378,
      "eval_loss": 0.5317940711975098,
      "eval_runtime": 14.2499,
      "eval_samples_per_second": 68.632,
      "eval_steps_per_second": 8.632,
      "step": 800
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.857801914215088,
      "learning_rate": 7.10394265232975e-05,
      "loss": 0.415,
      "step": 810
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.023741722106934,
      "learning_rate": 7.06810035842294e-05,
      "loss": 0.3754,
      "step": 820
    },
    {
      "epoch": 2.97,
      "grad_norm": 7.845776081085205,
      "learning_rate": 7.03225806451613e-05,
      "loss": 0.464,
      "step": 830
    },
    {
      "epoch": 3.01,
      "grad_norm": 10.843958854675293,
      "learning_rate": 6.996415770609319e-05,
      "loss": 0.655,
      "step": 840
    },
    {
      "epoch": 3.05,
      "grad_norm": 2.662417411804199,
      "learning_rate": 6.960573476702509e-05,
      "loss": 0.4086,
      "step": 850
    },
    {
      "epoch": 3.08,
      "grad_norm": 6.569747447967529,
      "learning_rate": 6.924731182795699e-05,
      "loss": 0.4273,
      "step": 860
    },
    {
      "epoch": 3.12,
      "grad_norm": 3.297159433364868,
      "learning_rate": 6.88888888888889e-05,
      "loss": 0.4426,
      "step": 870
    },
    {
      "epoch": 3.15,
      "grad_norm": 5.205211639404297,
      "learning_rate": 6.853046594982078e-05,
      "loss": 0.3445,
      "step": 880
    },
    {
      "epoch": 3.19,
      "grad_norm": 6.75899076461792,
      "learning_rate": 6.81720430107527e-05,
      "loss": 0.466,
      "step": 890
    },
    {
      "epoch": 3.23,
      "grad_norm": 4.749955654144287,
      "learning_rate": 6.781362007168459e-05,
      "loss": 0.3835,
      "step": 900
    },
    {
      "epoch": 3.23,
      "eval_accuracy": 0.8343558282208589,
      "eval_loss": 0.5224626660346985,
      "eval_runtime": 14.3694,
      "eval_samples_per_second": 68.062,
      "eval_steps_per_second": 8.56,
      "step": 900
    },
    {
      "epoch": 3.26,
      "grad_norm": 5.145068645477295,
      "learning_rate": 6.745519713261649e-05,
      "loss": 0.3779,
      "step": 910
    },
    {
      "epoch": 3.3,
      "grad_norm": 4.831909656524658,
      "learning_rate": 6.709677419354839e-05,
      "loss": 0.3744,
      "step": 920
    },
    {
      "epoch": 3.33,
      "grad_norm": 3.3703436851501465,
      "learning_rate": 6.673835125448029e-05,
      "loss": 0.3374,
      "step": 930
    },
    {
      "epoch": 3.37,
      "grad_norm": 10.194363594055176,
      "learning_rate": 6.63799283154122e-05,
      "loss": 0.4076,
      "step": 940
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.0064858198165894,
      "learning_rate": 6.602150537634408e-05,
      "loss": 0.4434,
      "step": 950
    },
    {
      "epoch": 3.44,
      "grad_norm": 2.8392670154571533,
      "learning_rate": 6.5663082437276e-05,
      "loss": 0.318,
      "step": 960
    },
    {
      "epoch": 3.48,
      "grad_norm": 2.6734464168548584,
      "learning_rate": 6.530465949820789e-05,
      "loss": 0.5079,
      "step": 970
    },
    {
      "epoch": 3.51,
      "grad_norm": 2.2691242694854736,
      "learning_rate": 6.494623655913979e-05,
      "loss": 0.3849,
      "step": 980
    },
    {
      "epoch": 3.55,
      "grad_norm": 3.17016863822937,
      "learning_rate": 6.458781362007169e-05,
      "loss": 0.3427,
      "step": 990
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.534572720527649,
      "learning_rate": 6.422939068100359e-05,
      "loss": 0.3924,
      "step": 1000
    },
    {
      "epoch": 3.58,
      "eval_accuracy": 0.8302658486707567,
      "eval_loss": 0.5514972805976868,
      "eval_runtime": 14.2143,
      "eval_samples_per_second": 68.804,
      "eval_steps_per_second": 8.653,
      "step": 1000
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.171107292175293,
      "learning_rate": 6.387096774193548e-05,
      "loss": 0.3694,
      "step": 1010
    },
    {
      "epoch": 3.66,
      "grad_norm": 4.934769630432129,
      "learning_rate": 6.35125448028674e-05,
      "loss": 0.3607,
      "step": 1020
    },
    {
      "epoch": 3.69,
      "grad_norm": 7.391957759857178,
      "learning_rate": 6.315412186379928e-05,
      "loss": 0.3429,
      "step": 1030
    },
    {
      "epoch": 3.73,
      "grad_norm": 7.1492156982421875,
      "learning_rate": 6.279569892473119e-05,
      "loss": 0.4211,
      "step": 1040
    },
    {
      "epoch": 3.76,
      "grad_norm": 5.837622165679932,
      "learning_rate": 6.243727598566309e-05,
      "loss": 0.3857,
      "step": 1050
    },
    {
      "epoch": 3.8,
      "grad_norm": 5.638124465942383,
      "learning_rate": 6.207885304659499e-05,
      "loss": 0.4379,
      "step": 1060
    },
    {
      "epoch": 3.84,
      "grad_norm": 5.501093864440918,
      "learning_rate": 6.172043010752688e-05,
      "loss": 0.3903,
      "step": 1070
    },
    {
      "epoch": 3.87,
      "grad_norm": 6.717010021209717,
      "learning_rate": 6.136200716845878e-05,
      "loss": 0.3844,
      "step": 1080
    },
    {
      "epoch": 3.91,
      "grad_norm": 8.419461250305176,
      "learning_rate": 6.100358422939069e-05,
      "loss": 0.3705,
      "step": 1090
    },
    {
      "epoch": 3.94,
      "grad_norm": 7.757752895355225,
      "learning_rate": 6.064516129032258e-05,
      "loss": 0.5741,
      "step": 1100
    },
    {
      "epoch": 3.94,
      "eval_accuracy": 0.8251533742331288,
      "eval_loss": 0.5519386529922485,
      "eval_runtime": 13.907,
      "eval_samples_per_second": 70.324,
      "eval_steps_per_second": 8.844,
      "step": 1100
    },
    {
      "epoch": 3.98,
      "grad_norm": 3.006796360015869,
      "learning_rate": 6.028673835125448e-05,
      "loss": 0.5208,
      "step": 1110
    },
    {
      "epoch": 4.01,
      "grad_norm": 4.00032901763916,
      "learning_rate": 5.9928315412186386e-05,
      "loss": 0.3386,
      "step": 1120
    },
    {
      "epoch": 4.05,
      "grad_norm": 1.4214438199996948,
      "learning_rate": 5.956989247311828e-05,
      "loss": 0.2971,
      "step": 1130
    },
    {
      "epoch": 4.09,
      "grad_norm": 4.046026229858398,
      "learning_rate": 5.9211469534050176e-05,
      "loss": 0.4146,
      "step": 1140
    },
    {
      "epoch": 4.12,
      "grad_norm": 2.4826581478118896,
      "learning_rate": 5.8853046594982085e-05,
      "loss": 0.4243,
      "step": 1150
    },
    {
      "epoch": 4.16,
      "grad_norm": 4.898519515991211,
      "learning_rate": 5.849462365591398e-05,
      "loss": 0.3691,
      "step": 1160
    },
    {
      "epoch": 4.19,
      "grad_norm": 7.121424198150635,
      "learning_rate": 5.813620071684588e-05,
      "loss": 0.4513,
      "step": 1170
    },
    {
      "epoch": 4.23,
      "grad_norm": 5.212203025817871,
      "learning_rate": 5.7777777777777776e-05,
      "loss": 0.4,
      "step": 1180
    },
    {
      "epoch": 4.27,
      "grad_norm": 11.669204711914062,
      "learning_rate": 5.7419354838709685e-05,
      "loss": 0.3429,
      "step": 1190
    },
    {
      "epoch": 4.3,
      "grad_norm": 11.0433349609375,
      "learning_rate": 5.706093189964158e-05,
      "loss": 0.3991,
      "step": 1200
    },
    {
      "epoch": 4.3,
      "eval_accuracy": 0.8445807770961146,
      "eval_loss": 0.49900951981544495,
      "eval_runtime": 13.9712,
      "eval_samples_per_second": 70.001,
      "eval_steps_per_second": 8.804,
      "step": 1200
    },
    {
      "epoch": 4.34,
      "grad_norm": 3.952008008956909,
      "learning_rate": 5.6702508960573475e-05,
      "loss": 0.3259,
      "step": 1210
    },
    {
      "epoch": 4.37,
      "grad_norm": 2.535245895385742,
      "learning_rate": 5.6344086021505384e-05,
      "loss": 0.2201,
      "step": 1220
    },
    {
      "epoch": 4.41,
      "grad_norm": 5.427143096923828,
      "learning_rate": 5.598566308243728e-05,
      "loss": 0.3986,
      "step": 1230
    },
    {
      "epoch": 4.44,
      "grad_norm": 6.882002353668213,
      "learning_rate": 5.5627240143369174e-05,
      "loss": 0.4718,
      "step": 1240
    },
    {
      "epoch": 4.48,
      "grad_norm": 3.7098076343536377,
      "learning_rate": 5.526881720430108e-05,
      "loss": 0.3653,
      "step": 1250
    },
    {
      "epoch": 4.52,
      "grad_norm": 3.8536739349365234,
      "learning_rate": 5.491039426523298e-05,
      "loss": 0.358,
      "step": 1260
    },
    {
      "epoch": 4.55,
      "grad_norm": 6.399543762207031,
      "learning_rate": 5.455197132616487e-05,
      "loss": 0.4017,
      "step": 1270
    },
    {
      "epoch": 4.59,
      "grad_norm": 5.091586112976074,
      "learning_rate": 5.419354838709678e-05,
      "loss": 0.493,
      "step": 1280
    },
    {
      "epoch": 4.62,
      "grad_norm": 5.683386325836182,
      "learning_rate": 5.3835125448028676e-05,
      "loss": 0.5526,
      "step": 1290
    },
    {
      "epoch": 4.66,
      "grad_norm": 3.3343377113342285,
      "learning_rate": 5.347670250896057e-05,
      "loss": 0.4732,
      "step": 1300
    },
    {
      "epoch": 4.66,
      "eval_accuracy": 0.8302658486707567,
      "eval_loss": 0.5335860252380371,
      "eval_runtime": 14.2962,
      "eval_samples_per_second": 68.41,
      "eval_steps_per_second": 8.604,
      "step": 1300
    },
    {
      "epoch": 4.7,
      "grad_norm": 5.0791521072387695,
      "learning_rate": 5.311827956989247e-05,
      "loss": 0.3433,
      "step": 1310
    },
    {
      "epoch": 4.73,
      "grad_norm": 4.7593607902526855,
      "learning_rate": 5.2759856630824375e-05,
      "loss": 0.3617,
      "step": 1320
    },
    {
      "epoch": 4.77,
      "grad_norm": 3.0220556259155273,
      "learning_rate": 5.2401433691756277e-05,
      "loss": 0.3243,
      "step": 1330
    },
    {
      "epoch": 4.8,
      "grad_norm": 1.736549735069275,
      "learning_rate": 5.204301075268817e-05,
      "loss": 0.3488,
      "step": 1340
    },
    {
      "epoch": 4.84,
      "grad_norm": 4.916406154632568,
      "learning_rate": 5.168458781362008e-05,
      "loss": 0.2955,
      "step": 1350
    },
    {
      "epoch": 4.87,
      "grad_norm": 9.70033073425293,
      "learning_rate": 5.1326164874551975e-05,
      "loss": 0.392,
      "step": 1360
    },
    {
      "epoch": 4.91,
      "grad_norm": 2.5204570293426514,
      "learning_rate": 5.096774193548387e-05,
      "loss": 0.2706,
      "step": 1370
    },
    {
      "epoch": 4.95,
      "grad_norm": 4.673634052276611,
      "learning_rate": 5.060931899641578e-05,
      "loss": 0.3257,
      "step": 1380
    },
    {
      "epoch": 4.98,
      "grad_norm": 5.239686012268066,
      "learning_rate": 5.0250896057347674e-05,
      "loss": 0.3738,
      "step": 1390
    },
    {
      "epoch": 5.02,
      "grad_norm": 3.337782144546509,
      "learning_rate": 4.989247311827957e-05,
      "loss": 0.3324,
      "step": 1400
    },
    {
      "epoch": 5.02,
      "eval_accuracy": 0.8282208588957055,
      "eval_loss": 0.5350744128227234,
      "eval_runtime": 14.1615,
      "eval_samples_per_second": 69.061,
      "eval_steps_per_second": 8.686,
      "step": 1400
    },
    {
      "epoch": 5.05,
      "grad_norm": 2.9601023197174072,
      "learning_rate": 4.953405017921147e-05,
      "loss": 0.2966,
      "step": 1410
    },
    {
      "epoch": 5.09,
      "grad_norm": 10.412677764892578,
      "learning_rate": 4.917562724014337e-05,
      "loss": 0.3838,
      "step": 1420
    },
    {
      "epoch": 5.13,
      "grad_norm": 1.986757755279541,
      "learning_rate": 4.881720430107527e-05,
      "loss": 0.4096,
      "step": 1430
    },
    {
      "epoch": 5.16,
      "grad_norm": 1.2032238245010376,
      "learning_rate": 4.845878136200717e-05,
      "loss": 0.3183,
      "step": 1440
    },
    {
      "epoch": 5.2,
      "grad_norm": 1.7991982698440552,
      "learning_rate": 4.810035842293907e-05,
      "loss": 0.2713,
      "step": 1450
    },
    {
      "epoch": 5.23,
      "grad_norm": 7.0565385818481445,
      "learning_rate": 4.774193548387097e-05,
      "loss": 0.2414,
      "step": 1460
    },
    {
      "epoch": 5.27,
      "grad_norm": 3.8213694095611572,
      "learning_rate": 4.7383512544802875e-05,
      "loss": 0.363,
      "step": 1470
    },
    {
      "epoch": 5.3,
      "grad_norm": 2.4153404235839844,
      "learning_rate": 4.702508960573477e-05,
      "loss": 0.349,
      "step": 1480
    },
    {
      "epoch": 5.34,
      "grad_norm": 2.161435842514038,
      "learning_rate": 4.666666666666667e-05,
      "loss": 0.2626,
      "step": 1490
    },
    {
      "epoch": 5.38,
      "grad_norm": 5.8719282150268555,
      "learning_rate": 4.630824372759857e-05,
      "loss": 0.3433,
      "step": 1500
    },
    {
      "epoch": 5.38,
      "eval_accuracy": 0.8517382413087935,
      "eval_loss": 0.47253116965293884,
      "eval_runtime": 14.1174,
      "eval_samples_per_second": 69.276,
      "eval_steps_per_second": 8.713,
      "step": 1500
    },
    {
      "epoch": 5.41,
      "grad_norm": 5.560650825500488,
      "learning_rate": 4.594982078853047e-05,
      "loss": 0.2775,
      "step": 1510
    },
    {
      "epoch": 5.45,
      "grad_norm": 2.5716190338134766,
      "learning_rate": 4.559139784946237e-05,
      "loss": 0.2537,
      "step": 1520
    },
    {
      "epoch": 5.48,
      "grad_norm": 5.602371692657471,
      "learning_rate": 4.5232974910394265e-05,
      "loss": 0.3127,
      "step": 1530
    },
    {
      "epoch": 5.52,
      "grad_norm": 5.320743083953857,
      "learning_rate": 4.487455197132617e-05,
      "loss": 0.2878,
      "step": 1540
    },
    {
      "epoch": 5.56,
      "grad_norm": 4.925574779510498,
      "learning_rate": 4.451612903225807e-05,
      "loss": 0.2842,
      "step": 1550
    },
    {
      "epoch": 5.59,
      "grad_norm": 3.862708568572998,
      "learning_rate": 4.4157706093189964e-05,
      "loss": 0.3004,
      "step": 1560
    },
    {
      "epoch": 5.63,
      "grad_norm": 3.0864365100860596,
      "learning_rate": 4.3799283154121866e-05,
      "loss": 0.2638,
      "step": 1570
    },
    {
      "epoch": 5.66,
      "grad_norm": 3.5530340671539307,
      "learning_rate": 4.344086021505376e-05,
      "loss": 0.2439,
      "step": 1580
    },
    {
      "epoch": 5.7,
      "grad_norm": 5.929757118225098,
      "learning_rate": 4.308243727598566e-05,
      "loss": 0.3235,
      "step": 1590
    },
    {
      "epoch": 5.73,
      "grad_norm": 3.567291736602783,
      "learning_rate": 4.2724014336917564e-05,
      "loss": 0.2187,
      "step": 1600
    },
    {
      "epoch": 5.73,
      "eval_accuracy": 0.8466257668711656,
      "eval_loss": 0.5041967034339905,
      "eval_runtime": 14.1345,
      "eval_samples_per_second": 69.193,
      "eval_steps_per_second": 8.702,
      "step": 1600
    },
    {
      "epoch": 5.77,
      "grad_norm": 0.8105212450027466,
      "learning_rate": 4.2365591397849466e-05,
      "loss": 0.2792,
      "step": 1610
    },
    {
      "epoch": 5.81,
      "grad_norm": 3.8311142921447754,
      "learning_rate": 4.200716845878137e-05,
      "loss": 0.2313,
      "step": 1620
    },
    {
      "epoch": 5.84,
      "grad_norm": 6.618185043334961,
      "learning_rate": 4.164874551971326e-05,
      "loss": 0.2668,
      "step": 1630
    },
    {
      "epoch": 5.88,
      "grad_norm": 7.174586772918701,
      "learning_rate": 4.1290322580645165e-05,
      "loss": 0.2684,
      "step": 1640
    },
    {
      "epoch": 5.91,
      "grad_norm": 6.7463788986206055,
      "learning_rate": 4.093189964157707e-05,
      "loss": 0.2825,
      "step": 1650
    },
    {
      "epoch": 5.95,
      "grad_norm": 2.402876853942871,
      "learning_rate": 4.057347670250896e-05,
      "loss": 0.2986,
      "step": 1660
    },
    {
      "epoch": 5.99,
      "grad_norm": 2.343830108642578,
      "learning_rate": 4.0215053763440864e-05,
      "loss": 0.3086,
      "step": 1670
    },
    {
      "epoch": 6.02,
      "grad_norm": 3.3489632606506348,
      "learning_rate": 3.985663082437276e-05,
      "loss": 0.2844,
      "step": 1680
    },
    {
      "epoch": 6.06,
      "grad_norm": 2.076068162918091,
      "learning_rate": 3.949820788530466e-05,
      "loss": 0.1609,
      "step": 1690
    },
    {
      "epoch": 6.09,
      "grad_norm": 3.9822871685028076,
      "learning_rate": 3.913978494623656e-05,
      "loss": 0.2952,
      "step": 1700
    },
    {
      "epoch": 6.09,
      "eval_accuracy": 0.8548057259713702,
      "eval_loss": 0.5240095257759094,
      "eval_runtime": 14.3375,
      "eval_samples_per_second": 68.213,
      "eval_steps_per_second": 8.579,
      "step": 1700
    },
    {
      "epoch": 6.13,
      "grad_norm": 3.8955540657043457,
      "learning_rate": 3.878136200716846e-05,
      "loss": 0.3564,
      "step": 1710
    },
    {
      "epoch": 6.16,
      "grad_norm": 2.374730348587036,
      "learning_rate": 3.842293906810036e-05,
      "loss": 0.2467,
      "step": 1720
    },
    {
      "epoch": 6.2,
      "grad_norm": 6.376945495605469,
      "learning_rate": 3.8064516129032254e-05,
      "loss": 0.2584,
      "step": 1730
    },
    {
      "epoch": 6.24,
      "grad_norm": 5.645608901977539,
      "learning_rate": 3.7706093189964156e-05,
      "loss": 0.3091,
      "step": 1740
    },
    {
      "epoch": 6.27,
      "grad_norm": 2.882622241973877,
      "learning_rate": 3.734767025089606e-05,
      "loss": 0.2471,
      "step": 1750
    },
    {
      "epoch": 6.31,
      "grad_norm": 8.7712984085083,
      "learning_rate": 3.698924731182796e-05,
      "loss": 0.2595,
      "step": 1760
    },
    {
      "epoch": 6.34,
      "grad_norm": 2.5900111198425293,
      "learning_rate": 3.663082437275986e-05,
      "loss": 0.3474,
      "step": 1770
    },
    {
      "epoch": 6.38,
      "grad_norm": 0.14616957306861877,
      "learning_rate": 3.627240143369176e-05,
      "loss": 0.2777,
      "step": 1780
    },
    {
      "epoch": 6.42,
      "grad_norm": 5.120488166809082,
      "learning_rate": 3.591397849462366e-05,
      "loss": 0.2867,
      "step": 1790
    },
    {
      "epoch": 6.45,
      "grad_norm": 4.957128524780273,
      "learning_rate": 3.555555555555556e-05,
      "loss": 0.2687,
      "step": 1800
    },
    {
      "epoch": 6.45,
      "eval_accuracy": 0.83640081799591,
      "eval_loss": 0.5522516965866089,
      "eval_runtime": 14.281,
      "eval_samples_per_second": 68.482,
      "eval_steps_per_second": 8.613,
      "step": 1800
    },
    {
      "epoch": 6.49,
      "grad_norm": 6.335272312164307,
      "learning_rate": 3.5197132616487455e-05,
      "loss": 0.2486,
      "step": 1810
    },
    {
      "epoch": 6.52,
      "grad_norm": 3.3235371112823486,
      "learning_rate": 3.483870967741936e-05,
      "loss": 0.1797,
      "step": 1820
    },
    {
      "epoch": 6.56,
      "grad_norm": 0.23497584462165833,
      "learning_rate": 3.448028673835126e-05,
      "loss": 0.2214,
      "step": 1830
    },
    {
      "epoch": 6.59,
      "grad_norm": 1.0797181129455566,
      "learning_rate": 3.4121863799283154e-05,
      "loss": 0.2369,
      "step": 1840
    },
    {
      "epoch": 6.63,
      "grad_norm": 3.9156014919281006,
      "learning_rate": 3.3763440860215055e-05,
      "loss": 0.2826,
      "step": 1850
    },
    {
      "epoch": 6.67,
      "grad_norm": 9.87369155883789,
      "learning_rate": 3.340501792114695e-05,
      "loss": 0.3399,
      "step": 1860
    },
    {
      "epoch": 6.7,
      "grad_norm": 3.916306972503662,
      "learning_rate": 3.304659498207885e-05,
      "loss": 0.3758,
      "step": 1870
    },
    {
      "epoch": 6.74,
      "grad_norm": 7.597679615020752,
      "learning_rate": 3.2688172043010754e-05,
      "loss": 0.2505,
      "step": 1880
    },
    {
      "epoch": 6.77,
      "grad_norm": 4.102473735809326,
      "learning_rate": 3.2329749103942656e-05,
      "loss": 0.2692,
      "step": 1890
    },
    {
      "epoch": 6.81,
      "grad_norm": 8.36500358581543,
      "learning_rate": 3.197132616487455e-05,
      "loss": 0.3111,
      "step": 1900
    },
    {
      "epoch": 6.81,
      "eval_accuracy": 0.8496932515337423,
      "eval_loss": 0.53035569190979,
      "eval_runtime": 14.3586,
      "eval_samples_per_second": 68.112,
      "eval_steps_per_second": 8.566,
      "step": 1900
    },
    {
      "epoch": 6.85,
      "grad_norm": 5.454440593719482,
      "learning_rate": 3.161290322580645e-05,
      "loss": 0.2373,
      "step": 1910
    },
    {
      "epoch": 6.88,
      "grad_norm": 0.6554473638534546,
      "learning_rate": 3.1254480286738355e-05,
      "loss": 0.1937,
      "step": 1920
    },
    {
      "epoch": 6.92,
      "grad_norm": 0.24610257148742676,
      "learning_rate": 3.0896057347670256e-05,
      "loss": 0.2149,
      "step": 1930
    },
    {
      "epoch": 6.95,
      "grad_norm": 4.611593246459961,
      "learning_rate": 3.053763440860215e-05,
      "loss": 0.2791,
      "step": 1940
    },
    {
      "epoch": 6.99,
      "grad_norm": 8.88548755645752,
      "learning_rate": 3.0179211469534053e-05,
      "loss": 0.3238,
      "step": 1950
    },
    {
      "epoch": 7.03,
      "grad_norm": 2.9050915241241455,
      "learning_rate": 2.9820788530465955e-05,
      "loss": 0.2527,
      "step": 1960
    },
    {
      "epoch": 7.06,
      "grad_norm": 6.722965717315674,
      "learning_rate": 2.946236559139785e-05,
      "loss": 0.2451,
      "step": 1970
    },
    {
      "epoch": 7.1,
      "grad_norm": 4.379875183105469,
      "learning_rate": 2.9103942652329752e-05,
      "loss": 0.269,
      "step": 1980
    },
    {
      "epoch": 7.13,
      "grad_norm": 2.847017288208008,
      "learning_rate": 2.8745519713261647e-05,
      "loss": 0.2776,
      "step": 1990
    },
    {
      "epoch": 7.17,
      "grad_norm": 5.179224491119385,
      "learning_rate": 2.838709677419355e-05,
      "loss": 0.2431,
      "step": 2000
    },
    {
      "epoch": 7.17,
      "eval_accuracy": 0.8568507157464212,
      "eval_loss": 0.5104272365570068,
      "eval_runtime": 14.3416,
      "eval_samples_per_second": 68.193,
      "eval_steps_per_second": 8.576,
      "step": 2000
    },
    {
      "epoch": 7.2,
      "grad_norm": 2.521162986755371,
      "learning_rate": 2.802867383512545e-05,
      "loss": 0.1631,
      "step": 2010
    },
    {
      "epoch": 7.24,
      "grad_norm": 4.0884246826171875,
      "learning_rate": 2.767025089605735e-05,
      "loss": 0.2731,
      "step": 2020
    },
    {
      "epoch": 7.28,
      "grad_norm": 4.443497180938721,
      "learning_rate": 2.731182795698925e-05,
      "loss": 0.2594,
      "step": 2030
    },
    {
      "epoch": 7.31,
      "grad_norm": 2.3641021251678467,
      "learning_rate": 2.6953405017921146e-05,
      "loss": 0.2477,
      "step": 2040
    },
    {
      "epoch": 7.35,
      "grad_norm": 0.09834872931241989,
      "learning_rate": 2.6594982078853048e-05,
      "loss": 0.2298,
      "step": 2050
    },
    {
      "epoch": 7.38,
      "grad_norm": 3.825798749923706,
      "learning_rate": 2.623655913978495e-05,
      "loss": 0.2955,
      "step": 2060
    },
    {
      "epoch": 7.42,
      "grad_norm": 1.7917243242263794,
      "learning_rate": 2.5878136200716844e-05,
      "loss": 0.3034,
      "step": 2070
    },
    {
      "epoch": 7.46,
      "grad_norm": 2.0665640830993652,
      "learning_rate": 2.5519713261648746e-05,
      "loss": 0.2204,
      "step": 2080
    },
    {
      "epoch": 7.49,
      "grad_norm": 1.5993545055389404,
      "learning_rate": 2.5161290322580645e-05,
      "loss": 0.2114,
      "step": 2090
    },
    {
      "epoch": 7.53,
      "grad_norm": 7.3012871742248535,
      "learning_rate": 2.4802867383512547e-05,
      "loss": 0.3265,
      "step": 2100
    },
    {
      "epoch": 7.53,
      "eval_accuracy": 0.869120654396728,
      "eval_loss": 0.5085219740867615,
      "eval_runtime": 14.1317,
      "eval_samples_per_second": 69.206,
      "eval_steps_per_second": 8.704,
      "step": 2100
    },
    {
      "epoch": 7.56,
      "grad_norm": 1.1089341640472412,
      "learning_rate": 2.4444444444444445e-05,
      "loss": 0.2762,
      "step": 2110
    },
    {
      "epoch": 7.6,
      "grad_norm": 4.004220485687256,
      "learning_rate": 2.4086021505376347e-05,
      "loss": 0.2453,
      "step": 2120
    },
    {
      "epoch": 7.63,
      "grad_norm": 7.738345623016357,
      "learning_rate": 2.3727598566308245e-05,
      "loss": 0.2757,
      "step": 2130
    },
    {
      "epoch": 7.67,
      "grad_norm": 3.836663007736206,
      "learning_rate": 2.3405017921146955e-05,
      "loss": 0.298,
      "step": 2140
    },
    {
      "epoch": 7.71,
      "grad_norm": 1.0338927507400513,
      "learning_rate": 2.3046594982078853e-05,
      "loss": 0.1816,
      "step": 2150
    },
    {
      "epoch": 7.74,
      "grad_norm": 1.0525509119033813,
      "learning_rate": 2.268817204301075e-05,
      "loss": 0.145,
      "step": 2160
    },
    {
      "epoch": 7.78,
      "grad_norm": 5.745477676391602,
      "learning_rate": 2.2329749103942653e-05,
      "loss": 0.2143,
      "step": 2170
    },
    {
      "epoch": 7.81,
      "grad_norm": 4.428494930267334,
      "learning_rate": 2.1971326164874552e-05,
      "loss": 0.2301,
      "step": 2180
    },
    {
      "epoch": 7.85,
      "grad_norm": 5.054660320281982,
      "learning_rate": 2.1612903225806454e-05,
      "loss": 0.2317,
      "step": 2190
    },
    {
      "epoch": 7.89,
      "grad_norm": 1.9087740182876587,
      "learning_rate": 2.1254480286738352e-05,
      "loss": 0.2595,
      "step": 2200
    },
    {
      "epoch": 7.89,
      "eval_accuracy": 0.8568507157464212,
      "eval_loss": 0.5014929175376892,
      "eval_runtime": 14.0273,
      "eval_samples_per_second": 69.721,
      "eval_steps_per_second": 8.769,
      "step": 2200
    },
    {
      "epoch": 7.92,
      "grad_norm": 3.2280988693237305,
      "learning_rate": 2.0896057347670254e-05,
      "loss": 0.1503,
      "step": 2210
    },
    {
      "epoch": 7.96,
      "grad_norm": 7.906033039093018,
      "learning_rate": 2.0537634408602152e-05,
      "loss": 0.2864,
      "step": 2220
    },
    {
      "epoch": 7.99,
      "grad_norm": 8.353536605834961,
      "learning_rate": 2.017921146953405e-05,
      "loss": 0.2027,
      "step": 2230
    },
    {
      "epoch": 8.03,
      "grad_norm": 6.651542663574219,
      "learning_rate": 1.982078853046595e-05,
      "loss": 0.2513,
      "step": 2240
    },
    {
      "epoch": 8.06,
      "grad_norm": 4.423330307006836,
      "learning_rate": 1.9462365591397848e-05,
      "loss": 0.2591,
      "step": 2250
    },
    {
      "epoch": 8.1,
      "grad_norm": 7.127723217010498,
      "learning_rate": 1.910394265232975e-05,
      "loss": 0.2341,
      "step": 2260
    },
    {
      "epoch": 8.14,
      "grad_norm": 1.9292550086975098,
      "learning_rate": 1.874551971326165e-05,
      "loss": 0.1744,
      "step": 2270
    },
    {
      "epoch": 8.17,
      "grad_norm": 3.043701648712158,
      "learning_rate": 1.838709677419355e-05,
      "loss": 0.3194,
      "step": 2280
    },
    {
      "epoch": 8.21,
      "grad_norm": 5.514953136444092,
      "learning_rate": 1.8028673835125448e-05,
      "loss": 0.2454,
      "step": 2290
    },
    {
      "epoch": 8.24,
      "grad_norm": 1.1831002235412598,
      "learning_rate": 1.767025089605735e-05,
      "loss": 0.1825,
      "step": 2300
    },
    {
      "epoch": 8.24,
      "eval_accuracy": 0.8619631901840491,
      "eval_loss": 0.4919591546058655,
      "eval_runtime": 14.1344,
      "eval_samples_per_second": 69.193,
      "eval_steps_per_second": 8.702,
      "step": 2300
    },
    {
      "epoch": 8.28,
      "grad_norm": 3.9401707649230957,
      "learning_rate": 1.7311827956989248e-05,
      "loss": 0.1456,
      "step": 2310
    },
    {
      "epoch": 8.32,
      "grad_norm": 7.038700103759766,
      "learning_rate": 1.6989247311827958e-05,
      "loss": 0.1804,
      "step": 2320
    },
    {
      "epoch": 8.35,
      "grad_norm": 9.696996688842773,
      "learning_rate": 1.6630824372759856e-05,
      "loss": 0.144,
      "step": 2330
    },
    {
      "epoch": 8.39,
      "grad_norm": 1.7609977722167969,
      "learning_rate": 1.6272401433691755e-05,
      "loss": 0.1852,
      "step": 2340
    },
    {
      "epoch": 8.42,
      "grad_norm": 0.1569293588399887,
      "learning_rate": 1.5913978494623657e-05,
      "loss": 0.302,
      "step": 2350
    },
    {
      "epoch": 8.46,
      "grad_norm": 6.627655506134033,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 0.2418,
      "step": 2360
    },
    {
      "epoch": 8.49,
      "grad_norm": 4.680299282073975,
      "learning_rate": 1.5197132616487455e-05,
      "loss": 0.2042,
      "step": 2370
    },
    {
      "epoch": 8.53,
      "grad_norm": 3.413140058517456,
      "learning_rate": 1.4838709677419355e-05,
      "loss": 0.2608,
      "step": 2380
    },
    {
      "epoch": 8.57,
      "grad_norm": 0.6003228425979614,
      "learning_rate": 1.4480286738351257e-05,
      "loss": 0.2494,
      "step": 2390
    },
    {
      "epoch": 8.6,
      "grad_norm": 9.677336692810059,
      "learning_rate": 1.4121863799283155e-05,
      "loss": 0.2602,
      "step": 2400
    },
    {
      "epoch": 8.6,
      "eval_accuracy": 0.8619631901840491,
      "eval_loss": 0.5016282200813293,
      "eval_runtime": 14.4248,
      "eval_samples_per_second": 67.8,
      "eval_steps_per_second": 8.527,
      "step": 2400
    },
    {
      "epoch": 8.64,
      "grad_norm": 3.607775926589966,
      "learning_rate": 1.3763440860215054e-05,
      "loss": 0.2171,
      "step": 2410
    },
    {
      "epoch": 8.67,
      "grad_norm": 0.4323180019855499,
      "learning_rate": 1.3405017921146954e-05,
      "loss": 0.1981,
      "step": 2420
    },
    {
      "epoch": 8.71,
      "grad_norm": 9.15799331665039,
      "learning_rate": 1.3046594982078852e-05,
      "loss": 0.2206,
      "step": 2430
    },
    {
      "epoch": 8.75,
      "grad_norm": 3.574291229248047,
      "learning_rate": 1.2688172043010754e-05,
      "loss": 0.2388,
      "step": 2440
    },
    {
      "epoch": 8.78,
      "grad_norm": 4.433730125427246,
      "learning_rate": 1.2329749103942653e-05,
      "loss": 0.1705,
      "step": 2450
    },
    {
      "epoch": 8.82,
      "grad_norm": 9.609087944030762,
      "learning_rate": 1.1971326164874553e-05,
      "loss": 0.3498,
      "step": 2460
    },
    {
      "epoch": 8.85,
      "grad_norm": 5.335048675537109,
      "learning_rate": 1.1612903225806453e-05,
      "loss": 0.2864,
      "step": 2470
    },
    {
      "epoch": 8.89,
      "grad_norm": 1.8026251792907715,
      "learning_rate": 1.1254480286738351e-05,
      "loss": 0.2135,
      "step": 2480
    },
    {
      "epoch": 8.92,
      "grad_norm": 1.9891669750213623,
      "learning_rate": 1.0896057347670251e-05,
      "loss": 0.2214,
      "step": 2490
    },
    {
      "epoch": 8.96,
      "grad_norm": 0.7803227305412292,
      "learning_rate": 1.0537634408602151e-05,
      "loss": 0.2628,
      "step": 2500
    },
    {
      "epoch": 8.96,
      "eval_accuracy": 0.8680981595092024,
      "eval_loss": 0.47455406188964844,
      "eval_runtime": 14.3927,
      "eval_samples_per_second": 67.951,
      "eval_steps_per_second": 8.546,
      "step": 2500
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.4858655035495758,
      "learning_rate": 1.017921146953405e-05,
      "loss": 0.1861,
      "step": 2510
    },
    {
      "epoch": 9.03,
      "grad_norm": 6.302251815795898,
      "learning_rate": 9.82078853046595e-06,
      "loss": 0.2009,
      "step": 2520
    },
    {
      "epoch": 9.07,
      "grad_norm": 2.5692965984344482,
      "learning_rate": 9.46236559139785e-06,
      "loss": 0.1785,
      "step": 2530
    },
    {
      "epoch": 9.1,
      "grad_norm": 6.564579010009766,
      "learning_rate": 9.10394265232975e-06,
      "loss": 0.1284,
      "step": 2540
    },
    {
      "epoch": 9.14,
      "grad_norm": 5.155548572540283,
      "learning_rate": 8.745519713261649e-06,
      "loss": 0.2494,
      "step": 2550
    },
    {
      "epoch": 9.18,
      "grad_norm": 4.653995990753174,
      "learning_rate": 8.387096774193549e-06,
      "loss": 0.1142,
      "step": 2560
    },
    {
      "epoch": 9.21,
      "grad_norm": 5.491698741912842,
      "learning_rate": 8.028673835125449e-06,
      "loss": 0.1887,
      "step": 2570
    },
    {
      "epoch": 9.25,
      "grad_norm": 1.841364860534668,
      "learning_rate": 7.670250896057347e-06,
      "loss": 0.2236,
      "step": 2580
    },
    {
      "epoch": 9.28,
      "grad_norm": 4.787356376647949,
      "learning_rate": 7.3118279569892475e-06,
      "loss": 0.2155,
      "step": 2590
    },
    {
      "epoch": 9.32,
      "grad_norm": 0.1550242006778717,
      "learning_rate": 6.953405017921147e-06,
      "loss": 0.1024,
      "step": 2600
    },
    {
      "epoch": 9.32,
      "eval_accuracy": 0.869120654396728,
      "eval_loss": 0.48181313276290894,
      "eval_runtime": 14.3216,
      "eval_samples_per_second": 68.288,
      "eval_steps_per_second": 8.588,
      "step": 2600
    },
    {
      "epoch": 9.35,
      "grad_norm": 4.603553771972656,
      "learning_rate": 6.594982078853047e-06,
      "loss": 0.292,
      "step": 2610
    },
    {
      "epoch": 9.39,
      "grad_norm": 6.353521823883057,
      "learning_rate": 6.236559139784946e-06,
      "loss": 0.1462,
      "step": 2620
    },
    {
      "epoch": 9.43,
      "grad_norm": 3.8513760566711426,
      "learning_rate": 5.878136200716846e-06,
      "loss": 0.228,
      "step": 2630
    },
    {
      "epoch": 9.46,
      "grad_norm": 6.585672378540039,
      "learning_rate": 5.5197132616487455e-06,
      "loss": 0.1751,
      "step": 2640
    },
    {
      "epoch": 9.5,
      "grad_norm": 0.19121739268302917,
      "learning_rate": 5.161290322580646e-06,
      "loss": 0.2495,
      "step": 2650
    },
    {
      "epoch": 9.53,
      "grad_norm": 9.971940994262695,
      "learning_rate": 4.802867383512545e-06,
      "loss": 0.3073,
      "step": 2660
    },
    {
      "epoch": 9.57,
      "grad_norm": 0.3317970037460327,
      "learning_rate": 4.444444444444445e-06,
      "loss": 0.1477,
      "step": 2670
    },
    {
      "epoch": 9.61,
      "grad_norm": 4.959763526916504,
      "learning_rate": 4.086021505376344e-06,
      "loss": 0.2397,
      "step": 2680
    },
    {
      "epoch": 9.64,
      "grad_norm": 1.1729767322540283,
      "learning_rate": 3.7275985663082444e-06,
      "loss": 0.2094,
      "step": 2690
    },
    {
      "epoch": 9.68,
      "grad_norm": 5.1204118728637695,
      "learning_rate": 3.3691756272401432e-06,
      "loss": 0.1468,
      "step": 2700
    },
    {
      "epoch": 9.68,
      "eval_accuracy": 0.8680981595092024,
      "eval_loss": 0.4764532148838043,
      "eval_runtime": 14.3404,
      "eval_samples_per_second": 68.199,
      "eval_steps_per_second": 8.577,
      "step": 2700
    },
    {
      "epoch": 9.71,
      "grad_norm": 1.941149115562439,
      "learning_rate": 3.0107526881720433e-06,
      "loss": 0.1418,
      "step": 2710
    },
    {
      "epoch": 9.75,
      "grad_norm": 1.0805838108062744,
      "learning_rate": 2.6523297491039426e-06,
      "loss": 0.1762,
      "step": 2720
    },
    {
      "epoch": 9.78,
      "grad_norm": 0.28982478380203247,
      "learning_rate": 2.2939068100358423e-06,
      "loss": 0.1462,
      "step": 2730
    },
    {
      "epoch": 9.82,
      "grad_norm": 0.11841225624084473,
      "learning_rate": 1.935483870967742e-06,
      "loss": 0.1495,
      "step": 2740
    },
    {
      "epoch": 9.86,
      "grad_norm": 4.167394161224365,
      "learning_rate": 1.577060931899642e-06,
      "loss": 0.2596,
      "step": 2750
    },
    {
      "epoch": 9.89,
      "grad_norm": 4.835020542144775,
      "learning_rate": 1.2186379928315414e-06,
      "loss": 0.1955,
      "step": 2760
    },
    {
      "epoch": 9.93,
      "grad_norm": 7.901321887969971,
      "learning_rate": 8.602150537634409e-07,
      "loss": 0.2867,
      "step": 2770
    },
    {
      "epoch": 9.96,
      "grad_norm": 8.663835525512695,
      "learning_rate": 5.017921146953406e-07,
      "loss": 0.1797,
      "step": 2780
    },
    {
      "epoch": 10.0,
      "grad_norm": 13.646035194396973,
      "learning_rate": 1.4336917562724014e-07,
      "loss": 0.2222,
      "step": 2790
    },
    {
      "epoch": 10.0,
      "step": 2790,
      "total_flos": 3.451740694569861e+18,
      "train_loss": 0.44341087110580935,
      "train_runtime": 1600.5784,
      "train_samples_per_second": 27.827,
      "train_steps_per_second": 1.743
    }
  ],
  "logging_steps": 10,
  "max_steps": 2790,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 100,
  "total_flos": 3.451740694569861e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}