{
  "best_metric": 0.9935,
  "best_model_checkpoint": "beit-base-patch16-224-pt22k-ft22k-finetuned-mnist/checkpoint-2811",
  "epoch": 2.9994666666666667,
  "global_step": 2811,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 1.7730496453900712e-06,
      "loss": 2.3843,
      "step": 10
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.5460992907801423e-06,
      "loss": 2.27,
      "step": 20
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.319148936170213e-06,
      "loss": 2.1311,
      "step": 30
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.092198581560285e-06,
      "loss": 1.9981,
      "step": 40
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.865248226950355e-06,
      "loss": 1.761,
      "step": 50
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.0638297872340426e-05,
      "loss": 1.4946,
      "step": 60
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.2411347517730498e-05,
      "loss": 1.2439,
      "step": 70
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.418439716312057e-05,
      "loss": 1.0823,
      "step": 80
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.595744680851064e-05,
      "loss": 0.9241,
      "step": 90
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.773049645390071e-05,
      "loss": 0.8283,
      "step": 100
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.950354609929078e-05,
      "loss": 0.8041,
      "step": 110
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.1276595744680852e-05,
      "loss": 0.6482,
      "step": 120
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.3049645390070924e-05,
      "loss": 0.6448,
      "step": 130
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.4822695035460995e-05,
      "loss": 0.5983,
      "step": 140
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.6595744680851064e-05,
      "loss": 0.5754,
      "step": 150
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.836879432624114e-05,
      "loss": 0.5485,
      "step": 160
    },
    {
      "epoch": 0.18,
      "learning_rate": 3.0141843971631207e-05,
      "loss": 0.5681,
      "step": 170
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.191489361702128e-05,
      "loss": 0.48,
      "step": 180
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.3687943262411347e-05,
      "loss": 0.5461,
      "step": 190
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.546099290780142e-05,
      "loss": 0.5761,
      "step": 200
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.723404255319149e-05,
      "loss": 0.4957,
      "step": 210
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.900709219858156e-05,
      "loss": 0.4965,
      "step": 220
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.078014184397163e-05,
      "loss": 0.473,
      "step": 230
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.2553191489361704e-05,
      "loss": 0.4948,
      "step": 240
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.432624113475177e-05,
      "loss": 0.5289,
      "step": 250
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.609929078014185e-05,
      "loss": 0.4895,
      "step": 260
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.787234042553192e-05,
      "loss": 0.5204,
      "step": 270
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.964539007092199e-05,
      "loss": 0.5589,
      "step": 280
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.984183471727956e-05,
      "loss": 0.4661,
      "step": 290
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.964412811387901e-05,
      "loss": 0.4783,
      "step": 300
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.944642151047845e-05,
      "loss": 0.5219,
      "step": 310
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.92487149070779e-05,
      "loss": 0.5128,
      "step": 320
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.905100830367734e-05,
      "loss": 0.4896,
      "step": 330
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.8853301700276796e-05,
      "loss": 0.4123,
      "step": 340
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.865559509687624e-05,
      "loss": 0.4216,
      "step": 350
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.8457888493475686e-05,
      "loss": 0.4611,
      "step": 360
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.826018189007513e-05,
      "loss": 0.469,
      "step": 370
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.806247528667458e-05,
      "loss": 0.4072,
      "step": 380
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.786476868327402e-05,
      "loss": 0.4476,
      "step": 390
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.766706207987347e-05,
      "loss": 0.4469,
      "step": 400
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.746935547647291e-05,
      "loss": 0.4529,
      "step": 410
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.727164887307236e-05,
      "loss": 0.4699,
      "step": 420
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.707394226967181e-05,
      "loss": 0.4249,
      "step": 430
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.687623566627126e-05,
      "loss": 0.439,
      "step": 440
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.66785290628707e-05,
      "loss": 0.4137,
      "step": 450
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.6480822459470146e-05,
      "loss": 0.5142,
      "step": 460
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.6283115856069595e-05,
      "loss": 0.4532,
      "step": 470
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.608540925266904e-05,
      "loss": 0.4514,
      "step": 480
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.588770264926849e-05,
      "loss": 0.4166,
      "step": 490
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.568999604586793e-05,
      "loss": 0.4584,
      "step": 500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.549228944246738e-05,
      "loss": 0.3461,
      "step": 510
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.529458283906683e-05,
      "loss": 0.4783,
      "step": 520
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.5096876235666277e-05,
      "loss": 0.4474,
      "step": 530
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.489916963226572e-05,
      "loss": 0.4403,
      "step": 540
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.4701463028865166e-05,
      "loss": 0.3663,
      "step": 550
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.450375642546461e-05,
      "loss": 0.3791,
      "step": 560
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.430604982206406e-05,
      "loss": 0.4573,
      "step": 570
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.4108343218663504e-05,
      "loss": 0.3962,
      "step": 580
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.391063661526295e-05,
      "loss": 0.3705,
      "step": 590
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.371293001186239e-05,
      "loss": 0.4082,
      "step": 600
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.351522340846185e-05,
      "loss": 0.4173,
      "step": 610
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.331751680506129e-05,
      "loss": 0.4122,
      "step": 620
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.311981020166074e-05,
      "loss": 0.4279,
      "step": 630
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.2922103598260186e-05,
      "loss": 0.3369,
      "step": 640
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.272439699485963e-05,
      "loss": 0.3995,
      "step": 650
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.252669039145908e-05,
      "loss": 0.3735,
      "step": 660
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.232898378805852e-05,
      "loss": 0.356,
      "step": 670
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.213127718465797e-05,
      "loss": 0.3741,
      "step": 680
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.193357058125741e-05,
      "loss": 0.4291,
      "step": 690
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.173586397785687e-05,
      "loss": 0.4473,
      "step": 700
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.153815737445631e-05,
      "loss": 0.3914,
      "step": 710
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.134045077105576e-05,
      "loss": 0.3241,
      "step": 720
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.11427441676552e-05,
      "loss": 0.3986,
      "step": 730
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.0945037564254647e-05,
      "loss": 0.4287,
      "step": 740
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.0747330960854095e-05,
      "loss": 0.4011,
      "step": 750
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.054962435745354e-05,
      "loss": 0.388,
      "step": 760
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.0351917754052984e-05,
      "loss": 0.3904,
      "step": 770
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.015421115065243e-05,
      "loss": 0.3528,
      "step": 780
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.995650454725188e-05,
      "loss": 0.3805,
      "step": 790
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.975879794385133e-05,
      "loss": 0.3698,
      "step": 800
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.956109134045078e-05,
      "loss": 0.3454,
      "step": 810
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.936338473705022e-05,
      "loss": 0.4112,
      "step": 820
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.9165678133649666e-05,
      "loss": 0.344,
      "step": 830
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.8967971530249114e-05,
      "loss": 0.3694,
      "step": 840
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.877026492684856e-05,
      "loss": 0.3867,
      "step": 850
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.8572558323448004e-05,
      "loss": 0.3531,
      "step": 860
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.837485172004745e-05,
      "loss": 0.397,
      "step": 870
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.817714511664689e-05,
      "loss": 0.3702,
      "step": 880
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.797943851324635e-05,
      "loss": 0.3751,
      "step": 890
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.778173190984579e-05,
      "loss": 0.3294,
      "step": 900
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.758402530644524e-05,
      "loss": 0.3028,
      "step": 910
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.738631870304468e-05,
      "loss": 0.4089,
      "step": 920
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.7188612099644134e-05,
      "loss": 0.3376,
      "step": 930
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9855,
      "eval_loss": 0.044587597250938416,
      "eval_runtime": 131.1221,
      "eval_samples_per_second": 76.265,
      "eval_steps_per_second": 4.767,
      "step": 937
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.6990905496243575e-05,
      "loss": 0.4091,
      "step": 940
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.679319889284302e-05,
      "loss": 0.3623,
      "step": 950
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.6595492289442465e-05,
      "loss": 0.3532,
      "step": 960
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.639778568604191e-05,
      "loss": 0.3457,
      "step": 970
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.620007908264136e-05,
      "loss": 0.3359,
      "step": 980
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.600237247924081e-05,
      "loss": 0.3162,
      "step": 990
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.580466587584026e-05,
      "loss": 0.3081,
      "step": 1000
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.56069592724397e-05,
      "loss": 0.4274,
      "step": 1010
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.540925266903915e-05,
      "loss": 0.3806,
      "step": 1020
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.5211546065638595e-05,
      "loss": 0.4025,
      "step": 1030
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.501383946223804e-05,
      "loss": 0.3973,
      "step": 1040
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.4816132858837484e-05,
      "loss": 0.3603,
      "step": 1050
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.461842625543693e-05,
      "loss": 0.3805,
      "step": 1060
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.442071965203638e-05,
      "loss": 0.3542,
      "step": 1070
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.422301304863583e-05,
      "loss": 0.3334,
      "step": 1080
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.402530644523527e-05,
      "loss": 0.3736,
      "step": 1090
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.382759984183472e-05,
      "loss": 0.3524,
      "step": 1100
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.3629893238434166e-05,
      "loss": 0.4211,
      "step": 1110
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.3432186635033614e-05,
      "loss": 0.3224,
      "step": 1120
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.3234480031633056e-05,
      "loss": 0.3446,
      "step": 1130
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.3036773428232504e-05,
      "loss": 0.3062,
      "step": 1140
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.283906682483195e-05,
      "loss": 0.3114,
      "step": 1150
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.26413602214314e-05,
      "loss": 0.3816,
      "step": 1160
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.244365361803085e-05,
      "loss": 0.3216,
      "step": 1170
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.224594701463029e-05,
      "loss": 0.3618,
      "step": 1180
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.204824041122974e-05,
      "loss": 0.3515,
      "step": 1190
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.185053380782918e-05,
      "loss": 0.3023,
      "step": 1200
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.1652827204428634e-05,
      "loss": 0.3348,
      "step": 1210
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.1455120601028075e-05,
      "loss": 0.3047,
      "step": 1220
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.125741399762752e-05,
      "loss": 0.4023,
      "step": 1230
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.1059707394226965e-05,
      "loss": 0.3379,
      "step": 1240
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.086200079082642e-05,
      "loss": 0.2975,
      "step": 1250
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.066429418742586e-05,
      "loss": 0.3445,
      "step": 1260
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.046658758402531e-05,
      "loss": 0.3401,
      "step": 1270
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.0268880980624754e-05,
      "loss": 0.2998,
      "step": 1280
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.00711743772242e-05,
      "loss": 0.289,
      "step": 1290
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.987346777382365e-05,
      "loss": 0.3816,
      "step": 1300
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.9675761170423095e-05,
      "loss": 0.3213,
      "step": 1310
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.947805456702254e-05,
      "loss": 0.3604,
      "step": 1320
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.9280347963621984e-05,
      "loss": 0.2953,
      "step": 1330
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.9082641360221436e-05,
      "loss": 0.3474,
      "step": 1340
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.888493475682088e-05,
      "loss": 0.3291,
      "step": 1350
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.8687228153420325e-05,
      "loss": 0.3229,
      "step": 1360
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.848952155001977e-05,
      "loss": 0.3118,
      "step": 1370
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.8291814946619215e-05,
      "loss": 0.2957,
      "step": 1380
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.8094108343218666e-05,
      "loss": 0.3061,
      "step": 1390
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.789640173981811e-05,
      "loss": 0.3489,
      "step": 1400
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.7698695136417556e-05,
      "loss": 0.3754,
      "step": 1410
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.7500988533017004e-05,
      "loss": 0.3633,
      "step": 1420
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.7303281929616452e-05,
      "loss": 0.3709,
      "step": 1430
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.71055753262159e-05,
      "loss": 0.3616,
      "step": 1440
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.6907868722815345e-05,
      "loss": 0.3712,
      "step": 1450
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.671016211941479e-05,
      "loss": 0.2578,
      "step": 1460
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.6512455516014234e-05,
      "loss": 0.3222,
      "step": 1470
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.6314748912613686e-05,
      "loss": 0.323,
      "step": 1480
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.611704230921313e-05,
      "loss": 0.3349,
      "step": 1490
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.5919335705812575e-05,
      "loss": 0.3152,
      "step": 1500
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.572162910241202e-05,
      "loss": 0.3529,
      "step": 1510
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.5523922499011465e-05,
      "loss": 0.2843,
      "step": 1520
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.5326215895610916e-05,
      "loss": 0.3035,
      "step": 1530
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.512850929221036e-05,
      "loss": 0.3331,
      "step": 1540
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.4930802688809806e-05,
      "loss": 0.4147,
      "step": 1550
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.4733096085409254e-05,
      "loss": 0.2968,
      "step": 1560
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.45353894820087e-05,
      "loss": 0.3199,
      "step": 1570
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.4337682878608147e-05,
      "loss": 0.3143,
      "step": 1580
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.413997627520759e-05,
      "loss": 0.3115,
      "step": 1590
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.394226967180704e-05,
      "loss": 0.3396,
      "step": 1600
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.3744563068406488e-05,
      "loss": 0.375,
      "step": 1610
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.3546856465005932e-05,
      "loss": 0.3709,
      "step": 1620
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.334914986160538e-05,
      "loss": 0.2714,
      "step": 1630
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.3151443258204825e-05,
      "loss": 0.3532,
      "step": 1640
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.2953736654804273e-05,
      "loss": 0.3799,
      "step": 1650
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.2756030051403718e-05,
      "loss": 0.3241,
      "step": 1660
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.2558323448003166e-05,
      "loss": 0.368,
      "step": 1670
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.236061684460261e-05,
      "loss": 0.3099,
      "step": 1680
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.2162910241202056e-05,
      "loss": 0.3517,
      "step": 1690
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.1965203637801504e-05,
      "loss": 0.2803,
      "step": 1700
    },
    {
      "epoch": 1.82,
      "learning_rate": 2.176749703440095e-05,
      "loss": 0.3505,
      "step": 1710
    },
    {
      "epoch": 1.84,
      "learning_rate": 2.1569790431000397e-05,
      "loss": 0.3268,
      "step": 1720
    },
    {
      "epoch": 1.85,
      "learning_rate": 2.137208382759984e-05,
      "loss": 0.2774,
      "step": 1730
    },
    {
      "epoch": 1.86,
      "learning_rate": 2.117437722419929e-05,
      "loss": 0.3158,
      "step": 1740
    },
    {
      "epoch": 1.87,
      "learning_rate": 2.0976670620798734e-05,
      "loss": 0.3687,
      "step": 1750
    },
    {
      "epoch": 1.88,
      "learning_rate": 2.0778964017398182e-05,
      "loss": 0.3239,
      "step": 1760
    },
    {
      "epoch": 1.89,
      "learning_rate": 2.058125741399763e-05,
      "loss": 0.3351,
      "step": 1770
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.0383550810597075e-05,
      "loss": 0.341,
      "step": 1780
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.0185844207196523e-05,
      "loss": 0.2696,
      "step": 1790
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.9988137603795968e-05,
      "loss": 0.3837,
      "step": 1800
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.9790431000395416e-05,
      "loss": 0.3441,
      "step": 1810
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.959272439699486e-05,
      "loss": 0.258,
      "step": 1820
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.9395017793594306e-05,
      "loss": 0.3381,
      "step": 1830
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.9197311190193754e-05,
      "loss": 0.2887,
      "step": 1840
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.89996045867932e-05,
      "loss": 0.3031,
      "step": 1850
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.8801897983392647e-05,
      "loss": 0.3168,
      "step": 1860
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.860419137999209e-05,
      "loss": 0.318,
      "step": 1870
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9916,
      "eval_loss": 0.026178531348705292,
      "eval_runtime": 131.1473,
      "eval_samples_per_second": 76.25,
      "eval_steps_per_second": 4.766,
      "step": 1874
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.840648477659154e-05,
      "loss": 0.3821,
      "step": 1880
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.8208778173190984e-05,
      "loss": 0.3132,
      "step": 1890
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.8011071569790432e-05,
      "loss": 0.3423,
      "step": 1900
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.7813364966389877e-05,
      "loss": 0.3407,
      "step": 1910
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.7615658362989322e-05,
      "loss": 0.2871,
      "step": 1920
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.741795175958877e-05,
      "loss": 0.2987,
      "step": 1930
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.7220245156188218e-05,
      "loss": 0.3316,
      "step": 1940
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.7022538552787666e-05,
      "loss": 0.263,
      "step": 1950
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.682483194938711e-05,
      "loss": 0.2606,
      "step": 1960
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.662712534598656e-05,
      "loss": 0.2799,
      "step": 1970
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.6429418742586004e-05,
      "loss": 0.3024,
      "step": 1980
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.623171213918545e-05,
      "loss": 0.2587,
      "step": 1990
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.6034005535784897e-05,
      "loss": 0.2674,
      "step": 2000
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.583629893238434e-05,
      "loss": 0.292,
      "step": 2010
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.563859232898379e-05,
      "loss": 0.2998,
      "step": 2020
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.5440885725583234e-05,
      "loss": 0.2867,
      "step": 2030
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.5243179122182682e-05,
      "loss": 0.2567,
      "step": 2040
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.5045472518782127e-05,
      "loss": 0.3127,
      "step": 2050
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.4847765915381575e-05,
      "loss": 0.3159,
      "step": 2060
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.4650059311981022e-05,
      "loss": 0.2666,
      "step": 2070
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.4452352708580466e-05,
      "loss": 0.3152,
      "step": 2080
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.4254646105179915e-05,
      "loss": 0.2772,
      "step": 2090
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.405693950177936e-05,
      "loss": 0.3362,
      "step": 2100
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.3859232898378807e-05,
      "loss": 0.2909,
      "step": 2110
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.3661526294978252e-05,
      "loss": 0.3371,
      "step": 2120
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.34638196915777e-05,
      "loss": 0.3314,
      "step": 2130
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3266113088177145e-05,
      "loss": 0.3195,
      "step": 2140
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3068406484776591e-05,
      "loss": 0.2603,
      "step": 2150
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.287069988137604e-05,
      "loss": 0.2797,
      "step": 2160
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2672993277975484e-05,
      "loss": 0.2906,
      "step": 2170
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.2475286674574932e-05,
      "loss": 0.342,
      "step": 2180
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.2277580071174377e-05,
      "loss": 0.3176,
      "step": 2190
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.2079873467773824e-05,
      "loss": 0.271,
      "step": 2200
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.188216686437327e-05,
      "loss": 0.3251,
      "step": 2210
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.1684460260972716e-05,
      "loss": 0.2421,
      "step": 2220
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.1486753657572163e-05,
      "loss": 0.3071,
      "step": 2230
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.128904705417161e-05,
      "loss": 0.3017,
      "step": 2240
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.1091340450771057e-05,
      "loss": 0.3107,
      "step": 2250
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.0893633847370504e-05,
      "loss": 0.2682,
      "step": 2260
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.0695927243969949e-05,
      "loss": 0.3018,
      "step": 2270
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.0498220640569395e-05,
      "loss": 0.3179,
      "step": 2280
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.0300514037168841e-05,
      "loss": 0.3172,
      "step": 2290
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.0102807433768288e-05,
      "loss": 0.3421,
      "step": 2300
    },
    {
      "epoch": 2.47,
      "learning_rate": 9.905100830367734e-06,
      "loss": 0.2955,
      "step": 2310
    },
    {
      "epoch": 2.48,
      "learning_rate": 9.70739422696718e-06,
      "loss": 0.3146,
      "step": 2320
    },
    {
      "epoch": 2.49,
      "learning_rate": 9.509687623566627e-06,
      "loss": 0.2883,
      "step": 2330
    },
    {
      "epoch": 2.5,
      "learning_rate": 9.311981020166075e-06,
      "loss": 0.3252,
      "step": 2340
    },
    {
      "epoch": 2.51,
      "learning_rate": 9.11427441676552e-06,
      "loss": 0.3297,
      "step": 2350
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.916567813364967e-06,
      "loss": 0.3002,
      "step": 2360
    },
    {
      "epoch": 2.53,
      "learning_rate": 8.718861209964413e-06,
      "loss": 0.2354,
      "step": 2370
    },
    {
      "epoch": 2.54,
      "learning_rate": 8.52115460656386e-06,
      "loss": 0.2808,
      "step": 2380
    },
    {
      "epoch": 2.55,
      "learning_rate": 8.323448003163306e-06,
      "loss": 0.3158,
      "step": 2390
    },
    {
      "epoch": 2.56,
      "learning_rate": 8.125741399762752e-06,
      "loss": 0.3408,
      "step": 2400
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.928034796362199e-06,
      "loss": 0.3253,
      "step": 2410
    },
    {
      "epoch": 2.58,
      "learning_rate": 7.730328192961645e-06,
      "loss": 0.2898,
      "step": 2420
    },
    {
      "epoch": 2.59,
      "learning_rate": 7.532621589561091e-06,
      "loss": 0.3144,
      "step": 2430
    },
    {
      "epoch": 2.6,
      "learning_rate": 7.334914986160538e-06,
      "loss": 0.3143,
      "step": 2440
    },
    {
      "epoch": 2.61,
      "learning_rate": 7.137208382759984e-06,
      "loss": 0.2758,
      "step": 2450
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.939501779359431e-06,
      "loss": 0.2996,
      "step": 2460
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.741795175958877e-06,
      "loss": 0.3002,
      "step": 2470
    },
    {
      "epoch": 2.65,
      "learning_rate": 6.544088572558324e-06,
      "loss": 0.3579,
      "step": 2480
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.346381969157771e-06,
      "loss": 0.2395,
      "step": 2490
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.1486753657572165e-06,
      "loss": 0.242,
      "step": 2500
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.950968762356663e-06,
      "loss": 0.2256,
      "step": 2510
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.753262158956109e-06,
      "loss": 0.2807,
      "step": 2520
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.555555555555556e-06,
      "loss": 0.3264,
      "step": 2530
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.357848952155002e-06,
      "loss": 0.323,
      "step": 2540
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.160142348754449e-06,
      "loss": 0.2416,
      "step": 2550
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.962435745353895e-06,
      "loss": 0.2796,
      "step": 2560
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.7647291419533415e-06,
      "loss": 0.2513,
      "step": 2570
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.567022538552788e-06,
      "loss": 0.2681,
      "step": 2580
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.369315935152234e-06,
      "loss": 0.3121,
      "step": 2590
    },
    {
      "epoch": 2.77,
      "learning_rate": 4.171609331751681e-06,
      "loss": 0.2579,
      "step": 2600
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.973902728351126e-06,
      "loss": 0.2899,
      "step": 2610
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.7761961249505733e-06,
      "loss": 0.2676,
      "step": 2620
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.57848952155002e-06,
      "loss": 0.2652,
      "step": 2630
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.3807829181494666e-06,
      "loss": 0.2804,
      "step": 2640
    },
    {
      "epoch": 2.83,
      "learning_rate": 3.1830763147489126e-06,
      "loss": 0.2689,
      "step": 2650
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.985369711348359e-06,
      "loss": 0.2918,
      "step": 2660
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.787663107947806e-06,
      "loss": 0.2602,
      "step": 2670
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.589956504547252e-06,
      "loss": 0.2595,
      "step": 2680
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.3922499011466983e-06,
      "loss": 0.268,
      "step": 2690
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.1945432977461447e-06,
      "loss": 0.238,
      "step": 2700
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.996836694345591e-06,
      "loss": 0.2787,
      "step": 2710
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.7991300909450376e-06,
      "loss": 0.2723,
      "step": 2720
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.6014234875444842e-06,
      "loss": 0.2974,
      "step": 2730
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.4037168841439304e-06,
      "loss": 0.2928,
      "step": 2740
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.2060102807433769e-06,
      "loss": 0.2526,
      "step": 2750
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.0083036773428233e-06,
      "loss": 0.2957,
      "step": 2760
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.105970739422697e-07,
      "loss": 0.2813,
      "step": 2770
    },
    {
      "epoch": 2.97,
      "learning_rate": 6.12890470541716e-07,
      "loss": 0.3048,
      "step": 2780
    },
    {
      "epoch": 2.98,
      "learning_rate": 4.151838671411625e-07,
      "loss": 0.2735,
      "step": 2790
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.1747726374060896e-07,
      "loss": 0.2681,
      "step": 2800
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.977066034005536e-08,
      "loss": 0.2374,
      "step": 2810
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9935,
      "eval_loss": 0.02018115483224392,
      "eval_runtime": 131.0943,
      "eval_samples_per_second": 76.281,
      "eval_steps_per_second": 4.768,
      "step": 2811
    },
    {
      "epoch": 3.0,
      "step": 2811,
      "total_flos": 1.3941117625304089e+19,
      "train_loss": 0.3983713053842746,
      "train_runtime": 6676.1657,
      "train_samples_per_second": 26.962,
      "train_steps_per_second": 0.421
    }
  ],
  "max_steps": 2811,
  "num_train_epochs": 3,
  "total_flos": 1.3941117625304089e+19,
  "trial_name": null,
  "trial_params": null
}