{
  "best_metric": 0.5103544592857361,
  "best_model_checkpoint": "../../experiments_checkpoints/MAdAiLab/microsoft/phi_2_twitter/checkpoint-750",
  "epoch": 3.0,
  "eval_steps": 50,
  "global_step": 816,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "grad_norm": 164.05126953125,
      "learning_rate": 4.938725490196079e-06,
      "loss": 1.2318,
      "step": 10
    },
    {
      "epoch": 0.07,
      "grad_norm": 104.96498107910156,
      "learning_rate": 4.8774509803921576e-06,
      "loss": 1.0923,
      "step": 20
    },
    {
      "epoch": 0.11,
      "grad_norm": 83.26679229736328,
      "learning_rate": 4.816176470588236e-06,
      "loss": 0.7558,
      "step": 30
    },
    {
      "epoch": 0.15,
      "grad_norm": 51.119388580322266,
      "learning_rate": 4.754901960784314e-06,
      "loss": 0.6983,
      "step": 40
    },
    {
      "epoch": 0.18,
      "grad_norm": 38.16554260253906,
      "learning_rate": 4.693627450980393e-06,
      "loss": 0.5921,
      "step": 50
    },
    {
      "epoch": 0.18,
      "eval_accuracy": 0.6957720588235294,
      "eval_f1_macro": 0.5053995294725175,
      "eval_f1_micro": 0.6957720588235294,
      "eval_loss": 0.6037454009056091,
      "eval_runtime": 5.1072,
      "eval_samples_per_second": 213.033,
      "eval_steps_per_second": 6.657,
      "step": 50
    },
    {
      "epoch": 0.22,
      "grad_norm": 168.71607971191406,
      "learning_rate": 4.632352941176471e-06,
      "loss": 0.6117,
      "step": 60
    },
    {
      "epoch": 0.26,
      "grad_norm": 39.83993148803711,
      "learning_rate": 4.571078431372549e-06,
      "loss": 0.6139,
      "step": 70
    },
    {
      "epoch": 0.29,
      "grad_norm": 51.992225646972656,
      "learning_rate": 4.509803921568628e-06,
      "loss": 0.554,
      "step": 80
    },
    {
      "epoch": 0.33,
      "grad_norm": 40.58168029785156,
      "learning_rate": 4.448529411764706e-06,
      "loss": 0.5429,
      "step": 90
    },
    {
      "epoch": 0.37,
      "grad_norm": 77.49382019042969,
      "learning_rate": 4.3872549019607845e-06,
      "loss": 0.5904,
      "step": 100
    },
    {
      "epoch": 0.37,
      "eval_accuracy": 0.71875,
      "eval_f1_macro": 0.6277222408587723,
      "eval_f1_micro": 0.71875,
      "eval_loss": 0.5537109375,
      "eval_runtime": 5.1256,
      "eval_samples_per_second": 212.27,
      "eval_steps_per_second": 6.633,
      "step": 100
    },
    {
      "epoch": 0.4,
      "grad_norm": 55.94441604614258,
      "learning_rate": 4.3259803921568635e-06,
      "loss": 0.5684,
      "step": 110
    },
    {
      "epoch": 0.44,
      "grad_norm": 53.36739730834961,
      "learning_rate": 4.264705882352942e-06,
      "loss": 0.5263,
      "step": 120
    },
    {
      "epoch": 0.48,
      "grad_norm": 44.8289680480957,
      "learning_rate": 4.20343137254902e-06,
      "loss": 0.5239,
      "step": 130
    },
    {
      "epoch": 0.51,
      "grad_norm": 43.29486083984375,
      "learning_rate": 4.142156862745099e-06,
      "loss": 0.6123,
      "step": 140
    },
    {
      "epoch": 0.55,
      "grad_norm": 114.31592559814453,
      "learning_rate": 4.080882352941177e-06,
      "loss": 0.5021,
      "step": 150
    },
    {
      "epoch": 0.55,
      "eval_accuracy": 0.7159926470588235,
      "eval_f1_macro": 0.5774438641805396,
      "eval_f1_micro": 0.7159926470588235,
      "eval_loss": 0.6040900945663452,
      "eval_runtime": 5.1289,
      "eval_samples_per_second": 212.13,
      "eval_steps_per_second": 6.629,
      "step": 150
    },
    {
      "epoch": 0.59,
      "grad_norm": 99.46711730957031,
      "learning_rate": 4.019607843137255e-06,
      "loss": 0.5639,
      "step": 160
    },
    {
      "epoch": 0.62,
      "grad_norm": 90.892578125,
      "learning_rate": 3.958333333333333e-06,
      "loss": 0.5736,
      "step": 170
    },
    {
      "epoch": 0.66,
      "grad_norm": 70.59735107421875,
      "learning_rate": 3.897058823529412e-06,
      "loss": 0.5759,
      "step": 180
    },
    {
      "epoch": 0.7,
      "grad_norm": 61.23074722290039,
      "learning_rate": 3.8357843137254904e-06,
      "loss": 0.5252,
      "step": 190
    },
    {
      "epoch": 0.74,
      "grad_norm": 48.55831527709961,
      "learning_rate": 3.774509803921569e-06,
      "loss": 0.5266,
      "step": 200
    },
    {
      "epoch": 0.74,
      "eval_accuracy": 0.7095588235294118,
      "eval_f1_macro": 0.6496202779329097,
      "eval_f1_micro": 0.7095588235294118,
      "eval_loss": 0.5544002652168274,
      "eval_runtime": 5.139,
      "eval_samples_per_second": 211.713,
      "eval_steps_per_second": 6.616,
      "step": 200
    },
    {
      "epoch": 0.77,
      "grad_norm": 40.87614440917969,
      "learning_rate": 3.7132352941176476e-06,
      "loss": 0.5181,
      "step": 210
    },
    {
      "epoch": 0.81,
      "grad_norm": 42.869300842285156,
      "learning_rate": 3.6519607843137257e-06,
      "loss": 0.5433,
      "step": 220
    },
    {
      "epoch": 0.85,
      "grad_norm": 37.196022033691406,
      "learning_rate": 3.5906862745098043e-06,
      "loss": 0.5088,
      "step": 230
    },
    {
      "epoch": 0.88,
      "grad_norm": 26.8739013671875,
      "learning_rate": 3.529411764705883e-06,
      "loss": 0.515,
      "step": 240
    },
    {
      "epoch": 0.92,
      "grad_norm": 74.13105010986328,
      "learning_rate": 3.468137254901961e-06,
      "loss": 0.5427,
      "step": 250
    },
    {
      "epoch": 0.92,
      "eval_accuracy": 0.7398897058823529,
      "eval_f1_macro": 0.6914737024578574,
      "eval_f1_micro": 0.7398897058823529,
      "eval_loss": 0.533088207244873,
      "eval_runtime": 5.1312,
      "eval_samples_per_second": 212.038,
      "eval_steps_per_second": 6.626,
      "step": 250
    },
    {
      "epoch": 0.96,
      "grad_norm": 59.509281158447266,
      "learning_rate": 3.4068627450980396e-06,
      "loss": 0.4859,
      "step": 260
    },
    {
      "epoch": 0.99,
      "grad_norm": 26.682819366455078,
      "learning_rate": 3.3455882352941178e-06,
      "loss": 0.5441,
      "step": 270
    },
    {
      "epoch": 1.03,
      "grad_norm": 60.43280792236328,
      "learning_rate": 3.2843137254901964e-06,
      "loss": 0.487,
      "step": 280
    },
    {
      "epoch": 1.07,
      "grad_norm": 33.994571685791016,
      "learning_rate": 3.223039215686275e-06,
      "loss": 0.4848,
      "step": 290
    },
    {
      "epoch": 1.1,
      "grad_norm": 43.1735954284668,
      "learning_rate": 3.161764705882353e-06,
      "loss": 0.4715,
      "step": 300
    },
    {
      "epoch": 1.1,
      "eval_accuracy": 0.7398897058823529,
      "eval_f1_macro": 0.6361259916590146,
      "eval_f1_micro": 0.7398897058823529,
      "eval_loss": 0.5436149835586548,
      "eval_runtime": 5.1329,
      "eval_samples_per_second": 211.966,
      "eval_steps_per_second": 6.624,
      "step": 300
    },
    {
      "epoch": 1.14,
      "grad_norm": 43.79911804199219,
      "learning_rate": 3.1004901960784317e-06,
      "loss": 0.4838,
      "step": 310
    },
    {
      "epoch": 1.18,
      "grad_norm": 107.44807434082031,
      "learning_rate": 3.03921568627451e-06,
      "loss": 0.47,
      "step": 320
    },
    {
      "epoch": 1.21,
      "grad_norm": 33.9593620300293,
      "learning_rate": 2.9779411764705884e-06,
      "loss": 0.5337,
      "step": 330
    },
    {
      "epoch": 1.25,
      "grad_norm": 49.85417556762695,
      "learning_rate": 2.916666666666667e-06,
      "loss": 0.4914,
      "step": 340
    },
    {
      "epoch": 1.29,
      "grad_norm": 27.120018005371094,
      "learning_rate": 2.855392156862745e-06,
      "loss": 0.4829,
      "step": 350
    },
    {
      "epoch": 1.29,
      "eval_accuracy": 0.7564338235294118,
      "eval_f1_macro": 0.7135428309702008,
      "eval_f1_micro": 0.7564338235294118,
      "eval_loss": 0.5216710567474365,
      "eval_runtime": 5.1418,
      "eval_samples_per_second": 211.6,
      "eval_steps_per_second": 6.613,
      "step": 350
    },
    {
      "epoch": 1.32,
      "grad_norm": 42.45214080810547,
      "learning_rate": 2.7941176470588237e-06,
      "loss": 0.4851,
      "step": 360
    },
    {
      "epoch": 1.36,
      "grad_norm": 68.97784423828125,
      "learning_rate": 2.732843137254902e-06,
      "loss": 0.4166,
      "step": 370
    },
    {
      "epoch": 1.4,
      "grad_norm": 39.31782913208008,
      "learning_rate": 2.6715686274509804e-06,
      "loss": 0.5542,
      "step": 380
    },
    {
      "epoch": 1.43,
      "grad_norm": 46.57756805419922,
      "learning_rate": 2.610294117647059e-06,
      "loss": 0.4718,
      "step": 390
    },
    {
      "epoch": 1.47,
      "grad_norm": 70.58038330078125,
      "learning_rate": 2.549019607843137e-06,
      "loss": 0.4676,
      "step": 400
    },
    {
      "epoch": 1.47,
      "eval_accuracy": 0.7536764705882353,
      "eval_f1_macro": 0.6829050794520786,
      "eval_f1_micro": 0.7536764705882353,
      "eval_loss": 0.5224753022193909,
      "eval_runtime": 5.1324,
      "eval_samples_per_second": 211.985,
      "eval_steps_per_second": 6.625,
      "step": 400
    },
    {
      "epoch": 1.51,
      "grad_norm": 50.325191497802734,
      "learning_rate": 2.4877450980392158e-06,
      "loss": 0.5109,
      "step": 410
    },
    {
      "epoch": 1.54,
      "grad_norm": 74.2918701171875,
      "learning_rate": 2.4264705882352943e-06,
      "loss": 0.4521,
      "step": 420
    },
    {
      "epoch": 1.58,
      "grad_norm": 36.432159423828125,
      "learning_rate": 2.3651960784313725e-06,
      "loss": 0.44,
      "step": 430
    },
    {
      "epoch": 1.62,
      "grad_norm": 45.3189811706543,
      "learning_rate": 2.303921568627451e-06,
      "loss": 0.434,
      "step": 440
    },
    {
      "epoch": 1.65,
      "grad_norm": 38.40291213989258,
      "learning_rate": 2.2426470588235296e-06,
      "loss": 0.5196,
      "step": 450
    },
    {
      "epoch": 1.65,
      "eval_accuracy": 0.7628676470588235,
      "eval_f1_macro": 0.709592918596134,
      "eval_f1_micro": 0.7628676470588235,
      "eval_loss": 0.5163430571556091,
      "eval_runtime": 5.141,
      "eval_samples_per_second": 211.631,
      "eval_steps_per_second": 6.613,
      "step": 450
    },
    {
      "epoch": 1.69,
      "grad_norm": 34.863059997558594,
      "learning_rate": 2.1813725490196082e-06,
      "loss": 0.4809,
      "step": 460
    },
    {
      "epoch": 1.73,
      "grad_norm": 42.53833770751953,
      "learning_rate": 2.1200980392156864e-06,
      "loss": 0.4883,
      "step": 470
    },
    {
      "epoch": 1.76,
      "grad_norm": 34.529541015625,
      "learning_rate": 2.058823529411765e-06,
      "loss": 0.4098,
      "step": 480
    },
    {
      "epoch": 1.8,
      "grad_norm": 30.016141891479492,
      "learning_rate": 1.9975490196078435e-06,
      "loss": 0.4292,
      "step": 490
    },
    {
      "epoch": 1.84,
      "grad_norm": 39.967533111572266,
      "learning_rate": 1.9362745098039217e-06,
      "loss": 0.4815,
      "step": 500
    },
    {
      "epoch": 1.84,
      "eval_accuracy": 0.765625,
      "eval_f1_macro": 0.7215170814701055,
      "eval_f1_micro": 0.765625,
      "eval_loss": 0.5213120579719543,
      "eval_runtime": 5.1359,
      "eval_samples_per_second": 211.843,
      "eval_steps_per_second": 6.62,
      "step": 500
    },
    {
      "epoch": 1.88,
      "grad_norm": 43.9462776184082,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 0.5156,
      "step": 510
    },
    {
      "epoch": 1.91,
      "grad_norm": 26.374074935913086,
      "learning_rate": 1.8137254901960786e-06,
      "loss": 0.5271,
      "step": 520
    },
    {
      "epoch": 1.95,
      "grad_norm": 58.75647735595703,
      "learning_rate": 1.752450980392157e-06,
      "loss": 0.5166,
      "step": 530
    },
    {
      "epoch": 1.99,
      "grad_norm": 32.93243408203125,
      "learning_rate": 1.6911764705882356e-06,
      "loss": 0.4624,
      "step": 540
    },
    {
      "epoch": 2.02,
      "grad_norm": 39.54354476928711,
      "learning_rate": 1.629901960784314e-06,
      "loss": 0.4836,
      "step": 550
    },
    {
      "epoch": 2.02,
      "eval_accuracy": 0.7619485294117647,
      "eval_f1_macro": 0.7190841764536411,
      "eval_f1_micro": 0.7619485294117647,
      "eval_loss": 0.522087574005127,
      "eval_runtime": 5.1446,
      "eval_samples_per_second": 211.482,
      "eval_steps_per_second": 6.609,
      "step": 550
    },
    {
      "epoch": 2.06,
      "grad_norm": 27.993928909301758,
      "learning_rate": 1.5686274509803923e-06,
      "loss": 0.4759,
      "step": 560
    },
    {
      "epoch": 2.1,
      "grad_norm": 24.092933654785156,
      "learning_rate": 1.5073529411764707e-06,
      "loss": 0.4096,
      "step": 570
    },
    {
      "epoch": 2.13,
      "grad_norm": 26.48084831237793,
      "learning_rate": 1.4460784313725492e-06,
      "loss": 0.4636,
      "step": 580
    },
    {
      "epoch": 2.17,
      "grad_norm": 23.85141372680664,
      "learning_rate": 1.3848039215686276e-06,
      "loss": 0.4349,
      "step": 590
    },
    {
      "epoch": 2.21,
      "grad_norm": 28.8481502532959,
      "learning_rate": 1.323529411764706e-06,
      "loss": 0.4945,
      "step": 600
    },
    {
      "epoch": 2.21,
      "eval_accuracy": 0.7637867647058824,
      "eval_f1_macro": 0.7173491781255339,
      "eval_f1_micro": 0.7637867647058824,
      "eval_loss": 0.5133702754974365,
      "eval_runtime": 5.1414,
      "eval_samples_per_second": 211.615,
      "eval_steps_per_second": 6.613,
      "step": 600
    },
    {
      "epoch": 2.24,
      "grad_norm": 20.64706039428711,
      "learning_rate": 1.2622549019607843e-06,
      "loss": 0.4625,
      "step": 610
    },
    {
      "epoch": 2.28,
      "grad_norm": 29.8828067779541,
      "learning_rate": 1.200980392156863e-06,
      "loss": 0.4622,
      "step": 620
    },
    {
      "epoch": 2.32,
      "grad_norm": 30.85140037536621,
      "learning_rate": 1.1397058823529413e-06,
      "loss": 0.4505,
      "step": 630
    },
    {
      "epoch": 2.35,
      "grad_norm": 58.30624008178711,
      "learning_rate": 1.0784313725490197e-06,
      "loss": 0.4761,
      "step": 640
    },
    {
      "epoch": 2.39,
      "grad_norm": 32.4758186340332,
      "learning_rate": 1.017156862745098e-06,
      "loss": 0.4103,
      "step": 650
    },
    {
      "epoch": 2.39,
      "eval_accuracy": 0.7683823529411765,
      "eval_f1_macro": 0.72109375,
      "eval_f1_micro": 0.7683823529411765,
      "eval_loss": 0.5125373601913452,
      "eval_runtime": 5.1343,
      "eval_samples_per_second": 211.91,
      "eval_steps_per_second": 6.622,
      "step": 650
    },
    {
      "epoch": 2.43,
      "grad_norm": 20.597454071044922,
      "learning_rate": 9.558823529411764e-07,
      "loss": 0.4986,
      "step": 660
    },
    {
      "epoch": 2.46,
      "grad_norm": 41.76366424560547,
      "learning_rate": 8.94607843137255e-07,
      "loss": 0.4589,
      "step": 670
    },
    {
      "epoch": 2.5,
      "grad_norm": 18.32976531982422,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.4294,
      "step": 680
    },
    {
      "epoch": 2.54,
      "grad_norm": 25.26702117919922,
      "learning_rate": 7.720588235294119e-07,
      "loss": 0.4579,
      "step": 690
    },
    {
      "epoch": 2.57,
      "grad_norm": 30.386232376098633,
      "learning_rate": 7.107843137254903e-07,
      "loss": 0.4191,
      "step": 700
    },
    {
      "epoch": 2.57,
      "eval_accuracy": 0.7683823529411765,
      "eval_f1_macro": 0.7226006191950465,
      "eval_f1_micro": 0.7683823529411765,
      "eval_loss": 0.5108427405357361,
      "eval_runtime": 5.1417,
      "eval_samples_per_second": 211.604,
      "eval_steps_per_second": 6.613,
      "step": 700
    },
    {
      "epoch": 2.61,
      "grad_norm": 21.484479904174805,
      "learning_rate": 6.495098039215687e-07,
      "loss": 0.4528,
      "step": 710
    },
    {
      "epoch": 2.65,
      "grad_norm": 29.71825408935547,
      "learning_rate": 5.882352941176471e-07,
      "loss": 0.4237,
      "step": 720
    },
    {
      "epoch": 2.68,
      "grad_norm": 29.686370849609375,
      "learning_rate": 5.269607843137256e-07,
      "loss": 0.4671,
      "step": 730
    },
    {
      "epoch": 2.72,
      "grad_norm": 29.55408477783203,
      "learning_rate": 4.6568627450980395e-07,
      "loss": 0.484,
      "step": 740
    },
    {
      "epoch": 2.76,
      "grad_norm": 43.04338836669922,
      "learning_rate": 4.044117647058824e-07,
      "loss": 0.5004,
      "step": 750
    },
    {
      "epoch": 2.76,
      "eval_accuracy": 0.7647058823529411,
      "eval_f1_macro": 0.717181071212435,
      "eval_f1_micro": 0.7647058823529411,
      "eval_loss": 0.5103544592857361,
      "eval_runtime": 5.1424,
      "eval_samples_per_second": 211.576,
      "eval_steps_per_second": 6.612,
      "step": 750
    },
    {
      "epoch": 2.79,
      "grad_norm": 37.452056884765625,
      "learning_rate": 3.4313725490196084e-07,
      "loss": 0.3917,
      "step": 760
    },
    {
      "epoch": 2.83,
      "grad_norm": 31.18016815185547,
      "learning_rate": 2.8186274509803926e-07,
      "loss": 0.4226,
      "step": 770
    },
    {
      "epoch": 2.87,
      "grad_norm": 41.16299819946289,
      "learning_rate": 2.2058823529411768e-07,
      "loss": 0.4242,
      "step": 780
    },
    {
      "epoch": 2.9,
      "grad_norm": 18.450733184814453,
      "learning_rate": 1.5931372549019607e-07,
      "loss": 0.4482,
      "step": 790
    },
    {
      "epoch": 2.94,
      "grad_norm": 37.13800811767578,
      "learning_rate": 9.803921568627452e-08,
      "loss": 0.4398,
      "step": 800
    },
    {
      "epoch": 2.94,
      "eval_accuracy": 0.7674632352941176,
      "eval_f1_macro": 0.7138339312799731,
      "eval_f1_micro": 0.7674632352941176,
      "eval_loss": 0.511359691619873,
      "eval_runtime": 5.1453,
      "eval_samples_per_second": 211.454,
      "eval_steps_per_second": 6.608,
      "step": 800
    },
    {
      "epoch": 2.98,
      "grad_norm": 28.77461814880371,
      "learning_rate": 3.6764705882352945e-08,
      "loss": 0.4237,
      "step": 810
    },
    {
      "epoch": 3.0,
      "step": 816,
      "total_flos": 5.048586286084915e+16,
      "train_loss": 0.5124810910692402,
      "train_runtime": 1553.1677,
      "train_samples_per_second": 16.804,
      "train_steps_per_second": 0.525
    }
  ],
  "logging_steps": 10,
  "max_steps": 816,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 50,
  "total_flos": 5.048586286084915e+16,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}