{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 102000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.014705882352941176,
      "grad_norm": 3.0940372943878174,
      "learning_rate": 4.975490196078432e-05,
      "loss": 2.042,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.029411764705882353,
      "grad_norm": 3.2075328826904297,
      "learning_rate": 4.9509803921568634e-05,
      "loss": 1.5982,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 1000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.04411764705882353,
      "grad_norm": 4.371872425079346,
      "learning_rate": 4.9264705882352944e-05,
      "loss": 1.4943,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 1500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.058823529411764705,
      "grad_norm": 5.18107271194458,
      "learning_rate": 4.901960784313725e-05,
      "loss": 1.4003,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 2000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.07352941176470588,
      "grad_norm": 4.011467933654785,
      "learning_rate": 4.877450980392157e-05,
      "loss": 1.3578,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 2500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.08823529411764706,
      "grad_norm": 3.5927531719207764,
      "learning_rate": 4.8529411764705885e-05,
      "loss": 1.2995,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 3000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.10294117647058823,
      "grad_norm": 4.120026111602783,
      "learning_rate": 4.82843137254902e-05,
      "loss": 1.2598,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 3500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 6.63548469543457,
      "learning_rate": 4.803921568627452e-05,
      "loss": 1.2075,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 4000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.1323529411764706,
      "grad_norm": 4.13482666015625,
      "learning_rate": 4.7794117647058826e-05,
      "loss": 1.2405,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 4500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.14705882352941177,
      "grad_norm": 5.699426174163818,
      "learning_rate": 4.7549019607843135e-05,
      "loss": 1.2242,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 5000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.16176470588235295,
      "grad_norm": 6.220912933349609,
      "learning_rate": 4.730392156862745e-05,
      "loss": 1.1701,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 5500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.17647058823529413,
      "grad_norm": 5.078765869140625,
      "learning_rate": 4.705882352941177e-05,
      "loss": 1.1904,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 6000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.19117647058823528,
      "grad_norm": 1.5997480154037476,
      "learning_rate": 4.681372549019608e-05,
      "loss": 1.1321,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 6500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.20588235294117646,
      "grad_norm": 3.2184040546417236,
      "learning_rate": 4.656862745098039e-05,
      "loss": 1.1452,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 7000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.22058823529411764,
      "grad_norm": 5.408567905426025,
      "learning_rate": 4.632352941176471e-05,
      "loss": 1.0764,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 7500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 6.1005539894104,
      "learning_rate": 4.607843137254902e-05,
      "loss": 1.0756,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 8000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.336429595947266,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 1.084,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 8500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.2647058823529412,
      "grad_norm": 5.947171211242676,
      "learning_rate": 4.558823529411765e-05,
      "loss": 1.095,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 9000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.27941176470588236,
      "grad_norm": 3.9784250259399414,
      "learning_rate": 4.5343137254901966e-05,
      "loss": 1.0716,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 9500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.29411764705882354,
      "grad_norm": 7.9583420753479,
      "learning_rate": 4.5098039215686275e-05,
      "loss": 1.0725,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 10000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.3088235294117647,
      "grad_norm": 5.341268062591553,
      "learning_rate": 4.485294117647059e-05,
      "loss": 1.0656,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 10500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.3235294117647059,
      "grad_norm": 8.68230152130127,
      "learning_rate": 4.460784313725491e-05,
      "loss": 0.9894,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 11000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.3382352941176471,
      "grad_norm": 3.516399621963501,
      "learning_rate": 4.4362745098039216e-05,
      "loss": 1.0527,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 11500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 5.295083045959473,
      "learning_rate": 4.411764705882353e-05,
      "loss": 1.0073,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 12000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.36764705882352944,
      "grad_norm": 3.7748844623565674,
      "learning_rate": 4.387254901960784e-05,
      "loss": 1.0035,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 12500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.38235294117647056,
      "grad_norm": 7.610727310180664,
      "learning_rate": 4.362745098039216e-05,
      "loss": 0.9888,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 13000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.39705882352941174,
      "grad_norm": 5.661556243896484,
      "learning_rate": 4.3382352941176474e-05,
      "loss": 1.005,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 13500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.4117647058823529,
      "grad_norm": 7.081089496612549,
      "learning_rate": 4.313725490196079e-05,
      "loss": 0.9848,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 14000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.4264705882352941,
      "grad_norm": 3.7689192295074463,
      "learning_rate": 4.28921568627451e-05,
      "loss": 0.9745,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 14500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.4411764705882353,
      "grad_norm": 3.3003389835357666,
      "learning_rate": 4.2647058823529415e-05,
      "loss": 0.9684,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 15000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.45588235294117646,
      "grad_norm": 5.606069564819336,
      "learning_rate": 4.2401960784313724e-05,
      "loss": 0.9821,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 15500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 5.626067638397217,
      "learning_rate": 4.215686274509804e-05,
      "loss": 0.9537,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 16000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.4852941176470588,
      "grad_norm": 5.848409175872803,
      "learning_rate": 4.1911764705882356e-05,
      "loss": 0.9675,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 16500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.057644844055176,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.9714,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 17000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.5147058823529411,
      "grad_norm": 3.4304769039154053,
      "learning_rate": 4.142156862745099e-05,
      "loss": 0.9857,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 17500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.5294117647058824,
      "grad_norm": 3.0925090312957764,
      "learning_rate": 4.11764705882353e-05,
      "loss": 0.9716,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 18000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.5441176470588235,
      "grad_norm": 7.13212776184082,
      "learning_rate": 4.0931372549019607e-05,
      "loss": 0.9471,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 18500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.5588235294117647,
      "grad_norm": 3.437065362930298,
      "learning_rate": 4.068627450980392e-05,
      "loss": 0.9125,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 19000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.5735294117647058,
      "grad_norm": 3.033743143081665,
      "learning_rate": 4.044117647058824e-05,
      "loss": 0.9608,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 19500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 4.206902980804443,
      "learning_rate": 4.0196078431372555e-05,
      "loss": 0.9494,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 20000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.6029411764705882,
      "grad_norm": 4.862374305725098,
      "learning_rate": 3.9950980392156864e-05,
      "loss": 0.9636,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 20500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.6176470588235294,
      "grad_norm": 2.435167074203491,
      "learning_rate": 3.970588235294117e-05,
      "loss": 0.947,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 21000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.6323529411764706,
      "grad_norm": 3.3232932090759277,
      "learning_rate": 3.946078431372549e-05,
      "loss": 0.9104,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 21500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.6470588235294118,
      "grad_norm": 6.478214740753174,
      "learning_rate": 3.9215686274509805e-05,
      "loss": 0.9457,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 22000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.6617647058823529,
      "grad_norm": 4.218869209289551,
      "learning_rate": 3.897058823529412e-05,
      "loss": 0.9154,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 22500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.6764705882352942,
      "grad_norm": 3.7748098373413086,
      "learning_rate": 3.872549019607844e-05,
      "loss": 0.9,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 23000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.6911764705882353,
      "grad_norm": 4.164327621459961,
      "learning_rate": 3.8480392156862746e-05,
      "loss": 0.9025,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 23500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 4.087977886199951,
      "learning_rate": 3.8235294117647055e-05,
      "loss": 0.9147,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 24000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.7205882352941176,
      "grad_norm": 7.114366054534912,
      "learning_rate": 3.799019607843137e-05,
      "loss": 0.8656,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 24500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.7352941176470589,
      "grad_norm": 6.7599101066589355,
      "learning_rate": 3.774509803921569e-05,
      "loss": 0.8887,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 25000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.847716808319092,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.8807,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 25500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.7647058823529411,
      "grad_norm": 9.692086219787598,
      "learning_rate": 3.725490196078432e-05,
      "loss": 0.8615,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 26000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.7794117647058824,
      "grad_norm": 0.6128278970718384,
      "learning_rate": 3.700980392156863e-05,
      "loss": 0.8675,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 26500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.7941176470588235,
      "grad_norm": 8.303850173950195,
      "learning_rate": 3.6764705882352945e-05,
      "loss": 0.8809,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 27000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.8088235294117647,
      "grad_norm": 3.569289445877075,
      "learning_rate": 3.6519607843137254e-05,
      "loss": 0.8891,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 27500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.8235294117647058,
      "grad_norm": 0.8430338501930237,
      "learning_rate": 3.627450980392157e-05,
      "loss": 0.8613,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 28000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.8382352941176471,
      "grad_norm": 3.0527398586273193,
      "learning_rate": 3.6029411764705886e-05,
      "loss": 0.8622,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 28500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.8529411764705882,
      "grad_norm": 7.135374069213867,
      "learning_rate": 3.5784313725490195e-05,
      "loss": 0.8671,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 29000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.8676470588235294,
      "grad_norm": 3.676905393600464,
      "learning_rate": 3.553921568627451e-05,
      "loss": 0.8788,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 29500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.8823529411764706,
      "grad_norm": 7.306654453277588,
      "learning_rate": 3.529411764705883e-05,
      "loss": 0.8283,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 30000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.8970588235294118,
      "grad_norm": 6.978466510772705,
      "learning_rate": 3.5049019607843136e-05,
      "loss": 0.8755,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 30500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.9117647058823529,
      "grad_norm": 8.922977447509766,
      "learning_rate": 3.480392156862745e-05,
      "loss": 0.8398,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 31000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.9264705882352942,
      "grad_norm": 3.4424359798431396,
      "learning_rate": 3.455882352941177e-05,
      "loss": 0.8565,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 31500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 4.229028701782227,
      "learning_rate": 3.431372549019608e-05,
      "loss": 0.851,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 32000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.9558823529411765,
      "grad_norm": 12.309649467468262,
      "learning_rate": 3.4068627450980394e-05,
      "loss": 0.8563,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 32500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.9705882352941176,
      "grad_norm": 0.5380654335021973,
      "learning_rate": 3.382352941176471e-05,
      "loss": 0.8338,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 33000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 0.9852941176470589,
      "grad_norm": 4.774703025817871,
      "learning_rate": 3.357843137254902e-05,
      "loss": 0.8425,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 33500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.663310527801514,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.8711,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 34000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.0147058823529411,
      "grad_norm": 5.9944562911987305,
      "learning_rate": 3.308823529411765e-05,
      "loss": 0.8495,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 34500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.0294117647058822,
      "grad_norm": 4.68066930770874,
      "learning_rate": 3.284313725490196e-05,
      "loss": 0.7897,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 35000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.0441176470588236,
      "grad_norm": 1.811689853668213,
      "learning_rate": 3.2598039215686276e-05,
      "loss": 0.8056,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 35500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.0588235294117647,
      "grad_norm": 5.509186744689941,
      "learning_rate": 3.235294117647059e-05,
      "loss": 0.8267,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 36000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.0735294117647058,
      "grad_norm": 9.382353782653809,
      "learning_rate": 3.210784313725491e-05,
      "loss": 0.8298,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 36500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.088235294117647,
      "grad_norm": 5.186063766479492,
      "learning_rate": 3.186274509803922e-05,
      "loss": 0.8466,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 37000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.1029411764705883,
      "grad_norm": 6.467650413513184,
      "learning_rate": 3.161764705882353e-05,
      "loss": 0.8027,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 37500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.1176470588235294,
      "grad_norm": 4.107296943664551,
      "learning_rate": 3.137254901960784e-05,
      "loss": 0.7936,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 38000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.1323529411764706,
      "grad_norm": 2.7116880416870117,
      "learning_rate": 3.112745098039216e-05,
      "loss": 0.7658,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 38500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.1470588235294117,
      "grad_norm": 4.101169109344482,
      "learning_rate": 3.0882352941176475e-05,
      "loss": 0.8117,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 39000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.161764705882353,
      "grad_norm": 4.81936502456665,
      "learning_rate": 3.063725490196079e-05,
      "loss": 0.8001,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 39500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 7.343831539154053,
      "learning_rate": 3.0392156862745097e-05,
      "loss": 0.7704,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 40000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.1911764705882353,
      "grad_norm": 8.532153129577637,
      "learning_rate": 3.0147058823529413e-05,
      "loss": 0.7917,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 40500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.2058823529411764,
      "grad_norm": 7.05020809173584,
      "learning_rate": 2.9901960784313725e-05,
      "loss": 0.808,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 41000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.2205882352941178,
      "grad_norm": 3.989778995513916,
      "learning_rate": 2.965686274509804e-05,
      "loss": 0.8121,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 41500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.2352941176470589,
      "grad_norm": 6.2639689445495605,
      "learning_rate": 2.9411764705882354e-05,
      "loss": 0.7748,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 42000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.25,
      "grad_norm": 3.3618924617767334,
      "learning_rate": 2.916666666666667e-05,
      "loss": 0.813,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 42500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.2647058823529411,
      "grad_norm": 4.954437255859375,
      "learning_rate": 2.8921568627450986e-05,
      "loss": 0.7693,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 43000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.2794117647058822,
      "grad_norm": 3.4438722133636475,
      "learning_rate": 2.8676470588235295e-05,
      "loss": 0.7994,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 43500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.2941176470588236,
      "grad_norm": 4.17124080657959,
      "learning_rate": 2.8431372549019608e-05,
      "loss": 0.7764,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 44000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.3088235294117647,
      "grad_norm": 8.348847389221191,
      "learning_rate": 2.8186274509803924e-05,
      "loss": 0.7478,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 44500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.3235294117647058,
      "grad_norm": 1.2507061958312988,
      "learning_rate": 2.7941176470588236e-05,
      "loss": 0.7974,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 45000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.3382352941176472,
      "grad_norm": 13.898884773254395,
      "learning_rate": 2.7696078431372552e-05,
      "loss": 0.7541,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 45500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.3529411764705883,
      "grad_norm": 7.0425872802734375,
      "learning_rate": 2.7450980392156865e-05,
      "loss": 0.7837,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 46000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.3676470588235294,
      "grad_norm": 5.521500587463379,
      "learning_rate": 2.7205882352941174e-05,
      "loss": 0.7847,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 46500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.3823529411764706,
      "grad_norm": 7.491474628448486,
      "learning_rate": 2.696078431372549e-05,
      "loss": 0.748,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 47000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.3970588235294117,
      "grad_norm": 10.513397216796875,
      "learning_rate": 2.6715686274509806e-05,
      "loss": 0.7801,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 47500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.4117647058823528,
      "grad_norm": 15.570920944213867,
      "learning_rate": 2.647058823529412e-05,
      "loss": 0.7697,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 48000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.4264705882352942,
      "grad_norm": 4.1163010597229,
      "learning_rate": 2.6225490196078435e-05,
      "loss": 0.749,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 48500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.4411764705882353,
      "grad_norm": 7.289860248565674,
      "learning_rate": 2.5980392156862747e-05,
      "loss": 0.7788,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 49000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.4558823529411764,
      "grad_norm": 5.316174507141113,
      "learning_rate": 2.5735294117647057e-05,
      "loss": 0.8029,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 49500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.4705882352941178,
      "grad_norm": 3.60591721534729,
      "learning_rate": 2.5490196078431373e-05,
      "loss": 0.7372,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 50000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.4852941176470589,
      "grad_norm": 5.769986152648926,
      "learning_rate": 2.5245098039215685e-05,
      "loss": 0.765,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 50500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.5,
      "grad_norm": 3.587907552719116,
      "learning_rate": 2.5e-05,
      "loss": 0.7608,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 51000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.5147058823529411,
      "grad_norm": 2.3503806591033936,
      "learning_rate": 2.4754901960784317e-05,
      "loss": 0.7619,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 51500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.5294117647058822,
      "grad_norm": 9.60110855102539,
      "learning_rate": 2.4509803921568626e-05,
      "loss": 0.7633,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 52000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.5441176470588234,
      "grad_norm": 9.7755708694458,
      "learning_rate": 2.4264705882352942e-05,
      "loss": 0.7282,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 52500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.5588235294117647,
      "grad_norm": 6.872705936431885,
      "learning_rate": 2.401960784313726e-05,
      "loss": 0.7501,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 53000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.5735294117647058,
      "grad_norm": 4.416250228881836,
      "learning_rate": 2.3774509803921568e-05,
      "loss": 0.7622,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 53500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.5882352941176472,
      "grad_norm": 3.751142978668213,
      "learning_rate": 2.3529411764705884e-05,
      "loss": 0.7469,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 54000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.6029411764705883,
      "grad_norm": 2.0974745750427246,
      "learning_rate": 2.3284313725490196e-05,
      "loss": 0.7557,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 54500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.6176470588235294,
      "grad_norm": 3.7702689170837402,
      "learning_rate": 2.303921568627451e-05,
      "loss": 0.7468,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 55000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.6323529411764706,
      "grad_norm": 5.6461968421936035,
      "learning_rate": 2.2794117647058825e-05,
      "loss": 0.7501,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 55500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.6470588235294117,
      "grad_norm": 2.914978504180908,
      "learning_rate": 2.2549019607843138e-05,
      "loss": 0.7126,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 56000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.6617647058823528,
      "grad_norm": 5.506552219390869,
      "learning_rate": 2.2303921568627454e-05,
      "loss": 0.7176,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 56500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.6764705882352942,
      "grad_norm": 3.4189107418060303,
      "learning_rate": 2.2058823529411766e-05,
      "loss": 0.756,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 57000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.6911764705882353,
      "grad_norm": 10.335149765014648,
      "learning_rate": 2.181372549019608e-05,
      "loss": 0.7443,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 57500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.7058823529411766,
      "grad_norm": 1.2579532861709595,
      "learning_rate": 2.1568627450980395e-05,
      "loss": 0.7317,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 58000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.7205882352941178,
      "grad_norm": 5.039107322692871,
      "learning_rate": 2.1323529411764707e-05,
      "loss": 0.7626,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 58500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.7352941176470589,
      "grad_norm": 7.849141597747803,
      "learning_rate": 2.107843137254902e-05,
      "loss": 0.7506,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 59000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.323431491851807,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 0.7366,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 59500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.7647058823529411,
      "grad_norm": 4.651150703430176,
      "learning_rate": 2.058823529411765e-05,
      "loss": 0.7467,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 60000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.7794117647058822,
      "grad_norm": 2.9276676177978516,
      "learning_rate": 2.034313725490196e-05,
      "loss": 0.7328,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 60500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.7941176470588234,
      "grad_norm": 6.4653401374816895,
      "learning_rate": 2.0098039215686277e-05,
      "loss": 0.7213,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 61000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.8088235294117647,
      "grad_norm": 5.86958122253418,
      "learning_rate": 1.9852941176470586e-05,
      "loss": 0.726,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 61500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.8235294117647058,
      "grad_norm": 1.54404878616333,
      "learning_rate": 1.9607843137254903e-05,
      "loss": 0.7228,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 62000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.8382352941176472,
      "grad_norm": 3.6029295921325684,
      "learning_rate": 1.936274509803922e-05,
      "loss": 0.7194,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 62500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.8529411764705883,
      "grad_norm": 13.423847198486328,
      "learning_rate": 1.9117647058823528e-05,
      "loss": 0.745,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 63000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.8676470588235294,
      "grad_norm": 2.904106855392456,
      "learning_rate": 1.8872549019607844e-05,
      "loss": 0.7321,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 63500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.8823529411764706,
      "grad_norm": 2.838707447052002,
      "learning_rate": 1.862745098039216e-05,
      "loss": 0.7012,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 64000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.8970588235294117,
      "grad_norm": 3.0846309661865234,
      "learning_rate": 1.8382352941176472e-05,
      "loss": 0.6912,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 64500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.9117647058823528,
      "grad_norm": 5.948828220367432,
      "learning_rate": 1.8137254901960785e-05,
      "loss": 0.6854,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 65000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.9264705882352942,
      "grad_norm": 13.732516288757324,
      "learning_rate": 1.7892156862745098e-05,
      "loss": 0.6873,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 65500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.9411764705882353,
      "grad_norm": 10.169962882995605,
      "learning_rate": 1.7647058823529414e-05,
      "loss": 0.7158,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 66000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.9558823529411766,
      "grad_norm": 8.26551628112793,
      "learning_rate": 1.7401960784313726e-05,
      "loss": 0.7263,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 66500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.9705882352941178,
      "grad_norm": 8.526754379272461,
      "learning_rate": 1.715686274509804e-05,
      "loss": 0.6879,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 67000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 1.9852941176470589,
      "grad_norm": 11.833305358886719,
      "learning_rate": 1.6911764705882355e-05,
      "loss": 0.6696,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 67500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.8232524394989014,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.6767,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 68000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.014705882352941,
      "grad_norm": 7.861486434936523,
      "learning_rate": 1.642156862745098e-05,
      "loss": 0.688,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 68500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.0294117647058822,
      "grad_norm": 5.00877046585083,
      "learning_rate": 1.6176470588235296e-05,
      "loss": 0.6502,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 69000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.0441176470588234,
      "grad_norm": 3.252054214477539,
      "learning_rate": 1.593137254901961e-05,
      "loss": 0.6735,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 69500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.0588235294117645,
      "grad_norm": 8.655946731567383,
      "learning_rate": 1.568627450980392e-05,
      "loss": 0.7076,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 70000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.073529411764706,
      "grad_norm": 5.065808296203613,
      "learning_rate": 1.5441176470588237e-05,
      "loss": 0.6808,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 70500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.088235294117647,
      "grad_norm": 7.614666938781738,
      "learning_rate": 1.5196078431372548e-05,
      "loss": 0.6565,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 71000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.1029411764705883,
      "grad_norm": 5.335823059082031,
      "learning_rate": 1.4950980392156863e-05,
      "loss": 0.6929,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 71500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.1176470588235294,
      "grad_norm": 8.399900436401367,
      "learning_rate": 1.4705882352941177e-05,
      "loss": 0.6575,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 72000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.1323529411764706,
      "grad_norm": 4.3644537925720215,
      "learning_rate": 1.4460784313725493e-05,
      "loss": 0.7093,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 72500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.1470588235294117,
      "grad_norm": 5.606255054473877,
      "learning_rate": 1.4215686274509804e-05,
      "loss": 0.6497,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 73000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.161764705882353,
      "grad_norm": 8.045038223266602,
      "learning_rate": 1.3970588235294118e-05,
      "loss": 0.7041,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 73500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.176470588235294,
      "grad_norm": 10.108804702758789,
      "learning_rate": 1.3725490196078432e-05,
      "loss": 0.6512,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 74000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.1911764705882355,
      "grad_norm": 4.9611992835998535,
      "learning_rate": 1.3480392156862745e-05,
      "loss": 0.6748,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 74500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.2058823529411766,
      "grad_norm": 2.3439674377441406,
      "learning_rate": 1.323529411764706e-05,
      "loss": 0.6671,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 75000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.2205882352941178,
      "grad_norm": 5.686610698699951,
      "learning_rate": 1.2990196078431374e-05,
      "loss": 0.6246,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 75500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.235294117647059,
      "grad_norm": 3.122039794921875,
      "learning_rate": 1.2745098039215686e-05,
      "loss": 0.7125,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 76000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.25,
      "grad_norm": 12.2803373336792,
      "learning_rate": 1.25e-05,
      "loss": 0.6599,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 76500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.264705882352941,
      "grad_norm": 6.284506797790527,
      "learning_rate": 1.2254901960784313e-05,
      "loss": 0.6759,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 77000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.2794117647058822,
      "grad_norm": 3.540226697921753,
      "learning_rate": 1.200980392156863e-05,
      "loss": 0.6832,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 77500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.2941176470588234,
      "grad_norm": 7.354804992675781,
      "learning_rate": 1.1764705882352942e-05,
      "loss": 0.6289,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 78000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.3088235294117645,
      "grad_norm": 4.479936122894287,
      "learning_rate": 1.1519607843137254e-05,
      "loss": 0.6235,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 78500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.323529411764706,
      "grad_norm": 5.0150299072265625,
      "learning_rate": 1.1274509803921569e-05,
      "loss": 0.6257,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 79000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.338235294117647,
      "grad_norm": 6.954928874969482,
      "learning_rate": 1.1029411764705883e-05,
      "loss": 0.6543,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 79500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.3529411764705883,
      "grad_norm": 1.8401598930358887,
      "learning_rate": 1.0784313725490197e-05,
      "loss": 0.6433,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 80000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.3676470588235294,
      "grad_norm": 0.5422260761260986,
      "learning_rate": 1.053921568627451e-05,
      "loss": 0.652,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 80500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.3823529411764706,
      "grad_norm": 6.412265300750732,
      "learning_rate": 1.0294117647058824e-05,
      "loss": 0.6637,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 81000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.3970588235294117,
      "grad_norm": 5.444230556488037,
      "learning_rate": 1.0049019607843139e-05,
      "loss": 0.6218,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 81500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.411764705882353,
      "grad_norm": 3.0248329639434814,
      "learning_rate": 9.803921568627451e-06,
      "loss": 0.6663,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 82000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.426470588235294,
      "grad_norm": 2.8557262420654297,
      "learning_rate": 9.558823529411764e-06,
      "loss": 0.6768,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 82500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.4411764705882355,
      "grad_norm": 3.229672908782959,
      "learning_rate": 9.31372549019608e-06,
      "loss": 0.6566,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 83000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.4558823529411766,
      "grad_norm": 6.890370845794678,
      "learning_rate": 9.068627450980392e-06,
      "loss": 0.6589,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 83500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.4705882352941178,
      "grad_norm": 7.899547100067139,
      "learning_rate": 8.823529411764707e-06,
      "loss": 0.6775,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 84000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.485294117647059,
      "grad_norm": 7.249077796936035,
      "learning_rate": 8.57843137254902e-06,
      "loss": 0.6572,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 84500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.5,
      "grad_norm": 5.247090816497803,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.6348,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 85000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.514705882352941,
      "grad_norm": 5.179773330688477,
      "learning_rate": 8.088235294117648e-06,
      "loss": 0.6224,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 85500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.5294117647058822,
      "grad_norm": 8.414857864379883,
      "learning_rate": 7.84313725490196e-06,
      "loss": 0.6136,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 86000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.5441176470588234,
      "grad_norm": 1.9255198240280151,
      "learning_rate": 7.598039215686274e-06,
      "loss": 0.6418,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 86500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.5588235294117645,
      "grad_norm": 4.827670574188232,
      "learning_rate": 7.3529411764705884e-06,
      "loss": 0.6562,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 87000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.5735294117647056,
      "grad_norm": 5.845033645629883,
      "learning_rate": 7.107843137254902e-06,
      "loss": 0.6471,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 87500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.588235294117647,
      "grad_norm": 6.777614593505859,
      "learning_rate": 6.862745098039216e-06,
      "loss": 0.6311,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 88000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.6029411764705883,
      "grad_norm": 11.79529094696045,
      "learning_rate": 6.61764705882353e-06,
      "loss": 0.6289,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 88500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.6176470588235294,
      "grad_norm": 8.824967384338379,
      "learning_rate": 6.372549019607843e-06,
      "loss": 0.619,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 89000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.6323529411764706,
      "grad_norm": 6.913949012756348,
      "learning_rate": 6.127450980392157e-06,
      "loss": 0.652,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 89500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.6470588235294117,
      "grad_norm": 3.8313751220703125,
      "learning_rate": 5.882352941176471e-06,
      "loss": 0.6701,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 90000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.661764705882353,
      "grad_norm": 8.215780258178711,
      "learning_rate": 5.637254901960784e-06,
      "loss": 0.6433,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 90500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.6764705882352944,
      "grad_norm": 6.524360179901123,
      "learning_rate": 5.392156862745099e-06,
      "loss": 0.6048,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 91000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.6911764705882355,
      "grad_norm": 15.196798324584961,
      "learning_rate": 5.147058823529412e-06,
      "loss": 0.6427,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 91500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.7058823529411766,
      "grad_norm": 1.6884765625,
      "learning_rate": 4.901960784313726e-06,
      "loss": 0.6236,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 92000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.7205882352941178,
      "grad_norm": 2.3077046871185303,
      "learning_rate": 4.65686274509804e-06,
      "loss": 0.6163,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 92500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.735294117647059,
      "grad_norm": 5.261138916015625,
      "learning_rate": 4.411764705882353e-06,
      "loss": 0.6361,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 93000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.75,
      "grad_norm": 6.308595657348633,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.6245,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 93500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.764705882352941,
      "grad_norm": 5.467650413513184,
      "learning_rate": 3.92156862745098e-06,
      "loss": 0.621,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 94000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.7794117647058822,
      "grad_norm": 5.5255632400512695,
      "learning_rate": 3.6764705882352942e-06,
      "loss": 0.6446,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 94500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.7941176470588234,
      "grad_norm": 9.804576873779297,
      "learning_rate": 3.431372549019608e-06,
      "loss": 0.6226,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 95000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.8088235294117645,
      "grad_norm": 7.767389297485352,
      "learning_rate": 3.1862745098039216e-06,
      "loss": 0.6147,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 95500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.8235294117647056,
      "grad_norm": 2.8358209133148193,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 0.6416,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 96000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.838235294117647,
      "grad_norm": 7.5477190017700195,
      "learning_rate": 2.6960784313725493e-06,
      "loss": 0.6328,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 96500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.8529411764705883,
      "grad_norm": 13.355735778808594,
      "learning_rate": 2.450980392156863e-06,
      "loss": 0.6087,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 97000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.8676470588235294,
      "grad_norm": 5.399519443511963,
      "learning_rate": 2.2058823529411767e-06,
      "loss": 0.6063,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 97500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.8823529411764706,
      "grad_norm": 7.818796634674072,
      "learning_rate": 1.96078431372549e-06,
      "loss": 0.6211,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 98000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.8970588235294117,
      "grad_norm": 4.248461723327637,
      "learning_rate": 1.715686274509804e-06,
      "loss": 0.6054,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 98500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.911764705882353,
      "grad_norm": 6.80518102645874,
      "learning_rate": 1.4705882352941177e-06,
      "loss": 0.6261,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 99000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.9264705882352944,
      "grad_norm": 0.6795050501823425,
      "learning_rate": 1.2254901960784314e-06,
      "loss": 0.6062,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 99500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.9411764705882355,
      "grad_norm": 6.518540382385254,
      "learning_rate": 9.80392156862745e-07,
      "loss": 0.6304,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 100000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.9558823529411766,
      "grad_norm": 5.016809463500977,
      "learning_rate": 7.352941176470589e-07,
      "loss": 0.636,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 100500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.9705882352941178,
      "grad_norm": 7.016627788543701,
      "learning_rate": 4.901960784313725e-07,
      "loss": 0.6178,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 101000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 2.985294117647059,
      "grad_norm": 12.064628601074219,
      "learning_rate": 2.4509803921568627e-07,
      "loss": 0.6343,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 101500,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.9333810806274414,
      "learning_rate": 0.0,
      "loss": 0.5962,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 102000,
      "total_memory_available (GB)": 126.62
    },
    {
      "epoch": 3.0,
      "max_memory_allocated (GB)": 2.1,
      "memory_allocated (GB)": 1.46,
      "step": 102000,
      "total_flos": 6.324139790696448e+19,
      "total_memory_available (GB)": 126.62,
      "train_loss": 0.8088104343788297,
      "train_runtime": 4169.6612,
      "train_samples_per_second": 195.699,
      "train_steps_per_second": 24.462
    }
  ],
  "logging_steps": 500,
  "max_steps": 102000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.324139790696448e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}