{
  "best_metric": 0.5101504325866699,
  "best_model_checkpoint": "./vit-beta1-0.85/checkpoint-5778",
  "epoch": 28.0,
  "eval_steps": 500,
  "global_step": 8988,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 23.742420196533203,
      "learning_rate": 1.8291979226774382e-05,
      "loss": 1.763,
      "step": 321
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6952149791955617,
      "eval_f1": 0.6201672138002952,
      "eval_loss": 0.9504804015159607,
      "eval_precision": 0.6345772764957709,
      "eval_recall": 0.6952149791955617,
      "eval_runtime": 23.0549,
      "eval_samples_per_second": 125.093,
      "eval_steps_per_second": 15.658,
      "step": 321
    },
    {
      "epoch": 2.0,
      "grad_norm": 20.97748374938965,
      "learning_rate": 3.681477207155222e-05,
      "loss": 1.149,
      "step": 642
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7444521497919556,
      "eval_f1": 0.7229775407881395,
      "eval_loss": 0.714677631855011,
      "eval_precision": 0.74566856205471,
      "eval_recall": 0.7444521497919556,
      "eval_runtime": 23.0018,
      "eval_samples_per_second": 125.381,
      "eval_steps_per_second": 15.694,
      "step": 642
    },
    {
      "epoch": 3.0,
      "grad_norm": 8.279248237609863,
      "learning_rate": 5.5337564916330066e-05,
      "loss": 1.0452,
      "step": 963
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7572815533980582,
      "eval_f1": 0.7320657787389255,
      "eval_loss": 0.6250461935997009,
      "eval_precision": 0.7591347407203577,
      "eval_recall": 0.7572815533980582,
      "eval_runtime": 23.1015,
      "eval_samples_per_second": 124.841,
      "eval_steps_per_second": 15.627,
      "step": 963
    },
    {
      "epoch": 4.0,
      "grad_norm": 41.02663803100586,
      "learning_rate": 7.386035776110792e-05,
      "loss": 1.0048,
      "step": 1284
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.7784327323162274,
      "eval_f1": 0.7737493702586475,
      "eval_loss": 0.5614432692527771,
      "eval_precision": 0.779177150378666,
      "eval_recall": 0.7784327323162274,
      "eval_runtime": 23.126,
      "eval_samples_per_second": 124.708,
      "eval_steps_per_second": 15.61,
      "step": 1284
    },
    {
      "epoch": 5.0,
      "grad_norm": 7.507669448852539,
      "learning_rate": 9.238315060588575e-05,
      "loss": 0.931,
      "step": 1605
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.7739251040221914,
      "eval_f1": 0.7822626472987246,
      "eval_loss": 0.6081557869911194,
      "eval_precision": 0.8019901444431083,
      "eval_recall": 0.7739251040221914,
      "eval_runtime": 22.3445,
      "eval_samples_per_second": 129.07,
      "eval_steps_per_second": 16.156,
      "step": 1605
    },
    {
      "epoch": 6.0,
      "grad_norm": 8.154308319091797,
      "learning_rate": 9.984715255878176e-05,
      "loss": 0.9808,
      "step": 1926
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.7981969486823856,
      "eval_f1": 0.7950945949242283,
      "eval_loss": 0.554189145565033,
      "eval_precision": 0.7983952215842433,
      "eval_recall": 0.7981969486823856,
      "eval_runtime": 22.8586,
      "eval_samples_per_second": 126.167,
      "eval_steps_per_second": 15.793,
      "step": 1926
    },
    {
      "epoch": 7.0,
      "grad_norm": 11.704543113708496,
      "learning_rate": 9.889061131437471e-05,
      "loss": 0.8908,
      "step": 2247
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.7545076282940361,
      "eval_f1": 0.7708612112738275,
      "eval_loss": 0.5956947207450867,
      "eval_precision": 0.820205289282276,
      "eval_recall": 0.7545076282940361,
      "eval_runtime": 23.7447,
      "eval_samples_per_second": 121.459,
      "eval_steps_per_second": 15.203,
      "step": 2247
    },
    {
      "epoch": 8.0,
      "grad_norm": 3.4443271160125732,
      "learning_rate": 9.707265436104638e-05,
      "loss": 0.7747,
      "step": 2568
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.7694174757281553,
      "eval_f1": 0.7836405012298657,
      "eval_loss": 0.5766288638114929,
      "eval_precision": 0.8155227380188221,
      "eval_recall": 0.7694174757281553,
      "eval_runtime": 23.1168,
      "eval_samples_per_second": 124.758,
      "eval_steps_per_second": 15.616,
      "step": 2568
    },
    {
      "epoch": 9.0,
      "grad_norm": 7.433899879455566,
      "learning_rate": 9.44253127296151e-05,
      "loss": 0.741,
      "step": 2889
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.7995839112343966,
      "eval_f1": 0.804684275131234,
      "eval_loss": 0.5430790185928345,
      "eval_precision": 0.8189836827516774,
      "eval_recall": 0.7995839112343966,
      "eval_runtime": 23.7114,
      "eval_samples_per_second": 121.629,
      "eval_steps_per_second": 15.225,
      "step": 2889
    },
    {
      "epoch": 10.0,
      "grad_norm": 12.33841609954834,
      "learning_rate": 9.099523058358976e-05,
      "loss": 0.7179,
      "step": 3210
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.7773925104022191,
      "eval_f1": 0.7904480536083208,
      "eval_loss": 0.5864874720573425,
      "eval_precision": 0.831331916729736,
      "eval_recall": 0.7773925104022191,
      "eval_runtime": 23.035,
      "eval_samples_per_second": 125.201,
      "eval_steps_per_second": 15.672,
      "step": 3210
    },
    {
      "epoch": 11.0,
      "grad_norm": 21.430646896362305,
      "learning_rate": 8.684284338417735e-05,
      "loss": 0.6102,
      "step": 3531
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.8096393897364771,
      "eval_f1": 0.8180305563212311,
      "eval_loss": 0.5288417935371399,
      "eval_precision": 0.8361295244369513,
      "eval_recall": 0.8096393897364771,
      "eval_runtime": 23.7358,
      "eval_samples_per_second": 121.504,
      "eval_steps_per_second": 15.209,
      "step": 3531
    },
    {
      "epoch": 12.0,
      "grad_norm": 75.45726013183594,
      "learning_rate": 8.204131306302357e-05,
      "loss": 0.574,
      "step": 3852
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.7995839112343966,
      "eval_f1": 0.8096217027832384,
      "eval_loss": 0.599109947681427,
      "eval_precision": 0.8332319256993149,
      "eval_recall": 0.7995839112343966,
      "eval_runtime": 23.369,
      "eval_samples_per_second": 123.411,
      "eval_steps_per_second": 15.448,
      "step": 3852
    },
    {
      "epoch": 13.0,
      "grad_norm": 3.8113694190979004,
      "learning_rate": 7.667523896413962e-05,
      "loss": 0.4515,
      "step": 4173
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.8370319001386962,
      "eval_f1": 0.8292572001875775,
      "eval_loss": 0.5889743566513062,
      "eval_precision": 0.8333558129412982,
      "eval_recall": 0.8370319001386962,
      "eval_runtime": 23.6944,
      "eval_samples_per_second": 121.717,
      "eval_steps_per_second": 15.236,
      "step": 4173
    },
    {
      "epoch": 14.0,
      "grad_norm": 5.727066516876221,
      "learning_rate": 7.083916726724684e-05,
      "loss": 0.4629,
      "step": 4494
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.8120665742024965,
      "eval_f1": 0.8205347890470436,
      "eval_loss": 0.5573027729988098,
      "eval_precision": 0.8462670503631797,
      "eval_recall": 0.8120665742024965,
      "eval_runtime": 24.8143,
      "eval_samples_per_second": 116.223,
      "eval_steps_per_second": 14.548,
      "step": 4494
    },
    {
      "epoch": 15.0,
      "grad_norm": 4.605546474456787,
      "learning_rate": 6.463592515537568e-05,
      "loss": 0.3927,
      "step": 4815
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.8332177531206657,
      "eval_f1": 0.835722282090202,
      "eval_loss": 0.5279321670532227,
      "eval_precision": 0.8506448765217454,
      "eval_recall": 0.8332177531206657,
      "eval_runtime": 25.6733,
      "eval_samples_per_second": 112.335,
      "eval_steps_per_second": 14.061,
      "step": 4815
    },
    {
      "epoch": 16.0,
      "grad_norm": 5.051961898803711,
      "learning_rate": 5.8174809077430184e-05,
      "loss": 0.3535,
      "step": 5136
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.8356449375866851,
      "eval_f1": 0.8404866875551775,
      "eval_loss": 0.5363761782646179,
      "eval_precision": 0.8494425873765888,
      "eval_recall": 0.8356449375866851,
      "eval_runtime": 23.6271,
      "eval_samples_per_second": 122.063,
      "eval_steps_per_second": 15.279,
      "step": 5136
    },
    {
      "epoch": 17.0,
      "grad_norm": 28.548500061035156,
      "learning_rate": 5.156965902716534e-05,
      "loss": 0.2635,
      "step": 5457
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.8547156726768377,
      "eval_f1": 0.853209116290087,
      "eval_loss": 0.5475490093231201,
      "eval_precision": 0.8625841938606562,
      "eval_recall": 0.8547156726768377,
      "eval_runtime": 25.2986,
      "eval_samples_per_second": 113.999,
      "eval_steps_per_second": 14.27,
      "step": 5457
    },
    {
      "epoch": 18.0,
      "grad_norm": 16.80175018310547,
      "learning_rate": 4.493685276832998e-05,
      "loss": 0.2493,
      "step": 5778
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.855755894590846,
      "eval_f1": 0.8553441136144744,
      "eval_loss": 0.5101504325866699,
      "eval_precision": 0.8568218210572909,
      "eval_recall": 0.855755894590846,
      "eval_runtime": 23.831,
      "eval_samples_per_second": 121.019,
      "eval_steps_per_second": 15.148,
      "step": 5778
    },
    {
      "epoch": 19.0,
      "grad_norm": 0.11270666122436523,
      "learning_rate": 3.839325534621579e-05,
      "loss": 0.2125,
      "step": 6099
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.8328710124826629,
      "eval_f1": 0.8418273768641612,
      "eval_loss": 0.6120281219482422,
      "eval_precision": 0.8622992078235902,
      "eval_recall": 0.8328710124826629,
      "eval_runtime": 23.1564,
      "eval_samples_per_second": 124.544,
      "eval_steps_per_second": 15.59,
      "step": 6099
    },
    {
      "epoch": 20.0,
      "grad_norm": 0.06091728433966637,
      "learning_rate": 3.205416001367289e-05,
      "loss": 0.2179,
      "step": 6420
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.8567961165048543,
      "eval_f1": 0.85627072801028,
      "eval_loss": 0.5720838904380798,
      "eval_precision": 0.8563063857199564,
      "eval_recall": 0.8567961165048543,
      "eval_runtime": 23.124,
      "eval_samples_per_second": 124.719,
      "eval_steps_per_second": 15.611,
      "step": 6420
    },
    {
      "epoch": 21.0,
      "grad_norm": 2.7066216468811035,
      "learning_rate": 2.6067582030088143e-05,
      "loss": 0.1598,
      "step": 6741
    },
    {
      "epoch": 21.0,
      "eval_accuracy": 0.8651178918169209,
      "eval_f1": 0.8632510833215378,
      "eval_loss": 0.5503374338150024,
      "eval_precision": 0.8623213572654986,
      "eval_recall": 0.8651178918169209,
      "eval_runtime": 23.2584,
      "eval_samples_per_second": 123.998,
      "eval_steps_per_second": 15.521,
      "step": 6741
    },
    {
      "epoch": 22.0,
      "grad_norm": 37.88532257080078,
      "learning_rate": 2.0464044677195966e-05,
      "loss": 0.1194,
      "step": 7062
    },
    {
      "epoch": 22.0,
      "eval_accuracy": 0.8678918169209431,
      "eval_f1": 0.8668972179399497,
      "eval_loss": 0.5829476714134216,
      "eval_precision": 0.8672496533540811,
      "eval_recall": 0.8678918169209431,
      "eval_runtime": 23.2515,
      "eval_samples_per_second": 124.035,
      "eval_steps_per_second": 15.526,
      "step": 7062
    },
    {
      "epoch": 23.0,
      "grad_norm": 0.11761299520730972,
      "learning_rate": 1.538090858763548e-05,
      "loss": 0.1245,
      "step": 7383
    },
    {
      "epoch": 23.0,
      "eval_accuracy": 0.8682385575589459,
      "eval_f1": 0.8629371653200789,
      "eval_loss": 0.6137722134590149,
      "eval_precision": 0.8631764321501026,
      "eval_recall": 0.8682385575589459,
      "eval_runtime": 24.6364,
      "eval_samples_per_second": 117.063,
      "eval_steps_per_second": 14.653,
      "step": 7383
    },
    {
      "epoch": 24.0,
      "grad_norm": 0.2003249078989029,
      "learning_rate": 1.0907734786732799e-05,
      "loss": 0.1239,
      "step": 7704
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.8730929264909847,
      "eval_f1": 0.8694766569722234,
      "eval_loss": 0.6135957837104797,
      "eval_precision": 0.8702946096941098,
      "eval_recall": 0.8730929264909847,
      "eval_runtime": 23.1398,
      "eval_samples_per_second": 124.633,
      "eval_steps_per_second": 15.601,
      "step": 7704
    },
    {
      "epoch": 25.0,
      "grad_norm": 0.018545789644122124,
      "learning_rate": 7.123337224185811e-06,
      "loss": 0.1159,
      "step": 8025
    },
    {
      "epoch": 25.0,
      "eval_accuracy": 0.8751733703190014,
      "eval_f1": 0.8726121524469985,
      "eval_loss": 0.5931476354598999,
      "eval_precision": 0.8724487380217695,
      "eval_recall": 0.8751733703190014,
      "eval_runtime": 23.8052,
      "eval_samples_per_second": 121.15,
      "eval_steps_per_second": 15.165,
      "step": 8025
    },
    {
      "epoch": 26.0,
      "grad_norm": 13.244468688964844,
      "learning_rate": 4.094394131694684e-06,
      "loss": 0.089,
      "step": 8346
    },
    {
      "epoch": 26.0,
      "eval_accuracy": 0.8776005547850209,
      "eval_f1": 0.8749752761914638,
      "eval_loss": 0.5846670866012573,
      "eval_precision": 0.8742768994280256,
      "eval_recall": 0.8776005547850209,
      "eval_runtime": 23.0803,
      "eval_samples_per_second": 124.955,
      "eval_steps_per_second": 15.641,
      "step": 8346
    },
    {
      "epoch": 27.0,
      "grad_norm": 0.4510207176208496,
      "learning_rate": 1.8742732027931087e-06,
      "loss": 0.1123,
      "step": 8667
    },
    {
      "epoch": 27.0,
      "eval_accuracy": 0.8751733703190014,
      "eval_f1": 0.8718954079920697,
      "eval_loss": 0.594070553779602,
      "eval_precision": 0.8710238593528318,
      "eval_recall": 0.8751733703190014,
      "eval_runtime": 22.9979,
      "eval_samples_per_second": 125.403,
      "eval_steps_per_second": 15.697,
      "step": 8667
    },
    {
      "epoch": 28.0,
      "grad_norm": 0.05787363648414612,
      "learning_rate": 5.020912943263345e-07,
      "loss": 0.0779,
      "step": 8988
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.8765603328710125,
      "eval_f1": 0.8729435556323918,
      "eval_loss": 0.6037870049476624,
      "eval_precision": 0.872156471246016,
      "eval_recall": 0.8765603328710125,
      "eval_runtime": 24.3339,
      "eval_samples_per_second": 118.518,
      "eval_steps_per_second": 14.835,
      "step": 8988
    },
    {
      "epoch": 28.0,
      "step": 8988,
      "total_flos": 1.1127108458244538e+19,
      "train_loss": 0.5253127112513815,
      "train_runtime": 3374.2787,
      "train_samples_per_second": 151.973,
      "train_steps_per_second": 9.513
    }
  ],
  "logging_steps": 500,
  "max_steps": 32100,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 100,
  "save_steps": 500,
  "total_flos": 1.1127108458244538e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}