{
  "best_metric": 0.8737932435834382,
  "best_model_checkpoint": "result/roberta-base-cls_before_pooler-sym_mlp-mlp_bert-bs64-gpu8-gs1-lr5e-5-m=stsb-cross_contra-norm0.05-l32",
  "epoch": 3.0,
  "global_step": 1617,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 4.969078540507112e-05,
      "loss": 6.1836,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.938157081014224e-05,
      "loss": 2.6504,
      "step": 20
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.907235621521336e-05,
      "loss": 1.6665,
      "step": 30
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.876314162028448e-05,
      "loss": 1.3732,
      "step": 40
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.84539270253556e-05,
      "loss": 1.2483,
      "step": 50
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.814471243042672e-05,
      "loss": 1.1685,
      "step": 60
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.783549783549784e-05,
      "loss": 1.0963,
      "step": 70
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.752628324056896e-05,
      "loss": 1.0671,
      "step": 80
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.721706864564008e-05,
      "loss": 1.0623,
      "step": 90
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.69078540507112e-05,
      "loss": 0.9754,
      "step": 100
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.6598639455782315e-05,
      "loss": 1.0497,
      "step": 110
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.628942486085344e-05,
      "loss": 1.0097,
      "step": 120
    },
    {
      "epoch": 0.23,
      "eval_avg_sts": 0.8361329859275041,
      "eval_sickr_spearman": 0.8112100964923686,
      "eval_stsb_spearman": 0.8610558753626395,
      "step": 125
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.5980210265924555e-05,
      "loss": 0.981,
      "step": 130
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.567099567099568e-05,
      "loss": 0.9648,
      "step": 140
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.5361781076066796e-05,
      "loss": 0.9035,
      "step": 150
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.505256648113791e-05,
      "loss": 0.9459,
      "step": 160
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.4743351886209036e-05,
      "loss": 0.9296,
      "step": 170
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.4434137291280146e-05,
      "loss": 0.9249,
      "step": 180
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.412492269635127e-05,
      "loss": 0.8699,
      "step": 190
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.381570810142239e-05,
      "loss": 0.88,
      "step": 200
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.3506493506493503e-05,
      "loss": 0.8992,
      "step": 210
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.319727891156463e-05,
      "loss": 0.893,
      "step": 220
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.2888064316635744e-05,
      "loss": 0.8977,
      "step": 230
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.257884972170687e-05,
      "loss": 0.8957,
      "step": 240
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.2269635126777984e-05,
      "loss": 0.8969,
      "step": 250
    },
    {
      "epoch": 0.46,
      "eval_avg_sts": 0.8389081420973826,
      "eval_sickr_spearman": 0.8101423170762979,
      "eval_stsb_spearman": 0.8676739671184673,
      "step": 250
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.19604205318491e-05,
      "loss": 0.8434,
      "step": 260
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1651205936920225e-05,
      "loss": 0.8362,
      "step": 270
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.134199134199134e-05,
      "loss": 0.8513,
      "step": 280
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.1032776747062465e-05,
      "loss": 0.8384,
      "step": 290
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.072356215213358e-05,
      "loss": 0.8653,
      "step": 300
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.04143475572047e-05,
      "loss": 0.8558,
      "step": 310
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.010513296227582e-05,
      "loss": 0.8241,
      "step": 320
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.979591836734694e-05,
      "loss": 0.8324,
      "step": 330
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.948670377241806e-05,
      "loss": 0.8396,
      "step": 340
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.917748917748918e-05,
      "loss": 0.8361,
      "step": 350
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.88682745825603e-05,
      "loss": 0.8737,
      "step": 360
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.855905998763142e-05,
      "loss": 0.84,
      "step": 370
    },
    {
      "epoch": 0.7,
      "eval_avg_sts": 0.8445458159728749,
      "eval_sickr_spearman": 0.8180373333129849,
      "eval_stsb_spearman": 0.8710542986327651,
      "step": 375
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.824984539270254e-05,
      "loss": 0.8384,
      "step": 380
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.794063079777366e-05,
      "loss": 0.8047,
      "step": 390
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.763141620284478e-05,
      "loss": 0.8039,
      "step": 400
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.7322201607915894e-05,
      "loss": 0.8474,
      "step": 410
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.701298701298702e-05,
      "loss": 0.7912,
      "step": 420
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.6703772418058135e-05,
      "loss": 0.8061,
      "step": 430
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.639455782312925e-05,
      "loss": 0.8132,
      "step": 440
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.6085343228200375e-05,
      "loss": 0.7815,
      "step": 450
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.577612863327149e-05,
      "loss": 0.7839,
      "step": 460
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.5466914038342616e-05,
      "loss": 0.7892,
      "step": 470
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.515769944341373e-05,
      "loss": 0.7738,
      "step": 480
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.484848484848485e-05,
      "loss": 0.7702,
      "step": 490
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.453927025355597e-05,
      "loss": 0.7715,
      "step": 500
    },
    {
      "epoch": 0.93,
      "eval_avg_sts": 0.8402489492013288,
      "eval_sickr_spearman": 0.8102146999461652,
      "eval_stsb_spearman": 0.8702831984564924,
      "step": 500
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.423005565862709e-05,
      "loss": 0.7891,
      "step": 510
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.392084106369821e-05,
      "loss": 0.7646,
      "step": 520
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.3611626468769324e-05,
      "loss": 0.7798,
      "step": 530
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.330241187384045e-05,
      "loss": 0.7456,
      "step": 540
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2993197278911564e-05,
      "loss": 0.6767,
      "step": 550
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.268398268398268e-05,
      "loss": 0.6897,
      "step": 560
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.2374768089053805e-05,
      "loss": 0.6593,
      "step": 570
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.206555349412492e-05,
      "loss": 0.697,
      "step": 580
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.1756338899196045e-05,
      "loss": 0.6816,
      "step": 590
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.144712430426716e-05,
      "loss": 0.6675,
      "step": 600
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.113790970933828e-05,
      "loss": 0.6661,
      "step": 610
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.08286951144094e-05,
      "loss": 0.6719,
      "step": 620
    },
    {
      "epoch": 1.16,
      "eval_avg_sts": 0.8420328093315328,
      "eval_sickr_spearman": 0.8171519279890747,
      "eval_stsb_spearman": 0.8669136906739908,
      "step": 625
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.051948051948052e-05,
      "loss": 0.7076,
      "step": 630
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.021026592455164e-05,
      "loss": 0.6515,
      "step": 640
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.990105132962276e-05,
      "loss": 0.6893,
      "step": 650
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.959183673469388e-05,
      "loss": 0.6714,
      "step": 660
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9282622139764997e-05,
      "loss": 0.7052,
      "step": 670
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.8973407544836117e-05,
      "loss": 0.6671,
      "step": 680
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8664192949907237e-05,
      "loss": 0.6735,
      "step": 690
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8354978354978357e-05,
      "loss": 0.6778,
      "step": 700
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.8045763760049478e-05,
      "loss": 0.6832,
      "step": 710
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7736549165120594e-05,
      "loss": 0.6587,
      "step": 720
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.7427334570191715e-05,
      "loss": 0.6567,
      "step": 730
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.7118119975262835e-05,
      "loss": 0.648,
      "step": 740
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.6808905380333955e-05,
      "loss": 0.6488,
      "step": 750
    },
    {
      "epoch": 1.39,
      "eval_avg_sts": 0.8417559206280472,
      "eval_sickr_spearman": 0.8131766819096506,
      "eval_stsb_spearman": 0.8703351593464438,
      "step": 750
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6499690785405072e-05,
      "loss": 0.6445,
      "step": 760
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6190476190476192e-05,
      "loss": 0.6763,
      "step": 770
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5881261595547312e-05,
      "loss": 0.6585,
      "step": 780
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5572047000618433e-05,
      "loss": 0.6492,
      "step": 790
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.5262832405689553e-05,
      "loss": 0.6634,
      "step": 800
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.495361781076067e-05,
      "loss": 0.6757,
      "step": 810
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4644403215831787e-05,
      "loss": 0.6805,
      "step": 820
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4335188620902907e-05,
      "loss": 0.6639,
      "step": 830
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.4025974025974027e-05,
      "loss": 0.6576,
      "step": 840
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3716759431045147e-05,
      "loss": 0.6849,
      "step": 850
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3407544836116267e-05,
      "loss": 0.7029,
      "step": 860
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.3098330241187384e-05,
      "loss": 0.6384,
      "step": 870
    },
    {
      "epoch": 1.62,
      "eval_avg_sts": 0.8415267776775321,
      "eval_sickr_spearman": 0.8138372056166275,
      "eval_stsb_spearman": 0.8692163497384369,
      "step": 875
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2789115646258505e-05,
      "loss": 0.6228,
      "step": 880
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.2479901051329625e-05,
      "loss": 0.6844,
      "step": 890
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2170686456400745e-05,
      "loss": 0.646,
      "step": 900
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1861471861471862e-05,
      "loss": 0.6328,
      "step": 910
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1552257266542982e-05,
      "loss": 0.6774,
      "step": 920
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.1243042671614102e-05,
      "loss": 0.6613,
      "step": 930
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.0933828076685223e-05,
      "loss": 0.6602,
      "step": 940
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.062461348175634e-05,
      "loss": 0.6581,
      "step": 950
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.031539888682746e-05,
      "loss": 0.6729,
      "step": 960
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.0006184291898576e-05,
      "loss": 0.6428,
      "step": 970
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9696969696969697e-05,
      "loss": 0.6552,
      "step": 980
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9387755102040817e-05,
      "loss": 0.6501,
      "step": 990
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.9078540507111937e-05,
      "loss": 0.6839,
      "step": 1000
    },
    {
      "epoch": 1.86,
      "eval_avg_sts": 0.8442180854633452,
      "eval_sickr_spearman": 0.814642927343252,
      "eval_stsb_spearman": 0.8737932435834382,
      "step": 1000
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8769325912183054e-05,
      "loss": 0.646,
      "step": 1010
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8460111317254174e-05,
      "loss": 0.6217,
      "step": 1020
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8150896722325294e-05,
      "loss": 0.6278,
      "step": 1030
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7841682127396415e-05,
      "loss": 0.6456,
      "step": 1040
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7532467532467535e-05,
      "loss": 0.6459,
      "step": 1050
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.7223252937538652e-05,
      "loss": 0.6223,
      "step": 1060
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6914038342609772e-05,
      "loss": 0.6357,
      "step": 1070
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6604823747680892e-05,
      "loss": 0.6018,
      "step": 1080
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.6295609152752012e-05,
      "loss": 0.5888,
      "step": 1090
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.5986394557823133e-05,
      "loss": 0.5913,
      "step": 1100
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.567717996289425e-05,
      "loss": 0.5581,
      "step": 1110
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5367965367965366e-05,
      "loss": 0.5899,
      "step": 1120
    },
    {
      "epoch": 2.09,
      "eval_avg_sts": 0.8410988161028077,
      "eval_sickr_spearman": 0.8122731208293968,
      "eval_stsb_spearman": 0.8699245113762184,
      "step": 1125
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.5058750773036487e-05,
      "loss": 0.5456,
      "step": 1130
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4749536178107607e-05,
      "loss": 0.6023,
      "step": 1140
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.4440321583178725e-05,
      "loss": 0.6004,
      "step": 1150
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4131106988249846e-05,
      "loss": 0.588,
      "step": 1160
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3821892393320964e-05,
      "loss": 0.5716,
      "step": 1170
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3512677798392084e-05,
      "loss": 0.5813,
      "step": 1180
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3203463203463205e-05,
      "loss": 0.5615,
      "step": 1190
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2894248608534323e-05,
      "loss": 0.5874,
      "step": 1200
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2585034013605443e-05,
      "loss": 0.5983,
      "step": 1210
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2275819418676562e-05,
      "loss": 0.5922,
      "step": 1220
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.1966604823747682e-05,
      "loss": 0.5772,
      "step": 1230
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.16573902288188e-05,
      "loss": 0.5772,
      "step": 1240
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1348175633889919e-05,
      "loss": 0.5828,
      "step": 1250
    },
    {
      "epoch": 2.32,
      "eval_avg_sts": 0.8417651170306701,
      "eval_sickr_spearman": 0.8143169402577914,
      "eval_stsb_spearman": 0.869213293803549,
      "step": 1250
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.103896103896104e-05,
      "loss": 0.5919,
      "step": 1260
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0729746444032158e-05,
      "loss": 0.5587,
      "step": 1270
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0420531849103278e-05,
      "loss": 0.5779,
      "step": 1280
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0111317254174398e-05,
      "loss": 0.5422,
      "step": 1290
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.802102659245517e-06,
      "loss": 0.5915,
      "step": 1300
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.492888064316637e-06,
      "loss": 0.5794,
      "step": 1310
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.183673469387756e-06,
      "loss": 0.5832,
      "step": 1320
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.874458874458876e-06,
      "loss": 0.5644,
      "step": 1330
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.565244279529994e-06,
      "loss": 0.5704,
      "step": 1340
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.256029684601113e-06,
      "loss": 0.5692,
      "step": 1350
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.946815089672233e-06,
      "loss": 0.5794,
      "step": 1360
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.637600494743352e-06,
      "loss": 0.5793,
      "step": 1370
    },
    {
      "epoch": 2.55,
      "eval_avg_sts": 0.8438379193565275,
      "eval_sickr_spearman": 0.8155493222584906,
      "eval_stsb_spearman": 0.8721265164545645,
      "step": 1375
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.328385899814472e-06,
      "loss": 0.57,
      "step": 1380
    },
    {
      "epoch": 2.58,
      "learning_rate": 7.019171304885591e-06,
      "loss": 0.5759,
      "step": 1390
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.709956709956711e-06,
      "loss": 0.5518,
      "step": 1400
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.40074211502783e-06,
      "loss": 0.5608,
      "step": 1410
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.091527520098949e-06,
      "loss": 0.5544,
      "step": 1420
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.782312925170069e-06,
      "loss": 0.5701,
      "step": 1430
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.473098330241188e-06,
      "loss": 0.5886,
      "step": 1440
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.163883735312307e-06,
      "loss": 0.5781,
      "step": 1450
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.854669140383426e-06,
      "loss": 0.5767,
      "step": 1460
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 0.5859,
      "step": 1470
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.236239950525665e-06,
      "loss": 0.5906,
      "step": 1480
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.927025355596784e-06,
      "loss": 0.5621,
      "step": 1490
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.6178107606679037e-06,
      "loss": 0.5505,
      "step": 1500
    },
    {
      "epoch": 2.78,
      "eval_avg_sts": 0.8425113202508864,
      "eval_sickr_spearman": 0.813630912035951,
      "eval_stsb_spearman": 0.8713917284658218,
      "step": 1500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.308596165739023e-06,
      "loss": 0.5843,
      "step": 1510
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.9993815708101424e-06,
      "loss": 0.5518,
      "step": 1520
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.690166975881262e-06,
      "loss": 0.5837,
      "step": 1530
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.3809523809523808e-06,
      "loss": 0.5736,
      "step": 1540
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.0717377860235006e-06,
      "loss": 0.5655,
      "step": 1550
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.7625231910946195e-06,
      "loss": 0.5563,
      "step": 1560
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.4533085961657391e-06,
      "loss": 0.5646,
      "step": 1570
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.1440940012368585e-06,
      "loss": 0.5493,
      "step": 1580
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.348794063079778e-07,
      "loss": 0.5594,
      "step": 1590
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.25664811379097e-07,
      "loss": 0.5628,
      "step": 1600
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.1645021645021646e-07,
      "loss": 0.59,
      "step": 1610
    },
    {
      "epoch": 3.0,
      "step": 1617,
      "train_runtime": 1218.5135,
      "train_samples_per_second": 1.327
    }
  ],
  "max_steps": 1617,
  "num_train_epochs": 3,
  "total_flos": 59643369959989248,
  "trial_name": null,
  "trial_params": null
}