{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.8449049873931491,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.5801353078493935e-06,
      "loss": 0.8726,
      "step": 10
    },
    {
      "epoch": 0.02,
      "learning_rate": 7.087352805422317e-06,
      "loss": 0.8672,
      "step": 20
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.228161798644422e-06,
      "loss": 0.8565,
      "step": 30
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.014161010104347e-06,
      "loss": 0.8213,
      "step": 40
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.673229499590088e-06,
      "loss": 0.7977,
      "step": 50
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.987442444537463e-06,
      "loss": 0.7521,
      "step": 60
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.94558392632901e-06,
      "loss": 0.7423,
      "step": 70
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.903725408120554e-06,
      "loss": 0.7026,
      "step": 80
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.861866889912098e-06,
      "loss": 0.6808,
      "step": 90
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.820008371703642e-06,
      "loss": 0.6901,
      "step": 100
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.778149853495186e-06,
      "loss": 0.6514,
      "step": 110
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.736291335286732e-06,
      "loss": 0.6467,
      "step": 120
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.694432817078277e-06,
      "loss": 0.6641,
      "step": 130
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.652574298869821e-06,
      "loss": 0.6616,
      "step": 140
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.610715780661365e-06,
      "loss": 0.6617,
      "step": 150
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.56885726245291e-06,
      "loss": 0.6295,
      "step": 160
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.526998744244454e-06,
      "loss": 0.6071,
      "step": 170
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.485140226036e-06,
      "loss": 0.6354,
      "step": 180
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.443281707827544e-06,
      "loss": 0.6211,
      "step": 190
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.401423189619088e-06,
      "loss": 0.6272,
      "step": 200
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.359564671410633e-06,
      "loss": 0.641,
      "step": 210
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.317706153202177e-06,
      "loss": 0.6205,
      "step": 220
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.275847634993721e-06,
      "loss": 0.6263,
      "step": 230
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.233989116785267e-06,
      "loss": 0.6148,
      "step": 240
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.192130598576812e-06,
      "loss": 0.6521,
      "step": 250
    },
    {
      "epoch": 0.31,
      "eval_oasst_export_w_label_accuracy": 0.7346153846153847,
      "eval_oasst_export_w_label_kendalltau": 0.3859706959706959,
      "eval_oasst_export_w_label_loss": 0.626953125,
      "eval_oasst_export_w_label_neg_score": 0.2529296875,
      "eval_oasst_export_w_label_pos_score": 0.91064453125,
      "eval_oasst_export_w_label_runtime": 190.5015,
      "eval_oasst_export_w_label_samples_per_second": 9.554,
      "eval_oasst_export_w_label_score_diff": 0.65771484375,
      "eval_oasst_export_w_label_steps_per_second": 2.388,
      "step": 251
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.150272080368356e-06,
      "loss": 0.602,
      "step": 260
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.1084135621599e-06,
      "loss": 0.5637,
      "step": 270
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.066555043951444e-06,
      "loss": 0.5948,
      "step": 280
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.02469652574299e-06,
      "loss": 0.6245,
      "step": 290
    },
    {
      "epoch": 0.37,
      "learning_rate": 8.982838007534535e-06,
      "loss": 0.5943,
      "step": 300
    },
    {
      "epoch": 0.38,
      "learning_rate": 8.940979489326079e-06,
      "loss": 0.5934,
      "step": 310
    },
    {
      "epoch": 0.39,
      "learning_rate": 8.899120971117623e-06,
      "loss": 0.5426,
      "step": 320
    },
    {
      "epoch": 0.41,
      "learning_rate": 8.857262452909168e-06,
      "loss": 0.6191,
      "step": 330
    },
    {
      "epoch": 0.42,
      "learning_rate": 8.815403934700712e-06,
      "loss": 0.6316,
      "step": 340
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.773545416492258e-06,
      "loss": 0.611,
      "step": 350
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.7316868982838e-06,
      "loss": 0.6434,
      "step": 360
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.689828380075346e-06,
      "loss": 0.6327,
      "step": 370
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.64796986186689e-06,
      "loss": 0.6021,
      "step": 380
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.606111343658435e-06,
      "loss": 0.5986,
      "step": 390
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.564252825449981e-06,
      "loss": 0.5798,
      "step": 400
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.522394307241524e-06,
      "loss": 0.6055,
      "step": 410
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.480535789033068e-06,
      "loss": 0.6172,
      "step": 420
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.438677270824614e-06,
      "loss": 0.6657,
      "step": 430
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.396818752616158e-06,
      "loss": 0.6206,
      "step": 440
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.354960234407702e-06,
      "loss": 0.6177,
      "step": 450
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.313101716199248e-06,
      "loss": 0.6201,
      "step": 460
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.271243197990791e-06,
      "loss": 0.5299,
      "step": 470
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.229384679782337e-06,
      "loss": 0.6733,
      "step": 480
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.187526161573881e-06,
      "loss": 0.6546,
      "step": 490
    },
    {
      "epoch": 0.61,
      "learning_rate": 8.145667643365426e-06,
      "loss": 0.6234,
      "step": 500
    },
    {
      "epoch": 0.62,
      "eval_oasst_export_w_label_accuracy": 0.7423076923076923,
      "eval_oasst_export_w_label_kendalltau": 0.3931501831501834,
      "eval_oasst_export_w_label_loss": 0.60888671875,
      "eval_oasst_export_w_label_neg_score": 0.53515625,
      "eval_oasst_export_w_label_pos_score": 1.1513671875,
      "eval_oasst_export_w_label_runtime": 190.2556,
      "eval_oasst_export_w_label_samples_per_second": 9.566,
      "eval_oasst_export_w_label_score_diff": 0.6162109375,
      "eval_oasst_export_w_label_steps_per_second": 2.392,
      "step": 502
    },
    {
      "epoch": 0.63,
      "learning_rate": 8.10380912515697e-06,
      "loss": 0.6087,
      "step": 510
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.061950606948514e-06,
      "loss": 0.6345,
      "step": 520
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.020092088740058e-06,
      "loss": 0.5881,
      "step": 530
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.978233570531604e-06,
      "loss": 0.5954,
      "step": 540
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.936375052323149e-06,
      "loss": 0.61,
      "step": 550
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.894516534114693e-06,
      "loss": 0.5808,
      "step": 560
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.852658015906237e-06,
      "loss": 0.6579,
      "step": 570
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.810799497697782e-06,
      "loss": 0.6155,
      "step": 580
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.768940979489327e-06,
      "loss": 0.6049,
      "step": 590
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.727082461280872e-06,
      "loss": 0.6135,
      "step": 600
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.685223943072416e-06,
      "loss": 0.5793,
      "step": 610
    },
    {
      "epoch": 0.76,
      "learning_rate": 7.64336542486396e-06,
      "loss": 0.6091,
      "step": 620
    },
    {
      "epoch": 0.77,
      "learning_rate": 7.601506906655505e-06,
      "loss": 0.6196,
      "step": 630
    },
    {
      "epoch": 0.79,
      "learning_rate": 7.55964838844705e-06,
      "loss": 0.5951,
      "step": 640
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.517789870238594e-06,
      "loss": 0.6366,
      "step": 650
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.475931352030139e-06,
      "loss": 0.6109,
      "step": 660
    },
    {
      "epoch": 0.82,
      "learning_rate": 7.434072833821683e-06,
      "loss": 0.5869,
      "step": 670
    },
    {
      "epoch": 0.84,
      "learning_rate": 7.392214315613228e-06,
      "loss": 0.612,
      "step": 680
    },
    {
      "epoch": 0.85,
      "learning_rate": 7.350355797404772e-06,
      "loss": 0.6019,
      "step": 690
    },
    {
      "epoch": 0.86,
      "learning_rate": 7.308497279196317e-06,
      "loss": 0.6265,
      "step": 700
    },
    {
      "epoch": 0.87,
      "learning_rate": 7.2666387609878615e-06,
      "loss": 0.5994,
      "step": 710
    },
    {
      "epoch": 0.89,
      "learning_rate": 7.224780242779407e-06,
      "loss": 0.5607,
      "step": 720
    },
    {
      "epoch": 0.9,
      "learning_rate": 7.18292172457095e-06,
      "loss": 0.5985,
      "step": 730
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.141063206362495e-06,
      "loss": 0.5915,
      "step": 740
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.09920468815404e-06,
      "loss": 0.5442,
      "step": 750
    },
    {
      "epoch": 0.93,
      "eval_oasst_export_w_label_accuracy": 0.7478021978021978,
      "eval_oasst_export_w_label_kendalltau": 0.4016483516483517,
      "eval_oasst_export_w_label_loss": 0.62255859375,
      "eval_oasst_export_w_label_neg_score": -0.1964111328125,
      "eval_oasst_export_w_label_pos_score": 0.82470703125,
      "eval_oasst_export_w_label_runtime": 190.3739,
      "eval_oasst_export_w_label_samples_per_second": 9.56,
      "eval_oasst_export_w_label_score_diff": 1.021484375,
      "eval_oasst_export_w_label_steps_per_second": 2.39,
      "step": 753
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.057346169945585e-06,
      "loss": 0.6559,
      "step": 760
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.01548765173713e-06,
      "loss": 0.5909,
      "step": 770
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.973629133528673e-06,
      "loss": 0.6263,
      "step": 780
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.931770615320218e-06,
      "loss": 0.6288,
      "step": 790
    },
    {
      "epoch": 0.98,
      "learning_rate": 6.889912097111763e-06,
      "loss": 0.5627,
      "step": 800
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.848053578903308e-06,
      "loss": 0.6076,
      "step": 810
    },
    {
      "epoch": 1.01,
      "learning_rate": 6.806195060694852e-06,
      "loss": 0.567,
      "step": 820
    },
    {
      "epoch": 1.02,
      "learning_rate": 6.764336542486396e-06,
      "loss": 0.565,
      "step": 830
    },
    {
      "epoch": 1.03,
      "learning_rate": 6.722478024277941e-06,
      "loss": 0.6091,
      "step": 840
    },
    {
      "epoch": 1.05,
      "learning_rate": 6.680619506069486e-06,
      "loss": 0.5579,
      "step": 850
    },
    {
      "epoch": 1.06,
      "learning_rate": 6.63876098786103e-06,
      "loss": 0.5943,
      "step": 860
    },
    {
      "epoch": 1.07,
      "learning_rate": 6.596902469652575e-06,
      "loss": 0.566,
      "step": 870
    },
    {
      "epoch": 1.08,
      "learning_rate": 6.555043951444119e-06,
      "loss": 0.5757,
      "step": 880
    },
    {
      "epoch": 1.09,
      "learning_rate": 6.513185433235664e-06,
      "loss": 0.5752,
      "step": 890
    },
    {
      "epoch": 1.11,
      "learning_rate": 6.471326915027208e-06,
      "loss": 0.5595,
      "step": 900
    },
    {
      "epoch": 1.12,
      "learning_rate": 6.429468396818753e-06,
      "loss": 0.6051,
      "step": 910
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.387609878610298e-06,
      "loss": 0.5308,
      "step": 920
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.345751360401843e-06,
      "loss": 0.5341,
      "step": 930
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.303892842193386e-06,
      "loss": 0.5976,
      "step": 940
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.262034323984931e-06,
      "loss": 0.568,
      "step": 950
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.220175805776476e-06,
      "loss": 0.5865,
      "step": 960
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.178317287568021e-06,
      "loss": 0.5486,
      "step": 970
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.136458769359566e-06,
      "loss": 0.5699,
      "step": 980
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.094600251151109e-06,
      "loss": 0.5831,
      "step": 990
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.052741732942654e-06,
      "loss": 0.5558,
      "step": 1000
    },
    {
      "epoch": 1.23,
      "eval_oasst_export_w_label_accuracy": 0.743956043956044,
      "eval_oasst_export_w_label_kendalltau": 0.3847985347985353,
      "eval_oasst_export_w_label_loss": 0.61865234375,
      "eval_oasst_export_w_label_neg_score": -0.09515380859375,
      "eval_oasst_export_w_label_pos_score": 0.93212890625,
      "eval_oasst_export_w_label_runtime": 190.324,
      "eval_oasst_export_w_label_samples_per_second": 9.563,
      "eval_oasst_export_w_label_score_diff": 1.02734375,
      "eval_oasst_export_w_label_steps_per_second": 2.391,
      "step": 1004
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.010883214734199e-06,
      "loss": 0.5286,
      "step": 1010
    },
    {
      "epoch": 1.25,
      "learning_rate": 5.969024696525744e-06,
      "loss": 0.5657,
      "step": 1020
    },
    {
      "epoch": 1.27,
      "learning_rate": 5.927166178317288e-06,
      "loss": 0.5376,
      "step": 1030
    },
    {
      "epoch": 1.28,
      "learning_rate": 5.885307660108832e-06,
      "loss": 0.5375,
      "step": 1040
    },
    {
      "epoch": 1.29,
      "learning_rate": 5.843449141900377e-06,
      "loss": 0.5513,
      "step": 1050
    },
    {
      "epoch": 1.3,
      "learning_rate": 5.801590623691922e-06,
      "loss": 0.5778,
      "step": 1060
    },
    {
      "epoch": 1.32,
      "learning_rate": 5.759732105483467e-06,
      "loss": 0.5445,
      "step": 1070
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.717873587275011e-06,
      "loss": 0.5657,
      "step": 1080
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.676015069066555e-06,
      "loss": 0.5808,
      "step": 1090
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.6341565508581e-06,
      "loss": 0.5723,
      "step": 1100
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.592298032649645e-06,
      "loss": 0.5413,
      "step": 1110
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.550439514441189e-06,
      "loss": 0.6384,
      "step": 1120
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.508580996232734e-06,
      "loss": 0.5336,
      "step": 1130
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.466722478024279e-06,
      "loss": 0.5754,
      "step": 1140
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.424863959815823e-06,
      "loss": 0.5171,
      "step": 1150
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.383005441607367e-06,
      "loss": 0.5536,
      "step": 1160
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.341146923398912e-06,
      "loss": 0.5812,
      "step": 1170
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.299288405190457e-06,
      "loss": 0.5604,
      "step": 1180
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.257429886982002e-06,
      "loss": 0.5433,
      "step": 1190
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.215571368773545e-06,
      "loss": 0.5704,
      "step": 1200
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.17371285056509e-06,
      "loss": 0.592,
      "step": 1210
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.131854332356635e-06,
      "loss": 0.5406,
      "step": 1220
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.08999581414818e-06,
      "loss": 0.5597,
      "step": 1230
    },
    {
      "epoch": 1.53,
      "learning_rate": 5.048137295939725e-06,
      "loss": 0.6011,
      "step": 1240
    },
    {
      "epoch": 1.54,
      "learning_rate": 5.006278777731268e-06,
      "loss": 0.5071,
      "step": 1250
    },
    {
      "epoch": 1.54,
      "eval_oasst_export_w_label_accuracy": 0.7478021978021978,
      "eval_oasst_export_w_label_kendalltau": 0.4045054945054951,
      "eval_oasst_export_w_label_loss": 0.61669921875,
      "eval_oasst_export_w_label_neg_score": -0.1842041015625,
      "eval_oasst_export_w_label_pos_score": 1.02734375,
      "eval_oasst_export_w_label_runtime": 190.3207,
      "eval_oasst_export_w_label_samples_per_second": 9.563,
      "eval_oasst_export_w_label_score_diff": 1.2109375,
      "eval_oasst_export_w_label_steps_per_second": 2.391,
      "step": 1255
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.964420259522813e-06,
      "loss": 0.5374,
      "step": 1260
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.922561741314358e-06,
      "loss": 0.5613,
      "step": 1270
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.880703223105903e-06,
      "loss": 0.5643,
      "step": 1280
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.838844704897447e-06,
      "loss": 0.5755,
      "step": 1290
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.7969861866889915e-06,
      "loss": 0.5631,
      "step": 1300
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.755127668480537e-06,
      "loss": 0.5557,
      "step": 1310
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.713269150272081e-06,
      "loss": 0.5173,
      "step": 1320
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.671410632063625e-06,
      "loss": 0.5938,
      "step": 1330
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.6295521138551695e-06,
      "loss": 0.6386,
      "step": 1340
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.587693595646715e-06,
      "loss": 0.5652,
      "step": 1350
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.545835077438259e-06,
      "loss": 0.6097,
      "step": 1360
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.503976559229803e-06,
      "loss": 0.5369,
      "step": 1370
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.462118041021348e-06,
      "loss": 0.539,
      "step": 1380
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.4202595228128935e-06,
      "loss": 0.5854,
      "step": 1390
    },
    {
      "epoch": 1.72,
      "learning_rate": 4.378401004604437e-06,
      "loss": 0.5574,
      "step": 1400
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.336542486395982e-06,
      "loss": 0.5079,
      "step": 1410
    },
    {
      "epoch": 1.75,
      "learning_rate": 4.294683968187526e-06,
      "loss": 0.5889,
      "step": 1420
    },
    {
      "epoch": 1.76,
      "learning_rate": 4.2528254499790715e-06,
      "loss": 0.5908,
      "step": 1430
    },
    {
      "epoch": 1.77,
      "learning_rate": 4.210966931770616e-06,
      "loss": 0.5254,
      "step": 1440
    },
    {
      "epoch": 1.78,
      "learning_rate": 4.16910841356216e-06,
      "loss": 0.5566,
      "step": 1450
    },
    {
      "epoch": 1.8,
      "learning_rate": 4.127249895353705e-06,
      "loss": 0.5743,
      "step": 1460
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.0853913771452495e-06,
      "loss": 0.5208,
      "step": 1470
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.043532858936794e-06,
      "loss": 0.5486,
      "step": 1480
    },
    {
      "epoch": 1.83,
      "learning_rate": 4.001674340728339e-06,
      "loss": 0.587,
      "step": 1490
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.959815822519883e-06,
      "loss": 0.5781,
      "step": 1500
    }
  ],
  "max_steps": 2439,
  "num_train_epochs": 3,
  "total_flos": 0.0,
  "trial_name": null,
  "trial_params": null
}