{
  "best_metric": 0.6899755597114563,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.3401360544217687,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0017006802721088435,
      "grad_norm": 0.22153174877166748,
      "learning_rate": 1e-05,
      "loss": 0.8125,
      "step": 1
    },
    {
      "epoch": 0.0017006802721088435,
      "eval_loss": 1.186415433883667,
      "eval_runtime": 124.9498,
      "eval_samples_per_second": 7.923,
      "eval_steps_per_second": 1.985,
      "step": 1
    },
    {
      "epoch": 0.003401360544217687,
      "grad_norm": 0.24999311566352844,
      "learning_rate": 2e-05,
      "loss": 0.8426,
      "step": 2
    },
    {
      "epoch": 0.00510204081632653,
      "grad_norm": 0.31139490008354187,
      "learning_rate": 3e-05,
      "loss": 0.9179,
      "step": 3
    },
    {
      "epoch": 0.006802721088435374,
      "grad_norm": 0.2909492254257202,
      "learning_rate": 4e-05,
      "loss": 0.9067,
      "step": 4
    },
    {
      "epoch": 0.008503401360544218,
      "grad_norm": 0.2692619860172272,
      "learning_rate": 5e-05,
      "loss": 0.8637,
      "step": 5
    },
    {
      "epoch": 0.01020408163265306,
      "grad_norm": 0.26725709438323975,
      "learning_rate": 6e-05,
      "loss": 0.8966,
      "step": 6
    },
    {
      "epoch": 0.011904761904761904,
      "grad_norm": 0.2664909064769745,
      "learning_rate": 7e-05,
      "loss": 0.818,
      "step": 7
    },
    {
      "epoch": 0.013605442176870748,
      "grad_norm": 0.24496795237064362,
      "learning_rate": 8e-05,
      "loss": 0.894,
      "step": 8
    },
    {
      "epoch": 0.015306122448979591,
      "grad_norm": 0.20702075958251953,
      "learning_rate": 9e-05,
      "loss": 0.849,
      "step": 9
    },
    {
      "epoch": 0.017006802721088437,
      "grad_norm": 0.20149566233158112,
      "learning_rate": 0.0001,
      "loss": 0.9358,
      "step": 10
    },
    {
      "epoch": 0.01870748299319728,
      "grad_norm": 0.2666966915130615,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.7822,
      "step": 11
    },
    {
      "epoch": 0.02040816326530612,
      "grad_norm": 0.2660152018070221,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.7644,
      "step": 12
    },
    {
      "epoch": 0.022108843537414966,
      "grad_norm": 0.28873929381370544,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.8288,
      "step": 13
    },
    {
      "epoch": 0.023809523809523808,
      "grad_norm": 0.20585861802101135,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.7245,
      "step": 14
    },
    {
      "epoch": 0.025510204081632654,
      "grad_norm": 0.20767712593078613,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.8772,
      "step": 15
    },
    {
      "epoch": 0.027210884353741496,
      "grad_norm": 0.1987304538488388,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.9127,
      "step": 16
    },
    {
      "epoch": 0.02891156462585034,
      "grad_norm": 0.2005297690629959,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.8824,
      "step": 17
    },
    {
      "epoch": 0.030612244897959183,
      "grad_norm": 0.19833321869373322,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.856,
      "step": 18
    },
    {
      "epoch": 0.03231292517006803,
      "grad_norm": 0.21213646233081818,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.9459,
      "step": 19
    },
    {
      "epoch": 0.034013605442176874,
      "grad_norm": 0.22667399048805237,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.9033,
      "step": 20
    },
    {
      "epoch": 0.03571428571428571,
      "grad_norm": 0.20015716552734375,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.8934,
      "step": 21
    },
    {
      "epoch": 0.03741496598639456,
      "grad_norm": 0.24498829245567322,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.8923,
      "step": 22
    },
    {
      "epoch": 0.0391156462585034,
      "grad_norm": 0.197325199842453,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.8122,
      "step": 23
    },
    {
      "epoch": 0.04081632653061224,
      "grad_norm": 0.22649633884429932,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.8663,
      "step": 24
    },
    {
      "epoch": 0.04251700680272109,
      "grad_norm": 0.20791395008563995,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.8384,
      "step": 25
    },
    {
      "epoch": 0.04421768707482993,
      "grad_norm": 0.21198329329490662,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.8571,
      "step": 26
    },
    {
      "epoch": 0.04591836734693878,
      "grad_norm": 0.23952525854110718,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.8493,
      "step": 27
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 0.23386171460151672,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.8947,
      "step": 28
    },
    {
      "epoch": 0.04931972789115646,
      "grad_norm": 0.2588789761066437,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.9538,
      "step": 29
    },
    {
      "epoch": 0.05102040816326531,
      "grad_norm": 0.2704077959060669,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.9163,
      "step": 30
    },
    {
      "epoch": 0.05272108843537415,
      "grad_norm": 0.26958054304122925,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.8531,
      "step": 31
    },
    {
      "epoch": 0.05442176870748299,
      "grad_norm": 0.296394944190979,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.9323,
      "step": 32
    },
    {
      "epoch": 0.05612244897959184,
      "grad_norm": 0.28837698698043823,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.8862,
      "step": 33
    },
    {
      "epoch": 0.05782312925170068,
      "grad_norm": 0.3194785416126251,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.8426,
      "step": 34
    },
    {
      "epoch": 0.05952380952380952,
      "grad_norm": 0.33932724595069885,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.9595,
      "step": 35
    },
    {
      "epoch": 0.061224489795918366,
      "grad_norm": 0.3013361096382141,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.8495,
      "step": 36
    },
    {
      "epoch": 0.06292517006802721,
      "grad_norm": 0.38077646493911743,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.8389,
      "step": 37
    },
    {
      "epoch": 0.06462585034013606,
      "grad_norm": 0.34454333782196045,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.9841,
      "step": 38
    },
    {
      "epoch": 0.0663265306122449,
      "grad_norm": 0.3004533052444458,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.7992,
      "step": 39
    },
    {
      "epoch": 0.06802721088435375,
      "grad_norm": 0.293108195066452,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.8365,
      "step": 40
    },
    {
      "epoch": 0.06972789115646258,
      "grad_norm": 0.26349979639053345,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.9266,
      "step": 41
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 0.24968861043453217,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.848,
      "step": 42
    },
    {
      "epoch": 0.07312925170068027,
      "grad_norm": 0.25499433279037476,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.773,
      "step": 43
    },
    {
      "epoch": 0.07482993197278912,
      "grad_norm": 0.2647852897644043,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.8602,
      "step": 44
    },
    {
      "epoch": 0.07653061224489796,
      "grad_norm": 0.29176974296569824,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.8878,
      "step": 45
    },
    {
      "epoch": 0.0782312925170068,
      "grad_norm": 0.262540340423584,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.7934,
      "step": 46
    },
    {
      "epoch": 0.07993197278911565,
      "grad_norm": 0.28180527687072754,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.7712,
      "step": 47
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 0.2878844141960144,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.7283,
      "step": 48
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 0.31438517570495605,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.7335,
      "step": 49
    },
    {
      "epoch": 0.08503401360544217,
      "grad_norm": 0.3259439468383789,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.6936,
      "step": 50
    },
    {
      "epoch": 0.08503401360544217,
      "eval_loss": 0.7770039439201355,
      "eval_runtime": 127.3416,
      "eval_samples_per_second": 7.774,
      "eval_steps_per_second": 1.948,
      "step": 50
    },
    {
      "epoch": 0.08673469387755102,
      "grad_norm": 0.22468984127044678,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.6047,
      "step": 51
    },
    {
      "epoch": 0.08843537414965986,
      "grad_norm": 0.19191181659698486,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.6422,
      "step": 52
    },
    {
      "epoch": 0.09013605442176871,
      "grad_norm": 0.17332300543785095,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.6352,
      "step": 53
    },
    {
      "epoch": 0.09183673469387756,
      "grad_norm": 0.15703777968883514,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.6882,
      "step": 54
    },
    {
      "epoch": 0.0935374149659864,
      "grad_norm": 0.14402662217617035,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.68,
      "step": 55
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 0.14448726177215576,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.6176,
      "step": 56
    },
    {
      "epoch": 0.09693877551020408,
      "grad_norm": 0.14335256814956665,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.6866,
      "step": 57
    },
    {
      "epoch": 0.09863945578231292,
      "grad_norm": 0.14909979701042175,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.7024,
      "step": 58
    },
    {
      "epoch": 0.10034013605442177,
      "grad_norm": 0.15612825751304626,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.6926,
      "step": 59
    },
    {
      "epoch": 0.10204081632653061,
      "grad_norm": 0.18259373307228088,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.7686,
      "step": 60
    },
    {
      "epoch": 0.10374149659863946,
      "grad_norm": 0.16720153391361237,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.6996,
      "step": 61
    },
    {
      "epoch": 0.1054421768707483,
      "grad_norm": 0.1870233118534088,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.748,
      "step": 62
    },
    {
      "epoch": 0.10714285714285714,
      "grad_norm": 0.18254293501377106,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.7073,
      "step": 63
    },
    {
      "epoch": 0.10884353741496598,
      "grad_norm": 0.2915474772453308,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.6906,
      "step": 64
    },
    {
      "epoch": 0.11054421768707483,
      "grad_norm": 0.19223552942276,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.7241,
      "step": 65
    },
    {
      "epoch": 0.11224489795918367,
      "grad_norm": 0.19417499005794525,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.8011,
      "step": 66
    },
    {
      "epoch": 0.11394557823129252,
      "grad_norm": 0.20863336324691772,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.7768,
      "step": 67
    },
    {
      "epoch": 0.11564625850340136,
      "grad_norm": 0.20013408362865448,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.8215,
      "step": 68
    },
    {
      "epoch": 0.11734693877551021,
      "grad_norm": 0.2080213874578476,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.7709,
      "step": 69
    },
    {
      "epoch": 0.11904761904761904,
      "grad_norm": 0.22706864774227142,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.8945,
      "step": 70
    },
    {
      "epoch": 0.12074829931972789,
      "grad_norm": 0.23297883570194244,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.7946,
      "step": 71
    },
    {
      "epoch": 0.12244897959183673,
      "grad_norm": 0.1977277547121048,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.737,
      "step": 72
    },
    {
      "epoch": 0.12414965986394558,
      "grad_norm": 0.20646902918815613,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.7431,
      "step": 73
    },
    {
      "epoch": 0.12585034013605442,
      "grad_norm": 0.2172570377588272,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.7489,
      "step": 74
    },
    {
      "epoch": 0.12755102040816327,
      "grad_norm": 0.22185806930065155,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.7936,
      "step": 75
    },
    {
      "epoch": 0.1292517006802721,
      "grad_norm": 0.22498106956481934,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.7585,
      "step": 76
    },
    {
      "epoch": 0.13095238095238096,
      "grad_norm": 0.2507253587245941,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.6934,
      "step": 77
    },
    {
      "epoch": 0.1326530612244898,
      "grad_norm": 0.2425968497991562,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.7123,
      "step": 78
    },
    {
      "epoch": 0.13435374149659865,
      "grad_norm": 0.25330182909965515,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.7705,
      "step": 79
    },
    {
      "epoch": 0.1360544217687075,
      "grad_norm": 0.2874208390712738,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.7719,
      "step": 80
    },
    {
      "epoch": 0.1377551020408163,
      "grad_norm": 0.4115007817745209,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.8468,
      "step": 81
    },
    {
      "epoch": 0.13945578231292516,
      "grad_norm": 0.5738610029220581,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.7712,
      "step": 82
    },
    {
      "epoch": 0.141156462585034,
      "grad_norm": 0.31313905119895935,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.777,
      "step": 83
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.33824095129966736,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.7802,
      "step": 84
    },
    {
      "epoch": 0.1445578231292517,
      "grad_norm": 0.3063613772392273,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.6995,
      "step": 85
    },
    {
      "epoch": 0.14625850340136054,
      "grad_norm": 0.2932402193546295,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.7355,
      "step": 86
    },
    {
      "epoch": 0.14795918367346939,
      "grad_norm": 0.3320849537849426,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.775,
      "step": 87
    },
    {
      "epoch": 0.14965986394557823,
      "grad_norm": 0.3260820508003235,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.7322,
      "step": 88
    },
    {
      "epoch": 0.15136054421768708,
      "grad_norm": 0.3097473382949829,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.748,
      "step": 89
    },
    {
      "epoch": 0.15306122448979592,
      "grad_norm": 0.2701698839664459,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.7404,
      "step": 90
    },
    {
      "epoch": 0.15476190476190477,
      "grad_norm": 0.2577556073665619,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.7713,
      "step": 91
    },
    {
      "epoch": 0.1564625850340136,
      "grad_norm": 0.2623687982559204,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.7626,
      "step": 92
    },
    {
      "epoch": 0.15816326530612246,
      "grad_norm": 0.26207149028778076,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.6944,
      "step": 93
    },
    {
      "epoch": 0.1598639455782313,
      "grad_norm": 0.25475212931632996,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.7277,
      "step": 94
    },
    {
      "epoch": 0.16156462585034015,
      "grad_norm": 0.3018037974834442,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.7631,
      "step": 95
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 0.26409912109375,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.6952,
      "step": 96
    },
    {
      "epoch": 0.1649659863945578,
      "grad_norm": 0.3402869403362274,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.7449,
      "step": 97
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.31432080268859863,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.6847,
      "step": 98
    },
    {
      "epoch": 0.1683673469387755,
      "grad_norm": 0.35394060611724854,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.7377,
      "step": 99
    },
    {
      "epoch": 0.17006802721088435,
      "grad_norm": 0.397417277097702,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.7064,
      "step": 100
    },
    {
      "epoch": 0.17006802721088435,
      "eval_loss": 0.7253766655921936,
      "eval_runtime": 127.2121,
      "eval_samples_per_second": 7.782,
      "eval_steps_per_second": 1.95,
      "step": 100
    },
    {
      "epoch": 0.1717687074829932,
      "grad_norm": 0.15172697603702545,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.589,
      "step": 101
    },
    {
      "epoch": 0.17346938775510204,
      "grad_norm": 0.15769332647323608,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.5558,
      "step": 102
    },
    {
      "epoch": 0.17517006802721088,
      "grad_norm": 0.15128929913043976,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.6205,
      "step": 103
    },
    {
      "epoch": 0.17687074829931973,
      "grad_norm": 0.16581159830093384,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.6377,
      "step": 104
    },
    {
      "epoch": 0.17857142857142858,
      "grad_norm": 0.1451852023601532,
      "learning_rate": 5e-05,
      "loss": 0.565,
      "step": 105
    },
    {
      "epoch": 0.18027210884353742,
      "grad_norm": 0.16264212131500244,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.6301,
      "step": 106
    },
    {
      "epoch": 0.18197278911564627,
      "grad_norm": 0.15103355050086975,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.6943,
      "step": 107
    },
    {
      "epoch": 0.1836734693877551,
      "grad_norm": 0.1626369208097458,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.6535,
      "step": 108
    },
    {
      "epoch": 0.18537414965986396,
      "grad_norm": 0.15834422409534454,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.6697,
      "step": 109
    },
    {
      "epoch": 0.1870748299319728,
      "grad_norm": 0.1552390158176422,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.5966,
      "step": 110
    },
    {
      "epoch": 0.18877551020408162,
      "grad_norm": 0.16435950994491577,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.6343,
      "step": 111
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 0.18012185394763947,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.7195,
      "step": 112
    },
    {
      "epoch": 0.1921768707482993,
      "grad_norm": 0.19753223657608032,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.7592,
      "step": 113
    },
    {
      "epoch": 0.19387755102040816,
      "grad_norm": 0.18591929972171783,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.7626,
      "step": 114
    },
    {
      "epoch": 0.195578231292517,
      "grad_norm": 0.20835751295089722,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.7074,
      "step": 115
    },
    {
      "epoch": 0.19727891156462585,
      "grad_norm": 0.1931455284357071,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.696,
      "step": 116
    },
    {
      "epoch": 0.1989795918367347,
      "grad_norm": 0.2061833143234253,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.725,
      "step": 117
    },
    {
      "epoch": 0.20068027210884354,
      "grad_norm": 0.20548807084560394,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.6552,
      "step": 118
    },
    {
      "epoch": 0.20238095238095238,
      "grad_norm": 0.20919598639011383,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.7117,
      "step": 119
    },
    {
      "epoch": 0.20408163265306123,
      "grad_norm": 0.2350170612335205,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.6716,
      "step": 120
    },
    {
      "epoch": 0.20578231292517007,
      "grad_norm": 0.2034604698419571,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.6757,
      "step": 121
    },
    {
      "epoch": 0.20748299319727892,
      "grad_norm": 0.2228289246559143,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.719,
      "step": 122
    },
    {
      "epoch": 0.20918367346938777,
      "grad_norm": 0.2372211068868637,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.6627,
      "step": 123
    },
    {
      "epoch": 0.2108843537414966,
      "grad_norm": 0.23014475405216217,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.7278,
      "step": 124
    },
    {
      "epoch": 0.21258503401360543,
      "grad_norm": 0.231934055685997,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.8059,
      "step": 125
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 0.2525619566440582,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.7829,
      "step": 126
    },
    {
      "epoch": 0.21598639455782312,
      "grad_norm": 0.22821013629436493,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.703,
      "step": 127
    },
    {
      "epoch": 0.21768707482993196,
      "grad_norm": 0.2759142220020294,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.831,
      "step": 128
    },
    {
      "epoch": 0.2193877551020408,
      "grad_norm": 0.2888336479663849,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.7977,
      "step": 129
    },
    {
      "epoch": 0.22108843537414966,
      "grad_norm": 0.2612399458885193,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.7792,
      "step": 130
    },
    {
      "epoch": 0.2227891156462585,
      "grad_norm": 0.2800489664077759,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.7285,
      "step": 131
    },
    {
      "epoch": 0.22448979591836735,
      "grad_norm": 0.3610658645629883,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.7198,
      "step": 132
    },
    {
      "epoch": 0.2261904761904762,
      "grad_norm": 0.2689877450466156,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.742,
      "step": 133
    },
    {
      "epoch": 0.22789115646258504,
      "grad_norm": 0.29836830496788025,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.7214,
      "step": 134
    },
    {
      "epoch": 0.22959183673469388,
      "grad_norm": 0.3338813781738281,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.7189,
      "step": 135
    },
    {
      "epoch": 0.23129251700680273,
      "grad_norm": 0.3424853980541229,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.7628,
      "step": 136
    },
    {
      "epoch": 0.23299319727891157,
      "grad_norm": 0.3546268939971924,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.6624,
      "step": 137
    },
    {
      "epoch": 0.23469387755102042,
      "grad_norm": 0.40582168102264404,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.6435,
      "step": 138
    },
    {
      "epoch": 0.23639455782312926,
      "grad_norm": 0.3407474160194397,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.7789,
      "step": 139
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 0.4899592697620392,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.8151,
      "step": 140
    },
    {
      "epoch": 0.23979591836734693,
      "grad_norm": 0.2846715450286865,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.7211,
      "step": 141
    },
    {
      "epoch": 0.24149659863945577,
      "grad_norm": 0.25135180354118347,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.6928,
      "step": 142
    },
    {
      "epoch": 0.24319727891156462,
      "grad_norm": 0.2946225106716156,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.7354,
      "step": 143
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 0.2937641143798828,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.6778,
      "step": 144
    },
    {
      "epoch": 0.2465986394557823,
      "grad_norm": 0.3299817740917206,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.8131,
      "step": 145
    },
    {
      "epoch": 0.24829931972789115,
      "grad_norm": 0.3105611503124237,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.6599,
      "step": 146
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.31812918186187744,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.6576,
      "step": 147
    },
    {
      "epoch": 0.25170068027210885,
      "grad_norm": 0.2982231080532074,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.5782,
      "step": 148
    },
    {
      "epoch": 0.2534013605442177,
      "grad_norm": 0.3875696361064911,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.6566,
      "step": 149
    },
    {
      "epoch": 0.25510204081632654,
      "grad_norm": 0.41503244638442993,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.6468,
      "step": 150
    },
    {
      "epoch": 0.25510204081632654,
      "eval_loss": 0.6988990306854248,
      "eval_runtime": 126.8575,
      "eval_samples_per_second": 7.804,
      "eval_steps_per_second": 1.955,
      "step": 150
    },
    {
      "epoch": 0.2568027210884354,
      "grad_norm": 0.13521897792816162,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.561,
      "step": 151
    },
    {
      "epoch": 0.2585034013605442,
      "grad_norm": 0.13190492987632751,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.5361,
      "step": 152
    },
    {
      "epoch": 0.2602040816326531,
      "grad_norm": 0.1558769941329956,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.5745,
      "step": 153
    },
    {
      "epoch": 0.2619047619047619,
      "grad_norm": 0.158156618475914,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.6341,
      "step": 154
    },
    {
      "epoch": 0.26360544217687076,
      "grad_norm": 0.16553521156311035,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.5872,
      "step": 155
    },
    {
      "epoch": 0.2653061224489796,
      "grad_norm": 0.16343246400356293,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.5714,
      "step": 156
    },
    {
      "epoch": 0.26700680272108845,
      "grad_norm": 0.17622269690036774,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.5801,
      "step": 157
    },
    {
      "epoch": 0.2687074829931973,
      "grad_norm": 0.17592722177505493,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.6579,
      "step": 158
    },
    {
      "epoch": 0.27040816326530615,
      "grad_norm": 0.18341432511806488,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.6813,
      "step": 159
    },
    {
      "epoch": 0.272108843537415,
      "grad_norm": 0.17723487317562103,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.5837,
      "step": 160
    },
    {
      "epoch": 0.27380952380952384,
      "grad_norm": 0.19495318830013275,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.6575,
      "step": 161
    },
    {
      "epoch": 0.2755102040816326,
      "grad_norm": 0.18905970454216003,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.6932,
      "step": 162
    },
    {
      "epoch": 0.27721088435374147,
      "grad_norm": 0.18781432509422302,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.5665,
      "step": 163
    },
    {
      "epoch": 0.2789115646258503,
      "grad_norm": 0.22015555202960968,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.664,
      "step": 164
    },
    {
      "epoch": 0.28061224489795916,
      "grad_norm": 0.21366287767887115,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.7489,
      "step": 165
    },
    {
      "epoch": 0.282312925170068,
      "grad_norm": 0.2087860256433487,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.7017,
      "step": 166
    },
    {
      "epoch": 0.28401360544217685,
      "grad_norm": 0.234105184674263,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.6215,
      "step": 167
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.23136542737483978,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.7564,
      "step": 168
    },
    {
      "epoch": 0.28741496598639454,
      "grad_norm": 0.21522806584835052,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.7532,
      "step": 169
    },
    {
      "epoch": 0.2891156462585034,
      "grad_norm": 0.23693765699863434,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.7102,
      "step": 170
    },
    {
      "epoch": 0.29081632653061223,
      "grad_norm": 0.2175215631723404,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.6938,
      "step": 171
    },
    {
      "epoch": 0.2925170068027211,
      "grad_norm": 0.25002777576446533,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.6949,
      "step": 172
    },
    {
      "epoch": 0.2942176870748299,
      "grad_norm": 0.24212975800037384,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.6643,
      "step": 173
    },
    {
      "epoch": 0.29591836734693877,
      "grad_norm": 0.2649815082550049,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.7444,
      "step": 174
    },
    {
      "epoch": 0.2976190476190476,
      "grad_norm": 0.26538190245628357,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.7986,
      "step": 175
    },
    {
      "epoch": 0.29931972789115646,
      "grad_norm": 0.2911010980606079,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.781,
      "step": 176
    },
    {
      "epoch": 0.3010204081632653,
      "grad_norm": 0.2879495322704315,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.7473,
      "step": 177
    },
    {
      "epoch": 0.30272108843537415,
      "grad_norm": 0.33281680941581726,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.7358,
      "step": 178
    },
    {
      "epoch": 0.304421768707483,
      "grad_norm": 0.3067537248134613,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.7274,
      "step": 179
    },
    {
      "epoch": 0.30612244897959184,
      "grad_norm": 0.3485519587993622,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.8132,
      "step": 180
    },
    {
      "epoch": 0.3078231292517007,
      "grad_norm": 0.37323057651519775,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.7217,
      "step": 181
    },
    {
      "epoch": 0.30952380952380953,
      "grad_norm": 0.29564332962036133,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.7569,
      "step": 182
    },
    {
      "epoch": 0.3112244897959184,
      "grad_norm": 0.3461090326309204,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.8656,
      "step": 183
    },
    {
      "epoch": 0.3129251700680272,
      "grad_norm": 0.3658357560634613,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.7307,
      "step": 184
    },
    {
      "epoch": 0.31462585034013607,
      "grad_norm": 0.3515385091304779,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.7559,
      "step": 185
    },
    {
      "epoch": 0.3163265306122449,
      "grad_norm": 0.41622504591941833,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.7898,
      "step": 186
    },
    {
      "epoch": 0.31802721088435376,
      "grad_norm": 0.35886716842651367,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.76,
      "step": 187
    },
    {
      "epoch": 0.3197278911564626,
      "grad_norm": 0.30330008268356323,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.6006,
      "step": 188
    },
    {
      "epoch": 0.32142857142857145,
      "grad_norm": 0.31331560015678406,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.8091,
      "step": 189
    },
    {
      "epoch": 0.3231292517006803,
      "grad_norm": 0.2833898961544037,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.7019,
      "step": 190
    },
    {
      "epoch": 0.32482993197278914,
      "grad_norm": 0.3017280697822571,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.7392,
      "step": 191
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 0.2699677050113678,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.664,
      "step": 192
    },
    {
      "epoch": 0.3282312925170068,
      "grad_norm": 0.2956189513206482,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.7738,
      "step": 193
    },
    {
      "epoch": 0.3299319727891156,
      "grad_norm": 0.2937477231025696,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.6918,
      "step": 194
    },
    {
      "epoch": 0.33163265306122447,
      "grad_norm": 0.3136081099510193,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.6406,
      "step": 195
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.335477739572525,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.7295,
      "step": 196
    },
    {
      "epoch": 0.33503401360544216,
      "grad_norm": 0.3291155993938446,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.6883,
      "step": 197
    },
    {
      "epoch": 0.336734693877551,
      "grad_norm": 0.3627217710018158,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.5751,
      "step": 198
    },
    {
      "epoch": 0.33843537414965985,
      "grad_norm": 0.43677857518196106,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.6766,
      "step": 199
    },
    {
      "epoch": 0.3401360544217687,
      "grad_norm": 0.47067245841026306,
      "learning_rate": 0.0,
      "loss": 0.6891,
      "step": 200
    },
    {
      "epoch": 0.3401360544217687,
      "eval_loss": 0.6899755597114563,
      "eval_runtime": 127.1971,
      "eval_samples_per_second": 7.783,
      "eval_steps_per_second": 1.95,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.823093889466368e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}