{
  "best_metric": 0.3733478784561157,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.095890410958904,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.005479452054794521,
      "grad_norm": 17.7918758392334,
      "learning_rate": 1e-05,
      "loss": 8.3013,
      "step": 1
    },
    {
      "epoch": 0.005479452054794521,
      "eval_loss": 1.8490040302276611,
      "eval_runtime": 45.4625,
      "eval_samples_per_second": 6.775,
      "eval_steps_per_second": 1.694,
      "step": 1
    },
    {
      "epoch": 0.010958904109589041,
      "grad_norm": 11.438009262084961,
      "learning_rate": 2e-05,
      "loss": 8.8417,
      "step": 2
    },
    {
      "epoch": 0.01643835616438356,
      "grad_norm": 14.303898811340332,
      "learning_rate": 3e-05,
      "loss": 9.6507,
      "step": 3
    },
    {
      "epoch": 0.021917808219178082,
      "grad_norm": 16.360658645629883,
      "learning_rate": 4e-05,
      "loss": 8.7899,
      "step": 4
    },
    {
      "epoch": 0.0273972602739726,
      "grad_norm": 14.200637817382812,
      "learning_rate": 5e-05,
      "loss": 9.143,
      "step": 5
    },
    {
      "epoch": 0.03287671232876712,
      "grad_norm": 12.905475616455078,
      "learning_rate": 6e-05,
      "loss": 8.7342,
      "step": 6
    },
    {
      "epoch": 0.038356164383561646,
      "grad_norm": 8.813647270202637,
      "learning_rate": 7e-05,
      "loss": 7.2467,
      "step": 7
    },
    {
      "epoch": 0.043835616438356165,
      "grad_norm": 7.976632118225098,
      "learning_rate": 8e-05,
      "loss": 6.7988,
      "step": 8
    },
    {
      "epoch": 0.049315068493150684,
      "grad_norm": 8.724935531616211,
      "learning_rate": 9e-05,
      "loss": 5.6702,
      "step": 9
    },
    {
      "epoch": 0.0547945205479452,
      "grad_norm": 7.649383544921875,
      "learning_rate": 0.0001,
      "loss": 5.9402,
      "step": 10
    },
    {
      "epoch": 0.06027397260273973,
      "grad_norm": 6.983597278594971,
      "learning_rate": 9.999316524962345e-05,
      "loss": 5.7171,
      "step": 11
    },
    {
      "epoch": 0.06575342465753424,
      "grad_norm": 7.777904033660889,
      "learning_rate": 9.997266286704631e-05,
      "loss": 5.4167,
      "step": 12
    },
    {
      "epoch": 0.07123287671232877,
      "grad_norm": 9.056642532348633,
      "learning_rate": 9.993849845741524e-05,
      "loss": 5.2841,
      "step": 13
    },
    {
      "epoch": 0.07671232876712329,
      "grad_norm": 27.901456832885742,
      "learning_rate": 9.989068136093873e-05,
      "loss": 6.8328,
      "step": 14
    },
    {
      "epoch": 0.0821917808219178,
      "grad_norm": 13.157306671142578,
      "learning_rate": 9.98292246503335e-05,
      "loss": 6.256,
      "step": 15
    },
    {
      "epoch": 0.08767123287671233,
      "grad_norm": 11.654566764831543,
      "learning_rate": 9.975414512725057e-05,
      "loss": 4.5427,
      "step": 16
    },
    {
      "epoch": 0.09315068493150686,
      "grad_norm": 16.894699096679688,
      "learning_rate": 9.966546331768191e-05,
      "loss": 7.1066,
      "step": 17
    },
    {
      "epoch": 0.09863013698630137,
      "grad_norm": 11.487627983093262,
      "learning_rate": 9.956320346634876e-05,
      "loss": 4.5795,
      "step": 18
    },
    {
      "epoch": 0.10410958904109589,
      "grad_norm": 9.62430191040039,
      "learning_rate": 9.944739353007344e-05,
      "loss": 5.7696,
      "step": 19
    },
    {
      "epoch": 0.1095890410958904,
      "grad_norm": 9.206327438354492,
      "learning_rate": 9.931806517013612e-05,
      "loss": 4.9274,
      "step": 20
    },
    {
      "epoch": 0.11506849315068493,
      "grad_norm": 7.153225421905518,
      "learning_rate": 9.917525374361912e-05,
      "loss": 3.4898,
      "step": 21
    },
    {
      "epoch": 0.12054794520547946,
      "grad_norm": 13.710307121276855,
      "learning_rate": 9.901899829374047e-05,
      "loss": 6.2841,
      "step": 22
    },
    {
      "epoch": 0.12602739726027398,
      "grad_norm": 6.4639573097229,
      "learning_rate": 9.884934153917997e-05,
      "loss": 4.5236,
      "step": 23
    },
    {
      "epoch": 0.13150684931506848,
      "grad_norm": 10.647743225097656,
      "learning_rate": 9.86663298624003e-05,
      "loss": 4.3312,
      "step": 24
    },
    {
      "epoch": 0.136986301369863,
      "grad_norm": 6.536044120788574,
      "learning_rate": 9.847001329696653e-05,
      "loss": 4.5427,
      "step": 25
    },
    {
      "epoch": 0.14246575342465753,
      "grad_norm": 5.443116664886475,
      "learning_rate": 9.826044551386744e-05,
      "loss": 3.938,
      "step": 26
    },
    {
      "epoch": 0.14794520547945206,
      "grad_norm": 6.665347099304199,
      "learning_rate": 9.803768380684242e-05,
      "loss": 4.5373,
      "step": 27
    },
    {
      "epoch": 0.15342465753424658,
      "grad_norm": 7.1282572746276855,
      "learning_rate": 9.780178907671789e-05,
      "loss": 4.1605,
      "step": 28
    },
    {
      "epoch": 0.1589041095890411,
      "grad_norm": 8.673735618591309,
      "learning_rate": 9.755282581475769e-05,
      "loss": 4.3192,
      "step": 29
    },
    {
      "epoch": 0.1643835616438356,
      "grad_norm": 8.7310209274292,
      "learning_rate": 9.729086208503174e-05,
      "loss": 4.7612,
      "step": 30
    },
    {
      "epoch": 0.16986301369863013,
      "grad_norm": 10.135071754455566,
      "learning_rate": 9.701596950580806e-05,
      "loss": 5.0488,
      "step": 31
    },
    {
      "epoch": 0.17534246575342466,
      "grad_norm": 7.562803268432617,
      "learning_rate": 9.672822322997305e-05,
      "loss": 3.1319,
      "step": 32
    },
    {
      "epoch": 0.18082191780821918,
      "grad_norm": 7.11598539352417,
      "learning_rate": 9.642770192448536e-05,
      "loss": 4.2384,
      "step": 33
    },
    {
      "epoch": 0.1863013698630137,
      "grad_norm": 7.557225704193115,
      "learning_rate": 9.611448774886924e-05,
      "loss": 4.2195,
      "step": 34
    },
    {
      "epoch": 0.1917808219178082,
      "grad_norm": 12.648408889770508,
      "learning_rate": 9.578866633275288e-05,
      "loss": 5.6305,
      "step": 35
    },
    {
      "epoch": 0.19726027397260273,
      "grad_norm": 9.483174324035645,
      "learning_rate": 9.545032675245813e-05,
      "loss": 4.193,
      "step": 36
    },
    {
      "epoch": 0.20273972602739726,
      "grad_norm": 12.605467796325684,
      "learning_rate": 9.509956150664796e-05,
      "loss": 5.9955,
      "step": 37
    },
    {
      "epoch": 0.20821917808219179,
      "grad_norm": 9.469303131103516,
      "learning_rate": 9.473646649103818e-05,
      "loss": 6.1838,
      "step": 38
    },
    {
      "epoch": 0.2136986301369863,
      "grad_norm": 9.724604606628418,
      "learning_rate": 9.43611409721806e-05,
      "loss": 3.3981,
      "step": 39
    },
    {
      "epoch": 0.2191780821917808,
      "grad_norm": 6.6114044189453125,
      "learning_rate": 9.397368756032445e-05,
      "loss": 4.1157,
      "step": 40
    },
    {
      "epoch": 0.22465753424657534,
      "grad_norm": 8.522857666015625,
      "learning_rate": 9.357421218136386e-05,
      "loss": 6.0396,
      "step": 41
    },
    {
      "epoch": 0.23013698630136986,
      "grad_norm": 7.6315484046936035,
      "learning_rate": 9.316282404787871e-05,
      "loss": 3.157,
      "step": 42
    },
    {
      "epoch": 0.2356164383561644,
      "grad_norm": 7.825123310089111,
      "learning_rate": 9.273963562927695e-05,
      "loss": 4.1319,
      "step": 43
    },
    {
      "epoch": 0.2410958904109589,
      "grad_norm": 13.693933486938477,
      "learning_rate": 9.230476262104677e-05,
      "loss": 3.8697,
      "step": 44
    },
    {
      "epoch": 0.2465753424657534,
      "grad_norm": 12.450105667114258,
      "learning_rate": 9.185832391312644e-05,
      "loss": 4.3434,
      "step": 45
    },
    {
      "epoch": 0.25205479452054796,
      "grad_norm": 20.672178268432617,
      "learning_rate": 9.140044155740101e-05,
      "loss": 7.1262,
      "step": 46
    },
    {
      "epoch": 0.25753424657534246,
      "grad_norm": 15.176887512207031,
      "learning_rate": 9.093124073433463e-05,
      "loss": 5.9313,
      "step": 47
    },
    {
      "epoch": 0.26301369863013696,
      "grad_norm": 11.982175827026367,
      "learning_rate": 9.045084971874738e-05,
      "loss": 6.2673,
      "step": 48
    },
    {
      "epoch": 0.2684931506849315,
      "grad_norm": 11.685881614685059,
      "learning_rate": 8.995939984474624e-05,
      "loss": 4.8815,
      "step": 49
    },
    {
      "epoch": 0.273972602739726,
      "grad_norm": 7.684195041656494,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.3478,
      "step": 50
    },
    {
      "epoch": 0.273972602739726,
      "eval_loss": 0.8194975256919861,
      "eval_runtime": 45.1941,
      "eval_samples_per_second": 6.815,
      "eval_steps_per_second": 1.704,
      "step": 50
    },
    {
      "epoch": 0.27945205479452057,
      "grad_norm": 7.425979137420654,
      "learning_rate": 8.894386393810563e-05,
      "loss": 4.607,
      "step": 51
    },
    {
      "epoch": 0.28493150684931506,
      "grad_norm": 5.817314624786377,
      "learning_rate": 8.842005554284296e-05,
      "loss": 3.4188,
      "step": 52
    },
    {
      "epoch": 0.29041095890410956,
      "grad_norm": 6.273691177368164,
      "learning_rate": 8.788574348801675e-05,
      "loss": 3.8499,
      "step": 53
    },
    {
      "epoch": 0.2958904109589041,
      "grad_norm": 4.588272571563721,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.4864,
      "step": 54
    },
    {
      "epoch": 0.3013698630136986,
      "grad_norm": 6.175556182861328,
      "learning_rate": 8.678619553365659e-05,
      "loss": 3.4496,
      "step": 55
    },
    {
      "epoch": 0.30684931506849317,
      "grad_norm": 6.041647434234619,
      "learning_rate": 8.622126023955446e-05,
      "loss": 3.4513,
      "step": 56
    },
    {
      "epoch": 0.31232876712328766,
      "grad_norm": 6.051809310913086,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.8065,
      "step": 57
    },
    {
      "epoch": 0.3178082191780822,
      "grad_norm": 5.9933085441589355,
      "learning_rate": 8.506183921362443e-05,
      "loss": 3.0325,
      "step": 58
    },
    {
      "epoch": 0.3232876712328767,
      "grad_norm": 7.8386054039001465,
      "learning_rate": 8.44676704559283e-05,
      "loss": 4.0459,
      "step": 59
    },
    {
      "epoch": 0.3287671232876712,
      "grad_norm": 7.559600830078125,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.9939,
      "step": 60
    },
    {
      "epoch": 0.33424657534246577,
      "grad_norm": 5.800102710723877,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.7337,
      "step": 61
    },
    {
      "epoch": 0.33972602739726027,
      "grad_norm": 8.976499557495117,
      "learning_rate": 8.262928807620843e-05,
      "loss": 4.0395,
      "step": 62
    },
    {
      "epoch": 0.3452054794520548,
      "grad_norm": 8.108333587646484,
      "learning_rate": 8.199842702516583e-05,
      "loss": 3.0305,
      "step": 63
    },
    {
      "epoch": 0.3506849315068493,
      "grad_norm": 7.5701165199279785,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.9634,
      "step": 64
    },
    {
      "epoch": 0.3561643835616438,
      "grad_norm": 6.603546619415283,
      "learning_rate": 8.07106356344834e-05,
      "loss": 3.0778,
      "step": 65
    },
    {
      "epoch": 0.36164383561643837,
      "grad_norm": 5.186285495758057,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.3056,
      "step": 66
    },
    {
      "epoch": 0.36712328767123287,
      "grad_norm": 7.2703375816345215,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.7308,
      "step": 67
    },
    {
      "epoch": 0.3726027397260274,
      "grad_norm": 6.352651119232178,
      "learning_rate": 7.871643313414718e-05,
      "loss": 3.3469,
      "step": 68
    },
    {
      "epoch": 0.3780821917808219,
      "grad_norm": 4.730004787445068,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.4582,
      "step": 69
    },
    {
      "epoch": 0.3835616438356164,
      "grad_norm": 7.018665313720703,
      "learning_rate": 7.734740790612136e-05,
      "loss": 3.1767,
      "step": 70
    },
    {
      "epoch": 0.38904109589041097,
      "grad_norm": 4.880072116851807,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.5717,
      "step": 71
    },
    {
      "epoch": 0.39452054794520547,
      "grad_norm": 6.6348724365234375,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.8853,
      "step": 72
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.321679592132568,
      "learning_rate": 7.52382768867422e-05,
      "loss": 3.0309,
      "step": 73
    },
    {
      "epoch": 0.4054794520547945,
      "grad_norm": 5.9580607414245605,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.9534,
      "step": 74
    },
    {
      "epoch": 0.410958904109589,
      "grad_norm": 7.704655170440674,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.8195,
      "step": 75
    },
    {
      "epoch": 0.41643835616438357,
      "grad_norm": 6.245016098022461,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.1046,
      "step": 76
    },
    {
      "epoch": 0.42191780821917807,
      "grad_norm": 6.447231769561768,
      "learning_rate": 7.233044034264034e-05,
      "loss": 3.0456,
      "step": 77
    },
    {
      "epoch": 0.4273972602739726,
      "grad_norm": 6.793859481811523,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.9127,
      "step": 78
    },
    {
      "epoch": 0.4328767123287671,
      "grad_norm": 7.1470255851745605,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.6011,
      "step": 79
    },
    {
      "epoch": 0.4383561643835616,
      "grad_norm": 7.638210296630859,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.8653,
      "step": 80
    },
    {
      "epoch": 0.4438356164383562,
      "grad_norm": 8.20581340789795,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.5194,
      "step": 81
    },
    {
      "epoch": 0.44931506849315067,
      "grad_norm": 7.874269962310791,
      "learning_rate": 6.855986244591104e-05,
      "loss": 3.2452,
      "step": 82
    },
    {
      "epoch": 0.4547945205479452,
      "grad_norm": 8.305530548095703,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.7718,
      "step": 83
    },
    {
      "epoch": 0.4602739726027397,
      "grad_norm": 12.325811386108398,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.7301,
      "step": 84
    },
    {
      "epoch": 0.4657534246575342,
      "grad_norm": 7.521613597869873,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.0855,
      "step": 85
    },
    {
      "epoch": 0.4712328767123288,
      "grad_norm": 6.810982704162598,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.8149,
      "step": 86
    },
    {
      "epoch": 0.4767123287671233,
      "grad_norm": 5.873499393463135,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.8463,
      "step": 87
    },
    {
      "epoch": 0.4821917808219178,
      "grad_norm": 8.883861541748047,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.6975,
      "step": 88
    },
    {
      "epoch": 0.4876712328767123,
      "grad_norm": 7.85122537612915,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.1403,
      "step": 89
    },
    {
      "epoch": 0.4931506849315068,
      "grad_norm": 9.159356117248535,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.6294,
      "step": 90
    },
    {
      "epoch": 0.4986301369863014,
      "grad_norm": 11.891826629638672,
      "learning_rate": 6.147119600233758e-05,
      "loss": 4.7088,
      "step": 91
    },
    {
      "epoch": 0.5041095890410959,
      "grad_norm": 8.85066032409668,
      "learning_rate": 6.066498153718735e-05,
      "loss": 3.2274,
      "step": 92
    },
    {
      "epoch": 0.5095890410958904,
      "grad_norm": 9.487573623657227,
      "learning_rate": 5.985585137257401e-05,
      "loss": 3.4859,
      "step": 93
    },
    {
      "epoch": 0.5150684931506849,
      "grad_norm": 6.103641510009766,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.5251,
      "step": 94
    },
    {
      "epoch": 0.5205479452054794,
      "grad_norm": 5.212517738342285,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.3956,
      "step": 95
    },
    {
      "epoch": 0.5260273972602739,
      "grad_norm": 6.487666130065918,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.2505,
      "step": 96
    },
    {
      "epoch": 0.5315068493150685,
      "grad_norm": 4.33821964263916,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.9735,
      "step": 97
    },
    {
      "epoch": 0.536986301369863,
      "grad_norm": 5.6743621826171875,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.1636,
      "step": 98
    },
    {
      "epoch": 0.5424657534246575,
      "grad_norm": 4.478014945983887,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.4474,
      "step": 99
    },
    {
      "epoch": 0.547945205479452,
      "grad_norm": 4.142763137817383,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.6394,
      "step": 100
    },
    {
      "epoch": 0.547945205479452,
      "eval_loss": 0.5193338990211487,
      "eval_runtime": 45.6069,
      "eval_samples_per_second": 6.753,
      "eval_steps_per_second": 1.688,
      "step": 100
    },
    {
      "epoch": 0.5534246575342465,
      "grad_norm": 4.596429347991943,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.1229,
      "step": 101
    },
    {
      "epoch": 0.5589041095890411,
      "grad_norm": 4.113990783691406,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.4998,
      "step": 102
    },
    {
      "epoch": 0.5643835616438356,
      "grad_norm": 5.312803745269775,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.9848,
      "step": 103
    },
    {
      "epoch": 0.5698630136986301,
      "grad_norm": 4.280724048614502,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.2866,
      "step": 104
    },
    {
      "epoch": 0.5753424657534246,
      "grad_norm": 4.694958209991455,
      "learning_rate": 5e-05,
      "loss": 1.9803,
      "step": 105
    },
    {
      "epoch": 0.5808219178082191,
      "grad_norm": 4.337661266326904,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.1724,
      "step": 106
    },
    {
      "epoch": 0.5863013698630137,
      "grad_norm": 5.090383529663086,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.2949,
      "step": 107
    },
    {
      "epoch": 0.5917808219178082,
      "grad_norm": 4.136338233947754,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.1583,
      "step": 108
    },
    {
      "epoch": 0.5972602739726027,
      "grad_norm": 7.187709331512451,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.1602,
      "step": 109
    },
    {
      "epoch": 0.6027397260273972,
      "grad_norm": 5.790441989898682,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.3462,
      "step": 110
    },
    {
      "epoch": 0.6082191780821918,
      "grad_norm": 3.950591564178467,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.765,
      "step": 111
    },
    {
      "epoch": 0.6136986301369863,
      "grad_norm": 5.5668253898620605,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.5192,
      "step": 112
    },
    {
      "epoch": 0.6191780821917808,
      "grad_norm": 4.116091251373291,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.9688,
      "step": 113
    },
    {
      "epoch": 0.6246575342465753,
      "grad_norm": 3.303114414215088,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.594,
      "step": 114
    },
    {
      "epoch": 0.6301369863013698,
      "grad_norm": 3.47721004486084,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.0963,
      "step": 115
    },
    {
      "epoch": 0.6356164383561644,
      "grad_norm": 3.724168539047241,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.8037,
      "step": 116
    },
    {
      "epoch": 0.6410958904109589,
      "grad_norm": 4.004659175872803,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.9184,
      "step": 117
    },
    {
      "epoch": 0.6465753424657534,
      "grad_norm": 4.6142778396606445,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.1329,
      "step": 118
    },
    {
      "epoch": 0.6520547945205479,
      "grad_norm": 5.693519592285156,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.2582,
      "step": 119
    },
    {
      "epoch": 0.6575342465753424,
      "grad_norm": 4.891607761383057,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.4932,
      "step": 120
    },
    {
      "epoch": 0.663013698630137,
      "grad_norm": 10.162339210510254,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.0571,
      "step": 121
    },
    {
      "epoch": 0.6684931506849315,
      "grad_norm": 4.338406085968018,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.5289,
      "step": 122
    },
    {
      "epoch": 0.673972602739726,
      "grad_norm": 4.223052978515625,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.4907,
      "step": 123
    },
    {
      "epoch": 0.6794520547945205,
      "grad_norm": 7.396172523498535,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.093,
      "step": 124
    },
    {
      "epoch": 0.684931506849315,
      "grad_norm": 5.221099376678467,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.8771,
      "step": 125
    },
    {
      "epoch": 0.6904109589041096,
      "grad_norm": 7.64324951171875,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.0021,
      "step": 126
    },
    {
      "epoch": 0.6958904109589041,
      "grad_norm": 6.441079139709473,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.9577,
      "step": 127
    },
    {
      "epoch": 0.7013698630136986,
      "grad_norm": 7.423572063446045,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.9837,
      "step": 128
    },
    {
      "epoch": 0.7068493150684931,
      "grad_norm": 8.613947868347168,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.1062,
      "step": 129
    },
    {
      "epoch": 0.7123287671232876,
      "grad_norm": 4.238228797912598,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.7656,
      "step": 130
    },
    {
      "epoch": 0.7178082191780822,
      "grad_norm": 4.765012264251709,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.9459,
      "step": 131
    },
    {
      "epoch": 0.7232876712328767,
      "grad_norm": 4.682638645172119,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.391,
      "step": 132
    },
    {
      "epoch": 0.7287671232876712,
      "grad_norm": 5.603198528289795,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.7678,
      "step": 133
    },
    {
      "epoch": 0.7342465753424657,
      "grad_norm": 8.716992378234863,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.9654,
      "step": 134
    },
    {
      "epoch": 0.7397260273972602,
      "grad_norm": 5.090101718902588,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.8412,
      "step": 135
    },
    {
      "epoch": 0.7452054794520548,
      "grad_norm": 15.500847816467285,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.5837,
      "step": 136
    },
    {
      "epoch": 0.7506849315068493,
      "grad_norm": 6.448434352874756,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.3614,
      "step": 137
    },
    {
      "epoch": 0.7561643835616438,
      "grad_norm": 7.797610759735107,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.1721,
      "step": 138
    },
    {
      "epoch": 0.7616438356164383,
      "grad_norm": 7.393301486968994,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.5321,
      "step": 139
    },
    {
      "epoch": 0.7671232876712328,
      "grad_norm": 7.083054065704346,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.1688,
      "step": 140
    },
    {
      "epoch": 0.7726027397260274,
      "grad_norm": 5.309166431427002,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.3836,
      "step": 141
    },
    {
      "epoch": 0.7780821917808219,
      "grad_norm": 4.6967034339904785,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.7332,
      "step": 142
    },
    {
      "epoch": 0.7835616438356164,
      "grad_norm": 4.093077659606934,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.7852,
      "step": 143
    },
    {
      "epoch": 0.7890410958904109,
      "grad_norm": 4.156311511993408,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.4856,
      "step": 144
    },
    {
      "epoch": 0.7945205479452054,
      "grad_norm": 3.1240689754486084,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.3244,
      "step": 145
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.971726655960083,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.3327,
      "step": 146
    },
    {
      "epoch": 0.8054794520547945,
      "grad_norm": 5.024684429168701,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.8613,
      "step": 147
    },
    {
      "epoch": 0.810958904109589,
      "grad_norm": 3.268922805786133,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.4694,
      "step": 148
    },
    {
      "epoch": 0.8164383561643835,
      "grad_norm": 4.612654209136963,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.7357,
      "step": 149
    },
    {
      "epoch": 0.821917808219178,
      "grad_norm": 4.326077461242676,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.8367,
      "step": 150
    },
    {
      "epoch": 0.821917808219178,
      "eval_loss": 0.4166771471500397,
      "eval_runtime": 44.7901,
      "eval_samples_per_second": 6.877,
      "eval_steps_per_second": 1.719,
      "step": 150
    },
    {
      "epoch": 0.8273972602739726,
      "grad_norm": 4.419493675231934,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.7479,
      "step": 151
    },
    {
      "epoch": 0.8328767123287671,
      "grad_norm": 4.18338680267334,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.4463,
      "step": 152
    },
    {
      "epoch": 0.8383561643835616,
      "grad_norm": 4.952473163604736,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.018,
      "step": 153
    },
    {
      "epoch": 0.8438356164383561,
      "grad_norm": 5.12903356552124,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.5896,
      "step": 154
    },
    {
      "epoch": 0.8493150684931506,
      "grad_norm": 3.8325114250183105,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.4944,
      "step": 155
    },
    {
      "epoch": 0.8547945205479452,
      "grad_norm": 4.2050676345825195,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.5034,
      "step": 156
    },
    {
      "epoch": 0.8602739726027397,
      "grad_norm": 4.155175685882568,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.8342,
      "step": 157
    },
    {
      "epoch": 0.8657534246575342,
      "grad_norm": 5.488875389099121,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.2855,
      "step": 158
    },
    {
      "epoch": 0.8712328767123287,
      "grad_norm": 3.545876979827881,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.8454,
      "step": 159
    },
    {
      "epoch": 0.8767123287671232,
      "grad_norm": 3.3381426334381104,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.0274,
      "step": 160
    },
    {
      "epoch": 0.8821917808219178,
      "grad_norm": 2.7987794876098633,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.7089,
      "step": 161
    },
    {
      "epoch": 0.8876712328767123,
      "grad_norm": 3.7850358486175537,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.0865,
      "step": 162
    },
    {
      "epoch": 0.8931506849315068,
      "grad_norm": 3.633331298828125,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.603,
      "step": 163
    },
    {
      "epoch": 0.8986301369863013,
      "grad_norm": 3.7027995586395264,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.5317,
      "step": 164
    },
    {
      "epoch": 0.9041095890410958,
      "grad_norm": 4.084996223449707,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.8023,
      "step": 165
    },
    {
      "epoch": 0.9095890410958904,
      "grad_norm": 3.995504379272461,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.7174,
      "step": 166
    },
    {
      "epoch": 0.915068493150685,
      "grad_norm": 3.251267194747925,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.6239,
      "step": 167
    },
    {
      "epoch": 0.9205479452054794,
      "grad_norm": 2.9762394428253174,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.6132,
      "step": 168
    },
    {
      "epoch": 0.9260273972602739,
      "grad_norm": 3.431431293487549,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.4326,
      "step": 169
    },
    {
      "epoch": 0.9315068493150684,
      "grad_norm": 3.6486892700195312,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.7485,
      "step": 170
    },
    {
      "epoch": 0.936986301369863,
      "grad_norm": 3.8881452083587646,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.2369,
      "step": 171
    },
    {
      "epoch": 0.9424657534246575,
      "grad_norm": 4.1761088371276855,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.6678,
      "step": 172
    },
    {
      "epoch": 0.947945205479452,
      "grad_norm": 4.164760112762451,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.9532,
      "step": 173
    },
    {
      "epoch": 0.9534246575342465,
      "grad_norm": 4.184694766998291,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.5455,
      "step": 174
    },
    {
      "epoch": 0.958904109589041,
      "grad_norm": 3.3591256141662598,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.6262,
      "step": 175
    },
    {
      "epoch": 0.9643835616438357,
      "grad_norm": 3.6442887783050537,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.9719,
      "step": 176
    },
    {
      "epoch": 0.9698630136986301,
      "grad_norm": 3.1420350074768066,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.3151,
      "step": 177
    },
    {
      "epoch": 0.9753424657534246,
      "grad_norm": 2.8509936332702637,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.209,
      "step": 178
    },
    {
      "epoch": 0.9808219178082191,
      "grad_norm": 4.512632369995117,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.9398,
      "step": 179
    },
    {
      "epoch": 0.9863013698630136,
      "grad_norm": 3.7434775829315186,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.7185,
      "step": 180
    },
    {
      "epoch": 0.9917808219178083,
      "grad_norm": 2.7593531608581543,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.7369,
      "step": 181
    },
    {
      "epoch": 0.9972602739726028,
      "grad_norm": 2.2325077056884766,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.3785,
      "step": 182
    },
    {
      "epoch": 1.0027397260273974,
      "grad_norm": 3.4822206497192383,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.6205,
      "step": 183
    },
    {
      "epoch": 1.0082191780821919,
      "grad_norm": 4.383573055267334,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.8946,
      "step": 184
    },
    {
      "epoch": 1.0136986301369864,
      "grad_norm": 3.8309481143951416,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.7482,
      "step": 185
    },
    {
      "epoch": 1.0191780821917809,
      "grad_norm": 3.276615858078003,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.4716,
      "step": 186
    },
    {
      "epoch": 1.0246575342465754,
      "grad_norm": 2.762657642364502,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.3637,
      "step": 187
    },
    {
      "epoch": 1.0301369863013699,
      "grad_norm": 3.1694130897521973,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.5265,
      "step": 188
    },
    {
      "epoch": 1.0356164383561643,
      "grad_norm": 3.03877592086792,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.5109,
      "step": 189
    },
    {
      "epoch": 1.0410958904109588,
      "grad_norm": 2.6092803478240967,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.0861,
      "step": 190
    },
    {
      "epoch": 1.0465753424657533,
      "grad_norm": 2.6065003871917725,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.2288,
      "step": 191
    },
    {
      "epoch": 1.0520547945205478,
      "grad_norm": 2.6849613189697266,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.9436,
      "step": 192
    },
    {
      "epoch": 1.0575342465753426,
      "grad_norm": 2.085724115371704,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.9763,
      "step": 193
    },
    {
      "epoch": 1.063013698630137,
      "grad_norm": 2.7171568870544434,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.4067,
      "step": 194
    },
    {
      "epoch": 1.0684931506849316,
      "grad_norm": 2.5242607593536377,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.2058,
      "step": 195
    },
    {
      "epoch": 1.073972602739726,
      "grad_norm": 2.8951051235198975,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.3486,
      "step": 196
    },
    {
      "epoch": 1.0794520547945206,
      "grad_norm": 3.127889633178711,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.4576,
      "step": 197
    },
    {
      "epoch": 1.084931506849315,
      "grad_norm": 2.652359962463379,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.2039,
      "step": 198
    },
    {
      "epoch": 1.0904109589041096,
      "grad_norm": 2.599365711212158,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.3376,
      "step": 199
    },
    {
      "epoch": 1.095890410958904,
      "grad_norm": 2.9625699520111084,
      "learning_rate": 0.0,
      "loss": 1.3877,
      "step": 200
    },
    {
      "epoch": 1.095890410958904,
      "eval_loss": 0.3733478784561157,
      "eval_runtime": 45.5982,
      "eval_samples_per_second": 6.755,
      "eval_steps_per_second": 1.689,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 6.562373209738445e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}