{
  "best_metric": 1.9505361318588257,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.8163265306122449,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004081632653061225,
      "grad_norm": 3.6789000034332275,
      "learning_rate": 1e-05,
      "loss": 2.8167,
      "step": 1
    },
    {
      "epoch": 0.004081632653061225,
      "eval_loss": 3.210695505142212,
      "eval_runtime": 18.3056,
      "eval_samples_per_second": 22.561,
      "eval_steps_per_second": 5.681,
      "step": 1
    },
    {
      "epoch": 0.00816326530612245,
      "grad_norm": 3.684469699859619,
      "learning_rate": 2e-05,
      "loss": 2.4153,
      "step": 2
    },
    {
      "epoch": 0.012244897959183673,
      "grad_norm": 3.655902862548828,
      "learning_rate": 3e-05,
      "loss": 2.74,
      "step": 3
    },
    {
      "epoch": 0.0163265306122449,
      "grad_norm": 3.6202504634857178,
      "learning_rate": 4e-05,
      "loss": 2.5746,
      "step": 4
    },
    {
      "epoch": 0.02040816326530612,
      "grad_norm": 3.588803768157959,
      "learning_rate": 5e-05,
      "loss": 2.3748,
      "step": 5
    },
    {
      "epoch": 0.024489795918367346,
      "grad_norm": 3.5844037532806396,
      "learning_rate": 6e-05,
      "loss": 2.5779,
      "step": 6
    },
    {
      "epoch": 0.02857142857142857,
      "grad_norm": 3.7049906253814697,
      "learning_rate": 7e-05,
      "loss": 2.2273,
      "step": 7
    },
    {
      "epoch": 0.0326530612244898,
      "grad_norm": 3.7319467067718506,
      "learning_rate": 8e-05,
      "loss": 2.1541,
      "step": 8
    },
    {
      "epoch": 0.036734693877551024,
      "grad_norm": 3.831761360168457,
      "learning_rate": 9e-05,
      "loss": 2.2977,
      "step": 9
    },
    {
      "epoch": 0.04081632653061224,
      "grad_norm": 3.496588945388794,
      "learning_rate": 0.0001,
      "loss": 2.245,
      "step": 10
    },
    {
      "epoch": 0.044897959183673466,
      "grad_norm": 3.2340826988220215,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.0979,
      "step": 11
    },
    {
      "epoch": 0.04897959183673469,
      "grad_norm": 2.9853811264038086,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.1769,
      "step": 12
    },
    {
      "epoch": 0.053061224489795916,
      "grad_norm": 2.966261863708496,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.1394,
      "step": 13
    },
    {
      "epoch": 0.05714285714285714,
      "grad_norm": 2.848217248916626,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.026,
      "step": 14
    },
    {
      "epoch": 0.061224489795918366,
      "grad_norm": 3.0951640605926514,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.2256,
      "step": 15
    },
    {
      "epoch": 0.0653061224489796,
      "grad_norm": 3.018355131149292,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.0871,
      "step": 16
    },
    {
      "epoch": 0.06938775510204082,
      "grad_norm": 2.919481039047241,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.0776,
      "step": 17
    },
    {
      "epoch": 0.07346938775510205,
      "grad_norm": 2.7707157135009766,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.0685,
      "step": 18
    },
    {
      "epoch": 0.07755102040816327,
      "grad_norm": 3.1321136951446533,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.1922,
      "step": 19
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 3.047600746154785,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.9852,
      "step": 20
    },
    {
      "epoch": 0.08571428571428572,
      "grad_norm": 3.1659388542175293,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.1217,
      "step": 21
    },
    {
      "epoch": 0.08979591836734693,
      "grad_norm": 3.2751541137695312,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.1238,
      "step": 22
    },
    {
      "epoch": 0.09387755102040816,
      "grad_norm": 3.0182857513427734,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.9642,
      "step": 23
    },
    {
      "epoch": 0.09795918367346938,
      "grad_norm": 2.7492477893829346,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.011,
      "step": 24
    },
    {
      "epoch": 0.10204081632653061,
      "grad_norm": 2.752206563949585,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.0494,
      "step": 25
    },
    {
      "epoch": 0.10612244897959183,
      "grad_norm": 2.752488851547241,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.0802,
      "step": 26
    },
    {
      "epoch": 0.11020408163265306,
      "grad_norm": 2.7888200283050537,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.0808,
      "step": 27
    },
    {
      "epoch": 0.11428571428571428,
      "grad_norm": 3.1208605766296387,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.0811,
      "step": 28
    },
    {
      "epoch": 0.11836734693877551,
      "grad_norm": 2.9246163368225098,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.0366,
      "step": 29
    },
    {
      "epoch": 0.12244897959183673,
      "grad_norm": 2.9753007888793945,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.0554,
      "step": 30
    },
    {
      "epoch": 0.12653061224489795,
      "grad_norm": 2.928884267807007,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.1449,
      "step": 31
    },
    {
      "epoch": 0.1306122448979592,
      "grad_norm": 2.882300853729248,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.0154,
      "step": 32
    },
    {
      "epoch": 0.1346938775510204,
      "grad_norm": 3.268815755844116,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.0103,
      "step": 33
    },
    {
      "epoch": 0.13877551020408163,
      "grad_norm": 2.8984081745147705,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.9235,
      "step": 34
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 2.6884477138519287,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.9561,
      "step": 35
    },
    {
      "epoch": 0.1469387755102041,
      "grad_norm": 2.773634910583496,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.9172,
      "step": 36
    },
    {
      "epoch": 0.1510204081632653,
      "grad_norm": 2.7971017360687256,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.9637,
      "step": 37
    },
    {
      "epoch": 0.15510204081632653,
      "grad_norm": 3.950688123703003,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.0465,
      "step": 38
    },
    {
      "epoch": 0.15918367346938775,
      "grad_norm": 3.422671318054199,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.0933,
      "step": 39
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 3.50972056388855,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.9924,
      "step": 40
    },
    {
      "epoch": 0.1673469387755102,
      "grad_norm": 3.0816891193389893,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.1197,
      "step": 41
    },
    {
      "epoch": 0.17142857142857143,
      "grad_norm": 3.1695454120635986,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.0399,
      "step": 42
    },
    {
      "epoch": 0.17551020408163265,
      "grad_norm": 2.9866249561309814,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.855,
      "step": 43
    },
    {
      "epoch": 0.17959183673469387,
      "grad_norm": 2.940147876739502,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.1183,
      "step": 44
    },
    {
      "epoch": 0.1836734693877551,
      "grad_norm": 3.456336259841919,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.1747,
      "step": 45
    },
    {
      "epoch": 0.18775510204081633,
      "grad_norm": 3.18509840965271,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.1518,
      "step": 46
    },
    {
      "epoch": 0.19183673469387755,
      "grad_norm": 4.01165771484375,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.1929,
      "step": 47
    },
    {
      "epoch": 0.19591836734693877,
      "grad_norm": 3.3220551013946533,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.3228,
      "step": 48
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.385434627532959,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.1331,
      "step": 49
    },
    {
      "epoch": 0.20408163265306123,
      "grad_norm": 4.125483512878418,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.1296,
      "step": 50
    },
    {
      "epoch": 0.20408163265306123,
      "eval_loss": 2.2572968006134033,
      "eval_runtime": 18.4107,
      "eval_samples_per_second": 22.433,
      "eval_steps_per_second": 5.649,
      "step": 50
    },
    {
      "epoch": 0.20816326530612245,
      "grad_norm": 4.218966484069824,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.227,
      "step": 51
    },
    {
      "epoch": 0.21224489795918366,
      "grad_norm": 3.6228156089782715,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.1094,
      "step": 52
    },
    {
      "epoch": 0.2163265306122449,
      "grad_norm": 2.5655953884124756,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.8853,
      "step": 53
    },
    {
      "epoch": 0.22040816326530613,
      "grad_norm": 1.9767965078353882,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.7629,
      "step": 54
    },
    {
      "epoch": 0.22448979591836735,
      "grad_norm": 2.2300515174865723,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.8235,
      "step": 55
    },
    {
      "epoch": 0.22857142857142856,
      "grad_norm": 2.356839656829834,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.7778,
      "step": 56
    },
    {
      "epoch": 0.23265306122448978,
      "grad_norm": 2.4376683235168457,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.796,
      "step": 57
    },
    {
      "epoch": 0.23673469387755103,
      "grad_norm": 2.6358144283294678,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.0317,
      "step": 58
    },
    {
      "epoch": 0.24081632653061225,
      "grad_norm": 2.4748549461364746,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.1629,
      "step": 59
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 2.625441789627075,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.9798,
      "step": 60
    },
    {
      "epoch": 0.24897959183673468,
      "grad_norm": 2.1958751678466797,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.8895,
      "step": 61
    },
    {
      "epoch": 0.2530612244897959,
      "grad_norm": 2.5427582263946533,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.0609,
      "step": 62
    },
    {
      "epoch": 0.2571428571428571,
      "grad_norm": 2.3777239322662354,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.9472,
      "step": 63
    },
    {
      "epoch": 0.2612244897959184,
      "grad_norm": 2.361562967300415,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.984,
      "step": 64
    },
    {
      "epoch": 0.2653061224489796,
      "grad_norm": 2.377790689468384,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.912,
      "step": 65
    },
    {
      "epoch": 0.2693877551020408,
      "grad_norm": 2.3698973655700684,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.0221,
      "step": 66
    },
    {
      "epoch": 0.27346938775510204,
      "grad_norm": 2.270094871520996,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.8288,
      "step": 67
    },
    {
      "epoch": 0.27755102040816326,
      "grad_norm": 2.254058599472046,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.981,
      "step": 68
    },
    {
      "epoch": 0.2816326530612245,
      "grad_norm": 2.5751991271972656,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.107,
      "step": 69
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 3.247217893600464,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.8935,
      "step": 70
    },
    {
      "epoch": 0.2897959183673469,
      "grad_norm": 2.267015218734741,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.8044,
      "step": 71
    },
    {
      "epoch": 0.2938775510204082,
      "grad_norm": 2.542069435119629,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.9103,
      "step": 72
    },
    {
      "epoch": 0.2979591836734694,
      "grad_norm": 2.367809295654297,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.9915,
      "step": 73
    },
    {
      "epoch": 0.3020408163265306,
      "grad_norm": 2.3319528102874756,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.9277,
      "step": 74
    },
    {
      "epoch": 0.30612244897959184,
      "grad_norm": 2.4607839584350586,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.9147,
      "step": 75
    },
    {
      "epoch": 0.31020408163265306,
      "grad_norm": 2.590803861618042,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.1734,
      "step": 76
    },
    {
      "epoch": 0.3142857142857143,
      "grad_norm": 2.5294189453125,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.8926,
      "step": 77
    },
    {
      "epoch": 0.3183673469387755,
      "grad_norm": 3.961461305618286,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.8666,
      "step": 78
    },
    {
      "epoch": 0.3224489795918367,
      "grad_norm": 2.4877421855926514,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.021,
      "step": 79
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 2.5108823776245117,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.0798,
      "step": 80
    },
    {
      "epoch": 0.3306122448979592,
      "grad_norm": 2.6391265392303467,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.0319,
      "step": 81
    },
    {
      "epoch": 0.3346938775510204,
      "grad_norm": 2.3597116470336914,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.9081,
      "step": 82
    },
    {
      "epoch": 0.33877551020408164,
      "grad_norm": 2.3042685985565186,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.7745,
      "step": 83
    },
    {
      "epoch": 0.34285714285714286,
      "grad_norm": 2.576266288757324,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.007,
      "step": 84
    },
    {
      "epoch": 0.3469387755102041,
      "grad_norm": 2.5641062259674072,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.969,
      "step": 85
    },
    {
      "epoch": 0.3510204081632653,
      "grad_norm": 2.6728813648223877,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.0079,
      "step": 86
    },
    {
      "epoch": 0.3551020408163265,
      "grad_norm": 2.360001802444458,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.9353,
      "step": 87
    },
    {
      "epoch": 0.35918367346938773,
      "grad_norm": 2.9285295009613037,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.9953,
      "step": 88
    },
    {
      "epoch": 0.363265306122449,
      "grad_norm": 2.799461603164673,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.0193,
      "step": 89
    },
    {
      "epoch": 0.3673469387755102,
      "grad_norm": 2.408754348754883,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.9315,
      "step": 90
    },
    {
      "epoch": 0.37142857142857144,
      "grad_norm": 6.137120723724365,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.0991,
      "step": 91
    },
    {
      "epoch": 0.37551020408163266,
      "grad_norm": 3.0358972549438477,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.154,
      "step": 92
    },
    {
      "epoch": 0.3795918367346939,
      "grad_norm": 2.8257830142974854,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.0584,
      "step": 93
    },
    {
      "epoch": 0.3836734693877551,
      "grad_norm": 2.6828041076660156,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.9637,
      "step": 94
    },
    {
      "epoch": 0.3877551020408163,
      "grad_norm": 2.435418128967285,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.7421,
      "step": 95
    },
    {
      "epoch": 0.39183673469387753,
      "grad_norm": 2.78086519241333,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.9335,
      "step": 96
    },
    {
      "epoch": 0.39591836734693875,
      "grad_norm": 2.709322929382324,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.0148,
      "step": 97
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.8029189109802246,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.0737,
      "step": 98
    },
    {
      "epoch": 0.40408163265306124,
      "grad_norm": 3.100375175476074,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.1987,
      "step": 99
    },
    {
      "epoch": 0.40816326530612246,
      "grad_norm": 3.068387269973755,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.0296,
      "step": 100
    },
    {
      "epoch": 0.40816326530612246,
      "eval_loss": 2.0559909343719482,
      "eval_runtime": 18.4037,
      "eval_samples_per_second": 22.441,
      "eval_steps_per_second": 5.651,
      "step": 100
    },
    {
      "epoch": 0.4122448979591837,
      "grad_norm": 2.3000049591064453,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.7501,
      "step": 101
    },
    {
      "epoch": 0.4163265306122449,
      "grad_norm": 2.3735408782958984,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.9529,
      "step": 102
    },
    {
      "epoch": 0.4204081632653061,
      "grad_norm": 2.064789295196533,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.7665,
      "step": 103
    },
    {
      "epoch": 0.42448979591836733,
      "grad_norm": 2.6209146976470947,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.1263,
      "step": 104
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 1.8736414909362793,
      "learning_rate": 5e-05,
      "loss": 1.7112,
      "step": 105
    },
    {
      "epoch": 0.4326530612244898,
      "grad_norm": 2.061260938644409,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.9565,
      "step": 106
    },
    {
      "epoch": 0.43673469387755104,
      "grad_norm": 2.093158006668091,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.8642,
      "step": 107
    },
    {
      "epoch": 0.44081632653061226,
      "grad_norm": 2.1666319370269775,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.8207,
      "step": 108
    },
    {
      "epoch": 0.4448979591836735,
      "grad_norm": 2.077009677886963,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.6903,
      "step": 109
    },
    {
      "epoch": 0.4489795918367347,
      "grad_norm": 2.0196053981781006,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.7871,
      "step": 110
    },
    {
      "epoch": 0.4530612244897959,
      "grad_norm": 2.1568984985351562,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.8875,
      "step": 111
    },
    {
      "epoch": 0.45714285714285713,
      "grad_norm": 2.2310187816619873,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.9676,
      "step": 112
    },
    {
      "epoch": 0.46122448979591835,
      "grad_norm": 2.1009225845336914,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.947,
      "step": 113
    },
    {
      "epoch": 0.46530612244897956,
      "grad_norm": 2.098377227783203,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.8617,
      "step": 114
    },
    {
      "epoch": 0.46938775510204084,
      "grad_norm": 2.551858901977539,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.1065,
      "step": 115
    },
    {
      "epoch": 0.47346938775510206,
      "grad_norm": 2.151590585708618,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.8597,
      "step": 116
    },
    {
      "epoch": 0.4775510204081633,
      "grad_norm": 2.2898778915405273,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.9697,
      "step": 117
    },
    {
      "epoch": 0.4816326530612245,
      "grad_norm": 2.2017312049865723,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.0406,
      "step": 118
    },
    {
      "epoch": 0.4857142857142857,
      "grad_norm": 1.9686689376831055,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.8614,
      "step": 119
    },
    {
      "epoch": 0.4897959183673469,
      "grad_norm": 2.0317113399505615,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.8346,
      "step": 120
    },
    {
      "epoch": 0.49387755102040815,
      "grad_norm": 2.0290732383728027,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.8742,
      "step": 121
    },
    {
      "epoch": 0.49795918367346936,
      "grad_norm": 2.1589717864990234,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.9403,
      "step": 122
    },
    {
      "epoch": 0.5020408163265306,
      "grad_norm": 2.178767681121826,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.9272,
      "step": 123
    },
    {
      "epoch": 0.5061224489795918,
      "grad_norm": 2.227872848510742,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.9365,
      "step": 124
    },
    {
      "epoch": 0.5102040816326531,
      "grad_norm": 2.3334078788757324,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.9581,
      "step": 125
    },
    {
      "epoch": 0.5142857142857142,
      "grad_norm": 2.417819023132324,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.0406,
      "step": 126
    },
    {
      "epoch": 0.5183673469387755,
      "grad_norm": 2.3735945224761963,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.9759,
      "step": 127
    },
    {
      "epoch": 0.5224489795918368,
      "grad_norm": 2.378300428390503,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.0378,
      "step": 128
    },
    {
      "epoch": 0.5265306122448979,
      "grad_norm": 2.266714572906494,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.7289,
      "step": 129
    },
    {
      "epoch": 0.5306122448979592,
      "grad_norm": 2.1906001567840576,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.9047,
      "step": 130
    },
    {
      "epoch": 0.5346938775510204,
      "grad_norm": 2.3547041416168213,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.7346,
      "step": 131
    },
    {
      "epoch": 0.5387755102040817,
      "grad_norm": 2.500988483428955,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.8472,
      "step": 132
    },
    {
      "epoch": 0.5428571428571428,
      "grad_norm": 2.3619797229766846,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.816,
      "step": 133
    },
    {
      "epoch": 0.5469387755102041,
      "grad_norm": 2.3436827659606934,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.9699,
      "step": 134
    },
    {
      "epoch": 0.5510204081632653,
      "grad_norm": 2.3023483753204346,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.892,
      "step": 135
    },
    {
      "epoch": 0.5551020408163265,
      "grad_norm": 2.3355298042297363,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.9183,
      "step": 136
    },
    {
      "epoch": 0.5591836734693878,
      "grad_norm": 2.3187716007232666,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.9059,
      "step": 137
    },
    {
      "epoch": 0.563265306122449,
      "grad_norm": 2.4979822635650635,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.9173,
      "step": 138
    },
    {
      "epoch": 0.5673469387755102,
      "grad_norm": 2.6041839122772217,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.067,
      "step": 139
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 2.4150474071502686,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.8418,
      "step": 140
    },
    {
      "epoch": 0.5755102040816327,
      "grad_norm": 2.300887107849121,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.8603,
      "step": 141
    },
    {
      "epoch": 0.5795918367346938,
      "grad_norm": 2.7207343578338623,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.1383,
      "step": 142
    },
    {
      "epoch": 0.5836734693877551,
      "grad_norm": 2.454503297805786,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.0441,
      "step": 143
    },
    {
      "epoch": 0.5877551020408164,
      "grad_norm": 2.73113751411438,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.0183,
      "step": 144
    },
    {
      "epoch": 0.5918367346938775,
      "grad_norm": 2.6498947143554688,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.01,
      "step": 145
    },
    {
      "epoch": 0.5959183673469388,
      "grad_norm": 2.7176923751831055,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.0281,
      "step": 146
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.631112575531006,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.9935,
      "step": 147
    },
    {
      "epoch": 0.6040816326530613,
      "grad_norm": 3.0244274139404297,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.265,
      "step": 148
    },
    {
      "epoch": 0.6081632653061224,
      "grad_norm": 3.076306104660034,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.1485,
      "step": 149
    },
    {
      "epoch": 0.6122448979591837,
      "grad_norm": 3.135551691055298,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.0589,
      "step": 150
    },
    {
      "epoch": 0.6122448979591837,
      "eval_loss": 1.9783351421356201,
      "eval_runtime": 18.3985,
      "eval_samples_per_second": 22.447,
      "eval_steps_per_second": 5.653,
      "step": 150
    },
    {
      "epoch": 0.6163265306122448,
      "grad_norm": 1.955858826637268,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.9187,
      "step": 151
    },
    {
      "epoch": 0.6204081632653061,
      "grad_norm": 2.0378377437591553,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.8139,
      "step": 152
    },
    {
      "epoch": 0.6244897959183674,
      "grad_norm": 2.14668607711792,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.9795,
      "step": 153
    },
    {
      "epoch": 0.6285714285714286,
      "grad_norm": 2.1166129112243652,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.9859,
      "step": 154
    },
    {
      "epoch": 0.6326530612244898,
      "grad_norm": 2.075705051422119,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.9211,
      "step": 155
    },
    {
      "epoch": 0.636734693877551,
      "grad_norm": 2.0989394187927246,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.9829,
      "step": 156
    },
    {
      "epoch": 0.6408163265306123,
      "grad_norm": 2.0178780555725098,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.8917,
      "step": 157
    },
    {
      "epoch": 0.6448979591836734,
      "grad_norm": 2.1808199882507324,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.0833,
      "step": 158
    },
    {
      "epoch": 0.6489795918367347,
      "grad_norm": 2.24410080909729,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.9879,
      "step": 159
    },
    {
      "epoch": 0.6530612244897959,
      "grad_norm": 1.9626398086547852,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.8779,
      "step": 160
    },
    {
      "epoch": 0.6571428571428571,
      "grad_norm": 2.096698522567749,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.8731,
      "step": 161
    },
    {
      "epoch": 0.6612244897959184,
      "grad_norm": 2.1191563606262207,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.0416,
      "step": 162
    },
    {
      "epoch": 0.6653061224489796,
      "grad_norm": 2.0757346153259277,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.823,
      "step": 163
    },
    {
      "epoch": 0.6693877551020408,
      "grad_norm": 2.017198085784912,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.8042,
      "step": 164
    },
    {
      "epoch": 0.673469387755102,
      "grad_norm": 2.0933871269226074,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.9693,
      "step": 165
    },
    {
      "epoch": 0.6775510204081633,
      "grad_norm": 2.2173516750335693,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.0697,
      "step": 166
    },
    {
      "epoch": 0.6816326530612244,
      "grad_norm": 1.891088843345642,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.7225,
      "step": 167
    },
    {
      "epoch": 0.6857142857142857,
      "grad_norm": 1.9167189598083496,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.6583,
      "step": 168
    },
    {
      "epoch": 0.689795918367347,
      "grad_norm": 2.224860429763794,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.0403,
      "step": 169
    },
    {
      "epoch": 0.6938775510204082,
      "grad_norm": 2.0640337467193604,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.7965,
      "step": 170
    },
    {
      "epoch": 0.6979591836734694,
      "grad_norm": 2.1124682426452637,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.8304,
      "step": 171
    },
    {
      "epoch": 0.7020408163265306,
      "grad_norm": 2.1509499549865723,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.9436,
      "step": 172
    },
    {
      "epoch": 0.7061224489795919,
      "grad_norm": 2.3938586711883545,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.079,
      "step": 173
    },
    {
      "epoch": 0.710204081632653,
      "grad_norm": 2.048996686935425,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.7752,
      "step": 174
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 2.1311025619506836,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.8363,
      "step": 175
    },
    {
      "epoch": 0.7183673469387755,
      "grad_norm": 2.100555896759033,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.7995,
      "step": 176
    },
    {
      "epoch": 0.7224489795918367,
      "grad_norm": 2.238762140274048,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.8128,
      "step": 177
    },
    {
      "epoch": 0.726530612244898,
      "grad_norm": 2.4262523651123047,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.8977,
      "step": 178
    },
    {
      "epoch": 0.7306122448979592,
      "grad_norm": 2.3008034229278564,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.9233,
      "step": 179
    },
    {
      "epoch": 0.7346938775510204,
      "grad_norm": 2.2035138607025146,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.9273,
      "step": 180
    },
    {
      "epoch": 0.7387755102040816,
      "grad_norm": 2.267024517059326,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.97,
      "step": 181
    },
    {
      "epoch": 0.7428571428571429,
      "grad_norm": 2.276695489883423,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.0129,
      "step": 182
    },
    {
      "epoch": 0.746938775510204,
      "grad_norm": 2.3708064556121826,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.0156,
      "step": 183
    },
    {
      "epoch": 0.7510204081632653,
      "grad_norm": 2.2846715450286865,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.9481,
      "step": 184
    },
    {
      "epoch": 0.7551020408163265,
      "grad_norm": 2.314390182495117,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.8879,
      "step": 185
    },
    {
      "epoch": 0.7591836734693878,
      "grad_norm": 2.4787652492523193,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.9558,
      "step": 186
    },
    {
      "epoch": 0.763265306122449,
      "grad_norm": 2.526646852493286,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.99,
      "step": 187
    },
    {
      "epoch": 0.7673469387755102,
      "grad_norm": 2.5436453819274902,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.195,
      "step": 188
    },
    {
      "epoch": 0.7714285714285715,
      "grad_norm": 2.4905717372894287,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.9338,
      "step": 189
    },
    {
      "epoch": 0.7755102040816326,
      "grad_norm": 2.830731153488159,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.8599,
      "step": 190
    },
    {
      "epoch": 0.7795918367346939,
      "grad_norm": 2.484801769256592,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.9787,
      "step": 191
    },
    {
      "epoch": 0.7836734693877551,
      "grad_norm": 2.8095600605010986,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.0859,
      "step": 192
    },
    {
      "epoch": 0.7877551020408163,
      "grad_norm": 2.7070090770721436,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.9888,
      "step": 193
    },
    {
      "epoch": 0.7918367346938775,
      "grad_norm": 2.8377037048339844,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.1436,
      "step": 194
    },
    {
      "epoch": 0.7959183673469388,
      "grad_norm": 2.7143778800964355,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.0589,
      "step": 195
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.522444486618042,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.9077,
      "step": 196
    },
    {
      "epoch": 0.8040816326530612,
      "grad_norm": 2.848640203475952,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.829,
      "step": 197
    },
    {
      "epoch": 0.8081632653061225,
      "grad_norm": 2.8264479637145996,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.9952,
      "step": 198
    },
    {
      "epoch": 0.8122448979591836,
      "grad_norm": 4.156278133392334,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.3854,
      "step": 199
    },
    {
      "epoch": 0.8163265306122449,
      "grad_norm": 3.49161696434021,
      "learning_rate": 0.0,
      "loss": 2.1724,
      "step": 200
    },
    {
      "epoch": 0.8163265306122449,
      "eval_loss": 1.9505361318588257,
      "eval_runtime": 18.4125,
      "eval_samples_per_second": 22.43,
      "eval_steps_per_second": 5.648,
      "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": 8.28736052133888e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}