{
  "best_metric": 1.7463113069534302,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.01861374159473231,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.306870797366156e-05,
      "grad_norm": 5.207387447357178,
      "learning_rate": 1e-05,
      "loss": 4.2325,
      "step": 1
    },
    {
      "epoch": 9.306870797366156e-05,
      "eval_loss": 5.123601913452148,
      "eval_runtime": 1312.4272,
      "eval_samples_per_second": 13.789,
      "eval_steps_per_second": 3.448,
      "step": 1
    },
    {
      "epoch": 0.00018613741594732312,
      "grad_norm": 5.114199638366699,
      "learning_rate": 2e-05,
      "loss": 4.9671,
      "step": 2
    },
    {
      "epoch": 0.0002792061239209847,
      "grad_norm": 11.131013870239258,
      "learning_rate": 3e-05,
      "loss": 4.927,
      "step": 3
    },
    {
      "epoch": 0.00037227483189464623,
      "grad_norm": 5.8381476402282715,
      "learning_rate": 4e-05,
      "loss": 4.7925,
      "step": 4
    },
    {
      "epoch": 0.0004653435398683078,
      "grad_norm": 5.491620063781738,
      "learning_rate": 5e-05,
      "loss": 4.4039,
      "step": 5
    },
    {
      "epoch": 0.0005584122478419694,
      "grad_norm": 5.235612392425537,
      "learning_rate": 6e-05,
      "loss": 4.4379,
      "step": 6
    },
    {
      "epoch": 0.0006514809558156309,
      "grad_norm": 4.4761223793029785,
      "learning_rate": 7e-05,
      "loss": 3.226,
      "step": 7
    },
    {
      "epoch": 0.0007445496637892925,
      "grad_norm": 3.6422791481018066,
      "learning_rate": 8e-05,
      "loss": 3.0557,
      "step": 8
    },
    {
      "epoch": 0.000837618371762954,
      "grad_norm": 3.777402400970459,
      "learning_rate": 9e-05,
      "loss": 2.7272,
      "step": 9
    },
    {
      "epoch": 0.0009306870797366156,
      "grad_norm": 4.208980083465576,
      "learning_rate": 0.0001,
      "loss": 2.2452,
      "step": 10
    },
    {
      "epoch": 0.001023755787710277,
      "grad_norm": 3.99601674079895,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.7339,
      "step": 11
    },
    {
      "epoch": 0.0011168244956839388,
      "grad_norm": 3.705526113510132,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.4497,
      "step": 12
    },
    {
      "epoch": 0.0012098932036576002,
      "grad_norm": 5.578863143920898,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.3502,
      "step": 13
    },
    {
      "epoch": 0.0013029619116312618,
      "grad_norm": 2.7705371379852295,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.887,
      "step": 14
    },
    {
      "epoch": 0.0013960306196049233,
      "grad_norm": 2.6053831577301025,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.7888,
      "step": 15
    },
    {
      "epoch": 0.001489099327578585,
      "grad_norm": 2.449625015258789,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.1012,
      "step": 16
    },
    {
      "epoch": 0.0015821680355522464,
      "grad_norm": 2.1639928817749023,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.8619,
      "step": 17
    },
    {
      "epoch": 0.001675236743525908,
      "grad_norm": 1.8968380689620972,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.7503,
      "step": 18
    },
    {
      "epoch": 0.0017683054514995695,
      "grad_norm": 1.9328277111053467,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.9054,
      "step": 19
    },
    {
      "epoch": 0.0018613741594732311,
      "grad_norm": 1.9170423746109009,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.007,
      "step": 20
    },
    {
      "epoch": 0.0019544428674468928,
      "grad_norm": 1.7488279342651367,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.9752,
      "step": 21
    },
    {
      "epoch": 0.002047511575420554,
      "grad_norm": 1.788096308708191,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.9762,
      "step": 22
    },
    {
      "epoch": 0.0021405802833942156,
      "grad_norm": 1.6053792238235474,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.0492,
      "step": 23
    },
    {
      "epoch": 0.0022336489913678775,
      "grad_norm": 1.8559181690216064,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.8378,
      "step": 24
    },
    {
      "epoch": 0.002326717699341539,
      "grad_norm": 1.677232265472412,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.0239,
      "step": 25
    },
    {
      "epoch": 0.0024197864073152004,
      "grad_norm": 1.486190676689148,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.8766,
      "step": 26
    },
    {
      "epoch": 0.002512855115288862,
      "grad_norm": 1.5576579570770264,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.7009,
      "step": 27
    },
    {
      "epoch": 0.0026059238232625237,
      "grad_norm": 1.5862576961517334,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.0082,
      "step": 28
    },
    {
      "epoch": 0.002698992531236185,
      "grad_norm": 2.048056125640869,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.838,
      "step": 29
    },
    {
      "epoch": 0.0027920612392098466,
      "grad_norm": 1.6328774690628052,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.7657,
      "step": 30
    },
    {
      "epoch": 0.0028851299471835084,
      "grad_norm": 1.4930105209350586,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.6227,
      "step": 31
    },
    {
      "epoch": 0.00297819865515717,
      "grad_norm": 1.784622311592102,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.6952,
      "step": 32
    },
    {
      "epoch": 0.0030712673631308313,
      "grad_norm": 1.6426860094070435,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.8683,
      "step": 33
    },
    {
      "epoch": 0.0031643360711044928,
      "grad_norm": 1.7322124242782593,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.888,
      "step": 34
    },
    {
      "epoch": 0.0032574047790781546,
      "grad_norm": 2.0428249835968018,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.7147,
      "step": 35
    },
    {
      "epoch": 0.003350473487051816,
      "grad_norm": 1.4184378385543823,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.7758,
      "step": 36
    },
    {
      "epoch": 0.0034435421950254775,
      "grad_norm": 1.818867802619934,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.0707,
      "step": 37
    },
    {
      "epoch": 0.003536610902999139,
      "grad_norm": 1.587225079536438,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.5689,
      "step": 38
    },
    {
      "epoch": 0.003629679610972801,
      "grad_norm": 1.6806753873825073,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.2385,
      "step": 39
    },
    {
      "epoch": 0.0037227483189464622,
      "grad_norm": 1.712263822555542,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.8868,
      "step": 40
    },
    {
      "epoch": 0.0038158170269201237,
      "grad_norm": 1.796549916267395,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.8235,
      "step": 41
    },
    {
      "epoch": 0.0039088857348937855,
      "grad_norm": 1.7432290315628052,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.9347,
      "step": 42
    },
    {
      "epoch": 0.004001954442867447,
      "grad_norm": 2.0825459957122803,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.8276,
      "step": 43
    },
    {
      "epoch": 0.004095023150841108,
      "grad_norm": 1.9126439094543457,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.9491,
      "step": 44
    },
    {
      "epoch": 0.00418809185881477,
      "grad_norm": 1.721409797668457,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.9496,
      "step": 45
    },
    {
      "epoch": 0.004281160566788431,
      "grad_norm": 2.0497400760650635,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.9325,
      "step": 46
    },
    {
      "epoch": 0.004374229274762093,
      "grad_norm": 1.7738792896270752,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.9584,
      "step": 47
    },
    {
      "epoch": 0.004467297982735755,
      "grad_norm": 1.8696461915969849,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.9787,
      "step": 48
    },
    {
      "epoch": 0.0045603666907094165,
      "grad_norm": 1.9008206129074097,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.7074,
      "step": 49
    },
    {
      "epoch": 0.004653435398683078,
      "grad_norm": 2.8567354679107666,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.1529,
      "step": 50
    },
    {
      "epoch": 0.004653435398683078,
      "eval_loss": 1.895154356956482,
      "eval_runtime": 1318.796,
      "eval_samples_per_second": 13.722,
      "eval_steps_per_second": 3.431,
      "step": 50
    },
    {
      "epoch": 0.004746504106656739,
      "grad_norm": 2.5106000900268555,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.6629,
      "step": 51
    },
    {
      "epoch": 0.004839572814630401,
      "grad_norm": 1.8666657209396362,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.8131,
      "step": 52
    },
    {
      "epoch": 0.004932641522604062,
      "grad_norm": 1.975517749786377,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.8937,
      "step": 53
    },
    {
      "epoch": 0.005025710230577724,
      "grad_norm": 1.793887972831726,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.836,
      "step": 54
    },
    {
      "epoch": 0.005118778938551386,
      "grad_norm": 1.6891287565231323,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.9566,
      "step": 55
    },
    {
      "epoch": 0.005211847646525047,
      "grad_norm": 1.4570671319961548,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.5209,
      "step": 56
    },
    {
      "epoch": 0.005304916354498709,
      "grad_norm": 1.575348138809204,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.6971,
      "step": 57
    },
    {
      "epoch": 0.00539798506247237,
      "grad_norm": 1.690929889678955,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.7764,
      "step": 58
    },
    {
      "epoch": 0.005491053770446032,
      "grad_norm": 1.7088589668273926,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.0965,
      "step": 59
    },
    {
      "epoch": 0.005584122478419693,
      "grad_norm": 1.5418660640716553,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.7433,
      "step": 60
    },
    {
      "epoch": 0.005677191186393355,
      "grad_norm": 1.3604835271835327,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.6691,
      "step": 61
    },
    {
      "epoch": 0.005770259894367017,
      "grad_norm": 1.6632291078567505,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.0291,
      "step": 62
    },
    {
      "epoch": 0.005863328602340678,
      "grad_norm": 1.51374351978302,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.7343,
      "step": 63
    },
    {
      "epoch": 0.00595639731031434,
      "grad_norm": 1.6543316841125488,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.092,
      "step": 64
    },
    {
      "epoch": 0.006049466018288001,
      "grad_norm": 1.4366066455841064,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.6421,
      "step": 65
    },
    {
      "epoch": 0.006142534726261663,
      "grad_norm": 1.6704673767089844,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.1059,
      "step": 66
    },
    {
      "epoch": 0.006235603434235324,
      "grad_norm": 1.408278465270996,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.7576,
      "step": 67
    },
    {
      "epoch": 0.0063286721422089855,
      "grad_norm": 1.5189968347549438,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.1127,
      "step": 68
    },
    {
      "epoch": 0.006421740850182647,
      "grad_norm": 1.4850586652755737,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.6378,
      "step": 69
    },
    {
      "epoch": 0.006514809558156309,
      "grad_norm": 1.3093808889389038,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.4265,
      "step": 70
    },
    {
      "epoch": 0.006607878266129971,
      "grad_norm": 1.2844609022140503,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.7181,
      "step": 71
    },
    {
      "epoch": 0.006700946974103632,
      "grad_norm": 1.2631522417068481,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.6354,
      "step": 72
    },
    {
      "epoch": 0.0067940156820772936,
      "grad_norm": 1.588806390762329,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.1476,
      "step": 73
    },
    {
      "epoch": 0.006887084390050955,
      "grad_norm": 1.418257474899292,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.9464,
      "step": 74
    },
    {
      "epoch": 0.006980153098024616,
      "grad_norm": 1.4015493392944336,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.4929,
      "step": 75
    },
    {
      "epoch": 0.007073221805998278,
      "grad_norm": 1.4926618337631226,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.9594,
      "step": 76
    },
    {
      "epoch": 0.00716629051397194,
      "grad_norm": 1.4897574186325073,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.7255,
      "step": 77
    },
    {
      "epoch": 0.007259359221945602,
      "grad_norm": 1.4887974262237549,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.0148,
      "step": 78
    },
    {
      "epoch": 0.007352427929919263,
      "grad_norm": 1.9028290510177612,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.9502,
      "step": 79
    },
    {
      "epoch": 0.0074454966378929245,
      "grad_norm": 1.2643802165985107,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.6995,
      "step": 80
    },
    {
      "epoch": 0.007538565345866586,
      "grad_norm": 1.452819585800171,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.7471,
      "step": 81
    },
    {
      "epoch": 0.007631634053840247,
      "grad_norm": 1.3838050365447998,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.828,
      "step": 82
    },
    {
      "epoch": 0.007724702761813909,
      "grad_norm": 1.4136393070220947,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.8672,
      "step": 83
    },
    {
      "epoch": 0.007817771469787571,
      "grad_norm": 1.4885083436965942,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.8153,
      "step": 84
    },
    {
      "epoch": 0.007910840177761232,
      "grad_norm": 1.40280020236969,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.6192,
      "step": 85
    },
    {
      "epoch": 0.008003908885734894,
      "grad_norm": 1.5277206897735596,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.7702,
      "step": 86
    },
    {
      "epoch": 0.008096977593708555,
      "grad_norm": 1.5335403680801392,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.0259,
      "step": 87
    },
    {
      "epoch": 0.008190046301682217,
      "grad_norm": 1.6681487560272217,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.8646,
      "step": 88
    },
    {
      "epoch": 0.00828311500965588,
      "grad_norm": 1.5285978317260742,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.6269,
      "step": 89
    },
    {
      "epoch": 0.00837618371762954,
      "grad_norm": 1.6006323099136353,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.0429,
      "step": 90
    },
    {
      "epoch": 0.008469252425603202,
      "grad_norm": 1.5587456226348877,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.685,
      "step": 91
    },
    {
      "epoch": 0.008562321133576863,
      "grad_norm": 1.6037956476211548,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.5068,
      "step": 92
    },
    {
      "epoch": 0.008655389841550525,
      "grad_norm": 1.6179325580596924,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.8092,
      "step": 93
    },
    {
      "epoch": 0.008748458549524185,
      "grad_norm": 1.955496072769165,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.6188,
      "step": 94
    },
    {
      "epoch": 0.008841527257497848,
      "grad_norm": 1.6679211854934692,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.9612,
      "step": 95
    },
    {
      "epoch": 0.00893459596547151,
      "grad_norm": 1.6616119146347046,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.993,
      "step": 96
    },
    {
      "epoch": 0.00902766467344517,
      "grad_norm": 1.4882142543792725,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.6337,
      "step": 97
    },
    {
      "epoch": 0.009120733381418833,
      "grad_norm": 2.1910977363586426,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.7137,
      "step": 98
    },
    {
      "epoch": 0.009213802089392494,
      "grad_norm": 2.2215073108673096,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.6943,
      "step": 99
    },
    {
      "epoch": 0.009306870797366156,
      "grad_norm": 5.864289283752441,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.2214,
      "step": 100
    },
    {
      "epoch": 0.009306870797366156,
      "eval_loss": 1.7882341146469116,
      "eval_runtime": 1318.1193,
      "eval_samples_per_second": 13.729,
      "eval_steps_per_second": 3.433,
      "step": 100
    },
    {
      "epoch": 0.009399939505339816,
      "grad_norm": 1.483485221862793,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.5424,
      "step": 101
    },
    {
      "epoch": 0.009493008213313479,
      "grad_norm": 1.2999660968780518,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.3147,
      "step": 102
    },
    {
      "epoch": 0.009586076921287141,
      "grad_norm": 1.446432113647461,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.5708,
      "step": 103
    },
    {
      "epoch": 0.009679145629260802,
      "grad_norm": 1.3734368085861206,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.2507,
      "step": 104
    },
    {
      "epoch": 0.009772214337234464,
      "grad_norm": 1.6593358516693115,
      "learning_rate": 5e-05,
      "loss": 1.9986,
      "step": 105
    },
    {
      "epoch": 0.009865283045208124,
      "grad_norm": 2.377887725830078,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.3354,
      "step": 106
    },
    {
      "epoch": 0.009958351753181787,
      "grad_norm": 1.229383111000061,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.2883,
      "step": 107
    },
    {
      "epoch": 0.010051420461155447,
      "grad_norm": 2.3325161933898926,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.3863,
      "step": 108
    },
    {
      "epoch": 0.01014448916912911,
      "grad_norm": 1.4902580976486206,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.7693,
      "step": 109
    },
    {
      "epoch": 0.010237557877102772,
      "grad_norm": 1.4714196920394897,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.734,
      "step": 110
    },
    {
      "epoch": 0.010330626585076432,
      "grad_norm": 1.8626681566238403,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.5525,
      "step": 111
    },
    {
      "epoch": 0.010423695293050095,
      "grad_norm": 1.296229362487793,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.5855,
      "step": 112
    },
    {
      "epoch": 0.010516764001023755,
      "grad_norm": 1.5485410690307617,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.983,
      "step": 113
    },
    {
      "epoch": 0.010609832708997418,
      "grad_norm": 1.2655266523361206,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.4546,
      "step": 114
    },
    {
      "epoch": 0.010702901416971078,
      "grad_norm": 1.541294813156128,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.7385,
      "step": 115
    },
    {
      "epoch": 0.01079597012494474,
      "grad_norm": 1.6061317920684814,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.0427,
      "step": 116
    },
    {
      "epoch": 0.010889038832918403,
      "grad_norm": 1.7228378057479858,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.9554,
      "step": 117
    },
    {
      "epoch": 0.010982107540892063,
      "grad_norm": 1.3780680894851685,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.8207,
      "step": 118
    },
    {
      "epoch": 0.011075176248865726,
      "grad_norm": 1.3306137323379517,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.7031,
      "step": 119
    },
    {
      "epoch": 0.011168244956839386,
      "grad_norm": 1.062326431274414,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.3092,
      "step": 120
    },
    {
      "epoch": 0.011261313664813049,
      "grad_norm": 1.190127968788147,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.404,
      "step": 121
    },
    {
      "epoch": 0.01135438237278671,
      "grad_norm": 1.2720649242401123,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.3766,
      "step": 122
    },
    {
      "epoch": 0.011447451080760371,
      "grad_norm": 1.203975796699524,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.4408,
      "step": 123
    },
    {
      "epoch": 0.011540519788734034,
      "grad_norm": 1.2768726348876953,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.7104,
      "step": 124
    },
    {
      "epoch": 0.011633588496707694,
      "grad_norm": 1.4720231294631958,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.9706,
      "step": 125
    },
    {
      "epoch": 0.011726657204681357,
      "grad_norm": 1.5098546743392944,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.7356,
      "step": 126
    },
    {
      "epoch": 0.011819725912655017,
      "grad_norm": 1.4503053426742554,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.5427,
      "step": 127
    },
    {
      "epoch": 0.01191279462062868,
      "grad_norm": 1.446598768234253,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.6832,
      "step": 128
    },
    {
      "epoch": 0.01200586332860234,
      "grad_norm": 1.3927395343780518,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.7382,
      "step": 129
    },
    {
      "epoch": 0.012098932036576002,
      "grad_norm": 1.3049274682998657,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.779,
      "step": 130
    },
    {
      "epoch": 0.012192000744549663,
      "grad_norm": 1.267709732055664,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.6302,
      "step": 131
    },
    {
      "epoch": 0.012285069452523325,
      "grad_norm": 1.3769845962524414,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.7908,
      "step": 132
    },
    {
      "epoch": 0.012378138160496988,
      "grad_norm": 1.4618409872055054,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.0161,
      "step": 133
    },
    {
      "epoch": 0.012471206868470648,
      "grad_norm": 1.4933005571365356,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.8368,
      "step": 134
    },
    {
      "epoch": 0.01256427557644431,
      "grad_norm": 1.4526944160461426,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.7678,
      "step": 135
    },
    {
      "epoch": 0.012657344284417971,
      "grad_norm": 1.45134699344635,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.7374,
      "step": 136
    },
    {
      "epoch": 0.012750412992391633,
      "grad_norm": 1.4074360132217407,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.7355,
      "step": 137
    },
    {
      "epoch": 0.012843481700365294,
      "grad_norm": 1.381651759147644,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.9154,
      "step": 138
    },
    {
      "epoch": 0.012936550408338956,
      "grad_norm": 1.3270832300186157,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.7018,
      "step": 139
    },
    {
      "epoch": 0.013029619116312618,
      "grad_norm": 1.3010722398757935,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.5919,
      "step": 140
    },
    {
      "epoch": 0.013122687824286279,
      "grad_norm": 1.3202168941497803,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.5657,
      "step": 141
    },
    {
      "epoch": 0.013215756532259941,
      "grad_norm": 1.7393354177474976,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.1907,
      "step": 142
    },
    {
      "epoch": 0.013308825240233602,
      "grad_norm": 1.3629181385040283,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.6485,
      "step": 143
    },
    {
      "epoch": 0.013401893948207264,
      "grad_norm": 1.502760648727417,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.9964,
      "step": 144
    },
    {
      "epoch": 0.013494962656180925,
      "grad_norm": 1.5563199520111084,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.0839,
      "step": 145
    },
    {
      "epoch": 0.013588031364154587,
      "grad_norm": 1.4114282131195068,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.5819,
      "step": 146
    },
    {
      "epoch": 0.01368110007212825,
      "grad_norm": 1.5543055534362793,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.9491,
      "step": 147
    },
    {
      "epoch": 0.01377416878010191,
      "grad_norm": 1.4594930410385132,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.2757,
      "step": 148
    },
    {
      "epoch": 0.013867237488075572,
      "grad_norm": 1.8859738111495972,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.8625,
      "step": 149
    },
    {
      "epoch": 0.013960306196049233,
      "grad_norm": 1.8530824184417725,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.856,
      "step": 150
    },
    {
      "epoch": 0.013960306196049233,
      "eval_loss": 1.7509338855743408,
      "eval_runtime": 1318.3415,
      "eval_samples_per_second": 13.727,
      "eval_steps_per_second": 3.432,
      "step": 150
    },
    {
      "epoch": 0.014053374904022895,
      "grad_norm": 1.2731139659881592,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.1452,
      "step": 151
    },
    {
      "epoch": 0.014146443611996556,
      "grad_norm": 1.4139740467071533,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.4442,
      "step": 152
    },
    {
      "epoch": 0.014239512319970218,
      "grad_norm": 1.4596621990203857,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.3637,
      "step": 153
    },
    {
      "epoch": 0.01433258102794388,
      "grad_norm": 1.2959932088851929,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.3565,
      "step": 154
    },
    {
      "epoch": 0.014425649735917541,
      "grad_norm": 1.4797643423080444,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.9072,
      "step": 155
    },
    {
      "epoch": 0.014518718443891203,
      "grad_norm": 1.5254017114639282,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.0478,
      "step": 156
    },
    {
      "epoch": 0.014611787151864864,
      "grad_norm": 1.2887506484985352,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.6207,
      "step": 157
    },
    {
      "epoch": 0.014704855859838526,
      "grad_norm": 1.4283150434494019,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.7568,
      "step": 158
    },
    {
      "epoch": 0.014797924567812187,
      "grad_norm": 1.4353879690170288,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.6969,
      "step": 159
    },
    {
      "epoch": 0.014890993275785849,
      "grad_norm": 1.4240456819534302,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.6948,
      "step": 160
    },
    {
      "epoch": 0.014984061983759511,
      "grad_norm": 1.3843494653701782,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.5348,
      "step": 161
    },
    {
      "epoch": 0.015077130691733172,
      "grad_norm": 1.3916226625442505,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.6416,
      "step": 162
    },
    {
      "epoch": 0.015170199399706834,
      "grad_norm": 1.4566906690597534,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.7614,
      "step": 163
    },
    {
      "epoch": 0.015263268107680495,
      "grad_norm": 1.3425648212432861,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.5419,
      "step": 164
    },
    {
      "epoch": 0.015356336815654157,
      "grad_norm": 1.3166548013687134,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.7445,
      "step": 165
    },
    {
      "epoch": 0.015449405523627818,
      "grad_norm": 1.4770258665084839,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.8934,
      "step": 166
    },
    {
      "epoch": 0.01554247423160148,
      "grad_norm": 1.4499372243881226,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.8445,
      "step": 167
    },
    {
      "epoch": 0.015635542939575142,
      "grad_norm": 1.4570034742355347,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.7429,
      "step": 168
    },
    {
      "epoch": 0.015728611647548803,
      "grad_norm": 1.3112519979476929,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.6166,
      "step": 169
    },
    {
      "epoch": 0.015821680355522463,
      "grad_norm": 1.433048129081726,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.7226,
      "step": 170
    },
    {
      "epoch": 0.015914749063496127,
      "grad_norm": 1.4483195543289185,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.9978,
      "step": 171
    },
    {
      "epoch": 0.016007817771469788,
      "grad_norm": 1.3582394123077393,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.674,
      "step": 172
    },
    {
      "epoch": 0.01610088647944345,
      "grad_norm": 1.192657470703125,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.3831,
      "step": 173
    },
    {
      "epoch": 0.01619395518741711,
      "grad_norm": 1.5846621990203857,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.9037,
      "step": 174
    },
    {
      "epoch": 0.016287023895390773,
      "grad_norm": 1.310686469078064,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.7105,
      "step": 175
    },
    {
      "epoch": 0.016380092603364434,
      "grad_norm": 1.380845546722412,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.7245,
      "step": 176
    },
    {
      "epoch": 0.016473161311338094,
      "grad_norm": 1.2821667194366455,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.5099,
      "step": 177
    },
    {
      "epoch": 0.01656623001931176,
      "grad_norm": 1.5495095252990723,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.0022,
      "step": 178
    },
    {
      "epoch": 0.01665929872728542,
      "grad_norm": 1.4604781866073608,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.0229,
      "step": 179
    },
    {
      "epoch": 0.01675236743525908,
      "grad_norm": 1.3071576356887817,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.528,
      "step": 180
    },
    {
      "epoch": 0.01684543614323274,
      "grad_norm": 1.5683079957962036,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.7324,
      "step": 181
    },
    {
      "epoch": 0.016938504851206404,
      "grad_norm": 1.6445461511611938,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.5118,
      "step": 182
    },
    {
      "epoch": 0.017031573559180065,
      "grad_norm": 1.539658784866333,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.0175,
      "step": 183
    },
    {
      "epoch": 0.017124642267153725,
      "grad_norm": 1.4990078210830688,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.9876,
      "step": 184
    },
    {
      "epoch": 0.01721771097512739,
      "grad_norm": 1.666442632675171,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.7982,
      "step": 185
    },
    {
      "epoch": 0.01731077968310105,
      "grad_norm": 1.583263874053955,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.1048,
      "step": 186
    },
    {
      "epoch": 0.01740384839107471,
      "grad_norm": 1.2608344554901123,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.5226,
      "step": 187
    },
    {
      "epoch": 0.01749691709904837,
      "grad_norm": 1.5084272623062134,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.9098,
      "step": 188
    },
    {
      "epoch": 0.017589985807022035,
      "grad_norm": 1.494791865348816,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.8247,
      "step": 189
    },
    {
      "epoch": 0.017683054514995696,
      "grad_norm": 1.4276938438415527,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.6332,
      "step": 190
    },
    {
      "epoch": 0.017776123222969356,
      "grad_norm": 1.4337233304977417,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.7772,
      "step": 191
    },
    {
      "epoch": 0.01786919193094302,
      "grad_norm": 1.4559675455093384,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.6239,
      "step": 192
    },
    {
      "epoch": 0.01796226063891668,
      "grad_norm": 1.8160728216171265,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.767,
      "step": 193
    },
    {
      "epoch": 0.01805532934689034,
      "grad_norm": 1.5655581951141357,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.8079,
      "step": 194
    },
    {
      "epoch": 0.018148398054864002,
      "grad_norm": 2.16974139213562,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.7176,
      "step": 195
    },
    {
      "epoch": 0.018241466762837666,
      "grad_norm": 1.643149733543396,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.7749,
      "step": 196
    },
    {
      "epoch": 0.018334535470811326,
      "grad_norm": 1.7206534147262573,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.044,
      "step": 197
    },
    {
      "epoch": 0.018427604178784987,
      "grad_norm": 1.5860283374786377,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.8028,
      "step": 198
    },
    {
      "epoch": 0.01852067288675865,
      "grad_norm": 1.635237455368042,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.7463,
      "step": 199
    },
    {
      "epoch": 0.01861374159473231,
      "grad_norm": 2.653092622756958,
      "learning_rate": 0.0,
      "loss": 1.7286,
      "step": 200
    },
    {
      "epoch": 0.01861374159473231,
      "eval_loss": 1.7463113069534302,
      "eval_runtime": 1318.4484,
      "eval_samples_per_second": 13.726,
      "eval_steps_per_second": 3.432,
      "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": 2.69826404646912e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}