{
  "best_metric": 2.353553533554077,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.2290294875465216,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001145147437732608,
      "grad_norm": 1.1920368671417236,
      "learning_rate": 1e-05,
      "loss": 1.7224,
      "step": 1
    },
    {
      "epoch": 0.001145147437732608,
      "eval_loss": 3.6989080905914307,
      "eval_runtime": 65.1194,
      "eval_samples_per_second": 22.589,
      "eval_steps_per_second": 5.651,
      "step": 1
    },
    {
      "epoch": 0.002290294875465216,
      "grad_norm": 1.3984811305999756,
      "learning_rate": 2e-05,
      "loss": 1.9717,
      "step": 2
    },
    {
      "epoch": 0.0034354423131978244,
      "grad_norm": 1.4595022201538086,
      "learning_rate": 3e-05,
      "loss": 1.9944,
      "step": 3
    },
    {
      "epoch": 0.004580589750930432,
      "grad_norm": 1.4127777814865112,
      "learning_rate": 4e-05,
      "loss": 2.089,
      "step": 4
    },
    {
      "epoch": 0.005725737188663041,
      "grad_norm": 1.3036189079284668,
      "learning_rate": 5e-05,
      "loss": 2.0604,
      "step": 5
    },
    {
      "epoch": 0.006870884626395649,
      "grad_norm": 1.3643872737884521,
      "learning_rate": 6e-05,
      "loss": 2.0559,
      "step": 6
    },
    {
      "epoch": 0.008016032064128256,
      "grad_norm": 1.3958029747009277,
      "learning_rate": 7e-05,
      "loss": 1.956,
      "step": 7
    },
    {
      "epoch": 0.009161179501860864,
      "grad_norm": 1.3658332824707031,
      "learning_rate": 8e-05,
      "loss": 1.8905,
      "step": 8
    },
    {
      "epoch": 0.010306326939593472,
      "grad_norm": 1.3912417888641357,
      "learning_rate": 9e-05,
      "loss": 2.1969,
      "step": 9
    },
    {
      "epoch": 0.011451474377326082,
      "grad_norm": 1.2657861709594727,
      "learning_rate": 0.0001,
      "loss": 1.9844,
      "step": 10
    },
    {
      "epoch": 0.01259662181505869,
      "grad_norm": 1.3974820375442505,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.3125,
      "step": 11
    },
    {
      "epoch": 0.013741769252791297,
      "grad_norm": 1.3555654287338257,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.202,
      "step": 12
    },
    {
      "epoch": 0.014886916690523905,
      "grad_norm": 1.4771543741226196,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.2342,
      "step": 13
    },
    {
      "epoch": 0.01603206412825651,
      "grad_norm": 1.253482460975647,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.209,
      "step": 14
    },
    {
      "epoch": 0.01717721156598912,
      "grad_norm": 1.3795892000198364,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.1611,
      "step": 15
    },
    {
      "epoch": 0.018322359003721728,
      "grad_norm": 1.3504819869995117,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.2543,
      "step": 16
    },
    {
      "epoch": 0.019467506441454337,
      "grad_norm": 1.4682098627090454,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.3384,
      "step": 17
    },
    {
      "epoch": 0.020612653879186944,
      "grad_norm": 1.4126142263412476,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.3074,
      "step": 18
    },
    {
      "epoch": 0.021757801316919553,
      "grad_norm": 1.498470664024353,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.1771,
      "step": 19
    },
    {
      "epoch": 0.022902948754652163,
      "grad_norm": 1.4822009801864624,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.1938,
      "step": 20
    },
    {
      "epoch": 0.02404809619238477,
      "grad_norm": 1.5245144367218018,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.1559,
      "step": 21
    },
    {
      "epoch": 0.02519324363011738,
      "grad_norm": 1.5489428043365479,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.1203,
      "step": 22
    },
    {
      "epoch": 0.026338391067849985,
      "grad_norm": 1.6767163276672363,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.2803,
      "step": 23
    },
    {
      "epoch": 0.027483538505582595,
      "grad_norm": 1.7486976385116577,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.3883,
      "step": 24
    },
    {
      "epoch": 0.0286286859433152,
      "grad_norm": 1.8233695030212402,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.4714,
      "step": 25
    },
    {
      "epoch": 0.02977383338104781,
      "grad_norm": 1.860403060913086,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.4733,
      "step": 26
    },
    {
      "epoch": 0.030918980818780417,
      "grad_norm": 2.008307456970215,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.6319,
      "step": 27
    },
    {
      "epoch": 0.03206412825651302,
      "grad_norm": 2.029810667037964,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.6611,
      "step": 28
    },
    {
      "epoch": 0.03320927569424564,
      "grad_norm": 2.267798900604248,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.7188,
      "step": 29
    },
    {
      "epoch": 0.03435442313197824,
      "grad_norm": 2.241194009780884,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.4464,
      "step": 30
    },
    {
      "epoch": 0.03549957056971085,
      "grad_norm": 2.4929938316345215,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.6818,
      "step": 31
    },
    {
      "epoch": 0.036644718007443455,
      "grad_norm": 2.8612117767333984,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.8701,
      "step": 32
    },
    {
      "epoch": 0.03778986544517607,
      "grad_norm": 2.8577001094818115,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.7105,
      "step": 33
    },
    {
      "epoch": 0.038935012882908675,
      "grad_norm": 2.9048681259155273,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.7013,
      "step": 34
    },
    {
      "epoch": 0.04008016032064128,
      "grad_norm": 3.3334455490112305,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.8755,
      "step": 35
    },
    {
      "epoch": 0.04122530775837389,
      "grad_norm": 3.2485523223876953,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.8467,
      "step": 36
    },
    {
      "epoch": 0.0423704551961065,
      "grad_norm": 3.1462554931640625,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.6991,
      "step": 37
    },
    {
      "epoch": 0.04351560263383911,
      "grad_norm": 3.066805362701416,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.589,
      "step": 38
    },
    {
      "epoch": 0.04466075007157171,
      "grad_norm": 3.45009446144104,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.6086,
      "step": 39
    },
    {
      "epoch": 0.045805897509304326,
      "grad_norm": 3.630845308303833,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.822,
      "step": 40
    },
    {
      "epoch": 0.04695104494703693,
      "grad_norm": 3.7101216316223145,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.6875,
      "step": 41
    },
    {
      "epoch": 0.04809619238476954,
      "grad_norm": 3.9603166580200195,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.618,
      "step": 42
    },
    {
      "epoch": 0.049241339822502145,
      "grad_norm": 3.8295705318450928,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.8453,
      "step": 43
    },
    {
      "epoch": 0.05038648726023476,
      "grad_norm": 4.498661041259766,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.6142,
      "step": 44
    },
    {
      "epoch": 0.051531634697967364,
      "grad_norm": 4.3905930519104,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.8931,
      "step": 45
    },
    {
      "epoch": 0.05267678213569997,
      "grad_norm": 5.010626792907715,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.9706,
      "step": 46
    },
    {
      "epoch": 0.05382192957343258,
      "grad_norm": 4.599480152130127,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.6548,
      "step": 47
    },
    {
      "epoch": 0.05496707701116519,
      "grad_norm": 5.539959907531738,
      "learning_rate": 9.045084971874738e-05,
      "loss": 3.0943,
      "step": 48
    },
    {
      "epoch": 0.056112224448897796,
      "grad_norm": 6.1842265129089355,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.7171,
      "step": 49
    },
    {
      "epoch": 0.0572573718866304,
      "grad_norm": 15.531563758850098,
      "learning_rate": 8.945702546981969e-05,
      "loss": 3.4802,
      "step": 50
    },
    {
      "epoch": 0.0572573718866304,
      "eval_loss": 2.8572957515716553,
      "eval_runtime": 65.4195,
      "eval_samples_per_second": 22.486,
      "eval_steps_per_second": 5.625,
      "step": 50
    },
    {
      "epoch": 0.05840251932436301,
      "grad_norm": 2.859070062637329,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.0059,
      "step": 51
    },
    {
      "epoch": 0.05954766676209562,
      "grad_norm": 2.8047566413879395,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.1934,
      "step": 52
    },
    {
      "epoch": 0.06069281419982823,
      "grad_norm": 1.9153670072555542,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.1558,
      "step": 53
    },
    {
      "epoch": 0.061837961637560834,
      "grad_norm": 1.4249013662338257,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.1183,
      "step": 54
    },
    {
      "epoch": 0.06298310907529345,
      "grad_norm": 1.234809160232544,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.1326,
      "step": 55
    },
    {
      "epoch": 0.06412825651302605,
      "grad_norm": 1.1947890520095825,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.1988,
      "step": 56
    },
    {
      "epoch": 0.06527340395075866,
      "grad_norm": 1.1203817129135132,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.1225,
      "step": 57
    },
    {
      "epoch": 0.06641855138849127,
      "grad_norm": 1.0745985507965088,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.128,
      "step": 58
    },
    {
      "epoch": 0.06756369882622387,
      "grad_norm": 1.1263383626937866,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.2567,
      "step": 59
    },
    {
      "epoch": 0.06870884626395649,
      "grad_norm": 1.070082187652588,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.0725,
      "step": 60
    },
    {
      "epoch": 0.0698539937016891,
      "grad_norm": 1.1051433086395264,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.9879,
      "step": 61
    },
    {
      "epoch": 0.0709991411394217,
      "grad_norm": 1.1032260656356812,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.9213,
      "step": 62
    },
    {
      "epoch": 0.07214428857715431,
      "grad_norm": 1.1471400260925293,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.0203,
      "step": 63
    },
    {
      "epoch": 0.07328943601488691,
      "grad_norm": 1.1910686492919922,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.1165,
      "step": 64
    },
    {
      "epoch": 0.07443458345261952,
      "grad_norm": 1.345789909362793,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.2556,
      "step": 65
    },
    {
      "epoch": 0.07557973089035214,
      "grad_norm": 1.3378866910934448,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.3452,
      "step": 66
    },
    {
      "epoch": 0.07672487832808474,
      "grad_norm": 1.3747169971466064,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.2314,
      "step": 67
    },
    {
      "epoch": 0.07787002576581735,
      "grad_norm": 1.430034875869751,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.3874,
      "step": 68
    },
    {
      "epoch": 0.07901517320354996,
      "grad_norm": 1.4959536790847778,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.2472,
      "step": 69
    },
    {
      "epoch": 0.08016032064128256,
      "grad_norm": 1.4410640001296997,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.2507,
      "step": 70
    },
    {
      "epoch": 0.08130546807901518,
      "grad_norm": 1.5648914575576782,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.2848,
      "step": 71
    },
    {
      "epoch": 0.08245061551674777,
      "grad_norm": 1.618811011314392,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.5163,
      "step": 72
    },
    {
      "epoch": 0.08359576295448039,
      "grad_norm": 1.6799885034561157,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.389,
      "step": 73
    },
    {
      "epoch": 0.084740910392213,
      "grad_norm": 1.8219566345214844,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.4905,
      "step": 74
    },
    {
      "epoch": 0.0858860578299456,
      "grad_norm": 1.8733477592468262,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.3988,
      "step": 75
    },
    {
      "epoch": 0.08703120526767821,
      "grad_norm": 1.8254079818725586,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.6282,
      "step": 76
    },
    {
      "epoch": 0.08817635270541083,
      "grad_norm": 1.9588936567306519,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.5544,
      "step": 77
    },
    {
      "epoch": 0.08932150014314343,
      "grad_norm": 2.1041502952575684,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.649,
      "step": 78
    },
    {
      "epoch": 0.09046664758087604,
      "grad_norm": 2.1702065467834473,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.6443,
      "step": 79
    },
    {
      "epoch": 0.09161179501860865,
      "grad_norm": 2.4502837657928467,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.663,
      "step": 80
    },
    {
      "epoch": 0.09275694245634125,
      "grad_norm": 2.2540695667266846,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.6443,
      "step": 81
    },
    {
      "epoch": 0.09390208989407386,
      "grad_norm": 2.5956249237060547,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.9245,
      "step": 82
    },
    {
      "epoch": 0.09504723733180646,
      "grad_norm": 2.5950498580932617,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.6324,
      "step": 83
    },
    {
      "epoch": 0.09619238476953908,
      "grad_norm": 2.605154037475586,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.5983,
      "step": 84
    },
    {
      "epoch": 0.09733753220727169,
      "grad_norm": 2.8390369415283203,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.7569,
      "step": 85
    },
    {
      "epoch": 0.09848267964500429,
      "grad_norm": 2.915346145629883,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.7026,
      "step": 86
    },
    {
      "epoch": 0.0996278270827369,
      "grad_norm": 3.138815402984619,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.8381,
      "step": 87
    },
    {
      "epoch": 0.10077297452046952,
      "grad_norm": 3.196565866470337,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.6399,
      "step": 88
    },
    {
      "epoch": 0.10191812195820212,
      "grad_norm": 3.280749559402466,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.7087,
      "step": 89
    },
    {
      "epoch": 0.10306326939593473,
      "grad_norm": 3.1474597454071045,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.6212,
      "step": 90
    },
    {
      "epoch": 0.10420841683366733,
      "grad_norm": 3.85715389251709,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.7165,
      "step": 91
    },
    {
      "epoch": 0.10535356427139994,
      "grad_norm": 3.252838134765625,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.3891,
      "step": 92
    },
    {
      "epoch": 0.10649871170913255,
      "grad_norm": 3.2739152908325195,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.5296,
      "step": 93
    },
    {
      "epoch": 0.10764385914686515,
      "grad_norm": 3.9459891319274902,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.7765,
      "step": 94
    },
    {
      "epoch": 0.10878900658459777,
      "grad_norm": 4.354531288146973,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.6944,
      "step": 95
    },
    {
      "epoch": 0.10993415402233038,
      "grad_norm": 3.6282012462615967,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.4549,
      "step": 96
    },
    {
      "epoch": 0.11107930146006298,
      "grad_norm": 4.309568881988525,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.7927,
      "step": 97
    },
    {
      "epoch": 0.11222444889779559,
      "grad_norm": 4.669059753417969,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.7934,
      "step": 98
    },
    {
      "epoch": 0.1133695963355282,
      "grad_norm": 4.723805904388428,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.5294,
      "step": 99
    },
    {
      "epoch": 0.1145147437732608,
      "grad_norm": 8.176491737365723,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 3.1986,
      "step": 100
    },
    {
      "epoch": 0.1145147437732608,
      "eval_loss": 2.6328210830688477,
      "eval_runtime": 65.4663,
      "eval_samples_per_second": 22.47,
      "eval_steps_per_second": 5.621,
      "step": 100
    },
    {
      "epoch": 0.11565989121099342,
      "grad_norm": 2.0501911640167236,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.7066,
      "step": 101
    },
    {
      "epoch": 0.11680503864872602,
      "grad_norm": 2.434068441390991,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.9807,
      "step": 102
    },
    {
      "epoch": 0.11795018608645863,
      "grad_norm": 1.9924542903900146,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.1546,
      "step": 103
    },
    {
      "epoch": 0.11909533352419124,
      "grad_norm": 1.59428071975708,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.0418,
      "step": 104
    },
    {
      "epoch": 0.12024048096192384,
      "grad_norm": 1.3242534399032593,
      "learning_rate": 5e-05,
      "loss": 1.9606,
      "step": 105
    },
    {
      "epoch": 0.12138562839965646,
      "grad_norm": 1.093428373336792,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.8867,
      "step": 106
    },
    {
      "epoch": 0.12253077583738907,
      "grad_norm": 1.0202025175094604,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.9641,
      "step": 107
    },
    {
      "epoch": 0.12367592327512167,
      "grad_norm": 1.0236577987670898,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.001,
      "step": 108
    },
    {
      "epoch": 0.12482107071285428,
      "grad_norm": 1.0549591779708862,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.0084,
      "step": 109
    },
    {
      "epoch": 0.1259662181505869,
      "grad_norm": 1.060852289199829,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.1009,
      "step": 110
    },
    {
      "epoch": 0.1271113655883195,
      "grad_norm": 1.1274445056915283,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.127,
      "step": 111
    },
    {
      "epoch": 0.1282565130260521,
      "grad_norm": 1.1930686235427856,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.2591,
      "step": 112
    },
    {
      "epoch": 0.1294016604637847,
      "grad_norm": 1.1260312795639038,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.0188,
      "step": 113
    },
    {
      "epoch": 0.13054680790151732,
      "grad_norm": 1.2100647687911987,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.1751,
      "step": 114
    },
    {
      "epoch": 0.13169195533924993,
      "grad_norm": 1.233555555343628,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.1692,
      "step": 115
    },
    {
      "epoch": 0.13283710277698255,
      "grad_norm": 1.2247536182403564,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.1314,
      "step": 116
    },
    {
      "epoch": 0.13398225021471513,
      "grad_norm": 1.3735865354537964,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.2494,
      "step": 117
    },
    {
      "epoch": 0.13512739765244774,
      "grad_norm": 1.3731098175048828,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.3256,
      "step": 118
    },
    {
      "epoch": 0.13627254509018036,
      "grad_norm": 1.3202950954437256,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.0712,
      "step": 119
    },
    {
      "epoch": 0.13741769252791297,
      "grad_norm": 1.3336747884750366,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.079,
      "step": 120
    },
    {
      "epoch": 0.13856283996564558,
      "grad_norm": 1.433763027191162,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.3229,
      "step": 121
    },
    {
      "epoch": 0.1397079874033782,
      "grad_norm": 1.534828782081604,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.1689,
      "step": 122
    },
    {
      "epoch": 0.14085313484111078,
      "grad_norm": 1.5490010976791382,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.3611,
      "step": 123
    },
    {
      "epoch": 0.1419982822788434,
      "grad_norm": 1.7407718896865845,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.5073,
      "step": 124
    },
    {
      "epoch": 0.143143429716576,
      "grad_norm": 1.7451822757720947,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.4768,
      "step": 125
    },
    {
      "epoch": 0.14428857715430862,
      "grad_norm": 1.7998104095458984,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.4813,
      "step": 126
    },
    {
      "epoch": 0.14543372459204124,
      "grad_norm": 1.7736880779266357,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.4303,
      "step": 127
    },
    {
      "epoch": 0.14657887202977382,
      "grad_norm": 2.1772241592407227,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.596,
      "step": 128
    },
    {
      "epoch": 0.14772401946750643,
      "grad_norm": 2.2178280353546143,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.5961,
      "step": 129
    },
    {
      "epoch": 0.14886916690523905,
      "grad_norm": 2.141711711883545,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.4554,
      "step": 130
    },
    {
      "epoch": 0.15001431434297166,
      "grad_norm": 2.382219076156616,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.605,
      "step": 131
    },
    {
      "epoch": 0.15115946178070427,
      "grad_norm": 2.470813512802124,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.6563,
      "step": 132
    },
    {
      "epoch": 0.1523046092184369,
      "grad_norm": 2.446791410446167,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.6634,
      "step": 133
    },
    {
      "epoch": 0.15344975665616947,
      "grad_norm": 2.6642603874206543,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.6763,
      "step": 134
    },
    {
      "epoch": 0.15459490409390209,
      "grad_norm": 2.6905620098114014,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.7467,
      "step": 135
    },
    {
      "epoch": 0.1557400515316347,
      "grad_norm": 2.6977028846740723,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.5506,
      "step": 136
    },
    {
      "epoch": 0.1568851989693673,
      "grad_norm": 2.8402292728424072,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.8719,
      "step": 137
    },
    {
      "epoch": 0.15803034640709993,
      "grad_norm": 2.9274494647979736,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.648,
      "step": 138
    },
    {
      "epoch": 0.1591754938448325,
      "grad_norm": 2.880117416381836,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.8494,
      "step": 139
    },
    {
      "epoch": 0.16032064128256512,
      "grad_norm": 3.2345147132873535,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.8267,
      "step": 140
    },
    {
      "epoch": 0.16146578872029774,
      "grad_norm": 3.5025017261505127,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.6794,
      "step": 141
    },
    {
      "epoch": 0.16261093615803035,
      "grad_norm": 3.119682788848877,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.5615,
      "step": 142
    },
    {
      "epoch": 0.16375608359576296,
      "grad_norm": 3.5150766372680664,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.8336,
      "step": 143
    },
    {
      "epoch": 0.16490123103349555,
      "grad_norm": 8.45030689239502,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.9348,
      "step": 144
    },
    {
      "epoch": 0.16604637847122816,
      "grad_norm": 3.675546646118164,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.7121,
      "step": 145
    },
    {
      "epoch": 0.16719152590896078,
      "grad_norm": 4.131383895874023,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.9267,
      "step": 146
    },
    {
      "epoch": 0.1683366733466934,
      "grad_norm": 3.6341030597686768,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.4231,
      "step": 147
    },
    {
      "epoch": 0.169481820784426,
      "grad_norm": 5.185886383056641,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 3.219,
      "step": 148
    },
    {
      "epoch": 0.17062696822215861,
      "grad_norm": 5.132345199584961,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.569,
      "step": 149
    },
    {
      "epoch": 0.1717721156598912,
      "grad_norm": 9.172524452209473,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 4.5991,
      "step": 150
    },
    {
      "epoch": 0.1717721156598912,
      "eval_loss": 2.3809421062469482,
      "eval_runtime": 65.4659,
      "eval_samples_per_second": 22.47,
      "eval_steps_per_second": 5.621,
      "step": 150
    },
    {
      "epoch": 0.1729172630976238,
      "grad_norm": 0.906279981136322,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.7776,
      "step": 151
    },
    {
      "epoch": 0.17406241053535643,
      "grad_norm": 1.0710793733596802,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.7752,
      "step": 152
    },
    {
      "epoch": 0.17520755797308904,
      "grad_norm": 1.1325005292892456,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.0484,
      "step": 153
    },
    {
      "epoch": 0.17635270541082165,
      "grad_norm": 1.1274832487106323,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.0185,
      "step": 154
    },
    {
      "epoch": 0.17749785284855424,
      "grad_norm": 1.102135181427002,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.8403,
      "step": 155
    },
    {
      "epoch": 0.17864300028628685,
      "grad_norm": 1.0935771465301514,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.1674,
      "step": 156
    },
    {
      "epoch": 0.17978814772401946,
      "grad_norm": 0.9937973618507385,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.8085,
      "step": 157
    },
    {
      "epoch": 0.18093329516175208,
      "grad_norm": 0.9920843839645386,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.9231,
      "step": 158
    },
    {
      "epoch": 0.1820784425994847,
      "grad_norm": 1.0586398839950562,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.0345,
      "step": 159
    },
    {
      "epoch": 0.1832235900372173,
      "grad_norm": 1.0832324028015137,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.1197,
      "step": 160
    },
    {
      "epoch": 0.1843687374749499,
      "grad_norm": 1.1074697971343994,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.0555,
      "step": 161
    },
    {
      "epoch": 0.1855138849126825,
      "grad_norm": 1.1114755868911743,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.1232,
      "step": 162
    },
    {
      "epoch": 0.18665903235041512,
      "grad_norm": 1.151540994644165,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.2332,
      "step": 163
    },
    {
      "epoch": 0.18780417978814773,
      "grad_norm": 1.2318562269210815,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.3171,
      "step": 164
    },
    {
      "epoch": 0.18894932722588034,
      "grad_norm": 1.2227367162704468,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.2778,
      "step": 165
    },
    {
      "epoch": 0.19009447466361293,
      "grad_norm": 1.198326587677002,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.9871,
      "step": 166
    },
    {
      "epoch": 0.19123962210134554,
      "grad_norm": 1.3991447687149048,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.3381,
      "step": 167
    },
    {
      "epoch": 0.19238476953907815,
      "grad_norm": 1.3925669193267822,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.1529,
      "step": 168
    },
    {
      "epoch": 0.19352991697681077,
      "grad_norm": 1.5126734972000122,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.3885,
      "step": 169
    },
    {
      "epoch": 0.19467506441454338,
      "grad_norm": 1.4755884408950806,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.1708,
      "step": 170
    },
    {
      "epoch": 0.195820211852276,
      "grad_norm": 1.6294785737991333,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.3729,
      "step": 171
    },
    {
      "epoch": 0.19696535929000858,
      "grad_norm": 1.5729774236679077,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.3608,
      "step": 172
    },
    {
      "epoch": 0.1981105067277412,
      "grad_norm": 1.6711604595184326,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.243,
      "step": 173
    },
    {
      "epoch": 0.1992556541654738,
      "grad_norm": 1.694813847541809,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.251,
      "step": 174
    },
    {
      "epoch": 0.20040080160320642,
      "grad_norm": 1.7610057592391968,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.1883,
      "step": 175
    },
    {
      "epoch": 0.20154594904093903,
      "grad_norm": 1.905004858970642,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.614,
      "step": 176
    },
    {
      "epoch": 0.20269109647867162,
      "grad_norm": 1.9833762645721436,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.523,
      "step": 177
    },
    {
      "epoch": 0.20383624391640423,
      "grad_norm": 2.0892560482025146,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.5729,
      "step": 178
    },
    {
      "epoch": 0.20498139135413684,
      "grad_norm": 2.169820785522461,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.5836,
      "step": 179
    },
    {
      "epoch": 0.20612653879186946,
      "grad_norm": 2.2675445079803467,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.472,
      "step": 180
    },
    {
      "epoch": 0.20727168622960207,
      "grad_norm": 2.3433666229248047,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.5145,
      "step": 181
    },
    {
      "epoch": 0.20841683366733466,
      "grad_norm": 2.3637661933898926,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.5846,
      "step": 182
    },
    {
      "epoch": 0.20956198110506727,
      "grad_norm": 2.585118532180786,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.6088,
      "step": 183
    },
    {
      "epoch": 0.21070712854279988,
      "grad_norm": 2.700228452682495,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.6961,
      "step": 184
    },
    {
      "epoch": 0.2118522759805325,
      "grad_norm": 2.7650864124298096,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.6333,
      "step": 185
    },
    {
      "epoch": 0.2129974234182651,
      "grad_norm": 2.896402597427368,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.6837,
      "step": 186
    },
    {
      "epoch": 0.21414257085599772,
      "grad_norm": 2.8594272136688232,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.6529,
      "step": 187
    },
    {
      "epoch": 0.2152877182937303,
      "grad_norm": 2.67228627204895,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.3681,
      "step": 188
    },
    {
      "epoch": 0.21643286573146292,
      "grad_norm": 3.1715033054351807,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.6362,
      "step": 189
    },
    {
      "epoch": 0.21757801316919553,
      "grad_norm": 3.1333723068237305,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.5085,
      "step": 190
    },
    {
      "epoch": 0.21872316060692815,
      "grad_norm": 2.865365743637085,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.6215,
      "step": 191
    },
    {
      "epoch": 0.21986830804466076,
      "grad_norm": 3.3263680934906006,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.5116,
      "step": 192
    },
    {
      "epoch": 0.22101345548239335,
      "grad_norm": 3.456552505493164,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.4736,
      "step": 193
    },
    {
      "epoch": 0.22215860292012596,
      "grad_norm": 3.5772175788879395,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.5855,
      "step": 194
    },
    {
      "epoch": 0.22330375035785857,
      "grad_norm": 3.368765115737915,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.5883,
      "step": 195
    },
    {
      "epoch": 0.22444889779559118,
      "grad_norm": 4.1975178718566895,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.9944,
      "step": 196
    },
    {
      "epoch": 0.2255940452333238,
      "grad_norm": 3.9533374309539795,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.985,
      "step": 197
    },
    {
      "epoch": 0.2267391926710564,
      "grad_norm": 4.194900035858154,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.9013,
      "step": 198
    },
    {
      "epoch": 0.227884340108789,
      "grad_norm": 5.727431297302246,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.8366,
      "step": 199
    },
    {
      "epoch": 0.2290294875465216,
      "grad_norm": 9.029935836791992,
      "learning_rate": 0.0,
      "loss": 3.4315,
      "step": 200
    },
    {
      "epoch": 0.2290294875465216,
      "eval_loss": 2.353553533554077,
      "eval_runtime": 65.4596,
      "eval_samples_per_second": 22.472,
      "eval_steps_per_second": 5.622,
      "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
}