{
  "best_metric": 2.2710835310135735e-06,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 1.941747572815534,
  "eval_steps": 50,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.012944983818770227,
      "grad_norm": 1.0300663709640503,
      "learning_rate": 1e-05,
      "loss": 0.1373,
      "step": 1
    },
    {
      "epoch": 0.012944983818770227,
      "eval_loss": 0.2981894910335541,
      "eval_runtime": 9.5829,
      "eval_samples_per_second": 13.566,
      "eval_steps_per_second": 3.444,
      "step": 1
    },
    {
      "epoch": 0.025889967637540454,
      "grad_norm": 1.242613434791565,
      "learning_rate": 2e-05,
      "loss": 0.1639,
      "step": 2
    },
    {
      "epoch": 0.038834951456310676,
      "grad_norm": 1.2580989599227905,
      "learning_rate": 3e-05,
      "loss": 0.1689,
      "step": 3
    },
    {
      "epoch": 0.05177993527508091,
      "grad_norm": 1.0557060241699219,
      "learning_rate": 4e-05,
      "loss": 0.1523,
      "step": 4
    },
    {
      "epoch": 0.06472491909385113,
      "grad_norm": 0.6219015121459961,
      "learning_rate": 5e-05,
      "loss": 0.0967,
      "step": 5
    },
    {
      "epoch": 0.07766990291262135,
      "grad_norm": 0.5410397052764893,
      "learning_rate": 6e-05,
      "loss": 0.0613,
      "step": 6
    },
    {
      "epoch": 0.09061488673139159,
      "grad_norm": 0.6054167151451111,
      "learning_rate": 7e-05,
      "loss": 0.0308,
      "step": 7
    },
    {
      "epoch": 0.10355987055016182,
      "grad_norm": 0.26588791608810425,
      "learning_rate": 8e-05,
      "loss": 0.011,
      "step": 8
    },
    {
      "epoch": 0.11650485436893204,
      "grad_norm": 0.1244489848613739,
      "learning_rate": 9e-05,
      "loss": 0.004,
      "step": 9
    },
    {
      "epoch": 0.12944983818770225,
      "grad_norm": 0.06946226954460144,
      "learning_rate": 0.0001,
      "loss": 0.0012,
      "step": 10
    },
    {
      "epoch": 0.1423948220064725,
      "grad_norm": 0.05331278219819069,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.0008,
      "step": 11
    },
    {
      "epoch": 0.1553398058252427,
      "grad_norm": 0.07835312187671661,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.0004,
      "step": 12
    },
    {
      "epoch": 0.16828478964401294,
      "grad_norm": 0.010071509517729282,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.0001,
      "step": 13
    },
    {
      "epoch": 0.18122977346278318,
      "grad_norm": 0.2724490165710449,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.0015,
      "step": 14
    },
    {
      "epoch": 0.1941747572815534,
      "grad_norm": 0.005802796222269535,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.0001,
      "step": 15
    },
    {
      "epoch": 0.20711974110032363,
      "grad_norm": 0.0032133806962519884,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.0001,
      "step": 16
    },
    {
      "epoch": 0.22006472491909385,
      "grad_norm": 0.006188486702740192,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.0001,
      "step": 17
    },
    {
      "epoch": 0.23300970873786409,
      "grad_norm": 2.9183788299560547,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.0014,
      "step": 18
    },
    {
      "epoch": 0.2459546925566343,
      "grad_norm": 0.009266923181712627,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.0001,
      "step": 19
    },
    {
      "epoch": 0.2588996763754045,
      "grad_norm": 0.11878190189599991,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.001,
      "step": 20
    },
    {
      "epoch": 0.27184466019417475,
      "grad_norm": 0.05568773299455643,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.0003,
      "step": 21
    },
    {
      "epoch": 0.284789644012945,
      "grad_norm": 0.27257758378982544,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.0003,
      "step": 22
    },
    {
      "epoch": 0.2977346278317152,
      "grad_norm": 0.0010619190288707614,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.0,
      "step": 23
    },
    {
      "epoch": 0.3106796116504854,
      "grad_norm": 0.0012464552419260144,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.0,
      "step": 24
    },
    {
      "epoch": 0.32362459546925565,
      "grad_norm": 0.004392554517835379,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.0,
      "step": 25
    },
    {
      "epoch": 0.3365695792880259,
      "grad_norm": 0.0003919773153029382,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.0,
      "step": 26
    },
    {
      "epoch": 0.34951456310679613,
      "grad_norm": 0.10447962582111359,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.0001,
      "step": 27
    },
    {
      "epoch": 0.36245954692556637,
      "grad_norm": 0.1830519437789917,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.0004,
      "step": 28
    },
    {
      "epoch": 0.37540453074433655,
      "grad_norm": 0.0003561489866115153,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.0,
      "step": 29
    },
    {
      "epoch": 0.3883495145631068,
      "grad_norm": 0.004686282481998205,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.0,
      "step": 30
    },
    {
      "epoch": 0.40129449838187703,
      "grad_norm": 0.0013320472789928317,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.0,
      "step": 31
    },
    {
      "epoch": 0.41423948220064727,
      "grad_norm": 0.025852827355265617,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.0,
      "step": 32
    },
    {
      "epoch": 0.42718446601941745,
      "grad_norm": 0.00043161975918337703,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.0,
      "step": 33
    },
    {
      "epoch": 0.4401294498381877,
      "grad_norm": 0.001204935135319829,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.0,
      "step": 34
    },
    {
      "epoch": 0.45307443365695793,
      "grad_norm": 0.00046805289457552135,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.0,
      "step": 35
    },
    {
      "epoch": 0.46601941747572817,
      "grad_norm": 0.0006769143510609865,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.0,
      "step": 36
    },
    {
      "epoch": 0.47896440129449835,
      "grad_norm": 0.0007127629360184073,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.0,
      "step": 37
    },
    {
      "epoch": 0.4919093851132686,
      "grad_norm": 0.001910505467094481,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.0,
      "step": 38
    },
    {
      "epoch": 0.5048543689320388,
      "grad_norm": 0.0006861760630272329,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.0,
      "step": 39
    },
    {
      "epoch": 0.517799352750809,
      "grad_norm": 0.0065710097551345825,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.0,
      "step": 40
    },
    {
      "epoch": 0.5307443365695793,
      "grad_norm": 0.14705024659633636,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.0005,
      "step": 41
    },
    {
      "epoch": 0.5436893203883495,
      "grad_norm": 0.02524726092815399,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.0,
      "step": 42
    },
    {
      "epoch": 0.5566343042071198,
      "grad_norm": 0.00010728214692790061,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.0,
      "step": 43
    },
    {
      "epoch": 0.56957928802589,
      "grad_norm": 0.0009991212282329798,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.0,
      "step": 44
    },
    {
      "epoch": 0.5825242718446602,
      "grad_norm": 0.00020374900486785918,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.0,
      "step": 45
    },
    {
      "epoch": 0.5954692556634305,
      "grad_norm": 0.00016235416114795953,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.0,
      "step": 46
    },
    {
      "epoch": 0.6084142394822006,
      "grad_norm": 0.00022325626923702657,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.0,
      "step": 47
    },
    {
      "epoch": 0.6213592233009708,
      "grad_norm": 0.0018907163757830858,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.0,
      "step": 48
    },
    {
      "epoch": 0.6343042071197411,
      "grad_norm": 0.00039835559437051415,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.0,
      "step": 49
    },
    {
      "epoch": 0.6472491909385113,
      "grad_norm": 0.0002363417879678309,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.0,
      "step": 50
    },
    {
      "epoch": 0.6472491909385113,
      "eval_loss": 5.708453318220563e-05,
      "eval_runtime": 9.8109,
      "eval_samples_per_second": 13.251,
      "eval_steps_per_second": 3.364,
      "step": 50
    },
    {
      "epoch": 0.6601941747572816,
      "grad_norm": 0.5545428395271301,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.0024,
      "step": 51
    },
    {
      "epoch": 0.6731391585760518,
      "grad_norm": 0.0015667704865336418,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.0,
      "step": 52
    },
    {
      "epoch": 0.686084142394822,
      "grad_norm": 0.011077708564698696,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.0,
      "step": 53
    },
    {
      "epoch": 0.6990291262135923,
      "grad_norm": 0.010326799936592579,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.0,
      "step": 54
    },
    {
      "epoch": 0.7119741100323624,
      "grad_norm": 0.0014523833524435759,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.0,
      "step": 55
    },
    {
      "epoch": 0.7249190938511327,
      "grad_norm": 0.06251269578933716,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.0002,
      "step": 56
    },
    {
      "epoch": 0.7378640776699029,
      "grad_norm": 0.003645202610641718,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.0,
      "step": 57
    },
    {
      "epoch": 0.7508090614886731,
      "grad_norm": 0.2278345227241516,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.0013,
      "step": 58
    },
    {
      "epoch": 0.7637540453074434,
      "grad_norm": 0.003915652632713318,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.0,
      "step": 59
    },
    {
      "epoch": 0.7766990291262136,
      "grad_norm": 0.0031239991076290607,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.0,
      "step": 60
    },
    {
      "epoch": 0.7896440129449838,
      "grad_norm": 0.006219485308974981,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.0001,
      "step": 61
    },
    {
      "epoch": 0.8025889967637541,
      "grad_norm": 0.005084617529064417,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.0,
      "step": 62
    },
    {
      "epoch": 0.8155339805825242,
      "grad_norm": 0.003509129863232374,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.0,
      "step": 63
    },
    {
      "epoch": 0.8284789644012945,
      "grad_norm": 0.0037375520914793015,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.0,
      "step": 64
    },
    {
      "epoch": 0.8414239482200647,
      "grad_norm": 0.00264919875189662,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.0,
      "step": 65
    },
    {
      "epoch": 0.8543689320388349,
      "grad_norm": 0.10307146608829498,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.0002,
      "step": 66
    },
    {
      "epoch": 0.8673139158576052,
      "grad_norm": 0.001447496935725212,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.0,
      "step": 67
    },
    {
      "epoch": 0.8802588996763754,
      "grad_norm": 0.0005993408849462867,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.0,
      "step": 68
    },
    {
      "epoch": 0.8932038834951457,
      "grad_norm": 0.0018503112951293588,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.0,
      "step": 69
    },
    {
      "epoch": 0.9061488673139159,
      "grad_norm": 0.0011176472762599587,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.0,
      "step": 70
    },
    {
      "epoch": 0.919093851132686,
      "grad_norm": 0.0071342382580041885,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.0,
      "step": 71
    },
    {
      "epoch": 0.9320388349514563,
      "grad_norm": 0.0045208255760371685,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.0,
      "step": 72
    },
    {
      "epoch": 0.9449838187702265,
      "grad_norm": 0.0016186069697141647,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.0,
      "step": 73
    },
    {
      "epoch": 0.9579288025889967,
      "grad_norm": 0.0014599431306123734,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.0,
      "step": 74
    },
    {
      "epoch": 0.970873786407767,
      "grad_norm": 0.0028509364929050207,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.0,
      "step": 75
    },
    {
      "epoch": 0.9838187702265372,
      "grad_norm": 0.0015371855115517974,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.0,
      "step": 76
    },
    {
      "epoch": 0.9967637540453075,
      "grad_norm": 0.0928664579987526,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.0001,
      "step": 77
    },
    {
      "epoch": 1.0097087378640777,
      "grad_norm": 0.0019110508728772402,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.0,
      "step": 78
    },
    {
      "epoch": 1.022653721682848,
      "grad_norm": 0.0001286536717088893,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.0,
      "step": 79
    },
    {
      "epoch": 1.035598705501618,
      "grad_norm": 0.00010072574514197186,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.0,
      "step": 80
    },
    {
      "epoch": 1.0485436893203883,
      "grad_norm": 8.10532656032592e-05,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.0,
      "step": 81
    },
    {
      "epoch": 1.0614886731391586,
      "grad_norm": 0.00010899638436967507,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.0,
      "step": 82
    },
    {
      "epoch": 1.074433656957929,
      "grad_norm": 7.373379776254296e-05,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.0,
      "step": 83
    },
    {
      "epoch": 1.087378640776699,
      "grad_norm": 0.00044416377204470336,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.0,
      "step": 84
    },
    {
      "epoch": 1.1003236245954693,
      "grad_norm": 0.00014110577467363328,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.0,
      "step": 85
    },
    {
      "epoch": 1.1132686084142396,
      "grad_norm": 7.416921289404854e-05,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.0,
      "step": 86
    },
    {
      "epoch": 1.1262135922330097,
      "grad_norm": 6.953041884116828e-05,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.0,
      "step": 87
    },
    {
      "epoch": 1.13915857605178,
      "grad_norm": 0.00018072547391057014,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.0,
      "step": 88
    },
    {
      "epoch": 1.1521035598705502,
      "grad_norm": 0.00014231581008061767,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.0,
      "step": 89
    },
    {
      "epoch": 1.1650485436893203,
      "grad_norm": 0.049870043992996216,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.0001,
      "step": 90
    },
    {
      "epoch": 1.1779935275080906,
      "grad_norm": 7.035312592051923e-05,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.0,
      "step": 91
    },
    {
      "epoch": 1.190938511326861,
      "grad_norm": 0.00027099126600660384,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.0,
      "step": 92
    },
    {
      "epoch": 1.203883495145631,
      "grad_norm": 0.00015066895866766572,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.0,
      "step": 93
    },
    {
      "epoch": 1.2168284789644013,
      "grad_norm": 0.0002044395951088518,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.0,
      "step": 94
    },
    {
      "epoch": 1.2297734627831716,
      "grad_norm": 0.00016389232769142836,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.0,
      "step": 95
    },
    {
      "epoch": 1.2427184466019416,
      "grad_norm": 0.00033923389855772257,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.0,
      "step": 96
    },
    {
      "epoch": 1.255663430420712,
      "grad_norm": 0.00017732290143612772,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.0,
      "step": 97
    },
    {
      "epoch": 1.2686084142394822,
      "grad_norm": 0.010200129821896553,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.0,
      "step": 98
    },
    {
      "epoch": 1.2815533980582523,
      "grad_norm": 0.00042962911538779736,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.0,
      "step": 99
    },
    {
      "epoch": 1.2944983818770226,
      "grad_norm": 0.0004276715044397861,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.0,
      "step": 100
    },
    {
      "epoch": 1.2944983818770226,
      "eval_loss": 2.6809143491846044e-06,
      "eval_runtime": 9.7982,
      "eval_samples_per_second": 13.268,
      "eval_steps_per_second": 3.368,
      "step": 100
    },
    {
      "epoch": 1.307443365695793,
      "grad_norm": 6.501084862975404e-05,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.0,
      "step": 101
    },
    {
      "epoch": 1.3203883495145632,
      "grad_norm": 0.00010038226173492149,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.0,
      "step": 102
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.0001285731268581003,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.0,
      "step": 103
    },
    {
      "epoch": 1.3462783171521036,
      "grad_norm": 0.001511538284830749,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.0,
      "step": 104
    },
    {
      "epoch": 1.3592233009708738,
      "grad_norm": 0.0001728609058773145,
      "learning_rate": 5e-05,
      "loss": 0.0,
      "step": 105
    },
    {
      "epoch": 1.3721682847896441,
      "grad_norm": 0.00020497982040978968,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.0,
      "step": 106
    },
    {
      "epoch": 1.3851132686084142,
      "grad_norm": 0.00012128037633374333,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.0,
      "step": 107
    },
    {
      "epoch": 1.3980582524271845,
      "grad_norm": 0.00011459392408141866,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.0,
      "step": 108
    },
    {
      "epoch": 1.4110032362459548,
      "grad_norm": 0.0001164645073004067,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.0,
      "step": 109
    },
    {
      "epoch": 1.4239482200647249,
      "grad_norm": 0.0002743539516814053,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.0,
      "step": 110
    },
    {
      "epoch": 1.4368932038834952,
      "grad_norm": 0.00023465685080736876,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.0,
      "step": 111
    },
    {
      "epoch": 1.4498381877022655,
      "grad_norm": 0.0001482328079873696,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.0,
      "step": 112
    },
    {
      "epoch": 1.4627831715210355,
      "grad_norm": 0.00019140676886308938,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.0,
      "step": 113
    },
    {
      "epoch": 1.4757281553398058,
      "grad_norm": 0.0003757222439162433,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.0,
      "step": 114
    },
    {
      "epoch": 1.4886731391585761,
      "grad_norm": 0.00025924082729034126,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.0,
      "step": 115
    },
    {
      "epoch": 1.5016181229773462,
      "grad_norm": 0.0408693365752697,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.0001,
      "step": 116
    },
    {
      "epoch": 1.5145631067961165,
      "grad_norm": 0.0011624947655946016,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.0,
      "step": 117
    },
    {
      "epoch": 1.5275080906148868,
      "grad_norm": 8.236700523411855e-05,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.0,
      "step": 118
    },
    {
      "epoch": 1.5404530744336569,
      "grad_norm": 0.00018579971219878644,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.0,
      "step": 119
    },
    {
      "epoch": 1.5533980582524272,
      "grad_norm": 0.0002899975224863738,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.0,
      "step": 120
    },
    {
      "epoch": 1.5663430420711975,
      "grad_norm": 0.001745874877087772,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.0,
      "step": 121
    },
    {
      "epoch": 1.5792880258899675,
      "grad_norm": 0.0003118854365311563,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.0,
      "step": 122
    },
    {
      "epoch": 1.5922330097087378,
      "grad_norm": 0.00011779765918618068,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.0,
      "step": 123
    },
    {
      "epoch": 1.6051779935275081,
      "grad_norm": 0.0001423739013262093,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.0,
      "step": 124
    },
    {
      "epoch": 1.6181229773462782,
      "grad_norm": 0.0001056610417435877,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.0,
      "step": 125
    },
    {
      "epoch": 1.6310679611650487,
      "grad_norm": 9.993094863602892e-05,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.0,
      "step": 126
    },
    {
      "epoch": 1.6440129449838188,
      "grad_norm": 0.00013124274846632034,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.0,
      "step": 127
    },
    {
      "epoch": 1.6569579288025889,
      "grad_norm": 9.836336539592594e-05,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.0,
      "step": 128
    },
    {
      "epoch": 1.6699029126213594,
      "grad_norm": 0.0002020129031734541,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.0,
      "step": 129
    },
    {
      "epoch": 1.6828478964401294,
      "grad_norm": 0.0007685688906349242,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.0,
      "step": 130
    },
    {
      "epoch": 1.6957928802588995,
      "grad_norm": 0.0014894054038450122,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.0,
      "step": 131
    },
    {
      "epoch": 1.70873786407767,
      "grad_norm": 0.0004369337111711502,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.0,
      "step": 132
    },
    {
      "epoch": 1.72168284789644,
      "grad_norm": 0.000166458121384494,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.0,
      "step": 133
    },
    {
      "epoch": 1.7346278317152104,
      "grad_norm": 0.0002028081362368539,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.0,
      "step": 134
    },
    {
      "epoch": 1.7475728155339807,
      "grad_norm": 0.0001326661295024678,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.0,
      "step": 135
    },
    {
      "epoch": 1.7605177993527508,
      "grad_norm": 6.024070535204373e-05,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.0,
      "step": 136
    },
    {
      "epoch": 1.773462783171521,
      "grad_norm": 0.0004106041742488742,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.0,
      "step": 137
    },
    {
      "epoch": 1.7864077669902914,
      "grad_norm": 5.501594932866283e-05,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.0,
      "step": 138
    },
    {
      "epoch": 1.7993527508090614,
      "grad_norm": 8.930330659495667e-05,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.0,
      "step": 139
    },
    {
      "epoch": 1.8122977346278317,
      "grad_norm": 0.00012645589595194906,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.0,
      "step": 140
    },
    {
      "epoch": 1.825242718446602,
      "grad_norm": 6.120719626778737e-05,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.0,
      "step": 141
    },
    {
      "epoch": 1.838187702265372,
      "grad_norm": 0.0001415311126038432,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.0,
      "step": 142
    },
    {
      "epoch": 1.8511326860841424,
      "grad_norm": 0.00016300934657920152,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.0,
      "step": 143
    },
    {
      "epoch": 1.8640776699029127,
      "grad_norm": 6.67195490677841e-05,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.0,
      "step": 144
    },
    {
      "epoch": 1.8770226537216828,
      "grad_norm": 0.000838777341414243,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.0,
      "step": 145
    },
    {
      "epoch": 1.889967637540453,
      "grad_norm": 8.979487756732851e-05,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.0,
      "step": 146
    },
    {
      "epoch": 1.9029126213592233,
      "grad_norm": 0.0001125104317907244,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.0,
      "step": 147
    },
    {
      "epoch": 1.9158576051779934,
      "grad_norm": 9.87684543360956e-05,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.0,
      "step": 148
    },
    {
      "epoch": 1.9288025889967637,
      "grad_norm": 0.0005952689098194242,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.0,
      "step": 149
    },
    {
      "epoch": 1.941747572815534,
      "grad_norm": 0.006320950109511614,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.0,
      "step": 150
    },
    {
      "epoch": 1.941747572815534,
      "eval_loss": 2.2710835310135735e-06,
      "eval_runtime": 9.8147,
      "eval_samples_per_second": 13.245,
      "eval_steps_per_second": 3.362,
      "step": 150
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.261827547132068e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}