{
  "best_metric": 1.1731460094451904,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 0.5401755570560433,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0027008777852802163,
      "grad_norm": 0.3624476194381714,
      "learning_rate": 1e-05,
      "loss": 1.4218,
      "step": 1
    },
    {
      "epoch": 0.0027008777852802163,
      "eval_loss": 1.4606833457946777,
      "eval_runtime": 15.6141,
      "eval_samples_per_second": 39.964,
      "eval_steps_per_second": 9.991,
      "step": 1
    },
    {
      "epoch": 0.0054017555705604325,
      "grad_norm": 0.3938450813293457,
      "learning_rate": 2e-05,
      "loss": 1.3709,
      "step": 2
    },
    {
      "epoch": 0.008102633355840648,
      "grad_norm": 0.39943069219589233,
      "learning_rate": 3e-05,
      "loss": 1.4067,
      "step": 3
    },
    {
      "epoch": 0.010803511141120865,
      "grad_norm": 0.40101104974746704,
      "learning_rate": 4e-05,
      "loss": 1.4221,
      "step": 4
    },
    {
      "epoch": 0.01350438892640108,
      "grad_norm": 0.36832377314567566,
      "learning_rate": 5e-05,
      "loss": 1.4441,
      "step": 5
    },
    {
      "epoch": 0.016205266711681297,
      "grad_norm": 0.41918250918388367,
      "learning_rate": 6e-05,
      "loss": 1.4905,
      "step": 6
    },
    {
      "epoch": 0.018906144496961513,
      "grad_norm": 0.4318130612373352,
      "learning_rate": 7e-05,
      "loss": 1.3951,
      "step": 7
    },
    {
      "epoch": 0.02160702228224173,
      "grad_norm": 0.524061381816864,
      "learning_rate": 8e-05,
      "loss": 1.3791,
      "step": 8
    },
    {
      "epoch": 0.024307900067521943,
      "grad_norm": 0.5278139710426331,
      "learning_rate": 9e-05,
      "loss": 1.3826,
      "step": 9
    },
    {
      "epoch": 0.02700877785280216,
      "grad_norm": 0.5666940808296204,
      "learning_rate": 0.0001,
      "loss": 1.3379,
      "step": 10
    },
    {
      "epoch": 0.029709655638082377,
      "grad_norm": 0.5191312432289124,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.1892,
      "step": 11
    },
    {
      "epoch": 0.03241053342336259,
      "grad_norm": 0.430105596780777,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.1462,
      "step": 12
    },
    {
      "epoch": 0.035111411208642807,
      "grad_norm": 0.4412747621536255,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.2241,
      "step": 13
    },
    {
      "epoch": 0.03781228899392303,
      "grad_norm": 0.4131179451942444,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.2011,
      "step": 14
    },
    {
      "epoch": 0.04051316677920324,
      "grad_norm": 0.4340050518512726,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.1766,
      "step": 15
    },
    {
      "epoch": 0.04321404456448346,
      "grad_norm": 0.42998600006103516,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.2414,
      "step": 16
    },
    {
      "epoch": 0.04591492234976367,
      "grad_norm": 0.42034226655960083,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.0788,
      "step": 17
    },
    {
      "epoch": 0.048615800135043886,
      "grad_norm": 0.5337233543395996,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.1981,
      "step": 18
    },
    {
      "epoch": 0.05131667792032411,
      "grad_norm": 0.4583562910556793,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.2196,
      "step": 19
    },
    {
      "epoch": 0.05401755570560432,
      "grad_norm": 0.49332761764526367,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.1968,
      "step": 20
    },
    {
      "epoch": 0.05671843349088454,
      "grad_norm": 0.48594799637794495,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.1309,
      "step": 21
    },
    {
      "epoch": 0.05941931127616475,
      "grad_norm": 0.5126538276672363,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.1186,
      "step": 22
    },
    {
      "epoch": 0.062120189061444966,
      "grad_norm": 0.47921475768089294,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.0795,
      "step": 23
    },
    {
      "epoch": 0.06482106684672519,
      "grad_norm": 0.4876593053340912,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.0949,
      "step": 24
    },
    {
      "epoch": 0.0675219446320054,
      "grad_norm": 0.536691427230835,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.0373,
      "step": 25
    },
    {
      "epoch": 0.07022282241728561,
      "grad_norm": 0.6566392779350281,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.2434,
      "step": 26
    },
    {
      "epoch": 0.07292370020256583,
      "grad_norm": 0.5158787369728088,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.1554,
      "step": 27
    },
    {
      "epoch": 0.07562457798784605,
      "grad_norm": 0.5759814977645874,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.3289,
      "step": 28
    },
    {
      "epoch": 0.07832545577312626,
      "grad_norm": 0.5752044320106506,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.1889,
      "step": 29
    },
    {
      "epoch": 0.08102633355840648,
      "grad_norm": 0.6152178645133972,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.1266,
      "step": 30
    },
    {
      "epoch": 0.0837272113436867,
      "grad_norm": 0.6457420587539673,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.1839,
      "step": 31
    },
    {
      "epoch": 0.08642808912896692,
      "grad_norm": 0.6329084038734436,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.234,
      "step": 32
    },
    {
      "epoch": 0.08912896691424713,
      "grad_norm": 0.6863877773284912,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.2342,
      "step": 33
    },
    {
      "epoch": 0.09182984469952735,
      "grad_norm": 0.6431719660758972,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.177,
      "step": 34
    },
    {
      "epoch": 0.09453072248480757,
      "grad_norm": 0.6808779835700989,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.3163,
      "step": 35
    },
    {
      "epoch": 0.09723160027008777,
      "grad_norm": 0.7064206600189209,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.3071,
      "step": 36
    },
    {
      "epoch": 0.099932478055368,
      "grad_norm": 0.7828763723373413,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.3091,
      "step": 37
    },
    {
      "epoch": 0.10263335584064821,
      "grad_norm": 0.8526866436004639,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.2666,
      "step": 38
    },
    {
      "epoch": 0.10533423362592843,
      "grad_norm": 0.7964354157447815,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.4333,
      "step": 39
    },
    {
      "epoch": 0.10803511141120864,
      "grad_norm": 0.8185365796089172,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.3689,
      "step": 40
    },
    {
      "epoch": 0.11073598919648886,
      "grad_norm": 0.8150967955589294,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.3536,
      "step": 41
    },
    {
      "epoch": 0.11343686698176908,
      "grad_norm": 0.8503111004829407,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.3862,
      "step": 42
    },
    {
      "epoch": 0.11613774476704929,
      "grad_norm": 0.8694619536399841,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.3512,
      "step": 43
    },
    {
      "epoch": 0.1188386225523295,
      "grad_norm": 1.0029643774032593,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.2742,
      "step": 44
    },
    {
      "epoch": 0.12153950033760973,
      "grad_norm": 1.017909049987793,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.4303,
      "step": 45
    },
    {
      "epoch": 0.12424037812288993,
      "grad_norm": 1.519902229309082,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.4151,
      "step": 46
    },
    {
      "epoch": 0.12694125590817015,
      "grad_norm": 1.4137510061264038,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.4488,
      "step": 47
    },
    {
      "epoch": 0.12964213369345037,
      "grad_norm": 1.4305354356765747,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.3568,
      "step": 48
    },
    {
      "epoch": 0.1323430114787306,
      "grad_norm": 1.5225553512573242,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.2391,
      "step": 49
    },
    {
      "epoch": 0.1350438892640108,
      "grad_norm": 2.4049932956695557,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.4481,
      "step": 50
    },
    {
      "epoch": 0.1350438892640108,
      "eval_loss": 1.2866495847702026,
      "eval_runtime": 15.5441,
      "eval_samples_per_second": 40.144,
      "eval_steps_per_second": 10.036,
      "step": 50
    },
    {
      "epoch": 0.137744767049291,
      "grad_norm": 0.6834385395050049,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.4087,
      "step": 51
    },
    {
      "epoch": 0.14044564483457123,
      "grad_norm": 0.6383047699928284,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.4105,
      "step": 52
    },
    {
      "epoch": 0.14314652261985145,
      "grad_norm": 0.4831598997116089,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.4986,
      "step": 53
    },
    {
      "epoch": 0.14584740040513167,
      "grad_norm": 0.42735621333122253,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.3571,
      "step": 54
    },
    {
      "epoch": 0.1485482781904119,
      "grad_norm": 0.3566948473453522,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.3558,
      "step": 55
    },
    {
      "epoch": 0.1512491559756921,
      "grad_norm": 0.37861260771751404,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.2901,
      "step": 56
    },
    {
      "epoch": 0.15395003376097233,
      "grad_norm": 0.3618011176586151,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.2906,
      "step": 57
    },
    {
      "epoch": 0.15665091154625252,
      "grad_norm": 0.42364272475242615,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.2269,
      "step": 58
    },
    {
      "epoch": 0.15935178933153274,
      "grad_norm": 0.39519202709198,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.3802,
      "step": 59
    },
    {
      "epoch": 0.16205266711681296,
      "grad_norm": 0.39644855260849,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.2644,
      "step": 60
    },
    {
      "epoch": 0.16475354490209318,
      "grad_norm": 0.4740903973579407,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.4338,
      "step": 61
    },
    {
      "epoch": 0.1674544226873734,
      "grad_norm": 0.40571337938308716,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.0527,
      "step": 62
    },
    {
      "epoch": 0.17015530047265362,
      "grad_norm": 0.37622109055519104,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.2377,
      "step": 63
    },
    {
      "epoch": 0.17285617825793384,
      "grad_norm": 0.3867974579334259,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.1328,
      "step": 64
    },
    {
      "epoch": 0.17555705604321403,
      "grad_norm": 0.3665173649787903,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.0941,
      "step": 65
    },
    {
      "epoch": 0.17825793382849425,
      "grad_norm": 0.3812106251716614,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.1682,
      "step": 66
    },
    {
      "epoch": 0.18095881161377447,
      "grad_norm": 0.396047979593277,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.0769,
      "step": 67
    },
    {
      "epoch": 0.1836596893990547,
      "grad_norm": 0.4016658663749695,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.2621,
      "step": 68
    },
    {
      "epoch": 0.1863605671843349,
      "grad_norm": 0.42150306701660156,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.0641,
      "step": 69
    },
    {
      "epoch": 0.18906144496961513,
      "grad_norm": 0.4154197573661804,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.0688,
      "step": 70
    },
    {
      "epoch": 0.19176232275489535,
      "grad_norm": 0.43228766322135925,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.0701,
      "step": 71
    },
    {
      "epoch": 0.19446320054017555,
      "grad_norm": 0.5811237096786499,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.1266,
      "step": 72
    },
    {
      "epoch": 0.19716407832545577,
      "grad_norm": 0.4404866397380829,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.0782,
      "step": 73
    },
    {
      "epoch": 0.199864956110736,
      "grad_norm": 0.5220156908035278,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.1294,
      "step": 74
    },
    {
      "epoch": 0.2025658338960162,
      "grad_norm": 0.5216073989868164,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.1765,
      "step": 75
    },
    {
      "epoch": 0.20526671168129643,
      "grad_norm": 0.5543369054794312,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.0467,
      "step": 76
    },
    {
      "epoch": 0.20796758946657665,
      "grad_norm": 0.4874238669872284,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.164,
      "step": 77
    },
    {
      "epoch": 0.21066846725185687,
      "grad_norm": 0.5671214461326599,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.0665,
      "step": 78
    },
    {
      "epoch": 0.21336934503713706,
      "grad_norm": 0.5563834309577942,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.1413,
      "step": 79
    },
    {
      "epoch": 0.21607022282241728,
      "grad_norm": 0.5251386761665344,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.0943,
      "step": 80
    },
    {
      "epoch": 0.2187711006076975,
      "grad_norm": 0.6318755745887756,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.0754,
      "step": 81
    },
    {
      "epoch": 0.22147197839297772,
      "grad_norm": 0.6159390211105347,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.2784,
      "step": 82
    },
    {
      "epoch": 0.22417285617825794,
      "grad_norm": 0.6226337552070618,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.2542,
      "step": 83
    },
    {
      "epoch": 0.22687373396353816,
      "grad_norm": 0.6405479311943054,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.2991,
      "step": 84
    },
    {
      "epoch": 0.22957461174881835,
      "grad_norm": 0.5985673069953918,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.2481,
      "step": 85
    },
    {
      "epoch": 0.23227548953409857,
      "grad_norm": 0.6607664823532104,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.1372,
      "step": 86
    },
    {
      "epoch": 0.2349763673193788,
      "grad_norm": 0.6545572280883789,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.2627,
      "step": 87
    },
    {
      "epoch": 0.237677245104659,
      "grad_norm": 0.7357593774795532,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.1535,
      "step": 88
    },
    {
      "epoch": 0.24037812288993923,
      "grad_norm": 0.7484399676322937,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.3262,
      "step": 89
    },
    {
      "epoch": 0.24307900067521945,
      "grad_norm": 0.7360692024230957,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.2027,
      "step": 90
    },
    {
      "epoch": 0.24577987846049967,
      "grad_norm": 0.7954505681991577,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.3984,
      "step": 91
    },
    {
      "epoch": 0.24848075624577987,
      "grad_norm": 0.8479906320571899,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.3639,
      "step": 92
    },
    {
      "epoch": 0.2511816340310601,
      "grad_norm": 0.9463720917701721,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.2841,
      "step": 93
    },
    {
      "epoch": 0.2538825118163403,
      "grad_norm": 1.0730533599853516,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.3653,
      "step": 94
    },
    {
      "epoch": 0.2565833896016205,
      "grad_norm": 0.95060795545578,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.3841,
      "step": 95
    },
    {
      "epoch": 0.25928426738690075,
      "grad_norm": 1.2638509273529053,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.3912,
      "step": 96
    },
    {
      "epoch": 0.26198514517218097,
      "grad_norm": 1.0868135690689087,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.4162,
      "step": 97
    },
    {
      "epoch": 0.2646860229574612,
      "grad_norm": 1.3574937582015991,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.2467,
      "step": 98
    },
    {
      "epoch": 0.2673869007427414,
      "grad_norm": 1.4365593194961548,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.1534,
      "step": 99
    },
    {
      "epoch": 0.2700877785280216,
      "grad_norm": 2.1243996620178223,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.2386,
      "step": 100
    },
    {
      "epoch": 0.2700877785280216,
      "eval_loss": 1.2199846506118774,
      "eval_runtime": 15.5244,
      "eval_samples_per_second": 40.195,
      "eval_steps_per_second": 10.049,
      "step": 100
    },
    {
      "epoch": 0.27278865631330185,
      "grad_norm": 0.35616037249565125,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.4098,
      "step": 101
    },
    {
      "epoch": 0.275489534098582,
      "grad_norm": 0.4488758146762848,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.3327,
      "step": 102
    },
    {
      "epoch": 0.27819041188386223,
      "grad_norm": 0.36774784326553345,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.3351,
      "step": 103
    },
    {
      "epoch": 0.28089128966914245,
      "grad_norm": 0.3631817400455475,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.3699,
      "step": 104
    },
    {
      "epoch": 0.2835921674544227,
      "grad_norm": 0.4073306620121002,
      "learning_rate": 5e-05,
      "loss": 1.2573,
      "step": 105
    },
    {
      "epoch": 0.2862930452397029,
      "grad_norm": 0.38760390877723694,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.3056,
      "step": 106
    },
    {
      "epoch": 0.2889939230249831,
      "grad_norm": 0.4187258183956146,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.2195,
      "step": 107
    },
    {
      "epoch": 0.29169480081026333,
      "grad_norm": 0.4538334310054779,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.2329,
      "step": 108
    },
    {
      "epoch": 0.29439567859554355,
      "grad_norm": 0.4429304897785187,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.2568,
      "step": 109
    },
    {
      "epoch": 0.2970965563808238,
      "grad_norm": 0.4676951766014099,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.1481,
      "step": 110
    },
    {
      "epoch": 0.299797434166104,
      "grad_norm": 0.37878647446632385,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.09,
      "step": 111
    },
    {
      "epoch": 0.3024983119513842,
      "grad_norm": 0.3448541760444641,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.17,
      "step": 112
    },
    {
      "epoch": 0.30519918973666443,
      "grad_norm": 0.391757994890213,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.1818,
      "step": 113
    },
    {
      "epoch": 0.30790006752194465,
      "grad_norm": 0.3847673535346985,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.1785,
      "step": 114
    },
    {
      "epoch": 0.3106009453072249,
      "grad_norm": 0.409818559885025,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.0581,
      "step": 115
    },
    {
      "epoch": 0.31330182309250504,
      "grad_norm": 0.39335235953330994,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.1005,
      "step": 116
    },
    {
      "epoch": 0.31600270087778526,
      "grad_norm": 0.4226117730140686,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.2566,
      "step": 117
    },
    {
      "epoch": 0.3187035786630655,
      "grad_norm": 0.4220075309276581,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.1277,
      "step": 118
    },
    {
      "epoch": 0.3214044564483457,
      "grad_norm": 0.4282516837120056,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.2025,
      "step": 119
    },
    {
      "epoch": 0.3241053342336259,
      "grad_norm": 0.4248524010181427,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.0884,
      "step": 120
    },
    {
      "epoch": 0.32680621201890614,
      "grad_norm": 0.4239839017391205,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.0171,
      "step": 121
    },
    {
      "epoch": 0.32950708980418636,
      "grad_norm": 0.461810439825058,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.1447,
      "step": 122
    },
    {
      "epoch": 0.3322079675894666,
      "grad_norm": 0.3951714038848877,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.0092,
      "step": 123
    },
    {
      "epoch": 0.3349088453747468,
      "grad_norm": 0.44695621728897095,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.0149,
      "step": 124
    },
    {
      "epoch": 0.337609723160027,
      "grad_norm": 0.4358537495136261,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.0545,
      "step": 125
    },
    {
      "epoch": 0.34031060094530724,
      "grad_norm": 0.46858829259872437,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.1025,
      "step": 126
    },
    {
      "epoch": 0.34301147873058746,
      "grad_norm": 0.4878123998641968,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.0073,
      "step": 127
    },
    {
      "epoch": 0.3457123565158677,
      "grad_norm": 0.48034295439720154,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.0226,
      "step": 128
    },
    {
      "epoch": 0.3484132343011479,
      "grad_norm": 0.49730291962623596,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.158,
      "step": 129
    },
    {
      "epoch": 0.35111411208642807,
      "grad_norm": 0.5154497027397156,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.098,
      "step": 130
    },
    {
      "epoch": 0.3538149898717083,
      "grad_norm": 0.5487515330314636,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.082,
      "step": 131
    },
    {
      "epoch": 0.3565158676569885,
      "grad_norm": 0.5983529686927795,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.188,
      "step": 132
    },
    {
      "epoch": 0.3592167454422687,
      "grad_norm": 0.6114199161529541,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.1883,
      "step": 133
    },
    {
      "epoch": 0.36191762322754895,
      "grad_norm": 0.5855720043182373,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.1963,
      "step": 134
    },
    {
      "epoch": 0.36461850101282917,
      "grad_norm": 0.633309543132782,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.1824,
      "step": 135
    },
    {
      "epoch": 0.3673193787981094,
      "grad_norm": 0.6567472815513611,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.1252,
      "step": 136
    },
    {
      "epoch": 0.3700202565833896,
      "grad_norm": 0.7787711024284363,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.2533,
      "step": 137
    },
    {
      "epoch": 0.3727211343686698,
      "grad_norm": 0.6865463852882385,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.2705,
      "step": 138
    },
    {
      "epoch": 0.37542201215395005,
      "grad_norm": 0.7248463034629822,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.3203,
      "step": 139
    },
    {
      "epoch": 0.37812288993923027,
      "grad_norm": 0.7092664837837219,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.408,
      "step": 140
    },
    {
      "epoch": 0.3808237677245105,
      "grad_norm": 0.8950397372245789,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.3258,
      "step": 141
    },
    {
      "epoch": 0.3835246455097907,
      "grad_norm": 0.7758725881576538,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.2531,
      "step": 142
    },
    {
      "epoch": 0.38622552329507087,
      "grad_norm": 0.8862745761871338,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.3664,
      "step": 143
    },
    {
      "epoch": 0.3889264010803511,
      "grad_norm": 0.8553701043128967,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.3519,
      "step": 144
    },
    {
      "epoch": 0.3916272788656313,
      "grad_norm": 0.8899438977241516,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1241,
      "step": 145
    },
    {
      "epoch": 0.39432815665091153,
      "grad_norm": 1.0587637424468994,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.3975,
      "step": 146
    },
    {
      "epoch": 0.39702903443619175,
      "grad_norm": 0.9895782470703125,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.2027,
      "step": 147
    },
    {
      "epoch": 0.399729912221472,
      "grad_norm": 1.0888241529464722,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.3854,
      "step": 148
    },
    {
      "epoch": 0.4024307900067522,
      "grad_norm": 1.2862181663513184,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.146,
      "step": 149
    },
    {
      "epoch": 0.4051316677920324,
      "grad_norm": 2.0095698833465576,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.4912,
      "step": 150
    },
    {
      "epoch": 0.4051316677920324,
      "eval_loss": 1.1731460094451904,
      "eval_runtime": 15.5301,
      "eval_samples_per_second": 40.18,
      "eval_steps_per_second": 10.045,
      "step": 150
    },
    {
      "epoch": 0.40783254557731263,
      "grad_norm": 0.2538085877895355,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.3162,
      "step": 151
    },
    {
      "epoch": 0.41053342336259285,
      "grad_norm": 0.2953255772590637,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.2597,
      "step": 152
    },
    {
      "epoch": 0.4132343011478731,
      "grad_norm": 0.2751544415950775,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.3478,
      "step": 153
    },
    {
      "epoch": 0.4159351789331533,
      "grad_norm": 0.29152625799179077,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.3893,
      "step": 154
    },
    {
      "epoch": 0.4186360567184335,
      "grad_norm": 0.3238907754421234,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.2884,
      "step": 155
    },
    {
      "epoch": 0.42133693450371373,
      "grad_norm": 0.36897632479667664,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.2411,
      "step": 156
    },
    {
      "epoch": 0.4240378122889939,
      "grad_norm": 0.40534698963165283,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.0328,
      "step": 157
    },
    {
      "epoch": 0.4267386900742741,
      "grad_norm": 0.44237667322158813,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.2544,
      "step": 158
    },
    {
      "epoch": 0.42943956785955434,
      "grad_norm": 0.3954787254333496,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.1782,
      "step": 159
    },
    {
      "epoch": 0.43214044564483456,
      "grad_norm": 0.3908039331436157,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.2553,
      "step": 160
    },
    {
      "epoch": 0.4348413234301148,
      "grad_norm": 0.3938920795917511,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.164,
      "step": 161
    },
    {
      "epoch": 0.437542201215395,
      "grad_norm": 0.3663087785243988,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.1487,
      "step": 162
    },
    {
      "epoch": 0.4402430790006752,
      "grad_norm": 0.4079456031322479,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.2359,
      "step": 163
    },
    {
      "epoch": 0.44294395678595544,
      "grad_norm": 0.3637031316757202,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.1947,
      "step": 164
    },
    {
      "epoch": 0.44564483457123566,
      "grad_norm": 0.36981093883514404,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.1892,
      "step": 165
    },
    {
      "epoch": 0.4483457123565159,
      "grad_norm": 0.34960412979125977,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.116,
      "step": 166
    },
    {
      "epoch": 0.4510465901417961,
      "grad_norm": 0.3739708364009857,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.1281,
      "step": 167
    },
    {
      "epoch": 0.4537474679270763,
      "grad_norm": 0.4456571042537689,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.0629,
      "step": 168
    },
    {
      "epoch": 0.45644834571235654,
      "grad_norm": 0.4303252398967743,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.0095,
      "step": 169
    },
    {
      "epoch": 0.4591492234976367,
      "grad_norm": 0.3988953232765198,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.0987,
      "step": 170
    },
    {
      "epoch": 0.4618501012829169,
      "grad_norm": 0.46439293026924133,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.1222,
      "step": 171
    },
    {
      "epoch": 0.46455097906819715,
      "grad_norm": 0.42667528986930847,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.1285,
      "step": 172
    },
    {
      "epoch": 0.46725185685347737,
      "grad_norm": 0.4452310800552368,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.0251,
      "step": 173
    },
    {
      "epoch": 0.4699527346387576,
      "grad_norm": 0.4917750358581543,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.0654,
      "step": 174
    },
    {
      "epoch": 0.4726536124240378,
      "grad_norm": 0.4726669490337372,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.0525,
      "step": 175
    },
    {
      "epoch": 0.475354490209318,
      "grad_norm": 0.46483397483825684,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.1452,
      "step": 176
    },
    {
      "epoch": 0.47805536799459825,
      "grad_norm": 0.5226284265518188,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.176,
      "step": 177
    },
    {
      "epoch": 0.48075624577987847,
      "grad_norm": 0.47096461057662964,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.0988,
      "step": 178
    },
    {
      "epoch": 0.4834571235651587,
      "grad_norm": 0.5327020883560181,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.1441,
      "step": 179
    },
    {
      "epoch": 0.4861580013504389,
      "grad_norm": 0.592038094997406,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.0014,
      "step": 180
    },
    {
      "epoch": 0.4888588791357191,
      "grad_norm": 0.5378474593162537,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.0981,
      "step": 181
    },
    {
      "epoch": 0.49155975692099935,
      "grad_norm": 0.5592798590660095,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.1337,
      "step": 182
    },
    {
      "epoch": 0.49426063470627957,
      "grad_norm": 0.5733963847160339,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.0944,
      "step": 183
    },
    {
      "epoch": 0.49696151249155973,
      "grad_norm": 0.612891674041748,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.2933,
      "step": 184
    },
    {
      "epoch": 0.49966239027683995,
      "grad_norm": 0.59614098072052,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.2098,
      "step": 185
    },
    {
      "epoch": 0.5023632680621202,
      "grad_norm": 0.633982241153717,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.226,
      "step": 186
    },
    {
      "epoch": 0.5050641458474004,
      "grad_norm": 0.7396524548530579,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.2522,
      "step": 187
    },
    {
      "epoch": 0.5077650236326806,
      "grad_norm": 0.6871618032455444,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.2784,
      "step": 188
    },
    {
      "epoch": 0.5104659014179609,
      "grad_norm": 0.705457329750061,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.1914,
      "step": 189
    },
    {
      "epoch": 0.513166779203241,
      "grad_norm": 0.7759872674942017,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.3097,
      "step": 190
    },
    {
      "epoch": 0.5158676569885212,
      "grad_norm": 0.7772696018218994,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.104,
      "step": 191
    },
    {
      "epoch": 0.5185685347738015,
      "grad_norm": 0.851283848285675,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.3883,
      "step": 192
    },
    {
      "epoch": 0.5212694125590817,
      "grad_norm": 0.8198362588882446,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.2874,
      "step": 193
    },
    {
      "epoch": 0.5239702903443619,
      "grad_norm": 0.8983657360076904,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.3215,
      "step": 194
    },
    {
      "epoch": 0.5266711681296421,
      "grad_norm": 0.8976041078567505,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.3496,
      "step": 195
    },
    {
      "epoch": 0.5293720459149224,
      "grad_norm": 0.9548887610435486,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.2875,
      "step": 196
    },
    {
      "epoch": 0.5320729237002025,
      "grad_norm": 1.1276315450668335,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.2384,
      "step": 197
    },
    {
      "epoch": 0.5347738014854828,
      "grad_norm": 1.1375982761383057,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.0811,
      "step": 198
    },
    {
      "epoch": 0.537474679270763,
      "grad_norm": 1.3477256298065186,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.1775,
      "step": 199
    },
    {
      "epoch": 0.5401755570560433,
      "grad_norm": 2.2407636642456055,
      "learning_rate": 0.0,
      "loss": 1.5276,
      "step": 200
    },
    {
      "epoch": 0.5401755570560433,
      "eval_loss": 1.1766870021820068,
      "eval_runtime": 15.4666,
      "eval_samples_per_second": 40.345,
      "eval_steps_per_second": 10.086,
      "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": 1
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.789879145332736e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}