{
  "best_metric": 2.718132734298706,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.6019563581640331,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0030097817908201654,
      "grad_norm": 1.0383433103561401,
      "learning_rate": 1e-05,
      "loss": 2.837,
      "step": 1
    },
    {
      "epoch": 0.0030097817908201654,
      "eval_loss": 3.62335467338562,
      "eval_runtime": 42.8751,
      "eval_samples_per_second": 13.061,
      "eval_steps_per_second": 3.265,
      "step": 1
    },
    {
      "epoch": 0.006019563581640331,
      "grad_norm": 1.223947525024414,
      "learning_rate": 2e-05,
      "loss": 2.924,
      "step": 2
    },
    {
      "epoch": 0.009029345372460496,
      "grad_norm": 1.2525657415390015,
      "learning_rate": 3e-05,
      "loss": 2.8023,
      "step": 3
    },
    {
      "epoch": 0.012039127163280662,
      "grad_norm": 1.2499439716339111,
      "learning_rate": 4e-05,
      "loss": 2.8144,
      "step": 4
    },
    {
      "epoch": 0.015048908954100828,
      "grad_norm": 0.8609808683395386,
      "learning_rate": 5e-05,
      "loss": 2.7237,
      "step": 5
    },
    {
      "epoch": 0.01805869074492099,
      "grad_norm": 1.0881868600845337,
      "learning_rate": 6e-05,
      "loss": 2.9225,
      "step": 6
    },
    {
      "epoch": 0.021068472535741158,
      "grad_norm": 1.000299334526062,
      "learning_rate": 7e-05,
      "loss": 2.8644,
      "step": 7
    },
    {
      "epoch": 0.024078254326561323,
      "grad_norm": 0.9951403141021729,
      "learning_rate": 8e-05,
      "loss": 2.7948,
      "step": 8
    },
    {
      "epoch": 0.02708803611738149,
      "grad_norm": 1.1025782823562622,
      "learning_rate": 9e-05,
      "loss": 2.853,
      "step": 9
    },
    {
      "epoch": 0.030097817908201655,
      "grad_norm": 1.0384840965270996,
      "learning_rate": 0.0001,
      "loss": 2.8641,
      "step": 10
    },
    {
      "epoch": 0.03310759969902182,
      "grad_norm": 1.3857022523880005,
      "learning_rate": 9.999316524962345e-05,
      "loss": 3.0502,
      "step": 11
    },
    {
      "epoch": 0.03611738148984198,
      "grad_norm": 1.0399494171142578,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.8511,
      "step": 12
    },
    {
      "epoch": 0.03912716328066215,
      "grad_norm": 1.0963470935821533,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.9582,
      "step": 13
    },
    {
      "epoch": 0.042136945071482315,
      "grad_norm": 1.1283156871795654,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.7851,
      "step": 14
    },
    {
      "epoch": 0.045146726862302484,
      "grad_norm": 1.0826599597930908,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.768,
      "step": 15
    },
    {
      "epoch": 0.04815650865312265,
      "grad_norm": 0.9662471413612366,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.7946,
      "step": 16
    },
    {
      "epoch": 0.051166290443942816,
      "grad_norm": 0.9118244051933289,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.7911,
      "step": 17
    },
    {
      "epoch": 0.05417607223476298,
      "grad_norm": 1.045340895652771,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.8433,
      "step": 18
    },
    {
      "epoch": 0.05718585402558315,
      "grad_norm": 1.2076919078826904,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.9204,
      "step": 19
    },
    {
      "epoch": 0.06019563581640331,
      "grad_norm": 0.9516964554786682,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.8679,
      "step": 20
    },
    {
      "epoch": 0.06320541760722348,
      "grad_norm": 0.9742953181266785,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.6819,
      "step": 21
    },
    {
      "epoch": 0.06621519939804364,
      "grad_norm": 1.2952927350997925,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.9203,
      "step": 22
    },
    {
      "epoch": 0.0692249811888638,
      "grad_norm": 1.0710830688476562,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.918,
      "step": 23
    },
    {
      "epoch": 0.07223476297968397,
      "grad_norm": 1.1441292762756348,
      "learning_rate": 9.86663298624003e-05,
      "loss": 3.0047,
      "step": 24
    },
    {
      "epoch": 0.07524454477050414,
      "grad_norm": 0.9753203988075256,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.6872,
      "step": 25
    },
    {
      "epoch": 0.0782543265613243,
      "grad_norm": 1.068655014038086,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.8628,
      "step": 26
    },
    {
      "epoch": 0.08126410835214447,
      "grad_norm": 1.1840782165527344,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.8454,
      "step": 27
    },
    {
      "epoch": 0.08427389014296463,
      "grad_norm": 1.118066430091858,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.646,
      "step": 28
    },
    {
      "epoch": 0.0872836719337848,
      "grad_norm": 1.0265779495239258,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.5202,
      "step": 29
    },
    {
      "epoch": 0.09029345372460497,
      "grad_norm": 1.3051583766937256,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.8238,
      "step": 30
    },
    {
      "epoch": 0.09330323551542513,
      "grad_norm": 1.3814913034439087,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.7794,
      "step": 31
    },
    {
      "epoch": 0.0963130173062453,
      "grad_norm": 1.313609004020691,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.8425,
      "step": 32
    },
    {
      "epoch": 0.09932279909706546,
      "grad_norm": 1.2244532108306885,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.8177,
      "step": 33
    },
    {
      "epoch": 0.10233258088788563,
      "grad_norm": 1.4679851531982422,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.8303,
      "step": 34
    },
    {
      "epoch": 0.1053423626787058,
      "grad_norm": 1.2653416395187378,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.8379,
      "step": 35
    },
    {
      "epoch": 0.10835214446952596,
      "grad_norm": 1.4733144044876099,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.7795,
      "step": 36
    },
    {
      "epoch": 0.11136192626034612,
      "grad_norm": 1.387394905090332,
      "learning_rate": 9.509956150664796e-05,
      "loss": 3.0197,
      "step": 37
    },
    {
      "epoch": 0.1143717080511663,
      "grad_norm": 1.3538862466812134,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.9811,
      "step": 38
    },
    {
      "epoch": 0.11738148984198646,
      "grad_norm": 1.4648830890655518,
      "learning_rate": 9.43611409721806e-05,
      "loss": 3.0393,
      "step": 39
    },
    {
      "epoch": 0.12039127163280662,
      "grad_norm": 1.4998297691345215,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.983,
      "step": 40
    },
    {
      "epoch": 0.12340105342362678,
      "grad_norm": 1.3303641080856323,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.8374,
      "step": 41
    },
    {
      "epoch": 0.12641083521444696,
      "grad_norm": 1.434880018234253,
      "learning_rate": 9.316282404787871e-05,
      "loss": 3.001,
      "step": 42
    },
    {
      "epoch": 0.1294206170052671,
      "grad_norm": 1.4882705211639404,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.9935,
      "step": 43
    },
    {
      "epoch": 0.13243039879608728,
      "grad_norm": 1.495313048362732,
      "learning_rate": 9.230476262104677e-05,
      "loss": 3.062,
      "step": 44
    },
    {
      "epoch": 0.13544018058690746,
      "grad_norm": 1.6354619264602661,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.9735,
      "step": 45
    },
    {
      "epoch": 0.1384499623777276,
      "grad_norm": 1.6975343227386475,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.8529,
      "step": 46
    },
    {
      "epoch": 0.14145974416854779,
      "grad_norm": 1.8099063634872437,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.931,
      "step": 47
    },
    {
      "epoch": 0.14446952595936793,
      "grad_norm": 1.795233964920044,
      "learning_rate": 9.045084971874738e-05,
      "loss": 3.1318,
      "step": 48
    },
    {
      "epoch": 0.1474793077501881,
      "grad_norm": 2.3223793506622314,
      "learning_rate": 8.995939984474624e-05,
      "loss": 3.4612,
      "step": 49
    },
    {
      "epoch": 0.1504890895410083,
      "grad_norm": 2.675429582595825,
      "learning_rate": 8.945702546981969e-05,
      "loss": 3.2712,
      "step": 50
    },
    {
      "epoch": 0.1504890895410083,
      "eval_loss": 2.8597769737243652,
      "eval_runtime": 43.5315,
      "eval_samples_per_second": 12.864,
      "eval_steps_per_second": 3.216,
      "step": 50
    },
    {
      "epoch": 0.15349887133182843,
      "grad_norm": 0.9883130788803101,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.1225,
      "step": 51
    },
    {
      "epoch": 0.1565086531226486,
      "grad_norm": 1.146127462387085,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.7079,
      "step": 52
    },
    {
      "epoch": 0.1595184349134688,
      "grad_norm": 0.7247859239578247,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.4953,
      "step": 53
    },
    {
      "epoch": 0.16252821670428894,
      "grad_norm": 0.6606523394584656,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.6936,
      "step": 54
    },
    {
      "epoch": 0.1655379984951091,
      "grad_norm": 0.661456286907196,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.6561,
      "step": 55
    },
    {
      "epoch": 0.16854778028592926,
      "grad_norm": 0.6059918999671936,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.529,
      "step": 56
    },
    {
      "epoch": 0.17155756207674944,
      "grad_norm": 0.7299861311912537,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.8652,
      "step": 57
    },
    {
      "epoch": 0.1745673438675696,
      "grad_norm": 0.6570218205451965,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.6003,
      "step": 58
    },
    {
      "epoch": 0.17757712565838976,
      "grad_norm": 0.7697155475616455,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.9703,
      "step": 59
    },
    {
      "epoch": 0.18058690744920994,
      "grad_norm": 0.8205830454826355,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.719,
      "step": 60
    },
    {
      "epoch": 0.1835966892400301,
      "grad_norm": 0.8287568092346191,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.7445,
      "step": 61
    },
    {
      "epoch": 0.18660647103085026,
      "grad_norm": 0.7720220685005188,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.6071,
      "step": 62
    },
    {
      "epoch": 0.18961625282167044,
      "grad_norm": 0.8769961595535278,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.9422,
      "step": 63
    },
    {
      "epoch": 0.1926260346124906,
      "grad_norm": 0.7806358337402344,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.7808,
      "step": 64
    },
    {
      "epoch": 0.19563581640331076,
      "grad_norm": 0.7959004044532776,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.6023,
      "step": 65
    },
    {
      "epoch": 0.1986455981941309,
      "grad_norm": 0.9200003147125244,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.8079,
      "step": 66
    },
    {
      "epoch": 0.2016553799849511,
      "grad_norm": 0.9827006459236145,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.9431,
      "step": 67
    },
    {
      "epoch": 0.20466516177577126,
      "grad_norm": 0.9591490626335144,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.7521,
      "step": 68
    },
    {
      "epoch": 0.2076749435665914,
      "grad_norm": 0.9347128868103027,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.6955,
      "step": 69
    },
    {
      "epoch": 0.2106847253574116,
      "grad_norm": 0.8477599620819092,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.8735,
      "step": 70
    },
    {
      "epoch": 0.21369450714823177,
      "grad_norm": 0.8589199781417847,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.7021,
      "step": 71
    },
    {
      "epoch": 0.21670428893905191,
      "grad_norm": 0.8839547634124756,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.725,
      "step": 72
    },
    {
      "epoch": 0.2197140707298721,
      "grad_norm": 1.0816020965576172,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.9448,
      "step": 73
    },
    {
      "epoch": 0.22272385252069224,
      "grad_norm": 0.9550137519836426,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.673,
      "step": 74
    },
    {
      "epoch": 0.22573363431151242,
      "grad_norm": 1.0094174146652222,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.6478,
      "step": 75
    },
    {
      "epoch": 0.2287434161023326,
      "grad_norm": 0.9577383995056152,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.949,
      "step": 76
    },
    {
      "epoch": 0.23175319789315274,
      "grad_norm": 1.0763578414916992,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.752,
      "step": 77
    },
    {
      "epoch": 0.23476297968397292,
      "grad_norm": 1.019482970237732,
      "learning_rate": 7.158771761692464e-05,
      "loss": 3.0192,
      "step": 78
    },
    {
      "epoch": 0.23777276147479307,
      "grad_norm": 1.2120007276535034,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.958,
      "step": 79
    },
    {
      "epoch": 0.24078254326561324,
      "grad_norm": 1.0528912544250488,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.6808,
      "step": 80
    },
    {
      "epoch": 0.24379232505643342,
      "grad_norm": 1.070916771888733,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.8994,
      "step": 81
    },
    {
      "epoch": 0.24680210684725357,
      "grad_norm": 0.9918401837348938,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.6755,
      "step": 82
    },
    {
      "epoch": 0.24981188863807374,
      "grad_norm": 1.4494421482086182,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.9853,
      "step": 83
    },
    {
      "epoch": 0.2528216704288939,
      "grad_norm": 1.0939158201217651,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.7875,
      "step": 84
    },
    {
      "epoch": 0.2558314522197141,
      "grad_norm": 1.1803234815597534,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.87,
      "step": 85
    },
    {
      "epoch": 0.2588412340105342,
      "grad_norm": 1.224190354347229,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.7292,
      "step": 86
    },
    {
      "epoch": 0.2618510158013544,
      "grad_norm": 1.1455833911895752,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.7313,
      "step": 87
    },
    {
      "epoch": 0.26486079759217457,
      "grad_norm": 1.1967684030532837,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.8791,
      "step": 88
    },
    {
      "epoch": 0.26787057938299474,
      "grad_norm": 1.4523427486419678,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.8266,
      "step": 89
    },
    {
      "epoch": 0.2708803611738149,
      "grad_norm": 1.6074488162994385,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.7747,
      "step": 90
    },
    {
      "epoch": 0.27389014296463504,
      "grad_norm": 1.4523072242736816,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.9463,
      "step": 91
    },
    {
      "epoch": 0.2768999247554552,
      "grad_norm": 1.4732896089553833,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.864,
      "step": 92
    },
    {
      "epoch": 0.2799097065462754,
      "grad_norm": 1.4440211057662964,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.5142,
      "step": 93
    },
    {
      "epoch": 0.28291948833709557,
      "grad_norm": 1.569728970527649,
      "learning_rate": 5.90440267166055e-05,
      "loss": 3.1005,
      "step": 94
    },
    {
      "epoch": 0.28592927012791575,
      "grad_norm": 1.6300673484802246,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.9026,
      "step": 95
    },
    {
      "epoch": 0.28893905191873587,
      "grad_norm": 1.6838175058364868,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.9086,
      "step": 96
    },
    {
      "epoch": 0.29194883370955604,
      "grad_norm": 1.7202688455581665,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.7964,
      "step": 97
    },
    {
      "epoch": 0.2949586155003762,
      "grad_norm": 1.8333368301391602,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.7565,
      "step": 98
    },
    {
      "epoch": 0.2979683972911964,
      "grad_norm": 1.8898142576217651,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.7308,
      "step": 99
    },
    {
      "epoch": 0.3009781790820166,
      "grad_norm": 3.0437440872192383,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 3.1335,
      "step": 100
    },
    {
      "epoch": 0.3009781790820166,
      "eval_loss": 2.7928898334503174,
      "eval_runtime": 43.5272,
      "eval_samples_per_second": 12.866,
      "eval_steps_per_second": 3.216,
      "step": 100
    },
    {
      "epoch": 0.3039879608728367,
      "grad_norm": 0.9572544693946838,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.59,
      "step": 101
    },
    {
      "epoch": 0.30699774266365687,
      "grad_norm": 0.8997783064842224,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.5526,
      "step": 102
    },
    {
      "epoch": 0.31000752445447705,
      "grad_norm": 0.906589925289154,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.5865,
      "step": 103
    },
    {
      "epoch": 0.3130173062452972,
      "grad_norm": 0.6696784496307373,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.5878,
      "step": 104
    },
    {
      "epoch": 0.3160270880361174,
      "grad_norm": 0.6746740341186523,
      "learning_rate": 5e-05,
      "loss": 2.6586,
      "step": 105
    },
    {
      "epoch": 0.3190368698269376,
      "grad_norm": 0.6169286966323853,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.4803,
      "step": 106
    },
    {
      "epoch": 0.3220466516177577,
      "grad_norm": 0.6375830173492432,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.7054,
      "step": 107
    },
    {
      "epoch": 0.32505643340857787,
      "grad_norm": 0.6389443278312683,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.5025,
      "step": 108
    },
    {
      "epoch": 0.32806621519939805,
      "grad_norm": 0.6818410754203796,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.4627,
      "step": 109
    },
    {
      "epoch": 0.3310759969902182,
      "grad_norm": 0.7902951836585999,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.6796,
      "step": 110
    },
    {
      "epoch": 0.3340857787810384,
      "grad_norm": 0.8217026591300964,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.4694,
      "step": 111
    },
    {
      "epoch": 0.3370955605718585,
      "grad_norm": 0.8203257918357849,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.7957,
      "step": 112
    },
    {
      "epoch": 0.3401053423626787,
      "grad_norm": 0.7445793747901917,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.6715,
      "step": 113
    },
    {
      "epoch": 0.3431151241534989,
      "grad_norm": 0.9404602646827698,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.8712,
      "step": 114
    },
    {
      "epoch": 0.34612490594431905,
      "grad_norm": 0.8628761768341064,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.6455,
      "step": 115
    },
    {
      "epoch": 0.3491346877351392,
      "grad_norm": 0.8209057450294495,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.7571,
      "step": 116
    },
    {
      "epoch": 0.35214446952595935,
      "grad_norm": 0.7805319428443909,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.5658,
      "step": 117
    },
    {
      "epoch": 0.3551542513167795,
      "grad_norm": 0.8453297019004822,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.8783,
      "step": 118
    },
    {
      "epoch": 0.3581640331075997,
      "grad_norm": 0.814822256565094,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.7575,
      "step": 119
    },
    {
      "epoch": 0.3611738148984199,
      "grad_norm": 1.0434857606887817,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.7282,
      "step": 120
    },
    {
      "epoch": 0.36418359668924005,
      "grad_norm": 0.8157389163970947,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.5727,
      "step": 121
    },
    {
      "epoch": 0.3671933784800602,
      "grad_norm": 0.870448887348175,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.643,
      "step": 122
    },
    {
      "epoch": 0.37020316027088035,
      "grad_norm": 0.8797793984413147,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.706,
      "step": 123
    },
    {
      "epoch": 0.3732129420617005,
      "grad_norm": 0.9377344846725464,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.6047,
      "step": 124
    },
    {
      "epoch": 0.3762227238525207,
      "grad_norm": 0.9191809296607971,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.7021,
      "step": 125
    },
    {
      "epoch": 0.3792325056433409,
      "grad_norm": 0.9445900917053223,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.8099,
      "step": 126
    },
    {
      "epoch": 0.382242287434161,
      "grad_norm": 0.9417356848716736,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.8348,
      "step": 127
    },
    {
      "epoch": 0.3852520692249812,
      "grad_norm": 1.145680546760559,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.755,
      "step": 128
    },
    {
      "epoch": 0.38826185101580135,
      "grad_norm": 0.967311441898346,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.7202,
      "step": 129
    },
    {
      "epoch": 0.3912716328066215,
      "grad_norm": 0.9585645198822021,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.8266,
      "step": 130
    },
    {
      "epoch": 0.3942814145974417,
      "grad_norm": 1.0192813873291016,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.7238,
      "step": 131
    },
    {
      "epoch": 0.3972911963882618,
      "grad_norm": 1.1950159072875977,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 3.0407,
      "step": 132
    },
    {
      "epoch": 0.400300978179082,
      "grad_norm": 1.1695165634155273,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.8674,
      "step": 133
    },
    {
      "epoch": 0.4033107599699022,
      "grad_norm": 1.1528035402297974,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.9491,
      "step": 134
    },
    {
      "epoch": 0.40632054176072235,
      "grad_norm": 1.1426727771759033,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.7404,
      "step": 135
    },
    {
      "epoch": 0.40933032355154253,
      "grad_norm": 1.325439453125,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.7394,
      "step": 136
    },
    {
      "epoch": 0.4123401053423627,
      "grad_norm": 1.2166088819503784,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.6504,
      "step": 137
    },
    {
      "epoch": 0.4153498871331828,
      "grad_norm": 1.3485374450683594,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.7567,
      "step": 138
    },
    {
      "epoch": 0.418359668924003,
      "grad_norm": 1.1951407194137573,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.9476,
      "step": 139
    },
    {
      "epoch": 0.4213694507148232,
      "grad_norm": 1.1799910068511963,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.4991,
      "step": 140
    },
    {
      "epoch": 0.42437923250564336,
      "grad_norm": 1.217898964881897,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.7609,
      "step": 141
    },
    {
      "epoch": 0.42738901429646353,
      "grad_norm": 1.485028862953186,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.8643,
      "step": 142
    },
    {
      "epoch": 0.43039879608728365,
      "grad_norm": 1.5121876001358032,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.8997,
      "step": 143
    },
    {
      "epoch": 0.43340857787810383,
      "grad_norm": 1.3564367294311523,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.6157,
      "step": 144
    },
    {
      "epoch": 0.436418359668924,
      "grad_norm": 1.3653913736343384,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.883,
      "step": 145
    },
    {
      "epoch": 0.4394281414597442,
      "grad_norm": 1.441187858581543,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.8226,
      "step": 146
    },
    {
      "epoch": 0.44243792325056436,
      "grad_norm": 1.439168930053711,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.8308,
      "step": 147
    },
    {
      "epoch": 0.4454477050413845,
      "grad_norm": 1.5764445066452026,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.7922,
      "step": 148
    },
    {
      "epoch": 0.44845748683220465,
      "grad_norm": 2.1895360946655273,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.9549,
      "step": 149
    },
    {
      "epoch": 0.45146726862302483,
      "grad_norm": 2.6794886589050293,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.9266,
      "step": 150
    },
    {
      "epoch": 0.45146726862302483,
      "eval_loss": 2.7207462787628174,
      "eval_runtime": 43.5205,
      "eval_samples_per_second": 12.867,
      "eval_steps_per_second": 3.217,
      "step": 150
    },
    {
      "epoch": 0.454477050413845,
      "grad_norm": 0.5827016830444336,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.6062,
      "step": 151
    },
    {
      "epoch": 0.4574868322046652,
      "grad_norm": 0.6507757902145386,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.5759,
      "step": 152
    },
    {
      "epoch": 0.4604966139954853,
      "grad_norm": 0.6897706389427185,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.5331,
      "step": 153
    },
    {
      "epoch": 0.4635063957863055,
      "grad_norm": 0.7376187443733215,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.6529,
      "step": 154
    },
    {
      "epoch": 0.46651617757712566,
      "grad_norm": 4.104633331298828,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.7607,
      "step": 155
    },
    {
      "epoch": 0.46952595936794583,
      "grad_norm": 0.7048830389976501,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.7224,
      "step": 156
    },
    {
      "epoch": 0.472535741158766,
      "grad_norm": 0.6888774037361145,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.6512,
      "step": 157
    },
    {
      "epoch": 0.47554552294958613,
      "grad_norm": 0.7250568270683289,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.767,
      "step": 158
    },
    {
      "epoch": 0.4785553047404063,
      "grad_norm": 0.6772659420967102,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.6853,
      "step": 159
    },
    {
      "epoch": 0.4815650865312265,
      "grad_norm": 0.6967433094978333,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.6472,
      "step": 160
    },
    {
      "epoch": 0.48457486832204666,
      "grad_norm": 0.7274630069732666,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.6822,
      "step": 161
    },
    {
      "epoch": 0.48758465011286684,
      "grad_norm": 0.8335841298103333,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.6068,
      "step": 162
    },
    {
      "epoch": 0.49059443190368696,
      "grad_norm": 0.7825259566307068,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.7789,
      "step": 163
    },
    {
      "epoch": 0.49360421369450713,
      "grad_norm": 0.7136161923408508,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.4088,
      "step": 164
    },
    {
      "epoch": 0.4966139954853273,
      "grad_norm": 0.7300835847854614,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.6801,
      "step": 165
    },
    {
      "epoch": 0.4996237772761475,
      "grad_norm": 0.7578651309013367,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.6223,
      "step": 166
    },
    {
      "epoch": 0.5026335590669676,
      "grad_norm": 0.9058271646499634,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.6618,
      "step": 167
    },
    {
      "epoch": 0.5056433408577878,
      "grad_norm": 0.7805584669113159,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.7429,
      "step": 168
    },
    {
      "epoch": 0.508653122648608,
      "grad_norm": 0.8842304944992065,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.5051,
      "step": 169
    },
    {
      "epoch": 0.5116629044394282,
      "grad_norm": 0.8665125966072083,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.7933,
      "step": 170
    },
    {
      "epoch": 0.5146726862302483,
      "grad_norm": 0.9622229337692261,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.7892,
      "step": 171
    },
    {
      "epoch": 0.5176824680210684,
      "grad_norm": 0.9020377993583679,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.6627,
      "step": 172
    },
    {
      "epoch": 0.5206922498118887,
      "grad_norm": 0.7896916270256042,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.4676,
      "step": 173
    },
    {
      "epoch": 0.5237020316027088,
      "grad_norm": 0.8899149894714355,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.5136,
      "step": 174
    },
    {
      "epoch": 0.526711813393529,
      "grad_norm": 0.9398487210273743,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.6724,
      "step": 175
    },
    {
      "epoch": 0.5297215951843491,
      "grad_norm": 0.8152235746383667,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.4499,
      "step": 176
    },
    {
      "epoch": 0.5327313769751693,
      "grad_norm": 1.0111510753631592,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.7443,
      "step": 177
    },
    {
      "epoch": 0.5357411587659895,
      "grad_norm": 0.97441166639328,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.5807,
      "step": 178
    },
    {
      "epoch": 0.5387509405568096,
      "grad_norm": 0.9852139949798584,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.787,
      "step": 179
    },
    {
      "epoch": 0.5417607223476298,
      "grad_norm": 1.0015703439712524,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.5823,
      "step": 180
    },
    {
      "epoch": 0.54477050413845,
      "grad_norm": 1.0196603536605835,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.596,
      "step": 181
    },
    {
      "epoch": 0.5477802859292701,
      "grad_norm": 1.1556131839752197,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.8026,
      "step": 182
    },
    {
      "epoch": 0.5507900677200903,
      "grad_norm": 1.1140217781066895,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.7128,
      "step": 183
    },
    {
      "epoch": 0.5537998495109104,
      "grad_norm": 1.183419942855835,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.814,
      "step": 184
    },
    {
      "epoch": 0.5568096313017307,
      "grad_norm": 1.10748291015625,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.676,
      "step": 185
    },
    {
      "epoch": 0.5598194130925508,
      "grad_norm": 1.1944704055786133,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.5415,
      "step": 186
    },
    {
      "epoch": 0.5628291948833709,
      "grad_norm": 1.0653362274169922,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.6195,
      "step": 187
    },
    {
      "epoch": 0.5658389766741911,
      "grad_norm": 1.2596079111099243,
      "learning_rate": 9.810017062595322e-07,
      "loss": 3.0578,
      "step": 188
    },
    {
      "epoch": 0.5688487584650113,
      "grad_norm": 1.1765263080596924,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.8044,
      "step": 189
    },
    {
      "epoch": 0.5718585402558315,
      "grad_norm": 1.3503178358078003,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.9753,
      "step": 190
    },
    {
      "epoch": 0.5748683220466516,
      "grad_norm": 1.3773014545440674,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.8056,
      "step": 191
    },
    {
      "epoch": 0.5778781038374717,
      "grad_norm": 1.4599443674087524,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.94,
      "step": 192
    },
    {
      "epoch": 0.580887885628292,
      "grad_norm": 1.3997405767440796,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.6367,
      "step": 193
    },
    {
      "epoch": 0.5838976674191121,
      "grad_norm": 1.42356538772583,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.9428,
      "step": 194
    },
    {
      "epoch": 0.5869074492099323,
      "grad_norm": 1.5371516942977905,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.755,
      "step": 195
    },
    {
      "epoch": 0.5899172310007524,
      "grad_norm": 1.7925596237182617,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 3.057,
      "step": 196
    },
    {
      "epoch": 0.5929270127915726,
      "grad_norm": 2.0468907356262207,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.9634,
      "step": 197
    },
    {
      "epoch": 0.5959367945823928,
      "grad_norm": 2.079108953475952,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 3.0522,
      "step": 198
    },
    {
      "epoch": 0.5989465763732129,
      "grad_norm": 2.6386282444000244,
      "learning_rate": 6.834750376549792e-09,
      "loss": 3.3413,
      "step": 199
    },
    {
      "epoch": 0.6019563581640331,
      "grad_norm": 3.1302988529205322,
      "learning_rate": 0.0,
      "loss": 2.98,
      "step": 200
    },
    {
      "epoch": 0.6019563581640331,
      "eval_loss": 2.718132734298706,
      "eval_runtime": 43.5214,
      "eval_samples_per_second": 12.867,
      "eval_steps_per_second": 3.217,
      "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": 3.047197429066629e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}