{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 9.652075471698113,
  "eval_steps": 500,
  "global_step": 270,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.036226415094339624,
      "grad_norm": 4.074844837188721,
      "learning_rate": 3.7037037037037036e-07,
      "loss": 0.9495,
      "step": 1
    },
    {
      "epoch": 0.07245283018867925,
      "grad_norm": 4.034702301025391,
      "learning_rate": 7.407407407407407e-07,
      "loss": 0.9536,
      "step": 2
    },
    {
      "epoch": 0.10867924528301887,
      "grad_norm": 4.080567359924316,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.9267,
      "step": 3
    },
    {
      "epoch": 0.1449056603773585,
      "grad_norm": 3.9866671562194824,
      "learning_rate": 1.4814814814814815e-06,
      "loss": 0.95,
      "step": 4
    },
    {
      "epoch": 0.1811320754716981,
      "grad_norm": 4.042410373687744,
      "learning_rate": 1.8518518518518519e-06,
      "loss": 0.995,
      "step": 5
    },
    {
      "epoch": 0.21735849056603773,
      "grad_norm": 4.1409783363342285,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.9579,
      "step": 6
    },
    {
      "epoch": 0.25358490566037734,
      "grad_norm": 3.998582601547241,
      "learning_rate": 2.5925925925925925e-06,
      "loss": 0.9625,
      "step": 7
    },
    {
      "epoch": 0.289811320754717,
      "grad_norm": 4.141905307769775,
      "learning_rate": 2.962962962962963e-06,
      "loss": 0.9506,
      "step": 8
    },
    {
      "epoch": 0.3260377358490566,
      "grad_norm": 4.01569128036499,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.9591,
      "step": 9
    },
    {
      "epoch": 0.3622641509433962,
      "grad_norm": 3.662905216217041,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 0.9519,
      "step": 10
    },
    {
      "epoch": 0.39849056603773586,
      "grad_norm": 2.892972707748413,
      "learning_rate": 4.074074074074074e-06,
      "loss": 0.9484,
      "step": 11
    },
    {
      "epoch": 0.43471698113207546,
      "grad_norm": 2.5980327129364014,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.9322,
      "step": 12
    },
    {
      "epoch": 0.47094339622641507,
      "grad_norm": 2.2705729007720947,
      "learning_rate": 4.814814814814815e-06,
      "loss": 0.9016,
      "step": 13
    },
    {
      "epoch": 0.5071698113207547,
      "grad_norm": 2.027101755142212,
      "learning_rate": 5.185185185185185e-06,
      "loss": 0.9084,
      "step": 14
    },
    {
      "epoch": 0.5433962264150943,
      "grad_norm": 1.2326773405075073,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.9035,
      "step": 15
    },
    {
      "epoch": 0.579622641509434,
      "grad_norm": 1.5498149394989014,
      "learning_rate": 5.925925925925926e-06,
      "loss": 0.8692,
      "step": 16
    },
    {
      "epoch": 0.6158490566037735,
      "grad_norm": 1.6801735162734985,
      "learning_rate": 6.296296296296297e-06,
      "loss": 0.8736,
      "step": 17
    },
    {
      "epoch": 0.6520754716981132,
      "grad_norm": 1.998213529586792,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.8728,
      "step": 18
    },
    {
      "epoch": 0.6883018867924529,
      "grad_norm": 1.8951051235198975,
      "learning_rate": 7.0370370370370375e-06,
      "loss": 0.8715,
      "step": 19
    },
    {
      "epoch": 0.7245283018867924,
      "grad_norm": 1.7730218172073364,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 0.8358,
      "step": 20
    },
    {
      "epoch": 0.7607547169811321,
      "grad_norm": 1.758504867553711,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.8289,
      "step": 21
    },
    {
      "epoch": 0.7969811320754717,
      "grad_norm": 1.978657841682434,
      "learning_rate": 8.148148148148148e-06,
      "loss": 0.8432,
      "step": 22
    },
    {
      "epoch": 0.8332075471698113,
      "grad_norm": 1.5869137048721313,
      "learning_rate": 8.518518518518519e-06,
      "loss": 0.8255,
      "step": 23
    },
    {
      "epoch": 0.8694339622641509,
      "grad_norm": 1.3577728271484375,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.8272,
      "step": 24
    },
    {
      "epoch": 0.9056603773584906,
      "grad_norm": 0.984247088432312,
      "learning_rate": 9.25925925925926e-06,
      "loss": 0.8071,
      "step": 25
    },
    {
      "epoch": 0.9418867924528301,
      "grad_norm": 0.8904944062232971,
      "learning_rate": 9.62962962962963e-06,
      "loss": 0.8054,
      "step": 26
    },
    {
      "epoch": 0.9781132075471698,
      "grad_norm": 0.8401544094085693,
      "learning_rate": 1e-05,
      "loss": 0.8313,
      "step": 27
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8401544094085693,
      "learning_rate": 9.999582149277188e-06,
      "loss": 0.8047,
      "step": 28
    },
    {
      "epoch": 1.0362264150943397,
      "grad_norm": 1.6306538581848145,
      "learning_rate": 9.998328666948437e-06,
      "loss": 0.8103,
      "step": 29
    },
    {
      "epoch": 1.0724528301886793,
      "grad_norm": 1.0208637714385986,
      "learning_rate": 9.996239762521152e-06,
      "loss": 0.7616,
      "step": 30
    },
    {
      "epoch": 1.1086792452830188,
      "grad_norm": 1.045002818107605,
      "learning_rate": 9.993315785135417e-06,
      "loss": 0.7736,
      "step": 31
    },
    {
      "epoch": 1.1449056603773584,
      "grad_norm": 0.8551422357559204,
      "learning_rate": 9.989557223505661e-06,
      "loss": 0.762,
      "step": 32
    },
    {
      "epoch": 1.181132075471698,
      "grad_norm": 0.7436487078666687,
      "learning_rate": 9.98496470583896e-06,
      "loss": 0.7793,
      "step": 33
    },
    {
      "epoch": 1.2173584905660377,
      "grad_norm": 0.6822688579559326,
      "learning_rate": 9.979538999730047e-06,
      "loss": 0.7728,
      "step": 34
    },
    {
      "epoch": 1.2535849056603774,
      "grad_norm": 0.6098222732543945,
      "learning_rate": 9.973281012033009e-06,
      "loss": 0.7362,
      "step": 35
    },
    {
      "epoch": 1.289811320754717,
      "grad_norm": 0.6669561266899109,
      "learning_rate": 9.966191788709716e-06,
      "loss": 0.7052,
      "step": 36
    },
    {
      "epoch": 1.3260377358490567,
      "grad_norm": 0.6505219340324402,
      "learning_rate": 9.958272514655006e-06,
      "loss": 0.7609,
      "step": 37
    },
    {
      "epoch": 1.3622641509433961,
      "grad_norm": 0.737601637840271,
      "learning_rate": 9.949524513498636e-06,
      "loss": 0.7477,
      "step": 38
    },
    {
      "epoch": 1.3984905660377358,
      "grad_norm": 0.5156015753746033,
      "learning_rate": 9.939949247384046e-06,
      "loss": 0.7554,
      "step": 39
    },
    {
      "epoch": 1.4347169811320755,
      "grad_norm": 0.5583709478378296,
      "learning_rate": 9.929548316723983e-06,
      "loss": 0.7526,
      "step": 40
    },
    {
      "epoch": 1.4709433962264151,
      "grad_norm": 0.5988060832023621,
      "learning_rate": 9.918323459933006e-06,
      "loss": 0.7516,
      "step": 41
    },
    {
      "epoch": 1.5071698113207548,
      "grad_norm": 0.5599033832550049,
      "learning_rate": 9.906276553136924e-06,
      "loss": 0.7294,
      "step": 42
    },
    {
      "epoch": 1.5433962264150942,
      "grad_norm": 0.6027616858482361,
      "learning_rate": 9.893409609859221e-06,
      "loss": 0.72,
      "step": 43
    },
    {
      "epoch": 1.579622641509434,
      "grad_norm": 0.6933006048202515,
      "learning_rate": 9.879724780684518e-06,
      "loss": 0.7176,
      "step": 44
    },
    {
      "epoch": 1.6158490566037735,
      "grad_norm": 0.5711022615432739,
      "learning_rate": 9.86522435289912e-06,
      "loss": 0.716,
      "step": 45
    },
    {
      "epoch": 1.6520754716981132,
      "grad_norm": 0.48980799317359924,
      "learning_rate": 9.849910750108718e-06,
      "loss": 0.7333,
      "step": 46
    },
    {
      "epoch": 1.6883018867924529,
      "grad_norm": 0.4940769076347351,
      "learning_rate": 9.833786531833311e-06,
      "loss": 0.7368,
      "step": 47
    },
    {
      "epoch": 1.7245283018867923,
      "grad_norm": 0.4914894998073578,
      "learning_rate": 9.816854393079402e-06,
      "loss": 0.7323,
      "step": 48
    },
    {
      "epoch": 1.7607547169811322,
      "grad_norm": 0.5431811213493347,
      "learning_rate": 9.79911716388956e-06,
      "loss": 0.7226,
      "step": 49
    },
    {
      "epoch": 1.7969811320754716,
      "grad_norm": 0.4794186055660248,
      "learning_rate": 9.7805778088694e-06,
      "loss": 0.7277,
      "step": 50
    },
    {
      "epoch": 1.8332075471698113,
      "grad_norm": 0.5294991731643677,
      "learning_rate": 9.761239426692077e-06,
      "loss": 0.7193,
      "step": 51
    },
    {
      "epoch": 1.869433962264151,
      "grad_norm": 0.48333099484443665,
      "learning_rate": 9.741105249580383e-06,
      "loss": 0.6936,
      "step": 52
    },
    {
      "epoch": 1.9056603773584906,
      "grad_norm": 0.42380550503730774,
      "learning_rate": 9.7201786427665e-06,
      "loss": 0.7183,
      "step": 53
    },
    {
      "epoch": 1.9418867924528302,
      "grad_norm": 0.49902644753456116,
      "learning_rate": 9.698463103929542e-06,
      "loss": 0.7288,
      "step": 54
    },
    {
      "epoch": 1.9781132075471697,
      "grad_norm": 0.44469350576400757,
      "learning_rate": 9.67596226261095e-06,
      "loss": 0.7469,
      "step": 55
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.6417510509490967,
      "learning_rate": 9.652679879607843e-06,
      "loss": 0.6903,
      "step": 56
    },
    {
      "epoch": 2.0362264150943394,
      "grad_norm": 0.5208938121795654,
      "learning_rate": 9.628619846344453e-06,
      "loss": 0.6569,
      "step": 57
    },
    {
      "epoch": 2.0724528301886793,
      "grad_norm": 0.41403067111968994,
      "learning_rate": 9.603786184221693e-06,
      "loss": 0.6837,
      "step": 58
    },
    {
      "epoch": 2.1086792452830188,
      "grad_norm": 0.4388677775859833,
      "learning_rate": 9.578183043945031e-06,
      "loss": 0.6577,
      "step": 59
    },
    {
      "epoch": 2.1449056603773586,
      "grad_norm": 0.38680893182754517,
      "learning_rate": 9.551814704830734e-06,
      "loss": 0.6647,
      "step": 60
    },
    {
      "epoch": 2.181132075471698,
      "grad_norm": 0.41955962777137756,
      "learning_rate": 9.524685574090627e-06,
      "loss": 0.6414,
      "step": 61
    },
    {
      "epoch": 2.2173584905660375,
      "grad_norm": 0.5132808089256287,
      "learning_rate": 9.496800186095466e-06,
      "loss": 0.6397,
      "step": 62
    },
    {
      "epoch": 2.2535849056603774,
      "grad_norm": 0.41187506914138794,
      "learning_rate": 9.468163201617063e-06,
      "loss": 0.6654,
      "step": 63
    },
    {
      "epoch": 2.289811320754717,
      "grad_norm": 0.40094566345214844,
      "learning_rate": 9.438779407049282e-06,
      "loss": 0.6483,
      "step": 64
    },
    {
      "epoch": 2.3260377358490567,
      "grad_norm": 0.42385968565940857,
      "learning_rate": 9.40865371360804e-06,
      "loss": 0.6382,
      "step": 65
    },
    {
      "epoch": 2.362264150943396,
      "grad_norm": 0.42505592107772827,
      "learning_rate": 9.377791156510456e-06,
      "loss": 0.621,
      "step": 66
    },
    {
      "epoch": 2.398490566037736,
      "grad_norm": 0.4004240036010742,
      "learning_rate": 9.346196894133239e-06,
      "loss": 0.6137,
      "step": 67
    },
    {
      "epoch": 2.4347169811320755,
      "grad_norm": 0.44593772292137146,
      "learning_rate": 9.313876207150544e-06,
      "loss": 0.6299,
      "step": 68
    },
    {
      "epoch": 2.470943396226415,
      "grad_norm": 0.40830758213996887,
      "learning_rate": 9.280834497651334e-06,
      "loss": 0.6504,
      "step": 69
    },
    {
      "epoch": 2.507169811320755,
      "grad_norm": 0.4714842438697815,
      "learning_rate": 9.247077288236488e-06,
      "loss": 0.626,
      "step": 70
    },
    {
      "epoch": 2.543396226415094,
      "grad_norm": 0.4523302912712097,
      "learning_rate": 9.212610221095748e-06,
      "loss": 0.6386,
      "step": 71
    },
    {
      "epoch": 2.579622641509434,
      "grad_norm": 0.4274156093597412,
      "learning_rate": 9.177439057064684e-06,
      "loss": 0.6368,
      "step": 72
    },
    {
      "epoch": 2.6158490566037735,
      "grad_norm": 0.3964768350124359,
      "learning_rate": 9.141569674661816e-06,
      "loss": 0.6161,
      "step": 73
    },
    {
      "epoch": 2.6520754716981134,
      "grad_norm": 0.4652266502380371,
      "learning_rate": 9.105008069106093e-06,
      "loss": 0.6208,
      "step": 74
    },
    {
      "epoch": 2.688301886792453,
      "grad_norm": 0.4457603394985199,
      "learning_rate": 9.067760351314838e-06,
      "loss": 0.6407,
      "step": 75
    },
    {
      "epoch": 2.7245283018867923,
      "grad_norm": 0.47072410583496094,
      "learning_rate": 9.029832746882372e-06,
      "loss": 0.6345,
      "step": 76
    },
    {
      "epoch": 2.760754716981132,
      "grad_norm": 0.41980573534965515,
      "learning_rate": 8.991231595039464e-06,
      "loss": 0.6242,
      "step": 77
    },
    {
      "epoch": 2.7969811320754716,
      "grad_norm": 0.8126304745674133,
      "learning_rate": 8.951963347593797e-06,
      "loss": 0.6368,
      "step": 78
    },
    {
      "epoch": 2.8332075471698115,
      "grad_norm": 0.4141775667667389,
      "learning_rate": 8.9120345678516e-06,
      "loss": 0.6299,
      "step": 79
    },
    {
      "epoch": 2.869433962264151,
      "grad_norm": 0.4654167592525482,
      "learning_rate": 8.871451929520662e-06,
      "loss": 0.6064,
      "step": 80
    },
    {
      "epoch": 2.9056603773584904,
      "grad_norm": 0.4674675464630127,
      "learning_rate": 8.83022221559489e-06,
      "loss": 0.6252,
      "step": 81
    },
    {
      "epoch": 2.9418867924528302,
      "grad_norm": 0.46518832445144653,
      "learning_rate": 8.78835231722059e-06,
      "loss": 0.6304,
      "step": 82
    },
    {
      "epoch": 2.9781132075471697,
      "grad_norm": 0.40669646859169006,
      "learning_rate": 8.74584923254468e-06,
      "loss": 0.6323,
      "step": 83
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.40669646859169006,
      "learning_rate": 8.702720065545024e-06,
      "loss": 0.6371,
      "step": 84
    },
    {
      "epoch": 3.0362264150943394,
      "grad_norm": 0.5793619155883789,
      "learning_rate": 8.658972024843063e-06,
      "loss": 0.5837,
      "step": 85
    },
    {
      "epoch": 3.0724528301886793,
      "grad_norm": 0.42956608533859253,
      "learning_rate": 8.614612422498965e-06,
      "loss": 0.5657,
      "step": 86
    },
    {
      "epoch": 3.1086792452830188,
      "grad_norm": 0.3798193037509918,
      "learning_rate": 8.569648672789496e-06,
      "loss": 0.5788,
      "step": 87
    },
    {
      "epoch": 3.1449056603773586,
      "grad_norm": 0.43145015835762024,
      "learning_rate": 8.524088290968781e-06,
      "loss": 0.57,
      "step": 88
    },
    {
      "epoch": 3.181132075471698,
      "grad_norm": 0.40505528450012207,
      "learning_rate": 8.477938892012209e-06,
      "loss": 0.5556,
      "step": 89
    },
    {
      "epoch": 3.2173584905660375,
      "grad_norm": 0.36875954270362854,
      "learning_rate": 8.43120818934367e-06,
      "loss": 0.5566,
      "step": 90
    },
    {
      "epoch": 3.2535849056603774,
      "grad_norm": 0.45734313130378723,
      "learning_rate": 8.38390399354631e-06,
      "loss": 0.5388,
      "step": 91
    },
    {
      "epoch": 3.289811320754717,
      "grad_norm": 0.40550005435943604,
      "learning_rate": 8.336034211057098e-06,
      "loss": 0.5569,
      "step": 92
    },
    {
      "epoch": 3.3260377358490567,
      "grad_norm": 0.3886796832084656,
      "learning_rate": 8.28760684284532e-06,
      "loss": 0.546,
      "step": 93
    },
    {
      "epoch": 3.362264150943396,
      "grad_norm": 0.39407268166542053,
      "learning_rate": 8.238629983075296e-06,
      "loss": 0.5513,
      "step": 94
    },
    {
      "epoch": 3.398490566037736,
      "grad_norm": 0.44064226746559143,
      "learning_rate": 8.18911181775353e-06,
      "loss": 0.5524,
      "step": 95
    },
    {
      "epoch": 3.4347169811320755,
      "grad_norm": 0.45655596256256104,
      "learning_rate": 8.139060623360494e-06,
      "loss": 0.5402,
      "step": 96
    },
    {
      "epoch": 3.470943396226415,
      "grad_norm": 0.44399574398994446,
      "learning_rate": 8.088484765467286e-06,
      "loss": 0.5403,
      "step": 97
    },
    {
      "epoch": 3.507169811320755,
      "grad_norm": 0.46749594807624817,
      "learning_rate": 8.037392697337418e-06,
      "loss": 0.5609,
      "step": 98
    },
    {
      "epoch": 3.543396226415094,
      "grad_norm": 0.39268767833709717,
      "learning_rate": 7.985792958513932e-06,
      "loss": 0.5616,
      "step": 99
    },
    {
      "epoch": 3.579622641509434,
      "grad_norm": 0.4343853294849396,
      "learning_rate": 7.93369417339209e-06,
      "loss": 0.5493,
      "step": 100
    },
    {
      "epoch": 3.6158490566037735,
      "grad_norm": 0.4399295151233673,
      "learning_rate": 7.881105049777902e-06,
      "loss": 0.5118,
      "step": 101
    },
    {
      "epoch": 3.6520754716981134,
      "grad_norm": 0.5278156995773315,
      "learning_rate": 7.828034377432694e-06,
      "loss": 0.5592,
      "step": 102
    },
    {
      "epoch": 3.688301886792453,
      "grad_norm": 0.46402445435523987,
      "learning_rate": 7.774491026603985e-06,
      "loss": 0.5389,
      "step": 103
    },
    {
      "epoch": 3.7245283018867923,
      "grad_norm": 0.4599299430847168,
      "learning_rate": 7.720483946542913e-06,
      "loss": 0.5429,
      "step": 104
    },
    {
      "epoch": 3.760754716981132,
      "grad_norm": 0.4408857822418213,
      "learning_rate": 7.666022164008458e-06,
      "loss": 0.557,
      "step": 105
    },
    {
      "epoch": 3.7969811320754716,
      "grad_norm": 0.41980406641960144,
      "learning_rate": 7.6111147817586925e-06,
      "loss": 0.5434,
      "step": 106
    },
    {
      "epoch": 3.8332075471698115,
      "grad_norm": 0.5317021012306213,
      "learning_rate": 7.5557709770293664e-06,
      "loss": 0.552,
      "step": 107
    },
    {
      "epoch": 3.869433962264151,
      "grad_norm": 0.5827487111091614,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.5203,
      "step": 108
    },
    {
      "epoch": 3.9056603773584904,
      "grad_norm": 0.4034237265586853,
      "learning_rate": 7.443811172247822e-06,
      "loss": 0.5418,
      "step": 109
    },
    {
      "epoch": 3.9418867924528302,
      "grad_norm": 0.4142571985721588,
      "learning_rate": 7.387213885189746e-06,
      "loss": 0.5517,
      "step": 110
    },
    {
      "epoch": 3.9781132075471697,
      "grad_norm": 0.4551447033882141,
      "learning_rate": 7.330217598512696e-06,
      "loss": 0.5426,
      "step": 111
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.5489457845687866,
      "learning_rate": 7.2728318385925035e-06,
      "loss": 0.5209,
      "step": 112
    },
    {
      "epoch": 4.036226415094339,
      "grad_norm": 0.5183960795402527,
      "learning_rate": 7.215066196901676e-06,
      "loss": 0.4952,
      "step": 113
    },
    {
      "epoch": 4.072452830188679,
      "grad_norm": 0.4089732766151428,
      "learning_rate": 7.156930328406268e-06,
      "loss": 0.4918,
      "step": 114
    },
    {
      "epoch": 4.108679245283019,
      "grad_norm": 0.47553887963294983,
      "learning_rate": 7.098433949952146e-06,
      "loss": 0.4745,
      "step": 115
    },
    {
      "epoch": 4.144905660377359,
      "grad_norm": 0.43845227360725403,
      "learning_rate": 7.039586838640918e-06,
      "loss": 0.463,
      "step": 116
    },
    {
      "epoch": 4.181132075471698,
      "grad_norm": 0.4647291898727417,
      "learning_rate": 6.980398830195785e-06,
      "loss": 0.5002,
      "step": 117
    },
    {
      "epoch": 4.2173584905660375,
      "grad_norm": 0.42274776101112366,
      "learning_rate": 6.920879817317588e-06,
      "loss": 0.4634,
      "step": 118
    },
    {
      "epoch": 4.253584905660377,
      "grad_norm": 0.46993109583854675,
      "learning_rate": 6.861039748031351e-06,
      "loss": 0.4527,
      "step": 119
    },
    {
      "epoch": 4.289811320754717,
      "grad_norm": 0.49515578150749207,
      "learning_rate": 6.800888624023552e-06,
      "loss": 0.473,
      "step": 120
    },
    {
      "epoch": 4.326037735849057,
      "grad_norm": 0.4713442027568817,
      "learning_rate": 6.740436498970453e-06,
      "loss": 0.4611,
      "step": 121
    },
    {
      "epoch": 4.362264150943396,
      "grad_norm": 0.4236859977245331,
      "learning_rate": 6.679693476857712e-06,
      "loss": 0.4632,
      "step": 122
    },
    {
      "epoch": 4.398490566037736,
      "grad_norm": 0.4841013252735138,
      "learning_rate": 6.618669710291607e-06,
      "loss": 0.4825,
      "step": 123
    },
    {
      "epoch": 4.434716981132075,
      "grad_norm": 0.5489901304244995,
      "learning_rate": 6.557375398802124e-06,
      "loss": 0.4679,
      "step": 124
    },
    {
      "epoch": 4.470943396226415,
      "grad_norm": 0.49812740087509155,
      "learning_rate": 6.495820787138209e-06,
      "loss": 0.4615,
      "step": 125
    },
    {
      "epoch": 4.507169811320755,
      "grad_norm": 0.3921999931335449,
      "learning_rate": 6.434016163555452e-06,
      "loss": 0.4664,
      "step": 126
    },
    {
      "epoch": 4.543396226415094,
      "grad_norm": 0.478518545627594,
      "learning_rate": 6.371971858096509e-06,
      "loss": 0.4754,
      "step": 127
    },
    {
      "epoch": 4.579622641509434,
      "grad_norm": 0.5080537796020508,
      "learning_rate": 6.30969824086453e-06,
      "loss": 0.442,
      "step": 128
    },
    {
      "epoch": 4.615849056603773,
      "grad_norm": 0.44495469331741333,
      "learning_rate": 6.247205720289907e-06,
      "loss": 0.4527,
      "step": 129
    },
    {
      "epoch": 4.652075471698113,
      "grad_norm": 0.4623711407184601,
      "learning_rate": 6.184504741390596e-06,
      "loss": 0.434,
      "step": 130
    },
    {
      "epoch": 4.688301886792453,
      "grad_norm": 0.38758838176727295,
      "learning_rate": 6.121605784026339e-06,
      "loss": 0.45,
      "step": 131
    },
    {
      "epoch": 4.724528301886792,
      "grad_norm": 0.4601069688796997,
      "learning_rate": 6.058519361147055e-06,
      "loss": 0.4655,
      "step": 132
    },
    {
      "epoch": 4.760754716981132,
      "grad_norm": 0.46330124139785767,
      "learning_rate": 5.995256017035703e-06,
      "loss": 0.4531,
      "step": 133
    },
    {
      "epoch": 4.796981132075472,
      "grad_norm": 0.5022570490837097,
      "learning_rate": 5.931826325545912e-06,
      "loss": 0.465,
      "step": 134
    },
    {
      "epoch": 4.8332075471698115,
      "grad_norm": 0.4626692533493042,
      "learning_rate": 5.8682408883346535e-06,
      "loss": 0.4511,
      "step": 135
    },
    {
      "epoch": 4.869433962264151,
      "grad_norm": 0.3865698575973511,
      "learning_rate": 5.804510333090287e-06,
      "loss": 0.46,
      "step": 136
    },
    {
      "epoch": 4.90566037735849,
      "grad_norm": 0.4084339737892151,
      "learning_rate": 5.740645311756246e-06,
      "loss": 0.4587,
      "step": 137
    },
    {
      "epoch": 4.94188679245283,
      "grad_norm": 0.42995211482048035,
      "learning_rate": 5.6766564987506564e-06,
      "loss": 0.4516,
      "step": 138
    },
    {
      "epoch": 4.97811320754717,
      "grad_norm": 0.41004374623298645,
      "learning_rate": 5.612554589182228e-06,
      "loss": 0.4644,
      "step": 139
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.5633000135421753,
      "learning_rate": 5.548350297062659e-06,
      "loss": 0.4656,
      "step": 140
    },
    {
      "epoch": 5.036226415094339,
      "grad_norm": 0.4353588819503784,
      "learning_rate": 5.484054353515896e-06,
      "loss": 0.3881,
      "step": 141
    },
    {
      "epoch": 5.072452830188679,
      "grad_norm": 0.4253406822681427,
      "learning_rate": 5.419677504984534e-06,
      "loss": 0.3969,
      "step": 142
    },
    {
      "epoch": 5.108679245283019,
      "grad_norm": 0.46061569452285767,
      "learning_rate": 5.3552305114336515e-06,
      "loss": 0.3986,
      "step": 143
    },
    {
      "epoch": 5.144905660377359,
      "grad_norm": 0.4197911024093628,
      "learning_rate": 5.290724144552379e-06,
      "loss": 0.4135,
      "step": 144
    },
    {
      "epoch": 5.181132075471698,
      "grad_norm": 0.40193256735801697,
      "learning_rate": 5.2261691859535325e-06,
      "loss": 0.3908,
      "step": 145
    },
    {
      "epoch": 5.2173584905660375,
      "grad_norm": 0.4688466787338257,
      "learning_rate": 5.161576425371554e-06,
      "loss": 0.3923,
      "step": 146
    },
    {
      "epoch": 5.253584905660377,
      "grad_norm": 0.5070242881774902,
      "learning_rate": 5.096956658859122e-06,
      "loss": 0.3837,
      "step": 147
    },
    {
      "epoch": 5.289811320754717,
      "grad_norm": 0.46500205993652344,
      "learning_rate": 5.032320686982697e-06,
      "loss": 0.3959,
      "step": 148
    },
    {
      "epoch": 5.326037735849057,
      "grad_norm": 1.2324038743972778,
      "learning_rate": 4.967679313017304e-06,
      "loss": 0.3785,
      "step": 149
    },
    {
      "epoch": 5.362264150943396,
      "grad_norm": 0.4382549524307251,
      "learning_rate": 4.903043341140879e-06,
      "loss": 0.3595,
      "step": 150
    },
    {
      "epoch": 5.398490566037736,
      "grad_norm": 0.4779147803783417,
      "learning_rate": 4.838423574628447e-06,
      "loss": 0.369,
      "step": 151
    },
    {
      "epoch": 5.434716981132075,
      "grad_norm": 0.506566047668457,
      "learning_rate": 4.773830814046469e-06,
      "loss": 0.3752,
      "step": 152
    },
    {
      "epoch": 5.470943396226415,
      "grad_norm": 0.4850460886955261,
      "learning_rate": 4.7092758554476215e-06,
      "loss": 0.3805,
      "step": 153
    },
    {
      "epoch": 5.507169811320755,
      "grad_norm": 0.4675082266330719,
      "learning_rate": 4.644769488566351e-06,
      "loss": 0.3696,
      "step": 154
    },
    {
      "epoch": 5.543396226415094,
      "grad_norm": 0.4534352123737335,
      "learning_rate": 4.580322495015466e-06,
      "loss": 0.3937,
      "step": 155
    },
    {
      "epoch": 5.579622641509434,
      "grad_norm": 0.4178565442562103,
      "learning_rate": 4.515945646484105e-06,
      "loss": 0.3673,
      "step": 156
    },
    {
      "epoch": 5.615849056603773,
      "grad_norm": 0.43837058544158936,
      "learning_rate": 4.451649702937343e-06,
      "loss": 0.3932,
      "step": 157
    },
    {
      "epoch": 5.652075471698113,
      "grad_norm": 0.5009051561355591,
      "learning_rate": 4.387445410817774e-06,
      "loss": 0.3767,
      "step": 158
    },
    {
      "epoch": 5.688301886792453,
      "grad_norm": 0.4401375651359558,
      "learning_rate": 4.323343501249346e-06,
      "loss": 0.3836,
      "step": 159
    },
    {
      "epoch": 5.724528301886792,
      "grad_norm": 0.49288874864578247,
      "learning_rate": 4.259354688243758e-06,
      "loss": 0.3735,
      "step": 160
    },
    {
      "epoch": 5.760754716981132,
      "grad_norm": 0.41812247037887573,
      "learning_rate": 4.195489666909714e-06,
      "loss": 0.3632,
      "step": 161
    },
    {
      "epoch": 5.796981132075472,
      "grad_norm": 0.4192414879798889,
      "learning_rate": 4.131759111665349e-06,
      "loss": 0.3742,
      "step": 162
    },
    {
      "epoch": 5.8332075471698115,
      "grad_norm": 0.46856996417045593,
      "learning_rate": 4.06817367445409e-06,
      "loss": 0.3842,
      "step": 163
    },
    {
      "epoch": 5.869433962264151,
      "grad_norm": 0.4367072582244873,
      "learning_rate": 4.004743982964298e-06,
      "loss": 0.3778,
      "step": 164
    },
    {
      "epoch": 5.90566037735849,
      "grad_norm": 0.4130428433418274,
      "learning_rate": 3.941480638852948e-06,
      "loss": 0.392,
      "step": 165
    },
    {
      "epoch": 5.94188679245283,
      "grad_norm": 0.44017598032951355,
      "learning_rate": 3.878394215973663e-06,
      "loss": 0.3997,
      "step": 166
    },
    {
      "epoch": 5.97811320754717,
      "grad_norm": 0.44622254371643066,
      "learning_rate": 3.815495258609404e-06,
      "loss": 0.3827,
      "step": 167
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.44622254371643066,
      "learning_rate": 3.752794279710094e-06,
      "loss": 0.4119,
      "step": 168
    },
    {
      "epoch": 6.036226415094339,
      "grad_norm": 0.6544888615608215,
      "learning_rate": 3.690301759135471e-06,
      "loss": 0.3342,
      "step": 169
    },
    {
      "epoch": 6.072452830188679,
      "grad_norm": 0.48705703020095825,
      "learning_rate": 3.6280281419034934e-06,
      "loss": 0.3163,
      "step": 170
    },
    {
      "epoch": 6.108679245283019,
      "grad_norm": 0.42752087116241455,
      "learning_rate": 3.5659838364445505e-06,
      "loss": 0.3442,
      "step": 171
    },
    {
      "epoch": 6.144905660377359,
      "grad_norm": 0.3857191205024719,
      "learning_rate": 3.504179212861793e-06,
      "loss": 0.3319,
      "step": 172
    },
    {
      "epoch": 6.181132075471698,
      "grad_norm": 0.46637335419654846,
      "learning_rate": 3.442624601197877e-06,
      "loss": 0.3167,
      "step": 173
    },
    {
      "epoch": 6.2173584905660375,
      "grad_norm": 0.44155874848365784,
      "learning_rate": 3.3813302897083955e-06,
      "loss": 0.3032,
      "step": 174
    },
    {
      "epoch": 6.253584905660377,
      "grad_norm": 0.4230786859989166,
      "learning_rate": 3.3203065231422904e-06,
      "loss": 0.3082,
      "step": 175
    },
    {
      "epoch": 6.289811320754717,
      "grad_norm": 0.45516237616539,
      "learning_rate": 3.259563501029548e-06,
      "loss": 0.3323,
      "step": 176
    },
    {
      "epoch": 6.326037735849057,
      "grad_norm": 0.4669758081436157,
      "learning_rate": 3.1991113759764493e-06,
      "loss": 0.3143,
      "step": 177
    },
    {
      "epoch": 6.362264150943396,
      "grad_norm": 0.5168375968933105,
      "learning_rate": 3.1389602519686515e-06,
      "loss": 0.3154,
      "step": 178
    },
    {
      "epoch": 6.398490566037736,
      "grad_norm": 0.4606865644454956,
      "learning_rate": 3.0791201826824117e-06,
      "loss": 0.3067,
      "step": 179
    },
    {
      "epoch": 6.434716981132075,
      "grad_norm": 0.4242306053638458,
      "learning_rate": 3.019601169804216e-06,
      "loss": 0.3326,
      "step": 180
    },
    {
      "epoch": 6.470943396226415,
      "grad_norm": 0.4492356777191162,
      "learning_rate": 2.9604131613590825e-06,
      "loss": 0.3422,
      "step": 181
    },
    {
      "epoch": 6.507169811320755,
      "grad_norm": 0.4301685690879822,
      "learning_rate": 2.901566050047855e-06,
      "loss": 0.3071,
      "step": 182
    },
    {
      "epoch": 6.543396226415094,
      "grad_norm": 0.46162980794906616,
      "learning_rate": 2.843069671593734e-06,
      "loss": 0.3084,
      "step": 183
    },
    {
      "epoch": 6.579622641509434,
      "grad_norm": 0.43854185938835144,
      "learning_rate": 2.784933803098326e-06,
      "loss": 0.3256,
      "step": 184
    },
    {
      "epoch": 6.615849056603773,
      "grad_norm": 0.46818047761917114,
      "learning_rate": 2.7271681614074973e-06,
      "loss": 0.3109,
      "step": 185
    },
    {
      "epoch": 6.652075471698113,
      "grad_norm": 0.3902846872806549,
      "learning_rate": 2.6697824014873076e-06,
      "loss": 0.303,
      "step": 186
    },
    {
      "epoch": 6.688301886792453,
      "grad_norm": 0.4288512170314789,
      "learning_rate": 2.6127861148102552e-06,
      "loss": 0.2988,
      "step": 187
    },
    {
      "epoch": 6.724528301886792,
      "grad_norm": 0.3952915668487549,
      "learning_rate": 2.5561888277521797e-06,
      "loss": 0.3003,
      "step": 188
    },
    {
      "epoch": 6.760754716981132,
      "grad_norm": 0.4569963812828064,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.3145,
      "step": 189
    },
    {
      "epoch": 6.796981132075472,
      "grad_norm": 0.448163241147995,
      "learning_rate": 2.4442290229706344e-06,
      "loss": 0.3302,
      "step": 190
    },
    {
      "epoch": 6.8332075471698115,
      "grad_norm": 0.44628778100013733,
      "learning_rate": 2.3888852182413087e-06,
      "loss": 0.3256,
      "step": 191
    },
    {
      "epoch": 6.869433962264151,
      "grad_norm": 0.4496423900127411,
      "learning_rate": 2.333977835991545e-06,
      "loss": 0.3399,
      "step": 192
    },
    {
      "epoch": 6.90566037735849,
      "grad_norm": 0.4480762481689453,
      "learning_rate": 2.2795160534570866e-06,
      "loss": 0.3202,
      "step": 193
    },
    {
      "epoch": 6.94188679245283,
      "grad_norm": 0.4092267155647278,
      "learning_rate": 2.2255089733960162e-06,
      "loss": 0.3302,
      "step": 194
    },
    {
      "epoch": 6.97811320754717,
      "grad_norm": 0.43138426542282104,
      "learning_rate": 2.171965622567308e-06,
      "loss": 0.3106,
      "step": 195
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.5887420773506165,
      "learning_rate": 2.1188949502220987e-06,
      "loss": 0.3219,
      "step": 196
    },
    {
      "epoch": 7.036226415094339,
      "grad_norm": 0.556282103061676,
      "learning_rate": 2.066305826607911e-06,
      "loss": 0.3141,
      "step": 197
    },
    {
      "epoch": 7.072452830188679,
      "grad_norm": 0.4548152983188629,
      "learning_rate": 2.0142070414860704e-06,
      "loss": 0.2697,
      "step": 198
    },
    {
      "epoch": 7.108679245283019,
      "grad_norm": 0.45477020740509033,
      "learning_rate": 1.962607302662582e-06,
      "loss": 0.27,
      "step": 199
    },
    {
      "epoch": 7.144905660377359,
      "grad_norm": 0.3835766911506653,
      "learning_rate": 1.9115152345327154e-06,
      "loss": 0.2718,
      "step": 200
    },
    {
      "epoch": 7.181132075471698,
      "grad_norm": 0.4106207489967346,
      "learning_rate": 1.8609393766395083e-06,
      "loss": 0.2902,
      "step": 201
    },
    {
      "epoch": 7.2173584905660375,
      "grad_norm": 0.40980860590934753,
      "learning_rate": 1.8108881822464697e-06,
      "loss": 0.294,
      "step": 202
    },
    {
      "epoch": 7.253584905660377,
      "grad_norm": 0.4507136940956116,
      "learning_rate": 1.7613700169247055e-06,
      "loss": 0.2941,
      "step": 203
    },
    {
      "epoch": 7.289811320754717,
      "grad_norm": 0.3944286108016968,
      "learning_rate": 1.7123931571546826e-06,
      "loss": 0.2596,
      "step": 204
    },
    {
      "epoch": 7.326037735849057,
      "grad_norm": 0.42168113589286804,
      "learning_rate": 1.6639657889429017e-06,
      "loss": 0.2757,
      "step": 205
    },
    {
      "epoch": 7.362264150943396,
      "grad_norm": 0.45149892568588257,
      "learning_rate": 1.6160960064536907e-06,
      "loss": 0.2618,
      "step": 206
    },
    {
      "epoch": 7.398490566037736,
      "grad_norm": 0.4606323540210724,
      "learning_rate": 1.5687918106563326e-06,
      "loss": 0.2756,
      "step": 207
    },
    {
      "epoch": 7.434716981132075,
      "grad_norm": 0.5498378872871399,
      "learning_rate": 1.52206110798779e-06,
      "loss": 0.2818,
      "step": 208
    },
    {
      "epoch": 7.470943396226415,
      "grad_norm": 0.4124191403388977,
      "learning_rate": 1.4759117090312197e-06,
      "loss": 0.2731,
      "step": 209
    },
    {
      "epoch": 7.507169811320755,
      "grad_norm": 0.44834864139556885,
      "learning_rate": 1.4303513272105057e-06,
      "loss": 0.2687,
      "step": 210
    },
    {
      "epoch": 7.543396226415094,
      "grad_norm": 0.39862528443336487,
      "learning_rate": 1.3853875775010355e-06,
      "loss": 0.2688,
      "step": 211
    },
    {
      "epoch": 7.579622641509434,
      "grad_norm": 0.48807084560394287,
      "learning_rate": 1.3410279751569399e-06,
      "loss": 0.3119,
      "step": 212
    },
    {
      "epoch": 7.615849056603773,
      "grad_norm": 0.44604766368865967,
      "learning_rate": 1.297279934454978e-06,
      "loss": 0.2621,
      "step": 213
    },
    {
      "epoch": 7.652075471698113,
      "grad_norm": 0.3903985619544983,
      "learning_rate": 1.25415076745532e-06,
      "loss": 0.2745,
      "step": 214
    },
    {
      "epoch": 7.688301886792453,
      "grad_norm": 0.4241688549518585,
      "learning_rate": 1.2116476827794104e-06,
      "loss": 0.264,
      "step": 215
    },
    {
      "epoch": 7.724528301886792,
      "grad_norm": 0.4157409369945526,
      "learning_rate": 1.1697777844051105e-06,
      "loss": 0.2695,
      "step": 216
    },
    {
      "epoch": 7.760754716981132,
      "grad_norm": 0.42209798097610474,
      "learning_rate": 1.1285480704793378e-06,
      "loss": 0.293,
      "step": 217
    },
    {
      "epoch": 7.796981132075472,
      "grad_norm": 0.39040008187294006,
      "learning_rate": 1.0879654321484012e-06,
      "loss": 0.2443,
      "step": 218
    },
    {
      "epoch": 7.8332075471698115,
      "grad_norm": 0.4144212603569031,
      "learning_rate": 1.0480366524062041e-06,
      "loss": 0.2747,
      "step": 219
    },
    {
      "epoch": 7.869433962264151,
      "grad_norm": 0.38053098320961,
      "learning_rate": 1.008768404960535e-06,
      "loss": 0.2634,
      "step": 220
    },
    {
      "epoch": 7.90566037735849,
      "grad_norm": 0.4866553246974945,
      "learning_rate": 9.701672531176287e-07,
      "loss": 0.2883,
      "step": 221
    },
    {
      "epoch": 7.94188679245283,
      "grad_norm": 0.401796817779541,
      "learning_rate": 9.322396486851626e-07,
      "loss": 0.2885,
      "step": 222
    },
    {
      "epoch": 7.97811320754717,
      "grad_norm": 0.4356318414211273,
      "learning_rate": 8.949919308939081e-07,
      "loss": 0.2985,
      "step": 223
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.4356318414211273,
      "learning_rate": 8.584303253381848e-07,
      "loss": 0.2454,
      "step": 224
    },
    {
      "epoch": 8.03622641509434,
      "grad_norm": 0.5875634551048279,
      "learning_rate": 8.225609429353187e-07,
      "loss": 0.2752,
      "step": 225
    },
    {
      "epoch": 8.072452830188679,
      "grad_norm": 0.4065380394458771,
      "learning_rate": 7.873897789042523e-07,
      "loss": 0.2725,
      "step": 226
    },
    {
      "epoch": 8.10867924528302,
      "grad_norm": 0.47121724486351013,
      "learning_rate": 7.529227117635135e-07,
      "loss": 0.2802,
      "step": 227
    },
    {
      "epoch": 8.144905660377358,
      "grad_norm": 0.41164955496788025,
      "learning_rate": 7.191655023486682e-07,
      "loss": 0.2475,
      "step": 228
    },
    {
      "epoch": 8.181132075471698,
      "grad_norm": 0.40498465299606323,
      "learning_rate": 6.86123792849458e-07,
      "loss": 0.2585,
      "step": 229
    },
    {
      "epoch": 8.217358490566038,
      "grad_norm": 0.39466381072998047,
      "learning_rate": 6.53803105866761e-07,
      "loss": 0.2552,
      "step": 230
    },
    {
      "epoch": 8.253584905660377,
      "grad_norm": 0.3694005310535431,
      "learning_rate": 6.222088434895462e-07,
      "loss": 0.266,
      "step": 231
    },
    {
      "epoch": 8.289811320754717,
      "grad_norm": 0.3654123544692993,
      "learning_rate": 5.9134628639196e-07,
      "loss": 0.261,
      "step": 232
    },
    {
      "epoch": 8.326037735849056,
      "grad_norm": 0.3835010230541229,
      "learning_rate": 5.612205929507209e-07,
      "loss": 0.262,
      "step": 233
    },
    {
      "epoch": 8.362264150943396,
      "grad_norm": 0.4232535660266876,
      "learning_rate": 5.318367983829393e-07,
      "loss": 0.2443,
      "step": 234
    },
    {
      "epoch": 8.398490566037736,
      "grad_norm": 0.3865967392921448,
      "learning_rate": 5.031998139045352e-07,
      "loss": 0.2479,
      "step": 235
    },
    {
      "epoch": 8.434716981132075,
      "grad_norm": 0.3988674581050873,
      "learning_rate": 4.753144259093734e-07,
      "loss": 0.2569,
      "step": 236
    },
    {
      "epoch": 8.470943396226415,
      "grad_norm": 0.40713047981262207,
      "learning_rate": 4.481852951692672e-07,
      "loss": 0.2446,
      "step": 237
    },
    {
      "epoch": 8.507169811320754,
      "grad_norm": 0.3784768283367157,
      "learning_rate": 4.2181695605497066e-07,
      "loss": 0.2561,
      "step": 238
    },
    {
      "epoch": 8.543396226415094,
      "grad_norm": 0.39897289872169495,
      "learning_rate": 3.9621381577830855e-07,
      "loss": 0.2504,
      "step": 239
    },
    {
      "epoch": 8.579622641509435,
      "grad_norm": 0.40435102581977844,
      "learning_rate": 3.7138015365554834e-07,
      "loss": 0.2572,
      "step": 240
    },
    {
      "epoch": 8.615849056603773,
      "grad_norm": 0.3976576328277588,
      "learning_rate": 3.473201203921578e-07,
      "loss": 0.2622,
      "step": 241
    },
    {
      "epoch": 8.652075471698113,
      "grad_norm": 0.40374553203582764,
      "learning_rate": 3.2403773738905185e-07,
      "loss": 0.2302,
      "step": 242
    },
    {
      "epoch": 8.688301886792452,
      "grad_norm": 0.39839479327201843,
      "learning_rate": 3.015368960704584e-07,
      "loss": 0.2658,
      "step": 243
    },
    {
      "epoch": 8.724528301886792,
      "grad_norm": 0.3632482588291168,
      "learning_rate": 2.798213572335001e-07,
      "loss": 0.2435,
      "step": 244
    },
    {
      "epoch": 8.760754716981133,
      "grad_norm": 0.37311646342277527,
      "learning_rate": 2.5889475041961767e-07,
      "loss": 0.2427,
      "step": 245
    },
    {
      "epoch": 8.796981132075471,
      "grad_norm": 0.43890783190727234,
      "learning_rate": 2.3876057330792344e-07,
      "loss": 0.2576,
      "step": 246
    },
    {
      "epoch": 8.833207547169811,
      "grad_norm": 0.3749382197856903,
      "learning_rate": 2.1942219113060215e-07,
      "loss": 0.2588,
      "step": 247
    },
    {
      "epoch": 8.86943396226415,
      "grad_norm": 0.41522374749183655,
      "learning_rate": 2.0088283611044034e-07,
      "loss": 0.2526,
      "step": 248
    },
    {
      "epoch": 8.90566037735849,
      "grad_norm": 0.3923836648464203,
      "learning_rate": 1.8314560692059836e-07,
      "loss": 0.2504,
      "step": 249
    },
    {
      "epoch": 8.94188679245283,
      "grad_norm": 0.40502244234085083,
      "learning_rate": 1.6621346816668993e-07,
      "loss": 0.2409,
      "step": 250
    },
    {
      "epoch": 8.97811320754717,
      "grad_norm": 0.391658753156662,
      "learning_rate": 1.500892498912826e-07,
      "loss": 0.2607,
      "step": 251
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.5146093964576721,
      "learning_rate": 1.3477564710088097e-07,
      "loss": 0.2573,
      "step": 252
    },
    {
      "epoch": 9.03622641509434,
      "grad_norm": 0.5422486662864685,
      "learning_rate": 1.2027521931548214e-07,
      "loss": 0.2379,
      "step": 253
    },
    {
      "epoch": 9.072452830188679,
      "grad_norm": 0.40020889043807983,
      "learning_rate": 1.0659039014077943e-07,
      "loss": 0.2412,
      "step": 254
    },
    {
      "epoch": 9.10867924528302,
      "grad_norm": 0.3687814772129059,
      "learning_rate": 9.372344686307655e-08,
      "loss": 0.2409,
      "step": 255
    },
    {
      "epoch": 9.144905660377358,
      "grad_norm": 0.380043089389801,
      "learning_rate": 8.167654006699444e-08,
      "loss": 0.2699,
      "step": 256
    },
    {
      "epoch": 9.181132075471698,
      "grad_norm": 0.39074811339378357,
      "learning_rate": 7.04516832760177e-08,
      "loss": 0.2489,
      "step": 257
    },
    {
      "epoch": 9.217358490566038,
      "grad_norm": 0.38397496938705444,
      "learning_rate": 6.005075261595495e-08,
      "loss": 0.2585,
      "step": 258
    },
    {
      "epoch": 9.253584905660377,
      "grad_norm": 0.4563198685646057,
      "learning_rate": 5.047548650136513e-08,
      "loss": 0.2351,
      "step": 259
    },
    {
      "epoch": 9.289811320754717,
      "grad_norm": 0.3548761308193207,
      "learning_rate": 4.172748534499449e-08,
      "loss": 0.2508,
      "step": 260
    },
    {
      "epoch": 9.326037735849056,
      "grad_norm": 0.41531455516815186,
      "learning_rate": 3.3808211290284886e-08,
      "loss": 0.2534,
      "step": 261
    },
    {
      "epoch": 9.362264150943396,
      "grad_norm": 0.34948405623435974,
      "learning_rate": 2.6718987966992683e-08,
      "loss": 0.2497,
      "step": 262
    },
    {
      "epoch": 9.398490566037736,
      "grad_norm": 0.3671889305114746,
      "learning_rate": 2.0461000269953457e-08,
      "loss": 0.2317,
      "step": 263
    },
    {
      "epoch": 9.434716981132075,
      "grad_norm": 0.43335914611816406,
      "learning_rate": 1.5035294161039882e-08,
      "loss": 0.2353,
      "step": 264
    },
    {
      "epoch": 9.470943396226415,
      "grad_norm": 0.41739046573638916,
      "learning_rate": 1.044277649433989e-08,
      "loss": 0.2589,
      "step": 265
    },
    {
      "epoch": 9.507169811320754,
      "grad_norm": 0.36336734890937805,
      "learning_rate": 6.6842148645840374e-09,
      "loss": 0.2228,
      "step": 266
    },
    {
      "epoch": 9.543396226415094,
      "grad_norm": 0.3958076238632202,
      "learning_rate": 3.760237478849793e-09,
      "loss": 0.2405,
      "step": 267
    },
    {
      "epoch": 9.579622641509435,
      "grad_norm": 0.40326353907585144,
      "learning_rate": 1.6713330515627512e-09,
      "loss": 0.2645,
      "step": 268
    },
    {
      "epoch": 9.615849056603773,
      "grad_norm": 0.38379284739494324,
      "learning_rate": 4.178507228136397e-10,
      "loss": 0.2699,
      "step": 269
    },
    {
      "epoch": 9.652075471698113,
      "grad_norm": 0.4049612283706665,
      "learning_rate": 0.0,
      "loss": 0.2758,
      "step": 270
    },
    {
      "epoch": 9.652075471698113,
      "step": 270,
      "total_flos": 99687939342336.0,
      "train_loss": 0.48553379895510496,
      "train_runtime": 12266.0559,
      "train_samples_per_second": 1.08,
      "train_steps_per_second": 0.022
    }
  ],
  "logging_steps": 1,
  "max_steps": 270,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 100000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 99687939342336.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}