{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.2997002997002997,
  "eval_steps": 375,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001998001998001998,
      "grad_norm": 1.1832072734832764,
      "learning_rate": 2e-05,
      "loss": 3.4001,
      "step": 1
    },
    {
      "epoch": 0.0003996003996003996,
      "grad_norm": 4.091126441955566,
      "learning_rate": 4e-05,
      "loss": 8.7337,
      "step": 2
    },
    {
      "epoch": 0.0005994005994005994,
      "grad_norm": 2.625993490219116,
      "learning_rate": 6e-05,
      "loss": 5.4228,
      "step": 3
    },
    {
      "epoch": 0.0007992007992007992,
      "grad_norm": 2.678379535675049,
      "learning_rate": 8e-05,
      "loss": 5.4312,
      "step": 4
    },
    {
      "epoch": 0.000999000999000999,
      "grad_norm": 2.724587917327881,
      "learning_rate": 0.0001,
      "loss": 4.7989,
      "step": 5
    },
    {
      "epoch": 0.0011988011988011988,
      "grad_norm": 1.209551453590393,
      "learning_rate": 0.00012,
      "loss": 2.959,
      "step": 6
    },
    {
      "epoch": 0.0013986013986013986,
      "grad_norm": 3.961056709289551,
      "learning_rate": 0.00014,
      "loss": 4.9851,
      "step": 7
    },
    {
      "epoch": 0.0015984015984015984,
      "grad_norm": 1.544677972793579,
      "learning_rate": 0.00016,
      "loss": 3.3173,
      "step": 8
    },
    {
      "epoch": 0.0017982017982017982,
      "grad_norm": 1.7183178663253784,
      "learning_rate": 0.00018,
      "loss": 2.6315,
      "step": 9
    },
    {
      "epoch": 0.001998001998001998,
      "grad_norm": 2.1965956687927246,
      "learning_rate": 0.0002,
      "loss": 3.1788,
      "step": 10
    },
    {
      "epoch": 0.002197802197802198,
      "grad_norm": 2.1976754665374756,
      "learning_rate": 0.00019999977772170748,
      "loss": 2.515,
      "step": 11
    },
    {
      "epoch": 0.0023976023976023976,
      "grad_norm": 3.281036853790283,
      "learning_rate": 0.00019999911088781805,
      "loss": 2.6452,
      "step": 12
    },
    {
      "epoch": 0.0025974025974025974,
      "grad_norm": 5.566474914550781,
      "learning_rate": 0.0001999979995012962,
      "loss": 3.5328,
      "step": 13
    },
    {
      "epoch": 0.002797202797202797,
      "grad_norm": 6.349402904510498,
      "learning_rate": 0.00019999644356708261,
      "loss": 2.7048,
      "step": 14
    },
    {
      "epoch": 0.002997002997002997,
      "grad_norm": 3.808734178543091,
      "learning_rate": 0.00019999444309209432,
      "loss": 2.47,
      "step": 15
    },
    {
      "epoch": 0.003196803196803197,
      "grad_norm": 6.101389408111572,
      "learning_rate": 0.0001999919980852246,
      "loss": 2.5584,
      "step": 16
    },
    {
      "epoch": 0.0033966033966033966,
      "grad_norm": 3.816959857940674,
      "learning_rate": 0.00019998910855734288,
      "loss": 2.1296,
      "step": 17
    },
    {
      "epoch": 0.0035964035964035964,
      "grad_norm": 4.79175329208374,
      "learning_rate": 0.0001999857745212947,
      "loss": 2.5676,
      "step": 18
    },
    {
      "epoch": 0.003796203796203796,
      "grad_norm": 4.2344183921813965,
      "learning_rate": 0.00019998199599190178,
      "loss": 2.9587,
      "step": 19
    },
    {
      "epoch": 0.003996003996003996,
      "grad_norm": 2.9176828861236572,
      "learning_rate": 0.0001999777729859618,
      "loss": 2.4689,
      "step": 20
    },
    {
      "epoch": 0.004195804195804196,
      "grad_norm": 3.460623025894165,
      "learning_rate": 0.00019997310552224846,
      "loss": 2.1157,
      "step": 21
    },
    {
      "epoch": 0.004395604395604396,
      "grad_norm": 3.876563310623169,
      "learning_rate": 0.00019996799362151122,
      "loss": 2.6294,
      "step": 22
    },
    {
      "epoch": 0.004595404595404595,
      "grad_norm": 3.7633607387542725,
      "learning_rate": 0.00019996243730647538,
      "loss": 2.8489,
      "step": 23
    },
    {
      "epoch": 0.004795204795204795,
      "grad_norm": 2.670142412185669,
      "learning_rate": 0.00019995643660184191,
      "loss": 2.5668,
      "step": 24
    },
    {
      "epoch": 0.004995004995004995,
      "grad_norm": 2.6036922931671143,
      "learning_rate": 0.00019994999153428737,
      "loss": 1.9949,
      "step": 25
    },
    {
      "epoch": 0.005194805194805195,
      "grad_norm": 2.1248209476470947,
      "learning_rate": 0.00019994310213246368,
      "loss": 2.061,
      "step": 26
    },
    {
      "epoch": 0.005394605394605395,
      "grad_norm": 2.9935054779052734,
      "learning_rate": 0.00019993576842699816,
      "loss": 2.31,
      "step": 27
    },
    {
      "epoch": 0.005594405594405594,
      "grad_norm": 3.189983606338501,
      "learning_rate": 0.0001999279904504933,
      "loss": 2.7955,
      "step": 28
    },
    {
      "epoch": 0.005794205794205794,
      "grad_norm": 2.2985377311706543,
      "learning_rate": 0.00019991976823752653,
      "loss": 2.9887,
      "step": 29
    },
    {
      "epoch": 0.005994005994005994,
      "grad_norm": 1.364545226097107,
      "learning_rate": 0.00019991110182465032,
      "loss": 2.9207,
      "step": 30
    },
    {
      "epoch": 0.006193806193806194,
      "grad_norm": 2.546379327774048,
      "learning_rate": 0.00019990199125039174,
      "loss": 2.2884,
      "step": 31
    },
    {
      "epoch": 0.006393606393606394,
      "grad_norm": 1.2872494459152222,
      "learning_rate": 0.00019989243655525247,
      "loss": 1.9876,
      "step": 32
    },
    {
      "epoch": 0.006593406593406593,
      "grad_norm": 1.1356850862503052,
      "learning_rate": 0.00019988243778170853,
      "loss": 2.8054,
      "step": 33
    },
    {
      "epoch": 0.006793206793206793,
      "grad_norm": 4.005868911743164,
      "learning_rate": 0.0001998719949742101,
      "loss": 2.4908,
      "step": 34
    },
    {
      "epoch": 0.006993006993006993,
      "grad_norm": 2.206594705581665,
      "learning_rate": 0.0001998611081791814,
      "loss": 1.8848,
      "step": 35
    },
    {
      "epoch": 0.007192807192807193,
      "grad_norm": 4.710446834564209,
      "learning_rate": 0.00019984977744502038,
      "loss": 3.1834,
      "step": 36
    },
    {
      "epoch": 0.007392607392607393,
      "grad_norm": 3.1490848064422607,
      "learning_rate": 0.00019983800282209857,
      "loss": 2.3447,
      "step": 37
    },
    {
      "epoch": 0.007592407592407592,
      "grad_norm": 2.809861660003662,
      "learning_rate": 0.00019982578436276082,
      "loss": 2.6769,
      "step": 38
    },
    {
      "epoch": 0.007792207792207792,
      "grad_norm": 1.5440502166748047,
      "learning_rate": 0.00019981312212132512,
      "loss": 1.8814,
      "step": 39
    },
    {
      "epoch": 0.007992007992007992,
      "grad_norm": 4.79085636138916,
      "learning_rate": 0.00019980001615408228,
      "loss": 2.6759,
      "step": 40
    },
    {
      "epoch": 0.008191808191808193,
      "grad_norm": 2.4709253311157227,
      "learning_rate": 0.00019978646651929572,
      "loss": 2.243,
      "step": 41
    },
    {
      "epoch": 0.008391608391608392,
      "grad_norm": 0.944856584072113,
      "learning_rate": 0.00019977247327720128,
      "loss": 2.0041,
      "step": 42
    },
    {
      "epoch": 0.008591408591408592,
      "grad_norm": 2.451951742172241,
      "learning_rate": 0.0001997580364900068,
      "loss": 1.8252,
      "step": 43
    },
    {
      "epoch": 0.008791208791208791,
      "grad_norm": 1.4323277473449707,
      "learning_rate": 0.000199743156221892,
      "loss": 1.452,
      "step": 44
    },
    {
      "epoch": 0.008991008991008992,
      "grad_norm": 2.0564022064208984,
      "learning_rate": 0.00019972783253900808,
      "loss": 2.2977,
      "step": 45
    },
    {
      "epoch": 0.00919080919080919,
      "grad_norm": 5.169833183288574,
      "learning_rate": 0.00019971206550947748,
      "loss": 1.824,
      "step": 46
    },
    {
      "epoch": 0.009390609390609391,
      "grad_norm": 2.6448116302490234,
      "learning_rate": 0.00019969585520339354,
      "loss": 2.1913,
      "step": 47
    },
    {
      "epoch": 0.00959040959040959,
      "grad_norm": 2.1816165447235107,
      "learning_rate": 0.0001996792016928203,
      "loss": 2.8424,
      "step": 48
    },
    {
      "epoch": 0.009790209790209791,
      "grad_norm": 2.257312774658203,
      "learning_rate": 0.00019966210505179197,
      "loss": 2.8312,
      "step": 49
    },
    {
      "epoch": 0.00999000999000999,
      "grad_norm": 1.8191273212432861,
      "learning_rate": 0.00019964456535631286,
      "loss": 2.3105,
      "step": 50
    },
    {
      "epoch": 0.01018981018981019,
      "grad_norm": 1.68135666847229,
      "learning_rate": 0.0001996265826843568,
      "loss": 2.4698,
      "step": 51
    },
    {
      "epoch": 0.01038961038961039,
      "grad_norm": 2.976138114929199,
      "learning_rate": 0.00019960815711586696,
      "loss": 2.3114,
      "step": 52
    },
    {
      "epoch": 0.01058941058941059,
      "grad_norm": 3.3976998329162598,
      "learning_rate": 0.00019958928873275539,
      "loss": 2.236,
      "step": 53
    },
    {
      "epoch": 0.01078921078921079,
      "grad_norm": 2.0569653511047363,
      "learning_rate": 0.00019956997761890277,
      "loss": 1.266,
      "step": 54
    },
    {
      "epoch": 0.01098901098901099,
      "grad_norm": 3.5636274814605713,
      "learning_rate": 0.00019955022386015792,
      "loss": 3.0822,
      "step": 55
    },
    {
      "epoch": 0.011188811188811189,
      "grad_norm": 3.3449692726135254,
      "learning_rate": 0.00019953002754433743,
      "loss": 2.4288,
      "step": 56
    },
    {
      "epoch": 0.01138861138861139,
      "grad_norm": 3.5546164512634277,
      "learning_rate": 0.00019950938876122542,
      "loss": 2.0962,
      "step": 57
    },
    {
      "epoch": 0.011588411588411588,
      "grad_norm": 2.1120035648345947,
      "learning_rate": 0.00019948830760257291,
      "loss": 1.8609,
      "step": 58
    },
    {
      "epoch": 0.011788211788211789,
      "grad_norm": 1.241166114807129,
      "learning_rate": 0.0001994667841620976,
      "loss": 2.0666,
      "step": 59
    },
    {
      "epoch": 0.011988011988011988,
      "grad_norm": 4.637640476226807,
      "learning_rate": 0.00019944481853548335,
      "loss": 1.9308,
      "step": 60
    },
    {
      "epoch": 0.012187812187812189,
      "grad_norm": 2.2608180046081543,
      "learning_rate": 0.00019942241082037982,
      "loss": 2.0089,
      "step": 61
    },
    {
      "epoch": 0.012387612387612388,
      "grad_norm": 1.5937212705612183,
      "learning_rate": 0.00019939956111640197,
      "loss": 1.8042,
      "step": 62
    },
    {
      "epoch": 0.012587412587412588,
      "grad_norm": 1.371766209602356,
      "learning_rate": 0.00019937626952512964,
      "loss": 2.2133,
      "step": 63
    },
    {
      "epoch": 0.012787212787212787,
      "grad_norm": 3.777963876724243,
      "learning_rate": 0.0001993525361501072,
      "loss": 2.4498,
      "step": 64
    },
    {
      "epoch": 0.012987012987012988,
      "grad_norm": 3.4710748195648193,
      "learning_rate": 0.00019932836109684286,
      "loss": 2.6978,
      "step": 65
    },
    {
      "epoch": 0.013186813186813187,
      "grad_norm": 1.7983630895614624,
      "learning_rate": 0.00019930374447280845,
      "loss": 1.9646,
      "step": 66
    },
    {
      "epoch": 0.013386613386613387,
      "grad_norm": 1.0987855195999146,
      "learning_rate": 0.00019927868638743875,
      "loss": 2.2936,
      "step": 67
    },
    {
      "epoch": 0.013586413586413586,
      "grad_norm": 2.6548454761505127,
      "learning_rate": 0.0001992531869521312,
      "loss": 2.542,
      "step": 68
    },
    {
      "epoch": 0.013786213786213787,
      "grad_norm": 1.5722721815109253,
      "learning_rate": 0.00019922724628024515,
      "loss": 2.5417,
      "step": 69
    },
    {
      "epoch": 0.013986013986013986,
      "grad_norm": 2.8036539554595947,
      "learning_rate": 0.0001992008644871016,
      "loss": 2.3866,
      "step": 70
    },
    {
      "epoch": 0.014185814185814187,
      "grad_norm": 2.3543879985809326,
      "learning_rate": 0.00019917404168998256,
      "loss": 2.2646,
      "step": 71
    },
    {
      "epoch": 0.014385614385614386,
      "grad_norm": 1.909787893295288,
      "learning_rate": 0.0001991467780081305,
      "loss": 2.4832,
      "step": 72
    },
    {
      "epoch": 0.014585414585414586,
      "grad_norm": 2.962636709213257,
      "learning_rate": 0.00019911907356274795,
      "loss": 2.4872,
      "step": 73
    },
    {
      "epoch": 0.014785214785214785,
      "grad_norm": 1.4696192741394043,
      "learning_rate": 0.00019909092847699683,
      "loss": 2.5372,
      "step": 74
    },
    {
      "epoch": 0.014985014985014986,
      "grad_norm": 5.570610046386719,
      "learning_rate": 0.00019906234287599798,
      "loss": 1.4486,
      "step": 75
    },
    {
      "epoch": 0.015184815184815185,
      "grad_norm": 2.3279919624328613,
      "learning_rate": 0.00019903331688683057,
      "loss": 2.7787,
      "step": 76
    },
    {
      "epoch": 0.015384615384615385,
      "grad_norm": 2.3371896743774414,
      "learning_rate": 0.00019900385063853154,
      "loss": 1.1298,
      "step": 77
    },
    {
      "epoch": 0.015584415584415584,
      "grad_norm": 2.1514484882354736,
      "learning_rate": 0.00019897394426209505,
      "loss": 2.6728,
      "step": 78
    },
    {
      "epoch": 0.015784215784215783,
      "grad_norm": 3.018923759460449,
      "learning_rate": 0.00019894359789047187,
      "loss": 2.6985,
      "step": 79
    },
    {
      "epoch": 0.015984015984015984,
      "grad_norm": 2.568561553955078,
      "learning_rate": 0.00019891281165856873,
      "loss": 2.3459,
      "step": 80
    },
    {
      "epoch": 0.016183816183816185,
      "grad_norm": 3.6998348236083984,
      "learning_rate": 0.00019888158570324795,
      "loss": 2.4016,
      "step": 81
    },
    {
      "epoch": 0.016383616383616385,
      "grad_norm": 2.2671546936035156,
      "learning_rate": 0.0001988499201633265,
      "loss": 1.5604,
      "step": 82
    },
    {
      "epoch": 0.016583416583416583,
      "grad_norm": 3.0175819396972656,
      "learning_rate": 0.00019881781517957562,
      "loss": 1.0567,
      "step": 83
    },
    {
      "epoch": 0.016783216783216783,
      "grad_norm": 3.607232093811035,
      "learning_rate": 0.0001987852708947202,
      "loss": 1.933,
      "step": 84
    },
    {
      "epoch": 0.016983016983016984,
      "grad_norm": 1.392553687095642,
      "learning_rate": 0.00019875228745343794,
      "loss": 1.6011,
      "step": 85
    },
    {
      "epoch": 0.017182817182817185,
      "grad_norm": 3.5552830696105957,
      "learning_rate": 0.0001987188650023589,
      "loss": 1.9676,
      "step": 86
    },
    {
      "epoch": 0.017382617382617382,
      "grad_norm": 3.608072519302368,
      "learning_rate": 0.0001986850036900648,
      "loss": 1.6936,
      "step": 87
    },
    {
      "epoch": 0.017582417582417582,
      "grad_norm": 1.7247647047042847,
      "learning_rate": 0.00019865070366708836,
      "loss": 2.0497,
      "step": 88
    },
    {
      "epoch": 0.017782217782217783,
      "grad_norm": 2.6731009483337402,
      "learning_rate": 0.00019861596508591255,
      "loss": 1.5382,
      "step": 89
    },
    {
      "epoch": 0.017982017982017984,
      "grad_norm": 2.112288236618042,
      "learning_rate": 0.00019858078810097002,
      "loss": 2.7381,
      "step": 90
    },
    {
      "epoch": 0.01818181818181818,
      "grad_norm": 3.9500579833984375,
      "learning_rate": 0.00019854517286864245,
      "loss": 1.6446,
      "step": 91
    },
    {
      "epoch": 0.01838161838161838,
      "grad_norm": 3.410757303237915,
      "learning_rate": 0.0001985091195472596,
      "loss": 2.2337,
      "step": 92
    },
    {
      "epoch": 0.018581418581418582,
      "grad_norm": 1.9765585660934448,
      "learning_rate": 0.0001984726282970989,
      "loss": 1.3086,
      "step": 93
    },
    {
      "epoch": 0.018781218781218783,
      "grad_norm": 2.698464870452881,
      "learning_rate": 0.0001984356992803847,
      "loss": 1.9353,
      "step": 94
    },
    {
      "epoch": 0.01898101898101898,
      "grad_norm": 2.4799211025238037,
      "learning_rate": 0.00019839833266128724,
      "loss": 2.2466,
      "step": 95
    },
    {
      "epoch": 0.01918081918081918,
      "grad_norm": 1.8451018333435059,
      "learning_rate": 0.00019836052860592237,
      "loss": 1.6693,
      "step": 96
    },
    {
      "epoch": 0.01938061938061938,
      "grad_norm": 1.6218140125274658,
      "learning_rate": 0.0001983222872823505,
      "loss": 2.1556,
      "step": 97
    },
    {
      "epoch": 0.019580419580419582,
      "grad_norm": 7.524921417236328,
      "learning_rate": 0.00019828360886057594,
      "loss": 2.3376,
      "step": 98
    },
    {
      "epoch": 0.01978021978021978,
      "grad_norm": 2.4295475482940674,
      "learning_rate": 0.00019824449351254616,
      "loss": 2.5976,
      "step": 99
    },
    {
      "epoch": 0.01998001998001998,
      "grad_norm": 1.1456009149551392,
      "learning_rate": 0.00019820494141215104,
      "loss": 1.1917,
      "step": 100
    },
    {
      "epoch": 0.02017982017982018,
      "grad_norm": 2.11057186126709,
      "learning_rate": 0.000198164952735222,
      "loss": 2.259,
      "step": 101
    },
    {
      "epoch": 0.02037962037962038,
      "grad_norm": 1.591711401939392,
      "learning_rate": 0.00019812452765953135,
      "loss": 2.7036,
      "step": 102
    },
    {
      "epoch": 0.02057942057942058,
      "grad_norm": 2.012918710708618,
      "learning_rate": 0.00019808366636479147,
      "loss": 2.3368,
      "step": 103
    },
    {
      "epoch": 0.02077922077922078,
      "grad_norm": 1.3877614736557007,
      "learning_rate": 0.00019804236903265388,
      "loss": 2.61,
      "step": 104
    },
    {
      "epoch": 0.02097902097902098,
      "grad_norm": 1.5335688591003418,
      "learning_rate": 0.00019800063584670863,
      "loss": 2.4632,
      "step": 105
    },
    {
      "epoch": 0.02117882117882118,
      "grad_norm": 2.5433783531188965,
      "learning_rate": 0.00019795846699248332,
      "loss": 2.2444,
      "step": 106
    },
    {
      "epoch": 0.021378621378621378,
      "grad_norm": 2.0772433280944824,
      "learning_rate": 0.00019791586265744237,
      "loss": 2.0821,
      "step": 107
    },
    {
      "epoch": 0.02157842157842158,
      "grad_norm": 2.1214754581451416,
      "learning_rate": 0.00019787282303098617,
      "loss": 2.7248,
      "step": 108
    },
    {
      "epoch": 0.02177822177822178,
      "grad_norm": 1.8279271125793457,
      "learning_rate": 0.0001978293483044502,
      "loss": 1.2373,
      "step": 109
    },
    {
      "epoch": 0.02197802197802198,
      "grad_norm": 2.581543207168579,
      "learning_rate": 0.00019778543867110426,
      "loss": 2.1578,
      "step": 110
    },
    {
      "epoch": 0.022177822177822177,
      "grad_norm": 1.9022427797317505,
      "learning_rate": 0.00019774109432615147,
      "loss": 1.9701,
      "step": 111
    },
    {
      "epoch": 0.022377622377622378,
      "grad_norm": 1.2144060134887695,
      "learning_rate": 0.00019769631546672756,
      "loss": 1.1096,
      "step": 112
    },
    {
      "epoch": 0.02257742257742258,
      "grad_norm": 1.591853141784668,
      "learning_rate": 0.00019765110229189988,
      "loss": 2.3145,
      "step": 113
    },
    {
      "epoch": 0.02277722277722278,
      "grad_norm": 1.9778727293014526,
      "learning_rate": 0.00019760545500266657,
      "loss": 1.8132,
      "step": 114
    },
    {
      "epoch": 0.022977022977022976,
      "grad_norm": 1.6417161226272583,
      "learning_rate": 0.00019755937380195568,
      "loss": 2.1357,
      "step": 115
    },
    {
      "epoch": 0.023176823176823177,
      "grad_norm": 2.216238498687744,
      "learning_rate": 0.00019751285889462423,
      "loss": 2.0035,
      "step": 116
    },
    {
      "epoch": 0.023376623376623377,
      "grad_norm": 1.5507177114486694,
      "learning_rate": 0.0001974659104874573,
      "loss": 2.0934,
      "step": 117
    },
    {
      "epoch": 0.023576423576423578,
      "grad_norm": 1.891466498374939,
      "learning_rate": 0.0001974185287891671,
      "loss": 1.9964,
      "step": 118
    },
    {
      "epoch": 0.023776223776223775,
      "grad_norm": 2.101008653640747,
      "learning_rate": 0.0001973707140103921,
      "loss": 2.6874,
      "step": 119
    },
    {
      "epoch": 0.023976023976023976,
      "grad_norm": 3.775693655014038,
      "learning_rate": 0.00019732246636369605,
      "loss": 1.7318,
      "step": 120
    },
    {
      "epoch": 0.024175824175824177,
      "grad_norm": 1.4058136940002441,
      "learning_rate": 0.00019727378606356703,
      "loss": 2.3822,
      "step": 121
    },
    {
      "epoch": 0.024375624375624377,
      "grad_norm": 1.977903962135315,
      "learning_rate": 0.00019722467332641656,
      "loss": 1.7467,
      "step": 122
    },
    {
      "epoch": 0.024575424575424575,
      "grad_norm": 2.8408825397491455,
      "learning_rate": 0.00019717512837057855,
      "loss": 2.2329,
      "step": 123
    },
    {
      "epoch": 0.024775224775224775,
      "grad_norm": 2.27626895904541,
      "learning_rate": 0.0001971251514163083,
      "loss": 2.221,
      "step": 124
    },
    {
      "epoch": 0.024975024975024976,
      "grad_norm": 2.598998785018921,
      "learning_rate": 0.0001970747426857817,
      "loss": 2.6603,
      "step": 125
    },
    {
      "epoch": 0.025174825174825177,
      "grad_norm": 3.1240956783294678,
      "learning_rate": 0.00019702390240309404,
      "loss": 2.4393,
      "step": 126
    },
    {
      "epoch": 0.025374625374625374,
      "grad_norm": 2.0257961750030518,
      "learning_rate": 0.0001969726307942592,
      "loss": 2.8588,
      "step": 127
    },
    {
      "epoch": 0.025574425574425574,
      "grad_norm": 1.3819655179977417,
      "learning_rate": 0.00019692092808720846,
      "loss": 2.7077,
      "step": 128
    },
    {
      "epoch": 0.025774225774225775,
      "grad_norm": 3.106044292449951,
      "learning_rate": 0.0001968687945117896,
      "loss": 2.0123,
      "step": 129
    },
    {
      "epoch": 0.025974025974025976,
      "grad_norm": 1.9946179389953613,
      "learning_rate": 0.00019681623029976588,
      "loss": 2.0447,
      "step": 130
    },
    {
      "epoch": 0.026173826173826173,
      "grad_norm": 1.3225812911987305,
      "learning_rate": 0.00019676323568481498,
      "loss": 2.6258,
      "step": 131
    },
    {
      "epoch": 0.026373626373626374,
      "grad_norm": 3.2038662433624268,
      "learning_rate": 0.00019670981090252792,
      "loss": 2.6692,
      "step": 132
    },
    {
      "epoch": 0.026573426573426574,
      "grad_norm": 1.4215757846832275,
      "learning_rate": 0.00019665595619040808,
      "loss": 1.6513,
      "step": 133
    },
    {
      "epoch": 0.026773226773226775,
      "grad_norm": 1.862005352973938,
      "learning_rate": 0.0001966016717878702,
      "loss": 2.2465,
      "step": 134
    },
    {
      "epoch": 0.026973026973026972,
      "grad_norm": 1.2574067115783691,
      "learning_rate": 0.00019654695793623907,
      "loss": 2.571,
      "step": 135
    },
    {
      "epoch": 0.027172827172827173,
      "grad_norm": 2.859692096710205,
      "learning_rate": 0.0001964918148787488,
      "loss": 1.7874,
      "step": 136
    },
    {
      "epoch": 0.027372627372627373,
      "grad_norm": 1.8328845500946045,
      "learning_rate": 0.00019643624286054144,
      "loss": 0.8426,
      "step": 137
    },
    {
      "epoch": 0.027572427572427574,
      "grad_norm": 1.9008418321609497,
      "learning_rate": 0.00019638024212866606,
      "loss": 1.9568,
      "step": 138
    },
    {
      "epoch": 0.02777222777222777,
      "grad_norm": 2.183302164077759,
      "learning_rate": 0.0001963238129320776,
      "loss": 2.1374,
      "step": 139
    },
    {
      "epoch": 0.027972027972027972,
      "grad_norm": 2.028078556060791,
      "learning_rate": 0.00019626695552163578,
      "loss": 0.9724,
      "step": 140
    },
    {
      "epoch": 0.028171828171828173,
      "grad_norm": 1.4799890518188477,
      "learning_rate": 0.00019620967015010395,
      "loss": 1.846,
      "step": 141
    },
    {
      "epoch": 0.028371628371628373,
      "grad_norm": 1.7850462198257446,
      "learning_rate": 0.00019615195707214803,
      "loss": 2.8311,
      "step": 142
    },
    {
      "epoch": 0.02857142857142857,
      "grad_norm": 9.003202438354492,
      "learning_rate": 0.0001960938165443353,
      "loss": 2.7836,
      "step": 143
    },
    {
      "epoch": 0.02877122877122877,
      "grad_norm": 1.9001516103744507,
      "learning_rate": 0.00019603524882513327,
      "loss": 1.869,
      "step": 144
    },
    {
      "epoch": 0.028971028971028972,
      "grad_norm": 1.9313069581985474,
      "learning_rate": 0.0001959762541749086,
      "loss": 2.452,
      "step": 145
    },
    {
      "epoch": 0.029170829170829173,
      "grad_norm": 4.486754894256592,
      "learning_rate": 0.00019591683285592593,
      "loss": 1.9237,
      "step": 146
    },
    {
      "epoch": 0.02937062937062937,
      "grad_norm": 1.7127289772033691,
      "learning_rate": 0.00019585698513234663,
      "loss": 2.2313,
      "step": 147
    },
    {
      "epoch": 0.02957042957042957,
      "grad_norm": 0.8598572015762329,
      "learning_rate": 0.0001957967112702277,
      "loss": 2.219,
      "step": 148
    },
    {
      "epoch": 0.02977022977022977,
      "grad_norm": 1.0183742046356201,
      "learning_rate": 0.00019573601153752052,
      "loss": 1.2486,
      "step": 149
    },
    {
      "epoch": 0.029970029970029972,
      "grad_norm": 3.059025526046753,
      "learning_rate": 0.00019567488620406983,
      "loss": 2.4259,
      "step": 150
    },
    {
      "epoch": 0.03016983016983017,
      "grad_norm": 3.22623872756958,
      "learning_rate": 0.00019561333554161224,
      "loss": 2.7449,
      "step": 151
    },
    {
      "epoch": 0.03036963036963037,
      "grad_norm": 2.9344961643218994,
      "learning_rate": 0.0001955513598237753,
      "loss": 1.8362,
      "step": 152
    },
    {
      "epoch": 0.03056943056943057,
      "grad_norm": 1.9411219358444214,
      "learning_rate": 0.00019548895932607621,
      "loss": 2.6107,
      "step": 153
    },
    {
      "epoch": 0.03076923076923077,
      "grad_norm": 2.5881571769714355,
      "learning_rate": 0.00019542613432592038,
      "loss": 2.5389,
      "step": 154
    },
    {
      "epoch": 0.030969030969030968,
      "grad_norm": 0.9415623545646667,
      "learning_rate": 0.00019536288510260056,
      "loss": 0.6916,
      "step": 155
    },
    {
      "epoch": 0.03116883116883117,
      "grad_norm": 2.4108664989471436,
      "learning_rate": 0.00019529921193729534,
      "loss": 2.0332,
      "step": 156
    },
    {
      "epoch": 0.03136863136863137,
      "grad_norm": 4.4062819480896,
      "learning_rate": 0.00019523511511306793,
      "loss": 2.6123,
      "step": 157
    },
    {
      "epoch": 0.03156843156843157,
      "grad_norm": 3.420725107192993,
      "learning_rate": 0.000195170594914865,
      "loss": 2.2125,
      "step": 158
    },
    {
      "epoch": 0.03176823176823177,
      "grad_norm": 1.1376384496688843,
      "learning_rate": 0.00019510565162951537,
      "loss": 1.3457,
      "step": 159
    },
    {
      "epoch": 0.03196803196803197,
      "grad_norm": 3.2714099884033203,
      "learning_rate": 0.00019504028554572864,
      "loss": 1.8065,
      "step": 160
    },
    {
      "epoch": 0.032167832167832165,
      "grad_norm": 1.7296656370162964,
      "learning_rate": 0.00019497449695409408,
      "loss": 1.5025,
      "step": 161
    },
    {
      "epoch": 0.03236763236763237,
      "grad_norm": 1.5529906749725342,
      "learning_rate": 0.00019490828614707916,
      "loss": 1.9524,
      "step": 162
    },
    {
      "epoch": 0.03256743256743257,
      "grad_norm": 2.1530702114105225,
      "learning_rate": 0.00019484165341902845,
      "loss": 2.5262,
      "step": 163
    },
    {
      "epoch": 0.03276723276723277,
      "grad_norm": 1.9110842943191528,
      "learning_rate": 0.00019477459906616206,
      "loss": 2.3581,
      "step": 164
    },
    {
      "epoch": 0.03296703296703297,
      "grad_norm": 1.6430325508117676,
      "learning_rate": 0.00019470712338657458,
      "loss": 2.2753,
      "step": 165
    },
    {
      "epoch": 0.033166833166833165,
      "grad_norm": 3.6344187259674072,
      "learning_rate": 0.0001946392266802336,
      "loss": 1.7072,
      "step": 166
    },
    {
      "epoch": 0.03336663336663337,
      "grad_norm": 2.4185080528259277,
      "learning_rate": 0.0001945709092489783,
      "loss": 1.9346,
      "step": 167
    },
    {
      "epoch": 0.033566433566433566,
      "grad_norm": 2.780256986618042,
      "learning_rate": 0.00019450217139651844,
      "loss": 2.7294,
      "step": 168
    },
    {
      "epoch": 0.033766233766233764,
      "grad_norm": 1.3963232040405273,
      "learning_rate": 0.0001944330134284326,
      "loss": 1.0797,
      "step": 169
    },
    {
      "epoch": 0.03396603396603397,
      "grad_norm": 2.0075647830963135,
      "learning_rate": 0.00019436343565216711,
      "loss": 1.9338,
      "step": 170
    },
    {
      "epoch": 0.034165834165834165,
      "grad_norm": 3.7039761543273926,
      "learning_rate": 0.00019429343837703455,
      "loss": 2.5623,
      "step": 171
    },
    {
      "epoch": 0.03436563436563437,
      "grad_norm": 2.2598161697387695,
      "learning_rate": 0.0001942230219142124,
      "loss": 2.6052,
      "step": 172
    },
    {
      "epoch": 0.034565434565434566,
      "grad_norm": 4.2248454093933105,
      "learning_rate": 0.0001941521865767417,
      "loss": 1.8662,
      "step": 173
    },
    {
      "epoch": 0.034765234765234763,
      "grad_norm": 2.460540771484375,
      "learning_rate": 0.0001940809326795256,
      "loss": 1.9277,
      "step": 174
    },
    {
      "epoch": 0.03496503496503497,
      "grad_norm": 1.9257768392562866,
      "learning_rate": 0.000194009260539328,
      "loss": 2.2591,
      "step": 175
    },
    {
      "epoch": 0.035164835164835165,
      "grad_norm": 4.052698612213135,
      "learning_rate": 0.0001939371704747721,
      "loss": 1.7993,
      "step": 176
    },
    {
      "epoch": 0.03536463536463536,
      "grad_norm": 1.953648567199707,
      "learning_rate": 0.00019386466280633906,
      "loss": 2.5142,
      "step": 177
    },
    {
      "epoch": 0.035564435564435566,
      "grad_norm": 3.8480732440948486,
      "learning_rate": 0.00019379173785636646,
      "loss": 1.7734,
      "step": 178
    },
    {
      "epoch": 0.03576423576423576,
      "grad_norm": 1.224303960800171,
      "learning_rate": 0.000193718395949047,
      "loss": 2.6781,
      "step": 179
    },
    {
      "epoch": 0.03596403596403597,
      "grad_norm": 1.1210733652114868,
      "learning_rate": 0.00019364463741042694,
      "loss": 0.9148,
      "step": 180
    },
    {
      "epoch": 0.036163836163836165,
      "grad_norm": 2.1987240314483643,
      "learning_rate": 0.00019357046256840473,
      "loss": 2.4853,
      "step": 181
    },
    {
      "epoch": 0.03636363636363636,
      "grad_norm": 1.2381153106689453,
      "learning_rate": 0.00019349587175272948,
      "loss": 1.7151,
      "step": 182
    },
    {
      "epoch": 0.036563436563436566,
      "grad_norm": 3.117276191711426,
      "learning_rate": 0.0001934208652949996,
      "loss": 2.8313,
      "step": 183
    },
    {
      "epoch": 0.03676323676323676,
      "grad_norm": 1.350603461265564,
      "learning_rate": 0.00019334544352866127,
      "loss": 2.447,
      "step": 184
    },
    {
      "epoch": 0.03696303696303696,
      "grad_norm": 2.444509983062744,
      "learning_rate": 0.00019326960678900688,
      "loss": 1.7039,
      "step": 185
    },
    {
      "epoch": 0.037162837162837165,
      "grad_norm": 1.6882402896881104,
      "learning_rate": 0.00019319335541317361,
      "loss": 2.4264,
      "step": 186
    },
    {
      "epoch": 0.03736263736263736,
      "grad_norm": 0.994568407535553,
      "learning_rate": 0.00019311668974014208,
      "loss": 2.3419,
      "step": 187
    },
    {
      "epoch": 0.037562437562437566,
      "grad_norm": 2.3215484619140625,
      "learning_rate": 0.00019303961011073447,
      "loss": 2.0939,
      "step": 188
    },
    {
      "epoch": 0.03776223776223776,
      "grad_norm": 2.9770655632019043,
      "learning_rate": 0.00019296211686761346,
      "loss": 1.6391,
      "step": 189
    },
    {
      "epoch": 0.03796203796203796,
      "grad_norm": 1.8096888065338135,
      "learning_rate": 0.00019288421035528028,
      "loss": 1.9464,
      "step": 190
    },
    {
      "epoch": 0.038161838161838164,
      "grad_norm": 0.8935004472732544,
      "learning_rate": 0.00019280589092007352,
      "loss": 2.0536,
      "step": 191
    },
    {
      "epoch": 0.03836163836163836,
      "grad_norm": 2.0284087657928467,
      "learning_rate": 0.00019272715891016735,
      "loss": 2.2272,
      "step": 192
    },
    {
      "epoch": 0.03856143856143856,
      "grad_norm": 2.158893585205078,
      "learning_rate": 0.00019264801467557007,
      "loss": 1.9132,
      "step": 193
    },
    {
      "epoch": 0.03876123876123876,
      "grad_norm": 2.1813151836395264,
      "learning_rate": 0.00019256845856812266,
      "loss": 2.0057,
      "step": 194
    },
    {
      "epoch": 0.03896103896103896,
      "grad_norm": 2.5834860801696777,
      "learning_rate": 0.000192488490941497,
      "loss": 1.9996,
      "step": 195
    },
    {
      "epoch": 0.039160839160839164,
      "grad_norm": 2.374650239944458,
      "learning_rate": 0.00019240811215119448,
      "loss": 2.1538,
      "step": 196
    },
    {
      "epoch": 0.03936063936063936,
      "grad_norm": 2.7119600772857666,
      "learning_rate": 0.00019232732255454422,
      "loss": 1.6647,
      "step": 197
    },
    {
      "epoch": 0.03956043956043956,
      "grad_norm": 3.71632981300354,
      "learning_rate": 0.00019224612251070175,
      "loss": 2.187,
      "step": 198
    },
    {
      "epoch": 0.03976023976023976,
      "grad_norm": 2.8582239151000977,
      "learning_rate": 0.0001921645123806472,
      "loss": 1.5865,
      "step": 199
    },
    {
      "epoch": 0.03996003996003996,
      "grad_norm": 4.3827924728393555,
      "learning_rate": 0.0001920824925271838,
      "loss": 3.0931,
      "step": 200
    },
    {
      "epoch": 0.04015984015984016,
      "grad_norm": 2.314863443374634,
      "learning_rate": 0.0001920000633149362,
      "loss": 1.5053,
      "step": 201
    },
    {
      "epoch": 0.04035964035964036,
      "grad_norm": 2.615989923477173,
      "learning_rate": 0.00019191722511034884,
      "loss": 1.9091,
      "step": 202
    },
    {
      "epoch": 0.04055944055944056,
      "grad_norm": 1.3818212747573853,
      "learning_rate": 0.00019183397828168448,
      "loss": 1.1688,
      "step": 203
    },
    {
      "epoch": 0.04075924075924076,
      "grad_norm": 2.1217761039733887,
      "learning_rate": 0.00019175032319902234,
      "loss": 2.1323,
      "step": 204
    },
    {
      "epoch": 0.04095904095904096,
      "grad_norm": 1.6765371561050415,
      "learning_rate": 0.00019166626023425662,
      "loss": 1.1012,
      "step": 205
    },
    {
      "epoch": 0.04115884115884116,
      "grad_norm": 2.3769619464874268,
      "learning_rate": 0.00019158178976109476,
      "loss": 1.4703,
      "step": 206
    },
    {
      "epoch": 0.04135864135864136,
      "grad_norm": 0.9167413711547852,
      "learning_rate": 0.0001914969121550558,
      "loss": 1.2618,
      "step": 207
    },
    {
      "epoch": 0.04155844155844156,
      "grad_norm": 2.132364273071289,
      "learning_rate": 0.00019141162779346874,
      "loss": 1.0209,
      "step": 208
    },
    {
      "epoch": 0.041758241758241756,
      "grad_norm": 1.2914800643920898,
      "learning_rate": 0.00019132593705547082,
      "loss": 0.7191,
      "step": 209
    },
    {
      "epoch": 0.04195804195804196,
      "grad_norm": 2.249603033065796,
      "learning_rate": 0.00019123984032200586,
      "loss": 1.8325,
      "step": 210
    },
    {
      "epoch": 0.04215784215784216,
      "grad_norm": 2.487809896469116,
      "learning_rate": 0.00019115333797582254,
      "loss": 2.1842,
      "step": 211
    },
    {
      "epoch": 0.04235764235764236,
      "grad_norm": 2.6654603481292725,
      "learning_rate": 0.00019106643040147278,
      "loss": 2.8571,
      "step": 212
    },
    {
      "epoch": 0.04255744255744256,
      "grad_norm": 2.1209607124328613,
      "learning_rate": 0.00019097911798530987,
      "loss": 2.4967,
      "step": 213
    },
    {
      "epoch": 0.042757242757242755,
      "grad_norm": 2.520019054412842,
      "learning_rate": 0.00019089140111548696,
      "loss": 2.3591,
      "step": 214
    },
    {
      "epoch": 0.04295704295704296,
      "grad_norm": 5.732324600219727,
      "learning_rate": 0.00019080328018195513,
      "loss": 2.0895,
      "step": 215
    },
    {
      "epoch": 0.04315684315684316,
      "grad_norm": 3.0174124240875244,
      "learning_rate": 0.0001907147555764618,
      "loss": 2.3704,
      "step": 216
    },
    {
      "epoch": 0.043356643356643354,
      "grad_norm": 2.2944223880767822,
      "learning_rate": 0.00019062582769254895,
      "loss": 2.1149,
      "step": 217
    },
    {
      "epoch": 0.04355644355644356,
      "grad_norm": 3.592420816421509,
      "learning_rate": 0.00019053649692555135,
      "loss": 2.1119,
      "step": 218
    },
    {
      "epoch": 0.043756243756243755,
      "grad_norm": 2.6429970264434814,
      "learning_rate": 0.00019044676367259476,
      "loss": 1.4201,
      "step": 219
    },
    {
      "epoch": 0.04395604395604396,
      "grad_norm": 1.792651653289795,
      "learning_rate": 0.00019035662833259432,
      "loss": 2.1131,
      "step": 220
    },
    {
      "epoch": 0.04415584415584416,
      "grad_norm": 1.9210007190704346,
      "learning_rate": 0.00019026609130625257,
      "loss": 1.694,
      "step": 221
    },
    {
      "epoch": 0.044355644355644354,
      "grad_norm": 1.4148427248001099,
      "learning_rate": 0.00019017515299605788,
      "loss": 2.5437,
      "step": 222
    },
    {
      "epoch": 0.04455544455544456,
      "grad_norm": 2.8444790840148926,
      "learning_rate": 0.00019008381380628247,
      "loss": 1.7215,
      "step": 223
    },
    {
      "epoch": 0.044755244755244755,
      "grad_norm": 3.1274664402008057,
      "learning_rate": 0.00018999207414298067,
      "loss": 2.4997,
      "step": 224
    },
    {
      "epoch": 0.04495504495504495,
      "grad_norm": 1.4046162366867065,
      "learning_rate": 0.00018989993441398726,
      "loss": 2.3084,
      "step": 225
    },
    {
      "epoch": 0.04515484515484516,
      "grad_norm": 3.3667502403259277,
      "learning_rate": 0.00018980739502891546,
      "loss": 2.0695,
      "step": 226
    },
    {
      "epoch": 0.045354645354645354,
      "grad_norm": 1.766465187072754,
      "learning_rate": 0.0001897144563991552,
      "loss": 2.1368,
      "step": 227
    },
    {
      "epoch": 0.04555444555444556,
      "grad_norm": 1.8711544275283813,
      "learning_rate": 0.00018962111893787128,
      "loss": 1.7168,
      "step": 228
    },
    {
      "epoch": 0.045754245754245755,
      "grad_norm": 2.4305419921875,
      "learning_rate": 0.00018952738306000151,
      "loss": 2.2354,
      "step": 229
    },
    {
      "epoch": 0.04595404595404595,
      "grad_norm": 3.5147576332092285,
      "learning_rate": 0.00018943324918225494,
      "loss": 2.0449,
      "step": 230
    },
    {
      "epoch": 0.046153846153846156,
      "grad_norm": 3.226665496826172,
      "learning_rate": 0.0001893387177231099,
      "loss": 2.1509,
      "step": 231
    },
    {
      "epoch": 0.046353646353646354,
      "grad_norm": 2.7920920848846436,
      "learning_rate": 0.0001892437891028122,
      "loss": 2.2344,
      "step": 232
    },
    {
      "epoch": 0.04655344655344655,
      "grad_norm": 1.891906499862671,
      "learning_rate": 0.0001891484637433733,
      "loss": 2.082,
      "step": 233
    },
    {
      "epoch": 0.046753246753246755,
      "grad_norm": 1.0820852518081665,
      "learning_rate": 0.00018905274206856837,
      "loss": 1.3008,
      "step": 234
    },
    {
      "epoch": 0.04695304695304695,
      "grad_norm": 2.658303737640381,
      "learning_rate": 0.00018895662450393438,
      "loss": 2.8221,
      "step": 235
    },
    {
      "epoch": 0.047152847152847156,
      "grad_norm": 1.8654121160507202,
      "learning_rate": 0.00018886011147676833,
      "loss": 2.1363,
      "step": 236
    },
    {
      "epoch": 0.047352647352647353,
      "grad_norm": 1.44706392288208,
      "learning_rate": 0.00018876320341612522,
      "loss": 2.1883,
      "step": 237
    },
    {
      "epoch": 0.04755244755244755,
      "grad_norm": 2.80721116065979,
      "learning_rate": 0.00018866590075281624,
      "loss": 2.0924,
      "step": 238
    },
    {
      "epoch": 0.047752247752247755,
      "grad_norm": 3.1160550117492676,
      "learning_rate": 0.00018856820391940674,
      "loss": 2.4193,
      "step": 239
    },
    {
      "epoch": 0.04795204795204795,
      "grad_norm": 2.4184887409210205,
      "learning_rate": 0.00018847011335021449,
      "loss": 2.7246,
      "step": 240
    },
    {
      "epoch": 0.04815184815184815,
      "grad_norm": 2.4509494304656982,
      "learning_rate": 0.00018837162948130752,
      "loss": 2.1728,
      "step": 241
    },
    {
      "epoch": 0.04835164835164835,
      "grad_norm": 2.5162031650543213,
      "learning_rate": 0.00018827275275050233,
      "loss": 1.604,
      "step": 242
    },
    {
      "epoch": 0.04855144855144855,
      "grad_norm": 2.5790669918060303,
      "learning_rate": 0.00018817348359736203,
      "loss": 2.421,
      "step": 243
    },
    {
      "epoch": 0.048751248751248755,
      "grad_norm": 4.03166389465332,
      "learning_rate": 0.00018807382246319412,
      "loss": 1.9682,
      "step": 244
    },
    {
      "epoch": 0.04895104895104895,
      "grad_norm": 4.638970375061035,
      "learning_rate": 0.00018797376979104872,
      "loss": 2.3701,
      "step": 245
    },
    {
      "epoch": 0.04915084915084915,
      "grad_norm": 2.4261131286621094,
      "learning_rate": 0.00018787332602571662,
      "loss": 1.8785,
      "step": 246
    },
    {
      "epoch": 0.04935064935064935,
      "grad_norm": 3.2700650691986084,
      "learning_rate": 0.00018777249161372713,
      "loss": 1.8044,
      "step": 247
    },
    {
      "epoch": 0.04955044955044955,
      "grad_norm": 2.148256778717041,
      "learning_rate": 0.00018767126700334634,
      "loss": 2.3476,
      "step": 248
    },
    {
      "epoch": 0.04975024975024975,
      "grad_norm": 1.7313624620437622,
      "learning_rate": 0.0001875696526445749,
      "loss": 2.4342,
      "step": 249
    },
    {
      "epoch": 0.04995004995004995,
      "grad_norm": 2.8317391872406006,
      "learning_rate": 0.0001874676489891461,
      "loss": 1.3805,
      "step": 250
    },
    {
      "epoch": 0.05014985014985015,
      "grad_norm": 1.993682861328125,
      "learning_rate": 0.00018736525649052394,
      "loss": 3.0423,
      "step": 251
    },
    {
      "epoch": 0.05034965034965035,
      "grad_norm": 1.751766324043274,
      "learning_rate": 0.00018726247560390099,
      "loss": 2.0693,
      "step": 252
    },
    {
      "epoch": 0.05054945054945055,
      "grad_norm": 4.182394981384277,
      "learning_rate": 0.00018715930678619644,
      "loss": 2.2779,
      "step": 253
    },
    {
      "epoch": 0.05074925074925075,
      "grad_norm": 1.9574288129806519,
      "learning_rate": 0.00018705575049605413,
      "loss": 1.8734,
      "step": 254
    },
    {
      "epoch": 0.05094905094905095,
      "grad_norm": 1.9037193059921265,
      "learning_rate": 0.00018695180719384029,
      "loss": 1.9662,
      "step": 255
    },
    {
      "epoch": 0.05114885114885115,
      "grad_norm": 2.9709396362304688,
      "learning_rate": 0.00018684747734164177,
      "loss": 2.3722,
      "step": 256
    },
    {
      "epoch": 0.051348651348651346,
      "grad_norm": 2.1774497032165527,
      "learning_rate": 0.00018674276140326376,
      "loss": 1.8733,
      "step": 257
    },
    {
      "epoch": 0.05154845154845155,
      "grad_norm": 3.7941067218780518,
      "learning_rate": 0.00018663765984422786,
      "loss": 2.3216,
      "step": 258
    },
    {
      "epoch": 0.05174825174825175,
      "grad_norm": 1.0956095457077026,
      "learning_rate": 0.00018653217313177004,
      "loss": 2.4467,
      "step": 259
    },
    {
      "epoch": 0.05194805194805195,
      "grad_norm": 3.7411015033721924,
      "learning_rate": 0.00018642630173483832,
      "loss": 2.5227,
      "step": 260
    },
    {
      "epoch": 0.05214785214785215,
      "grad_norm": 3.4958276748657227,
      "learning_rate": 0.00018632004612409103,
      "loss": 2.3791,
      "step": 261
    },
    {
      "epoch": 0.052347652347652346,
      "grad_norm": 2.0506348609924316,
      "learning_rate": 0.00018621340677189453,
      "loss": 2.3625,
      "step": 262
    },
    {
      "epoch": 0.05254745254745255,
      "grad_norm": 1.0599132776260376,
      "learning_rate": 0.00018610638415232097,
      "loss": 2.5274,
      "step": 263
    },
    {
      "epoch": 0.05274725274725275,
      "grad_norm": 1.7403405904769897,
      "learning_rate": 0.00018599897874114652,
      "loss": 1.4018,
      "step": 264
    },
    {
      "epoch": 0.052947052947052944,
      "grad_norm": 1.5512832403182983,
      "learning_rate": 0.00018589119101584898,
      "loss": 2.2585,
      "step": 265
    },
    {
      "epoch": 0.05314685314685315,
      "grad_norm": 1.4400017261505127,
      "learning_rate": 0.00018578302145560584,
      "loss": 2.1167,
      "step": 266
    },
    {
      "epoch": 0.053346653346653346,
      "grad_norm": 3.778263807296753,
      "learning_rate": 0.00018567447054129195,
      "loss": 1.664,
      "step": 267
    },
    {
      "epoch": 0.05354645354645355,
      "grad_norm": 1.875180959701538,
      "learning_rate": 0.00018556553875547754,
      "loss": 2.3964,
      "step": 268
    },
    {
      "epoch": 0.05374625374625375,
      "grad_norm": 0.5713325142860413,
      "learning_rate": 0.00018545622658242607,
      "loss": 0.738,
      "step": 269
    },
    {
      "epoch": 0.053946053946053944,
      "grad_norm": 1.7285752296447754,
      "learning_rate": 0.00018534653450809197,
      "loss": 2.3473,
      "step": 270
    },
    {
      "epoch": 0.05414585414585415,
      "grad_norm": 1.3165370225906372,
      "learning_rate": 0.00018523646302011867,
      "loss": 2.7335,
      "step": 271
    },
    {
      "epoch": 0.054345654345654346,
      "grad_norm": 3.232527732849121,
      "learning_rate": 0.00018512601260783606,
      "loss": 1.9704,
      "step": 272
    },
    {
      "epoch": 0.05454545454545454,
      "grad_norm": 2.0769572257995605,
      "learning_rate": 0.00018501518376225887,
      "loss": 1.6432,
      "step": 273
    },
    {
      "epoch": 0.05474525474525475,
      "grad_norm": 4.582975387573242,
      "learning_rate": 0.00018490397697608395,
      "loss": 1.8136,
      "step": 274
    },
    {
      "epoch": 0.054945054945054944,
      "grad_norm": 2.2063236236572266,
      "learning_rate": 0.0001847923927436884,
      "loss": 2.1067,
      "step": 275
    },
    {
      "epoch": 0.05514485514485515,
      "grad_norm": 1.857171893119812,
      "learning_rate": 0.00018468043156112728,
      "loss": 0.817,
      "step": 276
    },
    {
      "epoch": 0.055344655344655345,
      "grad_norm": 2.6284055709838867,
      "learning_rate": 0.0001845680939261314,
      "loss": 1.7915,
      "step": 277
    },
    {
      "epoch": 0.05554445554445554,
      "grad_norm": 2.6198198795318604,
      "learning_rate": 0.00018445538033810515,
      "loss": 1.8117,
      "step": 278
    },
    {
      "epoch": 0.05574425574425575,
      "grad_norm": 1.622298240661621,
      "learning_rate": 0.00018434229129812418,
      "loss": 1.9287,
      "step": 279
    },
    {
      "epoch": 0.055944055944055944,
      "grad_norm": 2.0141146183013916,
      "learning_rate": 0.0001842288273089332,
      "loss": 2.2417,
      "step": 280
    },
    {
      "epoch": 0.05614385614385614,
      "grad_norm": 2.390171766281128,
      "learning_rate": 0.00018411498887494396,
      "loss": 2.1587,
      "step": 281
    },
    {
      "epoch": 0.056343656343656345,
      "grad_norm": 1.1146891117095947,
      "learning_rate": 0.00018400077650223263,
      "loss": 2.0318,
      "step": 282
    },
    {
      "epoch": 0.05654345654345654,
      "grad_norm": 3.3207759857177734,
      "learning_rate": 0.0001838861906985379,
      "loss": 1.9269,
      "step": 283
    },
    {
      "epoch": 0.05674325674325675,
      "grad_norm": 1.6940569877624512,
      "learning_rate": 0.00018377123197325842,
      "loss": 1.6321,
      "step": 284
    },
    {
      "epoch": 0.056943056943056944,
      "grad_norm": 2.2602603435516357,
      "learning_rate": 0.00018365590083745085,
      "loss": 2.0847,
      "step": 285
    },
    {
      "epoch": 0.05714285714285714,
      "grad_norm": 2.543576717376709,
      "learning_rate": 0.00018354019780382735,
      "loss": 1.6949,
      "step": 286
    },
    {
      "epoch": 0.057342657342657345,
      "grad_norm": 1.543125867843628,
      "learning_rate": 0.0001834241233867533,
      "loss": 2.2949,
      "step": 287
    },
    {
      "epoch": 0.05754245754245754,
      "grad_norm": 1.9820222854614258,
      "learning_rate": 0.00018330767810224524,
      "loss": 1.8283,
      "step": 288
    },
    {
      "epoch": 0.05774225774225774,
      "grad_norm": 1.9447818994522095,
      "learning_rate": 0.0001831908624679683,
      "loss": 2.2786,
      "step": 289
    },
    {
      "epoch": 0.057942057942057944,
      "grad_norm": 2.6755576133728027,
      "learning_rate": 0.0001830736770032341,
      "loss": 1.8216,
      "step": 290
    },
    {
      "epoch": 0.05814185814185814,
      "grad_norm": 1.4908809661865234,
      "learning_rate": 0.0001829561222289984,
      "loss": 2.2231,
      "step": 291
    },
    {
      "epoch": 0.058341658341658345,
      "grad_norm": 2.0485293865203857,
      "learning_rate": 0.00018283819866785853,
      "loss": 2.2628,
      "step": 292
    },
    {
      "epoch": 0.05854145854145854,
      "grad_norm": 2.956082344055176,
      "learning_rate": 0.0001827199068440516,
      "loss": 2.1004,
      "step": 293
    },
    {
      "epoch": 0.05874125874125874,
      "grad_norm": 3.019245147705078,
      "learning_rate": 0.00018260124728345162,
      "loss": 2.7525,
      "step": 294
    },
    {
      "epoch": 0.058941058941058944,
      "grad_norm": 1.5601719617843628,
      "learning_rate": 0.00018248222051356754,
      "loss": 2.7708,
      "step": 295
    },
    {
      "epoch": 0.05914085914085914,
      "grad_norm": 2.817288637161255,
      "learning_rate": 0.00018236282706354063,
      "loss": 2.5394,
      "step": 296
    },
    {
      "epoch": 0.05934065934065934,
      "grad_norm": 1.7105910778045654,
      "learning_rate": 0.00018224306746414238,
      "loss": 2.0009,
      "step": 297
    },
    {
      "epoch": 0.05954045954045954,
      "grad_norm": 2.5918288230895996,
      "learning_rate": 0.00018212294224777197,
      "loss": 2.0588,
      "step": 298
    },
    {
      "epoch": 0.05974025974025974,
      "grad_norm": 2.5453405380249023,
      "learning_rate": 0.00018200245194845399,
      "loss": 1.5728,
      "step": 299
    },
    {
      "epoch": 0.059940059940059943,
      "grad_norm": 1.9029632806777954,
      "learning_rate": 0.00018188159710183594,
      "loss": 2.1116,
      "step": 300
    },
    {
      "epoch": 0.06013986013986014,
      "grad_norm": 1.0984117984771729,
      "learning_rate": 0.000181760378245186,
      "loss": 2.3532,
      "step": 301
    },
    {
      "epoch": 0.06033966033966034,
      "grad_norm": 4.011904716491699,
      "learning_rate": 0.00018163879591739067,
      "loss": 2.1492,
      "step": 302
    },
    {
      "epoch": 0.06053946053946054,
      "grad_norm": 4.592916011810303,
      "learning_rate": 0.0001815168506589521,
      "loss": 2.1651,
      "step": 303
    },
    {
      "epoch": 0.06073926073926074,
      "grad_norm": 1.7433207035064697,
      "learning_rate": 0.000181394543011986,
      "loss": 2.0739,
      "step": 304
    },
    {
      "epoch": 0.060939060939060936,
      "grad_norm": 1.7145335674285889,
      "learning_rate": 0.00018127187352021907,
      "loss": 2.4062,
      "step": 305
    },
    {
      "epoch": 0.06113886113886114,
      "grad_norm": 1.4776666164398193,
      "learning_rate": 0.0001811488427289866,
      "loss": 2.4075,
      "step": 306
    },
    {
      "epoch": 0.06133866133866134,
      "grad_norm": 1.0966976881027222,
      "learning_rate": 0.00018102545118523007,
      "loss": 2.1613,
      "step": 307
    },
    {
      "epoch": 0.06153846153846154,
      "grad_norm": 1.688111662864685,
      "learning_rate": 0.00018090169943749476,
      "loss": 1.0875,
      "step": 308
    },
    {
      "epoch": 0.06173826173826174,
      "grad_norm": 1.445436716079712,
      "learning_rate": 0.00018077758803592718,
      "loss": 1.9274,
      "step": 309
    },
    {
      "epoch": 0.061938061938061936,
      "grad_norm": 3.0315604209899902,
      "learning_rate": 0.00018065311753227273,
      "loss": 1.8366,
      "step": 310
    },
    {
      "epoch": 0.06213786213786214,
      "grad_norm": 2.3171865940093994,
      "learning_rate": 0.0001805282884798732,
      "loss": 2.3319,
      "step": 311
    },
    {
      "epoch": 0.06233766233766234,
      "grad_norm": 2.7450220584869385,
      "learning_rate": 0.00018040310143366446,
      "loss": 2.5023,
      "step": 312
    },
    {
      "epoch": 0.06253746253746253,
      "grad_norm": 0.9302834868431091,
      "learning_rate": 0.00018027755695017368,
      "loss": 2.1774,
      "step": 313
    },
    {
      "epoch": 0.06273726273726274,
      "grad_norm": 2.9321112632751465,
      "learning_rate": 0.00018015165558751717,
      "loss": 2.093,
      "step": 314
    },
    {
      "epoch": 0.06293706293706294,
      "grad_norm": 1.0208370685577393,
      "learning_rate": 0.00018002539790539773,
      "loss": 2.3578,
      "step": 315
    },
    {
      "epoch": 0.06313686313686313,
      "grad_norm": 2.0822482109069824,
      "learning_rate": 0.00017989878446510215,
      "loss": 1.4541,
      "step": 316
    },
    {
      "epoch": 0.06333666333666334,
      "grad_norm": 3.1926510334014893,
      "learning_rate": 0.00017977181582949888,
      "loss": 2.4907,
      "step": 317
    },
    {
      "epoch": 0.06353646353646354,
      "grad_norm": 3.1711392402648926,
      "learning_rate": 0.0001796444925630353,
      "loss": 2.4275,
      "step": 318
    },
    {
      "epoch": 0.06373626373626373,
      "grad_norm": 3.0191235542297363,
      "learning_rate": 0.00017951681523173542,
      "loss": 2.4277,
      "step": 319
    },
    {
      "epoch": 0.06393606393606394,
      "grad_norm": 1.5903990268707275,
      "learning_rate": 0.0001793887844031972,
      "loss": 2.2421,
      "step": 320
    },
    {
      "epoch": 0.06413586413586414,
      "grad_norm": 1.7100311517715454,
      "learning_rate": 0.00017926040064659014,
      "loss": 1.6193,
      "step": 321
    },
    {
      "epoch": 0.06433566433566433,
      "grad_norm": 1.3281675577163696,
      "learning_rate": 0.0001791316645326526,
      "loss": 2.3229,
      "step": 322
    },
    {
      "epoch": 0.06453546453546453,
      "grad_norm": 3.218231201171875,
      "learning_rate": 0.00017900257663368963,
      "loss": 1.6293,
      "step": 323
    },
    {
      "epoch": 0.06473526473526474,
      "grad_norm": 2.695230484008789,
      "learning_rate": 0.0001788731375235698,
      "loss": 2.502,
      "step": 324
    },
    {
      "epoch": 0.06493506493506493,
      "grad_norm": 2.5918002128601074,
      "learning_rate": 0.00017874334777772327,
      "loss": 1.5196,
      "step": 325
    },
    {
      "epoch": 0.06513486513486513,
      "grad_norm": 2.4473190307617188,
      "learning_rate": 0.00017861320797313892,
      "loss": 2.9103,
      "step": 326
    },
    {
      "epoch": 0.06533466533466534,
      "grad_norm": 2.6190431118011475,
      "learning_rate": 0.0001784827186883618,
      "loss": 1.3661,
      "step": 327
    },
    {
      "epoch": 0.06553446553446554,
      "grad_norm": 3.114339828491211,
      "learning_rate": 0.00017835188050349064,
      "loss": 2.2666,
      "step": 328
    },
    {
      "epoch": 0.06573426573426573,
      "grad_norm": 3.648681879043579,
      "learning_rate": 0.00017822069400017516,
      "loss": 2.0938,
      "step": 329
    },
    {
      "epoch": 0.06593406593406594,
      "grad_norm": 2.7650301456451416,
      "learning_rate": 0.00017808915976161362,
      "loss": 2.4264,
      "step": 330
    },
    {
      "epoch": 0.06613386613386614,
      "grad_norm": 1.3321884870529175,
      "learning_rate": 0.00017795727837255015,
      "loss": 1.8049,
      "step": 331
    },
    {
      "epoch": 0.06633366633366633,
      "grad_norm": 2.5818018913269043,
      "learning_rate": 0.00017782505041927216,
      "loss": 2.1749,
      "step": 332
    },
    {
      "epoch": 0.06653346653346653,
      "grad_norm": 4.567615985870361,
      "learning_rate": 0.00017769247648960774,
      "loss": 2.2728,
      "step": 333
    },
    {
      "epoch": 0.06673326673326674,
      "grad_norm": 1.749512791633606,
      "learning_rate": 0.00017755955717292296,
      "loss": 1.7944,
      "step": 334
    },
    {
      "epoch": 0.06693306693306693,
      "grad_norm": 2.3047711849212646,
      "learning_rate": 0.00017742629306011944,
      "loss": 1.9256,
      "step": 335
    },
    {
      "epoch": 0.06713286713286713,
      "grad_norm": 0.9987353086471558,
      "learning_rate": 0.00017729268474363154,
      "loss": 2.3319,
      "step": 336
    },
    {
      "epoch": 0.06733266733266734,
      "grad_norm": 2.1943697929382324,
      "learning_rate": 0.0001771587328174239,
      "loss": 1.1735,
      "step": 337
    },
    {
      "epoch": 0.06753246753246753,
      "grad_norm": 3.154210329055786,
      "learning_rate": 0.0001770244378769885,
      "loss": 2.0831,
      "step": 338
    },
    {
      "epoch": 0.06773226773226773,
      "grad_norm": 2.9076945781707764,
      "learning_rate": 0.0001768898005193425,
      "loss": 2.3403,
      "step": 339
    },
    {
      "epoch": 0.06793206793206794,
      "grad_norm": 1.7863624095916748,
      "learning_rate": 0.000176754821343025,
      "loss": 2.445,
      "step": 340
    },
    {
      "epoch": 0.06813186813186813,
      "grad_norm": 1.7532540559768677,
      "learning_rate": 0.0001766195009480949,
      "loss": 1.8649,
      "step": 341
    },
    {
      "epoch": 0.06833166833166833,
      "grad_norm": 2.2405738830566406,
      "learning_rate": 0.0001764838399361279,
      "loss": 2.0464,
      "step": 342
    },
    {
      "epoch": 0.06853146853146853,
      "grad_norm": 2.160360813140869,
      "learning_rate": 0.00017634783891021393,
      "loss": 2.0295,
      "step": 343
    },
    {
      "epoch": 0.06873126873126874,
      "grad_norm": 0.5503923296928406,
      "learning_rate": 0.00017621149847495458,
      "loss": 0.3897,
      "step": 344
    },
    {
      "epoch": 0.06893106893106893,
      "grad_norm": 1.2185351848602295,
      "learning_rate": 0.00017607481923646016,
      "loss": 0.8051,
      "step": 345
    },
    {
      "epoch": 0.06913086913086913,
      "grad_norm": 2.6080684661865234,
      "learning_rate": 0.0001759378018023473,
      "loss": 1.5721,
      "step": 346
    },
    {
      "epoch": 0.06933066933066934,
      "grad_norm": 3.3662521839141846,
      "learning_rate": 0.00017580044678173592,
      "loss": 2.3202,
      "step": 347
    },
    {
      "epoch": 0.06953046953046953,
      "grad_norm": 1.9234858751296997,
      "learning_rate": 0.00017566275478524693,
      "loss": 1.6161,
      "step": 348
    },
    {
      "epoch": 0.06973026973026973,
      "grad_norm": 2.5417191982269287,
      "learning_rate": 0.0001755247264249991,
      "loss": 2.1661,
      "step": 349
    },
    {
      "epoch": 0.06993006993006994,
      "grad_norm": 2.1425821781158447,
      "learning_rate": 0.0001753863623146066,
      "loss": 2.2618,
      "step": 350
    },
    {
      "epoch": 0.07012987012987013,
      "grad_norm": 1.4929770231246948,
      "learning_rate": 0.00017524766306917618,
      "loss": 1.9333,
      "step": 351
    },
    {
      "epoch": 0.07032967032967033,
      "grad_norm": 3.8221333026885986,
      "learning_rate": 0.0001751086293053045,
      "loss": 1.5715,
      "step": 352
    },
    {
      "epoch": 0.07052947052947053,
      "grad_norm": 4.9178547859191895,
      "learning_rate": 0.0001749692616410753,
      "loss": 1.6718,
      "step": 353
    },
    {
      "epoch": 0.07072927072927072,
      "grad_norm": 1.3113629817962646,
      "learning_rate": 0.00017482956069605668,
      "loss": 1.5194,
      "step": 354
    },
    {
      "epoch": 0.07092907092907093,
      "grad_norm": 1.7924879789352417,
      "learning_rate": 0.00017468952709129846,
      "loss": 1.8977,
      "step": 355
    },
    {
      "epoch": 0.07112887112887113,
      "grad_norm": 2.833829879760742,
      "learning_rate": 0.00017454916144932922,
      "loss": 1.931,
      "step": 356
    },
    {
      "epoch": 0.07132867132867132,
      "grad_norm": 2.381495475769043,
      "learning_rate": 0.0001744084643941536,
      "loss": 2.2419,
      "step": 357
    },
    {
      "epoch": 0.07152847152847153,
      "grad_norm": 4.041555404663086,
      "learning_rate": 0.00017426743655124974,
      "loss": 1.2209,
      "step": 358
    },
    {
      "epoch": 0.07172827172827173,
      "grad_norm": 1.4068500995635986,
      "learning_rate": 0.0001741260785475661,
      "loss": 1.4168,
      "step": 359
    },
    {
      "epoch": 0.07192807192807193,
      "grad_norm": 0.9596447944641113,
      "learning_rate": 0.00017398439101151905,
      "loss": 0.6973,
      "step": 360
    },
    {
      "epoch": 0.07212787212787213,
      "grad_norm": 2.131685495376587,
      "learning_rate": 0.00017384237457298987,
      "loss": 1.8757,
      "step": 361
    },
    {
      "epoch": 0.07232767232767233,
      "grad_norm": 1.3789016008377075,
      "learning_rate": 0.00017370002986332193,
      "loss": 2.2286,
      "step": 362
    },
    {
      "epoch": 0.07252747252747253,
      "grad_norm": 1.696277141571045,
      "learning_rate": 0.00017355735751531807,
      "loss": 1.9328,
      "step": 363
    },
    {
      "epoch": 0.07272727272727272,
      "grad_norm": 1.42643404006958,
      "learning_rate": 0.00017341435816323756,
      "loss": 2.9264,
      "step": 364
    },
    {
      "epoch": 0.07292707292707293,
      "grad_norm": 2.217338800430298,
      "learning_rate": 0.00017327103244279348,
      "loss": 1.8554,
      "step": 365
    },
    {
      "epoch": 0.07312687312687313,
      "grad_norm": 1.3762191534042358,
      "learning_rate": 0.00017312738099114973,
      "loss": 2.6073,
      "step": 366
    },
    {
      "epoch": 0.07332667332667332,
      "grad_norm": 4.520292282104492,
      "learning_rate": 0.00017298340444691835,
      "loss": 1.9503,
      "step": 367
    },
    {
      "epoch": 0.07352647352647353,
      "grad_norm": 2.198173761367798,
      "learning_rate": 0.00017283910345015647,
      "loss": 1.9313,
      "step": 368
    },
    {
      "epoch": 0.07372627372627373,
      "grad_norm": 3.1280393600463867,
      "learning_rate": 0.0001726944786423637,
      "loss": 2.2458,
      "step": 369
    },
    {
      "epoch": 0.07392607392607392,
      "grad_norm": 1.7793433666229248,
      "learning_rate": 0.00017254953066647913,
      "loss": 1.704,
      "step": 370
    },
    {
      "epoch": 0.07412587412587412,
      "grad_norm": 3.164799928665161,
      "learning_rate": 0.00017240426016687863,
      "loss": 1.6993,
      "step": 371
    },
    {
      "epoch": 0.07432567432567433,
      "grad_norm": 3.2174603939056396,
      "learning_rate": 0.00017225866778937165,
      "loss": 1.9844,
      "step": 372
    },
    {
      "epoch": 0.07452547452547452,
      "grad_norm": 1.709800124168396,
      "learning_rate": 0.00017211275418119876,
      "loss": 2.0286,
      "step": 373
    },
    {
      "epoch": 0.07472527472527472,
      "grad_norm": 2.832578182220459,
      "learning_rate": 0.0001719665199910285,
      "loss": 2.0779,
      "step": 374
    },
    {
      "epoch": 0.07492507492507493,
      "grad_norm": 2.4686119556427,
      "learning_rate": 0.00017181996586895454,
      "loss": 2.4825,
      "step": 375
    },
    {
      "epoch": 0.07492507492507493,
      "eval_loss": 1.9845601320266724,
      "eval_runtime": 124.2695,
      "eval_samples_per_second": 33.918,
      "eval_steps_per_second": 8.482,
      "step": 375
    },
    {
      "epoch": 0.07512487512487513,
      "grad_norm": 2.2116332054138184,
      "learning_rate": 0.00017167309246649297,
      "loss": 2.1225,
      "step": 376
    },
    {
      "epoch": 0.07532467532467532,
      "grad_norm": 2.4644813537597656,
      "learning_rate": 0.0001715259004365791,
      "loss": 2.0706,
      "step": 377
    },
    {
      "epoch": 0.07552447552447553,
      "grad_norm": 1.862250804901123,
      "learning_rate": 0.00017137839043356484,
      "loss": 1.848,
      "step": 378
    },
    {
      "epoch": 0.07572427572427573,
      "grad_norm": 3.2452962398529053,
      "learning_rate": 0.00017123056311321562,
      "loss": 2.7598,
      "step": 379
    },
    {
      "epoch": 0.07592407592407592,
      "grad_norm": 2.1849207878112793,
      "learning_rate": 0.0001710824191327075,
      "loss": 1.5791,
      "step": 380
    },
    {
      "epoch": 0.07612387612387612,
      "grad_norm": 2.926933526992798,
      "learning_rate": 0.00017093395915062428,
      "loss": 2.5355,
      "step": 381
    },
    {
      "epoch": 0.07632367632367633,
      "grad_norm": 1.501541256904602,
      "learning_rate": 0.00017078518382695465,
      "loss": 2.3292,
      "step": 382
    },
    {
      "epoch": 0.07652347652347652,
      "grad_norm": 1.471026062965393,
      "learning_rate": 0.00017063609382308908,
      "loss": 1.426,
      "step": 383
    },
    {
      "epoch": 0.07672327672327672,
      "grad_norm": 2.449251651763916,
      "learning_rate": 0.00017048668980181698,
      "loss": 1.5493,
      "step": 384
    },
    {
      "epoch": 0.07692307692307693,
      "grad_norm": 1.6696304082870483,
      "learning_rate": 0.00017033697242732377,
      "loss": 2.0322,
      "step": 385
    },
    {
      "epoch": 0.07712287712287712,
      "grad_norm": 2.985755205154419,
      "learning_rate": 0.0001701869423651879,
      "loss": 1.8804,
      "step": 386
    },
    {
      "epoch": 0.07732267732267732,
      "grad_norm": 3.3332691192626953,
      "learning_rate": 0.00017003660028237793,
      "loss": 1.4652,
      "step": 387
    },
    {
      "epoch": 0.07752247752247753,
      "grad_norm": 4.17718505859375,
      "learning_rate": 0.00016988594684724947,
      "loss": 2.1647,
      "step": 388
    },
    {
      "epoch": 0.07772227772227772,
      "grad_norm": 3.0072596073150635,
      "learning_rate": 0.00016973498272954222,
      "loss": 1.9708,
      "step": 389
    },
    {
      "epoch": 0.07792207792207792,
      "grad_norm": 4.344801902770996,
      "learning_rate": 0.00016958370860037717,
      "loss": 1.5509,
      "step": 390
    },
    {
      "epoch": 0.07812187812187812,
      "grad_norm": 1.5232011079788208,
      "learning_rate": 0.00016943212513225345,
      "loss": 1.4471,
      "step": 391
    },
    {
      "epoch": 0.07832167832167833,
      "grad_norm": 3.0849902629852295,
      "learning_rate": 0.00016928023299904533,
      "loss": 1.9911,
      "step": 392
    },
    {
      "epoch": 0.07852147852147852,
      "grad_norm": 2.610097646713257,
      "learning_rate": 0.0001691280328759992,
      "loss": 1.7388,
      "step": 393
    },
    {
      "epoch": 0.07872127872127872,
      "grad_norm": 2.6489884853363037,
      "learning_rate": 0.00016897552543973084,
      "loss": 2.0695,
      "step": 394
    },
    {
      "epoch": 0.07892107892107893,
      "grad_norm": 2.018738031387329,
      "learning_rate": 0.00016882271136822206,
      "loss": 2.249,
      "step": 395
    },
    {
      "epoch": 0.07912087912087912,
      "grad_norm": 1.2624651193618774,
      "learning_rate": 0.0001686695913408179,
      "loss": 2.3341,
      "step": 396
    },
    {
      "epoch": 0.07932067932067932,
      "grad_norm": 2.4896445274353027,
      "learning_rate": 0.0001685161660382235,
      "loss": 2.1028,
      "step": 397
    },
    {
      "epoch": 0.07952047952047953,
      "grad_norm": 1.4938240051269531,
      "learning_rate": 0.00016836243614250113,
      "loss": 1.757,
      "step": 398
    },
    {
      "epoch": 0.07972027972027972,
      "grad_norm": 2.2095422744750977,
      "learning_rate": 0.00016820840233706719,
      "loss": 0.9808,
      "step": 399
    },
    {
      "epoch": 0.07992007992007992,
      "grad_norm": 2.796496868133545,
      "learning_rate": 0.0001680540653066891,
      "loss": 1.67,
      "step": 400
    },
    {
      "epoch": 0.08011988011988012,
      "grad_norm": 1.573728322982788,
      "learning_rate": 0.00016789942573748232,
      "loss": 2.2848,
      "step": 401
    },
    {
      "epoch": 0.08031968031968031,
      "grad_norm": 2.2960400581359863,
      "learning_rate": 0.0001677444843169072,
      "loss": 2.4782,
      "step": 402
    },
    {
      "epoch": 0.08051948051948052,
      "grad_norm": 3.4365086555480957,
      "learning_rate": 0.00016758924173376603,
      "loss": 1.9694,
      "step": 403
    },
    {
      "epoch": 0.08071928071928072,
      "grad_norm": 2.5896055698394775,
      "learning_rate": 0.0001674336986781999,
      "loss": 2.3352,
      "step": 404
    },
    {
      "epoch": 0.08091908091908091,
      "grad_norm": 3.2374393939971924,
      "learning_rate": 0.00016727785584168581,
      "loss": 1.8858,
      "step": 405
    },
    {
      "epoch": 0.08111888111888112,
      "grad_norm": 3.144979476928711,
      "learning_rate": 0.0001671217139170333,
      "loss": 1.3404,
      "step": 406
    },
    {
      "epoch": 0.08131868131868132,
      "grad_norm": 2.6684107780456543,
      "learning_rate": 0.00016696527359838154,
      "loss": 1.2872,
      "step": 407
    },
    {
      "epoch": 0.08151848151848153,
      "grad_norm": 3.5947470664978027,
      "learning_rate": 0.00016680853558119632,
      "loss": 2.1397,
      "step": 408
    },
    {
      "epoch": 0.08171828171828172,
      "grad_norm": 2.4237396717071533,
      "learning_rate": 0.0001666515005622668,
      "loss": 1.4599,
      "step": 409
    },
    {
      "epoch": 0.08191808191808192,
      "grad_norm": 2.8178930282592773,
      "learning_rate": 0.0001664941692397025,
      "loss": 2.29,
      "step": 410
    },
    {
      "epoch": 0.08211788211788212,
      "grad_norm": 2.091677665710449,
      "learning_rate": 0.00016633654231293013,
      "loss": 2.1065,
      "step": 411
    },
    {
      "epoch": 0.08231768231768231,
      "grad_norm": 0.9981814026832581,
      "learning_rate": 0.00016617862048269065,
      "loss": 2.2154,
      "step": 412
    },
    {
      "epoch": 0.08251748251748252,
      "grad_norm": 2.296738862991333,
      "learning_rate": 0.00016602040445103588,
      "loss": 2.2789,
      "step": 413
    },
    {
      "epoch": 0.08271728271728272,
      "grad_norm": 2.143744707107544,
      "learning_rate": 0.00016586189492132566,
      "loss": 1.4546,
      "step": 414
    },
    {
      "epoch": 0.08291708291708291,
      "grad_norm": 2.216097354888916,
      "learning_rate": 0.00016570309259822453,
      "loss": 1.6481,
      "step": 415
    },
    {
      "epoch": 0.08311688311688312,
      "grad_norm": 1.868310809135437,
      "learning_rate": 0.0001655439981876987,
      "loss": 1.3699,
      "step": 416
    },
    {
      "epoch": 0.08331668331668332,
      "grad_norm": 2.7406275272369385,
      "learning_rate": 0.00016538461239701277,
      "loss": 1.8958,
      "step": 417
    },
    {
      "epoch": 0.08351648351648351,
      "grad_norm": 3.3770556449890137,
      "learning_rate": 0.00016522493593472683,
      "loss": 2.5294,
      "step": 418
    },
    {
      "epoch": 0.08371628371628372,
      "grad_norm": 2.410560131072998,
      "learning_rate": 0.0001650649695106931,
      "loss": 1.594,
      "step": 419
    },
    {
      "epoch": 0.08391608391608392,
      "grad_norm": 1.2297099828720093,
      "learning_rate": 0.00016490471383605288,
      "loss": 2.0286,
      "step": 420
    },
    {
      "epoch": 0.08411588411588411,
      "grad_norm": 2.4507522583007812,
      "learning_rate": 0.00016474416962323325,
      "loss": 2.1023,
      "step": 421
    },
    {
      "epoch": 0.08431568431568431,
      "grad_norm": 1.9183270931243896,
      "learning_rate": 0.00016458333758594414,
      "loss": 2.5265,
      "step": 422
    },
    {
      "epoch": 0.08451548451548452,
      "grad_norm": 4.113091468811035,
      "learning_rate": 0.00016442221843917496,
      "loss": 1.8602,
      "step": 423
    },
    {
      "epoch": 0.08471528471528472,
      "grad_norm": 1.0285725593566895,
      "learning_rate": 0.00016426081289919143,
      "loss": 2.2546,
      "step": 424
    },
    {
      "epoch": 0.08491508491508491,
      "grad_norm": 2.8106017112731934,
      "learning_rate": 0.0001640991216835326,
      "loss": 3.0031,
      "step": 425
    },
    {
      "epoch": 0.08511488511488512,
      "grad_norm": 1.6420671939849854,
      "learning_rate": 0.00016393714551100734,
      "loss": 1.4067,
      "step": 426
    },
    {
      "epoch": 0.08531468531468532,
      "grad_norm": 1.8953157663345337,
      "learning_rate": 0.0001637748851016914,
      "loss": 2.1225,
      "step": 427
    },
    {
      "epoch": 0.08551448551448551,
      "grad_norm": 1.4816550016403198,
      "learning_rate": 0.00016361234117692413,
      "loss": 1.8523,
      "step": 428
    },
    {
      "epoch": 0.08571428571428572,
      "grad_norm": 1.441401720046997,
      "learning_rate": 0.00016344951445930526,
      "loss": 1.7483,
      "step": 429
    },
    {
      "epoch": 0.08591408591408592,
      "grad_norm": 0.9544094800949097,
      "learning_rate": 0.0001632864056726917,
      "loss": 0.9705,
      "step": 430
    },
    {
      "epoch": 0.08611388611388611,
      "grad_norm": 1.1085466146469116,
      "learning_rate": 0.00016312301554219426,
      "loss": 2.4607,
      "step": 431
    },
    {
      "epoch": 0.08631368631368631,
      "grad_norm": 2.34993839263916,
      "learning_rate": 0.00016295934479417453,
      "loss": 2.1435,
      "step": 432
    },
    {
      "epoch": 0.08651348651348652,
      "grad_norm": 1.7900936603546143,
      "learning_rate": 0.00016279539415624164,
      "loss": 1.2572,
      "step": 433
    },
    {
      "epoch": 0.08671328671328671,
      "grad_norm": 1.2673487663269043,
      "learning_rate": 0.0001626311643572489,
      "loss": 1.0396,
      "step": 434
    },
    {
      "epoch": 0.08691308691308691,
      "grad_norm": 2.250947952270508,
      "learning_rate": 0.00016246665612729074,
      "loss": 2.1033,
      "step": 435
    },
    {
      "epoch": 0.08711288711288712,
      "grad_norm": 1.7594767808914185,
      "learning_rate": 0.00016230187019769928,
      "loss": 2.3538,
      "step": 436
    },
    {
      "epoch": 0.0873126873126873,
      "grad_norm": 1.8499858379364014,
      "learning_rate": 0.00016213680730104124,
      "loss": 2.4163,
      "step": 437
    },
    {
      "epoch": 0.08751248751248751,
      "grad_norm": 1.0088387727737427,
      "learning_rate": 0.0001619714681711146,
      "loss": 2.0424,
      "step": 438
    },
    {
      "epoch": 0.08771228771228771,
      "grad_norm": 3.1308746337890625,
      "learning_rate": 0.00016180585354294536,
      "loss": 1.5136,
      "step": 439
    },
    {
      "epoch": 0.08791208791208792,
      "grad_norm": 2.5227863788604736,
      "learning_rate": 0.00016163996415278424,
      "loss": 2.2188,
      "step": 440
    },
    {
      "epoch": 0.08811188811188811,
      "grad_norm": 2.1084585189819336,
      "learning_rate": 0.00016147380073810346,
      "loss": 2.0339,
      "step": 441
    },
    {
      "epoch": 0.08831168831168831,
      "grad_norm": 1.4168033599853516,
      "learning_rate": 0.0001613073640375934,
      "loss": 0.8119,
      "step": 442
    },
    {
      "epoch": 0.08851148851148852,
      "grad_norm": 1.6228835582733154,
      "learning_rate": 0.00016114065479115946,
      "loss": 1.3543,
      "step": 443
    },
    {
      "epoch": 0.08871128871128871,
      "grad_norm": 1.6214308738708496,
      "learning_rate": 0.00016097367373991842,
      "loss": 1.1844,
      "step": 444
    },
    {
      "epoch": 0.08891108891108891,
      "grad_norm": 2.1330323219299316,
      "learning_rate": 0.00016080642162619565,
      "loss": 1.6831,
      "step": 445
    },
    {
      "epoch": 0.08911088911088912,
      "grad_norm": 2.502420425415039,
      "learning_rate": 0.0001606388991935214,
      "loss": 2.0339,
      "step": 446
    },
    {
      "epoch": 0.0893106893106893,
      "grad_norm": 2.0865745544433594,
      "learning_rate": 0.0001604711071866277,
      "loss": 1.5245,
      "step": 447
    },
    {
      "epoch": 0.08951048951048951,
      "grad_norm": 3.114449977874756,
      "learning_rate": 0.00016030304635144494,
      "loss": 2.4604,
      "step": 448
    },
    {
      "epoch": 0.08971028971028971,
      "grad_norm": 1.802714228630066,
      "learning_rate": 0.00016013471743509862,
      "loss": 1.9101,
      "step": 449
    },
    {
      "epoch": 0.0899100899100899,
      "grad_norm": 1.377928614616394,
      "learning_rate": 0.00015996612118590603,
      "loss": 1.2138,
      "step": 450
    },
    {
      "epoch": 0.09010989010989011,
      "grad_norm": 1.6390742063522339,
      "learning_rate": 0.00015979725835337294,
      "loss": 1.6222,
      "step": 451
    },
    {
      "epoch": 0.09030969030969031,
      "grad_norm": 2.1084725856781006,
      "learning_rate": 0.00015962812968819016,
      "loss": 1.988,
      "step": 452
    },
    {
      "epoch": 0.0905094905094905,
      "grad_norm": 3.1766951084136963,
      "learning_rate": 0.0001594587359422303,
      "loss": 1.7531,
      "step": 453
    },
    {
      "epoch": 0.09070929070929071,
      "grad_norm": 1.4322423934936523,
      "learning_rate": 0.0001592890778685444,
      "loss": 1.0366,
      "step": 454
    },
    {
      "epoch": 0.09090909090909091,
      "grad_norm": 1.6303942203521729,
      "learning_rate": 0.00015911915622135862,
      "loss": 1.4366,
      "step": 455
    },
    {
      "epoch": 0.09110889110889112,
      "grad_norm": 2.784449815750122,
      "learning_rate": 0.00015894897175607086,
      "loss": 1.9676,
      "step": 456
    },
    {
      "epoch": 0.0913086913086913,
      "grad_norm": 3.8181662559509277,
      "learning_rate": 0.00015877852522924732,
      "loss": 1.5624,
      "step": 457
    },
    {
      "epoch": 0.09150849150849151,
      "grad_norm": 2.6872711181640625,
      "learning_rate": 0.00015860781739861928,
      "loss": 2.2326,
      "step": 458
    },
    {
      "epoch": 0.09170829170829171,
      "grad_norm": 3.5563430786132812,
      "learning_rate": 0.00015843684902307962,
      "loss": 2.1697,
      "step": 459
    },
    {
      "epoch": 0.0919080919080919,
      "grad_norm": 1.6350597143173218,
      "learning_rate": 0.00015826562086267956,
      "loss": 1.942,
      "step": 460
    },
    {
      "epoch": 0.09210789210789211,
      "grad_norm": 4.524228572845459,
      "learning_rate": 0.00015809413367862512,
      "loss": 2.6147,
      "step": 461
    },
    {
      "epoch": 0.09230769230769231,
      "grad_norm": 3.938884735107422,
      "learning_rate": 0.00015792238823327388,
      "loss": 1.0589,
      "step": 462
    },
    {
      "epoch": 0.0925074925074925,
      "grad_norm": 1.4143668413162231,
      "learning_rate": 0.00015775038529013152,
      "loss": 1.4808,
      "step": 463
    },
    {
      "epoch": 0.09270729270729271,
      "grad_norm": 2.842125654220581,
      "learning_rate": 0.0001575781256138485,
      "loss": 1.847,
      "step": 464
    },
    {
      "epoch": 0.09290709290709291,
      "grad_norm": 1.4472795724868774,
      "learning_rate": 0.00015740560997021648,
      "loss": 2.4868,
      "step": 465
    },
    {
      "epoch": 0.0931068931068931,
      "grad_norm": 2.628919839859009,
      "learning_rate": 0.00015723283912616513,
      "loss": 1.879,
      "step": 466
    },
    {
      "epoch": 0.0933066933066933,
      "grad_norm": 0.9058033227920532,
      "learning_rate": 0.00015705981384975866,
      "loss": 2.0875,
      "step": 467
    },
    {
      "epoch": 0.09350649350649351,
      "grad_norm": 3.1023600101470947,
      "learning_rate": 0.0001568865349101923,
      "loss": 2.335,
      "step": 468
    },
    {
      "epoch": 0.0937062937062937,
      "grad_norm": 2.380549907684326,
      "learning_rate": 0.00015671300307778898,
      "loss": 2.0823,
      "step": 469
    },
    {
      "epoch": 0.0939060939060939,
      "grad_norm": 1.7399860620498657,
      "learning_rate": 0.00015653921912399589,
      "loss": 2.1513,
      "step": 470
    },
    {
      "epoch": 0.09410589410589411,
      "grad_norm": 3.1824607849121094,
      "learning_rate": 0.00015636518382138107,
      "loss": 1.9068,
      "step": 471
    },
    {
      "epoch": 0.09430569430569431,
      "grad_norm": 1.4199103116989136,
      "learning_rate": 0.0001561908979436299,
      "loss": 1.6113,
      "step": 472
    },
    {
      "epoch": 0.0945054945054945,
      "grad_norm": 2.7813353538513184,
      "learning_rate": 0.00015601636226554168,
      "loss": 2.5938,
      "step": 473
    },
    {
      "epoch": 0.09470529470529471,
      "grad_norm": 4.041593074798584,
      "learning_rate": 0.00015584157756302634,
      "loss": 2.2223,
      "step": 474
    },
    {
      "epoch": 0.09490509490509491,
      "grad_norm": 1.5122382640838623,
      "learning_rate": 0.0001556665446131007,
      "loss": 1.7002,
      "step": 475
    },
    {
      "epoch": 0.0951048951048951,
      "grad_norm": 2.3305814266204834,
      "learning_rate": 0.00015549126419388536,
      "loss": 1.5787,
      "step": 476
    },
    {
      "epoch": 0.0953046953046953,
      "grad_norm": 2.2571403980255127,
      "learning_rate": 0.0001553157370846009,
      "loss": 2.4153,
      "step": 477
    },
    {
      "epoch": 0.09550449550449551,
      "grad_norm": 4.9194560050964355,
      "learning_rate": 0.00015513996406556465,
      "loss": 1.4335,
      "step": 478
    },
    {
      "epoch": 0.0957042957042957,
      "grad_norm": 4.778641700744629,
      "learning_rate": 0.00015496394591818716,
      "loss": 2.2141,
      "step": 479
    },
    {
      "epoch": 0.0959040959040959,
      "grad_norm": 1.8974075317382812,
      "learning_rate": 0.0001547876834249687,
      "loss": 1.9282,
      "step": 480
    },
    {
      "epoch": 0.09610389610389611,
      "grad_norm": 0.8870239853858948,
      "learning_rate": 0.00015461117736949577,
      "loss": 2.2461,
      "step": 481
    },
    {
      "epoch": 0.0963036963036963,
      "grad_norm": 2.656989336013794,
      "learning_rate": 0.00015443442853643762,
      "loss": 2.2703,
      "step": 482
    },
    {
      "epoch": 0.0965034965034965,
      "grad_norm": 1.7446268796920776,
      "learning_rate": 0.00015425743771154294,
      "loss": 2.0833,
      "step": 483
    },
    {
      "epoch": 0.0967032967032967,
      "grad_norm": 2.6155710220336914,
      "learning_rate": 0.00015408020568163602,
      "loss": 1.268,
      "step": 484
    },
    {
      "epoch": 0.0969030969030969,
      "grad_norm": 2.1306145191192627,
      "learning_rate": 0.00015390273323461352,
      "loss": 1.8515,
      "step": 485
    },
    {
      "epoch": 0.0971028971028971,
      "grad_norm": 1.6019847393035889,
      "learning_rate": 0.0001537250211594409,
      "loss": 1.5859,
      "step": 486
    },
    {
      "epoch": 0.0973026973026973,
      "grad_norm": 1.953979730606079,
      "learning_rate": 0.0001535470702461489,
      "loss": 2.5188,
      "step": 487
    },
    {
      "epoch": 0.09750249750249751,
      "grad_norm": 2.7269322872161865,
      "learning_rate": 0.00015336888128583,
      "loss": 0.8919,
      "step": 488
    },
    {
      "epoch": 0.0977022977022977,
      "grad_norm": 2.2593624591827393,
      "learning_rate": 0.000153190455070635,
      "loss": 1.7071,
      "step": 489
    },
    {
      "epoch": 0.0979020979020979,
      "grad_norm": 1.0241385698318481,
      "learning_rate": 0.00015301179239376938,
      "loss": 1.3668,
      "step": 490
    },
    {
      "epoch": 0.09810189810189811,
      "grad_norm": 2.260610580444336,
      "learning_rate": 0.00015283289404948976,
      "loss": 1.9843,
      "step": 491
    },
    {
      "epoch": 0.0983016983016983,
      "grad_norm": 1.9773402214050293,
      "learning_rate": 0.0001526537608331006,
      "loss": 2.3181,
      "step": 492
    },
    {
      "epoch": 0.0985014985014985,
      "grad_norm": 3.522320508956909,
      "learning_rate": 0.00015247439354095041,
      "loss": 2.0929,
      "step": 493
    },
    {
      "epoch": 0.0987012987012987,
      "grad_norm": 3.586347818374634,
      "learning_rate": 0.00015229479297042823,
      "loss": 1.7053,
      "step": 494
    },
    {
      "epoch": 0.0989010989010989,
      "grad_norm": 2.394479274749756,
      "learning_rate": 0.00015211495991996027,
      "loss": 2.2255,
      "step": 495
    },
    {
      "epoch": 0.0991008991008991,
      "grad_norm": 3.0646183490753174,
      "learning_rate": 0.0001519348951890062,
      "loss": 1.7046,
      "step": 496
    },
    {
      "epoch": 0.0993006993006993,
      "grad_norm": 3.735779285430908,
      "learning_rate": 0.0001517545995780556,
      "loss": 2.729,
      "step": 497
    },
    {
      "epoch": 0.0995004995004995,
      "grad_norm": 4.216831207275391,
      "learning_rate": 0.00015157407388862452,
      "loss": 1.9828,
      "step": 498
    },
    {
      "epoch": 0.0997002997002997,
      "grad_norm": 2.256601333618164,
      "learning_rate": 0.00015139331892325179,
      "loss": 1.3887,
      "step": 499
    },
    {
      "epoch": 0.0999000999000999,
      "grad_norm": 3.887983798980713,
      "learning_rate": 0.0001512123354854955,
      "loss": 2.4133,
      "step": 500
    },
    {
      "epoch": 0.1000999000999001,
      "grad_norm": 1.5903120040893555,
      "learning_rate": 0.0001510311243799295,
      "loss": 1.6903,
      "step": 501
    },
    {
      "epoch": 0.1002997002997003,
      "grad_norm": 1.9416152238845825,
      "learning_rate": 0.00015084968641213958,
      "loss": 1.8254,
      "step": 502
    },
    {
      "epoch": 0.1004995004995005,
      "grad_norm": 1.4919044971466064,
      "learning_rate": 0.00015066802238872023,
      "loss": 1.7577,
      "step": 503
    },
    {
      "epoch": 0.1006993006993007,
      "grad_norm": 2.300889730453491,
      "learning_rate": 0.0001504861331172709,
      "loss": 2.8784,
      "step": 504
    },
    {
      "epoch": 0.1008991008991009,
      "grad_norm": 3.6400606632232666,
      "learning_rate": 0.0001503040194063922,
      "loss": 2.1229,
      "step": 505
    },
    {
      "epoch": 0.1010989010989011,
      "grad_norm": 2.3643879890441895,
      "learning_rate": 0.00015012168206568268,
      "loss": 1.7666,
      "step": 506
    },
    {
      "epoch": 0.1012987012987013,
      "grad_norm": 2.124890089035034,
      "learning_rate": 0.00014993912190573505,
      "loss": 0.8823,
      "step": 507
    },
    {
      "epoch": 0.1014985014985015,
      "grad_norm": 2.8733973503112793,
      "learning_rate": 0.00014975633973813242,
      "loss": 2.4345,
      "step": 508
    },
    {
      "epoch": 0.1016983016983017,
      "grad_norm": 3.874695301055908,
      "learning_rate": 0.00014957333637544503,
      "loss": 1.5017,
      "step": 509
    },
    {
      "epoch": 0.1018981018981019,
      "grad_norm": 2.798830270767212,
      "learning_rate": 0.00014939011263122634,
      "loss": 1.172,
      "step": 510
    },
    {
      "epoch": 0.1020979020979021,
      "grad_norm": 1.0153292417526245,
      "learning_rate": 0.0001492066693200096,
      "loss": 0.3735,
      "step": 511
    },
    {
      "epoch": 0.1022977022977023,
      "grad_norm": 5.334494113922119,
      "learning_rate": 0.00014902300725730413,
      "loss": 1.4743,
      "step": 512
    },
    {
      "epoch": 0.1024975024975025,
      "grad_norm": 2.3830583095550537,
      "learning_rate": 0.00014883912725959167,
      "loss": 2.3663,
      "step": 513
    },
    {
      "epoch": 0.10269730269730269,
      "grad_norm": 3.354517936706543,
      "learning_rate": 0.00014865503014432292,
      "loss": 2.1836,
      "step": 514
    },
    {
      "epoch": 0.1028971028971029,
      "grad_norm": 2.732463836669922,
      "learning_rate": 0.00014847071672991367,
      "loss": 2.3941,
      "step": 515
    },
    {
      "epoch": 0.1030969030969031,
      "grad_norm": 2.213967800140381,
      "learning_rate": 0.0001482861878357414,
      "loss": 1.6571,
      "step": 516
    },
    {
      "epoch": 0.10329670329670329,
      "grad_norm": 3.7244977951049805,
      "learning_rate": 0.00014810144428214144,
      "loss": 2.1976,
      "step": 517
    },
    {
      "epoch": 0.1034965034965035,
      "grad_norm": 3.412705183029175,
      "learning_rate": 0.0001479164868904034,
      "loss": 2.9507,
      "step": 518
    },
    {
      "epoch": 0.1036963036963037,
      "grad_norm": 1.5558984279632568,
      "learning_rate": 0.00014773131648276758,
      "loss": 2.6026,
      "step": 519
    },
    {
      "epoch": 0.1038961038961039,
      "grad_norm": 2.067119836807251,
      "learning_rate": 0.00014754593388242117,
      "loss": 1.8502,
      "step": 520
    },
    {
      "epoch": 0.1040959040959041,
      "grad_norm": 0.7480472922325134,
      "learning_rate": 0.0001473603399134948,
      "loss": 1.0575,
      "step": 521
    },
    {
      "epoch": 0.1042957042957043,
      "grad_norm": 3.0084283351898193,
      "learning_rate": 0.0001471745354010586,
      "loss": 2.097,
      "step": 522
    },
    {
      "epoch": 0.1044955044955045,
      "grad_norm": 1.0576696395874023,
      "learning_rate": 0.00014698852117111884,
      "loss": 1.3047,
      "step": 523
    },
    {
      "epoch": 0.10469530469530469,
      "grad_norm": 2.0393102169036865,
      "learning_rate": 0.000146802298050614,
      "loss": 1.6959,
      "step": 524
    },
    {
      "epoch": 0.1048951048951049,
      "grad_norm": 2.7678613662719727,
      "learning_rate": 0.0001466158668674112,
      "loss": 1.6035,
      "step": 525
    },
    {
      "epoch": 0.1050949050949051,
      "grad_norm": 2.7693581581115723,
      "learning_rate": 0.00014642922845030257,
      "loss": 1.1587,
      "step": 526
    },
    {
      "epoch": 0.10529470529470529,
      "grad_norm": 1.911803126335144,
      "learning_rate": 0.0001462423836290015,
      "loss": 2.1149,
      "step": 527
    },
    {
      "epoch": 0.1054945054945055,
      "grad_norm": 1.862065076828003,
      "learning_rate": 0.00014605533323413887,
      "loss": 1.504,
      "step": 528
    },
    {
      "epoch": 0.1056943056943057,
      "grad_norm": 2.4451651573181152,
      "learning_rate": 0.00014586807809725962,
      "loss": 2.44,
      "step": 529
    },
    {
      "epoch": 0.10589410589410589,
      "grad_norm": 1.9186125993728638,
      "learning_rate": 0.00014568061905081875,
      "loss": 2.2828,
      "step": 530
    },
    {
      "epoch": 0.10609390609390609,
      "grad_norm": 3.4011685848236084,
      "learning_rate": 0.00014549295692817778,
      "loss": 2.0329,
      "step": 531
    },
    {
      "epoch": 0.1062937062937063,
      "grad_norm": 2.4106664657592773,
      "learning_rate": 0.00014530509256360102,
      "loss": 2.237,
      "step": 532
    },
    {
      "epoch": 0.10649350649350649,
      "grad_norm": 2.6649742126464844,
      "learning_rate": 0.00014511702679225193,
      "loss": 1.918,
      "step": 533
    },
    {
      "epoch": 0.10669330669330669,
      "grad_norm": 1.6149412393569946,
      "learning_rate": 0.0001449287604501893,
      "loss": 1.2352,
      "step": 534
    },
    {
      "epoch": 0.1068931068931069,
      "grad_norm": 0.7237920761108398,
      "learning_rate": 0.00014474029437436348,
      "loss": 0.9355,
      "step": 535
    },
    {
      "epoch": 0.1070929070929071,
      "grad_norm": 2.466742753982544,
      "learning_rate": 0.00014455162940261285,
      "loss": 2.0733,
      "step": 536
    },
    {
      "epoch": 0.10729270729270729,
      "grad_norm": 2.180974006652832,
      "learning_rate": 0.0001443627663736599,
      "loss": 2.0414,
      "step": 537
    },
    {
      "epoch": 0.1074925074925075,
      "grad_norm": 2.411487102508545,
      "learning_rate": 0.00014417370612710778,
      "loss": 1.9468,
      "step": 538
    },
    {
      "epoch": 0.1076923076923077,
      "grad_norm": 2.308522939682007,
      "learning_rate": 0.00014398444950343623,
      "loss": 1.6725,
      "step": 539
    },
    {
      "epoch": 0.10789210789210789,
      "grad_norm": 2.054513931274414,
      "learning_rate": 0.00014379499734399798,
      "loss": 1.1317,
      "step": 540
    },
    {
      "epoch": 0.10809190809190809,
      "grad_norm": 2.5630626678466797,
      "learning_rate": 0.0001436053504910151,
      "loss": 1.651,
      "step": 541
    },
    {
      "epoch": 0.1082917082917083,
      "grad_norm": 3.0410654544830322,
      "learning_rate": 0.0001434155097875752,
      "loss": 2.0497,
      "step": 542
    },
    {
      "epoch": 0.10849150849150849,
      "grad_norm": 1.6541826725006104,
      "learning_rate": 0.00014322547607762762,
      "loss": 2.0056,
      "step": 543
    },
    {
      "epoch": 0.10869130869130869,
      "grad_norm": 3.749985933303833,
      "learning_rate": 0.0001430352502059797,
      "loss": 2.0926,
      "step": 544
    },
    {
      "epoch": 0.1088911088911089,
      "grad_norm": 1.9526426792144775,
      "learning_rate": 0.0001428448330182931,
      "loss": 2.177,
      "step": 545
    },
    {
      "epoch": 0.10909090909090909,
      "grad_norm": 2.2504589557647705,
      "learning_rate": 0.00014265422536107993,
      "loss": 2.119,
      "step": 546
    },
    {
      "epoch": 0.10929070929070929,
      "grad_norm": 1.8519624471664429,
      "learning_rate": 0.00014246342808169914,
      "loss": 2.4311,
      "step": 547
    },
    {
      "epoch": 0.1094905094905095,
      "grad_norm": 2.740975856781006,
      "learning_rate": 0.00014227244202835257,
      "loss": 1.737,
      "step": 548
    },
    {
      "epoch": 0.10969030969030968,
      "grad_norm": 2.296621799468994,
      "learning_rate": 0.0001420812680500813,
      "loss": 1.1392,
      "step": 549
    },
    {
      "epoch": 0.10989010989010989,
      "grad_norm": 1.7081011533737183,
      "learning_rate": 0.00014188990699676184,
      "loss": 2.0664,
      "step": 550
    },
    {
      "epoch": 0.11008991008991009,
      "grad_norm": 2.170546531677246,
      "learning_rate": 0.00014169835971910238,
      "loss": 1.2886,
      "step": 551
    },
    {
      "epoch": 0.1102897102897103,
      "grad_norm": 3.480308771133423,
      "learning_rate": 0.0001415066270686389,
      "loss": 1.7403,
      "step": 552
    },
    {
      "epoch": 0.11048951048951049,
      "grad_norm": 3.370931386947632,
      "learning_rate": 0.00014131470989773158,
      "loss": 1.6485,
      "step": 553
    },
    {
      "epoch": 0.11068931068931069,
      "grad_norm": 3.272455930709839,
      "learning_rate": 0.0001411226090595608,
      "loss": 2.2648,
      "step": 554
    },
    {
      "epoch": 0.1108891108891109,
      "grad_norm": 3.4591856002807617,
      "learning_rate": 0.00014093032540812348,
      "loss": 2.0151,
      "step": 555
    },
    {
      "epoch": 0.11108891108891109,
      "grad_norm": 1.3422034978866577,
      "learning_rate": 0.0001407378597982293,
      "loss": 2.5451,
      "step": 556
    },
    {
      "epoch": 0.11128871128871129,
      "grad_norm": 1.0069397687911987,
      "learning_rate": 0.00014054521308549673,
      "loss": 0.7346,
      "step": 557
    },
    {
      "epoch": 0.1114885114885115,
      "grad_norm": 1.197633147239685,
      "learning_rate": 0.0001403523861263495,
      "loss": 1.0659,
      "step": 558
    },
    {
      "epoch": 0.11168831168831168,
      "grad_norm": 2.9967596530914307,
      "learning_rate": 0.00014015937977801256,
      "loss": 1.9391,
      "step": 559
    },
    {
      "epoch": 0.11188811188811189,
      "grad_norm": 2.999469041824341,
      "learning_rate": 0.00013996619489850822,
      "loss": 1.5757,
      "step": 560
    },
    {
      "epoch": 0.11208791208791209,
      "grad_norm": 2.1205127239227295,
      "learning_rate": 0.00013977283234665273,
      "loss": 2.4094,
      "step": 561
    },
    {
      "epoch": 0.11228771228771228,
      "grad_norm": 3.0170955657958984,
      "learning_rate": 0.00013957929298205195,
      "loss": 2.1521,
      "step": 562
    },
    {
      "epoch": 0.11248751248751249,
      "grad_norm": 3.175715446472168,
      "learning_rate": 0.00013938557766509792,
      "loss": 2.041,
      "step": 563
    },
    {
      "epoch": 0.11268731268731269,
      "grad_norm": 3.651431083679199,
      "learning_rate": 0.0001391916872569648,
      "loss": 2.5195,
      "step": 564
    },
    {
      "epoch": 0.11288711288711288,
      "grad_norm": 4.294022560119629,
      "learning_rate": 0.00013899762261960518,
      "loss": 2.8541,
      "step": 565
    },
    {
      "epoch": 0.11308691308691309,
      "grad_norm": 1.0030198097229004,
      "learning_rate": 0.0001388033846157462,
      "loss": 0.723,
      "step": 566
    },
    {
      "epoch": 0.11328671328671329,
      "grad_norm": 1.8119988441467285,
      "learning_rate": 0.0001386089741088857,
      "loss": 1.9167,
      "step": 567
    },
    {
      "epoch": 0.1134865134865135,
      "grad_norm": 1.31764817237854,
      "learning_rate": 0.00013841439196328836,
      "loss": 2.2044,
      "step": 568
    },
    {
      "epoch": 0.11368631368631368,
      "grad_norm": 1.3381454944610596,
      "learning_rate": 0.00013821963904398193,
      "loss": 1.4934,
      "step": 569
    },
    {
      "epoch": 0.11388611388611389,
      "grad_norm": 2.025158643722534,
      "learning_rate": 0.00013802471621675338,
      "loss": 1.7147,
      "step": 570
    },
    {
      "epoch": 0.11408591408591409,
      "grad_norm": 1.6919548511505127,
      "learning_rate": 0.00013782962434814492,
      "loss": 2.3647,
      "step": 571
    },
    {
      "epoch": 0.11428571428571428,
      "grad_norm": 3.70638370513916,
      "learning_rate": 0.00013763436430545034,
      "loss": 2.6152,
      "step": 572
    },
    {
      "epoch": 0.11448551448551449,
      "grad_norm": 2.3201558589935303,
      "learning_rate": 0.00013743893695671096,
      "loss": 2.5022,
      "step": 573
    },
    {
      "epoch": 0.11468531468531469,
      "grad_norm": 3.132354497909546,
      "learning_rate": 0.00013724334317071198,
      "loss": 2.4926,
      "step": 574
    },
    {
      "epoch": 0.11488511488511488,
      "grad_norm": 2.2933902740478516,
      "learning_rate": 0.00013704758381697844,
      "loss": 1.9873,
      "step": 575
    },
    {
      "epoch": 0.11508491508491508,
      "grad_norm": 1.0434370040893555,
      "learning_rate": 0.00013685165976577146,
      "loss": 1.2052,
      "step": 576
    },
    {
      "epoch": 0.11528471528471529,
      "grad_norm": 2.1708195209503174,
      "learning_rate": 0.0001366555718880843,
      "loss": 1.6191,
      "step": 577
    },
    {
      "epoch": 0.11548451548451548,
      "grad_norm": 0.9505724906921387,
      "learning_rate": 0.00013645932105563844,
      "loss": 2.1538,
      "step": 578
    },
    {
      "epoch": 0.11568431568431568,
      "grad_norm": 2.761343479156494,
      "learning_rate": 0.00013626290814088005,
      "loss": 1.966,
      "step": 579
    },
    {
      "epoch": 0.11588411588411589,
      "grad_norm": 4.136228084564209,
      "learning_rate": 0.00013606633401697557,
      "loss": 2.4474,
      "step": 580
    },
    {
      "epoch": 0.11608391608391608,
      "grad_norm": 1.4657450914382935,
      "learning_rate": 0.00013586959955780824,
      "loss": 2.0576,
      "step": 581
    },
    {
      "epoch": 0.11628371628371628,
      "grad_norm": 0.8768656253814697,
      "learning_rate": 0.00013567270563797398,
      "loss": 2.1151,
      "step": 582
    },
    {
      "epoch": 0.11648351648351649,
      "grad_norm": 2.8868937492370605,
      "learning_rate": 0.00013547565313277776,
      "loss": 2.455,
      "step": 583
    },
    {
      "epoch": 0.11668331668331669,
      "grad_norm": 3.389960765838623,
      "learning_rate": 0.00013527844291822948,
      "loss": 2.6885,
      "step": 584
    },
    {
      "epoch": 0.11688311688311688,
      "grad_norm": 2.080451011657715,
      "learning_rate": 0.0001350810758710401,
      "loss": 2.6304,
      "step": 585
    },
    {
      "epoch": 0.11708291708291708,
      "grad_norm": 2.384458303451538,
      "learning_rate": 0.00013488355286861783,
      "loss": 1.715,
      "step": 586
    },
    {
      "epoch": 0.11728271728271729,
      "grad_norm": 3.0337419509887695,
      "learning_rate": 0.0001346858747890642,
      "loss": 1.9547,
      "step": 587
    },
    {
      "epoch": 0.11748251748251748,
      "grad_norm": 2.653172492980957,
      "learning_rate": 0.00013448804251117003,
      "loss": 2.0736,
      "step": 588
    },
    {
      "epoch": 0.11768231768231768,
      "grad_norm": 1.4786633253097534,
      "learning_rate": 0.0001342900569144119,
      "loss": 1.6303,
      "step": 589
    },
    {
      "epoch": 0.11788211788211789,
      "grad_norm": 3.670684814453125,
      "learning_rate": 0.0001340919188789477,
      "loss": 2.5406,
      "step": 590
    },
    {
      "epoch": 0.11808191808191808,
      "grad_norm": 3.596186637878418,
      "learning_rate": 0.00013389362928561317,
      "loss": 1.9135,
      "step": 591
    },
    {
      "epoch": 0.11828171828171828,
      "grad_norm": 1.9404085874557495,
      "learning_rate": 0.00013369518901591772,
      "loss": 0.9342,
      "step": 592
    },
    {
      "epoch": 0.11848151848151849,
      "grad_norm": 3.885128974914551,
      "learning_rate": 0.00013349659895204067,
      "loss": 1.9232,
      "step": 593
    },
    {
      "epoch": 0.11868131868131868,
      "grad_norm": 1.6739671230316162,
      "learning_rate": 0.0001332978599768272,
      "loss": 1.9674,
      "step": 594
    },
    {
      "epoch": 0.11888111888111888,
      "grad_norm": 2.8616886138916016,
      "learning_rate": 0.00013309897297378455,
      "loss": 2.5587,
      "step": 595
    },
    {
      "epoch": 0.11908091908091908,
      "grad_norm": 2.5372321605682373,
      "learning_rate": 0.00013289993882707797,
      "loss": 2.0233,
      "step": 596
    },
    {
      "epoch": 0.11928071928071927,
      "grad_norm": 1.5624561309814453,
      "learning_rate": 0.00013270075842152678,
      "loss": 1.6396,
      "step": 597
    },
    {
      "epoch": 0.11948051948051948,
      "grad_norm": 1.3420947790145874,
      "learning_rate": 0.00013250143264260074,
      "loss": 1.693,
      "step": 598
    },
    {
      "epoch": 0.11968031968031968,
      "grad_norm": 2.562721014022827,
      "learning_rate": 0.0001323019623764156,
      "loss": 1.0703,
      "step": 599
    },
    {
      "epoch": 0.11988011988011989,
      "grad_norm": 2.2994723320007324,
      "learning_rate": 0.00013210234850972964,
      "loss": 1.8199,
      "step": 600
    },
    {
      "epoch": 0.12007992007992008,
      "grad_norm": 2.231781005859375,
      "learning_rate": 0.0001319025919299394,
      "loss": 2.3085,
      "step": 601
    },
    {
      "epoch": 0.12027972027972028,
      "grad_norm": 0.7306497693061829,
      "learning_rate": 0.00013170269352507597,
      "loss": 0.7058,
      "step": 602
    },
    {
      "epoch": 0.12047952047952049,
      "grad_norm": 3.6294503211975098,
      "learning_rate": 0.0001315026541838008,
      "loss": 2.427,
      "step": 603
    },
    {
      "epoch": 0.12067932067932068,
      "grad_norm": 1.1597257852554321,
      "learning_rate": 0.00013130247479540202,
      "loss": 2.5514,
      "step": 604
    },
    {
      "epoch": 0.12087912087912088,
      "grad_norm": 1.6528184413909912,
      "learning_rate": 0.00013110215624979025,
      "loss": 1.3491,
      "step": 605
    },
    {
      "epoch": 0.12107892107892108,
      "grad_norm": 1.5021053552627563,
      "learning_rate": 0.00013090169943749476,
      "loss": 1.5478,
      "step": 606
    },
    {
      "epoch": 0.12127872127872127,
      "grad_norm": 1.1097666025161743,
      "learning_rate": 0.00013070110524965954,
      "loss": 1.1394,
      "step": 607
    },
    {
      "epoch": 0.12147852147852148,
      "grad_norm": 1.7656742334365845,
      "learning_rate": 0.00013050037457803924,
      "loss": 0.9063,
      "step": 608
    },
    {
      "epoch": 0.12167832167832168,
      "grad_norm": 1.1813184022903442,
      "learning_rate": 0.0001302995083149953,
      "loss": 2.2282,
      "step": 609
    },
    {
      "epoch": 0.12187812187812187,
      "grad_norm": 3.9542620182037354,
      "learning_rate": 0.0001300985073534919,
      "loss": 1.6272,
      "step": 610
    },
    {
      "epoch": 0.12207792207792208,
      "grad_norm": 2.0715441703796387,
      "learning_rate": 0.00012989737258709203,
      "loss": 1.5492,
      "step": 611
    },
    {
      "epoch": 0.12227772227772228,
      "grad_norm": 1.8157408237457275,
      "learning_rate": 0.00012969610490995358,
      "loss": 1.7746,
      "step": 612
    },
    {
      "epoch": 0.12247752247752247,
      "grad_norm": 3.63930344581604,
      "learning_rate": 0.00012949470521682528,
      "loss": 2.1908,
      "step": 613
    },
    {
      "epoch": 0.12267732267732268,
      "grad_norm": 2.6896297931671143,
      "learning_rate": 0.0001292931744030427,
      "loss": 1.8941,
      "step": 614
    },
    {
      "epoch": 0.12287712287712288,
      "grad_norm": 3.2297921180725098,
      "learning_rate": 0.0001290915133645243,
      "loss": 1.7448,
      "step": 615
    },
    {
      "epoch": 0.12307692307692308,
      "grad_norm": 1.6754640340805054,
      "learning_rate": 0.00012888972299776754,
      "loss": 1.4958,
      "step": 616
    },
    {
      "epoch": 0.12327672327672327,
      "grad_norm": 2.0380616188049316,
      "learning_rate": 0.00012868780419984482,
      "loss": 2.0344,
      "step": 617
    },
    {
      "epoch": 0.12347652347652348,
      "grad_norm": 2.275420665740967,
      "learning_rate": 0.00012848575786839943,
      "loss": 1.8151,
      "step": 618
    },
    {
      "epoch": 0.12367632367632368,
      "grad_norm": 2.516294479370117,
      "learning_rate": 0.0001282835849016416,
      "loss": 1.1641,
      "step": 619
    },
    {
      "epoch": 0.12387612387612387,
      "grad_norm": 2.093801259994507,
      "learning_rate": 0.00012808128619834461,
      "loss": 2.2376,
      "step": 620
    },
    {
      "epoch": 0.12407592407592408,
      "grad_norm": 2.514761447906494,
      "learning_rate": 0.0001278788626578407,
      "loss": 2.7257,
      "step": 621
    },
    {
      "epoch": 0.12427572427572428,
      "grad_norm": 2.149421215057373,
      "learning_rate": 0.00012767631518001698,
      "loss": 2.4683,
      "step": 622
    },
    {
      "epoch": 0.12447552447552447,
      "grad_norm": 1.1154927015304565,
      "learning_rate": 0.00012747364466531163,
      "loss": 0.5757,
      "step": 623
    },
    {
      "epoch": 0.12467532467532468,
      "grad_norm": 2.641796112060547,
      "learning_rate": 0.00012727085201470973,
      "loss": 2.2154,
      "step": 624
    },
    {
      "epoch": 0.12487512487512488,
      "grad_norm": 2.961432933807373,
      "learning_rate": 0.00012706793812973941,
      "loss": 1.5613,
      "step": 625
    },
    {
      "epoch": 0.12507492507492507,
      "grad_norm": 3.3777637481689453,
      "learning_rate": 0.0001268649039124677,
      "loss": 1.9995,
      "step": 626
    },
    {
      "epoch": 0.12527472527472527,
      "grad_norm": 2.4651803970336914,
      "learning_rate": 0.00012666175026549662,
      "loss": 2.5503,
      "step": 627
    },
    {
      "epoch": 0.12547452547452548,
      "grad_norm": 1.7901756763458252,
      "learning_rate": 0.000126458478091959,
      "loss": 2.0579,
      "step": 628
    },
    {
      "epoch": 0.12567432567432568,
      "grad_norm": 3.7804133892059326,
      "learning_rate": 0.00012625508829551473,
      "loss": 2.0917,
      "step": 629
    },
    {
      "epoch": 0.1258741258741259,
      "grad_norm": 2.7492520809173584,
      "learning_rate": 0.00012605158178034654,
      "loss": 1.5869,
      "step": 630
    },
    {
      "epoch": 0.12607392607392606,
      "grad_norm": 1.8400332927703857,
      "learning_rate": 0.00012584795945115603,
      "loss": 1.7763,
      "step": 631
    },
    {
      "epoch": 0.12627372627372627,
      "grad_norm": 1.9279481172561646,
      "learning_rate": 0.0001256442222131597,
      "loss": 2.1263,
      "step": 632
    },
    {
      "epoch": 0.12647352647352647,
      "grad_norm": 3.282273530960083,
      "learning_rate": 0.0001254403709720848,
      "loss": 1.558,
      "step": 633
    },
    {
      "epoch": 0.12667332667332667,
      "grad_norm": 4.137247085571289,
      "learning_rate": 0.0001252364066341655,
      "loss": 1.8655,
      "step": 634
    },
    {
      "epoch": 0.12687312687312688,
      "grad_norm": 1.7836055755615234,
      "learning_rate": 0.00012503233010613865,
      "loss": 2.6928,
      "step": 635
    },
    {
      "epoch": 0.12707292707292708,
      "grad_norm": 3.0559756755828857,
      "learning_rate": 0.00012482814229523997,
      "loss": 1.7622,
      "step": 636
    },
    {
      "epoch": 0.12727272727272726,
      "grad_norm": 0.9399649500846863,
      "learning_rate": 0.00012462384410919975,
      "loss": 0.5781,
      "step": 637
    },
    {
      "epoch": 0.12747252747252746,
      "grad_norm": 1.5372148752212524,
      "learning_rate": 0.00012441943645623903,
      "loss": 1.6115,
      "step": 638
    },
    {
      "epoch": 0.12767232767232767,
      "grad_norm": 1.7965742349624634,
      "learning_rate": 0.00012421492024506555,
      "loss": 0.9174,
      "step": 639
    },
    {
      "epoch": 0.12787212787212787,
      "grad_norm": 2.1579504013061523,
      "learning_rate": 0.00012401029638486953,
      "loss": 2.1106,
      "step": 640
    },
    {
      "epoch": 0.12807192807192808,
      "grad_norm": 2.24662709236145,
      "learning_rate": 0.0001238055657853198,
      "loss": 1.5097,
      "step": 641
    },
    {
      "epoch": 0.12827172827172828,
      "grad_norm": 1.7009669542312622,
      "learning_rate": 0.00012360072935655982,
      "loss": 1.1661,
      "step": 642
    },
    {
      "epoch": 0.12847152847152848,
      "grad_norm": 2.8562543392181396,
      "learning_rate": 0.00012339578800920332,
      "loss": 1.4536,
      "step": 643
    },
    {
      "epoch": 0.12867132867132866,
      "grad_norm": 2.3831987380981445,
      "learning_rate": 0.00012319074265433063,
      "loss": 1.386,
      "step": 644
    },
    {
      "epoch": 0.12887112887112886,
      "grad_norm": 3.724574089050293,
      "learning_rate": 0.00012298559420348437,
      "loss": 2.641,
      "step": 645
    },
    {
      "epoch": 0.12907092907092907,
      "grad_norm": 1.9762496948242188,
      "learning_rate": 0.00012278034356866545,
      "loss": 2.2945,
      "step": 646
    },
    {
      "epoch": 0.12927072927072927,
      "grad_norm": 3.153521776199341,
      "learning_rate": 0.00012257499166232907,
      "loss": 2.8244,
      "step": 647
    },
    {
      "epoch": 0.12947052947052948,
      "grad_norm": 2.0080628395080566,
      "learning_rate": 0.0001223695393973807,
      "loss": 1.9367,
      "step": 648
    },
    {
      "epoch": 0.12967032967032968,
      "grad_norm": 2.840522050857544,
      "learning_rate": 0.0001221639876871719,
      "loss": 1.6678,
      "step": 649
    },
    {
      "epoch": 0.12987012987012986,
      "grad_norm": 2.415473699569702,
      "learning_rate": 0.0001219583374454963,
      "loss": 1.4156,
      "step": 650
    },
    {
      "epoch": 0.13006993006993006,
      "grad_norm": 1.6977438926696777,
      "learning_rate": 0.00012175258958658564,
      "loss": 1.4141,
      "step": 651
    },
    {
      "epoch": 0.13026973026973027,
      "grad_norm": 3.044330596923828,
      "learning_rate": 0.00012154674502510555,
      "loss": 2.0283,
      "step": 652
    },
    {
      "epoch": 0.13046953046953047,
      "grad_norm": 0.9067225456237793,
      "learning_rate": 0.00012134080467615159,
      "loss": 1.089,
      "step": 653
    },
    {
      "epoch": 0.13066933066933067,
      "grad_norm": 0.9506210684776306,
      "learning_rate": 0.00012113476945524513,
      "loss": 0.9395,
      "step": 654
    },
    {
      "epoch": 0.13086913086913088,
      "grad_norm": 2.784560203552246,
      "learning_rate": 0.00012092864027832933,
      "loss": 2.1588,
      "step": 655
    },
    {
      "epoch": 0.13106893106893108,
      "grad_norm": 3.0516326427459717,
      "learning_rate": 0.000120722418061765,
      "loss": 2.3833,
      "step": 656
    },
    {
      "epoch": 0.13126873126873126,
      "grad_norm": 3.062882900238037,
      "learning_rate": 0.0001205161037223266,
      "loss": 1.7638,
      "step": 657
    },
    {
      "epoch": 0.13146853146853146,
      "grad_norm": 2.81889009475708,
      "learning_rate": 0.00012030969817719808,
      "loss": 2.0823,
      "step": 658
    },
    {
      "epoch": 0.13166833166833167,
      "grad_norm": 2.2523906230926514,
      "learning_rate": 0.00012010320234396894,
      "loss": 1.7256,
      "step": 659
    },
    {
      "epoch": 0.13186813186813187,
      "grad_norm": 3.8378658294677734,
      "learning_rate": 0.00011989661714062999,
      "loss": 2.3166,
      "step": 660
    },
    {
      "epoch": 0.13206793206793208,
      "grad_norm": 0.8830970525741577,
      "learning_rate": 0.0001196899434855693,
      "loss": 0.9049,
      "step": 661
    },
    {
      "epoch": 0.13226773226773228,
      "grad_norm": 3.724817991256714,
      "learning_rate": 0.00011948318229756827,
      "loss": 2.1269,
      "step": 662
    },
    {
      "epoch": 0.13246753246753246,
      "grad_norm": 5.8385210037231445,
      "learning_rate": 0.00011927633449579735,
      "loss": 2.0566,
      "step": 663
    },
    {
      "epoch": 0.13266733266733266,
      "grad_norm": 3.483747959136963,
      "learning_rate": 0.0001190694009998121,
      "loss": 2.1309,
      "step": 664
    },
    {
      "epoch": 0.13286713286713286,
      "grad_norm": 1.1793261766433716,
      "learning_rate": 0.00011886238272954897,
      "loss": 0.5984,
      "step": 665
    },
    {
      "epoch": 0.13306693306693307,
      "grad_norm": 2.0061557292938232,
      "learning_rate": 0.00011865528060532127,
      "loss": 1.1805,
      "step": 666
    },
    {
      "epoch": 0.13326673326673327,
      "grad_norm": 1.343865156173706,
      "learning_rate": 0.0001184480955478152,
      "loss": 1.2641,
      "step": 667
    },
    {
      "epoch": 0.13346653346653348,
      "grad_norm": 3.0216474533081055,
      "learning_rate": 0.00011824082847808558,
      "loss": 2.4026,
      "step": 668
    },
    {
      "epoch": 0.13366633366633365,
      "grad_norm": 1.6855766773223877,
      "learning_rate": 0.00011803348031755179,
      "loss": 2.4982,
      "step": 669
    },
    {
      "epoch": 0.13386613386613386,
      "grad_norm": 2.2669217586517334,
      "learning_rate": 0.0001178260519879937,
      "loss": 2.1515,
      "step": 670
    },
    {
      "epoch": 0.13406593406593406,
      "grad_norm": 1.919115424156189,
      "learning_rate": 0.00011761854441154767,
      "loss": 1.6673,
      "step": 671
    },
    {
      "epoch": 0.13426573426573427,
      "grad_norm": 1.8461419343948364,
      "learning_rate": 0.00011741095851070228,
      "loss": 2.1981,
      "step": 672
    },
    {
      "epoch": 0.13446553446553447,
      "grad_norm": 2.5164804458618164,
      "learning_rate": 0.00011720329520829429,
      "loss": 1.9067,
      "step": 673
    },
    {
      "epoch": 0.13466533466533467,
      "grad_norm": 1.815097451210022,
      "learning_rate": 0.0001169955554275046,
      "loss": 2.1429,
      "step": 674
    },
    {
      "epoch": 0.13486513486513488,
      "grad_norm": 2.3510656356811523,
      "learning_rate": 0.0001167877400918541,
      "loss": 1.3227,
      "step": 675
    },
    {
      "epoch": 0.13506493506493505,
      "grad_norm": 2.166733503341675,
      "learning_rate": 0.00011657985012519952,
      "loss": 2.4444,
      "step": 676
    },
    {
      "epoch": 0.13526473526473526,
      "grad_norm": 3.075360059738159,
      "learning_rate": 0.00011637188645172944,
      "loss": 2.2248,
      "step": 677
    },
    {
      "epoch": 0.13546453546453546,
      "grad_norm": 3.347303867340088,
      "learning_rate": 0.00011616384999596006,
      "loss": 1.3446,
      "step": 678
    },
    {
      "epoch": 0.13566433566433567,
      "grad_norm": 1.3850624561309814,
      "learning_rate": 0.00011595574168273111,
      "loss": 1.7409,
      "step": 679
    },
    {
      "epoch": 0.13586413586413587,
      "grad_norm": 2.635024309158325,
      "learning_rate": 0.0001157475624372018,
      "loss": 1.7902,
      "step": 680
    },
    {
      "epoch": 0.13606393606393608,
      "grad_norm": 2.3166441917419434,
      "learning_rate": 0.0001155393131848467,
      "loss": 0.9,
      "step": 681
    },
    {
      "epoch": 0.13626373626373625,
      "grad_norm": 1.494813084602356,
      "learning_rate": 0.00011533099485145155,
      "loss": 1.9292,
      "step": 682
    },
    {
      "epoch": 0.13646353646353646,
      "grad_norm": 0.8544937968254089,
      "learning_rate": 0.00011512260836310924,
      "loss": 0.7824,
      "step": 683
    },
    {
      "epoch": 0.13666333666333666,
      "grad_norm": 1.836879849433899,
      "learning_rate": 0.00011491415464621562,
      "loss": 1.7824,
      "step": 684
    },
    {
      "epoch": 0.13686313686313686,
      "grad_norm": 1.7532786130905151,
      "learning_rate": 0.00011470563462746541,
      "loss": 1.1342,
      "step": 685
    },
    {
      "epoch": 0.13706293706293707,
      "grad_norm": 2.1970179080963135,
      "learning_rate": 0.00011449704923384812,
      "loss": 2.1155,
      "step": 686
    },
    {
      "epoch": 0.13726273726273727,
      "grad_norm": 2.904566526412964,
      "learning_rate": 0.00011428839939264382,
      "loss": 2.2281,
      "step": 687
    },
    {
      "epoch": 0.13746253746253748,
      "grad_norm": 1.1816821098327637,
      "learning_rate": 0.0001140796860314191,
      "loss": 1.7829,
      "step": 688
    },
    {
      "epoch": 0.13766233766233765,
      "grad_norm": 3.3773508071899414,
      "learning_rate": 0.00011387091007802297,
      "loss": 1.516,
      "step": 689
    },
    {
      "epoch": 0.13786213786213786,
      "grad_norm": 0.7771274447441101,
      "learning_rate": 0.0001136620724605827,
      "loss": 1.017,
      "step": 690
    },
    {
      "epoch": 0.13806193806193806,
      "grad_norm": 1.6280626058578491,
      "learning_rate": 0.00011345317410749964,
      "loss": 2.2575,
      "step": 691
    },
    {
      "epoch": 0.13826173826173827,
      "grad_norm": 2.046172857284546,
      "learning_rate": 0.00011324421594744516,
      "loss": 2.0589,
      "step": 692
    },
    {
      "epoch": 0.13846153846153847,
      "grad_norm": 2.468099594116211,
      "learning_rate": 0.00011303519890935656,
      "loss": 1.9645,
      "step": 693
    },
    {
      "epoch": 0.13866133866133867,
      "grad_norm": 2.076395034790039,
      "learning_rate": 0.00011282612392243286,
      "loss": 1.8338,
      "step": 694
    },
    {
      "epoch": 0.13886113886113885,
      "grad_norm": 2.757429599761963,
      "learning_rate": 0.00011261699191613066,
      "loss": 2.118,
      "step": 695
    },
    {
      "epoch": 0.13906093906093905,
      "grad_norm": 1.4514985084533691,
      "learning_rate": 0.00011240780382016005,
      "loss": 1.0355,
      "step": 696
    },
    {
      "epoch": 0.13926073926073926,
      "grad_norm": 1.5118156671524048,
      "learning_rate": 0.00011219856056448051,
      "loss": 2.0643,
      "step": 697
    },
    {
      "epoch": 0.13946053946053946,
      "grad_norm": 2.423887014389038,
      "learning_rate": 0.00011198926307929664,
      "loss": 0.9427,
      "step": 698
    },
    {
      "epoch": 0.13966033966033967,
      "grad_norm": 3.055157423019409,
      "learning_rate": 0.00011177991229505431,
      "loss": 2.1605,
      "step": 699
    },
    {
      "epoch": 0.13986013986013987,
      "grad_norm": 2.7236547470092773,
      "learning_rate": 0.00011157050914243614,
      "loss": 1.933,
      "step": 700
    },
    {
      "epoch": 0.14005994005994005,
      "grad_norm": 1.828993558883667,
      "learning_rate": 0.00011136105455235766,
      "loss": 1.8005,
      "step": 701
    },
    {
      "epoch": 0.14025974025974025,
      "grad_norm": 1.3731114864349365,
      "learning_rate": 0.00011115154945596305,
      "loss": 2.7275,
      "step": 702
    },
    {
      "epoch": 0.14045954045954046,
      "grad_norm": 1.934890627861023,
      "learning_rate": 0.00011094199478462095,
      "loss": 2.5253,
      "step": 703
    },
    {
      "epoch": 0.14065934065934066,
      "grad_norm": 1.7792664766311646,
      "learning_rate": 0.00011073239146992054,
      "loss": 1.7212,
      "step": 704
    },
    {
      "epoch": 0.14085914085914086,
      "grad_norm": 2.2803008556365967,
      "learning_rate": 0.00011052274044366711,
      "loss": 2.2409,
      "step": 705
    },
    {
      "epoch": 0.14105894105894107,
      "grad_norm": 3.5123517513275146,
      "learning_rate": 0.00011031304263787812,
      "loss": 1.2577,
      "step": 706
    },
    {
      "epoch": 0.14125874125874127,
      "grad_norm": 3.04335618019104,
      "learning_rate": 0.00011010329898477891,
      "loss": 2.1747,
      "step": 707
    },
    {
      "epoch": 0.14145854145854145,
      "grad_norm": 1.5780164003372192,
      "learning_rate": 0.0001098935104167988,
      "loss": 2.0881,
      "step": 708
    },
    {
      "epoch": 0.14165834165834165,
      "grad_norm": 1.4799244403839111,
      "learning_rate": 0.00010968367786656663,
      "loss": 0.7509,
      "step": 709
    },
    {
      "epoch": 0.14185814185814186,
      "grad_norm": 3.0631210803985596,
      "learning_rate": 0.00010947380226690684,
      "loss": 1.3353,
      "step": 710
    },
    {
      "epoch": 0.14205794205794206,
      "grad_norm": 2.7948882579803467,
      "learning_rate": 0.00010926388455083522,
      "loss": 2.1038,
      "step": 711
    },
    {
      "epoch": 0.14225774225774226,
      "grad_norm": 1.5561167001724243,
      "learning_rate": 0.00010905392565155477,
      "loss": 2.024,
      "step": 712
    },
    {
      "epoch": 0.14245754245754247,
      "grad_norm": 2.1947484016418457,
      "learning_rate": 0.00010884392650245165,
      "loss": 1.511,
      "step": 713
    },
    {
      "epoch": 0.14265734265734265,
      "grad_norm": 2.972017526626587,
      "learning_rate": 0.00010863388803709089,
      "loss": 2.0617,
      "step": 714
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 3.507697105407715,
      "learning_rate": 0.00010842381118921232,
      "loss": 2.3228,
      "step": 715
    },
    {
      "epoch": 0.14305694305694305,
      "grad_norm": 3.8943817615509033,
      "learning_rate": 0.00010821369689272638,
      "loss": 1.8042,
      "step": 716
    },
    {
      "epoch": 0.14325674325674326,
      "grad_norm": 1.8386285305023193,
      "learning_rate": 0.00010800354608171003,
      "loss": 2.4331,
      "step": 717
    },
    {
      "epoch": 0.14345654345654346,
      "grad_norm": 2.6339380741119385,
      "learning_rate": 0.00010779335969040252,
      "loss": 2.0043,
      "step": 718
    },
    {
      "epoch": 0.14365634365634367,
      "grad_norm": 2.3876147270202637,
      "learning_rate": 0.00010758313865320134,
      "loss": 2.152,
      "step": 719
    },
    {
      "epoch": 0.14385614385614387,
      "grad_norm": 2.3183741569519043,
      "learning_rate": 0.00010737288390465792,
      "loss": 2.4199,
      "step": 720
    },
    {
      "epoch": 0.14405594405594405,
      "grad_norm": 2.8862783908843994,
      "learning_rate": 0.00010716259637947357,
      "loss": 1.8511,
      "step": 721
    },
    {
      "epoch": 0.14425574425574425,
      "grad_norm": 2.358572483062744,
      "learning_rate": 0.00010695227701249537,
      "loss": 1.7872,
      "step": 722
    },
    {
      "epoch": 0.14445554445554445,
      "grad_norm": 1.7442290782928467,
      "learning_rate": 0.00010674192673871191,
      "loss": 2.0053,
      "step": 723
    },
    {
      "epoch": 0.14465534465534466,
      "grad_norm": 2.9464476108551025,
      "learning_rate": 0.00010653154649324917,
      "loss": 2.5362,
      "step": 724
    },
    {
      "epoch": 0.14485514485514486,
      "grad_norm": 0.9960998296737671,
      "learning_rate": 0.00010632113721136636,
      "loss": 0.816,
      "step": 725
    },
    {
      "epoch": 0.14505494505494507,
      "grad_norm": 1.9500738382339478,
      "learning_rate": 0.00010611069982845183,
      "loss": 2.3313,
      "step": 726
    },
    {
      "epoch": 0.14525474525474524,
      "grad_norm": 1.7992953062057495,
      "learning_rate": 0.00010590023528001884,
      "loss": 1.3785,
      "step": 727
    },
    {
      "epoch": 0.14545454545454545,
      "grad_norm": 2.0980491638183594,
      "learning_rate": 0.00010568974450170139,
      "loss": 1.4056,
      "step": 728
    },
    {
      "epoch": 0.14565434565434565,
      "grad_norm": 1.7661640644073486,
      "learning_rate": 0.00010547922842925008,
      "loss": 2.2884,
      "step": 729
    },
    {
      "epoch": 0.14585414585414586,
      "grad_norm": 2.7894139289855957,
      "learning_rate": 0.00010526868799852796,
      "loss": 2.2122,
      "step": 730
    },
    {
      "epoch": 0.14605394605394606,
      "grad_norm": 0.9134557247161865,
      "learning_rate": 0.0001050581241455064,
      "loss": 2.2506,
      "step": 731
    },
    {
      "epoch": 0.14625374625374626,
      "grad_norm": 4.833321571350098,
      "learning_rate": 0.00010484753780626089,
      "loss": 1.9774,
      "step": 732
    },
    {
      "epoch": 0.14645354645354644,
      "grad_norm": 1.5540224313735962,
      "learning_rate": 0.00010463692991696685,
      "loss": 1.5717,
      "step": 733
    },
    {
      "epoch": 0.14665334665334664,
      "grad_norm": 4.989921569824219,
      "learning_rate": 0.00010442630141389549,
      "loss": 2.0861,
      "step": 734
    },
    {
      "epoch": 0.14685314685314685,
      "grad_norm": 2.135622978210449,
      "learning_rate": 0.00010421565323340971,
      "loss": 1.9362,
      "step": 735
    },
    {
      "epoch": 0.14705294705294705,
      "grad_norm": 2.7071893215179443,
      "learning_rate": 0.00010400498631195992,
      "loss": 2.04,
      "step": 736
    },
    {
      "epoch": 0.14725274725274726,
      "grad_norm": 0.9868791103363037,
      "learning_rate": 0.00010379430158607975,
      "loss": 0.8817,
      "step": 737
    },
    {
      "epoch": 0.14745254745254746,
      "grad_norm": 2.976081132888794,
      "learning_rate": 0.000103583599992382,
      "loss": 2.0827,
      "step": 738
    },
    {
      "epoch": 0.14765234765234767,
      "grad_norm": 3.71689772605896,
      "learning_rate": 0.0001033728824675545,
      "loss": 1.9136,
      "step": 739
    },
    {
      "epoch": 0.14785214785214784,
      "grad_norm": 2.528423547744751,
      "learning_rate": 0.0001031621499483559,
      "loss": 1.7226,
      "step": 740
    },
    {
      "epoch": 0.14805194805194805,
      "grad_norm": 0.9580516219139099,
      "learning_rate": 0.00010295140337161146,
      "loss": 1.0615,
      "step": 741
    },
    {
      "epoch": 0.14825174825174825,
      "grad_norm": 3.30519700050354,
      "learning_rate": 0.00010274064367420897,
      "loss": 2.4017,
      "step": 742
    },
    {
      "epoch": 0.14845154845154845,
      "grad_norm": 3.217101573944092,
      "learning_rate": 0.00010252987179309459,
      "loss": 1.1689,
      "step": 743
    },
    {
      "epoch": 0.14865134865134866,
      "grad_norm": 3.9076790809631348,
      "learning_rate": 0.00010231908866526851,
      "loss": 1.6547,
      "step": 744
    },
    {
      "epoch": 0.14885114885114886,
      "grad_norm": 2.3255348205566406,
      "learning_rate": 0.00010210829522778111,
      "loss": 1.7629,
      "step": 745
    },
    {
      "epoch": 0.14905094905094904,
      "grad_norm": 2.0988636016845703,
      "learning_rate": 0.00010189749241772844,
      "loss": 1.6059,
      "step": 746
    },
    {
      "epoch": 0.14925074925074924,
      "grad_norm": 2.2119386196136475,
      "learning_rate": 0.00010168668117224825,
      "loss": 2.4522,
      "step": 747
    },
    {
      "epoch": 0.14945054945054945,
      "grad_norm": 2.3840341567993164,
      "learning_rate": 0.00010147586242851585,
      "loss": 2.0973,
      "step": 748
    },
    {
      "epoch": 0.14965034965034965,
      "grad_norm": 2.3250715732574463,
      "learning_rate": 0.00010126503712373982,
      "loss": 1.5185,
      "step": 749
    },
    {
      "epoch": 0.14985014985014986,
      "grad_norm": 3.1672186851501465,
      "learning_rate": 0.00010105420619515798,
      "loss": 2.3471,
      "step": 750
    },
    {
      "epoch": 0.14985014985014986,
      "eval_loss": 1.8448344469070435,
      "eval_runtime": 123.9861,
      "eval_samples_per_second": 33.996,
      "eval_steps_per_second": 8.501,
      "step": 750
    },
    {
      "epoch": 0.15004995004995006,
      "grad_norm": 1.7308681011199951,
      "learning_rate": 0.00010084337058003303,
      "loss": 2.242,
      "step": 751
    },
    {
      "epoch": 0.15024975024975026,
      "grad_norm": 1.2347564697265625,
      "learning_rate": 0.00010063253121564868,
      "loss": 1.9644,
      "step": 752
    },
    {
      "epoch": 0.15044955044955044,
      "grad_norm": 1.0844910144805908,
      "learning_rate": 0.00010042168903930514,
      "loss": 2.9764,
      "step": 753
    },
    {
      "epoch": 0.15064935064935064,
      "grad_norm": 2.45151686668396,
      "learning_rate": 0.00010021084498831522,
      "loss": 2.0986,
      "step": 754
    },
    {
      "epoch": 0.15084915084915085,
      "grad_norm": 2.3400442600250244,
      "learning_rate": 0.0001,
      "loss": 2.3916,
      "step": 755
    },
    {
      "epoch": 0.15104895104895105,
      "grad_norm": 2.8387551307678223,
      "learning_rate": 9.97891550116848e-05,
      "loss": 2.2081,
      "step": 756
    },
    {
      "epoch": 0.15124875124875126,
      "grad_norm": 3.754849672317505,
      "learning_rate": 9.957831096069488e-05,
      "loss": 1.7333,
      "step": 757
    },
    {
      "epoch": 0.15144855144855146,
      "grad_norm": 1.065299153327942,
      "learning_rate": 9.936746878435136e-05,
      "loss": 2.2731,
      "step": 758
    },
    {
      "epoch": 0.15164835164835164,
      "grad_norm": 0.9390467405319214,
      "learning_rate": 9.915662941996699e-05,
      "loss": 1.6865,
      "step": 759
    },
    {
      "epoch": 0.15184815184815184,
      "grad_norm": 4.233647346496582,
      "learning_rate": 9.894579380484204e-05,
      "loss": 2.0562,
      "step": 760
    },
    {
      "epoch": 0.15204795204795205,
      "grad_norm": 1.5831232070922852,
      "learning_rate": 9.873496287626019e-05,
      "loss": 1.5952,
      "step": 761
    },
    {
      "epoch": 0.15224775224775225,
      "grad_norm": 2.283733367919922,
      "learning_rate": 9.852413757148417e-05,
      "loss": 2.5759,
      "step": 762
    },
    {
      "epoch": 0.15244755244755245,
      "grad_norm": 1.3199553489685059,
      "learning_rate": 9.831331882775178e-05,
      "loss": 0.8512,
      "step": 763
    },
    {
      "epoch": 0.15264735264735266,
      "grad_norm": 1.984217882156372,
      "learning_rate": 9.81025075822716e-05,
      "loss": 1.4683,
      "step": 764
    },
    {
      "epoch": 0.15284715284715283,
      "grad_norm": 2.4857113361358643,
      "learning_rate": 9.789170477221891e-05,
      "loss": 2.1919,
      "step": 765
    },
    {
      "epoch": 0.15304695304695304,
      "grad_norm": 1.8946733474731445,
      "learning_rate": 9.76809113347315e-05,
      "loss": 1.9192,
      "step": 766
    },
    {
      "epoch": 0.15324675324675324,
      "grad_norm": 1.534278392791748,
      "learning_rate": 9.747012820690543e-05,
      "loss": 2.1031,
      "step": 767
    },
    {
      "epoch": 0.15344655344655345,
      "grad_norm": 2.929818630218506,
      "learning_rate": 9.725935632579104e-05,
      "loss": 1.9064,
      "step": 768
    },
    {
      "epoch": 0.15364635364635365,
      "grad_norm": 2.5807180404663086,
      "learning_rate": 9.704859662838855e-05,
      "loss": 1.7496,
      "step": 769
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 1.138712763786316,
      "learning_rate": 9.683785005164411e-05,
      "loss": 2.2042,
      "step": 770
    },
    {
      "epoch": 0.15404595404595406,
      "grad_norm": 3.88594388961792,
      "learning_rate": 9.662711753244551e-05,
      "loss": 1.9389,
      "step": 771
    },
    {
      "epoch": 0.15424575424575424,
      "grad_norm": 2.8658790588378906,
      "learning_rate": 9.641640000761802e-05,
      "loss": 1.5136,
      "step": 772
    },
    {
      "epoch": 0.15444555444555444,
      "grad_norm": 2.6935622692108154,
      "learning_rate": 9.620569841392029e-05,
      "loss": 2.2158,
      "step": 773
    },
    {
      "epoch": 0.15464535464535464,
      "grad_norm": 3.106640100479126,
      "learning_rate": 9.59950136880401e-05,
      "loss": 1.8374,
      "step": 774
    },
    {
      "epoch": 0.15484515484515485,
      "grad_norm": 1.9321558475494385,
      "learning_rate": 9.57843467665903e-05,
      "loss": 1.4841,
      "step": 775
    },
    {
      "epoch": 0.15504495504495505,
      "grad_norm": 7.649879455566406,
      "learning_rate": 9.557369858610453e-05,
      "loss": 2.5617,
      "step": 776
    },
    {
      "epoch": 0.15524475524475526,
      "grad_norm": 3.1792640686035156,
      "learning_rate": 9.53630700830332e-05,
      "loss": 1.7893,
      "step": 777
    },
    {
      "epoch": 0.15544455544455543,
      "grad_norm": 1.8094089031219482,
      "learning_rate": 9.51524621937391e-05,
      "loss": 1.6416,
      "step": 778
    },
    {
      "epoch": 0.15564435564435564,
      "grad_norm": 1.9329922199249268,
      "learning_rate": 9.494187585449358e-05,
      "loss": 2.0382,
      "step": 779
    },
    {
      "epoch": 0.15584415584415584,
      "grad_norm": 3.7754721641540527,
      "learning_rate": 9.473131200147205e-05,
      "loss": 1.7667,
      "step": 780
    },
    {
      "epoch": 0.15604395604395604,
      "grad_norm": 1.700691819190979,
      "learning_rate": 9.452077157074994e-05,
      "loss": 1.4737,
      "step": 781
    },
    {
      "epoch": 0.15624375624375625,
      "grad_norm": 1.6430381536483765,
      "learning_rate": 9.431025549829862e-05,
      "loss": 2.1351,
      "step": 782
    },
    {
      "epoch": 0.15644355644355645,
      "grad_norm": 1.5432300567626953,
      "learning_rate": 9.409976471998118e-05,
      "loss": 2.2282,
      "step": 783
    },
    {
      "epoch": 0.15664335664335666,
      "grad_norm": 2.441065549850464,
      "learning_rate": 9.388930017154819e-05,
      "loss": 2.2547,
      "step": 784
    },
    {
      "epoch": 0.15684315684315683,
      "grad_norm": 2.7178735733032227,
      "learning_rate": 9.367886278863366e-05,
      "loss": 2.4387,
      "step": 785
    },
    {
      "epoch": 0.15704295704295704,
      "grad_norm": 1.0630104541778564,
      "learning_rate": 9.346845350675088e-05,
      "loss": 2.078,
      "step": 786
    },
    {
      "epoch": 0.15724275724275724,
      "grad_norm": 0.9285669922828674,
      "learning_rate": 9.325807326128814e-05,
      "loss": 1.041,
      "step": 787
    },
    {
      "epoch": 0.15744255744255745,
      "grad_norm": 1.3061896562576294,
      "learning_rate": 9.304772298750463e-05,
      "loss": 1.4156,
      "step": 788
    },
    {
      "epoch": 0.15764235764235765,
      "grad_norm": 1.7859020233154297,
      "learning_rate": 9.283740362052642e-05,
      "loss": 2.1194,
      "step": 789
    },
    {
      "epoch": 0.15784215784215785,
      "grad_norm": 1.053816318511963,
      "learning_rate": 9.26271160953421e-05,
      "loss": 2.242,
      "step": 790
    },
    {
      "epoch": 0.15804195804195803,
      "grad_norm": 0.930632472038269,
      "learning_rate": 9.241686134679867e-05,
      "loss": 0.8732,
      "step": 791
    },
    {
      "epoch": 0.15824175824175823,
      "grad_norm": 2.3721718788146973,
      "learning_rate": 9.220664030959749e-05,
      "loss": 2.3831,
      "step": 792
    },
    {
      "epoch": 0.15844155844155844,
      "grad_norm": 1.4319783449172974,
      "learning_rate": 9.199645391828999e-05,
      "loss": 1.5982,
      "step": 793
    },
    {
      "epoch": 0.15864135864135864,
      "grad_norm": 3.923182964324951,
      "learning_rate": 9.178630310727365e-05,
      "loss": 1.5537,
      "step": 794
    },
    {
      "epoch": 0.15884115884115885,
      "grad_norm": 4.757667064666748,
      "learning_rate": 9.157618881078772e-05,
      "loss": 2.0941,
      "step": 795
    },
    {
      "epoch": 0.15904095904095905,
      "grad_norm": 3.2333312034606934,
      "learning_rate": 9.136611196290915e-05,
      "loss": 1.9799,
      "step": 796
    },
    {
      "epoch": 0.15924075924075923,
      "grad_norm": 2.746279001235962,
      "learning_rate": 9.115607349754834e-05,
      "loss": 1.6287,
      "step": 797
    },
    {
      "epoch": 0.15944055944055943,
      "grad_norm": 4.113913059234619,
      "learning_rate": 9.094607434844523e-05,
      "loss": 2.3051,
      "step": 798
    },
    {
      "epoch": 0.15964035964035964,
      "grad_norm": 2.265540599822998,
      "learning_rate": 9.07361154491648e-05,
      "loss": 1.7245,
      "step": 799
    },
    {
      "epoch": 0.15984015984015984,
      "grad_norm": 4.110149383544922,
      "learning_rate": 9.052619773309317e-05,
      "loss": 2.2289,
      "step": 800
    },
    {
      "epoch": 0.16003996003996004,
      "grad_norm": 3.2626516819000244,
      "learning_rate": 9.031632213343339e-05,
      "loss": 2.4915,
      "step": 801
    },
    {
      "epoch": 0.16023976023976025,
      "grad_norm": 1.6996110677719116,
      "learning_rate": 9.01064895832012e-05,
      "loss": 1.5196,
      "step": 802
    },
    {
      "epoch": 0.16043956043956045,
      "grad_norm": 1.7402448654174805,
      "learning_rate": 8.98967010152211e-05,
      "loss": 1.6111,
      "step": 803
    },
    {
      "epoch": 0.16063936063936063,
      "grad_norm": 3.9737424850463867,
      "learning_rate": 8.968695736212193e-05,
      "loss": 2.8393,
      "step": 804
    },
    {
      "epoch": 0.16083916083916083,
      "grad_norm": 2.0351781845092773,
      "learning_rate": 8.947725955633294e-05,
      "loss": 1.5922,
      "step": 805
    },
    {
      "epoch": 0.16103896103896104,
      "grad_norm": 1.0423582792282104,
      "learning_rate": 8.926760853007946e-05,
      "loss": 1.0445,
      "step": 806
    },
    {
      "epoch": 0.16123876123876124,
      "grad_norm": 2.211793899536133,
      "learning_rate": 8.905800521537905e-05,
      "loss": 1.6032,
      "step": 807
    },
    {
      "epoch": 0.16143856143856145,
      "grad_norm": 2.598252058029175,
      "learning_rate": 8.884845054403699e-05,
      "loss": 2.4403,
      "step": 808
    },
    {
      "epoch": 0.16163836163836165,
      "grad_norm": 2.621938943862915,
      "learning_rate": 8.863894544764236e-05,
      "loss": 1.4848,
      "step": 809
    },
    {
      "epoch": 0.16183816183816183,
      "grad_norm": 1.799440860748291,
      "learning_rate": 8.84294908575639e-05,
      "loss": 2.0561,
      "step": 810
    },
    {
      "epoch": 0.16203796203796203,
      "grad_norm": 2.9993185997009277,
      "learning_rate": 8.822008770494572e-05,
      "loss": 1.7681,
      "step": 811
    },
    {
      "epoch": 0.16223776223776223,
      "grad_norm": 2.2242822647094727,
      "learning_rate": 8.801073692070337e-05,
      "loss": 1.1821,
      "step": 812
    },
    {
      "epoch": 0.16243756243756244,
      "grad_norm": 1.5802218914031982,
      "learning_rate": 8.780143943551954e-05,
      "loss": 1.376,
      "step": 813
    },
    {
      "epoch": 0.16263736263736264,
      "grad_norm": 1.6632424592971802,
      "learning_rate": 8.759219617983999e-05,
      "loss": 1.8141,
      "step": 814
    },
    {
      "epoch": 0.16283716283716285,
      "grad_norm": 1.0167920589447021,
      "learning_rate": 8.738300808386935e-05,
      "loss": 0.9921,
      "step": 815
    },
    {
      "epoch": 0.16303696303696305,
      "grad_norm": 2.7525556087493896,
      "learning_rate": 8.717387607756713e-05,
      "loss": 2.3079,
      "step": 816
    },
    {
      "epoch": 0.16323676323676323,
      "grad_norm": 4.766961097717285,
      "learning_rate": 8.696480109064342e-05,
      "loss": 2.4908,
      "step": 817
    },
    {
      "epoch": 0.16343656343656343,
      "grad_norm": 2.824397325515747,
      "learning_rate": 8.675578405255485e-05,
      "loss": 1.9116,
      "step": 818
    },
    {
      "epoch": 0.16363636363636364,
      "grad_norm": 0.7965782880783081,
      "learning_rate": 8.654682589250038e-05,
      "loss": 1.1729,
      "step": 819
    },
    {
      "epoch": 0.16383616383616384,
      "grad_norm": 4.128576755523682,
      "learning_rate": 8.633792753941733e-05,
      "loss": 2.6802,
      "step": 820
    },
    {
      "epoch": 0.16403596403596404,
      "grad_norm": 2.4867055416107178,
      "learning_rate": 8.612908992197705e-05,
      "loss": 2.5261,
      "step": 821
    },
    {
      "epoch": 0.16423576423576425,
      "grad_norm": 1.550986647605896,
      "learning_rate": 8.592031396858093e-05,
      "loss": 1.9977,
      "step": 822
    },
    {
      "epoch": 0.16443556443556442,
      "grad_norm": 0.7057918906211853,
      "learning_rate": 8.571160060735624e-05,
      "loss": 1.7502,
      "step": 823
    },
    {
      "epoch": 0.16463536463536463,
      "grad_norm": 0.8527300953865051,
      "learning_rate": 8.550295076615188e-05,
      "loss": 1.2357,
      "step": 824
    },
    {
      "epoch": 0.16483516483516483,
      "grad_norm": 1.7797188758850098,
      "learning_rate": 8.529436537253458e-05,
      "loss": 1.9051,
      "step": 825
    },
    {
      "epoch": 0.16503496503496504,
      "grad_norm": 1.4820665121078491,
      "learning_rate": 8.508584535378439e-05,
      "loss": 2.2235,
      "step": 826
    },
    {
      "epoch": 0.16523476523476524,
      "grad_norm": 2.056929111480713,
      "learning_rate": 8.487739163689079e-05,
      "loss": 1.0167,
      "step": 827
    },
    {
      "epoch": 0.16543456543456544,
      "grad_norm": 1.8054448366165161,
      "learning_rate": 8.466900514854847e-05,
      "loss": 2.042,
      "step": 828
    },
    {
      "epoch": 0.16563436563436562,
      "grad_norm": 2.090838670730591,
      "learning_rate": 8.446068681515334e-05,
      "loss": 1.7507,
      "step": 829
    },
    {
      "epoch": 0.16583416583416583,
      "grad_norm": 1.4770698547363281,
      "learning_rate": 8.425243756279824e-05,
      "loss": 2.6725,
      "step": 830
    },
    {
      "epoch": 0.16603396603396603,
      "grad_norm": 2.1057870388031006,
      "learning_rate": 8.404425831726894e-05,
      "loss": 2.0415,
      "step": 831
    },
    {
      "epoch": 0.16623376623376623,
      "grad_norm": 1.5713011026382446,
      "learning_rate": 8.383615000404e-05,
      "loss": 2.3562,
      "step": 832
    },
    {
      "epoch": 0.16643356643356644,
      "grad_norm": 3.174250841140747,
      "learning_rate": 8.362811354827059e-05,
      "loss": 1.0954,
      "step": 833
    },
    {
      "epoch": 0.16663336663336664,
      "grad_norm": 2.3101999759674072,
      "learning_rate": 8.342014987480047e-05,
      "loss": 2.1411,
      "step": 834
    },
    {
      "epoch": 0.16683316683316685,
      "grad_norm": 3.392484664916992,
      "learning_rate": 8.321225990814591e-05,
      "loss": 2.0549,
      "step": 835
    },
    {
      "epoch": 0.16703296703296702,
      "grad_norm": 1.68133544921875,
      "learning_rate": 8.300444457249543e-05,
      "loss": 1.6829,
      "step": 836
    },
    {
      "epoch": 0.16723276723276723,
      "grad_norm": 1.998247504234314,
      "learning_rate": 8.279670479170573e-05,
      "loss": 1.5719,
      "step": 837
    },
    {
      "epoch": 0.16743256743256743,
      "grad_norm": 2.060650110244751,
      "learning_rate": 8.258904148929775e-05,
      "loss": 1.6291,
      "step": 838
    },
    {
      "epoch": 0.16763236763236763,
      "grad_norm": 1.1461197137832642,
      "learning_rate": 8.238145558845235e-05,
      "loss": 1.8434,
      "step": 839
    },
    {
      "epoch": 0.16783216783216784,
      "grad_norm": 3.0855860710144043,
      "learning_rate": 8.217394801200631e-05,
      "loss": 1.5676,
      "step": 840
    },
    {
      "epoch": 0.16803196803196804,
      "grad_norm": 2.1299405097961426,
      "learning_rate": 8.196651968244826e-05,
      "loss": 2.3561,
      "step": 841
    },
    {
      "epoch": 0.16823176823176822,
      "grad_norm": 0.9916551113128662,
      "learning_rate": 8.175917152191447e-05,
      "loss": 2.5623,
      "step": 842
    },
    {
      "epoch": 0.16843156843156842,
      "grad_norm": 1.9672701358795166,
      "learning_rate": 8.15519044521848e-05,
      "loss": 1.799,
      "step": 843
    },
    {
      "epoch": 0.16863136863136863,
      "grad_norm": 1.842760443687439,
      "learning_rate": 8.134471939467874e-05,
      "loss": 1.9993,
      "step": 844
    },
    {
      "epoch": 0.16883116883116883,
      "grad_norm": 2.120115041732788,
      "learning_rate": 8.113761727045105e-05,
      "loss": 2.3167,
      "step": 845
    },
    {
      "epoch": 0.16903096903096904,
      "grad_norm": 2.6375300884246826,
      "learning_rate": 8.093059900018792e-05,
      "loss": 2.199,
      "step": 846
    },
    {
      "epoch": 0.16923076923076924,
      "grad_norm": 2.5650014877319336,
      "learning_rate": 8.072366550420266e-05,
      "loss": 1.1527,
      "step": 847
    },
    {
      "epoch": 0.16943056943056944,
      "grad_norm": 3.7919082641601562,
      "learning_rate": 8.051681770243175e-05,
      "loss": 1.7206,
      "step": 848
    },
    {
      "epoch": 0.16963036963036962,
      "grad_norm": 1.9254869222640991,
      "learning_rate": 8.031005651443073e-05,
      "loss": 2.3636,
      "step": 849
    },
    {
      "epoch": 0.16983016983016982,
      "grad_norm": 1.154807686805725,
      "learning_rate": 8.010338285937006e-05,
      "loss": 0.8404,
      "step": 850
    },
    {
      "epoch": 0.17002997002997003,
      "grad_norm": 2.823699712753296,
      "learning_rate": 7.989679765603108e-05,
      "loss": 2.0125,
      "step": 851
    },
    {
      "epoch": 0.17022977022977023,
      "grad_norm": 4.296392440795898,
      "learning_rate": 7.969030182280192e-05,
      "loss": 1.7983,
      "step": 852
    },
    {
      "epoch": 0.17042957042957044,
      "grad_norm": 2.7911388874053955,
      "learning_rate": 7.948389627767343e-05,
      "loss": 1.9678,
      "step": 853
    },
    {
      "epoch": 0.17062937062937064,
      "grad_norm": 1.058059811592102,
      "learning_rate": 7.927758193823501e-05,
      "loss": 0.8794,
      "step": 854
    },
    {
      "epoch": 0.17082917082917082,
      "grad_norm": 1.316698670387268,
      "learning_rate": 7.907135972167069e-05,
      "loss": 0.9544,
      "step": 855
    },
    {
      "epoch": 0.17102897102897102,
      "grad_norm": 2.0101001262664795,
      "learning_rate": 7.88652305447549e-05,
      "loss": 1.5735,
      "step": 856
    },
    {
      "epoch": 0.17122877122877123,
      "grad_norm": 1.0396119356155396,
      "learning_rate": 7.865919532384844e-05,
      "loss": 1.1007,
      "step": 857
    },
    {
      "epoch": 0.17142857142857143,
      "grad_norm": 3.773207902908325,
      "learning_rate": 7.845325497489449e-05,
      "loss": 2.0574,
      "step": 858
    },
    {
      "epoch": 0.17162837162837163,
      "grad_norm": 2.882340908050537,
      "learning_rate": 7.82474104134144e-05,
      "loss": 1.1172,
      "step": 859
    },
    {
      "epoch": 0.17182817182817184,
      "grad_norm": 2.4804646968841553,
      "learning_rate": 7.804166255450373e-05,
      "loss": 1.9873,
      "step": 860
    },
    {
      "epoch": 0.17202797202797201,
      "grad_norm": 2.495528221130371,
      "learning_rate": 7.783601231282812e-05,
      "loss": 2.027,
      "step": 861
    },
    {
      "epoch": 0.17222777222777222,
      "grad_norm": 3.018609046936035,
      "learning_rate": 7.763046060261932e-05,
      "loss": 2.5494,
      "step": 862
    },
    {
      "epoch": 0.17242757242757242,
      "grad_norm": 1.4497827291488647,
      "learning_rate": 7.742500833767094e-05,
      "loss": 1.7581,
      "step": 863
    },
    {
      "epoch": 0.17262737262737263,
      "grad_norm": 3.0562055110931396,
      "learning_rate": 7.721965643133458e-05,
      "loss": 2.0972,
      "step": 864
    },
    {
      "epoch": 0.17282717282717283,
      "grad_norm": 1.315582513809204,
      "learning_rate": 7.701440579651566e-05,
      "loss": 1.6615,
      "step": 865
    },
    {
      "epoch": 0.17302697302697304,
      "grad_norm": 2.1207053661346436,
      "learning_rate": 7.680925734566937e-05,
      "loss": 1.7759,
      "step": 866
    },
    {
      "epoch": 0.17322677322677324,
      "grad_norm": 1.8030495643615723,
      "learning_rate": 7.660421199079669e-05,
      "loss": 2.4271,
      "step": 867
    },
    {
      "epoch": 0.17342657342657342,
      "grad_norm": 1.5998008251190186,
      "learning_rate": 7.639927064344022e-05,
      "loss": 1.2568,
      "step": 868
    },
    {
      "epoch": 0.17362637362637362,
      "grad_norm": 2.0377871990203857,
      "learning_rate": 7.619443421468021e-05,
      "loss": 1.6366,
      "step": 869
    },
    {
      "epoch": 0.17382617382617382,
      "grad_norm": 1.1051067113876343,
      "learning_rate": 7.598970361513051e-05,
      "loss": 2.0432,
      "step": 870
    },
    {
      "epoch": 0.17402597402597403,
      "grad_norm": 2.2950568199157715,
      "learning_rate": 7.578507975493448e-05,
      "loss": 2.4907,
      "step": 871
    },
    {
      "epoch": 0.17422577422577423,
      "grad_norm": 2.4267780780792236,
      "learning_rate": 7.558056354376098e-05,
      "loss": 1.7446,
      "step": 872
    },
    {
      "epoch": 0.17442557442557444,
      "grad_norm": 4.489258766174316,
      "learning_rate": 7.537615589080027e-05,
      "loss": 1.7765,
      "step": 873
    },
    {
      "epoch": 0.1746253746253746,
      "grad_norm": 1.4550879001617432,
      "learning_rate": 7.517185770476006e-05,
      "loss": 2.2539,
      "step": 874
    },
    {
      "epoch": 0.17482517482517482,
      "grad_norm": 3.1140196323394775,
      "learning_rate": 7.496766989386136e-05,
      "loss": 1.9919,
      "step": 875
    },
    {
      "epoch": 0.17502497502497502,
      "grad_norm": 1.915116548538208,
      "learning_rate": 7.476359336583454e-05,
      "loss": 2.2443,
      "step": 876
    },
    {
      "epoch": 0.17522477522477523,
      "grad_norm": 3.4786338806152344,
      "learning_rate": 7.455962902791522e-05,
      "loss": 1.7265,
      "step": 877
    },
    {
      "epoch": 0.17542457542457543,
      "grad_norm": 2.0269017219543457,
      "learning_rate": 7.435577778684033e-05,
      "loss": 1.9271,
      "step": 878
    },
    {
      "epoch": 0.17562437562437563,
      "grad_norm": 1.6195753812789917,
      "learning_rate": 7.415204054884399e-05,
      "loss": 1.8042,
      "step": 879
    },
    {
      "epoch": 0.17582417582417584,
      "grad_norm": 1.4486581087112427,
      "learning_rate": 7.394841821965345e-05,
      "loss": 2.41,
      "step": 880
    },
    {
      "epoch": 0.17602397602397601,
      "grad_norm": 3.3392367362976074,
      "learning_rate": 7.374491170448525e-05,
      "loss": 0.6347,
      "step": 881
    },
    {
      "epoch": 0.17622377622377622,
      "grad_norm": 1.3322975635528564,
      "learning_rate": 7.3541521908041e-05,
      "loss": 1.0243,
      "step": 882
    },
    {
      "epoch": 0.17642357642357642,
      "grad_norm": 2.3097293376922607,
      "learning_rate": 7.33382497345034e-05,
      "loss": 1.6005,
      "step": 883
    },
    {
      "epoch": 0.17662337662337663,
      "grad_norm": 2.557023763656616,
      "learning_rate": 7.313509608753231e-05,
      "loss": 2.3908,
      "step": 884
    },
    {
      "epoch": 0.17682317682317683,
      "grad_norm": 2.509622812271118,
      "learning_rate": 7.293206187026061e-05,
      "loss": 2.185,
      "step": 885
    },
    {
      "epoch": 0.17702297702297703,
      "grad_norm": 1.0252410173416138,
      "learning_rate": 7.27291479852903e-05,
      "loss": 1.6726,
      "step": 886
    },
    {
      "epoch": 0.1772227772227772,
      "grad_norm": 2.901326894760132,
      "learning_rate": 7.252635533468843e-05,
      "loss": 2.6824,
      "step": 887
    },
    {
      "epoch": 0.17742257742257742,
      "grad_norm": 1.665132999420166,
      "learning_rate": 7.232368481998309e-05,
      "loss": 1.5114,
      "step": 888
    },
    {
      "epoch": 0.17762237762237762,
      "grad_norm": 3.7579219341278076,
      "learning_rate": 7.212113734215932e-05,
      "loss": 1.5069,
      "step": 889
    },
    {
      "epoch": 0.17782217782217782,
      "grad_norm": 1.725112795829773,
      "learning_rate": 7.191871380165538e-05,
      "loss": 2.1623,
      "step": 890
    },
    {
      "epoch": 0.17802197802197803,
      "grad_norm": 4.477016925811768,
      "learning_rate": 7.17164150983584e-05,
      "loss": 2.2632,
      "step": 891
    },
    {
      "epoch": 0.17822177822177823,
      "grad_norm": 3.0874006748199463,
      "learning_rate": 7.151424213160061e-05,
      "loss": 1.5272,
      "step": 892
    },
    {
      "epoch": 0.1784215784215784,
      "grad_norm": 1.9160354137420654,
      "learning_rate": 7.131219580015521e-05,
      "loss": 0.9831,
      "step": 893
    },
    {
      "epoch": 0.1786213786213786,
      "grad_norm": 3.2237794399261475,
      "learning_rate": 7.11102770022325e-05,
      "loss": 1.7172,
      "step": 894
    },
    {
      "epoch": 0.17882117882117882,
      "grad_norm": 1.0210037231445312,
      "learning_rate": 7.090848663547574e-05,
      "loss": 1.6034,
      "step": 895
    },
    {
      "epoch": 0.17902097902097902,
      "grad_norm": 1.9740357398986816,
      "learning_rate": 7.070682559695736e-05,
      "loss": 2.038,
      "step": 896
    },
    {
      "epoch": 0.17922077922077922,
      "grad_norm": 1.5184401273727417,
      "learning_rate": 7.050529478317476e-05,
      "loss": 1.9848,
      "step": 897
    },
    {
      "epoch": 0.17942057942057943,
      "grad_norm": 3.8241817951202393,
      "learning_rate": 7.03038950900464e-05,
      "loss": 1.6818,
      "step": 898
    },
    {
      "epoch": 0.17962037962037963,
      "grad_norm": 1.622344970703125,
      "learning_rate": 7.010262741290798e-05,
      "loss": 1.5116,
      "step": 899
    },
    {
      "epoch": 0.1798201798201798,
      "grad_norm": 4.112119197845459,
      "learning_rate": 6.990149264650814e-05,
      "loss": 1.8638,
      "step": 900
    },
    {
      "epoch": 0.18001998001998,
      "grad_norm": 2.0355026721954346,
      "learning_rate": 6.970049168500474e-05,
      "loss": 1.8184,
      "step": 901
    },
    {
      "epoch": 0.18021978021978022,
      "grad_norm": 1.9191902875900269,
      "learning_rate": 6.94996254219608e-05,
      "loss": 2.1724,
      "step": 902
    },
    {
      "epoch": 0.18041958041958042,
      "grad_norm": 1.8613873720169067,
      "learning_rate": 6.929889475034048e-05,
      "loss": 1.1559,
      "step": 903
    },
    {
      "epoch": 0.18061938061938063,
      "grad_norm": 5.070361614227295,
      "learning_rate": 6.909830056250527e-05,
      "loss": 1.7484,
      "step": 904
    },
    {
      "epoch": 0.18081918081918083,
      "grad_norm": 1.503034234046936,
      "learning_rate": 6.889784375020978e-05,
      "loss": 1.6054,
      "step": 905
    },
    {
      "epoch": 0.181018981018981,
      "grad_norm": 2.453569173812866,
      "learning_rate": 6.869752520459803e-05,
      "loss": 1.7363,
      "step": 906
    },
    {
      "epoch": 0.1812187812187812,
      "grad_norm": 2.095191478729248,
      "learning_rate": 6.849734581619918e-05,
      "loss": 0.8846,
      "step": 907
    },
    {
      "epoch": 0.18141858141858141,
      "grad_norm": 0.9337109327316284,
      "learning_rate": 6.829730647492404e-05,
      "loss": 1.4785,
      "step": 908
    },
    {
      "epoch": 0.18161838161838162,
      "grad_norm": 2.3453924655914307,
      "learning_rate": 6.80974080700606e-05,
      "loss": 1.3436,
      "step": 909
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 1.6251871585845947,
      "learning_rate": 6.789765149027039e-05,
      "loss": 1.7181,
      "step": 910
    },
    {
      "epoch": 0.18201798201798203,
      "grad_norm": 1.8028430938720703,
      "learning_rate": 6.769803762358443e-05,
      "loss": 2.2534,
      "step": 911
    },
    {
      "epoch": 0.18221778221778223,
      "grad_norm": 2.8331806659698486,
      "learning_rate": 6.749856735739928e-05,
      "loss": 1.8221,
      "step": 912
    },
    {
      "epoch": 0.1824175824175824,
      "grad_norm": 2.668593168258667,
      "learning_rate": 6.729924157847323e-05,
      "loss": 2.5403,
      "step": 913
    },
    {
      "epoch": 0.1826173826173826,
      "grad_norm": 1.7772200107574463,
      "learning_rate": 6.710006117292209e-05,
      "loss": 1.8535,
      "step": 914
    },
    {
      "epoch": 0.18281718281718282,
      "grad_norm": 1.2930259704589844,
      "learning_rate": 6.690102702621548e-05,
      "loss": 1.0175,
      "step": 915
    },
    {
      "epoch": 0.18301698301698302,
      "grad_norm": 2.2263996601104736,
      "learning_rate": 6.670214002317278e-05,
      "loss": 2.2472,
      "step": 916
    },
    {
      "epoch": 0.18321678321678322,
      "grad_norm": 2.658219814300537,
      "learning_rate": 6.650340104795932e-05,
      "loss": 2.1643,
      "step": 917
    },
    {
      "epoch": 0.18341658341658343,
      "grad_norm": 3.1427218914031982,
      "learning_rate": 6.630481098408228e-05,
      "loss": 1.8815,
      "step": 918
    },
    {
      "epoch": 0.1836163836163836,
      "grad_norm": 0.7484788298606873,
      "learning_rate": 6.610637071438686e-05,
      "loss": 1.4903,
      "step": 919
    },
    {
      "epoch": 0.1838161838161838,
      "grad_norm": 0.882391631603241,
      "learning_rate": 6.590808112105232e-05,
      "loss": 0.9156,
      "step": 920
    },
    {
      "epoch": 0.184015984015984,
      "grad_norm": 2.8732683658599854,
      "learning_rate": 6.570994308558812e-05,
      "loss": 1.2664,
      "step": 921
    },
    {
      "epoch": 0.18421578421578422,
      "grad_norm": 2.232675075531006,
      "learning_rate": 6.551195748882997e-05,
      "loss": 3.305,
      "step": 922
    },
    {
      "epoch": 0.18441558441558442,
      "grad_norm": 2.037015199661255,
      "learning_rate": 6.531412521093586e-05,
      "loss": 1.5656,
      "step": 923
    },
    {
      "epoch": 0.18461538461538463,
      "grad_norm": 2.632323980331421,
      "learning_rate": 6.51164471313822e-05,
      "loss": 1.4403,
      "step": 924
    },
    {
      "epoch": 0.1848151848151848,
      "grad_norm": 3.4172160625457764,
      "learning_rate": 6.491892412895995e-05,
      "loss": 1.9409,
      "step": 925
    },
    {
      "epoch": 0.185014985014985,
      "grad_norm": 1.2638347148895264,
      "learning_rate": 6.472155708177052e-05,
      "loss": 2.2422,
      "step": 926
    },
    {
      "epoch": 0.1852147852147852,
      "grad_norm": 3.2859091758728027,
      "learning_rate": 6.452434686722224e-05,
      "loss": 1.6946,
      "step": 927
    },
    {
      "epoch": 0.18541458541458541,
      "grad_norm": 2.0068798065185547,
      "learning_rate": 6.432729436202604e-05,
      "loss": 1.9101,
      "step": 928
    },
    {
      "epoch": 0.18561438561438562,
      "grad_norm": 2.224031448364258,
      "learning_rate": 6.41304004421918e-05,
      "loss": 2.9881,
      "step": 929
    },
    {
      "epoch": 0.18581418581418582,
      "grad_norm": 2.896193742752075,
      "learning_rate": 6.393366598302446e-05,
      "loss": 1.9255,
      "step": 930
    },
    {
      "epoch": 0.18601398601398603,
      "grad_norm": 2.338862657546997,
      "learning_rate": 6.373709185911998e-05,
      "loss": 1.7118,
      "step": 931
    },
    {
      "epoch": 0.1862137862137862,
      "grad_norm": 1.872319221496582,
      "learning_rate": 6.354067894436155e-05,
      "loss": 1.8273,
      "step": 932
    },
    {
      "epoch": 0.1864135864135864,
      "grad_norm": 1.1004530191421509,
      "learning_rate": 6.334442811191576e-05,
      "loss": 0.4674,
      "step": 933
    },
    {
      "epoch": 0.1866133866133866,
      "grad_norm": 1.543302297592163,
      "learning_rate": 6.314834023422858e-05,
      "loss": 2.0194,
      "step": 934
    },
    {
      "epoch": 0.18681318681318682,
      "grad_norm": 0.8261705636978149,
      "learning_rate": 6.295241618302156e-05,
      "loss": 2.0305,
      "step": 935
    },
    {
      "epoch": 0.18701298701298702,
      "grad_norm": 1.89741849899292,
      "learning_rate": 6.275665682928803e-05,
      "loss": 1.6427,
      "step": 936
    },
    {
      "epoch": 0.18721278721278722,
      "grad_norm": 3.9334094524383545,
      "learning_rate": 6.256106304328905e-05,
      "loss": 1.8507,
      "step": 937
    },
    {
      "epoch": 0.1874125874125874,
      "grad_norm": 1.5562832355499268,
      "learning_rate": 6.23656356945497e-05,
      "loss": 2.1231,
      "step": 938
    },
    {
      "epoch": 0.1876123876123876,
      "grad_norm": 1.2202882766723633,
      "learning_rate": 6.21703756518551e-05,
      "loss": 1.9889,
      "step": 939
    },
    {
      "epoch": 0.1878121878121878,
      "grad_norm": 1.6650581359863281,
      "learning_rate": 6.197528378324665e-05,
      "loss": 2.7896,
      "step": 940
    },
    {
      "epoch": 0.188011988011988,
      "grad_norm": 1.0490233898162842,
      "learning_rate": 6.17803609560181e-05,
      "loss": 2.0745,
      "step": 941
    },
    {
      "epoch": 0.18821178821178822,
      "grad_norm": 1.733917474746704,
      "learning_rate": 6.158560803671168e-05,
      "loss": 1.947,
      "step": 942
    },
    {
      "epoch": 0.18841158841158842,
      "grad_norm": 1.6134380102157593,
      "learning_rate": 6.139102589111435e-05,
      "loss": 2.1088,
      "step": 943
    },
    {
      "epoch": 0.18861138861138863,
      "grad_norm": 1.0397019386291504,
      "learning_rate": 6.119661538425381e-05,
      "loss": 1.3949,
      "step": 944
    },
    {
      "epoch": 0.1888111888111888,
      "grad_norm": 2.5493083000183105,
      "learning_rate": 6.100237738039484e-05,
      "loss": 1.9434,
      "step": 945
    },
    {
      "epoch": 0.189010989010989,
      "grad_norm": 2.427701234817505,
      "learning_rate": 6.0808312743035236e-05,
      "loss": 2.0539,
      "step": 946
    },
    {
      "epoch": 0.1892107892107892,
      "grad_norm": 2.8848845958709717,
      "learning_rate": 6.061442233490211e-05,
      "loss": 2.0288,
      "step": 947
    },
    {
      "epoch": 0.18941058941058941,
      "grad_norm": 0.766930103302002,
      "learning_rate": 6.042070701794806e-05,
      "loss": 0.5258,
      "step": 948
    },
    {
      "epoch": 0.18961038961038962,
      "grad_norm": 2.0311174392700195,
      "learning_rate": 6.0227167653347305e-05,
      "loss": 2.3811,
      "step": 949
    },
    {
      "epoch": 0.18981018981018982,
      "grad_norm": 1.199125051498413,
      "learning_rate": 6.0033805101491794e-05,
      "loss": 1.1648,
      "step": 950
    },
    {
      "epoch": 0.19000999000999,
      "grad_norm": 1.56325364112854,
      "learning_rate": 5.98406202219875e-05,
      "loss": 1.9673,
      "step": 951
    },
    {
      "epoch": 0.1902097902097902,
      "grad_norm": 2.7365026473999023,
      "learning_rate": 5.964761387365052e-05,
      "loss": 2.0553,
      "step": 952
    },
    {
      "epoch": 0.1904095904095904,
      "grad_norm": 2.7366909980773926,
      "learning_rate": 5.9454786914503255e-05,
      "loss": 1.9732,
      "step": 953
    },
    {
      "epoch": 0.1906093906093906,
      "grad_norm": 2.500722646713257,
      "learning_rate": 5.926214020177074e-05,
      "loss": 1.7507,
      "step": 954
    },
    {
      "epoch": 0.19080919080919082,
      "grad_norm": 2.0231411457061768,
      "learning_rate": 5.9069674591876534e-05,
      "loss": 1.5741,
      "step": 955
    },
    {
      "epoch": 0.19100899100899102,
      "grad_norm": 2.577286720275879,
      "learning_rate": 5.887739094043923e-05,
      "loss": 2.126,
      "step": 956
    },
    {
      "epoch": 0.1912087912087912,
      "grad_norm": 1.1040592193603516,
      "learning_rate": 5.868529010226845e-05,
      "loss": 0.4903,
      "step": 957
    },
    {
      "epoch": 0.1914085914085914,
      "grad_norm": 1.2436195611953735,
      "learning_rate": 5.849337293136112e-05,
      "loss": 0.7412,
      "step": 958
    },
    {
      "epoch": 0.1916083916083916,
      "grad_norm": 1.7570528984069824,
      "learning_rate": 5.830164028089766e-05,
      "loss": 1.3942,
      "step": 959
    },
    {
      "epoch": 0.1918081918081918,
      "grad_norm": 1.1048418283462524,
      "learning_rate": 5.811009300323818e-05,
      "loss": 0.9357,
      "step": 960
    },
    {
      "epoch": 0.192007992007992,
      "grad_norm": 1.1667903661727905,
      "learning_rate": 5.791873194991872e-05,
      "loss": 2.8031,
      "step": 961
    },
    {
      "epoch": 0.19220779220779222,
      "grad_norm": 2.8910741806030273,
      "learning_rate": 5.7727557971647427e-05,
      "loss": 2.815,
      "step": 962
    },
    {
      "epoch": 0.19240759240759242,
      "grad_norm": 1.4528439044952393,
      "learning_rate": 5.7536571918300864e-05,
      "loss": 1.7111,
      "step": 963
    },
    {
      "epoch": 0.1926073926073926,
      "grad_norm": 1.8478124141693115,
      "learning_rate": 5.734577463892008e-05,
      "loss": 2.0092,
      "step": 964
    },
    {
      "epoch": 0.1928071928071928,
      "grad_norm": 2.86433482170105,
      "learning_rate": 5.7155166981706956e-05,
      "loss": 2.1948,
      "step": 965
    },
    {
      "epoch": 0.193006993006993,
      "grad_norm": 1.377597689628601,
      "learning_rate": 5.6964749794020354e-05,
      "loss": 2.3825,
      "step": 966
    },
    {
      "epoch": 0.1932067932067932,
      "grad_norm": 2.2262918949127197,
      "learning_rate": 5.6774523922372394e-05,
      "loss": 2.4155,
      "step": 967
    },
    {
      "epoch": 0.1934065934065934,
      "grad_norm": 2.356685161590576,
      "learning_rate": 5.6584490212424804e-05,
      "loss": 1.9483,
      "step": 968
    },
    {
      "epoch": 0.19360639360639362,
      "grad_norm": 1.5135438442230225,
      "learning_rate": 5.639464950898491e-05,
      "loss": 2.3864,
      "step": 969
    },
    {
      "epoch": 0.1938061938061938,
      "grad_norm": 2.031242609024048,
      "learning_rate": 5.620500265600206e-05,
      "loss": 1.9185,
      "step": 970
    },
    {
      "epoch": 0.194005994005994,
      "grad_norm": 2.5787153244018555,
      "learning_rate": 5.601555049656382e-05,
      "loss": 2.2412,
      "step": 971
    },
    {
      "epoch": 0.1942057942057942,
      "grad_norm": 3.1233463287353516,
      "learning_rate": 5.58262938728922e-05,
      "loss": 1.833,
      "step": 972
    },
    {
      "epoch": 0.1944055944055944,
      "grad_norm": 4.2123942375183105,
      "learning_rate": 5.563723362634008e-05,
      "loss": 1.4143,
      "step": 973
    },
    {
      "epoch": 0.1946053946053946,
      "grad_norm": 2.417607545852661,
      "learning_rate": 5.544837059738719e-05,
      "loss": 1.8439,
      "step": 974
    },
    {
      "epoch": 0.19480519480519481,
      "grad_norm": 4.222956657409668,
      "learning_rate": 5.525970562563656e-05,
      "loss": 2.5378,
      "step": 975
    },
    {
      "epoch": 0.19500499500499502,
      "grad_norm": 0.9048491716384888,
      "learning_rate": 5.507123954981073e-05,
      "loss": 2.1016,
      "step": 976
    },
    {
      "epoch": 0.1952047952047952,
      "grad_norm": 3.101668357849121,
      "learning_rate": 5.488297320774807e-05,
      "loss": 2.2974,
      "step": 977
    },
    {
      "epoch": 0.1954045954045954,
      "grad_norm": 3.9237775802612305,
      "learning_rate": 5.4694907436399e-05,
      "loss": 1.938,
      "step": 978
    },
    {
      "epoch": 0.1956043956043956,
      "grad_norm": 2.4713480472564697,
      "learning_rate": 5.4507043071822284e-05,
      "loss": 2.1542,
      "step": 979
    },
    {
      "epoch": 0.1958041958041958,
      "grad_norm": 1.4540680646896362,
      "learning_rate": 5.431938094918132e-05,
      "loss": 1.5431,
      "step": 980
    },
    {
      "epoch": 0.196003996003996,
      "grad_norm": 1.0258973836898804,
      "learning_rate": 5.41319219027404e-05,
      "loss": 1.0439,
      "step": 981
    },
    {
      "epoch": 0.19620379620379622,
      "grad_norm": 1.9177066087722778,
      "learning_rate": 5.394466676586114e-05,
      "loss": 2.378,
      "step": 982
    },
    {
      "epoch": 0.1964035964035964,
      "grad_norm": 2.840031147003174,
      "learning_rate": 5.375761637099854e-05,
      "loss": 2.1754,
      "step": 983
    },
    {
      "epoch": 0.1966033966033966,
      "grad_norm": 0.9176260232925415,
      "learning_rate": 5.357077154969742e-05,
      "loss": 1.0646,
      "step": 984
    },
    {
      "epoch": 0.1968031968031968,
      "grad_norm": 2.0221989154815674,
      "learning_rate": 5.3384133132588784e-05,
      "loss": 1.8957,
      "step": 985
    },
    {
      "epoch": 0.197002997002997,
      "grad_norm": 3.0448691844940186,
      "learning_rate": 5.3197701949386e-05,
      "loss": 2.2055,
      "step": 986
    },
    {
      "epoch": 0.1972027972027972,
      "grad_norm": 1.1111698150634766,
      "learning_rate": 5.301147882888116e-05,
      "loss": 0.9532,
      "step": 987
    },
    {
      "epoch": 0.1974025974025974,
      "grad_norm": 3.0166213512420654,
      "learning_rate": 5.28254645989414e-05,
      "loss": 1.82,
      "step": 988
    },
    {
      "epoch": 0.1976023976023976,
      "grad_norm": 1.057173728942871,
      "learning_rate": 5.2639660086505226e-05,
      "loss": 1.1411,
      "step": 989
    },
    {
      "epoch": 0.1978021978021978,
      "grad_norm": 3.8735439777374268,
      "learning_rate": 5.2454066117578815e-05,
      "loss": 1.4259,
      "step": 990
    },
    {
      "epoch": 0.198001998001998,
      "grad_norm": 2.4388513565063477,
      "learning_rate": 5.226868351723244e-05,
      "loss": 1.1446,
      "step": 991
    },
    {
      "epoch": 0.1982017982017982,
      "grad_norm": 1.200182318687439,
      "learning_rate": 5.2083513109596616e-05,
      "loss": 2.8787,
      "step": 992
    },
    {
      "epoch": 0.1984015984015984,
      "grad_norm": 2.173332929611206,
      "learning_rate": 5.189855571785859e-05,
      "loss": 1.2858,
      "step": 993
    },
    {
      "epoch": 0.1986013986013986,
      "grad_norm": 2.6679093837738037,
      "learning_rate": 5.171381216425863e-05,
      "loss": 1.9196,
      "step": 994
    },
    {
      "epoch": 0.19880119880119881,
      "grad_norm": 2.906251907348633,
      "learning_rate": 5.152928327008635e-05,
      "loss": 1.7141,
      "step": 995
    },
    {
      "epoch": 0.199000999000999,
      "grad_norm": 2.181311845779419,
      "learning_rate": 5.134496985567714e-05,
      "loss": 2.225,
      "step": 996
    },
    {
      "epoch": 0.1992007992007992,
      "grad_norm": 3.3263535499572754,
      "learning_rate": 5.116087274040837e-05,
      "loss": 1.1601,
      "step": 997
    },
    {
      "epoch": 0.1994005994005994,
      "grad_norm": 1.6543798446655273,
      "learning_rate": 5.0976992742695925e-05,
      "loss": 1.2074,
      "step": 998
    },
    {
      "epoch": 0.1996003996003996,
      "grad_norm": 1.208321213722229,
      "learning_rate": 5.07933306799904e-05,
      "loss": 2.2189,
      "step": 999
    },
    {
      "epoch": 0.1998001998001998,
      "grad_norm": 3.908557891845703,
      "learning_rate": 5.060988736877366e-05,
      "loss": 2.0459,
      "step": 1000
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9158048629760742,
      "learning_rate": 5.042666362455498e-05,
      "loss": 2.284,
      "step": 1001
    },
    {
      "epoch": 0.2001998001998002,
      "grad_norm": 0.9646271467208862,
      "learning_rate": 5.024366026186755e-05,
      "loss": 2.1987,
      "step": 1002
    },
    {
      "epoch": 0.2003996003996004,
      "grad_norm": 1.368666410446167,
      "learning_rate": 5.006087809426496e-05,
      "loss": 1.004,
      "step": 1003
    },
    {
      "epoch": 0.2005994005994006,
      "grad_norm": 3.646068811416626,
      "learning_rate": 4.987831793431731e-05,
      "loss": 2.2703,
      "step": 1004
    },
    {
      "epoch": 0.2007992007992008,
      "grad_norm": 1.7744122743606567,
      "learning_rate": 4.9695980593607817e-05,
      "loss": 2.2441,
      "step": 1005
    },
    {
      "epoch": 0.200999000999001,
      "grad_norm": 2.6374502182006836,
      "learning_rate": 4.9513866882729146e-05,
      "loss": 1.8811,
      "step": 1006
    },
    {
      "epoch": 0.2011988011988012,
      "grad_norm": 2.3681869506835938,
      "learning_rate": 4.9331977611279777e-05,
      "loss": 2.3007,
      "step": 1007
    },
    {
      "epoch": 0.2013986013986014,
      "grad_norm": 4.361027240753174,
      "learning_rate": 4.9150313587860433e-05,
      "loss": 1.3333,
      "step": 1008
    },
    {
      "epoch": 0.2015984015984016,
      "grad_norm": 2.489013910293579,
      "learning_rate": 4.896887562007054e-05,
      "loss": 2.2896,
      "step": 1009
    },
    {
      "epoch": 0.2017982017982018,
      "grad_norm": 2.1032354831695557,
      "learning_rate": 4.8787664514504504e-05,
      "loss": 1.371,
      "step": 1010
    },
    {
      "epoch": 0.201998001998002,
      "grad_norm": 2.0713963508605957,
      "learning_rate": 4.860668107674823e-05,
      "loss": 2.0109,
      "step": 1011
    },
    {
      "epoch": 0.2021978021978022,
      "grad_norm": 3.5446667671203613,
      "learning_rate": 4.8425926111375506e-05,
      "loss": 1.7487,
      "step": 1012
    },
    {
      "epoch": 0.2023976023976024,
      "grad_norm": 3.0079853534698486,
      "learning_rate": 4.824540042194443e-05,
      "loss": 1.9346,
      "step": 1013
    },
    {
      "epoch": 0.2025974025974026,
      "grad_norm": 1.0646512508392334,
      "learning_rate": 4.8065104810993856e-05,
      "loss": 1.93,
      "step": 1014
    },
    {
      "epoch": 0.20279720279720279,
      "grad_norm": 2.148587942123413,
      "learning_rate": 4.788504008003978e-05,
      "loss": 2.1741,
      "step": 1015
    },
    {
      "epoch": 0.202997002997003,
      "grad_norm": 1.3568068742752075,
      "learning_rate": 4.770520702957182e-05,
      "loss": 1.7085,
      "step": 1016
    },
    {
      "epoch": 0.2031968031968032,
      "grad_norm": 1.0131592750549316,
      "learning_rate": 4.752560645904962e-05,
      "loss": 2.539,
      "step": 1017
    },
    {
      "epoch": 0.2033966033966034,
      "grad_norm": 1.6859813928604126,
      "learning_rate": 4.734623916689941e-05,
      "loss": 1.6041,
      "step": 1018
    },
    {
      "epoch": 0.2035964035964036,
      "grad_norm": 1.8035125732421875,
      "learning_rate": 4.716710595051022e-05,
      "loss": 1.3656,
      "step": 1019
    },
    {
      "epoch": 0.2037962037962038,
      "grad_norm": 1.4954320192337036,
      "learning_rate": 4.698820760623064e-05,
      "loss": 2.2809,
      "step": 1020
    },
    {
      "epoch": 0.20399600399600398,
      "grad_norm": 4.195502758026123,
      "learning_rate": 4.6809544929365004e-05,
      "loss": 1.6693,
      "step": 1021
    },
    {
      "epoch": 0.2041958041958042,
      "grad_norm": 1.2347359657287598,
      "learning_rate": 4.663111871417e-05,
      "loss": 2.0528,
      "step": 1022
    },
    {
      "epoch": 0.2043956043956044,
      "grad_norm": 1.281032919883728,
      "learning_rate": 4.645292975385111e-05,
      "loss": 2.6097,
      "step": 1023
    },
    {
      "epoch": 0.2045954045954046,
      "grad_norm": 1.843165397644043,
      "learning_rate": 4.627497884055912e-05,
      "loss": 1.5506,
      "step": 1024
    },
    {
      "epoch": 0.2047952047952048,
      "grad_norm": 1.4382215738296509,
      "learning_rate": 4.609726676538652e-05,
      "loss": 1.4086,
      "step": 1025
    },
    {
      "epoch": 0.204995004995005,
      "grad_norm": 1.5982941389083862,
      "learning_rate": 4.591979431836402e-05,
      "loss": 2.1207,
      "step": 1026
    },
    {
      "epoch": 0.2051948051948052,
      "grad_norm": 0.7260691523551941,
      "learning_rate": 4.574256228845706e-05,
      "loss": 2.0259,
      "step": 1027
    },
    {
      "epoch": 0.20539460539460538,
      "grad_norm": 1.1398053169250488,
      "learning_rate": 4.5565571463562365e-05,
      "loss": 0.8232,
      "step": 1028
    },
    {
      "epoch": 0.2055944055944056,
      "grad_norm": 1.3954448699951172,
      "learning_rate": 4.5388822630504256e-05,
      "loss": 1.739,
      "step": 1029
    },
    {
      "epoch": 0.2057942057942058,
      "grad_norm": 1.9886215925216675,
      "learning_rate": 4.521231657503132e-05,
      "loss": 1.4644,
      "step": 1030
    },
    {
      "epoch": 0.205994005994006,
      "grad_norm": 1.9017658233642578,
      "learning_rate": 4.503605408181286e-05,
      "loss": 2.9676,
      "step": 1031
    },
    {
      "epoch": 0.2061938061938062,
      "grad_norm": 1.7367621660232544,
      "learning_rate": 4.486003593443537e-05,
      "loss": 2.2194,
      "step": 1032
    },
    {
      "epoch": 0.2063936063936064,
      "grad_norm": 3.36721134185791,
      "learning_rate": 4.468426291539914e-05,
      "loss": 2.4375,
      "step": 1033
    },
    {
      "epoch": 0.20659340659340658,
      "grad_norm": 2.9212398529052734,
      "learning_rate": 4.4508735806114654e-05,
      "loss": 2.1653,
      "step": 1034
    },
    {
      "epoch": 0.20679320679320679,
      "grad_norm": 2.392639636993408,
      "learning_rate": 4.433345538689929e-05,
      "loss": 2.1505,
      "step": 1035
    },
    {
      "epoch": 0.206993006993007,
      "grad_norm": 4.3227643966674805,
      "learning_rate": 4.415842243697369e-05,
      "loss": 1.913,
      "step": 1036
    },
    {
      "epoch": 0.2071928071928072,
      "grad_norm": 4.071188449859619,
      "learning_rate": 4.39836377344583e-05,
      "loss": 1.4823,
      "step": 1037
    },
    {
      "epoch": 0.2073926073926074,
      "grad_norm": 1.7091764211654663,
      "learning_rate": 4.380910205637012e-05,
      "loss": 1.8441,
      "step": 1038
    },
    {
      "epoch": 0.2075924075924076,
      "grad_norm": 2.6567697525024414,
      "learning_rate": 4.363481617861893e-05,
      "loss": 1.7593,
      "step": 1039
    },
    {
      "epoch": 0.2077922077922078,
      "grad_norm": 1.2343820333480835,
      "learning_rate": 4.346078087600412e-05,
      "loss": 1.0608,
      "step": 1040
    },
    {
      "epoch": 0.20799200799200798,
      "grad_norm": 2.9564220905303955,
      "learning_rate": 4.3286996922211034e-05,
      "loss": 1.9438,
      "step": 1041
    },
    {
      "epoch": 0.2081918081918082,
      "grad_norm": 1.8759785890579224,
      "learning_rate": 4.311346508980772e-05,
      "loss": 1.9417,
      "step": 1042
    },
    {
      "epoch": 0.2083916083916084,
      "grad_norm": 2.478180170059204,
      "learning_rate": 4.2940186150241365e-05,
      "loss": 2.1175,
      "step": 1043
    },
    {
      "epoch": 0.2085914085914086,
      "grad_norm": 1.2920022010803223,
      "learning_rate": 4.27671608738349e-05,
      "loss": 2.0684,
      "step": 1044
    },
    {
      "epoch": 0.2087912087912088,
      "grad_norm": 2.5861194133758545,
      "learning_rate": 4.2594390029783534e-05,
      "loss": 2.4857,
      "step": 1045
    },
    {
      "epoch": 0.208991008991009,
      "grad_norm": 2.5244009494781494,
      "learning_rate": 4.242187438615153e-05,
      "loss": 1.4436,
      "step": 1046
    },
    {
      "epoch": 0.20919080919080918,
      "grad_norm": 1.3091806173324585,
      "learning_rate": 4.224961470986849e-05,
      "loss": 1.4478,
      "step": 1047
    },
    {
      "epoch": 0.20939060939060938,
      "grad_norm": 1.3066073656082153,
      "learning_rate": 4.207761176672614e-05,
      "loss": 1.5521,
      "step": 1048
    },
    {
      "epoch": 0.2095904095904096,
      "grad_norm": 4.555966377258301,
      "learning_rate": 4.190586632137491e-05,
      "loss": 1.7281,
      "step": 1049
    },
    {
      "epoch": 0.2097902097902098,
      "grad_norm": 1.361045241355896,
      "learning_rate": 4.173437913732048e-05,
      "loss": 0.8749,
      "step": 1050
    },
    {
      "epoch": 0.20999000999001,
      "grad_norm": 4.6330413818359375,
      "learning_rate": 4.156315097692037e-05,
      "loss": 2.4297,
      "step": 1051
    },
    {
      "epoch": 0.2101898101898102,
      "grad_norm": 1.065653681755066,
      "learning_rate": 4.139218260138074e-05,
      "loss": 0.9002,
      "step": 1052
    },
    {
      "epoch": 0.21038961038961038,
      "grad_norm": 2.372692108154297,
      "learning_rate": 4.12214747707527e-05,
      "loss": 2.177,
      "step": 1053
    },
    {
      "epoch": 0.21058941058941058,
      "grad_norm": 2.7710981369018555,
      "learning_rate": 4.1051028243929125e-05,
      "loss": 2.5115,
      "step": 1054
    },
    {
      "epoch": 0.21078921078921078,
      "grad_norm": 3.0162222385406494,
      "learning_rate": 4.088084377864135e-05,
      "loss": 2.4858,
      "step": 1055
    },
    {
      "epoch": 0.210989010989011,
      "grad_norm": 1.809080719947815,
      "learning_rate": 4.07109221314556e-05,
      "loss": 1.9607,
      "step": 1056
    },
    {
      "epoch": 0.2111888111888112,
      "grad_norm": 0.943491518497467,
      "learning_rate": 4.054126405776971e-05,
      "loss": 0.6693,
      "step": 1057
    },
    {
      "epoch": 0.2113886113886114,
      "grad_norm": 3.1297006607055664,
      "learning_rate": 4.037187031180985e-05,
      "loss": 2.0412,
      "step": 1058
    },
    {
      "epoch": 0.2115884115884116,
      "grad_norm": 3.102771043777466,
      "learning_rate": 4.020274164662707e-05,
      "loss": 2.5403,
      "step": 1059
    },
    {
      "epoch": 0.21178821178821178,
      "grad_norm": 3.033423662185669,
      "learning_rate": 4.003387881409397e-05,
      "loss": 1.4481,
      "step": 1060
    },
    {
      "epoch": 0.21198801198801198,
      "grad_norm": 3.0310895442962646,
      "learning_rate": 3.986528256490141e-05,
      "loss": 2.0771,
      "step": 1061
    },
    {
      "epoch": 0.21218781218781219,
      "grad_norm": 2.808224678039551,
      "learning_rate": 3.969695364855511e-05,
      "loss": 2.0607,
      "step": 1062
    },
    {
      "epoch": 0.2123876123876124,
      "grad_norm": 1.1163580417633057,
      "learning_rate": 3.952889281337235e-05,
      "loss": 0.9773,
      "step": 1063
    },
    {
      "epoch": 0.2125874125874126,
      "grad_norm": 2.9041547775268555,
      "learning_rate": 3.93611008064786e-05,
      "loss": 2.4248,
      "step": 1064
    },
    {
      "epoch": 0.2127872127872128,
      "grad_norm": 1.8523093461990356,
      "learning_rate": 3.9193578373804364e-05,
      "loss": 1.3205,
      "step": 1065
    },
    {
      "epoch": 0.21298701298701297,
      "grad_norm": 2.4304003715515137,
      "learning_rate": 3.90263262600816e-05,
      "loss": 0.9685,
      "step": 1066
    },
    {
      "epoch": 0.21318681318681318,
      "grad_norm": 2.8599562644958496,
      "learning_rate": 3.88593452088406e-05,
      "loss": 1.612,
      "step": 1067
    },
    {
      "epoch": 0.21338661338661338,
      "grad_norm": 3.161896228790283,
      "learning_rate": 3.869263596240661e-05,
      "loss": 1.8184,
      "step": 1068
    },
    {
      "epoch": 0.2135864135864136,
      "grad_norm": 2.987400531768799,
      "learning_rate": 3.8526199261896544e-05,
      "loss": 2.4745,
      "step": 1069
    },
    {
      "epoch": 0.2137862137862138,
      "grad_norm": 1.4229843616485596,
      "learning_rate": 3.836003584721577e-05,
      "loss": 2.1778,
      "step": 1070
    },
    {
      "epoch": 0.213986013986014,
      "grad_norm": 2.435032367706299,
      "learning_rate": 3.8194146457054655e-05,
      "loss": 2.7151,
      "step": 1071
    },
    {
      "epoch": 0.2141858141858142,
      "grad_norm": 1.1492829322814941,
      "learning_rate": 3.802853182888543e-05,
      "loss": 1.2627,
      "step": 1072
    },
    {
      "epoch": 0.21438561438561438,
      "grad_norm": 2.142901659011841,
      "learning_rate": 3.786319269895877e-05,
      "loss": 1.7842,
      "step": 1073
    },
    {
      "epoch": 0.21458541458541458,
      "grad_norm": 1.140316367149353,
      "learning_rate": 3.769812980230074e-05,
      "loss": 2.2437,
      "step": 1074
    },
    {
      "epoch": 0.21478521478521478,
      "grad_norm": 2.691150665283203,
      "learning_rate": 3.7533343872709294e-05,
      "loss": 1.0087,
      "step": 1075
    },
    {
      "epoch": 0.214985014985015,
      "grad_norm": 2.1415464878082275,
      "learning_rate": 3.736883564275112e-05,
      "loss": 1.7191,
      "step": 1076
    },
    {
      "epoch": 0.2151848151848152,
      "grad_norm": 3.1888504028320312,
      "learning_rate": 3.7204605843758386e-05,
      "loss": 1.7471,
      "step": 1077
    },
    {
      "epoch": 0.2153846153846154,
      "grad_norm": 2.23016619682312,
      "learning_rate": 3.704065520582549e-05,
      "loss": 1.9055,
      "step": 1078
    },
    {
      "epoch": 0.21558441558441557,
      "grad_norm": 2.4021713733673096,
      "learning_rate": 3.6876984457805786e-05,
      "loss": 1.4407,
      "step": 1079
    },
    {
      "epoch": 0.21578421578421578,
      "grad_norm": 2.4166436195373535,
      "learning_rate": 3.671359432730834e-05,
      "loss": 1.9342,
      "step": 1080
    },
    {
      "epoch": 0.21598401598401598,
      "grad_norm": 3.7197539806365967,
      "learning_rate": 3.655048554069478e-05,
      "loss": 2.1429,
      "step": 1081
    },
    {
      "epoch": 0.21618381618381619,
      "grad_norm": 1.0120347738265991,
      "learning_rate": 3.638765882307589e-05,
      "loss": 2.3855,
      "step": 1082
    },
    {
      "epoch": 0.2163836163836164,
      "grad_norm": 1.7315000295639038,
      "learning_rate": 3.6225114898308634e-05,
      "loss": 1.8736,
      "step": 1083
    },
    {
      "epoch": 0.2165834165834166,
      "grad_norm": 2.6747043132781982,
      "learning_rate": 3.6062854488992714e-05,
      "loss": 1.7091,
      "step": 1084
    },
    {
      "epoch": 0.21678321678321677,
      "grad_norm": 2.637986898422241,
      "learning_rate": 3.5900878316467454e-05,
      "loss": 1.3522,
      "step": 1085
    },
    {
      "epoch": 0.21698301698301697,
      "grad_norm": 0.9538910984992981,
      "learning_rate": 3.573918710080857e-05,
      "loss": 0.6764,
      "step": 1086
    },
    {
      "epoch": 0.21718281718281718,
      "grad_norm": 2.6533548831939697,
      "learning_rate": 3.5577781560825066e-05,
      "loss": 1.8517,
      "step": 1087
    },
    {
      "epoch": 0.21738261738261738,
      "grad_norm": 1.782597541809082,
      "learning_rate": 3.541666241405588e-05,
      "loss": 1.2182,
      "step": 1088
    },
    {
      "epoch": 0.2175824175824176,
      "grad_norm": 3.9720616340637207,
      "learning_rate": 3.5255830376766764e-05,
      "loss": 1.1809,
      "step": 1089
    },
    {
      "epoch": 0.2177822177822178,
      "grad_norm": 2.3164825439453125,
      "learning_rate": 3.509528616394716e-05,
      "loss": 1.6263,
      "step": 1090
    },
    {
      "epoch": 0.217982017982018,
      "grad_norm": 2.402500629425049,
      "learning_rate": 3.4935030489306883e-05,
      "loss": 1.2161,
      "step": 1091
    },
    {
      "epoch": 0.21818181818181817,
      "grad_norm": 2.864975690841675,
      "learning_rate": 3.4775064065273165e-05,
      "loss": 2.3578,
      "step": 1092
    },
    {
      "epoch": 0.21838161838161838,
      "grad_norm": 2.8155481815338135,
      "learning_rate": 3.4615387602987236e-05,
      "loss": 1.88,
      "step": 1093
    },
    {
      "epoch": 0.21858141858141858,
      "grad_norm": 2.084221839904785,
      "learning_rate": 3.445600181230134e-05,
      "loss": 1.2613,
      "step": 1094
    },
    {
      "epoch": 0.21878121878121878,
      "grad_norm": 1.5177751779556274,
      "learning_rate": 3.429690740177549e-05,
      "loss": 2.1246,
      "step": 1095
    },
    {
      "epoch": 0.218981018981019,
      "grad_norm": 1.3483099937438965,
      "learning_rate": 3.413810507867436e-05,
      "loss": 2.49,
      "step": 1096
    },
    {
      "epoch": 0.2191808191808192,
      "grad_norm": 1.0184931755065918,
      "learning_rate": 3.397959554896415e-05,
      "loss": 2.3546,
      "step": 1097
    },
    {
      "epoch": 0.21938061938061937,
      "grad_norm": 1.4430732727050781,
      "learning_rate": 3.3821379517309405e-05,
      "loss": 1.9969,
      "step": 1098
    },
    {
      "epoch": 0.21958041958041957,
      "grad_norm": 1.0055434703826904,
      "learning_rate": 3.3663457687069924e-05,
      "loss": 1.1596,
      "step": 1099
    },
    {
      "epoch": 0.21978021978021978,
      "grad_norm": 2.2729146480560303,
      "learning_rate": 3.350583076029754e-05,
      "loss": 1.1651,
      "step": 1100
    },
    {
      "epoch": 0.21998001998001998,
      "grad_norm": 3.3770875930786133,
      "learning_rate": 3.334849943773323e-05,
      "loss": 1.69,
      "step": 1101
    },
    {
      "epoch": 0.22017982017982018,
      "grad_norm": 1.5588802099227905,
      "learning_rate": 3.319146441880371e-05,
      "loss": 2.1901,
      "step": 1102
    },
    {
      "epoch": 0.2203796203796204,
      "grad_norm": 1.8612388372421265,
      "learning_rate": 3.3034726401618444e-05,
      "loss": 2.362,
      "step": 1103
    },
    {
      "epoch": 0.2205794205794206,
      "grad_norm": 3.244528293609619,
      "learning_rate": 3.28782860829667e-05,
      "loss": 1.9285,
      "step": 1104
    },
    {
      "epoch": 0.22077922077922077,
      "grad_norm": 2.2998971939086914,
      "learning_rate": 3.272214415831418e-05,
      "loss": 1.8675,
      "step": 1105
    },
    {
      "epoch": 0.22097902097902097,
      "grad_norm": 1.0410702228546143,
      "learning_rate": 3.2566301321800085e-05,
      "loss": 0.8278,
      "step": 1106
    },
    {
      "epoch": 0.22117882117882118,
      "grad_norm": 2.2678260803222656,
      "learning_rate": 3.241075826623401e-05,
      "loss": 2.5433,
      "step": 1107
    },
    {
      "epoch": 0.22137862137862138,
      "grad_norm": 0.6878364086151123,
      "learning_rate": 3.225551568309284e-05,
      "loss": 0.3681,
      "step": 1108
    },
    {
      "epoch": 0.22157842157842159,
      "grad_norm": 1.4128341674804688,
      "learning_rate": 3.210057426251773e-05,
      "loss": 2.3949,
      "step": 1109
    },
    {
      "epoch": 0.2217782217782218,
      "grad_norm": 3.2122933864593506,
      "learning_rate": 3.1945934693310896e-05,
      "loss": 1.533,
      "step": 1110
    },
    {
      "epoch": 0.22197802197802197,
      "grad_norm": 1.1943044662475586,
      "learning_rate": 3.179159766293282e-05,
      "loss": 2.5525,
      "step": 1111
    },
    {
      "epoch": 0.22217782217782217,
      "grad_norm": 2.964670181274414,
      "learning_rate": 3.163756385749889e-05,
      "loss": 2.0801,
      "step": 1112
    },
    {
      "epoch": 0.22237762237762237,
      "grad_norm": 3.8190441131591797,
      "learning_rate": 3.148383396177653e-05,
      "loss": 1.6675,
      "step": 1113
    },
    {
      "epoch": 0.22257742257742258,
      "grad_norm": 1.440389633178711,
      "learning_rate": 3.133040865918213e-05,
      "loss": 1.2342,
      "step": 1114
    },
    {
      "epoch": 0.22277722277722278,
      "grad_norm": 2.0289714336395264,
      "learning_rate": 3.117728863177796e-05,
      "loss": 1.102,
      "step": 1115
    },
    {
      "epoch": 0.222977022977023,
      "grad_norm": 2.587390661239624,
      "learning_rate": 3.102447456026919e-05,
      "loss": 1.6792,
      "step": 1116
    },
    {
      "epoch": 0.22317682317682316,
      "grad_norm": 2.102611541748047,
      "learning_rate": 3.0871967124000834e-05,
      "loss": 1.7596,
      "step": 1117
    },
    {
      "epoch": 0.22337662337662337,
      "grad_norm": 2.009068250656128,
      "learning_rate": 3.0719767000954714e-05,
      "loss": 1.9345,
      "step": 1118
    },
    {
      "epoch": 0.22357642357642357,
      "grad_norm": 2.5964889526367188,
      "learning_rate": 3.056787486774656e-05,
      "loss": 2.1152,
      "step": 1119
    },
    {
      "epoch": 0.22377622377622378,
      "grad_norm": 2.0954530239105225,
      "learning_rate": 3.041629139962283e-05,
      "loss": 2.7949,
      "step": 1120
    },
    {
      "epoch": 0.22397602397602398,
      "grad_norm": 2.1943793296813965,
      "learning_rate": 3.0265017270457775e-05,
      "loss": 1.7696,
      "step": 1121
    },
    {
      "epoch": 0.22417582417582418,
      "grad_norm": 2.569697141647339,
      "learning_rate": 3.0114053152750556e-05,
      "loss": 1.6453,
      "step": 1122
    },
    {
      "epoch": 0.2243756243756244,
      "grad_norm": 2.638148307800293,
      "learning_rate": 2.9963399717622077e-05,
      "loss": 2.0991,
      "step": 1123
    },
    {
      "epoch": 0.22457542457542456,
      "grad_norm": 2.01450252532959,
      "learning_rate": 2.98130576348121e-05,
      "loss": 1.7859,
      "step": 1124
    },
    {
      "epoch": 0.22477522477522477,
      "grad_norm": 2.278923988342285,
      "learning_rate": 2.966302757267625e-05,
      "loss": 1.8646,
      "step": 1125
    },
    {
      "epoch": 0.22477522477522477,
      "eval_loss": 1.7687376737594604,
      "eval_runtime": 124.0933,
      "eval_samples_per_second": 33.966,
      "eval_steps_per_second": 8.494,
      "step": 1125
    },
    {
      "epoch": 0.22497502497502497,
      "grad_norm": 2.8302416801452637,
      "learning_rate": 2.9513310198183065e-05,
      "loss": 2.0466,
      "step": 1126
    },
    {
      "epoch": 0.22517482517482518,
      "grad_norm": 2.9561195373535156,
      "learning_rate": 2.936390617691097e-05,
      "loss": 2.156,
      "step": 1127
    },
    {
      "epoch": 0.22537462537462538,
      "grad_norm": 1.7244153022766113,
      "learning_rate": 2.9214816173045356e-05,
      "loss": 1.633,
      "step": 1128
    },
    {
      "epoch": 0.22557442557442559,
      "grad_norm": 2.5604629516601562,
      "learning_rate": 2.906604084937572e-05,
      "loss": 1.4548,
      "step": 1129
    },
    {
      "epoch": 0.22577422577422576,
      "grad_norm": 0.8351600766181946,
      "learning_rate": 2.8917580867292526e-05,
      "loss": 0.595,
      "step": 1130
    },
    {
      "epoch": 0.22597402597402597,
      "grad_norm": 2.61030912399292,
      "learning_rate": 2.8769436886784408e-05,
      "loss": 1.2764,
      "step": 1131
    },
    {
      "epoch": 0.22617382617382617,
      "grad_norm": 2.402665376663208,
      "learning_rate": 2.862160956643517e-05,
      "loss": 1.9498,
      "step": 1132
    },
    {
      "epoch": 0.22637362637362637,
      "grad_norm": 3.01234769821167,
      "learning_rate": 2.847409956342092e-05,
      "loss": 2.2981,
      "step": 1133
    },
    {
      "epoch": 0.22657342657342658,
      "grad_norm": 1.9059995412826538,
      "learning_rate": 2.8326907533507074e-05,
      "loss": 1.3305,
      "step": 1134
    },
    {
      "epoch": 0.22677322677322678,
      "grad_norm": 1.4675731658935547,
      "learning_rate": 2.8180034131045464e-05,
      "loss": 1.379,
      "step": 1135
    },
    {
      "epoch": 0.226973026973027,
      "grad_norm": 1.9874333143234253,
      "learning_rate": 2.8033480008971546e-05,
      "loss": 2.3787,
      "step": 1136
    },
    {
      "epoch": 0.22717282717282716,
      "grad_norm": 1.1071186065673828,
      "learning_rate": 2.7887245818801277e-05,
      "loss": 1.548,
      "step": 1137
    },
    {
      "epoch": 0.22737262737262737,
      "grad_norm": 1.9095979928970337,
      "learning_rate": 2.7741332210628345e-05,
      "loss": 1.453,
      "step": 1138
    },
    {
      "epoch": 0.22757242757242757,
      "grad_norm": 3.9741580486297607,
      "learning_rate": 2.759573983312138e-05,
      "loss": 1.5055,
      "step": 1139
    },
    {
      "epoch": 0.22777222777222778,
      "grad_norm": 2.0852982997894287,
      "learning_rate": 2.7450469333520855e-05,
      "loss": 1.9138,
      "step": 1140
    },
    {
      "epoch": 0.22797202797202798,
      "grad_norm": 2.784945249557495,
      "learning_rate": 2.730552135763632e-05,
      "loss": 2.2776,
      "step": 1141
    },
    {
      "epoch": 0.22817182817182818,
      "grad_norm": 3.37286639213562,
      "learning_rate": 2.7160896549843562e-05,
      "loss": 1.5284,
      "step": 1142
    },
    {
      "epoch": 0.22837162837162836,
      "grad_norm": 1.4662927389144897,
      "learning_rate": 2.701659555308169e-05,
      "loss": 1.0145,
      "step": 1143
    },
    {
      "epoch": 0.22857142857142856,
      "grad_norm": 2.640758514404297,
      "learning_rate": 2.6872619008850274e-05,
      "loss": 2.4774,
      "step": 1144
    },
    {
      "epoch": 0.22877122877122877,
      "grad_norm": 1.9995914697647095,
      "learning_rate": 2.672896755720654e-05,
      "loss": 0.6398,
      "step": 1145
    },
    {
      "epoch": 0.22897102897102897,
      "grad_norm": 1.8506040573120117,
      "learning_rate": 2.6585641836762433e-05,
      "loss": 1.6269,
      "step": 1146
    },
    {
      "epoch": 0.22917082917082918,
      "grad_norm": 1.2241052389144897,
      "learning_rate": 2.6442642484681944e-05,
      "loss": 1.3596,
      "step": 1147
    },
    {
      "epoch": 0.22937062937062938,
      "grad_norm": 1.5470662117004395,
      "learning_rate": 2.6299970136678077e-05,
      "loss": 1.8979,
      "step": 1148
    },
    {
      "epoch": 0.22957042957042956,
      "grad_norm": 1.2168165445327759,
      "learning_rate": 2.6157625427010156e-05,
      "loss": 2.1255,
      "step": 1149
    },
    {
      "epoch": 0.22977022977022976,
      "grad_norm": 3.269280433654785,
      "learning_rate": 2.6015608988480955e-05,
      "loss": 1.6236,
      "step": 1150
    },
    {
      "epoch": 0.22997002997002997,
      "grad_norm": 2.274259567260742,
      "learning_rate": 2.5873921452433915e-05,
      "loss": 2.2945,
      "step": 1151
    },
    {
      "epoch": 0.23016983016983017,
      "grad_norm": 6.463413715362549,
      "learning_rate": 2.57325634487503e-05,
      "loss": 1.2932,
      "step": 1152
    },
    {
      "epoch": 0.23036963036963037,
      "grad_norm": 4.268492698669434,
      "learning_rate": 2.5591535605846383e-05,
      "loss": 1.6785,
      "step": 1153
    },
    {
      "epoch": 0.23056943056943058,
      "grad_norm": 1.3531522750854492,
      "learning_rate": 2.5450838550670808e-05,
      "loss": 1.6665,
      "step": 1154
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 3.3075051307678223,
      "learning_rate": 2.5310472908701555e-05,
      "loss": 1.7962,
      "step": 1155
    },
    {
      "epoch": 0.23096903096903096,
      "grad_norm": 3.319352149963379,
      "learning_rate": 2.5170439303943294e-05,
      "loss": 2.5966,
      "step": 1156
    },
    {
      "epoch": 0.23116883116883116,
      "grad_norm": 2.1255486011505127,
      "learning_rate": 2.503073835892471e-05,
      "loss": 1.7878,
      "step": 1157
    },
    {
      "epoch": 0.23136863136863137,
      "grad_norm": 2.436182975769043,
      "learning_rate": 2.4891370694695517e-05,
      "loss": 1.681,
      "step": 1158
    },
    {
      "epoch": 0.23156843156843157,
      "grad_norm": 3.0848090648651123,
      "learning_rate": 2.4752336930823837e-05,
      "loss": 1.366,
      "step": 1159
    },
    {
      "epoch": 0.23176823176823177,
      "grad_norm": 2.6933181285858154,
      "learning_rate": 2.4613637685393432e-05,
      "loss": 2.0712,
      "step": 1160
    },
    {
      "epoch": 0.23196803196803198,
      "grad_norm": 1.505959391593933,
      "learning_rate": 2.4475273575000936e-05,
      "loss": 1.3355,
      "step": 1161
    },
    {
      "epoch": 0.23216783216783216,
      "grad_norm": 2.8885483741760254,
      "learning_rate": 2.4337245214753103e-05,
      "loss": 1.8612,
      "step": 1162
    },
    {
      "epoch": 0.23236763236763236,
      "grad_norm": 3.609788179397583,
      "learning_rate": 2.4199553218264093e-05,
      "loss": 0.9082,
      "step": 1163
    },
    {
      "epoch": 0.23256743256743256,
      "grad_norm": 2.3863630294799805,
      "learning_rate": 2.4062198197652752e-05,
      "loss": 2.035,
      "step": 1164
    },
    {
      "epoch": 0.23276723276723277,
      "grad_norm": 1.7970850467681885,
      "learning_rate": 2.3925180763539844e-05,
      "loss": 1.4872,
      "step": 1165
    },
    {
      "epoch": 0.23296703296703297,
      "grad_norm": 2.523423910140991,
      "learning_rate": 2.3788501525045438e-05,
      "loss": 2.415,
      "step": 1166
    },
    {
      "epoch": 0.23316683316683318,
      "grad_norm": 2.222416400909424,
      "learning_rate": 2.3652161089786086e-05,
      "loss": 0.883,
      "step": 1167
    },
    {
      "epoch": 0.23336663336663338,
      "grad_norm": 2.1163458824157715,
      "learning_rate": 2.351616006387214e-05,
      "loss": 1.2108,
      "step": 1168
    },
    {
      "epoch": 0.23356643356643356,
      "grad_norm": 2.3124494552612305,
      "learning_rate": 2.3380499051905137e-05,
      "loss": 1.8916,
      "step": 1169
    },
    {
      "epoch": 0.23376623376623376,
      "grad_norm": 2.8991920948028564,
      "learning_rate": 2.324517865697501e-05,
      "loss": 1.7406,
      "step": 1170
    },
    {
      "epoch": 0.23396603396603396,
      "grad_norm": 1.7772811651229858,
      "learning_rate": 2.3110199480657525e-05,
      "loss": 2.246,
      "step": 1171
    },
    {
      "epoch": 0.23416583416583417,
      "grad_norm": 1.3433674573898315,
      "learning_rate": 2.2975562123011495e-05,
      "loss": 1.6542,
      "step": 1172
    },
    {
      "epoch": 0.23436563436563437,
      "grad_norm": 0.7129135131835938,
      "learning_rate": 2.2841267182576143e-05,
      "loss": 1.0992,
      "step": 1173
    },
    {
      "epoch": 0.23456543456543458,
      "grad_norm": 2.047239065170288,
      "learning_rate": 2.2707315256368433e-05,
      "loss": 1.9429,
      "step": 1174
    },
    {
      "epoch": 0.23476523476523475,
      "grad_norm": 1.8302230834960938,
      "learning_rate": 2.2573706939880555e-05,
      "loss": 2.6155,
      "step": 1175
    },
    {
      "epoch": 0.23496503496503496,
      "grad_norm": 1.5473768711090088,
      "learning_rate": 2.2440442827077045e-05,
      "loss": 2.5393,
      "step": 1176
    },
    {
      "epoch": 0.23516483516483516,
      "grad_norm": 3.2450220584869385,
      "learning_rate": 2.230752351039228e-05,
      "loss": 1.8841,
      "step": 1177
    },
    {
      "epoch": 0.23536463536463537,
      "grad_norm": 1.6718326807022095,
      "learning_rate": 2.2174949580727832e-05,
      "loss": 1.9738,
      "step": 1178
    },
    {
      "epoch": 0.23556443556443557,
      "grad_norm": 1.4383682012557983,
      "learning_rate": 2.2042721627449846e-05,
      "loss": 1.9168,
      "step": 1179
    },
    {
      "epoch": 0.23576423576423577,
      "grad_norm": 1.3591008186340332,
      "learning_rate": 2.1910840238386398e-05,
      "loss": 1.3081,
      "step": 1180
    },
    {
      "epoch": 0.23596403596403595,
      "grad_norm": 1.5118911266326904,
      "learning_rate": 2.1779305999824884e-05,
      "loss": 0.657,
      "step": 1181
    },
    {
      "epoch": 0.23616383616383615,
      "grad_norm": 2.038100242614746,
      "learning_rate": 2.164811949650942e-05,
      "loss": 1.8349,
      "step": 1182
    },
    {
      "epoch": 0.23636363636363636,
      "grad_norm": 0.5831717848777771,
      "learning_rate": 2.1517281311638217e-05,
      "loss": 0.6895,
      "step": 1183
    },
    {
      "epoch": 0.23656343656343656,
      "grad_norm": 4.429853439331055,
      "learning_rate": 2.1386792026861103e-05,
      "loss": 2.4122,
      "step": 1184
    },
    {
      "epoch": 0.23676323676323677,
      "grad_norm": 2.6585118770599365,
      "learning_rate": 2.125665222227675e-05,
      "loss": 1.7502,
      "step": 1185
    },
    {
      "epoch": 0.23696303696303697,
      "grad_norm": 2.5498976707458496,
      "learning_rate": 2.112686247643024e-05,
      "loss": 1.5595,
      "step": 1186
    },
    {
      "epoch": 0.23716283716283718,
      "grad_norm": 1.235526204109192,
      "learning_rate": 2.09974233663104e-05,
      "loss": 1.037,
      "step": 1187
    },
    {
      "epoch": 0.23736263736263735,
      "grad_norm": 1.0872644186019897,
      "learning_rate": 2.0868335467347366e-05,
      "loss": 2.0625,
      "step": 1188
    },
    {
      "epoch": 0.23756243756243756,
      "grad_norm": 3.6115143299102783,
      "learning_rate": 2.073959935340988e-05,
      "loss": 1.772,
      "step": 1189
    },
    {
      "epoch": 0.23776223776223776,
      "grad_norm": 2.576842784881592,
      "learning_rate": 2.06112155968028e-05,
      "loss": 1.4477,
      "step": 1190
    },
    {
      "epoch": 0.23796203796203796,
      "grad_norm": 2.361271381378174,
      "learning_rate": 2.0483184768264596e-05,
      "loss": 1.5352,
      "step": 1191
    },
    {
      "epoch": 0.23816183816183817,
      "grad_norm": 1.4435462951660156,
      "learning_rate": 2.035550743696468e-05,
      "loss": 2.1469,
      "step": 1192
    },
    {
      "epoch": 0.23836163836163837,
      "grad_norm": 3.190134048461914,
      "learning_rate": 2.022818417050113e-05,
      "loss": 1.9878,
      "step": 1193
    },
    {
      "epoch": 0.23856143856143855,
      "grad_norm": 5.2222771644592285,
      "learning_rate": 2.0101215534897855e-05,
      "loss": 2.1835,
      "step": 1194
    },
    {
      "epoch": 0.23876123876123875,
      "grad_norm": 2.298099994659424,
      "learning_rate": 1.99746020946023e-05,
      "loss": 2.1838,
      "step": 1195
    },
    {
      "epoch": 0.23896103896103896,
      "grad_norm": 1.3423162698745728,
      "learning_rate": 1.9848344412482854e-05,
      "loss": 1.9843,
      "step": 1196
    },
    {
      "epoch": 0.23916083916083916,
      "grad_norm": 0.77281254529953,
      "learning_rate": 1.9722443049826344e-05,
      "loss": 0.6409,
      "step": 1197
    },
    {
      "epoch": 0.23936063936063937,
      "grad_norm": 3.027249813079834,
      "learning_rate": 1.9596898566335576e-05,
      "loss": 1.0255,
      "step": 1198
    },
    {
      "epoch": 0.23956043956043957,
      "grad_norm": 2.236252784729004,
      "learning_rate": 1.9471711520126824e-05,
      "loss": 1.5678,
      "step": 1199
    },
    {
      "epoch": 0.23976023976023977,
      "grad_norm": 3.8324012756347656,
      "learning_rate": 1.9346882467727325e-05,
      "loss": 1.6375,
      "step": 1200
    },
    {
      "epoch": 0.23996003996003995,
      "grad_norm": 0.8493699431419373,
      "learning_rate": 1.9222411964072884e-05,
      "loss": 2.3111,
      "step": 1201
    },
    {
      "epoch": 0.24015984015984015,
      "grad_norm": 5.565759658813477,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 2.0174,
      "step": 1202
    },
    {
      "epoch": 0.24035964035964036,
      "grad_norm": 3.1273725032806396,
      "learning_rate": 1.8974548814769944e-05,
      "loss": 2.2839,
      "step": 1203
    },
    {
      "epoch": 0.24055944055944056,
      "grad_norm": 2.09582257270813,
      "learning_rate": 1.8851157271013442e-05,
      "loss": 2.12,
      "step": 1204
    },
    {
      "epoch": 0.24075924075924077,
      "grad_norm": 1.4521441459655762,
      "learning_rate": 1.872812647978095e-05,
      "loss": 1.7027,
      "step": 1205
    },
    {
      "epoch": 0.24095904095904097,
      "grad_norm": 3.6881027221679688,
      "learning_rate": 1.8605456988014015e-05,
      "loss": 2.2427,
      "step": 1206
    },
    {
      "epoch": 0.24115884115884115,
      "grad_norm": 1.068831205368042,
      "learning_rate": 1.8483149341047923e-05,
      "loss": 2.4938,
      "step": 1207
    },
    {
      "epoch": 0.24135864135864135,
      "grad_norm": 3.0096325874328613,
      "learning_rate": 1.8361204082609352e-05,
      "loss": 2.1785,
      "step": 1208
    },
    {
      "epoch": 0.24155844155844156,
      "grad_norm": 1.125986099243164,
      "learning_rate": 1.8239621754813995e-05,
      "loss": 0.7494,
      "step": 1209
    },
    {
      "epoch": 0.24175824175824176,
      "grad_norm": 4.480381965637207,
      "learning_rate": 1.811840289816409e-05,
      "loss": 1.6715,
      "step": 1210
    },
    {
      "epoch": 0.24195804195804196,
      "grad_norm": 4.932903289794922,
      "learning_rate": 1.799754805154603e-05,
      "loss": 2.1218,
      "step": 1211
    },
    {
      "epoch": 0.24215784215784217,
      "grad_norm": 2.4358506202697754,
      "learning_rate": 1.787705775222802e-05,
      "loss": 2.0109,
      "step": 1212
    },
    {
      "epoch": 0.24235764235764234,
      "grad_norm": 1.8057516813278198,
      "learning_rate": 1.775693253585763e-05,
      "loss": 2.4749,
      "step": 1213
    },
    {
      "epoch": 0.24255744255744255,
      "grad_norm": 1.180400013923645,
      "learning_rate": 1.763717293645939e-05,
      "loss": 1.2958,
      "step": 1214
    },
    {
      "epoch": 0.24275724275724275,
      "grad_norm": 2.089881658554077,
      "learning_rate": 1.7517779486432495e-05,
      "loss": 1.9762,
      "step": 1215
    },
    {
      "epoch": 0.24295704295704296,
      "grad_norm": 2.9989922046661377,
      "learning_rate": 1.7398752716548395e-05,
      "loss": 2.2775,
      "step": 1216
    },
    {
      "epoch": 0.24315684315684316,
      "grad_norm": 0.7617408633232117,
      "learning_rate": 1.728009315594843e-05,
      "loss": 0.4476,
      "step": 1217
    },
    {
      "epoch": 0.24335664335664337,
      "grad_norm": 3.2402334213256836,
      "learning_rate": 1.716180133214149e-05,
      "loss": 2.0595,
      "step": 1218
    },
    {
      "epoch": 0.24355644355644357,
      "grad_norm": 1.1942652463912964,
      "learning_rate": 1.704387777100165e-05,
      "loss": 1.7,
      "step": 1219
    },
    {
      "epoch": 0.24375624375624375,
      "grad_norm": 1.7374550104141235,
      "learning_rate": 1.6926322996765897e-05,
      "loss": 1.1294,
      "step": 1220
    },
    {
      "epoch": 0.24395604395604395,
      "grad_norm": 1.8940194845199585,
      "learning_rate": 1.6809137532031704e-05,
      "loss": 1.9679,
      "step": 1221
    },
    {
      "epoch": 0.24415584415584415,
      "grad_norm": 2.657851219177246,
      "learning_rate": 1.6692321897754758e-05,
      "loss": 2.0838,
      "step": 1222
    },
    {
      "epoch": 0.24435564435564436,
      "grad_norm": 2.7410597801208496,
      "learning_rate": 1.65758766132467e-05,
      "loss": 1.3946,
      "step": 1223
    },
    {
      "epoch": 0.24455544455544456,
      "grad_norm": 1.3236883878707886,
      "learning_rate": 1.6459802196172668e-05,
      "loss": 2.2545,
      "step": 1224
    },
    {
      "epoch": 0.24475524475524477,
      "grad_norm": 0.9156760573387146,
      "learning_rate": 1.634409916254914e-05,
      "loss": 2.2094,
      "step": 1225
    },
    {
      "epoch": 0.24495504495504494,
      "grad_norm": 2.15793776512146,
      "learning_rate": 1.622876802674158e-05,
      "loss": 1.7692,
      "step": 1226
    },
    {
      "epoch": 0.24515484515484515,
      "grad_norm": 2.207073926925659,
      "learning_rate": 1.6113809301462125e-05,
      "loss": 1.4758,
      "step": 1227
    },
    {
      "epoch": 0.24535464535464535,
      "grad_norm": 1.7459187507629395,
      "learning_rate": 1.599922349776738e-05,
      "loss": 2.1694,
      "step": 1228
    },
    {
      "epoch": 0.24555444555444556,
      "grad_norm": 1.7124254703521729,
      "learning_rate": 1.5885011125056047e-05,
      "loss": 1.3,
      "step": 1229
    },
    {
      "epoch": 0.24575424575424576,
      "grad_norm": 1.9106528759002686,
      "learning_rate": 1.5771172691066794e-05,
      "loss": 1.8295,
      "step": 1230
    },
    {
      "epoch": 0.24595404595404596,
      "grad_norm": 0.9798733592033386,
      "learning_rate": 1.565770870187585e-05,
      "loss": 1.1027,
      "step": 1231
    },
    {
      "epoch": 0.24615384615384617,
      "grad_norm": 0.9579630494117737,
      "learning_rate": 1.5544619661894864e-05,
      "loss": 1.6364,
      "step": 1232
    },
    {
      "epoch": 0.24635364635364634,
      "grad_norm": 1.3720325231552124,
      "learning_rate": 1.543190607386861e-05,
      "loss": 0.653,
      "step": 1233
    },
    {
      "epoch": 0.24655344655344655,
      "grad_norm": 1.4701471328735352,
      "learning_rate": 1.5319568438872745e-05,
      "loss": 0.8227,
      "step": 1234
    },
    {
      "epoch": 0.24675324675324675,
      "grad_norm": 1.8065595626831055,
      "learning_rate": 1.520760725631164e-05,
      "loss": 1.4186,
      "step": 1235
    },
    {
      "epoch": 0.24695304695304696,
      "grad_norm": 3.4482524394989014,
      "learning_rate": 1.5096023023916094e-05,
      "loss": 2.1503,
      "step": 1236
    },
    {
      "epoch": 0.24715284715284716,
      "grad_norm": 1.537431001663208,
      "learning_rate": 1.498481623774115e-05,
      "loss": 1.4033,
      "step": 1237
    },
    {
      "epoch": 0.24735264735264736,
      "grad_norm": 2.0567922592163086,
      "learning_rate": 1.4873987392163947e-05,
      "loss": 0.8103,
      "step": 1238
    },
    {
      "epoch": 0.24755244755244754,
      "grad_norm": 2.463792085647583,
      "learning_rate": 1.4763536979881354e-05,
      "loss": 1.7582,
      "step": 1239
    },
    {
      "epoch": 0.24775224775224775,
      "grad_norm": 2.161283493041992,
      "learning_rate": 1.4653465491908003e-05,
      "loss": 1.7228,
      "step": 1240
    },
    {
      "epoch": 0.24795204795204795,
      "grad_norm": 2.042328357696533,
      "learning_rate": 1.4543773417573925e-05,
      "loss": 1.968,
      "step": 1241
    },
    {
      "epoch": 0.24815184815184815,
      "grad_norm": 4.177019119262695,
      "learning_rate": 1.4434461244522458e-05,
      "loss": 2.2524,
      "step": 1242
    },
    {
      "epoch": 0.24835164835164836,
      "grad_norm": 1.0806447267532349,
      "learning_rate": 1.4325529458708065e-05,
      "loss": 2.3989,
      "step": 1243
    },
    {
      "epoch": 0.24855144855144856,
      "grad_norm": 1.9708497524261475,
      "learning_rate": 1.4216978544394177e-05,
      "loss": 1.3812,
      "step": 1244
    },
    {
      "epoch": 0.24875124875124874,
      "grad_norm": 3.028118848800659,
      "learning_rate": 1.4108808984151023e-05,
      "loss": 2.1919,
      "step": 1245
    },
    {
      "epoch": 0.24895104895104894,
      "grad_norm": 1.7838475704193115,
      "learning_rate": 1.4001021258853509e-05,
      "loss": 1.1901,
      "step": 1246
    },
    {
      "epoch": 0.24915084915084915,
      "grad_norm": 3.597397804260254,
      "learning_rate": 1.3893615847679065e-05,
      "loss": 2.4456,
      "step": 1247
    },
    {
      "epoch": 0.24935064935064935,
      "grad_norm": 2.764890193939209,
      "learning_rate": 1.3786593228105494e-05,
      "loss": 1.766,
      "step": 1248
    },
    {
      "epoch": 0.24955044955044955,
      "grad_norm": 3.7665417194366455,
      "learning_rate": 1.3679953875908957e-05,
      "loss": 2.1258,
      "step": 1249
    },
    {
      "epoch": 0.24975024975024976,
      "grad_norm": 3.09995174407959,
      "learning_rate": 1.3573698265161683e-05,
      "loss": 2.3165,
      "step": 1250
    },
    {
      "epoch": 0.24995004995004996,
      "grad_norm": 4.309656620025635,
      "learning_rate": 1.3467826868229994e-05,
      "loss": 1.4685,
      "step": 1251
    },
    {
      "epoch": 0.25014985014985014,
      "grad_norm": 0.837217390537262,
      "learning_rate": 1.3362340155772146e-05,
      "loss": 0.4723,
      "step": 1252
    },
    {
      "epoch": 0.25034965034965034,
      "grad_norm": 1.6218265295028687,
      "learning_rate": 1.3257238596736266e-05,
      "loss": 1.5754,
      "step": 1253
    },
    {
      "epoch": 0.25054945054945055,
      "grad_norm": 3.8803489208221436,
      "learning_rate": 1.3152522658358245e-05,
      "loss": 2.9792,
      "step": 1254
    },
    {
      "epoch": 0.25074925074925075,
      "grad_norm": 1.7271703481674194,
      "learning_rate": 1.3048192806159721e-05,
      "loss": 2.3918,
      "step": 1255
    },
    {
      "epoch": 0.25094905094905096,
      "grad_norm": 1.5603059530258179,
      "learning_rate": 1.2944249503945894e-05,
      "loss": 1.4444,
      "step": 1256
    },
    {
      "epoch": 0.25114885114885116,
      "grad_norm": 2.0375001430511475,
      "learning_rate": 1.2840693213803545e-05,
      "loss": 1.7138,
      "step": 1257
    },
    {
      "epoch": 0.25134865134865136,
      "grad_norm": 1.6352300643920898,
      "learning_rate": 1.2737524396099032e-05,
      "loss": 2.0868,
      "step": 1258
    },
    {
      "epoch": 0.25154845154845157,
      "grad_norm": 3.6839704513549805,
      "learning_rate": 1.2634743509476088e-05,
      "loss": 1.6174,
      "step": 1259
    },
    {
      "epoch": 0.2517482517482518,
      "grad_norm": 2.788311719894409,
      "learning_rate": 1.2532351010853916e-05,
      "loss": 2.0182,
      "step": 1260
    },
    {
      "epoch": 0.2519480519480519,
      "grad_norm": 1.302727222442627,
      "learning_rate": 1.243034735542512e-05,
      "loss": 1.8616,
      "step": 1261
    },
    {
      "epoch": 0.2521478521478521,
      "grad_norm": 1.4308948516845703,
      "learning_rate": 1.2328732996653669e-05,
      "loss": 1.91,
      "step": 1262
    },
    {
      "epoch": 0.25234765234765233,
      "grad_norm": 3.203817367553711,
      "learning_rate": 1.2227508386272878e-05,
      "loss": 1.5516,
      "step": 1263
    },
    {
      "epoch": 0.25254745254745253,
      "grad_norm": 2.260446786880493,
      "learning_rate": 1.212667397428342e-05,
      "loss": 1.9414,
      "step": 1264
    },
    {
      "epoch": 0.25274725274725274,
      "grad_norm": 0.8816061019897461,
      "learning_rate": 1.2026230208951306e-05,
      "loss": 1.2953,
      "step": 1265
    },
    {
      "epoch": 0.25294705294705294,
      "grad_norm": 2.2827537059783936,
      "learning_rate": 1.1926177536805905e-05,
      "loss": 1.6162,
      "step": 1266
    },
    {
      "epoch": 0.25314685314685315,
      "grad_norm": 2.400808811187744,
      "learning_rate": 1.1826516402637989e-05,
      "loss": 2.0287,
      "step": 1267
    },
    {
      "epoch": 0.25334665334665335,
      "grad_norm": 2.2918334007263184,
      "learning_rate": 1.1727247249497685e-05,
      "loss": 1.9541,
      "step": 1268
    },
    {
      "epoch": 0.25354645354645355,
      "grad_norm": 2.2755861282348633,
      "learning_rate": 1.1628370518692533e-05,
      "loss": 2.5971,
      "step": 1269
    },
    {
      "epoch": 0.25374625374625376,
      "grad_norm": 2.0948331356048584,
      "learning_rate": 1.152988664978556e-05,
      "loss": 1.9792,
      "step": 1270
    },
    {
      "epoch": 0.25394605394605396,
      "grad_norm": 1.1644953489303589,
      "learning_rate": 1.1431796080593283e-05,
      "loss": 0.9605,
      "step": 1271
    },
    {
      "epoch": 0.25414585414585417,
      "grad_norm": 2.0577688217163086,
      "learning_rate": 1.1334099247183783e-05,
      "loss": 2.2076,
      "step": 1272
    },
    {
      "epoch": 0.25434565434565437,
      "grad_norm": 1.69131600856781,
      "learning_rate": 1.1236796583874787e-05,
      "loss": 2.296,
      "step": 1273
    },
    {
      "epoch": 0.2545454545454545,
      "grad_norm": 0.931521475315094,
      "learning_rate": 1.1139888523231678e-05,
      "loss": 2.4176,
      "step": 1274
    },
    {
      "epoch": 0.2547452547452547,
      "grad_norm": 2.744114398956299,
      "learning_rate": 1.1043375496065611e-05,
      "loss": 2.294,
      "step": 1275
    },
    {
      "epoch": 0.2549450549450549,
      "grad_norm": 1.786939263343811,
      "learning_rate": 1.0947257931431642e-05,
      "loss": 1.5299,
      "step": 1276
    },
    {
      "epoch": 0.25514485514485513,
      "grad_norm": 2.49094295501709,
      "learning_rate": 1.0851536256626705e-05,
      "loss": 2.1603,
      "step": 1277
    },
    {
      "epoch": 0.25534465534465534,
      "grad_norm": 3.4830093383789062,
      "learning_rate": 1.0756210897187812e-05,
      "loss": 1.2179,
      "step": 1278
    },
    {
      "epoch": 0.25554445554445554,
      "grad_norm": 2.826843023300171,
      "learning_rate": 1.0661282276890127e-05,
      "loss": 2.1316,
      "step": 1279
    },
    {
      "epoch": 0.25574425574425574,
      "grad_norm": 4.040721893310547,
      "learning_rate": 1.0566750817745074e-05,
      "loss": 1.5632,
      "step": 1280
    },
    {
      "epoch": 0.25594405594405595,
      "grad_norm": 0.9824175238609314,
      "learning_rate": 1.0472616939998492e-05,
      "loss": 2.5201,
      "step": 1281
    },
    {
      "epoch": 0.25614385614385615,
      "grad_norm": 1.5525962114334106,
      "learning_rate": 1.0378881062128731e-05,
      "loss": 1.848,
      "step": 1282
    },
    {
      "epoch": 0.25634365634365636,
      "grad_norm": 0.9476394057273865,
      "learning_rate": 1.0285543600844804e-05,
      "loss": 0.5524,
      "step": 1283
    },
    {
      "epoch": 0.25654345654345656,
      "grad_norm": 2.9578473567962646,
      "learning_rate": 1.019260497108453e-05,
      "loss": 1.4792,
      "step": 1284
    },
    {
      "epoch": 0.25674325674325676,
      "grad_norm": 3.725837230682373,
      "learning_rate": 1.010006558601274e-05,
      "loss": 2.078,
      "step": 1285
    },
    {
      "epoch": 0.25694305694305697,
      "grad_norm": 3.3834123611450195,
      "learning_rate": 1.000792585701934e-05,
      "loss": 2.016,
      "step": 1286
    },
    {
      "epoch": 0.2571428571428571,
      "grad_norm": 2.421738624572754,
      "learning_rate": 9.91618619371757e-06,
      "loss": 2.1414,
      "step": 1287
    },
    {
      "epoch": 0.2573426573426573,
      "grad_norm": 1.947953701019287,
      "learning_rate": 9.82484700394215e-06,
      "loss": 2.1062,
      "step": 1288
    },
    {
      "epoch": 0.2575424575424575,
      "grad_norm": 3.0603771209716797,
      "learning_rate": 9.73390869374743e-06,
      "loss": 1.5206,
      "step": 1289
    },
    {
      "epoch": 0.25774225774225773,
      "grad_norm": 1.848644495010376,
      "learning_rate": 9.643371667405698e-06,
      "loss": 2.1113,
      "step": 1290
    },
    {
      "epoch": 0.25794205794205793,
      "grad_norm": 2.4660403728485107,
      "learning_rate": 9.553236327405246e-06,
      "loss": 1.1462,
      "step": 1291
    },
    {
      "epoch": 0.25814185814185814,
      "grad_norm": 2.070410966873169,
      "learning_rate": 9.463503074448677e-06,
      "loss": 1.0255,
      "step": 1292
    },
    {
      "epoch": 0.25834165834165834,
      "grad_norm": 2.081519365310669,
      "learning_rate": 9.374172307451068e-06,
      "loss": 2.0029,
      "step": 1293
    },
    {
      "epoch": 0.25854145854145855,
      "grad_norm": 2.7274789810180664,
      "learning_rate": 9.285244423538197e-06,
      "loss": 2.1392,
      "step": 1294
    },
    {
      "epoch": 0.25874125874125875,
      "grad_norm": 2.6383161544799805,
      "learning_rate": 9.196719818044886e-06,
      "loss": 1.4673,
      "step": 1295
    },
    {
      "epoch": 0.25894105894105895,
      "grad_norm": 2.089522361755371,
      "learning_rate": 9.108598884513053e-06,
      "loss": 1.7534,
      "step": 1296
    },
    {
      "epoch": 0.25914085914085916,
      "grad_norm": 2.858562469482422,
      "learning_rate": 9.020882014690136e-06,
      "loss": 2.636,
      "step": 1297
    },
    {
      "epoch": 0.25934065934065936,
      "grad_norm": 1.8407063484191895,
      "learning_rate": 8.933569598527247e-06,
      "loss": 2.1835,
      "step": 1298
    },
    {
      "epoch": 0.25954045954045957,
      "grad_norm": 2.2388453483581543,
      "learning_rate": 8.846662024177477e-06,
      "loss": 1.7534,
      "step": 1299
    },
    {
      "epoch": 0.2597402597402597,
      "grad_norm": 2.081683397293091,
      "learning_rate": 8.760159677994172e-06,
      "loss": 1.7676,
      "step": 1300
    },
    {
      "epoch": 0.2599400599400599,
      "grad_norm": 1.5982688665390015,
      "learning_rate": 8.674062944529216e-06,
      "loss": 1.7244,
      "step": 1301
    },
    {
      "epoch": 0.2601398601398601,
      "grad_norm": 3.0909926891326904,
      "learning_rate": 8.588372206531292e-06,
      "loss": 2.1874,
      "step": 1302
    },
    {
      "epoch": 0.26033966033966033,
      "grad_norm": 2.329345703125,
      "learning_rate": 8.503087844944213e-06,
      "loss": 2.3697,
      "step": 1303
    },
    {
      "epoch": 0.26053946053946053,
      "grad_norm": 1.3546773195266724,
      "learning_rate": 8.418210238905256e-06,
      "loss": 2.0347,
      "step": 1304
    },
    {
      "epoch": 0.26073926073926074,
      "grad_norm": 1.8661444187164307,
      "learning_rate": 8.333739765743398e-06,
      "loss": 2.0956,
      "step": 1305
    },
    {
      "epoch": 0.26093906093906094,
      "grad_norm": 2.413288116455078,
      "learning_rate": 8.249676800977658e-06,
      "loss": 1.8304,
      "step": 1306
    },
    {
      "epoch": 0.26113886113886114,
      "grad_norm": 3.680356979370117,
      "learning_rate": 8.16602171831553e-06,
      "loss": 1.7851,
      "step": 1307
    },
    {
      "epoch": 0.26133866133866135,
      "grad_norm": 1.1345601081848145,
      "learning_rate": 8.082774889651168e-06,
      "loss": 0.5807,
      "step": 1308
    },
    {
      "epoch": 0.26153846153846155,
      "grad_norm": 4.087169170379639,
      "learning_rate": 7.999936685063835e-06,
      "loss": 1.6081,
      "step": 1309
    },
    {
      "epoch": 0.26173826173826176,
      "grad_norm": 1.6901847124099731,
      "learning_rate": 7.91750747281621e-06,
      "loss": 1.5404,
      "step": 1310
    },
    {
      "epoch": 0.26193806193806196,
      "grad_norm": 2.7638628482818604,
      "learning_rate": 7.835487619352811e-06,
      "loss": 1.3727,
      "step": 1311
    },
    {
      "epoch": 0.26213786213786217,
      "grad_norm": 1.9739452600479126,
      "learning_rate": 7.753877489298244e-06,
      "loss": 1.2838,
      "step": 1312
    },
    {
      "epoch": 0.2623376623376623,
      "grad_norm": 3.3962326049804688,
      "learning_rate": 7.67267744545579e-06,
      "loss": 1.8961,
      "step": 1313
    },
    {
      "epoch": 0.2625374625374625,
      "grad_norm": 2.853637933731079,
      "learning_rate": 7.591887848805545e-06,
      "loss": 1.2217,
      "step": 1314
    },
    {
      "epoch": 0.2627372627372627,
      "grad_norm": 3.4392106533050537,
      "learning_rate": 7.5115090585029966e-06,
      "loss": 2.1989,
      "step": 1315
    },
    {
      "epoch": 0.2629370629370629,
      "grad_norm": 0.8740386366844177,
      "learning_rate": 7.431541431877342e-06,
      "loss": 0.928,
      "step": 1316
    },
    {
      "epoch": 0.26313686313686313,
      "grad_norm": 3.347041368484497,
      "learning_rate": 7.351985324429933e-06,
      "loss": 2.2959,
      "step": 1317
    },
    {
      "epoch": 0.26333666333666333,
      "grad_norm": 1.3547163009643555,
      "learning_rate": 7.272841089832694e-06,
      "loss": 1.9116,
      "step": 1318
    },
    {
      "epoch": 0.26353646353646354,
      "grad_norm": 3.4543821811676025,
      "learning_rate": 7.194109079926514e-06,
      "loss": 1.3869,
      "step": 1319
    },
    {
      "epoch": 0.26373626373626374,
      "grad_norm": 1.8126962184906006,
      "learning_rate": 7.115789644719728e-06,
      "loss": 1.6983,
      "step": 1320
    },
    {
      "epoch": 0.26393606393606395,
      "grad_norm": 0.9362018704414368,
      "learning_rate": 7.037883132386547e-06,
      "loss": 0.7012,
      "step": 1321
    },
    {
      "epoch": 0.26413586413586415,
      "grad_norm": 2.6125810146331787,
      "learning_rate": 6.960389889265517e-06,
      "loss": 1.8272,
      "step": 1322
    },
    {
      "epoch": 0.26433566433566436,
      "grad_norm": 1.98081374168396,
      "learning_rate": 6.883310259857944e-06,
      "loss": 1.7277,
      "step": 1323
    },
    {
      "epoch": 0.26453546453546456,
      "grad_norm": 1.7236369848251343,
      "learning_rate": 6.806644586826383e-06,
      "loss": 2.1807,
      "step": 1324
    },
    {
      "epoch": 0.2647352647352647,
      "grad_norm": 4.869028568267822,
      "learning_rate": 6.730393210993147e-06,
      "loss": 1.8986,
      "step": 1325
    },
    {
      "epoch": 0.2649350649350649,
      "grad_norm": 1.9536770582199097,
      "learning_rate": 6.654556471338746e-06,
      "loss": 2.1679,
      "step": 1326
    },
    {
      "epoch": 0.2651348651348651,
      "grad_norm": 2.2516820430755615,
      "learning_rate": 6.579134705000412e-06,
      "loss": 1.7203,
      "step": 1327
    },
    {
      "epoch": 0.2653346653346653,
      "grad_norm": 2.0443434715270996,
      "learning_rate": 6.504128247270546e-06,
      "loss": 1.3692,
      "step": 1328
    },
    {
      "epoch": 0.2655344655344655,
      "grad_norm": 2.5579569339752197,
      "learning_rate": 6.429537431595312e-06,
      "loss": 1.223,
      "step": 1329
    },
    {
      "epoch": 0.26573426573426573,
      "grad_norm": 3.26993465423584,
      "learning_rate": 6.355362589573077e-06,
      "loss": 2.2871,
      "step": 1330
    },
    {
      "epoch": 0.26593406593406593,
      "grad_norm": 0.905366837978363,
      "learning_rate": 6.2816040509530165e-06,
      "loss": 1.4402,
      "step": 1331
    },
    {
      "epoch": 0.26613386613386614,
      "grad_norm": 2.851557731628418,
      "learning_rate": 6.2082621436335475e-06,
      "loss": 2.6764,
      "step": 1332
    },
    {
      "epoch": 0.26633366633366634,
      "grad_norm": 1.4785799980163574,
      "learning_rate": 6.135337193660962e-06,
      "loss": 2.6353,
      "step": 1333
    },
    {
      "epoch": 0.26653346653346655,
      "grad_norm": 3.1786277294158936,
      "learning_rate": 6.062829525227909e-06,
      "loss": 1.6176,
      "step": 1334
    },
    {
      "epoch": 0.26673326673326675,
      "grad_norm": 3.2823326587677,
      "learning_rate": 5.990739460672024e-06,
      "loss": 1.7004,
      "step": 1335
    },
    {
      "epoch": 0.26693306693306695,
      "grad_norm": 4.533176422119141,
      "learning_rate": 5.9190673204744255e-06,
      "loss": 2.4638,
      "step": 1336
    },
    {
      "epoch": 0.26713286713286716,
      "grad_norm": 1.3544318675994873,
      "learning_rate": 5.84781342325833e-06,
      "loss": 1.8745,
      "step": 1337
    },
    {
      "epoch": 0.2673326673326673,
      "grad_norm": 2.2955806255340576,
      "learning_rate": 5.77697808578761e-06,
      "loss": 2.2414,
      "step": 1338
    },
    {
      "epoch": 0.2675324675324675,
      "grad_norm": 3.539999008178711,
      "learning_rate": 5.706561622965467e-06,
      "loss": 1.4898,
      "step": 1339
    },
    {
      "epoch": 0.2677322677322677,
      "grad_norm": 0.6196832060813904,
      "learning_rate": 5.636564347832907e-06,
      "loss": 1.5225,
      "step": 1340
    },
    {
      "epoch": 0.2679320679320679,
      "grad_norm": 2.341278076171875,
      "learning_rate": 5.566986571567401e-06,
      "loss": 1.5889,
      "step": 1341
    },
    {
      "epoch": 0.2681318681318681,
      "grad_norm": 2.839869499206543,
      "learning_rate": 5.497828603481569e-06,
      "loss": 1.4987,
      "step": 1342
    },
    {
      "epoch": 0.2683316683316683,
      "grad_norm": 1.3282711505889893,
      "learning_rate": 5.429090751021704e-06,
      "loss": 0.9929,
      "step": 1343
    },
    {
      "epoch": 0.26853146853146853,
      "grad_norm": 3.2844274044036865,
      "learning_rate": 5.3607733197664436e-06,
      "loss": 1.7294,
      "step": 1344
    },
    {
      "epoch": 0.26873126873126874,
      "grad_norm": 3.284437894821167,
      "learning_rate": 5.2928766134254345e-06,
      "loss": 1.3519,
      "step": 1345
    },
    {
      "epoch": 0.26893106893106894,
      "grad_norm": 1.025913953781128,
      "learning_rate": 5.225400933837954e-06,
      "loss": 2.2182,
      "step": 1346
    },
    {
      "epoch": 0.26913086913086914,
      "grad_norm": 1.518723964691162,
      "learning_rate": 5.158346580971573e-06,
      "loss": 0.9363,
      "step": 1347
    },
    {
      "epoch": 0.26933066933066935,
      "grad_norm": 1.851684331893921,
      "learning_rate": 5.091713852920854e-06,
      "loss": 2.5102,
      "step": 1348
    },
    {
      "epoch": 0.26953046953046955,
      "grad_norm": 2.563770055770874,
      "learning_rate": 5.025503045905933e-06,
      "loss": 2.0675,
      "step": 1349
    },
    {
      "epoch": 0.26973026973026976,
      "grad_norm": 1.3160310983657837,
      "learning_rate": 4.959714454271369e-06,
      "loss": 2.275,
      "step": 1350
    },
    {
      "epoch": 0.2699300699300699,
      "grad_norm": 2.4515297412872314,
      "learning_rate": 4.8943483704846475e-06,
      "loss": 2.1559,
      "step": 1351
    },
    {
      "epoch": 0.2701298701298701,
      "grad_norm": 3.2289884090423584,
      "learning_rate": 4.829405085134997e-06,
      "loss": 2.7193,
      "step": 1352
    },
    {
      "epoch": 0.2703296703296703,
      "grad_norm": 2.6481146812438965,
      "learning_rate": 4.764884886932086e-06,
      "loss": 1.4372,
      "step": 1353
    },
    {
      "epoch": 0.2705294705294705,
      "grad_norm": 1.7307050228118896,
      "learning_rate": 4.700788062704687e-06,
      "loss": 1.8172,
      "step": 1354
    },
    {
      "epoch": 0.2707292707292707,
      "grad_norm": 1.7471261024475098,
      "learning_rate": 4.6371148973994525e-06,
      "loss": 0.6416,
      "step": 1355
    },
    {
      "epoch": 0.2709290709290709,
      "grad_norm": 2.2157058715820312,
      "learning_rate": 4.573865674079625e-06,
      "loss": 2.0902,
      "step": 1356
    },
    {
      "epoch": 0.27112887112887113,
      "grad_norm": 3.4105029106140137,
      "learning_rate": 4.511040673923828e-06,
      "loss": 1.8246,
      "step": 1357
    },
    {
      "epoch": 0.27132867132867133,
      "grad_norm": 1.661271572113037,
      "learning_rate": 4.448640176224694e-06,
      "loss": 1.9919,
      "step": 1358
    },
    {
      "epoch": 0.27152847152847154,
      "grad_norm": 2.2584166526794434,
      "learning_rate": 4.386664458387779e-06,
      "loss": 2.5084,
      "step": 1359
    },
    {
      "epoch": 0.27172827172827174,
      "grad_norm": 3.6093215942382812,
      "learning_rate": 4.325113795930203e-06,
      "loss": 1.774,
      "step": 1360
    },
    {
      "epoch": 0.27192807192807195,
      "grad_norm": 1.6291857957839966,
      "learning_rate": 4.263988462479484e-06,
      "loss": 1.8534,
      "step": 1361
    },
    {
      "epoch": 0.27212787212787215,
      "grad_norm": 2.7186968326568604,
      "learning_rate": 4.203288729772326e-06,
      "loss": 1.4413,
      "step": 1362
    },
    {
      "epoch": 0.27232767232767235,
      "grad_norm": 1.776559829711914,
      "learning_rate": 4.143014867653383e-06,
      "loss": 2.1646,
      "step": 1363
    },
    {
      "epoch": 0.2725274725274725,
      "grad_norm": 1.9078245162963867,
      "learning_rate": 4.083167144074073e-06,
      "loss": 2.4272,
      "step": 1364
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 2.954379081726074,
      "learning_rate": 4.023745825091407e-06,
      "loss": 1.9718,
      "step": 1365
    },
    {
      "epoch": 0.2729270729270729,
      "grad_norm": 3.34550142288208,
      "learning_rate": 3.964751174866765e-06,
      "loss": 2.0609,
      "step": 1366
    },
    {
      "epoch": 0.2731268731268731,
      "grad_norm": 2.446174383163452,
      "learning_rate": 3.906183455664725e-06,
      "loss": 2.0143,
      "step": 1367
    },
    {
      "epoch": 0.2733266733266733,
      "grad_norm": 2.4035749435424805,
      "learning_rate": 3.84804292785198e-06,
      "loss": 2.3776,
      "step": 1368
    },
    {
      "epoch": 0.2735264735264735,
      "grad_norm": 2.542775869369507,
      "learning_rate": 3.7903298498960572e-06,
      "loss": 2.1156,
      "step": 1369
    },
    {
      "epoch": 0.27372627372627373,
      "grad_norm": 2.3180062770843506,
      "learning_rate": 3.7330444783642338e-06,
      "loss": 1.7976,
      "step": 1370
    },
    {
      "epoch": 0.27392607392607393,
      "grad_norm": 3.594348430633545,
      "learning_rate": 3.676187067922421e-06,
      "loss": 2.2641,
      "step": 1371
    },
    {
      "epoch": 0.27412587412587414,
      "grad_norm": 1.0998749732971191,
      "learning_rate": 3.619757871333973e-06,
      "loss": 1.7389,
      "step": 1372
    },
    {
      "epoch": 0.27432567432567434,
      "grad_norm": 0.9753875136375427,
      "learning_rate": 3.563757139458579e-06,
      "loss": 0.6612,
      "step": 1373
    },
    {
      "epoch": 0.27452547452547454,
      "grad_norm": 2.0982937812805176,
      "learning_rate": 3.5081851212512175e-06,
      "loss": 1.6762,
      "step": 1374
    },
    {
      "epoch": 0.27472527472527475,
      "grad_norm": 2.134904623031616,
      "learning_rate": 3.4530420637609363e-06,
      "loss": 1.9316,
      "step": 1375
    },
    {
      "epoch": 0.27492507492507495,
      "grad_norm": 2.623469591140747,
      "learning_rate": 3.3983282121298086e-06,
      "loss": 1.7784,
      "step": 1376
    },
    {
      "epoch": 0.2751248751248751,
      "grad_norm": 2.5506207942962646,
      "learning_rate": 3.3440438095919126e-06,
      "loss": 1.527,
      "step": 1377
    },
    {
      "epoch": 0.2753246753246753,
      "grad_norm": 4.523069858551025,
      "learning_rate": 3.290189097472096e-06,
      "loss": 2.4427,
      "step": 1378
    },
    {
      "epoch": 0.2755244755244755,
      "grad_norm": 2.0411832332611084,
      "learning_rate": 3.236764315185037e-06,
      "loss": 2.4423,
      "step": 1379
    },
    {
      "epoch": 0.2757242757242757,
      "grad_norm": 2.3199424743652344,
      "learning_rate": 3.1837697002341293e-06,
      "loss": 1.4518,
      "step": 1380
    },
    {
      "epoch": 0.2759240759240759,
      "grad_norm": 1.213091254234314,
      "learning_rate": 3.131205488210409e-06,
      "loss": 0.8921,
      "step": 1381
    },
    {
      "epoch": 0.2761238761238761,
      "grad_norm": 2.2103261947631836,
      "learning_rate": 3.0790719127915646e-06,
      "loss": 1.9313,
      "step": 1382
    },
    {
      "epoch": 0.2763236763236763,
      "grad_norm": 1.526326060295105,
      "learning_rate": 3.0273692057408265e-06,
      "loss": 2.6484,
      "step": 1383
    },
    {
      "epoch": 0.27652347652347653,
      "grad_norm": 4.652854919433594,
      "learning_rate": 2.976097596905969e-06,
      "loss": 2.3698,
      "step": 1384
    },
    {
      "epoch": 0.27672327672327673,
      "grad_norm": 3.125795841217041,
      "learning_rate": 2.9252573142183326e-06,
      "loss": 1.7599,
      "step": 1385
    },
    {
      "epoch": 0.27692307692307694,
      "grad_norm": 2.8028552532196045,
      "learning_rate": 2.874848583691714e-06,
      "loss": 2.3416,
      "step": 1386
    },
    {
      "epoch": 0.27712287712287714,
      "grad_norm": 2.5300536155700684,
      "learning_rate": 2.8248716294214774e-06,
      "loss": 1.6237,
      "step": 1387
    },
    {
      "epoch": 0.27732267732267735,
      "grad_norm": 2.9590065479278564,
      "learning_rate": 2.7753266735834338e-06,
      "loss": 1.2637,
      "step": 1388
    },
    {
      "epoch": 0.2775224775224775,
      "grad_norm": 1.386031985282898,
      "learning_rate": 2.7262139364329643e-06,
      "loss": 1.9034,
      "step": 1389
    },
    {
      "epoch": 0.2777222777222777,
      "grad_norm": 2.2058627605438232,
      "learning_rate": 2.677533636303964e-06,
      "loss": 2.2217,
      "step": 1390
    },
    {
      "epoch": 0.2779220779220779,
      "grad_norm": 2.222290515899658,
      "learning_rate": 2.6292859896079213e-06,
      "loss": 1.8867,
      "step": 1391
    },
    {
      "epoch": 0.2781218781218781,
      "grad_norm": 3.3234455585479736,
      "learning_rate": 2.581471210832931e-06,
      "loss": 2.1031,
      "step": 1392
    },
    {
      "epoch": 0.2783216783216783,
      "grad_norm": 3.229383945465088,
      "learning_rate": 2.5340895125427364e-06,
      "loss": 2.2038,
      "step": 1393
    },
    {
      "epoch": 0.2785214785214785,
      "grad_norm": 2.6980013847351074,
      "learning_rate": 2.4871411053757898e-06,
      "loss": 2.1244,
      "step": 1394
    },
    {
      "epoch": 0.2787212787212787,
      "grad_norm": 3.165888547897339,
      "learning_rate": 2.440626198044327e-06,
      "loss": 2.2845,
      "step": 1395
    },
    {
      "epoch": 0.2789210789210789,
      "grad_norm": 1.5466876029968262,
      "learning_rate": 2.394544997333437e-06,
      "loss": 1.4165,
      "step": 1396
    },
    {
      "epoch": 0.27912087912087913,
      "grad_norm": 3.3853037357330322,
      "learning_rate": 2.3488977081001394e-06,
      "loss": 2.1611,
      "step": 1397
    },
    {
      "epoch": 0.27932067932067933,
      "grad_norm": 1.183717966079712,
      "learning_rate": 2.3036845332724543e-06,
      "loss": 2.5048,
      "step": 1398
    },
    {
      "epoch": 0.27952047952047954,
      "grad_norm": 1.6473960876464844,
      "learning_rate": 2.2589056738485324e-06,
      "loss": 1.5302,
      "step": 1399
    },
    {
      "epoch": 0.27972027972027974,
      "grad_norm": 2.360764741897583,
      "learning_rate": 2.2145613288957478e-06,
      "loss": 1.0742,
      "step": 1400
    },
    {
      "epoch": 0.27992007992007994,
      "grad_norm": 2.673153877258301,
      "learning_rate": 2.170651695549786e-06,
      "loss": 0.9324,
      "step": 1401
    },
    {
      "epoch": 0.2801198801198801,
      "grad_norm": 0.7940964698791504,
      "learning_rate": 2.1271769690138332e-06,
      "loss": 0.6881,
      "step": 1402
    },
    {
      "epoch": 0.2803196803196803,
      "grad_norm": 3.0873773097991943,
      "learning_rate": 2.084137342557646e-06,
      "loss": 1.9818,
      "step": 1403
    },
    {
      "epoch": 0.2805194805194805,
      "grad_norm": 0.8106163740158081,
      "learning_rate": 2.0415330075166937e-06,
      "loss": 1.0546,
      "step": 1404
    },
    {
      "epoch": 0.2807192807192807,
      "grad_norm": 1.07546865940094,
      "learning_rate": 1.9993641532913833e-06,
      "loss": 0.852,
      "step": 1405
    },
    {
      "epoch": 0.2809190809190809,
      "grad_norm": 3.639585494995117,
      "learning_rate": 1.9576309673461357e-06,
      "loss": 1.8313,
      "step": 1406
    },
    {
      "epoch": 0.2811188811188811,
      "grad_norm": 1.9846417903900146,
      "learning_rate": 1.916333635208556e-06,
      "loss": 0.9261,
      "step": 1407
    },
    {
      "epoch": 0.2813186813186813,
      "grad_norm": 1.1532965898513794,
      "learning_rate": 1.8754723404686425e-06,
      "loss": 0.9697,
      "step": 1408
    },
    {
      "epoch": 0.2815184815184815,
      "grad_norm": 2.3780477046966553,
      "learning_rate": 1.8350472647780116e-06,
      "loss": 2.3483,
      "step": 1409
    },
    {
      "epoch": 0.2817182817182817,
      "grad_norm": 0.9089410901069641,
      "learning_rate": 1.7950585878489856e-06,
      "loss": 1.3832,
      "step": 1410
    },
    {
      "epoch": 0.28191808191808193,
      "grad_norm": 3.343585729598999,
      "learning_rate": 1.7555064874538397e-06,
      "loss": 1.3732,
      "step": 1411
    },
    {
      "epoch": 0.28211788211788213,
      "grad_norm": 2.4793474674224854,
      "learning_rate": 1.7163911394240672e-06,
      "loss": 1.9582,
      "step": 1412
    },
    {
      "epoch": 0.28231768231768234,
      "grad_norm": 0.8928383588790894,
      "learning_rate": 1.6777127176495043e-06,
      "loss": 0.3872,
      "step": 1413
    },
    {
      "epoch": 0.28251748251748254,
      "grad_norm": 1.337222933769226,
      "learning_rate": 1.6394713940776296e-06,
      "loss": 1.3356,
      "step": 1414
    },
    {
      "epoch": 0.2827172827172827,
      "grad_norm": 2.4888858795166016,
      "learning_rate": 1.6016673387127646e-06,
      "loss": 1.7655,
      "step": 1415
    },
    {
      "epoch": 0.2829170829170829,
      "grad_norm": 2.25498104095459,
      "learning_rate": 1.5643007196153302e-06,
      "loss": 0.9709,
      "step": 1416
    },
    {
      "epoch": 0.2831168831168831,
      "grad_norm": 1.0651980638504028,
      "learning_rate": 1.5273717029010925e-06,
      "loss": 1.0313,
      "step": 1417
    },
    {
      "epoch": 0.2833166833166833,
      "grad_norm": 1.4574512243270874,
      "learning_rate": 1.4908804527404286e-06,
      "loss": 1.5113,
      "step": 1418
    },
    {
      "epoch": 0.2835164835164835,
      "grad_norm": 2.873931407928467,
      "learning_rate": 1.4548271313575835e-06,
      "loss": 1.5666,
      "step": 1419
    },
    {
      "epoch": 0.2837162837162837,
      "grad_norm": 1.449967622756958,
      "learning_rate": 1.4192118990299707e-06,
      "loss": 1.8286,
      "step": 1420
    },
    {
      "epoch": 0.2839160839160839,
      "grad_norm": 0.8263225555419922,
      "learning_rate": 1.3840349140874619e-06,
      "loss": 0.4797,
      "step": 1421
    },
    {
      "epoch": 0.2841158841158841,
      "grad_norm": 2.279428720474243,
      "learning_rate": 1.3492963329116537e-06,
      "loss": 2.0568,
      "step": 1422
    },
    {
      "epoch": 0.2843156843156843,
      "grad_norm": 2.3729288578033447,
      "learning_rate": 1.3149963099352014e-06,
      "loss": 2.0375,
      "step": 1423
    },
    {
      "epoch": 0.28451548451548453,
      "grad_norm": 2.2034716606140137,
      "learning_rate": 1.2811349976411202e-06,
      "loss": 1.5301,
      "step": 1424
    },
    {
      "epoch": 0.28471528471528473,
      "grad_norm": 2.6854982376098633,
      "learning_rate": 1.2477125465620853e-06,
      "loss": 1.552,
      "step": 1425
    },
    {
      "epoch": 0.28491508491508494,
      "grad_norm": 2.346436023712158,
      "learning_rate": 1.2147291052798216e-06,
      "loss": 1.5142,
      "step": 1426
    },
    {
      "epoch": 0.28511488511488514,
      "grad_norm": 1.835170865058899,
      "learning_rate": 1.1821848204243814e-06,
      "loss": 1.8177,
      "step": 1427
    },
    {
      "epoch": 0.2853146853146853,
      "grad_norm": 1.6144003868103027,
      "learning_rate": 1.1500798366735233e-06,
      "loss": 1.3589,
      "step": 1428
    },
    {
      "epoch": 0.2855144855144855,
      "grad_norm": 0.9738304018974304,
      "learning_rate": 1.1184142967520794e-06,
      "loss": 0.8807,
      "step": 1429
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 2.713076591491699,
      "learning_rate": 1.0871883414312777e-06,
      "loss": 1.4394,
      "step": 1430
    },
    {
      "epoch": 0.2859140859140859,
      "grad_norm": 2.2603237628936768,
      "learning_rate": 1.0564021095281652e-06,
      "loss": 1.8323,
      "step": 1431
    },
    {
      "epoch": 0.2861138861138861,
      "grad_norm": 2.9957311153411865,
      "learning_rate": 1.0260557379049519e-06,
      "loss": 1.5152,
      "step": 1432
    },
    {
      "epoch": 0.2863136863136863,
      "grad_norm": 1.4863415956497192,
      "learning_rate": 9.96149361468457e-07,
      "loss": 1.3349,
      "step": 1433
    },
    {
      "epoch": 0.2865134865134865,
      "grad_norm": 1.7435582876205444,
      "learning_rate": 9.66683113169431e-07,
      "loss": 1.9929,
      "step": 1434
    },
    {
      "epoch": 0.2867132867132867,
      "grad_norm": 1.1338272094726562,
      "learning_rate": 9.376571240020227e-07,
      "loss": 1.2289,
      "step": 1435
    },
    {
      "epoch": 0.2869130869130869,
      "grad_norm": 1.5133880376815796,
      "learning_rate": 9.090715230031688e-07,
      "loss": 1.1967,
      "step": 1436
    },
    {
      "epoch": 0.2871128871128871,
      "grad_norm": 2.0323219299316406,
      "learning_rate": 8.809264372520609e-07,
      "loss": 2.1973,
      "step": 1437
    },
    {
      "epoch": 0.28731268731268733,
      "grad_norm": 1.3181827068328857,
      "learning_rate": 8.532219918695128e-07,
      "loss": 1.067,
      "step": 1438
    },
    {
      "epoch": 0.28751248751248754,
      "grad_norm": 4.754121780395508,
      "learning_rate": 8.259583100174606e-07,
      "loss": 1.7461,
      "step": 1439
    },
    {
      "epoch": 0.28771228771228774,
      "grad_norm": 2.407866954803467,
      "learning_rate": 7.991355128984079e-07,
      "loss": 1.6676,
      "step": 1440
    },
    {
      "epoch": 0.2879120879120879,
      "grad_norm": 3.4123401641845703,
      "learning_rate": 7.727537197548707e-07,
      "loss": 1.2812,
      "step": 1441
    },
    {
      "epoch": 0.2881118881118881,
      "grad_norm": 1.7936171293258667,
      "learning_rate": 7.468130478688218e-07,
      "loss": 2.5672,
      "step": 1442
    },
    {
      "epoch": 0.2883116883116883,
      "grad_norm": 2.5005099773406982,
      "learning_rate": 7.213136125612586e-07,
      "loss": 2.0952,
      "step": 1443
    },
    {
      "epoch": 0.2885114885114885,
      "grad_norm": 1.5314844846725464,
      "learning_rate": 6.962555271915805e-07,
      "loss": 1.4739,
      "step": 1444
    },
    {
      "epoch": 0.2887112887112887,
      "grad_norm": 1.0852973461151123,
      "learning_rate": 6.716389031571568e-07,
      "loss": 0.9738,
      "step": 1445
    },
    {
      "epoch": 0.2889110889110889,
      "grad_norm": 1.1234794855117798,
      "learning_rate": 6.474638498928265e-07,
      "loss": 3.0619,
      "step": 1446
    },
    {
      "epoch": 0.2891108891108891,
      "grad_norm": 2.889023542404175,
      "learning_rate": 6.237304748703543e-07,
      "loss": 2.2604,
      "step": 1447
    },
    {
      "epoch": 0.2893106893106893,
      "grad_norm": 2.0557050704956055,
      "learning_rate": 6.004388835980423e-07,
      "loss": 2.5481,
      "step": 1448
    },
    {
      "epoch": 0.2895104895104895,
      "grad_norm": 1.2526904344558716,
      "learning_rate": 5.77589179620186e-07,
      "loss": 1.1593,
      "step": 1449
    },
    {
      "epoch": 0.2897102897102897,
      "grad_norm": 1.1131049394607544,
      "learning_rate": 5.55181464516652e-07,
      "loss": 0.7075,
      "step": 1450
    },
    {
      "epoch": 0.28991008991008993,
      "grad_norm": 2.641251564025879,
      "learning_rate": 5.332158379024122e-07,
      "loss": 2.1809,
      "step": 1451
    },
    {
      "epoch": 0.29010989010989013,
      "grad_norm": 2.545504093170166,
      "learning_rate": 5.116923974270993e-07,
      "loss": 1.888,
      "step": 1452
    },
    {
      "epoch": 0.2903096903096903,
      "grad_norm": 2.052154302597046,
      "learning_rate": 4.906112387745965e-07,
      "loss": 1.5378,
      "step": 1453
    },
    {
      "epoch": 0.2905094905094905,
      "grad_norm": 1.9360885620117188,
      "learning_rate": 4.6997245566257064e-07,
      "loss": 2.423,
      "step": 1454
    },
    {
      "epoch": 0.2907092907092907,
      "grad_norm": 1.0807054042816162,
      "learning_rate": 4.497761398421063e-07,
      "loss": 0.6,
      "step": 1455
    },
    {
      "epoch": 0.2909090909090909,
      "grad_norm": 2.006211042404175,
      "learning_rate": 4.3002238109723927e-07,
      "loss": 1.9224,
      "step": 1456
    },
    {
      "epoch": 0.2911088911088911,
      "grad_norm": 1.7676888704299927,
      "learning_rate": 4.107112672446123e-07,
      "loss": 1.6368,
      "step": 1457
    },
    {
      "epoch": 0.2913086913086913,
      "grad_norm": 3.1097540855407715,
      "learning_rate": 3.9184288413306456e-07,
      "loss": 1.6453,
      "step": 1458
    },
    {
      "epoch": 0.2915084915084915,
      "grad_norm": 2.098031520843506,
      "learning_rate": 3.734173156432208e-07,
      "loss": 1.7041,
      "step": 1459
    },
    {
      "epoch": 0.2917082917082917,
      "grad_norm": 1.3085335493087769,
      "learning_rate": 3.554346436871581e-07,
      "loss": 1.2218,
      "step": 1460
    },
    {
      "epoch": 0.2919080919080919,
      "grad_norm": 4.150809288024902,
      "learning_rate": 3.3789494820803957e-07,
      "loss": 2.6837,
      "step": 1461
    },
    {
      "epoch": 0.2921078921078921,
      "grad_norm": 1.909101963043213,
      "learning_rate": 3.2079830717972606e-07,
      "loss": 1.5835,
      "step": 1462
    },
    {
      "epoch": 0.2923076923076923,
      "grad_norm": 2.7022693157196045,
      "learning_rate": 3.041447966064648e-07,
      "loss": 1.8666,
      "step": 1463
    },
    {
      "epoch": 0.29250749250749253,
      "grad_norm": 2.2920103073120117,
      "learning_rate": 2.8793449052254563e-07,
      "loss": 1.2492,
      "step": 1464
    },
    {
      "epoch": 0.29270729270729273,
      "grad_norm": 2.985042095184326,
      "learning_rate": 2.721674609919345e-07,
      "loss": 1.5306,
      "step": 1465
    },
    {
      "epoch": 0.2929070929070929,
      "grad_norm": 3.8514034748077393,
      "learning_rate": 2.568437781080069e-07,
      "loss": 1.5403,
      "step": 1466
    },
    {
      "epoch": 0.2931068931068931,
      "grad_norm": 5.368758678436279,
      "learning_rate": 2.4196350999320384e-07,
      "loss": 0.9026,
      "step": 1467
    },
    {
      "epoch": 0.2933066933066933,
      "grad_norm": 2.310803174972534,
      "learning_rate": 2.275267227987321e-07,
      "loss": 1.3887,
      "step": 1468
    },
    {
      "epoch": 0.2935064935064935,
      "grad_norm": 2.153717517852783,
      "learning_rate": 2.135334807042866e-07,
      "loss": 1.4491,
      "step": 1469
    },
    {
      "epoch": 0.2937062937062937,
      "grad_norm": 4.270534992218018,
      "learning_rate": 1.9998384591773944e-07,
      "loss": 1.7351,
      "step": 1470
    },
    {
      "epoch": 0.2939060939060939,
      "grad_norm": 2.4280519485473633,
      "learning_rate": 1.8687787867489592e-07,
      "loss": 1.6183,
      "step": 1471
    },
    {
      "epoch": 0.2941058941058941,
      "grad_norm": 1.6690927743911743,
      "learning_rate": 1.7421563723919454e-07,
      "loss": 2.2096,
      "step": 1472
    },
    {
      "epoch": 0.2943056943056943,
      "grad_norm": 1.5953704118728638,
      "learning_rate": 1.6199717790145174e-07,
      "loss": 1.9958,
      "step": 1473
    },
    {
      "epoch": 0.2945054945054945,
      "grad_norm": 1.1822435855865479,
      "learning_rate": 1.5022255497962879e-07,
      "loss": 2.7805,
      "step": 1474
    },
    {
      "epoch": 0.2947052947052947,
      "grad_norm": 1.871971607208252,
      "learning_rate": 1.3889182081860962e-07,
      "loss": 0.8616,
      "step": 1475
    },
    {
      "epoch": 0.2949050949050949,
      "grad_norm": 2.493915557861328,
      "learning_rate": 1.2800502578991235e-07,
      "loss": 2.4126,
      "step": 1476
    },
    {
      "epoch": 0.2951048951048951,
      "grad_norm": 2.5208139419555664,
      "learning_rate": 1.1756221829148928e-07,
      "loss": 1.293,
      "step": 1477
    },
    {
      "epoch": 0.29530469530469533,
      "grad_norm": 1.7042564153671265,
      "learning_rate": 1.0756344474753821e-07,
      "loss": 2.3608,
      "step": 1478
    },
    {
      "epoch": 0.2955044955044955,
      "grad_norm": 3.2595224380493164,
      "learning_rate": 9.800874960826933e-08,
      "loss": 2.197,
      "step": 1479
    },
    {
      "epoch": 0.2957042957042957,
      "grad_norm": 2.5350639820098877,
      "learning_rate": 8.889817534969425e-08,
      "loss": 1.8273,
      "step": 1480
    },
    {
      "epoch": 0.2959040959040959,
      "grad_norm": 2.675603151321411,
      "learning_rate": 8.023176247348163e-08,
      "loss": 1.0578,
      "step": 1481
    },
    {
      "epoch": 0.2961038961038961,
      "grad_norm": 1.3462543487548828,
      "learning_rate": 7.200954950673522e-08,
      "loss": 1.8099,
      "step": 1482
    },
    {
      "epoch": 0.2963036963036963,
      "grad_norm": 3.072120428085327,
      "learning_rate": 6.423157300184946e-08,
      "loss": 1.5421,
      "step": 1483
    },
    {
      "epoch": 0.2965034965034965,
      "grad_norm": 0.7590976357460022,
      "learning_rate": 5.6897867536331864e-08,
      "loss": 0.5953,
      "step": 1484
    },
    {
      "epoch": 0.2967032967032967,
      "grad_norm": 2.021287679672241,
      "learning_rate": 5.000846571264761e-08,
      "loss": 1.9083,
      "step": 1485
    },
    {
      "epoch": 0.2969030969030969,
      "grad_norm": 1.2576559782028198,
      "learning_rate": 4.35633981580974e-08,
      "loss": 1.4796,
      "step": 1486
    },
    {
      "epoch": 0.2971028971028971,
      "grad_norm": 2.597594976425171,
      "learning_rate": 3.756269352462871e-08,
      "loss": 2.1462,
      "step": 1487
    },
    {
      "epoch": 0.2973026973026973,
      "grad_norm": 2.201155424118042,
      "learning_rate": 3.20063784888025e-08,
      "loss": 1.5498,
      "step": 1488
    },
    {
      "epoch": 0.2975024975024975,
      "grad_norm": 2.3125736713409424,
      "learning_rate": 2.6894477751548964e-08,
      "loss": 1.6882,
      "step": 1489
    },
    {
      "epoch": 0.2977022977022977,
      "grad_norm": 3.154930591583252,
      "learning_rate": 2.222701403818972e-08,
      "loss": 1.9849,
      "step": 1490
    },
    {
      "epoch": 0.29790209790209793,
      "grad_norm": 2.069775104522705,
      "learning_rate": 1.8004008098226887e-08,
      "loss": 1.5192,
      "step": 1491
    },
    {
      "epoch": 0.2981018981018981,
      "grad_norm": 3.4867522716522217,
      "learning_rate": 1.4225478705309769e-08,
      "loss": 1.6957,
      "step": 1492
    },
    {
      "epoch": 0.2983016983016983,
      "grad_norm": 2.8967080116271973,
      "learning_rate": 1.0891442657134932e-08,
      "loss": 2.0213,
      "step": 1493
    },
    {
      "epoch": 0.2985014985014985,
      "grad_norm": 1.6692050695419312,
      "learning_rate": 8.001914775401798e-09,
      "loss": 1.9052,
      "step": 1494
    },
    {
      "epoch": 0.2987012987012987,
      "grad_norm": 3.3736367225646973,
      "learning_rate": 5.5569079056794206e-09,
      "loss": 2.2479,
      "step": 1495
    },
    {
      "epoch": 0.2989010989010989,
      "grad_norm": 1.6611956357955933,
      "learning_rate": 3.5564329174064824e-09,
      "loss": 1.7885,
      "step": 1496
    },
    {
      "epoch": 0.2991008991008991,
      "grad_norm": 2.8642051219940186,
      "learning_rate": 2.0004987038246824e-09,
      "loss": 1.1887,
      "step": 1497
    },
    {
      "epoch": 0.2993006993006993,
      "grad_norm": 1.234829068183899,
      "learning_rate": 8.891121819565306e-10,
      "loss": 0.7075,
      "step": 1498
    },
    {
      "epoch": 0.2995004995004995,
      "grad_norm": 2.8221359252929688,
      "learning_rate": 2.2227829252763344e-10,
      "loss": 1.7144,
      "step": 1499
    },
    {
      "epoch": 0.2997002997002997,
      "grad_norm": 1.8620914220809937,
      "learning_rate": 0.0,
      "loss": 0.7334,
      "step": 1500
    },
    {
      "epoch": 0.2997002997002997,
      "eval_loss": 1.7515219449996948,
      "eval_runtime": 124.2353,
      "eval_samples_per_second": 33.928,
      "eval_steps_per_second": 8.484,
      "step": 1500
    }
  ],
  "logging_steps": 1,
  "max_steps": 1500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 375,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.660591381450588e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}