{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 200,
  "global_step": 1624,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006157635467980296,
      "grad_norm": 1.721225001438228,
      "learning_rate": 1.226993865030675e-07,
      "loss": 0.8773,
      "step": 1
    },
    {
      "epoch": 0.0012315270935960591,
      "grad_norm": 1.761449591811476,
      "learning_rate": 2.45398773006135e-07,
      "loss": 0.9148,
      "step": 2
    },
    {
      "epoch": 0.0018472906403940886,
      "grad_norm": 1.7418986975438806,
      "learning_rate": 3.680981595092025e-07,
      "loss": 0.9107,
      "step": 3
    },
    {
      "epoch": 0.0024630541871921183,
      "grad_norm": 1.802626307204644,
      "learning_rate": 4.9079754601227e-07,
      "loss": 0.9222,
      "step": 4
    },
    {
      "epoch": 0.003078817733990148,
      "grad_norm": 2.0097560577889477,
      "learning_rate": 6.134969325153375e-07,
      "loss": 0.963,
      "step": 5
    },
    {
      "epoch": 0.003694581280788177,
      "grad_norm": 1.6592882432475673,
      "learning_rate": 7.36196319018405e-07,
      "loss": 0.9265,
      "step": 6
    },
    {
      "epoch": 0.004310344827586207,
      "grad_norm": 1.610410489360794,
      "learning_rate": 8.588957055214725e-07,
      "loss": 0.9672,
      "step": 7
    },
    {
      "epoch": 0.0049261083743842365,
      "grad_norm": 1.386093689587479,
      "learning_rate": 9.8159509202454e-07,
      "loss": 0.8782,
      "step": 8
    },
    {
      "epoch": 0.005541871921182266,
      "grad_norm": 1.4652023534386536,
      "learning_rate": 1.1042944785276075e-06,
      "loss": 0.9299,
      "step": 9
    },
    {
      "epoch": 0.006157635467980296,
      "grad_norm": 1.0723118235360465,
      "learning_rate": 1.226993865030675e-06,
      "loss": 0.8725,
      "step": 10
    },
    {
      "epoch": 0.0067733990147783255,
      "grad_norm": 0.9484974898811056,
      "learning_rate": 1.3496932515337425e-06,
      "loss": 0.8709,
      "step": 11
    },
    {
      "epoch": 0.007389162561576354,
      "grad_norm": 0.8784689208658663,
      "learning_rate": 1.47239263803681e-06,
      "loss": 0.9015,
      "step": 12
    },
    {
      "epoch": 0.008004926108374385,
      "grad_norm": 0.7508371524800348,
      "learning_rate": 1.5950920245398775e-06,
      "loss": 0.8792,
      "step": 13
    },
    {
      "epoch": 0.008620689655172414,
      "grad_norm": 0.8647640920710259,
      "learning_rate": 1.717791411042945e-06,
      "loss": 0.893,
      "step": 14
    },
    {
      "epoch": 0.009236453201970444,
      "grad_norm": 0.9674971150888206,
      "learning_rate": 1.8404907975460124e-06,
      "loss": 0.8959,
      "step": 15
    },
    {
      "epoch": 0.009852216748768473,
      "grad_norm": 0.8880139035914942,
      "learning_rate": 1.96319018404908e-06,
      "loss": 0.9241,
      "step": 16
    },
    {
      "epoch": 0.010467980295566502,
      "grad_norm": 0.9041223500018419,
      "learning_rate": 2.085889570552147e-06,
      "loss": 0.8673,
      "step": 17
    },
    {
      "epoch": 0.011083743842364532,
      "grad_norm": 0.9403938406911142,
      "learning_rate": 2.208588957055215e-06,
      "loss": 0.8984,
      "step": 18
    },
    {
      "epoch": 0.011699507389162561,
      "grad_norm": 0.846897420506937,
      "learning_rate": 2.331288343558282e-06,
      "loss": 0.9279,
      "step": 19
    },
    {
      "epoch": 0.012315270935960592,
      "grad_norm": 0.8532580650052666,
      "learning_rate": 2.45398773006135e-06,
      "loss": 0.8785,
      "step": 20
    },
    {
      "epoch": 0.01293103448275862,
      "grad_norm": 0.8994737748819118,
      "learning_rate": 2.5766871165644175e-06,
      "loss": 0.8702,
      "step": 21
    },
    {
      "epoch": 0.013546798029556651,
      "grad_norm": 0.8199875578238895,
      "learning_rate": 2.699386503067485e-06,
      "loss": 0.923,
      "step": 22
    },
    {
      "epoch": 0.01416256157635468,
      "grad_norm": 0.7694128402470037,
      "learning_rate": 2.822085889570552e-06,
      "loss": 0.936,
      "step": 23
    },
    {
      "epoch": 0.014778325123152709,
      "grad_norm": 0.7673585316550291,
      "learning_rate": 2.94478527607362e-06,
      "loss": 0.9027,
      "step": 24
    },
    {
      "epoch": 0.01539408866995074,
      "grad_norm": 0.756912037400029,
      "learning_rate": 3.0674846625766875e-06,
      "loss": 0.8824,
      "step": 25
    },
    {
      "epoch": 0.01600985221674877,
      "grad_norm": 0.7792563927083785,
      "learning_rate": 3.190184049079755e-06,
      "loss": 0.8887,
      "step": 26
    },
    {
      "epoch": 0.0166256157635468,
      "grad_norm": 0.775019720655459,
      "learning_rate": 3.312883435582822e-06,
      "loss": 0.8917,
      "step": 27
    },
    {
      "epoch": 0.017241379310344827,
      "grad_norm": 0.8632777951943599,
      "learning_rate": 3.43558282208589e-06,
      "loss": 0.876,
      "step": 28
    },
    {
      "epoch": 0.017857142857142856,
      "grad_norm": 0.7802699496594686,
      "learning_rate": 3.5582822085889574e-06,
      "loss": 0.9636,
      "step": 29
    },
    {
      "epoch": 0.01847290640394089,
      "grad_norm": 0.7701837553810241,
      "learning_rate": 3.680981595092025e-06,
      "loss": 0.8895,
      "step": 30
    },
    {
      "epoch": 0.019088669950738917,
      "grad_norm": 0.845385357225407,
      "learning_rate": 3.8036809815950928e-06,
      "loss": 0.8944,
      "step": 31
    },
    {
      "epoch": 0.019704433497536946,
      "grad_norm": 0.793922285216402,
      "learning_rate": 3.92638036809816e-06,
      "loss": 0.9025,
      "step": 32
    },
    {
      "epoch": 0.020320197044334975,
      "grad_norm": 0.7781600045718331,
      "learning_rate": 4.049079754601227e-06,
      "loss": 0.9418,
      "step": 33
    },
    {
      "epoch": 0.020935960591133004,
      "grad_norm": 0.7421532346498493,
      "learning_rate": 4.171779141104294e-06,
      "loss": 0.9037,
      "step": 34
    },
    {
      "epoch": 0.021551724137931036,
      "grad_norm": 0.7360494717765897,
      "learning_rate": 4.294478527607362e-06,
      "loss": 0.8747,
      "step": 35
    },
    {
      "epoch": 0.022167487684729065,
      "grad_norm": 0.7806149241033189,
      "learning_rate": 4.41717791411043e-06,
      "loss": 0.9108,
      "step": 36
    },
    {
      "epoch": 0.022783251231527094,
      "grad_norm": 0.7701250199737361,
      "learning_rate": 4.539877300613497e-06,
      "loss": 0.8905,
      "step": 37
    },
    {
      "epoch": 0.023399014778325122,
      "grad_norm": 0.7909203010788056,
      "learning_rate": 4.662576687116564e-06,
      "loss": 0.9038,
      "step": 38
    },
    {
      "epoch": 0.02401477832512315,
      "grad_norm": 0.7319943745852359,
      "learning_rate": 4.785276073619632e-06,
      "loss": 0.8356,
      "step": 39
    },
    {
      "epoch": 0.024630541871921183,
      "grad_norm": 0.7921307319145839,
      "learning_rate": 4.9079754601227e-06,
      "loss": 0.8527,
      "step": 40
    },
    {
      "epoch": 0.025246305418719212,
      "grad_norm": 0.7591963616721327,
      "learning_rate": 5.030674846625767e-06,
      "loss": 0.93,
      "step": 41
    },
    {
      "epoch": 0.02586206896551724,
      "grad_norm": 0.7673054455461492,
      "learning_rate": 5.153374233128835e-06,
      "loss": 0.8716,
      "step": 42
    },
    {
      "epoch": 0.02647783251231527,
      "grad_norm": 0.7364994534307109,
      "learning_rate": 5.276073619631902e-06,
      "loss": 0.8739,
      "step": 43
    },
    {
      "epoch": 0.027093596059113302,
      "grad_norm": 0.7471446919347251,
      "learning_rate": 5.39877300613497e-06,
      "loss": 0.8572,
      "step": 44
    },
    {
      "epoch": 0.02770935960591133,
      "grad_norm": 0.7797552243345709,
      "learning_rate": 5.521472392638038e-06,
      "loss": 0.8847,
      "step": 45
    },
    {
      "epoch": 0.02832512315270936,
      "grad_norm": 0.7989373604586089,
      "learning_rate": 5.644171779141104e-06,
      "loss": 0.8981,
      "step": 46
    },
    {
      "epoch": 0.02894088669950739,
      "grad_norm": 0.7920594484602965,
      "learning_rate": 5.766871165644172e-06,
      "loss": 0.8737,
      "step": 47
    },
    {
      "epoch": 0.029556650246305417,
      "grad_norm": 0.7640713620796211,
      "learning_rate": 5.88957055214724e-06,
      "loss": 0.8771,
      "step": 48
    },
    {
      "epoch": 0.03017241379310345,
      "grad_norm": 0.8377200333112498,
      "learning_rate": 6.012269938650307e-06,
      "loss": 0.8989,
      "step": 49
    },
    {
      "epoch": 0.03078817733990148,
      "grad_norm": 0.8054367108202074,
      "learning_rate": 6.134969325153375e-06,
      "loss": 0.8662,
      "step": 50
    },
    {
      "epoch": 0.03140394088669951,
      "grad_norm": 0.7947954181333833,
      "learning_rate": 6.257668711656443e-06,
      "loss": 0.9014,
      "step": 51
    },
    {
      "epoch": 0.03201970443349754,
      "grad_norm": 0.8411782441989318,
      "learning_rate": 6.38036809815951e-06,
      "loss": 0.9038,
      "step": 52
    },
    {
      "epoch": 0.03263546798029557,
      "grad_norm": 0.7700814176980257,
      "learning_rate": 6.503067484662578e-06,
      "loss": 0.9061,
      "step": 53
    },
    {
      "epoch": 0.0332512315270936,
      "grad_norm": 0.7675571806944624,
      "learning_rate": 6.625766871165644e-06,
      "loss": 0.9009,
      "step": 54
    },
    {
      "epoch": 0.033866995073891626,
      "grad_norm": 0.7433329957685538,
      "learning_rate": 6.748466257668712e-06,
      "loss": 0.8998,
      "step": 55
    },
    {
      "epoch": 0.034482758620689655,
      "grad_norm": 0.8133958703190483,
      "learning_rate": 6.87116564417178e-06,
      "loss": 0.9023,
      "step": 56
    },
    {
      "epoch": 0.035098522167487683,
      "grad_norm": 0.7948870165656131,
      "learning_rate": 6.993865030674847e-06,
      "loss": 0.8892,
      "step": 57
    },
    {
      "epoch": 0.03571428571428571,
      "grad_norm": 0.7731371312531223,
      "learning_rate": 7.116564417177915e-06,
      "loss": 0.825,
      "step": 58
    },
    {
      "epoch": 0.03633004926108374,
      "grad_norm": 0.8469677819360523,
      "learning_rate": 7.239263803680983e-06,
      "loss": 0.8899,
      "step": 59
    },
    {
      "epoch": 0.03694581280788178,
      "grad_norm": 0.7965637902163071,
      "learning_rate": 7.36196319018405e-06,
      "loss": 0.9305,
      "step": 60
    },
    {
      "epoch": 0.037561576354679806,
      "grad_norm": 0.8041102084151476,
      "learning_rate": 7.484662576687118e-06,
      "loss": 0.9698,
      "step": 61
    },
    {
      "epoch": 0.038177339901477834,
      "grad_norm": 0.7731752588890729,
      "learning_rate": 7.6073619631901856e-06,
      "loss": 0.9213,
      "step": 62
    },
    {
      "epoch": 0.03879310344827586,
      "grad_norm": 0.8242709018183485,
      "learning_rate": 7.730061349693252e-06,
      "loss": 0.9086,
      "step": 63
    },
    {
      "epoch": 0.03940886699507389,
      "grad_norm": 0.8800117795352059,
      "learning_rate": 7.85276073619632e-06,
      "loss": 0.8597,
      "step": 64
    },
    {
      "epoch": 0.04002463054187192,
      "grad_norm": 0.8276446887500438,
      "learning_rate": 7.975460122699386e-06,
      "loss": 0.8635,
      "step": 65
    },
    {
      "epoch": 0.04064039408866995,
      "grad_norm": 0.7785733603298536,
      "learning_rate": 8.098159509202455e-06,
      "loss": 0.87,
      "step": 66
    },
    {
      "epoch": 0.04125615763546798,
      "grad_norm": 0.7505345016880394,
      "learning_rate": 8.220858895705522e-06,
      "loss": 0.8677,
      "step": 67
    },
    {
      "epoch": 0.04187192118226601,
      "grad_norm": 0.7818170918133651,
      "learning_rate": 8.343558282208589e-06,
      "loss": 0.8876,
      "step": 68
    },
    {
      "epoch": 0.042487684729064036,
      "grad_norm": 0.7482286373466502,
      "learning_rate": 8.466257668711658e-06,
      "loss": 0.8995,
      "step": 69
    },
    {
      "epoch": 0.04310344827586207,
      "grad_norm": 0.7349347321553887,
      "learning_rate": 8.588957055214725e-06,
      "loss": 0.9233,
      "step": 70
    },
    {
      "epoch": 0.0437192118226601,
      "grad_norm": 0.7945898211870677,
      "learning_rate": 8.711656441717792e-06,
      "loss": 0.8836,
      "step": 71
    },
    {
      "epoch": 0.04433497536945813,
      "grad_norm": 0.8266255954816315,
      "learning_rate": 8.83435582822086e-06,
      "loss": 0.9117,
      "step": 72
    },
    {
      "epoch": 0.04495073891625616,
      "grad_norm": 0.8401830732652382,
      "learning_rate": 8.957055214723927e-06,
      "loss": 0.8882,
      "step": 73
    },
    {
      "epoch": 0.04556650246305419,
      "grad_norm": 0.7736490316489866,
      "learning_rate": 9.079754601226994e-06,
      "loss": 0.8904,
      "step": 74
    },
    {
      "epoch": 0.046182266009852216,
      "grad_norm": 0.7883338735183056,
      "learning_rate": 9.202453987730062e-06,
      "loss": 0.8935,
      "step": 75
    },
    {
      "epoch": 0.046798029556650245,
      "grad_norm": 0.8438813461088808,
      "learning_rate": 9.325153374233129e-06,
      "loss": 0.8952,
      "step": 76
    },
    {
      "epoch": 0.04741379310344827,
      "grad_norm": 0.7956332654635961,
      "learning_rate": 9.447852760736197e-06,
      "loss": 0.9039,
      "step": 77
    },
    {
      "epoch": 0.0480295566502463,
      "grad_norm": 0.7950278546472513,
      "learning_rate": 9.570552147239264e-06,
      "loss": 0.9451,
      "step": 78
    },
    {
      "epoch": 0.04864532019704434,
      "grad_norm": 0.8859768808291926,
      "learning_rate": 9.693251533742331e-06,
      "loss": 0.9319,
      "step": 79
    },
    {
      "epoch": 0.04926108374384237,
      "grad_norm": 0.8041801798316897,
      "learning_rate": 9.8159509202454e-06,
      "loss": 0.8606,
      "step": 80
    },
    {
      "epoch": 0.049876847290640396,
      "grad_norm": 0.7404593218302198,
      "learning_rate": 9.938650306748467e-06,
      "loss": 0.8148,
      "step": 81
    },
    {
      "epoch": 0.050492610837438424,
      "grad_norm": 0.8235236985048631,
      "learning_rate": 1.0061349693251534e-05,
      "loss": 0.893,
      "step": 82
    },
    {
      "epoch": 0.05110837438423645,
      "grad_norm": 0.846297255087323,
      "learning_rate": 1.0184049079754601e-05,
      "loss": 0.906,
      "step": 83
    },
    {
      "epoch": 0.05172413793103448,
      "grad_norm": 0.8346901591685469,
      "learning_rate": 1.030674846625767e-05,
      "loss": 0.9136,
      "step": 84
    },
    {
      "epoch": 0.05233990147783251,
      "grad_norm": 0.7971002204204433,
      "learning_rate": 1.0429447852760737e-05,
      "loss": 0.8915,
      "step": 85
    },
    {
      "epoch": 0.05295566502463054,
      "grad_norm": 0.8339421780179564,
      "learning_rate": 1.0552147239263804e-05,
      "loss": 0.8798,
      "step": 86
    },
    {
      "epoch": 0.05357142857142857,
      "grad_norm": 0.8472098001071048,
      "learning_rate": 1.0674846625766873e-05,
      "loss": 0.9065,
      "step": 87
    },
    {
      "epoch": 0.054187192118226604,
      "grad_norm": 0.7623695216595175,
      "learning_rate": 1.079754601226994e-05,
      "loss": 0.8876,
      "step": 88
    },
    {
      "epoch": 0.05480295566502463,
      "grad_norm": 0.8257432776800313,
      "learning_rate": 1.0920245398773005e-05,
      "loss": 0.8921,
      "step": 89
    },
    {
      "epoch": 0.05541871921182266,
      "grad_norm": 0.8152833868044597,
      "learning_rate": 1.1042944785276076e-05,
      "loss": 0.9345,
      "step": 90
    },
    {
      "epoch": 0.05603448275862069,
      "grad_norm": 0.8456383837736814,
      "learning_rate": 1.1165644171779141e-05,
      "loss": 0.9218,
      "step": 91
    },
    {
      "epoch": 0.05665024630541872,
      "grad_norm": 0.8430873162814652,
      "learning_rate": 1.1288343558282208e-05,
      "loss": 0.8605,
      "step": 92
    },
    {
      "epoch": 0.05726600985221675,
      "grad_norm": 0.7804923367501427,
      "learning_rate": 1.1411042944785277e-05,
      "loss": 0.9085,
      "step": 93
    },
    {
      "epoch": 0.05788177339901478,
      "grad_norm": 0.8671700737828291,
      "learning_rate": 1.1533742331288344e-05,
      "loss": 0.9419,
      "step": 94
    },
    {
      "epoch": 0.058497536945812806,
      "grad_norm": 0.8192357717815728,
      "learning_rate": 1.1656441717791411e-05,
      "loss": 0.905,
      "step": 95
    },
    {
      "epoch": 0.059113300492610835,
      "grad_norm": 0.8067399598953795,
      "learning_rate": 1.177914110429448e-05,
      "loss": 0.8799,
      "step": 96
    },
    {
      "epoch": 0.05972906403940887,
      "grad_norm": 0.8003319045036887,
      "learning_rate": 1.1901840490797547e-05,
      "loss": 0.9008,
      "step": 97
    },
    {
      "epoch": 0.0603448275862069,
      "grad_norm": 0.7383782044087496,
      "learning_rate": 1.2024539877300614e-05,
      "loss": 0.8512,
      "step": 98
    },
    {
      "epoch": 0.06096059113300493,
      "grad_norm": 0.7945989594972028,
      "learning_rate": 1.2147239263803683e-05,
      "loss": 0.8698,
      "step": 99
    },
    {
      "epoch": 0.06157635467980296,
      "grad_norm": 0.8254712573654962,
      "learning_rate": 1.226993865030675e-05,
      "loss": 0.8933,
      "step": 100
    },
    {
      "epoch": 0.062192118226600986,
      "grad_norm": 0.8497458128005192,
      "learning_rate": 1.2392638036809817e-05,
      "loss": 0.8242,
      "step": 101
    },
    {
      "epoch": 0.06280788177339902,
      "grad_norm": 0.7848574521153052,
      "learning_rate": 1.2515337423312886e-05,
      "loss": 0.8868,
      "step": 102
    },
    {
      "epoch": 0.06342364532019705,
      "grad_norm": 0.7803779830006228,
      "learning_rate": 1.2638036809815953e-05,
      "loss": 0.8941,
      "step": 103
    },
    {
      "epoch": 0.06403940886699508,
      "grad_norm": 0.9103498769563041,
      "learning_rate": 1.276073619631902e-05,
      "loss": 0.9045,
      "step": 104
    },
    {
      "epoch": 0.06465517241379311,
      "grad_norm": 0.8578836360189689,
      "learning_rate": 1.2883435582822085e-05,
      "loss": 0.9033,
      "step": 105
    },
    {
      "epoch": 0.06527093596059114,
      "grad_norm": 0.8227794055234684,
      "learning_rate": 1.3006134969325156e-05,
      "loss": 0.9086,
      "step": 106
    },
    {
      "epoch": 0.06588669950738917,
      "grad_norm": 0.8508853078867162,
      "learning_rate": 1.3128834355828221e-05,
      "loss": 0.8819,
      "step": 107
    },
    {
      "epoch": 0.0665024630541872,
      "grad_norm": 0.8568135819152922,
      "learning_rate": 1.3251533742331288e-05,
      "loss": 0.8535,
      "step": 108
    },
    {
      "epoch": 0.06711822660098522,
      "grad_norm": 0.8503777280182665,
      "learning_rate": 1.3374233128834357e-05,
      "loss": 0.927,
      "step": 109
    },
    {
      "epoch": 0.06773399014778325,
      "grad_norm": 0.8886923200565565,
      "learning_rate": 1.3496932515337424e-05,
      "loss": 0.9484,
      "step": 110
    },
    {
      "epoch": 0.06834975369458128,
      "grad_norm": 0.8343761427704829,
      "learning_rate": 1.3619631901840491e-05,
      "loss": 0.8959,
      "step": 111
    },
    {
      "epoch": 0.06896551724137931,
      "grad_norm": 0.8906695043655418,
      "learning_rate": 1.374233128834356e-05,
      "loss": 0.9292,
      "step": 112
    },
    {
      "epoch": 0.06958128078817734,
      "grad_norm": 0.9112066001699903,
      "learning_rate": 1.3865030674846627e-05,
      "loss": 0.957,
      "step": 113
    },
    {
      "epoch": 0.07019704433497537,
      "grad_norm": 0.8831695287042961,
      "learning_rate": 1.3987730061349694e-05,
      "loss": 0.9032,
      "step": 114
    },
    {
      "epoch": 0.0708128078817734,
      "grad_norm": 0.9466099851920295,
      "learning_rate": 1.4110429447852763e-05,
      "loss": 0.8709,
      "step": 115
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 0.8991797980803579,
      "learning_rate": 1.423312883435583e-05,
      "loss": 0.9885,
      "step": 116
    },
    {
      "epoch": 0.07204433497536945,
      "grad_norm": 0.9132091687913635,
      "learning_rate": 1.4355828220858897e-05,
      "loss": 0.9268,
      "step": 117
    },
    {
      "epoch": 0.07266009852216748,
      "grad_norm": 0.8575510233248194,
      "learning_rate": 1.4478527607361965e-05,
      "loss": 0.9083,
      "step": 118
    },
    {
      "epoch": 0.07327586206896551,
      "grad_norm": 0.7797718430354518,
      "learning_rate": 1.4601226993865032e-05,
      "loss": 0.9136,
      "step": 119
    },
    {
      "epoch": 0.07389162561576355,
      "grad_norm": 0.956063784302148,
      "learning_rate": 1.47239263803681e-05,
      "loss": 0.933,
      "step": 120
    },
    {
      "epoch": 0.07450738916256158,
      "grad_norm": 0.7917170004637722,
      "learning_rate": 1.4846625766871168e-05,
      "loss": 0.9421,
      "step": 121
    },
    {
      "epoch": 0.07512315270935961,
      "grad_norm": 0.8029451774284248,
      "learning_rate": 1.4969325153374235e-05,
      "loss": 0.9199,
      "step": 122
    },
    {
      "epoch": 0.07573891625615764,
      "grad_norm": 0.8186285812457579,
      "learning_rate": 1.50920245398773e-05,
      "loss": 0.864,
      "step": 123
    },
    {
      "epoch": 0.07635467980295567,
      "grad_norm": 0.8622955066595351,
      "learning_rate": 1.5214723926380371e-05,
      "loss": 0.9103,
      "step": 124
    },
    {
      "epoch": 0.0769704433497537,
      "grad_norm": 0.8309941842113272,
      "learning_rate": 1.5337423312883436e-05,
      "loss": 0.9673,
      "step": 125
    },
    {
      "epoch": 0.07758620689655173,
      "grad_norm": 0.77864206025243,
      "learning_rate": 1.5460122699386504e-05,
      "loss": 0.8376,
      "step": 126
    },
    {
      "epoch": 0.07820197044334976,
      "grad_norm": 0.983707276357124,
      "learning_rate": 1.5582822085889574e-05,
      "loss": 0.9278,
      "step": 127
    },
    {
      "epoch": 0.07881773399014778,
      "grad_norm": 0.8732821583207065,
      "learning_rate": 1.570552147239264e-05,
      "loss": 0.9049,
      "step": 128
    },
    {
      "epoch": 0.07943349753694581,
      "grad_norm": 0.8445050698838605,
      "learning_rate": 1.5828220858895708e-05,
      "loss": 0.9281,
      "step": 129
    },
    {
      "epoch": 0.08004926108374384,
      "grad_norm": 0.9196087517410864,
      "learning_rate": 1.5950920245398772e-05,
      "loss": 0.9245,
      "step": 130
    },
    {
      "epoch": 0.08066502463054187,
      "grad_norm": 0.8750914056845331,
      "learning_rate": 1.6073619631901842e-05,
      "loss": 0.8744,
      "step": 131
    },
    {
      "epoch": 0.0812807881773399,
      "grad_norm": 0.8181948757724183,
      "learning_rate": 1.619631901840491e-05,
      "loss": 0.8532,
      "step": 132
    },
    {
      "epoch": 0.08189655172413793,
      "grad_norm": 0.8098171912457909,
      "learning_rate": 1.6319018404907976e-05,
      "loss": 0.9194,
      "step": 133
    },
    {
      "epoch": 0.08251231527093596,
      "grad_norm": 0.7989212941758933,
      "learning_rate": 1.6441717791411043e-05,
      "loss": 0.8426,
      "step": 134
    },
    {
      "epoch": 0.08312807881773399,
      "grad_norm": 0.872001279696351,
      "learning_rate": 1.656441717791411e-05,
      "loss": 0.863,
      "step": 135
    },
    {
      "epoch": 0.08374384236453201,
      "grad_norm": 0.7977844268913128,
      "learning_rate": 1.6687116564417178e-05,
      "loss": 0.9064,
      "step": 136
    },
    {
      "epoch": 0.08435960591133004,
      "grad_norm": 0.8022802398821214,
      "learning_rate": 1.6809815950920248e-05,
      "loss": 0.901,
      "step": 137
    },
    {
      "epoch": 0.08497536945812807,
      "grad_norm": 0.8386236574175275,
      "learning_rate": 1.6932515337423315e-05,
      "loss": 0.8712,
      "step": 138
    },
    {
      "epoch": 0.08559113300492611,
      "grad_norm": 0.8698883383954628,
      "learning_rate": 1.7055214723926382e-05,
      "loss": 0.9301,
      "step": 139
    },
    {
      "epoch": 0.08620689655172414,
      "grad_norm": 0.7462752000102024,
      "learning_rate": 1.717791411042945e-05,
      "loss": 0.9006,
      "step": 140
    },
    {
      "epoch": 0.08682266009852217,
      "grad_norm": 0.7836471599091207,
      "learning_rate": 1.7300613496932516e-05,
      "loss": 0.939,
      "step": 141
    },
    {
      "epoch": 0.0874384236453202,
      "grad_norm": 0.955610975008346,
      "learning_rate": 1.7423312883435583e-05,
      "loss": 0.9306,
      "step": 142
    },
    {
      "epoch": 0.08805418719211823,
      "grad_norm": 0.7896130653383984,
      "learning_rate": 1.7546012269938654e-05,
      "loss": 0.8856,
      "step": 143
    },
    {
      "epoch": 0.08866995073891626,
      "grad_norm": 0.7780248917712866,
      "learning_rate": 1.766871165644172e-05,
      "loss": 0.937,
      "step": 144
    },
    {
      "epoch": 0.08928571428571429,
      "grad_norm": 0.8350347144786314,
      "learning_rate": 1.7791411042944788e-05,
      "loss": 0.9989,
      "step": 145
    },
    {
      "epoch": 0.08990147783251232,
      "grad_norm": 0.8705246009370556,
      "learning_rate": 1.7914110429447855e-05,
      "loss": 0.9146,
      "step": 146
    },
    {
      "epoch": 0.09051724137931035,
      "grad_norm": 0.7120216917284158,
      "learning_rate": 1.8036809815950922e-05,
      "loss": 0.8781,
      "step": 147
    },
    {
      "epoch": 0.09113300492610837,
      "grad_norm": 0.7345460292740787,
      "learning_rate": 1.815950920245399e-05,
      "loss": 0.9056,
      "step": 148
    },
    {
      "epoch": 0.0917487684729064,
      "grad_norm": 0.8572024484100429,
      "learning_rate": 1.828220858895706e-05,
      "loss": 0.9271,
      "step": 149
    },
    {
      "epoch": 0.09236453201970443,
      "grad_norm": 0.7609473964134008,
      "learning_rate": 1.8404907975460123e-05,
      "loss": 0.884,
      "step": 150
    },
    {
      "epoch": 0.09298029556650246,
      "grad_norm": 0.8426420339461909,
      "learning_rate": 1.852760736196319e-05,
      "loss": 0.9388,
      "step": 151
    },
    {
      "epoch": 0.09359605911330049,
      "grad_norm": 0.7927397876577357,
      "learning_rate": 1.8650306748466257e-05,
      "loss": 0.8917,
      "step": 152
    },
    {
      "epoch": 0.09421182266009852,
      "grad_norm": 0.7792181849218192,
      "learning_rate": 1.8773006134969328e-05,
      "loss": 0.8968,
      "step": 153
    },
    {
      "epoch": 0.09482758620689655,
      "grad_norm": 0.8009884843245464,
      "learning_rate": 1.8895705521472395e-05,
      "loss": 0.904,
      "step": 154
    },
    {
      "epoch": 0.09544334975369458,
      "grad_norm": 0.8778291721261292,
      "learning_rate": 1.9018404907975462e-05,
      "loss": 0.9187,
      "step": 155
    },
    {
      "epoch": 0.0960591133004926,
      "grad_norm": 0.9016618452466743,
      "learning_rate": 1.914110429447853e-05,
      "loss": 0.897,
      "step": 156
    },
    {
      "epoch": 0.09667487684729065,
      "grad_norm": 0.7995635359625541,
      "learning_rate": 1.9263803680981596e-05,
      "loss": 0.9004,
      "step": 157
    },
    {
      "epoch": 0.09729064039408868,
      "grad_norm": 0.9011543731495566,
      "learning_rate": 1.9386503067484663e-05,
      "loss": 0.9323,
      "step": 158
    },
    {
      "epoch": 0.0979064039408867,
      "grad_norm": 0.7929095016787937,
      "learning_rate": 1.9509202453987733e-05,
      "loss": 0.8634,
      "step": 159
    },
    {
      "epoch": 0.09852216748768473,
      "grad_norm": 0.8515972803071793,
      "learning_rate": 1.96319018404908e-05,
      "loss": 0.9281,
      "step": 160
    },
    {
      "epoch": 0.09913793103448276,
      "grad_norm": 0.790402021927727,
      "learning_rate": 1.9754601226993868e-05,
      "loss": 0.8292,
      "step": 161
    },
    {
      "epoch": 0.09975369458128079,
      "grad_norm": 0.9004428859686308,
      "learning_rate": 1.9877300613496935e-05,
      "loss": 0.9487,
      "step": 162
    },
    {
      "epoch": 0.10036945812807882,
      "grad_norm": 0.8031744588347524,
      "learning_rate": 2e-05,
      "loss": 0.8809,
      "step": 163
    },
    {
      "epoch": 0.10098522167487685,
      "grad_norm": 0.8346631589347174,
      "learning_rate": 1.9999976880994384e-05,
      "loss": 0.9359,
      "step": 164
    },
    {
      "epoch": 0.10160098522167488,
      "grad_norm": 0.8696639320439042,
      "learning_rate": 1.999990752408443e-05,
      "loss": 0.9256,
      "step": 165
    },
    {
      "epoch": 0.1022167487684729,
      "grad_norm": 1.0772236328827147,
      "learning_rate": 1.9999791929590823e-05,
      "loss": 0.9179,
      "step": 166
    },
    {
      "epoch": 0.10283251231527094,
      "grad_norm": 0.9297112478665194,
      "learning_rate": 1.999963009804806e-05,
      "loss": 0.9515,
      "step": 167
    },
    {
      "epoch": 0.10344827586206896,
      "grad_norm": 0.9073442916178279,
      "learning_rate": 1.999942203020442e-05,
      "loss": 0.8692,
      "step": 168
    },
    {
      "epoch": 0.10406403940886699,
      "grad_norm": 1.0432606858253322,
      "learning_rate": 1.9999167727021954e-05,
      "loss": 0.946,
      "step": 169
    },
    {
      "epoch": 0.10467980295566502,
      "grad_norm": 0.8007998350459348,
      "learning_rate": 1.9998867189676517e-05,
      "loss": 0.9476,
      "step": 170
    },
    {
      "epoch": 0.10529556650246305,
      "grad_norm": 0.9617632974869773,
      "learning_rate": 1.9998520419557736e-05,
      "loss": 0.9276,
      "step": 171
    },
    {
      "epoch": 0.10591133004926108,
      "grad_norm": 0.8118624217932321,
      "learning_rate": 1.9998127418269004e-05,
      "loss": 0.9155,
      "step": 172
    },
    {
      "epoch": 0.10652709359605911,
      "grad_norm": 0.8675105478680718,
      "learning_rate": 1.999768818762748e-05,
      "loss": 0.8802,
      "step": 173
    },
    {
      "epoch": 0.10714285714285714,
      "grad_norm": 0.7944597632181472,
      "learning_rate": 1.9997202729664085e-05,
      "loss": 0.9195,
      "step": 174
    },
    {
      "epoch": 0.10775862068965517,
      "grad_norm": 0.7971011424920242,
      "learning_rate": 1.999667104662347e-05,
      "loss": 0.9636,
      "step": 175
    },
    {
      "epoch": 0.10837438423645321,
      "grad_norm": 0.8277594905566759,
      "learning_rate": 1.999609314096404e-05,
      "loss": 0.8944,
      "step": 176
    },
    {
      "epoch": 0.10899014778325124,
      "grad_norm": 0.8689097698007539,
      "learning_rate": 1.9995469015357914e-05,
      "loss": 0.9175,
      "step": 177
    },
    {
      "epoch": 0.10960591133004927,
      "grad_norm": 0.7264382983824488,
      "learning_rate": 1.9994798672690922e-05,
      "loss": 0.8957,
      "step": 178
    },
    {
      "epoch": 0.1102216748768473,
      "grad_norm": 0.7766617644792241,
      "learning_rate": 1.9994082116062602e-05,
      "loss": 0.9004,
      "step": 179
    },
    {
      "epoch": 0.11083743842364532,
      "grad_norm": 0.8058619703120573,
      "learning_rate": 1.9993319348786157e-05,
      "loss": 0.9362,
      "step": 180
    },
    {
      "epoch": 0.11145320197044335,
      "grad_norm": 0.8148471866820708,
      "learning_rate": 1.9992510374388484e-05,
      "loss": 0.9279,
      "step": 181
    },
    {
      "epoch": 0.11206896551724138,
      "grad_norm": 0.756890977599957,
      "learning_rate": 1.999165519661011e-05,
      "loss": 0.9286,
      "step": 182
    },
    {
      "epoch": 0.11268472906403941,
      "grad_norm": 0.9195579508192184,
      "learning_rate": 1.9990753819405212e-05,
      "loss": 0.8916,
      "step": 183
    },
    {
      "epoch": 0.11330049261083744,
      "grad_norm": 0.8406498485025024,
      "learning_rate": 1.9989806246941583e-05,
      "loss": 0.9197,
      "step": 184
    },
    {
      "epoch": 0.11391625615763547,
      "grad_norm": 0.8092498206198186,
      "learning_rate": 1.9988812483600597e-05,
      "loss": 0.9119,
      "step": 185
    },
    {
      "epoch": 0.1145320197044335,
      "grad_norm": 0.9790541044001935,
      "learning_rate": 1.998777253397723e-05,
      "loss": 0.8968,
      "step": 186
    },
    {
      "epoch": 0.11514778325123153,
      "grad_norm": 0.7552530913453138,
      "learning_rate": 1.998668640288e-05,
      "loss": 0.8868,
      "step": 187
    },
    {
      "epoch": 0.11576354679802955,
      "grad_norm": 0.8747687673198808,
      "learning_rate": 1.9985554095330958e-05,
      "loss": 0.9162,
      "step": 188
    },
    {
      "epoch": 0.11637931034482758,
      "grad_norm": 0.7168082556197526,
      "learning_rate": 1.9984375616565666e-05,
      "loss": 0.8906,
      "step": 189
    },
    {
      "epoch": 0.11699507389162561,
      "grad_norm": 0.7679004666241568,
      "learning_rate": 1.9983150972033186e-05,
      "loss": 0.8908,
      "step": 190
    },
    {
      "epoch": 0.11761083743842364,
      "grad_norm": 0.8115726000340714,
      "learning_rate": 1.9981880167396022e-05,
      "loss": 0.9494,
      "step": 191
    },
    {
      "epoch": 0.11822660098522167,
      "grad_norm": 0.7453180759937623,
      "learning_rate": 1.998056320853012e-05,
      "loss": 0.8865,
      "step": 192
    },
    {
      "epoch": 0.1188423645320197,
      "grad_norm": 0.8897450510132522,
      "learning_rate": 1.9979200101524844e-05,
      "loss": 0.8903,
      "step": 193
    },
    {
      "epoch": 0.11945812807881774,
      "grad_norm": 0.8294717021117937,
      "learning_rate": 1.9977790852682924e-05,
      "loss": 0.9442,
      "step": 194
    },
    {
      "epoch": 0.12007389162561577,
      "grad_norm": 0.8266768581136608,
      "learning_rate": 1.9976335468520452e-05,
      "loss": 0.9609,
      "step": 195
    },
    {
      "epoch": 0.1206896551724138,
      "grad_norm": 0.79378271436016,
      "learning_rate": 1.9974833955766832e-05,
      "loss": 0.9521,
      "step": 196
    },
    {
      "epoch": 0.12130541871921183,
      "grad_norm": 0.8744256884049737,
      "learning_rate": 1.9973286321364753e-05,
      "loss": 0.947,
      "step": 197
    },
    {
      "epoch": 0.12192118226600986,
      "grad_norm": 0.821879502649211,
      "learning_rate": 1.997169257247018e-05,
      "loss": 0.9816,
      "step": 198
    },
    {
      "epoch": 0.12253694581280788,
      "grad_norm": 0.7679678189425322,
      "learning_rate": 1.9970052716452284e-05,
      "loss": 0.8986,
      "step": 199
    },
    {
      "epoch": 0.12315270935960591,
      "grad_norm": 0.8019476384563389,
      "learning_rate": 1.9968366760893437e-05,
      "loss": 0.9206,
      "step": 200
    },
    {
      "epoch": 0.12315270935960591,
      "eval_loss": 0.9238145351409912,
      "eval_runtime": 523.5533,
      "eval_samples_per_second": 44.141,
      "eval_steps_per_second": 0.691,
      "step": 200
    },
    {
      "epoch": 0.12376847290640394,
      "grad_norm": 0.766811135897515,
      "learning_rate": 1.996663471358916e-05,
      "loss": 0.898,
      "step": 201
    },
    {
      "epoch": 0.12438423645320197,
      "grad_norm": 0.8025368514352377,
      "learning_rate": 1.9964856582548094e-05,
      "loss": 0.9178,
      "step": 202
    },
    {
      "epoch": 0.125,
      "grad_norm": 0.8006454024411117,
      "learning_rate": 1.9963032375991965e-05,
      "loss": 0.8903,
      "step": 203
    },
    {
      "epoch": 0.12561576354679804,
      "grad_norm": 0.8301861874891336,
      "learning_rate": 1.9961162102355543e-05,
      "loss": 0.8789,
      "step": 204
    },
    {
      "epoch": 0.12623152709359606,
      "grad_norm": 0.8115077255795144,
      "learning_rate": 1.9959245770286602e-05,
      "loss": 0.8971,
      "step": 205
    },
    {
      "epoch": 0.1268472906403941,
      "grad_norm": 0.7739623153787399,
      "learning_rate": 1.9957283388645875e-05,
      "loss": 0.8547,
      "step": 206
    },
    {
      "epoch": 0.12746305418719212,
      "grad_norm": 0.8307477255865857,
      "learning_rate": 1.9955274966507027e-05,
      "loss": 0.9061,
      "step": 207
    },
    {
      "epoch": 0.12807881773399016,
      "grad_norm": 0.8064119508356158,
      "learning_rate": 1.9953220513156604e-05,
      "loss": 0.8874,
      "step": 208
    },
    {
      "epoch": 0.12869458128078817,
      "grad_norm": 0.9855998170217997,
      "learning_rate": 1.995112003809399e-05,
      "loss": 0.97,
      "step": 209
    },
    {
      "epoch": 0.12931034482758622,
      "grad_norm": 0.8483606995649495,
      "learning_rate": 1.994897355103136e-05,
      "loss": 0.9274,
      "step": 210
    },
    {
      "epoch": 0.12992610837438423,
      "grad_norm": 0.8483234985696375,
      "learning_rate": 1.9946781061893648e-05,
      "loss": 0.9238,
      "step": 211
    },
    {
      "epoch": 0.13054187192118227,
      "grad_norm": 0.9025464403448759,
      "learning_rate": 1.9944542580818485e-05,
      "loss": 0.9107,
      "step": 212
    },
    {
      "epoch": 0.1311576354679803,
      "grad_norm": 0.7717164729261441,
      "learning_rate": 1.9942258118156165e-05,
      "loss": 0.8782,
      "step": 213
    },
    {
      "epoch": 0.13177339901477833,
      "grad_norm": 0.8325383713764134,
      "learning_rate": 1.9939927684469584e-05,
      "loss": 0.9398,
      "step": 214
    },
    {
      "epoch": 0.13238916256157635,
      "grad_norm": 0.7845058822605753,
      "learning_rate": 1.9937551290534208e-05,
      "loss": 0.9165,
      "step": 215
    },
    {
      "epoch": 0.1330049261083744,
      "grad_norm": 0.8175874643010415,
      "learning_rate": 1.993512894733801e-05,
      "loss": 0.961,
      "step": 216
    },
    {
      "epoch": 0.1336206896551724,
      "grad_norm": 0.8023686474356596,
      "learning_rate": 1.993266066608142e-05,
      "loss": 0.9108,
      "step": 217
    },
    {
      "epoch": 0.13423645320197045,
      "grad_norm": 0.8049109886988418,
      "learning_rate": 1.9930146458177282e-05,
      "loss": 0.8972,
      "step": 218
    },
    {
      "epoch": 0.13485221674876846,
      "grad_norm": 0.7542682686422236,
      "learning_rate": 1.9927586335250796e-05,
      "loss": 0.9325,
      "step": 219
    },
    {
      "epoch": 0.1354679802955665,
      "grad_norm": 0.817376472890674,
      "learning_rate": 1.9924980309139455e-05,
      "loss": 0.9378,
      "step": 220
    },
    {
      "epoch": 0.13608374384236452,
      "grad_norm": 0.8326949710974465,
      "learning_rate": 1.9922328391893007e-05,
      "loss": 0.9712,
      "step": 221
    },
    {
      "epoch": 0.13669950738916256,
      "grad_norm": 0.7526484263871717,
      "learning_rate": 1.9919630595773397e-05,
      "loss": 0.916,
      "step": 222
    },
    {
      "epoch": 0.1373152709359606,
      "grad_norm": 0.8571902278119561,
      "learning_rate": 1.991688693325469e-05,
      "loss": 0.8949,
      "step": 223
    },
    {
      "epoch": 0.13793103448275862,
      "grad_norm": 0.7776264560636764,
      "learning_rate": 1.9914097417023043e-05,
      "loss": 0.9634,
      "step": 224
    },
    {
      "epoch": 0.13854679802955666,
      "grad_norm": 0.7445247398683744,
      "learning_rate": 1.9911262059976614e-05,
      "loss": 0.8878,
      "step": 225
    },
    {
      "epoch": 0.13916256157635468,
      "grad_norm": 0.8354061596531734,
      "learning_rate": 1.9908380875225534e-05,
      "loss": 0.9097,
      "step": 226
    },
    {
      "epoch": 0.13977832512315272,
      "grad_norm": 0.815634553353213,
      "learning_rate": 1.9905453876091835e-05,
      "loss": 0.9234,
      "step": 227
    },
    {
      "epoch": 0.14039408866995073,
      "grad_norm": 0.8155359417632289,
      "learning_rate": 1.9902481076109372e-05,
      "loss": 0.9224,
      "step": 228
    },
    {
      "epoch": 0.14100985221674878,
      "grad_norm": 0.7829625206563378,
      "learning_rate": 1.9899462489023783e-05,
      "loss": 0.9598,
      "step": 229
    },
    {
      "epoch": 0.1416256157635468,
      "grad_norm": 0.8381266262278113,
      "learning_rate": 1.9896398128792413e-05,
      "loss": 0.9105,
      "step": 230
    },
    {
      "epoch": 0.14224137931034483,
      "grad_norm": 0.8080891695346878,
      "learning_rate": 1.9893288009584258e-05,
      "loss": 0.9319,
      "step": 231
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.8667201910162531,
      "learning_rate": 1.9890132145779885e-05,
      "loss": 0.8645,
      "step": 232
    },
    {
      "epoch": 0.1434729064039409,
      "grad_norm": 0.7778523524198537,
      "learning_rate": 1.9886930551971385e-05,
      "loss": 0.9667,
      "step": 233
    },
    {
      "epoch": 0.1440886699507389,
      "grad_norm": 0.8745858831442975,
      "learning_rate": 1.988368324296229e-05,
      "loss": 0.9563,
      "step": 234
    },
    {
      "epoch": 0.14470443349753695,
      "grad_norm": 0.8523609460802929,
      "learning_rate": 1.988039023376751e-05,
      "loss": 0.9612,
      "step": 235
    },
    {
      "epoch": 0.14532019704433496,
      "grad_norm": 0.7878303743097008,
      "learning_rate": 1.9877051539613265e-05,
      "loss": 0.9468,
      "step": 236
    },
    {
      "epoch": 0.145935960591133,
      "grad_norm": 0.7676878713019638,
      "learning_rate": 1.9873667175937016e-05,
      "loss": 0.9307,
      "step": 237
    },
    {
      "epoch": 0.14655172413793102,
      "grad_norm": 0.7774897452328057,
      "learning_rate": 1.9870237158387385e-05,
      "loss": 0.8995,
      "step": 238
    },
    {
      "epoch": 0.14716748768472906,
      "grad_norm": 0.7470787648877826,
      "learning_rate": 1.9866761502824088e-05,
      "loss": 0.8784,
      "step": 239
    },
    {
      "epoch": 0.1477832512315271,
      "grad_norm": 0.8007647097261282,
      "learning_rate": 1.9863240225317868e-05,
      "loss": 0.9312,
      "step": 240
    },
    {
      "epoch": 0.14839901477832512,
      "grad_norm": 0.8355981351102507,
      "learning_rate": 1.9859673342150413e-05,
      "loss": 0.9232,
      "step": 241
    },
    {
      "epoch": 0.14901477832512317,
      "grad_norm": 0.761247632333827,
      "learning_rate": 1.985606086981428e-05,
      "loss": 0.9733,
      "step": 242
    },
    {
      "epoch": 0.14963054187192118,
      "grad_norm": 0.7575740270221389,
      "learning_rate": 1.9852402825012823e-05,
      "loss": 0.8951,
      "step": 243
    },
    {
      "epoch": 0.15024630541871922,
      "grad_norm": 0.765160914764374,
      "learning_rate": 1.984869922466011e-05,
      "loss": 0.9506,
      "step": 244
    },
    {
      "epoch": 0.15086206896551724,
      "grad_norm": 0.7918428235596552,
      "learning_rate": 1.984495008588086e-05,
      "loss": 0.9271,
      "step": 245
    },
    {
      "epoch": 0.15147783251231528,
      "grad_norm": 0.8171106876940029,
      "learning_rate": 1.984115542601034e-05,
      "loss": 0.9018,
      "step": 246
    },
    {
      "epoch": 0.1520935960591133,
      "grad_norm": 0.8115563785082668,
      "learning_rate": 1.9837315262594307e-05,
      "loss": 0.9174,
      "step": 247
    },
    {
      "epoch": 0.15270935960591134,
      "grad_norm": 0.7278823333385184,
      "learning_rate": 1.9833429613388905e-05,
      "loss": 0.9074,
      "step": 248
    },
    {
      "epoch": 0.15332512315270935,
      "grad_norm": 0.7855417352170823,
      "learning_rate": 1.9829498496360607e-05,
      "loss": 0.9517,
      "step": 249
    },
    {
      "epoch": 0.1539408866995074,
      "grad_norm": 0.8174477151964407,
      "learning_rate": 1.982552192968612e-05,
      "loss": 0.9525,
      "step": 250
    },
    {
      "epoch": 0.1545566502463054,
      "grad_norm": 0.8054432322387395,
      "learning_rate": 1.9821499931752292e-05,
      "loss": 0.8825,
      "step": 251
    },
    {
      "epoch": 0.15517241379310345,
      "grad_norm": 0.7982735324252316,
      "learning_rate": 1.9817432521156047e-05,
      "loss": 0.9389,
      "step": 252
    },
    {
      "epoch": 0.15578817733990147,
      "grad_norm": 0.7805107135393612,
      "learning_rate": 1.9813319716704278e-05,
      "loss": 0.9004,
      "step": 253
    },
    {
      "epoch": 0.1564039408866995,
      "grad_norm": 0.7195404709538322,
      "learning_rate": 1.9809161537413775e-05,
      "loss": 0.9323,
      "step": 254
    },
    {
      "epoch": 0.15701970443349753,
      "grad_norm": 0.73218208865218,
      "learning_rate": 1.9804958002511137e-05,
      "loss": 0.8722,
      "step": 255
    },
    {
      "epoch": 0.15763546798029557,
      "grad_norm": 0.7860295483837922,
      "learning_rate": 1.9800709131432665e-05,
      "loss": 0.9731,
      "step": 256
    },
    {
      "epoch": 0.15825123152709358,
      "grad_norm": 0.7556943337361351,
      "learning_rate": 1.9796414943824304e-05,
      "loss": 0.9175,
      "step": 257
    },
    {
      "epoch": 0.15886699507389163,
      "grad_norm": 0.8019976991885589,
      "learning_rate": 1.9792075459541518e-05,
      "loss": 0.9774,
      "step": 258
    },
    {
      "epoch": 0.15948275862068967,
      "grad_norm": 0.8059364032971281,
      "learning_rate": 1.9787690698649217e-05,
      "loss": 0.9063,
      "step": 259
    },
    {
      "epoch": 0.16009852216748768,
      "grad_norm": 0.773456856479164,
      "learning_rate": 1.9783260681421667e-05,
      "loss": 0.9499,
      "step": 260
    },
    {
      "epoch": 0.16071428571428573,
      "grad_norm": 0.7300533905246162,
      "learning_rate": 1.9778785428342386e-05,
      "loss": 0.9439,
      "step": 261
    },
    {
      "epoch": 0.16133004926108374,
      "grad_norm": 0.738792779812286,
      "learning_rate": 1.9774264960104056e-05,
      "loss": 0.912,
      "step": 262
    },
    {
      "epoch": 0.16194581280788178,
      "grad_norm": 0.8729352219027641,
      "learning_rate": 1.9769699297608416e-05,
      "loss": 0.9536,
      "step": 263
    },
    {
      "epoch": 0.1625615763546798,
      "grad_norm": 0.7757847205133515,
      "learning_rate": 1.976508846196619e-05,
      "loss": 0.9161,
      "step": 264
    },
    {
      "epoch": 0.16317733990147784,
      "grad_norm": 0.8166704319869404,
      "learning_rate": 1.9760432474496963e-05,
      "loss": 0.8949,
      "step": 265
    },
    {
      "epoch": 0.16379310344827586,
      "grad_norm": 0.8409421246118747,
      "learning_rate": 1.9755731356729094e-05,
      "loss": 0.9499,
      "step": 266
    },
    {
      "epoch": 0.1644088669950739,
      "grad_norm": 0.808399920067878,
      "learning_rate": 1.975098513039961e-05,
      "loss": 0.9131,
      "step": 267
    },
    {
      "epoch": 0.16502463054187191,
      "grad_norm": 0.823446607809563,
      "learning_rate": 1.9746193817454128e-05,
      "loss": 0.9412,
      "step": 268
    },
    {
      "epoch": 0.16564039408866996,
      "grad_norm": 0.9512153296743574,
      "learning_rate": 1.974135744004672e-05,
      "loss": 0.9435,
      "step": 269
    },
    {
      "epoch": 0.16625615763546797,
      "grad_norm": 0.7740819240310834,
      "learning_rate": 1.973647602053984e-05,
      "loss": 0.9556,
      "step": 270
    },
    {
      "epoch": 0.16687192118226601,
      "grad_norm": 0.818502116185871,
      "learning_rate": 1.973154958150419e-05,
      "loss": 0.9096,
      "step": 271
    },
    {
      "epoch": 0.16748768472906403,
      "grad_norm": 0.8946263806150646,
      "learning_rate": 1.9726578145718654e-05,
      "loss": 0.96,
      "step": 272
    },
    {
      "epoch": 0.16810344827586207,
      "grad_norm": 0.7705057444844188,
      "learning_rate": 1.972156173617016e-05,
      "loss": 0.9149,
      "step": 273
    },
    {
      "epoch": 0.1687192118226601,
      "grad_norm": 0.91081171321787,
      "learning_rate": 1.9716500376053586e-05,
      "loss": 0.9513,
      "step": 274
    },
    {
      "epoch": 0.16933497536945813,
      "grad_norm": 0.7725867627601574,
      "learning_rate": 1.9711394088771658e-05,
      "loss": 0.9454,
      "step": 275
    },
    {
      "epoch": 0.16995073891625614,
      "grad_norm": 0.8957482040725308,
      "learning_rate": 1.9706242897934826e-05,
      "loss": 0.9286,
      "step": 276
    },
    {
      "epoch": 0.1705665024630542,
      "grad_norm": 0.8039307343476086,
      "learning_rate": 1.9701046827361175e-05,
      "loss": 0.9573,
      "step": 277
    },
    {
      "epoch": 0.17118226600985223,
      "grad_norm": 0.8782048108817353,
      "learning_rate": 1.9695805901076308e-05,
      "loss": 0.9396,
      "step": 278
    },
    {
      "epoch": 0.17179802955665024,
      "grad_norm": 0.8366624163131032,
      "learning_rate": 1.969052014331322e-05,
      "loss": 0.9913,
      "step": 279
    },
    {
      "epoch": 0.1724137931034483,
      "grad_norm": 0.7465725236732216,
      "learning_rate": 1.9685189578512206e-05,
      "loss": 0.8888,
      "step": 280
    },
    {
      "epoch": 0.1730295566502463,
      "grad_norm": 0.9467399300342781,
      "learning_rate": 1.9679814231320736e-05,
      "loss": 0.9284,
      "step": 281
    },
    {
      "epoch": 0.17364532019704434,
      "grad_norm": 0.8255017526755606,
      "learning_rate": 1.9674394126593346e-05,
      "loss": 0.9136,
      "step": 282
    },
    {
      "epoch": 0.17426108374384236,
      "grad_norm": 0.8200143146236559,
      "learning_rate": 1.9668929289391524e-05,
      "loss": 0.9674,
      "step": 283
    },
    {
      "epoch": 0.1748768472906404,
      "grad_norm": 0.874681073246467,
      "learning_rate": 1.966341974498359e-05,
      "loss": 0.9228,
      "step": 284
    },
    {
      "epoch": 0.17549261083743842,
      "grad_norm": 0.8341482214962955,
      "learning_rate": 1.9657865518844578e-05,
      "loss": 0.9479,
      "step": 285
    },
    {
      "epoch": 0.17610837438423646,
      "grad_norm": 0.8080442181933382,
      "learning_rate": 1.965226663665613e-05,
      "loss": 0.9237,
      "step": 286
    },
    {
      "epoch": 0.17672413793103448,
      "grad_norm": 0.7939094419784439,
      "learning_rate": 1.9646623124306363e-05,
      "loss": 0.9581,
      "step": 287
    },
    {
      "epoch": 0.17733990147783252,
      "grad_norm": 0.9050370699095787,
      "learning_rate": 1.9640935007889755e-05,
      "loss": 0.9371,
      "step": 288
    },
    {
      "epoch": 0.17795566502463053,
      "grad_norm": 0.7724341183951936,
      "learning_rate": 1.9635202313707028e-05,
      "loss": 0.8789,
      "step": 289
    },
    {
      "epoch": 0.17857142857142858,
      "grad_norm": 0.8270914299894135,
      "learning_rate": 1.962942506826501e-05,
      "loss": 0.919,
      "step": 290
    },
    {
      "epoch": 0.1791871921182266,
      "grad_norm": 0.9544375937270435,
      "learning_rate": 1.9623603298276545e-05,
      "loss": 0.9342,
      "step": 291
    },
    {
      "epoch": 0.17980295566502463,
      "grad_norm": 0.7864959982532571,
      "learning_rate": 1.9617737030660338e-05,
      "loss": 0.9177,
      "step": 292
    },
    {
      "epoch": 0.18041871921182265,
      "grad_norm": 0.7167581959694044,
      "learning_rate": 1.961182629254084e-05,
      "loss": 0.9093,
      "step": 293
    },
    {
      "epoch": 0.1810344827586207,
      "grad_norm": 0.8317338487072818,
      "learning_rate": 1.960587111124813e-05,
      "loss": 0.9792,
      "step": 294
    },
    {
      "epoch": 0.1816502463054187,
      "grad_norm": 0.9178255727144404,
      "learning_rate": 1.9599871514317785e-05,
      "loss": 0.8945,
      "step": 295
    },
    {
      "epoch": 0.18226600985221675,
      "grad_norm": 0.8371627605450963,
      "learning_rate": 1.9593827529490744e-05,
      "loss": 0.9692,
      "step": 296
    },
    {
      "epoch": 0.1828817733990148,
      "grad_norm": 0.765795824426098,
      "learning_rate": 1.9587739184713192e-05,
      "loss": 0.9245,
      "step": 297
    },
    {
      "epoch": 0.1834975369458128,
      "grad_norm": 0.8995499278257717,
      "learning_rate": 1.9581606508136426e-05,
      "loss": 0.9227,
      "step": 298
    },
    {
      "epoch": 0.18411330049261085,
      "grad_norm": 0.9235508417646728,
      "learning_rate": 1.957542952811672e-05,
      "loss": 0.9356,
      "step": 299
    },
    {
      "epoch": 0.18472906403940886,
      "grad_norm": 0.750784718869211,
      "learning_rate": 1.9569208273215204e-05,
      "loss": 0.9471,
      "step": 300
    },
    {
      "epoch": 0.1853448275862069,
      "grad_norm": 0.8993985497856888,
      "learning_rate": 1.9562942772197726e-05,
      "loss": 0.8914,
      "step": 301
    },
    {
      "epoch": 0.18596059113300492,
      "grad_norm": 0.8471075255507667,
      "learning_rate": 1.9556633054034707e-05,
      "loss": 0.9204,
      "step": 302
    },
    {
      "epoch": 0.18657635467980296,
      "grad_norm": 0.7668611103994605,
      "learning_rate": 1.9550279147901037e-05,
      "loss": 0.9682,
      "step": 303
    },
    {
      "epoch": 0.18719211822660098,
      "grad_norm": 0.8621277046658785,
      "learning_rate": 1.9543881083175914e-05,
      "loss": 0.9374,
      "step": 304
    },
    {
      "epoch": 0.18780788177339902,
      "grad_norm": 0.7757854412466699,
      "learning_rate": 1.953743888944271e-05,
      "loss": 0.8969,
      "step": 305
    },
    {
      "epoch": 0.18842364532019704,
      "grad_norm": 0.7976085486664728,
      "learning_rate": 1.9530952596488857e-05,
      "loss": 0.9164,
      "step": 306
    },
    {
      "epoch": 0.18903940886699508,
      "grad_norm": 0.880211871033142,
      "learning_rate": 1.9524422234305677e-05,
      "loss": 0.9479,
      "step": 307
    },
    {
      "epoch": 0.1896551724137931,
      "grad_norm": 0.6950384560075606,
      "learning_rate": 1.951784783308827e-05,
      "loss": 0.9316,
      "step": 308
    },
    {
      "epoch": 0.19027093596059114,
      "grad_norm": 0.7336537914805706,
      "learning_rate": 1.9511229423235357e-05,
      "loss": 0.9249,
      "step": 309
    },
    {
      "epoch": 0.19088669950738915,
      "grad_norm": 0.7413111577564022,
      "learning_rate": 1.950456703534915e-05,
      "loss": 0.9088,
      "step": 310
    },
    {
      "epoch": 0.1915024630541872,
      "grad_norm": 0.7778753120819494,
      "learning_rate": 1.9497860700235208e-05,
      "loss": 0.9009,
      "step": 311
    },
    {
      "epoch": 0.1921182266009852,
      "grad_norm": 0.7618619042194296,
      "learning_rate": 1.9491110448902286e-05,
      "loss": 0.8736,
      "step": 312
    },
    {
      "epoch": 0.19273399014778325,
      "grad_norm": 0.7916462914577406,
      "learning_rate": 1.9484316312562204e-05,
      "loss": 0.9398,
      "step": 313
    },
    {
      "epoch": 0.1933497536945813,
      "grad_norm": 0.7664024948322994,
      "learning_rate": 1.9477478322629704e-05,
      "loss": 0.9214,
      "step": 314
    },
    {
      "epoch": 0.1939655172413793,
      "grad_norm": 0.7189134485376704,
      "learning_rate": 1.9470596510722285e-05,
      "loss": 0.9892,
      "step": 315
    },
    {
      "epoch": 0.19458128078817735,
      "grad_norm": 0.7591770640338076,
      "learning_rate": 1.946367090866008e-05,
      "loss": 0.923,
      "step": 316
    },
    {
      "epoch": 0.19519704433497537,
      "grad_norm": 0.8212741084944802,
      "learning_rate": 1.945670154846569e-05,
      "loss": 0.9619,
      "step": 317
    },
    {
      "epoch": 0.1958128078817734,
      "grad_norm": 0.7630368384976093,
      "learning_rate": 1.944968846236406e-05,
      "loss": 0.9252,
      "step": 318
    },
    {
      "epoch": 0.19642857142857142,
      "grad_norm": 0.7280853853328945,
      "learning_rate": 1.9442631682782295e-05,
      "loss": 0.9373,
      "step": 319
    },
    {
      "epoch": 0.19704433497536947,
      "grad_norm": 0.837083415681938,
      "learning_rate": 1.9435531242349545e-05,
      "loss": 0.9409,
      "step": 320
    },
    {
      "epoch": 0.19766009852216748,
      "grad_norm": 0.7407559970272712,
      "learning_rate": 1.942838717389684e-05,
      "loss": 0.9056,
      "step": 321
    },
    {
      "epoch": 0.19827586206896552,
      "grad_norm": 0.7856813603543013,
      "learning_rate": 1.942119951045692e-05,
      "loss": 0.9413,
      "step": 322
    },
    {
      "epoch": 0.19889162561576354,
      "grad_norm": 0.7599899793244124,
      "learning_rate": 1.9413968285264122e-05,
      "loss": 0.909,
      "step": 323
    },
    {
      "epoch": 0.19950738916256158,
      "grad_norm": 0.8050859430409764,
      "learning_rate": 1.9406693531754192e-05,
      "loss": 0.9248,
      "step": 324
    },
    {
      "epoch": 0.2001231527093596,
      "grad_norm": 0.792477964968095,
      "learning_rate": 1.9399375283564134e-05,
      "loss": 0.9464,
      "step": 325
    },
    {
      "epoch": 0.20073891625615764,
      "grad_norm": 0.7341704081770669,
      "learning_rate": 1.9392013574532083e-05,
      "loss": 0.9011,
      "step": 326
    },
    {
      "epoch": 0.20135467980295566,
      "grad_norm": 0.7832095347072934,
      "learning_rate": 1.938460843869711e-05,
      "loss": 0.9463,
      "step": 327
    },
    {
      "epoch": 0.2019704433497537,
      "grad_norm": 0.8610313703423595,
      "learning_rate": 1.9377159910299094e-05,
      "loss": 0.9009,
      "step": 328
    },
    {
      "epoch": 0.2025862068965517,
      "grad_norm": 0.7300077266098178,
      "learning_rate": 1.936966802377855e-05,
      "loss": 0.8944,
      "step": 329
    },
    {
      "epoch": 0.20320197044334976,
      "grad_norm": 0.8002375322280387,
      "learning_rate": 1.9362132813776472e-05,
      "loss": 0.9446,
      "step": 330
    },
    {
      "epoch": 0.20381773399014777,
      "grad_norm": 0.7074002589406563,
      "learning_rate": 1.9354554315134166e-05,
      "loss": 0.87,
      "step": 331
    },
    {
      "epoch": 0.2044334975369458,
      "grad_norm": 0.8128426199274593,
      "learning_rate": 1.934693256289311e-05,
      "loss": 0.8935,
      "step": 332
    },
    {
      "epoch": 0.20504926108374386,
      "grad_norm": 0.7981416322416451,
      "learning_rate": 1.9339267592294764e-05,
      "loss": 0.9022,
      "step": 333
    },
    {
      "epoch": 0.20566502463054187,
      "grad_norm": 0.7261881007203638,
      "learning_rate": 1.9331559438780432e-05,
      "loss": 0.9036,
      "step": 334
    },
    {
      "epoch": 0.2062807881773399,
      "grad_norm": 0.753404506913586,
      "learning_rate": 1.9323808137991084e-05,
      "loss": 0.9322,
      "step": 335
    },
    {
      "epoch": 0.20689655172413793,
      "grad_norm": 0.8277311418719314,
      "learning_rate": 1.9316013725767192e-05,
      "loss": 0.8618,
      "step": 336
    },
    {
      "epoch": 0.20751231527093597,
      "grad_norm": 0.7698315798263946,
      "learning_rate": 1.9308176238148565e-05,
      "loss": 0.927,
      "step": 337
    },
    {
      "epoch": 0.20812807881773399,
      "grad_norm": 0.916642639190395,
      "learning_rate": 1.930029571137419e-05,
      "loss": 0.9205,
      "step": 338
    },
    {
      "epoch": 0.20874384236453203,
      "grad_norm": 0.8554735254916017,
      "learning_rate": 1.9292372181882052e-05,
      "loss": 0.9071,
      "step": 339
    },
    {
      "epoch": 0.20935960591133004,
      "grad_norm": 0.7720387299280486,
      "learning_rate": 1.9284405686308982e-05,
      "loss": 0.9236,
      "step": 340
    },
    {
      "epoch": 0.20997536945812809,
      "grad_norm": 0.9268896580061773,
      "learning_rate": 1.9276396261490468e-05,
      "loss": 0.9078,
      "step": 341
    },
    {
      "epoch": 0.2105911330049261,
      "grad_norm": 0.8011928317198055,
      "learning_rate": 1.9268343944460494e-05,
      "loss": 0.9274,
      "step": 342
    },
    {
      "epoch": 0.21120689655172414,
      "grad_norm": 0.9662939159527071,
      "learning_rate": 1.9260248772451378e-05,
      "loss": 0.9724,
      "step": 343
    },
    {
      "epoch": 0.21182266009852216,
      "grad_norm": 0.9326833322336923,
      "learning_rate": 1.9252110782893582e-05,
      "loss": 0.9021,
      "step": 344
    },
    {
      "epoch": 0.2124384236453202,
      "grad_norm": 0.7223127977809748,
      "learning_rate": 1.924393001341555e-05,
      "loss": 0.924,
      "step": 345
    },
    {
      "epoch": 0.21305418719211822,
      "grad_norm": 0.8404194229060159,
      "learning_rate": 1.923570650184354e-05,
      "loss": 0.9589,
      "step": 346
    },
    {
      "epoch": 0.21366995073891626,
      "grad_norm": 0.8165681660557852,
      "learning_rate": 1.922744028620143e-05,
      "loss": 0.8703,
      "step": 347
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 0.7159791882313546,
      "learning_rate": 1.9219131404710555e-05,
      "loss": 0.9003,
      "step": 348
    },
    {
      "epoch": 0.21490147783251232,
      "grad_norm": 0.8262503553805308,
      "learning_rate": 1.921077989578953e-05,
      "loss": 0.9327,
      "step": 349
    },
    {
      "epoch": 0.21551724137931033,
      "grad_norm": 0.7764923443898881,
      "learning_rate": 1.9202385798054073e-05,
      "loss": 0.9183,
      "step": 350
    },
    {
      "epoch": 0.21613300492610837,
      "grad_norm": 0.7819486205187545,
      "learning_rate": 1.9193949150316825e-05,
      "loss": 0.8979,
      "step": 351
    },
    {
      "epoch": 0.21674876847290642,
      "grad_norm": 0.8543827092991148,
      "learning_rate": 1.9185469991587166e-05,
      "loss": 0.9009,
      "step": 352
    },
    {
      "epoch": 0.21736453201970443,
      "grad_norm": 0.9991842007470291,
      "learning_rate": 1.917694836107104e-05,
      "loss": 0.9458,
      "step": 353
    },
    {
      "epoch": 0.21798029556650247,
      "grad_norm": 0.779877584954099,
      "learning_rate": 1.916838429817077e-05,
      "loss": 0.9498,
      "step": 354
    },
    {
      "epoch": 0.2185960591133005,
      "grad_norm": 0.8983037497197904,
      "learning_rate": 1.9159777842484878e-05,
      "loss": 0.9272,
      "step": 355
    },
    {
      "epoch": 0.21921182266009853,
      "grad_norm": 0.9349132060291829,
      "learning_rate": 1.915112903380791e-05,
      "loss": 0.901,
      "step": 356
    },
    {
      "epoch": 0.21982758620689655,
      "grad_norm": 0.7759637787072661,
      "learning_rate": 1.914243791213023e-05,
      "loss": 0.9209,
      "step": 357
    },
    {
      "epoch": 0.2204433497536946,
      "grad_norm": 0.8478190370330694,
      "learning_rate": 1.913370451763786e-05,
      "loss": 0.9557,
      "step": 358
    },
    {
      "epoch": 0.2210591133004926,
      "grad_norm": 0.8773931356634249,
      "learning_rate": 1.912492889071228e-05,
      "loss": 0.9782,
      "step": 359
    },
    {
      "epoch": 0.22167487684729065,
      "grad_norm": 0.7630942518293452,
      "learning_rate": 1.911611107193024e-05,
      "loss": 0.9208,
      "step": 360
    },
    {
      "epoch": 0.22229064039408866,
      "grad_norm": 0.7392236830022609,
      "learning_rate": 1.9107251102063582e-05,
      "loss": 0.9122,
      "step": 361
    },
    {
      "epoch": 0.2229064039408867,
      "grad_norm": 0.859701378409601,
      "learning_rate": 1.909834902207905e-05,
      "loss": 0.954,
      "step": 362
    },
    {
      "epoch": 0.22352216748768472,
      "grad_norm": 0.7782181419421879,
      "learning_rate": 1.908940487313808e-05,
      "loss": 0.9485,
      "step": 363
    },
    {
      "epoch": 0.22413793103448276,
      "grad_norm": 0.680528341187256,
      "learning_rate": 1.908041869659665e-05,
      "loss": 0.9053,
      "step": 364
    },
    {
      "epoch": 0.22475369458128078,
      "grad_norm": 0.7869725983044522,
      "learning_rate": 1.9071390534005045e-05,
      "loss": 0.8733,
      "step": 365
    },
    {
      "epoch": 0.22536945812807882,
      "grad_norm": 0.8609706951671159,
      "learning_rate": 1.90623204271077e-05,
      "loss": 0.9484,
      "step": 366
    },
    {
      "epoch": 0.22598522167487683,
      "grad_norm": 0.8093685289807101,
      "learning_rate": 1.905320841784298e-05,
      "loss": 0.9453,
      "step": 367
    },
    {
      "epoch": 0.22660098522167488,
      "grad_norm": 0.8203240519063807,
      "learning_rate": 1.9044054548343005e-05,
      "loss": 0.9228,
      "step": 368
    },
    {
      "epoch": 0.2272167487684729,
      "grad_norm": 0.7703630132377759,
      "learning_rate": 1.9034858860933444e-05,
      "loss": 0.9252,
      "step": 369
    },
    {
      "epoch": 0.22783251231527094,
      "grad_norm": 0.7571931245301118,
      "learning_rate": 1.9025621398133333e-05,
      "loss": 0.9229,
      "step": 370
    },
    {
      "epoch": 0.22844827586206898,
      "grad_norm": 0.7690719964753099,
      "learning_rate": 1.901634220265486e-05,
      "loss": 0.9104,
      "step": 371
    },
    {
      "epoch": 0.229064039408867,
      "grad_norm": 0.7310690413909582,
      "learning_rate": 1.900702131740318e-05,
      "loss": 0.9104,
      "step": 372
    },
    {
      "epoch": 0.22967980295566504,
      "grad_norm": 0.747404494929283,
      "learning_rate": 1.8997658785476215e-05,
      "loss": 0.9293,
      "step": 373
    },
    {
      "epoch": 0.23029556650246305,
      "grad_norm": 0.7603377333004162,
      "learning_rate": 1.898825465016444e-05,
      "loss": 0.8788,
      "step": 374
    },
    {
      "epoch": 0.2309113300492611,
      "grad_norm": 0.8283410721747364,
      "learning_rate": 1.8978808954950722e-05,
      "loss": 0.9282,
      "step": 375
    },
    {
      "epoch": 0.2315270935960591,
      "grad_norm": 0.8346967454923524,
      "learning_rate": 1.8969321743510067e-05,
      "loss": 0.9055,
      "step": 376
    },
    {
      "epoch": 0.23214285714285715,
      "grad_norm": 0.7423389981901787,
      "learning_rate": 1.895979305970946e-05,
      "loss": 0.933,
      "step": 377
    },
    {
      "epoch": 0.23275862068965517,
      "grad_norm": 0.8131608926254003,
      "learning_rate": 1.8950222947607627e-05,
      "loss": 0.9455,
      "step": 378
    },
    {
      "epoch": 0.2333743842364532,
      "grad_norm": 0.7432295853974047,
      "learning_rate": 1.894061145145488e-05,
      "loss": 0.9306,
      "step": 379
    },
    {
      "epoch": 0.23399014778325122,
      "grad_norm": 0.7766812381516162,
      "learning_rate": 1.8930958615692854e-05,
      "loss": 0.8855,
      "step": 380
    },
    {
      "epoch": 0.23460591133004927,
      "grad_norm": 0.763506016269915,
      "learning_rate": 1.8921264484954344e-05,
      "loss": 0.9159,
      "step": 381
    },
    {
      "epoch": 0.23522167487684728,
      "grad_norm": 0.747606693417055,
      "learning_rate": 1.8911529104063093e-05,
      "loss": 0.9729,
      "step": 382
    },
    {
      "epoch": 0.23583743842364532,
      "grad_norm": 0.7787579301569807,
      "learning_rate": 1.890175251803355e-05,
      "loss": 0.9278,
      "step": 383
    },
    {
      "epoch": 0.23645320197044334,
      "grad_norm": 0.8277651409665697,
      "learning_rate": 1.8891934772070715e-05,
      "loss": 0.9291,
      "step": 384
    },
    {
      "epoch": 0.23706896551724138,
      "grad_norm": 0.7758168413372164,
      "learning_rate": 1.8882075911569887e-05,
      "loss": 0.9513,
      "step": 385
    },
    {
      "epoch": 0.2376847290640394,
      "grad_norm": 0.7224831737449906,
      "learning_rate": 1.887217598211648e-05,
      "loss": 0.9219,
      "step": 386
    },
    {
      "epoch": 0.23830049261083744,
      "grad_norm": 0.7709800434131463,
      "learning_rate": 1.8862235029485803e-05,
      "loss": 0.9393,
      "step": 387
    },
    {
      "epoch": 0.23891625615763548,
      "grad_norm": 0.7443011866053137,
      "learning_rate": 1.8852253099642835e-05,
      "loss": 0.9137,
      "step": 388
    },
    {
      "epoch": 0.2395320197044335,
      "grad_norm": 0.8275945285666486,
      "learning_rate": 1.884223023874204e-05,
      "loss": 0.9315,
      "step": 389
    },
    {
      "epoch": 0.24014778325123154,
      "grad_norm": 0.7403460886051261,
      "learning_rate": 1.8832166493127128e-05,
      "loss": 0.9118,
      "step": 390
    },
    {
      "epoch": 0.24076354679802955,
      "grad_norm": 0.842531009552115,
      "learning_rate": 1.8822061909330866e-05,
      "loss": 0.9073,
      "step": 391
    },
    {
      "epoch": 0.2413793103448276,
      "grad_norm": 0.8282790621426415,
      "learning_rate": 1.881191653407483e-05,
      "loss": 0.9107,
      "step": 392
    },
    {
      "epoch": 0.2419950738916256,
      "grad_norm": 0.7759272313606862,
      "learning_rate": 1.8801730414269227e-05,
      "loss": 0.9677,
      "step": 393
    },
    {
      "epoch": 0.24261083743842365,
      "grad_norm": 0.7220067629347077,
      "learning_rate": 1.8791503597012643e-05,
      "loss": 0.8769,
      "step": 394
    },
    {
      "epoch": 0.24322660098522167,
      "grad_norm": 0.7167124451104708,
      "learning_rate": 1.8781236129591847e-05,
      "loss": 0.8859,
      "step": 395
    },
    {
      "epoch": 0.2438423645320197,
      "grad_norm": 0.8164790916428831,
      "learning_rate": 1.8770928059481574e-05,
      "loss": 0.8929,
      "step": 396
    },
    {
      "epoch": 0.24445812807881773,
      "grad_norm": 0.7155020874194057,
      "learning_rate": 1.8760579434344283e-05,
      "loss": 0.931,
      "step": 397
    },
    {
      "epoch": 0.24507389162561577,
      "grad_norm": 0.7387318912377534,
      "learning_rate": 1.8750190302029955e-05,
      "loss": 0.9126,
      "step": 398
    },
    {
      "epoch": 0.24568965517241378,
      "grad_norm": 0.7681415946686942,
      "learning_rate": 1.8739760710575882e-05,
      "loss": 0.9454,
      "step": 399
    },
    {
      "epoch": 0.24630541871921183,
      "grad_norm": 0.805324566571223,
      "learning_rate": 1.8729290708206412e-05,
      "loss": 0.9521,
      "step": 400
    },
    {
      "epoch": 0.24630541871921183,
      "eval_loss": 0.9253853559494019,
      "eval_runtime": 524.641,
      "eval_samples_per_second": 44.049,
      "eval_steps_per_second": 0.69,
      "step": 400
    },
    {
      "epoch": 0.24692118226600984,
      "grad_norm": 0.7651031068473965,
      "learning_rate": 1.8718780343332757e-05,
      "loss": 0.8913,
      "step": 401
    },
    {
      "epoch": 0.24753694581280788,
      "grad_norm": 0.8751132705691878,
      "learning_rate": 1.8708229664552754e-05,
      "loss": 0.9002,
      "step": 402
    },
    {
      "epoch": 0.2481527093596059,
      "grad_norm": 0.7560523940392909,
      "learning_rate": 1.8697638720650647e-05,
      "loss": 0.9632,
      "step": 403
    },
    {
      "epoch": 0.24876847290640394,
      "grad_norm": 0.8495415128192424,
      "learning_rate": 1.868700756059685e-05,
      "loss": 0.9064,
      "step": 404
    },
    {
      "epoch": 0.24938423645320196,
      "grad_norm": 0.8057339202439634,
      "learning_rate": 1.867633623354773e-05,
      "loss": 0.9532,
      "step": 405
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.722529164479765,
      "learning_rate": 1.8665624788845387e-05,
      "loss": 0.9043,
      "step": 406
    },
    {
      "epoch": 0.25061576354679804,
      "grad_norm": 0.724134246312139,
      "learning_rate": 1.8654873276017408e-05,
      "loss": 0.8773,
      "step": 407
    },
    {
      "epoch": 0.2512315270935961,
      "grad_norm": 0.760768629407117,
      "learning_rate": 1.864408174477665e-05,
      "loss": 0.9036,
      "step": 408
    },
    {
      "epoch": 0.2518472906403941,
      "grad_norm": 0.7187391975091129,
      "learning_rate": 1.863325024502101e-05,
      "loss": 0.9698,
      "step": 409
    },
    {
      "epoch": 0.2524630541871921,
      "grad_norm": 0.7672210570107554,
      "learning_rate": 1.8622378826833186e-05,
      "loss": 0.8766,
      "step": 410
    },
    {
      "epoch": 0.25307881773399016,
      "grad_norm": 0.7333361144397894,
      "learning_rate": 1.861146754048045e-05,
      "loss": 0.9552,
      "step": 411
    },
    {
      "epoch": 0.2536945812807882,
      "grad_norm": 0.7852630888641706,
      "learning_rate": 1.860051643641443e-05,
      "loss": 0.9361,
      "step": 412
    },
    {
      "epoch": 0.2543103448275862,
      "grad_norm": 0.7183389654359763,
      "learning_rate": 1.8589525565270846e-05,
      "loss": 0.9022,
      "step": 413
    },
    {
      "epoch": 0.25492610837438423,
      "grad_norm": 0.8862705004271282,
      "learning_rate": 1.8578494977869304e-05,
      "loss": 0.9004,
      "step": 414
    },
    {
      "epoch": 0.2555418719211823,
      "grad_norm": 0.6839679489296381,
      "learning_rate": 1.856742472521304e-05,
      "loss": 0.8995,
      "step": 415
    },
    {
      "epoch": 0.2561576354679803,
      "grad_norm": 0.8585477637248662,
      "learning_rate": 1.8556314858488708e-05,
      "loss": 0.913,
      "step": 416
    },
    {
      "epoch": 0.2567733990147783,
      "grad_norm": 0.8011725272856874,
      "learning_rate": 1.8545165429066123e-05,
      "loss": 0.8776,
      "step": 417
    },
    {
      "epoch": 0.25738916256157635,
      "grad_norm": 0.7136614773367724,
      "learning_rate": 1.853397648849802e-05,
      "loss": 0.908,
      "step": 418
    },
    {
      "epoch": 0.2580049261083744,
      "grad_norm": 0.7405916379143913,
      "learning_rate": 1.8522748088519838e-05,
      "loss": 0.9376,
      "step": 419
    },
    {
      "epoch": 0.25862068965517243,
      "grad_norm": 0.7553681510234944,
      "learning_rate": 1.8511480281049475e-05,
      "loss": 0.9249,
      "step": 420
    },
    {
      "epoch": 0.2592364532019704,
      "grad_norm": 0.757017524040465,
      "learning_rate": 1.8500173118187018e-05,
      "loss": 0.8685,
      "step": 421
    },
    {
      "epoch": 0.25985221674876846,
      "grad_norm": 0.715093764749469,
      "learning_rate": 1.8488826652214552e-05,
      "loss": 0.9511,
      "step": 422
    },
    {
      "epoch": 0.2604679802955665,
      "grad_norm": 0.7102025891312678,
      "learning_rate": 1.8477440935595873e-05,
      "loss": 0.9329,
      "step": 423
    },
    {
      "epoch": 0.26108374384236455,
      "grad_norm": 0.7396501098787446,
      "learning_rate": 1.846601602097627e-05,
      "loss": 0.9283,
      "step": 424
    },
    {
      "epoch": 0.2616995073891626,
      "grad_norm": 0.6850859109013904,
      "learning_rate": 1.8454551961182276e-05,
      "loss": 0.9121,
      "step": 425
    },
    {
      "epoch": 0.2623152709359606,
      "grad_norm": 0.7230918606529195,
      "learning_rate": 1.8443048809221424e-05,
      "loss": 0.9187,
      "step": 426
    },
    {
      "epoch": 0.2629310344827586,
      "grad_norm": 0.8157545823958939,
      "learning_rate": 1.8431506618282e-05,
      "loss": 0.8825,
      "step": 427
    },
    {
      "epoch": 0.26354679802955666,
      "grad_norm": 0.6872674322074096,
      "learning_rate": 1.8419925441732803e-05,
      "loss": 0.8942,
      "step": 428
    },
    {
      "epoch": 0.2641625615763547,
      "grad_norm": 0.6924203363192046,
      "learning_rate": 1.8408305333122888e-05,
      "loss": 0.972,
      "step": 429
    },
    {
      "epoch": 0.2647783251231527,
      "grad_norm": 0.745262935793078,
      "learning_rate": 1.8396646346181327e-05,
      "loss": 0.922,
      "step": 430
    },
    {
      "epoch": 0.26539408866995073,
      "grad_norm": 0.6990227237779921,
      "learning_rate": 1.8384948534816954e-05,
      "loss": 0.9083,
      "step": 431
    },
    {
      "epoch": 0.2660098522167488,
      "grad_norm": 0.6827363974377901,
      "learning_rate": 1.8373211953118124e-05,
      "loss": 0.8711,
      "step": 432
    },
    {
      "epoch": 0.2666256157635468,
      "grad_norm": 0.730141138924111,
      "learning_rate": 1.8361436655352456e-05,
      "loss": 0.9084,
      "step": 433
    },
    {
      "epoch": 0.2672413793103448,
      "grad_norm": 0.7187148098691404,
      "learning_rate": 1.8349622695966588e-05,
      "loss": 0.8877,
      "step": 434
    },
    {
      "epoch": 0.26785714285714285,
      "grad_norm": 0.7787673238959798,
      "learning_rate": 1.8337770129585918e-05,
      "loss": 0.9261,
      "step": 435
    },
    {
      "epoch": 0.2684729064039409,
      "grad_norm": 0.7167114341803934,
      "learning_rate": 1.832587901101435e-05,
      "loss": 0.9418,
      "step": 436
    },
    {
      "epoch": 0.26908866995073893,
      "grad_norm": 0.726836999095157,
      "learning_rate": 1.831394939523406e-05,
      "loss": 0.919,
      "step": 437
    },
    {
      "epoch": 0.2697044334975369,
      "grad_norm": 0.7654179615840212,
      "learning_rate": 1.8301981337405214e-05,
      "loss": 0.9626,
      "step": 438
    },
    {
      "epoch": 0.27032019704433496,
      "grad_norm": 0.7548674750482893,
      "learning_rate": 1.828997489286573e-05,
      "loss": 0.9226,
      "step": 439
    },
    {
      "epoch": 0.270935960591133,
      "grad_norm": 0.7383366366799822,
      "learning_rate": 1.8277930117131025e-05,
      "loss": 0.8887,
      "step": 440
    },
    {
      "epoch": 0.27155172413793105,
      "grad_norm": 0.7482032524119224,
      "learning_rate": 1.8265847065893738e-05,
      "loss": 0.8925,
      "step": 441
    },
    {
      "epoch": 0.27216748768472904,
      "grad_norm": 0.7261926105967714,
      "learning_rate": 1.8253725795023504e-05,
      "loss": 0.9039,
      "step": 442
    },
    {
      "epoch": 0.2727832512315271,
      "grad_norm": 0.7402671023507161,
      "learning_rate": 1.8241566360566664e-05,
      "loss": 0.9051,
      "step": 443
    },
    {
      "epoch": 0.2733990147783251,
      "grad_norm": 0.8040983636955732,
      "learning_rate": 1.8229368818746026e-05,
      "loss": 0.9687,
      "step": 444
    },
    {
      "epoch": 0.27401477832512317,
      "grad_norm": 0.6933707732651354,
      "learning_rate": 1.8217133225960597e-05,
      "loss": 0.9379,
      "step": 445
    },
    {
      "epoch": 0.2746305418719212,
      "grad_norm": 0.758817648533376,
      "learning_rate": 1.8204859638785327e-05,
      "loss": 0.9228,
      "step": 446
    },
    {
      "epoch": 0.2752463054187192,
      "grad_norm": 0.7705552882313036,
      "learning_rate": 1.8192548113970838e-05,
      "loss": 0.91,
      "step": 447
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 0.7914279766663801,
      "learning_rate": 1.8180198708443172e-05,
      "loss": 0.9302,
      "step": 448
    },
    {
      "epoch": 0.2764778325123153,
      "grad_norm": 0.7535845437080304,
      "learning_rate": 1.8167811479303533e-05,
      "loss": 0.9072,
      "step": 449
    },
    {
      "epoch": 0.2770935960591133,
      "grad_norm": 0.8934482585040108,
      "learning_rate": 1.8155386483827995e-05,
      "loss": 0.9296,
      "step": 450
    },
    {
      "epoch": 0.2777093596059113,
      "grad_norm": 0.7820381898673626,
      "learning_rate": 1.814292377946727e-05,
      "loss": 0.8739,
      "step": 451
    },
    {
      "epoch": 0.27832512315270935,
      "grad_norm": 0.7542784926748579,
      "learning_rate": 1.813042342384642e-05,
      "loss": 0.9258,
      "step": 452
    },
    {
      "epoch": 0.2789408866995074,
      "grad_norm": 0.7816544357236844,
      "learning_rate": 1.8117885474764613e-05,
      "loss": 0.9143,
      "step": 453
    },
    {
      "epoch": 0.27955665024630544,
      "grad_norm": 0.774924239743098,
      "learning_rate": 1.8105309990194823e-05,
      "loss": 0.9265,
      "step": 454
    },
    {
      "epoch": 0.2801724137931034,
      "grad_norm": 0.7549508497203472,
      "learning_rate": 1.8092697028283598e-05,
      "loss": 0.9488,
      "step": 455
    },
    {
      "epoch": 0.28078817733990147,
      "grad_norm": 0.8666578103437991,
      "learning_rate": 1.8080046647350756e-05,
      "loss": 0.917,
      "step": 456
    },
    {
      "epoch": 0.2814039408866995,
      "grad_norm": 0.7760132448909304,
      "learning_rate": 1.8067358905889148e-05,
      "loss": 0.8987,
      "step": 457
    },
    {
      "epoch": 0.28201970443349755,
      "grad_norm": 0.7188144207941528,
      "learning_rate": 1.805463386256437e-05,
      "loss": 0.955,
      "step": 458
    },
    {
      "epoch": 0.28263546798029554,
      "grad_norm": 0.7535389245587262,
      "learning_rate": 1.8041871576214483e-05,
      "loss": 0.9581,
      "step": 459
    },
    {
      "epoch": 0.2832512315270936,
      "grad_norm": 0.7924484208520991,
      "learning_rate": 1.8029072105849767e-05,
      "loss": 0.9161,
      "step": 460
    },
    {
      "epoch": 0.2838669950738916,
      "grad_norm": 0.7694823049018515,
      "learning_rate": 1.8016235510652428e-05,
      "loss": 0.9411,
      "step": 461
    },
    {
      "epoch": 0.28448275862068967,
      "grad_norm": 0.7325938606007679,
      "learning_rate": 1.8003361849976327e-05,
      "loss": 0.9729,
      "step": 462
    },
    {
      "epoch": 0.2850985221674877,
      "grad_norm": 0.7406292579367867,
      "learning_rate": 1.799045118334671e-05,
      "loss": 0.9271,
      "step": 463
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.8363801686373288,
      "learning_rate": 1.7977503570459936e-05,
      "loss": 0.9236,
      "step": 464
    },
    {
      "epoch": 0.28633004926108374,
      "grad_norm": 0.7092004889728918,
      "learning_rate": 1.7964519071183188e-05,
      "loss": 0.9049,
      "step": 465
    },
    {
      "epoch": 0.2869458128078818,
      "grad_norm": 0.7003790493078939,
      "learning_rate": 1.795149774555421e-05,
      "loss": 0.9383,
      "step": 466
    },
    {
      "epoch": 0.2875615763546798,
      "grad_norm": 0.7608809585849616,
      "learning_rate": 1.793843965378102e-05,
      "loss": 0.9521,
      "step": 467
    },
    {
      "epoch": 0.2881773399014778,
      "grad_norm": 0.7736517455654832,
      "learning_rate": 1.792534485624164e-05,
      "loss": 0.9454,
      "step": 468
    },
    {
      "epoch": 0.28879310344827586,
      "grad_norm": 0.7410646849144276,
      "learning_rate": 1.791221341348381e-05,
      "loss": 0.9125,
      "step": 469
    },
    {
      "epoch": 0.2894088669950739,
      "grad_norm": 0.7367949558811753,
      "learning_rate": 1.789904538622471e-05,
      "loss": 0.9171,
      "step": 470
    },
    {
      "epoch": 0.29002463054187194,
      "grad_norm": 0.8211950855042173,
      "learning_rate": 1.7885840835350672e-05,
      "loss": 0.9387,
      "step": 471
    },
    {
      "epoch": 0.29064039408866993,
      "grad_norm": 0.7041817084762423,
      "learning_rate": 1.7872599821916922e-05,
      "loss": 0.9215,
      "step": 472
    },
    {
      "epoch": 0.29125615763546797,
      "grad_norm": 0.7119333586965787,
      "learning_rate": 1.785932240714727e-05,
      "loss": 0.9019,
      "step": 473
    },
    {
      "epoch": 0.291871921182266,
      "grad_norm": 0.813591755339563,
      "learning_rate": 1.7846008652433843e-05,
      "loss": 0.8907,
      "step": 474
    },
    {
      "epoch": 0.29248768472906406,
      "grad_norm": 0.8500781287246608,
      "learning_rate": 1.7832658619336794e-05,
      "loss": 0.8937,
      "step": 475
    },
    {
      "epoch": 0.29310344827586204,
      "grad_norm": 0.7812008749561193,
      "learning_rate": 1.7819272369584016e-05,
      "loss": 0.9434,
      "step": 476
    },
    {
      "epoch": 0.2937192118226601,
      "grad_norm": 0.7761760175721167,
      "learning_rate": 1.780584996507087e-05,
      "loss": 0.8875,
      "step": 477
    },
    {
      "epoch": 0.29433497536945813,
      "grad_norm": 0.8202424395402113,
      "learning_rate": 1.7792391467859886e-05,
      "loss": 0.9817,
      "step": 478
    },
    {
      "epoch": 0.29495073891625617,
      "grad_norm": 0.7849150123641426,
      "learning_rate": 1.777889694018048e-05,
      "loss": 0.9418,
      "step": 479
    },
    {
      "epoch": 0.2955665024630542,
      "grad_norm": 0.7738889237481468,
      "learning_rate": 1.7765366444428655e-05,
      "loss": 0.8751,
      "step": 480
    },
    {
      "epoch": 0.2961822660098522,
      "grad_norm": 0.7478640182382722,
      "learning_rate": 1.7751800043166745e-05,
      "loss": 0.8999,
      "step": 481
    },
    {
      "epoch": 0.29679802955665024,
      "grad_norm": 0.8773238515711607,
      "learning_rate": 1.7738197799123084e-05,
      "loss": 0.9561,
      "step": 482
    },
    {
      "epoch": 0.2974137931034483,
      "grad_norm": 0.8850935845127047,
      "learning_rate": 1.7724559775191744e-05,
      "loss": 0.9393,
      "step": 483
    },
    {
      "epoch": 0.29802955665024633,
      "grad_norm": 0.7795625038443592,
      "learning_rate": 1.7710886034432237e-05,
      "loss": 0.9038,
      "step": 484
    },
    {
      "epoch": 0.2986453201970443,
      "grad_norm": 0.7040977797551838,
      "learning_rate": 1.7697176640069217e-05,
      "loss": 0.8836,
      "step": 485
    },
    {
      "epoch": 0.29926108374384236,
      "grad_norm": 0.7972284526710359,
      "learning_rate": 1.7683431655492203e-05,
      "loss": 0.9541,
      "step": 486
    },
    {
      "epoch": 0.2998768472906404,
      "grad_norm": 0.7518324916303603,
      "learning_rate": 1.7669651144255265e-05,
      "loss": 0.8789,
      "step": 487
    },
    {
      "epoch": 0.30049261083743845,
      "grad_norm": 0.8417130710296228,
      "learning_rate": 1.765583517007675e-05,
      "loss": 0.9184,
      "step": 488
    },
    {
      "epoch": 0.30110837438423643,
      "grad_norm": 0.8309159872330002,
      "learning_rate": 1.7641983796838972e-05,
      "loss": 0.8796,
      "step": 489
    },
    {
      "epoch": 0.3017241379310345,
      "grad_norm": 0.6882113439982166,
      "learning_rate": 1.762809708858793e-05,
      "loss": 0.886,
      "step": 490
    },
    {
      "epoch": 0.3023399014778325,
      "grad_norm": 0.7754495631088804,
      "learning_rate": 1.7614175109532998e-05,
      "loss": 0.9314,
      "step": 491
    },
    {
      "epoch": 0.30295566502463056,
      "grad_norm": 0.8296703137132061,
      "learning_rate": 1.7600217924046637e-05,
      "loss": 0.9077,
      "step": 492
    },
    {
      "epoch": 0.30357142857142855,
      "grad_norm": 0.7348496889665732,
      "learning_rate": 1.7586225596664102e-05,
      "loss": 0.933,
      "step": 493
    },
    {
      "epoch": 0.3041871921182266,
      "grad_norm": 0.7490850578235511,
      "learning_rate": 1.757219819208313e-05,
      "loss": 0.9347,
      "step": 494
    },
    {
      "epoch": 0.30480295566502463,
      "grad_norm": 0.8135504690965311,
      "learning_rate": 1.7558135775163645e-05,
      "loss": 0.8954,
      "step": 495
    },
    {
      "epoch": 0.3054187192118227,
      "grad_norm": 0.7498863578943734,
      "learning_rate": 1.7544038410927474e-05,
      "loss": 0.925,
      "step": 496
    },
    {
      "epoch": 0.30603448275862066,
      "grad_norm": 0.7479015833484532,
      "learning_rate": 1.7529906164558023e-05,
      "loss": 0.9116,
      "step": 497
    },
    {
      "epoch": 0.3066502463054187,
      "grad_norm": 0.738997604422002,
      "learning_rate": 1.7515739101399983e-05,
      "loss": 0.9043,
      "step": 498
    },
    {
      "epoch": 0.30726600985221675,
      "grad_norm": 0.7459637211359954,
      "learning_rate": 1.750153728695904e-05,
      "loss": 0.9305,
      "step": 499
    },
    {
      "epoch": 0.3078817733990148,
      "grad_norm": 0.7473917446351895,
      "learning_rate": 1.7487300786901568e-05,
      "loss": 0.9034,
      "step": 500
    },
    {
      "epoch": 0.30849753694581283,
      "grad_norm": 0.7356205768831505,
      "learning_rate": 1.74730296670543e-05,
      "loss": 0.9193,
      "step": 501
    },
    {
      "epoch": 0.3091133004926108,
      "grad_norm": 0.7460044938292174,
      "learning_rate": 1.7458723993404065e-05,
      "loss": 0.8836,
      "step": 502
    },
    {
      "epoch": 0.30972906403940886,
      "grad_norm": 0.747021975062783,
      "learning_rate": 1.744438383209744e-05,
      "loss": 0.9221,
      "step": 503
    },
    {
      "epoch": 0.3103448275862069,
      "grad_norm": 0.7111229022452777,
      "learning_rate": 1.74300092494405e-05,
      "loss": 0.8649,
      "step": 504
    },
    {
      "epoch": 0.31096059113300495,
      "grad_norm": 0.7717529766195572,
      "learning_rate": 1.7415600311898436e-05,
      "loss": 0.9131,
      "step": 505
    },
    {
      "epoch": 0.31157635467980294,
      "grad_norm": 0.7645033165009212,
      "learning_rate": 1.7401157086095317e-05,
      "loss": 0.8813,
      "step": 506
    },
    {
      "epoch": 0.312192118226601,
      "grad_norm": 0.749147400411528,
      "learning_rate": 1.738667963881375e-05,
      "loss": 0.8959,
      "step": 507
    },
    {
      "epoch": 0.312807881773399,
      "grad_norm": 0.6943278079600544,
      "learning_rate": 1.7372168036994566e-05,
      "loss": 0.8799,
      "step": 508
    },
    {
      "epoch": 0.31342364532019706,
      "grad_norm": 0.7200313987360141,
      "learning_rate": 1.7357622347736534e-05,
      "loss": 0.8842,
      "step": 509
    },
    {
      "epoch": 0.31403940886699505,
      "grad_norm": 0.7499095441711879,
      "learning_rate": 1.734304263829602e-05,
      "loss": 0.8797,
      "step": 510
    },
    {
      "epoch": 0.3146551724137931,
      "grad_norm": 0.762281113036068,
      "learning_rate": 1.7328428976086706e-05,
      "loss": 0.9081,
      "step": 511
    },
    {
      "epoch": 0.31527093596059114,
      "grad_norm": 0.7074231017960704,
      "learning_rate": 1.7313781428679256e-05,
      "loss": 0.8983,
      "step": 512
    },
    {
      "epoch": 0.3158866995073892,
      "grad_norm": 0.7413278955038624,
      "learning_rate": 1.729910006380102e-05,
      "loss": 0.8847,
      "step": 513
    },
    {
      "epoch": 0.31650246305418717,
      "grad_norm": 0.7626501910071514,
      "learning_rate": 1.728438494933571e-05,
      "loss": 0.9088,
      "step": 514
    },
    {
      "epoch": 0.3171182266009852,
      "grad_norm": 0.7743187244086897,
      "learning_rate": 1.726963615332308e-05,
      "loss": 0.9959,
      "step": 515
    },
    {
      "epoch": 0.31773399014778325,
      "grad_norm": 0.7742186492821854,
      "learning_rate": 1.725485374395864e-05,
      "loss": 0.9134,
      "step": 516
    },
    {
      "epoch": 0.3183497536945813,
      "grad_norm": 0.6748102318157325,
      "learning_rate": 1.7240037789593307e-05,
      "loss": 0.946,
      "step": 517
    },
    {
      "epoch": 0.31896551724137934,
      "grad_norm": 0.7748190774101086,
      "learning_rate": 1.7225188358733106e-05,
      "loss": 0.9262,
      "step": 518
    },
    {
      "epoch": 0.3195812807881773,
      "grad_norm": 0.6933035589894282,
      "learning_rate": 1.7210305520038857e-05,
      "loss": 0.961,
      "step": 519
    },
    {
      "epoch": 0.32019704433497537,
      "grad_norm": 0.7591073547227345,
      "learning_rate": 1.7195389342325843e-05,
      "loss": 0.9101,
      "step": 520
    },
    {
      "epoch": 0.3208128078817734,
      "grad_norm": 0.6940430496358198,
      "learning_rate": 1.71804398945635e-05,
      "loss": 0.8773,
      "step": 521
    },
    {
      "epoch": 0.32142857142857145,
      "grad_norm": 0.718356114152716,
      "learning_rate": 1.7165457245875107e-05,
      "loss": 0.8525,
      "step": 522
    },
    {
      "epoch": 0.32204433497536944,
      "grad_norm": 0.7907598131269848,
      "learning_rate": 1.7150441465537446e-05,
      "loss": 0.9189,
      "step": 523
    },
    {
      "epoch": 0.3226600985221675,
      "grad_norm": 0.7379759596247293,
      "learning_rate": 1.7135392622980505e-05,
      "loss": 0.9246,
      "step": 524
    },
    {
      "epoch": 0.3232758620689655,
      "grad_norm": 0.7282792993241147,
      "learning_rate": 1.7120310787787136e-05,
      "loss": 0.9219,
      "step": 525
    },
    {
      "epoch": 0.32389162561576357,
      "grad_norm": 0.7794267158816651,
      "learning_rate": 1.7105196029692743e-05,
      "loss": 0.8776,
      "step": 526
    },
    {
      "epoch": 0.32450738916256155,
      "grad_norm": 0.7234726123744878,
      "learning_rate": 1.7090048418584972e-05,
      "loss": 0.9422,
      "step": 527
    },
    {
      "epoch": 0.3251231527093596,
      "grad_norm": 0.7275204190473751,
      "learning_rate": 1.707486802450335e-05,
      "loss": 0.9431,
      "step": 528
    },
    {
      "epoch": 0.32573891625615764,
      "grad_norm": 0.7066903044219974,
      "learning_rate": 1.7059654917639008e-05,
      "loss": 0.9059,
      "step": 529
    },
    {
      "epoch": 0.3263546798029557,
      "grad_norm": 0.783812008371628,
      "learning_rate": 1.7044409168334327e-05,
      "loss": 0.9406,
      "step": 530
    },
    {
      "epoch": 0.32697044334975367,
      "grad_norm": 0.7324495421663059,
      "learning_rate": 1.7029130847082615e-05,
      "loss": 0.9235,
      "step": 531
    },
    {
      "epoch": 0.3275862068965517,
      "grad_norm": 0.728827363994343,
      "learning_rate": 1.70138200245278e-05,
      "loss": 0.9367,
      "step": 532
    },
    {
      "epoch": 0.32820197044334976,
      "grad_norm": 0.712961786760796,
      "learning_rate": 1.699847677146407e-05,
      "loss": 0.8206,
      "step": 533
    },
    {
      "epoch": 0.3288177339901478,
      "grad_norm": 0.7226672366718142,
      "learning_rate": 1.698310115883558e-05,
      "loss": 0.9269,
      "step": 534
    },
    {
      "epoch": 0.3294334975369458,
      "grad_norm": 0.7121050711154516,
      "learning_rate": 1.696769325773611e-05,
      "loss": 0.942,
      "step": 535
    },
    {
      "epoch": 0.33004926108374383,
      "grad_norm": 0.7271801428738688,
      "learning_rate": 1.6952253139408723e-05,
      "loss": 0.9796,
      "step": 536
    },
    {
      "epoch": 0.33066502463054187,
      "grad_norm": 0.724444711960148,
      "learning_rate": 1.6936780875245462e-05,
      "loss": 0.9108,
      "step": 537
    },
    {
      "epoch": 0.3312807881773399,
      "grad_norm": 0.7472370675148189,
      "learning_rate": 1.692127653678699e-05,
      "loss": 0.9077,
      "step": 538
    },
    {
      "epoch": 0.33189655172413796,
      "grad_norm": 0.7726764183134254,
      "learning_rate": 1.6905740195722296e-05,
      "loss": 0.9096,
      "step": 539
    },
    {
      "epoch": 0.33251231527093594,
      "grad_norm": 0.728277812610702,
      "learning_rate": 1.6890171923888323e-05,
      "loss": 0.9049,
      "step": 540
    },
    {
      "epoch": 0.333128078817734,
      "grad_norm": 0.7523289114776415,
      "learning_rate": 1.6874571793269665e-05,
      "loss": 0.9274,
      "step": 541
    },
    {
      "epoch": 0.33374384236453203,
      "grad_norm": 0.7160550390783534,
      "learning_rate": 1.6858939875998227e-05,
      "loss": 0.9015,
      "step": 542
    },
    {
      "epoch": 0.33435960591133007,
      "grad_norm": 0.7155404673275594,
      "learning_rate": 1.6843276244352884e-05,
      "loss": 0.898,
      "step": 543
    },
    {
      "epoch": 0.33497536945812806,
      "grad_norm": 0.7445508686988879,
      "learning_rate": 1.682758097075915e-05,
      "loss": 0.8942,
      "step": 544
    },
    {
      "epoch": 0.3355911330049261,
      "grad_norm": 0.8020818566666468,
      "learning_rate": 1.6811854127788857e-05,
      "loss": 1.0005,
      "step": 545
    },
    {
      "epoch": 0.33620689655172414,
      "grad_norm": 0.7012228962421047,
      "learning_rate": 1.679609578815979e-05,
      "loss": 0.8829,
      "step": 546
    },
    {
      "epoch": 0.3368226600985222,
      "grad_norm": 0.7010885266300754,
      "learning_rate": 1.6780306024735384e-05,
      "loss": 0.9358,
      "step": 547
    },
    {
      "epoch": 0.3374384236453202,
      "grad_norm": 0.7360081698341392,
      "learning_rate": 1.676448491052436e-05,
      "loss": 0.9054,
      "step": 548
    },
    {
      "epoch": 0.3380541871921182,
      "grad_norm": 0.7286772954760785,
      "learning_rate": 1.6748632518680407e-05,
      "loss": 0.8995,
      "step": 549
    },
    {
      "epoch": 0.33866995073891626,
      "grad_norm": 0.7015473992476906,
      "learning_rate": 1.6732748922501832e-05,
      "loss": 0.9369,
      "step": 550
    },
    {
      "epoch": 0.3392857142857143,
      "grad_norm": 0.7069702591934885,
      "learning_rate": 1.6716834195431225e-05,
      "loss": 0.8633,
      "step": 551
    },
    {
      "epoch": 0.3399014778325123,
      "grad_norm": 0.7269374577094286,
      "learning_rate": 1.6700888411055113e-05,
      "loss": 0.8997,
      "step": 552
    },
    {
      "epoch": 0.34051724137931033,
      "grad_norm": 0.7479511971499142,
      "learning_rate": 1.6684911643103643e-05,
      "loss": 0.8989,
      "step": 553
    },
    {
      "epoch": 0.3411330049261084,
      "grad_norm": 0.7210893606666264,
      "learning_rate": 1.6668903965450204e-05,
      "loss": 0.8753,
      "step": 554
    },
    {
      "epoch": 0.3417487684729064,
      "grad_norm": 0.7268174203081782,
      "learning_rate": 1.6652865452111115e-05,
      "loss": 0.9274,
      "step": 555
    },
    {
      "epoch": 0.34236453201970446,
      "grad_norm": 0.8049597192307807,
      "learning_rate": 1.663679617724528e-05,
      "loss": 0.892,
      "step": 556
    },
    {
      "epoch": 0.34298029556650245,
      "grad_norm": 0.7489226139785651,
      "learning_rate": 1.6620696215153816e-05,
      "loss": 0.947,
      "step": 557
    },
    {
      "epoch": 0.3435960591133005,
      "grad_norm": 0.758510538689753,
      "learning_rate": 1.6604565640279755e-05,
      "loss": 0.8993,
      "step": 558
    },
    {
      "epoch": 0.34421182266009853,
      "grad_norm": 0.7732653735571087,
      "learning_rate": 1.6588404527207665e-05,
      "loss": 0.9314,
      "step": 559
    },
    {
      "epoch": 0.3448275862068966,
      "grad_norm": 0.8053398930223878,
      "learning_rate": 1.657221295066332e-05,
      "loss": 0.986,
      "step": 560
    },
    {
      "epoch": 0.34544334975369456,
      "grad_norm": 0.7179093218040571,
      "learning_rate": 1.6555990985513352e-05,
      "loss": 0.8983,
      "step": 561
    },
    {
      "epoch": 0.3460591133004926,
      "grad_norm": 0.7785963959186433,
      "learning_rate": 1.6539738706764895e-05,
      "loss": 0.9348,
      "step": 562
    },
    {
      "epoch": 0.34667487684729065,
      "grad_norm": 0.7423671621854595,
      "learning_rate": 1.652345618956526e-05,
      "loss": 0.9418,
      "step": 563
    },
    {
      "epoch": 0.3472906403940887,
      "grad_norm": 0.8369399491047526,
      "learning_rate": 1.6507143509201565e-05,
      "loss": 0.9218,
      "step": 564
    },
    {
      "epoch": 0.3479064039408867,
      "grad_norm": 0.7726139817087707,
      "learning_rate": 1.6490800741100396e-05,
      "loss": 0.9247,
      "step": 565
    },
    {
      "epoch": 0.3485221674876847,
      "grad_norm": 0.6890448721464069,
      "learning_rate": 1.6474427960827473e-05,
      "loss": 0.8839,
      "step": 566
    },
    {
      "epoch": 0.34913793103448276,
      "grad_norm": 0.8046033074674707,
      "learning_rate": 1.645802524408727e-05,
      "loss": 0.9638,
      "step": 567
    },
    {
      "epoch": 0.3497536945812808,
      "grad_norm": 0.852693357185336,
      "learning_rate": 1.6441592666722686e-05,
      "loss": 0.9289,
      "step": 568
    },
    {
      "epoch": 0.3503694581280788,
      "grad_norm": 0.6936317682648182,
      "learning_rate": 1.642513030471469e-05,
      "loss": 0.9376,
      "step": 569
    },
    {
      "epoch": 0.35098522167487683,
      "grad_norm": 0.8359120347696283,
      "learning_rate": 1.6408638234181975e-05,
      "loss": 0.9258,
      "step": 570
    },
    {
      "epoch": 0.3516009852216749,
      "grad_norm": 0.7902217296638873,
      "learning_rate": 1.6392116531380592e-05,
      "loss": 0.9249,
      "step": 571
    },
    {
      "epoch": 0.3522167487684729,
      "grad_norm": 0.8429997933042143,
      "learning_rate": 1.6375565272703603e-05,
      "loss": 0.9235,
      "step": 572
    },
    {
      "epoch": 0.35283251231527096,
      "grad_norm": 0.7599739111864708,
      "learning_rate": 1.635898453468075e-05,
      "loss": 0.9341,
      "step": 573
    },
    {
      "epoch": 0.35344827586206895,
      "grad_norm": 0.8147136588774577,
      "learning_rate": 1.6342374393978056e-05,
      "loss": 0.9339,
      "step": 574
    },
    {
      "epoch": 0.354064039408867,
      "grad_norm": 0.68549077002484,
      "learning_rate": 1.6325734927397514e-05,
      "loss": 0.9114,
      "step": 575
    },
    {
      "epoch": 0.35467980295566504,
      "grad_norm": 0.7647400428163069,
      "learning_rate": 1.6309066211876706e-05,
      "loss": 0.8837,
      "step": 576
    },
    {
      "epoch": 0.3552955665024631,
      "grad_norm": 0.758906658259343,
      "learning_rate": 1.6292368324488462e-05,
      "loss": 0.9221,
      "step": 577
    },
    {
      "epoch": 0.35591133004926107,
      "grad_norm": 0.7004070808672317,
      "learning_rate": 1.6275641342440485e-05,
      "loss": 0.8985,
      "step": 578
    },
    {
      "epoch": 0.3565270935960591,
      "grad_norm": 0.8364878249721586,
      "learning_rate": 1.625888534307502e-05,
      "loss": 0.9436,
      "step": 579
    },
    {
      "epoch": 0.35714285714285715,
      "grad_norm": 0.7479863126690329,
      "learning_rate": 1.6242100403868472e-05,
      "loss": 0.9245,
      "step": 580
    },
    {
      "epoch": 0.3577586206896552,
      "grad_norm": 0.6748153387155272,
      "learning_rate": 1.6225286602431062e-05,
      "loss": 0.8839,
      "step": 581
    },
    {
      "epoch": 0.3583743842364532,
      "grad_norm": 0.6800983370411355,
      "learning_rate": 1.6208444016506466e-05,
      "loss": 0.9226,
      "step": 582
    },
    {
      "epoch": 0.3589901477832512,
      "grad_norm": 0.7318130111586356,
      "learning_rate": 1.6191572723971453e-05,
      "loss": 0.9593,
      "step": 583
    },
    {
      "epoch": 0.35960591133004927,
      "grad_norm": 0.7091040860289491,
      "learning_rate": 1.617467280283552e-05,
      "loss": 0.9159,
      "step": 584
    },
    {
      "epoch": 0.3602216748768473,
      "grad_norm": 0.6630825782873812,
      "learning_rate": 1.615774433124054e-05,
      "loss": 0.9188,
      "step": 585
    },
    {
      "epoch": 0.3608374384236453,
      "grad_norm": 0.7264937003279959,
      "learning_rate": 1.6140787387460406e-05,
      "loss": 0.8684,
      "step": 586
    },
    {
      "epoch": 0.36145320197044334,
      "grad_norm": 0.8216114100156174,
      "learning_rate": 1.612380204990065e-05,
      "loss": 0.9513,
      "step": 587
    },
    {
      "epoch": 0.3620689655172414,
      "grad_norm": 0.7747778468186207,
      "learning_rate": 1.6106788397098096e-05,
      "loss": 0.9143,
      "step": 588
    },
    {
      "epoch": 0.3626847290640394,
      "grad_norm": 0.7533543771366809,
      "learning_rate": 1.608974650772049e-05,
      "loss": 0.9104,
      "step": 589
    },
    {
      "epoch": 0.3633004926108374,
      "grad_norm": 0.6828432319324707,
      "learning_rate": 1.6072676460566136e-05,
      "loss": 0.8639,
      "step": 590
    },
    {
      "epoch": 0.36391625615763545,
      "grad_norm": 0.7652619880731669,
      "learning_rate": 1.605557833456354e-05,
      "loss": 0.9114,
      "step": 591
    },
    {
      "epoch": 0.3645320197044335,
      "grad_norm": 0.8800018906761232,
      "learning_rate": 1.6038452208771037e-05,
      "loss": 0.9514,
      "step": 592
    },
    {
      "epoch": 0.36514778325123154,
      "grad_norm": 0.7307219299855516,
      "learning_rate": 1.602129816237643e-05,
      "loss": 0.9175,
      "step": 593
    },
    {
      "epoch": 0.3657635467980296,
      "grad_norm": 0.8081051916331987,
      "learning_rate": 1.6004116274696612e-05,
      "loss": 0.9312,
      "step": 594
    },
    {
      "epoch": 0.36637931034482757,
      "grad_norm": 0.8852465096504152,
      "learning_rate": 1.5986906625177215e-05,
      "loss": 0.9165,
      "step": 595
    },
    {
      "epoch": 0.3669950738916256,
      "grad_norm": 0.7818806211103125,
      "learning_rate": 1.596966929339224e-05,
      "loss": 0.8795,
      "step": 596
    },
    {
      "epoch": 0.36761083743842365,
      "grad_norm": 0.7340620432050833,
      "learning_rate": 1.5952404359043677e-05,
      "loss": 0.935,
      "step": 597
    },
    {
      "epoch": 0.3682266009852217,
      "grad_norm": 0.7148992930021167,
      "learning_rate": 1.593511190196115e-05,
      "loss": 0.9124,
      "step": 598
    },
    {
      "epoch": 0.3688423645320197,
      "grad_norm": 0.7537062477024269,
      "learning_rate": 1.5917792002101543e-05,
      "loss": 0.9255,
      "step": 599
    },
    {
      "epoch": 0.3694581280788177,
      "grad_norm": 0.866813704134017,
      "learning_rate": 1.590044473954863e-05,
      "loss": 0.9654,
      "step": 600
    },
    {
      "epoch": 0.3694581280788177,
      "eval_loss": 0.9203624129295349,
      "eval_runtime": 523.2904,
      "eval_samples_per_second": 44.163,
      "eval_steps_per_second": 0.692,
      "step": 600
    },
    {
      "epoch": 0.37007389162561577,
      "grad_norm": 0.7666079621815708,
      "learning_rate": 1.5883070194512695e-05,
      "loss": 0.9504,
      "step": 601
    },
    {
      "epoch": 0.3706896551724138,
      "grad_norm": 0.768162955775916,
      "learning_rate": 1.586566844733019e-05,
      "loss": 0.9456,
      "step": 602
    },
    {
      "epoch": 0.3713054187192118,
      "grad_norm": 0.7542787312915412,
      "learning_rate": 1.5848239578463325e-05,
      "loss": 0.8882,
      "step": 603
    },
    {
      "epoch": 0.37192118226600984,
      "grad_norm": 0.7366063864182639,
      "learning_rate": 1.5830783668499728e-05,
      "loss": 0.9211,
      "step": 604
    },
    {
      "epoch": 0.3725369458128079,
      "grad_norm": 0.7965809717367995,
      "learning_rate": 1.5813300798152048e-05,
      "loss": 0.941,
      "step": 605
    },
    {
      "epoch": 0.3731527093596059,
      "grad_norm": 0.7498674103705482,
      "learning_rate": 1.579579104825761e-05,
      "loss": 0.946,
      "step": 606
    },
    {
      "epoch": 0.3737684729064039,
      "grad_norm": 0.7038586405178672,
      "learning_rate": 1.5778254499778006e-05,
      "loss": 0.8616,
      "step": 607
    },
    {
      "epoch": 0.37438423645320196,
      "grad_norm": 0.8097405489416537,
      "learning_rate": 1.5760691233798758e-05,
      "loss": 0.9579,
      "step": 608
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.8063846843137037,
      "learning_rate": 1.574310133152891e-05,
      "loss": 0.9551,
      "step": 609
    },
    {
      "epoch": 0.37561576354679804,
      "grad_norm": 0.724083436733293,
      "learning_rate": 1.5725484874300673e-05,
      "loss": 0.8762,
      "step": 610
    },
    {
      "epoch": 0.3762315270935961,
      "grad_norm": 0.7432110761254708,
      "learning_rate": 1.570784194356904e-05,
      "loss": 0.9508,
      "step": 611
    },
    {
      "epoch": 0.3768472906403941,
      "grad_norm": 0.8101556483372908,
      "learning_rate": 1.569017262091141e-05,
      "loss": 0.9226,
      "step": 612
    },
    {
      "epoch": 0.3774630541871921,
      "grad_norm": 0.8081611053152782,
      "learning_rate": 1.567247698802723e-05,
      "loss": 0.9345,
      "step": 613
    },
    {
      "epoch": 0.37807881773399016,
      "grad_norm": 0.7829024361706384,
      "learning_rate": 1.5654755126737574e-05,
      "loss": 0.9145,
      "step": 614
    },
    {
      "epoch": 0.3786945812807882,
      "grad_norm": 0.7791445046324124,
      "learning_rate": 1.5637007118984814e-05,
      "loss": 0.9158,
      "step": 615
    },
    {
      "epoch": 0.3793103448275862,
      "grad_norm": 0.784606422832932,
      "learning_rate": 1.56192330468322e-05,
      "loss": 0.9134,
      "step": 616
    },
    {
      "epoch": 0.37992610837438423,
      "grad_norm": 0.7219260756014515,
      "learning_rate": 1.560143299246351e-05,
      "loss": 0.9373,
      "step": 617
    },
    {
      "epoch": 0.3805418719211823,
      "grad_norm": 0.826664374796452,
      "learning_rate": 1.5583607038182656e-05,
      "loss": 0.9046,
      "step": 618
    },
    {
      "epoch": 0.3811576354679803,
      "grad_norm": 0.6819272677265947,
      "learning_rate": 1.55657552664133e-05,
      "loss": 0.8825,
      "step": 619
    },
    {
      "epoch": 0.3817733990147783,
      "grad_norm": 0.6669468387202399,
      "learning_rate": 1.5547877759698498e-05,
      "loss": 0.9311,
      "step": 620
    },
    {
      "epoch": 0.38238916256157635,
      "grad_norm": 0.7095027316903946,
      "learning_rate": 1.5529974600700277e-05,
      "loss": 0.8446,
      "step": 621
    },
    {
      "epoch": 0.3830049261083744,
      "grad_norm": 0.8165641898908078,
      "learning_rate": 1.551204587219928e-05,
      "loss": 0.9478,
      "step": 622
    },
    {
      "epoch": 0.38362068965517243,
      "grad_norm": 0.7490226706084134,
      "learning_rate": 1.5494091657094384e-05,
      "loss": 0.8809,
      "step": 623
    },
    {
      "epoch": 0.3842364532019704,
      "grad_norm": 0.7055231669213677,
      "learning_rate": 1.5476112038402314e-05,
      "loss": 0.9154,
      "step": 624
    },
    {
      "epoch": 0.38485221674876846,
      "grad_norm": 0.7154633220704486,
      "learning_rate": 1.5458107099257245e-05,
      "loss": 0.8891,
      "step": 625
    },
    {
      "epoch": 0.3854679802955665,
      "grad_norm": 0.7394384180399522,
      "learning_rate": 1.544007692291044e-05,
      "loss": 0.9237,
      "step": 626
    },
    {
      "epoch": 0.38608374384236455,
      "grad_norm": 0.7158409982073084,
      "learning_rate": 1.5422021592729845e-05,
      "loss": 0.9111,
      "step": 627
    },
    {
      "epoch": 0.3866995073891626,
      "grad_norm": 0.8389391973488234,
      "learning_rate": 1.540394119219972e-05,
      "loss": 0.9974,
      "step": 628
    },
    {
      "epoch": 0.3873152709359606,
      "grad_norm": 0.7083564669234742,
      "learning_rate": 1.5385835804920234e-05,
      "loss": 0.8757,
      "step": 629
    },
    {
      "epoch": 0.3879310344827586,
      "grad_norm": 0.7467115398187937,
      "learning_rate": 1.5367705514607107e-05,
      "loss": 0.8777,
      "step": 630
    },
    {
      "epoch": 0.38854679802955666,
      "grad_norm": 0.6895572413201174,
      "learning_rate": 1.534955040509119e-05,
      "loss": 0.8855,
      "step": 631
    },
    {
      "epoch": 0.3891625615763547,
      "grad_norm": 0.7298596258341112,
      "learning_rate": 1.5331370560318097e-05,
      "loss": 0.9275,
      "step": 632
    },
    {
      "epoch": 0.3897783251231527,
      "grad_norm": 0.7198657599128543,
      "learning_rate": 1.5313166064347814e-05,
      "loss": 0.9067,
      "step": 633
    },
    {
      "epoch": 0.39039408866995073,
      "grad_norm": 0.7127688696941393,
      "learning_rate": 1.5294937001354314e-05,
      "loss": 0.8741,
      "step": 634
    },
    {
      "epoch": 0.3910098522167488,
      "grad_norm": 0.6947564547366606,
      "learning_rate": 1.527668345562516e-05,
      "loss": 0.9062,
      "step": 635
    },
    {
      "epoch": 0.3916256157635468,
      "grad_norm": 0.8089478199124545,
      "learning_rate": 1.5258405511561117e-05,
      "loss": 0.9312,
      "step": 636
    },
    {
      "epoch": 0.3922413793103448,
      "grad_norm": 0.7045152856838085,
      "learning_rate": 1.5240103253675756e-05,
      "loss": 0.8762,
      "step": 637
    },
    {
      "epoch": 0.39285714285714285,
      "grad_norm": 0.7467057065510266,
      "learning_rate": 1.5221776766595082e-05,
      "loss": 0.8842,
      "step": 638
    },
    {
      "epoch": 0.3934729064039409,
      "grad_norm": 0.7026527662371164,
      "learning_rate": 1.5203426135057134e-05,
      "loss": 0.9171,
      "step": 639
    },
    {
      "epoch": 0.39408866995073893,
      "grad_norm": 0.7837461065473632,
      "learning_rate": 1.518505144391157e-05,
      "loss": 0.8989,
      "step": 640
    },
    {
      "epoch": 0.3947044334975369,
      "grad_norm": 0.8074879591147561,
      "learning_rate": 1.516665277811932e-05,
      "loss": 0.9284,
      "step": 641
    },
    {
      "epoch": 0.39532019704433496,
      "grad_norm": 0.8040848744255084,
      "learning_rate": 1.5148230222752146e-05,
      "loss": 0.9198,
      "step": 642
    },
    {
      "epoch": 0.395935960591133,
      "grad_norm": 0.8000900335268015,
      "learning_rate": 1.5129783862992282e-05,
      "loss": 0.9884,
      "step": 643
    },
    {
      "epoch": 0.39655172413793105,
      "grad_norm": 0.707826407308019,
      "learning_rate": 1.511131378413203e-05,
      "loss": 0.904,
      "step": 644
    },
    {
      "epoch": 0.39716748768472904,
      "grad_norm": 0.6831968272065668,
      "learning_rate": 1.5092820071573358e-05,
      "loss": 0.9306,
      "step": 645
    },
    {
      "epoch": 0.3977832512315271,
      "grad_norm": 0.7526525298398172,
      "learning_rate": 1.5074302810827516e-05,
      "loss": 0.8944,
      "step": 646
    },
    {
      "epoch": 0.3983990147783251,
      "grad_norm": 0.7259511252677896,
      "learning_rate": 1.5055762087514641e-05,
      "loss": 0.8717,
      "step": 647
    },
    {
      "epoch": 0.39901477832512317,
      "grad_norm": 0.7262593221006888,
      "learning_rate": 1.503719798736334e-05,
      "loss": 0.9329,
      "step": 648
    },
    {
      "epoch": 0.3996305418719212,
      "grad_norm": 0.761381185399442,
      "learning_rate": 1.5018610596210328e-05,
      "loss": 0.8441,
      "step": 649
    },
    {
      "epoch": 0.4002463054187192,
      "grad_norm": 0.7380672763641735,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.8626,
      "step": 650
    },
    {
      "epoch": 0.40086206896551724,
      "grad_norm": 0.7162617676427693,
      "learning_rate": 1.4981366284784058e-05,
      "loss": 0.9094,
      "step": 651
    },
    {
      "epoch": 0.4014778325123153,
      "grad_norm": 0.7001630092423472,
      "learning_rate": 1.4962709536721087e-05,
      "loss": 0.9038,
      "step": 652
    },
    {
      "epoch": 0.4020935960591133,
      "grad_norm": 0.7262642057511702,
      "learning_rate": 1.4944029842076184e-05,
      "loss": 0.9452,
      "step": 653
    },
    {
      "epoch": 0.4027093596059113,
      "grad_norm": 0.7731201626263766,
      "learning_rate": 1.4925327287220546e-05,
      "loss": 0.9049,
      "step": 654
    },
    {
      "epoch": 0.40332512315270935,
      "grad_norm": 0.7466778029205926,
      "learning_rate": 1.490660195863106e-05,
      "loss": 0.8923,
      "step": 655
    },
    {
      "epoch": 0.4039408866995074,
      "grad_norm": 0.7647145959939173,
      "learning_rate": 1.4887853942889928e-05,
      "loss": 0.9204,
      "step": 656
    },
    {
      "epoch": 0.40455665024630544,
      "grad_norm": 0.775571015601727,
      "learning_rate": 1.486908332668424e-05,
      "loss": 0.9405,
      "step": 657
    },
    {
      "epoch": 0.4051724137931034,
      "grad_norm": 0.8581067779979905,
      "learning_rate": 1.4850290196805595e-05,
      "loss": 0.9296,
      "step": 658
    },
    {
      "epoch": 0.40578817733990147,
      "grad_norm": 0.879168250647678,
      "learning_rate": 1.4831474640149686e-05,
      "loss": 0.8995,
      "step": 659
    },
    {
      "epoch": 0.4064039408866995,
      "grad_norm": 0.7379906668609233,
      "learning_rate": 1.4812636743715912e-05,
      "loss": 0.9031,
      "step": 660
    },
    {
      "epoch": 0.40701970443349755,
      "grad_norm": 0.7169990539973173,
      "learning_rate": 1.4793776594606953e-05,
      "loss": 0.9057,
      "step": 661
    },
    {
      "epoch": 0.40763546798029554,
      "grad_norm": 0.735574014453652,
      "learning_rate": 1.477489428002839e-05,
      "loss": 0.9258,
      "step": 662
    },
    {
      "epoch": 0.4082512315270936,
      "grad_norm": 0.7673748728077313,
      "learning_rate": 1.4755989887288287e-05,
      "loss": 0.9632,
      "step": 663
    },
    {
      "epoch": 0.4088669950738916,
      "grad_norm": 0.6967848282294777,
      "learning_rate": 1.4737063503796803e-05,
      "loss": 0.9178,
      "step": 664
    },
    {
      "epoch": 0.40948275862068967,
      "grad_norm": 0.759648736431933,
      "learning_rate": 1.4718115217065766e-05,
      "loss": 0.9208,
      "step": 665
    },
    {
      "epoch": 0.4100985221674877,
      "grad_norm": 0.7028106750849102,
      "learning_rate": 1.4699145114708286e-05,
      "loss": 0.9079,
      "step": 666
    },
    {
      "epoch": 0.4107142857142857,
      "grad_norm": 0.7286631476623967,
      "learning_rate": 1.4680153284438345e-05,
      "loss": 0.9374,
      "step": 667
    },
    {
      "epoch": 0.41133004926108374,
      "grad_norm": 0.7173261895659683,
      "learning_rate": 1.4661139814070388e-05,
      "loss": 0.904,
      "step": 668
    },
    {
      "epoch": 0.4119458128078818,
      "grad_norm": 0.8041189035448213,
      "learning_rate": 1.4642104791518925e-05,
      "loss": 0.977,
      "step": 669
    },
    {
      "epoch": 0.4125615763546798,
      "grad_norm": 0.8565943031002121,
      "learning_rate": 1.462304830479811e-05,
      "loss": 0.9435,
      "step": 670
    },
    {
      "epoch": 0.4131773399014778,
      "grad_norm": 0.8068780087093754,
      "learning_rate": 1.4603970442021347e-05,
      "loss": 0.9159,
      "step": 671
    },
    {
      "epoch": 0.41379310344827586,
      "grad_norm": 0.6878664508992455,
      "learning_rate": 1.458487129140088e-05,
      "loss": 0.8934,
      "step": 672
    },
    {
      "epoch": 0.4144088669950739,
      "grad_norm": 0.8854710422138209,
      "learning_rate": 1.4565750941247387e-05,
      "loss": 0.9294,
      "step": 673
    },
    {
      "epoch": 0.41502463054187194,
      "grad_norm": 0.7612026061508668,
      "learning_rate": 1.454660947996956e-05,
      "loss": 0.8515,
      "step": 674
    },
    {
      "epoch": 0.41564039408866993,
      "grad_norm": 0.7623965909921044,
      "learning_rate": 1.4527446996073714e-05,
      "loss": 0.8734,
      "step": 675
    },
    {
      "epoch": 0.41625615763546797,
      "grad_norm": 0.8924056763465215,
      "learning_rate": 1.4508263578163358e-05,
      "loss": 0.9217,
      "step": 676
    },
    {
      "epoch": 0.416871921182266,
      "grad_norm": 0.7642503352656572,
      "learning_rate": 1.4489059314938803e-05,
      "loss": 0.9419,
      "step": 677
    },
    {
      "epoch": 0.41748768472906406,
      "grad_norm": 0.7238412323618686,
      "learning_rate": 1.4469834295196743e-05,
      "loss": 0.9307,
      "step": 678
    },
    {
      "epoch": 0.41810344827586204,
      "grad_norm": 0.6886381697098398,
      "learning_rate": 1.4450588607829848e-05,
      "loss": 0.8495,
      "step": 679
    },
    {
      "epoch": 0.4187192118226601,
      "grad_norm": 0.7579337372764616,
      "learning_rate": 1.4431322341826348e-05,
      "loss": 0.9298,
      "step": 680
    },
    {
      "epoch": 0.41933497536945813,
      "grad_norm": 0.8184298477126075,
      "learning_rate": 1.4412035586269624e-05,
      "loss": 0.9229,
      "step": 681
    },
    {
      "epoch": 0.41995073891625617,
      "grad_norm": 0.7410630804312669,
      "learning_rate": 1.4392728430337801e-05,
      "loss": 0.9196,
      "step": 682
    },
    {
      "epoch": 0.4205665024630542,
      "grad_norm": 0.7862185272893113,
      "learning_rate": 1.437340096330332e-05,
      "loss": 0.9202,
      "step": 683
    },
    {
      "epoch": 0.4211822660098522,
      "grad_norm": 0.7702857384435966,
      "learning_rate": 1.4354053274532558e-05,
      "loss": 0.9126,
      "step": 684
    },
    {
      "epoch": 0.42179802955665024,
      "grad_norm": 0.7243725862161867,
      "learning_rate": 1.433468545348537e-05,
      "loss": 0.868,
      "step": 685
    },
    {
      "epoch": 0.4224137931034483,
      "grad_norm": 0.7138774034175982,
      "learning_rate": 1.4315297589714709e-05,
      "loss": 0.9238,
      "step": 686
    },
    {
      "epoch": 0.42302955665024633,
      "grad_norm": 0.8281148499170136,
      "learning_rate": 1.429588977286621e-05,
      "loss": 0.9972,
      "step": 687
    },
    {
      "epoch": 0.4236453201970443,
      "grad_norm": 0.9075920013527521,
      "learning_rate": 1.427646209267775e-05,
      "loss": 0.9195,
      "step": 688
    },
    {
      "epoch": 0.42426108374384236,
      "grad_norm": 0.7281268969343063,
      "learning_rate": 1.4257014638979056e-05,
      "loss": 0.8992,
      "step": 689
    },
    {
      "epoch": 0.4248768472906404,
      "grad_norm": 0.6762899043003119,
      "learning_rate": 1.4237547501691298e-05,
      "loss": 0.8421,
      "step": 690
    },
    {
      "epoch": 0.42549261083743845,
      "grad_norm": 0.7670526040062573,
      "learning_rate": 1.4218060770826636e-05,
      "loss": 0.8563,
      "step": 691
    },
    {
      "epoch": 0.42610837438423643,
      "grad_norm": 0.7333402772092343,
      "learning_rate": 1.4198554536487844e-05,
      "loss": 0.9477,
      "step": 692
    },
    {
      "epoch": 0.4267241379310345,
      "grad_norm": 0.7197172160635574,
      "learning_rate": 1.4179028888867868e-05,
      "loss": 0.9262,
      "step": 693
    },
    {
      "epoch": 0.4273399014778325,
      "grad_norm": 0.751828215778602,
      "learning_rate": 1.415948391824942e-05,
      "loss": 0.9373,
      "step": 694
    },
    {
      "epoch": 0.42795566502463056,
      "grad_norm": 0.6560075288777694,
      "learning_rate": 1.4139919715004558e-05,
      "loss": 0.9538,
      "step": 695
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.7026101697659416,
      "learning_rate": 1.4120336369594266e-05,
      "loss": 0.937,
      "step": 696
    },
    {
      "epoch": 0.4291871921182266,
      "grad_norm": 0.6450108202472301,
      "learning_rate": 1.4100733972568038e-05,
      "loss": 0.8941,
      "step": 697
    },
    {
      "epoch": 0.42980295566502463,
      "grad_norm": 0.7116290523833795,
      "learning_rate": 1.408111261456346e-05,
      "loss": 0.9012,
      "step": 698
    },
    {
      "epoch": 0.4304187192118227,
      "grad_norm": 0.7934496833808204,
      "learning_rate": 1.406147238630579e-05,
      "loss": 0.9038,
      "step": 699
    },
    {
      "epoch": 0.43103448275862066,
      "grad_norm": 0.7654978692961596,
      "learning_rate": 1.4041813378607534e-05,
      "loss": 0.8678,
      "step": 700
    },
    {
      "epoch": 0.4316502463054187,
      "grad_norm": 0.7022805868116359,
      "learning_rate": 1.4022135682368041e-05,
      "loss": 0.9431,
      "step": 701
    },
    {
      "epoch": 0.43226600985221675,
      "grad_norm": 0.795529174405464,
      "learning_rate": 1.400243938857306e-05,
      "loss": 0.9279,
      "step": 702
    },
    {
      "epoch": 0.4328817733990148,
      "grad_norm": 0.7165028003016511,
      "learning_rate": 1.3982724588294334e-05,
      "loss": 0.9409,
      "step": 703
    },
    {
      "epoch": 0.43349753694581283,
      "grad_norm": 0.7451919760046569,
      "learning_rate": 1.3962991372689181e-05,
      "loss": 0.9551,
      "step": 704
    },
    {
      "epoch": 0.4341133004926108,
      "grad_norm": 0.7674502321860064,
      "learning_rate": 1.3943239833000068e-05,
      "loss": 0.9008,
      "step": 705
    },
    {
      "epoch": 0.43472906403940886,
      "grad_norm": 0.7037219915764653,
      "learning_rate": 1.3923470060554185e-05,
      "loss": 0.9186,
      "step": 706
    },
    {
      "epoch": 0.4353448275862069,
      "grad_norm": 0.7573658811995292,
      "learning_rate": 1.3903682146763027e-05,
      "loss": 0.9629,
      "step": 707
    },
    {
      "epoch": 0.43596059113300495,
      "grad_norm": 0.7414293929086352,
      "learning_rate": 1.3883876183121974e-05,
      "loss": 0.8787,
      "step": 708
    },
    {
      "epoch": 0.43657635467980294,
      "grad_norm": 0.7339579410596888,
      "learning_rate": 1.3864052261209859e-05,
      "loss": 0.9121,
      "step": 709
    },
    {
      "epoch": 0.437192118226601,
      "grad_norm": 0.7876891230483778,
      "learning_rate": 1.3844210472688557e-05,
      "loss": 0.9093,
      "step": 710
    },
    {
      "epoch": 0.437807881773399,
      "grad_norm": 0.7434208786771694,
      "learning_rate": 1.3824350909302553e-05,
      "loss": 0.9282,
      "step": 711
    },
    {
      "epoch": 0.43842364532019706,
      "grad_norm": 0.7404901505784446,
      "learning_rate": 1.3804473662878519e-05,
      "loss": 0.9147,
      "step": 712
    },
    {
      "epoch": 0.43903940886699505,
      "grad_norm": 0.8042763661824709,
      "learning_rate": 1.3784578825324885e-05,
      "loss": 0.8942,
      "step": 713
    },
    {
      "epoch": 0.4396551724137931,
      "grad_norm": 0.7469166635025594,
      "learning_rate": 1.3764666488631431e-05,
      "loss": 0.9101,
      "step": 714
    },
    {
      "epoch": 0.44027093596059114,
      "grad_norm": 0.7217997379390045,
      "learning_rate": 1.3744736744868832e-05,
      "loss": 0.9133,
      "step": 715
    },
    {
      "epoch": 0.4408866995073892,
      "grad_norm": 0.7114878770374122,
      "learning_rate": 1.3724789686188266e-05,
      "loss": 0.9139,
      "step": 716
    },
    {
      "epoch": 0.44150246305418717,
      "grad_norm": 0.6785655094635878,
      "learning_rate": 1.3704825404820965e-05,
      "loss": 0.9236,
      "step": 717
    },
    {
      "epoch": 0.4421182266009852,
      "grad_norm": 0.7913942314503019,
      "learning_rate": 1.368484399307779e-05,
      "loss": 0.9677,
      "step": 718
    },
    {
      "epoch": 0.44273399014778325,
      "grad_norm": 0.7739611512073685,
      "learning_rate": 1.3664845543348823e-05,
      "loss": 0.8712,
      "step": 719
    },
    {
      "epoch": 0.4433497536945813,
      "grad_norm": 0.7195732459919049,
      "learning_rate": 1.3644830148102915e-05,
      "loss": 0.9226,
      "step": 720
    },
    {
      "epoch": 0.44396551724137934,
      "grad_norm": 0.7549396013565522,
      "learning_rate": 1.3624797899887272e-05,
      "loss": 0.9199,
      "step": 721
    },
    {
      "epoch": 0.4445812807881773,
      "grad_norm": 0.7547598843253903,
      "learning_rate": 1.360474889132703e-05,
      "loss": 0.9181,
      "step": 722
    },
    {
      "epoch": 0.44519704433497537,
      "grad_norm": 0.7213199375607754,
      "learning_rate": 1.358468321512481e-05,
      "loss": 0.8874,
      "step": 723
    },
    {
      "epoch": 0.4458128078817734,
      "grad_norm": 0.7346964706245546,
      "learning_rate": 1.356460096406031e-05,
      "loss": 0.9102,
      "step": 724
    },
    {
      "epoch": 0.44642857142857145,
      "grad_norm": 0.7901259633449156,
      "learning_rate": 1.3544502230989868e-05,
      "loss": 0.9343,
      "step": 725
    },
    {
      "epoch": 0.44704433497536944,
      "grad_norm": 0.7881231829269384,
      "learning_rate": 1.3524387108846032e-05,
      "loss": 0.8918,
      "step": 726
    },
    {
      "epoch": 0.4476600985221675,
      "grad_norm": 0.7864951596398998,
      "learning_rate": 1.3504255690637122e-05,
      "loss": 0.9304,
      "step": 727
    },
    {
      "epoch": 0.4482758620689655,
      "grad_norm": 0.7178339745294761,
      "learning_rate": 1.348410806944681e-05,
      "loss": 0.9121,
      "step": 728
    },
    {
      "epoch": 0.44889162561576357,
      "grad_norm": 0.7895000415658101,
      "learning_rate": 1.3463944338433692e-05,
      "loss": 0.8958,
      "step": 729
    },
    {
      "epoch": 0.44950738916256155,
      "grad_norm": 0.7087760457346979,
      "learning_rate": 1.344376459083085e-05,
      "loss": 0.9149,
      "step": 730
    },
    {
      "epoch": 0.4501231527093596,
      "grad_norm": 0.7741596941094813,
      "learning_rate": 1.342356891994542e-05,
      "loss": 0.9304,
      "step": 731
    },
    {
      "epoch": 0.45073891625615764,
      "grad_norm": 0.6702404195640927,
      "learning_rate": 1.3403357419158176e-05,
      "loss": 0.9481,
      "step": 732
    },
    {
      "epoch": 0.4513546798029557,
      "grad_norm": 0.6967525931961402,
      "learning_rate": 1.3383130181923073e-05,
      "loss": 0.902,
      "step": 733
    },
    {
      "epoch": 0.45197044334975367,
      "grad_norm": 0.7082409850587649,
      "learning_rate": 1.3362887301766834e-05,
      "loss": 0.899,
      "step": 734
    },
    {
      "epoch": 0.4525862068965517,
      "grad_norm": 0.7071614460916286,
      "learning_rate": 1.3342628872288505e-05,
      "loss": 0.9375,
      "step": 735
    },
    {
      "epoch": 0.45320197044334976,
      "grad_norm": 0.7502379126842741,
      "learning_rate": 1.3322354987159045e-05,
      "loss": 0.914,
      "step": 736
    },
    {
      "epoch": 0.4538177339901478,
      "grad_norm": 0.7111155880289425,
      "learning_rate": 1.3302065740120864e-05,
      "loss": 0.8948,
      "step": 737
    },
    {
      "epoch": 0.4544334975369458,
      "grad_norm": 0.7276748125885187,
      "learning_rate": 1.32817612249874e-05,
      "loss": 0.8688,
      "step": 738
    },
    {
      "epoch": 0.45504926108374383,
      "grad_norm": 0.7234583362482946,
      "learning_rate": 1.3261441535642696e-05,
      "loss": 0.927,
      "step": 739
    },
    {
      "epoch": 0.45566502463054187,
      "grad_norm": 0.6296219306039947,
      "learning_rate": 1.3241106766040957e-05,
      "loss": 0.8864,
      "step": 740
    },
    {
      "epoch": 0.4562807881773399,
      "grad_norm": 0.7332182321360358,
      "learning_rate": 1.3220757010206111e-05,
      "loss": 0.8997,
      "step": 741
    },
    {
      "epoch": 0.45689655172413796,
      "grad_norm": 0.7536049483836669,
      "learning_rate": 1.3200392362231385e-05,
      "loss": 0.9251,
      "step": 742
    },
    {
      "epoch": 0.45751231527093594,
      "grad_norm": 0.6855242096870591,
      "learning_rate": 1.3180012916278855e-05,
      "loss": 0.8972,
      "step": 743
    },
    {
      "epoch": 0.458128078817734,
      "grad_norm": 0.7001599160875688,
      "learning_rate": 1.3159618766579032e-05,
      "loss": 0.8955,
      "step": 744
    },
    {
      "epoch": 0.45874384236453203,
      "grad_norm": 0.7116392543374059,
      "learning_rate": 1.3139210007430404e-05,
      "loss": 0.9292,
      "step": 745
    },
    {
      "epoch": 0.45935960591133007,
      "grad_norm": 0.7166564981692325,
      "learning_rate": 1.3118786733199014e-05,
      "loss": 0.8929,
      "step": 746
    },
    {
      "epoch": 0.45997536945812806,
      "grad_norm": 0.6832089580595156,
      "learning_rate": 1.3098349038318026e-05,
      "loss": 0.903,
      "step": 747
    },
    {
      "epoch": 0.4605911330049261,
      "grad_norm": 0.7197387896602598,
      "learning_rate": 1.3077897017287274e-05,
      "loss": 0.9159,
      "step": 748
    },
    {
      "epoch": 0.46120689655172414,
      "grad_norm": 0.692155131115442,
      "learning_rate": 1.3057430764672834e-05,
      "loss": 0.9741,
      "step": 749
    },
    {
      "epoch": 0.4618226600985222,
      "grad_norm": 0.6854938343175595,
      "learning_rate": 1.3036950375106588e-05,
      "loss": 0.8568,
      "step": 750
    },
    {
      "epoch": 0.4624384236453202,
      "grad_norm": 0.7274248126040425,
      "learning_rate": 1.3016455943285787e-05,
      "loss": 0.8799,
      "step": 751
    },
    {
      "epoch": 0.4630541871921182,
      "grad_norm": 0.689157903141474,
      "learning_rate": 1.2995947563972604e-05,
      "loss": 0.8193,
      "step": 752
    },
    {
      "epoch": 0.46366995073891626,
      "grad_norm": 0.6983691875726622,
      "learning_rate": 1.297542533199371e-05,
      "loss": 0.8998,
      "step": 753
    },
    {
      "epoch": 0.4642857142857143,
      "grad_norm": 0.6847991398765221,
      "learning_rate": 1.2954889342239825e-05,
      "loss": 0.8771,
      "step": 754
    },
    {
      "epoch": 0.4649014778325123,
      "grad_norm": 0.8585339038785846,
      "learning_rate": 1.2934339689665274e-05,
      "loss": 0.9268,
      "step": 755
    },
    {
      "epoch": 0.46551724137931033,
      "grad_norm": 0.8505282121065763,
      "learning_rate": 1.2913776469287572e-05,
      "loss": 0.8925,
      "step": 756
    },
    {
      "epoch": 0.4661330049261084,
      "grad_norm": 0.7470693127913075,
      "learning_rate": 1.2893199776186957e-05,
      "loss": 0.9417,
      "step": 757
    },
    {
      "epoch": 0.4667487684729064,
      "grad_norm": 0.7822483289372357,
      "learning_rate": 1.2872609705505965e-05,
      "loss": 0.9429,
      "step": 758
    },
    {
      "epoch": 0.46736453201970446,
      "grad_norm": 0.8003604230792677,
      "learning_rate": 1.2852006352448991e-05,
      "loss": 0.8945,
      "step": 759
    },
    {
      "epoch": 0.46798029556650245,
      "grad_norm": 0.7879033714957672,
      "learning_rate": 1.283138981228184e-05,
      "loss": 0.9002,
      "step": 760
    },
    {
      "epoch": 0.4685960591133005,
      "grad_norm": 0.7637683111524548,
      "learning_rate": 1.281076018033129e-05,
      "loss": 0.892,
      "step": 761
    },
    {
      "epoch": 0.46921182266009853,
      "grad_norm": 0.9168263354462931,
      "learning_rate": 1.2790117551984664e-05,
      "loss": 0.9479,
      "step": 762
    },
    {
      "epoch": 0.4698275862068966,
      "grad_norm": 0.8006785643129495,
      "learning_rate": 1.2769462022689365e-05,
      "loss": 0.9406,
      "step": 763
    },
    {
      "epoch": 0.47044334975369456,
      "grad_norm": 0.7918262708716106,
      "learning_rate": 1.2748793687952455e-05,
      "loss": 0.8677,
      "step": 764
    },
    {
      "epoch": 0.4710591133004926,
      "grad_norm": 0.7704944713404307,
      "learning_rate": 1.2728112643340201e-05,
      "loss": 0.9085,
      "step": 765
    },
    {
      "epoch": 0.47167487684729065,
      "grad_norm": 0.6976627036514705,
      "learning_rate": 1.2707418984477637e-05,
      "loss": 0.9288,
      "step": 766
    },
    {
      "epoch": 0.4722906403940887,
      "grad_norm": 0.8343227464775321,
      "learning_rate": 1.2686712807048136e-05,
      "loss": 0.8582,
      "step": 767
    },
    {
      "epoch": 0.4729064039408867,
      "grad_norm": 0.6854883432742245,
      "learning_rate": 1.2665994206792937e-05,
      "loss": 0.8825,
      "step": 768
    },
    {
      "epoch": 0.4735221674876847,
      "grad_norm": 0.7456532369552783,
      "learning_rate": 1.2645263279510731e-05,
      "loss": 0.9299,
      "step": 769
    },
    {
      "epoch": 0.47413793103448276,
      "grad_norm": 0.8306336545606529,
      "learning_rate": 1.26245201210572e-05,
      "loss": 0.9638,
      "step": 770
    },
    {
      "epoch": 0.4747536945812808,
      "grad_norm": 0.7030381691849643,
      "learning_rate": 1.260376482734458e-05,
      "loss": 0.8786,
      "step": 771
    },
    {
      "epoch": 0.4753694581280788,
      "grad_norm": 0.8296074716392883,
      "learning_rate": 1.258299749434123e-05,
      "loss": 0.9136,
      "step": 772
    },
    {
      "epoch": 0.47598522167487683,
      "grad_norm": 0.7943998579944325,
      "learning_rate": 1.2562218218071162e-05,
      "loss": 0.8861,
      "step": 773
    },
    {
      "epoch": 0.4766009852216749,
      "grad_norm": 0.6659287111256073,
      "learning_rate": 1.2541427094613621e-05,
      "loss": 0.9332,
      "step": 774
    },
    {
      "epoch": 0.4772167487684729,
      "grad_norm": 0.7828823706483055,
      "learning_rate": 1.2520624220102623e-05,
      "loss": 0.9682,
      "step": 775
    },
    {
      "epoch": 0.47783251231527096,
      "grad_norm": 0.751724259581513,
      "learning_rate": 1.2499809690726523e-05,
      "loss": 0.8953,
      "step": 776
    },
    {
      "epoch": 0.47844827586206895,
      "grad_norm": 0.7709296438139199,
      "learning_rate": 1.2478983602727568e-05,
      "loss": 0.9217,
      "step": 777
    },
    {
      "epoch": 0.479064039408867,
      "grad_norm": 0.785247496487343,
      "learning_rate": 1.2458146052401444e-05,
      "loss": 0.9198,
      "step": 778
    },
    {
      "epoch": 0.47967980295566504,
      "grad_norm": 0.749054491735161,
      "learning_rate": 1.243729713609684e-05,
      "loss": 0.8791,
      "step": 779
    },
    {
      "epoch": 0.4802955665024631,
      "grad_norm": 0.7108264955630383,
      "learning_rate": 1.2416436950215001e-05,
      "loss": 0.8937,
      "step": 780
    },
    {
      "epoch": 0.48091133004926107,
      "grad_norm": 0.7475126916720256,
      "learning_rate": 1.2395565591209273e-05,
      "loss": 0.8799,
      "step": 781
    },
    {
      "epoch": 0.4815270935960591,
      "grad_norm": 0.8052386767876591,
      "learning_rate": 1.2374683155584677e-05,
      "loss": 0.9509,
      "step": 782
    },
    {
      "epoch": 0.48214285714285715,
      "grad_norm": 0.6859679272753326,
      "learning_rate": 1.2353789739897437e-05,
      "loss": 0.8518,
      "step": 783
    },
    {
      "epoch": 0.4827586206896552,
      "grad_norm": 0.7731564832079155,
      "learning_rate": 1.2332885440754552e-05,
      "loss": 0.8886,
      "step": 784
    },
    {
      "epoch": 0.4833743842364532,
      "grad_norm": 0.6574688278065861,
      "learning_rate": 1.2311970354813345e-05,
      "loss": 0.8859,
      "step": 785
    },
    {
      "epoch": 0.4839901477832512,
      "grad_norm": 0.7287958371716784,
      "learning_rate": 1.2291044578781015e-05,
      "loss": 0.9429,
      "step": 786
    },
    {
      "epoch": 0.48460591133004927,
      "grad_norm": 0.7763061554506329,
      "learning_rate": 1.2270108209414186e-05,
      "loss": 0.8956,
      "step": 787
    },
    {
      "epoch": 0.4852216748768473,
      "grad_norm": 0.7440089537078,
      "learning_rate": 1.2249161343518467e-05,
      "loss": 0.8517,
      "step": 788
    },
    {
      "epoch": 0.4858374384236453,
      "grad_norm": 0.7201825141632505,
      "learning_rate": 1.2228204077948002e-05,
      "loss": 0.8953,
      "step": 789
    },
    {
      "epoch": 0.48645320197044334,
      "grad_norm": 0.7222961232854457,
      "learning_rate": 1.220723650960502e-05,
      "loss": 0.9217,
      "step": 790
    },
    {
      "epoch": 0.4870689655172414,
      "grad_norm": 0.727265926531914,
      "learning_rate": 1.2186258735439381e-05,
      "loss": 0.9337,
      "step": 791
    },
    {
      "epoch": 0.4876847290640394,
      "grad_norm": 0.7477961961346524,
      "learning_rate": 1.2165270852448144e-05,
      "loss": 0.9751,
      "step": 792
    },
    {
      "epoch": 0.4883004926108374,
      "grad_norm": 0.7092375179687316,
      "learning_rate": 1.214427295767511e-05,
      "loss": 0.8721,
      "step": 793
    },
    {
      "epoch": 0.48891625615763545,
      "grad_norm": 0.7569508154972621,
      "learning_rate": 1.2123265148210363e-05,
      "loss": 0.8979,
      "step": 794
    },
    {
      "epoch": 0.4895320197044335,
      "grad_norm": 0.7258679222417804,
      "learning_rate": 1.2102247521189838e-05,
      "loss": 0.9572,
      "step": 795
    },
    {
      "epoch": 0.49014778325123154,
      "grad_norm": 0.7773125618692818,
      "learning_rate": 1.2081220173794862e-05,
      "loss": 0.8931,
      "step": 796
    },
    {
      "epoch": 0.4907635467980296,
      "grad_norm": 0.759218411113691,
      "learning_rate": 1.2060183203251709e-05,
      "loss": 0.8953,
      "step": 797
    },
    {
      "epoch": 0.49137931034482757,
      "grad_norm": 0.7557020335237933,
      "learning_rate": 1.2039136706831146e-05,
      "loss": 0.9233,
      "step": 798
    },
    {
      "epoch": 0.4919950738916256,
      "grad_norm": 0.8093515030007062,
      "learning_rate": 1.2018080781847985e-05,
      "loss": 0.9318,
      "step": 799
    },
    {
      "epoch": 0.49261083743842365,
      "grad_norm": 0.8277020506855606,
      "learning_rate": 1.199701552566064e-05,
      "loss": 0.9188,
      "step": 800
    },
    {
      "epoch": 0.49261083743842365,
      "eval_loss": 0.9125872254371643,
      "eval_runtime": 524.0363,
      "eval_samples_per_second": 44.1,
      "eval_steps_per_second": 0.691,
      "step": 800
    },
    {
      "epoch": 0.4932266009852217,
      "grad_norm": 0.8705091025306452,
      "learning_rate": 1.1975941035670664e-05,
      "loss": 0.954,
      "step": 801
    },
    {
      "epoch": 0.4938423645320197,
      "grad_norm": 0.8710895270894086,
      "learning_rate": 1.1954857409322302e-05,
      "loss": 0.9036,
      "step": 802
    },
    {
      "epoch": 0.4944581280788177,
      "grad_norm": 0.9122224081988305,
      "learning_rate": 1.1933764744102058e-05,
      "loss": 0.9197,
      "step": 803
    },
    {
      "epoch": 0.49507389162561577,
      "grad_norm": 0.7365837488952112,
      "learning_rate": 1.1912663137538219e-05,
      "loss": 0.912,
      "step": 804
    },
    {
      "epoch": 0.4956896551724138,
      "grad_norm": 0.7197397097811458,
      "learning_rate": 1.1891552687200414e-05,
      "loss": 0.9578,
      "step": 805
    },
    {
      "epoch": 0.4963054187192118,
      "grad_norm": 0.8191474262471599,
      "learning_rate": 1.1870433490699167e-05,
      "loss": 0.9079,
      "step": 806
    },
    {
      "epoch": 0.49692118226600984,
      "grad_norm": 0.7444398905235264,
      "learning_rate": 1.1849305645685446e-05,
      "loss": 0.8663,
      "step": 807
    },
    {
      "epoch": 0.4975369458128079,
      "grad_norm": 0.7797923574138393,
      "learning_rate": 1.1828169249850201e-05,
      "loss": 0.9071,
      "step": 808
    },
    {
      "epoch": 0.4981527093596059,
      "grad_norm": 0.726523629031649,
      "learning_rate": 1.1807024400923926e-05,
      "loss": 0.917,
      "step": 809
    },
    {
      "epoch": 0.4987684729064039,
      "grad_norm": 0.7352400808317435,
      "learning_rate": 1.1785871196676196e-05,
      "loss": 0.9266,
      "step": 810
    },
    {
      "epoch": 0.49938423645320196,
      "grad_norm": 0.8843951705916415,
      "learning_rate": 1.1764709734915218e-05,
      "loss": 0.8698,
      "step": 811
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7504513220566594,
      "learning_rate": 1.1743540113487383e-05,
      "loss": 0.9179,
      "step": 812
    },
    {
      "epoch": 0.500615763546798,
      "grad_norm": 0.8713684989796586,
      "learning_rate": 1.1722362430276817e-05,
      "loss": 0.974,
      "step": 813
    },
    {
      "epoch": 0.5012315270935961,
      "grad_norm": 0.8095447909522274,
      "learning_rate": 1.1701176783204908e-05,
      "loss": 0.9266,
      "step": 814
    },
    {
      "epoch": 0.5018472906403941,
      "grad_norm": 0.6682186131685554,
      "learning_rate": 1.167998327022988e-05,
      "loss": 0.8817,
      "step": 815
    },
    {
      "epoch": 0.5024630541871922,
      "grad_norm": 0.6615645080189052,
      "learning_rate": 1.165878198934632e-05,
      "loss": 0.9017,
      "step": 816
    },
    {
      "epoch": 0.5030788177339901,
      "grad_norm": 0.7383253800572941,
      "learning_rate": 1.1637573038584729e-05,
      "loss": 0.8618,
      "step": 817
    },
    {
      "epoch": 0.5036945812807881,
      "grad_norm": 0.7521204218690731,
      "learning_rate": 1.1616356516011083e-05,
      "loss": 0.8943,
      "step": 818
    },
    {
      "epoch": 0.5043103448275862,
      "grad_norm": 0.7852008644303985,
      "learning_rate": 1.1595132519726365e-05,
      "loss": 0.897,
      "step": 819
    },
    {
      "epoch": 0.5049261083743842,
      "grad_norm": 0.7297091027843033,
      "learning_rate": 1.1573901147866108e-05,
      "loss": 0.9922,
      "step": 820
    },
    {
      "epoch": 0.5055418719211823,
      "grad_norm": 0.7313783744949351,
      "learning_rate": 1.1552662498599957e-05,
      "loss": 0.8746,
      "step": 821
    },
    {
      "epoch": 0.5061576354679803,
      "grad_norm": 0.6685216911809002,
      "learning_rate": 1.1531416670131197e-05,
      "loss": 0.8604,
      "step": 822
    },
    {
      "epoch": 0.5067733990147784,
      "grad_norm": 0.7641288533323628,
      "learning_rate": 1.151016376069632e-05,
      "loss": 0.9158,
      "step": 823
    },
    {
      "epoch": 0.5073891625615764,
      "grad_norm": 0.7527574256781122,
      "learning_rate": 1.1488903868564548e-05,
      "loss": 0.9536,
      "step": 824
    },
    {
      "epoch": 0.5080049261083743,
      "grad_norm": 0.7078038295091879,
      "learning_rate": 1.1467637092037399e-05,
      "loss": 0.9382,
      "step": 825
    },
    {
      "epoch": 0.5086206896551724,
      "grad_norm": 0.6758486089355186,
      "learning_rate": 1.144636352944821e-05,
      "loss": 0.9298,
      "step": 826
    },
    {
      "epoch": 0.5092364532019704,
      "grad_norm": 0.7252374928700351,
      "learning_rate": 1.142508327916171e-05,
      "loss": 0.9076,
      "step": 827
    },
    {
      "epoch": 0.5098522167487685,
      "grad_norm": 0.7092032225089651,
      "learning_rate": 1.1403796439573545e-05,
      "loss": 0.9096,
      "step": 828
    },
    {
      "epoch": 0.5104679802955665,
      "grad_norm": 0.6867178874515946,
      "learning_rate": 1.1382503109109822e-05,
      "loss": 0.8684,
      "step": 829
    },
    {
      "epoch": 0.5110837438423645,
      "grad_norm": 0.6602056713969897,
      "learning_rate": 1.1361203386226672e-05,
      "loss": 0.8958,
      "step": 830
    },
    {
      "epoch": 0.5116995073891626,
      "grad_norm": 0.7206657833500593,
      "learning_rate": 1.1339897369409774e-05,
      "loss": 0.9132,
      "step": 831
    },
    {
      "epoch": 0.5123152709359606,
      "grad_norm": 0.7150647992144722,
      "learning_rate": 1.1318585157173913e-05,
      "loss": 0.8845,
      "step": 832
    },
    {
      "epoch": 0.5129310344827587,
      "grad_norm": 0.7396323559482656,
      "learning_rate": 1.129726684806252e-05,
      "loss": 0.8921,
      "step": 833
    },
    {
      "epoch": 0.5135467980295566,
      "grad_norm": 0.6873794205698561,
      "learning_rate": 1.127594254064722e-05,
      "loss": 0.8679,
      "step": 834
    },
    {
      "epoch": 0.5141625615763546,
      "grad_norm": 0.746882590359839,
      "learning_rate": 1.1254612333527368e-05,
      "loss": 0.8696,
      "step": 835
    },
    {
      "epoch": 0.5147783251231527,
      "grad_norm": 0.7311017740853613,
      "learning_rate": 1.1233276325329596e-05,
      "loss": 0.903,
      "step": 836
    },
    {
      "epoch": 0.5153940886699507,
      "grad_norm": 0.7088658398561933,
      "learning_rate": 1.1211934614707365e-05,
      "loss": 0.9134,
      "step": 837
    },
    {
      "epoch": 0.5160098522167488,
      "grad_norm": 0.7573210382991106,
      "learning_rate": 1.11905873003405e-05,
      "loss": 0.9008,
      "step": 838
    },
    {
      "epoch": 0.5166256157635468,
      "grad_norm": 0.7411917020528135,
      "learning_rate": 1.1169234480934741e-05,
      "loss": 0.9342,
      "step": 839
    },
    {
      "epoch": 0.5172413793103449,
      "grad_norm": 0.7245845248364069,
      "learning_rate": 1.1147876255221274e-05,
      "loss": 0.9355,
      "step": 840
    },
    {
      "epoch": 0.5178571428571429,
      "grad_norm": 0.7709650538053515,
      "learning_rate": 1.1126512721956288e-05,
      "loss": 0.925,
      "step": 841
    },
    {
      "epoch": 0.5184729064039408,
      "grad_norm": 0.791007172051326,
      "learning_rate": 1.1105143979920512e-05,
      "loss": 0.9426,
      "step": 842
    },
    {
      "epoch": 0.5190886699507389,
      "grad_norm": 0.7408004017390732,
      "learning_rate": 1.1083770127918763e-05,
      "loss": 0.935,
      "step": 843
    },
    {
      "epoch": 0.5197044334975369,
      "grad_norm": 0.6954555925090047,
      "learning_rate": 1.1062391264779473e-05,
      "loss": 0.8841,
      "step": 844
    },
    {
      "epoch": 0.520320197044335,
      "grad_norm": 0.7443571881222744,
      "learning_rate": 1.1041007489354263e-05,
      "loss": 0.8361,
      "step": 845
    },
    {
      "epoch": 0.520935960591133,
      "grad_norm": 0.7900646235188628,
      "learning_rate": 1.1019618900517455e-05,
      "loss": 0.9083,
      "step": 846
    },
    {
      "epoch": 0.521551724137931,
      "grad_norm": 0.7066477210918738,
      "learning_rate": 1.0998225597165628e-05,
      "loss": 0.887,
      "step": 847
    },
    {
      "epoch": 0.5221674876847291,
      "grad_norm": 0.735861566560009,
      "learning_rate": 1.0976827678217162e-05,
      "loss": 0.8925,
      "step": 848
    },
    {
      "epoch": 0.5227832512315271,
      "grad_norm": 0.7379444244185547,
      "learning_rate": 1.0955425242611781e-05,
      "loss": 0.8976,
      "step": 849
    },
    {
      "epoch": 0.5233990147783252,
      "grad_norm": 0.685087576780142,
      "learning_rate": 1.093401838931009e-05,
      "loss": 0.8627,
      "step": 850
    },
    {
      "epoch": 0.5240147783251231,
      "grad_norm": 0.6890717315459193,
      "learning_rate": 1.0912607217293121e-05,
      "loss": 0.8765,
      "step": 851
    },
    {
      "epoch": 0.5246305418719212,
      "grad_norm": 0.7950098705434928,
      "learning_rate": 1.0891191825561874e-05,
      "loss": 0.9747,
      "step": 852
    },
    {
      "epoch": 0.5252463054187192,
      "grad_norm": 0.7630868540865025,
      "learning_rate": 1.0869772313136861e-05,
      "loss": 0.9585,
      "step": 853
    },
    {
      "epoch": 0.5258620689655172,
      "grad_norm": 0.7100522083917626,
      "learning_rate": 1.0848348779057654e-05,
      "loss": 0.9302,
      "step": 854
    },
    {
      "epoch": 0.5264778325123153,
      "grad_norm": 0.8220642399730556,
      "learning_rate": 1.0826921322382407e-05,
      "loss": 0.9272,
      "step": 855
    },
    {
      "epoch": 0.5270935960591133,
      "grad_norm": 0.7284956872894887,
      "learning_rate": 1.080549004218742e-05,
      "loss": 0.9045,
      "step": 856
    },
    {
      "epoch": 0.5277093596059114,
      "grad_norm": 0.6902918868684503,
      "learning_rate": 1.078405503756667e-05,
      "loss": 0.8732,
      "step": 857
    },
    {
      "epoch": 0.5283251231527094,
      "grad_norm": 0.7362499254605412,
      "learning_rate": 1.0762616407631356e-05,
      "loss": 0.8532,
      "step": 858
    },
    {
      "epoch": 0.5289408866995073,
      "grad_norm": 0.8017503242675776,
      "learning_rate": 1.0741174251509443e-05,
      "loss": 0.8621,
      "step": 859
    },
    {
      "epoch": 0.5295566502463054,
      "grad_norm": 0.7475464361866181,
      "learning_rate": 1.071972866834519e-05,
      "loss": 0.9086,
      "step": 860
    },
    {
      "epoch": 0.5301724137931034,
      "grad_norm": 0.7585422872451106,
      "learning_rate": 1.0698279757298715e-05,
      "loss": 0.9238,
      "step": 861
    },
    {
      "epoch": 0.5307881773399015,
      "grad_norm": 0.6803722099054084,
      "learning_rate": 1.0676827617545511e-05,
      "loss": 0.9034,
      "step": 862
    },
    {
      "epoch": 0.5314039408866995,
      "grad_norm": 0.740736005000993,
      "learning_rate": 1.0655372348276007e-05,
      "loss": 0.9059,
      "step": 863
    },
    {
      "epoch": 0.5320197044334976,
      "grad_norm": 0.8021855381784778,
      "learning_rate": 1.0633914048695104e-05,
      "loss": 0.8599,
      "step": 864
    },
    {
      "epoch": 0.5326354679802956,
      "grad_norm": 0.7127165043924963,
      "learning_rate": 1.061245281802171e-05,
      "loss": 0.9014,
      "step": 865
    },
    {
      "epoch": 0.5332512315270936,
      "grad_norm": 0.6957371546213569,
      "learning_rate": 1.0590988755488292e-05,
      "loss": 0.8902,
      "step": 866
    },
    {
      "epoch": 0.5338669950738916,
      "grad_norm": 0.691570964127424,
      "learning_rate": 1.0569521960340399e-05,
      "loss": 0.8817,
      "step": 867
    },
    {
      "epoch": 0.5344827586206896,
      "grad_norm": 0.6896635826518177,
      "learning_rate": 1.0548052531836225e-05,
      "loss": 0.9121,
      "step": 868
    },
    {
      "epoch": 0.5350985221674877,
      "grad_norm": 0.6581526335230111,
      "learning_rate": 1.0526580569246138e-05,
      "loss": 0.877,
      "step": 869
    },
    {
      "epoch": 0.5357142857142857,
      "grad_norm": 0.7031472332257903,
      "learning_rate": 1.0505106171852226e-05,
      "loss": 0.901,
      "step": 870
    },
    {
      "epoch": 0.5363300492610837,
      "grad_norm": 0.7263699642263189,
      "learning_rate": 1.0483629438947827e-05,
      "loss": 0.9206,
      "step": 871
    },
    {
      "epoch": 0.5369458128078818,
      "grad_norm": 0.7400337248026725,
      "learning_rate": 1.0462150469837085e-05,
      "loss": 0.9399,
      "step": 872
    },
    {
      "epoch": 0.5375615763546798,
      "grad_norm": 0.6754764681503417,
      "learning_rate": 1.0440669363834483e-05,
      "loss": 0.9185,
      "step": 873
    },
    {
      "epoch": 0.5381773399014779,
      "grad_norm": 0.7132117928818225,
      "learning_rate": 1.0419186220264379e-05,
      "loss": 0.9177,
      "step": 874
    },
    {
      "epoch": 0.5387931034482759,
      "grad_norm": 0.6958653260599712,
      "learning_rate": 1.039770113846056e-05,
      "loss": 0.926,
      "step": 875
    },
    {
      "epoch": 0.5394088669950738,
      "grad_norm": 0.7080339280050165,
      "learning_rate": 1.0376214217765773e-05,
      "loss": 0.9092,
      "step": 876
    },
    {
      "epoch": 0.5400246305418719,
      "grad_norm": 0.679372205843852,
      "learning_rate": 1.0354725557531258e-05,
      "loss": 0.874,
      "step": 877
    },
    {
      "epoch": 0.5406403940886699,
      "grad_norm": 0.7038060172887204,
      "learning_rate": 1.0333235257116314e-05,
      "loss": 0.9355,
      "step": 878
    },
    {
      "epoch": 0.541256157635468,
      "grad_norm": 0.790889041221645,
      "learning_rate": 1.0311743415887811e-05,
      "loss": 0.9474,
      "step": 879
    },
    {
      "epoch": 0.541871921182266,
      "grad_norm": 0.7534118519908194,
      "learning_rate": 1.0290250133219755e-05,
      "loss": 0.9097,
      "step": 880
    },
    {
      "epoch": 0.5424876847290641,
      "grad_norm": 0.7291813566043958,
      "learning_rate": 1.0268755508492805e-05,
      "loss": 0.9161,
      "step": 881
    },
    {
      "epoch": 0.5431034482758621,
      "grad_norm": 0.6645908172012055,
      "learning_rate": 1.0247259641093834e-05,
      "loss": 0.8739,
      "step": 882
    },
    {
      "epoch": 0.5437192118226601,
      "grad_norm": 0.7667243948949509,
      "learning_rate": 1.0225762630415456e-05,
      "loss": 0.9201,
      "step": 883
    },
    {
      "epoch": 0.5443349753694581,
      "grad_norm": 0.6588971366250324,
      "learning_rate": 1.020426457585557e-05,
      "loss": 0.8972,
      "step": 884
    },
    {
      "epoch": 0.5449507389162561,
      "grad_norm": 0.7226635472351208,
      "learning_rate": 1.0182765576816916e-05,
      "loss": 0.9059,
      "step": 885
    },
    {
      "epoch": 0.5455665024630542,
      "grad_norm": 0.732962735187896,
      "learning_rate": 1.016126573270658e-05,
      "loss": 0.9349,
      "step": 886
    },
    {
      "epoch": 0.5461822660098522,
      "grad_norm": 0.7371413060542562,
      "learning_rate": 1.0139765142935566e-05,
      "loss": 0.8734,
      "step": 887
    },
    {
      "epoch": 0.5467980295566502,
      "grad_norm": 0.6667788574675981,
      "learning_rate": 1.011826390691833e-05,
      "loss": 0.8777,
      "step": 888
    },
    {
      "epoch": 0.5474137931034483,
      "grad_norm": 0.6821875909000377,
      "learning_rate": 1.0096762124072306e-05,
      "loss": 0.8724,
      "step": 889
    },
    {
      "epoch": 0.5480295566502463,
      "grad_norm": 0.7236226568165169,
      "learning_rate": 1.0075259893817465e-05,
      "loss": 0.8705,
      "step": 890
    },
    {
      "epoch": 0.5486453201970444,
      "grad_norm": 0.7029581397414398,
      "learning_rate": 1.005375731557584e-05,
      "loss": 0.8936,
      "step": 891
    },
    {
      "epoch": 0.5492610837438424,
      "grad_norm": 0.7106732672161409,
      "learning_rate": 1.003225448877108e-05,
      "loss": 0.9227,
      "step": 892
    },
    {
      "epoch": 0.5498768472906403,
      "grad_norm": 0.6925608446579796,
      "learning_rate": 1.001075151282798e-05,
      "loss": 0.9164,
      "step": 893
    },
    {
      "epoch": 0.5504926108374384,
      "grad_norm": 0.73525241680307,
      "learning_rate": 9.989248487172022e-06,
      "loss": 0.9308,
      "step": 894
    },
    {
      "epoch": 0.5511083743842364,
      "grad_norm": 0.7247177551259489,
      "learning_rate": 9.967745511228922e-06,
      "loss": 0.8929,
      "step": 895
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 0.6841093396955116,
      "learning_rate": 9.946242684424163e-06,
      "loss": 0.8865,
      "step": 896
    },
    {
      "epoch": 0.5523399014778325,
      "grad_norm": 0.7372258916619508,
      "learning_rate": 9.92474010618254e-06,
      "loss": 0.9217,
      "step": 897
    },
    {
      "epoch": 0.5529556650246306,
      "grad_norm": 0.6898975186159325,
      "learning_rate": 9.903237875927697e-06,
      "loss": 0.893,
      "step": 898
    },
    {
      "epoch": 0.5535714285714286,
      "grad_norm": 0.7003278063557198,
      "learning_rate": 9.881736093081674e-06,
      "loss": 0.9409,
      "step": 899
    },
    {
      "epoch": 0.5541871921182266,
      "grad_norm": 0.7110259237745857,
      "learning_rate": 9.860234857064439e-06,
      "loss": 0.8967,
      "step": 900
    },
    {
      "epoch": 0.5548029556650246,
      "grad_norm": 0.730759555736587,
      "learning_rate": 9.838734267293421e-06,
      "loss": 0.896,
      "step": 901
    },
    {
      "epoch": 0.5554187192118226,
      "grad_norm": 0.6970111403288594,
      "learning_rate": 9.817234423183086e-06,
      "loss": 0.8754,
      "step": 902
    },
    {
      "epoch": 0.5560344827586207,
      "grad_norm": 0.7113323795773189,
      "learning_rate": 9.795735424144427e-06,
      "loss": 0.9242,
      "step": 903
    },
    {
      "epoch": 0.5566502463054187,
      "grad_norm": 0.7554641597510858,
      "learning_rate": 9.774237369584547e-06,
      "loss": 0.8513,
      "step": 904
    },
    {
      "epoch": 0.5572660098522167,
      "grad_norm": 0.6963119030960164,
      "learning_rate": 9.752740358906168e-06,
      "loss": 0.9036,
      "step": 905
    },
    {
      "epoch": 0.5578817733990148,
      "grad_norm": 0.6857086776829809,
      "learning_rate": 9.731244491507196e-06,
      "loss": 0.94,
      "step": 906
    },
    {
      "epoch": 0.5584975369458128,
      "grad_norm": 0.6459304023601538,
      "learning_rate": 9.709749866780248e-06,
      "loss": 0.8537,
      "step": 907
    },
    {
      "epoch": 0.5591133004926109,
      "grad_norm": 0.6780251226828853,
      "learning_rate": 9.688256584112192e-06,
      "loss": 0.8963,
      "step": 908
    },
    {
      "epoch": 0.5597290640394089,
      "grad_norm": 0.6910629394481728,
      "learning_rate": 9.66676474288369e-06,
      "loss": 0.8666,
      "step": 909
    },
    {
      "epoch": 0.5603448275862069,
      "grad_norm": 0.6754288130144743,
      "learning_rate": 9.645274442468746e-06,
      "loss": 0.905,
      "step": 910
    },
    {
      "epoch": 0.5609605911330049,
      "grad_norm": 0.8186863835215311,
      "learning_rate": 9.623785782234234e-06,
      "loss": 0.908,
      "step": 911
    },
    {
      "epoch": 0.5615763546798029,
      "grad_norm": 0.6955165931156284,
      "learning_rate": 9.602298861539446e-06,
      "loss": 0.8692,
      "step": 912
    },
    {
      "epoch": 0.562192118226601,
      "grad_norm": 0.7480318106353373,
      "learning_rate": 9.580813779735624e-06,
      "loss": 0.9139,
      "step": 913
    },
    {
      "epoch": 0.562807881773399,
      "grad_norm": 0.708786870369378,
      "learning_rate": 9.55933063616552e-06,
      "loss": 0.943,
      "step": 914
    },
    {
      "epoch": 0.5634236453201971,
      "grad_norm": 0.6716768522183068,
      "learning_rate": 9.537849530162915e-06,
      "loss": 0.8275,
      "step": 915
    },
    {
      "epoch": 0.5640394088669951,
      "grad_norm": 0.7138827197394422,
      "learning_rate": 9.516370561052173e-06,
      "loss": 0.8949,
      "step": 916
    },
    {
      "epoch": 0.5646551724137931,
      "grad_norm": 0.7005350108310532,
      "learning_rate": 9.494893828147777e-06,
      "loss": 0.866,
      "step": 917
    },
    {
      "epoch": 0.5652709359605911,
      "grad_norm": 0.7635198603580495,
      "learning_rate": 9.473419430753864e-06,
      "loss": 0.9096,
      "step": 918
    },
    {
      "epoch": 0.5658866995073891,
      "grad_norm": 0.822348987809011,
      "learning_rate": 9.451947468163779e-06,
      "loss": 0.8957,
      "step": 919
    },
    {
      "epoch": 0.5665024630541872,
      "grad_norm": 0.6838276805124194,
      "learning_rate": 9.430478039659604e-06,
      "loss": 0.8718,
      "step": 920
    },
    {
      "epoch": 0.5671182266009852,
      "grad_norm": 0.756287728619952,
      "learning_rate": 9.409011244511713e-06,
      "loss": 0.8692,
      "step": 921
    },
    {
      "epoch": 0.5677339901477833,
      "grad_norm": 0.6867466189340491,
      "learning_rate": 9.387547181978291e-06,
      "loss": 0.8801,
      "step": 922
    },
    {
      "epoch": 0.5683497536945813,
      "grad_norm": 0.7697497206746393,
      "learning_rate": 9.3660859513049e-06,
      "loss": 0.8552,
      "step": 923
    },
    {
      "epoch": 0.5689655172413793,
      "grad_norm": 0.7726642759485614,
      "learning_rate": 9.344627651723995e-06,
      "loss": 0.9596,
      "step": 924
    },
    {
      "epoch": 0.5695812807881774,
      "grad_norm": 0.8683256748579654,
      "learning_rate": 9.323172382454494e-06,
      "loss": 0.926,
      "step": 925
    },
    {
      "epoch": 0.5701970443349754,
      "grad_norm": 0.7268267299623987,
      "learning_rate": 9.301720242701291e-06,
      "loss": 0.8669,
      "step": 926
    },
    {
      "epoch": 0.5708128078817734,
      "grad_norm": 0.7699568091695068,
      "learning_rate": 9.28027133165481e-06,
      "loss": 0.8932,
      "step": 927
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.7769664750978159,
      "learning_rate": 9.258825748490559e-06,
      "loss": 0.9258,
      "step": 928
    },
    {
      "epoch": 0.5720443349753694,
      "grad_norm": 0.7593995324299458,
      "learning_rate": 9.237383592368644e-06,
      "loss": 0.8906,
      "step": 929
    },
    {
      "epoch": 0.5726600985221675,
      "grad_norm": 0.7339900638688633,
      "learning_rate": 9.215944962433329e-06,
      "loss": 0.8784,
      "step": 930
    },
    {
      "epoch": 0.5732758620689655,
      "grad_norm": 0.7013492553982281,
      "learning_rate": 9.194509957812583e-06,
      "loss": 0.8955,
      "step": 931
    },
    {
      "epoch": 0.5738916256157636,
      "grad_norm": 0.7283087345271415,
      "learning_rate": 9.173078677617596e-06,
      "loss": 0.8903,
      "step": 932
    },
    {
      "epoch": 0.5745073891625616,
      "grad_norm": 0.7148119710335699,
      "learning_rate": 9.15165122094235e-06,
      "loss": 0.9049,
      "step": 933
    },
    {
      "epoch": 0.5751231527093597,
      "grad_norm": 0.75207246220578,
      "learning_rate": 9.130227686863142e-06,
      "loss": 0.9461,
      "step": 934
    },
    {
      "epoch": 0.5757389162561576,
      "grad_norm": 0.7958065562798486,
      "learning_rate": 9.108808174438128e-06,
      "loss": 0.9027,
      "step": 935
    },
    {
      "epoch": 0.5763546798029556,
      "grad_norm": 0.6788552583780807,
      "learning_rate": 9.087392782706884e-06,
      "loss": 0.9202,
      "step": 936
    },
    {
      "epoch": 0.5769704433497537,
      "grad_norm": 0.6810559421882132,
      "learning_rate": 9.065981610689915e-06,
      "loss": 0.865,
      "step": 937
    },
    {
      "epoch": 0.5775862068965517,
      "grad_norm": 0.7995092690604082,
      "learning_rate": 9.044574757388224e-06,
      "loss": 0.9062,
      "step": 938
    },
    {
      "epoch": 0.5782019704433498,
      "grad_norm": 0.7668211334590852,
      "learning_rate": 9.02317232178284e-06,
      "loss": 0.9052,
      "step": 939
    },
    {
      "epoch": 0.5788177339901478,
      "grad_norm": 0.8309802706938013,
      "learning_rate": 9.001774402834375e-06,
      "loss": 0.9421,
      "step": 940
    },
    {
      "epoch": 0.5794334975369458,
      "grad_norm": 0.775769275356401,
      "learning_rate": 8.980381099482545e-06,
      "loss": 0.9077,
      "step": 941
    },
    {
      "epoch": 0.5800492610837439,
      "grad_norm": 0.792605932859535,
      "learning_rate": 8.958992510645737e-06,
      "loss": 0.8848,
      "step": 942
    },
    {
      "epoch": 0.5806650246305419,
      "grad_norm": 0.7332658074922906,
      "learning_rate": 8.937608735220527e-06,
      "loss": 0.9379,
      "step": 943
    },
    {
      "epoch": 0.5812807881773399,
      "grad_norm": 0.7163507430062284,
      "learning_rate": 8.916229872081242e-06,
      "loss": 0.8797,
      "step": 944
    },
    {
      "epoch": 0.5818965517241379,
      "grad_norm": 0.8016070284136756,
      "learning_rate": 8.89485602007949e-06,
      "loss": 0.8745,
      "step": 945
    },
    {
      "epoch": 0.5825123152709359,
      "grad_norm": 0.7148711901711244,
      "learning_rate": 8.873487278043713e-06,
      "loss": 0.8856,
      "step": 946
    },
    {
      "epoch": 0.583128078817734,
      "grad_norm": 0.7353079455018744,
      "learning_rate": 8.852123744778729e-06,
      "loss": 0.8895,
      "step": 947
    },
    {
      "epoch": 0.583743842364532,
      "grad_norm": 0.69470589152061,
      "learning_rate": 8.830765519065262e-06,
      "loss": 0.8968,
      "step": 948
    },
    {
      "epoch": 0.5843596059113301,
      "grad_norm": 0.6999280065046031,
      "learning_rate": 8.809412699659503e-06,
      "loss": 0.8888,
      "step": 949
    },
    {
      "epoch": 0.5849753694581281,
      "grad_norm": 0.8274458542225668,
      "learning_rate": 8.788065385292637e-06,
      "loss": 0.9018,
      "step": 950
    },
    {
      "epoch": 0.5855911330049262,
      "grad_norm": 0.7045024288048022,
      "learning_rate": 8.766723674670407e-06,
      "loss": 0.8833,
      "step": 951
    },
    {
      "epoch": 0.5862068965517241,
      "grad_norm": 0.6993860744579231,
      "learning_rate": 8.745387666472639e-06,
      "loss": 0.9045,
      "step": 952
    },
    {
      "epoch": 0.5868226600985221,
      "grad_norm": 0.7436681792661218,
      "learning_rate": 8.724057459352783e-06,
      "loss": 0.8744,
      "step": 953
    },
    {
      "epoch": 0.5874384236453202,
      "grad_norm": 0.7532127413813657,
      "learning_rate": 8.70273315193748e-06,
      "loss": 0.9119,
      "step": 954
    },
    {
      "epoch": 0.5880541871921182,
      "grad_norm": 0.7011039982520797,
      "learning_rate": 8.68141484282609e-06,
      "loss": 0.8648,
      "step": 955
    },
    {
      "epoch": 0.5886699507389163,
      "grad_norm": 0.7056076516401616,
      "learning_rate": 8.660102630590228e-06,
      "loss": 0.934,
      "step": 956
    },
    {
      "epoch": 0.5892857142857143,
      "grad_norm": 0.7099288580757988,
      "learning_rate": 8.63879661377333e-06,
      "loss": 0.8382,
      "step": 957
    },
    {
      "epoch": 0.5899014778325123,
      "grad_norm": 0.725341688382521,
      "learning_rate": 8.61749689089018e-06,
      "loss": 0.8843,
      "step": 958
    },
    {
      "epoch": 0.5905172413793104,
      "grad_norm": 0.7237544882086393,
      "learning_rate": 8.59620356042646e-06,
      "loss": 0.8991,
      "step": 959
    },
    {
      "epoch": 0.5911330049261084,
      "grad_norm": 0.6699678262558691,
      "learning_rate": 8.574916720838293e-06,
      "loss": 0.8687,
      "step": 960
    },
    {
      "epoch": 0.5917487684729064,
      "grad_norm": 0.6832087254391259,
      "learning_rate": 8.553636470551792e-06,
      "loss": 0.9192,
      "step": 961
    },
    {
      "epoch": 0.5923645320197044,
      "grad_norm": 0.7362873695882963,
      "learning_rate": 8.532362907962606e-06,
      "loss": 0.901,
      "step": 962
    },
    {
      "epoch": 0.5929802955665024,
      "grad_norm": 0.7280051767941127,
      "learning_rate": 8.511096131435454e-06,
      "loss": 0.8985,
      "step": 963
    },
    {
      "epoch": 0.5935960591133005,
      "grad_norm": 0.6919078547929436,
      "learning_rate": 8.489836239303684e-06,
      "loss": 0.8474,
      "step": 964
    },
    {
      "epoch": 0.5942118226600985,
      "grad_norm": 0.755740958856751,
      "learning_rate": 8.468583329868805e-06,
      "loss": 0.9182,
      "step": 965
    },
    {
      "epoch": 0.5948275862068966,
      "grad_norm": 0.7180220462206086,
      "learning_rate": 8.447337501400048e-06,
      "loss": 0.9054,
      "step": 966
    },
    {
      "epoch": 0.5954433497536946,
      "grad_norm": 0.6845412538700358,
      "learning_rate": 8.426098852133892e-06,
      "loss": 0.859,
      "step": 967
    },
    {
      "epoch": 0.5960591133004927,
      "grad_norm": 0.6749128652402018,
      "learning_rate": 8.404867480273637e-06,
      "loss": 0.8574,
      "step": 968
    },
    {
      "epoch": 0.5966748768472906,
      "grad_norm": 0.6905816236736056,
      "learning_rate": 8.383643483988919e-06,
      "loss": 0.9057,
      "step": 969
    },
    {
      "epoch": 0.5972906403940886,
      "grad_norm": 0.7497805132215074,
      "learning_rate": 8.362426961415274e-06,
      "loss": 0.8879,
      "step": 970
    },
    {
      "epoch": 0.5979064039408867,
      "grad_norm": 0.6788135069181235,
      "learning_rate": 8.341218010653684e-06,
      "loss": 0.9156,
      "step": 971
    },
    {
      "epoch": 0.5985221674876847,
      "grad_norm": 0.6928404088637068,
      "learning_rate": 8.320016729770123e-06,
      "loss": 0.9211,
      "step": 972
    },
    {
      "epoch": 0.5991379310344828,
      "grad_norm": 0.7067871760344416,
      "learning_rate": 8.298823216795093e-06,
      "loss": 0.8492,
      "step": 973
    },
    {
      "epoch": 0.5997536945812808,
      "grad_norm": 0.7730840099850161,
      "learning_rate": 8.277637569723186e-06,
      "loss": 0.9035,
      "step": 974
    },
    {
      "epoch": 0.6003694581280788,
      "grad_norm": 0.6716186769932622,
      "learning_rate": 8.256459886512618e-06,
      "loss": 0.8575,
      "step": 975
    },
    {
      "epoch": 0.6009852216748769,
      "grad_norm": 0.7155148381986921,
      "learning_rate": 8.235290265084785e-06,
      "loss": 0.926,
      "step": 976
    },
    {
      "epoch": 0.6016009852216748,
      "grad_norm": 0.6964027473022767,
      "learning_rate": 8.214128803323809e-06,
      "loss": 0.8487,
      "step": 977
    },
    {
      "epoch": 0.6022167487684729,
      "grad_norm": 0.73094611687795,
      "learning_rate": 8.192975599076079e-06,
      "loss": 0.8629,
      "step": 978
    },
    {
      "epoch": 0.6028325123152709,
      "grad_norm": 0.7297141740884356,
      "learning_rate": 8.171830750149804e-06,
      "loss": 0.9034,
      "step": 979
    },
    {
      "epoch": 0.603448275862069,
      "grad_norm": 0.699969294705083,
      "learning_rate": 8.150694354314556e-06,
      "loss": 0.9028,
      "step": 980
    },
    {
      "epoch": 0.604064039408867,
      "grad_norm": 0.7343385645380645,
      "learning_rate": 8.129566509300834e-06,
      "loss": 0.8988,
      "step": 981
    },
    {
      "epoch": 0.604679802955665,
      "grad_norm": 0.712103939030973,
      "learning_rate": 8.108447312799588e-06,
      "loss": 0.9228,
      "step": 982
    },
    {
      "epoch": 0.6052955665024631,
      "grad_norm": 0.7240834667067108,
      "learning_rate": 8.087336862461783e-06,
      "loss": 0.8431,
      "step": 983
    },
    {
      "epoch": 0.6059113300492611,
      "grad_norm": 0.6556738033918945,
      "learning_rate": 8.066235255897943e-06,
      "loss": 0.9151,
      "step": 984
    },
    {
      "epoch": 0.6065270935960592,
      "grad_norm": 0.7380263211222373,
      "learning_rate": 8.0451425906777e-06,
      "loss": 0.8545,
      "step": 985
    },
    {
      "epoch": 0.6071428571428571,
      "grad_norm": 0.6934352406632428,
      "learning_rate": 8.02405896432934e-06,
      "loss": 0.8906,
      "step": 986
    },
    {
      "epoch": 0.6077586206896551,
      "grad_norm": 0.7191670006151082,
      "learning_rate": 8.002984474339363e-06,
      "loss": 0.8925,
      "step": 987
    },
    {
      "epoch": 0.6083743842364532,
      "grad_norm": 0.7131617682926624,
      "learning_rate": 7.981919218152017e-06,
      "loss": 0.8851,
      "step": 988
    },
    {
      "epoch": 0.6089901477832512,
      "grad_norm": 0.7516036870424501,
      "learning_rate": 7.960863293168858e-06,
      "loss": 0.8966,
      "step": 989
    },
    {
      "epoch": 0.6096059113300493,
      "grad_norm": 0.723040141127469,
      "learning_rate": 7.939816796748296e-06,
      "loss": 0.9118,
      "step": 990
    },
    {
      "epoch": 0.6102216748768473,
      "grad_norm": 0.694428648728375,
      "learning_rate": 7.918779826205141e-06,
      "loss": 0.9215,
      "step": 991
    },
    {
      "epoch": 0.6108374384236454,
      "grad_norm": 0.6803128863266967,
      "learning_rate": 7.897752478810166e-06,
      "loss": 0.9002,
      "step": 992
    },
    {
      "epoch": 0.6114532019704434,
      "grad_norm": 0.6953226292227428,
      "learning_rate": 7.876734851789642e-06,
      "loss": 0.9241,
      "step": 993
    },
    {
      "epoch": 0.6120689655172413,
      "grad_norm": 0.6595731764852006,
      "learning_rate": 7.855727042324892e-06,
      "loss": 0.9137,
      "step": 994
    },
    {
      "epoch": 0.6126847290640394,
      "grad_norm": 0.6464304991134646,
      "learning_rate": 7.834729147551858e-06,
      "loss": 0.8669,
      "step": 995
    },
    {
      "epoch": 0.6133004926108374,
      "grad_norm": 0.7000684577733264,
      "learning_rate": 7.813741264560622e-06,
      "loss": 0.9066,
      "step": 996
    },
    {
      "epoch": 0.6139162561576355,
      "grad_norm": 0.7418543104010067,
      "learning_rate": 7.792763490394983e-06,
      "loss": 0.9222,
      "step": 997
    },
    {
      "epoch": 0.6145320197044335,
      "grad_norm": 0.7200020574411867,
      "learning_rate": 7.771795922052e-06,
      "loss": 0.8988,
      "step": 998
    },
    {
      "epoch": 0.6151477832512315,
      "grad_norm": 0.6967270700744492,
      "learning_rate": 7.750838656481535e-06,
      "loss": 0.9504,
      "step": 999
    },
    {
      "epoch": 0.6157635467980296,
      "grad_norm": 0.728542891651735,
      "learning_rate": 7.729891790585817e-06,
      "loss": 0.967,
      "step": 1000
    },
    {
      "epoch": 0.6157635467980296,
      "eval_loss": 0.9036802053451538,
      "eval_runtime": 523.2874,
      "eval_samples_per_second": 44.163,
      "eval_steps_per_second": 0.692,
      "step": 1000
    },
    {
      "epoch": 0.6163793103448276,
      "grad_norm": 0.7399827647078763,
      "learning_rate": 7.708955421218987e-06,
      "loss": 0.8843,
      "step": 1001
    },
    {
      "epoch": 0.6169950738916257,
      "grad_norm": 0.7806304331026023,
      "learning_rate": 7.688029645186657e-06,
      "loss": 0.884,
      "step": 1002
    },
    {
      "epoch": 0.6176108374384236,
      "grad_norm": 0.7349953773834164,
      "learning_rate": 7.667114559245451e-06,
      "loss": 0.8698,
      "step": 1003
    },
    {
      "epoch": 0.6182266009852216,
      "grad_norm": 0.7077408048408157,
      "learning_rate": 7.646210260102566e-06,
      "loss": 0.8857,
      "step": 1004
    },
    {
      "epoch": 0.6188423645320197,
      "grad_norm": 0.7146919777327378,
      "learning_rate": 7.625316844415327e-06,
      "loss": 0.951,
      "step": 1005
    },
    {
      "epoch": 0.6194581280788177,
      "grad_norm": 0.7656501714607619,
      "learning_rate": 7.604434408790729e-06,
      "loss": 0.8663,
      "step": 1006
    },
    {
      "epoch": 0.6200738916256158,
      "grad_norm": 0.69577677743543,
      "learning_rate": 7.583563049785002e-06,
      "loss": 0.8841,
      "step": 1007
    },
    {
      "epoch": 0.6206896551724138,
      "grad_norm": 0.7996152982937744,
      "learning_rate": 7.562702863903161e-06,
      "loss": 0.9952,
      "step": 1008
    },
    {
      "epoch": 0.6213054187192119,
      "grad_norm": 0.7140005021266642,
      "learning_rate": 7.541853947598559e-06,
      "loss": 0.9028,
      "step": 1009
    },
    {
      "epoch": 0.6219211822660099,
      "grad_norm": 0.7399756292074203,
      "learning_rate": 7.521016397272436e-06,
      "loss": 0.891,
      "step": 1010
    },
    {
      "epoch": 0.6225369458128078,
      "grad_norm": 0.6842092636200658,
      "learning_rate": 7.50019030927348e-06,
      "loss": 0.8701,
      "step": 1011
    },
    {
      "epoch": 0.6231527093596059,
      "grad_norm": 0.7501039456328638,
      "learning_rate": 7.479375779897379e-06,
      "loss": 0.8708,
      "step": 1012
    },
    {
      "epoch": 0.6237684729064039,
      "grad_norm": 0.7152606388781787,
      "learning_rate": 7.458572905386381e-06,
      "loss": 0.901,
      "step": 1013
    },
    {
      "epoch": 0.624384236453202,
      "grad_norm": 0.734757620022921,
      "learning_rate": 7.437781781928839e-06,
      "loss": 0.8654,
      "step": 1014
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.6991097062782058,
      "learning_rate": 7.417002505658773e-06,
      "loss": 0.8687,
      "step": 1015
    },
    {
      "epoch": 0.625615763546798,
      "grad_norm": 0.685583466743735,
      "learning_rate": 7.396235172655423e-06,
      "loss": 0.9214,
      "step": 1016
    },
    {
      "epoch": 0.6262315270935961,
      "grad_norm": 0.7013999039879654,
      "learning_rate": 7.375479878942805e-06,
      "loss": 0.8807,
      "step": 1017
    },
    {
      "epoch": 0.6268472906403941,
      "grad_norm": 0.7277707896999317,
      "learning_rate": 7.354736720489274e-06,
      "loss": 0.9008,
      "step": 1018
    },
    {
      "epoch": 0.6274630541871922,
      "grad_norm": 0.7389829727982484,
      "learning_rate": 7.334005793207067e-06,
      "loss": 0.9063,
      "step": 1019
    },
    {
      "epoch": 0.6280788177339901,
      "grad_norm": 0.6863337645189733,
      "learning_rate": 7.313287192951866e-06,
      "loss": 0.9398,
      "step": 1020
    },
    {
      "epoch": 0.6286945812807881,
      "grad_norm": 0.7055571895559535,
      "learning_rate": 7.292581015522363e-06,
      "loss": 0.8933,
      "step": 1021
    },
    {
      "epoch": 0.6293103448275862,
      "grad_norm": 0.7286582787902749,
      "learning_rate": 7.271887356659804e-06,
      "loss": 0.9233,
      "step": 1022
    },
    {
      "epoch": 0.6299261083743842,
      "grad_norm": 0.6380393614129405,
      "learning_rate": 7.2512063120475475e-06,
      "loss": 0.8324,
      "step": 1023
    },
    {
      "epoch": 0.6305418719211823,
      "grad_norm": 0.6938127097867235,
      "learning_rate": 7.230537977310636e-06,
      "loss": 0.8657,
      "step": 1024
    },
    {
      "epoch": 0.6311576354679803,
      "grad_norm": 0.7169759107613062,
      "learning_rate": 7.209882448015338e-06,
      "loss": 0.9265,
      "step": 1025
    },
    {
      "epoch": 0.6317733990147784,
      "grad_norm": 0.6661974734595069,
      "learning_rate": 7.189239819668712e-06,
      "loss": 0.9092,
      "step": 1026
    },
    {
      "epoch": 0.6323891625615764,
      "grad_norm": 0.6839535306386562,
      "learning_rate": 7.168610187718164e-06,
      "loss": 0.9547,
      "step": 1027
    },
    {
      "epoch": 0.6330049261083743,
      "grad_norm": 0.7470239102226777,
      "learning_rate": 7.147993647551011e-06,
      "loss": 0.8929,
      "step": 1028
    },
    {
      "epoch": 0.6336206896551724,
      "grad_norm": 0.6724627375944633,
      "learning_rate": 7.1273902944940365e-06,
      "loss": 0.8487,
      "step": 1029
    },
    {
      "epoch": 0.6342364532019704,
      "grad_norm": 0.6590026701606456,
      "learning_rate": 7.1068002238130465e-06,
      "loss": 0.8592,
      "step": 1030
    },
    {
      "epoch": 0.6348522167487685,
      "grad_norm": 0.7616979831587749,
      "learning_rate": 7.086223530712433e-06,
      "loss": 0.9101,
      "step": 1031
    },
    {
      "epoch": 0.6354679802955665,
      "grad_norm": 0.7351013793945751,
      "learning_rate": 7.065660310334729e-06,
      "loss": 0.9059,
      "step": 1032
    },
    {
      "epoch": 0.6360837438423645,
      "grad_norm": 0.7526120972657393,
      "learning_rate": 7.045110657760179e-06,
      "loss": 0.9221,
      "step": 1033
    },
    {
      "epoch": 0.6366995073891626,
      "grad_norm": 0.8587735295905505,
      "learning_rate": 7.02457466800629e-06,
      "loss": 0.9059,
      "step": 1034
    },
    {
      "epoch": 0.6373152709359606,
      "grad_norm": 0.7436958251916905,
      "learning_rate": 7.004052436027397e-06,
      "loss": 0.8936,
      "step": 1035
    },
    {
      "epoch": 0.6379310344827587,
      "grad_norm": 0.7312142440915423,
      "learning_rate": 6.9835440567142155e-06,
      "loss": 0.8921,
      "step": 1036
    },
    {
      "epoch": 0.6385467980295566,
      "grad_norm": 0.7255786745849482,
      "learning_rate": 6.9630496248934144e-06,
      "loss": 0.9041,
      "step": 1037
    },
    {
      "epoch": 0.6391625615763546,
      "grad_norm": 0.7783713616066462,
      "learning_rate": 6.942569235327168e-06,
      "loss": 0.9981,
      "step": 1038
    },
    {
      "epoch": 0.6397783251231527,
      "grad_norm": 0.7549143343405446,
      "learning_rate": 6.9221029827127285e-06,
      "loss": 0.9755,
      "step": 1039
    },
    {
      "epoch": 0.6403940886699507,
      "grad_norm": 0.7185976163308054,
      "learning_rate": 6.901650961681976e-06,
      "loss": 0.8711,
      "step": 1040
    },
    {
      "epoch": 0.6410098522167488,
      "grad_norm": 0.7619837240226445,
      "learning_rate": 6.881213266800989e-06,
      "loss": 0.8693,
      "step": 1041
    },
    {
      "epoch": 0.6416256157635468,
      "grad_norm": 0.7049835078619207,
      "learning_rate": 6.860789992569601e-06,
      "loss": 0.9175,
      "step": 1042
    },
    {
      "epoch": 0.6422413793103449,
      "grad_norm": 0.7241732407965847,
      "learning_rate": 6.840381233420974e-06,
      "loss": 0.9022,
      "step": 1043
    },
    {
      "epoch": 0.6428571428571429,
      "grad_norm": 0.8082792824957932,
      "learning_rate": 6.819987083721149e-06,
      "loss": 0.8647,
      "step": 1044
    },
    {
      "epoch": 0.6434729064039408,
      "grad_norm": 0.6930734357056676,
      "learning_rate": 6.799607637768621e-06,
      "loss": 0.9028,
      "step": 1045
    },
    {
      "epoch": 0.6440886699507389,
      "grad_norm": 0.6696406912423044,
      "learning_rate": 6.779242989793893e-06,
      "loss": 0.9151,
      "step": 1046
    },
    {
      "epoch": 0.6447044334975369,
      "grad_norm": 0.7279160070259054,
      "learning_rate": 6.758893233959044e-06,
      "loss": 0.9625,
      "step": 1047
    },
    {
      "epoch": 0.645320197044335,
      "grad_norm": 0.7486147535235085,
      "learning_rate": 6.738558464357305e-06,
      "loss": 0.9269,
      "step": 1048
    },
    {
      "epoch": 0.645935960591133,
      "grad_norm": 0.6807206687890379,
      "learning_rate": 6.718238775012602e-06,
      "loss": 0.8511,
      "step": 1049
    },
    {
      "epoch": 0.646551724137931,
      "grad_norm": 0.7369388177653815,
      "learning_rate": 6.6979342598791395e-06,
      "loss": 0.8983,
      "step": 1050
    },
    {
      "epoch": 0.6471674876847291,
      "grad_norm": 0.7236067613688747,
      "learning_rate": 6.677645012840957e-06,
      "loss": 0.92,
      "step": 1051
    },
    {
      "epoch": 0.6477832512315271,
      "grad_norm": 0.6967572914790638,
      "learning_rate": 6.657371127711496e-06,
      "loss": 0.8985,
      "step": 1052
    },
    {
      "epoch": 0.6483990147783252,
      "grad_norm": 0.7289813792072826,
      "learning_rate": 6.63711269823317e-06,
      "loss": 0.8568,
      "step": 1053
    },
    {
      "epoch": 0.6490147783251231,
      "grad_norm": 0.7159803329095398,
      "learning_rate": 6.61686981807693e-06,
      "loss": 0.9394,
      "step": 1054
    },
    {
      "epoch": 0.6496305418719212,
      "grad_norm": 0.7653260760550493,
      "learning_rate": 6.596642580841827e-06,
      "loss": 0.8886,
      "step": 1055
    },
    {
      "epoch": 0.6502463054187192,
      "grad_norm": 0.7327112985445179,
      "learning_rate": 6.5764310800545815e-06,
      "loss": 0.9128,
      "step": 1056
    },
    {
      "epoch": 0.6508620689655172,
      "grad_norm": 0.735283286745817,
      "learning_rate": 6.556235409169154e-06,
      "loss": 0.9341,
      "step": 1057
    },
    {
      "epoch": 0.6514778325123153,
      "grad_norm": 0.7028369246241214,
      "learning_rate": 6.536055661566312e-06,
      "loss": 0.9216,
      "step": 1058
    },
    {
      "epoch": 0.6520935960591133,
      "grad_norm": 0.7059368127020033,
      "learning_rate": 6.515891930553195e-06,
      "loss": 0.8986,
      "step": 1059
    },
    {
      "epoch": 0.6527093596059114,
      "grad_norm": 0.6685419981100885,
      "learning_rate": 6.495744309362879e-06,
      "loss": 0.8877,
      "step": 1060
    },
    {
      "epoch": 0.6533251231527094,
      "grad_norm": 0.692945482931656,
      "learning_rate": 6.475612891153968e-06,
      "loss": 0.8984,
      "step": 1061
    },
    {
      "epoch": 0.6539408866995073,
      "grad_norm": 0.7203315145409438,
      "learning_rate": 6.455497769010131e-06,
      "loss": 0.9107,
      "step": 1062
    },
    {
      "epoch": 0.6545566502463054,
      "grad_norm": 0.7330519641347962,
      "learning_rate": 6.435399035939691e-06,
      "loss": 0.897,
      "step": 1063
    },
    {
      "epoch": 0.6551724137931034,
      "grad_norm": 0.8166539751884623,
      "learning_rate": 6.415316784875193e-06,
      "loss": 0.955,
      "step": 1064
    },
    {
      "epoch": 0.6557881773399015,
      "grad_norm": 0.7854759336100344,
      "learning_rate": 6.395251108672975e-06,
      "loss": 0.9026,
      "step": 1065
    },
    {
      "epoch": 0.6564039408866995,
      "grad_norm": 0.7070047922082041,
      "learning_rate": 6.37520210011273e-06,
      "loss": 0.9357,
      "step": 1066
    },
    {
      "epoch": 0.6570197044334976,
      "grad_norm": 0.6920386833572916,
      "learning_rate": 6.355169851897088e-06,
      "loss": 0.8781,
      "step": 1067
    },
    {
      "epoch": 0.6576354679802956,
      "grad_norm": 0.7768278458758424,
      "learning_rate": 6.335154456651178e-06,
      "loss": 0.8841,
      "step": 1068
    },
    {
      "epoch": 0.6582512315270936,
      "grad_norm": 0.736529077623641,
      "learning_rate": 6.315156006922211e-06,
      "loss": 0.8882,
      "step": 1069
    },
    {
      "epoch": 0.6588669950738916,
      "grad_norm": 0.757743515865678,
      "learning_rate": 6.295174595179041e-06,
      "loss": 0.9645,
      "step": 1070
    },
    {
      "epoch": 0.6594827586206896,
      "grad_norm": 0.7531804075740277,
      "learning_rate": 6.275210313811739e-06,
      "loss": 0.9117,
      "step": 1071
    },
    {
      "epoch": 0.6600985221674877,
      "grad_norm": 0.7093148408878822,
      "learning_rate": 6.255263255131172e-06,
      "loss": 0.8788,
      "step": 1072
    },
    {
      "epoch": 0.6607142857142857,
      "grad_norm": 0.7075222834833991,
      "learning_rate": 6.235333511368573e-06,
      "loss": 0.9123,
      "step": 1073
    },
    {
      "epoch": 0.6613300492610837,
      "grad_norm": 0.7342363961726721,
      "learning_rate": 6.215421174675114e-06,
      "loss": 0.933,
      "step": 1074
    },
    {
      "epoch": 0.6619458128078818,
      "grad_norm": 0.7308850961732138,
      "learning_rate": 6.195526337121483e-06,
      "loss": 0.9128,
      "step": 1075
    },
    {
      "epoch": 0.6625615763546798,
      "grad_norm": 0.760706735929803,
      "learning_rate": 6.1756490906974476e-06,
      "loss": 0.8871,
      "step": 1076
    },
    {
      "epoch": 0.6631773399014779,
      "grad_norm": 0.7897330459399138,
      "learning_rate": 6.155789527311444e-06,
      "loss": 0.9453,
      "step": 1077
    },
    {
      "epoch": 0.6637931034482759,
      "grad_norm": 0.75454419115083,
      "learning_rate": 6.135947738790145e-06,
      "loss": 0.9031,
      "step": 1078
    },
    {
      "epoch": 0.6644088669950738,
      "grad_norm": 0.7022688742424034,
      "learning_rate": 6.116123816878029e-06,
      "loss": 0.86,
      "step": 1079
    },
    {
      "epoch": 0.6650246305418719,
      "grad_norm": 0.7575331856996939,
      "learning_rate": 6.096317853236975e-06,
      "loss": 0.9359,
      "step": 1080
    },
    {
      "epoch": 0.6656403940886699,
      "grad_norm": 0.7932549758878508,
      "learning_rate": 6.0765299394458185e-06,
      "loss": 0.9009,
      "step": 1081
    },
    {
      "epoch": 0.666256157635468,
      "grad_norm": 0.6963101524993518,
      "learning_rate": 6.056760166999935e-06,
      "loss": 0.8696,
      "step": 1082
    },
    {
      "epoch": 0.666871921182266,
      "grad_norm": 0.7010640056450823,
      "learning_rate": 6.0370086273108205e-06,
      "loss": 0.8881,
      "step": 1083
    },
    {
      "epoch": 0.6674876847290641,
      "grad_norm": 0.6764799819916056,
      "learning_rate": 6.017275411705671e-06,
      "loss": 0.8943,
      "step": 1084
    },
    {
      "epoch": 0.6681034482758621,
      "grad_norm": 0.7083258760659875,
      "learning_rate": 5.997560611426947e-06,
      "loss": 0.8933,
      "step": 1085
    },
    {
      "epoch": 0.6687192118226601,
      "grad_norm": 0.7173990603121654,
      "learning_rate": 5.977864317631965e-06,
      "loss": 0.8934,
      "step": 1086
    },
    {
      "epoch": 0.6693349753694581,
      "grad_norm": 0.6941901037614218,
      "learning_rate": 5.9581866213924656e-06,
      "loss": 0.8989,
      "step": 1087
    },
    {
      "epoch": 0.6699507389162561,
      "grad_norm": 0.7113189481651018,
      "learning_rate": 5.938527613694214e-06,
      "loss": 0.9193,
      "step": 1088
    },
    {
      "epoch": 0.6705665024630542,
      "grad_norm": 0.6887362289870574,
      "learning_rate": 5.918887385436542e-06,
      "loss": 0.914,
      "step": 1089
    },
    {
      "epoch": 0.6711822660098522,
      "grad_norm": 0.6750588800805896,
      "learning_rate": 5.899266027431965e-06,
      "loss": 0.8969,
      "step": 1090
    },
    {
      "epoch": 0.6717980295566502,
      "grad_norm": 0.6517070127117931,
      "learning_rate": 5.879663630405737e-06,
      "loss": 0.8341,
      "step": 1091
    },
    {
      "epoch": 0.6724137931034483,
      "grad_norm": 0.6933178996461266,
      "learning_rate": 5.860080284995445e-06,
      "loss": 0.8895,
      "step": 1092
    },
    {
      "epoch": 0.6730295566502463,
      "grad_norm": 0.7368896707448154,
      "learning_rate": 5.8405160817505834e-06,
      "loss": 0.9422,
      "step": 1093
    },
    {
      "epoch": 0.6736453201970444,
      "grad_norm": 0.6861746459299171,
      "learning_rate": 5.820971111132137e-06,
      "loss": 0.8709,
      "step": 1094
    },
    {
      "epoch": 0.6742610837438424,
      "grad_norm": 0.6561959889319966,
      "learning_rate": 5.80144546351216e-06,
      "loss": 0.861,
      "step": 1095
    },
    {
      "epoch": 0.6748768472906403,
      "grad_norm": 0.7272105862793617,
      "learning_rate": 5.781939229173365e-06,
      "loss": 0.9215,
      "step": 1096
    },
    {
      "epoch": 0.6754926108374384,
      "grad_norm": 0.7272158836826056,
      "learning_rate": 5.7624524983087085e-06,
      "loss": 0.8861,
      "step": 1097
    },
    {
      "epoch": 0.6761083743842364,
      "grad_norm": 0.702461786137382,
      "learning_rate": 5.742985361020945e-06,
      "loss": 0.8752,
      "step": 1098
    },
    {
      "epoch": 0.6767241379310345,
      "grad_norm": 0.6837968977325968,
      "learning_rate": 5.723537907322259e-06,
      "loss": 0.8757,
      "step": 1099
    },
    {
      "epoch": 0.6773399014778325,
      "grad_norm": 0.7221776757845857,
      "learning_rate": 5.704110227133792e-06,
      "loss": 0.9082,
      "step": 1100
    },
    {
      "epoch": 0.6779556650246306,
      "grad_norm": 0.6604643140383945,
      "learning_rate": 5.684702410285292e-06,
      "loss": 0.8867,
      "step": 1101
    },
    {
      "epoch": 0.6785714285714286,
      "grad_norm": 0.6662328357944223,
      "learning_rate": 5.665314546514633e-06,
      "loss": 0.8823,
      "step": 1102
    },
    {
      "epoch": 0.6791871921182266,
      "grad_norm": 0.7370332676476488,
      "learning_rate": 5.645946725467444e-06,
      "loss": 0.9583,
      "step": 1103
    },
    {
      "epoch": 0.6798029556650246,
      "grad_norm": 0.691972382172292,
      "learning_rate": 5.62659903669668e-06,
      "loss": 0.9079,
      "step": 1104
    },
    {
      "epoch": 0.6804187192118226,
      "grad_norm": 0.71127498333362,
      "learning_rate": 5.607271569662203e-06,
      "loss": 0.9397,
      "step": 1105
    },
    {
      "epoch": 0.6810344827586207,
      "grad_norm": 0.7036095549670983,
      "learning_rate": 5.5879644137303805e-06,
      "loss": 0.8905,
      "step": 1106
    },
    {
      "epoch": 0.6816502463054187,
      "grad_norm": 0.7134693907206262,
      "learning_rate": 5.568677658173656e-06,
      "loss": 0.9265,
      "step": 1107
    },
    {
      "epoch": 0.6822660098522167,
      "grad_norm": 0.7301449551203517,
      "learning_rate": 5.549411392170154e-06,
      "loss": 0.8555,
      "step": 1108
    },
    {
      "epoch": 0.6828817733990148,
      "grad_norm": 0.789429912768648,
      "learning_rate": 5.530165704803261e-06,
      "loss": 0.9128,
      "step": 1109
    },
    {
      "epoch": 0.6834975369458128,
      "grad_norm": 0.6661566204651262,
      "learning_rate": 5.510940685061202e-06,
      "loss": 0.8585,
      "step": 1110
    },
    {
      "epoch": 0.6841133004926109,
      "grad_norm": 0.6970017845200686,
      "learning_rate": 5.491736421836647e-06,
      "loss": 0.879,
      "step": 1111
    },
    {
      "epoch": 0.6847290640394089,
      "grad_norm": 0.6635233449206236,
      "learning_rate": 5.472553003926293e-06,
      "loss": 0.8838,
      "step": 1112
    },
    {
      "epoch": 0.6853448275862069,
      "grad_norm": 0.6838529385150229,
      "learning_rate": 5.4533905200304395e-06,
      "loss": 0.887,
      "step": 1113
    },
    {
      "epoch": 0.6859605911330049,
      "grad_norm": 0.6742272263221694,
      "learning_rate": 5.434249058752615e-06,
      "loss": 0.9026,
      "step": 1114
    },
    {
      "epoch": 0.6865763546798029,
      "grad_norm": 0.7355508834817683,
      "learning_rate": 5.41512870859912e-06,
      "loss": 0.8607,
      "step": 1115
    },
    {
      "epoch": 0.687192118226601,
      "grad_norm": 0.7308847801416948,
      "learning_rate": 5.396029557978657e-06,
      "loss": 0.8928,
      "step": 1116
    },
    {
      "epoch": 0.687807881773399,
      "grad_norm": 0.6991166439799994,
      "learning_rate": 5.376951695201894e-06,
      "loss": 0.9009,
      "step": 1117
    },
    {
      "epoch": 0.6884236453201971,
      "grad_norm": 0.7170088732226073,
      "learning_rate": 5.357895208481076e-06,
      "loss": 0.8518,
      "step": 1118
    },
    {
      "epoch": 0.6890394088669951,
      "grad_norm": 0.7075792216704179,
      "learning_rate": 5.338860185929614e-06,
      "loss": 0.8601,
      "step": 1119
    },
    {
      "epoch": 0.6896551724137931,
      "grad_norm": 0.691429674102568,
      "learning_rate": 5.319846715561656e-06,
      "loss": 0.9249,
      "step": 1120
    },
    {
      "epoch": 0.6902709359605911,
      "grad_norm": 0.6774415461383786,
      "learning_rate": 5.30085488529172e-06,
      "loss": 0.9224,
      "step": 1121
    },
    {
      "epoch": 0.6908866995073891,
      "grad_norm": 0.6306919964634131,
      "learning_rate": 5.281884782934239e-06,
      "loss": 0.8289,
      "step": 1122
    },
    {
      "epoch": 0.6915024630541872,
      "grad_norm": 0.686166220476993,
      "learning_rate": 5.2629364962032e-06,
      "loss": 0.8806,
      "step": 1123
    },
    {
      "epoch": 0.6921182266009852,
      "grad_norm": 0.7069555127702706,
      "learning_rate": 5.244010112711717e-06,
      "loss": 0.9256,
      "step": 1124
    },
    {
      "epoch": 0.6927339901477833,
      "grad_norm": 0.690250285208387,
      "learning_rate": 5.225105719971615e-06,
      "loss": 0.8937,
      "step": 1125
    },
    {
      "epoch": 0.6933497536945813,
      "grad_norm": 0.650338536773774,
      "learning_rate": 5.206223405393051e-06,
      "loss": 0.9191,
      "step": 1126
    },
    {
      "epoch": 0.6939655172413793,
      "grad_norm": 0.7128659558364466,
      "learning_rate": 5.18736325628409e-06,
      "loss": 0.8805,
      "step": 1127
    },
    {
      "epoch": 0.6945812807881774,
      "grad_norm": 0.7577327494269358,
      "learning_rate": 5.168525359850312e-06,
      "loss": 0.9361,
      "step": 1128
    },
    {
      "epoch": 0.6951970443349754,
      "grad_norm": 0.7443917428668443,
      "learning_rate": 5.149709803194409e-06,
      "loss": 0.9046,
      "step": 1129
    },
    {
      "epoch": 0.6958128078817734,
      "grad_norm": 0.6911657482487051,
      "learning_rate": 5.130916673315762e-06,
      "loss": 0.8893,
      "step": 1130
    },
    {
      "epoch": 0.6964285714285714,
      "grad_norm": 0.7420299108317902,
      "learning_rate": 5.112146057110078e-06,
      "loss": 0.9544,
      "step": 1131
    },
    {
      "epoch": 0.6970443349753694,
      "grad_norm": 0.7532947094570263,
      "learning_rate": 5.093398041368942e-06,
      "loss": 0.8957,
      "step": 1132
    },
    {
      "epoch": 0.6976600985221675,
      "grad_norm": 0.6909150453724049,
      "learning_rate": 5.074672712779456e-06,
      "loss": 0.8864,
      "step": 1133
    },
    {
      "epoch": 0.6982758620689655,
      "grad_norm": 0.7108792926248048,
      "learning_rate": 5.055970157923818e-06,
      "loss": 0.8898,
      "step": 1134
    },
    {
      "epoch": 0.6988916256157636,
      "grad_norm": 0.7323086265043367,
      "learning_rate": 5.037290463278914e-06,
      "loss": 0.9028,
      "step": 1135
    },
    {
      "epoch": 0.6995073891625616,
      "grad_norm": 0.7272691206595147,
      "learning_rate": 5.018633715215948e-06,
      "loss": 0.9135,
      "step": 1136
    },
    {
      "epoch": 0.7001231527093597,
      "grad_norm": 0.7408524463657474,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.8876,
      "step": 1137
    },
    {
      "epoch": 0.7007389162561576,
      "grad_norm": 0.7533029224996899,
      "learning_rate": 4.9813894037896745e-06,
      "loss": 0.9073,
      "step": 1138
    },
    {
      "epoch": 0.7013546798029556,
      "grad_norm": 0.7028843289865291,
      "learning_rate": 4.962802012636664e-06,
      "loss": 0.9301,
      "step": 1139
    },
    {
      "epoch": 0.7019704433497537,
      "grad_norm": 0.7351315375065111,
      "learning_rate": 4.94423791248536e-06,
      "loss": 0.9147,
      "step": 1140
    },
    {
      "epoch": 0.7025862068965517,
      "grad_norm": 0.7172822165058504,
      "learning_rate": 4.925697189172485e-06,
      "loss": 0.8854,
      "step": 1141
    },
    {
      "epoch": 0.7032019704433498,
      "grad_norm": 0.7308483490747527,
      "learning_rate": 4.907179928426643e-06,
      "loss": 0.8813,
      "step": 1142
    },
    {
      "epoch": 0.7038177339901478,
      "grad_norm": 0.710804702366717,
      "learning_rate": 4.8886862158679716e-06,
      "loss": 0.8827,
      "step": 1143
    },
    {
      "epoch": 0.7044334975369458,
      "grad_norm": 0.7517885983645317,
      "learning_rate": 4.8702161370077215e-06,
      "loss": 0.9385,
      "step": 1144
    },
    {
      "epoch": 0.7050492610837439,
      "grad_norm": 0.6993761567919461,
      "learning_rate": 4.851769777247857e-06,
      "loss": 0.9061,
      "step": 1145
    },
    {
      "epoch": 0.7056650246305419,
      "grad_norm": 0.7428001822845043,
      "learning_rate": 4.833347221880686e-06,
      "loss": 0.8515,
      "step": 1146
    },
    {
      "epoch": 0.7062807881773399,
      "grad_norm": 0.7541052061241705,
      "learning_rate": 4.81494855608843e-06,
      "loss": 0.8839,
      "step": 1147
    },
    {
      "epoch": 0.7068965517241379,
      "grad_norm": 0.7549349737123539,
      "learning_rate": 4.796573864942868e-06,
      "loss": 0.9278,
      "step": 1148
    },
    {
      "epoch": 0.7075123152709359,
      "grad_norm": 0.7396361528050266,
      "learning_rate": 4.778223233404919e-06,
      "loss": 0.8995,
      "step": 1149
    },
    {
      "epoch": 0.708128078817734,
      "grad_norm": 0.7121619793668674,
      "learning_rate": 4.759896746324247e-06,
      "loss": 0.9132,
      "step": 1150
    },
    {
      "epoch": 0.708743842364532,
      "grad_norm": 0.7168507683951918,
      "learning_rate": 4.741594488438891e-06,
      "loss": 0.8811,
      "step": 1151
    },
    {
      "epoch": 0.7093596059113301,
      "grad_norm": 0.6935727448013648,
      "learning_rate": 4.723316544374844e-06,
      "loss": 0.9297,
      "step": 1152
    },
    {
      "epoch": 0.7099753694581281,
      "grad_norm": 0.7027157109735622,
      "learning_rate": 4.705062998645687e-06,
      "loss": 0.8732,
      "step": 1153
    },
    {
      "epoch": 0.7105911330049262,
      "grad_norm": 0.7017545358996017,
      "learning_rate": 4.686833935652187e-06,
      "loss": 0.9042,
      "step": 1154
    },
    {
      "epoch": 0.7112068965517241,
      "grad_norm": 0.7159832920676509,
      "learning_rate": 4.668629439681907e-06,
      "loss": 0.9472,
      "step": 1155
    },
    {
      "epoch": 0.7118226600985221,
      "grad_norm": 0.7021212769316445,
      "learning_rate": 4.650449594908816e-06,
      "loss": 0.8993,
      "step": 1156
    },
    {
      "epoch": 0.7124384236453202,
      "grad_norm": 0.7330627706461624,
      "learning_rate": 4.632294485392895e-06,
      "loss": 0.8808,
      "step": 1157
    },
    {
      "epoch": 0.7130541871921182,
      "grad_norm": 0.6197392246119906,
      "learning_rate": 4.614164195079765e-06,
      "loss": 0.8962,
      "step": 1158
    },
    {
      "epoch": 0.7136699507389163,
      "grad_norm": 0.6889361567439508,
      "learning_rate": 4.596058807800284e-06,
      "loss": 0.8952,
      "step": 1159
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.7215864654618344,
      "learning_rate": 4.577978407270156e-06,
      "loss": 0.9164,
      "step": 1160
    },
    {
      "epoch": 0.7149014778325123,
      "grad_norm": 0.760768811037527,
      "learning_rate": 4.559923077089564e-06,
      "loss": 0.956,
      "step": 1161
    },
    {
      "epoch": 0.7155172413793104,
      "grad_norm": 0.6825990754478561,
      "learning_rate": 4.541892900742757e-06,
      "loss": 0.8481,
      "step": 1162
    },
    {
      "epoch": 0.7161330049261084,
      "grad_norm": 0.7259504844924438,
      "learning_rate": 4.523887961597688e-06,
      "loss": 0.8778,
      "step": 1163
    },
    {
      "epoch": 0.7167487684729064,
      "grad_norm": 0.7034864096309602,
      "learning_rate": 4.505908342905621e-06,
      "loss": 0.8888,
      "step": 1164
    },
    {
      "epoch": 0.7173645320197044,
      "grad_norm": 0.7409145534041075,
      "learning_rate": 4.487954127800726e-06,
      "loss": 0.9153,
      "step": 1165
    },
    {
      "epoch": 0.7179802955665024,
      "grad_norm": 0.702018285358734,
      "learning_rate": 4.470025399299728e-06,
      "loss": 0.9206,
      "step": 1166
    },
    {
      "epoch": 0.7185960591133005,
      "grad_norm": 0.6613616108051624,
      "learning_rate": 4.452122240301502e-06,
      "loss": 0.8775,
      "step": 1167
    },
    {
      "epoch": 0.7192118226600985,
      "grad_norm": 0.7285310952951727,
      "learning_rate": 4.434244733586699e-06,
      "loss": 0.9435,
      "step": 1168
    },
    {
      "epoch": 0.7198275862068966,
      "grad_norm": 0.6940407917304187,
      "learning_rate": 4.416392961817347e-06,
      "loss": 0.9337,
      "step": 1169
    },
    {
      "epoch": 0.7204433497536946,
      "grad_norm": 0.747561617210193,
      "learning_rate": 4.398567007536493e-06,
      "loss": 0.8469,
      "step": 1170
    },
    {
      "epoch": 0.7210591133004927,
      "grad_norm": 0.6801012636742867,
      "learning_rate": 4.380766953167805e-06,
      "loss": 0.8621,
      "step": 1171
    },
    {
      "epoch": 0.7216748768472906,
      "grad_norm": 0.7031026654530009,
      "learning_rate": 4.36299288101519e-06,
      "loss": 0.9158,
      "step": 1172
    },
    {
      "epoch": 0.7222906403940886,
      "grad_norm": 0.7044838536109875,
      "learning_rate": 4.345244873262426e-06,
      "loss": 0.8718,
      "step": 1173
    },
    {
      "epoch": 0.7229064039408867,
      "grad_norm": 0.7548819146899708,
      "learning_rate": 4.327523011972774e-06,
      "loss": 0.8901,
      "step": 1174
    },
    {
      "epoch": 0.7235221674876847,
      "grad_norm": 0.678556950992195,
      "learning_rate": 4.309827379088589e-06,
      "loss": 0.8967,
      "step": 1175
    },
    {
      "epoch": 0.7241379310344828,
      "grad_norm": 0.7298694618133801,
      "learning_rate": 4.292158056430966e-06,
      "loss": 0.9126,
      "step": 1176
    },
    {
      "epoch": 0.7247536945812808,
      "grad_norm": 0.6947824903298965,
      "learning_rate": 4.2745151256993325e-06,
      "loss": 0.9134,
      "step": 1177
    },
    {
      "epoch": 0.7253694581280788,
      "grad_norm": 0.6803533556314053,
      "learning_rate": 4.256898668471092e-06,
      "loss": 0.846,
      "step": 1178
    },
    {
      "epoch": 0.7259852216748769,
      "grad_norm": 0.751649089217124,
      "learning_rate": 4.239308766201246e-06,
      "loss": 0.8955,
      "step": 1179
    },
    {
      "epoch": 0.7266009852216748,
      "grad_norm": 0.7087913948458853,
      "learning_rate": 4.221745500221992e-06,
      "loss": 0.8867,
      "step": 1180
    },
    {
      "epoch": 0.7272167487684729,
      "grad_norm": 0.7054724290766984,
      "learning_rate": 4.204208951742393e-06,
      "loss": 0.8921,
      "step": 1181
    },
    {
      "epoch": 0.7278325123152709,
      "grad_norm": 0.6910401524033906,
      "learning_rate": 4.186699201847954e-06,
      "loss": 0.8614,
      "step": 1182
    },
    {
      "epoch": 0.728448275862069,
      "grad_norm": 0.7360498266835115,
      "learning_rate": 4.169216331500279e-06,
      "loss": 0.9178,
      "step": 1183
    },
    {
      "epoch": 0.729064039408867,
      "grad_norm": 0.7030079852514759,
      "learning_rate": 4.151760421536678e-06,
      "loss": 0.8485,
      "step": 1184
    },
    {
      "epoch": 0.729679802955665,
      "grad_norm": 0.7873932523125778,
      "learning_rate": 4.134331552669812e-06,
      "loss": 0.8766,
      "step": 1185
    },
    {
      "epoch": 0.7302955665024631,
      "grad_norm": 0.6486202837777103,
      "learning_rate": 4.116929805487307e-06,
      "loss": 0.8753,
      "step": 1186
    },
    {
      "epoch": 0.7309113300492611,
      "grad_norm": 0.6946338088414719,
      "learning_rate": 4.099555260451373e-06,
      "loss": 0.8675,
      "step": 1187
    },
    {
      "epoch": 0.7315270935960592,
      "grad_norm": 0.706810909550085,
      "learning_rate": 4.082207997898458e-06,
      "loss": 0.8846,
      "step": 1188
    },
    {
      "epoch": 0.7321428571428571,
      "grad_norm": 0.6817543156499241,
      "learning_rate": 4.064888098038853e-06,
      "loss": 0.9066,
      "step": 1189
    },
    {
      "epoch": 0.7327586206896551,
      "grad_norm": 0.7573080005369632,
      "learning_rate": 4.047595640956326e-06,
      "loss": 0.9042,
      "step": 1190
    },
    {
      "epoch": 0.7333743842364532,
      "grad_norm": 0.7401167080857884,
      "learning_rate": 4.030330706607766e-06,
      "loss": 0.9008,
      "step": 1191
    },
    {
      "epoch": 0.7339901477832512,
      "grad_norm": 0.6944702353457608,
      "learning_rate": 4.013093374822789e-06,
      "loss": 0.8836,
      "step": 1192
    },
    {
      "epoch": 0.7346059113300493,
      "grad_norm": 0.6507739963925147,
      "learning_rate": 3.995883725303392e-06,
      "loss": 0.87,
      "step": 1193
    },
    {
      "epoch": 0.7352216748768473,
      "grad_norm": 0.6954253220979149,
      "learning_rate": 3.978701837623573e-06,
      "loss": 0.8953,
      "step": 1194
    },
    {
      "epoch": 0.7358374384236454,
      "grad_norm": 0.7003801085848937,
      "learning_rate": 3.961547791228963e-06,
      "loss": 0.8588,
      "step": 1195
    },
    {
      "epoch": 0.7364532019704434,
      "grad_norm": 0.7157685427077439,
      "learning_rate": 3.944421665436462e-06,
      "loss": 0.8553,
      "step": 1196
    },
    {
      "epoch": 0.7370689655172413,
      "grad_norm": 0.7503442845156104,
      "learning_rate": 3.927323539433867e-06,
      "loss": 0.8846,
      "step": 1197
    },
    {
      "epoch": 0.7376847290640394,
      "grad_norm": 0.7633209098067081,
      "learning_rate": 3.9102534922795166e-06,
      "loss": 0.9056,
      "step": 1198
    },
    {
      "epoch": 0.7383004926108374,
      "grad_norm": 0.7345409080446796,
      "learning_rate": 3.893211602901908e-06,
      "loss": 0.97,
      "step": 1199
    },
    {
      "epoch": 0.7389162561576355,
      "grad_norm": 0.7062568025599855,
      "learning_rate": 3.876197950099351e-06,
      "loss": 0.8783,
      "step": 1200
    },
    {
      "epoch": 0.7389162561576355,
      "eval_loss": 0.896390438079834,
      "eval_runtime": 522.2918,
      "eval_samples_per_second": 44.247,
      "eval_steps_per_second": 0.693,
      "step": 1200
    },
    {
      "epoch": 0.7395320197044335,
      "grad_norm": 0.7464837475252311,
      "learning_rate": 3.859212612539597e-06,
      "loss": 0.8747,
      "step": 1201
    },
    {
      "epoch": 0.7401477832512315,
      "grad_norm": 0.7009583955839648,
      "learning_rate": 3.842255668759462e-06,
      "loss": 0.937,
      "step": 1202
    },
    {
      "epoch": 0.7407635467980296,
      "grad_norm": 0.7244823893076702,
      "learning_rate": 3.8253271971644835e-06,
      "loss": 0.859,
      "step": 1203
    },
    {
      "epoch": 0.7413793103448276,
      "grad_norm": 0.7087461358242156,
      "learning_rate": 3.8084272760285523e-06,
      "loss": 0.9263,
      "step": 1204
    },
    {
      "epoch": 0.7419950738916257,
      "grad_norm": 0.7119062978634647,
      "learning_rate": 3.7915559834935355e-06,
      "loss": 0.864,
      "step": 1205
    },
    {
      "epoch": 0.7426108374384236,
      "grad_norm": 0.7177163219660962,
      "learning_rate": 3.7747133975689387e-06,
      "loss": 0.8938,
      "step": 1206
    },
    {
      "epoch": 0.7432266009852216,
      "grad_norm": 0.7217275689965462,
      "learning_rate": 3.757899596131529e-06,
      "loss": 0.8643,
      "step": 1207
    },
    {
      "epoch": 0.7438423645320197,
      "grad_norm": 0.7063620316514662,
      "learning_rate": 3.7411146569249836e-06,
      "loss": 0.8452,
      "step": 1208
    },
    {
      "epoch": 0.7444581280788177,
      "grad_norm": 0.7122128483763945,
      "learning_rate": 3.724358657559517e-06,
      "loss": 0.8802,
      "step": 1209
    },
    {
      "epoch": 0.7450738916256158,
      "grad_norm": 0.7618299656273183,
      "learning_rate": 3.7076316755115407e-06,
      "loss": 0.9277,
      "step": 1210
    },
    {
      "epoch": 0.7456896551724138,
      "grad_norm": 0.6994731836145274,
      "learning_rate": 3.6909337881232965e-06,
      "loss": 0.9218,
      "step": 1211
    },
    {
      "epoch": 0.7463054187192119,
      "grad_norm": 0.6674984746427913,
      "learning_rate": 3.6742650726024887e-06,
      "loss": 0.8131,
      "step": 1212
    },
    {
      "epoch": 0.7469211822660099,
      "grad_norm": 0.7127871816131648,
      "learning_rate": 3.6576256060219486e-06,
      "loss": 0.8522,
      "step": 1213
    },
    {
      "epoch": 0.7475369458128078,
      "grad_norm": 0.7296385455082367,
      "learning_rate": 3.6410154653192544e-06,
      "loss": 0.9387,
      "step": 1214
    },
    {
      "epoch": 0.7481527093596059,
      "grad_norm": 0.7013255042890396,
      "learning_rate": 3.6244347272963974e-06,
      "loss": 0.853,
      "step": 1215
    },
    {
      "epoch": 0.7487684729064039,
      "grad_norm": 0.687992580181443,
      "learning_rate": 3.607883468619414e-06,
      "loss": 0.8785,
      "step": 1216
    },
    {
      "epoch": 0.749384236453202,
      "grad_norm": 0.7197834541205442,
      "learning_rate": 3.591361765818029e-06,
      "loss": 0.9263,
      "step": 1217
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6971162171099686,
      "learning_rate": 3.5748696952853147e-06,
      "loss": 0.8644,
      "step": 1218
    },
    {
      "epoch": 0.750615763546798,
      "grad_norm": 0.681624330179704,
      "learning_rate": 3.558407333277315e-06,
      "loss": 0.8732,
      "step": 1219
    },
    {
      "epoch": 0.7512315270935961,
      "grad_norm": 0.6928733464146026,
      "learning_rate": 3.5419747559127294e-06,
      "loss": 0.9039,
      "step": 1220
    },
    {
      "epoch": 0.7518472906403941,
      "grad_norm": 0.6932073674827232,
      "learning_rate": 3.525572039172528e-06,
      "loss": 0.8528,
      "step": 1221
    },
    {
      "epoch": 0.7524630541871922,
      "grad_norm": 0.7148233747926745,
      "learning_rate": 3.5091992588996026e-06,
      "loss": 0.9072,
      "step": 1222
    },
    {
      "epoch": 0.7530788177339901,
      "grad_norm": 0.7053856426644688,
      "learning_rate": 3.4928564907984395e-06,
      "loss": 0.8794,
      "step": 1223
    },
    {
      "epoch": 0.7536945812807881,
      "grad_norm": 0.6850033250468565,
      "learning_rate": 3.476543810434744e-06,
      "loss": 0.8723,
      "step": 1224
    },
    {
      "epoch": 0.7543103448275862,
      "grad_norm": 0.7437383205410577,
      "learning_rate": 3.4602612932351065e-06,
      "loss": 0.8758,
      "step": 1225
    },
    {
      "epoch": 0.7549261083743842,
      "grad_norm": 0.6703635441826831,
      "learning_rate": 3.4440090144866533e-06,
      "loss": 0.8444,
      "step": 1226
    },
    {
      "epoch": 0.7555418719211823,
      "grad_norm": 0.7534721425412435,
      "learning_rate": 3.4277870493366816e-06,
      "loss": 0.9033,
      "step": 1227
    },
    {
      "epoch": 0.7561576354679803,
      "grad_norm": 0.7680180921115617,
      "learning_rate": 3.4115954727923394e-06,
      "loss": 0.9093,
      "step": 1228
    },
    {
      "epoch": 0.7567733990147784,
      "grad_norm": 0.7308869891853025,
      "learning_rate": 3.395434359720249e-06,
      "loss": 0.9511,
      "step": 1229
    },
    {
      "epoch": 0.7573891625615764,
      "grad_norm": 0.7042593700478477,
      "learning_rate": 3.3793037848461873e-06,
      "loss": 0.9011,
      "step": 1230
    },
    {
      "epoch": 0.7580049261083743,
      "grad_norm": 0.6560537624759148,
      "learning_rate": 3.3632038227547283e-06,
      "loss": 0.8963,
      "step": 1231
    },
    {
      "epoch": 0.7586206896551724,
      "grad_norm": 0.7551512515282361,
      "learning_rate": 3.3471345478888863e-06,
      "loss": 0.97,
      "step": 1232
    },
    {
      "epoch": 0.7592364532019704,
      "grad_norm": 0.7963625612938491,
      "learning_rate": 3.3310960345497976e-06,
      "loss": 0.95,
      "step": 1233
    },
    {
      "epoch": 0.7598522167487685,
      "grad_norm": 0.6928645038756635,
      "learning_rate": 3.3150883568963577e-06,
      "loss": 0.8814,
      "step": 1234
    },
    {
      "epoch": 0.7604679802955665,
      "grad_norm": 0.7398912822139294,
      "learning_rate": 3.2991115889448877e-06,
      "loss": 0.8922,
      "step": 1235
    },
    {
      "epoch": 0.7610837438423645,
      "grad_norm": 0.6784640829373988,
      "learning_rate": 3.2831658045687786e-06,
      "loss": 0.9481,
      "step": 1236
    },
    {
      "epoch": 0.7616995073891626,
      "grad_norm": 0.7399067008273665,
      "learning_rate": 3.2672510774981692e-06,
      "loss": 0.9125,
      "step": 1237
    },
    {
      "epoch": 0.7623152709359606,
      "grad_norm": 0.796355998359148,
      "learning_rate": 3.251367481319596e-06,
      "loss": 0.9144,
      "step": 1238
    },
    {
      "epoch": 0.7629310344827587,
      "grad_norm": 0.6898011716916692,
      "learning_rate": 3.2355150894756426e-06,
      "loss": 0.9062,
      "step": 1239
    },
    {
      "epoch": 0.7635467980295566,
      "grad_norm": 0.6723703055696021,
      "learning_rate": 3.2196939752646183e-06,
      "loss": 0.9128,
      "step": 1240
    },
    {
      "epoch": 0.7641625615763546,
      "grad_norm": 0.6924792649513672,
      "learning_rate": 3.203904211840213e-06,
      "loss": 0.942,
      "step": 1241
    },
    {
      "epoch": 0.7647783251231527,
      "grad_norm": 0.7379544482732634,
      "learning_rate": 3.188145872211147e-06,
      "loss": 0.9271,
      "step": 1242
    },
    {
      "epoch": 0.7653940886699507,
      "grad_norm": 0.6932002072845549,
      "learning_rate": 3.172419029240853e-06,
      "loss": 0.864,
      "step": 1243
    },
    {
      "epoch": 0.7660098522167488,
      "grad_norm": 0.7446840114851969,
      "learning_rate": 3.1567237556471197e-06,
      "loss": 0.8779,
      "step": 1244
    },
    {
      "epoch": 0.7666256157635468,
      "grad_norm": 0.7540193453221069,
      "learning_rate": 3.141060124001776e-06,
      "loss": 0.8923,
      "step": 1245
    },
    {
      "epoch": 0.7672413793103449,
      "grad_norm": 0.7209390754852734,
      "learning_rate": 3.1254282067303353e-06,
      "loss": 0.8838,
      "step": 1246
    },
    {
      "epoch": 0.7678571428571429,
      "grad_norm": 0.6669977841878069,
      "learning_rate": 3.1098280761116784e-06,
      "loss": 0.8813,
      "step": 1247
    },
    {
      "epoch": 0.7684729064039408,
      "grad_norm": 0.7030660757286702,
      "learning_rate": 3.0942598042777074e-06,
      "loss": 0.8875,
      "step": 1248
    },
    {
      "epoch": 0.7690886699507389,
      "grad_norm": 0.6520578055644021,
      "learning_rate": 3.07872346321301e-06,
      "loss": 0.8719,
      "step": 1249
    },
    {
      "epoch": 0.7697044334975369,
      "grad_norm": 0.6714050725194269,
      "learning_rate": 3.063219124754543e-06,
      "loss": 0.922,
      "step": 1250
    },
    {
      "epoch": 0.770320197044335,
      "grad_norm": 0.6923412946313824,
      "learning_rate": 3.0477468605912788e-06,
      "loss": 0.894,
      "step": 1251
    },
    {
      "epoch": 0.770935960591133,
      "grad_norm": 0.6801649746147456,
      "learning_rate": 3.032306742263891e-06,
      "loss": 0.9017,
      "step": 1252
    },
    {
      "epoch": 0.771551724137931,
      "grad_norm": 0.7130748387449891,
      "learning_rate": 3.0168988411644207e-06,
      "loss": 0.9047,
      "step": 1253
    },
    {
      "epoch": 0.7721674876847291,
      "grad_norm": 0.7384413832954502,
      "learning_rate": 3.001523228535932e-06,
      "loss": 0.9236,
      "step": 1254
    },
    {
      "epoch": 0.7727832512315271,
      "grad_norm": 0.7290448012401478,
      "learning_rate": 2.9861799754722033e-06,
      "loss": 0.8395,
      "step": 1255
    },
    {
      "epoch": 0.7733990147783252,
      "grad_norm": 0.7175730057044413,
      "learning_rate": 2.970869152917387e-06,
      "loss": 0.8382,
      "step": 1256
    },
    {
      "epoch": 0.7740147783251231,
      "grad_norm": 0.7098133641809817,
      "learning_rate": 2.955590831665677e-06,
      "loss": 0.8607,
      "step": 1257
    },
    {
      "epoch": 0.7746305418719212,
      "grad_norm": 0.6782673379034447,
      "learning_rate": 2.940345082360997e-06,
      "loss": 0.8552,
      "step": 1258
    },
    {
      "epoch": 0.7752463054187192,
      "grad_norm": 0.7423590499849376,
      "learning_rate": 2.9251319754966525e-06,
      "loss": 0.9408,
      "step": 1259
    },
    {
      "epoch": 0.7758620689655172,
      "grad_norm": 0.7149891274148373,
      "learning_rate": 2.9099515814150335e-06,
      "loss": 0.8508,
      "step": 1260
    },
    {
      "epoch": 0.7764778325123153,
      "grad_norm": 0.7389974848268119,
      "learning_rate": 2.8948039703072574e-06,
      "loss": 0.9461,
      "step": 1261
    },
    {
      "epoch": 0.7770935960591133,
      "grad_norm": 0.6638453507312181,
      "learning_rate": 2.879689212212866e-06,
      "loss": 0.8768,
      "step": 1262
    },
    {
      "epoch": 0.7777093596059114,
      "grad_norm": 0.6994415422179924,
      "learning_rate": 2.8646073770194983e-06,
      "loss": 0.9061,
      "step": 1263
    },
    {
      "epoch": 0.7783251231527094,
      "grad_norm": 0.7070391136292098,
      "learning_rate": 2.8495585344625566e-06,
      "loss": 0.8893,
      "step": 1264
    },
    {
      "epoch": 0.7789408866995073,
      "grad_norm": 0.7122601352832316,
      "learning_rate": 2.8345427541248993e-06,
      "loss": 0.8708,
      "step": 1265
    },
    {
      "epoch": 0.7795566502463054,
      "grad_norm": 0.704954331266899,
      "learning_rate": 2.819560105436504e-06,
      "loss": 0.8974,
      "step": 1266
    },
    {
      "epoch": 0.7801724137931034,
      "grad_norm": 0.6118800842561571,
      "learning_rate": 2.8046106576741605e-06,
      "loss": 0.8388,
      "step": 1267
    },
    {
      "epoch": 0.7807881773399015,
      "grad_norm": 0.6705591047582942,
      "learning_rate": 2.789694479961147e-06,
      "loss": 0.8696,
      "step": 1268
    },
    {
      "epoch": 0.7814039408866995,
      "grad_norm": 0.6725721669806741,
      "learning_rate": 2.7748116412668944e-06,
      "loss": 0.8815,
      "step": 1269
    },
    {
      "epoch": 0.7820197044334976,
      "grad_norm": 0.6915763816368014,
      "learning_rate": 2.7599622104066937e-06,
      "loss": 0.8794,
      "step": 1270
    },
    {
      "epoch": 0.7826354679802956,
      "grad_norm": 0.7148209548161062,
      "learning_rate": 2.7451462560413634e-06,
      "loss": 0.8807,
      "step": 1271
    },
    {
      "epoch": 0.7832512315270936,
      "grad_norm": 0.7104615021724215,
      "learning_rate": 2.730363846676922e-06,
      "loss": 0.8392,
      "step": 1272
    },
    {
      "epoch": 0.7838669950738916,
      "grad_norm": 0.7205601961052422,
      "learning_rate": 2.715615050664294e-06,
      "loss": 0.8916,
      "step": 1273
    },
    {
      "epoch": 0.7844827586206896,
      "grad_norm": 0.761627093447812,
      "learning_rate": 2.7008999361989808e-06,
      "loss": 0.8862,
      "step": 1274
    },
    {
      "epoch": 0.7850985221674877,
      "grad_norm": 0.6981721422989406,
      "learning_rate": 2.6862185713207467e-06,
      "loss": 0.9372,
      "step": 1275
    },
    {
      "epoch": 0.7857142857142857,
      "grad_norm": 0.6879745345008941,
      "learning_rate": 2.6715710239132976e-06,
      "loss": 0.8751,
      "step": 1276
    },
    {
      "epoch": 0.7863300492610837,
      "grad_norm": 0.7300207672745401,
      "learning_rate": 2.6569573617039813e-06,
      "loss": 0.8543,
      "step": 1277
    },
    {
      "epoch": 0.7869458128078818,
      "grad_norm": 0.6991783035818121,
      "learning_rate": 2.64237765226347e-06,
      "loss": 0.8621,
      "step": 1278
    },
    {
      "epoch": 0.7875615763546798,
      "grad_norm": 0.711494802058516,
      "learning_rate": 2.627831963005435e-06,
      "loss": 0.9202,
      "step": 1279
    },
    {
      "epoch": 0.7881773399014779,
      "grad_norm": 0.7167305100825402,
      "learning_rate": 2.6133203611862554e-06,
      "loss": 0.8943,
      "step": 1280
    },
    {
      "epoch": 0.7887931034482759,
      "grad_norm": 0.6625438446113113,
      "learning_rate": 2.598842913904687e-06,
      "loss": 0.934,
      "step": 1281
    },
    {
      "epoch": 0.7894088669950738,
      "grad_norm": 0.6963201684636584,
      "learning_rate": 2.5843996881015676e-06,
      "loss": 0.8896,
      "step": 1282
    },
    {
      "epoch": 0.7900246305418719,
      "grad_norm": 0.7231524568782347,
      "learning_rate": 2.569990750559507e-06,
      "loss": 0.8538,
      "step": 1283
    },
    {
      "epoch": 0.7906403940886699,
      "grad_norm": 0.7630109497585719,
      "learning_rate": 2.5556161679025594e-06,
      "loss": 0.8486,
      "step": 1284
    },
    {
      "epoch": 0.791256157635468,
      "grad_norm": 0.7417476925025379,
      "learning_rate": 2.5412760065959386e-06,
      "loss": 0.8377,
      "step": 1285
    },
    {
      "epoch": 0.791871921182266,
      "grad_norm": 0.7235668995600057,
      "learning_rate": 2.5269703329456996e-06,
      "loss": 0.9109,
      "step": 1286
    },
    {
      "epoch": 0.7924876847290641,
      "grad_norm": 0.7396287488922845,
      "learning_rate": 2.5126992130984317e-06,
      "loss": 0.8655,
      "step": 1287
    },
    {
      "epoch": 0.7931034482758621,
      "grad_norm": 0.7544218931009301,
      "learning_rate": 2.498462713040958e-06,
      "loss": 0.9521,
      "step": 1288
    },
    {
      "epoch": 0.7937192118226601,
      "grad_norm": 0.6836298206984881,
      "learning_rate": 2.484260898600018e-06,
      "loss": 0.8974,
      "step": 1289
    },
    {
      "epoch": 0.7943349753694581,
      "grad_norm": 0.6613662652656556,
      "learning_rate": 2.4700938354419823e-06,
      "loss": 0.8676,
      "step": 1290
    },
    {
      "epoch": 0.7949507389162561,
      "grad_norm": 0.6905161801297028,
      "learning_rate": 2.4559615890725286e-06,
      "loss": 0.8823,
      "step": 1291
    },
    {
      "epoch": 0.7955665024630542,
      "grad_norm": 0.7506342288421078,
      "learning_rate": 2.4418642248363554e-06,
      "loss": 0.9115,
      "step": 1292
    },
    {
      "epoch": 0.7961822660098522,
      "grad_norm": 0.73842150628241,
      "learning_rate": 2.427801807916874e-06,
      "loss": 0.8988,
      "step": 1293
    },
    {
      "epoch": 0.7967980295566502,
      "grad_norm": 0.6793391848494101,
      "learning_rate": 2.4137744033359e-06,
      "loss": 0.8715,
      "step": 1294
    },
    {
      "epoch": 0.7974137931034483,
      "grad_norm": 0.688924043991798,
      "learning_rate": 2.3997820759533654e-06,
      "loss": 0.8601,
      "step": 1295
    },
    {
      "epoch": 0.7980295566502463,
      "grad_norm": 0.726978793889228,
      "learning_rate": 2.3858248904670056e-06,
      "loss": 0.9083,
      "step": 1296
    },
    {
      "epoch": 0.7986453201970444,
      "grad_norm": 0.6849845840682508,
      "learning_rate": 2.3719029114120716e-06,
      "loss": 0.8833,
      "step": 1297
    },
    {
      "epoch": 0.7992610837438424,
      "grad_norm": 0.6903100210864577,
      "learning_rate": 2.358016203161031e-06,
      "loss": 0.9341,
      "step": 1298
    },
    {
      "epoch": 0.7998768472906403,
      "grad_norm": 0.6766698034358168,
      "learning_rate": 2.34416482992325e-06,
      "loss": 0.8583,
      "step": 1299
    },
    {
      "epoch": 0.8004926108374384,
      "grad_norm": 0.6699574090912118,
      "learning_rate": 2.3303488557447374e-06,
      "loss": 0.8359,
      "step": 1300
    },
    {
      "epoch": 0.8011083743842364,
      "grad_norm": 0.7282254569532907,
      "learning_rate": 2.316568344507799e-06,
      "loss": 0.8659,
      "step": 1301
    },
    {
      "epoch": 0.8017241379310345,
      "grad_norm": 0.7187594189195817,
      "learning_rate": 2.3028233599307837e-06,
      "loss": 0.9435,
      "step": 1302
    },
    {
      "epoch": 0.8023399014778325,
      "grad_norm": 0.687973971797822,
      "learning_rate": 2.2891139655677674e-06,
      "loss": 0.9088,
      "step": 1303
    },
    {
      "epoch": 0.8029556650246306,
      "grad_norm": 0.6612472279837367,
      "learning_rate": 2.2754402248082584e-06,
      "loss": 0.8567,
      "step": 1304
    },
    {
      "epoch": 0.8035714285714286,
      "grad_norm": 0.6920668215882422,
      "learning_rate": 2.26180220087692e-06,
      "loss": 0.8616,
      "step": 1305
    },
    {
      "epoch": 0.8041871921182266,
      "grad_norm": 0.6830448075411245,
      "learning_rate": 2.248199956833258e-06,
      "loss": 0.8696,
      "step": 1306
    },
    {
      "epoch": 0.8048029556650246,
      "grad_norm": 0.6569229837834815,
      "learning_rate": 2.234633555571345e-06,
      "loss": 0.8976,
      "step": 1307
    },
    {
      "epoch": 0.8054187192118226,
      "grad_norm": 0.6932543424034427,
      "learning_rate": 2.2211030598195247e-06,
      "loss": 0.8761,
      "step": 1308
    },
    {
      "epoch": 0.8060344827586207,
      "grad_norm": 0.6493937754591891,
      "learning_rate": 2.2076085321401152e-06,
      "loss": 0.8873,
      "step": 1309
    },
    {
      "epoch": 0.8066502463054187,
      "grad_norm": 0.7138308897777075,
      "learning_rate": 2.194150034929133e-06,
      "loss": 0.9312,
      "step": 1310
    },
    {
      "epoch": 0.8072660098522167,
      "grad_norm": 0.677096127162086,
      "learning_rate": 2.1807276304159874e-06,
      "loss": 0.832,
      "step": 1311
    },
    {
      "epoch": 0.8078817733990148,
      "grad_norm": 0.6863454675184343,
      "learning_rate": 2.1673413806632104e-06,
      "loss": 0.8672,
      "step": 1312
    },
    {
      "epoch": 0.8084975369458128,
      "grad_norm": 0.7187559551640595,
      "learning_rate": 2.1539913475661578e-06,
      "loss": 0.8916,
      "step": 1313
    },
    {
      "epoch": 0.8091133004926109,
      "grad_norm": 0.7692647968024777,
      "learning_rate": 2.140677592852728e-06,
      "loss": 0.8537,
      "step": 1314
    },
    {
      "epoch": 0.8097290640394089,
      "grad_norm": 0.6678530072291136,
      "learning_rate": 2.1274001780830776e-06,
      "loss": 0.8706,
      "step": 1315
    },
    {
      "epoch": 0.8103448275862069,
      "grad_norm": 0.6982182064654807,
      "learning_rate": 2.1141591646493275e-06,
      "loss": 0.8464,
      "step": 1316
    },
    {
      "epoch": 0.8109605911330049,
      "grad_norm": 0.6942044820541133,
      "learning_rate": 2.100954613775292e-06,
      "loss": 0.8659,
      "step": 1317
    },
    {
      "epoch": 0.8115763546798029,
      "grad_norm": 0.6911157403692272,
      "learning_rate": 2.0877865865161916e-06,
      "loss": 0.9033,
      "step": 1318
    },
    {
      "epoch": 0.812192118226601,
      "grad_norm": 0.7054627263142159,
      "learning_rate": 2.0746551437583607e-06,
      "loss": 0.8665,
      "step": 1319
    },
    {
      "epoch": 0.812807881773399,
      "grad_norm": 0.7616064483320234,
      "learning_rate": 2.0615603462189824e-06,
      "loss": 0.8422,
      "step": 1320
    },
    {
      "epoch": 0.8134236453201971,
      "grad_norm": 0.729714535998479,
      "learning_rate": 2.0485022544457933e-06,
      "loss": 0.8902,
      "step": 1321
    },
    {
      "epoch": 0.8140394088669951,
      "grad_norm": 0.7274064388625505,
      "learning_rate": 2.0354809288168143e-06,
      "loss": 0.939,
      "step": 1322
    },
    {
      "epoch": 0.8146551724137931,
      "grad_norm": 0.6901914435910567,
      "learning_rate": 2.0224964295400684e-06,
      "loss": 0.8814,
      "step": 1323
    },
    {
      "epoch": 0.8152709359605911,
      "grad_norm": 0.7033003360949591,
      "learning_rate": 2.009548816653293e-06,
      "loss": 0.8396,
      "step": 1324
    },
    {
      "epoch": 0.8158866995073891,
      "grad_norm": 0.6619644651209049,
      "learning_rate": 1.9966381500236786e-06,
      "loss": 0.8545,
      "step": 1325
    },
    {
      "epoch": 0.8165024630541872,
      "grad_norm": 0.689482030027885,
      "learning_rate": 1.983764489347574e-06,
      "loss": 0.9176,
      "step": 1326
    },
    {
      "epoch": 0.8171182266009852,
      "grad_norm": 0.7034970047312203,
      "learning_rate": 1.9709278941502363e-06,
      "loss": 0.889,
      "step": 1327
    },
    {
      "epoch": 0.8177339901477833,
      "grad_norm": 0.6921347780556869,
      "learning_rate": 1.9581284237855204e-06,
      "loss": 0.9315,
      "step": 1328
    },
    {
      "epoch": 0.8183497536945813,
      "grad_norm": 0.6803194702289825,
      "learning_rate": 1.945366137435635e-06,
      "loss": 0.8784,
      "step": 1329
    },
    {
      "epoch": 0.8189655172413793,
      "grad_norm": 0.6602505099556776,
      "learning_rate": 1.932641094110855e-06,
      "loss": 0.8498,
      "step": 1330
    },
    {
      "epoch": 0.8195812807881774,
      "grad_norm": 0.7370194018029482,
      "learning_rate": 1.9199533526492465e-06,
      "loss": 0.9022,
      "step": 1331
    },
    {
      "epoch": 0.8201970443349754,
      "grad_norm": 0.6673331489053921,
      "learning_rate": 1.9073029717164048e-06,
      "loss": 0.858,
      "step": 1332
    },
    {
      "epoch": 0.8208128078817734,
      "grad_norm": 0.7272649569029246,
      "learning_rate": 1.894690009805178e-06,
      "loss": 0.8682,
      "step": 1333
    },
    {
      "epoch": 0.8214285714285714,
      "grad_norm": 0.6874140609762676,
      "learning_rate": 1.882114525235389e-06,
      "loss": 0.8269,
      "step": 1334
    },
    {
      "epoch": 0.8220443349753694,
      "grad_norm": 0.6733544606617353,
      "learning_rate": 1.869576576153581e-06,
      "loss": 0.9205,
      "step": 1335
    },
    {
      "epoch": 0.8226600985221675,
      "grad_norm": 0.7420922856003039,
      "learning_rate": 1.857076220532734e-06,
      "loss": 0.8624,
      "step": 1336
    },
    {
      "epoch": 0.8232758620689655,
      "grad_norm": 0.7551120318304361,
      "learning_rate": 1.844613516172008e-06,
      "loss": 0.8885,
      "step": 1337
    },
    {
      "epoch": 0.8238916256157636,
      "grad_norm": 0.6703769887838777,
      "learning_rate": 1.8321885206964719e-06,
      "loss": 0.8895,
      "step": 1338
    },
    {
      "epoch": 0.8245073891625616,
      "grad_norm": 0.6759800940209398,
      "learning_rate": 1.8198012915568253e-06,
      "loss": 0.9117,
      "step": 1339
    },
    {
      "epoch": 0.8251231527093597,
      "grad_norm": 0.7571027934722131,
      "learning_rate": 1.8074518860291646e-06,
      "loss": 0.8399,
      "step": 1340
    },
    {
      "epoch": 0.8257389162561576,
      "grad_norm": 0.7025487067328403,
      "learning_rate": 1.7951403612146745e-06,
      "loss": 0.9115,
      "step": 1341
    },
    {
      "epoch": 0.8263546798029556,
      "grad_norm": 0.7561651951268618,
      "learning_rate": 1.7828667740394045e-06,
      "loss": 0.8787,
      "step": 1342
    },
    {
      "epoch": 0.8269704433497537,
      "grad_norm": 0.7161961848706409,
      "learning_rate": 1.7706311812539756e-06,
      "loss": 0.9154,
      "step": 1343
    },
    {
      "epoch": 0.8275862068965517,
      "grad_norm": 0.7501514694984982,
      "learning_rate": 1.7584336394333367e-06,
      "loss": 0.9315,
      "step": 1344
    },
    {
      "epoch": 0.8282019704433498,
      "grad_norm": 0.6890018753605649,
      "learning_rate": 1.746274204976498e-06,
      "loss": 0.9278,
      "step": 1345
    },
    {
      "epoch": 0.8288177339901478,
      "grad_norm": 0.6444120018866913,
      "learning_rate": 1.7341529341062624e-06,
      "loss": 0.8857,
      "step": 1346
    },
    {
      "epoch": 0.8294334975369458,
      "grad_norm": 0.6912333590884836,
      "learning_rate": 1.7220698828689785e-06,
      "loss": 0.8615,
      "step": 1347
    },
    {
      "epoch": 0.8300492610837439,
      "grad_norm": 0.6969383616749296,
      "learning_rate": 1.7100251071342721e-06,
      "loss": 0.9057,
      "step": 1348
    },
    {
      "epoch": 0.8306650246305419,
      "grad_norm": 0.6782028637332954,
      "learning_rate": 1.698018662594789e-06,
      "loss": 0.8943,
      "step": 1349
    },
    {
      "epoch": 0.8312807881773399,
      "grad_norm": 0.7353446227842368,
      "learning_rate": 1.6860506047659442e-06,
      "loss": 0.9455,
      "step": 1350
    },
    {
      "epoch": 0.8318965517241379,
      "grad_norm": 0.6719699140813302,
      "learning_rate": 1.6741209889856513e-06,
      "loss": 0.8889,
      "step": 1351
    },
    {
      "epoch": 0.8325123152709359,
      "grad_norm": 0.6672640673845985,
      "learning_rate": 1.662229870414086e-06,
      "loss": 0.8438,
      "step": 1352
    },
    {
      "epoch": 0.833128078817734,
      "grad_norm": 0.7467768623403751,
      "learning_rate": 1.6503773040334126e-06,
      "loss": 0.911,
      "step": 1353
    },
    {
      "epoch": 0.833743842364532,
      "grad_norm": 0.7106018711095955,
      "learning_rate": 1.6385633446475434e-06,
      "loss": 0.8988,
      "step": 1354
    },
    {
      "epoch": 0.8343596059113301,
      "grad_norm": 0.7013841851504359,
      "learning_rate": 1.6267880468818787e-06,
      "loss": 0.8584,
      "step": 1355
    },
    {
      "epoch": 0.8349753694581281,
      "grad_norm": 0.6824824816719666,
      "learning_rate": 1.6150514651830485e-06,
      "loss": 0.8621,
      "step": 1356
    },
    {
      "epoch": 0.8355911330049262,
      "grad_norm": 0.723220520763517,
      "learning_rate": 1.6033536538186778e-06,
      "loss": 0.8891,
      "step": 1357
    },
    {
      "epoch": 0.8362068965517241,
      "grad_norm": 0.735251427123691,
      "learning_rate": 1.591694666877114e-06,
      "loss": 0.8881,
      "step": 1358
    },
    {
      "epoch": 0.8368226600985221,
      "grad_norm": 0.7261763246358538,
      "learning_rate": 1.5800745582671973e-06,
      "loss": 0.8964,
      "step": 1359
    },
    {
      "epoch": 0.8374384236453202,
      "grad_norm": 0.6905242840613158,
      "learning_rate": 1.5684933817180014e-06,
      "loss": 0.8817,
      "step": 1360
    },
    {
      "epoch": 0.8380541871921182,
      "grad_norm": 0.7100074184680476,
      "learning_rate": 1.5569511907785783e-06,
      "loss": 0.8281,
      "step": 1361
    },
    {
      "epoch": 0.8386699507389163,
      "grad_norm": 0.7211718885899312,
      "learning_rate": 1.5454480388177263e-06,
      "loss": 0.8421,
      "step": 1362
    },
    {
      "epoch": 0.8392857142857143,
      "grad_norm": 0.7219332486217335,
      "learning_rate": 1.5339839790237333e-06,
      "loss": 0.9335,
      "step": 1363
    },
    {
      "epoch": 0.8399014778325123,
      "grad_norm": 0.7309447719905752,
      "learning_rate": 1.5225590644041287e-06,
      "loss": 0.852,
      "step": 1364
    },
    {
      "epoch": 0.8405172413793104,
      "grad_norm": 0.6659216018144737,
      "learning_rate": 1.5111733477854507e-06,
      "loss": 0.845,
      "step": 1365
    },
    {
      "epoch": 0.8411330049261084,
      "grad_norm": 0.7847721499011631,
      "learning_rate": 1.4998268818129801e-06,
      "loss": 0.9002,
      "step": 1366
    },
    {
      "epoch": 0.8417487684729064,
      "grad_norm": 0.6859509353756376,
      "learning_rate": 1.4885197189505285e-06,
      "loss": 0.8833,
      "step": 1367
    },
    {
      "epoch": 0.8423645320197044,
      "grad_norm": 0.6760794464391707,
      "learning_rate": 1.477251911480162e-06,
      "loss": 0.884,
      "step": 1368
    },
    {
      "epoch": 0.8429802955665024,
      "grad_norm": 0.6589254599898151,
      "learning_rate": 1.4660235115019838e-06,
      "loss": 0.9182,
      "step": 1369
    },
    {
      "epoch": 0.8435960591133005,
      "grad_norm": 0.6784828513131478,
      "learning_rate": 1.454834570933884e-06,
      "loss": 0.8939,
      "step": 1370
    },
    {
      "epoch": 0.8442118226600985,
      "grad_norm": 0.6693101153861184,
      "learning_rate": 1.4436851415112941e-06,
      "loss": 0.8049,
      "step": 1371
    },
    {
      "epoch": 0.8448275862068966,
      "grad_norm": 0.636945608355087,
      "learning_rate": 1.4325752747869626e-06,
      "loss": 0.8653,
      "step": 1372
    },
    {
      "epoch": 0.8454433497536946,
      "grad_norm": 0.6821098598510313,
      "learning_rate": 1.4215050221307002e-06,
      "loss": 0.8738,
      "step": 1373
    },
    {
      "epoch": 0.8460591133004927,
      "grad_norm": 0.7089859648752211,
      "learning_rate": 1.4104744347291555e-06,
      "loss": 0.8955,
      "step": 1374
    },
    {
      "epoch": 0.8466748768472906,
      "grad_norm": 0.7016088372921392,
      "learning_rate": 1.399483563585573e-06,
      "loss": 0.9048,
      "step": 1375
    },
    {
      "epoch": 0.8472906403940886,
      "grad_norm": 0.7521478775855152,
      "learning_rate": 1.3885324595195493e-06,
      "loss": 0.8713,
      "step": 1376
    },
    {
      "epoch": 0.8479064039408867,
      "grad_norm": 0.7144394129123897,
      "learning_rate": 1.3776211731668187e-06,
      "loss": 0.9508,
      "step": 1377
    },
    {
      "epoch": 0.8485221674876847,
      "grad_norm": 0.7259168913142005,
      "learning_rate": 1.3667497549789933e-06,
      "loss": 0.9321,
      "step": 1378
    },
    {
      "epoch": 0.8491379310344828,
      "grad_norm": 0.700674679476634,
      "learning_rate": 1.3559182552233497e-06,
      "loss": 0.9239,
      "step": 1379
    },
    {
      "epoch": 0.8497536945812808,
      "grad_norm": 0.7019193889530307,
      "learning_rate": 1.345126723982594e-06,
      "loss": 0.8675,
      "step": 1380
    },
    {
      "epoch": 0.8503694581280788,
      "grad_norm": 0.6890763944443243,
      "learning_rate": 1.3343752111546139e-06,
      "loss": 0.8132,
      "step": 1381
    },
    {
      "epoch": 0.8509852216748769,
      "grad_norm": 0.680448050970852,
      "learning_rate": 1.3236637664522722e-06,
      "loss": 0.9271,
      "step": 1382
    },
    {
      "epoch": 0.8516009852216748,
      "grad_norm": 0.6896367271716518,
      "learning_rate": 1.3129924394031535e-06,
      "loss": 0.8941,
      "step": 1383
    },
    {
      "epoch": 0.8522167487684729,
      "grad_norm": 0.7315695533677982,
      "learning_rate": 1.3023612793493535e-06,
      "loss": 0.8917,
      "step": 1384
    },
    {
      "epoch": 0.8528325123152709,
      "grad_norm": 0.6733301510685723,
      "learning_rate": 1.2917703354472467e-06,
      "loss": 0.8746,
      "step": 1385
    },
    {
      "epoch": 0.853448275862069,
      "grad_norm": 0.6971500073946438,
      "learning_rate": 1.2812196566672441e-06,
      "loss": 0.8366,
      "step": 1386
    },
    {
      "epoch": 0.854064039408867,
      "grad_norm": 0.6547238214161516,
      "learning_rate": 1.2707092917935914e-06,
      "loss": 0.8815,
      "step": 1387
    },
    {
      "epoch": 0.854679802955665,
      "grad_norm": 0.7037114792126304,
      "learning_rate": 1.2602392894241222e-06,
      "loss": 0.8984,
      "step": 1388
    },
    {
      "epoch": 0.8552955665024631,
      "grad_norm": 0.7077408491764982,
      "learning_rate": 1.2498096979700459e-06,
      "loss": 0.9105,
      "step": 1389
    },
    {
      "epoch": 0.8559113300492611,
      "grad_norm": 0.7047156583110215,
      "learning_rate": 1.2394205656557224e-06,
      "loss": 0.8682,
      "step": 1390
    },
    {
      "epoch": 0.8565270935960592,
      "grad_norm": 0.6880847497746563,
      "learning_rate": 1.2290719405184294e-06,
      "loss": 0.9168,
      "step": 1391
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.7427055319972887,
      "learning_rate": 1.2187638704081516e-06,
      "loss": 0.917,
      "step": 1392
    },
    {
      "epoch": 0.8577586206896551,
      "grad_norm": 0.6744677791729871,
      "learning_rate": 1.208496402987358e-06,
      "loss": 0.8557,
      "step": 1393
    },
    {
      "epoch": 0.8583743842364532,
      "grad_norm": 0.7446264201547833,
      "learning_rate": 1.1982695857307757e-06,
      "loss": 0.8749,
      "step": 1394
    },
    {
      "epoch": 0.8589901477832512,
      "grad_norm": 0.7102898821667599,
      "learning_rate": 1.1880834659251706e-06,
      "loss": 0.8604,
      "step": 1395
    },
    {
      "epoch": 0.8596059113300493,
      "grad_norm": 0.7258306808885525,
      "learning_rate": 1.1779380906691362e-06,
      "loss": 0.8999,
      "step": 1396
    },
    {
      "epoch": 0.8602216748768473,
      "grad_norm": 0.7147424949040373,
      "learning_rate": 1.167833506872874e-06,
      "loss": 0.8898,
      "step": 1397
    },
    {
      "epoch": 0.8608374384236454,
      "grad_norm": 0.6679190413989127,
      "learning_rate": 1.1577697612579642e-06,
      "loss": 0.8629,
      "step": 1398
    },
    {
      "epoch": 0.8614532019704434,
      "grad_norm": 0.7624338454985283,
      "learning_rate": 1.1477469003571672e-06,
      "loss": 0.8611,
      "step": 1399
    },
    {
      "epoch": 0.8620689655172413,
      "grad_norm": 0.7011040594053884,
      "learning_rate": 1.1377649705142012e-06,
      "loss": 0.8915,
      "step": 1400
    },
    {
      "epoch": 0.8620689655172413,
      "eval_loss": 0.8917863368988037,
      "eval_runtime": 523.1323,
      "eval_samples_per_second": 44.176,
      "eval_steps_per_second": 0.692,
      "step": 1400
    },
    {
      "epoch": 0.8626847290640394,
      "grad_norm": 0.7060647327425855,
      "learning_rate": 1.1278240178835197e-06,
      "loss": 0.8731,
      "step": 1401
    },
    {
      "epoch": 0.8633004926108374,
      "grad_norm": 0.7147211592219792,
      "learning_rate": 1.1179240884301158e-06,
      "loss": 0.8285,
      "step": 1402
    },
    {
      "epoch": 0.8639162561576355,
      "grad_norm": 0.7582881345547138,
      "learning_rate": 1.1080652279292892e-06,
      "loss": 0.8705,
      "step": 1403
    },
    {
      "epoch": 0.8645320197044335,
      "grad_norm": 0.71255709795896,
      "learning_rate": 1.0982474819664524e-06,
      "loss": 0.912,
      "step": 1404
    },
    {
      "epoch": 0.8651477832512315,
      "grad_norm": 0.7270415543443802,
      "learning_rate": 1.0884708959369116e-06,
      "loss": 0.8913,
      "step": 1405
    },
    {
      "epoch": 0.8657635467980296,
      "grad_norm": 0.6828239949234485,
      "learning_rate": 1.0787355150456546e-06,
      "loss": 0.8407,
      "step": 1406
    },
    {
      "epoch": 0.8663793103448276,
      "grad_norm": 0.6513420424298494,
      "learning_rate": 1.069041384307149e-06,
      "loss": 0.8498,
      "step": 1407
    },
    {
      "epoch": 0.8669950738916257,
      "grad_norm": 0.6792362993242139,
      "learning_rate": 1.0593885485451239e-06,
      "loss": 0.9355,
      "step": 1408
    },
    {
      "epoch": 0.8676108374384236,
      "grad_norm": 0.6783965452098765,
      "learning_rate": 1.0497770523923756e-06,
      "loss": 0.8838,
      "step": 1409
    },
    {
      "epoch": 0.8682266009852216,
      "grad_norm": 0.7035239435365321,
      "learning_rate": 1.040206940290547e-06,
      "loss": 0.9456,
      "step": 1410
    },
    {
      "epoch": 0.8688423645320197,
      "grad_norm": 0.6866172736096757,
      "learning_rate": 1.0306782564899353e-06,
      "loss": 0.9018,
      "step": 1411
    },
    {
      "epoch": 0.8694581280788177,
      "grad_norm": 0.7199186362813331,
      "learning_rate": 1.0211910450492813e-06,
      "loss": 0.8696,
      "step": 1412
    },
    {
      "epoch": 0.8700738916256158,
      "grad_norm": 0.6491077351810749,
      "learning_rate": 1.0117453498355589e-06,
      "loss": 0.854,
      "step": 1413
    },
    {
      "epoch": 0.8706896551724138,
      "grad_norm": 0.6634598436607574,
      "learning_rate": 1.0023412145237888e-06,
      "loss": 0.8612,
      "step": 1414
    },
    {
      "epoch": 0.8713054187192119,
      "grad_norm": 0.7331697823873177,
      "learning_rate": 9.929786825968213e-07,
      "loss": 0.9478,
      "step": 1415
    },
    {
      "epoch": 0.8719211822660099,
      "grad_norm": 0.7404794052541122,
      "learning_rate": 9.83657797345141e-07,
      "loss": 0.9266,
      "step": 1416
    },
    {
      "epoch": 0.8725369458128078,
      "grad_norm": 0.6850111320625838,
      "learning_rate": 9.74378601866669e-07,
      "loss": 0.8825,
      "step": 1417
    },
    {
      "epoch": 0.8731527093596059,
      "grad_norm": 0.6783314069895383,
      "learning_rate": 9.651411390665577e-07,
      "loss": 0.8231,
      "step": 1418
    },
    {
      "epoch": 0.8737684729064039,
      "grad_norm": 0.6897575805871837,
      "learning_rate": 9.559454516569988e-07,
      "loss": 0.8668,
      "step": 1419
    },
    {
      "epoch": 0.874384236453202,
      "grad_norm": 0.7127280264038853,
      "learning_rate": 9.467915821570228e-07,
      "loss": 0.8568,
      "step": 1420
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.7144772593038013,
      "learning_rate": 9.376795728923016e-07,
      "loss": 0.8677,
      "step": 1421
    },
    {
      "epoch": 0.875615763546798,
      "grad_norm": 0.6826760547460019,
      "learning_rate": 9.286094659949552e-07,
      "loss": 0.9087,
      "step": 1422
    },
    {
      "epoch": 0.8762315270935961,
      "grad_norm": 0.6739078329215882,
      "learning_rate": 9.195813034033507e-07,
      "loss": 0.8677,
      "step": 1423
    },
    {
      "epoch": 0.8768472906403941,
      "grad_norm": 0.7337514720697376,
      "learning_rate": 9.105951268619206e-07,
      "loss": 0.8861,
      "step": 1424
    },
    {
      "epoch": 0.8774630541871922,
      "grad_norm": 0.6886457729998411,
      "learning_rate": 9.016509779209536e-07,
      "loss": 0.8793,
      "step": 1425
    },
    {
      "epoch": 0.8780788177339901,
      "grad_norm": 0.7033627506226959,
      "learning_rate": 8.927488979364185e-07,
      "loss": 0.8492,
      "step": 1426
    },
    {
      "epoch": 0.8786945812807881,
      "grad_norm": 0.6894608357505778,
      "learning_rate": 8.838889280697637e-07,
      "loss": 0.8668,
      "step": 1427
    },
    {
      "epoch": 0.8793103448275862,
      "grad_norm": 0.7403749765302757,
      "learning_rate": 8.75071109287724e-07,
      "loss": 0.9062,
      "step": 1428
    },
    {
      "epoch": 0.8799261083743842,
      "grad_norm": 0.6868195072985124,
      "learning_rate": 8.662954823621428e-07,
      "loss": 0.9033,
      "step": 1429
    },
    {
      "epoch": 0.8805418719211823,
      "grad_norm": 0.7501136249611785,
      "learning_rate": 8.575620878697744e-07,
      "loss": 0.9595,
      "step": 1430
    },
    {
      "epoch": 0.8811576354679803,
      "grad_norm": 0.6524509996736709,
      "learning_rate": 8.488709661920946e-07,
      "loss": 0.8487,
      "step": 1431
    },
    {
      "epoch": 0.8817733990147784,
      "grad_norm": 0.7193559806670244,
      "learning_rate": 8.402221575151238e-07,
      "loss": 0.882,
      "step": 1432
    },
    {
      "epoch": 0.8823891625615764,
      "grad_norm": 0.6763306249667755,
      "learning_rate": 8.316157018292326e-07,
      "loss": 0.9101,
      "step": 1433
    },
    {
      "epoch": 0.8830049261083743,
      "grad_norm": 0.7235459063153095,
      "learning_rate": 8.230516389289634e-07,
      "loss": 0.8982,
      "step": 1434
    },
    {
      "epoch": 0.8836206896551724,
      "grad_norm": 0.6830268566684098,
      "learning_rate": 8.145300084128349e-07,
      "loss": 0.8571,
      "step": 1435
    },
    {
      "epoch": 0.8842364532019704,
      "grad_norm": 0.7349842223025389,
      "learning_rate": 8.060508496831742e-07,
      "loss": 0.85,
      "step": 1436
    },
    {
      "epoch": 0.8848522167487685,
      "grad_norm": 0.6503277303735258,
      "learning_rate": 7.976142019459277e-07,
      "loss": 0.8926,
      "step": 1437
    },
    {
      "epoch": 0.8854679802955665,
      "grad_norm": 0.6859194704963234,
      "learning_rate": 7.892201042104719e-07,
      "loss": 0.8635,
      "step": 1438
    },
    {
      "epoch": 0.8860837438423645,
      "grad_norm": 0.7381142683238526,
      "learning_rate": 7.808685952894501e-07,
      "loss": 0.9409,
      "step": 1439
    },
    {
      "epoch": 0.8866995073891626,
      "grad_norm": 0.7596321875248033,
      "learning_rate": 7.725597137985741e-07,
      "loss": 0.9409,
      "step": 1440
    },
    {
      "epoch": 0.8873152709359606,
      "grad_norm": 0.6910708742557097,
      "learning_rate": 7.642934981564609e-07,
      "loss": 0.9179,
      "step": 1441
    },
    {
      "epoch": 0.8879310344827587,
      "grad_norm": 0.7177307993884582,
      "learning_rate": 7.560699865844501e-07,
      "loss": 0.9113,
      "step": 1442
    },
    {
      "epoch": 0.8885467980295566,
      "grad_norm": 0.7106990315287217,
      "learning_rate": 7.478892171064201e-07,
      "loss": 0.882,
      "step": 1443
    },
    {
      "epoch": 0.8891625615763546,
      "grad_norm": 0.6990608092010604,
      "learning_rate": 7.397512275486241e-07,
      "loss": 0.8568,
      "step": 1444
    },
    {
      "epoch": 0.8897783251231527,
      "grad_norm": 0.7050176996842427,
      "learning_rate": 7.316560555395069e-07,
      "loss": 0.8745,
      "step": 1445
    },
    {
      "epoch": 0.8903940886699507,
      "grad_norm": 0.7042171582242612,
      "learning_rate": 7.23603738509534e-07,
      "loss": 0.8951,
      "step": 1446
    },
    {
      "epoch": 0.8910098522167488,
      "grad_norm": 0.69305634114784,
      "learning_rate": 7.155943136910193e-07,
      "loss": 0.9267,
      "step": 1447
    },
    {
      "epoch": 0.8916256157635468,
      "grad_norm": 0.7241262302877287,
      "learning_rate": 7.076278181179486e-07,
      "loss": 0.8283,
      "step": 1448
    },
    {
      "epoch": 0.8922413793103449,
      "grad_norm": 0.7263458292040635,
      "learning_rate": 6.997042886258143e-07,
      "loss": 0.885,
      "step": 1449
    },
    {
      "epoch": 0.8928571428571429,
      "grad_norm": 0.6775207884121133,
      "learning_rate": 6.918237618514378e-07,
      "loss": 0.9014,
      "step": 1450
    },
    {
      "epoch": 0.8934729064039408,
      "grad_norm": 0.6983204005232182,
      "learning_rate": 6.839862742328107e-07,
      "loss": 0.9372,
      "step": 1451
    },
    {
      "epoch": 0.8940886699507389,
      "grad_norm": 0.6761695805179136,
      "learning_rate": 6.761918620089181e-07,
      "loss": 0.8871,
      "step": 1452
    },
    {
      "epoch": 0.8947044334975369,
      "grad_norm": 0.7014201222832569,
      "learning_rate": 6.684405612195688e-07,
      "loss": 0.8813,
      "step": 1453
    },
    {
      "epoch": 0.895320197044335,
      "grad_norm": 0.649062919631672,
      "learning_rate": 6.60732407705239e-07,
      "loss": 0.8997,
      "step": 1454
    },
    {
      "epoch": 0.895935960591133,
      "grad_norm": 0.7218805943526583,
      "learning_rate": 6.530674371068946e-07,
      "loss": 0.9209,
      "step": 1455
    },
    {
      "epoch": 0.896551724137931,
      "grad_norm": 0.6888546682186595,
      "learning_rate": 6.454456848658364e-07,
      "loss": 0.8808,
      "step": 1456
    },
    {
      "epoch": 0.8971674876847291,
      "grad_norm": 0.6751838117283254,
      "learning_rate": 6.378671862235331e-07,
      "loss": 0.8363,
      "step": 1457
    },
    {
      "epoch": 0.8977832512315271,
      "grad_norm": 0.6916999590746955,
      "learning_rate": 6.3033197622145e-07,
      "loss": 0.8993,
      "step": 1458
    },
    {
      "epoch": 0.8983990147783252,
      "grad_norm": 0.7811410705308859,
      "learning_rate": 6.228400897009068e-07,
      "loss": 0.841,
      "step": 1459
    },
    {
      "epoch": 0.8990147783251231,
      "grad_norm": 0.6910995131031151,
      "learning_rate": 6.153915613028915e-07,
      "loss": 0.8487,
      "step": 1460
    },
    {
      "epoch": 0.8996305418719212,
      "grad_norm": 0.6948043639867122,
      "learning_rate": 6.07986425467919e-07,
      "loss": 0.8441,
      "step": 1461
    },
    {
      "epoch": 0.9002463054187192,
      "grad_norm": 0.701171216763334,
      "learning_rate": 6.00624716435868e-07,
      "loss": 0.9265,
      "step": 1462
    },
    {
      "epoch": 0.9008620689655172,
      "grad_norm": 0.6603754347835511,
      "learning_rate": 5.933064682458123e-07,
      "loss": 0.8573,
      "step": 1463
    },
    {
      "epoch": 0.9014778325123153,
      "grad_norm": 0.6587648260360551,
      "learning_rate": 5.860317147358796e-07,
      "loss": 0.8826,
      "step": 1464
    },
    {
      "epoch": 0.9020935960591133,
      "grad_norm": 0.6905223891036076,
      "learning_rate": 5.788004895430799e-07,
      "loss": 0.9048,
      "step": 1465
    },
    {
      "epoch": 0.9027093596059114,
      "grad_norm": 0.7099567495312544,
      "learning_rate": 5.716128261031628e-07,
      "loss": 0.9012,
      "step": 1466
    },
    {
      "epoch": 0.9033251231527094,
      "grad_norm": 0.71453664068664,
      "learning_rate": 5.64468757650456e-07,
      "loss": 0.846,
      "step": 1467
    },
    {
      "epoch": 0.9039408866995073,
      "grad_norm": 0.6972552213101737,
      "learning_rate": 5.57368317217708e-07,
      "loss": 0.9116,
      "step": 1468
    },
    {
      "epoch": 0.9045566502463054,
      "grad_norm": 0.6995589002067875,
      "learning_rate": 5.503115376359459e-07,
      "loss": 0.8701,
      "step": 1469
    },
    {
      "epoch": 0.9051724137931034,
      "grad_norm": 0.6679956253363852,
      "learning_rate": 5.43298451534312e-07,
      "loss": 0.8651,
      "step": 1470
    },
    {
      "epoch": 0.9057881773399015,
      "grad_norm": 0.6747146700049863,
      "learning_rate": 5.363290913399232e-07,
      "loss": 0.8844,
      "step": 1471
    },
    {
      "epoch": 0.9064039408866995,
      "grad_norm": 0.7295697806495524,
      "learning_rate": 5.294034892777167e-07,
      "loss": 0.9383,
      "step": 1472
    },
    {
      "epoch": 0.9070197044334976,
      "grad_norm": 0.6979939879357876,
      "learning_rate": 5.225216773702968e-07,
      "loss": 0.8472,
      "step": 1473
    },
    {
      "epoch": 0.9076354679802956,
      "grad_norm": 0.693617466920837,
      "learning_rate": 5.156836874377958e-07,
      "loss": 0.8696,
      "step": 1474
    },
    {
      "epoch": 0.9082512315270936,
      "grad_norm": 0.6711641682577815,
      "learning_rate": 5.088895510977154e-07,
      "loss": 0.8511,
      "step": 1475
    },
    {
      "epoch": 0.9088669950738916,
      "grad_norm": 0.6853178862122168,
      "learning_rate": 5.021392997647933e-07,
      "loss": 0.8422,
      "step": 1476
    },
    {
      "epoch": 0.9094827586206896,
      "grad_norm": 0.7012150409409456,
      "learning_rate": 4.954329646508505e-07,
      "loss": 0.8906,
      "step": 1477
    },
    {
      "epoch": 0.9100985221674877,
      "grad_norm": 0.7060134114635694,
      "learning_rate": 4.887705767646433e-07,
      "loss": 0.8757,
      "step": 1478
    },
    {
      "epoch": 0.9107142857142857,
      "grad_norm": 0.7040819052837094,
      "learning_rate": 4.821521669117324e-07,
      "loss": 0.9328,
      "step": 1479
    },
    {
      "epoch": 0.9113300492610837,
      "grad_norm": 0.6976672394279331,
      "learning_rate": 4.755777656943239e-07,
      "loss": 0.8872,
      "step": 1480
    },
    {
      "epoch": 0.9119458128078818,
      "grad_norm": 0.682379872487146,
      "learning_rate": 4.6904740351114477e-07,
      "loss": 0.8644,
      "step": 1481
    },
    {
      "epoch": 0.9125615763546798,
      "grad_norm": 0.6796317743344112,
      "learning_rate": 4.6256111055729134e-07,
      "loss": 0.9111,
      "step": 1482
    },
    {
      "epoch": 0.9131773399014779,
      "grad_norm": 0.6683479053267635,
      "learning_rate": 4.561189168240909e-07,
      "loss": 0.9148,
      "step": 1483
    },
    {
      "epoch": 0.9137931034482759,
      "grad_norm": 0.7483027549189329,
      "learning_rate": 4.497208520989671e-07,
      "loss": 0.86,
      "step": 1484
    },
    {
      "epoch": 0.9144088669950738,
      "grad_norm": 0.740498884963197,
      "learning_rate": 4.433669459652945e-07,
      "loss": 0.8652,
      "step": 1485
    },
    {
      "epoch": 0.9150246305418719,
      "grad_norm": 0.6923375319507666,
      "learning_rate": 4.3705722780227887e-07,
      "loss": 0.8981,
      "step": 1486
    },
    {
      "epoch": 0.9156403940886699,
      "grad_norm": 0.7230013794227099,
      "learning_rate": 4.3079172678479717e-07,
      "loss": 0.8874,
      "step": 1487
    },
    {
      "epoch": 0.916256157635468,
      "grad_norm": 0.7298204994350348,
      "learning_rate": 4.2457047188328104e-07,
      "loss": 0.9171,
      "step": 1488
    },
    {
      "epoch": 0.916871921182266,
      "grad_norm": 0.7174869388413678,
      "learning_rate": 4.183934918635768e-07,
      "loss": 0.9081,
      "step": 1489
    },
    {
      "epoch": 0.9174876847290641,
      "grad_norm": 0.6348144766379977,
      "learning_rate": 4.1226081528680907e-07,
      "loss": 0.8722,
      "step": 1490
    },
    {
      "epoch": 0.9181034482758621,
      "grad_norm": 0.7155705747566401,
      "learning_rate": 4.0617247050925733e-07,
      "loss": 0.9213,
      "step": 1491
    },
    {
      "epoch": 0.9187192118226601,
      "grad_norm": 0.6666023091000182,
      "learning_rate": 4.001284856822174e-07,
      "loss": 0.8758,
      "step": 1492
    },
    {
      "epoch": 0.9193349753694581,
      "grad_norm": 0.7001378738096707,
      "learning_rate": 3.9412888875187127e-07,
      "loss": 0.9012,
      "step": 1493
    },
    {
      "epoch": 0.9199507389162561,
      "grad_norm": 0.6906259199169451,
      "learning_rate": 3.881737074591629e-07,
      "loss": 0.9029,
      "step": 1494
    },
    {
      "epoch": 0.9205665024630542,
      "grad_norm": 0.7013707416960405,
      "learning_rate": 3.822629693396651e-07,
      "loss": 0.8983,
      "step": 1495
    },
    {
      "epoch": 0.9211822660098522,
      "grad_norm": 0.6850603198219924,
      "learning_rate": 3.76396701723456e-07,
      "loss": 0.8324,
      "step": 1496
    },
    {
      "epoch": 0.9217980295566502,
      "grad_norm": 0.682394069020597,
      "learning_rate": 3.705749317349916e-07,
      "loss": 0.9164,
      "step": 1497
    },
    {
      "epoch": 0.9224137931034483,
      "grad_norm": 0.6838573458872114,
      "learning_rate": 3.6479768629297473e-07,
      "loss": 0.8084,
      "step": 1498
    },
    {
      "epoch": 0.9230295566502463,
      "grad_norm": 0.746524942310327,
      "learning_rate": 3.5906499211024514e-07,
      "loss": 0.8909,
      "step": 1499
    },
    {
      "epoch": 0.9236453201970444,
      "grad_norm": 0.7338870656671379,
      "learning_rate": 3.5337687569363734e-07,
      "loss": 0.868,
      "step": 1500
    },
    {
      "epoch": 0.9242610837438424,
      "grad_norm": 0.7186354465149037,
      "learning_rate": 3.4773336334387063e-07,
      "loss": 0.8725,
      "step": 1501
    },
    {
      "epoch": 0.9248768472906403,
      "grad_norm": 0.6681117547595306,
      "learning_rate": 3.421344811554228e-07,
      "loss": 0.8624,
      "step": 1502
    },
    {
      "epoch": 0.9254926108374384,
      "grad_norm": 0.7108712296420937,
      "learning_rate": 3.365802550164132e-07,
      "loss": 0.9132,
      "step": 1503
    },
    {
      "epoch": 0.9261083743842364,
      "grad_norm": 0.7088008485446813,
      "learning_rate": 3.310707106084798e-07,
      "loss": 0.8749,
      "step": 1504
    },
    {
      "epoch": 0.9267241379310345,
      "grad_norm": 0.6409847358732418,
      "learning_rate": 3.2560587340665694e-07,
      "loss": 0.8913,
      "step": 1505
    },
    {
      "epoch": 0.9273399014778325,
      "grad_norm": 0.6740894215154577,
      "learning_rate": 3.2018576867926644e-07,
      "loss": 0.9168,
      "step": 1506
    },
    {
      "epoch": 0.9279556650246306,
      "grad_norm": 0.7018070380883923,
      "learning_rate": 3.1481042148779674e-07,
      "loss": 0.8626,
      "step": 1507
    },
    {
      "epoch": 0.9285714285714286,
      "grad_norm": 0.6965244265618198,
      "learning_rate": 3.0947985668678183e-07,
      "loss": 0.8943,
      "step": 1508
    },
    {
      "epoch": 0.9291871921182266,
      "grad_norm": 0.7478441799972412,
      "learning_rate": 3.041940989236947e-07,
      "loss": 0.8964,
      "step": 1509
    },
    {
      "epoch": 0.9298029556650246,
      "grad_norm": 0.643242854379963,
      "learning_rate": 2.989531726388262e-07,
      "loss": 0.9161,
      "step": 1510
    },
    {
      "epoch": 0.9304187192118226,
      "grad_norm": 0.7031789004815949,
      "learning_rate": 2.9375710206517747e-07,
      "loss": 0.8723,
      "step": 1511
    },
    {
      "epoch": 0.9310344827586207,
      "grad_norm": 0.721161449367227,
      "learning_rate": 2.886059112283468e-07,
      "loss": 0.9444,
      "step": 1512
    },
    {
      "epoch": 0.9316502463054187,
      "grad_norm": 0.6961343746994174,
      "learning_rate": 2.8349962394641604e-07,
      "loss": 0.9329,
      "step": 1513
    },
    {
      "epoch": 0.9322660098522167,
      "grad_norm": 0.6760304716348517,
      "learning_rate": 2.784382638298422e-07,
      "loss": 0.9431,
      "step": 1514
    },
    {
      "epoch": 0.9328817733990148,
      "grad_norm": 0.721971100849572,
      "learning_rate": 2.7342185428134604e-07,
      "loss": 0.8394,
      "step": 1515
    },
    {
      "epoch": 0.9334975369458128,
      "grad_norm": 0.6948058761420016,
      "learning_rate": 2.6845041849581034e-07,
      "loss": 0.9047,
      "step": 1516
    },
    {
      "epoch": 0.9341133004926109,
      "grad_norm": 0.7185599432627174,
      "learning_rate": 2.6352397946016293e-07,
      "loss": 0.8828,
      "step": 1517
    },
    {
      "epoch": 0.9347290640394089,
      "grad_norm": 0.6640589159569891,
      "learning_rate": 2.5864255995327937e-07,
      "loss": 0.8441,
      "step": 1518
    },
    {
      "epoch": 0.9353448275862069,
      "grad_norm": 0.6710651231946017,
      "learning_rate": 2.5380618254587376e-07,
      "loss": 0.881,
      "step": 1519
    },
    {
      "epoch": 0.9359605911330049,
      "grad_norm": 0.702607456704797,
      "learning_rate": 2.4901486960039025e-07,
      "loss": 0.8521,
      "step": 1520
    },
    {
      "epoch": 0.9365763546798029,
      "grad_norm": 0.6872356572873789,
      "learning_rate": 2.442686432709096e-07,
      "loss": 0.9035,
      "step": 1521
    },
    {
      "epoch": 0.937192118226601,
      "grad_norm": 0.7092189868566748,
      "learning_rate": 2.395675255030383e-07,
      "loss": 0.8858,
      "step": 1522
    },
    {
      "epoch": 0.937807881773399,
      "grad_norm": 0.7286347776073536,
      "learning_rate": 2.349115380338096e-07,
      "loss": 0.8933,
      "step": 1523
    },
    {
      "epoch": 0.9384236453201971,
      "grad_norm": 0.6621409199054071,
      "learning_rate": 2.3030070239158375e-07,
      "loss": 0.8825,
      "step": 1524
    },
    {
      "epoch": 0.9390394088669951,
      "grad_norm": 0.7262631266929793,
      "learning_rate": 2.257350398959457e-07,
      "loss": 0.8497,
      "step": 1525
    },
    {
      "epoch": 0.9396551724137931,
      "grad_norm": 0.7049920827290415,
      "learning_rate": 2.2121457165761416e-07,
      "loss": 0.909,
      "step": 1526
    },
    {
      "epoch": 0.9402709359605911,
      "grad_norm": 0.7015014845005071,
      "learning_rate": 2.1673931857833285e-07,
      "loss": 0.8616,
      "step": 1527
    },
    {
      "epoch": 0.9408866995073891,
      "grad_norm": 0.6785184870772237,
      "learning_rate": 2.1230930135078377e-07,
      "loss": 0.8777,
      "step": 1528
    },
    {
      "epoch": 0.9415024630541872,
      "grad_norm": 0.6734031817376145,
      "learning_rate": 2.0792454045848398e-07,
      "loss": 0.8798,
      "step": 1529
    },
    {
      "epoch": 0.9421182266009852,
      "grad_norm": 0.6892723295598365,
      "learning_rate": 2.035850561756969e-07,
      "loss": 0.8608,
      "step": 1530
    },
    {
      "epoch": 0.9427339901477833,
      "grad_norm": 0.6669344631861204,
      "learning_rate": 1.992908685673345e-07,
      "loss": 0.8705,
      "step": 1531
    },
    {
      "epoch": 0.9433497536945813,
      "grad_norm": 0.6798326473784124,
      "learning_rate": 1.950419974888662e-07,
      "loss": 0.8704,
      "step": 1532
    },
    {
      "epoch": 0.9439655172413793,
      "grad_norm": 0.7054188549310293,
      "learning_rate": 1.9083846258622585e-07,
      "loss": 0.8622,
      "step": 1533
    },
    {
      "epoch": 0.9445812807881774,
      "grad_norm": 0.6927907384758445,
      "learning_rate": 1.8668028329572485e-07,
      "loss": 0.8697,
      "step": 1534
    },
    {
      "epoch": 0.9451970443349754,
      "grad_norm": 0.693589422819117,
      "learning_rate": 1.8256747884395577e-07,
      "loss": 0.8813,
      "step": 1535
    },
    {
      "epoch": 0.9458128078817734,
      "grad_norm": 0.6593983465099507,
      "learning_rate": 1.7850006824771005e-07,
      "loss": 0.853,
      "step": 1536
    },
    {
      "epoch": 0.9464285714285714,
      "grad_norm": 0.7161520010795283,
      "learning_rate": 1.7447807031388264e-07,
      "loss": 0.9104,
      "step": 1537
    },
    {
      "epoch": 0.9470443349753694,
      "grad_norm": 0.7146587054904631,
      "learning_rate": 1.7050150363939311e-07,
      "loss": 0.9206,
      "step": 1538
    },
    {
      "epoch": 0.9476600985221675,
      "grad_norm": 0.7027733452018162,
      "learning_rate": 1.6657038661109791e-07,
      "loss": 0.8924,
      "step": 1539
    },
    {
      "epoch": 0.9482758620689655,
      "grad_norm": 0.7006351275025492,
      "learning_rate": 1.6268473740569723e-07,
      "loss": 0.9013,
      "step": 1540
    },
    {
      "epoch": 0.9488916256157636,
      "grad_norm": 0.6888828921543767,
      "learning_rate": 1.5884457398966046e-07,
      "loss": 0.8926,
      "step": 1541
    },
    {
      "epoch": 0.9495073891625616,
      "grad_norm": 0.6393524161828175,
      "learning_rate": 1.5504991411914083e-07,
      "loss": 0.8596,
      "step": 1542
    },
    {
      "epoch": 0.9501231527093597,
      "grad_norm": 0.727362627408383,
      "learning_rate": 1.5130077533988874e-07,
      "loss": 0.9267,
      "step": 1543
    },
    {
      "epoch": 0.9507389162561576,
      "grad_norm": 0.6662474702577719,
      "learning_rate": 1.4759717498717962e-07,
      "loss": 0.8332,
      "step": 1544
    },
    {
      "epoch": 0.9513546798029556,
      "grad_norm": 0.7018435385867428,
      "learning_rate": 1.4393913018572182e-07,
      "loss": 0.9332,
      "step": 1545
    },
    {
      "epoch": 0.9519704433497537,
      "grad_norm": 0.7278518815716029,
      "learning_rate": 1.4032665784958877e-07,
      "loss": 0.871,
      "step": 1546
    },
    {
      "epoch": 0.9525862068965517,
      "grad_norm": 0.6734817923630341,
      "learning_rate": 1.3675977468213365e-07,
      "loss": 0.8416,
      "step": 1547
    },
    {
      "epoch": 0.9532019704433498,
      "grad_norm": 0.7297014054654807,
      "learning_rate": 1.3323849717591376e-07,
      "loss": 0.9209,
      "step": 1548
    },
    {
      "epoch": 0.9538177339901478,
      "grad_norm": 0.694768261560907,
      "learning_rate": 1.2976284161261843e-07,
      "loss": 0.9443,
      "step": 1549
    },
    {
      "epoch": 0.9544334975369458,
      "grad_norm": 0.7461907609168345,
      "learning_rate": 1.2633282406298576e-07,
      "loss": 0.8861,
      "step": 1550
    },
    {
      "epoch": 0.9550492610837439,
      "grad_norm": 0.7117613900462639,
      "learning_rate": 1.229484603867348e-07,
      "loss": 0.8909,
      "step": 1551
    },
    {
      "epoch": 0.9556650246305419,
      "grad_norm": 0.6971181920108205,
      "learning_rate": 1.196097662324902e-07,
      "loss": 0.8397,
      "step": 1552
    },
    {
      "epoch": 0.9562807881773399,
      "grad_norm": 0.7143168700770987,
      "learning_rate": 1.1631675703771106e-07,
      "loss": 0.8918,
      "step": 1553
    },
    {
      "epoch": 0.9568965517241379,
      "grad_norm": 0.7692519095009186,
      "learning_rate": 1.1306944802861652e-07,
      "loss": 0.905,
      "step": 1554
    },
    {
      "epoch": 0.9575123152709359,
      "grad_norm": 0.727500502284882,
      "learning_rate": 1.0986785422011593e-07,
      "loss": 0.8912,
      "step": 1555
    },
    {
      "epoch": 0.958128078817734,
      "grad_norm": 0.6716895966813067,
      "learning_rate": 1.0671199041574542e-07,
      "loss": 0.9068,
      "step": 1556
    },
    {
      "epoch": 0.958743842364532,
      "grad_norm": 0.699308252538153,
      "learning_rate": 1.0360187120758813e-07,
      "loss": 0.8228,
      "step": 1557
    },
    {
      "epoch": 0.9593596059113301,
      "grad_norm": 0.6597662319126755,
      "learning_rate": 1.0053751097621855e-07,
      "loss": 0.8261,
      "step": 1558
    },
    {
      "epoch": 0.9599753694581281,
      "grad_norm": 0.687168640557222,
      "learning_rate": 9.751892389063045e-08,
      "loss": 0.9401,
      "step": 1559
    },
    {
      "epoch": 0.9605911330049262,
      "grad_norm": 0.7692909433658529,
      "learning_rate": 9.454612390816686e-08,
      "loss": 0.9055,
      "step": 1560
    },
    {
      "epoch": 0.9612068965517241,
      "grad_norm": 0.6800494280144185,
      "learning_rate": 9.161912477446689e-08,
      "loss": 0.8702,
      "step": 1561
    },
    {
      "epoch": 0.9618226600985221,
      "grad_norm": 0.6627212691700862,
      "learning_rate": 8.873794002338898e-08,
      "loss": 0.869,
      "step": 1562
    },
    {
      "epoch": 0.9624384236453202,
      "grad_norm": 0.6902969230966061,
      "learning_rate": 8.590258297696108e-08,
      "loss": 0.8704,
      "step": 1563
    },
    {
      "epoch": 0.9630541871921182,
      "grad_norm": 0.6870197136616305,
      "learning_rate": 8.311306674531062e-08,
      "loss": 0.9072,
      "step": 1564
    },
    {
      "epoch": 0.9636699507389163,
      "grad_norm": 0.6960189724137371,
      "learning_rate": 8.036940422660345e-08,
      "loss": 0.8793,
      "step": 1565
    },
    {
      "epoch": 0.9642857142857143,
      "grad_norm": 0.6746482088388054,
      "learning_rate": 7.76716081069917e-08,
      "loss": 0.8356,
      "step": 1566
    },
    {
      "epoch": 0.9649014778325123,
      "grad_norm": 0.713348066742654,
      "learning_rate": 7.501969086054717e-08,
      "loss": 0.9219,
      "step": 1567
    },
    {
      "epoch": 0.9655172413793104,
      "grad_norm": 0.7185118355729003,
      "learning_rate": 7.241366474920796e-08,
      "loss": 0.8704,
      "step": 1568
    },
    {
      "epoch": 0.9661330049261084,
      "grad_norm": 0.7705198627025334,
      "learning_rate": 6.985354182271975e-08,
      "loss": 0.8656,
      "step": 1569
    },
    {
      "epoch": 0.9667487684729064,
      "grad_norm": 0.6594105600402287,
      "learning_rate": 6.733933391858238e-08,
      "loss": 0.9008,
      "step": 1570
    },
    {
      "epoch": 0.9673645320197044,
      "grad_norm": 0.7238235153231951,
      "learning_rate": 6.487105266199333e-08,
      "loss": 0.882,
      "step": 1571
    },
    {
      "epoch": 0.9679802955665024,
      "grad_norm": 0.7132127686034845,
      "learning_rate": 6.24487094657944e-08,
      "loss": 0.82,
      "step": 1572
    },
    {
      "epoch": 0.9685960591133005,
      "grad_norm": 0.6955113108944352,
      "learning_rate": 6.007231553041837e-08,
      "loss": 0.952,
      "step": 1573
    },
    {
      "epoch": 0.9692118226600985,
      "grad_norm": 0.7459540070127851,
      "learning_rate": 5.774188184383911e-08,
      "loss": 0.9196,
      "step": 1574
    },
    {
      "epoch": 0.9698275862068966,
      "grad_norm": 0.6828373122148967,
      "learning_rate": 5.5457419181517145e-08,
      "loss": 0.8773,
      "step": 1575
    },
    {
      "epoch": 0.9704433497536946,
      "grad_norm": 0.6666318001956405,
      "learning_rate": 5.321893810635526e-08,
      "loss": 0.8663,
      "step": 1576
    },
    {
      "epoch": 0.9710591133004927,
      "grad_norm": 0.7000582766616417,
      "learning_rate": 5.102644896864184e-08,
      "loss": 0.8987,
      "step": 1577
    },
    {
      "epoch": 0.9716748768472906,
      "grad_norm": 0.7097762013954348,
      "learning_rate": 4.8879961906013184e-08,
      "loss": 0.8571,
      "step": 1578
    },
    {
      "epoch": 0.9722906403940886,
      "grad_norm": 0.7068082725032607,
      "learning_rate": 4.6779486843397946e-08,
      "loss": 0.8857,
      "step": 1579
    },
    {
      "epoch": 0.9729064039408867,
      "grad_norm": 0.7202824417674714,
      "learning_rate": 4.472503349297497e-08,
      "loss": 0.8913,
      "step": 1580
    },
    {
      "epoch": 0.9735221674876847,
      "grad_norm": 0.6933943614930186,
      "learning_rate": 4.271661135412775e-08,
      "loss": 0.8737,
      "step": 1581
    },
    {
      "epoch": 0.9741379310344828,
      "grad_norm": 0.6705760439434347,
      "learning_rate": 4.075422971340115e-08,
      "loss": 0.8232,
      "step": 1582
    },
    {
      "epoch": 0.9747536945812808,
      "grad_norm": 0.7050400634723398,
      "learning_rate": 3.8837897644457e-08,
      "loss": 0.8673,
      "step": 1583
    },
    {
      "epoch": 0.9753694581280788,
      "grad_norm": 0.685670666616804,
      "learning_rate": 3.6967624008035216e-08,
      "loss": 0.8886,
      "step": 1584
    },
    {
      "epoch": 0.9759852216748769,
      "grad_norm": 0.6824471684547423,
      "learning_rate": 3.5143417451907195e-08,
      "loss": 0.8312,
      "step": 1585
    },
    {
      "epoch": 0.9766009852216748,
      "grad_norm": 0.6517699814165081,
      "learning_rate": 3.3365286410842465e-08,
      "loss": 0.8679,
      "step": 1586
    },
    {
      "epoch": 0.9772167487684729,
      "grad_norm": 0.6981734406526013,
      "learning_rate": 3.163323910656546e-08,
      "loss": 0.9403,
      "step": 1587
    },
    {
      "epoch": 0.9778325123152709,
      "grad_norm": 0.6970696294310844,
      "learning_rate": 2.994728354771659e-08,
      "loss": 0.9131,
      "step": 1588
    },
    {
      "epoch": 0.978448275862069,
      "grad_norm": 0.7056345409991207,
      "learning_rate": 2.8307427529822295e-08,
      "loss": 0.8468,
      "step": 1589
    },
    {
      "epoch": 0.979064039408867,
      "grad_norm": 0.6860295323730407,
      "learning_rate": 2.671367863524732e-08,
      "loss": 0.9027,
      "step": 1590
    },
    {
      "epoch": 0.979679802955665,
      "grad_norm": 0.7410440395196958,
      "learning_rate": 2.5166044233172483e-08,
      "loss": 0.8856,
      "step": 1591
    },
    {
      "epoch": 0.9802955665024631,
      "grad_norm": 0.6886837575543354,
      "learning_rate": 2.366453147954917e-08,
      "loss": 0.848,
      "step": 1592
    },
    {
      "epoch": 0.9809113300492611,
      "grad_norm": 0.7182335819028924,
      "learning_rate": 2.220914731707491e-08,
      "loss": 0.8723,
      "step": 1593
    },
    {
      "epoch": 0.9815270935960592,
      "grad_norm": 0.6610002529109936,
      "learning_rate": 2.0799898475156732e-08,
      "loss": 0.8572,
      "step": 1594
    },
    {
      "epoch": 0.9821428571428571,
      "grad_norm": 0.6425156620391559,
      "learning_rate": 1.943679146988009e-08,
      "loss": 0.8567,
      "step": 1595
    },
    {
      "epoch": 0.9827586206896551,
      "grad_norm": 0.7063392697335805,
      "learning_rate": 1.8119832603982202e-08,
      "loss": 0.9266,
      "step": 1596
    },
    {
      "epoch": 0.9833743842364532,
      "grad_norm": 0.6750495620890997,
      "learning_rate": 1.6849027966816535e-08,
      "loss": 0.8926,
      "step": 1597
    },
    {
      "epoch": 0.9839901477832512,
      "grad_norm": 0.6905819839385934,
      "learning_rate": 1.5624383434333923e-08,
      "loss": 0.8758,
      "step": 1598
    },
    {
      "epoch": 0.9846059113300493,
      "grad_norm": 0.6968298829346535,
      "learning_rate": 1.444590466904483e-08,
      "loss": 0.8631,
      "step": 1599
    },
    {
      "epoch": 0.9852216748768473,
      "grad_norm": 0.6970086076232006,
      "learning_rate": 1.3313597120002686e-08,
      "loss": 0.9246,
      "step": 1600
    },
    {
      "epoch": 0.9852216748768473,
      "eval_loss": 0.8905929923057556,
      "eval_runtime": 523.0979,
      "eval_samples_per_second": 44.179,
      "eval_steps_per_second": 0.692,
      "step": 1600
    },
    {
      "epoch": 0.9858374384236454,
      "grad_norm": 0.7044541933434698,
      "learning_rate": 1.2227466022770584e-08,
      "loss": 0.8713,
      "step": 1601
    },
    {
      "epoch": 0.9864532019704434,
      "grad_norm": 0.7211895232951646,
      "learning_rate": 1.118751639940352e-08,
      "loss": 0.874,
      "step": 1602
    },
    {
      "epoch": 0.9870689655172413,
      "grad_norm": 0.6859497427720389,
      "learning_rate": 1.0193753058420631e-08,
      "loss": 0.919,
      "step": 1603
    },
    {
      "epoch": 0.9876847290640394,
      "grad_norm": 0.7037250343035977,
      "learning_rate": 9.246180594788546e-09,
      "loss": 0.8661,
      "step": 1604
    },
    {
      "epoch": 0.9883004926108374,
      "grad_norm": 0.66426229433972,
      "learning_rate": 8.34480338989141e-09,
      "loss": 0.8885,
      "step": 1605
    },
    {
      "epoch": 0.9889162561576355,
      "grad_norm": 0.7152692117367438,
      "learning_rate": 7.48962561151978e-09,
      "loss": 0.8795,
      "step": 1606
    },
    {
      "epoch": 0.9895320197044335,
      "grad_norm": 0.6826932370204328,
      "learning_rate": 6.68065121384398e-09,
      "loss": 0.8996,
      "step": 1607
    },
    {
      "epoch": 0.9901477832512315,
      "grad_norm": 0.7187080719818589,
      "learning_rate": 5.917883937401892e-09,
      "loss": 0.8929,
      "step": 1608
    },
    {
      "epoch": 0.9907635467980296,
      "grad_norm": 0.6839597096528248,
      "learning_rate": 5.201327309077853e-09,
      "loss": 0.9307,
      "step": 1609
    },
    {
      "epoch": 0.9913793103448276,
      "grad_norm": 0.6468850237270645,
      "learning_rate": 4.530984642087122e-09,
      "loss": 0.853,
      "step": 1610
    },
    {
      "epoch": 0.9919950738916257,
      "grad_norm": 0.7339171586736611,
      "learning_rate": 3.906859035960331e-09,
      "loss": 0.8851,
      "step": 1611
    },
    {
      "epoch": 0.9926108374384236,
      "grad_norm": 0.7000088311520395,
      "learning_rate": 3.328953376530164e-09,
      "loss": 0.8835,
      "step": 1612
    },
    {
      "epoch": 0.9932266009852216,
      "grad_norm": 0.6747988959503071,
      "learning_rate": 2.797270335916924e-09,
      "loss": 0.8757,
      "step": 1613
    },
    {
      "epoch": 0.9938423645320197,
      "grad_norm": 0.7189675115443626,
      "learning_rate": 2.3118123725196506e-09,
      "loss": 0.9226,
      "step": 1614
    },
    {
      "epoch": 0.9944581280788177,
      "grad_norm": 0.6634281215596808,
      "learning_rate": 1.872581730997247e-09,
      "loss": 0.8911,
      "step": 1615
    },
    {
      "epoch": 0.9950738916256158,
      "grad_norm": 0.6867852712876448,
      "learning_rate": 1.4795804422651494e-09,
      "loss": 0.8144,
      "step": 1616
    },
    {
      "epoch": 0.9956896551724138,
      "grad_norm": 0.7023732304198141,
      "learning_rate": 1.1328103234831133e-09,
      "loss": 0.8811,
      "step": 1617
    },
    {
      "epoch": 0.9963054187192119,
      "grad_norm": 0.7035953919786208,
      "learning_rate": 8.322729780474437e-10,
      "loss": 0.9123,
      "step": 1618
    },
    {
      "epoch": 0.9969211822660099,
      "grad_norm": 0.6655030979524406,
      "learning_rate": 5.779697955843322e-10,
      "loss": 0.8845,
      "step": 1619
    },
    {
      "epoch": 0.9975369458128078,
      "grad_norm": 0.747441606826873,
      "learning_rate": 3.699019519387559e-10,
      "loss": 0.8795,
      "step": 1620
    },
    {
      "epoch": 0.9981527093596059,
      "grad_norm": 0.6945129775662154,
      "learning_rate": 2.0807040917669719e-10,
      "loss": 0.8602,
      "step": 1621
    },
    {
      "epoch": 0.9987684729064039,
      "grad_norm": 0.7244212750611351,
      "learning_rate": 9.247591557404179e-11,
      "loss": 0.8385,
      "step": 1622
    },
    {
      "epoch": 0.999384236453202,
      "grad_norm": 0.7077357669801623,
      "learning_rate": 2.311900561768887e-11,
      "loss": 0.8834,
      "step": 1623
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6684415300974857,
      "learning_rate": 0.0,
      "loss": 0.8729,
      "step": 1624
    },
    {
      "epoch": 1.0,
      "step": 1624,
      "total_flos": 702000445095936.0,
      "train_loss": 0.9030163899944921,
      "train_runtime": 23225.8285,
      "train_samples_per_second": 8.95,
      "train_steps_per_second": 0.07
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1624,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 702000445095936.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}