{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9990426041168023,
  "eval_steps": 500,
  "global_step": 348,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0057443752991862135,
      "grad_norm": 5.363491058349609,
      "learning_rate": 2.8571428571428575e-07,
      "loss": 0.8587,
      "step": 1
    },
    {
      "epoch": 0.011488750598372427,
      "grad_norm": 5.751570701599121,
      "learning_rate": 5.714285714285715e-07,
      "loss": 0.8936,
      "step": 2
    },
    {
      "epoch": 0.01723312589755864,
      "grad_norm": 5.681423187255859,
      "learning_rate": 8.571428571428572e-07,
      "loss": 0.9174,
      "step": 3
    },
    {
      "epoch": 0.022977501196744854,
      "grad_norm": 5.466259956359863,
      "learning_rate": 1.142857142857143e-06,
      "loss": 0.8825,
      "step": 4
    },
    {
      "epoch": 0.028721876495931067,
      "grad_norm": 5.4661688804626465,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.8794,
      "step": 5
    },
    {
      "epoch": 0.03446625179511728,
      "grad_norm": 5.406000137329102,
      "learning_rate": 1.7142857142857145e-06,
      "loss": 0.8566,
      "step": 6
    },
    {
      "epoch": 0.040210627094303494,
      "grad_norm": 4.255201816558838,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.8527,
      "step": 7
    },
    {
      "epoch": 0.04595500239348971,
      "grad_norm": 3.8569490909576416,
      "learning_rate": 2.285714285714286e-06,
      "loss": 0.8273,
      "step": 8
    },
    {
      "epoch": 0.05169937769267592,
      "grad_norm": 2.5158278942108154,
      "learning_rate": 2.571428571428571e-06,
      "loss": 0.7763,
      "step": 9
    },
    {
      "epoch": 0.057443752991862135,
      "grad_norm": 2.4135913848876953,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.832,
      "step": 10
    },
    {
      "epoch": 0.06318812829104835,
      "grad_norm": 2.236311674118042,
      "learning_rate": 3.142857142857143e-06,
      "loss": 0.8195,
      "step": 11
    },
    {
      "epoch": 0.06893250359023456,
      "grad_norm": 2.0858547687530518,
      "learning_rate": 3.428571428571429e-06,
      "loss": 0.7293,
      "step": 12
    },
    {
      "epoch": 0.07467687888942078,
      "grad_norm": 3.858452320098877,
      "learning_rate": 3.7142857142857146e-06,
      "loss": 0.7993,
      "step": 13
    },
    {
      "epoch": 0.08042125418860699,
      "grad_norm": 3.8865373134613037,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.7821,
      "step": 14
    },
    {
      "epoch": 0.0861656294877932,
      "grad_norm": 3.8490350246429443,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.7953,
      "step": 15
    },
    {
      "epoch": 0.09191000478697942,
      "grad_norm": 3.2923967838287354,
      "learning_rate": 4.571428571428572e-06,
      "loss": 0.7904,
      "step": 16
    },
    {
      "epoch": 0.09765438008616563,
      "grad_norm": 2.2245757579803467,
      "learning_rate": 4.857142857142858e-06,
      "loss": 0.7397,
      "step": 17
    },
    {
      "epoch": 0.10339875538535184,
      "grad_norm": 1.9646612405776978,
      "learning_rate": 5.142857142857142e-06,
      "loss": 0.7405,
      "step": 18
    },
    {
      "epoch": 0.10914313068453806,
      "grad_norm": 1.5325367450714111,
      "learning_rate": 5.428571428571429e-06,
      "loss": 0.6697,
      "step": 19
    },
    {
      "epoch": 0.11488750598372427,
      "grad_norm": 1.4567933082580566,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.6977,
      "step": 20
    },
    {
      "epoch": 0.12063188128291048,
      "grad_norm": 1.4041451215744019,
      "learning_rate": 6e-06,
      "loss": 0.6796,
      "step": 21
    },
    {
      "epoch": 0.1263762565820967,
      "grad_norm": 1.1699495315551758,
      "learning_rate": 6.285714285714286e-06,
      "loss": 0.674,
      "step": 22
    },
    {
      "epoch": 0.13212063188128292,
      "grad_norm": 1.0591332912445068,
      "learning_rate": 6.571428571428572e-06,
      "loss": 0.6336,
      "step": 23
    },
    {
      "epoch": 0.13786500718046912,
      "grad_norm": 1.2895734310150146,
      "learning_rate": 6.857142857142858e-06,
      "loss": 0.6701,
      "step": 24
    },
    {
      "epoch": 0.14360938247965535,
      "grad_norm": 1.0462803840637207,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.6557,
      "step": 25
    },
    {
      "epoch": 0.14935375777884155,
      "grad_norm": 1.1943080425262451,
      "learning_rate": 7.428571428571429e-06,
      "loss": 0.6453,
      "step": 26
    },
    {
      "epoch": 0.15509813307802778,
      "grad_norm": 1.0741592645645142,
      "learning_rate": 7.714285714285716e-06,
      "loss": 0.6813,
      "step": 27
    },
    {
      "epoch": 0.16084250837721398,
      "grad_norm": 0.9222673773765564,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.6415,
      "step": 28
    },
    {
      "epoch": 0.1665868836764002,
      "grad_norm": 0.8281923532485962,
      "learning_rate": 8.285714285714287e-06,
      "loss": 0.6287,
      "step": 29
    },
    {
      "epoch": 0.1723312589755864,
      "grad_norm": 1.0101515054702759,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.6347,
      "step": 30
    },
    {
      "epoch": 0.17807563427477263,
      "grad_norm": 0.9029678702354431,
      "learning_rate": 8.857142857142858e-06,
      "loss": 0.5939,
      "step": 31
    },
    {
      "epoch": 0.18382000957395883,
      "grad_norm": 0.6645869016647339,
      "learning_rate": 9.142857142857144e-06,
      "loss": 0.5704,
      "step": 32
    },
    {
      "epoch": 0.18956438487314506,
      "grad_norm": 0.7702145576477051,
      "learning_rate": 9.42857142857143e-06,
      "loss": 0.6125,
      "step": 33
    },
    {
      "epoch": 0.19530876017233126,
      "grad_norm": 0.8352411985397339,
      "learning_rate": 9.714285714285715e-06,
      "loss": 0.6332,
      "step": 34
    },
    {
      "epoch": 0.20105313547151749,
      "grad_norm": 0.6543819308280945,
      "learning_rate": 1e-05,
      "loss": 0.6196,
      "step": 35
    },
    {
      "epoch": 0.20679751077070369,
      "grad_norm": 0.6806799173355103,
      "learning_rate": 9.999748146823376e-06,
      "loss": 0.6135,
      "step": 36
    },
    {
      "epoch": 0.2125418860698899,
      "grad_norm": 0.7468794584274292,
      "learning_rate": 9.99899261266551e-06,
      "loss": 0.5972,
      "step": 37
    },
    {
      "epoch": 0.2182862613690761,
      "grad_norm": 0.6796684861183167,
      "learning_rate": 9.997733473639876e-06,
      "loss": 0.6111,
      "step": 38
    },
    {
      "epoch": 0.22403063666826234,
      "grad_norm": 0.6619651913642883,
      "learning_rate": 9.995970856593739e-06,
      "loss": 0.6117,
      "step": 39
    },
    {
      "epoch": 0.22977501196744854,
      "grad_norm": 0.6917636394500732,
      "learning_rate": 9.993704939095376e-06,
      "loss": 0.5887,
      "step": 40
    },
    {
      "epoch": 0.23551938726663477,
      "grad_norm": 0.6591335535049438,
      "learning_rate": 9.9909359494162e-06,
      "loss": 0.631,
      "step": 41
    },
    {
      "epoch": 0.24126376256582097,
      "grad_norm": 0.7333484888076782,
      "learning_rate": 9.987664166507749e-06,
      "loss": 0.5682,
      "step": 42
    },
    {
      "epoch": 0.2470081378650072,
      "grad_norm": 0.7811906933784485,
      "learning_rate": 9.983889919973586e-06,
      "loss": 0.6056,
      "step": 43
    },
    {
      "epoch": 0.2527525131641934,
      "grad_norm": 0.691569447517395,
      "learning_rate": 9.979613590036108e-06,
      "loss": 0.6209,
      "step": 44
    },
    {
      "epoch": 0.2584968884633796,
      "grad_norm": 0.684109091758728,
      "learning_rate": 9.974835607498224e-06,
      "loss": 0.605,
      "step": 45
    },
    {
      "epoch": 0.26424126376256585,
      "grad_norm": 0.7513927817344666,
      "learning_rate": 9.969556453699966e-06,
      "loss": 0.556,
      "step": 46
    },
    {
      "epoch": 0.26998563906175205,
      "grad_norm": 0.7860096096992493,
      "learning_rate": 9.963776660469996e-06,
      "loss": 0.5852,
      "step": 47
    },
    {
      "epoch": 0.27573001436093825,
      "grad_norm": 0.555550754070282,
      "learning_rate": 9.957496810072027e-06,
      "loss": 0.5736,
      "step": 48
    },
    {
      "epoch": 0.28147438966012445,
      "grad_norm": 0.8144235014915466,
      "learning_rate": 9.95071753514617e-06,
      "loss": 0.6078,
      "step": 49
    },
    {
      "epoch": 0.2872187649593107,
      "grad_norm": 0.6273301839828491,
      "learning_rate": 9.943439518645193e-06,
      "loss": 0.5638,
      "step": 50
    },
    {
      "epoch": 0.2929631402584969,
      "grad_norm": 0.5855135321617126,
      "learning_rate": 9.935663493765726e-06,
      "loss": 0.5501,
      "step": 51
    },
    {
      "epoch": 0.2987075155576831,
      "grad_norm": 0.7060447931289673,
      "learning_rate": 9.9273902438744e-06,
      "loss": 0.596,
      "step": 52
    },
    {
      "epoch": 0.3044518908568693,
      "grad_norm": 0.6160895228385925,
      "learning_rate": 9.918620602428916e-06,
      "loss": 0.568,
      "step": 53
    },
    {
      "epoch": 0.31019626615605556,
      "grad_norm": 0.6151447296142578,
      "learning_rate": 9.909355452894098e-06,
      "loss": 0.5581,
      "step": 54
    },
    {
      "epoch": 0.31594064145524176,
      "grad_norm": 0.5906559824943542,
      "learning_rate": 9.899595728652883e-06,
      "loss": 0.5628,
      "step": 55
    },
    {
      "epoch": 0.32168501675442795,
      "grad_norm": 0.6064203977584839,
      "learning_rate": 9.889342412912296e-06,
      "loss": 0.5708,
      "step": 56
    },
    {
      "epoch": 0.32742939205361415,
      "grad_norm": 0.645464301109314,
      "learning_rate": 9.878596538604388e-06,
      "loss": 0.5792,
      "step": 57
    },
    {
      "epoch": 0.3331737673528004,
      "grad_norm": 0.6362553238868713,
      "learning_rate": 9.867359188282193e-06,
      "loss": 0.5733,
      "step": 58
    },
    {
      "epoch": 0.3389181426519866,
      "grad_norm": 0.680167019367218,
      "learning_rate": 9.855631494010661e-06,
      "loss": 0.5473,
      "step": 59
    },
    {
      "epoch": 0.3446625179511728,
      "grad_norm": 0.6046214699745178,
      "learning_rate": 9.843414637252615e-06,
      "loss": 0.5568,
      "step": 60
    },
    {
      "epoch": 0.350406893250359,
      "grad_norm": 0.6066192388534546,
      "learning_rate": 9.830709848749727e-06,
      "loss": 0.5732,
      "step": 61
    },
    {
      "epoch": 0.35615126854954526,
      "grad_norm": 0.5794113278388977,
      "learning_rate": 9.817518408398536e-06,
      "loss": 0.5825,
      "step": 62
    },
    {
      "epoch": 0.36189564384873146,
      "grad_norm": 0.5537358522415161,
      "learning_rate": 9.803841645121505e-06,
      "loss": 0.5664,
      "step": 63
    },
    {
      "epoch": 0.36764001914791766,
      "grad_norm": 0.6195133328437805,
      "learning_rate": 9.78968093673314e-06,
      "loss": 0.55,
      "step": 64
    },
    {
      "epoch": 0.37338439444710386,
      "grad_norm": 0.5777291655540466,
      "learning_rate": 9.775037709801206e-06,
      "loss": 0.572,
      "step": 65
    },
    {
      "epoch": 0.3791287697462901,
      "grad_norm": 0.6024558544158936,
      "learning_rate": 9.759913439502982e-06,
      "loss": 0.5657,
      "step": 66
    },
    {
      "epoch": 0.3848731450454763,
      "grad_norm": 0.5568858981132507,
      "learning_rate": 9.74430964947668e-06,
      "loss": 0.5261,
      "step": 67
    },
    {
      "epoch": 0.3906175203446625,
      "grad_norm": 0.5718939304351807,
      "learning_rate": 9.728227911667934e-06,
      "loss": 0.5494,
      "step": 68
    },
    {
      "epoch": 0.3963618956438487,
      "grad_norm": 0.5540202856063843,
      "learning_rate": 9.711669846171443e-06,
      "loss": 0.5771,
      "step": 69
    },
    {
      "epoch": 0.40210627094303497,
      "grad_norm": 0.5410825610160828,
      "learning_rate": 9.694637121067764e-06,
      "loss": 0.501,
      "step": 70
    },
    {
      "epoch": 0.40785064624222117,
      "grad_norm": 0.6998445987701416,
      "learning_rate": 9.677131452255272e-06,
      "loss": 0.555,
      "step": 71
    },
    {
      "epoch": 0.41359502154140737,
      "grad_norm": 0.5608072876930237,
      "learning_rate": 9.659154603277283e-06,
      "loss": 0.5344,
      "step": 72
    },
    {
      "epoch": 0.41933939684059357,
      "grad_norm": 0.5644481182098389,
      "learning_rate": 9.640708385144403e-06,
      "loss": 0.5547,
      "step": 73
    },
    {
      "epoch": 0.4250837721397798,
      "grad_norm": 0.643834114074707,
      "learning_rate": 9.62179465615209e-06,
      "loss": 0.5578,
      "step": 74
    },
    {
      "epoch": 0.430828147438966,
      "grad_norm": 0.6082519888877869,
      "learning_rate": 9.602415321693434e-06,
      "loss": 0.5687,
      "step": 75
    },
    {
      "epoch": 0.4365725227381522,
      "grad_norm": 0.4994487762451172,
      "learning_rate": 9.582572334067213e-06,
      "loss": 0.5447,
      "step": 76
    },
    {
      "epoch": 0.4423168980373384,
      "grad_norm": 0.52949458360672,
      "learning_rate": 9.562267692281212e-06,
      "loss": 0.5713,
      "step": 77
    },
    {
      "epoch": 0.4480612733365247,
      "grad_norm": 0.5970189571380615,
      "learning_rate": 9.541503441850844e-06,
      "loss": 0.5486,
      "step": 78
    },
    {
      "epoch": 0.4538056486357109,
      "grad_norm": 0.5419505834579468,
      "learning_rate": 9.520281674593084e-06,
      "loss": 0.5577,
      "step": 79
    },
    {
      "epoch": 0.4595500239348971,
      "grad_norm": 0.5340558886528015,
      "learning_rate": 9.498604528415731e-06,
      "loss": 0.5397,
      "step": 80
    },
    {
      "epoch": 0.4652943992340833,
      "grad_norm": 0.5403291583061218,
      "learning_rate": 9.476474187102033e-06,
      "loss": 0.5438,
      "step": 81
    },
    {
      "epoch": 0.47103877453326953,
      "grad_norm": 0.6180256605148315,
      "learning_rate": 9.453892880090696e-06,
      "loss": 0.5624,
      "step": 82
    },
    {
      "epoch": 0.47678314983245573,
      "grad_norm": 0.5336986780166626,
      "learning_rate": 9.430862882251279e-06,
      "loss": 0.5473,
      "step": 83
    },
    {
      "epoch": 0.48252752513164193,
      "grad_norm": 0.6668586134910583,
      "learning_rate": 9.40738651365503e-06,
      "loss": 0.5757,
      "step": 84
    },
    {
      "epoch": 0.48827190043082813,
      "grad_norm": 0.6107608675956726,
      "learning_rate": 9.38346613934115e-06,
      "loss": 0.5335,
      "step": 85
    },
    {
      "epoch": 0.4940162757300144,
      "grad_norm": 0.5242419838905334,
      "learning_rate": 9.359104169078541e-06,
      "loss": 0.5509,
      "step": 86
    },
    {
      "epoch": 0.4997606510292006,
      "grad_norm": 0.6356630921363831,
      "learning_rate": 9.334303057123044e-06,
      "loss": 0.5676,
      "step": 87
    },
    {
      "epoch": 0.5055050263283868,
      "grad_norm": 0.533847987651825,
      "learning_rate": 9.309065301970193e-06,
      "loss": 0.5515,
      "step": 88
    },
    {
      "epoch": 0.511249401627573,
      "grad_norm": 0.5083225965499878,
      "learning_rate": 9.283393446103506e-06,
      "loss": 0.5474,
      "step": 89
    },
    {
      "epoch": 0.5169937769267592,
      "grad_norm": 0.5779216885566711,
      "learning_rate": 9.257290075738365e-06,
      "loss": 0.5574,
      "step": 90
    },
    {
      "epoch": 0.5227381522259454,
      "grad_norm": 0.6124880313873291,
      "learning_rate": 9.23075782056147e-06,
      "loss": 0.535,
      "step": 91
    },
    {
      "epoch": 0.5284825275251317,
      "grad_norm": 0.598517119884491,
      "learning_rate": 9.20379935346592e-06,
      "loss": 0.5409,
      "step": 92
    },
    {
      "epoch": 0.5342269028243178,
      "grad_norm": 0.492055207490921,
      "learning_rate": 9.176417390281944e-06,
      "loss": 0.5293,
      "step": 93
    },
    {
      "epoch": 0.5399712781235041,
      "grad_norm": 0.4873235523700714,
      "learning_rate": 9.148614689503307e-06,
      "loss": 0.5585,
      "step": 94
    },
    {
      "epoch": 0.5457156534226902,
      "grad_norm": 0.5771122574806213,
      "learning_rate": 9.120394052009412e-06,
      "loss": 0.5464,
      "step": 95
    },
    {
      "epoch": 0.5514600287218765,
      "grad_norm": 0.5583028793334961,
      "learning_rate": 9.091758320783139e-06,
      "loss": 0.5603,
      "step": 96
    },
    {
      "epoch": 0.5572044040210627,
      "grad_norm": 0.5799354910850525,
      "learning_rate": 9.062710380624439e-06,
      "loss": 0.5307,
      "step": 97
    },
    {
      "epoch": 0.5629487793202489,
      "grad_norm": 0.5598204731941223,
      "learning_rate": 9.033253157859715e-06,
      "loss": 0.5733,
      "step": 98
    },
    {
      "epoch": 0.5686931546194351,
      "grad_norm": 0.5341041684150696,
      "learning_rate": 9.003389620047012e-06,
      "loss": 0.5288,
      "step": 99
    },
    {
      "epoch": 0.5744375299186214,
      "grad_norm": 0.5493537187576294,
      "learning_rate": 8.973122775677078e-06,
      "loss": 0.5322,
      "step": 100
    },
    {
      "epoch": 0.5801819052178075,
      "grad_norm": 0.7178077101707458,
      "learning_rate": 8.942455673870278e-06,
      "loss": 0.5638,
      "step": 101
    },
    {
      "epoch": 0.5859262805169938,
      "grad_norm": 0.5725718140602112,
      "learning_rate": 8.91139140406941e-06,
      "loss": 0.5603,
      "step": 102
    },
    {
      "epoch": 0.59167065581618,
      "grad_norm": 0.5904880166053772,
      "learning_rate": 8.879933095728485e-06,
      "loss": 0.5454,
      "step": 103
    },
    {
      "epoch": 0.5974150311153662,
      "grad_norm": 0.5282031893730164,
      "learning_rate": 8.848083917997463e-06,
      "loss": 0.5287,
      "step": 104
    },
    {
      "epoch": 0.6031594064145525,
      "grad_norm": 0.5110641121864319,
      "learning_rate": 8.815847079402972e-06,
      "loss": 0.5301,
      "step": 105
    },
    {
      "epoch": 0.6089037817137386,
      "grad_norm": 0.7060251235961914,
      "learning_rate": 8.783225827525098e-06,
      "loss": 0.5529,
      "step": 106
    },
    {
      "epoch": 0.6146481570129249,
      "grad_norm": 0.5038994550704956,
      "learning_rate": 8.750223448670204e-06,
      "loss": 0.5464,
      "step": 107
    },
    {
      "epoch": 0.6203925323121111,
      "grad_norm": 0.6462810039520264,
      "learning_rate": 8.716843267539868e-06,
      "loss": 0.5634,
      "step": 108
    },
    {
      "epoch": 0.6261369076112973,
      "grad_norm": 0.5832483768463135,
      "learning_rate": 8.683088646895955e-06,
      "loss": 0.5102,
      "step": 109
    },
    {
      "epoch": 0.6318812829104835,
      "grad_norm": 0.5752907991409302,
      "learning_rate": 8.648962987221837e-06,
      "loss": 0.5767,
      "step": 110
    },
    {
      "epoch": 0.6376256582096697,
      "grad_norm": 0.714330792427063,
      "learning_rate": 8.614469726379833e-06,
      "loss": 0.5567,
      "step": 111
    },
    {
      "epoch": 0.6433700335088559,
      "grad_norm": 0.5703815817832947,
      "learning_rate": 8.579612339264867e-06,
      "loss": 0.566,
      "step": 112
    },
    {
      "epoch": 0.6491144088080422,
      "grad_norm": 0.7698194980621338,
      "learning_rate": 8.544394337454409e-06,
      "loss": 0.5308,
      "step": 113
    },
    {
      "epoch": 0.6548587841072283,
      "grad_norm": 0.5629884600639343,
      "learning_rate": 8.508819268854713e-06,
      "loss": 0.555,
      "step": 114
    },
    {
      "epoch": 0.6606031594064146,
      "grad_norm": 0.6958817839622498,
      "learning_rate": 8.472890717343391e-06,
      "loss": 0.5623,
      "step": 115
    },
    {
      "epoch": 0.6663475347056008,
      "grad_norm": 0.6036946773529053,
      "learning_rate": 8.436612302408376e-06,
      "loss": 0.526,
      "step": 116
    },
    {
      "epoch": 0.672091910004787,
      "grad_norm": 0.511654794216156,
      "learning_rate": 8.399987678783285e-06,
      "loss": 0.5446,
      "step": 117
    },
    {
      "epoch": 0.6778362853039732,
      "grad_norm": 0.7489587068557739,
      "learning_rate": 8.36302053607924e-06,
      "loss": 0.5671,
      "step": 118
    },
    {
      "epoch": 0.6835806606031594,
      "grad_norm": 0.6197418570518494,
      "learning_rate": 8.325714598413169e-06,
      "loss": 0.5701,
      "step": 119
    },
    {
      "epoch": 0.6893250359023456,
      "grad_norm": 0.6502342820167542,
      "learning_rate": 8.288073624032634e-06,
      "loss": 0.5321,
      "step": 120
    },
    {
      "epoch": 0.6950694112015319,
      "grad_norm": 0.49902498722076416,
      "learning_rate": 8.250101404937223e-06,
      "loss": 0.5542,
      "step": 121
    },
    {
      "epoch": 0.700813786500718,
      "grad_norm": 0.5273813009262085,
      "learning_rate": 8.211801766496537e-06,
      "loss": 0.5311,
      "step": 122
    },
    {
      "epoch": 0.7065581617999043,
      "grad_norm": 0.5683380365371704,
      "learning_rate": 8.17317856706482e-06,
      "loss": 0.523,
      "step": 123
    },
    {
      "epoch": 0.7123025370990905,
      "grad_norm": 0.5648055076599121,
      "learning_rate": 8.13423569759226e-06,
      "loss": 0.5678,
      "step": 124
    },
    {
      "epoch": 0.7180469123982767,
      "grad_norm": 0.5629974007606506,
      "learning_rate": 8.094977081233006e-06,
      "loss": 0.5669,
      "step": 125
    },
    {
      "epoch": 0.7237912876974629,
      "grad_norm": 0.602506160736084,
      "learning_rate": 8.055406672949957e-06,
      "loss": 0.5275,
      "step": 126
    },
    {
      "epoch": 0.7295356629966491,
      "grad_norm": 0.572467029094696,
      "learning_rate": 8.015528459116321e-06,
      "loss": 0.5275,
      "step": 127
    },
    {
      "epoch": 0.7352800382958353,
      "grad_norm": 0.5794216394424438,
      "learning_rate": 7.975346457114034e-06,
      "loss": 0.5396,
      "step": 128
    },
    {
      "epoch": 0.7410244135950216,
      "grad_norm": 0.560357391834259,
      "learning_rate": 7.934864714929036e-06,
      "loss": 0.5483,
      "step": 129
    },
    {
      "epoch": 0.7467687888942077,
      "grad_norm": 0.5050742626190186,
      "learning_rate": 7.894087310743468e-06,
      "loss": 0.5617,
      "step": 130
    },
    {
      "epoch": 0.752513164193394,
      "grad_norm": 0.4990502595901489,
      "learning_rate": 7.853018352524845e-06,
      "loss": 0.5253,
      "step": 131
    },
    {
      "epoch": 0.7582575394925802,
      "grad_norm": 0.4985724985599518,
      "learning_rate": 7.811661977612202e-06,
      "loss": 0.5303,
      "step": 132
    },
    {
      "epoch": 0.7640019147917664,
      "grad_norm": 0.5744158029556274,
      "learning_rate": 7.770022352299294e-06,
      "loss": 0.5235,
      "step": 133
    },
    {
      "epoch": 0.7697462900909526,
      "grad_norm": 0.5937462449073792,
      "learning_rate": 7.728103671414889e-06,
      "loss": 0.5148,
      "step": 134
    },
    {
      "epoch": 0.7754906653901388,
      "grad_norm": 0.5738289952278137,
      "learning_rate": 7.685910157900158e-06,
      "loss": 0.5345,
      "step": 135
    },
    {
      "epoch": 0.781235040689325,
      "grad_norm": 0.5279061794281006,
      "learning_rate": 7.643446062383273e-06,
      "loss": 0.5451,
      "step": 136
    },
    {
      "epoch": 0.7869794159885113,
      "grad_norm": 0.5690545439720154,
      "learning_rate": 7.600715662751166e-06,
      "loss": 0.5083,
      "step": 137
    },
    {
      "epoch": 0.7927237912876974,
      "grad_norm": 0.5676944255828857,
      "learning_rate": 7.557723263718596e-06,
      "loss": 0.5516,
      "step": 138
    },
    {
      "epoch": 0.7984681665868837,
      "grad_norm": 0.6335099339485168,
      "learning_rate": 7.514473196394467e-06,
      "loss": 0.5583,
      "step": 139
    },
    {
      "epoch": 0.8042125418860699,
      "grad_norm": 0.6167879104614258,
      "learning_rate": 7.470969817845518e-06,
      "loss": 0.5675,
      "step": 140
    },
    {
      "epoch": 0.8099569171852561,
      "grad_norm": 0.6238246560096741,
      "learning_rate": 7.427217510657383e-06,
      "loss": 0.5242,
      "step": 141
    },
    {
      "epoch": 0.8157012924844423,
      "grad_norm": 0.6096833944320679,
      "learning_rate": 7.383220682493081e-06,
      "loss": 0.56,
      "step": 142
    },
    {
      "epoch": 0.8214456677836285,
      "grad_norm": 0.6408026814460754,
      "learning_rate": 7.338983765648985e-06,
      "loss": 0.517,
      "step": 143
    },
    {
      "epoch": 0.8271900430828147,
      "grad_norm": 0.6161112785339355,
      "learning_rate": 7.294511216608308e-06,
      "loss": 0.5568,
      "step": 144
    },
    {
      "epoch": 0.832934418382001,
      "grad_norm": 0.6821258664131165,
      "learning_rate": 7.249807515592149e-06,
      "loss": 0.5391,
      "step": 145
    },
    {
      "epoch": 0.8386787936811871,
      "grad_norm": 0.5701781511306763,
      "learning_rate": 7.2048771661081515e-06,
      "loss": 0.5393,
      "step": 146
    },
    {
      "epoch": 0.8444231689803734,
      "grad_norm": 0.573361873626709,
      "learning_rate": 7.159724694496815e-06,
      "loss": 0.5104,
      "step": 147
    },
    {
      "epoch": 0.8501675442795597,
      "grad_norm": 0.6415104269981384,
      "learning_rate": 7.114354649475499e-06,
      "loss": 0.5373,
      "step": 148
    },
    {
      "epoch": 0.8559119195787458,
      "grad_norm": 0.5799002647399902,
      "learning_rate": 7.068771601680191e-06,
      "loss": 0.516,
      "step": 149
    },
    {
      "epoch": 0.861656294877932,
      "grad_norm": 0.6010524034500122,
      "learning_rate": 7.022980143205046e-06,
      "loss": 0.5421,
      "step": 150
    },
    {
      "epoch": 0.8674006701771182,
      "grad_norm": 0.5849425196647644,
      "learning_rate": 6.976984887139775e-06,
      "loss": 0.5576,
      "step": 151
    },
    {
      "epoch": 0.8731450454763044,
      "grad_norm": 0.49850520491600037,
      "learning_rate": 6.930790467104916e-06,
      "loss": 0.4881,
      "step": 152
    },
    {
      "epoch": 0.8788894207754907,
      "grad_norm": 0.6236470937728882,
      "learning_rate": 6.884401536785045e-06,
      "loss": 0.5521,
      "step": 153
    },
    {
      "epoch": 0.8846337960746768,
      "grad_norm": 0.5845939517021179,
      "learning_rate": 6.837822769459942e-06,
      "loss": 0.5143,
      "step": 154
    },
    {
      "epoch": 0.8903781713738631,
      "grad_norm": 0.48170965909957886,
      "learning_rate": 6.791058857533814e-06,
      "loss": 0.5248,
      "step": 155
    },
    {
      "epoch": 0.8961225466730494,
      "grad_norm": 0.5586768984794617,
      "learning_rate": 6.744114512062571e-06,
      "loss": 0.543,
      "step": 156
    },
    {
      "epoch": 0.9018669219722355,
      "grad_norm": 0.5925922393798828,
      "learning_rate": 6.696994462279223e-06,
      "loss": 0.5272,
      "step": 157
    },
    {
      "epoch": 0.9076112972714218,
      "grad_norm": 0.5067867636680603,
      "learning_rate": 6.6497034551174585e-06,
      "loss": 0.5572,
      "step": 158
    },
    {
      "epoch": 0.9133556725706079,
      "grad_norm": 0.5289098024368286,
      "learning_rate": 6.602246254733431e-06,
      "loss": 0.5216,
      "step": 159
    },
    {
      "epoch": 0.9191000478697942,
      "grad_norm": 0.5776088833808899,
      "learning_rate": 6.554627642025807e-06,
      "loss": 0.5278,
      "step": 160
    },
    {
      "epoch": 0.9248444231689804,
      "grad_norm": 0.4834924340248108,
      "learning_rate": 6.506852414154138e-06,
      "loss": 0.5479,
      "step": 161
    },
    {
      "epoch": 0.9305887984681666,
      "grad_norm": 0.5455948114395142,
      "learning_rate": 6.4589253840555856e-06,
      "loss": 0.4813,
      "step": 162
    },
    {
      "epoch": 0.9363331737673528,
      "grad_norm": 0.5379589200019836,
      "learning_rate": 6.41085137996006e-06,
      "loss": 0.5391,
      "step": 163
    },
    {
      "epoch": 0.9420775490665391,
      "grad_norm": 0.5952578783035278,
      "learning_rate": 6.362635244903818e-06,
      "loss": 0.556,
      "step": 164
    },
    {
      "epoch": 0.9478219243657252,
      "grad_norm": 0.5639842748641968,
      "learning_rate": 6.314281836241573e-06,
      "loss": 0.552,
      "step": 165
    },
    {
      "epoch": 0.9535662996649115,
      "grad_norm": 0.5705066323280334,
      "learning_rate": 6.265796025157154e-06,
      "loss": 0.5028,
      "step": 166
    },
    {
      "epoch": 0.9593106749640976,
      "grad_norm": 0.5012354850769043,
      "learning_rate": 6.217182696172776e-06,
      "loss": 0.5229,
      "step": 167
    },
    {
      "epoch": 0.9650550502632839,
      "grad_norm": 0.5164003968238831,
      "learning_rate": 6.168446746656973e-06,
      "loss": 0.5166,
      "step": 168
    },
    {
      "epoch": 0.9707994255624701,
      "grad_norm": 0.5868187546730042,
      "learning_rate": 6.119593086331225e-06,
      "loss": 0.5342,
      "step": 169
    },
    {
      "epoch": 0.9765438008616563,
      "grad_norm": 0.5106255412101746,
      "learning_rate": 6.070626636775349e-06,
      "loss": 0.526,
      "step": 170
    },
    {
      "epoch": 0.9822881761608425,
      "grad_norm": 0.44993191957473755,
      "learning_rate": 6.021552330931693e-06,
      "loss": 0.5599,
      "step": 171
    },
    {
      "epoch": 0.9880325514600288,
      "grad_norm": 0.5432422161102295,
      "learning_rate": 5.972375112608182e-06,
      "loss": 0.5265,
      "step": 172
    },
    {
      "epoch": 0.9937769267592149,
      "grad_norm": 0.5361687541007996,
      "learning_rate": 5.923099935980278e-06,
      "loss": 0.5652,
      "step": 173
    },
    {
      "epoch": 0.9995213020584012,
      "grad_norm": 0.43177682161331177,
      "learning_rate": 5.8737317650918905e-06,
      "loss": 0.5462,
      "step": 174
    },
    {
      "epoch": 1.0052656773575874,
      "grad_norm": 1.580591082572937,
      "learning_rate": 5.824275573355278e-06,
      "loss": 0.9082,
      "step": 175
    },
    {
      "epoch": 1.0110100526567736,
      "grad_norm": 0.5395136475563049,
      "learning_rate": 5.7747363430500395e-06,
      "loss": 0.4953,
      "step": 176
    },
    {
      "epoch": 1.0167544279559597,
      "grad_norm": 0.5305489301681519,
      "learning_rate": 5.725119064821185e-06,
      "loss": 0.5229,
      "step": 177
    },
    {
      "epoch": 1.022498803255146,
      "grad_norm": 0.5417148470878601,
      "learning_rate": 5.675428737176367e-06,
      "loss": 0.461,
      "step": 178
    },
    {
      "epoch": 1.0282431785543322,
      "grad_norm": 0.5382701754570007,
      "learning_rate": 5.625670365982332e-06,
      "loss": 0.4442,
      "step": 179
    },
    {
      "epoch": 1.0339875538535184,
      "grad_norm": 0.5847955346107483,
      "learning_rate": 5.575848963960621e-06,
      "loss": 0.5286,
      "step": 180
    },
    {
      "epoch": 1.0397319291527047,
      "grad_norm": 0.4929944574832916,
      "learning_rate": 5.525969550182577e-06,
      "loss": 0.4789,
      "step": 181
    },
    {
      "epoch": 1.0454763044518909,
      "grad_norm": 0.5635377764701843,
      "learning_rate": 5.4760371495637256e-06,
      "loss": 0.4753,
      "step": 182
    },
    {
      "epoch": 1.051220679751077,
      "grad_norm": 0.4980822801589966,
      "learning_rate": 5.426056792357552e-06,
      "loss": 0.4895,
      "step": 183
    },
    {
      "epoch": 1.0569650550502634,
      "grad_norm": 0.5097235441207886,
      "learning_rate": 5.376033513648743e-06,
      "loss": 0.4794,
      "step": 184
    },
    {
      "epoch": 1.0627094303494495,
      "grad_norm": 0.5179345607757568,
      "learning_rate": 5.325972352845965e-06,
      "loss": 0.4957,
      "step": 185
    },
    {
      "epoch": 1.0684538056486357,
      "grad_norm": 0.4883691072463989,
      "learning_rate": 5.2758783531741655e-06,
      "loss": 0.5047,
      "step": 186
    },
    {
      "epoch": 1.0741981809478218,
      "grad_norm": 0.46270909905433655,
      "learning_rate": 5.225756561166521e-06,
      "loss": 0.5066,
      "step": 187
    },
    {
      "epoch": 1.0799425562470082,
      "grad_norm": 0.5014520883560181,
      "learning_rate": 5.175612026156045e-06,
      "loss": 0.5012,
      "step": 188
    },
    {
      "epoch": 1.0856869315461943,
      "grad_norm": 0.4673246443271637,
      "learning_rate": 5.125449799766916e-06,
      "loss": 0.4638,
      "step": 189
    },
    {
      "epoch": 1.0914313068453805,
      "grad_norm": 0.4951716363430023,
      "learning_rate": 5.075274935405554e-06,
      "loss": 0.4915,
      "step": 190
    },
    {
      "epoch": 1.0971756821445668,
      "grad_norm": 0.4571003317832947,
      "learning_rate": 5.025092487751552e-06,
      "loss": 0.494,
      "step": 191
    },
    {
      "epoch": 1.102920057443753,
      "grad_norm": 0.4891183078289032,
      "learning_rate": 4.974907512248451e-06,
      "loss": 0.4549,
      "step": 192
    },
    {
      "epoch": 1.1086644327429391,
      "grad_norm": 0.4218912422657013,
      "learning_rate": 4.924725064594448e-06,
      "loss": 0.489,
      "step": 193
    },
    {
      "epoch": 1.1144088080421255,
      "grad_norm": 0.45297980308532715,
      "learning_rate": 4.874550200233085e-06,
      "loss": 0.4907,
      "step": 194
    },
    {
      "epoch": 1.1201531833413116,
      "grad_norm": 0.5533784031867981,
      "learning_rate": 4.824387973843957e-06,
      "loss": 0.5129,
      "step": 195
    },
    {
      "epoch": 1.1258975586404978,
      "grad_norm": 0.45622408390045166,
      "learning_rate": 4.7742434388334815e-06,
      "loss": 0.4418,
      "step": 196
    },
    {
      "epoch": 1.1316419339396842,
      "grad_norm": 0.4435925781726837,
      "learning_rate": 4.724121646825838e-06,
      "loss": 0.5022,
      "step": 197
    },
    {
      "epoch": 1.1373863092388703,
      "grad_norm": 0.4371262490749359,
      "learning_rate": 4.674027647154037e-06,
      "loss": 0.4711,
      "step": 198
    },
    {
      "epoch": 1.1431306845380564,
      "grad_norm": 0.473237544298172,
      "learning_rate": 4.623966486351257e-06,
      "loss": 0.4465,
      "step": 199
    },
    {
      "epoch": 1.1488750598372426,
      "grad_norm": 0.4501272439956665,
      "learning_rate": 4.573943207642452e-06,
      "loss": 0.4659,
      "step": 200
    },
    {
      "epoch": 1.154619435136429,
      "grad_norm": 0.5544123649597168,
      "learning_rate": 4.523962850436276e-06,
      "loss": 0.4974,
      "step": 201
    },
    {
      "epoch": 1.160363810435615,
      "grad_norm": 0.519002377986908,
      "learning_rate": 4.474030449817423e-06,
      "loss": 0.5099,
      "step": 202
    },
    {
      "epoch": 1.1661081857348012,
      "grad_norm": 0.48358842730522156,
      "learning_rate": 4.424151036039381e-06,
      "loss": 0.4687,
      "step": 203
    },
    {
      "epoch": 1.1718525610339876,
      "grad_norm": 0.5397114753723145,
      "learning_rate": 4.3743296340176694e-06,
      "loss": 0.505,
      "step": 204
    },
    {
      "epoch": 1.1775969363331737,
      "grad_norm": 0.47014209628105164,
      "learning_rate": 4.3245712628236356e-06,
      "loss": 0.5201,
      "step": 205
    },
    {
      "epoch": 1.18334131163236,
      "grad_norm": 0.4648483395576477,
      "learning_rate": 4.274880935178817e-06,
      "loss": 0.4652,
      "step": 206
    },
    {
      "epoch": 1.1890856869315463,
      "grad_norm": 0.45514115691185,
      "learning_rate": 4.225263656949961e-06,
      "loss": 0.4692,
      "step": 207
    },
    {
      "epoch": 1.1948300622307324,
      "grad_norm": 0.48053470253944397,
      "learning_rate": 4.175724426644724e-06,
      "loss": 0.4842,
      "step": 208
    },
    {
      "epoch": 1.2005744375299185,
      "grad_norm": 0.45836907625198364,
      "learning_rate": 4.12626823490811e-06,
      "loss": 0.4535,
      "step": 209
    },
    {
      "epoch": 1.206318812829105,
      "grad_norm": 0.43734660744667053,
      "learning_rate": 4.076900064019721e-06,
      "loss": 0.5134,
      "step": 210
    },
    {
      "epoch": 1.212063188128291,
      "grad_norm": 0.4292330741882324,
      "learning_rate": 4.02762488739182e-06,
      "loss": 0.4933,
      "step": 211
    },
    {
      "epoch": 1.2178075634274772,
      "grad_norm": 0.48980435729026794,
      "learning_rate": 3.978447669068309e-06,
      "loss": 0.4924,
      "step": 212
    },
    {
      "epoch": 1.2235519387266636,
      "grad_norm": 0.415892630815506,
      "learning_rate": 3.929373363224654e-06,
      "loss": 0.4602,
      "step": 213
    },
    {
      "epoch": 1.2292963140258497,
      "grad_norm": 0.5073506236076355,
      "learning_rate": 3.8804069136687775e-06,
      "loss": 0.4469,
      "step": 214
    },
    {
      "epoch": 1.2350406893250359,
      "grad_norm": 0.4534899592399597,
      "learning_rate": 3.8315532533430285e-06,
      "loss": 0.4937,
      "step": 215
    },
    {
      "epoch": 1.2407850646242222,
      "grad_norm": 0.43552905321121216,
      "learning_rate": 3.7828173038272266e-06,
      "loss": 0.5012,
      "step": 216
    },
    {
      "epoch": 1.2465294399234084,
      "grad_norm": 0.45762044191360474,
      "learning_rate": 3.7342039748428473e-06,
      "loss": 0.448,
      "step": 217
    },
    {
      "epoch": 1.2522738152225945,
      "grad_norm": 0.4884515702724457,
      "learning_rate": 3.685718163758427e-06,
      "loss": 0.4803,
      "step": 218
    },
    {
      "epoch": 1.2580181905217809,
      "grad_norm": 0.414726585149765,
      "learning_rate": 3.6373647550961834e-06,
      "loss": 0.4844,
      "step": 219
    },
    {
      "epoch": 1.263762565820967,
      "grad_norm": 0.44502031803131104,
      "learning_rate": 3.5891486200399413e-06,
      "loss": 0.5254,
      "step": 220
    },
    {
      "epoch": 1.2695069411201532,
      "grad_norm": 0.46549510955810547,
      "learning_rate": 3.5410746159444165e-06,
      "loss": 0.503,
      "step": 221
    },
    {
      "epoch": 1.2752513164193395,
      "grad_norm": 0.480082631111145,
      "learning_rate": 3.4931475858458634e-06,
      "loss": 0.474,
      "step": 222
    },
    {
      "epoch": 1.2809956917185257,
      "grad_norm": 0.5088851451873779,
      "learning_rate": 3.445372357974194e-06,
      "loss": 0.5197,
      "step": 223
    },
    {
      "epoch": 1.2867400670177118,
      "grad_norm": 0.4297751486301422,
      "learning_rate": 3.397753745266571e-06,
      "loss": 0.4698,
      "step": 224
    },
    {
      "epoch": 1.292484442316898,
      "grad_norm": 0.436560720205307,
      "learning_rate": 3.350296544882543e-06,
      "loss": 0.4573,
      "step": 225
    },
    {
      "epoch": 1.2982288176160843,
      "grad_norm": 0.451806902885437,
      "learning_rate": 3.303005537720778e-06,
      "loss": 0.4654,
      "step": 226
    },
    {
      "epoch": 1.3039731929152705,
      "grad_norm": 0.4921393394470215,
      "learning_rate": 3.255885487937431e-06,
      "loss": 0.5075,
      "step": 227
    },
    {
      "epoch": 1.3097175682144566,
      "grad_norm": 0.4880327880382538,
      "learning_rate": 3.2089411424661864e-06,
      "loss": 0.5467,
      "step": 228
    },
    {
      "epoch": 1.3154619435136428,
      "grad_norm": 0.40174514055252075,
      "learning_rate": 3.1621772305400603e-06,
      "loss": 0.4965,
      "step": 229
    },
    {
      "epoch": 1.3212063188128291,
      "grad_norm": 0.41413503885269165,
      "learning_rate": 3.1155984632149565e-06,
      "loss": 0.4826,
      "step": 230
    },
    {
      "epoch": 1.3269506941120153,
      "grad_norm": 0.43642458319664,
      "learning_rate": 3.0692095328950843e-06,
      "loss": 0.4355,
      "step": 231
    },
    {
      "epoch": 1.3326950694112014,
      "grad_norm": 0.4538045823574066,
      "learning_rate": 3.023015112860228e-06,
      "loss": 0.5378,
      "step": 232
    },
    {
      "epoch": 1.3384394447103878,
      "grad_norm": 0.42121201753616333,
      "learning_rate": 2.977019856794955e-06,
      "loss": 0.4811,
      "step": 233
    },
    {
      "epoch": 1.344183820009574,
      "grad_norm": 0.3827037513256073,
      "learning_rate": 2.93122839831981e-06,
      "loss": 0.4956,
      "step": 234
    },
    {
      "epoch": 1.34992819530876,
      "grad_norm": 0.3937554359436035,
      "learning_rate": 2.8856453505245018e-06,
      "loss": 0.4729,
      "step": 235
    },
    {
      "epoch": 1.3556725706079464,
      "grad_norm": 0.4603598415851593,
      "learning_rate": 2.840275305503186e-06,
      "loss": 0.5352,
      "step": 236
    },
    {
      "epoch": 1.3614169459071326,
      "grad_norm": 0.4124448001384735,
      "learning_rate": 2.7951228338918506e-06,
      "loss": 0.4523,
      "step": 237
    },
    {
      "epoch": 1.3671613212063187,
      "grad_norm": 0.41025596857070923,
      "learning_rate": 2.7501924844078538e-06,
      "loss": 0.4824,
      "step": 238
    },
    {
      "epoch": 1.372905696505505,
      "grad_norm": 0.3917722702026367,
      "learning_rate": 2.7054887833916933e-06,
      "loss": 0.5059,
      "step": 239
    },
    {
      "epoch": 1.3786500718046912,
      "grad_norm": 0.4388699531555176,
      "learning_rate": 2.6610162343510183e-06,
      "loss": 0.5039,
      "step": 240
    },
    {
      "epoch": 1.3843944471038774,
      "grad_norm": 0.4307573139667511,
      "learning_rate": 2.616779317506921e-06,
      "loss": 0.4668,
      "step": 241
    },
    {
      "epoch": 1.3901388224030637,
      "grad_norm": 0.40513068437576294,
      "learning_rate": 2.572782489342617e-06,
      "loss": 0.4365,
      "step": 242
    },
    {
      "epoch": 1.39588319770225,
      "grad_norm": 0.43141913414001465,
      "learning_rate": 2.5290301821544826e-06,
      "loss": 0.5055,
      "step": 243
    },
    {
      "epoch": 1.401627573001436,
      "grad_norm": 0.4245367646217346,
      "learning_rate": 2.4855268036055346e-06,
      "loss": 0.5001,
      "step": 244
    },
    {
      "epoch": 1.4073719483006224,
      "grad_norm": 0.41667869687080383,
      "learning_rate": 2.4422767362814045e-06,
      "loss": 0.5056,
      "step": 245
    },
    {
      "epoch": 1.4131163235998085,
      "grad_norm": 0.41470879316329956,
      "learning_rate": 2.3992843372488357e-06,
      "loss": 0.477,
      "step": 246
    },
    {
      "epoch": 1.4188606988989947,
      "grad_norm": 0.43821245431900024,
      "learning_rate": 2.3565539376167295e-06,
      "loss": 0.5074,
      "step": 247
    },
    {
      "epoch": 1.424605074198181,
      "grad_norm": 0.38262587785720825,
      "learning_rate": 2.3140898420998425e-06,
      "loss": 0.4686,
      "step": 248
    },
    {
      "epoch": 1.4303494494973672,
      "grad_norm": 0.3910864591598511,
      "learning_rate": 2.271896328585114e-06,
      "loss": 0.4852,
      "step": 249
    },
    {
      "epoch": 1.4360938247965533,
      "grad_norm": 0.43671905994415283,
      "learning_rate": 2.2299776477007073e-06,
      "loss": 0.4729,
      "step": 250
    },
    {
      "epoch": 1.4418382000957397,
      "grad_norm": 0.3937380611896515,
      "learning_rate": 2.1883380223878004e-06,
      "loss": 0.5039,
      "step": 251
    },
    {
      "epoch": 1.4475825753949259,
      "grad_norm": 0.4383409023284912,
      "learning_rate": 2.1469816474751566e-06,
      "loss": 0.5264,
      "step": 252
    },
    {
      "epoch": 1.453326950694112,
      "grad_norm": 0.38024550676345825,
      "learning_rate": 2.105912689256533e-06,
      "loss": 0.4575,
      "step": 253
    },
    {
      "epoch": 1.4590713259932984,
      "grad_norm": 0.4220210313796997,
      "learning_rate": 2.0651352850709656e-06,
      "loss": 0.4846,
      "step": 254
    },
    {
      "epoch": 1.4648157012924845,
      "grad_norm": 0.4223257005214691,
      "learning_rate": 2.0246535428859652e-06,
      "loss": 0.4968,
      "step": 255
    },
    {
      "epoch": 1.4705600765916707,
      "grad_norm": 0.3835613429546356,
      "learning_rate": 1.984471540883679e-06,
      "loss": 0.4874,
      "step": 256
    },
    {
      "epoch": 1.4763044518908568,
      "grad_norm": 0.43379145860671997,
      "learning_rate": 1.9445933270500444e-06,
      "loss": 0.5144,
      "step": 257
    },
    {
      "epoch": 1.4820488271900432,
      "grad_norm": 0.3767804205417633,
      "learning_rate": 1.905022918766995e-06,
      "loss": 0.4621,
      "step": 258
    },
    {
      "epoch": 1.4877932024892293,
      "grad_norm": 0.4112907648086548,
      "learning_rate": 1.8657643024077431e-06,
      "loss": 0.5212,
      "step": 259
    },
    {
      "epoch": 1.4935375777884154,
      "grad_norm": 0.41141629219055176,
      "learning_rate": 1.8268214329351797e-06,
      "loss": 0.4859,
      "step": 260
    },
    {
      "epoch": 1.4992819530876016,
      "grad_norm": 0.3843127191066742,
      "learning_rate": 1.7881982335034625e-06,
      "loss": 0.4308,
      "step": 261
    },
    {
      "epoch": 1.505026328386788,
      "grad_norm": 0.3895561695098877,
      "learning_rate": 1.7498985950627794e-06,
      "loss": 0.4565,
      "step": 262
    },
    {
      "epoch": 1.510770703685974,
      "grad_norm": 0.4275520443916321,
      "learning_rate": 1.7119263759673677e-06,
      "loss": 0.4704,
      "step": 263
    },
    {
      "epoch": 1.5165150789851602,
      "grad_norm": 0.4638034999370575,
      "learning_rate": 1.6742854015868349e-06,
      "loss": 0.5137,
      "step": 264
    },
    {
      "epoch": 1.5222594542843466,
      "grad_norm": 0.35882094502449036,
      "learning_rate": 1.6369794639207626e-06,
      "loss": 0.4582,
      "step": 265
    },
    {
      "epoch": 1.5280038295835328,
      "grad_norm": 0.4297800362110138,
      "learning_rate": 1.6000123212167158e-06,
      "loss": 0.471,
      "step": 266
    },
    {
      "epoch": 1.533748204882719,
      "grad_norm": 0.37162676453590393,
      "learning_rate": 1.5633876975916261e-06,
      "loss": 0.479,
      "step": 267
    },
    {
      "epoch": 1.5394925801819053,
      "grad_norm": 0.4432937502861023,
      "learning_rate": 1.5271092826566108e-06,
      "loss": 0.4918,
      "step": 268
    },
    {
      "epoch": 1.5452369554810914,
      "grad_norm": 0.4131205976009369,
      "learning_rate": 1.4911807311452874e-06,
      "loss": 0.4862,
      "step": 269
    },
    {
      "epoch": 1.5509813307802776,
      "grad_norm": 0.39494404196739197,
      "learning_rate": 1.4556056625455922e-06,
      "loss": 0.472,
      "step": 270
    },
    {
      "epoch": 1.556725706079464,
      "grad_norm": 0.3842601776123047,
      "learning_rate": 1.4203876607351347e-06,
      "loss": 0.5468,
      "step": 271
    },
    {
      "epoch": 1.56247008137865,
      "grad_norm": 0.39018625020980835,
      "learning_rate": 1.3855302736201686e-06,
      "loss": 0.4695,
      "step": 272
    },
    {
      "epoch": 1.5682144566778362,
      "grad_norm": 0.4116652011871338,
      "learning_rate": 1.3510370127781635e-06,
      "loss": 0.5009,
      "step": 273
    },
    {
      "epoch": 1.5739588319770226,
      "grad_norm": 0.4268287420272827,
      "learning_rate": 1.3169113531040462e-06,
      "loss": 0.509,
      "step": 274
    },
    {
      "epoch": 1.5797032072762087,
      "grad_norm": 0.403522253036499,
      "learning_rate": 1.2831567324601325e-06,
      "loss": 0.4961,
      "step": 275
    },
    {
      "epoch": 1.5854475825753949,
      "grad_norm": 0.37210026383399963,
      "learning_rate": 1.2497765513297976e-06,
      "loss": 0.4923,
      "step": 276
    },
    {
      "epoch": 1.5911919578745812,
      "grad_norm": 0.4218740463256836,
      "learning_rate": 1.2167741724749026e-06,
      "loss": 0.4894,
      "step": 277
    },
    {
      "epoch": 1.5969363331737674,
      "grad_norm": 0.3306220769882202,
      "learning_rate": 1.1841529205970281e-06,
      "loss": 0.4224,
      "step": 278
    },
    {
      "epoch": 1.6026807084729535,
      "grad_norm": 0.4241751730442047,
      "learning_rate": 1.1519160820025382e-06,
      "loss": 0.4937,
      "step": 279
    },
    {
      "epoch": 1.6084250837721399,
      "grad_norm": 0.3803061842918396,
      "learning_rate": 1.1200669042715163e-06,
      "loss": 0.4921,
      "step": 280
    },
    {
      "epoch": 1.614169459071326,
      "grad_norm": 0.40259426832199097,
      "learning_rate": 1.0886085959305915e-06,
      "loss": 0.5068,
      "step": 281
    },
    {
      "epoch": 1.6199138343705122,
      "grad_norm": 0.3979838490486145,
      "learning_rate": 1.057544326129723e-06,
      "loss": 0.4888,
      "step": 282
    },
    {
      "epoch": 1.6256582096696985,
      "grad_norm": 0.39289093017578125,
      "learning_rate": 1.026877224322923e-06,
      "loss": 0.5025,
      "step": 283
    },
    {
      "epoch": 1.6314025849688847,
      "grad_norm": 0.3723068833351135,
      "learning_rate": 9.966103799529891e-07,
      "loss": 0.5262,
      "step": 284
    },
    {
      "epoch": 1.6371469602680708,
      "grad_norm": 0.3888184726238251,
      "learning_rate": 9.66746842140287e-07,
      "loss": 0.4623,
      "step": 285
    },
    {
      "epoch": 1.6428913355672572,
      "grad_norm": 0.38385969400405884,
      "learning_rate": 9.372896193755621e-07,
      "loss": 0.4814,
      "step": 286
    },
    {
      "epoch": 1.6486357108664431,
      "grad_norm": 0.386476069688797,
      "learning_rate": 9.082416792168608e-07,
      "loss": 0.5195,
      "step": 287
    },
    {
      "epoch": 1.6543800861656295,
      "grad_norm": 0.3499607741832733,
      "learning_rate": 8.7960594799059e-07,
      "loss": 0.4491,
      "step": 288
    },
    {
      "epoch": 1.6601244614648158,
      "grad_norm": 0.3445354402065277,
      "learning_rate": 8.513853104966951e-07,
      "loss": 0.5042,
      "step": 289
    },
    {
      "epoch": 1.6658688367640018,
      "grad_norm": 0.41543084383010864,
      "learning_rate": 8.235826097180566e-07,
      "loss": 0.4904,
      "step": 290
    },
    {
      "epoch": 1.6716132120631881,
      "grad_norm": 0.40791216492652893,
      "learning_rate": 7.962006465340821e-07,
      "loss": 0.4554,
      "step": 291
    },
    {
      "epoch": 1.6773575873623745,
      "grad_norm": 0.36069628596305847,
      "learning_rate": 7.692421794385313e-07,
      "loss": 0.473,
      "step": 292
    },
    {
      "epoch": 1.6831019626615604,
      "grad_norm": 0.44029319286346436,
      "learning_rate": 7.427099242616348e-07,
      "loss": 0.5168,
      "step": 293
    },
    {
      "epoch": 1.6888463379607468,
      "grad_norm": 0.4227386713027954,
      "learning_rate": 7.166065538964955e-07,
      "loss": 0.5405,
      "step": 294
    },
    {
      "epoch": 1.694590713259933,
      "grad_norm": 0.37983396649360657,
      "learning_rate": 6.909346980298093e-07,
      "loss": 0.4794,
      "step": 295
    },
    {
      "epoch": 1.700335088559119,
      "grad_norm": 0.3403533101081848,
      "learning_rate": 6.656969428769567e-07,
      "loss": 0.4514,
      "step": 296
    },
    {
      "epoch": 1.7060794638583054,
      "grad_norm": 0.3760150372982025,
      "learning_rate": 6.408958309214597e-07,
      "loss": 0.4801,
      "step": 297
    },
    {
      "epoch": 1.7118238391574916,
      "grad_norm": 0.4047829508781433,
      "learning_rate": 6.165338606588517e-07,
      "loss": 0.5115,
      "step": 298
    },
    {
      "epoch": 1.7175682144566777,
      "grad_norm": 0.38393792510032654,
      "learning_rate": 5.926134863449712e-07,
      "loss": 0.4826,
      "step": 299
    },
    {
      "epoch": 1.723312589755864,
      "grad_norm": 0.3590121865272522,
      "learning_rate": 5.691371177487215e-07,
      "loss": 0.4491,
      "step": 300
    },
    {
      "epoch": 1.7290569650550502,
      "grad_norm": 0.4042126536369324,
      "learning_rate": 5.461071199093048e-07,
      "loss": 0.5254,
      "step": 301
    },
    {
      "epoch": 1.7348013403542364,
      "grad_norm": 0.3463343381881714,
      "learning_rate": 5.235258128979676e-07,
      "loss": 0.4591,
      "step": 302
    },
    {
      "epoch": 1.7405457156534228,
      "grad_norm": 0.37464749813079834,
      "learning_rate": 5.0139547158427e-07,
      "loss": 0.4621,
      "step": 303
    },
    {
      "epoch": 1.746290090952609,
      "grad_norm": 0.386735737323761,
      "learning_rate": 4.797183254069176e-07,
      "loss": 0.499,
      "step": 304
    },
    {
      "epoch": 1.752034466251795,
      "grad_norm": 0.3763267695903778,
      "learning_rate": 4.5849655814915683e-07,
      "loss": 0.5162,
      "step": 305
    },
    {
      "epoch": 1.7577788415509814,
      "grad_norm": 0.37057819962501526,
      "learning_rate": 4.3773230771879004e-07,
      "loss": 0.4925,
      "step": 306
    },
    {
      "epoch": 1.7635232168501676,
      "grad_norm": 0.35524293780326843,
      "learning_rate": 4.1742766593278974e-07,
      "loss": 0.4587,
      "step": 307
    },
    {
      "epoch": 1.7692675921493537,
      "grad_norm": 0.40347564220428467,
      "learning_rate": 3.9758467830656623e-07,
      "loss": 0.5026,
      "step": 308
    },
    {
      "epoch": 1.77501196744854,
      "grad_norm": 0.3765867054462433,
      "learning_rate": 3.782053438479094e-07,
      "loss": 0.4692,
      "step": 309
    },
    {
      "epoch": 1.7807563427477262,
      "grad_norm": 0.35763201117515564,
      "learning_rate": 3.5929161485559694e-07,
      "loss": 0.4576,
      "step": 310
    },
    {
      "epoch": 1.7865007180469124,
      "grad_norm": 0.4578535556793213,
      "learning_rate": 3.4084539672271764e-07,
      "loss": 0.5168,
      "step": 311
    },
    {
      "epoch": 1.7922450933460987,
      "grad_norm": 0.36126625537872314,
      "learning_rate": 3.228685477447291e-07,
      "loss": 0.4769,
      "step": 312
    },
    {
      "epoch": 1.7979894686452849,
      "grad_norm": 0.40714967250823975,
      "learning_rate": 3.0536287893223603e-07,
      "loss": 0.4646,
      "step": 313
    },
    {
      "epoch": 1.803733843944471,
      "grad_norm": 0.35251492261886597,
      "learning_rate": 2.883301538285582e-07,
      "loss": 0.4506,
      "step": 314
    },
    {
      "epoch": 1.8094782192436574,
      "grad_norm": 0.36679378151893616,
      "learning_rate": 2.717720883320685e-07,
      "loss": 0.4858,
      "step": 315
    },
    {
      "epoch": 1.8152225945428435,
      "grad_norm": 0.39732909202575684,
      "learning_rate": 2.556903505233216e-07,
      "loss": 0.5302,
      "step": 316
    },
    {
      "epoch": 1.8209669698420297,
      "grad_norm": 0.37163034081459045,
      "learning_rate": 2.4008656049701875e-07,
      "loss": 0.4868,
      "step": 317
    },
    {
      "epoch": 1.826711345141216,
      "grad_norm": 0.37081223726272583,
      "learning_rate": 2.2496229019879635e-07,
      "loss": 0.4698,
      "step": 318
    },
    {
      "epoch": 1.832455720440402,
      "grad_norm": 0.3732491135597229,
      "learning_rate": 2.1031906326685946e-07,
      "loss": 0.4716,
      "step": 319
    },
    {
      "epoch": 1.8382000957395883,
      "grad_norm": 0.437520831823349,
      "learning_rate": 1.9615835487849677e-07,
      "loss": 0.4804,
      "step": 320
    },
    {
      "epoch": 1.8439444710387747,
      "grad_norm": 0.3673948645591736,
      "learning_rate": 1.824815916014644e-07,
      "loss": 0.5048,
      "step": 321
    },
    {
      "epoch": 1.8496888463379606,
      "grad_norm": 0.3330841064453125,
      "learning_rate": 1.6929015125027314e-07,
      "loss": 0.4454,
      "step": 322
    },
    {
      "epoch": 1.855433221637147,
      "grad_norm": 0.3728371560573578,
      "learning_rate": 1.5658536274738623e-07,
      "loss": 0.5037,
      "step": 323
    },
    {
      "epoch": 1.8611775969363333,
      "grad_norm": 0.36687803268432617,
      "learning_rate": 1.443685059893396e-07,
      "loss": 0.5074,
      "step": 324
    },
    {
      "epoch": 1.8669219722355193,
      "grad_norm": 0.3871447741985321,
      "learning_rate": 1.3264081171780797e-07,
      "loss": 0.4878,
      "step": 325
    },
    {
      "epoch": 1.8726663475347056,
      "grad_norm": 0.34846240282058716,
      "learning_rate": 1.2140346139561277e-07,
      "loss": 0.4694,
      "step": 326
    },
    {
      "epoch": 1.8784107228338918,
      "grad_norm": 0.37107598781585693,
      "learning_rate": 1.1065758708770468e-07,
      "loss": 0.4856,
      "step": 327
    },
    {
      "epoch": 1.884155098133078,
      "grad_norm": 0.38006165623664856,
      "learning_rate": 1.004042713471165e-07,
      "loss": 0.5022,
      "step": 328
    },
    {
      "epoch": 1.8898994734322643,
      "grad_norm": 0.36080440878868103,
      "learning_rate": 9.064454710590253e-08,
      "loss": 0.4619,
      "step": 329
    },
    {
      "epoch": 1.8956438487314504,
      "grad_norm": 0.350732684135437,
      "learning_rate": 8.137939757108526e-08,
      "loss": 0.4584,
      "step": 330
    },
    {
      "epoch": 1.9013882240306366,
      "grad_norm": 0.37510958313941956,
      "learning_rate": 7.260975612560173e-08,
      "loss": 0.4813,
      "step": 331
    },
    {
      "epoch": 1.907132599329823,
      "grad_norm": 0.3882618844509125,
      "learning_rate": 6.433650623427379e-08,
      "loss": 0.4304,
      "step": 332
    },
    {
      "epoch": 1.912876974629009,
      "grad_norm": 0.4088869094848633,
      "learning_rate": 5.6560481354807625e-08,
      "loss": 0.4624,
      "step": 333
    },
    {
      "epoch": 1.9186213499281952,
      "grad_norm": 0.3656178414821625,
      "learning_rate": 4.928246485383148e-08,
      "loss": 0.5049,
      "step": 334
    },
    {
      "epoch": 1.9243657252273816,
      "grad_norm": 0.39486929774284363,
      "learning_rate": 4.250318992797375e-08,
      "loss": 0.5062,
      "step": 335
    },
    {
      "epoch": 1.9301101005265677,
      "grad_norm": 0.39059191942214966,
      "learning_rate": 3.622333953000601e-08,
      "loss": 0.4485,
      "step": 336
    },
    {
      "epoch": 1.9358544758257539,
      "grad_norm": 0.3584195673465729,
      "learning_rate": 3.0443546300035764e-08,
      "loss": 0.4358,
      "step": 337
    },
    {
      "epoch": 1.9415988511249402,
      "grad_norm": 0.3339657187461853,
      "learning_rate": 2.516439250177749e-08,
      "loss": 0.46,
      "step": 338
    },
    {
      "epoch": 1.9473432264241264,
      "grad_norm": 0.36130189895629883,
      "learning_rate": 2.038640996389285e-08,
      "loss": 0.5263,
      "step": 339
    },
    {
      "epoch": 1.9530876017233125,
      "grad_norm": 0.3555588722229004,
      "learning_rate": 1.6110080026414123e-08,
      "loss": 0.4882,
      "step": 340
    },
    {
      "epoch": 1.958831977022499,
      "grad_norm": 0.38061201572418213,
      "learning_rate": 1.2335833492252425e-08,
      "loss": 0.4988,
      "step": 341
    },
    {
      "epoch": 1.964576352321685,
      "grad_norm": 0.3624252676963806,
      "learning_rate": 9.06405058380022e-09,
      "loss": 0.5049,
      "step": 342
    },
    {
      "epoch": 1.9703207276208712,
      "grad_norm": 0.39548906683921814,
      "learning_rate": 6.295060904623618e-09,
      "loss": 0.4578,
      "step": 343
    },
    {
      "epoch": 1.9760651029200575,
      "grad_norm": 0.37694376707077026,
      "learning_rate": 4.02914340626226e-09,
      "loss": 0.4775,
      "step": 344
    },
    {
      "epoch": 1.9818094782192437,
      "grad_norm": 0.3581026494503021,
      "learning_rate": 2.2665263601240328e-09,
      "loss": 0.491,
      "step": 345
    },
    {
      "epoch": 1.9875538535184298,
      "grad_norm": 0.37684017419815063,
      "learning_rate": 1.0073873344895735e-09,
      "loss": 0.5071,
      "step": 346
    },
    {
      "epoch": 1.9932982288176162,
      "grad_norm": 0.3794161379337311,
      "learning_rate": 2.5185317662490547e-10,
      "loss": 0.5096,
      "step": 347
    },
    {
      "epoch": 1.9990426041168023,
      "grad_norm": 0.35005468130111694,
      "learning_rate": 0.0,
      "loss": 0.436,
      "step": 348
    },
    {
      "epoch": 1.9990426041168023,
      "step": 348,
      "total_flos": 358710641426432.0,
      "train_loss": 0.5378552473139488,
      "train_runtime": 32465.3196,
      "train_samples_per_second": 1.029,
      "train_steps_per_second": 0.011
    }
  ],
  "logging_steps": 1,
  "max_steps": 348,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 358710641426432.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}