{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 1977, "global_step": 19769, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 5.0584248065152515e-05, "grad_norm": 50.5782356262207, "learning_rate": 5.050505050505051e-08, "loss": 1.1121, "step": 1 }, { "epoch": 0.00010116849613030503, "grad_norm": 60.670440673828125, "learning_rate": 1.0101010101010103e-07, "loss": 1.1474, "step": 2 }, { "epoch": 0.00015175274419545754, "grad_norm": 104.18495178222656, "learning_rate": 1.5151515151515152e-07, "loss": 1.1878, "step": 3 }, { "epoch": 0.00020233699226061006, "grad_norm": 67.87019348144531, "learning_rate": 2.0202020202020205e-07, "loss": 1.0376, "step": 4 }, { "epoch": 0.00025292124032576257, "grad_norm": 72.19010925292969, "learning_rate": 2.525252525252525e-07, "loss": 1.0976, "step": 5 }, { "epoch": 0.0003035054883909151, "grad_norm": 69.45247650146484, "learning_rate": 3.0303030303030305e-07, "loss": 1.0494, "step": 6 }, { "epoch": 0.0003540897364560676, "grad_norm": 51.570796966552734, "learning_rate": 3.535353535353536e-07, "loss": 1.0321, "step": 7 }, { "epoch": 0.0004046739845212201, "grad_norm": 46.872520446777344, "learning_rate": 4.040404040404041e-07, "loss": 1.0329, "step": 8 }, { "epoch": 0.00045525823258637263, "grad_norm": 378.68212890625, "learning_rate": 4.5454545454545457e-07, "loss": 1.2824, "step": 9 }, { "epoch": 0.0005058424806515251, "grad_norm": 103.92955780029297, "learning_rate": 5.05050505050505e-07, "loss": 1.0873, "step": 10 }, { "epoch": 0.0005564267287166777, "grad_norm": 84.58622741699219, "learning_rate": 5.555555555555555e-07, "loss": 1.2509, "step": 11 }, { "epoch": 0.0006070109767818302, "grad_norm": 63.53932571411133, "learning_rate": 6.060606060606061e-07, "loss": 1.0863, "step": 12 }, { "epoch": 0.0006575952248469827, "grad_norm": 44.65436553955078, "learning_rate": 6.565656565656567e-07, "loss": 0.944, "step": 13 }, { "epoch": 0.0007081794729121352, "grad_norm": 65.78855895996094, "learning_rate": 7.070707070707071e-07, "loss": 1.1806, "step": 14 }, { "epoch": 0.0007587637209772877, "grad_norm": 76.02718353271484, "learning_rate": 7.575757575757576e-07, "loss": 1.1739, "step": 15 }, { "epoch": 0.0008093479690424402, "grad_norm": 69.14771270751953, "learning_rate": 8.080808080808082e-07, "loss": 1.0629, "step": 16 }, { "epoch": 0.0008599322171075927, "grad_norm": 46.85115051269531, "learning_rate": 8.585858585858587e-07, "loss": 1.0142, "step": 17 }, { "epoch": 0.0009105164651727453, "grad_norm": 44.566654205322266, "learning_rate": 9.090909090909091e-07, "loss": 1.0526, "step": 18 }, { "epoch": 0.0009611007132378977, "grad_norm": 45.50556564331055, "learning_rate": 9.595959595959596e-07, "loss": 0.9666, "step": 19 }, { "epoch": 0.0010116849613030503, "grad_norm": 48.91423034667969, "learning_rate": 1.01010101010101e-06, "loss": 0.9167, "step": 20 }, { "epoch": 0.0010622692093682027, "grad_norm": 37.258235931396484, "learning_rate": 1.0606060606060608e-06, "loss": 0.9107, "step": 21 }, { "epoch": 0.0011128534574333553, "grad_norm": 54.91384506225586, "learning_rate": 1.111111111111111e-06, "loss": 1.028, "step": 22 }, { "epoch": 0.0011634377054985077, "grad_norm": 56.112796783447266, "learning_rate": 1.1616161616161617e-06, "loss": 1.0174, "step": 23 }, { "epoch": 0.0012140219535636603, "grad_norm": 27.967382431030273, "learning_rate": 1.2121212121212122e-06, "loss": 0.9006, "step": 24 }, { "epoch": 0.0012646062016288128, "grad_norm": 26.931421279907227, "learning_rate": 1.2626262626262629e-06, "loss": 0.9414, "step": 25 }, { "epoch": 0.0013151904496939654, "grad_norm": 38.67788314819336, "learning_rate": 1.3131313131313134e-06, "loss": 0.9898, "step": 26 }, { "epoch": 0.0013657746977591178, "grad_norm": 17.901771545410156, "learning_rate": 1.3636363636363636e-06, "loss": 0.857, "step": 27 }, { "epoch": 0.0014163589458242704, "grad_norm": 26.71088409423828, "learning_rate": 1.4141414141414143e-06, "loss": 0.8012, "step": 28 }, { "epoch": 0.0014669431938894228, "grad_norm": 26.47563934326172, "learning_rate": 1.4646464646464648e-06, "loss": 0.8503, "step": 29 }, { "epoch": 0.0015175274419545754, "grad_norm": 44.35660934448242, "learning_rate": 1.5151515151515152e-06, "loss": 0.8765, "step": 30 }, { "epoch": 0.0015681116900197278, "grad_norm": 22.451826095581055, "learning_rate": 1.565656565656566e-06, "loss": 0.9015, "step": 31 }, { "epoch": 0.0016186959380848805, "grad_norm": 27.244152069091797, "learning_rate": 1.6161616161616164e-06, "loss": 0.9022, "step": 32 }, { "epoch": 0.0016692801861500329, "grad_norm": 18.157508850097656, "learning_rate": 1.6666666666666667e-06, "loss": 0.8887, "step": 33 }, { "epoch": 0.0017198644342151855, "grad_norm": 30.875972747802734, "learning_rate": 1.7171717171717173e-06, "loss": 0.7552, "step": 34 }, { "epoch": 0.001770448682280338, "grad_norm": 28.64618682861328, "learning_rate": 1.7676767676767678e-06, "loss": 0.9693, "step": 35 }, { "epoch": 0.0018210329303454905, "grad_norm": 13.883614540100098, "learning_rate": 1.8181818181818183e-06, "loss": 0.8683, "step": 36 }, { "epoch": 0.001871617178410643, "grad_norm": 20.347986221313477, "learning_rate": 1.868686868686869e-06, "loss": 0.8658, "step": 37 }, { "epoch": 0.0019222014264757953, "grad_norm": 23.266088485717773, "learning_rate": 1.9191919191919192e-06, "loss": 0.8289, "step": 38 }, { "epoch": 0.0019727856745409477, "grad_norm": 17.252241134643555, "learning_rate": 1.96969696969697e-06, "loss": 0.8629, "step": 39 }, { "epoch": 0.0020233699226061006, "grad_norm": 31.024005889892578, "learning_rate": 2.02020202020202e-06, "loss": 0.9608, "step": 40 }, { "epoch": 0.002073954170671253, "grad_norm": 9.084613800048828, "learning_rate": 2.070707070707071e-06, "loss": 0.8545, "step": 41 }, { "epoch": 0.0021245384187364054, "grad_norm": 17.46491241455078, "learning_rate": 2.1212121212121216e-06, "loss": 0.9101, "step": 42 }, { "epoch": 0.002175122666801558, "grad_norm": 18.7784423828125, "learning_rate": 2.171717171717172e-06, "loss": 0.8602, "step": 43 }, { "epoch": 0.0022257069148667106, "grad_norm": 15.92908000946045, "learning_rate": 2.222222222222222e-06, "loss": 0.9321, "step": 44 }, { "epoch": 0.002276291162931863, "grad_norm": 13.930456161499023, "learning_rate": 2.2727272727272728e-06, "loss": 0.8713, "step": 45 }, { "epoch": 0.0023268754109970155, "grad_norm": 20.67007827758789, "learning_rate": 2.3232323232323234e-06, "loss": 0.8195, "step": 46 }, { "epoch": 0.002377459659062168, "grad_norm": 20.0476016998291, "learning_rate": 2.373737373737374e-06, "loss": 0.9206, "step": 47 }, { "epoch": 0.0024280439071273207, "grad_norm": 18.066862106323242, "learning_rate": 2.4242424242424244e-06, "loss": 0.904, "step": 48 }, { "epoch": 0.002478628155192473, "grad_norm": 15.198264122009277, "learning_rate": 2.474747474747475e-06, "loss": 0.8272, "step": 49 }, { "epoch": 0.0025292124032576255, "grad_norm": 28.15618324279785, "learning_rate": 2.5252525252525258e-06, "loss": 0.9647, "step": 50 }, { "epoch": 0.002579796651322778, "grad_norm": 20.124496459960938, "learning_rate": 2.575757575757576e-06, "loss": 0.8217, "step": 51 }, { "epoch": 0.0026303808993879308, "grad_norm": 14.76452350616455, "learning_rate": 2.6262626262626267e-06, "loss": 0.8211, "step": 52 }, { "epoch": 0.002680965147453083, "grad_norm": 23.33282470703125, "learning_rate": 2.676767676767677e-06, "loss": 0.9072, "step": 53 }, { "epoch": 0.0027315493955182356, "grad_norm": 17.225427627563477, "learning_rate": 2.7272727272727272e-06, "loss": 0.8683, "step": 54 }, { "epoch": 0.002782133643583388, "grad_norm": 28.340431213378906, "learning_rate": 2.7777777777777783e-06, "loss": 0.9274, "step": 55 }, { "epoch": 0.002832717891648541, "grad_norm": 22.247989654541016, "learning_rate": 2.8282828282828286e-06, "loss": 0.877, "step": 56 }, { "epoch": 0.0028833021397136932, "grad_norm": 39.226043701171875, "learning_rate": 2.8787878787878793e-06, "loss": 0.8464, "step": 57 }, { "epoch": 0.0029338863877788456, "grad_norm": 21.533878326416016, "learning_rate": 2.9292929292929295e-06, "loss": 0.9234, "step": 58 }, { "epoch": 0.002984470635843998, "grad_norm": 16.03548240661621, "learning_rate": 2.97979797979798e-06, "loss": 0.8588, "step": 59 }, { "epoch": 0.003035054883909151, "grad_norm": 25.47140121459961, "learning_rate": 3.0303030303030305e-06, "loss": 0.7863, "step": 60 }, { "epoch": 0.0030856391319743033, "grad_norm": 13.238901138305664, "learning_rate": 3.0808080808080807e-06, "loss": 0.8352, "step": 61 }, { "epoch": 0.0031362233800394557, "grad_norm": 33.06532287597656, "learning_rate": 3.131313131313132e-06, "loss": 0.7741, "step": 62 }, { "epoch": 0.003186807628104608, "grad_norm": 18.614259719848633, "learning_rate": 3.181818181818182e-06, "loss": 0.8586, "step": 63 }, { "epoch": 0.003237391876169761, "grad_norm": 27.09310531616211, "learning_rate": 3.232323232323233e-06, "loss": 0.846, "step": 64 }, { "epoch": 0.0032879761242349133, "grad_norm": 20.295303344726562, "learning_rate": 3.282828282828283e-06, "loss": 0.7887, "step": 65 }, { "epoch": 0.0033385603723000657, "grad_norm": 20.94864845275879, "learning_rate": 3.3333333333333333e-06, "loss": 0.9103, "step": 66 }, { "epoch": 0.003389144620365218, "grad_norm": 25.763141632080078, "learning_rate": 3.3838383838383844e-06, "loss": 0.8935, "step": 67 }, { "epoch": 0.003439728868430371, "grad_norm": 24.435693740844727, "learning_rate": 3.4343434343434347e-06, "loss": 0.9108, "step": 68 }, { "epoch": 0.0034903131164955234, "grad_norm": 23.003145217895508, "learning_rate": 3.4848484848484854e-06, "loss": 0.8094, "step": 69 }, { "epoch": 0.003540897364560676, "grad_norm": 15.974407196044922, "learning_rate": 3.5353535353535356e-06, "loss": 0.825, "step": 70 }, { "epoch": 0.003591481612625828, "grad_norm": 22.897077560424805, "learning_rate": 3.585858585858586e-06, "loss": 0.9144, "step": 71 }, { "epoch": 0.003642065860690981, "grad_norm": 18.80307388305664, "learning_rate": 3.6363636363636366e-06, "loss": 0.7924, "step": 72 }, { "epoch": 0.0036926501087561335, "grad_norm": 23.70836067199707, "learning_rate": 3.686868686868687e-06, "loss": 0.8593, "step": 73 }, { "epoch": 0.003743234356821286, "grad_norm": 28.325363159179688, "learning_rate": 3.737373737373738e-06, "loss": 0.8027, "step": 74 }, { "epoch": 0.0037938186048864383, "grad_norm": 15.749767303466797, "learning_rate": 3.7878787878787882e-06, "loss": 0.8629, "step": 75 }, { "epoch": 0.0038444028529515907, "grad_norm": 16.129369735717773, "learning_rate": 3.8383838383838385e-06, "loss": 0.8582, "step": 76 }, { "epoch": 0.0038949871010167435, "grad_norm": 17.864240646362305, "learning_rate": 3.88888888888889e-06, "loss": 0.8503, "step": 77 }, { "epoch": 0.0039455713490818955, "grad_norm": 224.7100372314453, "learning_rate": 3.93939393939394e-06, "loss": 1.3872, "step": 78 }, { "epoch": 0.003996155597147048, "grad_norm": 16.356430053710938, "learning_rate": 3.98989898989899e-06, "loss": 0.7829, "step": 79 }, { "epoch": 0.004046739845212201, "grad_norm": 17.523164749145508, "learning_rate": 4.04040404040404e-06, "loss": 0.7901, "step": 80 }, { "epoch": 0.004097324093277353, "grad_norm": 16.012954711914062, "learning_rate": 4.0909090909090915e-06, "loss": 0.8018, "step": 81 }, { "epoch": 0.004147908341342506, "grad_norm": 14.424223899841309, "learning_rate": 4.141414141414142e-06, "loss": 0.745, "step": 82 }, { "epoch": 0.004198492589407659, "grad_norm": 22.008962631225586, "learning_rate": 4.191919191919192e-06, "loss": 0.7462, "step": 83 }, { "epoch": 0.004249076837472811, "grad_norm": 16.8351993560791, "learning_rate": 4.242424242424243e-06, "loss": 0.7592, "step": 84 }, { "epoch": 0.004299661085537964, "grad_norm": 37.292877197265625, "learning_rate": 4.292929292929293e-06, "loss": 0.7998, "step": 85 }, { "epoch": 0.004350245333603116, "grad_norm": 29.97257423400879, "learning_rate": 4.343434343434344e-06, "loss": 0.8272, "step": 86 }, { "epoch": 0.0044008295816682684, "grad_norm": 35.998138427734375, "learning_rate": 4.393939393939394e-06, "loss": 0.7571, "step": 87 }, { "epoch": 0.004451413829733421, "grad_norm": 30.172080993652344, "learning_rate": 4.444444444444444e-06, "loss": 0.7593, "step": 88 }, { "epoch": 0.004501998077798573, "grad_norm": 24.991302490234375, "learning_rate": 4.494949494949495e-06, "loss": 0.7729, "step": 89 }, { "epoch": 0.004552582325863726, "grad_norm": 25.77794075012207, "learning_rate": 4.5454545454545455e-06, "loss": 0.7902, "step": 90 }, { "epoch": 0.004603166573928879, "grad_norm": 35.670166015625, "learning_rate": 4.595959595959597e-06, "loss": 0.8492, "step": 91 }, { "epoch": 0.004653750821994031, "grad_norm": 25.642837524414062, "learning_rate": 4.646464646464647e-06, "loss": 0.8804, "step": 92 }, { "epoch": 0.004704335070059184, "grad_norm": 33.92631912231445, "learning_rate": 4.696969696969698e-06, "loss": 0.8682, "step": 93 }, { "epoch": 0.004754919318124336, "grad_norm": 29.576330184936523, "learning_rate": 4.747474747474748e-06, "loss": 0.7035, "step": 94 }, { "epoch": 0.0048055035661894886, "grad_norm": 29.576967239379883, "learning_rate": 4.7979797979797985e-06, "loss": 0.6663, "step": 95 }, { "epoch": 0.004856087814254641, "grad_norm": 27.043075561523438, "learning_rate": 4.848484848484849e-06, "loss": 0.7649, "step": 96 }, { "epoch": 0.004906672062319793, "grad_norm": 17.66451072692871, "learning_rate": 4.898989898989899e-06, "loss": 0.7139, "step": 97 }, { "epoch": 0.004957256310384946, "grad_norm": 16.692092895507812, "learning_rate": 4.94949494949495e-06, "loss": 0.6626, "step": 98 }, { "epoch": 0.005007840558450099, "grad_norm": 37.80686950683594, "learning_rate": 5e-06, "loss": 0.568, "step": 99 }, { "epoch": 0.005058424806515251, "grad_norm": 27.32476043701172, "learning_rate": 5.0505050505050515e-06, "loss": 0.6879, "step": 100 }, { "epoch": 0.005109009054580404, "grad_norm": 24.884183883666992, "learning_rate": 5.101010101010101e-06, "loss": 0.691, "step": 101 }, { "epoch": 0.005159593302645556, "grad_norm": 59.504737854003906, "learning_rate": 5.151515151515152e-06, "loss": 0.8777, "step": 102 }, { "epoch": 0.005210177550710709, "grad_norm": 26.967599868774414, "learning_rate": 5.202020202020202e-06, "loss": 0.5092, "step": 103 }, { "epoch": 0.0052607617987758615, "grad_norm": 29.929872512817383, "learning_rate": 5.252525252525253e-06, "loss": 0.6947, "step": 104 }, { "epoch": 0.0053113460468410135, "grad_norm": 30.953746795654297, "learning_rate": 5.303030303030303e-06, "loss": 0.6463, "step": 105 }, { "epoch": 0.005361930294906166, "grad_norm": 27.775087356567383, "learning_rate": 5.353535353535354e-06, "loss": 0.7681, "step": 106 }, { "epoch": 0.005412514542971319, "grad_norm": 25.878192901611328, "learning_rate": 5.404040404040405e-06, "loss": 0.6898, "step": 107 }, { "epoch": 0.005463098791036471, "grad_norm": 23.883628845214844, "learning_rate": 5.4545454545454545e-06, "loss": 0.7956, "step": 108 }, { "epoch": 0.005513683039101624, "grad_norm": 25.763141632080078, "learning_rate": 5.5050505050505056e-06, "loss": 0.6388, "step": 109 }, { "epoch": 0.005564267287166776, "grad_norm": 39.20890808105469, "learning_rate": 5.555555555555557e-06, "loss": 0.5935, "step": 110 }, { "epoch": 0.005614851535231929, "grad_norm": 25.359010696411133, "learning_rate": 5.606060606060606e-06, "loss": 0.7161, "step": 111 }, { "epoch": 0.005665435783297082, "grad_norm": 28.834171295166016, "learning_rate": 5.656565656565657e-06, "loss": 0.4962, "step": 112 }, { "epoch": 0.005716020031362234, "grad_norm": 27.819393157958984, "learning_rate": 5.7070707070707075e-06, "loss": 0.5369, "step": 113 }, { "epoch": 0.0057666042794273864, "grad_norm": 36.43419647216797, "learning_rate": 5.7575757575757586e-06, "loss": 0.7811, "step": 114 }, { "epoch": 0.005817188527492538, "grad_norm": 21.626033782958984, "learning_rate": 5.808080808080808e-06, "loss": 0.6441, "step": 115 }, { "epoch": 0.005867772775557691, "grad_norm": 23.996023178100586, "learning_rate": 5.858585858585859e-06, "loss": 0.6542, "step": 116 }, { "epoch": 0.005918357023622844, "grad_norm": 27.249528884887695, "learning_rate": 5.90909090909091e-06, "loss": 0.7374, "step": 117 }, { "epoch": 0.005968941271687996, "grad_norm": 21.578083038330078, "learning_rate": 5.95959595959596e-06, "loss": 0.6842, "step": 118 }, { "epoch": 0.006019525519753149, "grad_norm": 21.514127731323242, "learning_rate": 6.010101010101011e-06, "loss": 0.5835, "step": 119 }, { "epoch": 0.006070109767818302, "grad_norm": 22.144807815551758, "learning_rate": 6.060606060606061e-06, "loss": 0.7269, "step": 120 }, { "epoch": 0.006120694015883454, "grad_norm": 22.411209106445312, "learning_rate": 6.111111111111112e-06, "loss": 0.7505, "step": 121 }, { "epoch": 0.0061712782639486066, "grad_norm": 38.29654312133789, "learning_rate": 6.1616161616161615e-06, "loss": 0.6834, "step": 122 }, { "epoch": 0.0062218625120137585, "grad_norm": 29.639760971069336, "learning_rate": 6.212121212121213e-06, "loss": 0.5401, "step": 123 }, { "epoch": 0.006272446760078911, "grad_norm": 17.06061553955078, "learning_rate": 6.262626262626264e-06, "loss": 0.5621, "step": 124 }, { "epoch": 0.006323031008144064, "grad_norm": 43.462799072265625, "learning_rate": 6.313131313131313e-06, "loss": 0.7536, "step": 125 }, { "epoch": 0.006373615256209216, "grad_norm": 26.978931427001953, "learning_rate": 6.363636363636364e-06, "loss": 0.5106, "step": 126 }, { "epoch": 0.006424199504274369, "grad_norm": 30.509258270263672, "learning_rate": 6.4141414141414145e-06, "loss": 0.668, "step": 127 }, { "epoch": 0.006474783752339522, "grad_norm": 36.92049789428711, "learning_rate": 6.464646464646466e-06, "loss": 0.7861, "step": 128 }, { "epoch": 0.006525368000404674, "grad_norm": 27.675073623657227, "learning_rate": 6.515151515151516e-06, "loss": 0.5674, "step": 129 }, { "epoch": 0.006575952248469827, "grad_norm": 20.432708740234375, "learning_rate": 6.565656565656566e-06, "loss": 0.5205, "step": 130 }, { "epoch": 0.006626536496534979, "grad_norm": 27.352642059326172, "learning_rate": 6.616161616161617e-06, "loss": 0.4855, "step": 131 }, { "epoch": 0.0066771207446001315, "grad_norm": 17.609615325927734, "learning_rate": 6.666666666666667e-06, "loss": 0.5658, "step": 132 }, { "epoch": 0.006727704992665284, "grad_norm": 18.96381378173828, "learning_rate": 6.717171717171718e-06, "loss": 0.6414, "step": 133 }, { "epoch": 0.006778289240730436, "grad_norm": 17.264280319213867, "learning_rate": 6.767676767676769e-06, "loss": 0.682, "step": 134 }, { "epoch": 0.006828873488795589, "grad_norm": 22.183008193969727, "learning_rate": 6.818181818181818e-06, "loss": 0.59, "step": 135 }, { "epoch": 0.006879457736860742, "grad_norm": 21.94026756286621, "learning_rate": 6.868686868686869e-06, "loss": 0.7015, "step": 136 }, { "epoch": 0.006930041984925894, "grad_norm": 22.582977294921875, "learning_rate": 6.91919191919192e-06, "loss": 0.495, "step": 137 }, { "epoch": 0.006980626232991047, "grad_norm": 15.771801948547363, "learning_rate": 6.969696969696971e-06, "loss": 0.625, "step": 138 }, { "epoch": 0.007031210481056199, "grad_norm": 14.919344902038574, "learning_rate": 7.02020202020202e-06, "loss": 0.5367, "step": 139 }, { "epoch": 0.007081794729121352, "grad_norm": 19.314231872558594, "learning_rate": 7.070707070707071e-06, "loss": 0.6514, "step": 140 }, { "epoch": 0.0071323789771865044, "grad_norm": 30.883546829223633, "learning_rate": 7.121212121212122e-06, "loss": 0.7386, "step": 141 }, { "epoch": 0.007182963225251656, "grad_norm": 26.993871688842773, "learning_rate": 7.171717171717172e-06, "loss": 0.5671, "step": 142 }, { "epoch": 0.007233547473316809, "grad_norm": 35.63401412963867, "learning_rate": 7.222222222222223e-06, "loss": 0.6853, "step": 143 }, { "epoch": 0.007284131721381962, "grad_norm": 36.72064971923828, "learning_rate": 7.272727272727273e-06, "loss": 0.5843, "step": 144 }, { "epoch": 0.007334715969447114, "grad_norm": 21.811140060424805, "learning_rate": 7.323232323232324e-06, "loss": 0.5436, "step": 145 }, { "epoch": 0.007385300217512267, "grad_norm": 26.727733612060547, "learning_rate": 7.373737373737374e-06, "loss": 0.8076, "step": 146 }, { "epoch": 0.007435884465577419, "grad_norm": 26.84775733947754, "learning_rate": 7.424242424242425e-06, "loss": 0.7409, "step": 147 }, { "epoch": 0.007486468713642572, "grad_norm": 25.79094123840332, "learning_rate": 7.474747474747476e-06, "loss": 0.6482, "step": 148 }, { "epoch": 0.0075370529617077246, "grad_norm": 26.548110961914062, "learning_rate": 7.525252525252525e-06, "loss": 0.621, "step": 149 }, { "epoch": 0.0075876372097728765, "grad_norm": 50.205657958984375, "learning_rate": 7.5757575757575764e-06, "loss": 0.6445, "step": 150 }, { "epoch": 0.007638221457838029, "grad_norm": 30.54366111755371, "learning_rate": 7.6262626262626275e-06, "loss": 0.6271, "step": 151 }, { "epoch": 0.007688805705903181, "grad_norm": 19.769853591918945, "learning_rate": 7.676767676767677e-06, "loss": 0.6067, "step": 152 }, { "epoch": 0.007739389953968334, "grad_norm": 17.965391159057617, "learning_rate": 7.727272727272727e-06, "loss": 0.5416, "step": 153 }, { "epoch": 0.007789974202033487, "grad_norm": 15.6107816696167, "learning_rate": 7.77777777777778e-06, "loss": 0.6566, "step": 154 }, { "epoch": 0.00784055845009864, "grad_norm": 21.21352195739746, "learning_rate": 7.82828282828283e-06, "loss": 0.5619, "step": 155 }, { "epoch": 0.007891142698163791, "grad_norm": 24.089017868041992, "learning_rate": 7.87878787878788e-06, "loss": 0.6513, "step": 156 }, { "epoch": 0.007941726946228944, "grad_norm": 22.450519561767578, "learning_rate": 7.92929292929293e-06, "loss": 0.6536, "step": 157 }, { "epoch": 0.007992311194294097, "grad_norm": 22.72298240661621, "learning_rate": 7.97979797979798e-06, "loss": 0.5981, "step": 158 }, { "epoch": 0.00804289544235925, "grad_norm": 14.06804084777832, "learning_rate": 8.03030303030303e-06, "loss": 0.5339, "step": 159 }, { "epoch": 0.008093479690424402, "grad_norm": 24.473007202148438, "learning_rate": 8.08080808080808e-06, "loss": 0.6634, "step": 160 }, { "epoch": 0.008144063938489555, "grad_norm": 26.42620086669922, "learning_rate": 8.131313131313133e-06, "loss": 0.6558, "step": 161 }, { "epoch": 0.008194648186554706, "grad_norm": 36.58184814453125, "learning_rate": 8.181818181818183e-06, "loss": 0.7288, "step": 162 }, { "epoch": 0.008245232434619859, "grad_norm": 20.66307258605957, "learning_rate": 8.232323232323233e-06, "loss": 0.6211, "step": 163 }, { "epoch": 0.008295816682685012, "grad_norm": 17.057435989379883, "learning_rate": 8.282828282828283e-06, "loss": 0.4814, "step": 164 }, { "epoch": 0.008346400930750165, "grad_norm": 25.118968963623047, "learning_rate": 8.333333333333334e-06, "loss": 0.6334, "step": 165 }, { "epoch": 0.008396985178815318, "grad_norm": 20.36933708190918, "learning_rate": 8.383838383838384e-06, "loss": 0.6281, "step": 166 }, { "epoch": 0.008447569426880469, "grad_norm": 14.90415096282959, "learning_rate": 8.434343434343434e-06, "loss": 0.7382, "step": 167 }, { "epoch": 0.008498153674945622, "grad_norm": 25.610687255859375, "learning_rate": 8.484848484848486e-06, "loss": 0.6551, "step": 168 }, { "epoch": 0.008548737923010774, "grad_norm": 34.631587982177734, "learning_rate": 8.535353535353535e-06, "loss": 0.6097, "step": 169 }, { "epoch": 0.008599322171075927, "grad_norm": 32.757301330566406, "learning_rate": 8.585858585858587e-06, "loss": 0.6602, "step": 170 }, { "epoch": 0.00864990641914108, "grad_norm": 21.76865577697754, "learning_rate": 8.636363636363637e-06, "loss": 0.5711, "step": 171 }, { "epoch": 0.008700490667206231, "grad_norm": 19.133024215698242, "learning_rate": 8.686868686868687e-06, "loss": 0.5733, "step": 172 }, { "epoch": 0.008751074915271384, "grad_norm": 11.9871187210083, "learning_rate": 8.737373737373738e-06, "loss": 0.6085, "step": 173 }, { "epoch": 0.008801659163336537, "grad_norm": 33.54527282714844, "learning_rate": 8.787878787878788e-06, "loss": 0.7065, "step": 174 }, { "epoch": 0.00885224341140169, "grad_norm": 30.19257926940918, "learning_rate": 8.83838383838384e-06, "loss": 0.6561, "step": 175 }, { "epoch": 0.008902827659466843, "grad_norm": 20.798595428466797, "learning_rate": 8.888888888888888e-06, "loss": 0.6796, "step": 176 }, { "epoch": 0.008953411907531995, "grad_norm": 18.61840057373047, "learning_rate": 8.93939393939394e-06, "loss": 0.6547, "step": 177 }, { "epoch": 0.009003996155597147, "grad_norm": 12.906478881835938, "learning_rate": 8.98989898989899e-06, "loss": 0.5022, "step": 178 }, { "epoch": 0.0090545804036623, "grad_norm": 22.467634201049805, "learning_rate": 9.040404040404042e-06, "loss": 0.5592, "step": 179 }, { "epoch": 0.009105164651727452, "grad_norm": 17.99599838256836, "learning_rate": 9.090909090909091e-06, "loss": 0.5751, "step": 180 }, { "epoch": 0.009155748899792605, "grad_norm": 30.549976348876953, "learning_rate": 9.141414141414143e-06, "loss": 0.558, "step": 181 }, { "epoch": 0.009206333147857758, "grad_norm": 20.566837310791016, "learning_rate": 9.191919191919193e-06, "loss": 0.5925, "step": 182 }, { "epoch": 0.009256917395922909, "grad_norm": 32.81270980834961, "learning_rate": 9.242424242424244e-06, "loss": 0.5524, "step": 183 }, { "epoch": 0.009307501643988062, "grad_norm": 29.030202865600586, "learning_rate": 9.292929292929294e-06, "loss": 0.5629, "step": 184 }, { "epoch": 0.009358085892053215, "grad_norm": 26.472116470336914, "learning_rate": 9.343434343434344e-06, "loss": 0.5574, "step": 185 }, { "epoch": 0.009408670140118367, "grad_norm": 15.020588874816895, "learning_rate": 9.393939393939396e-06, "loss": 0.4887, "step": 186 }, { "epoch": 0.00945925438818352, "grad_norm": 36.62296676635742, "learning_rate": 9.444444444444445e-06, "loss": 0.8239, "step": 187 }, { "epoch": 0.009509838636248671, "grad_norm": 14.52802848815918, "learning_rate": 9.494949494949497e-06, "loss": 0.5408, "step": 188 }, { "epoch": 0.009560422884313824, "grad_norm": 22.754674911499023, "learning_rate": 9.545454545454547e-06, "loss": 0.5521, "step": 189 }, { "epoch": 0.009611007132378977, "grad_norm": 20.58759117126465, "learning_rate": 9.595959595959597e-06, "loss": 0.669, "step": 190 }, { "epoch": 0.00966159138044413, "grad_norm": 15.862406730651855, "learning_rate": 9.646464646464647e-06, "loss": 0.7316, "step": 191 }, { "epoch": 0.009712175628509283, "grad_norm": 24.044376373291016, "learning_rate": 9.696969696969698e-06, "loss": 0.5534, "step": 192 }, { "epoch": 0.009762759876574434, "grad_norm": 28.88985824584961, "learning_rate": 9.747474747474748e-06, "loss": 0.5342, "step": 193 }, { "epoch": 0.009813344124639587, "grad_norm": 19.496883392333984, "learning_rate": 9.797979797979798e-06, "loss": 0.7268, "step": 194 }, { "epoch": 0.00986392837270474, "grad_norm": 15.816526412963867, "learning_rate": 9.84848484848485e-06, "loss": 0.5817, "step": 195 }, { "epoch": 0.009914512620769892, "grad_norm": 17.75444984436035, "learning_rate": 9.8989898989899e-06, "loss": 0.7115, "step": 196 }, { "epoch": 0.009965096868835045, "grad_norm": 21.01492691040039, "learning_rate": 9.94949494949495e-06, "loss": 0.6263, "step": 197 }, { "epoch": 0.010015681116900198, "grad_norm": 12.768424987792969, "learning_rate": 1e-05, "loss": 0.484, "step": 198 }, { "epoch": 0.01006626536496535, "grad_norm": 22.86803436279297, "learning_rate": 9.99999993558104e-06, "loss": 0.5808, "step": 199 }, { "epoch": 0.010116849613030502, "grad_norm": 19.393211364746094, "learning_rate": 9.999999742324155e-06, "loss": 0.5373, "step": 200 }, { "epoch": 0.010167433861095655, "grad_norm": 15.989542961120605, "learning_rate": 9.999999420229357e-06, "loss": 0.4862, "step": 201 }, { "epoch": 0.010218018109160808, "grad_norm": 12.684467315673828, "learning_rate": 9.99999896929665e-06, "loss": 0.5334, "step": 202 }, { "epoch": 0.01026860235722596, "grad_norm": 41.0718994140625, "learning_rate": 9.999998389526046e-06, "loss": 0.7022, "step": 203 }, { "epoch": 0.010319186605291112, "grad_norm": 12.49349308013916, "learning_rate": 9.999997680917561e-06, "loss": 0.546, "step": 204 }, { "epoch": 0.010369770853356265, "grad_norm": 18.726335525512695, "learning_rate": 9.999996843471214e-06, "loss": 0.6882, "step": 205 }, { "epoch": 0.010420355101421417, "grad_norm": 19.217233657836914, "learning_rate": 9.999995877187022e-06, "loss": 0.4872, "step": 206 }, { "epoch": 0.01047093934948657, "grad_norm": 20.728586196899414, "learning_rate": 9.999994782065016e-06, "loss": 0.6746, "step": 207 }, { "epoch": 0.010521523597551723, "grad_norm": 17.297653198242188, "learning_rate": 9.999993558105222e-06, "loss": 0.5946, "step": 208 }, { "epoch": 0.010572107845616874, "grad_norm": 54.1619758605957, "learning_rate": 9.999992205307668e-06, "loss": 0.7518, "step": 209 }, { "epoch": 0.010622692093682027, "grad_norm": 26.165761947631836, "learning_rate": 9.999990723672394e-06, "loss": 0.5621, "step": 210 }, { "epoch": 0.01067327634174718, "grad_norm": 19.489540100097656, "learning_rate": 9.999989113199435e-06, "loss": 0.477, "step": 211 }, { "epoch": 0.010723860589812333, "grad_norm": 15.422208786010742, "learning_rate": 9.999987373888834e-06, "loss": 0.4626, "step": 212 }, { "epoch": 0.010774444837877485, "grad_norm": 25.755903244018555, "learning_rate": 9.999985505740635e-06, "loss": 0.6255, "step": 213 }, { "epoch": 0.010825029085942638, "grad_norm": 27.325439453125, "learning_rate": 9.999983508754887e-06, "loss": 0.7381, "step": 214 }, { "epoch": 0.01087561333400779, "grad_norm": 35.46827697753906, "learning_rate": 9.999981382931641e-06, "loss": 0.7855, "step": 215 }, { "epoch": 0.010926197582072942, "grad_norm": 33.008033752441406, "learning_rate": 9.999979128270951e-06, "loss": 0.6108, "step": 216 }, { "epoch": 0.010976781830138095, "grad_norm": 37.022735595703125, "learning_rate": 9.999976744772877e-06, "loss": 0.5364, "step": 217 }, { "epoch": 0.011027366078203248, "grad_norm": 33.454837799072266, "learning_rate": 9.999974232437479e-06, "loss": 0.6652, "step": 218 }, { "epoch": 0.0110779503262684, "grad_norm": 13.358001708984375, "learning_rate": 9.999971591264821e-06, "loss": 0.6419, "step": 219 }, { "epoch": 0.011128534574333552, "grad_norm": 15.801761627197266, "learning_rate": 9.999968821254973e-06, "loss": 0.6306, "step": 220 }, { "epoch": 0.011179118822398705, "grad_norm": 32.37762451171875, "learning_rate": 9.999965922408005e-06, "loss": 0.6569, "step": 221 }, { "epoch": 0.011229703070463858, "grad_norm": 19.906774520874023, "learning_rate": 9.999962894723992e-06, "loss": 0.6483, "step": 222 }, { "epoch": 0.01128028731852901, "grad_norm": 17.420452117919922, "learning_rate": 9.999959738203012e-06, "loss": 0.5755, "step": 223 }, { "epoch": 0.011330871566594163, "grad_norm": 24.519367218017578, "learning_rate": 9.999956452845147e-06, "loss": 0.6043, "step": 224 }, { "epoch": 0.011381455814659314, "grad_norm": 18.717021942138672, "learning_rate": 9.99995303865048e-06, "loss": 0.6784, "step": 225 }, { "epoch": 0.011432040062724467, "grad_norm": 28.122739791870117, "learning_rate": 9.999949495619101e-06, "loss": 0.5132, "step": 226 }, { "epoch": 0.01148262431078962, "grad_norm": 8.879480361938477, "learning_rate": 9.9999458237511e-06, "loss": 0.5479, "step": 227 }, { "epoch": 0.011533208558854773, "grad_norm": 17.9887752532959, "learning_rate": 9.999942023046574e-06, "loss": 0.6893, "step": 228 }, { "epoch": 0.011583792806919926, "grad_norm": 30.12117576599121, "learning_rate": 9.999938093505617e-06, "loss": 0.7929, "step": 229 }, { "epoch": 0.011634377054985077, "grad_norm": 22.058338165283203, "learning_rate": 9.999934035128332e-06, "loss": 0.5912, "step": 230 }, { "epoch": 0.01168496130305023, "grad_norm": 13.299651145935059, "learning_rate": 9.999929847914824e-06, "loss": 0.5523, "step": 231 }, { "epoch": 0.011735545551115383, "grad_norm": 42.8521728515625, "learning_rate": 9.9999255318652e-06, "loss": 0.5734, "step": 232 }, { "epoch": 0.011786129799180535, "grad_norm": 18.515382766723633, "learning_rate": 9.999921086979573e-06, "loss": 0.6017, "step": 233 }, { "epoch": 0.011836714047245688, "grad_norm": 13.719735145568848, "learning_rate": 9.999916513258057e-06, "loss": 0.4899, "step": 234 }, { "epoch": 0.011887298295310841, "grad_norm": 27.967578887939453, "learning_rate": 9.999911810700769e-06, "loss": 0.5971, "step": 235 }, { "epoch": 0.011937882543375992, "grad_norm": 19.38290023803711, "learning_rate": 9.999906979307828e-06, "loss": 0.5784, "step": 236 }, { "epoch": 0.011988466791441145, "grad_norm": 20.670820236206055, "learning_rate": 9.999902019079364e-06, "loss": 0.5611, "step": 237 }, { "epoch": 0.012039051039506298, "grad_norm": 35.95936584472656, "learning_rate": 9.999896930015501e-06, "loss": 0.8376, "step": 238 }, { "epoch": 0.01208963528757145, "grad_norm": 20.55226707458496, "learning_rate": 9.99989171211637e-06, "loss": 0.6866, "step": 239 }, { "epoch": 0.012140219535636603, "grad_norm": 18.43644905090332, "learning_rate": 9.999886365382106e-06, "loss": 0.3801, "step": 240 }, { "epoch": 0.012190803783701755, "grad_norm": 19.466358184814453, "learning_rate": 9.999880889812847e-06, "loss": 0.7955, "step": 241 }, { "epoch": 0.012241388031766907, "grad_norm": 15.9744234085083, "learning_rate": 9.999875285408736e-06, "loss": 0.4555, "step": 242 }, { "epoch": 0.01229197227983206, "grad_norm": 8.98836612701416, "learning_rate": 9.999869552169913e-06, "loss": 0.4794, "step": 243 }, { "epoch": 0.012342556527897213, "grad_norm": 14.708904266357422, "learning_rate": 9.99986369009653e-06, "loss": 0.5995, "step": 244 }, { "epoch": 0.012393140775962366, "grad_norm": 12.213827133178711, "learning_rate": 9.999857699188738e-06, "loss": 0.5717, "step": 245 }, { "epoch": 0.012443725024027517, "grad_norm": 28.877830505371094, "learning_rate": 9.999851579446687e-06, "loss": 0.6875, "step": 246 }, { "epoch": 0.01249430927209267, "grad_norm": 20.180156707763672, "learning_rate": 9.999845330870538e-06, "loss": 0.6882, "step": 247 }, { "epoch": 0.012544893520157823, "grad_norm": 21.49137306213379, "learning_rate": 9.999838953460453e-06, "loss": 0.6496, "step": 248 }, { "epoch": 0.012595477768222976, "grad_norm": 14.763259887695312, "learning_rate": 9.999832447216592e-06, "loss": 0.5908, "step": 249 }, { "epoch": 0.012646062016288128, "grad_norm": 34.85430145263672, "learning_rate": 9.999825812139127e-06, "loss": 0.7474, "step": 250 }, { "epoch": 0.012696646264353281, "grad_norm": 15.414175033569336, "learning_rate": 9.999819048228227e-06, "loss": 0.5912, "step": 251 }, { "epoch": 0.012747230512418432, "grad_norm": 18.208133697509766, "learning_rate": 9.999812155484068e-06, "loss": 0.6322, "step": 252 }, { "epoch": 0.012797814760483585, "grad_norm": 29.04851531982422, "learning_rate": 9.999805133906826e-06, "loss": 0.6505, "step": 253 }, { "epoch": 0.012848399008548738, "grad_norm": 15.870619773864746, "learning_rate": 9.999797983496681e-06, "loss": 0.6181, "step": 254 }, { "epoch": 0.012898983256613891, "grad_norm": 15.969467163085938, "learning_rate": 9.99979070425382e-06, "loss": 0.5919, "step": 255 }, { "epoch": 0.012949567504679044, "grad_norm": 18.499616622924805, "learning_rate": 9.999783296178428e-06, "loss": 0.4902, "step": 256 }, { "epoch": 0.013000151752744195, "grad_norm": 22.959575653076172, "learning_rate": 9.999775759270698e-06, "loss": 0.4916, "step": 257 }, { "epoch": 0.013050736000809348, "grad_norm": 19.2589054107666, "learning_rate": 9.999768093530821e-06, "loss": 0.5994, "step": 258 }, { "epoch": 0.0131013202488745, "grad_norm": 16.01880645751953, "learning_rate": 9.999760298958997e-06, "loss": 0.4513, "step": 259 }, { "epoch": 0.013151904496939653, "grad_norm": 9.932026863098145, "learning_rate": 9.99975237555543e-06, "loss": 0.4202, "step": 260 }, { "epoch": 0.013202488745004806, "grad_norm": 19.732858657836914, "learning_rate": 9.999744323320317e-06, "loss": 0.5639, "step": 261 }, { "epoch": 0.013253072993069957, "grad_norm": 17.810331344604492, "learning_rate": 9.999736142253871e-06, "loss": 0.6159, "step": 262 }, { "epoch": 0.01330365724113511, "grad_norm": 18.924135208129883, "learning_rate": 9.999727832356301e-06, "loss": 0.5314, "step": 263 }, { "epoch": 0.013354241489200263, "grad_norm": 27.7332820892334, "learning_rate": 9.99971939362782e-06, "loss": 0.4753, "step": 264 }, { "epoch": 0.013404825737265416, "grad_norm": 23.05564308166504, "learning_rate": 9.999710826068647e-06, "loss": 0.7342, "step": 265 }, { "epoch": 0.013455409985330569, "grad_norm": 14.822151184082031, "learning_rate": 9.999702129679002e-06, "loss": 0.5281, "step": 266 }, { "epoch": 0.01350599423339572, "grad_norm": 21.867015838623047, "learning_rate": 9.999693304459111e-06, "loss": 0.5938, "step": 267 }, { "epoch": 0.013556578481460873, "grad_norm": 28.065805435180664, "learning_rate": 9.999684350409199e-06, "loss": 0.6756, "step": 268 }, { "epoch": 0.013607162729526025, "grad_norm": 21.686079025268555, "learning_rate": 9.999675267529497e-06, "loss": 0.5495, "step": 269 }, { "epoch": 0.013657746977591178, "grad_norm": 11.835439682006836, "learning_rate": 9.999666055820239e-06, "loss": 0.5796, "step": 270 }, { "epoch": 0.013708331225656331, "grad_norm": 20.891103744506836, "learning_rate": 9.999656715281663e-06, "loss": 0.5071, "step": 271 }, { "epoch": 0.013758915473721484, "grad_norm": 17.45075798034668, "learning_rate": 9.99964724591401e-06, "loss": 0.6347, "step": 272 }, { "epoch": 0.013809499721786635, "grad_norm": 14.682421684265137, "learning_rate": 9.999637647717524e-06, "loss": 0.6315, "step": 273 }, { "epoch": 0.013860083969851788, "grad_norm": 7.979848384857178, "learning_rate": 9.999627920692452e-06, "loss": 0.5144, "step": 274 }, { "epoch": 0.01391066821791694, "grad_norm": 34.56169128417969, "learning_rate": 9.999618064839043e-06, "loss": 0.6801, "step": 275 }, { "epoch": 0.013961252465982094, "grad_norm": 16.20729637145996, "learning_rate": 9.999608080157555e-06, "loss": 0.5897, "step": 276 }, { "epoch": 0.014011836714047246, "grad_norm": 22.143362045288086, "learning_rate": 9.999597966648243e-06, "loss": 0.6984, "step": 277 }, { "epoch": 0.014062420962112398, "grad_norm": 17.71419906616211, "learning_rate": 9.999587724311366e-06, "loss": 0.5051, "step": 278 }, { "epoch": 0.01411300521017755, "grad_norm": 12.204113006591797, "learning_rate": 9.99957735314719e-06, "loss": 0.6064, "step": 279 }, { "epoch": 0.014163589458242703, "grad_norm": 11.478131294250488, "learning_rate": 9.99956685315598e-06, "loss": 0.5676, "step": 280 }, { "epoch": 0.014214173706307856, "grad_norm": 17.0579776763916, "learning_rate": 9.999556224338009e-06, "loss": 0.6751, "step": 281 }, { "epoch": 0.014264757954373009, "grad_norm": 10.81912612915039, "learning_rate": 9.99954546669355e-06, "loss": 0.7483, "step": 282 }, { "epoch": 0.01431534220243816, "grad_norm": 32.51858901977539, "learning_rate": 9.999534580222881e-06, "loss": 0.5632, "step": 283 }, { "epoch": 0.014365926450503313, "grad_norm": 18.37827491760254, "learning_rate": 9.99952356492628e-06, "loss": 0.568, "step": 284 }, { "epoch": 0.014416510698568466, "grad_norm": 12.54236125946045, "learning_rate": 9.999512420804036e-06, "loss": 0.5535, "step": 285 }, { "epoch": 0.014467094946633619, "grad_norm": 11.604639053344727, "learning_rate": 9.999501147856429e-06, "loss": 0.5451, "step": 286 }, { "epoch": 0.014517679194698771, "grad_norm": 17.320377349853516, "learning_rate": 9.999489746083755e-06, "loss": 0.5696, "step": 287 }, { "epoch": 0.014568263442763924, "grad_norm": 16.880413055419922, "learning_rate": 9.999478215486306e-06, "loss": 0.4869, "step": 288 }, { "epoch": 0.014618847690829075, "grad_norm": 14.824078559875488, "learning_rate": 9.99946655606438e-06, "loss": 0.4709, "step": 289 }, { "epoch": 0.014669431938894228, "grad_norm": 29.817581176757812, "learning_rate": 9.999454767818276e-06, "loss": 0.6747, "step": 290 }, { "epoch": 0.014720016186959381, "grad_norm": 18.520177841186523, "learning_rate": 9.999442850748297e-06, "loss": 0.6828, "step": 291 }, { "epoch": 0.014770600435024534, "grad_norm": 14.114188194274902, "learning_rate": 9.999430804854754e-06, "loss": 0.5738, "step": 292 }, { "epoch": 0.014821184683089687, "grad_norm": 38.45853042602539, "learning_rate": 9.99941863013795e-06, "loss": 0.7881, "step": 293 }, { "epoch": 0.014871768931154838, "grad_norm": 23.42243766784668, "learning_rate": 9.99940632659821e-06, "loss": 0.5349, "step": 294 }, { "epoch": 0.01492235317921999, "grad_norm": 25.361377716064453, "learning_rate": 9.99939389423584e-06, "loss": 0.6304, "step": 295 }, { "epoch": 0.014972937427285143, "grad_norm": 14.699026107788086, "learning_rate": 9.999381333051165e-06, "loss": 0.5534, "step": 296 }, { "epoch": 0.015023521675350296, "grad_norm": 13.398818016052246, "learning_rate": 9.99936864304451e-06, "loss": 0.58, "step": 297 }, { "epoch": 0.015074105923415449, "grad_norm": 44.953277587890625, "learning_rate": 9.9993558242162e-06, "loss": 0.6487, "step": 298 }, { "epoch": 0.0151246901714806, "grad_norm": 10.804400444030762, "learning_rate": 9.999342876566566e-06, "loss": 0.5221, "step": 299 }, { "epoch": 0.015175274419545753, "grad_norm": 18.35785675048828, "learning_rate": 9.999329800095943e-06, "loss": 0.6759, "step": 300 }, { "epoch": 0.015225858667610906, "grad_norm": 23.048254013061523, "learning_rate": 9.999316594804665e-06, "loss": 0.5722, "step": 301 }, { "epoch": 0.015276442915676059, "grad_norm": 47.66334533691406, "learning_rate": 9.999303260693073e-06, "loss": 0.5148, "step": 302 }, { "epoch": 0.015327027163741212, "grad_norm": 27.353897094726562, "learning_rate": 9.999289797761511e-06, "loss": 0.6027, "step": 303 }, { "epoch": 0.015377611411806363, "grad_norm": 18.29271697998047, "learning_rate": 9.999276206010327e-06, "loss": 0.5604, "step": 304 }, { "epoch": 0.015428195659871516, "grad_norm": 13.002845764160156, "learning_rate": 9.999262485439871e-06, "loss": 0.4954, "step": 305 }, { "epoch": 0.015478779907936668, "grad_norm": 21.431432723999023, "learning_rate": 9.999248636050496e-06, "loss": 0.6751, "step": 306 }, { "epoch": 0.015529364156001821, "grad_norm": 12.129515647888184, "learning_rate": 9.999234657842559e-06, "loss": 0.5221, "step": 307 }, { "epoch": 0.015579948404066974, "grad_norm": 17.040306091308594, "learning_rate": 9.999220550816419e-06, "loss": 0.5137, "step": 308 }, { "epoch": 0.015630532652132125, "grad_norm": 15.912343978881836, "learning_rate": 9.999206314972441e-06, "loss": 0.5455, "step": 309 }, { "epoch": 0.01568111690019728, "grad_norm": 23.0745792388916, "learning_rate": 9.999191950310991e-06, "loss": 0.5271, "step": 310 }, { "epoch": 0.01573170114826243, "grad_norm": 18.99973487854004, "learning_rate": 9.999177456832439e-06, "loss": 0.5562, "step": 311 }, { "epoch": 0.015782285396327582, "grad_norm": 23.031417846679688, "learning_rate": 9.99916283453716e-06, "loss": 0.5571, "step": 312 }, { "epoch": 0.015832869644392737, "grad_norm": 19.578399658203125, "learning_rate": 9.999148083425529e-06, "loss": 0.5533, "step": 313 }, { "epoch": 0.015883453892457888, "grad_norm": 25.25925064086914, "learning_rate": 9.999133203497928e-06, "loss": 0.7421, "step": 314 }, { "epoch": 0.015934038140523042, "grad_norm": 18.31416130065918, "learning_rate": 9.99911819475474e-06, "loss": 0.5723, "step": 315 }, { "epoch": 0.015984622388588193, "grad_norm": 19.468034744262695, "learning_rate": 9.999103057196349e-06, "loss": 0.6137, "step": 316 }, { "epoch": 0.016035206636653348, "grad_norm": 9.924004554748535, "learning_rate": 9.999087790823148e-06, "loss": 0.6055, "step": 317 }, { "epoch": 0.0160857908847185, "grad_norm": 25.801158905029297, "learning_rate": 9.999072395635528e-06, "loss": 0.5419, "step": 318 }, { "epoch": 0.01613637513278365, "grad_norm": 21.38816261291504, "learning_rate": 9.999056871633888e-06, "loss": 0.4922, "step": 319 }, { "epoch": 0.016186959380848805, "grad_norm": 17.09770393371582, "learning_rate": 9.999041218818628e-06, "loss": 0.5048, "step": 320 }, { "epoch": 0.016237543628913956, "grad_norm": 20.38344955444336, "learning_rate": 9.999025437190149e-06, "loss": 0.6221, "step": 321 }, { "epoch": 0.01628812787697911, "grad_norm": 25.058883666992188, "learning_rate": 9.99900952674886e-06, "loss": 0.4766, "step": 322 }, { "epoch": 0.01633871212504426, "grad_norm": 8.712764739990234, "learning_rate": 9.99899348749517e-06, "loss": 0.5666, "step": 323 }, { "epoch": 0.016389296373109413, "grad_norm": 21.486536026000977, "learning_rate": 9.998977319429493e-06, "loss": 0.631, "step": 324 }, { "epoch": 0.016439880621174567, "grad_norm": 18.457901000976562, "learning_rate": 9.998961022552244e-06, "loss": 0.6736, "step": 325 }, { "epoch": 0.016490464869239718, "grad_norm": 11.119332313537598, "learning_rate": 9.998944596863844e-06, "loss": 0.5473, "step": 326 }, { "epoch": 0.016541049117304873, "grad_norm": 19.883148193359375, "learning_rate": 9.998928042364717e-06, "loss": 0.5924, "step": 327 }, { "epoch": 0.016591633365370024, "grad_norm": 26.815746307373047, "learning_rate": 9.998911359055288e-06, "loss": 0.4823, "step": 328 }, { "epoch": 0.016642217613435175, "grad_norm": 15.875190734863281, "learning_rate": 9.99889454693599e-06, "loss": 0.6268, "step": 329 }, { "epoch": 0.01669280186150033, "grad_norm": 28.572433471679688, "learning_rate": 9.998877606007251e-06, "loss": 0.5739, "step": 330 }, { "epoch": 0.01674338610956548, "grad_norm": 18.750572204589844, "learning_rate": 9.99886053626951e-06, "loss": 0.6373, "step": 331 }, { "epoch": 0.016793970357630635, "grad_norm": 8.417898178100586, "learning_rate": 9.99884333772321e-06, "loss": 0.4933, "step": 332 }, { "epoch": 0.016844554605695786, "grad_norm": 15.385995864868164, "learning_rate": 9.998826010368789e-06, "loss": 0.6953, "step": 333 }, { "epoch": 0.016895138853760937, "grad_norm": 19.328392028808594, "learning_rate": 9.998808554206697e-06, "loss": 0.5275, "step": 334 }, { "epoch": 0.016945723101826092, "grad_norm": 18.831592559814453, "learning_rate": 9.998790969237381e-06, "loss": 0.458, "step": 335 }, { "epoch": 0.016996307349891243, "grad_norm": 8.082682609558105, "learning_rate": 9.998773255461297e-06, "loss": 0.555, "step": 336 }, { "epoch": 0.017046891597956398, "grad_norm": 12.12751293182373, "learning_rate": 9.998755412878902e-06, "loss": 0.5465, "step": 337 }, { "epoch": 0.01709747584602155, "grad_norm": 16.989463806152344, "learning_rate": 9.998737441490653e-06, "loss": 0.6377, "step": 338 }, { "epoch": 0.0171480600940867, "grad_norm": 26.76637840270996, "learning_rate": 9.998719341297013e-06, "loss": 0.6367, "step": 339 }, { "epoch": 0.017198644342151855, "grad_norm": 12.162381172180176, "learning_rate": 9.99870111229845e-06, "loss": 0.5668, "step": 340 }, { "epoch": 0.017249228590217006, "grad_norm": 11.3539457321167, "learning_rate": 9.998682754495433e-06, "loss": 0.5662, "step": 341 }, { "epoch": 0.01729981283828216, "grad_norm": 15.632756233215332, "learning_rate": 9.998664267888437e-06, "loss": 0.6056, "step": 342 }, { "epoch": 0.01735039708634731, "grad_norm": 10.255614280700684, "learning_rate": 9.998645652477934e-06, "loss": 0.4626, "step": 343 }, { "epoch": 0.017400981334412462, "grad_norm": 25.228322982788086, "learning_rate": 9.998626908264409e-06, "loss": 0.5987, "step": 344 }, { "epoch": 0.017451565582477617, "grad_norm": 30.583538055419922, "learning_rate": 9.998608035248339e-06, "loss": 0.5104, "step": 345 }, { "epoch": 0.017502149830542768, "grad_norm": 12.900336265563965, "learning_rate": 9.998589033430215e-06, "loss": 0.5044, "step": 346 }, { "epoch": 0.017552734078607923, "grad_norm": 23.66718101501465, "learning_rate": 9.998569902810527e-06, "loss": 0.4461, "step": 347 }, { "epoch": 0.017603318326673074, "grad_norm": 8.258745193481445, "learning_rate": 9.998550643389763e-06, "loss": 0.5049, "step": 348 }, { "epoch": 0.017653902574738225, "grad_norm": 21.105329513549805, "learning_rate": 9.998531255168424e-06, "loss": 0.5941, "step": 349 }, { "epoch": 0.01770448682280338, "grad_norm": 24.564308166503906, "learning_rate": 9.998511738147009e-06, "loss": 0.6902, "step": 350 }, { "epoch": 0.01775507107086853, "grad_norm": 19.23177146911621, "learning_rate": 9.99849209232602e-06, "loss": 0.6706, "step": 351 }, { "epoch": 0.017805655318933685, "grad_norm": 32.71662902832031, "learning_rate": 9.99847231770596e-06, "loss": 0.5884, "step": 352 }, { "epoch": 0.017856239566998836, "grad_norm": 13.286932945251465, "learning_rate": 9.998452414287345e-06, "loss": 0.5089, "step": 353 }, { "epoch": 0.01790682381506399, "grad_norm": 11.034104347229004, "learning_rate": 9.998432382070684e-06, "loss": 0.4209, "step": 354 }, { "epoch": 0.017957408063129142, "grad_norm": 16.126462936401367, "learning_rate": 9.998412221056492e-06, "loss": 0.6226, "step": 355 }, { "epoch": 0.018007992311194293, "grad_norm": 12.66288948059082, "learning_rate": 9.998391931245291e-06, "loss": 0.4846, "step": 356 }, { "epoch": 0.018058576559259448, "grad_norm": 14.728238105773926, "learning_rate": 9.998371512637604e-06, "loss": 0.4396, "step": 357 }, { "epoch": 0.0181091608073246, "grad_norm": 17.62960433959961, "learning_rate": 9.998350965233956e-06, "loss": 0.5952, "step": 358 }, { "epoch": 0.018159745055389753, "grad_norm": 11.953987121582031, "learning_rate": 9.998330289034876e-06, "loss": 0.5831, "step": 359 }, { "epoch": 0.018210329303454904, "grad_norm": 15.44891357421875, "learning_rate": 9.998309484040898e-06, "loss": 0.6008, "step": 360 }, { "epoch": 0.018260913551520055, "grad_norm": 24.083465576171875, "learning_rate": 9.998288550252558e-06, "loss": 0.6569, "step": 361 }, { "epoch": 0.01831149779958521, "grad_norm": 24.86124610900879, "learning_rate": 9.998267487670392e-06, "loss": 0.7049, "step": 362 }, { "epoch": 0.01836208204765036, "grad_norm": 20.4520263671875, "learning_rate": 9.998246296294948e-06, "loss": 0.6152, "step": 363 }, { "epoch": 0.018412666295715516, "grad_norm": 20.59589195251465, "learning_rate": 9.99822497612677e-06, "loss": 0.6243, "step": 364 }, { "epoch": 0.018463250543780667, "grad_norm": 16.62360954284668, "learning_rate": 9.998203527166406e-06, "loss": 0.6041, "step": 365 }, { "epoch": 0.018513834791845818, "grad_norm": 18.745952606201172, "learning_rate": 9.99818194941441e-06, "loss": 0.5981, "step": 366 }, { "epoch": 0.018564419039910973, "grad_norm": 22.469358444213867, "learning_rate": 9.998160242871338e-06, "loss": 0.6665, "step": 367 }, { "epoch": 0.018615003287976124, "grad_norm": 16.410886764526367, "learning_rate": 9.99813840753775e-06, "loss": 0.5298, "step": 368 }, { "epoch": 0.018665587536041278, "grad_norm": 14.819952011108398, "learning_rate": 9.998116443414208e-06, "loss": 0.4376, "step": 369 }, { "epoch": 0.01871617178410643, "grad_norm": 32.903717041015625, "learning_rate": 9.998094350501275e-06, "loss": 0.5686, "step": 370 }, { "epoch": 0.01876675603217158, "grad_norm": 14.534067153930664, "learning_rate": 9.998072128799525e-06, "loss": 0.5004, "step": 371 }, { "epoch": 0.018817340280236735, "grad_norm": 14.174397468566895, "learning_rate": 9.99804977830953e-06, "loss": 0.5577, "step": 372 }, { "epoch": 0.018867924528301886, "grad_norm": 22.148029327392578, "learning_rate": 9.99802729903186e-06, "loss": 0.6859, "step": 373 }, { "epoch": 0.01891850877636704, "grad_norm": 9.29059886932373, "learning_rate": 9.998004690967102e-06, "loss": 0.5801, "step": 374 }, { "epoch": 0.018969093024432192, "grad_norm": 26.019575119018555, "learning_rate": 9.997981954115835e-06, "loss": 0.7295, "step": 375 }, { "epoch": 0.019019677272497343, "grad_norm": 22.406734466552734, "learning_rate": 9.997959088478645e-06, "loss": 0.6683, "step": 376 }, { "epoch": 0.019070261520562497, "grad_norm": 14.723047256469727, "learning_rate": 9.997936094056122e-06, "loss": 0.6232, "step": 377 }, { "epoch": 0.01912084576862765, "grad_norm": 19.532161712646484, "learning_rate": 9.997912970848858e-06, "loss": 0.6095, "step": 378 }, { "epoch": 0.019171430016692803, "grad_norm": 18.357751846313477, "learning_rate": 9.997889718857448e-06, "loss": 0.631, "step": 379 }, { "epoch": 0.019222014264757954, "grad_norm": 32.95282745361328, "learning_rate": 9.997866338082491e-06, "loss": 0.7684, "step": 380 }, { "epoch": 0.019272598512823105, "grad_norm": 13.637768745422363, "learning_rate": 9.99784282852459e-06, "loss": 0.5153, "step": 381 }, { "epoch": 0.01932318276088826, "grad_norm": 19.892004013061523, "learning_rate": 9.997819190184353e-06, "loss": 0.6246, "step": 382 }, { "epoch": 0.01937376700895341, "grad_norm": 12.545245170593262, "learning_rate": 9.997795423062387e-06, "loss": 0.4939, "step": 383 }, { "epoch": 0.019424351257018566, "grad_norm": 13.777259826660156, "learning_rate": 9.997771527159303e-06, "loss": 0.6092, "step": 384 }, { "epoch": 0.019474935505083717, "grad_norm": 11.841228485107422, "learning_rate": 9.99774750247572e-06, "loss": 0.5852, "step": 385 }, { "epoch": 0.019525519753148868, "grad_norm": 16.577835083007812, "learning_rate": 9.997723349012254e-06, "loss": 0.538, "step": 386 }, { "epoch": 0.019576104001214022, "grad_norm": 9.043819427490234, "learning_rate": 9.99769906676953e-06, "loss": 0.4545, "step": 387 }, { "epoch": 0.019626688249279173, "grad_norm": 19.322219848632812, "learning_rate": 9.997674655748172e-06, "loss": 0.5977, "step": 388 }, { "epoch": 0.019677272497344328, "grad_norm": 34.73544692993164, "learning_rate": 9.99765011594881e-06, "loss": 0.5977, "step": 389 }, { "epoch": 0.01972785674540948, "grad_norm": 21.77810287475586, "learning_rate": 9.997625447372075e-06, "loss": 0.6472, "step": 390 }, { "epoch": 0.019778440993474634, "grad_norm": 22.654064178466797, "learning_rate": 9.997600650018604e-06, "loss": 0.542, "step": 391 }, { "epoch": 0.019829025241539785, "grad_norm": 16.178050994873047, "learning_rate": 9.997575723889036e-06, "loss": 0.5283, "step": 392 }, { "epoch": 0.019879609489604936, "grad_norm": 31.826595306396484, "learning_rate": 9.997550668984012e-06, "loss": 0.6439, "step": 393 }, { "epoch": 0.01993019373767009, "grad_norm": 30.46768569946289, "learning_rate": 9.997525485304178e-06, "loss": 0.5601, "step": 394 }, { "epoch": 0.01998077798573524, "grad_norm": 15.708309173583984, "learning_rate": 9.997500172850184e-06, "loss": 0.5705, "step": 395 }, { "epoch": 0.020031362233800396, "grad_norm": 10.784492492675781, "learning_rate": 9.997474731622681e-06, "loss": 0.4789, "step": 396 }, { "epoch": 0.020081946481865547, "grad_norm": 12.048617362976074, "learning_rate": 9.997449161622325e-06, "loss": 0.5187, "step": 397 }, { "epoch": 0.0201325307299307, "grad_norm": 11.856405258178711, "learning_rate": 9.997423462849776e-06, "loss": 0.4429, "step": 398 }, { "epoch": 0.020183114977995853, "grad_norm": 18.457937240600586, "learning_rate": 9.997397635305695e-06, "loss": 0.5745, "step": 399 }, { "epoch": 0.020233699226061004, "grad_norm": 34.42298889160156, "learning_rate": 9.997371678990746e-06, "loss": 0.509, "step": 400 }, { "epoch": 0.02028428347412616, "grad_norm": 20.744239807128906, "learning_rate": 9.9973455939056e-06, "loss": 0.5081, "step": 401 }, { "epoch": 0.02033486772219131, "grad_norm": 12.944624900817871, "learning_rate": 9.99731938005093e-06, "loss": 0.4991, "step": 402 }, { "epoch": 0.02038545197025646, "grad_norm": 12.514148712158203, "learning_rate": 9.997293037427409e-06, "loss": 0.4355, "step": 403 }, { "epoch": 0.020436036218321615, "grad_norm": 22.374168395996094, "learning_rate": 9.997266566035719e-06, "loss": 0.5927, "step": 404 }, { "epoch": 0.020486620466386767, "grad_norm": 17.56817626953125, "learning_rate": 9.997239965876537e-06, "loss": 0.5337, "step": 405 }, { "epoch": 0.02053720471445192, "grad_norm": 12.445320129394531, "learning_rate": 9.997213236950554e-06, "loss": 0.4846, "step": 406 }, { "epoch": 0.020587788962517072, "grad_norm": 18.908184051513672, "learning_rate": 9.997186379258455e-06, "loss": 0.6511, "step": 407 }, { "epoch": 0.020638373210582223, "grad_norm": 20.390335083007812, "learning_rate": 9.997159392800932e-06, "loss": 0.5218, "step": 408 }, { "epoch": 0.020688957458647378, "grad_norm": 19.873157501220703, "learning_rate": 9.997132277578682e-06, "loss": 0.5587, "step": 409 }, { "epoch": 0.02073954170671253, "grad_norm": 14.750218391418457, "learning_rate": 9.997105033592405e-06, "loss": 0.489, "step": 410 }, { "epoch": 0.020790125954777684, "grad_norm": 14.891282081604004, "learning_rate": 9.9970776608428e-06, "loss": 0.4184, "step": 411 }, { "epoch": 0.020840710202842835, "grad_norm": 10.735186576843262, "learning_rate": 9.997050159330573e-06, "loss": 0.5287, "step": 412 }, { "epoch": 0.020891294450907986, "grad_norm": 31.88671112060547, "learning_rate": 9.997022529056434e-06, "loss": 0.6215, "step": 413 }, { "epoch": 0.02094187869897314, "grad_norm": 7.449845790863037, "learning_rate": 9.996994770021094e-06, "loss": 0.4839, "step": 414 }, { "epoch": 0.02099246294703829, "grad_norm": 10.056912422180176, "learning_rate": 9.99696688222527e-06, "loss": 0.5663, "step": 415 }, { "epoch": 0.021043047195103446, "grad_norm": 27.354934692382812, "learning_rate": 9.996938865669678e-06, "loss": 0.7546, "step": 416 }, { "epoch": 0.021093631443168597, "grad_norm": 13.545953750610352, "learning_rate": 9.996910720355041e-06, "loss": 0.5987, "step": 417 }, { "epoch": 0.021144215691233748, "grad_norm": 20.91629409790039, "learning_rate": 9.996882446282083e-06, "loss": 0.5235, "step": 418 }, { "epoch": 0.021194799939298903, "grad_norm": 13.922904968261719, "learning_rate": 9.996854043451537e-06, "loss": 0.5674, "step": 419 }, { "epoch": 0.021245384187364054, "grad_norm": 13.367424011230469, "learning_rate": 9.996825511864128e-06, "loss": 0.5279, "step": 420 }, { "epoch": 0.02129596843542921, "grad_norm": 28.46970558166504, "learning_rate": 9.996796851520599e-06, "loss": 0.5223, "step": 421 }, { "epoch": 0.02134655268349436, "grad_norm": 22.233675003051758, "learning_rate": 9.99676806242168e-06, "loss": 0.6247, "step": 422 }, { "epoch": 0.02139713693155951, "grad_norm": 16.897573471069336, "learning_rate": 9.99673914456812e-06, "loss": 0.5413, "step": 423 }, { "epoch": 0.021447721179624665, "grad_norm": 18.229829788208008, "learning_rate": 9.99671009796066e-06, "loss": 0.5095, "step": 424 }, { "epoch": 0.021498305427689816, "grad_norm": 15.868843078613281, "learning_rate": 9.996680922600052e-06, "loss": 0.5273, "step": 425 }, { "epoch": 0.02154888967575497, "grad_norm": 16.795560836791992, "learning_rate": 9.996651618487044e-06, "loss": 0.6312, "step": 426 }, { "epoch": 0.021599473923820122, "grad_norm": 20.17876625061035, "learning_rate": 9.996622185622393e-06, "loss": 0.5912, "step": 427 }, { "epoch": 0.021650058171885277, "grad_norm": 17.764883041381836, "learning_rate": 9.996592624006856e-06, "loss": 0.5855, "step": 428 }, { "epoch": 0.021700642419950428, "grad_norm": 14.359132766723633, "learning_rate": 9.996562933641197e-06, "loss": 0.4965, "step": 429 }, { "epoch": 0.02175122666801558, "grad_norm": 12.988452911376953, "learning_rate": 9.99653311452618e-06, "loss": 0.5045, "step": 430 }, { "epoch": 0.021801810916080733, "grad_norm": 27.219852447509766, "learning_rate": 9.996503166662573e-06, "loss": 0.5354, "step": 431 }, { "epoch": 0.021852395164145885, "grad_norm": 14.920003890991211, "learning_rate": 9.996473090051148e-06, "loss": 0.6281, "step": 432 }, { "epoch": 0.02190297941221104, "grad_norm": 22.014047622680664, "learning_rate": 9.99644288469268e-06, "loss": 0.6746, "step": 433 }, { "epoch": 0.02195356366027619, "grad_norm": 8.352063179016113, "learning_rate": 9.996412550587949e-06, "loss": 0.434, "step": 434 }, { "epoch": 0.02200414790834134, "grad_norm": 16.943754196166992, "learning_rate": 9.996382087737733e-06, "loss": 0.5132, "step": 435 }, { "epoch": 0.022054732156406496, "grad_norm": 6.603067398071289, "learning_rate": 9.996351496142817e-06, "loss": 0.3706, "step": 436 }, { "epoch": 0.022105316404471647, "grad_norm": 9.1964693069458, "learning_rate": 9.996320775803994e-06, "loss": 0.4364, "step": 437 }, { "epoch": 0.0221559006525368, "grad_norm": 13.152020454406738, "learning_rate": 9.996289926722052e-06, "loss": 0.4457, "step": 438 }, { "epoch": 0.022206484900601953, "grad_norm": 15.642483711242676, "learning_rate": 9.996258948897786e-06, "loss": 0.6366, "step": 439 }, { "epoch": 0.022257069148667104, "grad_norm": 14.627484321594238, "learning_rate": 9.996227842331996e-06, "loss": 0.527, "step": 440 }, { "epoch": 0.02230765339673226, "grad_norm": 16.215803146362305, "learning_rate": 9.99619660702548e-06, "loss": 0.4663, "step": 441 }, { "epoch": 0.02235823764479741, "grad_norm": 19.17774200439453, "learning_rate": 9.996165242979048e-06, "loss": 0.536, "step": 442 }, { "epoch": 0.022408821892862564, "grad_norm": 29.526744842529297, "learning_rate": 9.996133750193504e-06, "loss": 0.7298, "step": 443 }, { "epoch": 0.022459406140927715, "grad_norm": 21.034481048583984, "learning_rate": 9.996102128669663e-06, "loss": 0.4493, "step": 444 }, { "epoch": 0.022509990388992866, "grad_norm": 12.984509468078613, "learning_rate": 9.996070378408336e-06, "loss": 0.4214, "step": 445 }, { "epoch": 0.02256057463705802, "grad_norm": 16.177465438842773, "learning_rate": 9.996038499410344e-06, "loss": 0.5817, "step": 446 }, { "epoch": 0.022611158885123172, "grad_norm": 22.95356559753418, "learning_rate": 9.996006491676504e-06, "loss": 0.5921, "step": 447 }, { "epoch": 0.022661743133188327, "grad_norm": 19.19211769104004, "learning_rate": 9.995974355207647e-06, "loss": 0.5337, "step": 448 }, { "epoch": 0.022712327381253478, "grad_norm": 10.212876319885254, "learning_rate": 9.995942090004598e-06, "loss": 0.5547, "step": 449 }, { "epoch": 0.02276291162931863, "grad_norm": 23.198890686035156, "learning_rate": 9.995909696068188e-06, "loss": 0.4771, "step": 450 }, { "epoch": 0.022813495877383783, "grad_norm": 21.496679306030273, "learning_rate": 9.995877173399253e-06, "loss": 0.6643, "step": 451 }, { "epoch": 0.022864080125448934, "grad_norm": 19.022438049316406, "learning_rate": 9.995844521998628e-06, "loss": 0.5494, "step": 452 }, { "epoch": 0.02291466437351409, "grad_norm": 19.632640838623047, "learning_rate": 9.995811741867158e-06, "loss": 0.5727, "step": 453 }, { "epoch": 0.02296524862157924, "grad_norm": 30.35894012451172, "learning_rate": 9.995778833005687e-06, "loss": 0.679, "step": 454 }, { "epoch": 0.02301583286964439, "grad_norm": 17.40985870361328, "learning_rate": 9.995745795415062e-06, "loss": 0.4717, "step": 455 }, { "epoch": 0.023066417117709546, "grad_norm": 14.632824897766113, "learning_rate": 9.995712629096135e-06, "loss": 0.3792, "step": 456 }, { "epoch": 0.023117001365774697, "grad_norm": 10.62977123260498, "learning_rate": 9.99567933404976e-06, "loss": 0.4409, "step": 457 }, { "epoch": 0.02316758561383985, "grad_norm": 27.798810958862305, "learning_rate": 9.995645910276794e-06, "loss": 0.628, "step": 458 }, { "epoch": 0.023218169861905003, "grad_norm": 13.799121856689453, "learning_rate": 9.9956123577781e-06, "loss": 0.5498, "step": 459 }, { "epoch": 0.023268754109970154, "grad_norm": 8.058122634887695, "learning_rate": 9.995578676554542e-06, "loss": 0.5458, "step": 460 }, { "epoch": 0.023319338358035308, "grad_norm": 39.08647537231445, "learning_rate": 9.995544866606987e-06, "loss": 0.5539, "step": 461 }, { "epoch": 0.02336992260610046, "grad_norm": 17.676462173461914, "learning_rate": 9.99551092793631e-06, "loss": 0.6682, "step": 462 }, { "epoch": 0.023420506854165614, "grad_norm": 20.122283935546875, "learning_rate": 9.995476860543379e-06, "loss": 0.5759, "step": 463 }, { "epoch": 0.023471091102230765, "grad_norm": 17.70441436767578, "learning_rate": 9.995442664429079e-06, "loss": 0.5742, "step": 464 }, { "epoch": 0.02352167535029592, "grad_norm": 11.49309253692627, "learning_rate": 9.995408339594283e-06, "loss": 0.4088, "step": 465 }, { "epoch": 0.02357225959836107, "grad_norm": 11.051969528198242, "learning_rate": 9.995373886039884e-06, "loss": 0.5516, "step": 466 }, { "epoch": 0.023622843846426222, "grad_norm": 12.02979564666748, "learning_rate": 9.995339303766763e-06, "loss": 0.5961, "step": 467 }, { "epoch": 0.023673428094491376, "grad_norm": 14.179569244384766, "learning_rate": 9.995304592775815e-06, "loss": 0.41, "step": 468 }, { "epoch": 0.023724012342556527, "grad_norm": 17.457332611083984, "learning_rate": 9.995269753067933e-06, "loss": 0.6448, "step": 469 }, { "epoch": 0.023774596590621682, "grad_norm": 20.71563720703125, "learning_rate": 9.995234784644017e-06, "loss": 0.6169, "step": 470 }, { "epoch": 0.023825180838686833, "grad_norm": 17.395965576171875, "learning_rate": 9.995199687504964e-06, "loss": 0.5675, "step": 471 }, { "epoch": 0.023875765086751984, "grad_norm": 15.869020462036133, "learning_rate": 9.995164461651681e-06, "loss": 0.439, "step": 472 }, { "epoch": 0.02392634933481714, "grad_norm": 14.059713363647461, "learning_rate": 9.995129107085076e-06, "loss": 0.5314, "step": 473 }, { "epoch": 0.02397693358288229, "grad_norm": 12.601426124572754, "learning_rate": 9.995093623806057e-06, "loss": 0.4781, "step": 474 }, { "epoch": 0.024027517830947445, "grad_norm": 17.220657348632812, "learning_rate": 9.99505801181554e-06, "loss": 0.464, "step": 475 }, { "epoch": 0.024078102079012596, "grad_norm": 16.243547439575195, "learning_rate": 9.995022271114447e-06, "loss": 0.6293, "step": 476 }, { "epoch": 0.024128686327077747, "grad_norm": 20.31680679321289, "learning_rate": 9.994986401703693e-06, "loss": 0.5654, "step": 477 }, { "epoch": 0.0241792705751429, "grad_norm": 12.16711711883545, "learning_rate": 9.994950403584204e-06, "loss": 0.5205, "step": 478 }, { "epoch": 0.024229854823208052, "grad_norm": 11.018013954162598, "learning_rate": 9.994914276756909e-06, "loss": 0.5488, "step": 479 }, { "epoch": 0.024280439071273207, "grad_norm": 10.623589515686035, "learning_rate": 9.994878021222735e-06, "loss": 0.58, "step": 480 }, { "epoch": 0.024331023319338358, "grad_norm": 24.486631393432617, "learning_rate": 9.994841636982622e-06, "loss": 0.6391, "step": 481 }, { "epoch": 0.02438160756740351, "grad_norm": 27.240694046020508, "learning_rate": 9.994805124037504e-06, "loss": 0.6834, "step": 482 }, { "epoch": 0.024432191815468664, "grad_norm": 16.820154190063477, "learning_rate": 9.994768482388322e-06, "loss": 0.4588, "step": 483 }, { "epoch": 0.024482776063533815, "grad_norm": 15.267021179199219, "learning_rate": 9.99473171203602e-06, "loss": 0.4939, "step": 484 }, { "epoch": 0.02453336031159897, "grad_norm": 13.957478523254395, "learning_rate": 9.994694812981546e-06, "loss": 0.5678, "step": 485 }, { "epoch": 0.02458394455966412, "grad_norm": 13.836231231689453, "learning_rate": 9.994657785225853e-06, "loss": 0.5213, "step": 486 }, { "epoch": 0.02463452880772927, "grad_norm": 15.820413589477539, "learning_rate": 9.994620628769891e-06, "loss": 0.5688, "step": 487 }, { "epoch": 0.024685113055794426, "grad_norm": 18.453712463378906, "learning_rate": 9.994583343614618e-06, "loss": 0.5428, "step": 488 }, { "epoch": 0.024735697303859577, "grad_norm": 19.584468841552734, "learning_rate": 9.994545929760999e-06, "loss": 0.4672, "step": 489 }, { "epoch": 0.024786281551924732, "grad_norm": 25.030935287475586, "learning_rate": 9.994508387209995e-06, "loss": 0.6802, "step": 490 }, { "epoch": 0.024836865799989883, "grad_norm": 24.441417694091797, "learning_rate": 9.994470715962573e-06, "loss": 0.7106, "step": 491 }, { "epoch": 0.024887450048055034, "grad_norm": 15.519182205200195, "learning_rate": 9.994432916019703e-06, "loss": 0.5351, "step": 492 }, { "epoch": 0.02493803429612019, "grad_norm": 32.44498062133789, "learning_rate": 9.994394987382362e-06, "loss": 0.5375, "step": 493 }, { "epoch": 0.02498861854418534, "grad_norm": 17.906469345092773, "learning_rate": 9.994356930051526e-06, "loss": 0.598, "step": 494 }, { "epoch": 0.025039202792250494, "grad_norm": 12.755962371826172, "learning_rate": 9.994318744028174e-06, "loss": 0.339, "step": 495 }, { "epoch": 0.025089787040315645, "grad_norm": 15.831110954284668, "learning_rate": 9.99428042931329e-06, "loss": 0.6415, "step": 496 }, { "epoch": 0.025140371288380797, "grad_norm": 11.517849922180176, "learning_rate": 9.994241985907864e-06, "loss": 0.5405, "step": 497 }, { "epoch": 0.02519095553644595, "grad_norm": 8.979331970214844, "learning_rate": 9.994203413812886e-06, "loss": 0.5269, "step": 498 }, { "epoch": 0.025241539784511102, "grad_norm": 12.223047256469727, "learning_rate": 9.994164713029348e-06, "loss": 0.4325, "step": 499 }, { "epoch": 0.025292124032576257, "grad_norm": 76.44385528564453, "learning_rate": 9.99412588355825e-06, "loss": 0.7679, "step": 500 }, { "epoch": 0.025342708280641408, "grad_norm": 17.999616622924805, "learning_rate": 9.994086925400588e-06, "loss": 0.5244, "step": 501 }, { "epoch": 0.025393292528706563, "grad_norm": 12.982962608337402, "learning_rate": 9.994047838557371e-06, "loss": 0.487, "step": 502 }, { "epoch": 0.025443876776771714, "grad_norm": 10.408404350280762, "learning_rate": 9.994008623029604e-06, "loss": 0.6003, "step": 503 }, { "epoch": 0.025494461024836865, "grad_norm": 18.66061782836914, "learning_rate": 9.993969278818294e-06, "loss": 0.664, "step": 504 }, { "epoch": 0.02554504527290202, "grad_norm": 26.679096221923828, "learning_rate": 9.99392980592446e-06, "loss": 0.6143, "step": 505 }, { "epoch": 0.02559562952096717, "grad_norm": 14.215124130249023, "learning_rate": 9.993890204349118e-06, "loss": 0.5161, "step": 506 }, { "epoch": 0.025646213769032325, "grad_norm": 17.866512298583984, "learning_rate": 9.993850474093288e-06, "loss": 0.5895, "step": 507 }, { "epoch": 0.025696798017097476, "grad_norm": 14.661191940307617, "learning_rate": 9.993810615157991e-06, "loss": 0.6927, "step": 508 }, { "epoch": 0.025747382265162627, "grad_norm": 26.257061004638672, "learning_rate": 9.99377062754426e-06, "loss": 0.5969, "step": 509 }, { "epoch": 0.025797966513227782, "grad_norm": 16.05788230895996, "learning_rate": 9.99373051125312e-06, "loss": 0.481, "step": 510 }, { "epoch": 0.025848550761292933, "grad_norm": 16.222455978393555, "learning_rate": 9.993690266285605e-06, "loss": 0.5777, "step": 511 }, { "epoch": 0.025899135009358087, "grad_norm": 35.14080047607422, "learning_rate": 9.993649892642755e-06, "loss": 0.637, "step": 512 }, { "epoch": 0.02594971925742324, "grad_norm": 19.467554092407227, "learning_rate": 9.99360939032561e-06, "loss": 0.5936, "step": 513 }, { "epoch": 0.02600030350548839, "grad_norm": 10.580787658691406, "learning_rate": 9.993568759335211e-06, "loss": 0.5091, "step": 514 }, { "epoch": 0.026050887753553544, "grad_norm": 13.058818817138672, "learning_rate": 9.993527999672608e-06, "loss": 0.4661, "step": 515 }, { "epoch": 0.026101472001618695, "grad_norm": 16.934995651245117, "learning_rate": 9.993487111338848e-06, "loss": 0.4585, "step": 516 }, { "epoch": 0.02615205624968385, "grad_norm": 26.13129234313965, "learning_rate": 9.993446094334988e-06, "loss": 0.6991, "step": 517 }, { "epoch": 0.026202640497749, "grad_norm": 14.119146347045898, "learning_rate": 9.993404948662084e-06, "loss": 0.434, "step": 518 }, { "epoch": 0.026253224745814152, "grad_norm": 21.154077529907227, "learning_rate": 9.993363674321193e-06, "loss": 0.4801, "step": 519 }, { "epoch": 0.026303808993879307, "grad_norm": 10.76091480255127, "learning_rate": 9.993322271313384e-06, "loss": 0.4374, "step": 520 }, { "epoch": 0.026354393241944458, "grad_norm": 39.2489013671875, "learning_rate": 9.993280739639718e-06, "loss": 0.7472, "step": 521 }, { "epoch": 0.026404977490009612, "grad_norm": 18.33047866821289, "learning_rate": 9.99323907930127e-06, "loss": 0.4018, "step": 522 }, { "epoch": 0.026455561738074763, "grad_norm": 21.33974266052246, "learning_rate": 9.99319729029911e-06, "loss": 0.6454, "step": 523 }, { "epoch": 0.026506145986139915, "grad_norm": 12.480508804321289, "learning_rate": 9.993155372634318e-06, "loss": 0.5891, "step": 524 }, { "epoch": 0.02655673023420507, "grad_norm": 16.726516723632812, "learning_rate": 9.993113326307972e-06, "loss": 0.5869, "step": 525 }, { "epoch": 0.02660731448227022, "grad_norm": 12.088958740234375, "learning_rate": 9.993071151321157e-06, "loss": 0.5323, "step": 526 }, { "epoch": 0.026657898730335375, "grad_norm": 23.118804931640625, "learning_rate": 9.993028847674957e-06, "loss": 0.5839, "step": 527 }, { "epoch": 0.026708482978400526, "grad_norm": 14.318805694580078, "learning_rate": 9.992986415370465e-06, "loss": 0.7057, "step": 528 }, { "epoch": 0.026759067226465677, "grad_norm": 24.8510684967041, "learning_rate": 9.992943854408773e-06, "loss": 0.5787, "step": 529 }, { "epoch": 0.02680965147453083, "grad_norm": 22.258304595947266, "learning_rate": 9.992901164790976e-06, "loss": 0.6393, "step": 530 }, { "epoch": 0.026860235722595983, "grad_norm": 19.549049377441406, "learning_rate": 9.992858346518179e-06, "loss": 0.5275, "step": 531 }, { "epoch": 0.026910819970661137, "grad_norm": 18.49915885925293, "learning_rate": 9.99281539959148e-06, "loss": 0.5144, "step": 532 }, { "epoch": 0.02696140421872629, "grad_norm": 7.883636474609375, "learning_rate": 9.992772324011988e-06, "loss": 0.4233, "step": 533 }, { "epoch": 0.02701198846679144, "grad_norm": 35.28264617919922, "learning_rate": 9.992729119780812e-06, "loss": 0.7573, "step": 534 }, { "epoch": 0.027062572714856594, "grad_norm": 16.979537963867188, "learning_rate": 9.992685786899069e-06, "loss": 0.4, "step": 535 }, { "epoch": 0.027113156962921745, "grad_norm": 14.427353858947754, "learning_rate": 9.992642325367869e-06, "loss": 0.4429, "step": 536 }, { "epoch": 0.0271637412109869, "grad_norm": 18.041688919067383, "learning_rate": 9.992598735188336e-06, "loss": 0.543, "step": 537 }, { "epoch": 0.02721432545905205, "grad_norm": 11.759900093078613, "learning_rate": 9.992555016361595e-06, "loss": 0.545, "step": 538 }, { "epoch": 0.027264909707117205, "grad_norm": 16.02208137512207, "learning_rate": 9.992511168888769e-06, "loss": 0.4108, "step": 539 }, { "epoch": 0.027315493955182357, "grad_norm": 9.61088752746582, "learning_rate": 9.99246719277099e-06, "loss": 0.5508, "step": 540 }, { "epoch": 0.027366078203247508, "grad_norm": 16.499446868896484, "learning_rate": 9.992423088009389e-06, "loss": 0.4099, "step": 541 }, { "epoch": 0.027416662451312662, "grad_norm": 10.64089298248291, "learning_rate": 9.992378854605104e-06, "loss": 0.4666, "step": 542 }, { "epoch": 0.027467246699377813, "grad_norm": 15.439763069152832, "learning_rate": 9.992334492559275e-06, "loss": 0.5193, "step": 543 }, { "epoch": 0.027517830947442968, "grad_norm": 22.2291259765625, "learning_rate": 9.992290001873044e-06, "loss": 0.6016, "step": 544 }, { "epoch": 0.02756841519550812, "grad_norm": 13.374086380004883, "learning_rate": 9.992245382547559e-06, "loss": 0.5437, "step": 545 }, { "epoch": 0.02761899944357327, "grad_norm": 18.3164005279541, "learning_rate": 9.992200634583968e-06, "loss": 0.6833, "step": 546 }, { "epoch": 0.027669583691638425, "grad_norm": 14.294161796569824, "learning_rate": 9.992155757983426e-06, "loss": 0.5025, "step": 547 }, { "epoch": 0.027720167939703576, "grad_norm": 15.075506210327148, "learning_rate": 9.992110752747087e-06, "loss": 0.4746, "step": 548 }, { "epoch": 0.02777075218776873, "grad_norm": 15.20297622680664, "learning_rate": 9.992065618876112e-06, "loss": 0.6088, "step": 549 }, { "epoch": 0.02782133643583388, "grad_norm": 11.594330787658691, "learning_rate": 9.992020356371664e-06, "loss": 0.431, "step": 550 }, { "epoch": 0.027871920683899033, "grad_norm": 17.86979103088379, "learning_rate": 9.991974965234907e-06, "loss": 0.5953, "step": 551 }, { "epoch": 0.027922504931964187, "grad_norm": 19.36792755126953, "learning_rate": 9.991929445467015e-06, "loss": 0.5744, "step": 552 }, { "epoch": 0.027973089180029338, "grad_norm": 12.836052894592285, "learning_rate": 9.991883797069159e-06, "loss": 0.5339, "step": 553 }, { "epoch": 0.028023673428094493, "grad_norm": 22.026653289794922, "learning_rate": 9.991838020042514e-06, "loss": 0.6758, "step": 554 }, { "epoch": 0.028074257676159644, "grad_norm": 12.689266204833984, "learning_rate": 9.991792114388262e-06, "loss": 0.526, "step": 555 }, { "epoch": 0.028124841924224795, "grad_norm": 19.83156394958496, "learning_rate": 9.991746080107582e-06, "loss": 0.6033, "step": 556 }, { "epoch": 0.02817542617228995, "grad_norm": 17.308691024780273, "learning_rate": 9.991699917201665e-06, "loss": 0.5839, "step": 557 }, { "epoch": 0.0282260104203551, "grad_norm": 10.75741195678711, "learning_rate": 9.991653625671696e-06, "loss": 0.5601, "step": 558 }, { "epoch": 0.028276594668420255, "grad_norm": 23.622529983520508, "learning_rate": 9.991607205518872e-06, "loss": 0.7037, "step": 559 }, { "epoch": 0.028327178916485406, "grad_norm": 18.806568145751953, "learning_rate": 9.991560656744385e-06, "loss": 0.4664, "step": 560 }, { "epoch": 0.028377763164550558, "grad_norm": 11.953832626342773, "learning_rate": 9.991513979349437e-06, "loss": 0.6709, "step": 561 }, { "epoch": 0.028428347412615712, "grad_norm": 8.828137397766113, "learning_rate": 9.99146717333523e-06, "loss": 0.4147, "step": 562 }, { "epoch": 0.028478931660680863, "grad_norm": 14.030008316040039, "learning_rate": 9.991420238702972e-06, "loss": 0.5013, "step": 563 }, { "epoch": 0.028529515908746018, "grad_norm": 18.13528060913086, "learning_rate": 9.99137317545387e-06, "loss": 0.4594, "step": 564 }, { "epoch": 0.02858010015681117, "grad_norm": 21.26542854309082, "learning_rate": 9.991325983589138e-06, "loss": 0.6183, "step": 565 }, { "epoch": 0.02863068440487632, "grad_norm": 17.3687801361084, "learning_rate": 9.991278663109989e-06, "loss": 0.66, "step": 566 }, { "epoch": 0.028681268652941475, "grad_norm": 19.1029109954834, "learning_rate": 9.991231214017646e-06, "loss": 0.5528, "step": 567 }, { "epoch": 0.028731852901006626, "grad_norm": 17.70476531982422, "learning_rate": 9.991183636313332e-06, "loss": 0.6206, "step": 568 }, { "epoch": 0.02878243714907178, "grad_norm": 17.248069763183594, "learning_rate": 9.99113592999827e-06, "loss": 0.6166, "step": 569 }, { "epoch": 0.02883302139713693, "grad_norm": 9.378047943115234, "learning_rate": 9.99108809507369e-06, "loss": 0.5284, "step": 570 }, { "epoch": 0.028883605645202082, "grad_norm": 11.9326171875, "learning_rate": 9.991040131540828e-06, "loss": 0.5282, "step": 571 }, { "epoch": 0.028934189893267237, "grad_norm": 12.815873146057129, "learning_rate": 9.990992039400914e-06, "loss": 0.5554, "step": 572 }, { "epoch": 0.028984774141332388, "grad_norm": 14.775552749633789, "learning_rate": 9.990943818655193e-06, "loss": 0.4202, "step": 573 }, { "epoch": 0.029035358389397543, "grad_norm": 12.653632164001465, "learning_rate": 9.990895469304904e-06, "loss": 0.4797, "step": 574 }, { "epoch": 0.029085942637462694, "grad_norm": 30.22648811340332, "learning_rate": 9.990846991351293e-06, "loss": 0.6164, "step": 575 }, { "epoch": 0.02913652688552785, "grad_norm": 14.695141792297363, "learning_rate": 9.99079838479561e-06, "loss": 0.5773, "step": 576 }, { "epoch": 0.029187111133593, "grad_norm": 13.011380195617676, "learning_rate": 9.990749649639108e-06, "loss": 0.4682, "step": 577 }, { "epoch": 0.02923769538165815, "grad_norm": 7.9724812507629395, "learning_rate": 9.990700785883043e-06, "loss": 0.3295, "step": 578 }, { "epoch": 0.029288279629723305, "grad_norm": 23.565244674682617, "learning_rate": 9.990651793528671e-06, "loss": 0.6469, "step": 579 }, { "epoch": 0.029338863877788456, "grad_norm": 19.976789474487305, "learning_rate": 9.990602672577258e-06, "loss": 0.6086, "step": 580 }, { "epoch": 0.02938944812585361, "grad_norm": 11.882352828979492, "learning_rate": 9.990553423030069e-06, "loss": 0.4459, "step": 581 }, { "epoch": 0.029440032373918762, "grad_norm": 8.952604293823242, "learning_rate": 9.990504044888372e-06, "loss": 0.5066, "step": 582 }, { "epoch": 0.029490616621983913, "grad_norm": 11.395858764648438, "learning_rate": 9.99045453815344e-06, "loss": 0.4963, "step": 583 }, { "epoch": 0.029541200870049068, "grad_norm": 9.674945831298828, "learning_rate": 9.990404902826548e-06, "loss": 0.4572, "step": 584 }, { "epoch": 0.02959178511811422, "grad_norm": 11.939826011657715, "learning_rate": 9.990355138908975e-06, "loss": 0.4584, "step": 585 }, { "epoch": 0.029642369366179373, "grad_norm": 24.477920532226562, "learning_rate": 9.990305246402005e-06, "loss": 0.5845, "step": 586 }, { "epoch": 0.029692953614244524, "grad_norm": 13.20823860168457, "learning_rate": 9.990255225306922e-06, "loss": 0.5827, "step": 587 }, { "epoch": 0.029743537862309676, "grad_norm": 20.04751205444336, "learning_rate": 9.990205075625014e-06, "loss": 0.5271, "step": 588 }, { "epoch": 0.02979412211037483, "grad_norm": 26.921968460083008, "learning_rate": 9.990154797357576e-06, "loss": 0.6601, "step": 589 }, { "epoch": 0.02984470635843998, "grad_norm": 20.528675079345703, "learning_rate": 9.990104390505903e-06, "loss": 0.4944, "step": 590 }, { "epoch": 0.029895290606505136, "grad_norm": 21.37859535217285, "learning_rate": 9.99005385507129e-06, "loss": 0.5313, "step": 591 }, { "epoch": 0.029945874854570287, "grad_norm": 8.360918998718262, "learning_rate": 9.990003191055045e-06, "loss": 0.5509, "step": 592 }, { "epoch": 0.029996459102635438, "grad_norm": 11.17050838470459, "learning_rate": 9.98995239845847e-06, "loss": 0.5665, "step": 593 }, { "epoch": 0.030047043350700593, "grad_norm": 15.601547241210938, "learning_rate": 9.989901477282874e-06, "loss": 0.4704, "step": 594 }, { "epoch": 0.030097627598765744, "grad_norm": 10.721697807312012, "learning_rate": 9.989850427529569e-06, "loss": 0.5054, "step": 595 }, { "epoch": 0.030148211846830898, "grad_norm": 15.333925247192383, "learning_rate": 9.989799249199872e-06, "loss": 0.4793, "step": 596 }, { "epoch": 0.03019879609489605, "grad_norm": 22.24761962890625, "learning_rate": 9.989747942295101e-06, "loss": 0.5267, "step": 597 }, { "epoch": 0.0302493803429612, "grad_norm": 17.65778350830078, "learning_rate": 9.989696506816576e-06, "loss": 0.6203, "step": 598 }, { "epoch": 0.030299964591026355, "grad_norm": 15.272322654724121, "learning_rate": 9.989644942765624e-06, "loss": 0.5926, "step": 599 }, { "epoch": 0.030350548839091506, "grad_norm": 11.953412055969238, "learning_rate": 9.989593250143577e-06, "loss": 0.5927, "step": 600 }, { "epoch": 0.03040113308715666, "grad_norm": 10.285418510437012, "learning_rate": 9.98954142895176e-06, "loss": 0.5691, "step": 601 }, { "epoch": 0.030451717335221812, "grad_norm": 18.698139190673828, "learning_rate": 9.989489479191514e-06, "loss": 0.5467, "step": 602 }, { "epoch": 0.030502301583286963, "grad_norm": 12.583427429199219, "learning_rate": 9.989437400864177e-06, "loss": 0.4875, "step": 603 }, { "epoch": 0.030552885831352117, "grad_norm": 11.780808448791504, "learning_rate": 9.989385193971089e-06, "loss": 0.4572, "step": 604 }, { "epoch": 0.03060347007941727, "grad_norm": 9.291247367858887, "learning_rate": 9.989332858513595e-06, "loss": 0.5296, "step": 605 }, { "epoch": 0.030654054327482423, "grad_norm": 10.280113220214844, "learning_rate": 9.989280394493044e-06, "loss": 0.5538, "step": 606 }, { "epoch": 0.030704638575547574, "grad_norm": 20.57911491394043, "learning_rate": 9.989227801910789e-06, "loss": 0.5759, "step": 607 }, { "epoch": 0.030755222823612725, "grad_norm": 24.99956512451172, "learning_rate": 9.989175080768184e-06, "loss": 0.6079, "step": 608 }, { "epoch": 0.03080580707167788, "grad_norm": 15.80003547668457, "learning_rate": 9.989122231066587e-06, "loss": 0.4687, "step": 609 }, { "epoch": 0.03085639131974303, "grad_norm": 24.860578536987305, "learning_rate": 9.989069252807363e-06, "loss": 0.5882, "step": 610 }, { "epoch": 0.030906975567808186, "grad_norm": 11.37317943572998, "learning_rate": 9.989016145991876e-06, "loss": 0.4496, "step": 611 }, { "epoch": 0.030957559815873337, "grad_norm": 15.668007850646973, "learning_rate": 9.988962910621491e-06, "loss": 0.5357, "step": 612 }, { "epoch": 0.031008144063938488, "grad_norm": 17.801803588867188, "learning_rate": 9.988909546697583e-06, "loss": 0.5547, "step": 613 }, { "epoch": 0.031058728312003642, "grad_norm": 17.42266845703125, "learning_rate": 9.988856054221528e-06, "loss": 0.5658, "step": 614 }, { "epoch": 0.031109312560068794, "grad_norm": 21.201950073242188, "learning_rate": 9.9888024331947e-06, "loss": 0.6227, "step": 615 }, { "epoch": 0.031159896808133948, "grad_norm": 13.362656593322754, "learning_rate": 9.988748683618483e-06, "loss": 0.5163, "step": 616 }, { "epoch": 0.0312104810561991, "grad_norm": 13.131528854370117, "learning_rate": 9.988694805494263e-06, "loss": 0.5538, "step": 617 }, { "epoch": 0.03126106530426425, "grad_norm": 14.06868839263916, "learning_rate": 9.988640798823427e-06, "loss": 0.5943, "step": 618 }, { "epoch": 0.0313116495523294, "grad_norm": 16.12552833557129, "learning_rate": 9.98858666360737e-06, "loss": 0.6122, "step": 619 }, { "epoch": 0.03136223380039456, "grad_norm": 9.694734573364258, "learning_rate": 9.988532399847482e-06, "loss": 0.5613, "step": 620 }, { "epoch": 0.03141281804845971, "grad_norm": 12.366657257080078, "learning_rate": 9.988478007545162e-06, "loss": 0.5456, "step": 621 }, { "epoch": 0.03146340229652486, "grad_norm": 7.858946800231934, "learning_rate": 9.988423486701815e-06, "loss": 0.548, "step": 622 }, { "epoch": 0.03151398654459001, "grad_norm": 14.554610252380371, "learning_rate": 9.988368837318843e-06, "loss": 0.5279, "step": 623 }, { "epoch": 0.031564570792655164, "grad_norm": 14.450069427490234, "learning_rate": 9.988314059397654e-06, "loss": 0.5447, "step": 624 }, { "epoch": 0.03161515504072032, "grad_norm": 12.717538833618164, "learning_rate": 9.988259152939663e-06, "loss": 0.5284, "step": 625 }, { "epoch": 0.03166573928878547, "grad_norm": 9.273475646972656, "learning_rate": 9.988204117946281e-06, "loss": 0.2889, "step": 626 }, { "epoch": 0.031716323536850624, "grad_norm": 12.814962387084961, "learning_rate": 9.988148954418927e-06, "loss": 0.6502, "step": 627 }, { "epoch": 0.031766907784915775, "grad_norm": 22.35105323791504, "learning_rate": 9.988093662359024e-06, "loss": 0.6338, "step": 628 }, { "epoch": 0.031817492032980926, "grad_norm": 15.308331489562988, "learning_rate": 9.988038241767995e-06, "loss": 0.5993, "step": 629 }, { "epoch": 0.031868076281046084, "grad_norm": 37.02248764038086, "learning_rate": 9.987982692647266e-06, "loss": 0.5052, "step": 630 }, { "epoch": 0.031918660529111235, "grad_norm": 26.863739013671875, "learning_rate": 9.987927014998274e-06, "loss": 0.7437, "step": 631 }, { "epoch": 0.03196924477717639, "grad_norm": 7.9905524253845215, "learning_rate": 9.98787120882245e-06, "loss": 0.4143, "step": 632 }, { "epoch": 0.03201982902524154, "grad_norm": 15.933112144470215, "learning_rate": 9.987815274121233e-06, "loss": 0.6174, "step": 633 }, { "epoch": 0.032070413273306696, "grad_norm": 20.471435546875, "learning_rate": 9.987759210896063e-06, "loss": 0.5563, "step": 634 }, { "epoch": 0.03212099752137185, "grad_norm": 16.153392791748047, "learning_rate": 9.987703019148387e-06, "loss": 0.4434, "step": 635 }, { "epoch": 0.032171581769437, "grad_norm": 15.296040534973145, "learning_rate": 9.98764669887965e-06, "loss": 0.5159, "step": 636 }, { "epoch": 0.03222216601750215, "grad_norm": 18.817232131958008, "learning_rate": 9.987590250091306e-06, "loss": 0.5515, "step": 637 }, { "epoch": 0.0322727502655673, "grad_norm": 19.7265625, "learning_rate": 9.987533672784806e-06, "loss": 0.6451, "step": 638 }, { "epoch": 0.03232333451363246, "grad_norm": 23.5718936920166, "learning_rate": 9.987476966961613e-06, "loss": 0.626, "step": 639 }, { "epoch": 0.03237391876169761, "grad_norm": 12.790105819702148, "learning_rate": 9.987420132623184e-06, "loss": 0.5212, "step": 640 }, { "epoch": 0.03242450300976276, "grad_norm": 11.860028266906738, "learning_rate": 9.987363169770985e-06, "loss": 0.5119, "step": 641 }, { "epoch": 0.03247508725782791, "grad_norm": 41.81549072265625, "learning_rate": 9.987306078406482e-06, "loss": 0.7371, "step": 642 }, { "epoch": 0.03252567150589306, "grad_norm": 13.943751335144043, "learning_rate": 9.987248858531151e-06, "loss": 0.5111, "step": 643 }, { "epoch": 0.03257625575395822, "grad_norm": 24.652267456054688, "learning_rate": 9.987191510146462e-06, "loss": 0.5197, "step": 644 }, { "epoch": 0.03262684000202337, "grad_norm": 14.081793785095215, "learning_rate": 9.987134033253891e-06, "loss": 0.5049, "step": 645 }, { "epoch": 0.03267742425008852, "grad_norm": 21.024354934692383, "learning_rate": 9.987076427854925e-06, "loss": 0.5378, "step": 646 }, { "epoch": 0.032728008498153674, "grad_norm": 15.869294166564941, "learning_rate": 9.987018693951045e-06, "loss": 0.5948, "step": 647 }, { "epoch": 0.032778592746218825, "grad_norm": 12.61564826965332, "learning_rate": 9.986960831543737e-06, "loss": 0.6223, "step": 648 }, { "epoch": 0.03282917699428398, "grad_norm": 12.248970031738281, "learning_rate": 9.986902840634496e-06, "loss": 0.5909, "step": 649 }, { "epoch": 0.032879761242349134, "grad_norm": 15.602255821228027, "learning_rate": 9.986844721224813e-06, "loss": 0.4169, "step": 650 }, { "epoch": 0.032930345490414285, "grad_norm": 13.123071670532227, "learning_rate": 9.986786473316187e-06, "loss": 0.4091, "step": 651 }, { "epoch": 0.032980929738479436, "grad_norm": 14.498602867126465, "learning_rate": 9.986728096910119e-06, "loss": 0.5328, "step": 652 }, { "epoch": 0.03303151398654459, "grad_norm": 20.485336303710938, "learning_rate": 9.98666959200811e-06, "loss": 0.5327, "step": 653 }, { "epoch": 0.033082098234609746, "grad_norm": 14.356918334960938, "learning_rate": 9.986610958611675e-06, "loss": 0.4938, "step": 654 }, { "epoch": 0.0331326824826749, "grad_norm": 15.311623573303223, "learning_rate": 9.986552196722318e-06, "loss": 0.5165, "step": 655 }, { "epoch": 0.03318326673074005, "grad_norm": 12.682887077331543, "learning_rate": 9.986493306341555e-06, "loss": 0.5384, "step": 656 }, { "epoch": 0.0332338509788052, "grad_norm": 10.113044738769531, "learning_rate": 9.986434287470905e-06, "loss": 0.3728, "step": 657 }, { "epoch": 0.03328443522687035, "grad_norm": 20.273855209350586, "learning_rate": 9.986375140111889e-06, "loss": 0.5639, "step": 658 }, { "epoch": 0.03333501947493551, "grad_norm": 9.651646614074707, "learning_rate": 9.986315864266028e-06, "loss": 0.5573, "step": 659 }, { "epoch": 0.03338560372300066, "grad_norm": 14.181355476379395, "learning_rate": 9.98625645993485e-06, "loss": 0.4962, "step": 660 }, { "epoch": 0.03343618797106581, "grad_norm": 33.57608413696289, "learning_rate": 9.986196927119887e-06, "loss": 0.5453, "step": 661 }, { "epoch": 0.03348677221913096, "grad_norm": 32.077667236328125, "learning_rate": 9.986137265822675e-06, "loss": 0.6297, "step": 662 }, { "epoch": 0.03353735646719611, "grad_norm": 17.2042179107666, "learning_rate": 9.986077476044747e-06, "loss": 0.6551, "step": 663 }, { "epoch": 0.03358794071526127, "grad_norm": 13.004806518554688, "learning_rate": 9.986017557787647e-06, "loss": 0.5249, "step": 664 }, { "epoch": 0.03363852496332642, "grad_norm": 13.50832462310791, "learning_rate": 9.985957511052917e-06, "loss": 0.5519, "step": 665 }, { "epoch": 0.03368910921139157, "grad_norm": 13.024831771850586, "learning_rate": 9.985897335842106e-06, "loss": 0.552, "step": 666 }, { "epoch": 0.033739693459456724, "grad_norm": 9.697739601135254, "learning_rate": 9.985837032156764e-06, "loss": 0.5056, "step": 667 }, { "epoch": 0.033790277707521875, "grad_norm": 16.46345329284668, "learning_rate": 9.985776599998444e-06, "loss": 0.4714, "step": 668 }, { "epoch": 0.03384086195558703, "grad_norm": 24.08216667175293, "learning_rate": 9.985716039368703e-06, "loss": 0.4717, "step": 669 }, { "epoch": 0.033891446203652184, "grad_norm": 11.624739646911621, "learning_rate": 9.985655350269102e-06, "loss": 0.5428, "step": 670 }, { "epoch": 0.033942030451717335, "grad_norm": 19.130840301513672, "learning_rate": 9.985594532701206e-06, "loss": 0.4933, "step": 671 }, { "epoch": 0.033992614699782486, "grad_norm": 13.245088577270508, "learning_rate": 9.985533586666581e-06, "loss": 0.4555, "step": 672 }, { "epoch": 0.03404319894784764, "grad_norm": 18.55948257446289, "learning_rate": 9.985472512166796e-06, "loss": 0.5422, "step": 673 }, { "epoch": 0.034093783195912795, "grad_norm": 22.121658325195312, "learning_rate": 9.985411309203427e-06, "loss": 0.6397, "step": 674 }, { "epoch": 0.03414436744397795, "grad_norm": 20.848569869995117, "learning_rate": 9.985349977778051e-06, "loss": 0.5336, "step": 675 }, { "epoch": 0.0341949516920431, "grad_norm": 15.1366548538208, "learning_rate": 9.985288517892247e-06, "loss": 0.5489, "step": 676 }, { "epoch": 0.03424553594010825, "grad_norm": 25.5090274810791, "learning_rate": 9.9852269295476e-06, "loss": 0.6395, "step": 677 }, { "epoch": 0.0342961201881734, "grad_norm": 23.620702743530273, "learning_rate": 9.985165212745697e-06, "loss": 0.6031, "step": 678 }, { "epoch": 0.03434670443623856, "grad_norm": 16.138309478759766, "learning_rate": 9.985103367488128e-06, "loss": 0.4668, "step": 679 }, { "epoch": 0.03439728868430371, "grad_norm": 12.130837440490723, "learning_rate": 9.985041393776483e-06, "loss": 0.5312, "step": 680 }, { "epoch": 0.03444787293236886, "grad_norm": 25.603464126586914, "learning_rate": 9.984979291612364e-06, "loss": 0.7049, "step": 681 }, { "epoch": 0.03449845718043401, "grad_norm": 9.197319030761719, "learning_rate": 9.98491706099737e-06, "loss": 0.5596, "step": 682 }, { "epoch": 0.03454904142849916, "grad_norm": 13.361732482910156, "learning_rate": 9.984854701933104e-06, "loss": 0.4712, "step": 683 }, { "epoch": 0.03459962567656432, "grad_norm": 13.008930206298828, "learning_rate": 9.984792214421172e-06, "loss": 0.5105, "step": 684 }, { "epoch": 0.03465020992462947, "grad_norm": 8.513949394226074, "learning_rate": 9.984729598463186e-06, "loss": 0.5073, "step": 685 }, { "epoch": 0.03470079417269462, "grad_norm": 24.15869903564453, "learning_rate": 9.984666854060756e-06, "loss": 0.4961, "step": 686 }, { "epoch": 0.034751378420759774, "grad_norm": 23.505374908447266, "learning_rate": 9.984603981215503e-06, "loss": 0.641, "step": 687 }, { "epoch": 0.034801962668824925, "grad_norm": 13.907588958740234, "learning_rate": 9.984540979929046e-06, "loss": 0.5706, "step": 688 }, { "epoch": 0.03485254691689008, "grad_norm": 13.788132667541504, "learning_rate": 9.984477850203005e-06, "loss": 0.5379, "step": 689 }, { "epoch": 0.034903131164955234, "grad_norm": 11.186384201049805, "learning_rate": 9.984414592039012e-06, "loss": 0.5407, "step": 690 }, { "epoch": 0.034953715413020385, "grad_norm": 8.504643440246582, "learning_rate": 9.984351205438691e-06, "loss": 0.4126, "step": 691 }, { "epoch": 0.035004299661085536, "grad_norm": 37.72002029418945, "learning_rate": 9.984287690403682e-06, "loss": 0.443, "step": 692 }, { "epoch": 0.03505488390915069, "grad_norm": 10.672569274902344, "learning_rate": 9.984224046935616e-06, "loss": 0.4833, "step": 693 }, { "epoch": 0.035105468157215845, "grad_norm": 12.600421905517578, "learning_rate": 9.984160275036137e-06, "loss": 0.4291, "step": 694 }, { "epoch": 0.035156052405280996, "grad_norm": 21.77521514892578, "learning_rate": 9.984096374706885e-06, "loss": 0.5782, "step": 695 }, { "epoch": 0.03520663665334615, "grad_norm": 12.410987854003906, "learning_rate": 9.98403234594951e-06, "loss": 0.5368, "step": 696 }, { "epoch": 0.0352572209014113, "grad_norm": 15.909688949584961, "learning_rate": 9.983968188765658e-06, "loss": 0.5845, "step": 697 }, { "epoch": 0.03530780514947645, "grad_norm": 13.341547012329102, "learning_rate": 9.983903903156984e-06, "loss": 0.5513, "step": 698 }, { "epoch": 0.03535838939754161, "grad_norm": 15.003726959228516, "learning_rate": 9.983839489125146e-06, "loss": 0.6005, "step": 699 }, { "epoch": 0.03540897364560676, "grad_norm": 18.884248733520508, "learning_rate": 9.983774946671801e-06, "loss": 0.6363, "step": 700 }, { "epoch": 0.03545955789367191, "grad_norm": 19.176807403564453, "learning_rate": 9.983710275798615e-06, "loss": 0.5684, "step": 701 }, { "epoch": 0.03551014214173706, "grad_norm": 9.755188941955566, "learning_rate": 9.983645476507251e-06, "loss": 0.5073, "step": 702 }, { "epoch": 0.03556072638980221, "grad_norm": 14.216994285583496, "learning_rate": 9.983580548799381e-06, "loss": 0.3821, "step": 703 }, { "epoch": 0.03561131063786737, "grad_norm": 21.756887435913086, "learning_rate": 9.98351549267668e-06, "loss": 0.5838, "step": 704 }, { "epoch": 0.03566189488593252, "grad_norm": 13.026229858398438, "learning_rate": 9.983450308140818e-06, "loss": 0.5567, "step": 705 }, { "epoch": 0.03571247913399767, "grad_norm": 10.365445137023926, "learning_rate": 9.983384995193482e-06, "loss": 0.4816, "step": 706 }, { "epoch": 0.035763063382062824, "grad_norm": 15.541449546813965, "learning_rate": 9.983319553836348e-06, "loss": 0.5495, "step": 707 }, { "epoch": 0.03581364763012798, "grad_norm": 32.63273239135742, "learning_rate": 9.983253984071109e-06, "loss": 0.9119, "step": 708 }, { "epoch": 0.03586423187819313, "grad_norm": 13.682848930358887, "learning_rate": 9.98318828589945e-06, "loss": 0.4991, "step": 709 }, { "epoch": 0.035914816126258284, "grad_norm": 14.257963180541992, "learning_rate": 9.983122459323065e-06, "loss": 0.5393, "step": 710 }, { "epoch": 0.035965400374323435, "grad_norm": 10.977893829345703, "learning_rate": 9.983056504343651e-06, "loss": 0.4968, "step": 711 }, { "epoch": 0.036015984622388586, "grad_norm": 16.276737213134766, "learning_rate": 9.982990420962907e-06, "loss": 0.5824, "step": 712 }, { "epoch": 0.036066568870453744, "grad_norm": 9.747637748718262, "learning_rate": 9.982924209182536e-06, "loss": 0.3123, "step": 713 }, { "epoch": 0.036117153118518895, "grad_norm": 21.14165496826172, "learning_rate": 9.982857869004243e-06, "loss": 0.5428, "step": 714 }, { "epoch": 0.036167737366584046, "grad_norm": 20.230636596679688, "learning_rate": 9.98279140042974e-06, "loss": 0.5891, "step": 715 }, { "epoch": 0.0362183216146492, "grad_norm": 9.92956256866455, "learning_rate": 9.982724803460736e-06, "loss": 0.5316, "step": 716 }, { "epoch": 0.03626890586271435, "grad_norm": 15.905277252197266, "learning_rate": 9.98265807809895e-06, "loss": 0.5538, "step": 717 }, { "epoch": 0.036319490110779507, "grad_norm": 23.79247283935547, "learning_rate": 9.9825912243461e-06, "loss": 0.5498, "step": 718 }, { "epoch": 0.03637007435884466, "grad_norm": 12.539578437805176, "learning_rate": 9.98252424220391e-06, "loss": 0.4477, "step": 719 }, { "epoch": 0.03642065860690981, "grad_norm": 19.692306518554688, "learning_rate": 9.982457131674106e-06, "loss": 0.5341, "step": 720 }, { "epoch": 0.03647124285497496, "grad_norm": 22.909521102905273, "learning_rate": 9.982389892758414e-06, "loss": 0.4632, "step": 721 }, { "epoch": 0.03652182710304011, "grad_norm": 8.190949440002441, "learning_rate": 9.98232252545857e-06, "loss": 0.5154, "step": 722 }, { "epoch": 0.03657241135110527, "grad_norm": 16.792034149169922, "learning_rate": 9.98225502977631e-06, "loss": 0.444, "step": 723 }, { "epoch": 0.03662299559917042, "grad_norm": 17.070024490356445, "learning_rate": 9.982187405713371e-06, "loss": 0.5438, "step": 724 }, { "epoch": 0.03667357984723557, "grad_norm": 12.307943344116211, "learning_rate": 9.982119653271497e-06, "loss": 0.4931, "step": 725 }, { "epoch": 0.03672416409530072, "grad_norm": 14.323333740234375, "learning_rate": 9.982051772452431e-06, "loss": 0.5068, "step": 726 }, { "epoch": 0.03677474834336587, "grad_norm": 12.866251945495605, "learning_rate": 9.981983763257927e-06, "loss": 0.6118, "step": 727 }, { "epoch": 0.03682533259143103, "grad_norm": 15.488363265991211, "learning_rate": 9.981915625689732e-06, "loss": 0.5206, "step": 728 }, { "epoch": 0.03687591683949618, "grad_norm": 18.031713485717773, "learning_rate": 9.981847359749607e-06, "loss": 0.6972, "step": 729 }, { "epoch": 0.036926501087561334, "grad_norm": 23.05552864074707, "learning_rate": 9.98177896543931e-06, "loss": 0.6019, "step": 730 }, { "epoch": 0.036977085335626485, "grad_norm": 11.299986839294434, "learning_rate": 9.9817104427606e-06, "loss": 0.5541, "step": 731 }, { "epoch": 0.037027669583691636, "grad_norm": 16.101839065551758, "learning_rate": 9.981641791715245e-06, "loss": 0.6437, "step": 732 }, { "epoch": 0.037078253831756794, "grad_norm": 16.28394889831543, "learning_rate": 9.981573012305014e-06, "loss": 0.6092, "step": 733 }, { "epoch": 0.037128838079821945, "grad_norm": 12.366171836853027, "learning_rate": 9.981504104531679e-06, "loss": 0.4854, "step": 734 }, { "epoch": 0.037179422327887096, "grad_norm": 9.35709285736084, "learning_rate": 9.981435068397016e-06, "loss": 0.4817, "step": 735 }, { "epoch": 0.03723000657595225, "grad_norm": 17.124187469482422, "learning_rate": 9.981365903902802e-06, "loss": 0.6307, "step": 736 }, { "epoch": 0.0372805908240174, "grad_norm": 14.019539833068848, "learning_rate": 9.981296611050821e-06, "loss": 0.4737, "step": 737 }, { "epoch": 0.037331175072082556, "grad_norm": 14.287375450134277, "learning_rate": 9.98122718984286e-06, "loss": 0.499, "step": 738 }, { "epoch": 0.03738175932014771, "grad_norm": 20.706439971923828, "learning_rate": 9.981157640280705e-06, "loss": 0.63, "step": 739 }, { "epoch": 0.03743234356821286, "grad_norm": 17.022808074951172, "learning_rate": 9.981087962366149e-06, "loss": 0.5182, "step": 740 }, { "epoch": 0.03748292781627801, "grad_norm": 15.117709159851074, "learning_rate": 9.981018156100987e-06, "loss": 0.556, "step": 741 }, { "epoch": 0.03753351206434316, "grad_norm": 11.30010986328125, "learning_rate": 9.980948221487018e-06, "loss": 0.5498, "step": 742 }, { "epoch": 0.03758409631240832, "grad_norm": 12.529576301574707, "learning_rate": 9.980878158526046e-06, "loss": 0.5807, "step": 743 }, { "epoch": 0.03763468056047347, "grad_norm": 12.85781478881836, "learning_rate": 9.980807967219872e-06, "loss": 0.458, "step": 744 }, { "epoch": 0.03768526480853862, "grad_norm": 9.939268112182617, "learning_rate": 9.980737647570309e-06, "loss": 0.4466, "step": 745 }, { "epoch": 0.03773584905660377, "grad_norm": 15.897769927978516, "learning_rate": 9.980667199579167e-06, "loss": 0.6187, "step": 746 }, { "epoch": 0.03778643330466892, "grad_norm": 15.4701509475708, "learning_rate": 9.980596623248262e-06, "loss": 0.5765, "step": 747 }, { "epoch": 0.03783701755273408, "grad_norm": 11.95398235321045, "learning_rate": 9.980525918579412e-06, "loss": 0.614, "step": 748 }, { "epoch": 0.03788760180079923, "grad_norm": 6.5373358726501465, "learning_rate": 9.980455085574438e-06, "loss": 0.4078, "step": 749 }, { "epoch": 0.037938186048864384, "grad_norm": 14.46668815612793, "learning_rate": 9.98038412423517e-06, "loss": 0.4968, "step": 750 }, { "epoch": 0.037988770296929535, "grad_norm": 18.294658660888672, "learning_rate": 9.980313034563429e-06, "loss": 0.5467, "step": 751 }, { "epoch": 0.038039354544994686, "grad_norm": 16.62605094909668, "learning_rate": 9.98024181656105e-06, "loss": 0.5476, "step": 752 }, { "epoch": 0.038089938793059844, "grad_norm": 14.737811088562012, "learning_rate": 9.98017047022987e-06, "loss": 0.6149, "step": 753 }, { "epoch": 0.038140523041124995, "grad_norm": 13.521196365356445, "learning_rate": 9.980098995571727e-06, "loss": 0.5258, "step": 754 }, { "epoch": 0.038191107289190146, "grad_norm": 14.407252311706543, "learning_rate": 9.980027392588461e-06, "loss": 0.476, "step": 755 }, { "epoch": 0.0382416915372553, "grad_norm": 8.636924743652344, "learning_rate": 9.97995566128192e-06, "loss": 0.5637, "step": 756 }, { "epoch": 0.03829227578532045, "grad_norm": 10.806689262390137, "learning_rate": 9.979883801653947e-06, "loss": 0.5, "step": 757 }, { "epoch": 0.038342860033385606, "grad_norm": 11.399323463439941, "learning_rate": 9.979811813706398e-06, "loss": 0.5892, "step": 758 }, { "epoch": 0.03839344428145076, "grad_norm": 12.457462310791016, "learning_rate": 9.979739697441127e-06, "loss": 0.4776, "step": 759 }, { "epoch": 0.03844402852951591, "grad_norm": 10.528178215026855, "learning_rate": 9.97966745285999e-06, "loss": 0.4649, "step": 760 }, { "epoch": 0.03849461277758106, "grad_norm": 21.1678409576416, "learning_rate": 9.979595079964855e-06, "loss": 0.69, "step": 761 }, { "epoch": 0.03854519702564621, "grad_norm": 12.16005802154541, "learning_rate": 9.979522578757579e-06, "loss": 0.3709, "step": 762 }, { "epoch": 0.03859578127371137, "grad_norm": 23.015411376953125, "learning_rate": 9.979449949240034e-06, "loss": 0.4531, "step": 763 }, { "epoch": 0.03864636552177652, "grad_norm": 26.106353759765625, "learning_rate": 9.979377191414093e-06, "loss": 0.4471, "step": 764 }, { "epoch": 0.03869694976984167, "grad_norm": 14.358628273010254, "learning_rate": 9.979304305281627e-06, "loss": 0.5918, "step": 765 }, { "epoch": 0.03874753401790682, "grad_norm": 12.303658485412598, "learning_rate": 9.979231290844518e-06, "loss": 0.5667, "step": 766 }, { "epoch": 0.03879811826597197, "grad_norm": 31.32155418395996, "learning_rate": 9.979158148104644e-06, "loss": 0.6641, "step": 767 }, { "epoch": 0.03884870251403713, "grad_norm": 16.403913497924805, "learning_rate": 9.97908487706389e-06, "loss": 0.5466, "step": 768 }, { "epoch": 0.03889928676210228, "grad_norm": 15.124350547790527, "learning_rate": 9.979011477724148e-06, "loss": 0.6223, "step": 769 }, { "epoch": 0.03894987101016743, "grad_norm": 13.702037811279297, "learning_rate": 9.978937950087305e-06, "loss": 0.5632, "step": 770 }, { "epoch": 0.039000455258232584, "grad_norm": 9.703871726989746, "learning_rate": 9.978864294155257e-06, "loss": 0.5139, "step": 771 }, { "epoch": 0.039051039506297736, "grad_norm": 13.659268379211426, "learning_rate": 9.978790509929904e-06, "loss": 0.4748, "step": 772 }, { "epoch": 0.039101623754362894, "grad_norm": 16.963512420654297, "learning_rate": 9.978716597413144e-06, "loss": 0.5452, "step": 773 }, { "epoch": 0.039152208002428045, "grad_norm": 10.021419525146484, "learning_rate": 9.978642556606884e-06, "loss": 0.5058, "step": 774 }, { "epoch": 0.039202792250493196, "grad_norm": 13.15272331237793, "learning_rate": 9.978568387513029e-06, "loss": 0.5562, "step": 775 }, { "epoch": 0.03925337649855835, "grad_norm": 9.901442527770996, "learning_rate": 9.978494090133492e-06, "loss": 0.5482, "step": 776 }, { "epoch": 0.0393039607466235, "grad_norm": 9.890872955322266, "learning_rate": 9.978419664470187e-06, "loss": 0.459, "step": 777 }, { "epoch": 0.039354544994688656, "grad_norm": 13.456164360046387, "learning_rate": 9.978345110525034e-06, "loss": 0.4815, "step": 778 }, { "epoch": 0.03940512924275381, "grad_norm": 13.87784194946289, "learning_rate": 9.97827042829995e-06, "loss": 0.6034, "step": 779 }, { "epoch": 0.03945571349081896, "grad_norm": 14.107504844665527, "learning_rate": 9.978195617796863e-06, "loss": 0.4705, "step": 780 }, { "epoch": 0.03950629773888411, "grad_norm": 13.252358436584473, "learning_rate": 9.978120679017699e-06, "loss": 0.5336, "step": 781 }, { "epoch": 0.03955688198694927, "grad_norm": 17.05323600769043, "learning_rate": 9.978045611964389e-06, "loss": 0.5373, "step": 782 }, { "epoch": 0.03960746623501442, "grad_norm": 6.2303547859191895, "learning_rate": 9.977970416638867e-06, "loss": 0.4085, "step": 783 }, { "epoch": 0.03965805048307957, "grad_norm": 11.293170928955078, "learning_rate": 9.977895093043072e-06, "loss": 0.5165, "step": 784 }, { "epoch": 0.03970863473114472, "grad_norm": 11.540691375732422, "learning_rate": 9.977819641178945e-06, "loss": 0.5338, "step": 785 }, { "epoch": 0.03975921897920987, "grad_norm": 12.920918464660645, "learning_rate": 9.977744061048429e-06, "loss": 0.5291, "step": 786 }, { "epoch": 0.03980980322727503, "grad_norm": 9.098905563354492, "learning_rate": 9.97766835265347e-06, "loss": 0.5087, "step": 787 }, { "epoch": 0.03986038747534018, "grad_norm": 12.40237045288086, "learning_rate": 9.977592515996023e-06, "loss": 0.5081, "step": 788 }, { "epoch": 0.03991097172340533, "grad_norm": 18.232845306396484, "learning_rate": 9.977516551078037e-06, "loss": 0.6556, "step": 789 }, { "epoch": 0.03996155597147048, "grad_norm": 9.395824432373047, "learning_rate": 9.977440457901474e-06, "loss": 0.4878, "step": 790 }, { "epoch": 0.040012140219535634, "grad_norm": 12.2406644821167, "learning_rate": 9.977364236468292e-06, "loss": 0.5662, "step": 791 }, { "epoch": 0.04006272446760079, "grad_norm": 22.391077041625977, "learning_rate": 9.977287886780457e-06, "loss": 0.5501, "step": 792 }, { "epoch": 0.040113308715665943, "grad_norm": 11.471529006958008, "learning_rate": 9.977211408839935e-06, "loss": 0.4198, "step": 793 }, { "epoch": 0.040163892963731095, "grad_norm": 8.245503425598145, "learning_rate": 9.977134802648696e-06, "loss": 0.5101, "step": 794 }, { "epoch": 0.040214477211796246, "grad_norm": 31.616619110107422, "learning_rate": 9.977058068208715e-06, "loss": 0.56, "step": 795 }, { "epoch": 0.0402650614598614, "grad_norm": 11.268345832824707, "learning_rate": 9.976981205521971e-06, "loss": 0.4046, "step": 796 }, { "epoch": 0.040315645707926555, "grad_norm": 15.9933443069458, "learning_rate": 9.97690421459044e-06, "loss": 0.5421, "step": 797 }, { "epoch": 0.040366229955991706, "grad_norm": 17.500228881835938, "learning_rate": 9.976827095416113e-06, "loss": 0.5572, "step": 798 }, { "epoch": 0.04041681420405686, "grad_norm": 9.937352180480957, "learning_rate": 9.97674984800097e-06, "loss": 0.4816, "step": 799 }, { "epoch": 0.04046739845212201, "grad_norm": 16.865089416503906, "learning_rate": 9.976672472347004e-06, "loss": 0.5459, "step": 800 }, { "epoch": 0.04051798270018716, "grad_norm": 13.221504211425781, "learning_rate": 9.976594968456208e-06, "loss": 0.5646, "step": 801 }, { "epoch": 0.04056856694825232, "grad_norm": 8.530221939086914, "learning_rate": 9.976517336330583e-06, "loss": 0.4759, "step": 802 }, { "epoch": 0.04061915119631747, "grad_norm": 10.148481369018555, "learning_rate": 9.976439575972123e-06, "loss": 0.4237, "step": 803 }, { "epoch": 0.04066973544438262, "grad_norm": 16.289262771606445, "learning_rate": 9.976361687382839e-06, "loss": 0.4494, "step": 804 }, { "epoch": 0.04072031969244777, "grad_norm": 6.571033477783203, "learning_rate": 9.976283670564732e-06, "loss": 0.5337, "step": 805 }, { "epoch": 0.04077090394051292, "grad_norm": 8.266644477844238, "learning_rate": 9.976205525519815e-06, "loss": 0.4552, "step": 806 }, { "epoch": 0.04082148818857808, "grad_norm": 16.05561065673828, "learning_rate": 9.976127252250101e-06, "loss": 0.5524, "step": 807 }, { "epoch": 0.04087207243664323, "grad_norm": 15.877915382385254, "learning_rate": 9.976048850757607e-06, "loss": 0.5742, "step": 808 }, { "epoch": 0.04092265668470838, "grad_norm": 25.722869873046875, "learning_rate": 9.975970321044352e-06, "loss": 0.637, "step": 809 }, { "epoch": 0.04097324093277353, "grad_norm": 16.22998809814453, "learning_rate": 9.975891663112363e-06, "loss": 0.5359, "step": 810 }, { "epoch": 0.041023825180838684, "grad_norm": 49.13504409790039, "learning_rate": 9.975812876963664e-06, "loss": 0.6927, "step": 811 }, { "epoch": 0.04107440942890384, "grad_norm": 9.890056610107422, "learning_rate": 9.975733962600287e-06, "loss": 0.4979, "step": 812 }, { "epoch": 0.04112499367696899, "grad_norm": 12.326412200927734, "learning_rate": 9.97565492002426e-06, "loss": 0.506, "step": 813 }, { "epoch": 0.041175577925034144, "grad_norm": 13.532567977905273, "learning_rate": 9.975575749237628e-06, "loss": 0.5064, "step": 814 }, { "epoch": 0.041226162173099296, "grad_norm": 13.602232933044434, "learning_rate": 9.975496450242425e-06, "loss": 0.5252, "step": 815 }, { "epoch": 0.04127674642116445, "grad_norm": 10.280238151550293, "learning_rate": 9.975417023040698e-06, "loss": 0.4903, "step": 816 }, { "epoch": 0.041327330669229605, "grad_norm": 15.700023651123047, "learning_rate": 9.97533746763449e-06, "loss": 0.5983, "step": 817 }, { "epoch": 0.041377914917294756, "grad_norm": 17.5142765045166, "learning_rate": 9.975257784025854e-06, "loss": 0.599, "step": 818 }, { "epoch": 0.04142849916535991, "grad_norm": 15.49537181854248, "learning_rate": 9.975177972216842e-06, "loss": 0.4833, "step": 819 }, { "epoch": 0.04147908341342506, "grad_norm": 29.755191802978516, "learning_rate": 9.97509803220951e-06, "loss": 0.6198, "step": 820 }, { "epoch": 0.04152966766149021, "grad_norm": 20.04233741760254, "learning_rate": 9.97501796400592e-06, "loss": 0.5494, "step": 821 }, { "epoch": 0.04158025190955537, "grad_norm": 21.202220916748047, "learning_rate": 9.974937767608133e-06, "loss": 0.6735, "step": 822 }, { "epoch": 0.04163083615762052, "grad_norm": 7.873749256134033, "learning_rate": 9.974857443018215e-06, "loss": 0.4689, "step": 823 }, { "epoch": 0.04168142040568567, "grad_norm": 11.914811134338379, "learning_rate": 9.974776990238239e-06, "loss": 0.529, "step": 824 }, { "epoch": 0.04173200465375082, "grad_norm": 10.4163236618042, "learning_rate": 9.974696409270274e-06, "loss": 0.4488, "step": 825 }, { "epoch": 0.04178258890181597, "grad_norm": 9.47976016998291, "learning_rate": 9.974615700116401e-06, "loss": 0.5092, "step": 826 }, { "epoch": 0.04183317314988113, "grad_norm": 18.05750274658203, "learning_rate": 9.974534862778696e-06, "loss": 0.5172, "step": 827 }, { "epoch": 0.04188375739794628, "grad_norm": 9.459668159484863, "learning_rate": 9.974453897259242e-06, "loss": 0.5117, "step": 828 }, { "epoch": 0.04193434164601143, "grad_norm": 8.325526237487793, "learning_rate": 9.974372803560125e-06, "loss": 0.4608, "step": 829 }, { "epoch": 0.04198492589407658, "grad_norm": 22.71495246887207, "learning_rate": 9.97429158168344e-06, "loss": 0.6473, "step": 830 }, { "epoch": 0.042035510142141734, "grad_norm": 20.229827880859375, "learning_rate": 9.974210231631274e-06, "loss": 0.6017, "step": 831 }, { "epoch": 0.04208609439020689, "grad_norm": 11.560651779174805, "learning_rate": 9.974128753405725e-06, "loss": 0.5187, "step": 832 }, { "epoch": 0.04213667863827204, "grad_norm": 22.96588706970215, "learning_rate": 9.974047147008892e-06, "loss": 0.6094, "step": 833 }, { "epoch": 0.042187262886337194, "grad_norm": 18.286617279052734, "learning_rate": 9.973965412442879e-06, "loss": 0.6195, "step": 834 }, { "epoch": 0.042237847134402345, "grad_norm": 13.53775691986084, "learning_rate": 9.973883549709791e-06, "loss": 0.4997, "step": 835 }, { "epoch": 0.042288431382467496, "grad_norm": 10.115012168884277, "learning_rate": 9.973801558811737e-06, "loss": 0.5833, "step": 836 }, { "epoch": 0.042339015630532655, "grad_norm": 14.779763221740723, "learning_rate": 9.973719439750834e-06, "loss": 0.381, "step": 837 }, { "epoch": 0.042389599878597806, "grad_norm": 12.072443962097168, "learning_rate": 9.973637192529191e-06, "loss": 0.5317, "step": 838 }, { "epoch": 0.04244018412666296, "grad_norm": 15.426185607910156, "learning_rate": 9.973554817148932e-06, "loss": 0.429, "step": 839 }, { "epoch": 0.04249076837472811, "grad_norm": 15.787618637084961, "learning_rate": 9.973472313612179e-06, "loss": 0.5741, "step": 840 }, { "epoch": 0.04254135262279326, "grad_norm": 49.02326583862305, "learning_rate": 9.973389681921057e-06, "loss": 0.5266, "step": 841 }, { "epoch": 0.04259193687085842, "grad_norm": 13.008159637451172, "learning_rate": 9.973306922077694e-06, "loss": 0.4488, "step": 842 }, { "epoch": 0.04264252111892357, "grad_norm": 15.301721572875977, "learning_rate": 9.973224034084227e-06, "loss": 0.6406, "step": 843 }, { "epoch": 0.04269310536698872, "grad_norm": 13.677992820739746, "learning_rate": 9.973141017942787e-06, "loss": 0.3941, "step": 844 }, { "epoch": 0.04274368961505387, "grad_norm": 14.438179969787598, "learning_rate": 9.973057873655517e-06, "loss": 0.6866, "step": 845 }, { "epoch": 0.04279427386311902, "grad_norm": 16.88158416748047, "learning_rate": 9.972974601224555e-06, "loss": 0.4758, "step": 846 }, { "epoch": 0.04284485811118418, "grad_norm": 11.454526901245117, "learning_rate": 9.97289120065205e-06, "loss": 0.474, "step": 847 }, { "epoch": 0.04289544235924933, "grad_norm": 9.838461875915527, "learning_rate": 9.972807671940152e-06, "loss": 0.5096, "step": 848 }, { "epoch": 0.04294602660731448, "grad_norm": 12.26262378692627, "learning_rate": 9.972724015091011e-06, "loss": 0.4809, "step": 849 }, { "epoch": 0.04299661085537963, "grad_norm": 10.555598258972168, "learning_rate": 9.972640230106783e-06, "loss": 0.4766, "step": 850 }, { "epoch": 0.043047195103444784, "grad_norm": 8.765241622924805, "learning_rate": 9.972556316989625e-06, "loss": 0.559, "step": 851 }, { "epoch": 0.04309777935150994, "grad_norm": 17.685035705566406, "learning_rate": 9.972472275741705e-06, "loss": 0.5214, "step": 852 }, { "epoch": 0.04314836359957509, "grad_norm": 11.459569931030273, "learning_rate": 9.972388106365183e-06, "loss": 0.5981, "step": 853 }, { "epoch": 0.043198947847640244, "grad_norm": 9.627740859985352, "learning_rate": 9.97230380886223e-06, "loss": 0.3864, "step": 854 }, { "epoch": 0.043249532095705395, "grad_norm": 44.398433685302734, "learning_rate": 9.972219383235018e-06, "loss": 0.7987, "step": 855 }, { "epoch": 0.04330011634377055, "grad_norm": 13.152433395385742, "learning_rate": 9.972134829485722e-06, "loss": 0.6656, "step": 856 }, { "epoch": 0.043350700591835704, "grad_norm": 19.52175521850586, "learning_rate": 9.972050147616523e-06, "loss": 0.5603, "step": 857 }, { "epoch": 0.043401284839900856, "grad_norm": 27.553630828857422, "learning_rate": 9.971965337629599e-06, "loss": 0.599, "step": 858 }, { "epoch": 0.04345186908796601, "grad_norm": 13.051039695739746, "learning_rate": 9.971880399527139e-06, "loss": 0.6136, "step": 859 }, { "epoch": 0.04350245333603116, "grad_norm": 16.40554428100586, "learning_rate": 9.971795333311329e-06, "loss": 0.5534, "step": 860 }, { "epoch": 0.043553037584096316, "grad_norm": 15.75973892211914, "learning_rate": 9.971710138984363e-06, "loss": 0.4947, "step": 861 }, { "epoch": 0.04360362183216147, "grad_norm": 22.067773818969727, "learning_rate": 9.971624816548435e-06, "loss": 0.6059, "step": 862 }, { "epoch": 0.04365420608022662, "grad_norm": 11.488348960876465, "learning_rate": 9.971539366005745e-06, "loss": 0.5269, "step": 863 }, { "epoch": 0.04370479032829177, "grad_norm": 9.0611572265625, "learning_rate": 9.971453787358491e-06, "loss": 0.4845, "step": 864 }, { "epoch": 0.04375537457635692, "grad_norm": 13.868473052978516, "learning_rate": 9.971368080608883e-06, "loss": 0.5437, "step": 865 }, { "epoch": 0.04380595882442208, "grad_norm": 8.791155815124512, "learning_rate": 9.971282245759128e-06, "loss": 0.4604, "step": 866 }, { "epoch": 0.04385654307248723, "grad_norm": 16.424535751342773, "learning_rate": 9.971196282811437e-06, "loss": 0.5901, "step": 867 }, { "epoch": 0.04390712732055238, "grad_norm": 9.693004608154297, "learning_rate": 9.971110191768025e-06, "loss": 0.554, "step": 868 }, { "epoch": 0.04395771156861753, "grad_norm": 18.00870704650879, "learning_rate": 9.971023972631109e-06, "loss": 0.5508, "step": 869 }, { "epoch": 0.04400829581668268, "grad_norm": 17.04701805114746, "learning_rate": 9.970937625402915e-06, "loss": 0.5584, "step": 870 }, { "epoch": 0.04405888006474784, "grad_norm": 15.478034019470215, "learning_rate": 9.970851150085664e-06, "loss": 0.548, "step": 871 }, { "epoch": 0.04410946431281299, "grad_norm": 9.573702812194824, "learning_rate": 9.970764546681584e-06, "loss": 0.3831, "step": 872 }, { "epoch": 0.04416004856087814, "grad_norm": 8.045393943786621, "learning_rate": 9.970677815192909e-06, "loss": 0.5403, "step": 873 }, { "epoch": 0.044210632808943294, "grad_norm": 5.587521076202393, "learning_rate": 9.970590955621873e-06, "loss": 0.5232, "step": 874 }, { "epoch": 0.044261217057008445, "grad_norm": 11.402976989746094, "learning_rate": 9.970503967970714e-06, "loss": 0.5277, "step": 875 }, { "epoch": 0.0443118013050736, "grad_norm": 10.142276763916016, "learning_rate": 9.970416852241674e-06, "loss": 0.4714, "step": 876 }, { "epoch": 0.044362385553138754, "grad_norm": 6.480471611022949, "learning_rate": 9.970329608436996e-06, "loss": 0.577, "step": 877 }, { "epoch": 0.044412969801203905, "grad_norm": 10.420592308044434, "learning_rate": 9.970242236558931e-06, "loss": 0.5335, "step": 878 }, { "epoch": 0.044463554049269056, "grad_norm": 8.213703155517578, "learning_rate": 9.970154736609726e-06, "loss": 0.4857, "step": 879 }, { "epoch": 0.04451413829733421, "grad_norm": 9.61079216003418, "learning_rate": 9.97006710859164e-06, "loss": 0.517, "step": 880 }, { "epoch": 0.044564722545399366, "grad_norm": 18.296472549438477, "learning_rate": 9.96997935250693e-06, "loss": 0.5769, "step": 881 }, { "epoch": 0.04461530679346452, "grad_norm": 13.347856521606445, "learning_rate": 9.969891468357857e-06, "loss": 0.6004, "step": 882 }, { "epoch": 0.04466589104152967, "grad_norm": 8.750860214233398, "learning_rate": 9.969803456146684e-06, "loss": 0.4694, "step": 883 }, { "epoch": 0.04471647528959482, "grad_norm": 8.071752548217773, "learning_rate": 9.96971531587568e-06, "loss": 0.4494, "step": 884 }, { "epoch": 0.04476705953765997, "grad_norm": 8.951651573181152, "learning_rate": 9.969627047547116e-06, "loss": 0.4954, "step": 885 }, { "epoch": 0.04481764378572513, "grad_norm": 9.137948989868164, "learning_rate": 9.969538651163267e-06, "loss": 0.5319, "step": 886 }, { "epoch": 0.04486822803379028, "grad_norm": 20.689241409301758, "learning_rate": 9.969450126726412e-06, "loss": 0.5131, "step": 887 }, { "epoch": 0.04491881228185543, "grad_norm": 15.080933570861816, "learning_rate": 9.969361474238827e-06, "loss": 0.5745, "step": 888 }, { "epoch": 0.04496939652992058, "grad_norm": 30.206743240356445, "learning_rate": 9.969272693702802e-06, "loss": 0.526, "step": 889 }, { "epoch": 0.04501998077798573, "grad_norm": 8.9621000289917, "learning_rate": 9.969183785120624e-06, "loss": 0.3469, "step": 890 }, { "epoch": 0.04507056502605089, "grad_norm": 7.656215667724609, "learning_rate": 9.969094748494578e-06, "loss": 0.5141, "step": 891 }, { "epoch": 0.04512114927411604, "grad_norm": 13.074280738830566, "learning_rate": 9.969005583826967e-06, "loss": 0.5239, "step": 892 }, { "epoch": 0.04517173352218119, "grad_norm": 14.372026443481445, "learning_rate": 9.968916291120082e-06, "loss": 0.4996, "step": 893 }, { "epoch": 0.045222317770246344, "grad_norm": 12.610551834106445, "learning_rate": 9.968826870376226e-06, "loss": 0.528, "step": 894 }, { "epoch": 0.045272902018311495, "grad_norm": 18.274696350097656, "learning_rate": 9.968737321597705e-06, "loss": 0.4796, "step": 895 }, { "epoch": 0.04532348626637665, "grad_norm": 14.153239250183105, "learning_rate": 9.968647644786823e-06, "loss": 0.5209, "step": 896 }, { "epoch": 0.045374070514441804, "grad_norm": 20.702442169189453, "learning_rate": 9.968557839945894e-06, "loss": 0.4937, "step": 897 }, { "epoch": 0.045424654762506955, "grad_norm": 18.75519371032715, "learning_rate": 9.968467907077228e-06, "loss": 0.6158, "step": 898 }, { "epoch": 0.045475239010572106, "grad_norm": 16.916563034057617, "learning_rate": 9.968377846183148e-06, "loss": 0.472, "step": 899 }, { "epoch": 0.04552582325863726, "grad_norm": 10.982499122619629, "learning_rate": 9.96828765726597e-06, "loss": 0.5424, "step": 900 }, { "epoch": 0.045576407506702415, "grad_norm": 20.041780471801758, "learning_rate": 9.968197340328018e-06, "loss": 0.6322, "step": 901 }, { "epoch": 0.04562699175476757, "grad_norm": 15.328235626220703, "learning_rate": 9.968106895371623e-06, "loss": 0.5637, "step": 902 }, { "epoch": 0.04567757600283272, "grad_norm": 18.385295867919922, "learning_rate": 9.968016322399112e-06, "loss": 0.6178, "step": 903 }, { "epoch": 0.04572816025089787, "grad_norm": 11.29980182647705, "learning_rate": 9.967925621412821e-06, "loss": 0.5323, "step": 904 }, { "epoch": 0.04577874449896302, "grad_norm": 17.596567153930664, "learning_rate": 9.967834792415085e-06, "loss": 0.4768, "step": 905 }, { "epoch": 0.04582932874702818, "grad_norm": 24.7388858795166, "learning_rate": 9.967743835408247e-06, "loss": 0.6621, "step": 906 }, { "epoch": 0.04587991299509333, "grad_norm": 8.843932151794434, "learning_rate": 9.96765275039465e-06, "loss": 0.4687, "step": 907 }, { "epoch": 0.04593049724315848, "grad_norm": 12.726505279541016, "learning_rate": 9.96756153737664e-06, "loss": 0.4946, "step": 908 }, { "epoch": 0.04598108149122363, "grad_norm": 9.025435447692871, "learning_rate": 9.967470196356566e-06, "loss": 0.5806, "step": 909 }, { "epoch": 0.04603166573928878, "grad_norm": 15.206401824951172, "learning_rate": 9.967378727336785e-06, "loss": 0.6634, "step": 910 }, { "epoch": 0.04608224998735394, "grad_norm": 11.286425590515137, "learning_rate": 9.967287130319652e-06, "loss": 0.5359, "step": 911 }, { "epoch": 0.04613283423541909, "grad_norm": 11.216166496276855, "learning_rate": 9.967195405307528e-06, "loss": 0.5562, "step": 912 }, { "epoch": 0.04618341848348424, "grad_norm": 9.701642990112305, "learning_rate": 9.967103552302778e-06, "loss": 0.4017, "step": 913 }, { "epoch": 0.046234002731549394, "grad_norm": 11.106766700744629, "learning_rate": 9.967011571307763e-06, "loss": 0.3485, "step": 914 }, { "epoch": 0.046284586979614545, "grad_norm": 10.455779075622559, "learning_rate": 9.96691946232486e-06, "loss": 0.4681, "step": 915 }, { "epoch": 0.0463351712276797, "grad_norm": 9.300071716308594, "learning_rate": 9.966827225356439e-06, "loss": 0.5796, "step": 916 }, { "epoch": 0.046385755475744854, "grad_norm": 15.263216018676758, "learning_rate": 9.966734860404877e-06, "loss": 0.5759, "step": 917 }, { "epoch": 0.046436339723810005, "grad_norm": 9.328479766845703, "learning_rate": 9.966642367472554e-06, "loss": 0.5422, "step": 918 }, { "epoch": 0.046486923971875156, "grad_norm": 11.857305526733398, "learning_rate": 9.966549746561854e-06, "loss": 0.4784, "step": 919 }, { "epoch": 0.04653750821994031, "grad_norm": 10.494166374206543, "learning_rate": 9.966456997675163e-06, "loss": 0.4025, "step": 920 }, { "epoch": 0.046588092468005465, "grad_norm": 9.896831512451172, "learning_rate": 9.96636412081487e-06, "loss": 0.5294, "step": 921 }, { "epoch": 0.046638676716070616, "grad_norm": 12.961734771728516, "learning_rate": 9.966271115983373e-06, "loss": 0.5612, "step": 922 }, { "epoch": 0.04668926096413577, "grad_norm": 12.235654830932617, "learning_rate": 9.966177983183061e-06, "loss": 0.5373, "step": 923 }, { "epoch": 0.04673984521220092, "grad_norm": 12.542612075805664, "learning_rate": 9.966084722416341e-06, "loss": 0.5316, "step": 924 }, { "epoch": 0.04679042946026607, "grad_norm": 8.791284561157227, "learning_rate": 9.96599133368561e-06, "loss": 0.4908, "step": 925 }, { "epoch": 0.04684101370833123, "grad_norm": 14.87885570526123, "learning_rate": 9.965897816993279e-06, "loss": 0.5266, "step": 926 }, { "epoch": 0.04689159795639638, "grad_norm": 10.789162635803223, "learning_rate": 9.965804172341755e-06, "loss": 0.4531, "step": 927 }, { "epoch": 0.04694218220446153, "grad_norm": 13.902609825134277, "learning_rate": 9.965710399733451e-06, "loss": 0.5697, "step": 928 }, { "epoch": 0.04699276645252668, "grad_norm": 20.211977005004883, "learning_rate": 9.965616499170785e-06, "loss": 0.5437, "step": 929 }, { "epoch": 0.04704335070059184, "grad_norm": 17.389652252197266, "learning_rate": 9.965522470656178e-06, "loss": 0.6487, "step": 930 }, { "epoch": 0.04709393494865699, "grad_norm": 12.725837707519531, "learning_rate": 9.965428314192046e-06, "loss": 0.554, "step": 931 }, { "epoch": 0.04714451919672214, "grad_norm": 13.594388961791992, "learning_rate": 9.965334029780824e-06, "loss": 0.5137, "step": 932 }, { "epoch": 0.04719510344478729, "grad_norm": 11.621698379516602, "learning_rate": 9.965239617424935e-06, "loss": 0.5017, "step": 933 }, { "epoch": 0.047245687692852444, "grad_norm": 16.036964416503906, "learning_rate": 9.965145077126817e-06, "loss": 0.4779, "step": 934 }, { "epoch": 0.0472962719409176, "grad_norm": 9.655816078186035, "learning_rate": 9.9650504088889e-06, "loss": 0.4848, "step": 935 }, { "epoch": 0.04734685618898275, "grad_norm": 15.929134368896484, "learning_rate": 9.964955612713628e-06, "loss": 0.6356, "step": 936 }, { "epoch": 0.047397440437047904, "grad_norm": 10.773250579833984, "learning_rate": 9.964860688603441e-06, "loss": 0.4183, "step": 937 }, { "epoch": 0.047448024685113055, "grad_norm": 15.554764747619629, "learning_rate": 9.964765636560787e-06, "loss": 0.4252, "step": 938 }, { "epoch": 0.047498608933178206, "grad_norm": 15.256807327270508, "learning_rate": 9.964670456588115e-06, "loss": 0.6299, "step": 939 }, { "epoch": 0.047549193181243364, "grad_norm": 14.493980407714844, "learning_rate": 9.964575148687877e-06, "loss": 0.4046, "step": 940 }, { "epoch": 0.047599777429308515, "grad_norm": 10.354348182678223, "learning_rate": 9.964479712862529e-06, "loss": 0.4229, "step": 941 }, { "epoch": 0.047650361677373666, "grad_norm": 11.21027946472168, "learning_rate": 9.964384149114528e-06, "loss": 0.4612, "step": 942 }, { "epoch": 0.04770094592543882, "grad_norm": 14.417557716369629, "learning_rate": 9.964288457446341e-06, "loss": 0.3741, "step": 943 }, { "epoch": 0.04775153017350397, "grad_norm": 20.586750030517578, "learning_rate": 9.96419263786043e-06, "loss": 0.4725, "step": 944 }, { "epoch": 0.04780211442156913, "grad_norm": 12.036109924316406, "learning_rate": 9.964096690359264e-06, "loss": 0.4739, "step": 945 }, { "epoch": 0.04785269866963428, "grad_norm": 24.724834442138672, "learning_rate": 9.964000614945318e-06, "loss": 0.7709, "step": 946 }, { "epoch": 0.04790328291769943, "grad_norm": 51.18813705444336, "learning_rate": 9.963904411621065e-06, "loss": 0.5244, "step": 947 }, { "epoch": 0.04795386716576458, "grad_norm": 11.361013412475586, "learning_rate": 9.963808080388988e-06, "loss": 0.4875, "step": 948 }, { "epoch": 0.04800445141382973, "grad_norm": 14.173946380615234, "learning_rate": 9.963711621251564e-06, "loss": 0.5505, "step": 949 }, { "epoch": 0.04805503566189489, "grad_norm": 14.589447975158691, "learning_rate": 9.963615034211283e-06, "loss": 0.5104, "step": 950 }, { "epoch": 0.04810561990996004, "grad_norm": 26.674238204956055, "learning_rate": 9.96351831927063e-06, "loss": 0.4816, "step": 951 }, { "epoch": 0.04815620415802519, "grad_norm": 13.22658920288086, "learning_rate": 9.963421476432098e-06, "loss": 0.5636, "step": 952 }, { "epoch": 0.04820678840609034, "grad_norm": 15.064470291137695, "learning_rate": 9.963324505698184e-06, "loss": 0.5249, "step": 953 }, { "epoch": 0.04825737265415549, "grad_norm": 16.945783615112305, "learning_rate": 9.963227407071385e-06, "loss": 0.6287, "step": 954 }, { "epoch": 0.04830795690222065, "grad_norm": 8.167441368103027, "learning_rate": 9.963130180554205e-06, "loss": 0.5142, "step": 955 }, { "epoch": 0.0483585411502858, "grad_norm": 12.820926666259766, "learning_rate": 9.963032826149149e-06, "loss": 0.4348, "step": 956 }, { "epoch": 0.048409125398350954, "grad_norm": 9.472015380859375, "learning_rate": 9.962935343858724e-06, "loss": 0.5759, "step": 957 }, { "epoch": 0.048459709646416105, "grad_norm": 8.025760650634766, "learning_rate": 9.962837733685443e-06, "loss": 0.495, "step": 958 }, { "epoch": 0.048510293894481256, "grad_norm": 5.839386940002441, "learning_rate": 9.96273999563182e-06, "loss": 0.3971, "step": 959 }, { "epoch": 0.048560878142546414, "grad_norm": 6.05701208114624, "learning_rate": 9.962642129700375e-06, "loss": 0.5011, "step": 960 }, { "epoch": 0.048611462390611565, "grad_norm": 10.830724716186523, "learning_rate": 9.962544135893628e-06, "loss": 0.4561, "step": 961 }, { "epoch": 0.048662046638676716, "grad_norm": 19.778181076049805, "learning_rate": 9.962446014214106e-06, "loss": 0.5347, "step": 962 }, { "epoch": 0.04871263088674187, "grad_norm": 14.602739334106445, "learning_rate": 9.962347764664338e-06, "loss": 0.5578, "step": 963 }, { "epoch": 0.04876321513480702, "grad_norm": 20.57383155822754, "learning_rate": 9.962249387246851e-06, "loss": 0.6219, "step": 964 }, { "epoch": 0.048813799382872176, "grad_norm": 17.058612823486328, "learning_rate": 9.962150881964185e-06, "loss": 0.3976, "step": 965 }, { "epoch": 0.04886438363093733, "grad_norm": 10.435853004455566, "learning_rate": 9.962052248818877e-06, "loss": 0.5194, "step": 966 }, { "epoch": 0.04891496787900248, "grad_norm": 23.97966194152832, "learning_rate": 9.961953487813468e-06, "loss": 0.582, "step": 967 }, { "epoch": 0.04896555212706763, "grad_norm": 14.730279922485352, "learning_rate": 9.9618545989505e-06, "loss": 0.5714, "step": 968 }, { "epoch": 0.04901613637513278, "grad_norm": 10.136567115783691, "learning_rate": 9.961755582232527e-06, "loss": 0.4782, "step": 969 }, { "epoch": 0.04906672062319794, "grad_norm": 17.37933349609375, "learning_rate": 9.961656437662094e-06, "loss": 0.6229, "step": 970 }, { "epoch": 0.04911730487126309, "grad_norm": 5.955059051513672, "learning_rate": 9.961557165241763e-06, "loss": 0.4257, "step": 971 }, { "epoch": 0.04916788911932824, "grad_norm": 10.97181510925293, "learning_rate": 9.961457764974084e-06, "loss": 0.421, "step": 972 }, { "epoch": 0.04921847336739339, "grad_norm": 16.543750762939453, "learning_rate": 9.961358236861625e-06, "loss": 0.5167, "step": 973 }, { "epoch": 0.04926905761545854, "grad_norm": 13.430973052978516, "learning_rate": 9.961258580906948e-06, "loss": 0.4658, "step": 974 }, { "epoch": 0.0493196418635237, "grad_norm": 20.534582138061523, "learning_rate": 9.961158797112619e-06, "loss": 0.4507, "step": 975 }, { "epoch": 0.04937022611158885, "grad_norm": 33.99174880981445, "learning_rate": 9.961058885481211e-06, "loss": 0.4634, "step": 976 }, { "epoch": 0.049420810359654004, "grad_norm": 17.73912811279297, "learning_rate": 9.9609588460153e-06, "loss": 0.4433, "step": 977 }, { "epoch": 0.049471394607719155, "grad_norm": 14.736488342285156, "learning_rate": 9.960858678717461e-06, "loss": 0.4595, "step": 978 }, { "epoch": 0.049521978855784306, "grad_norm": 37.54460144042969, "learning_rate": 9.960758383590278e-06, "loss": 0.6104, "step": 979 }, { "epoch": 0.049572563103849464, "grad_norm": 13.081777572631836, "learning_rate": 9.960657960636334e-06, "loss": 0.4894, "step": 980 }, { "epoch": 0.049623147351914615, "grad_norm": 14.846126556396484, "learning_rate": 9.960557409858213e-06, "loss": 0.5608, "step": 981 }, { "epoch": 0.049673731599979766, "grad_norm": 19.66621208190918, "learning_rate": 9.960456731258512e-06, "loss": 0.5495, "step": 982 }, { "epoch": 0.04972431584804492, "grad_norm": 10.835927963256836, "learning_rate": 9.960355924839822e-06, "loss": 0.5649, "step": 983 }, { "epoch": 0.04977490009611007, "grad_norm": 11.932440757751465, "learning_rate": 9.96025499060474e-06, "loss": 0.5498, "step": 984 }, { "epoch": 0.049825484344175226, "grad_norm": 14.27657413482666, "learning_rate": 9.960153928555868e-06, "loss": 0.4628, "step": 985 }, { "epoch": 0.04987606859224038, "grad_norm": 10.412836074829102, "learning_rate": 9.96005273869581e-06, "loss": 0.4513, "step": 986 }, { "epoch": 0.04992665284030553, "grad_norm": 18.489974975585938, "learning_rate": 9.959951421027175e-06, "loss": 0.4834, "step": 987 }, { "epoch": 0.04997723708837068, "grad_norm": 13.137764930725098, "learning_rate": 9.959849975552572e-06, "loss": 0.5684, "step": 988 }, { "epoch": 0.05002782133643583, "grad_norm": 12.459959983825684, "learning_rate": 9.959748402274614e-06, "loss": 0.5568, "step": 989 }, { "epoch": 0.05007840558450099, "grad_norm": 12.571123123168945, "learning_rate": 9.95964670119592e-06, "loss": 0.4499, "step": 990 }, { "epoch": 0.05012898983256614, "grad_norm": 16.10092544555664, "learning_rate": 9.95954487231911e-06, "loss": 0.5202, "step": 991 }, { "epoch": 0.05017957408063129, "grad_norm": 15.502323150634766, "learning_rate": 9.959442915646809e-06, "loss": 0.546, "step": 992 }, { "epoch": 0.05023015832869644, "grad_norm": 18.5846004486084, "learning_rate": 9.959340831181641e-06, "loss": 0.5719, "step": 993 }, { "epoch": 0.05028074257676159, "grad_norm": 11.177858352661133, "learning_rate": 9.95923861892624e-06, "loss": 0.5123, "step": 994 }, { "epoch": 0.05033132682482675, "grad_norm": 14.330747604370117, "learning_rate": 9.959136278883237e-06, "loss": 0.5335, "step": 995 }, { "epoch": 0.0503819110728919, "grad_norm": 39.295509338378906, "learning_rate": 9.959033811055271e-06, "loss": 0.5746, "step": 996 }, { "epoch": 0.05043249532095705, "grad_norm": 8.81405258178711, "learning_rate": 9.958931215444982e-06, "loss": 0.4822, "step": 997 }, { "epoch": 0.050483079569022205, "grad_norm": 10.221955299377441, "learning_rate": 9.958828492055015e-06, "loss": 0.4034, "step": 998 }, { "epoch": 0.050533663817087356, "grad_norm": 15.929289817810059, "learning_rate": 9.958725640888013e-06, "loss": 0.5184, "step": 999 }, { "epoch": 0.050584248065152514, "grad_norm": 14.079445838928223, "learning_rate": 9.95862266194663e-06, "loss": 0.4051, "step": 1000 }, { "epoch": 0.050634832313217665, "grad_norm": 20.313657760620117, "learning_rate": 9.958519555233517e-06, "loss": 0.5835, "step": 1001 }, { "epoch": 0.050685416561282816, "grad_norm": 10.543672561645508, "learning_rate": 9.958416320751331e-06, "loss": 0.5246, "step": 1002 }, { "epoch": 0.05073600080934797, "grad_norm": 7.314309597015381, "learning_rate": 9.958312958502736e-06, "loss": 0.5414, "step": 1003 }, { "epoch": 0.050786585057413125, "grad_norm": 12.874102592468262, "learning_rate": 9.95820946849039e-06, "loss": 0.4686, "step": 1004 }, { "epoch": 0.050837169305478276, "grad_norm": 10.385843276977539, "learning_rate": 9.958105850716962e-06, "loss": 0.5175, "step": 1005 }, { "epoch": 0.05088775355354343, "grad_norm": 14.231083869934082, "learning_rate": 9.958002105185123e-06, "loss": 0.5402, "step": 1006 }, { "epoch": 0.05093833780160858, "grad_norm": 26.8487491607666, "learning_rate": 9.957898231897544e-06, "loss": 0.7583, "step": 1007 }, { "epoch": 0.05098892204967373, "grad_norm": 11.567831993103027, "learning_rate": 9.957794230856905e-06, "loss": 0.5286, "step": 1008 }, { "epoch": 0.05103950629773889, "grad_norm": 6.296516418457031, "learning_rate": 9.957690102065882e-06, "loss": 0.4264, "step": 1009 }, { "epoch": 0.05109009054580404, "grad_norm": 11.30919075012207, "learning_rate": 9.957585845527162e-06, "loss": 0.432, "step": 1010 }, { "epoch": 0.05114067479386919, "grad_norm": 23.9431209564209, "learning_rate": 9.957481461243428e-06, "loss": 0.5659, "step": 1011 }, { "epoch": 0.05119125904193434, "grad_norm": 9.171560287475586, "learning_rate": 9.95737694921737e-06, "loss": 0.4598, "step": 1012 }, { "epoch": 0.05124184328999949, "grad_norm": 13.07469654083252, "learning_rate": 9.957272309451682e-06, "loss": 0.5428, "step": 1013 }, { "epoch": 0.05129242753806465, "grad_norm": 9.931997299194336, "learning_rate": 9.957167541949062e-06, "loss": 0.5147, "step": 1014 }, { "epoch": 0.0513430117861298, "grad_norm": 15.219779968261719, "learning_rate": 9.957062646712207e-06, "loss": 0.5267, "step": 1015 }, { "epoch": 0.05139359603419495, "grad_norm": 13.496488571166992, "learning_rate": 9.956957623743821e-06, "loss": 0.5692, "step": 1016 }, { "epoch": 0.0514441802822601, "grad_norm": 16.903223037719727, "learning_rate": 9.956852473046612e-06, "loss": 0.6558, "step": 1017 }, { "epoch": 0.051494764530325254, "grad_norm": 10.641776084899902, "learning_rate": 9.956747194623285e-06, "loss": 0.4581, "step": 1018 }, { "epoch": 0.05154534877839041, "grad_norm": 11.297843933105469, "learning_rate": 9.956641788476556e-06, "loss": 0.4746, "step": 1019 }, { "epoch": 0.051595933026455564, "grad_norm": 8.104442596435547, "learning_rate": 9.956536254609141e-06, "loss": 0.4935, "step": 1020 }, { "epoch": 0.051646517274520715, "grad_norm": 12.489080429077148, "learning_rate": 9.956430593023759e-06, "loss": 0.5544, "step": 1021 }, { "epoch": 0.051697101522585866, "grad_norm": 26.130146026611328, "learning_rate": 9.95632480372313e-06, "loss": 0.5269, "step": 1022 }, { "epoch": 0.05174768577065102, "grad_norm": 8.709075927734375, "learning_rate": 9.956218886709985e-06, "loss": 0.4958, "step": 1023 }, { "epoch": 0.051798270018716175, "grad_norm": 9.610030174255371, "learning_rate": 9.956112841987049e-06, "loss": 0.482, "step": 1024 }, { "epoch": 0.051848854266781326, "grad_norm": 8.743727684020996, "learning_rate": 9.956006669557057e-06, "loss": 0.5522, "step": 1025 }, { "epoch": 0.05189943851484648, "grad_norm": 5.919762134552002, "learning_rate": 9.955900369422745e-06, "loss": 0.4857, "step": 1026 }, { "epoch": 0.05195002276291163, "grad_norm": 8.976361274719238, "learning_rate": 9.955793941586847e-06, "loss": 0.4051, "step": 1027 }, { "epoch": 0.05200060701097678, "grad_norm": 25.71986198425293, "learning_rate": 9.955687386052113e-06, "loss": 0.5225, "step": 1028 }, { "epoch": 0.05205119125904194, "grad_norm": 7.300754070281982, "learning_rate": 9.955580702821284e-06, "loss": 0.4693, "step": 1029 }, { "epoch": 0.05210177550710709, "grad_norm": 10.064770698547363, "learning_rate": 9.955473891897109e-06, "loss": 0.4929, "step": 1030 }, { "epoch": 0.05215235975517224, "grad_norm": 9.058404922485352, "learning_rate": 9.955366953282341e-06, "loss": 0.4655, "step": 1031 }, { "epoch": 0.05220294400323739, "grad_norm": 18.768321990966797, "learning_rate": 9.955259886979738e-06, "loss": 0.6228, "step": 1032 }, { "epoch": 0.05225352825130254, "grad_norm": 20.089397430419922, "learning_rate": 9.955152692992057e-06, "loss": 0.4954, "step": 1033 }, { "epoch": 0.0523041124993677, "grad_norm": 9.067694664001465, "learning_rate": 9.955045371322058e-06, "loss": 0.489, "step": 1034 }, { "epoch": 0.05235469674743285, "grad_norm": 9.617795944213867, "learning_rate": 9.954937921972506e-06, "loss": 0.4909, "step": 1035 }, { "epoch": 0.052405280995498, "grad_norm": 9.16403579711914, "learning_rate": 9.954830344946176e-06, "loss": 0.4342, "step": 1036 }, { "epoch": 0.05245586524356315, "grad_norm": 15.477633476257324, "learning_rate": 9.954722640245835e-06, "loss": 0.6455, "step": 1037 }, { "epoch": 0.052506449491628304, "grad_norm": 16.10988998413086, "learning_rate": 9.95461480787426e-06, "loss": 0.3893, "step": 1038 }, { "epoch": 0.05255703373969346, "grad_norm": 27.73767852783203, "learning_rate": 9.954506847834226e-06, "loss": 0.4511, "step": 1039 }, { "epoch": 0.05260761798775861, "grad_norm": 22.52994728088379, "learning_rate": 9.954398760128519e-06, "loss": 0.5336, "step": 1040 }, { "epoch": 0.052658202235823764, "grad_norm": 17.175148010253906, "learning_rate": 9.954290544759923e-06, "loss": 0.4654, "step": 1041 }, { "epoch": 0.052708786483888916, "grad_norm": 15.868907928466797, "learning_rate": 9.954182201731226e-06, "loss": 0.507, "step": 1042 }, { "epoch": 0.05275937073195407, "grad_norm": 7.306356430053711, "learning_rate": 9.95407373104522e-06, "loss": 0.5365, "step": 1043 }, { "epoch": 0.052809954980019225, "grad_norm": 5.124794006347656, "learning_rate": 9.9539651327047e-06, "loss": 0.3223, "step": 1044 }, { "epoch": 0.052860539228084376, "grad_norm": 10.405784606933594, "learning_rate": 9.953856406712464e-06, "loss": 0.4465, "step": 1045 }, { "epoch": 0.05291112347614953, "grad_norm": 11.77574634552002, "learning_rate": 9.953747553071314e-06, "loss": 0.5027, "step": 1046 }, { "epoch": 0.05296170772421468, "grad_norm": 24.11672019958496, "learning_rate": 9.953638571784056e-06, "loss": 0.5028, "step": 1047 }, { "epoch": 0.05301229197227983, "grad_norm": 12.803863525390625, "learning_rate": 9.953529462853497e-06, "loss": 0.4635, "step": 1048 }, { "epoch": 0.05306287622034499, "grad_norm": 13.922802925109863, "learning_rate": 9.953420226282446e-06, "loss": 0.5817, "step": 1049 }, { "epoch": 0.05311346046841014, "grad_norm": 11.40183162689209, "learning_rate": 9.953310862073722e-06, "loss": 0.5075, "step": 1050 }, { "epoch": 0.05316404471647529, "grad_norm": 23.122806549072266, "learning_rate": 9.953201370230142e-06, "loss": 0.605, "step": 1051 }, { "epoch": 0.05321462896454044, "grad_norm": 24.32611846923828, "learning_rate": 9.953091750754527e-06, "loss": 0.6761, "step": 1052 }, { "epoch": 0.05326521321260559, "grad_norm": 24.961389541625977, "learning_rate": 9.9529820036497e-06, "loss": 0.7574, "step": 1053 }, { "epoch": 0.05331579746067075, "grad_norm": 10.7555570602417, "learning_rate": 9.95287212891849e-06, "loss": 0.6035, "step": 1054 }, { "epoch": 0.0533663817087359, "grad_norm": 17.21748924255371, "learning_rate": 9.952762126563729e-06, "loss": 0.5977, "step": 1055 }, { "epoch": 0.05341696595680105, "grad_norm": 15.59583568572998, "learning_rate": 9.952651996588252e-06, "loss": 0.462, "step": 1056 }, { "epoch": 0.0534675502048662, "grad_norm": 11.395957946777344, "learning_rate": 9.952541738994895e-06, "loss": 0.5339, "step": 1057 }, { "epoch": 0.053518134452931354, "grad_norm": 19.60047721862793, "learning_rate": 9.952431353786498e-06, "loss": 0.5896, "step": 1058 }, { "epoch": 0.05356871870099651, "grad_norm": 12.50003433227539, "learning_rate": 9.952320840965907e-06, "loss": 0.5945, "step": 1059 }, { "epoch": 0.05361930294906166, "grad_norm": 18.92740249633789, "learning_rate": 9.952210200535972e-06, "loss": 0.7049, "step": 1060 }, { "epoch": 0.053669887197126814, "grad_norm": 11.410401344299316, "learning_rate": 9.95209943249954e-06, "loss": 0.559, "step": 1061 }, { "epoch": 0.053720471445191965, "grad_norm": 13.833223342895508, "learning_rate": 9.95198853685947e-06, "loss": 0.5365, "step": 1062 }, { "epoch": 0.05377105569325712, "grad_norm": 17.20979118347168, "learning_rate": 9.951877513618613e-06, "loss": 0.5369, "step": 1063 }, { "epoch": 0.053821639941322275, "grad_norm": 9.816272735595703, "learning_rate": 9.951766362779836e-06, "loss": 0.4698, "step": 1064 }, { "epoch": 0.053872224189387426, "grad_norm": 13.474077224731445, "learning_rate": 9.951655084345996e-06, "loss": 0.5931, "step": 1065 }, { "epoch": 0.05392280843745258, "grad_norm": 11.962610244750977, "learning_rate": 9.95154367831997e-06, "loss": 0.5722, "step": 1066 }, { "epoch": 0.05397339268551773, "grad_norm": 8.701945304870605, "learning_rate": 9.95143214470462e-06, "loss": 0.5092, "step": 1067 }, { "epoch": 0.05402397693358288, "grad_norm": 15.024698257446289, "learning_rate": 9.951320483502824e-06, "loss": 0.4593, "step": 1068 }, { "epoch": 0.05407456118164804, "grad_norm": 9.607105255126953, "learning_rate": 9.951208694717459e-06, "loss": 0.5356, "step": 1069 }, { "epoch": 0.05412514542971319, "grad_norm": 11.614667892456055, "learning_rate": 9.951096778351405e-06, "loss": 0.4174, "step": 1070 }, { "epoch": 0.05417572967777834, "grad_norm": 16.60467529296875, "learning_rate": 9.950984734407545e-06, "loss": 0.6418, "step": 1071 }, { "epoch": 0.05422631392584349, "grad_norm": 10.73070240020752, "learning_rate": 9.950872562888769e-06, "loss": 0.5821, "step": 1072 }, { "epoch": 0.05427689817390864, "grad_norm": 8.498597145080566, "learning_rate": 9.950760263797964e-06, "loss": 0.4781, "step": 1073 }, { "epoch": 0.0543274824219738, "grad_norm": 11.248486518859863, "learning_rate": 9.950647837138025e-06, "loss": 0.4239, "step": 1074 }, { "epoch": 0.05437806667003895, "grad_norm": 17.297698974609375, "learning_rate": 9.95053528291185e-06, "loss": 0.6337, "step": 1075 }, { "epoch": 0.0544286509181041, "grad_norm": 13.093476295471191, "learning_rate": 9.950422601122338e-06, "loss": 0.5706, "step": 1076 }, { "epoch": 0.05447923516616925, "grad_norm": 9.779497146606445, "learning_rate": 9.95030979177239e-06, "loss": 0.6072, "step": 1077 }, { "epoch": 0.05452981941423441, "grad_norm": 14.113748550415039, "learning_rate": 9.95019685486492e-06, "loss": 0.4734, "step": 1078 }, { "epoch": 0.05458040366229956, "grad_norm": 11.699620246887207, "learning_rate": 9.950083790402832e-06, "loss": 0.5252, "step": 1079 }, { "epoch": 0.05463098791036471, "grad_norm": 13.485969543457031, "learning_rate": 9.94997059838904e-06, "loss": 0.559, "step": 1080 }, { "epoch": 0.054681572158429864, "grad_norm": 13.832254409790039, "learning_rate": 9.949857278826463e-06, "loss": 0.535, "step": 1081 }, { "epoch": 0.054732156406495015, "grad_norm": 15.964289665222168, "learning_rate": 9.949743831718019e-06, "loss": 0.5427, "step": 1082 }, { "epoch": 0.05478274065456017, "grad_norm": 13.166296005249023, "learning_rate": 9.949630257066633e-06, "loss": 0.464, "step": 1083 }, { "epoch": 0.054833324902625324, "grad_norm": 9.582018852233887, "learning_rate": 9.949516554875231e-06, "loss": 0.5796, "step": 1084 }, { "epoch": 0.054883909150690476, "grad_norm": 9.10732650756836, "learning_rate": 9.949402725146741e-06, "loss": 0.5107, "step": 1085 }, { "epoch": 0.05493449339875563, "grad_norm": 18.408946990966797, "learning_rate": 9.949288767884098e-06, "loss": 0.5801, "step": 1086 }, { "epoch": 0.05498507764682078, "grad_norm": 11.023999214172363, "learning_rate": 9.949174683090239e-06, "loss": 0.5079, "step": 1087 }, { "epoch": 0.055035661894885936, "grad_norm": 42.768802642822266, "learning_rate": 9.949060470768101e-06, "loss": 0.5131, "step": 1088 }, { "epoch": 0.05508624614295109, "grad_norm": 15.793190002441406, "learning_rate": 9.94894613092063e-06, "loss": 0.595, "step": 1089 }, { "epoch": 0.05513683039101624, "grad_norm": 8.319008827209473, "learning_rate": 9.948831663550771e-06, "loss": 0.347, "step": 1090 }, { "epoch": 0.05518741463908139, "grad_norm": 11.595958709716797, "learning_rate": 9.948717068661473e-06, "loss": 0.4743, "step": 1091 }, { "epoch": 0.05523799888714654, "grad_norm": 13.776129722595215, "learning_rate": 9.948602346255689e-06, "loss": 0.4948, "step": 1092 }, { "epoch": 0.0552885831352117, "grad_norm": 16.618427276611328, "learning_rate": 9.948487496336376e-06, "loss": 0.5245, "step": 1093 }, { "epoch": 0.05533916738327685, "grad_norm": 12.3923921585083, "learning_rate": 9.948372518906493e-06, "loss": 0.5497, "step": 1094 }, { "epoch": 0.055389751631342, "grad_norm": 12.543841361999512, "learning_rate": 9.948257413969003e-06, "loss": 0.45, "step": 1095 }, { "epoch": 0.05544033587940715, "grad_norm": 14.199183464050293, "learning_rate": 9.948142181526869e-06, "loss": 0.4852, "step": 1096 }, { "epoch": 0.0554909201274723, "grad_norm": 16.787912368774414, "learning_rate": 9.948026821583065e-06, "loss": 0.5619, "step": 1097 }, { "epoch": 0.05554150437553746, "grad_norm": 11.45170783996582, "learning_rate": 9.947911334140559e-06, "loss": 0.4686, "step": 1098 }, { "epoch": 0.05559208862360261, "grad_norm": 11.788008689880371, "learning_rate": 9.947795719202331e-06, "loss": 0.5862, "step": 1099 }, { "epoch": 0.05564267287166776, "grad_norm": 6.338517189025879, "learning_rate": 9.947679976771358e-06, "loss": 0.5477, "step": 1100 }, { "epoch": 0.055693257119732914, "grad_norm": 9.336585998535156, "learning_rate": 9.947564106850622e-06, "loss": 0.4587, "step": 1101 }, { "epoch": 0.055743841367798065, "grad_norm": 8.225591659545898, "learning_rate": 9.94744810944311e-06, "loss": 0.543, "step": 1102 }, { "epoch": 0.05579442561586322, "grad_norm": 12.434741973876953, "learning_rate": 9.94733198455181e-06, "loss": 0.5966, "step": 1103 }, { "epoch": 0.055845009863928374, "grad_norm": 10.190448760986328, "learning_rate": 9.947215732179714e-06, "loss": 0.5149, "step": 1104 }, { "epoch": 0.055895594111993525, "grad_norm": 10.420694351196289, "learning_rate": 9.947099352329818e-06, "loss": 0.4501, "step": 1105 }, { "epoch": 0.055946178360058677, "grad_norm": 13.56445598602295, "learning_rate": 9.946982845005122e-06, "loss": 0.4741, "step": 1106 }, { "epoch": 0.05599676260812383, "grad_norm": 15.922347068786621, "learning_rate": 9.946866210208627e-06, "loss": 0.4845, "step": 1107 }, { "epoch": 0.056047346856188986, "grad_norm": 15.246184349060059, "learning_rate": 9.946749447943338e-06, "loss": 0.7475, "step": 1108 }, { "epoch": 0.05609793110425414, "grad_norm": 10.008994102478027, "learning_rate": 9.946632558212264e-06, "loss": 0.4288, "step": 1109 }, { "epoch": 0.05614851535231929, "grad_norm": 12.086153030395508, "learning_rate": 9.946515541018417e-06, "loss": 0.4953, "step": 1110 }, { "epoch": 0.05619909960038444, "grad_norm": 13.177242279052734, "learning_rate": 9.946398396364814e-06, "loss": 0.4385, "step": 1111 }, { "epoch": 0.05624968384844959, "grad_norm": 21.867435455322266, "learning_rate": 9.94628112425447e-06, "loss": 0.6142, "step": 1112 }, { "epoch": 0.05630026809651475, "grad_norm": 9.58033275604248, "learning_rate": 9.946163724690409e-06, "loss": 0.4136, "step": 1113 }, { "epoch": 0.0563508523445799, "grad_norm": 13.42786693572998, "learning_rate": 9.946046197675655e-06, "loss": 0.5065, "step": 1114 }, { "epoch": 0.05640143659264505, "grad_norm": 13.467802047729492, "learning_rate": 9.945928543213238e-06, "loss": 0.5376, "step": 1115 }, { "epoch": 0.0564520208407102, "grad_norm": 17.498594284057617, "learning_rate": 9.945810761306189e-06, "loss": 0.4566, "step": 1116 }, { "epoch": 0.05650260508877535, "grad_norm": 7.00752067565918, "learning_rate": 9.945692851957543e-06, "loss": 0.475, "step": 1117 }, { "epoch": 0.05655318933684051, "grad_norm": 12.759499549865723, "learning_rate": 9.945574815170338e-06, "loss": 0.5703, "step": 1118 }, { "epoch": 0.05660377358490566, "grad_norm": 18.60797691345215, "learning_rate": 9.945456650947615e-06, "loss": 0.6667, "step": 1119 }, { "epoch": 0.05665435783297081, "grad_norm": 11.094538688659668, "learning_rate": 9.94533835929242e-06, "loss": 0.4665, "step": 1120 }, { "epoch": 0.056704942081035964, "grad_norm": 7.671059608459473, "learning_rate": 9.9452199402078e-06, "loss": 0.4969, "step": 1121 }, { "epoch": 0.056755526329101115, "grad_norm": 10.990020751953125, "learning_rate": 9.945101393696806e-06, "loss": 0.428, "step": 1122 }, { "epoch": 0.05680611057716627, "grad_norm": 10.171147346496582, "learning_rate": 9.944982719762494e-06, "loss": 0.467, "step": 1123 }, { "epoch": 0.056856694825231424, "grad_norm": 10.866326332092285, "learning_rate": 9.944863918407923e-06, "loss": 0.5422, "step": 1124 }, { "epoch": 0.056907279073296575, "grad_norm": 12.967378616333008, "learning_rate": 9.94474498963615e-06, "loss": 0.4659, "step": 1125 }, { "epoch": 0.056957863321361726, "grad_norm": 25.663684844970703, "learning_rate": 9.944625933450245e-06, "loss": 0.667, "step": 1126 }, { "epoch": 0.05700844756942688, "grad_norm": 12.892967224121094, "learning_rate": 9.944506749853271e-06, "loss": 0.4974, "step": 1127 }, { "epoch": 0.057059031817492036, "grad_norm": 12.893927574157715, "learning_rate": 9.944387438848302e-06, "loss": 0.52, "step": 1128 }, { "epoch": 0.05710961606555719, "grad_norm": 15.808509826660156, "learning_rate": 9.94426800043841e-06, "loss": 0.5671, "step": 1129 }, { "epoch": 0.05716020031362234, "grad_norm": 13.123173713684082, "learning_rate": 9.944148434626676e-06, "loss": 0.6298, "step": 1130 }, { "epoch": 0.05721078456168749, "grad_norm": 13.7112398147583, "learning_rate": 9.944028741416177e-06, "loss": 0.5425, "step": 1131 }, { "epoch": 0.05726136880975264, "grad_norm": 19.35982322692871, "learning_rate": 9.943908920810001e-06, "loss": 0.6661, "step": 1132 }, { "epoch": 0.0573119530578178, "grad_norm": 11.23141098022461, "learning_rate": 9.943788972811233e-06, "loss": 0.4746, "step": 1133 }, { "epoch": 0.05736253730588295, "grad_norm": 12.572214126586914, "learning_rate": 9.943668897422963e-06, "loss": 0.5133, "step": 1134 }, { "epoch": 0.0574131215539481, "grad_norm": 9.118813514709473, "learning_rate": 9.943548694648288e-06, "loss": 0.4499, "step": 1135 }, { "epoch": 0.05746370580201325, "grad_norm": 13.385607719421387, "learning_rate": 9.943428364490304e-06, "loss": 0.4898, "step": 1136 }, { "epoch": 0.0575142900500784, "grad_norm": 9.13769817352295, "learning_rate": 9.943307906952111e-06, "loss": 0.4984, "step": 1137 }, { "epoch": 0.05756487429814356, "grad_norm": 19.367216110229492, "learning_rate": 9.943187322036813e-06, "loss": 0.6146, "step": 1138 }, { "epoch": 0.05761545854620871, "grad_norm": 13.424570083618164, "learning_rate": 9.943066609747518e-06, "loss": 0.4916, "step": 1139 }, { "epoch": 0.05766604279427386, "grad_norm": 14.334085464477539, "learning_rate": 9.942945770087335e-06, "loss": 0.5451, "step": 1140 }, { "epoch": 0.057716627042339014, "grad_norm": 14.5894136428833, "learning_rate": 9.942824803059379e-06, "loss": 0.4702, "step": 1141 }, { "epoch": 0.057767211290404165, "grad_norm": 10.12897777557373, "learning_rate": 9.942703708666767e-06, "loss": 0.5112, "step": 1142 }, { "epoch": 0.05781779553846932, "grad_norm": 9.102018356323242, "learning_rate": 9.942582486912618e-06, "loss": 0.5333, "step": 1143 }, { "epoch": 0.057868379786534474, "grad_norm": 7.736358165740967, "learning_rate": 9.942461137800058e-06, "loss": 0.4641, "step": 1144 }, { "epoch": 0.057918964034599625, "grad_norm": 14.641006469726562, "learning_rate": 9.942339661332212e-06, "loss": 0.5787, "step": 1145 }, { "epoch": 0.057969548282664776, "grad_norm": 7.778643608093262, "learning_rate": 9.942218057512211e-06, "loss": 0.4995, "step": 1146 }, { "epoch": 0.05802013253072993, "grad_norm": 15.011602401733398, "learning_rate": 9.942096326343189e-06, "loss": 0.5093, "step": 1147 }, { "epoch": 0.058070716778795085, "grad_norm": 10.30348014831543, "learning_rate": 9.941974467828278e-06, "loss": 0.4924, "step": 1148 }, { "epoch": 0.058121301026860236, "grad_norm": 14.220394134521484, "learning_rate": 9.941852481970626e-06, "loss": 0.4578, "step": 1149 }, { "epoch": 0.05817188527492539, "grad_norm": 10.421085357666016, "learning_rate": 9.941730368773372e-06, "loss": 0.4357, "step": 1150 }, { "epoch": 0.05822246952299054, "grad_norm": 32.26780700683594, "learning_rate": 9.94160812823966e-06, "loss": 0.5752, "step": 1151 }, { "epoch": 0.0582730537710557, "grad_norm": 17.379243850708008, "learning_rate": 9.941485760372644e-06, "loss": 0.6416, "step": 1152 }, { "epoch": 0.05832363801912085, "grad_norm": 16.7251033782959, "learning_rate": 9.941363265175475e-06, "loss": 0.5693, "step": 1153 }, { "epoch": 0.058374222267186, "grad_norm": 10.012630462646484, "learning_rate": 9.94124064265131e-06, "loss": 0.5195, "step": 1154 }, { "epoch": 0.05842480651525115, "grad_norm": 12.95511245727539, "learning_rate": 9.94111789280331e-06, "loss": 0.51, "step": 1155 }, { "epoch": 0.0584753907633163, "grad_norm": 9.966464042663574, "learning_rate": 9.940995015634636e-06, "loss": 0.4878, "step": 1156 }, { "epoch": 0.05852597501138146, "grad_norm": 11.835882186889648, "learning_rate": 9.940872011148455e-06, "loss": 0.5058, "step": 1157 }, { "epoch": 0.05857655925944661, "grad_norm": 10.044069290161133, "learning_rate": 9.940748879347936e-06, "loss": 0.4582, "step": 1158 }, { "epoch": 0.05862714350751176, "grad_norm": 15.546363830566406, "learning_rate": 9.940625620236253e-06, "loss": 0.5688, "step": 1159 }, { "epoch": 0.05867772775557691, "grad_norm": 22.438243865966797, "learning_rate": 9.94050223381658e-06, "loss": 0.5417, "step": 1160 }, { "epoch": 0.058728312003642064, "grad_norm": 18.24222183227539, "learning_rate": 9.9403787200921e-06, "loss": 0.5467, "step": 1161 }, { "epoch": 0.05877889625170722, "grad_norm": 10.061891555786133, "learning_rate": 9.94025507906599e-06, "loss": 0.5599, "step": 1162 }, { "epoch": 0.05882948049977237, "grad_norm": 14.156319618225098, "learning_rate": 9.940131310741443e-06, "loss": 0.5333, "step": 1163 }, { "epoch": 0.058880064747837524, "grad_norm": 8.35550308227539, "learning_rate": 9.940007415121642e-06, "loss": 0.4767, "step": 1164 }, { "epoch": 0.058930648995902675, "grad_norm": 19.208330154418945, "learning_rate": 9.939883392209783e-06, "loss": 0.6336, "step": 1165 }, { "epoch": 0.058981233243967826, "grad_norm": 8.58966064453125, "learning_rate": 9.93975924200906e-06, "loss": 0.472, "step": 1166 }, { "epoch": 0.059031817492032984, "grad_norm": 10.530050277709961, "learning_rate": 9.939634964522673e-06, "loss": 0.5348, "step": 1167 }, { "epoch": 0.059082401740098135, "grad_norm": 7.880033016204834, "learning_rate": 9.939510559753825e-06, "loss": 0.4827, "step": 1168 }, { "epoch": 0.059132985988163286, "grad_norm": 7.699892997741699, "learning_rate": 9.939386027705721e-06, "loss": 0.4529, "step": 1169 }, { "epoch": 0.05918357023622844, "grad_norm": 8.400718688964844, "learning_rate": 9.93926136838157e-06, "loss": 0.4877, "step": 1170 }, { "epoch": 0.05923415448429359, "grad_norm": 8.765338897705078, "learning_rate": 9.939136581784581e-06, "loss": 0.4692, "step": 1171 }, { "epoch": 0.05928473873235875, "grad_norm": 8.169957160949707, "learning_rate": 9.939011667917975e-06, "loss": 0.5162, "step": 1172 }, { "epoch": 0.0593353229804239, "grad_norm": 8.430106163024902, "learning_rate": 9.938886626784968e-06, "loss": 0.4813, "step": 1173 }, { "epoch": 0.05938590722848905, "grad_norm": 8.126947402954102, "learning_rate": 9.93876145838878e-06, "loss": 0.5572, "step": 1174 }, { "epoch": 0.0594364914765542, "grad_norm": 17.53326988220215, "learning_rate": 9.938636162732639e-06, "loss": 0.4503, "step": 1175 }, { "epoch": 0.05948707572461935, "grad_norm": 13.437838554382324, "learning_rate": 9.938510739819773e-06, "loss": 0.5003, "step": 1176 }, { "epoch": 0.05953765997268451, "grad_norm": 13.924017906188965, "learning_rate": 9.938385189653415e-06, "loss": 0.5428, "step": 1177 }, { "epoch": 0.05958824422074966, "grad_norm": 9.71427059173584, "learning_rate": 9.938259512236799e-06, "loss": 0.5489, "step": 1178 }, { "epoch": 0.05963882846881481, "grad_norm": 6.711885929107666, "learning_rate": 9.938133707573161e-06, "loss": 0.4796, "step": 1179 }, { "epoch": 0.05968941271687996, "grad_norm": 19.294950485229492, "learning_rate": 9.938007775665745e-06, "loss": 0.6709, "step": 1180 }, { "epoch": 0.05973999696494511, "grad_norm": 22.437284469604492, "learning_rate": 9.9378817165178e-06, "loss": 0.5283, "step": 1181 }, { "epoch": 0.05979058121301027, "grad_norm": 22.318086624145508, "learning_rate": 9.937755530132565e-06, "loss": 0.7069, "step": 1182 }, { "epoch": 0.05984116546107542, "grad_norm": 13.728970527648926, "learning_rate": 9.937629216513297e-06, "loss": 0.5353, "step": 1183 }, { "epoch": 0.059891749709140574, "grad_norm": 11.0049467086792, "learning_rate": 9.937502775663255e-06, "loss": 0.5563, "step": 1184 }, { "epoch": 0.059942333957205725, "grad_norm": 21.42020606994629, "learning_rate": 9.93737620758569e-06, "loss": 0.5212, "step": 1185 }, { "epoch": 0.059992918205270876, "grad_norm": 19.633195877075195, "learning_rate": 9.937249512283864e-06, "loss": 0.5783, "step": 1186 }, { "epoch": 0.060043502453336034, "grad_norm": 9.421856880187988, "learning_rate": 9.937122689761045e-06, "loss": 0.441, "step": 1187 }, { "epoch": 0.060094086701401185, "grad_norm": 14.165992736816406, "learning_rate": 9.9369957400205e-06, "loss": 0.5646, "step": 1188 }, { "epoch": 0.060144670949466336, "grad_norm": 13.513474464416504, "learning_rate": 9.9368686630655e-06, "loss": 0.4812, "step": 1189 }, { "epoch": 0.06019525519753149, "grad_norm": 15.590638160705566, "learning_rate": 9.936741458899316e-06, "loss": 0.4494, "step": 1190 }, { "epoch": 0.06024583944559664, "grad_norm": 13.392678260803223, "learning_rate": 9.936614127525232e-06, "loss": 0.5535, "step": 1191 }, { "epoch": 0.060296423693661796, "grad_norm": 16.616254806518555, "learning_rate": 9.936486668946525e-06, "loss": 0.5105, "step": 1192 }, { "epoch": 0.06034700794172695, "grad_norm": 7.597771167755127, "learning_rate": 9.93635908316648e-06, "loss": 0.4275, "step": 1193 }, { "epoch": 0.0603975921897921, "grad_norm": 9.217510223388672, "learning_rate": 9.936231370188385e-06, "loss": 0.5285, "step": 1194 }, { "epoch": 0.06044817643785725, "grad_norm": 10.04349136352539, "learning_rate": 9.93610353001553e-06, "loss": 0.4804, "step": 1195 }, { "epoch": 0.0604987606859224, "grad_norm": 12.112603187561035, "learning_rate": 9.93597556265121e-06, "loss": 0.5457, "step": 1196 }, { "epoch": 0.06054934493398756, "grad_norm": 10.869625091552734, "learning_rate": 9.935847468098722e-06, "loss": 0.4695, "step": 1197 }, { "epoch": 0.06059992918205271, "grad_norm": 16.88268280029297, "learning_rate": 9.935719246361366e-06, "loss": 0.5692, "step": 1198 }, { "epoch": 0.06065051343011786, "grad_norm": 12.36493968963623, "learning_rate": 9.935590897442449e-06, "loss": 0.5046, "step": 1199 }, { "epoch": 0.06070109767818301, "grad_norm": 6.703972816467285, "learning_rate": 9.935462421345275e-06, "loss": 0.3873, "step": 1200 }, { "epoch": 0.06075168192624816, "grad_norm": 5.217097282409668, "learning_rate": 9.935333818073153e-06, "loss": 0.331, "step": 1201 }, { "epoch": 0.06080226617431332, "grad_norm": 19.79229736328125, "learning_rate": 9.935205087629402e-06, "loss": 0.7438, "step": 1202 }, { "epoch": 0.06085285042237847, "grad_norm": 10.796804428100586, "learning_rate": 9.935076230017338e-06, "loss": 0.5358, "step": 1203 }, { "epoch": 0.060903434670443624, "grad_norm": 17.936893463134766, "learning_rate": 9.934947245240276e-06, "loss": 0.4836, "step": 1204 }, { "epoch": 0.060954018918508775, "grad_norm": 26.143234252929688, "learning_rate": 9.934818133301545e-06, "loss": 0.6968, "step": 1205 }, { "epoch": 0.061004603166573926, "grad_norm": 14.274460792541504, "learning_rate": 9.93468889420447e-06, "loss": 0.5915, "step": 1206 }, { "epoch": 0.061055187414639084, "grad_norm": 8.3154296875, "learning_rate": 9.934559527952383e-06, "loss": 0.3991, "step": 1207 }, { "epoch": 0.061105771662704235, "grad_norm": 15.38583755493164, "learning_rate": 9.934430034548615e-06, "loss": 0.6115, "step": 1208 }, { "epoch": 0.061156355910769386, "grad_norm": 9.947072982788086, "learning_rate": 9.934300413996503e-06, "loss": 0.4039, "step": 1209 }, { "epoch": 0.06120694015883454, "grad_norm": 14.858601570129395, "learning_rate": 9.934170666299388e-06, "loss": 0.5941, "step": 1210 }, { "epoch": 0.06125752440689969, "grad_norm": 9.98437786102295, "learning_rate": 9.934040791460613e-06, "loss": 0.4658, "step": 1211 }, { "epoch": 0.061308108654964846, "grad_norm": 6.940086841583252, "learning_rate": 9.933910789483524e-06, "loss": 0.5473, "step": 1212 }, { "epoch": 0.06135869290303, "grad_norm": 13.527108192443848, "learning_rate": 9.93378066037147e-06, "loss": 0.5641, "step": 1213 }, { "epoch": 0.06140927715109515, "grad_norm": 14.218056678771973, "learning_rate": 9.933650404127807e-06, "loss": 0.5137, "step": 1214 }, { "epoch": 0.0614598613991603, "grad_norm": 6.9768829345703125, "learning_rate": 9.93352002075589e-06, "loss": 0.3305, "step": 1215 }, { "epoch": 0.06151044564722545, "grad_norm": 19.77399253845215, "learning_rate": 9.93338951025908e-06, "loss": 0.6367, "step": 1216 }, { "epoch": 0.06156102989529061, "grad_norm": 9.007976531982422, "learning_rate": 9.933258872640736e-06, "loss": 0.4834, "step": 1217 }, { "epoch": 0.06161161414335576, "grad_norm": 8.714609146118164, "learning_rate": 9.933128107904228e-06, "loss": 0.4984, "step": 1218 }, { "epoch": 0.06166219839142091, "grad_norm": 16.88357162475586, "learning_rate": 9.932997216052924e-06, "loss": 0.488, "step": 1219 }, { "epoch": 0.06171278263948606, "grad_norm": 6.175335884094238, "learning_rate": 9.932866197090196e-06, "loss": 0.545, "step": 1220 }, { "epoch": 0.06176336688755121, "grad_norm": 7.473507404327393, "learning_rate": 9.932735051019422e-06, "loss": 0.5084, "step": 1221 }, { "epoch": 0.06181395113561637, "grad_norm": 7.9406046867370605, "learning_rate": 9.93260377784398e-06, "loss": 0.5243, "step": 1222 }, { "epoch": 0.06186453538368152, "grad_norm": 7.472501277923584, "learning_rate": 9.932472377567252e-06, "loss": 0.4722, "step": 1223 }, { "epoch": 0.06191511963174667, "grad_norm": 6.4848151206970215, "learning_rate": 9.932340850192625e-06, "loss": 0.3839, "step": 1224 }, { "epoch": 0.061965703879811825, "grad_norm": 14.325953483581543, "learning_rate": 9.932209195723487e-06, "loss": 0.4673, "step": 1225 }, { "epoch": 0.062016288127876976, "grad_norm": 12.145095825195312, "learning_rate": 9.932077414163234e-06, "loss": 0.5116, "step": 1226 }, { "epoch": 0.062066872375942134, "grad_norm": 11.28935432434082, "learning_rate": 9.931945505515257e-06, "loss": 0.4699, "step": 1227 }, { "epoch": 0.062117456624007285, "grad_norm": 17.59033966064453, "learning_rate": 9.931813469782957e-06, "loss": 0.5479, "step": 1228 }, { "epoch": 0.062168040872072436, "grad_norm": 20.31255340576172, "learning_rate": 9.931681306969737e-06, "loss": 0.6521, "step": 1229 }, { "epoch": 0.06221862512013759, "grad_norm": 7.303000450134277, "learning_rate": 9.931549017079002e-06, "loss": 0.5045, "step": 1230 }, { "epoch": 0.062269209368202745, "grad_norm": 11.405509948730469, "learning_rate": 9.93141660011416e-06, "loss": 0.4795, "step": 1231 }, { "epoch": 0.062319793616267896, "grad_norm": 16.224124908447266, "learning_rate": 9.931284056078623e-06, "loss": 0.6304, "step": 1232 }, { "epoch": 0.06237037786433305, "grad_norm": 9.308422088623047, "learning_rate": 9.931151384975807e-06, "loss": 0.4919, "step": 1233 }, { "epoch": 0.0624209621123982, "grad_norm": 18.375354766845703, "learning_rate": 9.931018586809131e-06, "loss": 0.556, "step": 1234 }, { "epoch": 0.06247154636046335, "grad_norm": 13.283058166503906, "learning_rate": 9.930885661582016e-06, "loss": 0.508, "step": 1235 }, { "epoch": 0.0625221306085285, "grad_norm": 12.844522476196289, "learning_rate": 9.930752609297889e-06, "loss": 0.5289, "step": 1236 }, { "epoch": 0.06257271485659366, "grad_norm": 20.363203048706055, "learning_rate": 9.930619429960175e-06, "loss": 0.4704, "step": 1237 }, { "epoch": 0.0626232991046588, "grad_norm": 7.303799152374268, "learning_rate": 9.93048612357231e-06, "loss": 0.4558, "step": 1238 }, { "epoch": 0.06267388335272396, "grad_norm": 13.163534164428711, "learning_rate": 9.930352690137723e-06, "loss": 0.6163, "step": 1239 }, { "epoch": 0.06272446760078912, "grad_norm": 7.028366565704346, "learning_rate": 9.930219129659859e-06, "loss": 0.4609, "step": 1240 }, { "epoch": 0.06277505184885426, "grad_norm": 11.2509765625, "learning_rate": 9.930085442142156e-06, "loss": 0.427, "step": 1241 }, { "epoch": 0.06282563609691942, "grad_norm": 9.93203353881836, "learning_rate": 9.92995162758806e-06, "loss": 0.5306, "step": 1242 }, { "epoch": 0.06287622034498457, "grad_norm": 11.509961128234863, "learning_rate": 9.929817686001016e-06, "loss": 0.5637, "step": 1243 }, { "epoch": 0.06292680459304972, "grad_norm": 13.3330717086792, "learning_rate": 9.92968361738448e-06, "loss": 0.6585, "step": 1244 }, { "epoch": 0.06297738884111488, "grad_norm": 9.757505416870117, "learning_rate": 9.929549421741902e-06, "loss": 0.4383, "step": 1245 }, { "epoch": 0.06302797308918003, "grad_norm": 15.093791961669922, "learning_rate": 9.929415099076745e-06, "loss": 0.5904, "step": 1246 }, { "epoch": 0.06307855733724518, "grad_norm": 13.021904945373535, "learning_rate": 9.929280649392465e-06, "loss": 0.4872, "step": 1247 }, { "epoch": 0.06312914158531033, "grad_norm": 10.85635757446289, "learning_rate": 9.92914607269253e-06, "loss": 0.4791, "step": 1248 }, { "epoch": 0.06317972583337549, "grad_norm": 13.292964935302734, "learning_rate": 9.929011368980406e-06, "loss": 0.5491, "step": 1249 }, { "epoch": 0.06323031008144064, "grad_norm": 10.687962532043457, "learning_rate": 9.928876538259565e-06, "loss": 0.5098, "step": 1250 }, { "epoch": 0.06328089432950579, "grad_norm": 10.947681427001953, "learning_rate": 9.92874158053348e-06, "loss": 0.5063, "step": 1251 }, { "epoch": 0.06333147857757095, "grad_norm": 16.368881225585938, "learning_rate": 9.928606495805631e-06, "loss": 0.5467, "step": 1252 }, { "epoch": 0.06338206282563609, "grad_norm": 9.509716033935547, "learning_rate": 9.928471284079498e-06, "loss": 0.4899, "step": 1253 }, { "epoch": 0.06343264707370125, "grad_norm": 13.662806510925293, "learning_rate": 9.928335945358561e-06, "loss": 0.4815, "step": 1254 }, { "epoch": 0.0634832313217664, "grad_norm": 12.122023582458496, "learning_rate": 9.928200479646312e-06, "loss": 0.5364, "step": 1255 }, { "epoch": 0.06353381556983155, "grad_norm": 11.734512329101562, "learning_rate": 9.92806488694624e-06, "loss": 0.5335, "step": 1256 }, { "epoch": 0.06358439981789671, "grad_norm": 11.576454162597656, "learning_rate": 9.92792916726184e-06, "loss": 0.5005, "step": 1257 }, { "epoch": 0.06363498406596185, "grad_norm": 15.20301628112793, "learning_rate": 9.927793320596608e-06, "loss": 0.6398, "step": 1258 }, { "epoch": 0.06368556831402701, "grad_norm": 11.115067481994629, "learning_rate": 9.927657346954045e-06, "loss": 0.4846, "step": 1259 }, { "epoch": 0.06373615256209217, "grad_norm": 13.102632522583008, "learning_rate": 9.927521246337654e-06, "loss": 0.5039, "step": 1260 }, { "epoch": 0.06378673681015731, "grad_norm": 8.081887245178223, "learning_rate": 9.927385018750943e-06, "loss": 0.4764, "step": 1261 }, { "epoch": 0.06383732105822247, "grad_norm": 8.780571937561035, "learning_rate": 9.92724866419742e-06, "loss": 0.4239, "step": 1262 }, { "epoch": 0.06388790530628762, "grad_norm": 6.305699825286865, "learning_rate": 9.927112182680602e-06, "loss": 0.4934, "step": 1263 }, { "epoch": 0.06393848955435277, "grad_norm": 9.154419898986816, "learning_rate": 9.926975574204002e-06, "loss": 0.5612, "step": 1264 }, { "epoch": 0.06398907380241793, "grad_norm": 11.191035270690918, "learning_rate": 9.926838838771144e-06, "loss": 0.4245, "step": 1265 }, { "epoch": 0.06403965805048308, "grad_norm": 10.640531539916992, "learning_rate": 9.926701976385548e-06, "loss": 0.3657, "step": 1266 }, { "epoch": 0.06409024229854823, "grad_norm": 16.78085708618164, "learning_rate": 9.926564987050743e-06, "loss": 0.641, "step": 1267 }, { "epoch": 0.06414082654661339, "grad_norm": 12.091856956481934, "learning_rate": 9.926427870770259e-06, "loss": 0.5476, "step": 1268 }, { "epoch": 0.06419141079467854, "grad_norm": 8.969335556030273, "learning_rate": 9.926290627547626e-06, "loss": 0.47, "step": 1269 }, { "epoch": 0.0642419950427437, "grad_norm": 8.487530708312988, "learning_rate": 9.926153257386383e-06, "loss": 0.4897, "step": 1270 }, { "epoch": 0.06429257929080884, "grad_norm": 11.655051231384277, "learning_rate": 9.92601576029007e-06, "loss": 0.5143, "step": 1271 }, { "epoch": 0.064343163538874, "grad_norm": 12.03453254699707, "learning_rate": 9.92587813626223e-06, "loss": 0.4301, "step": 1272 }, { "epoch": 0.06439374778693915, "grad_norm": 7.2350568771362305, "learning_rate": 9.925740385306406e-06, "loss": 0.4864, "step": 1273 }, { "epoch": 0.0644443320350043, "grad_norm": 10.796408653259277, "learning_rate": 9.92560250742615e-06, "loss": 0.5121, "step": 1274 }, { "epoch": 0.06449491628306946, "grad_norm": 16.980344772338867, "learning_rate": 9.925464502625017e-06, "loss": 0.5051, "step": 1275 }, { "epoch": 0.0645455005311346, "grad_norm": 11.525541305541992, "learning_rate": 9.925326370906558e-06, "loss": 0.6775, "step": 1276 }, { "epoch": 0.06459608477919976, "grad_norm": 7.40805196762085, "learning_rate": 9.925188112274337e-06, "loss": 0.4183, "step": 1277 }, { "epoch": 0.06464666902726492, "grad_norm": 10.669075012207031, "learning_rate": 9.925049726731915e-06, "loss": 0.4126, "step": 1278 }, { "epoch": 0.06469725327533006, "grad_norm": 7.101309299468994, "learning_rate": 9.924911214282856e-06, "loss": 0.5289, "step": 1279 }, { "epoch": 0.06474783752339522, "grad_norm": 9.376482009887695, "learning_rate": 9.924772574930732e-06, "loss": 0.5581, "step": 1280 }, { "epoch": 0.06479842177146036, "grad_norm": 14.156128883361816, "learning_rate": 9.924633808679113e-06, "loss": 0.4749, "step": 1281 }, { "epoch": 0.06484900601952552, "grad_norm": 8.66835880279541, "learning_rate": 9.924494915531575e-06, "loss": 0.5273, "step": 1282 }, { "epoch": 0.06489959026759068, "grad_norm": 13.569887161254883, "learning_rate": 9.9243558954917e-06, "loss": 0.45, "step": 1283 }, { "epoch": 0.06495017451565582, "grad_norm": 14.307456970214844, "learning_rate": 9.924216748563066e-06, "loss": 0.6234, "step": 1284 }, { "epoch": 0.06500075876372098, "grad_norm": 7.469395637512207, "learning_rate": 9.92407747474926e-06, "loss": 0.3777, "step": 1285 }, { "epoch": 0.06505134301178613, "grad_norm": 10.43185806274414, "learning_rate": 9.923938074053872e-06, "loss": 0.4449, "step": 1286 }, { "epoch": 0.06510192725985128, "grad_norm": 5.970028877258301, "learning_rate": 9.923798546480495e-06, "loss": 0.412, "step": 1287 }, { "epoch": 0.06515251150791644, "grad_norm": 7.398985385894775, "learning_rate": 9.923658892032719e-06, "loss": 0.4666, "step": 1288 }, { "epoch": 0.06520309575598159, "grad_norm": 9.149250984191895, "learning_rate": 9.923519110714148e-06, "loss": 0.3997, "step": 1289 }, { "epoch": 0.06525368000404674, "grad_norm": 14.66727066040039, "learning_rate": 9.923379202528381e-06, "loss": 0.5683, "step": 1290 }, { "epoch": 0.06530426425211189, "grad_norm": 14.356999397277832, "learning_rate": 9.923239167479024e-06, "loss": 0.6231, "step": 1291 }, { "epoch": 0.06535484850017705, "grad_norm": 14.419325828552246, "learning_rate": 9.923099005569686e-06, "loss": 0.568, "step": 1292 }, { "epoch": 0.0654054327482422, "grad_norm": 7.207144737243652, "learning_rate": 9.922958716803976e-06, "loss": 0.4421, "step": 1293 }, { "epoch": 0.06545601699630735, "grad_norm": 10.056734085083008, "learning_rate": 9.922818301185512e-06, "loss": 0.5312, "step": 1294 }, { "epoch": 0.0655066012443725, "grad_norm": 7.793803691864014, "learning_rate": 9.922677758717912e-06, "loss": 0.3585, "step": 1295 }, { "epoch": 0.06555718549243765, "grad_norm": 13.889998435974121, "learning_rate": 9.922537089404795e-06, "loss": 0.527, "step": 1296 }, { "epoch": 0.06560776974050281, "grad_norm": 8.90810775756836, "learning_rate": 9.922396293249787e-06, "loss": 0.4362, "step": 1297 }, { "epoch": 0.06565835398856797, "grad_norm": 11.683246612548828, "learning_rate": 9.922255370256518e-06, "loss": 0.522, "step": 1298 }, { "epoch": 0.06570893823663311, "grad_norm": 18.20769691467285, "learning_rate": 9.922114320428614e-06, "loss": 0.6048, "step": 1299 }, { "epoch": 0.06575952248469827, "grad_norm": 10.071601867675781, "learning_rate": 9.921973143769716e-06, "loss": 0.4782, "step": 1300 }, { "epoch": 0.06581010673276341, "grad_norm": 13.351356506347656, "learning_rate": 9.921831840283459e-06, "loss": 0.5404, "step": 1301 }, { "epoch": 0.06586069098082857, "grad_norm": 11.307851791381836, "learning_rate": 9.921690409973482e-06, "loss": 0.5127, "step": 1302 }, { "epoch": 0.06591127522889373, "grad_norm": 9.486740112304688, "learning_rate": 9.92154885284343e-06, "loss": 0.4518, "step": 1303 }, { "epoch": 0.06596185947695887, "grad_norm": 6.878830909729004, "learning_rate": 9.921407168896952e-06, "loss": 0.3759, "step": 1304 }, { "epoch": 0.06601244372502403, "grad_norm": 8.90121078491211, "learning_rate": 9.921265358137699e-06, "loss": 0.4435, "step": 1305 }, { "epoch": 0.06606302797308918, "grad_norm": 12.92795181274414, "learning_rate": 9.921123420569326e-06, "loss": 0.5629, "step": 1306 }, { "epoch": 0.06611361222115433, "grad_norm": 13.496315002441406, "learning_rate": 9.920981356195488e-06, "loss": 0.6296, "step": 1307 }, { "epoch": 0.06616419646921949, "grad_norm": 8.538675308227539, "learning_rate": 9.920839165019845e-06, "loss": 0.5446, "step": 1308 }, { "epoch": 0.06621478071728464, "grad_norm": 19.590734481811523, "learning_rate": 9.920696847046063e-06, "loss": 0.5975, "step": 1309 }, { "epoch": 0.0662653649653498, "grad_norm": 22.223051071166992, "learning_rate": 9.92055440227781e-06, "loss": 0.6274, "step": 1310 }, { "epoch": 0.06631594921341494, "grad_norm": 12.954475402832031, "learning_rate": 9.920411830718755e-06, "loss": 0.4793, "step": 1311 }, { "epoch": 0.0663665334614801, "grad_norm": 19.63374137878418, "learning_rate": 9.920269132372572e-06, "loss": 0.4919, "step": 1312 }, { "epoch": 0.06641711770954525, "grad_norm": 12.276275634765625, "learning_rate": 9.920126307242936e-06, "loss": 0.5172, "step": 1313 }, { "epoch": 0.0664677019576104, "grad_norm": 10.220870971679688, "learning_rate": 9.919983355333532e-06, "loss": 0.4752, "step": 1314 }, { "epoch": 0.06651828620567556, "grad_norm": 9.86029052734375, "learning_rate": 9.919840276648039e-06, "loss": 0.5082, "step": 1315 }, { "epoch": 0.0665688704537407, "grad_norm": 9.013038635253906, "learning_rate": 9.919697071190144e-06, "loss": 0.5596, "step": 1316 }, { "epoch": 0.06661945470180586, "grad_norm": 10.528997421264648, "learning_rate": 9.919553738963541e-06, "loss": 0.4333, "step": 1317 }, { "epoch": 0.06667003894987102, "grad_norm": 7.469677448272705, "learning_rate": 9.919410279971919e-06, "loss": 0.5306, "step": 1318 }, { "epoch": 0.06672062319793616, "grad_norm": 7.609323024749756, "learning_rate": 9.919266694218978e-06, "loss": 0.4303, "step": 1319 }, { "epoch": 0.06677120744600132, "grad_norm": 11.729119300842285, "learning_rate": 9.919122981708416e-06, "loss": 0.5601, "step": 1320 }, { "epoch": 0.06682179169406646, "grad_norm": 13.395929336547852, "learning_rate": 9.918979142443937e-06, "loss": 0.4387, "step": 1321 }, { "epoch": 0.06687237594213162, "grad_norm": 22.017898559570312, "learning_rate": 9.918835176429244e-06, "loss": 0.4423, "step": 1322 }, { "epoch": 0.06692296019019678, "grad_norm": 16.82500457763672, "learning_rate": 9.918691083668054e-06, "loss": 0.528, "step": 1323 }, { "epoch": 0.06697354443826192, "grad_norm": 8.312012672424316, "learning_rate": 9.918546864164071e-06, "loss": 0.4198, "step": 1324 }, { "epoch": 0.06702412868632708, "grad_norm": 11.09088134765625, "learning_rate": 9.918402517921018e-06, "loss": 0.4532, "step": 1325 }, { "epoch": 0.06707471293439222, "grad_norm": 8.911177635192871, "learning_rate": 9.91825804494261e-06, "loss": 0.3105, "step": 1326 }, { "epoch": 0.06712529718245738, "grad_norm": 15.425115585327148, "learning_rate": 9.918113445232574e-06, "loss": 0.4672, "step": 1327 }, { "epoch": 0.06717588143052254, "grad_norm": 14.50921630859375, "learning_rate": 9.917968718794632e-06, "loss": 0.5677, "step": 1328 }, { "epoch": 0.06722646567858769, "grad_norm": 32.957862854003906, "learning_rate": 9.917823865632517e-06, "loss": 0.6038, "step": 1329 }, { "epoch": 0.06727704992665284, "grad_norm": 13.366103172302246, "learning_rate": 9.917678885749957e-06, "loss": 0.5258, "step": 1330 }, { "epoch": 0.06732763417471799, "grad_norm": 15.953904151916504, "learning_rate": 9.917533779150692e-06, "loss": 0.5921, "step": 1331 }, { "epoch": 0.06737821842278315, "grad_norm": 6.246720314025879, "learning_rate": 9.91738854583846e-06, "loss": 0.4269, "step": 1332 }, { "epoch": 0.0674288026708483, "grad_norm": 11.570836067199707, "learning_rate": 9.917243185817e-06, "loss": 0.4214, "step": 1333 }, { "epoch": 0.06747938691891345, "grad_norm": 10.993982315063477, "learning_rate": 9.917097699090063e-06, "loss": 0.4928, "step": 1334 }, { "epoch": 0.0675299711669786, "grad_norm": 32.02973937988281, "learning_rate": 9.916952085661393e-06, "loss": 0.734, "step": 1335 }, { "epoch": 0.06758055541504375, "grad_norm": 13.04419231414795, "learning_rate": 9.916806345534746e-06, "loss": 0.49, "step": 1336 }, { "epoch": 0.06763113966310891, "grad_norm": 22.90138053894043, "learning_rate": 9.916660478713874e-06, "loss": 0.4431, "step": 1337 }, { "epoch": 0.06768172391117407, "grad_norm": 27.0816707611084, "learning_rate": 9.916514485202538e-06, "loss": 0.6742, "step": 1338 }, { "epoch": 0.06773230815923921, "grad_norm": 17.725521087646484, "learning_rate": 9.916368365004498e-06, "loss": 0.4411, "step": 1339 }, { "epoch": 0.06778289240730437, "grad_norm": 9.103404998779297, "learning_rate": 9.916222118123522e-06, "loss": 0.5078, "step": 1340 }, { "epoch": 0.06783347665536951, "grad_norm": 12.48015308380127, "learning_rate": 9.916075744563377e-06, "loss": 0.4974, "step": 1341 }, { "epoch": 0.06788406090343467, "grad_norm": 13.418211936950684, "learning_rate": 9.915929244327831e-06, "loss": 0.4641, "step": 1342 }, { "epoch": 0.06793464515149983, "grad_norm": 25.801376342773438, "learning_rate": 9.915782617420664e-06, "loss": 0.6155, "step": 1343 }, { "epoch": 0.06798522939956497, "grad_norm": 8.693182945251465, "learning_rate": 9.915635863845655e-06, "loss": 0.5621, "step": 1344 }, { "epoch": 0.06803581364763013, "grad_norm": 14.271254539489746, "learning_rate": 9.915488983606579e-06, "loss": 0.539, "step": 1345 }, { "epoch": 0.06808639789569527, "grad_norm": 26.20655632019043, "learning_rate": 9.915341976707227e-06, "loss": 0.6047, "step": 1346 }, { "epoch": 0.06813698214376043, "grad_norm": 7.909338474273682, "learning_rate": 9.915194843151385e-06, "loss": 0.4545, "step": 1347 }, { "epoch": 0.06818756639182559, "grad_norm": 14.144987106323242, "learning_rate": 9.915047582942842e-06, "loss": 0.4756, "step": 1348 }, { "epoch": 0.06823815063989074, "grad_norm": 22.08143424987793, "learning_rate": 9.914900196085397e-06, "loss": 0.5938, "step": 1349 }, { "epoch": 0.0682887348879559, "grad_norm": 23.354598999023438, "learning_rate": 9.914752682582845e-06, "loss": 0.6363, "step": 1350 }, { "epoch": 0.06833931913602104, "grad_norm": 14.408279418945312, "learning_rate": 9.914605042438987e-06, "loss": 0.6055, "step": 1351 }, { "epoch": 0.0683899033840862, "grad_norm": 7.651090145111084, "learning_rate": 9.914457275657627e-06, "loss": 0.44, "step": 1352 }, { "epoch": 0.06844048763215135, "grad_norm": 10.840312004089355, "learning_rate": 9.914309382242575e-06, "loss": 0.5343, "step": 1353 }, { "epoch": 0.0684910718802165, "grad_norm": 29.024757385253906, "learning_rate": 9.914161362197638e-06, "loss": 0.5021, "step": 1354 }, { "epoch": 0.06854165612828166, "grad_norm": 12.185113906860352, "learning_rate": 9.914013215526635e-06, "loss": 0.5751, "step": 1355 }, { "epoch": 0.0685922403763468, "grad_norm": 19.779468536376953, "learning_rate": 9.91386494223338e-06, "loss": 0.4388, "step": 1356 }, { "epoch": 0.06864282462441196, "grad_norm": 8.077066421508789, "learning_rate": 9.913716542321694e-06, "loss": 0.4562, "step": 1357 }, { "epoch": 0.06869340887247712, "grad_norm": 18.200664520263672, "learning_rate": 9.913568015795401e-06, "loss": 0.4827, "step": 1358 }, { "epoch": 0.06874399312054226, "grad_norm": 17.855709075927734, "learning_rate": 9.913419362658328e-06, "loss": 0.5331, "step": 1359 }, { "epoch": 0.06879457736860742, "grad_norm": 11.211417198181152, "learning_rate": 9.913270582914307e-06, "loss": 0.4236, "step": 1360 }, { "epoch": 0.06884516161667256, "grad_norm": 8.320298194885254, "learning_rate": 9.91312167656717e-06, "loss": 0.5188, "step": 1361 }, { "epoch": 0.06889574586473772, "grad_norm": 12.383490562438965, "learning_rate": 9.912972643620755e-06, "loss": 0.4161, "step": 1362 }, { "epoch": 0.06894633011280288, "grad_norm": 17.84209632873535, "learning_rate": 9.912823484078901e-06, "loss": 0.5929, "step": 1363 }, { "epoch": 0.06899691436086802, "grad_norm": 18.026926040649414, "learning_rate": 9.912674197945453e-06, "loss": 0.5741, "step": 1364 }, { "epoch": 0.06904749860893318, "grad_norm": 26.226125717163086, "learning_rate": 9.912524785224256e-06, "loss": 0.6815, "step": 1365 }, { "epoch": 0.06909808285699832, "grad_norm": 9.32882308959961, "learning_rate": 9.912375245919164e-06, "loss": 0.5217, "step": 1366 }, { "epoch": 0.06914866710506348, "grad_norm": 13.712133407592773, "learning_rate": 9.912225580034023e-06, "loss": 0.5446, "step": 1367 }, { "epoch": 0.06919925135312864, "grad_norm": 8.505687713623047, "learning_rate": 9.912075787572697e-06, "loss": 0.4583, "step": 1368 }, { "epoch": 0.06924983560119378, "grad_norm": 9.327201843261719, "learning_rate": 9.91192586853904e-06, "loss": 0.5027, "step": 1369 }, { "epoch": 0.06930041984925894, "grad_norm": 10.104897499084473, "learning_rate": 9.91177582293692e-06, "loss": 0.542, "step": 1370 }, { "epoch": 0.06935100409732409, "grad_norm": 14.358742713928223, "learning_rate": 9.911625650770201e-06, "loss": 0.5876, "step": 1371 }, { "epoch": 0.06940158834538925, "grad_norm": 17.77823257446289, "learning_rate": 9.911475352042751e-06, "loss": 0.5714, "step": 1372 }, { "epoch": 0.0694521725934544, "grad_norm": 13.079364776611328, "learning_rate": 9.911324926758446e-06, "loss": 0.4702, "step": 1373 }, { "epoch": 0.06950275684151955, "grad_norm": 10.523650169372559, "learning_rate": 9.911174374921158e-06, "loss": 0.4107, "step": 1374 }, { "epoch": 0.0695533410895847, "grad_norm": 9.73084831237793, "learning_rate": 9.91102369653477e-06, "loss": 0.4926, "step": 1375 }, { "epoch": 0.06960392533764985, "grad_norm": 12.150979042053223, "learning_rate": 9.910872891603164e-06, "loss": 0.5459, "step": 1376 }, { "epoch": 0.06965450958571501, "grad_norm": 15.030858039855957, "learning_rate": 9.910721960130225e-06, "loss": 0.4524, "step": 1377 }, { "epoch": 0.06970509383378017, "grad_norm": 9.687091827392578, "learning_rate": 9.910570902119843e-06, "loss": 0.4806, "step": 1378 }, { "epoch": 0.06975567808184531, "grad_norm": 10.982922554016113, "learning_rate": 9.910419717575907e-06, "loss": 0.4402, "step": 1379 }, { "epoch": 0.06980626232991047, "grad_norm": 17.964031219482422, "learning_rate": 9.910268406502319e-06, "loss": 0.56, "step": 1380 }, { "epoch": 0.06985684657797561, "grad_norm": 10.552005767822266, "learning_rate": 9.910116968902974e-06, "loss": 0.4448, "step": 1381 }, { "epoch": 0.06990743082604077, "grad_norm": 12.835192680358887, "learning_rate": 9.909965404781773e-06, "loss": 0.5525, "step": 1382 }, { "epoch": 0.06995801507410593, "grad_norm": 14.9146728515625, "learning_rate": 9.909813714142624e-06, "loss": 0.5101, "step": 1383 }, { "epoch": 0.07000859932217107, "grad_norm": 9.188117027282715, "learning_rate": 9.909661896989434e-06, "loss": 0.4684, "step": 1384 }, { "epoch": 0.07005918357023623, "grad_norm": 13.567268371582031, "learning_rate": 9.909509953326118e-06, "loss": 0.4733, "step": 1385 }, { "epoch": 0.07010976781830137, "grad_norm": 8.50709342956543, "learning_rate": 9.909357883156586e-06, "loss": 0.4659, "step": 1386 }, { "epoch": 0.07016035206636653, "grad_norm": 13.906353950500488, "learning_rate": 9.90920568648476e-06, "loss": 0.5917, "step": 1387 }, { "epoch": 0.07021093631443169, "grad_norm": 7.364969730377197, "learning_rate": 9.909053363314562e-06, "loss": 0.4276, "step": 1388 }, { "epoch": 0.07026152056249683, "grad_norm": 9.005088806152344, "learning_rate": 9.908900913649918e-06, "loss": 0.4368, "step": 1389 }, { "epoch": 0.07031210481056199, "grad_norm": 14.917234420776367, "learning_rate": 9.908748337494752e-06, "loss": 0.5805, "step": 1390 }, { "epoch": 0.07036268905862714, "grad_norm": 10.650565147399902, "learning_rate": 9.908595634852998e-06, "loss": 0.4074, "step": 1391 }, { "epoch": 0.0704132733066923, "grad_norm": 16.83431053161621, "learning_rate": 9.908442805728592e-06, "loss": 0.6793, "step": 1392 }, { "epoch": 0.07046385755475745, "grad_norm": 11.939702987670898, "learning_rate": 9.908289850125467e-06, "loss": 0.4711, "step": 1393 }, { "epoch": 0.0705144418028226, "grad_norm": 10.160499572753906, "learning_rate": 9.908136768047572e-06, "loss": 0.5856, "step": 1394 }, { "epoch": 0.07056502605088776, "grad_norm": 11.265661239624023, "learning_rate": 9.907983559498846e-06, "loss": 0.4757, "step": 1395 }, { "epoch": 0.0706156102989529, "grad_norm": 10.07454776763916, "learning_rate": 9.907830224483239e-06, "loss": 0.5281, "step": 1396 }, { "epoch": 0.07066619454701806, "grad_norm": 11.038941383361816, "learning_rate": 9.907676763004702e-06, "loss": 0.4166, "step": 1397 }, { "epoch": 0.07071677879508322, "grad_norm": 11.843646049499512, "learning_rate": 9.907523175067188e-06, "loss": 0.531, "step": 1398 }, { "epoch": 0.07076736304314836, "grad_norm": 9.646636009216309, "learning_rate": 9.907369460674657e-06, "loss": 0.5402, "step": 1399 }, { "epoch": 0.07081794729121352, "grad_norm": 15.14129638671875, "learning_rate": 9.907215619831065e-06, "loss": 0.5028, "step": 1400 }, { "epoch": 0.07086853153927866, "grad_norm": 14.454742431640625, "learning_rate": 9.907061652540381e-06, "loss": 0.5415, "step": 1401 }, { "epoch": 0.07091911578734382, "grad_norm": 13.973467826843262, "learning_rate": 9.906907558806572e-06, "loss": 0.5322, "step": 1402 }, { "epoch": 0.07096970003540898, "grad_norm": 10.822958946228027, "learning_rate": 9.906753338633606e-06, "loss": 0.5355, "step": 1403 }, { "epoch": 0.07102028428347412, "grad_norm": 13.027571678161621, "learning_rate": 9.90659899202546e-06, "loss": 0.5423, "step": 1404 }, { "epoch": 0.07107086853153928, "grad_norm": 15.324679374694824, "learning_rate": 9.906444518986107e-06, "loss": 0.5671, "step": 1405 }, { "epoch": 0.07112145277960442, "grad_norm": 10.383803367614746, "learning_rate": 9.90628991951953e-06, "loss": 0.5172, "step": 1406 }, { "epoch": 0.07117203702766958, "grad_norm": 16.32122230529785, "learning_rate": 9.906135193629713e-06, "loss": 0.4473, "step": 1407 }, { "epoch": 0.07122262127573474, "grad_norm": 11.705016136169434, "learning_rate": 9.905980341320644e-06, "loss": 0.517, "step": 1408 }, { "epoch": 0.07127320552379988, "grad_norm": 9.235435485839844, "learning_rate": 9.90582536259631e-06, "loss": 0.5197, "step": 1409 }, { "epoch": 0.07132378977186504, "grad_norm": 7.839149475097656, "learning_rate": 9.905670257460706e-06, "loss": 0.4425, "step": 1410 }, { "epoch": 0.07137437401993019, "grad_norm": 18.763423919677734, "learning_rate": 9.905515025917827e-06, "loss": 0.5437, "step": 1411 }, { "epoch": 0.07142495826799534, "grad_norm": 14.440435409545898, "learning_rate": 9.905359667971678e-06, "loss": 0.5495, "step": 1412 }, { "epoch": 0.0714755425160605, "grad_norm": 5.424169063568115, "learning_rate": 9.905204183626256e-06, "loss": 0.4676, "step": 1413 }, { "epoch": 0.07152612676412565, "grad_norm": 12.314888954162598, "learning_rate": 9.905048572885571e-06, "loss": 0.4233, "step": 1414 }, { "epoch": 0.0715767110121908, "grad_norm": 11.801104545593262, "learning_rate": 9.904892835753632e-06, "loss": 0.5614, "step": 1415 }, { "epoch": 0.07162729526025596, "grad_norm": 23.114334106445312, "learning_rate": 9.904736972234453e-06, "loss": 0.647, "step": 1416 }, { "epoch": 0.07167787950832111, "grad_norm": 11.803875923156738, "learning_rate": 9.904580982332048e-06, "loss": 0.546, "step": 1417 }, { "epoch": 0.07172846375638627, "grad_norm": 10.486919403076172, "learning_rate": 9.904424866050436e-06, "loss": 0.5509, "step": 1418 }, { "epoch": 0.07177904800445141, "grad_norm": 10.297253608703613, "learning_rate": 9.904268623393645e-06, "loss": 0.4498, "step": 1419 }, { "epoch": 0.07182963225251657, "grad_norm": 9.229914665222168, "learning_rate": 9.904112254365695e-06, "loss": 0.5473, "step": 1420 }, { "epoch": 0.07188021650058173, "grad_norm": 15.51598072052002, "learning_rate": 9.903955758970619e-06, "loss": 0.5087, "step": 1421 }, { "epoch": 0.07193080074864687, "grad_norm": 9.16920280456543, "learning_rate": 9.903799137212447e-06, "loss": 0.543, "step": 1422 }, { "epoch": 0.07198138499671203, "grad_norm": 18.152843475341797, "learning_rate": 9.903642389095216e-06, "loss": 0.6256, "step": 1423 }, { "epoch": 0.07203196924477717, "grad_norm": 9.21558952331543, "learning_rate": 9.903485514622963e-06, "loss": 0.5496, "step": 1424 }, { "epoch": 0.07208255349284233, "grad_norm": 18.606212615966797, "learning_rate": 9.903328513799736e-06, "loss": 0.6754, "step": 1425 }, { "epoch": 0.07213313774090749, "grad_norm": 8.072100639343262, "learning_rate": 9.903171386629574e-06, "loss": 0.4719, "step": 1426 }, { "epoch": 0.07218372198897263, "grad_norm": 18.91046905517578, "learning_rate": 9.90301413311653e-06, "loss": 0.5791, "step": 1427 }, { "epoch": 0.07223430623703779, "grad_norm": 12.992467880249023, "learning_rate": 9.902856753264654e-06, "loss": 0.4884, "step": 1428 }, { "epoch": 0.07228489048510293, "grad_norm": 13.12845516204834, "learning_rate": 9.902699247078001e-06, "loss": 0.4871, "step": 1429 }, { "epoch": 0.07233547473316809, "grad_norm": 7.903231143951416, "learning_rate": 9.90254161456063e-06, "loss": 0.4796, "step": 1430 }, { "epoch": 0.07238605898123325, "grad_norm": 10.463624954223633, "learning_rate": 9.902383855716606e-06, "loss": 0.4493, "step": 1431 }, { "epoch": 0.0724366432292984, "grad_norm": 11.173397064208984, "learning_rate": 9.902225970549988e-06, "loss": 0.4034, "step": 1432 }, { "epoch": 0.07248722747736355, "grad_norm": 8.191490173339844, "learning_rate": 9.90206795906485e-06, "loss": 0.4739, "step": 1433 }, { "epoch": 0.0725378117254287, "grad_norm": 11.38820743560791, "learning_rate": 9.90190982126526e-06, "loss": 0.5179, "step": 1434 }, { "epoch": 0.07258839597349385, "grad_norm": 13.027939796447754, "learning_rate": 9.901751557155294e-06, "loss": 0.575, "step": 1435 }, { "epoch": 0.07263898022155901, "grad_norm": 13.006616592407227, "learning_rate": 9.901593166739031e-06, "loss": 0.5725, "step": 1436 }, { "epoch": 0.07268956446962416, "grad_norm": 12.81505012512207, "learning_rate": 9.90143465002055e-06, "loss": 0.5786, "step": 1437 }, { "epoch": 0.07274014871768932, "grad_norm": 10.010563850402832, "learning_rate": 9.901276007003938e-06, "loss": 0.4518, "step": 1438 }, { "epoch": 0.07279073296575446, "grad_norm": 8.598267555236816, "learning_rate": 9.901117237693282e-06, "loss": 0.3705, "step": 1439 }, { "epoch": 0.07284131721381962, "grad_norm": 8.415310859680176, "learning_rate": 9.900958342092673e-06, "loss": 0.5471, "step": 1440 }, { "epoch": 0.07289190146188478, "grad_norm": 13.520275115966797, "learning_rate": 9.900799320206205e-06, "loss": 0.4997, "step": 1441 }, { "epoch": 0.07294248570994992, "grad_norm": 15.150456428527832, "learning_rate": 9.900640172037976e-06, "loss": 0.5012, "step": 1442 }, { "epoch": 0.07299306995801508, "grad_norm": 9.785075187683105, "learning_rate": 9.900480897592086e-06, "loss": 0.4314, "step": 1443 }, { "epoch": 0.07304365420608022, "grad_norm": 11.455830574035645, "learning_rate": 9.90032149687264e-06, "loss": 0.4974, "step": 1444 }, { "epoch": 0.07309423845414538, "grad_norm": 11.210528373718262, "learning_rate": 9.900161969883745e-06, "loss": 0.5333, "step": 1445 }, { "epoch": 0.07314482270221054, "grad_norm": 9.734731674194336, "learning_rate": 9.900002316629514e-06, "loss": 0.4813, "step": 1446 }, { "epoch": 0.07319540695027568, "grad_norm": 9.362186431884766, "learning_rate": 9.899842537114057e-06, "loss": 0.3596, "step": 1447 }, { "epoch": 0.07324599119834084, "grad_norm": 13.514559745788574, "learning_rate": 9.899682631341493e-06, "loss": 0.4926, "step": 1448 }, { "epoch": 0.07329657544640598, "grad_norm": 7.3577117919921875, "learning_rate": 9.899522599315943e-06, "loss": 0.457, "step": 1449 }, { "epoch": 0.07334715969447114, "grad_norm": 10.39692497253418, "learning_rate": 9.89936244104153e-06, "loss": 0.5851, "step": 1450 }, { "epoch": 0.0733977439425363, "grad_norm": 10.276847839355469, "learning_rate": 9.89920215652238e-06, "loss": 0.5083, "step": 1451 }, { "epoch": 0.07344832819060144, "grad_norm": 7.891593933105469, "learning_rate": 9.899041745762623e-06, "loss": 0.4074, "step": 1452 }, { "epoch": 0.0734989124386666, "grad_norm": 8.373456001281738, "learning_rate": 9.898881208766395e-06, "loss": 0.5026, "step": 1453 }, { "epoch": 0.07354949668673175, "grad_norm": 11.380097389221191, "learning_rate": 9.898720545537831e-06, "loss": 0.4312, "step": 1454 }, { "epoch": 0.0736000809347969, "grad_norm": 25.118715286254883, "learning_rate": 9.89855975608107e-06, "loss": 0.4966, "step": 1455 }, { "epoch": 0.07365066518286206, "grad_norm": 10.501276016235352, "learning_rate": 9.898398840400257e-06, "loss": 0.4359, "step": 1456 }, { "epoch": 0.07370124943092721, "grad_norm": 11.701130867004395, "learning_rate": 9.898237798499537e-06, "loss": 0.4847, "step": 1457 }, { "epoch": 0.07375183367899237, "grad_norm": 9.00269889831543, "learning_rate": 9.89807663038306e-06, "loss": 0.4445, "step": 1458 }, { "epoch": 0.07380241792705751, "grad_norm": 32.7095947265625, "learning_rate": 9.89791533605498e-06, "loss": 0.6287, "step": 1459 }, { "epoch": 0.07385300217512267, "grad_norm": 8.203532218933105, "learning_rate": 9.89775391551945e-06, "loss": 0.557, "step": 1460 }, { "epoch": 0.07390358642318783, "grad_norm": 10.286773681640625, "learning_rate": 9.897592368780633e-06, "loss": 0.4894, "step": 1461 }, { "epoch": 0.07395417067125297, "grad_norm": 12.389739990234375, "learning_rate": 9.89743069584269e-06, "loss": 0.5596, "step": 1462 }, { "epoch": 0.07400475491931813, "grad_norm": 21.556848526000977, "learning_rate": 9.89726889670979e-06, "loss": 0.6235, "step": 1463 }, { "epoch": 0.07405533916738327, "grad_norm": 10.508918762207031, "learning_rate": 9.897106971386097e-06, "loss": 0.5229, "step": 1464 }, { "epoch": 0.07410592341544843, "grad_norm": 9.8958740234375, "learning_rate": 9.896944919875786e-06, "loss": 0.5455, "step": 1465 }, { "epoch": 0.07415650766351359, "grad_norm": 12.12164306640625, "learning_rate": 9.896782742183032e-06, "loss": 0.4723, "step": 1466 }, { "epoch": 0.07420709191157873, "grad_norm": 25.39642333984375, "learning_rate": 9.896620438312014e-06, "loss": 0.5893, "step": 1467 }, { "epoch": 0.07425767615964389, "grad_norm": 7.343916893005371, "learning_rate": 9.896458008266915e-06, "loss": 0.4325, "step": 1468 }, { "epoch": 0.07430826040770903, "grad_norm": 9.497089385986328, "learning_rate": 9.896295452051921e-06, "loss": 0.5042, "step": 1469 }, { "epoch": 0.07435884465577419, "grad_norm": 7.862093448638916, "learning_rate": 9.89613276967122e-06, "loss": 0.5448, "step": 1470 }, { "epoch": 0.07440942890383935, "grad_norm": 11.40882682800293, "learning_rate": 9.895969961129004e-06, "loss": 0.494, "step": 1471 }, { "epoch": 0.0744600131519045, "grad_norm": 4.667178630828857, "learning_rate": 9.895807026429464e-06, "loss": 0.4618, "step": 1472 }, { "epoch": 0.07451059739996965, "grad_norm": 7.143405437469482, "learning_rate": 9.895643965576807e-06, "loss": 0.5565, "step": 1473 }, { "epoch": 0.0745611816480348, "grad_norm": 15.816152572631836, "learning_rate": 9.895480778575227e-06, "loss": 0.5803, "step": 1474 }, { "epoch": 0.07461176589609995, "grad_norm": 13.041996002197266, "learning_rate": 9.895317465428932e-06, "loss": 0.6399, "step": 1475 }, { "epoch": 0.07466235014416511, "grad_norm": 12.899856567382812, "learning_rate": 9.895154026142133e-06, "loss": 0.6623, "step": 1476 }, { "epoch": 0.07471293439223026, "grad_norm": 10.5542573928833, "learning_rate": 9.894990460719037e-06, "loss": 0.4413, "step": 1477 }, { "epoch": 0.07476351864029541, "grad_norm": 8.907977104187012, "learning_rate": 9.894826769163858e-06, "loss": 0.5282, "step": 1478 }, { "epoch": 0.07481410288836056, "grad_norm": 8.901573181152344, "learning_rate": 9.894662951480819e-06, "loss": 0.4656, "step": 1479 }, { "epoch": 0.07486468713642572, "grad_norm": 6.495519161224365, "learning_rate": 9.894499007674136e-06, "loss": 0.3986, "step": 1480 }, { "epoch": 0.07491527138449088, "grad_norm": 14.460328102111816, "learning_rate": 9.894334937748035e-06, "loss": 0.5911, "step": 1481 }, { "epoch": 0.07496585563255602, "grad_norm": 11.738529205322266, "learning_rate": 9.894170741706747e-06, "loss": 0.5825, "step": 1482 }, { "epoch": 0.07501643988062118, "grad_norm": 19.884403228759766, "learning_rate": 9.894006419554501e-06, "loss": 0.6096, "step": 1483 }, { "epoch": 0.07506702412868632, "grad_norm": 14.238449096679688, "learning_rate": 9.893841971295529e-06, "loss": 0.5477, "step": 1484 }, { "epoch": 0.07511760837675148, "grad_norm": 11.643157958984375, "learning_rate": 9.89367739693407e-06, "loss": 0.4662, "step": 1485 }, { "epoch": 0.07516819262481664, "grad_norm": 6.311812400817871, "learning_rate": 9.893512696474363e-06, "loss": 0.4749, "step": 1486 }, { "epoch": 0.07521877687288178, "grad_norm": 15.866637229919434, "learning_rate": 9.893347869920656e-06, "loss": 0.5698, "step": 1487 }, { "epoch": 0.07526936112094694, "grad_norm": 7.672445774078369, "learning_rate": 9.893182917277193e-06, "loss": 0.5287, "step": 1488 }, { "epoch": 0.07531994536901208, "grad_norm": 38.304317474365234, "learning_rate": 9.893017838548227e-06, "loss": 0.5059, "step": 1489 }, { "epoch": 0.07537052961707724, "grad_norm": 11.276658058166504, "learning_rate": 9.892852633738006e-06, "loss": 0.5815, "step": 1490 }, { "epoch": 0.0754211138651424, "grad_norm": 12.761786460876465, "learning_rate": 9.892687302850795e-06, "loss": 0.4534, "step": 1491 }, { "epoch": 0.07547169811320754, "grad_norm": 14.965200424194336, "learning_rate": 9.892521845890848e-06, "loss": 0.5626, "step": 1492 }, { "epoch": 0.0755222823612727, "grad_norm": 7.560397148132324, "learning_rate": 9.89235626286243e-06, "loss": 0.5152, "step": 1493 }, { "epoch": 0.07557286660933785, "grad_norm": 17.05508041381836, "learning_rate": 9.89219055376981e-06, "loss": 0.4517, "step": 1494 }, { "epoch": 0.075623450857403, "grad_norm": 13.583589553833008, "learning_rate": 9.892024718617254e-06, "loss": 0.5326, "step": 1495 }, { "epoch": 0.07567403510546816, "grad_norm": 17.437902450561523, "learning_rate": 9.891858757409038e-06, "loss": 0.5767, "step": 1496 }, { "epoch": 0.0757246193535333, "grad_norm": 11.283146858215332, "learning_rate": 9.891692670149437e-06, "loss": 0.5169, "step": 1497 }, { "epoch": 0.07577520360159846, "grad_norm": 8.11658000946045, "learning_rate": 9.891526456842732e-06, "loss": 0.3921, "step": 1498 }, { "epoch": 0.07582578784966361, "grad_norm": 10.968807220458984, "learning_rate": 9.891360117493204e-06, "loss": 0.4998, "step": 1499 }, { "epoch": 0.07587637209772877, "grad_norm": 14.70217514038086, "learning_rate": 9.891193652105142e-06, "loss": 0.495, "step": 1500 }, { "epoch": 0.07592695634579393, "grad_norm": 14.779682159423828, "learning_rate": 9.891027060682834e-06, "loss": 0.403, "step": 1501 }, { "epoch": 0.07597754059385907, "grad_norm": 13.9553804397583, "learning_rate": 9.89086034323057e-06, "loss": 0.586, "step": 1502 }, { "epoch": 0.07602812484192423, "grad_norm": 10.302974700927734, "learning_rate": 9.89069349975265e-06, "loss": 0.5021, "step": 1503 }, { "epoch": 0.07607870908998937, "grad_norm": 11.307626724243164, "learning_rate": 9.890526530253373e-06, "loss": 0.6057, "step": 1504 }, { "epoch": 0.07612929333805453, "grad_norm": 14.0724515914917, "learning_rate": 9.890359434737038e-06, "loss": 0.5923, "step": 1505 }, { "epoch": 0.07617987758611969, "grad_norm": 6.753888130187988, "learning_rate": 9.890192213207953e-06, "loss": 0.4975, "step": 1506 }, { "epoch": 0.07623046183418483, "grad_norm": 12.13442611694336, "learning_rate": 9.890024865670427e-06, "loss": 0.3852, "step": 1507 }, { "epoch": 0.07628104608224999, "grad_norm": 8.099991798400879, "learning_rate": 9.88985739212877e-06, "loss": 0.4757, "step": 1508 }, { "epoch": 0.07633163033031513, "grad_norm": 17.459142684936523, "learning_rate": 9.8896897925873e-06, "loss": 0.5063, "step": 1509 }, { "epoch": 0.07638221457838029, "grad_norm": 6.701237678527832, "learning_rate": 9.889522067050335e-06, "loss": 0.5565, "step": 1510 }, { "epoch": 0.07643279882644545, "grad_norm": 7.759278774261475, "learning_rate": 9.889354215522195e-06, "loss": 0.4607, "step": 1511 }, { "epoch": 0.0764833830745106, "grad_norm": 11.028894424438477, "learning_rate": 9.889186238007209e-06, "loss": 0.416, "step": 1512 }, { "epoch": 0.07653396732257575, "grad_norm": 6.620573997497559, "learning_rate": 9.8890181345097e-06, "loss": 0.5027, "step": 1513 }, { "epoch": 0.0765845515706409, "grad_norm": 5.344559669494629, "learning_rate": 9.888849905034004e-06, "loss": 0.4822, "step": 1514 }, { "epoch": 0.07663513581870605, "grad_norm": 14.123164176940918, "learning_rate": 9.888681549584455e-06, "loss": 0.5568, "step": 1515 }, { "epoch": 0.07668572006677121, "grad_norm": 15.213176727294922, "learning_rate": 9.88851306816539e-06, "loss": 0.4786, "step": 1516 }, { "epoch": 0.07673630431483636, "grad_norm": 14.423091888427734, "learning_rate": 9.888344460781151e-06, "loss": 0.5341, "step": 1517 }, { "epoch": 0.07678688856290151, "grad_norm": 6.943253517150879, "learning_rate": 9.888175727436084e-06, "loss": 0.3777, "step": 1518 }, { "epoch": 0.07683747281096666, "grad_norm": 7.043682098388672, "learning_rate": 9.888006868134534e-06, "loss": 0.4112, "step": 1519 }, { "epoch": 0.07688805705903182, "grad_norm": 12.046351432800293, "learning_rate": 9.887837882880854e-06, "loss": 0.3696, "step": 1520 }, { "epoch": 0.07693864130709697, "grad_norm": 16.51590347290039, "learning_rate": 9.887668771679398e-06, "loss": 0.4906, "step": 1521 }, { "epoch": 0.07698922555516212, "grad_norm": 15.276084899902344, "learning_rate": 9.887499534534523e-06, "loss": 0.7009, "step": 1522 }, { "epoch": 0.07703980980322728, "grad_norm": 14.881805419921875, "learning_rate": 9.88733017145059e-06, "loss": 0.5226, "step": 1523 }, { "epoch": 0.07709039405129242, "grad_norm": 7.678349494934082, "learning_rate": 9.887160682431963e-06, "loss": 0.4575, "step": 1524 }, { "epoch": 0.07714097829935758, "grad_norm": 9.243696212768555, "learning_rate": 9.88699106748301e-06, "loss": 0.4138, "step": 1525 }, { "epoch": 0.07719156254742274, "grad_norm": 13.691131591796875, "learning_rate": 9.886821326608102e-06, "loss": 0.5422, "step": 1526 }, { "epoch": 0.07724214679548788, "grad_norm": 14.240006446838379, "learning_rate": 9.88665145981161e-06, "loss": 0.3921, "step": 1527 }, { "epoch": 0.07729273104355304, "grad_norm": 10.828779220581055, "learning_rate": 9.886481467097917e-06, "loss": 0.4183, "step": 1528 }, { "epoch": 0.07734331529161818, "grad_norm": 10.113965034484863, "learning_rate": 9.886311348471396e-06, "loss": 0.5517, "step": 1529 }, { "epoch": 0.07739389953968334, "grad_norm": 12.428647994995117, "learning_rate": 9.886141103936435e-06, "loss": 0.5591, "step": 1530 }, { "epoch": 0.0774444837877485, "grad_norm": 17.158950805664062, "learning_rate": 9.885970733497421e-06, "loss": 0.4294, "step": 1531 }, { "epoch": 0.07749506803581364, "grad_norm": 11.734905242919922, "learning_rate": 9.88580023715874e-06, "loss": 0.4308, "step": 1532 }, { "epoch": 0.0775456522838788, "grad_norm": 6.401408672332764, "learning_rate": 9.88562961492479e-06, "loss": 0.5117, "step": 1533 }, { "epoch": 0.07759623653194395, "grad_norm": 20.560699462890625, "learning_rate": 9.885458866799966e-06, "loss": 0.5758, "step": 1534 }, { "epoch": 0.0776468207800091, "grad_norm": 7.940581798553467, "learning_rate": 9.885287992788665e-06, "loss": 0.4727, "step": 1535 }, { "epoch": 0.07769740502807426, "grad_norm": 8.146711349487305, "learning_rate": 9.885116992895295e-06, "loss": 0.492, "step": 1536 }, { "epoch": 0.0777479892761394, "grad_norm": 11.871431350708008, "learning_rate": 9.884945867124258e-06, "loss": 0.5839, "step": 1537 }, { "epoch": 0.07779857352420456, "grad_norm": 11.089533805847168, "learning_rate": 9.884774615479967e-06, "loss": 0.5095, "step": 1538 }, { "epoch": 0.07784915777226971, "grad_norm": 17.3154354095459, "learning_rate": 9.88460323796683e-06, "loss": 0.6915, "step": 1539 }, { "epoch": 0.07789974202033487, "grad_norm": 10.522151947021484, "learning_rate": 9.884431734589269e-06, "loss": 0.506, "step": 1540 }, { "epoch": 0.07795032626840002, "grad_norm": 6.46327018737793, "learning_rate": 9.884260105351697e-06, "loss": 0.3529, "step": 1541 }, { "epoch": 0.07800091051646517, "grad_norm": 12.463274955749512, "learning_rate": 9.88408835025854e-06, "loss": 0.6119, "step": 1542 }, { "epoch": 0.07805149476453033, "grad_norm": 9.744842529296875, "learning_rate": 9.883916469314225e-06, "loss": 0.5231, "step": 1543 }, { "epoch": 0.07810207901259547, "grad_norm": 8.340315818786621, "learning_rate": 9.883744462523178e-06, "loss": 0.4962, "step": 1544 }, { "epoch": 0.07815266326066063, "grad_norm": 7.7607927322387695, "learning_rate": 9.883572329889832e-06, "loss": 0.4441, "step": 1545 }, { "epoch": 0.07820324750872579, "grad_norm": 8.123613357543945, "learning_rate": 9.883400071418626e-06, "loss": 0.5673, "step": 1546 }, { "epoch": 0.07825383175679093, "grad_norm": 8.633678436279297, "learning_rate": 9.883227687113992e-06, "loss": 0.5444, "step": 1547 }, { "epoch": 0.07830441600485609, "grad_norm": 12.127372741699219, "learning_rate": 9.883055176980376e-06, "loss": 0.5257, "step": 1548 }, { "epoch": 0.07835500025292123, "grad_norm": 6.563934803009033, "learning_rate": 9.882882541022223e-06, "loss": 0.5343, "step": 1549 }, { "epoch": 0.07840558450098639, "grad_norm": 7.16607141494751, "learning_rate": 9.882709779243981e-06, "loss": 0.4571, "step": 1550 }, { "epoch": 0.07845616874905155, "grad_norm": 9.016817092895508, "learning_rate": 9.882536891650104e-06, "loss": 0.6516, "step": 1551 }, { "epoch": 0.0785067529971167, "grad_norm": 7.259747505187988, "learning_rate": 9.88236387824504e-06, "loss": 0.5169, "step": 1552 }, { "epoch": 0.07855733724518185, "grad_norm": 10.357641220092773, "learning_rate": 9.882190739033256e-06, "loss": 0.5802, "step": 1553 }, { "epoch": 0.078607921493247, "grad_norm": 19.596355438232422, "learning_rate": 9.882017474019207e-06, "loss": 0.5643, "step": 1554 }, { "epoch": 0.07865850574131215, "grad_norm": 14.16342830657959, "learning_rate": 9.88184408320736e-06, "loss": 0.5863, "step": 1555 }, { "epoch": 0.07870908998937731, "grad_norm": 11.302835464477539, "learning_rate": 9.881670566602182e-06, "loss": 0.5516, "step": 1556 }, { "epoch": 0.07875967423744246, "grad_norm": 13.522135734558105, "learning_rate": 9.881496924208144e-06, "loss": 0.4936, "step": 1557 }, { "epoch": 0.07881025848550761, "grad_norm": 17.716094970703125, "learning_rate": 9.881323156029724e-06, "loss": 0.5609, "step": 1558 }, { "epoch": 0.07886084273357276, "grad_norm": 9.475680351257324, "learning_rate": 9.881149262071394e-06, "loss": 0.4881, "step": 1559 }, { "epoch": 0.07891142698163792, "grad_norm": 7.8265485763549805, "learning_rate": 9.88097524233764e-06, "loss": 0.442, "step": 1560 }, { "epoch": 0.07896201122970307, "grad_norm": 7.516635417938232, "learning_rate": 9.880801096832942e-06, "loss": 0.4888, "step": 1561 }, { "epoch": 0.07901259547776822, "grad_norm": 11.65626049041748, "learning_rate": 9.880626825561788e-06, "loss": 0.539, "step": 1562 }, { "epoch": 0.07906317972583338, "grad_norm": 17.569744110107422, "learning_rate": 9.880452428528671e-06, "loss": 0.5946, "step": 1563 }, { "epoch": 0.07911376397389853, "grad_norm": 8.105836868286133, "learning_rate": 9.880277905738082e-06, "loss": 0.4949, "step": 1564 }, { "epoch": 0.07916434822196368, "grad_norm": 17.565397262573242, "learning_rate": 9.88010325719452e-06, "loss": 0.6302, "step": 1565 }, { "epoch": 0.07921493247002884, "grad_norm": 9.884222030639648, "learning_rate": 9.879928482902484e-06, "loss": 0.5974, "step": 1566 }, { "epoch": 0.07926551671809398, "grad_norm": 7.082926273345947, "learning_rate": 9.879753582866478e-06, "loss": 0.4007, "step": 1567 }, { "epoch": 0.07931610096615914, "grad_norm": 19.953054428100586, "learning_rate": 9.879578557091007e-06, "loss": 0.5374, "step": 1568 }, { "epoch": 0.0793666852142243, "grad_norm": 14.321699142456055, "learning_rate": 9.879403405580585e-06, "loss": 0.5463, "step": 1569 }, { "epoch": 0.07941726946228944, "grad_norm": 13.608661651611328, "learning_rate": 9.879228128339722e-06, "loss": 0.3894, "step": 1570 }, { "epoch": 0.0794678537103546, "grad_norm": 6.801243782043457, "learning_rate": 9.879052725372936e-06, "loss": 0.4583, "step": 1571 }, { "epoch": 0.07951843795841974, "grad_norm": 9.172856330871582, "learning_rate": 9.878877196684747e-06, "loss": 0.5061, "step": 1572 }, { "epoch": 0.0795690222064849, "grad_norm": 16.098037719726562, "learning_rate": 9.878701542279675e-06, "loss": 0.7036, "step": 1573 }, { "epoch": 0.07961960645455006, "grad_norm": 8.648968696594238, "learning_rate": 9.87852576216225e-06, "loss": 0.5543, "step": 1574 }, { "epoch": 0.0796701907026152, "grad_norm": 11.80416488647461, "learning_rate": 9.878349856336998e-06, "loss": 0.5191, "step": 1575 }, { "epoch": 0.07972077495068036, "grad_norm": 12.240236282348633, "learning_rate": 9.878173824808455e-06, "loss": 0.4646, "step": 1576 }, { "epoch": 0.0797713591987455, "grad_norm": 10.84026050567627, "learning_rate": 9.877997667581155e-06, "loss": 0.5112, "step": 1577 }, { "epoch": 0.07982194344681066, "grad_norm": 6.468084335327148, "learning_rate": 9.877821384659636e-06, "loss": 0.5531, "step": 1578 }, { "epoch": 0.07987252769487582, "grad_norm": 9.265419960021973, "learning_rate": 9.877644976048444e-06, "loss": 0.5316, "step": 1579 }, { "epoch": 0.07992311194294097, "grad_norm": 10.800164222717285, "learning_rate": 9.877468441752121e-06, "loss": 0.4357, "step": 1580 }, { "epoch": 0.07997369619100612, "grad_norm": 10.594705581665039, "learning_rate": 9.877291781775218e-06, "loss": 0.4078, "step": 1581 }, { "epoch": 0.08002428043907127, "grad_norm": 8.963818550109863, "learning_rate": 9.877114996122285e-06, "loss": 0.5015, "step": 1582 }, { "epoch": 0.08007486468713643, "grad_norm": 8.202963829040527, "learning_rate": 9.87693808479788e-06, "loss": 0.5362, "step": 1583 }, { "epoch": 0.08012544893520158, "grad_norm": 9.405852317810059, "learning_rate": 9.87676104780656e-06, "loss": 0.3757, "step": 1584 }, { "epoch": 0.08017603318326673, "grad_norm": 10.808770179748535, "learning_rate": 9.876583885152887e-06, "loss": 0.4495, "step": 1585 }, { "epoch": 0.08022661743133189, "grad_norm": 30.800342559814453, "learning_rate": 9.876406596841426e-06, "loss": 0.6593, "step": 1586 }, { "epoch": 0.08027720167939703, "grad_norm": 14.43838882446289, "learning_rate": 9.876229182876746e-06, "loss": 0.5743, "step": 1587 }, { "epoch": 0.08032778592746219, "grad_norm": 14.408217430114746, "learning_rate": 9.876051643263418e-06, "loss": 0.4507, "step": 1588 }, { "epoch": 0.08037837017552735, "grad_norm": 20.383882522583008, "learning_rate": 9.875873978006017e-06, "loss": 0.6922, "step": 1589 }, { "epoch": 0.08042895442359249, "grad_norm": 16.194534301757812, "learning_rate": 9.87569618710912e-06, "loss": 0.4937, "step": 1590 }, { "epoch": 0.08047953867165765, "grad_norm": 9.041959762573242, "learning_rate": 9.87551827057731e-06, "loss": 0.4643, "step": 1591 }, { "epoch": 0.0805301229197228, "grad_norm": 10.109902381896973, "learning_rate": 9.87534022841517e-06, "loss": 0.4441, "step": 1592 }, { "epoch": 0.08058070716778795, "grad_norm": 20.741579055786133, "learning_rate": 9.875162060627289e-06, "loss": 0.6854, "step": 1593 }, { "epoch": 0.08063129141585311, "grad_norm": 19.153860092163086, "learning_rate": 9.874983767218255e-06, "loss": 0.6215, "step": 1594 }, { "epoch": 0.08068187566391825, "grad_norm": 12.614603042602539, "learning_rate": 9.874805348192667e-06, "loss": 0.4318, "step": 1595 }, { "epoch": 0.08073245991198341, "grad_norm": 10.385489463806152, "learning_rate": 9.874626803555119e-06, "loss": 0.4895, "step": 1596 }, { "epoch": 0.08078304416004856, "grad_norm": 9.869991302490234, "learning_rate": 9.87444813331021e-06, "loss": 0.5023, "step": 1597 }, { "epoch": 0.08083362840811371, "grad_norm": 12.125442504882812, "learning_rate": 9.874269337462549e-06, "loss": 0.442, "step": 1598 }, { "epoch": 0.08088421265617887, "grad_norm": 13.063725471496582, "learning_rate": 9.874090416016739e-06, "loss": 0.5597, "step": 1599 }, { "epoch": 0.08093479690424402, "grad_norm": 14.553729057312012, "learning_rate": 9.873911368977393e-06, "loss": 0.379, "step": 1600 }, { "epoch": 0.08098538115230917, "grad_norm": 14.063051223754883, "learning_rate": 9.873732196349124e-06, "loss": 0.5888, "step": 1601 }, { "epoch": 0.08103596540037432, "grad_norm": 12.66696548461914, "learning_rate": 9.873552898136547e-06, "loss": 0.4061, "step": 1602 }, { "epoch": 0.08108654964843948, "grad_norm": 8.696595191955566, "learning_rate": 9.873373474344284e-06, "loss": 0.448, "step": 1603 }, { "epoch": 0.08113713389650463, "grad_norm": 10.390074729919434, "learning_rate": 9.873193924976958e-06, "loss": 0.4184, "step": 1604 }, { "epoch": 0.08118771814456978, "grad_norm": 14.22169017791748, "learning_rate": 9.873014250039195e-06, "loss": 0.5523, "step": 1605 }, { "epoch": 0.08123830239263494, "grad_norm": 8.238658905029297, "learning_rate": 9.872834449535624e-06, "loss": 0.4493, "step": 1606 }, { "epoch": 0.08128888664070008, "grad_norm": 8.542798042297363, "learning_rate": 9.87265452347088e-06, "loss": 0.4821, "step": 1607 }, { "epoch": 0.08133947088876524, "grad_norm": 8.162860870361328, "learning_rate": 9.872474471849598e-06, "loss": 0.5323, "step": 1608 }, { "epoch": 0.0813900551368304, "grad_norm": 9.622842788696289, "learning_rate": 9.872294294676419e-06, "loss": 0.4689, "step": 1609 }, { "epoch": 0.08144063938489554, "grad_norm": 10.325462341308594, "learning_rate": 9.872113991955982e-06, "loss": 0.487, "step": 1610 }, { "epoch": 0.0814912236329607, "grad_norm": 16.573978424072266, "learning_rate": 9.87193356369294e-06, "loss": 0.4965, "step": 1611 }, { "epoch": 0.08154180788102584, "grad_norm": 39.1019287109375, "learning_rate": 9.871753009891933e-06, "loss": 0.6882, "step": 1612 }, { "epoch": 0.081592392129091, "grad_norm": 10.513895988464355, "learning_rate": 9.87157233055762e-06, "loss": 0.5716, "step": 1613 }, { "epoch": 0.08164297637715616, "grad_norm": 28.033824920654297, "learning_rate": 9.871391525694656e-06, "loss": 0.7874, "step": 1614 }, { "epoch": 0.0816935606252213, "grad_norm": 9.591083526611328, "learning_rate": 9.871210595307698e-06, "loss": 0.4309, "step": 1615 }, { "epoch": 0.08174414487328646, "grad_norm": 8.618520736694336, "learning_rate": 9.87102953940141e-06, "loss": 0.4385, "step": 1616 }, { "epoch": 0.0817947291213516, "grad_norm": 6.385439395904541, "learning_rate": 9.870848357980454e-06, "loss": 0.5047, "step": 1617 }, { "epoch": 0.08184531336941676, "grad_norm": 9.645421028137207, "learning_rate": 9.8706670510495e-06, "loss": 0.5235, "step": 1618 }, { "epoch": 0.08189589761748192, "grad_norm": 7.312952518463135, "learning_rate": 9.870485618613224e-06, "loss": 0.552, "step": 1619 }, { "epoch": 0.08194648186554707, "grad_norm": 13.21914005279541, "learning_rate": 9.870304060676296e-06, "loss": 0.5481, "step": 1620 }, { "epoch": 0.08199706611361222, "grad_norm": 13.400589942932129, "learning_rate": 9.870122377243397e-06, "loss": 0.471, "step": 1621 }, { "epoch": 0.08204765036167737, "grad_norm": 10.142589569091797, "learning_rate": 9.869940568319205e-06, "loss": 0.5117, "step": 1622 }, { "epoch": 0.08209823460974253, "grad_norm": 8.343387603759766, "learning_rate": 9.86975863390841e-06, "loss": 0.4289, "step": 1623 }, { "epoch": 0.08214881885780768, "grad_norm": 10.570852279663086, "learning_rate": 9.869576574015697e-06, "loss": 0.6098, "step": 1624 }, { "epoch": 0.08219940310587283, "grad_norm": 6.9624481201171875, "learning_rate": 9.869394388645758e-06, "loss": 0.4627, "step": 1625 }, { "epoch": 0.08224998735393799, "grad_norm": 9.692605972290039, "learning_rate": 9.869212077803286e-06, "loss": 0.3684, "step": 1626 }, { "epoch": 0.08230057160200313, "grad_norm": 14.469637870788574, "learning_rate": 9.86902964149298e-06, "loss": 0.4248, "step": 1627 }, { "epoch": 0.08235115585006829, "grad_norm": 10.5553617477417, "learning_rate": 9.86884707971954e-06, "loss": 0.4323, "step": 1628 }, { "epoch": 0.08240174009813345, "grad_norm": 19.259475708007812, "learning_rate": 9.868664392487674e-06, "loss": 0.4927, "step": 1629 }, { "epoch": 0.08245232434619859, "grad_norm": 9.449013710021973, "learning_rate": 9.868481579802086e-06, "loss": 0.563, "step": 1630 }, { "epoch": 0.08250290859426375, "grad_norm": 8.142119407653809, "learning_rate": 9.868298641667484e-06, "loss": 0.3711, "step": 1631 }, { "epoch": 0.0825534928423289, "grad_norm": 9.216270446777344, "learning_rate": 9.868115578088588e-06, "loss": 0.421, "step": 1632 }, { "epoch": 0.08260407709039405, "grad_norm": 11.130351066589355, "learning_rate": 9.86793238907011e-06, "loss": 0.5085, "step": 1633 }, { "epoch": 0.08265466133845921, "grad_norm": 9.645954132080078, "learning_rate": 9.867749074616773e-06, "loss": 0.5262, "step": 1634 }, { "epoch": 0.08270524558652435, "grad_norm": 6.657686233520508, "learning_rate": 9.8675656347333e-06, "loss": 0.5161, "step": 1635 }, { "epoch": 0.08275582983458951, "grad_norm": 10.802709579467773, "learning_rate": 9.867382069424417e-06, "loss": 0.5593, "step": 1636 }, { "epoch": 0.08280641408265466, "grad_norm": 7.465345859527588, "learning_rate": 9.867198378694855e-06, "loss": 0.5029, "step": 1637 }, { "epoch": 0.08285699833071981, "grad_norm": 7.853663444519043, "learning_rate": 9.867014562549348e-06, "loss": 0.5189, "step": 1638 }, { "epoch": 0.08290758257878497, "grad_norm": 10.573230743408203, "learning_rate": 9.866830620992631e-06, "loss": 0.4093, "step": 1639 }, { "epoch": 0.08295816682685012, "grad_norm": 8.567730903625488, "learning_rate": 9.866646554029444e-06, "loss": 0.5188, "step": 1640 }, { "epoch": 0.08300875107491527, "grad_norm": 9.640490531921387, "learning_rate": 9.866462361664531e-06, "loss": 0.4974, "step": 1641 }, { "epoch": 0.08305933532298042, "grad_norm": 10.981006622314453, "learning_rate": 9.866278043902635e-06, "loss": 0.4995, "step": 1642 }, { "epoch": 0.08310991957104558, "grad_norm": 9.220463752746582, "learning_rate": 9.86609360074851e-06, "loss": 0.5245, "step": 1643 }, { "epoch": 0.08316050381911073, "grad_norm": 5.616707801818848, "learning_rate": 9.865909032206906e-06, "loss": 0.4631, "step": 1644 }, { "epoch": 0.08321108806717588, "grad_norm": 16.36834144592285, "learning_rate": 9.865724338282579e-06, "loss": 0.4323, "step": 1645 }, { "epoch": 0.08326167231524104, "grad_norm": 10.30675983428955, "learning_rate": 9.86553951898029e-06, "loss": 0.5456, "step": 1646 }, { "epoch": 0.08331225656330618, "grad_norm": 9.434503555297852, "learning_rate": 9.865354574304797e-06, "loss": 0.4745, "step": 1647 }, { "epoch": 0.08336284081137134, "grad_norm": 9.370694160461426, "learning_rate": 9.86516950426087e-06, "loss": 0.4592, "step": 1648 }, { "epoch": 0.0834134250594365, "grad_norm": 4.53615665435791, "learning_rate": 9.864984308853277e-06, "loss": 0.2823, "step": 1649 }, { "epoch": 0.08346400930750164, "grad_norm": 16.134021759033203, "learning_rate": 9.86479898808679e-06, "loss": 0.5791, "step": 1650 }, { "epoch": 0.0835145935555668, "grad_norm": 8.315022468566895, "learning_rate": 9.864613541966182e-06, "loss": 0.4538, "step": 1651 }, { "epoch": 0.08356517780363194, "grad_norm": 12.31552505493164, "learning_rate": 9.864427970496231e-06, "loss": 0.5176, "step": 1652 }, { "epoch": 0.0836157620516971, "grad_norm": 18.456315994262695, "learning_rate": 9.864242273681723e-06, "loss": 0.662, "step": 1653 }, { "epoch": 0.08366634629976226, "grad_norm": 14.414620399475098, "learning_rate": 9.864056451527442e-06, "loss": 0.5324, "step": 1654 }, { "epoch": 0.0837169305478274, "grad_norm": 5.92703104019165, "learning_rate": 9.863870504038174e-06, "loss": 0.4375, "step": 1655 }, { "epoch": 0.08376751479589256, "grad_norm": 10.485411643981934, "learning_rate": 9.86368443121871e-06, "loss": 0.5223, "step": 1656 }, { "epoch": 0.0838180990439577, "grad_norm": 12.772544860839844, "learning_rate": 9.863498233073846e-06, "loss": 0.561, "step": 1657 }, { "epoch": 0.08386868329202286, "grad_norm": 11.932833671569824, "learning_rate": 9.86331190960838e-06, "loss": 0.5161, "step": 1658 }, { "epoch": 0.08391926754008802, "grad_norm": 10.090291023254395, "learning_rate": 9.863125460827115e-06, "loss": 0.4299, "step": 1659 }, { "epoch": 0.08396985178815317, "grad_norm": 8.957674026489258, "learning_rate": 9.86293888673485e-06, "loss": 0.4803, "step": 1660 }, { "epoch": 0.08402043603621832, "grad_norm": 10.980063438415527, "learning_rate": 9.862752187336397e-06, "loss": 0.5518, "step": 1661 }, { "epoch": 0.08407102028428347, "grad_norm": 9.36278247833252, "learning_rate": 9.862565362636567e-06, "loss": 0.4628, "step": 1662 }, { "epoch": 0.08412160453234863, "grad_norm": 17.179414749145508, "learning_rate": 9.862378412640173e-06, "loss": 0.6015, "step": 1663 }, { "epoch": 0.08417218878041378, "grad_norm": 7.555184364318848, "learning_rate": 9.862191337352029e-06, "loss": 0.4405, "step": 1664 }, { "epoch": 0.08422277302847893, "grad_norm": 10.692133903503418, "learning_rate": 9.862004136776962e-06, "loss": 0.3534, "step": 1665 }, { "epoch": 0.08427335727654409, "grad_norm": 13.886868476867676, "learning_rate": 9.861816810919789e-06, "loss": 0.5419, "step": 1666 }, { "epoch": 0.08432394152460923, "grad_norm": 5.75516939163208, "learning_rate": 9.86162935978534e-06, "loss": 0.4016, "step": 1667 }, { "epoch": 0.08437452577267439, "grad_norm": 11.943541526794434, "learning_rate": 9.861441783378447e-06, "loss": 0.5658, "step": 1668 }, { "epoch": 0.08442511002073955, "grad_norm": 16.396875381469727, "learning_rate": 9.861254081703942e-06, "loss": 0.4814, "step": 1669 }, { "epoch": 0.08447569426880469, "grad_norm": 10.183768272399902, "learning_rate": 9.86106625476666e-06, "loss": 0.4645, "step": 1670 }, { "epoch": 0.08452627851686985, "grad_norm": 9.546126365661621, "learning_rate": 9.860878302571442e-06, "loss": 0.4959, "step": 1671 }, { "epoch": 0.08457686276493499, "grad_norm": 8.880634307861328, "learning_rate": 9.86069022512313e-06, "loss": 0.4937, "step": 1672 }, { "epoch": 0.08462744701300015, "grad_norm": 14.658769607543945, "learning_rate": 9.860502022426573e-06, "loss": 0.5022, "step": 1673 }, { "epoch": 0.08467803126106531, "grad_norm": 16.75995445251465, "learning_rate": 9.86031369448662e-06, "loss": 0.5174, "step": 1674 }, { "epoch": 0.08472861550913045, "grad_norm": 6.366328716278076, "learning_rate": 9.86012524130812e-06, "loss": 0.5016, "step": 1675 }, { "epoch": 0.08477919975719561, "grad_norm": 15.52210521697998, "learning_rate": 9.859936662895933e-06, "loss": 0.6235, "step": 1676 }, { "epoch": 0.08482978400526076, "grad_norm": 8.274341583251953, "learning_rate": 9.859747959254918e-06, "loss": 0.4254, "step": 1677 }, { "epoch": 0.08488036825332591, "grad_norm": 11.788444519042969, "learning_rate": 9.859559130389935e-06, "loss": 0.4907, "step": 1678 }, { "epoch": 0.08493095250139107, "grad_norm": 8.461584091186523, "learning_rate": 9.859370176305851e-06, "loss": 0.4696, "step": 1679 }, { "epoch": 0.08498153674945622, "grad_norm": 16.719221115112305, "learning_rate": 9.859181097007536e-06, "loss": 0.4796, "step": 1680 }, { "epoch": 0.08503212099752137, "grad_norm": 10.637688636779785, "learning_rate": 9.85899189249986e-06, "loss": 0.5104, "step": 1681 }, { "epoch": 0.08508270524558652, "grad_norm": 17.8576602935791, "learning_rate": 9.8588025627877e-06, "loss": 0.4752, "step": 1682 }, { "epoch": 0.08513328949365168, "grad_norm": 10.982163429260254, "learning_rate": 9.858613107875932e-06, "loss": 0.4215, "step": 1683 }, { "epoch": 0.08518387374171683, "grad_norm": 12.68514633178711, "learning_rate": 9.858423527769441e-06, "loss": 0.4355, "step": 1684 }, { "epoch": 0.08523445798978198, "grad_norm": 6.287144660949707, "learning_rate": 9.858233822473112e-06, "loss": 0.4584, "step": 1685 }, { "epoch": 0.08528504223784714, "grad_norm": 14.257061958312988, "learning_rate": 9.85804399199183e-06, "loss": 0.3097, "step": 1686 }, { "epoch": 0.08533562648591228, "grad_norm": 7.457954406738281, "learning_rate": 9.85785403633049e-06, "loss": 0.4442, "step": 1687 }, { "epoch": 0.08538621073397744, "grad_norm": 10.703139305114746, "learning_rate": 9.857663955493984e-06, "loss": 0.4171, "step": 1688 }, { "epoch": 0.0854367949820426, "grad_norm": 6.526852130889893, "learning_rate": 9.857473749487211e-06, "loss": 0.4291, "step": 1689 }, { "epoch": 0.08548737923010774, "grad_norm": 7.452974319458008, "learning_rate": 9.857283418315073e-06, "loss": 0.4431, "step": 1690 }, { "epoch": 0.0855379634781729, "grad_norm": 8.21695327758789, "learning_rate": 9.857092961982473e-06, "loss": 0.3996, "step": 1691 }, { "epoch": 0.08558854772623804, "grad_norm": 10.00966739654541, "learning_rate": 9.85690238049432e-06, "loss": 0.5371, "step": 1692 }, { "epoch": 0.0856391319743032, "grad_norm": 14.638986587524414, "learning_rate": 9.856711673855523e-06, "loss": 0.5614, "step": 1693 }, { "epoch": 0.08568971622236836, "grad_norm": 9.458870887756348, "learning_rate": 9.856520842070999e-06, "loss": 0.451, "step": 1694 }, { "epoch": 0.0857403004704335, "grad_norm": 12.458974838256836, "learning_rate": 9.85632988514566e-06, "loss": 0.5618, "step": 1695 }, { "epoch": 0.08579088471849866, "grad_norm": 20.730493545532227, "learning_rate": 9.856138803084434e-06, "loss": 0.6732, "step": 1696 }, { "epoch": 0.0858414689665638, "grad_norm": 14.221606254577637, "learning_rate": 9.855947595892237e-06, "loss": 0.4968, "step": 1697 }, { "epoch": 0.08589205321462896, "grad_norm": 16.507354736328125, "learning_rate": 9.855756263574001e-06, "loss": 0.4852, "step": 1698 }, { "epoch": 0.08594263746269412, "grad_norm": 14.91327953338623, "learning_rate": 9.855564806134655e-06, "loss": 0.4512, "step": 1699 }, { "epoch": 0.08599322171075927, "grad_norm": 12.827080726623535, "learning_rate": 9.85537322357913e-06, "loss": 0.5363, "step": 1700 }, { "epoch": 0.08604380595882442, "grad_norm": 8.325942993164062, "learning_rate": 9.855181515912367e-06, "loss": 0.5006, "step": 1701 }, { "epoch": 0.08609439020688957, "grad_norm": 11.95489501953125, "learning_rate": 9.854989683139304e-06, "loss": 0.5918, "step": 1702 }, { "epoch": 0.08614497445495473, "grad_norm": 14.576555252075195, "learning_rate": 9.854797725264883e-06, "loss": 0.5326, "step": 1703 }, { "epoch": 0.08619555870301988, "grad_norm": 9.190546035766602, "learning_rate": 9.85460564229405e-06, "loss": 0.4415, "step": 1704 }, { "epoch": 0.08624614295108503, "grad_norm": 13.476530075073242, "learning_rate": 9.854413434231756e-06, "loss": 0.626, "step": 1705 }, { "epoch": 0.08629672719915019, "grad_norm": 18.12171173095703, "learning_rate": 9.854221101082952e-06, "loss": 0.5045, "step": 1706 }, { "epoch": 0.08634731144721533, "grad_norm": 11.908918380737305, "learning_rate": 9.854028642852595e-06, "loss": 0.4814, "step": 1707 }, { "epoch": 0.08639789569528049, "grad_norm": 12.901350975036621, "learning_rate": 9.853836059545648e-06, "loss": 0.3891, "step": 1708 }, { "epoch": 0.08644847994334565, "grad_norm": 8.069640159606934, "learning_rate": 9.853643351167066e-06, "loss": 0.5112, "step": 1709 }, { "epoch": 0.08649906419141079, "grad_norm": 12.139983177185059, "learning_rate": 9.85345051772182e-06, "loss": 0.4922, "step": 1710 }, { "epoch": 0.08654964843947595, "grad_norm": 5.194583892822266, "learning_rate": 9.853257559214875e-06, "loss": 0.3972, "step": 1711 }, { "epoch": 0.0866002326875411, "grad_norm": 9.303813934326172, "learning_rate": 9.853064475651208e-06, "loss": 0.5814, "step": 1712 }, { "epoch": 0.08665081693560625, "grad_norm": 12.86970043182373, "learning_rate": 9.85287126703579e-06, "loss": 0.6313, "step": 1713 }, { "epoch": 0.08670140118367141, "grad_norm": 9.018763542175293, "learning_rate": 9.852677933373603e-06, "loss": 0.3755, "step": 1714 }, { "epoch": 0.08675198543173655, "grad_norm": 9.596274375915527, "learning_rate": 9.852484474669624e-06, "loss": 0.3911, "step": 1715 }, { "epoch": 0.08680256967980171, "grad_norm": 6.979623317718506, "learning_rate": 9.852290890928842e-06, "loss": 0.4308, "step": 1716 }, { "epoch": 0.08685315392786687, "grad_norm": 17.332172393798828, "learning_rate": 9.852097182156243e-06, "loss": 0.7387, "step": 1717 }, { "epoch": 0.08690373817593201, "grad_norm": 14.078479766845703, "learning_rate": 9.851903348356822e-06, "loss": 0.4657, "step": 1718 }, { "epoch": 0.08695432242399717, "grad_norm": 13.157388687133789, "learning_rate": 9.851709389535569e-06, "loss": 0.5674, "step": 1719 }, { "epoch": 0.08700490667206232, "grad_norm": 9.092456817626953, "learning_rate": 9.851515305697484e-06, "loss": 0.5217, "step": 1720 }, { "epoch": 0.08705549092012747, "grad_norm": 9.344330787658691, "learning_rate": 9.851321096847568e-06, "loss": 0.4379, "step": 1721 }, { "epoch": 0.08710607516819263, "grad_norm": 8.355547904968262, "learning_rate": 9.851126762990825e-06, "loss": 0.4578, "step": 1722 }, { "epoch": 0.08715665941625778, "grad_norm": 13.022180557250977, "learning_rate": 9.850932304132265e-06, "loss": 0.4041, "step": 1723 }, { "epoch": 0.08720724366432293, "grad_norm": 12.869723320007324, "learning_rate": 9.850737720276894e-06, "loss": 0.4994, "step": 1724 }, { "epoch": 0.08725782791238808, "grad_norm": 9.493541717529297, "learning_rate": 9.85054301142973e-06, "loss": 0.4415, "step": 1725 }, { "epoch": 0.08730841216045324, "grad_norm": 12.707379341125488, "learning_rate": 9.850348177595788e-06, "loss": 0.5143, "step": 1726 }, { "epoch": 0.0873589964085184, "grad_norm": 12.73519229888916, "learning_rate": 9.850153218780088e-06, "loss": 0.5759, "step": 1727 }, { "epoch": 0.08740958065658354, "grad_norm": 10.505592346191406, "learning_rate": 9.849958134987655e-06, "loss": 0.4902, "step": 1728 }, { "epoch": 0.0874601649046487, "grad_norm": 14.56128978729248, "learning_rate": 9.849762926223516e-06, "loss": 0.584, "step": 1729 }, { "epoch": 0.08751074915271384, "grad_norm": 7.624418258666992, "learning_rate": 9.849567592492699e-06, "loss": 0.4241, "step": 1730 }, { "epoch": 0.087561333400779, "grad_norm": 19.3909854888916, "learning_rate": 9.849372133800241e-06, "loss": 0.5449, "step": 1731 }, { "epoch": 0.08761191764884416, "grad_norm": 18.787322998046875, "learning_rate": 9.849176550151175e-06, "loss": 0.3905, "step": 1732 }, { "epoch": 0.0876625018969093, "grad_norm": 9.3512601852417, "learning_rate": 9.848980841550542e-06, "loss": 0.4775, "step": 1733 }, { "epoch": 0.08771308614497446, "grad_norm": 8.896815299987793, "learning_rate": 9.848785008003386e-06, "loss": 0.5077, "step": 1734 }, { "epoch": 0.0877636703930396, "grad_norm": 13.93103313446045, "learning_rate": 9.848589049514752e-06, "loss": 0.5271, "step": 1735 }, { "epoch": 0.08781425464110476, "grad_norm": 9.27192497253418, "learning_rate": 9.848392966089688e-06, "loss": 0.5019, "step": 1736 }, { "epoch": 0.08786483888916992, "grad_norm": 12.893123626708984, "learning_rate": 9.84819675773325e-06, "loss": 0.4357, "step": 1737 }, { "epoch": 0.08791542313723506, "grad_norm": 12.739590644836426, "learning_rate": 9.84800042445049e-06, "loss": 0.6154, "step": 1738 }, { "epoch": 0.08796600738530022, "grad_norm": 14.043575286865234, "learning_rate": 9.847803966246469e-06, "loss": 0.6008, "step": 1739 }, { "epoch": 0.08801659163336537, "grad_norm": 9.40376091003418, "learning_rate": 9.847607383126253e-06, "loss": 0.3675, "step": 1740 }, { "epoch": 0.08806717588143052, "grad_norm": 13.856907844543457, "learning_rate": 9.847410675094899e-06, "loss": 0.4472, "step": 1741 }, { "epoch": 0.08811776012949568, "grad_norm": 12.42581558227539, "learning_rate": 9.847213842157482e-06, "loss": 0.45, "step": 1742 }, { "epoch": 0.08816834437756083, "grad_norm": 11.46610164642334, "learning_rate": 9.847016884319071e-06, "loss": 0.4971, "step": 1743 }, { "epoch": 0.08821892862562598, "grad_norm": 10.670817375183105, "learning_rate": 9.846819801584745e-06, "loss": 0.5284, "step": 1744 }, { "epoch": 0.08826951287369113, "grad_norm": 16.644208908081055, "learning_rate": 9.846622593959579e-06, "loss": 0.5883, "step": 1745 }, { "epoch": 0.08832009712175629, "grad_norm": 9.972620010375977, "learning_rate": 9.846425261448653e-06, "loss": 0.485, "step": 1746 }, { "epoch": 0.08837068136982144, "grad_norm": 8.1404447555542, "learning_rate": 9.846227804057056e-06, "loss": 0.5097, "step": 1747 }, { "epoch": 0.08842126561788659, "grad_norm": 12.236173629760742, "learning_rate": 9.846030221789876e-06, "loss": 0.6408, "step": 1748 }, { "epoch": 0.08847184986595175, "grad_norm": 8.379069328308105, "learning_rate": 9.845832514652198e-06, "loss": 0.4403, "step": 1749 }, { "epoch": 0.08852243411401689, "grad_norm": 10.678789138793945, "learning_rate": 9.845634682649123e-06, "loss": 0.5788, "step": 1750 }, { "epoch": 0.08857301836208205, "grad_norm": 16.027225494384766, "learning_rate": 9.845436725785747e-06, "loss": 0.732, "step": 1751 }, { "epoch": 0.0886236026101472, "grad_norm": 12.644323348999023, "learning_rate": 9.845238644067169e-06, "loss": 0.4795, "step": 1752 }, { "epoch": 0.08867418685821235, "grad_norm": 8.36074447631836, "learning_rate": 9.845040437498495e-06, "loss": 0.449, "step": 1753 }, { "epoch": 0.08872477110627751, "grad_norm": 9.956668853759766, "learning_rate": 9.844842106084831e-06, "loss": 0.409, "step": 1754 }, { "epoch": 0.08877535535434265, "grad_norm": 9.40852165222168, "learning_rate": 9.84464364983129e-06, "loss": 0.5308, "step": 1755 }, { "epoch": 0.08882593960240781, "grad_norm": 9.071903228759766, "learning_rate": 9.84444506874298e-06, "loss": 0.4743, "step": 1756 }, { "epoch": 0.08887652385047297, "grad_norm": 9.104323387145996, "learning_rate": 9.844246362825025e-06, "loss": 0.5277, "step": 1757 }, { "epoch": 0.08892710809853811, "grad_norm": 7.392155647277832, "learning_rate": 9.844047532082541e-06, "loss": 0.401, "step": 1758 }, { "epoch": 0.08897769234660327, "grad_norm": 7.536657810211182, "learning_rate": 9.843848576520653e-06, "loss": 0.4099, "step": 1759 }, { "epoch": 0.08902827659466842, "grad_norm": 7.142374038696289, "learning_rate": 9.843649496144488e-06, "loss": 0.4542, "step": 1760 }, { "epoch": 0.08907886084273357, "grad_norm": 10.122427940368652, "learning_rate": 9.843450290959173e-06, "loss": 0.5181, "step": 1761 }, { "epoch": 0.08912944509079873, "grad_norm": 12.045560836791992, "learning_rate": 9.843250960969844e-06, "loss": 0.472, "step": 1762 }, { "epoch": 0.08918002933886388, "grad_norm": 7.841624736785889, "learning_rate": 9.843051506181636e-06, "loss": 0.4394, "step": 1763 }, { "epoch": 0.08923061358692903, "grad_norm": 9.60123062133789, "learning_rate": 9.842851926599688e-06, "loss": 0.5156, "step": 1764 }, { "epoch": 0.08928119783499418, "grad_norm": 9.035720825195312, "learning_rate": 9.842652222229143e-06, "loss": 0.462, "step": 1765 }, { "epoch": 0.08933178208305934, "grad_norm": 11.8569917678833, "learning_rate": 9.842452393075148e-06, "loss": 0.5379, "step": 1766 }, { "epoch": 0.0893823663311245, "grad_norm": 6.8375244140625, "learning_rate": 9.84225243914285e-06, "loss": 0.496, "step": 1767 }, { "epoch": 0.08943295057918964, "grad_norm": 15.856664657592773, "learning_rate": 9.842052360437404e-06, "loss": 0.6173, "step": 1768 }, { "epoch": 0.0894835348272548, "grad_norm": 13.28905200958252, "learning_rate": 9.841852156963965e-06, "loss": 0.4567, "step": 1769 }, { "epoch": 0.08953411907531994, "grad_norm": 6.489736080169678, "learning_rate": 9.84165182872769e-06, "loss": 0.5128, "step": 1770 }, { "epoch": 0.0895847033233851, "grad_norm": 9.065711975097656, "learning_rate": 9.84145137573374e-06, "loss": 0.4211, "step": 1771 }, { "epoch": 0.08963528757145026, "grad_norm": 7.836298942565918, "learning_rate": 9.841250797987285e-06, "loss": 0.4803, "step": 1772 }, { "epoch": 0.0896858718195154, "grad_norm": 9.091583251953125, "learning_rate": 9.841050095493488e-06, "loss": 0.561, "step": 1773 }, { "epoch": 0.08973645606758056, "grad_norm": 9.222970008850098, "learning_rate": 9.840849268257525e-06, "loss": 0.4616, "step": 1774 }, { "epoch": 0.0897870403156457, "grad_norm": 5.644326686859131, "learning_rate": 9.840648316284569e-06, "loss": 0.5258, "step": 1775 }, { "epoch": 0.08983762456371086, "grad_norm": 8.530130386352539, "learning_rate": 9.840447239579796e-06, "loss": 0.401, "step": 1776 }, { "epoch": 0.08988820881177602, "grad_norm": 10.711544036865234, "learning_rate": 9.84024603814839e-06, "loss": 0.4373, "step": 1777 }, { "epoch": 0.08993879305984116, "grad_norm": 8.336236000061035, "learning_rate": 9.840044711995536e-06, "loss": 0.5072, "step": 1778 }, { "epoch": 0.08998937730790632, "grad_norm": 11.19351577758789, "learning_rate": 9.83984326112642e-06, "loss": 0.5583, "step": 1779 }, { "epoch": 0.09003996155597147, "grad_norm": 11.919151306152344, "learning_rate": 9.839641685546233e-06, "loss": 0.4853, "step": 1780 }, { "epoch": 0.09009054580403662, "grad_norm": 10.886941909790039, "learning_rate": 9.83943998526017e-06, "loss": 0.4896, "step": 1781 }, { "epoch": 0.09014113005210178, "grad_norm": 12.51097583770752, "learning_rate": 9.839238160273427e-06, "loss": 0.5321, "step": 1782 }, { "epoch": 0.09019171430016693, "grad_norm": 10.092133522033691, "learning_rate": 9.839036210591205e-06, "loss": 0.4455, "step": 1783 }, { "epoch": 0.09024229854823208, "grad_norm": 19.01357078552246, "learning_rate": 9.83883413621871e-06, "loss": 0.6716, "step": 1784 }, { "epoch": 0.09029288279629723, "grad_norm": 11.467944145202637, "learning_rate": 9.838631937161145e-06, "loss": 0.5828, "step": 1785 }, { "epoch": 0.09034346704436239, "grad_norm": 12.450713157653809, "learning_rate": 9.838429613423722e-06, "loss": 0.4688, "step": 1786 }, { "epoch": 0.09039405129242754, "grad_norm": 9.690985679626465, "learning_rate": 9.838227165011656e-06, "loss": 0.443, "step": 1787 }, { "epoch": 0.09044463554049269, "grad_norm": 7.85269021987915, "learning_rate": 9.838024591930161e-06, "loss": 0.5609, "step": 1788 }, { "epoch": 0.09049521978855785, "grad_norm": 15.973312377929688, "learning_rate": 9.83782189418446e-06, "loss": 0.4326, "step": 1789 }, { "epoch": 0.09054580403662299, "grad_norm": 10.375959396362305, "learning_rate": 9.837619071779772e-06, "loss": 0.4445, "step": 1790 }, { "epoch": 0.09059638828468815, "grad_norm": 7.465676307678223, "learning_rate": 9.837416124721326e-06, "loss": 0.426, "step": 1791 }, { "epoch": 0.0906469725327533, "grad_norm": 9.436019897460938, "learning_rate": 9.837213053014352e-06, "loss": 0.5225, "step": 1792 }, { "epoch": 0.09069755678081845, "grad_norm": 8.088079452514648, "learning_rate": 9.83700985666408e-06, "loss": 0.4004, "step": 1793 }, { "epoch": 0.09074814102888361, "grad_norm": 8.512917518615723, "learning_rate": 9.836806535675747e-06, "loss": 0.4943, "step": 1794 }, { "epoch": 0.09079872527694875, "grad_norm": 19.410520553588867, "learning_rate": 9.836603090054593e-06, "loss": 0.5158, "step": 1795 }, { "epoch": 0.09084930952501391, "grad_norm": 6.467487335205078, "learning_rate": 9.83639951980586e-06, "loss": 0.4242, "step": 1796 }, { "epoch": 0.09089989377307907, "grad_norm": 8.10726547241211, "learning_rate": 9.836195824934793e-06, "loss": 0.5418, "step": 1797 }, { "epoch": 0.09095047802114421, "grad_norm": 12.708356857299805, "learning_rate": 9.83599200544664e-06, "loss": 0.5799, "step": 1798 }, { "epoch": 0.09100106226920937, "grad_norm": 11.67446231842041, "learning_rate": 9.835788061346655e-06, "loss": 0.5021, "step": 1799 }, { "epoch": 0.09105164651727451, "grad_norm": 7.9521918296813965, "learning_rate": 9.835583992640092e-06, "loss": 0.4578, "step": 1800 }, { "epoch": 0.09110223076533967, "grad_norm": 10.613757133483887, "learning_rate": 9.83537979933221e-06, "loss": 0.4163, "step": 1801 }, { "epoch": 0.09115281501340483, "grad_norm": 17.171493530273438, "learning_rate": 9.835175481428269e-06, "loss": 0.6397, "step": 1802 }, { "epoch": 0.09120339926146998, "grad_norm": 8.689393043518066, "learning_rate": 9.834971038933535e-06, "loss": 0.4365, "step": 1803 }, { "epoch": 0.09125398350953513, "grad_norm": 8.427176475524902, "learning_rate": 9.834766471853274e-06, "loss": 0.3848, "step": 1804 }, { "epoch": 0.09130456775760028, "grad_norm": 11.767460823059082, "learning_rate": 9.834561780192762e-06, "loss": 0.4161, "step": 1805 }, { "epoch": 0.09135515200566544, "grad_norm": 16.42729949951172, "learning_rate": 9.834356963957268e-06, "loss": 0.5952, "step": 1806 }, { "epoch": 0.0914057362537306, "grad_norm": 7.51518440246582, "learning_rate": 9.834152023152072e-06, "loss": 0.4828, "step": 1807 }, { "epoch": 0.09145632050179574, "grad_norm": 11.630207061767578, "learning_rate": 9.833946957782456e-06, "loss": 0.4459, "step": 1808 }, { "epoch": 0.0915069047498609, "grad_norm": 10.663467407226562, "learning_rate": 9.833741767853702e-06, "loss": 0.5003, "step": 1809 }, { "epoch": 0.09155748899792604, "grad_norm": 8.113441467285156, "learning_rate": 9.8335364533711e-06, "loss": 0.4695, "step": 1810 }, { "epoch": 0.0916080732459912, "grad_norm": 8.572530746459961, "learning_rate": 9.833331014339937e-06, "loss": 0.5455, "step": 1811 }, { "epoch": 0.09165865749405636, "grad_norm": 16.65833282470703, "learning_rate": 9.833125450765507e-06, "loss": 0.5961, "step": 1812 }, { "epoch": 0.0917092417421215, "grad_norm": 12.924875259399414, "learning_rate": 9.83291976265311e-06, "loss": 0.53, "step": 1813 }, { "epoch": 0.09175982599018666, "grad_norm": 9.008926391601562, "learning_rate": 9.832713950008044e-06, "loss": 0.4541, "step": 1814 }, { "epoch": 0.0918104102382518, "grad_norm": 20.675893783569336, "learning_rate": 9.832508012835612e-06, "loss": 0.6478, "step": 1815 }, { "epoch": 0.09186099448631696, "grad_norm": 9.74485969543457, "learning_rate": 9.832301951141118e-06, "loss": 0.5684, "step": 1816 }, { "epoch": 0.09191157873438212, "grad_norm": 18.566997528076172, "learning_rate": 9.83209576492988e-06, "loss": 0.3679, "step": 1817 }, { "epoch": 0.09196216298244726, "grad_norm": 6.978800296783447, "learning_rate": 9.831889454207201e-06, "loss": 0.4932, "step": 1818 }, { "epoch": 0.09201274723051242, "grad_norm": 14.998855590820312, "learning_rate": 9.831683018978405e-06, "loss": 0.5254, "step": 1819 }, { "epoch": 0.09206333147857756, "grad_norm": 12.234270095825195, "learning_rate": 9.831476459248806e-06, "loss": 0.4336, "step": 1820 }, { "epoch": 0.09211391572664272, "grad_norm": 16.19879913330078, "learning_rate": 9.83126977502373e-06, "loss": 0.7488, "step": 1821 }, { "epoch": 0.09216449997470788, "grad_norm": 17.93642807006836, "learning_rate": 9.8310629663085e-06, "loss": 0.7417, "step": 1822 }, { "epoch": 0.09221508422277302, "grad_norm": 8.805078506469727, "learning_rate": 9.830856033108448e-06, "loss": 0.4743, "step": 1823 }, { "epoch": 0.09226566847083818, "grad_norm": 7.6706438064575195, "learning_rate": 9.830648975428903e-06, "loss": 0.474, "step": 1824 }, { "epoch": 0.09231625271890333, "grad_norm": 11.729391098022461, "learning_rate": 9.830441793275202e-06, "loss": 0.4449, "step": 1825 }, { "epoch": 0.09236683696696849, "grad_norm": 8.141900062561035, "learning_rate": 9.830234486652686e-06, "loss": 0.5457, "step": 1826 }, { "epoch": 0.09241742121503364, "grad_norm": 6.955993175506592, "learning_rate": 9.83002705556669e-06, "loss": 0.3714, "step": 1827 }, { "epoch": 0.09246800546309879, "grad_norm": 8.736125946044922, "learning_rate": 9.829819500022565e-06, "loss": 0.5612, "step": 1828 }, { "epoch": 0.09251858971116395, "grad_norm": 12.809720039367676, "learning_rate": 9.829611820025659e-06, "loss": 0.5213, "step": 1829 }, { "epoch": 0.09256917395922909, "grad_norm": 7.759527206420898, "learning_rate": 9.82940401558132e-06, "loss": 0.4084, "step": 1830 }, { "epoch": 0.09261975820729425, "grad_norm": 9.15369987487793, "learning_rate": 9.829196086694906e-06, "loss": 0.5761, "step": 1831 }, { "epoch": 0.0926703424553594, "grad_norm": 12.761154174804688, "learning_rate": 9.82898803337177e-06, "loss": 0.4727, "step": 1832 }, { "epoch": 0.09272092670342455, "grad_norm": 6.302578449249268, "learning_rate": 9.828779855617281e-06, "loss": 0.3989, "step": 1833 }, { "epoch": 0.09277151095148971, "grad_norm": 6.652553558349609, "learning_rate": 9.828571553436796e-06, "loss": 0.4916, "step": 1834 }, { "epoch": 0.09282209519955485, "grad_norm": 13.420143127441406, "learning_rate": 9.828363126835685e-06, "loss": 0.5788, "step": 1835 }, { "epoch": 0.09287267944762001, "grad_norm": 7.323451995849609, "learning_rate": 9.828154575819319e-06, "loss": 0.5502, "step": 1836 }, { "epoch": 0.09292326369568517, "grad_norm": 12.20038890838623, "learning_rate": 9.827945900393073e-06, "loss": 0.5695, "step": 1837 }, { "epoch": 0.09297384794375031, "grad_norm": 14.470549583435059, "learning_rate": 9.82773710056232e-06, "loss": 0.5919, "step": 1838 }, { "epoch": 0.09302443219181547, "grad_norm": 11.5158109664917, "learning_rate": 9.827528176332444e-06, "loss": 0.5057, "step": 1839 }, { "epoch": 0.09307501643988061, "grad_norm": 9.472334861755371, "learning_rate": 9.827319127708826e-06, "loss": 0.4579, "step": 1840 }, { "epoch": 0.09312560068794577, "grad_norm": 9.129237174987793, "learning_rate": 9.827109954696856e-06, "loss": 0.3295, "step": 1841 }, { "epoch": 0.09317618493601093, "grad_norm": 21.527746200561523, "learning_rate": 9.826900657301922e-06, "loss": 0.6508, "step": 1842 }, { "epoch": 0.09322676918407607, "grad_norm": 9.495650291442871, "learning_rate": 9.826691235529416e-06, "loss": 0.4743, "step": 1843 }, { "epoch": 0.09327735343214123, "grad_norm": 10.239481925964355, "learning_rate": 9.826481689384733e-06, "loss": 0.4853, "step": 1844 }, { "epoch": 0.09332793768020638, "grad_norm": 7.751674175262451, "learning_rate": 9.826272018873278e-06, "loss": 0.4663, "step": 1845 }, { "epoch": 0.09337852192827154, "grad_norm": 6.262358665466309, "learning_rate": 9.82606222400045e-06, "loss": 0.4384, "step": 1846 }, { "epoch": 0.0934291061763367, "grad_norm": 10.125840187072754, "learning_rate": 9.825852304771655e-06, "loss": 0.5579, "step": 1847 }, { "epoch": 0.09347969042440184, "grad_norm": 13.567803382873535, "learning_rate": 9.8256422611923e-06, "loss": 0.679, "step": 1848 }, { "epoch": 0.093530274672467, "grad_norm": 4.946768760681152, "learning_rate": 9.825432093267804e-06, "loss": 0.4336, "step": 1849 }, { "epoch": 0.09358085892053214, "grad_norm": 11.216443061828613, "learning_rate": 9.825221801003575e-06, "loss": 0.4335, "step": 1850 }, { "epoch": 0.0936314431685973, "grad_norm": 11.516378402709961, "learning_rate": 9.825011384405037e-06, "loss": 0.5877, "step": 1851 }, { "epoch": 0.09368202741666246, "grad_norm": 7.568297386169434, "learning_rate": 9.824800843477609e-06, "loss": 0.4233, "step": 1852 }, { "epoch": 0.0937326116647276, "grad_norm": 11.702662467956543, "learning_rate": 9.824590178226718e-06, "loss": 0.6176, "step": 1853 }, { "epoch": 0.09378319591279276, "grad_norm": 7.256410121917725, "learning_rate": 9.824379388657791e-06, "loss": 0.4633, "step": 1854 }, { "epoch": 0.0938337801608579, "grad_norm": 9.77432632446289, "learning_rate": 9.82416847477626e-06, "loss": 0.4812, "step": 1855 }, { "epoch": 0.09388436440892306, "grad_norm": 6.498785495758057, "learning_rate": 9.82395743658756e-06, "loss": 0.3896, "step": 1856 }, { "epoch": 0.09393494865698822, "grad_norm": 8.011085510253906, "learning_rate": 9.823746274097127e-06, "loss": 0.4625, "step": 1857 }, { "epoch": 0.09398553290505336, "grad_norm": 14.926204681396484, "learning_rate": 9.823534987310405e-06, "loss": 0.5252, "step": 1858 }, { "epoch": 0.09403611715311852, "grad_norm": 17.5052490234375, "learning_rate": 9.823323576232837e-06, "loss": 0.3625, "step": 1859 }, { "epoch": 0.09408670140118368, "grad_norm": 11.944583892822266, "learning_rate": 9.82311204086987e-06, "loss": 0.4414, "step": 1860 }, { "epoch": 0.09413728564924882, "grad_norm": 15.355494499206543, "learning_rate": 9.822900381226958e-06, "loss": 0.6403, "step": 1861 }, { "epoch": 0.09418786989731398, "grad_norm": 9.136028289794922, "learning_rate": 9.82268859730955e-06, "loss": 0.4976, "step": 1862 }, { "epoch": 0.09423845414537912, "grad_norm": 8.527779579162598, "learning_rate": 9.822476689123108e-06, "loss": 0.4866, "step": 1863 }, { "epoch": 0.09428903839344428, "grad_norm": 7.516037940979004, "learning_rate": 9.822264656673088e-06, "loss": 0.5115, "step": 1864 }, { "epoch": 0.09433962264150944, "grad_norm": 7.250953674316406, "learning_rate": 9.822052499964955e-06, "loss": 0.454, "step": 1865 }, { "epoch": 0.09439020688957458, "grad_norm": 12.017525672912598, "learning_rate": 9.821840219004178e-06, "loss": 0.5076, "step": 1866 }, { "epoch": 0.09444079113763974, "grad_norm": 7.629183292388916, "learning_rate": 9.821627813796223e-06, "loss": 0.4604, "step": 1867 }, { "epoch": 0.09449137538570489, "grad_norm": 7.1090779304504395, "learning_rate": 9.821415284346567e-06, "loss": 0.4032, "step": 1868 }, { "epoch": 0.09454195963377005, "grad_norm": 11.823036193847656, "learning_rate": 9.821202630660684e-06, "loss": 0.4101, "step": 1869 }, { "epoch": 0.0945925438818352, "grad_norm": 13.950338363647461, "learning_rate": 9.820989852744055e-06, "loss": 0.5311, "step": 1870 }, { "epoch": 0.09464312812990035, "grad_norm": 9.446733474731445, "learning_rate": 9.820776950602163e-06, "loss": 0.5211, "step": 1871 }, { "epoch": 0.0946937123779655, "grad_norm": 16.114530563354492, "learning_rate": 9.820563924240491e-06, "loss": 0.4814, "step": 1872 }, { "epoch": 0.09474429662603065, "grad_norm": 6.713069915771484, "learning_rate": 9.82035077366453e-06, "loss": 0.4912, "step": 1873 }, { "epoch": 0.09479488087409581, "grad_norm": 7.393362998962402, "learning_rate": 9.820137498879776e-06, "loss": 0.487, "step": 1874 }, { "epoch": 0.09484546512216097, "grad_norm": 6.875153064727783, "learning_rate": 9.819924099891718e-06, "loss": 0.3503, "step": 1875 }, { "epoch": 0.09489604937022611, "grad_norm": 18.824573516845703, "learning_rate": 9.81971057670586e-06, "loss": 0.6416, "step": 1876 }, { "epoch": 0.09494663361829127, "grad_norm": 18.337360382080078, "learning_rate": 9.819496929327701e-06, "loss": 0.626, "step": 1877 }, { "epoch": 0.09499721786635641, "grad_norm": 7.972190856933594, "learning_rate": 9.819283157762747e-06, "loss": 0.3977, "step": 1878 }, { "epoch": 0.09504780211442157, "grad_norm": 10.513592720031738, "learning_rate": 9.819069262016506e-06, "loss": 0.5348, "step": 1879 }, { "epoch": 0.09509838636248673, "grad_norm": 10.091261863708496, "learning_rate": 9.818855242094492e-06, "loss": 0.4871, "step": 1880 }, { "epoch": 0.09514897061055187, "grad_norm": 13.487826347351074, "learning_rate": 9.818641098002215e-06, "loss": 0.4714, "step": 1881 }, { "epoch": 0.09519955485861703, "grad_norm": 8.16492748260498, "learning_rate": 9.818426829745198e-06, "loss": 0.4828, "step": 1882 }, { "epoch": 0.09525013910668217, "grad_norm": 13.980719566345215, "learning_rate": 9.81821243732896e-06, "loss": 0.5063, "step": 1883 }, { "epoch": 0.09530072335474733, "grad_norm": 9.641166687011719, "learning_rate": 9.817997920759025e-06, "loss": 0.5997, "step": 1884 }, { "epoch": 0.09535130760281249, "grad_norm": 10.946761131286621, "learning_rate": 9.817783280040922e-06, "loss": 0.6523, "step": 1885 }, { "epoch": 0.09540189185087763, "grad_norm": 8.383543968200684, "learning_rate": 9.817568515180178e-06, "loss": 0.5262, "step": 1886 }, { "epoch": 0.09545247609894279, "grad_norm": 8.403871536254883, "learning_rate": 9.817353626182332e-06, "loss": 0.4541, "step": 1887 }, { "epoch": 0.09550306034700794, "grad_norm": 7.123899936676025, "learning_rate": 9.817138613052918e-06, "loss": 0.4884, "step": 1888 }, { "epoch": 0.0955536445950731, "grad_norm": 13.346063613891602, "learning_rate": 9.816923475797479e-06, "loss": 0.5839, "step": 1889 }, { "epoch": 0.09560422884313825, "grad_norm": 9.789551734924316, "learning_rate": 9.816708214421555e-06, "loss": 0.4632, "step": 1890 }, { "epoch": 0.0956548130912034, "grad_norm": 13.275967597961426, "learning_rate": 9.816492828930694e-06, "loss": 0.5385, "step": 1891 }, { "epoch": 0.09570539733926856, "grad_norm": 16.57866096496582, "learning_rate": 9.816277319330449e-06, "loss": 0.5978, "step": 1892 }, { "epoch": 0.0957559815873337, "grad_norm": 14.067394256591797, "learning_rate": 9.81606168562637e-06, "loss": 0.4032, "step": 1893 }, { "epoch": 0.09580656583539886, "grad_norm": 9.30111026763916, "learning_rate": 9.815845927824012e-06, "loss": 0.443, "step": 1894 }, { "epoch": 0.09585715008346402, "grad_norm": 8.314431190490723, "learning_rate": 9.815630045928939e-06, "loss": 0.4762, "step": 1895 }, { "epoch": 0.09590773433152916, "grad_norm": 5.977445602416992, "learning_rate": 9.815414039946708e-06, "loss": 0.4908, "step": 1896 }, { "epoch": 0.09595831857959432, "grad_norm": 7.342129707336426, "learning_rate": 9.815197909882893e-06, "loss": 0.4293, "step": 1897 }, { "epoch": 0.09600890282765946, "grad_norm": 9.834312438964844, "learning_rate": 9.814981655743056e-06, "loss": 0.3624, "step": 1898 }, { "epoch": 0.09605948707572462, "grad_norm": 12.090872764587402, "learning_rate": 9.814765277532772e-06, "loss": 0.5221, "step": 1899 }, { "epoch": 0.09611007132378978, "grad_norm": 10.075422286987305, "learning_rate": 9.814548775257617e-06, "loss": 0.4394, "step": 1900 }, { "epoch": 0.09616065557185492, "grad_norm": 9.293360710144043, "learning_rate": 9.814332148923167e-06, "loss": 0.5746, "step": 1901 }, { "epoch": 0.09621123981992008, "grad_norm": 8.640926361083984, "learning_rate": 9.81411539853501e-06, "loss": 0.5137, "step": 1902 }, { "epoch": 0.09626182406798522, "grad_norm": 15.158053398132324, "learning_rate": 9.813898524098725e-06, "loss": 0.4789, "step": 1903 }, { "epoch": 0.09631240831605038, "grad_norm": 10.41715145111084, "learning_rate": 9.8136815256199e-06, "loss": 0.5719, "step": 1904 }, { "epoch": 0.09636299256411554, "grad_norm": 19.22887420654297, "learning_rate": 9.813464403104133e-06, "loss": 0.7261, "step": 1905 }, { "epoch": 0.09641357681218068, "grad_norm": 8.171589851379395, "learning_rate": 9.813247156557013e-06, "loss": 0.5455, "step": 1906 }, { "epoch": 0.09646416106024584, "grad_norm": 16.281505584716797, "learning_rate": 9.81302978598414e-06, "loss": 0.4221, "step": 1907 }, { "epoch": 0.09651474530831099, "grad_norm": 10.637383460998535, "learning_rate": 9.812812291391114e-06, "loss": 0.4736, "step": 1908 }, { "epoch": 0.09656532955637614, "grad_norm": 20.53169059753418, "learning_rate": 9.812594672783542e-06, "loss": 0.5901, "step": 1909 }, { "epoch": 0.0966159138044413, "grad_norm": 9.325838088989258, "learning_rate": 9.81237693016703e-06, "loss": 0.6011, "step": 1910 }, { "epoch": 0.09666649805250645, "grad_norm": 10.107133865356445, "learning_rate": 9.812159063547185e-06, "loss": 0.4121, "step": 1911 }, { "epoch": 0.0967170823005716, "grad_norm": 7.9617719650268555, "learning_rate": 9.811941072929627e-06, "loss": 0.4765, "step": 1912 }, { "epoch": 0.09676766654863675, "grad_norm": 18.45640754699707, "learning_rate": 9.81172295831997e-06, "loss": 0.5618, "step": 1913 }, { "epoch": 0.09681825079670191, "grad_norm": 8.344522476196289, "learning_rate": 9.811504719723833e-06, "loss": 0.3511, "step": 1914 }, { "epoch": 0.09686883504476707, "grad_norm": 11.672009468078613, "learning_rate": 9.811286357146842e-06, "loss": 0.6123, "step": 1915 }, { "epoch": 0.09691941929283221, "grad_norm": 12.63421630859375, "learning_rate": 9.811067870594624e-06, "loss": 0.5828, "step": 1916 }, { "epoch": 0.09697000354089737, "grad_norm": 18.679584503173828, "learning_rate": 9.810849260072806e-06, "loss": 0.5845, "step": 1917 }, { "epoch": 0.09702058778896251, "grad_norm": 13.420619010925293, "learning_rate": 9.810630525587023e-06, "loss": 0.558, "step": 1918 }, { "epoch": 0.09707117203702767, "grad_norm": 6.636417388916016, "learning_rate": 9.81041166714291e-06, "loss": 0.4616, "step": 1919 }, { "epoch": 0.09712175628509283, "grad_norm": 10.630330085754395, "learning_rate": 9.81019268474611e-06, "loss": 0.4994, "step": 1920 }, { "epoch": 0.09717234053315797, "grad_norm": 26.324792861938477, "learning_rate": 9.80997357840226e-06, "loss": 0.6249, "step": 1921 }, { "epoch": 0.09722292478122313, "grad_norm": 12.397912979125977, "learning_rate": 9.80975434811701e-06, "loss": 0.6176, "step": 1922 }, { "epoch": 0.09727350902928827, "grad_norm": 11.902633666992188, "learning_rate": 9.80953499389601e-06, "loss": 0.5297, "step": 1923 }, { "epoch": 0.09732409327735343, "grad_norm": 7.6537370681762695, "learning_rate": 9.809315515744908e-06, "loss": 0.4654, "step": 1924 }, { "epoch": 0.09737467752541859, "grad_norm": 10.08556842803955, "learning_rate": 9.809095913669363e-06, "loss": 0.5059, "step": 1925 }, { "epoch": 0.09742526177348373, "grad_norm": 5.816115856170654, "learning_rate": 9.808876187675031e-06, "loss": 0.3848, "step": 1926 }, { "epoch": 0.09747584602154889, "grad_norm": 10.92093276977539, "learning_rate": 9.808656337767575e-06, "loss": 0.5342, "step": 1927 }, { "epoch": 0.09752643026961404, "grad_norm": 15.470698356628418, "learning_rate": 9.808436363952662e-06, "loss": 0.5374, "step": 1928 }, { "epoch": 0.0975770145176792, "grad_norm": 11.789746284484863, "learning_rate": 9.808216266235956e-06, "loss": 0.5854, "step": 1929 }, { "epoch": 0.09762759876574435, "grad_norm": 8.558606147766113, "learning_rate": 9.807996044623131e-06, "loss": 0.5262, "step": 1930 }, { "epoch": 0.0976781830138095, "grad_norm": 8.618255615234375, "learning_rate": 9.80777569911986e-06, "loss": 0.4373, "step": 1931 }, { "epoch": 0.09772876726187466, "grad_norm": 13.053874015808105, "learning_rate": 9.807555229731825e-06, "loss": 0.5438, "step": 1932 }, { "epoch": 0.0977793515099398, "grad_norm": 5.514989852905273, "learning_rate": 9.807334636464703e-06, "loss": 0.4222, "step": 1933 }, { "epoch": 0.09782993575800496, "grad_norm": 10.154090881347656, "learning_rate": 9.807113919324177e-06, "loss": 0.4063, "step": 1934 }, { "epoch": 0.09788052000607012, "grad_norm": 21.011388778686523, "learning_rate": 9.806893078315938e-06, "loss": 0.506, "step": 1935 }, { "epoch": 0.09793110425413526, "grad_norm": 8.636472702026367, "learning_rate": 9.806672113445674e-06, "loss": 0.5174, "step": 1936 }, { "epoch": 0.09798168850220042, "grad_norm": 5.753244400024414, "learning_rate": 9.806451024719082e-06, "loss": 0.5026, "step": 1937 }, { "epoch": 0.09803227275026556, "grad_norm": 15.261526107788086, "learning_rate": 9.806229812141855e-06, "loss": 0.513, "step": 1938 }, { "epoch": 0.09808285699833072, "grad_norm": 5.101361274719238, "learning_rate": 9.806008475719694e-06, "loss": 0.4218, "step": 1939 }, { "epoch": 0.09813344124639588, "grad_norm": 11.018599510192871, "learning_rate": 9.805787015458303e-06, "loss": 0.4859, "step": 1940 }, { "epoch": 0.09818402549446102, "grad_norm": 10.02345085144043, "learning_rate": 9.80556543136339e-06, "loss": 0.4533, "step": 1941 }, { "epoch": 0.09823460974252618, "grad_norm": 16.28225326538086, "learning_rate": 9.80534372344066e-06, "loss": 0.6178, "step": 1942 }, { "epoch": 0.09828519399059132, "grad_norm": 8.073914527893066, "learning_rate": 9.805121891695834e-06, "loss": 0.3987, "step": 1943 }, { "epoch": 0.09833577823865648, "grad_norm": 9.568089485168457, "learning_rate": 9.80489993613462e-06, "loss": 0.4569, "step": 1944 }, { "epoch": 0.09838636248672164, "grad_norm": 25.739356994628906, "learning_rate": 9.804677856762742e-06, "loss": 0.5914, "step": 1945 }, { "epoch": 0.09843694673478678, "grad_norm": 12.744311332702637, "learning_rate": 9.804455653585919e-06, "loss": 0.5171, "step": 1946 }, { "epoch": 0.09848753098285194, "grad_norm": 14.650885581970215, "learning_rate": 9.80423332660988e-06, "loss": 0.6043, "step": 1947 }, { "epoch": 0.09853811523091709, "grad_norm": 23.180339813232422, "learning_rate": 9.80401087584035e-06, "loss": 0.4485, "step": 1948 }, { "epoch": 0.09858869947898224, "grad_norm": 13.059709548950195, "learning_rate": 9.803788301283066e-06, "loss": 0.4075, "step": 1949 }, { "epoch": 0.0986392837270474, "grad_norm": 13.865548133850098, "learning_rate": 9.803565602943759e-06, "loss": 0.4598, "step": 1950 }, { "epoch": 0.09868986797511255, "grad_norm": 23.39909553527832, "learning_rate": 9.80334278082817e-06, "loss": 0.5096, "step": 1951 }, { "epoch": 0.0987404522231777, "grad_norm": 9.189618110656738, "learning_rate": 9.80311983494204e-06, "loss": 0.5111, "step": 1952 }, { "epoch": 0.09879103647124285, "grad_norm": 17.602615356445312, "learning_rate": 9.802896765291111e-06, "loss": 0.4604, "step": 1953 }, { "epoch": 0.09884162071930801, "grad_norm": 9.021650314331055, "learning_rate": 9.802673571881135e-06, "loss": 0.4943, "step": 1954 }, { "epoch": 0.09889220496737317, "grad_norm": 15.464653015136719, "learning_rate": 9.802450254717862e-06, "loss": 0.6022, "step": 1955 }, { "epoch": 0.09894278921543831, "grad_norm": 12.291083335876465, "learning_rate": 9.802226813807045e-06, "loss": 0.5434, "step": 1956 }, { "epoch": 0.09899337346350347, "grad_norm": 7.825693130493164, "learning_rate": 9.802003249154442e-06, "loss": 0.4265, "step": 1957 }, { "epoch": 0.09904395771156861, "grad_norm": 5.148998260498047, "learning_rate": 9.801779560765814e-06, "loss": 0.4378, "step": 1958 }, { "epoch": 0.09909454195963377, "grad_norm": 9.082344055175781, "learning_rate": 9.801555748646924e-06, "loss": 0.5409, "step": 1959 }, { "epoch": 0.09914512620769893, "grad_norm": 12.716058731079102, "learning_rate": 9.801331812803543e-06, "loss": 0.4281, "step": 1960 }, { "epoch": 0.09919571045576407, "grad_norm": 7.306156635284424, "learning_rate": 9.801107753241435e-06, "loss": 0.3633, "step": 1961 }, { "epoch": 0.09924629470382923, "grad_norm": 7.2243452072143555, "learning_rate": 9.80088356996638e-06, "loss": 0.4233, "step": 1962 }, { "epoch": 0.09929687895189437, "grad_norm": 9.270750999450684, "learning_rate": 9.800659262984149e-06, "loss": 0.4304, "step": 1963 }, { "epoch": 0.09934746319995953, "grad_norm": 17.503276824951172, "learning_rate": 9.800434832300525e-06, "loss": 0.5861, "step": 1964 }, { "epoch": 0.09939804744802469, "grad_norm": 15.768264770507812, "learning_rate": 9.80021027792129e-06, "loss": 0.5573, "step": 1965 }, { "epoch": 0.09944863169608983, "grad_norm": 8.52269172668457, "learning_rate": 9.799985599852231e-06, "loss": 0.4167, "step": 1966 }, { "epoch": 0.09949921594415499, "grad_norm": 5.8297576904296875, "learning_rate": 9.799760798099137e-06, "loss": 0.307, "step": 1967 }, { "epoch": 0.09954980019222014, "grad_norm": 9.199021339416504, "learning_rate": 9.7995358726678e-06, "loss": 0.519, "step": 1968 }, { "epoch": 0.0996003844402853, "grad_norm": 5.8656535148620605, "learning_rate": 9.799310823564018e-06, "loss": 0.5071, "step": 1969 }, { "epoch": 0.09965096868835045, "grad_norm": 10.240312576293945, "learning_rate": 9.799085650793587e-06, "loss": 0.4109, "step": 1970 }, { "epoch": 0.0997015529364156, "grad_norm": 9.36715030670166, "learning_rate": 9.798860354362311e-06, "loss": 0.3858, "step": 1971 }, { "epoch": 0.09975213718448075, "grad_norm": 12.606526374816895, "learning_rate": 9.798634934275995e-06, "loss": 0.3874, "step": 1972 }, { "epoch": 0.0998027214325459, "grad_norm": 7.826398849487305, "learning_rate": 9.798409390540446e-06, "loss": 0.4391, "step": 1973 }, { "epoch": 0.09985330568061106, "grad_norm": 13.2358980178833, "learning_rate": 9.79818372316148e-06, "loss": 0.4971, "step": 1974 }, { "epoch": 0.09990388992867621, "grad_norm": 10.186567306518555, "learning_rate": 9.797957932144906e-06, "loss": 0.4292, "step": 1975 }, { "epoch": 0.09995447417674136, "grad_norm": 10.508633613586426, "learning_rate": 9.797732017496547e-06, "loss": 0.514, "step": 1976 }, { "epoch": 0.10000505842480652, "grad_norm": 7.5608625411987305, "learning_rate": 9.797505979222221e-06, "loss": 0.403, "step": 1977 }, { "epoch": 0.10000505842480652, "eval_loss": 0.4782736897468567, "eval_runtime": 34.9563, "eval_samples_per_second": 45.714, "eval_steps_per_second": 5.721, "step": 1977 }, { "epoch": 0.10005564267287166, "grad_norm": 4.179031848907471, "learning_rate": 9.797279817327754e-06, "loss": 0.4148, "step": 1978 }, { "epoch": 0.10010622692093682, "grad_norm": 12.942991256713867, "learning_rate": 9.797053531818975e-06, "loss": 0.5063, "step": 1979 }, { "epoch": 0.10015681116900198, "grad_norm": 6.536281108856201, "learning_rate": 9.79682712270171e-06, "loss": 0.507, "step": 1980 }, { "epoch": 0.10020739541706712, "grad_norm": 10.175827026367188, "learning_rate": 9.796600589981799e-06, "loss": 0.4488, "step": 1981 }, { "epoch": 0.10025797966513228, "grad_norm": 11.51638412475586, "learning_rate": 9.796373933665076e-06, "loss": 0.2353, "step": 1982 }, { "epoch": 0.10030856391319742, "grad_norm": 9.153254508972168, "learning_rate": 9.79614715375738e-06, "loss": 0.4324, "step": 1983 }, { "epoch": 0.10035914816126258, "grad_norm": 9.618071556091309, "learning_rate": 9.79592025026456e-06, "loss": 0.5214, "step": 1984 }, { "epoch": 0.10040973240932774, "grad_norm": 13.452919006347656, "learning_rate": 9.795693223192457e-06, "loss": 0.5776, "step": 1985 }, { "epoch": 0.10046031665739288, "grad_norm": 14.024735450744629, "learning_rate": 9.795466072546923e-06, "loss": 0.4833, "step": 1986 }, { "epoch": 0.10051090090545804, "grad_norm": 21.06290054321289, "learning_rate": 9.795238798333812e-06, "loss": 0.4839, "step": 1987 }, { "epoch": 0.10056148515352319, "grad_norm": 14.408293724060059, "learning_rate": 9.795011400558981e-06, "loss": 0.5258, "step": 1988 }, { "epoch": 0.10061206940158834, "grad_norm": 9.025083541870117, "learning_rate": 9.794783879228287e-06, "loss": 0.4442, "step": 1989 }, { "epoch": 0.1006626536496535, "grad_norm": 17.822019577026367, "learning_rate": 9.794556234347591e-06, "loss": 0.6597, "step": 1990 }, { "epoch": 0.10071323789771865, "grad_norm": 18.939096450805664, "learning_rate": 9.794328465922765e-06, "loss": 0.6239, "step": 1991 }, { "epoch": 0.1007638221457838, "grad_norm": 27.593351364135742, "learning_rate": 9.794100573959676e-06, "loss": 0.5884, "step": 1992 }, { "epoch": 0.10081440639384895, "grad_norm": 7.188124179840088, "learning_rate": 9.79387255846419e-06, "loss": 0.5295, "step": 1993 }, { "epoch": 0.1008649906419141, "grad_norm": 7.564386367797852, "learning_rate": 9.79364441944219e-06, "loss": 0.5373, "step": 1994 }, { "epoch": 0.10091557488997926, "grad_norm": 8.787016868591309, "learning_rate": 9.793416156899552e-06, "loss": 0.4996, "step": 1995 }, { "epoch": 0.10096615913804441, "grad_norm": 7.182962894439697, "learning_rate": 9.793187770842158e-06, "loss": 0.5084, "step": 1996 }, { "epoch": 0.10101674338610957, "grad_norm": 6.584283351898193, "learning_rate": 9.792959261275892e-06, "loss": 0.5255, "step": 1997 }, { "epoch": 0.10106732763417471, "grad_norm": 7.576235771179199, "learning_rate": 9.792730628206643e-06, "loss": 0.4814, "step": 1998 }, { "epoch": 0.10111791188223987, "grad_norm": 8.250005722045898, "learning_rate": 9.7925018716403e-06, "loss": 0.45, "step": 1999 }, { "epoch": 0.10116849613030503, "grad_norm": 7.413142681121826, "learning_rate": 9.792272991582763e-06, "loss": 0.5218, "step": 2000 }, { "epoch": 0.10121908037837017, "grad_norm": 9.411563873291016, "learning_rate": 9.792043988039924e-06, "loss": 0.4538, "step": 2001 }, { "epoch": 0.10126966462643533, "grad_norm": 6.926442623138428, "learning_rate": 9.791814861017688e-06, "loss": 0.4484, "step": 2002 }, { "epoch": 0.10132024887450047, "grad_norm": 7.648702621459961, "learning_rate": 9.791585610521955e-06, "loss": 0.5049, "step": 2003 }, { "epoch": 0.10137083312256563, "grad_norm": 9.282817840576172, "learning_rate": 9.791356236558635e-06, "loss": 0.3504, "step": 2004 }, { "epoch": 0.10142141737063079, "grad_norm": 10.756082534790039, "learning_rate": 9.791126739133638e-06, "loss": 0.406, "step": 2005 }, { "epoch": 0.10147200161869593, "grad_norm": 10.491487503051758, "learning_rate": 9.790897118252877e-06, "loss": 0.4086, "step": 2006 }, { "epoch": 0.10152258586676109, "grad_norm": 9.772181510925293, "learning_rate": 9.790667373922272e-06, "loss": 0.5, "step": 2007 }, { "epoch": 0.10157317011482625, "grad_norm": 16.61635971069336, "learning_rate": 9.790437506147738e-06, "loss": 0.6865, "step": 2008 }, { "epoch": 0.1016237543628914, "grad_norm": 12.660127639770508, "learning_rate": 9.790207514935199e-06, "loss": 0.487, "step": 2009 }, { "epoch": 0.10167433861095655, "grad_norm": 11.419099807739258, "learning_rate": 9.789977400290584e-06, "loss": 0.4158, "step": 2010 }, { "epoch": 0.1017249228590217, "grad_norm": 25.886985778808594, "learning_rate": 9.789747162219822e-06, "loss": 0.4824, "step": 2011 }, { "epoch": 0.10177550710708685, "grad_norm": 11.7368745803833, "learning_rate": 9.789516800728843e-06, "loss": 0.5245, "step": 2012 }, { "epoch": 0.10182609135515201, "grad_norm": 22.71005630493164, "learning_rate": 9.789286315823585e-06, "loss": 0.6043, "step": 2013 }, { "epoch": 0.10187667560321716, "grad_norm": 10.579380989074707, "learning_rate": 9.789055707509988e-06, "loss": 0.5185, "step": 2014 }, { "epoch": 0.10192725985128231, "grad_norm": 11.895326614379883, "learning_rate": 9.788824975793992e-06, "loss": 0.4856, "step": 2015 }, { "epoch": 0.10197784409934746, "grad_norm": 13.739741325378418, "learning_rate": 9.788594120681544e-06, "loss": 0.565, "step": 2016 }, { "epoch": 0.10202842834741262, "grad_norm": 11.931597709655762, "learning_rate": 9.788363142178591e-06, "loss": 0.4513, "step": 2017 }, { "epoch": 0.10207901259547777, "grad_norm": 8.315810203552246, "learning_rate": 9.788132040291085e-06, "loss": 0.538, "step": 2018 }, { "epoch": 0.10212959684354292, "grad_norm": 11.949161529541016, "learning_rate": 9.787900815024983e-06, "loss": 0.4478, "step": 2019 }, { "epoch": 0.10218018109160808, "grad_norm": 5.031383037567139, "learning_rate": 9.787669466386242e-06, "loss": 0.4536, "step": 2020 }, { "epoch": 0.10223076533967322, "grad_norm": 9.111480712890625, "learning_rate": 9.787437994380822e-06, "loss": 0.5285, "step": 2021 }, { "epoch": 0.10228134958773838, "grad_norm": 7.069510459899902, "learning_rate": 9.787206399014689e-06, "loss": 0.3427, "step": 2022 }, { "epoch": 0.10233193383580354, "grad_norm": 33.940853118896484, "learning_rate": 9.786974680293809e-06, "loss": 0.5934, "step": 2023 }, { "epoch": 0.10238251808386868, "grad_norm": 9.826305389404297, "learning_rate": 9.786742838224154e-06, "loss": 0.4109, "step": 2024 }, { "epoch": 0.10243310233193384, "grad_norm": 12.540133476257324, "learning_rate": 9.786510872811698e-06, "loss": 0.5526, "step": 2025 }, { "epoch": 0.10248368657999898, "grad_norm": 12.441570281982422, "learning_rate": 9.78627878406242e-06, "loss": 0.5188, "step": 2026 }, { "epoch": 0.10253427082806414, "grad_norm": 10.976491928100586, "learning_rate": 9.786046571982298e-06, "loss": 0.4529, "step": 2027 }, { "epoch": 0.1025848550761293, "grad_norm": 22.982383728027344, "learning_rate": 9.785814236577316e-06, "loss": 0.7552, "step": 2028 }, { "epoch": 0.10263543932419444, "grad_norm": 12.941311836242676, "learning_rate": 9.785581777853459e-06, "loss": 0.5798, "step": 2029 }, { "epoch": 0.1026860235722596, "grad_norm": 11.35175895690918, "learning_rate": 9.78534919581672e-06, "loss": 0.6339, "step": 2030 }, { "epoch": 0.10273660782032475, "grad_norm": 7.063699722290039, "learning_rate": 9.78511649047309e-06, "loss": 0.3756, "step": 2031 }, { "epoch": 0.1027871920683899, "grad_norm": 10.18331241607666, "learning_rate": 9.784883661828569e-06, "loss": 0.5475, "step": 2032 }, { "epoch": 0.10283777631645506, "grad_norm": 16.931882858276367, "learning_rate": 9.78465070988915e-06, "loss": 0.6627, "step": 2033 }, { "epoch": 0.1028883605645202, "grad_norm": 10.686186790466309, "learning_rate": 9.784417634660839e-06, "loss": 0.5326, "step": 2034 }, { "epoch": 0.10293894481258536, "grad_norm": 11.885024070739746, "learning_rate": 9.784184436149645e-06, "loss": 0.6078, "step": 2035 }, { "epoch": 0.10298952906065051, "grad_norm": 7.799513816833496, "learning_rate": 9.783951114361571e-06, "loss": 0.4755, "step": 2036 }, { "epoch": 0.10304011330871567, "grad_norm": 11.492841720581055, "learning_rate": 9.783717669302632e-06, "loss": 0.4997, "step": 2037 }, { "epoch": 0.10309069755678082, "grad_norm": 11.384532928466797, "learning_rate": 9.783484100978845e-06, "loss": 0.43, "step": 2038 }, { "epoch": 0.10314128180484597, "grad_norm": 13.204156875610352, "learning_rate": 9.783250409396226e-06, "loss": 0.5537, "step": 2039 }, { "epoch": 0.10319186605291113, "grad_norm": 10.848970413208008, "learning_rate": 9.783016594560798e-06, "loss": 0.3012, "step": 2040 }, { "epoch": 0.10324245030097627, "grad_norm": 9.188164710998535, "learning_rate": 9.782782656478582e-06, "loss": 0.4228, "step": 2041 }, { "epoch": 0.10329303454904143, "grad_norm": 16.3311710357666, "learning_rate": 9.782548595155613e-06, "loss": 0.4904, "step": 2042 }, { "epoch": 0.10334361879710659, "grad_norm": 19.582242965698242, "learning_rate": 9.782314410597918e-06, "loss": 0.5494, "step": 2043 }, { "epoch": 0.10339420304517173, "grad_norm": 6.8755388259887695, "learning_rate": 9.78208010281153e-06, "loss": 0.3947, "step": 2044 }, { "epoch": 0.10344478729323689, "grad_norm": 10.601800918579102, "learning_rate": 9.78184567180249e-06, "loss": 0.3648, "step": 2045 }, { "epoch": 0.10349537154130203, "grad_norm": 8.661100387573242, "learning_rate": 9.781611117576836e-06, "loss": 0.4812, "step": 2046 }, { "epoch": 0.10354595578936719, "grad_norm": 11.129374504089355, "learning_rate": 9.781376440140614e-06, "loss": 0.5206, "step": 2047 }, { "epoch": 0.10359654003743235, "grad_norm": 8.393204689025879, "learning_rate": 9.781141639499869e-06, "loss": 0.3699, "step": 2048 }, { "epoch": 0.1036471242854975, "grad_norm": 8.791845321655273, "learning_rate": 9.780906715660652e-06, "loss": 0.4711, "step": 2049 }, { "epoch": 0.10369770853356265, "grad_norm": 9.690587043762207, "learning_rate": 9.780671668629018e-06, "loss": 0.3537, "step": 2050 }, { "epoch": 0.1037482927816278, "grad_norm": 28.83854103088379, "learning_rate": 9.780436498411024e-06, "loss": 0.5273, "step": 2051 }, { "epoch": 0.10379887702969295, "grad_norm": 12.620728492736816, "learning_rate": 9.780201205012725e-06, "loss": 0.5074, "step": 2052 }, { "epoch": 0.10384946127775811, "grad_norm": 15.638106346130371, "learning_rate": 9.779965788440187e-06, "loss": 0.4943, "step": 2053 }, { "epoch": 0.10390004552582326, "grad_norm": 15.452524185180664, "learning_rate": 9.779730248699477e-06, "loss": 0.3594, "step": 2054 }, { "epoch": 0.10395062977388841, "grad_norm": 11.115311622619629, "learning_rate": 9.779494585796665e-06, "loss": 0.5578, "step": 2055 }, { "epoch": 0.10400121402195356, "grad_norm": 12.903158187866211, "learning_rate": 9.77925879973782e-06, "loss": 0.5907, "step": 2056 }, { "epoch": 0.10405179827001872, "grad_norm": 12.779358863830566, "learning_rate": 9.779022890529022e-06, "loss": 0.5327, "step": 2057 }, { "epoch": 0.10410238251808387, "grad_norm": 16.93378257751465, "learning_rate": 9.778786858176345e-06, "loss": 0.52, "step": 2058 }, { "epoch": 0.10415296676614902, "grad_norm": 18.87198257446289, "learning_rate": 9.778550702685875e-06, "loss": 0.5405, "step": 2059 }, { "epoch": 0.10420355101421418, "grad_norm": 10.48476505279541, "learning_rate": 9.778314424063693e-06, "loss": 0.5732, "step": 2060 }, { "epoch": 0.10425413526227932, "grad_norm": 9.487783432006836, "learning_rate": 9.778078022315891e-06, "loss": 0.5729, "step": 2061 }, { "epoch": 0.10430471951034448, "grad_norm": 7.3859357833862305, "learning_rate": 9.77784149744856e-06, "loss": 0.4351, "step": 2062 }, { "epoch": 0.10435530375840964, "grad_norm": 8.97900390625, "learning_rate": 9.777604849467795e-06, "loss": 0.4775, "step": 2063 }, { "epoch": 0.10440588800647478, "grad_norm": 12.68799114227295, "learning_rate": 9.77736807837969e-06, "loss": 0.379, "step": 2064 }, { "epoch": 0.10445647225453994, "grad_norm": 11.024117469787598, "learning_rate": 9.777131184190352e-06, "loss": 0.4701, "step": 2065 }, { "epoch": 0.10450705650260508, "grad_norm": 7.723320484161377, "learning_rate": 9.776894166905879e-06, "loss": 0.3615, "step": 2066 }, { "epoch": 0.10455764075067024, "grad_norm": 7.8604960441589355, "learning_rate": 9.776657026532384e-06, "loss": 0.4762, "step": 2067 }, { "epoch": 0.1046082249987354, "grad_norm": 8.780795097351074, "learning_rate": 9.776419763075974e-06, "loss": 0.4065, "step": 2068 }, { "epoch": 0.10465880924680054, "grad_norm": 10.161285400390625, "learning_rate": 9.776182376542763e-06, "loss": 0.4198, "step": 2069 }, { "epoch": 0.1047093934948657, "grad_norm": 9.013846397399902, "learning_rate": 9.77594486693887e-06, "loss": 0.4508, "step": 2070 }, { "epoch": 0.10475997774293085, "grad_norm": 16.26988983154297, "learning_rate": 9.775707234270414e-06, "loss": 0.5402, "step": 2071 }, { "epoch": 0.104810561990996, "grad_norm": 9.659322738647461, "learning_rate": 9.775469478543515e-06, "loss": 0.4997, "step": 2072 }, { "epoch": 0.10486114623906116, "grad_norm": 11.831114768981934, "learning_rate": 9.775231599764303e-06, "loss": 0.3871, "step": 2073 }, { "epoch": 0.1049117304871263, "grad_norm": 15.0466947555542, "learning_rate": 9.774993597938909e-06, "loss": 0.5618, "step": 2074 }, { "epoch": 0.10496231473519146, "grad_norm": 7.558920860290527, "learning_rate": 9.774755473073461e-06, "loss": 0.5178, "step": 2075 }, { "epoch": 0.10501289898325661, "grad_norm": 18.449052810668945, "learning_rate": 9.774517225174098e-06, "loss": 0.4328, "step": 2076 }, { "epoch": 0.10506348323132177, "grad_norm": 14.364030838012695, "learning_rate": 9.77427885424696e-06, "loss": 0.547, "step": 2077 }, { "epoch": 0.10511406747938692, "grad_norm": 19.8599796295166, "learning_rate": 9.774040360298188e-06, "loss": 0.6657, "step": 2078 }, { "epoch": 0.10516465172745207, "grad_norm": 17.147432327270508, "learning_rate": 9.773801743333925e-06, "loss": 0.5076, "step": 2079 }, { "epoch": 0.10521523597551723, "grad_norm": 17.50420379638672, "learning_rate": 9.773563003360323e-06, "loss": 0.4962, "step": 2080 }, { "epoch": 0.10526582022358237, "grad_norm": 13.528446197509766, "learning_rate": 9.773324140383531e-06, "loss": 0.4895, "step": 2081 }, { "epoch": 0.10531640447164753, "grad_norm": 10.219524383544922, "learning_rate": 9.773085154409707e-06, "loss": 0.5421, "step": 2082 }, { "epoch": 0.10536698871971269, "grad_norm": 7.6654887199401855, "learning_rate": 9.772846045445006e-06, "loss": 0.4909, "step": 2083 }, { "epoch": 0.10541757296777783, "grad_norm": 8.807527542114258, "learning_rate": 9.772606813495592e-06, "loss": 0.4678, "step": 2084 }, { "epoch": 0.10546815721584299, "grad_norm": 9.428638458251953, "learning_rate": 9.772367458567628e-06, "loss": 0.4678, "step": 2085 }, { "epoch": 0.10551874146390813, "grad_norm": 8.962078094482422, "learning_rate": 9.77212798066728e-06, "loss": 0.5261, "step": 2086 }, { "epoch": 0.10556932571197329, "grad_norm": 18.17207145690918, "learning_rate": 9.771888379800723e-06, "loss": 0.6195, "step": 2087 }, { "epoch": 0.10561990996003845, "grad_norm": 17.05950355529785, "learning_rate": 9.771648655974128e-06, "loss": 0.577, "step": 2088 }, { "epoch": 0.1056704942081036, "grad_norm": 12.564291954040527, "learning_rate": 9.771408809193673e-06, "loss": 0.4067, "step": 2089 }, { "epoch": 0.10572107845616875, "grad_norm": 12.395662307739258, "learning_rate": 9.771168839465538e-06, "loss": 0.5452, "step": 2090 }, { "epoch": 0.1057716627042339, "grad_norm": 12.172588348388672, "learning_rate": 9.770928746795903e-06, "loss": 0.4095, "step": 2091 }, { "epoch": 0.10582224695229905, "grad_norm": 5.6374335289001465, "learning_rate": 9.770688531190961e-06, "loss": 0.3519, "step": 2092 }, { "epoch": 0.10587283120036421, "grad_norm": 12.210537910461426, "learning_rate": 9.770448192656899e-06, "loss": 0.5756, "step": 2093 }, { "epoch": 0.10592341544842936, "grad_norm": 8.771955490112305, "learning_rate": 9.770207731199909e-06, "loss": 0.4615, "step": 2094 }, { "epoch": 0.10597399969649451, "grad_norm": 5.875935077667236, "learning_rate": 9.769967146826186e-06, "loss": 0.4315, "step": 2095 }, { "epoch": 0.10602458394455966, "grad_norm": 9.495190620422363, "learning_rate": 9.769726439541932e-06, "loss": 0.5497, "step": 2096 }, { "epoch": 0.10607516819262482, "grad_norm": 5.742330074310303, "learning_rate": 9.769485609353348e-06, "loss": 0.3985, "step": 2097 }, { "epoch": 0.10612575244068997, "grad_norm": 13.297308921813965, "learning_rate": 9.769244656266642e-06, "loss": 0.5761, "step": 2098 }, { "epoch": 0.10617633668875512, "grad_norm": 9.65923023223877, "learning_rate": 9.769003580288019e-06, "loss": 0.5234, "step": 2099 }, { "epoch": 0.10622692093682028, "grad_norm": 6.164853096008301, "learning_rate": 9.768762381423693e-06, "loss": 0.4478, "step": 2100 }, { "epoch": 0.10627750518488542, "grad_norm": 6.040792942047119, "learning_rate": 9.76852105967988e-06, "loss": 0.5094, "step": 2101 }, { "epoch": 0.10632808943295058, "grad_norm": 8.5952787399292, "learning_rate": 9.768279615062797e-06, "loss": 0.3831, "step": 2102 }, { "epoch": 0.10637867368101574, "grad_norm": 21.968334197998047, "learning_rate": 9.768038047578665e-06, "loss": 0.5469, "step": 2103 }, { "epoch": 0.10642925792908088, "grad_norm": 13.281744956970215, "learning_rate": 9.76779635723371e-06, "loss": 0.5172, "step": 2104 }, { "epoch": 0.10647984217714604, "grad_norm": 14.36513900756836, "learning_rate": 9.767554544034158e-06, "loss": 0.5623, "step": 2105 }, { "epoch": 0.10653042642521118, "grad_norm": 15.347131729125977, "learning_rate": 9.767312607986241e-06, "loss": 0.5912, "step": 2106 }, { "epoch": 0.10658101067327634, "grad_norm": 7.913651466369629, "learning_rate": 9.767070549096194e-06, "loss": 0.5382, "step": 2107 }, { "epoch": 0.1066315949213415, "grad_norm": 6.846064567565918, "learning_rate": 9.766828367370252e-06, "loss": 0.4803, "step": 2108 }, { "epoch": 0.10668217916940664, "grad_norm": 5.769557476043701, "learning_rate": 9.766586062814658e-06, "loss": 0.393, "step": 2109 }, { "epoch": 0.1067327634174718, "grad_norm": 7.321330547332764, "learning_rate": 9.766343635435653e-06, "loss": 0.432, "step": 2110 }, { "epoch": 0.10678334766553695, "grad_norm": 7.33292293548584, "learning_rate": 9.766101085239486e-06, "loss": 0.43, "step": 2111 }, { "epoch": 0.1068339319136021, "grad_norm": 7.9695725440979, "learning_rate": 9.765858412232407e-06, "loss": 0.4749, "step": 2112 }, { "epoch": 0.10688451616166726, "grad_norm": 9.450716018676758, "learning_rate": 9.765615616420668e-06, "loss": 0.4464, "step": 2113 }, { "epoch": 0.1069351004097324, "grad_norm": 7.4050774574279785, "learning_rate": 9.765372697810524e-06, "loss": 0.4632, "step": 2114 }, { "epoch": 0.10698568465779756, "grad_norm": 8.70212173461914, "learning_rate": 9.765129656408237e-06, "loss": 0.4923, "step": 2115 }, { "epoch": 0.10703626890586271, "grad_norm": 7.051092624664307, "learning_rate": 9.764886492220067e-06, "loss": 0.4503, "step": 2116 }, { "epoch": 0.10708685315392787, "grad_norm": 10.015761375427246, "learning_rate": 9.764643205252284e-06, "loss": 0.4311, "step": 2117 }, { "epoch": 0.10713743740199302, "grad_norm": 10.537524223327637, "learning_rate": 9.764399795511152e-06, "loss": 0.5027, "step": 2118 }, { "epoch": 0.10718802165005817, "grad_norm": 7.942960739135742, "learning_rate": 9.764156263002946e-06, "loss": 0.3845, "step": 2119 }, { "epoch": 0.10723860589812333, "grad_norm": 8.32974624633789, "learning_rate": 9.763912607733938e-06, "loss": 0.4754, "step": 2120 }, { "epoch": 0.10728919014618847, "grad_norm": 7.884248733520508, "learning_rate": 9.763668829710412e-06, "loss": 0.4405, "step": 2121 }, { "epoch": 0.10733977439425363, "grad_norm": 6.066192150115967, "learning_rate": 9.763424928938646e-06, "loss": 0.3644, "step": 2122 }, { "epoch": 0.10739035864231879, "grad_norm": 21.611675262451172, "learning_rate": 9.763180905424924e-06, "loss": 0.513, "step": 2123 }, { "epoch": 0.10744094289038393, "grad_norm": 10.261446952819824, "learning_rate": 9.762936759175535e-06, "loss": 0.5205, "step": 2124 }, { "epoch": 0.10749152713844909, "grad_norm": 11.062623977661133, "learning_rate": 9.762692490196771e-06, "loss": 0.465, "step": 2125 }, { "epoch": 0.10754211138651423, "grad_norm": 16.520017623901367, "learning_rate": 9.762448098494925e-06, "loss": 0.4313, "step": 2126 }, { "epoch": 0.10759269563457939, "grad_norm": 13.535219192504883, "learning_rate": 9.762203584076295e-06, "loss": 0.5416, "step": 2127 }, { "epoch": 0.10764327988264455, "grad_norm": 9.097967147827148, "learning_rate": 9.761958946947181e-06, "loss": 0.4152, "step": 2128 }, { "epoch": 0.1076938641307097, "grad_norm": 9.011199951171875, "learning_rate": 9.761714187113886e-06, "loss": 0.3639, "step": 2129 }, { "epoch": 0.10774444837877485, "grad_norm": 8.952476501464844, "learning_rate": 9.76146930458272e-06, "loss": 0.5006, "step": 2130 }, { "epoch": 0.10779503262684, "grad_norm": 8.771080017089844, "learning_rate": 9.76122429935999e-06, "loss": 0.499, "step": 2131 }, { "epoch": 0.10784561687490515, "grad_norm": 7.256201267242432, "learning_rate": 9.760979171452009e-06, "loss": 0.4018, "step": 2132 }, { "epoch": 0.10789620112297031, "grad_norm": 14.930139541625977, "learning_rate": 9.760733920865096e-06, "loss": 0.5537, "step": 2133 }, { "epoch": 0.10794678537103546, "grad_norm": 12.672710418701172, "learning_rate": 9.760488547605568e-06, "loss": 0.642, "step": 2134 }, { "epoch": 0.10799736961910061, "grad_norm": 12.00405216217041, "learning_rate": 9.760243051679748e-06, "loss": 0.5824, "step": 2135 }, { "epoch": 0.10804795386716576, "grad_norm": 9.718962669372559, "learning_rate": 9.759997433093965e-06, "loss": 0.4515, "step": 2136 }, { "epoch": 0.10809853811523092, "grad_norm": 9.775050163269043, "learning_rate": 9.759751691854541e-06, "loss": 0.4419, "step": 2137 }, { "epoch": 0.10814912236329607, "grad_norm": 14.808868408203125, "learning_rate": 9.759505827967817e-06, "loss": 0.5043, "step": 2138 }, { "epoch": 0.10819970661136122, "grad_norm": 13.780189514160156, "learning_rate": 9.759259841440124e-06, "loss": 0.4517, "step": 2139 }, { "epoch": 0.10825029085942638, "grad_norm": 14.044647216796875, "learning_rate": 9.759013732277797e-06, "loss": 0.5712, "step": 2140 }, { "epoch": 0.10830087510749152, "grad_norm": 6.550115585327148, "learning_rate": 9.758767500487183e-06, "loss": 0.4672, "step": 2141 }, { "epoch": 0.10835145935555668, "grad_norm": 7.76632833480835, "learning_rate": 9.758521146074624e-06, "loss": 0.3885, "step": 2142 }, { "epoch": 0.10840204360362184, "grad_norm": 8.462224006652832, "learning_rate": 9.75827466904647e-06, "loss": 0.4863, "step": 2143 }, { "epoch": 0.10845262785168698, "grad_norm": 6.1619977951049805, "learning_rate": 9.758028069409068e-06, "loss": 0.4253, "step": 2144 }, { "epoch": 0.10850321209975214, "grad_norm": 13.115635871887207, "learning_rate": 9.757781347168779e-06, "loss": 0.5906, "step": 2145 }, { "epoch": 0.10855379634781728, "grad_norm": 6.215761184692383, "learning_rate": 9.757534502331955e-06, "loss": 0.4325, "step": 2146 }, { "epoch": 0.10860438059588244, "grad_norm": 15.199505805969238, "learning_rate": 9.757287534904959e-06, "loss": 0.5235, "step": 2147 }, { "epoch": 0.1086549648439476, "grad_norm": 15.715638160705566, "learning_rate": 9.757040444894153e-06, "loss": 0.6697, "step": 2148 }, { "epoch": 0.10870554909201274, "grad_norm": 5.29465389251709, "learning_rate": 9.756793232305906e-06, "loss": 0.4499, "step": 2149 }, { "epoch": 0.1087561333400779, "grad_norm": 8.912266731262207, "learning_rate": 9.756545897146585e-06, "loss": 0.4854, "step": 2150 }, { "epoch": 0.10880671758814305, "grad_norm": 10.857187271118164, "learning_rate": 9.756298439422566e-06, "loss": 0.5174, "step": 2151 }, { "epoch": 0.1088573018362082, "grad_norm": 10.31348991394043, "learning_rate": 9.756050859140224e-06, "loss": 0.5303, "step": 2152 }, { "epoch": 0.10890788608427336, "grad_norm": 5.934322357177734, "learning_rate": 9.755803156305941e-06, "loss": 0.4267, "step": 2153 }, { "epoch": 0.1089584703323385, "grad_norm": 9.409659385681152, "learning_rate": 9.755555330926098e-06, "loss": 0.4202, "step": 2154 }, { "epoch": 0.10900905458040366, "grad_norm": 6.953386306762695, "learning_rate": 9.75530738300708e-06, "loss": 0.4772, "step": 2155 }, { "epoch": 0.10905963882846882, "grad_norm": 9.24351692199707, "learning_rate": 9.755059312555275e-06, "loss": 0.5446, "step": 2156 }, { "epoch": 0.10911022307653397, "grad_norm": 7.160998344421387, "learning_rate": 9.75481111957708e-06, "loss": 0.3688, "step": 2157 }, { "epoch": 0.10916080732459912, "grad_norm": 6.230129718780518, "learning_rate": 9.754562804078884e-06, "loss": 0.4547, "step": 2158 }, { "epoch": 0.10921139157266427, "grad_norm": 20.944562911987305, "learning_rate": 9.75431436606709e-06, "loss": 0.6354, "step": 2159 }, { "epoch": 0.10926197582072943, "grad_norm": 18.12847137451172, "learning_rate": 9.754065805548099e-06, "loss": 0.5776, "step": 2160 }, { "epoch": 0.10931256006879458, "grad_norm": 9.864290237426758, "learning_rate": 9.753817122528315e-06, "loss": 0.5148, "step": 2161 }, { "epoch": 0.10936314431685973, "grad_norm": 8.28518295288086, "learning_rate": 9.753568317014146e-06, "loss": 0.5522, "step": 2162 }, { "epoch": 0.10941372856492489, "grad_norm": 8.135408401489258, "learning_rate": 9.753319389012003e-06, "loss": 0.5553, "step": 2163 }, { "epoch": 0.10946431281299003, "grad_norm": 10.926722526550293, "learning_rate": 9.753070338528299e-06, "loss": 0.4178, "step": 2164 }, { "epoch": 0.10951489706105519, "grad_norm": 10.448451042175293, "learning_rate": 9.752821165569456e-06, "loss": 0.41, "step": 2165 }, { "epoch": 0.10956548130912035, "grad_norm": 8.098381042480469, "learning_rate": 9.75257187014189e-06, "loss": 0.5673, "step": 2166 }, { "epoch": 0.10961606555718549, "grad_norm": 11.802475929260254, "learning_rate": 9.752322452252026e-06, "loss": 0.4564, "step": 2167 }, { "epoch": 0.10966664980525065, "grad_norm": 11.37027359008789, "learning_rate": 9.752072911906292e-06, "loss": 0.5726, "step": 2168 }, { "epoch": 0.10971723405331579, "grad_norm": 7.426273345947266, "learning_rate": 9.751823249111116e-06, "loss": 0.4561, "step": 2169 }, { "epoch": 0.10976781830138095, "grad_norm": 14.46198558807373, "learning_rate": 9.751573463872933e-06, "loss": 0.4912, "step": 2170 }, { "epoch": 0.10981840254944611, "grad_norm": 10.683104515075684, "learning_rate": 9.751323556198178e-06, "loss": 0.3917, "step": 2171 }, { "epoch": 0.10986898679751125, "grad_norm": 8.94572639465332, "learning_rate": 9.751073526093293e-06, "loss": 0.4597, "step": 2172 }, { "epoch": 0.10991957104557641, "grad_norm": 13.084985733032227, "learning_rate": 9.750823373564718e-06, "loss": 0.4754, "step": 2173 }, { "epoch": 0.10997015529364156, "grad_norm": 7.292754650115967, "learning_rate": 9.750573098618901e-06, "loss": 0.5121, "step": 2174 }, { "epoch": 0.11002073954170671, "grad_norm": 13.881087303161621, "learning_rate": 9.75032270126229e-06, "loss": 0.5654, "step": 2175 }, { "epoch": 0.11007132378977187, "grad_norm": 13.71902847290039, "learning_rate": 9.750072181501334e-06, "loss": 0.5661, "step": 2176 }, { "epoch": 0.11012190803783702, "grad_norm": 18.192245483398438, "learning_rate": 9.749821539342493e-06, "loss": 0.4322, "step": 2177 }, { "epoch": 0.11017249228590217, "grad_norm": 5.949455261230469, "learning_rate": 9.749570774792223e-06, "loss": 0.4593, "step": 2178 }, { "epoch": 0.11022307653396732, "grad_norm": 9.50487995147705, "learning_rate": 9.749319887856988e-06, "loss": 0.4525, "step": 2179 }, { "epoch": 0.11027366078203248, "grad_norm": 8.30917739868164, "learning_rate": 9.74906887854325e-06, "loss": 0.4642, "step": 2180 }, { "epoch": 0.11032424503009763, "grad_norm": 8.65867805480957, "learning_rate": 9.748817746857478e-06, "loss": 0.5558, "step": 2181 }, { "epoch": 0.11037482927816278, "grad_norm": 8.979269027709961, "learning_rate": 9.748566492806143e-06, "loss": 0.4243, "step": 2182 }, { "epoch": 0.11042541352622794, "grad_norm": 19.102025985717773, "learning_rate": 9.748315116395717e-06, "loss": 0.6012, "step": 2183 }, { "epoch": 0.11047599777429308, "grad_norm": 8.491166114807129, "learning_rate": 9.748063617632682e-06, "loss": 0.5519, "step": 2184 }, { "epoch": 0.11052658202235824, "grad_norm": 12.860454559326172, "learning_rate": 9.747811996523515e-06, "loss": 0.5821, "step": 2185 }, { "epoch": 0.1105771662704234, "grad_norm": 18.50994110107422, "learning_rate": 9.747560253074702e-06, "loss": 0.5454, "step": 2186 }, { "epoch": 0.11062775051848854, "grad_norm": 8.978483200073242, "learning_rate": 9.747308387292728e-06, "loss": 0.5075, "step": 2187 }, { "epoch": 0.1106783347665537, "grad_norm": 11.429389953613281, "learning_rate": 9.747056399184083e-06, "loss": 0.5548, "step": 2188 }, { "epoch": 0.11072891901461884, "grad_norm": 9.52747631072998, "learning_rate": 9.74680428875526e-06, "loss": 0.5459, "step": 2189 }, { "epoch": 0.110779503262684, "grad_norm": 12.435273170471191, "learning_rate": 9.746552056012759e-06, "loss": 0.6823, "step": 2190 }, { "epoch": 0.11083008751074916, "grad_norm": 14.130777359008789, "learning_rate": 9.746299700963073e-06, "loss": 0.5384, "step": 2191 }, { "epoch": 0.1108806717588143, "grad_norm": 12.253547668457031, "learning_rate": 9.746047223612708e-06, "loss": 0.5118, "step": 2192 }, { "epoch": 0.11093125600687946, "grad_norm": 10.725736618041992, "learning_rate": 9.745794623968172e-06, "loss": 0.5167, "step": 2193 }, { "epoch": 0.1109818402549446, "grad_norm": 13.313794136047363, "learning_rate": 9.745541902035969e-06, "loss": 0.5818, "step": 2194 }, { "epoch": 0.11103242450300976, "grad_norm": 10.839015007019043, "learning_rate": 9.745289057822616e-06, "loss": 0.5837, "step": 2195 }, { "epoch": 0.11108300875107492, "grad_norm": 9.244467735290527, "learning_rate": 9.745036091334625e-06, "loss": 0.5075, "step": 2196 }, { "epoch": 0.11113359299914007, "grad_norm": 11.289786338806152, "learning_rate": 9.744783002578514e-06, "loss": 0.5126, "step": 2197 }, { "epoch": 0.11118417724720522, "grad_norm": 11.520411491394043, "learning_rate": 9.744529791560806e-06, "loss": 0.4428, "step": 2198 }, { "epoch": 0.11123476149527037, "grad_norm": 5.803370952606201, "learning_rate": 9.744276458288027e-06, "loss": 0.4373, "step": 2199 }, { "epoch": 0.11128534574333553, "grad_norm": 10.819036483764648, "learning_rate": 9.7440230027667e-06, "loss": 0.3977, "step": 2200 }, { "epoch": 0.11133592999140068, "grad_norm": 16.106117248535156, "learning_rate": 9.74376942500336e-06, "loss": 0.6111, "step": 2201 }, { "epoch": 0.11138651423946583, "grad_norm": 11.226428985595703, "learning_rate": 9.743515725004543e-06, "loss": 0.4546, "step": 2202 }, { "epoch": 0.11143709848753099, "grad_norm": 17.40626335144043, "learning_rate": 9.743261902776779e-06, "loss": 0.7543, "step": 2203 }, { "epoch": 0.11148768273559613, "grad_norm": 19.264081954956055, "learning_rate": 9.743007958326614e-06, "loss": 0.4665, "step": 2204 }, { "epoch": 0.11153826698366129, "grad_norm": 10.292668342590332, "learning_rate": 9.74275389166059e-06, "loss": 0.5679, "step": 2205 }, { "epoch": 0.11158885123172645, "grad_norm": 13.18447208404541, "learning_rate": 9.742499702785255e-06, "loss": 0.5591, "step": 2206 }, { "epoch": 0.11163943547979159, "grad_norm": 10.861893653869629, "learning_rate": 9.742245391707157e-06, "loss": 0.6313, "step": 2207 }, { "epoch": 0.11169001972785675, "grad_norm": 11.203600883483887, "learning_rate": 9.74199095843285e-06, "loss": 0.6395, "step": 2208 }, { "epoch": 0.11174060397592189, "grad_norm": 10.536338806152344, "learning_rate": 9.74173640296889e-06, "loss": 0.4837, "step": 2209 }, { "epoch": 0.11179118822398705, "grad_norm": 5.417582988739014, "learning_rate": 9.741481725321836e-06, "loss": 0.4276, "step": 2210 }, { "epoch": 0.11184177247205221, "grad_norm": 13.117378234863281, "learning_rate": 9.741226925498251e-06, "loss": 0.6024, "step": 2211 }, { "epoch": 0.11189235672011735, "grad_norm": 13.5956392288208, "learning_rate": 9.740972003504698e-06, "loss": 0.4619, "step": 2212 }, { "epoch": 0.11194294096818251, "grad_norm": 7.76848840713501, "learning_rate": 9.74071695934775e-06, "loss": 0.526, "step": 2213 }, { "epoch": 0.11199352521624766, "grad_norm": 12.182363510131836, "learning_rate": 9.740461793033974e-06, "loss": 0.3795, "step": 2214 }, { "epoch": 0.11204410946431281, "grad_norm": 11.705594062805176, "learning_rate": 9.740206504569951e-06, "loss": 0.5329, "step": 2215 }, { "epoch": 0.11209469371237797, "grad_norm": 18.297475814819336, "learning_rate": 9.739951093962255e-06, "loss": 0.6764, "step": 2216 }, { "epoch": 0.11214527796044312, "grad_norm": 13.074421882629395, "learning_rate": 9.739695561217465e-06, "loss": 0.4916, "step": 2217 }, { "epoch": 0.11219586220850827, "grad_norm": 6.567411422729492, "learning_rate": 9.739439906342173e-06, "loss": 0.4541, "step": 2218 }, { "epoch": 0.11224644645657342, "grad_norm": 11.865763664245605, "learning_rate": 9.73918412934296e-06, "loss": 0.4636, "step": 2219 }, { "epoch": 0.11229703070463858, "grad_norm": 8.247868537902832, "learning_rate": 9.73892823022642e-06, "loss": 0.4901, "step": 2220 }, { "epoch": 0.11234761495270373, "grad_norm": 11.68879222869873, "learning_rate": 9.738672208999144e-06, "loss": 0.5191, "step": 2221 }, { "epoch": 0.11239819920076888, "grad_norm": 6.081917762756348, "learning_rate": 9.738416065667732e-06, "loss": 0.3897, "step": 2222 }, { "epoch": 0.11244878344883404, "grad_norm": 15.777430534362793, "learning_rate": 9.738159800238784e-06, "loss": 0.5853, "step": 2223 }, { "epoch": 0.11249936769689918, "grad_norm": 10.23561954498291, "learning_rate": 9.737903412718903e-06, "loss": 0.5811, "step": 2224 }, { "epoch": 0.11254995194496434, "grad_norm": 9.628497123718262, "learning_rate": 9.737646903114694e-06, "loss": 0.3343, "step": 2225 }, { "epoch": 0.1126005361930295, "grad_norm": 10.70766544342041, "learning_rate": 9.737390271432767e-06, "loss": 0.5083, "step": 2226 }, { "epoch": 0.11265112044109464, "grad_norm": 8.637734413146973, "learning_rate": 9.737133517679738e-06, "loss": 0.5611, "step": 2227 }, { "epoch": 0.1127017046891598, "grad_norm": 16.98990249633789, "learning_rate": 9.736876641862217e-06, "loss": 0.4138, "step": 2228 }, { "epoch": 0.11275228893722494, "grad_norm": 9.776063919067383, "learning_rate": 9.736619643986828e-06, "loss": 0.541, "step": 2229 }, { "epoch": 0.1128028731852901, "grad_norm": 7.284682273864746, "learning_rate": 9.736362524060192e-06, "loss": 0.5087, "step": 2230 }, { "epoch": 0.11285345743335526, "grad_norm": 9.753669738769531, "learning_rate": 9.736105282088933e-06, "loss": 0.4351, "step": 2231 }, { "epoch": 0.1129040416814204, "grad_norm": 18.445043563842773, "learning_rate": 9.735847918079682e-06, "loss": 0.5441, "step": 2232 }, { "epoch": 0.11295462592948556, "grad_norm": 4.520391464233398, "learning_rate": 9.735590432039066e-06, "loss": 0.3519, "step": 2233 }, { "epoch": 0.1130052101775507, "grad_norm": 13.279611587524414, "learning_rate": 9.735332823973726e-06, "loss": 0.6168, "step": 2234 }, { "epoch": 0.11305579442561586, "grad_norm": 11.95324420928955, "learning_rate": 9.735075093890296e-06, "loss": 0.5918, "step": 2235 }, { "epoch": 0.11310637867368102, "grad_norm": 8.235193252563477, "learning_rate": 9.734817241795418e-06, "loss": 0.4504, "step": 2236 }, { "epoch": 0.11315696292174617, "grad_norm": 11.503511428833008, "learning_rate": 9.734559267695736e-06, "loss": 0.6159, "step": 2237 }, { "epoch": 0.11320754716981132, "grad_norm": 8.315666198730469, "learning_rate": 9.734301171597896e-06, "loss": 0.4923, "step": 2238 }, { "epoch": 0.11325813141787647, "grad_norm": 9.782812118530273, "learning_rate": 9.734042953508552e-06, "loss": 0.5229, "step": 2239 }, { "epoch": 0.11330871566594163, "grad_norm": 11.478137016296387, "learning_rate": 9.733784613434355e-06, "loss": 0.4654, "step": 2240 }, { "epoch": 0.11335929991400678, "grad_norm": 8.007735252380371, "learning_rate": 9.733526151381962e-06, "loss": 0.4314, "step": 2241 }, { "epoch": 0.11340988416207193, "grad_norm": 8.243045806884766, "learning_rate": 9.733267567358035e-06, "loss": 0.3933, "step": 2242 }, { "epoch": 0.11346046841013709, "grad_norm": 6.298995494842529, "learning_rate": 9.733008861369234e-06, "loss": 0.462, "step": 2243 }, { "epoch": 0.11351105265820223, "grad_norm": 9.919722557067871, "learning_rate": 9.732750033422227e-06, "loss": 0.5643, "step": 2244 }, { "epoch": 0.11356163690626739, "grad_norm": 9.556256294250488, "learning_rate": 9.732491083523684e-06, "loss": 0.4908, "step": 2245 }, { "epoch": 0.11361222115433255, "grad_norm": 6.763632774353027, "learning_rate": 9.732232011680277e-06, "loss": 0.4851, "step": 2246 }, { "epoch": 0.11366280540239769, "grad_norm": 8.182405471801758, "learning_rate": 9.73197281789868e-06, "loss": 0.4351, "step": 2247 }, { "epoch": 0.11371338965046285, "grad_norm": 9.826265335083008, "learning_rate": 9.731713502185574e-06, "loss": 0.4902, "step": 2248 }, { "epoch": 0.11376397389852799, "grad_norm": 19.831212997436523, "learning_rate": 9.731454064547638e-06, "loss": 0.5184, "step": 2249 }, { "epoch": 0.11381455814659315, "grad_norm": 8.631776809692383, "learning_rate": 9.731194504991561e-06, "loss": 0.4444, "step": 2250 }, { "epoch": 0.11386514239465831, "grad_norm": 7.2525634765625, "learning_rate": 9.73093482352403e-06, "loss": 0.5474, "step": 2251 }, { "epoch": 0.11391572664272345, "grad_norm": 8.525755882263184, "learning_rate": 9.730675020151734e-06, "loss": 0.4259, "step": 2252 }, { "epoch": 0.11396631089078861, "grad_norm": 14.325681686401367, "learning_rate": 9.73041509488137e-06, "loss": 0.4402, "step": 2253 }, { "epoch": 0.11401689513885375, "grad_norm": 9.503738403320312, "learning_rate": 9.730155047719635e-06, "loss": 0.4572, "step": 2254 }, { "epoch": 0.11406747938691891, "grad_norm": 13.997736930847168, "learning_rate": 9.72989487867323e-06, "loss": 0.3317, "step": 2255 }, { "epoch": 0.11411806363498407, "grad_norm": 12.55298900604248, "learning_rate": 9.729634587748858e-06, "loss": 0.3977, "step": 2256 }, { "epoch": 0.11416864788304922, "grad_norm": 21.076610565185547, "learning_rate": 9.729374174953225e-06, "loss": 0.5707, "step": 2257 }, { "epoch": 0.11421923213111437, "grad_norm": 16.40089225769043, "learning_rate": 9.729113640293046e-06, "loss": 0.5424, "step": 2258 }, { "epoch": 0.11426981637917952, "grad_norm": 13.386125564575195, "learning_rate": 9.728852983775029e-06, "loss": 0.5414, "step": 2259 }, { "epoch": 0.11432040062724468, "grad_norm": 13.020681381225586, "learning_rate": 9.728592205405893e-06, "loss": 0.4834, "step": 2260 }, { "epoch": 0.11437098487530983, "grad_norm": 22.542076110839844, "learning_rate": 9.728331305192357e-06, "loss": 0.4913, "step": 2261 }, { "epoch": 0.11442156912337498, "grad_norm": 14.130237579345703, "learning_rate": 9.728070283141146e-06, "loss": 0.4779, "step": 2262 }, { "epoch": 0.11447215337144014, "grad_norm": 10.009299278259277, "learning_rate": 9.727809139258983e-06, "loss": 0.5109, "step": 2263 }, { "epoch": 0.11452273761950528, "grad_norm": 13.485845565795898, "learning_rate": 9.727547873552597e-06, "loss": 0.4953, "step": 2264 }, { "epoch": 0.11457332186757044, "grad_norm": 9.905128479003906, "learning_rate": 9.727286486028722e-06, "loss": 0.4575, "step": 2265 }, { "epoch": 0.1146239061156356, "grad_norm": 6.169199466705322, "learning_rate": 9.727024976694092e-06, "loss": 0.457, "step": 2266 }, { "epoch": 0.11467449036370074, "grad_norm": 12.136810302734375, "learning_rate": 9.726763345555448e-06, "loss": 0.6024, "step": 2267 }, { "epoch": 0.1147250746117659, "grad_norm": 11.8016939163208, "learning_rate": 9.726501592619529e-06, "loss": 0.53, "step": 2268 }, { "epoch": 0.11477565885983104, "grad_norm": 17.157989501953125, "learning_rate": 9.726239717893078e-06, "loss": 0.5668, "step": 2269 }, { "epoch": 0.1148262431078962, "grad_norm": 11.676033020019531, "learning_rate": 9.725977721382847e-06, "loss": 0.554, "step": 2270 }, { "epoch": 0.11487682735596136, "grad_norm": 11.039328575134277, "learning_rate": 9.725715603095585e-06, "loss": 0.5762, "step": 2271 }, { "epoch": 0.1149274116040265, "grad_norm": 16.63336944580078, "learning_rate": 9.725453363038047e-06, "loss": 0.4081, "step": 2272 }, { "epoch": 0.11497799585209166, "grad_norm": 8.418282508850098, "learning_rate": 9.725191001216988e-06, "loss": 0.5547, "step": 2273 }, { "epoch": 0.1150285801001568, "grad_norm": 16.647232055664062, "learning_rate": 9.724928517639172e-06, "loss": 0.5003, "step": 2274 }, { "epoch": 0.11507916434822196, "grad_norm": 16.472412109375, "learning_rate": 9.72466591231136e-06, "loss": 0.3761, "step": 2275 }, { "epoch": 0.11512974859628712, "grad_norm": 7.196225166320801, "learning_rate": 9.724403185240318e-06, "loss": 0.4487, "step": 2276 }, { "epoch": 0.11518033284435227, "grad_norm": 8.41779613494873, "learning_rate": 9.72414033643282e-06, "loss": 0.4895, "step": 2277 }, { "epoch": 0.11523091709241742, "grad_norm": 11.71787166595459, "learning_rate": 9.723877365895635e-06, "loss": 0.407, "step": 2278 }, { "epoch": 0.11528150134048257, "grad_norm": 13.572104454040527, "learning_rate": 9.72361427363554e-06, "loss": 0.5149, "step": 2279 }, { "epoch": 0.11533208558854773, "grad_norm": 14.325418472290039, "learning_rate": 9.723351059659314e-06, "loss": 0.5247, "step": 2280 }, { "epoch": 0.11538266983661288, "grad_norm": 8.102592468261719, "learning_rate": 9.72308772397374e-06, "loss": 0.4707, "step": 2281 }, { "epoch": 0.11543325408467803, "grad_norm": 8.513775825500488, "learning_rate": 9.722824266585606e-06, "loss": 0.5056, "step": 2282 }, { "epoch": 0.11548383833274319, "grad_norm": 8.550285339355469, "learning_rate": 9.722560687501696e-06, "loss": 0.4399, "step": 2283 }, { "epoch": 0.11553442258080833, "grad_norm": 6.8590922355651855, "learning_rate": 9.722296986728805e-06, "loss": 0.5255, "step": 2284 }, { "epoch": 0.11558500682887349, "grad_norm": 17.567228317260742, "learning_rate": 9.722033164273725e-06, "loss": 0.6697, "step": 2285 }, { "epoch": 0.11563559107693865, "grad_norm": 11.528342247009277, "learning_rate": 9.721769220143257e-06, "loss": 0.5746, "step": 2286 }, { "epoch": 0.11568617532500379, "grad_norm": 10.067700386047363, "learning_rate": 9.721505154344201e-06, "loss": 0.5197, "step": 2287 }, { "epoch": 0.11573675957306895, "grad_norm": 10.71081256866455, "learning_rate": 9.721240966883363e-06, "loss": 0.5725, "step": 2288 }, { "epoch": 0.11578734382113409, "grad_norm": 7.920322895050049, "learning_rate": 9.720976657767547e-06, "loss": 0.44, "step": 2289 }, { "epoch": 0.11583792806919925, "grad_norm": 8.396538734436035, "learning_rate": 9.720712227003567e-06, "loss": 0.4872, "step": 2290 }, { "epoch": 0.11588851231726441, "grad_norm": 7.0252532958984375, "learning_rate": 9.720447674598235e-06, "loss": 0.4509, "step": 2291 }, { "epoch": 0.11593909656532955, "grad_norm": 14.28758430480957, "learning_rate": 9.720183000558368e-06, "loss": 0.5004, "step": 2292 }, { "epoch": 0.11598968081339471, "grad_norm": 10.199484825134277, "learning_rate": 9.719918204890786e-06, "loss": 0.4191, "step": 2293 }, { "epoch": 0.11604026506145985, "grad_norm": 10.154881477355957, "learning_rate": 9.719653287602312e-06, "loss": 0.4688, "step": 2294 }, { "epoch": 0.11609084930952501, "grad_norm": 8.56124496459961, "learning_rate": 9.719388248699774e-06, "loss": 0.5099, "step": 2295 }, { "epoch": 0.11614143355759017, "grad_norm": 6.382133483886719, "learning_rate": 9.719123088189999e-06, "loss": 0.4348, "step": 2296 }, { "epoch": 0.11619201780565531, "grad_norm": 6.538751125335693, "learning_rate": 9.71885780607982e-06, "loss": 0.4849, "step": 2297 }, { "epoch": 0.11624260205372047, "grad_norm": 7.341673851013184, "learning_rate": 9.718592402376075e-06, "loss": 0.4057, "step": 2298 }, { "epoch": 0.11629318630178562, "grad_norm": 8.358845710754395, "learning_rate": 9.7183268770856e-06, "loss": 0.4001, "step": 2299 }, { "epoch": 0.11634377054985078, "grad_norm": 15.732513427734375, "learning_rate": 9.718061230215238e-06, "loss": 0.5376, "step": 2300 }, { "epoch": 0.11639435479791593, "grad_norm": 14.99863052368164, "learning_rate": 9.717795461771834e-06, "loss": 0.6703, "step": 2301 }, { "epoch": 0.11644493904598108, "grad_norm": 13.539957046508789, "learning_rate": 9.717529571762236e-06, "loss": 0.4648, "step": 2302 }, { "epoch": 0.11649552329404624, "grad_norm": 8.290730476379395, "learning_rate": 9.717263560193298e-06, "loss": 0.5351, "step": 2303 }, { "epoch": 0.1165461075421114, "grad_norm": 7.885566711425781, "learning_rate": 9.71699742707187e-06, "loss": 0.4264, "step": 2304 }, { "epoch": 0.11659669179017654, "grad_norm": 5.384448528289795, "learning_rate": 9.716731172404811e-06, "loss": 0.4775, "step": 2305 }, { "epoch": 0.1166472760382417, "grad_norm": 5.82820987701416, "learning_rate": 9.716464796198984e-06, "loss": 0.4322, "step": 2306 }, { "epoch": 0.11669786028630684, "grad_norm": 13.130861282348633, "learning_rate": 9.71619829846125e-06, "loss": 0.4672, "step": 2307 }, { "epoch": 0.116748444534372, "grad_norm": 8.417285919189453, "learning_rate": 9.715931679198479e-06, "loss": 0.4924, "step": 2308 }, { "epoch": 0.11679902878243716, "grad_norm": 6.610389709472656, "learning_rate": 9.715664938417539e-06, "loss": 0.4762, "step": 2309 }, { "epoch": 0.1168496130305023, "grad_norm": 6.66260290145874, "learning_rate": 9.715398076125304e-06, "loss": 0.4116, "step": 2310 }, { "epoch": 0.11690019727856746, "grad_norm": 11.499792098999023, "learning_rate": 9.715131092328649e-06, "loss": 0.4372, "step": 2311 }, { "epoch": 0.1169507815266326, "grad_norm": 8.50525951385498, "learning_rate": 9.714863987034455e-06, "loss": 0.4753, "step": 2312 }, { "epoch": 0.11700136577469776, "grad_norm": 10.423168182373047, "learning_rate": 9.714596760249606e-06, "loss": 0.5575, "step": 2313 }, { "epoch": 0.11705195002276292, "grad_norm": 10.405198097229004, "learning_rate": 9.714329411980985e-06, "loss": 0.446, "step": 2314 }, { "epoch": 0.11710253427082806, "grad_norm": 7.669521808624268, "learning_rate": 9.714061942235482e-06, "loss": 0.3236, "step": 2315 }, { "epoch": 0.11715311851889322, "grad_norm": 11.511848449707031, "learning_rate": 9.71379435101999e-06, "loss": 0.608, "step": 2316 }, { "epoch": 0.11720370276695836, "grad_norm": 10.383912086486816, "learning_rate": 9.713526638341402e-06, "loss": 0.389, "step": 2317 }, { "epoch": 0.11725428701502352, "grad_norm": 7.072097301483154, "learning_rate": 9.713258804206618e-06, "loss": 0.4566, "step": 2318 }, { "epoch": 0.11730487126308868, "grad_norm": 36.4439582824707, "learning_rate": 9.712990848622538e-06, "loss": 0.5965, "step": 2319 }, { "epoch": 0.11735545551115383, "grad_norm": 18.118017196655273, "learning_rate": 9.712722771596069e-06, "loss": 0.5955, "step": 2320 }, { "epoch": 0.11740603975921898, "grad_norm": 8.527251243591309, "learning_rate": 9.712454573134115e-06, "loss": 0.5264, "step": 2321 }, { "epoch": 0.11745662400728413, "grad_norm": 6.8289289474487305, "learning_rate": 9.712186253243591e-06, "loss": 0.4647, "step": 2322 }, { "epoch": 0.11750720825534929, "grad_norm": 9.510627746582031, "learning_rate": 9.71191781193141e-06, "loss": 0.5196, "step": 2323 }, { "epoch": 0.11755779250341444, "grad_norm": 8.285612106323242, "learning_rate": 9.711649249204487e-06, "loss": 0.5054, "step": 2324 }, { "epoch": 0.11760837675147959, "grad_norm": 12.374939918518066, "learning_rate": 9.711380565069743e-06, "loss": 0.609, "step": 2325 }, { "epoch": 0.11765896099954475, "grad_norm": 9.93558406829834, "learning_rate": 9.711111759534103e-06, "loss": 0.4553, "step": 2326 }, { "epoch": 0.11770954524760989, "grad_norm": 9.199418067932129, "learning_rate": 9.710842832604492e-06, "loss": 0.5007, "step": 2327 }, { "epoch": 0.11776012949567505, "grad_norm": 9.200591087341309, "learning_rate": 9.71057378428784e-06, "loss": 0.4543, "step": 2328 }, { "epoch": 0.1178107137437402, "grad_norm": 11.60881519317627, "learning_rate": 9.710304614591077e-06, "loss": 0.5771, "step": 2329 }, { "epoch": 0.11786129799180535, "grad_norm": 6.850183010101318, "learning_rate": 9.710035323521143e-06, "loss": 0.4844, "step": 2330 }, { "epoch": 0.11791188223987051, "grad_norm": 8.88841438293457, "learning_rate": 9.709765911084976e-06, "loss": 0.4637, "step": 2331 }, { "epoch": 0.11796246648793565, "grad_norm": 6.661958694458008, "learning_rate": 9.709496377289517e-06, "loss": 0.4365, "step": 2332 }, { "epoch": 0.11801305073600081, "grad_norm": 6.379390239715576, "learning_rate": 9.709226722141711e-06, "loss": 0.4395, "step": 2333 }, { "epoch": 0.11806363498406597, "grad_norm": 6.397888660430908, "learning_rate": 9.708956945648508e-06, "loss": 0.3141, "step": 2334 }, { "epoch": 0.11811421923213111, "grad_norm": 13.183162689208984, "learning_rate": 9.708687047816858e-06, "loss": 0.5107, "step": 2335 }, { "epoch": 0.11816480348019627, "grad_norm": 22.115036010742188, "learning_rate": 9.708417028653717e-06, "loss": 0.4749, "step": 2336 }, { "epoch": 0.11821538772826141, "grad_norm": 8.799613952636719, "learning_rate": 9.708146888166041e-06, "loss": 0.5406, "step": 2337 }, { "epoch": 0.11826597197632657, "grad_norm": 9.922798156738281, "learning_rate": 9.707876626360792e-06, "loss": 0.4842, "step": 2338 }, { "epoch": 0.11831655622439173, "grad_norm": 8.972850799560547, "learning_rate": 9.707606243244933e-06, "loss": 0.5436, "step": 2339 }, { "epoch": 0.11836714047245687, "grad_norm": 11.49515438079834, "learning_rate": 9.707335738825434e-06, "loss": 0.5354, "step": 2340 }, { "epoch": 0.11841772472052203, "grad_norm": 11.851741790771484, "learning_rate": 9.70706511310926e-06, "loss": 0.4959, "step": 2341 }, { "epoch": 0.11846830896858718, "grad_norm": 6.67958402633667, "learning_rate": 9.70679436610339e-06, "loss": 0.3553, "step": 2342 }, { "epoch": 0.11851889321665234, "grad_norm": 11.722823143005371, "learning_rate": 9.706523497814798e-06, "loss": 0.5179, "step": 2343 }, { "epoch": 0.1185694774647175, "grad_norm": 8.230915069580078, "learning_rate": 9.706252508250462e-06, "loss": 0.5467, "step": 2344 }, { "epoch": 0.11862006171278264, "grad_norm": 9.793401718139648, "learning_rate": 9.705981397417367e-06, "loss": 0.4659, "step": 2345 }, { "epoch": 0.1186706459608478, "grad_norm": 10.929908752441406, "learning_rate": 9.705710165322496e-06, "loss": 0.5051, "step": 2346 }, { "epoch": 0.11872123020891294, "grad_norm": 9.826921463012695, "learning_rate": 9.705438811972845e-06, "loss": 0.3865, "step": 2347 }, { "epoch": 0.1187718144569781, "grad_norm": 9.048709869384766, "learning_rate": 9.7051673373754e-06, "loss": 0.4399, "step": 2348 }, { "epoch": 0.11882239870504326, "grad_norm": 6.386701583862305, "learning_rate": 9.704895741537154e-06, "loss": 0.4131, "step": 2349 }, { "epoch": 0.1188729829531084, "grad_norm": 16.686389923095703, "learning_rate": 9.704624024465113e-06, "loss": 0.4335, "step": 2350 }, { "epoch": 0.11892356720117356, "grad_norm": 10.726837158203125, "learning_rate": 9.704352186166271e-06, "loss": 0.5436, "step": 2351 }, { "epoch": 0.1189741514492387, "grad_norm": 12.244184494018555, "learning_rate": 9.704080226647639e-06, "loss": 0.4348, "step": 2352 }, { "epoch": 0.11902473569730386, "grad_norm": 14.658895492553711, "learning_rate": 9.70380814591622e-06, "loss": 0.5424, "step": 2353 }, { "epoch": 0.11907531994536902, "grad_norm": 10.181828498840332, "learning_rate": 9.703535943979028e-06, "loss": 0.535, "step": 2354 }, { "epoch": 0.11912590419343416, "grad_norm": 7.0725417137146, "learning_rate": 9.703263620843076e-06, "loss": 0.4744, "step": 2355 }, { "epoch": 0.11917648844149932, "grad_norm": 17.542320251464844, "learning_rate": 9.702991176515378e-06, "loss": 0.5487, "step": 2356 }, { "epoch": 0.11922707268956446, "grad_norm": 19.24229621887207, "learning_rate": 9.70271861100296e-06, "loss": 0.5266, "step": 2357 }, { "epoch": 0.11927765693762962, "grad_norm": 8.544342041015625, "learning_rate": 9.70244592431284e-06, "loss": 0.4495, "step": 2358 }, { "epoch": 0.11932824118569478, "grad_norm": 15.38912582397461, "learning_rate": 9.702173116452047e-06, "loss": 0.5186, "step": 2359 }, { "epoch": 0.11937882543375992, "grad_norm": 7.846489429473877, "learning_rate": 9.701900187427611e-06, "loss": 0.5298, "step": 2360 }, { "epoch": 0.11942940968182508, "grad_norm": 10.080306053161621, "learning_rate": 9.701627137246565e-06, "loss": 0.4216, "step": 2361 }, { "epoch": 0.11947999392989023, "grad_norm": 9.113515853881836, "learning_rate": 9.701353965915944e-06, "loss": 0.5342, "step": 2362 }, { "epoch": 0.11953057817795538, "grad_norm": 9.065251350402832, "learning_rate": 9.701080673442787e-06, "loss": 0.6051, "step": 2363 }, { "epoch": 0.11958116242602054, "grad_norm": 15.131909370422363, "learning_rate": 9.700807259834136e-06, "loss": 0.704, "step": 2364 }, { "epoch": 0.11963174667408569, "grad_norm": 8.289603233337402, "learning_rate": 9.700533725097035e-06, "loss": 0.3725, "step": 2365 }, { "epoch": 0.11968233092215085, "grad_norm": 7.29202127456665, "learning_rate": 9.700260069238535e-06, "loss": 0.4871, "step": 2366 }, { "epoch": 0.11973291517021599, "grad_norm": 9.034862518310547, "learning_rate": 9.699986292265685e-06, "loss": 0.4636, "step": 2367 }, { "epoch": 0.11978349941828115, "grad_norm": 8.85631275177002, "learning_rate": 9.699712394185541e-06, "loss": 0.4728, "step": 2368 }, { "epoch": 0.1198340836663463, "grad_norm": 5.731250762939453, "learning_rate": 9.699438375005158e-06, "loss": 0.3838, "step": 2369 }, { "epoch": 0.11988466791441145, "grad_norm": 8.640137672424316, "learning_rate": 9.699164234731603e-06, "loss": 0.4088, "step": 2370 }, { "epoch": 0.11993525216247661, "grad_norm": 15.944292068481445, "learning_rate": 9.698889973371934e-06, "loss": 0.5468, "step": 2371 }, { "epoch": 0.11998583641054175, "grad_norm": 8.373595237731934, "learning_rate": 9.69861559093322e-06, "loss": 0.465, "step": 2372 }, { "epoch": 0.12003642065860691, "grad_norm": 19.746122360229492, "learning_rate": 9.698341087422532e-06, "loss": 0.6641, "step": 2373 }, { "epoch": 0.12008700490667207, "grad_norm": 11.758955955505371, "learning_rate": 9.698066462846942e-06, "loss": 0.4975, "step": 2374 }, { "epoch": 0.12013758915473721, "grad_norm": 18.772884368896484, "learning_rate": 9.697791717213528e-06, "loss": 0.6639, "step": 2375 }, { "epoch": 0.12018817340280237, "grad_norm": 11.97056770324707, "learning_rate": 9.697516850529367e-06, "loss": 0.6259, "step": 2376 }, { "epoch": 0.12023875765086751, "grad_norm": 7.038887023925781, "learning_rate": 9.697241862801544e-06, "loss": 0.5259, "step": 2377 }, { "epoch": 0.12028934189893267, "grad_norm": 7.2394537925720215, "learning_rate": 9.696966754037143e-06, "loss": 0.5457, "step": 2378 }, { "epoch": 0.12033992614699783, "grad_norm": 8.819025039672852, "learning_rate": 9.696691524243255e-06, "loss": 0.4433, "step": 2379 }, { "epoch": 0.12039051039506297, "grad_norm": 11.564252853393555, "learning_rate": 9.69641617342697e-06, "loss": 0.4465, "step": 2380 }, { "epoch": 0.12044109464312813, "grad_norm": 8.735262870788574, "learning_rate": 9.696140701595384e-06, "loss": 0.4874, "step": 2381 }, { "epoch": 0.12049167889119328, "grad_norm": 5.1947784423828125, "learning_rate": 9.695865108755597e-06, "loss": 0.5522, "step": 2382 }, { "epoch": 0.12054226313925843, "grad_norm": 12.13407039642334, "learning_rate": 9.695589394914706e-06, "loss": 0.4739, "step": 2383 }, { "epoch": 0.12059284738732359, "grad_norm": 10.71971607208252, "learning_rate": 9.69531356007982e-06, "loss": 0.5045, "step": 2384 }, { "epoch": 0.12064343163538874, "grad_norm": 12.482538223266602, "learning_rate": 9.695037604258043e-06, "loss": 0.4254, "step": 2385 }, { "epoch": 0.1206940158834539, "grad_norm": 6.829817295074463, "learning_rate": 9.694761527456489e-06, "loss": 0.5324, "step": 2386 }, { "epoch": 0.12074460013151904, "grad_norm": 6.87022590637207, "learning_rate": 9.694485329682268e-06, "loss": 0.5209, "step": 2387 }, { "epoch": 0.1207951843795842, "grad_norm": 8.641921043395996, "learning_rate": 9.694209010942501e-06, "loss": 0.3349, "step": 2388 }, { "epoch": 0.12084576862764936, "grad_norm": 9.97567081451416, "learning_rate": 9.693932571244306e-06, "loss": 0.4802, "step": 2389 }, { "epoch": 0.1208963528757145, "grad_norm": 5.86590051651001, "learning_rate": 9.693656010594805e-06, "loss": 0.3697, "step": 2390 }, { "epoch": 0.12094693712377966, "grad_norm": 10.922499656677246, "learning_rate": 9.693379329001128e-06, "loss": 0.6848, "step": 2391 }, { "epoch": 0.1209975213718448, "grad_norm": 8.814624786376953, "learning_rate": 9.693102526470399e-06, "loss": 0.4366, "step": 2392 }, { "epoch": 0.12104810561990996, "grad_norm": 12.741928100585938, "learning_rate": 9.692825603009757e-06, "loss": 0.5709, "step": 2393 }, { "epoch": 0.12109868986797512, "grad_norm": 13.49649715423584, "learning_rate": 9.692548558626332e-06, "loss": 0.5032, "step": 2394 }, { "epoch": 0.12114927411604026, "grad_norm": 11.628276824951172, "learning_rate": 9.692271393327267e-06, "loss": 0.4298, "step": 2395 }, { "epoch": 0.12119985836410542, "grad_norm": 13.863476753234863, "learning_rate": 9.691994107119698e-06, "loss": 0.5573, "step": 2396 }, { "epoch": 0.12125044261217056, "grad_norm": 6.331872463226318, "learning_rate": 9.691716700010777e-06, "loss": 0.5191, "step": 2397 }, { "epoch": 0.12130102686023572, "grad_norm": 9.801365852355957, "learning_rate": 9.691439172007648e-06, "loss": 0.519, "step": 2398 }, { "epoch": 0.12135161110830088, "grad_norm": 6.951061248779297, "learning_rate": 9.691161523117465e-06, "loss": 0.4181, "step": 2399 }, { "epoch": 0.12140219535636602, "grad_norm": 8.215188980102539, "learning_rate": 9.690883753347379e-06, "loss": 0.4452, "step": 2400 }, { "epoch": 0.12145277960443118, "grad_norm": 11.311378479003906, "learning_rate": 9.690605862704548e-06, "loss": 0.4327, "step": 2401 }, { "epoch": 0.12150336385249633, "grad_norm": 10.231246948242188, "learning_rate": 9.690327851196134e-06, "loss": 0.4158, "step": 2402 }, { "epoch": 0.12155394810056148, "grad_norm": 6.801124095916748, "learning_rate": 9.690049718829303e-06, "loss": 0.4605, "step": 2403 }, { "epoch": 0.12160453234862664, "grad_norm": 8.586295127868652, "learning_rate": 9.689771465611218e-06, "loss": 0.4994, "step": 2404 }, { "epoch": 0.12165511659669179, "grad_norm": 10.70340347290039, "learning_rate": 9.689493091549047e-06, "loss": 0.4045, "step": 2405 }, { "epoch": 0.12170570084475694, "grad_norm": 10.960251808166504, "learning_rate": 9.68921459664997e-06, "loss": 0.4237, "step": 2406 }, { "epoch": 0.12175628509282209, "grad_norm": 5.910664081573486, "learning_rate": 9.688935980921157e-06, "loss": 0.3759, "step": 2407 }, { "epoch": 0.12180686934088725, "grad_norm": 17.271821975708008, "learning_rate": 9.68865724436979e-06, "loss": 0.5823, "step": 2408 }, { "epoch": 0.1218574535889524, "grad_norm": 8.427387237548828, "learning_rate": 9.68837838700305e-06, "loss": 0.327, "step": 2409 }, { "epoch": 0.12190803783701755, "grad_norm": 7.623230457305908, "learning_rate": 9.688099408828125e-06, "loss": 0.4602, "step": 2410 }, { "epoch": 0.12195862208508271, "grad_norm": 14.833612442016602, "learning_rate": 9.687820309852201e-06, "loss": 0.5009, "step": 2411 }, { "epoch": 0.12200920633314785, "grad_norm": 9.297395706176758, "learning_rate": 9.687541090082471e-06, "loss": 0.4, "step": 2412 }, { "epoch": 0.12205979058121301, "grad_norm": 11.386247634887695, "learning_rate": 9.687261749526127e-06, "loss": 0.531, "step": 2413 }, { "epoch": 0.12211037482927817, "grad_norm": 12.62019157409668, "learning_rate": 9.686982288190371e-06, "loss": 0.5471, "step": 2414 }, { "epoch": 0.12216095907734331, "grad_norm": 9.364335060119629, "learning_rate": 9.686702706082403e-06, "loss": 0.5193, "step": 2415 }, { "epoch": 0.12221154332540847, "grad_norm": 8.074071884155273, "learning_rate": 9.686423003209427e-06, "loss": 0.5521, "step": 2416 }, { "epoch": 0.12226212757347361, "grad_norm": 17.873985290527344, "learning_rate": 9.686143179578648e-06, "loss": 0.5479, "step": 2417 }, { "epoch": 0.12231271182153877, "grad_norm": 11.05881118774414, "learning_rate": 9.68586323519728e-06, "loss": 0.4381, "step": 2418 }, { "epoch": 0.12236329606960393, "grad_norm": 9.093664169311523, "learning_rate": 9.685583170072535e-06, "loss": 0.5887, "step": 2419 }, { "epoch": 0.12241388031766907, "grad_norm": 11.202329635620117, "learning_rate": 9.685302984211626e-06, "loss": 0.5348, "step": 2420 }, { "epoch": 0.12246446456573423, "grad_norm": 9.368311882019043, "learning_rate": 9.685022677621779e-06, "loss": 0.482, "step": 2421 }, { "epoch": 0.12251504881379938, "grad_norm": 6.934504508972168, "learning_rate": 9.684742250310213e-06, "loss": 0.4255, "step": 2422 }, { "epoch": 0.12256563306186453, "grad_norm": 10.966324806213379, "learning_rate": 9.684461702284153e-06, "loss": 0.418, "step": 2423 }, { "epoch": 0.12261621730992969, "grad_norm": 9.613936424255371, "learning_rate": 9.684181033550832e-06, "loss": 0.5348, "step": 2424 }, { "epoch": 0.12266680155799484, "grad_norm": 8.821162223815918, "learning_rate": 9.68390024411748e-06, "loss": 0.4767, "step": 2425 }, { "epoch": 0.12271738580606, "grad_norm": 10.206795692443848, "learning_rate": 9.683619333991332e-06, "loss": 0.544, "step": 2426 }, { "epoch": 0.12276797005412514, "grad_norm": 19.053232192993164, "learning_rate": 9.683338303179626e-06, "loss": 0.5722, "step": 2427 }, { "epoch": 0.1228185543021903, "grad_norm": 15.6214017868042, "learning_rate": 9.683057151689604e-06, "loss": 0.554, "step": 2428 }, { "epoch": 0.12286913855025546, "grad_norm": 7.970694065093994, "learning_rate": 9.68277587952851e-06, "loss": 0.4298, "step": 2429 }, { "epoch": 0.1229197227983206, "grad_norm": 14.383167266845703, "learning_rate": 9.682494486703593e-06, "loss": 0.4575, "step": 2430 }, { "epoch": 0.12297030704638576, "grad_norm": 7.69623327255249, "learning_rate": 9.682212973222104e-06, "loss": 0.4137, "step": 2431 }, { "epoch": 0.1230208912944509, "grad_norm": 8.214822769165039, "learning_rate": 9.681931339091295e-06, "loss": 0.5501, "step": 2432 }, { "epoch": 0.12307147554251606, "grad_norm": 5.527608871459961, "learning_rate": 9.681649584318424e-06, "loss": 0.2818, "step": 2433 }, { "epoch": 0.12312205979058122, "grad_norm": 5.942363262176514, "learning_rate": 9.681367708910752e-06, "loss": 0.4399, "step": 2434 }, { "epoch": 0.12317264403864636, "grad_norm": 16.35713768005371, "learning_rate": 9.681085712875543e-06, "loss": 0.4727, "step": 2435 }, { "epoch": 0.12322322828671152, "grad_norm": 14.918388366699219, "learning_rate": 9.680803596220058e-06, "loss": 0.5697, "step": 2436 }, { "epoch": 0.12327381253477666, "grad_norm": 11.639963150024414, "learning_rate": 9.680521358951574e-06, "loss": 0.4123, "step": 2437 }, { "epoch": 0.12332439678284182, "grad_norm": 7.96528959274292, "learning_rate": 9.680239001077359e-06, "loss": 0.5143, "step": 2438 }, { "epoch": 0.12337498103090698, "grad_norm": 12.627415657043457, "learning_rate": 9.679956522604688e-06, "loss": 0.4981, "step": 2439 }, { "epoch": 0.12342556527897212, "grad_norm": 9.824830055236816, "learning_rate": 9.679673923540843e-06, "loss": 0.4496, "step": 2440 }, { "epoch": 0.12347614952703728, "grad_norm": 10.073741912841797, "learning_rate": 9.679391203893103e-06, "loss": 0.5152, "step": 2441 }, { "epoch": 0.12352673377510243, "grad_norm": 6.0554890632629395, "learning_rate": 9.679108363668756e-06, "loss": 0.4397, "step": 2442 }, { "epoch": 0.12357731802316758, "grad_norm": 14.580350875854492, "learning_rate": 9.678825402875087e-06, "loss": 0.5376, "step": 2443 }, { "epoch": 0.12362790227123274, "grad_norm": 12.609935760498047, "learning_rate": 9.678542321519389e-06, "loss": 0.5543, "step": 2444 }, { "epoch": 0.12367848651929789, "grad_norm": 6.708312034606934, "learning_rate": 9.678259119608956e-06, "loss": 0.4658, "step": 2445 }, { "epoch": 0.12372907076736304, "grad_norm": 13.211852073669434, "learning_rate": 9.677975797151085e-06, "loss": 0.5545, "step": 2446 }, { "epoch": 0.12377965501542819, "grad_norm": 15.739047050476074, "learning_rate": 9.677692354153078e-06, "loss": 0.6522, "step": 2447 }, { "epoch": 0.12383023926349335, "grad_norm": 9.312605857849121, "learning_rate": 9.677408790622237e-06, "loss": 0.5105, "step": 2448 }, { "epoch": 0.1238808235115585, "grad_norm": 6.876385688781738, "learning_rate": 9.677125106565868e-06, "loss": 0.4589, "step": 2449 }, { "epoch": 0.12393140775962365, "grad_norm": 6.187245845794678, "learning_rate": 9.676841301991284e-06, "loss": 0.4523, "step": 2450 }, { "epoch": 0.12398199200768881, "grad_norm": 5.520021915435791, "learning_rate": 9.676557376905796e-06, "loss": 0.5203, "step": 2451 }, { "epoch": 0.12403257625575395, "grad_norm": 6.273560523986816, "learning_rate": 9.676273331316719e-06, "loss": 0.4477, "step": 2452 }, { "epoch": 0.12408316050381911, "grad_norm": 8.973950386047363, "learning_rate": 9.675989165231374e-06, "loss": 0.4104, "step": 2453 }, { "epoch": 0.12413374475188427, "grad_norm": 6.437857151031494, "learning_rate": 9.675704878657082e-06, "loss": 0.4511, "step": 2454 }, { "epoch": 0.12418432899994941, "grad_norm": 6.909470558166504, "learning_rate": 9.675420471601169e-06, "loss": 0.4404, "step": 2455 }, { "epoch": 0.12423491324801457, "grad_norm": 13.870248794555664, "learning_rate": 9.675135944070963e-06, "loss": 0.5274, "step": 2456 }, { "epoch": 0.12428549749607973, "grad_norm": 9.140334129333496, "learning_rate": 9.674851296073798e-06, "loss": 0.4388, "step": 2457 }, { "epoch": 0.12433608174414487, "grad_norm": 6.9515485763549805, "learning_rate": 9.674566527617007e-06, "loss": 0.4412, "step": 2458 }, { "epoch": 0.12438666599221003, "grad_norm": 10.413031578063965, "learning_rate": 9.674281638707926e-06, "loss": 0.3356, "step": 2459 }, { "epoch": 0.12443725024027517, "grad_norm": 12.305898666381836, "learning_rate": 9.673996629353898e-06, "loss": 0.5957, "step": 2460 }, { "epoch": 0.12448783448834033, "grad_norm": 7.560647964477539, "learning_rate": 9.673711499562264e-06, "loss": 0.4363, "step": 2461 }, { "epoch": 0.12453841873640549, "grad_norm": 9.168752670288086, "learning_rate": 9.673426249340376e-06, "loss": 0.4552, "step": 2462 }, { "epoch": 0.12458900298447063, "grad_norm": 9.05239486694336, "learning_rate": 9.673140878695581e-06, "loss": 0.527, "step": 2463 }, { "epoch": 0.12463958723253579, "grad_norm": 9.398942947387695, "learning_rate": 9.672855387635233e-06, "loss": 0.5102, "step": 2464 }, { "epoch": 0.12469017148060094, "grad_norm": 7.200027942657471, "learning_rate": 9.67256977616669e-06, "loss": 0.4548, "step": 2465 }, { "epoch": 0.1247407557286661, "grad_norm": 8.113397598266602, "learning_rate": 9.67228404429731e-06, "loss": 0.5152, "step": 2466 }, { "epoch": 0.12479133997673125, "grad_norm": 7.7112135887146, "learning_rate": 9.671998192034452e-06, "loss": 0.4089, "step": 2467 }, { "epoch": 0.1248419242247964, "grad_norm": 21.899328231811523, "learning_rate": 9.67171221938549e-06, "loss": 0.434, "step": 2468 }, { "epoch": 0.12489250847286155, "grad_norm": 10.143519401550293, "learning_rate": 9.671426126357785e-06, "loss": 0.4939, "step": 2469 }, { "epoch": 0.1249430927209267, "grad_norm": 8.599384307861328, "learning_rate": 9.671139912958714e-06, "loss": 0.4236, "step": 2470 }, { "epoch": 0.12499367696899186, "grad_norm": 8.211816787719727, "learning_rate": 9.670853579195648e-06, "loss": 0.5128, "step": 2471 }, { "epoch": 0.125044261217057, "grad_norm": 9.34001350402832, "learning_rate": 9.67056712507597e-06, "loss": 0.4838, "step": 2472 }, { "epoch": 0.12509484546512217, "grad_norm": 7.096646785736084, "learning_rate": 9.670280550607055e-06, "loss": 0.4951, "step": 2473 }, { "epoch": 0.12514542971318732, "grad_norm": 15.458888053894043, "learning_rate": 9.669993855796294e-06, "loss": 0.5363, "step": 2474 }, { "epoch": 0.12519601396125246, "grad_norm": 16.107948303222656, "learning_rate": 9.669707040651069e-06, "loss": 0.5648, "step": 2475 }, { "epoch": 0.1252465982093176, "grad_norm": 10.61740493774414, "learning_rate": 9.669420105178774e-06, "loss": 0.4933, "step": 2476 }, { "epoch": 0.12529718245738278, "grad_norm": 11.826822280883789, "learning_rate": 9.6691330493868e-06, "loss": 0.3234, "step": 2477 }, { "epoch": 0.12534776670544792, "grad_norm": 5.898537635803223, "learning_rate": 9.668845873282547e-06, "loss": 0.4722, "step": 2478 }, { "epoch": 0.12539835095351307, "grad_norm": 13.974885940551758, "learning_rate": 9.668558576873413e-06, "loss": 0.5944, "step": 2479 }, { "epoch": 0.12544893520157824, "grad_norm": 10.515462875366211, "learning_rate": 9.6682711601668e-06, "loss": 0.5953, "step": 2480 }, { "epoch": 0.12549951944964338, "grad_norm": 7.978176593780518, "learning_rate": 9.667983623170115e-06, "loss": 0.4644, "step": 2481 }, { "epoch": 0.12555010369770853, "grad_norm": 8.466141700744629, "learning_rate": 9.667695965890767e-06, "loss": 0.4714, "step": 2482 }, { "epoch": 0.1256006879457737, "grad_norm": 11.381327629089355, "learning_rate": 9.66740818833617e-06, "loss": 0.6068, "step": 2483 }, { "epoch": 0.12565127219383884, "grad_norm": 9.323442459106445, "learning_rate": 9.667120290513735e-06, "loss": 0.4682, "step": 2484 }, { "epoch": 0.125701856441904, "grad_norm": 14.17958927154541, "learning_rate": 9.666832272430884e-06, "loss": 0.5179, "step": 2485 }, { "epoch": 0.12575244068996913, "grad_norm": 7.596869468688965, "learning_rate": 9.666544134095038e-06, "loss": 0.52, "step": 2486 }, { "epoch": 0.1258030249380343, "grad_norm": 6.846745491027832, "learning_rate": 9.666255875513622e-06, "loss": 0.4327, "step": 2487 }, { "epoch": 0.12585360918609945, "grad_norm": 7.971548557281494, "learning_rate": 9.665967496694062e-06, "loss": 0.3197, "step": 2488 }, { "epoch": 0.1259041934341646, "grad_norm": 6.790052890777588, "learning_rate": 9.66567899764379e-06, "loss": 0.4037, "step": 2489 }, { "epoch": 0.12595477768222976, "grad_norm": 10.31688404083252, "learning_rate": 9.66539037837024e-06, "loss": 0.433, "step": 2490 }, { "epoch": 0.1260053619302949, "grad_norm": 16.772977828979492, "learning_rate": 9.665101638880848e-06, "loss": 0.4358, "step": 2491 }, { "epoch": 0.12605594617836005, "grad_norm": 9.846858978271484, "learning_rate": 9.664812779183054e-06, "loss": 0.5594, "step": 2492 }, { "epoch": 0.12610653042642522, "grad_norm": 7.246566295623779, "learning_rate": 9.664523799284304e-06, "loss": 0.4823, "step": 2493 }, { "epoch": 0.12615711467449037, "grad_norm": 24.724388122558594, "learning_rate": 9.664234699192041e-06, "loss": 0.7349, "step": 2494 }, { "epoch": 0.1262076989225555, "grad_norm": 10.112449645996094, "learning_rate": 9.663945478913716e-06, "loss": 0.4086, "step": 2495 }, { "epoch": 0.12625828317062066, "grad_norm": 10.42047119140625, "learning_rate": 9.663656138456783e-06, "loss": 0.5242, "step": 2496 }, { "epoch": 0.12630886741868583, "grad_norm": 9.652129173278809, "learning_rate": 9.663366677828695e-06, "loss": 0.4421, "step": 2497 }, { "epoch": 0.12635945166675097, "grad_norm": 6.007265567779541, "learning_rate": 9.66307709703691e-06, "loss": 0.3539, "step": 2498 }, { "epoch": 0.12641003591481612, "grad_norm": 6.31332540512085, "learning_rate": 9.662787396088894e-06, "loss": 0.4586, "step": 2499 }, { "epoch": 0.1264606201628813, "grad_norm": 9.266637802124023, "learning_rate": 9.662497574992109e-06, "loss": 0.3462, "step": 2500 }, { "epoch": 0.12651120441094643, "grad_norm": 6.658364772796631, "learning_rate": 9.662207633754022e-06, "loss": 0.4908, "step": 2501 }, { "epoch": 0.12656178865901158, "grad_norm": 8.648271560668945, "learning_rate": 9.661917572382104e-06, "loss": 0.5289, "step": 2502 }, { "epoch": 0.12661237290707675, "grad_norm": 10.70396900177002, "learning_rate": 9.661627390883833e-06, "loss": 0.405, "step": 2503 }, { "epoch": 0.1266629571551419, "grad_norm": 9.030415534973145, "learning_rate": 9.661337089266683e-06, "loss": 0.4227, "step": 2504 }, { "epoch": 0.12671354140320704, "grad_norm": 7.356062412261963, "learning_rate": 9.661046667538135e-06, "loss": 0.4764, "step": 2505 }, { "epoch": 0.12676412565127218, "grad_norm": 20.34926414489746, "learning_rate": 9.660756125705673e-06, "loss": 0.6312, "step": 2506 }, { "epoch": 0.12681470989933735, "grad_norm": 11.231744766235352, "learning_rate": 9.660465463776784e-06, "loss": 0.5101, "step": 2507 }, { "epoch": 0.1268652941474025, "grad_norm": 22.482961654663086, "learning_rate": 9.660174681758955e-06, "loss": 0.6556, "step": 2508 }, { "epoch": 0.12691587839546764, "grad_norm": 11.245149612426758, "learning_rate": 9.659883779659683e-06, "loss": 0.6741, "step": 2509 }, { "epoch": 0.1269664626435328, "grad_norm": 7.776262283325195, "learning_rate": 9.65959275748646e-06, "loss": 0.4142, "step": 2510 }, { "epoch": 0.12701704689159796, "grad_norm": 9.95458984375, "learning_rate": 9.659301615246786e-06, "loss": 0.5456, "step": 2511 }, { "epoch": 0.1270676311396631, "grad_norm": 12.919581413269043, "learning_rate": 9.659010352948164e-06, "loss": 0.576, "step": 2512 }, { "epoch": 0.12711821538772827, "grad_norm": 12.349946975708008, "learning_rate": 9.6587189705981e-06, "loss": 0.4833, "step": 2513 }, { "epoch": 0.12716879963579342, "grad_norm": 10.003448486328125, "learning_rate": 9.658427468204098e-06, "loss": 0.4429, "step": 2514 }, { "epoch": 0.12721938388385856, "grad_norm": 5.745459079742432, "learning_rate": 9.658135845773676e-06, "loss": 0.4115, "step": 2515 }, { "epoch": 0.1272699681319237, "grad_norm": 7.057268142700195, "learning_rate": 9.65784410331434e-06, "loss": 0.5642, "step": 2516 }, { "epoch": 0.12732055237998888, "grad_norm": 15.231910705566406, "learning_rate": 9.657552240833614e-06, "loss": 0.4406, "step": 2517 }, { "epoch": 0.12737113662805402, "grad_norm": 7.419838905334473, "learning_rate": 9.657260258339017e-06, "loss": 0.4274, "step": 2518 }, { "epoch": 0.12742172087611917, "grad_norm": 8.076871871948242, "learning_rate": 9.656968155838074e-06, "loss": 0.4553, "step": 2519 }, { "epoch": 0.12747230512418434, "grad_norm": 5.645204544067383, "learning_rate": 9.656675933338307e-06, "loss": 0.4989, "step": 2520 }, { "epoch": 0.12752288937224948, "grad_norm": 8.172967910766602, "learning_rate": 9.65638359084725e-06, "loss": 0.4576, "step": 2521 }, { "epoch": 0.12757347362031463, "grad_norm": 7.017653942108154, "learning_rate": 9.656091128372435e-06, "loss": 0.3607, "step": 2522 }, { "epoch": 0.1276240578683798, "grad_norm": 7.387547016143799, "learning_rate": 9.655798545921397e-06, "loss": 0.4253, "step": 2523 }, { "epoch": 0.12767464211644494, "grad_norm": 9.571832656860352, "learning_rate": 9.655505843501678e-06, "loss": 0.503, "step": 2524 }, { "epoch": 0.12772522636451009, "grad_norm": 10.433820724487305, "learning_rate": 9.655213021120817e-06, "loss": 0.4844, "step": 2525 }, { "epoch": 0.12777581061257523, "grad_norm": 11.062895774841309, "learning_rate": 9.65492007878636e-06, "loss": 0.4284, "step": 2526 }, { "epoch": 0.1278263948606404, "grad_norm": 10.14175033569336, "learning_rate": 9.654627016505858e-06, "loss": 0.5064, "step": 2527 }, { "epoch": 0.12787697910870555, "grad_norm": 8.435357093811035, "learning_rate": 9.65433383428686e-06, "loss": 0.5683, "step": 2528 }, { "epoch": 0.1279275633567707, "grad_norm": 12.511487007141113, "learning_rate": 9.654040532136921e-06, "loss": 0.5738, "step": 2529 }, { "epoch": 0.12797814760483586, "grad_norm": 5.613940238952637, "learning_rate": 9.653747110063599e-06, "loss": 0.4413, "step": 2530 }, { "epoch": 0.128028731852901, "grad_norm": 9.183568000793457, "learning_rate": 9.653453568074453e-06, "loss": 0.5003, "step": 2531 }, { "epoch": 0.12807931610096615, "grad_norm": 8.009278297424316, "learning_rate": 9.65315990617705e-06, "loss": 0.4322, "step": 2532 }, { "epoch": 0.12812990034903132, "grad_norm": 15.755513191223145, "learning_rate": 9.652866124378956e-06, "loss": 0.5231, "step": 2533 }, { "epoch": 0.12818048459709647, "grad_norm": 6.13640832901001, "learning_rate": 9.65257222268774e-06, "loss": 0.4698, "step": 2534 }, { "epoch": 0.1282310688451616, "grad_norm": 10.460559844970703, "learning_rate": 9.652278201110976e-06, "loss": 0.4703, "step": 2535 }, { "epoch": 0.12828165309322678, "grad_norm": 6.723852634429932, "learning_rate": 9.651984059656239e-06, "loss": 0.4879, "step": 2536 }, { "epoch": 0.12833223734129193, "grad_norm": 9.267922401428223, "learning_rate": 9.651689798331108e-06, "loss": 0.4476, "step": 2537 }, { "epoch": 0.12838282158935707, "grad_norm": 7.276064395904541, "learning_rate": 9.651395417143169e-06, "loss": 0.4584, "step": 2538 }, { "epoch": 0.12843340583742222, "grad_norm": 10.128077507019043, "learning_rate": 9.651100916100005e-06, "loss": 0.3618, "step": 2539 }, { "epoch": 0.1284839900854874, "grad_norm": 9.259227752685547, "learning_rate": 9.650806295209202e-06, "loss": 0.5375, "step": 2540 }, { "epoch": 0.12853457433355253, "grad_norm": 8.611254692077637, "learning_rate": 9.650511554478357e-06, "loss": 0.4615, "step": 2541 }, { "epoch": 0.12858515858161768, "grad_norm": 21.085901260375977, "learning_rate": 9.650216693915062e-06, "loss": 0.8056, "step": 2542 }, { "epoch": 0.12863574282968285, "grad_norm": 6.598304271697998, "learning_rate": 9.649921713526914e-06, "loss": 0.399, "step": 2543 }, { "epoch": 0.128686327077748, "grad_norm": 7.9935150146484375, "learning_rate": 9.649626613321515e-06, "loss": 0.4445, "step": 2544 }, { "epoch": 0.12873691132581314, "grad_norm": 14.075374603271484, "learning_rate": 9.64933139330647e-06, "loss": 0.5397, "step": 2545 }, { "epoch": 0.1287874955738783, "grad_norm": 11.391390800476074, "learning_rate": 9.649036053489382e-06, "loss": 0.4593, "step": 2546 }, { "epoch": 0.12883807982194345, "grad_norm": 9.619405746459961, "learning_rate": 9.648740593877866e-06, "loss": 0.4511, "step": 2547 }, { "epoch": 0.1288886640700086, "grad_norm": 11.586901664733887, "learning_rate": 9.648445014479536e-06, "loss": 0.5199, "step": 2548 }, { "epoch": 0.12893924831807374, "grad_norm": 12.615777969360352, "learning_rate": 9.648149315302003e-06, "loss": 0.5622, "step": 2549 }, { "epoch": 0.1289898325661389, "grad_norm": 7.934619903564453, "learning_rate": 9.647853496352889e-06, "loss": 0.3903, "step": 2550 }, { "epoch": 0.12904041681420406, "grad_norm": 10.882477760314941, "learning_rate": 9.647557557639817e-06, "loss": 0.4834, "step": 2551 }, { "epoch": 0.1290910010622692, "grad_norm": 8.090531349182129, "learning_rate": 9.647261499170412e-06, "loss": 0.4913, "step": 2552 }, { "epoch": 0.12914158531033437, "grad_norm": 11.052908897399902, "learning_rate": 9.646965320952307e-06, "loss": 0.3851, "step": 2553 }, { "epoch": 0.12919216955839952, "grad_norm": 13.558370590209961, "learning_rate": 9.646669022993125e-06, "loss": 0.557, "step": 2554 }, { "epoch": 0.12924275380646466, "grad_norm": 17.815326690673828, "learning_rate": 9.64637260530051e-06, "loss": 0.486, "step": 2555 }, { "epoch": 0.12929333805452983, "grad_norm": 14.99227523803711, "learning_rate": 9.646076067882093e-06, "loss": 0.5902, "step": 2556 }, { "epoch": 0.12934392230259498, "grad_norm": 13.00120735168457, "learning_rate": 9.645779410745518e-06, "loss": 0.5346, "step": 2557 }, { "epoch": 0.12939450655066012, "grad_norm": 17.50697898864746, "learning_rate": 9.64548263389843e-06, "loss": 0.4651, "step": 2558 }, { "epoch": 0.12944509079872527, "grad_norm": 11.622659683227539, "learning_rate": 9.645185737348475e-06, "loss": 0.4128, "step": 2559 }, { "epoch": 0.12949567504679044, "grad_norm": 6.819965839385986, "learning_rate": 9.644888721103302e-06, "loss": 0.3507, "step": 2560 }, { "epoch": 0.12954625929485558, "grad_norm": 13.307674407958984, "learning_rate": 9.644591585170568e-06, "loss": 0.4208, "step": 2561 }, { "epoch": 0.12959684354292073, "grad_norm": 11.119824409484863, "learning_rate": 9.644294329557926e-06, "loss": 0.6205, "step": 2562 }, { "epoch": 0.1296474277909859, "grad_norm": 6.66431188583374, "learning_rate": 9.643996954273038e-06, "loss": 0.4504, "step": 2563 }, { "epoch": 0.12969801203905104, "grad_norm": 6.960129737854004, "learning_rate": 9.643699459323566e-06, "loss": 0.3553, "step": 2564 }, { "epoch": 0.12974859628711619, "grad_norm": 12.766446113586426, "learning_rate": 9.643401844717175e-06, "loss": 0.5373, "step": 2565 }, { "epoch": 0.12979918053518136, "grad_norm": 6.996239185333252, "learning_rate": 9.643104110461535e-06, "loss": 0.4232, "step": 2566 }, { "epoch": 0.1298497647832465, "grad_norm": 23.52566146850586, "learning_rate": 9.642806256564315e-06, "loss": 0.6535, "step": 2567 }, { "epoch": 0.12990034903131165, "grad_norm": 7.415971279144287, "learning_rate": 9.642508283033193e-06, "loss": 0.5096, "step": 2568 }, { "epoch": 0.1299509332793768, "grad_norm": 24.765167236328125, "learning_rate": 9.642210189875845e-06, "loss": 0.5525, "step": 2569 }, { "epoch": 0.13000151752744196, "grad_norm": 8.810222625732422, "learning_rate": 9.641911977099955e-06, "loss": 0.3891, "step": 2570 }, { "epoch": 0.1300521017755071, "grad_norm": 11.167922973632812, "learning_rate": 9.641613644713204e-06, "loss": 0.4516, "step": 2571 }, { "epoch": 0.13010268602357225, "grad_norm": 16.498537063598633, "learning_rate": 9.641315192723282e-06, "loss": 0.4581, "step": 2572 }, { "epoch": 0.13015327027163742, "grad_norm": 7.615708351135254, "learning_rate": 9.641016621137878e-06, "loss": 0.4461, "step": 2573 }, { "epoch": 0.13020385451970257, "grad_norm": 9.551617622375488, "learning_rate": 9.640717929964685e-06, "loss": 0.4378, "step": 2574 }, { "epoch": 0.1302544387677677, "grad_norm": 8.0595121383667, "learning_rate": 9.640419119211399e-06, "loss": 0.5201, "step": 2575 }, { "epoch": 0.13030502301583288, "grad_norm": 8.866692543029785, "learning_rate": 9.640120188885722e-06, "loss": 0.4394, "step": 2576 }, { "epoch": 0.13035560726389803, "grad_norm": 11.00913143157959, "learning_rate": 9.639821138995355e-06, "loss": 0.468, "step": 2577 }, { "epoch": 0.13040619151196317, "grad_norm": 5.275252342224121, "learning_rate": 9.639521969548005e-06, "loss": 0.4878, "step": 2578 }, { "epoch": 0.13045677576002832, "grad_norm": 6.79263973236084, "learning_rate": 9.63922268055138e-06, "loss": 0.461, "step": 2579 }, { "epoch": 0.1305073600080935, "grad_norm": 8.63471794128418, "learning_rate": 9.638923272013192e-06, "loss": 0.4451, "step": 2580 }, { "epoch": 0.13055794425615863, "grad_norm": 16.127065658569336, "learning_rate": 9.638623743941156e-06, "loss": 0.4574, "step": 2581 }, { "epoch": 0.13060852850422378, "grad_norm": 8.728680610656738, "learning_rate": 9.63832409634299e-06, "loss": 0.4825, "step": 2582 }, { "epoch": 0.13065911275228895, "grad_norm": 9.373953819274902, "learning_rate": 9.638024329226417e-06, "loss": 0.4885, "step": 2583 }, { "epoch": 0.1307096970003541, "grad_norm": 16.382028579711914, "learning_rate": 9.637724442599159e-06, "loss": 0.6116, "step": 2584 }, { "epoch": 0.13076028124841924, "grad_norm": 13.594433784484863, "learning_rate": 9.637424436468943e-06, "loss": 0.5273, "step": 2585 }, { "epoch": 0.1308108654964844, "grad_norm": 10.827180862426758, "learning_rate": 9.637124310843499e-06, "loss": 0.5214, "step": 2586 }, { "epoch": 0.13086144974454955, "grad_norm": 10.571057319641113, "learning_rate": 9.636824065730566e-06, "loss": 0.554, "step": 2587 }, { "epoch": 0.1309120339926147, "grad_norm": 6.534485340118408, "learning_rate": 9.636523701137874e-06, "loss": 0.4426, "step": 2588 }, { "epoch": 0.13096261824067984, "grad_norm": 9.5880126953125, "learning_rate": 9.636223217073166e-06, "loss": 0.3871, "step": 2589 }, { "epoch": 0.131013202488745, "grad_norm": 8.363626480102539, "learning_rate": 9.635922613544184e-06, "loss": 0.3735, "step": 2590 }, { "epoch": 0.13106378673681016, "grad_norm": 9.655451774597168, "learning_rate": 9.635621890558674e-06, "loss": 0.4422, "step": 2591 }, { "epoch": 0.1311143709848753, "grad_norm": 10.38428020477295, "learning_rate": 9.635321048124383e-06, "loss": 0.5399, "step": 2592 }, { "epoch": 0.13116495523294047, "grad_norm": 13.103337287902832, "learning_rate": 9.635020086249069e-06, "loss": 0.4796, "step": 2593 }, { "epoch": 0.13121553948100562, "grad_norm": 17.615333557128906, "learning_rate": 9.63471900494048e-06, "loss": 0.6303, "step": 2594 }, { "epoch": 0.13126612372907076, "grad_norm": 5.646346092224121, "learning_rate": 9.634417804206377e-06, "loss": 0.3891, "step": 2595 }, { "epoch": 0.13131670797713593, "grad_norm": 12.944222450256348, "learning_rate": 9.634116484054522e-06, "loss": 0.511, "step": 2596 }, { "epoch": 0.13136729222520108, "grad_norm": 8.859893798828125, "learning_rate": 9.633815044492677e-06, "loss": 0.392, "step": 2597 }, { "epoch": 0.13141787647326622, "grad_norm": 39.81491470336914, "learning_rate": 9.633513485528612e-06, "loss": 0.6124, "step": 2598 }, { "epoch": 0.13146846072133136, "grad_norm": 8.932940483093262, "learning_rate": 9.633211807170096e-06, "loss": 0.5023, "step": 2599 }, { "epoch": 0.13151904496939654, "grad_norm": 12.385199546813965, "learning_rate": 9.632910009424903e-06, "loss": 0.4787, "step": 2600 }, { "epoch": 0.13156962921746168, "grad_norm": 7.904058933258057, "learning_rate": 9.632608092300808e-06, "loss": 0.4518, "step": 2601 }, { "epoch": 0.13162021346552683, "grad_norm": 9.717973709106445, "learning_rate": 9.632306055805594e-06, "loss": 0.4193, "step": 2602 }, { "epoch": 0.131670797713592, "grad_norm": 8.474749565124512, "learning_rate": 9.63200389994704e-06, "loss": 0.4058, "step": 2603 }, { "epoch": 0.13172138196165714, "grad_norm": 19.917293548583984, "learning_rate": 9.631701624732935e-06, "loss": 0.4524, "step": 2604 }, { "epoch": 0.13177196620972229, "grad_norm": 8.203598022460938, "learning_rate": 9.631399230171066e-06, "loss": 0.4248, "step": 2605 }, { "epoch": 0.13182255045778746, "grad_norm": 18.384462356567383, "learning_rate": 9.631096716269226e-06, "loss": 0.8051, "step": 2606 }, { "epoch": 0.1318731347058526, "grad_norm": 8.753592491149902, "learning_rate": 9.630794083035209e-06, "loss": 0.458, "step": 2607 }, { "epoch": 0.13192371895391775, "grad_norm": 11.73144245147705, "learning_rate": 9.630491330476815e-06, "loss": 0.4476, "step": 2608 }, { "epoch": 0.1319743032019829, "grad_norm": 14.944291114807129, "learning_rate": 9.630188458601842e-06, "loss": 0.4636, "step": 2609 }, { "epoch": 0.13202488745004806, "grad_norm": 14.85410213470459, "learning_rate": 9.629885467418098e-06, "loss": 0.5524, "step": 2610 }, { "epoch": 0.1320754716981132, "grad_norm": 6.6043572425842285, "learning_rate": 9.629582356933386e-06, "loss": 0.4387, "step": 2611 }, { "epoch": 0.13212605594617835, "grad_norm": 7.711194038391113, "learning_rate": 9.629279127155523e-06, "loss": 0.4479, "step": 2612 }, { "epoch": 0.13217664019424352, "grad_norm": 8.41914176940918, "learning_rate": 9.628975778092314e-06, "loss": 0.4372, "step": 2613 }, { "epoch": 0.13222722444230867, "grad_norm": 7.477046012878418, "learning_rate": 9.628672309751583e-06, "loss": 0.4438, "step": 2614 }, { "epoch": 0.1322778086903738, "grad_norm": 9.612503051757812, "learning_rate": 9.628368722141146e-06, "loss": 0.3806, "step": 2615 }, { "epoch": 0.13232839293843898, "grad_norm": 9.306982040405273, "learning_rate": 9.628065015268825e-06, "loss": 0.47, "step": 2616 }, { "epoch": 0.13237897718650413, "grad_norm": 11.653129577636719, "learning_rate": 9.627761189142448e-06, "loss": 0.497, "step": 2617 }, { "epoch": 0.13242956143456927, "grad_norm": 10.497214317321777, "learning_rate": 9.627457243769844e-06, "loss": 0.4863, "step": 2618 }, { "epoch": 0.13248014568263441, "grad_norm": 14.283693313598633, "learning_rate": 9.627153179158844e-06, "loss": 0.5332, "step": 2619 }, { "epoch": 0.1325307299306996, "grad_norm": 11.651822090148926, "learning_rate": 9.626848995317281e-06, "loss": 0.5207, "step": 2620 }, { "epoch": 0.13258131417876473, "grad_norm": 7.35696268081665, "learning_rate": 9.626544692252998e-06, "loss": 0.5168, "step": 2621 }, { "epoch": 0.13263189842682988, "grad_norm": 15.221226692199707, "learning_rate": 9.626240269973831e-06, "loss": 0.5154, "step": 2622 }, { "epoch": 0.13268248267489505, "grad_norm": 9.355064392089844, "learning_rate": 9.625935728487628e-06, "loss": 0.5446, "step": 2623 }, { "epoch": 0.1327330669229602, "grad_norm": 6.883498668670654, "learning_rate": 9.625631067802232e-06, "loss": 0.3418, "step": 2624 }, { "epoch": 0.13278365117102534, "grad_norm": 15.384919166564941, "learning_rate": 9.6253262879255e-06, "loss": 0.4552, "step": 2625 }, { "epoch": 0.1328342354190905, "grad_norm": 10.035958290100098, "learning_rate": 9.625021388865278e-06, "loss": 0.505, "step": 2626 }, { "epoch": 0.13288481966715565, "grad_norm": 7.467972755432129, "learning_rate": 9.624716370629429e-06, "loss": 0.3159, "step": 2627 }, { "epoch": 0.1329354039152208, "grad_norm": 12.198954582214355, "learning_rate": 9.624411233225808e-06, "loss": 0.5691, "step": 2628 }, { "epoch": 0.13298598816328594, "grad_norm": 12.322981834411621, "learning_rate": 9.624105976662279e-06, "loss": 0.4934, "step": 2629 }, { "epoch": 0.1330365724113511, "grad_norm": 6.095273017883301, "learning_rate": 9.623800600946709e-06, "loss": 0.4599, "step": 2630 }, { "epoch": 0.13308715665941626, "grad_norm": 9.945419311523438, "learning_rate": 9.623495106086967e-06, "loss": 0.4699, "step": 2631 }, { "epoch": 0.1331377409074814, "grad_norm": 6.613058090209961, "learning_rate": 9.62318949209092e-06, "loss": 0.4798, "step": 2632 }, { "epoch": 0.13318832515554657, "grad_norm": 8.568361282348633, "learning_rate": 9.62288375896645e-06, "loss": 0.4555, "step": 2633 }, { "epoch": 0.13323890940361172, "grad_norm": 6.878365993499756, "learning_rate": 9.622577906721432e-06, "loss": 0.4929, "step": 2634 }, { "epoch": 0.13328949365167686, "grad_norm": 10.610742568969727, "learning_rate": 9.622271935363745e-06, "loss": 0.4643, "step": 2635 }, { "epoch": 0.13334007789974203, "grad_norm": 7.030160903930664, "learning_rate": 9.621965844901275e-06, "loss": 0.3332, "step": 2636 }, { "epoch": 0.13339066214780718, "grad_norm": 10.014638900756836, "learning_rate": 9.621659635341908e-06, "loss": 0.4302, "step": 2637 }, { "epoch": 0.13344124639587232, "grad_norm": 9.19017219543457, "learning_rate": 9.621353306693538e-06, "loss": 0.4537, "step": 2638 }, { "epoch": 0.13349183064393746, "grad_norm": 12.63023567199707, "learning_rate": 9.621046858964054e-06, "loss": 0.6049, "step": 2639 }, { "epoch": 0.13354241489200264, "grad_norm": 6.96691370010376, "learning_rate": 9.620740292161354e-06, "loss": 0.4164, "step": 2640 }, { "epoch": 0.13359299914006778, "grad_norm": 7.751502990722656, "learning_rate": 9.62043360629334e-06, "loss": 0.3537, "step": 2641 }, { "epoch": 0.13364358338813292, "grad_norm": 8.62781047821045, "learning_rate": 9.62012680136791e-06, "loss": 0.4449, "step": 2642 }, { "epoch": 0.1336941676361981, "grad_norm": 7.8449602127075195, "learning_rate": 9.61981987739297e-06, "loss": 0.3094, "step": 2643 }, { "epoch": 0.13374475188426324, "grad_norm": 13.232071876525879, "learning_rate": 9.619512834376435e-06, "loss": 0.4853, "step": 2644 }, { "epoch": 0.13379533613232839, "grad_norm": 7.94553279876709, "learning_rate": 9.619205672326208e-06, "loss": 0.441, "step": 2645 }, { "epoch": 0.13384592038039356, "grad_norm": 8.526204109191895, "learning_rate": 9.61889839125021e-06, "loss": 0.4508, "step": 2646 }, { "epoch": 0.1338965046284587, "grad_norm": 10.32691478729248, "learning_rate": 9.618590991156358e-06, "loss": 0.5182, "step": 2647 }, { "epoch": 0.13394708887652385, "grad_norm": 7.964968204498291, "learning_rate": 9.61828347205257e-06, "loss": 0.4092, "step": 2648 }, { "epoch": 0.133997673124589, "grad_norm": 11.57717514038086, "learning_rate": 9.617975833946773e-06, "loss": 0.562, "step": 2649 }, { "epoch": 0.13404825737265416, "grad_norm": 17.460506439208984, "learning_rate": 9.617668076846892e-06, "loss": 0.4616, "step": 2650 }, { "epoch": 0.1340988416207193, "grad_norm": 19.62324333190918, "learning_rate": 9.617360200760858e-06, "loss": 0.6184, "step": 2651 }, { "epoch": 0.13414942586878445, "grad_norm": 9.106337547302246, "learning_rate": 9.617052205696607e-06, "loss": 0.4899, "step": 2652 }, { "epoch": 0.13420001011684962, "grad_norm": 7.074557781219482, "learning_rate": 9.61674409166207e-06, "loss": 0.4663, "step": 2653 }, { "epoch": 0.13425059436491477, "grad_norm": 9.855308532714844, "learning_rate": 9.616435858665189e-06, "loss": 0.4258, "step": 2654 }, { "epoch": 0.1343011786129799, "grad_norm": 15.158440589904785, "learning_rate": 9.616127506713908e-06, "loss": 0.521, "step": 2655 }, { "epoch": 0.13435176286104508, "grad_norm": 15.648083686828613, "learning_rate": 9.615819035816172e-06, "loss": 0.5782, "step": 2656 }, { "epoch": 0.13440234710911023, "grad_norm": 11.402193069458008, "learning_rate": 9.615510445979925e-06, "loss": 0.4797, "step": 2657 }, { "epoch": 0.13445293135717537, "grad_norm": 14.187507629394531, "learning_rate": 9.615201737213125e-06, "loss": 0.5944, "step": 2658 }, { "epoch": 0.13450351560524051, "grad_norm": 8.963325500488281, "learning_rate": 9.614892909523723e-06, "loss": 0.5369, "step": 2659 }, { "epoch": 0.1345540998533057, "grad_norm": 11.592888832092285, "learning_rate": 9.614583962919676e-06, "loss": 0.4135, "step": 2660 }, { "epoch": 0.13460468410137083, "grad_norm": 10.299789428710938, "learning_rate": 9.614274897408949e-06, "loss": 0.4813, "step": 2661 }, { "epoch": 0.13465526834943597, "grad_norm": 8.410541534423828, "learning_rate": 9.613965712999503e-06, "loss": 0.4322, "step": 2662 }, { "epoch": 0.13470585259750115, "grad_norm": 14.311128616333008, "learning_rate": 9.613656409699303e-06, "loss": 0.507, "step": 2663 }, { "epoch": 0.1347564368455663, "grad_norm": 8.135560989379883, "learning_rate": 9.613346987516322e-06, "loss": 0.5334, "step": 2664 }, { "epoch": 0.13480702109363144, "grad_norm": 15.295074462890625, "learning_rate": 9.613037446458533e-06, "loss": 0.5511, "step": 2665 }, { "epoch": 0.1348576053416966, "grad_norm": 6.580499172210693, "learning_rate": 9.612727786533911e-06, "loss": 0.3918, "step": 2666 }, { "epoch": 0.13490818958976175, "grad_norm": 6.999154567718506, "learning_rate": 9.612418007750435e-06, "loss": 0.4167, "step": 2667 }, { "epoch": 0.1349587738378269, "grad_norm": 8.198392868041992, "learning_rate": 9.612108110116087e-06, "loss": 0.4974, "step": 2668 }, { "epoch": 0.13500935808589204, "grad_norm": 8.08952808380127, "learning_rate": 9.611798093638855e-06, "loss": 0.6012, "step": 2669 }, { "epoch": 0.1350599423339572, "grad_norm": 8.612844467163086, "learning_rate": 9.611487958326724e-06, "loss": 0.519, "step": 2670 }, { "epoch": 0.13511052658202236, "grad_norm": 9.908483505249023, "learning_rate": 9.611177704187689e-06, "loss": 0.4481, "step": 2671 }, { "epoch": 0.1351611108300875, "grad_norm": 6.549832344055176, "learning_rate": 9.610867331229742e-06, "loss": 0.424, "step": 2672 }, { "epoch": 0.13521169507815267, "grad_norm": 6.642735481262207, "learning_rate": 9.610556839460878e-06, "loss": 0.4391, "step": 2673 }, { "epoch": 0.13526227932621782, "grad_norm": 5.609287261962891, "learning_rate": 9.610246228889101e-06, "loss": 0.465, "step": 2674 }, { "epoch": 0.13531286357428296, "grad_norm": 9.741748809814453, "learning_rate": 9.609935499522417e-06, "loss": 0.4453, "step": 2675 }, { "epoch": 0.13536344782234813, "grad_norm": 20.444244384765625, "learning_rate": 9.60962465136883e-06, "loss": 0.5644, "step": 2676 }, { "epoch": 0.13541403207041328, "grad_norm": 11.794920921325684, "learning_rate": 9.609313684436347e-06, "loss": 0.606, "step": 2677 }, { "epoch": 0.13546461631847842, "grad_norm": 11.244303703308105, "learning_rate": 9.609002598732985e-06, "loss": 0.4092, "step": 2678 }, { "epoch": 0.1355152005665436, "grad_norm": 9.037519454956055, "learning_rate": 9.608691394266761e-06, "loss": 0.5239, "step": 2679 }, { "epoch": 0.13556578481460874, "grad_norm": 6.583184242248535, "learning_rate": 9.60838007104569e-06, "loss": 0.4037, "step": 2680 }, { "epoch": 0.13561636906267388, "grad_norm": 11.021442413330078, "learning_rate": 9.608068629077793e-06, "loss": 0.546, "step": 2681 }, { "epoch": 0.13566695331073902, "grad_norm": 11.144291877746582, "learning_rate": 9.6077570683711e-06, "loss": 0.5841, "step": 2682 }, { "epoch": 0.1357175375588042, "grad_norm": 8.138762474060059, "learning_rate": 9.607445388933638e-06, "loss": 0.4146, "step": 2683 }, { "epoch": 0.13576812180686934, "grad_norm": 9.011852264404297, "learning_rate": 9.607133590773436e-06, "loss": 0.4827, "step": 2684 }, { "epoch": 0.13581870605493448, "grad_norm": 18.06553077697754, "learning_rate": 9.60682167389853e-06, "loss": 0.7043, "step": 2685 }, { "epoch": 0.13586929030299966, "grad_norm": 9.6708402633667, "learning_rate": 9.606509638316956e-06, "loss": 0.468, "step": 2686 }, { "epoch": 0.1359198745510648, "grad_norm": 13.696710586547852, "learning_rate": 9.606197484036754e-06, "loss": 0.5144, "step": 2687 }, { "epoch": 0.13597045879912995, "grad_norm": 9.763338088989258, "learning_rate": 9.60588521106597e-06, "loss": 0.3518, "step": 2688 }, { "epoch": 0.13602104304719512, "grad_norm": 8.145801544189453, "learning_rate": 9.60557281941265e-06, "loss": 0.4415, "step": 2689 }, { "epoch": 0.13607162729526026, "grad_norm": 25.039955139160156, "learning_rate": 9.605260309084843e-06, "loss": 0.594, "step": 2690 }, { "epoch": 0.1361222115433254, "grad_norm": 10.66859245300293, "learning_rate": 9.604947680090599e-06, "loss": 0.5239, "step": 2691 }, { "epoch": 0.13617279579139055, "grad_norm": 8.784505844116211, "learning_rate": 9.604634932437979e-06, "loss": 0.4762, "step": 2692 }, { "epoch": 0.13622338003945572, "grad_norm": 9.701227188110352, "learning_rate": 9.604322066135037e-06, "loss": 0.4657, "step": 2693 }, { "epoch": 0.13627396428752087, "grad_norm": 4.926166534423828, "learning_rate": 9.604009081189837e-06, "loss": 0.4173, "step": 2694 }, { "epoch": 0.136324548535586, "grad_norm": 9.745553016662598, "learning_rate": 9.603695977610442e-06, "loss": 0.4076, "step": 2695 }, { "epoch": 0.13637513278365118, "grad_norm": 7.568397521972656, "learning_rate": 9.603382755404924e-06, "loss": 0.4441, "step": 2696 }, { "epoch": 0.13642571703171633, "grad_norm": 9.07861042022705, "learning_rate": 9.603069414581352e-06, "loss": 0.5402, "step": 2697 }, { "epoch": 0.13647630127978147, "grad_norm": 7.204952716827393, "learning_rate": 9.602755955147796e-06, "loss": 0.5046, "step": 2698 }, { "epoch": 0.13652688552784664, "grad_norm": 10.302748680114746, "learning_rate": 9.602442377112339e-06, "loss": 0.5627, "step": 2699 }, { "epoch": 0.1365774697759118, "grad_norm": 8.381875038146973, "learning_rate": 9.602128680483058e-06, "loss": 0.3785, "step": 2700 }, { "epoch": 0.13662805402397693, "grad_norm": 13.105636596679688, "learning_rate": 9.601814865268037e-06, "loss": 0.4381, "step": 2701 }, { "epoch": 0.13667863827204207, "grad_norm": 8.685811042785645, "learning_rate": 9.601500931475362e-06, "loss": 0.3083, "step": 2702 }, { "epoch": 0.13672922252010725, "grad_norm": 6.679476737976074, "learning_rate": 9.601186879113123e-06, "loss": 0.4675, "step": 2703 }, { "epoch": 0.1367798067681724, "grad_norm": 8.97934627532959, "learning_rate": 9.60087270818941e-06, "loss": 0.4411, "step": 2704 }, { "epoch": 0.13683039101623753, "grad_norm": 12.814411163330078, "learning_rate": 9.600558418712322e-06, "loss": 0.512, "step": 2705 }, { "epoch": 0.1368809752643027, "grad_norm": 8.125356674194336, "learning_rate": 9.600244010689956e-06, "loss": 0.5265, "step": 2706 }, { "epoch": 0.13693155951236785, "grad_norm": 8.556879043579102, "learning_rate": 9.599929484130413e-06, "loss": 0.4802, "step": 2707 }, { "epoch": 0.136982143760433, "grad_norm": 11.421368598937988, "learning_rate": 9.599614839041797e-06, "loss": 0.4193, "step": 2708 }, { "epoch": 0.13703272800849817, "grad_norm": 8.21834945678711, "learning_rate": 9.599300075432216e-06, "loss": 0.5662, "step": 2709 }, { "epoch": 0.1370833122565633, "grad_norm": 8.79266357421875, "learning_rate": 9.598985193309783e-06, "loss": 0.501, "step": 2710 }, { "epoch": 0.13713389650462846, "grad_norm": 6.712255954742432, "learning_rate": 9.598670192682609e-06, "loss": 0.5125, "step": 2711 }, { "epoch": 0.1371844807526936, "grad_norm": 14.579849243164062, "learning_rate": 9.59835507355881e-06, "loss": 0.6207, "step": 2712 }, { "epoch": 0.13723506500075877, "grad_norm": 11.51487922668457, "learning_rate": 9.598039835946511e-06, "loss": 0.4604, "step": 2713 }, { "epoch": 0.13728564924882392, "grad_norm": 10.919276237487793, "learning_rate": 9.597724479853831e-06, "loss": 0.4477, "step": 2714 }, { "epoch": 0.13733623349688906, "grad_norm": 10.799629211425781, "learning_rate": 9.597409005288896e-06, "loss": 0.4884, "step": 2715 }, { "epoch": 0.13738681774495423, "grad_norm": 8.438838958740234, "learning_rate": 9.597093412259835e-06, "loss": 0.4439, "step": 2716 }, { "epoch": 0.13743740199301938, "grad_norm": 10.448816299438477, "learning_rate": 9.59677770077478e-06, "loss": 0.5891, "step": 2717 }, { "epoch": 0.13748798624108452, "grad_norm": 7.128586769104004, "learning_rate": 9.596461870841868e-06, "loss": 0.4698, "step": 2718 }, { "epoch": 0.1375385704891497, "grad_norm": 6.464155197143555, "learning_rate": 9.596145922469235e-06, "loss": 0.4118, "step": 2719 }, { "epoch": 0.13758915473721484, "grad_norm": 15.734929084777832, "learning_rate": 9.595829855665023e-06, "loss": 0.5828, "step": 2720 }, { "epoch": 0.13763973898527998, "grad_norm": 8.171501159667969, "learning_rate": 9.595513670437378e-06, "loss": 0.4852, "step": 2721 }, { "epoch": 0.13769032323334512, "grad_norm": 7.108001232147217, "learning_rate": 9.595197366794444e-06, "loss": 0.4606, "step": 2722 }, { "epoch": 0.1377409074814103, "grad_norm": 5.738302707672119, "learning_rate": 9.594880944744373e-06, "loss": 0.3721, "step": 2723 }, { "epoch": 0.13779149172947544, "grad_norm": 8.443678855895996, "learning_rate": 9.594564404295319e-06, "loss": 0.4554, "step": 2724 }, { "epoch": 0.13784207597754058, "grad_norm": 11.010359764099121, "learning_rate": 9.594247745455439e-06, "loss": 0.3951, "step": 2725 }, { "epoch": 0.13789266022560576, "grad_norm": 7.2123589515686035, "learning_rate": 9.593930968232891e-06, "loss": 0.5056, "step": 2726 }, { "epoch": 0.1379432444736709, "grad_norm": 10.402933120727539, "learning_rate": 9.593614072635837e-06, "loss": 0.5832, "step": 2727 }, { "epoch": 0.13799382872173604, "grad_norm": 8.211827278137207, "learning_rate": 9.593297058672444e-06, "loss": 0.5158, "step": 2728 }, { "epoch": 0.13804441296980122, "grad_norm": 9.481792449951172, "learning_rate": 9.592979926350883e-06, "loss": 0.5674, "step": 2729 }, { "epoch": 0.13809499721786636, "grad_norm": 8.385343551635742, "learning_rate": 9.59266267567932e-06, "loss": 0.4349, "step": 2730 }, { "epoch": 0.1381455814659315, "grad_norm": 6.742096900939941, "learning_rate": 9.592345306665935e-06, "loss": 0.3806, "step": 2731 }, { "epoch": 0.13819616571399665, "grad_norm": 29.97891616821289, "learning_rate": 9.592027819318903e-06, "loss": 0.5892, "step": 2732 }, { "epoch": 0.13824674996206182, "grad_norm": 6.013443470001221, "learning_rate": 9.591710213646406e-06, "loss": 0.4816, "step": 2733 }, { "epoch": 0.13829733421012697, "grad_norm": 9.124791145324707, "learning_rate": 9.591392489656627e-06, "loss": 0.5189, "step": 2734 }, { "epoch": 0.1383479184581921, "grad_norm": 19.03909683227539, "learning_rate": 9.591074647357755e-06, "loss": 0.5548, "step": 2735 }, { "epoch": 0.13839850270625728, "grad_norm": 13.602429389953613, "learning_rate": 9.590756686757975e-06, "loss": 0.4671, "step": 2736 }, { "epoch": 0.13844908695432243, "grad_norm": 7.03361701965332, "learning_rate": 9.590438607865488e-06, "loss": 0.3724, "step": 2737 }, { "epoch": 0.13849967120238757, "grad_norm": 7.161037921905518, "learning_rate": 9.590120410688484e-06, "loss": 0.3947, "step": 2738 }, { "epoch": 0.13855025545045274, "grad_norm": 22.021167755126953, "learning_rate": 9.589802095235163e-06, "loss": 0.5143, "step": 2739 }, { "epoch": 0.13860083969851789, "grad_norm": 8.580223083496094, "learning_rate": 9.58948366151373e-06, "loss": 0.4955, "step": 2740 }, { "epoch": 0.13865142394658303, "grad_norm": 10.326412200927734, "learning_rate": 9.589165109532387e-06, "loss": 0.5054, "step": 2741 }, { "epoch": 0.13870200819464817, "grad_norm": 10.6058988571167, "learning_rate": 9.588846439299347e-06, "loss": 0.4297, "step": 2742 }, { "epoch": 0.13875259244271335, "grad_norm": 11.97280216217041, "learning_rate": 9.588527650822814e-06, "loss": 0.5131, "step": 2743 }, { "epoch": 0.1388031766907785, "grad_norm": 11.411886215209961, "learning_rate": 9.58820874411101e-06, "loss": 0.5805, "step": 2744 }, { "epoch": 0.13885376093884363, "grad_norm": 14.001184463500977, "learning_rate": 9.587889719172147e-06, "loss": 0.6704, "step": 2745 }, { "epoch": 0.1389043451869088, "grad_norm": 7.651937007904053, "learning_rate": 9.587570576014446e-06, "loss": 0.3922, "step": 2746 }, { "epoch": 0.13895492943497395, "grad_norm": 10.955451965332031, "learning_rate": 9.587251314646136e-06, "loss": 0.4966, "step": 2747 }, { "epoch": 0.1390055136830391, "grad_norm": 8.433907508850098, "learning_rate": 9.586931935075437e-06, "loss": 0.498, "step": 2748 }, { "epoch": 0.13905609793110427, "grad_norm": 7.715710639953613, "learning_rate": 9.586612437310583e-06, "loss": 0.4211, "step": 2749 }, { "epoch": 0.1391066821791694, "grad_norm": 11.249920845031738, "learning_rate": 9.586292821359804e-06, "loss": 0.4559, "step": 2750 }, { "epoch": 0.13915726642723455, "grad_norm": 10.609482765197754, "learning_rate": 9.585973087231336e-06, "loss": 0.4823, "step": 2751 }, { "epoch": 0.1392078506752997, "grad_norm": 19.533945083618164, "learning_rate": 9.58565323493342e-06, "loss": 0.6498, "step": 2752 }, { "epoch": 0.13925843492336487, "grad_norm": 6.103965759277344, "learning_rate": 9.585333264474294e-06, "loss": 0.3736, "step": 2753 }, { "epoch": 0.13930901917143002, "grad_norm": 5.719281196594238, "learning_rate": 9.585013175862208e-06, "loss": 0.4889, "step": 2754 }, { "epoch": 0.13935960341949516, "grad_norm": 10.84570598602295, "learning_rate": 9.584692969105404e-06, "loss": 0.4228, "step": 2755 }, { "epoch": 0.13941018766756033, "grad_norm": 9.997023582458496, "learning_rate": 9.584372644212139e-06, "loss": 0.4825, "step": 2756 }, { "epoch": 0.13946077191562548, "grad_norm": 7.833487510681152, "learning_rate": 9.584052201190663e-06, "loss": 0.4782, "step": 2757 }, { "epoch": 0.13951135616369062, "grad_norm": 10.85949420928955, "learning_rate": 9.583731640049234e-06, "loss": 0.4156, "step": 2758 }, { "epoch": 0.1395619404117558, "grad_norm": 9.578252792358398, "learning_rate": 9.583410960796112e-06, "loss": 0.5211, "step": 2759 }, { "epoch": 0.13961252465982094, "grad_norm": 8.502196311950684, "learning_rate": 9.583090163439562e-06, "loss": 0.4075, "step": 2760 }, { "epoch": 0.13966310890788608, "grad_norm": 8.01046371459961, "learning_rate": 9.582769247987846e-06, "loss": 0.4865, "step": 2761 }, { "epoch": 0.13971369315595122, "grad_norm": 7.611371994018555, "learning_rate": 9.582448214449236e-06, "loss": 0.4659, "step": 2762 }, { "epoch": 0.1397642774040164, "grad_norm": 12.821462631225586, "learning_rate": 9.582127062832004e-06, "loss": 0.4492, "step": 2763 }, { "epoch": 0.13981486165208154, "grad_norm": 10.319415092468262, "learning_rate": 9.581805793144427e-06, "loss": 0.4427, "step": 2764 }, { "epoch": 0.13986544590014668, "grad_norm": 10.181615829467773, "learning_rate": 9.58148440539478e-06, "loss": 0.3995, "step": 2765 }, { "epoch": 0.13991603014821186, "grad_norm": 7.604337215423584, "learning_rate": 9.581162899591347e-06, "loss": 0.3964, "step": 2766 }, { "epoch": 0.139966614396277, "grad_norm": 13.890591621398926, "learning_rate": 9.580841275742411e-06, "loss": 0.5125, "step": 2767 }, { "epoch": 0.14001719864434214, "grad_norm": 11.181181907653809, "learning_rate": 9.58051953385626e-06, "loss": 0.5237, "step": 2768 }, { "epoch": 0.14006778289240732, "grad_norm": 13.956037521362305, "learning_rate": 9.580197673941185e-06, "loss": 0.5242, "step": 2769 }, { "epoch": 0.14011836714047246, "grad_norm": 6.0984673500061035, "learning_rate": 9.579875696005478e-06, "loss": 0.4649, "step": 2770 }, { "epoch": 0.1401689513885376, "grad_norm": 9.442109107971191, "learning_rate": 9.579553600057436e-06, "loss": 0.5005, "step": 2771 }, { "epoch": 0.14021953563660275, "grad_norm": 6.742983341217041, "learning_rate": 9.57923138610536e-06, "loss": 0.4827, "step": 2772 }, { "epoch": 0.14027011988466792, "grad_norm": 11.330294609069824, "learning_rate": 9.578909054157552e-06, "loss": 0.4981, "step": 2773 }, { "epoch": 0.14032070413273307, "grad_norm": 10.935599327087402, "learning_rate": 9.578586604222318e-06, "loss": 0.4907, "step": 2774 }, { "epoch": 0.1403712883807982, "grad_norm": 13.770026206970215, "learning_rate": 9.578264036307967e-06, "loss": 0.6104, "step": 2775 }, { "epoch": 0.14042187262886338, "grad_norm": 11.709412574768066, "learning_rate": 9.577941350422808e-06, "loss": 0.4332, "step": 2776 }, { "epoch": 0.14047245687692853, "grad_norm": 8.61789608001709, "learning_rate": 9.577618546575159e-06, "loss": 0.4253, "step": 2777 }, { "epoch": 0.14052304112499367, "grad_norm": 6.826059341430664, "learning_rate": 9.577295624773337e-06, "loss": 0.3849, "step": 2778 }, { "epoch": 0.14057362537305884, "grad_norm": 10.126702308654785, "learning_rate": 9.576972585025662e-06, "loss": 0.3603, "step": 2779 }, { "epoch": 0.14062420962112399, "grad_norm": 12.966280937194824, "learning_rate": 9.576649427340458e-06, "loss": 0.4464, "step": 2780 }, { "epoch": 0.14067479386918913, "grad_norm": 10.599810600280762, "learning_rate": 9.576326151726055e-06, "loss": 0.4171, "step": 2781 }, { "epoch": 0.14072537811725427, "grad_norm": 8.923599243164062, "learning_rate": 9.576002758190778e-06, "loss": 0.5429, "step": 2782 }, { "epoch": 0.14077596236531945, "grad_norm": 14.839617729187012, "learning_rate": 9.575679246742966e-06, "loss": 0.5639, "step": 2783 }, { "epoch": 0.1408265466133846, "grad_norm": 10.6455717086792, "learning_rate": 9.57535561739095e-06, "loss": 0.539, "step": 2784 }, { "epoch": 0.14087713086144973, "grad_norm": 9.47364616394043, "learning_rate": 9.575031870143073e-06, "loss": 0.4752, "step": 2785 }, { "epoch": 0.1409277151095149, "grad_norm": 8.869380950927734, "learning_rate": 9.574708005007673e-06, "loss": 0.4174, "step": 2786 }, { "epoch": 0.14097829935758005, "grad_norm": 14.632685661315918, "learning_rate": 9.574384021993098e-06, "loss": 0.4215, "step": 2787 }, { "epoch": 0.1410288836056452, "grad_norm": 7.74962043762207, "learning_rate": 9.574059921107694e-06, "loss": 0.4117, "step": 2788 }, { "epoch": 0.14107946785371037, "grad_norm": 8.945911407470703, "learning_rate": 9.573735702359817e-06, "loss": 0.5125, "step": 2789 }, { "epoch": 0.1411300521017755, "grad_norm": 13.356398582458496, "learning_rate": 9.573411365757818e-06, "loss": 0.6099, "step": 2790 }, { "epoch": 0.14118063634984065, "grad_norm": 15.82447624206543, "learning_rate": 9.573086911310053e-06, "loss": 0.5478, "step": 2791 }, { "epoch": 0.1412312205979058, "grad_norm": 11.049214363098145, "learning_rate": 9.572762339024887e-06, "loss": 0.3627, "step": 2792 }, { "epoch": 0.14128180484597097, "grad_norm": 12.886861801147461, "learning_rate": 9.572437648910677e-06, "loss": 0.5433, "step": 2793 }, { "epoch": 0.14133238909403611, "grad_norm": 6.999958515167236, "learning_rate": 9.572112840975795e-06, "loss": 0.6013, "step": 2794 }, { "epoch": 0.14138297334210126, "grad_norm": 6.481417655944824, "learning_rate": 9.571787915228608e-06, "loss": 0.4544, "step": 2795 }, { "epoch": 0.14143355759016643, "grad_norm": 6.62939977645874, "learning_rate": 9.571462871677489e-06, "loss": 0.4683, "step": 2796 }, { "epoch": 0.14148414183823158, "grad_norm": 7.287919521331787, "learning_rate": 9.571137710330815e-06, "loss": 0.4902, "step": 2797 }, { "epoch": 0.14153472608629672, "grad_norm": 8.62877368927002, "learning_rate": 9.570812431196961e-06, "loss": 0.4476, "step": 2798 }, { "epoch": 0.1415853103343619, "grad_norm": 7.6471333503723145, "learning_rate": 9.570487034284313e-06, "loss": 0.5143, "step": 2799 }, { "epoch": 0.14163589458242704, "grad_norm": 5.571504592895508, "learning_rate": 9.570161519601254e-06, "loss": 0.4641, "step": 2800 }, { "epoch": 0.14168647883049218, "grad_norm": 6.345476150512695, "learning_rate": 9.56983588715617e-06, "loss": 0.3593, "step": 2801 }, { "epoch": 0.14173706307855732, "grad_norm": 9.697271347045898, "learning_rate": 9.569510136957455e-06, "loss": 0.3669, "step": 2802 }, { "epoch": 0.1417876473266225, "grad_norm": 6.292819976806641, "learning_rate": 9.569184269013498e-06, "loss": 0.4376, "step": 2803 }, { "epoch": 0.14183823157468764, "grad_norm": 10.344324111938477, "learning_rate": 9.5688582833327e-06, "loss": 0.526, "step": 2804 }, { "epoch": 0.14188881582275278, "grad_norm": 6.2691121101379395, "learning_rate": 9.568532179923461e-06, "loss": 0.4465, "step": 2805 }, { "epoch": 0.14193940007081796, "grad_norm": 5.4055562019348145, "learning_rate": 9.56820595879418e-06, "loss": 0.4566, "step": 2806 }, { "epoch": 0.1419899843188831, "grad_norm": 6.6113762855529785, "learning_rate": 9.567879619953268e-06, "loss": 0.4041, "step": 2807 }, { "epoch": 0.14204056856694824, "grad_norm": 11.078485488891602, "learning_rate": 9.567553163409131e-06, "loss": 0.4597, "step": 2808 }, { "epoch": 0.14209115281501342, "grad_norm": 11.310093879699707, "learning_rate": 9.567226589170182e-06, "loss": 0.552, "step": 2809 }, { "epoch": 0.14214173706307856, "grad_norm": 4.78338623046875, "learning_rate": 9.566899897244834e-06, "loss": 0.3593, "step": 2810 }, { "epoch": 0.1421923213111437, "grad_norm": 8.052964210510254, "learning_rate": 9.566573087641507e-06, "loss": 0.5268, "step": 2811 }, { "epoch": 0.14224290555920885, "grad_norm": 4.977423191070557, "learning_rate": 9.566246160368622e-06, "loss": 0.3686, "step": 2812 }, { "epoch": 0.14229348980727402, "grad_norm": 8.436458587646484, "learning_rate": 9.565919115434603e-06, "loss": 0.4937, "step": 2813 }, { "epoch": 0.14234407405533916, "grad_norm": 7.7243499755859375, "learning_rate": 9.565591952847877e-06, "loss": 0.4708, "step": 2814 }, { "epoch": 0.1423946583034043, "grad_norm": 7.4344353675842285, "learning_rate": 9.565264672616872e-06, "loss": 0.4489, "step": 2815 }, { "epoch": 0.14244524255146948, "grad_norm": 8.678932189941406, "learning_rate": 9.564937274750026e-06, "loss": 0.5043, "step": 2816 }, { "epoch": 0.14249582679953463, "grad_norm": 10.45868968963623, "learning_rate": 9.564609759255771e-06, "loss": 0.447, "step": 2817 }, { "epoch": 0.14254641104759977, "grad_norm": 11.351037979125977, "learning_rate": 9.564282126142547e-06, "loss": 0.4016, "step": 2818 }, { "epoch": 0.14259699529566494, "grad_norm": 3.9948654174804688, "learning_rate": 9.5639543754188e-06, "loss": 0.4195, "step": 2819 }, { "epoch": 0.14264757954373009, "grad_norm": 12.786715507507324, "learning_rate": 9.563626507092971e-06, "loss": 0.569, "step": 2820 }, { "epoch": 0.14269816379179523, "grad_norm": 7.452816486358643, "learning_rate": 9.56329852117351e-06, "loss": 0.494, "step": 2821 }, { "epoch": 0.14274874803986037, "grad_norm": 6.890173435211182, "learning_rate": 9.56297041766887e-06, "loss": 0.4856, "step": 2822 }, { "epoch": 0.14279933228792555, "grad_norm": 6.877353668212891, "learning_rate": 9.5626421965875e-06, "loss": 0.4405, "step": 2823 }, { "epoch": 0.1428499165359907, "grad_norm": 6.607927322387695, "learning_rate": 9.562313857937863e-06, "loss": 0.5127, "step": 2824 }, { "epoch": 0.14290050078405583, "grad_norm": 8.020859718322754, "learning_rate": 9.561985401728418e-06, "loss": 0.5005, "step": 2825 }, { "epoch": 0.142951085032121, "grad_norm": 17.778947830200195, "learning_rate": 9.561656827967628e-06, "loss": 0.5873, "step": 2826 }, { "epoch": 0.14300166928018615, "grad_norm": 27.310100555419922, "learning_rate": 9.56132813666396e-06, "loss": 0.711, "step": 2827 }, { "epoch": 0.1430522535282513, "grad_norm": 23.98604393005371, "learning_rate": 9.560999327825884e-06, "loss": 0.5222, "step": 2828 }, { "epoch": 0.14310283777631647, "grad_norm": 5.801028728485107, "learning_rate": 9.560670401461871e-06, "loss": 0.3908, "step": 2829 }, { "epoch": 0.1431534220243816, "grad_norm": 6.538027763366699, "learning_rate": 9.560341357580398e-06, "loss": 0.5141, "step": 2830 }, { "epoch": 0.14320400627244675, "grad_norm": 11.440335273742676, "learning_rate": 9.560012196189943e-06, "loss": 0.434, "step": 2831 }, { "epoch": 0.14325459052051193, "grad_norm": 6.4189958572387695, "learning_rate": 9.559682917298987e-06, "loss": 0.3994, "step": 2832 }, { "epoch": 0.14330517476857707, "grad_norm": 5.896032333374023, "learning_rate": 9.559353520916016e-06, "loss": 0.4709, "step": 2833 }, { "epoch": 0.14335575901664221, "grad_norm": 11.147425651550293, "learning_rate": 9.559024007049518e-06, "loss": 0.5451, "step": 2834 }, { "epoch": 0.14340634326470736, "grad_norm": 11.486295700073242, "learning_rate": 9.558694375707985e-06, "loss": 0.5755, "step": 2835 }, { "epoch": 0.14345692751277253, "grad_norm": 9.938319206237793, "learning_rate": 9.558364626899907e-06, "loss": 0.4508, "step": 2836 }, { "epoch": 0.14350751176083767, "grad_norm": 8.16849136352539, "learning_rate": 9.558034760633783e-06, "loss": 0.3851, "step": 2837 }, { "epoch": 0.14355809600890282, "grad_norm": 6.178287506103516, "learning_rate": 9.557704776918114e-06, "loss": 0.3777, "step": 2838 }, { "epoch": 0.143608680256968, "grad_norm": 7.692359924316406, "learning_rate": 9.5573746757614e-06, "loss": 0.4112, "step": 2839 }, { "epoch": 0.14365926450503314, "grad_norm": 10.204294204711914, "learning_rate": 9.55704445717215e-06, "loss": 0.4058, "step": 2840 }, { "epoch": 0.14370984875309828, "grad_norm": 18.570693969726562, "learning_rate": 9.55671412115887e-06, "loss": 0.4586, "step": 2841 }, { "epoch": 0.14376043300116345, "grad_norm": 14.298173904418945, "learning_rate": 9.556383667730074e-06, "loss": 0.4041, "step": 2842 }, { "epoch": 0.1438110172492286, "grad_norm": 6.440999507904053, "learning_rate": 9.556053096894275e-06, "loss": 0.5149, "step": 2843 }, { "epoch": 0.14386160149729374, "grad_norm": 7.699947834014893, "learning_rate": 9.555722408659993e-06, "loss": 0.5428, "step": 2844 }, { "epoch": 0.14391218574535888, "grad_norm": 14.19523811340332, "learning_rate": 9.55539160303575e-06, "loss": 0.6073, "step": 2845 }, { "epoch": 0.14396276999342406, "grad_norm": 7.584588527679443, "learning_rate": 9.555060680030068e-06, "loss": 0.3434, "step": 2846 }, { "epoch": 0.1440133542414892, "grad_norm": 10.4926176071167, "learning_rate": 9.554729639651473e-06, "loss": 0.5278, "step": 2847 }, { "epoch": 0.14406393848955434, "grad_norm": 12.955260276794434, "learning_rate": 9.554398481908498e-06, "loss": 0.4483, "step": 2848 }, { "epoch": 0.14411452273761952, "grad_norm": 17.155601501464844, "learning_rate": 9.554067206809674e-06, "loss": 0.5375, "step": 2849 }, { "epoch": 0.14416510698568466, "grad_norm": 10.34434700012207, "learning_rate": 9.553735814363539e-06, "loss": 0.4417, "step": 2850 }, { "epoch": 0.1442156912337498, "grad_norm": 7.721675872802734, "learning_rate": 9.553404304578631e-06, "loss": 0.4321, "step": 2851 }, { "epoch": 0.14426627548181498, "grad_norm": 8.529255867004395, "learning_rate": 9.553072677463492e-06, "loss": 0.3754, "step": 2852 }, { "epoch": 0.14431685972988012, "grad_norm": 12.756741523742676, "learning_rate": 9.552740933026668e-06, "loss": 0.5658, "step": 2853 }, { "epoch": 0.14436744397794526, "grad_norm": 7.494439125061035, "learning_rate": 9.552409071276706e-06, "loss": 0.4095, "step": 2854 }, { "epoch": 0.1444180282260104, "grad_norm": 14.776984214782715, "learning_rate": 9.552077092222159e-06, "loss": 0.594, "step": 2855 }, { "epoch": 0.14446861247407558, "grad_norm": 9.929322242736816, "learning_rate": 9.551744995871578e-06, "loss": 0.5048, "step": 2856 }, { "epoch": 0.14451919672214072, "grad_norm": 8.467772483825684, "learning_rate": 9.551412782233527e-06, "loss": 0.5231, "step": 2857 }, { "epoch": 0.14456978097020587, "grad_norm": 12.653153419494629, "learning_rate": 9.551080451316558e-06, "loss": 0.5489, "step": 2858 }, { "epoch": 0.14462036521827104, "grad_norm": 7.890932559967041, "learning_rate": 9.550748003129242e-06, "loss": 0.5086, "step": 2859 }, { "epoch": 0.14467094946633619, "grad_norm": 9.30056095123291, "learning_rate": 9.55041543768014e-06, "loss": 0.4457, "step": 2860 }, { "epoch": 0.14472153371440133, "grad_norm": 12.090890884399414, "learning_rate": 9.550082754977823e-06, "loss": 0.4399, "step": 2861 }, { "epoch": 0.1447721179624665, "grad_norm": 9.015406608581543, "learning_rate": 9.549749955030864e-06, "loss": 0.3677, "step": 2862 }, { "epoch": 0.14482270221053165, "grad_norm": 7.99516487121582, "learning_rate": 9.54941703784784e-06, "loss": 0.4116, "step": 2863 }, { "epoch": 0.1448732864585968, "grad_norm": 21.89977264404297, "learning_rate": 9.549084003437325e-06, "loss": 0.6377, "step": 2864 }, { "epoch": 0.14492387070666193, "grad_norm": 14.241257667541504, "learning_rate": 9.548750851807905e-06, "loss": 0.4942, "step": 2865 }, { "epoch": 0.1449744549547271, "grad_norm": 11.639986991882324, "learning_rate": 9.54841758296816e-06, "loss": 0.4879, "step": 2866 }, { "epoch": 0.14502503920279225, "grad_norm": 15.593644142150879, "learning_rate": 9.548084196926682e-06, "loss": 0.5525, "step": 2867 }, { "epoch": 0.1450756234508574, "grad_norm": 11.934558868408203, "learning_rate": 9.547750693692059e-06, "loss": 0.5762, "step": 2868 }, { "epoch": 0.14512620769892257, "grad_norm": 8.143073081970215, "learning_rate": 9.547417073272886e-06, "loss": 0.5585, "step": 2869 }, { "epoch": 0.1451767919469877, "grad_norm": 7.726799964904785, "learning_rate": 9.547083335677758e-06, "loss": 0.4647, "step": 2870 }, { "epoch": 0.14522737619505285, "grad_norm": 10.725544929504395, "learning_rate": 9.546749480915277e-06, "loss": 0.4837, "step": 2871 }, { "epoch": 0.14527796044311803, "grad_norm": 8.504953384399414, "learning_rate": 9.546415508994043e-06, "loss": 0.4967, "step": 2872 }, { "epoch": 0.14532854469118317, "grad_norm": 7.3630900382995605, "learning_rate": 9.546081419922663e-06, "loss": 0.5255, "step": 2873 }, { "epoch": 0.14537912893924831, "grad_norm": 6.613464832305908, "learning_rate": 9.545747213709746e-06, "loss": 0.4849, "step": 2874 }, { "epoch": 0.14542971318731346, "grad_norm": 10.57125186920166, "learning_rate": 9.545412890363902e-06, "loss": 0.5305, "step": 2875 }, { "epoch": 0.14548029743537863, "grad_norm": 11.218017578125, "learning_rate": 9.545078449893746e-06, "loss": 0.4711, "step": 2876 }, { "epoch": 0.14553088168344377, "grad_norm": 6.260952472686768, "learning_rate": 9.544743892307898e-06, "loss": 0.3787, "step": 2877 }, { "epoch": 0.14558146593150892, "grad_norm": 10.093235969543457, "learning_rate": 9.544409217614978e-06, "loss": 0.5687, "step": 2878 }, { "epoch": 0.1456320501795741, "grad_norm": 7.9731831550598145, "learning_rate": 9.544074425823608e-06, "loss": 0.4297, "step": 2879 }, { "epoch": 0.14568263442763923, "grad_norm": 8.9869384765625, "learning_rate": 9.543739516942415e-06, "loss": 0.4125, "step": 2880 }, { "epoch": 0.14573321867570438, "grad_norm": 6.285625457763672, "learning_rate": 9.54340449098003e-06, "loss": 0.3876, "step": 2881 }, { "epoch": 0.14578380292376955, "grad_norm": 11.772497177124023, "learning_rate": 9.543069347945087e-06, "loss": 0.6975, "step": 2882 }, { "epoch": 0.1458343871718347, "grad_norm": 5.973378658294678, "learning_rate": 9.54273408784622e-06, "loss": 0.4427, "step": 2883 }, { "epoch": 0.14588497141989984, "grad_norm": 6.731135368347168, "learning_rate": 9.542398710692066e-06, "loss": 0.4701, "step": 2884 }, { "epoch": 0.14593555566796498, "grad_norm": 8.684868812561035, "learning_rate": 9.54206321649127e-06, "loss": 0.3113, "step": 2885 }, { "epoch": 0.14598613991603016, "grad_norm": 11.585838317871094, "learning_rate": 9.541727605252476e-06, "loss": 0.4568, "step": 2886 }, { "epoch": 0.1460367241640953, "grad_norm": 10.883811950683594, "learning_rate": 9.541391876984332e-06, "loss": 0.5311, "step": 2887 }, { "epoch": 0.14608730841216044, "grad_norm": 10.60999870300293, "learning_rate": 9.541056031695489e-06, "loss": 0.4566, "step": 2888 }, { "epoch": 0.14613789266022562, "grad_norm": 9.800457000732422, "learning_rate": 9.540720069394599e-06, "loss": 0.5193, "step": 2889 }, { "epoch": 0.14618847690829076, "grad_norm": 10.275102615356445, "learning_rate": 9.54038399009032e-06, "loss": 0.4128, "step": 2890 }, { "epoch": 0.1462390611563559, "grad_norm": 9.560405731201172, "learning_rate": 9.540047793791314e-06, "loss": 0.4418, "step": 2891 }, { "epoch": 0.14628964540442108, "grad_norm": 12.696699142456055, "learning_rate": 9.539711480506242e-06, "loss": 0.5196, "step": 2892 }, { "epoch": 0.14634022965248622, "grad_norm": 8.844297409057617, "learning_rate": 9.53937505024377e-06, "loss": 0.4265, "step": 2893 }, { "epoch": 0.14639081390055136, "grad_norm": 7.498210430145264, "learning_rate": 9.53903850301257e-06, "loss": 0.385, "step": 2894 }, { "epoch": 0.1464413981486165, "grad_norm": 6.944670677185059, "learning_rate": 9.538701838821307e-06, "loss": 0.4989, "step": 2895 }, { "epoch": 0.14649198239668168, "grad_norm": 7.5561842918396, "learning_rate": 9.538365057678664e-06, "loss": 0.3771, "step": 2896 }, { "epoch": 0.14654256664474682, "grad_norm": 6.758105278015137, "learning_rate": 9.538028159593314e-06, "loss": 0.4677, "step": 2897 }, { "epoch": 0.14659315089281197, "grad_norm": 5.851391792297363, "learning_rate": 9.537691144573939e-06, "loss": 0.4119, "step": 2898 }, { "epoch": 0.14664373514087714, "grad_norm": 8.807991981506348, "learning_rate": 9.537354012629224e-06, "loss": 0.3856, "step": 2899 }, { "epoch": 0.14669431938894228, "grad_norm": 16.946260452270508, "learning_rate": 9.537016763767856e-06, "loss": 0.4952, "step": 2900 }, { "epoch": 0.14674490363700743, "grad_norm": 7.919156074523926, "learning_rate": 9.536679397998524e-06, "loss": 0.3412, "step": 2901 }, { "epoch": 0.1467954878850726, "grad_norm": 10.266600608825684, "learning_rate": 9.536341915329923e-06, "loss": 0.4142, "step": 2902 }, { "epoch": 0.14684607213313774, "grad_norm": 14.363058090209961, "learning_rate": 9.536004315770746e-06, "loss": 0.6169, "step": 2903 }, { "epoch": 0.1468966563812029, "grad_norm": 14.63154125213623, "learning_rate": 9.535666599329696e-06, "loss": 0.5074, "step": 2904 }, { "epoch": 0.14694724062926803, "grad_norm": 8.491539001464844, "learning_rate": 9.535328766015472e-06, "loss": 0.4939, "step": 2905 }, { "epoch": 0.1469978248773332, "grad_norm": 27.6695556640625, "learning_rate": 9.53499081583678e-06, "loss": 0.5702, "step": 2906 }, { "epoch": 0.14704840912539835, "grad_norm": 9.434350967407227, "learning_rate": 9.53465274880233e-06, "loss": 0.4284, "step": 2907 }, { "epoch": 0.1470989933734635, "grad_norm": 13.71735668182373, "learning_rate": 9.534314564920831e-06, "loss": 0.5128, "step": 2908 }, { "epoch": 0.14714957762152867, "grad_norm": 10.764137268066406, "learning_rate": 9.533976264200998e-06, "loss": 0.4647, "step": 2909 }, { "epoch": 0.1472001618695938, "grad_norm": 14.22591781616211, "learning_rate": 9.533637846651546e-06, "loss": 0.5709, "step": 2910 }, { "epoch": 0.14725074611765895, "grad_norm": 10.908642768859863, "learning_rate": 9.5332993122812e-06, "loss": 0.4176, "step": 2911 }, { "epoch": 0.14730133036572413, "grad_norm": 9.196746826171875, "learning_rate": 9.53296066109868e-06, "loss": 0.4994, "step": 2912 }, { "epoch": 0.14735191461378927, "grad_norm": 8.731649398803711, "learning_rate": 9.532621893112712e-06, "loss": 0.4884, "step": 2913 }, { "epoch": 0.14740249886185441, "grad_norm": 8.105534553527832, "learning_rate": 9.532283008332026e-06, "loss": 0.4362, "step": 2914 }, { "epoch": 0.14745308310991956, "grad_norm": 8.665393829345703, "learning_rate": 9.531944006765355e-06, "loss": 0.4217, "step": 2915 }, { "epoch": 0.14750366735798473, "grad_norm": 13.198077201843262, "learning_rate": 9.531604888421431e-06, "loss": 0.5241, "step": 2916 }, { "epoch": 0.14755425160604987, "grad_norm": 8.533960342407227, "learning_rate": 9.531265653308996e-06, "loss": 0.4437, "step": 2917 }, { "epoch": 0.14760483585411502, "grad_norm": 10.137703895568848, "learning_rate": 9.530926301436791e-06, "loss": 0.472, "step": 2918 }, { "epoch": 0.1476554201021802, "grad_norm": 12.577054023742676, "learning_rate": 9.53058683281356e-06, "loss": 0.4849, "step": 2919 }, { "epoch": 0.14770600435024533, "grad_norm": 25.31034278869629, "learning_rate": 9.530247247448047e-06, "loss": 0.6185, "step": 2920 }, { "epoch": 0.14775658859831048, "grad_norm": 5.90010404586792, "learning_rate": 9.529907545349006e-06, "loss": 0.5508, "step": 2921 }, { "epoch": 0.14780717284637565, "grad_norm": 7.211288928985596, "learning_rate": 9.529567726525188e-06, "loss": 0.4582, "step": 2922 }, { "epoch": 0.1478577570944408, "grad_norm": 10.357319831848145, "learning_rate": 9.529227790985352e-06, "loss": 0.4981, "step": 2923 }, { "epoch": 0.14790834134250594, "grad_norm": 6.497555732727051, "learning_rate": 9.528887738738256e-06, "loss": 0.5096, "step": 2924 }, { "epoch": 0.14795892559057108, "grad_norm": 11.32254695892334, "learning_rate": 9.528547569792659e-06, "loss": 0.4814, "step": 2925 }, { "epoch": 0.14800950983863626, "grad_norm": 12.642356872558594, "learning_rate": 9.528207284157332e-06, "loss": 0.5133, "step": 2926 }, { "epoch": 0.1480600940867014, "grad_norm": 18.66560173034668, "learning_rate": 9.52786688184104e-06, "loss": 0.5876, "step": 2927 }, { "epoch": 0.14811067833476654, "grad_norm": 13.080534934997559, "learning_rate": 9.527526362852554e-06, "loss": 0.5754, "step": 2928 }, { "epoch": 0.14816126258283172, "grad_norm": 7.286337375640869, "learning_rate": 9.527185727200652e-06, "loss": 0.4512, "step": 2929 }, { "epoch": 0.14821184683089686, "grad_norm": 21.09616470336914, "learning_rate": 9.526844974894108e-06, "loss": 0.4757, "step": 2930 }, { "epoch": 0.148262431078962, "grad_norm": 15.262909889221191, "learning_rate": 9.526504105941703e-06, "loss": 0.5736, "step": 2931 }, { "epoch": 0.14831301532702718, "grad_norm": 10.81488037109375, "learning_rate": 9.526163120352219e-06, "loss": 0.5424, "step": 2932 }, { "epoch": 0.14836359957509232, "grad_norm": 12.025740623474121, "learning_rate": 9.525822018134445e-06, "loss": 0.521, "step": 2933 }, { "epoch": 0.14841418382315746, "grad_norm": 7.204867362976074, "learning_rate": 9.525480799297168e-06, "loss": 0.4693, "step": 2934 }, { "epoch": 0.1484647680712226, "grad_norm": 7.71729850769043, "learning_rate": 9.525139463849184e-06, "loss": 0.4251, "step": 2935 }, { "epoch": 0.14851535231928778, "grad_norm": 7.777822971343994, "learning_rate": 9.524798011799286e-06, "loss": 0.4101, "step": 2936 }, { "epoch": 0.14856593656735292, "grad_norm": 10.367815017700195, "learning_rate": 9.52445644315627e-06, "loss": 0.4555, "step": 2937 }, { "epoch": 0.14861652081541807, "grad_norm": 11.174304008483887, "learning_rate": 9.524114757928941e-06, "loss": 0.6213, "step": 2938 }, { "epoch": 0.14866710506348324, "grad_norm": 8.548616409301758, "learning_rate": 9.523772956126102e-06, "loss": 0.3678, "step": 2939 }, { "epoch": 0.14871768931154838, "grad_norm": 6.6798906326293945, "learning_rate": 9.523431037756562e-06, "loss": 0.4932, "step": 2940 }, { "epoch": 0.14876827355961353, "grad_norm": 7.556405067443848, "learning_rate": 9.52308900282913e-06, "loss": 0.4336, "step": 2941 }, { "epoch": 0.1488188578076787, "grad_norm": 4.704728126525879, "learning_rate": 9.522746851352617e-06, "loss": 0.3897, "step": 2942 }, { "epoch": 0.14886944205574384, "grad_norm": 6.697819232940674, "learning_rate": 9.522404583335844e-06, "loss": 0.4647, "step": 2943 }, { "epoch": 0.148920026303809, "grad_norm": 7.942695140838623, "learning_rate": 9.522062198787629e-06, "loss": 0.5192, "step": 2944 }, { "epoch": 0.14897061055187413, "grad_norm": 7.457113265991211, "learning_rate": 9.521719697716791e-06, "loss": 0.3841, "step": 2945 }, { "epoch": 0.1490211947999393, "grad_norm": 14.921011924743652, "learning_rate": 9.52137708013216e-06, "loss": 0.6727, "step": 2946 }, { "epoch": 0.14907177904800445, "grad_norm": 7.45298433303833, "learning_rate": 9.521034346042563e-06, "loss": 0.4992, "step": 2947 }, { "epoch": 0.1491223632960696, "grad_norm": 10.005240440368652, "learning_rate": 9.52069149545683e-06, "loss": 0.511, "step": 2948 }, { "epoch": 0.14917294754413477, "grad_norm": 8.023069381713867, "learning_rate": 9.520348528383797e-06, "loss": 0.4268, "step": 2949 }, { "epoch": 0.1492235317921999, "grad_norm": 6.327906131744385, "learning_rate": 9.5200054448323e-06, "loss": 0.502, "step": 2950 }, { "epoch": 0.14927411604026505, "grad_norm": 10.120878219604492, "learning_rate": 9.519662244811181e-06, "loss": 0.4712, "step": 2951 }, { "epoch": 0.14932470028833023, "grad_norm": 15.919347763061523, "learning_rate": 9.519318928329284e-06, "loss": 0.5036, "step": 2952 }, { "epoch": 0.14937528453639537, "grad_norm": 7.797972202301025, "learning_rate": 9.518975495395452e-06, "loss": 0.4633, "step": 2953 }, { "epoch": 0.1494258687844605, "grad_norm": 14.459306716918945, "learning_rate": 9.518631946018538e-06, "loss": 0.4898, "step": 2954 }, { "epoch": 0.14947645303252566, "grad_norm": 10.296091079711914, "learning_rate": 9.518288280207392e-06, "loss": 0.5037, "step": 2955 }, { "epoch": 0.14952703728059083, "grad_norm": 10.373791694641113, "learning_rate": 9.51794449797087e-06, "loss": 0.3988, "step": 2956 }, { "epoch": 0.14957762152865597, "grad_norm": 15.195780754089355, "learning_rate": 9.517600599317833e-06, "loss": 0.4027, "step": 2957 }, { "epoch": 0.14962820577672112, "grad_norm": 6.0347185134887695, "learning_rate": 9.51725658425714e-06, "loss": 0.4083, "step": 2958 }, { "epoch": 0.1496787900247863, "grad_norm": 17.57097625732422, "learning_rate": 9.516912452797654e-06, "loss": 0.4915, "step": 2959 }, { "epoch": 0.14972937427285143, "grad_norm": 8.739011764526367, "learning_rate": 9.516568204948247e-06, "loss": 0.5247, "step": 2960 }, { "epoch": 0.14977995852091658, "grad_norm": 8.235038757324219, "learning_rate": 9.516223840717785e-06, "loss": 0.4428, "step": 2961 }, { "epoch": 0.14983054276898175, "grad_norm": 6.481448650360107, "learning_rate": 9.515879360115142e-06, "loss": 0.464, "step": 2962 }, { "epoch": 0.1498811270170469, "grad_norm": 8.526955604553223, "learning_rate": 9.515534763149196e-06, "loss": 0.3775, "step": 2963 }, { "epoch": 0.14993171126511204, "grad_norm": 7.996408462524414, "learning_rate": 9.515190049828828e-06, "loss": 0.4202, "step": 2964 }, { "epoch": 0.14998229551317718, "grad_norm": 13.881686210632324, "learning_rate": 9.514845220162917e-06, "loss": 0.6041, "step": 2965 }, { "epoch": 0.15003287976124235, "grad_norm": 6.587958335876465, "learning_rate": 9.51450027416035e-06, "loss": 0.4229, "step": 2966 }, { "epoch": 0.1500834640093075, "grad_norm": 17.15700912475586, "learning_rate": 9.514155211830016e-06, "loss": 0.5571, "step": 2967 }, { "epoch": 0.15013404825737264, "grad_norm": 42.578208923339844, "learning_rate": 9.513810033180804e-06, "loss": 0.6891, "step": 2968 }, { "epoch": 0.15018463250543782, "grad_norm": 6.613854885101318, "learning_rate": 9.513464738221612e-06, "loss": 0.4219, "step": 2969 }, { "epoch": 0.15023521675350296, "grad_norm": 6.840424537658691, "learning_rate": 9.513119326961334e-06, "loss": 0.4086, "step": 2970 }, { "epoch": 0.1502858010015681, "grad_norm": 14.082215309143066, "learning_rate": 9.512773799408871e-06, "loss": 0.467, "step": 2971 }, { "epoch": 0.15033638524963328, "grad_norm": 8.434679985046387, "learning_rate": 9.51242815557313e-06, "loss": 0.4541, "step": 2972 }, { "epoch": 0.15038696949769842, "grad_norm": 7.356993675231934, "learning_rate": 9.512082395463014e-06, "loss": 0.5149, "step": 2973 }, { "epoch": 0.15043755374576356, "grad_norm": 8.411009788513184, "learning_rate": 9.511736519087432e-06, "loss": 0.542, "step": 2974 }, { "epoch": 0.15048813799382874, "grad_norm": 9.073976516723633, "learning_rate": 9.511390526455298e-06, "loss": 0.4457, "step": 2975 }, { "epoch": 0.15053872224189388, "grad_norm": 6.170853614807129, "learning_rate": 9.511044417575528e-06, "loss": 0.3861, "step": 2976 }, { "epoch": 0.15058930648995902, "grad_norm": 7.435972213745117, "learning_rate": 9.510698192457039e-06, "loss": 0.5445, "step": 2977 }, { "epoch": 0.15063989073802417, "grad_norm": 10.380951881408691, "learning_rate": 9.510351851108752e-06, "loss": 0.4729, "step": 2978 }, { "epoch": 0.15069047498608934, "grad_norm": 7.568857669830322, "learning_rate": 9.510005393539592e-06, "loss": 0.4684, "step": 2979 }, { "epoch": 0.15074105923415448, "grad_norm": 6.133627891540527, "learning_rate": 9.509658819758488e-06, "loss": 0.5187, "step": 2980 }, { "epoch": 0.15079164348221963, "grad_norm": 10.179302215576172, "learning_rate": 9.509312129774367e-06, "loss": 0.4178, "step": 2981 }, { "epoch": 0.1508422277302848, "grad_norm": 8.830522537231445, "learning_rate": 9.508965323596165e-06, "loss": 0.344, "step": 2982 }, { "epoch": 0.15089281197834994, "grad_norm": 9.594337463378906, "learning_rate": 9.508618401232817e-06, "loss": 0.4849, "step": 2983 }, { "epoch": 0.1509433962264151, "grad_norm": 7.525082111358643, "learning_rate": 9.508271362693264e-06, "loss": 0.43, "step": 2984 }, { "epoch": 0.15099398047448026, "grad_norm": 10.565555572509766, "learning_rate": 9.507924207986448e-06, "loss": 0.4527, "step": 2985 }, { "epoch": 0.1510445647225454, "grad_norm": 5.917895317077637, "learning_rate": 9.507576937121311e-06, "loss": 0.4469, "step": 2986 }, { "epoch": 0.15109514897061055, "grad_norm": 9.642991065979004, "learning_rate": 9.507229550106805e-06, "loss": 0.4134, "step": 2987 }, { "epoch": 0.1511457332186757, "grad_norm": 9.3915433883667, "learning_rate": 9.506882046951881e-06, "loss": 0.5233, "step": 2988 }, { "epoch": 0.15119631746674086, "grad_norm": 5.3598952293396, "learning_rate": 9.506534427665493e-06, "loss": 0.4082, "step": 2989 }, { "epoch": 0.151246901714806, "grad_norm": 9.701088905334473, "learning_rate": 9.506186692256596e-06, "loss": 0.6025, "step": 2990 }, { "epoch": 0.15129748596287115, "grad_norm": 9.591188430786133, "learning_rate": 9.505838840734155e-06, "loss": 0.3872, "step": 2991 }, { "epoch": 0.15134807021093633, "grad_norm": 11.883481979370117, "learning_rate": 9.505490873107127e-06, "loss": 0.4795, "step": 2992 }, { "epoch": 0.15139865445900147, "grad_norm": 21.80036735534668, "learning_rate": 9.505142789384483e-06, "loss": 0.4844, "step": 2993 }, { "epoch": 0.1514492387070666, "grad_norm": 11.287397384643555, "learning_rate": 9.504794589575192e-06, "loss": 0.5421, "step": 2994 }, { "epoch": 0.15149982295513179, "grad_norm": 12.910318374633789, "learning_rate": 9.504446273688225e-06, "loss": 0.4668, "step": 2995 }, { "epoch": 0.15155040720319693, "grad_norm": 4.965362071990967, "learning_rate": 9.504097841732557e-06, "loss": 0.4458, "step": 2996 }, { "epoch": 0.15160099145126207, "grad_norm": 8.281818389892578, "learning_rate": 9.503749293717168e-06, "loss": 0.3936, "step": 2997 }, { "epoch": 0.15165157569932722, "grad_norm": 13.708946228027344, "learning_rate": 9.503400629651036e-06, "loss": 0.5938, "step": 2998 }, { "epoch": 0.1517021599473924, "grad_norm": 10.338566780090332, "learning_rate": 9.50305184954315e-06, "loss": 0.6159, "step": 2999 }, { "epoch": 0.15175274419545753, "grad_norm": 10.658479690551758, "learning_rate": 9.502702953402491e-06, "loss": 0.5471, "step": 3000 }, { "epoch": 0.15180332844352268, "grad_norm": 7.682033538818359, "learning_rate": 9.502353941238056e-06, "loss": 0.501, "step": 3001 }, { "epoch": 0.15185391269158785, "grad_norm": 8.760491371154785, "learning_rate": 9.502004813058834e-06, "loss": 0.4795, "step": 3002 }, { "epoch": 0.151904496939653, "grad_norm": 8.2421875, "learning_rate": 9.501655568873822e-06, "loss": 0.4981, "step": 3003 }, { "epoch": 0.15195508118771814, "grad_norm": 8.186349868774414, "learning_rate": 9.501306208692018e-06, "loss": 0.5567, "step": 3004 }, { "epoch": 0.1520056654357833, "grad_norm": 8.118148803710938, "learning_rate": 9.500956732522427e-06, "loss": 0.5035, "step": 3005 }, { "epoch": 0.15205624968384845, "grad_norm": 11.36609172821045, "learning_rate": 9.500607140374052e-06, "loss": 0.4604, "step": 3006 }, { "epoch": 0.1521068339319136, "grad_norm": 4.407533645629883, "learning_rate": 9.500257432255902e-06, "loss": 0.4305, "step": 3007 }, { "epoch": 0.15215741817997874, "grad_norm": 9.483789443969727, "learning_rate": 9.499907608176987e-06, "loss": 0.4452, "step": 3008 }, { "epoch": 0.15220800242804391, "grad_norm": 10.954751014709473, "learning_rate": 9.499557668146325e-06, "loss": 0.5442, "step": 3009 }, { "epoch": 0.15225858667610906, "grad_norm": 7.07267427444458, "learning_rate": 9.499207612172927e-06, "loss": 0.4429, "step": 3010 }, { "epoch": 0.1523091709241742, "grad_norm": 8.052515983581543, "learning_rate": 9.498857440265817e-06, "loss": 0.4919, "step": 3011 }, { "epoch": 0.15235975517223938, "grad_norm": 16.265810012817383, "learning_rate": 9.498507152434017e-06, "loss": 0.5812, "step": 3012 }, { "epoch": 0.15241033942030452, "grad_norm": 10.223368644714355, "learning_rate": 9.498156748686555e-06, "loss": 0.5597, "step": 3013 }, { "epoch": 0.15246092366836966, "grad_norm": 6.435176849365234, "learning_rate": 9.497806229032456e-06, "loss": 0.4166, "step": 3014 }, { "epoch": 0.15251150791643484, "grad_norm": 9.604792594909668, "learning_rate": 9.497455593480758e-06, "loss": 0.3963, "step": 3015 }, { "epoch": 0.15256209216449998, "grad_norm": 5.593103885650635, "learning_rate": 9.49710484204049e-06, "loss": 0.3525, "step": 3016 }, { "epoch": 0.15261267641256512, "grad_norm": 8.746834754943848, "learning_rate": 9.49675397472069e-06, "loss": 0.3996, "step": 3017 }, { "epoch": 0.15266326066063027, "grad_norm": 10.241218566894531, "learning_rate": 9.496402991530407e-06, "loss": 0.5459, "step": 3018 }, { "epoch": 0.15271384490869544, "grad_norm": 6.426721096038818, "learning_rate": 9.496051892478677e-06, "loss": 0.4679, "step": 3019 }, { "epoch": 0.15276442915676058, "grad_norm": 8.575931549072266, "learning_rate": 9.49570067757455e-06, "loss": 0.4059, "step": 3020 }, { "epoch": 0.15281501340482573, "grad_norm": 13.973137855529785, "learning_rate": 9.495349346827074e-06, "loss": 0.4678, "step": 3021 }, { "epoch": 0.1528655976528909, "grad_norm": 12.438680648803711, "learning_rate": 9.494997900245303e-06, "loss": 0.5233, "step": 3022 }, { "epoch": 0.15291618190095604, "grad_norm": 7.134104251861572, "learning_rate": 9.494646337838296e-06, "loss": 0.503, "step": 3023 }, { "epoch": 0.1529667661490212, "grad_norm": 12.061946868896484, "learning_rate": 9.494294659615109e-06, "loss": 0.6556, "step": 3024 }, { "epoch": 0.15301735039708636, "grad_norm": 10.17318344116211, "learning_rate": 9.493942865584802e-06, "loss": 0.5355, "step": 3025 }, { "epoch": 0.1530679346451515, "grad_norm": 7.562304496765137, "learning_rate": 9.493590955756444e-06, "loss": 0.5273, "step": 3026 }, { "epoch": 0.15311851889321665, "grad_norm": 14.775965690612793, "learning_rate": 9.493238930139098e-06, "loss": 0.4918, "step": 3027 }, { "epoch": 0.1531691031412818, "grad_norm": 11.965868949890137, "learning_rate": 9.492886788741841e-06, "loss": 0.4752, "step": 3028 }, { "epoch": 0.15321968738934696, "grad_norm": 11.878995895385742, "learning_rate": 9.492534531573743e-06, "loss": 0.4833, "step": 3029 }, { "epoch": 0.1532702716374121, "grad_norm": 10.443851470947266, "learning_rate": 9.492182158643878e-06, "loss": 0.429, "step": 3030 }, { "epoch": 0.15332085588547725, "grad_norm": 14.757951736450195, "learning_rate": 9.491829669961334e-06, "loss": 0.4878, "step": 3031 }, { "epoch": 0.15337144013354242, "grad_norm": 11.050363540649414, "learning_rate": 9.491477065535186e-06, "loss": 0.4776, "step": 3032 }, { "epoch": 0.15342202438160757, "grad_norm": 5.955247402191162, "learning_rate": 9.491124345374523e-06, "loss": 0.4999, "step": 3033 }, { "epoch": 0.1534726086296727, "grad_norm": 5.70510196685791, "learning_rate": 9.490771509488435e-06, "loss": 0.4026, "step": 3034 }, { "epoch": 0.15352319287773789, "grad_norm": 10.954742431640625, "learning_rate": 9.490418557886012e-06, "loss": 0.5452, "step": 3035 }, { "epoch": 0.15357377712580303, "grad_norm": 8.440250396728516, "learning_rate": 9.490065490576348e-06, "loss": 0.4665, "step": 3036 }, { "epoch": 0.15362436137386817, "grad_norm": 10.151714324951172, "learning_rate": 9.489712307568542e-06, "loss": 0.5085, "step": 3037 }, { "epoch": 0.15367494562193332, "grad_norm": 12.90852165222168, "learning_rate": 9.489359008871693e-06, "loss": 0.441, "step": 3038 }, { "epoch": 0.1537255298699985, "grad_norm": 8.731480598449707, "learning_rate": 9.489005594494908e-06, "loss": 0.468, "step": 3039 }, { "epoch": 0.15377611411806363, "grad_norm": 20.86188316345215, "learning_rate": 9.48865206444729e-06, "loss": 0.4975, "step": 3040 }, { "epoch": 0.15382669836612878, "grad_norm": 8.228894233703613, "learning_rate": 9.48829841873795e-06, "loss": 0.503, "step": 3041 }, { "epoch": 0.15387728261419395, "grad_norm": 10.187457084655762, "learning_rate": 9.487944657376002e-06, "loss": 0.5654, "step": 3042 }, { "epoch": 0.1539278668622591, "grad_norm": 7.763041019439697, "learning_rate": 9.487590780370561e-06, "loss": 0.4813, "step": 3043 }, { "epoch": 0.15397845111032424, "grad_norm": 10.220061302185059, "learning_rate": 9.487236787730744e-06, "loss": 0.4446, "step": 3044 }, { "epoch": 0.1540290353583894, "grad_norm": 7.888998508453369, "learning_rate": 9.486882679465673e-06, "loss": 0.4324, "step": 3045 }, { "epoch": 0.15407961960645455, "grad_norm": 10.657424926757812, "learning_rate": 9.486528455584473e-06, "loss": 0.431, "step": 3046 }, { "epoch": 0.1541302038545197, "grad_norm": 6.651118278503418, "learning_rate": 9.486174116096272e-06, "loss": 0.5005, "step": 3047 }, { "epoch": 0.15418078810258484, "grad_norm": 11.475905418395996, "learning_rate": 9.4858196610102e-06, "loss": 0.3713, "step": 3048 }, { "epoch": 0.15423137235065001, "grad_norm": 17.183349609375, "learning_rate": 9.485465090335391e-06, "loss": 0.5617, "step": 3049 }, { "epoch": 0.15428195659871516, "grad_norm": 11.702765464782715, "learning_rate": 9.48511040408098e-06, "loss": 0.4255, "step": 3050 }, { "epoch": 0.1543325408467803, "grad_norm": 13.807353973388672, "learning_rate": 9.484755602256105e-06, "loss": 0.5997, "step": 3051 }, { "epoch": 0.15438312509484547, "grad_norm": 10.718708038330078, "learning_rate": 9.484400684869914e-06, "loss": 0.504, "step": 3052 }, { "epoch": 0.15443370934291062, "grad_norm": 12.480674743652344, "learning_rate": 9.484045651931548e-06, "loss": 0.472, "step": 3053 }, { "epoch": 0.15448429359097576, "grad_norm": 14.35213565826416, "learning_rate": 9.483690503450156e-06, "loss": 0.5475, "step": 3054 }, { "epoch": 0.15453487783904093, "grad_norm": 7.2211785316467285, "learning_rate": 9.48333523943489e-06, "loss": 0.3736, "step": 3055 }, { "epoch": 0.15458546208710608, "grad_norm": 12.897369384765625, "learning_rate": 9.482979859894903e-06, "loss": 0.5323, "step": 3056 }, { "epoch": 0.15463604633517122, "grad_norm": 6.899827480316162, "learning_rate": 9.482624364839354e-06, "loss": 0.4414, "step": 3057 }, { "epoch": 0.15468663058323637, "grad_norm": 9.630447387695312, "learning_rate": 9.482268754277403e-06, "loss": 0.5744, "step": 3058 }, { "epoch": 0.15473721483130154, "grad_norm": 15.445781707763672, "learning_rate": 9.481913028218212e-06, "loss": 0.5558, "step": 3059 }, { "epoch": 0.15478779907936668, "grad_norm": 11.97347354888916, "learning_rate": 9.481557186670947e-06, "loss": 0.5684, "step": 3060 }, { "epoch": 0.15483838332743183, "grad_norm": 15.933579444885254, "learning_rate": 9.48120122964478e-06, "loss": 0.4345, "step": 3061 }, { "epoch": 0.154888967575497, "grad_norm": 6.522799015045166, "learning_rate": 9.48084515714888e-06, "loss": 0.4746, "step": 3062 }, { "epoch": 0.15493955182356214, "grad_norm": 6.273479461669922, "learning_rate": 9.480488969192422e-06, "loss": 0.3526, "step": 3063 }, { "epoch": 0.1549901360716273, "grad_norm": 5.778529644012451, "learning_rate": 9.480132665784588e-06, "loss": 0.3794, "step": 3064 }, { "epoch": 0.15504072031969246, "grad_norm": 10.904513359069824, "learning_rate": 9.479776246934556e-06, "loss": 0.557, "step": 3065 }, { "epoch": 0.1550913045677576, "grad_norm": 9.90593147277832, "learning_rate": 9.47941971265151e-06, "loss": 0.4575, "step": 3066 }, { "epoch": 0.15514188881582275, "grad_norm": 9.557957649230957, "learning_rate": 9.479063062944637e-06, "loss": 0.4758, "step": 3067 }, { "epoch": 0.1551924730638879, "grad_norm": 4.912463188171387, "learning_rate": 9.478706297823128e-06, "loss": 0.4892, "step": 3068 }, { "epoch": 0.15524305731195306, "grad_norm": 8.363768577575684, "learning_rate": 9.478349417296176e-06, "loss": 0.4155, "step": 3069 }, { "epoch": 0.1552936415600182, "grad_norm": 9.829324722290039, "learning_rate": 9.477992421372977e-06, "loss": 0.5143, "step": 3070 }, { "epoch": 0.15534422580808335, "grad_norm": 8.124412536621094, "learning_rate": 9.477635310062729e-06, "loss": 0.473, "step": 3071 }, { "epoch": 0.15539481005614852, "grad_norm": 16.821935653686523, "learning_rate": 9.477278083374632e-06, "loss": 0.6134, "step": 3072 }, { "epoch": 0.15544539430421367, "grad_norm": 13.184733390808105, "learning_rate": 9.476920741317896e-06, "loss": 0.5607, "step": 3073 }, { "epoch": 0.1554959785522788, "grad_norm": 9.59594440460205, "learning_rate": 9.476563283901726e-06, "loss": 0.3957, "step": 3074 }, { "epoch": 0.15554656280034398, "grad_norm": 7.79537296295166, "learning_rate": 9.476205711135333e-06, "loss": 0.5025, "step": 3075 }, { "epoch": 0.15559714704840913, "grad_norm": 11.484285354614258, "learning_rate": 9.475848023027929e-06, "loss": 0.5079, "step": 3076 }, { "epoch": 0.15564773129647427, "grad_norm": 11.187613487243652, "learning_rate": 9.475490219588735e-06, "loss": 0.4773, "step": 3077 }, { "epoch": 0.15569831554453942, "grad_norm": 7.7339959144592285, "learning_rate": 9.475132300826966e-06, "loss": 0.4033, "step": 3078 }, { "epoch": 0.1557488997926046, "grad_norm": 9.177081108093262, "learning_rate": 9.474774266751848e-06, "loss": 0.2972, "step": 3079 }, { "epoch": 0.15579948404066973, "grad_norm": 11.875391006469727, "learning_rate": 9.474416117372604e-06, "loss": 0.4419, "step": 3080 }, { "epoch": 0.15585006828873488, "grad_norm": 7.996640205383301, "learning_rate": 9.474057852698465e-06, "loss": 0.4968, "step": 3081 }, { "epoch": 0.15590065253680005, "grad_norm": 16.102153778076172, "learning_rate": 9.47369947273866e-06, "loss": 0.5938, "step": 3082 }, { "epoch": 0.1559512367848652, "grad_norm": 8.693050384521484, "learning_rate": 9.473340977502427e-06, "loss": 0.4676, "step": 3083 }, { "epoch": 0.15600182103293034, "grad_norm": 14.39088249206543, "learning_rate": 9.472982366999005e-06, "loss": 0.5441, "step": 3084 }, { "epoch": 0.1560524052809955, "grad_norm": 12.88427734375, "learning_rate": 9.472623641237625e-06, "loss": 0.5081, "step": 3085 }, { "epoch": 0.15610298952906065, "grad_norm": 6.489086151123047, "learning_rate": 9.472264800227542e-06, "loss": 0.3873, "step": 3086 }, { "epoch": 0.1561535737771258, "grad_norm": 7.249180793762207, "learning_rate": 9.471905843977996e-06, "loss": 0.3502, "step": 3087 }, { "epoch": 0.15620415802519094, "grad_norm": 5.775835990905762, "learning_rate": 9.471546772498237e-06, "loss": 0.473, "step": 3088 }, { "epoch": 0.15625474227325611, "grad_norm": 10.437461853027344, "learning_rate": 9.47118758579752e-06, "loss": 0.5132, "step": 3089 }, { "epoch": 0.15630532652132126, "grad_norm": 7.514313220977783, "learning_rate": 9.470828283885097e-06, "loss": 0.4728, "step": 3090 }, { "epoch": 0.1563559107693864, "grad_norm": 14.440935134887695, "learning_rate": 9.470468866770229e-06, "loss": 0.4109, "step": 3091 }, { "epoch": 0.15640649501745157, "grad_norm": 9.335968971252441, "learning_rate": 9.470109334462176e-06, "loss": 0.3673, "step": 3092 }, { "epoch": 0.15645707926551672, "grad_norm": 8.533196449279785, "learning_rate": 9.469749686970203e-06, "loss": 0.5374, "step": 3093 }, { "epoch": 0.15650766351358186, "grad_norm": 9.243367195129395, "learning_rate": 9.469389924303576e-06, "loss": 0.3845, "step": 3094 }, { "epoch": 0.15655824776164703, "grad_norm": 10.22287654876709, "learning_rate": 9.469030046471566e-06, "loss": 0.3851, "step": 3095 }, { "epoch": 0.15660883200971218, "grad_norm": 9.792787551879883, "learning_rate": 9.468670053483447e-06, "loss": 0.3698, "step": 3096 }, { "epoch": 0.15665941625777732, "grad_norm": 8.314349174499512, "learning_rate": 9.468309945348494e-06, "loss": 0.4531, "step": 3097 }, { "epoch": 0.15671000050584247, "grad_norm": 30.760080337524414, "learning_rate": 9.467949722075986e-06, "loss": 0.6646, "step": 3098 }, { "epoch": 0.15676058475390764, "grad_norm": 13.129281044006348, "learning_rate": 9.467589383675205e-06, "loss": 0.4793, "step": 3099 }, { "epoch": 0.15681116900197278, "grad_norm": 9.457557678222656, "learning_rate": 9.467228930155438e-06, "loss": 0.4294, "step": 3100 }, { "epoch": 0.15686175325003793, "grad_norm": 6.1102190017700195, "learning_rate": 9.466868361525971e-06, "loss": 0.4293, "step": 3101 }, { "epoch": 0.1569123374981031, "grad_norm": 14.933953285217285, "learning_rate": 9.466507677796095e-06, "loss": 0.6321, "step": 3102 }, { "epoch": 0.15696292174616824, "grad_norm": 14.775077819824219, "learning_rate": 9.466146878975105e-06, "loss": 0.5792, "step": 3103 }, { "epoch": 0.1570135059942334, "grad_norm": 6.892970085144043, "learning_rate": 9.465785965072298e-06, "loss": 0.4798, "step": 3104 }, { "epoch": 0.15706409024229856, "grad_norm": 11.602743148803711, "learning_rate": 9.465424936096972e-06, "loss": 0.4083, "step": 3105 }, { "epoch": 0.1571146744903637, "grad_norm": 20.84882164001465, "learning_rate": 9.465063792058433e-06, "loss": 0.5782, "step": 3106 }, { "epoch": 0.15716525873842885, "grad_norm": 15.18336296081543, "learning_rate": 9.464702532965983e-06, "loss": 0.4421, "step": 3107 }, { "epoch": 0.157215842986494, "grad_norm": 6.6060566902160645, "learning_rate": 9.464341158828936e-06, "loss": 0.4145, "step": 3108 }, { "epoch": 0.15726642723455916, "grad_norm": 10.337128639221191, "learning_rate": 9.463979669656598e-06, "loss": 0.5293, "step": 3109 }, { "epoch": 0.1573170114826243, "grad_norm": 6.821113586425781, "learning_rate": 9.463618065458287e-06, "loss": 0.403, "step": 3110 }, { "epoch": 0.15736759573068945, "grad_norm": 9.574531555175781, "learning_rate": 9.463256346243319e-06, "loss": 0.4846, "step": 3111 }, { "epoch": 0.15741817997875462, "grad_norm": 10.225210189819336, "learning_rate": 9.462894512021016e-06, "loss": 0.4601, "step": 3112 }, { "epoch": 0.15746876422681977, "grad_norm": 8.859102249145508, "learning_rate": 9.462532562800702e-06, "loss": 0.4791, "step": 3113 }, { "epoch": 0.1575193484748849, "grad_norm": 12.63991928100586, "learning_rate": 9.462170498591701e-06, "loss": 0.6808, "step": 3114 }, { "epoch": 0.15756993272295008, "grad_norm": 16.570842742919922, "learning_rate": 9.461808319403344e-06, "loss": 0.547, "step": 3115 }, { "epoch": 0.15762051697101523, "grad_norm": 10.526623725891113, "learning_rate": 9.461446025244966e-06, "loss": 0.4493, "step": 3116 }, { "epoch": 0.15767110121908037, "grad_norm": 19.263120651245117, "learning_rate": 9.4610836161259e-06, "loss": 0.4718, "step": 3117 }, { "epoch": 0.15772168546714552, "grad_norm": 7.91985559463501, "learning_rate": 9.460721092055483e-06, "loss": 0.3932, "step": 3118 }, { "epoch": 0.1577722697152107, "grad_norm": 14.333861351013184, "learning_rate": 9.460358453043056e-06, "loss": 0.5576, "step": 3119 }, { "epoch": 0.15782285396327583, "grad_norm": 5.925724506378174, "learning_rate": 9.459995699097968e-06, "loss": 0.4813, "step": 3120 }, { "epoch": 0.15787343821134098, "grad_norm": 9.144554138183594, "learning_rate": 9.459632830229564e-06, "loss": 0.4709, "step": 3121 }, { "epoch": 0.15792402245940615, "grad_norm": 11.809309959411621, "learning_rate": 9.459269846447191e-06, "loss": 0.4858, "step": 3122 }, { "epoch": 0.1579746067074713, "grad_norm": 12.681998252868652, "learning_rate": 9.458906747760208e-06, "loss": 0.5021, "step": 3123 }, { "epoch": 0.15802519095553644, "grad_norm": 10.183238983154297, "learning_rate": 9.458543534177966e-06, "loss": 0.5247, "step": 3124 }, { "epoch": 0.1580757752036016, "grad_norm": 58.90282440185547, "learning_rate": 9.458180205709828e-06, "loss": 0.738, "step": 3125 }, { "epoch": 0.15812635945166675, "grad_norm": 8.739496231079102, "learning_rate": 9.457816762365153e-06, "loss": 0.4005, "step": 3126 }, { "epoch": 0.1581769436997319, "grad_norm": 7.569066524505615, "learning_rate": 9.45745320415331e-06, "loss": 0.39, "step": 3127 }, { "epoch": 0.15822752794779707, "grad_norm": 8.557510375976562, "learning_rate": 9.45708953108366e-06, "loss": 0.549, "step": 3128 }, { "epoch": 0.15827811219586221, "grad_norm": 8.672228813171387, "learning_rate": 9.456725743165582e-06, "loss": 0.5322, "step": 3129 }, { "epoch": 0.15832869644392736, "grad_norm": 10.705666542053223, "learning_rate": 9.456361840408446e-06, "loss": 0.5194, "step": 3130 }, { "epoch": 0.1583792806919925, "grad_norm": 8.398672103881836, "learning_rate": 9.455997822821628e-06, "loss": 0.3052, "step": 3131 }, { "epoch": 0.15842986494005767, "grad_norm": 7.177128314971924, "learning_rate": 9.455633690414509e-06, "loss": 0.3664, "step": 3132 }, { "epoch": 0.15848044918812282, "grad_norm": 11.09376335144043, "learning_rate": 9.455269443196474e-06, "loss": 0.4666, "step": 3133 }, { "epoch": 0.15853103343618796, "grad_norm": 13.73413372039795, "learning_rate": 9.454905081176904e-06, "loss": 0.4818, "step": 3134 }, { "epoch": 0.15858161768425313, "grad_norm": 9.231024742126465, "learning_rate": 9.454540604365193e-06, "loss": 0.5585, "step": 3135 }, { "epoch": 0.15863220193231828, "grad_norm": 11.428778648376465, "learning_rate": 9.454176012770729e-06, "loss": 0.5121, "step": 3136 }, { "epoch": 0.15868278618038342, "grad_norm": 6.650980472564697, "learning_rate": 9.453811306402907e-06, "loss": 0.4419, "step": 3137 }, { "epoch": 0.1587333704284486, "grad_norm": 12.257040977478027, "learning_rate": 9.453446485271124e-06, "loss": 0.4572, "step": 3138 }, { "epoch": 0.15878395467651374, "grad_norm": 8.93397331237793, "learning_rate": 9.453081549384786e-06, "loss": 0.5144, "step": 3139 }, { "epoch": 0.15883453892457888, "grad_norm": 7.773750305175781, "learning_rate": 9.45271649875329e-06, "loss": 0.3197, "step": 3140 }, { "epoch": 0.15888512317264403, "grad_norm": 11.59897518157959, "learning_rate": 9.452351333386044e-06, "loss": 0.472, "step": 3141 }, { "epoch": 0.1589357074207092, "grad_norm": 8.623126029968262, "learning_rate": 9.45198605329246e-06, "loss": 0.4174, "step": 3142 }, { "epoch": 0.15898629166877434, "grad_norm": 12.087570190429688, "learning_rate": 9.451620658481949e-06, "loss": 0.4626, "step": 3143 }, { "epoch": 0.1590368759168395, "grad_norm": 14.085315704345703, "learning_rate": 9.451255148963926e-06, "loss": 0.4699, "step": 3144 }, { "epoch": 0.15908746016490466, "grad_norm": 8.069189071655273, "learning_rate": 9.450889524747808e-06, "loss": 0.37, "step": 3145 }, { "epoch": 0.1591380444129698, "grad_norm": 12.189523696899414, "learning_rate": 9.45052378584302e-06, "loss": 0.4575, "step": 3146 }, { "epoch": 0.15918862866103495, "grad_norm": 10.264270782470703, "learning_rate": 9.45015793225898e-06, "loss": 0.4737, "step": 3147 }, { "epoch": 0.15923921290910012, "grad_norm": 9.454313278198242, "learning_rate": 9.449791964005123e-06, "loss": 0.5307, "step": 3148 }, { "epoch": 0.15928979715716526, "grad_norm": 7.732309341430664, "learning_rate": 9.449425881090873e-06, "loss": 0.5134, "step": 3149 }, { "epoch": 0.1593403814052304, "grad_norm": 15.685099601745605, "learning_rate": 9.449059683525665e-06, "loss": 0.5114, "step": 3150 }, { "epoch": 0.15939096565329555, "grad_norm": 16.536632537841797, "learning_rate": 9.448693371318938e-06, "loss": 0.5306, "step": 3151 }, { "epoch": 0.15944154990136072, "grad_norm": 16.784854888916016, "learning_rate": 9.448326944480126e-06, "loss": 0.5768, "step": 3152 }, { "epoch": 0.15949213414942587, "grad_norm": 8.33241081237793, "learning_rate": 9.447960403018672e-06, "loss": 0.4672, "step": 3153 }, { "epoch": 0.159542718397491, "grad_norm": 11.57249927520752, "learning_rate": 9.447593746944024e-06, "loss": 0.5542, "step": 3154 }, { "epoch": 0.15959330264555618, "grad_norm": 10.819228172302246, "learning_rate": 9.447226976265627e-06, "loss": 0.5056, "step": 3155 }, { "epoch": 0.15964388689362133, "grad_norm": 9.281899452209473, "learning_rate": 9.446860090992933e-06, "loss": 0.5206, "step": 3156 }, { "epoch": 0.15969447114168647, "grad_norm": 8.475191116333008, "learning_rate": 9.446493091135395e-06, "loss": 0.4197, "step": 3157 }, { "epoch": 0.15974505538975164, "grad_norm": 10.528175354003906, "learning_rate": 9.44612597670247e-06, "loss": 0.4193, "step": 3158 }, { "epoch": 0.1597956396378168, "grad_norm": 6.81652307510376, "learning_rate": 9.445758747703617e-06, "loss": 0.4383, "step": 3159 }, { "epoch": 0.15984622388588193, "grad_norm": 12.109183311462402, "learning_rate": 9.445391404148301e-06, "loss": 0.4234, "step": 3160 }, { "epoch": 0.15989680813394708, "grad_norm": 6.676385402679443, "learning_rate": 9.445023946045986e-06, "loss": 0.4706, "step": 3161 }, { "epoch": 0.15994739238201225, "grad_norm": 9.786731719970703, "learning_rate": 9.44465637340614e-06, "loss": 0.425, "step": 3162 }, { "epoch": 0.1599979766300774, "grad_norm": 6.179423809051514, "learning_rate": 9.444288686238234e-06, "loss": 0.3971, "step": 3163 }, { "epoch": 0.16004856087814254, "grad_norm": 4.672389030456543, "learning_rate": 9.443920884551744e-06, "loss": 0.334, "step": 3164 }, { "epoch": 0.1600991451262077, "grad_norm": 9.286048889160156, "learning_rate": 9.443552968356147e-06, "loss": 0.4899, "step": 3165 }, { "epoch": 0.16014972937427285, "grad_norm": 6.695357322692871, "learning_rate": 9.443184937660923e-06, "loss": 0.4865, "step": 3166 }, { "epoch": 0.160200313622338, "grad_norm": 8.770208358764648, "learning_rate": 9.442816792475553e-06, "loss": 0.2699, "step": 3167 }, { "epoch": 0.16025089787040317, "grad_norm": 6.591130256652832, "learning_rate": 9.442448532809529e-06, "loss": 0.4112, "step": 3168 }, { "epoch": 0.1603014821184683, "grad_norm": 5.5093770027160645, "learning_rate": 9.442080158672335e-06, "loss": 0.4616, "step": 3169 }, { "epoch": 0.16035206636653346, "grad_norm": 10.474923133850098, "learning_rate": 9.441711670073463e-06, "loss": 0.4618, "step": 3170 }, { "epoch": 0.1604026506145986, "grad_norm": 13.335351943969727, "learning_rate": 9.441343067022412e-06, "loss": 0.4713, "step": 3171 }, { "epoch": 0.16045323486266377, "grad_norm": 12.897665977478027, "learning_rate": 9.440974349528678e-06, "loss": 0.4581, "step": 3172 }, { "epoch": 0.16050381911072892, "grad_norm": 6.0948286056518555, "learning_rate": 9.44060551760176e-06, "loss": 0.4197, "step": 3173 }, { "epoch": 0.16055440335879406, "grad_norm": 19.168516159057617, "learning_rate": 9.440236571251164e-06, "loss": 0.5269, "step": 3174 }, { "epoch": 0.16060498760685923, "grad_norm": 13.312847137451172, "learning_rate": 9.439867510486399e-06, "loss": 0.5011, "step": 3175 }, { "epoch": 0.16065557185492438, "grad_norm": 12.307700157165527, "learning_rate": 9.439498335316969e-06, "loss": 0.5417, "step": 3176 }, { "epoch": 0.16070615610298952, "grad_norm": 9.324883460998535, "learning_rate": 9.439129045752392e-06, "loss": 0.4353, "step": 3177 }, { "epoch": 0.1607567403510547, "grad_norm": 11.317509651184082, "learning_rate": 9.43875964180218e-06, "loss": 0.4225, "step": 3178 }, { "epoch": 0.16080732459911984, "grad_norm": 7.733705520629883, "learning_rate": 9.438390123475855e-06, "loss": 0.2361, "step": 3179 }, { "epoch": 0.16085790884718498, "grad_norm": 8.201820373535156, "learning_rate": 9.438020490782936e-06, "loss": 0.4973, "step": 3180 }, { "epoch": 0.16090849309525013, "grad_norm": 10.779780387878418, "learning_rate": 9.437650743732949e-06, "loss": 0.588, "step": 3181 }, { "epoch": 0.1609590773433153, "grad_norm": 9.147663116455078, "learning_rate": 9.437280882335423e-06, "loss": 0.4047, "step": 3182 }, { "epoch": 0.16100966159138044, "grad_norm": 9.04656982421875, "learning_rate": 9.436910906599884e-06, "loss": 0.4455, "step": 3183 }, { "epoch": 0.1610602458394456, "grad_norm": 13.989007949829102, "learning_rate": 9.436540816535868e-06, "loss": 0.422, "step": 3184 }, { "epoch": 0.16111083008751076, "grad_norm": 11.242412567138672, "learning_rate": 9.436170612152912e-06, "loss": 0.5603, "step": 3185 }, { "epoch": 0.1611614143355759, "grad_norm": 10.098798751831055, "learning_rate": 9.435800293460553e-06, "loss": 0.448, "step": 3186 }, { "epoch": 0.16121199858364105, "grad_norm": 9.558272361755371, "learning_rate": 9.435429860468337e-06, "loss": 0.4302, "step": 3187 }, { "epoch": 0.16126258283170622, "grad_norm": 6.9160661697387695, "learning_rate": 9.435059313185806e-06, "loss": 0.4687, "step": 3188 }, { "epoch": 0.16131316707977136, "grad_norm": 8.714235305786133, "learning_rate": 9.434688651622507e-06, "loss": 0.4971, "step": 3189 }, { "epoch": 0.1613637513278365, "grad_norm": 7.099761009216309, "learning_rate": 9.434317875787995e-06, "loss": 0.4542, "step": 3190 }, { "epoch": 0.16141433557590165, "grad_norm": 11.531915664672852, "learning_rate": 9.433946985691823e-06, "loss": 0.572, "step": 3191 }, { "epoch": 0.16146491982396682, "grad_norm": 6.888375759124756, "learning_rate": 9.433575981343545e-06, "loss": 0.356, "step": 3192 }, { "epoch": 0.16151550407203197, "grad_norm": 8.423609733581543, "learning_rate": 9.433204862752722e-06, "loss": 0.41, "step": 3193 }, { "epoch": 0.1615660883200971, "grad_norm": 8.11059284210205, "learning_rate": 9.43283362992892e-06, "loss": 0.4521, "step": 3194 }, { "epoch": 0.16161667256816228, "grad_norm": 6.478438377380371, "learning_rate": 9.432462282881702e-06, "loss": 0.409, "step": 3195 }, { "epoch": 0.16166725681622743, "grad_norm": 7.807971477508545, "learning_rate": 9.432090821620636e-06, "loss": 0.5303, "step": 3196 }, { "epoch": 0.16171784106429257, "grad_norm": 7.841345310211182, "learning_rate": 9.431719246155295e-06, "loss": 0.376, "step": 3197 }, { "epoch": 0.16176842531235774, "grad_norm": 7.490889072418213, "learning_rate": 9.431347556495253e-06, "loss": 0.382, "step": 3198 }, { "epoch": 0.1618190095604229, "grad_norm": 9.189288139343262, "learning_rate": 9.43097575265009e-06, "loss": 0.4246, "step": 3199 }, { "epoch": 0.16186959380848803, "grad_norm": 12.878530502319336, "learning_rate": 9.430603834629384e-06, "loss": 0.5096, "step": 3200 }, { "epoch": 0.16192017805655318, "grad_norm": 6.864965915679932, "learning_rate": 9.430231802442717e-06, "loss": 0.4143, "step": 3201 }, { "epoch": 0.16197076230461835, "grad_norm": 13.320576667785645, "learning_rate": 9.429859656099678e-06, "loss": 0.5036, "step": 3202 }, { "epoch": 0.1620213465526835, "grad_norm": 9.223467826843262, "learning_rate": 9.429487395609858e-06, "loss": 0.4465, "step": 3203 }, { "epoch": 0.16207193080074864, "grad_norm": 5.197959899902344, "learning_rate": 9.429115020982844e-06, "loss": 0.3736, "step": 3204 }, { "epoch": 0.1621225150488138, "grad_norm": 11.134688377380371, "learning_rate": 9.428742532228234e-06, "loss": 0.4887, "step": 3205 }, { "epoch": 0.16217309929687895, "grad_norm": 8.748082160949707, "learning_rate": 9.428369929355627e-06, "loss": 0.3547, "step": 3206 }, { "epoch": 0.1622236835449441, "grad_norm": 8.362249374389648, "learning_rate": 9.427997212374623e-06, "loss": 0.4825, "step": 3207 }, { "epoch": 0.16227426779300927, "grad_norm": 14.59416389465332, "learning_rate": 9.427624381294827e-06, "loss": 0.5078, "step": 3208 }, { "epoch": 0.1623248520410744, "grad_norm": 9.266891479492188, "learning_rate": 9.427251436125844e-06, "loss": 0.4384, "step": 3209 }, { "epoch": 0.16237543628913956, "grad_norm": 9.963676452636719, "learning_rate": 9.426878376877284e-06, "loss": 0.4509, "step": 3210 }, { "epoch": 0.1624260205372047, "grad_norm": 10.916851997375488, "learning_rate": 9.426505203558764e-06, "loss": 0.518, "step": 3211 }, { "epoch": 0.16247660478526987, "grad_norm": 14.090651512145996, "learning_rate": 9.426131916179895e-06, "loss": 0.4591, "step": 3212 }, { "epoch": 0.16252718903333502, "grad_norm": 8.102749824523926, "learning_rate": 9.425758514750298e-06, "loss": 0.4577, "step": 3213 }, { "epoch": 0.16257777328140016, "grad_norm": 10.06946849822998, "learning_rate": 9.425384999279592e-06, "loss": 0.4494, "step": 3214 }, { "epoch": 0.16262835752946533, "grad_norm": 6.112576007843018, "learning_rate": 9.425011369777406e-06, "loss": 0.3788, "step": 3215 }, { "epoch": 0.16267894177753048, "grad_norm": 13.296661376953125, "learning_rate": 9.424637626253363e-06, "loss": 0.5742, "step": 3216 }, { "epoch": 0.16272952602559562, "grad_norm": 9.608368873596191, "learning_rate": 9.424263768717099e-06, "loss": 0.4934, "step": 3217 }, { "epoch": 0.1627801102736608, "grad_norm": 8.226719856262207, "learning_rate": 9.423889797178241e-06, "loss": 0.514, "step": 3218 }, { "epoch": 0.16283069452172594, "grad_norm": 31.728511810302734, "learning_rate": 9.42351571164643e-06, "loss": 0.4568, "step": 3219 }, { "epoch": 0.16288127876979108, "grad_norm": 11.422670364379883, "learning_rate": 9.423141512131302e-06, "loss": 0.4826, "step": 3220 }, { "epoch": 0.16293186301785623, "grad_norm": 10.352518081665039, "learning_rate": 9.422767198642503e-06, "loss": 0.5289, "step": 3221 }, { "epoch": 0.1629824472659214, "grad_norm": 7.829034328460693, "learning_rate": 9.422392771189674e-06, "loss": 0.3793, "step": 3222 }, { "epoch": 0.16303303151398654, "grad_norm": 10.712295532226562, "learning_rate": 9.422018229782468e-06, "loss": 0.4555, "step": 3223 }, { "epoch": 0.1630836157620517, "grad_norm": 11.780617713928223, "learning_rate": 9.42164357443053e-06, "loss": 0.4566, "step": 3224 }, { "epoch": 0.16313420001011686, "grad_norm": 9.064240455627441, "learning_rate": 9.42126880514352e-06, "loss": 0.4235, "step": 3225 }, { "epoch": 0.163184784258182, "grad_norm": 18.960540771484375, "learning_rate": 9.42089392193109e-06, "loss": 0.4149, "step": 3226 }, { "epoch": 0.16323536850624715, "grad_norm": 10.591105461120605, "learning_rate": 9.420518924802901e-06, "loss": 0.5164, "step": 3227 }, { "epoch": 0.16328595275431232, "grad_norm": 16.112478256225586, "learning_rate": 9.420143813768619e-06, "loss": 0.5781, "step": 3228 }, { "epoch": 0.16333653700237746, "grad_norm": 7.383039474487305, "learning_rate": 9.419768588837906e-06, "loss": 0.4662, "step": 3229 }, { "epoch": 0.1633871212504426, "grad_norm": 10.289396286010742, "learning_rate": 9.419393250020434e-06, "loss": 0.4597, "step": 3230 }, { "epoch": 0.16343770549850775, "grad_norm": 7.111055374145508, "learning_rate": 9.41901779732587e-06, "loss": 0.5468, "step": 3231 }, { "epoch": 0.16348828974657292, "grad_norm": 10.43169116973877, "learning_rate": 9.418642230763892e-06, "loss": 0.4373, "step": 3232 }, { "epoch": 0.16353887399463807, "grad_norm": 13.837230682373047, "learning_rate": 9.418266550344177e-06, "loss": 0.4956, "step": 3233 }, { "epoch": 0.1635894582427032, "grad_norm": 12.416698455810547, "learning_rate": 9.417890756076404e-06, "loss": 0.5236, "step": 3234 }, { "epoch": 0.16364004249076838, "grad_norm": 6.544328212738037, "learning_rate": 9.417514847970258e-06, "loss": 0.2787, "step": 3235 }, { "epoch": 0.16369062673883353, "grad_norm": 7.8244805335998535, "learning_rate": 9.417138826035423e-06, "loss": 0.3282, "step": 3236 }, { "epoch": 0.16374121098689867, "grad_norm": 9.300235748291016, "learning_rate": 9.416762690281591e-06, "loss": 0.4192, "step": 3237 }, { "epoch": 0.16379179523496384, "grad_norm": 17.506229400634766, "learning_rate": 9.416386440718453e-06, "loss": 0.4042, "step": 3238 }, { "epoch": 0.163842379483029, "grad_norm": 6.385549545288086, "learning_rate": 9.416010077355702e-06, "loss": 0.4294, "step": 3239 }, { "epoch": 0.16389296373109413, "grad_norm": 10.923323631286621, "learning_rate": 9.41563360020304e-06, "loss": 0.5358, "step": 3240 }, { "epoch": 0.16394354797915928, "grad_norm": 8.245593070983887, "learning_rate": 9.415257009270164e-06, "loss": 0.3246, "step": 3241 }, { "epoch": 0.16399413222722445, "grad_norm": 7.082493782043457, "learning_rate": 9.41488030456678e-06, "loss": 0.4397, "step": 3242 }, { "epoch": 0.1640447164752896, "grad_norm": 10.825267791748047, "learning_rate": 9.414503486102594e-06, "loss": 0.5169, "step": 3243 }, { "epoch": 0.16409530072335474, "grad_norm": 10.068388938903809, "learning_rate": 9.414126553887317e-06, "loss": 0.4147, "step": 3244 }, { "epoch": 0.1641458849714199, "grad_norm": 19.002155303955078, "learning_rate": 9.413749507930659e-06, "loss": 0.5162, "step": 3245 }, { "epoch": 0.16419646921948505, "grad_norm": 8.001690864562988, "learning_rate": 9.413372348242338e-06, "loss": 0.4017, "step": 3246 }, { "epoch": 0.1642470534675502, "grad_norm": 16.20397186279297, "learning_rate": 9.412995074832072e-06, "loss": 0.5281, "step": 3247 }, { "epoch": 0.16429763771561537, "grad_norm": 12.628341674804688, "learning_rate": 9.412617687709582e-06, "loss": 0.4882, "step": 3248 }, { "epoch": 0.1643482219636805, "grad_norm": 7.505413055419922, "learning_rate": 9.412240186884591e-06, "loss": 0.4158, "step": 3249 }, { "epoch": 0.16439880621174566, "grad_norm": 7.987827301025391, "learning_rate": 9.41186257236683e-06, "loss": 0.5064, "step": 3250 }, { "epoch": 0.1644493904598108, "grad_norm": 10.2925443649292, "learning_rate": 9.411484844166026e-06, "loss": 0.3437, "step": 3251 }, { "epoch": 0.16449997470787597, "grad_norm": 8.138376235961914, "learning_rate": 9.411107002291913e-06, "loss": 0.3613, "step": 3252 }, { "epoch": 0.16455055895594112, "grad_norm": 8.292080879211426, "learning_rate": 9.410729046754225e-06, "loss": 0.4119, "step": 3253 }, { "epoch": 0.16460114320400626, "grad_norm": 7.614290714263916, "learning_rate": 9.410350977562705e-06, "loss": 0.4036, "step": 3254 }, { "epoch": 0.16465172745207143, "grad_norm": 8.21430778503418, "learning_rate": 9.409972794727094e-06, "loss": 0.487, "step": 3255 }, { "epoch": 0.16470231170013658, "grad_norm": 8.015901565551758, "learning_rate": 9.409594498257135e-06, "loss": 0.3941, "step": 3256 }, { "epoch": 0.16475289594820172, "grad_norm": 15.35873031616211, "learning_rate": 9.409216088162577e-06, "loss": 0.455, "step": 3257 }, { "epoch": 0.1648034801962669, "grad_norm": 21.99161720275879, "learning_rate": 9.408837564453171e-06, "loss": 0.6743, "step": 3258 }, { "epoch": 0.16485406444433204, "grad_norm": 15.512730598449707, "learning_rate": 9.408458927138669e-06, "loss": 0.4992, "step": 3259 }, { "epoch": 0.16490464869239718, "grad_norm": 15.271824836730957, "learning_rate": 9.40808017622883e-06, "loss": 0.4485, "step": 3260 }, { "epoch": 0.16495523294046233, "grad_norm": 24.881296157836914, "learning_rate": 9.407701311733411e-06, "loss": 0.4649, "step": 3261 }, { "epoch": 0.1650058171885275, "grad_norm": 11.868205070495605, "learning_rate": 9.407322333662174e-06, "loss": 0.6307, "step": 3262 }, { "epoch": 0.16505640143659264, "grad_norm": 7.3106770515441895, "learning_rate": 9.406943242024888e-06, "loss": 0.3232, "step": 3263 }, { "epoch": 0.1651069856846578, "grad_norm": 15.786205291748047, "learning_rate": 9.406564036831319e-06, "loss": 0.4708, "step": 3264 }, { "epoch": 0.16515756993272296, "grad_norm": 11.00301456451416, "learning_rate": 9.406184718091239e-06, "loss": 0.4217, "step": 3265 }, { "epoch": 0.1652081541807881, "grad_norm": 7.212739944458008, "learning_rate": 9.40580528581442e-06, "loss": 0.4483, "step": 3266 }, { "epoch": 0.16525873842885325, "grad_norm": 9.139092445373535, "learning_rate": 9.40542574001064e-06, "loss": 0.4761, "step": 3267 }, { "epoch": 0.16530932267691842, "grad_norm": 12.778363227844238, "learning_rate": 9.405046080689681e-06, "loss": 0.5108, "step": 3268 }, { "epoch": 0.16535990692498356, "grad_norm": 8.164222717285156, "learning_rate": 9.404666307861325e-06, "loss": 0.5868, "step": 3269 }, { "epoch": 0.1654104911730487, "grad_norm": 9.96465015411377, "learning_rate": 9.404286421535358e-06, "loss": 0.4659, "step": 3270 }, { "epoch": 0.16546107542111388, "grad_norm": 11.408652305603027, "learning_rate": 9.403906421721565e-06, "loss": 0.4291, "step": 3271 }, { "epoch": 0.16551165966917902, "grad_norm": 13.665359497070312, "learning_rate": 9.403526308429743e-06, "loss": 0.6034, "step": 3272 }, { "epoch": 0.16556224391724417, "grad_norm": 6.739560127258301, "learning_rate": 9.403146081669682e-06, "loss": 0.412, "step": 3273 }, { "epoch": 0.1656128281653093, "grad_norm": 9.966141700744629, "learning_rate": 9.402765741451185e-06, "loss": 0.4743, "step": 3274 }, { "epoch": 0.16566341241337448, "grad_norm": 8.971683502197266, "learning_rate": 9.402385287784049e-06, "loss": 0.4483, "step": 3275 }, { "epoch": 0.16571399666143963, "grad_norm": 10.096282005310059, "learning_rate": 9.402004720678075e-06, "loss": 0.5653, "step": 3276 }, { "epoch": 0.16576458090950477, "grad_norm": 10.35494327545166, "learning_rate": 9.401624040143074e-06, "loss": 0.4984, "step": 3277 }, { "epoch": 0.16581516515756994, "grad_norm": 8.644102096557617, "learning_rate": 9.401243246188852e-06, "loss": 0.41, "step": 3278 }, { "epoch": 0.1658657494056351, "grad_norm": 8.065709114074707, "learning_rate": 9.400862338825223e-06, "loss": 0.5128, "step": 3279 }, { "epoch": 0.16591633365370023, "grad_norm": 7.409126281738281, "learning_rate": 9.400481318062e-06, "loss": 0.4774, "step": 3280 }, { "epoch": 0.1659669179017654, "grad_norm": 6.3422932624816895, "learning_rate": 9.400100183909005e-06, "loss": 0.4438, "step": 3281 }, { "epoch": 0.16601750214983055, "grad_norm": 12.022601127624512, "learning_rate": 9.399718936376055e-06, "loss": 0.6183, "step": 3282 }, { "epoch": 0.1660680863978957, "grad_norm": 13.704512596130371, "learning_rate": 9.399337575472974e-06, "loss": 0.5378, "step": 3283 }, { "epoch": 0.16611867064596084, "grad_norm": 6.464690208435059, "learning_rate": 9.39895610120959e-06, "loss": 0.5083, "step": 3284 }, { "epoch": 0.166169254894026, "grad_norm": 7.990697383880615, "learning_rate": 9.398574513595735e-06, "loss": 0.4491, "step": 3285 }, { "epoch": 0.16621983914209115, "grad_norm": 7.8918328285217285, "learning_rate": 9.398192812641237e-06, "loss": 0.4404, "step": 3286 }, { "epoch": 0.1662704233901563, "grad_norm": 7.084381103515625, "learning_rate": 9.397810998355935e-06, "loss": 0.5084, "step": 3287 }, { "epoch": 0.16632100763822147, "grad_norm": 12.742818832397461, "learning_rate": 9.397429070749664e-06, "loss": 0.5568, "step": 3288 }, { "epoch": 0.1663715918862866, "grad_norm": 9.430807113647461, "learning_rate": 9.397047029832269e-06, "loss": 0.4073, "step": 3289 }, { "epoch": 0.16642217613435176, "grad_norm": 5.701009273529053, "learning_rate": 9.396664875613593e-06, "loss": 0.3397, "step": 3290 }, { "epoch": 0.16647276038241693, "grad_norm": 6.458398818969727, "learning_rate": 9.39628260810348e-06, "loss": 0.4316, "step": 3291 }, { "epoch": 0.16652334463048207, "grad_norm": 10.574760437011719, "learning_rate": 9.395900227311785e-06, "loss": 0.3767, "step": 3292 }, { "epoch": 0.16657392887854722, "grad_norm": 10.473546028137207, "learning_rate": 9.39551773324836e-06, "loss": 0.49, "step": 3293 }, { "epoch": 0.16662451312661236, "grad_norm": 6.894680023193359, "learning_rate": 9.395135125923059e-06, "loss": 0.4057, "step": 3294 }, { "epoch": 0.16667509737467753, "grad_norm": 10.188333511352539, "learning_rate": 9.394752405345742e-06, "loss": 0.3874, "step": 3295 }, { "epoch": 0.16672568162274268, "grad_norm": 9.397384643554688, "learning_rate": 9.39436957152627e-06, "loss": 0.5397, "step": 3296 }, { "epoch": 0.16677626587080782, "grad_norm": 11.02453899383545, "learning_rate": 9.393986624474509e-06, "loss": 0.4703, "step": 3297 }, { "epoch": 0.166826850118873, "grad_norm": 7.847651958465576, "learning_rate": 9.393603564200326e-06, "loss": 0.4942, "step": 3298 }, { "epoch": 0.16687743436693814, "grad_norm": 15.17343807220459, "learning_rate": 9.393220390713591e-06, "loss": 0.5002, "step": 3299 }, { "epoch": 0.16692801861500328, "grad_norm": 9.423730850219727, "learning_rate": 9.392837104024178e-06, "loss": 0.3751, "step": 3300 }, { "epoch": 0.16697860286306845, "grad_norm": 17.390893936157227, "learning_rate": 9.392453704141964e-06, "loss": 0.517, "step": 3301 }, { "epoch": 0.1670291871111336, "grad_norm": 12.445940017700195, "learning_rate": 9.392070191076825e-06, "loss": 0.4416, "step": 3302 }, { "epoch": 0.16707977135919874, "grad_norm": 11.935840606689453, "learning_rate": 9.391686564838648e-06, "loss": 0.5054, "step": 3303 }, { "epoch": 0.1671303556072639, "grad_norm": 8.173124313354492, "learning_rate": 9.391302825437317e-06, "loss": 0.3969, "step": 3304 }, { "epoch": 0.16718093985532906, "grad_norm": 12.556380271911621, "learning_rate": 9.390918972882717e-06, "loss": 0.4587, "step": 3305 }, { "epoch": 0.1672315241033942, "grad_norm": 10.558906555175781, "learning_rate": 9.39053500718474e-06, "loss": 0.474, "step": 3306 }, { "epoch": 0.16728210835145935, "grad_norm": 8.312544822692871, "learning_rate": 9.390150928353284e-06, "loss": 0.3593, "step": 3307 }, { "epoch": 0.16733269259952452, "grad_norm": 6.3918609619140625, "learning_rate": 9.38976673639824e-06, "loss": 0.4226, "step": 3308 }, { "epoch": 0.16738327684758966, "grad_norm": 27.85953140258789, "learning_rate": 9.38938243132951e-06, "loss": 0.4374, "step": 3309 }, { "epoch": 0.1674338610956548, "grad_norm": 24.575448989868164, "learning_rate": 9.388998013156997e-06, "loss": 0.6617, "step": 3310 }, { "epoch": 0.16748444534371998, "grad_norm": 14.824808120727539, "learning_rate": 9.388613481890609e-06, "loss": 0.4561, "step": 3311 }, { "epoch": 0.16753502959178512, "grad_norm": 4.1407318115234375, "learning_rate": 9.38822883754025e-06, "loss": 0.3763, "step": 3312 }, { "epoch": 0.16758561383985027, "grad_norm": 19.327735900878906, "learning_rate": 9.387844080115833e-06, "loss": 0.6378, "step": 3313 }, { "epoch": 0.1676361980879154, "grad_norm": 11.643011093139648, "learning_rate": 9.387459209627272e-06, "loss": 0.4806, "step": 3314 }, { "epoch": 0.16768678233598058, "grad_norm": 13.610259056091309, "learning_rate": 9.387074226084485e-06, "loss": 0.5608, "step": 3315 }, { "epoch": 0.16773736658404573, "grad_norm": 9.187433242797852, "learning_rate": 9.386689129497393e-06, "loss": 0.3369, "step": 3316 }, { "epoch": 0.16778795083211087, "grad_norm": 8.252169609069824, "learning_rate": 9.386303919875916e-06, "loss": 0.4219, "step": 3317 }, { "epoch": 0.16783853508017604, "grad_norm": 10.561477661132812, "learning_rate": 9.385918597229984e-06, "loss": 0.5605, "step": 3318 }, { "epoch": 0.1678891193282412, "grad_norm": 11.688583374023438, "learning_rate": 9.385533161569523e-06, "loss": 0.4861, "step": 3319 }, { "epoch": 0.16793970357630633, "grad_norm": 6.684213638305664, "learning_rate": 9.385147612904463e-06, "loss": 0.4509, "step": 3320 }, { "epoch": 0.1679902878243715, "grad_norm": 12.469842910766602, "learning_rate": 9.384761951244743e-06, "loss": 0.4206, "step": 3321 }, { "epoch": 0.16804087207243665, "grad_norm": 10.384591102600098, "learning_rate": 9.384376176600297e-06, "loss": 0.5321, "step": 3322 }, { "epoch": 0.1680914563205018, "grad_norm": 19.900331497192383, "learning_rate": 9.383990288981066e-06, "loss": 0.7197, "step": 3323 }, { "epoch": 0.16814204056856694, "grad_norm": 5.089353561401367, "learning_rate": 9.383604288396997e-06, "loss": 0.4369, "step": 3324 }, { "epoch": 0.1681926248166321, "grad_norm": 9.20161247253418, "learning_rate": 9.383218174858034e-06, "loss": 0.4484, "step": 3325 }, { "epoch": 0.16824320906469725, "grad_norm": 8.845062255859375, "learning_rate": 9.382831948374123e-06, "loss": 0.3239, "step": 3326 }, { "epoch": 0.1682937933127624, "grad_norm": 10.587998390197754, "learning_rate": 9.382445608955222e-06, "loss": 0.6687, "step": 3327 }, { "epoch": 0.16834437756082757, "grad_norm": 8.357654571533203, "learning_rate": 9.38205915661128e-06, "loss": 0.5442, "step": 3328 }, { "epoch": 0.1683949618088927, "grad_norm": 7.544602870941162, "learning_rate": 9.38167259135226e-06, "loss": 0.4715, "step": 3329 }, { "epoch": 0.16844554605695786, "grad_norm": 23.48465347290039, "learning_rate": 9.381285913188121e-06, "loss": 0.6174, "step": 3330 }, { "epoch": 0.16849613030502303, "grad_norm": 12.229659080505371, "learning_rate": 9.380899122128825e-06, "loss": 0.318, "step": 3331 }, { "epoch": 0.16854671455308817, "grad_norm": 8.424824714660645, "learning_rate": 9.380512218184341e-06, "loss": 0.444, "step": 3332 }, { "epoch": 0.16859729880115332, "grad_norm": 14.331108093261719, "learning_rate": 9.380125201364638e-06, "loss": 0.5419, "step": 3333 }, { "epoch": 0.16864788304921846, "grad_norm": 7.011729717254639, "learning_rate": 9.379738071679689e-06, "loss": 0.4436, "step": 3334 }, { "epoch": 0.16869846729728363, "grad_norm": 11.823339462280273, "learning_rate": 9.379350829139466e-06, "loss": 0.5575, "step": 3335 }, { "epoch": 0.16874905154534878, "grad_norm": 10.796591758728027, "learning_rate": 9.378963473753952e-06, "loss": 0.5522, "step": 3336 }, { "epoch": 0.16879963579341392, "grad_norm": 6.259024620056152, "learning_rate": 9.378576005533127e-06, "loss": 0.4463, "step": 3337 }, { "epoch": 0.1688502200414791, "grad_norm": 6.273628234863281, "learning_rate": 9.378188424486973e-06, "loss": 0.4237, "step": 3338 }, { "epoch": 0.16890080428954424, "grad_norm": 8.8661527633667, "learning_rate": 9.377800730625479e-06, "loss": 0.4355, "step": 3339 }, { "epoch": 0.16895138853760938, "grad_norm": 6.682534694671631, "learning_rate": 9.377412923958635e-06, "loss": 0.4665, "step": 3340 }, { "epoch": 0.16900197278567455, "grad_norm": 7.784976482391357, "learning_rate": 9.377025004496432e-06, "loss": 0.4651, "step": 3341 }, { "epoch": 0.1690525570337397, "grad_norm": 7.734186172485352, "learning_rate": 9.376636972248867e-06, "loss": 0.4808, "step": 3342 }, { "epoch": 0.16910314128180484, "grad_norm": 10.27120304107666, "learning_rate": 9.37624882722594e-06, "loss": 0.3371, "step": 3343 }, { "epoch": 0.16915372552986999, "grad_norm": 7.341376781463623, "learning_rate": 9.37586056943765e-06, "loss": 0.4603, "step": 3344 }, { "epoch": 0.16920430977793516, "grad_norm": 12.56796646118164, "learning_rate": 9.375472198894002e-06, "loss": 0.6087, "step": 3345 }, { "epoch": 0.1692548940260003, "grad_norm": 17.05315589904785, "learning_rate": 9.375083715605004e-06, "loss": 0.5057, "step": 3346 }, { "epoch": 0.16930547827406545, "grad_norm": 10.229101181030273, "learning_rate": 9.374695119580667e-06, "loss": 0.5056, "step": 3347 }, { "epoch": 0.16935606252213062, "grad_norm": 14.034632682800293, "learning_rate": 9.374306410831004e-06, "loss": 0.6084, "step": 3348 }, { "epoch": 0.16940664677019576, "grad_norm": 8.498635292053223, "learning_rate": 9.37391758936603e-06, "loss": 0.5618, "step": 3349 }, { "epoch": 0.1694572310182609, "grad_norm": 10.06885051727295, "learning_rate": 9.373528655195762e-06, "loss": 0.5088, "step": 3350 }, { "epoch": 0.16950781526632608, "grad_norm": 9.726898193359375, "learning_rate": 9.373139608330228e-06, "loss": 0.4465, "step": 3351 }, { "epoch": 0.16955839951439122, "grad_norm": 17.656339645385742, "learning_rate": 9.372750448779447e-06, "loss": 0.5447, "step": 3352 }, { "epoch": 0.16960898376245637, "grad_norm": 9.220627784729004, "learning_rate": 9.37236117655345e-06, "loss": 0.5181, "step": 3353 }, { "epoch": 0.1696595680105215, "grad_norm": 7.178214073181152, "learning_rate": 9.371971791662266e-06, "loss": 0.5034, "step": 3354 }, { "epoch": 0.16971015225858668, "grad_norm": 9.261283874511719, "learning_rate": 9.371582294115929e-06, "loss": 0.4365, "step": 3355 }, { "epoch": 0.16976073650665183, "grad_norm": 12.056245803833008, "learning_rate": 9.371192683924475e-06, "loss": 0.7161, "step": 3356 }, { "epoch": 0.16981132075471697, "grad_norm": 10.087730407714844, "learning_rate": 9.370802961097945e-06, "loss": 0.517, "step": 3357 }, { "epoch": 0.16986190500278214, "grad_norm": 8.73394775390625, "learning_rate": 9.370413125646379e-06, "loss": 0.4811, "step": 3358 }, { "epoch": 0.1699124892508473, "grad_norm": 9.408086776733398, "learning_rate": 9.370023177579823e-06, "loss": 0.5425, "step": 3359 }, { "epoch": 0.16996307349891243, "grad_norm": 14.910614967346191, "learning_rate": 9.369633116908325e-06, "loss": 0.4483, "step": 3360 }, { "epoch": 0.1700136577469776, "grad_norm": 11.197736740112305, "learning_rate": 9.369242943641937e-06, "loss": 0.5458, "step": 3361 }, { "epoch": 0.17006424199504275, "grad_norm": 9.365995407104492, "learning_rate": 9.368852657790711e-06, "loss": 0.5167, "step": 3362 }, { "epoch": 0.1701148262431079, "grad_norm": 8.32110595703125, "learning_rate": 9.368462259364705e-06, "loss": 0.5005, "step": 3363 }, { "epoch": 0.17016541049117304, "grad_norm": 10.212093353271484, "learning_rate": 9.368071748373977e-06, "loss": 0.5161, "step": 3364 }, { "epoch": 0.1702159947392382, "grad_norm": 7.53515100479126, "learning_rate": 9.367681124828592e-06, "loss": 0.4221, "step": 3365 }, { "epoch": 0.17026657898730335, "grad_norm": 8.658324241638184, "learning_rate": 9.367290388738614e-06, "loss": 0.4472, "step": 3366 }, { "epoch": 0.1703171632353685, "grad_norm": 15.779009819030762, "learning_rate": 9.36689954011411e-06, "loss": 0.4656, "step": 3367 }, { "epoch": 0.17036774748343367, "grad_norm": 9.677587509155273, "learning_rate": 9.366508578965154e-06, "loss": 0.5574, "step": 3368 }, { "epoch": 0.1704183317314988, "grad_norm": 7.55058479309082, "learning_rate": 9.36611750530182e-06, "loss": 0.3576, "step": 3369 }, { "epoch": 0.17046891597956396, "grad_norm": 17.348464965820312, "learning_rate": 9.365726319134181e-06, "loss": 0.5467, "step": 3370 }, { "epoch": 0.17051950022762913, "grad_norm": 6.524267673492432, "learning_rate": 9.365335020472321e-06, "loss": 0.4402, "step": 3371 }, { "epoch": 0.17057008447569427, "grad_norm": 10.475005149841309, "learning_rate": 9.364943609326322e-06, "loss": 0.5152, "step": 3372 }, { "epoch": 0.17062066872375942, "grad_norm": 8.400655746459961, "learning_rate": 9.364552085706269e-06, "loss": 0.4602, "step": 3373 }, { "epoch": 0.17067125297182456, "grad_norm": 8.039794921875, "learning_rate": 9.36416044962225e-06, "loss": 0.4134, "step": 3374 }, { "epoch": 0.17072183721988973, "grad_norm": 7.464919567108154, "learning_rate": 9.363768701084358e-06, "loss": 0.4487, "step": 3375 }, { "epoch": 0.17077242146795488, "grad_norm": 9.7378511428833, "learning_rate": 9.363376840102686e-06, "loss": 0.4781, "step": 3376 }, { "epoch": 0.17082300571602002, "grad_norm": 15.690027236938477, "learning_rate": 9.362984866687335e-06, "loss": 0.6901, "step": 3377 }, { "epoch": 0.1708735899640852, "grad_norm": 9.840265274047852, "learning_rate": 9.3625927808484e-06, "loss": 0.3978, "step": 3378 }, { "epoch": 0.17092417421215034, "grad_norm": 28.000438690185547, "learning_rate": 9.362200582595986e-06, "loss": 0.5905, "step": 3379 }, { "epoch": 0.17097475846021548, "grad_norm": 4.646181583404541, "learning_rate": 9.361808271940201e-06, "loss": 0.3345, "step": 3380 }, { "epoch": 0.17102534270828065, "grad_norm": 10.442221641540527, "learning_rate": 9.361415848891152e-06, "loss": 0.4733, "step": 3381 }, { "epoch": 0.1710759269563458, "grad_norm": 7.734528541564941, "learning_rate": 9.36102331345895e-06, "loss": 0.4456, "step": 3382 }, { "epoch": 0.17112651120441094, "grad_norm": 8.482087135314941, "learning_rate": 9.360630665653713e-06, "loss": 0.3736, "step": 3383 }, { "epoch": 0.17117709545247609, "grad_norm": 9.657513618469238, "learning_rate": 9.360237905485554e-06, "loss": 0.4367, "step": 3384 }, { "epoch": 0.17122767970054126, "grad_norm": 12.60423755645752, "learning_rate": 9.359845032964599e-06, "loss": 0.5326, "step": 3385 }, { "epoch": 0.1712782639486064, "grad_norm": 9.010221481323242, "learning_rate": 9.359452048100966e-06, "loss": 0.4189, "step": 3386 }, { "epoch": 0.17132884819667155, "grad_norm": 7.68747091293335, "learning_rate": 9.359058950904782e-06, "loss": 0.4341, "step": 3387 }, { "epoch": 0.17137943244473672, "grad_norm": 10.120718955993652, "learning_rate": 9.35866574138618e-06, "loss": 0.4209, "step": 3388 }, { "epoch": 0.17143001669280186, "grad_norm": 9.13103199005127, "learning_rate": 9.35827241955529e-06, "loss": 0.4584, "step": 3389 }, { "epoch": 0.171480600940867, "grad_norm": 11.900222778320312, "learning_rate": 9.357878985422246e-06, "loss": 0.4509, "step": 3390 }, { "epoch": 0.17153118518893218, "grad_norm": 8.684012413024902, "learning_rate": 9.357485438997187e-06, "loss": 0.4553, "step": 3391 }, { "epoch": 0.17158176943699732, "grad_norm": 11.724474906921387, "learning_rate": 9.357091780290254e-06, "loss": 0.3982, "step": 3392 }, { "epoch": 0.17163235368506247, "grad_norm": 14.696884155273438, "learning_rate": 9.356698009311587e-06, "loss": 0.4762, "step": 3393 }, { "epoch": 0.1716829379331276, "grad_norm": 17.324188232421875, "learning_rate": 9.356304126071338e-06, "loss": 0.4265, "step": 3394 }, { "epoch": 0.17173352218119278, "grad_norm": 10.648634910583496, "learning_rate": 9.355910130579653e-06, "loss": 0.4779, "step": 3395 }, { "epoch": 0.17178410642925793, "grad_norm": 12.592236518859863, "learning_rate": 9.355516022846685e-06, "loss": 0.581, "step": 3396 }, { "epoch": 0.17183469067732307, "grad_norm": 5.852007865905762, "learning_rate": 9.35512180288259e-06, "loss": 0.4281, "step": 3397 }, { "epoch": 0.17188527492538824, "grad_norm": 9.326430320739746, "learning_rate": 9.354727470697526e-06, "loss": 0.5459, "step": 3398 }, { "epoch": 0.1719358591734534, "grad_norm": 13.069726943969727, "learning_rate": 9.354333026301652e-06, "loss": 0.5017, "step": 3399 }, { "epoch": 0.17198644342151853, "grad_norm": 9.225650787353516, "learning_rate": 9.353938469705135e-06, "loss": 0.4279, "step": 3400 }, { "epoch": 0.1720370276695837, "grad_norm": 8.57927417755127, "learning_rate": 9.35354380091814e-06, "loss": 0.4684, "step": 3401 }, { "epoch": 0.17208761191764885, "grad_norm": 10.575455665588379, "learning_rate": 9.353149019950835e-06, "loss": 0.5265, "step": 3402 }, { "epoch": 0.172138196165714, "grad_norm": 11.77735710144043, "learning_rate": 9.352754126813394e-06, "loss": 0.4655, "step": 3403 }, { "epoch": 0.17218878041377914, "grad_norm": 10.079852104187012, "learning_rate": 9.352359121515995e-06, "loss": 0.454, "step": 3404 }, { "epoch": 0.1722393646618443, "grad_norm": 12.068502426147461, "learning_rate": 9.351964004068814e-06, "loss": 0.708, "step": 3405 }, { "epoch": 0.17228994890990945, "grad_norm": 9.468277931213379, "learning_rate": 9.351568774482032e-06, "loss": 0.4659, "step": 3406 }, { "epoch": 0.1723405331579746, "grad_norm": 20.25524139404297, "learning_rate": 9.351173432765834e-06, "loss": 0.5963, "step": 3407 }, { "epoch": 0.17239111740603977, "grad_norm": 9.345309257507324, "learning_rate": 9.350777978930405e-06, "loss": 0.4951, "step": 3408 }, { "epoch": 0.1724417016541049, "grad_norm": 6.488471031188965, "learning_rate": 9.350382412985937e-06, "loss": 0.4182, "step": 3409 }, { "epoch": 0.17249228590217006, "grad_norm": 7.776607036590576, "learning_rate": 9.349986734942623e-06, "loss": 0.4043, "step": 3410 }, { "epoch": 0.17254287015023523, "grad_norm": 13.112222671508789, "learning_rate": 9.349590944810657e-06, "loss": 0.5659, "step": 3411 }, { "epoch": 0.17259345439830037, "grad_norm": 8.319196701049805, "learning_rate": 9.349195042600238e-06, "loss": 0.4776, "step": 3412 }, { "epoch": 0.17264403864636552, "grad_norm": 8.2225923538208, "learning_rate": 9.348799028321568e-06, "loss": 0.4525, "step": 3413 }, { "epoch": 0.17269462289443066, "grad_norm": 8.653058052062988, "learning_rate": 9.348402901984852e-06, "loss": 0.4747, "step": 3414 }, { "epoch": 0.17274520714249583, "grad_norm": 8.9251127243042, "learning_rate": 9.348006663600296e-06, "loss": 0.4033, "step": 3415 }, { "epoch": 0.17279579139056098, "grad_norm": 7.71573543548584, "learning_rate": 9.347610313178109e-06, "loss": 0.453, "step": 3416 }, { "epoch": 0.17284637563862612, "grad_norm": 9.11149787902832, "learning_rate": 9.347213850728508e-06, "loss": 0.5525, "step": 3417 }, { "epoch": 0.1728969598866913, "grad_norm": 11.596963882446289, "learning_rate": 9.346817276261703e-06, "loss": 0.6438, "step": 3418 }, { "epoch": 0.17294754413475644, "grad_norm": 9.376319885253906, "learning_rate": 9.346420589787919e-06, "loss": 0.4416, "step": 3419 }, { "epoch": 0.17299812838282158, "grad_norm": 9.915688514709473, "learning_rate": 9.346023791317373e-06, "loss": 0.4445, "step": 3420 }, { "epoch": 0.17304871263088675, "grad_norm": 12.053570747375488, "learning_rate": 9.345626880860292e-06, "loss": 0.5955, "step": 3421 }, { "epoch": 0.1730992968789519, "grad_norm": 8.016953468322754, "learning_rate": 9.345229858426903e-06, "loss": 0.3494, "step": 3422 }, { "epoch": 0.17314988112701704, "grad_norm": 7.997037887573242, "learning_rate": 9.344832724027436e-06, "loss": 0.4304, "step": 3423 }, { "epoch": 0.1732004653750822, "grad_norm": 6.353649139404297, "learning_rate": 9.344435477672125e-06, "loss": 0.4482, "step": 3424 }, { "epoch": 0.17325104962314736, "grad_norm": 5.163069725036621, "learning_rate": 9.344038119371204e-06, "loss": 0.357, "step": 3425 }, { "epoch": 0.1733016338712125, "grad_norm": 7.354866027832031, "learning_rate": 9.343640649134913e-06, "loss": 0.3536, "step": 3426 }, { "epoch": 0.17335221811927765, "grad_norm": 9.343457221984863, "learning_rate": 9.343243066973496e-06, "loss": 0.4318, "step": 3427 }, { "epoch": 0.17340280236734282, "grad_norm": 10.402984619140625, "learning_rate": 9.342845372897193e-06, "loss": 0.4466, "step": 3428 }, { "epoch": 0.17345338661540796, "grad_norm": 16.261600494384766, "learning_rate": 9.342447566916256e-06, "loss": 0.801, "step": 3429 }, { "epoch": 0.1735039708634731, "grad_norm": 5.657705783843994, "learning_rate": 9.342049649040934e-06, "loss": 0.4396, "step": 3430 }, { "epoch": 0.17355455511153828, "grad_norm": 6.38917875289917, "learning_rate": 9.34165161928148e-06, "loss": 0.3843, "step": 3431 }, { "epoch": 0.17360513935960342, "grad_norm": 9.274833679199219, "learning_rate": 9.341253477648152e-06, "loss": 0.4003, "step": 3432 }, { "epoch": 0.17365572360766857, "grad_norm": 10.883769035339355, "learning_rate": 9.340855224151206e-06, "loss": 0.4969, "step": 3433 }, { "epoch": 0.17370630785573374, "grad_norm": 6.548292636871338, "learning_rate": 9.340456858800906e-06, "loss": 0.4064, "step": 3434 }, { "epoch": 0.17375689210379888, "grad_norm": 4.8474321365356445, "learning_rate": 9.340058381607517e-06, "loss": 0.4205, "step": 3435 }, { "epoch": 0.17380747635186403, "grad_norm": 10.602739334106445, "learning_rate": 9.339659792581306e-06, "loss": 0.5759, "step": 3436 }, { "epoch": 0.17385806059992917, "grad_norm": 9.714492797851562, "learning_rate": 9.339261091732546e-06, "loss": 0.487, "step": 3437 }, { "epoch": 0.17390864484799434, "grad_norm": 15.967663764953613, "learning_rate": 9.338862279071507e-06, "loss": 0.646, "step": 3438 }, { "epoch": 0.1739592290960595, "grad_norm": 5.862614154815674, "learning_rate": 9.338463354608467e-06, "loss": 0.4467, "step": 3439 }, { "epoch": 0.17400981334412463, "grad_norm": 6.697502136230469, "learning_rate": 9.338064318353707e-06, "loss": 0.4686, "step": 3440 }, { "epoch": 0.1740603975921898, "grad_norm": 10.720683097839355, "learning_rate": 9.337665170317507e-06, "loss": 0.615, "step": 3441 }, { "epoch": 0.17411098184025495, "grad_norm": 7.084919452667236, "learning_rate": 9.33726591051015e-06, "loss": 0.447, "step": 3442 }, { "epoch": 0.1741615660883201, "grad_norm": 9.771202087402344, "learning_rate": 9.336866538941932e-06, "loss": 0.361, "step": 3443 }, { "epoch": 0.17421215033638526, "grad_norm": 12.052083969116211, "learning_rate": 9.336467055623135e-06, "loss": 0.431, "step": 3444 }, { "epoch": 0.1742627345844504, "grad_norm": 9.475692749023438, "learning_rate": 9.336067460564058e-06, "loss": 0.3535, "step": 3445 }, { "epoch": 0.17431331883251555, "grad_norm": 12.682440757751465, "learning_rate": 9.335667753774995e-06, "loss": 0.572, "step": 3446 }, { "epoch": 0.1743639030805807, "grad_norm": 11.388386726379395, "learning_rate": 9.335267935266245e-06, "loss": 0.4151, "step": 3447 }, { "epoch": 0.17441448732864587, "grad_norm": 6.695384979248047, "learning_rate": 9.334868005048115e-06, "loss": 0.3458, "step": 3448 }, { "epoch": 0.174465071576711, "grad_norm": 10.414569854736328, "learning_rate": 9.334467963130905e-06, "loss": 0.4192, "step": 3449 }, { "epoch": 0.17451565582477616, "grad_norm": 8.562041282653809, "learning_rate": 9.334067809524923e-06, "loss": 0.4046, "step": 3450 }, { "epoch": 0.17456624007284133, "grad_norm": 11.1470365524292, "learning_rate": 9.333667544240486e-06, "loss": 0.4559, "step": 3451 }, { "epoch": 0.17461682432090647, "grad_norm": 8.096911430358887, "learning_rate": 9.3332671672879e-06, "loss": 0.5005, "step": 3452 }, { "epoch": 0.17466740856897162, "grad_norm": 8.310463905334473, "learning_rate": 9.332866678677489e-06, "loss": 0.5007, "step": 3453 }, { "epoch": 0.1747179928170368, "grad_norm": 14.159165382385254, "learning_rate": 9.332466078419568e-06, "loss": 0.4437, "step": 3454 }, { "epoch": 0.17476857706510193, "grad_norm": 5.418168067932129, "learning_rate": 9.33206536652446e-06, "loss": 0.3998, "step": 3455 }, { "epoch": 0.17481916131316708, "grad_norm": 8.403716087341309, "learning_rate": 9.331664543002493e-06, "loss": 0.4143, "step": 3456 }, { "epoch": 0.17486974556123222, "grad_norm": 12.687678337097168, "learning_rate": 9.331263607863991e-06, "loss": 0.5353, "step": 3457 }, { "epoch": 0.1749203298092974, "grad_norm": 16.67671775817871, "learning_rate": 9.33086256111929e-06, "loss": 0.6135, "step": 3458 }, { "epoch": 0.17497091405736254, "grad_norm": 11.075359344482422, "learning_rate": 9.330461402778722e-06, "loss": 0.377, "step": 3459 }, { "epoch": 0.17502149830542768, "grad_norm": 10.378146171569824, "learning_rate": 9.330060132852621e-06, "loss": 0.5277, "step": 3460 }, { "epoch": 0.17507208255349285, "grad_norm": 12.192529678344727, "learning_rate": 9.32965875135133e-06, "loss": 0.4289, "step": 3461 }, { "epoch": 0.175122666801558, "grad_norm": 10.036965370178223, "learning_rate": 9.329257258285192e-06, "loss": 0.4837, "step": 3462 }, { "epoch": 0.17517325104962314, "grad_norm": 16.98793601989746, "learning_rate": 9.32885565366455e-06, "loss": 0.6815, "step": 3463 }, { "epoch": 0.1752238352976883, "grad_norm": 7.2117509841918945, "learning_rate": 9.328453937499754e-06, "loss": 0.4025, "step": 3464 }, { "epoch": 0.17527441954575346, "grad_norm": 11.13096809387207, "learning_rate": 9.328052109801155e-06, "loss": 0.4668, "step": 3465 }, { "epoch": 0.1753250037938186, "grad_norm": 11.143497467041016, "learning_rate": 9.327650170579107e-06, "loss": 0.4233, "step": 3466 }, { "epoch": 0.17537558804188375, "grad_norm": 9.047602653503418, "learning_rate": 9.327248119843968e-06, "loss": 0.4239, "step": 3467 }, { "epoch": 0.17542617228994892, "grad_norm": 7.892157554626465, "learning_rate": 9.326845957606097e-06, "loss": 0.5055, "step": 3468 }, { "epoch": 0.17547675653801406, "grad_norm": 8.596273422241211, "learning_rate": 9.326443683875855e-06, "loss": 0.4128, "step": 3469 }, { "epoch": 0.1755273407860792, "grad_norm": 10.907880783081055, "learning_rate": 9.326041298663611e-06, "loss": 0.4315, "step": 3470 }, { "epoch": 0.17557792503414438, "grad_norm": 15.680948257446289, "learning_rate": 9.32563880197973e-06, "loss": 0.5995, "step": 3471 }, { "epoch": 0.17562850928220952, "grad_norm": 9.215152740478516, "learning_rate": 9.325236193834588e-06, "loss": 0.4485, "step": 3472 }, { "epoch": 0.17567909353027467, "grad_norm": 8.560029029846191, "learning_rate": 9.324833474238554e-06, "loss": 0.3572, "step": 3473 }, { "epoch": 0.17572967777833984, "grad_norm": 8.547810554504395, "learning_rate": 9.32443064320201e-06, "loss": 0.3687, "step": 3474 }, { "epoch": 0.17578026202640498, "grad_norm": 6.270383358001709, "learning_rate": 9.324027700735332e-06, "loss": 0.3299, "step": 3475 }, { "epoch": 0.17583084627447013, "grad_norm": 13.251982688903809, "learning_rate": 9.323624646848905e-06, "loss": 0.4028, "step": 3476 }, { "epoch": 0.17588143052253527, "grad_norm": 8.743889808654785, "learning_rate": 9.323221481553113e-06, "loss": 0.4994, "step": 3477 }, { "epoch": 0.17593201477060044, "grad_norm": 12.781493186950684, "learning_rate": 9.322818204858346e-06, "loss": 0.5817, "step": 3478 }, { "epoch": 0.1759825990186656, "grad_norm": 5.8871331214904785, "learning_rate": 9.322414816774996e-06, "loss": 0.4144, "step": 3479 }, { "epoch": 0.17603318326673073, "grad_norm": 5.332548141479492, "learning_rate": 9.322011317313457e-06, "loss": 0.4097, "step": 3480 }, { "epoch": 0.1760837675147959, "grad_norm": 9.150001525878906, "learning_rate": 9.321607706484123e-06, "loss": 0.4928, "step": 3481 }, { "epoch": 0.17613435176286105, "grad_norm": 11.512145042419434, "learning_rate": 9.3212039842974e-06, "loss": 0.5186, "step": 3482 }, { "epoch": 0.1761849360109262, "grad_norm": 10.656283378601074, "learning_rate": 9.320800150763686e-06, "loss": 0.5904, "step": 3483 }, { "epoch": 0.17623552025899136, "grad_norm": 8.273035049438477, "learning_rate": 9.32039620589339e-06, "loss": 0.2877, "step": 3484 }, { "epoch": 0.1762861045070565, "grad_norm": 9.130280494689941, "learning_rate": 9.319992149696917e-06, "loss": 0.3876, "step": 3485 }, { "epoch": 0.17633668875512165, "grad_norm": 29.408367156982422, "learning_rate": 9.319587982184683e-06, "loss": 0.7455, "step": 3486 }, { "epoch": 0.1763872730031868, "grad_norm": 9.779464721679688, "learning_rate": 9.3191837033671e-06, "loss": 0.4016, "step": 3487 }, { "epoch": 0.17643785725125197, "grad_norm": 6.652312755584717, "learning_rate": 9.318779313254585e-06, "loss": 0.4551, "step": 3488 }, { "epoch": 0.1764884414993171, "grad_norm": 10.521788597106934, "learning_rate": 9.318374811857557e-06, "loss": 0.4434, "step": 3489 }, { "epoch": 0.17653902574738226, "grad_norm": 7.627398490905762, "learning_rate": 9.317970199186444e-06, "loss": 0.3753, "step": 3490 }, { "epoch": 0.17658960999544743, "grad_norm": 6.230451583862305, "learning_rate": 9.317565475251667e-06, "loss": 0.4742, "step": 3491 }, { "epoch": 0.17664019424351257, "grad_norm": 8.981308937072754, "learning_rate": 9.317160640063655e-06, "loss": 0.4988, "step": 3492 }, { "epoch": 0.17669077849157772, "grad_norm": 12.587623596191406, "learning_rate": 9.316755693632843e-06, "loss": 0.4914, "step": 3493 }, { "epoch": 0.1767413627396429, "grad_norm": 10.70553970336914, "learning_rate": 9.316350635969661e-06, "loss": 0.4772, "step": 3494 }, { "epoch": 0.17679194698770803, "grad_norm": 12.889168739318848, "learning_rate": 9.315945467084553e-06, "loss": 0.4346, "step": 3495 }, { "epoch": 0.17684253123577318, "grad_norm": 8.390643119812012, "learning_rate": 9.315540186987953e-06, "loss": 0.4413, "step": 3496 }, { "epoch": 0.17689311548383832, "grad_norm": 9.131729125976562, "learning_rate": 9.315134795690306e-06, "loss": 0.4559, "step": 3497 }, { "epoch": 0.1769436997319035, "grad_norm": 7.330708980560303, "learning_rate": 9.314729293202056e-06, "loss": 0.42, "step": 3498 }, { "epoch": 0.17699428397996864, "grad_norm": 7.459205150604248, "learning_rate": 9.314323679533657e-06, "loss": 0.3979, "step": 3499 }, { "epoch": 0.17704486822803378, "grad_norm": 9.83029842376709, "learning_rate": 9.313917954695555e-06, "loss": 0.4279, "step": 3500 }, { "epoch": 0.17709545247609895, "grad_norm": 11.67037582397461, "learning_rate": 9.31351211869821e-06, "loss": 0.3903, "step": 3501 }, { "epoch": 0.1771460367241641, "grad_norm": 6.35844612121582, "learning_rate": 9.313106171552073e-06, "loss": 0.3983, "step": 3502 }, { "epoch": 0.17719662097222924, "grad_norm": 5.253466606140137, "learning_rate": 9.31270011326761e-06, "loss": 0.4018, "step": 3503 }, { "epoch": 0.1772472052202944, "grad_norm": 8.098241806030273, "learning_rate": 9.312293943855284e-06, "loss": 0.5172, "step": 3504 }, { "epoch": 0.17729778946835956, "grad_norm": 8.919795036315918, "learning_rate": 9.311887663325556e-06, "loss": 0.5179, "step": 3505 }, { "epoch": 0.1773483737164247, "grad_norm": 9.695789337158203, "learning_rate": 9.311481271688898e-06, "loss": 0.4408, "step": 3506 }, { "epoch": 0.17739895796448985, "grad_norm": 5.720642566680908, "learning_rate": 9.311074768955783e-06, "loss": 0.4806, "step": 3507 }, { "epoch": 0.17744954221255502, "grad_norm": 10.101889610290527, "learning_rate": 9.310668155136684e-06, "loss": 0.4518, "step": 3508 }, { "epoch": 0.17750012646062016, "grad_norm": 9.832707405090332, "learning_rate": 9.310261430242077e-06, "loss": 0.4778, "step": 3509 }, { "epoch": 0.1775507107086853, "grad_norm": 11.673789024353027, "learning_rate": 9.309854594282446e-06, "loss": 0.4682, "step": 3510 }, { "epoch": 0.17760129495675048, "grad_norm": 8.004232406616211, "learning_rate": 9.30944764726827e-06, "loss": 0.3702, "step": 3511 }, { "epoch": 0.17765187920481562, "grad_norm": 12.042089462280273, "learning_rate": 9.30904058921004e-06, "loss": 0.5793, "step": 3512 }, { "epoch": 0.17770246345288077, "grad_norm": 6.497157096862793, "learning_rate": 9.30863342011824e-06, "loss": 0.3998, "step": 3513 }, { "epoch": 0.17775304770094594, "grad_norm": 6.914909839630127, "learning_rate": 9.308226140003365e-06, "loss": 0.3621, "step": 3514 }, { "epoch": 0.17780363194901108, "grad_norm": 6.175652980804443, "learning_rate": 9.307818748875908e-06, "loss": 0.4509, "step": 3515 }, { "epoch": 0.17785421619707623, "grad_norm": 12.319896697998047, "learning_rate": 9.307411246746364e-06, "loss": 0.5314, "step": 3516 }, { "epoch": 0.17790480044514137, "grad_norm": 8.489130020141602, "learning_rate": 9.30700363362524e-06, "loss": 0.4585, "step": 3517 }, { "epoch": 0.17795538469320654, "grad_norm": 10.471256256103516, "learning_rate": 9.306595909523035e-06, "loss": 0.53, "step": 3518 }, { "epoch": 0.1780059689412717, "grad_norm": 7.6393609046936035, "learning_rate": 9.306188074450255e-06, "loss": 0.4439, "step": 3519 }, { "epoch": 0.17805655318933683, "grad_norm": 6.912095069885254, "learning_rate": 9.30578012841741e-06, "loss": 0.4411, "step": 3520 }, { "epoch": 0.178107137437402, "grad_norm": 14.878555297851562, "learning_rate": 9.30537207143501e-06, "loss": 0.454, "step": 3521 }, { "epoch": 0.17815772168546715, "grad_norm": 8.969094276428223, "learning_rate": 9.304963903513574e-06, "loss": 0.5928, "step": 3522 }, { "epoch": 0.1782083059335323, "grad_norm": 7.707851886749268, "learning_rate": 9.304555624663615e-06, "loss": 0.3824, "step": 3523 }, { "epoch": 0.17825889018159746, "grad_norm": 8.085144996643066, "learning_rate": 9.304147234895651e-06, "loss": 0.537, "step": 3524 }, { "epoch": 0.1783094744296626, "grad_norm": 7.0207905769348145, "learning_rate": 9.303738734220213e-06, "loss": 0.4902, "step": 3525 }, { "epoch": 0.17836005867772775, "grad_norm": 7.5803608894348145, "learning_rate": 9.303330122647823e-06, "loss": 0.4702, "step": 3526 }, { "epoch": 0.1784106429257929, "grad_norm": 8.381265640258789, "learning_rate": 9.302921400189008e-06, "loss": 0.312, "step": 3527 }, { "epoch": 0.17846122717385807, "grad_norm": 9.048574447631836, "learning_rate": 9.302512566854303e-06, "loss": 0.503, "step": 3528 }, { "epoch": 0.1785118114219232, "grad_norm": 9.10919189453125, "learning_rate": 9.302103622654242e-06, "loss": 0.434, "step": 3529 }, { "epoch": 0.17856239566998836, "grad_norm": 8.22009563446045, "learning_rate": 9.30169456759936e-06, "loss": 0.4152, "step": 3530 }, { "epoch": 0.17861297991805353, "grad_norm": 10.77917194366455, "learning_rate": 9.3012854017002e-06, "loss": 0.5761, "step": 3531 }, { "epoch": 0.17866356416611867, "grad_norm": 9.730171203613281, "learning_rate": 9.300876124967306e-06, "loss": 0.5068, "step": 3532 }, { "epoch": 0.17871414841418382, "grad_norm": 21.03130531311035, "learning_rate": 9.300466737411223e-06, "loss": 0.4883, "step": 3533 }, { "epoch": 0.178764732662249, "grad_norm": 14.981977462768555, "learning_rate": 9.300057239042496e-06, "loss": 0.641, "step": 3534 }, { "epoch": 0.17881531691031413, "grad_norm": 7.280197620391846, "learning_rate": 9.299647629871684e-06, "loss": 0.4887, "step": 3535 }, { "epoch": 0.17886590115837928, "grad_norm": 9.180305480957031, "learning_rate": 9.299237909909335e-06, "loss": 0.4911, "step": 3536 }, { "epoch": 0.17891648540644442, "grad_norm": 13.160411834716797, "learning_rate": 9.298828079166013e-06, "loss": 0.5124, "step": 3537 }, { "epoch": 0.1789670696545096, "grad_norm": 9.769425392150879, "learning_rate": 9.298418137652273e-06, "loss": 0.5118, "step": 3538 }, { "epoch": 0.17901765390257474, "grad_norm": 11.095818519592285, "learning_rate": 9.298008085378678e-06, "loss": 0.6094, "step": 3539 }, { "epoch": 0.17906823815063988, "grad_norm": 10.856010437011719, "learning_rate": 9.297597922355798e-06, "loss": 0.3597, "step": 3540 }, { "epoch": 0.17911882239870505, "grad_norm": 9.021493911743164, "learning_rate": 9.2971876485942e-06, "loss": 0.5584, "step": 3541 }, { "epoch": 0.1791694066467702, "grad_norm": 10.797184944152832, "learning_rate": 9.296777264104455e-06, "loss": 0.5995, "step": 3542 }, { "epoch": 0.17921999089483534, "grad_norm": 8.362074851989746, "learning_rate": 9.29636676889714e-06, "loss": 0.4534, "step": 3543 }, { "epoch": 0.1792705751429005, "grad_norm": 10.087600708007812, "learning_rate": 9.295956162982829e-06, "loss": 0.3956, "step": 3544 }, { "epoch": 0.17932115939096566, "grad_norm": 9.83051586151123, "learning_rate": 9.295545446372105e-06, "loss": 0.4825, "step": 3545 }, { "epoch": 0.1793717436390308, "grad_norm": 9.358521461486816, "learning_rate": 9.295134619075549e-06, "loss": 0.5885, "step": 3546 }, { "epoch": 0.17942232788709594, "grad_norm": 14.40074634552002, "learning_rate": 9.29472368110375e-06, "loss": 0.4108, "step": 3547 }, { "epoch": 0.17947291213516112, "grad_norm": 17.01844596862793, "learning_rate": 9.294312632467296e-06, "loss": 0.6001, "step": 3548 }, { "epoch": 0.17952349638322626, "grad_norm": 8.518318176269531, "learning_rate": 9.293901473176776e-06, "loss": 0.4689, "step": 3549 }, { "epoch": 0.1795740806312914, "grad_norm": 12.346857070922852, "learning_rate": 9.293490203242787e-06, "loss": 0.4363, "step": 3550 }, { "epoch": 0.17962466487935658, "grad_norm": 11.56093692779541, "learning_rate": 9.293078822675926e-06, "loss": 0.5305, "step": 3551 }, { "epoch": 0.17967524912742172, "grad_norm": 6.467161655426025, "learning_rate": 9.292667331486793e-06, "loss": 0.4035, "step": 3552 }, { "epoch": 0.17972583337548687, "grad_norm": 11.838562965393066, "learning_rate": 9.292255729685991e-06, "loss": 0.6005, "step": 3553 }, { "epoch": 0.17977641762355204, "grad_norm": 13.128499031066895, "learning_rate": 9.29184401728413e-06, "loss": 0.5559, "step": 3554 }, { "epoch": 0.17982700187161718, "grad_norm": 10.617609024047852, "learning_rate": 9.291432194291809e-06, "loss": 0.4752, "step": 3555 }, { "epoch": 0.17987758611968233, "grad_norm": 10.070242881774902, "learning_rate": 9.29102026071965e-06, "loss": 0.447, "step": 3556 }, { "epoch": 0.17992817036774747, "grad_norm": 6.68286657333374, "learning_rate": 9.290608216578263e-06, "loss": 0.4831, "step": 3557 }, { "epoch": 0.17997875461581264, "grad_norm": 5.943473815917969, "learning_rate": 9.290196061878264e-06, "loss": 0.5106, "step": 3558 }, { "epoch": 0.18002933886387779, "grad_norm": 7.69053316116333, "learning_rate": 9.289783796630278e-06, "loss": 0.4134, "step": 3559 }, { "epoch": 0.18007992311194293, "grad_norm": 11.59210205078125, "learning_rate": 9.289371420844923e-06, "loss": 0.5234, "step": 3560 }, { "epoch": 0.1801305073600081, "grad_norm": 8.533042907714844, "learning_rate": 9.288958934532828e-06, "loss": 0.5048, "step": 3561 }, { "epoch": 0.18018109160807325, "grad_norm": 7.865940093994141, "learning_rate": 9.28854633770462e-06, "loss": 0.3997, "step": 3562 }, { "epoch": 0.1802316758561384, "grad_norm": 6.24387264251709, "learning_rate": 9.288133630370932e-06, "loss": 0.3742, "step": 3563 }, { "epoch": 0.18028226010420356, "grad_norm": 6.028845310211182, "learning_rate": 9.287720812542398e-06, "loss": 0.4669, "step": 3564 }, { "epoch": 0.1803328443522687, "grad_norm": 8.905265808105469, "learning_rate": 9.287307884229655e-06, "loss": 0.6314, "step": 3565 }, { "epoch": 0.18038342860033385, "grad_norm": 5.94718074798584, "learning_rate": 9.286894845443342e-06, "loss": 0.4312, "step": 3566 }, { "epoch": 0.180434012848399, "grad_norm": 8.698209762573242, "learning_rate": 9.286481696194106e-06, "loss": 0.5016, "step": 3567 }, { "epoch": 0.18048459709646417, "grad_norm": 10.788451194763184, "learning_rate": 9.286068436492588e-06, "loss": 0.5793, "step": 3568 }, { "epoch": 0.1805351813445293, "grad_norm": 10.440962791442871, "learning_rate": 9.285655066349441e-06, "loss": 0.4121, "step": 3569 }, { "epoch": 0.18058576559259445, "grad_norm": 8.677688598632812, "learning_rate": 9.285241585775314e-06, "loss": 0.4073, "step": 3570 }, { "epoch": 0.18063634984065963, "grad_norm": 7.0053229331970215, "learning_rate": 9.28482799478086e-06, "loss": 0.3856, "step": 3571 }, { "epoch": 0.18068693408872477, "grad_norm": 9.7716703414917, "learning_rate": 9.28441429337674e-06, "loss": 0.4657, "step": 3572 }, { "epoch": 0.18073751833678992, "grad_norm": 11.192479133605957, "learning_rate": 9.284000481573612e-06, "loss": 0.4856, "step": 3573 }, { "epoch": 0.1807881025848551, "grad_norm": 7.519443511962891, "learning_rate": 9.283586559382138e-06, "loss": 0.4537, "step": 3574 }, { "epoch": 0.18083868683292023, "grad_norm": 6.882574081420898, "learning_rate": 9.283172526812986e-06, "loss": 0.3044, "step": 3575 }, { "epoch": 0.18088927108098538, "grad_norm": 8.812511444091797, "learning_rate": 9.282758383876825e-06, "loss": 0.4629, "step": 3576 }, { "epoch": 0.18093985532905055, "grad_norm": 7.526209354400635, "learning_rate": 9.282344130584321e-06, "loss": 0.4322, "step": 3577 }, { "epoch": 0.1809904395771157, "grad_norm": 18.358144760131836, "learning_rate": 9.281929766946156e-06, "loss": 0.4353, "step": 3578 }, { "epoch": 0.18104102382518084, "grad_norm": 6.663575649261475, "learning_rate": 9.281515292973003e-06, "loss": 0.3976, "step": 3579 }, { "epoch": 0.18109160807324598, "grad_norm": 11.315512657165527, "learning_rate": 9.281100708675541e-06, "loss": 0.5231, "step": 3580 }, { "epoch": 0.18114219232131115, "grad_norm": 9.027338981628418, "learning_rate": 9.280686014064454e-06, "loss": 0.5151, "step": 3581 }, { "epoch": 0.1811927765693763, "grad_norm": 13.40421199798584, "learning_rate": 9.280271209150431e-06, "loss": 0.4806, "step": 3582 }, { "epoch": 0.18124336081744144, "grad_norm": 8.825467109680176, "learning_rate": 9.279856293944155e-06, "loss": 0.4853, "step": 3583 }, { "epoch": 0.1812939450655066, "grad_norm": 11.330227851867676, "learning_rate": 9.27944126845632e-06, "loss": 0.4029, "step": 3584 }, { "epoch": 0.18134452931357176, "grad_norm": 7.9033989906311035, "learning_rate": 9.279026132697621e-06, "loss": 0.4252, "step": 3585 }, { "epoch": 0.1813951135616369, "grad_norm": 8.623307228088379, "learning_rate": 9.278610886678753e-06, "loss": 0.432, "step": 3586 }, { "epoch": 0.18144569780970207, "grad_norm": 7.588760852813721, "learning_rate": 9.278195530410419e-06, "loss": 0.3471, "step": 3587 }, { "epoch": 0.18149628205776722, "grad_norm": 10.801573753356934, "learning_rate": 9.27778006390332e-06, "loss": 0.439, "step": 3588 }, { "epoch": 0.18154686630583236, "grad_norm": 6.715818881988525, "learning_rate": 9.277364487168157e-06, "loss": 0.3603, "step": 3589 }, { "epoch": 0.1815974505538975, "grad_norm": 6.7060160636901855, "learning_rate": 9.276948800215646e-06, "loss": 0.3961, "step": 3590 }, { "epoch": 0.18164803480196268, "grad_norm": 24.02515983581543, "learning_rate": 9.276533003056495e-06, "loss": 0.614, "step": 3591 }, { "epoch": 0.18169861905002782, "grad_norm": 6.788844108581543, "learning_rate": 9.276117095701417e-06, "loss": 0.4762, "step": 3592 }, { "epoch": 0.18174920329809297, "grad_norm": 10.553295135498047, "learning_rate": 9.27570107816113e-06, "loss": 0.5265, "step": 3593 }, { "epoch": 0.18179978754615814, "grad_norm": 11.043865203857422, "learning_rate": 9.275284950446355e-06, "loss": 0.4268, "step": 3594 }, { "epoch": 0.18185037179422328, "grad_norm": 5.0002312660217285, "learning_rate": 9.274868712567813e-06, "loss": 0.3932, "step": 3595 }, { "epoch": 0.18190095604228843, "grad_norm": 14.572787284851074, "learning_rate": 9.274452364536229e-06, "loss": 0.4753, "step": 3596 }, { "epoch": 0.1819515402903536, "grad_norm": 6.200220108032227, "learning_rate": 9.274035906362332e-06, "loss": 0.526, "step": 3597 }, { "epoch": 0.18200212453841874, "grad_norm": 11.244318008422852, "learning_rate": 9.273619338056854e-06, "loss": 0.4856, "step": 3598 }, { "epoch": 0.18205270878648389, "grad_norm": 10.996413230895996, "learning_rate": 9.273202659630526e-06, "loss": 0.3926, "step": 3599 }, { "epoch": 0.18210329303454903, "grad_norm": 6.762655735015869, "learning_rate": 9.27278587109409e-06, "loss": 0.365, "step": 3600 }, { "epoch": 0.1821538772826142, "grad_norm": 8.758360862731934, "learning_rate": 9.272368972458281e-06, "loss": 0.4955, "step": 3601 }, { "epoch": 0.18220446153067935, "grad_norm": 8.40256118774414, "learning_rate": 9.271951963733845e-06, "loss": 0.4684, "step": 3602 }, { "epoch": 0.1822550457787445, "grad_norm": 6.125500679016113, "learning_rate": 9.271534844931522e-06, "loss": 0.392, "step": 3603 }, { "epoch": 0.18230563002680966, "grad_norm": 6.498897552490234, "learning_rate": 9.271117616062067e-06, "loss": 0.4086, "step": 3604 }, { "epoch": 0.1823562142748748, "grad_norm": 9.703608512878418, "learning_rate": 9.270700277136225e-06, "loss": 0.4763, "step": 3605 }, { "epoch": 0.18240679852293995, "grad_norm": 5.633080959320068, "learning_rate": 9.270282828164752e-06, "loss": 0.4562, "step": 3606 }, { "epoch": 0.18245738277100512, "grad_norm": 14.307056427001953, "learning_rate": 9.269865269158407e-06, "loss": 0.7238, "step": 3607 }, { "epoch": 0.18250796701907027, "grad_norm": 13.053706169128418, "learning_rate": 9.269447600127947e-06, "loss": 0.6864, "step": 3608 }, { "epoch": 0.1825585512671354, "grad_norm": 5.218372344970703, "learning_rate": 9.269029821084133e-06, "loss": 0.2638, "step": 3609 }, { "epoch": 0.18260913551520055, "grad_norm": 6.554940223693848, "learning_rate": 9.268611932037735e-06, "loss": 0.4487, "step": 3610 }, { "epoch": 0.18265971976326573, "grad_norm": 7.619945526123047, "learning_rate": 9.268193932999515e-06, "loss": 0.5196, "step": 3611 }, { "epoch": 0.18271030401133087, "grad_norm": 10.818193435668945, "learning_rate": 9.267775823980247e-06, "loss": 0.4136, "step": 3612 }, { "epoch": 0.18276088825939601, "grad_norm": 8.340424537658691, "learning_rate": 9.267357604990707e-06, "loss": 0.5368, "step": 3613 }, { "epoch": 0.1828114725074612, "grad_norm": 12.390009880065918, "learning_rate": 9.266939276041667e-06, "loss": 0.4319, "step": 3614 }, { "epoch": 0.18286205675552633, "grad_norm": 8.81187629699707, "learning_rate": 9.266520837143909e-06, "loss": 0.5131, "step": 3615 }, { "epoch": 0.18291264100359148, "grad_norm": 8.983011245727539, "learning_rate": 9.266102288308212e-06, "loss": 0.3523, "step": 3616 }, { "epoch": 0.18296322525165665, "grad_norm": 9.743983268737793, "learning_rate": 9.265683629545366e-06, "loss": 0.3288, "step": 3617 }, { "epoch": 0.1830138094997218, "grad_norm": 6.326700687408447, "learning_rate": 9.265264860866155e-06, "loss": 0.474, "step": 3618 }, { "epoch": 0.18306439374778694, "grad_norm": 8.087652206420898, "learning_rate": 9.264845982281372e-06, "loss": 0.3838, "step": 3619 }, { "epoch": 0.18311497799585208, "grad_norm": 12.875266075134277, "learning_rate": 9.264426993801809e-06, "loss": 0.5649, "step": 3620 }, { "epoch": 0.18316556224391725, "grad_norm": 24.43814468383789, "learning_rate": 9.264007895438262e-06, "loss": 0.5347, "step": 3621 }, { "epoch": 0.1832161464919824, "grad_norm": 7.481565952301025, "learning_rate": 9.26358868720153e-06, "loss": 0.4796, "step": 3622 }, { "epoch": 0.18326673074004754, "grad_norm": 6.850042819976807, "learning_rate": 9.263169369102418e-06, "loss": 0.4759, "step": 3623 }, { "epoch": 0.1833173149881127, "grad_norm": 8.712267875671387, "learning_rate": 9.262749941151727e-06, "loss": 0.3204, "step": 3624 }, { "epoch": 0.18336789923617786, "grad_norm": 14.737786293029785, "learning_rate": 9.262330403360267e-06, "loss": 0.4776, "step": 3625 }, { "epoch": 0.183418483484243, "grad_norm": 5.5009307861328125, "learning_rate": 9.261910755738848e-06, "loss": 0.247, "step": 3626 }, { "epoch": 0.18346906773230817, "grad_norm": 18.8294677734375, "learning_rate": 9.261490998298282e-06, "loss": 0.5116, "step": 3627 }, { "epoch": 0.18351965198037332, "grad_norm": 10.234001159667969, "learning_rate": 9.261071131049386e-06, "loss": 0.4236, "step": 3628 }, { "epoch": 0.18357023622843846, "grad_norm": 4.305499076843262, "learning_rate": 9.26065115400298e-06, "loss": 0.2778, "step": 3629 }, { "epoch": 0.1836208204765036, "grad_norm": 11.887226104736328, "learning_rate": 9.260231067169885e-06, "loss": 0.5304, "step": 3630 }, { "epoch": 0.18367140472456878, "grad_norm": 6.533990383148193, "learning_rate": 9.259810870560925e-06, "loss": 0.4833, "step": 3631 }, { "epoch": 0.18372198897263392, "grad_norm": 9.219510078430176, "learning_rate": 9.259390564186929e-06, "loss": 0.564, "step": 3632 }, { "epoch": 0.18377257322069906, "grad_norm": 7.078641891479492, "learning_rate": 9.258970148058725e-06, "loss": 0.4435, "step": 3633 }, { "epoch": 0.18382315746876424, "grad_norm": 6.246804237365723, "learning_rate": 9.25854962218715e-06, "loss": 0.4304, "step": 3634 }, { "epoch": 0.18387374171682938, "grad_norm": 11.717530250549316, "learning_rate": 9.258128986583035e-06, "loss": 0.5298, "step": 3635 }, { "epoch": 0.18392432596489453, "grad_norm": 12.539466857910156, "learning_rate": 9.257708241257221e-06, "loss": 0.4066, "step": 3636 }, { "epoch": 0.1839749102129597, "grad_norm": 16.266584396362305, "learning_rate": 9.25728738622055e-06, "loss": 0.4826, "step": 3637 }, { "epoch": 0.18402549446102484, "grad_norm": 11.39923095703125, "learning_rate": 9.256866421483867e-06, "loss": 0.5611, "step": 3638 }, { "epoch": 0.18407607870908999, "grad_norm": 7.0642595291137695, "learning_rate": 9.256445347058018e-06, "loss": 0.4731, "step": 3639 }, { "epoch": 0.18412666295715513, "grad_norm": 6.089073181152344, "learning_rate": 9.256024162953853e-06, "loss": 0.5097, "step": 3640 }, { "epoch": 0.1841772472052203, "grad_norm": 6.586595058441162, "learning_rate": 9.255602869182226e-06, "loss": 0.4296, "step": 3641 }, { "epoch": 0.18422783145328545, "grad_norm": 12.633833885192871, "learning_rate": 9.25518146575399e-06, "loss": 0.6079, "step": 3642 }, { "epoch": 0.1842784157013506, "grad_norm": 9.377156257629395, "learning_rate": 9.254759952680009e-06, "loss": 0.4632, "step": 3643 }, { "epoch": 0.18432899994941576, "grad_norm": 7.974698543548584, "learning_rate": 9.254338329971137e-06, "loss": 0.425, "step": 3644 }, { "epoch": 0.1843795841974809, "grad_norm": 9.15591049194336, "learning_rate": 9.253916597638247e-06, "loss": 0.4235, "step": 3645 }, { "epoch": 0.18443016844554605, "grad_norm": 11.612629890441895, "learning_rate": 9.253494755692198e-06, "loss": 0.4585, "step": 3646 }, { "epoch": 0.18448075269361122, "grad_norm": 6.609070301055908, "learning_rate": 9.253072804143864e-06, "loss": 0.4016, "step": 3647 }, { "epoch": 0.18453133694167637, "grad_norm": 20.112356185913086, "learning_rate": 9.252650743004115e-06, "loss": 0.4636, "step": 3648 }, { "epoch": 0.1845819211897415, "grad_norm": 7.490131378173828, "learning_rate": 9.25222857228383e-06, "loss": 0.4073, "step": 3649 }, { "epoch": 0.18463250543780665, "grad_norm": 7.724233150482178, "learning_rate": 9.251806291993885e-06, "loss": 0.5451, "step": 3650 }, { "epoch": 0.18468308968587183, "grad_norm": 10.97632884979248, "learning_rate": 9.25138390214516e-06, "loss": 0.5225, "step": 3651 }, { "epoch": 0.18473367393393697, "grad_norm": 7.097207069396973, "learning_rate": 9.250961402748544e-06, "loss": 0.4768, "step": 3652 }, { "epoch": 0.18478425818200211, "grad_norm": 10.498035430908203, "learning_rate": 9.250538793814919e-06, "loss": 0.4484, "step": 3653 }, { "epoch": 0.1848348424300673, "grad_norm": 17.18030548095703, "learning_rate": 9.250116075355175e-06, "loss": 0.566, "step": 3654 }, { "epoch": 0.18488542667813243, "grad_norm": 7.335501670837402, "learning_rate": 9.249693247380205e-06, "loss": 0.4305, "step": 3655 }, { "epoch": 0.18493601092619757, "grad_norm": 6.4563069343566895, "learning_rate": 9.249270309900908e-06, "loss": 0.3767, "step": 3656 }, { "epoch": 0.18498659517426275, "grad_norm": 7.347417831420898, "learning_rate": 9.248847262928175e-06, "loss": 0.4049, "step": 3657 }, { "epoch": 0.1850371794223279, "grad_norm": 5.612344264984131, "learning_rate": 9.248424106472912e-06, "loss": 0.3205, "step": 3658 }, { "epoch": 0.18508776367039304, "grad_norm": 6.8376383781433105, "learning_rate": 9.248000840546024e-06, "loss": 0.3324, "step": 3659 }, { "epoch": 0.18513834791845818, "grad_norm": 6.851396083831787, "learning_rate": 9.247577465158412e-06, "loss": 0.4045, "step": 3660 }, { "epoch": 0.18518893216652335, "grad_norm": 16.753808975219727, "learning_rate": 9.247153980320988e-06, "loss": 0.6106, "step": 3661 }, { "epoch": 0.1852395164145885, "grad_norm": 4.939150810241699, "learning_rate": 9.246730386044667e-06, "loss": 0.3948, "step": 3662 }, { "epoch": 0.18529010066265364, "grad_norm": 9.158352851867676, "learning_rate": 9.24630668234036e-06, "loss": 0.346, "step": 3663 }, { "epoch": 0.1853406849107188, "grad_norm": 14.717411041259766, "learning_rate": 9.245882869218988e-06, "loss": 0.5495, "step": 3664 }, { "epoch": 0.18539126915878396, "grad_norm": 8.805215835571289, "learning_rate": 9.245458946691467e-06, "loss": 0.4428, "step": 3665 }, { "epoch": 0.1854418534068491, "grad_norm": 16.964309692382812, "learning_rate": 9.245034914768725e-06, "loss": 0.4839, "step": 3666 }, { "epoch": 0.18549243765491427, "grad_norm": 11.720338821411133, "learning_rate": 9.244610773461687e-06, "loss": 0.5137, "step": 3667 }, { "epoch": 0.18554302190297942, "grad_norm": 6.861763000488281, "learning_rate": 9.244186522781282e-06, "loss": 0.4777, "step": 3668 }, { "epoch": 0.18559360615104456, "grad_norm": 18.8847599029541, "learning_rate": 9.243762162738442e-06, "loss": 0.5829, "step": 3669 }, { "epoch": 0.1856441903991097, "grad_norm": 11.624095916748047, "learning_rate": 9.243337693344099e-06, "loss": 0.4168, "step": 3670 }, { "epoch": 0.18569477464717488, "grad_norm": 9.997960090637207, "learning_rate": 9.242913114609195e-06, "loss": 0.5645, "step": 3671 }, { "epoch": 0.18574535889524002, "grad_norm": 8.100739479064941, "learning_rate": 9.242488426544668e-06, "loss": 0.4911, "step": 3672 }, { "epoch": 0.18579594314330516, "grad_norm": 6.590935230255127, "learning_rate": 9.242063629161462e-06, "loss": 0.3912, "step": 3673 }, { "epoch": 0.18584652739137034, "grad_norm": 6.394538879394531, "learning_rate": 9.241638722470522e-06, "loss": 0.469, "step": 3674 }, { "epoch": 0.18589711163943548, "grad_norm": 8.257227897644043, "learning_rate": 9.241213706482796e-06, "loss": 0.4849, "step": 3675 }, { "epoch": 0.18594769588750062, "grad_norm": 8.674476623535156, "learning_rate": 9.24078858120924e-06, "loss": 0.4544, "step": 3676 }, { "epoch": 0.1859982801355658, "grad_norm": 11.805425643920898, "learning_rate": 9.240363346660803e-06, "loss": 0.4811, "step": 3677 }, { "epoch": 0.18604886438363094, "grad_norm": 8.232301712036133, "learning_rate": 9.239938002848446e-06, "loss": 0.4758, "step": 3678 }, { "epoch": 0.18609944863169608, "grad_norm": 10.330370903015137, "learning_rate": 9.239512549783128e-06, "loss": 0.4689, "step": 3679 }, { "epoch": 0.18615003287976123, "grad_norm": 12.763068199157715, "learning_rate": 9.239086987475812e-06, "loss": 0.5431, "step": 3680 }, { "epoch": 0.1862006171278264, "grad_norm": 26.577037811279297, "learning_rate": 9.238661315937464e-06, "loss": 0.5781, "step": 3681 }, { "epoch": 0.18625120137589155, "grad_norm": 9.804917335510254, "learning_rate": 9.238235535179048e-06, "loss": 0.5559, "step": 3682 }, { "epoch": 0.1863017856239567, "grad_norm": 8.599852561950684, "learning_rate": 9.237809645211541e-06, "loss": 0.3861, "step": 3683 }, { "epoch": 0.18635236987202186, "grad_norm": 7.792324542999268, "learning_rate": 9.23738364604592e-06, "loss": 0.4657, "step": 3684 }, { "epoch": 0.186402954120087, "grad_norm": 5.229898929595947, "learning_rate": 9.236957537693153e-06, "loss": 0.3916, "step": 3685 }, { "epoch": 0.18645353836815215, "grad_norm": 17.685243606567383, "learning_rate": 9.236531320164224e-06, "loss": 0.5454, "step": 3686 }, { "epoch": 0.18650412261621732, "grad_norm": 7.875107288360596, "learning_rate": 9.236104993470116e-06, "loss": 0.5438, "step": 3687 }, { "epoch": 0.18655470686428247, "grad_norm": 7.995726108551025, "learning_rate": 9.235678557621815e-06, "loss": 0.4359, "step": 3688 }, { "epoch": 0.1866052911123476, "grad_norm": 12.57959270477295, "learning_rate": 9.23525201263031e-06, "loss": 0.516, "step": 3689 }, { "epoch": 0.18665587536041275, "grad_norm": 5.604857444763184, "learning_rate": 9.234825358506587e-06, "loss": 0.4263, "step": 3690 }, { "epoch": 0.18670645960847793, "grad_norm": 38.174949645996094, "learning_rate": 9.234398595261647e-06, "loss": 0.5558, "step": 3691 }, { "epoch": 0.18675704385654307, "grad_norm": 5.635927677154541, "learning_rate": 9.23397172290648e-06, "loss": 0.47, "step": 3692 }, { "epoch": 0.18680762810460821, "grad_norm": 10.872624397277832, "learning_rate": 9.233544741452089e-06, "loss": 0.5202, "step": 3693 }, { "epoch": 0.1868582123526734, "grad_norm": 9.226764678955078, "learning_rate": 9.233117650909478e-06, "loss": 0.5254, "step": 3694 }, { "epoch": 0.18690879660073853, "grad_norm": 11.482514381408691, "learning_rate": 9.232690451289647e-06, "loss": 0.5992, "step": 3695 }, { "epoch": 0.18695938084880367, "grad_norm": 11.807490348815918, "learning_rate": 9.232263142603608e-06, "loss": 0.5032, "step": 3696 }, { "epoch": 0.18700996509686885, "grad_norm": 4.839501857757568, "learning_rate": 9.231835724862371e-06, "loss": 0.419, "step": 3697 }, { "epoch": 0.187060549344934, "grad_norm": 9.04245662689209, "learning_rate": 9.231408198076948e-06, "loss": 0.5478, "step": 3698 }, { "epoch": 0.18711113359299913, "grad_norm": 5.008039474487305, "learning_rate": 9.230980562258357e-06, "loss": 0.4306, "step": 3699 }, { "epoch": 0.18716171784106428, "grad_norm": 9.17542839050293, "learning_rate": 9.230552817417616e-06, "loss": 0.4199, "step": 3700 }, { "epoch": 0.18721230208912945, "grad_norm": 7.689932823181152, "learning_rate": 9.230124963565747e-06, "loss": 0.4944, "step": 3701 }, { "epoch": 0.1872628863371946, "grad_norm": 7.034068584442139, "learning_rate": 9.229697000713775e-06, "loss": 0.4063, "step": 3702 }, { "epoch": 0.18731347058525974, "grad_norm": 6.617611408233643, "learning_rate": 9.22926892887273e-06, "loss": 0.4312, "step": 3703 }, { "epoch": 0.1873640548333249, "grad_norm": 6.74686861038208, "learning_rate": 9.228840748053637e-06, "loss": 0.4003, "step": 3704 }, { "epoch": 0.18741463908139006, "grad_norm": 11.63471508026123, "learning_rate": 9.228412458267534e-06, "loss": 0.5195, "step": 3705 }, { "epoch": 0.1874652233294552, "grad_norm": 10.384795188903809, "learning_rate": 9.227984059525454e-06, "loss": 0.4584, "step": 3706 }, { "epoch": 0.18751580757752037, "grad_norm": 6.9211530685424805, "learning_rate": 9.227555551838438e-06, "loss": 0.4546, "step": 3707 }, { "epoch": 0.18756639182558552, "grad_norm": 11.464111328125, "learning_rate": 9.227126935217527e-06, "loss": 0.4746, "step": 3708 }, { "epoch": 0.18761697607365066, "grad_norm": 36.17207717895508, "learning_rate": 9.226698209673764e-06, "loss": 0.5403, "step": 3709 }, { "epoch": 0.1876675603217158, "grad_norm": 11.986858367919922, "learning_rate": 9.226269375218198e-06, "loss": 0.5426, "step": 3710 }, { "epoch": 0.18771814456978098, "grad_norm": 6.392386436462402, "learning_rate": 9.22584043186188e-06, "loss": 0.4375, "step": 3711 }, { "epoch": 0.18776872881784612, "grad_norm": 9.470037460327148, "learning_rate": 9.225411379615859e-06, "loss": 0.4077, "step": 3712 }, { "epoch": 0.18781931306591126, "grad_norm": 15.713162422180176, "learning_rate": 9.224982218491191e-06, "loss": 0.5523, "step": 3713 }, { "epoch": 0.18786989731397644, "grad_norm": 12.503597259521484, "learning_rate": 9.224552948498941e-06, "loss": 0.4695, "step": 3714 }, { "epoch": 0.18792048156204158, "grad_norm": 7.598849773406982, "learning_rate": 9.224123569650162e-06, "loss": 0.5193, "step": 3715 }, { "epoch": 0.18797106581010672, "grad_norm": 6.839876651763916, "learning_rate": 9.223694081955923e-06, "loss": 0.3853, "step": 3716 }, { "epoch": 0.1880216500581719, "grad_norm": 11.740396499633789, "learning_rate": 9.22326448542729e-06, "loss": 0.4458, "step": 3717 }, { "epoch": 0.18807223430623704, "grad_norm": 11.611669540405273, "learning_rate": 9.222834780075329e-06, "loss": 0.5423, "step": 3718 }, { "epoch": 0.18812281855430218, "grad_norm": 9.50759506225586, "learning_rate": 9.222404965911119e-06, "loss": 0.4324, "step": 3719 }, { "epoch": 0.18817340280236736, "grad_norm": 10.020515441894531, "learning_rate": 9.22197504294573e-06, "loss": 0.5092, "step": 3720 }, { "epoch": 0.1882239870504325, "grad_norm": 22.110496520996094, "learning_rate": 9.221545011190243e-06, "loss": 0.6409, "step": 3721 }, { "epoch": 0.18827457129849764, "grad_norm": 5.512803077697754, "learning_rate": 9.221114870655736e-06, "loss": 0.4514, "step": 3722 }, { "epoch": 0.1883251555465628, "grad_norm": 8.909692764282227, "learning_rate": 9.220684621353296e-06, "loss": 0.4647, "step": 3723 }, { "epoch": 0.18837573979462796, "grad_norm": 7.303452968597412, "learning_rate": 9.220254263294008e-06, "loss": 0.3499, "step": 3724 }, { "epoch": 0.1884263240426931, "grad_norm": 7.805664539337158, "learning_rate": 9.21982379648896e-06, "loss": 0.4383, "step": 3725 }, { "epoch": 0.18847690829075825, "grad_norm": 8.311938285827637, "learning_rate": 9.219393220949247e-06, "loss": 0.5292, "step": 3726 }, { "epoch": 0.18852749253882342, "grad_norm": 16.774412155151367, "learning_rate": 9.218962536685961e-06, "loss": 0.5167, "step": 3727 }, { "epoch": 0.18857807678688857, "grad_norm": 13.631425857543945, "learning_rate": 9.218531743710201e-06, "loss": 0.6784, "step": 3728 }, { "epoch": 0.1886286610349537, "grad_norm": 11.83047866821289, "learning_rate": 9.218100842033068e-06, "loss": 0.4993, "step": 3729 }, { "epoch": 0.18867924528301888, "grad_norm": 6.557270526885986, "learning_rate": 9.217669831665666e-06, "loss": 0.4745, "step": 3730 }, { "epoch": 0.18872982953108403, "grad_norm": 7.804358005523682, "learning_rate": 9.217238712619096e-06, "loss": 0.4929, "step": 3731 }, { "epoch": 0.18878041377914917, "grad_norm": 9.346092224121094, "learning_rate": 9.216807484904474e-06, "loss": 0.3349, "step": 3732 }, { "epoch": 0.18883099802721431, "grad_norm": 10.348397254943848, "learning_rate": 9.216376148532907e-06, "loss": 0.4715, "step": 3733 }, { "epoch": 0.18888158227527949, "grad_norm": 10.669370651245117, "learning_rate": 9.215944703515511e-06, "loss": 0.4609, "step": 3734 }, { "epoch": 0.18893216652334463, "grad_norm": 8.556875228881836, "learning_rate": 9.215513149863404e-06, "loss": 0.5126, "step": 3735 }, { "epoch": 0.18898275077140977, "grad_norm": 4.862789630889893, "learning_rate": 9.215081487587705e-06, "loss": 0.3597, "step": 3736 }, { "epoch": 0.18903333501947495, "grad_norm": 9.22217082977295, "learning_rate": 9.214649716699536e-06, "loss": 0.5384, "step": 3737 }, { "epoch": 0.1890839192675401, "grad_norm": 9.175010681152344, "learning_rate": 9.214217837210025e-06, "loss": 0.5713, "step": 3738 }, { "epoch": 0.18913450351560523, "grad_norm": 13.226832389831543, "learning_rate": 9.2137858491303e-06, "loss": 0.5512, "step": 3739 }, { "epoch": 0.1891850877636704, "grad_norm": 8.35545825958252, "learning_rate": 9.21335375247149e-06, "loss": 0.4138, "step": 3740 }, { "epoch": 0.18923567201173555, "grad_norm": 7.351724147796631, "learning_rate": 9.212921547244731e-06, "loss": 0.3951, "step": 3741 }, { "epoch": 0.1892862562598007, "grad_norm": 6.962212562561035, "learning_rate": 9.21248923346116e-06, "loss": 0.4636, "step": 3742 }, { "epoch": 0.18933684050786584, "grad_norm": 8.808246612548828, "learning_rate": 9.212056811131917e-06, "loss": 0.5381, "step": 3743 }, { "epoch": 0.189387424755931, "grad_norm": 15.223170280456543, "learning_rate": 9.211624280268143e-06, "loss": 0.5484, "step": 3744 }, { "epoch": 0.18943800900399616, "grad_norm": 5.9360175132751465, "learning_rate": 9.211191640880986e-06, "loss": 0.4173, "step": 3745 }, { "epoch": 0.1894885932520613, "grad_norm": 6.700013637542725, "learning_rate": 9.21075889298159e-06, "loss": 0.3979, "step": 3746 }, { "epoch": 0.18953917750012647, "grad_norm": 9.7107515335083, "learning_rate": 9.210326036581108e-06, "loss": 0.5566, "step": 3747 }, { "epoch": 0.18958976174819162, "grad_norm": 9.63204288482666, "learning_rate": 9.209893071690694e-06, "loss": 0.4627, "step": 3748 }, { "epoch": 0.18964034599625676, "grad_norm": 8.005133628845215, "learning_rate": 9.209459998321504e-06, "loss": 0.3614, "step": 3749 }, { "epoch": 0.18969093024432193, "grad_norm": 9.749958038330078, "learning_rate": 9.209026816484698e-06, "loss": 0.4482, "step": 3750 }, { "epoch": 0.18974151449238708, "grad_norm": 6.3161702156066895, "learning_rate": 9.208593526191437e-06, "loss": 0.4823, "step": 3751 }, { "epoch": 0.18979209874045222, "grad_norm": 8.750696182250977, "learning_rate": 9.208160127452885e-06, "loss": 0.5393, "step": 3752 }, { "epoch": 0.18984268298851736, "grad_norm": 11.814230918884277, "learning_rate": 9.207726620280212e-06, "loss": 0.4901, "step": 3753 }, { "epoch": 0.18989326723658254, "grad_norm": 18.43352699279785, "learning_rate": 9.207293004684587e-06, "loss": 0.7085, "step": 3754 }, { "epoch": 0.18994385148464768, "grad_norm": 6.2038679122924805, "learning_rate": 9.206859280677185e-06, "loss": 0.4621, "step": 3755 }, { "epoch": 0.18999443573271282, "grad_norm": 10.43114948272705, "learning_rate": 9.20642544826918e-06, "loss": 0.4914, "step": 3756 }, { "epoch": 0.190045019980778, "grad_norm": 10.576021194458008, "learning_rate": 9.20599150747175e-06, "loss": 0.5129, "step": 3757 }, { "epoch": 0.19009560422884314, "grad_norm": 7.879589557647705, "learning_rate": 9.205557458296081e-06, "loss": 0.4878, "step": 3758 }, { "epoch": 0.19014618847690828, "grad_norm": 7.308587551116943, "learning_rate": 9.205123300753352e-06, "loss": 0.4118, "step": 3759 }, { "epoch": 0.19019677272497346, "grad_norm": 15.001818656921387, "learning_rate": 9.204689034854752e-06, "loss": 0.6726, "step": 3760 }, { "epoch": 0.1902473569730386, "grad_norm": 7.526678085327148, "learning_rate": 9.204254660611473e-06, "loss": 0.3495, "step": 3761 }, { "epoch": 0.19029794122110374, "grad_norm": 9.061304092407227, "learning_rate": 9.203820178034705e-06, "loss": 0.4608, "step": 3762 }, { "epoch": 0.1903485254691689, "grad_norm": 10.945981979370117, "learning_rate": 9.203385587135647e-06, "loss": 0.5553, "step": 3763 }, { "epoch": 0.19039910971723406, "grad_norm": 12.812996864318848, "learning_rate": 9.202950887925495e-06, "loss": 0.5885, "step": 3764 }, { "epoch": 0.1904496939652992, "grad_norm": 4.539328098297119, "learning_rate": 9.202516080415449e-06, "loss": 0.3913, "step": 3765 }, { "epoch": 0.19050027821336435, "grad_norm": 8.759444236755371, "learning_rate": 9.202081164616716e-06, "loss": 0.5685, "step": 3766 }, { "epoch": 0.19055086246142952, "grad_norm": 11.491006851196289, "learning_rate": 9.201646140540499e-06, "loss": 0.5084, "step": 3767 }, { "epoch": 0.19060144670949467, "grad_norm": 8.128985404968262, "learning_rate": 9.201211008198013e-06, "loss": 0.4533, "step": 3768 }, { "epoch": 0.1906520309575598, "grad_norm": 6.165907859802246, "learning_rate": 9.200775767600462e-06, "loss": 0.4301, "step": 3769 }, { "epoch": 0.19070261520562498, "grad_norm": 7.126716613769531, "learning_rate": 9.200340418759069e-06, "loss": 0.506, "step": 3770 }, { "epoch": 0.19075319945369013, "grad_norm": 10.255927085876465, "learning_rate": 9.19990496168505e-06, "loss": 0.4765, "step": 3771 }, { "epoch": 0.19080378370175527, "grad_norm": 9.158127784729004, "learning_rate": 9.199469396389623e-06, "loss": 0.5214, "step": 3772 }, { "epoch": 0.1908543679498204, "grad_norm": 10.356210708618164, "learning_rate": 9.199033722884012e-06, "loss": 0.4302, "step": 3773 }, { "epoch": 0.19090495219788559, "grad_norm": 10.288068771362305, "learning_rate": 9.198597941179447e-06, "loss": 0.5138, "step": 3774 }, { "epoch": 0.19095553644595073, "grad_norm": 11.409392356872559, "learning_rate": 9.198162051287151e-06, "loss": 0.3393, "step": 3775 }, { "epoch": 0.19100612069401587, "grad_norm": 6.50680685043335, "learning_rate": 9.197726053218363e-06, "loss": 0.4777, "step": 3776 }, { "epoch": 0.19105670494208105, "grad_norm": 6.661404132843018, "learning_rate": 9.197289946984311e-06, "loss": 0.514, "step": 3777 }, { "epoch": 0.1911072891901462, "grad_norm": 10.523578643798828, "learning_rate": 9.196853732596236e-06, "loss": 0.4801, "step": 3778 }, { "epoch": 0.19115787343821133, "grad_norm": 11.258031845092773, "learning_rate": 9.196417410065377e-06, "loss": 0.5257, "step": 3779 }, { "epoch": 0.1912084576862765, "grad_norm": 7.254378795623779, "learning_rate": 9.195980979402979e-06, "loss": 0.4302, "step": 3780 }, { "epoch": 0.19125904193434165, "grad_norm": 14.5985107421875, "learning_rate": 9.195544440620285e-06, "loss": 0.6139, "step": 3781 }, { "epoch": 0.1913096261824068, "grad_norm": 6.649686813354492, "learning_rate": 9.195107793728544e-06, "loss": 0.442, "step": 3782 }, { "epoch": 0.19136021043047194, "grad_norm": 8.525997161865234, "learning_rate": 9.194671038739009e-06, "loss": 0.3809, "step": 3783 }, { "epoch": 0.1914107946785371, "grad_norm": 6.143191814422607, "learning_rate": 9.194234175662933e-06, "loss": 0.4136, "step": 3784 }, { "epoch": 0.19146137892660225, "grad_norm": 9.207015037536621, "learning_rate": 9.193797204511574e-06, "loss": 0.5097, "step": 3785 }, { "epoch": 0.1915119631746674, "grad_norm": 13.907605171203613, "learning_rate": 9.19336012529619e-06, "loss": 0.4964, "step": 3786 }, { "epoch": 0.19156254742273257, "grad_norm": 18.30555534362793, "learning_rate": 9.192922938028043e-06, "loss": 0.72, "step": 3787 }, { "epoch": 0.19161313167079772, "grad_norm": 11.476895332336426, "learning_rate": 9.1924856427184e-06, "loss": 0.5555, "step": 3788 }, { "epoch": 0.19166371591886286, "grad_norm": 11.147019386291504, "learning_rate": 9.192048239378528e-06, "loss": 0.5046, "step": 3789 }, { "epoch": 0.19171430016692803, "grad_norm": 6.240572452545166, "learning_rate": 9.1916107280197e-06, "loss": 0.3691, "step": 3790 }, { "epoch": 0.19176488441499318, "grad_norm": 6.386966228485107, "learning_rate": 9.191173108653189e-06, "loss": 0.4315, "step": 3791 }, { "epoch": 0.19181546866305832, "grad_norm": 7.0922532081604, "learning_rate": 9.190735381290267e-06, "loss": 0.4092, "step": 3792 }, { "epoch": 0.19186605291112346, "grad_norm": 9.664498329162598, "learning_rate": 9.190297545942219e-06, "loss": 0.407, "step": 3793 }, { "epoch": 0.19191663715918864, "grad_norm": 7.623237133026123, "learning_rate": 9.189859602620324e-06, "loss": 0.4024, "step": 3794 }, { "epoch": 0.19196722140725378, "grad_norm": 6.826162815093994, "learning_rate": 9.189421551335866e-06, "loss": 0.5353, "step": 3795 }, { "epoch": 0.19201780565531892, "grad_norm": 6.827606678009033, "learning_rate": 9.188983392100135e-06, "loss": 0.4753, "step": 3796 }, { "epoch": 0.1920683899033841, "grad_norm": 7.47191858291626, "learning_rate": 9.188545124924419e-06, "loss": 0.4508, "step": 3797 }, { "epoch": 0.19211897415144924, "grad_norm": 7.578713417053223, "learning_rate": 9.188106749820013e-06, "loss": 0.435, "step": 3798 }, { "epoch": 0.19216955839951438, "grad_norm": 15.737775802612305, "learning_rate": 9.187668266798212e-06, "loss": 0.5605, "step": 3799 }, { "epoch": 0.19222014264757956, "grad_norm": 8.678544044494629, "learning_rate": 9.187229675870315e-06, "loss": 0.5906, "step": 3800 }, { "epoch": 0.1922707268956447, "grad_norm": 11.292884826660156, "learning_rate": 9.186790977047622e-06, "loss": 0.5875, "step": 3801 }, { "epoch": 0.19232131114370984, "grad_norm": 7.363112449645996, "learning_rate": 9.186352170341439e-06, "loss": 0.4345, "step": 3802 }, { "epoch": 0.192371895391775, "grad_norm": 18.102880477905273, "learning_rate": 9.18591325576307e-06, "loss": 0.6317, "step": 3803 }, { "epoch": 0.19242247963984016, "grad_norm": 7.02587890625, "learning_rate": 9.18547423332383e-06, "loss": 0.4328, "step": 3804 }, { "epoch": 0.1924730638879053, "grad_norm": 7.533231258392334, "learning_rate": 9.18503510303503e-06, "loss": 0.4128, "step": 3805 }, { "epoch": 0.19252364813597045, "grad_norm": 9.015339851379395, "learning_rate": 9.18459586490798e-06, "loss": 0.4413, "step": 3806 }, { "epoch": 0.19257423238403562, "grad_norm": 6.539431095123291, "learning_rate": 9.184156518954004e-06, "loss": 0.4804, "step": 3807 }, { "epoch": 0.19262481663210076, "grad_norm": 7.019802093505859, "learning_rate": 9.183717065184421e-06, "loss": 0.4223, "step": 3808 }, { "epoch": 0.1926754008801659, "grad_norm": 11.695196151733398, "learning_rate": 9.183277503610555e-06, "loss": 0.3559, "step": 3809 }, { "epoch": 0.19272598512823108, "grad_norm": 5.86047887802124, "learning_rate": 9.18283783424373e-06, "loss": 0.4189, "step": 3810 }, { "epoch": 0.19277656937629623, "grad_norm": 7.230424880981445, "learning_rate": 9.182398057095279e-06, "loss": 0.543, "step": 3811 }, { "epoch": 0.19282715362436137, "grad_norm": 14.068099021911621, "learning_rate": 9.18195817217653e-06, "loss": 0.5985, "step": 3812 }, { "epoch": 0.1928777378724265, "grad_norm": 8.738609313964844, "learning_rate": 9.181518179498822e-06, "loss": 0.5211, "step": 3813 }, { "epoch": 0.19292832212049169, "grad_norm": 9.612886428833008, "learning_rate": 9.181078079073491e-06, "loss": 0.4158, "step": 3814 }, { "epoch": 0.19297890636855683, "grad_norm": 11.043137550354004, "learning_rate": 9.180637870911878e-06, "loss": 0.4798, "step": 3815 }, { "epoch": 0.19302949061662197, "grad_norm": 8.060087203979492, "learning_rate": 9.180197555025322e-06, "loss": 0.4743, "step": 3816 }, { "epoch": 0.19308007486468715, "grad_norm": 6.353874206542969, "learning_rate": 9.179757131425173e-06, "loss": 0.3971, "step": 3817 }, { "epoch": 0.1931306591127523, "grad_norm": 26.850265502929688, "learning_rate": 9.179316600122778e-06, "loss": 0.5323, "step": 3818 }, { "epoch": 0.19318124336081743, "grad_norm": 8.65318775177002, "learning_rate": 9.17887596112949e-06, "loss": 0.4462, "step": 3819 }, { "epoch": 0.1932318276088826, "grad_norm": 4.390823841094971, "learning_rate": 9.178435214456663e-06, "loss": 0.2749, "step": 3820 }, { "epoch": 0.19328241185694775, "grad_norm": 8.52634334564209, "learning_rate": 9.17799436011565e-06, "loss": 0.5179, "step": 3821 }, { "epoch": 0.1933329961050129, "grad_norm": 9.812446594238281, "learning_rate": 9.177553398117816e-06, "loss": 0.4591, "step": 3822 }, { "epoch": 0.19338358035307804, "grad_norm": 8.993816375732422, "learning_rate": 9.177112328474522e-06, "loss": 0.5042, "step": 3823 }, { "epoch": 0.1934341646011432, "grad_norm": 6.213815689086914, "learning_rate": 9.176671151197131e-06, "loss": 0.4322, "step": 3824 }, { "epoch": 0.19348474884920835, "grad_norm": 9.602079391479492, "learning_rate": 9.176229866297014e-06, "loss": 0.4167, "step": 3825 }, { "epoch": 0.1935353330972735, "grad_norm": 9.004448890686035, "learning_rate": 9.175788473785541e-06, "loss": 0.4332, "step": 3826 }, { "epoch": 0.19358591734533867, "grad_norm": 12.215804100036621, "learning_rate": 9.175346973674082e-06, "loss": 0.5681, "step": 3827 }, { "epoch": 0.19363650159340381, "grad_norm": 5.656301021575928, "learning_rate": 9.17490536597402e-06, "loss": 0.4785, "step": 3828 }, { "epoch": 0.19368708584146896, "grad_norm": 6.912182331085205, "learning_rate": 9.17446365069673e-06, "loss": 0.4547, "step": 3829 }, { "epoch": 0.19373767008953413, "grad_norm": 6.141901969909668, "learning_rate": 9.174021827853593e-06, "loss": 0.4261, "step": 3830 }, { "epoch": 0.19378825433759927, "grad_norm": 12.541425704956055, "learning_rate": 9.173579897456e-06, "loss": 0.4322, "step": 3831 }, { "epoch": 0.19383883858566442, "grad_norm": 9.429183959960938, "learning_rate": 9.17313785951533e-06, "loss": 0.4404, "step": 3832 }, { "epoch": 0.19388942283372956, "grad_norm": 23.788053512573242, "learning_rate": 9.172695714042977e-06, "loss": 0.5831, "step": 3833 }, { "epoch": 0.19394000708179474, "grad_norm": 9.7805814743042, "learning_rate": 9.172253461050335e-06, "loss": 0.4949, "step": 3834 }, { "epoch": 0.19399059132985988, "grad_norm": 11.073572158813477, "learning_rate": 9.1718111005488e-06, "loss": 0.4561, "step": 3835 }, { "epoch": 0.19404117557792502, "grad_norm": 9.371878623962402, "learning_rate": 9.171368632549769e-06, "loss": 0.5141, "step": 3836 }, { "epoch": 0.1940917598259902, "grad_norm": 9.357481002807617, "learning_rate": 9.170926057064644e-06, "loss": 0.5279, "step": 3837 }, { "epoch": 0.19414234407405534, "grad_norm": 9.954877853393555, "learning_rate": 9.170483374104829e-06, "loss": 0.357, "step": 3838 }, { "epoch": 0.19419292832212048, "grad_norm": 10.478208541870117, "learning_rate": 9.17004058368173e-06, "loss": 0.5126, "step": 3839 }, { "epoch": 0.19424351257018566, "grad_norm": 11.713189125061035, "learning_rate": 9.169597685806758e-06, "loss": 0.4113, "step": 3840 }, { "epoch": 0.1942940968182508, "grad_norm": 9.475699424743652, "learning_rate": 9.169154680491324e-06, "loss": 0.478, "step": 3841 }, { "epoch": 0.19434468106631594, "grad_norm": 10.569730758666992, "learning_rate": 9.168711567746844e-06, "loss": 0.4818, "step": 3842 }, { "epoch": 0.1943952653143811, "grad_norm": 13.14992904663086, "learning_rate": 9.168268347584738e-06, "loss": 0.5215, "step": 3843 }, { "epoch": 0.19444584956244626, "grad_norm": 9.190136909484863, "learning_rate": 9.167825020016423e-06, "loss": 0.5134, "step": 3844 }, { "epoch": 0.1944964338105114, "grad_norm": 13.496003150939941, "learning_rate": 9.167381585053325e-06, "loss": 0.5181, "step": 3845 }, { "epoch": 0.19454701805857655, "grad_norm": 8.838666915893555, "learning_rate": 9.166938042706868e-06, "loss": 0.4409, "step": 3846 }, { "epoch": 0.19459760230664172, "grad_norm": 13.078628540039062, "learning_rate": 9.166494392988486e-06, "loss": 0.4514, "step": 3847 }, { "epoch": 0.19464818655470686, "grad_norm": 9.380093574523926, "learning_rate": 9.166050635909604e-06, "loss": 0.4818, "step": 3848 }, { "epoch": 0.194698770802772, "grad_norm": 11.502660751342773, "learning_rate": 9.165606771481661e-06, "loss": 0.3665, "step": 3849 }, { "epoch": 0.19474935505083718, "grad_norm": 6.779660224914551, "learning_rate": 9.165162799716092e-06, "loss": 0.4226, "step": 3850 }, { "epoch": 0.19479993929890232, "grad_norm": 6.6331562995910645, "learning_rate": 9.164718720624338e-06, "loss": 0.5093, "step": 3851 }, { "epoch": 0.19485052354696747, "grad_norm": 9.628421783447266, "learning_rate": 9.164274534217844e-06, "loss": 0.5038, "step": 3852 }, { "epoch": 0.1949011077950326, "grad_norm": 13.960169792175293, "learning_rate": 9.163830240508054e-06, "loss": 0.4607, "step": 3853 }, { "epoch": 0.19495169204309779, "grad_norm": 10.04497241973877, "learning_rate": 9.163385839506414e-06, "loss": 0.523, "step": 3854 }, { "epoch": 0.19500227629116293, "grad_norm": 9.858222961425781, "learning_rate": 9.162941331224377e-06, "loss": 0.4845, "step": 3855 }, { "epoch": 0.19505286053922807, "grad_norm": 8.058899879455566, "learning_rate": 9.162496715673399e-06, "loss": 0.4042, "step": 3856 }, { "epoch": 0.19510344478729325, "grad_norm": 6.419954776763916, "learning_rate": 9.162051992864934e-06, "loss": 0.4201, "step": 3857 }, { "epoch": 0.1951540290353584, "grad_norm": 14.914642333984375, "learning_rate": 9.161607162810442e-06, "loss": 0.4261, "step": 3858 }, { "epoch": 0.19520461328342353, "grad_norm": 9.056029319763184, "learning_rate": 9.161162225521384e-06, "loss": 0.4526, "step": 3859 }, { "epoch": 0.1952551975314887, "grad_norm": 5.6134257316589355, "learning_rate": 9.160717181009228e-06, "loss": 0.3578, "step": 3860 }, { "epoch": 0.19530578177955385, "grad_norm": 10.308826446533203, "learning_rate": 9.16027202928544e-06, "loss": 0.4152, "step": 3861 }, { "epoch": 0.195356366027619, "grad_norm": 8.17160415649414, "learning_rate": 9.15982677036149e-06, "loss": 0.4707, "step": 3862 }, { "epoch": 0.19540695027568414, "grad_norm": 7.31645393371582, "learning_rate": 9.159381404248851e-06, "loss": 0.4447, "step": 3863 }, { "epoch": 0.1954575345237493, "grad_norm": 7.830051898956299, "learning_rate": 9.158935930959002e-06, "loss": 0.5416, "step": 3864 }, { "epoch": 0.19550811877181445, "grad_norm": 10.043015480041504, "learning_rate": 9.158490350503418e-06, "loss": 0.528, "step": 3865 }, { "epoch": 0.1955587030198796, "grad_norm": 4.946963310241699, "learning_rate": 9.158044662893581e-06, "loss": 0.4513, "step": 3866 }, { "epoch": 0.19560928726794477, "grad_norm": 7.530138969421387, "learning_rate": 9.15759886814098e-06, "loss": 0.4619, "step": 3867 }, { "epoch": 0.19565987151600991, "grad_norm": 6.896417617797852, "learning_rate": 9.157152966257095e-06, "loss": 0.4123, "step": 3868 }, { "epoch": 0.19571045576407506, "grad_norm": 7.517919063568115, "learning_rate": 9.156706957253421e-06, "loss": 0.4374, "step": 3869 }, { "epoch": 0.19576104001214023, "grad_norm": 8.58736515045166, "learning_rate": 9.156260841141448e-06, "loss": 0.4533, "step": 3870 }, { "epoch": 0.19581162426020537, "grad_norm": 7.512094497680664, "learning_rate": 9.155814617932673e-06, "loss": 0.4222, "step": 3871 }, { "epoch": 0.19586220850827052, "grad_norm": 8.205207824707031, "learning_rate": 9.15536828763859e-06, "loss": 0.4347, "step": 3872 }, { "epoch": 0.1959127927563357, "grad_norm": 15.02078628540039, "learning_rate": 9.154921850270707e-06, "loss": 0.4228, "step": 3873 }, { "epoch": 0.19596337700440083, "grad_norm": 6.252734184265137, "learning_rate": 9.154475305840523e-06, "loss": 0.3493, "step": 3874 }, { "epoch": 0.19601396125246598, "grad_norm": 6.321897029876709, "learning_rate": 9.154028654359544e-06, "loss": 0.4994, "step": 3875 }, { "epoch": 0.19606454550053112, "grad_norm": 9.064288139343262, "learning_rate": 9.15358189583928e-06, "loss": 0.5023, "step": 3876 }, { "epoch": 0.1961151297485963, "grad_norm": 5.867040157318115, "learning_rate": 9.153135030291245e-06, "loss": 0.4185, "step": 3877 }, { "epoch": 0.19616571399666144, "grad_norm": 10.322940826416016, "learning_rate": 9.15268805772695e-06, "loss": 0.5155, "step": 3878 }, { "epoch": 0.19621629824472658, "grad_norm": 10.94483757019043, "learning_rate": 9.152240978157916e-06, "loss": 0.4973, "step": 3879 }, { "epoch": 0.19626688249279176, "grad_norm": 8.845471382141113, "learning_rate": 9.151793791595661e-06, "loss": 0.526, "step": 3880 }, { "epoch": 0.1963174667408569, "grad_norm": 8.540541648864746, "learning_rate": 9.151346498051706e-06, "loss": 0.4104, "step": 3881 }, { "epoch": 0.19636805098892204, "grad_norm": 13.535531997680664, "learning_rate": 9.15089909753758e-06, "loss": 0.5129, "step": 3882 }, { "epoch": 0.19641863523698722, "grad_norm": 7.383550643920898, "learning_rate": 9.15045159006481e-06, "loss": 0.3328, "step": 3883 }, { "epoch": 0.19646921948505236, "grad_norm": 10.809625625610352, "learning_rate": 9.15000397564493e-06, "loss": 0.4355, "step": 3884 }, { "epoch": 0.1965198037331175, "grad_norm": 12.907096862792969, "learning_rate": 9.14955625428947e-06, "loss": 0.4963, "step": 3885 }, { "epoch": 0.19657038798118265, "grad_norm": 12.10734748840332, "learning_rate": 9.149108426009968e-06, "loss": 0.4397, "step": 3886 }, { "epoch": 0.19662097222924782, "grad_norm": 9.361889839172363, "learning_rate": 9.148660490817963e-06, "loss": 0.4424, "step": 3887 }, { "epoch": 0.19667155647731296, "grad_norm": 7.419383525848389, "learning_rate": 9.148212448724999e-06, "loss": 0.5002, "step": 3888 }, { "epoch": 0.1967221407253781, "grad_norm": 6.488208293914795, "learning_rate": 9.147764299742618e-06, "loss": 0.394, "step": 3889 }, { "epoch": 0.19677272497344328, "grad_norm": 10.620582580566406, "learning_rate": 9.14731604388237e-06, "loss": 0.4479, "step": 3890 }, { "epoch": 0.19682330922150842, "grad_norm": 11.050948143005371, "learning_rate": 9.146867681155805e-06, "loss": 0.5071, "step": 3891 }, { "epoch": 0.19687389346957357, "grad_norm": 8.033568382263184, "learning_rate": 9.146419211574478e-06, "loss": 0.3586, "step": 3892 }, { "epoch": 0.19692447771763874, "grad_norm": 10.673334121704102, "learning_rate": 9.14597063514994e-06, "loss": 0.5726, "step": 3893 }, { "epoch": 0.19697506196570388, "grad_norm": 52.42234420776367, "learning_rate": 9.145521951893756e-06, "loss": 0.5092, "step": 3894 }, { "epoch": 0.19702564621376903, "grad_norm": 9.200180053710938, "learning_rate": 9.145073161817484e-06, "loss": 0.512, "step": 3895 }, { "epoch": 0.19707623046183417, "grad_norm": 7.049967288970947, "learning_rate": 9.144624264932686e-06, "loss": 0.4665, "step": 3896 }, { "epoch": 0.19712681470989935, "grad_norm": 7.455905437469482, "learning_rate": 9.144175261250936e-06, "loss": 0.4409, "step": 3897 }, { "epoch": 0.1971773989579645, "grad_norm": 11.011543273925781, "learning_rate": 9.143726150783796e-06, "loss": 0.4889, "step": 3898 }, { "epoch": 0.19722798320602963, "grad_norm": 10.720088005065918, "learning_rate": 9.143276933542842e-06, "loss": 0.4629, "step": 3899 }, { "epoch": 0.1972785674540948, "grad_norm": 5.471521377563477, "learning_rate": 9.142827609539651e-06, "loss": 0.2695, "step": 3900 }, { "epoch": 0.19732915170215995, "grad_norm": 26.458635330200195, "learning_rate": 9.142378178785796e-06, "loss": 0.4999, "step": 3901 }, { "epoch": 0.1973797359502251, "grad_norm": 9.304418563842773, "learning_rate": 9.141928641292864e-06, "loss": 0.6153, "step": 3902 }, { "epoch": 0.19743032019829027, "grad_norm": 8.88347339630127, "learning_rate": 9.141478997072434e-06, "loss": 0.5172, "step": 3903 }, { "epoch": 0.1974809044463554, "grad_norm": 14.366424560546875, "learning_rate": 9.141029246136093e-06, "loss": 0.5575, "step": 3904 }, { "epoch": 0.19753148869442055, "grad_norm": 10.724090576171875, "learning_rate": 9.140579388495432e-06, "loss": 0.525, "step": 3905 }, { "epoch": 0.1975820729424857, "grad_norm": 12.855965614318848, "learning_rate": 9.140129424162042e-06, "loss": 0.4819, "step": 3906 }, { "epoch": 0.19763265719055087, "grad_norm": 8.364846229553223, "learning_rate": 9.139679353147513e-06, "loss": 0.49, "step": 3907 }, { "epoch": 0.19768324143861601, "grad_norm": 9.545039176940918, "learning_rate": 9.139229175463449e-06, "loss": 0.5325, "step": 3908 }, { "epoch": 0.19773382568668116, "grad_norm": 8.88047981262207, "learning_rate": 9.138778891121447e-06, "loss": 0.434, "step": 3909 }, { "epoch": 0.19778440993474633, "grad_norm": 12.223468780517578, "learning_rate": 9.13832850013311e-06, "loss": 0.526, "step": 3910 }, { "epoch": 0.19783499418281147, "grad_norm": 8.054129600524902, "learning_rate": 9.13787800251004e-06, "loss": 0.4573, "step": 3911 }, { "epoch": 0.19788557843087662, "grad_norm": 7.2136054039001465, "learning_rate": 9.137427398263852e-06, "loss": 0.4797, "step": 3912 }, { "epoch": 0.1979361626789418, "grad_norm": 10.25277328491211, "learning_rate": 9.136976687406153e-06, "loss": 0.4692, "step": 3913 }, { "epoch": 0.19798674692700693, "grad_norm": 12.943209648132324, "learning_rate": 9.136525869948556e-06, "loss": 0.5603, "step": 3914 }, { "epoch": 0.19803733117507208, "grad_norm": 14.60353946685791, "learning_rate": 9.13607494590268e-06, "loss": 0.4674, "step": 3915 }, { "epoch": 0.19808791542313722, "grad_norm": 7.044250011444092, "learning_rate": 9.135623915280142e-06, "loss": 0.4473, "step": 3916 }, { "epoch": 0.1981384996712024, "grad_norm": 7.528008937835693, "learning_rate": 9.135172778092565e-06, "loss": 0.3897, "step": 3917 }, { "epoch": 0.19818908391926754, "grad_norm": 6.465843677520752, "learning_rate": 9.134721534351573e-06, "loss": 0.4343, "step": 3918 }, { "epoch": 0.19823966816733268, "grad_norm": 7.57539701461792, "learning_rate": 9.134270184068797e-06, "loss": 0.343, "step": 3919 }, { "epoch": 0.19829025241539786, "grad_norm": 6.689176082611084, "learning_rate": 9.133818727255861e-06, "loss": 0.3695, "step": 3920 }, { "epoch": 0.198340836663463, "grad_norm": 10.104609489440918, "learning_rate": 9.133367163924402e-06, "loss": 0.4858, "step": 3921 }, { "epoch": 0.19839142091152814, "grad_norm": 7.904773235321045, "learning_rate": 9.132915494086056e-06, "loss": 0.418, "step": 3922 }, { "epoch": 0.19844200515959332, "grad_norm": 10.977496147155762, "learning_rate": 9.13246371775246e-06, "loss": 0.4819, "step": 3923 }, { "epoch": 0.19849258940765846, "grad_norm": 13.188556671142578, "learning_rate": 9.132011834935255e-06, "loss": 0.5322, "step": 3924 }, { "epoch": 0.1985431736557236, "grad_norm": 10.231035232543945, "learning_rate": 9.131559845646087e-06, "loss": 0.4507, "step": 3925 }, { "epoch": 0.19859375790378875, "grad_norm": 12.30190372467041, "learning_rate": 9.1311077498966e-06, "loss": 0.3133, "step": 3926 }, { "epoch": 0.19864434215185392, "grad_norm": 9.199724197387695, "learning_rate": 9.130655547698443e-06, "loss": 0.4484, "step": 3927 }, { "epoch": 0.19869492639991906, "grad_norm": 8.6375093460083, "learning_rate": 9.130203239063272e-06, "loss": 0.4973, "step": 3928 }, { "epoch": 0.1987455106479842, "grad_norm": 8.595280647277832, "learning_rate": 9.129750824002738e-06, "loss": 0.3682, "step": 3929 }, { "epoch": 0.19879609489604938, "grad_norm": 10.412153244018555, "learning_rate": 9.129298302528502e-06, "loss": 0.5077, "step": 3930 }, { "epoch": 0.19884667914411452, "grad_norm": 6.711459636688232, "learning_rate": 9.128845674652221e-06, "loss": 0.4014, "step": 3931 }, { "epoch": 0.19889726339217967, "grad_norm": 5.8456196784973145, "learning_rate": 9.128392940385562e-06, "loss": 0.4695, "step": 3932 }, { "epoch": 0.19894784764024484, "grad_norm": 4.415282726287842, "learning_rate": 9.127940099740187e-06, "loss": 0.4152, "step": 3933 }, { "epoch": 0.19899843188830998, "grad_norm": 6.769670009613037, "learning_rate": 9.127487152727765e-06, "loss": 0.3829, "step": 3934 }, { "epoch": 0.19904901613637513, "grad_norm": 5.265360355377197, "learning_rate": 9.12703409935997e-06, "loss": 0.3376, "step": 3935 }, { "epoch": 0.19909960038444027, "grad_norm": 6.309384346008301, "learning_rate": 9.126580939648475e-06, "loss": 0.4803, "step": 3936 }, { "epoch": 0.19915018463250544, "grad_norm": 9.94727611541748, "learning_rate": 9.126127673604957e-06, "loss": 0.4732, "step": 3937 }, { "epoch": 0.1992007688805706, "grad_norm": 7.887248992919922, "learning_rate": 9.125674301241095e-06, "loss": 0.3969, "step": 3938 }, { "epoch": 0.19925135312863573, "grad_norm": 6.838386058807373, "learning_rate": 9.12522082256857e-06, "loss": 0.4271, "step": 3939 }, { "epoch": 0.1993019373767009, "grad_norm": 8.273966789245605, "learning_rate": 9.12476723759907e-06, "loss": 0.4618, "step": 3940 }, { "epoch": 0.19935252162476605, "grad_norm": 15.734883308410645, "learning_rate": 9.12431354634428e-06, "loss": 0.404, "step": 3941 }, { "epoch": 0.1994031058728312, "grad_norm": 6.663207054138184, "learning_rate": 9.123859748815893e-06, "loss": 0.4797, "step": 3942 }, { "epoch": 0.19945369012089637, "grad_norm": 7.263354778289795, "learning_rate": 9.1234058450256e-06, "loss": 0.4558, "step": 3943 }, { "epoch": 0.1995042743689615, "grad_norm": 9.286980628967285, "learning_rate": 9.1229518349851e-06, "loss": 0.6238, "step": 3944 }, { "epoch": 0.19955485861702665, "grad_norm": 13.95743465423584, "learning_rate": 9.122497718706087e-06, "loss": 0.4467, "step": 3945 }, { "epoch": 0.1996054428650918, "grad_norm": 8.536444664001465, "learning_rate": 9.122043496200265e-06, "loss": 0.483, "step": 3946 }, { "epoch": 0.19965602711315697, "grad_norm": 22.935144424438477, "learning_rate": 9.121589167479341e-06, "loss": 0.5161, "step": 3947 }, { "epoch": 0.1997066113612221, "grad_norm": 7.871023178100586, "learning_rate": 9.121134732555017e-06, "loss": 0.5701, "step": 3948 }, { "epoch": 0.19975719560928726, "grad_norm": 7.1896796226501465, "learning_rate": 9.120680191439008e-06, "loss": 0.4977, "step": 3949 }, { "epoch": 0.19980777985735243, "grad_norm": 12.309907913208008, "learning_rate": 9.120225544143021e-06, "loss": 0.5319, "step": 3950 }, { "epoch": 0.19985836410541757, "grad_norm": 10.801752090454102, "learning_rate": 9.119770790678776e-06, "loss": 0.5489, "step": 3951 }, { "epoch": 0.19990894835348272, "grad_norm": 5.774764537811279, "learning_rate": 9.119315931057987e-06, "loss": 0.3699, "step": 3952 }, { "epoch": 0.1999595326015479, "grad_norm": 10.416803359985352, "learning_rate": 9.118860965292378e-06, "loss": 0.6305, "step": 3953 }, { "epoch": 0.20001011684961303, "grad_norm": 8.03298282623291, "learning_rate": 9.118405893393669e-06, "loss": 0.5192, "step": 3954 }, { "epoch": 0.20001011684961303, "eval_loss": 0.4523782432079315, "eval_runtime": 34.9081, "eval_samples_per_second": 45.777, "eval_steps_per_second": 5.729, "step": 3954 }, { "epoch": 0.20006070109767818, "grad_norm": 5.350276470184326, "learning_rate": 9.117950715373588e-06, "loss": 0.4657, "step": 3955 }, { "epoch": 0.20011128534574332, "grad_norm": 5.8954291343688965, "learning_rate": 9.117495431243864e-06, "loss": 0.4223, "step": 3956 }, { "epoch": 0.2001618695938085, "grad_norm": 6.898488998413086, "learning_rate": 9.117040041016227e-06, "loss": 0.3273, "step": 3957 }, { "epoch": 0.20021245384187364, "grad_norm": 7.920068264007568, "learning_rate": 9.116584544702414e-06, "loss": 0.547, "step": 3958 }, { "epoch": 0.20026303808993878, "grad_norm": 14.939473152160645, "learning_rate": 9.11612894231416e-06, "loss": 0.5343, "step": 3959 }, { "epoch": 0.20031362233800395, "grad_norm": 9.193106651306152, "learning_rate": 9.115673233863204e-06, "loss": 0.4448, "step": 3960 }, { "epoch": 0.2003642065860691, "grad_norm": 6.275464057922363, "learning_rate": 9.115217419361291e-06, "loss": 0.4215, "step": 3961 }, { "epoch": 0.20041479083413424, "grad_norm": 11.802879333496094, "learning_rate": 9.114761498820164e-06, "loss": 0.4551, "step": 3962 }, { "epoch": 0.20046537508219942, "grad_norm": 7.261431694030762, "learning_rate": 9.114305472251574e-06, "loss": 0.4596, "step": 3963 }, { "epoch": 0.20051595933026456, "grad_norm": 12.56523609161377, "learning_rate": 9.113849339667267e-06, "loss": 0.4451, "step": 3964 }, { "epoch": 0.2005665435783297, "grad_norm": 5.937071800231934, "learning_rate": 9.113393101079e-06, "loss": 0.3839, "step": 3965 }, { "epoch": 0.20061712782639485, "grad_norm": 6.563784122467041, "learning_rate": 9.11293675649853e-06, "loss": 0.4174, "step": 3966 }, { "epoch": 0.20066771207446002, "grad_norm": 7.496607780456543, "learning_rate": 9.112480305937613e-06, "loss": 0.3015, "step": 3967 }, { "epoch": 0.20071829632252516, "grad_norm": 9.640836715698242, "learning_rate": 9.112023749408012e-06, "loss": 0.5282, "step": 3968 }, { "epoch": 0.2007688805705903, "grad_norm": 7.921989917755127, "learning_rate": 9.111567086921492e-06, "loss": 0.5844, "step": 3969 }, { "epoch": 0.20081946481865548, "grad_norm": 13.119050025939941, "learning_rate": 9.111110318489818e-06, "loss": 0.4212, "step": 3970 }, { "epoch": 0.20087004906672062, "grad_norm": 10.789443016052246, "learning_rate": 9.11065344412476e-06, "loss": 0.5278, "step": 3971 }, { "epoch": 0.20092063331478577, "grad_norm": 12.917082786560059, "learning_rate": 9.110196463838092e-06, "loss": 0.5456, "step": 3972 }, { "epoch": 0.20097121756285094, "grad_norm": 6.275032043457031, "learning_rate": 9.109739377641591e-06, "loss": 0.3458, "step": 3973 }, { "epoch": 0.20102180181091608, "grad_norm": 8.382367134094238, "learning_rate": 9.109282185547033e-06, "loss": 0.4797, "step": 3974 }, { "epoch": 0.20107238605898123, "grad_norm": 11.32568645477295, "learning_rate": 9.108824887566197e-06, "loss": 0.4461, "step": 3975 }, { "epoch": 0.20112297030704637, "grad_norm": 7.611416816711426, "learning_rate": 9.108367483710867e-06, "loss": 0.3843, "step": 3976 }, { "epoch": 0.20117355455511154, "grad_norm": 10.877978324890137, "learning_rate": 9.107909973992833e-06, "loss": 0.463, "step": 3977 }, { "epoch": 0.2012241388031767, "grad_norm": 7.574005126953125, "learning_rate": 9.107452358423878e-06, "loss": 0.5128, "step": 3978 }, { "epoch": 0.20127472305124183, "grad_norm": 10.3985013961792, "learning_rate": 9.1069946370158e-06, "loss": 0.6136, "step": 3979 }, { "epoch": 0.201325307299307, "grad_norm": 11.888108253479004, "learning_rate": 9.10653680978039e-06, "loss": 0.5653, "step": 3980 }, { "epoch": 0.20137589154737215, "grad_norm": 4.491320610046387, "learning_rate": 9.106078876729444e-06, "loss": 0.5012, "step": 3981 }, { "epoch": 0.2014264757954373, "grad_norm": 8.541851043701172, "learning_rate": 9.105620837874762e-06, "loss": 0.5158, "step": 3982 }, { "epoch": 0.20147706004350247, "grad_norm": 12.422272682189941, "learning_rate": 9.10516269322815e-06, "loss": 0.5367, "step": 3983 }, { "epoch": 0.2015276442915676, "grad_norm": 7.4157633781433105, "learning_rate": 9.10470444280141e-06, "loss": 0.4123, "step": 3984 }, { "epoch": 0.20157822853963275, "grad_norm": 7.457277774810791, "learning_rate": 9.104246086606352e-06, "loss": 0.4655, "step": 3985 }, { "epoch": 0.2016288127876979, "grad_norm": 7.015416145324707, "learning_rate": 9.103787624654784e-06, "loss": 0.4372, "step": 3986 }, { "epoch": 0.20167939703576307, "grad_norm": 21.07842445373535, "learning_rate": 9.103329056958523e-06, "loss": 0.6167, "step": 3987 }, { "epoch": 0.2017299812838282, "grad_norm": 7.774074077606201, "learning_rate": 9.102870383529383e-06, "loss": 0.4628, "step": 3988 }, { "epoch": 0.20178056553189336, "grad_norm": 12.210125923156738, "learning_rate": 9.102411604379183e-06, "loss": 0.5463, "step": 3989 }, { "epoch": 0.20183114977995853, "grad_norm": 8.256561279296875, "learning_rate": 9.101952719519746e-06, "loss": 0.4462, "step": 3990 }, { "epoch": 0.20188173402802367, "grad_norm": 5.671535491943359, "learning_rate": 9.101493728962894e-06, "loss": 0.4357, "step": 3991 }, { "epoch": 0.20193231827608882, "grad_norm": 7.397793292999268, "learning_rate": 9.101034632720455e-06, "loss": 0.5685, "step": 3992 }, { "epoch": 0.201982902524154, "grad_norm": 4.63519811630249, "learning_rate": 9.10057543080426e-06, "loss": 0.3367, "step": 3993 }, { "epoch": 0.20203348677221913, "grad_norm": 20.731386184692383, "learning_rate": 9.10011612322614e-06, "loss": 0.4825, "step": 3994 }, { "epoch": 0.20208407102028428, "grad_norm": 7.79056978225708, "learning_rate": 9.099656709997931e-06, "loss": 0.4918, "step": 3995 }, { "epoch": 0.20213465526834942, "grad_norm": 11.83269214630127, "learning_rate": 9.099197191131471e-06, "loss": 0.4415, "step": 3996 }, { "epoch": 0.2021852395164146, "grad_norm": 9.889793395996094, "learning_rate": 9.098737566638602e-06, "loss": 0.4485, "step": 3997 }, { "epoch": 0.20223582376447974, "grad_norm": 6.434690952301025, "learning_rate": 9.098277836531164e-06, "loss": 0.3877, "step": 3998 }, { "epoch": 0.20228640801254488, "grad_norm": 8.412029266357422, "learning_rate": 9.097818000821006e-06, "loss": 0.4384, "step": 3999 }, { "epoch": 0.20233699226061005, "grad_norm": 14.809927940368652, "learning_rate": 9.097358059519976e-06, "loss": 0.6211, "step": 4000 }, { "epoch": 0.2023875765086752, "grad_norm": 10.122481346130371, "learning_rate": 9.096898012639926e-06, "loss": 0.3975, "step": 4001 }, { "epoch": 0.20243816075674034, "grad_norm": 18.95960807800293, "learning_rate": 9.096437860192708e-06, "loss": 0.4556, "step": 4002 }, { "epoch": 0.20248874500480551, "grad_norm": 8.905576705932617, "learning_rate": 9.095977602190184e-06, "loss": 0.4725, "step": 4003 }, { "epoch": 0.20253932925287066, "grad_norm": 6.724025249481201, "learning_rate": 9.095517238644209e-06, "loss": 0.3849, "step": 4004 }, { "epoch": 0.2025899135009358, "grad_norm": 6.4943695068359375, "learning_rate": 9.095056769566646e-06, "loss": 0.5188, "step": 4005 }, { "epoch": 0.20264049774900095, "grad_norm": 16.915620803833008, "learning_rate": 9.094596194969363e-06, "loss": 0.6772, "step": 4006 }, { "epoch": 0.20269108199706612, "grad_norm": 5.613072395324707, "learning_rate": 9.094135514864224e-06, "loss": 0.3964, "step": 4007 }, { "epoch": 0.20274166624513126, "grad_norm": 6.883028507232666, "learning_rate": 9.093674729263104e-06, "loss": 0.4755, "step": 4008 }, { "epoch": 0.2027922504931964, "grad_norm": 9.263545989990234, "learning_rate": 9.093213838177874e-06, "loss": 0.3609, "step": 4009 }, { "epoch": 0.20284283474126158, "grad_norm": 10.489068984985352, "learning_rate": 9.09275284162041e-06, "loss": 0.5131, "step": 4010 }, { "epoch": 0.20289341898932672, "grad_norm": 7.402095317840576, "learning_rate": 9.092291739602588e-06, "loss": 0.3445, "step": 4011 }, { "epoch": 0.20294400323739187, "grad_norm": 21.131492614746094, "learning_rate": 9.091830532136296e-06, "loss": 0.549, "step": 4012 }, { "epoch": 0.20299458748545704, "grad_norm": 10.4517822265625, "learning_rate": 9.091369219233412e-06, "loss": 0.4316, "step": 4013 }, { "epoch": 0.20304517173352218, "grad_norm": 11.789033889770508, "learning_rate": 9.090907800905827e-06, "loss": 0.5541, "step": 4014 }, { "epoch": 0.20309575598158733, "grad_norm": 13.269169807434082, "learning_rate": 9.09044627716543e-06, "loss": 0.5675, "step": 4015 }, { "epoch": 0.2031463402296525, "grad_norm": 9.835539817810059, "learning_rate": 9.08998464802411e-06, "loss": 0.4419, "step": 4016 }, { "epoch": 0.20319692447771764, "grad_norm": 13.386138916015625, "learning_rate": 9.089522913493766e-06, "loss": 0.6237, "step": 4017 }, { "epoch": 0.2032475087257828, "grad_norm": 9.924381256103516, "learning_rate": 9.089061073586296e-06, "loss": 0.6013, "step": 4018 }, { "epoch": 0.20329809297384793, "grad_norm": 8.84299373626709, "learning_rate": 9.088599128313596e-06, "loss": 0.461, "step": 4019 }, { "epoch": 0.2033486772219131, "grad_norm": 8.583649635314941, "learning_rate": 9.088137077687573e-06, "loss": 0.5322, "step": 4020 }, { "epoch": 0.20339926146997825, "grad_norm": 7.563568592071533, "learning_rate": 9.087674921720133e-06, "loss": 0.4699, "step": 4021 }, { "epoch": 0.2034498457180434, "grad_norm": 5.913738250732422, "learning_rate": 9.087212660423184e-06, "loss": 0.4376, "step": 4022 }, { "epoch": 0.20350042996610856, "grad_norm": 7.759945392608643, "learning_rate": 9.086750293808635e-06, "loss": 0.4292, "step": 4023 }, { "epoch": 0.2035510142141737, "grad_norm": 12.661818504333496, "learning_rate": 9.086287821888402e-06, "loss": 0.3608, "step": 4024 }, { "epoch": 0.20360159846223885, "grad_norm": 7.145476341247559, "learning_rate": 9.085825244674405e-06, "loss": 0.2782, "step": 4025 }, { "epoch": 0.20365218271030402, "grad_norm": 5.613845348358154, "learning_rate": 9.085362562178556e-06, "loss": 0.3589, "step": 4026 }, { "epoch": 0.20370276695836917, "grad_norm": 8.732438087463379, "learning_rate": 9.084899774412785e-06, "loss": 0.343, "step": 4027 }, { "epoch": 0.2037533512064343, "grad_norm": 4.937623023986816, "learning_rate": 9.084436881389011e-06, "loss": 0.4301, "step": 4028 }, { "epoch": 0.20380393545449946, "grad_norm": 6.933420658111572, "learning_rate": 9.083973883119166e-06, "loss": 0.396, "step": 4029 }, { "epoch": 0.20385451970256463, "grad_norm": 4.643924713134766, "learning_rate": 9.083510779615179e-06, "loss": 0.3826, "step": 4030 }, { "epoch": 0.20390510395062977, "grad_norm": 8.198810577392578, "learning_rate": 9.083047570888981e-06, "loss": 0.3954, "step": 4031 }, { "epoch": 0.20395568819869492, "grad_norm": 11.464033126831055, "learning_rate": 9.08258425695251e-06, "loss": 0.4488, "step": 4032 }, { "epoch": 0.2040062724467601, "grad_norm": 9.707904815673828, "learning_rate": 9.082120837817704e-06, "loss": 0.4528, "step": 4033 }, { "epoch": 0.20405685669482523, "grad_norm": 8.610208511352539, "learning_rate": 9.081657313496501e-06, "loss": 0.491, "step": 4034 }, { "epoch": 0.20410744094289038, "grad_norm": 9.77192211151123, "learning_rate": 9.081193684000851e-06, "loss": 0.4548, "step": 4035 }, { "epoch": 0.20415802519095555, "grad_norm": 11.197297096252441, "learning_rate": 9.080729949342696e-06, "loss": 0.6029, "step": 4036 }, { "epoch": 0.2042086094390207, "grad_norm": 13.896806716918945, "learning_rate": 9.080266109533988e-06, "loss": 0.5345, "step": 4037 }, { "epoch": 0.20425919368708584, "grad_norm": 12.785571098327637, "learning_rate": 9.079802164586678e-06, "loss": 0.4519, "step": 4038 }, { "epoch": 0.20430977793515098, "grad_norm": 10.293599128723145, "learning_rate": 9.07933811451272e-06, "loss": 0.534, "step": 4039 }, { "epoch": 0.20436036218321615, "grad_norm": 7.624035358428955, "learning_rate": 9.078873959324073e-06, "loss": 0.4148, "step": 4040 }, { "epoch": 0.2044109464312813, "grad_norm": 8.346275329589844, "learning_rate": 9.078409699032695e-06, "loss": 0.4213, "step": 4041 }, { "epoch": 0.20446153067934644, "grad_norm": 10.791561126708984, "learning_rate": 9.07794533365055e-06, "loss": 0.3846, "step": 4042 }, { "epoch": 0.20451211492741161, "grad_norm": 6.369240760803223, "learning_rate": 9.077480863189602e-06, "loss": 0.4367, "step": 4043 }, { "epoch": 0.20456269917547676, "grad_norm": 15.18474006652832, "learning_rate": 9.077016287661823e-06, "loss": 0.5115, "step": 4044 }, { "epoch": 0.2046132834235419, "grad_norm": 8.772899627685547, "learning_rate": 9.076551607079182e-06, "loss": 0.4876, "step": 4045 }, { "epoch": 0.20466386767160707, "grad_norm": 10.324912071228027, "learning_rate": 9.076086821453653e-06, "loss": 0.4814, "step": 4046 }, { "epoch": 0.20471445191967222, "grad_norm": 9.988123893737793, "learning_rate": 9.075621930797212e-06, "loss": 0.3515, "step": 4047 }, { "epoch": 0.20476503616773736, "grad_norm": 7.117979526519775, "learning_rate": 9.075156935121836e-06, "loss": 0.3008, "step": 4048 }, { "epoch": 0.2048156204158025, "grad_norm": 7.193786144256592, "learning_rate": 9.07469183443951e-06, "loss": 0.4007, "step": 4049 }, { "epoch": 0.20486620466386768, "grad_norm": 7.3401994705200195, "learning_rate": 9.07422662876222e-06, "loss": 0.4792, "step": 4050 }, { "epoch": 0.20491678891193282, "grad_norm": 8.873666763305664, "learning_rate": 9.073761318101947e-06, "loss": 0.4669, "step": 4051 }, { "epoch": 0.20496737315999797, "grad_norm": 19.94728660583496, "learning_rate": 9.073295902470688e-06, "loss": 0.6245, "step": 4052 }, { "epoch": 0.20501795740806314, "grad_norm": 12.247271537780762, "learning_rate": 9.072830381880428e-06, "loss": 0.5193, "step": 4053 }, { "epoch": 0.20506854165612828, "grad_norm": 9.999773979187012, "learning_rate": 9.072364756343169e-06, "loss": 0.2909, "step": 4054 }, { "epoch": 0.20511912590419343, "grad_norm": 8.08050537109375, "learning_rate": 9.071899025870908e-06, "loss": 0.4134, "step": 4055 }, { "epoch": 0.2051697101522586, "grad_norm": 11.055403709411621, "learning_rate": 9.071433190475642e-06, "loss": 0.4411, "step": 4056 }, { "epoch": 0.20522029440032374, "grad_norm": 10.211535453796387, "learning_rate": 9.070967250169378e-06, "loss": 0.3756, "step": 4057 }, { "epoch": 0.2052708786483889, "grad_norm": 9.744540214538574, "learning_rate": 9.07050120496412e-06, "loss": 0.3773, "step": 4058 }, { "epoch": 0.20532146289645403, "grad_norm": 7.145182132720947, "learning_rate": 9.070035054871879e-06, "loss": 0.3661, "step": 4059 }, { "epoch": 0.2053720471445192, "grad_norm": 14.580443382263184, "learning_rate": 9.069568799904664e-06, "loss": 0.5776, "step": 4060 }, { "epoch": 0.20542263139258435, "grad_norm": 10.783658027648926, "learning_rate": 9.069102440074492e-06, "loss": 0.3722, "step": 4061 }, { "epoch": 0.2054732156406495, "grad_norm": 9.206596374511719, "learning_rate": 9.068635975393378e-06, "loss": 0.3655, "step": 4062 }, { "epoch": 0.20552379988871466, "grad_norm": 7.08213996887207, "learning_rate": 9.068169405873342e-06, "loss": 0.5052, "step": 4063 }, { "epoch": 0.2055743841367798, "grad_norm": 8.930624961853027, "learning_rate": 9.067702731526408e-06, "loss": 0.5527, "step": 4064 }, { "epoch": 0.20562496838484495, "grad_norm": 9.576080322265625, "learning_rate": 9.067235952364599e-06, "loss": 0.5027, "step": 4065 }, { "epoch": 0.20567555263291012, "grad_norm": 9.134217262268066, "learning_rate": 9.066769068399943e-06, "loss": 0.4894, "step": 4066 }, { "epoch": 0.20572613688097527, "grad_norm": 8.23537826538086, "learning_rate": 9.06630207964447e-06, "loss": 0.4088, "step": 4067 }, { "epoch": 0.2057767211290404, "grad_norm": 8.37390422821045, "learning_rate": 9.065834986110216e-06, "loss": 0.5137, "step": 4068 }, { "epoch": 0.20582730537710556, "grad_norm": 22.907543182373047, "learning_rate": 9.065367787809214e-06, "loss": 0.43, "step": 4069 }, { "epoch": 0.20587788962517073, "grad_norm": 6.878835201263428, "learning_rate": 9.064900484753505e-06, "loss": 0.3307, "step": 4070 }, { "epoch": 0.20592847387323587, "grad_norm": 12.478408813476562, "learning_rate": 9.064433076955128e-06, "loss": 0.5165, "step": 4071 }, { "epoch": 0.20597905812130102, "grad_norm": 8.34234619140625, "learning_rate": 9.063965564426128e-06, "loss": 0.5102, "step": 4072 }, { "epoch": 0.2060296423693662, "grad_norm": 8.800150871276855, "learning_rate": 9.063497947178553e-06, "loss": 0.5154, "step": 4073 }, { "epoch": 0.20608022661743133, "grad_norm": 8.533613204956055, "learning_rate": 9.063030225224448e-06, "loss": 0.3852, "step": 4074 }, { "epoch": 0.20613081086549648, "grad_norm": 10.538766860961914, "learning_rate": 9.06256239857587e-06, "loss": 0.4605, "step": 4075 }, { "epoch": 0.20618139511356165, "grad_norm": 7.582081317901611, "learning_rate": 9.06209446724487e-06, "loss": 0.3984, "step": 4076 }, { "epoch": 0.2062319793616268, "grad_norm": 5.731342315673828, "learning_rate": 9.061626431243509e-06, "loss": 0.4764, "step": 4077 }, { "epoch": 0.20628256360969194, "grad_norm": 4.855332851409912, "learning_rate": 9.061158290583844e-06, "loss": 0.4118, "step": 4078 }, { "epoch": 0.20633314785775708, "grad_norm": 5.174513339996338, "learning_rate": 9.060690045277938e-06, "loss": 0.4152, "step": 4079 }, { "epoch": 0.20638373210582225, "grad_norm": 6.5175042152404785, "learning_rate": 9.06022169533786e-06, "loss": 0.4165, "step": 4080 }, { "epoch": 0.2064343163538874, "grad_norm": 8.864618301391602, "learning_rate": 9.059753240775675e-06, "loss": 0.4164, "step": 4081 }, { "epoch": 0.20648490060195254, "grad_norm": 19.945859909057617, "learning_rate": 9.059284681603455e-06, "loss": 0.5473, "step": 4082 }, { "epoch": 0.20653548485001771, "grad_norm": 6.322196006774902, "learning_rate": 9.058816017833275e-06, "loss": 0.4793, "step": 4083 }, { "epoch": 0.20658606909808286, "grad_norm": 8.817431449890137, "learning_rate": 9.058347249477208e-06, "loss": 0.343, "step": 4084 }, { "epoch": 0.206636653346148, "grad_norm": 10.235054016113281, "learning_rate": 9.057878376547336e-06, "loss": 0.4521, "step": 4085 }, { "epoch": 0.20668723759421317, "grad_norm": 17.12354850769043, "learning_rate": 9.05740939905574e-06, "loss": 0.6611, "step": 4086 }, { "epoch": 0.20673782184227832, "grad_norm": 7.947188854217529, "learning_rate": 9.056940317014502e-06, "loss": 0.501, "step": 4087 }, { "epoch": 0.20678840609034346, "grad_norm": 8.05476188659668, "learning_rate": 9.056471130435713e-06, "loss": 0.4508, "step": 4088 }, { "epoch": 0.2068389903384086, "grad_norm": 13.622574806213379, "learning_rate": 9.056001839331462e-06, "loss": 0.3697, "step": 4089 }, { "epoch": 0.20688957458647378, "grad_norm": 14.867514610290527, "learning_rate": 9.055532443713836e-06, "loss": 0.5344, "step": 4090 }, { "epoch": 0.20694015883453892, "grad_norm": 8.49520492553711, "learning_rate": 9.055062943594939e-06, "loss": 0.3727, "step": 4091 }, { "epoch": 0.20699074308260407, "grad_norm": 11.385383605957031, "learning_rate": 9.054593338986864e-06, "loss": 0.5219, "step": 4092 }, { "epoch": 0.20704132733066924, "grad_norm": 12.648709297180176, "learning_rate": 9.05412362990171e-06, "loss": 0.4149, "step": 4093 }, { "epoch": 0.20709191157873438, "grad_norm": 6.068985462188721, "learning_rate": 9.053653816351583e-06, "loss": 0.4302, "step": 4094 }, { "epoch": 0.20714249582679953, "grad_norm": 12.341389656066895, "learning_rate": 9.05318389834859e-06, "loss": 0.4099, "step": 4095 }, { "epoch": 0.2071930800748647, "grad_norm": 11.564019203186035, "learning_rate": 9.052713875904837e-06, "loss": 0.5559, "step": 4096 }, { "epoch": 0.20724366432292984, "grad_norm": 10.769416809082031, "learning_rate": 9.052243749032437e-06, "loss": 0.4629, "step": 4097 }, { "epoch": 0.207294248570995, "grad_norm": 5.90496826171875, "learning_rate": 9.0517735177435e-06, "loss": 0.4728, "step": 4098 }, { "epoch": 0.20734483281906013, "grad_norm": 8.359103202819824, "learning_rate": 9.051303182050151e-06, "loss": 0.48, "step": 4099 }, { "epoch": 0.2073954170671253, "grad_norm": 5.613210201263428, "learning_rate": 9.050832741964501e-06, "loss": 0.4157, "step": 4100 }, { "epoch": 0.20744600131519045, "grad_norm": 5.763238430023193, "learning_rate": 9.050362197498678e-06, "loss": 0.3119, "step": 4101 }, { "epoch": 0.2074965855632556, "grad_norm": 10.74956226348877, "learning_rate": 9.049891548664804e-06, "loss": 0.3607, "step": 4102 }, { "epoch": 0.20754716981132076, "grad_norm": 8.001729011535645, "learning_rate": 9.049420795475005e-06, "loss": 0.351, "step": 4103 }, { "epoch": 0.2075977540593859, "grad_norm": 8.981711387634277, "learning_rate": 9.048949937941415e-06, "loss": 0.5099, "step": 4104 }, { "epoch": 0.20764833830745105, "grad_norm": 13.815671920776367, "learning_rate": 9.048478976076165e-06, "loss": 0.5207, "step": 4105 }, { "epoch": 0.20769892255551622, "grad_norm": 9.132412910461426, "learning_rate": 9.048007909891389e-06, "loss": 0.452, "step": 4106 }, { "epoch": 0.20774950680358137, "grad_norm": 10.453986167907715, "learning_rate": 9.047536739399227e-06, "loss": 0.5108, "step": 4107 }, { "epoch": 0.2078000910516465, "grad_norm": 11.192808151245117, "learning_rate": 9.04706546461182e-06, "loss": 0.5255, "step": 4108 }, { "epoch": 0.20785067529971166, "grad_norm": 11.134544372558594, "learning_rate": 9.046594085541311e-06, "loss": 0.5158, "step": 4109 }, { "epoch": 0.20790125954777683, "grad_norm": 7.605023384094238, "learning_rate": 9.046122602199847e-06, "loss": 0.4775, "step": 4110 }, { "epoch": 0.20795184379584197, "grad_norm": 26.548812866210938, "learning_rate": 9.045651014599577e-06, "loss": 0.5369, "step": 4111 }, { "epoch": 0.20800242804390712, "grad_norm": 8.582932472229004, "learning_rate": 9.045179322752652e-06, "loss": 0.4912, "step": 4112 }, { "epoch": 0.2080530122919723, "grad_norm": 11.02960205078125, "learning_rate": 9.044707526671227e-06, "loss": 0.4403, "step": 4113 }, { "epoch": 0.20810359654003743, "grad_norm": 9.130041122436523, "learning_rate": 9.044235626367457e-06, "loss": 0.2909, "step": 4114 }, { "epoch": 0.20815418078810258, "grad_norm": 13.23021411895752, "learning_rate": 9.043763621853506e-06, "loss": 0.524, "step": 4115 }, { "epoch": 0.20820476503616775, "grad_norm": 15.203640937805176, "learning_rate": 9.043291513141532e-06, "loss": 0.3755, "step": 4116 }, { "epoch": 0.2082553492842329, "grad_norm": 6.797627925872803, "learning_rate": 9.042819300243701e-06, "loss": 0.4886, "step": 4117 }, { "epoch": 0.20830593353229804, "grad_norm": 15.51530647277832, "learning_rate": 9.042346983172183e-06, "loss": 0.6457, "step": 4118 }, { "epoch": 0.20835651778036318, "grad_norm": 9.445301055908203, "learning_rate": 9.041874561939146e-06, "loss": 0.3979, "step": 4119 }, { "epoch": 0.20840710202842835, "grad_norm": 8.334548950195312, "learning_rate": 9.041402036556766e-06, "loss": 0.3964, "step": 4120 }, { "epoch": 0.2084576862764935, "grad_norm": 12.85669994354248, "learning_rate": 9.040929407037215e-06, "loss": 0.4974, "step": 4121 }, { "epoch": 0.20850827052455864, "grad_norm": 13.537421226501465, "learning_rate": 9.040456673392674e-06, "loss": 0.4339, "step": 4122 }, { "epoch": 0.20855885477262381, "grad_norm": 6.711885452270508, "learning_rate": 9.039983835635326e-06, "loss": 0.4671, "step": 4123 }, { "epoch": 0.20860943902068896, "grad_norm": 16.73857307434082, "learning_rate": 9.039510893777351e-06, "loss": 0.5517, "step": 4124 }, { "epoch": 0.2086600232687541, "grad_norm": 5.476842880249023, "learning_rate": 9.039037847830938e-06, "loss": 0.4108, "step": 4125 }, { "epoch": 0.20871060751681927, "grad_norm": 8.132807731628418, "learning_rate": 9.038564697808275e-06, "loss": 0.5126, "step": 4126 }, { "epoch": 0.20876119176488442, "grad_norm": 11.665657997131348, "learning_rate": 9.038091443721555e-06, "loss": 0.3759, "step": 4127 }, { "epoch": 0.20881177601294956, "grad_norm": 11.43483829498291, "learning_rate": 9.037618085582971e-06, "loss": 0.5005, "step": 4128 }, { "epoch": 0.2088623602610147, "grad_norm": 10.823204040527344, "learning_rate": 9.037144623404723e-06, "loss": 0.4184, "step": 4129 }, { "epoch": 0.20891294450907988, "grad_norm": 10.571650505065918, "learning_rate": 9.036671057199008e-06, "loss": 0.5588, "step": 4130 }, { "epoch": 0.20896352875714502, "grad_norm": 12.636672019958496, "learning_rate": 9.036197386978032e-06, "loss": 0.5879, "step": 4131 }, { "epoch": 0.20901411300521017, "grad_norm": 7.287863731384277, "learning_rate": 9.035723612753997e-06, "loss": 0.3667, "step": 4132 }, { "epoch": 0.20906469725327534, "grad_norm": 6.603115081787109, "learning_rate": 9.035249734539112e-06, "loss": 0.4479, "step": 4133 }, { "epoch": 0.20911528150134048, "grad_norm": 6.864531517028809, "learning_rate": 9.034775752345587e-06, "loss": 0.4894, "step": 4134 }, { "epoch": 0.20916586574940563, "grad_norm": 5.827361583709717, "learning_rate": 9.03430166618564e-06, "loss": 0.4743, "step": 4135 }, { "epoch": 0.2092164499974708, "grad_norm": 9.738542556762695, "learning_rate": 9.03382747607148e-06, "loss": 0.4987, "step": 4136 }, { "epoch": 0.20926703424553594, "grad_norm": 8.35457706451416, "learning_rate": 9.033353182015333e-06, "loss": 0.5145, "step": 4137 }, { "epoch": 0.2093176184936011, "grad_norm": 7.686325550079346, "learning_rate": 9.032878784029413e-06, "loss": 0.4147, "step": 4138 }, { "epoch": 0.20936820274166623, "grad_norm": 6.444817543029785, "learning_rate": 9.032404282125949e-06, "loss": 0.4267, "step": 4139 }, { "epoch": 0.2094187869897314, "grad_norm": 14.984153747558594, "learning_rate": 9.031929676317167e-06, "loss": 0.5974, "step": 4140 }, { "epoch": 0.20946937123779655, "grad_norm": 8.47382640838623, "learning_rate": 9.031454966615296e-06, "loss": 0.4255, "step": 4141 }, { "epoch": 0.2095199554858617, "grad_norm": 16.414094924926758, "learning_rate": 9.030980153032567e-06, "loss": 0.4012, "step": 4142 }, { "epoch": 0.20957053973392686, "grad_norm": 5.997035026550293, "learning_rate": 9.030505235581218e-06, "loss": 0.503, "step": 4143 }, { "epoch": 0.209621123981992, "grad_norm": 13.788922309875488, "learning_rate": 9.030030214273481e-06, "loss": 0.3345, "step": 4144 }, { "epoch": 0.20967170823005715, "grad_norm": 7.491602420806885, "learning_rate": 9.029555089121603e-06, "loss": 0.4896, "step": 4145 }, { "epoch": 0.20972229247812232, "grad_norm": 11.633307456970215, "learning_rate": 9.029079860137821e-06, "loss": 0.4629, "step": 4146 }, { "epoch": 0.20977287672618747, "grad_norm": 6.767633438110352, "learning_rate": 9.028604527334384e-06, "loss": 0.4435, "step": 4147 }, { "epoch": 0.2098234609742526, "grad_norm": 6.561529159545898, "learning_rate": 9.02812909072354e-06, "loss": 0.4617, "step": 4148 }, { "epoch": 0.20987404522231776, "grad_norm": 5.052733421325684, "learning_rate": 9.027653550317536e-06, "loss": 0.3854, "step": 4149 }, { "epoch": 0.20992462947038293, "grad_norm": 14.353282928466797, "learning_rate": 9.02717790612863e-06, "loss": 0.4669, "step": 4150 }, { "epoch": 0.20997521371844807, "grad_norm": 17.995380401611328, "learning_rate": 9.026702158169078e-06, "loss": 0.5531, "step": 4151 }, { "epoch": 0.21002579796651322, "grad_norm": 16.604106903076172, "learning_rate": 9.026226306451136e-06, "loss": 0.6052, "step": 4152 }, { "epoch": 0.2100763822145784, "grad_norm": 6.232485294342041, "learning_rate": 9.025750350987066e-06, "loss": 0.4456, "step": 4153 }, { "epoch": 0.21012696646264353, "grad_norm": 10.283802032470703, "learning_rate": 9.025274291789136e-06, "loss": 0.4867, "step": 4154 }, { "epoch": 0.21017755071070868, "grad_norm": 10.326654434204102, "learning_rate": 9.024798128869607e-06, "loss": 0.4935, "step": 4155 }, { "epoch": 0.21022813495877385, "grad_norm": 5.683557033538818, "learning_rate": 9.024321862240755e-06, "loss": 0.4234, "step": 4156 }, { "epoch": 0.210278719206839, "grad_norm": 5.6138105392456055, "learning_rate": 9.023845491914846e-06, "loss": 0.4111, "step": 4157 }, { "epoch": 0.21032930345490414, "grad_norm": 5.244898796081543, "learning_rate": 9.02336901790416e-06, "loss": 0.4408, "step": 4158 }, { "epoch": 0.21037988770296928, "grad_norm": 6.689334392547607, "learning_rate": 9.022892440220972e-06, "loss": 0.4532, "step": 4159 }, { "epoch": 0.21043047195103445, "grad_norm": 8.722267150878906, "learning_rate": 9.022415758877561e-06, "loss": 0.4486, "step": 4160 }, { "epoch": 0.2104810561990996, "grad_norm": 7.2273783683776855, "learning_rate": 9.021938973886212e-06, "loss": 0.4283, "step": 4161 }, { "epoch": 0.21053164044716474, "grad_norm": 9.495361328125, "learning_rate": 9.021462085259212e-06, "loss": 0.4802, "step": 4162 }, { "epoch": 0.2105822246952299, "grad_norm": 8.827736854553223, "learning_rate": 9.020985093008847e-06, "loss": 0.3915, "step": 4163 }, { "epoch": 0.21063280894329506, "grad_norm": 11.833066940307617, "learning_rate": 9.020507997147408e-06, "loss": 0.5524, "step": 4164 }, { "epoch": 0.2106833931913602, "grad_norm": 7.547205448150635, "learning_rate": 9.02003079768719e-06, "loss": 0.4091, "step": 4165 }, { "epoch": 0.21073397743942537, "grad_norm": 9.428545951843262, "learning_rate": 9.019553494640487e-06, "loss": 0.4777, "step": 4166 }, { "epoch": 0.21078456168749052, "grad_norm": 9.024853706359863, "learning_rate": 9.0190760880196e-06, "loss": 0.5586, "step": 4167 }, { "epoch": 0.21083514593555566, "grad_norm": 13.069509506225586, "learning_rate": 9.01859857783683e-06, "loss": 0.5329, "step": 4168 }, { "epoch": 0.21088573018362083, "grad_norm": 7.334753513336182, "learning_rate": 9.01812096410448e-06, "loss": 0.4926, "step": 4169 }, { "epoch": 0.21093631443168598, "grad_norm": 10.775872230529785, "learning_rate": 9.01764324683486e-06, "loss": 0.4175, "step": 4170 }, { "epoch": 0.21098689867975112, "grad_norm": 7.151773452758789, "learning_rate": 9.017165426040276e-06, "loss": 0.4254, "step": 4171 }, { "epoch": 0.21103748292781627, "grad_norm": 10.019336700439453, "learning_rate": 9.016687501733043e-06, "loss": 0.4338, "step": 4172 }, { "epoch": 0.21108806717588144, "grad_norm": 8.619831085205078, "learning_rate": 9.016209473925474e-06, "loss": 0.4006, "step": 4173 }, { "epoch": 0.21113865142394658, "grad_norm": 8.38760757446289, "learning_rate": 9.01573134262989e-06, "loss": 0.4269, "step": 4174 }, { "epoch": 0.21118923567201173, "grad_norm": 14.276132583618164, "learning_rate": 9.015253107858605e-06, "loss": 0.5114, "step": 4175 }, { "epoch": 0.2112398199200769, "grad_norm": 11.706526756286621, "learning_rate": 9.01477476962395e-06, "loss": 0.515, "step": 4176 }, { "epoch": 0.21129040416814204, "grad_norm": 10.148431777954102, "learning_rate": 9.014296327938245e-06, "loss": 0.42, "step": 4177 }, { "epoch": 0.2113409884162072, "grad_norm": 10.042110443115234, "learning_rate": 9.013817782813818e-06, "loss": 0.3925, "step": 4178 }, { "epoch": 0.21139157266427236, "grad_norm": 11.027179718017578, "learning_rate": 9.013339134263003e-06, "loss": 0.5643, "step": 4179 }, { "epoch": 0.2114421569123375, "grad_norm": 7.1163177490234375, "learning_rate": 9.012860382298133e-06, "loss": 0.4521, "step": 4180 }, { "epoch": 0.21149274116040265, "grad_norm": 5.54844331741333, "learning_rate": 9.012381526931544e-06, "loss": 0.3537, "step": 4181 }, { "epoch": 0.2115433254084678, "grad_norm": 17.19648551940918, "learning_rate": 9.011902568175572e-06, "loss": 0.4732, "step": 4182 }, { "epoch": 0.21159390965653296, "grad_norm": 6.62727689743042, "learning_rate": 9.011423506042564e-06, "loss": 0.4117, "step": 4183 }, { "epoch": 0.2116444939045981, "grad_norm": 4.718914031982422, "learning_rate": 9.01094434054486e-06, "loss": 0.43, "step": 4184 }, { "epoch": 0.21169507815266325, "grad_norm": 6.8674516677856445, "learning_rate": 9.010465071694809e-06, "loss": 0.5146, "step": 4185 }, { "epoch": 0.21174566240072842, "grad_norm": 10.779648780822754, "learning_rate": 9.009985699504759e-06, "loss": 0.5866, "step": 4186 }, { "epoch": 0.21179624664879357, "grad_norm": 7.185124397277832, "learning_rate": 9.009506223987064e-06, "loss": 0.4508, "step": 4187 }, { "epoch": 0.2118468308968587, "grad_norm": 8.623733520507812, "learning_rate": 9.009026645154078e-06, "loss": 0.3644, "step": 4188 }, { "epoch": 0.21189741514492388, "grad_norm": 13.468622207641602, "learning_rate": 9.008546963018159e-06, "loss": 0.3502, "step": 4189 }, { "epoch": 0.21194799939298903, "grad_norm": 12.826045036315918, "learning_rate": 9.008067177591666e-06, "loss": 0.534, "step": 4190 }, { "epoch": 0.21199858364105417, "grad_norm": 28.759523391723633, "learning_rate": 9.007587288886964e-06, "loss": 0.6594, "step": 4191 }, { "epoch": 0.21204916788911932, "grad_norm": 9.19704818725586, "learning_rate": 9.007107296916417e-06, "loss": 0.4174, "step": 4192 }, { "epoch": 0.2120997521371845, "grad_norm": 7.905323028564453, "learning_rate": 9.006627201692395e-06, "loss": 0.4645, "step": 4193 }, { "epoch": 0.21215033638524963, "grad_norm": 8.565245628356934, "learning_rate": 9.006147003227265e-06, "loss": 0.3886, "step": 4194 }, { "epoch": 0.21220092063331478, "grad_norm": 8.117122650146484, "learning_rate": 9.005666701533407e-06, "loss": 0.3754, "step": 4195 }, { "epoch": 0.21225150488137995, "grad_norm": 6.568051338195801, "learning_rate": 9.00518629662319e-06, "loss": 0.2947, "step": 4196 }, { "epoch": 0.2123020891294451, "grad_norm": 10.261815071105957, "learning_rate": 9.004705788508997e-06, "loss": 0.5269, "step": 4197 }, { "epoch": 0.21235267337751024, "grad_norm": 13.44098949432373, "learning_rate": 9.00422517720321e-06, "loss": 0.6239, "step": 4198 }, { "epoch": 0.2124032576255754, "grad_norm": 9.096574783325195, "learning_rate": 9.00374446271821e-06, "loss": 0.4104, "step": 4199 }, { "epoch": 0.21245384187364055, "grad_norm": 8.444701194763184, "learning_rate": 9.003263645066389e-06, "loss": 0.4448, "step": 4200 }, { "epoch": 0.2125044261217057, "grad_norm": 7.828876972198486, "learning_rate": 9.002782724260131e-06, "loss": 0.4443, "step": 4201 }, { "epoch": 0.21255501036977084, "grad_norm": 8.090555191040039, "learning_rate": 9.002301700311831e-06, "loss": 0.3719, "step": 4202 }, { "epoch": 0.212605594617836, "grad_norm": 6.651251792907715, "learning_rate": 9.001820573233883e-06, "loss": 0.4698, "step": 4203 }, { "epoch": 0.21265617886590116, "grad_norm": 8.982278823852539, "learning_rate": 9.001339343038684e-06, "loss": 0.4328, "step": 4204 }, { "epoch": 0.2127067631139663, "grad_norm": 14.021044731140137, "learning_rate": 9.000858009738637e-06, "loss": 0.5109, "step": 4205 }, { "epoch": 0.21275734736203147, "grad_norm": 19.732555389404297, "learning_rate": 9.000376573346142e-06, "loss": 0.5268, "step": 4206 }, { "epoch": 0.21280793161009662, "grad_norm": 17.334972381591797, "learning_rate": 8.999895033873604e-06, "loss": 0.3916, "step": 4207 }, { "epoch": 0.21285851585816176, "grad_norm": 16.602392196655273, "learning_rate": 8.999413391333434e-06, "loss": 0.5513, "step": 4208 }, { "epoch": 0.21290910010622693, "grad_norm": 8.562079429626465, "learning_rate": 8.99893164573804e-06, "loss": 0.3222, "step": 4209 }, { "epoch": 0.21295968435429208, "grad_norm": 8.662273406982422, "learning_rate": 8.998449797099837e-06, "loss": 0.3639, "step": 4210 }, { "epoch": 0.21301026860235722, "grad_norm": 7.831334590911865, "learning_rate": 8.99796784543124e-06, "loss": 0.4539, "step": 4211 }, { "epoch": 0.21306085285042237, "grad_norm": 12.579673767089844, "learning_rate": 8.99748579074467e-06, "loss": 0.6309, "step": 4212 }, { "epoch": 0.21311143709848754, "grad_norm": 7.332551002502441, "learning_rate": 8.997003633052545e-06, "loss": 0.4954, "step": 4213 }, { "epoch": 0.21316202134655268, "grad_norm": 10.24221134185791, "learning_rate": 8.996521372367291e-06, "loss": 0.4996, "step": 4214 }, { "epoch": 0.21321260559461783, "grad_norm": 15.458159446716309, "learning_rate": 8.996039008701336e-06, "loss": 0.7445, "step": 4215 }, { "epoch": 0.213263189842683, "grad_norm": 13.10577392578125, "learning_rate": 8.995556542067105e-06, "loss": 0.612, "step": 4216 }, { "epoch": 0.21331377409074814, "grad_norm": 14.852594375610352, "learning_rate": 8.995073972477035e-06, "loss": 0.6109, "step": 4217 }, { "epoch": 0.2133643583388133, "grad_norm": 5.577162742614746, "learning_rate": 8.994591299943558e-06, "loss": 0.4077, "step": 4218 }, { "epoch": 0.21341494258687846, "grad_norm": 6.133635997772217, "learning_rate": 8.994108524479112e-06, "loss": 0.3912, "step": 4219 }, { "epoch": 0.2134655268349436, "grad_norm": 8.254143714904785, "learning_rate": 8.993625646096136e-06, "loss": 0.5478, "step": 4220 }, { "epoch": 0.21351611108300875, "grad_norm": 9.123884201049805, "learning_rate": 8.993142664807073e-06, "loss": 0.4938, "step": 4221 }, { "epoch": 0.2135666953310739, "grad_norm": 5.668403625488281, "learning_rate": 8.99265958062437e-06, "loss": 0.4746, "step": 4222 }, { "epoch": 0.21361727957913906, "grad_norm": 5.515415668487549, "learning_rate": 8.992176393560472e-06, "loss": 0.4804, "step": 4223 }, { "epoch": 0.2136678638272042, "grad_norm": 8.731595993041992, "learning_rate": 8.991693103627832e-06, "loss": 0.3656, "step": 4224 }, { "epoch": 0.21371844807526935, "grad_norm": 6.666874408721924, "learning_rate": 8.991209710838902e-06, "loss": 0.4705, "step": 4225 }, { "epoch": 0.21376903232333452, "grad_norm": 8.516875267028809, "learning_rate": 8.990726215206138e-06, "loss": 0.422, "step": 4226 }, { "epoch": 0.21381961657139967, "grad_norm": 10.11315631866455, "learning_rate": 8.990242616741998e-06, "loss": 0.3854, "step": 4227 }, { "epoch": 0.2138702008194648, "grad_norm": 5.617693901062012, "learning_rate": 8.989758915458945e-06, "loss": 0.4068, "step": 4228 }, { "epoch": 0.21392078506752998, "grad_norm": 12.029091835021973, "learning_rate": 8.98927511136944e-06, "loss": 0.5594, "step": 4229 }, { "epoch": 0.21397136931559513, "grad_norm": 5.966885089874268, "learning_rate": 8.988791204485954e-06, "loss": 0.415, "step": 4230 }, { "epoch": 0.21402195356366027, "grad_norm": 5.006441593170166, "learning_rate": 8.988307194820952e-06, "loss": 0.4132, "step": 4231 }, { "epoch": 0.21407253781172542, "grad_norm": 7.595274925231934, "learning_rate": 8.987823082386904e-06, "loss": 0.4218, "step": 4232 }, { "epoch": 0.2141231220597906, "grad_norm": 9.666437149047852, "learning_rate": 8.98733886719629e-06, "loss": 0.5088, "step": 4233 }, { "epoch": 0.21417370630785573, "grad_norm": 13.948370933532715, "learning_rate": 8.986854549261584e-06, "loss": 0.5868, "step": 4234 }, { "epoch": 0.21422429055592088, "grad_norm": 13.201823234558105, "learning_rate": 8.986370128595267e-06, "loss": 0.5731, "step": 4235 }, { "epoch": 0.21427487480398605, "grad_norm": 9.417500495910645, "learning_rate": 8.985885605209819e-06, "loss": 0.408, "step": 4236 }, { "epoch": 0.2143254590520512, "grad_norm": 7.50322961807251, "learning_rate": 8.985400979117727e-06, "loss": 0.3661, "step": 4237 }, { "epoch": 0.21437604330011634, "grad_norm": 12.27418041229248, "learning_rate": 8.984916250331478e-06, "loss": 0.4071, "step": 4238 }, { "epoch": 0.2144266275481815, "grad_norm": 9.04523754119873, "learning_rate": 8.984431418863564e-06, "loss": 0.3652, "step": 4239 }, { "epoch": 0.21447721179624665, "grad_norm": 7.038347244262695, "learning_rate": 8.983946484726474e-06, "loss": 0.3908, "step": 4240 }, { "epoch": 0.2145277960443118, "grad_norm": 6.577509880065918, "learning_rate": 8.983461447932706e-06, "loss": 0.4212, "step": 4241 }, { "epoch": 0.21457838029237694, "grad_norm": 9.93826961517334, "learning_rate": 8.982976308494759e-06, "loss": 0.4744, "step": 4242 }, { "epoch": 0.2146289645404421, "grad_norm": 7.9664716720581055, "learning_rate": 8.982491066425135e-06, "loss": 0.4623, "step": 4243 }, { "epoch": 0.21467954878850726, "grad_norm": 6.144927501678467, "learning_rate": 8.982005721736334e-06, "loss": 0.3978, "step": 4244 }, { "epoch": 0.2147301330365724, "grad_norm": 7.852479457855225, "learning_rate": 8.981520274440862e-06, "loss": 0.4216, "step": 4245 }, { "epoch": 0.21478071728463757, "grad_norm": 12.982328414916992, "learning_rate": 8.98103472455123e-06, "loss": 0.5245, "step": 4246 }, { "epoch": 0.21483130153270272, "grad_norm": 10.376016616821289, "learning_rate": 8.98054907207995e-06, "loss": 0.4519, "step": 4247 }, { "epoch": 0.21488188578076786, "grad_norm": 9.679394721984863, "learning_rate": 8.980063317039536e-06, "loss": 0.5007, "step": 4248 }, { "epoch": 0.21493247002883303, "grad_norm": 12.549959182739258, "learning_rate": 8.979577459442502e-06, "loss": 0.5136, "step": 4249 }, { "epoch": 0.21498305427689818, "grad_norm": 7.832541465759277, "learning_rate": 8.979091499301369e-06, "loss": 0.3772, "step": 4250 }, { "epoch": 0.21503363852496332, "grad_norm": 10.31074333190918, "learning_rate": 8.97860543662866e-06, "loss": 0.5813, "step": 4251 }, { "epoch": 0.21508422277302847, "grad_norm": 10.62185287475586, "learning_rate": 8.978119271436899e-06, "loss": 0.3996, "step": 4252 }, { "epoch": 0.21513480702109364, "grad_norm": 8.768609046936035, "learning_rate": 8.977633003738613e-06, "loss": 0.4363, "step": 4253 }, { "epoch": 0.21518539126915878, "grad_norm": 7.430494785308838, "learning_rate": 8.977146633546331e-06, "loss": 0.449, "step": 4254 }, { "epoch": 0.21523597551722393, "grad_norm": 10.574590682983398, "learning_rate": 8.976660160872587e-06, "loss": 0.4711, "step": 4255 }, { "epoch": 0.2152865597652891, "grad_norm": 10.758245468139648, "learning_rate": 8.976173585729917e-06, "loss": 0.4526, "step": 4256 }, { "epoch": 0.21533714401335424, "grad_norm": 8.45864200592041, "learning_rate": 8.975686908130858e-06, "loss": 0.4636, "step": 4257 }, { "epoch": 0.2153877282614194, "grad_norm": 5.541993141174316, "learning_rate": 8.975200128087948e-06, "loss": 0.3516, "step": 4258 }, { "epoch": 0.21543831250948456, "grad_norm": 10.063243865966797, "learning_rate": 8.974713245613735e-06, "loss": 0.3502, "step": 4259 }, { "epoch": 0.2154888967575497, "grad_norm": 11.380456924438477, "learning_rate": 8.97422626072076e-06, "loss": 0.4192, "step": 4260 }, { "epoch": 0.21553948100561485, "grad_norm": 9.79228687286377, "learning_rate": 8.973739173421575e-06, "loss": 0.6025, "step": 4261 }, { "epoch": 0.21559006525368, "grad_norm": 13.302950859069824, "learning_rate": 8.973251983728729e-06, "loss": 0.4299, "step": 4262 }, { "epoch": 0.21564064950174516, "grad_norm": 15.171270370483398, "learning_rate": 8.972764691654777e-06, "loss": 0.6176, "step": 4263 }, { "epoch": 0.2156912337498103, "grad_norm": 5.30990743637085, "learning_rate": 8.972277297212275e-06, "loss": 0.3792, "step": 4264 }, { "epoch": 0.21574181799787545, "grad_norm": 9.197519302368164, "learning_rate": 8.97178980041378e-06, "loss": 0.4959, "step": 4265 }, { "epoch": 0.21579240224594062, "grad_norm": 9.690413475036621, "learning_rate": 8.971302201271856e-06, "loss": 0.5505, "step": 4266 }, { "epoch": 0.21584298649400577, "grad_norm": 5.483847618103027, "learning_rate": 8.970814499799067e-06, "loss": 0.3708, "step": 4267 }, { "epoch": 0.2158935707420709, "grad_norm": 15.436423301696777, "learning_rate": 8.97032669600798e-06, "loss": 0.5396, "step": 4268 }, { "epoch": 0.21594415499013608, "grad_norm": 6.794244766235352, "learning_rate": 8.969838789911164e-06, "loss": 0.3071, "step": 4269 }, { "epoch": 0.21599473923820123, "grad_norm": 11.676375389099121, "learning_rate": 8.969350781521189e-06, "loss": 0.5146, "step": 4270 }, { "epoch": 0.21604532348626637, "grad_norm": 6.104249477386475, "learning_rate": 8.968862670850634e-06, "loss": 0.4465, "step": 4271 }, { "epoch": 0.21609590773433152, "grad_norm": 5.14879035949707, "learning_rate": 8.968374457912075e-06, "loss": 0.4027, "step": 4272 }, { "epoch": 0.2161464919823967, "grad_norm": 12.392626762390137, "learning_rate": 8.96788614271809e-06, "loss": 0.5064, "step": 4273 }, { "epoch": 0.21619707623046183, "grad_norm": 9.385167121887207, "learning_rate": 8.967397725281263e-06, "loss": 0.4907, "step": 4274 }, { "epoch": 0.21624766047852698, "grad_norm": 10.735107421875, "learning_rate": 8.966909205614178e-06, "loss": 0.4452, "step": 4275 }, { "epoch": 0.21629824472659215, "grad_norm": 12.423799514770508, "learning_rate": 8.966420583729428e-06, "loss": 0.4822, "step": 4276 }, { "epoch": 0.2163488289746573, "grad_norm": 7.9268975257873535, "learning_rate": 8.965931859639598e-06, "loss": 0.4439, "step": 4277 }, { "epoch": 0.21639941322272244, "grad_norm": 10.247221946716309, "learning_rate": 8.965443033357284e-06, "loss": 0.5121, "step": 4278 }, { "epoch": 0.2164499974707876, "grad_norm": 12.71044635772705, "learning_rate": 8.96495410489508e-06, "loss": 0.5485, "step": 4279 }, { "epoch": 0.21650058171885275, "grad_norm": 6.77185583114624, "learning_rate": 8.964465074265587e-06, "loss": 0.4417, "step": 4280 }, { "epoch": 0.2165511659669179, "grad_norm": 9.948813438415527, "learning_rate": 8.963975941481404e-06, "loss": 0.4802, "step": 4281 }, { "epoch": 0.21660175021498304, "grad_norm": 11.580443382263184, "learning_rate": 8.963486706555135e-06, "loss": 0.4551, "step": 4282 }, { "epoch": 0.2166523344630482, "grad_norm": 8.375235557556152, "learning_rate": 8.962997369499389e-06, "loss": 0.4503, "step": 4283 }, { "epoch": 0.21670291871111336, "grad_norm": 5.756341457366943, "learning_rate": 8.96250793032677e-06, "loss": 0.3283, "step": 4284 }, { "epoch": 0.2167535029591785, "grad_norm": 5.476240158081055, "learning_rate": 8.962018389049896e-06, "loss": 0.4951, "step": 4285 }, { "epoch": 0.21680408720724367, "grad_norm": 13.31181812286377, "learning_rate": 8.961528745681376e-06, "loss": 0.5621, "step": 4286 }, { "epoch": 0.21685467145530882, "grad_norm": 6.595071315765381, "learning_rate": 8.96103900023383e-06, "loss": 0.4376, "step": 4287 }, { "epoch": 0.21690525570337396, "grad_norm": 7.573383331298828, "learning_rate": 8.960549152719875e-06, "loss": 0.4327, "step": 4288 }, { "epoch": 0.21695583995143913, "grad_norm": 8.951606750488281, "learning_rate": 8.960059203152137e-06, "loss": 0.4798, "step": 4289 }, { "epoch": 0.21700642419950428, "grad_norm": 12.888080596923828, "learning_rate": 8.959569151543237e-06, "loss": 0.3608, "step": 4290 }, { "epoch": 0.21705700844756942, "grad_norm": 12.74803638458252, "learning_rate": 8.959078997905804e-06, "loss": 0.4729, "step": 4291 }, { "epoch": 0.21710759269563457, "grad_norm": 4.5838799476623535, "learning_rate": 8.958588742252468e-06, "loss": 0.365, "step": 4292 }, { "epoch": 0.21715817694369974, "grad_norm": 9.365005493164062, "learning_rate": 8.958098384595861e-06, "loss": 0.4519, "step": 4293 }, { "epoch": 0.21720876119176488, "grad_norm": 10.810423851013184, "learning_rate": 8.95760792494862e-06, "loss": 0.4065, "step": 4294 }, { "epoch": 0.21725934543983003, "grad_norm": 15.44809627532959, "learning_rate": 8.95711736332338e-06, "loss": 0.4157, "step": 4295 }, { "epoch": 0.2173099296878952, "grad_norm": 9.292004585266113, "learning_rate": 8.956626699732787e-06, "loss": 0.569, "step": 4296 }, { "epoch": 0.21736051393596034, "grad_norm": 7.551836013793945, "learning_rate": 8.956135934189478e-06, "loss": 0.5145, "step": 4297 }, { "epoch": 0.2174110981840255, "grad_norm": 8.555191993713379, "learning_rate": 8.955645066706102e-06, "loss": 0.5058, "step": 4298 }, { "epoch": 0.21746168243209066, "grad_norm": 13.322113990783691, "learning_rate": 8.95515409729531e-06, "loss": 0.5848, "step": 4299 }, { "epoch": 0.2175122666801558, "grad_norm": 12.960978507995605, "learning_rate": 8.954663025969745e-06, "loss": 0.5546, "step": 4300 }, { "epoch": 0.21756285092822095, "grad_norm": 10.212995529174805, "learning_rate": 8.95417185274207e-06, "loss": 0.5184, "step": 4301 }, { "epoch": 0.2176134351762861, "grad_norm": 9.853837966918945, "learning_rate": 8.953680577624935e-06, "loss": 0.5231, "step": 4302 }, { "epoch": 0.21766401942435126, "grad_norm": 6.6368913650512695, "learning_rate": 8.953189200631003e-06, "loss": 0.3797, "step": 4303 }, { "epoch": 0.2177146036724164, "grad_norm": 11.28317928314209, "learning_rate": 8.952697721772934e-06, "loss": 0.4231, "step": 4304 }, { "epoch": 0.21776518792048155, "grad_norm": 11.516200065612793, "learning_rate": 8.952206141063392e-06, "loss": 0.4747, "step": 4305 }, { "epoch": 0.21781577216854672, "grad_norm": 7.093265056610107, "learning_rate": 8.951714458515043e-06, "loss": 0.4229, "step": 4306 }, { "epoch": 0.21786635641661187, "grad_norm": 8.43027114868164, "learning_rate": 8.951222674140556e-06, "loss": 0.371, "step": 4307 }, { "epoch": 0.217916940664677, "grad_norm": 10.482519149780273, "learning_rate": 8.950730787952607e-06, "loss": 0.4732, "step": 4308 }, { "epoch": 0.21796752491274218, "grad_norm": 10.807901382446289, "learning_rate": 8.950238799963867e-06, "loss": 0.542, "step": 4309 }, { "epoch": 0.21801810916080733, "grad_norm": 11.744548797607422, "learning_rate": 8.949746710187013e-06, "loss": 0.4705, "step": 4310 }, { "epoch": 0.21806869340887247, "grad_norm": 11.88855266571045, "learning_rate": 8.94925451863473e-06, "loss": 0.4686, "step": 4311 }, { "epoch": 0.21811927765693764, "grad_norm": 5.1142191886901855, "learning_rate": 8.948762225319694e-06, "loss": 0.4138, "step": 4312 }, { "epoch": 0.2181698619050028, "grad_norm": 6.13362979888916, "learning_rate": 8.948269830254594e-06, "loss": 0.4686, "step": 4313 }, { "epoch": 0.21822044615306793, "grad_norm": 8.473983764648438, "learning_rate": 8.947777333452119e-06, "loss": 0.4837, "step": 4314 }, { "epoch": 0.21827103040113308, "grad_norm": 10.049715995788574, "learning_rate": 8.947284734924956e-06, "loss": 0.4968, "step": 4315 }, { "epoch": 0.21832161464919825, "grad_norm": 9.016568183898926, "learning_rate": 8.946792034685801e-06, "loss": 0.4513, "step": 4316 }, { "epoch": 0.2183721988972634, "grad_norm": 8.36361026763916, "learning_rate": 8.946299232747346e-06, "loss": 0.4825, "step": 4317 }, { "epoch": 0.21842278314532854, "grad_norm": 8.66204833984375, "learning_rate": 8.945806329122293e-06, "loss": 0.4392, "step": 4318 }, { "epoch": 0.2184733673933937, "grad_norm": 8.536328315734863, "learning_rate": 8.945313323823341e-06, "loss": 0.4937, "step": 4319 }, { "epoch": 0.21852395164145885, "grad_norm": 15.047514915466309, "learning_rate": 8.944820216863195e-06, "loss": 0.6458, "step": 4320 }, { "epoch": 0.218574535889524, "grad_norm": 13.889508247375488, "learning_rate": 8.94432700825456e-06, "loss": 0.5094, "step": 4321 }, { "epoch": 0.21862512013758917, "grad_norm": 10.625664710998535, "learning_rate": 8.943833698010146e-06, "loss": 0.6155, "step": 4322 }, { "epoch": 0.2186757043856543, "grad_norm": 9.499403953552246, "learning_rate": 8.943340286142663e-06, "loss": 0.4082, "step": 4323 }, { "epoch": 0.21872628863371946, "grad_norm": 6.538654327392578, "learning_rate": 8.942846772664824e-06, "loss": 0.5049, "step": 4324 }, { "epoch": 0.2187768728817846, "grad_norm": 5.990056991577148, "learning_rate": 8.942353157589349e-06, "loss": 0.4128, "step": 4325 }, { "epoch": 0.21882745712984977, "grad_norm": 12.63615894317627, "learning_rate": 8.941859440928955e-06, "loss": 0.4957, "step": 4326 }, { "epoch": 0.21887804137791492, "grad_norm": 9.888038635253906, "learning_rate": 8.941365622696366e-06, "loss": 0.5539, "step": 4327 }, { "epoch": 0.21892862562598006, "grad_norm": 7.798803329467773, "learning_rate": 8.940871702904301e-06, "loss": 0.5322, "step": 4328 }, { "epoch": 0.21897920987404523, "grad_norm": 6.3409295082092285, "learning_rate": 8.940377681565493e-06, "loss": 0.4832, "step": 4329 }, { "epoch": 0.21902979412211038, "grad_norm": 5.272809028625488, "learning_rate": 8.939883558692672e-06, "loss": 0.4597, "step": 4330 }, { "epoch": 0.21908037837017552, "grad_norm": 5.808504104614258, "learning_rate": 8.939389334298566e-06, "loss": 0.3858, "step": 4331 }, { "epoch": 0.2191309626182407, "grad_norm": 7.325862884521484, "learning_rate": 8.93889500839591e-06, "loss": 0.4492, "step": 4332 }, { "epoch": 0.21918154686630584, "grad_norm": 10.600041389465332, "learning_rate": 8.938400580997448e-06, "loss": 0.4421, "step": 4333 }, { "epoch": 0.21923213111437098, "grad_norm": 9.775690078735352, "learning_rate": 8.937906052115914e-06, "loss": 0.5516, "step": 4334 }, { "epoch": 0.21928271536243613, "grad_norm": 7.590155124664307, "learning_rate": 8.93741142176405e-06, "loss": 0.4156, "step": 4335 }, { "epoch": 0.2193332996105013, "grad_norm": 7.792145729064941, "learning_rate": 8.936916689954608e-06, "loss": 0.4359, "step": 4336 }, { "epoch": 0.21938388385856644, "grad_norm": 7.81179666519165, "learning_rate": 8.93642185670033e-06, "loss": 0.4745, "step": 4337 }, { "epoch": 0.21943446810663159, "grad_norm": 7.639970302581787, "learning_rate": 8.935926922013969e-06, "loss": 0.5145, "step": 4338 }, { "epoch": 0.21948505235469676, "grad_norm": 7.502389430999756, "learning_rate": 8.935431885908279e-06, "loss": 0.5194, "step": 4339 }, { "epoch": 0.2195356366027619, "grad_norm": 6.208034038543701, "learning_rate": 8.934936748396015e-06, "loss": 0.4193, "step": 4340 }, { "epoch": 0.21958622085082705, "grad_norm": 8.168228149414062, "learning_rate": 8.934441509489934e-06, "loss": 0.4636, "step": 4341 }, { "epoch": 0.21963680509889222, "grad_norm": 8.56778335571289, "learning_rate": 8.9339461692028e-06, "loss": 0.3777, "step": 4342 }, { "epoch": 0.21968738934695736, "grad_norm": 7.2308268547058105, "learning_rate": 8.933450727547374e-06, "loss": 0.4169, "step": 4343 }, { "epoch": 0.2197379735950225, "grad_norm": 10.961156845092773, "learning_rate": 8.932955184536425e-06, "loss": 0.4396, "step": 4344 }, { "epoch": 0.21978855784308765, "grad_norm": 6.4934306144714355, "learning_rate": 8.932459540182718e-06, "loss": 0.4584, "step": 4345 }, { "epoch": 0.21983914209115282, "grad_norm": 9.968791007995605, "learning_rate": 8.93196379449903e-06, "loss": 0.4591, "step": 4346 }, { "epoch": 0.21988972633921797, "grad_norm": 7.82755708694458, "learning_rate": 8.931467947498131e-06, "loss": 0.4787, "step": 4347 }, { "epoch": 0.2199403105872831, "grad_norm": 11.064516067504883, "learning_rate": 8.9309719991928e-06, "loss": 0.3755, "step": 4348 }, { "epoch": 0.21999089483534828, "grad_norm": 18.793027877807617, "learning_rate": 8.930475949595813e-06, "loss": 0.5549, "step": 4349 }, { "epoch": 0.22004147908341343, "grad_norm": 9.446364402770996, "learning_rate": 8.929979798719958e-06, "loss": 0.5362, "step": 4350 }, { "epoch": 0.22009206333147857, "grad_norm": 14.247244834899902, "learning_rate": 8.929483546578014e-06, "loss": 0.6417, "step": 4351 }, { "epoch": 0.22014264757954374, "grad_norm": 8.550080299377441, "learning_rate": 8.92898719318277e-06, "loss": 0.3785, "step": 4352 }, { "epoch": 0.2201932318276089, "grad_norm": 5.010010719299316, "learning_rate": 8.928490738547017e-06, "loss": 0.3748, "step": 4353 }, { "epoch": 0.22024381607567403, "grad_norm": 8.323515892028809, "learning_rate": 8.927994182683546e-06, "loss": 0.5292, "step": 4354 }, { "epoch": 0.22029440032373918, "grad_norm": 13.395918846130371, "learning_rate": 8.927497525605154e-06, "loss": 0.4193, "step": 4355 }, { "epoch": 0.22034498457180435, "grad_norm": 12.688130378723145, "learning_rate": 8.927000767324636e-06, "loss": 0.4828, "step": 4356 }, { "epoch": 0.2203955688198695, "grad_norm": 5.91054105758667, "learning_rate": 8.926503907854792e-06, "loss": 0.3662, "step": 4357 }, { "epoch": 0.22044615306793464, "grad_norm": 8.1702241897583, "learning_rate": 8.926006947208427e-06, "loss": 0.5418, "step": 4358 }, { "epoch": 0.2204967373159998, "grad_norm": 6.887679100036621, "learning_rate": 8.925509885398348e-06, "loss": 0.3845, "step": 4359 }, { "epoch": 0.22054732156406495, "grad_norm": 12.969137191772461, "learning_rate": 8.92501272243736e-06, "loss": 0.4406, "step": 4360 }, { "epoch": 0.2205979058121301, "grad_norm": 9.506442070007324, "learning_rate": 8.924515458338274e-06, "loss": 0.5273, "step": 4361 }, { "epoch": 0.22064849006019527, "grad_norm": 10.219681739807129, "learning_rate": 8.924018093113904e-06, "loss": 0.5327, "step": 4362 }, { "epoch": 0.2206990743082604, "grad_norm": 8.714673042297363, "learning_rate": 8.923520626777067e-06, "loss": 0.3852, "step": 4363 }, { "epoch": 0.22074965855632556, "grad_norm": 11.690849304199219, "learning_rate": 8.923023059340579e-06, "loss": 0.3729, "step": 4364 }, { "epoch": 0.2208002428043907, "grad_norm": 6.916945457458496, "learning_rate": 8.922525390817263e-06, "loss": 0.4306, "step": 4365 }, { "epoch": 0.22085082705245587, "grad_norm": 10.485986709594727, "learning_rate": 8.922027621219943e-06, "loss": 0.4928, "step": 4366 }, { "epoch": 0.22090141130052102, "grad_norm": 7.585412979125977, "learning_rate": 8.921529750561441e-06, "loss": 0.4491, "step": 4367 }, { "epoch": 0.22095199554858616, "grad_norm": 7.19830322265625, "learning_rate": 8.921031778854592e-06, "loss": 0.3986, "step": 4368 }, { "epoch": 0.22100257979665133, "grad_norm": 5.549030780792236, "learning_rate": 8.920533706112226e-06, "loss": 0.3947, "step": 4369 }, { "epoch": 0.22105316404471648, "grad_norm": 13.033204078674316, "learning_rate": 8.920035532347175e-06, "loss": 0.5894, "step": 4370 }, { "epoch": 0.22110374829278162, "grad_norm": 8.674509048461914, "learning_rate": 8.919537257572275e-06, "loss": 0.3732, "step": 4371 }, { "epoch": 0.2211543325408468, "grad_norm": 7.798692226409912, "learning_rate": 8.919038881800371e-06, "loss": 0.4872, "step": 4372 }, { "epoch": 0.22120491678891194, "grad_norm": 10.883064270019531, "learning_rate": 8.918540405044297e-06, "loss": 0.5221, "step": 4373 }, { "epoch": 0.22125550103697708, "grad_norm": 20.19695472717285, "learning_rate": 8.918041827316904e-06, "loss": 0.4894, "step": 4374 }, { "epoch": 0.22130608528504223, "grad_norm": 7.67157506942749, "learning_rate": 8.917543148631036e-06, "loss": 0.5396, "step": 4375 }, { "epoch": 0.2213566695331074, "grad_norm": 8.363077163696289, "learning_rate": 8.917044368999544e-06, "loss": 0.4498, "step": 4376 }, { "epoch": 0.22140725378117254, "grad_norm": 8.348625183105469, "learning_rate": 8.916545488435279e-06, "loss": 0.4817, "step": 4377 }, { "epoch": 0.22145783802923769, "grad_norm": 9.181126594543457, "learning_rate": 8.916046506951097e-06, "loss": 0.4323, "step": 4378 }, { "epoch": 0.22150842227730286, "grad_norm": 13.288385391235352, "learning_rate": 8.915547424559855e-06, "loss": 0.3915, "step": 4379 }, { "epoch": 0.221559006525368, "grad_norm": 12.435251235961914, "learning_rate": 8.915048241274416e-06, "loss": 0.5147, "step": 4380 }, { "epoch": 0.22160959077343315, "grad_norm": 10.40613842010498, "learning_rate": 8.914548957107637e-06, "loss": 0.4441, "step": 4381 }, { "epoch": 0.22166017502149832, "grad_norm": 8.457386016845703, "learning_rate": 8.914049572072387e-06, "loss": 0.4699, "step": 4382 }, { "epoch": 0.22171075926956346, "grad_norm": 6.875565052032471, "learning_rate": 8.913550086181535e-06, "loss": 0.364, "step": 4383 }, { "epoch": 0.2217613435176286, "grad_norm": 7.280909061431885, "learning_rate": 8.91305049944795e-06, "loss": 0.4336, "step": 4384 }, { "epoch": 0.22181192776569375, "grad_norm": 6.875636577606201, "learning_rate": 8.912550811884505e-06, "loss": 0.4815, "step": 4385 }, { "epoch": 0.22186251201375892, "grad_norm": 7.520130157470703, "learning_rate": 8.912051023504077e-06, "loss": 0.4974, "step": 4386 }, { "epoch": 0.22191309626182407, "grad_norm": 8.805042266845703, "learning_rate": 8.911551134319541e-06, "loss": 0.4713, "step": 4387 }, { "epoch": 0.2219636805098892, "grad_norm": 29.722476959228516, "learning_rate": 8.91105114434378e-06, "loss": 0.4939, "step": 4388 }, { "epoch": 0.22201426475795438, "grad_norm": 3.9759926795959473, "learning_rate": 8.91055105358968e-06, "loss": 0.4819, "step": 4389 }, { "epoch": 0.22206484900601953, "grad_norm": 5.203561305999756, "learning_rate": 8.910050862070123e-06, "loss": 0.4106, "step": 4390 }, { "epoch": 0.22211543325408467, "grad_norm": 7.131525993347168, "learning_rate": 8.909550569798e-06, "loss": 0.4854, "step": 4391 }, { "epoch": 0.22216601750214984, "grad_norm": 6.815838813781738, "learning_rate": 8.909050176786204e-06, "loss": 0.3857, "step": 4392 }, { "epoch": 0.222216601750215, "grad_norm": 7.260682582855225, "learning_rate": 8.908549683047625e-06, "loss": 0.3997, "step": 4393 }, { "epoch": 0.22226718599828013, "grad_norm": 7.088644504547119, "learning_rate": 8.908049088595161e-06, "loss": 0.3857, "step": 4394 }, { "epoch": 0.22231777024634528, "grad_norm": 11.207452774047852, "learning_rate": 8.907548393441712e-06, "loss": 0.4031, "step": 4395 }, { "epoch": 0.22236835449441045, "grad_norm": 7.177798271179199, "learning_rate": 8.90704759760018e-06, "loss": 0.4386, "step": 4396 }, { "epoch": 0.2224189387424756, "grad_norm": 10.373217582702637, "learning_rate": 8.906546701083468e-06, "loss": 0.4071, "step": 4397 }, { "epoch": 0.22246952299054074, "grad_norm": 11.608933448791504, "learning_rate": 8.906045703904483e-06, "loss": 0.4562, "step": 4398 }, { "epoch": 0.2225201072386059, "grad_norm": 10.703642845153809, "learning_rate": 8.905544606076135e-06, "loss": 0.4685, "step": 4399 }, { "epoch": 0.22257069148667105, "grad_norm": 10.15883731842041, "learning_rate": 8.905043407611336e-06, "loss": 0.5345, "step": 4400 }, { "epoch": 0.2226212757347362, "grad_norm": 14.961158752441406, "learning_rate": 8.904542108523e-06, "loss": 0.4546, "step": 4401 }, { "epoch": 0.22267185998280137, "grad_norm": 13.18478775024414, "learning_rate": 8.904040708824046e-06, "loss": 0.4838, "step": 4402 }, { "epoch": 0.2227224442308665, "grad_norm": 9.3077392578125, "learning_rate": 8.903539208527394e-06, "loss": 0.4291, "step": 4403 }, { "epoch": 0.22277302847893166, "grad_norm": 9.960761070251465, "learning_rate": 8.903037607645963e-06, "loss": 0.5105, "step": 4404 }, { "epoch": 0.2228236127269968, "grad_norm": 5.83266019821167, "learning_rate": 8.90253590619268e-06, "loss": 0.3952, "step": 4405 }, { "epoch": 0.22287419697506197, "grad_norm": 8.847212791442871, "learning_rate": 8.902034104180475e-06, "loss": 0.3425, "step": 4406 }, { "epoch": 0.22292478122312712, "grad_norm": 9.435040473937988, "learning_rate": 8.901532201622273e-06, "loss": 0.4937, "step": 4407 }, { "epoch": 0.22297536547119226, "grad_norm": 6.4028801918029785, "learning_rate": 8.901030198531013e-06, "loss": 0.4098, "step": 4408 }, { "epoch": 0.22302594971925743, "grad_norm": 9.196577072143555, "learning_rate": 8.900528094919626e-06, "loss": 0.5373, "step": 4409 }, { "epoch": 0.22307653396732258, "grad_norm": 12.272722244262695, "learning_rate": 8.900025890801052e-06, "loss": 0.5494, "step": 4410 }, { "epoch": 0.22312711821538772, "grad_norm": 6.655494213104248, "learning_rate": 8.899523586188231e-06, "loss": 0.4664, "step": 4411 }, { "epoch": 0.2231777024634529, "grad_norm": 22.622983932495117, "learning_rate": 8.899021181094106e-06, "loss": 0.544, "step": 4412 }, { "epoch": 0.22322828671151804, "grad_norm": 10.913012504577637, "learning_rate": 8.89851867553162e-06, "loss": 0.4218, "step": 4413 }, { "epoch": 0.22327887095958318, "grad_norm": 6.827154636383057, "learning_rate": 8.898016069513728e-06, "loss": 0.3137, "step": 4414 }, { "epoch": 0.22332945520764833, "grad_norm": 7.403804779052734, "learning_rate": 8.897513363053374e-06, "loss": 0.4695, "step": 4415 }, { "epoch": 0.2233800394557135, "grad_norm": 5.657413482666016, "learning_rate": 8.897010556163517e-06, "loss": 0.4177, "step": 4416 }, { "epoch": 0.22343062370377864, "grad_norm": 7.68583869934082, "learning_rate": 8.896507648857112e-06, "loss": 0.4023, "step": 4417 }, { "epoch": 0.22348120795184379, "grad_norm": 7.872446060180664, "learning_rate": 8.896004641147114e-06, "loss": 0.3682, "step": 4418 }, { "epoch": 0.22353179219990896, "grad_norm": 8.306832313537598, "learning_rate": 8.895501533046486e-06, "loss": 0.478, "step": 4419 }, { "epoch": 0.2235823764479741, "grad_norm": 4.8768815994262695, "learning_rate": 8.894998324568195e-06, "loss": 0.4429, "step": 4420 }, { "epoch": 0.22363296069603925, "grad_norm": 6.137423515319824, "learning_rate": 8.894495015725205e-06, "loss": 0.4793, "step": 4421 }, { "epoch": 0.22368354494410442, "grad_norm": 8.01335620880127, "learning_rate": 8.893991606530485e-06, "loss": 0.4184, "step": 4422 }, { "epoch": 0.22373412919216956, "grad_norm": 12.600041389465332, "learning_rate": 8.893488096997005e-06, "loss": 0.3333, "step": 4423 }, { "epoch": 0.2237847134402347, "grad_norm": 9.632163047790527, "learning_rate": 8.892984487137744e-06, "loss": 0.4794, "step": 4424 }, { "epoch": 0.22383529768829985, "grad_norm": 13.669656753540039, "learning_rate": 8.892480776965672e-06, "loss": 0.6138, "step": 4425 }, { "epoch": 0.22388588193636502, "grad_norm": 8.73986530303955, "learning_rate": 8.891976966493776e-06, "loss": 0.3883, "step": 4426 }, { "epoch": 0.22393646618443017, "grad_norm": 6.882490158081055, "learning_rate": 8.891473055735035e-06, "loss": 0.3688, "step": 4427 }, { "epoch": 0.2239870504324953, "grad_norm": 7.062494277954102, "learning_rate": 8.890969044702428e-06, "loss": 0.451, "step": 4428 }, { "epoch": 0.22403763468056048, "grad_norm": 13.689258575439453, "learning_rate": 8.890464933408951e-06, "loss": 0.3727, "step": 4429 }, { "epoch": 0.22408821892862563, "grad_norm": 18.717782974243164, "learning_rate": 8.889960721867588e-06, "loss": 0.4978, "step": 4430 }, { "epoch": 0.22413880317669077, "grad_norm": 9.097362518310547, "learning_rate": 8.889456410091333e-06, "loss": 0.3985, "step": 4431 }, { "epoch": 0.22418938742475594, "grad_norm": 7.49066686630249, "learning_rate": 8.888951998093183e-06, "loss": 0.5728, "step": 4432 }, { "epoch": 0.2242399716728211, "grad_norm": 11.376731872558594, "learning_rate": 8.888447485886132e-06, "loss": 0.5066, "step": 4433 }, { "epoch": 0.22429055592088623, "grad_norm": 6.555022239685059, "learning_rate": 8.887942873483179e-06, "loss": 0.4065, "step": 4434 }, { "epoch": 0.22434114016895138, "grad_norm": 8.512404441833496, "learning_rate": 8.887438160897333e-06, "loss": 0.4404, "step": 4435 }, { "epoch": 0.22439172441701655, "grad_norm": 5.7249226570129395, "learning_rate": 8.886933348141593e-06, "loss": 0.41, "step": 4436 }, { "epoch": 0.2244423086650817, "grad_norm": 7.649599552154541, "learning_rate": 8.886428435228971e-06, "loss": 0.4539, "step": 4437 }, { "epoch": 0.22449289291314684, "grad_norm": 6.855589389801025, "learning_rate": 8.885923422172473e-06, "loss": 0.4527, "step": 4438 }, { "epoch": 0.224543477161212, "grad_norm": 16.454097747802734, "learning_rate": 8.885418308985116e-06, "loss": 0.565, "step": 4439 }, { "epoch": 0.22459406140927715, "grad_norm": 10.774162292480469, "learning_rate": 8.884913095679914e-06, "loss": 0.5178, "step": 4440 }, { "epoch": 0.2246446456573423, "grad_norm": 6.377718925476074, "learning_rate": 8.884407782269886e-06, "loss": 0.4583, "step": 4441 }, { "epoch": 0.22469522990540747, "grad_norm": 8.302244186401367, "learning_rate": 8.883902368768049e-06, "loss": 0.3017, "step": 4442 }, { "epoch": 0.2247458141534726, "grad_norm": 5.600888252258301, "learning_rate": 8.883396855187431e-06, "loss": 0.3597, "step": 4443 }, { "epoch": 0.22479639840153776, "grad_norm": 8.087051391601562, "learning_rate": 8.882891241541057e-06, "loss": 0.5262, "step": 4444 }, { "epoch": 0.2248469826496029, "grad_norm": 6.05983304977417, "learning_rate": 8.882385527841953e-06, "loss": 0.428, "step": 4445 }, { "epoch": 0.22489756689766807, "grad_norm": 9.319855690002441, "learning_rate": 8.881879714103152e-06, "loss": 0.4855, "step": 4446 }, { "epoch": 0.22494815114573322, "grad_norm": 12.734335899353027, "learning_rate": 8.881373800337687e-06, "loss": 0.4553, "step": 4447 }, { "epoch": 0.22499873539379836, "grad_norm": 8.310052871704102, "learning_rate": 8.880867786558594e-06, "loss": 0.4177, "step": 4448 }, { "epoch": 0.22504931964186353, "grad_norm": 8.089274406433105, "learning_rate": 8.880361672778912e-06, "loss": 0.4822, "step": 4449 }, { "epoch": 0.22509990388992868, "grad_norm": 11.3814058303833, "learning_rate": 8.879855459011682e-06, "loss": 0.4193, "step": 4450 }, { "epoch": 0.22515048813799382, "grad_norm": 16.550722122192383, "learning_rate": 8.879349145269949e-06, "loss": 0.5011, "step": 4451 }, { "epoch": 0.225201072386059, "grad_norm": 13.401872634887695, "learning_rate": 8.878842731566758e-06, "loss": 0.6981, "step": 4452 }, { "epoch": 0.22525165663412414, "grad_norm": 73.37538146972656, "learning_rate": 8.87833621791516e-06, "loss": 0.5165, "step": 4453 }, { "epoch": 0.22530224088218928, "grad_norm": 6.643610954284668, "learning_rate": 8.877829604328203e-06, "loss": 0.347, "step": 4454 }, { "epoch": 0.22535282513025442, "grad_norm": 8.871471405029297, "learning_rate": 8.877322890818946e-06, "loss": 0.5515, "step": 4455 }, { "epoch": 0.2254034093783196, "grad_norm": 12.515632629394531, "learning_rate": 8.876816077400443e-06, "loss": 0.5775, "step": 4456 }, { "epoch": 0.22545399362638474, "grad_norm": 17.700958251953125, "learning_rate": 8.876309164085752e-06, "loss": 0.537, "step": 4457 }, { "epoch": 0.22550457787444989, "grad_norm": 8.62712574005127, "learning_rate": 8.875802150887938e-06, "loss": 0.4534, "step": 4458 }, { "epoch": 0.22555516212251506, "grad_norm": 4.884189128875732, "learning_rate": 8.875295037820064e-06, "loss": 0.4486, "step": 4459 }, { "epoch": 0.2256057463705802, "grad_norm": 10.573970794677734, "learning_rate": 8.874787824895199e-06, "loss": 0.4586, "step": 4460 }, { "epoch": 0.22565633061864535, "grad_norm": 7.956830024719238, "learning_rate": 8.874280512126407e-06, "loss": 0.3726, "step": 4461 }, { "epoch": 0.22570691486671052, "grad_norm": 8.890522003173828, "learning_rate": 8.873773099526769e-06, "loss": 0.3738, "step": 4462 }, { "epoch": 0.22575749911477566, "grad_norm": 21.60013771057129, "learning_rate": 8.873265587109352e-06, "loss": 0.3771, "step": 4463 }, { "epoch": 0.2258080833628408, "grad_norm": 9.094650268554688, "learning_rate": 8.872757974887236e-06, "loss": 0.4211, "step": 4464 }, { "epoch": 0.22585866761090598, "grad_norm": 7.861051559448242, "learning_rate": 8.872250262873502e-06, "loss": 0.5214, "step": 4465 }, { "epoch": 0.22590925185897112, "grad_norm": 8.54092788696289, "learning_rate": 8.871742451081231e-06, "loss": 0.5954, "step": 4466 }, { "epoch": 0.22595983610703627, "grad_norm": 14.031390190124512, "learning_rate": 8.87123453952351e-06, "loss": 0.4247, "step": 4467 }, { "epoch": 0.2260104203551014, "grad_norm": 5.240217208862305, "learning_rate": 8.870726528213424e-06, "loss": 0.4009, "step": 4468 }, { "epoch": 0.22606100460316658, "grad_norm": 8.898626327514648, "learning_rate": 8.870218417164066e-06, "loss": 0.5678, "step": 4469 }, { "epoch": 0.22611158885123173, "grad_norm": 6.537588596343994, "learning_rate": 8.869710206388529e-06, "loss": 0.4201, "step": 4470 }, { "epoch": 0.22616217309929687, "grad_norm": 10.613000869750977, "learning_rate": 8.869201895899905e-06, "loss": 0.4624, "step": 4471 }, { "epoch": 0.22621275734736204, "grad_norm": 11.730061531066895, "learning_rate": 8.868693485711294e-06, "loss": 0.4456, "step": 4472 }, { "epoch": 0.2262633415954272, "grad_norm": 6.389041900634766, "learning_rate": 8.868184975835797e-06, "loss": 0.4064, "step": 4473 }, { "epoch": 0.22631392584349233, "grad_norm": 9.68272876739502, "learning_rate": 8.867676366286516e-06, "loss": 0.4942, "step": 4474 }, { "epoch": 0.2263645100915575, "grad_norm": 8.11806583404541, "learning_rate": 8.867167657076556e-06, "loss": 0.2779, "step": 4475 }, { "epoch": 0.22641509433962265, "grad_norm": 13.038342475891113, "learning_rate": 8.866658848219028e-06, "loss": 0.5619, "step": 4476 }, { "epoch": 0.2264656785876878, "grad_norm": 6.179069519042969, "learning_rate": 8.866149939727038e-06, "loss": 0.459, "step": 4477 }, { "epoch": 0.22651626283575294, "grad_norm": 8.970610618591309, "learning_rate": 8.865640931613707e-06, "loss": 0.5388, "step": 4478 }, { "epoch": 0.2265668470838181, "grad_norm": 8.613036155700684, "learning_rate": 8.865131823892142e-06, "loss": 0.4594, "step": 4479 }, { "epoch": 0.22661743133188325, "grad_norm": 12.577343940734863, "learning_rate": 8.864622616575467e-06, "loss": 0.56, "step": 4480 }, { "epoch": 0.2266680155799484, "grad_norm": 9.86139965057373, "learning_rate": 8.864113309676803e-06, "loss": 0.4754, "step": 4481 }, { "epoch": 0.22671859982801357, "grad_norm": 8.947598457336426, "learning_rate": 8.863603903209271e-06, "loss": 0.421, "step": 4482 }, { "epoch": 0.2267691840760787, "grad_norm": 13.66969108581543, "learning_rate": 8.863094397185999e-06, "loss": 0.4741, "step": 4483 }, { "epoch": 0.22681976832414386, "grad_norm": 11.871048927307129, "learning_rate": 8.862584791620116e-06, "loss": 0.5092, "step": 4484 }, { "epoch": 0.22687035257220903, "grad_norm": 4.98246431350708, "learning_rate": 8.862075086524752e-06, "loss": 0.4012, "step": 4485 }, { "epoch": 0.22692093682027417, "grad_norm": 6.248410701751709, "learning_rate": 8.86156528191304e-06, "loss": 0.4069, "step": 4486 }, { "epoch": 0.22697152106833932, "grad_norm": 10.41528034210205, "learning_rate": 8.86105537779812e-06, "loss": 0.5543, "step": 4487 }, { "epoch": 0.22702210531640446, "grad_norm": 6.80615758895874, "learning_rate": 8.86054537419313e-06, "loss": 0.2928, "step": 4488 }, { "epoch": 0.22707268956446963, "grad_norm": 6.327208042144775, "learning_rate": 8.860035271111208e-06, "loss": 0.419, "step": 4489 }, { "epoch": 0.22712327381253478, "grad_norm": 8.46734619140625, "learning_rate": 8.859525068565502e-06, "loss": 0.4577, "step": 4490 }, { "epoch": 0.22717385806059992, "grad_norm": 7.657981872558594, "learning_rate": 8.859014766569157e-06, "loss": 0.51, "step": 4491 }, { "epoch": 0.2272244423086651, "grad_norm": 12.044445037841797, "learning_rate": 8.858504365135322e-06, "loss": 0.4626, "step": 4492 }, { "epoch": 0.22727502655673024, "grad_norm": 7.462561130523682, "learning_rate": 8.857993864277149e-06, "loss": 0.4745, "step": 4493 }, { "epoch": 0.22732561080479538, "grad_norm": 9.227614402770996, "learning_rate": 8.857483264007794e-06, "loss": 0.3896, "step": 4494 }, { "epoch": 0.22737619505286055, "grad_norm": 14.166614532470703, "learning_rate": 8.85697256434041e-06, "loss": 0.5879, "step": 4495 }, { "epoch": 0.2274267793009257, "grad_norm": 6.40573787689209, "learning_rate": 8.856461765288163e-06, "loss": 0.4167, "step": 4496 }, { "epoch": 0.22747736354899084, "grad_norm": 4.622775554656982, "learning_rate": 8.855950866864209e-06, "loss": 0.3124, "step": 4497 }, { "epoch": 0.22752794779705598, "grad_norm": 11.964210510253906, "learning_rate": 8.855439869081715e-06, "loss": 0.5022, "step": 4498 }, { "epoch": 0.22757853204512116, "grad_norm": 9.791966438293457, "learning_rate": 8.854928771953848e-06, "loss": 0.4589, "step": 4499 }, { "epoch": 0.2276291162931863, "grad_norm": 10.957745552062988, "learning_rate": 8.854417575493777e-06, "loss": 0.6718, "step": 4500 }, { "epoch": 0.22767970054125145, "grad_norm": 5.612011909484863, "learning_rate": 8.853906279714677e-06, "loss": 0.4285, "step": 4501 }, { "epoch": 0.22773028478931662, "grad_norm": 12.395450592041016, "learning_rate": 8.85339488462972e-06, "loss": 0.423, "step": 4502 }, { "epoch": 0.22778086903738176, "grad_norm": 7.745953559875488, "learning_rate": 8.852883390252083e-06, "loss": 0.4537, "step": 4503 }, { "epoch": 0.2278314532854469, "grad_norm": 9.051981925964355, "learning_rate": 8.85237179659495e-06, "loss": 0.4809, "step": 4504 }, { "epoch": 0.22788203753351208, "grad_norm": 3.850710391998291, "learning_rate": 8.8518601036715e-06, "loss": 0.338, "step": 4505 }, { "epoch": 0.22793262178157722, "grad_norm": 7.688952445983887, "learning_rate": 8.851348311494915e-06, "loss": 0.3987, "step": 4506 }, { "epoch": 0.22798320602964237, "grad_norm": 6.0847368240356445, "learning_rate": 8.850836420078392e-06, "loss": 0.3984, "step": 4507 }, { "epoch": 0.2280337902777075, "grad_norm": 6.756451606750488, "learning_rate": 8.850324429435113e-06, "loss": 0.3791, "step": 4508 }, { "epoch": 0.22808437452577268, "grad_norm": 11.255313873291016, "learning_rate": 8.849812339578274e-06, "loss": 0.5873, "step": 4509 }, { "epoch": 0.22813495877383783, "grad_norm": 19.03148078918457, "learning_rate": 8.84930015052107e-06, "loss": 0.5019, "step": 4510 }, { "epoch": 0.22818554302190297, "grad_norm": 7.449468612670898, "learning_rate": 8.8487878622767e-06, "loss": 0.4536, "step": 4511 }, { "epoch": 0.22823612726996814, "grad_norm": 7.421839714050293, "learning_rate": 8.84827547485836e-06, "loss": 0.4724, "step": 4512 }, { "epoch": 0.2282867115180333, "grad_norm": 11.077615737915039, "learning_rate": 8.847762988279259e-06, "loss": 0.6121, "step": 4513 }, { "epoch": 0.22833729576609843, "grad_norm": 9.31972885131836, "learning_rate": 8.847250402552599e-06, "loss": 0.4168, "step": 4514 }, { "epoch": 0.2283878800141636, "grad_norm": 8.731095314025879, "learning_rate": 8.846737717691588e-06, "loss": 0.599, "step": 4515 }, { "epoch": 0.22843846426222875, "grad_norm": 8.353120803833008, "learning_rate": 8.846224933709438e-06, "loss": 0.439, "step": 4516 }, { "epoch": 0.2284890485102939, "grad_norm": 10.968067169189453, "learning_rate": 8.845712050619362e-06, "loss": 0.4373, "step": 4517 }, { "epoch": 0.22853963275835903, "grad_norm": 6.3430328369140625, "learning_rate": 8.845199068434574e-06, "loss": 0.4016, "step": 4518 }, { "epoch": 0.2285902170064242, "grad_norm": 4.958012104034424, "learning_rate": 8.844685987168295e-06, "loss": 0.3984, "step": 4519 }, { "epoch": 0.22864080125448935, "grad_norm": 10.42190933227539, "learning_rate": 8.844172806833744e-06, "loss": 0.5578, "step": 4520 }, { "epoch": 0.2286913855025545, "grad_norm": 9.482877731323242, "learning_rate": 8.843659527444144e-06, "loss": 0.5222, "step": 4521 }, { "epoch": 0.22874196975061967, "grad_norm": 6.529405117034912, "learning_rate": 8.843146149012724e-06, "loss": 0.4042, "step": 4522 }, { "epoch": 0.2287925539986848, "grad_norm": 9.175847053527832, "learning_rate": 8.84263267155271e-06, "loss": 0.3927, "step": 4523 }, { "epoch": 0.22884313824674996, "grad_norm": 6.322876930236816, "learning_rate": 8.842119095077332e-06, "loss": 0.4133, "step": 4524 }, { "epoch": 0.22889372249481513, "grad_norm": 7.111395835876465, "learning_rate": 8.841605419599828e-06, "loss": 0.4493, "step": 4525 }, { "epoch": 0.22894430674288027, "grad_norm": 7.679853916168213, "learning_rate": 8.841091645133428e-06, "loss": 0.4397, "step": 4526 }, { "epoch": 0.22899489099094542, "grad_norm": 26.311073303222656, "learning_rate": 8.840577771691376e-06, "loss": 0.5535, "step": 4527 }, { "epoch": 0.22904547523901056, "grad_norm": 9.57934856414795, "learning_rate": 8.84006379928691e-06, "loss": 0.561, "step": 4528 }, { "epoch": 0.22909605948707573, "grad_norm": 15.958495140075684, "learning_rate": 8.839549727933277e-06, "loss": 0.6642, "step": 4529 }, { "epoch": 0.22914664373514088, "grad_norm": 5.918814659118652, "learning_rate": 8.839035557643721e-06, "loss": 0.4079, "step": 4530 }, { "epoch": 0.22919722798320602, "grad_norm": 7.2392096519470215, "learning_rate": 8.838521288431492e-06, "loss": 0.4793, "step": 4531 }, { "epoch": 0.2292478122312712, "grad_norm": 13.492436408996582, "learning_rate": 8.83800692030984e-06, "loss": 0.5333, "step": 4532 }, { "epoch": 0.22929839647933634, "grad_norm": 9.386963844299316, "learning_rate": 8.83749245329202e-06, "loss": 0.4519, "step": 4533 }, { "epoch": 0.22934898072740148, "grad_norm": 12.018856048583984, "learning_rate": 8.83697788739129e-06, "loss": 0.4098, "step": 4534 }, { "epoch": 0.22939956497546665, "grad_norm": 8.178521156311035, "learning_rate": 8.836463222620905e-06, "loss": 0.4783, "step": 4535 }, { "epoch": 0.2294501492235318, "grad_norm": 7.047786235809326, "learning_rate": 8.835948458994133e-06, "loss": 0.4448, "step": 4536 }, { "epoch": 0.22950073347159694, "grad_norm": 11.018999099731445, "learning_rate": 8.835433596524231e-06, "loss": 0.436, "step": 4537 }, { "epoch": 0.22955131771966208, "grad_norm": 12.689018249511719, "learning_rate": 8.83491863522447e-06, "loss": 0.4033, "step": 4538 }, { "epoch": 0.22960190196772726, "grad_norm": 8.816472053527832, "learning_rate": 8.83440357510812e-06, "loss": 0.4999, "step": 4539 }, { "epoch": 0.2296524862157924, "grad_norm": 6.511161804199219, "learning_rate": 8.83388841618845e-06, "loss": 0.4568, "step": 4540 }, { "epoch": 0.22970307046385754, "grad_norm": 7.97078800201416, "learning_rate": 8.833373158478738e-06, "loss": 0.4823, "step": 4541 }, { "epoch": 0.22975365471192272, "grad_norm": 6.295597076416016, "learning_rate": 8.832857801992257e-06, "loss": 0.3258, "step": 4542 }, { "epoch": 0.22980423895998786, "grad_norm": 14.0820951461792, "learning_rate": 8.832342346742288e-06, "loss": 0.5423, "step": 4543 }, { "epoch": 0.229854823208053, "grad_norm": 8.587796211242676, "learning_rate": 8.831826792742115e-06, "loss": 0.4473, "step": 4544 }, { "epoch": 0.22990540745611818, "grad_norm": 8.64513874053955, "learning_rate": 8.831311140005022e-06, "loss": 0.4795, "step": 4545 }, { "epoch": 0.22995599170418332, "grad_norm": 10.080338478088379, "learning_rate": 8.830795388544292e-06, "loss": 0.4828, "step": 4546 }, { "epoch": 0.23000657595224847, "grad_norm": 15.174569129943848, "learning_rate": 8.83027953837322e-06, "loss": 0.5516, "step": 4547 }, { "epoch": 0.2300571602003136, "grad_norm": 8.979662895202637, "learning_rate": 8.829763589505096e-06, "loss": 0.4754, "step": 4548 }, { "epoch": 0.23010774444837878, "grad_norm": 9.384176254272461, "learning_rate": 8.829247541953212e-06, "loss": 0.4857, "step": 4549 }, { "epoch": 0.23015832869644393, "grad_norm": 6.564158916473389, "learning_rate": 8.82873139573087e-06, "loss": 0.4195, "step": 4550 }, { "epoch": 0.23020891294450907, "grad_norm": 7.562704086303711, "learning_rate": 8.828215150851368e-06, "loss": 0.4733, "step": 4551 }, { "epoch": 0.23025949719257424, "grad_norm": 5.994026184082031, "learning_rate": 8.827698807328008e-06, "loss": 0.4843, "step": 4552 }, { "epoch": 0.23031008144063939, "grad_norm": 9.758024215698242, "learning_rate": 8.827182365174094e-06, "loss": 0.4171, "step": 4553 }, { "epoch": 0.23036066568870453, "grad_norm": 11.2714204788208, "learning_rate": 8.826665824402936e-06, "loss": 0.6207, "step": 4554 }, { "epoch": 0.2304112499367697, "grad_norm": 5.569642066955566, "learning_rate": 8.826149185027843e-06, "loss": 0.4693, "step": 4555 }, { "epoch": 0.23046183418483485, "grad_norm": 8.800989151000977, "learning_rate": 8.825632447062128e-06, "loss": 0.4732, "step": 4556 }, { "epoch": 0.2305124184329, "grad_norm": 9.198042869567871, "learning_rate": 8.8251156105191e-06, "loss": 0.4597, "step": 4557 }, { "epoch": 0.23056300268096513, "grad_norm": 5.418159008026123, "learning_rate": 8.824598675412088e-06, "loss": 0.3648, "step": 4558 }, { "epoch": 0.2306135869290303, "grad_norm": 8.108598709106445, "learning_rate": 8.824081641754403e-06, "loss": 0.5065, "step": 4559 }, { "epoch": 0.23066417117709545, "grad_norm": 9.67380142211914, "learning_rate": 8.823564509559373e-06, "loss": 0.3746, "step": 4560 }, { "epoch": 0.2307147554251606, "grad_norm": 6.987807750701904, "learning_rate": 8.823047278840317e-06, "loss": 0.4274, "step": 4561 }, { "epoch": 0.23076533967322577, "grad_norm": 8.33023738861084, "learning_rate": 8.822529949610572e-06, "loss": 0.4964, "step": 4562 }, { "epoch": 0.2308159239212909, "grad_norm": 10.019350051879883, "learning_rate": 8.82201252188346e-06, "loss": 0.5006, "step": 4563 }, { "epoch": 0.23086650816935606, "grad_norm": 12.874186515808105, "learning_rate": 8.821494995672317e-06, "loss": 0.5763, "step": 4564 }, { "epoch": 0.23091709241742123, "grad_norm": 11.259295463562012, "learning_rate": 8.82097737099048e-06, "loss": 0.6435, "step": 4565 }, { "epoch": 0.23096767666548637, "grad_norm": 9.601990699768066, "learning_rate": 8.820459647851282e-06, "loss": 0.5538, "step": 4566 }, { "epoch": 0.23101826091355152, "grad_norm": 5.589823246002197, "learning_rate": 8.81994182626807e-06, "loss": 0.4009, "step": 4567 }, { "epoch": 0.23106884516161666, "grad_norm": 12.004055976867676, "learning_rate": 8.819423906254183e-06, "loss": 0.4282, "step": 4568 }, { "epoch": 0.23111942940968183, "grad_norm": 16.456945419311523, "learning_rate": 8.818905887822969e-06, "loss": 0.4525, "step": 4569 }, { "epoch": 0.23117001365774698, "grad_norm": 13.688759803771973, "learning_rate": 8.81838777098777e-06, "loss": 0.5121, "step": 4570 }, { "epoch": 0.23122059790581212, "grad_norm": 7.256329536437988, "learning_rate": 8.817869555761945e-06, "loss": 0.5178, "step": 4571 }, { "epoch": 0.2312711821538773, "grad_norm": 6.444263458251953, "learning_rate": 8.817351242158843e-06, "loss": 0.3879, "step": 4572 }, { "epoch": 0.23132176640194244, "grad_norm": 9.879941940307617, "learning_rate": 8.816832830191817e-06, "loss": 0.4749, "step": 4573 }, { "epoch": 0.23137235065000758, "grad_norm": 7.600144863128662, "learning_rate": 8.816314319874231e-06, "loss": 0.4946, "step": 4574 }, { "epoch": 0.23142293489807275, "grad_norm": 9.884769439697266, "learning_rate": 8.815795711219441e-06, "loss": 0.5318, "step": 4575 }, { "epoch": 0.2314735191461379, "grad_norm": 8.777737617492676, "learning_rate": 8.815277004240815e-06, "loss": 0.4485, "step": 4576 }, { "epoch": 0.23152410339420304, "grad_norm": 13.5914888381958, "learning_rate": 8.814758198951714e-06, "loss": 0.5582, "step": 4577 }, { "epoch": 0.23157468764226818, "grad_norm": 12.428236961364746, "learning_rate": 8.814239295365508e-06, "loss": 0.5688, "step": 4578 }, { "epoch": 0.23162527189033336, "grad_norm": 7.825093746185303, "learning_rate": 8.813720293495569e-06, "loss": 0.5385, "step": 4579 }, { "epoch": 0.2316758561383985, "grad_norm": 8.82287311553955, "learning_rate": 8.81320119335527e-06, "loss": 0.5166, "step": 4580 }, { "epoch": 0.23172644038646364, "grad_norm": 7.998969554901123, "learning_rate": 8.812681994957986e-06, "loss": 0.5023, "step": 4581 }, { "epoch": 0.23177702463452882, "grad_norm": 5.144117832183838, "learning_rate": 8.812162698317097e-06, "loss": 0.4693, "step": 4582 }, { "epoch": 0.23182760888259396, "grad_norm": 5.514873504638672, "learning_rate": 8.811643303445982e-06, "loss": 0.45, "step": 4583 }, { "epoch": 0.2318781931306591, "grad_norm": 16.828676223754883, "learning_rate": 8.811123810358027e-06, "loss": 0.5377, "step": 4584 }, { "epoch": 0.23192877737872428, "grad_norm": 5.675995826721191, "learning_rate": 8.810604219066616e-06, "loss": 0.4309, "step": 4585 }, { "epoch": 0.23197936162678942, "grad_norm": 6.841122627258301, "learning_rate": 8.81008452958514e-06, "loss": 0.47, "step": 4586 }, { "epoch": 0.23202994587485457, "grad_norm": 6.467083930969238, "learning_rate": 8.809564741926986e-06, "loss": 0.5024, "step": 4587 }, { "epoch": 0.2320805301229197, "grad_norm": 5.024454116821289, "learning_rate": 8.809044856105551e-06, "loss": 0.425, "step": 4588 }, { "epoch": 0.23213111437098488, "grad_norm": 4.807401657104492, "learning_rate": 8.80852487213423e-06, "loss": 0.4922, "step": 4589 }, { "epoch": 0.23218169861905003, "grad_norm": 7.992223262786865, "learning_rate": 8.808004790026422e-06, "loss": 0.3917, "step": 4590 }, { "epoch": 0.23223228286711517, "grad_norm": 6.699181079864502, "learning_rate": 8.807484609795532e-06, "loss": 0.4316, "step": 4591 }, { "epoch": 0.23228286711518034, "grad_norm": 6.788060188293457, "learning_rate": 8.806964331454955e-06, "loss": 0.5175, "step": 4592 }, { "epoch": 0.23233345136324549, "grad_norm": 8.12827205657959, "learning_rate": 8.806443955018107e-06, "loss": 0.4193, "step": 4593 }, { "epoch": 0.23238403561131063, "grad_norm": 10.199098587036133, "learning_rate": 8.80592348049839e-06, "loss": 0.5514, "step": 4594 }, { "epoch": 0.2324346198593758, "grad_norm": 6.017265796661377, "learning_rate": 8.805402907909218e-06, "loss": 0.4762, "step": 4595 }, { "epoch": 0.23248520410744095, "grad_norm": 11.870481491088867, "learning_rate": 8.804882237264005e-06, "loss": 0.5647, "step": 4596 }, { "epoch": 0.2325357883555061, "grad_norm": 8.743090629577637, "learning_rate": 8.804361468576167e-06, "loss": 0.4065, "step": 4597 }, { "epoch": 0.23258637260357123, "grad_norm": 6.557283878326416, "learning_rate": 8.803840601859121e-06, "loss": 0.432, "step": 4598 }, { "epoch": 0.2326369568516364, "grad_norm": 12.08361530303955, "learning_rate": 8.803319637126293e-06, "loss": 0.5295, "step": 4599 }, { "epoch": 0.23268754109970155, "grad_norm": 9.244559288024902, "learning_rate": 8.802798574391105e-06, "loss": 0.4165, "step": 4600 }, { "epoch": 0.2327381253477667, "grad_norm": 10.524809837341309, "learning_rate": 8.802277413666982e-06, "loss": 0.4707, "step": 4601 }, { "epoch": 0.23278870959583187, "grad_norm": 5.154772758483887, "learning_rate": 8.801756154967353e-06, "loss": 0.4122, "step": 4602 }, { "epoch": 0.232839293843897, "grad_norm": 6.311365127563477, "learning_rate": 8.801234798305652e-06, "loss": 0.4549, "step": 4603 }, { "epoch": 0.23288987809196215, "grad_norm": 11.44355297088623, "learning_rate": 8.80071334369531e-06, "loss": 0.5192, "step": 4604 }, { "epoch": 0.23294046234002733, "grad_norm": 7.257091522216797, "learning_rate": 8.800191791149769e-06, "loss": 0.5575, "step": 4605 }, { "epoch": 0.23299104658809247, "grad_norm": 6.181006908416748, "learning_rate": 8.799670140682462e-06, "loss": 0.2985, "step": 4606 }, { "epoch": 0.23304163083615761, "grad_norm": 7.838433265686035, "learning_rate": 8.799148392306833e-06, "loss": 0.5079, "step": 4607 }, { "epoch": 0.2330922150842228, "grad_norm": 6.888723850250244, "learning_rate": 8.798626546036327e-06, "loss": 0.423, "step": 4608 }, { "epoch": 0.23314279933228793, "grad_norm": 11.738734245300293, "learning_rate": 8.79810460188439e-06, "loss": 0.5488, "step": 4609 }, { "epoch": 0.23319338358035308, "grad_norm": 10.436470985412598, "learning_rate": 8.797582559864471e-06, "loss": 0.5426, "step": 4610 }, { "epoch": 0.23324396782841822, "grad_norm": 5.5679216384887695, "learning_rate": 8.797060419990023e-06, "loss": 0.4316, "step": 4611 }, { "epoch": 0.2332945520764834, "grad_norm": 9.201582908630371, "learning_rate": 8.796538182274499e-06, "loss": 0.5741, "step": 4612 }, { "epoch": 0.23334513632454854, "grad_norm": 12.182738304138184, "learning_rate": 8.796015846731356e-06, "loss": 0.6514, "step": 4613 }, { "epoch": 0.23339572057261368, "grad_norm": 9.645469665527344, "learning_rate": 8.795493413374052e-06, "loss": 0.456, "step": 4614 }, { "epoch": 0.23344630482067885, "grad_norm": 8.48887825012207, "learning_rate": 8.794970882216052e-06, "loss": 0.4239, "step": 4615 }, { "epoch": 0.233496889068744, "grad_norm": 8.525306701660156, "learning_rate": 8.79444825327082e-06, "loss": 0.4375, "step": 4616 }, { "epoch": 0.23354747331680914, "grad_norm": 8.198470115661621, "learning_rate": 8.79392552655182e-06, "loss": 0.4464, "step": 4617 }, { "epoch": 0.2335980575648743, "grad_norm": 9.90870475769043, "learning_rate": 8.793402702072523e-06, "loss": 0.5041, "step": 4618 }, { "epoch": 0.23364864181293946, "grad_norm": 6.743056297302246, "learning_rate": 8.792879779846402e-06, "loss": 0.522, "step": 4619 }, { "epoch": 0.2336992260610046, "grad_norm": 7.72516393661499, "learning_rate": 8.79235675988693e-06, "loss": 0.4368, "step": 4620 }, { "epoch": 0.23374981030906974, "grad_norm": 4.929906368255615, "learning_rate": 8.791833642207583e-06, "loss": 0.428, "step": 4621 }, { "epoch": 0.23380039455713492, "grad_norm": 10.047039031982422, "learning_rate": 8.791310426821843e-06, "loss": 0.4238, "step": 4622 }, { "epoch": 0.23385097880520006, "grad_norm": 5.953879356384277, "learning_rate": 8.790787113743191e-06, "loss": 0.5106, "step": 4623 }, { "epoch": 0.2339015630532652, "grad_norm": 5.872336387634277, "learning_rate": 8.790263702985111e-06, "loss": 0.4421, "step": 4624 }, { "epoch": 0.23395214730133038, "grad_norm": 8.546175956726074, "learning_rate": 8.78974019456109e-06, "loss": 0.4562, "step": 4625 }, { "epoch": 0.23400273154939552, "grad_norm": 8.678784370422363, "learning_rate": 8.78921658848462e-06, "loss": 0.4898, "step": 4626 }, { "epoch": 0.23405331579746066, "grad_norm": 7.839034080505371, "learning_rate": 8.788692884769187e-06, "loss": 0.4297, "step": 4627 }, { "epoch": 0.23410390004552584, "grad_norm": 7.307806968688965, "learning_rate": 8.788169083428293e-06, "loss": 0.4539, "step": 4628 }, { "epoch": 0.23415448429359098, "grad_norm": 8.686964988708496, "learning_rate": 8.787645184475431e-06, "loss": 0.4708, "step": 4629 }, { "epoch": 0.23420506854165613, "grad_norm": 6.222655296325684, "learning_rate": 8.787121187924102e-06, "loss": 0.3515, "step": 4630 }, { "epoch": 0.23425565278972127, "grad_norm": 11.850990295410156, "learning_rate": 8.786597093787806e-06, "loss": 0.4052, "step": 4631 }, { "epoch": 0.23430623703778644, "grad_norm": 7.29567813873291, "learning_rate": 8.78607290208005e-06, "loss": 0.5221, "step": 4632 }, { "epoch": 0.23435682128585159, "grad_norm": 15.661820411682129, "learning_rate": 8.785548612814338e-06, "loss": 0.3953, "step": 4633 }, { "epoch": 0.23440740553391673, "grad_norm": 6.963427543640137, "learning_rate": 8.785024226004185e-06, "loss": 0.3712, "step": 4634 }, { "epoch": 0.2344579897819819, "grad_norm": 9.764535903930664, "learning_rate": 8.784499741663096e-06, "loss": 0.4909, "step": 4635 }, { "epoch": 0.23450857403004705, "grad_norm": 13.320724487304688, "learning_rate": 8.783975159804594e-06, "loss": 0.6153, "step": 4636 }, { "epoch": 0.2345591582781122, "grad_norm": 8.874939918518066, "learning_rate": 8.78345048044219e-06, "loss": 0.5193, "step": 4637 }, { "epoch": 0.23460974252617736, "grad_norm": 8.126392364501953, "learning_rate": 8.782925703589405e-06, "loss": 0.3594, "step": 4638 }, { "epoch": 0.2346603267742425, "grad_norm": 10.607478141784668, "learning_rate": 8.782400829259762e-06, "loss": 0.5173, "step": 4639 }, { "epoch": 0.23471091102230765, "grad_norm": 7.117203235626221, "learning_rate": 8.781875857466789e-06, "loss": 0.5124, "step": 4640 }, { "epoch": 0.2347614952703728, "grad_norm": 6.966404914855957, "learning_rate": 8.781350788224006e-06, "loss": 0.3842, "step": 4641 }, { "epoch": 0.23481207951843797, "grad_norm": 12.113780975341797, "learning_rate": 8.78082562154495e-06, "loss": 0.5568, "step": 4642 }, { "epoch": 0.2348626637665031, "grad_norm": 6.244749546051025, "learning_rate": 8.780300357443146e-06, "loss": 0.4525, "step": 4643 }, { "epoch": 0.23491324801456825, "grad_norm": 10.947423934936523, "learning_rate": 8.779774995932135e-06, "loss": 0.3729, "step": 4644 }, { "epoch": 0.23496383226263343, "grad_norm": 10.817037582397461, "learning_rate": 8.779249537025451e-06, "loss": 0.5185, "step": 4645 }, { "epoch": 0.23501441651069857, "grad_norm": 8.230255126953125, "learning_rate": 8.778723980736636e-06, "loss": 0.4582, "step": 4646 }, { "epoch": 0.23506500075876371, "grad_norm": 14.696187973022461, "learning_rate": 8.778198327079231e-06, "loss": 0.5126, "step": 4647 }, { "epoch": 0.2351155850068289, "grad_norm": 9.938519477844238, "learning_rate": 8.77767257606678e-06, "loss": 0.4307, "step": 4648 }, { "epoch": 0.23516616925489403, "grad_norm": 6.557076930999756, "learning_rate": 8.777146727712832e-06, "loss": 0.4027, "step": 4649 }, { "epoch": 0.23521675350295917, "grad_norm": 8.695418357849121, "learning_rate": 8.776620782030936e-06, "loss": 0.4174, "step": 4650 }, { "epoch": 0.23526733775102432, "grad_norm": 7.646541118621826, "learning_rate": 8.776094739034645e-06, "loss": 0.4335, "step": 4651 }, { "epoch": 0.2353179219990895, "grad_norm": 6.656401634216309, "learning_rate": 8.775568598737513e-06, "loss": 0.4991, "step": 4652 }, { "epoch": 0.23536850624715464, "grad_norm": 8.836849212646484, "learning_rate": 8.775042361153097e-06, "loss": 0.496, "step": 4653 }, { "epoch": 0.23541909049521978, "grad_norm": 6.521297931671143, "learning_rate": 8.77451602629496e-06, "loss": 0.4314, "step": 4654 }, { "epoch": 0.23546967474328495, "grad_norm": 7.706422805786133, "learning_rate": 8.77398959417666e-06, "loss": 0.4945, "step": 4655 }, { "epoch": 0.2355202589913501, "grad_norm": 14.269421577453613, "learning_rate": 8.773463064811763e-06, "loss": 0.4835, "step": 4656 }, { "epoch": 0.23557084323941524, "grad_norm": 13.840683937072754, "learning_rate": 8.772936438213837e-06, "loss": 0.4787, "step": 4657 }, { "epoch": 0.2356214274874804, "grad_norm": 14.17702865600586, "learning_rate": 8.772409714396455e-06, "loss": 0.458, "step": 4658 }, { "epoch": 0.23567201173554556, "grad_norm": 8.583686828613281, "learning_rate": 8.771882893373185e-06, "loss": 0.5031, "step": 4659 }, { "epoch": 0.2357225959836107, "grad_norm": 5.636236190795898, "learning_rate": 8.771355975157603e-06, "loss": 0.46, "step": 4660 }, { "epoch": 0.23577318023167584, "grad_norm": 7.740855693817139, "learning_rate": 8.770828959763287e-06, "loss": 0.4154, "step": 4661 }, { "epoch": 0.23582376447974102, "grad_norm": 8.205023765563965, "learning_rate": 8.770301847203817e-06, "loss": 0.4359, "step": 4662 }, { "epoch": 0.23587434872780616, "grad_norm": 5.592391014099121, "learning_rate": 8.769774637492775e-06, "loss": 0.5217, "step": 4663 }, { "epoch": 0.2359249329758713, "grad_norm": 8.518848419189453, "learning_rate": 8.769247330643747e-06, "loss": 0.4396, "step": 4664 }, { "epoch": 0.23597551722393648, "grad_norm": 6.649228572845459, "learning_rate": 8.768719926670319e-06, "loss": 0.4901, "step": 4665 }, { "epoch": 0.23602610147200162, "grad_norm": 5.312989711761475, "learning_rate": 8.768192425586082e-06, "loss": 0.4139, "step": 4666 }, { "epoch": 0.23607668572006676, "grad_norm": 7.883427619934082, "learning_rate": 8.767664827404628e-06, "loss": 0.5158, "step": 4667 }, { "epoch": 0.23612726996813194, "grad_norm": 6.8878889083862305, "learning_rate": 8.76713713213955e-06, "loss": 0.4965, "step": 4668 }, { "epoch": 0.23617785421619708, "grad_norm": 11.07699203491211, "learning_rate": 8.766609339804448e-06, "loss": 0.5062, "step": 4669 }, { "epoch": 0.23622843846426222, "grad_norm": 6.637448787689209, "learning_rate": 8.766081450412923e-06, "loss": 0.4678, "step": 4670 }, { "epoch": 0.23627902271232737, "grad_norm": 7.466274261474609, "learning_rate": 8.765553463978573e-06, "loss": 0.4707, "step": 4671 }, { "epoch": 0.23632960696039254, "grad_norm": 11.76791763305664, "learning_rate": 8.765025380515006e-06, "loss": 0.4325, "step": 4672 }, { "epoch": 0.23638019120845769, "grad_norm": 9.76258659362793, "learning_rate": 8.76449720003583e-06, "loss": 0.4416, "step": 4673 }, { "epoch": 0.23643077545652283, "grad_norm": 9.041799545288086, "learning_rate": 8.763968922554652e-06, "loss": 0.5032, "step": 4674 }, { "epoch": 0.236481359704588, "grad_norm": 11.813005447387695, "learning_rate": 8.76344054808509e-06, "loss": 0.359, "step": 4675 }, { "epoch": 0.23653194395265315, "grad_norm": 12.473986625671387, "learning_rate": 8.762912076640752e-06, "loss": 0.5234, "step": 4676 }, { "epoch": 0.2365825282007183, "grad_norm": 6.857988357543945, "learning_rate": 8.762383508235262e-06, "loss": 0.4453, "step": 4677 }, { "epoch": 0.23663311244878346, "grad_norm": 9.840937614440918, "learning_rate": 8.761854842882234e-06, "loss": 0.6742, "step": 4678 }, { "epoch": 0.2366836966968486, "grad_norm": 7.328730583190918, "learning_rate": 8.761326080595293e-06, "loss": 0.4556, "step": 4679 }, { "epoch": 0.23673428094491375, "grad_norm": 13.307848930358887, "learning_rate": 8.760797221388066e-06, "loss": 0.4318, "step": 4680 }, { "epoch": 0.2367848651929789, "grad_norm": 5.091266632080078, "learning_rate": 8.760268265274177e-06, "loss": 0.3886, "step": 4681 }, { "epoch": 0.23683544944104407, "grad_norm": 5.0212249755859375, "learning_rate": 8.759739212267258e-06, "loss": 0.3159, "step": 4682 }, { "epoch": 0.2368860336891092, "grad_norm": 8.342974662780762, "learning_rate": 8.759210062380939e-06, "loss": 0.5673, "step": 4683 }, { "epoch": 0.23693661793717435, "grad_norm": 9.320659637451172, "learning_rate": 8.75868081562886e-06, "loss": 0.3224, "step": 4684 }, { "epoch": 0.23698720218523953, "grad_norm": 12.102004051208496, "learning_rate": 8.758151472024652e-06, "loss": 0.5589, "step": 4685 }, { "epoch": 0.23703778643330467, "grad_norm": 6.414139747619629, "learning_rate": 8.757622031581958e-06, "loss": 0.3565, "step": 4686 }, { "epoch": 0.23708837068136981, "grad_norm": 7.173974990844727, "learning_rate": 8.757092494314423e-06, "loss": 0.3722, "step": 4687 }, { "epoch": 0.237138954929435, "grad_norm": 5.506181716918945, "learning_rate": 8.756562860235687e-06, "loss": 0.3644, "step": 4688 }, { "epoch": 0.23718953917750013, "grad_norm": 7.055268287658691, "learning_rate": 8.756033129359401e-06, "loss": 0.4734, "step": 4689 }, { "epoch": 0.23724012342556527, "grad_norm": 8.081241607666016, "learning_rate": 8.755503301699213e-06, "loss": 0.4574, "step": 4690 }, { "epoch": 0.23729070767363042, "grad_norm": 5.768343448638916, "learning_rate": 8.754973377268776e-06, "loss": 0.4123, "step": 4691 }, { "epoch": 0.2373412919216956, "grad_norm": 8.476634979248047, "learning_rate": 8.754443356081746e-06, "loss": 0.3868, "step": 4692 }, { "epoch": 0.23739187616976073, "grad_norm": 4.890838146209717, "learning_rate": 8.753913238151777e-06, "loss": 0.3665, "step": 4693 }, { "epoch": 0.23744246041782588, "grad_norm": 12.7626371383667, "learning_rate": 8.753383023492532e-06, "loss": 0.5429, "step": 4694 }, { "epoch": 0.23749304466589105, "grad_norm": 13.258387565612793, "learning_rate": 8.752852712117672e-06, "loss": 0.5924, "step": 4695 }, { "epoch": 0.2375436289139562, "grad_norm": 6.962084770202637, "learning_rate": 8.752322304040863e-06, "loss": 0.3853, "step": 4696 }, { "epoch": 0.23759421316202134, "grad_norm": 6.026009559631348, "learning_rate": 8.751791799275773e-06, "loss": 0.4642, "step": 4697 }, { "epoch": 0.2376447974100865, "grad_norm": 9.392342567443848, "learning_rate": 8.751261197836068e-06, "loss": 0.3885, "step": 4698 }, { "epoch": 0.23769538165815166, "grad_norm": 8.38981819152832, "learning_rate": 8.750730499735423e-06, "loss": 0.3894, "step": 4699 }, { "epoch": 0.2377459659062168, "grad_norm": 9.715864181518555, "learning_rate": 8.750199704987515e-06, "loss": 0.654, "step": 4700 }, { "epoch": 0.23779655015428194, "grad_norm": 8.789055824279785, "learning_rate": 8.749668813606016e-06, "loss": 0.4527, "step": 4701 }, { "epoch": 0.23784713440234712, "grad_norm": 7.1820149421691895, "learning_rate": 8.74913782560461e-06, "loss": 0.5145, "step": 4702 }, { "epoch": 0.23789771865041226, "grad_norm": 8.937294960021973, "learning_rate": 8.748606740996976e-06, "loss": 0.5135, "step": 4703 }, { "epoch": 0.2379483028984774, "grad_norm": 15.636924743652344, "learning_rate": 8.748075559796802e-06, "loss": 0.5187, "step": 4704 }, { "epoch": 0.23799888714654258, "grad_norm": 5.373327255249023, "learning_rate": 8.747544282017774e-06, "loss": 0.4007, "step": 4705 }, { "epoch": 0.23804947139460772, "grad_norm": 7.77722692489624, "learning_rate": 8.747012907673583e-06, "loss": 0.4988, "step": 4706 }, { "epoch": 0.23810005564267286, "grad_norm": 7.746901035308838, "learning_rate": 8.746481436777918e-06, "loss": 0.4251, "step": 4707 }, { "epoch": 0.23815063989073804, "grad_norm": 5.968937873840332, "learning_rate": 8.745949869344477e-06, "loss": 0.4071, "step": 4708 }, { "epoch": 0.23820122413880318, "grad_norm": 4.20145320892334, "learning_rate": 8.745418205386955e-06, "loss": 0.4198, "step": 4709 }, { "epoch": 0.23825180838686832, "grad_norm": 8.299509048461914, "learning_rate": 8.744886444919053e-06, "loss": 0.4912, "step": 4710 }, { "epoch": 0.23830239263493347, "grad_norm": 7.83273458480835, "learning_rate": 8.744354587954473e-06, "loss": 0.5232, "step": 4711 }, { "epoch": 0.23835297688299864, "grad_norm": 7.42100715637207, "learning_rate": 8.743822634506918e-06, "loss": 0.5361, "step": 4712 }, { "epoch": 0.23840356113106378, "grad_norm": 9.054523468017578, "learning_rate": 8.743290584590098e-06, "loss": 0.5155, "step": 4713 }, { "epoch": 0.23845414537912893, "grad_norm": 8.5525484085083, "learning_rate": 8.74275843821772e-06, "loss": 0.5717, "step": 4714 }, { "epoch": 0.2385047296271941, "grad_norm": 6.229284286499023, "learning_rate": 8.742226195403498e-06, "loss": 0.5034, "step": 4715 }, { "epoch": 0.23855531387525925, "grad_norm": 6.962900161743164, "learning_rate": 8.741693856161146e-06, "loss": 0.3845, "step": 4716 }, { "epoch": 0.2386058981233244, "grad_norm": 13.350899696350098, "learning_rate": 8.74116142050438e-06, "loss": 0.5888, "step": 4717 }, { "epoch": 0.23865648237138956, "grad_norm": 8.709590911865234, "learning_rate": 8.740628888446921e-06, "loss": 0.4645, "step": 4718 }, { "epoch": 0.2387070666194547, "grad_norm": 8.380260467529297, "learning_rate": 8.740096260002491e-06, "loss": 0.4549, "step": 4719 }, { "epoch": 0.23875765086751985, "grad_norm": 8.200057983398438, "learning_rate": 8.739563535184816e-06, "loss": 0.4158, "step": 4720 }, { "epoch": 0.238808235115585, "grad_norm": 7.965893268585205, "learning_rate": 8.73903071400762e-06, "loss": 0.436, "step": 4721 }, { "epoch": 0.23885881936365017, "grad_norm": 6.330131530761719, "learning_rate": 8.738497796484631e-06, "loss": 0.4913, "step": 4722 }, { "epoch": 0.2389094036117153, "grad_norm": 5.799527645111084, "learning_rate": 8.737964782629586e-06, "loss": 0.4573, "step": 4723 }, { "epoch": 0.23895998785978045, "grad_norm": 6.411377429962158, "learning_rate": 8.737431672456218e-06, "loss": 0.4646, "step": 4724 }, { "epoch": 0.23901057210784563, "grad_norm": 5.9344377517700195, "learning_rate": 8.736898465978263e-06, "loss": 0.4372, "step": 4725 }, { "epoch": 0.23906115635591077, "grad_norm": 6.92165994644165, "learning_rate": 8.73636516320946e-06, "loss": 0.4341, "step": 4726 }, { "epoch": 0.23911174060397591, "grad_norm": 8.567706108093262, "learning_rate": 8.735831764163552e-06, "loss": 0.3567, "step": 4727 }, { "epoch": 0.23916232485204109, "grad_norm": 7.034980773925781, "learning_rate": 8.735298268854282e-06, "loss": 0.5383, "step": 4728 }, { "epoch": 0.23921290910010623, "grad_norm": 8.123833656311035, "learning_rate": 8.734764677295398e-06, "loss": 0.4021, "step": 4729 }, { "epoch": 0.23926349334817137, "grad_norm": 6.196828365325928, "learning_rate": 8.734230989500649e-06, "loss": 0.4097, "step": 4730 }, { "epoch": 0.23931407759623652, "grad_norm": 6.690840244293213, "learning_rate": 8.733697205483789e-06, "loss": 0.4307, "step": 4731 }, { "epoch": 0.2393646618443017, "grad_norm": 14.387883186340332, "learning_rate": 8.73316332525857e-06, "loss": 0.6295, "step": 4732 }, { "epoch": 0.23941524609236683, "grad_norm": 7.42836856842041, "learning_rate": 8.732629348838748e-06, "loss": 0.3753, "step": 4733 }, { "epoch": 0.23946583034043198, "grad_norm": 14.721711158752441, "learning_rate": 8.732095276238083e-06, "loss": 0.6637, "step": 4734 }, { "epoch": 0.23951641458849715, "grad_norm": 9.709293365478516, "learning_rate": 8.731561107470339e-06, "loss": 0.6199, "step": 4735 }, { "epoch": 0.2395669988365623, "grad_norm": 9.850017547607422, "learning_rate": 8.731026842549278e-06, "loss": 0.4158, "step": 4736 }, { "epoch": 0.23961758308462744, "grad_norm": 8.276859283447266, "learning_rate": 8.730492481488666e-06, "loss": 0.3963, "step": 4737 }, { "epoch": 0.2396681673326926, "grad_norm": 10.116823196411133, "learning_rate": 8.729958024302274e-06, "loss": 0.4427, "step": 4738 }, { "epoch": 0.23971875158075776, "grad_norm": 7.031273365020752, "learning_rate": 8.729423471003873e-06, "loss": 0.4678, "step": 4739 }, { "epoch": 0.2397693358288229, "grad_norm": 6.74870491027832, "learning_rate": 8.728888821607238e-06, "loss": 0.3945, "step": 4740 }, { "epoch": 0.23981992007688804, "grad_norm": 4.948624610900879, "learning_rate": 8.728354076126143e-06, "loss": 0.4923, "step": 4741 }, { "epoch": 0.23987050432495322, "grad_norm": 8.469911575317383, "learning_rate": 8.72781923457437e-06, "loss": 0.503, "step": 4742 }, { "epoch": 0.23992108857301836, "grad_norm": 6.072285175323486, "learning_rate": 8.7272842969657e-06, "loss": 0.3594, "step": 4743 }, { "epoch": 0.2399716728210835, "grad_norm": 11.325740814208984, "learning_rate": 8.726749263313917e-06, "loss": 0.4666, "step": 4744 }, { "epoch": 0.24002225706914868, "grad_norm": 12.913395881652832, "learning_rate": 8.726214133632805e-06, "loss": 0.5217, "step": 4745 }, { "epoch": 0.24007284131721382, "grad_norm": 6.962669372558594, "learning_rate": 8.725678907936158e-06, "loss": 0.4986, "step": 4746 }, { "epoch": 0.24012342556527896, "grad_norm": 7.548620700836182, "learning_rate": 8.725143586237762e-06, "loss": 0.4585, "step": 4747 }, { "epoch": 0.24017400981334414, "grad_norm": 12.180323600769043, "learning_rate": 8.724608168551414e-06, "loss": 0.5302, "step": 4748 }, { "epoch": 0.24022459406140928, "grad_norm": 7.589359283447266, "learning_rate": 8.724072654890911e-06, "loss": 0.4493, "step": 4749 }, { "epoch": 0.24027517830947442, "grad_norm": 13.934592247009277, "learning_rate": 8.72353704527005e-06, "loss": 0.3209, "step": 4750 }, { "epoch": 0.24032576255753957, "grad_norm": 10.656230926513672, "learning_rate": 8.723001339702633e-06, "loss": 0.423, "step": 4751 }, { "epoch": 0.24037634680560474, "grad_norm": 5.624079704284668, "learning_rate": 8.722465538202465e-06, "loss": 0.4349, "step": 4752 }, { "epoch": 0.24042693105366988, "grad_norm": 6.203988075256348, "learning_rate": 8.72192964078335e-06, "loss": 0.3294, "step": 4753 }, { "epoch": 0.24047751530173503, "grad_norm": 7.486274242401123, "learning_rate": 8.721393647459098e-06, "loss": 0.4864, "step": 4754 }, { "epoch": 0.2405280995498002, "grad_norm": 6.444990634918213, "learning_rate": 8.720857558243522e-06, "loss": 0.4177, "step": 4755 }, { "epoch": 0.24057868379786534, "grad_norm": 16.33336067199707, "learning_rate": 8.720321373150434e-06, "loss": 0.4777, "step": 4756 }, { "epoch": 0.2406292680459305, "grad_norm": 7.268613815307617, "learning_rate": 8.71978509219365e-06, "loss": 0.451, "step": 4757 }, { "epoch": 0.24067985229399566, "grad_norm": 6.909975528717041, "learning_rate": 8.719248715386987e-06, "loss": 0.4837, "step": 4758 }, { "epoch": 0.2407304365420608, "grad_norm": 8.176836013793945, "learning_rate": 8.718712242744269e-06, "loss": 0.3903, "step": 4759 }, { "epoch": 0.24078102079012595, "grad_norm": 5.932125091552734, "learning_rate": 8.71817567427932e-06, "loss": 0.4398, "step": 4760 }, { "epoch": 0.24083160503819112, "grad_norm": 10.924095153808594, "learning_rate": 8.717639010005964e-06, "loss": 0.4004, "step": 4761 }, { "epoch": 0.24088218928625627, "grad_norm": 7.983185768127441, "learning_rate": 8.71710224993803e-06, "loss": 0.3457, "step": 4762 }, { "epoch": 0.2409327735343214, "grad_norm": 7.417212963104248, "learning_rate": 8.71656539408935e-06, "loss": 0.4315, "step": 4763 }, { "epoch": 0.24098335778238655, "grad_norm": 8.520715713500977, "learning_rate": 8.716028442473756e-06, "loss": 0.3421, "step": 4764 }, { "epoch": 0.24103394203045173, "grad_norm": 7.337554454803467, "learning_rate": 8.715491395105085e-06, "loss": 0.4858, "step": 4765 }, { "epoch": 0.24108452627851687, "grad_norm": 8.335981369018555, "learning_rate": 8.714954251997175e-06, "loss": 0.4459, "step": 4766 }, { "epoch": 0.241135110526582, "grad_norm": 9.184667587280273, "learning_rate": 8.714417013163867e-06, "loss": 0.4368, "step": 4767 }, { "epoch": 0.24118569477464719, "grad_norm": 11.87086296081543, "learning_rate": 8.713879678619006e-06, "loss": 0.4768, "step": 4768 }, { "epoch": 0.24123627902271233, "grad_norm": 16.39287567138672, "learning_rate": 8.713342248376435e-06, "loss": 0.5046, "step": 4769 }, { "epoch": 0.24128686327077747, "grad_norm": 11.468999862670898, "learning_rate": 8.712804722450003e-06, "loss": 0.5397, "step": 4770 }, { "epoch": 0.24133744751884265, "grad_norm": 9.836045265197754, "learning_rate": 8.712267100853563e-06, "loss": 0.3573, "step": 4771 }, { "epoch": 0.2413880317669078, "grad_norm": 11.587859153747559, "learning_rate": 8.711729383600963e-06, "loss": 0.4691, "step": 4772 }, { "epoch": 0.24143861601497293, "grad_norm": 8.326074600219727, "learning_rate": 8.711191570706064e-06, "loss": 0.5262, "step": 4773 }, { "epoch": 0.24148920026303808, "grad_norm": 10.197632789611816, "learning_rate": 8.710653662182722e-06, "loss": 0.4634, "step": 4774 }, { "epoch": 0.24153978451110325, "grad_norm": 7.763383865356445, "learning_rate": 8.710115658044798e-06, "loss": 0.4352, "step": 4775 }, { "epoch": 0.2415903687591684, "grad_norm": 7.6598968505859375, "learning_rate": 8.709577558306156e-06, "loss": 0.5074, "step": 4776 }, { "epoch": 0.24164095300723354, "grad_norm": 9.692883491516113, "learning_rate": 8.70903936298066e-06, "loss": 0.3895, "step": 4777 }, { "epoch": 0.2416915372552987, "grad_norm": 6.832589149475098, "learning_rate": 8.708501072082178e-06, "loss": 0.4971, "step": 4778 }, { "epoch": 0.24174212150336385, "grad_norm": 6.989622116088867, "learning_rate": 8.70796268562458e-06, "loss": 0.5304, "step": 4779 }, { "epoch": 0.241792705751429, "grad_norm": 8.153389930725098, "learning_rate": 8.70742420362174e-06, "loss": 0.4063, "step": 4780 }, { "epoch": 0.24184328999949417, "grad_norm": 7.874114513397217, "learning_rate": 8.706885626087534e-06, "loss": 0.4419, "step": 4781 }, { "epoch": 0.24189387424755932, "grad_norm": 6.314638137817383, "learning_rate": 8.706346953035838e-06, "loss": 0.3559, "step": 4782 }, { "epoch": 0.24194445849562446, "grad_norm": 6.335028648376465, "learning_rate": 8.705808184480536e-06, "loss": 0.426, "step": 4783 }, { "epoch": 0.2419950427436896, "grad_norm": 9.056761741638184, "learning_rate": 8.705269320435506e-06, "loss": 0.4351, "step": 4784 }, { "epoch": 0.24204562699175478, "grad_norm": 10.707076072692871, "learning_rate": 8.704730360914635e-06, "loss": 0.5275, "step": 4785 }, { "epoch": 0.24209621123981992, "grad_norm": 5.862399101257324, "learning_rate": 8.704191305931811e-06, "loss": 0.4741, "step": 4786 }, { "epoch": 0.24214679548788506, "grad_norm": 11.374796867370605, "learning_rate": 8.703652155500926e-06, "loss": 0.4538, "step": 4787 }, { "epoch": 0.24219737973595024, "grad_norm": 5.702219486236572, "learning_rate": 8.70311290963587e-06, "loss": 0.4837, "step": 4788 }, { "epoch": 0.24224796398401538, "grad_norm": 9.27982234954834, "learning_rate": 8.702573568350537e-06, "loss": 0.4072, "step": 4789 }, { "epoch": 0.24229854823208052, "grad_norm": 4.158278465270996, "learning_rate": 8.702034131658829e-06, "loss": 0.3256, "step": 4790 }, { "epoch": 0.2423491324801457, "grad_norm": 7.062221050262451, "learning_rate": 8.701494599574642e-06, "loss": 0.3595, "step": 4791 }, { "epoch": 0.24239971672821084, "grad_norm": 4.901157379150391, "learning_rate": 8.70095497211188e-06, "loss": 0.3577, "step": 4792 }, { "epoch": 0.24245030097627598, "grad_norm": 7.570026874542236, "learning_rate": 8.700415249284446e-06, "loss": 0.3866, "step": 4793 }, { "epoch": 0.24250088522434113, "grad_norm": 7.869311809539795, "learning_rate": 8.69987543110625e-06, "loss": 0.4029, "step": 4794 }, { "epoch": 0.2425514694724063, "grad_norm": 13.045199394226074, "learning_rate": 8.699335517591203e-06, "loss": 0.5123, "step": 4795 }, { "epoch": 0.24260205372047144, "grad_norm": 17.211956024169922, "learning_rate": 8.698795508753213e-06, "loss": 0.5064, "step": 4796 }, { "epoch": 0.2426526379685366, "grad_norm": 6.023369312286377, "learning_rate": 8.698255404606197e-06, "loss": 0.4195, "step": 4797 }, { "epoch": 0.24270322221660176, "grad_norm": 5.450740337371826, "learning_rate": 8.697715205164073e-06, "loss": 0.4086, "step": 4798 }, { "epoch": 0.2427538064646669, "grad_norm": 8.287321090698242, "learning_rate": 8.697174910440757e-06, "loss": 0.4352, "step": 4799 }, { "epoch": 0.24280439071273205, "grad_norm": 11.509634017944336, "learning_rate": 8.696634520450176e-06, "loss": 0.454, "step": 4800 }, { "epoch": 0.24285497496079722, "grad_norm": 8.739070892333984, "learning_rate": 8.696094035206253e-06, "loss": 0.4765, "step": 4801 }, { "epoch": 0.24290555920886236, "grad_norm": 13.71989631652832, "learning_rate": 8.695553454722912e-06, "loss": 0.5705, "step": 4802 }, { "epoch": 0.2429561434569275, "grad_norm": 7.590580940246582, "learning_rate": 8.695012779014086e-06, "loss": 0.5173, "step": 4803 }, { "epoch": 0.24300672770499265, "grad_norm": 8.08584213256836, "learning_rate": 8.694472008093706e-06, "loss": 0.4132, "step": 4804 }, { "epoch": 0.24305731195305783, "grad_norm": 10.020030975341797, "learning_rate": 8.693931141975705e-06, "loss": 0.596, "step": 4805 }, { "epoch": 0.24310789620112297, "grad_norm": 6.339267730712891, "learning_rate": 8.693390180674021e-06, "loss": 0.3768, "step": 4806 }, { "epoch": 0.2431584804491881, "grad_norm": 9.940568923950195, "learning_rate": 8.692849124202594e-06, "loss": 0.5239, "step": 4807 }, { "epoch": 0.24320906469725329, "grad_norm": 6.821156978607178, "learning_rate": 8.692307972575362e-06, "loss": 0.3834, "step": 4808 }, { "epoch": 0.24325964894531843, "grad_norm": 5.710146427154541, "learning_rate": 8.691766725806274e-06, "loss": 0.3928, "step": 4809 }, { "epoch": 0.24331023319338357, "grad_norm": 10.210664749145508, "learning_rate": 8.691225383909275e-06, "loss": 0.481, "step": 4810 }, { "epoch": 0.24336081744144875, "grad_norm": 7.837977886199951, "learning_rate": 8.690683946898315e-06, "loss": 0.3941, "step": 4811 }, { "epoch": 0.2434114016895139, "grad_norm": 10.752363204956055, "learning_rate": 8.690142414787342e-06, "loss": 0.5367, "step": 4812 }, { "epoch": 0.24346198593757903, "grad_norm": 9.357422828674316, "learning_rate": 8.68960078759031e-06, "loss": 0.445, "step": 4813 }, { "epoch": 0.24351257018564418, "grad_norm": 9.154314041137695, "learning_rate": 8.689059065321181e-06, "loss": 0.4834, "step": 4814 }, { "epoch": 0.24356315443370935, "grad_norm": 6.041956901550293, "learning_rate": 8.688517247993911e-06, "loss": 0.3508, "step": 4815 }, { "epoch": 0.2436137386817745, "grad_norm": 7.543399333953857, "learning_rate": 8.687975335622459e-06, "loss": 0.4598, "step": 4816 }, { "epoch": 0.24366432292983964, "grad_norm": 11.072132110595703, "learning_rate": 8.68743332822079e-06, "loss": 0.4323, "step": 4817 }, { "epoch": 0.2437149071779048, "grad_norm": 12.363423347473145, "learning_rate": 8.686891225802872e-06, "loss": 0.6181, "step": 4818 }, { "epoch": 0.24376549142596995, "grad_norm": 8.338308334350586, "learning_rate": 8.686349028382672e-06, "loss": 0.4224, "step": 4819 }, { "epoch": 0.2438160756740351, "grad_norm": 5.599056243896484, "learning_rate": 8.685806735974162e-06, "loss": 0.376, "step": 4820 }, { "epoch": 0.24386665992210027, "grad_norm": 16.836145401000977, "learning_rate": 8.685264348591313e-06, "loss": 0.4241, "step": 4821 }, { "epoch": 0.24391724417016541, "grad_norm": 7.242994785308838, "learning_rate": 8.684721866248106e-06, "loss": 0.5178, "step": 4822 }, { "epoch": 0.24396782841823056, "grad_norm": 5.903168201446533, "learning_rate": 8.684179288958515e-06, "loss": 0.4486, "step": 4823 }, { "epoch": 0.2440184126662957, "grad_norm": 7.6365180015563965, "learning_rate": 8.683636616736523e-06, "loss": 0.4727, "step": 4824 }, { "epoch": 0.24406899691436088, "grad_norm": 7.232969284057617, "learning_rate": 8.683093849596112e-06, "loss": 0.414, "step": 4825 }, { "epoch": 0.24411958116242602, "grad_norm": 6.877125263214111, "learning_rate": 8.68255098755127e-06, "loss": 0.3905, "step": 4826 }, { "epoch": 0.24417016541049116, "grad_norm": 9.176347732543945, "learning_rate": 8.682008030615983e-06, "loss": 0.3089, "step": 4827 }, { "epoch": 0.24422074965855634, "grad_norm": 8.935104370117188, "learning_rate": 8.681464978804243e-06, "loss": 0.5434, "step": 4828 }, { "epoch": 0.24427133390662148, "grad_norm": 3.9502739906311035, "learning_rate": 8.680921832130042e-06, "loss": 0.3596, "step": 4829 }, { "epoch": 0.24432191815468662, "grad_norm": 6.527868747711182, "learning_rate": 8.680378590607377e-06, "loss": 0.4266, "step": 4830 }, { "epoch": 0.2443725024027518, "grad_norm": 10.966021537780762, "learning_rate": 8.679835254250246e-06, "loss": 0.4867, "step": 4831 }, { "epoch": 0.24442308665081694, "grad_norm": 8.385690689086914, "learning_rate": 8.679291823072649e-06, "loss": 0.4753, "step": 4832 }, { "epoch": 0.24447367089888208, "grad_norm": 5.960915565490723, "learning_rate": 8.678748297088587e-06, "loss": 0.4307, "step": 4833 }, { "epoch": 0.24452425514694723, "grad_norm": 8.220162391662598, "learning_rate": 8.678204676312068e-06, "loss": 0.3921, "step": 4834 }, { "epoch": 0.2445748393950124, "grad_norm": 9.761397361755371, "learning_rate": 8.677660960757097e-06, "loss": 0.3942, "step": 4835 }, { "epoch": 0.24462542364307754, "grad_norm": 16.016986846923828, "learning_rate": 8.677117150437688e-06, "loss": 0.5828, "step": 4836 }, { "epoch": 0.2446760078911427, "grad_norm": 9.982354164123535, "learning_rate": 8.676573245367852e-06, "loss": 0.4894, "step": 4837 }, { "epoch": 0.24472659213920786, "grad_norm": 13.465288162231445, "learning_rate": 8.676029245561603e-06, "loss": 0.5572, "step": 4838 }, { "epoch": 0.244777176387273, "grad_norm": 10.420082092285156, "learning_rate": 8.67548515103296e-06, "loss": 0.4543, "step": 4839 }, { "epoch": 0.24482776063533815, "grad_norm": 7.2643256187438965, "learning_rate": 8.674940961795942e-06, "loss": 0.3862, "step": 4840 }, { "epoch": 0.24487834488340332, "grad_norm": 18.15914535522461, "learning_rate": 8.67439667786457e-06, "loss": 0.4849, "step": 4841 }, { "epoch": 0.24492892913146846, "grad_norm": 8.994636535644531, "learning_rate": 8.673852299252873e-06, "loss": 0.5607, "step": 4842 }, { "epoch": 0.2449795133795336, "grad_norm": 19.619754791259766, "learning_rate": 8.673307825974878e-06, "loss": 0.4537, "step": 4843 }, { "epoch": 0.24503009762759875, "grad_norm": 8.081705093383789, "learning_rate": 8.67276325804461e-06, "loss": 0.4776, "step": 4844 }, { "epoch": 0.24508068187566392, "grad_norm": 8.431921005249023, "learning_rate": 8.672218595476104e-06, "loss": 0.4092, "step": 4845 }, { "epoch": 0.24513126612372907, "grad_norm": 10.612285614013672, "learning_rate": 8.671673838283395e-06, "loss": 0.5125, "step": 4846 }, { "epoch": 0.2451818503717942, "grad_norm": 15.470487594604492, "learning_rate": 8.671128986480519e-06, "loss": 0.5682, "step": 4847 }, { "epoch": 0.24523243461985939, "grad_norm": 6.266279220581055, "learning_rate": 8.670584040081517e-06, "loss": 0.4524, "step": 4848 }, { "epoch": 0.24528301886792453, "grad_norm": 11.127076148986816, "learning_rate": 8.670038999100432e-06, "loss": 0.4536, "step": 4849 }, { "epoch": 0.24533360311598967, "grad_norm": 10.853476524353027, "learning_rate": 8.669493863551304e-06, "loss": 0.4886, "step": 4850 }, { "epoch": 0.24538418736405485, "grad_norm": 8.748043060302734, "learning_rate": 8.668948633448184e-06, "loss": 0.4379, "step": 4851 }, { "epoch": 0.24543477161212, "grad_norm": 12.001137733459473, "learning_rate": 8.66840330880512e-06, "loss": 0.4776, "step": 4852 }, { "epoch": 0.24548535586018513, "grad_norm": 9.970038414001465, "learning_rate": 8.667857889636161e-06, "loss": 0.4543, "step": 4853 }, { "epoch": 0.24553594010825028, "grad_norm": 9.202857971191406, "learning_rate": 8.667312375955366e-06, "loss": 0.4135, "step": 4854 }, { "epoch": 0.24558652435631545, "grad_norm": 12.575429916381836, "learning_rate": 8.66676676777679e-06, "loss": 0.5064, "step": 4855 }, { "epoch": 0.2456371086043806, "grad_norm": 5.940347194671631, "learning_rate": 8.666221065114488e-06, "loss": 0.4572, "step": 4856 }, { "epoch": 0.24568769285244574, "grad_norm": 10.678006172180176, "learning_rate": 8.665675267982524e-06, "loss": 0.4512, "step": 4857 }, { "epoch": 0.2457382771005109, "grad_norm": 7.959460258483887, "learning_rate": 8.665129376394967e-06, "loss": 0.3903, "step": 4858 }, { "epoch": 0.24578886134857605, "grad_norm": 8.329150199890137, "learning_rate": 8.664583390365874e-06, "loss": 0.3839, "step": 4859 }, { "epoch": 0.2458394455966412, "grad_norm": 6.026998043060303, "learning_rate": 8.664037309909321e-06, "loss": 0.4414, "step": 4860 }, { "epoch": 0.24589002984470637, "grad_norm": 5.831097602844238, "learning_rate": 8.663491135039378e-06, "loss": 0.502, "step": 4861 }, { "epoch": 0.24594061409277151, "grad_norm": 6.240731716156006, "learning_rate": 8.662944865770114e-06, "loss": 0.4031, "step": 4862 }, { "epoch": 0.24599119834083666, "grad_norm": 15.225998878479004, "learning_rate": 8.66239850211561e-06, "loss": 0.59, "step": 4863 }, { "epoch": 0.2460417825889018, "grad_norm": 8.715568542480469, "learning_rate": 8.661852044089943e-06, "loss": 0.3845, "step": 4864 }, { "epoch": 0.24609236683696697, "grad_norm": 11.165040969848633, "learning_rate": 8.661305491707193e-06, "loss": 0.4895, "step": 4865 }, { "epoch": 0.24614295108503212, "grad_norm": 15.138317108154297, "learning_rate": 8.660758844981444e-06, "loss": 0.5305, "step": 4866 }, { "epoch": 0.24619353533309726, "grad_norm": 7.5436811447143555, "learning_rate": 8.660212103926782e-06, "loss": 0.3774, "step": 4867 }, { "epoch": 0.24624411958116244, "grad_norm": 8.186564445495605, "learning_rate": 8.659665268557295e-06, "loss": 0.4171, "step": 4868 }, { "epoch": 0.24629470382922758, "grad_norm": 9.477527618408203, "learning_rate": 8.659118338887073e-06, "loss": 0.5414, "step": 4869 }, { "epoch": 0.24634528807729272, "grad_norm": 8.983785629272461, "learning_rate": 8.65857131493021e-06, "loss": 0.5652, "step": 4870 }, { "epoch": 0.2463958723253579, "grad_norm": 7.364232540130615, "learning_rate": 8.658024196700801e-06, "loss": 0.5159, "step": 4871 }, { "epoch": 0.24644645657342304, "grad_norm": 9.6350679397583, "learning_rate": 8.657476984212944e-06, "loss": 0.4289, "step": 4872 }, { "epoch": 0.24649704082148818, "grad_norm": 5.14323091506958, "learning_rate": 8.65692967748074e-06, "loss": 0.4128, "step": 4873 }, { "epoch": 0.24654762506955333, "grad_norm": 7.564570426940918, "learning_rate": 8.656382276518291e-06, "loss": 0.5631, "step": 4874 }, { "epoch": 0.2465982093176185, "grad_norm": 6.530716419219971, "learning_rate": 8.655834781339702e-06, "loss": 0.4367, "step": 4875 }, { "epoch": 0.24664879356568364, "grad_norm": 10.571221351623535, "learning_rate": 8.655287191959082e-06, "loss": 0.5665, "step": 4876 }, { "epoch": 0.2466993778137488, "grad_norm": 13.483902931213379, "learning_rate": 8.654739508390539e-06, "loss": 0.4947, "step": 4877 }, { "epoch": 0.24674996206181396, "grad_norm": 11.48198127746582, "learning_rate": 8.654191730648187e-06, "loss": 0.5138, "step": 4878 }, { "epoch": 0.2468005463098791, "grad_norm": 8.253395080566406, "learning_rate": 8.65364385874614e-06, "loss": 0.5231, "step": 4879 }, { "epoch": 0.24685113055794425, "grad_norm": 9.689273834228516, "learning_rate": 8.653095892698516e-06, "loss": 0.5113, "step": 4880 }, { "epoch": 0.24690171480600942, "grad_norm": 8.684228897094727, "learning_rate": 8.652547832519435e-06, "loss": 0.38, "step": 4881 }, { "epoch": 0.24695229905407456, "grad_norm": 8.385849952697754, "learning_rate": 8.651999678223018e-06, "loss": 0.4128, "step": 4882 }, { "epoch": 0.2470028833021397, "grad_norm": 7.5095014572143555, "learning_rate": 8.65145142982339e-06, "loss": 0.4838, "step": 4883 }, { "epoch": 0.24705346755020485, "grad_norm": 21.204782485961914, "learning_rate": 8.650903087334679e-06, "loss": 0.4652, "step": 4884 }, { "epoch": 0.24710405179827002, "grad_norm": 14.027755737304688, "learning_rate": 8.650354650771014e-06, "loss": 0.5362, "step": 4885 }, { "epoch": 0.24715463604633517, "grad_norm": 24.509410858154297, "learning_rate": 8.649806120146527e-06, "loss": 0.5226, "step": 4886 }, { "epoch": 0.2472052202944003, "grad_norm": 10.680588722229004, "learning_rate": 8.649257495475352e-06, "loss": 0.6034, "step": 4887 }, { "epoch": 0.24725580454246548, "grad_norm": 9.013022422790527, "learning_rate": 8.648708776771625e-06, "loss": 0.3503, "step": 4888 }, { "epoch": 0.24730638879053063, "grad_norm": 8.765779495239258, "learning_rate": 8.648159964049486e-06, "loss": 0.2871, "step": 4889 }, { "epoch": 0.24735697303859577, "grad_norm": 5.330460548400879, "learning_rate": 8.647611057323077e-06, "loss": 0.3554, "step": 4890 }, { "epoch": 0.24740755728666095, "grad_norm": 10.738597869873047, "learning_rate": 8.647062056606541e-06, "loss": 0.4805, "step": 4891 }, { "epoch": 0.2474581415347261, "grad_norm": 19.9798583984375, "learning_rate": 8.646512961914026e-06, "loss": 0.5458, "step": 4892 }, { "epoch": 0.24750872578279123, "grad_norm": 9.154176712036133, "learning_rate": 8.645963773259678e-06, "loss": 0.5106, "step": 4893 }, { "epoch": 0.24755931003085638, "grad_norm": 14.085766792297363, "learning_rate": 8.645414490657652e-06, "loss": 0.5182, "step": 4894 }, { "epoch": 0.24760989427892155, "grad_norm": 14.738943099975586, "learning_rate": 8.644865114122098e-06, "loss": 0.6619, "step": 4895 }, { "epoch": 0.2476604785269867, "grad_norm": 10.090096473693848, "learning_rate": 8.644315643667174e-06, "loss": 0.5349, "step": 4896 }, { "epoch": 0.24771106277505184, "grad_norm": 7.409601211547852, "learning_rate": 8.643766079307038e-06, "loss": 0.4893, "step": 4897 }, { "epoch": 0.247761647023117, "grad_norm": 7.8030009269714355, "learning_rate": 8.643216421055854e-06, "loss": 0.4973, "step": 4898 }, { "epoch": 0.24781223127118215, "grad_norm": 9.214841842651367, "learning_rate": 8.64266666892778e-06, "loss": 0.4237, "step": 4899 }, { "epoch": 0.2478628155192473, "grad_norm": 8.219000816345215, "learning_rate": 8.642116822936986e-06, "loss": 0.4873, "step": 4900 }, { "epoch": 0.24791339976731247, "grad_norm": 4.629998683929443, "learning_rate": 8.641566883097636e-06, "loss": 0.5003, "step": 4901 }, { "epoch": 0.24796398401537761, "grad_norm": 13.680392265319824, "learning_rate": 8.641016849423907e-06, "loss": 0.3892, "step": 4902 }, { "epoch": 0.24801456826344276, "grad_norm": 11.844822883605957, "learning_rate": 8.640466721929966e-06, "loss": 0.4878, "step": 4903 }, { "epoch": 0.2480651525115079, "grad_norm": 13.85254192352295, "learning_rate": 8.63991650062999e-06, "loss": 0.6587, "step": 4904 }, { "epoch": 0.24811573675957307, "grad_norm": 4.167117118835449, "learning_rate": 8.63936618553816e-06, "loss": 0.4417, "step": 4905 }, { "epoch": 0.24816632100763822, "grad_norm": 8.059578895568848, "learning_rate": 8.638815776668652e-06, "loss": 0.2855, "step": 4906 }, { "epoch": 0.24821690525570336, "grad_norm": 9.159540176391602, "learning_rate": 8.63826527403565e-06, "loss": 0.4971, "step": 4907 }, { "epoch": 0.24826748950376853, "grad_norm": 9.177541732788086, "learning_rate": 8.637714677653343e-06, "loss": 0.4763, "step": 4908 }, { "epoch": 0.24831807375183368, "grad_norm": 5.7189106941223145, "learning_rate": 8.637163987535913e-06, "loss": 0.4476, "step": 4909 }, { "epoch": 0.24836865799989882, "grad_norm": 23.90258026123047, "learning_rate": 8.636613203697553e-06, "loss": 0.5907, "step": 4910 }, { "epoch": 0.248419242247964, "grad_norm": 5.8250298500061035, "learning_rate": 8.636062326152455e-06, "loss": 0.3897, "step": 4911 }, { "epoch": 0.24846982649602914, "grad_norm": 11.61347770690918, "learning_rate": 8.635511354914812e-06, "loss": 0.5526, "step": 4912 }, { "epoch": 0.24852041074409428, "grad_norm": 9.348612785339355, "learning_rate": 8.634960289998824e-06, "loss": 0.5359, "step": 4913 }, { "epoch": 0.24857099499215946, "grad_norm": 12.826947212219238, "learning_rate": 8.634409131418688e-06, "loss": 0.5312, "step": 4914 }, { "epoch": 0.2486215792402246, "grad_norm": 9.894623756408691, "learning_rate": 8.633857879188608e-06, "loss": 0.3997, "step": 4915 }, { "epoch": 0.24867216348828974, "grad_norm": 6.396818161010742, "learning_rate": 8.633306533322788e-06, "loss": 0.4465, "step": 4916 }, { "epoch": 0.2487227477363549, "grad_norm": 14.968758583068848, "learning_rate": 8.632755093835431e-06, "loss": 0.3799, "step": 4917 }, { "epoch": 0.24877333198442006, "grad_norm": 8.421135902404785, "learning_rate": 8.632203560740754e-06, "loss": 0.5419, "step": 4918 }, { "epoch": 0.2488239162324852, "grad_norm": 6.921187400817871, "learning_rate": 8.631651934052965e-06, "loss": 0.3757, "step": 4919 }, { "epoch": 0.24887450048055035, "grad_norm": 8.96572494506836, "learning_rate": 8.631100213786274e-06, "loss": 0.5189, "step": 4920 }, { "epoch": 0.24892508472861552, "grad_norm": 7.076498031616211, "learning_rate": 8.630548399954903e-06, "loss": 0.456, "step": 4921 }, { "epoch": 0.24897566897668066, "grad_norm": 5.706106185913086, "learning_rate": 8.62999649257307e-06, "loss": 0.397, "step": 4922 }, { "epoch": 0.2490262532247458, "grad_norm": 4.415600299835205, "learning_rate": 8.629444491654993e-06, "loss": 0.4404, "step": 4923 }, { "epoch": 0.24907683747281098, "grad_norm": 13.41874885559082, "learning_rate": 8.628892397214899e-06, "loss": 0.4571, "step": 4924 }, { "epoch": 0.24912742172087612, "grad_norm": 9.95048999786377, "learning_rate": 8.628340209267014e-06, "loss": 0.4572, "step": 4925 }, { "epoch": 0.24917800596894127, "grad_norm": 7.52114725112915, "learning_rate": 8.627787927825565e-06, "loss": 0.4266, "step": 4926 }, { "epoch": 0.2492285902170064, "grad_norm": 7.80372428894043, "learning_rate": 8.627235552904783e-06, "loss": 0.5204, "step": 4927 }, { "epoch": 0.24927917446507158, "grad_norm": 11.518707275390625, "learning_rate": 8.626683084518903e-06, "loss": 0.4076, "step": 4928 }, { "epoch": 0.24932975871313673, "grad_norm": 17.88543701171875, "learning_rate": 8.626130522682159e-06, "loss": 0.5021, "step": 4929 }, { "epoch": 0.24938034296120187, "grad_norm": 10.08660888671875, "learning_rate": 8.625577867408792e-06, "loss": 0.394, "step": 4930 }, { "epoch": 0.24943092720926704, "grad_norm": 16.531085968017578, "learning_rate": 8.625025118713038e-06, "loss": 0.7693, "step": 4931 }, { "epoch": 0.2494815114573322, "grad_norm": 6.158487796783447, "learning_rate": 8.624472276609145e-06, "loss": 0.3367, "step": 4932 }, { "epoch": 0.24953209570539733, "grad_norm": 8.317028045654297, "learning_rate": 8.623919341111353e-06, "loss": 0.4615, "step": 4933 }, { "epoch": 0.2495826799534625, "grad_norm": 6.606589317321777, "learning_rate": 8.623366312233913e-06, "loss": 0.3595, "step": 4934 }, { "epoch": 0.24963326420152765, "grad_norm": 6.411056041717529, "learning_rate": 8.622813189991078e-06, "loss": 0.3677, "step": 4935 }, { "epoch": 0.2496838484495928, "grad_norm": 6.851714611053467, "learning_rate": 8.622259974397095e-06, "loss": 0.4728, "step": 4936 }, { "epoch": 0.24973443269765794, "grad_norm": 6.994766712188721, "learning_rate": 8.621706665466225e-06, "loss": 0.467, "step": 4937 }, { "epoch": 0.2497850169457231, "grad_norm": 9.786529541015625, "learning_rate": 8.62115326321272e-06, "loss": 0.4473, "step": 4938 }, { "epoch": 0.24983560119378825, "grad_norm": 14.68928050994873, "learning_rate": 8.620599767650841e-06, "loss": 0.5686, "step": 4939 }, { "epoch": 0.2498861854418534, "grad_norm": 7.177305221557617, "learning_rate": 8.620046178794852e-06, "loss": 0.4988, "step": 4940 }, { "epoch": 0.24993676968991857, "grad_norm": 8.621713638305664, "learning_rate": 8.619492496659016e-06, "loss": 0.4542, "step": 4941 }, { "epoch": 0.24998735393798371, "grad_norm": 7.516741752624512, "learning_rate": 8.618938721257601e-06, "loss": 0.4114, "step": 4942 }, { "epoch": 0.25003793818604886, "grad_norm": 9.038400650024414, "learning_rate": 8.618384852604877e-06, "loss": 0.4141, "step": 4943 }, { "epoch": 0.250088522434114, "grad_norm": 10.761434555053711, "learning_rate": 8.617830890715117e-06, "loss": 0.5002, "step": 4944 }, { "epoch": 0.25013910668217915, "grad_norm": 6.475780487060547, "learning_rate": 8.61727683560259e-06, "loss": 0.4818, "step": 4945 }, { "epoch": 0.25018969093024435, "grad_norm": 5.389318943023682, "learning_rate": 8.61672268728158e-06, "loss": 0.4077, "step": 4946 }, { "epoch": 0.2502402751783095, "grad_norm": 7.144037246704102, "learning_rate": 8.616168445766359e-06, "loss": 0.3937, "step": 4947 }, { "epoch": 0.25029085942637463, "grad_norm": 9.836610794067383, "learning_rate": 8.615614111071213e-06, "loss": 0.529, "step": 4948 }, { "epoch": 0.2503414436744398, "grad_norm": 7.173312664031982, "learning_rate": 8.615059683210424e-06, "loss": 0.4712, "step": 4949 }, { "epoch": 0.2503920279225049, "grad_norm": 5.462027072906494, "learning_rate": 8.614505162198279e-06, "loss": 0.4245, "step": 4950 }, { "epoch": 0.25044261217057007, "grad_norm": 10.039440155029297, "learning_rate": 8.613950548049066e-06, "loss": 0.5025, "step": 4951 }, { "epoch": 0.2504931964186352, "grad_norm": 11.904180526733398, "learning_rate": 8.613395840777076e-06, "loss": 0.4557, "step": 4952 }, { "epoch": 0.2505437806667004, "grad_norm": 9.54365062713623, "learning_rate": 8.612841040396604e-06, "loss": 0.4326, "step": 4953 }, { "epoch": 0.25059436491476555, "grad_norm": 12.682971954345703, "learning_rate": 8.612286146921945e-06, "loss": 0.4243, "step": 4954 }, { "epoch": 0.2506449491628307, "grad_norm": 10.073517799377441, "learning_rate": 8.611731160367398e-06, "loss": 0.3734, "step": 4955 }, { "epoch": 0.25069553341089584, "grad_norm": 9.237675666809082, "learning_rate": 8.61117608074726e-06, "loss": 0.4328, "step": 4956 }, { "epoch": 0.250746117658961, "grad_norm": 33.02617263793945, "learning_rate": 8.61062090807584e-06, "loss": 0.4586, "step": 4957 }, { "epoch": 0.25079670190702613, "grad_norm": 19.489009857177734, "learning_rate": 8.610065642367437e-06, "loss": 0.6152, "step": 4958 }, { "epoch": 0.25084728615509133, "grad_norm": 7.3278961181640625, "learning_rate": 8.609510283636365e-06, "loss": 0.3961, "step": 4959 }, { "epoch": 0.2508978704031565, "grad_norm": 8.267524719238281, "learning_rate": 8.60895483189693e-06, "loss": 0.5022, "step": 4960 }, { "epoch": 0.2509484546512216, "grad_norm": 5.964146614074707, "learning_rate": 8.608399287163447e-06, "loss": 0.3907, "step": 4961 }, { "epoch": 0.25099903889928676, "grad_norm": 6.520516395568848, "learning_rate": 8.607843649450229e-06, "loss": 0.2665, "step": 4962 }, { "epoch": 0.2510496231473519, "grad_norm": 9.801689147949219, "learning_rate": 8.607287918771596e-06, "loss": 0.4867, "step": 4963 }, { "epoch": 0.25110020739541705, "grad_norm": 7.733133792877197, "learning_rate": 8.606732095141865e-06, "loss": 0.5216, "step": 4964 }, { "epoch": 0.2511507916434822, "grad_norm": 13.09719467163086, "learning_rate": 8.60617617857536e-06, "loss": 0.5218, "step": 4965 }, { "epoch": 0.2512013758915474, "grad_norm": 10.437329292297363, "learning_rate": 8.605620169086405e-06, "loss": 0.4266, "step": 4966 }, { "epoch": 0.25125196013961254, "grad_norm": 12.473799705505371, "learning_rate": 8.605064066689328e-06, "loss": 0.638, "step": 4967 }, { "epoch": 0.2513025443876777, "grad_norm": 7.248542785644531, "learning_rate": 8.604507871398458e-06, "loss": 0.4145, "step": 4968 }, { "epoch": 0.25135312863574283, "grad_norm": 7.892492771148682, "learning_rate": 8.603951583228126e-06, "loss": 0.4996, "step": 4969 }, { "epoch": 0.251403712883808, "grad_norm": 7.017789363861084, "learning_rate": 8.603395202192664e-06, "loss": 0.3496, "step": 4970 }, { "epoch": 0.2514542971318731, "grad_norm": 9.4721040725708, "learning_rate": 8.602838728306415e-06, "loss": 0.3745, "step": 4971 }, { "epoch": 0.25150488137993826, "grad_norm": 11.284628868103027, "learning_rate": 8.602282161583713e-06, "loss": 0.5349, "step": 4972 }, { "epoch": 0.25155546562800346, "grad_norm": 8.267229080200195, "learning_rate": 8.6017255020389e-06, "loss": 0.4302, "step": 4973 }, { "epoch": 0.2516060498760686, "grad_norm": 9.542133331298828, "learning_rate": 8.601168749686321e-06, "loss": 0.4685, "step": 4974 }, { "epoch": 0.25165663412413375, "grad_norm": 11.633615493774414, "learning_rate": 8.600611904540322e-06, "loss": 0.5045, "step": 4975 }, { "epoch": 0.2517072183721989, "grad_norm": 9.012805938720703, "learning_rate": 8.60005496661525e-06, "loss": 0.5167, "step": 4976 }, { "epoch": 0.25175780262026404, "grad_norm": 22.935527801513672, "learning_rate": 8.599497935925458e-06, "loss": 0.409, "step": 4977 }, { "epoch": 0.2518083868683292, "grad_norm": 15.763984680175781, "learning_rate": 8.598940812485298e-06, "loss": 0.624, "step": 4978 }, { "epoch": 0.2518589711163944, "grad_norm": 4.551523208618164, "learning_rate": 8.598383596309127e-06, "loss": 0.2885, "step": 4979 }, { "epoch": 0.2519095553644595, "grad_norm": 7.259990692138672, "learning_rate": 8.597826287411302e-06, "loss": 0.4431, "step": 4980 }, { "epoch": 0.25196013961252467, "grad_norm": 11.39510726928711, "learning_rate": 8.597268885806183e-06, "loss": 0.4326, "step": 4981 }, { "epoch": 0.2520107238605898, "grad_norm": 9.605892181396484, "learning_rate": 8.596711391508134e-06, "loss": 0.5023, "step": 4982 }, { "epoch": 0.25206130810865496, "grad_norm": 10.195893287658691, "learning_rate": 8.59615380453152e-06, "loss": 0.4524, "step": 4983 }, { "epoch": 0.2521118923567201, "grad_norm": 7.204518795013428, "learning_rate": 8.595596124890709e-06, "loss": 0.3333, "step": 4984 }, { "epoch": 0.25216247660478525, "grad_norm": 5.299294471740723, "learning_rate": 8.595038352600069e-06, "loss": 0.4121, "step": 4985 }, { "epoch": 0.25221306085285045, "grad_norm": 9.040958404541016, "learning_rate": 8.594480487673977e-06, "loss": 0.366, "step": 4986 }, { "epoch": 0.2522636451009156, "grad_norm": 11.3199462890625, "learning_rate": 8.593922530126804e-06, "loss": 0.5101, "step": 4987 }, { "epoch": 0.25231422934898073, "grad_norm": 6.159660816192627, "learning_rate": 8.593364479972925e-06, "loss": 0.322, "step": 4988 }, { "epoch": 0.2523648135970459, "grad_norm": 5.224004745483398, "learning_rate": 8.592806337226727e-06, "loss": 0.4191, "step": 4989 }, { "epoch": 0.252415397845111, "grad_norm": 7.095064640045166, "learning_rate": 8.592248101902587e-06, "loss": 0.3703, "step": 4990 }, { "epoch": 0.25246598209317617, "grad_norm": 10.629749298095703, "learning_rate": 8.591689774014888e-06, "loss": 0.4952, "step": 4991 }, { "epoch": 0.2525165663412413, "grad_norm": 9.91455364227295, "learning_rate": 8.59113135357802e-06, "loss": 0.5024, "step": 4992 }, { "epoch": 0.2525671505893065, "grad_norm": 8.144364356994629, "learning_rate": 8.59057284060637e-06, "loss": 0.4444, "step": 4993 }, { "epoch": 0.25261773483737165, "grad_norm": 6.166378498077393, "learning_rate": 8.590014235114332e-06, "loss": 0.3739, "step": 4994 }, { "epoch": 0.2526683190854368, "grad_norm": 9.35401725769043, "learning_rate": 8.589455537116297e-06, "loss": 0.4353, "step": 4995 }, { "epoch": 0.25271890333350194, "grad_norm": 11.811739921569824, "learning_rate": 8.588896746626665e-06, "loss": 0.4783, "step": 4996 }, { "epoch": 0.2527694875815671, "grad_norm": 8.714507102966309, "learning_rate": 8.588337863659833e-06, "loss": 0.4283, "step": 4997 }, { "epoch": 0.25282007182963223, "grad_norm": 15.883503913879395, "learning_rate": 8.5877788882302e-06, "loss": 0.7129, "step": 4998 }, { "epoch": 0.25287065607769743, "grad_norm": 7.9423322677612305, "learning_rate": 8.58721982035217e-06, "loss": 0.4099, "step": 4999 }, { "epoch": 0.2529212403257626, "grad_norm": 11.941916465759277, "learning_rate": 8.586660660040152e-06, "loss": 0.4164, "step": 5000 }, { "epoch": 0.2529718245738277, "grad_norm": 16.71723175048828, "learning_rate": 8.586101407308549e-06, "loss": 0.5384, "step": 5001 }, { "epoch": 0.25302240882189286, "grad_norm": 11.406723022460938, "learning_rate": 8.585542062171777e-06, "loss": 0.4939, "step": 5002 }, { "epoch": 0.253072993069958, "grad_norm": 13.703204154968262, "learning_rate": 8.584982624644246e-06, "loss": 0.5325, "step": 5003 }, { "epoch": 0.25312357731802315, "grad_norm": 6.737446308135986, "learning_rate": 8.584423094740372e-06, "loss": 0.5407, "step": 5004 }, { "epoch": 0.2531741615660883, "grad_norm": 8.320106506347656, "learning_rate": 8.583863472474571e-06, "loss": 0.3323, "step": 5005 }, { "epoch": 0.2532247458141535, "grad_norm": 5.288793563842773, "learning_rate": 8.583303757861267e-06, "loss": 0.4941, "step": 5006 }, { "epoch": 0.25327533006221864, "grad_norm": 15.592029571533203, "learning_rate": 8.58274395091488e-06, "loss": 0.5638, "step": 5007 }, { "epoch": 0.2533259143102838, "grad_norm": 5.595839500427246, "learning_rate": 8.582184051649835e-06, "loss": 0.4187, "step": 5008 }, { "epoch": 0.25337649855834893, "grad_norm": 12.010811805725098, "learning_rate": 8.58162406008056e-06, "loss": 0.4615, "step": 5009 }, { "epoch": 0.2534270828064141, "grad_norm": 5.49500846862793, "learning_rate": 8.581063976221483e-06, "loss": 0.4213, "step": 5010 }, { "epoch": 0.2534776670544792, "grad_norm": 8.63156509399414, "learning_rate": 8.580503800087037e-06, "loss": 0.4609, "step": 5011 }, { "epoch": 0.25352825130254436, "grad_norm": 6.639068126678467, "learning_rate": 8.579943531691656e-06, "loss": 0.4626, "step": 5012 }, { "epoch": 0.25357883555060956, "grad_norm": 8.49321460723877, "learning_rate": 8.579383171049777e-06, "loss": 0.4878, "step": 5013 }, { "epoch": 0.2536294197986747, "grad_norm": 11.278731346130371, "learning_rate": 8.578822718175841e-06, "loss": 0.6298, "step": 5014 }, { "epoch": 0.25368000404673985, "grad_norm": 13.877036094665527, "learning_rate": 8.578262173084286e-06, "loss": 0.5639, "step": 5015 }, { "epoch": 0.253730588294805, "grad_norm": 8.815997123718262, "learning_rate": 8.57770153578956e-06, "loss": 0.5752, "step": 5016 }, { "epoch": 0.25378117254287014, "grad_norm": 6.141082286834717, "learning_rate": 8.577140806306106e-06, "loss": 0.3971, "step": 5017 }, { "epoch": 0.2538317567909353, "grad_norm": 12.198312759399414, "learning_rate": 8.576579984648372e-06, "loss": 0.5678, "step": 5018 }, { "epoch": 0.2538823410390005, "grad_norm": 11.270475387573242, "learning_rate": 8.576019070830812e-06, "loss": 0.5523, "step": 5019 }, { "epoch": 0.2539329252870656, "grad_norm": 7.908660411834717, "learning_rate": 8.575458064867878e-06, "loss": 0.3474, "step": 5020 }, { "epoch": 0.25398350953513077, "grad_norm": 7.339848518371582, "learning_rate": 8.574896966774026e-06, "loss": 0.4917, "step": 5021 }, { "epoch": 0.2540340937831959, "grad_norm": 12.637535095214844, "learning_rate": 8.574335776563715e-06, "loss": 0.6055, "step": 5022 }, { "epoch": 0.25408467803126106, "grad_norm": 8.85988998413086, "learning_rate": 8.573774494251404e-06, "loss": 0.4885, "step": 5023 }, { "epoch": 0.2541352622793262, "grad_norm": 9.357638359069824, "learning_rate": 8.573213119851555e-06, "loss": 0.4442, "step": 5024 }, { "epoch": 0.25418584652739135, "grad_norm": 7.426375865936279, "learning_rate": 8.572651653378636e-06, "loss": 0.4243, "step": 5025 }, { "epoch": 0.25423643077545655, "grad_norm": 6.850745677947998, "learning_rate": 8.572090094847112e-06, "loss": 0.5047, "step": 5026 }, { "epoch": 0.2542870150235217, "grad_norm": 12.054512977600098, "learning_rate": 8.571528444271455e-06, "loss": 0.5102, "step": 5027 }, { "epoch": 0.25433759927158683, "grad_norm": 6.292893886566162, "learning_rate": 8.570966701666137e-06, "loss": 0.3952, "step": 5028 }, { "epoch": 0.254388183519652, "grad_norm": 11.351149559020996, "learning_rate": 8.570404867045631e-06, "loss": 0.5845, "step": 5029 }, { "epoch": 0.2544387677677171, "grad_norm": 9.645157814025879, "learning_rate": 8.569842940424417e-06, "loss": 0.5446, "step": 5030 }, { "epoch": 0.25448935201578227, "grad_norm": 9.389283180236816, "learning_rate": 8.569280921816974e-06, "loss": 0.4694, "step": 5031 }, { "epoch": 0.2545399362638474, "grad_norm": 7.933084011077881, "learning_rate": 8.568718811237781e-06, "loss": 0.316, "step": 5032 }, { "epoch": 0.2545905205119126, "grad_norm": 6.3894429206848145, "learning_rate": 8.568156608701325e-06, "loss": 0.387, "step": 5033 }, { "epoch": 0.25464110475997775, "grad_norm": 6.979374885559082, "learning_rate": 8.56759431422209e-06, "loss": 0.515, "step": 5034 }, { "epoch": 0.2546916890080429, "grad_norm": 7.950040340423584, "learning_rate": 8.567031927814568e-06, "loss": 0.4726, "step": 5035 }, { "epoch": 0.25474227325610804, "grad_norm": 10.431882858276367, "learning_rate": 8.566469449493252e-06, "loss": 0.5014, "step": 5036 }, { "epoch": 0.2547928575041732, "grad_norm": 6.752458572387695, "learning_rate": 8.565906879272627e-06, "loss": 0.4466, "step": 5037 }, { "epoch": 0.25484344175223833, "grad_norm": 17.290327072143555, "learning_rate": 8.565344217167199e-06, "loss": 0.5462, "step": 5038 }, { "epoch": 0.25489402600030353, "grad_norm": 5.3102521896362305, "learning_rate": 8.56478146319146e-06, "loss": 0.4652, "step": 5039 }, { "epoch": 0.2549446102483687, "grad_norm": 6.1029767990112305, "learning_rate": 8.564218617359914e-06, "loss": 0.4617, "step": 5040 }, { "epoch": 0.2549951944964338, "grad_norm": 5.136343955993652, "learning_rate": 8.563655679687065e-06, "loss": 0.3906, "step": 5041 }, { "epoch": 0.25504577874449896, "grad_norm": 8.675369262695312, "learning_rate": 8.563092650187414e-06, "loss": 0.3286, "step": 5042 }, { "epoch": 0.2550963629925641, "grad_norm": 6.823602199554443, "learning_rate": 8.562529528875472e-06, "loss": 0.4349, "step": 5043 }, { "epoch": 0.25514694724062925, "grad_norm": 11.558302879333496, "learning_rate": 8.56196631576575e-06, "loss": 0.5167, "step": 5044 }, { "epoch": 0.2551975314886944, "grad_norm": 19.069011688232422, "learning_rate": 8.56140301087276e-06, "loss": 0.5701, "step": 5045 }, { "epoch": 0.2552481157367596, "grad_norm": 9.12344741821289, "learning_rate": 8.560839614211016e-06, "loss": 0.4017, "step": 5046 }, { "epoch": 0.25529869998482474, "grad_norm": 12.670618057250977, "learning_rate": 8.560276125795036e-06, "loss": 0.5578, "step": 5047 }, { "epoch": 0.2553492842328899, "grad_norm": 6.6251420974731445, "learning_rate": 8.55971254563934e-06, "loss": 0.3474, "step": 5048 }, { "epoch": 0.25539986848095503, "grad_norm": 11.282392501831055, "learning_rate": 8.55914887375845e-06, "loss": 0.4986, "step": 5049 }, { "epoch": 0.25545045272902017, "grad_norm": 7.690075397491455, "learning_rate": 8.558585110166888e-06, "loss": 0.3059, "step": 5050 }, { "epoch": 0.2555010369770853, "grad_norm": 8.514121055603027, "learning_rate": 8.558021254879186e-06, "loss": 0.5405, "step": 5051 }, { "epoch": 0.25555162122515046, "grad_norm": 8.537464141845703, "learning_rate": 8.55745730790987e-06, "loss": 0.4698, "step": 5052 }, { "epoch": 0.25560220547321566, "grad_norm": 4.4730072021484375, "learning_rate": 8.55689326927347e-06, "loss": 0.292, "step": 5053 }, { "epoch": 0.2556527897212808, "grad_norm": 6.721518516540527, "learning_rate": 8.556329138984523e-06, "loss": 0.4314, "step": 5054 }, { "epoch": 0.25570337396934595, "grad_norm": 10.138054847717285, "learning_rate": 8.555764917057564e-06, "loss": 0.5007, "step": 5055 }, { "epoch": 0.2557539582174111, "grad_norm": 10.427523612976074, "learning_rate": 8.555200603507132e-06, "loss": 0.3616, "step": 5056 }, { "epoch": 0.25580454246547624, "grad_norm": 9.227554321289062, "learning_rate": 8.554636198347768e-06, "loss": 0.5974, "step": 5057 }, { "epoch": 0.2558551267135414, "grad_norm": 7.866847515106201, "learning_rate": 8.554071701594013e-06, "loss": 0.4464, "step": 5058 }, { "epoch": 0.2559057109616066, "grad_norm": 10.501264572143555, "learning_rate": 8.553507113260417e-06, "loss": 0.4597, "step": 5059 }, { "epoch": 0.2559562952096717, "grad_norm": 5.003580093383789, "learning_rate": 8.552942433361525e-06, "loss": 0.4044, "step": 5060 }, { "epoch": 0.25600687945773687, "grad_norm": 13.545554161071777, "learning_rate": 8.552377661911888e-06, "loss": 0.4232, "step": 5061 }, { "epoch": 0.256057463705802, "grad_norm": 6.2445831298828125, "learning_rate": 8.551812798926058e-06, "loss": 0.4279, "step": 5062 }, { "epoch": 0.25610804795386716, "grad_norm": 6.621402263641357, "learning_rate": 8.551247844418593e-06, "loss": 0.3957, "step": 5063 }, { "epoch": 0.2561586322019323, "grad_norm": 8.090254783630371, "learning_rate": 8.550682798404049e-06, "loss": 0.4276, "step": 5064 }, { "epoch": 0.25620921644999745, "grad_norm": 5.598877906799316, "learning_rate": 8.550117660896984e-06, "loss": 0.3492, "step": 5065 }, { "epoch": 0.25625980069806265, "grad_norm": 7.280945301055908, "learning_rate": 8.549552431911963e-06, "loss": 0.4535, "step": 5066 }, { "epoch": 0.2563103849461278, "grad_norm": 5.465967178344727, "learning_rate": 8.548987111463549e-06, "loss": 0.4785, "step": 5067 }, { "epoch": 0.25636096919419293, "grad_norm": 11.679404258728027, "learning_rate": 8.548421699566309e-06, "loss": 0.5155, "step": 5068 }, { "epoch": 0.2564115534422581, "grad_norm": 10.579936027526855, "learning_rate": 8.547856196234814e-06, "loss": 0.4125, "step": 5069 }, { "epoch": 0.2564621376903232, "grad_norm": 8.971721649169922, "learning_rate": 8.547290601483633e-06, "loss": 0.4402, "step": 5070 }, { "epoch": 0.25651272193838837, "grad_norm": 5.792718887329102, "learning_rate": 8.546724915327343e-06, "loss": 0.4285, "step": 5071 }, { "epoch": 0.25656330618645357, "grad_norm": 8.30695915222168, "learning_rate": 8.546159137780516e-06, "loss": 0.3867, "step": 5072 }, { "epoch": 0.2566138904345187, "grad_norm": 6.528227806091309, "learning_rate": 8.545593268857735e-06, "loss": 0.4248, "step": 5073 }, { "epoch": 0.25666447468258385, "grad_norm": 6.790892124176025, "learning_rate": 8.54502730857358e-06, "loss": 0.4956, "step": 5074 }, { "epoch": 0.256715058930649, "grad_norm": 6.189891815185547, "learning_rate": 8.544461256942633e-06, "loss": 0.4131, "step": 5075 }, { "epoch": 0.25676564317871414, "grad_norm": 5.344735622406006, "learning_rate": 8.543895113979482e-06, "loss": 0.4086, "step": 5076 }, { "epoch": 0.2568162274267793, "grad_norm": 8.067519187927246, "learning_rate": 8.543328879698712e-06, "loss": 0.4418, "step": 5077 }, { "epoch": 0.25686681167484443, "grad_norm": 8.479135513305664, "learning_rate": 8.542762554114915e-06, "loss": 0.4802, "step": 5078 }, { "epoch": 0.25691739592290963, "grad_norm": 9.135034561157227, "learning_rate": 8.542196137242687e-06, "loss": 0.4435, "step": 5079 }, { "epoch": 0.2569679801709748, "grad_norm": 8.843071937561035, "learning_rate": 8.541629629096619e-06, "loss": 0.5415, "step": 5080 }, { "epoch": 0.2570185644190399, "grad_norm": 9.950191497802734, "learning_rate": 8.54106302969131e-06, "loss": 0.5585, "step": 5081 }, { "epoch": 0.25706914866710506, "grad_norm": 6.731178283691406, "learning_rate": 8.540496339041359e-06, "loss": 0.459, "step": 5082 }, { "epoch": 0.2571197329151702, "grad_norm": 7.010904312133789, "learning_rate": 8.539929557161368e-06, "loss": 0.3941, "step": 5083 }, { "epoch": 0.25717031716323535, "grad_norm": 15.305484771728516, "learning_rate": 8.539362684065945e-06, "loss": 0.6341, "step": 5084 }, { "epoch": 0.2572209014113005, "grad_norm": 8.69725513458252, "learning_rate": 8.538795719769694e-06, "loss": 0.435, "step": 5085 }, { "epoch": 0.2572714856593657, "grad_norm": 5.379043102264404, "learning_rate": 8.538228664287224e-06, "loss": 0.3445, "step": 5086 }, { "epoch": 0.25732206990743084, "grad_norm": 7.012641429901123, "learning_rate": 8.537661517633148e-06, "loss": 0.4358, "step": 5087 }, { "epoch": 0.257372654155496, "grad_norm": 11.20766544342041, "learning_rate": 8.537094279822081e-06, "loss": 0.5627, "step": 5088 }, { "epoch": 0.2574232384035611, "grad_norm": 8.990866661071777, "learning_rate": 8.536526950868637e-06, "loss": 0.4535, "step": 5089 }, { "epoch": 0.25747382265162627, "grad_norm": 7.756914138793945, "learning_rate": 8.535959530787434e-06, "loss": 0.4778, "step": 5090 }, { "epoch": 0.2575244068996914, "grad_norm": 6.048843860626221, "learning_rate": 8.535392019593098e-06, "loss": 0.4735, "step": 5091 }, { "epoch": 0.2575749911477566, "grad_norm": 7.857293128967285, "learning_rate": 8.534824417300247e-06, "loss": 0.4591, "step": 5092 }, { "epoch": 0.25762557539582176, "grad_norm": 7.903923511505127, "learning_rate": 8.534256723923509e-06, "loss": 0.5167, "step": 5093 }, { "epoch": 0.2576761596438869, "grad_norm": 9.816922187805176, "learning_rate": 8.533688939477511e-06, "loss": 0.4147, "step": 5094 }, { "epoch": 0.25772674389195205, "grad_norm": 6.420634746551514, "learning_rate": 8.533121063976885e-06, "loss": 0.3748, "step": 5095 }, { "epoch": 0.2577773281400172, "grad_norm": 5.729968547821045, "learning_rate": 8.532553097436263e-06, "loss": 0.3594, "step": 5096 }, { "epoch": 0.25782791238808234, "grad_norm": 8.77975082397461, "learning_rate": 8.53198503987028e-06, "loss": 0.4644, "step": 5097 }, { "epoch": 0.2578784966361475, "grad_norm": 9.913995742797852, "learning_rate": 8.531416891293573e-06, "loss": 0.535, "step": 5098 }, { "epoch": 0.2579290808842127, "grad_norm": 6.305706024169922, "learning_rate": 8.530848651720784e-06, "loss": 0.4264, "step": 5099 }, { "epoch": 0.2579796651322778, "grad_norm": 11.361210823059082, "learning_rate": 8.530280321166552e-06, "loss": 0.5109, "step": 5100 }, { "epoch": 0.25803024938034297, "grad_norm": 10.050493240356445, "learning_rate": 8.529711899645525e-06, "loss": 0.5201, "step": 5101 }, { "epoch": 0.2580808336284081, "grad_norm": 8.796829223632812, "learning_rate": 8.529143387172348e-06, "loss": 0.3581, "step": 5102 }, { "epoch": 0.25813141787647326, "grad_norm": 11.75301742553711, "learning_rate": 8.52857478376167e-06, "loss": 0.5692, "step": 5103 }, { "epoch": 0.2581820021245384, "grad_norm": 7.364997386932373, "learning_rate": 8.528006089428143e-06, "loss": 0.4825, "step": 5104 }, { "epoch": 0.25823258637260355, "grad_norm": 18.14057731628418, "learning_rate": 8.527437304186418e-06, "loss": 0.3999, "step": 5105 }, { "epoch": 0.25828317062066874, "grad_norm": 7.900181293487549, "learning_rate": 8.526868428051156e-06, "loss": 0.3673, "step": 5106 }, { "epoch": 0.2583337548687339, "grad_norm": 7.687961578369141, "learning_rate": 8.526299461037015e-06, "loss": 0.3817, "step": 5107 }, { "epoch": 0.25838433911679903, "grad_norm": 6.798724174499512, "learning_rate": 8.525730403158652e-06, "loss": 0.4142, "step": 5108 }, { "epoch": 0.2584349233648642, "grad_norm": 15.842968940734863, "learning_rate": 8.525161254430731e-06, "loss": 0.5857, "step": 5109 }, { "epoch": 0.2584855076129293, "grad_norm": 10.168281555175781, "learning_rate": 8.524592014867923e-06, "loss": 0.5275, "step": 5110 }, { "epoch": 0.25853609186099447, "grad_norm": 6.238214492797852, "learning_rate": 8.524022684484889e-06, "loss": 0.4819, "step": 5111 }, { "epoch": 0.25858667610905967, "grad_norm": 9.300206184387207, "learning_rate": 8.523453263296304e-06, "loss": 0.432, "step": 5112 }, { "epoch": 0.2586372603571248, "grad_norm": 7.260426998138428, "learning_rate": 8.52288375131684e-06, "loss": 0.4318, "step": 5113 }, { "epoch": 0.25868784460518995, "grad_norm": 7.342074394226074, "learning_rate": 8.522314148561168e-06, "loss": 0.4317, "step": 5114 }, { "epoch": 0.2587384288532551, "grad_norm": 6.4034953117370605, "learning_rate": 8.52174445504397e-06, "loss": 0.4944, "step": 5115 }, { "epoch": 0.25878901310132024, "grad_norm": 7.19274377822876, "learning_rate": 8.521174670779924e-06, "loss": 0.4043, "step": 5116 }, { "epoch": 0.2588395973493854, "grad_norm": 6.558006763458252, "learning_rate": 8.520604795783712e-06, "loss": 0.4696, "step": 5117 }, { "epoch": 0.25889018159745053, "grad_norm": 7.577282428741455, "learning_rate": 8.520034830070016e-06, "loss": 0.378, "step": 5118 }, { "epoch": 0.25894076584551573, "grad_norm": 17.615711212158203, "learning_rate": 8.519464773653528e-06, "loss": 0.5711, "step": 5119 }, { "epoch": 0.2589913500935809, "grad_norm": 7.604231357574463, "learning_rate": 8.51889462654893e-06, "loss": 0.4657, "step": 5120 }, { "epoch": 0.259041934341646, "grad_norm": 6.324631690979004, "learning_rate": 8.51832438877092e-06, "loss": 0.3794, "step": 5121 }, { "epoch": 0.25909251858971116, "grad_norm": 5.227373123168945, "learning_rate": 8.517754060334186e-06, "loss": 0.4652, "step": 5122 }, { "epoch": 0.2591431028377763, "grad_norm": 7.84155797958374, "learning_rate": 8.517183641253429e-06, "loss": 0.4763, "step": 5123 }, { "epoch": 0.25919368708584145, "grad_norm": 8.145167350769043, "learning_rate": 8.516613131543344e-06, "loss": 0.4568, "step": 5124 }, { "epoch": 0.2592442713339066, "grad_norm": 13.943140029907227, "learning_rate": 8.516042531218633e-06, "loss": 0.5949, "step": 5125 }, { "epoch": 0.2592948555819718, "grad_norm": 12.565858840942383, "learning_rate": 8.515471840293998e-06, "loss": 0.5529, "step": 5126 }, { "epoch": 0.25934543983003694, "grad_norm": 14.11624526977539, "learning_rate": 8.514901058784146e-06, "loss": 0.5435, "step": 5127 }, { "epoch": 0.2593960240781021, "grad_norm": 15.841254234313965, "learning_rate": 8.514330186703783e-06, "loss": 0.5746, "step": 5128 }, { "epoch": 0.2594466083261672, "grad_norm": 7.279994964599609, "learning_rate": 8.513759224067619e-06, "loss": 0.4745, "step": 5129 }, { "epoch": 0.25949719257423237, "grad_norm": 17.469045639038086, "learning_rate": 8.513188170890368e-06, "loss": 0.651, "step": 5130 }, { "epoch": 0.2595477768222975, "grad_norm": 13.652303695678711, "learning_rate": 8.512617027186742e-06, "loss": 0.6078, "step": 5131 }, { "epoch": 0.2595983610703627, "grad_norm": 12.736080169677734, "learning_rate": 8.512045792971461e-06, "loss": 0.6052, "step": 5132 }, { "epoch": 0.25964894531842786, "grad_norm": 10.234519958496094, "learning_rate": 8.511474468259243e-06, "loss": 0.582, "step": 5133 }, { "epoch": 0.259699529566493, "grad_norm": 8.1219482421875, "learning_rate": 8.51090305306481e-06, "loss": 0.2937, "step": 5134 }, { "epoch": 0.25975011381455815, "grad_norm": 9.483189582824707, "learning_rate": 8.510331547402884e-06, "loss": 0.5538, "step": 5135 }, { "epoch": 0.2598006980626233, "grad_norm": 6.464048385620117, "learning_rate": 8.509759951288194e-06, "loss": 0.4604, "step": 5136 }, { "epoch": 0.25985128231068844, "grad_norm": 9.998312950134277, "learning_rate": 8.509188264735466e-06, "loss": 0.4715, "step": 5137 }, { "epoch": 0.2599018665587536, "grad_norm": 10.27918529510498, "learning_rate": 8.508616487759434e-06, "loss": 0.4165, "step": 5138 }, { "epoch": 0.2599524508068188, "grad_norm": 6.134899139404297, "learning_rate": 8.50804462037483e-06, "loss": 0.5149, "step": 5139 }, { "epoch": 0.2600030350548839, "grad_norm": 8.19686222076416, "learning_rate": 8.507472662596388e-06, "loss": 0.4586, "step": 5140 }, { "epoch": 0.26005361930294907, "grad_norm": 6.632524490356445, "learning_rate": 8.506900614438848e-06, "loss": 0.4026, "step": 5141 }, { "epoch": 0.2601042035510142, "grad_norm": 6.674583911895752, "learning_rate": 8.50632847591695e-06, "loss": 0.4904, "step": 5142 }, { "epoch": 0.26015478779907936, "grad_norm": 7.5147905349731445, "learning_rate": 8.505756247045438e-06, "loss": 0.4538, "step": 5143 }, { "epoch": 0.2602053720471445, "grad_norm": 7.980885028839111, "learning_rate": 8.505183927839053e-06, "loss": 0.3648, "step": 5144 }, { "epoch": 0.26025595629520965, "grad_norm": 4.83426570892334, "learning_rate": 8.504611518312545e-06, "loss": 0.3967, "step": 5145 }, { "epoch": 0.26030654054327484, "grad_norm": 7.002583026885986, "learning_rate": 8.504039018480665e-06, "loss": 0.5149, "step": 5146 }, { "epoch": 0.26035712479134, "grad_norm": 5.753013610839844, "learning_rate": 8.503466428358161e-06, "loss": 0.4122, "step": 5147 }, { "epoch": 0.26040770903940513, "grad_norm": 9.2612886428833, "learning_rate": 8.50289374795979e-06, "loss": 0.395, "step": 5148 }, { "epoch": 0.2604582932874703, "grad_norm": 11.775895118713379, "learning_rate": 8.502320977300309e-06, "loss": 0.5385, "step": 5149 }, { "epoch": 0.2605088775355354, "grad_norm": 8.003339767456055, "learning_rate": 8.501748116394475e-06, "loss": 0.4896, "step": 5150 }, { "epoch": 0.26055946178360057, "grad_norm": 7.727896690368652, "learning_rate": 8.50117516525705e-06, "loss": 0.4376, "step": 5151 }, { "epoch": 0.26061004603166577, "grad_norm": 5.787542819976807, "learning_rate": 8.500602123902797e-06, "loss": 0.4149, "step": 5152 }, { "epoch": 0.2606606302797309, "grad_norm": 5.224888801574707, "learning_rate": 8.500028992346484e-06, "loss": 0.4499, "step": 5153 }, { "epoch": 0.26071121452779605, "grad_norm": 8.353772163391113, "learning_rate": 8.499455770602878e-06, "loss": 0.5335, "step": 5154 }, { "epoch": 0.2607617987758612, "grad_norm": 8.636323928833008, "learning_rate": 8.498882458686749e-06, "loss": 0.3989, "step": 5155 }, { "epoch": 0.26081238302392634, "grad_norm": 9.530654907226562, "learning_rate": 8.49830905661287e-06, "loss": 0.4628, "step": 5156 }, { "epoch": 0.2608629672719915, "grad_norm": 7.962884902954102, "learning_rate": 8.497735564396016e-06, "loss": 0.3361, "step": 5157 }, { "epoch": 0.26091355152005663, "grad_norm": 8.318962097167969, "learning_rate": 8.497161982050965e-06, "loss": 0.3994, "step": 5158 }, { "epoch": 0.26096413576812183, "grad_norm": 10.317829132080078, "learning_rate": 8.496588309592497e-06, "loss": 0.332, "step": 5159 }, { "epoch": 0.261014720016187, "grad_norm": 8.428543090820312, "learning_rate": 8.496014547035395e-06, "loss": 0.41, "step": 5160 }, { "epoch": 0.2610653042642521, "grad_norm": 8.082536697387695, "learning_rate": 8.495440694394442e-06, "loss": 0.4735, "step": 5161 }, { "epoch": 0.26111588851231726, "grad_norm": 7.729605197906494, "learning_rate": 8.494866751684425e-06, "loss": 0.4171, "step": 5162 }, { "epoch": 0.2611664727603824, "grad_norm": 6.331325531005859, "learning_rate": 8.494292718920131e-06, "loss": 0.4618, "step": 5163 }, { "epoch": 0.26121705700844755, "grad_norm": 10.229354858398438, "learning_rate": 8.493718596116358e-06, "loss": 0.4225, "step": 5164 }, { "epoch": 0.2612676412565127, "grad_norm": 8.78632926940918, "learning_rate": 8.493144383287893e-06, "loss": 0.5259, "step": 5165 }, { "epoch": 0.2613182255045779, "grad_norm": 8.646605491638184, "learning_rate": 8.492570080449534e-06, "loss": 0.48, "step": 5166 }, { "epoch": 0.26136880975264304, "grad_norm": 11.814103126525879, "learning_rate": 8.491995687616082e-06, "loss": 0.4485, "step": 5167 }, { "epoch": 0.2614193940007082, "grad_norm": 11.2662992477417, "learning_rate": 8.491421204802335e-06, "loss": 0.56, "step": 5168 }, { "epoch": 0.2614699782487733, "grad_norm": 11.303561210632324, "learning_rate": 8.490846632023096e-06, "loss": 0.4071, "step": 5169 }, { "epoch": 0.26152056249683847, "grad_norm": 6.315319061279297, "learning_rate": 8.490271969293171e-06, "loss": 0.3196, "step": 5170 }, { "epoch": 0.2615711467449036, "grad_norm": 8.158453941345215, "learning_rate": 8.489697216627368e-06, "loss": 0.3732, "step": 5171 }, { "epoch": 0.2616217309929688, "grad_norm": 6.852973461151123, "learning_rate": 8.489122374040496e-06, "loss": 0.4924, "step": 5172 }, { "epoch": 0.26167231524103396, "grad_norm": 6.806468486785889, "learning_rate": 8.48854744154737e-06, "loss": 0.3275, "step": 5173 }, { "epoch": 0.2617228994890991, "grad_norm": 8.169264793395996, "learning_rate": 8.487972419162801e-06, "loss": 0.4207, "step": 5174 }, { "epoch": 0.26177348373716425, "grad_norm": 8.52298355102539, "learning_rate": 8.487397306901609e-06, "loss": 0.3631, "step": 5175 }, { "epoch": 0.2618240679852294, "grad_norm": 5.903777599334717, "learning_rate": 8.48682210477861e-06, "loss": 0.331, "step": 5176 }, { "epoch": 0.26187465223329454, "grad_norm": 7.710724353790283, "learning_rate": 8.48624681280863e-06, "loss": 0.4863, "step": 5177 }, { "epoch": 0.2619252364813597, "grad_norm": 12.7266263961792, "learning_rate": 8.485671431006488e-06, "loss": 0.5935, "step": 5178 }, { "epoch": 0.2619758207294249, "grad_norm": 6.415971279144287, "learning_rate": 8.485095959387013e-06, "loss": 0.3137, "step": 5179 }, { "epoch": 0.26202640497749, "grad_norm": 10.051530838012695, "learning_rate": 8.484520397965033e-06, "loss": 0.4175, "step": 5180 }, { "epoch": 0.26207698922555517, "grad_norm": 10.97966194152832, "learning_rate": 8.483944746755382e-06, "loss": 0.4361, "step": 5181 }, { "epoch": 0.2621275734736203, "grad_norm": 7.023282527923584, "learning_rate": 8.483369005772886e-06, "loss": 0.4423, "step": 5182 }, { "epoch": 0.26217815772168546, "grad_norm": 11.032176971435547, "learning_rate": 8.482793175032386e-06, "loss": 0.5444, "step": 5183 }, { "epoch": 0.2622287419697506, "grad_norm": 11.622040748596191, "learning_rate": 8.482217254548716e-06, "loss": 0.3889, "step": 5184 }, { "epoch": 0.26227932621781574, "grad_norm": 6.545321464538574, "learning_rate": 8.48164124433672e-06, "loss": 0.2646, "step": 5185 }, { "epoch": 0.26232991046588094, "grad_norm": 12.840147018432617, "learning_rate": 8.48106514441124e-06, "loss": 0.4671, "step": 5186 }, { "epoch": 0.2623804947139461, "grad_norm": 12.277888298034668, "learning_rate": 8.480488954787119e-06, "loss": 0.4484, "step": 5187 }, { "epoch": 0.26243107896201123, "grad_norm": 10.604024887084961, "learning_rate": 8.479912675479203e-06, "loss": 0.4002, "step": 5188 }, { "epoch": 0.2624816632100764, "grad_norm": 14.052653312683105, "learning_rate": 8.479336306502345e-06, "loss": 0.3931, "step": 5189 }, { "epoch": 0.2625322474581415, "grad_norm": 7.562033653259277, "learning_rate": 8.478759847871392e-06, "loss": 0.3559, "step": 5190 }, { "epoch": 0.26258283170620667, "grad_norm": 13.90158748626709, "learning_rate": 8.478183299601201e-06, "loss": 0.5314, "step": 5191 }, { "epoch": 0.26263341595427186, "grad_norm": 9.911154747009277, "learning_rate": 8.477606661706627e-06, "loss": 0.4984, "step": 5192 }, { "epoch": 0.262684000202337, "grad_norm": 8.03626823425293, "learning_rate": 8.47702993420253e-06, "loss": 0.4256, "step": 5193 }, { "epoch": 0.26273458445040215, "grad_norm": 10.356060981750488, "learning_rate": 8.476453117103771e-06, "loss": 0.3848, "step": 5194 }, { "epoch": 0.2627851686984673, "grad_norm": 12.893452644348145, "learning_rate": 8.475876210425211e-06, "loss": 0.597, "step": 5195 }, { "epoch": 0.26283575294653244, "grad_norm": 8.597770690917969, "learning_rate": 8.475299214181717e-06, "loss": 0.3912, "step": 5196 }, { "epoch": 0.2628863371945976, "grad_norm": 12.76254940032959, "learning_rate": 8.474722128388157e-06, "loss": 0.5247, "step": 5197 }, { "epoch": 0.26293692144266273, "grad_norm": 11.757110595703125, "learning_rate": 8.4741449530594e-06, "loss": 0.6357, "step": 5198 }, { "epoch": 0.26298750569072793, "grad_norm": 8.641827583312988, "learning_rate": 8.473567688210318e-06, "loss": 0.475, "step": 5199 }, { "epoch": 0.2630380899387931, "grad_norm": 6.478572845458984, "learning_rate": 8.472990333855789e-06, "loss": 0.3413, "step": 5200 }, { "epoch": 0.2630886741868582, "grad_norm": 6.059112548828125, "learning_rate": 8.472412890010686e-06, "loss": 0.4154, "step": 5201 }, { "epoch": 0.26313925843492336, "grad_norm": 4.889173984527588, "learning_rate": 8.47183535668989e-06, "loss": 0.4383, "step": 5202 }, { "epoch": 0.2631898426829885, "grad_norm": 7.345656871795654, "learning_rate": 8.471257733908282e-06, "loss": 0.5022, "step": 5203 }, { "epoch": 0.26324042693105365, "grad_norm": 6.305444240570068, "learning_rate": 8.470680021680749e-06, "loss": 0.494, "step": 5204 }, { "epoch": 0.26329101117911885, "grad_norm": 7.530946731567383, "learning_rate": 8.470102220022174e-06, "loss": 0.3868, "step": 5205 }, { "epoch": 0.263341595427184, "grad_norm": 6.172507286071777, "learning_rate": 8.469524328947445e-06, "loss": 0.3952, "step": 5206 }, { "epoch": 0.26339217967524914, "grad_norm": 6.046676158905029, "learning_rate": 8.468946348471457e-06, "loss": 0.4297, "step": 5207 }, { "epoch": 0.2634427639233143, "grad_norm": 6.014155864715576, "learning_rate": 8.468368278609098e-06, "loss": 0.4308, "step": 5208 }, { "epoch": 0.2634933481713794, "grad_norm": 6.973775386810303, "learning_rate": 8.467790119375267e-06, "loss": 0.4602, "step": 5209 }, { "epoch": 0.26354393241944457, "grad_norm": 7.8605828285217285, "learning_rate": 8.467211870784862e-06, "loss": 0.422, "step": 5210 }, { "epoch": 0.2635945166675097, "grad_norm": 4.995382308959961, "learning_rate": 8.46663353285278e-06, "loss": 0.4245, "step": 5211 }, { "epoch": 0.2636451009155749, "grad_norm": 8.459527969360352, "learning_rate": 8.466055105593926e-06, "loss": 0.392, "step": 5212 }, { "epoch": 0.26369568516364006, "grad_norm": 9.794113159179688, "learning_rate": 8.465476589023205e-06, "loss": 0.4816, "step": 5213 }, { "epoch": 0.2637462694117052, "grad_norm": 8.589969635009766, "learning_rate": 8.46489798315552e-06, "loss": 0.4037, "step": 5214 }, { "epoch": 0.26379685365977035, "grad_norm": 5.276312351226807, "learning_rate": 8.464319288005784e-06, "loss": 0.4097, "step": 5215 }, { "epoch": 0.2638474379078355, "grad_norm": 14.695520401000977, "learning_rate": 8.46374050358891e-06, "loss": 0.46, "step": 5216 }, { "epoch": 0.26389802215590064, "grad_norm": 8.263490676879883, "learning_rate": 8.463161629919808e-06, "loss": 0.4057, "step": 5217 }, { "epoch": 0.2639486064039658, "grad_norm": 10.550346374511719, "learning_rate": 8.462582667013394e-06, "loss": 0.413, "step": 5218 }, { "epoch": 0.263999190652031, "grad_norm": 9.983317375183105, "learning_rate": 8.46200361488459e-06, "loss": 0.5137, "step": 5219 }, { "epoch": 0.2640497749000961, "grad_norm": 12.408967018127441, "learning_rate": 8.461424473548315e-06, "loss": 0.4286, "step": 5220 }, { "epoch": 0.26410035914816127, "grad_norm": 9.912548065185547, "learning_rate": 8.460845243019492e-06, "loss": 0.6377, "step": 5221 }, { "epoch": 0.2641509433962264, "grad_norm": 7.547399997711182, "learning_rate": 8.460265923313047e-06, "loss": 0.447, "step": 5222 }, { "epoch": 0.26420152764429156, "grad_norm": 9.322226524353027, "learning_rate": 8.459686514443905e-06, "loss": 0.3491, "step": 5223 }, { "epoch": 0.2642521118923567, "grad_norm": 9.2373628616333, "learning_rate": 8.459107016426999e-06, "loss": 0.5216, "step": 5224 }, { "epoch": 0.2643026961404219, "grad_norm": 8.455662727355957, "learning_rate": 8.45852742927726e-06, "loss": 0.3504, "step": 5225 }, { "epoch": 0.26435328038848704, "grad_norm": 7.251514434814453, "learning_rate": 8.457947753009625e-06, "loss": 0.2988, "step": 5226 }, { "epoch": 0.2644038646365522, "grad_norm": 8.663928985595703, "learning_rate": 8.457367987639027e-06, "loss": 0.3945, "step": 5227 }, { "epoch": 0.26445444888461733, "grad_norm": 12.294402122497559, "learning_rate": 8.456788133180407e-06, "loss": 0.4469, "step": 5228 }, { "epoch": 0.2645050331326825, "grad_norm": 8.395237922668457, "learning_rate": 8.456208189648707e-06, "loss": 0.4133, "step": 5229 }, { "epoch": 0.2645556173807476, "grad_norm": 8.56277084350586, "learning_rate": 8.455628157058867e-06, "loss": 0.5124, "step": 5230 }, { "epoch": 0.26460620162881276, "grad_norm": 9.219549179077148, "learning_rate": 8.455048035425841e-06, "loss": 0.6538, "step": 5231 }, { "epoch": 0.26465678587687796, "grad_norm": 11.805961608886719, "learning_rate": 8.45446782476457e-06, "loss": 0.5939, "step": 5232 }, { "epoch": 0.2647073701249431, "grad_norm": 10.853177070617676, "learning_rate": 8.453887525090008e-06, "loss": 0.4869, "step": 5233 }, { "epoch": 0.26475795437300825, "grad_norm": 8.359617233276367, "learning_rate": 8.453307136417104e-06, "loss": 0.5371, "step": 5234 }, { "epoch": 0.2648085386210734, "grad_norm": 10.52894401550293, "learning_rate": 8.452726658760819e-06, "loss": 0.5723, "step": 5235 }, { "epoch": 0.26485912286913854, "grad_norm": 10.627218246459961, "learning_rate": 8.452146092136108e-06, "loss": 0.6176, "step": 5236 }, { "epoch": 0.2649097071172037, "grad_norm": 8.099143981933594, "learning_rate": 8.45156543655793e-06, "loss": 0.5016, "step": 5237 }, { "epoch": 0.26496029136526883, "grad_norm": 12.51722240447998, "learning_rate": 8.450984692041246e-06, "loss": 0.6172, "step": 5238 }, { "epoch": 0.26501087561333403, "grad_norm": 5.173586368560791, "learning_rate": 8.450403858601025e-06, "loss": 0.433, "step": 5239 }, { "epoch": 0.2650614598613992, "grad_norm": 6.616931438446045, "learning_rate": 8.449822936252227e-06, "loss": 0.3753, "step": 5240 }, { "epoch": 0.2651120441094643, "grad_norm": 7.8103346824646, "learning_rate": 8.449241925009827e-06, "loss": 0.4032, "step": 5241 }, { "epoch": 0.26516262835752946, "grad_norm": 11.63348388671875, "learning_rate": 8.448660824888794e-06, "loss": 0.5042, "step": 5242 }, { "epoch": 0.2652132126055946, "grad_norm": 9.643542289733887, "learning_rate": 8.4480796359041e-06, "loss": 0.5065, "step": 5243 }, { "epoch": 0.26526379685365975, "grad_norm": 5.927483558654785, "learning_rate": 8.447498358070723e-06, "loss": 0.315, "step": 5244 }, { "epoch": 0.26531438110172495, "grad_norm": 7.649847030639648, "learning_rate": 8.44691699140364e-06, "loss": 0.4598, "step": 5245 }, { "epoch": 0.2653649653497901, "grad_norm": 6.519077301025391, "learning_rate": 8.446335535917834e-06, "loss": 0.4479, "step": 5246 }, { "epoch": 0.26541554959785524, "grad_norm": 9.1981782913208, "learning_rate": 8.445753991628284e-06, "loss": 0.4864, "step": 5247 }, { "epoch": 0.2654661338459204, "grad_norm": 5.435657024383545, "learning_rate": 8.445172358549975e-06, "loss": 0.3674, "step": 5248 }, { "epoch": 0.2655167180939855, "grad_norm": 6.884942054748535, "learning_rate": 8.444590636697898e-06, "loss": 0.3714, "step": 5249 }, { "epoch": 0.26556730234205067, "grad_norm": 8.488031387329102, "learning_rate": 8.44400882608704e-06, "loss": 0.3968, "step": 5250 }, { "epoch": 0.2656178865901158, "grad_norm": 7.11277437210083, "learning_rate": 8.443426926732392e-06, "loss": 0.3102, "step": 5251 }, { "epoch": 0.265668470838181, "grad_norm": 6.735693454742432, "learning_rate": 8.442844938648951e-06, "loss": 0.4263, "step": 5252 }, { "epoch": 0.26571905508624616, "grad_norm": 12.541386604309082, "learning_rate": 8.44226286185171e-06, "loss": 0.5976, "step": 5253 }, { "epoch": 0.2657696393343113, "grad_norm": 7.790648937225342, "learning_rate": 8.441680696355672e-06, "loss": 0.4757, "step": 5254 }, { "epoch": 0.26582022358237645, "grad_norm": 16.180511474609375, "learning_rate": 8.441098442175834e-06, "loss": 0.3819, "step": 5255 }, { "epoch": 0.2658708078304416, "grad_norm": 12.101717948913574, "learning_rate": 8.4405160993272e-06, "loss": 0.4579, "step": 5256 }, { "epoch": 0.26592139207850674, "grad_norm": 8.960000038146973, "learning_rate": 8.439933667824776e-06, "loss": 0.4911, "step": 5257 }, { "epoch": 0.2659719763265719, "grad_norm": 8.809356689453125, "learning_rate": 8.439351147683572e-06, "loss": 0.5075, "step": 5258 }, { "epoch": 0.2660225605746371, "grad_norm": 10.588370323181152, "learning_rate": 8.438768538918596e-06, "loss": 0.5333, "step": 5259 }, { "epoch": 0.2660731448227022, "grad_norm": 5.8151326179504395, "learning_rate": 8.43818584154486e-06, "loss": 0.4102, "step": 5260 }, { "epoch": 0.26612372907076737, "grad_norm": 7.635757923126221, "learning_rate": 8.43760305557738e-06, "loss": 0.3478, "step": 5261 }, { "epoch": 0.2661743133188325, "grad_norm": 13.04977035522461, "learning_rate": 8.437020181031172e-06, "loss": 0.4986, "step": 5262 }, { "epoch": 0.26622489756689766, "grad_norm": 11.460091590881348, "learning_rate": 8.436437217921256e-06, "loss": 0.4161, "step": 5263 }, { "epoch": 0.2662754818149628, "grad_norm": 7.544009685516357, "learning_rate": 8.435854166262653e-06, "loss": 0.4434, "step": 5264 }, { "epoch": 0.266326066063028, "grad_norm": 6.6070122718811035, "learning_rate": 8.435271026070388e-06, "loss": 0.4074, "step": 5265 }, { "epoch": 0.26637665031109314, "grad_norm": 10.983267784118652, "learning_rate": 8.434687797359485e-06, "loss": 0.5421, "step": 5266 }, { "epoch": 0.2664272345591583, "grad_norm": 5.844901084899902, "learning_rate": 8.434104480144975e-06, "loss": 0.3499, "step": 5267 }, { "epoch": 0.26647781880722343, "grad_norm": 9.295108795166016, "learning_rate": 8.433521074441886e-06, "loss": 0.5008, "step": 5268 }, { "epoch": 0.2665284030552886, "grad_norm": 11.47480297088623, "learning_rate": 8.432937580265253e-06, "loss": 0.4838, "step": 5269 }, { "epoch": 0.2665789873033537, "grad_norm": 7.263123989105225, "learning_rate": 8.43235399763011e-06, "loss": 0.3955, "step": 5270 }, { "epoch": 0.26662957155141886, "grad_norm": 5.724973678588867, "learning_rate": 8.431770326551497e-06, "loss": 0.4435, "step": 5271 }, { "epoch": 0.26668015579948406, "grad_norm": 6.896169185638428, "learning_rate": 8.431186567044448e-06, "loss": 0.3727, "step": 5272 }, { "epoch": 0.2667307400475492, "grad_norm": 10.295122146606445, "learning_rate": 8.43060271912401e-06, "loss": 0.4483, "step": 5273 }, { "epoch": 0.26678132429561435, "grad_norm": 8.25963020324707, "learning_rate": 8.430018782805227e-06, "loss": 0.4284, "step": 5274 }, { "epoch": 0.2668319085436795, "grad_norm": 7.760621070861816, "learning_rate": 8.429434758103145e-06, "loss": 0.4766, "step": 5275 }, { "epoch": 0.26688249279174464, "grad_norm": 6.825159072875977, "learning_rate": 8.428850645032813e-06, "loss": 0.4874, "step": 5276 }, { "epoch": 0.2669330770398098, "grad_norm": 7.232582092285156, "learning_rate": 8.428266443609282e-06, "loss": 0.3768, "step": 5277 }, { "epoch": 0.26698366128787493, "grad_norm": 9.268867492675781, "learning_rate": 8.427682153847605e-06, "loss": 0.5803, "step": 5278 }, { "epoch": 0.26703424553594013, "grad_norm": 6.249511241912842, "learning_rate": 8.427097775762837e-06, "loss": 0.47, "step": 5279 }, { "epoch": 0.2670848297840053, "grad_norm": 7.776778697967529, "learning_rate": 8.42651330937004e-06, "loss": 0.4806, "step": 5280 }, { "epoch": 0.2671354140320704, "grad_norm": 8.287246704101562, "learning_rate": 8.425928754684269e-06, "loss": 0.4683, "step": 5281 }, { "epoch": 0.26718599828013556, "grad_norm": 8.720663070678711, "learning_rate": 8.425344111720589e-06, "loss": 0.4346, "step": 5282 }, { "epoch": 0.2672365825282007, "grad_norm": 12.509140014648438, "learning_rate": 8.424759380494065e-06, "loss": 0.526, "step": 5283 }, { "epoch": 0.26728716677626585, "grad_norm": 9.614836692810059, "learning_rate": 8.424174561019764e-06, "loss": 0.4918, "step": 5284 }, { "epoch": 0.26733775102433105, "grad_norm": 5.74180269241333, "learning_rate": 8.423589653312757e-06, "loss": 0.4412, "step": 5285 }, { "epoch": 0.2673883352723962, "grad_norm": 8.420989036560059, "learning_rate": 8.423004657388111e-06, "loss": 0.5101, "step": 5286 }, { "epoch": 0.26743891952046134, "grad_norm": 7.040156364440918, "learning_rate": 8.422419573260905e-06, "loss": 0.4056, "step": 5287 }, { "epoch": 0.2674895037685265, "grad_norm": 12.890826225280762, "learning_rate": 8.421834400946214e-06, "loss": 0.4399, "step": 5288 }, { "epoch": 0.2675400880165916, "grad_norm": 7.918107509613037, "learning_rate": 8.421249140459114e-06, "loss": 0.548, "step": 5289 }, { "epoch": 0.26759067226465677, "grad_norm": 5.32254695892334, "learning_rate": 8.420663791814687e-06, "loss": 0.4191, "step": 5290 }, { "epoch": 0.2676412565127219, "grad_norm": 6.054841041564941, "learning_rate": 8.420078355028018e-06, "loss": 0.3842, "step": 5291 }, { "epoch": 0.2676918407607871, "grad_norm": 15.886216163635254, "learning_rate": 8.419492830114189e-06, "loss": 0.6036, "step": 5292 }, { "epoch": 0.26774242500885226, "grad_norm": 6.954150199890137, "learning_rate": 8.41890721708829e-06, "loss": 0.4073, "step": 5293 }, { "epoch": 0.2677930092569174, "grad_norm": 8.200471878051758, "learning_rate": 8.418321515965408e-06, "loss": 0.4951, "step": 5294 }, { "epoch": 0.26784359350498255, "grad_norm": 14.612700462341309, "learning_rate": 8.41773572676064e-06, "loss": 0.5644, "step": 5295 }, { "epoch": 0.2678941777530477, "grad_norm": 6.399186134338379, "learning_rate": 8.417149849489077e-06, "loss": 0.4473, "step": 5296 }, { "epoch": 0.26794476200111284, "grad_norm": 11.536606788635254, "learning_rate": 8.416563884165817e-06, "loss": 0.5206, "step": 5297 }, { "epoch": 0.267995346249178, "grad_norm": 8.338430404663086, "learning_rate": 8.415977830805955e-06, "loss": 0.3055, "step": 5298 }, { "epoch": 0.2680459304972432, "grad_norm": 6.495667934417725, "learning_rate": 8.415391689424596e-06, "loss": 0.5108, "step": 5299 }, { "epoch": 0.2680965147453083, "grad_norm": 4.709325790405273, "learning_rate": 8.414805460036846e-06, "loss": 0.4529, "step": 5300 }, { "epoch": 0.26814709899337347, "grad_norm": 8.235663414001465, "learning_rate": 8.414219142657804e-06, "loss": 0.5235, "step": 5301 }, { "epoch": 0.2681976832414386, "grad_norm": 5.950257301330566, "learning_rate": 8.413632737302583e-06, "loss": 0.4069, "step": 5302 }, { "epoch": 0.26824826748950376, "grad_norm": 5.942375183105469, "learning_rate": 8.41304624398629e-06, "loss": 0.428, "step": 5303 }, { "epoch": 0.2682988517375689, "grad_norm": 10.507915496826172, "learning_rate": 8.412459662724042e-06, "loss": 0.4108, "step": 5304 }, { "epoch": 0.2683494359856341, "grad_norm": 5.147900104522705, "learning_rate": 8.41187299353095e-06, "loss": 0.3363, "step": 5305 }, { "epoch": 0.26840002023369924, "grad_norm": 6.1329498291015625, "learning_rate": 8.411286236422129e-06, "loss": 0.4019, "step": 5306 }, { "epoch": 0.2684506044817644, "grad_norm": 9.768769264221191, "learning_rate": 8.410699391412705e-06, "loss": 0.4091, "step": 5307 }, { "epoch": 0.26850118872982953, "grad_norm": 7.814202308654785, "learning_rate": 8.410112458517793e-06, "loss": 0.4747, "step": 5308 }, { "epoch": 0.2685517729778947, "grad_norm": 7.6030426025390625, "learning_rate": 8.409525437752522e-06, "loss": 0.5104, "step": 5309 }, { "epoch": 0.2686023572259598, "grad_norm": 8.608837127685547, "learning_rate": 8.408938329132017e-06, "loss": 0.3867, "step": 5310 }, { "epoch": 0.26865294147402496, "grad_norm": 16.68494987487793, "learning_rate": 8.408351132671402e-06, "loss": 0.4607, "step": 5311 }, { "epoch": 0.26870352572209016, "grad_norm": 8.569564819335938, "learning_rate": 8.407763848385813e-06, "loss": 0.4657, "step": 5312 }, { "epoch": 0.2687541099701553, "grad_norm": 13.459155082702637, "learning_rate": 8.407176476290378e-06, "loss": 0.5517, "step": 5313 }, { "epoch": 0.26880469421822045, "grad_norm": 6.594313144683838, "learning_rate": 8.406589016400237e-06, "loss": 0.4177, "step": 5314 }, { "epoch": 0.2688552784662856, "grad_norm": 11.893468856811523, "learning_rate": 8.406001468730526e-06, "loss": 0.5, "step": 5315 }, { "epoch": 0.26890586271435074, "grad_norm": 10.918041229248047, "learning_rate": 8.405413833296383e-06, "loss": 0.4125, "step": 5316 }, { "epoch": 0.2689564469624159, "grad_norm": 30.813739776611328, "learning_rate": 8.40482611011295e-06, "loss": 0.6827, "step": 5317 }, { "epoch": 0.26900703121048103, "grad_norm": 8.531010627746582, "learning_rate": 8.404238299195375e-06, "loss": 0.4192, "step": 5318 }, { "epoch": 0.26905761545854623, "grad_norm": 7.029937267303467, "learning_rate": 8.4036504005588e-06, "loss": 0.4336, "step": 5319 }, { "epoch": 0.2691081997066114, "grad_norm": 7.558659553527832, "learning_rate": 8.403062414218375e-06, "loss": 0.4194, "step": 5320 }, { "epoch": 0.2691587839546765, "grad_norm": 10.122380256652832, "learning_rate": 8.40247434018925e-06, "loss": 0.4758, "step": 5321 }, { "epoch": 0.26920936820274166, "grad_norm": 7.542631149291992, "learning_rate": 8.401886178486583e-06, "loss": 0.4777, "step": 5322 }, { "epoch": 0.2692599524508068, "grad_norm": 13.583324432373047, "learning_rate": 8.401297929125524e-06, "loss": 0.5207, "step": 5323 }, { "epoch": 0.26931053669887195, "grad_norm": 7.014463424682617, "learning_rate": 8.400709592121232e-06, "loss": 0.3764, "step": 5324 }, { "epoch": 0.26936112094693715, "grad_norm": 6.5508551597595215, "learning_rate": 8.40012116748887e-06, "loss": 0.3775, "step": 5325 }, { "epoch": 0.2694117051950023, "grad_norm": 10.645429611206055, "learning_rate": 8.399532655243596e-06, "loss": 0.4546, "step": 5326 }, { "epoch": 0.26946228944306744, "grad_norm": 5.532618045806885, "learning_rate": 8.398944055400578e-06, "loss": 0.3752, "step": 5327 }, { "epoch": 0.2695128736911326, "grad_norm": 12.08182430267334, "learning_rate": 8.398355367974983e-06, "loss": 0.5094, "step": 5328 }, { "epoch": 0.2695634579391977, "grad_norm": 5.5471696853637695, "learning_rate": 8.397766592981976e-06, "loss": 0.4017, "step": 5329 }, { "epoch": 0.26961404218726287, "grad_norm": 11.503079414367676, "learning_rate": 8.397177730436731e-06, "loss": 0.4966, "step": 5330 }, { "epoch": 0.269664626435328, "grad_norm": 9.062625885009766, "learning_rate": 8.396588780354425e-06, "loss": 0.4827, "step": 5331 }, { "epoch": 0.2697152106833932, "grad_norm": 6.910575866699219, "learning_rate": 8.395999742750226e-06, "loss": 0.3061, "step": 5332 }, { "epoch": 0.26976579493145836, "grad_norm": 8.475658416748047, "learning_rate": 8.395410617639317e-06, "loss": 0.5011, "step": 5333 }, { "epoch": 0.2698163791795235, "grad_norm": 7.070809364318848, "learning_rate": 8.39482140503688e-06, "loss": 0.4589, "step": 5334 }, { "epoch": 0.26986696342758865, "grad_norm": 16.686405181884766, "learning_rate": 8.394232104958093e-06, "loss": 0.5407, "step": 5335 }, { "epoch": 0.2699175476756538, "grad_norm": 10.945512771606445, "learning_rate": 8.393642717418145e-06, "loss": 0.4745, "step": 5336 }, { "epoch": 0.26996813192371893, "grad_norm": 9.667449951171875, "learning_rate": 8.39305324243222e-06, "loss": 0.4591, "step": 5337 }, { "epoch": 0.2700187161717841, "grad_norm": 11.904147148132324, "learning_rate": 8.392463680015508e-06, "loss": 0.4917, "step": 5338 }, { "epoch": 0.2700693004198493, "grad_norm": 8.466934204101562, "learning_rate": 8.391874030183201e-06, "loss": 0.4482, "step": 5339 }, { "epoch": 0.2701198846679144, "grad_norm": 13.594059944152832, "learning_rate": 8.391284292950495e-06, "loss": 0.5512, "step": 5340 }, { "epoch": 0.27017046891597957, "grad_norm": 10.691641807556152, "learning_rate": 8.390694468332581e-06, "loss": 0.4728, "step": 5341 }, { "epoch": 0.2702210531640447, "grad_norm": 5.459009170532227, "learning_rate": 8.390104556344662e-06, "loss": 0.4008, "step": 5342 }, { "epoch": 0.27027163741210986, "grad_norm": 7.588048458099365, "learning_rate": 8.389514557001939e-06, "loss": 0.4588, "step": 5343 }, { "epoch": 0.270322221660175, "grad_norm": 7.026385307312012, "learning_rate": 8.38892447031961e-06, "loss": 0.4055, "step": 5344 }, { "epoch": 0.2703728059082402, "grad_norm": 7.94921350479126, "learning_rate": 8.388334296312884e-06, "loss": 0.5495, "step": 5345 }, { "epoch": 0.27042339015630534, "grad_norm": 7.044497013092041, "learning_rate": 8.387744034996967e-06, "loss": 0.4412, "step": 5346 }, { "epoch": 0.2704739744043705, "grad_norm": 7.1902546882629395, "learning_rate": 8.38715368638707e-06, "loss": 0.4815, "step": 5347 }, { "epoch": 0.27052455865243563, "grad_norm": 10.071913719177246, "learning_rate": 8.386563250498402e-06, "loss": 0.4928, "step": 5348 }, { "epoch": 0.2705751429005008, "grad_norm": 5.290224075317383, "learning_rate": 8.385972727346183e-06, "loss": 0.3836, "step": 5349 }, { "epoch": 0.2706257271485659, "grad_norm": 4.381806373596191, "learning_rate": 8.385382116945621e-06, "loss": 0.3912, "step": 5350 }, { "epoch": 0.27067631139663106, "grad_norm": 7.5235419273376465, "learning_rate": 8.384791419311941e-06, "loss": 0.3386, "step": 5351 }, { "epoch": 0.27072689564469626, "grad_norm": 7.579514503479004, "learning_rate": 8.384200634460363e-06, "loss": 0.3976, "step": 5352 }, { "epoch": 0.2707774798927614, "grad_norm": 5.009066104888916, "learning_rate": 8.383609762406106e-06, "loss": 0.4367, "step": 5353 }, { "epoch": 0.27082806414082655, "grad_norm": 8.783549308776855, "learning_rate": 8.3830188031644e-06, "loss": 0.4475, "step": 5354 }, { "epoch": 0.2708786483888917, "grad_norm": 6.785905838012695, "learning_rate": 8.382427756750469e-06, "loss": 0.4606, "step": 5355 }, { "epoch": 0.27092923263695684, "grad_norm": 14.590420722961426, "learning_rate": 8.381836623179546e-06, "loss": 0.6791, "step": 5356 }, { "epoch": 0.270979816885022, "grad_norm": 7.020775318145752, "learning_rate": 8.381245402466861e-06, "loss": 0.4542, "step": 5357 }, { "epoch": 0.2710304011330872, "grad_norm": 7.985334396362305, "learning_rate": 8.38065409462765e-06, "loss": 0.5256, "step": 5358 }, { "epoch": 0.27108098538115233, "grad_norm": 12.733025550842285, "learning_rate": 8.380062699677147e-06, "loss": 0.4082, "step": 5359 }, { "epoch": 0.2711315696292175, "grad_norm": 7.492889404296875, "learning_rate": 8.379471217630593e-06, "loss": 0.47, "step": 5360 }, { "epoch": 0.2711821538772826, "grad_norm": 8.044092178344727, "learning_rate": 8.37887964850323e-06, "loss": 0.5194, "step": 5361 }, { "epoch": 0.27123273812534776, "grad_norm": 9.242562294006348, "learning_rate": 8.378287992310299e-06, "loss": 0.3915, "step": 5362 }, { "epoch": 0.2712833223734129, "grad_norm": 8.043539047241211, "learning_rate": 8.377696249067047e-06, "loss": 0.4146, "step": 5363 }, { "epoch": 0.27133390662147805, "grad_norm": 7.636359214782715, "learning_rate": 8.37710441878872e-06, "loss": 0.4893, "step": 5364 }, { "epoch": 0.27138449086954325, "grad_norm": 7.664259910583496, "learning_rate": 8.376512501490567e-06, "loss": 0.4259, "step": 5365 }, { "epoch": 0.2714350751176084, "grad_norm": 6.769596099853516, "learning_rate": 8.375920497187845e-06, "loss": 0.4299, "step": 5366 }, { "epoch": 0.27148565936567354, "grad_norm": 7.2513532638549805, "learning_rate": 8.375328405895809e-06, "loss": 0.4483, "step": 5367 }, { "epoch": 0.2715362436137387, "grad_norm": 6.068502902984619, "learning_rate": 8.37473622762971e-06, "loss": 0.2722, "step": 5368 }, { "epoch": 0.2715868278618038, "grad_norm": 7.078057765960693, "learning_rate": 8.374143962404809e-06, "loss": 0.4124, "step": 5369 }, { "epoch": 0.27163741210986897, "grad_norm": 11.495575904846191, "learning_rate": 8.373551610236367e-06, "loss": 0.5052, "step": 5370 }, { "epoch": 0.2716879963579341, "grad_norm": 10.088532447814941, "learning_rate": 8.372959171139652e-06, "loss": 0.413, "step": 5371 }, { "epoch": 0.2717385806059993, "grad_norm": 10.080647468566895, "learning_rate": 8.372366645129924e-06, "loss": 0.2943, "step": 5372 }, { "epoch": 0.27178916485406446, "grad_norm": 8.29453182220459, "learning_rate": 8.371774032222454e-06, "loss": 0.4499, "step": 5373 }, { "epoch": 0.2718397491021296, "grad_norm": 7.324305057525635, "learning_rate": 8.37118133243251e-06, "loss": 0.4063, "step": 5374 }, { "epoch": 0.27189033335019475, "grad_norm": 7.843176364898682, "learning_rate": 8.370588545775367e-06, "loss": 0.468, "step": 5375 }, { "epoch": 0.2719409175982599, "grad_norm": 11.312411308288574, "learning_rate": 8.369995672266297e-06, "loss": 0.5521, "step": 5376 }, { "epoch": 0.27199150184632503, "grad_norm": 5.145047187805176, "learning_rate": 8.36940271192058e-06, "loss": 0.4208, "step": 5377 }, { "epoch": 0.27204208609439023, "grad_norm": 8.567181587219238, "learning_rate": 8.368809664753491e-06, "loss": 0.5132, "step": 5378 }, { "epoch": 0.2720926703424554, "grad_norm": 14.819189071655273, "learning_rate": 8.368216530780316e-06, "loss": 0.4896, "step": 5379 }, { "epoch": 0.2721432545905205, "grad_norm": 5.532103538513184, "learning_rate": 8.367623310016338e-06, "loss": 0.3932, "step": 5380 }, { "epoch": 0.27219383883858567, "grad_norm": 5.962536334991455, "learning_rate": 8.36703000247684e-06, "loss": 0.3706, "step": 5381 }, { "epoch": 0.2722444230866508, "grad_norm": 9.571455001831055, "learning_rate": 8.36643660817711e-06, "loss": 0.4384, "step": 5382 }, { "epoch": 0.27229500733471595, "grad_norm": 7.288273811340332, "learning_rate": 8.365843127132442e-06, "loss": 0.4835, "step": 5383 }, { "epoch": 0.2723455915827811, "grad_norm": 9.11156940460205, "learning_rate": 8.365249559358125e-06, "loss": 0.466, "step": 5384 }, { "epoch": 0.2723961758308463, "grad_norm": 7.61843204498291, "learning_rate": 8.364655904869458e-06, "loss": 0.4617, "step": 5385 }, { "epoch": 0.27244676007891144, "grad_norm": 10.302485466003418, "learning_rate": 8.364062163681734e-06, "loss": 0.3896, "step": 5386 }, { "epoch": 0.2724973443269766, "grad_norm": 10.366133689880371, "learning_rate": 8.363468335810254e-06, "loss": 0.3462, "step": 5387 }, { "epoch": 0.27254792857504173, "grad_norm": 12.200128555297852, "learning_rate": 8.362874421270316e-06, "loss": 0.5689, "step": 5388 }, { "epoch": 0.2725985128231069, "grad_norm": 12.246917724609375, "learning_rate": 8.36228042007723e-06, "loss": 0.4659, "step": 5389 }, { "epoch": 0.272649097071172, "grad_norm": 9.615516662597656, "learning_rate": 8.361686332246298e-06, "loss": 0.4971, "step": 5390 }, { "epoch": 0.27269968131923716, "grad_norm": 8.831957817077637, "learning_rate": 8.36109215779283e-06, "loss": 0.5663, "step": 5391 }, { "epoch": 0.27275026556730236, "grad_norm": 7.11870813369751, "learning_rate": 8.360497896732134e-06, "loss": 0.4194, "step": 5392 }, { "epoch": 0.2728008498153675, "grad_norm": 6.870093822479248, "learning_rate": 8.359903549079524e-06, "loss": 0.4113, "step": 5393 }, { "epoch": 0.27285143406343265, "grad_norm": 7.947272777557373, "learning_rate": 8.359309114850316e-06, "loss": 0.3371, "step": 5394 }, { "epoch": 0.2729020183114978, "grad_norm": 10.346086502075195, "learning_rate": 8.358714594059827e-06, "loss": 0.4443, "step": 5395 }, { "epoch": 0.27295260255956294, "grad_norm": 10.843132972717285, "learning_rate": 8.358119986723374e-06, "loss": 0.5128, "step": 5396 }, { "epoch": 0.2730031868076281, "grad_norm": 5.233538627624512, "learning_rate": 8.35752529285628e-06, "loss": 0.4459, "step": 5397 }, { "epoch": 0.2730537710556933, "grad_norm": 8.750540733337402, "learning_rate": 8.35693051247387e-06, "loss": 0.4067, "step": 5398 }, { "epoch": 0.27310435530375843, "grad_norm": 8.115045547485352, "learning_rate": 8.356335645591469e-06, "loss": 0.3799, "step": 5399 }, { "epoch": 0.2731549395518236, "grad_norm": 7.838285446166992, "learning_rate": 8.355740692224405e-06, "loss": 0.4676, "step": 5400 }, { "epoch": 0.2732055237998887, "grad_norm": 6.005498886108398, "learning_rate": 8.355145652388009e-06, "loss": 0.403, "step": 5401 }, { "epoch": 0.27325610804795386, "grad_norm": 8.967700004577637, "learning_rate": 8.354550526097613e-06, "loss": 0.5801, "step": 5402 }, { "epoch": 0.273306692296019, "grad_norm": 8.262856483459473, "learning_rate": 8.353955313368552e-06, "loss": 0.4223, "step": 5403 }, { "epoch": 0.27335727654408415, "grad_norm": 7.505570888519287, "learning_rate": 8.353360014216165e-06, "loss": 0.475, "step": 5404 }, { "epoch": 0.27340786079214935, "grad_norm": 6.329531669616699, "learning_rate": 8.352764628655789e-06, "loss": 0.4032, "step": 5405 }, { "epoch": 0.2734584450402145, "grad_norm": 20.088457107543945, "learning_rate": 8.352169156702769e-06, "loss": 0.5947, "step": 5406 }, { "epoch": 0.27350902928827964, "grad_norm": 7.6478400230407715, "learning_rate": 8.351573598372444e-06, "loss": 0.4186, "step": 5407 }, { "epoch": 0.2735596135363448, "grad_norm": 14.058696746826172, "learning_rate": 8.350977953680164e-06, "loss": 0.5956, "step": 5408 }, { "epoch": 0.2736101977844099, "grad_norm": 9.80154800415039, "learning_rate": 8.350382222641275e-06, "loss": 0.4762, "step": 5409 }, { "epoch": 0.27366078203247507, "grad_norm": 7.472720623016357, "learning_rate": 8.34978640527113e-06, "loss": 0.443, "step": 5410 }, { "epoch": 0.2737113662805402, "grad_norm": 9.926117897033691, "learning_rate": 8.34919050158508e-06, "loss": 0.4909, "step": 5411 }, { "epoch": 0.2737619505286054, "grad_norm": 14.776535034179688, "learning_rate": 8.348594511598481e-06, "loss": 0.5632, "step": 5412 }, { "epoch": 0.27381253477667056, "grad_norm": 13.422080993652344, "learning_rate": 8.34799843532669e-06, "loss": 0.4646, "step": 5413 }, { "epoch": 0.2738631190247357, "grad_norm": 9.66800308227539, "learning_rate": 8.347402272785065e-06, "loss": 0.5288, "step": 5414 }, { "epoch": 0.27391370327280085, "grad_norm": 6.896244049072266, "learning_rate": 8.34680602398897e-06, "loss": 0.4487, "step": 5415 }, { "epoch": 0.273964287520866, "grad_norm": 11.145813941955566, "learning_rate": 8.346209688953766e-06, "loss": 0.529, "step": 5416 }, { "epoch": 0.27401487176893113, "grad_norm": 6.438694953918457, "learning_rate": 8.345613267694821e-06, "loss": 0.4051, "step": 5417 }, { "epoch": 0.27406545601699633, "grad_norm": 13.474553108215332, "learning_rate": 8.345016760227502e-06, "loss": 0.5588, "step": 5418 }, { "epoch": 0.2741160402650615, "grad_norm": 5.829554557800293, "learning_rate": 8.344420166567183e-06, "loss": 0.4265, "step": 5419 }, { "epoch": 0.2741666245131266, "grad_norm": 5.949862957000732, "learning_rate": 8.343823486729234e-06, "loss": 0.4143, "step": 5420 }, { "epoch": 0.27421720876119177, "grad_norm": 7.01255989074707, "learning_rate": 8.34322672072903e-06, "loss": 0.4867, "step": 5421 }, { "epoch": 0.2742677930092569, "grad_norm": 7.207683086395264, "learning_rate": 8.342629868581948e-06, "loss": 0.4595, "step": 5422 }, { "epoch": 0.27431837725732205, "grad_norm": 4.560984134674072, "learning_rate": 8.342032930303369e-06, "loss": 0.3898, "step": 5423 }, { "epoch": 0.2743689615053872, "grad_norm": 11.35844898223877, "learning_rate": 8.341435905908673e-06, "loss": 0.5349, "step": 5424 }, { "epoch": 0.2744195457534524, "grad_norm": 5.743046283721924, "learning_rate": 8.340838795413244e-06, "loss": 0.4689, "step": 5425 }, { "epoch": 0.27447013000151754, "grad_norm": 9.903831481933594, "learning_rate": 8.34024159883247e-06, "loss": 0.5898, "step": 5426 }, { "epoch": 0.2745207142495827, "grad_norm": 6.757344722747803, "learning_rate": 8.339644316181739e-06, "loss": 0.4318, "step": 5427 }, { "epoch": 0.27457129849764783, "grad_norm": 5.970839500427246, "learning_rate": 8.339046947476439e-06, "loss": 0.4413, "step": 5428 }, { "epoch": 0.274621882745713, "grad_norm": 9.874963760375977, "learning_rate": 8.338449492731964e-06, "loss": 0.4174, "step": 5429 }, { "epoch": 0.2746724669937781, "grad_norm": 12.38375186920166, "learning_rate": 8.33785195196371e-06, "loss": 0.5583, "step": 5430 }, { "epoch": 0.27472305124184326, "grad_norm": 8.598273277282715, "learning_rate": 8.337254325187073e-06, "loss": 0.504, "step": 5431 }, { "epoch": 0.27477363548990846, "grad_norm": 7.159535884857178, "learning_rate": 8.336656612417454e-06, "loss": 0.5006, "step": 5432 }, { "epoch": 0.2748242197379736, "grad_norm": 5.0594563484191895, "learning_rate": 8.336058813670252e-06, "loss": 0.4474, "step": 5433 }, { "epoch": 0.27487480398603875, "grad_norm": 9.732131004333496, "learning_rate": 8.335460928960873e-06, "loss": 0.3583, "step": 5434 }, { "epoch": 0.2749253882341039, "grad_norm": 9.097726821899414, "learning_rate": 8.334862958304723e-06, "loss": 0.3785, "step": 5435 }, { "epoch": 0.27497597248216904, "grad_norm": 5.392683029174805, "learning_rate": 8.334264901717209e-06, "loss": 0.3934, "step": 5436 }, { "epoch": 0.2750265567302342, "grad_norm": 8.010957717895508, "learning_rate": 8.333666759213742e-06, "loss": 0.4922, "step": 5437 }, { "epoch": 0.2750771409782994, "grad_norm": 19.099693298339844, "learning_rate": 8.333068530809733e-06, "loss": 0.6544, "step": 5438 }, { "epoch": 0.27512772522636453, "grad_norm": 16.852434158325195, "learning_rate": 8.3324702165206e-06, "loss": 0.4911, "step": 5439 }, { "epoch": 0.27517830947442967, "grad_norm": 10.740509033203125, "learning_rate": 8.33187181636176e-06, "loss": 0.503, "step": 5440 }, { "epoch": 0.2752288937224948, "grad_norm": 10.33597469329834, "learning_rate": 8.33127333034863e-06, "loss": 0.4918, "step": 5441 }, { "epoch": 0.27527947797055996, "grad_norm": 11.751834869384766, "learning_rate": 8.330674758496632e-06, "loss": 0.511, "step": 5442 }, { "epoch": 0.2753300622186251, "grad_norm": 6.728593349456787, "learning_rate": 8.330076100821191e-06, "loss": 0.4364, "step": 5443 }, { "epoch": 0.27538064646669025, "grad_norm": 8.811002731323242, "learning_rate": 8.329477357337734e-06, "loss": 0.4532, "step": 5444 }, { "epoch": 0.27543123071475545, "grad_norm": 7.178900718688965, "learning_rate": 8.328878528061685e-06, "loss": 0.4182, "step": 5445 }, { "epoch": 0.2754818149628206, "grad_norm": 6.248692989349365, "learning_rate": 8.328279613008478e-06, "loss": 0.3706, "step": 5446 }, { "epoch": 0.27553239921088574, "grad_norm": 7.7663350105285645, "learning_rate": 8.327680612193546e-06, "loss": 0.4115, "step": 5447 }, { "epoch": 0.2755829834589509, "grad_norm": 7.441996097564697, "learning_rate": 8.32708152563232e-06, "loss": 0.4648, "step": 5448 }, { "epoch": 0.275633567707016, "grad_norm": 6.591432571411133, "learning_rate": 8.32648235334024e-06, "loss": 0.3678, "step": 5449 }, { "epoch": 0.27568415195508117, "grad_norm": 15.743865013122559, "learning_rate": 8.325883095332745e-06, "loss": 0.5271, "step": 5450 }, { "epoch": 0.2757347362031463, "grad_norm": 13.05367660522461, "learning_rate": 8.325283751625275e-06, "loss": 0.5759, "step": 5451 }, { "epoch": 0.2757853204512115, "grad_norm": 10.122241973876953, "learning_rate": 8.324684322233276e-06, "loss": 0.5251, "step": 5452 }, { "epoch": 0.27583590469927666, "grad_norm": 6.700644493103027, "learning_rate": 8.324084807172192e-06, "loss": 0.3986, "step": 5453 }, { "epoch": 0.2758864889473418, "grad_norm": 6.455223083496094, "learning_rate": 8.32348520645747e-06, "loss": 0.4634, "step": 5454 }, { "epoch": 0.27593707319540695, "grad_norm": 12.266178131103516, "learning_rate": 8.322885520104564e-06, "loss": 0.4821, "step": 5455 }, { "epoch": 0.2759876574434721, "grad_norm": 11.118383407592773, "learning_rate": 8.322285748128922e-06, "loss": 0.598, "step": 5456 }, { "epoch": 0.27603824169153723, "grad_norm": 6.590839862823486, "learning_rate": 8.321685890546003e-06, "loss": 0.4952, "step": 5457 }, { "epoch": 0.27608882593960243, "grad_norm": 9.622041702270508, "learning_rate": 8.321085947371262e-06, "loss": 0.5423, "step": 5458 }, { "epoch": 0.2761394101876676, "grad_norm": 7.015544891357422, "learning_rate": 8.320485918620155e-06, "loss": 0.4137, "step": 5459 }, { "epoch": 0.2761899944357327, "grad_norm": 7.92238712310791, "learning_rate": 8.31988580430815e-06, "loss": 0.4297, "step": 5460 }, { "epoch": 0.27624057868379787, "grad_norm": 5.859302043914795, "learning_rate": 8.319285604450705e-06, "loss": 0.4316, "step": 5461 }, { "epoch": 0.276291162931863, "grad_norm": 7.538707733154297, "learning_rate": 8.318685319063287e-06, "loss": 0.4705, "step": 5462 }, { "epoch": 0.27634174717992815, "grad_norm": 12.111161231994629, "learning_rate": 8.318084948161362e-06, "loss": 0.5031, "step": 5463 }, { "epoch": 0.2763923314279933, "grad_norm": 7.76350212097168, "learning_rate": 8.317484491760406e-06, "loss": 0.5192, "step": 5464 }, { "epoch": 0.2764429156760585, "grad_norm": 11.51979923248291, "learning_rate": 8.316883949875886e-06, "loss": 0.4593, "step": 5465 }, { "epoch": 0.27649349992412364, "grad_norm": 10.316705703735352, "learning_rate": 8.316283322523279e-06, "loss": 0.3709, "step": 5466 }, { "epoch": 0.2765440841721888, "grad_norm": 7.557793140411377, "learning_rate": 8.315682609718061e-06, "loss": 0.4878, "step": 5467 }, { "epoch": 0.27659466842025393, "grad_norm": 5.391087055206299, "learning_rate": 8.315081811475711e-06, "loss": 0.3712, "step": 5468 }, { "epoch": 0.2766452526683191, "grad_norm": 6.44757604598999, "learning_rate": 8.314480927811708e-06, "loss": 0.4385, "step": 5469 }, { "epoch": 0.2766958369163842, "grad_norm": 7.05917501449585, "learning_rate": 8.31387995874154e-06, "loss": 0.4214, "step": 5470 }, { "epoch": 0.27674642116444936, "grad_norm": 7.577714443206787, "learning_rate": 8.313278904280687e-06, "loss": 0.4531, "step": 5471 }, { "epoch": 0.27679700541251456, "grad_norm": 6.866298675537109, "learning_rate": 8.312677764444643e-06, "loss": 0.4417, "step": 5472 }, { "epoch": 0.2768475896605797, "grad_norm": 8.528971672058105, "learning_rate": 8.312076539248891e-06, "loss": 0.4547, "step": 5473 }, { "epoch": 0.27689817390864485, "grad_norm": 6.935538291931152, "learning_rate": 8.31147522870893e-06, "loss": 0.3732, "step": 5474 }, { "epoch": 0.27694875815671, "grad_norm": 11.73112678527832, "learning_rate": 8.310873832840248e-06, "loss": 0.5736, "step": 5475 }, { "epoch": 0.27699934240477514, "grad_norm": 6.706085205078125, "learning_rate": 8.310272351658345e-06, "loss": 0.3205, "step": 5476 }, { "epoch": 0.2770499266528403, "grad_norm": 14.025750160217285, "learning_rate": 8.309670785178721e-06, "loss": 0.5146, "step": 5477 }, { "epoch": 0.2771005109009055, "grad_norm": 8.524494171142578, "learning_rate": 8.309069133416873e-06, "loss": 0.412, "step": 5478 }, { "epoch": 0.2771510951489706, "grad_norm": 12.509429931640625, "learning_rate": 8.308467396388307e-06, "loss": 0.5159, "step": 5479 }, { "epoch": 0.27720167939703577, "grad_norm": 13.232701301574707, "learning_rate": 8.307865574108526e-06, "loss": 0.4544, "step": 5480 }, { "epoch": 0.2772522636451009, "grad_norm": 6.968139171600342, "learning_rate": 8.30726366659304e-06, "loss": 0.5168, "step": 5481 }, { "epoch": 0.27730284789316606, "grad_norm": 6.457491397857666, "learning_rate": 8.306661673857359e-06, "loss": 0.4275, "step": 5482 }, { "epoch": 0.2773534321412312, "grad_norm": 6.966491222381592, "learning_rate": 8.306059595916992e-06, "loss": 0.3118, "step": 5483 }, { "epoch": 0.27740401638929635, "grad_norm": 12.217826843261719, "learning_rate": 8.305457432787455e-06, "loss": 0.522, "step": 5484 }, { "epoch": 0.27745460063736155, "grad_norm": 10.212067604064941, "learning_rate": 8.304855184484265e-06, "loss": 0.5304, "step": 5485 }, { "epoch": 0.2775051848854267, "grad_norm": 6.726371765136719, "learning_rate": 8.304252851022938e-06, "loss": 0.3981, "step": 5486 }, { "epoch": 0.27755576913349184, "grad_norm": 7.51662015914917, "learning_rate": 8.303650432418996e-06, "loss": 0.4597, "step": 5487 }, { "epoch": 0.277606353381557, "grad_norm": 10.09660530090332, "learning_rate": 8.303047928687964e-06, "loss": 0.4027, "step": 5488 }, { "epoch": 0.2776569376296221, "grad_norm": 6.082779884338379, "learning_rate": 8.302445339845363e-06, "loss": 0.3877, "step": 5489 }, { "epoch": 0.27770752187768727, "grad_norm": 5.995169639587402, "learning_rate": 8.301842665906722e-06, "loss": 0.3953, "step": 5490 }, { "epoch": 0.2777581061257524, "grad_norm": 8.816116333007812, "learning_rate": 8.301239906887573e-06, "loss": 0.449, "step": 5491 }, { "epoch": 0.2778086903738176, "grad_norm": 9.099668502807617, "learning_rate": 8.300637062803445e-06, "loss": 0.503, "step": 5492 }, { "epoch": 0.27785927462188276, "grad_norm": 8.557486534118652, "learning_rate": 8.300034133669871e-06, "loss": 0.4683, "step": 5493 }, { "epoch": 0.2779098588699479, "grad_norm": 7.577928066253662, "learning_rate": 8.299431119502388e-06, "loss": 0.4901, "step": 5494 }, { "epoch": 0.27796044311801305, "grad_norm": 5.055172443389893, "learning_rate": 8.298828020316537e-06, "loss": 0.4102, "step": 5495 }, { "epoch": 0.2780110273660782, "grad_norm": 12.366851806640625, "learning_rate": 8.298224836127852e-06, "loss": 0.5615, "step": 5496 }, { "epoch": 0.27806161161414333, "grad_norm": 7.188662052154541, "learning_rate": 8.297621566951883e-06, "loss": 0.4596, "step": 5497 }, { "epoch": 0.27811219586220853, "grad_norm": 16.599885940551758, "learning_rate": 8.297018212804171e-06, "loss": 0.6395, "step": 5498 }, { "epoch": 0.2781627801102737, "grad_norm": 8.397521018981934, "learning_rate": 8.29641477370026e-06, "loss": 0.3856, "step": 5499 }, { "epoch": 0.2782133643583388, "grad_norm": 11.641761779785156, "learning_rate": 8.295811249655707e-06, "loss": 0.6615, "step": 5500 }, { "epoch": 0.27826394860640397, "grad_norm": 7.049093723297119, "learning_rate": 8.295207640686056e-06, "loss": 0.4876, "step": 5501 }, { "epoch": 0.2783145328544691, "grad_norm": 6.024382591247559, "learning_rate": 8.294603946806862e-06, "loss": 0.415, "step": 5502 }, { "epoch": 0.27836511710253425, "grad_norm": 11.512838363647461, "learning_rate": 8.294000168033687e-06, "loss": 0.5524, "step": 5503 }, { "epoch": 0.2784157013505994, "grad_norm": 6.397805690765381, "learning_rate": 8.293396304382079e-06, "loss": 0.5078, "step": 5504 }, { "epoch": 0.2784662855986646, "grad_norm": 7.114233493804932, "learning_rate": 8.292792355867606e-06, "loss": 0.4847, "step": 5505 }, { "epoch": 0.27851686984672974, "grad_norm": 12.877069473266602, "learning_rate": 8.292188322505827e-06, "loss": 0.5454, "step": 5506 }, { "epoch": 0.2785674540947949, "grad_norm": 5.22369909286499, "learning_rate": 8.291584204312306e-06, "loss": 0.3938, "step": 5507 }, { "epoch": 0.27861803834286003, "grad_norm": 6.7673821449279785, "learning_rate": 8.29098000130261e-06, "loss": 0.4323, "step": 5508 }, { "epoch": 0.2786686225909252, "grad_norm": 4.415411949157715, "learning_rate": 8.29037571349231e-06, "loss": 0.398, "step": 5509 }, { "epoch": 0.2787192068389903, "grad_norm": 8.770179748535156, "learning_rate": 8.289771340896975e-06, "loss": 0.3881, "step": 5510 }, { "epoch": 0.2787697910870555, "grad_norm": 10.446850776672363, "learning_rate": 8.289166883532179e-06, "loss": 0.5187, "step": 5511 }, { "epoch": 0.27882037533512066, "grad_norm": 6.616362571716309, "learning_rate": 8.288562341413495e-06, "loss": 0.3218, "step": 5512 }, { "epoch": 0.2788709595831858, "grad_norm": 5.698642730712891, "learning_rate": 8.287957714556505e-06, "loss": 0.3806, "step": 5513 }, { "epoch": 0.27892154383125095, "grad_norm": 7.981727600097656, "learning_rate": 8.287353002976785e-06, "loss": 0.5124, "step": 5514 }, { "epoch": 0.2789721280793161, "grad_norm": 7.249223709106445, "learning_rate": 8.286748206689918e-06, "loss": 0.4842, "step": 5515 }, { "epoch": 0.27902271232738124, "grad_norm": 13.805303573608398, "learning_rate": 8.286143325711491e-06, "loss": 0.4506, "step": 5516 }, { "epoch": 0.2790732965754464, "grad_norm": 7.705753326416016, "learning_rate": 8.285538360057087e-06, "loss": 0.3687, "step": 5517 }, { "epoch": 0.2791238808235116, "grad_norm": 9.118597030639648, "learning_rate": 8.284933309742293e-06, "loss": 0.4973, "step": 5518 }, { "epoch": 0.2791744650715767, "grad_norm": 17.66910743713379, "learning_rate": 8.284328174782705e-06, "loss": 0.4895, "step": 5519 }, { "epoch": 0.27922504931964187, "grad_norm": 6.494693279266357, "learning_rate": 8.283722955193912e-06, "loss": 0.4331, "step": 5520 }, { "epoch": 0.279275633567707, "grad_norm": 6.8934173583984375, "learning_rate": 8.283117650991509e-06, "loss": 0.3049, "step": 5521 }, { "epoch": 0.27932621781577216, "grad_norm": 3.805586338043213, "learning_rate": 8.282512262191095e-06, "loss": 0.3164, "step": 5522 }, { "epoch": 0.2793768020638373, "grad_norm": 6.988900661468506, "learning_rate": 8.281906788808268e-06, "loss": 0.4681, "step": 5523 }, { "epoch": 0.27942738631190245, "grad_norm": 10.024073600769043, "learning_rate": 8.28130123085863e-06, "loss": 0.437, "step": 5524 }, { "epoch": 0.27947797055996765, "grad_norm": 7.088413238525391, "learning_rate": 8.280695588357786e-06, "loss": 0.511, "step": 5525 }, { "epoch": 0.2795285548080328, "grad_norm": 17.064495086669922, "learning_rate": 8.280089861321342e-06, "loss": 0.6377, "step": 5526 }, { "epoch": 0.27957913905609794, "grad_norm": 8.398688316345215, "learning_rate": 8.2794840497649e-06, "loss": 0.2907, "step": 5527 }, { "epoch": 0.2796297233041631, "grad_norm": 11.75663948059082, "learning_rate": 8.27887815370408e-06, "loss": 0.4478, "step": 5528 }, { "epoch": 0.2796803075522282, "grad_norm": 6.727785587310791, "learning_rate": 8.278272173154488e-06, "loss": 0.3587, "step": 5529 }, { "epoch": 0.27973089180029337, "grad_norm": 5.416786193847656, "learning_rate": 8.277666108131742e-06, "loss": 0.3722, "step": 5530 }, { "epoch": 0.27978147604835857, "grad_norm": 6.876662254333496, "learning_rate": 8.277059958651456e-06, "loss": 0.3312, "step": 5531 }, { "epoch": 0.2798320602964237, "grad_norm": 13.548285484313965, "learning_rate": 8.27645372472925e-06, "loss": 0.4668, "step": 5532 }, { "epoch": 0.27988264454448886, "grad_norm": 6.842885494232178, "learning_rate": 8.275847406380746e-06, "loss": 0.3512, "step": 5533 }, { "epoch": 0.279933228792554, "grad_norm": 7.72330379486084, "learning_rate": 8.275241003621568e-06, "loss": 0.4776, "step": 5534 }, { "epoch": 0.27998381304061914, "grad_norm": 5.60444974899292, "learning_rate": 8.274634516467338e-06, "loss": 0.3949, "step": 5535 }, { "epoch": 0.2800343972886843, "grad_norm": 11.648076057434082, "learning_rate": 8.274027944933687e-06, "loss": 0.4361, "step": 5536 }, { "epoch": 0.28008498153674943, "grad_norm": 9.836464881896973, "learning_rate": 8.273421289036246e-06, "loss": 0.4769, "step": 5537 }, { "epoch": 0.28013556578481463, "grad_norm": 5.173454761505127, "learning_rate": 8.272814548790642e-06, "loss": 0.4201, "step": 5538 }, { "epoch": 0.2801861500328798, "grad_norm": 7.320765972137451, "learning_rate": 8.272207724212514e-06, "loss": 0.4831, "step": 5539 }, { "epoch": 0.2802367342809449, "grad_norm": 9.659708023071289, "learning_rate": 8.271600815317495e-06, "loss": 0.545, "step": 5540 }, { "epoch": 0.28028731852901007, "grad_norm": 13.696146011352539, "learning_rate": 8.270993822121226e-06, "loss": 0.5306, "step": 5541 }, { "epoch": 0.2803379027770752, "grad_norm": 7.582204341888428, "learning_rate": 8.270386744639347e-06, "loss": 0.4497, "step": 5542 }, { "epoch": 0.28038848702514035, "grad_norm": 14.028082847595215, "learning_rate": 8.269779582887502e-06, "loss": 0.6243, "step": 5543 }, { "epoch": 0.2804390712732055, "grad_norm": 6.979476451873779, "learning_rate": 8.269172336881334e-06, "loss": 0.4772, "step": 5544 }, { "epoch": 0.2804896555212707, "grad_norm": 8.765317916870117, "learning_rate": 8.268565006636491e-06, "loss": 0.4192, "step": 5545 }, { "epoch": 0.28054023976933584, "grad_norm": 8.69060230255127, "learning_rate": 8.267957592168623e-06, "loss": 0.2968, "step": 5546 }, { "epoch": 0.280590824017401, "grad_norm": 7.8962507247924805, "learning_rate": 8.26735009349338e-06, "loss": 0.4123, "step": 5547 }, { "epoch": 0.28064140826546613, "grad_norm": 9.517377853393555, "learning_rate": 8.26674251062642e-06, "loss": 0.4797, "step": 5548 }, { "epoch": 0.2806919925135313, "grad_norm": 9.22282886505127, "learning_rate": 8.266134843583394e-06, "loss": 0.3469, "step": 5549 }, { "epoch": 0.2807425767615964, "grad_norm": 8.15369701385498, "learning_rate": 8.265527092379962e-06, "loss": 0.3915, "step": 5550 }, { "epoch": 0.2807931610096616, "grad_norm": 7.538900852203369, "learning_rate": 8.264919257031785e-06, "loss": 0.4483, "step": 5551 }, { "epoch": 0.28084374525772676, "grad_norm": 10.922198295593262, "learning_rate": 8.264311337554525e-06, "loss": 0.4874, "step": 5552 }, { "epoch": 0.2808943295057919, "grad_norm": 5.287049293518066, "learning_rate": 8.263703333963846e-06, "loss": 0.3857, "step": 5553 }, { "epoch": 0.28094491375385705, "grad_norm": 12.479533195495605, "learning_rate": 8.263095246275416e-06, "loss": 0.5257, "step": 5554 }, { "epoch": 0.2809954980019222, "grad_norm": 11.0941162109375, "learning_rate": 8.262487074504902e-06, "loss": 0.4303, "step": 5555 }, { "epoch": 0.28104608224998734, "grad_norm": 5.943458557128906, "learning_rate": 8.261878818667977e-06, "loss": 0.3982, "step": 5556 }, { "epoch": 0.2810966664980525, "grad_norm": 9.669759750366211, "learning_rate": 8.261270478780313e-06, "loss": 0.3944, "step": 5557 }, { "epoch": 0.2811472507461177, "grad_norm": 8.070737838745117, "learning_rate": 8.260662054857587e-06, "loss": 0.4741, "step": 5558 }, { "epoch": 0.2811978349941828, "grad_norm": 9.185235977172852, "learning_rate": 8.260053546915475e-06, "loss": 0.4502, "step": 5559 }, { "epoch": 0.28124841924224797, "grad_norm": 17.86792755126953, "learning_rate": 8.259444954969656e-06, "loss": 0.5037, "step": 5560 }, { "epoch": 0.2812990034903131, "grad_norm": 6.469104290008545, "learning_rate": 8.258836279035816e-06, "loss": 0.4104, "step": 5561 }, { "epoch": 0.28134958773837826, "grad_norm": 5.8064985275268555, "learning_rate": 8.258227519129636e-06, "loss": 0.4477, "step": 5562 }, { "epoch": 0.2814001719864434, "grad_norm": 11.482146263122559, "learning_rate": 8.257618675266801e-06, "loss": 0.5411, "step": 5563 }, { "epoch": 0.28145075623450855, "grad_norm": 8.147767066955566, "learning_rate": 8.257009747463003e-06, "loss": 0.4267, "step": 5564 }, { "epoch": 0.28150134048257375, "grad_norm": 10.595839500427246, "learning_rate": 8.256400735733929e-06, "loss": 0.4222, "step": 5565 }, { "epoch": 0.2815519247306389, "grad_norm": 9.797677040100098, "learning_rate": 8.255791640095275e-06, "loss": 0.5268, "step": 5566 }, { "epoch": 0.28160250897870404, "grad_norm": 10.059908866882324, "learning_rate": 8.255182460562734e-06, "loss": 0.5026, "step": 5567 }, { "epoch": 0.2816530932267692, "grad_norm": 7.955201148986816, "learning_rate": 8.254573197152004e-06, "loss": 0.3581, "step": 5568 }, { "epoch": 0.2817036774748343, "grad_norm": 7.833637237548828, "learning_rate": 8.253963849878785e-06, "loss": 0.3966, "step": 5569 }, { "epoch": 0.28175426172289947, "grad_norm": 11.629232406616211, "learning_rate": 8.253354418758775e-06, "loss": 0.4712, "step": 5570 }, { "epoch": 0.28180484597096467, "grad_norm": 6.352349758148193, "learning_rate": 8.25274490380768e-06, "loss": 0.4526, "step": 5571 }, { "epoch": 0.2818554302190298, "grad_norm": 5.647150993347168, "learning_rate": 8.252135305041207e-06, "loss": 0.423, "step": 5572 }, { "epoch": 0.28190601446709496, "grad_norm": 9.494549751281738, "learning_rate": 8.25152562247506e-06, "loss": 0.4516, "step": 5573 }, { "epoch": 0.2819565987151601, "grad_norm": 12.835171699523926, "learning_rate": 8.250915856124955e-06, "loss": 0.4535, "step": 5574 }, { "epoch": 0.28200718296322524, "grad_norm": 7.56337308883667, "learning_rate": 8.250306006006596e-06, "loss": 0.5667, "step": 5575 }, { "epoch": 0.2820577672112904, "grad_norm": 7.218019485473633, "learning_rate": 8.249696072135705e-06, "loss": 0.466, "step": 5576 }, { "epoch": 0.28210835145935553, "grad_norm": 6.379152297973633, "learning_rate": 8.249086054527995e-06, "loss": 0.456, "step": 5577 }, { "epoch": 0.28215893570742073, "grad_norm": 7.709872722625732, "learning_rate": 8.248475953199187e-06, "loss": 0.3834, "step": 5578 }, { "epoch": 0.2822095199554859, "grad_norm": 10.840264320373535, "learning_rate": 8.247865768164998e-06, "loss": 0.4697, "step": 5579 }, { "epoch": 0.282260104203551, "grad_norm": 7.646017551422119, "learning_rate": 8.247255499441153e-06, "loss": 0.4182, "step": 5580 }, { "epoch": 0.28231068845161617, "grad_norm": 8.763309478759766, "learning_rate": 8.246645147043378e-06, "loss": 0.477, "step": 5581 }, { "epoch": 0.2823612726996813, "grad_norm": 11.509051322937012, "learning_rate": 8.246034710987398e-06, "loss": 0.5624, "step": 5582 }, { "epoch": 0.28241185694774645, "grad_norm": 5.593618392944336, "learning_rate": 8.245424191288947e-06, "loss": 0.4031, "step": 5583 }, { "epoch": 0.2824624411958116, "grad_norm": 16.39560317993164, "learning_rate": 8.24481358796375e-06, "loss": 0.6097, "step": 5584 }, { "epoch": 0.2825130254438768, "grad_norm": 7.960967540740967, "learning_rate": 8.244202901027547e-06, "loss": 0.4537, "step": 5585 }, { "epoch": 0.28256360969194194, "grad_norm": 7.449760913848877, "learning_rate": 8.24359213049607e-06, "loss": 0.4963, "step": 5586 }, { "epoch": 0.2826141939400071, "grad_norm": 9.250697135925293, "learning_rate": 8.242981276385059e-06, "loss": 0.3927, "step": 5587 }, { "epoch": 0.28266477818807223, "grad_norm": 7.528141021728516, "learning_rate": 8.242370338710253e-06, "loss": 0.4998, "step": 5588 }, { "epoch": 0.2827153624361374, "grad_norm": 6.014871120452881, "learning_rate": 8.241759317487396e-06, "loss": 0.4561, "step": 5589 }, { "epoch": 0.2827659466842025, "grad_norm": 12.260668754577637, "learning_rate": 8.241148212732231e-06, "loss": 0.5437, "step": 5590 }, { "epoch": 0.2828165309322677, "grad_norm": 8.001230239868164, "learning_rate": 8.240537024460504e-06, "loss": 0.4731, "step": 5591 }, { "epoch": 0.28286711518033286, "grad_norm": 6.485415935516357, "learning_rate": 8.239925752687967e-06, "loss": 0.4731, "step": 5592 }, { "epoch": 0.282917699428398, "grad_norm": 6.266460418701172, "learning_rate": 8.239314397430369e-06, "loss": 0.5144, "step": 5593 }, { "epoch": 0.28296828367646315, "grad_norm": 7.065286159515381, "learning_rate": 8.238702958703462e-06, "loss": 0.5312, "step": 5594 }, { "epoch": 0.2830188679245283, "grad_norm": 6.9417619705200195, "learning_rate": 8.238091436523002e-06, "loss": 0.4564, "step": 5595 }, { "epoch": 0.28306945217259344, "grad_norm": 16.23239517211914, "learning_rate": 8.237479830904748e-06, "loss": 0.5054, "step": 5596 }, { "epoch": 0.2831200364206586, "grad_norm": 7.145061016082764, "learning_rate": 8.236868141864457e-06, "loss": 0.4682, "step": 5597 }, { "epoch": 0.2831706206687238, "grad_norm": 7.629339218139648, "learning_rate": 8.236256369417896e-06, "loss": 0.4893, "step": 5598 }, { "epoch": 0.2832212049167889, "grad_norm": 5.03338098526001, "learning_rate": 8.235644513580823e-06, "loss": 0.3215, "step": 5599 }, { "epoch": 0.28327178916485407, "grad_norm": 8.508790969848633, "learning_rate": 8.235032574369007e-06, "loss": 0.4789, "step": 5600 }, { "epoch": 0.2833223734129192, "grad_norm": 5.727334499359131, "learning_rate": 8.234420551798214e-06, "loss": 0.3686, "step": 5601 }, { "epoch": 0.28337295766098436, "grad_norm": 6.510148525238037, "learning_rate": 8.233808445884218e-06, "loss": 0.4199, "step": 5602 }, { "epoch": 0.2834235419090495, "grad_norm": 10.106377601623535, "learning_rate": 8.233196256642786e-06, "loss": 0.5299, "step": 5603 }, { "epoch": 0.28347412615711465, "grad_norm": 27.303028106689453, "learning_rate": 8.232583984089698e-06, "loss": 0.473, "step": 5604 }, { "epoch": 0.28352471040517985, "grad_norm": 4.873515605926514, "learning_rate": 8.23197162824073e-06, "loss": 0.4165, "step": 5605 }, { "epoch": 0.283575294653245, "grad_norm": 12.623373985290527, "learning_rate": 8.231359189111657e-06, "loss": 0.4582, "step": 5606 }, { "epoch": 0.28362587890131014, "grad_norm": 14.641053199768066, "learning_rate": 8.230746666718264e-06, "loss": 0.5793, "step": 5607 }, { "epoch": 0.2836764631493753, "grad_norm": 6.896992206573486, "learning_rate": 8.230134061076333e-06, "loss": 0.3414, "step": 5608 }, { "epoch": 0.2837270473974404, "grad_norm": 6.4786176681518555, "learning_rate": 8.22952137220165e-06, "loss": 0.4966, "step": 5609 }, { "epoch": 0.28377763164550557, "grad_norm": 9.737853050231934, "learning_rate": 8.228908600109999e-06, "loss": 0.4912, "step": 5610 }, { "epoch": 0.28382821589357077, "grad_norm": 7.876208782196045, "learning_rate": 8.228295744817175e-06, "loss": 0.5493, "step": 5611 }, { "epoch": 0.2838788001416359, "grad_norm": 7.060464859008789, "learning_rate": 8.227682806338969e-06, "loss": 0.3549, "step": 5612 }, { "epoch": 0.28392938438970106, "grad_norm": 11.89411449432373, "learning_rate": 8.22706978469117e-06, "loss": 0.4842, "step": 5613 }, { "epoch": 0.2839799686377662, "grad_norm": 6.522781848907471, "learning_rate": 8.226456679889579e-06, "loss": 0.4548, "step": 5614 }, { "epoch": 0.28403055288583134, "grad_norm": 8.090069770812988, "learning_rate": 8.225843491949991e-06, "loss": 0.3682, "step": 5615 }, { "epoch": 0.2840811371338965, "grad_norm": 7.408231258392334, "learning_rate": 8.225230220888208e-06, "loss": 0.3806, "step": 5616 }, { "epoch": 0.28413172138196163, "grad_norm": 7.980017185211182, "learning_rate": 8.224616866720034e-06, "loss": 0.4657, "step": 5617 }, { "epoch": 0.28418230563002683, "grad_norm": 9.244678497314453, "learning_rate": 8.224003429461272e-06, "loss": 0.4376, "step": 5618 }, { "epoch": 0.284232889878092, "grad_norm": 7.1717071533203125, "learning_rate": 8.223389909127728e-06, "loss": 0.3562, "step": 5619 }, { "epoch": 0.2842834741261571, "grad_norm": 6.7245001792907715, "learning_rate": 8.222776305735212e-06, "loss": 0.4369, "step": 5620 }, { "epoch": 0.28433405837422226, "grad_norm": 10.63569164276123, "learning_rate": 8.222162619299534e-06, "loss": 0.3962, "step": 5621 }, { "epoch": 0.2843846426222874, "grad_norm": 7.833859920501709, "learning_rate": 8.22154884983651e-06, "loss": 0.3794, "step": 5622 }, { "epoch": 0.28443522687035255, "grad_norm": 8.771931648254395, "learning_rate": 8.220934997361952e-06, "loss": 0.5038, "step": 5623 }, { "epoch": 0.2844858111184177, "grad_norm": 7.026968479156494, "learning_rate": 8.220321061891678e-06, "loss": 0.5088, "step": 5624 }, { "epoch": 0.2845363953664829, "grad_norm": 13.830645561218262, "learning_rate": 8.219707043441512e-06, "loss": 0.4474, "step": 5625 }, { "epoch": 0.28458697961454804, "grad_norm": 10.331404685974121, "learning_rate": 8.21909294202727e-06, "loss": 0.5268, "step": 5626 }, { "epoch": 0.2846375638626132, "grad_norm": 9.755051612854004, "learning_rate": 8.218478757664779e-06, "loss": 0.5809, "step": 5627 }, { "epoch": 0.28468814811067833, "grad_norm": 7.212647438049316, "learning_rate": 8.217864490369863e-06, "loss": 0.4265, "step": 5628 }, { "epoch": 0.2847387323587435, "grad_norm": 10.734853744506836, "learning_rate": 8.217250140158352e-06, "loss": 0.4629, "step": 5629 }, { "epoch": 0.2847893166068086, "grad_norm": 6.848241806030273, "learning_rate": 8.216635707046076e-06, "loss": 0.3416, "step": 5630 }, { "epoch": 0.2848399008548738, "grad_norm": 7.354380130767822, "learning_rate": 8.216021191048869e-06, "loss": 0.4102, "step": 5631 }, { "epoch": 0.28489048510293896, "grad_norm": 12.988338470458984, "learning_rate": 8.215406592182562e-06, "loss": 0.4724, "step": 5632 }, { "epoch": 0.2849410693510041, "grad_norm": 12.8967866897583, "learning_rate": 8.214791910462996e-06, "loss": 0.7787, "step": 5633 }, { "epoch": 0.28499165359906925, "grad_norm": 10.72923755645752, "learning_rate": 8.214177145906005e-06, "loss": 0.4373, "step": 5634 }, { "epoch": 0.2850422378471344, "grad_norm": 6.041459083557129, "learning_rate": 8.213562298527433e-06, "loss": 0.4387, "step": 5635 }, { "epoch": 0.28509282209519954, "grad_norm": 6.467741966247559, "learning_rate": 8.212947368343123e-06, "loss": 0.4635, "step": 5636 }, { "epoch": 0.2851434063432647, "grad_norm": 5.349323272705078, "learning_rate": 8.212332355368919e-06, "loss": 0.3973, "step": 5637 }, { "epoch": 0.2851939905913299, "grad_norm": 5.178757667541504, "learning_rate": 8.21171725962067e-06, "loss": 0.3984, "step": 5638 }, { "epoch": 0.285244574839395, "grad_norm": 9.823919296264648, "learning_rate": 8.211102081114225e-06, "loss": 0.4214, "step": 5639 }, { "epoch": 0.28529515908746017, "grad_norm": 9.316353797912598, "learning_rate": 8.210486819865435e-06, "loss": 0.3994, "step": 5640 }, { "epoch": 0.2853457433355253, "grad_norm": 15.216643333435059, "learning_rate": 8.209871475890154e-06, "loss": 0.5267, "step": 5641 }, { "epoch": 0.28539632758359046, "grad_norm": 5.239596843719482, "learning_rate": 8.20925604920424e-06, "loss": 0.3168, "step": 5642 }, { "epoch": 0.2854469118316556, "grad_norm": 8.251516342163086, "learning_rate": 8.208640539823548e-06, "loss": 0.4237, "step": 5643 }, { "epoch": 0.28549749607972075, "grad_norm": 15.140859603881836, "learning_rate": 8.208024947763937e-06, "loss": 0.4577, "step": 5644 }, { "epoch": 0.28554808032778595, "grad_norm": 7.897657871246338, "learning_rate": 8.207409273041276e-06, "loss": 0.4715, "step": 5645 }, { "epoch": 0.2855986645758511, "grad_norm": 4.932610034942627, "learning_rate": 8.206793515671422e-06, "loss": 0.4833, "step": 5646 }, { "epoch": 0.28564924882391624, "grad_norm": 5.8245038986206055, "learning_rate": 8.206177675670247e-06, "loss": 0.4945, "step": 5647 }, { "epoch": 0.2856998330719814, "grad_norm": 18.836214065551758, "learning_rate": 8.205561753053617e-06, "loss": 0.5908, "step": 5648 }, { "epoch": 0.2857504173200465, "grad_norm": 7.023951530456543, "learning_rate": 8.204945747837403e-06, "loss": 0.5156, "step": 5649 }, { "epoch": 0.28580100156811167, "grad_norm": 8.867093086242676, "learning_rate": 8.204329660037478e-06, "loss": 0.4143, "step": 5650 }, { "epoch": 0.28585158581617687, "grad_norm": 13.845561981201172, "learning_rate": 8.203713489669718e-06, "loss": 0.4607, "step": 5651 }, { "epoch": 0.285902170064242, "grad_norm": 9.867918014526367, "learning_rate": 8.20309723675e-06, "loss": 0.5175, "step": 5652 }, { "epoch": 0.28595275431230716, "grad_norm": 5.02970027923584, "learning_rate": 8.202480901294202e-06, "loss": 0.3189, "step": 5653 }, { "epoch": 0.2860033385603723, "grad_norm": 6.307526588439941, "learning_rate": 8.201864483318206e-06, "loss": 0.4168, "step": 5654 }, { "epoch": 0.28605392280843744, "grad_norm": 11.988810539245605, "learning_rate": 8.201247982837896e-06, "loss": 0.549, "step": 5655 }, { "epoch": 0.2861045070565026, "grad_norm": 8.694364547729492, "learning_rate": 8.200631399869159e-06, "loss": 0.3957, "step": 5656 }, { "epoch": 0.28615509130456773, "grad_norm": 6.478387832641602, "learning_rate": 8.200014734427882e-06, "loss": 0.4009, "step": 5657 }, { "epoch": 0.28620567555263293, "grad_norm": 6.889084339141846, "learning_rate": 8.199397986529953e-06, "loss": 0.4024, "step": 5658 }, { "epoch": 0.2862562598006981, "grad_norm": 6.622251510620117, "learning_rate": 8.198781156191266e-06, "loss": 0.3834, "step": 5659 }, { "epoch": 0.2863068440487632, "grad_norm": 5.968278884887695, "learning_rate": 8.198164243427717e-06, "loss": 0.4401, "step": 5660 }, { "epoch": 0.28635742829682836, "grad_norm": 11.83598804473877, "learning_rate": 8.197547248255198e-06, "loss": 0.491, "step": 5661 }, { "epoch": 0.2864080125448935, "grad_norm": 7.212235927581787, "learning_rate": 8.19693017068961e-06, "loss": 0.4919, "step": 5662 }, { "epoch": 0.28645859679295865, "grad_norm": 8.47834587097168, "learning_rate": 8.196313010746855e-06, "loss": 0.462, "step": 5663 }, { "epoch": 0.28650918104102385, "grad_norm": 4.602464199066162, "learning_rate": 8.195695768442833e-06, "loss": 0.3988, "step": 5664 }, { "epoch": 0.286559765289089, "grad_norm": 5.9287519454956055, "learning_rate": 8.19507844379345e-06, "loss": 0.3821, "step": 5665 }, { "epoch": 0.28661034953715414, "grad_norm": 9.84375286102295, "learning_rate": 8.19446103681461e-06, "loss": 0.2851, "step": 5666 }, { "epoch": 0.2866609337852193, "grad_norm": 9.075300216674805, "learning_rate": 8.19384354752223e-06, "loss": 0.3971, "step": 5667 }, { "epoch": 0.28671151803328443, "grad_norm": 11.070756912231445, "learning_rate": 8.193225975932214e-06, "loss": 0.4515, "step": 5668 }, { "epoch": 0.2867621022813496, "grad_norm": 8.761316299438477, "learning_rate": 8.192608322060479e-06, "loss": 0.4232, "step": 5669 }, { "epoch": 0.2868126865294147, "grad_norm": 5.206174373626709, "learning_rate": 8.191990585922937e-06, "loss": 0.3669, "step": 5670 }, { "epoch": 0.2868632707774799, "grad_norm": 11.424545288085938, "learning_rate": 8.191372767535508e-06, "loss": 0.505, "step": 5671 }, { "epoch": 0.28691385502554506, "grad_norm": 7.676465034484863, "learning_rate": 8.19075486691411e-06, "loss": 0.4401, "step": 5672 }, { "epoch": 0.2869644392736102, "grad_norm": 7.060115337371826, "learning_rate": 8.190136884074669e-06, "loss": 0.38, "step": 5673 }, { "epoch": 0.28701502352167535, "grad_norm": 11.667723655700684, "learning_rate": 8.189518819033102e-06, "loss": 0.4667, "step": 5674 }, { "epoch": 0.2870656077697405, "grad_norm": 10.41306209564209, "learning_rate": 8.188900671805343e-06, "loss": 0.5215, "step": 5675 }, { "epoch": 0.28711619201780564, "grad_norm": 7.290041923522949, "learning_rate": 8.188282442407314e-06, "loss": 0.387, "step": 5676 }, { "epoch": 0.2871667762658708, "grad_norm": 7.1229071617126465, "learning_rate": 8.187664130854948e-06, "loss": 0.4782, "step": 5677 }, { "epoch": 0.287217360513936, "grad_norm": 6.882676601409912, "learning_rate": 8.187045737164176e-06, "loss": 0.2852, "step": 5678 }, { "epoch": 0.2872679447620011, "grad_norm": 10.061992645263672, "learning_rate": 8.186427261350934e-06, "loss": 0.5564, "step": 5679 }, { "epoch": 0.28731852901006627, "grad_norm": 9.4679594039917, "learning_rate": 8.185808703431158e-06, "loss": 0.4789, "step": 5680 }, { "epoch": 0.2873691132581314, "grad_norm": 10.51447582244873, "learning_rate": 8.185190063420787e-06, "loss": 0.4155, "step": 5681 }, { "epoch": 0.28741969750619656, "grad_norm": 11.25426959991455, "learning_rate": 8.18457134133576e-06, "loss": 0.3511, "step": 5682 }, { "epoch": 0.2874702817542617, "grad_norm": 11.797004699707031, "learning_rate": 8.183952537192024e-06, "loss": 0.4319, "step": 5683 }, { "epoch": 0.2875208660023269, "grad_norm": 7.221688270568848, "learning_rate": 8.18333365100552e-06, "loss": 0.4242, "step": 5684 }, { "epoch": 0.28757145025039205, "grad_norm": 6.627527713775635, "learning_rate": 8.182714682792199e-06, "loss": 0.3532, "step": 5685 }, { "epoch": 0.2876220344984572, "grad_norm": 10.005538940429688, "learning_rate": 8.182095632568005e-06, "loss": 0.3925, "step": 5686 }, { "epoch": 0.28767261874652233, "grad_norm": 10.253917694091797, "learning_rate": 8.181476500348895e-06, "loss": 0.5655, "step": 5687 }, { "epoch": 0.2877232029945875, "grad_norm": 7.027501106262207, "learning_rate": 8.180857286150819e-06, "loss": 0.334, "step": 5688 }, { "epoch": 0.2877737872426526, "grad_norm": 24.089521408081055, "learning_rate": 8.180237989989733e-06, "loss": 0.4702, "step": 5689 }, { "epoch": 0.28782437149071777, "grad_norm": 5.384641647338867, "learning_rate": 8.179618611881596e-06, "loss": 0.3934, "step": 5690 }, { "epoch": 0.28787495573878297, "grad_norm": 9.214178085327148, "learning_rate": 8.17899915184237e-06, "loss": 0.427, "step": 5691 }, { "epoch": 0.2879255399868481, "grad_norm": 6.76386833190918, "learning_rate": 8.178379609888012e-06, "loss": 0.4419, "step": 5692 }, { "epoch": 0.28797612423491326, "grad_norm": 6.245019912719727, "learning_rate": 8.177759986034489e-06, "loss": 0.4217, "step": 5693 }, { "epoch": 0.2880267084829784, "grad_norm": 5.155355930328369, "learning_rate": 8.177140280297766e-06, "loss": 0.3244, "step": 5694 }, { "epoch": 0.28807729273104354, "grad_norm": 7.300425052642822, "learning_rate": 8.176520492693812e-06, "loss": 0.55, "step": 5695 }, { "epoch": 0.2881278769791087, "grad_norm": 19.70598793029785, "learning_rate": 8.1759006232386e-06, "loss": 0.6297, "step": 5696 }, { "epoch": 0.28817846122717383, "grad_norm": 13.197835922241211, "learning_rate": 8.175280671948098e-06, "loss": 0.5337, "step": 5697 }, { "epoch": 0.28822904547523903, "grad_norm": 4.68305778503418, "learning_rate": 8.174660638838282e-06, "loss": 0.4443, "step": 5698 }, { "epoch": 0.2882796297233042, "grad_norm": 8.611186027526855, "learning_rate": 8.17404052392513e-06, "loss": 0.4355, "step": 5699 }, { "epoch": 0.2883302139713693, "grad_norm": 8.734399795532227, "learning_rate": 8.17342032722462e-06, "loss": 0.4527, "step": 5700 }, { "epoch": 0.28838079821943446, "grad_norm": 7.432015419006348, "learning_rate": 8.172800048752734e-06, "loss": 0.4979, "step": 5701 }, { "epoch": 0.2884313824674996, "grad_norm": 7.295537948608398, "learning_rate": 8.172179688525455e-06, "loss": 0.3902, "step": 5702 }, { "epoch": 0.28848196671556475, "grad_norm": 6.690683841705322, "learning_rate": 8.171559246558766e-06, "loss": 0.4631, "step": 5703 }, { "epoch": 0.28853255096362995, "grad_norm": 5.363518714904785, "learning_rate": 8.170938722868656e-06, "loss": 0.4811, "step": 5704 }, { "epoch": 0.2885831352116951, "grad_norm": 7.0548996925354, "learning_rate": 8.170318117471116e-06, "loss": 0.5102, "step": 5705 }, { "epoch": 0.28863371945976024, "grad_norm": 10.064351081848145, "learning_rate": 8.169697430382132e-06, "loss": 0.5036, "step": 5706 }, { "epoch": 0.2886843037078254, "grad_norm": 7.80318021774292, "learning_rate": 8.169076661617705e-06, "loss": 0.404, "step": 5707 }, { "epoch": 0.28873488795589053, "grad_norm": 13.670491218566895, "learning_rate": 8.168455811193825e-06, "loss": 0.5574, "step": 5708 }, { "epoch": 0.2887854722039557, "grad_norm": 5.3551225662231445, "learning_rate": 8.167834879126493e-06, "loss": 0.5183, "step": 5709 }, { "epoch": 0.2888360564520208, "grad_norm": 7.609818935394287, "learning_rate": 8.167213865431707e-06, "loss": 0.46, "step": 5710 }, { "epoch": 0.288886640700086, "grad_norm": 5.444582939147949, "learning_rate": 8.16659277012547e-06, "loss": 0.4241, "step": 5711 }, { "epoch": 0.28893722494815116, "grad_norm": 26.745187759399414, "learning_rate": 8.165971593223785e-06, "loss": 0.5485, "step": 5712 }, { "epoch": 0.2889878091962163, "grad_norm": 8.400130271911621, "learning_rate": 8.165350334742662e-06, "loss": 0.3945, "step": 5713 }, { "epoch": 0.28903839344428145, "grad_norm": 7.641487121582031, "learning_rate": 8.164728994698103e-06, "loss": 0.4937, "step": 5714 }, { "epoch": 0.2890889776923466, "grad_norm": 5.57889461517334, "learning_rate": 8.164107573106125e-06, "loss": 0.4372, "step": 5715 }, { "epoch": 0.28913956194041174, "grad_norm": 10.172964096069336, "learning_rate": 8.163486069982735e-06, "loss": 0.5543, "step": 5716 }, { "epoch": 0.2891901461884769, "grad_norm": 6.443807125091553, "learning_rate": 8.16286448534395e-06, "loss": 0.3854, "step": 5717 }, { "epoch": 0.2892407304365421, "grad_norm": 5.3843207359313965, "learning_rate": 8.162242819205788e-06, "loss": 0.3714, "step": 5718 }, { "epoch": 0.2892913146846072, "grad_norm": 5.804282188415527, "learning_rate": 8.161621071584266e-06, "loss": 0.425, "step": 5719 }, { "epoch": 0.28934189893267237, "grad_norm": 9.962668418884277, "learning_rate": 8.160999242495407e-06, "loss": 0.505, "step": 5720 }, { "epoch": 0.2893924831807375, "grad_norm": 7.6046342849731445, "learning_rate": 8.160377331955232e-06, "loss": 0.4859, "step": 5721 }, { "epoch": 0.28944306742880266, "grad_norm": 9.697992324829102, "learning_rate": 8.159755339979766e-06, "loss": 0.3944, "step": 5722 }, { "epoch": 0.2894936516768678, "grad_norm": 6.625580310821533, "learning_rate": 8.159133266585038e-06, "loss": 0.4196, "step": 5723 }, { "epoch": 0.289544235924933, "grad_norm": 6.3979692459106445, "learning_rate": 8.158511111787075e-06, "loss": 0.3248, "step": 5724 }, { "epoch": 0.28959482017299815, "grad_norm": 12.379545211791992, "learning_rate": 8.157888875601909e-06, "loss": 0.3837, "step": 5725 }, { "epoch": 0.2896454044210633, "grad_norm": 10.026116371154785, "learning_rate": 8.157266558045576e-06, "loss": 0.4094, "step": 5726 }, { "epoch": 0.28969598866912843, "grad_norm": 7.578281879425049, "learning_rate": 8.156644159134109e-06, "loss": 0.3768, "step": 5727 }, { "epoch": 0.2897465729171936, "grad_norm": 7.513118267059326, "learning_rate": 8.156021678883548e-06, "loss": 0.3642, "step": 5728 }, { "epoch": 0.2897971571652587, "grad_norm": 22.542333602905273, "learning_rate": 8.15539911730993e-06, "loss": 0.6152, "step": 5729 }, { "epoch": 0.28984774141332387, "grad_norm": 8.182236671447754, "learning_rate": 8.154776474429298e-06, "loss": 0.3756, "step": 5730 }, { "epoch": 0.28989832566138907, "grad_norm": 18.91889190673828, "learning_rate": 8.154153750257697e-06, "loss": 0.7189, "step": 5731 }, { "epoch": 0.2899489099094542, "grad_norm": 9.792463302612305, "learning_rate": 8.153530944811173e-06, "loss": 0.457, "step": 5732 }, { "epoch": 0.28999949415751936, "grad_norm": 8.952385902404785, "learning_rate": 8.152908058105772e-06, "loss": 0.4781, "step": 5733 }, { "epoch": 0.2900500784055845, "grad_norm": 8.677910804748535, "learning_rate": 8.152285090157547e-06, "loss": 0.4786, "step": 5734 }, { "epoch": 0.29010066265364964, "grad_norm": 6.142906665802002, "learning_rate": 8.15166204098255e-06, "loss": 0.4326, "step": 5735 }, { "epoch": 0.2901512469017148, "grad_norm": 6.885105133056641, "learning_rate": 8.151038910596833e-06, "loss": 0.5017, "step": 5736 }, { "epoch": 0.29020183114977993, "grad_norm": 13.965339660644531, "learning_rate": 8.150415699016456e-06, "loss": 0.5983, "step": 5737 }, { "epoch": 0.29025241539784513, "grad_norm": 8.627046585083008, "learning_rate": 8.149792406257474e-06, "loss": 0.4855, "step": 5738 }, { "epoch": 0.2903029996459103, "grad_norm": 6.97520637512207, "learning_rate": 8.14916903233595e-06, "loss": 0.4589, "step": 5739 }, { "epoch": 0.2903535838939754, "grad_norm": 5.637609481811523, "learning_rate": 8.148545577267947e-06, "loss": 0.4421, "step": 5740 }, { "epoch": 0.29040416814204056, "grad_norm": 6.259542465209961, "learning_rate": 8.147922041069531e-06, "loss": 0.3615, "step": 5741 }, { "epoch": 0.2904547523901057, "grad_norm": 11.249430656433105, "learning_rate": 8.147298423756767e-06, "loss": 0.4291, "step": 5742 }, { "epoch": 0.29050533663817085, "grad_norm": 6.809417724609375, "learning_rate": 8.146674725345722e-06, "loss": 0.3692, "step": 5743 }, { "epoch": 0.29055592088623605, "grad_norm": 11.443683624267578, "learning_rate": 8.146050945852473e-06, "loss": 0.4204, "step": 5744 }, { "epoch": 0.2906065051343012, "grad_norm": 20.459951400756836, "learning_rate": 8.145427085293088e-06, "loss": 0.4983, "step": 5745 }, { "epoch": 0.29065708938236634, "grad_norm": 8.380370140075684, "learning_rate": 8.144803143683646e-06, "loss": 0.5506, "step": 5746 }, { "epoch": 0.2907076736304315, "grad_norm": 10.259491920471191, "learning_rate": 8.144179121040222e-06, "loss": 0.4876, "step": 5747 }, { "epoch": 0.29075825787849663, "grad_norm": 7.862092971801758, "learning_rate": 8.143555017378897e-06, "loss": 0.4044, "step": 5748 }, { "epoch": 0.2908088421265618, "grad_norm": 5.822942733764648, "learning_rate": 8.142930832715753e-06, "loss": 0.4158, "step": 5749 }, { "epoch": 0.2908594263746269, "grad_norm": 8.898761749267578, "learning_rate": 8.14230656706687e-06, "loss": 0.4351, "step": 5750 }, { "epoch": 0.2909100106226921, "grad_norm": 9.02773380279541, "learning_rate": 8.141682220448339e-06, "loss": 0.411, "step": 5751 }, { "epoch": 0.29096059487075726, "grad_norm": 9.480170249938965, "learning_rate": 8.141057792876244e-06, "loss": 0.4844, "step": 5752 }, { "epoch": 0.2910111791188224, "grad_norm": 10.306635856628418, "learning_rate": 8.140433284366678e-06, "loss": 0.4693, "step": 5753 }, { "epoch": 0.29106176336688755, "grad_norm": 10.51589298248291, "learning_rate": 8.139808694935732e-06, "loss": 0.5093, "step": 5754 }, { "epoch": 0.2911123476149527, "grad_norm": 7.680710315704346, "learning_rate": 8.139184024599498e-06, "loss": 0.4946, "step": 5755 }, { "epoch": 0.29116293186301784, "grad_norm": 6.887127876281738, "learning_rate": 8.138559273374076e-06, "loss": 0.434, "step": 5756 }, { "epoch": 0.291213516111083, "grad_norm": 8.902727127075195, "learning_rate": 8.137934441275561e-06, "loss": 0.4252, "step": 5757 }, { "epoch": 0.2912641003591482, "grad_norm": 5.183740615844727, "learning_rate": 8.137309528320055e-06, "loss": 0.4219, "step": 5758 }, { "epoch": 0.2913146846072133, "grad_norm": 7.68123722076416, "learning_rate": 8.13668453452366e-06, "loss": 0.3656, "step": 5759 }, { "epoch": 0.29136526885527847, "grad_norm": 10.551568031311035, "learning_rate": 8.136059459902483e-06, "loss": 0.4385, "step": 5760 }, { "epoch": 0.2914158531033436, "grad_norm": 5.089813709259033, "learning_rate": 8.135434304472627e-06, "loss": 0.4525, "step": 5761 }, { "epoch": 0.29146643735140876, "grad_norm": 8.223614692687988, "learning_rate": 8.134809068250202e-06, "loss": 0.4394, "step": 5762 }, { "epoch": 0.2915170215994739, "grad_norm": 11.425751686096191, "learning_rate": 8.134183751251321e-06, "loss": 0.6042, "step": 5763 }, { "epoch": 0.2915676058475391, "grad_norm": 7.579277038574219, "learning_rate": 8.133558353492092e-06, "loss": 0.4871, "step": 5764 }, { "epoch": 0.29161819009560425, "grad_norm": 5.127289772033691, "learning_rate": 8.132932874988635e-06, "loss": 0.4474, "step": 5765 }, { "epoch": 0.2916687743436694, "grad_norm": 7.895031452178955, "learning_rate": 8.132307315757065e-06, "loss": 0.4597, "step": 5766 }, { "epoch": 0.29171935859173453, "grad_norm": 5.7362518310546875, "learning_rate": 8.1316816758135e-06, "loss": 0.3908, "step": 5767 }, { "epoch": 0.2917699428397997, "grad_norm": 9.14797592163086, "learning_rate": 8.131055955174065e-06, "loss": 0.5058, "step": 5768 }, { "epoch": 0.2918205270878648, "grad_norm": 8.64758586883545, "learning_rate": 8.13043015385488e-06, "loss": 0.461, "step": 5769 }, { "epoch": 0.29187111133592997, "grad_norm": 10.75687313079834, "learning_rate": 8.129804271872069e-06, "loss": 0.4643, "step": 5770 }, { "epoch": 0.29192169558399517, "grad_norm": 5.658822536468506, "learning_rate": 8.129178309241763e-06, "loss": 0.4606, "step": 5771 }, { "epoch": 0.2919722798320603, "grad_norm": 11.360907554626465, "learning_rate": 8.128552265980091e-06, "loss": 0.4434, "step": 5772 }, { "epoch": 0.29202286408012545, "grad_norm": 6.972225189208984, "learning_rate": 8.127926142103184e-06, "loss": 0.5198, "step": 5773 }, { "epoch": 0.2920734483281906, "grad_norm": 4.936942100524902, "learning_rate": 8.127299937627173e-06, "loss": 0.4374, "step": 5774 }, { "epoch": 0.29212403257625574, "grad_norm": 7.8530120849609375, "learning_rate": 8.126673652568198e-06, "loss": 0.379, "step": 5775 }, { "epoch": 0.2921746168243209, "grad_norm": 12.232060432434082, "learning_rate": 8.126047286942396e-06, "loss": 0.4832, "step": 5776 }, { "epoch": 0.29222520107238603, "grad_norm": 6.231693744659424, "learning_rate": 8.125420840765905e-06, "loss": 0.5105, "step": 5777 }, { "epoch": 0.29227578532045123, "grad_norm": 8.830803871154785, "learning_rate": 8.124794314054867e-06, "loss": 0.4821, "step": 5778 }, { "epoch": 0.2923263695685164, "grad_norm": 8.245718002319336, "learning_rate": 8.12416770682543e-06, "loss": 0.3576, "step": 5779 }, { "epoch": 0.2923769538165815, "grad_norm": 6.276103496551514, "learning_rate": 8.123541019093735e-06, "loss": 0.3714, "step": 5780 }, { "epoch": 0.29242753806464666, "grad_norm": 9.23774528503418, "learning_rate": 8.122914250875934e-06, "loss": 0.5313, "step": 5781 }, { "epoch": 0.2924781223127118, "grad_norm": 13.392374992370605, "learning_rate": 8.122287402188175e-06, "loss": 0.5006, "step": 5782 }, { "epoch": 0.29252870656077695, "grad_norm": 8.486950874328613, "learning_rate": 8.121660473046611e-06, "loss": 0.4333, "step": 5783 }, { "epoch": 0.29257929080884215, "grad_norm": 9.675769805908203, "learning_rate": 8.121033463467397e-06, "loss": 0.4131, "step": 5784 }, { "epoch": 0.2926298750569073, "grad_norm": 9.972137451171875, "learning_rate": 8.12040637346669e-06, "loss": 0.4273, "step": 5785 }, { "epoch": 0.29268045930497244, "grad_norm": 7.8404951095581055, "learning_rate": 8.119779203060647e-06, "loss": 0.4367, "step": 5786 }, { "epoch": 0.2927310435530376, "grad_norm": 9.047821998596191, "learning_rate": 8.11915195226543e-06, "loss": 0.6013, "step": 5787 }, { "epoch": 0.29278162780110273, "grad_norm": 7.427097797393799, "learning_rate": 8.1185246210972e-06, "loss": 0.4489, "step": 5788 }, { "epoch": 0.2928322120491679, "grad_norm": 7.9675679206848145, "learning_rate": 8.117897209572123e-06, "loss": 0.5288, "step": 5789 }, { "epoch": 0.292882796297233, "grad_norm": 5.666334629058838, "learning_rate": 8.117269717706366e-06, "loss": 0.309, "step": 5790 }, { "epoch": 0.2929333805452982, "grad_norm": 5.014064788818359, "learning_rate": 8.116642145516097e-06, "loss": 0.427, "step": 5791 }, { "epoch": 0.29298396479336336, "grad_norm": 8.703062057495117, "learning_rate": 8.11601449301749e-06, "loss": 0.4908, "step": 5792 }, { "epoch": 0.2930345490414285, "grad_norm": 11.125931739807129, "learning_rate": 8.115386760226717e-06, "loss": 0.5381, "step": 5793 }, { "epoch": 0.29308513328949365, "grad_norm": 9.126684188842773, "learning_rate": 8.114758947159949e-06, "loss": 0.4227, "step": 5794 }, { "epoch": 0.2931357175375588, "grad_norm": 4.422773361206055, "learning_rate": 8.114131053833368e-06, "loss": 0.297, "step": 5795 }, { "epoch": 0.29318630178562394, "grad_norm": 9.244109153747559, "learning_rate": 8.11350308026315e-06, "loss": 0.3689, "step": 5796 }, { "epoch": 0.29323688603368914, "grad_norm": 6.840124607086182, "learning_rate": 8.11287502646548e-06, "loss": 0.4421, "step": 5797 }, { "epoch": 0.2932874702817543, "grad_norm": 9.363075256347656, "learning_rate": 8.112246892456538e-06, "loss": 0.4239, "step": 5798 }, { "epoch": 0.2933380545298194, "grad_norm": 5.34712028503418, "learning_rate": 8.111618678252513e-06, "loss": 0.3937, "step": 5799 }, { "epoch": 0.29338863877788457, "grad_norm": 8.483467102050781, "learning_rate": 8.110990383869587e-06, "loss": 0.4828, "step": 5800 }, { "epoch": 0.2934392230259497, "grad_norm": 11.707868576049805, "learning_rate": 8.110362009323956e-06, "loss": 0.4445, "step": 5801 }, { "epoch": 0.29348980727401486, "grad_norm": 6.523952007293701, "learning_rate": 8.109733554631809e-06, "loss": 0.3953, "step": 5802 }, { "epoch": 0.29354039152208, "grad_norm": 7.323569297790527, "learning_rate": 8.109105019809338e-06, "loss": 0.3774, "step": 5803 }, { "epoch": 0.2935909757701452, "grad_norm": 17.298866271972656, "learning_rate": 8.108476404872741e-06, "loss": 0.5559, "step": 5804 }, { "epoch": 0.29364156001821035, "grad_norm": 7.5074076652526855, "learning_rate": 8.107847709838215e-06, "loss": 0.4249, "step": 5805 }, { "epoch": 0.2936921442662755, "grad_norm": 8.115252494812012, "learning_rate": 8.10721893472196e-06, "loss": 0.3838, "step": 5806 }, { "epoch": 0.29374272851434063, "grad_norm": 13.691169738769531, "learning_rate": 8.106590079540177e-06, "loss": 0.434, "step": 5807 }, { "epoch": 0.2937933127624058, "grad_norm": 16.976558685302734, "learning_rate": 8.105961144309074e-06, "loss": 0.6816, "step": 5808 }, { "epoch": 0.2938438970104709, "grad_norm": 8.405879020690918, "learning_rate": 8.105332129044852e-06, "loss": 0.4424, "step": 5809 }, { "epoch": 0.29389448125853607, "grad_norm": 10.343954086303711, "learning_rate": 8.104703033763723e-06, "loss": 0.2746, "step": 5810 }, { "epoch": 0.29394506550660127, "grad_norm": 7.010892391204834, "learning_rate": 8.104073858481896e-06, "loss": 0.3292, "step": 5811 }, { "epoch": 0.2939956497546664, "grad_norm": 6.373586177825928, "learning_rate": 8.103444603215584e-06, "loss": 0.3635, "step": 5812 }, { "epoch": 0.29404623400273155, "grad_norm": 7.1454243659973145, "learning_rate": 8.102815267981e-06, "loss": 0.455, "step": 5813 }, { "epoch": 0.2940968182507967, "grad_norm": 14.484448432922363, "learning_rate": 8.10218585279436e-06, "loss": 0.5201, "step": 5814 }, { "epoch": 0.29414740249886184, "grad_norm": 18.18463134765625, "learning_rate": 8.101556357671885e-06, "loss": 0.5086, "step": 5815 }, { "epoch": 0.294197986746927, "grad_norm": 23.66639518737793, "learning_rate": 8.100926782629792e-06, "loss": 0.5258, "step": 5816 }, { "epoch": 0.2942485709949922, "grad_norm": 5.552120685577393, "learning_rate": 8.100297127684307e-06, "loss": 0.3972, "step": 5817 }, { "epoch": 0.29429915524305733, "grad_norm": 3.9338631629943848, "learning_rate": 8.099667392851656e-06, "loss": 0.3515, "step": 5818 }, { "epoch": 0.2943497394911225, "grad_norm": 5.497037887573242, "learning_rate": 8.09903757814806e-06, "loss": 0.3285, "step": 5819 }, { "epoch": 0.2944003237391876, "grad_norm": 8.664398193359375, "learning_rate": 8.098407683589754e-06, "loss": 0.3824, "step": 5820 }, { "epoch": 0.29445090798725276, "grad_norm": 9.394624710083008, "learning_rate": 8.097777709192963e-06, "loss": 0.5362, "step": 5821 }, { "epoch": 0.2945014922353179, "grad_norm": 10.250057220458984, "learning_rate": 8.097147654973923e-06, "loss": 0.6356, "step": 5822 }, { "epoch": 0.29455207648338305, "grad_norm": 7.607235908508301, "learning_rate": 8.09651752094887e-06, "loss": 0.4549, "step": 5823 }, { "epoch": 0.29460266073144825, "grad_norm": 5.704712390899658, "learning_rate": 8.095887307134038e-06, "loss": 0.3447, "step": 5824 }, { "epoch": 0.2946532449795134, "grad_norm": 6.362423419952393, "learning_rate": 8.09525701354567e-06, "loss": 0.4886, "step": 5825 }, { "epoch": 0.29470382922757854, "grad_norm": 7.0829877853393555, "learning_rate": 8.094626640200005e-06, "loss": 0.378, "step": 5826 }, { "epoch": 0.2947544134756437, "grad_norm": 9.189632415771484, "learning_rate": 8.093996187113284e-06, "loss": 0.4894, "step": 5827 }, { "epoch": 0.29480499772370883, "grad_norm": 10.927373886108398, "learning_rate": 8.093365654301758e-06, "loss": 0.4391, "step": 5828 }, { "epoch": 0.294855581971774, "grad_norm": 14.380606651306152, "learning_rate": 8.092735041781667e-06, "loss": 0.4783, "step": 5829 }, { "epoch": 0.2949061662198391, "grad_norm": 7.9618353843688965, "learning_rate": 8.092104349569267e-06, "loss": 0.4159, "step": 5830 }, { "epoch": 0.2949567504679043, "grad_norm": 6.941181659698486, "learning_rate": 8.091473577680807e-06, "loss": 0.4187, "step": 5831 }, { "epoch": 0.29500733471596946, "grad_norm": 6.219784736633301, "learning_rate": 8.090842726132537e-06, "loss": 0.4174, "step": 5832 }, { "epoch": 0.2950579189640346, "grad_norm": 11.135316848754883, "learning_rate": 8.090211794940718e-06, "loss": 0.6437, "step": 5833 }, { "epoch": 0.29510850321209975, "grad_norm": 7.14982271194458, "learning_rate": 8.089580784121604e-06, "loss": 0.5087, "step": 5834 }, { "epoch": 0.2951590874601649, "grad_norm": 7.667275428771973, "learning_rate": 8.088949693691456e-06, "loss": 0.3788, "step": 5835 }, { "epoch": 0.29520967170823004, "grad_norm": 12.613032341003418, "learning_rate": 8.088318523666537e-06, "loss": 0.3933, "step": 5836 }, { "epoch": 0.29526025595629524, "grad_norm": 6.511054515838623, "learning_rate": 8.087687274063106e-06, "loss": 0.3616, "step": 5837 }, { "epoch": 0.2953108402043604, "grad_norm": 13.486564636230469, "learning_rate": 8.087055944897434e-06, "loss": 0.4588, "step": 5838 }, { "epoch": 0.2953614244524255, "grad_norm": 8.79050350189209, "learning_rate": 8.086424536185788e-06, "loss": 0.4461, "step": 5839 }, { "epoch": 0.29541200870049067, "grad_norm": 10.179166793823242, "learning_rate": 8.085793047944435e-06, "loss": 0.4832, "step": 5840 }, { "epoch": 0.2954625929485558, "grad_norm": 4.652525424957275, "learning_rate": 8.085161480189647e-06, "loss": 0.351, "step": 5841 }, { "epoch": 0.29551317719662096, "grad_norm": 6.398046970367432, "learning_rate": 8.084529832937702e-06, "loss": 0.3813, "step": 5842 }, { "epoch": 0.2955637614446861, "grad_norm": 9.912537574768066, "learning_rate": 8.083898106204872e-06, "loss": 0.4275, "step": 5843 }, { "epoch": 0.2956143456927513, "grad_norm": 9.253525733947754, "learning_rate": 8.08326630000744e-06, "loss": 0.4465, "step": 5844 }, { "epoch": 0.29566492994081645, "grad_norm": 6.781515121459961, "learning_rate": 8.082634414361678e-06, "loss": 0.4665, "step": 5845 }, { "epoch": 0.2957155141888816, "grad_norm": 5.172367572784424, "learning_rate": 8.082002449283877e-06, "loss": 0.3505, "step": 5846 }, { "epoch": 0.29576609843694673, "grad_norm": 4.428964138031006, "learning_rate": 8.081370404790315e-06, "loss": 0.4136, "step": 5847 }, { "epoch": 0.2958166826850119, "grad_norm": 14.206138610839844, "learning_rate": 8.08073828089728e-06, "loss": 0.4812, "step": 5848 }, { "epoch": 0.295867266933077, "grad_norm": 8.020392417907715, "learning_rate": 8.080106077621061e-06, "loss": 0.4581, "step": 5849 }, { "epoch": 0.29591785118114217, "grad_norm": 7.849334239959717, "learning_rate": 8.07947379497795e-06, "loss": 0.4178, "step": 5850 }, { "epoch": 0.29596843542920737, "grad_norm": 8.160533905029297, "learning_rate": 8.078841432984235e-06, "loss": 0.451, "step": 5851 }, { "epoch": 0.2960190196772725, "grad_norm": 8.81794548034668, "learning_rate": 8.078208991656215e-06, "loss": 0.4506, "step": 5852 }, { "epoch": 0.29606960392533765, "grad_norm": 9.605005264282227, "learning_rate": 8.077576471010183e-06, "loss": 0.4711, "step": 5853 }, { "epoch": 0.2961201881734028, "grad_norm": 10.70217514038086, "learning_rate": 8.07694387106244e-06, "loss": 0.4509, "step": 5854 }, { "epoch": 0.29617077242146794, "grad_norm": 4.862741947174072, "learning_rate": 8.076311191829284e-06, "loss": 0.4032, "step": 5855 }, { "epoch": 0.2962213566695331, "grad_norm": 8.064144134521484, "learning_rate": 8.07567843332702e-06, "loss": 0.4805, "step": 5856 }, { "epoch": 0.2962719409175983, "grad_norm": 13.632455825805664, "learning_rate": 8.075045595571951e-06, "loss": 0.4717, "step": 5857 }, { "epoch": 0.29632252516566343, "grad_norm": 11.213006019592285, "learning_rate": 8.074412678580386e-06, "loss": 0.5298, "step": 5858 }, { "epoch": 0.2963731094137286, "grad_norm": 11.576446533203125, "learning_rate": 8.073779682368632e-06, "loss": 0.53, "step": 5859 }, { "epoch": 0.2964236936617937, "grad_norm": 17.438865661621094, "learning_rate": 8.073146606952998e-06, "loss": 0.381, "step": 5860 }, { "epoch": 0.29647427790985886, "grad_norm": 9.379735946655273, "learning_rate": 8.072513452349802e-06, "loss": 0.4292, "step": 5861 }, { "epoch": 0.296524862157924, "grad_norm": 12.959620475769043, "learning_rate": 8.071880218575351e-06, "loss": 0.4602, "step": 5862 }, { "epoch": 0.29657544640598915, "grad_norm": 4.3365936279296875, "learning_rate": 8.071246905645971e-06, "loss": 0.2826, "step": 5863 }, { "epoch": 0.29662603065405435, "grad_norm": 6.019278526306152, "learning_rate": 8.070613513577976e-06, "loss": 0.44, "step": 5864 }, { "epoch": 0.2966766149021195, "grad_norm": 8.552559852600098, "learning_rate": 8.069980042387686e-06, "loss": 0.4254, "step": 5865 }, { "epoch": 0.29672719915018464, "grad_norm": 8.044715881347656, "learning_rate": 8.069346492091426e-06, "loss": 0.2679, "step": 5866 }, { "epoch": 0.2967777833982498, "grad_norm": 9.125775337219238, "learning_rate": 8.06871286270552e-06, "loss": 0.4703, "step": 5867 }, { "epoch": 0.29682836764631493, "grad_norm": 5.951451778411865, "learning_rate": 8.068079154246296e-06, "loss": 0.3558, "step": 5868 }, { "epoch": 0.29687895189438007, "grad_norm": 7.380189418792725, "learning_rate": 8.067445366730085e-06, "loss": 0.441, "step": 5869 }, { "epoch": 0.2969295361424452, "grad_norm": 7.442131519317627, "learning_rate": 8.066811500173213e-06, "loss": 0.4571, "step": 5870 }, { "epoch": 0.2969801203905104, "grad_norm": 13.693573951721191, "learning_rate": 8.066177554592018e-06, "loss": 0.3771, "step": 5871 }, { "epoch": 0.29703070463857556, "grad_norm": 8.96458625793457, "learning_rate": 8.065543530002834e-06, "loss": 0.4764, "step": 5872 }, { "epoch": 0.2970812888866407, "grad_norm": 7.2654194831848145, "learning_rate": 8.064909426421996e-06, "loss": 0.5243, "step": 5873 }, { "epoch": 0.29713187313470585, "grad_norm": 9.534188270568848, "learning_rate": 8.064275243865844e-06, "loss": 0.4998, "step": 5874 }, { "epoch": 0.297182457382771, "grad_norm": 8.701532363891602, "learning_rate": 8.063640982350722e-06, "loss": 0.4754, "step": 5875 }, { "epoch": 0.29723304163083614, "grad_norm": 6.014329433441162, "learning_rate": 8.063006641892974e-06, "loss": 0.3642, "step": 5876 }, { "epoch": 0.29728362587890134, "grad_norm": 19.62738609313965, "learning_rate": 8.062372222508941e-06, "loss": 0.4504, "step": 5877 }, { "epoch": 0.2973342101269665, "grad_norm": 6.160677433013916, "learning_rate": 8.061737724214971e-06, "loss": 0.402, "step": 5878 }, { "epoch": 0.2973847943750316, "grad_norm": 9.374322891235352, "learning_rate": 8.061103147027418e-06, "loss": 0.3708, "step": 5879 }, { "epoch": 0.29743537862309677, "grad_norm": 9.020705223083496, "learning_rate": 8.06046849096263e-06, "loss": 0.4658, "step": 5880 }, { "epoch": 0.2974859628711619, "grad_norm": 7.3279852867126465, "learning_rate": 8.05983375603696e-06, "loss": 0.434, "step": 5881 }, { "epoch": 0.29753654711922706, "grad_norm": 17.524919509887695, "learning_rate": 8.059198942266766e-06, "loss": 0.496, "step": 5882 }, { "epoch": 0.2975871313672922, "grad_norm": 10.63048267364502, "learning_rate": 8.058564049668405e-06, "loss": 0.4957, "step": 5883 }, { "epoch": 0.2976377156153574, "grad_norm": 9.804868698120117, "learning_rate": 8.057929078258235e-06, "loss": 0.4676, "step": 5884 }, { "epoch": 0.29768829986342255, "grad_norm": 10.893498420715332, "learning_rate": 8.05729402805262e-06, "loss": 0.4294, "step": 5885 }, { "epoch": 0.2977388841114877, "grad_norm": 11.555486679077148, "learning_rate": 8.056658899067921e-06, "loss": 0.4021, "step": 5886 }, { "epoch": 0.29778946835955283, "grad_norm": 8.318108558654785, "learning_rate": 8.056023691320506e-06, "loss": 0.4735, "step": 5887 }, { "epoch": 0.297840052607618, "grad_norm": 6.817720890045166, "learning_rate": 8.05538840482674e-06, "loss": 0.3871, "step": 5888 }, { "epoch": 0.2978906368556831, "grad_norm": 10.018046379089355, "learning_rate": 8.054753039602998e-06, "loss": 0.4135, "step": 5889 }, { "epoch": 0.29794122110374827, "grad_norm": 9.346559524536133, "learning_rate": 8.054117595665647e-06, "loss": 0.5014, "step": 5890 }, { "epoch": 0.29799180535181347, "grad_norm": 8.011326789855957, "learning_rate": 8.053482073031063e-06, "loss": 0.3975, "step": 5891 }, { "epoch": 0.2980423895998786, "grad_norm": 19.241430282592773, "learning_rate": 8.05284647171562e-06, "loss": 0.4928, "step": 5892 }, { "epoch": 0.29809297384794375, "grad_norm": 12.675604820251465, "learning_rate": 8.0522107917357e-06, "loss": 0.5389, "step": 5893 }, { "epoch": 0.2981435580960089, "grad_norm": 5.814464569091797, "learning_rate": 8.051575033107678e-06, "loss": 0.4022, "step": 5894 }, { "epoch": 0.29819414234407404, "grad_norm": 6.864404678344727, "learning_rate": 8.05093919584794e-06, "loss": 0.3536, "step": 5895 }, { "epoch": 0.2982447265921392, "grad_norm": 6.8153228759765625, "learning_rate": 8.050303279972866e-06, "loss": 0.4426, "step": 5896 }, { "epoch": 0.2982953108402044, "grad_norm": 11.441813468933105, "learning_rate": 8.049667285498845e-06, "loss": 0.5424, "step": 5897 }, { "epoch": 0.29834589508826953, "grad_norm": 11.974428176879883, "learning_rate": 8.049031212442264e-06, "loss": 0.5526, "step": 5898 }, { "epoch": 0.2983964793363347, "grad_norm": 8.31831169128418, "learning_rate": 8.048395060819514e-06, "loss": 0.4679, "step": 5899 }, { "epoch": 0.2984470635843998, "grad_norm": 6.042891502380371, "learning_rate": 8.047758830646986e-06, "loss": 0.358, "step": 5900 }, { "epoch": 0.29849764783246496, "grad_norm": 7.0181474685668945, "learning_rate": 8.047122521941075e-06, "loss": 0.4801, "step": 5901 }, { "epoch": 0.2985482320805301, "grad_norm": 7.838071823120117, "learning_rate": 8.046486134718176e-06, "loss": 0.3581, "step": 5902 }, { "epoch": 0.29859881632859525, "grad_norm": 6.076024532318115, "learning_rate": 8.045849668994688e-06, "loss": 0.4048, "step": 5903 }, { "epoch": 0.29864940057666045, "grad_norm": 6.556032180786133, "learning_rate": 8.04521312478701e-06, "loss": 0.4319, "step": 5904 }, { "epoch": 0.2986999848247256, "grad_norm": 6.398456573486328, "learning_rate": 8.044576502111548e-06, "loss": 0.4828, "step": 5905 }, { "epoch": 0.29875056907279074, "grad_norm": 10.466696739196777, "learning_rate": 8.043939800984698e-06, "loss": 0.5453, "step": 5906 }, { "epoch": 0.2988011533208559, "grad_norm": 4.373688220977783, "learning_rate": 8.043303021422877e-06, "loss": 0.3602, "step": 5907 }, { "epoch": 0.298851737568921, "grad_norm": 6.6228413581848145, "learning_rate": 8.042666163442487e-06, "loss": 0.4028, "step": 5908 }, { "epoch": 0.29890232181698617, "grad_norm": 11.090486526489258, "learning_rate": 8.042029227059937e-06, "loss": 0.5352, "step": 5909 }, { "epoch": 0.2989529060650513, "grad_norm": 5.350664138793945, "learning_rate": 8.041392212291643e-06, "loss": 0.3387, "step": 5910 }, { "epoch": 0.2990034903131165, "grad_norm": 7.11354923248291, "learning_rate": 8.040755119154018e-06, "loss": 0.4485, "step": 5911 }, { "epoch": 0.29905407456118166, "grad_norm": 10.393449783325195, "learning_rate": 8.040117947663478e-06, "loss": 0.521, "step": 5912 }, { "epoch": 0.2991046588092468, "grad_norm": 12.636792182922363, "learning_rate": 8.039480697836442e-06, "loss": 0.5562, "step": 5913 }, { "epoch": 0.29915524305731195, "grad_norm": 6.059262275695801, "learning_rate": 8.038843369689328e-06, "loss": 0.4013, "step": 5914 }, { "epoch": 0.2992058273053771, "grad_norm": 10.708521842956543, "learning_rate": 8.038205963238564e-06, "loss": 0.4449, "step": 5915 }, { "epoch": 0.29925641155344224, "grad_norm": 16.747852325439453, "learning_rate": 8.037568478500569e-06, "loss": 0.4616, "step": 5916 }, { "epoch": 0.29930699580150744, "grad_norm": 7.129931926727295, "learning_rate": 8.03693091549177e-06, "loss": 0.5018, "step": 5917 }, { "epoch": 0.2993575800495726, "grad_norm": 6.974902153015137, "learning_rate": 8.036293274228598e-06, "loss": 0.4906, "step": 5918 }, { "epoch": 0.2994081642976377, "grad_norm": 8.970781326293945, "learning_rate": 8.035655554727481e-06, "loss": 0.4404, "step": 5919 }, { "epoch": 0.29945874854570287, "grad_norm": 4.6770195960998535, "learning_rate": 8.035017757004853e-06, "loss": 0.4201, "step": 5920 }, { "epoch": 0.299509332793768, "grad_norm": 7.161413669586182, "learning_rate": 8.03437988107715e-06, "loss": 0.3978, "step": 5921 }, { "epoch": 0.29955991704183316, "grad_norm": 9.303136825561523, "learning_rate": 8.033741926960805e-06, "loss": 0.5183, "step": 5922 }, { "epoch": 0.2996105012898983, "grad_norm": 8.916744232177734, "learning_rate": 8.033103894672258e-06, "loss": 0.4587, "step": 5923 }, { "epoch": 0.2996610855379635, "grad_norm": 8.753273010253906, "learning_rate": 8.032465784227952e-06, "loss": 0.3255, "step": 5924 }, { "epoch": 0.29971166978602864, "grad_norm": 8.328261375427246, "learning_rate": 8.031827595644325e-06, "loss": 0.3696, "step": 5925 }, { "epoch": 0.2997622540340938, "grad_norm": 5.899117469787598, "learning_rate": 8.031189328937825e-06, "loss": 0.4267, "step": 5926 }, { "epoch": 0.29981283828215893, "grad_norm": 6.416265964508057, "learning_rate": 8.030550984124897e-06, "loss": 0.4328, "step": 5927 }, { "epoch": 0.2998634225302241, "grad_norm": 10.569604873657227, "learning_rate": 8.029912561221991e-06, "loss": 0.5615, "step": 5928 }, { "epoch": 0.2999140067782892, "grad_norm": 5.3758769035339355, "learning_rate": 8.029274060245556e-06, "loss": 0.4607, "step": 5929 }, { "epoch": 0.29996459102635437, "grad_norm": 6.648048400878906, "learning_rate": 8.028635481212045e-06, "loss": 0.5184, "step": 5930 }, { "epoch": 0.30001517527441957, "grad_norm": 10.213550567626953, "learning_rate": 8.027996824137913e-06, "loss": 0.3639, "step": 5931 }, { "epoch": 0.30001517527441957, "eval_loss": 0.43704280257225037, "eval_runtime": 34.9027, "eval_samples_per_second": 45.784, "eval_steps_per_second": 5.73, "step": 5931 }, { "epoch": 0.3000657595224847, "grad_norm": 7.982204437255859, "learning_rate": 8.027358089039618e-06, "loss": 0.4017, "step": 5932 }, { "epoch": 0.30011634377054985, "grad_norm": 13.84660816192627, "learning_rate": 8.026719275933616e-06, "loss": 0.4893, "step": 5933 }, { "epoch": 0.300166928018615, "grad_norm": 7.9099225997924805, "learning_rate": 8.026080384836371e-06, "loss": 0.4053, "step": 5934 }, { "epoch": 0.30021751226668014, "grad_norm": 6.6512980461120605, "learning_rate": 8.025441415764342e-06, "loss": 0.4187, "step": 5935 }, { "epoch": 0.3002680965147453, "grad_norm": 8.646206855773926, "learning_rate": 8.024802368733995e-06, "loss": 0.4675, "step": 5936 }, { "epoch": 0.3003186807628105, "grad_norm": 8.140741348266602, "learning_rate": 8.024163243761799e-06, "loss": 0.3859, "step": 5937 }, { "epoch": 0.30036926501087563, "grad_norm": 7.074759483337402, "learning_rate": 8.02352404086422e-06, "loss": 0.4795, "step": 5938 }, { "epoch": 0.3004198492589408, "grad_norm": 8.961745262145996, "learning_rate": 8.022884760057727e-06, "loss": 0.498, "step": 5939 }, { "epoch": 0.3004704335070059, "grad_norm": 8.209885597229004, "learning_rate": 8.022245401358799e-06, "loss": 0.4262, "step": 5940 }, { "epoch": 0.30052101775507106, "grad_norm": 8.105843544006348, "learning_rate": 8.021605964783905e-06, "loss": 0.4642, "step": 5941 }, { "epoch": 0.3005716020031362, "grad_norm": 5.817131042480469, "learning_rate": 8.020966450349524e-06, "loss": 0.3655, "step": 5942 }, { "epoch": 0.30062218625120135, "grad_norm": 9.978034019470215, "learning_rate": 8.020326858072134e-06, "loss": 0.3853, "step": 5943 }, { "epoch": 0.30067277049926655, "grad_norm": 5.970256328582764, "learning_rate": 8.019687187968215e-06, "loss": 0.4644, "step": 5944 }, { "epoch": 0.3007233547473317, "grad_norm": 14.792638778686523, "learning_rate": 8.019047440054252e-06, "loss": 0.4526, "step": 5945 }, { "epoch": 0.30077393899539684, "grad_norm": 8.35116195678711, "learning_rate": 8.01840761434673e-06, "loss": 0.4209, "step": 5946 }, { "epoch": 0.300824523243462, "grad_norm": 10.758857727050781, "learning_rate": 8.017767710862134e-06, "loss": 0.437, "step": 5947 }, { "epoch": 0.3008751074915271, "grad_norm": 10.726729393005371, "learning_rate": 8.017127729616953e-06, "loss": 0.4485, "step": 5948 }, { "epoch": 0.30092569173959227, "grad_norm": 10.307196617126465, "learning_rate": 8.016487670627677e-06, "loss": 0.4367, "step": 5949 }, { "epoch": 0.30097627598765747, "grad_norm": 7.660849571228027, "learning_rate": 8.0158475339108e-06, "loss": 0.3503, "step": 5950 }, { "epoch": 0.3010268602357226, "grad_norm": 11.938316345214844, "learning_rate": 8.015207319482818e-06, "loss": 0.4979, "step": 5951 }, { "epoch": 0.30107744448378776, "grad_norm": 7.989025115966797, "learning_rate": 8.014567027360226e-06, "loss": 0.3837, "step": 5952 }, { "epoch": 0.3011280287318529, "grad_norm": 10.843927383422852, "learning_rate": 8.013926657559521e-06, "loss": 0.423, "step": 5953 }, { "epoch": 0.30117861297991805, "grad_norm": 6.999736309051514, "learning_rate": 8.013286210097207e-06, "loss": 0.4487, "step": 5954 }, { "epoch": 0.3012291972279832, "grad_norm": 10.593591690063477, "learning_rate": 8.012645684989786e-06, "loss": 0.3286, "step": 5955 }, { "epoch": 0.30127978147604834, "grad_norm": 6.391602993011475, "learning_rate": 8.012005082253763e-06, "loss": 0.4193, "step": 5956 }, { "epoch": 0.30133036572411354, "grad_norm": 15.934300422668457, "learning_rate": 8.011364401905643e-06, "loss": 0.4459, "step": 5957 }, { "epoch": 0.3013809499721787, "grad_norm": 7.153628826141357, "learning_rate": 8.010723643961937e-06, "loss": 0.4307, "step": 5958 }, { "epoch": 0.3014315342202438, "grad_norm": 9.483433723449707, "learning_rate": 8.010082808439154e-06, "loss": 0.3915, "step": 5959 }, { "epoch": 0.30148211846830897, "grad_norm": 54.66865158081055, "learning_rate": 8.009441895353807e-06, "loss": 0.5022, "step": 5960 }, { "epoch": 0.3015327027163741, "grad_norm": 5.35835599899292, "learning_rate": 8.008800904722413e-06, "loss": 0.4178, "step": 5961 }, { "epoch": 0.30158328696443926, "grad_norm": 10.464736938476562, "learning_rate": 8.008159836561487e-06, "loss": 0.4522, "step": 5962 }, { "epoch": 0.3016338712125044, "grad_norm": 9.152505874633789, "learning_rate": 8.007518690887546e-06, "loss": 0.5248, "step": 5963 }, { "epoch": 0.3016844554605696, "grad_norm": 16.245729446411133, "learning_rate": 8.006877467717114e-06, "loss": 0.5094, "step": 5964 }, { "epoch": 0.30173503970863474, "grad_norm": 6.0188117027282715, "learning_rate": 8.006236167066713e-06, "loss": 0.4426, "step": 5965 }, { "epoch": 0.3017856239566999, "grad_norm": 6.006434440612793, "learning_rate": 8.005594788952869e-06, "loss": 0.4853, "step": 5966 }, { "epoch": 0.30183620820476503, "grad_norm": 4.786380767822266, "learning_rate": 8.004953333392104e-06, "loss": 0.4334, "step": 5967 }, { "epoch": 0.3018867924528302, "grad_norm": 5.605072975158691, "learning_rate": 8.00431180040095e-06, "loss": 0.3615, "step": 5968 }, { "epoch": 0.3019373767008953, "grad_norm": 6.002177715301514, "learning_rate": 8.003670189995938e-06, "loss": 0.3472, "step": 5969 }, { "epoch": 0.3019879609489605, "grad_norm": 8.937094688415527, "learning_rate": 8.003028502193601e-06, "loss": 0.4077, "step": 5970 }, { "epoch": 0.30203854519702567, "grad_norm": 6.916387557983398, "learning_rate": 8.002386737010472e-06, "loss": 0.4692, "step": 5971 }, { "epoch": 0.3020891294450908, "grad_norm": 8.403233528137207, "learning_rate": 8.001744894463088e-06, "loss": 0.4288, "step": 5972 }, { "epoch": 0.30213971369315595, "grad_norm": 16.693599700927734, "learning_rate": 8.001102974567992e-06, "loss": 0.6428, "step": 5973 }, { "epoch": 0.3021902979412211, "grad_norm": 8.972055435180664, "learning_rate": 8.000460977341717e-06, "loss": 0.4585, "step": 5974 }, { "epoch": 0.30224088218928624, "grad_norm": 10.20683479309082, "learning_rate": 7.999818902800812e-06, "loss": 0.4162, "step": 5975 }, { "epoch": 0.3022914664373514, "grad_norm": 9.560713768005371, "learning_rate": 7.999176750961819e-06, "loss": 0.4571, "step": 5976 }, { "epoch": 0.3023420506854166, "grad_norm": 11.84677791595459, "learning_rate": 7.998534521841286e-06, "loss": 0.4857, "step": 5977 }, { "epoch": 0.30239263493348173, "grad_norm": 8.537166595458984, "learning_rate": 7.99789221545576e-06, "loss": 0.3043, "step": 5978 }, { "epoch": 0.3024432191815469, "grad_norm": 5.613765716552734, "learning_rate": 7.997249831821795e-06, "loss": 0.4594, "step": 5979 }, { "epoch": 0.302493803429612, "grad_norm": 5.2333173751831055, "learning_rate": 7.99660737095594e-06, "loss": 0.4816, "step": 5980 }, { "epoch": 0.30254438767767716, "grad_norm": 5.553018093109131, "learning_rate": 7.99596483287475e-06, "loss": 0.3572, "step": 5981 }, { "epoch": 0.3025949719257423, "grad_norm": 9.101561546325684, "learning_rate": 7.995322217594785e-06, "loss": 0.4794, "step": 5982 }, { "epoch": 0.30264555617380745, "grad_norm": 13.189416885375977, "learning_rate": 7.9946795251326e-06, "loss": 0.5628, "step": 5983 }, { "epoch": 0.30269614042187265, "grad_norm": 6.336364269256592, "learning_rate": 7.994036755504756e-06, "loss": 0.3409, "step": 5984 }, { "epoch": 0.3027467246699378, "grad_norm": 6.334071636199951, "learning_rate": 7.993393908727818e-06, "loss": 0.5493, "step": 5985 }, { "epoch": 0.30279730891800294, "grad_norm": 8.085088729858398, "learning_rate": 7.992750984818349e-06, "loss": 0.3656, "step": 5986 }, { "epoch": 0.3028478931660681, "grad_norm": 10.398467063903809, "learning_rate": 7.992107983792917e-06, "loss": 0.4882, "step": 5987 }, { "epoch": 0.3028984774141332, "grad_norm": 9.013534545898438, "learning_rate": 7.991464905668088e-06, "loss": 0.5179, "step": 5988 }, { "epoch": 0.30294906166219837, "grad_norm": 5.732310771942139, "learning_rate": 7.990821750460433e-06, "loss": 0.3645, "step": 5989 }, { "epoch": 0.30299964591026357, "grad_norm": 6.149893760681152, "learning_rate": 7.990178518186527e-06, "loss": 0.388, "step": 5990 }, { "epoch": 0.3030502301583287, "grad_norm": 10.286028861999512, "learning_rate": 7.989535208862942e-06, "loss": 0.4146, "step": 5991 }, { "epoch": 0.30310081440639386, "grad_norm": 9.176424980163574, "learning_rate": 7.988891822506257e-06, "loss": 0.2661, "step": 5992 }, { "epoch": 0.303151398654459, "grad_norm": 5.751551628112793, "learning_rate": 7.988248359133047e-06, "loss": 0.3682, "step": 5993 }, { "epoch": 0.30320198290252415, "grad_norm": 6.301262855529785, "learning_rate": 7.987604818759896e-06, "loss": 0.4894, "step": 5994 }, { "epoch": 0.3032525671505893, "grad_norm": 7.546384334564209, "learning_rate": 7.986961201403384e-06, "loss": 0.4478, "step": 5995 }, { "epoch": 0.30330315139865444, "grad_norm": 7.077033996582031, "learning_rate": 7.986317507080099e-06, "loss": 0.4711, "step": 5996 }, { "epoch": 0.30335373564671964, "grad_norm": 10.607193946838379, "learning_rate": 7.985673735806623e-06, "loss": 0.4447, "step": 5997 }, { "epoch": 0.3034043198947848, "grad_norm": 6.278976917266846, "learning_rate": 7.985029887599547e-06, "loss": 0.3603, "step": 5998 }, { "epoch": 0.3034549041428499, "grad_norm": 6.326647758483887, "learning_rate": 7.984385962475462e-06, "loss": 0.475, "step": 5999 }, { "epoch": 0.30350548839091507, "grad_norm": 7.006811141967773, "learning_rate": 7.983741960450957e-06, "loss": 0.4682, "step": 6000 }, { "epoch": 0.3035560726389802, "grad_norm": 9.41908073425293, "learning_rate": 7.98309788154263e-06, "loss": 0.5464, "step": 6001 }, { "epoch": 0.30360665688704536, "grad_norm": 9.24552059173584, "learning_rate": 7.982453725767076e-06, "loss": 0.5198, "step": 6002 }, { "epoch": 0.3036572411351105, "grad_norm": 8.119335174560547, "learning_rate": 7.981809493140893e-06, "loss": 0.4464, "step": 6003 }, { "epoch": 0.3037078253831757, "grad_norm": 7.570228576660156, "learning_rate": 7.981165183680682e-06, "loss": 0.4783, "step": 6004 }, { "epoch": 0.30375840963124084, "grad_norm": 9.131942749023438, "learning_rate": 7.980520797403045e-06, "loss": 0.4518, "step": 6005 }, { "epoch": 0.303808993879306, "grad_norm": 5.9383392333984375, "learning_rate": 7.979876334324586e-06, "loss": 0.3982, "step": 6006 }, { "epoch": 0.30385957812737113, "grad_norm": 12.755876541137695, "learning_rate": 7.97923179446191e-06, "loss": 0.6049, "step": 6007 }, { "epoch": 0.3039101623754363, "grad_norm": 7.35593318939209, "learning_rate": 7.978587177831628e-06, "loss": 0.4158, "step": 6008 }, { "epoch": 0.3039607466235014, "grad_norm": 8.31298828125, "learning_rate": 7.977942484450349e-06, "loss": 0.4808, "step": 6009 }, { "epoch": 0.3040113308715666, "grad_norm": 11.214811325073242, "learning_rate": 7.977297714334685e-06, "loss": 0.5385, "step": 6010 }, { "epoch": 0.30406191511963176, "grad_norm": 11.2261381149292, "learning_rate": 7.97665286750125e-06, "loss": 0.3553, "step": 6011 }, { "epoch": 0.3041124993676969, "grad_norm": 5.3299641609191895, "learning_rate": 7.97600794396666e-06, "loss": 0.3793, "step": 6012 }, { "epoch": 0.30416308361576205, "grad_norm": 6.455000877380371, "learning_rate": 7.975362943747534e-06, "loss": 0.3231, "step": 6013 }, { "epoch": 0.3042136678638272, "grad_norm": 11.375055313110352, "learning_rate": 7.974717866860493e-06, "loss": 0.4578, "step": 6014 }, { "epoch": 0.30426425211189234, "grad_norm": 8.008880615234375, "learning_rate": 7.974072713322155e-06, "loss": 0.4492, "step": 6015 }, { "epoch": 0.3043148363599575, "grad_norm": 8.757057189941406, "learning_rate": 7.973427483149148e-06, "loss": 0.4253, "step": 6016 }, { "epoch": 0.3043654206080227, "grad_norm": 8.234402656555176, "learning_rate": 7.972782176358097e-06, "loss": 0.4719, "step": 6017 }, { "epoch": 0.30441600485608783, "grad_norm": 11.146475791931152, "learning_rate": 7.97213679296563e-06, "loss": 0.5039, "step": 6018 }, { "epoch": 0.304466589104153, "grad_norm": 13.237141609191895, "learning_rate": 7.971491332988377e-06, "loss": 0.4844, "step": 6019 }, { "epoch": 0.3045171733522181, "grad_norm": 10.003279685974121, "learning_rate": 7.970845796442969e-06, "loss": 0.5126, "step": 6020 }, { "epoch": 0.30456775760028326, "grad_norm": 4.740987300872803, "learning_rate": 7.97020018334604e-06, "loss": 0.3027, "step": 6021 }, { "epoch": 0.3046183418483484, "grad_norm": 5.327930450439453, "learning_rate": 7.969554493714227e-06, "loss": 0.4039, "step": 6022 }, { "epoch": 0.30466892609641355, "grad_norm": 9.773422241210938, "learning_rate": 7.968908727564169e-06, "loss": 0.4617, "step": 6023 }, { "epoch": 0.30471951034447875, "grad_norm": 7.703380584716797, "learning_rate": 7.968262884912504e-06, "loss": 0.3717, "step": 6024 }, { "epoch": 0.3047700945925439, "grad_norm": 6.884169578552246, "learning_rate": 7.967616965775872e-06, "loss": 0.5033, "step": 6025 }, { "epoch": 0.30482067884060904, "grad_norm": 8.323650360107422, "learning_rate": 7.966970970170919e-06, "loss": 0.4744, "step": 6026 }, { "epoch": 0.3048712630886742, "grad_norm": 6.277127265930176, "learning_rate": 7.966324898114292e-06, "loss": 0.3149, "step": 6027 }, { "epoch": 0.3049218473367393, "grad_norm": 4.924848556518555, "learning_rate": 7.965678749622636e-06, "loss": 0.4047, "step": 6028 }, { "epoch": 0.30497243158480447, "grad_norm": 7.650604724884033, "learning_rate": 7.965032524712601e-06, "loss": 0.3632, "step": 6029 }, { "epoch": 0.30502301583286967, "grad_norm": 7.103761196136475, "learning_rate": 7.964386223400841e-06, "loss": 0.298, "step": 6030 }, { "epoch": 0.3050736000809348, "grad_norm": 8.48161506652832, "learning_rate": 7.96373984570401e-06, "loss": 0.4625, "step": 6031 }, { "epoch": 0.30512418432899996, "grad_norm": 12.37572193145752, "learning_rate": 7.963093391638759e-06, "loss": 0.5585, "step": 6032 }, { "epoch": 0.3051747685770651, "grad_norm": 15.085165977478027, "learning_rate": 7.962446861221748e-06, "loss": 0.401, "step": 6033 }, { "epoch": 0.30522535282513025, "grad_norm": 8.077142715454102, "learning_rate": 7.96180025446964e-06, "loss": 0.4606, "step": 6034 }, { "epoch": 0.3052759370731954, "grad_norm": 8.219539642333984, "learning_rate": 7.961153571399091e-06, "loss": 0.4828, "step": 6035 }, { "epoch": 0.30532652132126054, "grad_norm": 10.865399360656738, "learning_rate": 7.960506812026767e-06, "loss": 0.4843, "step": 6036 }, { "epoch": 0.30537710556932574, "grad_norm": 6.924602508544922, "learning_rate": 7.959859976369335e-06, "loss": 0.5061, "step": 6037 }, { "epoch": 0.3054276898173909, "grad_norm": 13.210586547851562, "learning_rate": 7.959213064443458e-06, "loss": 0.6218, "step": 6038 }, { "epoch": 0.305478274065456, "grad_norm": 7.537676811218262, "learning_rate": 7.95856607626581e-06, "loss": 0.3495, "step": 6039 }, { "epoch": 0.30552885831352117, "grad_norm": 5.947803974151611, "learning_rate": 7.95791901185306e-06, "loss": 0.3855, "step": 6040 }, { "epoch": 0.3055794425615863, "grad_norm": 9.036003112792969, "learning_rate": 7.95727187122188e-06, "loss": 0.4483, "step": 6041 }, { "epoch": 0.30563002680965146, "grad_norm": 9.672781944274902, "learning_rate": 7.956624654388949e-06, "loss": 0.5173, "step": 6042 }, { "epoch": 0.3056806110577166, "grad_norm": 10.162386894226074, "learning_rate": 7.955977361370942e-06, "loss": 0.4547, "step": 6043 }, { "epoch": 0.3057311953057818, "grad_norm": 8.12589168548584, "learning_rate": 7.955329992184537e-06, "loss": 0.4331, "step": 6044 }, { "epoch": 0.30578177955384694, "grad_norm": 14.262454986572266, "learning_rate": 7.954682546846417e-06, "loss": 0.3642, "step": 6045 }, { "epoch": 0.3058323638019121, "grad_norm": 11.426432609558105, "learning_rate": 7.954035025373263e-06, "loss": 0.4963, "step": 6046 }, { "epoch": 0.30588294804997723, "grad_norm": 6.935757637023926, "learning_rate": 7.953387427781764e-06, "loss": 0.4721, "step": 6047 }, { "epoch": 0.3059335322980424, "grad_norm": 8.651371955871582, "learning_rate": 7.952739754088602e-06, "loss": 0.5238, "step": 6048 }, { "epoch": 0.3059841165461075, "grad_norm": 10.541333198547363, "learning_rate": 7.952092004310472e-06, "loss": 0.5799, "step": 6049 }, { "epoch": 0.3060347007941727, "grad_norm": 5.320841312408447, "learning_rate": 7.95144417846406e-06, "loss": 0.4395, "step": 6050 }, { "epoch": 0.30608528504223786, "grad_norm": 8.728021621704102, "learning_rate": 7.95079627656606e-06, "loss": 0.467, "step": 6051 }, { "epoch": 0.306135869290303, "grad_norm": 9.354416847229004, "learning_rate": 7.950148298633166e-06, "loss": 0.4509, "step": 6052 }, { "epoch": 0.30618645353836815, "grad_norm": 9.384142875671387, "learning_rate": 7.949500244682078e-06, "loss": 0.5408, "step": 6053 }, { "epoch": 0.3062370377864333, "grad_norm": 9.606720924377441, "learning_rate": 7.948852114729493e-06, "loss": 0.41, "step": 6054 }, { "epoch": 0.30628762203449844, "grad_norm": 6.367198467254639, "learning_rate": 7.948203908792111e-06, "loss": 0.3832, "step": 6055 }, { "epoch": 0.3063382062825636, "grad_norm": 4.87738561630249, "learning_rate": 7.947555626886637e-06, "loss": 0.4027, "step": 6056 }, { "epoch": 0.3063887905306288, "grad_norm": 9.78907299041748, "learning_rate": 7.946907269029772e-06, "loss": 0.4125, "step": 6057 }, { "epoch": 0.30643937477869393, "grad_norm": 6.449192523956299, "learning_rate": 7.946258835238226e-06, "loss": 0.397, "step": 6058 }, { "epoch": 0.3064899590267591, "grad_norm": 5.922034740447998, "learning_rate": 7.945610325528703e-06, "loss": 0.3241, "step": 6059 }, { "epoch": 0.3065405432748242, "grad_norm": 4.588937759399414, "learning_rate": 7.94496173991792e-06, "loss": 0.3732, "step": 6060 }, { "epoch": 0.30659112752288936, "grad_norm": 6.1643571853637695, "learning_rate": 7.944313078422584e-06, "loss": 0.3791, "step": 6061 }, { "epoch": 0.3066417117709545, "grad_norm": 9.422144889831543, "learning_rate": 7.943664341059413e-06, "loss": 0.3983, "step": 6062 }, { "epoch": 0.30669229601901965, "grad_norm": 6.295482635498047, "learning_rate": 7.943015527845123e-06, "loss": 0.3351, "step": 6063 }, { "epoch": 0.30674288026708485, "grad_norm": 8.443923950195312, "learning_rate": 7.942366638796432e-06, "loss": 0.4904, "step": 6064 }, { "epoch": 0.30679346451515, "grad_norm": 7.353925704956055, "learning_rate": 7.941717673930059e-06, "loss": 0.3992, "step": 6065 }, { "epoch": 0.30684404876321514, "grad_norm": 9.437769889831543, "learning_rate": 7.941068633262724e-06, "loss": 0.4972, "step": 6066 }, { "epoch": 0.3068946330112803, "grad_norm": 7.7591376304626465, "learning_rate": 7.940419516811157e-06, "loss": 0.3778, "step": 6067 }, { "epoch": 0.3069452172593454, "grad_norm": 7.674051761627197, "learning_rate": 7.939770324592081e-06, "loss": 0.3912, "step": 6068 }, { "epoch": 0.30699580150741057, "grad_norm": 7.824158668518066, "learning_rate": 7.939121056622225e-06, "loss": 0.4803, "step": 6069 }, { "epoch": 0.30704638575547577, "grad_norm": 8.158945083618164, "learning_rate": 7.938471712918317e-06, "loss": 0.3954, "step": 6070 }, { "epoch": 0.3070969700035409, "grad_norm": 8.58211898803711, "learning_rate": 7.937822293497091e-06, "loss": 0.3996, "step": 6071 }, { "epoch": 0.30714755425160606, "grad_norm": 8.609268188476562, "learning_rate": 7.937172798375281e-06, "loss": 0.4802, "step": 6072 }, { "epoch": 0.3071981384996712, "grad_norm": 7.337953090667725, "learning_rate": 7.936523227569622e-06, "loss": 0.4324, "step": 6073 }, { "epoch": 0.30724872274773635, "grad_norm": 8.28376293182373, "learning_rate": 7.935873581096854e-06, "loss": 0.4809, "step": 6074 }, { "epoch": 0.3072993069958015, "grad_norm": 10.854761123657227, "learning_rate": 7.935223858973712e-06, "loss": 0.4657, "step": 6075 }, { "epoch": 0.30734989124386664, "grad_norm": 8.855203628540039, "learning_rate": 7.934574061216944e-06, "loss": 0.4081, "step": 6076 }, { "epoch": 0.30740047549193183, "grad_norm": 6.877864360809326, "learning_rate": 7.93392418784329e-06, "loss": 0.4282, "step": 6077 }, { "epoch": 0.307451059739997, "grad_norm": 7.036245822906494, "learning_rate": 7.933274238869494e-06, "loss": 0.3431, "step": 6078 }, { "epoch": 0.3075016439880621, "grad_norm": 9.0259428024292, "learning_rate": 7.932624214312307e-06, "loss": 0.4334, "step": 6079 }, { "epoch": 0.30755222823612727, "grad_norm": 6.126731872558594, "learning_rate": 7.931974114188478e-06, "loss": 0.4214, "step": 6080 }, { "epoch": 0.3076028124841924, "grad_norm": 6.345946311950684, "learning_rate": 7.931323938514758e-06, "loss": 0.3367, "step": 6081 }, { "epoch": 0.30765339673225756, "grad_norm": 7.3548359870910645, "learning_rate": 7.930673687307902e-06, "loss": 0.3523, "step": 6082 }, { "epoch": 0.3077039809803227, "grad_norm": 8.396371841430664, "learning_rate": 7.93002336058466e-06, "loss": 0.4545, "step": 6083 }, { "epoch": 0.3077545652283879, "grad_norm": 19.39675521850586, "learning_rate": 7.929372958361797e-06, "loss": 0.3824, "step": 6084 }, { "epoch": 0.30780514947645304, "grad_norm": 8.143473625183105, "learning_rate": 7.928722480656065e-06, "loss": 0.3586, "step": 6085 }, { "epoch": 0.3078557337245182, "grad_norm": 10.544808387756348, "learning_rate": 7.928071927484232e-06, "loss": 0.4937, "step": 6086 }, { "epoch": 0.30790631797258333, "grad_norm": 10.808815956115723, "learning_rate": 7.927421298863055e-06, "loss": 0.5211, "step": 6087 }, { "epoch": 0.3079569022206485, "grad_norm": 13.626877784729004, "learning_rate": 7.926770594809303e-06, "loss": 0.4848, "step": 6088 }, { "epoch": 0.3080074864687136, "grad_norm": 6.0525712966918945, "learning_rate": 7.926119815339743e-06, "loss": 0.3965, "step": 6089 }, { "epoch": 0.3080580707167788, "grad_norm": 8.920275688171387, "learning_rate": 7.925468960471143e-06, "loss": 0.3958, "step": 6090 }, { "epoch": 0.30810865496484396, "grad_norm": 8.014904022216797, "learning_rate": 7.924818030220274e-06, "loss": 0.3552, "step": 6091 }, { "epoch": 0.3081592392129091, "grad_norm": 6.42428731918335, "learning_rate": 7.924167024603907e-06, "loss": 0.4175, "step": 6092 }, { "epoch": 0.30820982346097425, "grad_norm": 7.042340278625488, "learning_rate": 7.92351594363882e-06, "loss": 0.445, "step": 6093 }, { "epoch": 0.3082604077090394, "grad_norm": 5.818375110626221, "learning_rate": 7.922864787341792e-06, "loss": 0.3257, "step": 6094 }, { "epoch": 0.30831099195710454, "grad_norm": 6.951807022094727, "learning_rate": 7.922213555729595e-06, "loss": 0.4643, "step": 6095 }, { "epoch": 0.3083615762051697, "grad_norm": 9.097579956054688, "learning_rate": 7.921562248819013e-06, "loss": 0.4603, "step": 6096 }, { "epoch": 0.3084121604532349, "grad_norm": 14.30540943145752, "learning_rate": 7.920910866626828e-06, "loss": 0.458, "step": 6097 }, { "epoch": 0.30846274470130003, "grad_norm": 5.358302116394043, "learning_rate": 7.920259409169827e-06, "loss": 0.4378, "step": 6098 }, { "epoch": 0.3085133289493652, "grad_norm": 10.928791999816895, "learning_rate": 7.919607876464793e-06, "loss": 0.5189, "step": 6099 }, { "epoch": 0.3085639131974303, "grad_norm": 6.2427978515625, "learning_rate": 7.918956268528516e-06, "loss": 0.4137, "step": 6100 }, { "epoch": 0.30861449744549546, "grad_norm": 5.392481803894043, "learning_rate": 7.918304585377788e-06, "loss": 0.4285, "step": 6101 }, { "epoch": 0.3086650816935606, "grad_norm": 6.291604518890381, "learning_rate": 7.917652827029398e-06, "loss": 0.431, "step": 6102 }, { "epoch": 0.3087156659416258, "grad_norm": 7.772749423980713, "learning_rate": 7.917000993500144e-06, "loss": 0.4718, "step": 6103 }, { "epoch": 0.30876625018969095, "grad_norm": 9.139630317687988, "learning_rate": 7.916349084806819e-06, "loss": 0.3838, "step": 6104 }, { "epoch": 0.3088168344377561, "grad_norm": 6.55648136138916, "learning_rate": 7.91569710096622e-06, "loss": 0.4081, "step": 6105 }, { "epoch": 0.30886741868582124, "grad_norm": 5.479441165924072, "learning_rate": 7.915045041995153e-06, "loss": 0.4095, "step": 6106 }, { "epoch": 0.3089180029338864, "grad_norm": 7.3610944747924805, "learning_rate": 7.914392907910413e-06, "loss": 0.4981, "step": 6107 }, { "epoch": 0.3089685871819515, "grad_norm": 7.239968776702881, "learning_rate": 7.91374069872881e-06, "loss": 0.4347, "step": 6108 }, { "epoch": 0.30901917143001667, "grad_norm": 18.952922821044922, "learning_rate": 7.913088414467146e-06, "loss": 0.6153, "step": 6109 }, { "epoch": 0.30906975567808187, "grad_norm": 13.949522972106934, "learning_rate": 7.91243605514223e-06, "loss": 0.471, "step": 6110 }, { "epoch": 0.309120339926147, "grad_norm": 13.786664009094238, "learning_rate": 7.911783620770872e-06, "loss": 0.5092, "step": 6111 }, { "epoch": 0.30917092417421216, "grad_norm": 8.927530288696289, "learning_rate": 7.91113111136988e-06, "loss": 0.3846, "step": 6112 }, { "epoch": 0.3092215084222773, "grad_norm": 10.161872863769531, "learning_rate": 7.910478526956074e-06, "loss": 0.5556, "step": 6113 }, { "epoch": 0.30927209267034245, "grad_norm": 8.488269805908203, "learning_rate": 7.909825867546268e-06, "loss": 0.4683, "step": 6114 }, { "epoch": 0.3093226769184076, "grad_norm": 8.306230545043945, "learning_rate": 7.909173133157276e-06, "loss": 0.416, "step": 6115 }, { "epoch": 0.30937326116647273, "grad_norm": 5.122745990753174, "learning_rate": 7.908520323805917e-06, "loss": 0.4444, "step": 6116 }, { "epoch": 0.30942384541453793, "grad_norm": 6.934612274169922, "learning_rate": 7.907867439509015e-06, "loss": 0.339, "step": 6117 }, { "epoch": 0.3094744296626031, "grad_norm": 6.5775532722473145, "learning_rate": 7.907214480283395e-06, "loss": 0.3958, "step": 6118 }, { "epoch": 0.3095250139106682, "grad_norm": 10.024406433105469, "learning_rate": 7.906561446145878e-06, "loss": 0.4474, "step": 6119 }, { "epoch": 0.30957559815873337, "grad_norm": 5.334457874298096, "learning_rate": 7.905908337113296e-06, "loss": 0.4801, "step": 6120 }, { "epoch": 0.3096261824067985, "grad_norm": 8.899248123168945, "learning_rate": 7.905255153202472e-06, "loss": 0.4356, "step": 6121 }, { "epoch": 0.30967676665486366, "grad_norm": 11.523513793945312, "learning_rate": 7.904601894430241e-06, "loss": 0.4621, "step": 6122 }, { "epoch": 0.30972735090292886, "grad_norm": 6.682706832885742, "learning_rate": 7.903948560813435e-06, "loss": 0.3652, "step": 6123 }, { "epoch": 0.309777935150994, "grad_norm": 14.210916519165039, "learning_rate": 7.903295152368888e-06, "loss": 0.5484, "step": 6124 }, { "epoch": 0.30982851939905914, "grad_norm": 8.008096694946289, "learning_rate": 7.902641669113438e-06, "loss": 0.4957, "step": 6125 }, { "epoch": 0.3098791036471243, "grad_norm": 14.861429214477539, "learning_rate": 7.901988111063924e-06, "loss": 0.6072, "step": 6126 }, { "epoch": 0.30992968789518943, "grad_norm": 14.533639907836914, "learning_rate": 7.901334478237185e-06, "loss": 0.484, "step": 6127 }, { "epoch": 0.3099802721432546, "grad_norm": 8.43812084197998, "learning_rate": 7.900680770650066e-06, "loss": 0.4651, "step": 6128 }, { "epoch": 0.3100308563913197, "grad_norm": 20.383779525756836, "learning_rate": 7.90002698831941e-06, "loss": 0.6239, "step": 6129 }, { "epoch": 0.3100814406393849, "grad_norm": 14.943209648132324, "learning_rate": 7.89937313126206e-06, "loss": 0.5887, "step": 6130 }, { "epoch": 0.31013202488745006, "grad_norm": 7.413829326629639, "learning_rate": 7.898719199494868e-06, "loss": 0.479, "step": 6131 }, { "epoch": 0.3101826091355152, "grad_norm": 11.343477249145508, "learning_rate": 7.898065193034686e-06, "loss": 0.3222, "step": 6132 }, { "epoch": 0.31023319338358035, "grad_norm": 7.6032819747924805, "learning_rate": 7.897411111898365e-06, "loss": 0.35, "step": 6133 }, { "epoch": 0.3102837776316455, "grad_norm": 14.009100914001465, "learning_rate": 7.896756956102756e-06, "loss": 0.6924, "step": 6134 }, { "epoch": 0.31033436187971064, "grad_norm": 8.167391777038574, "learning_rate": 7.896102725664716e-06, "loss": 0.4254, "step": 6135 }, { "epoch": 0.3103849461277758, "grad_norm": 5.475343704223633, "learning_rate": 7.895448420601107e-06, "loss": 0.4318, "step": 6136 }, { "epoch": 0.310435530375841, "grad_norm": 8.638649940490723, "learning_rate": 7.894794040928784e-06, "loss": 0.5725, "step": 6137 }, { "epoch": 0.31048611462390613, "grad_norm": 6.333282947540283, "learning_rate": 7.894139586664609e-06, "loss": 0.4208, "step": 6138 }, { "epoch": 0.3105366988719713, "grad_norm": 13.766050338745117, "learning_rate": 7.893485057825449e-06, "loss": 0.6467, "step": 6139 }, { "epoch": 0.3105872831200364, "grad_norm": 6.286906719207764, "learning_rate": 7.892830454428169e-06, "loss": 0.5104, "step": 6140 }, { "epoch": 0.31063786736810156, "grad_norm": 11.737326622009277, "learning_rate": 7.892175776489635e-06, "loss": 0.4536, "step": 6141 }, { "epoch": 0.3106884516161667, "grad_norm": 5.390022277832031, "learning_rate": 7.891521024026718e-06, "loss": 0.4318, "step": 6142 }, { "epoch": 0.3107390358642319, "grad_norm": 6.3829450607299805, "learning_rate": 7.890866197056286e-06, "loss": 0.3531, "step": 6143 }, { "epoch": 0.31078962011229705, "grad_norm": 4.789051532745361, "learning_rate": 7.890211295595216e-06, "loss": 0.4099, "step": 6144 }, { "epoch": 0.3108402043603622, "grad_norm": 8.156082153320312, "learning_rate": 7.889556319660381e-06, "loss": 0.4402, "step": 6145 }, { "epoch": 0.31089078860842734, "grad_norm": 5.813032627105713, "learning_rate": 7.888901269268661e-06, "loss": 0.4317, "step": 6146 }, { "epoch": 0.3109413728564925, "grad_norm": 6.8845295906066895, "learning_rate": 7.888246144436932e-06, "loss": 0.432, "step": 6147 }, { "epoch": 0.3109919571045576, "grad_norm": 6.425919055938721, "learning_rate": 7.887590945182075e-06, "loss": 0.4432, "step": 6148 }, { "epoch": 0.31104254135262277, "grad_norm": 13.817578315734863, "learning_rate": 7.886935671520976e-06, "loss": 0.5619, "step": 6149 }, { "epoch": 0.31109312560068797, "grad_norm": 7.946162700653076, "learning_rate": 7.886280323470517e-06, "loss": 0.3929, "step": 6150 }, { "epoch": 0.3111437098487531, "grad_norm": 8.429594993591309, "learning_rate": 7.885624901047585e-06, "loss": 0.4301, "step": 6151 }, { "epoch": 0.31119429409681826, "grad_norm": 12.274910926818848, "learning_rate": 7.88496940426907e-06, "loss": 0.5399, "step": 6152 }, { "epoch": 0.3112448783448834, "grad_norm": 8.048140525817871, "learning_rate": 7.884313833151862e-06, "loss": 0.37, "step": 6153 }, { "epoch": 0.31129546259294855, "grad_norm": 7.5832014083862305, "learning_rate": 7.883658187712852e-06, "loss": 0.5353, "step": 6154 }, { "epoch": 0.3113460468410137, "grad_norm": 12.617145538330078, "learning_rate": 7.883002467968937e-06, "loss": 0.3933, "step": 6155 }, { "epoch": 0.31139663108907883, "grad_norm": 9.941563606262207, "learning_rate": 7.88234667393701e-06, "loss": 0.5325, "step": 6156 }, { "epoch": 0.31144721533714403, "grad_norm": 6.35325813293457, "learning_rate": 7.881690805633975e-06, "loss": 0.3921, "step": 6157 }, { "epoch": 0.3114977995852092, "grad_norm": 5.155187129974365, "learning_rate": 7.881034863076726e-06, "loss": 0.3546, "step": 6158 }, { "epoch": 0.3115483838332743, "grad_norm": 11.519293785095215, "learning_rate": 7.88037884628217e-06, "loss": 0.4899, "step": 6159 }, { "epoch": 0.31159896808133947, "grad_norm": 15.243768692016602, "learning_rate": 7.879722755267204e-06, "loss": 0.5371, "step": 6160 }, { "epoch": 0.3116495523294046, "grad_norm": 13.464319229125977, "learning_rate": 7.87906659004874e-06, "loss": 0.5292, "step": 6161 }, { "epoch": 0.31170013657746976, "grad_norm": 12.392582893371582, "learning_rate": 7.878410350643684e-06, "loss": 0.5925, "step": 6162 }, { "epoch": 0.31175072082553495, "grad_norm": 10.971121788024902, "learning_rate": 7.877754037068947e-06, "loss": 0.4304, "step": 6163 }, { "epoch": 0.3118013050736001, "grad_norm": 6.063545227050781, "learning_rate": 7.877097649341439e-06, "loss": 0.43, "step": 6164 }, { "epoch": 0.31185188932166524, "grad_norm": 8.89358139038086, "learning_rate": 7.876441187478071e-06, "loss": 0.4762, "step": 6165 }, { "epoch": 0.3119024735697304, "grad_norm": 6.428076267242432, "learning_rate": 7.875784651495765e-06, "loss": 0.4551, "step": 6166 }, { "epoch": 0.31195305781779553, "grad_norm": 9.152848243713379, "learning_rate": 7.875128041411436e-06, "loss": 0.4723, "step": 6167 }, { "epoch": 0.3120036420658607, "grad_norm": 6.866874694824219, "learning_rate": 7.874471357241998e-06, "loss": 0.4348, "step": 6168 }, { "epoch": 0.3120542263139258, "grad_norm": 8.739714622497559, "learning_rate": 7.873814599004377e-06, "loss": 0.4932, "step": 6169 }, { "epoch": 0.312104810561991, "grad_norm": 5.8505964279174805, "learning_rate": 7.873157766715496e-06, "loss": 0.3841, "step": 6170 }, { "epoch": 0.31215539481005616, "grad_norm": 10.221494674682617, "learning_rate": 7.87250086039228e-06, "loss": 0.4412, "step": 6171 }, { "epoch": 0.3122059790581213, "grad_norm": 8.326559066772461, "learning_rate": 7.871843880051654e-06, "loss": 0.4811, "step": 6172 }, { "epoch": 0.31225656330618645, "grad_norm": 8.54100513458252, "learning_rate": 7.871186825710548e-06, "loss": 0.4189, "step": 6173 }, { "epoch": 0.3123071475542516, "grad_norm": 15.392571449279785, "learning_rate": 7.870529697385893e-06, "loss": 0.5164, "step": 6174 }, { "epoch": 0.31235773180231674, "grad_norm": 9.8953857421875, "learning_rate": 7.869872495094621e-06, "loss": 0.4368, "step": 6175 }, { "epoch": 0.3124083160503819, "grad_norm": 7.3253092765808105, "learning_rate": 7.869215218853666e-06, "loss": 0.42, "step": 6176 }, { "epoch": 0.3124589002984471, "grad_norm": 8.389935493469238, "learning_rate": 7.868557868679965e-06, "loss": 0.5885, "step": 6177 }, { "epoch": 0.31250948454651223, "grad_norm": 9.189240455627441, "learning_rate": 7.867900444590457e-06, "loss": 0.5028, "step": 6178 }, { "epoch": 0.3125600687945774, "grad_norm": 5.605735778808594, "learning_rate": 7.867242946602083e-06, "loss": 0.4436, "step": 6179 }, { "epoch": 0.3126106530426425, "grad_norm": 9.8206148147583, "learning_rate": 7.866585374731781e-06, "loss": 0.431, "step": 6180 }, { "epoch": 0.31266123729070766, "grad_norm": 9.416422843933105, "learning_rate": 7.865927728996501e-06, "loss": 0.3003, "step": 6181 }, { "epoch": 0.3127118215387728, "grad_norm": 11.529187202453613, "learning_rate": 7.865270009413183e-06, "loss": 0.4722, "step": 6182 }, { "epoch": 0.312762405786838, "grad_norm": 7.250119209289551, "learning_rate": 7.864612215998779e-06, "loss": 0.4351, "step": 6183 }, { "epoch": 0.31281299003490315, "grad_norm": 6.203076362609863, "learning_rate": 7.863954348770238e-06, "loss": 0.4889, "step": 6184 }, { "epoch": 0.3128635742829683, "grad_norm": 5.599289417266846, "learning_rate": 7.863296407744509e-06, "loss": 0.4398, "step": 6185 }, { "epoch": 0.31291415853103344, "grad_norm": 7.633508205413818, "learning_rate": 7.86263839293855e-06, "loss": 0.4084, "step": 6186 }, { "epoch": 0.3129647427790986, "grad_norm": 9.333680152893066, "learning_rate": 7.861980304369311e-06, "loss": 0.6006, "step": 6187 }, { "epoch": 0.3130153270271637, "grad_norm": 9.46505069732666, "learning_rate": 7.861322142053755e-06, "loss": 0.5357, "step": 6188 }, { "epoch": 0.31306591127522887, "grad_norm": 5.3247761726379395, "learning_rate": 7.860663906008837e-06, "loss": 0.4821, "step": 6189 }, { "epoch": 0.31311649552329407, "grad_norm": 9.9971342086792, "learning_rate": 7.860005596251521e-06, "loss": 0.5067, "step": 6190 }, { "epoch": 0.3131670797713592, "grad_norm": 4.1945295333862305, "learning_rate": 7.859347212798769e-06, "loss": 0.406, "step": 6191 }, { "epoch": 0.31321766401942436, "grad_norm": 8.813994407653809, "learning_rate": 7.858688755667546e-06, "loss": 0.5002, "step": 6192 }, { "epoch": 0.3132682482674895, "grad_norm": 7.9427947998046875, "learning_rate": 7.858030224874817e-06, "loss": 0.391, "step": 6193 }, { "epoch": 0.31331883251555465, "grad_norm": 6.479122161865234, "learning_rate": 7.857371620437552e-06, "loss": 0.4246, "step": 6194 }, { "epoch": 0.3133694167636198, "grad_norm": 9.861743927001953, "learning_rate": 7.856712942372724e-06, "loss": 0.5768, "step": 6195 }, { "epoch": 0.31342000101168493, "grad_norm": 10.992057800292969, "learning_rate": 7.856054190697301e-06, "loss": 0.4651, "step": 6196 }, { "epoch": 0.31347058525975013, "grad_norm": 11.77402114868164, "learning_rate": 7.855395365428264e-06, "loss": 0.4299, "step": 6197 }, { "epoch": 0.3135211695078153, "grad_norm": 6.216953754425049, "learning_rate": 7.854736466582584e-06, "loss": 0.4673, "step": 6198 }, { "epoch": 0.3135717537558804, "grad_norm": 6.932822227478027, "learning_rate": 7.85407749417724e-06, "loss": 0.467, "step": 6199 }, { "epoch": 0.31362233800394557, "grad_norm": 7.654298305511475, "learning_rate": 7.853418448229211e-06, "loss": 0.5164, "step": 6200 }, { "epoch": 0.3136729222520107, "grad_norm": 6.778083324432373, "learning_rate": 7.852759328755484e-06, "loss": 0.4221, "step": 6201 }, { "epoch": 0.31372350650007585, "grad_norm": 6.409396171569824, "learning_rate": 7.852100135773038e-06, "loss": 0.4981, "step": 6202 }, { "epoch": 0.31377409074814105, "grad_norm": 7.3844380378723145, "learning_rate": 7.85144086929886e-06, "loss": 0.3496, "step": 6203 }, { "epoch": 0.3138246749962062, "grad_norm": 7.423678398132324, "learning_rate": 7.850781529349938e-06, "loss": 0.4302, "step": 6204 }, { "epoch": 0.31387525924427134, "grad_norm": 5.400659084320068, "learning_rate": 7.850122115943262e-06, "loss": 0.419, "step": 6205 }, { "epoch": 0.3139258434923365, "grad_norm": 4.182953357696533, "learning_rate": 7.849462629095825e-06, "loss": 0.4666, "step": 6206 }, { "epoch": 0.31397642774040163, "grad_norm": 13.788585662841797, "learning_rate": 7.848803068824617e-06, "loss": 0.4623, "step": 6207 }, { "epoch": 0.3140270119884668, "grad_norm": 14.131402015686035, "learning_rate": 7.848143435146635e-06, "loss": 0.4184, "step": 6208 }, { "epoch": 0.3140775962365319, "grad_norm": 12.828503608703613, "learning_rate": 7.847483728078875e-06, "loss": 0.4687, "step": 6209 }, { "epoch": 0.3141281804845971, "grad_norm": 6.422977924346924, "learning_rate": 7.846823947638339e-06, "loss": 0.4877, "step": 6210 }, { "epoch": 0.31417876473266226, "grad_norm": 6.673995018005371, "learning_rate": 7.846164093842026e-06, "loss": 0.4633, "step": 6211 }, { "epoch": 0.3142293489807274, "grad_norm": 8.236591339111328, "learning_rate": 7.845504166706939e-06, "loss": 0.4873, "step": 6212 }, { "epoch": 0.31427993322879255, "grad_norm": 7.499509811401367, "learning_rate": 7.844844166250082e-06, "loss": 0.3057, "step": 6213 }, { "epoch": 0.3143305174768577, "grad_norm": 13.038519859313965, "learning_rate": 7.844184092488462e-06, "loss": 0.5625, "step": 6214 }, { "epoch": 0.31438110172492284, "grad_norm": 10.64810848236084, "learning_rate": 7.843523945439087e-06, "loss": 0.5707, "step": 6215 }, { "epoch": 0.314431685972988, "grad_norm": 10.952207565307617, "learning_rate": 7.84286372511897e-06, "loss": 0.5197, "step": 6216 }, { "epoch": 0.3144822702210532, "grad_norm": 9.412259101867676, "learning_rate": 7.842203431545122e-06, "loss": 0.4836, "step": 6217 }, { "epoch": 0.31453285446911833, "grad_norm": 5.199155330657959, "learning_rate": 7.841543064734556e-06, "loss": 0.3896, "step": 6218 }, { "epoch": 0.3145834387171835, "grad_norm": 6.7365593910217285, "learning_rate": 7.840882624704287e-06, "loss": 0.5175, "step": 6219 }, { "epoch": 0.3146340229652486, "grad_norm": 5.305988311767578, "learning_rate": 7.840222111471337e-06, "loss": 0.5057, "step": 6220 }, { "epoch": 0.31468460721331376, "grad_norm": 9.265914916992188, "learning_rate": 7.83956152505272e-06, "loss": 0.4922, "step": 6221 }, { "epoch": 0.3147351914613789, "grad_norm": 12.850839614868164, "learning_rate": 7.838900865465464e-06, "loss": 0.5673, "step": 6222 }, { "epoch": 0.3147857757094441, "grad_norm": 4.517977714538574, "learning_rate": 7.838240132726588e-06, "loss": 0.4574, "step": 6223 }, { "epoch": 0.31483635995750925, "grad_norm": 7.904707908630371, "learning_rate": 7.837579326853121e-06, "loss": 0.4312, "step": 6224 }, { "epoch": 0.3148869442055744, "grad_norm": 4.994722843170166, "learning_rate": 7.83691844786209e-06, "loss": 0.4202, "step": 6225 }, { "epoch": 0.31493752845363954, "grad_norm": 5.523750305175781, "learning_rate": 7.83625749577052e-06, "loss": 0.3778, "step": 6226 }, { "epoch": 0.3149881127017047, "grad_norm": 5.667757511138916, "learning_rate": 7.835596470595446e-06, "loss": 0.4346, "step": 6227 }, { "epoch": 0.3150386969497698, "grad_norm": 6.7308526039123535, "learning_rate": 7.8349353723539e-06, "loss": 0.5047, "step": 6228 }, { "epoch": 0.31508928119783497, "grad_norm": 9.548501014709473, "learning_rate": 7.834274201062917e-06, "loss": 0.6139, "step": 6229 }, { "epoch": 0.31513986544590017, "grad_norm": 11.654341697692871, "learning_rate": 7.833612956739533e-06, "loss": 0.3809, "step": 6230 }, { "epoch": 0.3151904496939653, "grad_norm": 5.7027812004089355, "learning_rate": 7.83295163940079e-06, "loss": 0.3433, "step": 6231 }, { "epoch": 0.31524103394203046, "grad_norm": 10.346503257751465, "learning_rate": 7.832290249063723e-06, "loss": 0.5116, "step": 6232 }, { "epoch": 0.3152916181900956, "grad_norm": 5.917298793792725, "learning_rate": 7.83162878574538e-06, "loss": 0.4966, "step": 6233 }, { "epoch": 0.31534220243816075, "grad_norm": 8.620943069458008, "learning_rate": 7.8309672494628e-06, "loss": 0.4478, "step": 6234 }, { "epoch": 0.3153927866862259, "grad_norm": 9.53242301940918, "learning_rate": 7.830305640233034e-06, "loss": 0.4529, "step": 6235 }, { "epoch": 0.31544337093429103, "grad_norm": 7.36415433883667, "learning_rate": 7.829643958073127e-06, "loss": 0.4735, "step": 6236 }, { "epoch": 0.31549395518235623, "grad_norm": 11.927927017211914, "learning_rate": 7.828982203000131e-06, "loss": 0.5501, "step": 6237 }, { "epoch": 0.3155445394304214, "grad_norm": 8.288077354431152, "learning_rate": 7.828320375031095e-06, "loss": 0.4006, "step": 6238 }, { "epoch": 0.3155951236784865, "grad_norm": 10.880385398864746, "learning_rate": 7.827658474183076e-06, "loss": 0.4174, "step": 6239 }, { "epoch": 0.31564570792655167, "grad_norm": 11.771106719970703, "learning_rate": 7.826996500473126e-06, "loss": 0.6527, "step": 6240 }, { "epoch": 0.3156962921746168, "grad_norm": 10.176680564880371, "learning_rate": 7.826334453918307e-06, "loss": 0.4686, "step": 6241 }, { "epoch": 0.31574687642268195, "grad_norm": 8.484575271606445, "learning_rate": 7.825672334535673e-06, "loss": 0.2886, "step": 6242 }, { "epoch": 0.31579746067074715, "grad_norm": 19.461885452270508, "learning_rate": 7.82501014234229e-06, "loss": 0.5755, "step": 6243 }, { "epoch": 0.3158480449188123, "grad_norm": 14.30182933807373, "learning_rate": 7.82434787735522e-06, "loss": 0.4579, "step": 6244 }, { "epoch": 0.31589862916687744, "grad_norm": 9.482150077819824, "learning_rate": 7.823685539591526e-06, "loss": 0.5284, "step": 6245 }, { "epoch": 0.3159492134149426, "grad_norm": 7.895203590393066, "learning_rate": 7.823023129068276e-06, "loss": 0.4046, "step": 6246 }, { "epoch": 0.31599979766300773, "grad_norm": 10.494592666625977, "learning_rate": 7.822360645802538e-06, "loss": 0.5998, "step": 6247 }, { "epoch": 0.3160503819110729, "grad_norm": 5.809305667877197, "learning_rate": 7.821698089811384e-06, "loss": 0.3224, "step": 6248 }, { "epoch": 0.316100966159138, "grad_norm": 7.204652786254883, "learning_rate": 7.821035461111888e-06, "loss": 0.4334, "step": 6249 }, { "epoch": 0.3161515504072032, "grad_norm": 7.825764179229736, "learning_rate": 7.820372759721117e-06, "loss": 0.5052, "step": 6250 }, { "epoch": 0.31620213465526836, "grad_norm": 5.782710075378418, "learning_rate": 7.819709985656156e-06, "loss": 0.4116, "step": 6251 }, { "epoch": 0.3162527189033335, "grad_norm": 5.316450119018555, "learning_rate": 7.819047138934079e-06, "loss": 0.3816, "step": 6252 }, { "epoch": 0.31630330315139865, "grad_norm": 7.001629829406738, "learning_rate": 7.818384219571966e-06, "loss": 0.4269, "step": 6253 }, { "epoch": 0.3163538873994638, "grad_norm": 7.779061794281006, "learning_rate": 7.817721227586898e-06, "loss": 0.4192, "step": 6254 }, { "epoch": 0.31640447164752894, "grad_norm": 6.1284894943237305, "learning_rate": 7.817058162995961e-06, "loss": 0.4616, "step": 6255 }, { "epoch": 0.31645505589559414, "grad_norm": 4.769204139709473, "learning_rate": 7.816395025816241e-06, "loss": 0.3841, "step": 6256 }, { "epoch": 0.3165056401436593, "grad_norm": 9.662691116333008, "learning_rate": 7.815731816064822e-06, "loss": 0.4216, "step": 6257 }, { "epoch": 0.31655622439172443, "grad_norm": 5.038047790527344, "learning_rate": 7.815068533758796e-06, "loss": 0.4212, "step": 6258 }, { "epoch": 0.31660680863978957, "grad_norm": 6.656309127807617, "learning_rate": 7.814405178915254e-06, "loss": 0.4269, "step": 6259 }, { "epoch": 0.3166573928878547, "grad_norm": 7.00025749206543, "learning_rate": 7.813741751551287e-06, "loss": 0.4385, "step": 6260 }, { "epoch": 0.31670797713591986, "grad_norm": 6.432106971740723, "learning_rate": 7.813078251683992e-06, "loss": 0.444, "step": 6261 }, { "epoch": 0.316758561383985, "grad_norm": 10.659687995910645, "learning_rate": 7.812414679330464e-06, "loss": 0.5294, "step": 6262 }, { "epoch": 0.3168091456320502, "grad_norm": 5.162384033203125, "learning_rate": 7.811751034507803e-06, "loss": 0.4073, "step": 6263 }, { "epoch": 0.31685972988011535, "grad_norm": 9.9722900390625, "learning_rate": 7.811087317233112e-06, "loss": 0.5419, "step": 6264 }, { "epoch": 0.3169103141281805, "grad_norm": 6.122004985809326, "learning_rate": 7.810423527523489e-06, "loss": 0.4398, "step": 6265 }, { "epoch": 0.31696089837624564, "grad_norm": 7.964785099029541, "learning_rate": 7.80975966539604e-06, "loss": 0.3605, "step": 6266 }, { "epoch": 0.3170114826243108, "grad_norm": 12.021267890930176, "learning_rate": 7.80909573086787e-06, "loss": 0.51, "step": 6267 }, { "epoch": 0.3170620668723759, "grad_norm": 10.718018531799316, "learning_rate": 7.80843172395609e-06, "loss": 0.3904, "step": 6268 }, { "epoch": 0.31711265112044107, "grad_norm": 9.118997573852539, "learning_rate": 7.807767644677808e-06, "loss": 0.4467, "step": 6269 }, { "epoch": 0.31716323536850627, "grad_norm": 6.52795934677124, "learning_rate": 7.807103493050135e-06, "loss": 0.3738, "step": 6270 }, { "epoch": 0.3172138196165714, "grad_norm": 9.000229835510254, "learning_rate": 7.806439269090184e-06, "loss": 0.4339, "step": 6271 }, { "epoch": 0.31726440386463656, "grad_norm": 10.0491304397583, "learning_rate": 7.805774972815073e-06, "loss": 0.5194, "step": 6272 }, { "epoch": 0.3173149881127017, "grad_norm": 7.302181243896484, "learning_rate": 7.805110604241917e-06, "loss": 0.3991, "step": 6273 }, { "epoch": 0.31736557236076685, "grad_norm": 8.863304138183594, "learning_rate": 7.804446163387838e-06, "loss": 0.5594, "step": 6274 }, { "epoch": 0.317416156608832, "grad_norm": 11.172819137573242, "learning_rate": 7.803781650269954e-06, "loss": 0.6022, "step": 6275 }, { "epoch": 0.3174667408568972, "grad_norm": 8.523446083068848, "learning_rate": 7.80311706490539e-06, "loss": 0.5343, "step": 6276 }, { "epoch": 0.31751732510496233, "grad_norm": 8.779962539672852, "learning_rate": 7.802452407311269e-06, "loss": 0.4458, "step": 6277 }, { "epoch": 0.3175679093530275, "grad_norm": 12.406437873840332, "learning_rate": 7.801787677504717e-06, "loss": 0.5153, "step": 6278 }, { "epoch": 0.3176184936010926, "grad_norm": 6.948883533477783, "learning_rate": 7.801122875502867e-06, "loss": 0.3894, "step": 6279 }, { "epoch": 0.31766907784915777, "grad_norm": 11.530202865600586, "learning_rate": 7.800458001322846e-06, "loss": 0.4461, "step": 6280 }, { "epoch": 0.3177196620972229, "grad_norm": 8.99627685546875, "learning_rate": 7.799793054981786e-06, "loss": 0.4984, "step": 6281 }, { "epoch": 0.31777024634528805, "grad_norm": 5.382771015167236, "learning_rate": 7.79912803649682e-06, "loss": 0.4345, "step": 6282 }, { "epoch": 0.31782083059335325, "grad_norm": 8.834596633911133, "learning_rate": 7.798462945885086e-06, "loss": 0.4298, "step": 6283 }, { "epoch": 0.3178714148414184, "grad_norm": 7.7596917152404785, "learning_rate": 7.797797783163722e-06, "loss": 0.4322, "step": 6284 }, { "epoch": 0.31792199908948354, "grad_norm": 12.306995391845703, "learning_rate": 7.797132548349868e-06, "loss": 0.5619, "step": 6285 }, { "epoch": 0.3179725833375487, "grad_norm": 10.725567817687988, "learning_rate": 7.796467241460664e-06, "loss": 0.4407, "step": 6286 }, { "epoch": 0.31802316758561383, "grad_norm": 6.60920524597168, "learning_rate": 7.795801862513252e-06, "loss": 0.495, "step": 6287 }, { "epoch": 0.318073751833679, "grad_norm": 9.237217903137207, "learning_rate": 7.79513641152478e-06, "loss": 0.4619, "step": 6288 }, { "epoch": 0.3181243360817441, "grad_norm": 9.022578239440918, "learning_rate": 7.794470888512396e-06, "loss": 0.3744, "step": 6289 }, { "epoch": 0.3181749203298093, "grad_norm": 7.18382453918457, "learning_rate": 7.793805293493244e-06, "loss": 0.4538, "step": 6290 }, { "epoch": 0.31822550457787446, "grad_norm": 6.5045061111450195, "learning_rate": 7.793139626484479e-06, "loss": 0.449, "step": 6291 }, { "epoch": 0.3182760888259396, "grad_norm": 8.134251594543457, "learning_rate": 7.792473887503253e-06, "loss": 0.4256, "step": 6292 }, { "epoch": 0.31832667307400475, "grad_norm": 6.393719673156738, "learning_rate": 7.791808076566717e-06, "loss": 0.3919, "step": 6293 }, { "epoch": 0.3183772573220699, "grad_norm": 6.7530741691589355, "learning_rate": 7.791142193692034e-06, "loss": 0.4666, "step": 6294 }, { "epoch": 0.31842784157013504, "grad_norm": 12.916191101074219, "learning_rate": 7.790476238896358e-06, "loss": 0.5628, "step": 6295 }, { "epoch": 0.31847842581820024, "grad_norm": 11.13158893585205, "learning_rate": 7.789810212196846e-06, "loss": 0.5547, "step": 6296 }, { "epoch": 0.3185290100662654, "grad_norm": 15.334844589233398, "learning_rate": 7.789144113610667e-06, "loss": 0.5119, "step": 6297 }, { "epoch": 0.3185795943143305, "grad_norm": 12.064654350280762, "learning_rate": 7.78847794315498e-06, "loss": 0.5806, "step": 6298 }, { "epoch": 0.31863017856239567, "grad_norm": 8.68498706817627, "learning_rate": 7.78781170084695e-06, "loss": 0.2985, "step": 6299 }, { "epoch": 0.3186807628104608, "grad_norm": 6.034422874450684, "learning_rate": 7.787145386703748e-06, "loss": 0.4306, "step": 6300 }, { "epoch": 0.31873134705852596, "grad_norm": 6.660251617431641, "learning_rate": 7.786479000742541e-06, "loss": 0.4093, "step": 6301 }, { "epoch": 0.3187819313065911, "grad_norm": 7.392722129821777, "learning_rate": 7.785812542980502e-06, "loss": 0.4676, "step": 6302 }, { "epoch": 0.3188325155546563, "grad_norm": 17.667835235595703, "learning_rate": 7.7851460134348e-06, "loss": 0.6715, "step": 6303 }, { "epoch": 0.31888309980272145, "grad_norm": 5.094172954559326, "learning_rate": 7.784479412122614e-06, "loss": 0.4045, "step": 6304 }, { "epoch": 0.3189336840507866, "grad_norm": 4.728392601013184, "learning_rate": 7.783812739061118e-06, "loss": 0.3929, "step": 6305 }, { "epoch": 0.31898426829885174, "grad_norm": 6.1198344230651855, "learning_rate": 7.783145994267494e-06, "loss": 0.3447, "step": 6306 }, { "epoch": 0.3190348525469169, "grad_norm": 8.79847526550293, "learning_rate": 7.782479177758918e-06, "loss": 0.4953, "step": 6307 }, { "epoch": 0.319085436794982, "grad_norm": 10.568158149719238, "learning_rate": 7.781812289552575e-06, "loss": 0.4567, "step": 6308 }, { "epoch": 0.31913602104304717, "grad_norm": 7.070725440979004, "learning_rate": 7.781145329665649e-06, "loss": 0.4518, "step": 6309 }, { "epoch": 0.31918660529111237, "grad_norm": 9.502710342407227, "learning_rate": 7.780478298115325e-06, "loss": 0.4491, "step": 6310 }, { "epoch": 0.3192371895391775, "grad_norm": 12.267410278320312, "learning_rate": 7.779811194918791e-06, "loss": 0.5278, "step": 6311 }, { "epoch": 0.31928777378724266, "grad_norm": 9.651257514953613, "learning_rate": 7.779144020093236e-06, "loss": 0.3945, "step": 6312 }, { "epoch": 0.3193383580353078, "grad_norm": 9.671526908874512, "learning_rate": 7.778476773655855e-06, "loss": 0.52, "step": 6313 }, { "epoch": 0.31938894228337295, "grad_norm": 6.837371349334717, "learning_rate": 7.777809455623837e-06, "loss": 0.2727, "step": 6314 }, { "epoch": 0.3194395265314381, "grad_norm": 9.968255043029785, "learning_rate": 7.777142066014379e-06, "loss": 0.5286, "step": 6315 }, { "epoch": 0.3194901107795033, "grad_norm": 10.860815048217773, "learning_rate": 7.776474604844679e-06, "loss": 0.5513, "step": 6316 }, { "epoch": 0.31954069502756843, "grad_norm": 7.757415294647217, "learning_rate": 7.775807072131933e-06, "loss": 0.3762, "step": 6317 }, { "epoch": 0.3195912792756336, "grad_norm": 8.483531951904297, "learning_rate": 7.775139467893344e-06, "loss": 0.4581, "step": 6318 }, { "epoch": 0.3196418635236987, "grad_norm": 6.1545915603637695, "learning_rate": 7.774471792146114e-06, "loss": 0.3765, "step": 6319 }, { "epoch": 0.31969244777176387, "grad_norm": 6.925994873046875, "learning_rate": 7.773804044907448e-06, "loss": 0.4725, "step": 6320 }, { "epoch": 0.319743032019829, "grad_norm": 7.758349895477295, "learning_rate": 7.773136226194552e-06, "loss": 0.4249, "step": 6321 }, { "epoch": 0.31979361626789415, "grad_norm": 9.326862335205078, "learning_rate": 7.772468336024631e-06, "loss": 0.4972, "step": 6322 }, { "epoch": 0.31984420051595935, "grad_norm": 7.112741470336914, "learning_rate": 7.7718003744149e-06, "loss": 0.4404, "step": 6323 }, { "epoch": 0.3198947847640245, "grad_norm": 7.576022624969482, "learning_rate": 7.771132341382567e-06, "loss": 0.3394, "step": 6324 }, { "epoch": 0.31994536901208964, "grad_norm": 9.881499290466309, "learning_rate": 7.770464236944847e-06, "loss": 0.5178, "step": 6325 }, { "epoch": 0.3199959532601548, "grad_norm": 7.378927707672119, "learning_rate": 7.769796061118957e-06, "loss": 0.4466, "step": 6326 }, { "epoch": 0.32004653750821993, "grad_norm": 10.894770622253418, "learning_rate": 7.769127813922113e-06, "loss": 0.4495, "step": 6327 }, { "epoch": 0.3200971217562851, "grad_norm": 9.687901496887207, "learning_rate": 7.768459495371532e-06, "loss": 0.5606, "step": 6328 }, { "epoch": 0.3201477060043502, "grad_norm": 8.60284423828125, "learning_rate": 7.767791105484437e-06, "loss": 0.3829, "step": 6329 }, { "epoch": 0.3201982902524154, "grad_norm": 8.316123008728027, "learning_rate": 7.76712264427805e-06, "loss": 0.4098, "step": 6330 }, { "epoch": 0.32024887450048056, "grad_norm": 8.689545631408691, "learning_rate": 7.766454111769597e-06, "loss": 0.4485, "step": 6331 }, { "epoch": 0.3202994587485457, "grad_norm": 7.246978759765625, "learning_rate": 7.765785507976304e-06, "loss": 0.4284, "step": 6332 }, { "epoch": 0.32035004299661085, "grad_norm": 7.71929407119751, "learning_rate": 7.7651168329154e-06, "loss": 0.4804, "step": 6333 }, { "epoch": 0.320400627244676, "grad_norm": 6.694072246551514, "learning_rate": 7.76444808660411e-06, "loss": 0.3925, "step": 6334 }, { "epoch": 0.32045121149274114, "grad_norm": 8.752670288085938, "learning_rate": 7.763779269059675e-06, "loss": 0.3758, "step": 6335 }, { "epoch": 0.32050179574080634, "grad_norm": 11.756629943847656, "learning_rate": 7.76311038029932e-06, "loss": 0.6287, "step": 6336 }, { "epoch": 0.3205523799888715, "grad_norm": 9.543266296386719, "learning_rate": 7.762441420340286e-06, "loss": 0.6202, "step": 6337 }, { "epoch": 0.3206029642369366, "grad_norm": 6.8174614906311035, "learning_rate": 7.761772389199809e-06, "loss": 0.5387, "step": 6338 }, { "epoch": 0.32065354848500177, "grad_norm": 9.072153091430664, "learning_rate": 7.76110328689513e-06, "loss": 0.4331, "step": 6339 }, { "epoch": 0.3207041327330669, "grad_norm": 6.963715076446533, "learning_rate": 7.760434113443488e-06, "loss": 0.338, "step": 6340 }, { "epoch": 0.32075471698113206, "grad_norm": 8.977791786193848, "learning_rate": 7.759764868862123e-06, "loss": 0.4859, "step": 6341 }, { "epoch": 0.3208053012291972, "grad_norm": 6.577606201171875, "learning_rate": 7.759095553168286e-06, "loss": 0.4035, "step": 6342 }, { "epoch": 0.3208558854772624, "grad_norm": 6.571235656738281, "learning_rate": 7.75842616637922e-06, "loss": 0.429, "step": 6343 }, { "epoch": 0.32090646972532755, "grad_norm": 11.357522010803223, "learning_rate": 7.757756708512176e-06, "loss": 0.4879, "step": 6344 }, { "epoch": 0.3209570539733927, "grad_norm": 4.600313663482666, "learning_rate": 7.757087179584402e-06, "loss": 0.3104, "step": 6345 }, { "epoch": 0.32100763822145784, "grad_norm": 7.497420787811279, "learning_rate": 7.756417579613149e-06, "loss": 0.3976, "step": 6346 }, { "epoch": 0.321058222469523, "grad_norm": 6.996313095092773, "learning_rate": 7.755747908615674e-06, "loss": 0.3335, "step": 6347 }, { "epoch": 0.3211088067175881, "grad_norm": 5.710793495178223, "learning_rate": 7.755078166609234e-06, "loss": 0.441, "step": 6348 }, { "epoch": 0.32115939096565327, "grad_norm": 6.80225944519043, "learning_rate": 7.75440835361108e-06, "loss": 0.4824, "step": 6349 }, { "epoch": 0.32120997521371847, "grad_norm": 5.885562419891357, "learning_rate": 7.753738469638479e-06, "loss": 0.4557, "step": 6350 }, { "epoch": 0.3212605594617836, "grad_norm": 11.09011173248291, "learning_rate": 7.753068514708687e-06, "loss": 0.482, "step": 6351 }, { "epoch": 0.32131114370984876, "grad_norm": 15.751848220825195, "learning_rate": 7.752398488838972e-06, "loss": 0.5935, "step": 6352 }, { "epoch": 0.3213617279579139, "grad_norm": 7.340744495391846, "learning_rate": 7.751728392046593e-06, "loss": 0.436, "step": 6353 }, { "epoch": 0.32141231220597904, "grad_norm": 11.841148376464844, "learning_rate": 7.751058224348822e-06, "loss": 0.4128, "step": 6354 }, { "epoch": 0.3214628964540442, "grad_norm": 12.338224411010742, "learning_rate": 7.750387985762925e-06, "loss": 0.5609, "step": 6355 }, { "epoch": 0.3215134807021094, "grad_norm": 7.203478813171387, "learning_rate": 7.749717676306172e-06, "loss": 0.3413, "step": 6356 }, { "epoch": 0.32156406495017453, "grad_norm": 5.836124420166016, "learning_rate": 7.749047295995835e-06, "loss": 0.4113, "step": 6357 }, { "epoch": 0.3216146491982397, "grad_norm": 7.378907680511475, "learning_rate": 7.74837684484919e-06, "loss": 0.4694, "step": 6358 }, { "epoch": 0.3216652334463048, "grad_norm": 10.63175106048584, "learning_rate": 7.747706322883515e-06, "loss": 0.4392, "step": 6359 }, { "epoch": 0.32171581769436997, "grad_norm": 15.685336112976074, "learning_rate": 7.747035730116083e-06, "loss": 0.5438, "step": 6360 }, { "epoch": 0.3217664019424351, "grad_norm": 10.127689361572266, "learning_rate": 7.746365066564174e-06, "loss": 0.3587, "step": 6361 }, { "epoch": 0.32181698619050025, "grad_norm": 7.612424373626709, "learning_rate": 7.745694332245073e-06, "loss": 0.4039, "step": 6362 }, { "epoch": 0.32186757043856545, "grad_norm": 6.714548587799072, "learning_rate": 7.745023527176058e-06, "loss": 0.3841, "step": 6363 }, { "epoch": 0.3219181546866306, "grad_norm": 5.802762985229492, "learning_rate": 7.74435265137442e-06, "loss": 0.3262, "step": 6364 }, { "epoch": 0.32196873893469574, "grad_norm": 5.494406700134277, "learning_rate": 7.743681704857444e-06, "loss": 0.5265, "step": 6365 }, { "epoch": 0.3220193231827609, "grad_norm": 8.704086303710938, "learning_rate": 7.743010687642415e-06, "loss": 0.3521, "step": 6366 }, { "epoch": 0.32206990743082603, "grad_norm": 8.428496360778809, "learning_rate": 7.742339599746627e-06, "loss": 0.3873, "step": 6367 }, { "epoch": 0.3221204916788912, "grad_norm": 5.2553839683532715, "learning_rate": 7.74166844118737e-06, "loss": 0.3642, "step": 6368 }, { "epoch": 0.3221710759269563, "grad_norm": 8.233951568603516, "learning_rate": 7.74099721198194e-06, "loss": 0.4637, "step": 6369 }, { "epoch": 0.3222216601750215, "grad_norm": 13.875574111938477, "learning_rate": 7.740325912147632e-06, "loss": 0.5059, "step": 6370 }, { "epoch": 0.32227224442308666, "grad_norm": 7.794225215911865, "learning_rate": 7.739654541701746e-06, "loss": 0.5194, "step": 6371 }, { "epoch": 0.3223228286711518, "grad_norm": 5.623142719268799, "learning_rate": 7.73898310066158e-06, "loss": 0.3474, "step": 6372 }, { "epoch": 0.32237341291921695, "grad_norm": 9.795584678649902, "learning_rate": 7.738311589044433e-06, "loss": 0.5077, "step": 6373 }, { "epoch": 0.3224239971672821, "grad_norm": 8.125750541687012, "learning_rate": 7.737640006867612e-06, "loss": 0.4229, "step": 6374 }, { "epoch": 0.32247458141534724, "grad_norm": 6.947248935699463, "learning_rate": 7.73696835414842e-06, "loss": 0.464, "step": 6375 }, { "epoch": 0.32252516566341244, "grad_norm": 6.169379234313965, "learning_rate": 7.736296630904166e-06, "loss": 0.3988, "step": 6376 }, { "epoch": 0.3225757499114776, "grad_norm": 5.0757269859313965, "learning_rate": 7.735624837152154e-06, "loss": 0.4561, "step": 6377 }, { "epoch": 0.3226263341595427, "grad_norm": 6.240533828735352, "learning_rate": 7.734952972909699e-06, "loss": 0.3509, "step": 6378 }, { "epoch": 0.32267691840760787, "grad_norm": 7.603865623474121, "learning_rate": 7.734281038194114e-06, "loss": 0.4559, "step": 6379 }, { "epoch": 0.322727502655673, "grad_norm": 11.931055068969727, "learning_rate": 7.733609033022709e-06, "loss": 0.4652, "step": 6380 }, { "epoch": 0.32277808690373816, "grad_norm": 6.284057140350342, "learning_rate": 7.732936957412801e-06, "loss": 0.3971, "step": 6381 }, { "epoch": 0.3228286711518033, "grad_norm": 4.837466716766357, "learning_rate": 7.73226481138171e-06, "loss": 0.4186, "step": 6382 }, { "epoch": 0.3228792553998685, "grad_norm": 10.48304271697998, "learning_rate": 7.731592594946756e-06, "loss": 0.5111, "step": 6383 }, { "epoch": 0.32292983964793365, "grad_norm": 12.970213890075684, "learning_rate": 7.730920308125255e-06, "loss": 0.4404, "step": 6384 }, { "epoch": 0.3229804238959988, "grad_norm": 6.678482532501221, "learning_rate": 7.730247950934538e-06, "loss": 0.4306, "step": 6385 }, { "epoch": 0.32303100814406394, "grad_norm": 7.060308456420898, "learning_rate": 7.729575523391922e-06, "loss": 0.3936, "step": 6386 }, { "epoch": 0.3230815923921291, "grad_norm": 7.751643657684326, "learning_rate": 7.72890302551474e-06, "loss": 0.4811, "step": 6387 }, { "epoch": 0.3231321766401942, "grad_norm": 13.090523719787598, "learning_rate": 7.728230457320319e-06, "loss": 0.3556, "step": 6388 }, { "epoch": 0.32318276088825937, "grad_norm": 5.423998832702637, "learning_rate": 7.727557818825987e-06, "loss": 0.3919, "step": 6389 }, { "epoch": 0.32323334513632457, "grad_norm": 10.565674781799316, "learning_rate": 7.726885110049078e-06, "loss": 0.5369, "step": 6390 }, { "epoch": 0.3232839293843897, "grad_norm": 10.143373489379883, "learning_rate": 7.726212331006928e-06, "loss": 0.4402, "step": 6391 }, { "epoch": 0.32333451363245486, "grad_norm": 5.08350133895874, "learning_rate": 7.72553948171687e-06, "loss": 0.4098, "step": 6392 }, { "epoch": 0.32338509788052, "grad_norm": 7.097009658813477, "learning_rate": 7.724866562196243e-06, "loss": 0.43, "step": 6393 }, { "epoch": 0.32343568212858514, "grad_norm": 5.098017692565918, "learning_rate": 7.724193572462388e-06, "loss": 0.3205, "step": 6394 }, { "epoch": 0.3234862663766503, "grad_norm": 15.437145233154297, "learning_rate": 7.723520512532643e-06, "loss": 0.5419, "step": 6395 }, { "epoch": 0.3235368506247155, "grad_norm": 6.699018478393555, "learning_rate": 7.722847382424353e-06, "loss": 0.5065, "step": 6396 }, { "epoch": 0.32358743487278063, "grad_norm": 9.30225944519043, "learning_rate": 7.722174182154863e-06, "loss": 0.4404, "step": 6397 }, { "epoch": 0.3236380191208458, "grad_norm": 8.379687309265137, "learning_rate": 7.72150091174152e-06, "loss": 0.4929, "step": 6398 }, { "epoch": 0.3236886033689109, "grad_norm": 7.09626579284668, "learning_rate": 7.72082757120167e-06, "loss": 0.3689, "step": 6399 }, { "epoch": 0.32373918761697607, "grad_norm": 5.836361885070801, "learning_rate": 7.720154160552666e-06, "loss": 0.4891, "step": 6400 }, { "epoch": 0.3237897718650412, "grad_norm": 9.129384994506836, "learning_rate": 7.71948067981186e-06, "loss": 0.2986, "step": 6401 }, { "epoch": 0.32384035611310635, "grad_norm": 10.826351165771484, "learning_rate": 7.718807128996606e-06, "loss": 0.4484, "step": 6402 }, { "epoch": 0.32389094036117155, "grad_norm": 6.5613932609558105, "learning_rate": 7.718133508124259e-06, "loss": 0.4652, "step": 6403 }, { "epoch": 0.3239415246092367, "grad_norm": 7.222475528717041, "learning_rate": 7.717459817212178e-06, "loss": 0.5318, "step": 6404 }, { "epoch": 0.32399210885730184, "grad_norm": 5.572656631469727, "learning_rate": 7.71678605627772e-06, "loss": 0.4492, "step": 6405 }, { "epoch": 0.324042693105367, "grad_norm": 7.245270252227783, "learning_rate": 7.716112225338247e-06, "loss": 0.3934, "step": 6406 }, { "epoch": 0.32409327735343213, "grad_norm": 8.093957901000977, "learning_rate": 7.715438324411123e-06, "loss": 0.5195, "step": 6407 }, { "epoch": 0.3241438616014973, "grad_norm": 8.335916519165039, "learning_rate": 7.714764353513713e-06, "loss": 0.4317, "step": 6408 }, { "epoch": 0.3241944458495625, "grad_norm": 7.617021083831787, "learning_rate": 7.71409031266338e-06, "loss": 0.4465, "step": 6409 }, { "epoch": 0.3242450300976276, "grad_norm": 10.683697700500488, "learning_rate": 7.7134162018775e-06, "loss": 0.5089, "step": 6410 }, { "epoch": 0.32429561434569276, "grad_norm": 8.578983306884766, "learning_rate": 7.712742021173435e-06, "loss": 0.4389, "step": 6411 }, { "epoch": 0.3243461985937579, "grad_norm": 10.118396759033203, "learning_rate": 7.71206777056856e-06, "loss": 0.4318, "step": 6412 }, { "epoch": 0.32439678284182305, "grad_norm": 6.668187141418457, "learning_rate": 7.711393450080251e-06, "loss": 0.417, "step": 6413 }, { "epoch": 0.3244473670898882, "grad_norm": 7.586529731750488, "learning_rate": 7.710719059725881e-06, "loss": 0.4002, "step": 6414 }, { "epoch": 0.32449795133795334, "grad_norm": 8.276933670043945, "learning_rate": 7.71004459952283e-06, "loss": 0.4946, "step": 6415 }, { "epoch": 0.32454853558601854, "grad_norm": 7.030608177185059, "learning_rate": 7.709370069488475e-06, "loss": 0.4754, "step": 6416 }, { "epoch": 0.3245991198340837, "grad_norm": 13.002628326416016, "learning_rate": 7.708695469640197e-06, "loss": 0.4918, "step": 6417 }, { "epoch": 0.3246497040821488, "grad_norm": 17.289867401123047, "learning_rate": 7.70802079999538e-06, "loss": 0.5004, "step": 6418 }, { "epoch": 0.32470028833021397, "grad_norm": 6.981833457946777, "learning_rate": 7.707346060571406e-06, "loss": 0.3409, "step": 6419 }, { "epoch": 0.3247508725782791, "grad_norm": 12.359314918518066, "learning_rate": 7.706671251385665e-06, "loss": 0.4957, "step": 6420 }, { "epoch": 0.32480145682634426, "grad_norm": 6.872726917266846, "learning_rate": 7.705996372455541e-06, "loss": 0.4441, "step": 6421 }, { "epoch": 0.3248520410744094, "grad_norm": 12.622142791748047, "learning_rate": 7.70532142379843e-06, "loss": 0.4976, "step": 6422 }, { "epoch": 0.3249026253224746, "grad_norm": 7.1191582679748535, "learning_rate": 7.704646405431719e-06, "loss": 0.5035, "step": 6423 }, { "epoch": 0.32495320957053975, "grad_norm": 14.004661560058594, "learning_rate": 7.703971317372803e-06, "loss": 0.5426, "step": 6424 }, { "epoch": 0.3250037938186049, "grad_norm": 6.4063005447387695, "learning_rate": 7.703296159639078e-06, "loss": 0.4304, "step": 6425 }, { "epoch": 0.32505437806667004, "grad_norm": 6.144809722900391, "learning_rate": 7.702620932247939e-06, "loss": 0.362, "step": 6426 }, { "epoch": 0.3251049623147352, "grad_norm": 8.525860786437988, "learning_rate": 7.701945635216788e-06, "loss": 0.5114, "step": 6427 }, { "epoch": 0.3251555465628003, "grad_norm": 12.378823280334473, "learning_rate": 7.701270268563022e-06, "loss": 0.4253, "step": 6428 }, { "epoch": 0.3252061308108655, "grad_norm": 12.006501197814941, "learning_rate": 7.700594832304048e-06, "loss": 0.6372, "step": 6429 }, { "epoch": 0.32525671505893067, "grad_norm": 5.98776912689209, "learning_rate": 7.699919326457266e-06, "loss": 0.3541, "step": 6430 }, { "epoch": 0.3253072993069958, "grad_norm": 12.265064239501953, "learning_rate": 7.699243751040087e-06, "loss": 0.4549, "step": 6431 }, { "epoch": 0.32535788355506096, "grad_norm": 8.651217460632324, "learning_rate": 7.698568106069915e-06, "loss": 0.3872, "step": 6432 }, { "epoch": 0.3254084678031261, "grad_norm": 20.985557556152344, "learning_rate": 7.69789239156416e-06, "loss": 0.636, "step": 6433 }, { "epoch": 0.32545905205119124, "grad_norm": 11.990275382995605, "learning_rate": 7.697216607540236e-06, "loss": 0.3119, "step": 6434 }, { "epoch": 0.3255096362992564, "grad_norm": 16.0327091217041, "learning_rate": 7.696540754015556e-06, "loss": 0.6475, "step": 6435 }, { "epoch": 0.3255602205473216, "grad_norm": 8.6702241897583, "learning_rate": 7.695864831007533e-06, "loss": 0.5106, "step": 6436 }, { "epoch": 0.32561080479538673, "grad_norm": 5.390352725982666, "learning_rate": 7.695188838533582e-06, "loss": 0.3842, "step": 6437 }, { "epoch": 0.3256613890434519, "grad_norm": 8.848194122314453, "learning_rate": 7.69451277661113e-06, "loss": 0.5094, "step": 6438 }, { "epoch": 0.325711973291517, "grad_norm": 5.939291954040527, "learning_rate": 7.693836645257588e-06, "loss": 0.3613, "step": 6439 }, { "epoch": 0.32576255753958216, "grad_norm": 6.263391971588135, "learning_rate": 7.693160444490383e-06, "loss": 0.3925, "step": 6440 }, { "epoch": 0.3258131417876473, "grad_norm": 4.828439235687256, "learning_rate": 7.69248417432694e-06, "loss": 0.4894, "step": 6441 }, { "epoch": 0.32586372603571245, "grad_norm": 11.253973960876465, "learning_rate": 7.691807834784682e-06, "loss": 0.4604, "step": 6442 }, { "epoch": 0.32591431028377765, "grad_norm": 5.963430404663086, "learning_rate": 7.691131425881037e-06, "loss": 0.4675, "step": 6443 }, { "epoch": 0.3259648945318428, "grad_norm": 6.323659420013428, "learning_rate": 7.690454947633437e-06, "loss": 0.4974, "step": 6444 }, { "epoch": 0.32601547877990794, "grad_norm": 5.905526638031006, "learning_rate": 7.689778400059311e-06, "loss": 0.4904, "step": 6445 }, { "epoch": 0.3260660630279731, "grad_norm": 5.358658790588379, "learning_rate": 7.689101783176092e-06, "loss": 0.405, "step": 6446 }, { "epoch": 0.32611664727603823, "grad_norm": 7.088319778442383, "learning_rate": 7.688425097001214e-06, "loss": 0.4751, "step": 6447 }, { "epoch": 0.3261672315241034, "grad_norm": 8.880663871765137, "learning_rate": 7.687748341552118e-06, "loss": 0.4992, "step": 6448 }, { "epoch": 0.3262178157721686, "grad_norm": 9.068286895751953, "learning_rate": 7.687071516846239e-06, "loss": 0.5426, "step": 6449 }, { "epoch": 0.3262684000202337, "grad_norm": 9.490370750427246, "learning_rate": 7.686394622901016e-06, "loss": 0.325, "step": 6450 }, { "epoch": 0.32631898426829886, "grad_norm": 5.4406890869140625, "learning_rate": 7.685717659733893e-06, "loss": 0.432, "step": 6451 }, { "epoch": 0.326369568516364, "grad_norm": 9.323899269104004, "learning_rate": 7.685040627362311e-06, "loss": 0.4968, "step": 6452 }, { "epoch": 0.32642015276442915, "grad_norm": 7.054498195648193, "learning_rate": 7.68436352580372e-06, "loss": 0.4836, "step": 6453 }, { "epoch": 0.3264707370124943, "grad_norm": 7.448715686798096, "learning_rate": 7.683686355075563e-06, "loss": 0.3704, "step": 6454 }, { "epoch": 0.32652132126055944, "grad_norm": 8.77640438079834, "learning_rate": 7.683009115195292e-06, "loss": 0.4616, "step": 6455 }, { "epoch": 0.32657190550862464, "grad_norm": 7.069424629211426, "learning_rate": 7.682331806180356e-06, "loss": 0.424, "step": 6456 }, { "epoch": 0.3266224897566898, "grad_norm": 9.502291679382324, "learning_rate": 7.681654428048209e-06, "loss": 0.3759, "step": 6457 }, { "epoch": 0.3266730740047549, "grad_norm": 9.046375274658203, "learning_rate": 7.680976980816303e-06, "loss": 0.4835, "step": 6458 }, { "epoch": 0.32672365825282007, "grad_norm": 7.898292064666748, "learning_rate": 7.680299464502096e-06, "loss": 0.4451, "step": 6459 }, { "epoch": 0.3267742425008852, "grad_norm": 8.883004188537598, "learning_rate": 7.679621879123049e-06, "loss": 0.3963, "step": 6460 }, { "epoch": 0.32682482674895036, "grad_norm": 9.94593620300293, "learning_rate": 7.678944224696615e-06, "loss": 0.4637, "step": 6461 }, { "epoch": 0.3268754109970155, "grad_norm": 10.218611717224121, "learning_rate": 7.678266501240262e-06, "loss": 0.4099, "step": 6462 }, { "epoch": 0.3269259952450807, "grad_norm": 7.342185020446777, "learning_rate": 7.677588708771449e-06, "loss": 0.3975, "step": 6463 }, { "epoch": 0.32697657949314585, "grad_norm": 11.752289772033691, "learning_rate": 7.676910847307642e-06, "loss": 0.5135, "step": 6464 }, { "epoch": 0.327027163741211, "grad_norm": 8.183272361755371, "learning_rate": 7.676232916866306e-06, "loss": 0.3855, "step": 6465 }, { "epoch": 0.32707774798927614, "grad_norm": 9.366363525390625, "learning_rate": 7.675554917464917e-06, "loss": 0.4466, "step": 6466 }, { "epoch": 0.3271283322373413, "grad_norm": 9.035945892333984, "learning_rate": 7.674876849120935e-06, "loss": 0.4893, "step": 6467 }, { "epoch": 0.3271789164854064, "grad_norm": 6.931246280670166, "learning_rate": 7.674198711851842e-06, "loss": 0.4435, "step": 6468 }, { "epoch": 0.3272295007334716, "grad_norm": 9.24480152130127, "learning_rate": 7.673520505675105e-06, "loss": 0.3991, "step": 6469 }, { "epoch": 0.32728008498153677, "grad_norm": 8.863991737365723, "learning_rate": 7.672842230608203e-06, "loss": 0.3895, "step": 6470 }, { "epoch": 0.3273306692296019, "grad_norm": 9.184028625488281, "learning_rate": 7.67216388666861e-06, "loss": 0.5344, "step": 6471 }, { "epoch": 0.32738125347766706, "grad_norm": 6.558706760406494, "learning_rate": 7.671485473873812e-06, "loss": 0.3879, "step": 6472 }, { "epoch": 0.3274318377257322, "grad_norm": 10.410809516906738, "learning_rate": 7.670806992241284e-06, "loss": 0.5329, "step": 6473 }, { "epoch": 0.32748242197379734, "grad_norm": 8.659005165100098, "learning_rate": 7.670128441788512e-06, "loss": 0.3848, "step": 6474 }, { "epoch": 0.3275330062218625, "grad_norm": 8.229435920715332, "learning_rate": 7.669449822532979e-06, "loss": 0.3367, "step": 6475 }, { "epoch": 0.3275835904699277, "grad_norm": 14.629777908325195, "learning_rate": 7.668771134492172e-06, "loss": 0.597, "step": 6476 }, { "epoch": 0.32763417471799283, "grad_norm": 15.765360832214355, "learning_rate": 7.668092377683579e-06, "loss": 0.549, "step": 6477 }, { "epoch": 0.327684758966058, "grad_norm": 7.5312933921813965, "learning_rate": 7.667413552124689e-06, "loss": 0.4537, "step": 6478 }, { "epoch": 0.3277353432141231, "grad_norm": 5.080340385437012, "learning_rate": 7.666734657832996e-06, "loss": 0.383, "step": 6479 }, { "epoch": 0.32778592746218826, "grad_norm": 10.229613304138184, "learning_rate": 7.666055694825992e-06, "loss": 0.446, "step": 6480 }, { "epoch": 0.3278365117102534, "grad_norm": 8.91726016998291, "learning_rate": 7.665376663121172e-06, "loss": 0.3823, "step": 6481 }, { "epoch": 0.32788709595831855, "grad_norm": 14.977276802062988, "learning_rate": 7.664697562736032e-06, "loss": 0.4417, "step": 6482 }, { "epoch": 0.32793768020638375, "grad_norm": 5.75716495513916, "learning_rate": 7.664018393688074e-06, "loss": 0.4279, "step": 6483 }, { "epoch": 0.3279882644544489, "grad_norm": 17.955453872680664, "learning_rate": 7.663339155994795e-06, "loss": 0.4806, "step": 6484 }, { "epoch": 0.32803884870251404, "grad_norm": 5.655694484710693, "learning_rate": 7.6626598496737e-06, "loss": 0.4249, "step": 6485 }, { "epoch": 0.3280894329505792, "grad_norm": 14.561700820922852, "learning_rate": 7.661980474742293e-06, "loss": 0.5, "step": 6486 }, { "epoch": 0.32814001719864433, "grad_norm": 6.792402267456055, "learning_rate": 7.661301031218077e-06, "loss": 0.3726, "step": 6487 }, { "epoch": 0.3281906014467095, "grad_norm": 6.109333038330078, "learning_rate": 7.660621519118563e-06, "loss": 0.3736, "step": 6488 }, { "epoch": 0.3282411856947747, "grad_norm": 16.773115158081055, "learning_rate": 7.659941938461256e-06, "loss": 0.5131, "step": 6489 }, { "epoch": 0.3282917699428398, "grad_norm": 9.346324920654297, "learning_rate": 7.659262289263673e-06, "loss": 0.4102, "step": 6490 }, { "epoch": 0.32834235419090496, "grad_norm": 8.070551872253418, "learning_rate": 7.658582571543325e-06, "loss": 0.3654, "step": 6491 }, { "epoch": 0.3283929384389701, "grad_norm": 9.789061546325684, "learning_rate": 7.657902785317724e-06, "loss": 0.3292, "step": 6492 }, { "epoch": 0.32844352268703525, "grad_norm": 6.7033820152282715, "learning_rate": 7.657222930604386e-06, "loss": 0.4773, "step": 6493 }, { "epoch": 0.3284941069351004, "grad_norm": 5.5280609130859375, "learning_rate": 7.656543007420835e-06, "loss": 0.3684, "step": 6494 }, { "epoch": 0.32854469118316554, "grad_norm": 10.261480331420898, "learning_rate": 7.655863015784585e-06, "loss": 0.4718, "step": 6495 }, { "epoch": 0.32859527543123074, "grad_norm": 8.4717435836792, "learning_rate": 7.655182955713161e-06, "loss": 0.3975, "step": 6496 }, { "epoch": 0.3286458596792959, "grad_norm": 10.103107452392578, "learning_rate": 7.654502827224085e-06, "loss": 0.4288, "step": 6497 }, { "epoch": 0.328696443927361, "grad_norm": 9.832474708557129, "learning_rate": 7.653822630334883e-06, "loss": 0.4158, "step": 6498 }, { "epoch": 0.32874702817542617, "grad_norm": 5.438282489776611, "learning_rate": 7.653142365063084e-06, "loss": 0.3365, "step": 6499 }, { "epoch": 0.3287976124234913, "grad_norm": 5.242053031921387, "learning_rate": 7.652462031426213e-06, "loss": 0.2887, "step": 6500 }, { "epoch": 0.32884819667155646, "grad_norm": 7.375084400177002, "learning_rate": 7.651781629441801e-06, "loss": 0.3493, "step": 6501 }, { "epoch": 0.3288987809196216, "grad_norm": 7.655946254730225, "learning_rate": 7.651101159127384e-06, "loss": 0.4608, "step": 6502 }, { "epoch": 0.3289493651676868, "grad_norm": 8.704054832458496, "learning_rate": 7.650420620500491e-06, "loss": 0.4912, "step": 6503 }, { "epoch": 0.32899994941575195, "grad_norm": 7.986734867095947, "learning_rate": 7.649740013578662e-06, "loss": 0.4775, "step": 6504 }, { "epoch": 0.3290505336638171, "grad_norm": 15.714234352111816, "learning_rate": 7.649059338379432e-06, "loss": 0.5666, "step": 6505 }, { "epoch": 0.32910111791188223, "grad_norm": 8.552213668823242, "learning_rate": 7.648378594920342e-06, "loss": 0.4743, "step": 6506 }, { "epoch": 0.3291517021599474, "grad_norm": 17.406410217285156, "learning_rate": 7.647697783218933e-06, "loss": 0.5431, "step": 6507 }, { "epoch": 0.3292022864080125, "grad_norm": 20.61551284790039, "learning_rate": 7.647016903292747e-06, "loss": 0.5481, "step": 6508 }, { "epoch": 0.3292528706560777, "grad_norm": 23.580881118774414, "learning_rate": 7.646335955159328e-06, "loss": 0.4913, "step": 6509 }, { "epoch": 0.32930345490414287, "grad_norm": 6.997948169708252, "learning_rate": 7.645654938836224e-06, "loss": 0.4543, "step": 6510 }, { "epoch": 0.329354039152208, "grad_norm": 9.710107803344727, "learning_rate": 7.644973854340983e-06, "loss": 0.4259, "step": 6511 }, { "epoch": 0.32940462340027316, "grad_norm": 8.369824409484863, "learning_rate": 7.644292701691154e-06, "loss": 0.381, "step": 6512 }, { "epoch": 0.3294552076483383, "grad_norm": 10.428221702575684, "learning_rate": 7.64361148090429e-06, "loss": 0.3536, "step": 6513 }, { "epoch": 0.32950579189640344, "grad_norm": 9.475157737731934, "learning_rate": 7.642930191997944e-06, "loss": 0.4689, "step": 6514 }, { "epoch": 0.3295563761444686, "grad_norm": 12.403278350830078, "learning_rate": 7.64224883498967e-06, "loss": 0.4624, "step": 6515 }, { "epoch": 0.3296069603925338, "grad_norm": 7.122254371643066, "learning_rate": 7.641567409897024e-06, "loss": 0.432, "step": 6516 }, { "epoch": 0.32965754464059893, "grad_norm": 7.662106990814209, "learning_rate": 7.640885916737568e-06, "loss": 0.446, "step": 6517 }, { "epoch": 0.3297081288886641, "grad_norm": 14.698575019836426, "learning_rate": 7.64020435552886e-06, "loss": 0.5274, "step": 6518 }, { "epoch": 0.3297587131367292, "grad_norm": 6.3107147216796875, "learning_rate": 7.639522726288464e-06, "loss": 0.3386, "step": 6519 }, { "epoch": 0.32980929738479436, "grad_norm": 6.92738151550293, "learning_rate": 7.63884102903394e-06, "loss": 0.4772, "step": 6520 }, { "epoch": 0.3298598816328595, "grad_norm": 15.046688079833984, "learning_rate": 7.63815926378286e-06, "loss": 0.4384, "step": 6521 }, { "epoch": 0.32991046588092465, "grad_norm": 5.035706043243408, "learning_rate": 7.637477430552786e-06, "loss": 0.4262, "step": 6522 }, { "epoch": 0.32996105012898985, "grad_norm": 6.283407688140869, "learning_rate": 7.63679552936129e-06, "loss": 0.4003, "step": 6523 }, { "epoch": 0.330011634377055, "grad_norm": 5.05947732925415, "learning_rate": 7.636113560225941e-06, "loss": 0.378, "step": 6524 }, { "epoch": 0.33006221862512014, "grad_norm": 9.035636901855469, "learning_rate": 7.635431523164314e-06, "loss": 0.4263, "step": 6525 }, { "epoch": 0.3301128028731853, "grad_norm": 5.034605026245117, "learning_rate": 7.634749418193984e-06, "loss": 0.4108, "step": 6526 }, { "epoch": 0.33016338712125043, "grad_norm": 6.709367752075195, "learning_rate": 7.634067245332522e-06, "loss": 0.3474, "step": 6527 }, { "epoch": 0.3302139713693156, "grad_norm": 8.451828956604004, "learning_rate": 7.63338500459751e-06, "loss": 0.2733, "step": 6528 }, { "epoch": 0.3302645556173808, "grad_norm": 8.172794342041016, "learning_rate": 7.632702696006528e-06, "loss": 0.3586, "step": 6529 }, { "epoch": 0.3303151398654459, "grad_norm": 9.055837631225586, "learning_rate": 7.632020319577156e-06, "loss": 0.3872, "step": 6530 }, { "epoch": 0.33036572411351106, "grad_norm": 6.516030788421631, "learning_rate": 7.631337875326979e-06, "loss": 0.4343, "step": 6531 }, { "epoch": 0.3304163083615762, "grad_norm": 11.2788724899292, "learning_rate": 7.630655363273581e-06, "loss": 0.4435, "step": 6532 }, { "epoch": 0.33046689260964135, "grad_norm": 8.734397888183594, "learning_rate": 7.629972783434546e-06, "loss": 0.4795, "step": 6533 }, { "epoch": 0.3305174768577065, "grad_norm": 11.507547378540039, "learning_rate": 7.629290135827466e-06, "loss": 0.4531, "step": 6534 }, { "epoch": 0.33056806110577164, "grad_norm": 8.231717109680176, "learning_rate": 7.628607420469929e-06, "loss": 0.4493, "step": 6535 }, { "epoch": 0.33061864535383684, "grad_norm": 5.478990077972412, "learning_rate": 7.62792463737953e-06, "loss": 0.3646, "step": 6536 }, { "epoch": 0.330669229601902, "grad_norm": 5.127054214477539, "learning_rate": 7.627241786573859e-06, "loss": 0.3484, "step": 6537 }, { "epoch": 0.3307198138499671, "grad_norm": 6.865869522094727, "learning_rate": 7.626558868070515e-06, "loss": 0.5071, "step": 6538 }, { "epoch": 0.33077039809803227, "grad_norm": 8.689765930175781, "learning_rate": 7.625875881887092e-06, "loss": 0.4024, "step": 6539 }, { "epoch": 0.3308209823460974, "grad_norm": 9.819849014282227, "learning_rate": 7.625192828041189e-06, "loss": 0.4868, "step": 6540 }, { "epoch": 0.33087156659416256, "grad_norm": 8.401392936706543, "learning_rate": 7.624509706550408e-06, "loss": 0.4017, "step": 6541 }, { "epoch": 0.33092215084222776, "grad_norm": 16.877334594726562, "learning_rate": 7.623826517432352e-06, "loss": 0.4483, "step": 6542 }, { "epoch": 0.3309727350902929, "grad_norm": 7.160208702087402, "learning_rate": 7.623143260704624e-06, "loss": 0.4435, "step": 6543 }, { "epoch": 0.33102331933835805, "grad_norm": 7.2952070236206055, "learning_rate": 7.62245993638483e-06, "loss": 0.3598, "step": 6544 }, { "epoch": 0.3310739035864232, "grad_norm": 9.304813385009766, "learning_rate": 7.621776544490578e-06, "loss": 0.4784, "step": 6545 }, { "epoch": 0.33112448783448833, "grad_norm": 22.779476165771484, "learning_rate": 7.621093085039477e-06, "loss": 0.5632, "step": 6546 }, { "epoch": 0.3311750720825535, "grad_norm": 7.7709760665893555, "learning_rate": 7.620409558049138e-06, "loss": 0.3482, "step": 6547 }, { "epoch": 0.3312256563306186, "grad_norm": 10.813713073730469, "learning_rate": 7.619725963537174e-06, "loss": 0.3007, "step": 6548 }, { "epoch": 0.3312762405786838, "grad_norm": 6.112451076507568, "learning_rate": 7.619042301521199e-06, "loss": 0.4193, "step": 6549 }, { "epoch": 0.33132682482674897, "grad_norm": 7.978157997131348, "learning_rate": 7.61835857201883e-06, "loss": 0.3509, "step": 6550 }, { "epoch": 0.3313774090748141, "grad_norm": 7.609199047088623, "learning_rate": 7.617674775047685e-06, "loss": 0.4282, "step": 6551 }, { "epoch": 0.33142799332287926, "grad_norm": 8.664648056030273, "learning_rate": 7.616990910625386e-06, "loss": 0.4648, "step": 6552 }, { "epoch": 0.3314785775709444, "grad_norm": 7.311094284057617, "learning_rate": 7.61630697876955e-06, "loss": 0.4182, "step": 6553 }, { "epoch": 0.33152916181900954, "grad_norm": 17.541120529174805, "learning_rate": 7.615622979497803e-06, "loss": 0.5205, "step": 6554 }, { "epoch": 0.3315797460670747, "grad_norm": 9.722586631774902, "learning_rate": 7.614938912827769e-06, "loss": 0.4454, "step": 6555 }, { "epoch": 0.3316303303151399, "grad_norm": 12.53329086303711, "learning_rate": 7.614254778777077e-06, "loss": 0.596, "step": 6556 }, { "epoch": 0.33168091456320503, "grad_norm": 7.009939193725586, "learning_rate": 7.613570577363354e-06, "loss": 0.378, "step": 6557 }, { "epoch": 0.3317314988112702, "grad_norm": 6.002956867218018, "learning_rate": 7.6128863086042296e-06, "loss": 0.455, "step": 6558 }, { "epoch": 0.3317820830593353, "grad_norm": 6.317930221557617, "learning_rate": 7.612201972517334e-06, "loss": 0.3711, "step": 6559 }, { "epoch": 0.33183266730740046, "grad_norm": 5.93601655960083, "learning_rate": 7.611517569120306e-06, "loss": 0.3488, "step": 6560 }, { "epoch": 0.3318832515554656, "grad_norm": 5.990686416625977, "learning_rate": 7.610833098430777e-06, "loss": 0.4158, "step": 6561 }, { "epoch": 0.3319338358035308, "grad_norm": 9.732306480407715, "learning_rate": 7.610148560466384e-06, "loss": 0.494, "step": 6562 }, { "epoch": 0.33198442005159595, "grad_norm": 14.351360321044922, "learning_rate": 7.609463955244768e-06, "loss": 0.4946, "step": 6563 }, { "epoch": 0.3320350042996611, "grad_norm": 6.836007595062256, "learning_rate": 7.60877928278357e-06, "loss": 0.2711, "step": 6564 }, { "epoch": 0.33208558854772624, "grad_norm": 6.541057586669922, "learning_rate": 7.6080945431004304e-06, "loss": 0.4015, "step": 6565 }, { "epoch": 0.3321361727957914, "grad_norm": 16.074159622192383, "learning_rate": 7.607409736212993e-06, "loss": 0.5754, "step": 6566 }, { "epoch": 0.33218675704385653, "grad_norm": 7.642574310302734, "learning_rate": 7.606724862138906e-06, "loss": 0.38, "step": 6567 }, { "epoch": 0.3322373412919217, "grad_norm": 10.759511947631836, "learning_rate": 7.606039920895813e-06, "loss": 0.546, "step": 6568 }, { "epoch": 0.3322879255399869, "grad_norm": 11.413351058959961, "learning_rate": 7.605354912501368e-06, "loss": 0.4391, "step": 6569 }, { "epoch": 0.332338509788052, "grad_norm": 6.965322494506836, "learning_rate": 7.604669836973221e-06, "loss": 0.5108, "step": 6570 }, { "epoch": 0.33238909403611716, "grad_norm": 6.287825584411621, "learning_rate": 7.603984694329023e-06, "loss": 0.4677, "step": 6571 }, { "epoch": 0.3324396782841823, "grad_norm": 8.944384574890137, "learning_rate": 7.603299484586428e-06, "loss": 0.5512, "step": 6572 }, { "epoch": 0.33249026253224745, "grad_norm": 9.96212100982666, "learning_rate": 7.602614207763095e-06, "loss": 0.3631, "step": 6573 }, { "epoch": 0.3325408467803126, "grad_norm": 6.962059020996094, "learning_rate": 7.601928863876678e-06, "loss": 0.4224, "step": 6574 }, { "epoch": 0.33259143102837774, "grad_norm": 7.946589469909668, "learning_rate": 7.60124345294484e-06, "loss": 0.4678, "step": 6575 }, { "epoch": 0.33264201527644294, "grad_norm": 7.43447208404541, "learning_rate": 7.600557974985241e-06, "loss": 0.4743, "step": 6576 }, { "epoch": 0.3326925995245081, "grad_norm": 5.515531063079834, "learning_rate": 7.599872430015545e-06, "loss": 0.4266, "step": 6577 }, { "epoch": 0.3327431837725732, "grad_norm": 7.051076889038086, "learning_rate": 7.599186818053414e-06, "loss": 0.4899, "step": 6578 }, { "epoch": 0.33279376802063837, "grad_norm": 7.050043106079102, "learning_rate": 7.598501139116519e-06, "loss": 0.4275, "step": 6579 }, { "epoch": 0.3328443522687035, "grad_norm": 4.281708717346191, "learning_rate": 7.597815393222526e-06, "loss": 0.4472, "step": 6580 }, { "epoch": 0.33289493651676866, "grad_norm": 14.610193252563477, "learning_rate": 7.597129580389105e-06, "loss": 0.7246, "step": 6581 }, { "epoch": 0.33294552076483386, "grad_norm": 8.894437789916992, "learning_rate": 7.596443700633927e-06, "loss": 0.3764, "step": 6582 }, { "epoch": 0.332996105012899, "grad_norm": 10.812932968139648, "learning_rate": 7.595757753974666e-06, "loss": 0.51, "step": 6583 }, { "epoch": 0.33304668926096415, "grad_norm": 5.019397258758545, "learning_rate": 7.595071740428998e-06, "loss": 0.3881, "step": 6584 }, { "epoch": 0.3330972735090293, "grad_norm": 7.683717727661133, "learning_rate": 7.5943856600146e-06, "loss": 0.4297, "step": 6585 }, { "epoch": 0.33314785775709443, "grad_norm": 5.205839157104492, "learning_rate": 7.593699512749151e-06, "loss": 0.3656, "step": 6586 }, { "epoch": 0.3331984420051596, "grad_norm": 22.49258041381836, "learning_rate": 7.593013298650328e-06, "loss": 0.4384, "step": 6587 }, { "epoch": 0.3332490262532247, "grad_norm": 7.748059272766113, "learning_rate": 7.592327017735817e-06, "loss": 0.5695, "step": 6588 }, { "epoch": 0.3332996105012899, "grad_norm": 9.467133522033691, "learning_rate": 7.591640670023299e-06, "loss": 0.5489, "step": 6589 }, { "epoch": 0.33335019474935507, "grad_norm": 12.561470985412598, "learning_rate": 7.5909542555304624e-06, "loss": 0.4451, "step": 6590 }, { "epoch": 0.3334007789974202, "grad_norm": 10.5082426071167, "learning_rate": 7.590267774274991e-06, "loss": 0.5186, "step": 6591 }, { "epoch": 0.33345136324548535, "grad_norm": 23.93415069580078, "learning_rate": 7.589581226274578e-06, "loss": 0.4602, "step": 6592 }, { "epoch": 0.3335019474935505, "grad_norm": 7.581545352935791, "learning_rate": 7.588894611546908e-06, "loss": 0.4858, "step": 6593 }, { "epoch": 0.33355253174161564, "grad_norm": 5.702849864959717, "learning_rate": 7.588207930109681e-06, "loss": 0.3572, "step": 6594 }, { "epoch": 0.3336031159896808, "grad_norm": 7.1043548583984375, "learning_rate": 7.587521181980586e-06, "loss": 0.4054, "step": 6595 }, { "epoch": 0.333653700237746, "grad_norm": 9.093122482299805, "learning_rate": 7.5868343671773185e-06, "loss": 0.5048, "step": 6596 }, { "epoch": 0.33370428448581113, "grad_norm": 17.16496467590332, "learning_rate": 7.586147485717579e-06, "loss": 0.4175, "step": 6597 }, { "epoch": 0.3337548687338763, "grad_norm": 10.49229907989502, "learning_rate": 7.585460537619065e-06, "loss": 0.5365, "step": 6598 }, { "epoch": 0.3338054529819414, "grad_norm": 9.2655611038208, "learning_rate": 7.584773522899478e-06, "loss": 0.4532, "step": 6599 }, { "epoch": 0.33385603723000656, "grad_norm": 4.350186347961426, "learning_rate": 7.5840864415765205e-06, "loss": 0.4835, "step": 6600 }, { "epoch": 0.3339066214780717, "grad_norm": 7.40079927444458, "learning_rate": 7.5833992936678965e-06, "loss": 0.4279, "step": 6601 }, { "epoch": 0.3339572057261369, "grad_norm": 3.723832130432129, "learning_rate": 7.582712079191313e-06, "loss": 0.4225, "step": 6602 }, { "epoch": 0.33400778997420205, "grad_norm": 6.365222930908203, "learning_rate": 7.582024798164478e-06, "loss": 0.5023, "step": 6603 }, { "epoch": 0.3340583742222672, "grad_norm": 12.126845359802246, "learning_rate": 7.5813374506051e-06, "loss": 0.5276, "step": 6604 }, { "epoch": 0.33410895847033234, "grad_norm": 7.83262825012207, "learning_rate": 7.580650036530892e-06, "loss": 0.4402, "step": 6605 }, { "epoch": 0.3341595427183975, "grad_norm": 4.843143463134766, "learning_rate": 7.579962555959564e-06, "loss": 0.3439, "step": 6606 }, { "epoch": 0.33421012696646263, "grad_norm": 12.988786697387695, "learning_rate": 7.579275008908833e-06, "loss": 0.455, "step": 6607 }, { "epoch": 0.3342607112145278, "grad_norm": 7.845353126525879, "learning_rate": 7.578587395396416e-06, "loss": 0.4552, "step": 6608 }, { "epoch": 0.334311295462593, "grad_norm": 7.563396453857422, "learning_rate": 7.577899715440029e-06, "loss": 0.3981, "step": 6609 }, { "epoch": 0.3343618797106581, "grad_norm": 5.310657501220703, "learning_rate": 7.577211969057394e-06, "loss": 0.4647, "step": 6610 }, { "epoch": 0.33441246395872326, "grad_norm": 11.845695495605469, "learning_rate": 7.576524156266232e-06, "loss": 0.5189, "step": 6611 }, { "epoch": 0.3344630482067884, "grad_norm": 12.071866035461426, "learning_rate": 7.575836277084265e-06, "loss": 0.3696, "step": 6612 }, { "epoch": 0.33451363245485355, "grad_norm": 9.696246147155762, "learning_rate": 7.575148331529218e-06, "loss": 0.5328, "step": 6613 }, { "epoch": 0.3345642167029187, "grad_norm": 6.735287189483643, "learning_rate": 7.57446031961882e-06, "loss": 0.3987, "step": 6614 }, { "epoch": 0.33461480095098384, "grad_norm": 6.130788326263428, "learning_rate": 7.573772241370797e-06, "loss": 0.3591, "step": 6615 }, { "epoch": 0.33466538519904904, "grad_norm": 4.849286079406738, "learning_rate": 7.573084096802881e-06, "loss": 0.3529, "step": 6616 }, { "epoch": 0.3347159694471142, "grad_norm": 6.912206172943115, "learning_rate": 7.572395885932802e-06, "loss": 0.488, "step": 6617 }, { "epoch": 0.3347665536951793, "grad_norm": 8.968002319335938, "learning_rate": 7.571707608778295e-06, "loss": 0.3942, "step": 6618 }, { "epoch": 0.33481713794324447, "grad_norm": 16.115753173828125, "learning_rate": 7.571019265357094e-06, "loss": 0.5635, "step": 6619 }, { "epoch": 0.3348677221913096, "grad_norm": 5.752453327178955, "learning_rate": 7.570330855686936e-06, "loss": 0.3986, "step": 6620 }, { "epoch": 0.33491830643937476, "grad_norm": 10.366610527038574, "learning_rate": 7.569642379785561e-06, "loss": 0.4937, "step": 6621 }, { "epoch": 0.33496889068743996, "grad_norm": 9.72526741027832, "learning_rate": 7.5689538376707095e-06, "loss": 0.4044, "step": 6622 }, { "epoch": 0.3350194749355051, "grad_norm": 6.790070056915283, "learning_rate": 7.568265229360121e-06, "loss": 0.4408, "step": 6623 }, { "epoch": 0.33507005918357025, "grad_norm": 10.727900505065918, "learning_rate": 7.567576554871543e-06, "loss": 0.4335, "step": 6624 }, { "epoch": 0.3351206434316354, "grad_norm": 12.812908172607422, "learning_rate": 7.566887814222717e-06, "loss": 0.5893, "step": 6625 }, { "epoch": 0.33517122767970053, "grad_norm": 9.464866638183594, "learning_rate": 7.566199007431392e-06, "loss": 0.4243, "step": 6626 }, { "epoch": 0.3352218119277657, "grad_norm": 7.108940124511719, "learning_rate": 7.565510134515318e-06, "loss": 0.4444, "step": 6627 }, { "epoch": 0.3352723961758308, "grad_norm": 8.450647354125977, "learning_rate": 7.564821195492244e-06, "loss": 0.4717, "step": 6628 }, { "epoch": 0.335322980423896, "grad_norm": 7.282154560089111, "learning_rate": 7.5641321903799225e-06, "loss": 0.4471, "step": 6629 }, { "epoch": 0.33537356467196117, "grad_norm": 13.377398490905762, "learning_rate": 7.56344311919611e-06, "loss": 0.5403, "step": 6630 }, { "epoch": 0.3354241489200263, "grad_norm": 7.345751762390137, "learning_rate": 7.562753981958557e-06, "loss": 0.5185, "step": 6631 }, { "epoch": 0.33547473316809145, "grad_norm": 6.40824031829834, "learning_rate": 7.562064778685025e-06, "loss": 0.4816, "step": 6632 }, { "epoch": 0.3355253174161566, "grad_norm": 5.984322547912598, "learning_rate": 7.561375509393274e-06, "loss": 0.4122, "step": 6633 }, { "epoch": 0.33557590166422174, "grad_norm": 7.206704139709473, "learning_rate": 7.560686174101061e-06, "loss": 0.4227, "step": 6634 }, { "epoch": 0.3356264859122869, "grad_norm": 8.323949813842773, "learning_rate": 7.559996772826151e-06, "loss": 0.5335, "step": 6635 }, { "epoch": 0.3356770701603521, "grad_norm": 6.63358211517334, "learning_rate": 7.559307305586307e-06, "loss": 0.373, "step": 6636 }, { "epoch": 0.33572765440841723, "grad_norm": 5.317558765411377, "learning_rate": 7.558617772399297e-06, "loss": 0.4684, "step": 6637 }, { "epoch": 0.3357782386564824, "grad_norm": 7.903204441070557, "learning_rate": 7.557928173282885e-06, "loss": 0.3615, "step": 6638 }, { "epoch": 0.3358288229045475, "grad_norm": 7.7216410636901855, "learning_rate": 7.557238508254842e-06, "loss": 0.4506, "step": 6639 }, { "epoch": 0.33587940715261266, "grad_norm": 8.54256534576416, "learning_rate": 7.556548777332941e-06, "loss": 0.488, "step": 6640 }, { "epoch": 0.3359299914006778, "grad_norm": 7.647430419921875, "learning_rate": 7.555858980534954e-06, "loss": 0.3907, "step": 6641 }, { "epoch": 0.335980575648743, "grad_norm": 9.621562004089355, "learning_rate": 7.555169117878653e-06, "loss": 0.4841, "step": 6642 }, { "epoch": 0.33603115989680815, "grad_norm": 5.490088939666748, "learning_rate": 7.554479189381817e-06, "loss": 0.316, "step": 6643 }, { "epoch": 0.3360817441448733, "grad_norm": 17.607091903686523, "learning_rate": 7.55378919506222e-06, "loss": 0.3527, "step": 6644 }, { "epoch": 0.33613232839293844, "grad_norm": 15.184837341308594, "learning_rate": 7.5530991349376445e-06, "loss": 0.4532, "step": 6645 }, { "epoch": 0.3361829126410036, "grad_norm": 4.684782981872559, "learning_rate": 7.552409009025874e-06, "loss": 0.3658, "step": 6646 }, { "epoch": 0.33623349688906873, "grad_norm": 14.06425666809082, "learning_rate": 7.551718817344684e-06, "loss": 0.5009, "step": 6647 }, { "epoch": 0.3362840811371339, "grad_norm": 12.724336624145508, "learning_rate": 7.551028559911867e-06, "loss": 0.5115, "step": 6648 }, { "epoch": 0.33633466538519907, "grad_norm": 14.993216514587402, "learning_rate": 7.550338236745205e-06, "loss": 0.7608, "step": 6649 }, { "epoch": 0.3363852496332642, "grad_norm": 6.352065563201904, "learning_rate": 7.5496478478624855e-06, "loss": 0.4434, "step": 6650 }, { "epoch": 0.33643583388132936, "grad_norm": 4.740993022918701, "learning_rate": 7.5489573932814995e-06, "loss": 0.3187, "step": 6651 }, { "epoch": 0.3364864181293945, "grad_norm": 9.97136116027832, "learning_rate": 7.548266873020041e-06, "loss": 0.3339, "step": 6652 }, { "epoch": 0.33653700237745965, "grad_norm": 7.404289722442627, "learning_rate": 7.5475762870958964e-06, "loss": 0.4913, "step": 6653 }, { "epoch": 0.3365875866255248, "grad_norm": 13.168581008911133, "learning_rate": 7.546885635526866e-06, "loss": 0.4772, "step": 6654 }, { "epoch": 0.33663817087358994, "grad_norm": 7.118463039398193, "learning_rate": 7.546194918330745e-06, "loss": 0.4024, "step": 6655 }, { "epoch": 0.33668875512165514, "grad_norm": 7.7563276290893555, "learning_rate": 7.54550413552533e-06, "loss": 0.4673, "step": 6656 }, { "epoch": 0.3367393393697203, "grad_norm": 5.8416266441345215, "learning_rate": 7.544813287128423e-06, "loss": 0.3929, "step": 6657 }, { "epoch": 0.3367899236177854, "grad_norm": 13.167359352111816, "learning_rate": 7.544122373157823e-06, "loss": 0.4107, "step": 6658 }, { "epoch": 0.33684050786585057, "grad_norm": 18.608333587646484, "learning_rate": 7.5434313936313355e-06, "loss": 0.7633, "step": 6659 }, { "epoch": 0.3368910921139157, "grad_norm": 7.817391395568848, "learning_rate": 7.542740348566764e-06, "loss": 0.4514, "step": 6660 }, { "epoch": 0.33694167636198086, "grad_norm": 7.961796760559082, "learning_rate": 7.542049237981915e-06, "loss": 0.4889, "step": 6661 }, { "epoch": 0.33699226061004606, "grad_norm": 4.470808982849121, "learning_rate": 7.5413580618945995e-06, "loss": 0.4089, "step": 6662 }, { "epoch": 0.3370428448581112, "grad_norm": 8.464883804321289, "learning_rate": 7.540666820322622e-06, "loss": 0.5133, "step": 6663 }, { "epoch": 0.33709342910617635, "grad_norm": 13.101709365844727, "learning_rate": 7.539975513283798e-06, "loss": 0.5059, "step": 6664 }, { "epoch": 0.3371440133542415, "grad_norm": 6.475765705108643, "learning_rate": 7.539284140795941e-06, "loss": 0.3777, "step": 6665 }, { "epoch": 0.33719459760230663, "grad_norm": 6.57054328918457, "learning_rate": 7.538592702876865e-06, "loss": 0.3399, "step": 6666 }, { "epoch": 0.3372451818503718, "grad_norm": 7.848598957061768, "learning_rate": 7.5379011995443865e-06, "loss": 0.4383, "step": 6667 }, { "epoch": 0.3372957660984369, "grad_norm": 5.095029830932617, "learning_rate": 7.537209630816325e-06, "loss": 0.4259, "step": 6668 }, { "epoch": 0.3373463503465021, "grad_norm": 8.415901184082031, "learning_rate": 7.536517996710498e-06, "loss": 0.4794, "step": 6669 }, { "epoch": 0.33739693459456727, "grad_norm": 7.042738437652588, "learning_rate": 7.535826297244731e-06, "loss": 0.452, "step": 6670 }, { "epoch": 0.3374475188426324, "grad_norm": 8.915809631347656, "learning_rate": 7.535134532436843e-06, "loss": 0.4151, "step": 6671 }, { "epoch": 0.33749810309069755, "grad_norm": 11.277194023132324, "learning_rate": 7.534442702304663e-06, "loss": 0.4814, "step": 6672 }, { "epoch": 0.3375486873387627, "grad_norm": 8.808242797851562, "learning_rate": 7.533750806866016e-06, "loss": 0.4871, "step": 6673 }, { "epoch": 0.33759927158682784, "grad_norm": 5.087639808654785, "learning_rate": 7.53305884613873e-06, "loss": 0.4339, "step": 6674 }, { "epoch": 0.337649855834893, "grad_norm": 10.384955406188965, "learning_rate": 7.5323668201406375e-06, "loss": 0.5405, "step": 6675 }, { "epoch": 0.3377004400829582, "grad_norm": 9.181845664978027, "learning_rate": 7.531674728889568e-06, "loss": 0.4102, "step": 6676 }, { "epoch": 0.33775102433102333, "grad_norm": 6.303916931152344, "learning_rate": 7.530982572403356e-06, "loss": 0.3551, "step": 6677 }, { "epoch": 0.3378016085790885, "grad_norm": 23.857742309570312, "learning_rate": 7.530290350699836e-06, "loss": 0.6907, "step": 6678 }, { "epoch": 0.3378521928271536, "grad_norm": 6.468024253845215, "learning_rate": 7.529598063796844e-06, "loss": 0.3232, "step": 6679 }, { "epoch": 0.33790277707521876, "grad_norm": 4.660991668701172, "learning_rate": 7.528905711712222e-06, "loss": 0.371, "step": 6680 }, { "epoch": 0.3379533613232839, "grad_norm": 11.525581359863281, "learning_rate": 7.528213294463808e-06, "loss": 0.4058, "step": 6681 }, { "epoch": 0.3380039455713491, "grad_norm": 6.8741583824157715, "learning_rate": 7.527520812069442e-06, "loss": 0.4984, "step": 6682 }, { "epoch": 0.33805452981941425, "grad_norm": 7.0996222496032715, "learning_rate": 7.52682826454697e-06, "loss": 0.3987, "step": 6683 }, { "epoch": 0.3381051140674794, "grad_norm": 15.03231430053711, "learning_rate": 7.52613565191424e-06, "loss": 0.5587, "step": 6684 }, { "epoch": 0.33815569831554454, "grad_norm": 5.710363388061523, "learning_rate": 7.525442974189092e-06, "loss": 0.4106, "step": 6685 }, { "epoch": 0.3382062825636097, "grad_norm": 5.544496059417725, "learning_rate": 7.524750231389382e-06, "loss": 0.3364, "step": 6686 }, { "epoch": 0.33825686681167483, "grad_norm": 18.989240646362305, "learning_rate": 7.524057423532956e-06, "loss": 0.4154, "step": 6687 }, { "epoch": 0.33830745105973997, "grad_norm": 9.939719200134277, "learning_rate": 7.523364550637665e-06, "loss": 0.4822, "step": 6688 }, { "epoch": 0.33835803530780517, "grad_norm": 14.971184730529785, "learning_rate": 7.522671612721365e-06, "loss": 0.4312, "step": 6689 }, { "epoch": 0.3384086195558703, "grad_norm": 6.934784889221191, "learning_rate": 7.521978609801912e-06, "loss": 0.4163, "step": 6690 }, { "epoch": 0.33845920380393546, "grad_norm": 7.372857093811035, "learning_rate": 7.521285541897161e-06, "loss": 0.3977, "step": 6691 }, { "epoch": 0.3385097880520006, "grad_norm": 13.73111343383789, "learning_rate": 7.52059240902497e-06, "loss": 0.5342, "step": 6692 }, { "epoch": 0.33856037230006575, "grad_norm": 6.036685466766357, "learning_rate": 7.519899211203204e-06, "loss": 0.3784, "step": 6693 }, { "epoch": 0.3386109565481309, "grad_norm": 10.26761245727539, "learning_rate": 7.51920594844972e-06, "loss": 0.4463, "step": 6694 }, { "epoch": 0.3386615407961961, "grad_norm": 8.817689895629883, "learning_rate": 7.518512620782384e-06, "loss": 0.4405, "step": 6695 }, { "epoch": 0.33871212504426124, "grad_norm": 7.236274242401123, "learning_rate": 7.517819228219061e-06, "loss": 0.3095, "step": 6696 }, { "epoch": 0.3387627092923264, "grad_norm": 9.394721984863281, "learning_rate": 7.517125770777618e-06, "loss": 0.5294, "step": 6697 }, { "epoch": 0.3388132935403915, "grad_norm": 11.404447555541992, "learning_rate": 7.516432248475922e-06, "loss": 0.5179, "step": 6698 }, { "epoch": 0.33886387778845667, "grad_norm": 5.718979358673096, "learning_rate": 7.515738661331848e-06, "loss": 0.3232, "step": 6699 }, { "epoch": 0.3389144620365218, "grad_norm": 7.693319797515869, "learning_rate": 7.515045009363265e-06, "loss": 0.44, "step": 6700 }, { "epoch": 0.33896504628458696, "grad_norm": 11.144015312194824, "learning_rate": 7.5143512925880445e-06, "loss": 0.5413, "step": 6701 }, { "epoch": 0.33901563053265216, "grad_norm": 8.332488059997559, "learning_rate": 7.513657511024066e-06, "loss": 0.4751, "step": 6702 }, { "epoch": 0.3390662147807173, "grad_norm": 10.195785522460938, "learning_rate": 7.5129636646892055e-06, "loss": 0.3346, "step": 6703 }, { "epoch": 0.33911679902878245, "grad_norm": 13.288806915283203, "learning_rate": 7.51226975360134e-06, "loss": 0.6042, "step": 6704 }, { "epoch": 0.3391673832768476, "grad_norm": 8.585272789001465, "learning_rate": 7.51157577777835e-06, "loss": 0.4966, "step": 6705 }, { "epoch": 0.33921796752491273, "grad_norm": 7.789920330047607, "learning_rate": 7.510881737238123e-06, "loss": 0.3619, "step": 6706 }, { "epoch": 0.3392685517729779, "grad_norm": 6.525665760040283, "learning_rate": 7.510187631998536e-06, "loss": 0.4246, "step": 6707 }, { "epoch": 0.339319136021043, "grad_norm": 6.484234809875488, "learning_rate": 7.509493462077475e-06, "loss": 0.3968, "step": 6708 }, { "epoch": 0.3393697202691082, "grad_norm": 5.205955505371094, "learning_rate": 7.5087992274928316e-06, "loss": 0.4202, "step": 6709 }, { "epoch": 0.33942030451717337, "grad_norm": 7.9982218742370605, "learning_rate": 7.50810492826249e-06, "loss": 0.4468, "step": 6710 }, { "epoch": 0.3394708887652385, "grad_norm": 4.647693157196045, "learning_rate": 7.507410564404342e-06, "loss": 0.4265, "step": 6711 }, { "epoch": 0.33952147301330365, "grad_norm": 6.517237186431885, "learning_rate": 7.5067161359362825e-06, "loss": 0.3439, "step": 6712 }, { "epoch": 0.3395720572613688, "grad_norm": 9.276806831359863, "learning_rate": 7.506021642876202e-06, "loss": 0.4638, "step": 6713 }, { "epoch": 0.33962264150943394, "grad_norm": 8.564850807189941, "learning_rate": 7.5053270852419955e-06, "loss": 0.4131, "step": 6714 }, { "epoch": 0.33967322575749914, "grad_norm": 9.995831489562988, "learning_rate": 7.504632463051563e-06, "loss": 0.4522, "step": 6715 }, { "epoch": 0.3397238100055643, "grad_norm": 7.724773406982422, "learning_rate": 7.5039377763228e-06, "loss": 0.4859, "step": 6716 }, { "epoch": 0.33977439425362943, "grad_norm": 10.30307674407959, "learning_rate": 7.50324302507361e-06, "loss": 0.3812, "step": 6717 }, { "epoch": 0.3398249785016946, "grad_norm": 7.886294364929199, "learning_rate": 7.502548209321894e-06, "loss": 0.3516, "step": 6718 }, { "epoch": 0.3398755627497597, "grad_norm": 8.16425895690918, "learning_rate": 7.501853329085554e-06, "loss": 0.4294, "step": 6719 }, { "epoch": 0.33992614699782486, "grad_norm": 8.37209415435791, "learning_rate": 7.5011583843824965e-06, "loss": 0.4626, "step": 6720 }, { "epoch": 0.33997673124589, "grad_norm": 9.04401969909668, "learning_rate": 7.500463375230629e-06, "loss": 0.5356, "step": 6721 }, { "epoch": 0.3400273154939552, "grad_norm": 7.227890968322754, "learning_rate": 7.499768301647861e-06, "loss": 0.4124, "step": 6722 }, { "epoch": 0.34007789974202035, "grad_norm": 8.04024887084961, "learning_rate": 7.499073163652101e-06, "loss": 0.4418, "step": 6723 }, { "epoch": 0.3401284839900855, "grad_norm": 8.805158615112305, "learning_rate": 7.4983779612612614e-06, "loss": 0.4643, "step": 6724 }, { "epoch": 0.34017906823815064, "grad_norm": 7.378019332885742, "learning_rate": 7.497682694493258e-06, "loss": 0.4748, "step": 6725 }, { "epoch": 0.3402296524862158, "grad_norm": 9.463622093200684, "learning_rate": 7.4969873633660025e-06, "loss": 0.4223, "step": 6726 }, { "epoch": 0.3402802367342809, "grad_norm": 7.733773231506348, "learning_rate": 7.496291967897414e-06, "loss": 0.408, "step": 6727 }, { "epoch": 0.34033082098234607, "grad_norm": 9.861556053161621, "learning_rate": 7.495596508105413e-06, "loss": 0.5168, "step": 6728 }, { "epoch": 0.34038140523041127, "grad_norm": 6.179260730743408, "learning_rate": 7.494900984007916e-06, "loss": 0.4163, "step": 6729 }, { "epoch": 0.3404319894784764, "grad_norm": 9.265591621398926, "learning_rate": 7.4942053956228466e-06, "loss": 0.5035, "step": 6730 }, { "epoch": 0.34048257372654156, "grad_norm": 6.620155334472656, "learning_rate": 7.49350974296813e-06, "loss": 0.5316, "step": 6731 }, { "epoch": 0.3405331579746067, "grad_norm": 9.468582153320312, "learning_rate": 7.49281402606169e-06, "loss": 0.4549, "step": 6732 }, { "epoch": 0.34058374222267185, "grad_norm": 8.493244171142578, "learning_rate": 7.492118244921454e-06, "loss": 0.432, "step": 6733 }, { "epoch": 0.340634326470737, "grad_norm": 11.491135597229004, "learning_rate": 7.491422399565349e-06, "loss": 0.4229, "step": 6734 }, { "epoch": 0.3406849107188022, "grad_norm": 10.549452781677246, "learning_rate": 7.490726490011307e-06, "loss": 0.4388, "step": 6735 }, { "epoch": 0.34073549496686734, "grad_norm": 5.695113182067871, "learning_rate": 7.490030516277259e-06, "loss": 0.4482, "step": 6736 }, { "epoch": 0.3407860792149325, "grad_norm": 6.8720316886901855, "learning_rate": 7.489334478381141e-06, "loss": 0.4099, "step": 6737 }, { "epoch": 0.3408366634629976, "grad_norm": 6.481788158416748, "learning_rate": 7.488638376340884e-06, "loss": 0.3559, "step": 6738 }, { "epoch": 0.34088724771106277, "grad_norm": 6.877686023712158, "learning_rate": 7.487942210174428e-06, "loss": 0.3869, "step": 6739 }, { "epoch": 0.3409378319591279, "grad_norm": 15.065977096557617, "learning_rate": 7.487245979899711e-06, "loss": 0.4796, "step": 6740 }, { "epoch": 0.34098841620719306, "grad_norm": 5.6569390296936035, "learning_rate": 7.486549685534672e-06, "loss": 0.3601, "step": 6741 }, { "epoch": 0.34103900045525826, "grad_norm": 7.088983535766602, "learning_rate": 7.485853327097254e-06, "loss": 0.3987, "step": 6742 }, { "epoch": 0.3410895847033234, "grad_norm": 12.021478652954102, "learning_rate": 7.485156904605399e-06, "loss": 0.528, "step": 6743 }, { "epoch": 0.34114016895138854, "grad_norm": 5.655386447906494, "learning_rate": 7.484460418077055e-06, "loss": 0.4277, "step": 6744 }, { "epoch": 0.3411907531994537, "grad_norm": 12.272002220153809, "learning_rate": 7.483763867530167e-06, "loss": 0.5952, "step": 6745 }, { "epoch": 0.34124133744751883, "grad_norm": 7.376690864562988, "learning_rate": 7.483067252982683e-06, "loss": 0.4297, "step": 6746 }, { "epoch": 0.341291921695584, "grad_norm": 9.063998222351074, "learning_rate": 7.4823705744525535e-06, "loss": 0.5087, "step": 6747 }, { "epoch": 0.3413425059436491, "grad_norm": 7.109166145324707, "learning_rate": 7.4816738319577295e-06, "loss": 0.4668, "step": 6748 }, { "epoch": 0.3413930901917143, "grad_norm": 6.727160930633545, "learning_rate": 7.480977025516165e-06, "loss": 0.3257, "step": 6749 }, { "epoch": 0.34144367443977947, "grad_norm": 7.1086907386779785, "learning_rate": 7.480280155145816e-06, "loss": 0.4531, "step": 6750 }, { "epoch": 0.3414942586878446, "grad_norm": 7.165679931640625, "learning_rate": 7.479583220864638e-06, "loss": 0.4833, "step": 6751 }, { "epoch": 0.34154484293590975, "grad_norm": 4.854776859283447, "learning_rate": 7.478886222690589e-06, "loss": 0.4274, "step": 6752 }, { "epoch": 0.3415954271839749, "grad_norm": 7.947873115539551, "learning_rate": 7.478189160641631e-06, "loss": 0.4018, "step": 6753 }, { "epoch": 0.34164601143204004, "grad_norm": 17.451740264892578, "learning_rate": 7.477492034735722e-06, "loss": 0.6983, "step": 6754 }, { "epoch": 0.34169659568010524, "grad_norm": 8.080266952514648, "learning_rate": 7.476794844990829e-06, "loss": 0.4083, "step": 6755 }, { "epoch": 0.3417471799281704, "grad_norm": 8.70557975769043, "learning_rate": 7.4760975914249155e-06, "loss": 0.5212, "step": 6756 }, { "epoch": 0.34179776417623553, "grad_norm": 7.9646453857421875, "learning_rate": 7.475400274055946e-06, "loss": 0.369, "step": 6757 }, { "epoch": 0.3418483484243007, "grad_norm": 14.823925971984863, "learning_rate": 7.474702892901892e-06, "loss": 0.4771, "step": 6758 }, { "epoch": 0.3418989326723658, "grad_norm": 10.616230964660645, "learning_rate": 7.474005447980721e-06, "loss": 0.519, "step": 6759 }, { "epoch": 0.34194951692043096, "grad_norm": 9.558659553527832, "learning_rate": 7.473307939310407e-06, "loss": 0.4992, "step": 6760 }, { "epoch": 0.3420001011684961, "grad_norm": 6.848156929016113, "learning_rate": 7.47261036690892e-06, "loss": 0.4345, "step": 6761 }, { "epoch": 0.3420506854165613, "grad_norm": 5.5775299072265625, "learning_rate": 7.471912730794237e-06, "loss": 0.4405, "step": 6762 }, { "epoch": 0.34210126966462645, "grad_norm": 7.25241756439209, "learning_rate": 7.471215030984332e-06, "loss": 0.4939, "step": 6763 }, { "epoch": 0.3421518539126916, "grad_norm": 10.216257095336914, "learning_rate": 7.470517267497187e-06, "loss": 0.4886, "step": 6764 }, { "epoch": 0.34220243816075674, "grad_norm": 7.53640079498291, "learning_rate": 7.469819440350777e-06, "loss": 0.3814, "step": 6765 }, { "epoch": 0.3422530224088219, "grad_norm": 5.7386908531188965, "learning_rate": 7.4691215495630875e-06, "loss": 0.3257, "step": 6766 }, { "epoch": 0.342303606656887, "grad_norm": 6.204466819763184, "learning_rate": 7.468423595152099e-06, "loss": 0.4161, "step": 6767 }, { "epoch": 0.34235419090495217, "grad_norm": 10.487346649169922, "learning_rate": 7.467725577135796e-06, "loss": 0.4956, "step": 6768 }, { "epoch": 0.34240477515301737, "grad_norm": 9.518423080444336, "learning_rate": 7.467027495532166e-06, "loss": 0.4806, "step": 6769 }, { "epoch": 0.3424553594010825, "grad_norm": 15.036615371704102, "learning_rate": 7.466329350359196e-06, "loss": 0.5239, "step": 6770 }, { "epoch": 0.34250594364914766, "grad_norm": 14.223226547241211, "learning_rate": 7.465631141634876e-06, "loss": 0.3949, "step": 6771 }, { "epoch": 0.3425565278972128, "grad_norm": 11.465811729431152, "learning_rate": 7.464932869377197e-06, "loss": 0.4279, "step": 6772 }, { "epoch": 0.34260711214527795, "grad_norm": 8.083550453186035, "learning_rate": 7.4642345336041524e-06, "loss": 0.3101, "step": 6773 }, { "epoch": 0.3426576963933431, "grad_norm": 6.6128926277160645, "learning_rate": 7.463536134333734e-06, "loss": 0.5106, "step": 6774 }, { "epoch": 0.3427082806414083, "grad_norm": 7.140093803405762, "learning_rate": 7.462837671583943e-06, "loss": 0.5039, "step": 6775 }, { "epoch": 0.34275886488947344, "grad_norm": 6.950605392456055, "learning_rate": 7.462139145372771e-06, "loss": 0.38, "step": 6776 }, { "epoch": 0.3428094491375386, "grad_norm": 9.544285774230957, "learning_rate": 7.461440555718221e-06, "loss": 0.4886, "step": 6777 }, { "epoch": 0.3428600333856037, "grad_norm": 9.515202522277832, "learning_rate": 7.460741902638293e-06, "loss": 0.3905, "step": 6778 }, { "epoch": 0.34291061763366887, "grad_norm": 10.594524383544922, "learning_rate": 7.4600431861509916e-06, "loss": 0.5696, "step": 6779 }, { "epoch": 0.342961201881734, "grad_norm": 30.857677459716797, "learning_rate": 7.459344406274318e-06, "loss": 0.6458, "step": 6780 }, { "epoch": 0.34301178612979916, "grad_norm": 9.813668251037598, "learning_rate": 7.45864556302628e-06, "loss": 0.524, "step": 6781 }, { "epoch": 0.34306237037786436, "grad_norm": 14.339723587036133, "learning_rate": 7.457946656424884e-06, "loss": 0.5147, "step": 6782 }, { "epoch": 0.3431129546259295, "grad_norm": 6.700366497039795, "learning_rate": 7.457247686488139e-06, "loss": 0.3908, "step": 6783 }, { "epoch": 0.34316353887399464, "grad_norm": 19.411394119262695, "learning_rate": 7.456548653234057e-06, "loss": 0.456, "step": 6784 }, { "epoch": 0.3432141231220598, "grad_norm": 5.241639614105225, "learning_rate": 7.45584955668065e-06, "loss": 0.4345, "step": 6785 }, { "epoch": 0.34326470737012493, "grad_norm": 11.782526016235352, "learning_rate": 7.455150396845931e-06, "loss": 0.3438, "step": 6786 }, { "epoch": 0.3433152916181901, "grad_norm": 6.768352031707764, "learning_rate": 7.454451173747917e-06, "loss": 0.3772, "step": 6787 }, { "epoch": 0.3433658758662552, "grad_norm": 15.433774948120117, "learning_rate": 7.453751887404624e-06, "loss": 0.4939, "step": 6788 }, { "epoch": 0.3434164601143204, "grad_norm": 11.374080657958984, "learning_rate": 7.4530525378340724e-06, "loss": 0.5893, "step": 6789 }, { "epoch": 0.34346704436238557, "grad_norm": 6.263223648071289, "learning_rate": 7.452353125054282e-06, "loss": 0.4059, "step": 6790 }, { "epoch": 0.3435176286104507, "grad_norm": 6.945351600646973, "learning_rate": 7.451653649083276e-06, "loss": 0.5638, "step": 6791 }, { "epoch": 0.34356821285851585, "grad_norm": 11.18358039855957, "learning_rate": 7.450954109939075e-06, "loss": 0.4671, "step": 6792 }, { "epoch": 0.343618797106581, "grad_norm": 7.206847667694092, "learning_rate": 7.450254507639707e-06, "loss": 0.4465, "step": 6793 }, { "epoch": 0.34366938135464614, "grad_norm": 12.082908630371094, "learning_rate": 7.449554842203201e-06, "loss": 0.3448, "step": 6794 }, { "epoch": 0.34371996560271134, "grad_norm": 16.740215301513672, "learning_rate": 7.4488551136475805e-06, "loss": 0.5624, "step": 6795 }, { "epoch": 0.3437705498507765, "grad_norm": 10.984670639038086, "learning_rate": 7.4481553219908805e-06, "loss": 0.4103, "step": 6796 }, { "epoch": 0.34382113409884163, "grad_norm": 6.4764227867126465, "learning_rate": 7.44745546725113e-06, "loss": 0.344, "step": 6797 }, { "epoch": 0.3438717183469068, "grad_norm": 8.47128677368164, "learning_rate": 7.446755549446365e-06, "loss": 0.5735, "step": 6798 }, { "epoch": 0.3439223025949719, "grad_norm": 18.3724422454834, "learning_rate": 7.446055568594619e-06, "loss": 0.4106, "step": 6799 }, { "epoch": 0.34397288684303706, "grad_norm": 19.98798179626465, "learning_rate": 7.445355524713928e-06, "loss": 0.3298, "step": 6800 }, { "epoch": 0.3440234710911022, "grad_norm": 8.62033748626709, "learning_rate": 7.444655417822333e-06, "loss": 0.4637, "step": 6801 }, { "epoch": 0.3440740553391674, "grad_norm": 7.644872188568115, "learning_rate": 7.4439552479378715e-06, "loss": 0.3903, "step": 6802 }, { "epoch": 0.34412463958723255, "grad_norm": 4.83209753036499, "learning_rate": 7.443255015078587e-06, "loss": 0.4288, "step": 6803 }, { "epoch": 0.3441752238352977, "grad_norm": 8.227404594421387, "learning_rate": 7.442554719262524e-06, "loss": 0.4079, "step": 6804 }, { "epoch": 0.34422580808336284, "grad_norm": 10.904650688171387, "learning_rate": 7.441854360507723e-06, "loss": 0.3361, "step": 6805 }, { "epoch": 0.344276392331428, "grad_norm": 9.324617385864258, "learning_rate": 7.441153938832233e-06, "loss": 0.4155, "step": 6806 }, { "epoch": 0.3443269765794931, "grad_norm": 6.345114707946777, "learning_rate": 7.440453454254104e-06, "loss": 0.3997, "step": 6807 }, { "epoch": 0.34437756082755827, "grad_norm": 8.052871704101562, "learning_rate": 7.439752906791384e-06, "loss": 0.3827, "step": 6808 }, { "epoch": 0.34442814507562347, "grad_norm": 9.92869758605957, "learning_rate": 7.4390522964621245e-06, "loss": 0.3766, "step": 6809 }, { "epoch": 0.3444787293236886, "grad_norm": 6.6818013191223145, "learning_rate": 7.438351623284379e-06, "loss": 0.3077, "step": 6810 }, { "epoch": 0.34452931357175376, "grad_norm": 6.045687198638916, "learning_rate": 7.4376508872762e-06, "loss": 0.3498, "step": 6811 }, { "epoch": 0.3445798978198189, "grad_norm": 6.025667190551758, "learning_rate": 7.4369500884556465e-06, "loss": 0.3404, "step": 6812 }, { "epoch": 0.34463048206788405, "grad_norm": 6.491898536682129, "learning_rate": 7.436249226840776e-06, "loss": 0.4228, "step": 6813 }, { "epoch": 0.3446810663159492, "grad_norm": 18.01947784423828, "learning_rate": 7.435548302449647e-06, "loss": 0.3649, "step": 6814 }, { "epoch": 0.3447316505640144, "grad_norm": 12.403814315795898, "learning_rate": 7.434847315300321e-06, "loss": 0.6018, "step": 6815 }, { "epoch": 0.34478223481207954, "grad_norm": 7.9318366050720215, "learning_rate": 7.434146265410861e-06, "loss": 0.5088, "step": 6816 }, { "epoch": 0.3448328190601447, "grad_norm": 9.007026672363281, "learning_rate": 7.433445152799332e-06, "loss": 0.4627, "step": 6817 }, { "epoch": 0.3448834033082098, "grad_norm": 8.808420181274414, "learning_rate": 7.432743977483797e-06, "loss": 0.42, "step": 6818 }, { "epoch": 0.34493398755627497, "grad_norm": 11.894937515258789, "learning_rate": 7.432042739482328e-06, "loss": 0.4895, "step": 6819 }, { "epoch": 0.3449845718043401, "grad_norm": 10.968152046203613, "learning_rate": 7.43134143881299e-06, "loss": 0.471, "step": 6820 }, { "epoch": 0.34503515605240526, "grad_norm": 10.402361869812012, "learning_rate": 7.430640075493856e-06, "loss": 0.3807, "step": 6821 }, { "epoch": 0.34508574030047046, "grad_norm": 7.328420639038086, "learning_rate": 7.429938649543e-06, "loss": 0.4086, "step": 6822 }, { "epoch": 0.3451363245485356, "grad_norm": 13.189268112182617, "learning_rate": 7.429237160978493e-06, "loss": 0.4246, "step": 6823 }, { "epoch": 0.34518690879660074, "grad_norm": 18.259464263916016, "learning_rate": 7.4285356098184126e-06, "loss": 0.675, "step": 6824 }, { "epoch": 0.3452374930446659, "grad_norm": 7.727732181549072, "learning_rate": 7.427833996080834e-06, "loss": 0.3982, "step": 6825 }, { "epoch": 0.34528807729273103, "grad_norm": 10.544194221496582, "learning_rate": 7.427132319783839e-06, "loss": 0.4327, "step": 6826 }, { "epoch": 0.3453386615407962, "grad_norm": 5.699851036071777, "learning_rate": 7.426430580945507e-06, "loss": 0.4009, "step": 6827 }, { "epoch": 0.3453892457888613, "grad_norm": 6.421329021453857, "learning_rate": 7.4257287795839185e-06, "loss": 0.448, "step": 6828 }, { "epoch": 0.3454398300369265, "grad_norm": 11.860252380371094, "learning_rate": 7.4250269157171595e-06, "loss": 0.4649, "step": 6829 }, { "epoch": 0.34549041428499166, "grad_norm": 7.522774696350098, "learning_rate": 7.424324989363313e-06, "loss": 0.3831, "step": 6830 }, { "epoch": 0.3455409985330568, "grad_norm": 7.389225006103516, "learning_rate": 7.423623000540467e-06, "loss": 0.4515, "step": 6831 }, { "epoch": 0.34559158278112195, "grad_norm": 8.602344512939453, "learning_rate": 7.422920949266712e-06, "loss": 0.4435, "step": 6832 }, { "epoch": 0.3456421670291871, "grad_norm": 12.267740249633789, "learning_rate": 7.422218835560136e-06, "loss": 0.3959, "step": 6833 }, { "epoch": 0.34569275127725224, "grad_norm": 8.654945373535156, "learning_rate": 7.4215166594388295e-06, "loss": 0.4556, "step": 6834 }, { "epoch": 0.34574333552531744, "grad_norm": 11.865312576293945, "learning_rate": 7.4208144209208895e-06, "loss": 0.6679, "step": 6835 }, { "epoch": 0.3457939197733826, "grad_norm": 7.868284702301025, "learning_rate": 7.420112120024408e-06, "loss": 0.567, "step": 6836 }, { "epoch": 0.34584450402144773, "grad_norm": 8.009620666503906, "learning_rate": 7.419409756767484e-06, "loss": 0.4679, "step": 6837 }, { "epoch": 0.3458950882695129, "grad_norm": 7.864128112792969, "learning_rate": 7.418707331168213e-06, "loss": 0.3848, "step": 6838 }, { "epoch": 0.345945672517578, "grad_norm": 9.66479778289795, "learning_rate": 7.4180048432446975e-06, "loss": 0.4758, "step": 6839 }, { "epoch": 0.34599625676564316, "grad_norm": 5.2747344970703125, "learning_rate": 7.417302293015036e-06, "loss": 0.3749, "step": 6840 }, { "epoch": 0.3460468410137083, "grad_norm": 7.316106796264648, "learning_rate": 7.416599680497335e-06, "loss": 0.4889, "step": 6841 }, { "epoch": 0.3460974252617735, "grad_norm": 9.326606750488281, "learning_rate": 7.415897005709698e-06, "loss": 0.5371, "step": 6842 }, { "epoch": 0.34614800950983865, "grad_norm": 13.016218185424805, "learning_rate": 7.415194268670228e-06, "loss": 0.493, "step": 6843 }, { "epoch": 0.3461985937579038, "grad_norm": 4.277464389801025, "learning_rate": 7.414491469397036e-06, "loss": 0.3902, "step": 6844 }, { "epoch": 0.34624917800596894, "grad_norm": 11.669625282287598, "learning_rate": 7.413788607908232e-06, "loss": 0.6615, "step": 6845 }, { "epoch": 0.3462997622540341, "grad_norm": 9.004986763000488, "learning_rate": 7.413085684221925e-06, "loss": 0.5106, "step": 6846 }, { "epoch": 0.3463503465020992, "grad_norm": 10.035006523132324, "learning_rate": 7.412382698356227e-06, "loss": 0.458, "step": 6847 }, { "epoch": 0.3464009307501644, "grad_norm": 6.306484699249268, "learning_rate": 7.411679650329258e-06, "loss": 0.4028, "step": 6848 }, { "epoch": 0.34645151499822957, "grad_norm": 4.114457607269287, "learning_rate": 7.410976540159125e-06, "loss": 0.3908, "step": 6849 }, { "epoch": 0.3465020992462947, "grad_norm": 6.643782138824463, "learning_rate": 7.410273367863951e-06, "loss": 0.4539, "step": 6850 }, { "epoch": 0.34655268349435986, "grad_norm": 5.081124782562256, "learning_rate": 7.409570133461857e-06, "loss": 0.4033, "step": 6851 }, { "epoch": 0.346603267742425, "grad_norm": 7.457711219787598, "learning_rate": 7.408866836970958e-06, "loss": 0.4752, "step": 6852 }, { "epoch": 0.34665385199049015, "grad_norm": 6.624999046325684, "learning_rate": 7.408163478409378e-06, "loss": 0.5102, "step": 6853 }, { "epoch": 0.3467044362385553, "grad_norm": 8.639857292175293, "learning_rate": 7.407460057795246e-06, "loss": 0.4697, "step": 6854 }, { "epoch": 0.3467550204866205, "grad_norm": 16.43054962158203, "learning_rate": 7.40675657514668e-06, "loss": 0.4762, "step": 6855 }, { "epoch": 0.34680560473468564, "grad_norm": 9.93349838256836, "learning_rate": 7.406053030481811e-06, "loss": 0.5443, "step": 6856 }, { "epoch": 0.3468561889827508, "grad_norm": 6.0110182762146, "learning_rate": 7.405349423818767e-06, "loss": 0.3568, "step": 6857 }, { "epoch": 0.3469067732308159, "grad_norm": 9.025370597839355, "learning_rate": 7.404645755175679e-06, "loss": 0.5034, "step": 6858 }, { "epoch": 0.34695735747888107, "grad_norm": 8.18979263305664, "learning_rate": 7.403942024570676e-06, "loss": 0.4212, "step": 6859 }, { "epoch": 0.3470079417269462, "grad_norm": 7.507424831390381, "learning_rate": 7.403238232021896e-06, "loss": 0.474, "step": 6860 }, { "epoch": 0.34705852597501136, "grad_norm": 5.894152641296387, "learning_rate": 7.402534377547471e-06, "loss": 0.3457, "step": 6861 }, { "epoch": 0.34710911022307656, "grad_norm": 7.270450592041016, "learning_rate": 7.401830461165538e-06, "loss": 0.4387, "step": 6862 }, { "epoch": 0.3471596944711417, "grad_norm": 8.30997085571289, "learning_rate": 7.401126482894234e-06, "loss": 0.405, "step": 6863 }, { "epoch": 0.34721027871920684, "grad_norm": 5.491744518280029, "learning_rate": 7.400422442751702e-06, "loss": 0.3726, "step": 6864 }, { "epoch": 0.347260862967272, "grad_norm": 11.909520149230957, "learning_rate": 7.39971834075608e-06, "loss": 0.4954, "step": 6865 }, { "epoch": 0.34731144721533713, "grad_norm": 9.547587394714355, "learning_rate": 7.399014176925514e-06, "loss": 0.5863, "step": 6866 }, { "epoch": 0.3473620314634023, "grad_norm": 12.33639907836914, "learning_rate": 7.398309951278147e-06, "loss": 0.4249, "step": 6867 }, { "epoch": 0.3474126157114675, "grad_norm": 10.78829574584961, "learning_rate": 7.397605663832126e-06, "loss": 0.4485, "step": 6868 }, { "epoch": 0.3474631999595326, "grad_norm": 11.290823936462402, "learning_rate": 7.396901314605597e-06, "loss": 0.5198, "step": 6869 }, { "epoch": 0.34751378420759776, "grad_norm": 7.902760028839111, "learning_rate": 7.396196903616711e-06, "loss": 0.4131, "step": 6870 }, { "epoch": 0.3475643684556629, "grad_norm": 5.182350158691406, "learning_rate": 7.3954924308836184e-06, "loss": 0.338, "step": 6871 }, { "epoch": 0.34761495270372805, "grad_norm": 8.361994743347168, "learning_rate": 7.394787896424471e-06, "loss": 0.506, "step": 6872 }, { "epoch": 0.3476655369517932, "grad_norm": 9.789566993713379, "learning_rate": 7.394083300257428e-06, "loss": 0.3921, "step": 6873 }, { "epoch": 0.34771612119985834, "grad_norm": 5.617978096008301, "learning_rate": 7.393378642400637e-06, "loss": 0.3681, "step": 6874 }, { "epoch": 0.34776670544792354, "grad_norm": 6.2321085929870605, "learning_rate": 7.39267392287226e-06, "loss": 0.4498, "step": 6875 }, { "epoch": 0.3478172896959887, "grad_norm": 5.651655197143555, "learning_rate": 7.391969141690456e-06, "loss": 0.3651, "step": 6876 }, { "epoch": 0.34786787394405383, "grad_norm": 13.473709106445312, "learning_rate": 7.391264298873384e-06, "loss": 0.4725, "step": 6877 }, { "epoch": 0.347918458192119, "grad_norm": 11.292093276977539, "learning_rate": 7.390559394439208e-06, "loss": 0.6139, "step": 6878 }, { "epoch": 0.3479690424401841, "grad_norm": 12.362923622131348, "learning_rate": 7.389854428406089e-06, "loss": 0.4167, "step": 6879 }, { "epoch": 0.34801962668824926, "grad_norm": 7.9676032066345215, "learning_rate": 7.389149400792196e-06, "loss": 0.3991, "step": 6880 }, { "epoch": 0.3480702109363144, "grad_norm": 4.927303314208984, "learning_rate": 7.388444311615692e-06, "loss": 0.4414, "step": 6881 }, { "epoch": 0.3481207951843796, "grad_norm": 7.890200614929199, "learning_rate": 7.3877391608947465e-06, "loss": 0.3753, "step": 6882 }, { "epoch": 0.34817137943244475, "grad_norm": 5.27211332321167, "learning_rate": 7.387033948647533e-06, "loss": 0.3419, "step": 6883 }, { "epoch": 0.3482219636805099, "grad_norm": 10.287863731384277, "learning_rate": 7.386328674892218e-06, "loss": 0.4756, "step": 6884 }, { "epoch": 0.34827254792857504, "grad_norm": 5.283984184265137, "learning_rate": 7.3856233396469774e-06, "loss": 0.3896, "step": 6885 }, { "epoch": 0.3483231321766402, "grad_norm": 5.246545314788818, "learning_rate": 7.3849179429299866e-06, "loss": 0.3967, "step": 6886 }, { "epoch": 0.3483737164247053, "grad_norm": 7.156754016876221, "learning_rate": 7.38421248475942e-06, "loss": 0.5596, "step": 6887 }, { "epoch": 0.3484243006727705, "grad_norm": 7.507916450500488, "learning_rate": 7.383506965153456e-06, "loss": 0.4869, "step": 6888 }, { "epoch": 0.34847488492083567, "grad_norm": 7.416009902954102, "learning_rate": 7.382801384130276e-06, "loss": 0.4294, "step": 6889 }, { "epoch": 0.3485254691689008, "grad_norm": 6.632057189941406, "learning_rate": 7.382095741708058e-06, "loss": 0.3808, "step": 6890 }, { "epoch": 0.34857605341696596, "grad_norm": 7.922435283660889, "learning_rate": 7.381390037904987e-06, "loss": 0.4842, "step": 6891 }, { "epoch": 0.3486266376650311, "grad_norm": 7.880516052246094, "learning_rate": 7.380684272739248e-06, "loss": 0.3688, "step": 6892 }, { "epoch": 0.34867722191309625, "grad_norm": 7.936642646789551, "learning_rate": 7.3799784462290254e-06, "loss": 0.5254, "step": 6893 }, { "epoch": 0.3487278061611614, "grad_norm": 9.659621238708496, "learning_rate": 7.379272558392505e-06, "loss": 0.4247, "step": 6894 }, { "epoch": 0.3487783904092266, "grad_norm": 4.3279709815979, "learning_rate": 7.37856660924788e-06, "loss": 0.3862, "step": 6895 }, { "epoch": 0.34882897465729173, "grad_norm": 16.174428939819336, "learning_rate": 7.377860598813337e-06, "loss": 0.5583, "step": 6896 }, { "epoch": 0.3488795589053569, "grad_norm": 5.112790584564209, "learning_rate": 7.37715452710707e-06, "loss": 0.3569, "step": 6897 }, { "epoch": 0.348930143153422, "grad_norm": 4.337503910064697, "learning_rate": 7.376448394147273e-06, "loss": 0.2429, "step": 6898 }, { "epoch": 0.34898072740148717, "grad_norm": 7.406955718994141, "learning_rate": 7.375742199952142e-06, "loss": 0.4162, "step": 6899 }, { "epoch": 0.3490313116495523, "grad_norm": 9.782362937927246, "learning_rate": 7.375035944539871e-06, "loss": 0.4822, "step": 6900 }, { "epoch": 0.34908189589761746, "grad_norm": 7.343967437744141, "learning_rate": 7.374329627928661e-06, "loss": 0.5204, "step": 6901 }, { "epoch": 0.34913248014568266, "grad_norm": 6.832290172576904, "learning_rate": 7.373623250136712e-06, "loss": 0.4104, "step": 6902 }, { "epoch": 0.3491830643937478, "grad_norm": 7.886016845703125, "learning_rate": 7.372916811182225e-06, "loss": 0.3675, "step": 6903 }, { "epoch": 0.34923364864181294, "grad_norm": 8.833752632141113, "learning_rate": 7.372210311083403e-06, "loss": 0.42, "step": 6904 }, { "epoch": 0.3492842328898781, "grad_norm": 7.075765609741211, "learning_rate": 7.3715037498584505e-06, "loss": 0.4174, "step": 6905 }, { "epoch": 0.34933481713794323, "grad_norm": 6.759149074554443, "learning_rate": 7.370797127525576e-06, "loss": 0.4145, "step": 6906 }, { "epoch": 0.3493854013860084, "grad_norm": 5.7712178230285645, "learning_rate": 7.370090444102985e-06, "loss": 0.4231, "step": 6907 }, { "epoch": 0.3494359856340736, "grad_norm": 8.187453269958496, "learning_rate": 7.369383699608888e-06, "loss": 0.4902, "step": 6908 }, { "epoch": 0.3494865698821387, "grad_norm": 14.260992050170898, "learning_rate": 7.368676894061496e-06, "loss": 0.4407, "step": 6909 }, { "epoch": 0.34953715413020386, "grad_norm": 7.83270263671875, "learning_rate": 7.367970027479022e-06, "loss": 0.4946, "step": 6910 }, { "epoch": 0.349587738378269, "grad_norm": 5.822822093963623, "learning_rate": 7.36726309987968e-06, "loss": 0.3223, "step": 6911 }, { "epoch": 0.34963832262633415, "grad_norm": 10.341230392456055, "learning_rate": 7.366556111281686e-06, "loss": 0.5151, "step": 6912 }, { "epoch": 0.3496889068743993, "grad_norm": 10.214021682739258, "learning_rate": 7.365849061703257e-06, "loss": 0.4386, "step": 6913 }, { "epoch": 0.34973949112246444, "grad_norm": 9.044751167297363, "learning_rate": 7.365141951162614e-06, "loss": 0.4664, "step": 6914 }, { "epoch": 0.34979007537052964, "grad_norm": 9.31716537475586, "learning_rate": 7.364434779677972e-06, "loss": 0.3481, "step": 6915 }, { "epoch": 0.3498406596185948, "grad_norm": 7.5269670486450195, "learning_rate": 7.363727547267559e-06, "loss": 0.3405, "step": 6916 }, { "epoch": 0.34989124386665993, "grad_norm": 8.22316837310791, "learning_rate": 7.363020253949595e-06, "loss": 0.4668, "step": 6917 }, { "epoch": 0.3499418281147251, "grad_norm": 13.836281776428223, "learning_rate": 7.3623128997423084e-06, "loss": 0.4039, "step": 6918 }, { "epoch": 0.3499924123627902, "grad_norm": 7.592123031616211, "learning_rate": 7.361605484663924e-06, "loss": 0.4414, "step": 6919 }, { "epoch": 0.35004299661085536, "grad_norm": 7.130257606506348, "learning_rate": 7.36089800873267e-06, "loss": 0.3078, "step": 6920 }, { "epoch": 0.3500935808589205, "grad_norm": 10.828926086425781, "learning_rate": 7.360190471966777e-06, "loss": 0.4367, "step": 6921 }, { "epoch": 0.3501441651069857, "grad_norm": 14.70127010345459, "learning_rate": 7.359482874384476e-06, "loss": 0.5154, "step": 6922 }, { "epoch": 0.35019474935505085, "grad_norm": 5.870433807373047, "learning_rate": 7.358775216004002e-06, "loss": 0.4519, "step": 6923 }, { "epoch": 0.350245333603116, "grad_norm": 8.923174858093262, "learning_rate": 7.358067496843586e-06, "loss": 0.3385, "step": 6924 }, { "epoch": 0.35029591785118114, "grad_norm": 7.750060081481934, "learning_rate": 7.357359716921467e-06, "loss": 0.4711, "step": 6925 }, { "epoch": 0.3503465020992463, "grad_norm": 6.374121189117432, "learning_rate": 7.356651876255882e-06, "loss": 0.3374, "step": 6926 }, { "epoch": 0.3503970863473114, "grad_norm": 7.5253753662109375, "learning_rate": 7.35594397486507e-06, "loss": 0.4325, "step": 6927 }, { "epoch": 0.3504476705953766, "grad_norm": 6.787423610687256, "learning_rate": 7.355236012767273e-06, "loss": 0.4199, "step": 6928 }, { "epoch": 0.35049825484344177, "grad_norm": 14.426902770996094, "learning_rate": 7.354527989980732e-06, "loss": 0.4892, "step": 6929 }, { "epoch": 0.3505488390915069, "grad_norm": 5.608399868011475, "learning_rate": 7.353819906523694e-06, "loss": 0.3517, "step": 6930 }, { "epoch": 0.35059942333957206, "grad_norm": 4.528820991516113, "learning_rate": 7.353111762414401e-06, "loss": 0.3417, "step": 6931 }, { "epoch": 0.3506500075876372, "grad_norm": 6.255695819854736, "learning_rate": 7.3524035576711e-06, "loss": 0.3874, "step": 6932 }, { "epoch": 0.35070059183570235, "grad_norm": 7.633041858673096, "learning_rate": 7.351695292312044e-06, "loss": 0.4097, "step": 6933 }, { "epoch": 0.3507511760837675, "grad_norm": 9.899989128112793, "learning_rate": 7.350986966355479e-06, "loss": 0.4607, "step": 6934 }, { "epoch": 0.3508017603318327, "grad_norm": 6.208735466003418, "learning_rate": 7.350278579819659e-06, "loss": 0.3545, "step": 6935 }, { "epoch": 0.35085234457989783, "grad_norm": 9.136072158813477, "learning_rate": 7.349570132722836e-06, "loss": 0.492, "step": 6936 }, { "epoch": 0.350902928827963, "grad_norm": 16.608016967773438, "learning_rate": 7.348861625083267e-06, "loss": 0.596, "step": 6937 }, { "epoch": 0.3509535130760281, "grad_norm": 11.310152053833008, "learning_rate": 7.348153056919207e-06, "loss": 0.4604, "step": 6938 }, { "epoch": 0.35100409732409327, "grad_norm": 9.81706714630127, "learning_rate": 7.347444428248916e-06, "loss": 0.378, "step": 6939 }, { "epoch": 0.3510546815721584, "grad_norm": 5.493500232696533, "learning_rate": 7.346735739090649e-06, "loss": 0.4295, "step": 6940 }, { "epoch": 0.35110526582022356, "grad_norm": 5.069149494171143, "learning_rate": 7.3460269894626714e-06, "loss": 0.4444, "step": 6941 }, { "epoch": 0.35115585006828876, "grad_norm": 6.0371012687683105, "learning_rate": 7.345318179383247e-06, "loss": 0.3991, "step": 6942 }, { "epoch": 0.3512064343163539, "grad_norm": 6.337809085845947, "learning_rate": 7.344609308870635e-06, "loss": 0.3184, "step": 6943 }, { "epoch": 0.35125701856441904, "grad_norm": 10.326870918273926, "learning_rate": 7.343900377943106e-06, "loss": 0.4365, "step": 6944 }, { "epoch": 0.3513076028124842, "grad_norm": 10.95639419555664, "learning_rate": 7.3431913866189254e-06, "loss": 0.4561, "step": 6945 }, { "epoch": 0.35135818706054933, "grad_norm": 7.056919097900391, "learning_rate": 7.342482334916362e-06, "loss": 0.5119, "step": 6946 }, { "epoch": 0.3514087713086145, "grad_norm": 8.820734024047852, "learning_rate": 7.3417732228536875e-06, "loss": 0.365, "step": 6947 }, { "epoch": 0.3514593555566797, "grad_norm": 9.0468168258667, "learning_rate": 7.341064050449173e-06, "loss": 0.4562, "step": 6948 }, { "epoch": 0.3515099398047448, "grad_norm": 15.080588340759277, "learning_rate": 7.340354817721093e-06, "loss": 0.4964, "step": 6949 }, { "epoch": 0.35156052405280996, "grad_norm": 12.368219375610352, "learning_rate": 7.339645524687721e-06, "loss": 0.5101, "step": 6950 }, { "epoch": 0.3516111083008751, "grad_norm": 5.879303932189941, "learning_rate": 7.338936171367337e-06, "loss": 0.4092, "step": 6951 }, { "epoch": 0.35166169254894025, "grad_norm": 15.005956649780273, "learning_rate": 7.3382267577782165e-06, "loss": 0.5184, "step": 6952 }, { "epoch": 0.3517122767970054, "grad_norm": 9.513370513916016, "learning_rate": 7.33751728393864e-06, "loss": 0.3709, "step": 6953 }, { "epoch": 0.35176286104507054, "grad_norm": 9.69373893737793, "learning_rate": 7.336807749866888e-06, "loss": 0.44, "step": 6954 }, { "epoch": 0.35181344529313574, "grad_norm": 6.513266563415527, "learning_rate": 7.336098155581246e-06, "loss": 0.4289, "step": 6955 }, { "epoch": 0.3518640295412009, "grad_norm": 8.10660171508789, "learning_rate": 7.335388501099997e-06, "loss": 0.508, "step": 6956 }, { "epoch": 0.35191461378926603, "grad_norm": 12.027822494506836, "learning_rate": 7.334678786441427e-06, "loss": 0.4604, "step": 6957 }, { "epoch": 0.3519651980373312, "grad_norm": 6.596252918243408, "learning_rate": 7.333969011623825e-06, "loss": 0.4335, "step": 6958 }, { "epoch": 0.3520157822853963, "grad_norm": 19.252050399780273, "learning_rate": 7.333259176665477e-06, "loss": 0.5474, "step": 6959 }, { "epoch": 0.35206636653346146, "grad_norm": 6.604135513305664, "learning_rate": 7.332549281584676e-06, "loss": 0.4235, "step": 6960 }, { "epoch": 0.3521169507815266, "grad_norm": 6.822399139404297, "learning_rate": 7.331839326399715e-06, "loss": 0.4272, "step": 6961 }, { "epoch": 0.3521675350295918, "grad_norm": 7.451125144958496, "learning_rate": 7.331129311128887e-06, "loss": 0.4624, "step": 6962 }, { "epoch": 0.35221811927765695, "grad_norm": 13.75860595703125, "learning_rate": 7.3304192357904865e-06, "loss": 0.5277, "step": 6963 }, { "epoch": 0.3522687035257221, "grad_norm": 8.332832336425781, "learning_rate": 7.329709100402812e-06, "loss": 0.5392, "step": 6964 }, { "epoch": 0.35231928777378724, "grad_norm": 11.290595054626465, "learning_rate": 7.3289989049841615e-06, "loss": 0.3406, "step": 6965 }, { "epoch": 0.3523698720218524, "grad_norm": 7.694138526916504, "learning_rate": 7.3282886495528334e-06, "loss": 0.3498, "step": 6966 }, { "epoch": 0.3524204562699175, "grad_norm": 8.034457206726074, "learning_rate": 7.3275783341271315e-06, "loss": 0.3423, "step": 6967 }, { "epoch": 0.3524710405179827, "grad_norm": 7.871248722076416, "learning_rate": 7.326867958725358e-06, "loss": 0.4875, "step": 6968 }, { "epoch": 0.35252162476604787, "grad_norm": 8.81215763092041, "learning_rate": 7.326157523365817e-06, "loss": 0.4557, "step": 6969 }, { "epoch": 0.352572209014113, "grad_norm": 11.862058639526367, "learning_rate": 7.325447028066816e-06, "loss": 0.6178, "step": 6970 }, { "epoch": 0.35262279326217816, "grad_norm": 11.029802322387695, "learning_rate": 7.324736472846661e-06, "loss": 0.5306, "step": 6971 }, { "epoch": 0.3526733775102433, "grad_norm": 12.566083908081055, "learning_rate": 7.324025857723662e-06, "loss": 0.599, "step": 6972 }, { "epoch": 0.35272396175830845, "grad_norm": 5.158892631530762, "learning_rate": 7.323315182716131e-06, "loss": 0.374, "step": 6973 }, { "epoch": 0.3527745460063736, "grad_norm": 12.902929306030273, "learning_rate": 7.322604447842378e-06, "loss": 0.4663, "step": 6974 }, { "epoch": 0.3528251302544388, "grad_norm": 6.220435619354248, "learning_rate": 7.32189365312072e-06, "loss": 0.472, "step": 6975 }, { "epoch": 0.35287571450250393, "grad_norm": 6.6475348472595215, "learning_rate": 7.32118279856947e-06, "loss": 0.4949, "step": 6976 }, { "epoch": 0.3529262987505691, "grad_norm": 6.097779750823975, "learning_rate": 7.320471884206948e-06, "loss": 0.3838, "step": 6977 }, { "epoch": 0.3529768829986342, "grad_norm": 10.589828491210938, "learning_rate": 7.319760910051468e-06, "loss": 0.6013, "step": 6978 }, { "epoch": 0.35302746724669937, "grad_norm": 5.423465728759766, "learning_rate": 7.319049876121353e-06, "loss": 0.2802, "step": 6979 }, { "epoch": 0.3530780514947645, "grad_norm": 5.645786285400391, "learning_rate": 7.318338782434925e-06, "loss": 0.4451, "step": 6980 }, { "epoch": 0.35312863574282966, "grad_norm": 6.239836692810059, "learning_rate": 7.317627629010505e-06, "loss": 0.3974, "step": 6981 }, { "epoch": 0.35317921999089485, "grad_norm": 9.224185943603516, "learning_rate": 7.31691641586642e-06, "loss": 0.426, "step": 6982 }, { "epoch": 0.35322980423896, "grad_norm": 7.0979743003845215, "learning_rate": 7.316205143020995e-06, "loss": 0.394, "step": 6983 }, { "epoch": 0.35328038848702514, "grad_norm": 5.592524528503418, "learning_rate": 7.315493810492558e-06, "loss": 0.3101, "step": 6984 }, { "epoch": 0.3533309727350903, "grad_norm": 5.862204074859619, "learning_rate": 7.314782418299437e-06, "loss": 0.3699, "step": 6985 }, { "epoch": 0.35338155698315543, "grad_norm": 6.767961025238037, "learning_rate": 7.314070966459966e-06, "loss": 0.4373, "step": 6986 }, { "epoch": 0.3534321412312206, "grad_norm": 13.661718368530273, "learning_rate": 7.313359454992476e-06, "loss": 0.5615, "step": 6987 }, { "epoch": 0.3534827254792858, "grad_norm": 11.141369819641113, "learning_rate": 7.312647883915299e-06, "loss": 0.5168, "step": 6988 }, { "epoch": 0.3535333097273509, "grad_norm": 7.699002265930176, "learning_rate": 7.311936253246773e-06, "loss": 0.4216, "step": 6989 }, { "epoch": 0.35358389397541606, "grad_norm": 7.17071008682251, "learning_rate": 7.3112245630052345e-06, "loss": 0.4376, "step": 6990 }, { "epoch": 0.3536344782234812, "grad_norm": 4.259659767150879, "learning_rate": 7.31051281320902e-06, "loss": 0.2895, "step": 6991 }, { "epoch": 0.35368506247154635, "grad_norm": 24.17083168029785, "learning_rate": 7.309801003876471e-06, "loss": 0.5037, "step": 6992 }, { "epoch": 0.3537356467196115, "grad_norm": 7.573979377746582, "learning_rate": 7.309089135025931e-06, "loss": 0.4137, "step": 6993 }, { "epoch": 0.35378623096767664, "grad_norm": 6.283808708190918, "learning_rate": 7.30837720667574e-06, "loss": 0.4271, "step": 6994 }, { "epoch": 0.35383681521574184, "grad_norm": 6.465400695800781, "learning_rate": 7.307665218844245e-06, "loss": 0.3835, "step": 6995 }, { "epoch": 0.353887399463807, "grad_norm": 9.573286056518555, "learning_rate": 7.306953171549791e-06, "loss": 0.4328, "step": 6996 }, { "epoch": 0.35393798371187213, "grad_norm": 9.998456001281738, "learning_rate": 7.306241064810725e-06, "loss": 0.4397, "step": 6997 }, { "epoch": 0.3539885679599373, "grad_norm": 6.4700188636779785, "learning_rate": 7.305528898645398e-06, "loss": 0.4113, "step": 6998 }, { "epoch": 0.3540391522080024, "grad_norm": 7.153697490692139, "learning_rate": 7.30481667307216e-06, "loss": 0.4276, "step": 6999 }, { "epoch": 0.35408973645606756, "grad_norm": 7.144070625305176, "learning_rate": 7.304104388109363e-06, "loss": 0.4514, "step": 7000 }, { "epoch": 0.35414032070413276, "grad_norm": 10.132244110107422, "learning_rate": 7.3033920437753615e-06, "loss": 0.4518, "step": 7001 }, { "epoch": 0.3541909049521979, "grad_norm": 6.301732540130615, "learning_rate": 7.302679640088512e-06, "loss": 0.4839, "step": 7002 }, { "epoch": 0.35424148920026305, "grad_norm": 6.525094985961914, "learning_rate": 7.301967177067168e-06, "loss": 0.4435, "step": 7003 }, { "epoch": 0.3542920734483282, "grad_norm": 8.63839340209961, "learning_rate": 7.30125465472969e-06, "loss": 0.5822, "step": 7004 }, { "epoch": 0.35434265769639334, "grad_norm": 6.159735202789307, "learning_rate": 7.3005420730944385e-06, "loss": 0.3434, "step": 7005 }, { "epoch": 0.3543932419444585, "grad_norm": 8.969953536987305, "learning_rate": 7.299829432179776e-06, "loss": 0.4143, "step": 7006 }, { "epoch": 0.3544438261925236, "grad_norm": 6.381217956542969, "learning_rate": 7.299116732004063e-06, "loss": 0.3491, "step": 7007 }, { "epoch": 0.3544944104405888, "grad_norm": 8.258130073547363, "learning_rate": 7.298403972585665e-06, "loss": 0.5171, "step": 7008 }, { "epoch": 0.35454499468865397, "grad_norm": 9.38792610168457, "learning_rate": 7.297691153942949e-06, "loss": 0.3524, "step": 7009 }, { "epoch": 0.3545955789367191, "grad_norm": 16.47796058654785, "learning_rate": 7.29697827609428e-06, "loss": 0.5789, "step": 7010 }, { "epoch": 0.35464616318478426, "grad_norm": 8.367876052856445, "learning_rate": 7.29626533905803e-06, "loss": 0.4056, "step": 7011 }, { "epoch": 0.3546967474328494, "grad_norm": 8.940905570983887, "learning_rate": 7.2955523428525685e-06, "loss": 0.489, "step": 7012 }, { "epoch": 0.35474733168091455, "grad_norm": 6.701961517333984, "learning_rate": 7.294839287496268e-06, "loss": 0.4381, "step": 7013 }, { "epoch": 0.3547979159289797, "grad_norm": 8.261434555053711, "learning_rate": 7.294126173007501e-06, "loss": 0.4218, "step": 7014 }, { "epoch": 0.3548485001770449, "grad_norm": 9.47826862335205, "learning_rate": 7.293412999404647e-06, "loss": 0.4534, "step": 7015 }, { "epoch": 0.35489908442511003, "grad_norm": 6.815034866333008, "learning_rate": 7.292699766706076e-06, "loss": 0.4115, "step": 7016 }, { "epoch": 0.3549496686731752, "grad_norm": 8.580743789672852, "learning_rate": 7.291986474930171e-06, "loss": 0.3639, "step": 7017 }, { "epoch": 0.3550002529212403, "grad_norm": 7.536349296569824, "learning_rate": 7.29127312409531e-06, "loss": 0.4246, "step": 7018 }, { "epoch": 0.35505083716930547, "grad_norm": 6.037561416625977, "learning_rate": 7.290559714219876e-06, "loss": 0.4172, "step": 7019 }, { "epoch": 0.3551014214173706, "grad_norm": 6.2804694175720215, "learning_rate": 7.2898462453222495e-06, "loss": 0.4501, "step": 7020 }, { "epoch": 0.3551520056654358, "grad_norm": 4.82670259475708, "learning_rate": 7.289132717420818e-06, "loss": 0.4699, "step": 7021 }, { "epoch": 0.35520258991350095, "grad_norm": 9.705352783203125, "learning_rate": 7.288419130533965e-06, "loss": 0.4225, "step": 7022 }, { "epoch": 0.3552531741615661, "grad_norm": 24.637990951538086, "learning_rate": 7.287705484680077e-06, "loss": 0.3881, "step": 7023 }, { "epoch": 0.35530375840963124, "grad_norm": 5.5791916847229, "learning_rate": 7.286991779877545e-06, "loss": 0.4017, "step": 7024 }, { "epoch": 0.3553543426576964, "grad_norm": 5.627163410186768, "learning_rate": 7.2862780161447595e-06, "loss": 0.4344, "step": 7025 }, { "epoch": 0.35540492690576153, "grad_norm": 10.027948379516602, "learning_rate": 7.285564193500112e-06, "loss": 0.4157, "step": 7026 }, { "epoch": 0.3554555111538267, "grad_norm": 5.909028053283691, "learning_rate": 7.284850311961995e-06, "loss": 0.4228, "step": 7027 }, { "epoch": 0.3555060954018919, "grad_norm": 5.071579456329346, "learning_rate": 7.2841363715488055e-06, "loss": 0.351, "step": 7028 }, { "epoch": 0.355556679649957, "grad_norm": 9.927608489990234, "learning_rate": 7.283422372278938e-06, "loss": 0.5172, "step": 7029 }, { "epoch": 0.35560726389802216, "grad_norm": 11.244673728942871, "learning_rate": 7.28270831417079e-06, "loss": 0.4672, "step": 7030 }, { "epoch": 0.3556578481460873, "grad_norm": 9.85486888885498, "learning_rate": 7.281994197242765e-06, "loss": 0.4985, "step": 7031 }, { "epoch": 0.35570843239415245, "grad_norm": 12.836980819702148, "learning_rate": 7.28128002151326e-06, "loss": 0.5389, "step": 7032 }, { "epoch": 0.3557590166422176, "grad_norm": 16.203523635864258, "learning_rate": 7.2805657870006775e-06, "loss": 0.5029, "step": 7033 }, { "epoch": 0.35580960089028274, "grad_norm": 17.07863426208496, "learning_rate": 7.279851493723427e-06, "loss": 0.5913, "step": 7034 }, { "epoch": 0.35586018513834794, "grad_norm": 6.866184711456299, "learning_rate": 7.279137141699908e-06, "loss": 0.3421, "step": 7035 }, { "epoch": 0.3559107693864131, "grad_norm": 7.575542449951172, "learning_rate": 7.278422730948529e-06, "loss": 0.3757, "step": 7036 }, { "epoch": 0.35596135363447823, "grad_norm": 4.828701972961426, "learning_rate": 7.2777082614877e-06, "loss": 0.357, "step": 7037 }, { "epoch": 0.3560119378825434, "grad_norm": 7.839620113372803, "learning_rate": 7.276993733335831e-06, "loss": 0.4406, "step": 7038 }, { "epoch": 0.3560625221306085, "grad_norm": 5.1800994873046875, "learning_rate": 7.276279146511334e-06, "loss": 0.3292, "step": 7039 }, { "epoch": 0.35611310637867366, "grad_norm": 7.122854232788086, "learning_rate": 7.275564501032621e-06, "loss": 0.3077, "step": 7040 }, { "epoch": 0.35616369062673886, "grad_norm": 9.512986183166504, "learning_rate": 7.2748497969181075e-06, "loss": 0.3759, "step": 7041 }, { "epoch": 0.356214274874804, "grad_norm": 10.03920841217041, "learning_rate": 7.274135034186209e-06, "loss": 0.4035, "step": 7042 }, { "epoch": 0.35626485912286915, "grad_norm": 6.520224571228027, "learning_rate": 7.273420212855345e-06, "loss": 0.385, "step": 7043 }, { "epoch": 0.3563154433709343, "grad_norm": 8.104974746704102, "learning_rate": 7.272705332943931e-06, "loss": 0.414, "step": 7044 }, { "epoch": 0.35636602761899944, "grad_norm": 6.322412014007568, "learning_rate": 7.271990394470391e-06, "loss": 0.3905, "step": 7045 }, { "epoch": 0.3564166118670646, "grad_norm": 16.069059371948242, "learning_rate": 7.271275397453147e-06, "loss": 0.561, "step": 7046 }, { "epoch": 0.3564671961151297, "grad_norm": 8.561039924621582, "learning_rate": 7.270560341910622e-06, "loss": 0.4399, "step": 7047 }, { "epoch": 0.3565177803631949, "grad_norm": 11.702330589294434, "learning_rate": 7.2698452278612395e-06, "loss": 0.4722, "step": 7048 }, { "epoch": 0.35656836461126007, "grad_norm": 5.463146209716797, "learning_rate": 7.26913005532343e-06, "loss": 0.4213, "step": 7049 }, { "epoch": 0.3566189488593252, "grad_norm": 7.244116306304932, "learning_rate": 7.268414824315619e-06, "loss": 0.4035, "step": 7050 }, { "epoch": 0.35666953310739036, "grad_norm": 4.912014007568359, "learning_rate": 7.267699534856238e-06, "loss": 0.3428, "step": 7051 }, { "epoch": 0.3567201173554555, "grad_norm": 7.5218939781188965, "learning_rate": 7.266984186963714e-06, "loss": 0.4723, "step": 7052 }, { "epoch": 0.35677070160352065, "grad_norm": 6.841125011444092, "learning_rate": 7.266268780656489e-06, "loss": 0.4645, "step": 7053 }, { "epoch": 0.3568212858515858, "grad_norm": 7.3731913566589355, "learning_rate": 7.265553315952988e-06, "loss": 0.3978, "step": 7054 }, { "epoch": 0.356871870099651, "grad_norm": 12.082215309143066, "learning_rate": 7.26483779287165e-06, "loss": 0.5338, "step": 7055 }, { "epoch": 0.35692245434771613, "grad_norm": 8.430413246154785, "learning_rate": 7.264122211430914e-06, "loss": 0.4498, "step": 7056 }, { "epoch": 0.3569730385957813, "grad_norm": 9.021547317504883, "learning_rate": 7.263406571649216e-06, "loss": 0.4679, "step": 7057 }, { "epoch": 0.3570236228438464, "grad_norm": 8.773797988891602, "learning_rate": 7.262690873544997e-06, "loss": 0.3883, "step": 7058 }, { "epoch": 0.35707420709191157, "grad_norm": 9.396317481994629, "learning_rate": 7.261975117136701e-06, "loss": 0.4258, "step": 7059 }, { "epoch": 0.3571247913399767, "grad_norm": 6.573166370391846, "learning_rate": 7.2612593024427714e-06, "loss": 0.4552, "step": 7060 }, { "epoch": 0.3571753755880419, "grad_norm": 7.3658766746521, "learning_rate": 7.260543429481649e-06, "loss": 0.3367, "step": 7061 }, { "epoch": 0.35722595983610705, "grad_norm": 11.0278959274292, "learning_rate": 7.259827498271783e-06, "loss": 0.4378, "step": 7062 }, { "epoch": 0.3572765440841722, "grad_norm": 9.428013801574707, "learning_rate": 7.259111508831621e-06, "loss": 0.5, "step": 7063 }, { "epoch": 0.35732712833223734, "grad_norm": 6.754676818847656, "learning_rate": 7.258395461179612e-06, "loss": 0.4976, "step": 7064 }, { "epoch": 0.3573777125803025, "grad_norm": 6.452877521514893, "learning_rate": 7.257679355334206e-06, "loss": 0.4437, "step": 7065 }, { "epoch": 0.35742829682836763, "grad_norm": 14.582465171813965, "learning_rate": 7.256963191313858e-06, "loss": 0.5357, "step": 7066 }, { "epoch": 0.3574788810764328, "grad_norm": 5.851658821105957, "learning_rate": 7.256246969137018e-06, "loss": 0.4739, "step": 7067 }, { "epoch": 0.357529465324498, "grad_norm": 9.043363571166992, "learning_rate": 7.255530688822145e-06, "loss": 0.5457, "step": 7068 }, { "epoch": 0.3575800495725631, "grad_norm": 11.8287935256958, "learning_rate": 7.254814350387693e-06, "loss": 0.5273, "step": 7069 }, { "epoch": 0.35763063382062826, "grad_norm": 6.769567966461182, "learning_rate": 7.254097953852123e-06, "loss": 0.5007, "step": 7070 }, { "epoch": 0.3576812180686934, "grad_norm": 8.647875785827637, "learning_rate": 7.253381499233891e-06, "loss": 0.4451, "step": 7071 }, { "epoch": 0.35773180231675855, "grad_norm": 9.140161514282227, "learning_rate": 7.252664986551463e-06, "loss": 0.4429, "step": 7072 }, { "epoch": 0.3577823865648237, "grad_norm": 9.307234764099121, "learning_rate": 7.251948415823299e-06, "loss": 0.4897, "step": 7073 }, { "epoch": 0.35783297081288884, "grad_norm": 8.294917106628418, "learning_rate": 7.251231787067863e-06, "loss": 0.3747, "step": 7074 }, { "epoch": 0.35788355506095404, "grad_norm": 10.108695030212402, "learning_rate": 7.250515100303621e-06, "loss": 0.4521, "step": 7075 }, { "epoch": 0.3579341393090192, "grad_norm": 7.162308216094971, "learning_rate": 7.249798355549041e-06, "loss": 0.447, "step": 7076 }, { "epoch": 0.35798472355708433, "grad_norm": 6.865579128265381, "learning_rate": 7.249081552822592e-06, "loss": 0.4323, "step": 7077 }, { "epoch": 0.35803530780514947, "grad_norm": 6.792383193969727, "learning_rate": 7.248364692142744e-06, "loss": 0.4961, "step": 7078 }, { "epoch": 0.3580858920532146, "grad_norm": 13.529860496520996, "learning_rate": 7.247647773527969e-06, "loss": 0.5398, "step": 7079 }, { "epoch": 0.35813647630127976, "grad_norm": 9.21080207824707, "learning_rate": 7.246930796996739e-06, "loss": 0.4493, "step": 7080 }, { "epoch": 0.35818706054934496, "grad_norm": 12.566460609436035, "learning_rate": 7.246213762567529e-06, "loss": 0.4729, "step": 7081 }, { "epoch": 0.3582376447974101, "grad_norm": 11.965104103088379, "learning_rate": 7.2454966702588155e-06, "loss": 0.3876, "step": 7082 }, { "epoch": 0.35828822904547525, "grad_norm": 9.456819534301758, "learning_rate": 7.244779520089078e-06, "loss": 0.4198, "step": 7083 }, { "epoch": 0.3583388132935404, "grad_norm": 10.953750610351562, "learning_rate": 7.244062312076794e-06, "loss": 0.5026, "step": 7084 }, { "epoch": 0.35838939754160554, "grad_norm": 6.588737487792969, "learning_rate": 7.243345046240445e-06, "loss": 0.4528, "step": 7085 }, { "epoch": 0.3584399817896707, "grad_norm": 6.6329665184021, "learning_rate": 7.242627722598512e-06, "loss": 0.3881, "step": 7086 }, { "epoch": 0.3584905660377358, "grad_norm": 6.972047805786133, "learning_rate": 7.241910341169479e-06, "loss": 0.4449, "step": 7087 }, { "epoch": 0.358541150285801, "grad_norm": 7.266571044921875, "learning_rate": 7.241192901971833e-06, "loss": 0.4902, "step": 7088 }, { "epoch": 0.35859173453386617, "grad_norm": 10.339115142822266, "learning_rate": 7.2404754050240574e-06, "loss": 0.5384, "step": 7089 }, { "epoch": 0.3586423187819313, "grad_norm": 6.454623699188232, "learning_rate": 7.239757850344643e-06, "loss": 0.4474, "step": 7090 }, { "epoch": 0.35869290302999646, "grad_norm": 10.981416702270508, "learning_rate": 7.23904023795208e-06, "loss": 0.5335, "step": 7091 }, { "epoch": 0.3587434872780616, "grad_norm": 6.602825164794922, "learning_rate": 7.238322567864857e-06, "loss": 0.422, "step": 7092 }, { "epoch": 0.35879407152612675, "grad_norm": 8.13778018951416, "learning_rate": 7.237604840101468e-06, "loss": 0.4006, "step": 7093 }, { "epoch": 0.3588446557741919, "grad_norm": 7.320159435272217, "learning_rate": 7.236887054680408e-06, "loss": 0.3708, "step": 7094 }, { "epoch": 0.3588952400222571, "grad_norm": 7.246602535247803, "learning_rate": 7.236169211620171e-06, "loss": 0.4248, "step": 7095 }, { "epoch": 0.35894582427032223, "grad_norm": 6.511691093444824, "learning_rate": 7.235451310939253e-06, "loss": 0.3717, "step": 7096 }, { "epoch": 0.3589964085183874, "grad_norm": 7.556926250457764, "learning_rate": 7.234733352656157e-06, "loss": 0.4197, "step": 7097 }, { "epoch": 0.3590469927664525, "grad_norm": 11.132746696472168, "learning_rate": 7.234015336789378e-06, "loss": 0.5152, "step": 7098 }, { "epoch": 0.35909757701451767, "grad_norm": 9.42238712310791, "learning_rate": 7.233297263357421e-06, "loss": 0.5032, "step": 7099 }, { "epoch": 0.3591481612625828, "grad_norm": 9.551023483276367, "learning_rate": 7.232579132378789e-06, "loss": 0.4536, "step": 7100 }, { "epoch": 0.359198745510648, "grad_norm": 13.794462203979492, "learning_rate": 7.231860943871984e-06, "loss": 0.4692, "step": 7101 }, { "epoch": 0.35924932975871315, "grad_norm": 8.613757133483887, "learning_rate": 7.231142697855511e-06, "loss": 0.3322, "step": 7102 }, { "epoch": 0.3592999140067783, "grad_norm": 23.179256439208984, "learning_rate": 7.230424394347884e-06, "loss": 0.5138, "step": 7103 }, { "epoch": 0.35935049825484344, "grad_norm": 6.039730072021484, "learning_rate": 7.229706033367605e-06, "loss": 0.4603, "step": 7104 }, { "epoch": 0.3594010825029086, "grad_norm": 4.961038112640381, "learning_rate": 7.228987614933187e-06, "loss": 0.3594, "step": 7105 }, { "epoch": 0.35945166675097373, "grad_norm": 4.634253978729248, "learning_rate": 7.228269139063144e-06, "loss": 0.4038, "step": 7106 }, { "epoch": 0.3595022509990389, "grad_norm": 6.894516944885254, "learning_rate": 7.227550605775986e-06, "loss": 0.4012, "step": 7107 }, { "epoch": 0.3595528352471041, "grad_norm": 7.3035712242126465, "learning_rate": 7.226832015090229e-06, "loss": 0.5075, "step": 7108 }, { "epoch": 0.3596034194951692, "grad_norm": 8.120475769042969, "learning_rate": 7.2261133670243895e-06, "loss": 0.4211, "step": 7109 }, { "epoch": 0.35965400374323436, "grad_norm": 5.738796234130859, "learning_rate": 7.2253946615969874e-06, "loss": 0.3167, "step": 7110 }, { "epoch": 0.3597045879912995, "grad_norm": 7.727759838104248, "learning_rate": 7.22467589882654e-06, "loss": 0.5174, "step": 7111 }, { "epoch": 0.35975517223936465, "grad_norm": 12.574828147888184, "learning_rate": 7.223957078731567e-06, "loss": 0.4004, "step": 7112 }, { "epoch": 0.3598057564874298, "grad_norm": 6.494752407073975, "learning_rate": 7.2232382013305924e-06, "loss": 0.5053, "step": 7113 }, { "epoch": 0.35985634073549494, "grad_norm": 7.566823482513428, "learning_rate": 7.222519266642138e-06, "loss": 0.4588, "step": 7114 }, { "epoch": 0.35990692498356014, "grad_norm": 10.374892234802246, "learning_rate": 7.221800274684731e-06, "loss": 0.363, "step": 7115 }, { "epoch": 0.3599575092316253, "grad_norm": 8.730291366577148, "learning_rate": 7.221081225476899e-06, "loss": 0.415, "step": 7116 }, { "epoch": 0.3600080934796904, "grad_norm": 8.94627571105957, "learning_rate": 7.220362119037168e-06, "loss": 0.5504, "step": 7117 }, { "epoch": 0.36005867772775557, "grad_norm": 9.225939750671387, "learning_rate": 7.219642955384067e-06, "loss": 0.4212, "step": 7118 }, { "epoch": 0.3601092619758207, "grad_norm": 16.53221893310547, "learning_rate": 7.21892373453613e-06, "loss": 0.5253, "step": 7119 }, { "epoch": 0.36015984622388586, "grad_norm": 6.7428131103515625, "learning_rate": 7.2182044565118856e-06, "loss": 0.3713, "step": 7120 }, { "epoch": 0.36021043047195106, "grad_norm": 8.041769027709961, "learning_rate": 7.217485121329872e-06, "loss": 0.4282, "step": 7121 }, { "epoch": 0.3602610147200162, "grad_norm": 7.838534355163574, "learning_rate": 7.216765729008623e-06, "loss": 0.3643, "step": 7122 }, { "epoch": 0.36031159896808135, "grad_norm": 7.889418125152588, "learning_rate": 7.216046279566675e-06, "loss": 0.4303, "step": 7123 }, { "epoch": 0.3603621832161465, "grad_norm": 36.128475189208984, "learning_rate": 7.215326773022567e-06, "loss": 0.5392, "step": 7124 }, { "epoch": 0.36041276746421164, "grad_norm": 6.5951385498046875, "learning_rate": 7.214607209394839e-06, "loss": 0.3967, "step": 7125 }, { "epoch": 0.3604633517122768, "grad_norm": 6.640763282775879, "learning_rate": 7.2138875887020336e-06, "loss": 0.3081, "step": 7126 }, { "epoch": 0.3605139359603419, "grad_norm": 5.562182426452637, "learning_rate": 7.213167910962691e-06, "loss": 0.4339, "step": 7127 }, { "epoch": 0.3605645202084071, "grad_norm": 10.802045822143555, "learning_rate": 7.212448176195357e-06, "loss": 0.5196, "step": 7128 }, { "epoch": 0.36061510445647227, "grad_norm": 8.87634563446045, "learning_rate": 7.211728384418578e-06, "loss": 0.4288, "step": 7129 }, { "epoch": 0.3606656887045374, "grad_norm": 5.548431396484375, "learning_rate": 7.211008535650899e-06, "loss": 0.3421, "step": 7130 }, { "epoch": 0.36071627295260256, "grad_norm": 8.804852485656738, "learning_rate": 7.210288629910873e-06, "loss": 0.5408, "step": 7131 }, { "epoch": 0.3607668572006677, "grad_norm": 5.494693756103516, "learning_rate": 7.209568667217048e-06, "loss": 0.4726, "step": 7132 }, { "epoch": 0.36081744144873285, "grad_norm": 5.670834064483643, "learning_rate": 7.208848647587974e-06, "loss": 0.3941, "step": 7133 }, { "epoch": 0.360868025696798, "grad_norm": 16.00606346130371, "learning_rate": 7.208128571042205e-06, "loss": 0.4411, "step": 7134 }, { "epoch": 0.3609186099448632, "grad_norm": 27.61598777770996, "learning_rate": 7.207408437598298e-06, "loss": 0.4485, "step": 7135 }, { "epoch": 0.36096919419292833, "grad_norm": 11.589865684509277, "learning_rate": 7.206688247274806e-06, "loss": 0.5544, "step": 7136 }, { "epoch": 0.3610197784409935, "grad_norm": 10.750090599060059, "learning_rate": 7.205968000090289e-06, "loss": 0.5053, "step": 7137 }, { "epoch": 0.3610703626890586, "grad_norm": 7.378964900970459, "learning_rate": 7.205247696063304e-06, "loss": 0.4446, "step": 7138 }, { "epoch": 0.36112094693712377, "grad_norm": 8.10864543914795, "learning_rate": 7.204527335212412e-06, "loss": 0.5406, "step": 7139 }, { "epoch": 0.3611715311851889, "grad_norm": 9.111123085021973, "learning_rate": 7.203806917556176e-06, "loss": 0.3644, "step": 7140 }, { "epoch": 0.3612221154332541, "grad_norm": 10.271866798400879, "learning_rate": 7.20308644311316e-06, "loss": 0.4832, "step": 7141 }, { "epoch": 0.36127269968131925, "grad_norm": 7.77178955078125, "learning_rate": 7.202365911901926e-06, "loss": 0.4693, "step": 7142 }, { "epoch": 0.3613232839293844, "grad_norm": 6.630972862243652, "learning_rate": 7.201645323941043e-06, "loss": 0.4094, "step": 7143 }, { "epoch": 0.36137386817744954, "grad_norm": 5.76791524887085, "learning_rate": 7.200924679249077e-06, "loss": 0.4441, "step": 7144 }, { "epoch": 0.3614244524255147, "grad_norm": 5.343783855438232, "learning_rate": 7.2002039778446e-06, "loss": 0.4212, "step": 7145 }, { "epoch": 0.36147503667357983, "grad_norm": 7.614032745361328, "learning_rate": 7.199483219746178e-06, "loss": 0.4182, "step": 7146 }, { "epoch": 0.361525620921645, "grad_norm": 8.097485542297363, "learning_rate": 7.198762404972388e-06, "loss": 0.3938, "step": 7147 }, { "epoch": 0.3615762051697102, "grad_norm": 15.200104713439941, "learning_rate": 7.198041533541802e-06, "loss": 0.505, "step": 7148 }, { "epoch": 0.3616267894177753, "grad_norm": 11.009583473205566, "learning_rate": 7.197320605472992e-06, "loss": 0.4497, "step": 7149 }, { "epoch": 0.36167737366584046, "grad_norm": 4.682107925415039, "learning_rate": 7.196599620784541e-06, "loss": 0.382, "step": 7150 }, { "epoch": 0.3617279579139056, "grad_norm": 8.77627944946289, "learning_rate": 7.1958785794950225e-06, "loss": 0.5645, "step": 7151 }, { "epoch": 0.36177854216197075, "grad_norm": 4.788859844207764, "learning_rate": 7.195157481623016e-06, "loss": 0.4206, "step": 7152 }, { "epoch": 0.3618291264100359, "grad_norm": 7.114739418029785, "learning_rate": 7.194436327187104e-06, "loss": 0.4464, "step": 7153 }, { "epoch": 0.3618797106581011, "grad_norm": 8.133009910583496, "learning_rate": 7.1937151162058684e-06, "loss": 0.4346, "step": 7154 }, { "epoch": 0.36193029490616624, "grad_norm": 14.622722625732422, "learning_rate": 7.192993848697893e-06, "loss": 0.6273, "step": 7155 }, { "epoch": 0.3619808791542314, "grad_norm": 15.783268928527832, "learning_rate": 7.192272524681763e-06, "loss": 0.5971, "step": 7156 }, { "epoch": 0.3620314634022965, "grad_norm": 7.253253936767578, "learning_rate": 7.191551144176066e-06, "loss": 0.4241, "step": 7157 }, { "epoch": 0.36208204765036167, "grad_norm": 5.395839691162109, "learning_rate": 7.1908297071993874e-06, "loss": 0.3167, "step": 7158 }, { "epoch": 0.3621326318984268, "grad_norm": 5.497884273529053, "learning_rate": 7.190108213770319e-06, "loss": 0.4029, "step": 7159 }, { "epoch": 0.36218321614649196, "grad_norm": 13.053048133850098, "learning_rate": 7.189386663907454e-06, "loss": 0.4817, "step": 7160 }, { "epoch": 0.36223380039455716, "grad_norm": 6.963030815124512, "learning_rate": 7.18866505762938e-06, "loss": 0.4883, "step": 7161 }, { "epoch": 0.3622843846426223, "grad_norm": 5.92717170715332, "learning_rate": 7.187943394954696e-06, "loss": 0.4455, "step": 7162 }, { "epoch": 0.36233496889068745, "grad_norm": 6.084280014038086, "learning_rate": 7.1872216759019964e-06, "loss": 0.4848, "step": 7163 }, { "epoch": 0.3623855531387526, "grad_norm": 6.776468276977539, "learning_rate": 7.186499900489876e-06, "loss": 0.3634, "step": 7164 }, { "epoch": 0.36243613738681774, "grad_norm": 10.370353698730469, "learning_rate": 7.185778068736935e-06, "loss": 0.4093, "step": 7165 }, { "epoch": 0.3624867216348829, "grad_norm": 6.303755760192871, "learning_rate": 7.185056180661772e-06, "loss": 0.2738, "step": 7166 }, { "epoch": 0.362537305882948, "grad_norm": 8.228899002075195, "learning_rate": 7.1843342362829896e-06, "loss": 0.384, "step": 7167 }, { "epoch": 0.3625878901310132, "grad_norm": 8.831204414367676, "learning_rate": 7.183612235619189e-06, "loss": 0.4455, "step": 7168 }, { "epoch": 0.36263847437907837, "grad_norm": 18.86136245727539, "learning_rate": 7.1828901786889775e-06, "loss": 0.7543, "step": 7169 }, { "epoch": 0.3626890586271435, "grad_norm": 5.643636703491211, "learning_rate": 7.1821680655109575e-06, "loss": 0.3453, "step": 7170 }, { "epoch": 0.36273964287520866, "grad_norm": 7.180805683135986, "learning_rate": 7.181445896103736e-06, "loss": 0.4528, "step": 7171 }, { "epoch": 0.3627902271232738, "grad_norm": 13.705498695373535, "learning_rate": 7.180723670485923e-06, "loss": 0.6028, "step": 7172 }, { "epoch": 0.36284081137133894, "grad_norm": 7.843392372131348, "learning_rate": 7.18000138867613e-06, "loss": 0.401, "step": 7173 }, { "epoch": 0.36289139561940414, "grad_norm": 6.735057830810547, "learning_rate": 7.179279050692965e-06, "loss": 0.5083, "step": 7174 }, { "epoch": 0.3629419798674693, "grad_norm": 7.863984107971191, "learning_rate": 7.178556656555043e-06, "loss": 0.408, "step": 7175 }, { "epoch": 0.36299256411553443, "grad_norm": 7.167259216308594, "learning_rate": 7.17783420628098e-06, "loss": 0.2558, "step": 7176 }, { "epoch": 0.3630431483635996, "grad_norm": 7.217511177062988, "learning_rate": 7.177111699889389e-06, "loss": 0.5164, "step": 7177 }, { "epoch": 0.3630937326116647, "grad_norm": 8.903929710388184, "learning_rate": 7.176389137398887e-06, "loss": 0.5113, "step": 7178 }, { "epoch": 0.36314431685972987, "grad_norm": 9.869231224060059, "learning_rate": 7.175666518828095e-06, "loss": 0.5031, "step": 7179 }, { "epoch": 0.363194901107795, "grad_norm": 7.418919563293457, "learning_rate": 7.174943844195631e-06, "loss": 0.4211, "step": 7180 }, { "epoch": 0.3632454853558602, "grad_norm": 11.150193214416504, "learning_rate": 7.174221113520117e-06, "loss": 0.4785, "step": 7181 }, { "epoch": 0.36329606960392535, "grad_norm": 11.72746753692627, "learning_rate": 7.173498326820179e-06, "loss": 0.3854, "step": 7182 }, { "epoch": 0.3633466538519905, "grad_norm": 14.669071197509766, "learning_rate": 7.172775484114437e-06, "loss": 0.4091, "step": 7183 }, { "epoch": 0.36339723810005564, "grad_norm": 11.399188995361328, "learning_rate": 7.172052585421518e-06, "loss": 0.4125, "step": 7184 }, { "epoch": 0.3634478223481208, "grad_norm": 5.781165599822998, "learning_rate": 7.171329630760053e-06, "loss": 0.4348, "step": 7185 }, { "epoch": 0.36349840659618593, "grad_norm": 11.269478797912598, "learning_rate": 7.170606620148666e-06, "loss": 0.5424, "step": 7186 }, { "epoch": 0.3635489908442511, "grad_norm": 6.716498851776123, "learning_rate": 7.16988355360599e-06, "loss": 0.3252, "step": 7187 }, { "epoch": 0.3635995750923163, "grad_norm": 5.576808452606201, "learning_rate": 7.169160431150656e-06, "loss": 0.4151, "step": 7188 }, { "epoch": 0.3636501593403814, "grad_norm": 9.57594108581543, "learning_rate": 7.168437252801297e-06, "loss": 0.4588, "step": 7189 }, { "epoch": 0.36370074358844656, "grad_norm": 12.9010591506958, "learning_rate": 7.167714018576547e-06, "loss": 0.523, "step": 7190 }, { "epoch": 0.3637513278365117, "grad_norm": 7.146794319152832, "learning_rate": 7.166990728495042e-06, "loss": 0.4077, "step": 7191 }, { "epoch": 0.36380191208457685, "grad_norm": 5.555751323699951, "learning_rate": 7.1662673825754225e-06, "loss": 0.286, "step": 7192 }, { "epoch": 0.363852496332642, "grad_norm": 4.852283000946045, "learning_rate": 7.165543980836322e-06, "loss": 0.4373, "step": 7193 }, { "epoch": 0.3639030805807072, "grad_norm": 5.39841890335083, "learning_rate": 7.164820523296387e-06, "loss": 0.3855, "step": 7194 }, { "epoch": 0.36395366482877234, "grad_norm": 11.059326171875, "learning_rate": 7.164097009974256e-06, "loss": 0.4737, "step": 7195 }, { "epoch": 0.3640042490768375, "grad_norm": 10.305266380310059, "learning_rate": 7.163373440888569e-06, "loss": 0.4936, "step": 7196 }, { "epoch": 0.3640548333249026, "grad_norm": 7.336582660675049, "learning_rate": 7.162649816057976e-06, "loss": 0.3823, "step": 7197 }, { "epoch": 0.36410541757296777, "grad_norm": 5.9673967361450195, "learning_rate": 7.16192613550112e-06, "loss": 0.4184, "step": 7198 }, { "epoch": 0.3641560018210329, "grad_norm": 8.875822067260742, "learning_rate": 7.16120239923665e-06, "loss": 0.5271, "step": 7199 }, { "epoch": 0.36420658606909806, "grad_norm": 14.665183067321777, "learning_rate": 7.160478607283213e-06, "loss": 0.5128, "step": 7200 }, { "epoch": 0.36425717031716326, "grad_norm": 5.984831809997559, "learning_rate": 7.159754759659463e-06, "loss": 0.3704, "step": 7201 }, { "epoch": 0.3643077545652284, "grad_norm": 5.845013618469238, "learning_rate": 7.159030856384049e-06, "loss": 0.4225, "step": 7202 }, { "epoch": 0.36435833881329355, "grad_norm": 6.497091770172119, "learning_rate": 7.158306897475624e-06, "loss": 0.3563, "step": 7203 }, { "epoch": 0.3644089230613587, "grad_norm": 10.562369346618652, "learning_rate": 7.157582882952844e-06, "loss": 0.3676, "step": 7204 }, { "epoch": 0.36445950730942384, "grad_norm": 10.890714645385742, "learning_rate": 7.156858812834363e-06, "loss": 0.5246, "step": 7205 }, { "epoch": 0.364510091557489, "grad_norm": 6.490866661071777, "learning_rate": 7.1561346871388415e-06, "loss": 0.4899, "step": 7206 }, { "epoch": 0.3645606758055541, "grad_norm": 6.624200820922852, "learning_rate": 7.155410505884936e-06, "loss": 0.4126, "step": 7207 }, { "epoch": 0.3646112600536193, "grad_norm": 6.41942024230957, "learning_rate": 7.15468626909131e-06, "loss": 0.4025, "step": 7208 }, { "epoch": 0.36466184430168447, "grad_norm": 5.238823413848877, "learning_rate": 7.153961976776622e-06, "loss": 0.3984, "step": 7209 }, { "epoch": 0.3647124285497496, "grad_norm": 15.946549415588379, "learning_rate": 7.153237628959535e-06, "loss": 0.4537, "step": 7210 }, { "epoch": 0.36476301279781476, "grad_norm": 10.451493263244629, "learning_rate": 7.152513225658719e-06, "loss": 0.4739, "step": 7211 }, { "epoch": 0.3648135970458799, "grad_norm": 7.012585639953613, "learning_rate": 7.151788766892833e-06, "loss": 0.4321, "step": 7212 }, { "epoch": 0.36486418129394504, "grad_norm": 9.423529624938965, "learning_rate": 7.151064252680548e-06, "loss": 0.4298, "step": 7213 }, { "epoch": 0.36491476554201024, "grad_norm": 7.331805229187012, "learning_rate": 7.150339683040536e-06, "loss": 0.4829, "step": 7214 }, { "epoch": 0.3649653497900754, "grad_norm": 8.36086654663086, "learning_rate": 7.149615057991461e-06, "loss": 0.5049, "step": 7215 }, { "epoch": 0.36501593403814053, "grad_norm": 5.498964309692383, "learning_rate": 7.148890377551999e-06, "loss": 0.4103, "step": 7216 }, { "epoch": 0.3650665182862057, "grad_norm": 12.473894119262695, "learning_rate": 7.148165641740823e-06, "loss": 0.4443, "step": 7217 }, { "epoch": 0.3651171025342708, "grad_norm": 19.052978515625, "learning_rate": 7.147440850576607e-06, "loss": 0.5296, "step": 7218 }, { "epoch": 0.36516768678233597, "grad_norm": 8.834178924560547, "learning_rate": 7.146716004078027e-06, "loss": 0.4244, "step": 7219 }, { "epoch": 0.3652182710304011, "grad_norm": 6.859770774841309, "learning_rate": 7.145991102263761e-06, "loss": 0.4222, "step": 7220 }, { "epoch": 0.3652688552784663, "grad_norm": 13.767044067382812, "learning_rate": 7.1452661451524874e-06, "loss": 0.4444, "step": 7221 }, { "epoch": 0.36531943952653145, "grad_norm": 5.290274143218994, "learning_rate": 7.144541132762887e-06, "loss": 0.4195, "step": 7222 }, { "epoch": 0.3653700237745966, "grad_norm": 7.85618257522583, "learning_rate": 7.1438160651136424e-06, "loss": 0.4659, "step": 7223 }, { "epoch": 0.36542060802266174, "grad_norm": 9.015645027160645, "learning_rate": 7.143090942223434e-06, "loss": 0.5733, "step": 7224 }, { "epoch": 0.3654711922707269, "grad_norm": 9.227677345275879, "learning_rate": 7.14236576411095e-06, "loss": 0.4414, "step": 7225 }, { "epoch": 0.36552177651879203, "grad_norm": 7.822611331939697, "learning_rate": 7.141640530794874e-06, "loss": 0.4689, "step": 7226 }, { "epoch": 0.3655723607668572, "grad_norm": 6.762156009674072, "learning_rate": 7.140915242293896e-06, "loss": 0.4159, "step": 7227 }, { "epoch": 0.3656229450149224, "grad_norm": 9.227209091186523, "learning_rate": 7.140189898626701e-06, "loss": 0.3252, "step": 7228 }, { "epoch": 0.3656735292629875, "grad_norm": 14.914130210876465, "learning_rate": 7.139464499811984e-06, "loss": 0.3871, "step": 7229 }, { "epoch": 0.36572411351105266, "grad_norm": 6.462944507598877, "learning_rate": 7.138739045868433e-06, "loss": 0.4467, "step": 7230 }, { "epoch": 0.3657746977591178, "grad_norm": 7.491351127624512, "learning_rate": 7.138013536814743e-06, "loss": 0.4568, "step": 7231 }, { "epoch": 0.36582528200718295, "grad_norm": 9.7332181930542, "learning_rate": 7.137287972669609e-06, "loss": 0.4532, "step": 7232 }, { "epoch": 0.3658758662552481, "grad_norm": 8.049653053283691, "learning_rate": 7.136562353451726e-06, "loss": 0.4594, "step": 7233 }, { "epoch": 0.3659264505033133, "grad_norm": 11.73835563659668, "learning_rate": 7.135836679179793e-06, "loss": 0.4137, "step": 7234 }, { "epoch": 0.36597703475137844, "grad_norm": 7.548786640167236, "learning_rate": 7.135110949872506e-06, "loss": 0.4172, "step": 7235 }, { "epoch": 0.3660276189994436, "grad_norm": 5.871299743652344, "learning_rate": 7.1343851655485675e-06, "loss": 0.5555, "step": 7236 }, { "epoch": 0.3660782032475087, "grad_norm": 5.324813365936279, "learning_rate": 7.133659326226679e-06, "loss": 0.3968, "step": 7237 }, { "epoch": 0.36612878749557387, "grad_norm": 7.606365203857422, "learning_rate": 7.132933431925542e-06, "loss": 0.4483, "step": 7238 }, { "epoch": 0.366179371743639, "grad_norm": 5.314583778381348, "learning_rate": 7.132207482663863e-06, "loss": 0.3974, "step": 7239 }, { "epoch": 0.36622995599170416, "grad_norm": 7.547678470611572, "learning_rate": 7.131481478460348e-06, "loss": 0.4556, "step": 7240 }, { "epoch": 0.36628054023976936, "grad_norm": 7.12412166595459, "learning_rate": 7.130755419333704e-06, "loss": 0.4293, "step": 7241 }, { "epoch": 0.3663311244878345, "grad_norm": 7.302976608276367, "learning_rate": 7.130029305302638e-06, "loss": 0.4172, "step": 7242 }, { "epoch": 0.36638170873589965, "grad_norm": 13.678590774536133, "learning_rate": 7.1293031363858625e-06, "loss": 0.4658, "step": 7243 }, { "epoch": 0.3664322929839648, "grad_norm": 10.007983207702637, "learning_rate": 7.128576912602088e-06, "loss": 0.4704, "step": 7244 }, { "epoch": 0.36648287723202994, "grad_norm": 6.146446228027344, "learning_rate": 7.127850633970028e-06, "loss": 0.3685, "step": 7245 }, { "epoch": 0.3665334614800951, "grad_norm": 5.384579658508301, "learning_rate": 7.127124300508396e-06, "loss": 0.381, "step": 7246 }, { "epoch": 0.3665840457281602, "grad_norm": 6.340527057647705, "learning_rate": 7.126397912235911e-06, "loss": 0.5055, "step": 7247 }, { "epoch": 0.3666346299762254, "grad_norm": 5.998744010925293, "learning_rate": 7.1256714691712844e-06, "loss": 0.4129, "step": 7248 }, { "epoch": 0.36668521422429057, "grad_norm": 10.385416984558105, "learning_rate": 7.124944971333242e-06, "loss": 0.366, "step": 7249 }, { "epoch": 0.3667357984723557, "grad_norm": 4.65880012512207, "learning_rate": 7.124218418740498e-06, "loss": 0.3908, "step": 7250 }, { "epoch": 0.36678638272042086, "grad_norm": 24.632888793945312, "learning_rate": 7.1234918114117754e-06, "loss": 0.4448, "step": 7251 }, { "epoch": 0.366836966968486, "grad_norm": 5.427051544189453, "learning_rate": 7.1227651493658e-06, "loss": 0.376, "step": 7252 }, { "epoch": 0.36688755121655114, "grad_norm": 6.932285308837891, "learning_rate": 7.122038432621295e-06, "loss": 0.4837, "step": 7253 }, { "epoch": 0.36693813546461634, "grad_norm": 6.236761093139648, "learning_rate": 7.121311661196982e-06, "loss": 0.4855, "step": 7254 }, { "epoch": 0.3669887197126815, "grad_norm": 5.371333599090576, "learning_rate": 7.120584835111593e-06, "loss": 0.2729, "step": 7255 }, { "epoch": 0.36703930396074663, "grad_norm": 7.658029079437256, "learning_rate": 7.119857954383855e-06, "loss": 0.4305, "step": 7256 }, { "epoch": 0.3670898882088118, "grad_norm": 8.567641258239746, "learning_rate": 7.1191310190324966e-06, "loss": 0.4629, "step": 7257 }, { "epoch": 0.3671404724568769, "grad_norm": 7.853389739990234, "learning_rate": 7.118404029076252e-06, "loss": 0.4719, "step": 7258 }, { "epoch": 0.36719105670494206, "grad_norm": 6.979831695556641, "learning_rate": 7.117676984533852e-06, "loss": 0.4675, "step": 7259 }, { "epoch": 0.3672416409530072, "grad_norm": 7.775074481964111, "learning_rate": 7.116949885424031e-06, "loss": 0.4116, "step": 7260 }, { "epoch": 0.3672922252010724, "grad_norm": 6.910629749298096, "learning_rate": 7.116222731765526e-06, "loss": 0.3593, "step": 7261 }, { "epoch": 0.36734280944913755, "grad_norm": 6.29592227935791, "learning_rate": 7.11549552357707e-06, "loss": 0.3402, "step": 7262 }, { "epoch": 0.3673933936972027, "grad_norm": 7.861749172210693, "learning_rate": 7.114768260877405e-06, "loss": 0.3683, "step": 7263 }, { "epoch": 0.36744397794526784, "grad_norm": 9.036036491394043, "learning_rate": 7.11404094368527e-06, "loss": 0.4802, "step": 7264 }, { "epoch": 0.367494562193333, "grad_norm": 6.854830265045166, "learning_rate": 7.113313572019407e-06, "loss": 0.3454, "step": 7265 }, { "epoch": 0.36754514644139813, "grad_norm": 10.126832008361816, "learning_rate": 7.112586145898557e-06, "loss": 0.4515, "step": 7266 }, { "epoch": 0.3675957306894633, "grad_norm": 7.031952857971191, "learning_rate": 7.111858665341464e-06, "loss": 0.3895, "step": 7267 }, { "epoch": 0.3676463149375285, "grad_norm": 6.851478576660156, "learning_rate": 7.111131130366876e-06, "loss": 0.4363, "step": 7268 }, { "epoch": 0.3676968991855936, "grad_norm": 5.533620357513428, "learning_rate": 7.110403540993537e-06, "loss": 0.3621, "step": 7269 }, { "epoch": 0.36774748343365876, "grad_norm": 5.019972324371338, "learning_rate": 7.1096758972401965e-06, "loss": 0.3738, "step": 7270 }, { "epoch": 0.3677980676817239, "grad_norm": 7.997678756713867, "learning_rate": 7.108948199125604e-06, "loss": 0.3961, "step": 7271 }, { "epoch": 0.36784865192978905, "grad_norm": 7.015966892242432, "learning_rate": 7.10822044666851e-06, "loss": 0.3358, "step": 7272 }, { "epoch": 0.3678992361778542, "grad_norm": 12.306846618652344, "learning_rate": 7.107492639887666e-06, "loss": 0.507, "step": 7273 }, { "epoch": 0.3679498204259194, "grad_norm": 14.820356369018555, "learning_rate": 7.10676477880183e-06, "loss": 0.5693, "step": 7274 }, { "epoch": 0.36800040467398454, "grad_norm": 9.286060333251953, "learning_rate": 7.106036863429753e-06, "loss": 0.4523, "step": 7275 }, { "epoch": 0.3680509889220497, "grad_norm": 7.604959011077881, "learning_rate": 7.1053088937901915e-06, "loss": 0.3815, "step": 7276 }, { "epoch": 0.3681015731701148, "grad_norm": 6.246092796325684, "learning_rate": 7.104580869901907e-06, "loss": 0.3584, "step": 7277 }, { "epoch": 0.36815215741817997, "grad_norm": 7.575097560882568, "learning_rate": 7.103852791783656e-06, "loss": 0.4233, "step": 7278 }, { "epoch": 0.3682027416662451, "grad_norm": 8.424736976623535, "learning_rate": 7.103124659454201e-06, "loss": 0.5471, "step": 7279 }, { "epoch": 0.36825332591431026, "grad_norm": 7.0887885093688965, "learning_rate": 7.102396472932303e-06, "loss": 0.4952, "step": 7280 }, { "epoch": 0.36830391016237546, "grad_norm": 4.252700328826904, "learning_rate": 7.101668232236725e-06, "loss": 0.369, "step": 7281 }, { "epoch": 0.3683544944104406, "grad_norm": 7.182700157165527, "learning_rate": 7.100939937386233e-06, "loss": 0.4068, "step": 7282 }, { "epoch": 0.36840507865850575, "grad_norm": 10.48502254486084, "learning_rate": 7.100211588399595e-06, "loss": 0.4361, "step": 7283 }, { "epoch": 0.3684556629065709, "grad_norm": 6.182317733764648, "learning_rate": 7.099483185295576e-06, "loss": 0.373, "step": 7284 }, { "epoch": 0.36850624715463604, "grad_norm": 4.519266605377197, "learning_rate": 7.098754728092947e-06, "loss": 0.3485, "step": 7285 }, { "epoch": 0.3685568314027012, "grad_norm": 7.046545028686523, "learning_rate": 7.098026216810477e-06, "loss": 0.4033, "step": 7286 }, { "epoch": 0.3686074156507664, "grad_norm": 8.101428031921387, "learning_rate": 7.097297651466941e-06, "loss": 0.5186, "step": 7287 }, { "epoch": 0.3686579998988315, "grad_norm": 6.097434043884277, "learning_rate": 7.096569032081109e-06, "loss": 0.4486, "step": 7288 }, { "epoch": 0.36870858414689667, "grad_norm": 97.4621353149414, "learning_rate": 7.095840358671757e-06, "loss": 0.5132, "step": 7289 }, { "epoch": 0.3687591683949618, "grad_norm": 8.969474792480469, "learning_rate": 7.095111631257661e-06, "loss": 0.5462, "step": 7290 }, { "epoch": 0.36880975264302696, "grad_norm": 7.261470317840576, "learning_rate": 7.0943828498576e-06, "loss": 0.421, "step": 7291 }, { "epoch": 0.3688603368910921, "grad_norm": 7.985845565795898, "learning_rate": 7.0936540144903505e-06, "loss": 0.3869, "step": 7292 }, { "epoch": 0.36891092113915724, "grad_norm": 6.459444046020508, "learning_rate": 7.092925125174695e-06, "loss": 0.3332, "step": 7293 }, { "epoch": 0.36896150538722244, "grad_norm": 11.669432640075684, "learning_rate": 7.092196181929414e-06, "loss": 0.5391, "step": 7294 }, { "epoch": 0.3690120896352876, "grad_norm": 11.534123420715332, "learning_rate": 7.09146718477329e-06, "loss": 0.4923, "step": 7295 }, { "epoch": 0.36906267388335273, "grad_norm": 10.691431999206543, "learning_rate": 7.090738133725109e-06, "loss": 0.4878, "step": 7296 }, { "epoch": 0.3691132581314179, "grad_norm": 9.212586402893066, "learning_rate": 7.090009028803656e-06, "loss": 0.562, "step": 7297 }, { "epoch": 0.369163842379483, "grad_norm": 12.402060508728027, "learning_rate": 7.08927987002772e-06, "loss": 0.5367, "step": 7298 }, { "epoch": 0.36921442662754816, "grad_norm": 7.969493865966797, "learning_rate": 7.088550657416088e-06, "loss": 0.4434, "step": 7299 }, { "epoch": 0.3692650108756133, "grad_norm": 6.986035346984863, "learning_rate": 7.087821390987548e-06, "loss": 0.4225, "step": 7300 }, { "epoch": 0.3693155951236785, "grad_norm": 8.327058792114258, "learning_rate": 7.0870920707608945e-06, "loss": 0.4688, "step": 7301 }, { "epoch": 0.36936617937174365, "grad_norm": 6.308882236480713, "learning_rate": 7.0863626967549205e-06, "loss": 0.3429, "step": 7302 }, { "epoch": 0.3694167636198088, "grad_norm": 7.734405994415283, "learning_rate": 7.085633268988419e-06, "loss": 0.4839, "step": 7303 }, { "epoch": 0.36946734786787394, "grad_norm": 3.8384640216827393, "learning_rate": 7.084903787480186e-06, "loss": 0.39, "step": 7304 }, { "epoch": 0.3695179321159391, "grad_norm": 6.638646125793457, "learning_rate": 7.084174252249018e-06, "loss": 0.3856, "step": 7305 }, { "epoch": 0.36956851636400423, "grad_norm": 11.477777481079102, "learning_rate": 7.083444663313714e-06, "loss": 0.4625, "step": 7306 }, { "epoch": 0.36961910061206943, "grad_norm": 12.23459529876709, "learning_rate": 7.082715020693072e-06, "loss": 0.4736, "step": 7307 }, { "epoch": 0.3696696848601346, "grad_norm": 6.482019424438477, "learning_rate": 7.0819853244058955e-06, "loss": 0.4464, "step": 7308 }, { "epoch": 0.3697202691081997, "grad_norm": 5.397216320037842, "learning_rate": 7.081255574470985e-06, "loss": 0.4084, "step": 7309 }, { "epoch": 0.36977085335626486, "grad_norm": 7.208916187286377, "learning_rate": 7.080525770907146e-06, "loss": 0.4288, "step": 7310 }, { "epoch": 0.36982143760433, "grad_norm": 6.189490795135498, "learning_rate": 7.079795913733184e-06, "loss": 0.4415, "step": 7311 }, { "epoch": 0.36987202185239515, "grad_norm": 20.492523193359375, "learning_rate": 7.079066002967905e-06, "loss": 0.6028, "step": 7312 }, { "epoch": 0.3699226061004603, "grad_norm": 9.981329917907715, "learning_rate": 7.0783360386301156e-06, "loss": 0.3388, "step": 7313 }, { "epoch": 0.3699731903485255, "grad_norm": 8.004220962524414, "learning_rate": 7.077606020738626e-06, "loss": 0.3183, "step": 7314 }, { "epoch": 0.37002377459659064, "grad_norm": 5.230986595153809, "learning_rate": 7.076875949312249e-06, "loss": 0.4498, "step": 7315 }, { "epoch": 0.3700743588446558, "grad_norm": 7.017542362213135, "learning_rate": 7.076145824369795e-06, "loss": 0.4152, "step": 7316 }, { "epoch": 0.3701249430927209, "grad_norm": 10.854525566101074, "learning_rate": 7.0754156459300785e-06, "loss": 0.4043, "step": 7317 }, { "epoch": 0.37017552734078607, "grad_norm": 12.173341751098633, "learning_rate": 7.0746854140119135e-06, "loss": 0.4902, "step": 7318 }, { "epoch": 0.3702261115888512, "grad_norm": 5.6472673416137695, "learning_rate": 7.073955128634115e-06, "loss": 0.3544, "step": 7319 }, { "epoch": 0.37027669583691636, "grad_norm": 5.46494197845459, "learning_rate": 7.073224789815503e-06, "loss": 0.3899, "step": 7320 }, { "epoch": 0.37032728008498156, "grad_norm": 5.405341625213623, "learning_rate": 7.072494397574897e-06, "loss": 0.4269, "step": 7321 }, { "epoch": 0.3703778643330467, "grad_norm": 7.213343620300293, "learning_rate": 7.071763951931115e-06, "loss": 0.4073, "step": 7322 }, { "epoch": 0.37042844858111185, "grad_norm": 5.54286003112793, "learning_rate": 7.07103345290298e-06, "loss": 0.3951, "step": 7323 }, { "epoch": 0.370479032829177, "grad_norm": 6.5968194007873535, "learning_rate": 7.070302900509316e-06, "loss": 0.4428, "step": 7324 }, { "epoch": 0.37052961707724213, "grad_norm": 9.778217315673828, "learning_rate": 7.069572294768947e-06, "loss": 0.4444, "step": 7325 }, { "epoch": 0.3705802013253073, "grad_norm": 7.13428258895874, "learning_rate": 7.068841635700698e-06, "loss": 0.3818, "step": 7326 }, { "epoch": 0.3706307855733725, "grad_norm": 5.187729358673096, "learning_rate": 7.068110923323399e-06, "loss": 0.3505, "step": 7327 }, { "epoch": 0.3706813698214376, "grad_norm": 7.645251750946045, "learning_rate": 7.067380157655874e-06, "loss": 0.4107, "step": 7328 }, { "epoch": 0.37073195406950277, "grad_norm": 8.058476448059082, "learning_rate": 7.066649338716958e-06, "loss": 0.359, "step": 7329 }, { "epoch": 0.3707825383175679, "grad_norm": 8.56419849395752, "learning_rate": 7.06591846652548e-06, "loss": 0.4147, "step": 7330 }, { "epoch": 0.37083312256563306, "grad_norm": 10.041060447692871, "learning_rate": 7.065187541100274e-06, "loss": 0.3935, "step": 7331 }, { "epoch": 0.3708837068136982, "grad_norm": 5.888095378875732, "learning_rate": 7.0644565624601715e-06, "loss": 0.3929, "step": 7332 }, { "epoch": 0.37093429106176334, "grad_norm": 7.06428861618042, "learning_rate": 7.063725530624009e-06, "loss": 0.4534, "step": 7333 }, { "epoch": 0.37098487530982854, "grad_norm": 9.471659660339355, "learning_rate": 7.062994445610627e-06, "loss": 0.4839, "step": 7334 }, { "epoch": 0.3710354595578937, "grad_norm": 7.258330821990967, "learning_rate": 7.062263307438859e-06, "loss": 0.4135, "step": 7335 }, { "epoch": 0.37108604380595883, "grad_norm": 9.717671394348145, "learning_rate": 7.061532116127547e-06, "loss": 0.6472, "step": 7336 }, { "epoch": 0.371136628054024, "grad_norm": 7.027583599090576, "learning_rate": 7.060800871695535e-06, "loss": 0.4487, "step": 7337 }, { "epoch": 0.3711872123020891, "grad_norm": 10.806965827941895, "learning_rate": 7.060069574161659e-06, "loss": 0.5621, "step": 7338 }, { "epoch": 0.37123779655015426, "grad_norm": 5.3041605949401855, "learning_rate": 7.0593382235447664e-06, "loss": 0.3597, "step": 7339 }, { "epoch": 0.3712883807982194, "grad_norm": 4.712595462799072, "learning_rate": 7.0586068198637025e-06, "loss": 0.3527, "step": 7340 }, { "epoch": 0.3713389650462846, "grad_norm": 8.816646575927734, "learning_rate": 7.057875363137312e-06, "loss": 0.449, "step": 7341 }, { "epoch": 0.37138954929434975, "grad_norm": 14.363181114196777, "learning_rate": 7.057143853384445e-06, "loss": 0.5678, "step": 7342 }, { "epoch": 0.3714401335424149, "grad_norm": 11.773857116699219, "learning_rate": 7.056412290623951e-06, "loss": 0.4897, "step": 7343 }, { "epoch": 0.37149071779048004, "grad_norm": 8.305180549621582, "learning_rate": 7.055680674874677e-06, "loss": 0.4517, "step": 7344 }, { "epoch": 0.3715413020385452, "grad_norm": 6.24534273147583, "learning_rate": 7.054949006155478e-06, "loss": 0.4648, "step": 7345 }, { "epoch": 0.37159188628661033, "grad_norm": 8.732132911682129, "learning_rate": 7.054217284485207e-06, "loss": 0.4561, "step": 7346 }, { "epoch": 0.37164247053467553, "grad_norm": 13.39967155456543, "learning_rate": 7.053485509882719e-06, "loss": 0.6475, "step": 7347 }, { "epoch": 0.3716930547827407, "grad_norm": 5.898920059204102, "learning_rate": 7.052753682366866e-06, "loss": 0.3994, "step": 7348 }, { "epoch": 0.3717436390308058, "grad_norm": 11.438375473022461, "learning_rate": 7.052021801956512e-06, "loss": 0.476, "step": 7349 }, { "epoch": 0.37179422327887096, "grad_norm": 7.70365571975708, "learning_rate": 7.051289868670514e-06, "loss": 0.3462, "step": 7350 }, { "epoch": 0.3718448075269361, "grad_norm": 8.389705657958984, "learning_rate": 7.050557882527728e-06, "loss": 0.378, "step": 7351 }, { "epoch": 0.37189539177500125, "grad_norm": 7.306047439575195, "learning_rate": 7.049825843547018e-06, "loss": 0.4246, "step": 7352 }, { "epoch": 0.3719459760230664, "grad_norm": 12.727703094482422, "learning_rate": 7.049093751747249e-06, "loss": 0.5085, "step": 7353 }, { "epoch": 0.3719965602711316, "grad_norm": 9.895676612854004, "learning_rate": 7.0483616071472816e-06, "loss": 0.5087, "step": 7354 }, { "epoch": 0.37204714451919674, "grad_norm": 7.052361011505127, "learning_rate": 7.047629409765983e-06, "loss": 0.4412, "step": 7355 }, { "epoch": 0.3720977287672619, "grad_norm": 6.815237998962402, "learning_rate": 7.046897159622224e-06, "loss": 0.4325, "step": 7356 }, { "epoch": 0.372148313015327, "grad_norm": 8.76091480255127, "learning_rate": 7.046164856734867e-06, "loss": 0.4328, "step": 7357 }, { "epoch": 0.37219889726339217, "grad_norm": 11.5220308303833, "learning_rate": 7.045432501122783e-06, "loss": 0.6677, "step": 7358 }, { "epoch": 0.3722494815114573, "grad_norm": 9.494549751281738, "learning_rate": 7.044700092804846e-06, "loss": 0.428, "step": 7359 }, { "epoch": 0.37230006575952246, "grad_norm": 9.962597846984863, "learning_rate": 7.043967631799924e-06, "loss": 0.4353, "step": 7360 }, { "epoch": 0.37235065000758766, "grad_norm": 9.05825138092041, "learning_rate": 7.043235118126894e-06, "loss": 0.4184, "step": 7361 }, { "epoch": 0.3724012342556528, "grad_norm": 7.995223045349121, "learning_rate": 7.042502551804632e-06, "loss": 0.3506, "step": 7362 }, { "epoch": 0.37245181850371795, "grad_norm": 8.872065544128418, "learning_rate": 7.041769932852011e-06, "loss": 0.412, "step": 7363 }, { "epoch": 0.3725024027517831, "grad_norm": 5.327847003936768, "learning_rate": 7.0410372612879094e-06, "loss": 0.3985, "step": 7364 }, { "epoch": 0.37255298699984823, "grad_norm": 9.064194679260254, "learning_rate": 7.0403045371312105e-06, "loss": 0.4506, "step": 7365 }, { "epoch": 0.3726035712479134, "grad_norm": 12.511160850524902, "learning_rate": 7.039571760400789e-06, "loss": 0.5377, "step": 7366 }, { "epoch": 0.3726541554959786, "grad_norm": 6.813610076904297, "learning_rate": 7.03883893111553e-06, "loss": 0.4134, "step": 7367 }, { "epoch": 0.3727047397440437, "grad_norm": 10.921698570251465, "learning_rate": 7.038106049294318e-06, "loss": 0.4941, "step": 7368 }, { "epoch": 0.37275532399210887, "grad_norm": 9.02417278289795, "learning_rate": 7.037373114956036e-06, "loss": 0.5255, "step": 7369 }, { "epoch": 0.372805908240174, "grad_norm": 5.772407531738281, "learning_rate": 7.036640128119568e-06, "loss": 0.4549, "step": 7370 }, { "epoch": 0.37285649248823916, "grad_norm": 9.346429824829102, "learning_rate": 7.035907088803804e-06, "loss": 0.4367, "step": 7371 }, { "epoch": 0.3729070767363043, "grad_norm": 7.804226875305176, "learning_rate": 7.035173997027633e-06, "loss": 0.3933, "step": 7372 }, { "epoch": 0.37295766098436944, "grad_norm": 8.427736282348633, "learning_rate": 7.0344408528099425e-06, "loss": 0.4524, "step": 7373 }, { "epoch": 0.37300824523243464, "grad_norm": 8.157112121582031, "learning_rate": 7.033707656169626e-06, "loss": 0.4006, "step": 7374 }, { "epoch": 0.3730588294804998, "grad_norm": 4.731716632843018, "learning_rate": 7.0329744071255754e-06, "loss": 0.386, "step": 7375 }, { "epoch": 0.37310941372856493, "grad_norm": 11.0487699508667, "learning_rate": 7.032241105696684e-06, "loss": 0.5334, "step": 7376 }, { "epoch": 0.3731599979766301, "grad_norm": 8.118024826049805, "learning_rate": 7.031507751901848e-06, "loss": 0.4391, "step": 7377 }, { "epoch": 0.3732105822246952, "grad_norm": 10.4026517868042, "learning_rate": 7.030774345759966e-06, "loss": 0.4591, "step": 7378 }, { "epoch": 0.37326116647276036, "grad_norm": 11.287120819091797, "learning_rate": 7.030040887289932e-06, "loss": 0.483, "step": 7379 }, { "epoch": 0.3733117507208255, "grad_norm": 5.122019290924072, "learning_rate": 7.029307376510648e-06, "loss": 0.3524, "step": 7380 }, { "epoch": 0.3733623349688907, "grad_norm": 5.219818115234375, "learning_rate": 7.028573813441015e-06, "loss": 0.3483, "step": 7381 }, { "epoch": 0.37341291921695585, "grad_norm": 6.270233631134033, "learning_rate": 7.027840198099936e-06, "loss": 0.4354, "step": 7382 }, { "epoch": 0.373463503465021, "grad_norm": 18.0324649810791, "learning_rate": 7.027106530506311e-06, "loss": 0.5143, "step": 7383 }, { "epoch": 0.37351408771308614, "grad_norm": 9.296073913574219, "learning_rate": 7.026372810679048e-06, "loss": 0.2903, "step": 7384 }, { "epoch": 0.3735646719611513, "grad_norm": 8.126208305358887, "learning_rate": 7.025639038637052e-06, "loss": 0.4335, "step": 7385 }, { "epoch": 0.37361525620921643, "grad_norm": 8.690681457519531, "learning_rate": 7.02490521439923e-06, "loss": 0.4277, "step": 7386 }, { "epoch": 0.37366584045728163, "grad_norm": 6.0148539543151855, "learning_rate": 7.024171337984493e-06, "loss": 0.4028, "step": 7387 }, { "epoch": 0.3737164247053468, "grad_norm": 6.761460781097412, "learning_rate": 7.0234374094117495e-06, "loss": 0.4281, "step": 7388 }, { "epoch": 0.3737670089534119, "grad_norm": 8.656523704528809, "learning_rate": 7.022703428699911e-06, "loss": 0.4215, "step": 7389 }, { "epoch": 0.37381759320147706, "grad_norm": 5.07455587387085, "learning_rate": 7.021969395867889e-06, "loss": 0.4202, "step": 7390 }, { "epoch": 0.3738681774495422, "grad_norm": 6.273634433746338, "learning_rate": 7.021235310934602e-06, "loss": 0.4363, "step": 7391 }, { "epoch": 0.37391876169760735, "grad_norm": 12.399003982543945, "learning_rate": 7.020501173918962e-06, "loss": 0.5601, "step": 7392 }, { "epoch": 0.3739693459456725, "grad_norm": 6.80402946472168, "learning_rate": 7.019766984839889e-06, "loss": 0.3581, "step": 7393 }, { "epoch": 0.3740199301937377, "grad_norm": 5.276639461517334, "learning_rate": 7.0190327437162976e-06, "loss": 0.3563, "step": 7394 }, { "epoch": 0.37407051444180284, "grad_norm": 9.017439842224121, "learning_rate": 7.018298450567109e-06, "loss": 0.5047, "step": 7395 }, { "epoch": 0.374121098689868, "grad_norm": 5.767276287078857, "learning_rate": 7.017564105411244e-06, "loss": 0.3816, "step": 7396 }, { "epoch": 0.3741716829379331, "grad_norm": 6.614062786102295, "learning_rate": 7.016829708267626e-06, "loss": 0.4245, "step": 7397 }, { "epoch": 0.37422226718599827, "grad_norm": 9.295332908630371, "learning_rate": 7.016095259155178e-06, "loss": 0.3975, "step": 7398 }, { "epoch": 0.3742728514340634, "grad_norm": 7.3871870040893555, "learning_rate": 7.015360758092824e-06, "loss": 0.3809, "step": 7399 }, { "epoch": 0.37432343568212856, "grad_norm": 5.2889790534973145, "learning_rate": 7.014626205099493e-06, "loss": 0.4336, "step": 7400 }, { "epoch": 0.37437401993019376, "grad_norm": 9.386731147766113, "learning_rate": 7.013891600194109e-06, "loss": 0.5126, "step": 7401 }, { "epoch": 0.3744246041782589, "grad_norm": 9.96883773803711, "learning_rate": 7.013156943395604e-06, "loss": 0.4662, "step": 7402 }, { "epoch": 0.37447518842632405, "grad_norm": 16.745311737060547, "learning_rate": 7.012422234722906e-06, "loss": 0.5247, "step": 7403 }, { "epoch": 0.3745257726743892, "grad_norm": 10.264107704162598, "learning_rate": 7.011687474194948e-06, "loss": 0.3499, "step": 7404 }, { "epoch": 0.37457635692245433, "grad_norm": 7.9535932540893555, "learning_rate": 7.010952661830664e-06, "loss": 0.4948, "step": 7405 }, { "epoch": 0.3746269411705195, "grad_norm": 9.316139221191406, "learning_rate": 7.010217797648986e-06, "loss": 0.5894, "step": 7406 }, { "epoch": 0.3746775254185847, "grad_norm": 16.253442764282227, "learning_rate": 7.009482881668852e-06, "loss": 0.4627, "step": 7407 }, { "epoch": 0.3747281096666498, "grad_norm": 9.140399932861328, "learning_rate": 7.0087479139091975e-06, "loss": 0.3524, "step": 7408 }, { "epoch": 0.37477869391471497, "grad_norm": 9.20457649230957, "learning_rate": 7.00801289438896e-06, "loss": 0.4655, "step": 7409 }, { "epoch": 0.3748292781627801, "grad_norm": 4.5980963706970215, "learning_rate": 7.007277823127081e-06, "loss": 0.332, "step": 7410 }, { "epoch": 0.37487986241084525, "grad_norm": 11.965229988098145, "learning_rate": 7.0065427001425015e-06, "loss": 0.5397, "step": 7411 }, { "epoch": 0.3749304466589104, "grad_norm": 5.880317211151123, "learning_rate": 7.005807525454162e-06, "loss": 0.3656, "step": 7412 }, { "epoch": 0.37498103090697554, "grad_norm": 5.269329071044922, "learning_rate": 7.005072299081008e-06, "loss": 0.443, "step": 7413 }, { "epoch": 0.37503161515504074, "grad_norm": 9.736770629882812, "learning_rate": 7.004337021041984e-06, "loss": 0.4434, "step": 7414 }, { "epoch": 0.3750821994031059, "grad_norm": 7.043842315673828, "learning_rate": 7.003601691356036e-06, "loss": 0.3621, "step": 7415 }, { "epoch": 0.37513278365117103, "grad_norm": 7.365466117858887, "learning_rate": 7.002866310042113e-06, "loss": 0.3728, "step": 7416 }, { "epoch": 0.3751833678992362, "grad_norm": 10.36922550201416, "learning_rate": 7.002130877119162e-06, "loss": 0.4428, "step": 7417 }, { "epoch": 0.3752339521473013, "grad_norm": 6.686657905578613, "learning_rate": 7.001395392606134e-06, "loss": 0.3904, "step": 7418 }, { "epoch": 0.37528453639536646, "grad_norm": 8.0663480758667, "learning_rate": 7.000659856521982e-06, "loss": 0.4011, "step": 7419 }, { "epoch": 0.3753351206434316, "grad_norm": 9.378998756408691, "learning_rate": 6.9999242688856585e-06, "loss": 0.5331, "step": 7420 }, { "epoch": 0.3753857048914968, "grad_norm": 8.715238571166992, "learning_rate": 6.999188629716115e-06, "loss": 0.4711, "step": 7421 }, { "epoch": 0.37543628913956195, "grad_norm": 18.06079864501953, "learning_rate": 6.9984529390323106e-06, "loss": 0.3677, "step": 7422 }, { "epoch": 0.3754868733876271, "grad_norm": 8.502348899841309, "learning_rate": 6.9977171968532e-06, "loss": 0.4818, "step": 7423 }, { "epoch": 0.37553745763569224, "grad_norm": 7.225578308105469, "learning_rate": 6.996981403197742e-06, "loss": 0.4863, "step": 7424 }, { "epoch": 0.3755880418837574, "grad_norm": 6.978858947753906, "learning_rate": 6.9962455580849e-06, "loss": 0.4078, "step": 7425 }, { "epoch": 0.37563862613182253, "grad_norm": 5.223010540008545, "learning_rate": 6.995509661533629e-06, "loss": 0.4597, "step": 7426 }, { "epoch": 0.37568921037988773, "grad_norm": 9.388870239257812, "learning_rate": 6.994773713562894e-06, "loss": 0.5246, "step": 7427 }, { "epoch": 0.37573979462795287, "grad_norm": 7.648998260498047, "learning_rate": 6.9940377141916595e-06, "loss": 0.4391, "step": 7428 }, { "epoch": 0.375790378876018, "grad_norm": 6.873919486999512, "learning_rate": 6.9933016634388895e-06, "loss": 0.3341, "step": 7429 }, { "epoch": 0.37584096312408316, "grad_norm": 15.243257522583008, "learning_rate": 6.99256556132355e-06, "loss": 0.5265, "step": 7430 }, { "epoch": 0.3758915473721483, "grad_norm": 6.661818027496338, "learning_rate": 6.991829407864609e-06, "loss": 0.3816, "step": 7431 }, { "epoch": 0.37594213162021345, "grad_norm": 10.317720413208008, "learning_rate": 6.991093203081034e-06, "loss": 0.4351, "step": 7432 }, { "epoch": 0.3759927158682786, "grad_norm": 9.02324104309082, "learning_rate": 6.990356946991799e-06, "loss": 0.4901, "step": 7433 }, { "epoch": 0.3760433001163438, "grad_norm": 9.273063659667969, "learning_rate": 6.989620639615871e-06, "loss": 0.47, "step": 7434 }, { "epoch": 0.37609388436440894, "grad_norm": 8.741453170776367, "learning_rate": 6.988884280972226e-06, "loss": 0.4226, "step": 7435 }, { "epoch": 0.3761444686124741, "grad_norm": 4.7110090255737305, "learning_rate": 6.988147871079836e-06, "loss": 0.3678, "step": 7436 }, { "epoch": 0.3761950528605392, "grad_norm": 7.960603713989258, "learning_rate": 6.987411409957676e-06, "loss": 0.4823, "step": 7437 }, { "epoch": 0.37624563710860437, "grad_norm": 6.97502326965332, "learning_rate": 6.986674897624728e-06, "loss": 0.5195, "step": 7438 }, { "epoch": 0.3762962213566695, "grad_norm": 5.758822917938232, "learning_rate": 6.985938334099964e-06, "loss": 0.4003, "step": 7439 }, { "epoch": 0.3763468056047347, "grad_norm": 16.6807861328125, "learning_rate": 6.9852017194023665e-06, "loss": 0.493, "step": 7440 }, { "epoch": 0.37639738985279986, "grad_norm": 8.86595630645752, "learning_rate": 6.984465053550917e-06, "loss": 0.4532, "step": 7441 }, { "epoch": 0.376447974100865, "grad_norm": 5.678272724151611, "learning_rate": 6.983728336564595e-06, "loss": 0.3856, "step": 7442 }, { "epoch": 0.37649855834893015, "grad_norm": 8.971824645996094, "learning_rate": 6.982991568462385e-06, "loss": 0.4687, "step": 7443 }, { "epoch": 0.3765491425969953, "grad_norm": 21.579055786132812, "learning_rate": 6.982254749263273e-06, "loss": 0.5322, "step": 7444 }, { "epoch": 0.37659972684506043, "grad_norm": 10.25509262084961, "learning_rate": 6.9815178789862435e-06, "loss": 0.4465, "step": 7445 }, { "epoch": 0.3766503110931256, "grad_norm": 5.459258079528809, "learning_rate": 6.9807809576502846e-06, "loss": 0.3536, "step": 7446 }, { "epoch": 0.3767008953411908, "grad_norm": 11.009657859802246, "learning_rate": 6.980043985274385e-06, "loss": 0.4598, "step": 7447 }, { "epoch": 0.3767514795892559, "grad_norm": 8.545903205871582, "learning_rate": 6.979306961877534e-06, "loss": 0.4541, "step": 7448 }, { "epoch": 0.37680206383732107, "grad_norm": 7.41439151763916, "learning_rate": 6.978569887478724e-06, "loss": 0.3134, "step": 7449 }, { "epoch": 0.3768526480853862, "grad_norm": 8.601168632507324, "learning_rate": 6.977832762096947e-06, "loss": 0.4272, "step": 7450 }, { "epoch": 0.37690323233345135, "grad_norm": 5.022069931030273, "learning_rate": 6.9770955857511955e-06, "loss": 0.3473, "step": 7451 }, { "epoch": 0.3769538165815165, "grad_norm": 7.766541957855225, "learning_rate": 6.976358358460469e-06, "loss": 0.4713, "step": 7452 }, { "epoch": 0.37700440082958164, "grad_norm": 7.626211166381836, "learning_rate": 6.975621080243758e-06, "loss": 0.455, "step": 7453 }, { "epoch": 0.37705498507764684, "grad_norm": 16.946683883666992, "learning_rate": 6.974883751120066e-06, "loss": 0.512, "step": 7454 }, { "epoch": 0.377105569325712, "grad_norm": 9.024782180786133, "learning_rate": 6.974146371108389e-06, "loss": 0.4047, "step": 7455 }, { "epoch": 0.37715615357377713, "grad_norm": 7.6807451248168945, "learning_rate": 6.973408940227728e-06, "loss": 0.5368, "step": 7456 }, { "epoch": 0.3772067378218423, "grad_norm": 5.830257415771484, "learning_rate": 6.972671458497087e-06, "loss": 0.5061, "step": 7457 }, { "epoch": 0.3772573220699074, "grad_norm": 6.882201671600342, "learning_rate": 6.971933925935466e-06, "loss": 0.4078, "step": 7458 }, { "epoch": 0.37730790631797256, "grad_norm": 7.116455554962158, "learning_rate": 6.971196342561871e-06, "loss": 0.3517, "step": 7459 }, { "epoch": 0.37735849056603776, "grad_norm": 6.182440280914307, "learning_rate": 6.970458708395308e-06, "loss": 0.3826, "step": 7460 }, { "epoch": 0.3774090748141029, "grad_norm": 8.642278671264648, "learning_rate": 6.969721023454782e-06, "loss": 0.4559, "step": 7461 }, { "epoch": 0.37745965906216805, "grad_norm": 9.954180717468262, "learning_rate": 6.968983287759303e-06, "loss": 0.4547, "step": 7462 }, { "epoch": 0.3775102433102332, "grad_norm": 7.6111907958984375, "learning_rate": 6.968245501327882e-06, "loss": 0.4536, "step": 7463 }, { "epoch": 0.37756082755829834, "grad_norm": 13.763185501098633, "learning_rate": 6.9675076641795266e-06, "loss": 0.4436, "step": 7464 }, { "epoch": 0.3776114118063635, "grad_norm": 5.585670471191406, "learning_rate": 6.9667697763332545e-06, "loss": 0.3595, "step": 7465 }, { "epoch": 0.37766199605442863, "grad_norm": 9.819565773010254, "learning_rate": 6.966031837808074e-06, "loss": 0.455, "step": 7466 }, { "epoch": 0.3777125803024938, "grad_norm": 5.982438564300537, "learning_rate": 6.965293848623e-06, "loss": 0.3813, "step": 7467 }, { "epoch": 0.37776316455055897, "grad_norm": 12.198424339294434, "learning_rate": 6.964555808797051e-06, "loss": 0.4733, "step": 7468 }, { "epoch": 0.3778137487986241, "grad_norm": 9.739236831665039, "learning_rate": 6.963817718349246e-06, "loss": 0.4476, "step": 7469 }, { "epoch": 0.37786433304668926, "grad_norm": 7.794495105743408, "learning_rate": 6.9630795772986005e-06, "loss": 0.4129, "step": 7470 }, { "epoch": 0.3779149172947544, "grad_norm": 7.40865421295166, "learning_rate": 6.9623413856641356e-06, "loss": 0.45, "step": 7471 }, { "epoch": 0.37796550154281955, "grad_norm": 10.686768531799316, "learning_rate": 6.961603143464874e-06, "loss": 0.5046, "step": 7472 }, { "epoch": 0.3780160857908847, "grad_norm": 11.792746543884277, "learning_rate": 6.960864850719837e-06, "loss": 0.5537, "step": 7473 }, { "epoch": 0.3780666700389499, "grad_norm": 6.128517150878906, "learning_rate": 6.9601265074480484e-06, "loss": 0.4863, "step": 7474 }, { "epoch": 0.37811725428701504, "grad_norm": 14.36034107208252, "learning_rate": 6.9593881136685345e-06, "loss": 0.428, "step": 7475 }, { "epoch": 0.3781678385350802, "grad_norm": 12.065937042236328, "learning_rate": 6.958649669400323e-06, "loss": 0.4823, "step": 7476 }, { "epoch": 0.3782184227831453, "grad_norm": 7.184259414672852, "learning_rate": 6.95791117466244e-06, "loss": 0.485, "step": 7477 }, { "epoch": 0.37826900703121047, "grad_norm": 9.409575462341309, "learning_rate": 6.957172629473914e-06, "loss": 0.3748, "step": 7478 }, { "epoch": 0.3783195912792756, "grad_norm": 11.212884902954102, "learning_rate": 6.956434033853778e-06, "loss": 0.4745, "step": 7479 }, { "epoch": 0.3783701755273408, "grad_norm": 8.114727973937988, "learning_rate": 6.955695387821061e-06, "loss": 0.4243, "step": 7480 }, { "epoch": 0.37842075977540596, "grad_norm": 10.747086524963379, "learning_rate": 6.954956691394799e-06, "loss": 0.3253, "step": 7481 }, { "epoch": 0.3784713440234711, "grad_norm": 7.97703742980957, "learning_rate": 6.9542179445940235e-06, "loss": 0.4537, "step": 7482 }, { "epoch": 0.37852192827153625, "grad_norm": 13.766436576843262, "learning_rate": 6.953479147437773e-06, "loss": 0.6113, "step": 7483 }, { "epoch": 0.3785725125196014, "grad_norm": 6.125182151794434, "learning_rate": 6.952740299945082e-06, "loss": 0.4206, "step": 7484 }, { "epoch": 0.37862309676766653, "grad_norm": 7.248965740203857, "learning_rate": 6.952001402134994e-06, "loss": 0.5131, "step": 7485 }, { "epoch": 0.3786736810157317, "grad_norm": 7.402251720428467, "learning_rate": 6.95126245402654e-06, "loss": 0.4062, "step": 7486 }, { "epoch": 0.3787242652637969, "grad_norm": 4.985194206237793, "learning_rate": 6.950523455638768e-06, "loss": 0.4055, "step": 7487 }, { "epoch": 0.378774849511862, "grad_norm": 5.161146640777588, "learning_rate": 6.949784406990718e-06, "loss": 0.4453, "step": 7488 }, { "epoch": 0.37882543375992717, "grad_norm": 7.970603942871094, "learning_rate": 6.949045308101432e-06, "loss": 0.3708, "step": 7489 }, { "epoch": 0.3788760180079923, "grad_norm": 9.877020835876465, "learning_rate": 6.948306158989957e-06, "loss": 0.4827, "step": 7490 }, { "epoch": 0.37892660225605745, "grad_norm": 6.668606758117676, "learning_rate": 6.947566959675338e-06, "loss": 0.4264, "step": 7491 }, { "epoch": 0.3789771865041226, "grad_norm": 6.783577919006348, "learning_rate": 6.946827710176623e-06, "loss": 0.4501, "step": 7492 }, { "epoch": 0.37902777075218774, "grad_norm": 6.86146354675293, "learning_rate": 6.94608841051286e-06, "loss": 0.4642, "step": 7493 }, { "epoch": 0.37907835500025294, "grad_norm": 20.4364070892334, "learning_rate": 6.945349060703099e-06, "loss": 0.5913, "step": 7494 }, { "epoch": 0.3791289392483181, "grad_norm": 8.224077224731445, "learning_rate": 6.944609660766392e-06, "loss": 0.3203, "step": 7495 }, { "epoch": 0.37917952349638323, "grad_norm": 5.752328872680664, "learning_rate": 6.9438702107217905e-06, "loss": 0.4363, "step": 7496 }, { "epoch": 0.3792301077444484, "grad_norm": 8.393786430358887, "learning_rate": 6.94313071058835e-06, "loss": 0.4426, "step": 7497 }, { "epoch": 0.3792806919925135, "grad_norm": 6.412071228027344, "learning_rate": 6.942391160385125e-06, "loss": 0.4218, "step": 7498 }, { "epoch": 0.37933127624057866, "grad_norm": 7.764096260070801, "learning_rate": 6.9416515601311686e-06, "loss": 0.3256, "step": 7499 }, { "epoch": 0.37938186048864386, "grad_norm": 6.563656806945801, "learning_rate": 6.940911909845543e-06, "loss": 0.4662, "step": 7500 }, { "epoch": 0.379432444736709, "grad_norm": 6.1216349601745605, "learning_rate": 6.940172209547306e-06, "loss": 0.4066, "step": 7501 }, { "epoch": 0.37948302898477415, "grad_norm": 8.199795722961426, "learning_rate": 6.939432459255516e-06, "loss": 0.4404, "step": 7502 }, { "epoch": 0.3795336132328393, "grad_norm": 10.411615371704102, "learning_rate": 6.938692658989237e-06, "loss": 0.4362, "step": 7503 }, { "epoch": 0.37958419748090444, "grad_norm": 7.1422600746154785, "learning_rate": 6.9379528087675315e-06, "loss": 0.3896, "step": 7504 }, { "epoch": 0.3796347817289696, "grad_norm": 6.741068363189697, "learning_rate": 6.937212908609463e-06, "loss": 0.3571, "step": 7505 }, { "epoch": 0.37968536597703473, "grad_norm": 7.608179569244385, "learning_rate": 6.9364729585340966e-06, "loss": 0.4626, "step": 7506 }, { "epoch": 0.3797359502250999, "grad_norm": 5.205211162567139, "learning_rate": 6.935732958560499e-06, "loss": 0.4026, "step": 7507 }, { "epoch": 0.37978653447316507, "grad_norm": 15.829977035522461, "learning_rate": 6.934992908707738e-06, "loss": 0.5382, "step": 7508 }, { "epoch": 0.3798371187212302, "grad_norm": 10.343171119689941, "learning_rate": 6.934252808994885e-06, "loss": 0.4045, "step": 7509 }, { "epoch": 0.37988770296929536, "grad_norm": 10.416625022888184, "learning_rate": 6.933512659441009e-06, "loss": 0.434, "step": 7510 }, { "epoch": 0.3799382872173605, "grad_norm": 14.690793991088867, "learning_rate": 6.932772460065181e-06, "loss": 0.4889, "step": 7511 }, { "epoch": 0.37998887146542565, "grad_norm": 15.56773853302002, "learning_rate": 6.9320322108864755e-06, "loss": 0.4297, "step": 7512 }, { "epoch": 0.3800394557134908, "grad_norm": 11.978774070739746, "learning_rate": 6.931291911923966e-06, "loss": 0.5466, "step": 7513 }, { "epoch": 0.380090039961556, "grad_norm": 9.036718368530273, "learning_rate": 6.93055156319673e-06, "loss": 0.432, "step": 7514 }, { "epoch": 0.38014062420962114, "grad_norm": 12.052221298217773, "learning_rate": 6.929811164723843e-06, "loss": 0.4348, "step": 7515 }, { "epoch": 0.3801912084576863, "grad_norm": 14.894166946411133, "learning_rate": 6.929070716524383e-06, "loss": 0.5935, "step": 7516 }, { "epoch": 0.3802417927057514, "grad_norm": 6.119451999664307, "learning_rate": 6.9283302186174315e-06, "loss": 0.3642, "step": 7517 }, { "epoch": 0.38029237695381657, "grad_norm": 11.135507583618164, "learning_rate": 6.927589671022066e-06, "loss": 0.626, "step": 7518 }, { "epoch": 0.3803429612018817, "grad_norm": 6.5288166999816895, "learning_rate": 6.926849073757371e-06, "loss": 0.3663, "step": 7519 }, { "epoch": 0.3803935454499469, "grad_norm": 9.386127471923828, "learning_rate": 6.9261084268424305e-06, "loss": 0.566, "step": 7520 }, { "epoch": 0.38044412969801206, "grad_norm": 6.223455905914307, "learning_rate": 6.925367730296327e-06, "loss": 0.4232, "step": 7521 }, { "epoch": 0.3804947139460772, "grad_norm": 23.871767044067383, "learning_rate": 6.924626984138148e-06, "loss": 0.6088, "step": 7522 }, { "epoch": 0.38054529819414235, "grad_norm": 11.770772933959961, "learning_rate": 6.9238861883869824e-06, "loss": 0.5522, "step": 7523 }, { "epoch": 0.3805958824422075, "grad_norm": 8.964851379394531, "learning_rate": 6.923145343061914e-06, "loss": 0.458, "step": 7524 }, { "epoch": 0.38064646669027263, "grad_norm": 11.200517654418945, "learning_rate": 6.922404448182036e-06, "loss": 0.5108, "step": 7525 }, { "epoch": 0.3806970509383378, "grad_norm": 5.348020076751709, "learning_rate": 6.921663503766439e-06, "loss": 0.4861, "step": 7526 }, { "epoch": 0.380747635186403, "grad_norm": 6.881921291351318, "learning_rate": 6.920922509834215e-06, "loss": 0.4341, "step": 7527 }, { "epoch": 0.3807982194344681, "grad_norm": 7.068937301635742, "learning_rate": 6.920181466404456e-06, "loss": 0.4064, "step": 7528 }, { "epoch": 0.38084880368253327, "grad_norm": 9.39220142364502, "learning_rate": 6.919440373496262e-06, "loss": 0.403, "step": 7529 }, { "epoch": 0.3808993879305984, "grad_norm": 7.468896865844727, "learning_rate": 6.918699231128725e-06, "loss": 0.412, "step": 7530 }, { "epoch": 0.38094997217866355, "grad_norm": 10.135671615600586, "learning_rate": 6.917958039320941e-06, "loss": 0.449, "step": 7531 }, { "epoch": 0.3810005564267287, "grad_norm": 8.047903060913086, "learning_rate": 6.917216798092012e-06, "loss": 0.4689, "step": 7532 }, { "epoch": 0.38105114067479384, "grad_norm": 9.239740371704102, "learning_rate": 6.916475507461038e-06, "loss": 0.392, "step": 7533 }, { "epoch": 0.38110172492285904, "grad_norm": 3.7852327823638916, "learning_rate": 6.915734167447118e-06, "loss": 0.4077, "step": 7534 }, { "epoch": 0.3811523091709242, "grad_norm": 6.610409736633301, "learning_rate": 6.914992778069357e-06, "loss": 0.3796, "step": 7535 }, { "epoch": 0.38120289341898933, "grad_norm": 18.377504348754883, "learning_rate": 6.914251339346858e-06, "loss": 0.5266, "step": 7536 }, { "epoch": 0.3812534776670545, "grad_norm": 8.769515991210938, "learning_rate": 6.913509851298723e-06, "loss": 0.4428, "step": 7537 }, { "epoch": 0.3813040619151196, "grad_norm": 5.518862247467041, "learning_rate": 6.912768313944063e-06, "loss": 0.3305, "step": 7538 }, { "epoch": 0.38135464616318476, "grad_norm": 10.723577499389648, "learning_rate": 6.9120267273019835e-06, "loss": 0.4673, "step": 7539 }, { "epoch": 0.38140523041124996, "grad_norm": 7.7227396965026855, "learning_rate": 6.911285091391593e-06, "loss": 0.4806, "step": 7540 }, { "epoch": 0.3814558146593151, "grad_norm": 9.888964653015137, "learning_rate": 6.910543406232002e-06, "loss": 0.5026, "step": 7541 }, { "epoch": 0.38150639890738025, "grad_norm": 12.870532035827637, "learning_rate": 6.9098016718423244e-06, "loss": 0.4152, "step": 7542 }, { "epoch": 0.3815569831554454, "grad_norm": 8.884410858154297, "learning_rate": 6.9090598882416685e-06, "loss": 0.5068, "step": 7543 }, { "epoch": 0.38160756740351054, "grad_norm": 5.756654262542725, "learning_rate": 6.908318055449151e-06, "loss": 0.38, "step": 7544 }, { "epoch": 0.3816581516515757, "grad_norm": 8.896829605102539, "learning_rate": 6.907576173483885e-06, "loss": 0.3293, "step": 7545 }, { "epoch": 0.3817087358996408, "grad_norm": 7.757115364074707, "learning_rate": 6.906834242364991e-06, "loss": 0.4437, "step": 7546 }, { "epoch": 0.381759320147706, "grad_norm": 16.029949188232422, "learning_rate": 6.906092262111582e-06, "loss": 0.544, "step": 7547 }, { "epoch": 0.38180990439577117, "grad_norm": 7.995706081390381, "learning_rate": 6.905350232742781e-06, "loss": 0.3754, "step": 7548 }, { "epoch": 0.3818604886438363, "grad_norm": 7.4046454429626465, "learning_rate": 6.904608154277708e-06, "loss": 0.4604, "step": 7549 }, { "epoch": 0.38191107289190146, "grad_norm": 13.282474517822266, "learning_rate": 6.90386602673548e-06, "loss": 0.6514, "step": 7550 }, { "epoch": 0.3819616571399666, "grad_norm": 5.809009075164795, "learning_rate": 6.9031238501352245e-06, "loss": 0.4145, "step": 7551 }, { "epoch": 0.38201224138803175, "grad_norm": 3.8581817150115967, "learning_rate": 6.902381624496065e-06, "loss": 0.371, "step": 7552 }, { "epoch": 0.3820628256360969, "grad_norm": 5.303149223327637, "learning_rate": 6.901639349837125e-06, "loss": 0.4226, "step": 7553 }, { "epoch": 0.3821134098841621, "grad_norm": 8.18136215209961, "learning_rate": 6.900897026177533e-06, "loss": 0.4482, "step": 7554 }, { "epoch": 0.38216399413222724, "grad_norm": 12.015829086303711, "learning_rate": 6.900154653536416e-06, "loss": 0.484, "step": 7555 }, { "epoch": 0.3822145783802924, "grad_norm": 7.400139808654785, "learning_rate": 6.899412231932903e-06, "loss": 0.4141, "step": 7556 }, { "epoch": 0.3822651626283575, "grad_norm": 8.713963508605957, "learning_rate": 6.898669761386123e-06, "loss": 0.4285, "step": 7557 }, { "epoch": 0.38231574687642267, "grad_norm": 10.64313793182373, "learning_rate": 6.897927241915212e-06, "loss": 0.5274, "step": 7558 }, { "epoch": 0.3823663311244878, "grad_norm": 10.246651649475098, "learning_rate": 6.897184673539299e-06, "loss": 0.4927, "step": 7559 }, { "epoch": 0.382416915372553, "grad_norm": 7.854503154754639, "learning_rate": 6.896442056277519e-06, "loss": 0.4101, "step": 7560 }, { "epoch": 0.38246749962061816, "grad_norm": 6.748328685760498, "learning_rate": 6.895699390149009e-06, "loss": 0.4075, "step": 7561 }, { "epoch": 0.3825180838686833, "grad_norm": 9.414180755615234, "learning_rate": 6.894956675172904e-06, "loss": 0.3759, "step": 7562 }, { "epoch": 0.38256866811674844, "grad_norm": 4.672850608825684, "learning_rate": 6.894213911368342e-06, "loss": 0.3405, "step": 7563 }, { "epoch": 0.3826192523648136, "grad_norm": 9.760993003845215, "learning_rate": 6.893471098754464e-06, "loss": 0.4667, "step": 7564 }, { "epoch": 0.38266983661287873, "grad_norm": 7.219825744628906, "learning_rate": 6.892728237350408e-06, "loss": 0.3781, "step": 7565 }, { "epoch": 0.3827204208609439, "grad_norm": 6.479585647583008, "learning_rate": 6.891985327175316e-06, "loss": 0.471, "step": 7566 }, { "epoch": 0.3827710051090091, "grad_norm": 10.323261260986328, "learning_rate": 6.891242368248335e-06, "loss": 0.4264, "step": 7567 }, { "epoch": 0.3828215893570742, "grad_norm": 9.261945724487305, "learning_rate": 6.890499360588605e-06, "loss": 0.4526, "step": 7568 }, { "epoch": 0.38287217360513937, "grad_norm": 15.690455436706543, "learning_rate": 6.889756304215273e-06, "loss": 0.4767, "step": 7569 }, { "epoch": 0.3829227578532045, "grad_norm": 7.005249977111816, "learning_rate": 6.889013199147485e-06, "loss": 0.4312, "step": 7570 }, { "epoch": 0.38297334210126965, "grad_norm": 6.051903247833252, "learning_rate": 6.888270045404391e-06, "loss": 0.4286, "step": 7571 }, { "epoch": 0.3830239263493348, "grad_norm": 5.925224304199219, "learning_rate": 6.887526843005137e-06, "loss": 0.3263, "step": 7572 }, { "epoch": 0.38307451059739994, "grad_norm": 5.77925968170166, "learning_rate": 6.886783591968875e-06, "loss": 0.4189, "step": 7573 }, { "epoch": 0.38312509484546514, "grad_norm": 13.223135948181152, "learning_rate": 6.886040292314758e-06, "loss": 0.4504, "step": 7574 }, { "epoch": 0.3831756790935303, "grad_norm": 11.950836181640625, "learning_rate": 6.88529694406194e-06, "loss": 0.4372, "step": 7575 }, { "epoch": 0.38322626334159543, "grad_norm": 8.809745788574219, "learning_rate": 6.884553547229571e-06, "loss": 0.6052, "step": 7576 }, { "epoch": 0.3832768475896606, "grad_norm": 10.672477722167969, "learning_rate": 6.883810101836811e-06, "loss": 0.434, "step": 7577 }, { "epoch": 0.3833274318377257, "grad_norm": 8.061239242553711, "learning_rate": 6.883066607902813e-06, "loss": 0.4691, "step": 7578 }, { "epoch": 0.38337801608579086, "grad_norm": 6.826549053192139, "learning_rate": 6.882323065446737e-06, "loss": 0.452, "step": 7579 }, { "epoch": 0.38342860033385606, "grad_norm": 7.6065826416015625, "learning_rate": 6.881579474487744e-06, "loss": 0.3917, "step": 7580 }, { "epoch": 0.3834791845819212, "grad_norm": 13.704671859741211, "learning_rate": 6.880835835044993e-06, "loss": 0.6179, "step": 7581 }, { "epoch": 0.38352976882998635, "grad_norm": 5.6312713623046875, "learning_rate": 6.880092147137643e-06, "loss": 0.4202, "step": 7582 }, { "epoch": 0.3835803530780515, "grad_norm": 5.847779750823975, "learning_rate": 6.87934841078486e-06, "loss": 0.3152, "step": 7583 }, { "epoch": 0.38363093732611664, "grad_norm": 4.720727920532227, "learning_rate": 6.878604626005809e-06, "loss": 0.3902, "step": 7584 }, { "epoch": 0.3836815215741818, "grad_norm": 7.039353847503662, "learning_rate": 6.8778607928196525e-06, "loss": 0.4797, "step": 7585 }, { "epoch": 0.3837321058222469, "grad_norm": 10.866493225097656, "learning_rate": 6.8771169112455614e-06, "loss": 0.4316, "step": 7586 }, { "epoch": 0.3837826900703121, "grad_norm": 7.0012898445129395, "learning_rate": 6.876372981302702e-06, "loss": 0.5458, "step": 7587 }, { "epoch": 0.38383327431837727, "grad_norm": 6.569557189941406, "learning_rate": 6.87562900301024e-06, "loss": 0.4637, "step": 7588 }, { "epoch": 0.3838838585664424, "grad_norm": 7.616061687469482, "learning_rate": 6.874884976387351e-06, "loss": 0.3505, "step": 7589 }, { "epoch": 0.38393444281450756, "grad_norm": 11.839214324951172, "learning_rate": 6.874140901453205e-06, "loss": 0.4821, "step": 7590 }, { "epoch": 0.3839850270625727, "grad_norm": 13.678557395935059, "learning_rate": 6.873396778226973e-06, "loss": 0.4862, "step": 7591 }, { "epoch": 0.38403561131063785, "grad_norm": 12.161885261535645, "learning_rate": 6.872652606727833e-06, "loss": 0.5071, "step": 7592 }, { "epoch": 0.38408619555870305, "grad_norm": 5.423607349395752, "learning_rate": 6.871908386974957e-06, "loss": 0.3922, "step": 7593 }, { "epoch": 0.3841367798067682, "grad_norm": 11.728411674499512, "learning_rate": 6.871164118987525e-06, "loss": 0.4848, "step": 7594 }, { "epoch": 0.38418736405483334, "grad_norm": 7.670047283172607, "learning_rate": 6.870419802784712e-06, "loss": 0.4473, "step": 7595 }, { "epoch": 0.3842379483028985, "grad_norm": 7.109622955322266, "learning_rate": 6.869675438385699e-06, "loss": 0.3706, "step": 7596 }, { "epoch": 0.3842885325509636, "grad_norm": 13.054713249206543, "learning_rate": 6.868931025809665e-06, "loss": 0.5592, "step": 7597 }, { "epoch": 0.38433911679902877, "grad_norm": 5.105068683624268, "learning_rate": 6.868186565075794e-06, "loss": 0.34, "step": 7598 }, { "epoch": 0.3843897010470939, "grad_norm": 6.74371862411499, "learning_rate": 6.867442056203267e-06, "loss": 0.3402, "step": 7599 }, { "epoch": 0.3844402852951591, "grad_norm": 6.146301746368408, "learning_rate": 6.866697499211269e-06, "loss": 0.4237, "step": 7600 }, { "epoch": 0.38449086954322426, "grad_norm": 10.80604362487793, "learning_rate": 6.865952894118986e-06, "loss": 0.4953, "step": 7601 }, { "epoch": 0.3845414537912894, "grad_norm": 6.593811988830566, "learning_rate": 6.865208240945602e-06, "loss": 0.4303, "step": 7602 }, { "epoch": 0.38459203803935454, "grad_norm": 5.822253704071045, "learning_rate": 6.864463539710308e-06, "loss": 0.452, "step": 7603 }, { "epoch": 0.3846426222874197, "grad_norm": 13.523627281188965, "learning_rate": 6.863718790432292e-06, "loss": 0.5731, "step": 7604 }, { "epoch": 0.38469320653548483, "grad_norm": 7.458193302154541, "learning_rate": 6.862973993130745e-06, "loss": 0.4272, "step": 7605 }, { "epoch": 0.38474379078355, "grad_norm": 7.634190082550049, "learning_rate": 6.862229147824857e-06, "loss": 0.2738, "step": 7606 }, { "epoch": 0.3847943750316152, "grad_norm": 6.86320686340332, "learning_rate": 6.861484254533822e-06, "loss": 0.3485, "step": 7607 }, { "epoch": 0.3848449592796803, "grad_norm": 5.863392353057861, "learning_rate": 6.860739313276835e-06, "loss": 0.3546, "step": 7608 }, { "epoch": 0.38489554352774547, "grad_norm": 9.524243354797363, "learning_rate": 6.859994324073089e-06, "loss": 0.537, "step": 7609 }, { "epoch": 0.3849461277758106, "grad_norm": 13.791963577270508, "learning_rate": 6.859249286941783e-06, "loss": 0.5739, "step": 7610 }, { "epoch": 0.38499671202387575, "grad_norm": 13.670907020568848, "learning_rate": 6.858504201902113e-06, "loss": 0.6408, "step": 7611 }, { "epoch": 0.3850472962719409, "grad_norm": 5.671523571014404, "learning_rate": 6.857759068973279e-06, "loss": 0.3672, "step": 7612 }, { "epoch": 0.3850978805200061, "grad_norm": 16.829919815063477, "learning_rate": 6.8570138881744805e-06, "loss": 0.7859, "step": 7613 }, { "epoch": 0.38514846476807124, "grad_norm": 11.140496253967285, "learning_rate": 6.8562686595249205e-06, "loss": 0.5717, "step": 7614 }, { "epoch": 0.3851990490161364, "grad_norm": 7.1199188232421875, "learning_rate": 6.8555233830438006e-06, "loss": 0.4027, "step": 7615 }, { "epoch": 0.38524963326420153, "grad_norm": 4.059226989746094, "learning_rate": 6.8547780587503245e-06, "loss": 0.349, "step": 7616 }, { "epoch": 0.3853002175122667, "grad_norm": 5.677507400512695, "learning_rate": 6.854032686663698e-06, "loss": 0.3857, "step": 7617 }, { "epoch": 0.3853508017603318, "grad_norm": 8.880600929260254, "learning_rate": 6.853287266803127e-06, "loss": 0.4628, "step": 7618 }, { "epoch": 0.38540138600839696, "grad_norm": 7.651552677154541, "learning_rate": 6.85254179918782e-06, "loss": 0.4625, "step": 7619 }, { "epoch": 0.38545197025646216, "grad_norm": 9.73135757446289, "learning_rate": 6.851796283836986e-06, "loss": 0.5251, "step": 7620 }, { "epoch": 0.3855025545045273, "grad_norm": 9.784067153930664, "learning_rate": 6.851050720769834e-06, "loss": 0.3909, "step": 7621 }, { "epoch": 0.38555313875259245, "grad_norm": 8.731283187866211, "learning_rate": 6.8503051100055764e-06, "loss": 0.5349, "step": 7622 }, { "epoch": 0.3856037230006576, "grad_norm": 16.20456886291504, "learning_rate": 6.8495594515634235e-06, "loss": 0.461, "step": 7623 }, { "epoch": 0.38565430724872274, "grad_norm": 5.950830936431885, "learning_rate": 6.848813745462593e-06, "loss": 0.3994, "step": 7624 }, { "epoch": 0.3857048914967879, "grad_norm": 6.707911014556885, "learning_rate": 6.848067991722297e-06, "loss": 0.3774, "step": 7625 }, { "epoch": 0.385755475744853, "grad_norm": 5.34788703918457, "learning_rate": 6.847322190361754e-06, "loss": 0.4557, "step": 7626 }, { "epoch": 0.3858060599929182, "grad_norm": 7.053834438323975, "learning_rate": 6.846576341400179e-06, "loss": 0.5081, "step": 7627 }, { "epoch": 0.38585664424098337, "grad_norm": 12.881956100463867, "learning_rate": 6.845830444856791e-06, "loss": 0.4631, "step": 7628 }, { "epoch": 0.3859072284890485, "grad_norm": 8.130293846130371, "learning_rate": 6.8450845007508126e-06, "loss": 0.4348, "step": 7629 }, { "epoch": 0.38595781273711366, "grad_norm": 11.950116157531738, "learning_rate": 6.844338509101462e-06, "loss": 0.4878, "step": 7630 }, { "epoch": 0.3860083969851788, "grad_norm": 7.233832836151123, "learning_rate": 6.843592469927964e-06, "loss": 0.4336, "step": 7631 }, { "epoch": 0.38605898123324395, "grad_norm": 9.042679786682129, "learning_rate": 6.8428463832495395e-06, "loss": 0.4738, "step": 7632 }, { "epoch": 0.38610956548130915, "grad_norm": 5.6822285652160645, "learning_rate": 6.842100249085417e-06, "loss": 0.394, "step": 7633 }, { "epoch": 0.3861601497293743, "grad_norm": 5.960813999176025, "learning_rate": 6.841354067454819e-06, "loss": 0.36, "step": 7634 }, { "epoch": 0.38621073397743944, "grad_norm": 8.981549263000488, "learning_rate": 6.840607838376975e-06, "loss": 0.5528, "step": 7635 }, { "epoch": 0.3862613182255046, "grad_norm": 26.17678451538086, "learning_rate": 6.839861561871112e-06, "loss": 0.5041, "step": 7636 }, { "epoch": 0.3863119024735697, "grad_norm": 6.584303379058838, "learning_rate": 6.8391152379564615e-06, "loss": 0.4122, "step": 7637 }, { "epoch": 0.38636248672163487, "grad_norm": 13.232683181762695, "learning_rate": 6.838368866652253e-06, "loss": 0.4771, "step": 7638 }, { "epoch": 0.3864130709697, "grad_norm": 9.252927780151367, "learning_rate": 6.83762244797772e-06, "loss": 0.421, "step": 7639 }, { "epoch": 0.3864636552177652, "grad_norm": 8.15306568145752, "learning_rate": 6.836875981952094e-06, "loss": 0.4028, "step": 7640 }, { "epoch": 0.38651423946583036, "grad_norm": 10.306376457214355, "learning_rate": 6.8361294685946114e-06, "loss": 0.4247, "step": 7641 }, { "epoch": 0.3865648237138955, "grad_norm": 6.657684803009033, "learning_rate": 6.835382907924506e-06, "loss": 0.3004, "step": 7642 }, { "epoch": 0.38661540796196064, "grad_norm": 8.917852401733398, "learning_rate": 6.834636299961018e-06, "loss": 0.3638, "step": 7643 }, { "epoch": 0.3866659922100258, "grad_norm": 6.770850658416748, "learning_rate": 6.833889644723382e-06, "loss": 0.4606, "step": 7644 }, { "epoch": 0.38671657645809093, "grad_norm": 10.053492546081543, "learning_rate": 6.833142942230842e-06, "loss": 0.4293, "step": 7645 }, { "epoch": 0.3867671607061561, "grad_norm": 7.643006801605225, "learning_rate": 6.832396192502635e-06, "loss": 0.4201, "step": 7646 }, { "epoch": 0.3868177449542213, "grad_norm": 12.803149223327637, "learning_rate": 6.831649395558005e-06, "loss": 0.6428, "step": 7647 }, { "epoch": 0.3868683292022864, "grad_norm": 7.467730522155762, "learning_rate": 6.830902551416192e-06, "loss": 0.5395, "step": 7648 }, { "epoch": 0.38691891345035156, "grad_norm": 9.013041496276855, "learning_rate": 6.830155660096444e-06, "loss": 0.4975, "step": 7649 }, { "epoch": 0.3869694976984167, "grad_norm": 8.813968658447266, "learning_rate": 6.8294087216180045e-06, "loss": 0.4036, "step": 7650 }, { "epoch": 0.38702008194648185, "grad_norm": 5.577877998352051, "learning_rate": 6.828661736000122e-06, "loss": 0.4, "step": 7651 }, { "epoch": 0.387070666194547, "grad_norm": 8.15248966217041, "learning_rate": 6.827914703262043e-06, "loss": 0.3469, "step": 7652 }, { "epoch": 0.3871212504426122, "grad_norm": 7.612930774688721, "learning_rate": 6.827167623423018e-06, "loss": 0.4722, "step": 7653 }, { "epoch": 0.38717183469067734, "grad_norm": 5.834957599639893, "learning_rate": 6.826420496502295e-06, "loss": 0.4138, "step": 7654 }, { "epoch": 0.3872224189387425, "grad_norm": 10.59253215789795, "learning_rate": 6.825673322519129e-06, "loss": 0.4964, "step": 7655 }, { "epoch": 0.38727300318680763, "grad_norm": 8.253640174865723, "learning_rate": 6.82492610149277e-06, "loss": 0.422, "step": 7656 }, { "epoch": 0.3873235874348728, "grad_norm": 4.844549655914307, "learning_rate": 6.824178833442473e-06, "loss": 0.3885, "step": 7657 }, { "epoch": 0.3873741716829379, "grad_norm": 7.4770050048828125, "learning_rate": 6.8234315183874955e-06, "loss": 0.4761, "step": 7658 }, { "epoch": 0.38742475593100306, "grad_norm": 6.566333293914795, "learning_rate": 6.822684156347092e-06, "loss": 0.3805, "step": 7659 }, { "epoch": 0.38747534017906826, "grad_norm": 5.531732082366943, "learning_rate": 6.8219367473405185e-06, "loss": 0.3161, "step": 7660 }, { "epoch": 0.3875259244271334, "grad_norm": 7.184327602386475, "learning_rate": 6.821189291387036e-06, "loss": 0.4823, "step": 7661 }, { "epoch": 0.38757650867519855, "grad_norm": 6.423909664154053, "learning_rate": 6.820441788505905e-06, "loss": 0.4175, "step": 7662 }, { "epoch": 0.3876270929232637, "grad_norm": 17.54833984375, "learning_rate": 6.8196942387163855e-06, "loss": 0.6915, "step": 7663 }, { "epoch": 0.38767767717132884, "grad_norm": 6.45382022857666, "learning_rate": 6.81894664203774e-06, "loss": 0.3625, "step": 7664 }, { "epoch": 0.387728261419394, "grad_norm": 4.543482780456543, "learning_rate": 6.818198998489236e-06, "loss": 0.4313, "step": 7665 }, { "epoch": 0.3877788456674591, "grad_norm": 6.719714164733887, "learning_rate": 6.817451308090134e-06, "loss": 0.4616, "step": 7666 }, { "epoch": 0.3878294299155243, "grad_norm": 6.891731262207031, "learning_rate": 6.816703570859701e-06, "loss": 0.5021, "step": 7667 }, { "epoch": 0.38788001416358947, "grad_norm": 8.80501937866211, "learning_rate": 6.815955786817206e-06, "loss": 0.4392, "step": 7668 }, { "epoch": 0.3879305984116546, "grad_norm": 6.901387691497803, "learning_rate": 6.815207955981916e-06, "loss": 0.3745, "step": 7669 }, { "epoch": 0.38798118265971976, "grad_norm": 6.5457234382629395, "learning_rate": 6.814460078373101e-06, "loss": 0.3822, "step": 7670 }, { "epoch": 0.3880317669077849, "grad_norm": 11.40732192993164, "learning_rate": 6.8137121540100334e-06, "loss": 0.4833, "step": 7671 }, { "epoch": 0.38808235115585005, "grad_norm": 14.662036895751953, "learning_rate": 6.812964182911985e-06, "loss": 0.53, "step": 7672 }, { "epoch": 0.38813293540391525, "grad_norm": 6.350848197937012, "learning_rate": 6.812216165098227e-06, "loss": 0.429, "step": 7673 }, { "epoch": 0.3881835196519804, "grad_norm": 5.659634113311768, "learning_rate": 6.811468100588036e-06, "loss": 0.3611, "step": 7674 }, { "epoch": 0.38823410390004554, "grad_norm": 8.040172576904297, "learning_rate": 6.810719989400688e-06, "loss": 0.4156, "step": 7675 }, { "epoch": 0.3882846881481107, "grad_norm": 12.256255149841309, "learning_rate": 6.80997183155546e-06, "loss": 0.591, "step": 7676 }, { "epoch": 0.3883352723961758, "grad_norm": 7.101499080657959, "learning_rate": 6.809223627071629e-06, "loss": 0.4351, "step": 7677 }, { "epoch": 0.38838585664424097, "grad_norm": 9.148015975952148, "learning_rate": 6.808475375968475e-06, "loss": 0.425, "step": 7678 }, { "epoch": 0.3884364408923061, "grad_norm": 9.138928413391113, "learning_rate": 6.807727078265279e-06, "loss": 0.4655, "step": 7679 }, { "epoch": 0.3884870251403713, "grad_norm": 6.751618385314941, "learning_rate": 6.8069787339813224e-06, "loss": 0.3387, "step": 7680 }, { "epoch": 0.38853760938843646, "grad_norm": 18.18988609313965, "learning_rate": 6.806230343135889e-06, "loss": 0.552, "step": 7681 }, { "epoch": 0.3885881936365016, "grad_norm": 6.9569993019104, "learning_rate": 6.8054819057482615e-06, "loss": 0.4253, "step": 7682 }, { "epoch": 0.38863877788456674, "grad_norm": 11.582972526550293, "learning_rate": 6.804733421837726e-06, "loss": 0.3211, "step": 7683 }, { "epoch": 0.3886893621326319, "grad_norm": 21.91461753845215, "learning_rate": 6.8039848914235715e-06, "loss": 0.5197, "step": 7684 }, { "epoch": 0.38873994638069703, "grad_norm": 7.156205654144287, "learning_rate": 6.803236314525081e-06, "loss": 0.4024, "step": 7685 }, { "epoch": 0.3887905306287622, "grad_norm": 5.6363983154296875, "learning_rate": 6.802487691161548e-06, "loss": 0.3757, "step": 7686 }, { "epoch": 0.3888411148768274, "grad_norm": 3.6955952644348145, "learning_rate": 6.80173902135226e-06, "loss": 0.2929, "step": 7687 }, { "epoch": 0.3888916991248925, "grad_norm": 6.835087776184082, "learning_rate": 6.800990305116511e-06, "loss": 0.3969, "step": 7688 }, { "epoch": 0.38894228337295766, "grad_norm": 12.60265827178955, "learning_rate": 6.800241542473589e-06, "loss": 0.3688, "step": 7689 }, { "epoch": 0.3889928676210228, "grad_norm": 11.127192497253418, "learning_rate": 6.799492733442794e-06, "loss": 0.5619, "step": 7690 }, { "epoch": 0.38904345186908795, "grad_norm": 8.216744422912598, "learning_rate": 6.798743878043416e-06, "loss": 0.5351, "step": 7691 }, { "epoch": 0.3890940361171531, "grad_norm": 9.888434410095215, "learning_rate": 6.797994976294753e-06, "loss": 0.4186, "step": 7692 }, { "epoch": 0.3891446203652183, "grad_norm": 5.696649551391602, "learning_rate": 6.797246028216102e-06, "loss": 0.3955, "step": 7693 }, { "epoch": 0.38919520461328344, "grad_norm": 11.072778701782227, "learning_rate": 6.796497033826763e-06, "loss": 0.5184, "step": 7694 }, { "epoch": 0.3892457888613486, "grad_norm": 6.490318298339844, "learning_rate": 6.7957479931460335e-06, "loss": 0.4728, "step": 7695 }, { "epoch": 0.38929637310941373, "grad_norm": 8.293394088745117, "learning_rate": 6.794998906193216e-06, "loss": 0.4557, "step": 7696 }, { "epoch": 0.3893469573574789, "grad_norm": 7.958040237426758, "learning_rate": 6.794249772987613e-06, "loss": 0.4218, "step": 7697 }, { "epoch": 0.389397541605544, "grad_norm": 7.687464237213135, "learning_rate": 6.793500593548528e-06, "loss": 0.4476, "step": 7698 }, { "epoch": 0.38944812585360916, "grad_norm": 13.686935424804688, "learning_rate": 6.792751367895263e-06, "loss": 0.474, "step": 7699 }, { "epoch": 0.38949871010167436, "grad_norm": 8.225151062011719, "learning_rate": 6.792002096047126e-06, "loss": 0.4068, "step": 7700 }, { "epoch": 0.3895492943497395, "grad_norm": 7.469467639923096, "learning_rate": 6.791252778023423e-06, "loss": 0.3307, "step": 7701 }, { "epoch": 0.38959987859780465, "grad_norm": 8.143131256103516, "learning_rate": 6.790503413843462e-06, "loss": 0.4255, "step": 7702 }, { "epoch": 0.3896504628458698, "grad_norm": 7.340152263641357, "learning_rate": 6.789754003526555e-06, "loss": 0.4898, "step": 7703 }, { "epoch": 0.38970104709393494, "grad_norm": 25.44928550720215, "learning_rate": 6.789004547092008e-06, "loss": 0.4618, "step": 7704 }, { "epoch": 0.3897516313420001, "grad_norm": 6.502127170562744, "learning_rate": 6.7882550445591365e-06, "loss": 0.2965, "step": 7705 }, { "epoch": 0.3898022155900652, "grad_norm": 8.981977462768555, "learning_rate": 6.787505495947252e-06, "loss": 0.4989, "step": 7706 }, { "epoch": 0.3898527998381304, "grad_norm": 8.632796287536621, "learning_rate": 6.786755901275668e-06, "loss": 0.4801, "step": 7707 }, { "epoch": 0.38990338408619557, "grad_norm": 7.504095554351807, "learning_rate": 6.786006260563699e-06, "loss": 0.4363, "step": 7708 }, { "epoch": 0.3899539683342607, "grad_norm": 5.482292175292969, "learning_rate": 6.785256573830663e-06, "loss": 0.4126, "step": 7709 }, { "epoch": 0.39000455258232586, "grad_norm": 6.699093341827393, "learning_rate": 6.78450684109588e-06, "loss": 0.4318, "step": 7710 }, { "epoch": 0.390055136830391, "grad_norm": 6.423379898071289, "learning_rate": 6.783757062378662e-06, "loss": 0.4784, "step": 7711 }, { "epoch": 0.39010572107845615, "grad_norm": 5.541247844696045, "learning_rate": 6.7830072376983355e-06, "loss": 0.3714, "step": 7712 }, { "epoch": 0.39015630532652135, "grad_norm": 13.566838264465332, "learning_rate": 6.782257367074217e-06, "loss": 0.4472, "step": 7713 }, { "epoch": 0.3902068895745865, "grad_norm": 7.055011749267578, "learning_rate": 6.781507450525633e-06, "loss": 0.3513, "step": 7714 }, { "epoch": 0.39025747382265163, "grad_norm": 6.788811206817627, "learning_rate": 6.780757488071905e-06, "loss": 0.4327, "step": 7715 }, { "epoch": 0.3903080580707168, "grad_norm": 4.9634809494018555, "learning_rate": 6.780007479732358e-06, "loss": 0.2745, "step": 7716 }, { "epoch": 0.3903586423187819, "grad_norm": 7.180119514465332, "learning_rate": 6.7792574255263154e-06, "loss": 0.4236, "step": 7717 }, { "epoch": 0.39040922656684707, "grad_norm": 6.908402919769287, "learning_rate": 6.778507325473108e-06, "loss": 0.4275, "step": 7718 }, { "epoch": 0.3904598108149122, "grad_norm": 9.195338249206543, "learning_rate": 6.777757179592064e-06, "loss": 0.4645, "step": 7719 }, { "epoch": 0.3905103950629774, "grad_norm": 8.928738594055176, "learning_rate": 6.777006987902509e-06, "loss": 0.4034, "step": 7720 }, { "epoch": 0.39056097931104256, "grad_norm": 6.378804683685303, "learning_rate": 6.776256750423776e-06, "loss": 0.3863, "step": 7721 }, { "epoch": 0.3906115635591077, "grad_norm": 4.884373188018799, "learning_rate": 6.775506467175199e-06, "loss": 0.35, "step": 7722 }, { "epoch": 0.39066214780717284, "grad_norm": 6.732440948486328, "learning_rate": 6.774756138176108e-06, "loss": 0.3502, "step": 7723 }, { "epoch": 0.390712732055238, "grad_norm": 12.212295532226562, "learning_rate": 6.774005763445839e-06, "loss": 0.4207, "step": 7724 }, { "epoch": 0.39076331630330313, "grad_norm": 5.760292053222656, "learning_rate": 6.7732553430037264e-06, "loss": 0.4542, "step": 7725 }, { "epoch": 0.3908139005513683, "grad_norm": 9.678166389465332, "learning_rate": 6.772504876869105e-06, "loss": 0.553, "step": 7726 }, { "epoch": 0.3908644847994335, "grad_norm": 11.779495239257812, "learning_rate": 6.771754365061314e-06, "loss": 0.4668, "step": 7727 }, { "epoch": 0.3909150690474986, "grad_norm": 6.8801140785217285, "learning_rate": 6.771003807599694e-06, "loss": 0.4443, "step": 7728 }, { "epoch": 0.39096565329556376, "grad_norm": 8.748627662658691, "learning_rate": 6.770253204503584e-06, "loss": 0.4354, "step": 7729 }, { "epoch": 0.3910162375436289, "grad_norm": 6.544787883758545, "learning_rate": 6.769502555792324e-06, "loss": 0.4609, "step": 7730 }, { "epoch": 0.39106682179169405, "grad_norm": 7.1974945068359375, "learning_rate": 6.768751861485258e-06, "loss": 0.4005, "step": 7731 }, { "epoch": 0.3911174060397592, "grad_norm": 6.551551342010498, "learning_rate": 6.7680011216017275e-06, "loss": 0.4524, "step": 7732 }, { "epoch": 0.3911679902878244, "grad_norm": 8.510968208312988, "learning_rate": 6.7672503361610775e-06, "loss": 0.4649, "step": 7733 }, { "epoch": 0.39121857453588954, "grad_norm": 5.320348262786865, "learning_rate": 6.766499505182657e-06, "loss": 0.425, "step": 7734 }, { "epoch": 0.3912691587839547, "grad_norm": 13.965524673461914, "learning_rate": 6.765748628685812e-06, "loss": 0.5485, "step": 7735 }, { "epoch": 0.39131974303201983, "grad_norm": 9.411256790161133, "learning_rate": 6.7649977066898875e-06, "loss": 0.3849, "step": 7736 }, { "epoch": 0.391370327280085, "grad_norm": 12.111543655395508, "learning_rate": 6.764246739214236e-06, "loss": 0.5191, "step": 7737 }, { "epoch": 0.3914209115281501, "grad_norm": 8.002705574035645, "learning_rate": 6.763495726278208e-06, "loss": 0.4871, "step": 7738 }, { "epoch": 0.39147149577621526, "grad_norm": 6.852802753448486, "learning_rate": 6.762744667901153e-06, "loss": 0.4093, "step": 7739 }, { "epoch": 0.39152208002428046, "grad_norm": 7.133182048797607, "learning_rate": 6.761993564102427e-06, "loss": 0.4366, "step": 7740 }, { "epoch": 0.3915726642723456, "grad_norm": 8.288147926330566, "learning_rate": 6.761242414901383e-06, "loss": 0.4966, "step": 7741 }, { "epoch": 0.39162324852041075, "grad_norm": 10.762777328491211, "learning_rate": 6.7604912203173754e-06, "loss": 0.4834, "step": 7742 }, { "epoch": 0.3916738327684759, "grad_norm": 10.579426765441895, "learning_rate": 6.759739980369762e-06, "loss": 0.4029, "step": 7743 }, { "epoch": 0.39172441701654104, "grad_norm": 10.054239273071289, "learning_rate": 6.758988695077899e-06, "loss": 0.3716, "step": 7744 }, { "epoch": 0.3917750012646062, "grad_norm": 4.93813943862915, "learning_rate": 6.758237364461146e-06, "loss": 0.4375, "step": 7745 }, { "epoch": 0.3918255855126714, "grad_norm": 9.522631645202637, "learning_rate": 6.757485988538863e-06, "loss": 0.4899, "step": 7746 }, { "epoch": 0.3918761697607365, "grad_norm": 13.883734703063965, "learning_rate": 6.756734567330413e-06, "loss": 0.4435, "step": 7747 }, { "epoch": 0.39192675400880167, "grad_norm": 7.564229488372803, "learning_rate": 6.755983100855155e-06, "loss": 0.3839, "step": 7748 }, { "epoch": 0.3919773382568668, "grad_norm": 10.136404037475586, "learning_rate": 6.755231589132454e-06, "loss": 0.3993, "step": 7749 }, { "epoch": 0.39202792250493196, "grad_norm": 25.98173713684082, "learning_rate": 6.754480032181674e-06, "loss": 0.4957, "step": 7750 }, { "epoch": 0.3920785067529971, "grad_norm": 10.341848373413086, "learning_rate": 6.7537284300221815e-06, "loss": 0.4759, "step": 7751 }, { "epoch": 0.39212909100106225, "grad_norm": 15.59365177154541, "learning_rate": 6.752976782673343e-06, "loss": 0.56, "step": 7752 }, { "epoch": 0.39217967524912745, "grad_norm": 8.850019454956055, "learning_rate": 6.752225090154527e-06, "loss": 0.2888, "step": 7753 }, { "epoch": 0.3922302594971926, "grad_norm": 5.645358562469482, "learning_rate": 6.751473352485104e-06, "loss": 0.3827, "step": 7754 }, { "epoch": 0.39228084374525773, "grad_norm": 6.1129045486450195, "learning_rate": 6.750721569684442e-06, "loss": 0.5062, "step": 7755 }, { "epoch": 0.3923314279933229, "grad_norm": 7.9959917068481445, "learning_rate": 6.749969741771915e-06, "loss": 0.4152, "step": 7756 }, { "epoch": 0.392382012241388, "grad_norm": 23.223997116088867, "learning_rate": 6.749217868766894e-06, "loss": 0.5374, "step": 7757 }, { "epoch": 0.39243259648945317, "grad_norm": 10.055541038513184, "learning_rate": 6.748465950688753e-06, "loss": 0.4913, "step": 7758 }, { "epoch": 0.3924831807375183, "grad_norm": 10.352240562438965, "learning_rate": 6.747713987556867e-06, "loss": 0.4288, "step": 7759 }, { "epoch": 0.3925337649855835, "grad_norm": 6.729286193847656, "learning_rate": 6.7469619793906145e-06, "loss": 0.3677, "step": 7760 }, { "epoch": 0.39258434923364866, "grad_norm": 10.138834953308105, "learning_rate": 6.746209926209373e-06, "loss": 0.3838, "step": 7761 }, { "epoch": 0.3926349334817138, "grad_norm": 8.408227920532227, "learning_rate": 6.745457828032516e-06, "loss": 0.4391, "step": 7762 }, { "epoch": 0.39268551772977894, "grad_norm": 9.939213752746582, "learning_rate": 6.744705684879429e-06, "loss": 0.3817, "step": 7763 }, { "epoch": 0.3927361019778441, "grad_norm": 6.329154014587402, "learning_rate": 6.74395349676949e-06, "loss": 0.4828, "step": 7764 }, { "epoch": 0.39278668622590923, "grad_norm": 5.87501859664917, "learning_rate": 6.743201263722081e-06, "loss": 0.3676, "step": 7765 }, { "epoch": 0.39283727047397443, "grad_norm": 7.880684852600098, "learning_rate": 6.742448985756587e-06, "loss": 0.4419, "step": 7766 }, { "epoch": 0.3928878547220396, "grad_norm": 10.511953353881836, "learning_rate": 6.741696662892391e-06, "loss": 0.4832, "step": 7767 }, { "epoch": 0.3929384389701047, "grad_norm": 17.66217613220215, "learning_rate": 6.740944295148879e-06, "loss": 0.586, "step": 7768 }, { "epoch": 0.39298902321816986, "grad_norm": 13.695977210998535, "learning_rate": 6.740191882545438e-06, "loss": 0.4824, "step": 7769 }, { "epoch": 0.393039607466235, "grad_norm": 8.009742736816406, "learning_rate": 6.739439425101455e-06, "loss": 0.3832, "step": 7770 }, { "epoch": 0.39309019171430015, "grad_norm": 6.64504861831665, "learning_rate": 6.738686922836318e-06, "loss": 0.4332, "step": 7771 }, { "epoch": 0.3931407759623653, "grad_norm": 5.793552875518799, "learning_rate": 6.737934375769421e-06, "loss": 0.4397, "step": 7772 }, { "epoch": 0.3931913602104305, "grad_norm": 9.83305549621582, "learning_rate": 6.7371817839201515e-06, "loss": 0.5968, "step": 7773 }, { "epoch": 0.39324194445849564, "grad_norm": 9.753547668457031, "learning_rate": 6.7364291473079045e-06, "loss": 0.3861, "step": 7774 }, { "epoch": 0.3932925287065608, "grad_norm": 6.066319942474365, "learning_rate": 6.735676465952071e-06, "loss": 0.3735, "step": 7775 }, { "epoch": 0.39334311295462593, "grad_norm": 10.912406921386719, "learning_rate": 6.734923739872049e-06, "loss": 0.5329, "step": 7776 }, { "epoch": 0.3933936972026911, "grad_norm": 6.441005706787109, "learning_rate": 6.7341709690872316e-06, "loss": 0.3747, "step": 7777 }, { "epoch": 0.3934442814507562, "grad_norm": 9.821273803710938, "learning_rate": 6.733418153617018e-06, "loss": 0.3981, "step": 7778 }, { "epoch": 0.39349486569882136, "grad_norm": 10.394594192504883, "learning_rate": 6.732665293480805e-06, "loss": 0.5779, "step": 7779 }, { "epoch": 0.39354544994688656, "grad_norm": 6.834266185760498, "learning_rate": 6.731912388697992e-06, "loss": 0.4432, "step": 7780 }, { "epoch": 0.3935960341949517, "grad_norm": 7.651628494262695, "learning_rate": 6.73115943928798e-06, "loss": 0.4277, "step": 7781 }, { "epoch": 0.39364661844301685, "grad_norm": 8.459996223449707, "learning_rate": 6.7304064452701725e-06, "loss": 0.4866, "step": 7782 }, { "epoch": 0.393697202691082, "grad_norm": 5.8623948097229, "learning_rate": 6.729653406663969e-06, "loss": 0.3993, "step": 7783 }, { "epoch": 0.39374778693914714, "grad_norm": 6.333306789398193, "learning_rate": 6.728900323488774e-06, "loss": 0.415, "step": 7784 }, { "epoch": 0.3937983711872123, "grad_norm": 13.314870834350586, "learning_rate": 6.728147195763995e-06, "loss": 0.3094, "step": 7785 }, { "epoch": 0.3938489554352775, "grad_norm": 7.562681674957275, "learning_rate": 6.7273940235090365e-06, "loss": 0.3994, "step": 7786 }, { "epoch": 0.3938995396833426, "grad_norm": 9.64599895477295, "learning_rate": 6.726640806743308e-06, "loss": 0.4561, "step": 7787 }, { "epoch": 0.39395012393140777, "grad_norm": 9.774124145507812, "learning_rate": 6.725887545486215e-06, "loss": 0.384, "step": 7788 }, { "epoch": 0.3940007081794729, "grad_norm": 16.097814559936523, "learning_rate": 6.725134239757167e-06, "loss": 0.4781, "step": 7789 }, { "epoch": 0.39405129242753806, "grad_norm": 7.206085681915283, "learning_rate": 6.724380889575579e-06, "loss": 0.3306, "step": 7790 }, { "epoch": 0.3941018766756032, "grad_norm": 20.200328826904297, "learning_rate": 6.72362749496086e-06, "loss": 0.6348, "step": 7791 }, { "epoch": 0.39415246092366835, "grad_norm": 9.393414497375488, "learning_rate": 6.722874055932423e-06, "loss": 0.4597, "step": 7792 }, { "epoch": 0.39420304517173355, "grad_norm": 7.49368143081665, "learning_rate": 6.7221205725096835e-06, "loss": 0.369, "step": 7793 }, { "epoch": 0.3942536294197987, "grad_norm": 9.346161842346191, "learning_rate": 6.7213670447120574e-06, "loss": 0.3919, "step": 7794 }, { "epoch": 0.39430421366786383, "grad_norm": 9.340819358825684, "learning_rate": 6.720613472558959e-06, "loss": 0.4235, "step": 7795 }, { "epoch": 0.394354797915929, "grad_norm": 7.396360874176025, "learning_rate": 6.719859856069808e-06, "loss": 0.3714, "step": 7796 }, { "epoch": 0.3944053821639941, "grad_norm": 9.91458797454834, "learning_rate": 6.719106195264022e-06, "loss": 0.4776, "step": 7797 }, { "epoch": 0.39445596641205927, "grad_norm": 7.002449035644531, "learning_rate": 6.718352490161022e-06, "loss": 0.3509, "step": 7798 }, { "epoch": 0.3945065506601244, "grad_norm": 12.406329154968262, "learning_rate": 6.717598740780229e-06, "loss": 0.5174, "step": 7799 }, { "epoch": 0.3945571349081896, "grad_norm": 7.198647499084473, "learning_rate": 6.716844947141066e-06, "loss": 0.3535, "step": 7800 }, { "epoch": 0.39460771915625475, "grad_norm": 8.316359519958496, "learning_rate": 6.716091109262954e-06, "loss": 0.3646, "step": 7801 }, { "epoch": 0.3946583034043199, "grad_norm": 7.094764709472656, "learning_rate": 6.715337227165319e-06, "loss": 0.4327, "step": 7802 }, { "epoch": 0.39470888765238504, "grad_norm": 11.389453887939453, "learning_rate": 6.714583300867588e-06, "loss": 0.5055, "step": 7803 }, { "epoch": 0.3947594719004502, "grad_norm": 8.793198585510254, "learning_rate": 6.713829330389187e-06, "loss": 0.4462, "step": 7804 }, { "epoch": 0.39481005614851533, "grad_norm": 7.937404632568359, "learning_rate": 6.713075315749543e-06, "loss": 0.4083, "step": 7805 }, { "epoch": 0.39486064039658053, "grad_norm": 8.48311710357666, "learning_rate": 6.712321256968086e-06, "loss": 0.441, "step": 7806 }, { "epoch": 0.3949112246446457, "grad_norm": 10.20197868347168, "learning_rate": 6.711567154064247e-06, "loss": 0.5431, "step": 7807 }, { "epoch": 0.3949618088927108, "grad_norm": 11.073038101196289, "learning_rate": 6.710813007057456e-06, "loss": 0.5765, "step": 7808 }, { "epoch": 0.39501239314077596, "grad_norm": 6.289562702178955, "learning_rate": 6.710058815967145e-06, "loss": 0.3658, "step": 7809 }, { "epoch": 0.3950629773888411, "grad_norm": 7.775409698486328, "learning_rate": 6.70930458081275e-06, "loss": 0.3916, "step": 7810 }, { "epoch": 0.39511356163690625, "grad_norm": 6.334543704986572, "learning_rate": 6.708550301613705e-06, "loss": 0.4467, "step": 7811 }, { "epoch": 0.3951641458849714, "grad_norm": 5.864577770233154, "learning_rate": 6.7077959783894446e-06, "loss": 0.389, "step": 7812 }, { "epoch": 0.3952147301330366, "grad_norm": 15.826212882995605, "learning_rate": 6.707041611159407e-06, "loss": 0.63, "step": 7813 }, { "epoch": 0.39526531438110174, "grad_norm": 7.750992298126221, "learning_rate": 6.706287199943031e-06, "loss": 0.4469, "step": 7814 }, { "epoch": 0.3953158986291669, "grad_norm": 15.034005165100098, "learning_rate": 6.705532744759754e-06, "loss": 0.6308, "step": 7815 }, { "epoch": 0.39536648287723203, "grad_norm": 6.971179008483887, "learning_rate": 6.704778245629019e-06, "loss": 0.4158, "step": 7816 }, { "epoch": 0.3954170671252972, "grad_norm": 5.7400736808776855, "learning_rate": 6.704023702570264e-06, "loss": 0.4292, "step": 7817 }, { "epoch": 0.3954676513733623, "grad_norm": 4.1447014808654785, "learning_rate": 6.703269115602936e-06, "loss": 0.3736, "step": 7818 }, { "epoch": 0.39551823562142746, "grad_norm": 7.740550994873047, "learning_rate": 6.702514484746478e-06, "loss": 0.4451, "step": 7819 }, { "epoch": 0.39556881986949266, "grad_norm": 6.627840518951416, "learning_rate": 6.701759810020333e-06, "loss": 0.4424, "step": 7820 }, { "epoch": 0.3956194041175578, "grad_norm": 12.666667938232422, "learning_rate": 6.7010050914439476e-06, "loss": 0.4796, "step": 7821 }, { "epoch": 0.39566998836562295, "grad_norm": 7.401062488555908, "learning_rate": 6.7002503290367694e-06, "loss": 0.4828, "step": 7822 }, { "epoch": 0.3957205726136881, "grad_norm": 7.043314456939697, "learning_rate": 6.699495522818248e-06, "loss": 0.3219, "step": 7823 }, { "epoch": 0.39577115686175324, "grad_norm": 8.878384590148926, "learning_rate": 6.698740672807831e-06, "loss": 0.4643, "step": 7824 }, { "epoch": 0.3958217411098184, "grad_norm": 16.445552825927734, "learning_rate": 6.697985779024969e-06, "loss": 0.45, "step": 7825 }, { "epoch": 0.3958723253578836, "grad_norm": 8.442957878112793, "learning_rate": 6.697230841489119e-06, "loss": 0.5156, "step": 7826 }, { "epoch": 0.3959229096059487, "grad_norm": 9.071891784667969, "learning_rate": 6.696475860219726e-06, "loss": 0.2791, "step": 7827 }, { "epoch": 0.39597349385401387, "grad_norm": 7.761422634124756, "learning_rate": 6.695720835236248e-06, "loss": 0.2485, "step": 7828 }, { "epoch": 0.396024078102079, "grad_norm": 7.122035980224609, "learning_rate": 6.694965766558141e-06, "loss": 0.4189, "step": 7829 }, { "epoch": 0.39607466235014416, "grad_norm": 11.267160415649414, "learning_rate": 6.694210654204859e-06, "loss": 0.346, "step": 7830 }, { "epoch": 0.3961252465982093, "grad_norm": 10.244720458984375, "learning_rate": 6.693455498195861e-06, "loss": 0.3751, "step": 7831 }, { "epoch": 0.39617583084627445, "grad_norm": 7.604232311248779, "learning_rate": 6.6927002985506065e-06, "loss": 0.4065, "step": 7832 }, { "epoch": 0.39622641509433965, "grad_norm": 7.876231670379639, "learning_rate": 6.6919450552885535e-06, "loss": 0.4474, "step": 7833 }, { "epoch": 0.3962769993424048, "grad_norm": 9.197139739990234, "learning_rate": 6.6911897684291625e-06, "loss": 0.4228, "step": 7834 }, { "epoch": 0.39632758359046993, "grad_norm": 10.764097213745117, "learning_rate": 6.6904344379918966e-06, "loss": 0.4362, "step": 7835 }, { "epoch": 0.3963781678385351, "grad_norm": 9.496125221252441, "learning_rate": 6.689679063996217e-06, "loss": 0.4342, "step": 7836 }, { "epoch": 0.3964287520866002, "grad_norm": 8.016181945800781, "learning_rate": 6.688923646461589e-06, "loss": 0.4217, "step": 7837 }, { "epoch": 0.39647933633466537, "grad_norm": 13.575711250305176, "learning_rate": 6.6881681854074795e-06, "loss": 0.4606, "step": 7838 }, { "epoch": 0.3965299205827305, "grad_norm": 8.667621612548828, "learning_rate": 6.687412680853354e-06, "loss": 0.5097, "step": 7839 }, { "epoch": 0.3965805048307957, "grad_norm": 7.1692280769348145, "learning_rate": 6.686657132818679e-06, "loss": 0.4608, "step": 7840 }, { "epoch": 0.39663108907886085, "grad_norm": 10.158102989196777, "learning_rate": 6.685901541322922e-06, "loss": 0.4776, "step": 7841 }, { "epoch": 0.396681673326926, "grad_norm": 14.892370223999023, "learning_rate": 6.685145906385556e-06, "loss": 0.4196, "step": 7842 }, { "epoch": 0.39673225757499114, "grad_norm": 5.261710166931152, "learning_rate": 6.68439022802605e-06, "loss": 0.3421, "step": 7843 }, { "epoch": 0.3967828418230563, "grad_norm": 6.03233003616333, "learning_rate": 6.6836345062638765e-06, "loss": 0.3687, "step": 7844 }, { "epoch": 0.39683342607112143, "grad_norm": 5.123080253601074, "learning_rate": 6.682878741118509e-06, "loss": 0.3456, "step": 7845 }, { "epoch": 0.39688401031918663, "grad_norm": 12.7396879196167, "learning_rate": 6.6821229326094196e-06, "loss": 0.6237, "step": 7846 }, { "epoch": 0.3969345945672518, "grad_norm": 8.538000106811523, "learning_rate": 6.681367080756086e-06, "loss": 0.3295, "step": 7847 }, { "epoch": 0.3969851788153169, "grad_norm": 9.684488296508789, "learning_rate": 6.680611185577984e-06, "loss": 0.3231, "step": 7848 }, { "epoch": 0.39703576306338206, "grad_norm": 6.788203239440918, "learning_rate": 6.679855247094591e-06, "loss": 0.3858, "step": 7849 }, { "epoch": 0.3970863473114472, "grad_norm": 6.413044452667236, "learning_rate": 6.679099265325385e-06, "loss": 0.3652, "step": 7850 }, { "epoch": 0.39713693155951235, "grad_norm": 8.200804710388184, "learning_rate": 6.678343240289848e-06, "loss": 0.3534, "step": 7851 }, { "epoch": 0.3971875158075775, "grad_norm": 6.390496730804443, "learning_rate": 6.677587172007459e-06, "loss": 0.3803, "step": 7852 }, { "epoch": 0.3972381000556427, "grad_norm": 4.206459045410156, "learning_rate": 6.6768310604977e-06, "loss": 0.3202, "step": 7853 }, { "epoch": 0.39728868430370784, "grad_norm": 10.068382263183594, "learning_rate": 6.676074905780057e-06, "loss": 0.5405, "step": 7854 }, { "epoch": 0.397339268551773, "grad_norm": 5.608604431152344, "learning_rate": 6.675318707874009e-06, "loss": 0.3463, "step": 7855 }, { "epoch": 0.39738985279983813, "grad_norm": 6.623215198516846, "learning_rate": 6.6745624667990464e-06, "loss": 0.3484, "step": 7856 }, { "epoch": 0.39744043704790327, "grad_norm": 12.161540985107422, "learning_rate": 6.6738061825746536e-06, "loss": 0.5111, "step": 7857 }, { "epoch": 0.3974910212959684, "grad_norm": 6.516021728515625, "learning_rate": 6.673049855220318e-06, "loss": 0.3599, "step": 7858 }, { "epoch": 0.39754160554403356, "grad_norm": 8.79853630065918, "learning_rate": 6.672293484755529e-06, "loss": 0.4647, "step": 7859 }, { "epoch": 0.39759218979209876, "grad_norm": 5.387675762176514, "learning_rate": 6.671537071199775e-06, "loss": 0.4425, "step": 7860 }, { "epoch": 0.3976427740401639, "grad_norm": 7.237521648406982, "learning_rate": 6.670780614572549e-06, "loss": 0.3061, "step": 7861 }, { "epoch": 0.39769335828822905, "grad_norm": 9.427785873413086, "learning_rate": 6.670024114893342e-06, "loss": 0.4027, "step": 7862 }, { "epoch": 0.3977439425362942, "grad_norm": 13.049365043640137, "learning_rate": 6.669267572181648e-06, "loss": 0.4627, "step": 7863 }, { "epoch": 0.39779452678435934, "grad_norm": 14.953102111816406, "learning_rate": 6.668510986456961e-06, "loss": 0.6737, "step": 7864 }, { "epoch": 0.3978451110324245, "grad_norm": 8.385185241699219, "learning_rate": 6.667754357738775e-06, "loss": 0.4029, "step": 7865 }, { "epoch": 0.3978956952804897, "grad_norm": 8.389382362365723, "learning_rate": 6.666997686046587e-06, "loss": 0.3749, "step": 7866 }, { "epoch": 0.3979462795285548, "grad_norm": 8.925040245056152, "learning_rate": 6.666240971399896e-06, "loss": 0.3169, "step": 7867 }, { "epoch": 0.39799686377661997, "grad_norm": 9.644378662109375, "learning_rate": 6.665484213818199e-06, "loss": 0.5103, "step": 7868 }, { "epoch": 0.3980474480246851, "grad_norm": 11.318939208984375, "learning_rate": 6.664727413320997e-06, "loss": 0.6023, "step": 7869 }, { "epoch": 0.39809803227275026, "grad_norm": 6.512307167053223, "learning_rate": 6.663970569927791e-06, "loss": 0.4265, "step": 7870 }, { "epoch": 0.3981486165208154, "grad_norm": 9.232197761535645, "learning_rate": 6.663213683658083e-06, "loss": 0.3549, "step": 7871 }, { "epoch": 0.39819920076888055, "grad_norm": 7.83629035949707, "learning_rate": 6.662456754531375e-06, "loss": 0.495, "step": 7872 }, { "epoch": 0.39824978501694575, "grad_norm": 7.822333812713623, "learning_rate": 6.661699782567173e-06, "loss": 0.4147, "step": 7873 }, { "epoch": 0.3983003692650109, "grad_norm": 8.902022361755371, "learning_rate": 6.660942767784979e-06, "loss": 0.4087, "step": 7874 }, { "epoch": 0.39835095351307603, "grad_norm": 9.410804748535156, "learning_rate": 6.660185710204303e-06, "loss": 0.4318, "step": 7875 }, { "epoch": 0.3984015377611412, "grad_norm": 9.649283409118652, "learning_rate": 6.659428609844651e-06, "loss": 0.4735, "step": 7876 }, { "epoch": 0.3984521220092063, "grad_norm": 11.43265151977539, "learning_rate": 6.658671466725533e-06, "loss": 0.3598, "step": 7877 }, { "epoch": 0.39850270625727147, "grad_norm": 8.29748821258545, "learning_rate": 6.657914280866457e-06, "loss": 0.4283, "step": 7878 }, { "epoch": 0.39855329050533667, "grad_norm": 8.029820442199707, "learning_rate": 6.657157052286933e-06, "loss": 0.3678, "step": 7879 }, { "epoch": 0.3986038747534018, "grad_norm": 8.185449600219727, "learning_rate": 6.656399781006476e-06, "loss": 0.4393, "step": 7880 }, { "epoch": 0.39865445900146695, "grad_norm": 6.454680919647217, "learning_rate": 6.655642467044597e-06, "loss": 0.5181, "step": 7881 }, { "epoch": 0.3987050432495321, "grad_norm": 7.481328010559082, "learning_rate": 6.654885110420812e-06, "loss": 0.4111, "step": 7882 }, { "epoch": 0.39875562749759724, "grad_norm": 9.295512199401855, "learning_rate": 6.654127711154634e-06, "loss": 0.3521, "step": 7883 }, { "epoch": 0.3988062117456624, "grad_norm": 6.4631667137146, "learning_rate": 6.65337026926558e-06, "loss": 0.4373, "step": 7884 }, { "epoch": 0.39885679599372753, "grad_norm": 4.867062091827393, "learning_rate": 6.652612784773167e-06, "loss": 0.3472, "step": 7885 }, { "epoch": 0.39890738024179273, "grad_norm": 12.832844734191895, "learning_rate": 6.6518552576969154e-06, "loss": 0.3999, "step": 7886 }, { "epoch": 0.3989579644898579, "grad_norm": 6.027671813964844, "learning_rate": 6.651097688056343e-06, "loss": 0.3243, "step": 7887 }, { "epoch": 0.399008548737923, "grad_norm": 12.0097074508667, "learning_rate": 6.650340075870971e-06, "loss": 0.3988, "step": 7888 }, { "epoch": 0.39905913298598816, "grad_norm": 10.803269386291504, "learning_rate": 6.649582421160323e-06, "loss": 0.4941, "step": 7889 }, { "epoch": 0.3991097172340533, "grad_norm": 6.814968585968018, "learning_rate": 6.648824723943921e-06, "loss": 0.3771, "step": 7890 }, { "epoch": 0.39916030148211845, "grad_norm": 9.218514442443848, "learning_rate": 6.648066984241286e-06, "loss": 0.4451, "step": 7891 }, { "epoch": 0.3992108857301836, "grad_norm": 6.747323513031006, "learning_rate": 6.647309202071946e-06, "loss": 0.4249, "step": 7892 }, { "epoch": 0.3992614699782488, "grad_norm": 5.551401138305664, "learning_rate": 6.646551377455428e-06, "loss": 0.4168, "step": 7893 }, { "epoch": 0.39931205422631394, "grad_norm": 8.514056205749512, "learning_rate": 6.645793510411256e-06, "loss": 0.4636, "step": 7894 }, { "epoch": 0.3993626384743791, "grad_norm": 14.35159683227539, "learning_rate": 6.645035600958962e-06, "loss": 0.4994, "step": 7895 }, { "epoch": 0.3994132227224442, "grad_norm": 9.311736106872559, "learning_rate": 6.644277649118075e-06, "loss": 0.391, "step": 7896 }, { "epoch": 0.39946380697050937, "grad_norm": 11.432271957397461, "learning_rate": 6.643519654908123e-06, "loss": 0.4077, "step": 7897 }, { "epoch": 0.3995143912185745, "grad_norm": 8.5112886428833, "learning_rate": 6.642761618348638e-06, "loss": 0.4572, "step": 7898 }, { "epoch": 0.3995649754666397, "grad_norm": 8.194304466247559, "learning_rate": 6.642003539459157e-06, "loss": 0.5378, "step": 7899 }, { "epoch": 0.39961555971470486, "grad_norm": 11.658570289611816, "learning_rate": 6.64124541825921e-06, "loss": 0.6092, "step": 7900 }, { "epoch": 0.39966614396277, "grad_norm": 7.034437656402588, "learning_rate": 6.640487254768333e-06, "loss": 0.4216, "step": 7901 }, { "epoch": 0.39971672821083515, "grad_norm": 20.70886993408203, "learning_rate": 6.6397290490060605e-06, "loss": 0.6279, "step": 7902 }, { "epoch": 0.3997673124589003, "grad_norm": 8.907769203186035, "learning_rate": 6.638970800991932e-06, "loss": 0.4993, "step": 7903 }, { "epoch": 0.39981789670696544, "grad_norm": 6.184798717498779, "learning_rate": 6.638212510745484e-06, "loss": 0.456, "step": 7904 }, { "epoch": 0.3998684809550306, "grad_norm": 9.95751667022705, "learning_rate": 6.637454178286258e-06, "loss": 0.5209, "step": 7905 }, { "epoch": 0.3999190652030958, "grad_norm": 5.254542350769043, "learning_rate": 6.636695803633791e-06, "loss": 0.3566, "step": 7906 }, { "epoch": 0.3999696494511609, "grad_norm": 9.740638732910156, "learning_rate": 6.6359373868076274e-06, "loss": 0.4221, "step": 7907 }, { "epoch": 0.40002023369922607, "grad_norm": 9.522099494934082, "learning_rate": 6.63517892782731e-06, "loss": 0.4343, "step": 7908 }, { "epoch": 0.40002023369922607, "eval_loss": 0.42862823605537415, "eval_runtime": 34.9116, "eval_samples_per_second": 45.773, "eval_steps_per_second": 5.729, "step": 7908 }, { "epoch": 0.4000708179472912, "grad_norm": 7.892787933349609, "learning_rate": 6.63442042671238e-06, "loss": 0.4114, "step": 7909 }, { "epoch": 0.40012140219535636, "grad_norm": 7.463297367095947, "learning_rate": 6.633661883482383e-06, "loss": 0.361, "step": 7910 }, { "epoch": 0.4001719864434215, "grad_norm": 12.374155044555664, "learning_rate": 6.632903298156865e-06, "loss": 0.4394, "step": 7911 }, { "epoch": 0.40022257069148665, "grad_norm": 5.650111675262451, "learning_rate": 6.632144670755374e-06, "loss": 0.3847, "step": 7912 }, { "epoch": 0.40027315493955185, "grad_norm": 5.928441524505615, "learning_rate": 6.631386001297456e-06, "loss": 0.285, "step": 7913 }, { "epoch": 0.400323739187617, "grad_norm": 8.955100059509277, "learning_rate": 6.6306272898026635e-06, "loss": 0.4626, "step": 7914 }, { "epoch": 0.40037432343568213, "grad_norm": 8.781803131103516, "learning_rate": 6.6298685362905425e-06, "loss": 0.398, "step": 7915 }, { "epoch": 0.4004249076837473, "grad_norm": 5.256223201751709, "learning_rate": 6.629109740780647e-06, "loss": 0.4061, "step": 7916 }, { "epoch": 0.4004754919318124, "grad_norm": 8.92105770111084, "learning_rate": 6.628350903292527e-06, "loss": 0.433, "step": 7917 }, { "epoch": 0.40052607617987757, "grad_norm": 9.691153526306152, "learning_rate": 6.627592023845739e-06, "loss": 0.4427, "step": 7918 }, { "epoch": 0.40057666042794277, "grad_norm": 12.493799209594727, "learning_rate": 6.626833102459835e-06, "loss": 0.6069, "step": 7919 }, { "epoch": 0.4006272446760079, "grad_norm": 9.447303771972656, "learning_rate": 6.626074139154373e-06, "loss": 0.5414, "step": 7920 }, { "epoch": 0.40067782892407305, "grad_norm": 6.1375579833984375, "learning_rate": 6.625315133948905e-06, "loss": 0.3516, "step": 7921 }, { "epoch": 0.4007284131721382, "grad_norm": 12.692216873168945, "learning_rate": 6.624556086862995e-06, "loss": 0.4135, "step": 7922 }, { "epoch": 0.40077899742020334, "grad_norm": 10.843635559082031, "learning_rate": 6.623796997916196e-06, "loss": 0.5415, "step": 7923 }, { "epoch": 0.4008295816682685, "grad_norm": 5.959901332855225, "learning_rate": 6.623037867128072e-06, "loss": 0.3723, "step": 7924 }, { "epoch": 0.40088016591633363, "grad_norm": 7.631769180297852, "learning_rate": 6.62227869451818e-06, "loss": 0.4766, "step": 7925 }, { "epoch": 0.40093075016439883, "grad_norm": 6.572478771209717, "learning_rate": 6.621519480106087e-06, "loss": 0.4353, "step": 7926 }, { "epoch": 0.400981334412464, "grad_norm": 7.831518173217773, "learning_rate": 6.620760223911354e-06, "loss": 0.3676, "step": 7927 }, { "epoch": 0.4010319186605291, "grad_norm": 5.266305446624756, "learning_rate": 6.620000925953545e-06, "loss": 0.387, "step": 7928 }, { "epoch": 0.40108250290859426, "grad_norm": 6.423763751983643, "learning_rate": 6.619241586252223e-06, "loss": 0.4863, "step": 7929 }, { "epoch": 0.4011330871566594, "grad_norm": 4.973964691162109, "learning_rate": 6.618482204826957e-06, "loss": 0.3778, "step": 7930 }, { "epoch": 0.40118367140472455, "grad_norm": 7.66685152053833, "learning_rate": 6.6177227816973145e-06, "loss": 0.4084, "step": 7931 }, { "epoch": 0.4012342556527897, "grad_norm": 11.493206977844238, "learning_rate": 6.616963316882863e-06, "loss": 0.3863, "step": 7932 }, { "epoch": 0.4012848399008549, "grad_norm": 9.560957908630371, "learning_rate": 6.616203810403172e-06, "loss": 0.463, "step": 7933 }, { "epoch": 0.40133542414892004, "grad_norm": 5.419135570526123, "learning_rate": 6.6154442622778125e-06, "loss": 0.453, "step": 7934 }, { "epoch": 0.4013860083969852, "grad_norm": 7.93339729309082, "learning_rate": 6.614684672526358e-06, "loss": 0.4076, "step": 7935 }, { "epoch": 0.4014365926450503, "grad_norm": 8.55313777923584, "learning_rate": 6.613925041168378e-06, "loss": 0.4611, "step": 7936 }, { "epoch": 0.40148717689311547, "grad_norm": 13.716327667236328, "learning_rate": 6.61316536822345e-06, "loss": 0.478, "step": 7937 }, { "epoch": 0.4015377611411806, "grad_norm": 7.923116683959961, "learning_rate": 6.612405653711145e-06, "loss": 0.3913, "step": 7938 }, { "epoch": 0.4015883453892458, "grad_norm": 9.363443374633789, "learning_rate": 6.611645897651041e-06, "loss": 0.4548, "step": 7939 }, { "epoch": 0.40163892963731096, "grad_norm": 10.289056777954102, "learning_rate": 6.610886100062715e-06, "loss": 0.4187, "step": 7940 }, { "epoch": 0.4016895138853761, "grad_norm": 8.855284690856934, "learning_rate": 6.610126260965746e-06, "loss": 0.3238, "step": 7941 }, { "epoch": 0.40174009813344125, "grad_norm": 9.551033973693848, "learning_rate": 6.6093663803797115e-06, "loss": 0.3505, "step": 7942 }, { "epoch": 0.4017906823815064, "grad_norm": 9.250566482543945, "learning_rate": 6.6086064583241934e-06, "loss": 0.4169, "step": 7943 }, { "epoch": 0.40184126662957154, "grad_norm": 17.25838851928711, "learning_rate": 6.607846494818772e-06, "loss": 0.4694, "step": 7944 }, { "epoch": 0.4018918508776367, "grad_norm": 6.9316935539245605, "learning_rate": 6.607086489883029e-06, "loss": 0.4202, "step": 7945 }, { "epoch": 0.4019424351257019, "grad_norm": 29.50681495666504, "learning_rate": 6.6063264435365505e-06, "loss": 0.4651, "step": 7946 }, { "epoch": 0.401993019373767, "grad_norm": 10.342735290527344, "learning_rate": 6.605566355798918e-06, "loss": 0.5138, "step": 7947 }, { "epoch": 0.40204360362183217, "grad_norm": 7.1405510902404785, "learning_rate": 6.60480622668972e-06, "loss": 0.2707, "step": 7948 }, { "epoch": 0.4020941878698973, "grad_norm": 6.483708381652832, "learning_rate": 6.604046056228541e-06, "loss": 0.3786, "step": 7949 }, { "epoch": 0.40214477211796246, "grad_norm": 8.228035926818848, "learning_rate": 6.60328584443497e-06, "loss": 0.3747, "step": 7950 }, { "epoch": 0.4021953563660276, "grad_norm": 9.155488014221191, "learning_rate": 6.602525591328595e-06, "loss": 0.4099, "step": 7951 }, { "epoch": 0.40224594061409275, "grad_norm": 6.402221202850342, "learning_rate": 6.601765296929006e-06, "loss": 0.4132, "step": 7952 }, { "epoch": 0.40229652486215794, "grad_norm": 7.982851028442383, "learning_rate": 6.601004961255794e-06, "loss": 0.4135, "step": 7953 }, { "epoch": 0.4023471091102231, "grad_norm": 7.000295162200928, "learning_rate": 6.600244584328552e-06, "loss": 0.4485, "step": 7954 }, { "epoch": 0.40239769335828823, "grad_norm": 8.922225952148438, "learning_rate": 6.599484166166873e-06, "loss": 0.3652, "step": 7955 }, { "epoch": 0.4024482776063534, "grad_norm": 7.659760475158691, "learning_rate": 6.598723706790349e-06, "loss": 0.3993, "step": 7956 }, { "epoch": 0.4024988618544185, "grad_norm": 6.210092544555664, "learning_rate": 6.597963206218576e-06, "loss": 0.3596, "step": 7957 }, { "epoch": 0.40254944610248367, "grad_norm": 7.5667338371276855, "learning_rate": 6.597202664471154e-06, "loss": 0.4306, "step": 7958 }, { "epoch": 0.40260003035054887, "grad_norm": 16.888835906982422, "learning_rate": 6.5964420815676744e-06, "loss": 0.5613, "step": 7959 }, { "epoch": 0.402650614598614, "grad_norm": 7.036871433258057, "learning_rate": 6.595681457527738e-06, "loss": 0.3458, "step": 7960 }, { "epoch": 0.40270119884667915, "grad_norm": 6.586001873016357, "learning_rate": 6.594920792370946e-06, "loss": 0.4549, "step": 7961 }, { "epoch": 0.4027517830947443, "grad_norm": 10.153210639953613, "learning_rate": 6.594160086116899e-06, "loss": 0.5382, "step": 7962 }, { "epoch": 0.40280236734280944, "grad_norm": 4.879579544067383, "learning_rate": 6.593399338785195e-06, "loss": 0.3543, "step": 7963 }, { "epoch": 0.4028529515908746, "grad_norm": 7.474008083343506, "learning_rate": 6.592638550395438e-06, "loss": 0.3982, "step": 7964 }, { "epoch": 0.40290353583893973, "grad_norm": 7.4824347496032715, "learning_rate": 6.5918777209672346e-06, "loss": 0.4034, "step": 7965 }, { "epoch": 0.40295412008700493, "grad_norm": 8.931159973144531, "learning_rate": 6.591116850520186e-06, "loss": 0.4871, "step": 7966 }, { "epoch": 0.4030047043350701, "grad_norm": 13.350565910339355, "learning_rate": 6.5903559390739e-06, "loss": 0.4805, "step": 7967 }, { "epoch": 0.4030552885831352, "grad_norm": 6.944357395172119, "learning_rate": 6.589594986647983e-06, "loss": 0.4171, "step": 7968 }, { "epoch": 0.40310587283120036, "grad_norm": 9.77503490447998, "learning_rate": 6.588833993262043e-06, "loss": 0.4312, "step": 7969 }, { "epoch": 0.4031564570792655, "grad_norm": 9.6619873046875, "learning_rate": 6.588072958935688e-06, "loss": 0.2895, "step": 7970 }, { "epoch": 0.40320704132733065, "grad_norm": 9.206136703491211, "learning_rate": 6.587311883688529e-06, "loss": 0.4884, "step": 7971 }, { "epoch": 0.4032576255753958, "grad_norm": 8.852386474609375, "learning_rate": 6.586550767540175e-06, "loss": 0.4667, "step": 7972 }, { "epoch": 0.403308209823461, "grad_norm": 5.0488433837890625, "learning_rate": 6.5857896105102415e-06, "loss": 0.4594, "step": 7973 }, { "epoch": 0.40335879407152614, "grad_norm": 7.837985515594482, "learning_rate": 6.5850284126183415e-06, "loss": 0.3127, "step": 7974 }, { "epoch": 0.4034093783195913, "grad_norm": 7.024402141571045, "learning_rate": 6.584267173884085e-06, "loss": 0.3792, "step": 7975 }, { "epoch": 0.4034599625676564, "grad_norm": 6.293661117553711, "learning_rate": 6.583505894327091e-06, "loss": 0.2969, "step": 7976 }, { "epoch": 0.40351054681572157, "grad_norm": 11.546215057373047, "learning_rate": 6.582744573966976e-06, "loss": 0.4321, "step": 7977 }, { "epoch": 0.4035611310637867, "grad_norm": 9.058006286621094, "learning_rate": 6.581983212823355e-06, "loss": 0.451, "step": 7978 }, { "epoch": 0.4036117153118519, "grad_norm": 7.959939002990723, "learning_rate": 6.5812218109158475e-06, "loss": 0.4254, "step": 7979 }, { "epoch": 0.40366229955991706, "grad_norm": 13.47668170928955, "learning_rate": 6.5804603682640745e-06, "loss": 0.4049, "step": 7980 }, { "epoch": 0.4037128838079822, "grad_norm": 12.501386642456055, "learning_rate": 6.5796988848876554e-06, "loss": 0.3782, "step": 7981 }, { "epoch": 0.40376346805604735, "grad_norm": 6.270696640014648, "learning_rate": 6.57893736080621e-06, "loss": 0.4252, "step": 7982 }, { "epoch": 0.4038140523041125, "grad_norm": 10.162906646728516, "learning_rate": 6.578175796039362e-06, "loss": 0.5456, "step": 7983 }, { "epoch": 0.40386463655217764, "grad_norm": 4.718564987182617, "learning_rate": 6.577414190606738e-06, "loss": 0.2941, "step": 7984 }, { "epoch": 0.4039152208002428, "grad_norm": 5.113373756408691, "learning_rate": 6.57665254452796e-06, "loss": 0.3569, "step": 7985 }, { "epoch": 0.403965805048308, "grad_norm": 11.080913543701172, "learning_rate": 6.575890857822653e-06, "loss": 0.5017, "step": 7986 }, { "epoch": 0.4040163892963731, "grad_norm": 10.480944633483887, "learning_rate": 6.575129130510447e-06, "loss": 0.3915, "step": 7987 }, { "epoch": 0.40406697354443827, "grad_norm": 7.044002056121826, "learning_rate": 6.574367362610965e-06, "loss": 0.3872, "step": 7988 }, { "epoch": 0.4041175577925034, "grad_norm": 11.080570220947266, "learning_rate": 6.57360555414384e-06, "loss": 0.3927, "step": 7989 }, { "epoch": 0.40416814204056856, "grad_norm": 8.586432456970215, "learning_rate": 6.572843705128702e-06, "loss": 0.4923, "step": 7990 }, { "epoch": 0.4042187262886337, "grad_norm": 9.46144962310791, "learning_rate": 6.572081815585179e-06, "loss": 0.4205, "step": 7991 }, { "epoch": 0.40426931053669884, "grad_norm": 7.734278202056885, "learning_rate": 6.571319885532905e-06, "loss": 0.393, "step": 7992 }, { "epoch": 0.40431989478476404, "grad_norm": 10.421778678894043, "learning_rate": 6.5705579149915154e-06, "loss": 0.4113, "step": 7993 }, { "epoch": 0.4043704790328292, "grad_norm": 7.5151262283325195, "learning_rate": 6.56979590398064e-06, "loss": 0.3837, "step": 7994 }, { "epoch": 0.40442106328089433, "grad_norm": 4.942038536071777, "learning_rate": 6.569033852519915e-06, "loss": 0.3289, "step": 7995 }, { "epoch": 0.4044716475289595, "grad_norm": 9.053047180175781, "learning_rate": 6.568271760628979e-06, "loss": 0.4652, "step": 7996 }, { "epoch": 0.4045222317770246, "grad_norm": 11.045092582702637, "learning_rate": 6.567509628327466e-06, "loss": 0.4701, "step": 7997 }, { "epoch": 0.40457281602508977, "grad_norm": 8.532196044921875, "learning_rate": 6.566747455635017e-06, "loss": 0.4571, "step": 7998 }, { "epoch": 0.40462340027315497, "grad_norm": 12.847685813903809, "learning_rate": 6.565985242571271e-06, "loss": 0.5055, "step": 7999 }, { "epoch": 0.4046739845212201, "grad_norm": 7.5838704109191895, "learning_rate": 6.565222989155868e-06, "loss": 0.3475, "step": 8000 }, { "epoch": 0.40472456876928525, "grad_norm": 9.58516788482666, "learning_rate": 6.564460695408449e-06, "loss": 0.4299, "step": 8001 }, { "epoch": 0.4047751530173504, "grad_norm": 7.196671009063721, "learning_rate": 6.563698361348655e-06, "loss": 0.4273, "step": 8002 }, { "epoch": 0.40482573726541554, "grad_norm": 5.2089338302612305, "learning_rate": 6.562935986996133e-06, "loss": 0.3224, "step": 8003 }, { "epoch": 0.4048763215134807, "grad_norm": 6.304817199707031, "learning_rate": 6.562173572370525e-06, "loss": 0.4457, "step": 8004 }, { "epoch": 0.40492690576154583, "grad_norm": 6.46419095993042, "learning_rate": 6.561411117491477e-06, "loss": 0.371, "step": 8005 }, { "epoch": 0.40497749000961103, "grad_norm": 17.449403762817383, "learning_rate": 6.560648622378638e-06, "loss": 0.4349, "step": 8006 }, { "epoch": 0.4050280742576762, "grad_norm": 7.2103095054626465, "learning_rate": 6.559886087051651e-06, "loss": 0.3918, "step": 8007 }, { "epoch": 0.4050786585057413, "grad_norm": 8.032639503479004, "learning_rate": 6.559123511530168e-06, "loss": 0.4475, "step": 8008 }, { "epoch": 0.40512924275380646, "grad_norm": 18.050975799560547, "learning_rate": 6.558360895833839e-06, "loss": 0.5203, "step": 8009 }, { "epoch": 0.4051798270018716, "grad_norm": 13.608555793762207, "learning_rate": 6.557598239982312e-06, "loss": 0.575, "step": 8010 }, { "epoch": 0.40523041124993675, "grad_norm": 9.604135513305664, "learning_rate": 6.556835543995241e-06, "loss": 0.3771, "step": 8011 }, { "epoch": 0.4052809954980019, "grad_norm": 6.900638103485107, "learning_rate": 6.5560728078922805e-06, "loss": 0.3417, "step": 8012 }, { "epoch": 0.4053315797460671, "grad_norm": 8.915263175964355, "learning_rate": 6.555310031693081e-06, "loss": 0.4589, "step": 8013 }, { "epoch": 0.40538216399413224, "grad_norm": 9.023715019226074, "learning_rate": 6.554547215417297e-06, "loss": 0.4129, "step": 8014 }, { "epoch": 0.4054327482421974, "grad_norm": 14.015523910522461, "learning_rate": 6.553784359084589e-06, "loss": 0.3973, "step": 8015 }, { "epoch": 0.4054833324902625, "grad_norm": 18.440746307373047, "learning_rate": 6.553021462714609e-06, "loss": 0.4422, "step": 8016 }, { "epoch": 0.40553391673832767, "grad_norm": 5.430202960968018, "learning_rate": 6.552258526327018e-06, "loss": 0.38, "step": 8017 }, { "epoch": 0.4055845009863928, "grad_norm": 8.870139122009277, "learning_rate": 6.5514955499414734e-06, "loss": 0.4254, "step": 8018 }, { "epoch": 0.405635085234458, "grad_norm": 7.393499374389648, "learning_rate": 6.5507325335776375e-06, "loss": 0.2796, "step": 8019 }, { "epoch": 0.40568566948252316, "grad_norm": 8.392518043518066, "learning_rate": 6.549969477255168e-06, "loss": 0.3923, "step": 8020 }, { "epoch": 0.4057362537305883, "grad_norm": 6.230595111846924, "learning_rate": 6.549206380993729e-06, "loss": 0.4137, "step": 8021 }, { "epoch": 0.40578683797865345, "grad_norm": 11.010364532470703, "learning_rate": 6.548443244812985e-06, "loss": 0.5336, "step": 8022 }, { "epoch": 0.4058374222267186, "grad_norm": 5.844946384429932, "learning_rate": 6.547680068732598e-06, "loss": 0.343, "step": 8023 }, { "epoch": 0.40588800647478374, "grad_norm": 11.489815711975098, "learning_rate": 6.546916852772234e-06, "loss": 0.6254, "step": 8024 }, { "epoch": 0.4059385907228489, "grad_norm": 6.197476863861084, "learning_rate": 6.546153596951558e-06, "loss": 0.4276, "step": 8025 }, { "epoch": 0.4059891749709141, "grad_norm": 13.109374046325684, "learning_rate": 6.5453903012902396e-06, "loss": 0.4878, "step": 8026 }, { "epoch": 0.4060397592189792, "grad_norm": 8.717442512512207, "learning_rate": 6.544626965807945e-06, "loss": 0.5046, "step": 8027 }, { "epoch": 0.40609034346704437, "grad_norm": 8.768924713134766, "learning_rate": 6.543863590524346e-06, "loss": 0.4554, "step": 8028 }, { "epoch": 0.4061409277151095, "grad_norm": 5.144128799438477, "learning_rate": 6.543100175459109e-06, "loss": 0.4323, "step": 8029 }, { "epoch": 0.40619151196317466, "grad_norm": 7.421875953674316, "learning_rate": 6.54233672063191e-06, "loss": 0.3312, "step": 8030 }, { "epoch": 0.4062420962112398, "grad_norm": 5.773090362548828, "learning_rate": 6.541573226062419e-06, "loss": 0.3668, "step": 8031 }, { "epoch": 0.406292680459305, "grad_norm": 10.0983247756958, "learning_rate": 6.540809691770308e-06, "loss": 0.3977, "step": 8032 }, { "epoch": 0.40634326470737014, "grad_norm": 8.260614395141602, "learning_rate": 6.540046117775252e-06, "loss": 0.3057, "step": 8033 }, { "epoch": 0.4063938489554353, "grad_norm": 7.622390270233154, "learning_rate": 6.53928250409693e-06, "loss": 0.5719, "step": 8034 }, { "epoch": 0.40644443320350043, "grad_norm": 12.278560638427734, "learning_rate": 6.538518850755014e-06, "loss": 0.3773, "step": 8035 }, { "epoch": 0.4064950174515656, "grad_norm": 9.44477653503418, "learning_rate": 6.537755157769183e-06, "loss": 0.3451, "step": 8036 }, { "epoch": 0.4065456016996307, "grad_norm": 7.190261363983154, "learning_rate": 6.5369914251591185e-06, "loss": 0.3828, "step": 8037 }, { "epoch": 0.40659618594769587, "grad_norm": 10.26974868774414, "learning_rate": 6.536227652944496e-06, "loss": 0.4068, "step": 8038 }, { "epoch": 0.40664677019576106, "grad_norm": 5.688198089599609, "learning_rate": 6.535463841144997e-06, "loss": 0.3288, "step": 8039 }, { "epoch": 0.4066973544438262, "grad_norm": 7.210202693939209, "learning_rate": 6.534699989780303e-06, "loss": 0.3677, "step": 8040 }, { "epoch": 0.40674793869189135, "grad_norm": 11.331380844116211, "learning_rate": 6.5339360988701e-06, "loss": 0.4696, "step": 8041 }, { "epoch": 0.4067985229399565, "grad_norm": 32.97190475463867, "learning_rate": 6.533172168434066e-06, "loss": 0.5973, "step": 8042 }, { "epoch": 0.40684910718802164, "grad_norm": 10.705862998962402, "learning_rate": 6.532408198491889e-06, "loss": 0.5442, "step": 8043 }, { "epoch": 0.4068996914360868, "grad_norm": 8.572854995727539, "learning_rate": 6.531644189063255e-06, "loss": 0.4, "step": 8044 }, { "epoch": 0.40695027568415193, "grad_norm": 7.976503849029541, "learning_rate": 6.530880140167849e-06, "loss": 0.4898, "step": 8045 }, { "epoch": 0.40700085993221713, "grad_norm": 19.17531967163086, "learning_rate": 6.530116051825359e-06, "loss": 0.4535, "step": 8046 }, { "epoch": 0.4070514441802823, "grad_norm": 11.843985557556152, "learning_rate": 6.529351924055476e-06, "loss": 0.4746, "step": 8047 }, { "epoch": 0.4071020284283474, "grad_norm": 18.105318069458008, "learning_rate": 6.528587756877886e-06, "loss": 0.5726, "step": 8048 }, { "epoch": 0.40715261267641256, "grad_norm": 6.525039196014404, "learning_rate": 6.527823550312281e-06, "loss": 0.4651, "step": 8049 }, { "epoch": 0.4072031969244777, "grad_norm": 6.5673441886901855, "learning_rate": 6.5270593043783556e-06, "loss": 0.4018, "step": 8050 }, { "epoch": 0.40725378117254285, "grad_norm": 9.770676612854004, "learning_rate": 6.526295019095801e-06, "loss": 0.4301, "step": 8051 }, { "epoch": 0.40730436542060805, "grad_norm": 7.979015827178955, "learning_rate": 6.525530694484307e-06, "loss": 0.5693, "step": 8052 }, { "epoch": 0.4073549496686732, "grad_norm": 7.98754358291626, "learning_rate": 6.524766330563575e-06, "loss": 0.4068, "step": 8053 }, { "epoch": 0.40740553391673834, "grad_norm": 11.834210395812988, "learning_rate": 6.524001927353296e-06, "loss": 0.6367, "step": 8054 }, { "epoch": 0.4074561181648035, "grad_norm": 6.807636260986328, "learning_rate": 6.523237484873169e-06, "loss": 0.5005, "step": 8055 }, { "epoch": 0.4075067024128686, "grad_norm": 8.161848068237305, "learning_rate": 6.5224730031428905e-06, "loss": 0.347, "step": 8056 }, { "epoch": 0.40755728666093377, "grad_norm": 8.358624458312988, "learning_rate": 6.521708482182162e-06, "loss": 0.3903, "step": 8057 }, { "epoch": 0.4076078709089989, "grad_norm": 6.212766170501709, "learning_rate": 6.520943922010678e-06, "loss": 0.3735, "step": 8058 }, { "epoch": 0.4076584551570641, "grad_norm": 4.9097723960876465, "learning_rate": 6.520179322648145e-06, "loss": 0.3077, "step": 8059 }, { "epoch": 0.40770903940512926, "grad_norm": 6.882104873657227, "learning_rate": 6.519414684114264e-06, "loss": 0.3545, "step": 8060 }, { "epoch": 0.4077596236531944, "grad_norm": 12.785470008850098, "learning_rate": 6.518650006428734e-06, "loss": 0.5928, "step": 8061 }, { "epoch": 0.40781020790125955, "grad_norm": 9.489439964294434, "learning_rate": 6.517885289611263e-06, "loss": 0.3766, "step": 8062 }, { "epoch": 0.4078607921493247, "grad_norm": 4.014226913452148, "learning_rate": 6.517120533681554e-06, "loss": 0.4107, "step": 8063 }, { "epoch": 0.40791137639738984, "grad_norm": 15.36516284942627, "learning_rate": 6.516355738659314e-06, "loss": 0.5106, "step": 8064 }, { "epoch": 0.407961960645455, "grad_norm": 6.093475341796875, "learning_rate": 6.5155909045642476e-06, "loss": 0.4233, "step": 8065 }, { "epoch": 0.4080125448935202, "grad_norm": 10.299094200134277, "learning_rate": 6.514826031416066e-06, "loss": 0.542, "step": 8066 }, { "epoch": 0.4080631291415853, "grad_norm": 5.986876010894775, "learning_rate": 6.514061119234475e-06, "loss": 0.3175, "step": 8067 }, { "epoch": 0.40811371338965047, "grad_norm": 9.700737953186035, "learning_rate": 6.513296168039187e-06, "loss": 0.4754, "step": 8068 }, { "epoch": 0.4081642976377156, "grad_norm": 5.450881481170654, "learning_rate": 6.512531177849913e-06, "loss": 0.5029, "step": 8069 }, { "epoch": 0.40821488188578076, "grad_norm": 10.166229248046875, "learning_rate": 6.511766148686362e-06, "loss": 0.5119, "step": 8070 }, { "epoch": 0.4082654661338459, "grad_norm": 8.027186393737793, "learning_rate": 6.511001080568249e-06, "loss": 0.4361, "step": 8071 }, { "epoch": 0.4083160503819111, "grad_norm": 8.258332252502441, "learning_rate": 6.510235973515289e-06, "loss": 0.4984, "step": 8072 }, { "epoch": 0.40836663462997624, "grad_norm": 8.511754989624023, "learning_rate": 6.509470827547195e-06, "loss": 0.4245, "step": 8073 }, { "epoch": 0.4084172188780414, "grad_norm": 8.826804161071777, "learning_rate": 6.5087056426836835e-06, "loss": 0.3766, "step": 8074 }, { "epoch": 0.40846780312610653, "grad_norm": 5.365450859069824, "learning_rate": 6.507940418944473e-06, "loss": 0.3635, "step": 8075 }, { "epoch": 0.4085183873741717, "grad_norm": 8.098729133605957, "learning_rate": 6.507175156349279e-06, "loss": 0.3958, "step": 8076 }, { "epoch": 0.4085689716222368, "grad_norm": 12.820246696472168, "learning_rate": 6.506409854917821e-06, "loss": 0.4325, "step": 8077 }, { "epoch": 0.40861955587030196, "grad_norm": 9.046050071716309, "learning_rate": 6.505644514669822e-06, "loss": 0.4701, "step": 8078 }, { "epoch": 0.40867014011836716, "grad_norm": 7.967991352081299, "learning_rate": 6.504879135624999e-06, "loss": 0.4144, "step": 8079 }, { "epoch": 0.4087207243664323, "grad_norm": 5.5432024002075195, "learning_rate": 6.504113717803074e-06, "loss": 0.3023, "step": 8080 }, { "epoch": 0.40877130861449745, "grad_norm": 8.395339965820312, "learning_rate": 6.503348261223774e-06, "loss": 0.3865, "step": 8081 }, { "epoch": 0.4088218928625626, "grad_norm": 10.449378967285156, "learning_rate": 6.502582765906819e-06, "loss": 0.4502, "step": 8082 }, { "epoch": 0.40887247711062774, "grad_norm": 5.256753444671631, "learning_rate": 6.501817231871937e-06, "loss": 0.3897, "step": 8083 }, { "epoch": 0.4089230613586929, "grad_norm": 4.709334850311279, "learning_rate": 6.501051659138852e-06, "loss": 0.3378, "step": 8084 }, { "epoch": 0.40897364560675803, "grad_norm": 8.536371231079102, "learning_rate": 6.5002860477272914e-06, "loss": 0.4351, "step": 8085 }, { "epoch": 0.40902422985482323, "grad_norm": 7.945330619812012, "learning_rate": 6.499520397656982e-06, "loss": 0.485, "step": 8086 }, { "epoch": 0.4090748141028884, "grad_norm": 5.242333889007568, "learning_rate": 6.4987547089476545e-06, "loss": 0.4227, "step": 8087 }, { "epoch": 0.4091253983509535, "grad_norm": 12.010403633117676, "learning_rate": 6.497988981619039e-06, "loss": 0.6312, "step": 8088 }, { "epoch": 0.40917598259901866, "grad_norm": 12.085566520690918, "learning_rate": 6.497223215690866e-06, "loss": 0.5471, "step": 8089 }, { "epoch": 0.4092265668470838, "grad_norm": 7.524615287780762, "learning_rate": 6.496457411182866e-06, "loss": 0.4626, "step": 8090 }, { "epoch": 0.40927715109514895, "grad_norm": 5.9094109535217285, "learning_rate": 6.495691568114774e-06, "loss": 0.3973, "step": 8091 }, { "epoch": 0.40932773534321415, "grad_norm": 3.883040189743042, "learning_rate": 6.494925686506321e-06, "loss": 0.4309, "step": 8092 }, { "epoch": 0.4093783195912793, "grad_norm": 7.48361873626709, "learning_rate": 6.494159766377245e-06, "loss": 0.3045, "step": 8093 }, { "epoch": 0.40942890383934444, "grad_norm": 16.244647979736328, "learning_rate": 6.493393807747282e-06, "loss": 0.5207, "step": 8094 }, { "epoch": 0.4094794880874096, "grad_norm": 4.706594944000244, "learning_rate": 6.492627810636165e-06, "loss": 0.4288, "step": 8095 }, { "epoch": 0.4095300723354747, "grad_norm": 5.579041004180908, "learning_rate": 6.491861775063637e-06, "loss": 0.4399, "step": 8096 }, { "epoch": 0.40958065658353987, "grad_norm": 5.348215103149414, "learning_rate": 6.4910957010494345e-06, "loss": 0.4431, "step": 8097 }, { "epoch": 0.409631240831605, "grad_norm": 4.594571590423584, "learning_rate": 6.4903295886132964e-06, "loss": 0.3963, "step": 8098 }, { "epoch": 0.4096818250796702, "grad_norm": 8.688002586364746, "learning_rate": 6.489563437774965e-06, "loss": 0.4491, "step": 8099 }, { "epoch": 0.40973240932773536, "grad_norm": 9.605710983276367, "learning_rate": 6.488797248554182e-06, "loss": 0.3732, "step": 8100 }, { "epoch": 0.4097829935758005, "grad_norm": 4.9442548751831055, "learning_rate": 6.488031020970689e-06, "loss": 0.3426, "step": 8101 }, { "epoch": 0.40983357782386565, "grad_norm": 7.032223701477051, "learning_rate": 6.487264755044231e-06, "loss": 0.5169, "step": 8102 }, { "epoch": 0.4098841620719308, "grad_norm": 8.7217378616333, "learning_rate": 6.486498450794553e-06, "loss": 0.5358, "step": 8103 }, { "epoch": 0.40993474631999594, "grad_norm": 8.307950019836426, "learning_rate": 6.485732108241401e-06, "loss": 0.4849, "step": 8104 }, { "epoch": 0.4099853305680611, "grad_norm": 7.79520320892334, "learning_rate": 6.48496572740452e-06, "loss": 0.4419, "step": 8105 }, { "epoch": 0.4100359148161263, "grad_norm": 13.407938003540039, "learning_rate": 6.484199308303659e-06, "loss": 0.4654, "step": 8106 }, { "epoch": 0.4100864990641914, "grad_norm": 16.846118927001953, "learning_rate": 6.483432850958568e-06, "loss": 0.5903, "step": 8107 }, { "epoch": 0.41013708331225657, "grad_norm": 13.828113555908203, "learning_rate": 6.482666355388994e-06, "loss": 0.5882, "step": 8108 }, { "epoch": 0.4101876675603217, "grad_norm": 8.587301254272461, "learning_rate": 6.48189982161469e-06, "loss": 0.3856, "step": 8109 }, { "epoch": 0.41023825180838686, "grad_norm": 8.80925178527832, "learning_rate": 6.481133249655408e-06, "loss": 0.418, "step": 8110 }, { "epoch": 0.410288836056452, "grad_norm": 6.8570661544799805, "learning_rate": 6.480366639530899e-06, "loss": 0.3664, "step": 8111 }, { "epoch": 0.4103394203045172, "grad_norm": 13.84033489227295, "learning_rate": 6.479599991260916e-06, "loss": 0.4838, "step": 8112 }, { "epoch": 0.41039000455258234, "grad_norm": 8.740323066711426, "learning_rate": 6.4788333048652164e-06, "loss": 0.3122, "step": 8113 }, { "epoch": 0.4104405888006475, "grad_norm": 7.4607648849487305, "learning_rate": 6.4780665803635535e-06, "loss": 0.4144, "step": 8114 }, { "epoch": 0.41049117304871263, "grad_norm": 9.557879447937012, "learning_rate": 6.477299817775686e-06, "loss": 0.498, "step": 8115 }, { "epoch": 0.4105417572967778, "grad_norm": 5.96286678314209, "learning_rate": 6.476533017121372e-06, "loss": 0.4462, "step": 8116 }, { "epoch": 0.4105923415448429, "grad_norm": 9.051146507263184, "learning_rate": 6.475766178420366e-06, "loss": 0.4477, "step": 8117 }, { "epoch": 0.41064292579290806, "grad_norm": 7.12598991394043, "learning_rate": 6.4749993016924305e-06, "loss": 0.4103, "step": 8118 }, { "epoch": 0.41069351004097326, "grad_norm": 8.653116226196289, "learning_rate": 6.474232386957326e-06, "loss": 0.4275, "step": 8119 }, { "epoch": 0.4107440942890384, "grad_norm": 7.794053554534912, "learning_rate": 6.473465434234813e-06, "loss": 0.4532, "step": 8120 }, { "epoch": 0.41079467853710355, "grad_norm": 8.885144233703613, "learning_rate": 6.472698443544657e-06, "loss": 0.3449, "step": 8121 }, { "epoch": 0.4108452627851687, "grad_norm": 9.386737823486328, "learning_rate": 6.471931414906617e-06, "loss": 0.3621, "step": 8122 }, { "epoch": 0.41089584703323384, "grad_norm": 5.37992000579834, "learning_rate": 6.471164348340462e-06, "loss": 0.3379, "step": 8123 }, { "epoch": 0.410946431281299, "grad_norm": 7.589677333831787, "learning_rate": 6.4703972438659525e-06, "loss": 0.5142, "step": 8124 }, { "epoch": 0.41099701552936413, "grad_norm": 23.78457260131836, "learning_rate": 6.4696301015028586e-06, "loss": 0.3967, "step": 8125 }, { "epoch": 0.41104759977742933, "grad_norm": 7.441697597503662, "learning_rate": 6.468862921270948e-06, "loss": 0.39, "step": 8126 }, { "epoch": 0.4110981840254945, "grad_norm": 9.320706367492676, "learning_rate": 6.468095703189986e-06, "loss": 0.4009, "step": 8127 }, { "epoch": 0.4111487682735596, "grad_norm": 23.153419494628906, "learning_rate": 6.467328447279746e-06, "loss": 0.4804, "step": 8128 }, { "epoch": 0.41119935252162476, "grad_norm": 7.078683376312256, "learning_rate": 6.466561153559995e-06, "loss": 0.4468, "step": 8129 }, { "epoch": 0.4112499367696899, "grad_norm": 6.456906795501709, "learning_rate": 6.465793822050504e-06, "loss": 0.4122, "step": 8130 }, { "epoch": 0.41130052101775505, "grad_norm": 7.311640739440918, "learning_rate": 6.465026452771048e-06, "loss": 0.3987, "step": 8131 }, { "epoch": 0.41135110526582025, "grad_norm": 15.360024452209473, "learning_rate": 6.464259045741399e-06, "loss": 0.4545, "step": 8132 }, { "epoch": 0.4114016895138854, "grad_norm": 6.393542766571045, "learning_rate": 6.463491600981331e-06, "loss": 0.3552, "step": 8133 }, { "epoch": 0.41145227376195054, "grad_norm": 7.531096935272217, "learning_rate": 6.462724118510619e-06, "loss": 0.3286, "step": 8134 }, { "epoch": 0.4115028580100157, "grad_norm": 8.630330085754395, "learning_rate": 6.461956598349041e-06, "loss": 0.3794, "step": 8135 }, { "epoch": 0.4115534422580808, "grad_norm": 8.459562301635742, "learning_rate": 6.461189040516371e-06, "loss": 0.4712, "step": 8136 }, { "epoch": 0.41160402650614597, "grad_norm": 10.439318656921387, "learning_rate": 6.460421445032389e-06, "loss": 0.5683, "step": 8137 }, { "epoch": 0.4116546107542111, "grad_norm": 5.989672660827637, "learning_rate": 6.459653811916874e-06, "loss": 0.4068, "step": 8138 }, { "epoch": 0.4117051950022763, "grad_norm": 12.915084838867188, "learning_rate": 6.458886141189606e-06, "loss": 0.4961, "step": 8139 }, { "epoch": 0.41175577925034146, "grad_norm": 14.431985855102539, "learning_rate": 6.458118432870365e-06, "loss": 0.49, "step": 8140 }, { "epoch": 0.4118063634984066, "grad_norm": 6.825070381164551, "learning_rate": 6.457350686978936e-06, "loss": 0.4251, "step": 8141 }, { "epoch": 0.41185694774647175, "grad_norm": 17.102514266967773, "learning_rate": 6.456582903535099e-06, "loss": 0.4961, "step": 8142 }, { "epoch": 0.4119075319945369, "grad_norm": 23.393983840942383, "learning_rate": 6.455815082558639e-06, "loss": 0.4862, "step": 8143 }, { "epoch": 0.41195811624260203, "grad_norm": 27.6768741607666, "learning_rate": 6.455047224069339e-06, "loss": 0.4986, "step": 8144 }, { "epoch": 0.4120087004906672, "grad_norm": 5.9537129402160645, "learning_rate": 6.454279328086988e-06, "loss": 0.3491, "step": 8145 }, { "epoch": 0.4120592847387324, "grad_norm": 6.2546586990356445, "learning_rate": 6.453511394631371e-06, "loss": 0.3374, "step": 8146 }, { "epoch": 0.4121098689867975, "grad_norm": 9.256131172180176, "learning_rate": 6.452743423722277e-06, "loss": 0.459, "step": 8147 }, { "epoch": 0.41216045323486267, "grad_norm": 6.708637714385986, "learning_rate": 6.451975415379494e-06, "loss": 0.5009, "step": 8148 }, { "epoch": 0.4122110374829278, "grad_norm": 6.357979774475098, "learning_rate": 6.451207369622811e-06, "loss": 0.5047, "step": 8149 }, { "epoch": 0.41226162173099296, "grad_norm": 6.455109596252441, "learning_rate": 6.450439286472019e-06, "loss": 0.3034, "step": 8150 }, { "epoch": 0.4123122059790581, "grad_norm": 10.662761688232422, "learning_rate": 6.4496711659469114e-06, "loss": 0.5664, "step": 8151 }, { "epoch": 0.4123627902271233, "grad_norm": 10.439576148986816, "learning_rate": 6.448903008067277e-06, "loss": 0.4846, "step": 8152 }, { "epoch": 0.41241337447518844, "grad_norm": 6.664345741271973, "learning_rate": 6.448134812852913e-06, "loss": 0.3712, "step": 8153 }, { "epoch": 0.4124639587232536, "grad_norm": 10.647335052490234, "learning_rate": 6.447366580323614e-06, "loss": 0.3298, "step": 8154 }, { "epoch": 0.41251454297131873, "grad_norm": 6.0530219078063965, "learning_rate": 6.446598310499173e-06, "loss": 0.4812, "step": 8155 }, { "epoch": 0.4125651272193839, "grad_norm": 7.433631896972656, "learning_rate": 6.445830003399387e-06, "loss": 0.393, "step": 8156 }, { "epoch": 0.412615711467449, "grad_norm": 8.757074356079102, "learning_rate": 6.445061659044055e-06, "loss": 0.3562, "step": 8157 }, { "epoch": 0.41266629571551416, "grad_norm": 7.714808464050293, "learning_rate": 6.444293277452973e-06, "loss": 0.3289, "step": 8158 }, { "epoch": 0.41271687996357936, "grad_norm": 12.619873046875, "learning_rate": 6.4435248586459435e-06, "loss": 0.4411, "step": 8159 }, { "epoch": 0.4127674642116445, "grad_norm": 10.66104507446289, "learning_rate": 6.442756402642765e-06, "loss": 0.4895, "step": 8160 }, { "epoch": 0.41281804845970965, "grad_norm": 7.376105308532715, "learning_rate": 6.4419879094632385e-06, "loss": 0.3571, "step": 8161 }, { "epoch": 0.4128686327077748, "grad_norm": 6.087345600128174, "learning_rate": 6.441219379127166e-06, "loss": 0.4331, "step": 8162 }, { "epoch": 0.41291921695583994, "grad_norm": 5.135623931884766, "learning_rate": 6.440450811654353e-06, "loss": 0.3807, "step": 8163 }, { "epoch": 0.4129698012039051, "grad_norm": 7.149777412414551, "learning_rate": 6.439682207064599e-06, "loss": 0.3935, "step": 8164 }, { "epoch": 0.41302038545197023, "grad_norm": 9.035016059875488, "learning_rate": 6.4389135653777134e-06, "loss": 0.3669, "step": 8165 }, { "epoch": 0.41307096970003543, "grad_norm": 14.80953598022461, "learning_rate": 6.438144886613501e-06, "loss": 0.5222, "step": 8166 }, { "epoch": 0.4131215539481006, "grad_norm": 9.214401245117188, "learning_rate": 6.437376170791769e-06, "loss": 0.4841, "step": 8167 }, { "epoch": 0.4131721381961657, "grad_norm": 10.325187683105469, "learning_rate": 6.436607417932324e-06, "loss": 0.3671, "step": 8168 }, { "epoch": 0.41322272244423086, "grad_norm": 4.75564432144165, "learning_rate": 6.435838628054975e-06, "loss": 0.3018, "step": 8169 }, { "epoch": 0.413273306692296, "grad_norm": 9.856650352478027, "learning_rate": 6.435069801179534e-06, "loss": 0.4179, "step": 8170 }, { "epoch": 0.41332389094036115, "grad_norm": 11.049659729003906, "learning_rate": 6.43430093732581e-06, "loss": 0.5573, "step": 8171 }, { "epoch": 0.41337447518842635, "grad_norm": 8.979105949401855, "learning_rate": 6.4335320365136154e-06, "loss": 0.3587, "step": 8172 }, { "epoch": 0.4134250594364915, "grad_norm": 10.98637866973877, "learning_rate": 6.432763098762764e-06, "loss": 0.53, "step": 8173 }, { "epoch": 0.41347564368455664, "grad_norm": 10.489389419555664, "learning_rate": 6.431994124093068e-06, "loss": 0.5034, "step": 8174 }, { "epoch": 0.4135262279326218, "grad_norm": 12.182815551757812, "learning_rate": 6.4312251125243405e-06, "loss": 0.4398, "step": 8175 }, { "epoch": 0.4135768121806869, "grad_norm": 4.296504497528076, "learning_rate": 6.4304560640764e-06, "loss": 0.3231, "step": 8176 }, { "epoch": 0.41362739642875207, "grad_norm": 9.090055465698242, "learning_rate": 6.429686978769061e-06, "loss": 0.3108, "step": 8177 }, { "epoch": 0.4136779806768172, "grad_norm": 5.999547004699707, "learning_rate": 6.428917856622142e-06, "loss": 0.4206, "step": 8178 }, { "epoch": 0.4137285649248824, "grad_norm": 10.113682746887207, "learning_rate": 6.428148697655463e-06, "loss": 0.4275, "step": 8179 }, { "epoch": 0.41377914917294756, "grad_norm": 10.706503868103027, "learning_rate": 6.4273795018888405e-06, "loss": 0.4702, "step": 8180 }, { "epoch": 0.4138297334210127, "grad_norm": 13.937576293945312, "learning_rate": 6.426610269342095e-06, "loss": 0.5086, "step": 8181 }, { "epoch": 0.41388031766907785, "grad_norm": 11.18634033203125, "learning_rate": 6.42584100003505e-06, "loss": 0.4671, "step": 8182 }, { "epoch": 0.413930901917143, "grad_norm": 6.754908084869385, "learning_rate": 6.4250716939875255e-06, "loss": 0.4076, "step": 8183 }, { "epoch": 0.41398148616520813, "grad_norm": 6.036075115203857, "learning_rate": 6.4243023512193445e-06, "loss": 0.4084, "step": 8184 }, { "epoch": 0.41403207041327333, "grad_norm": 7.530466556549072, "learning_rate": 6.423532971750335e-06, "loss": 0.3916, "step": 8185 }, { "epoch": 0.4140826546613385, "grad_norm": 7.727636337280273, "learning_rate": 6.422763555600319e-06, "loss": 0.3542, "step": 8186 }, { "epoch": 0.4141332389094036, "grad_norm": 5.0848469734191895, "learning_rate": 6.42199410278912e-06, "loss": 0.429, "step": 8187 }, { "epoch": 0.41418382315746877, "grad_norm": 7.094573497772217, "learning_rate": 6.421224613336569e-06, "loss": 0.3467, "step": 8188 }, { "epoch": 0.4142344074055339, "grad_norm": 13.575668334960938, "learning_rate": 6.420455087262494e-06, "loss": 0.5706, "step": 8189 }, { "epoch": 0.41428499165359906, "grad_norm": 8.380942344665527, "learning_rate": 6.4196855245867195e-06, "loss": 0.404, "step": 8190 }, { "epoch": 0.4143355759016642, "grad_norm": 7.609431266784668, "learning_rate": 6.418915925329079e-06, "loss": 0.426, "step": 8191 }, { "epoch": 0.4143861601497294, "grad_norm": 8.149503707885742, "learning_rate": 6.418146289509406e-06, "loss": 0.4298, "step": 8192 }, { "epoch": 0.41443674439779454, "grad_norm": 7.253271102905273, "learning_rate": 6.4173766171475234e-06, "loss": 0.4566, "step": 8193 }, { "epoch": 0.4144873286458597, "grad_norm": 11.41616439819336, "learning_rate": 6.416606908263272e-06, "loss": 0.5406, "step": 8194 }, { "epoch": 0.41453791289392483, "grad_norm": 13.340764045715332, "learning_rate": 6.415837162876481e-06, "loss": 0.3985, "step": 8195 }, { "epoch": 0.41458849714199, "grad_norm": 6.293758392333984, "learning_rate": 6.415067381006986e-06, "loss": 0.3832, "step": 8196 }, { "epoch": 0.4146390813900551, "grad_norm": 10.167181968688965, "learning_rate": 6.414297562674623e-06, "loss": 0.4773, "step": 8197 }, { "epoch": 0.41468966563812026, "grad_norm": 11.227188110351562, "learning_rate": 6.413527707899228e-06, "loss": 0.4447, "step": 8198 }, { "epoch": 0.41474024988618546, "grad_norm": 12.008963584899902, "learning_rate": 6.412757816700639e-06, "loss": 0.3219, "step": 8199 }, { "epoch": 0.4147908341342506, "grad_norm": 9.227100372314453, "learning_rate": 6.411987889098692e-06, "loss": 0.4161, "step": 8200 }, { "epoch": 0.41484141838231575, "grad_norm": 10.314775466918945, "learning_rate": 6.411217925113229e-06, "loss": 0.5651, "step": 8201 }, { "epoch": 0.4148920026303809, "grad_norm": 5.570117950439453, "learning_rate": 6.4104479247640875e-06, "loss": 0.4517, "step": 8202 }, { "epoch": 0.41494258687844604, "grad_norm": 4.6477837562561035, "learning_rate": 6.409677888071109e-06, "loss": 0.4447, "step": 8203 }, { "epoch": 0.4149931711265112, "grad_norm": 27.029281616210938, "learning_rate": 6.408907815054139e-06, "loss": 0.5692, "step": 8204 }, { "epoch": 0.4150437553745764, "grad_norm": 4.359726428985596, "learning_rate": 6.408137705733016e-06, "loss": 0.3791, "step": 8205 }, { "epoch": 0.41509433962264153, "grad_norm": 8.362598419189453, "learning_rate": 6.407367560127585e-06, "loss": 0.3932, "step": 8206 }, { "epoch": 0.4151449238707067, "grad_norm": 8.545371055603027, "learning_rate": 6.406597378257692e-06, "loss": 0.409, "step": 8207 }, { "epoch": 0.4151955081187718, "grad_norm": 9.828593254089355, "learning_rate": 6.4058271601431824e-06, "loss": 0.4828, "step": 8208 }, { "epoch": 0.41524609236683696, "grad_norm": 13.170707702636719, "learning_rate": 6.405056905803902e-06, "loss": 0.4979, "step": 8209 }, { "epoch": 0.4152966766149021, "grad_norm": 20.508892059326172, "learning_rate": 6.404286615259699e-06, "loss": 0.3211, "step": 8210 }, { "epoch": 0.41534726086296725, "grad_norm": 7.142650604248047, "learning_rate": 6.403516288530422e-06, "loss": 0.5073, "step": 8211 }, { "epoch": 0.41539784511103245, "grad_norm": 7.226365089416504, "learning_rate": 6.40274592563592e-06, "loss": 0.3821, "step": 8212 }, { "epoch": 0.4154484293590976, "grad_norm": 8.515825271606445, "learning_rate": 6.401975526596044e-06, "loss": 0.4224, "step": 8213 }, { "epoch": 0.41549901360716274, "grad_norm": 7.384027004241943, "learning_rate": 6.401205091430646e-06, "loss": 0.3958, "step": 8214 }, { "epoch": 0.4155495978552279, "grad_norm": 13.034395217895508, "learning_rate": 6.400434620159575e-06, "loss": 0.4908, "step": 8215 }, { "epoch": 0.415600182103293, "grad_norm": 7.433323383331299, "learning_rate": 6.399664112802687e-06, "loss": 0.4609, "step": 8216 }, { "epoch": 0.41565076635135817, "grad_norm": 4.789209365844727, "learning_rate": 6.398893569379837e-06, "loss": 0.3847, "step": 8217 }, { "epoch": 0.4157013505994233, "grad_norm": 8.976400375366211, "learning_rate": 6.398122989910879e-06, "loss": 0.4307, "step": 8218 }, { "epoch": 0.4157519348474885, "grad_norm": 12.980770111083984, "learning_rate": 6.397352374415666e-06, "loss": 0.4083, "step": 8219 }, { "epoch": 0.41580251909555366, "grad_norm": 5.30658483505249, "learning_rate": 6.396581722914059e-06, "loss": 0.3854, "step": 8220 }, { "epoch": 0.4158531033436188, "grad_norm": 13.753509521484375, "learning_rate": 6.395811035425914e-06, "loss": 0.5679, "step": 8221 }, { "epoch": 0.41590368759168395, "grad_norm": 7.885075092315674, "learning_rate": 6.39504031197109e-06, "loss": 0.4176, "step": 8222 }, { "epoch": 0.4159542718397491, "grad_norm": 10.648398399353027, "learning_rate": 6.394269552569448e-06, "loss": 0.5547, "step": 8223 }, { "epoch": 0.41600485608781423, "grad_norm": 5.54617166519165, "learning_rate": 6.3934987572408465e-06, "loss": 0.3943, "step": 8224 }, { "epoch": 0.41605544033587943, "grad_norm": 5.978953838348389, "learning_rate": 6.392727926005147e-06, "loss": 0.3176, "step": 8225 }, { "epoch": 0.4161060245839446, "grad_norm": 6.034608364105225, "learning_rate": 6.3919570588822134e-06, "loss": 0.3772, "step": 8226 }, { "epoch": 0.4161566088320097, "grad_norm": 9.175884246826172, "learning_rate": 6.3911861558919085e-06, "loss": 0.5943, "step": 8227 }, { "epoch": 0.41620719308007487, "grad_norm": 8.579686164855957, "learning_rate": 6.390415217054097e-06, "loss": 0.4102, "step": 8228 }, { "epoch": 0.41625777732814, "grad_norm": 5.091846466064453, "learning_rate": 6.389644242388643e-06, "loss": 0.4201, "step": 8229 }, { "epoch": 0.41630836157620515, "grad_norm": 9.286952018737793, "learning_rate": 6.388873231915415e-06, "loss": 0.4433, "step": 8230 }, { "epoch": 0.4163589458242703, "grad_norm": 6.97910737991333, "learning_rate": 6.388102185654278e-06, "loss": 0.3754, "step": 8231 }, { "epoch": 0.4164095300723355, "grad_norm": 6.208345890045166, "learning_rate": 6.3873311036251e-06, "loss": 0.4602, "step": 8232 }, { "epoch": 0.41646011432040064, "grad_norm": 4.270662307739258, "learning_rate": 6.38655998584775e-06, "loss": 0.3423, "step": 8233 }, { "epoch": 0.4165106985684658, "grad_norm": 7.401851177215576, "learning_rate": 6.385788832342099e-06, "loss": 0.3888, "step": 8234 }, { "epoch": 0.41656128281653093, "grad_norm": 9.60147762298584, "learning_rate": 6.385017643128015e-06, "loss": 0.4601, "step": 8235 }, { "epoch": 0.4166118670645961, "grad_norm": 12.04753589630127, "learning_rate": 6.3842464182253745e-06, "loss": 0.4843, "step": 8236 }, { "epoch": 0.4166624513126612, "grad_norm": 8.185850143432617, "learning_rate": 6.383475157654046e-06, "loss": 0.4939, "step": 8237 }, { "epoch": 0.41671303556072636, "grad_norm": 9.419014930725098, "learning_rate": 6.3827038614339056e-06, "loss": 0.4205, "step": 8238 }, { "epoch": 0.41676361980879156, "grad_norm": 4.992865085601807, "learning_rate": 6.381932529584826e-06, "loss": 0.3764, "step": 8239 }, { "epoch": 0.4168142040568567, "grad_norm": 11.609025001525879, "learning_rate": 6.381161162126681e-06, "loss": 0.4281, "step": 8240 }, { "epoch": 0.41686478830492185, "grad_norm": 11.095036506652832, "learning_rate": 6.380389759079351e-06, "loss": 0.4832, "step": 8241 }, { "epoch": 0.416915372552987, "grad_norm": 11.084781646728516, "learning_rate": 6.379618320462711e-06, "loss": 0.5369, "step": 8242 }, { "epoch": 0.41696595680105214, "grad_norm": 8.199507713317871, "learning_rate": 6.378846846296638e-06, "loss": 0.4132, "step": 8243 }, { "epoch": 0.4170165410491173, "grad_norm": 8.513421058654785, "learning_rate": 6.3780753366010125e-06, "loss": 0.3086, "step": 8244 }, { "epoch": 0.4170671252971825, "grad_norm": 19.43545913696289, "learning_rate": 6.377303791395715e-06, "loss": 0.7333, "step": 8245 }, { "epoch": 0.41711770954524763, "grad_norm": 17.618906021118164, "learning_rate": 6.3765322107006254e-06, "loss": 0.5078, "step": 8246 }, { "epoch": 0.41716829379331277, "grad_norm": 7.4219770431518555, "learning_rate": 6.375760594535625e-06, "loss": 0.4537, "step": 8247 }, { "epoch": 0.4172188780413779, "grad_norm": 3.8709659576416016, "learning_rate": 6.374988942920597e-06, "loss": 0.4029, "step": 8248 }, { "epoch": 0.41726946228944306, "grad_norm": 12.808392524719238, "learning_rate": 6.374217255875426e-06, "loss": 0.4723, "step": 8249 }, { "epoch": 0.4173200465375082, "grad_norm": 7.503603935241699, "learning_rate": 6.373445533419995e-06, "loss": 0.5565, "step": 8250 }, { "epoch": 0.41737063078557335, "grad_norm": 5.662208557128906, "learning_rate": 6.37267377557419e-06, "loss": 0.4524, "step": 8251 }, { "epoch": 0.41742121503363855, "grad_norm": 11.809706687927246, "learning_rate": 6.371901982357899e-06, "loss": 0.5115, "step": 8252 }, { "epoch": 0.4174717992817037, "grad_norm": 8.825411796569824, "learning_rate": 6.371130153791005e-06, "loss": 0.4886, "step": 8253 }, { "epoch": 0.41752238352976884, "grad_norm": 7.842352867126465, "learning_rate": 6.370358289893399e-06, "loss": 0.3119, "step": 8254 }, { "epoch": 0.417572967777834, "grad_norm": 7.624742031097412, "learning_rate": 6.369586390684972e-06, "loss": 0.5075, "step": 8255 }, { "epoch": 0.4176235520258991, "grad_norm": 5.941114902496338, "learning_rate": 6.36881445618561e-06, "loss": 0.2987, "step": 8256 }, { "epoch": 0.41767413627396427, "grad_norm": 14.339951515197754, "learning_rate": 6.368042486415207e-06, "loss": 0.5588, "step": 8257 }, { "epoch": 0.4177247205220294, "grad_norm": 9.212797164916992, "learning_rate": 6.367270481393653e-06, "loss": 0.5348, "step": 8258 }, { "epoch": 0.4177753047700946, "grad_norm": 7.972133636474609, "learning_rate": 6.36649844114084e-06, "loss": 0.3602, "step": 8259 }, { "epoch": 0.41782588901815976, "grad_norm": 8.71007251739502, "learning_rate": 6.365726365676665e-06, "loss": 0.4781, "step": 8260 }, { "epoch": 0.4178764732662249, "grad_norm": 4.9956231117248535, "learning_rate": 6.364954255021019e-06, "loss": 0.3676, "step": 8261 }, { "epoch": 0.41792705751429005, "grad_norm": 9.404800415039062, "learning_rate": 6.364182109193798e-06, "loss": 0.4083, "step": 8262 }, { "epoch": 0.4179776417623552, "grad_norm": 6.2185821533203125, "learning_rate": 6.363409928214901e-06, "loss": 0.293, "step": 8263 }, { "epoch": 0.41802822601042033, "grad_norm": 6.7051897048950195, "learning_rate": 6.362637712104222e-06, "loss": 0.4542, "step": 8264 }, { "epoch": 0.41807881025848553, "grad_norm": 6.362833499908447, "learning_rate": 6.361865460881661e-06, "loss": 0.4136, "step": 8265 }, { "epoch": 0.4181293945065507, "grad_norm": 6.257495403289795, "learning_rate": 6.361093174567116e-06, "loss": 0.4362, "step": 8266 }, { "epoch": 0.4181799787546158, "grad_norm": 7.158471584320068, "learning_rate": 6.360320853180488e-06, "loss": 0.4407, "step": 8267 }, { "epoch": 0.41823056300268097, "grad_norm": 11.439289093017578, "learning_rate": 6.359548496741677e-06, "loss": 0.4136, "step": 8268 }, { "epoch": 0.4182811472507461, "grad_norm": 13.845067977905273, "learning_rate": 6.3587761052705856e-06, "loss": 0.4212, "step": 8269 }, { "epoch": 0.41833173149881125, "grad_norm": 8.91258716583252, "learning_rate": 6.358003678787115e-06, "loss": 0.4199, "step": 8270 }, { "epoch": 0.4183823157468764, "grad_norm": 9.079719543457031, "learning_rate": 6.35723121731117e-06, "loss": 0.4494, "step": 8271 }, { "epoch": 0.4184328999949416, "grad_norm": 11.340455055236816, "learning_rate": 6.356458720862655e-06, "loss": 0.4873, "step": 8272 }, { "epoch": 0.41848348424300674, "grad_norm": 6.005380630493164, "learning_rate": 6.355686189461474e-06, "loss": 0.4395, "step": 8273 }, { "epoch": 0.4185340684910719, "grad_norm": 5.729478359222412, "learning_rate": 6.354913623127537e-06, "loss": 0.3742, "step": 8274 }, { "epoch": 0.41858465273913703, "grad_norm": 6.762272834777832, "learning_rate": 6.354141021880746e-06, "loss": 0.4205, "step": 8275 }, { "epoch": 0.4186352369872022, "grad_norm": 6.115293502807617, "learning_rate": 6.353368385741014e-06, "loss": 0.3724, "step": 8276 }, { "epoch": 0.4186858212352673, "grad_norm": 6.319382190704346, "learning_rate": 6.352595714728247e-06, "loss": 0.3858, "step": 8277 }, { "epoch": 0.41873640548333246, "grad_norm": 8.649785995483398, "learning_rate": 6.3518230088623544e-06, "loss": 0.3801, "step": 8278 }, { "epoch": 0.41878698973139766, "grad_norm": 11.6384859085083, "learning_rate": 6.351050268163248e-06, "loss": 0.4159, "step": 8279 }, { "epoch": 0.4188375739794628, "grad_norm": 8.5343599319458, "learning_rate": 6.3502774926508416e-06, "loss": 0.4361, "step": 8280 }, { "epoch": 0.41888815822752795, "grad_norm": 5.756952285766602, "learning_rate": 6.349504682345044e-06, "loss": 0.4393, "step": 8281 }, { "epoch": 0.4189387424755931, "grad_norm": 15.623408317565918, "learning_rate": 6.34873183726577e-06, "loss": 0.4365, "step": 8282 }, { "epoch": 0.41898932672365824, "grad_norm": 9.2174711227417, "learning_rate": 6.347958957432935e-06, "loss": 0.4537, "step": 8283 }, { "epoch": 0.4190399109717234, "grad_norm": 5.603913307189941, "learning_rate": 6.347186042866454e-06, "loss": 0.3803, "step": 8284 }, { "epoch": 0.4190904952197886, "grad_norm": 8.212477684020996, "learning_rate": 6.3464130935862415e-06, "loss": 0.4536, "step": 8285 }, { "epoch": 0.4191410794678537, "grad_norm": 8.749022483825684, "learning_rate": 6.345640109612216e-06, "loss": 0.4053, "step": 8286 }, { "epoch": 0.41919166371591887, "grad_norm": 9.195435523986816, "learning_rate": 6.344867090964295e-06, "loss": 0.4168, "step": 8287 }, { "epoch": 0.419242247963984, "grad_norm": 8.700016021728516, "learning_rate": 6.344094037662399e-06, "loss": 0.4039, "step": 8288 }, { "epoch": 0.41929283221204916, "grad_norm": 4.908812999725342, "learning_rate": 6.343320949726445e-06, "loss": 0.3613, "step": 8289 }, { "epoch": 0.4193434164601143, "grad_norm": 6.557348251342773, "learning_rate": 6.3425478271763554e-06, "loss": 0.3849, "step": 8290 }, { "epoch": 0.41939400070817945, "grad_norm": 9.98143196105957, "learning_rate": 6.341774670032051e-06, "loss": 0.5144, "step": 8291 }, { "epoch": 0.41944458495624465, "grad_norm": 11.468581199645996, "learning_rate": 6.341001478313452e-06, "loss": 0.4944, "step": 8292 }, { "epoch": 0.4194951692043098, "grad_norm": 10.670770645141602, "learning_rate": 6.340228252040487e-06, "loss": 0.3096, "step": 8293 }, { "epoch": 0.41954575345237494, "grad_norm": 5.298659324645996, "learning_rate": 6.339454991233075e-06, "loss": 0.4189, "step": 8294 }, { "epoch": 0.4195963377004401, "grad_norm": 9.25459098815918, "learning_rate": 6.338681695911144e-06, "loss": 0.3534, "step": 8295 }, { "epoch": 0.4196469219485052, "grad_norm": 7.729344844818115, "learning_rate": 6.33790836609462e-06, "loss": 0.3804, "step": 8296 }, { "epoch": 0.41969750619657037, "grad_norm": 5.199252605438232, "learning_rate": 6.337135001803429e-06, "loss": 0.3562, "step": 8297 }, { "epoch": 0.4197480904446355, "grad_norm": 12.237428665161133, "learning_rate": 6.336361603057497e-06, "loss": 0.4708, "step": 8298 }, { "epoch": 0.4197986746927007, "grad_norm": 12.527971267700195, "learning_rate": 6.335588169876755e-06, "loss": 0.5571, "step": 8299 }, { "epoch": 0.41984925894076586, "grad_norm": 14.569414138793945, "learning_rate": 6.334814702281131e-06, "loss": 0.3564, "step": 8300 }, { "epoch": 0.419899843188831, "grad_norm": 11.44350814819336, "learning_rate": 6.334041200290557e-06, "loss": 0.4731, "step": 8301 }, { "epoch": 0.41995042743689615, "grad_norm": 16.94927406311035, "learning_rate": 6.333267663924964e-06, "loss": 0.5825, "step": 8302 }, { "epoch": 0.4200010116849613, "grad_norm": 5.802400588989258, "learning_rate": 6.332494093204285e-06, "loss": 0.3453, "step": 8303 }, { "epoch": 0.42005159593302643, "grad_norm": 6.001095294952393, "learning_rate": 6.33172048814845e-06, "loss": 0.4452, "step": 8304 }, { "epoch": 0.42010218018109163, "grad_norm": 6.106420993804932, "learning_rate": 6.330946848777395e-06, "loss": 0.4009, "step": 8305 }, { "epoch": 0.4201527644291568, "grad_norm": 10.655972480773926, "learning_rate": 6.330173175111054e-06, "loss": 0.5311, "step": 8306 }, { "epoch": 0.4202033486772219, "grad_norm": 4.910787582397461, "learning_rate": 6.329399467169364e-06, "loss": 0.238, "step": 8307 }, { "epoch": 0.42025393292528707, "grad_norm": 7.923670291900635, "learning_rate": 6.32862572497226e-06, "loss": 0.5425, "step": 8308 }, { "epoch": 0.4203045171733522, "grad_norm": 8.050741195678711, "learning_rate": 6.327851948539682e-06, "loss": 0.405, "step": 8309 }, { "epoch": 0.42035510142141735, "grad_norm": 7.86564826965332, "learning_rate": 6.327078137891565e-06, "loss": 0.4372, "step": 8310 }, { "epoch": 0.4204056856694825, "grad_norm": 8.08151626586914, "learning_rate": 6.326304293047849e-06, "loss": 0.4499, "step": 8311 }, { "epoch": 0.4204562699175477, "grad_norm": 6.0600714683532715, "learning_rate": 6.325530414028477e-06, "loss": 0.5281, "step": 8312 }, { "epoch": 0.42050685416561284, "grad_norm": 5.7438154220581055, "learning_rate": 6.324756500853386e-06, "loss": 0.3889, "step": 8313 }, { "epoch": 0.420557438413678, "grad_norm": 12.54399585723877, "learning_rate": 6.32398255354252e-06, "loss": 0.5091, "step": 8314 }, { "epoch": 0.42060802266174313, "grad_norm": 7.635797500610352, "learning_rate": 6.323208572115824e-06, "loss": 0.3886, "step": 8315 }, { "epoch": 0.4206586069098083, "grad_norm": 5.79901647567749, "learning_rate": 6.322434556593237e-06, "loss": 0.3718, "step": 8316 }, { "epoch": 0.4207091911578734, "grad_norm": 11.348711967468262, "learning_rate": 6.321660506994704e-06, "loss": 0.5023, "step": 8317 }, { "epoch": 0.42075977540593856, "grad_norm": 6.603202819824219, "learning_rate": 6.3208864233401755e-06, "loss": 0.4133, "step": 8318 }, { "epoch": 0.42081035965400376, "grad_norm": 7.623229503631592, "learning_rate": 6.320112305649593e-06, "loss": 0.3712, "step": 8319 }, { "epoch": 0.4208609439020689, "grad_norm": 18.502870559692383, "learning_rate": 6.319338153942905e-06, "loss": 0.4532, "step": 8320 }, { "epoch": 0.42091152815013405, "grad_norm": 6.19186544418335, "learning_rate": 6.318563968240059e-06, "loss": 0.4291, "step": 8321 }, { "epoch": 0.4209621123981992, "grad_norm": 11.148786544799805, "learning_rate": 6.317789748561005e-06, "loss": 0.5098, "step": 8322 }, { "epoch": 0.42101269664626434, "grad_norm": 6.393585205078125, "learning_rate": 6.317015494925692e-06, "loss": 0.4346, "step": 8323 }, { "epoch": 0.4210632808943295, "grad_norm": 7.528472423553467, "learning_rate": 6.3162412073540715e-06, "loss": 0.4274, "step": 8324 }, { "epoch": 0.4211138651423947, "grad_norm": 10.64428424835205, "learning_rate": 6.315466885866093e-06, "loss": 0.4897, "step": 8325 }, { "epoch": 0.4211644493904598, "grad_norm": 6.250580787658691, "learning_rate": 6.314692530481711e-06, "loss": 0.3713, "step": 8326 }, { "epoch": 0.42121503363852497, "grad_norm": 6.834151268005371, "learning_rate": 6.3139181412208785e-06, "loss": 0.4232, "step": 8327 }, { "epoch": 0.4212656178865901, "grad_norm": 9.206487655639648, "learning_rate": 6.313143718103548e-06, "loss": 0.3919, "step": 8328 }, { "epoch": 0.42131620213465526, "grad_norm": 6.870183944702148, "learning_rate": 6.312369261149676e-06, "loss": 0.4328, "step": 8329 }, { "epoch": 0.4213667863827204, "grad_norm": 11.613615989685059, "learning_rate": 6.311594770379218e-06, "loss": 0.3763, "step": 8330 }, { "epoch": 0.42141737063078555, "grad_norm": 9.120862007141113, "learning_rate": 6.310820245812131e-06, "loss": 0.3799, "step": 8331 }, { "epoch": 0.42146795487885075, "grad_norm": 6.354763031005859, "learning_rate": 6.310045687468373e-06, "loss": 0.3858, "step": 8332 }, { "epoch": 0.4215185391269159, "grad_norm": 6.0528178215026855, "learning_rate": 6.3092710953679e-06, "loss": 0.4137, "step": 8333 }, { "epoch": 0.42156912337498104, "grad_norm": 12.814476013183594, "learning_rate": 6.308496469530676e-06, "loss": 0.4626, "step": 8334 }, { "epoch": 0.4216197076230462, "grad_norm": 12.542794227600098, "learning_rate": 6.3077218099766565e-06, "loss": 0.5292, "step": 8335 }, { "epoch": 0.4216702918711113, "grad_norm": 10.074666023254395, "learning_rate": 6.3069471167258055e-06, "loss": 0.3477, "step": 8336 }, { "epoch": 0.42172087611917647, "grad_norm": 5.661407470703125, "learning_rate": 6.306172389798084e-06, "loss": 0.4122, "step": 8337 }, { "epoch": 0.42177146036724167, "grad_norm": 7.105926036834717, "learning_rate": 6.305397629213454e-06, "loss": 0.3367, "step": 8338 }, { "epoch": 0.4218220446153068, "grad_norm": 9.63887882232666, "learning_rate": 6.304622834991882e-06, "loss": 0.4465, "step": 8339 }, { "epoch": 0.42187262886337196, "grad_norm": 7.059132099151611, "learning_rate": 6.303848007153329e-06, "loss": 0.4734, "step": 8340 }, { "epoch": 0.4219232131114371, "grad_norm": 7.343297004699707, "learning_rate": 6.303073145717764e-06, "loss": 0.4085, "step": 8341 }, { "epoch": 0.42197379735950225, "grad_norm": 8.55286693572998, "learning_rate": 6.30229825070515e-06, "loss": 0.4438, "step": 8342 }, { "epoch": 0.4220243816075674, "grad_norm": 11.502968788146973, "learning_rate": 6.301523322135457e-06, "loss": 0.5013, "step": 8343 }, { "epoch": 0.42207496585563253, "grad_norm": 6.964874267578125, "learning_rate": 6.30074836002865e-06, "loss": 0.3633, "step": 8344 }, { "epoch": 0.42212555010369773, "grad_norm": 6.672020435333252, "learning_rate": 6.299973364404699e-06, "loss": 0.3134, "step": 8345 }, { "epoch": 0.4221761343517629, "grad_norm": 5.639354705810547, "learning_rate": 6.299198335283576e-06, "loss": 0.4497, "step": 8346 }, { "epoch": 0.422226718599828, "grad_norm": 9.699380874633789, "learning_rate": 6.298423272685252e-06, "loss": 0.392, "step": 8347 }, { "epoch": 0.42227730284789317, "grad_norm": 7.019670486450195, "learning_rate": 6.297648176629693e-06, "loss": 0.4111, "step": 8348 }, { "epoch": 0.4223278870959583, "grad_norm": 6.003820419311523, "learning_rate": 6.296873047136875e-06, "loss": 0.4391, "step": 8349 }, { "epoch": 0.42237847134402345, "grad_norm": 10.084284782409668, "learning_rate": 6.296097884226773e-06, "loss": 0.4636, "step": 8350 }, { "epoch": 0.4224290555920886, "grad_norm": 14.27921199798584, "learning_rate": 6.295322687919357e-06, "loss": 0.5177, "step": 8351 }, { "epoch": 0.4224796398401538, "grad_norm": 5.875887870788574, "learning_rate": 6.294547458234606e-06, "loss": 0.4509, "step": 8352 }, { "epoch": 0.42253022408821894, "grad_norm": 10.769984245300293, "learning_rate": 6.293772195192493e-06, "loss": 0.4774, "step": 8353 }, { "epoch": 0.4225808083362841, "grad_norm": 11.065953254699707, "learning_rate": 6.292996898812995e-06, "loss": 0.553, "step": 8354 }, { "epoch": 0.42263139258434923, "grad_norm": 6.0061469078063965, "learning_rate": 6.29222156911609e-06, "loss": 0.384, "step": 8355 }, { "epoch": 0.4226819768324144, "grad_norm": 6.491832256317139, "learning_rate": 6.291446206121758e-06, "loss": 0.3529, "step": 8356 }, { "epoch": 0.4227325610804795, "grad_norm": 8.465343475341797, "learning_rate": 6.290670809849975e-06, "loss": 0.4157, "step": 8357 }, { "epoch": 0.4227831453285447, "grad_norm": 5.976955890655518, "learning_rate": 6.289895380320723e-06, "loss": 0.4095, "step": 8358 }, { "epoch": 0.42283372957660986, "grad_norm": 8.678579330444336, "learning_rate": 6.289119917553983e-06, "loss": 0.417, "step": 8359 }, { "epoch": 0.422884313824675, "grad_norm": 8.296125411987305, "learning_rate": 6.288344421569738e-06, "loss": 0.4302, "step": 8360 }, { "epoch": 0.42293489807274015, "grad_norm": 12.25472640991211, "learning_rate": 6.2875688923879665e-06, "loss": 0.5795, "step": 8361 }, { "epoch": 0.4229854823208053, "grad_norm": 11.122004508972168, "learning_rate": 6.286793330028657e-06, "loss": 0.4484, "step": 8362 }, { "epoch": 0.42303606656887044, "grad_norm": 6.460672855377197, "learning_rate": 6.286017734511789e-06, "loss": 0.4038, "step": 8363 }, { "epoch": 0.4230866508169356, "grad_norm": 6.434424877166748, "learning_rate": 6.285242105857351e-06, "loss": 0.4162, "step": 8364 }, { "epoch": 0.4231372350650008, "grad_norm": 8.469286918640137, "learning_rate": 6.28446644408533e-06, "loss": 0.356, "step": 8365 }, { "epoch": 0.4231878193130659, "grad_norm": 7.9182939529418945, "learning_rate": 6.2836907492157095e-06, "loss": 0.3535, "step": 8366 }, { "epoch": 0.42323840356113107, "grad_norm": 4.799422740936279, "learning_rate": 6.2829150212684796e-06, "loss": 0.3519, "step": 8367 }, { "epoch": 0.4232889878091962, "grad_norm": 6.012901306152344, "learning_rate": 6.282139260263628e-06, "loss": 0.3195, "step": 8368 }, { "epoch": 0.42333957205726136, "grad_norm": 10.49556827545166, "learning_rate": 6.281363466221145e-06, "loss": 0.4065, "step": 8369 }, { "epoch": 0.4233901563053265, "grad_norm": 8.001941680908203, "learning_rate": 6.2805876391610195e-06, "loss": 0.4384, "step": 8370 }, { "epoch": 0.42344074055339165, "grad_norm": 9.916539192199707, "learning_rate": 6.279811779103244e-06, "loss": 0.4164, "step": 8371 }, { "epoch": 0.42349132480145685, "grad_norm": 7.946596145629883, "learning_rate": 6.279035886067812e-06, "loss": 0.462, "step": 8372 }, { "epoch": 0.423541909049522, "grad_norm": 5.869230270385742, "learning_rate": 6.278259960074713e-06, "loss": 0.3788, "step": 8373 }, { "epoch": 0.42359249329758714, "grad_norm": 10.032864570617676, "learning_rate": 6.277484001143942e-06, "loss": 0.5032, "step": 8374 }, { "epoch": 0.4236430775456523, "grad_norm": 8.609803199768066, "learning_rate": 6.276708009295496e-06, "loss": 0.3501, "step": 8375 }, { "epoch": 0.4236936617937174, "grad_norm": 9.941636085510254, "learning_rate": 6.275931984549366e-06, "loss": 0.4365, "step": 8376 }, { "epoch": 0.42374424604178257, "grad_norm": 4.63643741607666, "learning_rate": 6.275155926925551e-06, "loss": 0.3449, "step": 8377 }, { "epoch": 0.42379483028984777, "grad_norm": 5.22481107711792, "learning_rate": 6.274379836444049e-06, "loss": 0.3459, "step": 8378 }, { "epoch": 0.4238454145379129, "grad_norm": 7.630715370178223, "learning_rate": 6.273603713124856e-06, "loss": 0.4853, "step": 8379 }, { "epoch": 0.42389599878597806, "grad_norm": 14.943610191345215, "learning_rate": 6.2728275569879715e-06, "loss": 0.599, "step": 8380 }, { "epoch": 0.4239465830340432, "grad_norm": 10.301721572875977, "learning_rate": 6.2720513680533955e-06, "loss": 0.4525, "step": 8381 }, { "epoch": 0.42399716728210834, "grad_norm": 10.705804824829102, "learning_rate": 6.271275146341128e-06, "loss": 0.5201, "step": 8382 }, { "epoch": 0.4240477515301735, "grad_norm": 10.789219856262207, "learning_rate": 6.270498891871171e-06, "loss": 0.4598, "step": 8383 }, { "epoch": 0.42409833577823863, "grad_norm": 9.014171600341797, "learning_rate": 6.269722604663526e-06, "loss": 0.3141, "step": 8384 }, { "epoch": 0.42414892002630383, "grad_norm": 14.062967300415039, "learning_rate": 6.268946284738198e-06, "loss": 0.5321, "step": 8385 }, { "epoch": 0.424199504274369, "grad_norm": 6.2280354499816895, "learning_rate": 6.2681699321151865e-06, "loss": 0.404, "step": 8386 }, { "epoch": 0.4242500885224341, "grad_norm": 8.463215827941895, "learning_rate": 6.2673935468145e-06, "loss": 0.4323, "step": 8387 }, { "epoch": 0.42430067277049927, "grad_norm": 5.786350727081299, "learning_rate": 6.2666171288561425e-06, "loss": 0.3792, "step": 8388 }, { "epoch": 0.4243512570185644, "grad_norm": 11.001049041748047, "learning_rate": 6.265840678260122e-06, "loss": 0.4322, "step": 8389 }, { "epoch": 0.42440184126662955, "grad_norm": 8.577455520629883, "learning_rate": 6.2650641950464456e-06, "loss": 0.3608, "step": 8390 }, { "epoch": 0.4244524255146947, "grad_norm": 11.678069114685059, "learning_rate": 6.264287679235118e-06, "loss": 0.5282, "step": 8391 }, { "epoch": 0.4245030097627599, "grad_norm": 6.477408409118652, "learning_rate": 6.263511130846153e-06, "loss": 0.3967, "step": 8392 }, { "epoch": 0.42455359401082504, "grad_norm": 7.690115451812744, "learning_rate": 6.262734549899556e-06, "loss": 0.4232, "step": 8393 }, { "epoch": 0.4246041782588902, "grad_norm": 13.367118835449219, "learning_rate": 6.261957936415341e-06, "loss": 0.4407, "step": 8394 }, { "epoch": 0.42465476250695533, "grad_norm": 10.3423490524292, "learning_rate": 6.261181290413518e-06, "loss": 0.5384, "step": 8395 }, { "epoch": 0.4247053467550205, "grad_norm": 7.662858963012695, "learning_rate": 6.2604046119140994e-06, "loss": 0.4281, "step": 8396 }, { "epoch": 0.4247559310030856, "grad_norm": 12.697868347167969, "learning_rate": 6.259627900937099e-06, "loss": 0.4651, "step": 8397 }, { "epoch": 0.4248065152511508, "grad_norm": 7.259532928466797, "learning_rate": 6.25885115750253e-06, "loss": 0.4401, "step": 8398 }, { "epoch": 0.42485709949921596, "grad_norm": 7.600573539733887, "learning_rate": 6.2580743816304055e-06, "loss": 0.4488, "step": 8399 }, { "epoch": 0.4249076837472811, "grad_norm": 10.75174617767334, "learning_rate": 6.257297573340745e-06, "loss": 0.4879, "step": 8400 }, { "epoch": 0.42495826799534625, "grad_norm": 17.054582595825195, "learning_rate": 6.256520732653562e-06, "loss": 0.4188, "step": 8401 }, { "epoch": 0.4250088522434114, "grad_norm": 6.9820237159729, "learning_rate": 6.2557438595888735e-06, "loss": 0.4017, "step": 8402 }, { "epoch": 0.42505943649147654, "grad_norm": 9.42652702331543, "learning_rate": 6.2549669541667e-06, "loss": 0.4284, "step": 8403 }, { "epoch": 0.4251100207395417, "grad_norm": 6.103311061859131, "learning_rate": 6.25419001640706e-06, "loss": 0.3696, "step": 8404 }, { "epoch": 0.4251606049876069, "grad_norm": 6.053513526916504, "learning_rate": 6.253413046329972e-06, "loss": 0.3636, "step": 8405 }, { "epoch": 0.425211189235672, "grad_norm": 10.135078430175781, "learning_rate": 6.252636043955457e-06, "loss": 0.5161, "step": 8406 }, { "epoch": 0.42526177348373717, "grad_norm": 4.6578145027160645, "learning_rate": 6.251859009303538e-06, "loss": 0.3556, "step": 8407 }, { "epoch": 0.4253123577318023, "grad_norm": 7.077767848968506, "learning_rate": 6.251081942394235e-06, "loss": 0.4745, "step": 8408 }, { "epoch": 0.42536294197986746, "grad_norm": 8.576854705810547, "learning_rate": 6.2503048432475725e-06, "loss": 0.4761, "step": 8409 }, { "epoch": 0.4254135262279326, "grad_norm": 12.833751678466797, "learning_rate": 6.2495277118835726e-06, "loss": 0.65, "step": 8410 }, { "epoch": 0.42546411047599775, "grad_norm": 9.173463821411133, "learning_rate": 6.248750548322265e-06, "loss": 0.5264, "step": 8411 }, { "epoch": 0.42551469472406295, "grad_norm": 9.33885669708252, "learning_rate": 6.247973352583669e-06, "loss": 0.4394, "step": 8412 }, { "epoch": 0.4255652789721281, "grad_norm": 8.325191497802734, "learning_rate": 6.247196124687815e-06, "loss": 0.4242, "step": 8413 }, { "epoch": 0.42561586322019324, "grad_norm": 8.941665649414062, "learning_rate": 6.24641886465473e-06, "loss": 0.4086, "step": 8414 }, { "epoch": 0.4256664474682584, "grad_norm": 7.571623802185059, "learning_rate": 6.245641572504441e-06, "loss": 0.3748, "step": 8415 }, { "epoch": 0.4257170317163235, "grad_norm": 8.200724601745605, "learning_rate": 6.244864248256978e-06, "loss": 0.4575, "step": 8416 }, { "epoch": 0.42576761596438867, "grad_norm": 6.199142932891846, "learning_rate": 6.244086891932369e-06, "loss": 0.4918, "step": 8417 }, { "epoch": 0.42581820021245387, "grad_norm": 5.309046268463135, "learning_rate": 6.243309503550646e-06, "loss": 0.4105, "step": 8418 }, { "epoch": 0.425868784460519, "grad_norm": 8.646378517150879, "learning_rate": 6.242532083131841e-06, "loss": 0.4203, "step": 8419 }, { "epoch": 0.42591936870858416, "grad_norm": 9.358250617980957, "learning_rate": 6.241754630695985e-06, "loss": 0.4711, "step": 8420 }, { "epoch": 0.4259699529566493, "grad_norm": 5.416245460510254, "learning_rate": 6.24097714626311e-06, "loss": 0.375, "step": 8421 }, { "epoch": 0.42602053720471444, "grad_norm": 5.77104377746582, "learning_rate": 6.240199629853254e-06, "loss": 0.3885, "step": 8422 }, { "epoch": 0.4260711214527796, "grad_norm": 8.116619110107422, "learning_rate": 6.239422081486448e-06, "loss": 0.3562, "step": 8423 }, { "epoch": 0.42612170570084473, "grad_norm": 6.595596790313721, "learning_rate": 6.238644501182729e-06, "loss": 0.4164, "step": 8424 }, { "epoch": 0.42617228994890993, "grad_norm": 7.970637798309326, "learning_rate": 6.237866888962132e-06, "loss": 0.3771, "step": 8425 }, { "epoch": 0.4262228741969751, "grad_norm": 8.93491268157959, "learning_rate": 6.237089244844696e-06, "loss": 0.4665, "step": 8426 }, { "epoch": 0.4262734584450402, "grad_norm": 9.240571975708008, "learning_rate": 6.236311568850459e-06, "loss": 0.4791, "step": 8427 }, { "epoch": 0.42632404269310537, "grad_norm": 9.717509269714355, "learning_rate": 6.2355338609994585e-06, "loss": 0.3669, "step": 8428 }, { "epoch": 0.4263746269411705, "grad_norm": 8.510390281677246, "learning_rate": 6.234756121311734e-06, "loss": 0.4234, "step": 8429 }, { "epoch": 0.42642521118923565, "grad_norm": 10.114866256713867, "learning_rate": 6.233978349807327e-06, "loss": 0.4788, "step": 8430 }, { "epoch": 0.4264757954373008, "grad_norm": 8.563157081604004, "learning_rate": 6.233200546506278e-06, "loss": 0.4535, "step": 8431 }, { "epoch": 0.426526379685366, "grad_norm": 6.971102714538574, "learning_rate": 6.232422711428631e-06, "loss": 0.3931, "step": 8432 }, { "epoch": 0.42657696393343114, "grad_norm": 6.00922966003418, "learning_rate": 6.231644844594426e-06, "loss": 0.3869, "step": 8433 }, { "epoch": 0.4266275481814963, "grad_norm": 8.235626220703125, "learning_rate": 6.2308669460237085e-06, "loss": 0.4568, "step": 8434 }, { "epoch": 0.42667813242956143, "grad_norm": 9.3648099899292, "learning_rate": 6.2300890157365235e-06, "loss": 0.4314, "step": 8435 }, { "epoch": 0.4267287166776266, "grad_norm": 5.985743045806885, "learning_rate": 6.229311053752916e-06, "loss": 0.3159, "step": 8436 }, { "epoch": 0.4267793009256917, "grad_norm": 9.998019218444824, "learning_rate": 6.228533060092932e-06, "loss": 0.4681, "step": 8437 }, { "epoch": 0.4268298851737569, "grad_norm": 16.911720275878906, "learning_rate": 6.227755034776618e-06, "loss": 0.645, "step": 8438 }, { "epoch": 0.42688046942182206, "grad_norm": 10.370549201965332, "learning_rate": 6.226976977824022e-06, "loss": 0.5309, "step": 8439 }, { "epoch": 0.4269310536698872, "grad_norm": 10.872575759887695, "learning_rate": 6.226198889255192e-06, "loss": 0.552, "step": 8440 }, { "epoch": 0.42698163791795235, "grad_norm": 8.106927871704102, "learning_rate": 6.22542076909018e-06, "loss": 0.3805, "step": 8441 }, { "epoch": 0.4270322221660175, "grad_norm": 5.608200550079346, "learning_rate": 6.224642617349034e-06, "loss": 0.4285, "step": 8442 }, { "epoch": 0.42708280641408264, "grad_norm": 8.377554893493652, "learning_rate": 6.223864434051806e-06, "loss": 0.3871, "step": 8443 }, { "epoch": 0.4271333906621478, "grad_norm": 9.153497695922852, "learning_rate": 6.223086219218547e-06, "loss": 0.3681, "step": 8444 }, { "epoch": 0.427183974910213, "grad_norm": 12.840518951416016, "learning_rate": 6.222307972869312e-06, "loss": 0.538, "step": 8445 }, { "epoch": 0.4272345591582781, "grad_norm": 9.5133638381958, "learning_rate": 6.221529695024151e-06, "loss": 0.492, "step": 8446 }, { "epoch": 0.42728514340634327, "grad_norm": 8.564726829528809, "learning_rate": 6.220751385703121e-06, "loss": 0.3783, "step": 8447 }, { "epoch": 0.4273357276544084, "grad_norm": 7.001748085021973, "learning_rate": 6.219973044926276e-06, "loss": 0.4156, "step": 8448 }, { "epoch": 0.42738631190247356, "grad_norm": 9.856925010681152, "learning_rate": 6.219194672713672e-06, "loss": 0.3843, "step": 8449 }, { "epoch": 0.4274368961505387, "grad_norm": 5.5857086181640625, "learning_rate": 6.2184162690853675e-06, "loss": 0.362, "step": 8450 }, { "epoch": 0.42748748039860385, "grad_norm": 6.548710823059082, "learning_rate": 6.2176378340614194e-06, "loss": 0.4701, "step": 8451 }, { "epoch": 0.42753806464666905, "grad_norm": 5.380646705627441, "learning_rate": 6.216859367661884e-06, "loss": 0.5093, "step": 8452 }, { "epoch": 0.4275886488947342, "grad_norm": 5.7245049476623535, "learning_rate": 6.216080869906821e-06, "loss": 0.4453, "step": 8453 }, { "epoch": 0.42763923314279934, "grad_norm": 6.601527214050293, "learning_rate": 6.2153023408162925e-06, "loss": 0.3869, "step": 8454 }, { "epoch": 0.4276898173908645, "grad_norm": 6.554386615753174, "learning_rate": 6.214523780410358e-06, "loss": 0.4106, "step": 8455 }, { "epoch": 0.4277404016389296, "grad_norm": 13.960688591003418, "learning_rate": 6.2137451887090796e-06, "loss": 0.5487, "step": 8456 }, { "epoch": 0.42779098588699477, "grad_norm": 11.116192817687988, "learning_rate": 6.2129665657325185e-06, "loss": 0.3702, "step": 8457 }, { "epoch": 0.42784157013505997, "grad_norm": 9.62943172454834, "learning_rate": 6.21218791150074e-06, "loss": 0.5518, "step": 8458 }, { "epoch": 0.4278921543831251, "grad_norm": 7.190652847290039, "learning_rate": 6.211409226033806e-06, "loss": 0.4631, "step": 8459 }, { "epoch": 0.42794273863119026, "grad_norm": 8.866543769836426, "learning_rate": 6.210630509351783e-06, "loss": 0.4883, "step": 8460 }, { "epoch": 0.4279933228792554, "grad_norm": 8.756614685058594, "learning_rate": 6.209851761474734e-06, "loss": 0.458, "step": 8461 }, { "epoch": 0.42804390712732054, "grad_norm": 14.2724609375, "learning_rate": 6.2090729824227304e-06, "loss": 0.4961, "step": 8462 }, { "epoch": 0.4280944913753857, "grad_norm": 6.872296333312988, "learning_rate": 6.208294172215835e-06, "loss": 0.4304, "step": 8463 }, { "epoch": 0.42814507562345083, "grad_norm": 6.900112152099609, "learning_rate": 6.207515330874117e-06, "loss": 0.4675, "step": 8464 }, { "epoch": 0.42819565987151603, "grad_norm": 9.860116004943848, "learning_rate": 6.206736458417645e-06, "loss": 0.543, "step": 8465 }, { "epoch": 0.4282462441195812, "grad_norm": 8.915867805480957, "learning_rate": 6.20595755486649e-06, "loss": 0.4739, "step": 8466 }, { "epoch": 0.4282968283676463, "grad_norm": 8.707555770874023, "learning_rate": 6.205178620240722e-06, "loss": 0.4353, "step": 8467 }, { "epoch": 0.42834741261571146, "grad_norm": 5.966145038604736, "learning_rate": 6.20439965456041e-06, "loss": 0.4701, "step": 8468 }, { "epoch": 0.4283979968637766, "grad_norm": 6.773158073425293, "learning_rate": 6.20362065784563e-06, "loss": 0.3694, "step": 8469 }, { "epoch": 0.42844858111184175, "grad_norm": 7.47976541519165, "learning_rate": 6.202841630116453e-06, "loss": 0.3697, "step": 8470 }, { "epoch": 0.4284991653599069, "grad_norm": 9.073732376098633, "learning_rate": 6.2020625713929514e-06, "loss": 0.4616, "step": 8471 }, { "epoch": 0.4285497496079721, "grad_norm": 6.320772171020508, "learning_rate": 6.201283481695201e-06, "loss": 0.4364, "step": 8472 }, { "epoch": 0.42860033385603724, "grad_norm": 7.960508823394775, "learning_rate": 6.200504361043278e-06, "loss": 0.405, "step": 8473 }, { "epoch": 0.4286509181041024, "grad_norm": 9.074389457702637, "learning_rate": 6.199725209457256e-06, "loss": 0.4497, "step": 8474 }, { "epoch": 0.42870150235216753, "grad_norm": 6.094208717346191, "learning_rate": 6.198946026957214e-06, "loss": 0.4053, "step": 8475 }, { "epoch": 0.4287520866002327, "grad_norm": 9.287790298461914, "learning_rate": 6.198166813563229e-06, "loss": 0.5911, "step": 8476 }, { "epoch": 0.4288026708482978, "grad_norm": 8.628673553466797, "learning_rate": 6.197387569295379e-06, "loss": 0.4765, "step": 8477 }, { "epoch": 0.428853255096363, "grad_norm": 10.674250602722168, "learning_rate": 6.1966082941737425e-06, "loss": 0.5727, "step": 8478 }, { "epoch": 0.42890383934442816, "grad_norm": 7.532414436340332, "learning_rate": 6.195828988218402e-06, "loss": 0.4394, "step": 8479 }, { "epoch": 0.4289544235924933, "grad_norm": 9.029285430908203, "learning_rate": 6.195049651449436e-06, "loss": 0.4864, "step": 8480 }, { "epoch": 0.42900500784055845, "grad_norm": 13.342588424682617, "learning_rate": 6.1942702838869265e-06, "loss": 0.4287, "step": 8481 }, { "epoch": 0.4290555920886236, "grad_norm": 6.596540927886963, "learning_rate": 6.193490885550959e-06, "loss": 0.4575, "step": 8482 }, { "epoch": 0.42910617633668874, "grad_norm": 9.80318832397461, "learning_rate": 6.192711456461612e-06, "loss": 0.4926, "step": 8483 }, { "epoch": 0.4291567605847539, "grad_norm": 6.610768795013428, "learning_rate": 6.1919319966389716e-06, "loss": 0.3473, "step": 8484 }, { "epoch": 0.4292073448328191, "grad_norm": 7.262451171875, "learning_rate": 6.191152506103123e-06, "loss": 0.4638, "step": 8485 }, { "epoch": 0.4292579290808842, "grad_norm": 6.532690048217773, "learning_rate": 6.190372984874151e-06, "loss": 0.44, "step": 8486 }, { "epoch": 0.42930851332894937, "grad_norm": 11.204641342163086, "learning_rate": 6.189593432972142e-06, "loss": 0.5624, "step": 8487 }, { "epoch": 0.4293590975770145, "grad_norm": 6.489688396453857, "learning_rate": 6.188813850417186e-06, "loss": 0.3894, "step": 8488 }, { "epoch": 0.42940968182507966, "grad_norm": 7.771980285644531, "learning_rate": 6.188034237229367e-06, "loss": 0.478, "step": 8489 }, { "epoch": 0.4294602660731448, "grad_norm": 16.680675506591797, "learning_rate": 6.187254593428774e-06, "loss": 0.3988, "step": 8490 }, { "epoch": 0.42951085032121, "grad_norm": 8.515344619750977, "learning_rate": 6.186474919035499e-06, "loss": 0.3989, "step": 8491 }, { "epoch": 0.42956143456927515, "grad_norm": 6.434767246246338, "learning_rate": 6.185695214069632e-06, "loss": 0.3892, "step": 8492 }, { "epoch": 0.4296120188173403, "grad_norm": 8.580533981323242, "learning_rate": 6.184915478551261e-06, "loss": 0.499, "step": 8493 }, { "epoch": 0.42966260306540544, "grad_norm": 6.059906959533691, "learning_rate": 6.184135712500483e-06, "loss": 0.4638, "step": 8494 }, { "epoch": 0.4297131873134706, "grad_norm": 4.66043758392334, "learning_rate": 6.183355915937387e-06, "loss": 0.2954, "step": 8495 }, { "epoch": 0.4297637715615357, "grad_norm": 7.410721778869629, "learning_rate": 6.182576088882067e-06, "loss": 0.4554, "step": 8496 }, { "epoch": 0.42981435580960087, "grad_norm": 7.873166561126709, "learning_rate": 6.1817962313546165e-06, "loss": 0.417, "step": 8497 }, { "epoch": 0.42986494005766607, "grad_norm": 6.622179985046387, "learning_rate": 6.1810163433751326e-06, "loss": 0.3541, "step": 8498 }, { "epoch": 0.4299155243057312, "grad_norm": 7.040981769561768, "learning_rate": 6.180236424963709e-06, "loss": 0.3952, "step": 8499 }, { "epoch": 0.42996610855379636, "grad_norm": 5.130168914794922, "learning_rate": 6.179456476140444e-06, "loss": 0.3436, "step": 8500 }, { "epoch": 0.4300166928018615, "grad_norm": 8.136345863342285, "learning_rate": 6.178676496925438e-06, "loss": 0.4568, "step": 8501 }, { "epoch": 0.43006727704992664, "grad_norm": 6.958675861358643, "learning_rate": 6.177896487338781e-06, "loss": 0.3578, "step": 8502 }, { "epoch": 0.4301178612979918, "grad_norm": 7.517849922180176, "learning_rate": 6.1771164474005786e-06, "loss": 0.5127, "step": 8503 }, { "epoch": 0.43016844554605693, "grad_norm": 9.444709777832031, "learning_rate": 6.176336377130928e-06, "loss": 0.3968, "step": 8504 }, { "epoch": 0.43021902979412213, "grad_norm": 7.660740375518799, "learning_rate": 6.17555627654993e-06, "loss": 0.3931, "step": 8505 }, { "epoch": 0.4302696140421873, "grad_norm": 5.899432182312012, "learning_rate": 6.174776145677686e-06, "loss": 0.3387, "step": 8506 }, { "epoch": 0.4303201982902524, "grad_norm": 7.25689172744751, "learning_rate": 6.1739959845343e-06, "loss": 0.375, "step": 8507 }, { "epoch": 0.43037078253831756, "grad_norm": 8.702521324157715, "learning_rate": 6.173215793139873e-06, "loss": 0.4649, "step": 8508 }, { "epoch": 0.4304213667863827, "grad_norm": 10.89584732055664, "learning_rate": 6.172435571514508e-06, "loss": 0.3592, "step": 8509 }, { "epoch": 0.43047195103444785, "grad_norm": 9.706174850463867, "learning_rate": 6.171655319678309e-06, "loss": 0.477, "step": 8510 }, { "epoch": 0.43052253528251305, "grad_norm": 16.559722900390625, "learning_rate": 6.170875037651385e-06, "loss": 0.5177, "step": 8511 }, { "epoch": 0.4305731195305782, "grad_norm": 5.652673244476318, "learning_rate": 6.170094725453837e-06, "loss": 0.3443, "step": 8512 }, { "epoch": 0.43062370377864334, "grad_norm": 8.334904670715332, "learning_rate": 6.1693143831057755e-06, "loss": 0.3633, "step": 8513 }, { "epoch": 0.4306742880267085, "grad_norm": 8.420463562011719, "learning_rate": 6.168534010627307e-06, "loss": 0.3743, "step": 8514 }, { "epoch": 0.43072487227477363, "grad_norm": 7.142914295196533, "learning_rate": 6.167753608038538e-06, "loss": 0.5212, "step": 8515 }, { "epoch": 0.4307754565228388, "grad_norm": 11.234016418457031, "learning_rate": 6.166973175359579e-06, "loss": 0.5619, "step": 8516 }, { "epoch": 0.4308260407709039, "grad_norm": 11.302424430847168, "learning_rate": 6.166192712610542e-06, "loss": 0.4968, "step": 8517 }, { "epoch": 0.4308766250189691, "grad_norm": 5.337893486022949, "learning_rate": 6.165412219811535e-06, "loss": 0.4327, "step": 8518 }, { "epoch": 0.43092720926703426, "grad_norm": 6.906888008117676, "learning_rate": 6.164631696982669e-06, "loss": 0.3186, "step": 8519 }, { "epoch": 0.4309777935150994, "grad_norm": 7.983859062194824, "learning_rate": 6.163851144144059e-06, "loss": 0.4134, "step": 8520 }, { "epoch": 0.43102837776316455, "grad_norm": 5.223088264465332, "learning_rate": 6.163070561315815e-06, "loss": 0.4734, "step": 8521 }, { "epoch": 0.4310789620112297, "grad_norm": 9.273737907409668, "learning_rate": 6.162289948518052e-06, "loss": 0.4572, "step": 8522 }, { "epoch": 0.43112954625929484, "grad_norm": 12.392457962036133, "learning_rate": 6.161509305770884e-06, "loss": 0.3583, "step": 8523 }, { "epoch": 0.43118013050736, "grad_norm": 7.140173435211182, "learning_rate": 6.160728633094427e-06, "loss": 0.4309, "step": 8524 }, { "epoch": 0.4312307147554252, "grad_norm": 8.811921119689941, "learning_rate": 6.159947930508796e-06, "loss": 0.4407, "step": 8525 }, { "epoch": 0.4312812990034903, "grad_norm": 6.491818904876709, "learning_rate": 6.159167198034111e-06, "loss": 0.4442, "step": 8526 }, { "epoch": 0.43133188325155547, "grad_norm": 8.227235794067383, "learning_rate": 6.158386435690485e-06, "loss": 0.373, "step": 8527 }, { "epoch": 0.4313824674996206, "grad_norm": 10.077914237976074, "learning_rate": 6.157605643498039e-06, "loss": 0.5867, "step": 8528 }, { "epoch": 0.43143305174768576, "grad_norm": 7.307486057281494, "learning_rate": 6.156824821476891e-06, "loss": 0.4773, "step": 8529 }, { "epoch": 0.4314836359957509, "grad_norm": 7.174088478088379, "learning_rate": 6.156043969647163e-06, "loss": 0.3986, "step": 8530 }, { "epoch": 0.4315342202438161, "grad_norm": 6.067541599273682, "learning_rate": 6.155263088028973e-06, "loss": 0.419, "step": 8531 }, { "epoch": 0.43158480449188125, "grad_norm": 11.473814010620117, "learning_rate": 6.1544821766424455e-06, "loss": 0.5962, "step": 8532 }, { "epoch": 0.4316353887399464, "grad_norm": 6.61965274810791, "learning_rate": 6.1537012355077e-06, "loss": 0.3839, "step": 8533 }, { "epoch": 0.43168597298801153, "grad_norm": 6.288788318634033, "learning_rate": 6.1529202646448595e-06, "loss": 0.4256, "step": 8534 }, { "epoch": 0.4317365572360767, "grad_norm": 7.828334808349609, "learning_rate": 6.1521392640740485e-06, "loss": 0.3051, "step": 8535 }, { "epoch": 0.4317871414841418, "grad_norm": 4.911783218383789, "learning_rate": 6.151358233815392e-06, "loss": 0.3751, "step": 8536 }, { "epoch": 0.43183772573220697, "grad_norm": 13.369852066040039, "learning_rate": 6.150577173889014e-06, "loss": 0.6486, "step": 8537 }, { "epoch": 0.43188830998027217, "grad_norm": 7.285832405090332, "learning_rate": 6.149796084315042e-06, "loss": 0.4129, "step": 8538 }, { "epoch": 0.4319388942283373, "grad_norm": 4.878264904022217, "learning_rate": 6.149014965113602e-06, "loss": 0.4202, "step": 8539 }, { "epoch": 0.43198947847640246, "grad_norm": 5.0756988525390625, "learning_rate": 6.148233816304823e-06, "loss": 0.4476, "step": 8540 }, { "epoch": 0.4320400627244676, "grad_norm": 4.079211711883545, "learning_rate": 6.147452637908831e-06, "loss": 0.3834, "step": 8541 }, { "epoch": 0.43209064697253274, "grad_norm": 10.706700325012207, "learning_rate": 6.146671429945755e-06, "loss": 0.4956, "step": 8542 }, { "epoch": 0.4321412312205979, "grad_norm": 5.746606349945068, "learning_rate": 6.145890192435727e-06, "loss": 0.3475, "step": 8543 }, { "epoch": 0.43219181546866303, "grad_norm": 8.237696647644043, "learning_rate": 6.145108925398876e-06, "loss": 0.4927, "step": 8544 }, { "epoch": 0.43224239971672823, "grad_norm": 7.3821187019348145, "learning_rate": 6.144327628855334e-06, "loss": 0.2639, "step": 8545 }, { "epoch": 0.4322929839647934, "grad_norm": 8.048999786376953, "learning_rate": 6.143546302825235e-06, "loss": 0.4509, "step": 8546 }, { "epoch": 0.4323435682128585, "grad_norm": 8.854543685913086, "learning_rate": 6.142764947328708e-06, "loss": 0.4439, "step": 8547 }, { "epoch": 0.43239415246092366, "grad_norm": 8.177420616149902, "learning_rate": 6.141983562385887e-06, "loss": 0.4752, "step": 8548 }, { "epoch": 0.4324447367089888, "grad_norm": 8.704629898071289, "learning_rate": 6.1412021480169105e-06, "loss": 0.3984, "step": 8549 }, { "epoch": 0.43249532095705395, "grad_norm": 9.69304084777832, "learning_rate": 6.1404207042419095e-06, "loss": 0.4299, "step": 8550 }, { "epoch": 0.43254590520511915, "grad_norm": 7.879025459289551, "learning_rate": 6.139639231081024e-06, "loss": 0.4353, "step": 8551 }, { "epoch": 0.4325964894531843, "grad_norm": 7.3072614669799805, "learning_rate": 6.138857728554386e-06, "loss": 0.4271, "step": 8552 }, { "epoch": 0.43264707370124944, "grad_norm": 6.4803080558776855, "learning_rate": 6.138076196682135e-06, "loss": 0.3689, "step": 8553 }, { "epoch": 0.4326976579493146, "grad_norm": 9.563048362731934, "learning_rate": 6.1372946354844095e-06, "loss": 0.5323, "step": 8554 }, { "epoch": 0.43274824219737973, "grad_norm": 7.21331787109375, "learning_rate": 6.13651304498135e-06, "loss": 0.3669, "step": 8555 }, { "epoch": 0.4327988264454449, "grad_norm": 12.229747772216797, "learning_rate": 6.1357314251930925e-06, "loss": 0.4484, "step": 8556 }, { "epoch": 0.43284941069351, "grad_norm": 7.7092695236206055, "learning_rate": 6.13494977613978e-06, "loss": 0.4037, "step": 8557 }, { "epoch": 0.4328999949415752, "grad_norm": 8.310295104980469, "learning_rate": 6.134168097841554e-06, "loss": 0.4739, "step": 8558 }, { "epoch": 0.43295057918964036, "grad_norm": 11.009514808654785, "learning_rate": 6.133386390318555e-06, "loss": 0.6014, "step": 8559 }, { "epoch": 0.4330011634377055, "grad_norm": 6.423457145690918, "learning_rate": 6.132604653590926e-06, "loss": 0.4024, "step": 8560 }, { "epoch": 0.43305174768577065, "grad_norm": 9.055647850036621, "learning_rate": 6.131822887678811e-06, "loss": 0.4858, "step": 8561 }, { "epoch": 0.4331023319338358, "grad_norm": 6.2291340827941895, "learning_rate": 6.131041092602354e-06, "loss": 0.3293, "step": 8562 }, { "epoch": 0.43315291618190094, "grad_norm": 6.897702217102051, "learning_rate": 6.130259268381699e-06, "loss": 0.3271, "step": 8563 }, { "epoch": 0.4332035004299661, "grad_norm": 4.558736324310303, "learning_rate": 6.129477415036994e-06, "loss": 0.2902, "step": 8564 }, { "epoch": 0.4332540846780313, "grad_norm": 5.45925760269165, "learning_rate": 6.128695532588384e-06, "loss": 0.4007, "step": 8565 }, { "epoch": 0.4333046689260964, "grad_norm": 9.254961967468262, "learning_rate": 6.127913621056016e-06, "loss": 0.4716, "step": 8566 }, { "epoch": 0.43335525317416157, "grad_norm": 6.571776390075684, "learning_rate": 6.127131680460039e-06, "loss": 0.4591, "step": 8567 }, { "epoch": 0.4334058374222267, "grad_norm": 7.870950222015381, "learning_rate": 6.126349710820599e-06, "loss": 0.4491, "step": 8568 }, { "epoch": 0.43345642167029186, "grad_norm": 8.414287567138672, "learning_rate": 6.12556771215785e-06, "loss": 0.4688, "step": 8569 }, { "epoch": 0.433507005918357, "grad_norm": 7.9608354568481445, "learning_rate": 6.1247856844919385e-06, "loss": 0.435, "step": 8570 }, { "epoch": 0.4335575901664222, "grad_norm": 8.325706481933594, "learning_rate": 6.124003627843017e-06, "loss": 0.3915, "step": 8571 }, { "epoch": 0.43360817441448735, "grad_norm": 9.525257110595703, "learning_rate": 6.123221542231237e-06, "loss": 0.4559, "step": 8572 }, { "epoch": 0.4336587586625525, "grad_norm": 11.259235382080078, "learning_rate": 6.1224394276767505e-06, "loss": 0.4205, "step": 8573 }, { "epoch": 0.43370934291061763, "grad_norm": 6.281006336212158, "learning_rate": 6.121657284199712e-06, "loss": 0.3515, "step": 8574 }, { "epoch": 0.4337599271586828, "grad_norm": 5.881289958953857, "learning_rate": 6.120875111820274e-06, "loss": 0.407, "step": 8575 }, { "epoch": 0.4338105114067479, "grad_norm": 10.006317138671875, "learning_rate": 6.12009291055859e-06, "loss": 0.4404, "step": 8576 }, { "epoch": 0.43386109565481307, "grad_norm": 5.803414821624756, "learning_rate": 6.11931068043482e-06, "loss": 0.4301, "step": 8577 }, { "epoch": 0.43391167990287827, "grad_norm": 10.591765403747559, "learning_rate": 6.118528421469117e-06, "loss": 0.4756, "step": 8578 }, { "epoch": 0.4339622641509434, "grad_norm": 8.731236457824707, "learning_rate": 6.1177461336816365e-06, "loss": 0.5117, "step": 8579 }, { "epoch": 0.43401284839900856, "grad_norm": 9.238865852355957, "learning_rate": 6.1169638170925404e-06, "loss": 0.5004, "step": 8580 }, { "epoch": 0.4340634326470737, "grad_norm": 12.044771194458008, "learning_rate": 6.116181471721982e-06, "loss": 0.4191, "step": 8581 }, { "epoch": 0.43411401689513884, "grad_norm": 9.5150785446167, "learning_rate": 6.115399097590125e-06, "loss": 0.449, "step": 8582 }, { "epoch": 0.434164601143204, "grad_norm": 11.33057689666748, "learning_rate": 6.114616694717126e-06, "loss": 0.4585, "step": 8583 }, { "epoch": 0.43421518539126913, "grad_norm": 6.182518482208252, "learning_rate": 6.113834263123148e-06, "loss": 0.3998, "step": 8584 }, { "epoch": 0.43426576963933433, "grad_norm": 6.150864124298096, "learning_rate": 6.113051802828351e-06, "loss": 0.3411, "step": 8585 }, { "epoch": 0.4343163538873995, "grad_norm": 8.416909217834473, "learning_rate": 6.112269313852898e-06, "loss": 0.4286, "step": 8586 }, { "epoch": 0.4343669381354646, "grad_norm": 5.566344738006592, "learning_rate": 6.11148679621695e-06, "loss": 0.3791, "step": 8587 }, { "epoch": 0.43441752238352976, "grad_norm": 25.887643814086914, "learning_rate": 6.1107042499406736e-06, "loss": 0.5416, "step": 8588 }, { "epoch": 0.4344681066315949, "grad_norm": 8.01188850402832, "learning_rate": 6.10992167504423e-06, "loss": 0.3975, "step": 8589 }, { "epoch": 0.43451869087966005, "grad_norm": 5.215485572814941, "learning_rate": 6.109139071547787e-06, "loss": 0.378, "step": 8590 }, { "epoch": 0.43456927512772525, "grad_norm": 6.347697734832764, "learning_rate": 6.108356439471509e-06, "loss": 0.451, "step": 8591 }, { "epoch": 0.4346198593757904, "grad_norm": 10.607281684875488, "learning_rate": 6.107573778835562e-06, "loss": 0.469, "step": 8592 }, { "epoch": 0.43467044362385554, "grad_norm": 6.324544906616211, "learning_rate": 6.106791089660115e-06, "loss": 0.3658, "step": 8593 }, { "epoch": 0.4347210278719207, "grad_norm": 4.855567932128906, "learning_rate": 6.106008371965333e-06, "loss": 0.4245, "step": 8594 }, { "epoch": 0.43477161211998583, "grad_norm": 8.138607025146484, "learning_rate": 6.105225625771388e-06, "loss": 0.3136, "step": 8595 }, { "epoch": 0.434822196368051, "grad_norm": 5.592957973480225, "learning_rate": 6.104442851098449e-06, "loss": 0.4392, "step": 8596 }, { "epoch": 0.4348727806161161, "grad_norm": 11.600223541259766, "learning_rate": 6.103660047966684e-06, "loss": 0.3837, "step": 8597 }, { "epoch": 0.4349233648641813, "grad_norm": 6.557425022125244, "learning_rate": 6.102877216396266e-06, "loss": 0.3638, "step": 8598 }, { "epoch": 0.43497394911224646, "grad_norm": 9.184473037719727, "learning_rate": 6.102094356407367e-06, "loss": 0.4085, "step": 8599 }, { "epoch": 0.4350245333603116, "grad_norm": 12.675002098083496, "learning_rate": 6.101311468020157e-06, "loss": 0.5068, "step": 8600 }, { "epoch": 0.43507511760837675, "grad_norm": 8.252405166625977, "learning_rate": 6.10052855125481e-06, "loss": 0.4209, "step": 8601 }, { "epoch": 0.4351257018564419, "grad_norm": 7.838190078735352, "learning_rate": 6.099745606131502e-06, "loss": 0.42, "step": 8602 }, { "epoch": 0.43517628610450704, "grad_norm": 7.939213275909424, "learning_rate": 6.098962632670404e-06, "loss": 0.4656, "step": 8603 }, { "epoch": 0.4352268703525722, "grad_norm": 6.079672813415527, "learning_rate": 6.0981796308916965e-06, "loss": 0.4078, "step": 8604 }, { "epoch": 0.4352774546006374, "grad_norm": 12.02700138092041, "learning_rate": 6.09739660081555e-06, "loss": 0.4225, "step": 8605 }, { "epoch": 0.4353280388487025, "grad_norm": 7.871408462524414, "learning_rate": 6.096613542462145e-06, "loss": 0.466, "step": 8606 }, { "epoch": 0.43537862309676767, "grad_norm": 7.334045886993408, "learning_rate": 6.095830455851657e-06, "loss": 0.4272, "step": 8607 }, { "epoch": 0.4354292073448328, "grad_norm": 6.345044136047363, "learning_rate": 6.095047341004265e-06, "loss": 0.3789, "step": 8608 }, { "epoch": 0.43547979159289796, "grad_norm": 8.357096672058105, "learning_rate": 6.094264197940149e-06, "loss": 0.4624, "step": 8609 }, { "epoch": 0.4355303758409631, "grad_norm": 8.690570831298828, "learning_rate": 6.093481026679488e-06, "loss": 0.4241, "step": 8610 }, { "epoch": 0.4355809600890283, "grad_norm": 6.086642742156982, "learning_rate": 6.09269782724246e-06, "loss": 0.3937, "step": 8611 }, { "epoch": 0.43563154433709345, "grad_norm": 17.19099235534668, "learning_rate": 6.09191459964925e-06, "loss": 0.5252, "step": 8612 }, { "epoch": 0.4356821285851586, "grad_norm": 9.674683570861816, "learning_rate": 6.091131343920037e-06, "loss": 0.367, "step": 8613 }, { "epoch": 0.43573271283322373, "grad_norm": 9.869898796081543, "learning_rate": 6.090348060075004e-06, "loss": 0.4477, "step": 8614 }, { "epoch": 0.4357832970812889, "grad_norm": 10.406200408935547, "learning_rate": 6.0895647481343374e-06, "loss": 0.5237, "step": 8615 }, { "epoch": 0.435833881329354, "grad_norm": 10.597755432128906, "learning_rate": 6.088781408118218e-06, "loss": 0.4058, "step": 8616 }, { "epoch": 0.43588446557741917, "grad_norm": 9.15572738647461, "learning_rate": 6.087998040046832e-06, "loss": 0.4292, "step": 8617 }, { "epoch": 0.43593504982548437, "grad_norm": 13.685867309570312, "learning_rate": 6.087214643940365e-06, "loss": 0.6056, "step": 8618 }, { "epoch": 0.4359856340735495, "grad_norm": 6.139028072357178, "learning_rate": 6.086431219819001e-06, "loss": 0.431, "step": 8619 }, { "epoch": 0.43603621832161465, "grad_norm": 6.4235382080078125, "learning_rate": 6.085647767702929e-06, "loss": 0.4701, "step": 8620 }, { "epoch": 0.4360868025696798, "grad_norm": 6.306440353393555, "learning_rate": 6.084864287612338e-06, "loss": 0.387, "step": 8621 }, { "epoch": 0.43613738681774494, "grad_norm": 4.945326805114746, "learning_rate": 6.084080779567413e-06, "loss": 0.3008, "step": 8622 }, { "epoch": 0.4361879710658101, "grad_norm": 16.035747528076172, "learning_rate": 6.083297243588345e-06, "loss": 0.4929, "step": 8623 }, { "epoch": 0.4362385553138753, "grad_norm": 11.042341232299805, "learning_rate": 6.082513679695325e-06, "loss": 0.3576, "step": 8624 }, { "epoch": 0.43628913956194043, "grad_norm": 8.144747734069824, "learning_rate": 6.08173008790854e-06, "loss": 0.3671, "step": 8625 }, { "epoch": 0.4363397238100056, "grad_norm": 6.233855724334717, "learning_rate": 6.080946468248184e-06, "loss": 0.4485, "step": 8626 }, { "epoch": 0.4363903080580707, "grad_norm": 5.913686275482178, "learning_rate": 6.08016282073445e-06, "loss": 0.4068, "step": 8627 }, { "epoch": 0.43644089230613586, "grad_norm": 5.454015254974365, "learning_rate": 6.0793791453875276e-06, "loss": 0.3206, "step": 8628 }, { "epoch": 0.436491476554201, "grad_norm": 5.89342737197876, "learning_rate": 6.078595442227612e-06, "loss": 0.4157, "step": 8629 }, { "epoch": 0.43654206080226615, "grad_norm": 5.675137996673584, "learning_rate": 6.077811711274898e-06, "loss": 0.3671, "step": 8630 }, { "epoch": 0.43659264505033135, "grad_norm": 5.802275657653809, "learning_rate": 6.077027952549579e-06, "loss": 0.4022, "step": 8631 }, { "epoch": 0.4366432292983965, "grad_norm": 8.45527458190918, "learning_rate": 6.076244166071851e-06, "loss": 0.5166, "step": 8632 }, { "epoch": 0.43669381354646164, "grad_norm": 10.215497016906738, "learning_rate": 6.075460351861909e-06, "loss": 0.3761, "step": 8633 }, { "epoch": 0.4367443977945268, "grad_norm": 6.024811744689941, "learning_rate": 6.074676509939954e-06, "loss": 0.4135, "step": 8634 }, { "epoch": 0.43679498204259193, "grad_norm": 10.625531196594238, "learning_rate": 6.073892640326179e-06, "loss": 0.5809, "step": 8635 }, { "epoch": 0.4368455662906571, "grad_norm": 7.524846076965332, "learning_rate": 6.073108743040787e-06, "loss": 0.2985, "step": 8636 }, { "epoch": 0.4368961505387222, "grad_norm": 12.77003288269043, "learning_rate": 6.072324818103973e-06, "loss": 0.447, "step": 8637 }, { "epoch": 0.4369467347867874, "grad_norm": 9.6484375, "learning_rate": 6.071540865535939e-06, "loss": 0.3638, "step": 8638 }, { "epoch": 0.43699731903485256, "grad_norm": 6.632007122039795, "learning_rate": 6.0707568853568835e-06, "loss": 0.4798, "step": 8639 }, { "epoch": 0.4370479032829177, "grad_norm": 7.013364791870117, "learning_rate": 6.069972877587012e-06, "loss": 0.4379, "step": 8640 }, { "epoch": 0.43709848753098285, "grad_norm": 9.387937545776367, "learning_rate": 6.069188842246521e-06, "loss": 0.5381, "step": 8641 }, { "epoch": 0.437149071779048, "grad_norm": 19.5688419342041, "learning_rate": 6.0684047793556165e-06, "loss": 0.7279, "step": 8642 }, { "epoch": 0.43719965602711314, "grad_norm": 5.329463958740234, "learning_rate": 6.067620688934505e-06, "loss": 0.3911, "step": 8643 }, { "epoch": 0.43725024027517834, "grad_norm": 13.758804321289062, "learning_rate": 6.066836571003384e-06, "loss": 0.6544, "step": 8644 }, { "epoch": 0.4373008245232435, "grad_norm": 8.243001937866211, "learning_rate": 6.066052425582461e-06, "loss": 0.3688, "step": 8645 }, { "epoch": 0.4373514087713086, "grad_norm": 15.439208984375, "learning_rate": 6.0652682526919436e-06, "loss": 0.644, "step": 8646 }, { "epoch": 0.43740199301937377, "grad_norm": 8.717403411865234, "learning_rate": 6.064484052352036e-06, "loss": 0.4334, "step": 8647 }, { "epoch": 0.4374525772674389, "grad_norm": 7.500767230987549, "learning_rate": 6.0636998245829444e-06, "loss": 0.4646, "step": 8648 }, { "epoch": 0.43750316151550406, "grad_norm": 5.895767688751221, "learning_rate": 6.062915569404879e-06, "loss": 0.4424, "step": 8649 }, { "epoch": 0.4375537457635692, "grad_norm": 11.515382766723633, "learning_rate": 6.062131286838047e-06, "loss": 0.4863, "step": 8650 }, { "epoch": 0.4376043300116344, "grad_norm": 6.342658996582031, "learning_rate": 6.061346976902656e-06, "loss": 0.4216, "step": 8651 }, { "epoch": 0.43765491425969955, "grad_norm": 9.795693397521973, "learning_rate": 6.0605626396189165e-06, "loss": 0.3973, "step": 8652 }, { "epoch": 0.4377054985077647, "grad_norm": 4.891623497009277, "learning_rate": 6.0597782750070414e-06, "loss": 0.3362, "step": 8653 }, { "epoch": 0.43775608275582983, "grad_norm": 14.271879196166992, "learning_rate": 6.0589938830872375e-06, "loss": 0.59, "step": 8654 }, { "epoch": 0.437806667003895, "grad_norm": 5.9222517013549805, "learning_rate": 6.058209463879721e-06, "loss": 0.3272, "step": 8655 }, { "epoch": 0.4378572512519601, "grad_norm": 11.514071464538574, "learning_rate": 6.057425017404703e-06, "loss": 0.4284, "step": 8656 }, { "epoch": 0.43790783550002527, "grad_norm": 5.452831745147705, "learning_rate": 6.056640543682395e-06, "loss": 0.3731, "step": 8657 }, { "epoch": 0.43795841974809047, "grad_norm": 7.284152507781982, "learning_rate": 6.0558560427330125e-06, "loss": 0.409, "step": 8658 }, { "epoch": 0.4380090039961556, "grad_norm": 12.512473106384277, "learning_rate": 6.0550715145767715e-06, "loss": 0.5801, "step": 8659 }, { "epoch": 0.43805958824422075, "grad_norm": 7.952762603759766, "learning_rate": 6.054286959233884e-06, "loss": 0.4855, "step": 8660 }, { "epoch": 0.4381101724922859, "grad_norm": 9.047264099121094, "learning_rate": 6.053502376724568e-06, "loss": 0.4871, "step": 8661 }, { "epoch": 0.43816075674035104, "grad_norm": 6.691107749938965, "learning_rate": 6.052717767069043e-06, "loss": 0.3297, "step": 8662 }, { "epoch": 0.4382113409884162, "grad_norm": 25.487899780273438, "learning_rate": 6.051933130287522e-06, "loss": 0.5914, "step": 8663 }, { "epoch": 0.4382619252364814, "grad_norm": 7.382287979125977, "learning_rate": 6.051148466400225e-06, "loss": 0.3562, "step": 8664 }, { "epoch": 0.43831250948454653, "grad_norm": 6.42455530166626, "learning_rate": 6.050363775427372e-06, "loss": 0.4497, "step": 8665 }, { "epoch": 0.4383630937326117, "grad_norm": 13.514106750488281, "learning_rate": 6.049579057389181e-06, "loss": 0.5931, "step": 8666 }, { "epoch": 0.4384136779806768, "grad_norm": 14.6267728805542, "learning_rate": 6.048794312305872e-06, "loss": 0.5502, "step": 8667 }, { "epoch": 0.43846426222874196, "grad_norm": 7.748539924621582, "learning_rate": 6.048009540197668e-06, "loss": 0.3567, "step": 8668 }, { "epoch": 0.4385148464768071, "grad_norm": 7.563710689544678, "learning_rate": 6.04722474108479e-06, "loss": 0.3644, "step": 8669 }, { "epoch": 0.43856543072487225, "grad_norm": 28.20769691467285, "learning_rate": 6.046439914987458e-06, "loss": 0.5229, "step": 8670 }, { "epoch": 0.43861601497293745, "grad_norm": 11.587445259094238, "learning_rate": 6.045655061925899e-06, "loss": 0.4342, "step": 8671 }, { "epoch": 0.4386665992210026, "grad_norm": 6.424844264984131, "learning_rate": 6.044870181920332e-06, "loss": 0.4603, "step": 8672 }, { "epoch": 0.43871718346906774, "grad_norm": 10.823799133300781, "learning_rate": 6.044085274990986e-06, "loss": 0.5596, "step": 8673 }, { "epoch": 0.4387677677171329, "grad_norm": 5.8090500831604, "learning_rate": 6.043300341158085e-06, "loss": 0.3309, "step": 8674 }, { "epoch": 0.43881835196519803, "grad_norm": 9.64500617980957, "learning_rate": 6.042515380441854e-06, "loss": 0.3969, "step": 8675 }, { "epoch": 0.43886893621326317, "grad_norm": 9.24907112121582, "learning_rate": 6.0417303928625195e-06, "loss": 0.3186, "step": 8676 }, { "epoch": 0.4389195204613283, "grad_norm": 5.260766983032227, "learning_rate": 6.040945378440307e-06, "loss": 0.3481, "step": 8677 }, { "epoch": 0.4389701047093935, "grad_norm": 8.192567825317383, "learning_rate": 6.040160337195448e-06, "loss": 0.4338, "step": 8678 }, { "epoch": 0.43902068895745866, "grad_norm": 6.788429260253906, "learning_rate": 6.03937526914817e-06, "loss": 0.2698, "step": 8679 }, { "epoch": 0.4390712732055238, "grad_norm": 5.195126056671143, "learning_rate": 6.0385901743187015e-06, "loss": 0.3714, "step": 8680 }, { "epoch": 0.43912185745358895, "grad_norm": 8.892702102661133, "learning_rate": 6.037805052727274e-06, "loss": 0.4279, "step": 8681 }, { "epoch": 0.4391724417016541, "grad_norm": 5.771860599517822, "learning_rate": 6.0370199043941144e-06, "loss": 0.4046, "step": 8682 }, { "epoch": 0.43922302594971924, "grad_norm": 6.992618560791016, "learning_rate": 6.036234729339459e-06, "loss": 0.3742, "step": 8683 }, { "epoch": 0.43927361019778444, "grad_norm": 11.713456153869629, "learning_rate": 6.035449527583537e-06, "loss": 0.3688, "step": 8684 }, { "epoch": 0.4393241944458496, "grad_norm": 7.467779159545898, "learning_rate": 6.03466429914658e-06, "loss": 0.4192, "step": 8685 }, { "epoch": 0.4393747786939147, "grad_norm": 8.950392723083496, "learning_rate": 6.0338790440488236e-06, "loss": 0.4082, "step": 8686 }, { "epoch": 0.43942536294197987, "grad_norm": 8.309306144714355, "learning_rate": 6.0330937623105036e-06, "loss": 0.503, "step": 8687 }, { "epoch": 0.439475947190045, "grad_norm": 6.130568027496338, "learning_rate": 6.0323084539518515e-06, "loss": 0.4316, "step": 8688 }, { "epoch": 0.43952653143811016, "grad_norm": 11.56529712677002, "learning_rate": 6.031523118993104e-06, "loss": 0.448, "step": 8689 }, { "epoch": 0.4395771156861753, "grad_norm": 6.977741241455078, "learning_rate": 6.030737757454497e-06, "loss": 0.3314, "step": 8690 }, { "epoch": 0.4396276999342405, "grad_norm": 10.423212051391602, "learning_rate": 6.0299523693562665e-06, "loss": 0.4862, "step": 8691 }, { "epoch": 0.43967828418230565, "grad_norm": 6.210246562957764, "learning_rate": 6.029166954718652e-06, "loss": 0.4228, "step": 8692 }, { "epoch": 0.4397288684303708, "grad_norm": 5.987614154815674, "learning_rate": 6.028381513561893e-06, "loss": 0.4516, "step": 8693 }, { "epoch": 0.43977945267843593, "grad_norm": 8.759305000305176, "learning_rate": 6.027596045906224e-06, "loss": 0.4387, "step": 8694 }, { "epoch": 0.4398300369265011, "grad_norm": 10.68584156036377, "learning_rate": 6.026810551771888e-06, "loss": 0.4672, "step": 8695 }, { "epoch": 0.4398806211745662, "grad_norm": 8.993841171264648, "learning_rate": 6.026025031179123e-06, "loss": 0.4144, "step": 8696 }, { "epoch": 0.43993120542263137, "grad_norm": 7.743603706359863, "learning_rate": 6.025239484148173e-06, "loss": 0.4994, "step": 8697 }, { "epoch": 0.43998178967069657, "grad_norm": 11.071297645568848, "learning_rate": 6.024453910699276e-06, "loss": 0.4149, "step": 8698 }, { "epoch": 0.4400323739187617, "grad_norm": 10.871438026428223, "learning_rate": 6.023668310852678e-06, "loss": 0.3612, "step": 8699 }, { "epoch": 0.44008295816682685, "grad_norm": 7.10801887512207, "learning_rate": 6.02288268462862e-06, "loss": 0.3777, "step": 8700 }, { "epoch": 0.440133542414892, "grad_norm": 7.020914077758789, "learning_rate": 6.022097032047345e-06, "loss": 0.3772, "step": 8701 }, { "epoch": 0.44018412666295714, "grad_norm": 17.335180282592773, "learning_rate": 6.021311353129098e-06, "loss": 0.4569, "step": 8702 }, { "epoch": 0.4402347109110223, "grad_norm": 17.907934188842773, "learning_rate": 6.020525647894125e-06, "loss": 0.5811, "step": 8703 }, { "epoch": 0.4402852951590875, "grad_norm": 5.681807518005371, "learning_rate": 6.01973991636267e-06, "loss": 0.3705, "step": 8704 }, { "epoch": 0.44033587940715263, "grad_norm": 8.221270561218262, "learning_rate": 6.018954158554981e-06, "loss": 0.5248, "step": 8705 }, { "epoch": 0.4403864636552178, "grad_norm": 6.48970890045166, "learning_rate": 6.0181683744913055e-06, "loss": 0.3626, "step": 8706 }, { "epoch": 0.4404370479032829, "grad_norm": 7.92877197265625, "learning_rate": 6.01738256419189e-06, "loss": 0.4025, "step": 8707 }, { "epoch": 0.44048763215134806, "grad_norm": 6.723144054412842, "learning_rate": 6.016596727676982e-06, "loss": 0.4397, "step": 8708 }, { "epoch": 0.4405382163994132, "grad_norm": 10.858002662658691, "learning_rate": 6.0158108649668336e-06, "loss": 0.4638, "step": 8709 }, { "epoch": 0.44058880064747835, "grad_norm": 5.597957134246826, "learning_rate": 6.015024976081691e-06, "loss": 0.348, "step": 8710 }, { "epoch": 0.44063938489554355, "grad_norm": 5.47361421585083, "learning_rate": 6.014239061041806e-06, "loss": 0.4534, "step": 8711 }, { "epoch": 0.4406899691436087, "grad_norm": 7.336158752441406, "learning_rate": 6.013453119867433e-06, "loss": 0.4455, "step": 8712 }, { "epoch": 0.44074055339167384, "grad_norm": 6.222170829772949, "learning_rate": 6.0126671525788194e-06, "loss": 0.4152, "step": 8713 }, { "epoch": 0.440791137639739, "grad_norm": 10.17408561706543, "learning_rate": 6.011881159196219e-06, "loss": 0.4262, "step": 8714 }, { "epoch": 0.4408417218878041, "grad_norm": 8.579741477966309, "learning_rate": 6.011095139739885e-06, "loss": 0.4314, "step": 8715 }, { "epoch": 0.44089230613586927, "grad_norm": 11.08627700805664, "learning_rate": 6.010309094230072e-06, "loss": 0.5089, "step": 8716 }, { "epoch": 0.4409428903839344, "grad_norm": 11.260980606079102, "learning_rate": 6.009523022687034e-06, "loss": 0.4368, "step": 8717 }, { "epoch": 0.4409934746319996, "grad_norm": 10.626901626586914, "learning_rate": 6.008736925131026e-06, "loss": 0.4874, "step": 8718 }, { "epoch": 0.44104405888006476, "grad_norm": 7.219857692718506, "learning_rate": 6.007950801582304e-06, "loss": 0.5107, "step": 8719 }, { "epoch": 0.4410946431281299, "grad_norm": 20.529481887817383, "learning_rate": 6.007164652061125e-06, "loss": 0.4958, "step": 8720 }, { "epoch": 0.44114522737619505, "grad_norm": 5.371377944946289, "learning_rate": 6.006378476587744e-06, "loss": 0.4103, "step": 8721 }, { "epoch": 0.4411958116242602, "grad_norm": 5.785487174987793, "learning_rate": 6.005592275182423e-06, "loss": 0.4746, "step": 8722 }, { "epoch": 0.44124639587232534, "grad_norm": 4.838676929473877, "learning_rate": 6.004806047865415e-06, "loss": 0.4409, "step": 8723 }, { "epoch": 0.44129698012039054, "grad_norm": 11.754125595092773, "learning_rate": 6.004019794656983e-06, "loss": 0.5369, "step": 8724 }, { "epoch": 0.4413475643684557, "grad_norm": 7.733726978302002, "learning_rate": 6.003233515577388e-06, "loss": 0.5225, "step": 8725 }, { "epoch": 0.4413981486165208, "grad_norm": 7.698141098022461, "learning_rate": 6.002447210646888e-06, "loss": 0.3971, "step": 8726 }, { "epoch": 0.44144873286458597, "grad_norm": 8.754538536071777, "learning_rate": 6.001660879885744e-06, "loss": 0.4886, "step": 8727 }, { "epoch": 0.4414993171126511, "grad_norm": 8.000788688659668, "learning_rate": 6.000874523314218e-06, "loss": 0.3401, "step": 8728 }, { "epoch": 0.44154990136071626, "grad_norm": 7.715184211730957, "learning_rate": 6.000088140952573e-06, "loss": 0.4882, "step": 8729 }, { "epoch": 0.4416004856087814, "grad_norm": 9.95893383026123, "learning_rate": 5.9993017328210725e-06, "loss": 0.42, "step": 8730 }, { "epoch": 0.4416510698568466, "grad_norm": 5.901263236999512, "learning_rate": 5.998515298939981e-06, "loss": 0.4283, "step": 8731 }, { "epoch": 0.44170165410491175, "grad_norm": 6.082009792327881, "learning_rate": 5.99772883932956e-06, "loss": 0.4219, "step": 8732 }, { "epoch": 0.4417522383529769, "grad_norm": 8.943668365478516, "learning_rate": 5.996942354010079e-06, "loss": 0.4807, "step": 8733 }, { "epoch": 0.44180282260104203, "grad_norm": 6.395772457122803, "learning_rate": 5.996155843001799e-06, "loss": 0.443, "step": 8734 }, { "epoch": 0.4418534068491072, "grad_norm": 6.465811252593994, "learning_rate": 5.995369306324992e-06, "loss": 0.432, "step": 8735 }, { "epoch": 0.4419039910971723, "grad_norm": 10.348041534423828, "learning_rate": 5.994582743999921e-06, "loss": 0.5995, "step": 8736 }, { "epoch": 0.44195457534523747, "grad_norm": 5.598814487457275, "learning_rate": 5.993796156046854e-06, "loss": 0.4548, "step": 8737 }, { "epoch": 0.44200515959330267, "grad_norm": 10.991482734680176, "learning_rate": 5.9930095424860615e-06, "loss": 0.46, "step": 8738 }, { "epoch": 0.4420557438413678, "grad_norm": 6.922391414642334, "learning_rate": 5.992222903337812e-06, "loss": 0.4732, "step": 8739 }, { "epoch": 0.44210632808943295, "grad_norm": 10.517534255981445, "learning_rate": 5.991436238622374e-06, "loss": 0.5271, "step": 8740 }, { "epoch": 0.4421569123374981, "grad_norm": 10.84865951538086, "learning_rate": 5.990649548360019e-06, "loss": 0.3924, "step": 8741 }, { "epoch": 0.44220749658556324, "grad_norm": 5.150989532470703, "learning_rate": 5.989862832571019e-06, "loss": 0.3356, "step": 8742 }, { "epoch": 0.4422580808336284, "grad_norm": 10.304095268249512, "learning_rate": 5.989076091275642e-06, "loss": 0.5433, "step": 8743 }, { "epoch": 0.4423086650816936, "grad_norm": 11.81500244140625, "learning_rate": 5.9882893244941655e-06, "loss": 0.4949, "step": 8744 }, { "epoch": 0.44235924932975873, "grad_norm": 7.1520209312438965, "learning_rate": 5.987502532246859e-06, "loss": 0.3287, "step": 8745 }, { "epoch": 0.4424098335778239, "grad_norm": 7.424635887145996, "learning_rate": 5.986715714553999e-06, "loss": 0.4411, "step": 8746 }, { "epoch": 0.442460417825889, "grad_norm": 7.105843544006348, "learning_rate": 5.985928871435858e-06, "loss": 0.425, "step": 8747 }, { "epoch": 0.44251100207395416, "grad_norm": 8.14533519744873, "learning_rate": 5.98514200291271e-06, "loss": 0.443, "step": 8748 }, { "epoch": 0.4425615863220193, "grad_norm": 7.725929260253906, "learning_rate": 5.984355109004833e-06, "loss": 0.5034, "step": 8749 }, { "epoch": 0.44261217057008445, "grad_norm": 8.889961242675781, "learning_rate": 5.9835681897325025e-06, "loss": 0.3777, "step": 8750 }, { "epoch": 0.44266275481814965, "grad_norm": 6.250960826873779, "learning_rate": 5.982781245115995e-06, "loss": 0.4516, "step": 8751 }, { "epoch": 0.4427133390662148, "grad_norm": 9.43881893157959, "learning_rate": 5.98199427517559e-06, "loss": 0.4397, "step": 8752 }, { "epoch": 0.44276392331427994, "grad_norm": 7.8747639656066895, "learning_rate": 5.981207279931562e-06, "loss": 0.4323, "step": 8753 }, { "epoch": 0.4428145075623451, "grad_norm": 10.455653190612793, "learning_rate": 5.980420259404195e-06, "loss": 0.4161, "step": 8754 }, { "epoch": 0.4428650918104102, "grad_norm": 11.459546089172363, "learning_rate": 5.979633213613765e-06, "loss": 0.4788, "step": 8755 }, { "epoch": 0.44291567605847537, "grad_norm": 8.785660743713379, "learning_rate": 5.978846142580552e-06, "loss": 0.3491, "step": 8756 }, { "epoch": 0.4429662603065405, "grad_norm": 6.808652400970459, "learning_rate": 5.97805904632484e-06, "loss": 0.4103, "step": 8757 }, { "epoch": 0.4430168445546057, "grad_norm": 14.863933563232422, "learning_rate": 5.977271924866908e-06, "loss": 0.4381, "step": 8758 }, { "epoch": 0.44306742880267086, "grad_norm": 12.995134353637695, "learning_rate": 5.9764847782270395e-06, "loss": 0.5064, "step": 8759 }, { "epoch": 0.443118013050736, "grad_norm": 8.25352954864502, "learning_rate": 5.9756976064255165e-06, "loss": 0.516, "step": 8760 }, { "epoch": 0.44316859729880115, "grad_norm": 12.496833801269531, "learning_rate": 5.9749104094826226e-06, "loss": 0.455, "step": 8761 }, { "epoch": 0.4432191815468663, "grad_norm": 8.968677520751953, "learning_rate": 5.974123187418642e-06, "loss": 0.44, "step": 8762 }, { "epoch": 0.44326976579493144, "grad_norm": 7.041082382202148, "learning_rate": 5.973335940253861e-06, "loss": 0.4231, "step": 8763 }, { "epoch": 0.44332035004299664, "grad_norm": 7.334842681884766, "learning_rate": 5.9725486680085634e-06, "loss": 0.4471, "step": 8764 }, { "epoch": 0.4433709342910618, "grad_norm": 9.718924522399902, "learning_rate": 5.971761370703035e-06, "loss": 0.4249, "step": 8765 }, { "epoch": 0.4434215185391269, "grad_norm": 3.9130396842956543, "learning_rate": 5.970974048357565e-06, "loss": 0.3134, "step": 8766 }, { "epoch": 0.44347210278719207, "grad_norm": 4.850831508636475, "learning_rate": 5.970186700992438e-06, "loss": 0.4182, "step": 8767 }, { "epoch": 0.4435226870352572, "grad_norm": 15.915380477905273, "learning_rate": 5.969399328627942e-06, "loss": 0.4956, "step": 8768 }, { "epoch": 0.44357327128332236, "grad_norm": 8.197003364562988, "learning_rate": 5.968611931284367e-06, "loss": 0.5171, "step": 8769 }, { "epoch": 0.4436238555313875, "grad_norm": 5.902149677276611, "learning_rate": 5.967824508982003e-06, "loss": 0.374, "step": 8770 }, { "epoch": 0.4436744397794527, "grad_norm": 7.846671104431152, "learning_rate": 5.967037061741139e-06, "loss": 0.3072, "step": 8771 }, { "epoch": 0.44372502402751784, "grad_norm": 9.089445114135742, "learning_rate": 5.966249589582065e-06, "loss": 0.3525, "step": 8772 }, { "epoch": 0.443775608275583, "grad_norm": 5.701415538787842, "learning_rate": 5.965462092525074e-06, "loss": 0.3607, "step": 8773 }, { "epoch": 0.44382619252364813, "grad_norm": 7.816079616546631, "learning_rate": 5.964674570590456e-06, "loss": 0.4753, "step": 8774 }, { "epoch": 0.4438767767717133, "grad_norm": 7.415027618408203, "learning_rate": 5.963887023798505e-06, "loss": 0.4474, "step": 8775 }, { "epoch": 0.4439273610197784, "grad_norm": 6.695859432220459, "learning_rate": 5.963099452169514e-06, "loss": 0.457, "step": 8776 }, { "epoch": 0.4439779452678436, "grad_norm": 7.275197982788086, "learning_rate": 5.962311855723774e-06, "loss": 0.3935, "step": 8777 }, { "epoch": 0.44402852951590877, "grad_norm": 6.704047203063965, "learning_rate": 5.961524234481585e-06, "loss": 0.4257, "step": 8778 }, { "epoch": 0.4440791137639739, "grad_norm": 7.565069675445557, "learning_rate": 5.960736588463239e-06, "loss": 0.4994, "step": 8779 }, { "epoch": 0.44412969801203905, "grad_norm": 5.94016170501709, "learning_rate": 5.959948917689029e-06, "loss": 0.4228, "step": 8780 }, { "epoch": 0.4441802822601042, "grad_norm": 4.622339725494385, "learning_rate": 5.959161222179255e-06, "loss": 0.3825, "step": 8781 }, { "epoch": 0.44423086650816934, "grad_norm": 5.775205135345459, "learning_rate": 5.958373501954213e-06, "loss": 0.4628, "step": 8782 }, { "epoch": 0.4442814507562345, "grad_norm": 7.3230977058410645, "learning_rate": 5.957585757034201e-06, "loss": 0.3793, "step": 8783 }, { "epoch": 0.4443320350042997, "grad_norm": 18.079082489013672, "learning_rate": 5.956797987439518e-06, "loss": 0.4261, "step": 8784 }, { "epoch": 0.44438261925236483, "grad_norm": 5.319275856018066, "learning_rate": 5.9560101931904615e-06, "loss": 0.3298, "step": 8785 }, { "epoch": 0.44443320350043, "grad_norm": 8.175418853759766, "learning_rate": 5.95522237430733e-06, "loss": 0.3778, "step": 8786 }, { "epoch": 0.4444837877484951, "grad_norm": 6.636544704437256, "learning_rate": 5.954434530810426e-06, "loss": 0.404, "step": 8787 }, { "epoch": 0.44453437199656026, "grad_norm": 6.687603950500488, "learning_rate": 5.953646662720049e-06, "loss": 0.4129, "step": 8788 }, { "epoch": 0.4445849562446254, "grad_norm": 6.18648099899292, "learning_rate": 5.9528587700565025e-06, "loss": 0.4462, "step": 8789 }, { "epoch": 0.44463554049269055, "grad_norm": 8.682666778564453, "learning_rate": 5.952070852840085e-06, "loss": 0.4984, "step": 8790 }, { "epoch": 0.44468612474075575, "grad_norm": 6.53385066986084, "learning_rate": 5.951282911091103e-06, "loss": 0.4078, "step": 8791 }, { "epoch": 0.4447367089888209, "grad_norm": 10.768974304199219, "learning_rate": 5.950494944829858e-06, "loss": 0.5708, "step": 8792 }, { "epoch": 0.44478729323688604, "grad_norm": 5.850058555603027, "learning_rate": 5.949706954076653e-06, "loss": 0.4357, "step": 8793 }, { "epoch": 0.4448378774849512, "grad_norm": 10.34965705871582, "learning_rate": 5.948918938851794e-06, "loss": 0.4711, "step": 8794 }, { "epoch": 0.4448884617330163, "grad_norm": 9.858915328979492, "learning_rate": 5.948130899175586e-06, "loss": 0.5174, "step": 8795 }, { "epoch": 0.44493904598108147, "grad_norm": 5.27997350692749, "learning_rate": 5.947342835068335e-06, "loss": 0.403, "step": 8796 }, { "epoch": 0.44498963022914667, "grad_norm": 9.514239311218262, "learning_rate": 5.9465547465503486e-06, "loss": 0.5575, "step": 8797 }, { "epoch": 0.4450402144772118, "grad_norm": 8.232925415039062, "learning_rate": 5.945766633641931e-06, "loss": 0.3711, "step": 8798 }, { "epoch": 0.44509079872527696, "grad_norm": 9.291540145874023, "learning_rate": 5.944978496363391e-06, "loss": 0.5313, "step": 8799 }, { "epoch": 0.4451413829733421, "grad_norm": 14.081559181213379, "learning_rate": 5.944190334735038e-06, "loss": 0.5047, "step": 8800 }, { "epoch": 0.44519196722140725, "grad_norm": 8.087687492370605, "learning_rate": 5.9434021487771835e-06, "loss": 0.5283, "step": 8801 }, { "epoch": 0.4452425514694724, "grad_norm": 7.370234489440918, "learning_rate": 5.9426139385101315e-06, "loss": 0.4525, "step": 8802 }, { "epoch": 0.44529313571753754, "grad_norm": 10.313702583312988, "learning_rate": 5.941825703954196e-06, "loss": 0.4637, "step": 8803 }, { "epoch": 0.44534371996560274, "grad_norm": 14.091104507446289, "learning_rate": 5.941037445129687e-06, "loss": 0.4811, "step": 8804 }, { "epoch": 0.4453943042136679, "grad_norm": 8.135910987854004, "learning_rate": 5.940249162056916e-06, "loss": 0.4339, "step": 8805 }, { "epoch": 0.445444888461733, "grad_norm": 9.925566673278809, "learning_rate": 5.939460854756194e-06, "loss": 0.4411, "step": 8806 }, { "epoch": 0.44549547270979817, "grad_norm": 8.563852310180664, "learning_rate": 5.938672523247837e-06, "loss": 0.4141, "step": 8807 }, { "epoch": 0.4455460569578633, "grad_norm": 5.326432704925537, "learning_rate": 5.937884167552155e-06, "loss": 0.398, "step": 8808 }, { "epoch": 0.44559664120592846, "grad_norm": 8.656161308288574, "learning_rate": 5.937095787689463e-06, "loss": 0.3846, "step": 8809 }, { "epoch": 0.4456472254539936, "grad_norm": 10.065980911254883, "learning_rate": 5.936307383680078e-06, "loss": 0.4927, "step": 8810 }, { "epoch": 0.4456978097020588, "grad_norm": 10.149565696716309, "learning_rate": 5.9355189555443114e-06, "loss": 0.3685, "step": 8811 }, { "epoch": 0.44574839395012394, "grad_norm": 8.949780464172363, "learning_rate": 5.934730503302482e-06, "loss": 0.4796, "step": 8812 }, { "epoch": 0.4457989781981891, "grad_norm": 8.499448776245117, "learning_rate": 5.933942026974905e-06, "loss": 0.3451, "step": 8813 }, { "epoch": 0.44584956244625423, "grad_norm": 11.288535118103027, "learning_rate": 5.933153526581898e-06, "loss": 0.4248, "step": 8814 }, { "epoch": 0.4459001466943194, "grad_norm": 7.064999580383301, "learning_rate": 5.932365002143777e-06, "loss": 0.411, "step": 8815 }, { "epoch": 0.4459507309423845, "grad_norm": 5.94965934753418, "learning_rate": 5.931576453680864e-06, "loss": 0.3187, "step": 8816 }, { "epoch": 0.4460013151904497, "grad_norm": 6.240894317626953, "learning_rate": 5.930787881213476e-06, "loss": 0.3958, "step": 8817 }, { "epoch": 0.44605189943851487, "grad_norm": 17.545122146606445, "learning_rate": 5.92999928476193e-06, "loss": 0.5188, "step": 8818 }, { "epoch": 0.44610248368658, "grad_norm": 9.395859718322754, "learning_rate": 5.92921066434655e-06, "loss": 0.4008, "step": 8819 }, { "epoch": 0.44615306793464515, "grad_norm": 6.883078098297119, "learning_rate": 5.928422019987657e-06, "loss": 0.3798, "step": 8820 }, { "epoch": 0.4462036521827103, "grad_norm": 9.908941268920898, "learning_rate": 5.927633351705569e-06, "loss": 0.5309, "step": 8821 }, { "epoch": 0.44625423643077544, "grad_norm": 5.4954514503479, "learning_rate": 5.926844659520611e-06, "loss": 0.4008, "step": 8822 }, { "epoch": 0.4463048206788406, "grad_norm": 8.05820083618164, "learning_rate": 5.9260559434531066e-06, "loss": 0.5972, "step": 8823 }, { "epoch": 0.4463554049269058, "grad_norm": 4.400488376617432, "learning_rate": 5.925267203523375e-06, "loss": 0.3312, "step": 8824 }, { "epoch": 0.44640598917497093, "grad_norm": 11.326662063598633, "learning_rate": 5.924478439751743e-06, "loss": 0.4045, "step": 8825 }, { "epoch": 0.4464565734230361, "grad_norm": 6.732610702514648, "learning_rate": 5.923689652158535e-06, "loss": 0.4985, "step": 8826 }, { "epoch": 0.4465071576711012, "grad_norm": 6.200906276702881, "learning_rate": 5.9229008407640755e-06, "loss": 0.4843, "step": 8827 }, { "epoch": 0.44655774191916636, "grad_norm": 7.511178493499756, "learning_rate": 5.9221120055886894e-06, "loss": 0.3095, "step": 8828 }, { "epoch": 0.4466083261672315, "grad_norm": 8.53482437133789, "learning_rate": 5.921323146652707e-06, "loss": 0.3726, "step": 8829 }, { "epoch": 0.44665891041529665, "grad_norm": 6.696826934814453, "learning_rate": 5.920534263976451e-06, "loss": 0.4084, "step": 8830 }, { "epoch": 0.44670949466336185, "grad_norm": 6.0539093017578125, "learning_rate": 5.9197453575802484e-06, "loss": 0.3675, "step": 8831 }, { "epoch": 0.446760078911427, "grad_norm": 14.344647407531738, "learning_rate": 5.9189564274844326e-06, "loss": 0.5655, "step": 8832 }, { "epoch": 0.44681066315949214, "grad_norm": 17.127910614013672, "learning_rate": 5.918167473709328e-06, "loss": 0.4758, "step": 8833 }, { "epoch": 0.4468612474075573, "grad_norm": 5.327781677246094, "learning_rate": 5.917378496275265e-06, "loss": 0.409, "step": 8834 }, { "epoch": 0.4469118316556224, "grad_norm": 13.983959197998047, "learning_rate": 5.916589495202575e-06, "loss": 0.4034, "step": 8835 }, { "epoch": 0.44696241590368757, "grad_norm": 8.107378005981445, "learning_rate": 5.915800470511587e-06, "loss": 0.466, "step": 8836 }, { "epoch": 0.44701300015175277, "grad_norm": 7.24997615814209, "learning_rate": 5.915011422222633e-06, "loss": 0.3084, "step": 8837 }, { "epoch": 0.4470635843998179, "grad_norm": 8.299615859985352, "learning_rate": 5.9142223503560436e-06, "loss": 0.4289, "step": 8838 }, { "epoch": 0.44711416864788306, "grad_norm": 6.523895263671875, "learning_rate": 5.9134332549321535e-06, "loss": 0.4295, "step": 8839 }, { "epoch": 0.4471647528959482, "grad_norm": 8.246367454528809, "learning_rate": 5.912644135971295e-06, "loss": 0.4453, "step": 8840 }, { "epoch": 0.44721533714401335, "grad_norm": 7.44866418838501, "learning_rate": 5.9118549934938005e-06, "loss": 0.3851, "step": 8841 }, { "epoch": 0.4472659213920785, "grad_norm": 8.600111961364746, "learning_rate": 5.911065827520009e-06, "loss": 0.3787, "step": 8842 }, { "epoch": 0.44731650564014364, "grad_norm": 6.607243537902832, "learning_rate": 5.910276638070247e-06, "loss": 0.426, "step": 8843 }, { "epoch": 0.44736708988820884, "grad_norm": 9.56392765045166, "learning_rate": 5.909487425164856e-06, "loss": 0.385, "step": 8844 }, { "epoch": 0.447417674136274, "grad_norm": 8.938146591186523, "learning_rate": 5.908698188824172e-06, "loss": 0.3947, "step": 8845 }, { "epoch": 0.4474682583843391, "grad_norm": 6.174048900604248, "learning_rate": 5.9079089290685295e-06, "loss": 0.381, "step": 8846 }, { "epoch": 0.44751884263240427, "grad_norm": 8.105318069458008, "learning_rate": 5.9071196459182665e-06, "loss": 0.3879, "step": 8847 }, { "epoch": 0.4475694268804694, "grad_norm": 9.528095245361328, "learning_rate": 5.9063303393937235e-06, "loss": 0.5042, "step": 8848 }, { "epoch": 0.44762001112853456, "grad_norm": 7.6537628173828125, "learning_rate": 5.905541009515235e-06, "loss": 0.43, "step": 8849 }, { "epoch": 0.4476705953765997, "grad_norm": 8.188034057617188, "learning_rate": 5.904751656303143e-06, "loss": 0.4796, "step": 8850 }, { "epoch": 0.4477211796246649, "grad_norm": 8.796880722045898, "learning_rate": 5.903962279777785e-06, "loss": 0.4498, "step": 8851 }, { "epoch": 0.44777176387273004, "grad_norm": 8.448199272155762, "learning_rate": 5.903172879959502e-06, "loss": 0.4223, "step": 8852 }, { "epoch": 0.4478223481207952, "grad_norm": 13.298307418823242, "learning_rate": 5.902383456868637e-06, "loss": 0.4698, "step": 8853 }, { "epoch": 0.44787293236886033, "grad_norm": 7.75878381729126, "learning_rate": 5.9015940105255285e-06, "loss": 0.3943, "step": 8854 }, { "epoch": 0.4479235166169255, "grad_norm": 7.635089874267578, "learning_rate": 5.900804540950522e-06, "loss": 0.4438, "step": 8855 }, { "epoch": 0.4479741008649906, "grad_norm": 6.747097492218018, "learning_rate": 5.9000150481639566e-06, "loss": 0.4612, "step": 8856 }, { "epoch": 0.4480246851130558, "grad_norm": 6.391672611236572, "learning_rate": 5.899225532186177e-06, "loss": 0.3404, "step": 8857 }, { "epoch": 0.44807526936112096, "grad_norm": 10.383952140808105, "learning_rate": 5.898435993037529e-06, "loss": 0.3209, "step": 8858 }, { "epoch": 0.4481258536091861, "grad_norm": 13.512739181518555, "learning_rate": 5.897646430738355e-06, "loss": 0.4003, "step": 8859 }, { "epoch": 0.44817643785725125, "grad_norm": 12.93770980834961, "learning_rate": 5.896856845308999e-06, "loss": 0.4658, "step": 8860 }, { "epoch": 0.4482270221053164, "grad_norm": 19.852558135986328, "learning_rate": 5.896067236769813e-06, "loss": 0.4646, "step": 8861 }, { "epoch": 0.44827760635338154, "grad_norm": 8.34571361541748, "learning_rate": 5.895277605141134e-06, "loss": 0.4071, "step": 8862 }, { "epoch": 0.4483281906014467, "grad_norm": 8.318041801452637, "learning_rate": 5.894487950443315e-06, "loss": 0.4345, "step": 8863 }, { "epoch": 0.4483787748495119, "grad_norm": 12.8035888671875, "learning_rate": 5.893698272696702e-06, "loss": 0.3958, "step": 8864 }, { "epoch": 0.44842935909757703, "grad_norm": 8.603975296020508, "learning_rate": 5.892908571921643e-06, "loss": 0.3901, "step": 8865 }, { "epoch": 0.4484799433456422, "grad_norm": 19.460405349731445, "learning_rate": 5.892118848138486e-06, "loss": 0.5014, "step": 8866 }, { "epoch": 0.4485305275937073, "grad_norm": 27.566850662231445, "learning_rate": 5.891329101367583e-06, "loss": 0.4712, "step": 8867 }, { "epoch": 0.44858111184177246, "grad_norm": 8.66598129272461, "learning_rate": 5.890539331629282e-06, "loss": 0.5336, "step": 8868 }, { "epoch": 0.4486316960898376, "grad_norm": 6.404578685760498, "learning_rate": 5.889749538943932e-06, "loss": 0.2859, "step": 8869 }, { "epoch": 0.44868228033790275, "grad_norm": 6.471699237823486, "learning_rate": 5.8889597233318865e-06, "loss": 0.2691, "step": 8870 }, { "epoch": 0.44873286458596795, "grad_norm": 7.018553733825684, "learning_rate": 5.888169884813495e-06, "loss": 0.339, "step": 8871 }, { "epoch": 0.4487834488340331, "grad_norm": 9.059671401977539, "learning_rate": 5.887380023409112e-06, "loss": 0.4853, "step": 8872 }, { "epoch": 0.44883403308209824, "grad_norm": 13.151741981506348, "learning_rate": 5.886590139139089e-06, "loss": 0.5624, "step": 8873 }, { "epoch": 0.4488846173301634, "grad_norm": 14.990867614746094, "learning_rate": 5.885800232023779e-06, "loss": 0.4755, "step": 8874 }, { "epoch": 0.4489352015782285, "grad_norm": 6.799164295196533, "learning_rate": 5.885010302083537e-06, "loss": 0.4894, "step": 8875 }, { "epoch": 0.44898578582629367, "grad_norm": 8.0391845703125, "learning_rate": 5.884220349338716e-06, "loss": 0.371, "step": 8876 }, { "epoch": 0.44903637007435887, "grad_norm": 12.790291786193848, "learning_rate": 5.883430373809676e-06, "loss": 0.4474, "step": 8877 }, { "epoch": 0.449086954322424, "grad_norm": 6.675958156585693, "learning_rate": 5.882640375516765e-06, "loss": 0.4215, "step": 8878 }, { "epoch": 0.44913753857048916, "grad_norm": 5.589139461517334, "learning_rate": 5.881850354480344e-06, "loss": 0.3784, "step": 8879 }, { "epoch": 0.4491881228185543, "grad_norm": 8.304473876953125, "learning_rate": 5.8810603107207705e-06, "loss": 0.5245, "step": 8880 }, { "epoch": 0.44923870706661945, "grad_norm": 4.866262435913086, "learning_rate": 5.880270244258402e-06, "loss": 0.4078, "step": 8881 }, { "epoch": 0.4492892913146846, "grad_norm": 5.922057151794434, "learning_rate": 5.879480155113594e-06, "loss": 0.3943, "step": 8882 }, { "epoch": 0.44933987556274974, "grad_norm": 8.715753555297852, "learning_rate": 5.878690043306707e-06, "loss": 0.4213, "step": 8883 }, { "epoch": 0.44939045981081494, "grad_norm": 17.559906005859375, "learning_rate": 5.8778999088581005e-06, "loss": 0.6216, "step": 8884 }, { "epoch": 0.4494410440588801, "grad_norm": 8.667220115661621, "learning_rate": 5.877109751788132e-06, "loss": 0.4969, "step": 8885 }, { "epoch": 0.4494916283069452, "grad_norm": 7.295815467834473, "learning_rate": 5.876319572117166e-06, "loss": 0.3614, "step": 8886 }, { "epoch": 0.44954221255501037, "grad_norm": 9.871065139770508, "learning_rate": 5.87552936986556e-06, "loss": 0.3937, "step": 8887 }, { "epoch": 0.4495927968030755, "grad_norm": 7.445888042449951, "learning_rate": 5.874739145053677e-06, "loss": 0.3607, "step": 8888 }, { "epoch": 0.44964338105114066, "grad_norm": 9.432626724243164, "learning_rate": 5.87394889770188e-06, "loss": 0.3606, "step": 8889 }, { "epoch": 0.4496939652992058, "grad_norm": 4.789259910583496, "learning_rate": 5.87315862783053e-06, "loss": 0.3558, "step": 8890 }, { "epoch": 0.449744549547271, "grad_norm": 7.692309856414795, "learning_rate": 5.872368335459991e-06, "loss": 0.4067, "step": 8891 }, { "epoch": 0.44979513379533614, "grad_norm": 8.832622528076172, "learning_rate": 5.871578020610628e-06, "loss": 0.4206, "step": 8892 }, { "epoch": 0.4498457180434013, "grad_norm": 6.839741230010986, "learning_rate": 5.870787683302803e-06, "loss": 0.2537, "step": 8893 }, { "epoch": 0.44989630229146643, "grad_norm": 6.617712020874023, "learning_rate": 5.8699973235568855e-06, "loss": 0.437, "step": 8894 }, { "epoch": 0.4499468865395316, "grad_norm": 13.366531372070312, "learning_rate": 5.869206941393235e-06, "loss": 0.6522, "step": 8895 }, { "epoch": 0.4499974707875967, "grad_norm": 9.8009614944458, "learning_rate": 5.8684165368322245e-06, "loss": 0.4578, "step": 8896 }, { "epoch": 0.4500480550356619, "grad_norm": 10.605108261108398, "learning_rate": 5.8676261098942145e-06, "loss": 0.4007, "step": 8897 }, { "epoch": 0.45009863928372706, "grad_norm": 10.00155258178711, "learning_rate": 5.866835660599577e-06, "loss": 0.3606, "step": 8898 }, { "epoch": 0.4501492235317922, "grad_norm": 6.444452285766602, "learning_rate": 5.866045188968677e-06, "loss": 0.3888, "step": 8899 }, { "epoch": 0.45019980777985735, "grad_norm": 6.287414073944092, "learning_rate": 5.865254695021886e-06, "loss": 0.4689, "step": 8900 }, { "epoch": 0.4502503920279225, "grad_norm": 11.15884780883789, "learning_rate": 5.86446417877957e-06, "loss": 0.352, "step": 8901 }, { "epoch": 0.45030097627598764, "grad_norm": 17.078248977661133, "learning_rate": 5.863673640262102e-06, "loss": 0.6487, "step": 8902 }, { "epoch": 0.4503515605240528, "grad_norm": 8.477835655212402, "learning_rate": 5.862883079489848e-06, "loss": 0.4964, "step": 8903 }, { "epoch": 0.450402144772118, "grad_norm": 10.3068208694458, "learning_rate": 5.8620924964831815e-06, "loss": 0.4736, "step": 8904 }, { "epoch": 0.45045272902018313, "grad_norm": 10.461820602416992, "learning_rate": 5.8613018912624755e-06, "loss": 0.5045, "step": 8905 }, { "epoch": 0.4505033132682483, "grad_norm": 8.405814170837402, "learning_rate": 5.8605112638481e-06, "loss": 0.4052, "step": 8906 }, { "epoch": 0.4505538975163134, "grad_norm": 7.641960144042969, "learning_rate": 5.859720614260426e-06, "loss": 0.3258, "step": 8907 }, { "epoch": 0.45060448176437856, "grad_norm": 18.68429946899414, "learning_rate": 5.8589299425198295e-06, "loss": 0.4999, "step": 8908 }, { "epoch": 0.4506550660124437, "grad_norm": 15.2660493850708, "learning_rate": 5.8581392486466825e-06, "loss": 0.63, "step": 8909 }, { "epoch": 0.45070565026050885, "grad_norm": 9.025688171386719, "learning_rate": 5.85734853266136e-06, "loss": 0.4552, "step": 8910 }, { "epoch": 0.45075623450857405, "grad_norm": 7.6797051429748535, "learning_rate": 5.856557794584237e-06, "loss": 0.4731, "step": 8911 }, { "epoch": 0.4508068187566392, "grad_norm": 4.253947734832764, "learning_rate": 5.855767034435688e-06, "loss": 0.2818, "step": 8912 }, { "epoch": 0.45085740300470434, "grad_norm": 7.255660057067871, "learning_rate": 5.85497625223609e-06, "loss": 0.4479, "step": 8913 }, { "epoch": 0.4509079872527695, "grad_norm": 5.278113842010498, "learning_rate": 5.854185448005818e-06, "loss": 0.3414, "step": 8914 }, { "epoch": 0.4509585715008346, "grad_norm": 14.608171463012695, "learning_rate": 5.853394621765252e-06, "loss": 0.5837, "step": 8915 }, { "epoch": 0.45100915574889977, "grad_norm": 11.789657592773438, "learning_rate": 5.852603773534766e-06, "loss": 0.451, "step": 8916 }, { "epoch": 0.45105973999696497, "grad_norm": 6.055042266845703, "learning_rate": 5.851812903334743e-06, "loss": 0.3667, "step": 8917 }, { "epoch": 0.4511103242450301, "grad_norm": 5.730833530426025, "learning_rate": 5.851022011185557e-06, "loss": 0.4231, "step": 8918 }, { "epoch": 0.45116090849309526, "grad_norm": 6.196452617645264, "learning_rate": 5.850231097107589e-06, "loss": 0.4158, "step": 8919 }, { "epoch": 0.4512114927411604, "grad_norm": 8.066225051879883, "learning_rate": 5.8494401611212195e-06, "loss": 0.4676, "step": 8920 }, { "epoch": 0.45126207698922555, "grad_norm": 15.126927375793457, "learning_rate": 5.848649203246829e-06, "loss": 0.4895, "step": 8921 }, { "epoch": 0.4513126612372907, "grad_norm": 8.542800903320312, "learning_rate": 5.847858223504799e-06, "loss": 0.5052, "step": 8922 }, { "epoch": 0.45136324548535584, "grad_norm": 9.37102222442627, "learning_rate": 5.847067221915509e-06, "loss": 0.4354, "step": 8923 }, { "epoch": 0.45141382973342103, "grad_norm": 14.206452369689941, "learning_rate": 5.846276198499343e-06, "loss": 0.4162, "step": 8924 }, { "epoch": 0.4514644139814862, "grad_norm": 10.87554931640625, "learning_rate": 5.845485153276684e-06, "loss": 0.4143, "step": 8925 }, { "epoch": 0.4515149982295513, "grad_norm": 7.4068450927734375, "learning_rate": 5.844694086267916e-06, "loss": 0.4639, "step": 8926 }, { "epoch": 0.45156558247761647, "grad_norm": 5.381242752075195, "learning_rate": 5.84390299749342e-06, "loss": 0.3858, "step": 8927 }, { "epoch": 0.4516161667256816, "grad_norm": 10.225685119628906, "learning_rate": 5.843111886973582e-06, "loss": 0.504, "step": 8928 }, { "epoch": 0.45166675097374676, "grad_norm": 6.820442199707031, "learning_rate": 5.842320754728788e-06, "loss": 0.5326, "step": 8929 }, { "epoch": 0.45171733522181196, "grad_norm": 6.894443988800049, "learning_rate": 5.841529600779423e-06, "loss": 0.4619, "step": 8930 }, { "epoch": 0.4517679194698771, "grad_norm": 12.159321784973145, "learning_rate": 5.840738425145872e-06, "loss": 0.3789, "step": 8931 }, { "epoch": 0.45181850371794224, "grad_norm": 13.857996940612793, "learning_rate": 5.839947227848522e-06, "loss": 0.4625, "step": 8932 }, { "epoch": 0.4518690879660074, "grad_norm": 5.944266319274902, "learning_rate": 5.839156008907761e-06, "loss": 0.3275, "step": 8933 }, { "epoch": 0.45191967221407253, "grad_norm": 6.0716328620910645, "learning_rate": 5.838364768343978e-06, "loss": 0.5021, "step": 8934 }, { "epoch": 0.4519702564621377, "grad_norm": 5.281708240509033, "learning_rate": 5.8375735061775585e-06, "loss": 0.3959, "step": 8935 }, { "epoch": 0.4520208407102028, "grad_norm": 5.842961311340332, "learning_rate": 5.836782222428894e-06, "loss": 0.4042, "step": 8936 }, { "epoch": 0.452071424958268, "grad_norm": 7.251824378967285, "learning_rate": 5.835990917118372e-06, "loss": 0.4692, "step": 8937 }, { "epoch": 0.45212200920633316, "grad_norm": 7.365726470947266, "learning_rate": 5.835199590266383e-06, "loss": 0.4293, "step": 8938 }, { "epoch": 0.4521725934543983, "grad_norm": 8.88910961151123, "learning_rate": 5.834408241893319e-06, "loss": 0.3912, "step": 8939 }, { "epoch": 0.45222317770246345, "grad_norm": 12.03525161743164, "learning_rate": 5.8336168720195696e-06, "loss": 0.4851, "step": 8940 }, { "epoch": 0.4522737619505286, "grad_norm": 8.391142845153809, "learning_rate": 5.832825480665526e-06, "loss": 0.4385, "step": 8941 }, { "epoch": 0.45232434619859374, "grad_norm": 7.590365886688232, "learning_rate": 5.832034067851582e-06, "loss": 0.3297, "step": 8942 }, { "epoch": 0.4523749304466589, "grad_norm": 7.947304725646973, "learning_rate": 5.83124263359813e-06, "loss": 0.5178, "step": 8943 }, { "epoch": 0.4524255146947241, "grad_norm": 10.886862754821777, "learning_rate": 5.830451177925562e-06, "loss": 0.458, "step": 8944 }, { "epoch": 0.45247609894278923, "grad_norm": 7.6068830490112305, "learning_rate": 5.8296597008542744e-06, "loss": 0.4227, "step": 8945 }, { "epoch": 0.4525266831908544, "grad_norm": 9.186463356018066, "learning_rate": 5.8288682024046605e-06, "loss": 0.3946, "step": 8946 }, { "epoch": 0.4525772674389195, "grad_norm": 12.485508918762207, "learning_rate": 5.828076682597113e-06, "loss": 0.4527, "step": 8947 }, { "epoch": 0.45262785168698466, "grad_norm": 10.552764892578125, "learning_rate": 5.827285141452031e-06, "loss": 0.5116, "step": 8948 }, { "epoch": 0.4526784359350498, "grad_norm": 12.580143928527832, "learning_rate": 5.82649357898981e-06, "loss": 0.4534, "step": 8949 }, { "epoch": 0.452729020183115, "grad_norm": 8.616849899291992, "learning_rate": 5.825701995230845e-06, "loss": 0.4443, "step": 8950 }, { "epoch": 0.45277960443118015, "grad_norm": 8.44727611541748, "learning_rate": 5.824910390195534e-06, "loss": 0.4922, "step": 8951 }, { "epoch": 0.4528301886792453, "grad_norm": 5.420000076293945, "learning_rate": 5.824118763904274e-06, "loss": 0.3637, "step": 8952 }, { "epoch": 0.45288077292731044, "grad_norm": 4.94431209564209, "learning_rate": 5.823327116377465e-06, "loss": 0.3738, "step": 8953 }, { "epoch": 0.4529313571753756, "grad_norm": 8.556543350219727, "learning_rate": 5.8225354476355035e-06, "loss": 0.3909, "step": 8954 }, { "epoch": 0.4529819414234407, "grad_norm": 6.388820171356201, "learning_rate": 5.821743757698792e-06, "loss": 0.4108, "step": 8955 }, { "epoch": 0.45303252567150587, "grad_norm": 16.304960250854492, "learning_rate": 5.820952046587727e-06, "loss": 0.4406, "step": 8956 }, { "epoch": 0.45308310991957107, "grad_norm": 13.761982917785645, "learning_rate": 5.8201603143227115e-06, "loss": 0.4854, "step": 8957 }, { "epoch": 0.4531336941676362, "grad_norm": 11.596234321594238, "learning_rate": 5.819368560924146e-06, "loss": 0.4783, "step": 8958 }, { "epoch": 0.45318427841570136, "grad_norm": 5.701384544372559, "learning_rate": 5.818576786412432e-06, "loss": 0.4328, "step": 8959 }, { "epoch": 0.4532348626637665, "grad_norm": 6.158570289611816, "learning_rate": 5.817784990807971e-06, "loss": 0.3479, "step": 8960 }, { "epoch": 0.45328544691183165, "grad_norm": 6.631683826446533, "learning_rate": 5.816993174131166e-06, "loss": 0.4579, "step": 8961 }, { "epoch": 0.4533360311598968, "grad_norm": 13.665740013122559, "learning_rate": 5.81620133640242e-06, "loss": 0.3287, "step": 8962 }, { "epoch": 0.45338661540796193, "grad_norm": 9.990999221801758, "learning_rate": 5.815409477642138e-06, "loss": 0.4384, "step": 8963 }, { "epoch": 0.45343719965602713, "grad_norm": 8.625253677368164, "learning_rate": 5.814617597870722e-06, "loss": 0.4347, "step": 8964 }, { "epoch": 0.4534877839040923, "grad_norm": 7.332375526428223, "learning_rate": 5.813825697108579e-06, "loss": 0.4391, "step": 8965 }, { "epoch": 0.4535383681521574, "grad_norm": 11.564815521240234, "learning_rate": 5.8130337753761134e-06, "loss": 0.4226, "step": 8966 }, { "epoch": 0.45358895240022257, "grad_norm": 5.549868106842041, "learning_rate": 5.81224183269373e-06, "loss": 0.3461, "step": 8967 }, { "epoch": 0.4536395366482877, "grad_norm": 7.392411231994629, "learning_rate": 5.811449869081839e-06, "loss": 0.4528, "step": 8968 }, { "epoch": 0.45369012089635286, "grad_norm": 7.3043646812438965, "learning_rate": 5.810657884560843e-06, "loss": 0.4484, "step": 8969 }, { "epoch": 0.45374070514441806, "grad_norm": 14.324980735778809, "learning_rate": 5.809865879151151e-06, "loss": 0.6584, "step": 8970 }, { "epoch": 0.4537912893924832, "grad_norm": 9.561330795288086, "learning_rate": 5.809073852873174e-06, "loss": 0.4031, "step": 8971 }, { "epoch": 0.45384187364054834, "grad_norm": 9.1864595413208, "learning_rate": 5.808281805747316e-06, "loss": 0.4998, "step": 8972 }, { "epoch": 0.4538924578886135, "grad_norm": 7.051003456115723, "learning_rate": 5.807489737793988e-06, "loss": 0.3725, "step": 8973 }, { "epoch": 0.45394304213667863, "grad_norm": 7.165761470794678, "learning_rate": 5.8066976490336e-06, "loss": 0.2738, "step": 8974 }, { "epoch": 0.4539936263847438, "grad_norm": 9.53894329071045, "learning_rate": 5.8059055394865625e-06, "loss": 0.4214, "step": 8975 }, { "epoch": 0.4540442106328089, "grad_norm": 8.769575119018555, "learning_rate": 5.8051134091732855e-06, "loss": 0.3763, "step": 8976 }, { "epoch": 0.4540947948808741, "grad_norm": 7.155059814453125, "learning_rate": 5.80432125811418e-06, "loss": 0.4053, "step": 8977 }, { "epoch": 0.45414537912893926, "grad_norm": 9.306717872619629, "learning_rate": 5.803529086329661e-06, "loss": 0.5104, "step": 8978 }, { "epoch": 0.4541959633770044, "grad_norm": 7.346035957336426, "learning_rate": 5.802736893840137e-06, "loss": 0.4368, "step": 8979 }, { "epoch": 0.45424654762506955, "grad_norm": 7.272339344024658, "learning_rate": 5.801944680666021e-06, "loss": 0.4333, "step": 8980 }, { "epoch": 0.4542971318731347, "grad_norm": 7.124924182891846, "learning_rate": 5.801152446827728e-06, "loss": 0.3883, "step": 8981 }, { "epoch": 0.45434771612119984, "grad_norm": 4.77957820892334, "learning_rate": 5.800360192345671e-06, "loss": 0.4164, "step": 8982 }, { "epoch": 0.454398300369265, "grad_norm": 6.314075946807861, "learning_rate": 5.799567917240267e-06, "loss": 0.4288, "step": 8983 }, { "epoch": 0.4544488846173302, "grad_norm": 9.58513069152832, "learning_rate": 5.798775621531929e-06, "loss": 0.4305, "step": 8984 }, { "epoch": 0.45449946886539533, "grad_norm": 9.439181327819824, "learning_rate": 5.797983305241071e-06, "loss": 0.4075, "step": 8985 }, { "epoch": 0.4545500531134605, "grad_norm": 14.556285858154297, "learning_rate": 5.79719096838811e-06, "loss": 0.4717, "step": 8986 }, { "epoch": 0.4546006373615256, "grad_norm": 10.23111629486084, "learning_rate": 5.796398610993465e-06, "loss": 0.4674, "step": 8987 }, { "epoch": 0.45465122160959076, "grad_norm": 4.936832427978516, "learning_rate": 5.795606233077551e-06, "loss": 0.4065, "step": 8988 }, { "epoch": 0.4547018058576559, "grad_norm": 7.009214401245117, "learning_rate": 5.794813834660785e-06, "loss": 0.4514, "step": 8989 }, { "epoch": 0.4547523901057211, "grad_norm": 18.07126808166504, "learning_rate": 5.79402141576359e-06, "loss": 0.4556, "step": 8990 }, { "epoch": 0.45480297435378625, "grad_norm": 7.715585708618164, "learning_rate": 5.7932289764063775e-06, "loss": 0.411, "step": 8991 }, { "epoch": 0.4548535586018514, "grad_norm": 6.798415660858154, "learning_rate": 5.792436516609571e-06, "loss": 0.4832, "step": 8992 }, { "epoch": 0.45490414284991654, "grad_norm": 13.609694480895996, "learning_rate": 5.79164403639359e-06, "loss": 0.5234, "step": 8993 }, { "epoch": 0.4549547270979817, "grad_norm": 10.704076766967773, "learning_rate": 5.790851535778853e-06, "loss": 0.4541, "step": 8994 }, { "epoch": 0.4550053113460468, "grad_norm": 12.062478065490723, "learning_rate": 5.790059014785783e-06, "loss": 0.5129, "step": 8995 }, { "epoch": 0.45505589559411197, "grad_norm": 7.277263641357422, "learning_rate": 5.7892664734348e-06, "loss": 0.418, "step": 8996 }, { "epoch": 0.45510647984217717, "grad_norm": 5.29734468460083, "learning_rate": 5.788473911746329e-06, "loss": 0.4027, "step": 8997 }, { "epoch": 0.4551570640902423, "grad_norm": 14.10838508605957, "learning_rate": 5.787681329740786e-06, "loss": 0.4905, "step": 8998 }, { "epoch": 0.45520764833830746, "grad_norm": 7.2446393966674805, "learning_rate": 5.786888727438599e-06, "loss": 0.4951, "step": 8999 }, { "epoch": 0.4552582325863726, "grad_norm": 14.04411792755127, "learning_rate": 5.786096104860191e-06, "loss": 0.6147, "step": 9000 }, { "epoch": 0.45530881683443775, "grad_norm": 8.24498462677002, "learning_rate": 5.785303462025984e-06, "loss": 0.4207, "step": 9001 }, { "epoch": 0.4553594010825029, "grad_norm": 8.555501937866211, "learning_rate": 5.784510798956405e-06, "loss": 0.3709, "step": 9002 }, { "epoch": 0.45540998533056803, "grad_norm": 5.258581638336182, "learning_rate": 5.783718115671878e-06, "loss": 0.4511, "step": 9003 }, { "epoch": 0.45546056957863323, "grad_norm": 6.389657020568848, "learning_rate": 5.782925412192826e-06, "loss": 0.464, "step": 9004 }, { "epoch": 0.4555111538266984, "grad_norm": 7.545287132263184, "learning_rate": 5.782132688539678e-06, "loss": 0.473, "step": 9005 }, { "epoch": 0.4555617380747635, "grad_norm": 7.713746070861816, "learning_rate": 5.78133994473286e-06, "loss": 0.4445, "step": 9006 }, { "epoch": 0.45561232232282867, "grad_norm": 11.134859085083008, "learning_rate": 5.7805471807927996e-06, "loss": 0.4872, "step": 9007 }, { "epoch": 0.4556629065708938, "grad_norm": 15.275293350219727, "learning_rate": 5.779754396739922e-06, "loss": 0.5889, "step": 9008 }, { "epoch": 0.45571349081895896, "grad_norm": 9.408913612365723, "learning_rate": 5.77896159259466e-06, "loss": 0.3666, "step": 9009 }, { "epoch": 0.45576407506702415, "grad_norm": 8.429817199707031, "learning_rate": 5.778168768377436e-06, "loss": 0.3883, "step": 9010 }, { "epoch": 0.4558146593150893, "grad_norm": 10.246439933776855, "learning_rate": 5.777375924108684e-06, "loss": 0.4143, "step": 9011 }, { "epoch": 0.45586524356315444, "grad_norm": 6.502802848815918, "learning_rate": 5.776583059808831e-06, "loss": 0.33, "step": 9012 }, { "epoch": 0.4559158278112196, "grad_norm": 7.167872905731201, "learning_rate": 5.775790175498309e-06, "loss": 0.416, "step": 9013 }, { "epoch": 0.45596641205928473, "grad_norm": 8.249359130859375, "learning_rate": 5.774997271197548e-06, "loss": 0.5052, "step": 9014 }, { "epoch": 0.4560169963073499, "grad_norm": 13.104567527770996, "learning_rate": 5.774204346926979e-06, "loss": 0.3469, "step": 9015 }, { "epoch": 0.456067580555415, "grad_norm": 7.904541492462158, "learning_rate": 5.7734114027070355e-06, "loss": 0.3306, "step": 9016 }, { "epoch": 0.4561181648034802, "grad_norm": 7.598265647888184, "learning_rate": 5.772618438558147e-06, "loss": 0.4353, "step": 9017 }, { "epoch": 0.45616874905154536, "grad_norm": 6.479465007781982, "learning_rate": 5.771825454500747e-06, "loss": 0.3311, "step": 9018 }, { "epoch": 0.4562193332996105, "grad_norm": 7.458426475524902, "learning_rate": 5.771032450555271e-06, "loss": 0.3815, "step": 9019 }, { "epoch": 0.45626991754767565, "grad_norm": 8.91335678100586, "learning_rate": 5.770239426742149e-06, "loss": 0.3531, "step": 9020 }, { "epoch": 0.4563205017957408, "grad_norm": 6.8203935623168945, "learning_rate": 5.7694463830818196e-06, "loss": 0.2883, "step": 9021 }, { "epoch": 0.45637108604380594, "grad_norm": 5.619834899902344, "learning_rate": 5.768653319594715e-06, "loss": 0.4313, "step": 9022 }, { "epoch": 0.4564216702918711, "grad_norm": 8.219882011413574, "learning_rate": 5.767860236301269e-06, "loss": 0.4765, "step": 9023 }, { "epoch": 0.4564722545399363, "grad_norm": 7.108674049377441, "learning_rate": 5.76706713322192e-06, "loss": 0.3918, "step": 9024 }, { "epoch": 0.45652283878800143, "grad_norm": 6.681577682495117, "learning_rate": 5.7662740103771055e-06, "loss": 0.4396, "step": 9025 }, { "epoch": 0.4565734230360666, "grad_norm": 4.872593879699707, "learning_rate": 5.765480867787259e-06, "loss": 0.4283, "step": 9026 }, { "epoch": 0.4566240072841317, "grad_norm": 10.659469604492188, "learning_rate": 5.764687705472819e-06, "loss": 0.5452, "step": 9027 }, { "epoch": 0.45667459153219686, "grad_norm": 8.035318374633789, "learning_rate": 5.763894523454225e-06, "loss": 0.4384, "step": 9028 }, { "epoch": 0.456725175780262, "grad_norm": 6.011991024017334, "learning_rate": 5.763101321751915e-06, "loss": 0.378, "step": 9029 }, { "epoch": 0.4567757600283272, "grad_norm": 7.941108226776123, "learning_rate": 5.762308100386326e-06, "loss": 0.4769, "step": 9030 }, { "epoch": 0.45682634427639235, "grad_norm": 11.511884689331055, "learning_rate": 5.761514859377899e-06, "loss": 0.4917, "step": 9031 }, { "epoch": 0.4568769285244575, "grad_norm": 7.44468879699707, "learning_rate": 5.760721598747073e-06, "loss": 0.4875, "step": 9032 }, { "epoch": 0.45692751277252264, "grad_norm": 13.787568092346191, "learning_rate": 5.759928318514288e-06, "loss": 0.4236, "step": 9033 }, { "epoch": 0.4569780970205878, "grad_norm": 5.171320915222168, "learning_rate": 5.759135018699988e-06, "loss": 0.4148, "step": 9034 }, { "epoch": 0.4570286812686529, "grad_norm": 8.962047576904297, "learning_rate": 5.758341699324612e-06, "loss": 0.4898, "step": 9035 }, { "epoch": 0.45707926551671807, "grad_norm": 8.945616722106934, "learning_rate": 5.7575483604086e-06, "loss": 0.449, "step": 9036 }, { "epoch": 0.45712984976478327, "grad_norm": 7.392552375793457, "learning_rate": 5.7567550019723974e-06, "loss": 0.4375, "step": 9037 }, { "epoch": 0.4571804340128484, "grad_norm": 6.762667179107666, "learning_rate": 5.755961624036447e-06, "loss": 0.4461, "step": 9038 }, { "epoch": 0.45723101826091356, "grad_norm": 20.179126739501953, "learning_rate": 5.75516822662119e-06, "loss": 0.5841, "step": 9039 }, { "epoch": 0.4572816025089787, "grad_norm": 6.798849582672119, "learning_rate": 5.7543748097470745e-06, "loss": 0.345, "step": 9040 }, { "epoch": 0.45733218675704385, "grad_norm": 7.130910873413086, "learning_rate": 5.75358137343454e-06, "loss": 0.3476, "step": 9041 }, { "epoch": 0.457382771005109, "grad_norm": 16.433008193969727, "learning_rate": 5.752787917704035e-06, "loss": 0.5103, "step": 9042 }, { "epoch": 0.45743335525317413, "grad_norm": 7.121696472167969, "learning_rate": 5.751994442576002e-06, "loss": 0.4144, "step": 9043 }, { "epoch": 0.45748393950123933, "grad_norm": 8.273153305053711, "learning_rate": 5.75120094807089e-06, "loss": 0.4796, "step": 9044 }, { "epoch": 0.4575345237493045, "grad_norm": 5.918117523193359, "learning_rate": 5.750407434209143e-06, "loss": 0.4589, "step": 9045 }, { "epoch": 0.4575851079973696, "grad_norm": 5.628374099731445, "learning_rate": 5.74961390101121e-06, "loss": 0.4211, "step": 9046 }, { "epoch": 0.45763569224543477, "grad_norm": 11.329816818237305, "learning_rate": 5.748820348497537e-06, "loss": 0.5831, "step": 9047 }, { "epoch": 0.4576862764934999, "grad_norm": 6.10902738571167, "learning_rate": 5.748026776688572e-06, "loss": 0.3684, "step": 9048 }, { "epoch": 0.45773686074156505, "grad_norm": 12.194904327392578, "learning_rate": 5.747233185604764e-06, "loss": 0.5083, "step": 9049 }, { "epoch": 0.45778744498963025, "grad_norm": 6.794332027435303, "learning_rate": 5.746439575266563e-06, "loss": 0.4716, "step": 9050 }, { "epoch": 0.4578380292376954, "grad_norm": 14.450531005859375, "learning_rate": 5.745645945694414e-06, "loss": 0.5456, "step": 9051 }, { "epoch": 0.45788861348576054, "grad_norm": 12.489713668823242, "learning_rate": 5.744852296908772e-06, "loss": 0.4944, "step": 9052 }, { "epoch": 0.4579391977338257, "grad_norm": 8.572175979614258, "learning_rate": 5.7440586289300846e-06, "loss": 0.5397, "step": 9053 }, { "epoch": 0.45798978198189083, "grad_norm": 5.230024337768555, "learning_rate": 5.743264941778806e-06, "loss": 0.3961, "step": 9054 }, { "epoch": 0.458040366229956, "grad_norm": 7.686206817626953, "learning_rate": 5.742471235475382e-06, "loss": 0.4959, "step": 9055 }, { "epoch": 0.4580909504780211, "grad_norm": 8.402490615844727, "learning_rate": 5.741677510040269e-06, "loss": 0.4771, "step": 9056 }, { "epoch": 0.4581415347260863, "grad_norm": 8.12755298614502, "learning_rate": 5.7408837654939174e-06, "loss": 0.4365, "step": 9057 }, { "epoch": 0.45819211897415146, "grad_norm": 7.562368392944336, "learning_rate": 5.740090001856781e-06, "loss": 0.3669, "step": 9058 }, { "epoch": 0.4582427032222166, "grad_norm": 8.590124130249023, "learning_rate": 5.739296219149314e-06, "loss": 0.4884, "step": 9059 }, { "epoch": 0.45829328747028175, "grad_norm": 16.319412231445312, "learning_rate": 5.738502417391968e-06, "loss": 0.5766, "step": 9060 }, { "epoch": 0.4583438717183469, "grad_norm": 6.672562599182129, "learning_rate": 5.7377085966051995e-06, "loss": 0.4983, "step": 9061 }, { "epoch": 0.45839445596641204, "grad_norm": 7.837436199188232, "learning_rate": 5.736914756809461e-06, "loss": 0.4272, "step": 9062 }, { "epoch": 0.4584450402144772, "grad_norm": 11.049560546875, "learning_rate": 5.73612089802521e-06, "loss": 0.4338, "step": 9063 }, { "epoch": 0.4584956244625424, "grad_norm": 16.05181121826172, "learning_rate": 5.735327020272901e-06, "loss": 0.5555, "step": 9064 }, { "epoch": 0.45854620871060753, "grad_norm": 9.043118476867676, "learning_rate": 5.73453312357299e-06, "loss": 0.507, "step": 9065 }, { "epoch": 0.45859679295867267, "grad_norm": 4.491930961608887, "learning_rate": 5.733739207945935e-06, "loss": 0.3773, "step": 9066 }, { "epoch": 0.4586473772067378, "grad_norm": 7.337682723999023, "learning_rate": 5.732945273412195e-06, "loss": 0.4177, "step": 9067 }, { "epoch": 0.45869796145480296, "grad_norm": 10.477127075195312, "learning_rate": 5.732151319992223e-06, "loss": 0.3799, "step": 9068 }, { "epoch": 0.4587485457028681, "grad_norm": 11.76690673828125, "learning_rate": 5.7313573477064825e-06, "loss": 0.4733, "step": 9069 }, { "epoch": 0.4587991299509333, "grad_norm": 7.882683753967285, "learning_rate": 5.730563356575427e-06, "loss": 0.4305, "step": 9070 }, { "epoch": 0.45884971419899845, "grad_norm": 9.023940086364746, "learning_rate": 5.729769346619519e-06, "loss": 0.3703, "step": 9071 }, { "epoch": 0.4589002984470636, "grad_norm": 9.605460166931152, "learning_rate": 5.728975317859219e-06, "loss": 0.4975, "step": 9072 }, { "epoch": 0.45895088269512874, "grad_norm": 5.275416374206543, "learning_rate": 5.728181270314984e-06, "loss": 0.3166, "step": 9073 }, { "epoch": 0.4590014669431939, "grad_norm": 5.52048921585083, "learning_rate": 5.727387204007279e-06, "loss": 0.498, "step": 9074 }, { "epoch": 0.459052051191259, "grad_norm": 10.410521507263184, "learning_rate": 5.726593118956561e-06, "loss": 0.5327, "step": 9075 }, { "epoch": 0.45910263543932417, "grad_norm": 4.856059551239014, "learning_rate": 5.7257990151832935e-06, "loss": 0.2855, "step": 9076 }, { "epoch": 0.45915321968738937, "grad_norm": 17.07293701171875, "learning_rate": 5.725004892707938e-06, "loss": 0.4867, "step": 9077 }, { "epoch": 0.4592038039354545, "grad_norm": 6.5744147300720215, "learning_rate": 5.724210751550958e-06, "loss": 0.3415, "step": 9078 }, { "epoch": 0.45925438818351966, "grad_norm": 8.943692207336426, "learning_rate": 5.723416591732816e-06, "loss": 0.3963, "step": 9079 }, { "epoch": 0.4593049724315848, "grad_norm": 6.368921756744385, "learning_rate": 5.722622413273977e-06, "loss": 0.4213, "step": 9080 }, { "epoch": 0.45935555667964995, "grad_norm": 8.639649391174316, "learning_rate": 5.721828216194902e-06, "loss": 0.4312, "step": 9081 }, { "epoch": 0.4594061409277151, "grad_norm": 6.621675968170166, "learning_rate": 5.7210340005160595e-06, "loss": 0.3648, "step": 9082 }, { "epoch": 0.4594567251757803, "grad_norm": 7.520169258117676, "learning_rate": 5.720239766257911e-06, "loss": 0.312, "step": 9083 }, { "epoch": 0.45950730942384543, "grad_norm": 9.586394309997559, "learning_rate": 5.719445513440922e-06, "loss": 0.5213, "step": 9084 }, { "epoch": 0.4595578936719106, "grad_norm": 6.697306156158447, "learning_rate": 5.718651242085562e-06, "loss": 0.3981, "step": 9085 }, { "epoch": 0.4596084779199757, "grad_norm": 7.674910545349121, "learning_rate": 5.717856952212295e-06, "loss": 0.3885, "step": 9086 }, { "epoch": 0.45965906216804087, "grad_norm": 6.136214733123779, "learning_rate": 5.717062643841589e-06, "loss": 0.4111, "step": 9087 }, { "epoch": 0.459709646416106, "grad_norm": 5.393564701080322, "learning_rate": 5.716268316993911e-06, "loss": 0.3659, "step": 9088 }, { "epoch": 0.45976023066417115, "grad_norm": 11.466676712036133, "learning_rate": 5.715473971689726e-06, "loss": 0.4859, "step": 9089 }, { "epoch": 0.45981081491223635, "grad_norm": 8.82845401763916, "learning_rate": 5.714679607949505e-06, "loss": 0.4594, "step": 9090 }, { "epoch": 0.4598613991603015, "grad_norm": 9.798834800720215, "learning_rate": 5.713885225793719e-06, "loss": 0.4039, "step": 9091 }, { "epoch": 0.45991198340836664, "grad_norm": 12.024611473083496, "learning_rate": 5.713090825242833e-06, "loss": 0.5343, "step": 9092 }, { "epoch": 0.4599625676564318, "grad_norm": 7.253685474395752, "learning_rate": 5.71229640631732e-06, "loss": 0.4256, "step": 9093 }, { "epoch": 0.46001315190449693, "grad_norm": 7.944624423980713, "learning_rate": 5.711501969037649e-06, "loss": 0.3739, "step": 9094 }, { "epoch": 0.4600637361525621, "grad_norm": 15.771695137023926, "learning_rate": 5.710707513424291e-06, "loss": 0.4663, "step": 9095 }, { "epoch": 0.4601143204006272, "grad_norm": 5.590806007385254, "learning_rate": 5.709913039497716e-06, "loss": 0.4691, "step": 9096 }, { "epoch": 0.4601649046486924, "grad_norm": 12.13487720489502, "learning_rate": 5.709118547278398e-06, "loss": 0.5391, "step": 9097 }, { "epoch": 0.46021548889675756, "grad_norm": 7.770823001861572, "learning_rate": 5.708324036786808e-06, "loss": 0.4259, "step": 9098 }, { "epoch": 0.4602660731448227, "grad_norm": 7.216815948486328, "learning_rate": 5.707529508043418e-06, "loss": 0.456, "step": 9099 }, { "epoch": 0.46031665739288785, "grad_norm": 10.576047897338867, "learning_rate": 5.7067349610687e-06, "loss": 0.3483, "step": 9100 }, { "epoch": 0.460367241640953, "grad_norm": 15.005352973937988, "learning_rate": 5.705940395883132e-06, "loss": 0.4937, "step": 9101 }, { "epoch": 0.46041782588901814, "grad_norm": 7.011690616607666, "learning_rate": 5.7051458125071825e-06, "loss": 0.3857, "step": 9102 }, { "epoch": 0.46046841013708334, "grad_norm": 10.310887336730957, "learning_rate": 5.704351210961329e-06, "loss": 0.3553, "step": 9103 }, { "epoch": 0.4605189943851485, "grad_norm": 6.741755962371826, "learning_rate": 5.703556591266048e-06, "loss": 0.3052, "step": 9104 }, { "epoch": 0.4605695786332136, "grad_norm": 9.178657531738281, "learning_rate": 5.70276195344181e-06, "loss": 0.3588, "step": 9105 }, { "epoch": 0.46062016288127877, "grad_norm": 5.259054183959961, "learning_rate": 5.7019672975090965e-06, "loss": 0.3186, "step": 9106 }, { "epoch": 0.4606707471293439, "grad_norm": 7.001789093017578, "learning_rate": 5.70117262348838e-06, "loss": 0.425, "step": 9107 }, { "epoch": 0.46072133137740906, "grad_norm": 12.065558433532715, "learning_rate": 5.700377931400139e-06, "loss": 0.4575, "step": 9108 }, { "epoch": 0.4607719156254742, "grad_norm": 12.252248764038086, "learning_rate": 5.699583221264851e-06, "loss": 0.5832, "step": 9109 }, { "epoch": 0.4608224998735394, "grad_norm": 9.79577350616455, "learning_rate": 5.6987884931029925e-06, "loss": 0.4768, "step": 9110 }, { "epoch": 0.46087308412160455, "grad_norm": 9.171774864196777, "learning_rate": 5.6979937469350425e-06, "loss": 0.3423, "step": 9111 }, { "epoch": 0.4609236683696697, "grad_norm": 7.503808975219727, "learning_rate": 5.697198982781479e-06, "loss": 0.4068, "step": 9112 }, { "epoch": 0.46097425261773484, "grad_norm": 10.020682334899902, "learning_rate": 5.696404200662783e-06, "loss": 0.414, "step": 9113 }, { "epoch": 0.4610248368658, "grad_norm": 8.315991401672363, "learning_rate": 5.695609400599431e-06, "loss": 0.5032, "step": 9114 }, { "epoch": 0.4610754211138651, "grad_norm": 6.040711402893066, "learning_rate": 5.694814582611906e-06, "loss": 0.3806, "step": 9115 }, { "epoch": 0.46112600536193027, "grad_norm": 6.024463653564453, "learning_rate": 5.694019746720688e-06, "loss": 0.2919, "step": 9116 }, { "epoch": 0.46117658960999547, "grad_norm": 14.68617057800293, "learning_rate": 5.693224892946256e-06, "loss": 0.5817, "step": 9117 }, { "epoch": 0.4612271738580606, "grad_norm": 6.97390604019165, "learning_rate": 5.6924300213090945e-06, "loss": 0.3761, "step": 9118 }, { "epoch": 0.46127775810612576, "grad_norm": 7.528282165527344, "learning_rate": 5.6916351318296835e-06, "loss": 0.4508, "step": 9119 }, { "epoch": 0.4613283423541909, "grad_norm": 9.04688549041748, "learning_rate": 5.690840224528505e-06, "loss": 0.4575, "step": 9120 }, { "epoch": 0.46137892660225605, "grad_norm": 6.439747333526611, "learning_rate": 5.690045299426044e-06, "loss": 0.4313, "step": 9121 }, { "epoch": 0.4614295108503212, "grad_norm": 6.940822124481201, "learning_rate": 5.68925035654278e-06, "loss": 0.3603, "step": 9122 }, { "epoch": 0.4614800950983864, "grad_norm": 9.392651557922363, "learning_rate": 5.688455395899202e-06, "loss": 0.4069, "step": 9123 }, { "epoch": 0.46153067934645153, "grad_norm": 8.189751625061035, "learning_rate": 5.687660417515789e-06, "loss": 0.4831, "step": 9124 }, { "epoch": 0.4615812635945167, "grad_norm": 13.433880805969238, "learning_rate": 5.686865421413029e-06, "loss": 0.5113, "step": 9125 }, { "epoch": 0.4616318478425818, "grad_norm": 8.00397777557373, "learning_rate": 5.686070407611407e-06, "loss": 0.3342, "step": 9126 }, { "epoch": 0.46168243209064697, "grad_norm": 7.78349494934082, "learning_rate": 5.685275376131406e-06, "loss": 0.3766, "step": 9127 }, { "epoch": 0.4617330163387121, "grad_norm": 10.44558334350586, "learning_rate": 5.684480326993513e-06, "loss": 0.5224, "step": 9128 }, { "epoch": 0.46178360058677725, "grad_norm": 4.649426460266113, "learning_rate": 5.683685260218218e-06, "loss": 0.322, "step": 9129 }, { "epoch": 0.46183418483484245, "grad_norm": 9.624335289001465, "learning_rate": 5.6828901758260035e-06, "loss": 0.4228, "step": 9130 }, { "epoch": 0.4618847690829076, "grad_norm": 9.378620147705078, "learning_rate": 5.682095073837359e-06, "loss": 0.468, "step": 9131 }, { "epoch": 0.46193535333097274, "grad_norm": 8.208976745605469, "learning_rate": 5.681299954272774e-06, "loss": 0.4261, "step": 9132 }, { "epoch": 0.4619859375790379, "grad_norm": 10.737719535827637, "learning_rate": 5.680504817152732e-06, "loss": 0.4803, "step": 9133 }, { "epoch": 0.46203652182710303, "grad_norm": 7.885081768035889, "learning_rate": 5.679709662497726e-06, "loss": 0.3891, "step": 9134 }, { "epoch": 0.4620871060751682, "grad_norm": 8.329301834106445, "learning_rate": 5.678914490328242e-06, "loss": 0.4019, "step": 9135 }, { "epoch": 0.4621376903232333, "grad_norm": 6.346966743469238, "learning_rate": 5.678119300664773e-06, "loss": 0.384, "step": 9136 }, { "epoch": 0.4621882745712985, "grad_norm": 5.635665416717529, "learning_rate": 5.6773240935278064e-06, "loss": 0.428, "step": 9137 }, { "epoch": 0.46223885881936366, "grad_norm": 12.975035667419434, "learning_rate": 5.6765288689378365e-06, "loss": 0.5006, "step": 9138 }, { "epoch": 0.4622894430674288, "grad_norm": 6.32387638092041, "learning_rate": 5.67573362691535e-06, "loss": 0.5104, "step": 9139 }, { "epoch": 0.46234002731549395, "grad_norm": 7.723547458648682, "learning_rate": 5.674938367480839e-06, "loss": 0.4004, "step": 9140 }, { "epoch": 0.4623906115635591, "grad_norm": 6.152843475341797, "learning_rate": 5.674143090654798e-06, "loss": 0.3972, "step": 9141 }, { "epoch": 0.46244119581162424, "grad_norm": 10.012909889221191, "learning_rate": 5.673347796457718e-06, "loss": 0.5194, "step": 9142 }, { "epoch": 0.46249178005968944, "grad_norm": 7.906123638153076, "learning_rate": 5.67255248491009e-06, "loss": 0.4055, "step": 9143 }, { "epoch": 0.4625423643077546, "grad_norm": 4.693272113800049, "learning_rate": 5.6717571560324105e-06, "loss": 0.2655, "step": 9144 }, { "epoch": 0.4625929485558197, "grad_norm": 9.287363052368164, "learning_rate": 5.670961809845172e-06, "loss": 0.3269, "step": 9145 }, { "epoch": 0.46264353280388487, "grad_norm": 8.145771026611328, "learning_rate": 5.670166446368867e-06, "loss": 0.2747, "step": 9146 }, { "epoch": 0.46269411705195, "grad_norm": 5.366334915161133, "learning_rate": 5.669371065623992e-06, "loss": 0.3685, "step": 9147 }, { "epoch": 0.46274470130001516, "grad_norm": 11.814933776855469, "learning_rate": 5.668575667631041e-06, "loss": 0.5263, "step": 9148 }, { "epoch": 0.4627952855480803, "grad_norm": 23.847631454467773, "learning_rate": 5.66778025241051e-06, "loss": 0.4534, "step": 9149 }, { "epoch": 0.4628458697961455, "grad_norm": 6.516595840454102, "learning_rate": 5.666984819982895e-06, "loss": 0.3643, "step": 9150 }, { "epoch": 0.46289645404421065, "grad_norm": 9.358848571777344, "learning_rate": 5.666189370368694e-06, "loss": 0.3382, "step": 9151 }, { "epoch": 0.4629470382922758, "grad_norm": 8.327591896057129, "learning_rate": 5.6653939035883995e-06, "loss": 0.3986, "step": 9152 }, { "epoch": 0.46299762254034094, "grad_norm": 4.639569282531738, "learning_rate": 5.664598419662512e-06, "loss": 0.2982, "step": 9153 }, { "epoch": 0.4630482067884061, "grad_norm": 6.1985273361206055, "learning_rate": 5.6638029186115295e-06, "loss": 0.4445, "step": 9154 }, { "epoch": 0.4630987910364712, "grad_norm": 10.125971794128418, "learning_rate": 5.663007400455948e-06, "loss": 0.4692, "step": 9155 }, { "epoch": 0.46314937528453637, "grad_norm": 8.518667221069336, "learning_rate": 5.662211865216267e-06, "loss": 0.3089, "step": 9156 }, { "epoch": 0.46319995953260157, "grad_norm": 7.7351555824279785, "learning_rate": 5.661416312912986e-06, "loss": 0.464, "step": 9157 }, { "epoch": 0.4632505437806667, "grad_norm": 11.621603965759277, "learning_rate": 5.660620743566606e-06, "loss": 0.3394, "step": 9158 }, { "epoch": 0.46330112802873186, "grad_norm": 8.029328346252441, "learning_rate": 5.659825157197622e-06, "loss": 0.4923, "step": 9159 }, { "epoch": 0.463351712276797, "grad_norm": 8.480774879455566, "learning_rate": 5.659029553826539e-06, "loss": 0.5517, "step": 9160 }, { "epoch": 0.46340229652486215, "grad_norm": 11.539011001586914, "learning_rate": 5.658233933473858e-06, "loss": 0.4647, "step": 9161 }, { "epoch": 0.4634528807729273, "grad_norm": 7.243558406829834, "learning_rate": 5.657438296160076e-06, "loss": 0.4123, "step": 9162 }, { "epoch": 0.4635034650209925, "grad_norm": 11.301923751831055, "learning_rate": 5.656642641905699e-06, "loss": 0.3761, "step": 9163 }, { "epoch": 0.46355404926905763, "grad_norm": 8.992915153503418, "learning_rate": 5.655846970731226e-06, "loss": 0.4031, "step": 9164 }, { "epoch": 0.4636046335171228, "grad_norm": 5.802628993988037, "learning_rate": 5.65505128265716e-06, "loss": 0.389, "step": 9165 }, { "epoch": 0.4636552177651879, "grad_norm": 10.164658546447754, "learning_rate": 5.654255577704004e-06, "loss": 0.4448, "step": 9166 }, { "epoch": 0.46370580201325307, "grad_norm": 4.416214466094971, "learning_rate": 5.653459855892265e-06, "loss": 0.4032, "step": 9167 }, { "epoch": 0.4637563862613182, "grad_norm": 6.432437419891357, "learning_rate": 5.6526641172424414e-06, "loss": 0.4057, "step": 9168 }, { "epoch": 0.46380697050938335, "grad_norm": 7.739708423614502, "learning_rate": 5.65186836177504e-06, "loss": 0.3536, "step": 9169 }, { "epoch": 0.46385755475744855, "grad_norm": 9.606988906860352, "learning_rate": 5.651072589510568e-06, "loss": 0.4476, "step": 9170 }, { "epoch": 0.4639081390055137, "grad_norm": 14.12194538116455, "learning_rate": 5.650276800469526e-06, "loss": 0.5132, "step": 9171 }, { "epoch": 0.46395872325357884, "grad_norm": 6.581841468811035, "learning_rate": 5.649480994672421e-06, "loss": 0.2277, "step": 9172 }, { "epoch": 0.464009307501644, "grad_norm": 13.488012313842773, "learning_rate": 5.64868517213976e-06, "loss": 0.4641, "step": 9173 }, { "epoch": 0.46405989174970913, "grad_norm": 6.8047194480896, "learning_rate": 5.647889332892049e-06, "loss": 0.3529, "step": 9174 }, { "epoch": 0.4641104759977743, "grad_norm": 8.856744766235352, "learning_rate": 5.647093476949795e-06, "loss": 0.4205, "step": 9175 }, { "epoch": 0.4641610602458394, "grad_norm": 7.745422840118408, "learning_rate": 5.646297604333506e-06, "loss": 0.327, "step": 9176 }, { "epoch": 0.4642116444939046, "grad_norm": 7.35706901550293, "learning_rate": 5.645501715063688e-06, "loss": 0.3601, "step": 9177 }, { "epoch": 0.46426222874196976, "grad_norm": 13.598896980285645, "learning_rate": 5.64470580916085e-06, "loss": 0.4627, "step": 9178 }, { "epoch": 0.4643128129900349, "grad_norm": 10.074317932128906, "learning_rate": 5.643909886645501e-06, "loss": 0.51, "step": 9179 }, { "epoch": 0.46436339723810005, "grad_norm": 12.89260482788086, "learning_rate": 5.643113947538148e-06, "loss": 0.387, "step": 9180 }, { "epoch": 0.4644139814861652, "grad_norm": 8.587932586669922, "learning_rate": 5.642317991859303e-06, "loss": 0.4804, "step": 9181 }, { "epoch": 0.46446456573423034, "grad_norm": 7.309324741363525, "learning_rate": 5.641522019629475e-06, "loss": 0.4368, "step": 9182 }, { "epoch": 0.46451514998229554, "grad_norm": 5.890264511108398, "learning_rate": 5.640726030869175e-06, "loss": 0.3498, "step": 9183 }, { "epoch": 0.4645657342303607, "grad_norm": 7.0719709396362305, "learning_rate": 5.63993002559891e-06, "loss": 0.4642, "step": 9184 }, { "epoch": 0.4646163184784258, "grad_norm": 11.577032089233398, "learning_rate": 5.639134003839195e-06, "loss": 0.5015, "step": 9185 }, { "epoch": 0.46466690272649097, "grad_norm": 10.060405731201172, "learning_rate": 5.638337965610542e-06, "loss": 0.4004, "step": 9186 }, { "epoch": 0.4647174869745561, "grad_norm": 6.30727481842041, "learning_rate": 5.6375419109334595e-06, "loss": 0.3954, "step": 9187 }, { "epoch": 0.46476807122262126, "grad_norm": 9.287582397460938, "learning_rate": 5.636745839828462e-06, "loss": 0.375, "step": 9188 }, { "epoch": 0.4648186554706864, "grad_norm": 16.141878128051758, "learning_rate": 5.635949752316066e-06, "loss": 0.5333, "step": 9189 }, { "epoch": 0.4648692397187516, "grad_norm": 6.761744976043701, "learning_rate": 5.635153648416777e-06, "loss": 0.4049, "step": 9190 }, { "epoch": 0.46491982396681675, "grad_norm": 5.406529903411865, "learning_rate": 5.634357528151113e-06, "loss": 0.2768, "step": 9191 }, { "epoch": 0.4649704082148819, "grad_norm": 7.605826377868652, "learning_rate": 5.633561391539588e-06, "loss": 0.4643, "step": 9192 }, { "epoch": 0.46502099246294704, "grad_norm": 6.981164455413818, "learning_rate": 5.6327652386027164e-06, "loss": 0.4137, "step": 9193 }, { "epoch": 0.4650715767110122, "grad_norm": 4.068081378936768, "learning_rate": 5.631969069361013e-06, "loss": 0.3516, "step": 9194 }, { "epoch": 0.4651221609590773, "grad_norm": 7.694221019744873, "learning_rate": 5.631172883834994e-06, "loss": 0.4562, "step": 9195 }, { "epoch": 0.46517274520714247, "grad_norm": 12.948922157287598, "learning_rate": 5.630376682045173e-06, "loss": 0.5712, "step": 9196 }, { "epoch": 0.46522332945520767, "grad_norm": 14.835260391235352, "learning_rate": 5.629580464012069e-06, "loss": 0.5806, "step": 9197 }, { "epoch": 0.4652739137032728, "grad_norm": 12.401004791259766, "learning_rate": 5.628784229756197e-06, "loss": 0.5156, "step": 9198 }, { "epoch": 0.46532449795133796, "grad_norm": 6.412918567657471, "learning_rate": 5.627987979298072e-06, "loss": 0.3427, "step": 9199 }, { "epoch": 0.4653750821994031, "grad_norm": 11.374887466430664, "learning_rate": 5.6271917126582154e-06, "loss": 0.4467, "step": 9200 }, { "epoch": 0.46542566644746824, "grad_norm": 14.389666557312012, "learning_rate": 5.626395429857143e-06, "loss": 0.5434, "step": 9201 }, { "epoch": 0.4654762506955334, "grad_norm": 6.9513092041015625, "learning_rate": 5.625599130915373e-06, "loss": 0.4266, "step": 9202 }, { "epoch": 0.4655268349435986, "grad_norm": 12.691203117370605, "learning_rate": 5.624802815853425e-06, "loss": 0.4377, "step": 9203 }, { "epoch": 0.46557741919166373, "grad_norm": 5.086924076080322, "learning_rate": 5.624006484691818e-06, "loss": 0.2988, "step": 9204 }, { "epoch": 0.4656280034397289, "grad_norm": 8.23482608795166, "learning_rate": 5.623210137451069e-06, "loss": 0.5065, "step": 9205 }, { "epoch": 0.465678587687794, "grad_norm": 6.771098613739014, "learning_rate": 5.622413774151702e-06, "loss": 0.4406, "step": 9206 }, { "epoch": 0.46572917193585917, "grad_norm": 9.086363792419434, "learning_rate": 5.621617394814234e-06, "loss": 0.3106, "step": 9207 }, { "epoch": 0.4657797561839243, "grad_norm": 6.381585597991943, "learning_rate": 5.6208209994591885e-06, "loss": 0.4346, "step": 9208 }, { "epoch": 0.46583034043198945, "grad_norm": 7.219226837158203, "learning_rate": 5.6200245881070855e-06, "loss": 0.4057, "step": 9209 }, { "epoch": 0.46588092468005465, "grad_norm": 8.119060516357422, "learning_rate": 5.619228160778446e-06, "loss": 0.4944, "step": 9210 }, { "epoch": 0.4659315089281198, "grad_norm": 6.186088562011719, "learning_rate": 5.618431717493793e-06, "loss": 0.389, "step": 9211 }, { "epoch": 0.46598209317618494, "grad_norm": 6.346588611602783, "learning_rate": 5.617635258273648e-06, "loss": 0.4683, "step": 9212 }, { "epoch": 0.4660326774242501, "grad_norm": 7.436845779418945, "learning_rate": 5.616838783138533e-06, "loss": 0.3842, "step": 9213 }, { "epoch": 0.46608326167231523, "grad_norm": 7.376153469085693, "learning_rate": 5.616042292108974e-06, "loss": 0.4421, "step": 9214 }, { "epoch": 0.4661338459203804, "grad_norm": 4.888347625732422, "learning_rate": 5.615245785205494e-06, "loss": 0.3876, "step": 9215 }, { "epoch": 0.4661844301684456, "grad_norm": 5.280120372772217, "learning_rate": 5.614449262448613e-06, "loss": 0.389, "step": 9216 }, { "epoch": 0.4662350144165107, "grad_norm": 5.578779220581055, "learning_rate": 5.6136527238588625e-06, "loss": 0.3776, "step": 9217 }, { "epoch": 0.46628559866457586, "grad_norm": 14.92813777923584, "learning_rate": 5.612856169456761e-06, "loss": 0.4667, "step": 9218 }, { "epoch": 0.466336182912641, "grad_norm": 19.47090721130371, "learning_rate": 5.612059599262836e-06, "loss": 0.6478, "step": 9219 }, { "epoch": 0.46638676716070615, "grad_norm": 5.963635444641113, "learning_rate": 5.6112630132976156e-06, "loss": 0.4785, "step": 9220 }, { "epoch": 0.4664373514087713, "grad_norm": 9.132530212402344, "learning_rate": 5.610466411581621e-06, "loss": 0.3812, "step": 9221 }, { "epoch": 0.46648793565683644, "grad_norm": 9.690494537353516, "learning_rate": 5.609669794135384e-06, "loss": 0.382, "step": 9222 }, { "epoch": 0.46653851990490164, "grad_norm": 6.6975884437561035, "learning_rate": 5.608873160979427e-06, "loss": 0.4101, "step": 9223 }, { "epoch": 0.4665891041529668, "grad_norm": 4.810339450836182, "learning_rate": 5.608076512134282e-06, "loss": 0.4155, "step": 9224 }, { "epoch": 0.4666396884010319, "grad_norm": 10.11327838897705, "learning_rate": 5.607279847620472e-06, "loss": 0.4272, "step": 9225 }, { "epoch": 0.46669027264909707, "grad_norm": 6.840210914611816, "learning_rate": 5.606483167458526e-06, "loss": 0.2682, "step": 9226 }, { "epoch": 0.4667408568971622, "grad_norm": 7.2943525314331055, "learning_rate": 5.605686471668977e-06, "loss": 0.4361, "step": 9227 }, { "epoch": 0.46679144114522736, "grad_norm": 7.727958679199219, "learning_rate": 5.604889760272349e-06, "loss": 0.3454, "step": 9228 }, { "epoch": 0.4668420253932925, "grad_norm": 5.608309268951416, "learning_rate": 5.604093033289172e-06, "loss": 0.3799, "step": 9229 }, { "epoch": 0.4668926096413577, "grad_norm": 7.690146446228027, "learning_rate": 5.603296290739978e-06, "loss": 0.3737, "step": 9230 }, { "epoch": 0.46694319388942285, "grad_norm": 7.146229267120361, "learning_rate": 5.602499532645295e-06, "loss": 0.4793, "step": 9231 }, { "epoch": 0.466993778137488, "grad_norm": 10.559073448181152, "learning_rate": 5.601702759025654e-06, "loss": 0.5571, "step": 9232 }, { "epoch": 0.46704436238555314, "grad_norm": 7.402580261230469, "learning_rate": 5.600905969901587e-06, "loss": 0.4241, "step": 9233 }, { "epoch": 0.4670949466336183, "grad_norm": 5.952411651611328, "learning_rate": 5.600109165293625e-06, "loss": 0.4041, "step": 9234 }, { "epoch": 0.4671455308816834, "grad_norm": 4.5607590675354, "learning_rate": 5.599312345222297e-06, "loss": 0.3256, "step": 9235 }, { "epoch": 0.4671961151297486, "grad_norm": 10.46621322631836, "learning_rate": 5.59851550970814e-06, "loss": 0.4401, "step": 9236 }, { "epoch": 0.46724669937781377, "grad_norm": 8.866877555847168, "learning_rate": 5.597718658771684e-06, "loss": 0.4669, "step": 9237 }, { "epoch": 0.4672972836258789, "grad_norm": 8.280509948730469, "learning_rate": 5.59692179243346e-06, "loss": 0.2879, "step": 9238 }, { "epoch": 0.46734786787394406, "grad_norm": 13.931161880493164, "learning_rate": 5.596124910714005e-06, "loss": 0.5482, "step": 9239 }, { "epoch": 0.4673984521220092, "grad_norm": 11.951936721801758, "learning_rate": 5.5953280136338495e-06, "loss": 0.4473, "step": 9240 }, { "epoch": 0.46744903637007434, "grad_norm": 10.155652046203613, "learning_rate": 5.59453110121353e-06, "loss": 0.4695, "step": 9241 }, { "epoch": 0.4674996206181395, "grad_norm": 12.883792877197266, "learning_rate": 5.593734173473579e-06, "loss": 0.4113, "step": 9242 }, { "epoch": 0.4675502048662047, "grad_norm": 12.855264663696289, "learning_rate": 5.592937230434534e-06, "loss": 0.5968, "step": 9243 }, { "epoch": 0.46760078911426983, "grad_norm": 9.499349594116211, "learning_rate": 5.592140272116927e-06, "loss": 0.485, "step": 9244 }, { "epoch": 0.467651373362335, "grad_norm": 8.267281532287598, "learning_rate": 5.591343298541297e-06, "loss": 0.3115, "step": 9245 }, { "epoch": 0.4677019576104001, "grad_norm": 9.318991661071777, "learning_rate": 5.5905463097281775e-06, "loss": 0.5686, "step": 9246 }, { "epoch": 0.46775254185846527, "grad_norm": 8.647818565368652, "learning_rate": 5.589749305698106e-06, "loss": 0.3195, "step": 9247 }, { "epoch": 0.4678031261065304, "grad_norm": 19.379030227661133, "learning_rate": 5.5889522864716204e-06, "loss": 0.5803, "step": 9248 }, { "epoch": 0.46785371035459555, "grad_norm": 5.705512046813965, "learning_rate": 5.588155252069256e-06, "loss": 0.3667, "step": 9249 }, { "epoch": 0.46790429460266075, "grad_norm": 6.577765941619873, "learning_rate": 5.587358202511551e-06, "loss": 0.3562, "step": 9250 }, { "epoch": 0.4679548788507259, "grad_norm": 11.930070877075195, "learning_rate": 5.586561137819044e-06, "loss": 0.4965, "step": 9251 }, { "epoch": 0.46800546309879104, "grad_norm": 9.789021492004395, "learning_rate": 5.585764058012275e-06, "loss": 0.328, "step": 9252 }, { "epoch": 0.4680560473468562, "grad_norm": 5.924046516418457, "learning_rate": 5.58496696311178e-06, "loss": 0.3482, "step": 9253 }, { "epoch": 0.46810663159492133, "grad_norm": 8.809767723083496, "learning_rate": 5.584169853138099e-06, "loss": 0.3661, "step": 9254 }, { "epoch": 0.4681572158429865, "grad_norm": 14.935234069824219, "learning_rate": 5.5833727281117736e-06, "loss": 0.5521, "step": 9255 }, { "epoch": 0.4682078000910517, "grad_norm": 10.980951309204102, "learning_rate": 5.582575588053339e-06, "loss": 0.5334, "step": 9256 }, { "epoch": 0.4682583843391168, "grad_norm": 5.749450206756592, "learning_rate": 5.58177843298334e-06, "loss": 0.3604, "step": 9257 }, { "epoch": 0.46830896858718196, "grad_norm": 8.972342491149902, "learning_rate": 5.580981262922317e-06, "loss": 0.4602, "step": 9258 }, { "epoch": 0.4683595528352471, "grad_norm": 10.833308219909668, "learning_rate": 5.5801840778908086e-06, "loss": 0.4877, "step": 9259 }, { "epoch": 0.46841013708331225, "grad_norm": 4.569462299346924, "learning_rate": 5.579386877909359e-06, "loss": 0.3173, "step": 9260 }, { "epoch": 0.4684607213313774, "grad_norm": 7.137882709503174, "learning_rate": 5.578589662998508e-06, "loss": 0.415, "step": 9261 }, { "epoch": 0.46851130557944254, "grad_norm": 9.78503131866455, "learning_rate": 5.5777924331788e-06, "loss": 0.5091, "step": 9262 }, { "epoch": 0.46856188982750774, "grad_norm": 5.349451065063477, "learning_rate": 5.576995188470776e-06, "loss": 0.3901, "step": 9263 }, { "epoch": 0.4686124740755729, "grad_norm": 6.414158344268799, "learning_rate": 5.5761979288949794e-06, "loss": 0.3664, "step": 9264 }, { "epoch": 0.468663058323638, "grad_norm": 7.841503143310547, "learning_rate": 5.575400654471955e-06, "loss": 0.3481, "step": 9265 }, { "epoch": 0.46871364257170317, "grad_norm": 4.688436031341553, "learning_rate": 5.574603365222244e-06, "loss": 0.4081, "step": 9266 }, { "epoch": 0.4687642268197683, "grad_norm": 8.713043212890625, "learning_rate": 5.5738060611663934e-06, "loss": 0.3685, "step": 9267 }, { "epoch": 0.46881481106783346, "grad_norm": 9.667647361755371, "learning_rate": 5.573008742324947e-06, "loss": 0.48, "step": 9268 }, { "epoch": 0.4688653953158986, "grad_norm": 8.932868957519531, "learning_rate": 5.572211408718449e-06, "loss": 0.3596, "step": 9269 }, { "epoch": 0.4689159795639638, "grad_norm": 7.146431922912598, "learning_rate": 5.571414060367444e-06, "loss": 0.3886, "step": 9270 }, { "epoch": 0.46896656381202895, "grad_norm": 10.87608528137207, "learning_rate": 5.570616697292479e-06, "loss": 0.4756, "step": 9271 }, { "epoch": 0.4690171480600941, "grad_norm": 11.64745807647705, "learning_rate": 5.569819319514101e-06, "loss": 0.4786, "step": 9272 }, { "epoch": 0.46906773230815924, "grad_norm": 11.8764009475708, "learning_rate": 5.569021927052856e-06, "loss": 0.4759, "step": 9273 }, { "epoch": 0.4691183165562244, "grad_norm": 9.584053993225098, "learning_rate": 5.568224519929289e-06, "loss": 0.3382, "step": 9274 }, { "epoch": 0.4691689008042895, "grad_norm": 7.694417476654053, "learning_rate": 5.567427098163949e-06, "loss": 0.3902, "step": 9275 }, { "epoch": 0.4692194850523547, "grad_norm": 11.431159019470215, "learning_rate": 5.566629661777383e-06, "loss": 0.5124, "step": 9276 }, { "epoch": 0.46927006930041987, "grad_norm": 13.40822696685791, "learning_rate": 5.565832210790139e-06, "loss": 0.4715, "step": 9277 }, { "epoch": 0.469320653548485, "grad_norm": 3.7897045612335205, "learning_rate": 5.565034745222766e-06, "loss": 0.3675, "step": 9278 }, { "epoch": 0.46937123779655016, "grad_norm": 11.300882339477539, "learning_rate": 5.564237265095812e-06, "loss": 0.5242, "step": 9279 }, { "epoch": 0.4694218220446153, "grad_norm": 8.326640129089355, "learning_rate": 5.563439770429829e-06, "loss": 0.398, "step": 9280 }, { "epoch": 0.46947240629268044, "grad_norm": 5.056576251983643, "learning_rate": 5.5626422612453615e-06, "loss": 0.3919, "step": 9281 }, { "epoch": 0.4695229905407456, "grad_norm": 10.249526977539062, "learning_rate": 5.5618447375629625e-06, "loss": 0.488, "step": 9282 }, { "epoch": 0.4695735747888108, "grad_norm": 6.8069610595703125, "learning_rate": 5.561047199403182e-06, "loss": 0.3567, "step": 9283 }, { "epoch": 0.46962415903687593, "grad_norm": 12.172991752624512, "learning_rate": 5.56024964678657e-06, "loss": 0.4811, "step": 9284 }, { "epoch": 0.4696747432849411, "grad_norm": 5.671186447143555, "learning_rate": 5.559452079733677e-06, "loss": 0.399, "step": 9285 }, { "epoch": 0.4697253275330062, "grad_norm": 8.639252662658691, "learning_rate": 5.558654498265057e-06, "loss": 0.523, "step": 9286 }, { "epoch": 0.46977591178107136, "grad_norm": 11.570760726928711, "learning_rate": 5.557856902401259e-06, "loss": 0.5126, "step": 9287 }, { "epoch": 0.4698264960291365, "grad_norm": 6.870095729827881, "learning_rate": 5.557059292162835e-06, "loss": 0.3415, "step": 9288 }, { "epoch": 0.46987708027720165, "grad_norm": 8.216075897216797, "learning_rate": 5.55626166757034e-06, "loss": 0.3817, "step": 9289 }, { "epoch": 0.46992766452526685, "grad_norm": 5.046544075012207, "learning_rate": 5.555464028644326e-06, "loss": 0.3861, "step": 9290 }, { "epoch": 0.469978248773332, "grad_norm": 6.912389278411865, "learning_rate": 5.554666375405344e-06, "loss": 0.4006, "step": 9291 }, { "epoch": 0.47002883302139714, "grad_norm": 7.965055465698242, "learning_rate": 5.55386870787395e-06, "loss": 0.3754, "step": 9292 }, { "epoch": 0.4700794172694623, "grad_norm": 7.615141868591309, "learning_rate": 5.553071026070698e-06, "loss": 0.4124, "step": 9293 }, { "epoch": 0.47013000151752743, "grad_norm": 9.436665534973145, "learning_rate": 5.55227333001614e-06, "loss": 0.4738, "step": 9294 }, { "epoch": 0.4701805857655926, "grad_norm": 9.075166702270508, "learning_rate": 5.551475619730833e-06, "loss": 0.424, "step": 9295 }, { "epoch": 0.4702311700136578, "grad_norm": 10.054695129394531, "learning_rate": 5.550677895235332e-06, "loss": 0.4833, "step": 9296 }, { "epoch": 0.4702817542617229, "grad_norm": 6.564140796661377, "learning_rate": 5.549880156550191e-06, "loss": 0.3989, "step": 9297 }, { "epoch": 0.47033233850978806, "grad_norm": 13.649584770202637, "learning_rate": 5.549082403695965e-06, "loss": 0.4284, "step": 9298 }, { "epoch": 0.4703829227578532, "grad_norm": 4.914670467376709, "learning_rate": 5.548284636693214e-06, "loss": 0.3085, "step": 9299 }, { "epoch": 0.47043350700591835, "grad_norm": 7.09957218170166, "learning_rate": 5.547486855562493e-06, "loss": 0.4185, "step": 9300 }, { "epoch": 0.4704840912539835, "grad_norm": 8.171682357788086, "learning_rate": 5.546689060324356e-06, "loss": 0.3276, "step": 9301 }, { "epoch": 0.47053467550204864, "grad_norm": 6.8908185958862305, "learning_rate": 5.545891250999363e-06, "loss": 0.4442, "step": 9302 }, { "epoch": 0.47058525975011384, "grad_norm": 7.791441440582275, "learning_rate": 5.545093427608072e-06, "loss": 0.3843, "step": 9303 }, { "epoch": 0.470635843998179, "grad_norm": 5.209360599517822, "learning_rate": 5.544295590171038e-06, "loss": 0.4623, "step": 9304 }, { "epoch": 0.4706864282462441, "grad_norm": 6.499004364013672, "learning_rate": 5.543497738708823e-06, "loss": 0.3594, "step": 9305 }, { "epoch": 0.47073701249430927, "grad_norm": 4.381340026855469, "learning_rate": 5.542699873241983e-06, "loss": 0.3621, "step": 9306 }, { "epoch": 0.4707875967423744, "grad_norm": 8.341520309448242, "learning_rate": 5.541901993791079e-06, "loss": 0.4816, "step": 9307 }, { "epoch": 0.47083818099043956, "grad_norm": 7.2083635330200195, "learning_rate": 5.541104100376668e-06, "loss": 0.3753, "step": 9308 }, { "epoch": 0.4708887652385047, "grad_norm": 4.772383689880371, "learning_rate": 5.540306193019313e-06, "loss": 0.3648, "step": 9309 }, { "epoch": 0.4709393494865699, "grad_norm": 6.90474271774292, "learning_rate": 5.539508271739572e-06, "loss": 0.4029, "step": 9310 }, { "epoch": 0.47098993373463505, "grad_norm": 15.823171615600586, "learning_rate": 5.538710336558005e-06, "loss": 0.4662, "step": 9311 }, { "epoch": 0.4710405179827002, "grad_norm": 8.497941017150879, "learning_rate": 5.537912387495176e-06, "loss": 0.3483, "step": 9312 }, { "epoch": 0.47109110223076534, "grad_norm": 10.489697456359863, "learning_rate": 5.537114424571643e-06, "loss": 0.3572, "step": 9313 }, { "epoch": 0.4711416864788305, "grad_norm": 5.9379496574401855, "learning_rate": 5.536316447807968e-06, "loss": 0.4448, "step": 9314 }, { "epoch": 0.4711922707268956, "grad_norm": 7.709604263305664, "learning_rate": 5.535518457224715e-06, "loss": 0.354, "step": 9315 }, { "epoch": 0.4712428549749608, "grad_norm": 5.518184661865234, "learning_rate": 5.534720452842443e-06, "loss": 0.4543, "step": 9316 }, { "epoch": 0.47129343922302597, "grad_norm": 16.00798797607422, "learning_rate": 5.533922434681717e-06, "loss": 0.4784, "step": 9317 }, { "epoch": 0.4713440234710911, "grad_norm": 7.069802284240723, "learning_rate": 5.533124402763099e-06, "loss": 0.356, "step": 9318 }, { "epoch": 0.47139460771915626, "grad_norm": 8.229570388793945, "learning_rate": 5.532326357107155e-06, "loss": 0.4479, "step": 9319 }, { "epoch": 0.4714451919672214, "grad_norm": 6.573089122772217, "learning_rate": 5.5315282977344444e-06, "loss": 0.3315, "step": 9320 }, { "epoch": 0.47149577621528654, "grad_norm": 5.464256286621094, "learning_rate": 5.530730224665535e-06, "loss": 0.4819, "step": 9321 }, { "epoch": 0.4715463604633517, "grad_norm": 5.208344459533691, "learning_rate": 5.5299321379209884e-06, "loss": 0.2622, "step": 9322 }, { "epoch": 0.4715969447114169, "grad_norm": 9.437644958496094, "learning_rate": 5.52913403752137e-06, "loss": 0.5406, "step": 9323 }, { "epoch": 0.47164752895948203, "grad_norm": 6.051208019256592, "learning_rate": 5.528335923487247e-06, "loss": 0.3612, "step": 9324 }, { "epoch": 0.4716981132075472, "grad_norm": 12.56452465057373, "learning_rate": 5.527537795839184e-06, "loss": 0.4142, "step": 9325 }, { "epoch": 0.4717486974556123, "grad_norm": 8.937786102294922, "learning_rate": 5.526739654597744e-06, "loss": 0.5208, "step": 9326 }, { "epoch": 0.47179928170367746, "grad_norm": 8.056893348693848, "learning_rate": 5.525941499783497e-06, "loss": 0.4802, "step": 9327 }, { "epoch": 0.4718498659517426, "grad_norm": 15.217864036560059, "learning_rate": 5.5251433314170075e-06, "loss": 0.5855, "step": 9328 }, { "epoch": 0.47190045019980775, "grad_norm": 12.464061737060547, "learning_rate": 5.524345149518842e-06, "loss": 0.4908, "step": 9329 }, { "epoch": 0.47195103444787295, "grad_norm": 9.48903751373291, "learning_rate": 5.523546954109568e-06, "loss": 0.4336, "step": 9330 }, { "epoch": 0.4720016186959381, "grad_norm": 6.316868782043457, "learning_rate": 5.522748745209757e-06, "loss": 0.3107, "step": 9331 }, { "epoch": 0.47205220294400324, "grad_norm": 6.5984344482421875, "learning_rate": 5.521950522839972e-06, "loss": 0.3975, "step": 9332 }, { "epoch": 0.4721027871920684, "grad_norm": 6.035543918609619, "learning_rate": 5.521152287020781e-06, "loss": 0.3195, "step": 9333 }, { "epoch": 0.47215337144013353, "grad_norm": 10.607474327087402, "learning_rate": 5.520354037772756e-06, "loss": 0.4627, "step": 9334 }, { "epoch": 0.4722039556881987, "grad_norm": 14.464516639709473, "learning_rate": 5.519555775116463e-06, "loss": 0.434, "step": 9335 }, { "epoch": 0.4722545399362639, "grad_norm": 8.98386001586914, "learning_rate": 5.518757499072473e-06, "loss": 0.443, "step": 9336 }, { "epoch": 0.472305124184329, "grad_norm": 9.184977531433105, "learning_rate": 5.517959209661356e-06, "loss": 0.381, "step": 9337 }, { "epoch": 0.47235570843239416, "grad_norm": 22.95295524597168, "learning_rate": 5.517160906903681e-06, "loss": 0.7169, "step": 9338 }, { "epoch": 0.4724062926804593, "grad_norm": 6.886983871459961, "learning_rate": 5.516362590820018e-06, "loss": 0.4309, "step": 9339 }, { "epoch": 0.47245687692852445, "grad_norm": 10.178362846374512, "learning_rate": 5.515564261430939e-06, "loss": 0.414, "step": 9340 }, { "epoch": 0.4725074611765896, "grad_norm": 4.477383613586426, "learning_rate": 5.514765918757013e-06, "loss": 0.2724, "step": 9341 }, { "epoch": 0.47255804542465474, "grad_norm": 4.642017841339111, "learning_rate": 5.513967562818814e-06, "loss": 0.3685, "step": 9342 }, { "epoch": 0.47260862967271994, "grad_norm": 5.829061031341553, "learning_rate": 5.513169193636912e-06, "loss": 0.3535, "step": 9343 }, { "epoch": 0.4726592139207851, "grad_norm": 11.961609840393066, "learning_rate": 5.51237081123188e-06, "loss": 0.401, "step": 9344 }, { "epoch": 0.4727097981688502, "grad_norm": 4.547976016998291, "learning_rate": 5.511572415624289e-06, "loss": 0.3879, "step": 9345 }, { "epoch": 0.47276038241691537, "grad_norm": 12.683592796325684, "learning_rate": 5.510774006834712e-06, "loss": 0.4785, "step": 9346 }, { "epoch": 0.4728109666649805, "grad_norm": 4.7210588455200195, "learning_rate": 5.509975584883723e-06, "loss": 0.3783, "step": 9347 }, { "epoch": 0.47286155091304566, "grad_norm": 6.4557414054870605, "learning_rate": 5.509177149791896e-06, "loss": 0.4019, "step": 9348 }, { "epoch": 0.4729121351611108, "grad_norm": 10.970124244689941, "learning_rate": 5.508378701579802e-06, "loss": 0.4972, "step": 9349 }, { "epoch": 0.472962719409176, "grad_norm": 7.370970726013184, "learning_rate": 5.507580240268019e-06, "loss": 0.4142, "step": 9350 }, { "epoch": 0.47301330365724115, "grad_norm": 4.988685131072998, "learning_rate": 5.506781765877117e-06, "loss": 0.3857, "step": 9351 }, { "epoch": 0.4730638879053063, "grad_norm": 9.718592643737793, "learning_rate": 5.505983278427674e-06, "loss": 0.4411, "step": 9352 }, { "epoch": 0.47311447215337143, "grad_norm": 5.720420837402344, "learning_rate": 5.5051847779402646e-06, "loss": 0.3583, "step": 9353 }, { "epoch": 0.4731650564014366, "grad_norm": 6.423440933227539, "learning_rate": 5.504386264435462e-06, "loss": 0.4053, "step": 9354 }, { "epoch": 0.4732156406495017, "grad_norm": 7.866305828094482, "learning_rate": 5.5035877379338445e-06, "loss": 0.4542, "step": 9355 }, { "epoch": 0.4732662248975669, "grad_norm": 8.800623893737793, "learning_rate": 5.5027891984559875e-06, "loss": 0.3848, "step": 9356 }, { "epoch": 0.47331680914563207, "grad_norm": 11.389886856079102, "learning_rate": 5.501990646022468e-06, "loss": 0.4887, "step": 9357 }, { "epoch": 0.4733673933936972, "grad_norm": 13.500816345214844, "learning_rate": 5.501192080653861e-06, "loss": 0.5486, "step": 9358 }, { "epoch": 0.47341797764176236, "grad_norm": 6.878659248352051, "learning_rate": 5.500393502370746e-06, "loss": 0.3943, "step": 9359 }, { "epoch": 0.4734685618898275, "grad_norm": 7.221370220184326, "learning_rate": 5.4995949111936975e-06, "loss": 0.4901, "step": 9360 }, { "epoch": 0.47351914613789264, "grad_norm": 5.461787700653076, "learning_rate": 5.498796307143295e-06, "loss": 0.2582, "step": 9361 }, { "epoch": 0.4735697303859578, "grad_norm": 9.624372482299805, "learning_rate": 5.497997690240118e-06, "loss": 0.4225, "step": 9362 }, { "epoch": 0.473620314634023, "grad_norm": 4.678198337554932, "learning_rate": 5.497199060504742e-06, "loss": 0.2733, "step": 9363 }, { "epoch": 0.47367089888208813, "grad_norm": 5.080893039703369, "learning_rate": 5.496400417957747e-06, "loss": 0.4443, "step": 9364 }, { "epoch": 0.4737214831301533, "grad_norm": 7.7433013916015625, "learning_rate": 5.495601762619712e-06, "loss": 0.4126, "step": 9365 }, { "epoch": 0.4737720673782184, "grad_norm": 11.06888198852539, "learning_rate": 5.494803094511218e-06, "loss": 0.5217, "step": 9366 }, { "epoch": 0.47382265162628356, "grad_norm": 10.02381420135498, "learning_rate": 5.494004413652842e-06, "loss": 0.4699, "step": 9367 }, { "epoch": 0.4738732358743487, "grad_norm": 5.892396450042725, "learning_rate": 5.493205720065167e-06, "loss": 0.3491, "step": 9368 }, { "epoch": 0.4739238201224139, "grad_norm": 11.156192779541016, "learning_rate": 5.492407013768771e-06, "loss": 0.5308, "step": 9369 }, { "epoch": 0.47397440437047905, "grad_norm": 15.698452949523926, "learning_rate": 5.491608294784236e-06, "loss": 0.5178, "step": 9370 }, { "epoch": 0.4740249886185442, "grad_norm": 11.671287536621094, "learning_rate": 5.490809563132142e-06, "loss": 0.4804, "step": 9371 }, { "epoch": 0.47407557286660934, "grad_norm": 5.075978755950928, "learning_rate": 5.490010818833072e-06, "loss": 0.364, "step": 9372 }, { "epoch": 0.4741261571146745, "grad_norm": 7.205933570861816, "learning_rate": 5.489212061907606e-06, "loss": 0.3538, "step": 9373 }, { "epoch": 0.47417674136273963, "grad_norm": 6.635351181030273, "learning_rate": 5.488413292376327e-06, "loss": 0.3876, "step": 9374 }, { "epoch": 0.4742273256108048, "grad_norm": 6.617615699768066, "learning_rate": 5.487614510259818e-06, "loss": 0.3972, "step": 9375 }, { "epoch": 0.47427790985887, "grad_norm": 6.547781944274902, "learning_rate": 5.486815715578661e-06, "loss": 0.402, "step": 9376 }, { "epoch": 0.4743284941069351, "grad_norm": 8.755941390991211, "learning_rate": 5.486016908353437e-06, "loss": 0.49, "step": 9377 }, { "epoch": 0.47437907835500026, "grad_norm": 7.096338748931885, "learning_rate": 5.485218088604732e-06, "loss": 0.4254, "step": 9378 }, { "epoch": 0.4744296626030654, "grad_norm": 5.882934093475342, "learning_rate": 5.484419256353129e-06, "loss": 0.3406, "step": 9379 }, { "epoch": 0.47448024685113055, "grad_norm": 7.068517208099365, "learning_rate": 5.4836204116192115e-06, "loss": 0.4257, "step": 9380 }, { "epoch": 0.4745308310991957, "grad_norm": 6.3833842277526855, "learning_rate": 5.482821554423565e-06, "loss": 0.4191, "step": 9381 }, { "epoch": 0.47458141534726084, "grad_norm": 5.266036510467529, "learning_rate": 5.482022684786773e-06, "loss": 0.3928, "step": 9382 }, { "epoch": 0.47463199959532604, "grad_norm": 9.698126792907715, "learning_rate": 5.48122380272942e-06, "loss": 0.4737, "step": 9383 }, { "epoch": 0.4746825838433912, "grad_norm": 5.57393217086792, "learning_rate": 5.480424908272092e-06, "loss": 0.3853, "step": 9384 }, { "epoch": 0.4747331680914563, "grad_norm": 11.631593704223633, "learning_rate": 5.4796260014353756e-06, "loss": 0.3828, "step": 9385 }, { "epoch": 0.47478375233952147, "grad_norm": 9.606111526489258, "learning_rate": 5.478827082239854e-06, "loss": 0.3501, "step": 9386 }, { "epoch": 0.4748343365875866, "grad_norm": 7.185877799987793, "learning_rate": 5.478028150706116e-06, "loss": 0.3409, "step": 9387 }, { "epoch": 0.47488492083565176, "grad_norm": 7.234045505523682, "learning_rate": 5.477229206854747e-06, "loss": 0.4164, "step": 9388 }, { "epoch": 0.47493550508371696, "grad_norm": 9.30616283416748, "learning_rate": 5.476430250706335e-06, "loss": 0.4727, "step": 9389 }, { "epoch": 0.4749860893317821, "grad_norm": 6.144620895385742, "learning_rate": 5.475631282281465e-06, "loss": 0.3468, "step": 9390 }, { "epoch": 0.47503667357984725, "grad_norm": 6.4669718742370605, "learning_rate": 5.474832301600727e-06, "loss": 0.4561, "step": 9391 }, { "epoch": 0.4750872578279124, "grad_norm": 4.089354991912842, "learning_rate": 5.474033308684706e-06, "loss": 0.2231, "step": 9392 }, { "epoch": 0.47513784207597753, "grad_norm": 7.137676239013672, "learning_rate": 5.473234303553992e-06, "loss": 0.3974, "step": 9393 }, { "epoch": 0.4751884263240427, "grad_norm": 10.021272659301758, "learning_rate": 5.472435286229173e-06, "loss": 0.5588, "step": 9394 }, { "epoch": 0.4752390105721078, "grad_norm": 7.601344108581543, "learning_rate": 5.471636256730839e-06, "loss": 0.359, "step": 9395 }, { "epoch": 0.475289594820173, "grad_norm": 22.558446884155273, "learning_rate": 5.470837215079576e-06, "loss": 0.7357, "step": 9396 }, { "epoch": 0.47534017906823817, "grad_norm": 4.998476028442383, "learning_rate": 5.470038161295977e-06, "loss": 0.358, "step": 9397 }, { "epoch": 0.4753907633163033, "grad_norm": 6.826851844787598, "learning_rate": 5.469239095400628e-06, "loss": 0.4011, "step": 9398 }, { "epoch": 0.47544134756436846, "grad_norm": 6.011818885803223, "learning_rate": 5.468440017414122e-06, "loss": 0.2919, "step": 9399 }, { "epoch": 0.4754919318124336, "grad_norm": 7.203032493591309, "learning_rate": 5.467640927357049e-06, "loss": 0.3774, "step": 9400 }, { "epoch": 0.47554251606049874, "grad_norm": 7.641147136688232, "learning_rate": 5.466841825249998e-06, "loss": 0.5445, "step": 9401 }, { "epoch": 0.4755931003085639, "grad_norm": 5.334925651550293, "learning_rate": 5.46604271111356e-06, "loss": 0.2985, "step": 9402 }, { "epoch": 0.4756436845566291, "grad_norm": 14.06058120727539, "learning_rate": 5.465243584968328e-06, "loss": 0.5367, "step": 9403 }, { "epoch": 0.47569426880469423, "grad_norm": 5.618505477905273, "learning_rate": 5.4644444468348935e-06, "loss": 0.3012, "step": 9404 }, { "epoch": 0.4757448530527594, "grad_norm": 14.039566993713379, "learning_rate": 5.463645296733845e-06, "loss": 0.4984, "step": 9405 }, { "epoch": 0.4757954373008245, "grad_norm": 11.976394653320312, "learning_rate": 5.462846134685779e-06, "loss": 0.6338, "step": 9406 }, { "epoch": 0.47584602154888966, "grad_norm": 6.8651533126831055, "learning_rate": 5.4620469607112845e-06, "loss": 0.4106, "step": 9407 }, { "epoch": 0.4758966057969548, "grad_norm": 5.544517517089844, "learning_rate": 5.461247774830958e-06, "loss": 0.4733, "step": 9408 }, { "epoch": 0.47594719004502, "grad_norm": 9.510822296142578, "learning_rate": 5.460448577065389e-06, "loss": 0.5018, "step": 9409 }, { "epoch": 0.47599777429308515, "grad_norm": 7.14143180847168, "learning_rate": 5.459649367435173e-06, "loss": 0.4086, "step": 9410 }, { "epoch": 0.4760483585411503, "grad_norm": 5.533629894256592, "learning_rate": 5.458850145960902e-06, "loss": 0.3009, "step": 9411 }, { "epoch": 0.47609894278921544, "grad_norm": 12.039067268371582, "learning_rate": 5.458050912663171e-06, "loss": 0.3073, "step": 9412 }, { "epoch": 0.4761495270372806, "grad_norm": 9.453900337219238, "learning_rate": 5.457251667562577e-06, "loss": 0.464, "step": 9413 }, { "epoch": 0.47620011128534573, "grad_norm": 8.612677574157715, "learning_rate": 5.456452410679708e-06, "loss": 0.3755, "step": 9414 }, { "epoch": 0.4762506955334109, "grad_norm": 4.964635372161865, "learning_rate": 5.455653142035165e-06, "loss": 0.386, "step": 9415 }, { "epoch": 0.4763012797814761, "grad_norm": 8.560416221618652, "learning_rate": 5.4548538616495424e-06, "loss": 0.3434, "step": 9416 }, { "epoch": 0.4763518640295412, "grad_norm": 7.1044464111328125, "learning_rate": 5.4540545695434325e-06, "loss": 0.4072, "step": 9417 }, { "epoch": 0.47640244827760636, "grad_norm": 11.252508163452148, "learning_rate": 5.4532552657374334e-06, "loss": 0.5499, "step": 9418 }, { "epoch": 0.4764530325256715, "grad_norm": 8.484526634216309, "learning_rate": 5.452455950252141e-06, "loss": 0.4848, "step": 9419 }, { "epoch": 0.47650361677373665, "grad_norm": 4.995439052581787, "learning_rate": 5.451656623108151e-06, "loss": 0.3323, "step": 9420 }, { "epoch": 0.4765542010218018, "grad_norm": 8.449027061462402, "learning_rate": 5.450857284326063e-06, "loss": 0.4846, "step": 9421 }, { "epoch": 0.47660478526986694, "grad_norm": 10.238606452941895, "learning_rate": 5.450057933926469e-06, "loss": 0.4785, "step": 9422 }, { "epoch": 0.47665536951793214, "grad_norm": 4.949388027191162, "learning_rate": 5.449258571929972e-06, "loss": 0.3441, "step": 9423 }, { "epoch": 0.4767059537659973, "grad_norm": 7.4407196044921875, "learning_rate": 5.448459198357164e-06, "loss": 0.4362, "step": 9424 }, { "epoch": 0.4767565380140624, "grad_norm": 6.951644420623779, "learning_rate": 5.447659813228647e-06, "loss": 0.3574, "step": 9425 }, { "epoch": 0.47680712226212757, "grad_norm": 8.506781578063965, "learning_rate": 5.4468604165650174e-06, "loss": 0.3905, "step": 9426 }, { "epoch": 0.4768577065101927, "grad_norm": 7.612555503845215, "learning_rate": 5.446061008386876e-06, "loss": 0.4343, "step": 9427 }, { "epoch": 0.47690829075825786, "grad_norm": 43.140132904052734, "learning_rate": 5.445261588714817e-06, "loss": 0.5156, "step": 9428 }, { "epoch": 0.47695887500632306, "grad_norm": 7.427308559417725, "learning_rate": 5.444462157569444e-06, "loss": 0.3843, "step": 9429 }, { "epoch": 0.4770094592543882, "grad_norm": 10.25228214263916, "learning_rate": 5.443662714971355e-06, "loss": 0.4365, "step": 9430 }, { "epoch": 0.47706004350245335, "grad_norm": 11.099189758300781, "learning_rate": 5.442863260941148e-06, "loss": 0.5378, "step": 9431 }, { "epoch": 0.4771106277505185, "grad_norm": 5.147552013397217, "learning_rate": 5.442063795499426e-06, "loss": 0.4381, "step": 9432 }, { "epoch": 0.47716121199858363, "grad_norm": 8.65169620513916, "learning_rate": 5.441264318666787e-06, "loss": 0.4382, "step": 9433 }, { "epoch": 0.4772117962466488, "grad_norm": 7.149503231048584, "learning_rate": 5.440464830463833e-06, "loss": 0.4201, "step": 9434 }, { "epoch": 0.4772623804947139, "grad_norm": 8.03895378112793, "learning_rate": 5.4396653309111635e-06, "loss": 0.3532, "step": 9435 }, { "epoch": 0.4773129647427791, "grad_norm": 6.022177219390869, "learning_rate": 5.43886582002938e-06, "loss": 0.3883, "step": 9436 }, { "epoch": 0.47736354899084427, "grad_norm": 9.034889221191406, "learning_rate": 5.438066297839084e-06, "loss": 0.3686, "step": 9437 }, { "epoch": 0.4774141332389094, "grad_norm": 6.330482006072998, "learning_rate": 5.43726676436088e-06, "loss": 0.4196, "step": 9438 }, { "epoch": 0.47746471748697455, "grad_norm": 7.029146671295166, "learning_rate": 5.436467219615365e-06, "loss": 0.3709, "step": 9439 }, { "epoch": 0.4775153017350397, "grad_norm": 10.378159523010254, "learning_rate": 5.4356676636231465e-06, "loss": 0.4191, "step": 9440 }, { "epoch": 0.47756588598310484, "grad_norm": 6.693603038787842, "learning_rate": 5.434868096404823e-06, "loss": 0.4243, "step": 9441 }, { "epoch": 0.47761647023117, "grad_norm": 5.593337059020996, "learning_rate": 5.434068517980999e-06, "loss": 0.3116, "step": 9442 }, { "epoch": 0.4776670544792352, "grad_norm": 6.7765116691589355, "learning_rate": 5.433268928372279e-06, "loss": 0.3964, "step": 9443 }, { "epoch": 0.47771763872730033, "grad_norm": 9.160358428955078, "learning_rate": 5.432469327599263e-06, "loss": 0.3987, "step": 9444 }, { "epoch": 0.4777682229753655, "grad_norm": 10.802450180053711, "learning_rate": 5.431669715682558e-06, "loss": 0.4931, "step": 9445 }, { "epoch": 0.4778188072234306, "grad_norm": 10.101133346557617, "learning_rate": 5.430870092642767e-06, "loss": 0.5097, "step": 9446 }, { "epoch": 0.47786939147149576, "grad_norm": 7.533033847808838, "learning_rate": 5.430070458500496e-06, "loss": 0.4491, "step": 9447 }, { "epoch": 0.4779199757195609, "grad_norm": 8.442566871643066, "learning_rate": 5.429270813276347e-06, "loss": 0.4008, "step": 9448 }, { "epoch": 0.4779705599676261, "grad_norm": 9.404520034790039, "learning_rate": 5.428471156990926e-06, "loss": 0.4482, "step": 9449 }, { "epoch": 0.47802114421569125, "grad_norm": 8.579279899597168, "learning_rate": 5.427671489664838e-06, "loss": 0.3892, "step": 9450 }, { "epoch": 0.4780717284637564, "grad_norm": 5.98258113861084, "learning_rate": 5.426871811318689e-06, "loss": 0.3886, "step": 9451 }, { "epoch": 0.47812231271182154, "grad_norm": 13.578699111938477, "learning_rate": 5.426072121973084e-06, "loss": 0.5681, "step": 9452 }, { "epoch": 0.4781728969598867, "grad_norm": 9.471414566040039, "learning_rate": 5.425272421648631e-06, "loss": 0.4427, "step": 9453 }, { "epoch": 0.47822348120795183, "grad_norm": 16.197755813598633, "learning_rate": 5.424472710365933e-06, "loss": 0.3242, "step": 9454 }, { "epoch": 0.478274065456017, "grad_norm": 8.784417152404785, "learning_rate": 5.4236729881456e-06, "loss": 0.486, "step": 9455 }, { "epoch": 0.47832464970408217, "grad_norm": 11.14900016784668, "learning_rate": 5.422873255008235e-06, "loss": 0.5061, "step": 9456 }, { "epoch": 0.4783752339521473, "grad_norm": 14.167903900146484, "learning_rate": 5.42207351097445e-06, "loss": 0.5323, "step": 9457 }, { "epoch": 0.47842581820021246, "grad_norm": 10.409451484680176, "learning_rate": 5.42127375606485e-06, "loss": 0.5006, "step": 9458 }, { "epoch": 0.4784764024482776, "grad_norm": 9.721658706665039, "learning_rate": 5.420473990300041e-06, "loss": 0.5474, "step": 9459 }, { "epoch": 0.47852698669634275, "grad_norm": 7.298260688781738, "learning_rate": 5.4196742137006355e-06, "loss": 0.4473, "step": 9460 }, { "epoch": 0.4785775709444079, "grad_norm": 8.248444557189941, "learning_rate": 5.418874426287238e-06, "loss": 0.5036, "step": 9461 }, { "epoch": 0.47862815519247304, "grad_norm": 5.924102783203125, "learning_rate": 5.418074628080458e-06, "loss": 0.3329, "step": 9462 }, { "epoch": 0.47867873944053824, "grad_norm": 9.447997093200684, "learning_rate": 5.417274819100905e-06, "loss": 0.4579, "step": 9463 }, { "epoch": 0.4787293236886034, "grad_norm": 9.935853958129883, "learning_rate": 5.416474999369187e-06, "loss": 0.4016, "step": 9464 }, { "epoch": 0.4787799079366685, "grad_norm": 5.787993431091309, "learning_rate": 5.415675168905915e-06, "loss": 0.3664, "step": 9465 }, { "epoch": 0.47883049218473367, "grad_norm": 10.437138557434082, "learning_rate": 5.414875327731698e-06, "loss": 0.3852, "step": 9466 }, { "epoch": 0.4788810764327988, "grad_norm": 9.21010684967041, "learning_rate": 5.414075475867146e-06, "loss": 0.3603, "step": 9467 }, { "epoch": 0.47893166068086396, "grad_norm": 9.592157363891602, "learning_rate": 5.413275613332869e-06, "loss": 0.3968, "step": 9468 }, { "epoch": 0.47898224492892916, "grad_norm": 6.633944034576416, "learning_rate": 5.412475740149477e-06, "loss": 0.4507, "step": 9469 }, { "epoch": 0.4790328291769943, "grad_norm": 5.080099105834961, "learning_rate": 5.4116758563375815e-06, "loss": 0.4041, "step": 9470 }, { "epoch": 0.47908341342505945, "grad_norm": 11.031641960144043, "learning_rate": 5.410875961917795e-06, "loss": 0.4645, "step": 9471 }, { "epoch": 0.4791339976731246, "grad_norm": 9.820277214050293, "learning_rate": 5.410076056910724e-06, "loss": 0.4939, "step": 9472 }, { "epoch": 0.47918458192118973, "grad_norm": 14.2094087600708, "learning_rate": 5.4092761413369884e-06, "loss": 0.4345, "step": 9473 }, { "epoch": 0.4792351661692549, "grad_norm": 5.272684574127197, "learning_rate": 5.408476215217192e-06, "loss": 0.4233, "step": 9474 }, { "epoch": 0.47928575041732, "grad_norm": 8.674515724182129, "learning_rate": 5.40767627857195e-06, "loss": 0.4257, "step": 9475 }, { "epoch": 0.4793363346653852, "grad_norm": 9.53128719329834, "learning_rate": 5.406876331421877e-06, "loss": 0.5001, "step": 9476 }, { "epoch": 0.47938691891345037, "grad_norm": 5.433041095733643, "learning_rate": 5.406076373787582e-06, "loss": 0.4386, "step": 9477 }, { "epoch": 0.4794375031615155, "grad_norm": 4.224812030792236, "learning_rate": 5.40527640568968e-06, "loss": 0.327, "step": 9478 }, { "epoch": 0.47948808740958065, "grad_norm": 7.964758396148682, "learning_rate": 5.404476427148786e-06, "loss": 0.4262, "step": 9479 }, { "epoch": 0.4795386716576458, "grad_norm": 6.2427802085876465, "learning_rate": 5.403676438185509e-06, "loss": 0.3895, "step": 9480 }, { "epoch": 0.47958925590571094, "grad_norm": 7.20656156539917, "learning_rate": 5.402876438820466e-06, "loss": 0.4351, "step": 9481 }, { "epoch": 0.4796398401537761, "grad_norm": 7.9198527336120605, "learning_rate": 5.40207642907427e-06, "loss": 0.5417, "step": 9482 }, { "epoch": 0.4796904244018413, "grad_norm": 11.1668701171875, "learning_rate": 5.4012764089675355e-06, "loss": 0.544, "step": 9483 }, { "epoch": 0.47974100864990643, "grad_norm": 20.245412826538086, "learning_rate": 5.400476378520877e-06, "loss": 0.4241, "step": 9484 }, { "epoch": 0.4797915928979716, "grad_norm": 7.34009313583374, "learning_rate": 5.39967633775491e-06, "loss": 0.5195, "step": 9485 }, { "epoch": 0.4798421771460367, "grad_norm": 7.253830909729004, "learning_rate": 5.39887628669025e-06, "loss": 0.4359, "step": 9486 }, { "epoch": 0.47989276139410186, "grad_norm": 6.815582275390625, "learning_rate": 5.39807622534751e-06, "loss": 0.4313, "step": 9487 }, { "epoch": 0.479943345642167, "grad_norm": 9.724820137023926, "learning_rate": 5.397276153747308e-06, "loss": 0.3947, "step": 9488 }, { "epoch": 0.4799939298902322, "grad_norm": 5.550598621368408, "learning_rate": 5.39647607191026e-06, "loss": 0.2928, "step": 9489 }, { "epoch": 0.48004451413829735, "grad_norm": 6.986380100250244, "learning_rate": 5.395675979856979e-06, "loss": 0.4107, "step": 9490 }, { "epoch": 0.4800950983863625, "grad_norm": 9.202838897705078, "learning_rate": 5.394875877608084e-06, "loss": 0.4367, "step": 9491 }, { "epoch": 0.48014568263442764, "grad_norm": 18.005115509033203, "learning_rate": 5.394075765184195e-06, "loss": 0.4201, "step": 9492 }, { "epoch": 0.4801962668824928, "grad_norm": 5.499213218688965, "learning_rate": 5.393275642605922e-06, "loss": 0.4622, "step": 9493 }, { "epoch": 0.48024685113055793, "grad_norm": 12.932173728942871, "learning_rate": 5.392475509893885e-06, "loss": 0.3584, "step": 9494 }, { "epoch": 0.48029743537862307, "grad_norm": 9.292596817016602, "learning_rate": 5.391675367068704e-06, "loss": 0.4791, "step": 9495 }, { "epoch": 0.48034801962668827, "grad_norm": 6.011493682861328, "learning_rate": 5.390875214150993e-06, "loss": 0.3304, "step": 9496 }, { "epoch": 0.4803986038747534, "grad_norm": 7.911351203918457, "learning_rate": 5.390075051161373e-06, "loss": 0.4493, "step": 9497 }, { "epoch": 0.48044918812281856, "grad_norm": 5.011363506317139, "learning_rate": 5.389274878120464e-06, "loss": 0.3931, "step": 9498 }, { "epoch": 0.4804997723708837, "grad_norm": 17.873294830322266, "learning_rate": 5.3884746950488776e-06, "loss": 0.6065, "step": 9499 }, { "epoch": 0.48055035661894885, "grad_norm": 9.543889045715332, "learning_rate": 5.387674501967237e-06, "loss": 0.3987, "step": 9500 }, { "epoch": 0.480600940867014, "grad_norm": 5.701507091522217, "learning_rate": 5.386874298896163e-06, "loss": 0.4445, "step": 9501 }, { "epoch": 0.48065152511507914, "grad_norm": 5.564374923706055, "learning_rate": 5.3860740858562714e-06, "loss": 0.3553, "step": 9502 }, { "epoch": 0.48070210936314434, "grad_norm": 6.793651103973389, "learning_rate": 5.385273862868182e-06, "loss": 0.4828, "step": 9503 }, { "epoch": 0.4807526936112095, "grad_norm": 6.834071159362793, "learning_rate": 5.384473629952518e-06, "loss": 0.4573, "step": 9504 }, { "epoch": 0.4808032778592746, "grad_norm": 9.002474784851074, "learning_rate": 5.3836733871298975e-06, "loss": 0.3764, "step": 9505 }, { "epoch": 0.48085386210733977, "grad_norm": 11.966269493103027, "learning_rate": 5.382873134420938e-06, "loss": 0.5981, "step": 9506 }, { "epoch": 0.4809044463554049, "grad_norm": 7.143500328063965, "learning_rate": 5.382072871846263e-06, "loss": 0.4548, "step": 9507 }, { "epoch": 0.48095503060347006, "grad_norm": 7.214145183563232, "learning_rate": 5.3812725994264945e-06, "loss": 0.3955, "step": 9508 }, { "epoch": 0.48100561485153526, "grad_norm": 9.526359558105469, "learning_rate": 5.380472317182252e-06, "loss": 0.5476, "step": 9509 }, { "epoch": 0.4810561990996004, "grad_norm": 7.677830696105957, "learning_rate": 5.3796720251341564e-06, "loss": 0.4759, "step": 9510 }, { "epoch": 0.48110678334766555, "grad_norm": 6.26162576675415, "learning_rate": 5.378871723302831e-06, "loss": 0.4627, "step": 9511 }, { "epoch": 0.4811573675957307, "grad_norm": 7.583090305328369, "learning_rate": 5.378071411708894e-06, "loss": 0.4161, "step": 9512 }, { "epoch": 0.48120795184379583, "grad_norm": 7.147919654846191, "learning_rate": 5.377271090372971e-06, "loss": 0.3982, "step": 9513 }, { "epoch": 0.481258536091861, "grad_norm": 6.0684380531311035, "learning_rate": 5.376470759315683e-06, "loss": 0.3521, "step": 9514 }, { "epoch": 0.4813091203399261, "grad_norm": 6.858122825622559, "learning_rate": 5.375670418557653e-06, "loss": 0.3519, "step": 9515 }, { "epoch": 0.4813597045879913, "grad_norm": 5.173088073730469, "learning_rate": 5.374870068119503e-06, "loss": 0.3812, "step": 9516 }, { "epoch": 0.48141028883605647, "grad_norm": 5.050479412078857, "learning_rate": 5.374069708021858e-06, "loss": 0.3854, "step": 9517 }, { "epoch": 0.4814608730841216, "grad_norm": 4.360888957977295, "learning_rate": 5.373269338285341e-06, "loss": 0.4009, "step": 9518 }, { "epoch": 0.48151145733218675, "grad_norm": 6.367681980133057, "learning_rate": 5.3724689589305725e-06, "loss": 0.4379, "step": 9519 }, { "epoch": 0.4815620415802519, "grad_norm": 10.338736534118652, "learning_rate": 5.37166856997818e-06, "loss": 0.5343, "step": 9520 }, { "epoch": 0.48161262582831704, "grad_norm": 6.844425201416016, "learning_rate": 5.370868171448786e-06, "loss": 0.3674, "step": 9521 }, { "epoch": 0.48166321007638224, "grad_norm": 11.337713241577148, "learning_rate": 5.3700677633630134e-06, "loss": 0.423, "step": 9522 }, { "epoch": 0.4817137943244474, "grad_norm": 4.971085071563721, "learning_rate": 5.369267345741491e-06, "loss": 0.4043, "step": 9523 }, { "epoch": 0.48176437857251253, "grad_norm": 8.605067253112793, "learning_rate": 5.368466918604841e-06, "loss": 0.3998, "step": 9524 }, { "epoch": 0.4818149628205777, "grad_norm": 9.17033576965332, "learning_rate": 5.3676664819736865e-06, "loss": 0.4893, "step": 9525 }, { "epoch": 0.4818655470686428, "grad_norm": 6.532507419586182, "learning_rate": 5.366866035868655e-06, "loss": 0.3871, "step": 9526 }, { "epoch": 0.48191613131670796, "grad_norm": 13.145952224731445, "learning_rate": 5.3660655803103735e-06, "loss": 0.5053, "step": 9527 }, { "epoch": 0.4819667155647731, "grad_norm": 5.997602462768555, "learning_rate": 5.365265115319464e-06, "loss": 0.4266, "step": 9528 }, { "epoch": 0.4820172998128383, "grad_norm": 10.210704803466797, "learning_rate": 5.364464640916557e-06, "loss": 0.4026, "step": 9529 }, { "epoch": 0.48206788406090345, "grad_norm": 9.755319595336914, "learning_rate": 5.3636641571222754e-06, "loss": 0.5872, "step": 9530 }, { "epoch": 0.4821184683089686, "grad_norm": 6.280909061431885, "learning_rate": 5.362863663957247e-06, "loss": 0.4037, "step": 9531 }, { "epoch": 0.48216905255703374, "grad_norm": 5.463302135467529, "learning_rate": 5.362063161442098e-06, "loss": 0.3519, "step": 9532 }, { "epoch": 0.4822196368050989, "grad_norm": 5.044101238250732, "learning_rate": 5.361262649597457e-06, "loss": 0.4519, "step": 9533 }, { "epoch": 0.482270221053164, "grad_norm": 19.150346755981445, "learning_rate": 5.36046212844395e-06, "loss": 0.5448, "step": 9534 }, { "epoch": 0.48232080530122917, "grad_norm": 5.720119953155518, "learning_rate": 5.359661598002203e-06, "loss": 0.3915, "step": 9535 }, { "epoch": 0.48237138954929437, "grad_norm": 7.305422306060791, "learning_rate": 5.358861058292847e-06, "loss": 0.4432, "step": 9536 }, { "epoch": 0.4824219737973595, "grad_norm": 7.185011386871338, "learning_rate": 5.358060509336508e-06, "loss": 0.4406, "step": 9537 }, { "epoch": 0.48247255804542466, "grad_norm": 10.269646644592285, "learning_rate": 5.357259951153814e-06, "loss": 0.4228, "step": 9538 }, { "epoch": 0.4825231422934898, "grad_norm": 6.769292831420898, "learning_rate": 5.356459383765394e-06, "loss": 0.4027, "step": 9539 }, { "epoch": 0.48257372654155495, "grad_norm": 5.564431190490723, "learning_rate": 5.355658807191877e-06, "loss": 0.3961, "step": 9540 }, { "epoch": 0.4826243107896201, "grad_norm": 8.859281539916992, "learning_rate": 5.35485822145389e-06, "loss": 0.4111, "step": 9541 }, { "epoch": 0.4826748950376853, "grad_norm": 5.622995376586914, "learning_rate": 5.354057626572067e-06, "loss": 0.4411, "step": 9542 }, { "epoch": 0.48272547928575044, "grad_norm": 8.421003341674805, "learning_rate": 5.353257022567032e-06, "loss": 0.441, "step": 9543 }, { "epoch": 0.4827760635338156, "grad_norm": 8.201364517211914, "learning_rate": 5.352456409459417e-06, "loss": 0.4864, "step": 9544 }, { "epoch": 0.4828266477818807, "grad_norm": 6.437930583953857, "learning_rate": 5.351655787269852e-06, "loss": 0.384, "step": 9545 }, { "epoch": 0.48287723202994587, "grad_norm": 6.267667293548584, "learning_rate": 5.350855156018967e-06, "loss": 0.4615, "step": 9546 }, { "epoch": 0.482927816278011, "grad_norm": 8.586342811584473, "learning_rate": 5.3500545157273905e-06, "loss": 0.3802, "step": 9547 }, { "epoch": 0.48297840052607616, "grad_norm": 5.808287143707275, "learning_rate": 5.349253866415757e-06, "loss": 0.4192, "step": 9548 }, { "epoch": 0.48302898477414136, "grad_norm": 7.876040458679199, "learning_rate": 5.348453208104693e-06, "loss": 0.3694, "step": 9549 }, { "epoch": 0.4830795690222065, "grad_norm": 8.174931526184082, "learning_rate": 5.347652540814833e-06, "loss": 0.3989, "step": 9550 }, { "epoch": 0.48313015327027165, "grad_norm": 5.816524982452393, "learning_rate": 5.346851864566805e-06, "loss": 0.3228, "step": 9551 }, { "epoch": 0.4831807375183368, "grad_norm": 9.93396282196045, "learning_rate": 5.346051179381244e-06, "loss": 0.4217, "step": 9552 }, { "epoch": 0.48323132176640193, "grad_norm": 8.544099807739258, "learning_rate": 5.345250485278779e-06, "loss": 0.4329, "step": 9553 }, { "epoch": 0.4832819060144671, "grad_norm": 10.61819076538086, "learning_rate": 5.344449782280042e-06, "loss": 0.3872, "step": 9554 }, { "epoch": 0.4833324902625322, "grad_norm": 23.06550407409668, "learning_rate": 5.343649070405667e-06, "loss": 0.5987, "step": 9555 }, { "epoch": 0.4833830745105974, "grad_norm": 7.828853607177734, "learning_rate": 5.342848349676287e-06, "loss": 0.4596, "step": 9556 }, { "epoch": 0.48343365875866257, "grad_norm": 9.119848251342773, "learning_rate": 5.342047620112532e-06, "loss": 0.3654, "step": 9557 }, { "epoch": 0.4834842430067277, "grad_norm": 14.2045316696167, "learning_rate": 5.341246881735036e-06, "loss": 0.5171, "step": 9558 }, { "epoch": 0.48353482725479285, "grad_norm": 10.327038764953613, "learning_rate": 5.340446134564432e-06, "loss": 0.4377, "step": 9559 }, { "epoch": 0.483585411502858, "grad_norm": 8.02775764465332, "learning_rate": 5.339645378621352e-06, "loss": 0.3815, "step": 9560 }, { "epoch": 0.48363599575092314, "grad_norm": 12.768702507019043, "learning_rate": 5.3388446139264325e-06, "loss": 0.4161, "step": 9561 }, { "epoch": 0.48368657999898834, "grad_norm": 5.905368804931641, "learning_rate": 5.338043840500305e-06, "loss": 0.4323, "step": 9562 }, { "epoch": 0.4837371642470535, "grad_norm": 13.96184253692627, "learning_rate": 5.337243058363606e-06, "loss": 0.6019, "step": 9563 }, { "epoch": 0.48378774849511863, "grad_norm": 8.775938987731934, "learning_rate": 5.336442267536967e-06, "loss": 0.4558, "step": 9564 }, { "epoch": 0.4838383327431838, "grad_norm": 7.507713317871094, "learning_rate": 5.335641468041024e-06, "loss": 0.5213, "step": 9565 }, { "epoch": 0.4838889169912489, "grad_norm": 8.223579406738281, "learning_rate": 5.334840659896409e-06, "loss": 0.4361, "step": 9566 }, { "epoch": 0.48393950123931406, "grad_norm": 6.192048072814941, "learning_rate": 5.334039843123761e-06, "loss": 0.4627, "step": 9567 }, { "epoch": 0.4839900854873792, "grad_norm": 11.897294998168945, "learning_rate": 5.333239017743712e-06, "loss": 0.465, "step": 9568 }, { "epoch": 0.4840406697354444, "grad_norm": 8.446494102478027, "learning_rate": 5.332438183776899e-06, "loss": 0.4257, "step": 9569 }, { "epoch": 0.48409125398350955, "grad_norm": 7.345197677612305, "learning_rate": 5.331637341243956e-06, "loss": 0.4169, "step": 9570 }, { "epoch": 0.4841418382315747, "grad_norm": 5.62699556350708, "learning_rate": 5.330836490165522e-06, "loss": 0.4061, "step": 9571 }, { "epoch": 0.48419242247963984, "grad_norm": 10.580989837646484, "learning_rate": 5.330035630562228e-06, "loss": 0.5515, "step": 9572 }, { "epoch": 0.484243006727705, "grad_norm": 9.950418472290039, "learning_rate": 5.329234762454714e-06, "loss": 0.4882, "step": 9573 }, { "epoch": 0.4842935909757701, "grad_norm": 7.423442840576172, "learning_rate": 5.328433885863617e-06, "loss": 0.4592, "step": 9574 }, { "epoch": 0.48434417522383527, "grad_norm": 8.17259407043457, "learning_rate": 5.32763300080957e-06, "loss": 0.3701, "step": 9575 }, { "epoch": 0.48439475947190047, "grad_norm": 3.823683500289917, "learning_rate": 5.3268321073132125e-06, "loss": 0.36, "step": 9576 }, { "epoch": 0.4844453437199656, "grad_norm": 5.494839191436768, "learning_rate": 5.3260312053951815e-06, "loss": 0.2726, "step": 9577 }, { "epoch": 0.48449592796803076, "grad_norm": 8.220050811767578, "learning_rate": 5.325230295076113e-06, "loss": 0.4433, "step": 9578 }, { "epoch": 0.4845465122160959, "grad_norm": 7.2158050537109375, "learning_rate": 5.324429376376645e-06, "loss": 0.3735, "step": 9579 }, { "epoch": 0.48459709646416105, "grad_norm": 6.250274181365967, "learning_rate": 5.3236284493174175e-06, "loss": 0.4044, "step": 9580 }, { "epoch": 0.4846476807122262, "grad_norm": 14.53835391998291, "learning_rate": 5.322827513919066e-06, "loss": 0.5424, "step": 9581 }, { "epoch": 0.4846982649602914, "grad_norm": 6.5669474601745605, "learning_rate": 5.322026570202228e-06, "loss": 0.4262, "step": 9582 }, { "epoch": 0.48474884920835654, "grad_norm": 7.71107292175293, "learning_rate": 5.3212256181875445e-06, "loss": 0.4041, "step": 9583 }, { "epoch": 0.4847994334564217, "grad_norm": 10.098947525024414, "learning_rate": 5.320424657895652e-06, "loss": 0.5443, "step": 9584 }, { "epoch": 0.4848500177044868, "grad_norm": 5.538370132446289, "learning_rate": 5.31962368934719e-06, "loss": 0.3506, "step": 9585 }, { "epoch": 0.48490060195255197, "grad_norm": 11.670936584472656, "learning_rate": 5.3188227125627975e-06, "loss": 0.5413, "step": 9586 }, { "epoch": 0.4849511862006171, "grad_norm": 6.194093704223633, "learning_rate": 5.318021727563113e-06, "loss": 0.3988, "step": 9587 }, { "epoch": 0.48500177044868226, "grad_norm": 6.214161396026611, "learning_rate": 5.317220734368778e-06, "loss": 0.4333, "step": 9588 }, { "epoch": 0.48505235469674746, "grad_norm": 9.836843490600586, "learning_rate": 5.316419733000431e-06, "loss": 0.4863, "step": 9589 }, { "epoch": 0.4851029389448126, "grad_norm": 6.855584144592285, "learning_rate": 5.3156187234787115e-06, "loss": 0.3669, "step": 9590 }, { "epoch": 0.48515352319287774, "grad_norm": 6.285405158996582, "learning_rate": 5.314817705824259e-06, "loss": 0.3259, "step": 9591 }, { "epoch": 0.4852041074409429, "grad_norm": 14.105116844177246, "learning_rate": 5.314016680057714e-06, "loss": 0.5688, "step": 9592 }, { "epoch": 0.48525469168900803, "grad_norm": 9.33488655090332, "learning_rate": 5.31321564619972e-06, "loss": 0.4124, "step": 9593 }, { "epoch": 0.4853052759370732, "grad_norm": 5.499896049499512, "learning_rate": 5.3124146042709135e-06, "loss": 0.3764, "step": 9594 }, { "epoch": 0.4853558601851383, "grad_norm": 6.556337356567383, "learning_rate": 5.311613554291939e-06, "loss": 0.3746, "step": 9595 }, { "epoch": 0.4854064444332035, "grad_norm": 13.099534034729004, "learning_rate": 5.3108124962834355e-06, "loss": 0.5091, "step": 9596 }, { "epoch": 0.48545702868126867, "grad_norm": 4.840950965881348, "learning_rate": 5.310011430266043e-06, "loss": 0.3789, "step": 9597 }, { "epoch": 0.4855076129293338, "grad_norm": 7.197769641876221, "learning_rate": 5.309210356260406e-06, "loss": 0.4074, "step": 9598 }, { "epoch": 0.48555819717739895, "grad_norm": 8.264108657836914, "learning_rate": 5.308409274287165e-06, "loss": 0.3348, "step": 9599 }, { "epoch": 0.4856087814254641, "grad_norm": 11.960200309753418, "learning_rate": 5.307608184366962e-06, "loss": 0.4681, "step": 9600 }, { "epoch": 0.48565936567352924, "grad_norm": 10.147916793823242, "learning_rate": 5.306807086520438e-06, "loss": 0.3583, "step": 9601 }, { "epoch": 0.48570994992159444, "grad_norm": 8.215765953063965, "learning_rate": 5.306005980768238e-06, "loss": 0.4085, "step": 9602 }, { "epoch": 0.4857605341696596, "grad_norm": 10.668039321899414, "learning_rate": 5.305204867131002e-06, "loss": 0.4951, "step": 9603 }, { "epoch": 0.48581111841772473, "grad_norm": 4.424373149871826, "learning_rate": 5.304403745629374e-06, "loss": 0.3142, "step": 9604 }, { "epoch": 0.4858617026657899, "grad_norm": 6.585544109344482, "learning_rate": 5.303602616283997e-06, "loss": 0.3906, "step": 9605 }, { "epoch": 0.485912286913855, "grad_norm": 13.188560485839844, "learning_rate": 5.302801479115513e-06, "loss": 0.4849, "step": 9606 }, { "epoch": 0.48596287116192016, "grad_norm": 6.078419208526611, "learning_rate": 5.302000334144566e-06, "loss": 0.3632, "step": 9607 }, { "epoch": 0.4860134554099853, "grad_norm": 10.156218528747559, "learning_rate": 5.3011991813918015e-06, "loss": 0.4344, "step": 9608 }, { "epoch": 0.4860640396580505, "grad_norm": 7.130109786987305, "learning_rate": 5.300398020877862e-06, "loss": 0.4136, "step": 9609 }, { "epoch": 0.48611462390611565, "grad_norm": 8.618980407714844, "learning_rate": 5.299596852623388e-06, "loss": 0.4449, "step": 9610 }, { "epoch": 0.4861652081541808, "grad_norm": 16.388500213623047, "learning_rate": 5.298795676649029e-06, "loss": 0.3342, "step": 9611 }, { "epoch": 0.48621579240224594, "grad_norm": 22.030488967895508, "learning_rate": 5.297994492975428e-06, "loss": 0.3503, "step": 9612 }, { "epoch": 0.4862663766503111, "grad_norm": 6.383166790008545, "learning_rate": 5.2971933016232265e-06, "loss": 0.3719, "step": 9613 }, { "epoch": 0.4863169608983762, "grad_norm": 8.316350936889648, "learning_rate": 5.2963921026130725e-06, "loss": 0.3827, "step": 9614 }, { "epoch": 0.48636754514644137, "grad_norm": 6.820245265960693, "learning_rate": 5.295590895965611e-06, "loss": 0.3337, "step": 9615 }, { "epoch": 0.48641812939450657, "grad_norm": 16.459487915039062, "learning_rate": 5.294789681701485e-06, "loss": 0.6138, "step": 9616 }, { "epoch": 0.4864687136425717, "grad_norm": 8.416110038757324, "learning_rate": 5.29398845984134e-06, "loss": 0.3322, "step": 9617 }, { "epoch": 0.48651929789063686, "grad_norm": 8.752886772155762, "learning_rate": 5.293187230405824e-06, "loss": 0.3669, "step": 9618 }, { "epoch": 0.486569882138702, "grad_norm": 7.319307804107666, "learning_rate": 5.29238599341558e-06, "loss": 0.3592, "step": 9619 }, { "epoch": 0.48662046638676715, "grad_norm": 21.982423782348633, "learning_rate": 5.291584748891255e-06, "loss": 0.5601, "step": 9620 }, { "epoch": 0.4866710506348323, "grad_norm": 7.63236141204834, "learning_rate": 5.290783496853498e-06, "loss": 0.3841, "step": 9621 }, { "epoch": 0.4867216348828975, "grad_norm": 9.048072814941406, "learning_rate": 5.28998223732295e-06, "loss": 0.4535, "step": 9622 }, { "epoch": 0.48677221913096264, "grad_norm": 8.406920433044434, "learning_rate": 5.289180970320261e-06, "loss": 0.36, "step": 9623 }, { "epoch": 0.4868228033790278, "grad_norm": 10.1232328414917, "learning_rate": 5.288379695866077e-06, "loss": 0.4866, "step": 9624 }, { "epoch": 0.4868733876270929, "grad_norm": 9.442717552185059, "learning_rate": 5.287578413981045e-06, "loss": 0.4128, "step": 9625 }, { "epoch": 0.48692397187515807, "grad_norm": 9.225162506103516, "learning_rate": 5.28677712468581e-06, "loss": 0.462, "step": 9626 }, { "epoch": 0.4869745561232232, "grad_norm": 7.3123908042907715, "learning_rate": 5.285975828001023e-06, "loss": 0.4057, "step": 9627 }, { "epoch": 0.48702514037128836, "grad_norm": 8.99710464477539, "learning_rate": 5.285174523947331e-06, "loss": 0.4584, "step": 9628 }, { "epoch": 0.48707572461935356, "grad_norm": 5.033225059509277, "learning_rate": 5.284373212545377e-06, "loss": 0.3161, "step": 9629 }, { "epoch": 0.4871263088674187, "grad_norm": 6.640942096710205, "learning_rate": 5.283571893815814e-06, "loss": 0.5029, "step": 9630 }, { "epoch": 0.48717689311548384, "grad_norm": 7.6203742027282715, "learning_rate": 5.282770567779288e-06, "loss": 0.4235, "step": 9631 }, { "epoch": 0.487227477363549, "grad_norm": 14.457352638244629, "learning_rate": 5.281969234456448e-06, "loss": 0.3832, "step": 9632 }, { "epoch": 0.48727806161161413, "grad_norm": 9.255553245544434, "learning_rate": 5.281167893867943e-06, "loss": 0.4205, "step": 9633 }, { "epoch": 0.4873286458596793, "grad_norm": 8.427066802978516, "learning_rate": 5.280366546034419e-06, "loss": 0.3774, "step": 9634 }, { "epoch": 0.4873792301077444, "grad_norm": 12.294466018676758, "learning_rate": 5.279565190976525e-06, "loss": 0.5246, "step": 9635 }, { "epoch": 0.4874298143558096, "grad_norm": 9.837552070617676, "learning_rate": 5.278763828714913e-06, "loss": 0.4414, "step": 9636 }, { "epoch": 0.48748039860387476, "grad_norm": 6.013906478881836, "learning_rate": 5.277962459270229e-06, "loss": 0.4273, "step": 9637 }, { "epoch": 0.4875309828519399, "grad_norm": 6.970669746398926, "learning_rate": 5.2771610826631244e-06, "loss": 0.4786, "step": 9638 }, { "epoch": 0.48758156710000505, "grad_norm": 4.48018741607666, "learning_rate": 5.2763596989142485e-06, "loss": 0.246, "step": 9639 }, { "epoch": 0.4876321513480702, "grad_norm": 7.464449405670166, "learning_rate": 5.275558308044253e-06, "loss": 0.4271, "step": 9640 }, { "epoch": 0.48768273559613534, "grad_norm": 7.828527927398682, "learning_rate": 5.274756910073783e-06, "loss": 0.3797, "step": 9641 }, { "epoch": 0.48773331984420054, "grad_norm": 5.678576946258545, "learning_rate": 5.2739555050234904e-06, "loss": 0.4321, "step": 9642 }, { "epoch": 0.4877839040922657, "grad_norm": 12.408370018005371, "learning_rate": 5.273154092914028e-06, "loss": 0.407, "step": 9643 }, { "epoch": 0.48783448834033083, "grad_norm": 6.9096198081970215, "learning_rate": 5.2723526737660425e-06, "loss": 0.3646, "step": 9644 }, { "epoch": 0.487885072588396, "grad_norm": 6.720082759857178, "learning_rate": 5.2715512476001876e-06, "loss": 0.4199, "step": 9645 }, { "epoch": 0.4879356568364611, "grad_norm": 11.910959243774414, "learning_rate": 5.270749814437113e-06, "loss": 0.4617, "step": 9646 }, { "epoch": 0.48798624108452626, "grad_norm": 12.614995002746582, "learning_rate": 5.26994837429747e-06, "loss": 0.5948, "step": 9647 }, { "epoch": 0.4880368253325914, "grad_norm": 8.766846656799316, "learning_rate": 5.269146927201908e-06, "loss": 0.377, "step": 9648 }, { "epoch": 0.4880874095806566, "grad_norm": 6.473141193389893, "learning_rate": 5.2683454731710805e-06, "loss": 0.3809, "step": 9649 }, { "epoch": 0.48813799382872175, "grad_norm": 7.989511966705322, "learning_rate": 5.267544012225639e-06, "loss": 0.3855, "step": 9650 }, { "epoch": 0.4881885780767869, "grad_norm": 9.146416664123535, "learning_rate": 5.266742544386233e-06, "loss": 0.3918, "step": 9651 }, { "epoch": 0.48823916232485204, "grad_norm": 8.359224319458008, "learning_rate": 5.265941069673518e-06, "loss": 0.4139, "step": 9652 }, { "epoch": 0.4882897465729172, "grad_norm": 7.404886722564697, "learning_rate": 5.265139588108144e-06, "loss": 0.3979, "step": 9653 }, { "epoch": 0.4883403308209823, "grad_norm": 5.857920169830322, "learning_rate": 5.2643380997107606e-06, "loss": 0.4427, "step": 9654 }, { "epoch": 0.48839091506904747, "grad_norm": 9.174967765808105, "learning_rate": 5.2635366045020245e-06, "loss": 0.4648, "step": 9655 }, { "epoch": 0.48844149931711267, "grad_norm": 5.084825038909912, "learning_rate": 5.262735102502587e-06, "loss": 0.3664, "step": 9656 }, { "epoch": 0.4884920835651778, "grad_norm": 13.151983261108398, "learning_rate": 5.261933593733099e-06, "loss": 0.4333, "step": 9657 }, { "epoch": 0.48854266781324296, "grad_norm": 10.113083839416504, "learning_rate": 5.261132078214216e-06, "loss": 0.388, "step": 9658 }, { "epoch": 0.4885932520613081, "grad_norm": 9.047136306762695, "learning_rate": 5.260330555966592e-06, "loss": 0.5235, "step": 9659 }, { "epoch": 0.48864383630937325, "grad_norm": 5.802523136138916, "learning_rate": 5.259529027010876e-06, "loss": 0.4365, "step": 9660 }, { "epoch": 0.4886944205574384, "grad_norm": 8.692316055297852, "learning_rate": 5.258727491367723e-06, "loss": 0.5321, "step": 9661 }, { "epoch": 0.4887450048055036, "grad_norm": 9.470986366271973, "learning_rate": 5.25792594905779e-06, "loss": 0.3802, "step": 9662 }, { "epoch": 0.48879558905356874, "grad_norm": 5.23816442489624, "learning_rate": 5.2571244001017266e-06, "loss": 0.411, "step": 9663 }, { "epoch": 0.4888461733016339, "grad_norm": 5.8056254386901855, "learning_rate": 5.256322844520189e-06, "loss": 0.4149, "step": 9664 }, { "epoch": 0.488896757549699, "grad_norm": 7.6768479347229, "learning_rate": 5.255521282333831e-06, "loss": 0.4091, "step": 9665 }, { "epoch": 0.48894734179776417, "grad_norm": 10.921367645263672, "learning_rate": 5.254719713563306e-06, "loss": 0.4552, "step": 9666 }, { "epoch": 0.4889979260458293, "grad_norm": 8.653494834899902, "learning_rate": 5.25391813822927e-06, "loss": 0.5132, "step": 9667 }, { "epoch": 0.48904851029389446, "grad_norm": 7.831529140472412, "learning_rate": 5.253116556352376e-06, "loss": 0.4197, "step": 9668 }, { "epoch": 0.48909909454195966, "grad_norm": 8.334149360656738, "learning_rate": 5.252314967953282e-06, "loss": 0.4737, "step": 9669 }, { "epoch": 0.4891496787900248, "grad_norm": 9.784543991088867, "learning_rate": 5.251513373052638e-06, "loss": 0.4573, "step": 9670 }, { "epoch": 0.48920026303808994, "grad_norm": 5.701239585876465, "learning_rate": 5.250711771671103e-06, "loss": 0.3653, "step": 9671 }, { "epoch": 0.4892508472861551, "grad_norm": 7.531930446624756, "learning_rate": 5.249910163829332e-06, "loss": 0.3795, "step": 9672 }, { "epoch": 0.48930143153422023, "grad_norm": 11.705558776855469, "learning_rate": 5.2491085495479785e-06, "loss": 0.5285, "step": 9673 }, { "epoch": 0.4893520157822854, "grad_norm": 7.949028491973877, "learning_rate": 5.2483069288476995e-06, "loss": 0.4028, "step": 9674 }, { "epoch": 0.4894026000303506, "grad_norm": 7.181955337524414, "learning_rate": 5.2475053017491505e-06, "loss": 0.346, "step": 9675 }, { "epoch": 0.4894531842784157, "grad_norm": 7.517616271972656, "learning_rate": 5.246703668272988e-06, "loss": 0.4846, "step": 9676 }, { "epoch": 0.48950376852648086, "grad_norm": 7.520753860473633, "learning_rate": 5.245902028439868e-06, "loss": 0.3764, "step": 9677 }, { "epoch": 0.489554352774546, "grad_norm": 9.782757759094238, "learning_rate": 5.245100382270448e-06, "loss": 0.4335, "step": 9678 }, { "epoch": 0.48960493702261115, "grad_norm": 7.7022600173950195, "learning_rate": 5.244298729785382e-06, "loss": 0.4312, "step": 9679 }, { "epoch": 0.4896555212706763, "grad_norm": 7.085372447967529, "learning_rate": 5.243497071005327e-06, "loss": 0.4091, "step": 9680 }, { "epoch": 0.48970610551874144, "grad_norm": 6.294848918914795, "learning_rate": 5.242695405950942e-06, "loss": 0.4198, "step": 9681 }, { "epoch": 0.48975668976680664, "grad_norm": 7.162083148956299, "learning_rate": 5.241893734642881e-06, "loss": 0.4586, "step": 9682 }, { "epoch": 0.4898072740148718, "grad_norm": 8.331466674804688, "learning_rate": 5.241092057101803e-06, "loss": 0.3888, "step": 9683 }, { "epoch": 0.48985785826293693, "grad_norm": 5.476410865783691, "learning_rate": 5.240290373348367e-06, "loss": 0.3683, "step": 9684 }, { "epoch": 0.4899084425110021, "grad_norm": 5.834568500518799, "learning_rate": 5.239488683403227e-06, "loss": 0.3887, "step": 9685 }, { "epoch": 0.4899590267590672, "grad_norm": 9.712559700012207, "learning_rate": 5.2386869872870415e-06, "loss": 0.4797, "step": 9686 }, { "epoch": 0.49000961100713236, "grad_norm": 12.223578453063965, "learning_rate": 5.237885285020471e-06, "loss": 0.4898, "step": 9687 }, { "epoch": 0.4900601952551975, "grad_norm": 5.133445739746094, "learning_rate": 5.237083576624169e-06, "loss": 0.4185, "step": 9688 }, { "epoch": 0.4901107795032627, "grad_norm": 4.572971343994141, "learning_rate": 5.236281862118796e-06, "loss": 0.3181, "step": 9689 }, { "epoch": 0.49016136375132785, "grad_norm": 7.46997594833374, "learning_rate": 5.235480141525011e-06, "loss": 0.4371, "step": 9690 }, { "epoch": 0.490211947999393, "grad_norm": 5.6240234375, "learning_rate": 5.234678414863472e-06, "loss": 0.3952, "step": 9691 }, { "epoch": 0.49026253224745814, "grad_norm": 11.89235782623291, "learning_rate": 5.233876682154836e-06, "loss": 0.4678, "step": 9692 }, { "epoch": 0.4903131164955233, "grad_norm": 5.419736385345459, "learning_rate": 5.2330749434197635e-06, "loss": 0.4013, "step": 9693 }, { "epoch": 0.4903637007435884, "grad_norm": 4.905161380767822, "learning_rate": 5.232273198678913e-06, "loss": 0.3753, "step": 9694 }, { "epoch": 0.4904142849916536, "grad_norm": 5.087035655975342, "learning_rate": 5.231471447952942e-06, "loss": 0.363, "step": 9695 }, { "epoch": 0.49046486923971877, "grad_norm": 13.349833488464355, "learning_rate": 5.230669691262511e-06, "loss": 0.6257, "step": 9696 }, { "epoch": 0.4905154534877839, "grad_norm": 8.078004837036133, "learning_rate": 5.22986792862828e-06, "loss": 0.428, "step": 9697 }, { "epoch": 0.49056603773584906, "grad_norm": 7.871364116668701, "learning_rate": 5.229066160070908e-06, "loss": 0.4852, "step": 9698 }, { "epoch": 0.4906166219839142, "grad_norm": 10.061285972595215, "learning_rate": 5.228264385611054e-06, "loss": 0.4699, "step": 9699 }, { "epoch": 0.49066720623197935, "grad_norm": 5.755280017852783, "learning_rate": 5.227462605269378e-06, "loss": 0.4539, "step": 9700 }, { "epoch": 0.4907177904800445, "grad_norm": 7.584836959838867, "learning_rate": 5.22666081906654e-06, "loss": 0.399, "step": 9701 }, { "epoch": 0.4907683747281097, "grad_norm": 14.856471061706543, "learning_rate": 5.225859027023199e-06, "loss": 0.5085, "step": 9702 }, { "epoch": 0.49081895897617484, "grad_norm": 6.199364185333252, "learning_rate": 5.225057229160019e-06, "loss": 0.4658, "step": 9703 }, { "epoch": 0.49086954322424, "grad_norm": 5.425164699554443, "learning_rate": 5.224255425497656e-06, "loss": 0.3719, "step": 9704 }, { "epoch": 0.4909201274723051, "grad_norm": 8.86257266998291, "learning_rate": 5.223453616056773e-06, "loss": 0.4066, "step": 9705 }, { "epoch": 0.49097071172037027, "grad_norm": 13.258806228637695, "learning_rate": 5.2226518008580305e-06, "loss": 0.4533, "step": 9706 }, { "epoch": 0.4910212959684354, "grad_norm": 6.256863594055176, "learning_rate": 5.2218499799220875e-06, "loss": 0.4252, "step": 9707 }, { "epoch": 0.49107188021650056, "grad_norm": 4.885673999786377, "learning_rate": 5.2210481532696065e-06, "loss": 0.4097, "step": 9708 }, { "epoch": 0.49112246446456576, "grad_norm": 16.401140213012695, "learning_rate": 5.22024632092125e-06, "loss": 0.4553, "step": 9709 }, { "epoch": 0.4911730487126309, "grad_norm": 4.90416955947876, "learning_rate": 5.219444482897677e-06, "loss": 0.3192, "step": 9710 }, { "epoch": 0.49122363296069604, "grad_norm": 4.841923236846924, "learning_rate": 5.21864263921955e-06, "loss": 0.255, "step": 9711 }, { "epoch": 0.4912742172087612, "grad_norm": 13.155181884765625, "learning_rate": 5.217840789907529e-06, "loss": 0.5359, "step": 9712 }, { "epoch": 0.49132480145682633, "grad_norm": 11.122557640075684, "learning_rate": 5.217038934982278e-06, "loss": 0.4231, "step": 9713 }, { "epoch": 0.4913753857048915, "grad_norm": 7.477594375610352, "learning_rate": 5.2162370744644575e-06, "loss": 0.5055, "step": 9714 }, { "epoch": 0.4914259699529567, "grad_norm": 7.985776424407959, "learning_rate": 5.21543520837473e-06, "loss": 0.3824, "step": 9715 }, { "epoch": 0.4914765542010218, "grad_norm": 6.073765277862549, "learning_rate": 5.214633336733759e-06, "loss": 0.4489, "step": 9716 }, { "epoch": 0.49152713844908696, "grad_norm": 7.934134006500244, "learning_rate": 5.213831459562204e-06, "loss": 0.4803, "step": 9717 }, { "epoch": 0.4915777226971521, "grad_norm": 5.1529130935668945, "learning_rate": 5.213029576880728e-06, "loss": 0.3633, "step": 9718 }, { "epoch": 0.49162830694521725, "grad_norm": 7.664823532104492, "learning_rate": 5.212227688709997e-06, "loss": 0.452, "step": 9719 }, { "epoch": 0.4916788911932824, "grad_norm": 10.273609161376953, "learning_rate": 5.211425795070668e-06, "loss": 0.3634, "step": 9720 }, { "epoch": 0.49172947544134754, "grad_norm": 10.643684387207031, "learning_rate": 5.2106238959834075e-06, "loss": 0.5254, "step": 9721 }, { "epoch": 0.49178005968941274, "grad_norm": 9.053452491760254, "learning_rate": 5.20982199146888e-06, "loss": 0.42, "step": 9722 }, { "epoch": 0.4918306439374779, "grad_norm": 5.728879928588867, "learning_rate": 5.209020081547746e-06, "loss": 0.4287, "step": 9723 }, { "epoch": 0.49188122818554303, "grad_norm": 14.739737510681152, "learning_rate": 5.208218166240667e-06, "loss": 0.3631, "step": 9724 }, { "epoch": 0.4919318124336082, "grad_norm": 10.945347785949707, "learning_rate": 5.207416245568311e-06, "loss": 0.4612, "step": 9725 }, { "epoch": 0.4919823966816733, "grad_norm": 6.637511730194092, "learning_rate": 5.206614319551338e-06, "loss": 0.4312, "step": 9726 }, { "epoch": 0.49203298092973846, "grad_norm": 8.43374252319336, "learning_rate": 5.205812388210414e-06, "loss": 0.3039, "step": 9727 }, { "epoch": 0.4920835651778036, "grad_norm": 10.430550575256348, "learning_rate": 5.205010451566202e-06, "loss": 0.5076, "step": 9728 }, { "epoch": 0.4921341494258688, "grad_norm": 10.135024070739746, "learning_rate": 5.2042085096393645e-06, "loss": 0.4156, "step": 9729 }, { "epoch": 0.49218473367393395, "grad_norm": 6.172105312347412, "learning_rate": 5.203406562450569e-06, "loss": 0.3662, "step": 9730 }, { "epoch": 0.4922353179219991, "grad_norm": 10.344637870788574, "learning_rate": 5.2026046100204766e-06, "loss": 0.4118, "step": 9731 }, { "epoch": 0.49228590217006424, "grad_norm": 6.776928424835205, "learning_rate": 5.2018026523697525e-06, "loss": 0.424, "step": 9732 }, { "epoch": 0.4923364864181294, "grad_norm": 7.82218599319458, "learning_rate": 5.201000689519063e-06, "loss": 0.4566, "step": 9733 }, { "epoch": 0.4923870706661945, "grad_norm": 7.65178108215332, "learning_rate": 5.200198721489068e-06, "loss": 0.3428, "step": 9734 }, { "epoch": 0.4924376549142597, "grad_norm": 6.938331604003906, "learning_rate": 5.1993967483004386e-06, "loss": 0.3529, "step": 9735 }, { "epoch": 0.49248823916232487, "grad_norm": 6.647042274475098, "learning_rate": 5.198594769973836e-06, "loss": 0.4381, "step": 9736 }, { "epoch": 0.49253882341039, "grad_norm": 8.625614166259766, "learning_rate": 5.197792786529925e-06, "loss": 0.4062, "step": 9737 }, { "epoch": 0.49258940765845516, "grad_norm": 13.178269386291504, "learning_rate": 5.196990797989373e-06, "loss": 0.3794, "step": 9738 }, { "epoch": 0.4926399919065203, "grad_norm": 7.399637699127197, "learning_rate": 5.196188804372843e-06, "loss": 0.4074, "step": 9739 }, { "epoch": 0.49269057615458545, "grad_norm": 5.553203582763672, "learning_rate": 5.195386805701001e-06, "loss": 0.3986, "step": 9740 }, { "epoch": 0.4927411604026506, "grad_norm": 9.848705291748047, "learning_rate": 5.194584801994514e-06, "loss": 0.3362, "step": 9741 }, { "epoch": 0.4927917446507158, "grad_norm": 7.346928596496582, "learning_rate": 5.193782793274045e-06, "loss": 0.3645, "step": 9742 }, { "epoch": 0.49284232889878093, "grad_norm": 6.185661792755127, "learning_rate": 5.192980779560263e-06, "loss": 0.4159, "step": 9743 }, { "epoch": 0.4928929131468461, "grad_norm": 5.287821292877197, "learning_rate": 5.1921787608738314e-06, "loss": 0.3638, "step": 9744 }, { "epoch": 0.4929434973949112, "grad_norm": 6.125422954559326, "learning_rate": 5.191376737235418e-06, "loss": 0.403, "step": 9745 }, { "epoch": 0.49299408164297637, "grad_norm": 8.49548625946045, "learning_rate": 5.1905747086656875e-06, "loss": 0.4836, "step": 9746 }, { "epoch": 0.4930446658910415, "grad_norm": 10.095620155334473, "learning_rate": 5.189772675185309e-06, "loss": 0.4487, "step": 9747 }, { "epoch": 0.49309525013910666, "grad_norm": 15.270340919494629, "learning_rate": 5.188970636814945e-06, "loss": 0.4469, "step": 9748 }, { "epoch": 0.49314583438717186, "grad_norm": 6.9191131591796875, "learning_rate": 5.1881685935752654e-06, "loss": 0.435, "step": 9749 }, { "epoch": 0.493196418635237, "grad_norm": 12.810880661010742, "learning_rate": 5.187366545486935e-06, "loss": 0.7491, "step": 9750 }, { "epoch": 0.49324700288330214, "grad_norm": 7.629592418670654, "learning_rate": 5.1865644925706215e-06, "loss": 0.4282, "step": 9751 }, { "epoch": 0.4932975871313673, "grad_norm": 10.546704292297363, "learning_rate": 5.185762434846992e-06, "loss": 0.4147, "step": 9752 }, { "epoch": 0.49334817137943243, "grad_norm": 8.195021629333496, "learning_rate": 5.184960372336712e-06, "loss": 0.3987, "step": 9753 }, { "epoch": 0.4933987556274976, "grad_norm": 6.528144836425781, "learning_rate": 5.18415830506045e-06, "loss": 0.4422, "step": 9754 }, { "epoch": 0.4934493398755628, "grad_norm": 6.949896812438965, "learning_rate": 5.183356233038874e-06, "loss": 0.4047, "step": 9755 }, { "epoch": 0.4934999241236279, "grad_norm": 8.013535499572754, "learning_rate": 5.182554156292652e-06, "loss": 0.3964, "step": 9756 }, { "epoch": 0.49355050837169306, "grad_norm": 23.60531997680664, "learning_rate": 5.18175207484245e-06, "loss": 0.4435, "step": 9757 }, { "epoch": 0.4936010926197582, "grad_norm": 8.143754959106445, "learning_rate": 5.180949988708935e-06, "loss": 0.4015, "step": 9758 }, { "epoch": 0.49365167686782335, "grad_norm": 10.649435043334961, "learning_rate": 5.180147897912775e-06, "loss": 0.4105, "step": 9759 }, { "epoch": 0.4937022611158885, "grad_norm": 6.2387800216674805, "learning_rate": 5.1793458024746404e-06, "loss": 0.3985, "step": 9760 }, { "epoch": 0.49375284536395364, "grad_norm": 10.795408248901367, "learning_rate": 5.178543702415196e-06, "loss": 0.5464, "step": 9761 }, { "epoch": 0.49380342961201884, "grad_norm": 39.37309646606445, "learning_rate": 5.177741597755113e-06, "loss": 0.6406, "step": 9762 }, { "epoch": 0.493854013860084, "grad_norm": 6.977903842926025, "learning_rate": 5.17693948851506e-06, "loss": 0.3097, "step": 9763 }, { "epoch": 0.49390459810814913, "grad_norm": 9.789530754089355, "learning_rate": 5.176137374715702e-06, "loss": 0.4357, "step": 9764 }, { "epoch": 0.4939551823562143, "grad_norm": 5.571126461029053, "learning_rate": 5.175335256377708e-06, "loss": 0.4316, "step": 9765 }, { "epoch": 0.4940057666042794, "grad_norm": 9.136078834533691, "learning_rate": 5.174533133521749e-06, "loss": 0.3761, "step": 9766 }, { "epoch": 0.49405635085234456, "grad_norm": 6.948847770690918, "learning_rate": 5.173731006168493e-06, "loss": 0.3701, "step": 9767 }, { "epoch": 0.4941069351004097, "grad_norm": 7.155093669891357, "learning_rate": 5.172928874338609e-06, "loss": 0.4176, "step": 9768 }, { "epoch": 0.4941575193484749, "grad_norm": 9.209980964660645, "learning_rate": 5.172126738052765e-06, "loss": 0.4122, "step": 9769 }, { "epoch": 0.49420810359654005, "grad_norm": 5.120047092437744, "learning_rate": 5.1713245973316305e-06, "loss": 0.3681, "step": 9770 }, { "epoch": 0.4942586878446052, "grad_norm": 8.786460876464844, "learning_rate": 5.170522452195875e-06, "loss": 0.48, "step": 9771 }, { "epoch": 0.49430927209267034, "grad_norm": 8.703717231750488, "learning_rate": 5.169720302666168e-06, "loss": 0.4205, "step": 9772 }, { "epoch": 0.4943598563407355, "grad_norm": 8.881110191345215, "learning_rate": 5.168918148763179e-06, "loss": 0.4688, "step": 9773 }, { "epoch": 0.4944104405888006, "grad_norm": 3.70487904548645, "learning_rate": 5.168115990507576e-06, "loss": 0.3509, "step": 9774 }, { "epoch": 0.4944610248368658, "grad_norm": 44.17848587036133, "learning_rate": 5.167313827920032e-06, "loss": 0.6555, "step": 9775 }, { "epoch": 0.49451160908493097, "grad_norm": 10.084182739257812, "learning_rate": 5.166511661021215e-06, "loss": 0.4061, "step": 9776 }, { "epoch": 0.4945621933329961, "grad_norm": 5.8686065673828125, "learning_rate": 5.165709489831793e-06, "loss": 0.3111, "step": 9777 }, { "epoch": 0.49461277758106126, "grad_norm": 8.225114822387695, "learning_rate": 5.164907314372437e-06, "loss": 0.4256, "step": 9778 }, { "epoch": 0.4946633618291264, "grad_norm": 5.593372821807861, "learning_rate": 5.16410513466382e-06, "loss": 0.3797, "step": 9779 }, { "epoch": 0.49471394607719155, "grad_norm": 5.7628068923950195, "learning_rate": 5.163302950726609e-06, "loss": 0.4458, "step": 9780 }, { "epoch": 0.4947645303252567, "grad_norm": 7.445797443389893, "learning_rate": 5.162500762581476e-06, "loss": 0.4511, "step": 9781 }, { "epoch": 0.4948151145733219, "grad_norm": 9.623859405517578, "learning_rate": 5.161698570249091e-06, "loss": 0.4613, "step": 9782 }, { "epoch": 0.49486569882138703, "grad_norm": 6.145320415496826, "learning_rate": 5.1608963737501225e-06, "loss": 0.3522, "step": 9783 }, { "epoch": 0.4949162830694522, "grad_norm": 8.875248908996582, "learning_rate": 5.1600941731052446e-06, "loss": 0.3572, "step": 9784 }, { "epoch": 0.4949668673175173, "grad_norm": 8.993864059448242, "learning_rate": 5.1592919683351264e-06, "loss": 0.5413, "step": 9785 }, { "epoch": 0.49501745156558247, "grad_norm": 4.0725016593933105, "learning_rate": 5.158489759460439e-06, "loss": 0.3316, "step": 9786 }, { "epoch": 0.4950680358136476, "grad_norm": 12.937582969665527, "learning_rate": 5.157687546501852e-06, "loss": 0.5733, "step": 9787 }, { "epoch": 0.49511862006171276, "grad_norm": 7.508810043334961, "learning_rate": 5.156885329480039e-06, "loss": 0.4565, "step": 9788 }, { "epoch": 0.49516920430977795, "grad_norm": 6.570767402648926, "learning_rate": 5.156083108415672e-06, "loss": 0.3964, "step": 9789 }, { "epoch": 0.4952197885578431, "grad_norm": 14.663270950317383, "learning_rate": 5.155280883329417e-06, "loss": 0.5411, "step": 9790 }, { "epoch": 0.49527037280590824, "grad_norm": 5.29934549331665, "learning_rate": 5.15447865424195e-06, "loss": 0.4271, "step": 9791 }, { "epoch": 0.4953209570539734, "grad_norm": 4.354402542114258, "learning_rate": 5.153676421173941e-06, "loss": 0.3721, "step": 9792 }, { "epoch": 0.49537154130203853, "grad_norm": 6.625010013580322, "learning_rate": 5.152874184146061e-06, "loss": 0.4052, "step": 9793 }, { "epoch": 0.4954221255501037, "grad_norm": 10.594331741333008, "learning_rate": 5.152071943178985e-06, "loss": 0.4945, "step": 9794 }, { "epoch": 0.4954727097981689, "grad_norm": 5.025545120239258, "learning_rate": 5.151269698293381e-06, "loss": 0.3805, "step": 9795 }, { "epoch": 0.495523294046234, "grad_norm": 4.492533206939697, "learning_rate": 5.1504674495099215e-06, "loss": 0.3993, "step": 9796 }, { "epoch": 0.49557387829429916, "grad_norm": 5.683462142944336, "learning_rate": 5.14966519684928e-06, "loss": 0.4247, "step": 9797 }, { "epoch": 0.4956244625423643, "grad_norm": 7.433513641357422, "learning_rate": 5.148862940332128e-06, "loss": 0.4575, "step": 9798 }, { "epoch": 0.49567504679042945, "grad_norm": 5.608217716217041, "learning_rate": 5.148060679979136e-06, "loss": 0.3195, "step": 9799 }, { "epoch": 0.4957256310384946, "grad_norm": 7.745886325836182, "learning_rate": 5.147258415810979e-06, "loss": 0.4496, "step": 9800 }, { "epoch": 0.49577621528655974, "grad_norm": 6.172485828399658, "learning_rate": 5.146456147848331e-06, "loss": 0.402, "step": 9801 }, { "epoch": 0.49582679953462494, "grad_norm": 13.242355346679688, "learning_rate": 5.145653876111859e-06, "loss": 0.6744, "step": 9802 }, { "epoch": 0.4958773837826901, "grad_norm": 12.666581153869629, "learning_rate": 5.144851600622238e-06, "loss": 0.5609, "step": 9803 }, { "epoch": 0.49592796803075523, "grad_norm": 9.005125999450684, "learning_rate": 5.144049321400142e-06, "loss": 0.446, "step": 9804 }, { "epoch": 0.4959785522788204, "grad_norm": 9.835570335388184, "learning_rate": 5.143247038466242e-06, "loss": 0.424, "step": 9805 }, { "epoch": 0.4960291365268855, "grad_norm": 6.20929479598999, "learning_rate": 5.142444751841213e-06, "loss": 0.4302, "step": 9806 }, { "epoch": 0.49607972077495066, "grad_norm": 11.185535430908203, "learning_rate": 5.141642461545727e-06, "loss": 0.5093, "step": 9807 }, { "epoch": 0.4961303050230158, "grad_norm": 8.697735786437988, "learning_rate": 5.140840167600457e-06, "loss": 0.4749, "step": 9808 }, { "epoch": 0.496180889271081, "grad_norm": 6.6195831298828125, "learning_rate": 5.1400378700260745e-06, "loss": 0.4599, "step": 9809 }, { "epoch": 0.49623147351914615, "grad_norm": 8.792864799499512, "learning_rate": 5.139235568843256e-06, "loss": 0.4227, "step": 9810 }, { "epoch": 0.4962820577672113, "grad_norm": 6.326116561889648, "learning_rate": 5.138433264072672e-06, "loss": 0.4734, "step": 9811 }, { "epoch": 0.49633264201527644, "grad_norm": 6.965936660766602, "learning_rate": 5.137630955734997e-06, "loss": 0.3265, "step": 9812 }, { "epoch": 0.4963832262633416, "grad_norm": 5.4835052490234375, "learning_rate": 5.136828643850906e-06, "loss": 0.4752, "step": 9813 }, { "epoch": 0.4964338105114067, "grad_norm": 10.055061340332031, "learning_rate": 5.1360263284410715e-06, "loss": 0.4739, "step": 9814 }, { "epoch": 0.4964843947594719, "grad_norm": 9.317010879516602, "learning_rate": 5.1352240095261656e-06, "loss": 0.3388, "step": 9815 }, { "epoch": 0.49653497900753707, "grad_norm": 6.0780415534973145, "learning_rate": 5.1344216871268634e-06, "loss": 0.4264, "step": 9816 }, { "epoch": 0.4965855632556022, "grad_norm": 8.43632698059082, "learning_rate": 5.133619361263841e-06, "loss": 0.4722, "step": 9817 }, { "epoch": 0.49663614750366736, "grad_norm": 8.706271171569824, "learning_rate": 5.132817031957769e-06, "loss": 0.4669, "step": 9818 }, { "epoch": 0.4966867317517325, "grad_norm": 8.66635513305664, "learning_rate": 5.132014699229322e-06, "loss": 0.4105, "step": 9819 }, { "epoch": 0.49673731599979765, "grad_norm": 5.9770588874816895, "learning_rate": 5.131212363099178e-06, "loss": 0.4492, "step": 9820 }, { "epoch": 0.4967879002478628, "grad_norm": 7.538000583648682, "learning_rate": 5.130410023588006e-06, "loss": 0.3959, "step": 9821 }, { "epoch": 0.496838484495928, "grad_norm": 8.630340576171875, "learning_rate": 5.129607680716483e-06, "loss": 0.4723, "step": 9822 }, { "epoch": 0.49688906874399313, "grad_norm": 9.853896141052246, "learning_rate": 5.128805334505283e-06, "loss": 0.3788, "step": 9823 }, { "epoch": 0.4969396529920583, "grad_norm": 6.697261810302734, "learning_rate": 5.128002984975081e-06, "loss": 0.3832, "step": 9824 }, { "epoch": 0.4969902372401234, "grad_norm": 7.883578300476074, "learning_rate": 5.127200632146551e-06, "loss": 0.5013, "step": 9825 }, { "epoch": 0.49704082148818857, "grad_norm": 5.217052459716797, "learning_rate": 5.126398276040369e-06, "loss": 0.3957, "step": 9826 }, { "epoch": 0.4970914057362537, "grad_norm": 17.506540298461914, "learning_rate": 5.1255959166772085e-06, "loss": 0.7759, "step": 9827 }, { "epoch": 0.4971419899843189, "grad_norm": 4.812805652618408, "learning_rate": 5.124793554077744e-06, "loss": 0.3755, "step": 9828 }, { "epoch": 0.49719257423238405, "grad_norm": 8.228261947631836, "learning_rate": 5.1239911882626516e-06, "loss": 0.396, "step": 9829 }, { "epoch": 0.4972431584804492, "grad_norm": 9.16012191772461, "learning_rate": 5.123188819252604e-06, "loss": 0.4085, "step": 9830 }, { "epoch": 0.49729374272851434, "grad_norm": 6.93161153793335, "learning_rate": 5.122386447068279e-06, "loss": 0.3592, "step": 9831 }, { "epoch": 0.4973443269765795, "grad_norm": 11.303747177124023, "learning_rate": 5.121584071730352e-06, "loss": 0.5101, "step": 9832 }, { "epoch": 0.49739491122464463, "grad_norm": 10.20319938659668, "learning_rate": 5.120781693259497e-06, "loss": 0.3575, "step": 9833 }, { "epoch": 0.4974454954727098, "grad_norm": 6.608516693115234, "learning_rate": 5.119979311676388e-06, "loss": 0.3103, "step": 9834 }, { "epoch": 0.497496079720775, "grad_norm": 7.79498815536499, "learning_rate": 5.1191769270017024e-06, "loss": 0.4062, "step": 9835 }, { "epoch": 0.4975466639688401, "grad_norm": 70.89256286621094, "learning_rate": 5.118374539256116e-06, "loss": 0.6247, "step": 9836 }, { "epoch": 0.49759724821690526, "grad_norm": 7.797032833099365, "learning_rate": 5.117572148460302e-06, "loss": 0.5074, "step": 9837 }, { "epoch": 0.4976478324649704, "grad_norm": 4.662127494812012, "learning_rate": 5.116769754634939e-06, "loss": 0.3055, "step": 9838 }, { "epoch": 0.49769841671303555, "grad_norm": 6.27635383605957, "learning_rate": 5.115967357800703e-06, "loss": 0.3615, "step": 9839 }, { "epoch": 0.4977490009611007, "grad_norm": 7.350944519042969, "learning_rate": 5.115164957978267e-06, "loss": 0.4813, "step": 9840 }, { "epoch": 0.49779958520916584, "grad_norm": 8.44549560546875, "learning_rate": 5.114362555188307e-06, "loss": 0.3952, "step": 9841 }, { "epoch": 0.49785016945723104, "grad_norm": 5.781644344329834, "learning_rate": 5.113560149451502e-06, "loss": 0.4082, "step": 9842 }, { "epoch": 0.4979007537052962, "grad_norm": 6.888974666595459, "learning_rate": 5.112757740788524e-06, "loss": 0.4153, "step": 9843 }, { "epoch": 0.49795133795336133, "grad_norm": 10.186591148376465, "learning_rate": 5.111955329220053e-06, "loss": 0.4531, "step": 9844 }, { "epoch": 0.4980019222014265, "grad_norm": 10.005895614624023, "learning_rate": 5.111152914766763e-06, "loss": 0.3887, "step": 9845 }, { "epoch": 0.4980525064494916, "grad_norm": 9.28892707824707, "learning_rate": 5.110350497449332e-06, "loss": 0.4823, "step": 9846 }, { "epoch": 0.49810309069755676, "grad_norm": 4.773074150085449, "learning_rate": 5.109548077288434e-06, "loss": 0.3948, "step": 9847 }, { "epoch": 0.49815367494562196, "grad_norm": 16.477323532104492, "learning_rate": 5.108745654304749e-06, "loss": 0.6306, "step": 9848 }, { "epoch": 0.4982042591936871, "grad_norm": 9.73595905303955, "learning_rate": 5.107943228518948e-06, "loss": 0.5084, "step": 9849 }, { "epoch": 0.49825484344175225, "grad_norm": 8.29925537109375, "learning_rate": 5.107140799951711e-06, "loss": 0.337, "step": 9850 }, { "epoch": 0.4983054276898174, "grad_norm": 9.096826553344727, "learning_rate": 5.106338368623717e-06, "loss": 0.4492, "step": 9851 }, { "epoch": 0.49835601193788254, "grad_norm": 7.423526287078857, "learning_rate": 5.105535934555639e-06, "loss": 0.3414, "step": 9852 }, { "epoch": 0.4984065961859477, "grad_norm": 11.973224639892578, "learning_rate": 5.1047334977681525e-06, "loss": 0.4197, "step": 9853 }, { "epoch": 0.4984571804340128, "grad_norm": 11.67007064819336, "learning_rate": 5.103931058281939e-06, "loss": 0.3599, "step": 9854 }, { "epoch": 0.498507764682078, "grad_norm": 18.47770881652832, "learning_rate": 5.103128616117673e-06, "loss": 0.4114, "step": 9855 }, { "epoch": 0.49855834893014317, "grad_norm": 9.594934463500977, "learning_rate": 5.10232617129603e-06, "loss": 0.5013, "step": 9856 }, { "epoch": 0.4986089331782083, "grad_norm": 6.014290809631348, "learning_rate": 5.10152372383769e-06, "loss": 0.356, "step": 9857 }, { "epoch": 0.49865951742627346, "grad_norm": 7.118540287017822, "learning_rate": 5.10072127376333e-06, "loss": 0.3133, "step": 9858 }, { "epoch": 0.4987101016743386, "grad_norm": 6.2767229080200195, "learning_rate": 5.099918821093626e-06, "loss": 0.4029, "step": 9859 }, { "epoch": 0.49876068592240375, "grad_norm": 8.741042137145996, "learning_rate": 5.099116365849253e-06, "loss": 0.5137, "step": 9860 }, { "epoch": 0.4988112701704689, "grad_norm": 13.045846939086914, "learning_rate": 5.098313908050892e-06, "loss": 0.4635, "step": 9861 }, { "epoch": 0.4988618544185341, "grad_norm": 7.582655429840088, "learning_rate": 5.097511447719219e-06, "loss": 0.4238, "step": 9862 }, { "epoch": 0.49891243866659923, "grad_norm": 9.67236328125, "learning_rate": 5.096708984874911e-06, "loss": 0.3324, "step": 9863 }, { "epoch": 0.4989630229146644, "grad_norm": 7.674924373626709, "learning_rate": 5.095906519538647e-06, "loss": 0.4376, "step": 9864 }, { "epoch": 0.4990136071627295, "grad_norm": 10.254878044128418, "learning_rate": 5.095104051731105e-06, "loss": 0.4855, "step": 9865 }, { "epoch": 0.49906419141079467, "grad_norm": 9.150507926940918, "learning_rate": 5.094301581472958e-06, "loss": 0.4059, "step": 9866 }, { "epoch": 0.4991147756588598, "grad_norm": 6.436769485473633, "learning_rate": 5.09349910878489e-06, "loss": 0.5174, "step": 9867 }, { "epoch": 0.499165359906925, "grad_norm": 10.412915229797363, "learning_rate": 5.092696633687573e-06, "loss": 0.3953, "step": 9868 }, { "epoch": 0.49921594415499015, "grad_norm": 8.748455047607422, "learning_rate": 5.091894156201689e-06, "loss": 0.4143, "step": 9869 }, { "epoch": 0.4992665284030553, "grad_norm": 5.586085796356201, "learning_rate": 5.091091676347915e-06, "loss": 0.4727, "step": 9870 }, { "epoch": 0.49931711265112044, "grad_norm": 8.746078491210938, "learning_rate": 5.090289194146929e-06, "loss": 0.4441, "step": 9871 }, { "epoch": 0.4993676968991856, "grad_norm": 6.760805606842041, "learning_rate": 5.0894867096194065e-06, "loss": 0.3827, "step": 9872 }, { "epoch": 0.49941828114725073, "grad_norm": 9.055668830871582, "learning_rate": 5.088684222786029e-06, "loss": 0.5734, "step": 9873 }, { "epoch": 0.4994688653953159, "grad_norm": 5.91474723815918, "learning_rate": 5.0878817336674735e-06, "loss": 0.392, "step": 9874 }, { "epoch": 0.4995194496433811, "grad_norm": 8.602937698364258, "learning_rate": 5.087079242284417e-06, "loss": 0.4096, "step": 9875 }, { "epoch": 0.4995700338914462, "grad_norm": 7.34332275390625, "learning_rate": 5.086276748657538e-06, "loss": 0.453, "step": 9876 }, { "epoch": 0.49962061813951136, "grad_norm": 11.30978012084961, "learning_rate": 5.085474252807519e-06, "loss": 0.5618, "step": 9877 }, { "epoch": 0.4996712023875765, "grad_norm": 8.424067497253418, "learning_rate": 5.084671754755033e-06, "loss": 0.5049, "step": 9878 }, { "epoch": 0.49972178663564165, "grad_norm": 7.67383337020874, "learning_rate": 5.0838692545207595e-06, "loss": 0.3862, "step": 9879 }, { "epoch": 0.4997723708837068, "grad_norm": 5.554712772369385, "learning_rate": 5.08306675212538e-06, "loss": 0.4747, "step": 9880 }, { "epoch": 0.49982295513177194, "grad_norm": 6.432541847229004, "learning_rate": 5.082264247589568e-06, "loss": 0.4374, "step": 9881 }, { "epoch": 0.49987353937983714, "grad_norm": 6.761114597320557, "learning_rate": 5.081461740934006e-06, "loss": 0.4168, "step": 9882 }, { "epoch": 0.4999241236279023, "grad_norm": 7.0626726150512695, "learning_rate": 5.0806592321793735e-06, "loss": 0.4717, "step": 9883 }, { "epoch": 0.49997470787596743, "grad_norm": 8.383280754089355, "learning_rate": 5.079856721346347e-06, "loss": 0.3701, "step": 9884 }, { "epoch": 0.5000252921240326, "grad_norm": 5.964507579803467, "learning_rate": 5.079054208455604e-06, "loss": 0.3929, "step": 9885 }, { "epoch": 0.5000252921240326, "eval_loss": 0.41632401943206787, "eval_runtime": 34.8933, "eval_samples_per_second": 45.797, "eval_steps_per_second": 5.732, "step": 9885 }, { "epoch": 0.5000758763720977, "grad_norm": 17.50810432434082, "learning_rate": 5.078251693527827e-06, "loss": 0.3886, "step": 9886 }, { "epoch": 0.5001264606201629, "grad_norm": 4.769854545593262, "learning_rate": 5.077449176583691e-06, "loss": 0.3619, "step": 9887 }, { "epoch": 0.500177044868228, "grad_norm": 6.506598472595215, "learning_rate": 5.076646657643876e-06, "loss": 0.3438, "step": 9888 }, { "epoch": 0.5002276291162931, "grad_norm": 4.610020637512207, "learning_rate": 5.075844136729065e-06, "loss": 0.383, "step": 9889 }, { "epoch": 0.5002782133643583, "grad_norm": 8.868645668029785, "learning_rate": 5.075041613859931e-06, "loss": 0.3591, "step": 9890 }, { "epoch": 0.5003287976124234, "grad_norm": 9.56351375579834, "learning_rate": 5.074239089057158e-06, "loss": 0.5243, "step": 9891 }, { "epoch": 0.5003793818604887, "grad_norm": 12.300923347473145, "learning_rate": 5.073436562341422e-06, "loss": 0.4679, "step": 9892 }, { "epoch": 0.5004299661085538, "grad_norm": 10.063740730285645, "learning_rate": 5.072634033733403e-06, "loss": 0.417, "step": 9893 }, { "epoch": 0.500480550356619, "grad_norm": 10.255966186523438, "learning_rate": 5.07183150325378e-06, "loss": 0.526, "step": 9894 }, { "epoch": 0.5005311346046841, "grad_norm": 6.833789348602295, "learning_rate": 5.071028970923234e-06, "loss": 0.3334, "step": 9895 }, { "epoch": 0.5005817188527493, "grad_norm": 6.158719539642334, "learning_rate": 5.070226436762442e-06, "loss": 0.345, "step": 9896 }, { "epoch": 0.5006323031008144, "grad_norm": 5.5988874435424805, "learning_rate": 5.069423900792085e-06, "loss": 0.2727, "step": 9897 }, { "epoch": 0.5006828873488796, "grad_norm": 6.13860559463501, "learning_rate": 5.068621363032841e-06, "loss": 0.4438, "step": 9898 }, { "epoch": 0.5007334715969447, "grad_norm": 9.234786033630371, "learning_rate": 5.06781882350539e-06, "loss": 0.5007, "step": 9899 }, { "epoch": 0.5007840558450098, "grad_norm": 7.046781539916992, "learning_rate": 5.067016282230413e-06, "loss": 0.4395, "step": 9900 }, { "epoch": 0.500834640093075, "grad_norm": 13.136837005615234, "learning_rate": 5.066213739228586e-06, "loss": 0.4606, "step": 9901 }, { "epoch": 0.5008852243411401, "grad_norm": 6.180531024932861, "learning_rate": 5.065411194520593e-06, "loss": 0.3403, "step": 9902 }, { "epoch": 0.5009358085892053, "grad_norm": 7.808349132537842, "learning_rate": 5.064608648127111e-06, "loss": 0.4359, "step": 9903 }, { "epoch": 0.5009863928372704, "grad_norm": 6.92832088470459, "learning_rate": 5.063806100068821e-06, "loss": 0.3933, "step": 9904 }, { "epoch": 0.5010369770853357, "grad_norm": 4.524586200714111, "learning_rate": 5.063003550366401e-06, "loss": 0.3495, "step": 9905 }, { "epoch": 0.5010875613334008, "grad_norm": 15.242612838745117, "learning_rate": 5.062200999040532e-06, "loss": 0.504, "step": 9906 }, { "epoch": 0.501138145581466, "grad_norm": 7.904245853424072, "learning_rate": 5.061398446111892e-06, "loss": 0.3283, "step": 9907 }, { "epoch": 0.5011887298295311, "grad_norm": 6.46450138092041, "learning_rate": 5.060595891601165e-06, "loss": 0.4498, "step": 9908 }, { "epoch": 0.5012393140775963, "grad_norm": 8.762057304382324, "learning_rate": 5.059793335529025e-06, "loss": 0.4247, "step": 9909 }, { "epoch": 0.5012898983256614, "grad_norm": 5.935534954071045, "learning_rate": 5.0589907779161585e-06, "loss": 0.3825, "step": 9910 }, { "epoch": 0.5013404825737265, "grad_norm": 7.214865684509277, "learning_rate": 5.05818821878324e-06, "loss": 0.4647, "step": 9911 }, { "epoch": 0.5013910668217917, "grad_norm": 13.210421562194824, "learning_rate": 5.057385658150953e-06, "loss": 0.5919, "step": 9912 }, { "epoch": 0.5014416510698568, "grad_norm": 9.683685302734375, "learning_rate": 5.0565830960399745e-06, "loss": 0.5646, "step": 9913 }, { "epoch": 0.501492235317922, "grad_norm": 10.67961311340332, "learning_rate": 5.055780532470988e-06, "loss": 0.4763, "step": 9914 }, { "epoch": 0.5015428195659871, "grad_norm": 9.364950180053711, "learning_rate": 5.0549779674646705e-06, "loss": 0.4661, "step": 9915 }, { "epoch": 0.5015934038140523, "grad_norm": 4.740523338317871, "learning_rate": 5.054175401041704e-06, "loss": 0.4612, "step": 9916 }, { "epoch": 0.5016439880621174, "grad_norm": 6.046668529510498, "learning_rate": 5.053372833222769e-06, "loss": 0.3817, "step": 9917 }, { "epoch": 0.5016945723101827, "grad_norm": 9.66118049621582, "learning_rate": 5.052570264028544e-06, "loss": 0.3558, "step": 9918 }, { "epoch": 0.5017451565582478, "grad_norm": 36.169307708740234, "learning_rate": 5.051767693479711e-06, "loss": 0.4301, "step": 9919 }, { "epoch": 0.501795740806313, "grad_norm": 13.448856353759766, "learning_rate": 5.0509651215969494e-06, "loss": 0.4917, "step": 9920 }, { "epoch": 0.5018463250543781, "grad_norm": 8.646164894104004, "learning_rate": 5.05016254840094e-06, "loss": 0.2923, "step": 9921 }, { "epoch": 0.5018969093024432, "grad_norm": 13.58364486694336, "learning_rate": 5.049359973912362e-06, "loss": 0.443, "step": 9922 }, { "epoch": 0.5019474935505084, "grad_norm": 7.5783162117004395, "learning_rate": 5.048557398151898e-06, "loss": 0.4678, "step": 9923 }, { "epoch": 0.5019980777985735, "grad_norm": 7.939192771911621, "learning_rate": 5.047754821140226e-06, "loss": 0.489, "step": 9924 }, { "epoch": 0.5020486620466387, "grad_norm": 5.59691858291626, "learning_rate": 5.046952242898028e-06, "loss": 0.3732, "step": 9925 }, { "epoch": 0.5020992462947038, "grad_norm": 6.7336039543151855, "learning_rate": 5.046149663445984e-06, "loss": 0.4234, "step": 9926 }, { "epoch": 0.502149830542769, "grad_norm": 11.936838150024414, "learning_rate": 5.045347082804775e-06, "loss": 0.4761, "step": 9927 }, { "epoch": 0.5022004147908341, "grad_norm": 7.451305866241455, "learning_rate": 5.04454450099508e-06, "loss": 0.3743, "step": 9928 }, { "epoch": 0.5022509990388992, "grad_norm": 6.84387731552124, "learning_rate": 5.043741918037582e-06, "loss": 0.4171, "step": 9929 }, { "epoch": 0.5023015832869644, "grad_norm": 7.254943370819092, "learning_rate": 5.042939333952961e-06, "loss": 0.396, "step": 9930 }, { "epoch": 0.5023521675350295, "grad_norm": 5.695405006408691, "learning_rate": 5.042136748761896e-06, "loss": 0.4163, "step": 9931 }, { "epoch": 0.5024027517830948, "grad_norm": 13.078325271606445, "learning_rate": 5.041334162485069e-06, "loss": 0.4998, "step": 9932 }, { "epoch": 0.5024533360311599, "grad_norm": 8.078455924987793, "learning_rate": 5.0405315751431606e-06, "loss": 0.3435, "step": 9933 }, { "epoch": 0.5025039202792251, "grad_norm": 11.988458633422852, "learning_rate": 5.039728986756851e-06, "loss": 0.4324, "step": 9934 }, { "epoch": 0.5025545045272902, "grad_norm": 6.067167282104492, "learning_rate": 5.038926397346822e-06, "loss": 0.3864, "step": 9935 }, { "epoch": 0.5026050887753554, "grad_norm": 6.775343894958496, "learning_rate": 5.038123806933753e-06, "loss": 0.3215, "step": 9936 }, { "epoch": 0.5026556730234205, "grad_norm": 8.612154006958008, "learning_rate": 5.037321215538326e-06, "loss": 0.528, "step": 9937 }, { "epoch": 0.5027062572714857, "grad_norm": 5.794929027557373, "learning_rate": 5.036518623181221e-06, "loss": 0.4115, "step": 9938 }, { "epoch": 0.5027568415195508, "grad_norm": 9.797525405883789, "learning_rate": 5.0357160298831194e-06, "loss": 0.4678, "step": 9939 }, { "epoch": 0.502807425767616, "grad_norm": 7.621128559112549, "learning_rate": 5.034913435664704e-06, "loss": 0.4875, "step": 9940 }, { "epoch": 0.5028580100156811, "grad_norm": 7.28499174118042, "learning_rate": 5.034110840546651e-06, "loss": 0.4606, "step": 9941 }, { "epoch": 0.5029085942637462, "grad_norm": 6.260775089263916, "learning_rate": 5.033308244549645e-06, "loss": 0.3232, "step": 9942 }, { "epoch": 0.5029591785118114, "grad_norm": 9.651678085327148, "learning_rate": 5.032505647694367e-06, "loss": 0.4382, "step": 9943 }, { "epoch": 0.5030097627598765, "grad_norm": 6.988442897796631, "learning_rate": 5.031703050001495e-06, "loss": 0.4746, "step": 9944 }, { "epoch": 0.5030603470079418, "grad_norm": 9.488975524902344, "learning_rate": 5.030900451491712e-06, "loss": 0.2962, "step": 9945 }, { "epoch": 0.5031109312560069, "grad_norm": 8.943681716918945, "learning_rate": 5.0300978521857016e-06, "loss": 0.4467, "step": 9946 }, { "epoch": 0.5031615155040721, "grad_norm": 10.276779174804688, "learning_rate": 5.029295252104141e-06, "loss": 0.4168, "step": 9947 }, { "epoch": 0.5032120997521372, "grad_norm": 12.672872543334961, "learning_rate": 5.0284926512677115e-06, "loss": 0.3938, "step": 9948 }, { "epoch": 0.5032626840002024, "grad_norm": 6.677007675170898, "learning_rate": 5.027690049697096e-06, "loss": 0.3929, "step": 9949 }, { "epoch": 0.5033132682482675, "grad_norm": 6.330411911010742, "learning_rate": 5.026887447412975e-06, "loss": 0.4591, "step": 9950 }, { "epoch": 0.5033638524963326, "grad_norm": 9.583245277404785, "learning_rate": 5.0260848444360285e-06, "loss": 0.4324, "step": 9951 }, { "epoch": 0.5034144367443978, "grad_norm": 10.441584587097168, "learning_rate": 5.025282240786939e-06, "loss": 0.4205, "step": 9952 }, { "epoch": 0.5034650209924629, "grad_norm": 7.3715620040893555, "learning_rate": 5.024479636486388e-06, "loss": 0.4186, "step": 9953 }, { "epoch": 0.5035156052405281, "grad_norm": 9.510749816894531, "learning_rate": 5.023677031555054e-06, "loss": 0.4457, "step": 9954 }, { "epoch": 0.5035661894885932, "grad_norm": 4.955674171447754, "learning_rate": 5.022874426013622e-06, "loss": 0.3829, "step": 9955 }, { "epoch": 0.5036167737366584, "grad_norm": 6.656530380249023, "learning_rate": 5.022071819882771e-06, "loss": 0.4124, "step": 9956 }, { "epoch": 0.5036673579847235, "grad_norm": 7.4589080810546875, "learning_rate": 5.021269213183182e-06, "loss": 0.4003, "step": 9957 }, { "epoch": 0.5037179422327888, "grad_norm": 7.870881080627441, "learning_rate": 5.020466605935536e-06, "loss": 0.4383, "step": 9958 }, { "epoch": 0.5037685264808539, "grad_norm": 5.410549640655518, "learning_rate": 5.019663998160517e-06, "loss": 0.3368, "step": 9959 }, { "epoch": 0.503819110728919, "grad_norm": 12.204456329345703, "learning_rate": 5.018861389878802e-06, "loss": 0.5163, "step": 9960 }, { "epoch": 0.5038696949769842, "grad_norm": 8.685126304626465, "learning_rate": 5.018058781111077e-06, "loss": 0.4403, "step": 9961 }, { "epoch": 0.5039202792250493, "grad_norm": 8.978708267211914, "learning_rate": 5.01725617187802e-06, "loss": 0.4543, "step": 9962 }, { "epoch": 0.5039708634731145, "grad_norm": 5.321988582611084, "learning_rate": 5.0164535622003115e-06, "loss": 0.2739, "step": 9963 }, { "epoch": 0.5040214477211796, "grad_norm": 7.60606050491333, "learning_rate": 5.015650952098635e-06, "loss": 0.409, "step": 9964 }, { "epoch": 0.5040720319692448, "grad_norm": 6.3330397605896, "learning_rate": 5.014848341593673e-06, "loss": 0.3877, "step": 9965 }, { "epoch": 0.5041226162173099, "grad_norm": 7.4536309242248535, "learning_rate": 5.014045730706103e-06, "loss": 0.3855, "step": 9966 }, { "epoch": 0.5041732004653751, "grad_norm": 11.404427528381348, "learning_rate": 5.013243119456609e-06, "loss": 0.4639, "step": 9967 }, { "epoch": 0.5042237847134402, "grad_norm": 7.075272083282471, "learning_rate": 5.012440507865874e-06, "loss": 0.3331, "step": 9968 }, { "epoch": 0.5042743689615053, "grad_norm": 8.268036842346191, "learning_rate": 5.011637895954575e-06, "loss": 0.449, "step": 9969 }, { "epoch": 0.5043249532095705, "grad_norm": 5.95432710647583, "learning_rate": 5.010835283743395e-06, "loss": 0.3556, "step": 9970 }, { "epoch": 0.5043755374576357, "grad_norm": 5.825046062469482, "learning_rate": 5.010032671253018e-06, "loss": 0.3891, "step": 9971 }, { "epoch": 0.5044261217057009, "grad_norm": 5.215839862823486, "learning_rate": 5.009230058504121e-06, "loss": 0.324, "step": 9972 }, { "epoch": 0.504476705953766, "grad_norm": 5.918100357055664, "learning_rate": 5.008427445517389e-06, "loss": 0.3358, "step": 9973 }, { "epoch": 0.5045272902018312, "grad_norm": 13.009400367736816, "learning_rate": 5.0076248323135026e-06, "loss": 0.2883, "step": 9974 }, { "epoch": 0.5045778744498963, "grad_norm": 5.868866920471191, "learning_rate": 5.006822218913142e-06, "loss": 0.4643, "step": 9975 }, { "epoch": 0.5046284586979615, "grad_norm": 6.500502109527588, "learning_rate": 5.006019605336989e-06, "loss": 0.4423, "step": 9976 }, { "epoch": 0.5046790429460266, "grad_norm": 6.524890422821045, "learning_rate": 5.0052169916057246e-06, "loss": 0.2054, "step": 9977 }, { "epoch": 0.5047296271940918, "grad_norm": 4.959877967834473, "learning_rate": 5.004414377740032e-06, "loss": 0.373, "step": 9978 }, { "epoch": 0.5047802114421569, "grad_norm": 9.013981819152832, "learning_rate": 5.003611763760592e-06, "loss": 0.3969, "step": 9979 }, { "epoch": 0.504830795690222, "grad_norm": 7.5168776512146, "learning_rate": 5.002809149688084e-06, "loss": 0.4915, "step": 9980 }, { "epoch": 0.5048813799382872, "grad_norm": 6.915525436401367, "learning_rate": 5.002006535543193e-06, "loss": 0.3708, "step": 9981 }, { "epoch": 0.5049319641863523, "grad_norm": 7.736794948577881, "learning_rate": 5.001203921346598e-06, "loss": 0.4032, "step": 9982 }, { "epoch": 0.5049825484344175, "grad_norm": 10.767138481140137, "learning_rate": 5.0004013071189795e-06, "loss": 0.419, "step": 9983 }, { "epoch": 0.5050331326824826, "grad_norm": 9.334574699401855, "learning_rate": 4.999598692881022e-06, "loss": 0.3341, "step": 9984 }, { "epoch": 0.5050837169305479, "grad_norm": 16.87137794494629, "learning_rate": 4.9987960786534045e-06, "loss": 0.4579, "step": 9985 }, { "epoch": 0.505134301178613, "grad_norm": 4.457854270935059, "learning_rate": 4.9979934644568085e-06, "loss": 0.3359, "step": 9986 }, { "epoch": 0.5051848854266782, "grad_norm": 5.658257007598877, "learning_rate": 4.997190850311917e-06, "loss": 0.4034, "step": 9987 }, { "epoch": 0.5052354696747433, "grad_norm": 7.04995584487915, "learning_rate": 4.99638823623941e-06, "loss": 0.4511, "step": 9988 }, { "epoch": 0.5052860539228085, "grad_norm": 7.015823841094971, "learning_rate": 4.995585622259969e-06, "loss": 0.3765, "step": 9989 }, { "epoch": 0.5053366381708736, "grad_norm": 33.1942138671875, "learning_rate": 4.994783008394276e-06, "loss": 0.5616, "step": 9990 }, { "epoch": 0.5053872224189387, "grad_norm": 7.053206443786621, "learning_rate": 4.993980394663013e-06, "loss": 0.3573, "step": 9991 }, { "epoch": 0.5054378066670039, "grad_norm": 5.735273838043213, "learning_rate": 4.9931777810868595e-06, "loss": 0.3367, "step": 9992 }, { "epoch": 0.505488390915069, "grad_norm": 18.367103576660156, "learning_rate": 4.9923751676865e-06, "loss": 0.4427, "step": 9993 }, { "epoch": 0.5055389751631342, "grad_norm": 9.391948699951172, "learning_rate": 4.991572554482611e-06, "loss": 0.5427, "step": 9994 }, { "epoch": 0.5055895594111993, "grad_norm": 10.497010231018066, "learning_rate": 4.990769941495879e-06, "loss": 0.3864, "step": 9995 }, { "epoch": 0.5056401436592645, "grad_norm": 7.007731914520264, "learning_rate": 4.989967328746984e-06, "loss": 0.3405, "step": 9996 }, { "epoch": 0.5056907279073296, "grad_norm": 7.68689489364624, "learning_rate": 4.989164716256607e-06, "loss": 0.5298, "step": 9997 }, { "epoch": 0.5057413121553949, "grad_norm": 8.315173149108887, "learning_rate": 4.988362104045427e-06, "loss": 0.4605, "step": 9998 }, { "epoch": 0.50579189640346, "grad_norm": 10.354839324951172, "learning_rate": 4.987559492134129e-06, "loss": 0.4189, "step": 9999 }, { "epoch": 0.5058424806515252, "grad_norm": 6.167631149291992, "learning_rate": 4.986756880543392e-06, "loss": 0.4377, "step": 10000 }, { "epoch": 0.5058930648995903, "grad_norm": 6.190781593322754, "learning_rate": 4.985954269293898e-06, "loss": 0.4948, "step": 10001 }, { "epoch": 0.5059436491476554, "grad_norm": 14.223040580749512, "learning_rate": 4.985151658406328e-06, "loss": 0.5938, "step": 10002 }, { "epoch": 0.5059942333957206, "grad_norm": 8.351736068725586, "learning_rate": 4.9843490479013655e-06, "loss": 0.5123, "step": 10003 }, { "epoch": 0.5060448176437857, "grad_norm": 5.878732204437256, "learning_rate": 4.983546437799689e-06, "loss": 0.4304, "step": 10004 }, { "epoch": 0.5060954018918509, "grad_norm": 7.969093322753906, "learning_rate": 4.982743828121981e-06, "loss": 0.5282, "step": 10005 }, { "epoch": 0.506145986139916, "grad_norm": 4.444896697998047, "learning_rate": 4.9819412188889246e-06, "loss": 0.366, "step": 10006 }, { "epoch": 0.5061965703879812, "grad_norm": 5.125590801239014, "learning_rate": 4.981138610121197e-06, "loss": 0.4055, "step": 10007 }, { "epoch": 0.5062471546360463, "grad_norm": 11.7821626663208, "learning_rate": 4.980336001839483e-06, "loss": 0.4628, "step": 10008 }, { "epoch": 0.5062977388841114, "grad_norm": 10.841862678527832, "learning_rate": 4.9795333940644644e-06, "loss": 0.4787, "step": 10009 }, { "epoch": 0.5063483231321766, "grad_norm": 7.03364372253418, "learning_rate": 4.978730786816821e-06, "loss": 0.3862, "step": 10010 }, { "epoch": 0.5063989073802418, "grad_norm": 4.926862716674805, "learning_rate": 4.9779281801172314e-06, "loss": 0.4641, "step": 10011 }, { "epoch": 0.506449491628307, "grad_norm": 6.620747089385986, "learning_rate": 4.97712557398638e-06, "loss": 0.4166, "step": 10012 }, { "epoch": 0.5065000758763721, "grad_norm": 12.056138038635254, "learning_rate": 4.976322968444947e-06, "loss": 0.5324, "step": 10013 }, { "epoch": 0.5065506601244373, "grad_norm": 8.400981903076172, "learning_rate": 4.975520363513614e-06, "loss": 0.4449, "step": 10014 }, { "epoch": 0.5066012443725024, "grad_norm": 9.827523231506348, "learning_rate": 4.9747177592130615e-06, "loss": 0.4847, "step": 10015 }, { "epoch": 0.5066518286205676, "grad_norm": 7.750907897949219, "learning_rate": 4.973915155563973e-06, "loss": 0.3862, "step": 10016 }, { "epoch": 0.5067024128686327, "grad_norm": 7.82384729385376, "learning_rate": 4.9731125525870265e-06, "loss": 0.4826, "step": 10017 }, { "epoch": 0.5067529971166979, "grad_norm": 9.23061466217041, "learning_rate": 4.972309950302906e-06, "loss": 0.5099, "step": 10018 }, { "epoch": 0.506803581364763, "grad_norm": 7.574652671813965, "learning_rate": 4.971507348732289e-06, "loss": 0.3374, "step": 10019 }, { "epoch": 0.5068541656128281, "grad_norm": 7.351554870605469, "learning_rate": 4.97070474789586e-06, "loss": 0.3715, "step": 10020 }, { "epoch": 0.5069047498608933, "grad_norm": 8.016729354858398, "learning_rate": 4.9699021478143e-06, "loss": 0.5485, "step": 10021 }, { "epoch": 0.5069553341089584, "grad_norm": 8.405111312866211, "learning_rate": 4.969099548508287e-06, "loss": 0.4184, "step": 10022 }, { "epoch": 0.5070059183570236, "grad_norm": 7.7599568367004395, "learning_rate": 4.968296949998506e-06, "loss": 0.3799, "step": 10023 }, { "epoch": 0.5070565026050887, "grad_norm": 6.545772075653076, "learning_rate": 4.967494352305635e-06, "loss": 0.4562, "step": 10024 }, { "epoch": 0.507107086853154, "grad_norm": 8.638627052307129, "learning_rate": 4.9666917554503565e-06, "loss": 0.4301, "step": 10025 }, { "epoch": 0.5071576711012191, "grad_norm": 7.959926605224609, "learning_rate": 4.96588915945335e-06, "loss": 0.3916, "step": 10026 }, { "epoch": 0.5072082553492843, "grad_norm": 6.538083076477051, "learning_rate": 4.965086564335298e-06, "loss": 0.3931, "step": 10027 }, { "epoch": 0.5072588395973494, "grad_norm": 7.371477127075195, "learning_rate": 4.964283970116881e-06, "loss": 0.5348, "step": 10028 }, { "epoch": 0.5073094238454146, "grad_norm": 5.607303142547607, "learning_rate": 4.9634813768187795e-06, "loss": 0.3302, "step": 10029 }, { "epoch": 0.5073600080934797, "grad_norm": 10.60092830657959, "learning_rate": 4.9626787844616745e-06, "loss": 0.3962, "step": 10030 }, { "epoch": 0.5074105923415448, "grad_norm": 7.387380123138428, "learning_rate": 4.961876193066248e-06, "loss": 0.3221, "step": 10031 }, { "epoch": 0.50746117658961, "grad_norm": 5.71929931640625, "learning_rate": 4.961073602653179e-06, "loss": 0.3387, "step": 10032 }, { "epoch": 0.5075117608376751, "grad_norm": 6.454783916473389, "learning_rate": 4.960271013243149e-06, "loss": 0.4257, "step": 10033 }, { "epoch": 0.5075623450857403, "grad_norm": 8.159294128417969, "learning_rate": 4.95946842485684e-06, "loss": 0.3921, "step": 10034 }, { "epoch": 0.5076129293338054, "grad_norm": 5.372800827026367, "learning_rate": 4.9586658375149335e-06, "loss": 0.3671, "step": 10035 }, { "epoch": 0.5076635135818706, "grad_norm": 8.251035690307617, "learning_rate": 4.957863251238106e-06, "loss": 0.4184, "step": 10036 }, { "epoch": 0.5077140978299357, "grad_norm": 6.336057186126709, "learning_rate": 4.9570606660470414e-06, "loss": 0.4709, "step": 10037 }, { "epoch": 0.507764682078001, "grad_norm": 8.336583137512207, "learning_rate": 4.956258081962419e-06, "loss": 0.4466, "step": 10038 }, { "epoch": 0.5078152663260661, "grad_norm": 6.164731502532959, "learning_rate": 4.955455499004921e-06, "loss": 0.3398, "step": 10039 }, { "epoch": 0.5078658505741313, "grad_norm": 6.582147598266602, "learning_rate": 4.954652917195226e-06, "loss": 0.3836, "step": 10040 }, { "epoch": 0.5079164348221964, "grad_norm": 6.654188632965088, "learning_rate": 4.953850336554018e-06, "loss": 0.3586, "step": 10041 }, { "epoch": 0.5079670190702615, "grad_norm": 8.203961372375488, "learning_rate": 4.953047757101974e-06, "loss": 0.4589, "step": 10042 }, { "epoch": 0.5080176033183267, "grad_norm": 11.914027214050293, "learning_rate": 4.952245178859775e-06, "loss": 0.3897, "step": 10043 }, { "epoch": 0.5080681875663918, "grad_norm": 7.127358436584473, "learning_rate": 4.951442601848104e-06, "loss": 0.2941, "step": 10044 }, { "epoch": 0.508118771814457, "grad_norm": 7.292872428894043, "learning_rate": 4.950640026087639e-06, "loss": 0.3777, "step": 10045 }, { "epoch": 0.5081693560625221, "grad_norm": 8.04482650756836, "learning_rate": 4.949837451599061e-06, "loss": 0.3953, "step": 10046 }, { "epoch": 0.5082199403105873, "grad_norm": 3.817112922668457, "learning_rate": 4.949034878403051e-06, "loss": 0.3108, "step": 10047 }, { "epoch": 0.5082705245586524, "grad_norm": 6.4394612312316895, "learning_rate": 4.948232306520291e-06, "loss": 0.383, "step": 10048 }, { "epoch": 0.5083211088067175, "grad_norm": 10.620590209960938, "learning_rate": 4.947429735971458e-06, "loss": 0.4324, "step": 10049 }, { "epoch": 0.5083716930547827, "grad_norm": 6.733782768249512, "learning_rate": 4.946627166777232e-06, "loss": 0.4625, "step": 10050 }, { "epoch": 0.508422277302848, "grad_norm": 7.42392110824585, "learning_rate": 4.945824598958297e-06, "loss": 0.4265, "step": 10051 }, { "epoch": 0.5084728615509131, "grad_norm": 15.446381568908691, "learning_rate": 4.945022032535331e-06, "loss": 0.5354, "step": 10052 }, { "epoch": 0.5085234457989782, "grad_norm": 12.705818176269531, "learning_rate": 4.944219467529013e-06, "loss": 0.5262, "step": 10053 }, { "epoch": 0.5085740300470434, "grad_norm": 6.412447452545166, "learning_rate": 4.943416903960027e-06, "loss": 0.3812, "step": 10054 }, { "epoch": 0.5086246142951085, "grad_norm": 8.14838981628418, "learning_rate": 4.942614341849049e-06, "loss": 0.4597, "step": 10055 }, { "epoch": 0.5086751985431737, "grad_norm": 6.008846282958984, "learning_rate": 4.941811781216762e-06, "loss": 0.4021, "step": 10056 }, { "epoch": 0.5087257827912388, "grad_norm": 8.429365158081055, "learning_rate": 4.941009222083843e-06, "loss": 0.4523, "step": 10057 }, { "epoch": 0.508776367039304, "grad_norm": 6.765955448150635, "learning_rate": 4.940206664470974e-06, "loss": 0.4623, "step": 10058 }, { "epoch": 0.5088269512873691, "grad_norm": 14.097105979919434, "learning_rate": 4.939404108398837e-06, "loss": 0.5228, "step": 10059 }, { "epoch": 0.5088775355354342, "grad_norm": 4.98317289352417, "learning_rate": 4.938601553888109e-06, "loss": 0.4975, "step": 10060 }, { "epoch": 0.5089281197834994, "grad_norm": 16.386545181274414, "learning_rate": 4.937799000959471e-06, "loss": 0.6931, "step": 10061 }, { "epoch": 0.5089787040315645, "grad_norm": 8.32803726196289, "learning_rate": 4.9369964496336014e-06, "loss": 0.4604, "step": 10062 }, { "epoch": 0.5090292882796297, "grad_norm": 6.953734874725342, "learning_rate": 4.936193899931182e-06, "loss": 0.4177, "step": 10063 }, { "epoch": 0.5090798725276948, "grad_norm": 4.485250473022461, "learning_rate": 4.935391351872891e-06, "loss": 0.4001, "step": 10064 }, { "epoch": 0.5091304567757601, "grad_norm": 12.343631744384766, "learning_rate": 4.934588805479408e-06, "loss": 0.3685, "step": 10065 }, { "epoch": 0.5091810410238252, "grad_norm": 5.991939067840576, "learning_rate": 4.933786260771415e-06, "loss": 0.3574, "step": 10066 }, { "epoch": 0.5092316252718904, "grad_norm": 8.287256240844727, "learning_rate": 4.932983717769589e-06, "loss": 0.4518, "step": 10067 }, { "epoch": 0.5092822095199555, "grad_norm": 8.26500415802002, "learning_rate": 4.932181176494611e-06, "loss": 0.4181, "step": 10068 }, { "epoch": 0.5093327937680207, "grad_norm": 10.075034141540527, "learning_rate": 4.931378636967161e-06, "loss": 0.5786, "step": 10069 }, { "epoch": 0.5093833780160858, "grad_norm": 9.783387184143066, "learning_rate": 4.930576099207917e-06, "loss": 0.4265, "step": 10070 }, { "epoch": 0.5094339622641509, "grad_norm": 4.930639743804932, "learning_rate": 4.9297735632375586e-06, "loss": 0.3975, "step": 10071 }, { "epoch": 0.5094845465122161, "grad_norm": 9.604860305786133, "learning_rate": 4.928971029076767e-06, "loss": 0.4057, "step": 10072 }, { "epoch": 0.5095351307602812, "grad_norm": 5.964163780212402, "learning_rate": 4.9281684967462205e-06, "loss": 0.4132, "step": 10073 }, { "epoch": 0.5095857150083464, "grad_norm": 8.189557075500488, "learning_rate": 4.927365966266599e-06, "loss": 0.4692, "step": 10074 }, { "epoch": 0.5096362992564115, "grad_norm": 6.462892532348633, "learning_rate": 4.926563437658581e-06, "loss": 0.4142, "step": 10075 }, { "epoch": 0.5096868835044767, "grad_norm": 6.552626132965088, "learning_rate": 4.9257609109428444e-06, "loss": 0.3303, "step": 10076 }, { "epoch": 0.5097374677525418, "grad_norm": 10.210892677307129, "learning_rate": 4.92495838614007e-06, "loss": 0.4934, "step": 10077 }, { "epoch": 0.5097880520006071, "grad_norm": 9.058110237121582, "learning_rate": 4.924155863270937e-06, "loss": 0.399, "step": 10078 }, { "epoch": 0.5098386362486722, "grad_norm": 10.35682201385498, "learning_rate": 4.9233533423561245e-06, "loss": 0.4606, "step": 10079 }, { "epoch": 0.5098892204967373, "grad_norm": 5.828109264373779, "learning_rate": 4.922550823416311e-06, "loss": 0.4351, "step": 10080 }, { "epoch": 0.5099398047448025, "grad_norm": 6.581266403198242, "learning_rate": 4.921748306472175e-06, "loss": 0.2795, "step": 10081 }, { "epoch": 0.5099903889928676, "grad_norm": 8.181182861328125, "learning_rate": 4.920945791544398e-06, "loss": 0.4127, "step": 10082 }, { "epoch": 0.5100409732409328, "grad_norm": 6.682002067565918, "learning_rate": 4.920143278653655e-06, "loss": 0.4003, "step": 10083 }, { "epoch": 0.5100915574889979, "grad_norm": 5.377941608428955, "learning_rate": 4.919340767820627e-06, "loss": 0.3706, "step": 10084 }, { "epoch": 0.5101421417370631, "grad_norm": 6.895477771759033, "learning_rate": 4.9185382590659944e-06, "loss": 0.389, "step": 10085 }, { "epoch": 0.5101927259851282, "grad_norm": 10.8114013671875, "learning_rate": 4.917735752410432e-06, "loss": 0.4737, "step": 10086 }, { "epoch": 0.5102433102331934, "grad_norm": 10.704973220825195, "learning_rate": 4.916933247874624e-06, "loss": 0.5273, "step": 10087 }, { "epoch": 0.5102938944812585, "grad_norm": 9.133756637573242, "learning_rate": 4.916130745479242e-06, "loss": 0.4468, "step": 10088 }, { "epoch": 0.5103444787293236, "grad_norm": 5.807254314422607, "learning_rate": 4.91532824524497e-06, "loss": 0.475, "step": 10089 }, { "epoch": 0.5103950629773888, "grad_norm": 23.337852478027344, "learning_rate": 4.914525747192484e-06, "loss": 0.6749, "step": 10090 }, { "epoch": 0.510445647225454, "grad_norm": 7.983128547668457, "learning_rate": 4.913723251342463e-06, "loss": 0.3792, "step": 10091 }, { "epoch": 0.5104962314735192, "grad_norm": 6.228879451751709, "learning_rate": 4.912920757715584e-06, "loss": 0.333, "step": 10092 }, { "epoch": 0.5105468157215843, "grad_norm": 9.934088706970215, "learning_rate": 4.912118266332528e-06, "loss": 0.4627, "step": 10093 }, { "epoch": 0.5105973999696495, "grad_norm": 9.888764381408691, "learning_rate": 4.911315777213973e-06, "loss": 0.4735, "step": 10094 }, { "epoch": 0.5106479842177146, "grad_norm": 20.274919509887695, "learning_rate": 4.910513290380594e-06, "loss": 0.5929, "step": 10095 }, { "epoch": 0.5106985684657798, "grad_norm": 8.188774108886719, "learning_rate": 4.909710805853073e-06, "loss": 0.4395, "step": 10096 }, { "epoch": 0.5107491527138449, "grad_norm": 6.487425327301025, "learning_rate": 4.908908323652086e-06, "loss": 0.3637, "step": 10097 }, { "epoch": 0.5107997369619101, "grad_norm": 5.510702133178711, "learning_rate": 4.908105843798311e-06, "loss": 0.3764, "step": 10098 }, { "epoch": 0.5108503212099752, "grad_norm": 4.968888759613037, "learning_rate": 4.907303366312427e-06, "loss": 0.3648, "step": 10099 }, { "epoch": 0.5109009054580403, "grad_norm": 9.208209037780762, "learning_rate": 4.906500891215113e-06, "loss": 0.4599, "step": 10100 }, { "epoch": 0.5109514897061055, "grad_norm": 9.33668041229248, "learning_rate": 4.905698418527044e-06, "loss": 0.4091, "step": 10101 }, { "epoch": 0.5110020739541706, "grad_norm": 13.831673622131348, "learning_rate": 4.904895948268899e-06, "loss": 0.3999, "step": 10102 }, { "epoch": 0.5110526582022358, "grad_norm": 7.774698257446289, "learning_rate": 4.904093480461354e-06, "loss": 0.3011, "step": 10103 }, { "epoch": 0.5111032424503009, "grad_norm": 6.052530765533447, "learning_rate": 4.90329101512509e-06, "loss": 0.4469, "step": 10104 }, { "epoch": 0.5111538266983662, "grad_norm": 18.815696716308594, "learning_rate": 4.902488552280782e-06, "loss": 0.5934, "step": 10105 }, { "epoch": 0.5112044109464313, "grad_norm": 7.509600639343262, "learning_rate": 4.9016860919491085e-06, "loss": 0.3947, "step": 10106 }, { "epoch": 0.5112549951944965, "grad_norm": 5.327523708343506, "learning_rate": 4.900883634150748e-06, "loss": 0.4129, "step": 10107 }, { "epoch": 0.5113055794425616, "grad_norm": 5.83843994140625, "learning_rate": 4.900081178906376e-06, "loss": 0.4074, "step": 10108 }, { "epoch": 0.5113561636906268, "grad_norm": 8.466279983520508, "learning_rate": 4.89927872623667e-06, "loss": 0.3636, "step": 10109 }, { "epoch": 0.5114067479386919, "grad_norm": 8.686759948730469, "learning_rate": 4.89847627616231e-06, "loss": 0.349, "step": 10110 }, { "epoch": 0.511457332186757, "grad_norm": 4.300432205200195, "learning_rate": 4.89767382870397e-06, "loss": 0.3704, "step": 10111 }, { "epoch": 0.5115079164348222, "grad_norm": 8.494036674499512, "learning_rate": 4.896871383882328e-06, "loss": 0.487, "step": 10112 }, { "epoch": 0.5115585006828873, "grad_norm": 15.887585639953613, "learning_rate": 4.896068941718063e-06, "loss": 0.4925, "step": 10113 }, { "epoch": 0.5116090849309525, "grad_norm": 8.444125175476074, "learning_rate": 4.895266502231849e-06, "loss": 0.3737, "step": 10114 }, { "epoch": 0.5116596691790176, "grad_norm": 8.334081649780273, "learning_rate": 4.894464065444364e-06, "loss": 0.4992, "step": 10115 }, { "epoch": 0.5117102534270828, "grad_norm": 8.11704158782959, "learning_rate": 4.893661631376286e-06, "loss": 0.3956, "step": 10116 }, { "epoch": 0.5117608376751479, "grad_norm": 7.965924263000488, "learning_rate": 4.89285920004829e-06, "loss": 0.4325, "step": 10117 }, { "epoch": 0.5118114219232132, "grad_norm": 8.010968208312988, "learning_rate": 4.892056771481054e-06, "loss": 0.3357, "step": 10118 }, { "epoch": 0.5118620061712783, "grad_norm": 4.2715864181518555, "learning_rate": 4.891254345695253e-06, "loss": 0.3548, "step": 10119 }, { "epoch": 0.5119125904193434, "grad_norm": 8.27991771697998, "learning_rate": 4.890451922711567e-06, "loss": 0.3961, "step": 10120 }, { "epoch": 0.5119631746674086, "grad_norm": 9.770956039428711, "learning_rate": 4.889649502550669e-06, "loss": 0.3916, "step": 10121 }, { "epoch": 0.5120137589154737, "grad_norm": 10.842000961303711, "learning_rate": 4.888847085233237e-06, "loss": 0.5093, "step": 10122 }, { "epoch": 0.5120643431635389, "grad_norm": 20.711912155151367, "learning_rate": 4.888044670779947e-06, "loss": 0.477, "step": 10123 }, { "epoch": 0.512114927411604, "grad_norm": 12.636733055114746, "learning_rate": 4.887242259211476e-06, "loss": 0.4878, "step": 10124 }, { "epoch": 0.5121655116596692, "grad_norm": 8.21280574798584, "learning_rate": 4.886439850548499e-06, "loss": 0.4208, "step": 10125 }, { "epoch": 0.5122160959077343, "grad_norm": 7.502194881439209, "learning_rate": 4.885637444811694e-06, "loss": 0.4543, "step": 10126 }, { "epoch": 0.5122666801557995, "grad_norm": 10.621739387512207, "learning_rate": 4.884835042021736e-06, "loss": 0.4945, "step": 10127 }, { "epoch": 0.5123172644038646, "grad_norm": 10.07345962524414, "learning_rate": 4.884032642199299e-06, "loss": 0.4092, "step": 10128 }, { "epoch": 0.5123678486519297, "grad_norm": 4.823601722717285, "learning_rate": 4.883230245365062e-06, "loss": 0.3078, "step": 10129 }, { "epoch": 0.5124184328999949, "grad_norm": 7.491477012634277, "learning_rate": 4.8824278515396984e-06, "loss": 0.3265, "step": 10130 }, { "epoch": 0.5124690171480601, "grad_norm": 9.005670547485352, "learning_rate": 4.881625460743885e-06, "loss": 0.4692, "step": 10131 }, { "epoch": 0.5125196013961253, "grad_norm": 7.6382975578308105, "learning_rate": 4.880823072998299e-06, "loss": 0.322, "step": 10132 }, { "epoch": 0.5125701856441904, "grad_norm": 9.743782997131348, "learning_rate": 4.8800206883236135e-06, "loss": 0.3717, "step": 10133 }, { "epoch": 0.5126207698922556, "grad_norm": 11.125005722045898, "learning_rate": 4.879218306740505e-06, "loss": 0.4518, "step": 10134 }, { "epoch": 0.5126713541403207, "grad_norm": 8.931954383850098, "learning_rate": 4.8784159282696495e-06, "loss": 0.3684, "step": 10135 }, { "epoch": 0.5127219383883859, "grad_norm": 8.868249893188477, "learning_rate": 4.8776135529317216e-06, "loss": 0.4032, "step": 10136 }, { "epoch": 0.512772522636451, "grad_norm": 6.676638603210449, "learning_rate": 4.876811180747396e-06, "loss": 0.3151, "step": 10137 }, { "epoch": 0.5128231068845162, "grad_norm": 7.685858249664307, "learning_rate": 4.876008811737352e-06, "loss": 0.4928, "step": 10138 }, { "epoch": 0.5128736911325813, "grad_norm": 8.327526092529297, "learning_rate": 4.8752064459222595e-06, "loss": 0.4707, "step": 10139 }, { "epoch": 0.5129242753806464, "grad_norm": 8.581992149353027, "learning_rate": 4.874404083322794e-06, "loss": 0.483, "step": 10140 }, { "epoch": 0.5129748596287116, "grad_norm": 8.467451095581055, "learning_rate": 4.873601723959632e-06, "loss": 0.3398, "step": 10141 }, { "epoch": 0.5130254438767767, "grad_norm": 6.012983798980713, "learning_rate": 4.87279936785345e-06, "loss": 0.3846, "step": 10142 }, { "epoch": 0.5130760281248419, "grad_norm": 6.383946418762207, "learning_rate": 4.8719970150249205e-06, "loss": 0.4287, "step": 10143 }, { "epoch": 0.5131266123729071, "grad_norm": 8.471840858459473, "learning_rate": 4.871194665494718e-06, "loss": 0.5229, "step": 10144 }, { "epoch": 0.5131771966209723, "grad_norm": 8.432805061340332, "learning_rate": 4.870392319283519e-06, "loss": 0.4337, "step": 10145 }, { "epoch": 0.5132277808690374, "grad_norm": 16.81555938720703, "learning_rate": 4.869589976411995e-06, "loss": 0.5195, "step": 10146 }, { "epoch": 0.5132783651171026, "grad_norm": 27.8244686126709, "learning_rate": 4.868787636900823e-06, "loss": 0.3764, "step": 10147 }, { "epoch": 0.5133289493651677, "grad_norm": 13.752116203308105, "learning_rate": 4.867985300770678e-06, "loss": 0.4326, "step": 10148 }, { "epoch": 0.5133795336132329, "grad_norm": 7.333558082580566, "learning_rate": 4.867182968042232e-06, "loss": 0.3928, "step": 10149 }, { "epoch": 0.513430117861298, "grad_norm": 5.2074689865112305, "learning_rate": 4.86638063873616e-06, "loss": 0.3814, "step": 10150 }, { "epoch": 0.5134807021093631, "grad_norm": 6.279603958129883, "learning_rate": 4.865578312873137e-06, "loss": 0.3592, "step": 10151 }, { "epoch": 0.5135312863574283, "grad_norm": 7.70607328414917, "learning_rate": 4.864775990473837e-06, "loss": 0.3267, "step": 10152 }, { "epoch": 0.5135818706054934, "grad_norm": 6.052111625671387, "learning_rate": 4.863973671558931e-06, "loss": 0.3905, "step": 10153 }, { "epoch": 0.5136324548535586, "grad_norm": 6.605922698974609, "learning_rate": 4.863171356149095e-06, "loss": 0.4252, "step": 10154 }, { "epoch": 0.5136830391016237, "grad_norm": 15.631054878234863, "learning_rate": 4.862369044265004e-06, "loss": 0.5341, "step": 10155 }, { "epoch": 0.5137336233496889, "grad_norm": 7.018937587738037, "learning_rate": 4.8615667359273294e-06, "loss": 0.4001, "step": 10156 }, { "epoch": 0.513784207597754, "grad_norm": 11.78799819946289, "learning_rate": 4.8607644311567455e-06, "loss": 0.4586, "step": 10157 }, { "epoch": 0.5138347918458193, "grad_norm": 5.277341842651367, "learning_rate": 4.859962129973926e-06, "loss": 0.3677, "step": 10158 }, { "epoch": 0.5138853760938844, "grad_norm": 9.37635612487793, "learning_rate": 4.8591598323995446e-06, "loss": 0.41, "step": 10159 }, { "epoch": 0.5139359603419495, "grad_norm": 7.861697196960449, "learning_rate": 4.858357538454274e-06, "loss": 0.3898, "step": 10160 }, { "epoch": 0.5139865445900147, "grad_norm": 26.92216682434082, "learning_rate": 4.8575552481587875e-06, "loss": 0.4031, "step": 10161 }, { "epoch": 0.5140371288380798, "grad_norm": 6.930904388427734, "learning_rate": 4.8567529615337576e-06, "loss": 0.4152, "step": 10162 }, { "epoch": 0.514087713086145, "grad_norm": 5.783452033996582, "learning_rate": 4.855950678599859e-06, "loss": 0.3767, "step": 10163 }, { "epoch": 0.5141382973342101, "grad_norm": 5.1784210205078125, "learning_rate": 4.855148399377765e-06, "loss": 0.3976, "step": 10164 }, { "epoch": 0.5141888815822753, "grad_norm": 7.234426498413086, "learning_rate": 4.854346123888144e-06, "loss": 0.3677, "step": 10165 }, { "epoch": 0.5142394658303404, "grad_norm": 6.784688472747803, "learning_rate": 4.853543852151672e-06, "loss": 0.4167, "step": 10166 }, { "epoch": 0.5142900500784056, "grad_norm": 8.958209037780762, "learning_rate": 4.8527415841890215e-06, "loss": 0.3987, "step": 10167 }, { "epoch": 0.5143406343264707, "grad_norm": 8.795999526977539, "learning_rate": 4.8519393200208645e-06, "loss": 0.4612, "step": 10168 }, { "epoch": 0.5143912185745358, "grad_norm": 6.836859226226807, "learning_rate": 4.8511370596678735e-06, "loss": 0.4129, "step": 10169 }, { "epoch": 0.514441802822601, "grad_norm": 6.643554210662842, "learning_rate": 4.850334803150722e-06, "loss": 0.4096, "step": 10170 }, { "epoch": 0.5144923870706662, "grad_norm": 7.239140033721924, "learning_rate": 4.84953255049008e-06, "loss": 0.4799, "step": 10171 }, { "epoch": 0.5145429713187314, "grad_norm": 7.651954174041748, "learning_rate": 4.84873030170662e-06, "loss": 0.4619, "step": 10172 }, { "epoch": 0.5145935555667965, "grad_norm": 8.260632514953613, "learning_rate": 4.847928056821017e-06, "loss": 0.3901, "step": 10173 }, { "epoch": 0.5146441398148617, "grad_norm": 11.012079238891602, "learning_rate": 4.847125815853939e-06, "loss": 0.472, "step": 10174 }, { "epoch": 0.5146947240629268, "grad_norm": 9.49605655670166, "learning_rate": 4.846323578826059e-06, "loss": 0.3984, "step": 10175 }, { "epoch": 0.514745308310992, "grad_norm": 5.046606063842773, "learning_rate": 4.845521345758051e-06, "loss": 0.4623, "step": 10176 }, { "epoch": 0.5147958925590571, "grad_norm": 12.391059875488281, "learning_rate": 4.844719116670586e-06, "loss": 0.4392, "step": 10177 }, { "epoch": 0.5148464768071223, "grad_norm": 12.85062313079834, "learning_rate": 4.8439168915843315e-06, "loss": 0.3677, "step": 10178 }, { "epoch": 0.5148970610551874, "grad_norm": 10.4246826171875, "learning_rate": 4.843114670519962e-06, "loss": 0.417, "step": 10179 }, { "epoch": 0.5149476453032525, "grad_norm": 5.452922344207764, "learning_rate": 4.842312453498149e-06, "loss": 0.3893, "step": 10180 }, { "epoch": 0.5149982295513177, "grad_norm": 5.036471843719482, "learning_rate": 4.841510240539563e-06, "loss": 0.362, "step": 10181 }, { "epoch": 0.5150488137993828, "grad_norm": 9.75074291229248, "learning_rate": 4.840708031664874e-06, "loss": 0.3475, "step": 10182 }, { "epoch": 0.515099398047448, "grad_norm": 10.793243408203125, "learning_rate": 4.839905826894757e-06, "loss": 0.2983, "step": 10183 }, { "epoch": 0.5151499822955132, "grad_norm": 10.956767082214355, "learning_rate": 4.839103626249878e-06, "loss": 0.4797, "step": 10184 }, { "epoch": 0.5152005665435784, "grad_norm": 4.756375789642334, "learning_rate": 4.838301429750911e-06, "loss": 0.3756, "step": 10185 }, { "epoch": 0.5152511507916435, "grad_norm": 12.200736045837402, "learning_rate": 4.837499237418526e-06, "loss": 0.4116, "step": 10186 }, { "epoch": 0.5153017350397087, "grad_norm": 7.488938808441162, "learning_rate": 4.836697049273391e-06, "loss": 0.3799, "step": 10187 }, { "epoch": 0.5153523192877738, "grad_norm": 7.930645942687988, "learning_rate": 4.835894865336181e-06, "loss": 0.3589, "step": 10188 }, { "epoch": 0.515402903535839, "grad_norm": 5.0456085205078125, "learning_rate": 4.8350926856275635e-06, "loss": 0.4539, "step": 10189 }, { "epoch": 0.5154534877839041, "grad_norm": 6.99055814743042, "learning_rate": 4.83429051016821e-06, "loss": 0.4151, "step": 10190 }, { "epoch": 0.5155040720319692, "grad_norm": 9.766575813293457, "learning_rate": 4.833488338978788e-06, "loss": 0.4815, "step": 10191 }, { "epoch": 0.5155546562800344, "grad_norm": 5.92685604095459, "learning_rate": 4.83268617207997e-06, "loss": 0.4253, "step": 10192 }, { "epoch": 0.5156052405280995, "grad_norm": 11.127163887023926, "learning_rate": 4.8318840094924245e-06, "loss": 0.454, "step": 10193 }, { "epoch": 0.5156558247761647, "grad_norm": 6.338470935821533, "learning_rate": 4.8310818512368225e-06, "loss": 0.3331, "step": 10194 }, { "epoch": 0.5157064090242298, "grad_norm": 8.326360702514648, "learning_rate": 4.830279697333834e-06, "loss": 0.4655, "step": 10195 }, { "epoch": 0.515756993272295, "grad_norm": 9.608348846435547, "learning_rate": 4.8294775478041266e-06, "loss": 0.5032, "step": 10196 }, { "epoch": 0.5158075775203601, "grad_norm": 10.512133598327637, "learning_rate": 4.82867540266837e-06, "loss": 0.3515, "step": 10197 }, { "epoch": 0.5158581617684254, "grad_norm": 13.243011474609375, "learning_rate": 4.827873261947237e-06, "loss": 0.3189, "step": 10198 }, { "epoch": 0.5159087460164905, "grad_norm": 9.482030868530273, "learning_rate": 4.827071125661392e-06, "loss": 0.351, "step": 10199 }, { "epoch": 0.5159593302645556, "grad_norm": 11.440240859985352, "learning_rate": 4.826268993831507e-06, "loss": 0.415, "step": 10200 }, { "epoch": 0.5160099145126208, "grad_norm": 6.456771373748779, "learning_rate": 4.825466866478251e-06, "loss": 0.3922, "step": 10201 }, { "epoch": 0.5160604987606859, "grad_norm": 13.130891799926758, "learning_rate": 4.8246647436222925e-06, "loss": 0.6308, "step": 10202 }, { "epoch": 0.5161110830087511, "grad_norm": 5.755832195281982, "learning_rate": 4.8238626252843015e-06, "loss": 0.3952, "step": 10203 }, { "epoch": 0.5161616672568162, "grad_norm": 8.197938919067383, "learning_rate": 4.8230605114849425e-06, "loss": 0.4509, "step": 10204 }, { "epoch": 0.5162122515048814, "grad_norm": 7.928210258483887, "learning_rate": 4.822258402244888e-06, "loss": 0.4494, "step": 10205 }, { "epoch": 0.5162628357529465, "grad_norm": 5.277253150939941, "learning_rate": 4.821456297584805e-06, "loss": 0.4165, "step": 10206 }, { "epoch": 0.5163134200010117, "grad_norm": 5.962116241455078, "learning_rate": 4.82065419752536e-06, "loss": 0.3308, "step": 10207 }, { "epoch": 0.5163640042490768, "grad_norm": 10.498042106628418, "learning_rate": 4.819852102087226e-06, "loss": 0.3656, "step": 10208 }, { "epoch": 0.516414588497142, "grad_norm": 6.307939529418945, "learning_rate": 4.819050011291067e-06, "loss": 0.4153, "step": 10209 }, { "epoch": 0.5164651727452071, "grad_norm": 15.068181991577148, "learning_rate": 4.8182479251575515e-06, "loss": 0.5265, "step": 10210 }, { "epoch": 0.5165157569932723, "grad_norm": 6.545508861541748, "learning_rate": 4.81744584370735e-06, "loss": 0.3869, "step": 10211 }, { "epoch": 0.5165663412413375, "grad_norm": 18.295625686645508, "learning_rate": 4.816643766961126e-06, "loss": 0.5692, "step": 10212 }, { "epoch": 0.5166169254894026, "grad_norm": 11.752564430236816, "learning_rate": 4.815841694939549e-06, "loss": 0.4074, "step": 10213 }, { "epoch": 0.5166675097374678, "grad_norm": 7.183756351470947, "learning_rate": 4.815039627663289e-06, "loss": 0.4333, "step": 10214 }, { "epoch": 0.5167180939855329, "grad_norm": 9.73774528503418, "learning_rate": 4.814237565153009e-06, "loss": 0.5617, "step": 10215 }, { "epoch": 0.5167686782335981, "grad_norm": 8.817243576049805, "learning_rate": 4.813435507429381e-06, "loss": 0.4236, "step": 10216 }, { "epoch": 0.5168192624816632, "grad_norm": 5.726940155029297, "learning_rate": 4.812633454513067e-06, "loss": 0.3704, "step": 10217 }, { "epoch": 0.5168698467297284, "grad_norm": 9.014134407043457, "learning_rate": 4.811831406424737e-06, "loss": 0.4087, "step": 10218 }, { "epoch": 0.5169204309777935, "grad_norm": 10.447737693786621, "learning_rate": 4.8110293631850564e-06, "loss": 0.4392, "step": 10219 }, { "epoch": 0.5169710152258586, "grad_norm": 4.594350337982178, "learning_rate": 4.810227324814693e-06, "loss": 0.3925, "step": 10220 }, { "epoch": 0.5170215994739238, "grad_norm": 13.380561828613281, "learning_rate": 4.809425291334313e-06, "loss": 0.6514, "step": 10221 }, { "epoch": 0.5170721837219889, "grad_norm": 6.448160171508789, "learning_rate": 4.8086232627645835e-06, "loss": 0.4701, "step": 10222 }, { "epoch": 0.5171227679700541, "grad_norm": 8.273613929748535, "learning_rate": 4.807821239126169e-06, "loss": 0.3702, "step": 10223 }, { "epoch": 0.5171733522181193, "grad_norm": 10.789155006408691, "learning_rate": 4.8070192204397386e-06, "loss": 0.4027, "step": 10224 }, { "epoch": 0.5172239364661845, "grad_norm": 9.133283615112305, "learning_rate": 4.806217206725956e-06, "loss": 0.4452, "step": 10225 }, { "epoch": 0.5172745207142496, "grad_norm": 4.763652801513672, "learning_rate": 4.805415198005489e-06, "loss": 0.3636, "step": 10226 }, { "epoch": 0.5173251049623148, "grad_norm": 7.102541923522949, "learning_rate": 4.804613194299e-06, "loss": 0.4325, "step": 10227 }, { "epoch": 0.5173756892103799, "grad_norm": 7.821418285369873, "learning_rate": 4.803811195627161e-06, "loss": 0.4333, "step": 10228 }, { "epoch": 0.517426273458445, "grad_norm": 7.209293842315674, "learning_rate": 4.80300920201063e-06, "loss": 0.4163, "step": 10229 }, { "epoch": 0.5174768577065102, "grad_norm": 10.231245040893555, "learning_rate": 4.802207213470076e-06, "loss": 0.4866, "step": 10230 }, { "epoch": 0.5175274419545753, "grad_norm": 10.084232330322266, "learning_rate": 4.8014052300261666e-06, "loss": 0.4212, "step": 10231 }, { "epoch": 0.5175780262026405, "grad_norm": 6.834140300750732, "learning_rate": 4.800603251699563e-06, "loss": 0.3718, "step": 10232 }, { "epoch": 0.5176286104507056, "grad_norm": 9.390612602233887, "learning_rate": 4.799801278510934e-06, "loss": 0.5309, "step": 10233 }, { "epoch": 0.5176791946987708, "grad_norm": 10.534993171691895, "learning_rate": 4.79899931048094e-06, "loss": 0.3563, "step": 10234 }, { "epoch": 0.5177297789468359, "grad_norm": 4.955188274383545, "learning_rate": 4.798197347630248e-06, "loss": 0.3721, "step": 10235 }, { "epoch": 0.5177803631949011, "grad_norm": 6.452264308929443, "learning_rate": 4.797395389979525e-06, "loss": 0.367, "step": 10236 }, { "epoch": 0.5178309474429662, "grad_norm": 13.59408950805664, "learning_rate": 4.796593437549432e-06, "loss": 0.517, "step": 10237 }, { "epoch": 0.5178815316910315, "grad_norm": 11.290434837341309, "learning_rate": 4.7957914903606354e-06, "loss": 0.5806, "step": 10238 }, { "epoch": 0.5179321159390966, "grad_norm": 20.82898712158203, "learning_rate": 4.794989548433799e-06, "loss": 0.4468, "step": 10239 }, { "epoch": 0.5179827001871617, "grad_norm": 5.138726234436035, "learning_rate": 4.794187611789586e-06, "loss": 0.353, "step": 10240 }, { "epoch": 0.5180332844352269, "grad_norm": 6.676016330718994, "learning_rate": 4.793385680448664e-06, "loss": 0.4312, "step": 10241 }, { "epoch": 0.518083868683292, "grad_norm": 6.774391174316406, "learning_rate": 4.792583754431691e-06, "loss": 0.4206, "step": 10242 }, { "epoch": 0.5181344529313572, "grad_norm": 4.822888374328613, "learning_rate": 4.7917818337593355e-06, "loss": 0.3628, "step": 10243 }, { "epoch": 0.5181850371794223, "grad_norm": 15.959941864013672, "learning_rate": 4.790979918452258e-06, "loss": 0.443, "step": 10244 }, { "epoch": 0.5182356214274875, "grad_norm": 8.399356842041016, "learning_rate": 4.790178008531122e-06, "loss": 0.483, "step": 10245 }, { "epoch": 0.5182862056755526, "grad_norm": 7.174754619598389, "learning_rate": 4.789376104016593e-06, "loss": 0.41, "step": 10246 }, { "epoch": 0.5183367899236178, "grad_norm": 13.91004753112793, "learning_rate": 4.7885742049293334e-06, "loss": 0.4738, "step": 10247 }, { "epoch": 0.5183873741716829, "grad_norm": 5.2457661628723145, "learning_rate": 4.787772311290006e-06, "loss": 0.3529, "step": 10248 }, { "epoch": 0.518437958419748, "grad_norm": 9.374519348144531, "learning_rate": 4.786970423119273e-06, "loss": 0.4262, "step": 10249 }, { "epoch": 0.5184885426678132, "grad_norm": 10.556814193725586, "learning_rate": 4.786168540437798e-06, "loss": 0.5418, "step": 10250 }, { "epoch": 0.5185391269158784, "grad_norm": 6.902242660522461, "learning_rate": 4.785366663266242e-06, "loss": 0.2882, "step": 10251 }, { "epoch": 0.5185897111639436, "grad_norm": 5.950556755065918, "learning_rate": 4.78456479162527e-06, "loss": 0.4238, "step": 10252 }, { "epoch": 0.5186402954120087, "grad_norm": 6.661133289337158, "learning_rate": 4.7837629255355424e-06, "loss": 0.3793, "step": 10253 }, { "epoch": 0.5186908796600739, "grad_norm": 10.689380645751953, "learning_rate": 4.782961065017724e-06, "loss": 0.4364, "step": 10254 }, { "epoch": 0.518741463908139, "grad_norm": 7.858893871307373, "learning_rate": 4.782159210092472e-06, "loss": 0.4953, "step": 10255 }, { "epoch": 0.5187920481562042, "grad_norm": 9.27457332611084, "learning_rate": 4.781357360780453e-06, "loss": 0.3991, "step": 10256 }, { "epoch": 0.5188426324042693, "grad_norm": 9.58066463470459, "learning_rate": 4.780555517102325e-06, "loss": 0.4652, "step": 10257 }, { "epoch": 0.5188932166523345, "grad_norm": 5.873877048492432, "learning_rate": 4.779753679078751e-06, "loss": 0.3299, "step": 10258 }, { "epoch": 0.5189438009003996, "grad_norm": 5.94257116317749, "learning_rate": 4.778951846730394e-06, "loss": 0.355, "step": 10259 }, { "epoch": 0.5189943851484647, "grad_norm": 5.017705917358398, "learning_rate": 4.778150020077913e-06, "loss": 0.3845, "step": 10260 }, { "epoch": 0.5190449693965299, "grad_norm": 7.394810676574707, "learning_rate": 4.777348199141971e-06, "loss": 0.4261, "step": 10261 }, { "epoch": 0.519095553644595, "grad_norm": 6.30133581161499, "learning_rate": 4.776546383943229e-06, "loss": 0.4022, "step": 10262 }, { "epoch": 0.5191461378926602, "grad_norm": 5.588055610656738, "learning_rate": 4.775744574502345e-06, "loss": 0.3289, "step": 10263 }, { "epoch": 0.5191967221407254, "grad_norm": 9.01596736907959, "learning_rate": 4.774942770839983e-06, "loss": 0.4295, "step": 10264 }, { "epoch": 0.5192473063887906, "grad_norm": 11.537375450134277, "learning_rate": 4.774140972976801e-06, "loss": 0.4977, "step": 10265 }, { "epoch": 0.5192978906368557, "grad_norm": 5.641997814178467, "learning_rate": 4.773339180933461e-06, "loss": 0.401, "step": 10266 }, { "epoch": 0.5193484748849209, "grad_norm": 6.3416523933410645, "learning_rate": 4.772537394730624e-06, "loss": 0.4371, "step": 10267 }, { "epoch": 0.519399059132986, "grad_norm": 7.784975051879883, "learning_rate": 4.7717356143889484e-06, "loss": 0.5201, "step": 10268 }, { "epoch": 0.5194496433810512, "grad_norm": 9.136637687683105, "learning_rate": 4.7709338399290935e-06, "loss": 0.4905, "step": 10269 }, { "epoch": 0.5195002276291163, "grad_norm": 8.217427253723145, "learning_rate": 4.770132071371721e-06, "loss": 0.4184, "step": 10270 }, { "epoch": 0.5195508118771814, "grad_norm": 13.487214088439941, "learning_rate": 4.76933030873749e-06, "loss": 0.4102, "step": 10271 }, { "epoch": 0.5196013961252466, "grad_norm": 12.28852367401123, "learning_rate": 4.768528552047059e-06, "loss": 0.4413, "step": 10272 }, { "epoch": 0.5196519803733117, "grad_norm": 9.130176544189453, "learning_rate": 4.767726801321088e-06, "loss": 0.4308, "step": 10273 }, { "epoch": 0.5197025646213769, "grad_norm": 10.691414833068848, "learning_rate": 4.766925056580238e-06, "loss": 0.4535, "step": 10274 }, { "epoch": 0.519753148869442, "grad_norm": 7.805903911590576, "learning_rate": 4.766123317845165e-06, "loss": 0.3122, "step": 10275 }, { "epoch": 0.5198037331175072, "grad_norm": 7.552328109741211, "learning_rate": 4.76532158513653e-06, "loss": 0.3593, "step": 10276 }, { "epoch": 0.5198543173655724, "grad_norm": 7.903075218200684, "learning_rate": 4.76451985847499e-06, "loss": 0.4554, "step": 10277 }, { "epoch": 0.5199049016136376, "grad_norm": 7.102817535400391, "learning_rate": 4.763718137881204e-06, "loss": 0.3612, "step": 10278 }, { "epoch": 0.5199554858617027, "grad_norm": 6.429202556610107, "learning_rate": 4.762916423375832e-06, "loss": 0.3545, "step": 10279 }, { "epoch": 0.5200060701097678, "grad_norm": 5.219477653503418, "learning_rate": 4.762114714979532e-06, "loss": 0.3818, "step": 10280 }, { "epoch": 0.520056654357833, "grad_norm": 7.359745979309082, "learning_rate": 4.761313012712961e-06, "loss": 0.5132, "step": 10281 }, { "epoch": 0.5201072386058981, "grad_norm": 4.785268306732178, "learning_rate": 4.760511316596776e-06, "loss": 0.4045, "step": 10282 }, { "epoch": 0.5201578228539633, "grad_norm": 14.363387107849121, "learning_rate": 4.7597096266516355e-06, "loss": 0.3851, "step": 10283 }, { "epoch": 0.5202084071020284, "grad_norm": 6.913560390472412, "learning_rate": 4.7589079428981985e-06, "loss": 0.3642, "step": 10284 }, { "epoch": 0.5202589913500936, "grad_norm": 5.314249038696289, "learning_rate": 4.7581062653571205e-06, "loss": 0.3801, "step": 10285 }, { "epoch": 0.5203095755981587, "grad_norm": 10.399415969848633, "learning_rate": 4.75730459404906e-06, "loss": 0.576, "step": 10286 }, { "epoch": 0.5203601598462239, "grad_norm": 7.1160664558410645, "learning_rate": 4.756502928994675e-06, "loss": 0.4411, "step": 10287 }, { "epoch": 0.520410744094289, "grad_norm": 9.068716049194336, "learning_rate": 4.75570127021462e-06, "loss": 0.4913, "step": 10288 }, { "epoch": 0.5204613283423541, "grad_norm": 6.023880481719971, "learning_rate": 4.7548996177295535e-06, "loss": 0.3861, "step": 10289 }, { "epoch": 0.5205119125904193, "grad_norm": 7.418277263641357, "learning_rate": 4.754097971560133e-06, "loss": 0.4012, "step": 10290 }, { "epoch": 0.5205624968384845, "grad_norm": 7.271833419799805, "learning_rate": 4.753296331727012e-06, "loss": 0.3826, "step": 10291 }, { "epoch": 0.5206130810865497, "grad_norm": 6.192427158355713, "learning_rate": 4.752494698250849e-06, "loss": 0.3874, "step": 10292 }, { "epoch": 0.5206636653346148, "grad_norm": 7.472299098968506, "learning_rate": 4.751693071152302e-06, "loss": 0.3506, "step": 10293 }, { "epoch": 0.52071424958268, "grad_norm": 7.010046482086182, "learning_rate": 4.750891450452024e-06, "loss": 0.4388, "step": 10294 }, { "epoch": 0.5207648338307451, "grad_norm": 21.64181137084961, "learning_rate": 4.750089836170671e-06, "loss": 0.5179, "step": 10295 }, { "epoch": 0.5208154180788103, "grad_norm": 5.772072792053223, "learning_rate": 4.749288228328898e-06, "loss": 0.3823, "step": 10296 }, { "epoch": 0.5208660023268754, "grad_norm": 6.392542839050293, "learning_rate": 4.748486626947364e-06, "loss": 0.4237, "step": 10297 }, { "epoch": 0.5209165865749406, "grad_norm": 10.059220314025879, "learning_rate": 4.747685032046721e-06, "loss": 0.4299, "step": 10298 }, { "epoch": 0.5209671708230057, "grad_norm": 4.862689971923828, "learning_rate": 4.746883443647625e-06, "loss": 0.3659, "step": 10299 }, { "epoch": 0.5210177550710708, "grad_norm": 14.28610897064209, "learning_rate": 4.746081861770732e-06, "loss": 0.61, "step": 10300 }, { "epoch": 0.521068339319136, "grad_norm": 6.141342639923096, "learning_rate": 4.745280286436695e-06, "loss": 0.3867, "step": 10301 }, { "epoch": 0.5211189235672011, "grad_norm": 6.790042877197266, "learning_rate": 4.744478717666171e-06, "loss": 0.3668, "step": 10302 }, { "epoch": 0.5211695078152663, "grad_norm": 9.885299682617188, "learning_rate": 4.743677155479812e-06, "loss": 0.4573, "step": 10303 }, { "epoch": 0.5212200920633315, "grad_norm": 11.723315238952637, "learning_rate": 4.742875599898274e-06, "loss": 0.5252, "step": 10304 }, { "epoch": 0.5212706763113967, "grad_norm": 7.030248641967773, "learning_rate": 4.742074050942211e-06, "loss": 0.3258, "step": 10305 }, { "epoch": 0.5213212605594618, "grad_norm": 5.936456203460693, "learning_rate": 4.741272508632279e-06, "loss": 0.4502, "step": 10306 }, { "epoch": 0.521371844807527, "grad_norm": 7.102802753448486, "learning_rate": 4.740470972989127e-06, "loss": 0.2648, "step": 10307 }, { "epoch": 0.5214224290555921, "grad_norm": 5.349937438964844, "learning_rate": 4.739669444033411e-06, "loss": 0.3095, "step": 10308 }, { "epoch": 0.5214730133036573, "grad_norm": 5.356312274932861, "learning_rate": 4.738867921785785e-06, "loss": 0.42, "step": 10309 }, { "epoch": 0.5215235975517224, "grad_norm": 8.822232246398926, "learning_rate": 4.738066406266902e-06, "loss": 0.4963, "step": 10310 }, { "epoch": 0.5215741817997875, "grad_norm": 5.082265377044678, "learning_rate": 4.7372648974974146e-06, "loss": 0.3409, "step": 10311 }, { "epoch": 0.5216247660478527, "grad_norm": 10.05372142791748, "learning_rate": 4.736463395497977e-06, "loss": 0.3313, "step": 10312 }, { "epoch": 0.5216753502959178, "grad_norm": 6.8936381340026855, "learning_rate": 4.73566190028924e-06, "loss": 0.5048, "step": 10313 }, { "epoch": 0.521725934543983, "grad_norm": 9.769329071044922, "learning_rate": 4.734860411891858e-06, "loss": 0.446, "step": 10314 }, { "epoch": 0.5217765187920481, "grad_norm": 7.010583400726318, "learning_rate": 4.734058930326483e-06, "loss": 0.3533, "step": 10315 }, { "epoch": 0.5218271030401133, "grad_norm": 5.240139007568359, "learning_rate": 4.7332574556137664e-06, "loss": 0.4356, "step": 10316 }, { "epoch": 0.5218776872881785, "grad_norm": 7.86565637588501, "learning_rate": 4.7324559877743615e-06, "loss": 0.4095, "step": 10317 }, { "epoch": 0.5219282715362437, "grad_norm": 16.10089683532715, "learning_rate": 4.7316545268289195e-06, "loss": 0.5644, "step": 10318 }, { "epoch": 0.5219788557843088, "grad_norm": 25.489986419677734, "learning_rate": 4.730853072798095e-06, "loss": 0.368, "step": 10319 }, { "epoch": 0.522029440032374, "grad_norm": 8.3139009475708, "learning_rate": 4.7300516257025325e-06, "loss": 0.4253, "step": 10320 }, { "epoch": 0.5220800242804391, "grad_norm": 6.800834655761719, "learning_rate": 4.729250185562889e-06, "loss": 0.4334, "step": 10321 }, { "epoch": 0.5221306085285042, "grad_norm": 10.228983879089355, "learning_rate": 4.728448752399815e-06, "loss": 0.4537, "step": 10322 }, { "epoch": 0.5221811927765694, "grad_norm": 12.637307167053223, "learning_rate": 4.727647326233959e-06, "loss": 0.4185, "step": 10323 }, { "epoch": 0.5222317770246345, "grad_norm": 10.290121078491211, "learning_rate": 4.726845907085974e-06, "loss": 0.505, "step": 10324 }, { "epoch": 0.5222823612726997, "grad_norm": 8.081530570983887, "learning_rate": 4.726044494976511e-06, "loss": 0.5043, "step": 10325 }, { "epoch": 0.5223329455207648, "grad_norm": 12.030058860778809, "learning_rate": 4.725243089926219e-06, "loss": 0.4487, "step": 10326 }, { "epoch": 0.52238352976883, "grad_norm": 6.9055376052856445, "learning_rate": 4.7244416919557485e-06, "loss": 0.3666, "step": 10327 }, { "epoch": 0.5224341140168951, "grad_norm": 5.687886714935303, "learning_rate": 4.7236403010857514e-06, "loss": 0.3771, "step": 10328 }, { "epoch": 0.5224846982649602, "grad_norm": 11.44472885131836, "learning_rate": 4.722838917336875e-06, "loss": 0.3148, "step": 10329 }, { "epoch": 0.5225352825130254, "grad_norm": 8.54651165008545, "learning_rate": 4.7220375407297715e-06, "loss": 0.3839, "step": 10330 }, { "epoch": 0.5225858667610906, "grad_norm": 9.992850303649902, "learning_rate": 4.721236171285089e-06, "loss": 0.4455, "step": 10331 }, { "epoch": 0.5226364510091558, "grad_norm": 7.924637317657471, "learning_rate": 4.720434809023477e-06, "loss": 0.4405, "step": 10332 }, { "epoch": 0.5226870352572209, "grad_norm": 5.6415486335754395, "learning_rate": 4.719633453965584e-06, "loss": 0.3751, "step": 10333 }, { "epoch": 0.5227376195052861, "grad_norm": 5.127476692199707, "learning_rate": 4.71883210613206e-06, "loss": 0.3377, "step": 10334 }, { "epoch": 0.5227882037533512, "grad_norm": 6.242584228515625, "learning_rate": 4.718030765543554e-06, "loss": 0.4052, "step": 10335 }, { "epoch": 0.5228387880014164, "grad_norm": 6.338330268859863, "learning_rate": 4.717229432220713e-06, "loss": 0.3449, "step": 10336 }, { "epoch": 0.5228893722494815, "grad_norm": 6.668110370635986, "learning_rate": 4.716428106184188e-06, "loss": 0.3798, "step": 10337 }, { "epoch": 0.5229399564975467, "grad_norm": 19.26271629333496, "learning_rate": 4.715626787454624e-06, "loss": 0.4949, "step": 10338 }, { "epoch": 0.5229905407456118, "grad_norm": 10.914525985717773, "learning_rate": 4.714825476052671e-06, "loss": 0.464, "step": 10339 }, { "epoch": 0.5230411249936769, "grad_norm": 7.096123695373535, "learning_rate": 4.714024171998978e-06, "loss": 0.4637, "step": 10340 }, { "epoch": 0.5230917092417421, "grad_norm": 12.147222518920898, "learning_rate": 4.71322287531419e-06, "loss": 0.3789, "step": 10341 }, { "epoch": 0.5231422934898072, "grad_norm": 8.670419692993164, "learning_rate": 4.712421586018956e-06, "loss": 0.479, "step": 10342 }, { "epoch": 0.5231928777378724, "grad_norm": 4.618101596832275, "learning_rate": 4.711620304133924e-06, "loss": 0.383, "step": 10343 }, { "epoch": 0.5232434619859376, "grad_norm": 8.695323944091797, "learning_rate": 4.7108190296797416e-06, "loss": 0.494, "step": 10344 }, { "epoch": 0.5232940462340028, "grad_norm": 9.462397575378418, "learning_rate": 4.710017762677052e-06, "loss": 0.3911, "step": 10345 }, { "epoch": 0.5233446304820679, "grad_norm": 6.755958080291748, "learning_rate": 4.7092165031465045e-06, "loss": 0.5225, "step": 10346 }, { "epoch": 0.5233952147301331, "grad_norm": 5.124333381652832, "learning_rate": 4.708415251108746e-06, "loss": 0.397, "step": 10347 }, { "epoch": 0.5234457989781982, "grad_norm": 8.148284912109375, "learning_rate": 4.707614006584421e-06, "loss": 0.5003, "step": 10348 }, { "epoch": 0.5234963832262634, "grad_norm": 11.960585594177246, "learning_rate": 4.706812769594177e-06, "loss": 0.3943, "step": 10349 }, { "epoch": 0.5235469674743285, "grad_norm": 4.2051215171813965, "learning_rate": 4.706011540158661e-06, "loss": 0.4163, "step": 10350 }, { "epoch": 0.5235975517223936, "grad_norm": 11.433533668518066, "learning_rate": 4.705210318298517e-06, "loss": 0.4956, "step": 10351 }, { "epoch": 0.5236481359704588, "grad_norm": 5.511708736419678, "learning_rate": 4.704409104034391e-06, "loss": 0.4425, "step": 10352 }, { "epoch": 0.5236987202185239, "grad_norm": 10.758280754089355, "learning_rate": 4.703607897386928e-06, "loss": 0.4525, "step": 10353 }, { "epoch": 0.5237493044665891, "grad_norm": 4.4737420082092285, "learning_rate": 4.702806698376774e-06, "loss": 0.3268, "step": 10354 }, { "epoch": 0.5237998887146542, "grad_norm": 5.126239776611328, "learning_rate": 4.702005507024573e-06, "loss": 0.3565, "step": 10355 }, { "epoch": 0.5238504729627194, "grad_norm": 8.860404014587402, "learning_rate": 4.701204323350971e-06, "loss": 0.4425, "step": 10356 }, { "epoch": 0.5239010572107846, "grad_norm": 8.986756324768066, "learning_rate": 4.700403147376614e-06, "loss": 0.4048, "step": 10357 }, { "epoch": 0.5239516414588498, "grad_norm": 7.794259071350098, "learning_rate": 4.699601979122141e-06, "loss": 0.4065, "step": 10358 }, { "epoch": 0.5240022257069149, "grad_norm": 8.1976318359375, "learning_rate": 4.698800818608199e-06, "loss": 0.4303, "step": 10359 }, { "epoch": 0.52405280995498, "grad_norm": 6.987174034118652, "learning_rate": 4.697999665855435e-06, "loss": 0.3555, "step": 10360 }, { "epoch": 0.5241033942030452, "grad_norm": 10.556857109069824, "learning_rate": 4.697198520884488e-06, "loss": 0.4228, "step": 10361 }, { "epoch": 0.5241539784511103, "grad_norm": 6.830269813537598, "learning_rate": 4.6963973837160045e-06, "loss": 0.3875, "step": 10362 }, { "epoch": 0.5242045626991755, "grad_norm": 8.512336730957031, "learning_rate": 4.695596254370628e-06, "loss": 0.4664, "step": 10363 }, { "epoch": 0.5242551469472406, "grad_norm": 7.639380931854248, "learning_rate": 4.694795132868999e-06, "loss": 0.3574, "step": 10364 }, { "epoch": 0.5243057311953058, "grad_norm": 11.285816192626953, "learning_rate": 4.693994019231763e-06, "loss": 0.5686, "step": 10365 }, { "epoch": 0.5243563154433709, "grad_norm": 15.698262214660645, "learning_rate": 4.693192913479563e-06, "loss": 0.448, "step": 10366 }, { "epoch": 0.5244068996914361, "grad_norm": 7.031101703643799, "learning_rate": 4.692391815633039e-06, "loss": 0.3812, "step": 10367 }, { "epoch": 0.5244574839395012, "grad_norm": 4.150242328643799, "learning_rate": 4.691590725712837e-06, "loss": 0.3048, "step": 10368 }, { "epoch": 0.5245080681875663, "grad_norm": 7.709974765777588, "learning_rate": 4.690789643739595e-06, "loss": 0.4255, "step": 10369 }, { "epoch": 0.5245586524356315, "grad_norm": 12.971647262573242, "learning_rate": 4.6899885697339595e-06, "loss": 0.422, "step": 10370 }, { "epoch": 0.5246092366836967, "grad_norm": 7.732935905456543, "learning_rate": 4.689187503716567e-06, "loss": 0.4082, "step": 10371 }, { "epoch": 0.5246598209317619, "grad_norm": 5.177984714508057, "learning_rate": 4.688386445708063e-06, "loss": 0.3985, "step": 10372 }, { "epoch": 0.524710405179827, "grad_norm": 7.066631317138672, "learning_rate": 4.687585395729088e-06, "loss": 0.344, "step": 10373 }, { "epoch": 0.5247609894278922, "grad_norm": 6.261299133300781, "learning_rate": 4.686784353800281e-06, "loss": 0.3181, "step": 10374 }, { "epoch": 0.5248115736759573, "grad_norm": 7.134958267211914, "learning_rate": 4.685983319942287e-06, "loss": 0.5228, "step": 10375 }, { "epoch": 0.5248621579240225, "grad_norm": 5.29476261138916, "learning_rate": 4.685182294175742e-06, "loss": 0.4731, "step": 10376 }, { "epoch": 0.5249127421720876, "grad_norm": 8.473125457763672, "learning_rate": 4.68438127652129e-06, "loss": 0.3601, "step": 10377 }, { "epoch": 0.5249633264201528, "grad_norm": 7.752918720245361, "learning_rate": 4.6835802669995714e-06, "loss": 0.3403, "step": 10378 }, { "epoch": 0.5250139106682179, "grad_norm": 10.806048393249512, "learning_rate": 4.682779265631223e-06, "loss": 0.4685, "step": 10379 }, { "epoch": 0.525064494916283, "grad_norm": 5.797175407409668, "learning_rate": 4.681978272436887e-06, "loss": 0.4774, "step": 10380 }, { "epoch": 0.5251150791643482, "grad_norm": 5.271773338317871, "learning_rate": 4.681177287437204e-06, "loss": 0.3004, "step": 10381 }, { "epoch": 0.5251656634124133, "grad_norm": 15.509490013122559, "learning_rate": 4.680376310652811e-06, "loss": 0.4144, "step": 10382 }, { "epoch": 0.5252162476604785, "grad_norm": 8.194392204284668, "learning_rate": 4.679575342104351e-06, "loss": 0.3731, "step": 10383 }, { "epoch": 0.5252668319085437, "grad_norm": 8.315057754516602, "learning_rate": 4.678774381812457e-06, "loss": 0.3317, "step": 10384 }, { "epoch": 0.5253174161566089, "grad_norm": 7.693536758422852, "learning_rate": 4.677973429797774e-06, "loss": 0.3782, "step": 10385 }, { "epoch": 0.525368000404674, "grad_norm": 6.3024821281433105, "learning_rate": 4.677172486080937e-06, "loss": 0.4393, "step": 10386 }, { "epoch": 0.5254185846527392, "grad_norm": 6.163547992706299, "learning_rate": 4.676371550682584e-06, "loss": 0.3918, "step": 10387 }, { "epoch": 0.5254691689008043, "grad_norm": 7.184175968170166, "learning_rate": 4.675570623623356e-06, "loss": 0.4639, "step": 10388 }, { "epoch": 0.5255197531488695, "grad_norm": 13.386246681213379, "learning_rate": 4.674769704923888e-06, "loss": 0.6093, "step": 10389 }, { "epoch": 0.5255703373969346, "grad_norm": 5.9464592933654785, "learning_rate": 4.673968794604819e-06, "loss": 0.3293, "step": 10390 }, { "epoch": 0.5256209216449997, "grad_norm": 6.623647212982178, "learning_rate": 4.673167892686789e-06, "loss": 0.4134, "step": 10391 }, { "epoch": 0.5256715058930649, "grad_norm": 4.48013162612915, "learning_rate": 4.6723669991904306e-06, "loss": 0.4108, "step": 10392 }, { "epoch": 0.52572209014113, "grad_norm": 9.656803131103516, "learning_rate": 4.671566114136384e-06, "loss": 0.5654, "step": 10393 }, { "epoch": 0.5257726743891952, "grad_norm": 7.106534957885742, "learning_rate": 4.670765237545286e-06, "loss": 0.3781, "step": 10394 }, { "epoch": 0.5258232586372603, "grad_norm": 9.693117141723633, "learning_rate": 4.669964369437772e-06, "loss": 0.5292, "step": 10395 }, { "epoch": 0.5258738428853255, "grad_norm": 4.810830593109131, "learning_rate": 4.669163509834481e-06, "loss": 0.3287, "step": 10396 }, { "epoch": 0.5259244271333907, "grad_norm": 5.1158952713012695, "learning_rate": 4.668362658756045e-06, "loss": 0.2976, "step": 10397 }, { "epoch": 0.5259750113814559, "grad_norm": 6.465504169464111, "learning_rate": 4.667561816223104e-06, "loss": 0.3804, "step": 10398 }, { "epoch": 0.526025595629521, "grad_norm": 6.3455281257629395, "learning_rate": 4.66676098225629e-06, "loss": 0.4392, "step": 10399 }, { "epoch": 0.5260761798775861, "grad_norm": 6.883261203765869, "learning_rate": 4.66596015687624e-06, "loss": 0.4102, "step": 10400 }, { "epoch": 0.5261267641256513, "grad_norm": 9.829477310180664, "learning_rate": 4.665159340103592e-06, "loss": 0.4319, "step": 10401 }, { "epoch": 0.5261773483737164, "grad_norm": 10.760344505310059, "learning_rate": 4.664358531958978e-06, "loss": 0.4872, "step": 10402 }, { "epoch": 0.5262279326217816, "grad_norm": 10.906234741210938, "learning_rate": 4.663557732463035e-06, "loss": 0.3924, "step": 10403 }, { "epoch": 0.5262785168698467, "grad_norm": 11.991323471069336, "learning_rate": 4.6627569416363945e-06, "loss": 0.5932, "step": 10404 }, { "epoch": 0.5263291011179119, "grad_norm": 7.636390686035156, "learning_rate": 4.661956159499694e-06, "loss": 0.4718, "step": 10405 }, { "epoch": 0.526379685365977, "grad_norm": 7.354721546173096, "learning_rate": 4.661155386073568e-06, "loss": 0.3206, "step": 10406 }, { "epoch": 0.5264302696140422, "grad_norm": 12.035036087036133, "learning_rate": 4.660354621378647e-06, "loss": 0.4434, "step": 10407 }, { "epoch": 0.5264808538621073, "grad_norm": 6.807399272918701, "learning_rate": 4.659553865435569e-06, "loss": 0.3397, "step": 10408 }, { "epoch": 0.5265314381101724, "grad_norm": 9.915918350219727, "learning_rate": 4.658753118264966e-06, "loss": 0.4556, "step": 10409 }, { "epoch": 0.5265820223582377, "grad_norm": 7.057758808135986, "learning_rate": 4.657952379887471e-06, "loss": 0.3902, "step": 10410 }, { "epoch": 0.5266326066063028, "grad_norm": 4.627205848693848, "learning_rate": 4.6571516503237155e-06, "loss": 0.3798, "step": 10411 }, { "epoch": 0.526683190854368, "grad_norm": 7.938788414001465, "learning_rate": 4.6563509295943336e-06, "loss": 0.4323, "step": 10412 }, { "epoch": 0.5267337751024331, "grad_norm": 6.473911285400391, "learning_rate": 4.6555502177199595e-06, "loss": 0.3328, "step": 10413 }, { "epoch": 0.5267843593504983, "grad_norm": 5.780754089355469, "learning_rate": 4.654749514721223e-06, "loss": 0.4017, "step": 10414 }, { "epoch": 0.5268349435985634, "grad_norm": 10.4347562789917, "learning_rate": 4.653948820618758e-06, "loss": 0.4282, "step": 10415 }, { "epoch": 0.5268855278466286, "grad_norm": 5.960811614990234, "learning_rate": 4.6531481354331966e-06, "loss": 0.3895, "step": 10416 }, { "epoch": 0.5269361120946937, "grad_norm": 10.453632354736328, "learning_rate": 4.652347459185169e-06, "loss": 0.5169, "step": 10417 }, { "epoch": 0.5269866963427589, "grad_norm": 12.073513984680176, "learning_rate": 4.651546791895307e-06, "loss": 0.4741, "step": 10418 }, { "epoch": 0.527037280590824, "grad_norm": 6.1912970542907715, "learning_rate": 4.650746133584245e-06, "loss": 0.4349, "step": 10419 }, { "epoch": 0.5270878648388891, "grad_norm": 17.388187408447266, "learning_rate": 4.64994548427261e-06, "loss": 0.5222, "step": 10420 }, { "epoch": 0.5271384490869543, "grad_norm": 10.258012771606445, "learning_rate": 4.649144843981036e-06, "loss": 0.4688, "step": 10421 }, { "epoch": 0.5271890333350194, "grad_norm": 8.386177062988281, "learning_rate": 4.648344212730151e-06, "loss": 0.3819, "step": 10422 }, { "epoch": 0.5272396175830846, "grad_norm": 5.495429039001465, "learning_rate": 4.647543590540586e-06, "loss": 0.3281, "step": 10423 }, { "epoch": 0.5272902018311498, "grad_norm": 10.239148139953613, "learning_rate": 4.6467429774329705e-06, "loss": 0.3967, "step": 10424 }, { "epoch": 0.527340786079215, "grad_norm": 4.890440464019775, "learning_rate": 4.645942373427936e-06, "loss": 0.3281, "step": 10425 }, { "epoch": 0.5273913703272801, "grad_norm": 11.626340866088867, "learning_rate": 4.6451417785461105e-06, "loss": 0.4974, "step": 10426 }, { "epoch": 0.5274419545753453, "grad_norm": 6.357749938964844, "learning_rate": 4.644341192808125e-06, "loss": 0.3645, "step": 10427 }, { "epoch": 0.5274925388234104, "grad_norm": 6.494447708129883, "learning_rate": 4.6435406162346066e-06, "loss": 0.4886, "step": 10428 }, { "epoch": 0.5275431230714756, "grad_norm": 13.450461387634277, "learning_rate": 4.6427400488461885e-06, "loss": 0.4696, "step": 10429 }, { "epoch": 0.5275937073195407, "grad_norm": 9.364234924316406, "learning_rate": 4.641939490663493e-06, "loss": 0.4557, "step": 10430 }, { "epoch": 0.5276442915676058, "grad_norm": 7.345224380493164, "learning_rate": 4.641138941707154e-06, "loss": 0.4362, "step": 10431 }, { "epoch": 0.527694875815671, "grad_norm": 7.373775959014893, "learning_rate": 4.640338401997797e-06, "loss": 0.3653, "step": 10432 }, { "epoch": 0.5277454600637361, "grad_norm": 8.008434295654297, "learning_rate": 4.639537871556051e-06, "loss": 0.381, "step": 10433 }, { "epoch": 0.5277960443118013, "grad_norm": 8.456339836120605, "learning_rate": 4.638737350402546e-06, "loss": 0.4075, "step": 10434 }, { "epoch": 0.5278466285598664, "grad_norm": 5.97219181060791, "learning_rate": 4.6379368385579034e-06, "loss": 0.3715, "step": 10435 }, { "epoch": 0.5278972128079316, "grad_norm": 5.018878936767578, "learning_rate": 4.637136336042755e-06, "loss": 0.4092, "step": 10436 }, { "epoch": 0.5279477970559968, "grad_norm": 5.389441967010498, "learning_rate": 4.636335842877726e-06, "loss": 0.3087, "step": 10437 }, { "epoch": 0.527998381304062, "grad_norm": 9.053300857543945, "learning_rate": 4.635535359083444e-06, "loss": 0.3863, "step": 10438 }, { "epoch": 0.5280489655521271, "grad_norm": 7.090542316436768, "learning_rate": 4.6347348846805365e-06, "loss": 0.4215, "step": 10439 }, { "epoch": 0.5280995498001922, "grad_norm": 7.431148052215576, "learning_rate": 4.633934419689628e-06, "loss": 0.341, "step": 10440 }, { "epoch": 0.5281501340482574, "grad_norm": 6.410821914672852, "learning_rate": 4.633133964131346e-06, "loss": 0.4555, "step": 10441 }, { "epoch": 0.5282007182963225, "grad_norm": 8.052263259887695, "learning_rate": 4.632333518026315e-06, "loss": 0.4599, "step": 10442 }, { "epoch": 0.5282513025443877, "grad_norm": 7.70780086517334, "learning_rate": 4.631533081395161e-06, "loss": 0.4166, "step": 10443 }, { "epoch": 0.5283018867924528, "grad_norm": 8.438222885131836, "learning_rate": 4.63073265425851e-06, "loss": 0.4371, "step": 10444 }, { "epoch": 0.528352471040518, "grad_norm": 7.472383975982666, "learning_rate": 4.629932236636986e-06, "loss": 0.3732, "step": 10445 }, { "epoch": 0.5284030552885831, "grad_norm": 6.943765640258789, "learning_rate": 4.6291318285512144e-06, "loss": 0.3385, "step": 10446 }, { "epoch": 0.5284536395366483, "grad_norm": 23.594345092773438, "learning_rate": 4.628331430021822e-06, "loss": 0.4728, "step": 10447 }, { "epoch": 0.5285042237847134, "grad_norm": 6.5148138999938965, "learning_rate": 4.62753104106943e-06, "loss": 0.3535, "step": 10448 }, { "epoch": 0.5285548080327785, "grad_norm": 5.667871952056885, "learning_rate": 4.6267306617146614e-06, "loss": 0.3958, "step": 10449 }, { "epoch": 0.5286053922808438, "grad_norm": 7.240945339202881, "learning_rate": 4.6259302919781426e-06, "loss": 0.3344, "step": 10450 }, { "epoch": 0.5286559765289089, "grad_norm": 9.659480094909668, "learning_rate": 4.625129931880498e-06, "loss": 0.5174, "step": 10451 }, { "epoch": 0.5287065607769741, "grad_norm": 8.707024574279785, "learning_rate": 4.624329581442348e-06, "loss": 0.4154, "step": 10452 }, { "epoch": 0.5287571450250392, "grad_norm": 9.010648727416992, "learning_rate": 4.623529240684318e-06, "loss": 0.4424, "step": 10453 }, { "epoch": 0.5288077292731044, "grad_norm": 5.469205379486084, "learning_rate": 4.622728909627031e-06, "loss": 0.3578, "step": 10454 }, { "epoch": 0.5288583135211695, "grad_norm": 8.415434837341309, "learning_rate": 4.621928588291107e-06, "loss": 0.579, "step": 10455 }, { "epoch": 0.5289088977692347, "grad_norm": 9.335577964782715, "learning_rate": 4.621128276697171e-06, "loss": 0.5285, "step": 10456 }, { "epoch": 0.5289594820172998, "grad_norm": 8.009512901306152, "learning_rate": 4.620327974865844e-06, "loss": 0.4149, "step": 10457 }, { "epoch": 0.529010066265365, "grad_norm": 5.368649959564209, "learning_rate": 4.619527682817749e-06, "loss": 0.4097, "step": 10458 }, { "epoch": 0.5290606505134301, "grad_norm": 10.328878402709961, "learning_rate": 4.618727400573505e-06, "loss": 0.5739, "step": 10459 }, { "epoch": 0.5291112347614952, "grad_norm": 9.083451271057129, "learning_rate": 4.617927128153738e-06, "loss": 0.4112, "step": 10460 }, { "epoch": 0.5291618190095604, "grad_norm": 6.849255561828613, "learning_rate": 4.6171268655790645e-06, "loss": 0.2891, "step": 10461 }, { "epoch": 0.5292124032576255, "grad_norm": 8.839445114135742, "learning_rate": 4.616326612870106e-06, "loss": 0.3272, "step": 10462 }, { "epoch": 0.5292629875056907, "grad_norm": 7.9503350257873535, "learning_rate": 4.615526370047484e-06, "loss": 0.4034, "step": 10463 }, { "epoch": 0.5293135717537559, "grad_norm": 4.910615921020508, "learning_rate": 4.614726137131819e-06, "loss": 0.3843, "step": 10464 }, { "epoch": 0.5293641560018211, "grad_norm": 13.080726623535156, "learning_rate": 4.61392591414373e-06, "loss": 0.38, "step": 10465 }, { "epoch": 0.5294147402498862, "grad_norm": 9.492815017700195, "learning_rate": 4.6131257011038385e-06, "loss": 0.5508, "step": 10466 }, { "epoch": 0.5294653244979514, "grad_norm": 10.04240608215332, "learning_rate": 4.612325498032764e-06, "loss": 0.3921, "step": 10467 }, { "epoch": 0.5295159087460165, "grad_norm": 7.613317012786865, "learning_rate": 4.611525304951123e-06, "loss": 0.4688, "step": 10468 }, { "epoch": 0.5295664929940817, "grad_norm": 20.657291412353516, "learning_rate": 4.6107251218795375e-06, "loss": 0.53, "step": 10469 }, { "epoch": 0.5296170772421468, "grad_norm": 7.208471775054932, "learning_rate": 4.609924948838627e-06, "loss": 0.3354, "step": 10470 }, { "epoch": 0.5296676614902119, "grad_norm": 6.90926456451416, "learning_rate": 4.609124785849007e-06, "loss": 0.4111, "step": 10471 }, { "epoch": 0.5297182457382771, "grad_norm": 11.192066192626953, "learning_rate": 4.6083246329312975e-06, "loss": 0.5244, "step": 10472 }, { "epoch": 0.5297688299863422, "grad_norm": 9.082680702209473, "learning_rate": 4.607524490106116e-06, "loss": 0.3871, "step": 10473 }, { "epoch": 0.5298194142344074, "grad_norm": 10.03212833404541, "learning_rate": 4.606724357394081e-06, "loss": 0.4825, "step": 10474 }, { "epoch": 0.5298699984824725, "grad_norm": 6.807013034820557, "learning_rate": 4.605924234815809e-06, "loss": 0.4229, "step": 10475 }, { "epoch": 0.5299205827305377, "grad_norm": 7.113275527954102, "learning_rate": 4.605124122391917e-06, "loss": 0.4277, "step": 10476 }, { "epoch": 0.5299711669786029, "grad_norm": 8.50339126586914, "learning_rate": 4.604324020143023e-06, "loss": 0.3864, "step": 10477 }, { "epoch": 0.5300217512266681, "grad_norm": 9.744397163391113, "learning_rate": 4.603523928089743e-06, "loss": 0.4371, "step": 10478 }, { "epoch": 0.5300723354747332, "grad_norm": 9.241806030273438, "learning_rate": 4.602723846252694e-06, "loss": 0.4299, "step": 10479 }, { "epoch": 0.5301229197227983, "grad_norm": 5.486390113830566, "learning_rate": 4.6019237746524915e-06, "loss": 0.4497, "step": 10480 }, { "epoch": 0.5301735039708635, "grad_norm": 11.122154235839844, "learning_rate": 4.601123713309751e-06, "loss": 0.462, "step": 10481 }, { "epoch": 0.5302240882189286, "grad_norm": 5.0042924880981445, "learning_rate": 4.600323662245092e-06, "loss": 0.4461, "step": 10482 }, { "epoch": 0.5302746724669938, "grad_norm": 23.28680419921875, "learning_rate": 4.599523621479124e-06, "loss": 0.4511, "step": 10483 }, { "epoch": 0.5303252567150589, "grad_norm": 6.16456413269043, "learning_rate": 4.598723591032465e-06, "loss": 0.4407, "step": 10484 }, { "epoch": 0.5303758409631241, "grad_norm": 6.643478870391846, "learning_rate": 4.597923570925731e-06, "loss": 0.2776, "step": 10485 }, { "epoch": 0.5304264252111892, "grad_norm": 7.350184917449951, "learning_rate": 4.597123561179537e-06, "loss": 0.4178, "step": 10486 }, { "epoch": 0.5304770094592544, "grad_norm": 7.775844097137451, "learning_rate": 4.596323561814494e-06, "loss": 0.4134, "step": 10487 }, { "epoch": 0.5305275937073195, "grad_norm": 8.77013111114502, "learning_rate": 4.595523572851217e-06, "loss": 0.4919, "step": 10488 }, { "epoch": 0.5305781779553846, "grad_norm": 13.531841278076172, "learning_rate": 4.594723594310322e-06, "loss": 0.3997, "step": 10489 }, { "epoch": 0.5306287622034499, "grad_norm": 11.153297424316406, "learning_rate": 4.593923626212419e-06, "loss": 0.474, "step": 10490 }, { "epoch": 0.530679346451515, "grad_norm": 6.89568567276001, "learning_rate": 4.5931236685781246e-06, "loss": 0.3994, "step": 10491 }, { "epoch": 0.5307299306995802, "grad_norm": 8.582483291625977, "learning_rate": 4.592323721428051e-06, "loss": 0.3226, "step": 10492 }, { "epoch": 0.5307805149476453, "grad_norm": 10.899873733520508, "learning_rate": 4.591523784782809e-06, "loss": 0.3949, "step": 10493 }, { "epoch": 0.5308310991957105, "grad_norm": 6.071742057800293, "learning_rate": 4.590723858663013e-06, "loss": 0.4338, "step": 10494 }, { "epoch": 0.5308816834437756, "grad_norm": 12.285541534423828, "learning_rate": 4.589923943089276e-06, "loss": 0.4754, "step": 10495 }, { "epoch": 0.5309322676918408, "grad_norm": 17.129806518554688, "learning_rate": 4.589124038082206e-06, "loss": 0.4587, "step": 10496 }, { "epoch": 0.5309828519399059, "grad_norm": 7.403533458709717, "learning_rate": 4.5883241436624185e-06, "loss": 0.385, "step": 10497 }, { "epoch": 0.531033436187971, "grad_norm": 10.165214538574219, "learning_rate": 4.587524259850524e-06, "loss": 0.506, "step": 10498 }, { "epoch": 0.5310840204360362, "grad_norm": 7.345599174499512, "learning_rate": 4.586724386667134e-06, "loss": 0.4428, "step": 10499 }, { "epoch": 0.5311346046841013, "grad_norm": 14.351778984069824, "learning_rate": 4.585924524132856e-06, "loss": 0.4318, "step": 10500 }, { "epoch": 0.5311851889321665, "grad_norm": 6.2982282638549805, "learning_rate": 4.585124672268304e-06, "loss": 0.3851, "step": 10501 }, { "epoch": 0.5312357731802316, "grad_norm": 7.442999362945557, "learning_rate": 4.584324831094087e-06, "loss": 0.4281, "step": 10502 }, { "epoch": 0.5312863574282968, "grad_norm": 9.549666404724121, "learning_rate": 4.583525000630814e-06, "loss": 0.3992, "step": 10503 }, { "epoch": 0.531336941676362, "grad_norm": 9.294168472290039, "learning_rate": 4.582725180899096e-06, "loss": 0.44, "step": 10504 }, { "epoch": 0.5313875259244272, "grad_norm": 10.403459548950195, "learning_rate": 4.581925371919544e-06, "loss": 0.4178, "step": 10505 }, { "epoch": 0.5314381101724923, "grad_norm": 12.283903121948242, "learning_rate": 4.581125573712763e-06, "loss": 0.4411, "step": 10506 }, { "epoch": 0.5314886944205575, "grad_norm": 7.708209991455078, "learning_rate": 4.580325786299367e-06, "loss": 0.399, "step": 10507 }, { "epoch": 0.5315392786686226, "grad_norm": 8.657859802246094, "learning_rate": 4.579526009699959e-06, "loss": 0.4673, "step": 10508 }, { "epoch": 0.5315898629166877, "grad_norm": 10.053086280822754, "learning_rate": 4.578726243935151e-06, "loss": 0.4385, "step": 10509 }, { "epoch": 0.5316404471647529, "grad_norm": 11.211993217468262, "learning_rate": 4.577926489025551e-06, "loss": 0.4942, "step": 10510 }, { "epoch": 0.531691031412818, "grad_norm": 6.146015167236328, "learning_rate": 4.577126744991765e-06, "loss": 0.3442, "step": 10511 }, { "epoch": 0.5317416156608832, "grad_norm": 7.014988899230957, "learning_rate": 4.576327011854403e-06, "loss": 0.3524, "step": 10512 }, { "epoch": 0.5317921999089483, "grad_norm": 7.965632438659668, "learning_rate": 4.575527289634069e-06, "loss": 0.4839, "step": 10513 }, { "epoch": 0.5318427841570135, "grad_norm": 11.682513236999512, "learning_rate": 4.5747275783513724e-06, "loss": 0.5416, "step": 10514 }, { "epoch": 0.5318933684050786, "grad_norm": 5.995893478393555, "learning_rate": 4.573927878026918e-06, "loss": 0.3765, "step": 10515 }, { "epoch": 0.5319439526531438, "grad_norm": 5.093147277832031, "learning_rate": 4.573128188681313e-06, "loss": 0.4395, "step": 10516 }, { "epoch": 0.531994536901209, "grad_norm": 5.679017066955566, "learning_rate": 4.572328510335164e-06, "loss": 0.4197, "step": 10517 }, { "epoch": 0.5320451211492742, "grad_norm": 6.631054878234863, "learning_rate": 4.571528843009075e-06, "loss": 0.355, "step": 10518 }, { "epoch": 0.5320957053973393, "grad_norm": 7.522731781005859, "learning_rate": 4.570729186723654e-06, "loss": 0.4193, "step": 10519 }, { "epoch": 0.5321462896454044, "grad_norm": 4.945610523223877, "learning_rate": 4.569929541499505e-06, "loss": 0.3078, "step": 10520 }, { "epoch": 0.5321968738934696, "grad_norm": 6.439002513885498, "learning_rate": 4.569129907357233e-06, "loss": 0.4208, "step": 10521 }, { "epoch": 0.5322474581415347, "grad_norm": 7.5015764236450195, "learning_rate": 4.568330284317442e-06, "loss": 0.5082, "step": 10522 }, { "epoch": 0.5322980423895999, "grad_norm": 4.521364688873291, "learning_rate": 4.567530672400737e-06, "loss": 0.4368, "step": 10523 }, { "epoch": 0.532348626637665, "grad_norm": 12.781050682067871, "learning_rate": 4.566731071627725e-06, "loss": 0.4667, "step": 10524 }, { "epoch": 0.5323992108857302, "grad_norm": 12.300623893737793, "learning_rate": 4.5659314820190034e-06, "loss": 0.5435, "step": 10525 }, { "epoch": 0.5324497951337953, "grad_norm": 8.675545692443848, "learning_rate": 4.5651319035951795e-06, "loss": 0.2962, "step": 10526 }, { "epoch": 0.5325003793818605, "grad_norm": 5.091698169708252, "learning_rate": 4.564332336376857e-06, "loss": 0.3804, "step": 10527 }, { "epoch": 0.5325509636299256, "grad_norm": 9.21939754486084, "learning_rate": 4.5635327803846355e-06, "loss": 0.5674, "step": 10528 }, { "epoch": 0.5326015478779907, "grad_norm": 12.091937065124512, "learning_rate": 4.562733235639122e-06, "loss": 0.4884, "step": 10529 }, { "epoch": 0.532652132126056, "grad_norm": 8.383599281311035, "learning_rate": 4.561933702160917e-06, "loss": 0.4213, "step": 10530 }, { "epoch": 0.5327027163741211, "grad_norm": 5.642690658569336, "learning_rate": 4.561134179970621e-06, "loss": 0.398, "step": 10531 }, { "epoch": 0.5327533006221863, "grad_norm": 5.783313274383545, "learning_rate": 4.560334669088838e-06, "loss": 0.4166, "step": 10532 }, { "epoch": 0.5328038848702514, "grad_norm": 8.811098098754883, "learning_rate": 4.559535169536169e-06, "loss": 0.4409, "step": 10533 }, { "epoch": 0.5328544691183166, "grad_norm": 7.125101089477539, "learning_rate": 4.558735681333214e-06, "loss": 0.4451, "step": 10534 }, { "epoch": 0.5329050533663817, "grad_norm": 7.780281066894531, "learning_rate": 4.557936204500574e-06, "loss": 0.4541, "step": 10535 }, { "epoch": 0.5329556376144469, "grad_norm": 5.93162727355957, "learning_rate": 4.557136739058852e-06, "loss": 0.4273, "step": 10536 }, { "epoch": 0.533006221862512, "grad_norm": 10.495646476745605, "learning_rate": 4.556337285028648e-06, "loss": 0.5232, "step": 10537 }, { "epoch": 0.5330568061105772, "grad_norm": 6.947704792022705, "learning_rate": 4.555537842430558e-06, "loss": 0.4092, "step": 10538 }, { "epoch": 0.5331073903586423, "grad_norm": 7.925848484039307, "learning_rate": 4.554738411285184e-06, "loss": 0.503, "step": 10539 }, { "epoch": 0.5331579746067074, "grad_norm": 6.632322788238525, "learning_rate": 4.553938991613127e-06, "loss": 0.4234, "step": 10540 }, { "epoch": 0.5332085588547726, "grad_norm": 6.609585285186768, "learning_rate": 4.553139583434983e-06, "loss": 0.4626, "step": 10541 }, { "epoch": 0.5332591431028377, "grad_norm": 8.552027702331543, "learning_rate": 4.552340186771353e-06, "loss": 0.3434, "step": 10542 }, { "epoch": 0.5333097273509029, "grad_norm": 9.642379760742188, "learning_rate": 4.551540801642837e-06, "loss": 0.3727, "step": 10543 }, { "epoch": 0.5333603115989681, "grad_norm": 20.134763717651367, "learning_rate": 4.5507414280700296e-06, "loss": 0.4751, "step": 10544 }, { "epoch": 0.5334108958470333, "grad_norm": 23.327890396118164, "learning_rate": 4.549942066073532e-06, "loss": 0.5991, "step": 10545 }, { "epoch": 0.5334614800950984, "grad_norm": 6.112835884094238, "learning_rate": 4.549142715673939e-06, "loss": 0.3856, "step": 10546 }, { "epoch": 0.5335120643431636, "grad_norm": 5.442958354949951, "learning_rate": 4.5483433768918486e-06, "loss": 0.4305, "step": 10547 }, { "epoch": 0.5335626485912287, "grad_norm": 16.661287307739258, "learning_rate": 4.54754404974786e-06, "loss": 0.5721, "step": 10548 }, { "epoch": 0.5336132328392938, "grad_norm": 8.77371597290039, "learning_rate": 4.546744734262567e-06, "loss": 0.4228, "step": 10549 }, { "epoch": 0.533663817087359, "grad_norm": 6.178988456726074, "learning_rate": 4.54594543045657e-06, "loss": 0.3679, "step": 10550 }, { "epoch": 0.5337144013354241, "grad_norm": 8.34699821472168, "learning_rate": 4.54514613835046e-06, "loss": 0.465, "step": 10551 }, { "epoch": 0.5337649855834893, "grad_norm": 6.946324348449707, "learning_rate": 4.5443468579648365e-06, "loss": 0.4955, "step": 10552 }, { "epoch": 0.5338155698315544, "grad_norm": 10.685644149780273, "learning_rate": 4.543547589320293e-06, "loss": 0.5072, "step": 10553 }, { "epoch": 0.5338661540796196, "grad_norm": 5.5889458656311035, "learning_rate": 4.542748332437426e-06, "loss": 0.5018, "step": 10554 }, { "epoch": 0.5339167383276847, "grad_norm": 5.481109619140625, "learning_rate": 4.54194908733683e-06, "loss": 0.426, "step": 10555 }, { "epoch": 0.5339673225757499, "grad_norm": 6.033810615539551, "learning_rate": 4.541149854039099e-06, "loss": 0.3832, "step": 10556 }, { "epoch": 0.5340179068238151, "grad_norm": 5.690207004547119, "learning_rate": 4.540350632564828e-06, "loss": 0.4632, "step": 10557 }, { "epoch": 0.5340684910718803, "grad_norm": 5.579322814941406, "learning_rate": 4.539551422934613e-06, "loss": 0.3815, "step": 10558 }, { "epoch": 0.5341190753199454, "grad_norm": 11.867680549621582, "learning_rate": 4.5387522251690435e-06, "loss": 0.4615, "step": 10559 }, { "epoch": 0.5341696595680105, "grad_norm": 6.131507873535156, "learning_rate": 4.537953039288715e-06, "loss": 0.3959, "step": 10560 }, { "epoch": 0.5342202438160757, "grad_norm": 5.842953681945801, "learning_rate": 4.537153865314222e-06, "loss": 0.4617, "step": 10561 }, { "epoch": 0.5342708280641408, "grad_norm": 7.042934894561768, "learning_rate": 4.5363547032661555e-06, "loss": 0.437, "step": 10562 }, { "epoch": 0.534321412312206, "grad_norm": 7.755341053009033, "learning_rate": 4.53555555316511e-06, "loss": 0.3631, "step": 10563 }, { "epoch": 0.5343719965602711, "grad_norm": 6.922598838806152, "learning_rate": 4.534756415031673e-06, "loss": 0.452, "step": 10564 }, { "epoch": 0.5344225808083363, "grad_norm": 4.7723708152771, "learning_rate": 4.533957288886442e-06, "loss": 0.3238, "step": 10565 }, { "epoch": 0.5344731650564014, "grad_norm": 6.861602306365967, "learning_rate": 4.533158174750004e-06, "loss": 0.3867, "step": 10566 }, { "epoch": 0.5345237493044666, "grad_norm": 8.89985466003418, "learning_rate": 4.532359072642953e-06, "loss": 0.3948, "step": 10567 }, { "epoch": 0.5345743335525317, "grad_norm": 12.00377368927002, "learning_rate": 4.53155998258588e-06, "loss": 0.42, "step": 10568 }, { "epoch": 0.5346249178005968, "grad_norm": 21.653053283691406, "learning_rate": 4.530760904599373e-06, "loss": 0.464, "step": 10569 }, { "epoch": 0.5346755020486621, "grad_norm": 8.272324562072754, "learning_rate": 4.529961838704025e-06, "loss": 0.4579, "step": 10570 }, { "epoch": 0.5347260862967272, "grad_norm": 9.436948776245117, "learning_rate": 4.529162784920425e-06, "loss": 0.4346, "step": 10571 }, { "epoch": 0.5347766705447924, "grad_norm": 10.172754287719727, "learning_rate": 4.5283637432691625e-06, "loss": 0.4499, "step": 10572 }, { "epoch": 0.5348272547928575, "grad_norm": 14.059096336364746, "learning_rate": 4.527564713770827e-06, "loss": 0.4277, "step": 10573 }, { "epoch": 0.5348778390409227, "grad_norm": 6.04686975479126, "learning_rate": 4.526765696446009e-06, "loss": 0.3797, "step": 10574 }, { "epoch": 0.5349284232889878, "grad_norm": 9.615606307983398, "learning_rate": 4.525966691315294e-06, "loss": 0.383, "step": 10575 }, { "epoch": 0.534979007537053, "grad_norm": 11.188002586364746, "learning_rate": 4.525167698399276e-06, "loss": 0.4415, "step": 10576 }, { "epoch": 0.5350295917851181, "grad_norm": 16.241209030151367, "learning_rate": 4.5243687177185365e-06, "loss": 0.4824, "step": 10577 }, { "epoch": 0.5350801760331833, "grad_norm": 8.833191871643066, "learning_rate": 4.523569749293668e-06, "loss": 0.4586, "step": 10578 }, { "epoch": 0.5351307602812484, "grad_norm": 5.969705581665039, "learning_rate": 4.522770793145254e-06, "loss": 0.4325, "step": 10579 }, { "epoch": 0.5351813445293135, "grad_norm": 13.790863990783691, "learning_rate": 4.521971849293885e-06, "loss": 0.4693, "step": 10580 }, { "epoch": 0.5352319287773787, "grad_norm": 7.404759883880615, "learning_rate": 4.521172917760148e-06, "loss": 0.43, "step": 10581 }, { "epoch": 0.5352825130254438, "grad_norm": 15.604403495788574, "learning_rate": 4.520373998564626e-06, "loss": 0.4646, "step": 10582 }, { "epoch": 0.5353330972735091, "grad_norm": 25.081830978393555, "learning_rate": 4.51957509172791e-06, "loss": 0.4734, "step": 10583 }, { "epoch": 0.5353836815215742, "grad_norm": 10.92766284942627, "learning_rate": 4.518776197270582e-06, "loss": 0.579, "step": 10584 }, { "epoch": 0.5354342657696394, "grad_norm": 5.200039863586426, "learning_rate": 4.5179773152132284e-06, "loss": 0.4183, "step": 10585 }, { "epoch": 0.5354848500177045, "grad_norm": 8.127065658569336, "learning_rate": 4.517178445576437e-06, "loss": 0.4532, "step": 10586 }, { "epoch": 0.5355354342657697, "grad_norm": 6.261524200439453, "learning_rate": 4.516379588380789e-06, "loss": 0.3943, "step": 10587 }, { "epoch": 0.5355860185138348, "grad_norm": 6.888862609863281, "learning_rate": 4.515580743646871e-06, "loss": 0.4688, "step": 10588 }, { "epoch": 0.5356366027619, "grad_norm": 5.368368148803711, "learning_rate": 4.51478191139527e-06, "loss": 0.4125, "step": 10589 }, { "epoch": 0.5356871870099651, "grad_norm": 6.691574573516846, "learning_rate": 4.513983091646566e-06, "loss": 0.4993, "step": 10590 }, { "epoch": 0.5357377712580302, "grad_norm": 10.827692031860352, "learning_rate": 4.5131842844213426e-06, "loss": 0.4803, "step": 10591 }, { "epoch": 0.5357883555060954, "grad_norm": 9.224287033081055, "learning_rate": 4.512385489740184e-06, "loss": 0.434, "step": 10592 }, { "epoch": 0.5358389397541605, "grad_norm": 12.858952522277832, "learning_rate": 4.511586707623675e-06, "loss": 0.4486, "step": 10593 }, { "epoch": 0.5358895240022257, "grad_norm": 11.029178619384766, "learning_rate": 4.510787938092395e-06, "loss": 0.3325, "step": 10594 }, { "epoch": 0.5359401082502908, "grad_norm": 11.379203796386719, "learning_rate": 4.509989181166929e-06, "loss": 0.5728, "step": 10595 }, { "epoch": 0.535990692498356, "grad_norm": 8.00143814086914, "learning_rate": 4.509190436867859e-06, "loss": 0.4271, "step": 10596 }, { "epoch": 0.5360412767464212, "grad_norm": 7.795378684997559, "learning_rate": 4.508391705215765e-06, "loss": 0.3734, "step": 10597 }, { "epoch": 0.5360918609944864, "grad_norm": 6.0183587074279785, "learning_rate": 4.5075929862312294e-06, "loss": 0.4648, "step": 10598 }, { "epoch": 0.5361424452425515, "grad_norm": 14.147825241088867, "learning_rate": 4.506794279934834e-06, "loss": 0.5949, "step": 10599 }, { "epoch": 0.5361930294906166, "grad_norm": 5.994547367095947, "learning_rate": 4.505995586347158e-06, "loss": 0.3768, "step": 10600 }, { "epoch": 0.5362436137386818, "grad_norm": 10.134125709533691, "learning_rate": 4.505196905488783e-06, "loss": 0.4171, "step": 10601 }, { "epoch": 0.5362941979867469, "grad_norm": 6.561859130859375, "learning_rate": 4.504398237380289e-06, "loss": 0.3938, "step": 10602 }, { "epoch": 0.5363447822348121, "grad_norm": 3.9526052474975586, "learning_rate": 4.503599582042255e-06, "loss": 0.3019, "step": 10603 }, { "epoch": 0.5363953664828772, "grad_norm": 8.848518371582031, "learning_rate": 4.502800939495261e-06, "loss": 0.4378, "step": 10604 }, { "epoch": 0.5364459507309424, "grad_norm": 6.589232921600342, "learning_rate": 4.502002309759884e-06, "loss": 0.458, "step": 10605 }, { "epoch": 0.5364965349790075, "grad_norm": 7.237224102020264, "learning_rate": 4.5012036928567065e-06, "loss": 0.4673, "step": 10606 }, { "epoch": 0.5365471192270727, "grad_norm": 5.759692668914795, "learning_rate": 4.500405088806304e-06, "loss": 0.4246, "step": 10607 }, { "epoch": 0.5365977034751378, "grad_norm": 6.2587971687316895, "learning_rate": 4.499606497629256e-06, "loss": 0.4389, "step": 10608 }, { "epoch": 0.5366482877232029, "grad_norm": 6.384571075439453, "learning_rate": 4.4988079193461405e-06, "loss": 0.3861, "step": 10609 }, { "epoch": 0.5366988719712682, "grad_norm": 11.396997451782227, "learning_rate": 4.498009353977533e-06, "loss": 0.4737, "step": 10610 }, { "epoch": 0.5367494562193333, "grad_norm": 13.129639625549316, "learning_rate": 4.4972108015440124e-06, "loss": 0.6003, "step": 10611 }, { "epoch": 0.5368000404673985, "grad_norm": 14.399497032165527, "learning_rate": 4.496412262066156e-06, "loss": 0.4684, "step": 10612 }, { "epoch": 0.5368506247154636, "grad_norm": 12.92122745513916, "learning_rate": 4.495613735564538e-06, "loss": 0.4382, "step": 10613 }, { "epoch": 0.5369012089635288, "grad_norm": 7.867135524749756, "learning_rate": 4.494815222059739e-06, "loss": 0.3063, "step": 10614 }, { "epoch": 0.5369517932115939, "grad_norm": 10.145807266235352, "learning_rate": 4.494016721572328e-06, "loss": 0.3633, "step": 10615 }, { "epoch": 0.5370023774596591, "grad_norm": 5.583086967468262, "learning_rate": 4.493218234122885e-06, "loss": 0.3999, "step": 10616 }, { "epoch": 0.5370529617077242, "grad_norm": 7.597197532653809, "learning_rate": 4.492419759731984e-06, "loss": 0.4105, "step": 10617 }, { "epoch": 0.5371035459557894, "grad_norm": 6.530388832092285, "learning_rate": 4.4916212984202e-06, "loss": 0.4332, "step": 10618 }, { "epoch": 0.5371541302038545, "grad_norm": 8.556020736694336, "learning_rate": 4.490822850208107e-06, "loss": 0.3951, "step": 10619 }, { "epoch": 0.5372047144519196, "grad_norm": 8.310088157653809, "learning_rate": 4.490024415116278e-06, "loss": 0.4375, "step": 10620 }, { "epoch": 0.5372552986999848, "grad_norm": 12.179251670837402, "learning_rate": 4.489225993165289e-06, "loss": 0.463, "step": 10621 }, { "epoch": 0.5373058829480499, "grad_norm": 10.203583717346191, "learning_rate": 4.488427584375713e-06, "loss": 0.4676, "step": 10622 }, { "epoch": 0.5373564671961152, "grad_norm": 7.898662090301514, "learning_rate": 4.487629188768121e-06, "loss": 0.4443, "step": 10623 }, { "epoch": 0.5374070514441803, "grad_norm": 9.295134544372559, "learning_rate": 4.486830806363089e-06, "loss": 0.4771, "step": 10624 }, { "epoch": 0.5374576356922455, "grad_norm": 5.203171730041504, "learning_rate": 4.486032437181186e-06, "loss": 0.3948, "step": 10625 }, { "epoch": 0.5375082199403106, "grad_norm": 6.194190502166748, "learning_rate": 4.485234081242986e-06, "loss": 0.3466, "step": 10626 }, { "epoch": 0.5375588041883758, "grad_norm": 6.804534912109375, "learning_rate": 4.484435738569063e-06, "loss": 0.3989, "step": 10627 }, { "epoch": 0.5376093884364409, "grad_norm": 9.119612693786621, "learning_rate": 4.4836374091799846e-06, "loss": 0.4725, "step": 10628 }, { "epoch": 0.537659972684506, "grad_norm": 7.326145648956299, "learning_rate": 4.482839093096321e-06, "loss": 0.4025, "step": 10629 }, { "epoch": 0.5377105569325712, "grad_norm": 12.677690505981445, "learning_rate": 4.482040790338645e-06, "loss": 0.4866, "step": 10630 }, { "epoch": 0.5377611411806363, "grad_norm": 6.511077880859375, "learning_rate": 4.481242500927528e-06, "loss": 0.3243, "step": 10631 }, { "epoch": 0.5378117254287015, "grad_norm": 9.012083053588867, "learning_rate": 4.480444224883538e-06, "loss": 0.4668, "step": 10632 }, { "epoch": 0.5378623096767666, "grad_norm": 7.261988162994385, "learning_rate": 4.479645962227246e-06, "loss": 0.4265, "step": 10633 }, { "epoch": 0.5379128939248318, "grad_norm": 10.80502986907959, "learning_rate": 4.4788477129792206e-06, "loss": 0.4319, "step": 10634 }, { "epoch": 0.5379634781728969, "grad_norm": 6.946678161621094, "learning_rate": 4.47804947716003e-06, "loss": 0.4216, "step": 10635 }, { "epoch": 0.5380140624209621, "grad_norm": 9.618415832519531, "learning_rate": 4.477251254790244e-06, "loss": 0.4397, "step": 10636 }, { "epoch": 0.5380646466690273, "grad_norm": 7.459270477294922, "learning_rate": 4.476453045890432e-06, "loss": 0.4026, "step": 10637 }, { "epoch": 0.5381152309170925, "grad_norm": 7.211214065551758, "learning_rate": 4.4756548504811584e-06, "loss": 0.3652, "step": 10638 }, { "epoch": 0.5381658151651576, "grad_norm": 7.094802379608154, "learning_rate": 4.474856668582993e-06, "loss": 0.3901, "step": 10639 }, { "epoch": 0.5382163994132227, "grad_norm": 5.646320819854736, "learning_rate": 4.474058500216505e-06, "loss": 0.3786, "step": 10640 }, { "epoch": 0.5382669836612879, "grad_norm": 10.227884292602539, "learning_rate": 4.473260345402258e-06, "loss": 0.4493, "step": 10641 }, { "epoch": 0.538317567909353, "grad_norm": 6.954123497009277, "learning_rate": 4.472462204160818e-06, "loss": 0.4298, "step": 10642 }, { "epoch": 0.5383681521574182, "grad_norm": 9.319540977478027, "learning_rate": 4.4716640765127535e-06, "loss": 0.3552, "step": 10643 }, { "epoch": 0.5384187364054833, "grad_norm": 7.362995147705078, "learning_rate": 4.470865962478631e-06, "loss": 0.4363, "step": 10644 }, { "epoch": 0.5384693206535485, "grad_norm": 12.720049858093262, "learning_rate": 4.470067862079013e-06, "loss": 0.5714, "step": 10645 }, { "epoch": 0.5385199049016136, "grad_norm": 24.031713485717773, "learning_rate": 4.469269775334467e-06, "loss": 0.4323, "step": 10646 }, { "epoch": 0.5385704891496788, "grad_norm": 21.483877182006836, "learning_rate": 4.468471702265557e-06, "loss": 0.4378, "step": 10647 }, { "epoch": 0.5386210733977439, "grad_norm": 5.495678901672363, "learning_rate": 4.467673642892847e-06, "loss": 0.3521, "step": 10648 }, { "epoch": 0.538671657645809, "grad_norm": 6.547903537750244, "learning_rate": 4.466875597236901e-06, "loss": 0.3082, "step": 10649 }, { "epoch": 0.5387222418938743, "grad_norm": 5.925629138946533, "learning_rate": 4.466077565318284e-06, "loss": 0.4823, "step": 10650 }, { "epoch": 0.5387728261419394, "grad_norm": 4.384370803833008, "learning_rate": 4.465279547157558e-06, "loss": 0.374, "step": 10651 }, { "epoch": 0.5388234103900046, "grad_norm": 16.121129989624023, "learning_rate": 4.464481542775287e-06, "loss": 0.5719, "step": 10652 }, { "epoch": 0.5388739946380697, "grad_norm": 8.969339370727539, "learning_rate": 4.463683552192033e-06, "loss": 0.3462, "step": 10653 }, { "epoch": 0.5389245788861349, "grad_norm": 9.21535587310791, "learning_rate": 4.46288557542836e-06, "loss": 0.4432, "step": 10654 }, { "epoch": 0.5389751631342, "grad_norm": 4.9443206787109375, "learning_rate": 4.462087612504826e-06, "loss": 0.3201, "step": 10655 }, { "epoch": 0.5390257473822652, "grad_norm": 6.054194927215576, "learning_rate": 4.4612896634419965e-06, "loss": 0.4349, "step": 10656 }, { "epoch": 0.5390763316303303, "grad_norm": 6.946098327636719, "learning_rate": 4.460491728260429e-06, "loss": 0.46, "step": 10657 }, { "epoch": 0.5391269158783955, "grad_norm": 5.87300443649292, "learning_rate": 4.4596938069806876e-06, "loss": 0.3389, "step": 10658 }, { "epoch": 0.5391775001264606, "grad_norm": 6.569553852081299, "learning_rate": 4.458895899623333e-06, "loss": 0.342, "step": 10659 }, { "epoch": 0.5392280843745257, "grad_norm": 7.284534454345703, "learning_rate": 4.458098006208923e-06, "loss": 0.3942, "step": 10660 }, { "epoch": 0.5392786686225909, "grad_norm": 13.400203704833984, "learning_rate": 4.457300126758018e-06, "loss": 0.4861, "step": 10661 }, { "epoch": 0.539329252870656, "grad_norm": 9.516706466674805, "learning_rate": 4.456502261291179e-06, "loss": 0.4515, "step": 10662 }, { "epoch": 0.5393798371187213, "grad_norm": 7.055469036102295, "learning_rate": 4.455704409828963e-06, "loss": 0.3707, "step": 10663 }, { "epoch": 0.5394304213667864, "grad_norm": 6.113622665405273, "learning_rate": 4.454906572391929e-06, "loss": 0.3728, "step": 10664 }, { "epoch": 0.5394810056148516, "grad_norm": 5.327951431274414, "learning_rate": 4.4541087490006375e-06, "loss": 0.4119, "step": 10665 }, { "epoch": 0.5395315898629167, "grad_norm": 5.74273681640625, "learning_rate": 4.453310939675646e-06, "loss": 0.4293, "step": 10666 }, { "epoch": 0.5395821741109819, "grad_norm": 5.611219882965088, "learning_rate": 4.45251314443751e-06, "loss": 0.4352, "step": 10667 }, { "epoch": 0.539632758359047, "grad_norm": 10.42098617553711, "learning_rate": 4.451715363306787e-06, "loss": 0.4318, "step": 10668 }, { "epoch": 0.5396833426071121, "grad_norm": 5.616715908050537, "learning_rate": 4.450917596304036e-06, "loss": 0.3553, "step": 10669 }, { "epoch": 0.5397339268551773, "grad_norm": 10.876290321350098, "learning_rate": 4.450119843449811e-06, "loss": 0.3188, "step": 10670 }, { "epoch": 0.5397845111032424, "grad_norm": 4.984152317047119, "learning_rate": 4.44932210476467e-06, "loss": 0.4458, "step": 10671 }, { "epoch": 0.5398350953513076, "grad_norm": 8.359872817993164, "learning_rate": 4.448524380269168e-06, "loss": 0.2465, "step": 10672 }, { "epoch": 0.5398856795993727, "grad_norm": 6.979741096496582, "learning_rate": 4.447726669983861e-06, "loss": 0.4703, "step": 10673 }, { "epoch": 0.5399362638474379, "grad_norm": 4.305483818054199, "learning_rate": 4.446928973929303e-06, "loss": 0.3591, "step": 10674 }, { "epoch": 0.539986848095503, "grad_norm": 5.661098003387451, "learning_rate": 4.4461312921260505e-06, "loss": 0.2976, "step": 10675 }, { "epoch": 0.5400374323435682, "grad_norm": 8.223920822143555, "learning_rate": 4.4453336245946565e-06, "loss": 0.4227, "step": 10676 }, { "epoch": 0.5400880165916334, "grad_norm": 15.06428337097168, "learning_rate": 4.444535971355675e-06, "loss": 0.5293, "step": 10677 }, { "epoch": 0.5401386008396986, "grad_norm": 5.9479522705078125, "learning_rate": 4.44373833242966e-06, "loss": 0.3411, "step": 10678 }, { "epoch": 0.5401891850877637, "grad_norm": 6.122056484222412, "learning_rate": 4.4429407078371665e-06, "loss": 0.3726, "step": 10679 }, { "epoch": 0.5402397693358288, "grad_norm": 17.943450927734375, "learning_rate": 4.442143097598743e-06, "loss": 0.4963, "step": 10680 }, { "epoch": 0.540290353583894, "grad_norm": 21.602584838867188, "learning_rate": 4.441345501734945e-06, "loss": 0.5542, "step": 10681 }, { "epoch": 0.5403409378319591, "grad_norm": 11.390779495239258, "learning_rate": 4.440547920266325e-06, "loss": 0.5563, "step": 10682 }, { "epoch": 0.5403915220800243, "grad_norm": 6.501058578491211, "learning_rate": 4.439750353213432e-06, "loss": 0.4332, "step": 10683 }, { "epoch": 0.5404421063280894, "grad_norm": 7.977992534637451, "learning_rate": 4.4389528005968195e-06, "loss": 0.4476, "step": 10684 }, { "epoch": 0.5404926905761546, "grad_norm": 6.943356037139893, "learning_rate": 4.438155262437039e-06, "loss": 0.3938, "step": 10685 }, { "epoch": 0.5405432748242197, "grad_norm": 6.942516326904297, "learning_rate": 4.43735773875464e-06, "loss": 0.3915, "step": 10686 }, { "epoch": 0.5405938590722849, "grad_norm": 8.391701698303223, "learning_rate": 4.436560229570174e-06, "loss": 0.3958, "step": 10687 }, { "epoch": 0.54064444332035, "grad_norm": 6.188960075378418, "learning_rate": 4.435762734904188e-06, "loss": 0.307, "step": 10688 }, { "epoch": 0.5406950275684151, "grad_norm": 7.186991214752197, "learning_rate": 4.434965254777234e-06, "loss": 0.4498, "step": 10689 }, { "epoch": 0.5407456118164804, "grad_norm": 21.995037078857422, "learning_rate": 4.4341677892098615e-06, "loss": 0.7195, "step": 10690 }, { "epoch": 0.5407961960645455, "grad_norm": 10.676214218139648, "learning_rate": 4.433370338222618e-06, "loss": 0.5054, "step": 10691 }, { "epoch": 0.5408467803126107, "grad_norm": 8.21834945678711, "learning_rate": 4.432572901836053e-06, "loss": 0.4785, "step": 10692 }, { "epoch": 0.5408973645606758, "grad_norm": 8.188478469848633, "learning_rate": 4.431775480070713e-06, "loss": 0.4588, "step": 10693 }, { "epoch": 0.540947948808741, "grad_norm": 6.452823162078857, "learning_rate": 4.430978072947148e-06, "loss": 0.4139, "step": 10694 }, { "epoch": 0.5409985330568061, "grad_norm": 10.098071098327637, "learning_rate": 4.430180680485901e-06, "loss": 0.4411, "step": 10695 }, { "epoch": 0.5410491173048713, "grad_norm": 13.010622024536133, "learning_rate": 4.429383302707522e-06, "loss": 0.4312, "step": 10696 }, { "epoch": 0.5410997015529364, "grad_norm": 25.094764709472656, "learning_rate": 4.4285859396325585e-06, "loss": 0.6515, "step": 10697 }, { "epoch": 0.5411502858010016, "grad_norm": 7.327533721923828, "learning_rate": 4.427788591281553e-06, "loss": 0.3566, "step": 10698 }, { "epoch": 0.5412008700490667, "grad_norm": 5.982773780822754, "learning_rate": 4.426991257675054e-06, "loss": 0.4194, "step": 10699 }, { "epoch": 0.5412514542971318, "grad_norm": 7.037473678588867, "learning_rate": 4.426193938833608e-06, "loss": 0.474, "step": 10700 }, { "epoch": 0.541302038545197, "grad_norm": 18.443273544311523, "learning_rate": 4.425396634777756e-06, "loss": 0.6107, "step": 10701 }, { "epoch": 0.5413526227932621, "grad_norm": 10.749408721923828, "learning_rate": 4.424599345528046e-06, "loss": 0.4704, "step": 10702 }, { "epoch": 0.5414032070413274, "grad_norm": 10.040404319763184, "learning_rate": 4.423802071105021e-06, "loss": 0.3792, "step": 10703 }, { "epoch": 0.5414537912893925, "grad_norm": 16.73321533203125, "learning_rate": 4.423004811529225e-06, "loss": 0.2954, "step": 10704 }, { "epoch": 0.5415043755374577, "grad_norm": 7.06281852722168, "learning_rate": 4.422207566821202e-06, "loss": 0.3727, "step": 10705 }, { "epoch": 0.5415549597855228, "grad_norm": 7.5732645988464355, "learning_rate": 4.421410337001494e-06, "loss": 0.3848, "step": 10706 }, { "epoch": 0.541605544033588, "grad_norm": 7.279292106628418, "learning_rate": 4.420613122090643e-06, "loss": 0.3827, "step": 10707 }, { "epoch": 0.5416561282816531, "grad_norm": 11.863445281982422, "learning_rate": 4.419815922109193e-06, "loss": 0.5234, "step": 10708 }, { "epoch": 0.5417067125297182, "grad_norm": 7.403885364532471, "learning_rate": 4.419018737077685e-06, "loss": 0.4401, "step": 10709 }, { "epoch": 0.5417572967777834, "grad_norm": 10.475162506103516, "learning_rate": 4.4182215670166614e-06, "loss": 0.426, "step": 10710 }, { "epoch": 0.5418078810258485, "grad_norm": 8.162850379943848, "learning_rate": 4.417424411946662e-06, "loss": 0.4475, "step": 10711 }, { "epoch": 0.5418584652739137, "grad_norm": 10.824554443359375, "learning_rate": 4.416627271888229e-06, "loss": 0.449, "step": 10712 }, { "epoch": 0.5419090495219788, "grad_norm": 9.275571823120117, "learning_rate": 4.415830146861902e-06, "loss": 0.4473, "step": 10713 }, { "epoch": 0.541959633770044, "grad_norm": 5.849544525146484, "learning_rate": 4.415033036888221e-06, "loss": 0.4472, "step": 10714 }, { "epoch": 0.5420102180181091, "grad_norm": 9.964672088623047, "learning_rate": 4.414235941987725e-06, "loss": 0.3855, "step": 10715 }, { "epoch": 0.5420608022661744, "grad_norm": 19.3087158203125, "learning_rate": 4.4134388621809556e-06, "loss": 0.6206, "step": 10716 }, { "epoch": 0.5421113865142395, "grad_norm": 10.260724067687988, "learning_rate": 4.412641797488451e-06, "loss": 0.3954, "step": 10717 }, { "epoch": 0.5421619707623047, "grad_norm": 10.808606147766113, "learning_rate": 4.411844747930746e-06, "loss": 0.4497, "step": 10718 }, { "epoch": 0.5422125550103698, "grad_norm": 10.913670539855957, "learning_rate": 4.411047713528381e-06, "loss": 0.3422, "step": 10719 }, { "epoch": 0.542263139258435, "grad_norm": 12.342963218688965, "learning_rate": 4.410250694301896e-06, "loss": 0.4084, "step": 10720 }, { "epoch": 0.5423137235065001, "grad_norm": 7.032750606536865, "learning_rate": 4.409453690271824e-06, "loss": 0.3861, "step": 10721 }, { "epoch": 0.5423643077545652, "grad_norm": 7.167409420013428, "learning_rate": 4.408656701458706e-06, "loss": 0.336, "step": 10722 }, { "epoch": 0.5424148920026304, "grad_norm": 4.913217544555664, "learning_rate": 4.407859727883074e-06, "loss": 0.4387, "step": 10723 }, { "epoch": 0.5424654762506955, "grad_norm": 10.941547393798828, "learning_rate": 4.407062769565468e-06, "loss": 0.4307, "step": 10724 }, { "epoch": 0.5425160604987607, "grad_norm": 8.860135078430176, "learning_rate": 4.4062658265264225e-06, "loss": 0.4896, "step": 10725 }, { "epoch": 0.5425666447468258, "grad_norm": 5.699547290802002, "learning_rate": 4.405468898786472e-06, "loss": 0.3075, "step": 10726 }, { "epoch": 0.542617228994891, "grad_norm": 13.510647773742676, "learning_rate": 4.404671986366151e-06, "loss": 0.4635, "step": 10727 }, { "epoch": 0.5426678132429561, "grad_norm": 4.187182903289795, "learning_rate": 4.403875089285997e-06, "loss": 0.4361, "step": 10728 }, { "epoch": 0.5427183974910212, "grad_norm": 6.63147497177124, "learning_rate": 4.403078207566541e-06, "loss": 0.4525, "step": 10729 }, { "epoch": 0.5427689817390865, "grad_norm": 19.983930587768555, "learning_rate": 4.40228134122832e-06, "loss": 0.5051, "step": 10730 }, { "epoch": 0.5428195659871516, "grad_norm": 5.5715789794921875, "learning_rate": 4.401484490291862e-06, "loss": 0.4137, "step": 10731 }, { "epoch": 0.5428701502352168, "grad_norm": 7.125390529632568, "learning_rate": 4.4006876547777045e-06, "loss": 0.4352, "step": 10732 }, { "epoch": 0.5429207344832819, "grad_norm": 5.720056533813477, "learning_rate": 4.399890834706378e-06, "loss": 0.3282, "step": 10733 }, { "epoch": 0.5429713187313471, "grad_norm": 6.2631402015686035, "learning_rate": 4.399094030098415e-06, "loss": 0.4707, "step": 10734 }, { "epoch": 0.5430219029794122, "grad_norm": 7.107920169830322, "learning_rate": 4.398297240974348e-06, "loss": 0.3925, "step": 10735 }, { "epoch": 0.5430724872274774, "grad_norm": 5.233887195587158, "learning_rate": 4.397500467354707e-06, "loss": 0.3947, "step": 10736 }, { "epoch": 0.5431230714755425, "grad_norm": 8.219501495361328, "learning_rate": 4.3967037092600234e-06, "loss": 0.4472, "step": 10737 }, { "epoch": 0.5431736557236077, "grad_norm": 11.54841136932373, "learning_rate": 4.39590696671083e-06, "loss": 0.3876, "step": 10738 }, { "epoch": 0.5432242399716728, "grad_norm": 9.695158958435059, "learning_rate": 4.395110239727653e-06, "loss": 0.4557, "step": 10739 }, { "epoch": 0.5432748242197379, "grad_norm": 13.720306396484375, "learning_rate": 4.394313528331024e-06, "loss": 0.541, "step": 10740 }, { "epoch": 0.5433254084678031, "grad_norm": 5.211538314819336, "learning_rate": 4.393516832541474e-06, "loss": 0.4596, "step": 10741 }, { "epoch": 0.5433759927158682, "grad_norm": 7.364568710327148, "learning_rate": 4.392720152379529e-06, "loss": 0.3701, "step": 10742 }, { "epoch": 0.5434265769639335, "grad_norm": 6.694561958312988, "learning_rate": 4.3919234878657215e-06, "loss": 0.3349, "step": 10743 }, { "epoch": 0.5434771612119986, "grad_norm": 9.703927993774414, "learning_rate": 4.391126839020574e-06, "loss": 0.3943, "step": 10744 }, { "epoch": 0.5435277454600638, "grad_norm": 6.50121545791626, "learning_rate": 4.3903302058646185e-06, "loss": 0.3559, "step": 10745 }, { "epoch": 0.5435783297081289, "grad_norm": 10.540010452270508, "learning_rate": 4.38953358841838e-06, "loss": 0.4631, "step": 10746 }, { "epoch": 0.5436289139561941, "grad_norm": 6.352488994598389, "learning_rate": 4.388736986702387e-06, "loss": 0.3519, "step": 10747 }, { "epoch": 0.5436794982042592, "grad_norm": 7.027977466583252, "learning_rate": 4.387940400737165e-06, "loss": 0.3531, "step": 10748 }, { "epoch": 0.5437300824523243, "grad_norm": 5.127730846405029, "learning_rate": 4.387143830543241e-06, "loss": 0.4483, "step": 10749 }, { "epoch": 0.5437806667003895, "grad_norm": 5.797598361968994, "learning_rate": 4.386347276141139e-06, "loss": 0.3249, "step": 10750 }, { "epoch": 0.5438312509484546, "grad_norm": 6.878398895263672, "learning_rate": 4.3855507375513874e-06, "loss": 0.3794, "step": 10751 }, { "epoch": 0.5438818351965198, "grad_norm": 4.791615962982178, "learning_rate": 4.384754214794508e-06, "loss": 0.3469, "step": 10752 }, { "epoch": 0.5439324194445849, "grad_norm": 8.513718605041504, "learning_rate": 4.383957707891027e-06, "loss": 0.3837, "step": 10753 }, { "epoch": 0.5439830036926501, "grad_norm": 22.486793518066406, "learning_rate": 4.383161216861467e-06, "loss": 0.4047, "step": 10754 }, { "epoch": 0.5440335879407152, "grad_norm": 4.866349220275879, "learning_rate": 4.382364741726352e-06, "loss": 0.3827, "step": 10755 }, { "epoch": 0.5440841721887805, "grad_norm": 7.955525875091553, "learning_rate": 4.381568282506209e-06, "loss": 0.5008, "step": 10756 }, { "epoch": 0.5441347564368456, "grad_norm": 7.204239845275879, "learning_rate": 4.3807718392215555e-06, "loss": 0.3798, "step": 10757 }, { "epoch": 0.5441853406849108, "grad_norm": 5.8543009757995605, "learning_rate": 4.379975411892916e-06, "loss": 0.3217, "step": 10758 }, { "epoch": 0.5442359249329759, "grad_norm": 10.004942893981934, "learning_rate": 4.379179000540812e-06, "loss": 0.475, "step": 10759 }, { "epoch": 0.544286509181041, "grad_norm": 7.773905277252197, "learning_rate": 4.378382605185767e-06, "loss": 0.4427, "step": 10760 }, { "epoch": 0.5443370934291062, "grad_norm": 10.201285362243652, "learning_rate": 4.377586225848299e-06, "loss": 0.4102, "step": 10761 }, { "epoch": 0.5443876776771713, "grad_norm": 6.183211803436279, "learning_rate": 4.376789862548932e-06, "loss": 0.3882, "step": 10762 }, { "epoch": 0.5444382619252365, "grad_norm": 19.10573387145996, "learning_rate": 4.375993515308185e-06, "loss": 0.6136, "step": 10763 }, { "epoch": 0.5444888461733016, "grad_norm": 7.270832538604736, "learning_rate": 4.3751971841465765e-06, "loss": 0.3586, "step": 10764 }, { "epoch": 0.5445394304213668, "grad_norm": 8.619377136230469, "learning_rate": 4.3744008690846276e-06, "loss": 0.3323, "step": 10765 }, { "epoch": 0.5445900146694319, "grad_norm": 10.463276863098145, "learning_rate": 4.373604570142858e-06, "loss": 0.5015, "step": 10766 }, { "epoch": 0.544640598917497, "grad_norm": 7.953053951263428, "learning_rate": 4.372808287341785e-06, "loss": 0.4852, "step": 10767 }, { "epoch": 0.5446911831655622, "grad_norm": 5.484529972076416, "learning_rate": 4.3720120207019275e-06, "loss": 0.415, "step": 10768 }, { "epoch": 0.5447417674136273, "grad_norm": 9.877387046813965, "learning_rate": 4.3712157702438066e-06, "loss": 0.3948, "step": 10769 }, { "epoch": 0.5447923516616926, "grad_norm": 9.909465789794922, "learning_rate": 4.370419535987934e-06, "loss": 0.4126, "step": 10770 }, { "epoch": 0.5448429359097577, "grad_norm": 8.640460968017578, "learning_rate": 4.369623317954828e-06, "loss": 0.44, "step": 10771 }, { "epoch": 0.5448935201578229, "grad_norm": 9.959575653076172, "learning_rate": 4.368827116165007e-06, "loss": 0.4958, "step": 10772 }, { "epoch": 0.544944104405888, "grad_norm": 7.02064847946167, "learning_rate": 4.368030930638989e-06, "loss": 0.4261, "step": 10773 }, { "epoch": 0.5449946886539532, "grad_norm": 5.219463348388672, "learning_rate": 4.367234761397285e-06, "loss": 0.3668, "step": 10774 }, { "epoch": 0.5450452729020183, "grad_norm": 9.362998008728027, "learning_rate": 4.366438608460413e-06, "loss": 0.4317, "step": 10775 }, { "epoch": 0.5450958571500835, "grad_norm": 12.136094093322754, "learning_rate": 4.3656424718488886e-06, "loss": 0.4982, "step": 10776 }, { "epoch": 0.5451464413981486, "grad_norm": 10.317352294921875, "learning_rate": 4.364846351583225e-06, "loss": 0.4369, "step": 10777 }, { "epoch": 0.5451970256462138, "grad_norm": 8.168822288513184, "learning_rate": 4.364050247683937e-06, "loss": 0.4891, "step": 10778 }, { "epoch": 0.5452476098942789, "grad_norm": 7.373532295227051, "learning_rate": 4.363254160171538e-06, "loss": 0.3878, "step": 10779 }, { "epoch": 0.545298194142344, "grad_norm": 6.172713756561279, "learning_rate": 4.3624580890665404e-06, "loss": 0.4103, "step": 10780 }, { "epoch": 0.5453487783904092, "grad_norm": 9.958966255187988, "learning_rate": 4.361662034389459e-06, "loss": 0.4355, "step": 10781 }, { "epoch": 0.5453993626384743, "grad_norm": 13.98399543762207, "learning_rate": 4.360865996160807e-06, "loss": 0.3866, "step": 10782 }, { "epoch": 0.5454499468865396, "grad_norm": 14.794669151306152, "learning_rate": 4.360069974401092e-06, "loss": 0.6003, "step": 10783 }, { "epoch": 0.5455005311346047, "grad_norm": 5.273373603820801, "learning_rate": 4.359273969130828e-06, "loss": 0.3911, "step": 10784 }, { "epoch": 0.5455511153826699, "grad_norm": 12.720796585083008, "learning_rate": 4.358477980370527e-06, "loss": 0.5501, "step": 10785 }, { "epoch": 0.545601699630735, "grad_norm": 6.527102470397949, "learning_rate": 4.357682008140699e-06, "loss": 0.351, "step": 10786 }, { "epoch": 0.5456522838788002, "grad_norm": 9.85328483581543, "learning_rate": 4.3568860524618536e-06, "loss": 0.3885, "step": 10787 }, { "epoch": 0.5457028681268653, "grad_norm": 6.04904842376709, "learning_rate": 4.356090113354501e-06, "loss": 0.3584, "step": 10788 }, { "epoch": 0.5457534523749304, "grad_norm": 7.850252628326416, "learning_rate": 4.355294190839151e-06, "loss": 0.3426, "step": 10789 }, { "epoch": 0.5458040366229956, "grad_norm": 7.031247615814209, "learning_rate": 4.354498284936313e-06, "loss": 0.4362, "step": 10790 }, { "epoch": 0.5458546208710607, "grad_norm": 8.479918479919434, "learning_rate": 4.353702395666496e-06, "loss": 0.4703, "step": 10791 }, { "epoch": 0.5459052051191259, "grad_norm": 12.534040451049805, "learning_rate": 4.352906523050206e-06, "loss": 0.497, "step": 10792 }, { "epoch": 0.545955789367191, "grad_norm": 12.070446968078613, "learning_rate": 4.35211066710795e-06, "loss": 0.5595, "step": 10793 }, { "epoch": 0.5460063736152562, "grad_norm": 8.892965316772461, "learning_rate": 4.351314827860241e-06, "loss": 0.3977, "step": 10794 }, { "epoch": 0.5460569578633213, "grad_norm": 8.02793025970459, "learning_rate": 4.350519005327581e-06, "loss": 0.3701, "step": 10795 }, { "epoch": 0.5461075421113866, "grad_norm": 11.533021926879883, "learning_rate": 4.3497231995304764e-06, "loss": 0.5547, "step": 10796 }, { "epoch": 0.5461581263594517, "grad_norm": 6.9734063148498535, "learning_rate": 4.348927410489434e-06, "loss": 0.3357, "step": 10797 }, { "epoch": 0.5462087106075169, "grad_norm": 10.193535804748535, "learning_rate": 4.3481316382249615e-06, "loss": 0.4712, "step": 10798 }, { "epoch": 0.546259294855582, "grad_norm": 6.153002738952637, "learning_rate": 4.34733588275756e-06, "loss": 0.3706, "step": 10799 }, { "epoch": 0.5463098791036471, "grad_norm": 5.419948577880859, "learning_rate": 4.346540144107737e-06, "loss": 0.3224, "step": 10800 }, { "epoch": 0.5463604633517123, "grad_norm": 10.072693824768066, "learning_rate": 4.345744422295997e-06, "loss": 0.4565, "step": 10801 }, { "epoch": 0.5464110475997774, "grad_norm": 7.460104942321777, "learning_rate": 4.344948717342842e-06, "loss": 0.4571, "step": 10802 }, { "epoch": 0.5464616318478426, "grad_norm": 5.760234832763672, "learning_rate": 4.344153029268776e-06, "loss": 0.3473, "step": 10803 }, { "epoch": 0.5465122160959077, "grad_norm": 10.835986137390137, "learning_rate": 4.343357358094303e-06, "loss": 0.4563, "step": 10804 }, { "epoch": 0.5465628003439729, "grad_norm": 8.404960632324219, "learning_rate": 4.342561703839925e-06, "loss": 0.402, "step": 10805 }, { "epoch": 0.546613384592038, "grad_norm": 8.699759483337402, "learning_rate": 4.341766066526143e-06, "loss": 0.4383, "step": 10806 }, { "epoch": 0.5466639688401032, "grad_norm": 8.062456130981445, "learning_rate": 4.340970446173461e-06, "loss": 0.5228, "step": 10807 }, { "epoch": 0.5467145530881683, "grad_norm": 10.348481178283691, "learning_rate": 4.3401748428023795e-06, "loss": 0.466, "step": 10808 }, { "epoch": 0.5467651373362334, "grad_norm": 12.723587036132812, "learning_rate": 4.339379256433397e-06, "loss": 0.5609, "step": 10809 }, { "epoch": 0.5468157215842987, "grad_norm": 8.302559852600098, "learning_rate": 4.338583687087015e-06, "loss": 0.4872, "step": 10810 }, { "epoch": 0.5468663058323638, "grad_norm": 11.863876342773438, "learning_rate": 4.337788134783735e-06, "loss": 0.5195, "step": 10811 }, { "epoch": 0.546916890080429, "grad_norm": 8.02021312713623, "learning_rate": 4.336992599544054e-06, "loss": 0.3464, "step": 10812 }, { "epoch": 0.5469674743284941, "grad_norm": 8.544191360473633, "learning_rate": 4.336197081388472e-06, "loss": 0.3773, "step": 10813 }, { "epoch": 0.5470180585765593, "grad_norm": 9.125730514526367, "learning_rate": 4.3354015803374895e-06, "loss": 0.4871, "step": 10814 }, { "epoch": 0.5470686428246244, "grad_norm": 6.040026664733887, "learning_rate": 4.334606096411601e-06, "loss": 0.2621, "step": 10815 }, { "epoch": 0.5471192270726896, "grad_norm": 7.524200916290283, "learning_rate": 4.3338106296313075e-06, "loss": 0.3764, "step": 10816 }, { "epoch": 0.5471698113207547, "grad_norm": 5.130995750427246, "learning_rate": 4.333015180017105e-06, "loss": 0.3723, "step": 10817 }, { "epoch": 0.5472203955688199, "grad_norm": 11.350974082946777, "learning_rate": 4.33221974758949e-06, "loss": 0.4334, "step": 10818 }, { "epoch": 0.547270979816885, "grad_norm": 13.279322624206543, "learning_rate": 4.3314243323689586e-06, "loss": 0.5405, "step": 10819 }, { "epoch": 0.5473215640649501, "grad_norm": 5.759433269500732, "learning_rate": 4.33062893437601e-06, "loss": 0.3205, "step": 10820 }, { "epoch": 0.5473721483130153, "grad_norm": 10.739933013916016, "learning_rate": 4.329833553631135e-06, "loss": 0.4538, "step": 10821 }, { "epoch": 0.5474227325610804, "grad_norm": 8.366076469421387, "learning_rate": 4.32903819015483e-06, "loss": 0.4245, "step": 10822 }, { "epoch": 0.5474733168091457, "grad_norm": 6.573336601257324, "learning_rate": 4.32824284396759e-06, "loss": 0.3988, "step": 10823 }, { "epoch": 0.5475239010572108, "grad_norm": 7.825154781341553, "learning_rate": 4.327447515089912e-06, "loss": 0.4821, "step": 10824 }, { "epoch": 0.547574485305276, "grad_norm": 5.896254062652588, "learning_rate": 4.326652203542284e-06, "loss": 0.3421, "step": 10825 }, { "epoch": 0.5476250695533411, "grad_norm": 7.410268783569336, "learning_rate": 4.325856909345204e-06, "loss": 0.4624, "step": 10826 }, { "epoch": 0.5476756538014063, "grad_norm": 9.747552871704102, "learning_rate": 4.325061632519162e-06, "loss": 0.2768, "step": 10827 }, { "epoch": 0.5477262380494714, "grad_norm": 6.212347984313965, "learning_rate": 4.324266373084652e-06, "loss": 0.482, "step": 10828 }, { "epoch": 0.5477768222975365, "grad_norm": 7.278308391571045, "learning_rate": 4.323471131062166e-06, "loss": 0.4957, "step": 10829 }, { "epoch": 0.5478274065456017, "grad_norm": 5.727104187011719, "learning_rate": 4.3226759064721935e-06, "loss": 0.406, "step": 10830 }, { "epoch": 0.5478779907936668, "grad_norm": 6.140063285827637, "learning_rate": 4.321880699335227e-06, "loss": 0.4407, "step": 10831 }, { "epoch": 0.547928575041732, "grad_norm": 13.053082466125488, "learning_rate": 4.321085509671759e-06, "loss": 0.3801, "step": 10832 }, { "epoch": 0.5479791592897971, "grad_norm": 6.0736918449401855, "learning_rate": 4.320290337502278e-06, "loss": 0.3176, "step": 10833 }, { "epoch": 0.5480297435378623, "grad_norm": 7.420167446136475, "learning_rate": 4.3194951828472705e-06, "loss": 0.3796, "step": 10834 }, { "epoch": 0.5480803277859274, "grad_norm": 7.8789496421813965, "learning_rate": 4.3187000457272285e-06, "loss": 0.3717, "step": 10835 }, { "epoch": 0.5481309120339927, "grad_norm": 8.31695556640625, "learning_rate": 4.317904926162642e-06, "loss": 0.4432, "step": 10836 }, { "epoch": 0.5481814962820578, "grad_norm": 6.565267086029053, "learning_rate": 4.317109824173997e-06, "loss": 0.5154, "step": 10837 }, { "epoch": 0.548232080530123, "grad_norm": 9.996498107910156, "learning_rate": 4.3163147397817825e-06, "loss": 0.4396, "step": 10838 }, { "epoch": 0.5482826647781881, "grad_norm": 8.583395957946777, "learning_rate": 4.3155196730064875e-06, "loss": 0.469, "step": 10839 }, { "epoch": 0.5483332490262532, "grad_norm": 13.110922813415527, "learning_rate": 4.3147246238685955e-06, "loss": 0.4652, "step": 10840 }, { "epoch": 0.5483838332743184, "grad_norm": 8.986784934997559, "learning_rate": 4.313929592388595e-06, "loss": 0.3237, "step": 10841 }, { "epoch": 0.5484344175223835, "grad_norm": 7.216061115264893, "learning_rate": 4.313134578586972e-06, "loss": 0.4013, "step": 10842 }, { "epoch": 0.5484850017704487, "grad_norm": 10.098172187805176, "learning_rate": 4.3123395824842115e-06, "loss": 0.3462, "step": 10843 }, { "epoch": 0.5485355860185138, "grad_norm": 14.456459045410156, "learning_rate": 4.311544604100799e-06, "loss": 0.6372, "step": 10844 }, { "epoch": 0.548586170266579, "grad_norm": 5.915707588195801, "learning_rate": 4.31074964345722e-06, "loss": 0.2505, "step": 10845 }, { "epoch": 0.5486367545146441, "grad_norm": 6.577264785766602, "learning_rate": 4.3099547005739595e-06, "loss": 0.5185, "step": 10846 }, { "epoch": 0.5486873387627093, "grad_norm": 7.345544815063477, "learning_rate": 4.309159775471496e-06, "loss": 0.4253, "step": 10847 }, { "epoch": 0.5487379230107744, "grad_norm": 7.744265079498291, "learning_rate": 4.308364868170318e-06, "loss": 0.4703, "step": 10848 }, { "epoch": 0.5487885072588395, "grad_norm": 9.023236274719238, "learning_rate": 4.307569978690907e-06, "loss": 0.3643, "step": 10849 }, { "epoch": 0.5488390915069048, "grad_norm": 7.661370277404785, "learning_rate": 4.306775107053745e-06, "loss": 0.3305, "step": 10850 }, { "epoch": 0.5488896757549699, "grad_norm": 6.2435712814331055, "learning_rate": 4.3059802532793135e-06, "loss": 0.4213, "step": 10851 }, { "epoch": 0.5489402600030351, "grad_norm": 9.035460472106934, "learning_rate": 4.305185417388095e-06, "loss": 0.436, "step": 10852 }, { "epoch": 0.5489908442511002, "grad_norm": 8.633048057556152, "learning_rate": 4.30439059940057e-06, "loss": 0.4527, "step": 10853 }, { "epoch": 0.5490414284991654, "grad_norm": 8.382857322692871, "learning_rate": 4.3035957993372185e-06, "loss": 0.4769, "step": 10854 }, { "epoch": 0.5490920127472305, "grad_norm": 9.38303279876709, "learning_rate": 4.302801017218522e-06, "loss": 0.4668, "step": 10855 }, { "epoch": 0.5491425969952957, "grad_norm": 7.099395275115967, "learning_rate": 4.302006253064958e-06, "loss": 0.4566, "step": 10856 }, { "epoch": 0.5491931812433608, "grad_norm": 9.7990140914917, "learning_rate": 4.301211506897009e-06, "loss": 0.3476, "step": 10857 }, { "epoch": 0.549243765491426, "grad_norm": 7.178197383880615, "learning_rate": 4.30041677873515e-06, "loss": 0.4726, "step": 10858 }, { "epoch": 0.5492943497394911, "grad_norm": 6.106814861297607, "learning_rate": 4.299622068599863e-06, "loss": 0.3611, "step": 10859 }, { "epoch": 0.5493449339875562, "grad_norm": 6.176328659057617, "learning_rate": 4.298827376511622e-06, "loss": 0.4314, "step": 10860 }, { "epoch": 0.5493955182356214, "grad_norm": 7.809981346130371, "learning_rate": 4.298032702490905e-06, "loss": 0.4195, "step": 10861 }, { "epoch": 0.5494461024836865, "grad_norm": 7.101337909698486, "learning_rate": 4.297238046558192e-06, "loss": 0.3985, "step": 10862 }, { "epoch": 0.5494966867317518, "grad_norm": 7.429399490356445, "learning_rate": 4.296443408733955e-06, "loss": 0.3776, "step": 10863 }, { "epoch": 0.5495472709798169, "grad_norm": 18.649776458740234, "learning_rate": 4.295648789038673e-06, "loss": 0.4526, "step": 10864 }, { "epoch": 0.5495978552278821, "grad_norm": 8.922739028930664, "learning_rate": 4.29485418749282e-06, "loss": 0.5131, "step": 10865 }, { "epoch": 0.5496484394759472, "grad_norm": 5.790419101715088, "learning_rate": 4.29405960411687e-06, "loss": 0.3412, "step": 10866 }, { "epoch": 0.5496990237240124, "grad_norm": 7.543245792388916, "learning_rate": 4.293265038931301e-06, "loss": 0.4237, "step": 10867 }, { "epoch": 0.5497496079720775, "grad_norm": 11.911140441894531, "learning_rate": 4.292470491956584e-06, "loss": 0.4872, "step": 10868 }, { "epoch": 0.5498001922201426, "grad_norm": 10.381775856018066, "learning_rate": 4.291675963213193e-06, "loss": 0.336, "step": 10869 }, { "epoch": 0.5498507764682078, "grad_norm": 6.376782417297363, "learning_rate": 4.290881452721603e-06, "loss": 0.3576, "step": 10870 }, { "epoch": 0.5499013607162729, "grad_norm": 6.220269203186035, "learning_rate": 4.290086960502284e-06, "loss": 0.4743, "step": 10871 }, { "epoch": 0.5499519449643381, "grad_norm": 6.219159126281738, "learning_rate": 4.289292486575711e-06, "loss": 0.3872, "step": 10872 }, { "epoch": 0.5500025292124032, "grad_norm": 13.285812377929688, "learning_rate": 4.288498030962353e-06, "loss": 0.4673, "step": 10873 }, { "epoch": 0.5500531134604684, "grad_norm": 7.9738359451293945, "learning_rate": 4.287703593682682e-06, "loss": 0.4037, "step": 10874 }, { "epoch": 0.5501036977085335, "grad_norm": 16.34612464904785, "learning_rate": 4.286909174757169e-06, "loss": 0.5022, "step": 10875 }, { "epoch": 0.5501542819565988, "grad_norm": 18.228878021240234, "learning_rate": 4.286114774206282e-06, "loss": 0.4008, "step": 10876 }, { "epoch": 0.5502048662046639, "grad_norm": 11.291824340820312, "learning_rate": 4.285320392050496e-06, "loss": 0.437, "step": 10877 }, { "epoch": 0.5502554504527291, "grad_norm": 6.231892108917236, "learning_rate": 4.284526028310276e-06, "loss": 0.4105, "step": 10878 }, { "epoch": 0.5503060347007942, "grad_norm": 9.109623908996582, "learning_rate": 4.283731683006092e-06, "loss": 0.5204, "step": 10879 }, { "epoch": 0.5503566189488593, "grad_norm": 4.589740753173828, "learning_rate": 4.282937356158413e-06, "loss": 0.2635, "step": 10880 }, { "epoch": 0.5504072031969245, "grad_norm": 13.55935287475586, "learning_rate": 4.282143047787705e-06, "loss": 0.4973, "step": 10881 }, { "epoch": 0.5504577874449896, "grad_norm": 8.345733642578125, "learning_rate": 4.281348757914438e-06, "loss": 0.3515, "step": 10882 }, { "epoch": 0.5505083716930548, "grad_norm": 8.35527229309082, "learning_rate": 4.280554486559078e-06, "loss": 0.4381, "step": 10883 }, { "epoch": 0.5505589559411199, "grad_norm": 7.374518394470215, "learning_rate": 4.27976023374209e-06, "loss": 0.4015, "step": 10884 }, { "epoch": 0.5506095401891851, "grad_norm": 6.634505748748779, "learning_rate": 4.278965999483943e-06, "loss": 0.3936, "step": 10885 }, { "epoch": 0.5506601244372502, "grad_norm": 7.053276538848877, "learning_rate": 4.278171783805099e-06, "loss": 0.3648, "step": 10886 }, { "epoch": 0.5507107086853154, "grad_norm": 4.715599060058594, "learning_rate": 4.2773775867260255e-06, "loss": 0.3431, "step": 10887 }, { "epoch": 0.5507612929333805, "grad_norm": 7.64678955078125, "learning_rate": 4.2765834082671845e-06, "loss": 0.3718, "step": 10888 }, { "epoch": 0.5508118771814458, "grad_norm": 9.515226364135742, "learning_rate": 4.275789248449042e-06, "loss": 0.4118, "step": 10889 }, { "epoch": 0.5508624614295109, "grad_norm": 4.97853422164917, "learning_rate": 4.274995107292064e-06, "loss": 0.3367, "step": 10890 }, { "epoch": 0.550913045677576, "grad_norm": 13.8062744140625, "learning_rate": 4.274200984816708e-06, "loss": 0.5458, "step": 10891 }, { "epoch": 0.5509636299256412, "grad_norm": 9.747050285339355, "learning_rate": 4.27340688104344e-06, "loss": 0.3745, "step": 10892 }, { "epoch": 0.5510142141737063, "grad_norm": 8.372217178344727, "learning_rate": 4.272612795992723e-06, "loss": 0.4273, "step": 10893 }, { "epoch": 0.5510647984217715, "grad_norm": 15.69437026977539, "learning_rate": 4.271818729685016e-06, "loss": 0.4123, "step": 10894 }, { "epoch": 0.5511153826698366, "grad_norm": 8.495736122131348, "learning_rate": 4.271024682140783e-06, "loss": 0.439, "step": 10895 }, { "epoch": 0.5511659669179018, "grad_norm": 6.0518341064453125, "learning_rate": 4.270230653380481e-06, "loss": 0.3705, "step": 10896 }, { "epoch": 0.5512165511659669, "grad_norm": 10.357817649841309, "learning_rate": 4.269436643424573e-06, "loss": 0.4186, "step": 10897 }, { "epoch": 0.551267135414032, "grad_norm": 13.639337539672852, "learning_rate": 4.268642652293521e-06, "loss": 0.4505, "step": 10898 }, { "epoch": 0.5513177196620972, "grad_norm": 9.84136962890625, "learning_rate": 4.267848680007779e-06, "loss": 0.4568, "step": 10899 }, { "epoch": 0.5513683039101623, "grad_norm": 10.512557983398438, "learning_rate": 4.2670547265878075e-06, "loss": 0.5076, "step": 10900 }, { "epoch": 0.5514188881582275, "grad_norm": 5.816340923309326, "learning_rate": 4.2662607920540655e-06, "loss": 0.4062, "step": 10901 }, { "epoch": 0.5514694724062926, "grad_norm": 12.784210205078125, "learning_rate": 4.265466876427012e-06, "loss": 0.3472, "step": 10902 }, { "epoch": 0.5515200566543579, "grad_norm": 14.655776977539062, "learning_rate": 4.264672979727101e-06, "loss": 0.4516, "step": 10903 }, { "epoch": 0.551570640902423, "grad_norm": 7.748472690582275, "learning_rate": 4.263879101974791e-06, "loss": 0.3926, "step": 10904 }, { "epoch": 0.5516212251504882, "grad_norm": 7.746419429779053, "learning_rate": 4.263085243190542e-06, "loss": 0.4015, "step": 10905 }, { "epoch": 0.5516718093985533, "grad_norm": 5.258604526519775, "learning_rate": 4.262291403394803e-06, "loss": 0.347, "step": 10906 }, { "epoch": 0.5517223936466185, "grad_norm": 12.240668296813965, "learning_rate": 4.261497582608033e-06, "loss": 0.4793, "step": 10907 }, { "epoch": 0.5517729778946836, "grad_norm": 9.155834197998047, "learning_rate": 4.260703780850687e-06, "loss": 0.3838, "step": 10908 }, { "epoch": 0.5518235621427487, "grad_norm": 5.992954254150391, "learning_rate": 4.2599099981432194e-06, "loss": 0.3779, "step": 10909 }, { "epoch": 0.5518741463908139, "grad_norm": 9.322032928466797, "learning_rate": 4.259116234506084e-06, "loss": 0.3837, "step": 10910 }, { "epoch": 0.551924730638879, "grad_norm": 8.679101943969727, "learning_rate": 4.258322489959733e-06, "loss": 0.3938, "step": 10911 }, { "epoch": 0.5519753148869442, "grad_norm": 8.051755905151367, "learning_rate": 4.25752876452462e-06, "loss": 0.3968, "step": 10912 }, { "epoch": 0.5520258991350093, "grad_norm": 9.15034008026123, "learning_rate": 4.256735058221197e-06, "loss": 0.3825, "step": 10913 }, { "epoch": 0.5520764833830745, "grad_norm": 9.431267738342285, "learning_rate": 4.255941371069916e-06, "loss": 0.3046, "step": 10914 }, { "epoch": 0.5521270676311396, "grad_norm": 8.093422889709473, "learning_rate": 4.25514770309123e-06, "loss": 0.4461, "step": 10915 }, { "epoch": 0.5521776518792049, "grad_norm": 8.391510963439941, "learning_rate": 4.254354054305587e-06, "loss": 0.4178, "step": 10916 }, { "epoch": 0.55222823612727, "grad_norm": 9.651843070983887, "learning_rate": 4.25356042473344e-06, "loss": 0.4669, "step": 10917 }, { "epoch": 0.5522788203753352, "grad_norm": 9.553539276123047, "learning_rate": 4.252766814395238e-06, "loss": 0.4972, "step": 10918 }, { "epoch": 0.5523294046234003, "grad_norm": 6.328297138214111, "learning_rate": 4.2519732233114295e-06, "loss": 0.4508, "step": 10919 }, { "epoch": 0.5523799888714654, "grad_norm": 12.653605461120605, "learning_rate": 4.251179651502464e-06, "loss": 0.4385, "step": 10920 }, { "epoch": 0.5524305731195306, "grad_norm": 6.786198139190674, "learning_rate": 4.250386098988791e-06, "loss": 0.4245, "step": 10921 }, { "epoch": 0.5524811573675957, "grad_norm": 7.663055419921875, "learning_rate": 4.249592565790857e-06, "loss": 0.4013, "step": 10922 }, { "epoch": 0.5525317416156609, "grad_norm": 11.99577808380127, "learning_rate": 4.248799051929112e-06, "loss": 0.3319, "step": 10923 }, { "epoch": 0.552582325863726, "grad_norm": 6.606425762176514, "learning_rate": 4.248005557424e-06, "loss": 0.4731, "step": 10924 }, { "epoch": 0.5526329101117912, "grad_norm": 6.722163200378418, "learning_rate": 4.247212082295969e-06, "loss": 0.5131, "step": 10925 }, { "epoch": 0.5526834943598563, "grad_norm": 8.838726043701172, "learning_rate": 4.2464186265654615e-06, "loss": 0.4081, "step": 10926 }, { "epoch": 0.5527340786079215, "grad_norm": 8.78100299835205, "learning_rate": 4.245625190252928e-06, "loss": 0.3364, "step": 10927 }, { "epoch": 0.5527846628559866, "grad_norm": 5.708040714263916, "learning_rate": 4.2448317733788105e-06, "loss": 0.3255, "step": 10928 }, { "epoch": 0.5528352471040519, "grad_norm": 5.150373935699463, "learning_rate": 4.2440383759635545e-06, "loss": 0.3691, "step": 10929 }, { "epoch": 0.552885831352117, "grad_norm": 7.458450794219971, "learning_rate": 4.243244998027604e-06, "loss": 0.4835, "step": 10930 }, { "epoch": 0.5529364156001821, "grad_norm": 5.240680694580078, "learning_rate": 4.242451639591401e-06, "loss": 0.361, "step": 10931 }, { "epoch": 0.5529869998482473, "grad_norm": 7.2081756591796875, "learning_rate": 4.24165830067539e-06, "loss": 0.4072, "step": 10932 }, { "epoch": 0.5530375840963124, "grad_norm": 28.792943954467773, "learning_rate": 4.240864981300013e-06, "loss": 0.5196, "step": 10933 }, { "epoch": 0.5530881683443776, "grad_norm": 10.320226669311523, "learning_rate": 4.240071681485712e-06, "loss": 0.4706, "step": 10934 }, { "epoch": 0.5531387525924427, "grad_norm": 6.307401180267334, "learning_rate": 4.239278401252927e-06, "loss": 0.3584, "step": 10935 }, { "epoch": 0.5531893368405079, "grad_norm": 6.314393520355225, "learning_rate": 4.2384851406221025e-06, "loss": 0.3858, "step": 10936 }, { "epoch": 0.553239921088573, "grad_norm": 7.772695064544678, "learning_rate": 4.237691899613676e-06, "loss": 0.3887, "step": 10937 }, { "epoch": 0.5532905053366381, "grad_norm": 6.5579071044921875, "learning_rate": 4.236898678248087e-06, "loss": 0.3429, "step": 10938 }, { "epoch": 0.5533410895847033, "grad_norm": 4.932721138000488, "learning_rate": 4.2361054765457755e-06, "loss": 0.2677, "step": 10939 }, { "epoch": 0.5533916738327684, "grad_norm": 10.619160652160645, "learning_rate": 4.235312294527182e-06, "loss": 0.436, "step": 10940 }, { "epoch": 0.5534422580808336, "grad_norm": 7.060473442077637, "learning_rate": 4.234519132212743e-06, "loss": 0.4262, "step": 10941 }, { "epoch": 0.5534928423288987, "grad_norm": 11.499299049377441, "learning_rate": 4.233725989622896e-06, "loss": 0.4957, "step": 10942 }, { "epoch": 0.553543426576964, "grad_norm": 9.043414115905762, "learning_rate": 4.232932866778081e-06, "loss": 0.4009, "step": 10943 }, { "epoch": 0.5535940108250291, "grad_norm": 4.691682815551758, "learning_rate": 4.232139763698732e-06, "loss": 0.4335, "step": 10944 }, { "epoch": 0.5536445950730943, "grad_norm": 6.779820442199707, "learning_rate": 4.231346680405287e-06, "loss": 0.3933, "step": 10945 }, { "epoch": 0.5536951793211594, "grad_norm": 8.41563606262207, "learning_rate": 4.230553616918182e-06, "loss": 0.4332, "step": 10946 }, { "epoch": 0.5537457635692246, "grad_norm": 9.40576457977295, "learning_rate": 4.229760573257851e-06, "loss": 0.4499, "step": 10947 }, { "epoch": 0.5537963478172897, "grad_norm": 7.659003257751465, "learning_rate": 4.22896754944473e-06, "loss": 0.4633, "step": 10948 }, { "epoch": 0.5538469320653548, "grad_norm": 5.720489501953125, "learning_rate": 4.228174545499255e-06, "loss": 0.3727, "step": 10949 }, { "epoch": 0.55389751631342, "grad_norm": 16.839468002319336, "learning_rate": 4.2273815614418555e-06, "loss": 0.5198, "step": 10950 }, { "epoch": 0.5539481005614851, "grad_norm": 9.14621353149414, "learning_rate": 4.226588597292966e-06, "loss": 0.5366, "step": 10951 }, { "epoch": 0.5539986848095503, "grad_norm": 10.923858642578125, "learning_rate": 4.2257956530730215e-06, "loss": 0.4286, "step": 10952 }, { "epoch": 0.5540492690576154, "grad_norm": 19.895214080810547, "learning_rate": 4.2250027288024535e-06, "loss": 0.4232, "step": 10953 }, { "epoch": 0.5540998533056806, "grad_norm": 11.691580772399902, "learning_rate": 4.224209824501692e-06, "loss": 0.4378, "step": 10954 }, { "epoch": 0.5541504375537457, "grad_norm": 7.415050029754639, "learning_rate": 4.2234169401911695e-06, "loss": 0.3192, "step": 10955 }, { "epoch": 0.554201021801811, "grad_norm": 5.88274621963501, "learning_rate": 4.222624075891319e-06, "loss": 0.4747, "step": 10956 }, { "epoch": 0.5542516060498761, "grad_norm": 5.4789910316467285, "learning_rate": 4.221831231622565e-06, "loss": 0.2693, "step": 10957 }, { "epoch": 0.5543021902979413, "grad_norm": 8.593775749206543, "learning_rate": 4.2210384074053425e-06, "loss": 0.4228, "step": 10958 }, { "epoch": 0.5543527745460064, "grad_norm": 13.373632431030273, "learning_rate": 4.220245603260078e-06, "loss": 0.6649, "step": 10959 }, { "epoch": 0.5544033587940715, "grad_norm": 6.938271522521973, "learning_rate": 4.219452819207201e-06, "loss": 0.3967, "step": 10960 }, { "epoch": 0.5544539430421367, "grad_norm": 7.2413153648376465, "learning_rate": 4.218660055267141e-06, "loss": 0.452, "step": 10961 }, { "epoch": 0.5545045272902018, "grad_norm": 12.31396484375, "learning_rate": 4.217867311460324e-06, "loss": 0.4743, "step": 10962 }, { "epoch": 0.554555111538267, "grad_norm": 6.625682353973389, "learning_rate": 4.217074587807176e-06, "loss": 0.4145, "step": 10963 }, { "epoch": 0.5546056957863321, "grad_norm": 4.7190656661987305, "learning_rate": 4.216281884328125e-06, "loss": 0.3711, "step": 10964 }, { "epoch": 0.5546562800343973, "grad_norm": 5.783705711364746, "learning_rate": 4.215489201043596e-06, "loss": 0.4326, "step": 10965 }, { "epoch": 0.5547068642824624, "grad_norm": 6.068312644958496, "learning_rate": 4.214696537974017e-06, "loss": 0.3578, "step": 10966 }, { "epoch": 0.5547574485305276, "grad_norm": 5.694021224975586, "learning_rate": 4.21390389513981e-06, "loss": 0.3704, "step": 10967 }, { "epoch": 0.5548080327785927, "grad_norm": 11.042768478393555, "learning_rate": 4.213111272561402e-06, "loss": 0.4168, "step": 10968 }, { "epoch": 0.554858617026658, "grad_norm": 6.910201072692871, "learning_rate": 4.2123186702592154e-06, "loss": 0.3688, "step": 10969 }, { "epoch": 0.5549092012747231, "grad_norm": 8.728862762451172, "learning_rate": 4.211526088253674e-06, "loss": 0.411, "step": 10970 }, { "epoch": 0.5549597855227882, "grad_norm": 16.312437057495117, "learning_rate": 4.2107335265652006e-06, "loss": 0.4842, "step": 10971 }, { "epoch": 0.5550103697708534, "grad_norm": 8.586487770080566, "learning_rate": 4.209940985214218e-06, "loss": 0.3524, "step": 10972 }, { "epoch": 0.5550609540189185, "grad_norm": 6.123334884643555, "learning_rate": 4.209148464221147e-06, "loss": 0.3584, "step": 10973 }, { "epoch": 0.5551115382669837, "grad_norm": 11.39881706237793, "learning_rate": 4.208355963606412e-06, "loss": 0.4455, "step": 10974 }, { "epoch": 0.5551621225150488, "grad_norm": 9.84804916381836, "learning_rate": 4.207563483390432e-06, "loss": 0.4603, "step": 10975 }, { "epoch": 0.555212706763114, "grad_norm": 6.3609113693237305, "learning_rate": 4.206771023593626e-06, "loss": 0.386, "step": 10976 }, { "epoch": 0.5552632910111791, "grad_norm": 6.931696891784668, "learning_rate": 4.205978584236413e-06, "loss": 0.3862, "step": 10977 }, { "epoch": 0.5553138752592442, "grad_norm": 5.132296562194824, "learning_rate": 4.205186165339216e-06, "loss": 0.3738, "step": 10978 }, { "epoch": 0.5553644595073094, "grad_norm": 6.440504550933838, "learning_rate": 4.204393766922451e-06, "loss": 0.409, "step": 10979 }, { "epoch": 0.5554150437553745, "grad_norm": 9.500314712524414, "learning_rate": 4.2036013890065354e-06, "loss": 0.4227, "step": 10980 }, { "epoch": 0.5554656280034397, "grad_norm": 6.993231296539307, "learning_rate": 4.202809031611891e-06, "loss": 0.4193, "step": 10981 }, { "epoch": 0.5555162122515048, "grad_norm": 12.007779121398926, "learning_rate": 4.202016694758931e-06, "loss": 0.3723, "step": 10982 }, { "epoch": 0.5555667964995701, "grad_norm": 5.7215657234191895, "learning_rate": 4.201224378468073e-06, "loss": 0.3182, "step": 10983 }, { "epoch": 0.5556173807476352, "grad_norm": 6.464253902435303, "learning_rate": 4.200432082759734e-06, "loss": 0.3589, "step": 10984 }, { "epoch": 0.5556679649957004, "grad_norm": 7.068690299987793, "learning_rate": 4.199639807654328e-06, "loss": 0.4021, "step": 10985 }, { "epoch": 0.5557185492437655, "grad_norm": 7.281830787658691, "learning_rate": 4.1988475531722725e-06, "loss": 0.4135, "step": 10986 }, { "epoch": 0.5557691334918307, "grad_norm": 8.341625213623047, "learning_rate": 4.19805531933398e-06, "loss": 0.3484, "step": 10987 }, { "epoch": 0.5558197177398958, "grad_norm": 10.779455184936523, "learning_rate": 4.197263106159867e-06, "loss": 0.4364, "step": 10988 }, { "epoch": 0.555870301987961, "grad_norm": 9.795781135559082, "learning_rate": 4.196470913670342e-06, "loss": 0.5146, "step": 10989 }, { "epoch": 0.5559208862360261, "grad_norm": 7.346109867095947, "learning_rate": 4.1956787418858205e-06, "loss": 0.3741, "step": 10990 }, { "epoch": 0.5559714704840912, "grad_norm": 13.460704803466797, "learning_rate": 4.194886590826717e-06, "loss": 0.4004, "step": 10991 }, { "epoch": 0.5560220547321564, "grad_norm": 4.704157829284668, "learning_rate": 4.194094460513439e-06, "loss": 0.3718, "step": 10992 }, { "epoch": 0.5560726389802215, "grad_norm": 7.313210487365723, "learning_rate": 4.1933023509664015e-06, "loss": 0.3772, "step": 10993 }, { "epoch": 0.5561232232282867, "grad_norm": 21.007169723510742, "learning_rate": 4.192510262206014e-06, "loss": 0.491, "step": 10994 }, { "epoch": 0.5561738074763518, "grad_norm": 8.263952255249023, "learning_rate": 4.191718194252686e-06, "loss": 0.3445, "step": 10995 }, { "epoch": 0.5562243917244171, "grad_norm": 9.10934829711914, "learning_rate": 4.190926147126827e-06, "loss": 0.4413, "step": 10996 }, { "epoch": 0.5562749759724822, "grad_norm": 10.433396339416504, "learning_rate": 4.190134120848849e-06, "loss": 0.5303, "step": 10997 }, { "epoch": 0.5563255602205474, "grad_norm": 8.006710052490234, "learning_rate": 4.189342115439158e-06, "loss": 0.4521, "step": 10998 }, { "epoch": 0.5563761444686125, "grad_norm": 8.964855194091797, "learning_rate": 4.188550130918163e-06, "loss": 0.4015, "step": 10999 }, { "epoch": 0.5564267287166776, "grad_norm": 6.092023849487305, "learning_rate": 4.18775816730627e-06, "loss": 0.3976, "step": 11000 }, { "epoch": 0.5564773129647428, "grad_norm": 8.495538711547852, "learning_rate": 4.186966224623889e-06, "loss": 0.4026, "step": 11001 }, { "epoch": 0.5565278972128079, "grad_norm": 8.787693977355957, "learning_rate": 4.1861743028914226e-06, "loss": 0.4419, "step": 11002 }, { "epoch": 0.5565784814608731, "grad_norm": 5.795335292816162, "learning_rate": 4.18538240212928e-06, "loss": 0.4425, "step": 11003 }, { "epoch": 0.5566290657089382, "grad_norm": 6.729053497314453, "learning_rate": 4.1845905223578645e-06, "loss": 0.4129, "step": 11004 }, { "epoch": 0.5566796499570034, "grad_norm": 8.678536415100098, "learning_rate": 4.183798663597581e-06, "loss": 0.3879, "step": 11005 }, { "epoch": 0.5567302342050685, "grad_norm": 8.442492485046387, "learning_rate": 4.1830068258688365e-06, "loss": 0.405, "step": 11006 }, { "epoch": 0.5567808184531337, "grad_norm": 6.273405075073242, "learning_rate": 4.18221500919203e-06, "loss": 0.3664, "step": 11007 }, { "epoch": 0.5568314027011988, "grad_norm": 9.20767593383789, "learning_rate": 4.181423213587569e-06, "loss": 0.4744, "step": 11008 }, { "epoch": 0.556881986949264, "grad_norm": 9.505877494812012, "learning_rate": 4.180631439075855e-06, "loss": 0.544, "step": 11009 }, { "epoch": 0.5569325711973292, "grad_norm": 10.759815216064453, "learning_rate": 4.1798396856772884e-06, "loss": 0.5668, "step": 11010 }, { "epoch": 0.5569831554453943, "grad_norm": 4.9088521003723145, "learning_rate": 4.179047953412273e-06, "loss": 0.3515, "step": 11011 }, { "epoch": 0.5570337396934595, "grad_norm": 11.323935508728027, "learning_rate": 4.17825624230121e-06, "loss": 0.4847, "step": 11012 }, { "epoch": 0.5570843239415246, "grad_norm": 3.733591318130493, "learning_rate": 4.177464552364498e-06, "loss": 0.372, "step": 11013 }, { "epoch": 0.5571349081895898, "grad_norm": 7.606769561767578, "learning_rate": 4.176672883622538e-06, "loss": 0.3897, "step": 11014 }, { "epoch": 0.5571854924376549, "grad_norm": 8.469435691833496, "learning_rate": 4.175881236095728e-06, "loss": 0.4429, "step": 11015 }, { "epoch": 0.5572360766857201, "grad_norm": 8.518168449401855, "learning_rate": 4.1750896098044695e-06, "loss": 0.3794, "step": 11016 }, { "epoch": 0.5572866609337852, "grad_norm": 8.037226676940918, "learning_rate": 4.1742980047691575e-06, "loss": 0.4953, "step": 11017 }, { "epoch": 0.5573372451818503, "grad_norm": 8.646185874938965, "learning_rate": 4.173506421010192e-06, "loss": 0.5308, "step": 11018 }, { "epoch": 0.5573878294299155, "grad_norm": 8.894355773925781, "learning_rate": 4.1727148585479695e-06, "loss": 0.3479, "step": 11019 }, { "epoch": 0.5574384136779806, "grad_norm": 13.389678001403809, "learning_rate": 4.171923317402887e-06, "loss": 0.4429, "step": 11020 }, { "epoch": 0.5574889979260458, "grad_norm": 7.275275230407715, "learning_rate": 4.171131797595341e-06, "loss": 0.4057, "step": 11021 }, { "epoch": 0.557539582174111, "grad_norm": 5.747311115264893, "learning_rate": 4.170340299145726e-06, "loss": 0.4393, "step": 11022 }, { "epoch": 0.5575901664221762, "grad_norm": 10.281835556030273, "learning_rate": 4.169548822074439e-06, "loss": 0.4394, "step": 11023 }, { "epoch": 0.5576407506702413, "grad_norm": 5.478023529052734, "learning_rate": 4.16875736640187e-06, "loss": 0.4166, "step": 11024 }, { "epoch": 0.5576913349183065, "grad_norm": 7.920683860778809, "learning_rate": 4.167965932148419e-06, "loss": 0.4689, "step": 11025 }, { "epoch": 0.5577419191663716, "grad_norm": 12.171296119689941, "learning_rate": 4.1671745193344765e-06, "loss": 0.5564, "step": 11026 }, { "epoch": 0.5577925034144368, "grad_norm": 13.58632755279541, "learning_rate": 4.166383127980433e-06, "loss": 0.3939, "step": 11027 }, { "epoch": 0.5578430876625019, "grad_norm": 9.020140647888184, "learning_rate": 4.165591758106683e-06, "loss": 0.4113, "step": 11028 }, { "epoch": 0.557893671910567, "grad_norm": 11.257862091064453, "learning_rate": 4.164800409733619e-06, "loss": 0.4246, "step": 11029 }, { "epoch": 0.5579442561586322, "grad_norm": 6.087007522583008, "learning_rate": 4.164009082881629e-06, "loss": 0.3073, "step": 11030 }, { "epoch": 0.5579948404066973, "grad_norm": 26.273752212524414, "learning_rate": 4.163217777571107e-06, "loss": 0.4451, "step": 11031 }, { "epoch": 0.5580454246547625, "grad_norm": 15.9475679397583, "learning_rate": 4.162426493822442e-06, "loss": 0.4798, "step": 11032 }, { "epoch": 0.5580960089028276, "grad_norm": 8.166740417480469, "learning_rate": 4.1616352316560225e-06, "loss": 0.5095, "step": 11033 }, { "epoch": 0.5581465931508928, "grad_norm": 11.354814529418945, "learning_rate": 4.160843991092239e-06, "loss": 0.5492, "step": 11034 }, { "epoch": 0.5581971773989579, "grad_norm": 7.254438400268555, "learning_rate": 4.160052772151479e-06, "loss": 0.394, "step": 11035 }, { "epoch": 0.5582477616470232, "grad_norm": 7.737250328063965, "learning_rate": 4.159261574854128e-06, "loss": 0.3872, "step": 11036 }, { "epoch": 0.5582983458950883, "grad_norm": 4.447696685791016, "learning_rate": 4.158470399220579e-06, "loss": 0.4216, "step": 11037 }, { "epoch": 0.5583489301431535, "grad_norm": 11.098973274230957, "learning_rate": 4.157679245271212e-06, "loss": 0.4643, "step": 11038 }, { "epoch": 0.5583995143912186, "grad_norm": 13.459650039672852, "learning_rate": 4.15688811302642e-06, "loss": 0.3986, "step": 11039 }, { "epoch": 0.5584500986392837, "grad_norm": 5.9172163009643555, "learning_rate": 4.1560970025065814e-06, "loss": 0.3233, "step": 11040 }, { "epoch": 0.5585006828873489, "grad_norm": 7.51697301864624, "learning_rate": 4.155305913732088e-06, "loss": 0.3496, "step": 11041 }, { "epoch": 0.558551267135414, "grad_norm": 7.70165491104126, "learning_rate": 4.154514846723317e-06, "loss": 0.3724, "step": 11042 }, { "epoch": 0.5586018513834792, "grad_norm": 5.024940013885498, "learning_rate": 4.1537238015006574e-06, "loss": 0.3587, "step": 11043 }, { "epoch": 0.5586524356315443, "grad_norm": 8.586230278015137, "learning_rate": 4.152932778084493e-06, "loss": 0.4392, "step": 11044 }, { "epoch": 0.5587030198796095, "grad_norm": 12.034444808959961, "learning_rate": 4.152141776495203e-06, "loss": 0.5408, "step": 11045 }, { "epoch": 0.5587536041276746, "grad_norm": 4.885894298553467, "learning_rate": 4.151350796753172e-06, "loss": 0.4372, "step": 11046 }, { "epoch": 0.5588041883757398, "grad_norm": 11.31220531463623, "learning_rate": 4.150559838878782e-06, "loss": 0.5542, "step": 11047 }, { "epoch": 0.5588547726238049, "grad_norm": 9.964719772338867, "learning_rate": 4.1497689028924124e-06, "loss": 0.5367, "step": 11048 }, { "epoch": 0.5589053568718702, "grad_norm": 6.906467437744141, "learning_rate": 4.148977988814444e-06, "loss": 0.366, "step": 11049 }, { "epoch": 0.5589559411199353, "grad_norm": 7.585912704467773, "learning_rate": 4.148187096665259e-06, "loss": 0.3644, "step": 11050 }, { "epoch": 0.5590065253680004, "grad_norm": 7.461092948913574, "learning_rate": 4.147396226465234e-06, "loss": 0.4899, "step": 11051 }, { "epoch": 0.5590571096160656, "grad_norm": 7.792698383331299, "learning_rate": 4.14660537823475e-06, "loss": 0.375, "step": 11052 }, { "epoch": 0.5591076938641307, "grad_norm": 6.780781269073486, "learning_rate": 4.145814551994183e-06, "loss": 0.388, "step": 11053 }, { "epoch": 0.5591582781121959, "grad_norm": 8.461909294128418, "learning_rate": 4.145023747763912e-06, "loss": 0.4796, "step": 11054 }, { "epoch": 0.559208862360261, "grad_norm": 9.07801628112793, "learning_rate": 4.144232965564314e-06, "loss": 0.5169, "step": 11055 }, { "epoch": 0.5592594466083262, "grad_norm": 8.798441886901855, "learning_rate": 4.143442205415764e-06, "loss": 0.4744, "step": 11056 }, { "epoch": 0.5593100308563913, "grad_norm": 8.827777862548828, "learning_rate": 4.142651467338642e-06, "loss": 0.4705, "step": 11057 }, { "epoch": 0.5593606151044564, "grad_norm": 11.320256233215332, "learning_rate": 4.141860751353319e-06, "loss": 0.4459, "step": 11058 }, { "epoch": 0.5594111993525216, "grad_norm": 9.119997024536133, "learning_rate": 4.141070057480171e-06, "loss": 0.4266, "step": 11059 }, { "epoch": 0.5594617836005867, "grad_norm": 7.8506927490234375, "learning_rate": 4.140279385739575e-06, "loss": 0.4692, "step": 11060 }, { "epoch": 0.5595123678486519, "grad_norm": 8.177780151367188, "learning_rate": 4.139488736151902e-06, "loss": 0.3626, "step": 11061 }, { "epoch": 0.5595629520967171, "grad_norm": 6.535254955291748, "learning_rate": 4.138698108737525e-06, "loss": 0.4142, "step": 11062 }, { "epoch": 0.5596135363447823, "grad_norm": 5.471635818481445, "learning_rate": 4.1379075035168184e-06, "loss": 0.3798, "step": 11063 }, { "epoch": 0.5596641205928474, "grad_norm": 16.456714630126953, "learning_rate": 4.137116920510153e-06, "loss": 0.5414, "step": 11064 }, { "epoch": 0.5597147048409126, "grad_norm": 9.447092056274414, "learning_rate": 4.136326359737901e-06, "loss": 0.5974, "step": 11065 }, { "epoch": 0.5597652890889777, "grad_norm": 9.652474403381348, "learning_rate": 4.135535821220431e-06, "loss": 0.4123, "step": 11066 }, { "epoch": 0.5598158733370429, "grad_norm": 7.301605224609375, "learning_rate": 4.134745304978116e-06, "loss": 0.396, "step": 11067 }, { "epoch": 0.559866457585108, "grad_norm": 5.535146713256836, "learning_rate": 4.1339548110313235e-06, "loss": 0.4151, "step": 11068 }, { "epoch": 0.5599170418331731, "grad_norm": 7.113329887390137, "learning_rate": 4.133164339400424e-06, "loss": 0.436, "step": 11069 }, { "epoch": 0.5599676260812383, "grad_norm": 9.353029251098633, "learning_rate": 4.132373890105787e-06, "loss": 0.3853, "step": 11070 }, { "epoch": 0.5600182103293034, "grad_norm": 14.286489486694336, "learning_rate": 4.131583463167778e-06, "loss": 0.5454, "step": 11071 }, { "epoch": 0.5600687945773686, "grad_norm": 5.309208869934082, "learning_rate": 4.130793058606766e-06, "loss": 0.3665, "step": 11072 }, { "epoch": 0.5601193788254337, "grad_norm": 5.126429557800293, "learning_rate": 4.130002676443117e-06, "loss": 0.3619, "step": 11073 }, { "epoch": 0.5601699630734989, "grad_norm": 10.663544654846191, "learning_rate": 4.129212316697197e-06, "loss": 0.4753, "step": 11074 }, { "epoch": 0.560220547321564, "grad_norm": 8.663613319396973, "learning_rate": 4.128421979389373e-06, "loss": 0.3913, "step": 11075 }, { "epoch": 0.5602711315696293, "grad_norm": 8.380215644836426, "learning_rate": 4.1276316645400095e-06, "loss": 0.3795, "step": 11076 }, { "epoch": 0.5603217158176944, "grad_norm": 7.6593170166015625, "learning_rate": 4.12684137216947e-06, "loss": 0.3637, "step": 11077 }, { "epoch": 0.5603723000657596, "grad_norm": 10.274335861206055, "learning_rate": 4.126051102298122e-06, "loss": 0.4953, "step": 11078 }, { "epoch": 0.5604228843138247, "grad_norm": 5.8824076652526855, "learning_rate": 4.1252608549463245e-06, "loss": 0.336, "step": 11079 }, { "epoch": 0.5604734685618898, "grad_norm": 10.128301620483398, "learning_rate": 4.1244706301344415e-06, "loss": 0.4291, "step": 11080 }, { "epoch": 0.560524052809955, "grad_norm": 6.256049633026123, "learning_rate": 4.123680427882836e-06, "loss": 0.2887, "step": 11081 }, { "epoch": 0.5605746370580201, "grad_norm": 5.864253044128418, "learning_rate": 4.122890248211869e-06, "loss": 0.3602, "step": 11082 }, { "epoch": 0.5606252213060853, "grad_norm": 7.191356658935547, "learning_rate": 4.122100091141902e-06, "loss": 0.4538, "step": 11083 }, { "epoch": 0.5606758055541504, "grad_norm": 7.254965305328369, "learning_rate": 4.121309956693294e-06, "loss": 0.4417, "step": 11084 }, { "epoch": 0.5607263898022156, "grad_norm": 10.749311447143555, "learning_rate": 4.1205198448864074e-06, "loss": 0.3793, "step": 11085 }, { "epoch": 0.5607769740502807, "grad_norm": 7.335798263549805, "learning_rate": 4.1197297557415995e-06, "loss": 0.3606, "step": 11086 }, { "epoch": 0.5608275582983459, "grad_norm": 5.397397041320801, "learning_rate": 4.118939689279229e-06, "loss": 0.4111, "step": 11087 }, { "epoch": 0.560878142546411, "grad_norm": 9.884485244750977, "learning_rate": 4.118149645519656e-06, "loss": 0.4363, "step": 11088 }, { "epoch": 0.5609287267944763, "grad_norm": 14.686775207519531, "learning_rate": 4.1173596244832355e-06, "loss": 0.4739, "step": 11089 }, { "epoch": 0.5609793110425414, "grad_norm": 7.178137302398682, "learning_rate": 4.116569626190325e-06, "loss": 0.4221, "step": 11090 }, { "epoch": 0.5610298952906065, "grad_norm": 10.039715766906738, "learning_rate": 4.115779650661285e-06, "loss": 0.605, "step": 11091 }, { "epoch": 0.5610804795386717, "grad_norm": 6.475950717926025, "learning_rate": 4.114989697916465e-06, "loss": 0.3595, "step": 11092 }, { "epoch": 0.5611310637867368, "grad_norm": 6.994489669799805, "learning_rate": 4.1141997679762224e-06, "loss": 0.3589, "step": 11093 }, { "epoch": 0.561181648034802, "grad_norm": 6.87763786315918, "learning_rate": 4.1134098608609125e-06, "loss": 0.3062, "step": 11094 }, { "epoch": 0.5612322322828671, "grad_norm": 9.226722717285156, "learning_rate": 4.11261997659089e-06, "loss": 0.4268, "step": 11095 }, { "epoch": 0.5612828165309323, "grad_norm": 5.160466194152832, "learning_rate": 4.111830115186506e-06, "loss": 0.2881, "step": 11096 }, { "epoch": 0.5613334007789974, "grad_norm": 7.564407825469971, "learning_rate": 4.111040276668114e-06, "loss": 0.3834, "step": 11097 }, { "epoch": 0.5613839850270625, "grad_norm": 8.77158260345459, "learning_rate": 4.1102504610560685e-06, "loss": 0.4546, "step": 11098 }, { "epoch": 0.5614345692751277, "grad_norm": 6.786374092102051, "learning_rate": 4.1094606683707195e-06, "loss": 0.3799, "step": 11099 }, { "epoch": 0.5614851535231928, "grad_norm": 15.104809761047363, "learning_rate": 4.108670898632417e-06, "loss": 0.5329, "step": 11100 }, { "epoch": 0.561535737771258, "grad_norm": 7.062504768371582, "learning_rate": 4.107881151861514e-06, "loss": 0.3892, "step": 11101 }, { "epoch": 0.5615863220193232, "grad_norm": 11.223766326904297, "learning_rate": 4.107091428078357e-06, "loss": 0.465, "step": 11102 }, { "epoch": 0.5616369062673884, "grad_norm": 4.960606098175049, "learning_rate": 4.1063017273033005e-06, "loss": 0.3531, "step": 11103 }, { "epoch": 0.5616874905154535, "grad_norm": 13.648555755615234, "learning_rate": 4.105512049556687e-06, "loss": 0.4846, "step": 11104 }, { "epoch": 0.5617380747635187, "grad_norm": 5.178020000457764, "learning_rate": 4.104722394858869e-06, "loss": 0.3004, "step": 11105 }, { "epoch": 0.5617886590115838, "grad_norm": 25.45607566833496, "learning_rate": 4.103932763230191e-06, "loss": 0.5012, "step": 11106 }, { "epoch": 0.561839243259649, "grad_norm": 11.522644996643066, "learning_rate": 4.103143154691002e-06, "loss": 0.4314, "step": 11107 }, { "epoch": 0.5618898275077141, "grad_norm": 4.979147911071777, "learning_rate": 4.102353569261647e-06, "loss": 0.404, "step": 11108 }, { "epoch": 0.5619404117557792, "grad_norm": 11.604954719543457, "learning_rate": 4.101564006962472e-06, "loss": 0.4754, "step": 11109 }, { "epoch": 0.5619909960038444, "grad_norm": 9.161393165588379, "learning_rate": 4.100774467813824e-06, "loss": 0.3637, "step": 11110 }, { "epoch": 0.5620415802519095, "grad_norm": 10.651744842529297, "learning_rate": 4.099984951836044e-06, "loss": 0.3657, "step": 11111 }, { "epoch": 0.5620921644999747, "grad_norm": 5.025047779083252, "learning_rate": 4.09919545904948e-06, "loss": 0.336, "step": 11112 }, { "epoch": 0.5621427487480398, "grad_norm": 9.256253242492676, "learning_rate": 4.098405989474472e-06, "loss": 0.3646, "step": 11113 }, { "epoch": 0.562193332996105, "grad_norm": 11.720292091369629, "learning_rate": 4.097616543131364e-06, "loss": 0.4665, "step": 11114 }, { "epoch": 0.5622439172441701, "grad_norm": 6.482821941375732, "learning_rate": 4.096827120040497e-06, "loss": 0.4313, "step": 11115 }, { "epoch": 0.5622945014922354, "grad_norm": 9.338899612426758, "learning_rate": 4.096037720222217e-06, "loss": 0.4485, "step": 11116 }, { "epoch": 0.5623450857403005, "grad_norm": 8.512475967407227, "learning_rate": 4.095248343696861e-06, "loss": 0.449, "step": 11117 }, { "epoch": 0.5623956699883657, "grad_norm": 6.719726085662842, "learning_rate": 4.094458990484767e-06, "loss": 0.3499, "step": 11118 }, { "epoch": 0.5624462542364308, "grad_norm": 5.890857696533203, "learning_rate": 4.093669660606279e-06, "loss": 0.3653, "step": 11119 }, { "epoch": 0.5624968384844959, "grad_norm": 6.7922868728637695, "learning_rate": 4.092880354081734e-06, "loss": 0.479, "step": 11120 }, { "epoch": 0.5625474227325611, "grad_norm": 28.124940872192383, "learning_rate": 4.092091070931472e-06, "loss": 0.4617, "step": 11121 }, { "epoch": 0.5625980069806262, "grad_norm": 7.697853088378906, "learning_rate": 4.091301811175829e-06, "loss": 0.4339, "step": 11122 }, { "epoch": 0.5626485912286914, "grad_norm": 7.772347450256348, "learning_rate": 4.090512574835145e-06, "loss": 0.408, "step": 11123 }, { "epoch": 0.5626991754767565, "grad_norm": 9.196916580200195, "learning_rate": 4.089723361929754e-06, "loss": 0.4104, "step": 11124 }, { "epoch": 0.5627497597248217, "grad_norm": 9.186793327331543, "learning_rate": 4.088934172479994e-06, "loss": 0.3928, "step": 11125 }, { "epoch": 0.5628003439728868, "grad_norm": 8.657657623291016, "learning_rate": 4.0881450065061995e-06, "loss": 0.3349, "step": 11126 }, { "epoch": 0.562850928220952, "grad_norm": 7.976541519165039, "learning_rate": 4.087355864028705e-06, "loss": 0.3858, "step": 11127 }, { "epoch": 0.5629015124690171, "grad_norm": 11.930261611938477, "learning_rate": 4.086566745067846e-06, "loss": 0.3499, "step": 11128 }, { "epoch": 0.5629520967170824, "grad_norm": 7.145981788635254, "learning_rate": 4.085777649643958e-06, "loss": 0.3276, "step": 11129 }, { "epoch": 0.5630026809651475, "grad_norm": 9.527878761291504, "learning_rate": 4.08498857777737e-06, "loss": 0.356, "step": 11130 }, { "epoch": 0.5630532652132126, "grad_norm": 8.963592529296875, "learning_rate": 4.084199529488415e-06, "loss": 0.4541, "step": 11131 }, { "epoch": 0.5631038494612778, "grad_norm": 8.856009483337402, "learning_rate": 4.083410504797427e-06, "loss": 0.434, "step": 11132 }, { "epoch": 0.5631544337093429, "grad_norm": 6.680952548980713, "learning_rate": 4.082621503724737e-06, "loss": 0.4522, "step": 11133 }, { "epoch": 0.5632050179574081, "grad_norm": 9.452777862548828, "learning_rate": 4.081832526290674e-06, "loss": 0.5221, "step": 11134 }, { "epoch": 0.5632556022054732, "grad_norm": 7.853662014007568, "learning_rate": 4.081043572515568e-06, "loss": 0.3479, "step": 11135 }, { "epoch": 0.5633061864535384, "grad_norm": 10.846131324768066, "learning_rate": 4.080254642419752e-06, "loss": 0.3791, "step": 11136 }, { "epoch": 0.5633567707016035, "grad_norm": 6.9736328125, "learning_rate": 4.079465736023552e-06, "loss": 0.4377, "step": 11137 }, { "epoch": 0.5634073549496686, "grad_norm": 7.7753095626831055, "learning_rate": 4.078676853347296e-06, "loss": 0.3752, "step": 11138 }, { "epoch": 0.5634579391977338, "grad_norm": 7.526042938232422, "learning_rate": 4.0778879944113105e-06, "loss": 0.4371, "step": 11139 }, { "epoch": 0.5635085234457989, "grad_norm": 7.239795684814453, "learning_rate": 4.077099159235925e-06, "loss": 0.4106, "step": 11140 }, { "epoch": 0.5635591076938641, "grad_norm": 15.915800094604492, "learning_rate": 4.076310347841466e-06, "loss": 0.5941, "step": 11141 }, { "epoch": 0.5636096919419293, "grad_norm": 9.356232643127441, "learning_rate": 4.075521560248258e-06, "loss": 0.4146, "step": 11142 }, { "epoch": 0.5636602761899945, "grad_norm": 9.71642780303955, "learning_rate": 4.074732796476627e-06, "loss": 0.4176, "step": 11143 }, { "epoch": 0.5637108604380596, "grad_norm": 6.062382221221924, "learning_rate": 4.073944056546896e-06, "loss": 0.4079, "step": 11144 }, { "epoch": 0.5637614446861248, "grad_norm": 10.509622573852539, "learning_rate": 4.073155340479391e-06, "loss": 0.4397, "step": 11145 }, { "epoch": 0.5638120289341899, "grad_norm": 8.682971954345703, "learning_rate": 4.072366648294432e-06, "loss": 0.4206, "step": 11146 }, { "epoch": 0.5638626131822551, "grad_norm": 8.14902400970459, "learning_rate": 4.071577980012344e-06, "loss": 0.3467, "step": 11147 }, { "epoch": 0.5639131974303202, "grad_norm": 9.537331581115723, "learning_rate": 4.070789335653451e-06, "loss": 0.3934, "step": 11148 }, { "epoch": 0.5639637816783853, "grad_norm": 8.370652198791504, "learning_rate": 4.0700007152380715e-06, "loss": 0.5024, "step": 11149 }, { "epoch": 0.5640143659264505, "grad_norm": 5.792240619659424, "learning_rate": 4.069212118786526e-06, "loss": 0.3312, "step": 11150 }, { "epoch": 0.5640649501745156, "grad_norm": 8.143104553222656, "learning_rate": 4.068423546319138e-06, "loss": 0.3962, "step": 11151 }, { "epoch": 0.5641155344225808, "grad_norm": 6.533952236175537, "learning_rate": 4.067634997856224e-06, "loss": 0.3441, "step": 11152 }, { "epoch": 0.5641661186706459, "grad_norm": 5.973147392272949, "learning_rate": 4.066846473418103e-06, "loss": 0.3348, "step": 11153 }, { "epoch": 0.5642167029187111, "grad_norm": 6.442481517791748, "learning_rate": 4.0660579730250966e-06, "loss": 0.3646, "step": 11154 }, { "epoch": 0.5642672871667763, "grad_norm": 13.315258026123047, "learning_rate": 4.065269496697521e-06, "loss": 0.5827, "step": 11155 }, { "epoch": 0.5643178714148415, "grad_norm": 10.860992431640625, "learning_rate": 4.06448104445569e-06, "loss": 0.4651, "step": 11156 }, { "epoch": 0.5643684556629066, "grad_norm": 7.488465785980225, "learning_rate": 4.063692616319924e-06, "loss": 0.3402, "step": 11157 }, { "epoch": 0.5644190399109718, "grad_norm": 7.034938335418701, "learning_rate": 4.0629042123105385e-06, "loss": 0.3463, "step": 11158 }, { "epoch": 0.5644696241590369, "grad_norm": 6.748338222503662, "learning_rate": 4.062115832447847e-06, "loss": 0.4561, "step": 11159 }, { "epoch": 0.564520208407102, "grad_norm": 6.332359790802002, "learning_rate": 4.061327476752164e-06, "loss": 0.409, "step": 11160 }, { "epoch": 0.5645707926551672, "grad_norm": 5.3048415184021, "learning_rate": 4.060539145243807e-06, "loss": 0.3636, "step": 11161 }, { "epoch": 0.5646213769032323, "grad_norm": 11.082212448120117, "learning_rate": 4.0597508379430856e-06, "loss": 0.4634, "step": 11162 }, { "epoch": 0.5646719611512975, "grad_norm": 9.187224388122559, "learning_rate": 4.0589625548703136e-06, "loss": 0.425, "step": 11163 }, { "epoch": 0.5647225453993626, "grad_norm": 7.990313529968262, "learning_rate": 4.058174296045805e-06, "loss": 0.4323, "step": 11164 }, { "epoch": 0.5647731296474278, "grad_norm": 9.62938404083252, "learning_rate": 4.057386061489869e-06, "loss": 0.4102, "step": 11165 }, { "epoch": 0.5648237138954929, "grad_norm": 4.791843891143799, "learning_rate": 4.056597851222817e-06, "loss": 0.4, "step": 11166 }, { "epoch": 0.564874298143558, "grad_norm": 3.7318341732025146, "learning_rate": 4.055809665264961e-06, "loss": 0.3012, "step": 11167 }, { "epoch": 0.5649248823916232, "grad_norm": 6.655050754547119, "learning_rate": 4.055021503636611e-06, "loss": 0.3663, "step": 11168 }, { "epoch": 0.5649754666396885, "grad_norm": 5.115368843078613, "learning_rate": 4.0542333663580715e-06, "loss": 0.3327, "step": 11169 }, { "epoch": 0.5650260508877536, "grad_norm": 7.235857963562012, "learning_rate": 4.053445253449654e-06, "loss": 0.3762, "step": 11170 }, { "epoch": 0.5650766351358187, "grad_norm": 7.731409549713135, "learning_rate": 4.052657164931667e-06, "loss": 0.3708, "step": 11171 }, { "epoch": 0.5651272193838839, "grad_norm": 7.584218978881836, "learning_rate": 4.051869100824416e-06, "loss": 0.3298, "step": 11172 }, { "epoch": 0.565177803631949, "grad_norm": 6.362749099731445, "learning_rate": 4.051081061148207e-06, "loss": 0.4579, "step": 11173 }, { "epoch": 0.5652283878800142, "grad_norm": 21.640369415283203, "learning_rate": 4.050293045923349e-06, "loss": 0.5146, "step": 11174 }, { "epoch": 0.5652789721280793, "grad_norm": 10.452498435974121, "learning_rate": 4.049505055170144e-06, "loss": 0.4194, "step": 11175 }, { "epoch": 0.5653295563761445, "grad_norm": 10.185606002807617, "learning_rate": 4.0487170889088986e-06, "loss": 0.5555, "step": 11176 }, { "epoch": 0.5653801406242096, "grad_norm": 9.333983421325684, "learning_rate": 4.047929147159916e-06, "loss": 0.4556, "step": 11177 }, { "epoch": 0.5654307248722747, "grad_norm": 6.818216800689697, "learning_rate": 4.047141229943498e-06, "loss": 0.4955, "step": 11178 }, { "epoch": 0.5654813091203399, "grad_norm": 5.010903358459473, "learning_rate": 4.0463533372799515e-06, "loss": 0.3221, "step": 11179 }, { "epoch": 0.565531893368405, "grad_norm": 9.527398109436035, "learning_rate": 4.045565469189574e-06, "loss": 0.495, "step": 11180 }, { "epoch": 0.5655824776164702, "grad_norm": 23.196651458740234, "learning_rate": 4.0447776256926725e-06, "loss": 0.4219, "step": 11181 }, { "epoch": 0.5656330618645354, "grad_norm": 10.01579761505127, "learning_rate": 4.043989806809541e-06, "loss": 0.4762, "step": 11182 }, { "epoch": 0.5656836461126006, "grad_norm": 6.596877574920654, "learning_rate": 4.043202012560485e-06, "loss": 0.4337, "step": 11183 }, { "epoch": 0.5657342303606657, "grad_norm": 7.369583606719971, "learning_rate": 4.0424142429658e-06, "loss": 0.5052, "step": 11184 }, { "epoch": 0.5657848146087309, "grad_norm": 5.862614631652832, "learning_rate": 4.041626498045788e-06, "loss": 0.3855, "step": 11185 }, { "epoch": 0.565835398856796, "grad_norm": 6.841746807098389, "learning_rate": 4.040838777820747e-06, "loss": 0.4118, "step": 11186 }, { "epoch": 0.5658859831048612, "grad_norm": 9.49453353881836, "learning_rate": 4.040051082310973e-06, "loss": 0.3997, "step": 11187 }, { "epoch": 0.5659365673529263, "grad_norm": 7.0366692543029785, "learning_rate": 4.039263411536764e-06, "loss": 0.5198, "step": 11188 }, { "epoch": 0.5659871516009914, "grad_norm": 13.782426834106445, "learning_rate": 4.038475765518417e-06, "loss": 0.5887, "step": 11189 }, { "epoch": 0.5660377358490566, "grad_norm": 7.900674343109131, "learning_rate": 4.037688144276226e-06, "loss": 0.4837, "step": 11190 }, { "epoch": 0.5660883200971217, "grad_norm": 6.031805515289307, "learning_rate": 4.036900547830487e-06, "loss": 0.3962, "step": 11191 }, { "epoch": 0.5661389043451869, "grad_norm": 5.7885308265686035, "learning_rate": 4.0361129762014954e-06, "loss": 0.3814, "step": 11192 }, { "epoch": 0.566189488593252, "grad_norm": 4.69010591506958, "learning_rate": 4.035325429409544e-06, "loss": 0.3196, "step": 11193 }, { "epoch": 0.5662400728413172, "grad_norm": 7.709985256195068, "learning_rate": 4.034537907474928e-06, "loss": 0.414, "step": 11194 }, { "epoch": 0.5662906570893824, "grad_norm": 14.949455261230469, "learning_rate": 4.033750410417936e-06, "loss": 0.6053, "step": 11195 }, { "epoch": 0.5663412413374476, "grad_norm": 7.224057674407959, "learning_rate": 4.032962938258863e-06, "loss": 0.4059, "step": 11196 }, { "epoch": 0.5663918255855127, "grad_norm": 8.303173065185547, "learning_rate": 4.032175491017999e-06, "loss": 0.4769, "step": 11197 }, { "epoch": 0.5664424098335779, "grad_norm": 10.279760360717773, "learning_rate": 4.031388068715634e-06, "loss": 0.5467, "step": 11198 }, { "epoch": 0.566492994081643, "grad_norm": 5.6870832443237305, "learning_rate": 4.0306006713720605e-06, "loss": 0.3433, "step": 11199 }, { "epoch": 0.5665435783297081, "grad_norm": 5.839780330657959, "learning_rate": 4.029813299007565e-06, "loss": 0.4158, "step": 11200 }, { "epoch": 0.5665941625777733, "grad_norm": 5.260185718536377, "learning_rate": 4.029025951642437e-06, "loss": 0.4375, "step": 11201 }, { "epoch": 0.5666447468258384, "grad_norm": 7.213045120239258, "learning_rate": 4.0282386292969664e-06, "loss": 0.3725, "step": 11202 }, { "epoch": 0.5666953310739036, "grad_norm": 7.575104713439941, "learning_rate": 4.027451331991437e-06, "loss": 0.4633, "step": 11203 }, { "epoch": 0.5667459153219687, "grad_norm": 4.275501728057861, "learning_rate": 4.0266640597461396e-06, "loss": 0.3618, "step": 11204 }, { "epoch": 0.5667964995700339, "grad_norm": 8.280046463012695, "learning_rate": 4.025876812581358e-06, "loss": 0.4185, "step": 11205 }, { "epoch": 0.566847083818099, "grad_norm": 8.9747953414917, "learning_rate": 4.02508959051738e-06, "loss": 0.4292, "step": 11206 }, { "epoch": 0.5668976680661642, "grad_norm": 7.186443328857422, "learning_rate": 4.024302393574486e-06, "loss": 0.3377, "step": 11207 }, { "epoch": 0.5669482523142293, "grad_norm": 10.549230575561523, "learning_rate": 4.023515221772962e-06, "loss": 0.4737, "step": 11208 }, { "epoch": 0.5669988365622946, "grad_norm": 11.727112770080566, "learning_rate": 4.022728075133094e-06, "loss": 0.5672, "step": 11209 }, { "epoch": 0.5670494208103597, "grad_norm": 7.414634704589844, "learning_rate": 4.021940953675162e-06, "loss": 0.3386, "step": 11210 }, { "epoch": 0.5671000050584248, "grad_norm": 9.574975967407227, "learning_rate": 4.021153857419449e-06, "loss": 0.5441, "step": 11211 }, { "epoch": 0.56715058930649, "grad_norm": 8.930965423583984, "learning_rate": 4.020366786386237e-06, "loss": 0.4176, "step": 11212 }, { "epoch": 0.5672011735545551, "grad_norm": 10.744973182678223, "learning_rate": 4.019579740595807e-06, "loss": 0.4706, "step": 11213 }, { "epoch": 0.5672517578026203, "grad_norm": 13.49338436126709, "learning_rate": 4.018792720068439e-06, "loss": 0.5384, "step": 11214 }, { "epoch": 0.5673023420506854, "grad_norm": 14.624371528625488, "learning_rate": 4.0180057248244116e-06, "loss": 0.5535, "step": 11215 }, { "epoch": 0.5673529262987506, "grad_norm": 8.990986824035645, "learning_rate": 4.017218754884005e-06, "loss": 0.4163, "step": 11216 }, { "epoch": 0.5674035105468157, "grad_norm": 6.25588846206665, "learning_rate": 4.016431810267499e-06, "loss": 0.3967, "step": 11217 }, { "epoch": 0.5674540947948808, "grad_norm": 11.000308990478516, "learning_rate": 4.015644890995167e-06, "loss": 0.4514, "step": 11218 }, { "epoch": 0.567504679042946, "grad_norm": 6.2928853034973145, "learning_rate": 4.014857997087292e-06, "loss": 0.4232, "step": 11219 }, { "epoch": 0.5675552632910111, "grad_norm": 6.206925392150879, "learning_rate": 4.014071128564144e-06, "loss": 0.4523, "step": 11220 }, { "epoch": 0.5676058475390763, "grad_norm": 8.750236511230469, "learning_rate": 4.013284285446003e-06, "loss": 0.4999, "step": 11221 }, { "epoch": 0.5676564317871415, "grad_norm": 7.281906604766846, "learning_rate": 4.012497467753142e-06, "loss": 0.3957, "step": 11222 }, { "epoch": 0.5677070160352067, "grad_norm": 6.410316467285156, "learning_rate": 4.011710675505835e-06, "loss": 0.43, "step": 11223 }, { "epoch": 0.5677576002832718, "grad_norm": 5.711902618408203, "learning_rate": 4.010923908724359e-06, "loss": 0.4043, "step": 11224 }, { "epoch": 0.567808184531337, "grad_norm": 10.64289379119873, "learning_rate": 4.010137167428984e-06, "loss": 0.4945, "step": 11225 }, { "epoch": 0.5678587687794021, "grad_norm": 7.9675517082214355, "learning_rate": 4.0093504516399826e-06, "loss": 0.4579, "step": 11226 }, { "epoch": 0.5679093530274673, "grad_norm": 8.69527816772461, "learning_rate": 4.008563761377628e-06, "loss": 0.4201, "step": 11227 }, { "epoch": 0.5679599372755324, "grad_norm": 6.305741786956787, "learning_rate": 4.00777709666219e-06, "loss": 0.3929, "step": 11228 }, { "epoch": 0.5680105215235975, "grad_norm": 10.207465171813965, "learning_rate": 4.006990457513939e-06, "loss": 0.6054, "step": 11229 }, { "epoch": 0.5680611057716627, "grad_norm": 6.632786273956299, "learning_rate": 4.006203843953147e-06, "loss": 0.4627, "step": 11230 }, { "epoch": 0.5681116900197278, "grad_norm": 19.122108459472656, "learning_rate": 4.00541725600008e-06, "loss": 0.4212, "step": 11231 }, { "epoch": 0.568162274267793, "grad_norm": 8.803292274475098, "learning_rate": 4.004630693675011e-06, "loss": 0.4564, "step": 11232 }, { "epoch": 0.5682128585158581, "grad_norm": 5.8003950119018555, "learning_rate": 4.003844156998201e-06, "loss": 0.4212, "step": 11233 }, { "epoch": 0.5682634427639233, "grad_norm": 12.479046821594238, "learning_rate": 4.003057645989924e-06, "loss": 0.5414, "step": 11234 }, { "epoch": 0.5683140270119885, "grad_norm": 7.124624252319336, "learning_rate": 4.002271160670441e-06, "loss": 0.4118, "step": 11235 }, { "epoch": 0.5683646112600537, "grad_norm": 10.10234546661377, "learning_rate": 4.001484701060021e-06, "loss": 0.499, "step": 11236 }, { "epoch": 0.5684151955081188, "grad_norm": 9.909796714782715, "learning_rate": 4.00069826717893e-06, "loss": 0.4618, "step": 11237 }, { "epoch": 0.568465779756184, "grad_norm": 10.315932273864746, "learning_rate": 3.999911859047428e-06, "loss": 0.4485, "step": 11238 }, { "epoch": 0.5685163640042491, "grad_norm": 8.191400527954102, "learning_rate": 3.999125476685783e-06, "loss": 0.4094, "step": 11239 }, { "epoch": 0.5685669482523142, "grad_norm": 5.754266262054443, "learning_rate": 3.998339120114259e-06, "loss": 0.4211, "step": 11240 }, { "epoch": 0.5686175325003794, "grad_norm": 10.534490585327148, "learning_rate": 3.997552789353114e-06, "loss": 0.3825, "step": 11241 }, { "epoch": 0.5686681167484445, "grad_norm": 9.70028305053711, "learning_rate": 3.996766484422613e-06, "loss": 0.464, "step": 11242 }, { "epoch": 0.5687187009965097, "grad_norm": 9.741057395935059, "learning_rate": 3.995980205343017e-06, "loss": 0.5596, "step": 11243 }, { "epoch": 0.5687692852445748, "grad_norm": 8.235041618347168, "learning_rate": 3.995193952134585e-06, "loss": 0.3623, "step": 11244 }, { "epoch": 0.56881986949264, "grad_norm": 5.882604122161865, "learning_rate": 3.99440772481758e-06, "loss": 0.3237, "step": 11245 }, { "epoch": 0.5688704537407051, "grad_norm": 10.562812805175781, "learning_rate": 3.9936215234122565e-06, "loss": 0.4098, "step": 11246 }, { "epoch": 0.5689210379887703, "grad_norm": 8.98307991027832, "learning_rate": 3.992835347938877e-06, "loss": 0.4306, "step": 11247 }, { "epoch": 0.5689716222368354, "grad_norm": 7.919588088989258, "learning_rate": 3.992049198417697e-06, "loss": 0.4105, "step": 11248 }, { "epoch": 0.5690222064849007, "grad_norm": 6.889029026031494, "learning_rate": 3.991263074868976e-06, "loss": 0.311, "step": 11249 }, { "epoch": 0.5690727907329658, "grad_norm": 9.836906433105469, "learning_rate": 3.990476977312967e-06, "loss": 0.2886, "step": 11250 }, { "epoch": 0.5691233749810309, "grad_norm": 12.521599769592285, "learning_rate": 3.989690905769929e-06, "loss": 0.4944, "step": 11251 }, { "epoch": 0.5691739592290961, "grad_norm": 9.215225219726562, "learning_rate": 3.9889048602601165e-06, "loss": 0.4473, "step": 11252 }, { "epoch": 0.5692245434771612, "grad_norm": 8.874650955200195, "learning_rate": 3.988118840803782e-06, "loss": 0.4599, "step": 11253 }, { "epoch": 0.5692751277252264, "grad_norm": 8.176652908325195, "learning_rate": 3.987332847421181e-06, "loss": 0.3347, "step": 11254 }, { "epoch": 0.5693257119732915, "grad_norm": 9.141708374023438, "learning_rate": 3.986546880132569e-06, "loss": 0.3937, "step": 11255 }, { "epoch": 0.5693762962213567, "grad_norm": 5.302593231201172, "learning_rate": 3.985760938958194e-06, "loss": 0.3859, "step": 11256 }, { "epoch": 0.5694268804694218, "grad_norm": 9.760774612426758, "learning_rate": 3.98497502391831e-06, "loss": 0.4668, "step": 11257 }, { "epoch": 0.569477464717487, "grad_norm": 7.553163528442383, "learning_rate": 3.984189135033169e-06, "loss": 0.3773, "step": 11258 }, { "epoch": 0.5695280489655521, "grad_norm": 23.576452255249023, "learning_rate": 3.98340327232302e-06, "loss": 0.5096, "step": 11259 }, { "epoch": 0.5695786332136172, "grad_norm": 10.675267219543457, "learning_rate": 3.982617435808113e-06, "loss": 0.4521, "step": 11260 }, { "epoch": 0.5696292174616824, "grad_norm": 10.393298149108887, "learning_rate": 3.981831625508696e-06, "loss": 0.3529, "step": 11261 }, { "epoch": 0.5696798017097476, "grad_norm": 9.84874153137207, "learning_rate": 3.98104584144502e-06, "loss": 0.4112, "step": 11262 }, { "epoch": 0.5697303859578128, "grad_norm": 5.949604034423828, "learning_rate": 3.980260083637331e-06, "loss": 0.3583, "step": 11263 }, { "epoch": 0.5697809702058779, "grad_norm": 7.883159637451172, "learning_rate": 3.9794743521058765e-06, "loss": 0.4598, "step": 11264 }, { "epoch": 0.5698315544539431, "grad_norm": 12.592716217041016, "learning_rate": 3.978688646870904e-06, "loss": 0.4449, "step": 11265 }, { "epoch": 0.5698821387020082, "grad_norm": 7.58826208114624, "learning_rate": 3.977902967952657e-06, "loss": 0.4669, "step": 11266 }, { "epoch": 0.5699327229500734, "grad_norm": 6.188817024230957, "learning_rate": 3.977117315371381e-06, "loss": 0.4566, "step": 11267 }, { "epoch": 0.5699833071981385, "grad_norm": 6.386560916900635, "learning_rate": 3.976331689147323e-06, "loss": 0.3061, "step": 11268 }, { "epoch": 0.5700338914462036, "grad_norm": 6.788016319274902, "learning_rate": 3.975546089300724e-06, "loss": 0.4144, "step": 11269 }, { "epoch": 0.5700844756942688, "grad_norm": 7.6411213874816895, "learning_rate": 3.974760515851828e-06, "loss": 0.4677, "step": 11270 }, { "epoch": 0.5701350599423339, "grad_norm": 6.708393096923828, "learning_rate": 3.9739749688208785e-06, "loss": 0.4048, "step": 11271 }, { "epoch": 0.5701856441903991, "grad_norm": 7.939519882202148, "learning_rate": 3.973189448228115e-06, "loss": 0.4536, "step": 11272 }, { "epoch": 0.5702362284384642, "grad_norm": 6.903262138366699, "learning_rate": 3.972403954093778e-06, "loss": 0.3592, "step": 11273 }, { "epoch": 0.5702868126865294, "grad_norm": 8.195672035217285, "learning_rate": 3.971618486438109e-06, "loss": 0.3916, "step": 11274 }, { "epoch": 0.5703373969345946, "grad_norm": 8.24510383605957, "learning_rate": 3.970833045281348e-06, "loss": 0.3925, "step": 11275 }, { "epoch": 0.5703879811826598, "grad_norm": 7.045628547668457, "learning_rate": 3.970047630643734e-06, "loss": 0.3614, "step": 11276 }, { "epoch": 0.5704385654307249, "grad_norm": 7.326467990875244, "learning_rate": 3.969262242545505e-06, "loss": 0.3165, "step": 11277 }, { "epoch": 0.57048914967879, "grad_norm": 9.762459754943848, "learning_rate": 3.968476881006898e-06, "loss": 0.5019, "step": 11278 }, { "epoch": 0.5705397339268552, "grad_norm": 14.250651359558105, "learning_rate": 3.96769154604815e-06, "loss": 0.5426, "step": 11279 }, { "epoch": 0.5705903181749203, "grad_norm": 9.310458183288574, "learning_rate": 3.966906237689498e-06, "loss": 0.4591, "step": 11280 }, { "epoch": 0.5706409024229855, "grad_norm": 6.388774394989014, "learning_rate": 3.966120955951176e-06, "loss": 0.4066, "step": 11281 }, { "epoch": 0.5706914866710506, "grad_norm": 6.0446929931640625, "learning_rate": 3.96533570085342e-06, "loss": 0.3688, "step": 11282 }, { "epoch": 0.5707420709191158, "grad_norm": 19.678913116455078, "learning_rate": 3.964550472416465e-06, "loss": 0.4185, "step": 11283 }, { "epoch": 0.5707926551671809, "grad_norm": 17.823848724365234, "learning_rate": 3.963765270660545e-06, "loss": 0.3866, "step": 11284 }, { "epoch": 0.5708432394152461, "grad_norm": 7.113875865936279, "learning_rate": 3.962980095605887e-06, "loss": 0.4462, "step": 11285 }, { "epoch": 0.5708938236633112, "grad_norm": 5.663393974304199, "learning_rate": 3.962194947272728e-06, "loss": 0.3613, "step": 11286 }, { "epoch": 0.5709444079113764, "grad_norm": 9.552962303161621, "learning_rate": 3.9614098256813e-06, "loss": 0.433, "step": 11287 }, { "epoch": 0.5709949921594415, "grad_norm": 9.824562072753906, "learning_rate": 3.960624730851832e-06, "loss": 0.4051, "step": 11288 }, { "epoch": 0.5710455764075067, "grad_norm": 14.670913696289062, "learning_rate": 3.959839662804552e-06, "loss": 0.7223, "step": 11289 }, { "epoch": 0.5710961606555719, "grad_norm": 6.4916582107543945, "learning_rate": 3.9590546215596935e-06, "loss": 0.392, "step": 11290 }, { "epoch": 0.571146744903637, "grad_norm": 5.137347221374512, "learning_rate": 3.958269607137483e-06, "loss": 0.3552, "step": 11291 }, { "epoch": 0.5711973291517022, "grad_norm": 6.113890171051025, "learning_rate": 3.957484619558147e-06, "loss": 0.4141, "step": 11292 }, { "epoch": 0.5712479133997673, "grad_norm": 10.48730754852295, "learning_rate": 3.956699658841916e-06, "loss": 0.4245, "step": 11293 }, { "epoch": 0.5712984976478325, "grad_norm": 7.807571887969971, "learning_rate": 3.955914725009014e-06, "loss": 0.4397, "step": 11294 }, { "epoch": 0.5713490818958976, "grad_norm": 7.431417465209961, "learning_rate": 3.955129818079667e-06, "loss": 0.3736, "step": 11295 }, { "epoch": 0.5713996661439628, "grad_norm": 6.420337677001953, "learning_rate": 3.954344938074104e-06, "loss": 0.3938, "step": 11296 }, { "epoch": 0.5714502503920279, "grad_norm": 5.7154316902160645, "learning_rate": 3.953560085012544e-06, "loss": 0.3616, "step": 11297 }, { "epoch": 0.571500834640093, "grad_norm": 9.475991249084473, "learning_rate": 3.952775258915213e-06, "loss": 0.3856, "step": 11298 }, { "epoch": 0.5715514188881582, "grad_norm": 7.004607677459717, "learning_rate": 3.951990459802333e-06, "loss": 0.4643, "step": 11299 }, { "epoch": 0.5716020031362233, "grad_norm": 5.94625186920166, "learning_rate": 3.9512056876941295e-06, "loss": 0.408, "step": 11300 }, { "epoch": 0.5716525873842885, "grad_norm": 4.7182183265686035, "learning_rate": 3.950420942610821e-06, "loss": 0.232, "step": 11301 }, { "epoch": 0.5717031716323537, "grad_norm": 13.879292488098145, "learning_rate": 3.949636224572629e-06, "loss": 0.6243, "step": 11302 }, { "epoch": 0.5717537558804189, "grad_norm": 7.257801532745361, "learning_rate": 3.948851533599776e-06, "loss": 0.3498, "step": 11303 }, { "epoch": 0.571804340128484, "grad_norm": 6.7901611328125, "learning_rate": 3.948066869712479e-06, "loss": 0.3665, "step": 11304 }, { "epoch": 0.5718549243765492, "grad_norm": 8.100902557373047, "learning_rate": 3.947282232930958e-06, "loss": 0.3771, "step": 11305 }, { "epoch": 0.5719055086246143, "grad_norm": 6.367077350616455, "learning_rate": 3.946497623275432e-06, "loss": 0.366, "step": 11306 }, { "epoch": 0.5719560928726795, "grad_norm": 9.368453025817871, "learning_rate": 3.945713040766117e-06, "loss": 0.3306, "step": 11307 }, { "epoch": 0.5720066771207446, "grad_norm": 6.1091203689575195, "learning_rate": 3.944928485423229e-06, "loss": 0.3504, "step": 11308 }, { "epoch": 0.5720572613688097, "grad_norm": 9.22666072845459, "learning_rate": 3.944143957266989e-06, "loss": 0.4774, "step": 11309 }, { "epoch": 0.5721078456168749, "grad_norm": 13.222146034240723, "learning_rate": 3.943359456317607e-06, "loss": 0.4652, "step": 11310 }, { "epoch": 0.57215842986494, "grad_norm": 5.053945064544678, "learning_rate": 3.9425749825953e-06, "loss": 0.349, "step": 11311 }, { "epoch": 0.5722090141130052, "grad_norm": 9.044971466064453, "learning_rate": 3.94179053612028e-06, "loss": 0.4383, "step": 11312 }, { "epoch": 0.5722595983610703, "grad_norm": 7.753555774688721, "learning_rate": 3.941006116912764e-06, "loss": 0.4214, "step": 11313 }, { "epoch": 0.5723101826091355, "grad_norm": 9.979994773864746, "learning_rate": 3.940221724992961e-06, "loss": 0.3838, "step": 11314 }, { "epoch": 0.5723607668572007, "grad_norm": 11.938774108886719, "learning_rate": 3.939437360381085e-06, "loss": 0.3808, "step": 11315 }, { "epoch": 0.5724113511052659, "grad_norm": 10.218817710876465, "learning_rate": 3.9386530230973466e-06, "loss": 0.5258, "step": 11316 }, { "epoch": 0.572461935353331, "grad_norm": 11.334566116333008, "learning_rate": 3.937868713161955e-06, "loss": 0.4354, "step": 11317 }, { "epoch": 0.5725125196013962, "grad_norm": 6.095378875732422, "learning_rate": 3.937084430595122e-06, "loss": 0.4187, "step": 11318 }, { "epoch": 0.5725631038494613, "grad_norm": 11.037805557250977, "learning_rate": 3.936300175417056e-06, "loss": 0.4628, "step": 11319 }, { "epoch": 0.5726136880975264, "grad_norm": 12.876543998718262, "learning_rate": 3.935515947647965e-06, "loss": 0.4831, "step": 11320 }, { "epoch": 0.5726642723455916, "grad_norm": 8.631030082702637, "learning_rate": 3.934731747308057e-06, "loss": 0.4168, "step": 11321 }, { "epoch": 0.5727148565936567, "grad_norm": 8.566823959350586, "learning_rate": 3.933947574417541e-06, "loss": 0.4117, "step": 11322 }, { "epoch": 0.5727654408417219, "grad_norm": 12.909309387207031, "learning_rate": 3.933163428996619e-06, "loss": 0.4997, "step": 11323 }, { "epoch": 0.572816025089787, "grad_norm": 7.686760425567627, "learning_rate": 3.9323793110654975e-06, "loss": 0.4224, "step": 11324 }, { "epoch": 0.5728666093378522, "grad_norm": 9.570411682128906, "learning_rate": 3.931595220644384e-06, "loss": 0.3489, "step": 11325 }, { "epoch": 0.5729171935859173, "grad_norm": 9.904552459716797, "learning_rate": 3.93081115775348e-06, "loss": 0.4648, "step": 11326 }, { "epoch": 0.5729677778339825, "grad_norm": 6.936160564422607, "learning_rate": 3.930027122412991e-06, "loss": 0.4006, "step": 11327 }, { "epoch": 0.5730183620820477, "grad_norm": 7.442445755004883, "learning_rate": 3.929243114643118e-06, "loss": 0.3496, "step": 11328 }, { "epoch": 0.5730689463301128, "grad_norm": 6.4952168464660645, "learning_rate": 3.928459134464063e-06, "loss": 0.4322, "step": 11329 }, { "epoch": 0.573119530578178, "grad_norm": 9.993056297302246, "learning_rate": 3.927675181896029e-06, "loss": 0.4304, "step": 11330 }, { "epoch": 0.5731701148262431, "grad_norm": 9.56296443939209, "learning_rate": 3.926891256959215e-06, "loss": 0.5678, "step": 11331 }, { "epoch": 0.5732206990743083, "grad_norm": 4.30894660949707, "learning_rate": 3.926107359673821e-06, "loss": 0.4251, "step": 11332 }, { "epoch": 0.5732712833223734, "grad_norm": 7.307177543640137, "learning_rate": 3.925323490060047e-06, "loss": 0.4801, "step": 11333 }, { "epoch": 0.5733218675704386, "grad_norm": 6.205123424530029, "learning_rate": 3.924539648138091e-06, "loss": 0.3524, "step": 11334 }, { "epoch": 0.5733724518185037, "grad_norm": 7.001052379608154, "learning_rate": 3.923755833928151e-06, "loss": 0.5285, "step": 11335 }, { "epoch": 0.5734230360665689, "grad_norm": 6.128387928009033, "learning_rate": 3.922972047450424e-06, "loss": 0.4166, "step": 11336 }, { "epoch": 0.573473620314634, "grad_norm": 9.647215843200684, "learning_rate": 3.922188288725104e-06, "loss": 0.3931, "step": 11337 }, { "epoch": 0.5735242045626991, "grad_norm": 11.479736328125, "learning_rate": 3.92140455777239e-06, "loss": 0.5256, "step": 11338 }, { "epoch": 0.5735747888107643, "grad_norm": 8.362702369689941, "learning_rate": 3.920620854612473e-06, "loss": 0.4154, "step": 11339 }, { "epoch": 0.5736253730588294, "grad_norm": 7.8899078369140625, "learning_rate": 3.919837179265551e-06, "loss": 0.377, "step": 11340 }, { "epoch": 0.5736759573068946, "grad_norm": 10.248239517211914, "learning_rate": 3.919053531751817e-06, "loss": 0.4039, "step": 11341 }, { "epoch": 0.5737265415549598, "grad_norm": 6.868575096130371, "learning_rate": 3.9182699120914604e-06, "loss": 0.3968, "step": 11342 }, { "epoch": 0.573777125803025, "grad_norm": 6.749221324920654, "learning_rate": 3.917486320304677e-06, "loss": 0.4447, "step": 11343 }, { "epoch": 0.5738277100510901, "grad_norm": 8.115427017211914, "learning_rate": 3.916702756411656e-06, "loss": 0.3747, "step": 11344 }, { "epoch": 0.5738782942991553, "grad_norm": 4.727741718292236, "learning_rate": 3.915919220432587e-06, "loss": 0.4154, "step": 11345 }, { "epoch": 0.5739288785472204, "grad_norm": 7.734196662902832, "learning_rate": 3.915135712387664e-06, "loss": 0.4881, "step": 11346 }, { "epoch": 0.5739794627952856, "grad_norm": 7.863702297210693, "learning_rate": 3.91435223229707e-06, "loss": 0.4041, "step": 11347 }, { "epoch": 0.5740300470433507, "grad_norm": 8.41954517364502, "learning_rate": 3.913568780181001e-06, "loss": 0.485, "step": 11348 }, { "epoch": 0.5740806312914158, "grad_norm": 5.928646087646484, "learning_rate": 3.912785356059638e-06, "loss": 0.2448, "step": 11349 }, { "epoch": 0.574131215539481, "grad_norm": 5.013317108154297, "learning_rate": 3.912001959953169e-06, "loss": 0.3213, "step": 11350 }, { "epoch": 0.5741817997875461, "grad_norm": 6.73729944229126, "learning_rate": 3.9112185918817835e-06, "loss": 0.4046, "step": 11351 }, { "epoch": 0.5742323840356113, "grad_norm": 14.59788703918457, "learning_rate": 3.910435251865663e-06, "loss": 0.5908, "step": 11352 }, { "epoch": 0.5742829682836764, "grad_norm": 8.960172653198242, "learning_rate": 3.909651939924997e-06, "loss": 0.3626, "step": 11353 }, { "epoch": 0.5743335525317416, "grad_norm": 6.049250602722168, "learning_rate": 3.908868656079965e-06, "loss": 0.4734, "step": 11354 }, { "epoch": 0.5743841367798068, "grad_norm": 9.572025299072266, "learning_rate": 3.9080854003507516e-06, "loss": 0.5661, "step": 11355 }, { "epoch": 0.574434721027872, "grad_norm": 6.634969234466553, "learning_rate": 3.907302172757542e-06, "loss": 0.4184, "step": 11356 }, { "epoch": 0.5744853052759371, "grad_norm": 8.907075881958008, "learning_rate": 3.906518973320515e-06, "loss": 0.4425, "step": 11357 }, { "epoch": 0.5745358895240023, "grad_norm": 4.490241050720215, "learning_rate": 3.905735802059852e-06, "loss": 0.3685, "step": 11358 }, { "epoch": 0.5745864737720674, "grad_norm": 11.78683853149414, "learning_rate": 3.904952658995736e-06, "loss": 0.6121, "step": 11359 }, { "epoch": 0.5746370580201325, "grad_norm": 7.2141947746276855, "learning_rate": 3.904169544148344e-06, "loss": 0.3806, "step": 11360 }, { "epoch": 0.5746876422681977, "grad_norm": 13.019725799560547, "learning_rate": 3.9033864575378575e-06, "loss": 0.4936, "step": 11361 }, { "epoch": 0.5747382265162628, "grad_norm": 5.865396499633789, "learning_rate": 3.9026033991844514e-06, "loss": 0.3814, "step": 11362 }, { "epoch": 0.574788810764328, "grad_norm": 9.028019905090332, "learning_rate": 3.901820369108307e-06, "loss": 0.4139, "step": 11363 }, { "epoch": 0.5748393950123931, "grad_norm": 6.161552429199219, "learning_rate": 3.901037367329597e-06, "loss": 0.3936, "step": 11364 }, { "epoch": 0.5748899792604583, "grad_norm": 13.006813049316406, "learning_rate": 3.9002543938685e-06, "loss": 0.5153, "step": 11365 }, { "epoch": 0.5749405635085234, "grad_norm": 5.267831325531006, "learning_rate": 3.899471448745192e-06, "loss": 0.3025, "step": 11366 }, { "epoch": 0.5749911477565885, "grad_norm": 7.211945533752441, "learning_rate": 3.898688531979845e-06, "loss": 0.3376, "step": 11367 }, { "epoch": 0.5750417320046538, "grad_norm": 5.881272315979004, "learning_rate": 3.897905643592635e-06, "loss": 0.2471, "step": 11368 }, { "epoch": 0.575092316252719, "grad_norm": 6.604671001434326, "learning_rate": 3.897122783603735e-06, "loss": 0.4291, "step": 11369 }, { "epoch": 0.5751429005007841, "grad_norm": 9.072416305541992, "learning_rate": 3.896339952033317e-06, "loss": 0.401, "step": 11370 }, { "epoch": 0.5751934847488492, "grad_norm": 8.97005844116211, "learning_rate": 3.895557148901552e-06, "loss": 0.3874, "step": 11371 }, { "epoch": 0.5752440689969144, "grad_norm": 8.511549949645996, "learning_rate": 3.8947743742286125e-06, "loss": 0.4491, "step": 11372 }, { "epoch": 0.5752946532449795, "grad_norm": 10.386183738708496, "learning_rate": 3.8939916280346665e-06, "loss": 0.2986, "step": 11373 }, { "epoch": 0.5753452374930447, "grad_norm": 10.03888988494873, "learning_rate": 3.893208910339888e-06, "loss": 0.5183, "step": 11374 }, { "epoch": 0.5753958217411098, "grad_norm": 5.622151851654053, "learning_rate": 3.89242622116444e-06, "loss": 0.3793, "step": 11375 }, { "epoch": 0.575446405989175, "grad_norm": 4.799605369567871, "learning_rate": 3.891643560528493e-06, "loss": 0.4782, "step": 11376 }, { "epoch": 0.5754969902372401, "grad_norm": 8.683205604553223, "learning_rate": 3.890860928452215e-06, "loss": 0.3186, "step": 11377 }, { "epoch": 0.5755475744853052, "grad_norm": 8.654766082763672, "learning_rate": 3.890078324955771e-06, "loss": 0.3938, "step": 11378 }, { "epoch": 0.5755981587333704, "grad_norm": 10.045190811157227, "learning_rate": 3.889295750059329e-06, "loss": 0.5389, "step": 11379 }, { "epoch": 0.5756487429814355, "grad_norm": 15.420124053955078, "learning_rate": 3.8885132037830505e-06, "loss": 0.4922, "step": 11380 }, { "epoch": 0.5756993272295007, "grad_norm": 7.423239231109619, "learning_rate": 3.887730686147103e-06, "loss": 0.3986, "step": 11381 }, { "epoch": 0.5757499114775659, "grad_norm": 10.094261169433594, "learning_rate": 3.886948197171651e-06, "loss": 0.4058, "step": 11382 }, { "epoch": 0.5758004957256311, "grad_norm": 8.987061500549316, "learning_rate": 3.886165736876853e-06, "loss": 0.4028, "step": 11383 }, { "epoch": 0.5758510799736962, "grad_norm": 7.784372329711914, "learning_rate": 3.885383305282875e-06, "loss": 0.4714, "step": 11384 }, { "epoch": 0.5759016642217614, "grad_norm": 5.52001428604126, "learning_rate": 3.884600902409876e-06, "loss": 0.3788, "step": 11385 }, { "epoch": 0.5759522484698265, "grad_norm": 5.223046779632568, "learning_rate": 3.8838185282780185e-06, "loss": 0.3651, "step": 11386 }, { "epoch": 0.5760028327178917, "grad_norm": 7.098217487335205, "learning_rate": 3.883036182907463e-06, "loss": 0.4456, "step": 11387 }, { "epoch": 0.5760534169659568, "grad_norm": 5.15714693069458, "learning_rate": 3.882253866318366e-06, "loss": 0.3068, "step": 11388 }, { "epoch": 0.5761040012140219, "grad_norm": 12.960155487060547, "learning_rate": 3.8814715785308856e-06, "loss": 0.4413, "step": 11389 }, { "epoch": 0.5761545854620871, "grad_norm": 7.7057576179504395, "learning_rate": 3.880689319565181e-06, "loss": 0.4493, "step": 11390 }, { "epoch": 0.5762051697101522, "grad_norm": 7.261176109313965, "learning_rate": 3.8799070894414106e-06, "loss": 0.4543, "step": 11391 }, { "epoch": 0.5762557539582174, "grad_norm": 8.109831809997559, "learning_rate": 3.879124888179728e-06, "loss": 0.553, "step": 11392 }, { "epoch": 0.5763063382062825, "grad_norm": 7.835185527801514, "learning_rate": 3.878342715800289e-06, "loss": 0.3414, "step": 11393 }, { "epoch": 0.5763569224543477, "grad_norm": 6.436091423034668, "learning_rate": 3.87756057232325e-06, "loss": 0.487, "step": 11394 }, { "epoch": 0.5764075067024129, "grad_norm": 7.361361503601074, "learning_rate": 3.876778457768764e-06, "loss": 0.382, "step": 11395 }, { "epoch": 0.5764580909504781, "grad_norm": 7.399997711181641, "learning_rate": 3.875996372156984e-06, "loss": 0.3962, "step": 11396 }, { "epoch": 0.5765086751985432, "grad_norm": 4.810691833496094, "learning_rate": 3.875214315508062e-06, "loss": 0.4247, "step": 11397 }, { "epoch": 0.5765592594466084, "grad_norm": 6.115891933441162, "learning_rate": 3.874432287842151e-06, "loss": 0.4666, "step": 11398 }, { "epoch": 0.5766098436946735, "grad_norm": 5.911772727966309, "learning_rate": 3.873650289179403e-06, "loss": 0.3378, "step": 11399 }, { "epoch": 0.5766604279427386, "grad_norm": 9.452428817749023, "learning_rate": 3.872868319539964e-06, "loss": 0.4404, "step": 11400 }, { "epoch": 0.5767110121908038, "grad_norm": 5.968197345733643, "learning_rate": 3.872086378943987e-06, "loss": 0.3617, "step": 11401 }, { "epoch": 0.5767615964388689, "grad_norm": 6.176494598388672, "learning_rate": 3.871304467411618e-06, "loss": 0.4311, "step": 11402 }, { "epoch": 0.5768121806869341, "grad_norm": 5.634274959564209, "learning_rate": 3.870522584963007e-06, "loss": 0.4005, "step": 11403 }, { "epoch": 0.5768627649349992, "grad_norm": 10.061986923217773, "learning_rate": 3.869740731618302e-06, "loss": 0.3623, "step": 11404 }, { "epoch": 0.5769133491830644, "grad_norm": 7.7980523109436035, "learning_rate": 3.868958907397648e-06, "loss": 0.4131, "step": 11405 }, { "epoch": 0.5769639334311295, "grad_norm": 5.093960285186768, "learning_rate": 3.86817711232119e-06, "loss": 0.3076, "step": 11406 }, { "epoch": 0.5770145176791946, "grad_norm": 4.783902168273926, "learning_rate": 3.867395346409076e-06, "loss": 0.3593, "step": 11407 }, { "epoch": 0.5770651019272599, "grad_norm": 6.942966461181641, "learning_rate": 3.8666136096814466e-06, "loss": 0.5239, "step": 11408 }, { "epoch": 0.577115686175325, "grad_norm": 7.307939052581787, "learning_rate": 3.865831902158447e-06, "loss": 0.4908, "step": 11409 }, { "epoch": 0.5771662704233902, "grad_norm": 7.469294548034668, "learning_rate": 3.86505022386022e-06, "loss": 0.3889, "step": 11410 }, { "epoch": 0.5772168546714553, "grad_norm": 7.390053749084473, "learning_rate": 3.8642685748069074e-06, "loss": 0.3359, "step": 11411 }, { "epoch": 0.5772674389195205, "grad_norm": 5.112593650817871, "learning_rate": 3.863486955018653e-06, "loss": 0.4167, "step": 11412 }, { "epoch": 0.5773180231675856, "grad_norm": 7.117455959320068, "learning_rate": 3.862705364515591e-06, "loss": 0.3854, "step": 11413 }, { "epoch": 0.5773686074156508, "grad_norm": 7.399575233459473, "learning_rate": 3.861923803317867e-06, "loss": 0.3501, "step": 11414 }, { "epoch": 0.5774191916637159, "grad_norm": 5.116280555725098, "learning_rate": 3.861142271445616e-06, "loss": 0.3288, "step": 11415 }, { "epoch": 0.5774697759117811, "grad_norm": 7.993057727813721, "learning_rate": 3.8603607689189786e-06, "loss": 0.3931, "step": 11416 }, { "epoch": 0.5775203601598462, "grad_norm": 6.375626087188721, "learning_rate": 3.859579295758091e-06, "loss": 0.3172, "step": 11417 }, { "epoch": 0.5775709444079113, "grad_norm": 6.96371603012085, "learning_rate": 3.85879785198309e-06, "loss": 0.4751, "step": 11418 }, { "epoch": 0.5776215286559765, "grad_norm": 7.314760684967041, "learning_rate": 3.858016437614114e-06, "loss": 0.3698, "step": 11419 }, { "epoch": 0.5776721129040416, "grad_norm": 5.87187385559082, "learning_rate": 3.8572350526712946e-06, "loss": 0.4095, "step": 11420 }, { "epoch": 0.5777226971521068, "grad_norm": 7.140542507171631, "learning_rate": 3.856453697174767e-06, "loss": 0.4181, "step": 11421 }, { "epoch": 0.577773281400172, "grad_norm": 10.111054420471191, "learning_rate": 3.8556723711446665e-06, "loss": 0.416, "step": 11422 }, { "epoch": 0.5778238656482372, "grad_norm": 5.623944282531738, "learning_rate": 3.854891074601125e-06, "loss": 0.3205, "step": 11423 }, { "epoch": 0.5778744498963023, "grad_norm": 6.578541278839111, "learning_rate": 3.854109807564274e-06, "loss": 0.3492, "step": 11424 }, { "epoch": 0.5779250341443675, "grad_norm": 6.631300449371338, "learning_rate": 3.853328570054247e-06, "loss": 0.3848, "step": 11425 }, { "epoch": 0.5779756183924326, "grad_norm": 8.486434936523438, "learning_rate": 3.8525473620911725e-06, "loss": 0.5175, "step": 11426 }, { "epoch": 0.5780262026404978, "grad_norm": 11.641273498535156, "learning_rate": 3.851766183695179e-06, "loss": 0.4184, "step": 11427 }, { "epoch": 0.5780767868885629, "grad_norm": 9.401800155639648, "learning_rate": 3.850985034886399e-06, "loss": 0.3979, "step": 11428 }, { "epoch": 0.578127371136628, "grad_norm": 14.2437744140625, "learning_rate": 3.85020391568496e-06, "loss": 0.388, "step": 11429 }, { "epoch": 0.5781779553846932, "grad_norm": 9.574092864990234, "learning_rate": 3.849422826110987e-06, "loss": 0.3313, "step": 11430 }, { "epoch": 0.5782285396327583, "grad_norm": 14.652392387390137, "learning_rate": 3.84864176618461e-06, "loss": 0.587, "step": 11431 }, { "epoch": 0.5782791238808235, "grad_norm": 21.50137710571289, "learning_rate": 3.847860735925953e-06, "loss": 0.7298, "step": 11432 }, { "epoch": 0.5783297081288886, "grad_norm": 11.875441551208496, "learning_rate": 3.847079735355143e-06, "loss": 0.5491, "step": 11433 }, { "epoch": 0.5783802923769538, "grad_norm": 7.554602146148682, "learning_rate": 3.8462987644923016e-06, "loss": 0.3092, "step": 11434 }, { "epoch": 0.578430876625019, "grad_norm": 10.848341941833496, "learning_rate": 3.845517823357556e-06, "loss": 0.5937, "step": 11435 }, { "epoch": 0.5784814608730842, "grad_norm": 10.901277542114258, "learning_rate": 3.844736911971027e-06, "loss": 0.4753, "step": 11436 }, { "epoch": 0.5785320451211493, "grad_norm": 5.678575038909912, "learning_rate": 3.8439560303528366e-06, "loss": 0.3688, "step": 11437 }, { "epoch": 0.5785826293692145, "grad_norm": 8.000105857849121, "learning_rate": 3.84317517852311e-06, "loss": 0.3411, "step": 11438 }, { "epoch": 0.5786332136172796, "grad_norm": 6.396745681762695, "learning_rate": 3.842394356501964e-06, "loss": 0.4355, "step": 11439 }, { "epoch": 0.5786837978653447, "grad_norm": 5.273806571960449, "learning_rate": 3.841613564309517e-06, "loss": 0.4, "step": 11440 }, { "epoch": 0.5787343821134099, "grad_norm": 9.696516036987305, "learning_rate": 3.840832801965891e-06, "loss": 0.4861, "step": 11441 }, { "epoch": 0.578784966361475, "grad_norm": 4.380107402801514, "learning_rate": 3.840052069491205e-06, "loss": 0.3534, "step": 11442 }, { "epoch": 0.5788355506095402, "grad_norm": 8.290882110595703, "learning_rate": 3.839271366905574e-06, "loss": 0.3705, "step": 11443 }, { "epoch": 0.5788861348576053, "grad_norm": 6.696995735168457, "learning_rate": 3.8384906942291166e-06, "loss": 0.4477, "step": 11444 }, { "epoch": 0.5789367191056705, "grad_norm": 6.10418176651001, "learning_rate": 3.83771005148195e-06, "loss": 0.3726, "step": 11445 }, { "epoch": 0.5789873033537356, "grad_norm": 6.292423248291016, "learning_rate": 3.8369294386841865e-06, "loss": 0.4167, "step": 11446 }, { "epoch": 0.5790378876018007, "grad_norm": 3.466217041015625, "learning_rate": 3.8361488558559415e-06, "loss": 0.258, "step": 11447 }, { "epoch": 0.579088471849866, "grad_norm": 5.607745170593262, "learning_rate": 3.835368303017332e-06, "loss": 0.4078, "step": 11448 }, { "epoch": 0.5791390560979311, "grad_norm": 14.675663948059082, "learning_rate": 3.8345877801884655e-06, "loss": 0.405, "step": 11449 }, { "epoch": 0.5791896403459963, "grad_norm": 6.138299465179443, "learning_rate": 3.8338072873894584e-06, "loss": 0.3987, "step": 11450 }, { "epoch": 0.5792402245940614, "grad_norm": 9.025307655334473, "learning_rate": 3.833026824640422e-06, "loss": 0.3694, "step": 11451 }, { "epoch": 0.5792908088421266, "grad_norm": 11.697590827941895, "learning_rate": 3.8322463919614635e-06, "loss": 0.5266, "step": 11452 }, { "epoch": 0.5793413930901917, "grad_norm": 20.800392150878906, "learning_rate": 3.831465989372695e-06, "loss": 0.4279, "step": 11453 }, { "epoch": 0.5793919773382569, "grad_norm": 8.61080265045166, "learning_rate": 3.830685616894226e-06, "loss": 0.4921, "step": 11454 }, { "epoch": 0.579442561586322, "grad_norm": 8.874542236328125, "learning_rate": 3.829905274546165e-06, "loss": 0.4956, "step": 11455 }, { "epoch": 0.5794931458343872, "grad_norm": 9.180429458618164, "learning_rate": 3.829124962348617e-06, "loss": 0.4095, "step": 11456 }, { "epoch": 0.5795437300824523, "grad_norm": 6.648967266082764, "learning_rate": 3.828344680321693e-06, "loss": 0.3805, "step": 11457 }, { "epoch": 0.5795943143305174, "grad_norm": 9.275166511535645, "learning_rate": 3.8275644284854944e-06, "loss": 0.3932, "step": 11458 }, { "epoch": 0.5796448985785826, "grad_norm": 12.084602355957031, "learning_rate": 3.826784206860128e-06, "loss": 0.3864, "step": 11459 }, { "epoch": 0.5796954828266477, "grad_norm": 8.451821327209473, "learning_rate": 3.8260040154657006e-06, "loss": 0.3143, "step": 11460 }, { "epoch": 0.579746067074713, "grad_norm": 7.546178817749023, "learning_rate": 3.8252238543223136e-06, "loss": 0.3535, "step": 11461 }, { "epoch": 0.5797966513227781, "grad_norm": 5.52573299407959, "learning_rate": 3.82444372345007e-06, "loss": 0.3317, "step": 11462 }, { "epoch": 0.5798472355708433, "grad_norm": 9.462658882141113, "learning_rate": 3.823663622869073e-06, "loss": 0.4596, "step": 11463 }, { "epoch": 0.5798978198189084, "grad_norm": 13.617239952087402, "learning_rate": 3.822883552599424e-06, "loss": 0.4665, "step": 11464 }, { "epoch": 0.5799484040669736, "grad_norm": 6.354464054107666, "learning_rate": 3.822103512661221e-06, "loss": 0.4243, "step": 11465 }, { "epoch": 0.5799989883150387, "grad_norm": 7.436372756958008, "learning_rate": 3.8213235030745656e-06, "loss": 0.3638, "step": 11466 }, { "epoch": 0.5800495725631039, "grad_norm": 4.694099426269531, "learning_rate": 3.820543523859557e-06, "loss": 0.3064, "step": 11467 }, { "epoch": 0.580100156811169, "grad_norm": 6.563292026519775, "learning_rate": 3.819763575036292e-06, "loss": 0.4063, "step": 11468 }, { "epoch": 0.5801507410592341, "grad_norm": 6.091185569763184, "learning_rate": 3.818983656624869e-06, "loss": 0.4236, "step": 11469 }, { "epoch": 0.5802013253072993, "grad_norm": 8.672816276550293, "learning_rate": 3.818203768645385e-06, "loss": 0.4729, "step": 11470 }, { "epoch": 0.5802519095553644, "grad_norm": 10.429067611694336, "learning_rate": 3.817423911117935e-06, "loss": 0.3691, "step": 11471 }, { "epoch": 0.5803024938034296, "grad_norm": 11.824175834655762, "learning_rate": 3.816644084062615e-06, "loss": 0.5796, "step": 11472 }, { "epoch": 0.5803530780514947, "grad_norm": 7.5905938148498535, "learning_rate": 3.815864287499519e-06, "loss": 0.4085, "step": 11473 }, { "epoch": 0.5804036622995599, "grad_norm": 11.521164894104004, "learning_rate": 3.815084521448739e-06, "loss": 0.3895, "step": 11474 }, { "epoch": 0.5804542465476251, "grad_norm": 5.405662536621094, "learning_rate": 3.8143047859303682e-06, "loss": 0.4581, "step": 11475 }, { "epoch": 0.5805048307956903, "grad_norm": 10.474729537963867, "learning_rate": 3.8135250809645015e-06, "loss": 0.4885, "step": 11476 }, { "epoch": 0.5805554150437554, "grad_norm": 17.08453941345215, "learning_rate": 3.812745406571228e-06, "loss": 0.4267, "step": 11477 }, { "epoch": 0.5806059992918206, "grad_norm": 4.910190582275391, "learning_rate": 3.811965762770636e-06, "loss": 0.3964, "step": 11478 }, { "epoch": 0.5806565835398857, "grad_norm": 10.11577320098877, "learning_rate": 3.8111861495828163e-06, "loss": 0.4281, "step": 11479 }, { "epoch": 0.5807071677879508, "grad_norm": 6.329047679901123, "learning_rate": 3.810406567027859e-06, "loss": 0.4523, "step": 11480 }, { "epoch": 0.580757752036016, "grad_norm": 9.699315071105957, "learning_rate": 3.8096270151258506e-06, "loss": 0.4634, "step": 11481 }, { "epoch": 0.5808083362840811, "grad_norm": 5.988887786865234, "learning_rate": 3.8088474938968784e-06, "loss": 0.3184, "step": 11482 }, { "epoch": 0.5808589205321463, "grad_norm": 5.160221576690674, "learning_rate": 3.8080680033610305e-06, "loss": 0.4655, "step": 11483 }, { "epoch": 0.5809095047802114, "grad_norm": 5.389779090881348, "learning_rate": 3.80728854353839e-06, "loss": 0.4086, "step": 11484 }, { "epoch": 0.5809600890282766, "grad_norm": 6.947436809539795, "learning_rate": 3.806509114449043e-06, "loss": 0.4402, "step": 11485 }, { "epoch": 0.5810106732763417, "grad_norm": 8.849099159240723, "learning_rate": 3.805729716113074e-06, "loss": 0.4089, "step": 11486 }, { "epoch": 0.5810612575244068, "grad_norm": 13.785198211669922, "learning_rate": 3.8049503485505645e-06, "loss": 0.6827, "step": 11487 }, { "epoch": 0.5811118417724721, "grad_norm": 13.144610404968262, "learning_rate": 3.8041710117815996e-06, "loss": 0.407, "step": 11488 }, { "epoch": 0.5811624260205372, "grad_norm": 11.354619026184082, "learning_rate": 3.8033917058262588e-06, "loss": 0.4352, "step": 11489 }, { "epoch": 0.5812130102686024, "grad_norm": 6.098421573638916, "learning_rate": 3.8026124307046236e-06, "loss": 0.2857, "step": 11490 }, { "epoch": 0.5812635945166675, "grad_norm": 5.976833820343018, "learning_rate": 3.801833186436773e-06, "loss": 0.308, "step": 11491 }, { "epoch": 0.5813141787647327, "grad_norm": 12.710794448852539, "learning_rate": 3.801053973042788e-06, "loss": 0.4584, "step": 11492 }, { "epoch": 0.5813647630127978, "grad_norm": 16.12112808227539, "learning_rate": 3.800274790542745e-06, "loss": 0.6657, "step": 11493 }, { "epoch": 0.581415347260863, "grad_norm": 9.755173683166504, "learning_rate": 3.7994956389567236e-06, "loss": 0.4269, "step": 11494 }, { "epoch": 0.5814659315089281, "grad_norm": 34.12828063964844, "learning_rate": 3.7987165183048003e-06, "loss": 0.5487, "step": 11495 }, { "epoch": 0.5815165157569933, "grad_norm": 6.997263431549072, "learning_rate": 3.79793742860705e-06, "loss": 0.3822, "step": 11496 }, { "epoch": 0.5815671000050584, "grad_norm": 14.301907539367676, "learning_rate": 3.797158369883548e-06, "loss": 0.4333, "step": 11497 }, { "epoch": 0.5816176842531235, "grad_norm": 8.410120964050293, "learning_rate": 3.796379342154371e-06, "loss": 0.4793, "step": 11498 }, { "epoch": 0.5816682685011887, "grad_norm": 7.487990379333496, "learning_rate": 3.79560034543959e-06, "loss": 0.4416, "step": 11499 }, { "epoch": 0.5817188527492538, "grad_norm": 6.034703254699707, "learning_rate": 3.7948213797592793e-06, "loss": 0.4084, "step": 11500 }, { "epoch": 0.5817694369973191, "grad_norm": 15.026825904846191, "learning_rate": 3.794042445133511e-06, "loss": 0.4546, "step": 11501 }, { "epoch": 0.5818200212453842, "grad_norm": 11.240991592407227, "learning_rate": 3.7932635415823584e-06, "loss": 0.4392, "step": 11502 }, { "epoch": 0.5818706054934494, "grad_norm": 7.277524948120117, "learning_rate": 3.792484669125886e-06, "loss": 0.3941, "step": 11503 }, { "epoch": 0.5819211897415145, "grad_norm": 7.594748497009277, "learning_rate": 3.791705827784168e-06, "loss": 0.4359, "step": 11504 }, { "epoch": 0.5819717739895797, "grad_norm": 7.226965427398682, "learning_rate": 3.790927017577273e-06, "loss": 0.4026, "step": 11505 }, { "epoch": 0.5820223582376448, "grad_norm": 10.154815673828125, "learning_rate": 3.790148238525267e-06, "loss": 0.4288, "step": 11506 }, { "epoch": 0.58207294248571, "grad_norm": 15.32791519165039, "learning_rate": 3.7893694906482192e-06, "loss": 0.4685, "step": 11507 }, { "epoch": 0.5821235267337751, "grad_norm": 5.8077006340026855, "learning_rate": 3.788590773966196e-06, "loss": 0.3486, "step": 11508 }, { "epoch": 0.5821741109818402, "grad_norm": 5.594026565551758, "learning_rate": 3.787812088499262e-06, "loss": 0.4073, "step": 11509 }, { "epoch": 0.5822246952299054, "grad_norm": 4.068395137786865, "learning_rate": 3.787033434267482e-06, "loss": 0.4197, "step": 11510 }, { "epoch": 0.5822752794779705, "grad_norm": 5.829336166381836, "learning_rate": 3.786254811290922e-06, "loss": 0.4827, "step": 11511 }, { "epoch": 0.5823258637260357, "grad_norm": 9.577253341674805, "learning_rate": 3.7854762195896428e-06, "loss": 0.515, "step": 11512 }, { "epoch": 0.5823764479741008, "grad_norm": 9.91736888885498, "learning_rate": 3.7846976591837075e-06, "loss": 0.5767, "step": 11513 }, { "epoch": 0.582427032222166, "grad_norm": 6.325895309448242, "learning_rate": 3.783919130093179e-06, "loss": 0.486, "step": 11514 }, { "epoch": 0.5824776164702312, "grad_norm": 12.38249397277832, "learning_rate": 3.7831406323381193e-06, "loss": 0.5137, "step": 11515 }, { "epoch": 0.5825282007182964, "grad_norm": 9.005964279174805, "learning_rate": 3.7823621659385835e-06, "loss": 0.4063, "step": 11516 }, { "epoch": 0.5825787849663615, "grad_norm": 6.1445159912109375, "learning_rate": 3.781583730914634e-06, "loss": 0.4766, "step": 11517 }, { "epoch": 0.5826293692144267, "grad_norm": 8.657977104187012, "learning_rate": 3.7808053272863287e-06, "loss": 0.4686, "step": 11518 }, { "epoch": 0.5826799534624918, "grad_norm": 5.6264567375183105, "learning_rate": 3.780026955073725e-06, "loss": 0.4015, "step": 11519 }, { "epoch": 0.5827305377105569, "grad_norm": 10.463312149047852, "learning_rate": 3.7792486142968797e-06, "loss": 0.4445, "step": 11520 }, { "epoch": 0.5827811219586221, "grad_norm": 6.030263423919678, "learning_rate": 3.7784703049758507e-06, "loss": 0.3827, "step": 11521 }, { "epoch": 0.5828317062066872, "grad_norm": 9.631002426147461, "learning_rate": 3.77769202713069e-06, "loss": 0.4098, "step": 11522 }, { "epoch": 0.5828822904547524, "grad_norm": 4.707759380340576, "learning_rate": 3.776913780781454e-06, "loss": 0.3372, "step": 11523 }, { "epoch": 0.5829328747028175, "grad_norm": 7.533638000488281, "learning_rate": 3.7761355659481957e-06, "loss": 0.4376, "step": 11524 }, { "epoch": 0.5829834589508827, "grad_norm": 10.313429832458496, "learning_rate": 3.775357382650967e-06, "loss": 0.5364, "step": 11525 }, { "epoch": 0.5830340431989478, "grad_norm": 10.398137092590332, "learning_rate": 3.774579230909822e-06, "loss": 0.4486, "step": 11526 }, { "epoch": 0.583084627447013, "grad_norm": 10.11266803741455, "learning_rate": 3.7738011107448085e-06, "loss": 0.4315, "step": 11527 }, { "epoch": 0.5831352116950782, "grad_norm": 9.507301330566406, "learning_rate": 3.7730230221759813e-06, "loss": 0.3664, "step": 11528 }, { "epoch": 0.5831857959431433, "grad_norm": 7.332582950592041, "learning_rate": 3.772244965223385e-06, "loss": 0.453, "step": 11529 }, { "epoch": 0.5832363801912085, "grad_norm": 6.6014838218688965, "learning_rate": 3.7714669399070713e-06, "loss": 0.3828, "step": 11530 }, { "epoch": 0.5832869644392736, "grad_norm": 7.66663122177124, "learning_rate": 3.7706889462470864e-06, "loss": 0.3461, "step": 11531 }, { "epoch": 0.5833375486873388, "grad_norm": 7.769509792327881, "learning_rate": 3.7699109842634774e-06, "loss": 0.445, "step": 11532 }, { "epoch": 0.5833881329354039, "grad_norm": 5.6146392822265625, "learning_rate": 3.7691330539762927e-06, "loss": 0.3832, "step": 11533 }, { "epoch": 0.5834387171834691, "grad_norm": 6.113768577575684, "learning_rate": 3.7683551554055753e-06, "loss": 0.3958, "step": 11534 }, { "epoch": 0.5834893014315342, "grad_norm": 6.2315497398376465, "learning_rate": 3.7675772885713707e-06, "loss": 0.3838, "step": 11535 }, { "epoch": 0.5835398856795994, "grad_norm": 11.49805736541748, "learning_rate": 3.7667994534937238e-06, "loss": 0.4223, "step": 11536 }, { "epoch": 0.5835904699276645, "grad_norm": 11.158782958984375, "learning_rate": 3.7660216501926746e-06, "loss": 0.5565, "step": 11537 }, { "epoch": 0.5836410541757296, "grad_norm": 10.604385375976562, "learning_rate": 3.765243878688267e-06, "loss": 0.4389, "step": 11538 }, { "epoch": 0.5836916384237948, "grad_norm": 6.175649166107178, "learning_rate": 3.764466139000543e-06, "loss": 0.374, "step": 11539 }, { "epoch": 0.5837422226718599, "grad_norm": 8.2528715133667, "learning_rate": 3.763688431149542e-06, "loss": 0.4658, "step": 11540 }, { "epoch": 0.5837928069199252, "grad_norm": 6.393186092376709, "learning_rate": 3.7629107551553058e-06, "loss": 0.4184, "step": 11541 }, { "epoch": 0.5838433911679903, "grad_norm": 8.405097007751465, "learning_rate": 3.762133111037869e-06, "loss": 0.3677, "step": 11542 }, { "epoch": 0.5838939754160555, "grad_norm": 9.664284706115723, "learning_rate": 3.7613554988172735e-06, "loss": 0.4805, "step": 11543 }, { "epoch": 0.5839445596641206, "grad_norm": 13.354446411132812, "learning_rate": 3.7605779185135537e-06, "loss": 0.5749, "step": 11544 }, { "epoch": 0.5839951439121858, "grad_norm": 7.156918525695801, "learning_rate": 3.7598003701467466e-06, "loss": 0.3754, "step": 11545 }, { "epoch": 0.5840457281602509, "grad_norm": 8.064576148986816, "learning_rate": 3.7590228537368904e-06, "loss": 0.3323, "step": 11546 }, { "epoch": 0.584096312408316, "grad_norm": 5.054018974304199, "learning_rate": 3.7582453693040167e-06, "loss": 0.4476, "step": 11547 }, { "epoch": 0.5841468966563812, "grad_norm": 13.859373092651367, "learning_rate": 3.7574679168681597e-06, "loss": 0.6005, "step": 11548 }, { "epoch": 0.5841974809044463, "grad_norm": 7.962655544281006, "learning_rate": 3.756690496449355e-06, "loss": 0.3949, "step": 11549 }, { "epoch": 0.5842480651525115, "grad_norm": 6.549326419830322, "learning_rate": 3.7559131080676318e-06, "loss": 0.3739, "step": 11550 }, { "epoch": 0.5842986494005766, "grad_norm": 10.404773712158203, "learning_rate": 3.7551357517430233e-06, "loss": 0.524, "step": 11551 }, { "epoch": 0.5843492336486418, "grad_norm": 5.091421604156494, "learning_rate": 3.7543584274955597e-06, "loss": 0.3095, "step": 11552 }, { "epoch": 0.5843998178967069, "grad_norm": 17.693946838378906, "learning_rate": 3.7535811353452708e-06, "loss": 0.5755, "step": 11553 }, { "epoch": 0.5844504021447721, "grad_norm": 6.567553997039795, "learning_rate": 3.7528038753121863e-06, "loss": 0.3521, "step": 11554 }, { "epoch": 0.5845009863928373, "grad_norm": 6.893604755401611, "learning_rate": 3.7520266474163324e-06, "loss": 0.3981, "step": 11555 }, { "epoch": 0.5845515706409025, "grad_norm": 6.607245445251465, "learning_rate": 3.7512494516777383e-06, "loss": 0.4572, "step": 11556 }, { "epoch": 0.5846021548889676, "grad_norm": 10.656587600708008, "learning_rate": 3.750472288116428e-06, "loss": 0.4478, "step": 11557 }, { "epoch": 0.5846527391370328, "grad_norm": 5.653589725494385, "learning_rate": 3.749695156752429e-06, "loss": 0.3595, "step": 11558 }, { "epoch": 0.5847033233850979, "grad_norm": 8.400364875793457, "learning_rate": 3.748918057605767e-06, "loss": 0.4082, "step": 11559 }, { "epoch": 0.584753907633163, "grad_norm": 7.502399921417236, "learning_rate": 3.748140990696464e-06, "loss": 0.3699, "step": 11560 }, { "epoch": 0.5848044918812282, "grad_norm": 9.419837951660156, "learning_rate": 3.747363956044544e-06, "loss": 0.5277, "step": 11561 }, { "epoch": 0.5848550761292933, "grad_norm": 10.867985725402832, "learning_rate": 3.7465869536700296e-06, "loss": 0.5071, "step": 11562 }, { "epoch": 0.5849056603773585, "grad_norm": 7.915312767028809, "learning_rate": 3.7458099835929405e-06, "loss": 0.4686, "step": 11563 }, { "epoch": 0.5849562446254236, "grad_norm": 7.1993536949157715, "learning_rate": 3.7450330458333005e-06, "loss": 0.391, "step": 11564 }, { "epoch": 0.5850068288734888, "grad_norm": 8.289341926574707, "learning_rate": 3.7442561404111265e-06, "loss": 0.4082, "step": 11565 }, { "epoch": 0.5850574131215539, "grad_norm": 10.951737403869629, "learning_rate": 3.743479267346439e-06, "loss": 0.4664, "step": 11566 }, { "epoch": 0.585107997369619, "grad_norm": 8.526019096374512, "learning_rate": 3.7427024266592575e-06, "loss": 0.4419, "step": 11567 }, { "epoch": 0.5851585816176843, "grad_norm": 13.159737586975098, "learning_rate": 3.741925618369597e-06, "loss": 0.4625, "step": 11568 }, { "epoch": 0.5852091658657494, "grad_norm": 7.2219696044921875, "learning_rate": 3.741148842497473e-06, "loss": 0.4862, "step": 11569 }, { "epoch": 0.5852597501138146, "grad_norm": 9.529279708862305, "learning_rate": 3.7403720990629027e-06, "loss": 0.4296, "step": 11570 }, { "epoch": 0.5853103343618797, "grad_norm": 13.601604461669922, "learning_rate": 3.7395953880859026e-06, "loss": 0.3496, "step": 11571 }, { "epoch": 0.5853609186099449, "grad_norm": 5.303366661071777, "learning_rate": 3.738818709586483e-06, "loss": 0.3425, "step": 11572 }, { "epoch": 0.58541150285801, "grad_norm": 6.23898458480835, "learning_rate": 3.7380420635846592e-06, "loss": 0.4323, "step": 11573 }, { "epoch": 0.5854620871060752, "grad_norm": 13.369179725646973, "learning_rate": 3.7372654501004446e-06, "loss": 0.4932, "step": 11574 }, { "epoch": 0.5855126713541403, "grad_norm": 4.983399391174316, "learning_rate": 3.736488869153848e-06, "loss": 0.3324, "step": 11575 }, { "epoch": 0.5855632556022055, "grad_norm": 9.859533309936523, "learning_rate": 3.7357123207648816e-06, "loss": 0.3967, "step": 11576 }, { "epoch": 0.5856138398502706, "grad_norm": 5.780394554138184, "learning_rate": 3.7349358049535565e-06, "loss": 0.3609, "step": 11577 }, { "epoch": 0.5856644240983357, "grad_norm": 6.982817649841309, "learning_rate": 3.7341593217398785e-06, "loss": 0.4186, "step": 11578 }, { "epoch": 0.5857150083464009, "grad_norm": 8.472179412841797, "learning_rate": 3.733382871143857e-06, "loss": 0.3022, "step": 11579 }, { "epoch": 0.585765592594466, "grad_norm": 9.018446922302246, "learning_rate": 3.732606453185502e-06, "loss": 0.3104, "step": 11580 }, { "epoch": 0.5858161768425313, "grad_norm": 8.695148468017578, "learning_rate": 3.7318300678848165e-06, "loss": 0.3968, "step": 11581 }, { "epoch": 0.5858667610905964, "grad_norm": 6.529986381530762, "learning_rate": 3.7310537152618053e-06, "loss": 0.4091, "step": 11582 }, { "epoch": 0.5859173453386616, "grad_norm": 7.9485554695129395, "learning_rate": 3.7302773953364757e-06, "loss": 0.4125, "step": 11583 }, { "epoch": 0.5859679295867267, "grad_norm": 7.792884826660156, "learning_rate": 3.7295011081288314e-06, "loss": 0.3901, "step": 11584 }, { "epoch": 0.5860185138347919, "grad_norm": 7.73048734664917, "learning_rate": 3.7287248536588733e-06, "loss": 0.4359, "step": 11585 }, { "epoch": 0.586069098082857, "grad_norm": 7.633605003356934, "learning_rate": 3.7279486319466053e-06, "loss": 0.36, "step": 11586 }, { "epoch": 0.5861196823309222, "grad_norm": 6.628325462341309, "learning_rate": 3.7271724430120297e-06, "loss": 0.306, "step": 11587 }, { "epoch": 0.5861702665789873, "grad_norm": 8.788684844970703, "learning_rate": 3.7263962868751446e-06, "loss": 0.4881, "step": 11588 }, { "epoch": 0.5862208508270524, "grad_norm": 9.730466842651367, "learning_rate": 3.7256201635559517e-06, "loss": 0.4412, "step": 11589 }, { "epoch": 0.5862714350751176, "grad_norm": 12.775009155273438, "learning_rate": 3.7248440730744495e-06, "loss": 0.4481, "step": 11590 }, { "epoch": 0.5863220193231827, "grad_norm": 5.211648464202881, "learning_rate": 3.724068015450635e-06, "loss": 0.3626, "step": 11591 }, { "epoch": 0.5863726035712479, "grad_norm": 14.689495086669922, "learning_rate": 3.7232919907045073e-06, "loss": 0.5895, "step": 11592 }, { "epoch": 0.586423187819313, "grad_norm": 10.00430965423584, "learning_rate": 3.7225159988560587e-06, "loss": 0.4239, "step": 11593 }, { "epoch": 0.5864737720673783, "grad_norm": 9.414993286132812, "learning_rate": 3.721740039925289e-06, "loss": 0.3673, "step": 11594 }, { "epoch": 0.5865243563154434, "grad_norm": 8.250603675842285, "learning_rate": 3.72096411393219e-06, "loss": 0.4272, "step": 11595 }, { "epoch": 0.5865749405635086, "grad_norm": 8.76484203338623, "learning_rate": 3.720188220896757e-06, "loss": 0.3571, "step": 11596 }, { "epoch": 0.5866255248115737, "grad_norm": 9.49141788482666, "learning_rate": 3.719412360838982e-06, "loss": 0.3802, "step": 11597 }, { "epoch": 0.5866761090596389, "grad_norm": 6.173051357269287, "learning_rate": 3.7186365337788567e-06, "loss": 0.3785, "step": 11598 }, { "epoch": 0.586726693307704, "grad_norm": 5.299971580505371, "learning_rate": 3.717860739736374e-06, "loss": 0.4434, "step": 11599 }, { "epoch": 0.5867772775557691, "grad_norm": 10.834259033203125, "learning_rate": 3.717084978731522e-06, "loss": 0.5503, "step": 11600 }, { "epoch": 0.5868278618038343, "grad_norm": 6.98556661605835, "learning_rate": 3.7163092507842913e-06, "loss": 0.4494, "step": 11601 }, { "epoch": 0.5868784460518994, "grad_norm": 13.885543823242188, "learning_rate": 3.715533555914672e-06, "loss": 0.5387, "step": 11602 }, { "epoch": 0.5869290302999646, "grad_norm": 11.443456649780273, "learning_rate": 3.714757894142649e-06, "loss": 0.4543, "step": 11603 }, { "epoch": 0.5869796145480297, "grad_norm": 11.213051795959473, "learning_rate": 3.7139822654882114e-06, "loss": 0.4991, "step": 11604 }, { "epoch": 0.5870301987960949, "grad_norm": 7.003377437591553, "learning_rate": 3.713206669971346e-06, "loss": 0.377, "step": 11605 }, { "epoch": 0.58708078304416, "grad_norm": 6.9681501388549805, "learning_rate": 3.7124311076120356e-06, "loss": 0.3721, "step": 11606 }, { "epoch": 0.5871313672922251, "grad_norm": 6.647486209869385, "learning_rate": 3.7116555784302648e-06, "loss": 0.305, "step": 11607 }, { "epoch": 0.5871819515402904, "grad_norm": 6.81417989730835, "learning_rate": 3.7108800824460178e-06, "loss": 0.4419, "step": 11608 }, { "epoch": 0.5872325357883555, "grad_norm": 10.540947914123535, "learning_rate": 3.7101046196792783e-06, "loss": 0.3822, "step": 11609 }, { "epoch": 0.5872831200364207, "grad_norm": 7.17453145980835, "learning_rate": 3.709329190150026e-06, "loss": 0.3756, "step": 11610 }, { "epoch": 0.5873337042844858, "grad_norm": 7.02805233001709, "learning_rate": 3.7085537938782435e-06, "loss": 0.4339, "step": 11611 }, { "epoch": 0.587384288532551, "grad_norm": 7.6664652824401855, "learning_rate": 3.7077784308839104e-06, "loss": 0.3982, "step": 11612 }, { "epoch": 0.5874348727806161, "grad_norm": 10.490535736083984, "learning_rate": 3.7070031011870058e-06, "loss": 0.4498, "step": 11613 }, { "epoch": 0.5874854570286813, "grad_norm": 7.947263240814209, "learning_rate": 3.7062278048075077e-06, "loss": 0.2909, "step": 11614 }, { "epoch": 0.5875360412767464, "grad_norm": 8.531027793884277, "learning_rate": 3.705452541765395e-06, "loss": 0.404, "step": 11615 }, { "epoch": 0.5875866255248116, "grad_norm": 8.058480262756348, "learning_rate": 3.704677312080643e-06, "loss": 0.3687, "step": 11616 }, { "epoch": 0.5876372097728767, "grad_norm": 10.605055809020996, "learning_rate": 3.703902115773228e-06, "loss": 0.4715, "step": 11617 }, { "epoch": 0.5876877940209418, "grad_norm": 6.345395565032959, "learning_rate": 3.7031269528631263e-06, "loss": 0.3873, "step": 11618 }, { "epoch": 0.587738378269007, "grad_norm": 7.7904052734375, "learning_rate": 3.70235182337031e-06, "loss": 0.4327, "step": 11619 }, { "epoch": 0.5877889625170721, "grad_norm": 13.882904052734375, "learning_rate": 3.7015767273147517e-06, "loss": 0.4028, "step": 11620 }, { "epoch": 0.5878395467651374, "grad_norm": 7.752702236175537, "learning_rate": 3.7008016647164246e-06, "loss": 0.4364, "step": 11621 }, { "epoch": 0.5878901310132025, "grad_norm": 10.84182357788086, "learning_rate": 3.700026635595302e-06, "loss": 0.5009, "step": 11622 }, { "epoch": 0.5879407152612677, "grad_norm": 8.023730278015137, "learning_rate": 3.699251639971352e-06, "loss": 0.4529, "step": 11623 }, { "epoch": 0.5879912995093328, "grad_norm": 8.3284330368042, "learning_rate": 3.6984766778645447e-06, "loss": 0.4611, "step": 11624 }, { "epoch": 0.588041883757398, "grad_norm": 6.9411468505859375, "learning_rate": 3.6977017492948514e-06, "loss": 0.3991, "step": 11625 }, { "epoch": 0.5880924680054631, "grad_norm": 8.65446662902832, "learning_rate": 3.6969268542822372e-06, "loss": 0.4246, "step": 11626 }, { "epoch": 0.5881430522535283, "grad_norm": 8.877911567687988, "learning_rate": 3.6961519928466715e-06, "loss": 0.5585, "step": 11627 }, { "epoch": 0.5881936365015934, "grad_norm": 13.852989196777344, "learning_rate": 3.6953771650081195e-06, "loss": 0.4261, "step": 11628 }, { "epoch": 0.5882442207496585, "grad_norm": 5.606034755706787, "learning_rate": 3.694602370786545e-06, "loss": 0.3624, "step": 11629 }, { "epoch": 0.5882948049977237, "grad_norm": 8.75792121887207, "learning_rate": 3.6938276102019167e-06, "loss": 0.4356, "step": 11630 }, { "epoch": 0.5883453892457888, "grad_norm": 10.500347137451172, "learning_rate": 3.693052883274196e-06, "loss": 0.4098, "step": 11631 }, { "epoch": 0.588395973493854, "grad_norm": 5.73142147064209, "learning_rate": 3.692278190023345e-06, "loss": 0.4286, "step": 11632 }, { "epoch": 0.5884465577419191, "grad_norm": 11.43891716003418, "learning_rate": 3.6915035304693257e-06, "loss": 0.4275, "step": 11633 }, { "epoch": 0.5884971419899844, "grad_norm": 6.564849853515625, "learning_rate": 3.6907289046321016e-06, "loss": 0.3519, "step": 11634 }, { "epoch": 0.5885477262380495, "grad_norm": 6.72480583190918, "learning_rate": 3.689954312531629e-06, "loss": 0.4209, "step": 11635 }, { "epoch": 0.5885983104861147, "grad_norm": 5.024916172027588, "learning_rate": 3.68917975418787e-06, "loss": 0.4022, "step": 11636 }, { "epoch": 0.5886488947341798, "grad_norm": 11.996159553527832, "learning_rate": 3.688405229620784e-06, "loss": 0.4662, "step": 11637 }, { "epoch": 0.588699478982245, "grad_norm": 6.13362455368042, "learning_rate": 3.6876307388503253e-06, "loss": 0.4418, "step": 11638 }, { "epoch": 0.5887500632303101, "grad_norm": 7.4989519119262695, "learning_rate": 3.6868562818964526e-06, "loss": 0.4847, "step": 11639 }, { "epoch": 0.5888006474783752, "grad_norm": 7.347123146057129, "learning_rate": 3.6860818587791236e-06, "loss": 0.4244, "step": 11640 }, { "epoch": 0.5888512317264404, "grad_norm": 9.27435302734375, "learning_rate": 3.68530746951829e-06, "loss": 0.4147, "step": 11641 }, { "epoch": 0.5889018159745055, "grad_norm": 8.41747760772705, "learning_rate": 3.684533114133907e-06, "loss": 0.4215, "step": 11642 }, { "epoch": 0.5889524002225707, "grad_norm": 6.189330577850342, "learning_rate": 3.68375879264593e-06, "loss": 0.3977, "step": 11643 }, { "epoch": 0.5890029844706358, "grad_norm": 4.93019437789917, "learning_rate": 3.6829845050743108e-06, "loss": 0.3755, "step": 11644 }, { "epoch": 0.589053568718701, "grad_norm": 7.24228048324585, "learning_rate": 3.682210251438997e-06, "loss": 0.4908, "step": 11645 }, { "epoch": 0.5891041529667661, "grad_norm": 10.361708641052246, "learning_rate": 3.6814360317599423e-06, "loss": 0.4163, "step": 11646 }, { "epoch": 0.5891547372148312, "grad_norm": 6.58929443359375, "learning_rate": 3.6806618460570977e-06, "loss": 0.369, "step": 11647 }, { "epoch": 0.5892053214628965, "grad_norm": 7.053075313568115, "learning_rate": 3.6798876943504092e-06, "loss": 0.4474, "step": 11648 }, { "epoch": 0.5892559057109616, "grad_norm": 5.618311405181885, "learning_rate": 3.6791135766598257e-06, "loss": 0.281, "step": 11649 }, { "epoch": 0.5893064899590268, "grad_norm": 21.02972412109375, "learning_rate": 3.678339493005296e-06, "loss": 0.4128, "step": 11650 }, { "epoch": 0.5893570742070919, "grad_norm": 7.121920585632324, "learning_rate": 3.677565443406765e-06, "loss": 0.4138, "step": 11651 }, { "epoch": 0.5894076584551571, "grad_norm": 6.053375720977783, "learning_rate": 3.6767914278841775e-06, "loss": 0.3784, "step": 11652 }, { "epoch": 0.5894582427032222, "grad_norm": 11.291946411132812, "learning_rate": 3.6760174464574804e-06, "loss": 0.4155, "step": 11653 }, { "epoch": 0.5895088269512874, "grad_norm": 8.090156555175781, "learning_rate": 3.6752434991466145e-06, "loss": 0.4136, "step": 11654 }, { "epoch": 0.5895594111993525, "grad_norm": 7.190216541290283, "learning_rate": 3.674469585971524e-06, "loss": 0.4075, "step": 11655 }, { "epoch": 0.5896099954474177, "grad_norm": 6.805790424346924, "learning_rate": 3.673695706952152e-06, "loss": 0.4162, "step": 11656 }, { "epoch": 0.5896605796954828, "grad_norm": 7.493836879730225, "learning_rate": 3.6729218621084383e-06, "loss": 0.3388, "step": 11657 }, { "epoch": 0.589711163943548, "grad_norm": 7.40582799911499, "learning_rate": 3.6721480514603208e-06, "loss": 0.4311, "step": 11658 }, { "epoch": 0.5897617481916131, "grad_norm": 7.897820472717285, "learning_rate": 3.671374275027741e-06, "loss": 0.4309, "step": 11659 }, { "epoch": 0.5898123324396782, "grad_norm": 4.362250804901123, "learning_rate": 3.6706005328306376e-06, "loss": 0.2353, "step": 11660 }, { "epoch": 0.5898629166877435, "grad_norm": 9.245105743408203, "learning_rate": 3.669826824888947e-06, "loss": 0.3965, "step": 11661 }, { "epoch": 0.5899135009358086, "grad_norm": 5.247136116027832, "learning_rate": 3.669053151222606e-06, "loss": 0.3731, "step": 11662 }, { "epoch": 0.5899640851838738, "grad_norm": 10.005743026733398, "learning_rate": 3.6682795118515514e-06, "loss": 0.4717, "step": 11663 }, { "epoch": 0.5900146694319389, "grad_norm": 4.714787006378174, "learning_rate": 3.6675059067957164e-06, "loss": 0.4208, "step": 11664 }, { "epoch": 0.5900652536800041, "grad_norm": 9.291568756103516, "learning_rate": 3.666732336075036e-06, "loss": 0.4586, "step": 11665 }, { "epoch": 0.5901158379280692, "grad_norm": 7.557809829711914, "learning_rate": 3.6659587997094425e-06, "loss": 0.4946, "step": 11666 }, { "epoch": 0.5901664221761344, "grad_norm": 6.565572738647461, "learning_rate": 3.6651852977188684e-06, "loss": 0.3733, "step": 11667 }, { "epoch": 0.5902170064241995, "grad_norm": 8.675068855285645, "learning_rate": 3.664411830123246e-06, "loss": 0.3492, "step": 11668 }, { "epoch": 0.5902675906722646, "grad_norm": 7.507826805114746, "learning_rate": 3.6636383969425036e-06, "loss": 0.3503, "step": 11669 }, { "epoch": 0.5903181749203298, "grad_norm": 7.84519100189209, "learning_rate": 3.6628649981965747e-06, "loss": 0.447, "step": 11670 }, { "epoch": 0.5903687591683949, "grad_norm": 24.080320358276367, "learning_rate": 3.6620916339053826e-06, "loss": 0.6174, "step": 11671 }, { "epoch": 0.5904193434164601, "grad_norm": 5.111934661865234, "learning_rate": 3.661318304088858e-06, "loss": 0.4191, "step": 11672 }, { "epoch": 0.5904699276645252, "grad_norm": 6.941810607910156, "learning_rate": 3.660545008766927e-06, "loss": 0.3862, "step": 11673 }, { "epoch": 0.5905205119125905, "grad_norm": 8.290700912475586, "learning_rate": 3.6597717479595153e-06, "loss": 0.4171, "step": 11674 }, { "epoch": 0.5905710961606556, "grad_norm": 11.87760066986084, "learning_rate": 3.658998521686549e-06, "loss": 0.5083, "step": 11675 }, { "epoch": 0.5906216804087208, "grad_norm": 12.587823867797852, "learning_rate": 3.6582253299679514e-06, "loss": 0.5186, "step": 11676 }, { "epoch": 0.5906722646567859, "grad_norm": 8.897711753845215, "learning_rate": 3.6574521728236462e-06, "loss": 0.3865, "step": 11677 }, { "epoch": 0.590722848904851, "grad_norm": 6.7963433265686035, "learning_rate": 3.6566790502735566e-06, "loss": 0.3993, "step": 11678 }, { "epoch": 0.5907734331529162, "grad_norm": 6.760261058807373, "learning_rate": 3.6559059623376026e-06, "loss": 0.3788, "step": 11679 }, { "epoch": 0.5908240174009813, "grad_norm": 12.88317584991455, "learning_rate": 3.6551329090357048e-06, "loss": 0.5098, "step": 11680 }, { "epoch": 0.5908746016490465, "grad_norm": 9.41011905670166, "learning_rate": 3.654359890387784e-06, "loss": 0.3515, "step": 11681 }, { "epoch": 0.5909251858971116, "grad_norm": 6.751044750213623, "learning_rate": 3.653586906413761e-06, "loss": 0.4252, "step": 11682 }, { "epoch": 0.5909757701451768, "grad_norm": 6.847586631774902, "learning_rate": 3.6528139571335484e-06, "loss": 0.2912, "step": 11683 }, { "epoch": 0.5910263543932419, "grad_norm": 5.564126968383789, "learning_rate": 3.652041042567066e-06, "loss": 0.3056, "step": 11684 }, { "epoch": 0.5910769386413071, "grad_norm": 8.948784828186035, "learning_rate": 3.651268162734232e-06, "loss": 0.4315, "step": 11685 }, { "epoch": 0.5911275228893722, "grad_norm": 6.461855411529541, "learning_rate": 3.650495317654958e-06, "loss": 0.2797, "step": 11686 }, { "epoch": 0.5911781071374373, "grad_norm": 4.939322471618652, "learning_rate": 3.6497225073491605e-06, "loss": 0.4184, "step": 11687 }, { "epoch": 0.5912286913855026, "grad_norm": 6.104533672332764, "learning_rate": 3.648949731836753e-06, "loss": 0.374, "step": 11688 }, { "epoch": 0.5912792756335677, "grad_norm": 5.766260623931885, "learning_rate": 3.6481769911376464e-06, "loss": 0.4329, "step": 11689 }, { "epoch": 0.5913298598816329, "grad_norm": 10.658833503723145, "learning_rate": 3.6474042852717538e-06, "loss": 0.3468, "step": 11690 }, { "epoch": 0.591380444129698, "grad_norm": 6.314792156219482, "learning_rate": 3.6466316142589876e-06, "loss": 0.235, "step": 11691 }, { "epoch": 0.5914310283777632, "grad_norm": 7.0907487869262695, "learning_rate": 3.645858978119253e-06, "loss": 0.3423, "step": 11692 }, { "epoch": 0.5914816126258283, "grad_norm": 5.510745525360107, "learning_rate": 3.645086376872463e-06, "loss": 0.3574, "step": 11693 }, { "epoch": 0.5915321968738935, "grad_norm": 5.299161911010742, "learning_rate": 3.6443138105385257e-06, "loss": 0.348, "step": 11694 }, { "epoch": 0.5915827811219586, "grad_norm": 7.805335998535156, "learning_rate": 3.6435412791373475e-06, "loss": 0.3889, "step": 11695 }, { "epoch": 0.5916333653700238, "grad_norm": 12.589509010314941, "learning_rate": 3.642768782688832e-06, "loss": 0.481, "step": 11696 }, { "epoch": 0.5916839496180889, "grad_norm": 11.103790283203125, "learning_rate": 3.6419963212128866e-06, "loss": 0.5208, "step": 11697 }, { "epoch": 0.591734533866154, "grad_norm": 8.268939971923828, "learning_rate": 3.641223894729418e-06, "loss": 0.2768, "step": 11698 }, { "epoch": 0.5917851181142192, "grad_norm": 8.935236930847168, "learning_rate": 3.6404515032583252e-06, "loss": 0.3644, "step": 11699 }, { "epoch": 0.5918357023622843, "grad_norm": 10.427340507507324, "learning_rate": 3.6396791468195137e-06, "loss": 0.3827, "step": 11700 }, { "epoch": 0.5918862866103496, "grad_norm": 7.978565692901611, "learning_rate": 3.6389068254328864e-06, "loss": 0.4323, "step": 11701 }, { "epoch": 0.5919368708584147, "grad_norm": 7.390235900878906, "learning_rate": 3.638134539118341e-06, "loss": 0.4496, "step": 11702 }, { "epoch": 0.5919874551064799, "grad_norm": 14.459395408630371, "learning_rate": 3.6373622878957803e-06, "loss": 0.3739, "step": 11703 }, { "epoch": 0.592038039354545, "grad_norm": 6.074233055114746, "learning_rate": 3.636590071785101e-06, "loss": 0.319, "step": 11704 }, { "epoch": 0.5920886236026102, "grad_norm": 10.783327102661133, "learning_rate": 3.6358178908062024e-06, "loss": 0.4676, "step": 11705 }, { "epoch": 0.5921392078506753, "grad_norm": 7.5731987953186035, "learning_rate": 3.6350457449789827e-06, "loss": 0.3788, "step": 11706 }, { "epoch": 0.5921897920987405, "grad_norm": 11.0668306350708, "learning_rate": 3.6342736343233366e-06, "loss": 0.4606, "step": 11707 }, { "epoch": 0.5922403763468056, "grad_norm": 7.226195335388184, "learning_rate": 3.6335015588591616e-06, "loss": 0.4137, "step": 11708 }, { "epoch": 0.5922909605948707, "grad_norm": 9.647566795349121, "learning_rate": 3.6327295186063495e-06, "loss": 0.5293, "step": 11709 }, { "epoch": 0.5923415448429359, "grad_norm": 7.325069904327393, "learning_rate": 3.631957513584796e-06, "loss": 0.4122, "step": 11710 }, { "epoch": 0.592392129091001, "grad_norm": 5.816489219665527, "learning_rate": 3.6311855438143915e-06, "loss": 0.4057, "step": 11711 }, { "epoch": 0.5924427133390662, "grad_norm": 6.434671401977539, "learning_rate": 3.6304136093150295e-06, "loss": 0.3875, "step": 11712 }, { "epoch": 0.5924932975871313, "grad_norm": 10.67719554901123, "learning_rate": 3.629641710106602e-06, "loss": 0.4864, "step": 11713 }, { "epoch": 0.5925438818351966, "grad_norm": 7.401627063751221, "learning_rate": 3.6288698462089963e-06, "loss": 0.5022, "step": 11714 }, { "epoch": 0.5925944660832617, "grad_norm": 5.930440902709961, "learning_rate": 3.6280980176421026e-06, "loss": 0.3913, "step": 11715 }, { "epoch": 0.5926450503313269, "grad_norm": 19.29959487915039, "learning_rate": 3.6273262244258113e-06, "loss": 0.5449, "step": 11716 }, { "epoch": 0.592695634579392, "grad_norm": 8.907904624938965, "learning_rate": 3.6265544665800057e-06, "loss": 0.3802, "step": 11717 }, { "epoch": 0.5927462188274571, "grad_norm": 8.786761283874512, "learning_rate": 3.625782744124574e-06, "loss": 0.4474, "step": 11718 }, { "epoch": 0.5927968030755223, "grad_norm": 4.856817722320557, "learning_rate": 3.625011057079403e-06, "loss": 0.4098, "step": 11719 }, { "epoch": 0.5928473873235874, "grad_norm": 9.095174789428711, "learning_rate": 3.6242394054643753e-06, "loss": 0.4529, "step": 11720 }, { "epoch": 0.5928979715716526, "grad_norm": 14.41635513305664, "learning_rate": 3.623467789299377e-06, "loss": 0.5329, "step": 11721 }, { "epoch": 0.5929485558197177, "grad_norm": 13.777186393737793, "learning_rate": 3.6226962086042867e-06, "loss": 0.4652, "step": 11722 }, { "epoch": 0.5929991400677829, "grad_norm": 13.311214447021484, "learning_rate": 3.621924663398989e-06, "loss": 0.463, "step": 11723 }, { "epoch": 0.593049724315848, "grad_norm": 7.669860363006592, "learning_rate": 3.621153153703364e-06, "loss": 0.3815, "step": 11724 }, { "epoch": 0.5931003085639132, "grad_norm": 6.461414337158203, "learning_rate": 3.620381679537291e-06, "loss": 0.5021, "step": 11725 }, { "epoch": 0.5931508928119783, "grad_norm": 8.838677406311035, "learning_rate": 3.6196102409206515e-06, "loss": 0.49, "step": 11726 }, { "epoch": 0.5932014770600434, "grad_norm": 8.959809303283691, "learning_rate": 3.6188388378733195e-06, "loss": 0.5026, "step": 11727 }, { "epoch": 0.5932520613081087, "grad_norm": 6.6777873039245605, "learning_rate": 3.618067470415176e-06, "loss": 0.4454, "step": 11728 }, { "epoch": 0.5933026455561738, "grad_norm": 7.506287574768066, "learning_rate": 3.6172961385660965e-06, "loss": 0.4838, "step": 11729 }, { "epoch": 0.593353229804239, "grad_norm": 9.652868270874023, "learning_rate": 3.6165248423459542e-06, "loss": 0.5315, "step": 11730 }, { "epoch": 0.5934038140523041, "grad_norm": 9.3709135055542, "learning_rate": 3.615753581774627e-06, "loss": 0.4455, "step": 11731 }, { "epoch": 0.5934543983003693, "grad_norm": 13.108920097351074, "learning_rate": 3.6149823568719845e-06, "loss": 0.5991, "step": 11732 }, { "epoch": 0.5935049825484344, "grad_norm": 8.619277000427246, "learning_rate": 3.6142111676579015e-06, "loss": 0.4714, "step": 11733 }, { "epoch": 0.5935555667964996, "grad_norm": 6.476710319519043, "learning_rate": 3.6134400141522517e-06, "loss": 0.3852, "step": 11734 }, { "epoch": 0.5936061510445647, "grad_norm": 9.65482234954834, "learning_rate": 3.6126688963749023e-06, "loss": 0.3816, "step": 11735 }, { "epoch": 0.5936567352926299, "grad_norm": 7.7513227462768555, "learning_rate": 3.6118978143457245e-06, "loss": 0.3673, "step": 11736 }, { "epoch": 0.593707319540695, "grad_norm": 16.433238983154297, "learning_rate": 3.611126768084586e-06, "loss": 0.4694, "step": 11737 }, { "epoch": 0.5937579037887601, "grad_norm": 8.854571342468262, "learning_rate": 3.610355757611358e-06, "loss": 0.3797, "step": 11738 }, { "epoch": 0.5938084880368253, "grad_norm": 6.130041599273682, "learning_rate": 3.609584782945904e-06, "loss": 0.4213, "step": 11739 }, { "epoch": 0.5938590722848904, "grad_norm": 4.871250152587891, "learning_rate": 3.6088138441080923e-06, "loss": 0.3605, "step": 11740 }, { "epoch": 0.5939096565329557, "grad_norm": 5.018413543701172, "learning_rate": 3.6080429411177882e-06, "loss": 0.3235, "step": 11741 }, { "epoch": 0.5939602407810208, "grad_norm": 15.146007537841797, "learning_rate": 3.607272073994854e-06, "loss": 0.4547, "step": 11742 }, { "epoch": 0.594010825029086, "grad_norm": 7.638473987579346, "learning_rate": 3.606501242759155e-06, "loss": 0.4273, "step": 11743 }, { "epoch": 0.5940614092771511, "grad_norm": 11.391465187072754, "learning_rate": 3.6057304474305533e-06, "loss": 0.4279, "step": 11744 }, { "epoch": 0.5941119935252163, "grad_norm": 4.968838691711426, "learning_rate": 3.60495968802891e-06, "loss": 0.3938, "step": 11745 }, { "epoch": 0.5941625777732814, "grad_norm": 6.6395463943481445, "learning_rate": 3.604188964574086e-06, "loss": 0.3825, "step": 11746 }, { "epoch": 0.5942131620213466, "grad_norm": 5.838161945343018, "learning_rate": 3.6034182770859427e-06, "loss": 0.3846, "step": 11747 }, { "epoch": 0.5942637462694117, "grad_norm": 8.320832252502441, "learning_rate": 3.602647625584336e-06, "loss": 0.3421, "step": 11748 }, { "epoch": 0.5943143305174768, "grad_norm": 5.731321334838867, "learning_rate": 3.6018770100891244e-06, "loss": 0.4409, "step": 11749 }, { "epoch": 0.594364914765542, "grad_norm": 5.734219551086426, "learning_rate": 3.6011064306201644e-06, "loss": 0.4217, "step": 11750 }, { "epoch": 0.5944154990136071, "grad_norm": 9.567572593688965, "learning_rate": 3.6003358871973137e-06, "loss": 0.3665, "step": 11751 }, { "epoch": 0.5944660832616723, "grad_norm": 9.471253395080566, "learning_rate": 3.599565379840426e-06, "loss": 0.4683, "step": 11752 }, { "epoch": 0.5945166675097374, "grad_norm": 9.087013244628906, "learning_rate": 3.5987949085693564e-06, "loss": 0.4323, "step": 11753 }, { "epoch": 0.5945672517578027, "grad_norm": 6.921726703643799, "learning_rate": 3.598024473403957e-06, "loss": 0.3462, "step": 11754 }, { "epoch": 0.5946178360058678, "grad_norm": 6.835752964019775, "learning_rate": 3.5972540743640805e-06, "loss": 0.433, "step": 11755 }, { "epoch": 0.594668420253933, "grad_norm": 7.721312999725342, "learning_rate": 3.5964837114695782e-06, "loss": 0.4207, "step": 11756 }, { "epoch": 0.5947190045019981, "grad_norm": 7.076298236846924, "learning_rate": 3.595713384740302e-06, "loss": 0.3885, "step": 11757 }, { "epoch": 0.5947695887500632, "grad_norm": 4.161340713500977, "learning_rate": 3.5949430941960982e-06, "loss": 0.2463, "step": 11758 }, { "epoch": 0.5948201729981284, "grad_norm": 11.54409122467041, "learning_rate": 3.5941728398568175e-06, "loss": 0.3713, "step": 11759 }, { "epoch": 0.5948707572461935, "grad_norm": 7.33205509185791, "learning_rate": 3.5934026217423095e-06, "loss": 0.4842, "step": 11760 }, { "epoch": 0.5949213414942587, "grad_norm": 8.739376068115234, "learning_rate": 3.592632439872417e-06, "loss": 0.4424, "step": 11761 }, { "epoch": 0.5949719257423238, "grad_norm": 5.322826862335205, "learning_rate": 3.591862294266986e-06, "loss": 0.4184, "step": 11762 }, { "epoch": 0.595022509990389, "grad_norm": 7.8483781814575195, "learning_rate": 3.591092184945863e-06, "loss": 0.4236, "step": 11763 }, { "epoch": 0.5950730942384541, "grad_norm": 8.620266914367676, "learning_rate": 3.5903221119288923e-06, "loss": 0.4095, "step": 11764 }, { "epoch": 0.5951236784865193, "grad_norm": 10.994719505310059, "learning_rate": 3.589552075235914e-06, "loss": 0.4854, "step": 11765 }, { "epoch": 0.5951742627345844, "grad_norm": 8.704781532287598, "learning_rate": 3.5887820748867726e-06, "loss": 0.3838, "step": 11766 }, { "epoch": 0.5952248469826497, "grad_norm": 10.474069595336914, "learning_rate": 3.588012110901309e-06, "loss": 0.3771, "step": 11767 }, { "epoch": 0.5952754312307148, "grad_norm": 11.32917594909668, "learning_rate": 3.587242183299362e-06, "loss": 0.6031, "step": 11768 }, { "epoch": 0.59532601547878, "grad_norm": 5.909265518188477, "learning_rate": 3.5864722921007734e-06, "loss": 0.3701, "step": 11769 }, { "epoch": 0.5953765997268451, "grad_norm": 7.141290664672852, "learning_rate": 3.585702437325377e-06, "loss": 0.3307, "step": 11770 }, { "epoch": 0.5954271839749102, "grad_norm": 6.167169570922852, "learning_rate": 3.584932618993014e-06, "loss": 0.3337, "step": 11771 }, { "epoch": 0.5954777682229754, "grad_norm": 16.846603393554688, "learning_rate": 3.58416283712352e-06, "loss": 0.6375, "step": 11772 }, { "epoch": 0.5955283524710405, "grad_norm": 8.019662857055664, "learning_rate": 3.58339309173673e-06, "loss": 0.4472, "step": 11773 }, { "epoch": 0.5955789367191057, "grad_norm": 9.576883316040039, "learning_rate": 3.5826233828524782e-06, "loss": 0.4795, "step": 11774 }, { "epoch": 0.5956295209671708, "grad_norm": 7.359207630157471, "learning_rate": 3.5818537104905972e-06, "loss": 0.4888, "step": 11775 }, { "epoch": 0.595680105215236, "grad_norm": 8.162691116333008, "learning_rate": 3.581084074670922e-06, "loss": 0.4285, "step": 11776 }, { "epoch": 0.5957306894633011, "grad_norm": 6.113225936889648, "learning_rate": 3.5803144754132813e-06, "loss": 0.3837, "step": 11777 }, { "epoch": 0.5957812737113662, "grad_norm": 7.824382781982422, "learning_rate": 3.5795449127375083e-06, "loss": 0.2598, "step": 11778 }, { "epoch": 0.5958318579594314, "grad_norm": 9.771927833557129, "learning_rate": 3.5787753866634324e-06, "loss": 0.3696, "step": 11779 }, { "epoch": 0.5958824422074965, "grad_norm": 6.268984317779541, "learning_rate": 3.578005897210881e-06, "loss": 0.334, "step": 11780 }, { "epoch": 0.5959330264555618, "grad_norm": 5.784428119659424, "learning_rate": 3.5772364443996834e-06, "loss": 0.3667, "step": 11781 }, { "epoch": 0.5959836107036269, "grad_norm": 10.83769416809082, "learning_rate": 3.576467028249666e-06, "loss": 0.3953, "step": 11782 }, { "epoch": 0.5960341949516921, "grad_norm": 5.381788730621338, "learning_rate": 3.575697648780655e-06, "loss": 0.3656, "step": 11783 }, { "epoch": 0.5960847791997572, "grad_norm": 5.669276714324951, "learning_rate": 3.574928306012475e-06, "loss": 0.3775, "step": 11784 }, { "epoch": 0.5961353634478224, "grad_norm": 11.111661911010742, "learning_rate": 3.5741589999649523e-06, "loss": 0.5415, "step": 11785 }, { "epoch": 0.5961859476958875, "grad_norm": 9.195773124694824, "learning_rate": 3.5733897306579068e-06, "loss": 0.4523, "step": 11786 }, { "epoch": 0.5962365319439527, "grad_norm": 7.874078273773193, "learning_rate": 3.5726204981111624e-06, "loss": 0.48, "step": 11787 }, { "epoch": 0.5962871161920178, "grad_norm": 9.364233016967773, "learning_rate": 3.571851302344539e-06, "loss": 0.4121, "step": 11788 }, { "epoch": 0.5963377004400829, "grad_norm": 8.391743659973145, "learning_rate": 3.5710821433778586e-06, "loss": 0.4451, "step": 11789 }, { "epoch": 0.5963882846881481, "grad_norm": 4.719883441925049, "learning_rate": 3.57031302123094e-06, "loss": 0.3132, "step": 11790 }, { "epoch": 0.5964388689362132, "grad_norm": 9.181553840637207, "learning_rate": 3.569543935923601e-06, "loss": 0.4827, "step": 11791 }, { "epoch": 0.5964894531842784, "grad_norm": 6.436228275299072, "learning_rate": 3.568774887475661e-06, "loss": 0.463, "step": 11792 }, { "epoch": 0.5965400374323435, "grad_norm": 7.195648193359375, "learning_rate": 3.5680058759069336e-06, "loss": 0.4378, "step": 11793 }, { "epoch": 0.5965906216804088, "grad_norm": 8.4586763381958, "learning_rate": 3.5672369012372364e-06, "loss": 0.4291, "step": 11794 }, { "epoch": 0.5966412059284739, "grad_norm": 11.048384666442871, "learning_rate": 3.566467963486384e-06, "loss": 0.5443, "step": 11795 }, { "epoch": 0.5966917901765391, "grad_norm": 4.901044845581055, "learning_rate": 3.5656990626741893e-06, "loss": 0.3045, "step": 11796 }, { "epoch": 0.5967423744246042, "grad_norm": 6.865917682647705, "learning_rate": 3.5649301988204654e-06, "loss": 0.3492, "step": 11797 }, { "epoch": 0.5967929586726693, "grad_norm": 4.560728073120117, "learning_rate": 3.5641613719450256e-06, "loss": 0.3469, "step": 11798 }, { "epoch": 0.5968435429207345, "grad_norm": 8.408060073852539, "learning_rate": 3.5633925820676784e-06, "loss": 0.4758, "step": 11799 }, { "epoch": 0.5968941271687996, "grad_norm": 16.5594539642334, "learning_rate": 3.5626238292082332e-06, "loss": 0.5743, "step": 11800 }, { "epoch": 0.5969447114168648, "grad_norm": 5.498528480529785, "learning_rate": 3.5618551133864997e-06, "loss": 0.4758, "step": 11801 }, { "epoch": 0.5969952956649299, "grad_norm": 7.791516304016113, "learning_rate": 3.5610864346222882e-06, "loss": 0.3208, "step": 11802 }, { "epoch": 0.5970458799129951, "grad_norm": 7.520038604736328, "learning_rate": 3.560317792935402e-06, "loss": 0.4165, "step": 11803 }, { "epoch": 0.5970964641610602, "grad_norm": 7.533854007720947, "learning_rate": 3.559549188345649e-06, "loss": 0.3911, "step": 11804 }, { "epoch": 0.5971470484091254, "grad_norm": 10.640534400939941, "learning_rate": 3.5587806208728354e-06, "loss": 0.3403, "step": 11805 }, { "epoch": 0.5971976326571905, "grad_norm": 6.829623699188232, "learning_rate": 3.5580120905367628e-06, "loss": 0.3907, "step": 11806 }, { "epoch": 0.5972482169052558, "grad_norm": 7.711133003234863, "learning_rate": 3.5572435973572366e-06, "loss": 0.4337, "step": 11807 }, { "epoch": 0.5972988011533209, "grad_norm": 19.022846221923828, "learning_rate": 3.5564751413540573e-06, "loss": 0.5271, "step": 11808 }, { "epoch": 0.597349385401386, "grad_norm": 5.334953308105469, "learning_rate": 3.5557067225470267e-06, "loss": 0.4275, "step": 11809 }, { "epoch": 0.5973999696494512, "grad_norm": 8.819265365600586, "learning_rate": 3.5549383409559464e-06, "loss": 0.3777, "step": 11810 }, { "epoch": 0.5974505538975163, "grad_norm": 8.580976486206055, "learning_rate": 3.554169996600616e-06, "loss": 0.4063, "step": 11811 }, { "epoch": 0.5975011381455815, "grad_norm": 7.7870283126831055, "learning_rate": 3.5534016895008292e-06, "loss": 0.471, "step": 11812 }, { "epoch": 0.5975517223936466, "grad_norm": 10.278231620788574, "learning_rate": 3.5526334196763883e-06, "loss": 0.4391, "step": 11813 }, { "epoch": 0.5976023066417118, "grad_norm": 11.30829906463623, "learning_rate": 3.551865187147089e-06, "loss": 0.5767, "step": 11814 }, { "epoch": 0.5976528908897769, "grad_norm": 16.573566436767578, "learning_rate": 3.5510969919327243e-06, "loss": 0.656, "step": 11815 }, { "epoch": 0.597703475137842, "grad_norm": 10.856868743896484, "learning_rate": 3.5503288340530907e-06, "loss": 0.4174, "step": 11816 }, { "epoch": 0.5977540593859072, "grad_norm": 5.6249165534973145, "learning_rate": 3.5495607135279823e-06, "loss": 0.3887, "step": 11817 }, { "epoch": 0.5978046436339723, "grad_norm": 6.0041184425354, "learning_rate": 3.54879263037719e-06, "loss": 0.3389, "step": 11818 }, { "epoch": 0.5978552278820375, "grad_norm": 5.849336624145508, "learning_rate": 3.548024584620507e-06, "loss": 0.3419, "step": 11819 }, { "epoch": 0.5979058121301026, "grad_norm": 8.812188148498535, "learning_rate": 3.547256576277724e-06, "loss": 0.3051, "step": 11820 }, { "epoch": 0.5979563963781679, "grad_norm": 9.375046730041504, "learning_rate": 3.5464886053686286e-06, "loss": 0.4011, "step": 11821 }, { "epoch": 0.598006980626233, "grad_norm": 7.560469627380371, "learning_rate": 3.5457206719130116e-06, "loss": 0.4592, "step": 11822 }, { "epoch": 0.5980575648742982, "grad_norm": 7.2876362800598145, "learning_rate": 3.544952775930661e-06, "loss": 0.3956, "step": 11823 }, { "epoch": 0.5981081491223633, "grad_norm": 10.051515579223633, "learning_rate": 3.544184917441364e-06, "loss": 0.4428, "step": 11824 }, { "epoch": 0.5981587333704285, "grad_norm": 9.861940383911133, "learning_rate": 3.543417096464903e-06, "loss": 0.4342, "step": 11825 }, { "epoch": 0.5982093176184936, "grad_norm": 5.084065914154053, "learning_rate": 3.5426493130210655e-06, "loss": 0.4188, "step": 11826 }, { "epoch": 0.5982599018665588, "grad_norm": 7.491342067718506, "learning_rate": 3.541881567129636e-06, "loss": 0.4153, "step": 11827 }, { "epoch": 0.5983104861146239, "grad_norm": 6.390773773193359, "learning_rate": 3.5411138588103953e-06, "loss": 0.382, "step": 11828 }, { "epoch": 0.598361070362689, "grad_norm": 7.145532131195068, "learning_rate": 3.540346188083127e-06, "loss": 0.4404, "step": 11829 }, { "epoch": 0.5984116546107542, "grad_norm": 6.792956829071045, "learning_rate": 3.539578554967613e-06, "loss": 0.4321, "step": 11830 }, { "epoch": 0.5984622388588193, "grad_norm": 6.000840663909912, "learning_rate": 3.5388109594836308e-06, "loss": 0.3698, "step": 11831 }, { "epoch": 0.5985128231068845, "grad_norm": 7.067935466766357, "learning_rate": 3.5380434016509605e-06, "loss": 0.3872, "step": 11832 }, { "epoch": 0.5985634073549496, "grad_norm": 15.875763893127441, "learning_rate": 3.5372758814893822e-06, "loss": 0.5297, "step": 11833 }, { "epoch": 0.5986139916030149, "grad_norm": 4.90269660949707, "learning_rate": 3.53650839901867e-06, "loss": 0.4404, "step": 11834 }, { "epoch": 0.59866457585108, "grad_norm": 6.257713317871094, "learning_rate": 3.5357409542586013e-06, "loss": 0.3471, "step": 11835 }, { "epoch": 0.5987151600991452, "grad_norm": 11.100289344787598, "learning_rate": 3.5349735472289525e-06, "loss": 0.4956, "step": 11836 }, { "epoch": 0.5987657443472103, "grad_norm": 5.664422035217285, "learning_rate": 3.5342061779494984e-06, "loss": 0.3772, "step": 11837 }, { "epoch": 0.5988163285952754, "grad_norm": 9.656378746032715, "learning_rate": 3.533438846440008e-06, "loss": 0.3824, "step": 11838 }, { "epoch": 0.5988669128433406, "grad_norm": 6.709341526031494, "learning_rate": 3.532671552720256e-06, "loss": 0.4256, "step": 11839 }, { "epoch": 0.5989174970914057, "grad_norm": 10.916709899902344, "learning_rate": 3.5319042968100154e-06, "loss": 0.4479, "step": 11840 }, { "epoch": 0.5989680813394709, "grad_norm": 10.575008392333984, "learning_rate": 3.531137078729053e-06, "loss": 0.4361, "step": 11841 }, { "epoch": 0.599018665587536, "grad_norm": 7.967655658721924, "learning_rate": 3.5303698984971422e-06, "loss": 0.3444, "step": 11842 }, { "epoch": 0.5990692498356012, "grad_norm": 10.798943519592285, "learning_rate": 3.5296027561340484e-06, "loss": 0.387, "step": 11843 }, { "epoch": 0.5991198340836663, "grad_norm": 7.030792236328125, "learning_rate": 3.52883565165954e-06, "loss": 0.3689, "step": 11844 }, { "epoch": 0.5991704183317315, "grad_norm": 10.678497314453125, "learning_rate": 3.5280685850933837e-06, "loss": 0.3979, "step": 11845 }, { "epoch": 0.5992210025797966, "grad_norm": 9.342153549194336, "learning_rate": 3.527301556455345e-06, "loss": 0.3828, "step": 11846 }, { "epoch": 0.5992715868278619, "grad_norm": 12.463780403137207, "learning_rate": 3.5265345657651866e-06, "loss": 0.4105, "step": 11847 }, { "epoch": 0.599322171075927, "grad_norm": 11.512167930603027, "learning_rate": 3.5257676130426756e-06, "loss": 0.3987, "step": 11848 }, { "epoch": 0.5993727553239921, "grad_norm": 5.41763973236084, "learning_rate": 3.5250006983075703e-06, "loss": 0.4043, "step": 11849 }, { "epoch": 0.5994233395720573, "grad_norm": 6.71949577331543, "learning_rate": 3.5242338215796368e-06, "loss": 0.3669, "step": 11850 }, { "epoch": 0.5994739238201224, "grad_norm": 5.168011665344238, "learning_rate": 3.5234669828786303e-06, "loss": 0.3087, "step": 11851 }, { "epoch": 0.5995245080681876, "grad_norm": 8.636125564575195, "learning_rate": 3.5227001822243157e-06, "loss": 0.4064, "step": 11852 }, { "epoch": 0.5995750923162527, "grad_norm": 14.911266326904297, "learning_rate": 3.5219334196364473e-06, "loss": 0.4807, "step": 11853 }, { "epoch": 0.5996256765643179, "grad_norm": 6.287284851074219, "learning_rate": 3.5211666951347844e-06, "loss": 0.2983, "step": 11854 }, { "epoch": 0.599676260812383, "grad_norm": 7.797388553619385, "learning_rate": 3.5204000087390856e-06, "loss": 0.3932, "step": 11855 }, { "epoch": 0.5997268450604482, "grad_norm": 10.178038597106934, "learning_rate": 3.519633360469103e-06, "loss": 0.5741, "step": 11856 }, { "epoch": 0.5997774293085133, "grad_norm": 6.554142475128174, "learning_rate": 3.5188667503445927e-06, "loss": 0.3782, "step": 11857 }, { "epoch": 0.5998280135565784, "grad_norm": 6.361329555511475, "learning_rate": 3.518100178385311e-06, "loss": 0.4279, "step": 11858 }, { "epoch": 0.5998785978046436, "grad_norm": 10.394641876220703, "learning_rate": 3.517333644611006e-06, "loss": 0.4553, "step": 11859 }, { "epoch": 0.5999291820527087, "grad_norm": 13.373577117919922, "learning_rate": 3.5165671490414327e-06, "loss": 0.4738, "step": 11860 }, { "epoch": 0.599979766300774, "grad_norm": 6.049449920654297, "learning_rate": 3.5158006916963415e-06, "loss": 0.3929, "step": 11861 }, { "epoch": 0.6000303505488391, "grad_norm": 8.473231315612793, "learning_rate": 3.5150342725954807e-06, "loss": 0.4455, "step": 11862 }, { "epoch": 0.6000303505488391, "eval_loss": 0.4040122330188751, "eval_runtime": 34.9025, "eval_samples_per_second": 45.785, "eval_steps_per_second": 5.73, "step": 11862 }, { "epoch": 0.6000809347969043, "grad_norm": 17.384029388427734, "learning_rate": 3.5142678917586015e-06, "loss": 0.3696, "step": 11863 }, { "epoch": 0.6001315190449694, "grad_norm": 7.037545680999756, "learning_rate": 3.513501549205448e-06, "loss": 0.4402, "step": 11864 }, { "epoch": 0.6001821032930346, "grad_norm": 14.243949890136719, "learning_rate": 3.5127352449557706e-06, "loss": 0.4245, "step": 11865 }, { "epoch": 0.6002326875410997, "grad_norm": 10.054254531860352, "learning_rate": 3.5119689790293127e-06, "loss": 0.4963, "step": 11866 }, { "epoch": 0.6002832717891649, "grad_norm": 7.569592475891113, "learning_rate": 3.5112027514458196e-06, "loss": 0.4402, "step": 11867 }, { "epoch": 0.60033385603723, "grad_norm": 7.3360161781311035, "learning_rate": 3.510436562225037e-06, "loss": 0.5088, "step": 11868 }, { "epoch": 0.6003844402852951, "grad_norm": 20.11545181274414, "learning_rate": 3.509670411386705e-06, "loss": 0.4621, "step": 11869 }, { "epoch": 0.6004350245333603, "grad_norm": 7.190863609313965, "learning_rate": 3.5089042989505663e-06, "loss": 0.4431, "step": 11870 }, { "epoch": 0.6004856087814254, "grad_norm": 7.291831016540527, "learning_rate": 3.5081382249363637e-06, "loss": 0.5268, "step": 11871 }, { "epoch": 0.6005361930294906, "grad_norm": 11.765048027038574, "learning_rate": 3.5073721893638344e-06, "loss": 0.3993, "step": 11872 }, { "epoch": 0.6005867772775557, "grad_norm": 7.437768459320068, "learning_rate": 3.50660619225272e-06, "loss": 0.413, "step": 11873 }, { "epoch": 0.600637361525621, "grad_norm": 11.786062240600586, "learning_rate": 3.5058402336227552e-06, "loss": 0.3955, "step": 11874 }, { "epoch": 0.6006879457736861, "grad_norm": 9.411446571350098, "learning_rate": 3.5050743134936817e-06, "loss": 0.3219, "step": 11875 }, { "epoch": 0.6007385300217513, "grad_norm": 6.80676794052124, "learning_rate": 3.5043084318852293e-06, "loss": 0.3624, "step": 11876 }, { "epoch": 0.6007891142698164, "grad_norm": 9.170258522033691, "learning_rate": 3.503542588817136e-06, "loss": 0.3952, "step": 11877 }, { "epoch": 0.6008396985178815, "grad_norm": 10.76016616821289, "learning_rate": 3.5027767843091363e-06, "loss": 0.4367, "step": 11878 }, { "epoch": 0.6008902827659467, "grad_norm": 5.838324069976807, "learning_rate": 3.5020110183809624e-06, "loss": 0.3809, "step": 11879 }, { "epoch": 0.6009408670140118, "grad_norm": 7.118165016174316, "learning_rate": 3.5012452910523463e-06, "loss": 0.3431, "step": 11880 }, { "epoch": 0.600991451262077, "grad_norm": 9.467891693115234, "learning_rate": 3.500479602343019e-06, "loss": 0.4115, "step": 11881 }, { "epoch": 0.6010420355101421, "grad_norm": 5.404298305511475, "learning_rate": 3.49971395227271e-06, "loss": 0.2753, "step": 11882 }, { "epoch": 0.6010926197582073, "grad_norm": 7.5338134765625, "learning_rate": 3.4989483408611493e-06, "loss": 0.3581, "step": 11883 }, { "epoch": 0.6011432040062724, "grad_norm": 7.025312900543213, "learning_rate": 3.4981827681280633e-06, "loss": 0.3588, "step": 11884 }, { "epoch": 0.6011937882543376, "grad_norm": 6.774073123931885, "learning_rate": 3.49741723409318e-06, "loss": 0.3489, "step": 11885 }, { "epoch": 0.6012443725024027, "grad_norm": 21.82502555847168, "learning_rate": 3.4966517387762265e-06, "loss": 0.4792, "step": 11886 }, { "epoch": 0.601294956750468, "grad_norm": 11.52067756652832, "learning_rate": 3.4958862821969253e-06, "loss": 0.3779, "step": 11887 }, { "epoch": 0.6013455409985331, "grad_norm": 8.738964080810547, "learning_rate": 3.495120864375004e-06, "loss": 0.411, "step": 11888 }, { "epoch": 0.6013961252465982, "grad_norm": 8.2073974609375, "learning_rate": 3.494355485330181e-06, "loss": 0.397, "step": 11889 }, { "epoch": 0.6014467094946634, "grad_norm": 6.208498001098633, "learning_rate": 3.493590145082181e-06, "loss": 0.3474, "step": 11890 }, { "epoch": 0.6014972937427285, "grad_norm": 6.706328392028809, "learning_rate": 3.4928248436507227e-06, "loss": 0.4391, "step": 11891 }, { "epoch": 0.6015478779907937, "grad_norm": 9.784865379333496, "learning_rate": 3.492059581055529e-06, "loss": 0.4301, "step": 11892 }, { "epoch": 0.6015984622388588, "grad_norm": 6.290877819061279, "learning_rate": 3.491294357316318e-06, "loss": 0.3383, "step": 11893 }, { "epoch": 0.601649046486924, "grad_norm": 8.417438507080078, "learning_rate": 3.4905291724528067e-06, "loss": 0.3956, "step": 11894 }, { "epoch": 0.6016996307349891, "grad_norm": 9.922567367553711, "learning_rate": 3.4897640264847123e-06, "loss": 0.4855, "step": 11895 }, { "epoch": 0.6017502149830543, "grad_norm": 5.469772815704346, "learning_rate": 3.488998919431752e-06, "loss": 0.3839, "step": 11896 }, { "epoch": 0.6018007992311194, "grad_norm": 6.605167865753174, "learning_rate": 3.4882338513136393e-06, "loss": 0.3457, "step": 11897 }, { "epoch": 0.6018513834791845, "grad_norm": 7.756651878356934, "learning_rate": 3.4874688221500886e-06, "loss": 0.405, "step": 11898 }, { "epoch": 0.6019019677272497, "grad_norm": 12.742502212524414, "learning_rate": 3.4867038319608136e-06, "loss": 0.4759, "step": 11899 }, { "epoch": 0.6019525519753149, "grad_norm": 5.79623556137085, "learning_rate": 3.4859388807655246e-06, "loss": 0.3836, "step": 11900 }, { "epoch": 0.6020031362233801, "grad_norm": 6.251970291137695, "learning_rate": 3.4851739685839366e-06, "loss": 0.3935, "step": 11901 }, { "epoch": 0.6020537204714452, "grad_norm": 6.23960542678833, "learning_rate": 3.4844090954357532e-06, "loss": 0.3769, "step": 11902 }, { "epoch": 0.6021043047195104, "grad_norm": 6.540955543518066, "learning_rate": 3.4836442613406884e-06, "loss": 0.3505, "step": 11903 }, { "epoch": 0.6021548889675755, "grad_norm": 9.043177604675293, "learning_rate": 3.482879466318447e-06, "loss": 0.4476, "step": 11904 }, { "epoch": 0.6022054732156407, "grad_norm": 10.979931831359863, "learning_rate": 3.482114710388738e-06, "loss": 0.3979, "step": 11905 }, { "epoch": 0.6022560574637058, "grad_norm": 5.819793224334717, "learning_rate": 3.481349993571267e-06, "loss": 0.3384, "step": 11906 }, { "epoch": 0.602306641711771, "grad_norm": 13.764256477355957, "learning_rate": 3.4805853158857377e-06, "loss": 0.5501, "step": 11907 }, { "epoch": 0.6023572259598361, "grad_norm": 5.073709011077881, "learning_rate": 3.479820677351855e-06, "loss": 0.2985, "step": 11908 }, { "epoch": 0.6024078102079012, "grad_norm": 8.240219116210938, "learning_rate": 3.4790560779893223e-06, "loss": 0.4602, "step": 11909 }, { "epoch": 0.6024583944559664, "grad_norm": 6.9753737449646, "learning_rate": 3.4782915178178402e-06, "loss": 0.4294, "step": 11910 }, { "epoch": 0.6025089787040315, "grad_norm": 6.833995819091797, "learning_rate": 3.4775269968571104e-06, "loss": 0.3213, "step": 11911 }, { "epoch": 0.6025595629520967, "grad_norm": 6.967747211456299, "learning_rate": 3.476762515126832e-06, "loss": 0.4446, "step": 11912 }, { "epoch": 0.6026101472001618, "grad_norm": 5.718733310699463, "learning_rate": 3.4759980726467046e-06, "loss": 0.3939, "step": 11913 }, { "epoch": 0.6026607314482271, "grad_norm": 7.350438594818115, "learning_rate": 3.4752336694364276e-06, "loss": 0.4116, "step": 11914 }, { "epoch": 0.6027113156962922, "grad_norm": 8.299646377563477, "learning_rate": 3.4744693055156947e-06, "loss": 0.4022, "step": 11915 }, { "epoch": 0.6027618999443574, "grad_norm": 5.55868673324585, "learning_rate": 3.4737049809042023e-06, "loss": 0.4132, "step": 11916 }, { "epoch": 0.6028124841924225, "grad_norm": 6.33300256729126, "learning_rate": 3.4729406956216457e-06, "loss": 0.4499, "step": 11917 }, { "epoch": 0.6028630684404876, "grad_norm": 6.433618545532227, "learning_rate": 3.4721764496877196e-06, "loss": 0.4169, "step": 11918 }, { "epoch": 0.6029136526885528, "grad_norm": 6.231856822967529, "learning_rate": 3.471412243122116e-06, "loss": 0.4114, "step": 11919 }, { "epoch": 0.6029642369366179, "grad_norm": 7.767956733703613, "learning_rate": 3.470648075944526e-06, "loss": 0.3458, "step": 11920 }, { "epoch": 0.6030148211846831, "grad_norm": 8.647462844848633, "learning_rate": 3.4698839481746428e-06, "loss": 0.4493, "step": 11921 }, { "epoch": 0.6030654054327482, "grad_norm": 8.518511772155762, "learning_rate": 3.469119859832153e-06, "loss": 0.3819, "step": 11922 }, { "epoch": 0.6031159896808134, "grad_norm": 34.283836364746094, "learning_rate": 3.4683558109367466e-06, "loss": 0.3883, "step": 11923 }, { "epoch": 0.6031665739288785, "grad_norm": 6.492318153381348, "learning_rate": 3.467591801508112e-06, "loss": 0.3516, "step": 11924 }, { "epoch": 0.6032171581769437, "grad_norm": 7.611593246459961, "learning_rate": 3.4668278315659356e-06, "loss": 0.4166, "step": 11925 }, { "epoch": 0.6032677424250088, "grad_norm": 10.312641143798828, "learning_rate": 3.4660639011299014e-06, "loss": 0.5032, "step": 11926 }, { "epoch": 0.6033183266730741, "grad_norm": 12.969548225402832, "learning_rate": 3.4653000102196982e-06, "loss": 0.4621, "step": 11927 }, { "epoch": 0.6033689109211392, "grad_norm": 8.558551788330078, "learning_rate": 3.4645361588550053e-06, "loss": 0.4098, "step": 11928 }, { "epoch": 0.6034194951692043, "grad_norm": 6.227689266204834, "learning_rate": 3.4637723470555064e-06, "loss": 0.312, "step": 11929 }, { "epoch": 0.6034700794172695, "grad_norm": 9.754706382751465, "learning_rate": 3.463008574840883e-06, "loss": 0.4572, "step": 11930 }, { "epoch": 0.6035206636653346, "grad_norm": 6.022980690002441, "learning_rate": 3.4622448422308176e-06, "loss": 0.4021, "step": 11931 }, { "epoch": 0.6035712479133998, "grad_norm": 6.070736885070801, "learning_rate": 3.4614811492449873e-06, "loss": 0.4612, "step": 11932 }, { "epoch": 0.6036218321614649, "grad_norm": 7.296384811401367, "learning_rate": 3.4607174959030707e-06, "loss": 0.4324, "step": 11933 }, { "epoch": 0.6036724164095301, "grad_norm": 6.801690578460693, "learning_rate": 3.4599538822247484e-06, "loss": 0.4119, "step": 11934 }, { "epoch": 0.6037230006575952, "grad_norm": 7.683820724487305, "learning_rate": 3.4591903082296936e-06, "loss": 0.4081, "step": 11935 }, { "epoch": 0.6037735849056604, "grad_norm": 8.783742904663086, "learning_rate": 3.4584267739375827e-06, "loss": 0.3931, "step": 11936 }, { "epoch": 0.6038241691537255, "grad_norm": 5.892608642578125, "learning_rate": 3.4576632793680907e-06, "loss": 0.4003, "step": 11937 }, { "epoch": 0.6038747534017906, "grad_norm": 10.809584617614746, "learning_rate": 3.4568998245408903e-06, "loss": 0.4485, "step": 11938 }, { "epoch": 0.6039253376498558, "grad_norm": 23.555875778198242, "learning_rate": 3.4561364094756544e-06, "loss": 0.3682, "step": 11939 }, { "epoch": 0.603975921897921, "grad_norm": 12.30597972869873, "learning_rate": 3.4553730341920556e-06, "loss": 0.3507, "step": 11940 }, { "epoch": 0.6040265061459862, "grad_norm": 9.987524032592773, "learning_rate": 3.454609698709762e-06, "loss": 0.5489, "step": 11941 }, { "epoch": 0.6040770903940513, "grad_norm": 7.0442585945129395, "learning_rate": 3.4538464030484432e-06, "loss": 0.408, "step": 11942 }, { "epoch": 0.6041276746421165, "grad_norm": 4.0431623458862305, "learning_rate": 3.4530831472277676e-06, "loss": 0.3468, "step": 11943 }, { "epoch": 0.6041782588901816, "grad_norm": 10.797289848327637, "learning_rate": 3.452319931267404e-06, "loss": 0.5039, "step": 11944 }, { "epoch": 0.6042288431382468, "grad_norm": 10.563939094543457, "learning_rate": 3.451556755187016e-06, "loss": 0.4859, "step": 11945 }, { "epoch": 0.6042794273863119, "grad_norm": 9.604598999023438, "learning_rate": 3.4507936190062717e-06, "loss": 0.619, "step": 11946 }, { "epoch": 0.604330011634377, "grad_norm": 8.091526985168457, "learning_rate": 3.450030522744833e-06, "loss": 0.4307, "step": 11947 }, { "epoch": 0.6043805958824422, "grad_norm": 5.753750801086426, "learning_rate": 3.4492674664223633e-06, "loss": 0.4367, "step": 11948 }, { "epoch": 0.6044311801305073, "grad_norm": 19.531070709228516, "learning_rate": 3.4485044500585274e-06, "loss": 0.4516, "step": 11949 }, { "epoch": 0.6044817643785725, "grad_norm": 7.253159523010254, "learning_rate": 3.447741473672983e-06, "loss": 0.4, "step": 11950 }, { "epoch": 0.6045323486266376, "grad_norm": 9.695802688598633, "learning_rate": 3.4469785372853914e-06, "loss": 0.4713, "step": 11951 }, { "epoch": 0.6045829328747028, "grad_norm": 3.9586293697357178, "learning_rate": 3.4462156409154123e-06, "loss": 0.3421, "step": 11952 }, { "epoch": 0.6046335171227679, "grad_norm": 4.374165058135986, "learning_rate": 3.4454527845827046e-06, "loss": 0.4216, "step": 11953 }, { "epoch": 0.6046841013708332, "grad_norm": 14.626592636108398, "learning_rate": 3.4446899683069214e-06, "loss": 0.4968, "step": 11954 }, { "epoch": 0.6047346856188983, "grad_norm": 13.6405668258667, "learning_rate": 3.443927192107721e-06, "loss": 0.4596, "step": 11955 }, { "epoch": 0.6047852698669635, "grad_norm": 4.986231803894043, "learning_rate": 3.44316445600476e-06, "loss": 0.3735, "step": 11956 }, { "epoch": 0.6048358541150286, "grad_norm": 5.013978958129883, "learning_rate": 3.442401760017689e-06, "loss": 0.4241, "step": 11957 }, { "epoch": 0.6048864383630937, "grad_norm": 15.752145767211914, "learning_rate": 3.4416391041661622e-06, "loss": 0.8501, "step": 11958 }, { "epoch": 0.6049370226111589, "grad_norm": 5.189300060272217, "learning_rate": 3.4408764884698333e-06, "loss": 0.2801, "step": 11959 }, { "epoch": 0.604987606859224, "grad_norm": 7.671994209289551, "learning_rate": 3.4401139129483503e-06, "loss": 0.4333, "step": 11960 }, { "epoch": 0.6050381911072892, "grad_norm": 9.763909339904785, "learning_rate": 3.4393513776213636e-06, "loss": 0.4512, "step": 11961 }, { "epoch": 0.6050887753553543, "grad_norm": 6.452784061431885, "learning_rate": 3.4385888825085235e-06, "loss": 0.3666, "step": 11962 }, { "epoch": 0.6051393596034195, "grad_norm": 5.727955341339111, "learning_rate": 3.437826427629476e-06, "loss": 0.4342, "step": 11963 }, { "epoch": 0.6051899438514846, "grad_norm": 12.610010147094727, "learning_rate": 3.437064013003867e-06, "loss": 0.4288, "step": 11964 }, { "epoch": 0.6052405280995498, "grad_norm": 5.830944538116455, "learning_rate": 3.436301638651345e-06, "loss": 0.4165, "step": 11965 }, { "epoch": 0.6052911123476149, "grad_norm": 4.962033271789551, "learning_rate": 3.435539304591554e-06, "loss": 0.3348, "step": 11966 }, { "epoch": 0.6053416965956802, "grad_norm": 5.5975022315979, "learning_rate": 3.4347770108441335e-06, "loss": 0.4684, "step": 11967 }, { "epoch": 0.6053922808437453, "grad_norm": 7.039947032928467, "learning_rate": 3.4340147574287297e-06, "loss": 0.4195, "step": 11968 }, { "epoch": 0.6054428650918104, "grad_norm": 6.241726875305176, "learning_rate": 3.433252544364984e-06, "loss": 0.242, "step": 11969 }, { "epoch": 0.6054934493398756, "grad_norm": 9.104084014892578, "learning_rate": 3.4324903716725345e-06, "loss": 0.3701, "step": 11970 }, { "epoch": 0.6055440335879407, "grad_norm": 9.759550094604492, "learning_rate": 3.431728239371022e-06, "loss": 0.4906, "step": 11971 }, { "epoch": 0.6055946178360059, "grad_norm": 7.62793493270874, "learning_rate": 3.430966147480086e-06, "loss": 0.4384, "step": 11972 }, { "epoch": 0.605645202084071, "grad_norm": 8.046241760253906, "learning_rate": 3.4302040960193617e-06, "loss": 0.4006, "step": 11973 }, { "epoch": 0.6056957863321362, "grad_norm": 5.650490760803223, "learning_rate": 3.4294420850084854e-06, "loss": 0.474, "step": 11974 }, { "epoch": 0.6057463705802013, "grad_norm": 11.378294944763184, "learning_rate": 3.4286801144670946e-06, "loss": 0.5233, "step": 11975 }, { "epoch": 0.6057969548282665, "grad_norm": 7.850872039794922, "learning_rate": 3.4279181844148214e-06, "loss": 0.4179, "step": 11976 }, { "epoch": 0.6058475390763316, "grad_norm": 9.04237174987793, "learning_rate": 3.4271562948712992e-06, "loss": 0.3537, "step": 11977 }, { "epoch": 0.6058981233243967, "grad_norm": 8.925820350646973, "learning_rate": 3.426394445856161e-06, "loss": 0.5193, "step": 11978 }, { "epoch": 0.6059487075724619, "grad_norm": 9.413951873779297, "learning_rate": 3.425632637389037e-06, "loss": 0.4477, "step": 11979 }, { "epoch": 0.6059992918205271, "grad_norm": 8.984582901000977, "learning_rate": 3.424870869489556e-06, "loss": 0.4122, "step": 11980 }, { "epoch": 0.6060498760685923, "grad_norm": 8.065208435058594, "learning_rate": 3.4241091421773482e-06, "loss": 0.4017, "step": 11981 }, { "epoch": 0.6061004603166574, "grad_norm": 5.974669456481934, "learning_rate": 3.4233474554720423e-06, "loss": 0.3216, "step": 11982 }, { "epoch": 0.6061510445647226, "grad_norm": 6.017926216125488, "learning_rate": 3.4225858093932627e-06, "loss": 0.4366, "step": 11983 }, { "epoch": 0.6062016288127877, "grad_norm": 40.64524841308594, "learning_rate": 3.4218242039606385e-06, "loss": 0.5578, "step": 11984 }, { "epoch": 0.6062522130608529, "grad_norm": 9.712328910827637, "learning_rate": 3.4210626391937917e-06, "loss": 0.4351, "step": 11985 }, { "epoch": 0.606302797308918, "grad_norm": 6.610507965087891, "learning_rate": 3.4203011151123467e-06, "loss": 0.3744, "step": 11986 }, { "epoch": 0.6063533815569832, "grad_norm": 10.889909744262695, "learning_rate": 3.4195396317359275e-06, "loss": 0.4439, "step": 11987 }, { "epoch": 0.6064039658050483, "grad_norm": 6.001256942749023, "learning_rate": 3.4187781890841533e-06, "loss": 0.3642, "step": 11988 }, { "epoch": 0.6064545500531134, "grad_norm": 8.798335075378418, "learning_rate": 3.4180167871766455e-06, "loss": 0.3947, "step": 11989 }, { "epoch": 0.6065051343011786, "grad_norm": 9.75931167602539, "learning_rate": 3.417255426033026e-06, "loss": 0.434, "step": 11990 }, { "epoch": 0.6065557185492437, "grad_norm": 8.741473197937012, "learning_rate": 3.4164941056729117e-06, "loss": 0.4906, "step": 11991 }, { "epoch": 0.6066063027973089, "grad_norm": 7.95134162902832, "learning_rate": 3.4157328261159173e-06, "loss": 0.413, "step": 11992 }, { "epoch": 0.606656887045374, "grad_norm": 10.590519905090332, "learning_rate": 3.4149715873816615e-06, "loss": 0.4926, "step": 11993 }, { "epoch": 0.6067074712934393, "grad_norm": 6.513317108154297, "learning_rate": 3.4142103894897605e-06, "loss": 0.3252, "step": 11994 }, { "epoch": 0.6067580555415044, "grad_norm": 19.543102264404297, "learning_rate": 3.4134492324598263e-06, "loss": 0.4445, "step": 11995 }, { "epoch": 0.6068086397895696, "grad_norm": 4.852777481079102, "learning_rate": 3.4126881163114735e-06, "loss": 0.3466, "step": 11996 }, { "epoch": 0.6068592240376347, "grad_norm": 10.599191665649414, "learning_rate": 3.4119270410643144e-06, "loss": 0.4307, "step": 11997 }, { "epoch": 0.6069098082856998, "grad_norm": 9.274530410766602, "learning_rate": 3.4111660067379586e-06, "loss": 0.4552, "step": 11998 }, { "epoch": 0.606960392533765, "grad_norm": 6.633706569671631, "learning_rate": 3.4104050133520174e-06, "loss": 0.4616, "step": 11999 }, { "epoch": 0.6070109767818301, "grad_norm": 8.980658531188965, "learning_rate": 3.4096440609261006e-06, "loss": 0.5003, "step": 12000 }, { "epoch": 0.6070615610298953, "grad_norm": 6.132431983947754, "learning_rate": 3.4088831494798135e-06, "loss": 0.4118, "step": 12001 }, { "epoch": 0.6071121452779604, "grad_norm": 9.538656234741211, "learning_rate": 3.408122279032765e-06, "loss": 0.4327, "step": 12002 }, { "epoch": 0.6071627295260256, "grad_norm": 13.714798927307129, "learning_rate": 3.4073614496045613e-06, "loss": 0.3631, "step": 12003 }, { "epoch": 0.6072133137740907, "grad_norm": 5.915823936462402, "learning_rate": 3.4066006612148074e-06, "loss": 0.4357, "step": 12004 }, { "epoch": 0.6072638980221559, "grad_norm": 9.006443977355957, "learning_rate": 3.4058399138831033e-06, "loss": 0.5164, "step": 12005 }, { "epoch": 0.607314482270221, "grad_norm": 7.430482387542725, "learning_rate": 3.4050792076290544e-06, "loss": 0.3483, "step": 12006 }, { "epoch": 0.6073650665182863, "grad_norm": 6.892714023590088, "learning_rate": 3.404318542472263e-06, "loss": 0.4198, "step": 12007 }, { "epoch": 0.6074156507663514, "grad_norm": 5.611253261566162, "learning_rate": 3.4035579184323277e-06, "loss": 0.3058, "step": 12008 }, { "epoch": 0.6074662350144165, "grad_norm": 11.458163261413574, "learning_rate": 3.402797335528848e-06, "loss": 0.384, "step": 12009 }, { "epoch": 0.6075168192624817, "grad_norm": 4.735171318054199, "learning_rate": 3.402036793781425e-06, "loss": 0.3751, "step": 12010 }, { "epoch": 0.6075674035105468, "grad_norm": 10.778695106506348, "learning_rate": 3.4012762932096523e-06, "loss": 0.5189, "step": 12011 }, { "epoch": 0.607617987758612, "grad_norm": 6.927302837371826, "learning_rate": 3.400515833833128e-06, "loss": 0.3693, "step": 12012 }, { "epoch": 0.6076685720066771, "grad_norm": 10.8283109664917, "learning_rate": 3.3997554156714483e-06, "loss": 0.3331, "step": 12013 }, { "epoch": 0.6077191562547423, "grad_norm": 6.528535842895508, "learning_rate": 3.398995038744206e-06, "loss": 0.3951, "step": 12014 }, { "epoch": 0.6077697405028074, "grad_norm": 6.0701189041137695, "learning_rate": 3.398234703070995e-06, "loss": 0.3739, "step": 12015 }, { "epoch": 0.6078203247508726, "grad_norm": 3.878655195236206, "learning_rate": 3.397474408671406e-06, "loss": 0.2612, "step": 12016 }, { "epoch": 0.6078709089989377, "grad_norm": 7.089301109313965, "learning_rate": 3.396714155565033e-06, "loss": 0.4284, "step": 12017 }, { "epoch": 0.6079214932470028, "grad_norm": 13.514142990112305, "learning_rate": 3.3959539437714607e-06, "loss": 0.5567, "step": 12018 }, { "epoch": 0.607972077495068, "grad_norm": 8.953804969787598, "learning_rate": 3.3951937733102824e-06, "loss": 0.4133, "step": 12019 }, { "epoch": 0.6080226617431332, "grad_norm": 9.383489608764648, "learning_rate": 3.3944336442010832e-06, "loss": 0.4202, "step": 12020 }, { "epoch": 0.6080732459911984, "grad_norm": 9.576565742492676, "learning_rate": 3.3936735564634515e-06, "loss": 0.3403, "step": 12021 }, { "epoch": 0.6081238302392635, "grad_norm": 10.854558944702148, "learning_rate": 3.392913510116973e-06, "loss": 0.5414, "step": 12022 }, { "epoch": 0.6081744144873287, "grad_norm": 8.950464248657227, "learning_rate": 3.39215350518123e-06, "loss": 0.4603, "step": 12023 }, { "epoch": 0.6082249987353938, "grad_norm": 4.958380222320557, "learning_rate": 3.3913935416758082e-06, "loss": 0.4122, "step": 12024 }, { "epoch": 0.608275582983459, "grad_norm": 7.333427429199219, "learning_rate": 3.3906336196202897e-06, "loss": 0.3953, "step": 12025 }, { "epoch": 0.6083261672315241, "grad_norm": 6.750927448272705, "learning_rate": 3.389873739034255e-06, "loss": 0.4625, "step": 12026 }, { "epoch": 0.6083767514795893, "grad_norm": 15.089495658874512, "learning_rate": 3.389113899937285e-06, "loss": 0.4459, "step": 12027 }, { "epoch": 0.6084273357276544, "grad_norm": 6.879055023193359, "learning_rate": 3.3883541023489604e-06, "loss": 0.4555, "step": 12028 }, { "epoch": 0.6084779199757195, "grad_norm": 8.814923286437988, "learning_rate": 3.387594346288856e-06, "loss": 0.4814, "step": 12029 }, { "epoch": 0.6085285042237847, "grad_norm": 11.079963684082031, "learning_rate": 3.3868346317765533e-06, "loss": 0.4242, "step": 12030 }, { "epoch": 0.6085790884718498, "grad_norm": 11.119450569152832, "learning_rate": 3.3860749588316234e-06, "loss": 0.512, "step": 12031 }, { "epoch": 0.608629672719915, "grad_norm": 5.791540145874023, "learning_rate": 3.3853153274736437e-06, "loss": 0.3214, "step": 12032 }, { "epoch": 0.6086802569679801, "grad_norm": 5.163051605224609, "learning_rate": 3.3845557377221884e-06, "loss": 0.2924, "step": 12033 }, { "epoch": 0.6087308412160454, "grad_norm": 4.574185371398926, "learning_rate": 3.3837961895968285e-06, "loss": 0.3632, "step": 12034 }, { "epoch": 0.6087814254641105, "grad_norm": 6.012628078460693, "learning_rate": 3.383036683117139e-06, "loss": 0.3948, "step": 12035 }, { "epoch": 0.6088320097121757, "grad_norm": 8.805425643920898, "learning_rate": 3.382277218302687e-06, "loss": 0.3976, "step": 12036 }, { "epoch": 0.6088825939602408, "grad_norm": 6.062141418457031, "learning_rate": 3.3815177951730438e-06, "loss": 0.3935, "step": 12037 }, { "epoch": 0.608933178208306, "grad_norm": 9.913395881652832, "learning_rate": 3.380758413747779e-06, "loss": 0.4011, "step": 12038 }, { "epoch": 0.6089837624563711, "grad_norm": 18.056081771850586, "learning_rate": 3.3799990740464573e-06, "loss": 0.7526, "step": 12039 }, { "epoch": 0.6090343467044362, "grad_norm": 4.858914852142334, "learning_rate": 3.3792397760886463e-06, "loss": 0.3588, "step": 12040 }, { "epoch": 0.6090849309525014, "grad_norm": 4.625796794891357, "learning_rate": 3.378480519893913e-06, "loss": 0.3386, "step": 12041 }, { "epoch": 0.6091355152005665, "grad_norm": 13.425228118896484, "learning_rate": 3.3777213054818194e-06, "loss": 0.4369, "step": 12042 }, { "epoch": 0.6091860994486317, "grad_norm": 5.920383930206299, "learning_rate": 3.3769621328719305e-06, "loss": 0.3537, "step": 12043 }, { "epoch": 0.6092366836966968, "grad_norm": 4.790688514709473, "learning_rate": 3.3762030020838055e-06, "loss": 0.304, "step": 12044 }, { "epoch": 0.609287267944762, "grad_norm": 6.451690673828125, "learning_rate": 3.3754439131370087e-06, "loss": 0.3503, "step": 12045 }, { "epoch": 0.6093378521928271, "grad_norm": 7.453122138977051, "learning_rate": 3.374684866051096e-06, "loss": 0.3588, "step": 12046 }, { "epoch": 0.6093884364408924, "grad_norm": 13.910057067871094, "learning_rate": 3.373925860845629e-06, "loss": 0.5152, "step": 12047 }, { "epoch": 0.6094390206889575, "grad_norm": 5.978339672088623, "learning_rate": 3.3731668975401667e-06, "loss": 0.4152, "step": 12048 }, { "epoch": 0.6094896049370226, "grad_norm": 7.972186088562012, "learning_rate": 3.3724079761542623e-06, "loss": 0.4303, "step": 12049 }, { "epoch": 0.6095401891850878, "grad_norm": 11.124462127685547, "learning_rate": 3.3716490967074744e-06, "loss": 0.3791, "step": 12050 }, { "epoch": 0.6095907734331529, "grad_norm": 10.593046188354492, "learning_rate": 3.370890259219355e-06, "loss": 0.4779, "step": 12051 }, { "epoch": 0.6096413576812181, "grad_norm": 9.020940780639648, "learning_rate": 3.3701314637094583e-06, "loss": 0.4809, "step": 12052 }, { "epoch": 0.6096919419292832, "grad_norm": 5.855514049530029, "learning_rate": 3.3693727101973385e-06, "loss": 0.3342, "step": 12053 }, { "epoch": 0.6097425261773484, "grad_norm": 10.252256393432617, "learning_rate": 3.3686139987025435e-06, "loss": 0.5288, "step": 12054 }, { "epoch": 0.6097931104254135, "grad_norm": 6.153395652770996, "learning_rate": 3.3678553292446257e-06, "loss": 0.3144, "step": 12055 }, { "epoch": 0.6098436946734787, "grad_norm": 6.511349201202393, "learning_rate": 3.3670967018431367e-06, "loss": 0.3995, "step": 12056 }, { "epoch": 0.6098942789215438, "grad_norm": 9.17526912689209, "learning_rate": 3.3663381165176193e-06, "loss": 0.3494, "step": 12057 }, { "epoch": 0.6099448631696089, "grad_norm": 7.161346435546875, "learning_rate": 3.3655795732876227e-06, "loss": 0.3882, "step": 12058 }, { "epoch": 0.6099954474176741, "grad_norm": 8.527320861816406, "learning_rate": 3.364821072172692e-06, "loss": 0.4455, "step": 12059 }, { "epoch": 0.6100460316657393, "grad_norm": 11.887458801269531, "learning_rate": 3.3640626131923742e-06, "loss": 0.4246, "step": 12060 }, { "epoch": 0.6100966159138045, "grad_norm": 8.501938819885254, "learning_rate": 3.36330419636621e-06, "loss": 0.3186, "step": 12061 }, { "epoch": 0.6101472001618696, "grad_norm": 6.486177444458008, "learning_rate": 3.3625458217137435e-06, "loss": 0.3744, "step": 12062 }, { "epoch": 0.6101977844099348, "grad_norm": 9.244559288024902, "learning_rate": 3.3617874892545173e-06, "loss": 0.4584, "step": 12063 }, { "epoch": 0.6102483686579999, "grad_norm": 5.240727424621582, "learning_rate": 3.361029199008069e-06, "loss": 0.3366, "step": 12064 }, { "epoch": 0.6102989529060651, "grad_norm": 5.993310451507568, "learning_rate": 3.3602709509939403e-06, "loss": 0.4548, "step": 12065 }, { "epoch": 0.6103495371541302, "grad_norm": 10.420827865600586, "learning_rate": 3.3595127452316687e-06, "loss": 0.4374, "step": 12066 }, { "epoch": 0.6104001214021954, "grad_norm": 5.817418098449707, "learning_rate": 3.3587545817407907e-06, "loss": 0.3696, "step": 12067 }, { "epoch": 0.6104507056502605, "grad_norm": 7.878776550292969, "learning_rate": 3.357996460540845e-06, "loss": 0.3511, "step": 12068 }, { "epoch": 0.6105012898983256, "grad_norm": 9.426056861877441, "learning_rate": 3.357238381651362e-06, "loss": 0.323, "step": 12069 }, { "epoch": 0.6105518741463908, "grad_norm": 7.337830543518066, "learning_rate": 3.356480345091879e-06, "loss": 0.4119, "step": 12070 }, { "epoch": 0.6106024583944559, "grad_norm": 7.648326873779297, "learning_rate": 3.355722350881927e-06, "loss": 0.3355, "step": 12071 }, { "epoch": 0.6106530426425211, "grad_norm": 9.27066421508789, "learning_rate": 3.3549643990410385e-06, "loss": 0.4281, "step": 12072 }, { "epoch": 0.6107036268905863, "grad_norm": 9.386878967285156, "learning_rate": 3.354206489588745e-06, "loss": 0.3431, "step": 12073 }, { "epoch": 0.6107542111386515, "grad_norm": 9.082076072692871, "learning_rate": 3.353448622544574e-06, "loss": 0.4476, "step": 12074 }, { "epoch": 0.6108047953867166, "grad_norm": 8.673860549926758, "learning_rate": 3.3526907979280547e-06, "loss": 0.397, "step": 12075 }, { "epoch": 0.6108553796347818, "grad_norm": 7.134861469268799, "learning_rate": 3.351933015758716e-06, "loss": 0.2722, "step": 12076 }, { "epoch": 0.6109059638828469, "grad_norm": 8.698259353637695, "learning_rate": 3.351175276056081e-06, "loss": 0.4339, "step": 12077 }, { "epoch": 0.610956548130912, "grad_norm": 9.389533042907715, "learning_rate": 3.3504175788396775e-06, "loss": 0.4318, "step": 12078 }, { "epoch": 0.6110071323789772, "grad_norm": 11.077926635742188, "learning_rate": 3.349659924129029e-06, "loss": 0.4425, "step": 12079 }, { "epoch": 0.6110577166270423, "grad_norm": 6.522632598876953, "learning_rate": 3.348902311943657e-06, "loss": 0.2561, "step": 12080 }, { "epoch": 0.6111083008751075, "grad_norm": 11.879446029663086, "learning_rate": 3.3481447423030867e-06, "loss": 0.5169, "step": 12081 }, { "epoch": 0.6111588851231726, "grad_norm": 9.410961151123047, "learning_rate": 3.3473872152268344e-06, "loss": 0.4518, "step": 12082 }, { "epoch": 0.6112094693712378, "grad_norm": 7.790380477905273, "learning_rate": 3.3466297307344224e-06, "loss": 0.4536, "step": 12083 }, { "epoch": 0.6112600536193029, "grad_norm": 14.353248596191406, "learning_rate": 3.3458722888453684e-06, "loss": 0.5002, "step": 12084 }, { "epoch": 0.6113106378673681, "grad_norm": 7.826418876647949, "learning_rate": 3.3451148895791898e-06, "loss": 0.4891, "step": 12085 }, { "epoch": 0.6113612221154332, "grad_norm": 10.75424861907959, "learning_rate": 3.344357532955404e-06, "loss": 0.4698, "step": 12086 }, { "epoch": 0.6114118063634985, "grad_norm": 14.337512016296387, "learning_rate": 3.3436002189935245e-06, "loss": 0.4527, "step": 12087 }, { "epoch": 0.6114623906115636, "grad_norm": 8.944331169128418, "learning_rate": 3.342842947713069e-06, "loss": 0.4721, "step": 12088 }, { "epoch": 0.6115129748596287, "grad_norm": 5.462854862213135, "learning_rate": 3.3420857191335455e-06, "loss": 0.4244, "step": 12089 }, { "epoch": 0.6115635591076939, "grad_norm": 11.385384559631348, "learning_rate": 3.3413285332744685e-06, "loss": 0.4056, "step": 12090 }, { "epoch": 0.611614143355759, "grad_norm": 4.916652202606201, "learning_rate": 3.34057139015535e-06, "loss": 0.3985, "step": 12091 }, { "epoch": 0.6116647276038242, "grad_norm": 7.929208278656006, "learning_rate": 3.3398142897956975e-06, "loss": 0.4007, "step": 12092 }, { "epoch": 0.6117153118518893, "grad_norm": 7.841013431549072, "learning_rate": 3.3390572322150212e-06, "loss": 0.3181, "step": 12093 }, { "epoch": 0.6117658960999545, "grad_norm": 8.512808799743652, "learning_rate": 3.33830021743283e-06, "loss": 0.5119, "step": 12094 }, { "epoch": 0.6118164803480196, "grad_norm": 6.621626853942871, "learning_rate": 3.337543245468627e-06, "loss": 0.3604, "step": 12095 }, { "epoch": 0.6118670645960848, "grad_norm": 14.225420951843262, "learning_rate": 3.336786316341919e-06, "loss": 0.5412, "step": 12096 }, { "epoch": 0.6119176488441499, "grad_norm": 10.401052474975586, "learning_rate": 3.3360294300722097e-06, "loss": 0.4194, "step": 12097 }, { "epoch": 0.611968233092215, "grad_norm": 7.111323833465576, "learning_rate": 3.3352725866790044e-06, "loss": 0.339, "step": 12098 }, { "epoch": 0.6120188173402802, "grad_norm": 7.773711681365967, "learning_rate": 3.3345157861818024e-06, "loss": 0.3841, "step": 12099 }, { "epoch": 0.6120694015883454, "grad_norm": 8.680102348327637, "learning_rate": 3.3337590286001053e-06, "loss": 0.3178, "step": 12100 }, { "epoch": 0.6121199858364106, "grad_norm": 5.8011698722839355, "learning_rate": 3.333002313953414e-06, "loss": 0.3689, "step": 12101 }, { "epoch": 0.6121705700844757, "grad_norm": 5.8145928382873535, "learning_rate": 3.3322456422612265e-06, "loss": 0.4643, "step": 12102 }, { "epoch": 0.6122211543325409, "grad_norm": 13.601644515991211, "learning_rate": 3.33148901354304e-06, "loss": 0.521, "step": 12103 }, { "epoch": 0.612271738580606, "grad_norm": 9.453903198242188, "learning_rate": 3.330732427818353e-06, "loss": 0.4021, "step": 12104 }, { "epoch": 0.6123223228286712, "grad_norm": 8.212993621826172, "learning_rate": 3.3299758851066575e-06, "loss": 0.4198, "step": 12105 }, { "epoch": 0.6123729070767363, "grad_norm": 6.987191677093506, "learning_rate": 3.3292193854274507e-06, "loss": 0.4089, "step": 12106 }, { "epoch": 0.6124234913248015, "grad_norm": 7.044128894805908, "learning_rate": 3.328462928800226e-06, "loss": 0.3913, "step": 12107 }, { "epoch": 0.6124740755728666, "grad_norm": 5.402071952819824, "learning_rate": 3.3277065152444744e-06, "loss": 0.3382, "step": 12108 }, { "epoch": 0.6125246598209317, "grad_norm": 5.657069206237793, "learning_rate": 3.326950144779684e-06, "loss": 0.3743, "step": 12109 }, { "epoch": 0.6125752440689969, "grad_norm": 9.666288375854492, "learning_rate": 3.326193817425348e-06, "loss": 0.4059, "step": 12110 }, { "epoch": 0.612625828317062, "grad_norm": 7.810557842254639, "learning_rate": 3.3254375332009557e-06, "loss": 0.3984, "step": 12111 }, { "epoch": 0.6126764125651272, "grad_norm": 7.320105075836182, "learning_rate": 3.324681292125992e-06, "loss": 0.4013, "step": 12112 }, { "epoch": 0.6127269968131924, "grad_norm": 8.99299430847168, "learning_rate": 3.323925094219945e-06, "loss": 0.4582, "step": 12113 }, { "epoch": 0.6127775810612576, "grad_norm": 9.935751914978027, "learning_rate": 3.323168939502301e-06, "loss": 0.4571, "step": 12114 }, { "epoch": 0.6128281653093227, "grad_norm": 5.912107944488525, "learning_rate": 3.322412827992542e-06, "loss": 0.4049, "step": 12115 }, { "epoch": 0.6128787495573879, "grad_norm": 5.907578945159912, "learning_rate": 3.3216567597101523e-06, "loss": 0.415, "step": 12116 }, { "epoch": 0.612929333805453, "grad_norm": 8.462113380432129, "learning_rate": 3.3209007346746157e-06, "loss": 0.4033, "step": 12117 }, { "epoch": 0.6129799180535181, "grad_norm": 7.565276622772217, "learning_rate": 3.32014475290541e-06, "loss": 0.4056, "step": 12118 }, { "epoch": 0.6130305023015833, "grad_norm": 9.257014274597168, "learning_rate": 3.3193888144220174e-06, "loss": 0.428, "step": 12119 }, { "epoch": 0.6130810865496484, "grad_norm": 8.72403335571289, "learning_rate": 3.318632919243916e-06, "loss": 0.4169, "step": 12120 }, { "epoch": 0.6131316707977136, "grad_norm": 4.235568523406982, "learning_rate": 3.3178770673905825e-06, "loss": 0.2954, "step": 12121 }, { "epoch": 0.6131822550457787, "grad_norm": 9.646224975585938, "learning_rate": 3.317121258881494e-06, "loss": 0.4213, "step": 12122 }, { "epoch": 0.6132328392938439, "grad_norm": 11.033021926879883, "learning_rate": 3.3163654937361256e-06, "loss": 0.451, "step": 12123 }, { "epoch": 0.613283423541909, "grad_norm": 6.294144153594971, "learning_rate": 3.315609771973951e-06, "loss": 0.3913, "step": 12124 }, { "epoch": 0.6133340077899742, "grad_norm": 11.22424602508545, "learning_rate": 3.3148540936144445e-06, "loss": 0.4643, "step": 12125 }, { "epoch": 0.6133845920380393, "grad_norm": 6.0854573249816895, "learning_rate": 3.314098458677079e-06, "loss": 0.4118, "step": 12126 }, { "epoch": 0.6134351762861046, "grad_norm": 5.984163761138916, "learning_rate": 3.313342867181323e-06, "loss": 0.4527, "step": 12127 }, { "epoch": 0.6134857605341697, "grad_norm": 7.712687969207764, "learning_rate": 3.3125873191466474e-06, "loss": 0.3949, "step": 12128 }, { "epoch": 0.6135363447822348, "grad_norm": 6.690882682800293, "learning_rate": 3.311831814592521e-06, "loss": 0.4196, "step": 12129 }, { "epoch": 0.6135869290303, "grad_norm": 7.489247798919678, "learning_rate": 3.3110763535384106e-06, "loss": 0.3617, "step": 12130 }, { "epoch": 0.6136375132783651, "grad_norm": 7.894377708435059, "learning_rate": 3.3103209360037837e-06, "loss": 0.4501, "step": 12131 }, { "epoch": 0.6136880975264303, "grad_norm": 7.597317218780518, "learning_rate": 3.309565562008105e-06, "loss": 0.3981, "step": 12132 }, { "epoch": 0.6137386817744954, "grad_norm": 9.602899551391602, "learning_rate": 3.3088102315708404e-06, "loss": 0.3614, "step": 12133 }, { "epoch": 0.6137892660225606, "grad_norm": 7.206557750701904, "learning_rate": 3.308054944711449e-06, "loss": 0.3831, "step": 12134 }, { "epoch": 0.6138398502706257, "grad_norm": 8.065217971801758, "learning_rate": 3.307299701449395e-06, "loss": 0.3608, "step": 12135 }, { "epoch": 0.6138904345186909, "grad_norm": 4.223258018493652, "learning_rate": 3.3065445018041405e-06, "loss": 0.3359, "step": 12136 }, { "epoch": 0.613941018766756, "grad_norm": 7.458527565002441, "learning_rate": 3.305789345795142e-06, "loss": 0.3811, "step": 12137 }, { "epoch": 0.6139916030148211, "grad_norm": 5.456358432769775, "learning_rate": 3.3050342334418608e-06, "loss": 0.4422, "step": 12138 }, { "epoch": 0.6140421872628863, "grad_norm": 14.205037117004395, "learning_rate": 3.304279164763754e-06, "loss": 0.454, "step": 12139 }, { "epoch": 0.6140927715109515, "grad_norm": 4.768188953399658, "learning_rate": 3.303524139780276e-06, "loss": 0.2983, "step": 12140 }, { "epoch": 0.6141433557590167, "grad_norm": 7.028531551361084, "learning_rate": 3.3027691585108834e-06, "loss": 0.3083, "step": 12141 }, { "epoch": 0.6141939400070818, "grad_norm": 9.210467338562012, "learning_rate": 3.3020142209750304e-06, "loss": 0.4365, "step": 12142 }, { "epoch": 0.614244524255147, "grad_norm": 18.36562728881836, "learning_rate": 3.30125932719217e-06, "loss": 0.6165, "step": 12143 }, { "epoch": 0.6142951085032121, "grad_norm": 8.457381248474121, "learning_rate": 3.300504477181753e-06, "loss": 0.4083, "step": 12144 }, { "epoch": 0.6143456927512773, "grad_norm": 12.599651336669922, "learning_rate": 3.2997496709632314e-06, "loss": 0.6391, "step": 12145 }, { "epoch": 0.6143962769993424, "grad_norm": 12.194541931152344, "learning_rate": 3.2989949085560554e-06, "loss": 0.4725, "step": 12146 }, { "epoch": 0.6144468612474075, "grad_norm": 9.549396514892578, "learning_rate": 3.2982401899796694e-06, "loss": 0.4337, "step": 12147 }, { "epoch": 0.6144974454954727, "grad_norm": 8.981765747070312, "learning_rate": 3.2974855152535236e-06, "loss": 0.3882, "step": 12148 }, { "epoch": 0.6145480297435378, "grad_norm": 5.932254314422607, "learning_rate": 3.296730884397065e-06, "loss": 0.3786, "step": 12149 }, { "epoch": 0.614598613991603, "grad_norm": 5.7715935707092285, "learning_rate": 3.295976297429736e-06, "loss": 0.3878, "step": 12150 }, { "epoch": 0.6146491982396681, "grad_norm": 9.467992782592773, "learning_rate": 3.2952217543709824e-06, "loss": 0.4241, "step": 12151 }, { "epoch": 0.6146997824877333, "grad_norm": 6.833225250244141, "learning_rate": 3.2944672552402472e-06, "loss": 0.362, "step": 12152 }, { "epoch": 0.6147503667357985, "grad_norm": 6.697803974151611, "learning_rate": 3.2937128000569708e-06, "loss": 0.3116, "step": 12153 }, { "epoch": 0.6148009509838637, "grad_norm": 6.245572566986084, "learning_rate": 3.2929583888405945e-06, "loss": 0.36, "step": 12154 }, { "epoch": 0.6148515352319288, "grad_norm": 13.122684478759766, "learning_rate": 3.2922040216105567e-06, "loss": 0.3647, "step": 12155 }, { "epoch": 0.614902119479994, "grad_norm": 8.318031311035156, "learning_rate": 3.2914496983862957e-06, "loss": 0.5408, "step": 12156 }, { "epoch": 0.6149527037280591, "grad_norm": 6.299196243286133, "learning_rate": 3.2906954191872507e-06, "loss": 0.3709, "step": 12157 }, { "epoch": 0.6150032879761242, "grad_norm": 6.235662460327148, "learning_rate": 3.2899411840328555e-06, "loss": 0.3555, "step": 12158 }, { "epoch": 0.6150538722241894, "grad_norm": 5.750749111175537, "learning_rate": 3.289186992942547e-06, "loss": 0.4304, "step": 12159 }, { "epoch": 0.6151044564722545, "grad_norm": 10.030189514160156, "learning_rate": 3.288432845935755e-06, "loss": 0.4836, "step": 12160 }, { "epoch": 0.6151550407203197, "grad_norm": 6.52614688873291, "learning_rate": 3.2876787430319157e-06, "loss": 0.3749, "step": 12161 }, { "epoch": 0.6152056249683848, "grad_norm": 6.006502628326416, "learning_rate": 3.2869246842504587e-06, "loss": 0.3216, "step": 12162 }, { "epoch": 0.61525620921645, "grad_norm": 7.297877311706543, "learning_rate": 3.2861706696108143e-06, "loss": 0.3797, "step": 12163 }, { "epoch": 0.6153067934645151, "grad_norm": 5.5563645362854, "learning_rate": 3.2854166991324133e-06, "loss": 0.3323, "step": 12164 }, { "epoch": 0.6153573777125803, "grad_norm": 10.36800765991211, "learning_rate": 3.2846627728346812e-06, "loss": 0.4742, "step": 12165 }, { "epoch": 0.6154079619606454, "grad_norm": 7.821249485015869, "learning_rate": 3.2839088907370464e-06, "loss": 0.4528, "step": 12166 }, { "epoch": 0.6154585462087107, "grad_norm": 6.503544807434082, "learning_rate": 3.283155052858936e-06, "loss": 0.4759, "step": 12167 }, { "epoch": 0.6155091304567758, "grad_norm": 13.038447380065918, "learning_rate": 3.2824012592197717e-06, "loss": 0.384, "step": 12168 }, { "epoch": 0.6155597147048409, "grad_norm": 4.984539985656738, "learning_rate": 3.2816475098389784e-06, "loss": 0.3557, "step": 12169 }, { "epoch": 0.6156102989529061, "grad_norm": 6.670098304748535, "learning_rate": 3.280893804735979e-06, "loss": 0.3388, "step": 12170 }, { "epoch": 0.6156608832009712, "grad_norm": 5.3929033279418945, "learning_rate": 3.280140143930195e-06, "loss": 0.3182, "step": 12171 }, { "epoch": 0.6157114674490364, "grad_norm": 6.97678279876709, "learning_rate": 3.279386527441043e-06, "loss": 0.5102, "step": 12172 }, { "epoch": 0.6157620516971015, "grad_norm": 6.71592903137207, "learning_rate": 3.2786329552879447e-06, "loss": 0.3246, "step": 12173 }, { "epoch": 0.6158126359451667, "grad_norm": 8.784952163696289, "learning_rate": 3.277879427490318e-06, "loss": 0.5105, "step": 12174 }, { "epoch": 0.6158632201932318, "grad_norm": 6.425772190093994, "learning_rate": 3.2771259440675784e-06, "loss": 0.4102, "step": 12175 }, { "epoch": 0.615913804441297, "grad_norm": 3.760319232940674, "learning_rate": 3.2763725050391416e-06, "loss": 0.3725, "step": 12176 }, { "epoch": 0.6159643886893621, "grad_norm": 8.53166675567627, "learning_rate": 3.275619110424423e-06, "loss": 0.3403, "step": 12177 }, { "epoch": 0.6160149729374272, "grad_norm": 14.367924690246582, "learning_rate": 3.2748657602428334e-06, "loss": 0.5371, "step": 12178 }, { "epoch": 0.6160655571854924, "grad_norm": 9.483323097229004, "learning_rate": 3.274112454513787e-06, "loss": 0.5049, "step": 12179 }, { "epoch": 0.6161161414335576, "grad_norm": 6.370003700256348, "learning_rate": 3.2733591932566946e-06, "loss": 0.256, "step": 12180 }, { "epoch": 0.6161667256816228, "grad_norm": 12.062356948852539, "learning_rate": 3.272605976490964e-06, "loss": 0.43, "step": 12181 }, { "epoch": 0.6162173099296879, "grad_norm": 9.083231925964355, "learning_rate": 3.2718528042360053e-06, "loss": 0.3979, "step": 12182 }, { "epoch": 0.6162678941777531, "grad_norm": 9.194477081298828, "learning_rate": 3.2710996765112267e-06, "loss": 0.4423, "step": 12183 }, { "epoch": 0.6163184784258182, "grad_norm": 6.358370304107666, "learning_rate": 3.2703465933360347e-06, "loss": 0.3728, "step": 12184 }, { "epoch": 0.6163690626738834, "grad_norm": 8.25518798828125, "learning_rate": 3.269593554729831e-06, "loss": 0.4013, "step": 12185 }, { "epoch": 0.6164196469219485, "grad_norm": 5.764854431152344, "learning_rate": 3.268840560712021e-06, "loss": 0.335, "step": 12186 }, { "epoch": 0.6164702311700136, "grad_norm": 7.623055458068848, "learning_rate": 3.2680876113020098e-06, "loss": 0.436, "step": 12187 }, { "epoch": 0.6165208154180788, "grad_norm": 7.13407564163208, "learning_rate": 3.267334706519197e-06, "loss": 0.376, "step": 12188 }, { "epoch": 0.6165713996661439, "grad_norm": 22.744232177734375, "learning_rate": 3.2665818463829836e-06, "loss": 0.5585, "step": 12189 }, { "epoch": 0.6166219839142091, "grad_norm": 6.940526962280273, "learning_rate": 3.26582903091277e-06, "loss": 0.4646, "step": 12190 }, { "epoch": 0.6166725681622742, "grad_norm": 12.844735145568848, "learning_rate": 3.265076260127952e-06, "loss": 0.4553, "step": 12191 }, { "epoch": 0.6167231524103394, "grad_norm": 7.557556629180908, "learning_rate": 3.26432353404793e-06, "loss": 0.3897, "step": 12192 }, { "epoch": 0.6167737366584046, "grad_norm": 5.598568916320801, "learning_rate": 3.263570852692097e-06, "loss": 0.2864, "step": 12193 }, { "epoch": 0.6168243209064698, "grad_norm": 6.197070598602295, "learning_rate": 3.2628182160798493e-06, "loss": 0.4482, "step": 12194 }, { "epoch": 0.6168749051545349, "grad_norm": 5.856156826019287, "learning_rate": 3.2620656242305805e-06, "loss": 0.3598, "step": 12195 }, { "epoch": 0.6169254894026001, "grad_norm": 9.93928337097168, "learning_rate": 3.261313077163682e-06, "loss": 0.4147, "step": 12196 }, { "epoch": 0.6169760736506652, "grad_norm": 12.7765474319458, "learning_rate": 3.260560574898548e-06, "loss": 0.4769, "step": 12197 }, { "epoch": 0.6170266578987303, "grad_norm": 6.245941162109375, "learning_rate": 3.2598081174545647e-06, "loss": 0.3574, "step": 12198 }, { "epoch": 0.6170772421467955, "grad_norm": 13.249396324157715, "learning_rate": 3.2590557048511234e-06, "loss": 0.5083, "step": 12199 }, { "epoch": 0.6171278263948606, "grad_norm": 10.449814796447754, "learning_rate": 3.2583033371076112e-06, "loss": 0.4307, "step": 12200 }, { "epoch": 0.6171784106429258, "grad_norm": 6.567174434661865, "learning_rate": 3.2575510142434143e-06, "loss": 0.4205, "step": 12201 }, { "epoch": 0.6172289948909909, "grad_norm": 8.917777061462402, "learning_rate": 3.2567987362779208e-06, "loss": 0.4806, "step": 12202 }, { "epoch": 0.6172795791390561, "grad_norm": 12.214234352111816, "learning_rate": 3.256046503230512e-06, "loss": 0.4376, "step": 12203 }, { "epoch": 0.6173301633871212, "grad_norm": 6.243014812469482, "learning_rate": 3.2552943151205724e-06, "loss": 0.3861, "step": 12204 }, { "epoch": 0.6173807476351864, "grad_norm": 9.654217720031738, "learning_rate": 3.2545421719674854e-06, "loss": 0.3445, "step": 12205 }, { "epoch": 0.6174313318832516, "grad_norm": 9.089997291564941, "learning_rate": 3.253790073790629e-06, "loss": 0.416, "step": 12206 }, { "epoch": 0.6174819161313168, "grad_norm": 10.636730194091797, "learning_rate": 3.2530380206093854e-06, "loss": 0.4123, "step": 12207 }, { "epoch": 0.6175325003793819, "grad_norm": 6.686890602111816, "learning_rate": 3.252286012443133e-06, "loss": 0.4522, "step": 12208 }, { "epoch": 0.617583084627447, "grad_norm": 7.376917839050293, "learning_rate": 3.251534049311248e-06, "loss": 0.3823, "step": 12209 }, { "epoch": 0.6176336688755122, "grad_norm": 21.472768783569336, "learning_rate": 3.250782131233109e-06, "loss": 0.4897, "step": 12210 }, { "epoch": 0.6176842531235773, "grad_norm": 6.198086261749268, "learning_rate": 3.2500302582280874e-06, "loss": 0.2936, "step": 12211 }, { "epoch": 0.6177348373716425, "grad_norm": 15.846749305725098, "learning_rate": 3.24927843031556e-06, "loss": 0.5451, "step": 12212 }, { "epoch": 0.6177854216197076, "grad_norm": 7.3884782791137695, "learning_rate": 3.248526647514898e-06, "loss": 0.4403, "step": 12213 }, { "epoch": 0.6178360058677728, "grad_norm": 6.742948532104492, "learning_rate": 3.2477749098454736e-06, "loss": 0.3929, "step": 12214 }, { "epoch": 0.6178865901158379, "grad_norm": 13.525945663452148, "learning_rate": 3.2470232173266585e-06, "loss": 0.4583, "step": 12215 }, { "epoch": 0.617937174363903, "grad_norm": 5.1089186668396, "learning_rate": 3.24627156997782e-06, "loss": 0.4148, "step": 12216 }, { "epoch": 0.6179877586119682, "grad_norm": 10.18517780303955, "learning_rate": 3.245519967818327e-06, "loss": 0.4891, "step": 12217 }, { "epoch": 0.6180383428600333, "grad_norm": 7.348719596862793, "learning_rate": 3.2447684108675483e-06, "loss": 0.4206, "step": 12218 }, { "epoch": 0.6180889271080985, "grad_norm": 6.928554534912109, "learning_rate": 3.2440168991448462e-06, "loss": 0.3959, "step": 12219 }, { "epoch": 0.6181395113561637, "grad_norm": 5.660879135131836, "learning_rate": 3.243265432669588e-06, "loss": 0.3837, "step": 12220 }, { "epoch": 0.6181900956042289, "grad_norm": 9.189512252807617, "learning_rate": 3.2425140114611366e-06, "loss": 0.4014, "step": 12221 }, { "epoch": 0.618240679852294, "grad_norm": 9.680414199829102, "learning_rate": 3.2417626355388537e-06, "loss": 0.4354, "step": 12222 }, { "epoch": 0.6182912641003592, "grad_norm": 8.584362983703613, "learning_rate": 3.2410113049221024e-06, "loss": 0.3149, "step": 12223 }, { "epoch": 0.6183418483484243, "grad_norm": 6.214056015014648, "learning_rate": 3.240260019630239e-06, "loss": 0.3245, "step": 12224 }, { "epoch": 0.6183924325964895, "grad_norm": 6.9318742752075195, "learning_rate": 3.2395087796826262e-06, "loss": 0.4228, "step": 12225 }, { "epoch": 0.6184430168445546, "grad_norm": 9.725255966186523, "learning_rate": 3.2387575850986188e-06, "loss": 0.4783, "step": 12226 }, { "epoch": 0.6184936010926197, "grad_norm": 7.101161003112793, "learning_rate": 3.238006435897575e-06, "loss": 0.4164, "step": 12227 }, { "epoch": 0.6185441853406849, "grad_norm": 7.838691234588623, "learning_rate": 3.237255332098848e-06, "loss": 0.469, "step": 12228 }, { "epoch": 0.61859476958875, "grad_norm": 25.663896560668945, "learning_rate": 3.236504273721793e-06, "loss": 0.4803, "step": 12229 }, { "epoch": 0.6186453538368152, "grad_norm": 7.273881912231445, "learning_rate": 3.2357532607857656e-06, "loss": 0.3147, "step": 12230 }, { "epoch": 0.6186959380848803, "grad_norm": 8.421884536743164, "learning_rate": 3.2350022933101137e-06, "loss": 0.3638, "step": 12231 }, { "epoch": 0.6187465223329455, "grad_norm": 7.364358901977539, "learning_rate": 3.23425137131419e-06, "loss": 0.4347, "step": 12232 }, { "epoch": 0.6187971065810107, "grad_norm": 15.5269193649292, "learning_rate": 3.233500494817344e-06, "loss": 0.5685, "step": 12233 }, { "epoch": 0.6188476908290759, "grad_norm": 5.9424872398376465, "learning_rate": 3.232749663838922e-06, "loss": 0.4319, "step": 12234 }, { "epoch": 0.618898275077141, "grad_norm": 10.639778137207031, "learning_rate": 3.2319988783982737e-06, "loss": 0.4914, "step": 12235 }, { "epoch": 0.6189488593252062, "grad_norm": 14.662124633789062, "learning_rate": 3.231248138514745e-06, "loss": 0.508, "step": 12236 }, { "epoch": 0.6189994435732713, "grad_norm": 7.369888782501221, "learning_rate": 3.230497444207679e-06, "loss": 0.4168, "step": 12237 }, { "epoch": 0.6190500278213364, "grad_norm": 12.758049011230469, "learning_rate": 3.229746795496418e-06, "loss": 0.5744, "step": 12238 }, { "epoch": 0.6191006120694016, "grad_norm": 5.351385593414307, "learning_rate": 3.2289961924003066e-06, "loss": 0.305, "step": 12239 }, { "epoch": 0.6191511963174667, "grad_norm": 6.766417026519775, "learning_rate": 3.228245634938687e-06, "loss": 0.3667, "step": 12240 }, { "epoch": 0.6192017805655319, "grad_norm": 7.140470504760742, "learning_rate": 3.2274951231308965e-06, "loss": 0.3851, "step": 12241 }, { "epoch": 0.619252364813597, "grad_norm": 8.663803100585938, "learning_rate": 3.2267446569962757e-06, "loss": 0.4285, "step": 12242 }, { "epoch": 0.6193029490616622, "grad_norm": 7.748627185821533, "learning_rate": 3.225994236554163e-06, "loss": 0.3629, "step": 12243 }, { "epoch": 0.6193535333097273, "grad_norm": 9.626326560974121, "learning_rate": 3.2252438618238922e-06, "loss": 0.373, "step": 12244 }, { "epoch": 0.6194041175577925, "grad_norm": 7.129101753234863, "learning_rate": 3.2244935328248005e-06, "loss": 0.3919, "step": 12245 }, { "epoch": 0.6194547018058577, "grad_norm": 7.381924629211426, "learning_rate": 3.223743249576223e-06, "loss": 0.4806, "step": 12246 }, { "epoch": 0.6195052860539229, "grad_norm": 15.01070499420166, "learning_rate": 3.2229930120974917e-06, "loss": 0.5466, "step": 12247 }, { "epoch": 0.619555870301988, "grad_norm": 6.077045917510986, "learning_rate": 3.222242820407937e-06, "loss": 0.4181, "step": 12248 }, { "epoch": 0.6196064545500531, "grad_norm": 10.247421264648438, "learning_rate": 3.2214926745268938e-06, "loss": 0.4456, "step": 12249 }, { "epoch": 0.6196570387981183, "grad_norm": 7.125974655151367, "learning_rate": 3.2207425744736862e-06, "loss": 0.3928, "step": 12250 }, { "epoch": 0.6197076230461834, "grad_norm": 6.514589309692383, "learning_rate": 3.219992520267645e-06, "loss": 0.4417, "step": 12251 }, { "epoch": 0.6197582072942486, "grad_norm": 5.1505327224731445, "learning_rate": 3.2192425119280967e-06, "loss": 0.3962, "step": 12252 }, { "epoch": 0.6198087915423137, "grad_norm": 10.149476051330566, "learning_rate": 3.218492549474369e-06, "loss": 0.3925, "step": 12253 }, { "epoch": 0.6198593757903789, "grad_norm": 5.1641340255737305, "learning_rate": 3.217742632925783e-06, "loss": 0.3717, "step": 12254 }, { "epoch": 0.619909960038444, "grad_norm": 8.446938514709473, "learning_rate": 3.216992762301666e-06, "loss": 0.4282, "step": 12255 }, { "epoch": 0.6199605442865092, "grad_norm": 6.441927909851074, "learning_rate": 3.216242937621339e-06, "loss": 0.3705, "step": 12256 }, { "epoch": 0.6200111285345743, "grad_norm": 4.733137607574463, "learning_rate": 3.215493158904123e-06, "loss": 0.3943, "step": 12257 }, { "epoch": 0.6200617127826394, "grad_norm": 8.881302833557129, "learning_rate": 3.2147434261693376e-06, "loss": 0.3796, "step": 12258 }, { "epoch": 0.6201122970307046, "grad_norm": 9.65510082244873, "learning_rate": 3.213993739436302e-06, "loss": 0.4428, "step": 12259 }, { "epoch": 0.6201628812787698, "grad_norm": 9.086912155151367, "learning_rate": 3.213244098724333e-06, "loss": 0.4342, "step": 12260 }, { "epoch": 0.620213465526835, "grad_norm": 9.515865325927734, "learning_rate": 3.2124945040527504e-06, "loss": 0.4122, "step": 12261 }, { "epoch": 0.6202640497749001, "grad_norm": 6.208057403564453, "learning_rate": 3.2117449554408648e-06, "loss": 0.3454, "step": 12262 }, { "epoch": 0.6203146340229653, "grad_norm": 12.150116920471191, "learning_rate": 3.2109954529079933e-06, "loss": 0.5034, "step": 12263 }, { "epoch": 0.6203652182710304, "grad_norm": 6.982779502868652, "learning_rate": 3.2102459964734467e-06, "loss": 0.4048, "step": 12264 }, { "epoch": 0.6204158025190956, "grad_norm": 5.462303161621094, "learning_rate": 3.209496586156539e-06, "loss": 0.4557, "step": 12265 }, { "epoch": 0.6204663867671607, "grad_norm": 6.447343349456787, "learning_rate": 3.2087472219765786e-06, "loss": 0.4067, "step": 12266 }, { "epoch": 0.6205169710152258, "grad_norm": 4.234111785888672, "learning_rate": 3.2079979039528754e-06, "loss": 0.3892, "step": 12267 }, { "epoch": 0.620567555263291, "grad_norm": 6.380167484283447, "learning_rate": 3.2072486321047392e-06, "loss": 0.4583, "step": 12268 }, { "epoch": 0.6206181395113561, "grad_norm": 8.28424072265625, "learning_rate": 3.2064994064514744e-06, "loss": 0.4274, "step": 12269 }, { "epoch": 0.6206687237594213, "grad_norm": 5.964786052703857, "learning_rate": 3.2057502270123873e-06, "loss": 0.3105, "step": 12270 }, { "epoch": 0.6207193080074864, "grad_norm": 5.676056385040283, "learning_rate": 3.2050010938067843e-06, "loss": 0.2767, "step": 12271 }, { "epoch": 0.6207698922555516, "grad_norm": 4.827773571014404, "learning_rate": 3.204252006853967e-06, "loss": 0.2247, "step": 12272 }, { "epoch": 0.6208204765036168, "grad_norm": 5.3090972900390625, "learning_rate": 3.2035029661732376e-06, "loss": 0.3488, "step": 12273 }, { "epoch": 0.620871060751682, "grad_norm": 9.252399444580078, "learning_rate": 3.2027539717838994e-06, "loss": 0.434, "step": 12274 }, { "epoch": 0.6209216449997471, "grad_norm": 8.542974472045898, "learning_rate": 3.2020050237052495e-06, "loss": 0.3161, "step": 12275 }, { "epoch": 0.6209722292478123, "grad_norm": 7.425310134887695, "learning_rate": 3.201256121956586e-06, "loss": 0.3681, "step": 12276 }, { "epoch": 0.6210228134958774, "grad_norm": 17.675323486328125, "learning_rate": 3.200507266557208e-06, "loss": 0.4047, "step": 12277 }, { "epoch": 0.6210733977439425, "grad_norm": 6.269864082336426, "learning_rate": 3.1997584575264117e-06, "loss": 0.3766, "step": 12278 }, { "epoch": 0.6211239819920077, "grad_norm": 8.489294052124023, "learning_rate": 3.1990096948834914e-06, "loss": 0.3098, "step": 12279 }, { "epoch": 0.6211745662400728, "grad_norm": 10.572633743286133, "learning_rate": 3.19826097864774e-06, "loss": 0.4133, "step": 12280 }, { "epoch": 0.621225150488138, "grad_norm": 7.045102596282959, "learning_rate": 3.1975123088384534e-06, "loss": 0.4637, "step": 12281 }, { "epoch": 0.6212757347362031, "grad_norm": 6.722204685211182, "learning_rate": 3.1967636854749194e-06, "loss": 0.4052, "step": 12282 }, { "epoch": 0.6213263189842683, "grad_norm": 7.641253471374512, "learning_rate": 3.19601510857643e-06, "loss": 0.4001, "step": 12283 }, { "epoch": 0.6213769032323334, "grad_norm": 11.036076545715332, "learning_rate": 3.1952665781622744e-06, "loss": 0.5017, "step": 12284 }, { "epoch": 0.6214274874803986, "grad_norm": 8.455662727355957, "learning_rate": 3.194518094251739e-06, "loss": 0.4083, "step": 12285 }, { "epoch": 0.6214780717284638, "grad_norm": 10.281039237976074, "learning_rate": 3.193769656864112e-06, "loss": 0.3968, "step": 12286 }, { "epoch": 0.621528655976529, "grad_norm": 5.530038833618164, "learning_rate": 3.1930212660186792e-06, "loss": 0.3893, "step": 12287 }, { "epoch": 0.6215792402245941, "grad_norm": 10.10992431640625, "learning_rate": 3.192272921734724e-06, "loss": 0.3822, "step": 12288 }, { "epoch": 0.6216298244726592, "grad_norm": 8.35897445678711, "learning_rate": 3.191524624031527e-06, "loss": 0.3964, "step": 12289 }, { "epoch": 0.6216804087207244, "grad_norm": 10.662315368652344, "learning_rate": 3.1907763729283725e-06, "loss": 0.5988, "step": 12290 }, { "epoch": 0.6217309929687895, "grad_norm": 6.5705695152282715, "learning_rate": 3.190028168444542e-06, "loss": 0.3785, "step": 12291 }, { "epoch": 0.6217815772168547, "grad_norm": 5.771671772003174, "learning_rate": 3.189280010599313e-06, "loss": 0.437, "step": 12292 }, { "epoch": 0.6218321614649198, "grad_norm": 8.257856369018555, "learning_rate": 3.188531899411964e-06, "loss": 0.4302, "step": 12293 }, { "epoch": 0.621882745712985, "grad_norm": 6.895633697509766, "learning_rate": 3.1877838349017744e-06, "loss": 0.346, "step": 12294 }, { "epoch": 0.6219333299610501, "grad_norm": 8.840642929077148, "learning_rate": 3.187035817088017e-06, "loss": 0.337, "step": 12295 }, { "epoch": 0.6219839142091153, "grad_norm": 12.0940580368042, "learning_rate": 3.186287845989968e-06, "loss": 0.4459, "step": 12296 }, { "epoch": 0.6220344984571804, "grad_norm": 8.333904266357422, "learning_rate": 3.185539921626899e-06, "loss": 0.4478, "step": 12297 }, { "epoch": 0.6220850827052455, "grad_norm": 5.385551929473877, "learning_rate": 3.1847920440180845e-06, "loss": 0.3609, "step": 12298 }, { "epoch": 0.6221356669533107, "grad_norm": 6.618816375732422, "learning_rate": 3.1840442131827953e-06, "loss": 0.3621, "step": 12299 }, { "epoch": 0.6221862512013759, "grad_norm": 20.864389419555664, "learning_rate": 3.1832964291403014e-06, "loss": 0.519, "step": 12300 }, { "epoch": 0.6222368354494411, "grad_norm": 7.921503067016602, "learning_rate": 3.182548691909868e-06, "loss": 0.4405, "step": 12301 }, { "epoch": 0.6222874196975062, "grad_norm": 7.305840492248535, "learning_rate": 3.181801001510766e-06, "loss": 0.4873, "step": 12302 }, { "epoch": 0.6223380039455714, "grad_norm": 9.102079391479492, "learning_rate": 3.1810533579622604e-06, "loss": 0.3264, "step": 12303 }, { "epoch": 0.6223885881936365, "grad_norm": 9.414566993713379, "learning_rate": 3.180305761283616e-06, "loss": 0.4347, "step": 12304 }, { "epoch": 0.6224391724417017, "grad_norm": 9.139059066772461, "learning_rate": 3.1795582114940965e-06, "loss": 0.3664, "step": 12305 }, { "epoch": 0.6224897566897668, "grad_norm": 8.687432289123535, "learning_rate": 3.1788107086129662e-06, "loss": 0.4043, "step": 12306 }, { "epoch": 0.622540340937832, "grad_norm": 11.693238258361816, "learning_rate": 3.1780632526594836e-06, "loss": 0.4378, "step": 12307 }, { "epoch": 0.6225909251858971, "grad_norm": 11.558032035827637, "learning_rate": 3.1773158436529104e-06, "loss": 0.5375, "step": 12308 }, { "epoch": 0.6226415094339622, "grad_norm": 8.037166595458984, "learning_rate": 3.1765684816125066e-06, "loss": 0.4152, "step": 12309 }, { "epoch": 0.6226920936820274, "grad_norm": 5.84150505065918, "learning_rate": 3.1758211665575273e-06, "loss": 0.4621, "step": 12310 }, { "epoch": 0.6227426779300925, "grad_norm": 8.94779109954834, "learning_rate": 3.1750738985072303e-06, "loss": 0.4361, "step": 12311 }, { "epoch": 0.6227932621781577, "grad_norm": 14.61747932434082, "learning_rate": 3.1743266774808724e-06, "loss": 0.4835, "step": 12312 }, { "epoch": 0.6228438464262229, "grad_norm": 6.656270980834961, "learning_rate": 3.1735795034977075e-06, "loss": 0.4015, "step": 12313 }, { "epoch": 0.6228944306742881, "grad_norm": 14.030513763427734, "learning_rate": 3.1728323765769854e-06, "loss": 0.404, "step": 12314 }, { "epoch": 0.6229450149223532, "grad_norm": 6.61555290222168, "learning_rate": 3.172085296737959e-06, "loss": 0.3969, "step": 12315 }, { "epoch": 0.6229955991704184, "grad_norm": 7.817012786865234, "learning_rate": 3.17133826399988e-06, "loss": 0.2998, "step": 12316 }, { "epoch": 0.6230461834184835, "grad_norm": 8.273914337158203, "learning_rate": 3.170591278381997e-06, "loss": 0.4374, "step": 12317 }, { "epoch": 0.6230967676665486, "grad_norm": 7.734976768493652, "learning_rate": 3.1698443399035573e-06, "loss": 0.4075, "step": 12318 }, { "epoch": 0.6231473519146138, "grad_norm": 7.471263408660889, "learning_rate": 3.1690974485838098e-06, "loss": 0.3109, "step": 12319 }, { "epoch": 0.6231979361626789, "grad_norm": 7.309208393096924, "learning_rate": 3.1683506044419974e-06, "loss": 0.379, "step": 12320 }, { "epoch": 0.6232485204107441, "grad_norm": 9.419778823852539, "learning_rate": 3.167603807497366e-06, "loss": 0.4488, "step": 12321 }, { "epoch": 0.6232991046588092, "grad_norm": 6.5084428787231445, "learning_rate": 3.1668570577691594e-06, "loss": 0.4253, "step": 12322 }, { "epoch": 0.6233496889068744, "grad_norm": 4.556384563446045, "learning_rate": 3.1661103552766177e-06, "loss": 0.3365, "step": 12323 }, { "epoch": 0.6234002731549395, "grad_norm": 10.548491477966309, "learning_rate": 3.1653637000389824e-06, "loss": 0.4438, "step": 12324 }, { "epoch": 0.6234508574030047, "grad_norm": 7.3127055168151855, "learning_rate": 3.1646170920754943e-06, "loss": 0.3864, "step": 12325 }, { "epoch": 0.6235014416510699, "grad_norm": 7.985039710998535, "learning_rate": 3.163870531405392e-06, "loss": 0.3448, "step": 12326 }, { "epoch": 0.623552025899135, "grad_norm": 6.486395835876465, "learning_rate": 3.1631240180479085e-06, "loss": 0.4201, "step": 12327 }, { "epoch": 0.6236026101472002, "grad_norm": 10.760475158691406, "learning_rate": 3.162377552022282e-06, "loss": 0.4846, "step": 12328 }, { "epoch": 0.6236531943952653, "grad_norm": 6.777639389038086, "learning_rate": 3.161631133347749e-06, "loss": 0.4635, "step": 12329 }, { "epoch": 0.6237037786433305, "grad_norm": 5.921502590179443, "learning_rate": 3.16088476204354e-06, "loss": 0.3569, "step": 12330 }, { "epoch": 0.6237543628913956, "grad_norm": 4.870733261108398, "learning_rate": 3.160138438128889e-06, "loss": 0.3201, "step": 12331 }, { "epoch": 0.6238049471394608, "grad_norm": 8.933422088623047, "learning_rate": 3.1593921616230262e-06, "loss": 0.4577, "step": 12332 }, { "epoch": 0.6238555313875259, "grad_norm": 7.72914457321167, "learning_rate": 3.158645932545181e-06, "loss": 0.4641, "step": 12333 }, { "epoch": 0.6239061156355911, "grad_norm": 5.022200584411621, "learning_rate": 3.1578997509145847e-06, "loss": 0.4666, "step": 12334 }, { "epoch": 0.6239566998836562, "grad_norm": 5.790868282318115, "learning_rate": 3.15715361675046e-06, "loss": 0.378, "step": 12335 }, { "epoch": 0.6240072841317214, "grad_norm": 6.918194770812988, "learning_rate": 3.1564075300720363e-06, "loss": 0.4372, "step": 12336 }, { "epoch": 0.6240578683797865, "grad_norm": 11.630914688110352, "learning_rate": 3.1556614908985385e-06, "loss": 0.6275, "step": 12337 }, { "epoch": 0.6241084526278516, "grad_norm": 8.646805763244629, "learning_rate": 3.1549154992491883e-06, "loss": 0.3788, "step": 12338 }, { "epoch": 0.6241590368759169, "grad_norm": 8.783098220825195, "learning_rate": 3.1541695551432104e-06, "loss": 0.3625, "step": 12339 }, { "epoch": 0.624209621123982, "grad_norm": 12.15145206451416, "learning_rate": 3.153423658599823e-06, "loss": 0.3886, "step": 12340 }, { "epoch": 0.6242602053720472, "grad_norm": 12.699057579040527, "learning_rate": 3.152677809638249e-06, "loss": 0.4702, "step": 12341 }, { "epoch": 0.6243107896201123, "grad_norm": 6.556262016296387, "learning_rate": 3.151932008277704e-06, "loss": 0.381, "step": 12342 }, { "epoch": 0.6243613738681775, "grad_norm": 7.552707195281982, "learning_rate": 3.1511862545374084e-06, "loss": 0.4443, "step": 12343 }, { "epoch": 0.6244119581162426, "grad_norm": 22.06507682800293, "learning_rate": 3.1504405484365774e-06, "loss": 0.4502, "step": 12344 }, { "epoch": 0.6244625423643078, "grad_norm": 7.568631172180176, "learning_rate": 3.1496948899944257e-06, "loss": 0.4093, "step": 12345 }, { "epoch": 0.6245131266123729, "grad_norm": 8.949251174926758, "learning_rate": 3.148949279230167e-06, "loss": 0.4024, "step": 12346 }, { "epoch": 0.624563710860438, "grad_norm": 6.259746074676514, "learning_rate": 3.148203716163016e-06, "loss": 0.4168, "step": 12347 }, { "epoch": 0.6246142951085032, "grad_norm": 7.352405548095703, "learning_rate": 3.147458200812181e-06, "loss": 0.4653, "step": 12348 }, { "epoch": 0.6246648793565683, "grad_norm": 9.422810554504395, "learning_rate": 3.146712733196873e-06, "loss": 0.4435, "step": 12349 }, { "epoch": 0.6247154636046335, "grad_norm": 5.639313220977783, "learning_rate": 3.1459673133363034e-06, "loss": 0.2794, "step": 12350 }, { "epoch": 0.6247660478526986, "grad_norm": 8.650588035583496, "learning_rate": 3.1452219412496764e-06, "loss": 0.5741, "step": 12351 }, { "epoch": 0.6248166321007638, "grad_norm": 12.154268264770508, "learning_rate": 3.144476616956202e-06, "loss": 0.4989, "step": 12352 }, { "epoch": 0.624867216348829, "grad_norm": 6.6442036628723145, "learning_rate": 3.143731340475081e-06, "loss": 0.418, "step": 12353 }, { "epoch": 0.6249178005968942, "grad_norm": 7.545146465301514, "learning_rate": 3.142986111825521e-06, "loss": 0.4639, "step": 12354 }, { "epoch": 0.6249683848449593, "grad_norm": 5.0121541023254395, "learning_rate": 3.1422409310267227e-06, "loss": 0.3997, "step": 12355 }, { "epoch": 0.6250189690930245, "grad_norm": 7.307624340057373, "learning_rate": 3.1414957980978883e-06, "loss": 0.4238, "step": 12356 }, { "epoch": 0.6250695533410896, "grad_norm": 6.840898513793945, "learning_rate": 3.1407507130582194e-06, "loss": 0.4793, "step": 12357 }, { "epoch": 0.6251201375891547, "grad_norm": 5.13638162612915, "learning_rate": 3.140005675926912e-06, "loss": 0.314, "step": 12358 }, { "epoch": 0.6251707218372199, "grad_norm": 10.683555603027344, "learning_rate": 3.1392606867231663e-06, "loss": 0.4024, "step": 12359 }, { "epoch": 0.625221306085285, "grad_norm": 7.017210960388184, "learning_rate": 3.138515745466179e-06, "loss": 0.3624, "step": 12360 }, { "epoch": 0.6252718903333502, "grad_norm": 10.15817928314209, "learning_rate": 3.137770852175144e-06, "loss": 0.5202, "step": 12361 }, { "epoch": 0.6253224745814153, "grad_norm": 12.243061065673828, "learning_rate": 3.137026006869257e-06, "loss": 0.2983, "step": 12362 }, { "epoch": 0.6253730588294805, "grad_norm": 10.045759201049805, "learning_rate": 3.1362812095677084e-06, "loss": 0.5544, "step": 12363 }, { "epoch": 0.6254236430775456, "grad_norm": 5.427412033081055, "learning_rate": 3.135536460289694e-06, "loss": 0.4177, "step": 12364 }, { "epoch": 0.6254742273256108, "grad_norm": 5.160425662994385, "learning_rate": 3.1347917590543997e-06, "loss": 0.3491, "step": 12365 }, { "epoch": 0.625524811573676, "grad_norm": 14.160470008850098, "learning_rate": 3.1340471058810164e-06, "loss": 0.406, "step": 12366 }, { "epoch": 0.6255753958217412, "grad_norm": 7.750147342681885, "learning_rate": 3.133302500788733e-06, "loss": 0.3871, "step": 12367 }, { "epoch": 0.6256259800698063, "grad_norm": 4.577195644378662, "learning_rate": 3.1325579437967347e-06, "loss": 0.3813, "step": 12368 }, { "epoch": 0.6256765643178714, "grad_norm": 5.956208229064941, "learning_rate": 3.131813434924208e-06, "loss": 0.4228, "step": 12369 }, { "epoch": 0.6257271485659366, "grad_norm": 7.567938804626465, "learning_rate": 3.1310689741903362e-06, "loss": 0.3944, "step": 12370 }, { "epoch": 0.6257777328140017, "grad_norm": 7.681941509246826, "learning_rate": 3.1303245616143026e-06, "loss": 0.4348, "step": 12371 }, { "epoch": 0.6258283170620669, "grad_norm": 13.497103691101074, "learning_rate": 3.1295801972152896e-06, "loss": 0.5784, "step": 12372 }, { "epoch": 0.625878901310132, "grad_norm": 4.898360729217529, "learning_rate": 3.1288358810124764e-06, "loss": 0.4165, "step": 12373 }, { "epoch": 0.6259294855581972, "grad_norm": 8.844886779785156, "learning_rate": 3.1280916130250427e-06, "loss": 0.408, "step": 12374 }, { "epoch": 0.6259800698062623, "grad_norm": 6.703699588775635, "learning_rate": 3.127347393272168e-06, "loss": 0.3756, "step": 12375 }, { "epoch": 0.6260306540543275, "grad_norm": 8.96884822845459, "learning_rate": 3.126603221773027e-06, "loss": 0.3975, "step": 12376 }, { "epoch": 0.6260812383023926, "grad_norm": 9.289135932922363, "learning_rate": 3.1258590985467975e-06, "loss": 0.3677, "step": 12377 }, { "epoch": 0.6261318225504577, "grad_norm": 7.316531658172607, "learning_rate": 3.1251150236126504e-06, "loss": 0.3705, "step": 12378 }, { "epoch": 0.626182406798523, "grad_norm": 13.337645530700684, "learning_rate": 3.1243709969897617e-06, "loss": 0.5345, "step": 12379 }, { "epoch": 0.6262329910465881, "grad_norm": 8.567634582519531, "learning_rate": 3.1236270186973012e-06, "loss": 0.4702, "step": 12380 }, { "epoch": 0.6262835752946533, "grad_norm": 4.89150857925415, "learning_rate": 3.12288308875444e-06, "loss": 0.3326, "step": 12381 }, { "epoch": 0.6263341595427184, "grad_norm": 11.223089218139648, "learning_rate": 3.122139207180348e-06, "loss": 0.4262, "step": 12382 }, { "epoch": 0.6263847437907836, "grad_norm": 11.961299896240234, "learning_rate": 3.1213953739941926e-06, "loss": 0.4227, "step": 12383 }, { "epoch": 0.6264353280388487, "grad_norm": 11.357949256896973, "learning_rate": 3.1206515892151405e-06, "loss": 0.5075, "step": 12384 }, { "epoch": 0.6264859122869139, "grad_norm": 10.529030799865723, "learning_rate": 3.119907852862359e-06, "loss": 0.4209, "step": 12385 }, { "epoch": 0.626536496534979, "grad_norm": 7.910451889038086, "learning_rate": 3.11916416495501e-06, "loss": 0.4873, "step": 12386 }, { "epoch": 0.6265870807830441, "grad_norm": 5.362103462219238, "learning_rate": 3.1184205255122566e-06, "loss": 0.3673, "step": 12387 }, { "epoch": 0.6266376650311093, "grad_norm": 4.516971588134766, "learning_rate": 3.117676934553263e-06, "loss": 0.3492, "step": 12388 }, { "epoch": 0.6266882492791744, "grad_norm": 8.275897026062012, "learning_rate": 3.1169333920971875e-06, "loss": 0.3164, "step": 12389 }, { "epoch": 0.6267388335272396, "grad_norm": 10.459783554077148, "learning_rate": 3.116189898163192e-06, "loss": 0.3392, "step": 12390 }, { "epoch": 0.6267894177753047, "grad_norm": 9.041143417358398, "learning_rate": 3.1154464527704305e-06, "loss": 0.5285, "step": 12391 }, { "epoch": 0.6268400020233699, "grad_norm": 5.645458698272705, "learning_rate": 3.114703055938063e-06, "loss": 0.3466, "step": 12392 }, { "epoch": 0.6268905862714351, "grad_norm": 7.410915851593018, "learning_rate": 3.1139597076852423e-06, "loss": 0.3956, "step": 12393 }, { "epoch": 0.6269411705195003, "grad_norm": 9.764594078063965, "learning_rate": 3.1132164080311254e-06, "loss": 0.5217, "step": 12394 }, { "epoch": 0.6269917547675654, "grad_norm": 6.172876358032227, "learning_rate": 3.112473156994865e-06, "loss": 0.3405, "step": 12395 }, { "epoch": 0.6270423390156306, "grad_norm": 5.78227424621582, "learning_rate": 3.1117299545956114e-06, "loss": 0.4568, "step": 12396 }, { "epoch": 0.6270929232636957, "grad_norm": 6.745666027069092, "learning_rate": 3.1109868008525157e-06, "loss": 0.2778, "step": 12397 }, { "epoch": 0.6271435075117608, "grad_norm": 10.987543106079102, "learning_rate": 3.1102436957847283e-06, "loss": 0.407, "step": 12398 }, { "epoch": 0.627194091759826, "grad_norm": 7.172015190124512, "learning_rate": 3.1095006394113958e-06, "loss": 0.4438, "step": 12399 }, { "epoch": 0.6272446760078911, "grad_norm": 6.294021129608154, "learning_rate": 3.108757631751666e-06, "loss": 0.4163, "step": 12400 }, { "epoch": 0.6272952602559563, "grad_norm": 12.584951400756836, "learning_rate": 3.108014672824683e-06, "loss": 0.5355, "step": 12401 }, { "epoch": 0.6273458445040214, "grad_norm": 8.113691329956055, "learning_rate": 3.1072717626495923e-06, "loss": 0.3896, "step": 12402 }, { "epoch": 0.6273964287520866, "grad_norm": 9.297428131103516, "learning_rate": 3.1065289012455387e-06, "loss": 0.4759, "step": 12403 }, { "epoch": 0.6274470130001517, "grad_norm": 10.914955139160156, "learning_rate": 3.1057860886316603e-06, "loss": 0.5163, "step": 12404 }, { "epoch": 0.6274975972482169, "grad_norm": 9.602496147155762, "learning_rate": 3.1050433248270984e-06, "loss": 0.4238, "step": 12405 }, { "epoch": 0.6275481814962821, "grad_norm": 6.207921504974365, "learning_rate": 3.1043006098509933e-06, "loss": 0.3788, "step": 12406 }, { "epoch": 0.6275987657443473, "grad_norm": 5.389199256896973, "learning_rate": 3.103557943722483e-06, "loss": 0.441, "step": 12407 }, { "epoch": 0.6276493499924124, "grad_norm": 6.45003604888916, "learning_rate": 3.102815326460703e-06, "loss": 0.3077, "step": 12408 }, { "epoch": 0.6276999342404775, "grad_norm": 5.441888332366943, "learning_rate": 3.1020727580847893e-06, "loss": 0.4163, "step": 12409 }, { "epoch": 0.6277505184885427, "grad_norm": 8.440940856933594, "learning_rate": 3.101330238613878e-06, "loss": 0.5193, "step": 12410 }, { "epoch": 0.6278011027366078, "grad_norm": 5.897415637969971, "learning_rate": 3.1005877680670987e-06, "loss": 0.3172, "step": 12411 }, { "epoch": 0.627851686984673, "grad_norm": 5.574635028839111, "learning_rate": 3.0998453464635853e-06, "loss": 0.2954, "step": 12412 }, { "epoch": 0.6279022712327381, "grad_norm": 4.972814559936523, "learning_rate": 3.0991029738224688e-06, "loss": 0.3276, "step": 12413 }, { "epoch": 0.6279528554808033, "grad_norm": 9.312117576599121, "learning_rate": 3.0983606501628756e-06, "loss": 0.4856, "step": 12414 }, { "epoch": 0.6280034397288684, "grad_norm": 12.11952018737793, "learning_rate": 3.0976183755039355e-06, "loss": 0.4773, "step": 12415 }, { "epoch": 0.6280540239769336, "grad_norm": 8.647557258605957, "learning_rate": 3.096876149864777e-06, "loss": 0.5395, "step": 12416 }, { "epoch": 0.6281046082249987, "grad_norm": 9.039437294006348, "learning_rate": 3.0961339732645222e-06, "loss": 0.4324, "step": 12417 }, { "epoch": 0.6281551924730638, "grad_norm": 7.7271904945373535, "learning_rate": 3.0953918457222955e-06, "loss": 0.3465, "step": 12418 }, { "epoch": 0.6282057767211291, "grad_norm": 4.509310722351074, "learning_rate": 3.094649767257221e-06, "loss": 0.3973, "step": 12419 }, { "epoch": 0.6282563609691942, "grad_norm": 11.080878257751465, "learning_rate": 3.0939077378884196e-06, "loss": 0.5857, "step": 12420 }, { "epoch": 0.6283069452172594, "grad_norm": 7.603025436401367, "learning_rate": 3.0931657576350118e-06, "loss": 0.4574, "step": 12421 }, { "epoch": 0.6283575294653245, "grad_norm": 10.86634349822998, "learning_rate": 3.0924238265161154e-06, "loss": 0.4884, "step": 12422 }, { "epoch": 0.6284081137133897, "grad_norm": 4.649446964263916, "learning_rate": 3.0916819445508515e-06, "loss": 0.359, "step": 12423 }, { "epoch": 0.6284586979614548, "grad_norm": 5.177992820739746, "learning_rate": 3.0909401117583336e-06, "loss": 0.4065, "step": 12424 }, { "epoch": 0.62850928220952, "grad_norm": 7.887167930603027, "learning_rate": 3.0901983281576777e-06, "loss": 0.3652, "step": 12425 }, { "epoch": 0.6285598664575851, "grad_norm": 5.752506732940674, "learning_rate": 3.0894565937679986e-06, "loss": 0.3497, "step": 12426 }, { "epoch": 0.6286104507056502, "grad_norm": 9.648508071899414, "learning_rate": 3.0887149086084077e-06, "loss": 0.4208, "step": 12427 }, { "epoch": 0.6286610349537154, "grad_norm": 10.198369026184082, "learning_rate": 3.0879732726980173e-06, "loss": 0.5236, "step": 12428 }, { "epoch": 0.6287116192017805, "grad_norm": 7.398908615112305, "learning_rate": 3.0872316860559387e-06, "loss": 0.4414, "step": 12429 }, { "epoch": 0.6287622034498457, "grad_norm": 7.57255744934082, "learning_rate": 3.0864901487012788e-06, "loss": 0.548, "step": 12430 }, { "epoch": 0.6288127876979108, "grad_norm": 9.0541410446167, "learning_rate": 3.085748660653145e-06, "loss": 0.4415, "step": 12431 }, { "epoch": 0.628863371945976, "grad_norm": 12.418475151062012, "learning_rate": 3.0850072219306444e-06, "loss": 0.519, "step": 12432 }, { "epoch": 0.6289139561940412, "grad_norm": 6.07249641418457, "learning_rate": 3.084265832552883e-06, "loss": 0.3518, "step": 12433 }, { "epoch": 0.6289645404421064, "grad_norm": 8.197872161865234, "learning_rate": 3.083524492538963e-06, "loss": 0.3962, "step": 12434 }, { "epoch": 0.6290151246901715, "grad_norm": 7.1357340812683105, "learning_rate": 3.082783201907989e-06, "loss": 0.4304, "step": 12435 }, { "epoch": 0.6290657089382367, "grad_norm": 8.09432601928711, "learning_rate": 3.0820419606790596e-06, "loss": 0.4003, "step": 12436 }, { "epoch": 0.6291162931863018, "grad_norm": 6.781416416168213, "learning_rate": 3.081300768871277e-06, "loss": 0.3612, "step": 12437 }, { "epoch": 0.629166877434367, "grad_norm": 7.322079658508301, "learning_rate": 3.0805596265037395e-06, "loss": 0.4564, "step": 12438 }, { "epoch": 0.6292174616824321, "grad_norm": 8.88705062866211, "learning_rate": 3.0798185335955433e-06, "loss": 0.431, "step": 12439 }, { "epoch": 0.6292680459304972, "grad_norm": 13.76910400390625, "learning_rate": 3.0790774901657857e-06, "loss": 0.4317, "step": 12440 }, { "epoch": 0.6293186301785624, "grad_norm": 7.977394104003906, "learning_rate": 3.078336496233562e-06, "loss": 0.4295, "step": 12441 }, { "epoch": 0.6293692144266275, "grad_norm": 10.523490905761719, "learning_rate": 3.0775955518179666e-06, "loss": 0.4735, "step": 12442 }, { "epoch": 0.6294197986746927, "grad_norm": 7.819516658782959, "learning_rate": 3.076854656938088e-06, "loss": 0.434, "step": 12443 }, { "epoch": 0.6294703829227578, "grad_norm": 6.0495476722717285, "learning_rate": 3.07611381161302e-06, "loss": 0.4262, "step": 12444 }, { "epoch": 0.629520967170823, "grad_norm": 8.004480361938477, "learning_rate": 3.0753730158618533e-06, "loss": 0.4012, "step": 12445 }, { "epoch": 0.6295715514188882, "grad_norm": 12.726753234863281, "learning_rate": 3.074632269703674e-06, "loss": 0.4193, "step": 12446 }, { "epoch": 0.6296221356669534, "grad_norm": 9.361279487609863, "learning_rate": 3.0738915731575703e-06, "loss": 0.5447, "step": 12447 }, { "epoch": 0.6296727199150185, "grad_norm": 9.100626945495605, "learning_rate": 3.07315092624263e-06, "loss": 0.4864, "step": 12448 }, { "epoch": 0.6297233041630836, "grad_norm": 4.204693794250488, "learning_rate": 3.072410328977935e-06, "loss": 0.3871, "step": 12449 }, { "epoch": 0.6297738884111488, "grad_norm": 9.320661544799805, "learning_rate": 3.07166978138257e-06, "loss": 0.3997, "step": 12450 }, { "epoch": 0.6298244726592139, "grad_norm": 5.553530216217041, "learning_rate": 3.0709292834756176e-06, "loss": 0.3695, "step": 12451 }, { "epoch": 0.6298750569072791, "grad_norm": 8.720129013061523, "learning_rate": 3.070188835276158e-06, "loss": 0.4096, "step": 12452 }, { "epoch": 0.6299256411553442, "grad_norm": 7.864058494567871, "learning_rate": 3.0694484368032694e-06, "loss": 0.3872, "step": 12453 }, { "epoch": 0.6299762254034094, "grad_norm": 6.267116069793701, "learning_rate": 3.068708088076034e-06, "loss": 0.4385, "step": 12454 }, { "epoch": 0.6300268096514745, "grad_norm": 7.301961421966553, "learning_rate": 3.0679677891135266e-06, "loss": 0.3566, "step": 12455 }, { "epoch": 0.6300773938995397, "grad_norm": 5.475329399108887, "learning_rate": 3.0672275399348204e-06, "loss": 0.2934, "step": 12456 }, { "epoch": 0.6301279781476048, "grad_norm": 8.317110061645508, "learning_rate": 3.066487340558993e-06, "loss": 0.4376, "step": 12457 }, { "epoch": 0.6301785623956699, "grad_norm": 7.388470649719238, "learning_rate": 3.0657471910051173e-06, "loss": 0.4694, "step": 12458 }, { "epoch": 0.6302291466437352, "grad_norm": 8.462900161743164, "learning_rate": 3.065007091292263e-06, "loss": 0.4135, "step": 12459 }, { "epoch": 0.6302797308918003, "grad_norm": 6.458677768707275, "learning_rate": 3.0642670414395028e-06, "loss": 0.2929, "step": 12460 }, { "epoch": 0.6303303151398655, "grad_norm": 7.367126941680908, "learning_rate": 3.063527041465906e-06, "loss": 0.3864, "step": 12461 }, { "epoch": 0.6303808993879306, "grad_norm": 15.502408027648926, "learning_rate": 3.062787091390539e-06, "loss": 0.5092, "step": 12462 }, { "epoch": 0.6304314836359958, "grad_norm": 10.340920448303223, "learning_rate": 3.062047191232469e-06, "loss": 0.4991, "step": 12463 }, { "epoch": 0.6304820678840609, "grad_norm": 9.206384658813477, "learning_rate": 3.0613073410107637e-06, "loss": 0.5012, "step": 12464 }, { "epoch": 0.6305326521321261, "grad_norm": 7.034243106842041, "learning_rate": 3.060567540744484e-06, "loss": 0.4034, "step": 12465 }, { "epoch": 0.6305832363801912, "grad_norm": 5.679548263549805, "learning_rate": 3.059827790452695e-06, "loss": 0.4093, "step": 12466 }, { "epoch": 0.6306338206282563, "grad_norm": 4.792050361633301, "learning_rate": 3.0590880901544584e-06, "loss": 0.3808, "step": 12467 }, { "epoch": 0.6306844048763215, "grad_norm": 6.17864465713501, "learning_rate": 3.0583484398688335e-06, "loss": 0.4296, "step": 12468 }, { "epoch": 0.6307349891243866, "grad_norm": 7.521821022033691, "learning_rate": 3.0576088396148783e-06, "loss": 0.4105, "step": 12469 }, { "epoch": 0.6307855733724518, "grad_norm": 5.167011260986328, "learning_rate": 3.0568692894116514e-06, "loss": 0.3424, "step": 12470 }, { "epoch": 0.6308361576205169, "grad_norm": 9.934518814086914, "learning_rate": 3.0561297892782103e-06, "loss": 0.4446, "step": 12471 }, { "epoch": 0.6308867418685821, "grad_norm": 9.8706693649292, "learning_rate": 3.0553903392336094e-06, "loss": 0.4871, "step": 12472 }, { "epoch": 0.6309373261166473, "grad_norm": 18.344749450683594, "learning_rate": 3.0546509392969026e-06, "loss": 0.4173, "step": 12473 }, { "epoch": 0.6309879103647125, "grad_norm": 6.936984539031982, "learning_rate": 3.0539115894871415e-06, "loss": 0.3699, "step": 12474 }, { "epoch": 0.6310384946127776, "grad_norm": 9.283970832824707, "learning_rate": 3.053172289823378e-06, "loss": 0.4904, "step": 12475 }, { "epoch": 0.6310890788608428, "grad_norm": 4.325610637664795, "learning_rate": 3.052433040324663e-06, "loss": 0.369, "step": 12476 }, { "epoch": 0.6311396631089079, "grad_norm": 4.671260833740234, "learning_rate": 3.051693841010044e-06, "loss": 0.3108, "step": 12477 }, { "epoch": 0.631190247356973, "grad_norm": 8.23345947265625, "learning_rate": 3.050954691898568e-06, "loss": 0.3424, "step": 12478 }, { "epoch": 0.6312408316050382, "grad_norm": 11.236379623413086, "learning_rate": 3.0502155930092836e-06, "loss": 0.4245, "step": 12479 }, { "epoch": 0.6312914158531033, "grad_norm": 17.425724029541016, "learning_rate": 3.0494765443612344e-06, "loss": 0.5642, "step": 12480 }, { "epoch": 0.6313420001011685, "grad_norm": 7.371676921844482, "learning_rate": 3.0487375459734614e-06, "loss": 0.4028, "step": 12481 }, { "epoch": 0.6313925843492336, "grad_norm": 6.531951904296875, "learning_rate": 3.047998597865009e-06, "loss": 0.4234, "step": 12482 }, { "epoch": 0.6314431685972988, "grad_norm": 8.534799575805664, "learning_rate": 3.0472597000549186e-06, "loss": 0.4472, "step": 12483 }, { "epoch": 0.6314937528453639, "grad_norm": 5.697758197784424, "learning_rate": 3.046520852562228e-06, "loss": 0.339, "step": 12484 }, { "epoch": 0.631544337093429, "grad_norm": 7.602097034454346, "learning_rate": 3.045782055405977e-06, "loss": 0.3549, "step": 12485 }, { "epoch": 0.6315949213414943, "grad_norm": 12.579163551330566, "learning_rate": 3.0450433086052034e-06, "loss": 0.3776, "step": 12486 }, { "epoch": 0.6316455055895595, "grad_norm": 8.979464530944824, "learning_rate": 3.04430461217894e-06, "loss": 0.4132, "step": 12487 }, { "epoch": 0.6316960898376246, "grad_norm": 8.210894584655762, "learning_rate": 3.0435659661462237e-06, "loss": 0.5216, "step": 12488 }, { "epoch": 0.6317466740856897, "grad_norm": 7.36405086517334, "learning_rate": 3.0428273705260874e-06, "loss": 0.4712, "step": 12489 }, { "epoch": 0.6317972583337549, "grad_norm": 4.85470724105835, "learning_rate": 3.0420888253375613e-06, "loss": 0.3613, "step": 12490 }, { "epoch": 0.63184784258182, "grad_norm": 5.5293097496032715, "learning_rate": 3.041350330599677e-06, "loss": 0.437, "step": 12491 }, { "epoch": 0.6318984268298852, "grad_norm": 13.632567405700684, "learning_rate": 3.0406118863314655e-06, "loss": 0.4102, "step": 12492 }, { "epoch": 0.6319490110779503, "grad_norm": 7.6790080070495605, "learning_rate": 3.0398734925519537e-06, "loss": 0.4317, "step": 12493 }, { "epoch": 0.6319995953260155, "grad_norm": 10.491876602172852, "learning_rate": 3.039135149280165e-06, "loss": 0.4792, "step": 12494 }, { "epoch": 0.6320501795740806, "grad_norm": 10.838165283203125, "learning_rate": 3.0383968565351275e-06, "loss": 0.4139, "step": 12495 }, { "epoch": 0.6321007638221458, "grad_norm": 13.211053848266602, "learning_rate": 3.0376586143358657e-06, "loss": 0.4898, "step": 12496 }, { "epoch": 0.6321513480702109, "grad_norm": 8.239879608154297, "learning_rate": 3.036920422701401e-06, "loss": 0.4581, "step": 12497 }, { "epoch": 0.632201932318276, "grad_norm": 5.008341312408447, "learning_rate": 3.0361822816507553e-06, "loss": 0.3669, "step": 12498 }, { "epoch": 0.6322525165663413, "grad_norm": 5.761412143707275, "learning_rate": 3.0354441912029496e-06, "loss": 0.3812, "step": 12499 }, { "epoch": 0.6323031008144064, "grad_norm": 5.956238269805908, "learning_rate": 3.034706151377001e-06, "loss": 0.4426, "step": 12500 }, { "epoch": 0.6323536850624716, "grad_norm": 5.322170734405518, "learning_rate": 3.033968162191928e-06, "loss": 0.3172, "step": 12501 }, { "epoch": 0.6324042693105367, "grad_norm": 8.062934875488281, "learning_rate": 3.0332302236667476e-06, "loss": 0.3523, "step": 12502 }, { "epoch": 0.6324548535586019, "grad_norm": 6.402861595153809, "learning_rate": 3.0324923358204726e-06, "loss": 0.3078, "step": 12503 }, { "epoch": 0.632505437806667, "grad_norm": 7.129334449768066, "learning_rate": 3.031754498672119e-06, "loss": 0.3418, "step": 12504 }, { "epoch": 0.6325560220547322, "grad_norm": 7.570074081420898, "learning_rate": 3.031016712240697e-06, "loss": 0.3072, "step": 12505 }, { "epoch": 0.6326066063027973, "grad_norm": 12.529190063476562, "learning_rate": 3.0302789765452203e-06, "loss": 0.5225, "step": 12506 }, { "epoch": 0.6326571905508624, "grad_norm": 10.231783866882324, "learning_rate": 3.029541291604695e-06, "loss": 0.3977, "step": 12507 }, { "epoch": 0.6327077747989276, "grad_norm": 7.70994758605957, "learning_rate": 3.0288036574381318e-06, "loss": 0.4012, "step": 12508 }, { "epoch": 0.6327583590469927, "grad_norm": 4.741724967956543, "learning_rate": 3.028066074064536e-06, "loss": 0.4063, "step": 12509 }, { "epoch": 0.6328089432950579, "grad_norm": 8.687324523925781, "learning_rate": 3.0273285415029147e-06, "loss": 0.4014, "step": 12510 }, { "epoch": 0.632859527543123, "grad_norm": 9.900546073913574, "learning_rate": 3.0265910597722732e-06, "loss": 0.443, "step": 12511 }, { "epoch": 0.6329101117911883, "grad_norm": 8.832901000976562, "learning_rate": 3.025853628891612e-06, "loss": 0.3841, "step": 12512 }, { "epoch": 0.6329606960392534, "grad_norm": 10.371328353881836, "learning_rate": 3.025116248879935e-06, "loss": 0.4558, "step": 12513 }, { "epoch": 0.6330112802873186, "grad_norm": 7.6346588134765625, "learning_rate": 3.024378919756243e-06, "loss": 0.4041, "step": 12514 }, { "epoch": 0.6330618645353837, "grad_norm": 8.711395263671875, "learning_rate": 3.0236416415395332e-06, "loss": 0.5023, "step": 12515 }, { "epoch": 0.6331124487834489, "grad_norm": 5.232469081878662, "learning_rate": 3.0229044142488045e-06, "loss": 0.4301, "step": 12516 }, { "epoch": 0.633163033031514, "grad_norm": 13.563660621643066, "learning_rate": 3.022167237903055e-06, "loss": 0.4229, "step": 12517 }, { "epoch": 0.6332136172795791, "grad_norm": 13.073450088500977, "learning_rate": 3.0214301125212767e-06, "loss": 0.4552, "step": 12518 }, { "epoch": 0.6332642015276443, "grad_norm": 5.737964630126953, "learning_rate": 3.0206930381224688e-06, "loss": 0.3023, "step": 12519 }, { "epoch": 0.6333147857757094, "grad_norm": 10.75556755065918, "learning_rate": 3.019956014725617e-06, "loss": 0.3951, "step": 12520 }, { "epoch": 0.6333653700237746, "grad_norm": 7.600780487060547, "learning_rate": 3.019219042349718e-06, "loss": 0.3357, "step": 12521 }, { "epoch": 0.6334159542718397, "grad_norm": 8.39964485168457, "learning_rate": 3.018482121013758e-06, "loss": 0.4763, "step": 12522 }, { "epoch": 0.6334665385199049, "grad_norm": 6.052209854125977, "learning_rate": 3.017745250736729e-06, "loss": 0.3978, "step": 12523 }, { "epoch": 0.63351712276797, "grad_norm": 8.008759498596191, "learning_rate": 3.0170084315376167e-06, "loss": 0.4179, "step": 12524 }, { "epoch": 0.6335677070160352, "grad_norm": 6.110080718994141, "learning_rate": 3.0162716634354067e-06, "loss": 0.3535, "step": 12525 }, { "epoch": 0.6336182912641004, "grad_norm": 9.730738639831543, "learning_rate": 3.0155349464490847e-06, "loss": 0.4605, "step": 12526 }, { "epoch": 0.6336688755121656, "grad_norm": 5.162713050842285, "learning_rate": 3.0147982805976343e-06, "loss": 0.3653, "step": 12527 }, { "epoch": 0.6337194597602307, "grad_norm": 7.308144569396973, "learning_rate": 3.0140616659000367e-06, "loss": 0.2729, "step": 12528 }, { "epoch": 0.6337700440082958, "grad_norm": 5.105728626251221, "learning_rate": 3.013325102375273e-06, "loss": 0.3193, "step": 12529 }, { "epoch": 0.633820628256361, "grad_norm": 9.892019271850586, "learning_rate": 3.0125885900423234e-06, "loss": 0.4068, "step": 12530 }, { "epoch": 0.6338712125044261, "grad_norm": 6.056168079376221, "learning_rate": 3.0118521289201652e-06, "loss": 0.4556, "step": 12531 }, { "epoch": 0.6339217967524913, "grad_norm": 6.067975997924805, "learning_rate": 3.0111157190277774e-06, "loss": 0.3877, "step": 12532 }, { "epoch": 0.6339723810005564, "grad_norm": 5.753444194793701, "learning_rate": 3.010379360384131e-06, "loss": 0.4075, "step": 12533 }, { "epoch": 0.6340229652486216, "grad_norm": 4.624599933624268, "learning_rate": 3.0096430530082037e-06, "loss": 0.3539, "step": 12534 }, { "epoch": 0.6340735494966867, "grad_norm": 21.453697204589844, "learning_rate": 3.0089067969189666e-06, "loss": 0.3934, "step": 12535 }, { "epoch": 0.6341241337447519, "grad_norm": 7.637514591217041, "learning_rate": 3.008170592135392e-06, "loss": 0.3661, "step": 12536 }, { "epoch": 0.634174717992817, "grad_norm": 8.798431396484375, "learning_rate": 3.007434438676452e-06, "loss": 0.3878, "step": 12537 }, { "epoch": 0.6342253022408821, "grad_norm": 11.2471284866333, "learning_rate": 3.0066983365611117e-06, "loss": 0.3752, "step": 12538 }, { "epoch": 0.6342758864889474, "grad_norm": 8.121581077575684, "learning_rate": 3.005962285808342e-06, "loss": 0.4354, "step": 12539 }, { "epoch": 0.6343264707370125, "grad_norm": 10.72261905670166, "learning_rate": 3.0052262864371066e-06, "loss": 0.4336, "step": 12540 }, { "epoch": 0.6343770549850777, "grad_norm": 9.156031608581543, "learning_rate": 3.004490338466372e-06, "loss": 0.4015, "step": 12541 }, { "epoch": 0.6344276392331428, "grad_norm": 13.312541007995605, "learning_rate": 3.003754441915102e-06, "loss": 0.4672, "step": 12542 }, { "epoch": 0.634478223481208, "grad_norm": 7.749606132507324, "learning_rate": 3.0030185968022575e-06, "loss": 0.3865, "step": 12543 }, { "epoch": 0.6345288077292731, "grad_norm": 12.704068183898926, "learning_rate": 3.0022828031468006e-06, "loss": 0.4402, "step": 12544 }, { "epoch": 0.6345793919773383, "grad_norm": 7.543729305267334, "learning_rate": 3.0015470609676924e-06, "loss": 0.3157, "step": 12545 }, { "epoch": 0.6346299762254034, "grad_norm": 5.563939571380615, "learning_rate": 3.0008113702838883e-06, "loss": 0.2968, "step": 12546 }, { "epoch": 0.6346805604734685, "grad_norm": 9.348799705505371, "learning_rate": 3.000075731114345e-06, "loss": 0.4127, "step": 12547 }, { "epoch": 0.6347311447215337, "grad_norm": 6.437079906463623, "learning_rate": 2.9993401434780195e-06, "loss": 0.365, "step": 12548 }, { "epoch": 0.6347817289695988, "grad_norm": 4.97462272644043, "learning_rate": 2.9986046073938668e-06, "loss": 0.3459, "step": 12549 }, { "epoch": 0.634832313217664, "grad_norm": 5.2224321365356445, "learning_rate": 2.9978691228808386e-06, "loss": 0.3745, "step": 12550 }, { "epoch": 0.6348828974657291, "grad_norm": 6.259343147277832, "learning_rate": 2.9971336899578877e-06, "loss": 0.3545, "step": 12551 }, { "epoch": 0.6349334817137944, "grad_norm": 6.372990131378174, "learning_rate": 2.9963983086439644e-06, "loss": 0.367, "step": 12552 }, { "epoch": 0.6349840659618595, "grad_norm": 5.864536762237549, "learning_rate": 2.995662978958016e-06, "loss": 0.3609, "step": 12553 }, { "epoch": 0.6350346502099247, "grad_norm": 8.290332794189453, "learning_rate": 2.994927700918992e-06, "loss": 0.3968, "step": 12554 }, { "epoch": 0.6350852344579898, "grad_norm": 12.620104789733887, "learning_rate": 2.9941924745458383e-06, "loss": 0.5556, "step": 12555 }, { "epoch": 0.635135818706055, "grad_norm": 5.8905792236328125, "learning_rate": 2.9934572998574997e-06, "loss": 0.4255, "step": 12556 }, { "epoch": 0.6351864029541201, "grad_norm": 9.957784652709961, "learning_rate": 2.992722176872921e-06, "loss": 0.393, "step": 12557 }, { "epoch": 0.6352369872021852, "grad_norm": 6.083673000335693, "learning_rate": 2.9919871056110415e-06, "loss": 0.3329, "step": 12558 }, { "epoch": 0.6352875714502504, "grad_norm": 7.87135648727417, "learning_rate": 2.991252086090806e-06, "loss": 0.4965, "step": 12559 }, { "epoch": 0.6353381556983155, "grad_norm": 9.011783599853516, "learning_rate": 2.9905171183311497e-06, "loss": 0.4115, "step": 12560 }, { "epoch": 0.6353887399463807, "grad_norm": 11.182199478149414, "learning_rate": 2.989782202351015e-06, "loss": 0.3217, "step": 12561 }, { "epoch": 0.6354393241944458, "grad_norm": 6.7214274406433105, "learning_rate": 2.9890473381693376e-06, "loss": 0.3757, "step": 12562 }, { "epoch": 0.635489908442511, "grad_norm": 19.9953556060791, "learning_rate": 2.9883125258050526e-06, "loss": 0.5114, "step": 12563 }, { "epoch": 0.6355404926905761, "grad_norm": 6.9350714683532715, "learning_rate": 2.9875777652770942e-06, "loss": 0.3955, "step": 12564 }, { "epoch": 0.6355910769386413, "grad_norm": 11.889699935913086, "learning_rate": 2.9868430566043977e-06, "loss": 0.4023, "step": 12565 }, { "epoch": 0.6356416611867065, "grad_norm": 8.040771484375, "learning_rate": 2.9861083998058915e-06, "loss": 0.3297, "step": 12566 }, { "epoch": 0.6356922454347717, "grad_norm": 14.840651512145996, "learning_rate": 2.985373794900508e-06, "loss": 0.4306, "step": 12567 }, { "epoch": 0.6357428296828368, "grad_norm": 8.285385131835938, "learning_rate": 2.9846392419071762e-06, "loss": 0.3428, "step": 12568 }, { "epoch": 0.6357934139309019, "grad_norm": 16.965757369995117, "learning_rate": 2.9839047408448224e-06, "loss": 0.3743, "step": 12569 }, { "epoch": 0.6358439981789671, "grad_norm": 8.54147720336914, "learning_rate": 2.9831702917323756e-06, "loss": 0.4542, "step": 12570 }, { "epoch": 0.6358945824270322, "grad_norm": 6.542113780975342, "learning_rate": 2.9824358945887574e-06, "loss": 0.3408, "step": 12571 }, { "epoch": 0.6359451666750974, "grad_norm": 8.3240327835083, "learning_rate": 2.9817015494328937e-06, "loss": 0.4746, "step": 12572 }, { "epoch": 0.6359957509231625, "grad_norm": 7.09583854675293, "learning_rate": 2.9809672562837045e-06, "loss": 0.4061, "step": 12573 }, { "epoch": 0.6360463351712277, "grad_norm": 5.946650981903076, "learning_rate": 2.9802330151601132e-06, "loss": 0.3995, "step": 12574 }, { "epoch": 0.6360969194192928, "grad_norm": 6.117649078369141, "learning_rate": 2.979498826081039e-06, "loss": 0.4238, "step": 12575 }, { "epoch": 0.636147503667358, "grad_norm": 11.265509605407715, "learning_rate": 2.978764689065399e-06, "loss": 0.4909, "step": 12576 }, { "epoch": 0.6361980879154231, "grad_norm": 5.269223213195801, "learning_rate": 2.9780306041321115e-06, "loss": 0.3375, "step": 12577 }, { "epoch": 0.6362486721634882, "grad_norm": 16.748788833618164, "learning_rate": 2.9772965713000913e-06, "loss": 0.6562, "step": 12578 }, { "epoch": 0.6362992564115535, "grad_norm": 11.815683364868164, "learning_rate": 2.976562590588252e-06, "loss": 0.5943, "step": 12579 }, { "epoch": 0.6363498406596186, "grad_norm": 5.157867431640625, "learning_rate": 2.9758286620155087e-06, "loss": 0.32, "step": 12580 }, { "epoch": 0.6364004249076838, "grad_norm": 8.237980842590332, "learning_rate": 2.9750947856007705e-06, "loss": 0.3681, "step": 12581 }, { "epoch": 0.6364510091557489, "grad_norm": 10.22657299041748, "learning_rate": 2.9743609613629485e-06, "loss": 0.3946, "step": 12582 }, { "epoch": 0.6365015934038141, "grad_norm": 11.470673561096191, "learning_rate": 2.973627189320954e-06, "loss": 0.493, "step": 12583 }, { "epoch": 0.6365521776518792, "grad_norm": 11.536014556884766, "learning_rate": 2.9728934694936917e-06, "loss": 0.3919, "step": 12584 }, { "epoch": 0.6366027618999444, "grad_norm": 7.798485279083252, "learning_rate": 2.9721598019000664e-06, "loss": 0.4385, "step": 12585 }, { "epoch": 0.6366533461480095, "grad_norm": 10.167778968811035, "learning_rate": 2.9714261865589862e-06, "loss": 0.3966, "step": 12586 }, { "epoch": 0.6367039303960746, "grad_norm": 8.771833419799805, "learning_rate": 2.9706926234893536e-06, "loss": 0.4361, "step": 12587 }, { "epoch": 0.6367545146441398, "grad_norm": 4.964158535003662, "learning_rate": 2.9699591127100697e-06, "loss": 0.3734, "step": 12588 }, { "epoch": 0.6368050988922049, "grad_norm": 7.830482482910156, "learning_rate": 2.9692256542400356e-06, "loss": 0.3924, "step": 12589 }, { "epoch": 0.6368556831402701, "grad_norm": 6.344986438751221, "learning_rate": 2.968492248098153e-06, "loss": 0.3795, "step": 12590 }, { "epoch": 0.6369062673883352, "grad_norm": 8.873017311096191, "learning_rate": 2.967758894303317e-06, "loss": 0.4715, "step": 12591 }, { "epoch": 0.6369568516364005, "grad_norm": 12.636601448059082, "learning_rate": 2.967025592874426e-06, "loss": 0.5719, "step": 12592 }, { "epoch": 0.6370074358844656, "grad_norm": 5.033390998840332, "learning_rate": 2.9662923438303754e-06, "loss": 0.3413, "step": 12593 }, { "epoch": 0.6370580201325308, "grad_norm": 6.333237648010254, "learning_rate": 2.965559147190058e-06, "loss": 0.4076, "step": 12594 }, { "epoch": 0.6371086043805959, "grad_norm": 9.064974784851074, "learning_rate": 2.9648260029723676e-06, "loss": 0.3323, "step": 12595 }, { "epoch": 0.637159188628661, "grad_norm": 8.442377090454102, "learning_rate": 2.964092911196197e-06, "loss": 0.4583, "step": 12596 }, { "epoch": 0.6372097728767262, "grad_norm": 5.916069984436035, "learning_rate": 2.9633598718804337e-06, "loss": 0.396, "step": 12597 }, { "epoch": 0.6372603571247913, "grad_norm": 5.552792549133301, "learning_rate": 2.9626268850439668e-06, "loss": 0.3898, "step": 12598 }, { "epoch": 0.6373109413728565, "grad_norm": 8.17307186126709, "learning_rate": 2.9618939507056833e-06, "loss": 0.3695, "step": 12599 }, { "epoch": 0.6373615256209216, "grad_norm": 15.4221773147583, "learning_rate": 2.9611610688844705e-06, "loss": 0.4307, "step": 12600 }, { "epoch": 0.6374121098689868, "grad_norm": 11.16359806060791, "learning_rate": 2.9604282395992124e-06, "loss": 0.615, "step": 12601 }, { "epoch": 0.6374626941170519, "grad_norm": 7.006525039672852, "learning_rate": 2.959695462868791e-06, "loss": 0.4324, "step": 12602 }, { "epoch": 0.6375132783651171, "grad_norm": 11.061973571777344, "learning_rate": 2.958962738712091e-06, "loss": 0.5103, "step": 12603 }, { "epoch": 0.6375638626131822, "grad_norm": 8.846390724182129, "learning_rate": 2.958230067147991e-06, "loss": 0.3932, "step": 12604 }, { "epoch": 0.6376144468612474, "grad_norm": 6.628373146057129, "learning_rate": 2.9574974481953695e-06, "loss": 0.3356, "step": 12605 }, { "epoch": 0.6376650311093126, "grad_norm": 8.338190078735352, "learning_rate": 2.9567648818731065e-06, "loss": 0.4346, "step": 12606 }, { "epoch": 0.6377156153573778, "grad_norm": 6.25874137878418, "learning_rate": 2.9560323682000762e-06, "loss": 0.388, "step": 12607 }, { "epoch": 0.6377661996054429, "grad_norm": 10.607717514038086, "learning_rate": 2.955299907195156e-06, "loss": 0.4783, "step": 12608 }, { "epoch": 0.637816783853508, "grad_norm": 6.7677764892578125, "learning_rate": 2.9545674988772183e-06, "loss": 0.4269, "step": 12609 }, { "epoch": 0.6378673681015732, "grad_norm": 7.830479621887207, "learning_rate": 2.953835143265136e-06, "loss": 0.3884, "step": 12610 }, { "epoch": 0.6379179523496383, "grad_norm": 6.0675787925720215, "learning_rate": 2.9531028403777783e-06, "loss": 0.4147, "step": 12611 }, { "epoch": 0.6379685365977035, "grad_norm": 5.911930084228516, "learning_rate": 2.952370590234017e-06, "loss": 0.3629, "step": 12612 }, { "epoch": 0.6380191208457686, "grad_norm": 8.14421272277832, "learning_rate": 2.9516383928527193e-06, "loss": 0.3478, "step": 12613 }, { "epoch": 0.6380697050938338, "grad_norm": 6.100840091705322, "learning_rate": 2.9509062482527527e-06, "loss": 0.4328, "step": 12614 }, { "epoch": 0.6381202893418989, "grad_norm": 4.667459011077881, "learning_rate": 2.9501741564529835e-06, "loss": 0.4012, "step": 12615 }, { "epoch": 0.638170873589964, "grad_norm": 9.11440658569336, "learning_rate": 2.949442117472274e-06, "loss": 0.4193, "step": 12616 }, { "epoch": 0.6382214578380292, "grad_norm": 7.4224629402160645, "learning_rate": 2.9487101313294884e-06, "loss": 0.4429, "step": 12617 }, { "epoch": 0.6382720420860943, "grad_norm": 6.4531378746032715, "learning_rate": 2.9479781980434887e-06, "loss": 0.3116, "step": 12618 }, { "epoch": 0.6383226263341596, "grad_norm": 6.241519927978516, "learning_rate": 2.9472463176331335e-06, "loss": 0.3428, "step": 12619 }, { "epoch": 0.6383732105822247, "grad_norm": 8.484260559082031, "learning_rate": 2.9465144901172826e-06, "loss": 0.3813, "step": 12620 }, { "epoch": 0.6384237948302899, "grad_norm": 9.989777565002441, "learning_rate": 2.9457827155147943e-06, "loss": 0.3572, "step": 12621 }, { "epoch": 0.638474379078355, "grad_norm": 8.693360328674316, "learning_rate": 2.945050993844525e-06, "loss": 0.4177, "step": 12622 }, { "epoch": 0.6385249633264202, "grad_norm": 7.570591449737549, "learning_rate": 2.9443193251253253e-06, "loss": 0.3984, "step": 12623 }, { "epoch": 0.6385755475744853, "grad_norm": 8.497373580932617, "learning_rate": 2.9435877093760524e-06, "loss": 0.4079, "step": 12624 }, { "epoch": 0.6386261318225505, "grad_norm": 6.1981658935546875, "learning_rate": 2.942856146615557e-06, "loss": 0.3382, "step": 12625 }, { "epoch": 0.6386767160706156, "grad_norm": 14.031685829162598, "learning_rate": 2.9421246368626893e-06, "loss": 0.5209, "step": 12626 }, { "epoch": 0.6387273003186807, "grad_norm": 5.642564296722412, "learning_rate": 2.941393180136299e-06, "loss": 0.4752, "step": 12627 }, { "epoch": 0.6387778845667459, "grad_norm": 7.070592403411865, "learning_rate": 2.940661776455236e-06, "loss": 0.4512, "step": 12628 }, { "epoch": 0.638828468814811, "grad_norm": 7.100733280181885, "learning_rate": 2.9399304258383425e-06, "loss": 0.4006, "step": 12629 }, { "epoch": 0.6388790530628762, "grad_norm": 13.276642799377441, "learning_rate": 2.939199128304467e-06, "loss": 0.4983, "step": 12630 }, { "epoch": 0.6389296373109413, "grad_norm": 13.68086051940918, "learning_rate": 2.938467883872453e-06, "loss": 0.4584, "step": 12631 }, { "epoch": 0.6389802215590066, "grad_norm": 9.50634765625, "learning_rate": 2.937736692561141e-06, "loss": 0.3638, "step": 12632 }, { "epoch": 0.6390308058070717, "grad_norm": 12.530558586120605, "learning_rate": 2.937005554389374e-06, "loss": 0.5857, "step": 12633 }, { "epoch": 0.6390813900551369, "grad_norm": 7.520228385925293, "learning_rate": 2.936274469375991e-06, "loss": 0.3163, "step": 12634 }, { "epoch": 0.639131974303202, "grad_norm": 12.013687133789062, "learning_rate": 2.935543437539832e-06, "loss": 0.4999, "step": 12635 }, { "epoch": 0.6391825585512672, "grad_norm": 8.547592163085938, "learning_rate": 2.934812458899729e-06, "loss": 0.3492, "step": 12636 }, { "epoch": 0.6392331427993323, "grad_norm": 10.013566970825195, "learning_rate": 2.9340815334745216e-06, "loss": 0.5032, "step": 12637 }, { "epoch": 0.6392837270473974, "grad_norm": 5.637667179107666, "learning_rate": 2.9333506612830435e-06, "loss": 0.3675, "step": 12638 }, { "epoch": 0.6393343112954626, "grad_norm": 5.535999298095703, "learning_rate": 2.9326198423441265e-06, "loss": 0.3435, "step": 12639 }, { "epoch": 0.6393848955435277, "grad_norm": 10.072882652282715, "learning_rate": 2.9318890766766027e-06, "loss": 0.4983, "step": 12640 }, { "epoch": 0.6394354797915929, "grad_norm": 8.752718925476074, "learning_rate": 2.9311583642993026e-06, "loss": 0.5075, "step": 12641 }, { "epoch": 0.639486064039658, "grad_norm": 7.7942070960998535, "learning_rate": 2.9304277052310536e-06, "loss": 0.4646, "step": 12642 }, { "epoch": 0.6395366482877232, "grad_norm": 13.677504539489746, "learning_rate": 2.9296970994906848e-06, "loss": 0.4934, "step": 12643 }, { "epoch": 0.6395872325357883, "grad_norm": 5.937319278717041, "learning_rate": 2.92896654709702e-06, "loss": 0.4227, "step": 12644 }, { "epoch": 0.6396378167838536, "grad_norm": 6.782116889953613, "learning_rate": 2.9282360480688855e-06, "loss": 0.4601, "step": 12645 }, { "epoch": 0.6396884010319187, "grad_norm": 5.979361534118652, "learning_rate": 2.9275056024251044e-06, "loss": 0.3747, "step": 12646 }, { "epoch": 0.6397389852799839, "grad_norm": 7.81208610534668, "learning_rate": 2.926775210184497e-06, "loss": 0.4581, "step": 12647 }, { "epoch": 0.639789569528049, "grad_norm": 6.998337745666504, "learning_rate": 2.9260448713658867e-06, "loss": 0.3327, "step": 12648 }, { "epoch": 0.6398401537761141, "grad_norm": 7.470913887023926, "learning_rate": 2.925314585988089e-06, "loss": 0.4815, "step": 12649 }, { "epoch": 0.6398907380241793, "grad_norm": 16.709266662597656, "learning_rate": 2.9245843540699235e-06, "loss": 0.6268, "step": 12650 }, { "epoch": 0.6399413222722444, "grad_norm": 7.328620910644531, "learning_rate": 2.9238541756302065e-06, "loss": 0.3006, "step": 12651 }, { "epoch": 0.6399919065203096, "grad_norm": 6.0586934089660645, "learning_rate": 2.923124050687751e-06, "loss": 0.3819, "step": 12652 }, { "epoch": 0.6400424907683747, "grad_norm": 14.513379096984863, "learning_rate": 2.922393979261374e-06, "loss": 0.4443, "step": 12653 }, { "epoch": 0.6400930750164399, "grad_norm": 10.654691696166992, "learning_rate": 2.9216639613698857e-06, "loss": 0.4225, "step": 12654 }, { "epoch": 0.640143659264505, "grad_norm": 8.761181831359863, "learning_rate": 2.920933997032096e-06, "loss": 0.4172, "step": 12655 }, { "epoch": 0.6401942435125701, "grad_norm": 10.447986602783203, "learning_rate": 2.920204086266817e-06, "loss": 0.4155, "step": 12656 }, { "epoch": 0.6402448277606353, "grad_norm": 8.4111328125, "learning_rate": 2.919474229092854e-06, "loss": 0.3706, "step": 12657 }, { "epoch": 0.6402954120087004, "grad_norm": 9.140972137451172, "learning_rate": 2.918744425529016e-06, "loss": 0.474, "step": 12658 }, { "epoch": 0.6403459962567657, "grad_norm": 7.818044185638428, "learning_rate": 2.9180146755941053e-06, "loss": 0.4308, "step": 12659 }, { "epoch": 0.6403965805048308, "grad_norm": 6.65457820892334, "learning_rate": 2.9172849793069307e-06, "loss": 0.2512, "step": 12660 }, { "epoch": 0.640447164752896, "grad_norm": 8.717023849487305, "learning_rate": 2.916555336686288e-06, "loss": 0.3258, "step": 12661 }, { "epoch": 0.6404977490009611, "grad_norm": 4.097142696380615, "learning_rate": 2.9158257477509845e-06, "loss": 0.4162, "step": 12662 }, { "epoch": 0.6405483332490263, "grad_norm": 13.273998260498047, "learning_rate": 2.9150962125198166e-06, "loss": 0.4098, "step": 12663 }, { "epoch": 0.6405989174970914, "grad_norm": 9.174616813659668, "learning_rate": 2.9143667310115837e-06, "loss": 0.3889, "step": 12664 }, { "epoch": 0.6406495017451566, "grad_norm": 14.508645057678223, "learning_rate": 2.9136373032450803e-06, "loss": 0.4677, "step": 12665 }, { "epoch": 0.6407000859932217, "grad_norm": 7.381584644317627, "learning_rate": 2.912907929239106e-06, "loss": 0.4046, "step": 12666 }, { "epoch": 0.6407506702412868, "grad_norm": 15.346104621887207, "learning_rate": 2.9121786090124536e-06, "loss": 0.4842, "step": 12667 }, { "epoch": 0.640801254489352, "grad_norm": 5.1646504402160645, "learning_rate": 2.9114493425839143e-06, "loss": 0.327, "step": 12668 }, { "epoch": 0.6408518387374171, "grad_norm": 5.400955677032471, "learning_rate": 2.9107201299722816e-06, "loss": 0.3615, "step": 12669 }, { "epoch": 0.6409024229854823, "grad_norm": 8.821793556213379, "learning_rate": 2.909990971196345e-06, "loss": 0.4687, "step": 12670 }, { "epoch": 0.6409530072335474, "grad_norm": 5.621757507324219, "learning_rate": 2.9092618662748908e-06, "loss": 0.2467, "step": 12671 }, { "epoch": 0.6410035914816127, "grad_norm": 7.9813385009765625, "learning_rate": 2.908532815226711e-06, "loss": 0.3698, "step": 12672 }, { "epoch": 0.6410541757296778, "grad_norm": 6.8225884437561035, "learning_rate": 2.907803818070588e-06, "loss": 0.3214, "step": 12673 }, { "epoch": 0.641104759977743, "grad_norm": 7.805946350097656, "learning_rate": 2.9070748748253076e-06, "loss": 0.4042, "step": 12674 }, { "epoch": 0.6411553442258081, "grad_norm": 9.670101165771484, "learning_rate": 2.906345985509652e-06, "loss": 0.441, "step": 12675 }, { "epoch": 0.6412059284738733, "grad_norm": 4.702110290527344, "learning_rate": 2.9056171501424014e-06, "loss": 0.3513, "step": 12676 }, { "epoch": 0.6412565127219384, "grad_norm": 7.885443210601807, "learning_rate": 2.90488836874234e-06, "loss": 0.422, "step": 12677 }, { "epoch": 0.6413070969700035, "grad_norm": 12.683159828186035, "learning_rate": 2.9041596413282453e-06, "loss": 0.6091, "step": 12678 }, { "epoch": 0.6413576812180687, "grad_norm": 5.54789924621582, "learning_rate": 2.903430967918892e-06, "loss": 0.4126, "step": 12679 }, { "epoch": 0.6414082654661338, "grad_norm": 6.350280284881592, "learning_rate": 2.902702348533061e-06, "loss": 0.4985, "step": 12680 }, { "epoch": 0.641458849714199, "grad_norm": 10.318619728088379, "learning_rate": 2.901973783189524e-06, "loss": 0.4481, "step": 12681 }, { "epoch": 0.6415094339622641, "grad_norm": 7.550600051879883, "learning_rate": 2.901245271907053e-06, "loss": 0.4252, "step": 12682 }, { "epoch": 0.6415600182103293, "grad_norm": 5.814245700836182, "learning_rate": 2.900516814704425e-06, "loss": 0.3432, "step": 12683 }, { "epoch": 0.6416106024583944, "grad_norm": 21.877960205078125, "learning_rate": 2.8997884116004068e-06, "loss": 0.5161, "step": 12684 }, { "epoch": 0.6416611867064597, "grad_norm": 8.375909805297852, "learning_rate": 2.8990600626137665e-06, "loss": 0.3663, "step": 12685 }, { "epoch": 0.6417117709545248, "grad_norm": 8.47903823852539, "learning_rate": 2.898331767763278e-06, "loss": 0.5097, "step": 12686 }, { "epoch": 0.64176235520259, "grad_norm": 14.867531776428223, "learning_rate": 2.8976035270677006e-06, "loss": 0.5214, "step": 12687 }, { "epoch": 0.6418129394506551, "grad_norm": 8.18696403503418, "learning_rate": 2.896875340545801e-06, "loss": 0.4695, "step": 12688 }, { "epoch": 0.6418635236987202, "grad_norm": 7.217032432556152, "learning_rate": 2.896147208216346e-06, "loss": 0.4178, "step": 12689 }, { "epoch": 0.6419141079467854, "grad_norm": 12.090595245361328, "learning_rate": 2.895419130098096e-06, "loss": 0.4369, "step": 12690 }, { "epoch": 0.6419646921948505, "grad_norm": 14.124224662780762, "learning_rate": 2.8946911062098094e-06, "loss": 0.5111, "step": 12691 }, { "epoch": 0.6420152764429157, "grad_norm": 6.896515369415283, "learning_rate": 2.893963136570249e-06, "loss": 0.4152, "step": 12692 }, { "epoch": 0.6420658606909808, "grad_norm": 5.257626056671143, "learning_rate": 2.893235221198173e-06, "loss": 0.4098, "step": 12693 }, { "epoch": 0.642116444939046, "grad_norm": 9.114115715026855, "learning_rate": 2.8925073601123334e-06, "loss": 0.3375, "step": 12694 }, { "epoch": 0.6421670291871111, "grad_norm": 8.875996589660645, "learning_rate": 2.8917795533314913e-06, "loss": 0.4681, "step": 12695 }, { "epoch": 0.6422176134351762, "grad_norm": 9.82824420928955, "learning_rate": 2.8910518008743983e-06, "loss": 0.4549, "step": 12696 }, { "epoch": 0.6422681976832414, "grad_norm": 7.854160308837891, "learning_rate": 2.890324102759804e-06, "loss": 0.3704, "step": 12697 }, { "epoch": 0.6423187819313065, "grad_norm": 6.486033916473389, "learning_rate": 2.889596459006464e-06, "loss": 0.4433, "step": 12698 }, { "epoch": 0.6423693661793718, "grad_norm": 4.347931385040283, "learning_rate": 2.888868869633125e-06, "loss": 0.3757, "step": 12699 }, { "epoch": 0.6424199504274369, "grad_norm": 7.403172969818115, "learning_rate": 2.888141334658537e-06, "loss": 0.4779, "step": 12700 }, { "epoch": 0.6424705346755021, "grad_norm": 5.4427080154418945, "learning_rate": 2.887413854101445e-06, "loss": 0.353, "step": 12701 }, { "epoch": 0.6425211189235672, "grad_norm": 6.955048561096191, "learning_rate": 2.8866864279805963e-06, "loss": 0.4617, "step": 12702 }, { "epoch": 0.6425717031716324, "grad_norm": 6.404303550720215, "learning_rate": 2.885959056314731e-06, "loss": 0.3735, "step": 12703 }, { "epoch": 0.6426222874196975, "grad_norm": 6.341022491455078, "learning_rate": 2.885231739122597e-06, "loss": 0.3892, "step": 12704 }, { "epoch": 0.6426728716677627, "grad_norm": 7.002294540405273, "learning_rate": 2.884504476422933e-06, "loss": 0.3788, "step": 12705 }, { "epoch": 0.6427234559158278, "grad_norm": 8.700539588928223, "learning_rate": 2.883777268234477e-06, "loss": 0.4811, "step": 12706 }, { "epoch": 0.642774040163893, "grad_norm": 6.070414066314697, "learning_rate": 2.88305011457597e-06, "loss": 0.3906, "step": 12707 }, { "epoch": 0.6428246244119581, "grad_norm": 8.413270950317383, "learning_rate": 2.88232301546615e-06, "loss": 0.3532, "step": 12708 }, { "epoch": 0.6428752086600232, "grad_norm": 5.868719577789307, "learning_rate": 2.8815959709237484e-06, "loss": 0.4511, "step": 12709 }, { "epoch": 0.6429257929080884, "grad_norm": 9.95273494720459, "learning_rate": 2.8808689809675034e-06, "loss": 0.4321, "step": 12710 }, { "epoch": 0.6429763771561535, "grad_norm": 10.816601753234863, "learning_rate": 2.8801420456161467e-06, "loss": 0.4129, "step": 12711 }, { "epoch": 0.6430269614042188, "grad_norm": 15.289459228515625, "learning_rate": 2.8794151648884082e-06, "loss": 0.5118, "step": 12712 }, { "epoch": 0.6430775456522839, "grad_norm": 5.716133117675781, "learning_rate": 2.87868833880302e-06, "loss": 0.3598, "step": 12713 }, { "epoch": 0.6431281299003491, "grad_norm": 6.224432945251465, "learning_rate": 2.877961567378707e-06, "loss": 0.362, "step": 12714 }, { "epoch": 0.6431787141484142, "grad_norm": 10.290605545043945, "learning_rate": 2.877234850634201e-06, "loss": 0.4899, "step": 12715 }, { "epoch": 0.6432292983964794, "grad_norm": 8.360932350158691, "learning_rate": 2.876508188588225e-06, "loss": 0.3421, "step": 12716 }, { "epoch": 0.6432798826445445, "grad_norm": 16.420166015625, "learning_rate": 2.8757815812595026e-06, "loss": 0.6464, "step": 12717 }, { "epoch": 0.6433304668926096, "grad_norm": 7.802995204925537, "learning_rate": 2.87505502866676e-06, "loss": 0.3695, "step": 12718 }, { "epoch": 0.6433810511406748, "grad_norm": 11.763551712036133, "learning_rate": 2.874328530828716e-06, "loss": 0.4706, "step": 12719 }, { "epoch": 0.6434316353887399, "grad_norm": 5.66304349899292, "learning_rate": 2.87360208776409e-06, "loss": 0.3912, "step": 12720 }, { "epoch": 0.6434822196368051, "grad_norm": 12.689393997192383, "learning_rate": 2.872875699491604e-06, "loss": 0.4111, "step": 12721 }, { "epoch": 0.6435328038848702, "grad_norm": 4.863962650299072, "learning_rate": 2.8721493660299737e-06, "loss": 0.4228, "step": 12722 }, { "epoch": 0.6435833881329354, "grad_norm": 7.978920936584473, "learning_rate": 2.871423087397912e-06, "loss": 0.4334, "step": 12723 }, { "epoch": 0.6436339723810005, "grad_norm": 5.578015327453613, "learning_rate": 2.870696863614138e-06, "loss": 0.3465, "step": 12724 }, { "epoch": 0.6436845566290658, "grad_norm": 6.203045845031738, "learning_rate": 2.869970694697365e-06, "loss": 0.2836, "step": 12725 }, { "epoch": 0.6437351408771309, "grad_norm": 6.244227886199951, "learning_rate": 2.8692445806662985e-06, "loss": 0.3575, "step": 12726 }, { "epoch": 0.643785725125196, "grad_norm": 11.92608642578125, "learning_rate": 2.868518521539654e-06, "loss": 0.4298, "step": 12727 }, { "epoch": 0.6438363093732612, "grad_norm": 10.104618072509766, "learning_rate": 2.867792517336139e-06, "loss": 0.5165, "step": 12728 }, { "epoch": 0.6438868936213263, "grad_norm": 10.60081958770752, "learning_rate": 2.867066568074459e-06, "loss": 0.471, "step": 12729 }, { "epoch": 0.6439374778693915, "grad_norm": 7.135310173034668, "learning_rate": 2.8663406737733233e-06, "loss": 0.3163, "step": 12730 }, { "epoch": 0.6439880621174566, "grad_norm": 9.210079193115234, "learning_rate": 2.865614834451435e-06, "loss": 0.4784, "step": 12731 }, { "epoch": 0.6440386463655218, "grad_norm": 22.185619354248047, "learning_rate": 2.864889050127495e-06, "loss": 0.5808, "step": 12732 }, { "epoch": 0.6440892306135869, "grad_norm": 5.762813568115234, "learning_rate": 2.864163320820209e-06, "loss": 0.2983, "step": 12733 }, { "epoch": 0.6441398148616521, "grad_norm": 9.261116027832031, "learning_rate": 2.863437646548275e-06, "loss": 0.4485, "step": 12734 }, { "epoch": 0.6441903991097172, "grad_norm": 6.755384922027588, "learning_rate": 2.862712027330391e-06, "loss": 0.3887, "step": 12735 }, { "epoch": 0.6442409833577823, "grad_norm": 6.601161956787109, "learning_rate": 2.8619864631852567e-06, "loss": 0.3978, "step": 12736 }, { "epoch": 0.6442915676058475, "grad_norm": 11.627314567565918, "learning_rate": 2.861260954131567e-06, "loss": 0.4087, "step": 12737 }, { "epoch": 0.6443421518539126, "grad_norm": 11.062273979187012, "learning_rate": 2.8605355001880176e-06, "loss": 0.5781, "step": 12738 }, { "epoch": 0.6443927361019779, "grad_norm": 13.68985366821289, "learning_rate": 2.8598101013733004e-06, "loss": 0.4426, "step": 12739 }, { "epoch": 0.644443320350043, "grad_norm": 8.207540512084961, "learning_rate": 2.8590847577061078e-06, "loss": 0.4406, "step": 12740 }, { "epoch": 0.6444939045981082, "grad_norm": 8.29207706451416, "learning_rate": 2.8583594692051262e-06, "loss": 0.3604, "step": 12741 }, { "epoch": 0.6445444888461733, "grad_norm": 8.348740577697754, "learning_rate": 2.857634235889052e-06, "loss": 0.347, "step": 12742 }, { "epoch": 0.6445950730942385, "grad_norm": 13.23281192779541, "learning_rate": 2.856909057776568e-06, "loss": 0.3426, "step": 12743 }, { "epoch": 0.6446456573423036, "grad_norm": 14.016658782958984, "learning_rate": 2.8561839348863597e-06, "loss": 0.3903, "step": 12744 }, { "epoch": 0.6446962415903688, "grad_norm": 9.415715217590332, "learning_rate": 2.855458867237114e-06, "loss": 0.4098, "step": 12745 }, { "epoch": 0.6447468258384339, "grad_norm": 14.46558666229248, "learning_rate": 2.8547338548475146e-06, "loss": 0.5669, "step": 12746 }, { "epoch": 0.644797410086499, "grad_norm": 7.001288414001465, "learning_rate": 2.8540088977362394e-06, "loss": 0.439, "step": 12747 }, { "epoch": 0.6448479943345642, "grad_norm": 7.965309143066406, "learning_rate": 2.8532839959219737e-06, "loss": 0.3717, "step": 12748 }, { "epoch": 0.6448985785826293, "grad_norm": 10.91471004486084, "learning_rate": 2.8525591494233945e-06, "loss": 0.4138, "step": 12749 }, { "epoch": 0.6449491628306945, "grad_norm": 6.837155818939209, "learning_rate": 2.851834358259179e-06, "loss": 0.4139, "step": 12750 }, { "epoch": 0.6449997470787596, "grad_norm": 5.539379119873047, "learning_rate": 2.851109622448003e-06, "loss": 0.4259, "step": 12751 }, { "epoch": 0.6450503313268249, "grad_norm": 12.673676490783691, "learning_rate": 2.85038494200854e-06, "loss": 0.5426, "step": 12752 }, { "epoch": 0.64510091557489, "grad_norm": 9.557076454162598, "learning_rate": 2.8496603169594665e-06, "loss": 0.4817, "step": 12753 }, { "epoch": 0.6451514998229552, "grad_norm": 6.75907564163208, "learning_rate": 2.8489357473194535e-06, "loss": 0.4573, "step": 12754 }, { "epoch": 0.6452020840710203, "grad_norm": 9.341264724731445, "learning_rate": 2.8482112331071675e-06, "loss": 0.4728, "step": 12755 }, { "epoch": 0.6452526683190855, "grad_norm": 7.978724002838135, "learning_rate": 2.847486774341284e-06, "loss": 0.431, "step": 12756 }, { "epoch": 0.6453032525671506, "grad_norm": 7.429305553436279, "learning_rate": 2.8467623710404657e-06, "loss": 0.4462, "step": 12757 }, { "epoch": 0.6453538368152157, "grad_norm": 12.508317947387695, "learning_rate": 2.8460380232233786e-06, "loss": 0.5105, "step": 12758 }, { "epoch": 0.6454044210632809, "grad_norm": 4.5906901359558105, "learning_rate": 2.8453137309086907e-06, "loss": 0.3461, "step": 12759 }, { "epoch": 0.645455005311346, "grad_norm": 6.4338274002075195, "learning_rate": 2.8445894941150644e-06, "loss": 0.402, "step": 12760 }, { "epoch": 0.6455055895594112, "grad_norm": 6.778567314147949, "learning_rate": 2.843865312861158e-06, "loss": 0.3379, "step": 12761 }, { "epoch": 0.6455561738074763, "grad_norm": 7.067078113555908, "learning_rate": 2.843141187165637e-06, "loss": 0.4251, "step": 12762 }, { "epoch": 0.6456067580555415, "grad_norm": 6.0021138191223145, "learning_rate": 2.8424171170471592e-06, "loss": 0.3856, "step": 12763 }, { "epoch": 0.6456573423036066, "grad_norm": 9.030142784118652, "learning_rate": 2.8416931025243778e-06, "loss": 0.5038, "step": 12764 }, { "epoch": 0.6457079265516719, "grad_norm": 8.310461044311523, "learning_rate": 2.840969143615954e-06, "loss": 0.3781, "step": 12765 }, { "epoch": 0.645758510799737, "grad_norm": 8.38630485534668, "learning_rate": 2.8402452403405396e-06, "loss": 0.346, "step": 12766 }, { "epoch": 0.6458090950478022, "grad_norm": 9.4235258102417, "learning_rate": 2.839521392716787e-06, "loss": 0.3341, "step": 12767 }, { "epoch": 0.6458596792958673, "grad_norm": 15.056387901306152, "learning_rate": 2.8387976007633516e-06, "loss": 0.6242, "step": 12768 }, { "epoch": 0.6459102635439324, "grad_norm": 7.65696907043457, "learning_rate": 2.838073864498882e-06, "loss": 0.405, "step": 12769 }, { "epoch": 0.6459608477919976, "grad_norm": 6.3453521728515625, "learning_rate": 2.8373501839420247e-06, "loss": 0.3578, "step": 12770 }, { "epoch": 0.6460114320400627, "grad_norm": 3.841921806335449, "learning_rate": 2.836626559111432e-06, "loss": 0.3039, "step": 12771 }, { "epoch": 0.6460620162881279, "grad_norm": 6.141747951507568, "learning_rate": 2.835902990025747e-06, "loss": 0.3155, "step": 12772 }, { "epoch": 0.646112600536193, "grad_norm": 10.354227066040039, "learning_rate": 2.8351794767036134e-06, "loss": 0.492, "step": 12773 }, { "epoch": 0.6461631847842582, "grad_norm": 6.773136138916016, "learning_rate": 2.834456019163677e-06, "loss": 0.4171, "step": 12774 }, { "epoch": 0.6462137690323233, "grad_norm": 5.696592330932617, "learning_rate": 2.8337326174245788e-06, "loss": 0.3493, "step": 12775 }, { "epoch": 0.6462643532803884, "grad_norm": 8.037424087524414, "learning_rate": 2.8330092715049584e-06, "loss": 0.3977, "step": 12776 }, { "epoch": 0.6463149375284536, "grad_norm": 8.748023986816406, "learning_rate": 2.8322859814234553e-06, "loss": 0.3515, "step": 12777 }, { "epoch": 0.6463655217765187, "grad_norm": 9.993900299072266, "learning_rate": 2.831562747198706e-06, "loss": 0.3355, "step": 12778 }, { "epoch": 0.646416106024584, "grad_norm": 30.458717346191406, "learning_rate": 2.830839568849346e-06, "loss": 0.5858, "step": 12779 }, { "epoch": 0.6464666902726491, "grad_norm": 8.902369499206543, "learning_rate": 2.830116446394012e-06, "loss": 0.4157, "step": 12780 }, { "epoch": 0.6465172745207143, "grad_norm": 14.007990837097168, "learning_rate": 2.829393379851336e-06, "loss": 0.4794, "step": 12781 }, { "epoch": 0.6465678587687794, "grad_norm": 8.024587631225586, "learning_rate": 2.8286703692399484e-06, "loss": 0.4457, "step": 12782 }, { "epoch": 0.6466184430168446, "grad_norm": 7.098682880401611, "learning_rate": 2.8279474145784823e-06, "loss": 0.3378, "step": 12783 }, { "epoch": 0.6466690272649097, "grad_norm": 6.6799211502075195, "learning_rate": 2.827224515885565e-06, "loss": 0.411, "step": 12784 }, { "epoch": 0.6467196115129749, "grad_norm": 12.823835372924805, "learning_rate": 2.826501673179822e-06, "loss": 0.4821, "step": 12785 }, { "epoch": 0.64677019576104, "grad_norm": 13.326560020446777, "learning_rate": 2.825778886479883e-06, "loss": 0.338, "step": 12786 }, { "epoch": 0.6468207800091051, "grad_norm": 5.9377593994140625, "learning_rate": 2.8250561558043706e-06, "loss": 0.3791, "step": 12787 }, { "epoch": 0.6468713642571703, "grad_norm": 7.225045680999756, "learning_rate": 2.8243334811719054e-06, "loss": 0.3938, "step": 12788 }, { "epoch": 0.6469219485052354, "grad_norm": 16.41000747680664, "learning_rate": 2.823610862601115e-06, "loss": 0.4591, "step": 12789 }, { "epoch": 0.6469725327533006, "grad_norm": 5.333596229553223, "learning_rate": 2.8228883001106124e-06, "loss": 0.4104, "step": 12790 }, { "epoch": 0.6470231170013657, "grad_norm": 17.000465393066406, "learning_rate": 2.8221657937190215e-06, "loss": 0.5737, "step": 12791 }, { "epoch": 0.647073701249431, "grad_norm": 12.102569580078125, "learning_rate": 2.8214433434449574e-06, "loss": 0.4334, "step": 12792 }, { "epoch": 0.6471242854974961, "grad_norm": 4.34926176071167, "learning_rate": 2.8207209493070346e-06, "loss": 0.401, "step": 12793 }, { "epoch": 0.6471748697455613, "grad_norm": 5.7434916496276855, "learning_rate": 2.8199986113238713e-06, "loss": 0.262, "step": 12794 }, { "epoch": 0.6472254539936264, "grad_norm": 8.112190246582031, "learning_rate": 2.819276329514078e-06, "loss": 0.5228, "step": 12795 }, { "epoch": 0.6472760382416916, "grad_norm": 8.18864631652832, "learning_rate": 2.8185541038962643e-06, "loss": 0.3591, "step": 12796 }, { "epoch": 0.6473266224897567, "grad_norm": 9.566946983337402, "learning_rate": 2.817831934489044e-06, "loss": 0.4532, "step": 12797 }, { "epoch": 0.6473772067378218, "grad_norm": 9.042962074279785, "learning_rate": 2.817109821311025e-06, "loss": 0.4624, "step": 12798 }, { "epoch": 0.647427790985887, "grad_norm": 9.304644584655762, "learning_rate": 2.8163877643808124e-06, "loss": 0.503, "step": 12799 }, { "epoch": 0.6474783752339521, "grad_norm": 6.46265172958374, "learning_rate": 2.815665763717011e-06, "loss": 0.3623, "step": 12800 }, { "epoch": 0.6475289594820173, "grad_norm": 11.457961082458496, "learning_rate": 2.8149438193382285e-06, "loss": 0.3717, "step": 12801 }, { "epoch": 0.6475795437300824, "grad_norm": 8.630016326904297, "learning_rate": 2.814221931263067e-06, "loss": 0.3484, "step": 12802 }, { "epoch": 0.6476301279781476, "grad_norm": 6.8682861328125, "learning_rate": 2.813500099510126e-06, "loss": 0.4322, "step": 12803 }, { "epoch": 0.6476807122262127, "grad_norm": 5.85689640045166, "learning_rate": 2.812778324098006e-06, "loss": 0.3283, "step": 12804 }, { "epoch": 0.647731296474278, "grad_norm": 6.150325298309326, "learning_rate": 2.8120566050453037e-06, "loss": 0.4227, "step": 12805 }, { "epoch": 0.6477818807223431, "grad_norm": 5.144414901733398, "learning_rate": 2.8113349423706205e-06, "loss": 0.3088, "step": 12806 }, { "epoch": 0.6478324649704083, "grad_norm": 8.56081485748291, "learning_rate": 2.8106133360925484e-06, "loss": 0.4109, "step": 12807 }, { "epoch": 0.6478830492184734, "grad_norm": 8.162858963012695, "learning_rate": 2.8098917862296815e-06, "loss": 0.3983, "step": 12808 }, { "epoch": 0.6479336334665385, "grad_norm": 12.458921432495117, "learning_rate": 2.8091702928006146e-06, "loss": 0.4002, "step": 12809 }, { "epoch": 0.6479842177146037, "grad_norm": 11.777684211730957, "learning_rate": 2.8084488558239377e-06, "loss": 0.4801, "step": 12810 }, { "epoch": 0.6480348019626688, "grad_norm": 5.723401069641113, "learning_rate": 2.8077274753182384e-06, "loss": 0.4014, "step": 12811 }, { "epoch": 0.648085386210734, "grad_norm": 5.801550388336182, "learning_rate": 2.8070061513021086e-06, "loss": 0.4775, "step": 12812 }, { "epoch": 0.6481359704587991, "grad_norm": 12.455366134643555, "learning_rate": 2.8062848837941332e-06, "loss": 0.3887, "step": 12813 }, { "epoch": 0.6481865547068643, "grad_norm": 6.859375476837158, "learning_rate": 2.8055636728128956e-06, "loss": 0.3788, "step": 12814 }, { "epoch": 0.6482371389549294, "grad_norm": 10.894976615905762, "learning_rate": 2.8048425183769857e-06, "loss": 0.3512, "step": 12815 }, { "epoch": 0.6482877232029945, "grad_norm": 21.056581497192383, "learning_rate": 2.8041214205049805e-06, "loss": 0.505, "step": 12816 }, { "epoch": 0.6483383074510597, "grad_norm": 12.694046974182129, "learning_rate": 2.80340037921546e-06, "loss": 0.4106, "step": 12817 }, { "epoch": 0.648388891699125, "grad_norm": 6.75807523727417, "learning_rate": 2.802679394527008e-06, "loss": 0.4692, "step": 12818 }, { "epoch": 0.6484394759471901, "grad_norm": 11.309453010559082, "learning_rate": 2.801958466458201e-06, "loss": 0.4599, "step": 12819 }, { "epoch": 0.6484900601952552, "grad_norm": 13.500968933105469, "learning_rate": 2.801237595027613e-06, "loss": 0.4571, "step": 12820 }, { "epoch": 0.6485406444433204, "grad_norm": 6.1705241203308105, "learning_rate": 2.8005167802538223e-06, "loss": 0.2994, "step": 12821 }, { "epoch": 0.6485912286913855, "grad_norm": 5.296753883361816, "learning_rate": 2.7997960221554026e-06, "loss": 0.3921, "step": 12822 }, { "epoch": 0.6486418129394507, "grad_norm": 13.311442375183105, "learning_rate": 2.799075320750923e-06, "loss": 0.5402, "step": 12823 }, { "epoch": 0.6486923971875158, "grad_norm": 12.049827575683594, "learning_rate": 2.7983546760589582e-06, "loss": 0.4819, "step": 12824 }, { "epoch": 0.648742981435581, "grad_norm": 18.397544860839844, "learning_rate": 2.7976340880980755e-06, "loss": 0.6845, "step": 12825 }, { "epoch": 0.6487935656836461, "grad_norm": 8.997719764709473, "learning_rate": 2.7969135568868404e-06, "loss": 0.4567, "step": 12826 }, { "epoch": 0.6488441499317112, "grad_norm": 5.1942644119262695, "learning_rate": 2.7961930824438238e-06, "loss": 0.3577, "step": 12827 }, { "epoch": 0.6488947341797764, "grad_norm": 7.714773654937744, "learning_rate": 2.7954726647875883e-06, "loss": 0.4591, "step": 12828 }, { "epoch": 0.6489453184278415, "grad_norm": 5.412215232849121, "learning_rate": 2.7947523039366975e-06, "loss": 0.5141, "step": 12829 }, { "epoch": 0.6489959026759067, "grad_norm": 6.088029861450195, "learning_rate": 2.794031999909714e-06, "loss": 0.3213, "step": 12830 }, { "epoch": 0.6490464869239718, "grad_norm": 5.774286270141602, "learning_rate": 2.7933117527251947e-06, "loss": 0.3953, "step": 12831 }, { "epoch": 0.6490970711720371, "grad_norm": 8.149693489074707, "learning_rate": 2.7925915624017035e-06, "loss": 0.3868, "step": 12832 }, { "epoch": 0.6491476554201022, "grad_norm": 6.606420040130615, "learning_rate": 2.7918714289577965e-06, "loss": 0.3699, "step": 12833 }, { "epoch": 0.6491982396681674, "grad_norm": 6.161722660064697, "learning_rate": 2.7911513524120267e-06, "loss": 0.3655, "step": 12834 }, { "epoch": 0.6492488239162325, "grad_norm": 15.471918106079102, "learning_rate": 2.7904313327829534e-06, "loss": 0.5839, "step": 12835 }, { "epoch": 0.6492994081642977, "grad_norm": 8.605989456176758, "learning_rate": 2.7897113700891275e-06, "loss": 0.4084, "step": 12836 }, { "epoch": 0.6493499924123628, "grad_norm": 6.730690002441406, "learning_rate": 2.788991464349101e-06, "loss": 0.4804, "step": 12837 }, { "epoch": 0.6494005766604279, "grad_norm": 12.953554153442383, "learning_rate": 2.7882716155814226e-06, "loss": 0.5515, "step": 12838 }, { "epoch": 0.6494511609084931, "grad_norm": 9.087347984313965, "learning_rate": 2.787551823804644e-06, "loss": 0.3183, "step": 12839 }, { "epoch": 0.6495017451565582, "grad_norm": 9.365781784057617, "learning_rate": 2.786832089037311e-06, "loss": 0.2965, "step": 12840 }, { "epoch": 0.6495523294046234, "grad_norm": 9.474822044372559, "learning_rate": 2.7861124112979694e-06, "loss": 0.5244, "step": 12841 }, { "epoch": 0.6496029136526885, "grad_norm": 11.666592597961426, "learning_rate": 2.785392790605163e-06, "loss": 0.5414, "step": 12842 }, { "epoch": 0.6496534979007537, "grad_norm": 24.299516677856445, "learning_rate": 2.7846732269774335e-06, "loss": 0.3852, "step": 12843 }, { "epoch": 0.6497040821488188, "grad_norm": 14.710981369018555, "learning_rate": 2.7839537204333266e-06, "loss": 0.4583, "step": 12844 }, { "epoch": 0.6497546663968841, "grad_norm": 8.598642349243164, "learning_rate": 2.7832342709913795e-06, "loss": 0.4271, "step": 12845 }, { "epoch": 0.6498052506449492, "grad_norm": 7.161705493927002, "learning_rate": 2.7825148786701284e-06, "loss": 0.4475, "step": 12846 }, { "epoch": 0.6498558348930144, "grad_norm": 10.120142936706543, "learning_rate": 2.781795543488115e-06, "loss": 0.4229, "step": 12847 }, { "epoch": 0.6499064191410795, "grad_norm": 5.593669414520264, "learning_rate": 2.7810762654638722e-06, "loss": 0.4026, "step": 12848 }, { "epoch": 0.6499570033891446, "grad_norm": 9.142111778259277, "learning_rate": 2.7803570446159333e-06, "loss": 0.4284, "step": 12849 }, { "epoch": 0.6500075876372098, "grad_norm": 8.475272178649902, "learning_rate": 2.779637880962833e-06, "loss": 0.3119, "step": 12850 }, { "epoch": 0.6500581718852749, "grad_norm": 5.459552764892578, "learning_rate": 2.778918774523103e-06, "loss": 0.3357, "step": 12851 }, { "epoch": 0.6501087561333401, "grad_norm": 6.465392589569092, "learning_rate": 2.7781997253152683e-06, "loss": 0.3566, "step": 12852 }, { "epoch": 0.6501593403814052, "grad_norm": 9.085142135620117, "learning_rate": 2.7774807333578633e-06, "loss": 0.4903, "step": 12853 }, { "epoch": 0.6502099246294704, "grad_norm": 7.356816291809082, "learning_rate": 2.7767617986694105e-06, "loss": 0.4695, "step": 12854 }, { "epoch": 0.6502605088775355, "grad_norm": 8.232625961303711, "learning_rate": 2.776042921268435e-06, "loss": 0.4678, "step": 12855 }, { "epoch": 0.6503110931256006, "grad_norm": 6.8845906257629395, "learning_rate": 2.775324101173462e-06, "loss": 0.3419, "step": 12856 }, { "epoch": 0.6503616773736658, "grad_norm": 8.538861274719238, "learning_rate": 2.7746053384030146e-06, "loss": 0.3417, "step": 12857 }, { "epoch": 0.650412261621731, "grad_norm": 11.794041633605957, "learning_rate": 2.7738866329756096e-06, "loss": 0.5252, "step": 12858 }, { "epoch": 0.6504628458697962, "grad_norm": 26.188579559326172, "learning_rate": 2.7731679849097713e-06, "loss": 0.5051, "step": 12859 }, { "epoch": 0.6505134301178613, "grad_norm": 7.04525089263916, "learning_rate": 2.7724493942240162e-06, "loss": 0.4335, "step": 12860 }, { "epoch": 0.6505640143659265, "grad_norm": 10.375739097595215, "learning_rate": 2.771730860936857e-06, "loss": 0.3812, "step": 12861 }, { "epoch": 0.6506145986139916, "grad_norm": 6.170292377471924, "learning_rate": 2.7710123850668137e-06, "loss": 0.3611, "step": 12862 }, { "epoch": 0.6506651828620568, "grad_norm": 8.302926063537598, "learning_rate": 2.770293966632397e-06, "loss": 0.4367, "step": 12863 }, { "epoch": 0.6507157671101219, "grad_norm": 4.961328983306885, "learning_rate": 2.769575605652117e-06, "loss": 0.2374, "step": 12864 }, { "epoch": 0.6507663513581871, "grad_norm": 12.79229736328125, "learning_rate": 2.7688573021444888e-06, "loss": 0.4441, "step": 12865 }, { "epoch": 0.6508169356062522, "grad_norm": 10.294708251953125, "learning_rate": 2.7681390561280185e-06, "loss": 0.4632, "step": 12866 }, { "epoch": 0.6508675198543173, "grad_norm": 8.330667495727539, "learning_rate": 2.767420867621214e-06, "loss": 0.5737, "step": 12867 }, { "epoch": 0.6509181041023825, "grad_norm": 9.082901954650879, "learning_rate": 2.7667027366425807e-06, "loss": 0.3489, "step": 12868 }, { "epoch": 0.6509686883504476, "grad_norm": 5.429197311401367, "learning_rate": 2.7659846632106226e-06, "loss": 0.3705, "step": 12869 }, { "epoch": 0.6510192725985128, "grad_norm": 6.7133097648620605, "learning_rate": 2.765266647343845e-06, "loss": 0.3947, "step": 12870 }, { "epoch": 0.6510698568465779, "grad_norm": 7.143261909484863, "learning_rate": 2.7645486890607474e-06, "loss": 0.3329, "step": 12871 }, { "epoch": 0.6511204410946432, "grad_norm": 12.964766502380371, "learning_rate": 2.76383078837983e-06, "loss": 0.5484, "step": 12872 }, { "epoch": 0.6511710253427083, "grad_norm": 6.087304592132568, "learning_rate": 2.7631129453195933e-06, "loss": 0.3265, "step": 12873 }, { "epoch": 0.6512216095907735, "grad_norm": 10.723762512207031, "learning_rate": 2.7623951598985333e-06, "loss": 0.5814, "step": 12874 }, { "epoch": 0.6512721938388386, "grad_norm": 7.999849796295166, "learning_rate": 2.7616774321351446e-06, "loss": 0.419, "step": 12875 }, { "epoch": 0.6513227780869038, "grad_norm": 5.844414710998535, "learning_rate": 2.76095976204792e-06, "loss": 0.3686, "step": 12876 }, { "epoch": 0.6513733623349689, "grad_norm": 8.40384292602539, "learning_rate": 2.7602421496553568e-06, "loss": 0.3222, "step": 12877 }, { "epoch": 0.651423946583034, "grad_norm": 5.45389461517334, "learning_rate": 2.759524594975944e-06, "loss": 0.3217, "step": 12878 }, { "epoch": 0.6514745308310992, "grad_norm": 5.487911701202393, "learning_rate": 2.7588070980281696e-06, "loss": 0.2662, "step": 12879 }, { "epoch": 0.6515251150791643, "grad_norm": 7.785265922546387, "learning_rate": 2.7580896588305233e-06, "loss": 0.4257, "step": 12880 }, { "epoch": 0.6515756993272295, "grad_norm": 6.717867851257324, "learning_rate": 2.7573722774014896e-06, "loss": 0.3436, "step": 12881 }, { "epoch": 0.6516262835752946, "grad_norm": 5.632620334625244, "learning_rate": 2.756654953759557e-06, "loss": 0.3994, "step": 12882 }, { "epoch": 0.6516768678233598, "grad_norm": 10.273316383361816, "learning_rate": 2.755937687923208e-06, "loss": 0.4442, "step": 12883 }, { "epoch": 0.6517274520714249, "grad_norm": 8.036640167236328, "learning_rate": 2.7552204799109227e-06, "loss": 0.4118, "step": 12884 }, { "epoch": 0.6517780363194902, "grad_norm": 6.6981072425842285, "learning_rate": 2.754503329741185e-06, "loss": 0.3537, "step": 12885 }, { "epoch": 0.6518286205675553, "grad_norm": 7.7191667556762695, "learning_rate": 2.753786237432473e-06, "loss": 0.3925, "step": 12886 }, { "epoch": 0.6518792048156204, "grad_norm": 14.476279258728027, "learning_rate": 2.7530692030032625e-06, "loss": 0.5687, "step": 12887 }, { "epoch": 0.6519297890636856, "grad_norm": 8.521806716918945, "learning_rate": 2.7523522264720326e-06, "loss": 0.3829, "step": 12888 }, { "epoch": 0.6519803733117507, "grad_norm": 6.697693824768066, "learning_rate": 2.751635307857258e-06, "loss": 0.3313, "step": 12889 }, { "epoch": 0.6520309575598159, "grad_norm": 10.179585456848145, "learning_rate": 2.7509184471774076e-06, "loss": 0.4534, "step": 12890 }, { "epoch": 0.652081541807881, "grad_norm": 6.241157054901123, "learning_rate": 2.7502016444509595e-06, "loss": 0.4455, "step": 12891 }, { "epoch": 0.6521321260559462, "grad_norm": 7.528390884399414, "learning_rate": 2.749484899696382e-06, "loss": 0.4382, "step": 12892 }, { "epoch": 0.6521827103040113, "grad_norm": 6.920678615570068, "learning_rate": 2.748768212932139e-06, "loss": 0.4225, "step": 12893 }, { "epoch": 0.6522332945520765, "grad_norm": 11.130766868591309, "learning_rate": 2.7480515841767035e-06, "loss": 0.4651, "step": 12894 }, { "epoch": 0.6522838788001416, "grad_norm": 7.7132568359375, "learning_rate": 2.7473350134485393e-06, "loss": 0.3864, "step": 12895 }, { "epoch": 0.6523344630482067, "grad_norm": 8.473177909851074, "learning_rate": 2.746618500766109e-06, "loss": 0.385, "step": 12896 }, { "epoch": 0.6523850472962719, "grad_norm": 6.370852470397949, "learning_rate": 2.745902046147879e-06, "loss": 0.4273, "step": 12897 }, { "epoch": 0.6524356315443371, "grad_norm": 10.548028945922852, "learning_rate": 2.7451856496123086e-06, "loss": 0.3714, "step": 12898 }, { "epoch": 0.6524862157924023, "grad_norm": 10.221219062805176, "learning_rate": 2.7444693111778552e-06, "loss": 0.4434, "step": 12899 }, { "epoch": 0.6525368000404674, "grad_norm": 7.904365062713623, "learning_rate": 2.743753030862982e-06, "loss": 0.436, "step": 12900 }, { "epoch": 0.6525873842885326, "grad_norm": 12.074228286743164, "learning_rate": 2.743036808686144e-06, "loss": 0.4681, "step": 12901 }, { "epoch": 0.6526379685365977, "grad_norm": 7.745822906494141, "learning_rate": 2.7423206446657936e-06, "loss": 0.4311, "step": 12902 }, { "epoch": 0.6526885527846629, "grad_norm": 6.810887336730957, "learning_rate": 2.7416045388203894e-06, "loss": 0.417, "step": 12903 }, { "epoch": 0.652739137032728, "grad_norm": 12.974764823913574, "learning_rate": 2.740888491168381e-06, "loss": 0.3762, "step": 12904 }, { "epoch": 0.6527897212807932, "grad_norm": 8.14237117767334, "learning_rate": 2.740172501728219e-06, "loss": 0.3379, "step": 12905 }, { "epoch": 0.6528403055288583, "grad_norm": 6.644782543182373, "learning_rate": 2.739456570518354e-06, "loss": 0.4518, "step": 12906 }, { "epoch": 0.6528908897769234, "grad_norm": 5.608822345733643, "learning_rate": 2.7387406975572306e-06, "loss": 0.3727, "step": 12907 }, { "epoch": 0.6529414740249886, "grad_norm": 10.763813972473145, "learning_rate": 2.7380248828632995e-06, "loss": 0.5052, "step": 12908 }, { "epoch": 0.6529920582730537, "grad_norm": 5.620895862579346, "learning_rate": 2.737309126455004e-06, "loss": 0.337, "step": 12909 }, { "epoch": 0.6530426425211189, "grad_norm": 6.232489109039307, "learning_rate": 2.7365934283507865e-06, "loss": 0.3455, "step": 12910 }, { "epoch": 0.653093226769184, "grad_norm": 6.841986656188965, "learning_rate": 2.735877788569088e-06, "loss": 0.3389, "step": 12911 }, { "epoch": 0.6531438110172493, "grad_norm": 6.1805315017700195, "learning_rate": 2.7351622071283512e-06, "loss": 0.334, "step": 12912 }, { "epoch": 0.6531943952653144, "grad_norm": 7.286762714385986, "learning_rate": 2.7344466840470145e-06, "loss": 0.4716, "step": 12913 }, { "epoch": 0.6532449795133796, "grad_norm": 5.949853420257568, "learning_rate": 2.733731219343513e-06, "loss": 0.3252, "step": 12914 }, { "epoch": 0.6532955637614447, "grad_norm": 6.242796421051025, "learning_rate": 2.7330158130362848e-06, "loss": 0.3978, "step": 12915 }, { "epoch": 0.6533461480095099, "grad_norm": 14.642936706542969, "learning_rate": 2.7323004651437648e-06, "loss": 0.4787, "step": 12916 }, { "epoch": 0.653396732257575, "grad_norm": 8.678414344787598, "learning_rate": 2.7315851756843814e-06, "loss": 0.397, "step": 12917 }, { "epoch": 0.6534473165056401, "grad_norm": 10.022432327270508, "learning_rate": 2.730869944676573e-06, "loss": 0.5435, "step": 12918 }, { "epoch": 0.6534979007537053, "grad_norm": 6.807836055755615, "learning_rate": 2.7301547721387613e-06, "loss": 0.3785, "step": 12919 }, { "epoch": 0.6535484850017704, "grad_norm": 7.420170783996582, "learning_rate": 2.7294396580893808e-06, "loss": 0.4098, "step": 12920 }, { "epoch": 0.6535990692498356, "grad_norm": 9.02357006072998, "learning_rate": 2.7287246025468554e-06, "loss": 0.4565, "step": 12921 }, { "epoch": 0.6536496534979007, "grad_norm": 9.617171287536621, "learning_rate": 2.7280096055296094e-06, "loss": 0.4291, "step": 12922 }, { "epoch": 0.6537002377459659, "grad_norm": 10.802444458007812, "learning_rate": 2.7272946670560706e-06, "loss": 0.4293, "step": 12923 }, { "epoch": 0.653750821994031, "grad_norm": 8.0933256149292, "learning_rate": 2.726579787144658e-06, "loss": 0.446, "step": 12924 }, { "epoch": 0.6538014062420963, "grad_norm": 8.493917465209961, "learning_rate": 2.7258649658137914e-06, "loss": 0.37, "step": 12925 }, { "epoch": 0.6538519904901614, "grad_norm": 8.669864654541016, "learning_rate": 2.7251502030818933e-06, "loss": 0.3814, "step": 12926 }, { "epoch": 0.6539025747382265, "grad_norm": 7.141068935394287, "learning_rate": 2.7244354989673806e-06, "loss": 0.4047, "step": 12927 }, { "epoch": 0.6539531589862917, "grad_norm": 10.288318634033203, "learning_rate": 2.723720853488666e-06, "loss": 0.4333, "step": 12928 }, { "epoch": 0.6540037432343568, "grad_norm": 8.93335247039795, "learning_rate": 2.7230062666641688e-06, "loss": 0.4765, "step": 12929 }, { "epoch": 0.654054327482422, "grad_norm": 4.884488582611084, "learning_rate": 2.7222917385123004e-06, "loss": 0.3461, "step": 12930 }, { "epoch": 0.6541049117304871, "grad_norm": 4.699313640594482, "learning_rate": 2.721577269051473e-06, "loss": 0.296, "step": 12931 }, { "epoch": 0.6541554959785523, "grad_norm": 5.503199577331543, "learning_rate": 2.720862858300095e-06, "loss": 0.4538, "step": 12932 }, { "epoch": 0.6542060802266174, "grad_norm": 5.294456481933594, "learning_rate": 2.7201485062765766e-06, "loss": 0.3186, "step": 12933 }, { "epoch": 0.6542566644746826, "grad_norm": 5.892703533172607, "learning_rate": 2.7194342129993225e-06, "loss": 0.3438, "step": 12934 }, { "epoch": 0.6543072487227477, "grad_norm": 7.681188106536865, "learning_rate": 2.7187199784867426e-06, "loss": 0.3819, "step": 12935 }, { "epoch": 0.6543578329708128, "grad_norm": 4.682833671569824, "learning_rate": 2.7180058027572374e-06, "loss": 0.3645, "step": 12936 }, { "epoch": 0.654408417218878, "grad_norm": 4.836553573608398, "learning_rate": 2.7172916858292097e-06, "loss": 0.3345, "step": 12937 }, { "epoch": 0.6544590014669432, "grad_norm": 20.031085968017578, "learning_rate": 2.7165776277210637e-06, "loss": 0.4446, "step": 12938 }, { "epoch": 0.6545095857150084, "grad_norm": 5.022900581359863, "learning_rate": 2.715863628451196e-06, "loss": 0.3486, "step": 12939 }, { "epoch": 0.6545601699630735, "grad_norm": 9.40181827545166, "learning_rate": 2.7151496880380046e-06, "loss": 0.3855, "step": 12940 }, { "epoch": 0.6546107542111387, "grad_norm": 10.838528633117676, "learning_rate": 2.7144358064998886e-06, "loss": 0.4019, "step": 12941 }, { "epoch": 0.6546613384592038, "grad_norm": 4.081369876861572, "learning_rate": 2.713721983855241e-06, "loss": 0.3679, "step": 12942 }, { "epoch": 0.654711922707269, "grad_norm": 7.283792018890381, "learning_rate": 2.7130082201224563e-06, "loss": 0.4209, "step": 12943 }, { "epoch": 0.6547625069553341, "grad_norm": 12.742154121398926, "learning_rate": 2.7122945153199247e-06, "loss": 0.4759, "step": 12944 }, { "epoch": 0.6548130912033993, "grad_norm": 6.281408309936523, "learning_rate": 2.7115808694660362e-06, "loss": 0.3821, "step": 12945 }, { "epoch": 0.6548636754514644, "grad_norm": 12.01850700378418, "learning_rate": 2.7108672825791836e-06, "loss": 0.5718, "step": 12946 }, { "epoch": 0.6549142596995295, "grad_norm": 5.827223777770996, "learning_rate": 2.7101537546777514e-06, "loss": 0.3502, "step": 12947 }, { "epoch": 0.6549648439475947, "grad_norm": 11.391069412231445, "learning_rate": 2.709440285780126e-06, "loss": 0.3887, "step": 12948 }, { "epoch": 0.6550154281956598, "grad_norm": 6.872642517089844, "learning_rate": 2.7087268759046902e-06, "loss": 0.4094, "step": 12949 }, { "epoch": 0.655066012443725, "grad_norm": 6.878274440765381, "learning_rate": 2.7080135250698302e-06, "loss": 0.3993, "step": 12950 }, { "epoch": 0.6551165966917902, "grad_norm": 5.146446228027344, "learning_rate": 2.7073002332939257e-06, "loss": 0.4308, "step": 12951 }, { "epoch": 0.6551671809398554, "grad_norm": 12.856558799743652, "learning_rate": 2.7065870005953544e-06, "loss": 0.4865, "step": 12952 }, { "epoch": 0.6552177651879205, "grad_norm": 5.10082483291626, "learning_rate": 2.705873826992499e-06, "loss": 0.3164, "step": 12953 }, { "epoch": 0.6552683494359857, "grad_norm": 4.210910797119141, "learning_rate": 2.705160712503733e-06, "loss": 0.292, "step": 12954 }, { "epoch": 0.6553189336840508, "grad_norm": 9.629131317138672, "learning_rate": 2.704447657147431e-06, "loss": 0.3108, "step": 12955 }, { "epoch": 0.655369517932116, "grad_norm": 6.199118614196777, "learning_rate": 2.703734660941972e-06, "loss": 0.3625, "step": 12956 }, { "epoch": 0.6554201021801811, "grad_norm": 10.608258247375488, "learning_rate": 2.7030217239057212e-06, "loss": 0.3998, "step": 12957 }, { "epoch": 0.6554706864282462, "grad_norm": 7.394427299499512, "learning_rate": 2.7023088460570537e-06, "loss": 0.3618, "step": 12958 }, { "epoch": 0.6555212706763114, "grad_norm": 5.9580888748168945, "learning_rate": 2.701596027414337e-06, "loss": 0.4154, "step": 12959 }, { "epoch": 0.6555718549243765, "grad_norm": 6.961627960205078, "learning_rate": 2.7008832679959378e-06, "loss": 0.3847, "step": 12960 }, { "epoch": 0.6556224391724417, "grad_norm": 7.4371819496154785, "learning_rate": 2.700170567820225e-06, "loss": 0.4325, "step": 12961 }, { "epoch": 0.6556730234205068, "grad_norm": 6.406289100646973, "learning_rate": 2.699457926905562e-06, "loss": 0.3429, "step": 12962 }, { "epoch": 0.655723607668572, "grad_norm": 5.2178425788879395, "learning_rate": 2.6987453452703095e-06, "loss": 0.3467, "step": 12963 }, { "epoch": 0.6557741919166371, "grad_norm": 7.022045612335205, "learning_rate": 2.698032822932833e-06, "loss": 0.2941, "step": 12964 }, { "epoch": 0.6558247761647024, "grad_norm": 5.062689304351807, "learning_rate": 2.6973203599114907e-06, "loss": 0.3324, "step": 12965 }, { "epoch": 0.6558753604127675, "grad_norm": 7.405171871185303, "learning_rate": 2.6966079562246384e-06, "loss": 0.3529, "step": 12966 }, { "epoch": 0.6559259446608326, "grad_norm": 7.6481733322143555, "learning_rate": 2.6958956118906375e-06, "loss": 0.4623, "step": 12967 }, { "epoch": 0.6559765289088978, "grad_norm": 8.75910472869873, "learning_rate": 2.6951833269278415e-06, "loss": 0.4105, "step": 12968 }, { "epoch": 0.6560271131569629, "grad_norm": 9.644694328308105, "learning_rate": 2.694471101354604e-06, "loss": 0.3781, "step": 12969 }, { "epoch": 0.6560776974050281, "grad_norm": 6.429866790771484, "learning_rate": 2.693758935189277e-06, "loss": 0.408, "step": 12970 }, { "epoch": 0.6561282816530932, "grad_norm": 16.551380157470703, "learning_rate": 2.693046828450212e-06, "loss": 0.4022, "step": 12971 }, { "epoch": 0.6561788659011584, "grad_norm": 8.34476089477539, "learning_rate": 2.6923347811557565e-06, "loss": 0.4136, "step": 12972 }, { "epoch": 0.6562294501492235, "grad_norm": 7.661035060882568, "learning_rate": 2.6916227933242618e-06, "loss": 0.3349, "step": 12973 }, { "epoch": 0.6562800343972887, "grad_norm": 8.063077926635742, "learning_rate": 2.6909108649740713e-06, "loss": 0.4552, "step": 12974 }, { "epoch": 0.6563306186453538, "grad_norm": 10.732807159423828, "learning_rate": 2.690198996123529e-06, "loss": 0.4602, "step": 12975 }, { "epoch": 0.656381202893419, "grad_norm": 5.765448570251465, "learning_rate": 2.689487186790981e-06, "loss": 0.3614, "step": 12976 }, { "epoch": 0.6564317871414841, "grad_norm": 12.400007247924805, "learning_rate": 2.6887754369947684e-06, "loss": 0.4952, "step": 12977 }, { "epoch": 0.6564823713895493, "grad_norm": 8.137979507446289, "learning_rate": 2.688063746753227e-06, "loss": 0.5229, "step": 12978 }, { "epoch": 0.6565329556376145, "grad_norm": 6.9365339279174805, "learning_rate": 2.6873521160847014e-06, "loss": 0.4008, "step": 12979 }, { "epoch": 0.6565835398856796, "grad_norm": 6.403702735900879, "learning_rate": 2.6866405450075257e-06, "loss": 0.3584, "step": 12980 }, { "epoch": 0.6566341241337448, "grad_norm": 6.434298038482666, "learning_rate": 2.6859290335400332e-06, "loss": 0.4448, "step": 12981 }, { "epoch": 0.6566847083818099, "grad_norm": 10.515422821044922, "learning_rate": 2.6852175817005643e-06, "loss": 0.4597, "step": 12982 }, { "epoch": 0.6567352926298751, "grad_norm": 6.245938777923584, "learning_rate": 2.6845061895074454e-06, "loss": 0.2797, "step": 12983 }, { "epoch": 0.6567858768779402, "grad_norm": 9.8443021774292, "learning_rate": 2.683794856979006e-06, "loss": 0.4998, "step": 12984 }, { "epoch": 0.6568364611260054, "grad_norm": 11.313366889953613, "learning_rate": 2.683083584133582e-06, "loss": 0.4211, "step": 12985 }, { "epoch": 0.6568870453740705, "grad_norm": 9.253007888793945, "learning_rate": 2.6823723709894974e-06, "loss": 0.3552, "step": 12986 }, { "epoch": 0.6569376296221356, "grad_norm": 5.724282741546631, "learning_rate": 2.6816612175650763e-06, "loss": 0.3503, "step": 12987 }, { "epoch": 0.6569882138702008, "grad_norm": 11.99758529663086, "learning_rate": 2.6809501238786483e-06, "loss": 0.5227, "step": 12988 }, { "epoch": 0.6570387981182659, "grad_norm": 12.267704963684082, "learning_rate": 2.680239089948534e-06, "loss": 0.4571, "step": 12989 }, { "epoch": 0.6570893823663311, "grad_norm": 10.805877685546875, "learning_rate": 2.679528115793053e-06, "loss": 0.4005, "step": 12990 }, { "epoch": 0.6571399666143963, "grad_norm": 9.681516647338867, "learning_rate": 2.6788172014305304e-06, "loss": 0.498, "step": 12991 }, { "epoch": 0.6571905508624615, "grad_norm": 7.190853595733643, "learning_rate": 2.6781063468792814e-06, "loss": 0.456, "step": 12992 }, { "epoch": 0.6572411351105266, "grad_norm": 8.643207550048828, "learning_rate": 2.6773955521576213e-06, "loss": 0.4831, "step": 12993 }, { "epoch": 0.6572917193585918, "grad_norm": 13.961845397949219, "learning_rate": 2.67668481728387e-06, "loss": 0.3886, "step": 12994 }, { "epoch": 0.6573423036066569, "grad_norm": 8.931800842285156, "learning_rate": 2.6759741422763393e-06, "loss": 0.3942, "step": 12995 }, { "epoch": 0.657392887854722, "grad_norm": 9.539555549621582, "learning_rate": 2.6752635271533413e-06, "loss": 0.3403, "step": 12996 }, { "epoch": 0.6574434721027872, "grad_norm": 10.429085731506348, "learning_rate": 2.6745529719331876e-06, "loss": 0.4933, "step": 12997 }, { "epoch": 0.6574940563508523, "grad_norm": 5.249504089355469, "learning_rate": 2.673842476634184e-06, "loss": 0.4042, "step": 12998 }, { "epoch": 0.6575446405989175, "grad_norm": 11.799250602722168, "learning_rate": 2.673132041274644e-06, "loss": 0.4797, "step": 12999 }, { "epoch": 0.6575952248469826, "grad_norm": 6.659347057342529, "learning_rate": 2.6724216658728706e-06, "loss": 0.3446, "step": 13000 }, { "epoch": 0.6576458090950478, "grad_norm": 7.919773578643799, "learning_rate": 2.671711350447167e-06, "loss": 0.468, "step": 13001 }, { "epoch": 0.6576963933431129, "grad_norm": 6.614914894104004, "learning_rate": 2.6710010950158393e-06, "loss": 0.3762, "step": 13002 }, { "epoch": 0.6577469775911781, "grad_norm": 9.032417297363281, "learning_rate": 2.670290899597189e-06, "loss": 0.442, "step": 13003 }, { "epoch": 0.6577975618392432, "grad_norm": 21.019309997558594, "learning_rate": 2.6695807642095127e-06, "loss": 0.3441, "step": 13004 }, { "epoch": 0.6578481460873085, "grad_norm": 12.342016220092773, "learning_rate": 2.6688706888711136e-06, "loss": 0.4299, "step": 13005 }, { "epoch": 0.6578987303353736, "grad_norm": 8.603602409362793, "learning_rate": 2.668160673600285e-06, "loss": 0.4158, "step": 13006 }, { "epoch": 0.6579493145834387, "grad_norm": 9.300667762756348, "learning_rate": 2.667450718415323e-06, "loss": 0.392, "step": 13007 }, { "epoch": 0.6579998988315039, "grad_norm": 6.903639316558838, "learning_rate": 2.666740823334525e-06, "loss": 0.4558, "step": 13008 }, { "epoch": 0.658050483079569, "grad_norm": 8.217483520507812, "learning_rate": 2.6660309883761785e-06, "loss": 0.4058, "step": 13009 }, { "epoch": 0.6581010673276342, "grad_norm": 9.003304481506348, "learning_rate": 2.6653212135585737e-06, "loss": 0.2737, "step": 13010 }, { "epoch": 0.6581516515756993, "grad_norm": 7.243782997131348, "learning_rate": 2.664611498900005e-06, "loss": 0.346, "step": 13011 }, { "epoch": 0.6582022358237645, "grad_norm": 6.530452251434326, "learning_rate": 2.663901844418756e-06, "loss": 0.4668, "step": 13012 }, { "epoch": 0.6582528200718296, "grad_norm": 8.296147346496582, "learning_rate": 2.663192250133112e-06, "loss": 0.3907, "step": 13013 }, { "epoch": 0.6583034043198948, "grad_norm": 9.837695121765137, "learning_rate": 2.662482716061362e-06, "loss": 0.5591, "step": 13014 }, { "epoch": 0.6583539885679599, "grad_norm": 5.853817462921143, "learning_rate": 2.6617732422217856e-06, "loss": 0.3966, "step": 13015 }, { "epoch": 0.658404572816025, "grad_norm": 14.710929870605469, "learning_rate": 2.661063828632664e-06, "loss": 0.4266, "step": 13016 }, { "epoch": 0.6584551570640902, "grad_norm": 9.446219444274902, "learning_rate": 2.660354475312279e-06, "loss": 0.498, "step": 13017 }, { "epoch": 0.6585057413121554, "grad_norm": 8.805279731750488, "learning_rate": 2.6596451822789086e-06, "loss": 0.4841, "step": 13018 }, { "epoch": 0.6585563255602206, "grad_norm": 7.143296241760254, "learning_rate": 2.658935949550827e-06, "loss": 0.4613, "step": 13019 }, { "epoch": 0.6586069098082857, "grad_norm": 7.077054977416992, "learning_rate": 2.6582267771463134e-06, "loss": 0.4295, "step": 13020 }, { "epoch": 0.6586574940563509, "grad_norm": 8.705008506774902, "learning_rate": 2.657517665083641e-06, "loss": 0.4598, "step": 13021 }, { "epoch": 0.658708078304416, "grad_norm": 9.775656700134277, "learning_rate": 2.6568086133810767e-06, "loss": 0.4007, "step": 13022 }, { "epoch": 0.6587586625524812, "grad_norm": 6.07248592376709, "learning_rate": 2.6560996220568964e-06, "loss": 0.4181, "step": 13023 }, { "epoch": 0.6588092468005463, "grad_norm": 5.559722900390625, "learning_rate": 2.655390691129367e-06, "loss": 0.4077, "step": 13024 }, { "epoch": 0.6588598310486115, "grad_norm": 8.929080963134766, "learning_rate": 2.654681820616755e-06, "loss": 0.4141, "step": 13025 }, { "epoch": 0.6589104152966766, "grad_norm": 10.156850814819336, "learning_rate": 2.6539730105373294e-06, "loss": 0.3854, "step": 13026 }, { "epoch": 0.6589609995447417, "grad_norm": 6.191843509674072, "learning_rate": 2.653264260909353e-06, "loss": 0.3715, "step": 13027 }, { "epoch": 0.6590115837928069, "grad_norm": 6.896129131317139, "learning_rate": 2.652555571751086e-06, "loss": 0.3539, "step": 13028 }, { "epoch": 0.659062168040872, "grad_norm": 5.917990207672119, "learning_rate": 2.6518469430807937e-06, "loss": 0.4382, "step": 13029 }, { "epoch": 0.6591127522889372, "grad_norm": 8.058324813842773, "learning_rate": 2.651138374916734e-06, "loss": 0.3442, "step": 13030 }, { "epoch": 0.6591633365370024, "grad_norm": 12.05867862701416, "learning_rate": 2.6504298672771633e-06, "loss": 0.4068, "step": 13031 }, { "epoch": 0.6592139207850676, "grad_norm": 6.360105037689209, "learning_rate": 2.6497214201803418e-06, "loss": 0.2868, "step": 13032 }, { "epoch": 0.6592645050331327, "grad_norm": 7.53654670715332, "learning_rate": 2.649013033644522e-06, "loss": 0.373, "step": 13033 }, { "epoch": 0.6593150892811979, "grad_norm": 5.510149955749512, "learning_rate": 2.648304707687958e-06, "loss": 0.3823, "step": 13034 }, { "epoch": 0.659365673529263, "grad_norm": 9.12834358215332, "learning_rate": 2.6475964423289015e-06, "loss": 0.3925, "step": 13035 }, { "epoch": 0.6594162577773282, "grad_norm": 8.09786605834961, "learning_rate": 2.646888237585601e-06, "loss": 0.4211, "step": 13036 }, { "epoch": 0.6594668420253933, "grad_norm": 10.856805801391602, "learning_rate": 2.646180093476308e-06, "loss": 0.3788, "step": 13037 }, { "epoch": 0.6595174262734584, "grad_norm": 12.683270454406738, "learning_rate": 2.645472010019269e-06, "loss": 0.4627, "step": 13038 }, { "epoch": 0.6595680105215236, "grad_norm": 6.566129207611084, "learning_rate": 2.644763987232727e-06, "loss": 0.4079, "step": 13039 }, { "epoch": 0.6596185947695887, "grad_norm": 8.198301315307617, "learning_rate": 2.6440560251349305e-06, "loss": 0.4082, "step": 13040 }, { "epoch": 0.6596691790176539, "grad_norm": 11.007381439208984, "learning_rate": 2.6433481237441192e-06, "loss": 0.43, "step": 13041 }, { "epoch": 0.659719763265719, "grad_norm": 12.625516891479492, "learning_rate": 2.6426402830785335e-06, "loss": 0.4522, "step": 13042 }, { "epoch": 0.6597703475137842, "grad_norm": 9.072640419006348, "learning_rate": 2.6419325031564147e-06, "loss": 0.3778, "step": 13043 }, { "epoch": 0.6598209317618493, "grad_norm": 5.431128025054932, "learning_rate": 2.6412247839960002e-06, "loss": 0.3912, "step": 13044 }, { "epoch": 0.6598715160099146, "grad_norm": 7.450831413269043, "learning_rate": 2.6405171256155254e-06, "loss": 0.3969, "step": 13045 }, { "epoch": 0.6599221002579797, "grad_norm": 9.365501403808594, "learning_rate": 2.6398095280332252e-06, "loss": 0.434, "step": 13046 }, { "epoch": 0.6599726845060448, "grad_norm": 7.785852909088135, "learning_rate": 2.6391019912673325e-06, "loss": 0.3622, "step": 13047 }, { "epoch": 0.66002326875411, "grad_norm": 12.824789047241211, "learning_rate": 2.6383945153360774e-06, "loss": 0.4025, "step": 13048 }, { "epoch": 0.6600738530021751, "grad_norm": 17.138978958129883, "learning_rate": 2.6376871002576932e-06, "loss": 0.5281, "step": 13049 }, { "epoch": 0.6601244372502403, "grad_norm": 9.264842987060547, "learning_rate": 2.6369797460504064e-06, "loss": 0.3808, "step": 13050 }, { "epoch": 0.6601750214983054, "grad_norm": 5.711446285247803, "learning_rate": 2.636272452732442e-06, "loss": 0.3389, "step": 13051 }, { "epoch": 0.6602256057463706, "grad_norm": 6.338112831115723, "learning_rate": 2.6355652203220295e-06, "loss": 0.3394, "step": 13052 }, { "epoch": 0.6602761899944357, "grad_norm": 9.19366455078125, "learning_rate": 2.6348580488373897e-06, "loss": 0.4035, "step": 13053 }, { "epoch": 0.6603267742425009, "grad_norm": 5.916868209838867, "learning_rate": 2.6341509382967434e-06, "loss": 0.4299, "step": 13054 }, { "epoch": 0.660377358490566, "grad_norm": 4.921972751617432, "learning_rate": 2.6334438887183145e-06, "loss": 0.3997, "step": 13055 }, { "epoch": 0.6604279427386311, "grad_norm": 8.681156158447266, "learning_rate": 2.632736900120321e-06, "loss": 0.441, "step": 13056 }, { "epoch": 0.6604785269866963, "grad_norm": 18.08375358581543, "learning_rate": 2.6320299725209776e-06, "loss": 0.6166, "step": 13057 }, { "epoch": 0.6605291112347615, "grad_norm": 8.926491737365723, "learning_rate": 2.6313231059385036e-06, "loss": 0.4042, "step": 13058 }, { "epoch": 0.6605796954828267, "grad_norm": 4.759212970733643, "learning_rate": 2.6306163003911146e-06, "loss": 0.3613, "step": 13059 }, { "epoch": 0.6606302797308918, "grad_norm": 5.942121982574463, "learning_rate": 2.629909555897017e-06, "loss": 0.3469, "step": 13060 }, { "epoch": 0.660680863978957, "grad_norm": 7.181011199951172, "learning_rate": 2.6292028724744257e-06, "loss": 0.3449, "step": 13061 }, { "epoch": 0.6607314482270221, "grad_norm": 11.434632301330566, "learning_rate": 2.628496250141551e-06, "loss": 0.4975, "step": 13062 }, { "epoch": 0.6607820324750873, "grad_norm": 9.141717910766602, "learning_rate": 2.6277896889165976e-06, "loss": 0.4663, "step": 13063 }, { "epoch": 0.6608326167231524, "grad_norm": 7.127457141876221, "learning_rate": 2.6270831888177767e-06, "loss": 0.2667, "step": 13064 }, { "epoch": 0.6608832009712176, "grad_norm": 7.099367618560791, "learning_rate": 2.62637674986329e-06, "loss": 0.3816, "step": 13065 }, { "epoch": 0.6609337852192827, "grad_norm": 12.362937927246094, "learning_rate": 2.6256703720713394e-06, "loss": 0.355, "step": 13066 }, { "epoch": 0.6609843694673478, "grad_norm": 8.643762588500977, "learning_rate": 2.6249640554601295e-06, "loss": 0.3609, "step": 13067 }, { "epoch": 0.661034953715413, "grad_norm": 6.735847473144531, "learning_rate": 2.62425780004786e-06, "loss": 0.4458, "step": 13068 }, { "epoch": 0.6610855379634781, "grad_norm": 5.842170238494873, "learning_rate": 2.6235516058527266e-06, "loss": 0.3621, "step": 13069 }, { "epoch": 0.6611361222115433, "grad_norm": 8.508777618408203, "learning_rate": 2.62284547289293e-06, "loss": 0.5159, "step": 13070 }, { "epoch": 0.6611867064596085, "grad_norm": 6.782491207122803, "learning_rate": 2.622139401186664e-06, "loss": 0.4067, "step": 13071 }, { "epoch": 0.6612372907076737, "grad_norm": 5.052624225616455, "learning_rate": 2.6214333907521227e-06, "loss": 0.2932, "step": 13072 }, { "epoch": 0.6612878749557388, "grad_norm": 5.966973781585693, "learning_rate": 2.6207274416074968e-06, "loss": 0.3162, "step": 13073 }, { "epoch": 0.661338459203804, "grad_norm": 8.506579399108887, "learning_rate": 2.6200215537709762e-06, "loss": 0.4837, "step": 13074 }, { "epoch": 0.6613890434518691, "grad_norm": 9.07243824005127, "learning_rate": 2.619315727260754e-06, "loss": 0.4503, "step": 13075 }, { "epoch": 0.6614396276999343, "grad_norm": 6.289307117462158, "learning_rate": 2.6186099620950144e-06, "loss": 0.3749, "step": 13076 }, { "epoch": 0.6614902119479994, "grad_norm": 9.254393577575684, "learning_rate": 2.617904258291942e-06, "loss": 0.2452, "step": 13077 }, { "epoch": 0.6615407961960645, "grad_norm": 12.523303031921387, "learning_rate": 2.617198615869726e-06, "loss": 0.4073, "step": 13078 }, { "epoch": 0.6615913804441297, "grad_norm": 7.791682720184326, "learning_rate": 2.6164930348465457e-06, "loss": 0.4335, "step": 13079 }, { "epoch": 0.6616419646921948, "grad_norm": 12.600500106811523, "learning_rate": 2.615787515240581e-06, "loss": 0.3805, "step": 13080 }, { "epoch": 0.66169254894026, "grad_norm": 9.126923561096191, "learning_rate": 2.6150820570700147e-06, "loss": 0.3845, "step": 13081 }, { "epoch": 0.6617431331883251, "grad_norm": 8.118720054626465, "learning_rate": 2.6143766603530234e-06, "loss": 0.4878, "step": 13082 }, { "epoch": 0.6617937174363903, "grad_norm": 7.088134765625, "learning_rate": 2.613671325107784e-06, "loss": 0.4237, "step": 13083 }, { "epoch": 0.6618443016844555, "grad_norm": 6.8949761390686035, "learning_rate": 2.6129660513524678e-06, "loss": 0.3712, "step": 13084 }, { "epoch": 0.6618948859325207, "grad_norm": 7.6991119384765625, "learning_rate": 2.612260839105255e-06, "loss": 0.4306, "step": 13085 }, { "epoch": 0.6619454701805858, "grad_norm": 7.683018684387207, "learning_rate": 2.6115556883843097e-06, "loss": 0.4251, "step": 13086 }, { "epoch": 0.661996054428651, "grad_norm": 8.036598205566406, "learning_rate": 2.610850599207806e-06, "loss": 0.3697, "step": 13087 }, { "epoch": 0.6620466386767161, "grad_norm": 12.127533912658691, "learning_rate": 2.610145571593913e-06, "loss": 0.4821, "step": 13088 }, { "epoch": 0.6620972229247812, "grad_norm": 7.144867420196533, "learning_rate": 2.6094406055607936e-06, "loss": 0.3635, "step": 13089 }, { "epoch": 0.6621478071728464, "grad_norm": 11.2236909866333, "learning_rate": 2.608735701126617e-06, "loss": 0.456, "step": 13090 }, { "epoch": 0.6621983914209115, "grad_norm": 9.083884239196777, "learning_rate": 2.608030858309546e-06, "loss": 0.4654, "step": 13091 }, { "epoch": 0.6622489756689767, "grad_norm": 8.549880027770996, "learning_rate": 2.6073260771277402e-06, "loss": 0.3589, "step": 13092 }, { "epoch": 0.6622995599170418, "grad_norm": 9.015153884887695, "learning_rate": 2.6066213575993644e-06, "loss": 0.4314, "step": 13093 }, { "epoch": 0.662350144165107, "grad_norm": 9.863396644592285, "learning_rate": 2.605916699742575e-06, "loss": 0.4483, "step": 13094 }, { "epoch": 0.6624007284131721, "grad_norm": 6.599244594573975, "learning_rate": 2.605212103575527e-06, "loss": 0.3979, "step": 13095 }, { "epoch": 0.6624513126612372, "grad_norm": 7.627872943878174, "learning_rate": 2.604507569116382e-06, "loss": 0.4373, "step": 13096 }, { "epoch": 0.6625018969093024, "grad_norm": 10.240074157714844, "learning_rate": 2.6038030963832898e-06, "loss": 0.3501, "step": 13097 }, { "epoch": 0.6625524811573676, "grad_norm": 8.003796577453613, "learning_rate": 2.6030986853944047e-06, "loss": 0.4055, "step": 13098 }, { "epoch": 0.6626030654054328, "grad_norm": 6.076250076293945, "learning_rate": 2.6023943361678765e-06, "loss": 0.282, "step": 13099 }, { "epoch": 0.6626536496534979, "grad_norm": 6.32978630065918, "learning_rate": 2.6016900487218554e-06, "loss": 0.3818, "step": 13100 }, { "epoch": 0.6627042339015631, "grad_norm": 7.375278949737549, "learning_rate": 2.6009858230744865e-06, "loss": 0.3556, "step": 13101 }, { "epoch": 0.6627548181496282, "grad_norm": 5.8949384689331055, "learning_rate": 2.6002816592439205e-06, "loss": 0.3391, "step": 13102 }, { "epoch": 0.6628054023976934, "grad_norm": 5.953362464904785, "learning_rate": 2.599577557248301e-06, "loss": 0.3065, "step": 13103 }, { "epoch": 0.6628559866457585, "grad_norm": 10.11425495147705, "learning_rate": 2.5988735171057667e-06, "loss": 0.4374, "step": 13104 }, { "epoch": 0.6629065708938237, "grad_norm": 10.4467134475708, "learning_rate": 2.598169538834464e-06, "loss": 0.4318, "step": 13105 }, { "epoch": 0.6629571551418888, "grad_norm": 14.79785442352295, "learning_rate": 2.5974656224525313e-06, "loss": 0.3971, "step": 13106 }, { "epoch": 0.6630077393899539, "grad_norm": 10.969703674316406, "learning_rate": 2.5967617679781043e-06, "loss": 0.3909, "step": 13107 }, { "epoch": 0.6630583236380191, "grad_norm": 11.727088928222656, "learning_rate": 2.5960579754293234e-06, "loss": 0.3764, "step": 13108 }, { "epoch": 0.6631089078860842, "grad_norm": 14.889558792114258, "learning_rate": 2.595354244824323e-06, "loss": 0.498, "step": 13109 }, { "epoch": 0.6631594921341494, "grad_norm": 12.2052583694458, "learning_rate": 2.594650576181232e-06, "loss": 0.5359, "step": 13110 }, { "epoch": 0.6632100763822146, "grad_norm": 12.059144973754883, "learning_rate": 2.5939469695181918e-06, "loss": 0.3688, "step": 13111 }, { "epoch": 0.6632606606302798, "grad_norm": 6.703920364379883, "learning_rate": 2.5932434248533213e-06, "loss": 0.4178, "step": 13112 }, { "epoch": 0.6633112448783449, "grad_norm": 13.345723152160645, "learning_rate": 2.5925399422047567e-06, "loss": 0.4399, "step": 13113 }, { "epoch": 0.6633618291264101, "grad_norm": 6.154552459716797, "learning_rate": 2.5918365215906227e-06, "loss": 0.4256, "step": 13114 }, { "epoch": 0.6634124133744752, "grad_norm": 4.6918625831604, "learning_rate": 2.5911331630290424e-06, "loss": 0.3291, "step": 13115 }, { "epoch": 0.6634629976225404, "grad_norm": 9.638215065002441, "learning_rate": 2.590429866538145e-06, "loss": 0.5131, "step": 13116 }, { "epoch": 0.6635135818706055, "grad_norm": 8.277399063110352, "learning_rate": 2.589726632136049e-06, "loss": 0.4222, "step": 13117 }, { "epoch": 0.6635641661186706, "grad_norm": 6.2558722496032715, "learning_rate": 2.589023459840877e-06, "loss": 0.3564, "step": 13118 }, { "epoch": 0.6636147503667358, "grad_norm": 6.77132511138916, "learning_rate": 2.5883203496707445e-06, "loss": 0.326, "step": 13119 }, { "epoch": 0.6636653346148009, "grad_norm": 8.448311805725098, "learning_rate": 2.5876173016437734e-06, "loss": 0.5113, "step": 13120 }, { "epoch": 0.6637159188628661, "grad_norm": 8.141461372375488, "learning_rate": 2.5869143157780774e-06, "loss": 0.3579, "step": 13121 }, { "epoch": 0.6637665031109312, "grad_norm": 7.781853199005127, "learning_rate": 2.5862113920917685e-06, "loss": 0.4703, "step": 13122 }, { "epoch": 0.6638170873589964, "grad_norm": 6.492433071136475, "learning_rate": 2.585508530602965e-06, "loss": 0.4247, "step": 13123 }, { "epoch": 0.6638676716070616, "grad_norm": 6.173491477966309, "learning_rate": 2.5848057313297736e-06, "loss": 0.4882, "step": 13124 }, { "epoch": 0.6639182558551268, "grad_norm": 8.218122482299805, "learning_rate": 2.584102994290305e-06, "loss": 0.4112, "step": 13125 }, { "epoch": 0.6639688401031919, "grad_norm": 5.08157205581665, "learning_rate": 2.5834003195026667e-06, "loss": 0.2859, "step": 13126 }, { "epoch": 0.664019424351257, "grad_norm": 8.49783706665039, "learning_rate": 2.5826977069849636e-06, "loss": 0.3569, "step": 13127 }, { "epoch": 0.6640700085993222, "grad_norm": 6.815110683441162, "learning_rate": 2.5819951567553037e-06, "loss": 0.2858, "step": 13128 }, { "epoch": 0.6641205928473873, "grad_norm": 5.772686004638672, "learning_rate": 2.5812926688317887e-06, "loss": 0.2496, "step": 13129 }, { "epoch": 0.6641711770954525, "grad_norm": 8.818730354309082, "learning_rate": 2.580590243232517e-06, "loss": 0.3739, "step": 13130 }, { "epoch": 0.6642217613435176, "grad_norm": 6.202127456665039, "learning_rate": 2.579887879975592e-06, "loss": 0.3963, "step": 13131 }, { "epoch": 0.6642723455915828, "grad_norm": 4.766263008117676, "learning_rate": 2.579185579079112e-06, "loss": 0.409, "step": 13132 }, { "epoch": 0.6643229298396479, "grad_norm": 11.15356731414795, "learning_rate": 2.5784833405611696e-06, "loss": 0.4289, "step": 13133 }, { "epoch": 0.6643735140877131, "grad_norm": 7.978850364685059, "learning_rate": 2.5777811644398655e-06, "loss": 0.4082, "step": 13134 }, { "epoch": 0.6644240983357782, "grad_norm": 11.719243049621582, "learning_rate": 2.57707905073329e-06, "loss": 0.5464, "step": 13135 }, { "epoch": 0.6644746825838433, "grad_norm": 5.87013053894043, "learning_rate": 2.5763769994595345e-06, "loss": 0.3452, "step": 13136 }, { "epoch": 0.6645252668319085, "grad_norm": 9.446552276611328, "learning_rate": 2.5756750106366896e-06, "loss": 0.3949, "step": 13137 }, { "epoch": 0.6645758510799737, "grad_norm": 13.728826522827148, "learning_rate": 2.5749730842828447e-06, "loss": 0.5004, "step": 13138 }, { "epoch": 0.6646264353280389, "grad_norm": 5.722471237182617, "learning_rate": 2.5742712204160836e-06, "loss": 0.3148, "step": 13139 }, { "epoch": 0.664677019576104, "grad_norm": 5.675436496734619, "learning_rate": 2.573569419054496e-06, "loss": 0.3188, "step": 13140 }, { "epoch": 0.6647276038241692, "grad_norm": 12.085947036743164, "learning_rate": 2.5728676802161633e-06, "loss": 0.3391, "step": 13141 }, { "epoch": 0.6647781880722343, "grad_norm": 7.092342376708984, "learning_rate": 2.5721660039191664e-06, "loss": 0.34, "step": 13142 }, { "epoch": 0.6648287723202995, "grad_norm": 8.796948432922363, "learning_rate": 2.5714643901815895e-06, "loss": 0.4453, "step": 13143 }, { "epoch": 0.6648793565683646, "grad_norm": 9.413053512573242, "learning_rate": 2.5707628390215087e-06, "loss": 0.4358, "step": 13144 }, { "epoch": 0.6649299408164298, "grad_norm": 5.950229167938232, "learning_rate": 2.5700613504570005e-06, "loss": 0.3842, "step": 13145 }, { "epoch": 0.6649805250644949, "grad_norm": 8.580750465393066, "learning_rate": 2.569359924506144e-06, "loss": 0.4495, "step": 13146 }, { "epoch": 0.66503110931256, "grad_norm": 8.731904029846191, "learning_rate": 2.5686585611870113e-06, "loss": 0.4182, "step": 13147 }, { "epoch": 0.6650816935606252, "grad_norm": 4.939067363739014, "learning_rate": 2.567957260517673e-06, "loss": 0.3496, "step": 13148 }, { "epoch": 0.6651322778086903, "grad_norm": 11.622013092041016, "learning_rate": 2.567256022516205e-06, "loss": 0.5171, "step": 13149 }, { "epoch": 0.6651828620567555, "grad_norm": 10.368841171264648, "learning_rate": 2.56655484720067e-06, "loss": 0.466, "step": 13150 }, { "epoch": 0.6652334463048207, "grad_norm": 23.16364860534668, "learning_rate": 2.565853734589141e-06, "loss": 0.4835, "step": 13151 }, { "epoch": 0.6652840305528859, "grad_norm": 7.759617805480957, "learning_rate": 2.565152684699681e-06, "loss": 0.3556, "step": 13152 }, { "epoch": 0.665334614800951, "grad_norm": 5.998760223388672, "learning_rate": 2.5644516975503543e-06, "loss": 0.3614, "step": 13153 }, { "epoch": 0.6653851990490162, "grad_norm": 8.98951530456543, "learning_rate": 2.5637507731592252e-06, "loss": 0.4068, "step": 13154 }, { "epoch": 0.6654357832970813, "grad_norm": 10.9996919631958, "learning_rate": 2.5630499115443552e-06, "loss": 0.4326, "step": 13155 }, { "epoch": 0.6654863675451465, "grad_norm": 7.3634748458862305, "learning_rate": 2.5623491127238022e-06, "loss": 0.3739, "step": 13156 }, { "epoch": 0.6655369517932116, "grad_norm": 6.393908500671387, "learning_rate": 2.561648376715623e-06, "loss": 0.4281, "step": 13157 }, { "epoch": 0.6655875360412767, "grad_norm": 6.877094745635986, "learning_rate": 2.560947703537877e-06, "loss": 0.3409, "step": 13158 }, { "epoch": 0.6656381202893419, "grad_norm": 5.991779804229736, "learning_rate": 2.5602470932086176e-06, "loss": 0.3659, "step": 13159 }, { "epoch": 0.665688704537407, "grad_norm": 6.726159572601318, "learning_rate": 2.5595465457458955e-06, "loss": 0.4012, "step": 13160 }, { "epoch": 0.6657392887854722, "grad_norm": 8.786214828491211, "learning_rate": 2.558846061167767e-06, "loss": 0.3369, "step": 13161 }, { "epoch": 0.6657898730335373, "grad_norm": 10.939056396484375, "learning_rate": 2.558145639492279e-06, "loss": 0.4844, "step": 13162 }, { "epoch": 0.6658404572816025, "grad_norm": 6.049586296081543, "learning_rate": 2.5574452807374788e-06, "loss": 0.4101, "step": 13163 }, { "epoch": 0.6658910415296677, "grad_norm": 7.667741298675537, "learning_rate": 2.556744984921415e-06, "loss": 0.4544, "step": 13164 }, { "epoch": 0.6659416257777329, "grad_norm": 10.090062141418457, "learning_rate": 2.556044752062129e-06, "loss": 0.4381, "step": 13165 }, { "epoch": 0.665992210025798, "grad_norm": 6.266382694244385, "learning_rate": 2.555344582177669e-06, "loss": 0.3334, "step": 13166 }, { "epoch": 0.6660427942738631, "grad_norm": 7.281772613525391, "learning_rate": 2.5546444752860732e-06, "loss": 0.3686, "step": 13167 }, { "epoch": 0.6660933785219283, "grad_norm": 18.534021377563477, "learning_rate": 2.5539444314053816e-06, "loss": 0.4771, "step": 13168 }, { "epoch": 0.6661439627699934, "grad_norm": 15.64120101928711, "learning_rate": 2.5532444505536365e-06, "loss": 0.6668, "step": 13169 }, { "epoch": 0.6661945470180586, "grad_norm": 11.761331558227539, "learning_rate": 2.5525445327488706e-06, "loss": 0.4142, "step": 13170 }, { "epoch": 0.6662451312661237, "grad_norm": 10.538775444030762, "learning_rate": 2.55184467800912e-06, "loss": 0.4601, "step": 13171 }, { "epoch": 0.6662957155141889, "grad_norm": 7.532995223999023, "learning_rate": 2.5511448863524195e-06, "loss": 0.4474, "step": 13172 }, { "epoch": 0.666346299762254, "grad_norm": 9.067614555358887, "learning_rate": 2.550445157796801e-06, "loss": 0.4198, "step": 13173 }, { "epoch": 0.6663968840103192, "grad_norm": 12.741596221923828, "learning_rate": 2.5497454923602916e-06, "loss": 0.3489, "step": 13174 }, { "epoch": 0.6664474682583843, "grad_norm": 7.0774827003479, "learning_rate": 2.549045890060927e-06, "loss": 0.3508, "step": 13175 }, { "epoch": 0.6664980525064494, "grad_norm": 9.867298126220703, "learning_rate": 2.5483463509167273e-06, "loss": 0.4201, "step": 13176 }, { "epoch": 0.6665486367545146, "grad_norm": 5.924665927886963, "learning_rate": 2.547646874945719e-06, "loss": 0.4637, "step": 13177 }, { "epoch": 0.6665992210025798, "grad_norm": 8.574153900146484, "learning_rate": 2.5469474621659292e-06, "loss": 0.4179, "step": 13178 }, { "epoch": 0.666649805250645, "grad_norm": 13.513668060302734, "learning_rate": 2.5462481125953777e-06, "loss": 0.5616, "step": 13179 }, { "epoch": 0.6667003894987101, "grad_norm": 9.842164993286133, "learning_rate": 2.5455488262520837e-06, "loss": 0.3528, "step": 13180 }, { "epoch": 0.6667509737467753, "grad_norm": 13.381129264831543, "learning_rate": 2.5448496031540704e-06, "loss": 0.3682, "step": 13181 }, { "epoch": 0.6668015579948404, "grad_norm": 9.138741493225098, "learning_rate": 2.5441504433193522e-06, "loss": 0.4118, "step": 13182 }, { "epoch": 0.6668521422429056, "grad_norm": 7.279880523681641, "learning_rate": 2.5434513467659437e-06, "loss": 0.3671, "step": 13183 }, { "epoch": 0.6669027264909707, "grad_norm": 6.3672308921813965, "learning_rate": 2.542752313511862e-06, "loss": 0.3683, "step": 13184 }, { "epoch": 0.6669533107390359, "grad_norm": 5.059848785400391, "learning_rate": 2.542053343575118e-06, "loss": 0.3258, "step": 13185 }, { "epoch": 0.667003894987101, "grad_norm": 9.513976097106934, "learning_rate": 2.541354436973721e-06, "loss": 0.3826, "step": 13186 }, { "epoch": 0.6670544792351661, "grad_norm": 19.22484588623047, "learning_rate": 2.5406555937256826e-06, "loss": 0.6551, "step": 13187 }, { "epoch": 0.6671050634832313, "grad_norm": 8.367300033569336, "learning_rate": 2.5399568138490093e-06, "loss": 0.5366, "step": 13188 }, { "epoch": 0.6671556477312964, "grad_norm": 8.739523887634277, "learning_rate": 2.5392580973617075e-06, "loss": 0.4055, "step": 13189 }, { "epoch": 0.6672062319793616, "grad_norm": 7.365203380584717, "learning_rate": 2.5385594442817807e-06, "loss": 0.3659, "step": 13190 }, { "epoch": 0.6672568162274268, "grad_norm": 7.046950340270996, "learning_rate": 2.537860854627232e-06, "loss": 0.4417, "step": 13191 }, { "epoch": 0.667307400475492, "grad_norm": 5.324620246887207, "learning_rate": 2.53716232841606e-06, "loss": 0.3775, "step": 13192 }, { "epoch": 0.6673579847235571, "grad_norm": 8.415120124816895, "learning_rate": 2.536463865666267e-06, "loss": 0.4583, "step": 13193 }, { "epoch": 0.6674085689716223, "grad_norm": 11.739036560058594, "learning_rate": 2.5357654663958505e-06, "loss": 0.3596, "step": 13194 }, { "epoch": 0.6674591532196874, "grad_norm": 6.853958606719971, "learning_rate": 2.5350671306228035e-06, "loss": 0.2317, "step": 13195 }, { "epoch": 0.6675097374677526, "grad_norm": 6.205563068389893, "learning_rate": 2.5343688583651247e-06, "loss": 0.3939, "step": 13196 }, { "epoch": 0.6675603217158177, "grad_norm": 10.874692916870117, "learning_rate": 2.533670649640806e-06, "loss": 0.4754, "step": 13197 }, { "epoch": 0.6676109059638828, "grad_norm": 13.963449478149414, "learning_rate": 2.5329725044678344e-06, "loss": 0.5379, "step": 13198 }, { "epoch": 0.667661490211948, "grad_norm": 6.004893779754639, "learning_rate": 2.532274422864205e-06, "loss": 0.3164, "step": 13199 }, { "epoch": 0.6677120744600131, "grad_norm": 13.784607887268066, "learning_rate": 2.531576404847903e-06, "loss": 0.5754, "step": 13200 }, { "epoch": 0.6677626587080783, "grad_norm": 6.308990478515625, "learning_rate": 2.5308784504369146e-06, "loss": 0.3313, "step": 13201 }, { "epoch": 0.6678132429561434, "grad_norm": 5.070178031921387, "learning_rate": 2.5301805596492247e-06, "loss": 0.2822, "step": 13202 }, { "epoch": 0.6678638272042086, "grad_norm": 15.607210159301758, "learning_rate": 2.529482732502814e-06, "loss": 0.4745, "step": 13203 }, { "epoch": 0.6679144114522738, "grad_norm": 10.0017728805542, "learning_rate": 2.5287849690156687e-06, "loss": 0.4261, "step": 13204 }, { "epoch": 0.667964995700339, "grad_norm": 7.025292873382568, "learning_rate": 2.5280872692057658e-06, "loss": 0.3486, "step": 13205 }, { "epoch": 0.6680155799484041, "grad_norm": 5.84222936630249, "learning_rate": 2.5273896330910807e-06, "loss": 0.361, "step": 13206 }, { "epoch": 0.6680661641964692, "grad_norm": 7.225517272949219, "learning_rate": 2.526692060689594e-06, "loss": 0.4379, "step": 13207 }, { "epoch": 0.6681167484445344, "grad_norm": 5.493826866149902, "learning_rate": 2.5259945520192795e-06, "loss": 0.2938, "step": 13208 }, { "epoch": 0.6681673326925995, "grad_norm": 9.216073989868164, "learning_rate": 2.525297107098108e-06, "loss": 0.4543, "step": 13209 }, { "epoch": 0.6682179169406647, "grad_norm": 8.330507278442383, "learning_rate": 2.5245997259440543e-06, "loss": 0.4914, "step": 13210 }, { "epoch": 0.6682685011887298, "grad_norm": 11.62582778930664, "learning_rate": 2.523902408575086e-06, "loss": 0.4413, "step": 13211 }, { "epoch": 0.668319085436795, "grad_norm": 8.042136192321777, "learning_rate": 2.523205155009171e-06, "loss": 0.384, "step": 13212 }, { "epoch": 0.6683696696848601, "grad_norm": 8.571371078491211, "learning_rate": 2.522507965264278e-06, "loss": 0.4363, "step": 13213 }, { "epoch": 0.6684202539329253, "grad_norm": 11.938084602355957, "learning_rate": 2.521810839358373e-06, "loss": 0.4504, "step": 13214 }, { "epoch": 0.6684708381809904, "grad_norm": 8.642669677734375, "learning_rate": 2.521113777309412e-06, "loss": 0.5358, "step": 13215 }, { "epoch": 0.6685214224290555, "grad_norm": 5.5773234367370605, "learning_rate": 2.520416779135364e-06, "loss": 0.4368, "step": 13216 }, { "epoch": 0.6685720066771207, "grad_norm": 7.373754024505615, "learning_rate": 2.5197198448541863e-06, "loss": 0.3107, "step": 13217 }, { "epoch": 0.668622590925186, "grad_norm": 12.985015869140625, "learning_rate": 2.519022974483836e-06, "loss": 0.541, "step": 13218 }, { "epoch": 0.6686731751732511, "grad_norm": 11.582564353942871, "learning_rate": 2.518326168042272e-06, "loss": 0.4801, "step": 13219 }, { "epoch": 0.6687237594213162, "grad_norm": 7.447897911071777, "learning_rate": 2.5176294255474494e-06, "loss": 0.3765, "step": 13220 }, { "epoch": 0.6687743436693814, "grad_norm": 7.739511489868164, "learning_rate": 2.516932747017318e-06, "loss": 0.3598, "step": 13221 }, { "epoch": 0.6688249279174465, "grad_norm": 10.486753463745117, "learning_rate": 2.5162361324698343e-06, "loss": 0.3683, "step": 13222 }, { "epoch": 0.6688755121655117, "grad_norm": 5.4572038650512695, "learning_rate": 2.515539581922946e-06, "loss": 0.3465, "step": 13223 }, { "epoch": 0.6689260964135768, "grad_norm": 6.1736345291137695, "learning_rate": 2.5148430953946002e-06, "loss": 0.3248, "step": 13224 }, { "epoch": 0.668976680661642, "grad_norm": 6.994165897369385, "learning_rate": 2.5141466729027462e-06, "loss": 0.3599, "step": 13225 }, { "epoch": 0.6690272649097071, "grad_norm": 7.158275127410889, "learning_rate": 2.5134503144653287e-06, "loss": 0.4673, "step": 13226 }, { "epoch": 0.6690778491577722, "grad_norm": 4.715034484863281, "learning_rate": 2.5127540201002908e-06, "loss": 0.3741, "step": 13227 }, { "epoch": 0.6691284334058374, "grad_norm": 5.160700798034668, "learning_rate": 2.512057789825574e-06, "loss": 0.3706, "step": 13228 }, { "epoch": 0.6691790176539025, "grad_norm": 9.30040168762207, "learning_rate": 2.5113616236591187e-06, "loss": 0.4962, "step": 13229 }, { "epoch": 0.6692296019019677, "grad_norm": 5.4405517578125, "learning_rate": 2.5106655216188606e-06, "loss": 0.3661, "step": 13230 }, { "epoch": 0.6692801861500329, "grad_norm": 11.545384407043457, "learning_rate": 2.509969483722742e-06, "loss": 0.4604, "step": 13231 }, { "epoch": 0.6693307703980981, "grad_norm": 8.350296020507812, "learning_rate": 2.509273509988695e-06, "loss": 0.3283, "step": 13232 }, { "epoch": 0.6693813546461632, "grad_norm": 8.80345344543457, "learning_rate": 2.5085776004346516e-06, "loss": 0.3066, "step": 13233 }, { "epoch": 0.6694319388942284, "grad_norm": 9.895445823669434, "learning_rate": 2.5078817550785476e-06, "loss": 0.4042, "step": 13234 }, { "epoch": 0.6694825231422935, "grad_norm": 5.984574317932129, "learning_rate": 2.507185973938312e-06, "loss": 0.3972, "step": 13235 }, { "epoch": 0.6695331073903587, "grad_norm": 7.2183098793029785, "learning_rate": 2.50649025703187e-06, "loss": 0.3451, "step": 13236 }, { "epoch": 0.6695836916384238, "grad_norm": 7.21682071685791, "learning_rate": 2.505794604377153e-06, "loss": 0.3596, "step": 13237 }, { "epoch": 0.6696342758864889, "grad_norm": 12.183675765991211, "learning_rate": 2.505099015992085e-06, "loss": 0.4916, "step": 13238 }, { "epoch": 0.6696848601345541, "grad_norm": 5.990305423736572, "learning_rate": 2.5044034918945894e-06, "loss": 0.2889, "step": 13239 }, { "epoch": 0.6697354443826192, "grad_norm": 6.8202900886535645, "learning_rate": 2.5037080321025873e-06, "loss": 0.3276, "step": 13240 }, { "epoch": 0.6697860286306844, "grad_norm": 10.007292747497559, "learning_rate": 2.503012636633998e-06, "loss": 0.3899, "step": 13241 }, { "epoch": 0.6698366128787495, "grad_norm": 6.270829200744629, "learning_rate": 2.502317305506744e-06, "loss": 0.2572, "step": 13242 }, { "epoch": 0.6698871971268147, "grad_norm": 9.181107521057129, "learning_rate": 2.5016220387387402e-06, "loss": 0.4164, "step": 13243 }, { "epoch": 0.6699377813748799, "grad_norm": 9.155728340148926, "learning_rate": 2.5009268363478993e-06, "loss": 0.5168, "step": 13244 }, { "epoch": 0.6699883656229451, "grad_norm": 10.81943416595459, "learning_rate": 2.50023169835214e-06, "loss": 0.4685, "step": 13245 }, { "epoch": 0.6700389498710102, "grad_norm": 7.981943130493164, "learning_rate": 2.499536624769372e-06, "loss": 0.2979, "step": 13246 }, { "epoch": 0.6700895341190753, "grad_norm": 6.826574325561523, "learning_rate": 2.498841615617503e-06, "loss": 0.3733, "step": 13247 }, { "epoch": 0.6701401183671405, "grad_norm": 6.851012229919434, "learning_rate": 2.498146670914447e-06, "loss": 0.3338, "step": 13248 }, { "epoch": 0.6701907026152056, "grad_norm": 27.2289981842041, "learning_rate": 2.4974517906781075e-06, "loss": 0.3475, "step": 13249 }, { "epoch": 0.6702412868632708, "grad_norm": 7.083527565002441, "learning_rate": 2.4967569749263897e-06, "loss": 0.431, "step": 13250 }, { "epoch": 0.6702918711113359, "grad_norm": 9.552497863769531, "learning_rate": 2.4960622236772e-06, "loss": 0.4265, "step": 13251 }, { "epoch": 0.6703424553594011, "grad_norm": 6.659480571746826, "learning_rate": 2.49536753694844e-06, "loss": 0.3395, "step": 13252 }, { "epoch": 0.6703930396074662, "grad_norm": 8.222352027893066, "learning_rate": 2.4946729147580058e-06, "loss": 0.3205, "step": 13253 }, { "epoch": 0.6704436238555314, "grad_norm": 7.950098514556885, "learning_rate": 2.4939783571238006e-06, "loss": 0.3836, "step": 13254 }, { "epoch": 0.6704942081035965, "grad_norm": 7.612442493438721, "learning_rate": 2.4932838640637204e-06, "loss": 0.436, "step": 13255 }, { "epoch": 0.6705447923516616, "grad_norm": 7.95770788192749, "learning_rate": 2.492589435595658e-06, "loss": 0.3837, "step": 13256 }, { "epoch": 0.6705953765997269, "grad_norm": 9.495329856872559, "learning_rate": 2.491895071737512e-06, "loss": 0.449, "step": 13257 }, { "epoch": 0.670645960847792, "grad_norm": 6.792770862579346, "learning_rate": 2.4912007725071714e-06, "loss": 0.3367, "step": 13258 }, { "epoch": 0.6706965450958572, "grad_norm": 6.972211837768555, "learning_rate": 2.490506537922525e-06, "loss": 0.4431, "step": 13259 }, { "epoch": 0.6707471293439223, "grad_norm": 7.630728721618652, "learning_rate": 2.4898123680014665e-06, "loss": 0.3405, "step": 13260 }, { "epoch": 0.6707977135919875, "grad_norm": 7.198912620544434, "learning_rate": 2.4891182627618794e-06, "loss": 0.3545, "step": 13261 }, { "epoch": 0.6708482978400526, "grad_norm": 11.827189445495605, "learning_rate": 2.488424222221648e-06, "loss": 0.5071, "step": 13262 }, { "epoch": 0.6708988820881178, "grad_norm": 6.228579521179199, "learning_rate": 2.48773024639866e-06, "loss": 0.4005, "step": 13263 }, { "epoch": 0.6709494663361829, "grad_norm": 9.192242622375488, "learning_rate": 2.4870363353107957e-06, "loss": 0.5116, "step": 13264 }, { "epoch": 0.671000050584248, "grad_norm": 6.885652542114258, "learning_rate": 2.486342488975935e-06, "loss": 0.3547, "step": 13265 }, { "epoch": 0.6710506348323132, "grad_norm": 6.836886882781982, "learning_rate": 2.4856487074119567e-06, "loss": 0.3986, "step": 13266 }, { "epoch": 0.6711012190803783, "grad_norm": 4.777634620666504, "learning_rate": 2.4849549906367387e-06, "loss": 0.3151, "step": 13267 }, { "epoch": 0.6711518033284435, "grad_norm": 9.123109817504883, "learning_rate": 2.484261338668153e-06, "loss": 0.3275, "step": 13268 }, { "epoch": 0.6712023875765086, "grad_norm": 8.869312286376953, "learning_rate": 2.4835677515240786e-06, "loss": 0.3292, "step": 13269 }, { "epoch": 0.6712529718245738, "grad_norm": 6.979475498199463, "learning_rate": 2.4828742292223847e-06, "loss": 0.3865, "step": 13270 }, { "epoch": 0.671303556072639, "grad_norm": 5.32835054397583, "learning_rate": 2.4821807717809406e-06, "loss": 0.4314, "step": 13271 }, { "epoch": 0.6713541403207042, "grad_norm": 9.37427806854248, "learning_rate": 2.4814873792176175e-06, "loss": 0.4404, "step": 13272 }, { "epoch": 0.6714047245687693, "grad_norm": 7.566849231719971, "learning_rate": 2.480794051550282e-06, "loss": 0.3748, "step": 13273 }, { "epoch": 0.6714553088168345, "grad_norm": 6.287688255310059, "learning_rate": 2.480100788796797e-06, "loss": 0.4671, "step": 13274 }, { "epoch": 0.6715058930648996, "grad_norm": 13.913020133972168, "learning_rate": 2.47940759097503e-06, "loss": 0.4338, "step": 13275 }, { "epoch": 0.6715564773129648, "grad_norm": 10.665369033813477, "learning_rate": 2.478714458102841e-06, "loss": 0.4369, "step": 13276 }, { "epoch": 0.6716070615610299, "grad_norm": 6.438025951385498, "learning_rate": 2.478021390198089e-06, "loss": 0.3937, "step": 13277 }, { "epoch": 0.671657645809095, "grad_norm": 6.318982124328613, "learning_rate": 2.4773283872786376e-06, "loss": 0.3979, "step": 13278 }, { "epoch": 0.6717082300571602, "grad_norm": 8.877856254577637, "learning_rate": 2.4766354493623364e-06, "loss": 0.3754, "step": 13279 }, { "epoch": 0.6717588143052253, "grad_norm": 12.311195373535156, "learning_rate": 2.475942576467047e-06, "loss": 0.4663, "step": 13280 }, { "epoch": 0.6718093985532905, "grad_norm": 10.139065742492676, "learning_rate": 2.4752497686106203e-06, "loss": 0.3363, "step": 13281 }, { "epoch": 0.6718599828013556, "grad_norm": 6.794886589050293, "learning_rate": 2.4745570258109077e-06, "loss": 0.3203, "step": 13282 }, { "epoch": 0.6719105670494208, "grad_norm": 6.8472137451171875, "learning_rate": 2.4738643480857617e-06, "loss": 0.362, "step": 13283 }, { "epoch": 0.671961151297486, "grad_norm": 8.245051383972168, "learning_rate": 2.4731717354530304e-06, "loss": 0.4897, "step": 13284 }, { "epoch": 0.6720117355455512, "grad_norm": 7.52140998840332, "learning_rate": 2.4724791879305576e-06, "loss": 0.3074, "step": 13285 }, { "epoch": 0.6720623197936163, "grad_norm": 6.622494220733643, "learning_rate": 2.471786705536194e-06, "loss": 0.3418, "step": 13286 }, { "epoch": 0.6721129040416814, "grad_norm": 10.15329360961914, "learning_rate": 2.47109428828778e-06, "loss": 0.5037, "step": 13287 }, { "epoch": 0.6721634882897466, "grad_norm": 7.6136474609375, "learning_rate": 2.4704019362031557e-06, "loss": 0.3996, "step": 13288 }, { "epoch": 0.6722140725378117, "grad_norm": 10.589491844177246, "learning_rate": 2.4697096493001656e-06, "loss": 0.5839, "step": 13289 }, { "epoch": 0.6722646567858769, "grad_norm": 10.065564155578613, "learning_rate": 2.4690174275966456e-06, "loss": 0.47, "step": 13290 }, { "epoch": 0.672315241033942, "grad_norm": 9.385611534118652, "learning_rate": 2.4683252711104336e-06, "loss": 0.4195, "step": 13291 }, { "epoch": 0.6723658252820072, "grad_norm": 9.362775802612305, "learning_rate": 2.4676331798593645e-06, "loss": 0.4323, "step": 13292 }, { "epoch": 0.6724164095300723, "grad_norm": 9.166753768920898, "learning_rate": 2.4669411538612713e-06, "loss": 0.4299, "step": 13293 }, { "epoch": 0.6724669937781375, "grad_norm": 10.792315483093262, "learning_rate": 2.4662491931339848e-06, "loss": 0.4616, "step": 13294 }, { "epoch": 0.6725175780262026, "grad_norm": 10.59135627746582, "learning_rate": 2.4655572976953384e-06, "loss": 0.4091, "step": 13295 }, { "epoch": 0.6725681622742677, "grad_norm": 6.0968708992004395, "learning_rate": 2.4648654675631584e-06, "loss": 0.447, "step": 13296 }, { "epoch": 0.672618746522333, "grad_norm": 7.003658294677734, "learning_rate": 2.4641737027552702e-06, "loss": 0.4755, "step": 13297 }, { "epoch": 0.6726693307703981, "grad_norm": 6.420937538146973, "learning_rate": 2.4634820032895026e-06, "loss": 0.3905, "step": 13298 }, { "epoch": 0.6727199150184633, "grad_norm": 7.727128982543945, "learning_rate": 2.462790369183677e-06, "loss": 0.3722, "step": 13299 }, { "epoch": 0.6727704992665284, "grad_norm": 10.72866153717041, "learning_rate": 2.4620988004556135e-06, "loss": 0.4553, "step": 13300 }, { "epoch": 0.6728210835145936, "grad_norm": 8.42811393737793, "learning_rate": 2.4614072971231357e-06, "loss": 0.4121, "step": 13301 }, { "epoch": 0.6728716677626587, "grad_norm": 6.155718803405762, "learning_rate": 2.4607158592040596e-06, "loss": 0.4023, "step": 13302 }, { "epoch": 0.6729222520107239, "grad_norm": 6.6937031745910645, "learning_rate": 2.460024486716201e-06, "loss": 0.3735, "step": 13303 }, { "epoch": 0.672972836258789, "grad_norm": 16.74885368347168, "learning_rate": 2.45933317967738e-06, "loss": 0.6488, "step": 13304 }, { "epoch": 0.6730234205068542, "grad_norm": 14.058707237243652, "learning_rate": 2.4586419381054043e-06, "loss": 0.575, "step": 13305 }, { "epoch": 0.6730740047549193, "grad_norm": 5.296905994415283, "learning_rate": 2.4579507620180858e-06, "loss": 0.4508, "step": 13306 }, { "epoch": 0.6731245890029844, "grad_norm": 15.295157432556152, "learning_rate": 2.457259651433238e-06, "loss": 0.4156, "step": 13307 }, { "epoch": 0.6731751732510496, "grad_norm": 6.324324131011963, "learning_rate": 2.456568606368667e-06, "loss": 0.2417, "step": 13308 }, { "epoch": 0.6732257574991147, "grad_norm": 7.188587188720703, "learning_rate": 2.455877626842178e-06, "loss": 0.3588, "step": 13309 }, { "epoch": 0.6732763417471799, "grad_norm": 7.028690338134766, "learning_rate": 2.455186712871579e-06, "loss": 0.494, "step": 13310 }, { "epoch": 0.6733269259952451, "grad_norm": 6.060796737670898, "learning_rate": 2.454495864474672e-06, "loss": 0.3905, "step": 13311 }, { "epoch": 0.6733775102433103, "grad_norm": 5.559349536895752, "learning_rate": 2.4538050816692562e-06, "loss": 0.3322, "step": 13312 }, { "epoch": 0.6734280944913754, "grad_norm": 5.495506286621094, "learning_rate": 2.4531143644731355e-06, "loss": 0.4436, "step": 13313 }, { "epoch": 0.6734786787394406, "grad_norm": 16.394882202148438, "learning_rate": 2.4524237129041052e-06, "loss": 0.4282, "step": 13314 }, { "epoch": 0.6735292629875057, "grad_norm": 10.000578880310059, "learning_rate": 2.451733126979961e-06, "loss": 0.4377, "step": 13315 }, { "epoch": 0.6735798472355708, "grad_norm": 7.907893657684326, "learning_rate": 2.4510426067185004e-06, "loss": 0.3711, "step": 13316 }, { "epoch": 0.673630431483636, "grad_norm": 8.098909378051758, "learning_rate": 2.4503521521375157e-06, "loss": 0.4601, "step": 13317 }, { "epoch": 0.6736810157317011, "grad_norm": 9.608802795410156, "learning_rate": 2.4496617632547976e-06, "loss": 0.5033, "step": 13318 }, { "epoch": 0.6737315999797663, "grad_norm": 6.613749027252197, "learning_rate": 2.4489714400881354e-06, "loss": 0.4301, "step": 13319 }, { "epoch": 0.6737821842278314, "grad_norm": 15.146153450012207, "learning_rate": 2.4482811826553156e-06, "loss": 0.3858, "step": 13320 }, { "epoch": 0.6738327684758966, "grad_norm": 8.52612590789795, "learning_rate": 2.447590990974128e-06, "loss": 0.3474, "step": 13321 }, { "epoch": 0.6738833527239617, "grad_norm": 4.59497594833374, "learning_rate": 2.446900865062356e-06, "loss": 0.3359, "step": 13322 }, { "epoch": 0.6739339369720269, "grad_norm": 5.634446144104004, "learning_rate": 2.4462108049377796e-06, "loss": 0.3877, "step": 13323 }, { "epoch": 0.6739845212200921, "grad_norm": 11.719677925109863, "learning_rate": 2.445520810618184e-06, "loss": 0.6016, "step": 13324 }, { "epoch": 0.6740351054681573, "grad_norm": 9.03028678894043, "learning_rate": 2.444830882121348e-06, "loss": 0.4281, "step": 13325 }, { "epoch": 0.6740856897162224, "grad_norm": 4.702240943908691, "learning_rate": 2.4441410194650476e-06, "loss": 0.4449, "step": 13326 }, { "epoch": 0.6741362739642875, "grad_norm": 6.114233493804932, "learning_rate": 2.443451222667058e-06, "loss": 0.3528, "step": 13327 }, { "epoch": 0.6741868582123527, "grad_norm": 11.538116455078125, "learning_rate": 2.4427614917451576e-06, "loss": 0.4418, "step": 13328 }, { "epoch": 0.6742374424604178, "grad_norm": 10.182848930358887, "learning_rate": 2.4420718267171163e-06, "loss": 0.417, "step": 13329 }, { "epoch": 0.674288026708483, "grad_norm": 8.550455093383789, "learning_rate": 2.441382227600706e-06, "loss": 0.4557, "step": 13330 }, { "epoch": 0.6743386109565481, "grad_norm": 8.698002815246582, "learning_rate": 2.440692694413695e-06, "loss": 0.4259, "step": 13331 }, { "epoch": 0.6743891952046133, "grad_norm": 6.019649982452393, "learning_rate": 2.44000322717385e-06, "loss": 0.4313, "step": 13332 }, { "epoch": 0.6744397794526784, "grad_norm": 12.205236434936523, "learning_rate": 2.4393138258989406e-06, "loss": 0.5088, "step": 13333 }, { "epoch": 0.6744903637007436, "grad_norm": 11.734068870544434, "learning_rate": 2.4386244906067284e-06, "loss": 0.4145, "step": 13334 }, { "epoch": 0.6745409479488087, "grad_norm": 9.905016899108887, "learning_rate": 2.4379352213149744e-06, "loss": 0.4464, "step": 13335 }, { "epoch": 0.6745915321968738, "grad_norm": 7.961659908294678, "learning_rate": 2.4372460180414436e-06, "loss": 0.4195, "step": 13336 }, { "epoch": 0.6746421164449391, "grad_norm": 6.440505504608154, "learning_rate": 2.4365568808038927e-06, "loss": 0.4256, "step": 13337 }, { "epoch": 0.6746927006930042, "grad_norm": 8.610452651977539, "learning_rate": 2.435867809620077e-06, "loss": 0.3465, "step": 13338 }, { "epoch": 0.6747432849410694, "grad_norm": 8.532455444335938, "learning_rate": 2.4351788045077566e-06, "loss": 0.4491, "step": 13339 }, { "epoch": 0.6747938691891345, "grad_norm": 7.994866847991943, "learning_rate": 2.4344898654846834e-06, "loss": 0.3555, "step": 13340 }, { "epoch": 0.6748444534371997, "grad_norm": 7.858415603637695, "learning_rate": 2.4338009925686074e-06, "loss": 0.471, "step": 13341 }, { "epoch": 0.6748950376852648, "grad_norm": 12.583194732666016, "learning_rate": 2.4331121857772854e-06, "loss": 0.6103, "step": 13342 }, { "epoch": 0.67494562193333, "grad_norm": 5.937056541442871, "learning_rate": 2.4324234451284606e-06, "loss": 0.2771, "step": 13343 }, { "epoch": 0.6749962061813951, "grad_norm": 6.915391445159912, "learning_rate": 2.43173477063988e-06, "loss": 0.3545, "step": 13344 }, { "epoch": 0.6750467904294603, "grad_norm": 16.702152252197266, "learning_rate": 2.431046162329293e-06, "loss": 0.5733, "step": 13345 }, { "epoch": 0.6750973746775254, "grad_norm": 10.39098834991455, "learning_rate": 2.430357620214441e-06, "loss": 0.372, "step": 13346 }, { "epoch": 0.6751479589255905, "grad_norm": 20.557109832763672, "learning_rate": 2.4296691443130643e-06, "loss": 0.5136, "step": 13347 }, { "epoch": 0.6751985431736557, "grad_norm": 14.325553894042969, "learning_rate": 2.428980734642908e-06, "loss": 0.4637, "step": 13348 }, { "epoch": 0.6752491274217208, "grad_norm": 11.64123821258545, "learning_rate": 2.4282923912217076e-06, "loss": 0.5047, "step": 13349 }, { "epoch": 0.675299711669786, "grad_norm": 6.5524067878723145, "learning_rate": 2.4276041140671987e-06, "loss": 0.4533, "step": 13350 }, { "epoch": 0.6753502959178512, "grad_norm": 5.759267330169678, "learning_rate": 2.42691590319712e-06, "loss": 0.3986, "step": 13351 }, { "epoch": 0.6754008801659164, "grad_norm": 8.028854370117188, "learning_rate": 2.4262277586292043e-06, "loss": 0.4013, "step": 13352 }, { "epoch": 0.6754514644139815, "grad_norm": 6.528998851776123, "learning_rate": 2.42553968038118e-06, "loss": 0.3309, "step": 13353 }, { "epoch": 0.6755020486620467, "grad_norm": 7.8012495040893555, "learning_rate": 2.4248516684707816e-06, "loss": 0.2918, "step": 13354 }, { "epoch": 0.6755526329101118, "grad_norm": 6.234572410583496, "learning_rate": 2.424163722915736e-06, "loss": 0.4289, "step": 13355 }, { "epoch": 0.675603217158177, "grad_norm": 11.184062957763672, "learning_rate": 2.4234758437337697e-06, "loss": 0.439, "step": 13356 }, { "epoch": 0.6756538014062421, "grad_norm": 7.632632255554199, "learning_rate": 2.422788030942608e-06, "loss": 0.4205, "step": 13357 }, { "epoch": 0.6757043856543072, "grad_norm": 5.8680009841918945, "learning_rate": 2.4221002845599708e-06, "loss": 0.323, "step": 13358 }, { "epoch": 0.6757549699023724, "grad_norm": 9.63631534576416, "learning_rate": 2.4214126046035853e-06, "loss": 0.4897, "step": 13359 }, { "epoch": 0.6758055541504375, "grad_norm": 6.577164173126221, "learning_rate": 2.420724991091168e-06, "loss": 0.3664, "step": 13360 }, { "epoch": 0.6758561383985027, "grad_norm": 5.125473976135254, "learning_rate": 2.420037444040436e-06, "loss": 0.376, "step": 13361 }, { "epoch": 0.6759067226465678, "grad_norm": 9.687027931213379, "learning_rate": 2.4193499634691097e-06, "loss": 0.3852, "step": 13362 }, { "epoch": 0.675957306894633, "grad_norm": 8.448984146118164, "learning_rate": 2.4186625493949014e-06, "loss": 0.423, "step": 13363 }, { "epoch": 0.6760078911426982, "grad_norm": 8.869292259216309, "learning_rate": 2.417975201835524e-06, "loss": 0.5168, "step": 13364 }, { "epoch": 0.6760584753907634, "grad_norm": 5.935724258422852, "learning_rate": 2.417287920808687e-06, "loss": 0.3652, "step": 13365 }, { "epoch": 0.6761090596388285, "grad_norm": 9.06412410736084, "learning_rate": 2.416600706332104e-06, "loss": 0.3835, "step": 13366 }, { "epoch": 0.6761596438868936, "grad_norm": 6.015905380249023, "learning_rate": 2.415913558423481e-06, "loss": 0.3931, "step": 13367 }, { "epoch": 0.6762102281349588, "grad_norm": 5.087510108947754, "learning_rate": 2.415226477100524e-06, "loss": 0.3561, "step": 13368 }, { "epoch": 0.6762608123830239, "grad_norm": 4.942553520202637, "learning_rate": 2.4145394623809375e-06, "loss": 0.3033, "step": 13369 }, { "epoch": 0.6763113966310891, "grad_norm": 8.182356834411621, "learning_rate": 2.413852514282422e-06, "loss": 0.4056, "step": 13370 }, { "epoch": 0.6763619808791542, "grad_norm": 9.982475280761719, "learning_rate": 2.4131656328226827e-06, "loss": 0.4309, "step": 13371 }, { "epoch": 0.6764125651272194, "grad_norm": 6.659463882446289, "learning_rate": 2.4124788180194168e-06, "loss": 0.4442, "step": 13372 }, { "epoch": 0.6764631493752845, "grad_norm": 9.48493766784668, "learning_rate": 2.4117920698903204e-06, "loss": 0.4492, "step": 13373 }, { "epoch": 0.6765137336233497, "grad_norm": 7.059531211853027, "learning_rate": 2.4111053884530916e-06, "loss": 0.4074, "step": 13374 }, { "epoch": 0.6765643178714148, "grad_norm": 5.868401527404785, "learning_rate": 2.410418773725425e-06, "loss": 0.4503, "step": 13375 }, { "epoch": 0.6766149021194799, "grad_norm": 6.607141494750977, "learning_rate": 2.409732225725009e-06, "loss": 0.4013, "step": 13376 }, { "epoch": 0.6766654863675452, "grad_norm": 6.431273937225342, "learning_rate": 2.409045744469539e-06, "loss": 0.3423, "step": 13377 }, { "epoch": 0.6767160706156103, "grad_norm": 6.967508792877197, "learning_rate": 2.4083593299767017e-06, "loss": 0.3926, "step": 13378 }, { "epoch": 0.6767666548636755, "grad_norm": 9.420310974121094, "learning_rate": 2.407672982264183e-06, "loss": 0.5111, "step": 13379 }, { "epoch": 0.6768172391117406, "grad_norm": 5.120710372924805, "learning_rate": 2.406986701349672e-06, "loss": 0.3306, "step": 13380 }, { "epoch": 0.6768678233598058, "grad_norm": 15.02786636352539, "learning_rate": 2.406300487250853e-06, "loss": 0.465, "step": 13381 }, { "epoch": 0.6769184076078709, "grad_norm": 7.788782119750977, "learning_rate": 2.4056143399854008e-06, "loss": 0.4031, "step": 13382 }, { "epoch": 0.6769689918559361, "grad_norm": 9.529487609863281, "learning_rate": 2.404928259571003e-06, "loss": 0.4255, "step": 13383 }, { "epoch": 0.6770195761040012, "grad_norm": 8.51677131652832, "learning_rate": 2.4042422460253355e-06, "loss": 0.4981, "step": 13384 }, { "epoch": 0.6770701603520664, "grad_norm": 5.65763521194458, "learning_rate": 2.403556299366074e-06, "loss": 0.387, "step": 13385 }, { "epoch": 0.6771207446001315, "grad_norm": 8.91249942779541, "learning_rate": 2.402870419610897e-06, "loss": 0.325, "step": 13386 }, { "epoch": 0.6771713288481966, "grad_norm": 6.9928131103515625, "learning_rate": 2.402184606777476e-06, "loss": 0.4861, "step": 13387 }, { "epoch": 0.6772219130962618, "grad_norm": 5.859355926513672, "learning_rate": 2.4014988608834815e-06, "loss": 0.4635, "step": 13388 }, { "epoch": 0.6772724973443269, "grad_norm": 8.378761291503906, "learning_rate": 2.4008131819465862e-06, "loss": 0.3411, "step": 13389 }, { "epoch": 0.6773230815923922, "grad_norm": 17.115970611572266, "learning_rate": 2.4001275699844573e-06, "loss": 0.5204, "step": 13390 }, { "epoch": 0.6773736658404573, "grad_norm": 5.387894153594971, "learning_rate": 2.3994420250147597e-06, "loss": 0.4498, "step": 13391 }, { "epoch": 0.6774242500885225, "grad_norm": 8.127541542053223, "learning_rate": 2.398756547055161e-06, "loss": 0.4661, "step": 13392 }, { "epoch": 0.6774748343365876, "grad_norm": 6.13085412979126, "learning_rate": 2.398071136123324e-06, "loss": 0.3904, "step": 13393 }, { "epoch": 0.6775254185846528, "grad_norm": 8.662410736083984, "learning_rate": 2.397385792236908e-06, "loss": 0.4164, "step": 13394 }, { "epoch": 0.6775760028327179, "grad_norm": 5.696216583251953, "learning_rate": 2.396700515413574e-06, "loss": 0.3116, "step": 13395 }, { "epoch": 0.677626587080783, "grad_norm": 5.598812580108643, "learning_rate": 2.3960153056709784e-06, "loss": 0.3523, "step": 13396 }, { "epoch": 0.6776771713288482, "grad_norm": 10.075974464416504, "learning_rate": 2.3953301630267806e-06, "loss": 0.4473, "step": 13397 }, { "epoch": 0.6777277555769133, "grad_norm": 8.704951286315918, "learning_rate": 2.3946450874986327e-06, "loss": 0.5201, "step": 13398 }, { "epoch": 0.6777783398249785, "grad_norm": 6.765599727630615, "learning_rate": 2.393960079104188e-06, "loss": 0.3867, "step": 13399 }, { "epoch": 0.6778289240730436, "grad_norm": 5.395793914794922, "learning_rate": 2.3932751378610957e-06, "loss": 0.3366, "step": 13400 }, { "epoch": 0.6778795083211088, "grad_norm": 7.650719165802002, "learning_rate": 2.3925902637870085e-06, "loss": 0.4143, "step": 13401 }, { "epoch": 0.6779300925691739, "grad_norm": 5.854029655456543, "learning_rate": 2.3919054568995725e-06, "loss": 0.3347, "step": 13402 }, { "epoch": 0.6779806768172391, "grad_norm": 7.932086944580078, "learning_rate": 2.391220717216431e-06, "loss": 0.4676, "step": 13403 }, { "epoch": 0.6780312610653043, "grad_norm": 8.193005561828613, "learning_rate": 2.390536044755233e-06, "loss": 0.3672, "step": 13404 }, { "epoch": 0.6780818453133695, "grad_norm": 5.255403518676758, "learning_rate": 2.3898514395336175e-06, "loss": 0.4421, "step": 13405 }, { "epoch": 0.6781324295614346, "grad_norm": 6.648332118988037, "learning_rate": 2.3891669015692244e-06, "loss": 0.3038, "step": 13406 }, { "epoch": 0.6781830138094997, "grad_norm": 7.382088661193848, "learning_rate": 2.388482430879697e-06, "loss": 0.3873, "step": 13407 }, { "epoch": 0.6782335980575649, "grad_norm": 9.754962921142578, "learning_rate": 2.387798027482667e-06, "loss": 0.4334, "step": 13408 }, { "epoch": 0.67828418230563, "grad_norm": 7.0281171798706055, "learning_rate": 2.3871136913957734e-06, "loss": 0.3749, "step": 13409 }, { "epoch": 0.6783347665536952, "grad_norm": 7.601254940032959, "learning_rate": 2.3864294226366485e-06, "loss": 0.4623, "step": 13410 }, { "epoch": 0.6783853508017603, "grad_norm": 8.63067626953125, "learning_rate": 2.3857452212229233e-06, "loss": 0.2725, "step": 13411 }, { "epoch": 0.6784359350498255, "grad_norm": 14.547616958618164, "learning_rate": 2.3850610871722313e-06, "loss": 0.5313, "step": 13412 }, { "epoch": 0.6784865192978906, "grad_norm": 10.710927963256836, "learning_rate": 2.3843770205021983e-06, "loss": 0.4589, "step": 13413 }, { "epoch": 0.6785371035459558, "grad_norm": 8.808732986450195, "learning_rate": 2.3836930212304503e-06, "loss": 0.4232, "step": 13414 }, { "epoch": 0.6785876877940209, "grad_norm": 7.975308418273926, "learning_rate": 2.3830090893746157e-06, "loss": 0.4007, "step": 13415 }, { "epoch": 0.678638272042086, "grad_norm": 5.53261137008667, "learning_rate": 2.3823252249523154e-06, "loss": 0.4068, "step": 13416 }, { "epoch": 0.6786888562901513, "grad_norm": 7.827544689178467, "learning_rate": 2.3816414279811693e-06, "loss": 0.3721, "step": 13417 }, { "epoch": 0.6787394405382164, "grad_norm": 8.436697006225586, "learning_rate": 2.3809576984788013e-06, "loss": 0.364, "step": 13418 }, { "epoch": 0.6787900247862816, "grad_norm": 7.519589900970459, "learning_rate": 2.3802740364628274e-06, "loss": 0.3269, "step": 13419 }, { "epoch": 0.6788406090343467, "grad_norm": 10.170055389404297, "learning_rate": 2.3795904419508644e-06, "loss": 0.4364, "step": 13420 }, { "epoch": 0.6788911932824119, "grad_norm": 14.620129585266113, "learning_rate": 2.3789069149605253e-06, "loss": 0.4502, "step": 13421 }, { "epoch": 0.678941777530477, "grad_norm": 8.203983306884766, "learning_rate": 2.3782234555094247e-06, "loss": 0.3889, "step": 13422 }, { "epoch": 0.6789923617785422, "grad_norm": 6.668229579925537, "learning_rate": 2.3775400636151715e-06, "loss": 0.3432, "step": 13423 }, { "epoch": 0.6790429460266073, "grad_norm": 11.014605522155762, "learning_rate": 2.3768567392953777e-06, "loss": 0.4713, "step": 13424 }, { "epoch": 0.6790935302746725, "grad_norm": 7.680387020111084, "learning_rate": 2.37617348256765e-06, "loss": 0.3383, "step": 13425 }, { "epoch": 0.6791441145227376, "grad_norm": 10.835834503173828, "learning_rate": 2.375490293449592e-06, "loss": 0.407, "step": 13426 }, { "epoch": 0.6791946987708027, "grad_norm": 10.193556785583496, "learning_rate": 2.3748071719588125e-06, "loss": 0.4471, "step": 13427 }, { "epoch": 0.6792452830188679, "grad_norm": 7.937143325805664, "learning_rate": 2.3741241181129106e-06, "loss": 0.4197, "step": 13428 }, { "epoch": 0.679295867266933, "grad_norm": 8.964977264404297, "learning_rate": 2.373441131929486e-06, "loss": 0.3208, "step": 13429 }, { "epoch": 0.6793464515149983, "grad_norm": 10.767753601074219, "learning_rate": 2.3727582134261413e-06, "loss": 0.4807, "step": 13430 }, { "epoch": 0.6793970357630634, "grad_norm": 6.119144439697266, "learning_rate": 2.372075362620471e-06, "loss": 0.2553, "step": 13431 }, { "epoch": 0.6794476200111286, "grad_norm": 10.68321418762207, "learning_rate": 2.3713925795300717e-06, "loss": 0.3937, "step": 13432 }, { "epoch": 0.6794982042591937, "grad_norm": 8.140617370605469, "learning_rate": 2.370709864172536e-06, "loss": 0.3725, "step": 13433 }, { "epoch": 0.6795487885072589, "grad_norm": 5.991878032684326, "learning_rate": 2.370027216565455e-06, "loss": 0.4523, "step": 13434 }, { "epoch": 0.679599372755324, "grad_norm": 10.217729568481445, "learning_rate": 2.369344636726422e-06, "loss": 0.4391, "step": 13435 }, { "epoch": 0.6796499570033891, "grad_norm": 12.924407958984375, "learning_rate": 2.3686621246730233e-06, "loss": 0.4345, "step": 13436 }, { "epoch": 0.6797005412514543, "grad_norm": 12.180553436279297, "learning_rate": 2.367979680422846e-06, "loss": 0.4282, "step": 13437 }, { "epoch": 0.6797511254995194, "grad_norm": 26.80433464050293, "learning_rate": 2.3672973039934733e-06, "loss": 0.3811, "step": 13438 }, { "epoch": 0.6798017097475846, "grad_norm": 6.796029090881348, "learning_rate": 2.366614995402491e-06, "loss": 0.4333, "step": 13439 }, { "epoch": 0.6798522939956497, "grad_norm": 5.444977283477783, "learning_rate": 2.365932754667481e-06, "loss": 0.4218, "step": 13440 }, { "epoch": 0.6799028782437149, "grad_norm": 8.67966365814209, "learning_rate": 2.365250581806019e-06, "loss": 0.3993, "step": 13441 }, { "epoch": 0.67995346249178, "grad_norm": 8.044013977050781, "learning_rate": 2.3645684768356867e-06, "loss": 0.358, "step": 13442 }, { "epoch": 0.6800040467398452, "grad_norm": 9.204130172729492, "learning_rate": 2.36388643977406e-06, "loss": 0.5537, "step": 13443 }, { "epoch": 0.6800546309879104, "grad_norm": 24.03554916381836, "learning_rate": 2.3632044706387107e-06, "loss": 0.4332, "step": 13444 }, { "epoch": 0.6801052152359756, "grad_norm": 5.682765007019043, "learning_rate": 2.362522569447217e-06, "loss": 0.4186, "step": 13445 }, { "epoch": 0.6801557994840407, "grad_norm": 7.778204917907715, "learning_rate": 2.361840736217142e-06, "loss": 0.4192, "step": 13446 }, { "epoch": 0.6802063837321058, "grad_norm": 10.010477066040039, "learning_rate": 2.361158970966061e-06, "loss": 0.4424, "step": 13447 }, { "epoch": 0.680256967980171, "grad_norm": 6.341604232788086, "learning_rate": 2.3604772737115396e-06, "loss": 0.3146, "step": 13448 }, { "epoch": 0.6803075522282361, "grad_norm": 9.350868225097656, "learning_rate": 2.359795644471141e-06, "loss": 0.459, "step": 13449 }, { "epoch": 0.6803581364763013, "grad_norm": 8.403532981872559, "learning_rate": 2.3591140832624336e-06, "loss": 0.3596, "step": 13450 }, { "epoch": 0.6804087207243664, "grad_norm": 8.069618225097656, "learning_rate": 2.358432590102978e-06, "loss": 0.3886, "step": 13451 }, { "epoch": 0.6804593049724316, "grad_norm": 6.8678388595581055, "learning_rate": 2.357751165010332e-06, "loss": 0.4419, "step": 13452 }, { "epoch": 0.6805098892204967, "grad_norm": 5.80531644821167, "learning_rate": 2.3570698080020575e-06, "loss": 0.4218, "step": 13453 }, { "epoch": 0.6805604734685619, "grad_norm": 6.435197830200195, "learning_rate": 2.3563885190957113e-06, "loss": 0.3996, "step": 13454 }, { "epoch": 0.680611057716627, "grad_norm": 16.787952423095703, "learning_rate": 2.3557072983088453e-06, "loss": 0.4252, "step": 13455 }, { "epoch": 0.6806616419646921, "grad_norm": 7.500521183013916, "learning_rate": 2.355026145659017e-06, "loss": 0.3836, "step": 13456 }, { "epoch": 0.6807122262127574, "grad_norm": 8.523618698120117, "learning_rate": 2.3543450611637763e-06, "loss": 0.3331, "step": 13457 }, { "epoch": 0.6807628104608225, "grad_norm": 6.133610725402832, "learning_rate": 2.3536640448406726e-06, "loss": 0.3744, "step": 13458 }, { "epoch": 0.6808133947088877, "grad_norm": 5.926963806152344, "learning_rate": 2.3529830967072554e-06, "loss": 0.3594, "step": 13459 }, { "epoch": 0.6808639789569528, "grad_norm": 4.803000450134277, "learning_rate": 2.35230221678107e-06, "loss": 0.4029, "step": 13460 }, { "epoch": 0.680914563205018, "grad_norm": 4.687860488891602, "learning_rate": 2.351621405079659e-06, "loss": 0.3925, "step": 13461 }, { "epoch": 0.6809651474530831, "grad_norm": 9.381343841552734, "learning_rate": 2.350940661620569e-06, "loss": 0.4799, "step": 13462 }, { "epoch": 0.6810157317011483, "grad_norm": 5.335792064666748, "learning_rate": 2.3502599864213405e-06, "loss": 0.3266, "step": 13463 }, { "epoch": 0.6810663159492134, "grad_norm": 5.740725994110107, "learning_rate": 2.34957937949951e-06, "loss": 0.3847, "step": 13464 }, { "epoch": 0.6811169001972786, "grad_norm": 5.733452796936035, "learning_rate": 2.348898840872618e-06, "loss": 0.3599, "step": 13465 }, { "epoch": 0.6811674844453437, "grad_norm": 17.86054039001465, "learning_rate": 2.348218370558201e-06, "loss": 0.4903, "step": 13466 }, { "epoch": 0.6812180686934088, "grad_norm": 8.99863338470459, "learning_rate": 2.3475379685737882e-06, "loss": 0.2867, "step": 13467 }, { "epoch": 0.681268652941474, "grad_norm": 5.516175270080566, "learning_rate": 2.346857634936917e-06, "loss": 0.2977, "step": 13468 }, { "epoch": 0.6813192371895391, "grad_norm": 8.021891593933105, "learning_rate": 2.346177369665117e-06, "loss": 0.3803, "step": 13469 }, { "epoch": 0.6813698214376044, "grad_norm": 10.94009780883789, "learning_rate": 2.345497172775914e-06, "loss": 0.4402, "step": 13470 }, { "epoch": 0.6814204056856695, "grad_norm": 12.772378921508789, "learning_rate": 2.3448170442868405e-06, "loss": 0.3934, "step": 13471 }, { "epoch": 0.6814709899337347, "grad_norm": 8.117870330810547, "learning_rate": 2.344136984215417e-06, "loss": 0.3993, "step": 13472 }, { "epoch": 0.6815215741817998, "grad_norm": 7.055569171905518, "learning_rate": 2.3434569925791668e-06, "loss": 0.3754, "step": 13473 }, { "epoch": 0.681572158429865, "grad_norm": 11.150189399719238, "learning_rate": 2.3427770693956148e-06, "loss": 0.4479, "step": 13474 }, { "epoch": 0.6816227426779301, "grad_norm": 8.249679565429688, "learning_rate": 2.342097214682279e-06, "loss": 0.3668, "step": 13475 }, { "epoch": 0.6816733269259952, "grad_norm": 6.763067722320557, "learning_rate": 2.3414174284566772e-06, "loss": 0.3965, "step": 13476 }, { "epoch": 0.6817239111740604, "grad_norm": 11.282882690429688, "learning_rate": 2.3407377107363275e-06, "loss": 0.5296, "step": 13477 }, { "epoch": 0.6817744954221255, "grad_norm": 9.272120475769043, "learning_rate": 2.3400580615387448e-06, "loss": 0.4824, "step": 13478 }, { "epoch": 0.6818250796701907, "grad_norm": 7.400716304779053, "learning_rate": 2.3393784808814384e-06, "loss": 0.2986, "step": 13479 }, { "epoch": 0.6818756639182558, "grad_norm": 4.927995204925537, "learning_rate": 2.3386989687819243e-06, "loss": 0.3915, "step": 13480 }, { "epoch": 0.681926248166321, "grad_norm": 7.946291923522949, "learning_rate": 2.3380195252577097e-06, "loss": 0.4686, "step": 13481 }, { "epoch": 0.6819768324143861, "grad_norm": 6.150454998016357, "learning_rate": 2.3373401503263006e-06, "loss": 0.3039, "step": 13482 }, { "epoch": 0.6820274166624513, "grad_norm": 8.744576454162598, "learning_rate": 2.3366608440052057e-06, "loss": 0.3393, "step": 13483 }, { "epoch": 0.6820780009105165, "grad_norm": 4.342630386352539, "learning_rate": 2.3359816063119285e-06, "loss": 0.317, "step": 13484 }, { "epoch": 0.6821285851585817, "grad_norm": 9.07540225982666, "learning_rate": 2.33530243726397e-06, "loss": 0.4076, "step": 13485 }, { "epoch": 0.6821791694066468, "grad_norm": 6.831230163574219, "learning_rate": 2.3346233368788314e-06, "loss": 0.3566, "step": 13486 }, { "epoch": 0.682229753654712, "grad_norm": 10.428467750549316, "learning_rate": 2.3339443051740097e-06, "loss": 0.3125, "step": 13487 }, { "epoch": 0.6822803379027771, "grad_norm": 11.64312744140625, "learning_rate": 2.333265342167006e-06, "loss": 0.5503, "step": 13488 }, { "epoch": 0.6823309221508422, "grad_norm": 16.501134872436523, "learning_rate": 2.3325864478753124e-06, "loss": 0.4589, "step": 13489 }, { "epoch": 0.6823815063989074, "grad_norm": 10.16103744506836, "learning_rate": 2.331907622316422e-06, "loss": 0.5039, "step": 13490 }, { "epoch": 0.6824320906469725, "grad_norm": 8.975566864013672, "learning_rate": 2.331228865507829e-06, "loss": 0.3967, "step": 13491 }, { "epoch": 0.6824826748950377, "grad_norm": 8.314799308776855, "learning_rate": 2.3305501774670224e-06, "loss": 0.308, "step": 13492 }, { "epoch": 0.6825332591431028, "grad_norm": 5.985483646392822, "learning_rate": 2.329871558211488e-06, "loss": 0.4399, "step": 13493 }, { "epoch": 0.682583843391168, "grad_norm": 9.425260543823242, "learning_rate": 2.3291930077587156e-06, "loss": 0.4875, "step": 13494 }, { "epoch": 0.6826344276392331, "grad_norm": 8.069583892822266, "learning_rate": 2.3285145261261886e-06, "loss": 0.3508, "step": 13495 }, { "epoch": 0.6826850118872982, "grad_norm": 5.229923725128174, "learning_rate": 2.3278361133313877e-06, "loss": 0.3946, "step": 13496 }, { "epoch": 0.6827355961353635, "grad_norm": 10.190951347351074, "learning_rate": 2.3271577693917997e-06, "loss": 0.469, "step": 13497 }, { "epoch": 0.6827861803834286, "grad_norm": 6.970768928527832, "learning_rate": 2.326479494324898e-06, "loss": 0.4729, "step": 13498 }, { "epoch": 0.6828367646314938, "grad_norm": 9.94206714630127, "learning_rate": 2.32580128814816e-06, "loss": 0.3342, "step": 13499 }, { "epoch": 0.6828873488795589, "grad_norm": 7.167092323303223, "learning_rate": 2.3251231508790655e-06, "loss": 0.4225, "step": 13500 }, { "epoch": 0.6829379331276241, "grad_norm": 8.562329292297363, "learning_rate": 2.324445082535087e-06, "loss": 0.4412, "step": 13501 }, { "epoch": 0.6829885173756892, "grad_norm": 14.2152738571167, "learning_rate": 2.323767083133694e-06, "loss": 0.4255, "step": 13502 }, { "epoch": 0.6830391016237544, "grad_norm": 10.763371467590332, "learning_rate": 2.3230891526923604e-06, "loss": 0.4407, "step": 13503 }, { "epoch": 0.6830896858718195, "grad_norm": 9.232928276062012, "learning_rate": 2.322411291228554e-06, "loss": 0.4608, "step": 13504 }, { "epoch": 0.6831402701198847, "grad_norm": 6.239371299743652, "learning_rate": 2.3217334987597393e-06, "loss": 0.3559, "step": 13505 }, { "epoch": 0.6831908543679498, "grad_norm": 5.903054714202881, "learning_rate": 2.321055775303385e-06, "loss": 0.378, "step": 13506 }, { "epoch": 0.6832414386160149, "grad_norm": 7.887756824493408, "learning_rate": 2.320378120876953e-06, "loss": 0.3714, "step": 13507 }, { "epoch": 0.6832920228640801, "grad_norm": 7.357497215270996, "learning_rate": 2.3197005354979024e-06, "loss": 0.2535, "step": 13508 }, { "epoch": 0.6833426071121452, "grad_norm": 7.386750221252441, "learning_rate": 2.3190230191836972e-06, "loss": 0.366, "step": 13509 }, { "epoch": 0.6833931913602105, "grad_norm": 8.009525299072266, "learning_rate": 2.3183455719517945e-06, "loss": 0.3444, "step": 13510 }, { "epoch": 0.6834437756082756, "grad_norm": 16.894453048706055, "learning_rate": 2.3176681938196454e-06, "loss": 0.4665, "step": 13511 }, { "epoch": 0.6834943598563408, "grad_norm": 17.53318214416504, "learning_rate": 2.31699088480471e-06, "loss": 0.469, "step": 13512 }, { "epoch": 0.6835449441044059, "grad_norm": 7.818248271942139, "learning_rate": 2.3163136449244387e-06, "loss": 0.3788, "step": 13513 }, { "epoch": 0.6835955283524711, "grad_norm": 10.470901489257812, "learning_rate": 2.3156364741962815e-06, "loss": 0.4886, "step": 13514 }, { "epoch": 0.6836461126005362, "grad_norm": 12.586280822753906, "learning_rate": 2.3149593726376903e-06, "loss": 0.4451, "step": 13515 }, { "epoch": 0.6836966968486013, "grad_norm": 13.80263614654541, "learning_rate": 2.31428234026611e-06, "loss": 0.4805, "step": 13516 }, { "epoch": 0.6837472810966665, "grad_norm": 6.095423221588135, "learning_rate": 2.313605377098985e-06, "loss": 0.3822, "step": 13517 }, { "epoch": 0.6837978653447316, "grad_norm": 5.507772445678711, "learning_rate": 2.3129284831537627e-06, "loss": 0.3262, "step": 13518 }, { "epoch": 0.6838484495927968, "grad_norm": 8.549355506896973, "learning_rate": 2.312251658447883e-06, "loss": 0.3878, "step": 13519 }, { "epoch": 0.6838990338408619, "grad_norm": 5.382275581359863, "learning_rate": 2.3115749029987842e-06, "loss": 0.3658, "step": 13520 }, { "epoch": 0.6839496180889271, "grad_norm": 5.85458517074585, "learning_rate": 2.3108982168239086e-06, "loss": 0.372, "step": 13521 }, { "epoch": 0.6840002023369922, "grad_norm": 12.869187355041504, "learning_rate": 2.3102215999406907e-06, "loss": 0.4352, "step": 13522 }, { "epoch": 0.6840507865850574, "grad_norm": 9.981929779052734, "learning_rate": 2.3095450523665645e-06, "loss": 0.388, "step": 13523 }, { "epoch": 0.6841013708331226, "grad_norm": 9.48525619506836, "learning_rate": 2.3088685741189647e-06, "loss": 0.428, "step": 13524 }, { "epoch": 0.6841519550811878, "grad_norm": 14.01860237121582, "learning_rate": 2.308192165215319e-06, "loss": 0.5199, "step": 13525 }, { "epoch": 0.6842025393292529, "grad_norm": 8.723113059997559, "learning_rate": 2.307515825673062e-06, "loss": 0.3706, "step": 13526 }, { "epoch": 0.684253123577318, "grad_norm": 6.599364757537842, "learning_rate": 2.3068395555096183e-06, "loss": 0.3316, "step": 13527 }, { "epoch": 0.6843037078253832, "grad_norm": 8.482781410217285, "learning_rate": 2.306163354742413e-06, "loss": 0.4528, "step": 13528 }, { "epoch": 0.6843542920734483, "grad_norm": 18.293834686279297, "learning_rate": 2.3054872233888725e-06, "loss": 0.4593, "step": 13529 }, { "epoch": 0.6844048763215135, "grad_norm": 5.683192253112793, "learning_rate": 2.3048111614664184e-06, "loss": 0.3506, "step": 13530 }, { "epoch": 0.6844554605695786, "grad_norm": 14.92914867401123, "learning_rate": 2.304135168992468e-06, "loss": 0.4139, "step": 13531 }, { "epoch": 0.6845060448176438, "grad_norm": 6.729215145111084, "learning_rate": 2.3034592459844453e-06, "loss": 0.3254, "step": 13532 }, { "epoch": 0.6845566290657089, "grad_norm": 5.0006866455078125, "learning_rate": 2.302783392459764e-06, "loss": 0.3365, "step": 13533 }, { "epoch": 0.684607213313774, "grad_norm": 13.116907119750977, "learning_rate": 2.302107608435841e-06, "loss": 0.5893, "step": 13534 }, { "epoch": 0.6846577975618392, "grad_norm": 8.69969367980957, "learning_rate": 2.301431893930087e-06, "loss": 0.3834, "step": 13535 }, { "epoch": 0.6847083818099043, "grad_norm": 7.691412925720215, "learning_rate": 2.3007562489599155e-06, "loss": 0.3253, "step": 13536 }, { "epoch": 0.6847589660579696, "grad_norm": 5.694036483764648, "learning_rate": 2.3000806735427337e-06, "loss": 0.3878, "step": 13537 }, { "epoch": 0.6848095503060347, "grad_norm": 10.7494535446167, "learning_rate": 2.2994051676959538e-06, "loss": 0.3881, "step": 13538 }, { "epoch": 0.6848601345540999, "grad_norm": 8.627336502075195, "learning_rate": 2.29872973143698e-06, "loss": 0.4049, "step": 13539 }, { "epoch": 0.684910718802165, "grad_norm": 9.650700569152832, "learning_rate": 2.298054364783214e-06, "loss": 0.4638, "step": 13540 }, { "epoch": 0.6849613030502302, "grad_norm": 5.845538139343262, "learning_rate": 2.297379067752063e-06, "loss": 0.3233, "step": 13541 }, { "epoch": 0.6850118872982953, "grad_norm": 7.478440284729004, "learning_rate": 2.296703840360925e-06, "loss": 0.3474, "step": 13542 }, { "epoch": 0.6850624715463605, "grad_norm": 5.066465377807617, "learning_rate": 2.2960286826271976e-06, "loss": 0.3815, "step": 13543 }, { "epoch": 0.6851130557944256, "grad_norm": 6.705945014953613, "learning_rate": 2.295353594568282e-06, "loss": 0.4374, "step": 13544 }, { "epoch": 0.6851636400424908, "grad_norm": 12.699857711791992, "learning_rate": 2.294678576201572e-06, "loss": 0.5419, "step": 13545 }, { "epoch": 0.6852142242905559, "grad_norm": 12.751909255981445, "learning_rate": 2.2940036275444576e-06, "loss": 0.4446, "step": 13546 }, { "epoch": 0.685264808538621, "grad_norm": 6.148068428039551, "learning_rate": 2.2933287486143363e-06, "loss": 0.3597, "step": 13547 }, { "epoch": 0.6853153927866862, "grad_norm": 9.197866439819336, "learning_rate": 2.2926539394285974e-06, "loss": 0.527, "step": 13548 }, { "epoch": 0.6853659770347513, "grad_norm": 8.598684310913086, "learning_rate": 2.2919792000046233e-06, "loss": 0.4623, "step": 13549 }, { "epoch": 0.6854165612828166, "grad_norm": 14.961509704589844, "learning_rate": 2.2913045303598057e-06, "loss": 0.491, "step": 13550 }, { "epoch": 0.6854671455308817, "grad_norm": 8.28061294555664, "learning_rate": 2.2906299305115284e-06, "loss": 0.3376, "step": 13551 }, { "epoch": 0.6855177297789469, "grad_norm": 5.287593841552734, "learning_rate": 2.2899554004771706e-06, "loss": 0.388, "step": 13552 }, { "epoch": 0.685568314027012, "grad_norm": 6.1539306640625, "learning_rate": 2.2892809402741195e-06, "loss": 0.3812, "step": 13553 }, { "epoch": 0.6856188982750772, "grad_norm": 25.547279357910156, "learning_rate": 2.2886065499197502e-06, "loss": 0.5786, "step": 13554 }, { "epoch": 0.6856694825231423, "grad_norm": 12.010347366333008, "learning_rate": 2.2879322294314393e-06, "loss": 0.3309, "step": 13555 }, { "epoch": 0.6857200667712074, "grad_norm": 6.406189441680908, "learning_rate": 2.2872579788265664e-06, "loss": 0.4092, "step": 13556 }, { "epoch": 0.6857706510192726, "grad_norm": 7.873954772949219, "learning_rate": 2.2865837981225027e-06, "loss": 0.4143, "step": 13557 }, { "epoch": 0.6858212352673377, "grad_norm": 15.909750938415527, "learning_rate": 2.2859096873366186e-06, "loss": 0.5052, "step": 13558 }, { "epoch": 0.6858718195154029, "grad_norm": 6.773464202880859, "learning_rate": 2.2852356464862885e-06, "loss": 0.2959, "step": 13559 }, { "epoch": 0.685922403763468, "grad_norm": 5.2056803703308105, "learning_rate": 2.284561675588878e-06, "loss": 0.2516, "step": 13560 }, { "epoch": 0.6859729880115332, "grad_norm": 5.959519386291504, "learning_rate": 2.283887774661755e-06, "loss": 0.3026, "step": 13561 }, { "epoch": 0.6860235722595983, "grad_norm": 7.055663108825684, "learning_rate": 2.2832139437222827e-06, "loss": 0.2749, "step": 13562 }, { "epoch": 0.6860741565076636, "grad_norm": 5.411048412322998, "learning_rate": 2.2825401827878237e-06, "loss": 0.4891, "step": 13563 }, { "epoch": 0.6861247407557287, "grad_norm": 5.783266067504883, "learning_rate": 2.281866491875742e-06, "loss": 0.3714, "step": 13564 }, { "epoch": 0.6861753250037939, "grad_norm": 6.982443809509277, "learning_rate": 2.2811928710033953e-06, "loss": 0.4023, "step": 13565 }, { "epoch": 0.686225909251859, "grad_norm": 6.453383445739746, "learning_rate": 2.28051932018814e-06, "loss": 0.3931, "step": 13566 }, { "epoch": 0.6862764934999241, "grad_norm": 10.719766616821289, "learning_rate": 2.2798458394473343e-06, "loss": 0.3643, "step": 13567 }, { "epoch": 0.6863270777479893, "grad_norm": 10.017621994018555, "learning_rate": 2.279172428798331e-06, "loss": 0.4628, "step": 13568 }, { "epoch": 0.6863776619960544, "grad_norm": 10.09748649597168, "learning_rate": 2.2784990882584814e-06, "loss": 0.3586, "step": 13569 }, { "epoch": 0.6864282462441196, "grad_norm": 9.00239372253418, "learning_rate": 2.2778258178451385e-06, "loss": 0.3491, "step": 13570 }, { "epoch": 0.6864788304921847, "grad_norm": 9.903965950012207, "learning_rate": 2.2771526175756486e-06, "loss": 0.4074, "step": 13571 }, { "epoch": 0.6865294147402499, "grad_norm": 9.397294044494629, "learning_rate": 2.276479487467359e-06, "loss": 0.4659, "step": 13572 }, { "epoch": 0.686579998988315, "grad_norm": 8.609749794006348, "learning_rate": 2.275806427537613e-06, "loss": 0.373, "step": 13573 }, { "epoch": 0.6866305832363802, "grad_norm": 6.284247398376465, "learning_rate": 2.2751334378037586e-06, "loss": 0.4812, "step": 13574 }, { "epoch": 0.6866811674844453, "grad_norm": 6.782068729400635, "learning_rate": 2.274460518283131e-06, "loss": 0.4647, "step": 13575 }, { "epoch": 0.6867317517325104, "grad_norm": 6.334916591644287, "learning_rate": 2.2737876689930736e-06, "loss": 0.2782, "step": 13576 }, { "epoch": 0.6867823359805757, "grad_norm": 6.776771068572998, "learning_rate": 2.2731148899509237e-06, "loss": 0.3868, "step": 13577 }, { "epoch": 0.6868329202286408, "grad_norm": 12.044880867004395, "learning_rate": 2.2724421811740145e-06, "loss": 0.4175, "step": 13578 }, { "epoch": 0.686883504476706, "grad_norm": 7.354417324066162, "learning_rate": 2.2717695426796836e-06, "loss": 0.3573, "step": 13579 }, { "epoch": 0.6869340887247711, "grad_norm": 6.707510948181152, "learning_rate": 2.2710969744852618e-06, "loss": 0.4345, "step": 13580 }, { "epoch": 0.6869846729728363, "grad_norm": 13.068819999694824, "learning_rate": 2.2704244766080783e-06, "loss": 0.4348, "step": 13581 }, { "epoch": 0.6870352572209014, "grad_norm": 11.647561073303223, "learning_rate": 2.2697520490654645e-06, "loss": 0.4222, "step": 13582 }, { "epoch": 0.6870858414689666, "grad_norm": 8.863986015319824, "learning_rate": 2.269079691874746e-06, "loss": 0.4455, "step": 13583 }, { "epoch": 0.6871364257170317, "grad_norm": 8.372447967529297, "learning_rate": 2.2684074050532456e-06, "loss": 0.4133, "step": 13584 }, { "epoch": 0.6871870099650969, "grad_norm": 7.673799991607666, "learning_rate": 2.26773518861829e-06, "loss": 0.4278, "step": 13585 }, { "epoch": 0.687237594213162, "grad_norm": 6.991470813751221, "learning_rate": 2.2670630425872e-06, "loss": 0.4206, "step": 13586 }, { "epoch": 0.6872881784612271, "grad_norm": 5.21237850189209, "learning_rate": 2.266390966977293e-06, "loss": 0.4188, "step": 13587 }, { "epoch": 0.6873387627092923, "grad_norm": 10.715128898620605, "learning_rate": 2.265718961805889e-06, "loss": 0.3434, "step": 13588 }, { "epoch": 0.6873893469573574, "grad_norm": 5.194162845611572, "learning_rate": 2.2650470270903025e-06, "loss": 0.4342, "step": 13589 }, { "epoch": 0.6874399312054227, "grad_norm": 8.476462364196777, "learning_rate": 2.2643751628478467e-06, "loss": 0.4648, "step": 13590 }, { "epoch": 0.6874905154534878, "grad_norm": 6.340691566467285, "learning_rate": 2.2637033690958366e-06, "loss": 0.3716, "step": 13591 }, { "epoch": 0.687541099701553, "grad_norm": 7.386377811431885, "learning_rate": 2.2630316458515816e-06, "loss": 0.2997, "step": 13592 }, { "epoch": 0.6875916839496181, "grad_norm": 5.986424922943115, "learning_rate": 2.2623599931323885e-06, "loss": 0.2609, "step": 13593 }, { "epoch": 0.6876422681976833, "grad_norm": 5.31524133682251, "learning_rate": 2.2616884109555675e-06, "loss": 0.379, "step": 13594 }, { "epoch": 0.6876928524457484, "grad_norm": 9.030913352966309, "learning_rate": 2.2610168993384223e-06, "loss": 0.3373, "step": 13595 }, { "epoch": 0.6877434366938135, "grad_norm": 5.730037212371826, "learning_rate": 2.2603454582982544e-06, "loss": 0.3528, "step": 13596 }, { "epoch": 0.6877940209418787, "grad_norm": 15.061030387878418, "learning_rate": 2.259674087852368e-06, "loss": 0.4044, "step": 13597 }, { "epoch": 0.6878446051899438, "grad_norm": 11.869750022888184, "learning_rate": 2.259002788018061e-06, "loss": 0.4307, "step": 13598 }, { "epoch": 0.687895189438009, "grad_norm": 12.51315975189209, "learning_rate": 2.2583315588126302e-06, "loss": 0.5183, "step": 13599 }, { "epoch": 0.6879457736860741, "grad_norm": 10.246429443359375, "learning_rate": 2.2576604002533763e-06, "loss": 0.429, "step": 13600 }, { "epoch": 0.6879963579341393, "grad_norm": 7.902531147003174, "learning_rate": 2.2569893123575863e-06, "loss": 0.3695, "step": 13601 }, { "epoch": 0.6880469421822044, "grad_norm": 7.734940052032471, "learning_rate": 2.256318295142559e-06, "loss": 0.4228, "step": 13602 }, { "epoch": 0.6880975264302697, "grad_norm": 10.891798973083496, "learning_rate": 2.2556473486255816e-06, "loss": 0.4511, "step": 13603 }, { "epoch": 0.6881481106783348, "grad_norm": 6.0847859382629395, "learning_rate": 2.254976472823941e-06, "loss": 0.3252, "step": 13604 }, { "epoch": 0.6881986949264, "grad_norm": 9.592352867126465, "learning_rate": 2.2543056677549284e-06, "loss": 0.4759, "step": 13605 }, { "epoch": 0.6882492791744651, "grad_norm": 4.8629608154296875, "learning_rate": 2.253634933435827e-06, "loss": 0.2598, "step": 13606 }, { "epoch": 0.6882998634225302, "grad_norm": 9.837786674499512, "learning_rate": 2.25296426988392e-06, "loss": 0.3508, "step": 13607 }, { "epoch": 0.6883504476705954, "grad_norm": 10.638543128967285, "learning_rate": 2.252293677116486e-06, "loss": 0.3702, "step": 13608 }, { "epoch": 0.6884010319186605, "grad_norm": 9.532081604003906, "learning_rate": 2.2516231551508096e-06, "loss": 0.4981, "step": 13609 }, { "epoch": 0.6884516161667257, "grad_norm": 7.2854180335998535, "learning_rate": 2.250952704004166e-06, "loss": 0.3964, "step": 13610 }, { "epoch": 0.6885022004147908, "grad_norm": 6.898853778839111, "learning_rate": 2.250282323693829e-06, "loss": 0.3788, "step": 13611 }, { "epoch": 0.688552784662856, "grad_norm": 8.598671913146973, "learning_rate": 2.249612014237077e-06, "loss": 0.4797, "step": 13612 }, { "epoch": 0.6886033689109211, "grad_norm": 11.183305740356445, "learning_rate": 2.24894177565118e-06, "loss": 0.3746, "step": 13613 }, { "epoch": 0.6886539531589863, "grad_norm": 6.445113658905029, "learning_rate": 2.248271607953408e-06, "loss": 0.4066, "step": 13614 }, { "epoch": 0.6887045374070514, "grad_norm": 12.069202423095703, "learning_rate": 2.247601511161031e-06, "loss": 0.348, "step": 13615 }, { "epoch": 0.6887551216551165, "grad_norm": 6.676433563232422, "learning_rate": 2.246931485291313e-06, "loss": 0.3956, "step": 13616 }, { "epoch": 0.6888057059031818, "grad_norm": 4.48410177230835, "learning_rate": 2.2462615303615225e-06, "loss": 0.3265, "step": 13617 }, { "epoch": 0.6888562901512469, "grad_norm": 5.5395283699035645, "learning_rate": 2.245591646388921e-06, "loss": 0.4738, "step": 13618 }, { "epoch": 0.6889068743993121, "grad_norm": 8.7067289352417, "learning_rate": 2.244921833390768e-06, "loss": 0.4581, "step": 13619 }, { "epoch": 0.6889574586473772, "grad_norm": 10.197821617126465, "learning_rate": 2.244252091384326e-06, "loss": 0.3694, "step": 13620 }, { "epoch": 0.6890080428954424, "grad_norm": 8.39889907836914, "learning_rate": 2.2435824203868516e-06, "loss": 0.3702, "step": 13621 }, { "epoch": 0.6890586271435075, "grad_norm": 15.779508590698242, "learning_rate": 2.2429128204155993e-06, "loss": 0.5075, "step": 13622 }, { "epoch": 0.6891092113915727, "grad_norm": 7.169639587402344, "learning_rate": 2.2422432914878252e-06, "loss": 0.3369, "step": 13623 }, { "epoch": 0.6891597956396378, "grad_norm": 6.011453628540039, "learning_rate": 2.2415738336207804e-06, "loss": 0.3245, "step": 13624 }, { "epoch": 0.689210379887703, "grad_norm": 12.62270736694336, "learning_rate": 2.2409044468317155e-06, "loss": 0.5924, "step": 13625 }, { "epoch": 0.6892609641357681, "grad_norm": 7.951951026916504, "learning_rate": 2.240235131137879e-06, "loss": 0.3612, "step": 13626 }, { "epoch": 0.6893115483838332, "grad_norm": 13.159354209899902, "learning_rate": 2.2395658865565166e-06, "loss": 0.4536, "step": 13627 }, { "epoch": 0.6893621326318984, "grad_norm": 6.75895357131958, "learning_rate": 2.238896713104872e-06, "loss": 0.4065, "step": 13628 }, { "epoch": 0.6894127168799635, "grad_norm": 3.7928669452667236, "learning_rate": 2.238227610800192e-06, "loss": 0.2786, "step": 13629 }, { "epoch": 0.6894633011280288, "grad_norm": 9.658931732177734, "learning_rate": 2.2375585796597156e-06, "loss": 0.454, "step": 13630 }, { "epoch": 0.6895138853760939, "grad_norm": 6.14383602142334, "learning_rate": 2.23688961970068e-06, "loss": 0.3398, "step": 13631 }, { "epoch": 0.6895644696241591, "grad_norm": 7.515908241271973, "learning_rate": 2.236220730940327e-06, "loss": 0.4165, "step": 13632 }, { "epoch": 0.6896150538722242, "grad_norm": 11.90483283996582, "learning_rate": 2.2355519133958907e-06, "loss": 0.5223, "step": 13633 }, { "epoch": 0.6896656381202894, "grad_norm": 8.501395225524902, "learning_rate": 2.2348831670846017e-06, "loss": 0.4234, "step": 13634 }, { "epoch": 0.6897162223683545, "grad_norm": 8.276511192321777, "learning_rate": 2.2342144920236968e-06, "loss": 0.4263, "step": 13635 }, { "epoch": 0.6897668066164196, "grad_norm": 11.374330520629883, "learning_rate": 2.233545888230404e-06, "loss": 0.4985, "step": 13636 }, { "epoch": 0.6898173908644848, "grad_norm": 6.0207085609436035, "learning_rate": 2.2328773557219494e-06, "loss": 0.4786, "step": 13637 }, { "epoch": 0.6898679751125499, "grad_norm": 8.657999038696289, "learning_rate": 2.2322088945155655e-06, "loss": 0.3971, "step": 13638 }, { "epoch": 0.6899185593606151, "grad_norm": 6.063833713531494, "learning_rate": 2.2315405046284698e-06, "loss": 0.4262, "step": 13639 }, { "epoch": 0.6899691436086802, "grad_norm": 9.450815200805664, "learning_rate": 2.23087218607789e-06, "loss": 0.4591, "step": 13640 }, { "epoch": 0.6900197278567454, "grad_norm": 9.270583152770996, "learning_rate": 2.2302039388810446e-06, "loss": 0.4641, "step": 13641 }, { "epoch": 0.6900703121048105, "grad_norm": 4.131044864654541, "learning_rate": 2.2295357630551527e-06, "loss": 0.3126, "step": 13642 }, { "epoch": 0.6901208963528758, "grad_norm": 9.516491889953613, "learning_rate": 2.2288676586174342e-06, "loss": 0.5126, "step": 13643 }, { "epoch": 0.6901714806009409, "grad_norm": 9.862332344055176, "learning_rate": 2.2281996255851023e-06, "loss": 0.3881, "step": 13644 }, { "epoch": 0.6902220648490061, "grad_norm": 13.889995574951172, "learning_rate": 2.2275316639753708e-06, "loss": 0.4517, "step": 13645 }, { "epoch": 0.6902726490970712, "grad_norm": 8.285566329956055, "learning_rate": 2.2268637738054504e-06, "loss": 0.3717, "step": 13646 }, { "epoch": 0.6903232333451363, "grad_norm": 9.852684020996094, "learning_rate": 2.226195955092554e-06, "loss": 0.4221, "step": 13647 }, { "epoch": 0.6903738175932015, "grad_norm": 7.097771167755127, "learning_rate": 2.225528207853888e-06, "loss": 0.4304, "step": 13648 }, { "epoch": 0.6904244018412666, "grad_norm": 5.508431911468506, "learning_rate": 2.2248605321066568e-06, "loss": 0.3848, "step": 13649 }, { "epoch": 0.6904749860893318, "grad_norm": 5.880466938018799, "learning_rate": 2.224192927868068e-06, "loss": 0.3633, "step": 13650 }, { "epoch": 0.6905255703373969, "grad_norm": 9.245047569274902, "learning_rate": 2.2235253951553233e-06, "loss": 0.3916, "step": 13651 }, { "epoch": 0.6905761545854621, "grad_norm": 13.436973571777344, "learning_rate": 2.2228579339856225e-06, "loss": 0.3829, "step": 13652 }, { "epoch": 0.6906267388335272, "grad_norm": 6.404119968414307, "learning_rate": 2.2221905443761653e-06, "loss": 0.3553, "step": 13653 }, { "epoch": 0.6906773230815924, "grad_norm": 6.4631667137146, "learning_rate": 2.221523226344146e-06, "loss": 0.3939, "step": 13654 }, { "epoch": 0.6907279073296575, "grad_norm": 9.010368347167969, "learning_rate": 2.2208559799067643e-06, "loss": 0.4547, "step": 13655 }, { "epoch": 0.6907784915777226, "grad_norm": 7.7649712562561035, "learning_rate": 2.220188805081211e-06, "loss": 0.4302, "step": 13656 }, { "epoch": 0.6908290758257879, "grad_norm": 5.615181922912598, "learning_rate": 2.219521701884676e-06, "loss": 0.3733, "step": 13657 }, { "epoch": 0.690879660073853, "grad_norm": 8.606022834777832, "learning_rate": 2.218854670334352e-06, "loss": 0.3938, "step": 13658 }, { "epoch": 0.6909302443219182, "grad_norm": 8.348398208618164, "learning_rate": 2.2181877104474264e-06, "loss": 0.4307, "step": 13659 }, { "epoch": 0.6909808285699833, "grad_norm": 20.652061462402344, "learning_rate": 2.217520822241082e-06, "loss": 0.4108, "step": 13660 }, { "epoch": 0.6910314128180485, "grad_norm": 10.056336402893066, "learning_rate": 2.216854005732507e-06, "loss": 0.4319, "step": 13661 }, { "epoch": 0.6910819970661136, "grad_norm": 8.993515014648438, "learning_rate": 2.2161872609388822e-06, "loss": 0.5151, "step": 13662 }, { "epoch": 0.6911325813141788, "grad_norm": 11.71236515045166, "learning_rate": 2.215520587877386e-06, "loss": 0.3673, "step": 13663 }, { "epoch": 0.6911831655622439, "grad_norm": 7.486725807189941, "learning_rate": 2.2148539865652017e-06, "loss": 0.4666, "step": 13664 }, { "epoch": 0.691233749810309, "grad_norm": 5.67548131942749, "learning_rate": 2.2141874570195017e-06, "loss": 0.3435, "step": 13665 }, { "epoch": 0.6912843340583742, "grad_norm": 8.310525894165039, "learning_rate": 2.21352099925746e-06, "loss": 0.435, "step": 13666 }, { "epoch": 0.6913349183064393, "grad_norm": 12.815008163452148, "learning_rate": 2.2128546132962536e-06, "loss": 0.4841, "step": 13667 }, { "epoch": 0.6913855025545045, "grad_norm": 10.153970718383789, "learning_rate": 2.2121882991530514e-06, "loss": 0.4111, "step": 13668 }, { "epoch": 0.6914360868025696, "grad_norm": 12.838075637817383, "learning_rate": 2.211522056845021e-06, "loss": 0.4764, "step": 13669 }, { "epoch": 0.6914866710506349, "grad_norm": 6.044487476348877, "learning_rate": 2.2108558863893346e-06, "loss": 0.3101, "step": 13670 }, { "epoch": 0.6915372552987, "grad_norm": 6.190980434417725, "learning_rate": 2.210189787803155e-06, "loss": 0.3825, "step": 13671 }, { "epoch": 0.6915878395467652, "grad_norm": 10.147027015686035, "learning_rate": 2.2095237611036437e-06, "loss": 0.3928, "step": 13672 }, { "epoch": 0.6916384237948303, "grad_norm": 9.16355037689209, "learning_rate": 2.208857806307967e-06, "loss": 0.3857, "step": 13673 }, { "epoch": 0.6916890080428955, "grad_norm": 7.22718620300293, "learning_rate": 2.2081919234332827e-06, "loss": 0.4132, "step": 13674 }, { "epoch": 0.6917395922909606, "grad_norm": 6.632142066955566, "learning_rate": 2.207526112496748e-06, "loss": 0.3413, "step": 13675 }, { "epoch": 0.6917901765390257, "grad_norm": 14.100526809692383, "learning_rate": 2.2068603735155213e-06, "loss": 0.5259, "step": 13676 }, { "epoch": 0.6918407607870909, "grad_norm": 5.394461154937744, "learning_rate": 2.2061947065067567e-06, "loss": 0.3077, "step": 13677 }, { "epoch": 0.691891345035156, "grad_norm": 3.7760252952575684, "learning_rate": 2.2055291114876065e-06, "loss": 0.303, "step": 13678 }, { "epoch": 0.6919419292832212, "grad_norm": 9.943753242492676, "learning_rate": 2.2048635884752214e-06, "loss": 0.4074, "step": 13679 }, { "epoch": 0.6919925135312863, "grad_norm": 10.368643760681152, "learning_rate": 2.20419813748675e-06, "loss": 0.376, "step": 13680 }, { "epoch": 0.6920430977793515, "grad_norm": 9.148361206054688, "learning_rate": 2.203532758539338e-06, "loss": 0.3919, "step": 13681 }, { "epoch": 0.6920936820274166, "grad_norm": 8.059605598449707, "learning_rate": 2.2028674516501337e-06, "loss": 0.4173, "step": 13682 }, { "epoch": 0.6921442662754819, "grad_norm": 8.325124740600586, "learning_rate": 2.2022022168362793e-06, "loss": 0.4826, "step": 13683 }, { "epoch": 0.692194850523547, "grad_norm": 11.451325416564941, "learning_rate": 2.201537054114914e-06, "loss": 0.4871, "step": 13684 }, { "epoch": 0.6922454347716122, "grad_norm": 12.469846725463867, "learning_rate": 2.200871963503181e-06, "loss": 0.4186, "step": 13685 }, { "epoch": 0.6922960190196773, "grad_norm": 14.505462646484375, "learning_rate": 2.200206945018217e-06, "loss": 0.3908, "step": 13686 }, { "epoch": 0.6923466032677424, "grad_norm": 5.230672359466553, "learning_rate": 2.1995419986771554e-06, "loss": 0.3437, "step": 13687 }, { "epoch": 0.6923971875158076, "grad_norm": 10.73385238647461, "learning_rate": 2.1988771244971336e-06, "loss": 0.4198, "step": 13688 }, { "epoch": 0.6924477717638727, "grad_norm": 7.243312835693359, "learning_rate": 2.198212322495283e-06, "loss": 0.3609, "step": 13689 }, { "epoch": 0.6924983560119379, "grad_norm": 7.0316033363342285, "learning_rate": 2.1975475926887335e-06, "loss": 0.4817, "step": 13690 }, { "epoch": 0.692548940260003, "grad_norm": 6.1422929763793945, "learning_rate": 2.196882935094613e-06, "loss": 0.3454, "step": 13691 }, { "epoch": 0.6925995245080682, "grad_norm": 6.531137943267822, "learning_rate": 2.196218349730047e-06, "loss": 0.3033, "step": 13692 }, { "epoch": 0.6926501087561333, "grad_norm": 11.724799156188965, "learning_rate": 2.195553836612164e-06, "loss": 0.3846, "step": 13693 }, { "epoch": 0.6927006930041985, "grad_norm": 7.122897624969482, "learning_rate": 2.194889395758085e-06, "loss": 0.3924, "step": 13694 }, { "epoch": 0.6927512772522636, "grad_norm": 7.396969795227051, "learning_rate": 2.194225027184928e-06, "loss": 0.4117, "step": 13695 }, { "epoch": 0.6928018615003289, "grad_norm": 11.135844230651855, "learning_rate": 2.1935607309098173e-06, "loss": 0.4618, "step": 13696 }, { "epoch": 0.692852445748394, "grad_norm": 9.28914737701416, "learning_rate": 2.1928965069498674e-06, "loss": 0.4854, "step": 13697 }, { "epoch": 0.6929030299964591, "grad_norm": 13.925986289978027, "learning_rate": 2.1922323553221932e-06, "loss": 0.5643, "step": 13698 }, { "epoch": 0.6929536142445243, "grad_norm": 13.209060668945312, "learning_rate": 2.191568276043911e-06, "loss": 0.6329, "step": 13699 }, { "epoch": 0.6930041984925894, "grad_norm": 14.359769821166992, "learning_rate": 2.190904269132131e-06, "loss": 0.4622, "step": 13700 }, { "epoch": 0.6930547827406546, "grad_norm": 7.722962856292725, "learning_rate": 2.19024033460396e-06, "loss": 0.3657, "step": 13701 }, { "epoch": 0.6931053669887197, "grad_norm": 10.471123695373535, "learning_rate": 2.189576472476512e-06, "loss": 0.4484, "step": 13702 }, { "epoch": 0.6931559512367849, "grad_norm": 6.632311820983887, "learning_rate": 2.188912682766891e-06, "loss": 0.3483, "step": 13703 }, { "epoch": 0.69320653548485, "grad_norm": 10.260157585144043, "learning_rate": 2.1882489654921966e-06, "loss": 0.5221, "step": 13704 }, { "epoch": 0.6932571197329152, "grad_norm": 8.226298332214355, "learning_rate": 2.1875853206695373e-06, "loss": 0.4779, "step": 13705 }, { "epoch": 0.6933077039809803, "grad_norm": 10.044633865356445, "learning_rate": 2.186921748316011e-06, "loss": 0.3721, "step": 13706 }, { "epoch": 0.6933582882290454, "grad_norm": 13.489420890808105, "learning_rate": 2.1862582484487143e-06, "loss": 0.3708, "step": 13707 }, { "epoch": 0.6934088724771106, "grad_norm": 6.511799335479736, "learning_rate": 2.1855948210847482e-06, "loss": 0.3962, "step": 13708 }, { "epoch": 0.6934594567251757, "grad_norm": 6.7151312828063965, "learning_rate": 2.184931466241206e-06, "loss": 0.3909, "step": 13709 }, { "epoch": 0.693510040973241, "grad_norm": 13.815140724182129, "learning_rate": 2.184268183935178e-06, "loss": 0.3869, "step": 13710 }, { "epoch": 0.6935606252213061, "grad_norm": 6.815781593322754, "learning_rate": 2.18360497418376e-06, "loss": 0.3892, "step": 13711 }, { "epoch": 0.6936112094693713, "grad_norm": 4.735774993896484, "learning_rate": 2.1829418370040394e-06, "loss": 0.3455, "step": 13712 }, { "epoch": 0.6936617937174364, "grad_norm": 8.711958885192871, "learning_rate": 2.1822787724131013e-06, "loss": 0.481, "step": 13713 }, { "epoch": 0.6937123779655016, "grad_norm": 11.712064743041992, "learning_rate": 2.1816157804280346e-06, "loss": 0.4544, "step": 13714 }, { "epoch": 0.6937629622135667, "grad_norm": 7.31576681137085, "learning_rate": 2.180952861065922e-06, "loss": 0.4155, "step": 13715 }, { "epoch": 0.6938135464616318, "grad_norm": 5.3242621421813965, "learning_rate": 2.180290014343845e-06, "loss": 0.3033, "step": 13716 }, { "epoch": 0.693864130709697, "grad_norm": 5.673706531524658, "learning_rate": 2.179627240278884e-06, "loss": 0.37, "step": 13717 }, { "epoch": 0.6939147149577621, "grad_norm": 7.680704116821289, "learning_rate": 2.1789645388881163e-06, "loss": 0.3662, "step": 13718 }, { "epoch": 0.6939652992058273, "grad_norm": 5.17132043838501, "learning_rate": 2.1783019101886168e-06, "loss": 0.4003, "step": 13719 }, { "epoch": 0.6940158834538924, "grad_norm": 8.817790031433105, "learning_rate": 2.1776393541974632e-06, "loss": 0.3814, "step": 13720 }, { "epoch": 0.6940664677019576, "grad_norm": 6.047043800354004, "learning_rate": 2.1769768709317263e-06, "loss": 0.429, "step": 13721 }, { "epoch": 0.6941170519500227, "grad_norm": 8.07827377319336, "learning_rate": 2.176314460408475e-06, "loss": 0.4192, "step": 13722 }, { "epoch": 0.694167636198088, "grad_norm": 11.298940658569336, "learning_rate": 2.1756521226447818e-06, "loss": 0.4466, "step": 13723 }, { "epoch": 0.6942182204461531, "grad_norm": 6.883050918579102, "learning_rate": 2.1749898576577112e-06, "loss": 0.4322, "step": 13724 }, { "epoch": 0.6942688046942183, "grad_norm": 6.80880069732666, "learning_rate": 2.174327665464327e-06, "loss": 0.3451, "step": 13725 }, { "epoch": 0.6943193889422834, "grad_norm": 9.458560943603516, "learning_rate": 2.173665546081695e-06, "loss": 0.4869, "step": 13726 }, { "epoch": 0.6943699731903485, "grad_norm": 6.07839822769165, "learning_rate": 2.173003499526875e-06, "loss": 0.3645, "step": 13727 }, { "epoch": 0.6944205574384137, "grad_norm": 5.885734558105469, "learning_rate": 2.1723415258169274e-06, "loss": 0.3518, "step": 13728 }, { "epoch": 0.6944711416864788, "grad_norm": 10.26196575164795, "learning_rate": 2.1716796249689075e-06, "loss": 0.3892, "step": 13729 }, { "epoch": 0.694521725934544, "grad_norm": 15.709117889404297, "learning_rate": 2.1710177969998714e-06, "loss": 0.4526, "step": 13730 }, { "epoch": 0.6945723101826091, "grad_norm": 8.908862113952637, "learning_rate": 2.170356041926875e-06, "loss": 0.3601, "step": 13731 }, { "epoch": 0.6946228944306743, "grad_norm": 10.570597648620605, "learning_rate": 2.1696943597669686e-06, "loss": 0.4006, "step": 13732 }, { "epoch": 0.6946734786787394, "grad_norm": 6.287733554840088, "learning_rate": 2.1690327505372005e-06, "loss": 0.335, "step": 13733 }, { "epoch": 0.6947240629268046, "grad_norm": 5.378509044647217, "learning_rate": 2.168371214254622e-06, "loss": 0.3065, "step": 13734 }, { "epoch": 0.6947746471748697, "grad_norm": 10.046467781066895, "learning_rate": 2.167709750936278e-06, "loss": 0.3394, "step": 13735 }, { "epoch": 0.694825231422935, "grad_norm": 8.74075698852539, "learning_rate": 2.167048360599211e-06, "loss": 0.4465, "step": 13736 }, { "epoch": 0.6948758156710001, "grad_norm": 5.2587971687316895, "learning_rate": 2.1663870432604667e-06, "loss": 0.3044, "step": 13737 }, { "epoch": 0.6949263999190652, "grad_norm": 6.291714191436768, "learning_rate": 2.165725798937084e-06, "loss": 0.3437, "step": 13738 }, { "epoch": 0.6949769841671304, "grad_norm": 9.924784660339355, "learning_rate": 2.1650646276460997e-06, "loss": 0.418, "step": 13739 }, { "epoch": 0.6950275684151955, "grad_norm": 12.085264205932617, "learning_rate": 2.164403529404554e-06, "loss": 0.5318, "step": 13740 }, { "epoch": 0.6950781526632607, "grad_norm": 8.552431106567383, "learning_rate": 2.1637425042294826e-06, "loss": 0.458, "step": 13741 }, { "epoch": 0.6951287369113258, "grad_norm": 7.433813095092773, "learning_rate": 2.1630815521379123e-06, "loss": 0.3502, "step": 13742 }, { "epoch": 0.695179321159391, "grad_norm": 6.344608783721924, "learning_rate": 2.16242067314688e-06, "loss": 0.4317, "step": 13743 }, { "epoch": 0.6952299054074561, "grad_norm": 7.588046073913574, "learning_rate": 2.1617598672734125e-06, "loss": 0.5573, "step": 13744 }, { "epoch": 0.6952804896555212, "grad_norm": 12.671282768249512, "learning_rate": 2.1610991345345366e-06, "loss": 0.4643, "step": 13745 }, { "epoch": 0.6953310739035864, "grad_norm": 9.869109153747559, "learning_rate": 2.1604384749472807e-06, "loss": 0.3794, "step": 13746 }, { "epoch": 0.6953816581516515, "grad_norm": 7.123775482177734, "learning_rate": 2.159777888528666e-06, "loss": 0.2893, "step": 13747 }, { "epoch": 0.6954322423997167, "grad_norm": 6.517265319824219, "learning_rate": 2.1591173752957134e-06, "loss": 0.3704, "step": 13748 }, { "epoch": 0.6954828266477818, "grad_norm": 9.518499374389648, "learning_rate": 2.158456935265446e-06, "loss": 0.3875, "step": 13749 }, { "epoch": 0.6955334108958471, "grad_norm": 5.202544212341309, "learning_rate": 2.15779656845488e-06, "loss": 0.3592, "step": 13750 }, { "epoch": 0.6955839951439122, "grad_norm": 7.15998649597168, "learning_rate": 2.15713627488103e-06, "loss": 0.4252, "step": 13751 }, { "epoch": 0.6956345793919774, "grad_norm": 6.436666965484619, "learning_rate": 2.1564760545609126e-06, "loss": 0.3743, "step": 13752 }, { "epoch": 0.6956851636400425, "grad_norm": 9.401976585388184, "learning_rate": 2.1558159075115393e-06, "loss": 0.4727, "step": 13753 }, { "epoch": 0.6957357478881077, "grad_norm": 8.667255401611328, "learning_rate": 2.15515583374992e-06, "loss": 0.4201, "step": 13754 }, { "epoch": 0.6957863321361728, "grad_norm": 21.229412078857422, "learning_rate": 2.154495833293064e-06, "loss": 0.4209, "step": 13755 }, { "epoch": 0.695836916384238, "grad_norm": 8.898835182189941, "learning_rate": 2.1538359061579768e-06, "loss": 0.3706, "step": 13756 }, { "epoch": 0.6958875006323031, "grad_norm": 11.339818000793457, "learning_rate": 2.153176052361662e-06, "loss": 0.4297, "step": 13757 }, { "epoch": 0.6959380848803682, "grad_norm": 11.407245635986328, "learning_rate": 2.152516271921126e-06, "loss": 0.535, "step": 13758 }, { "epoch": 0.6959886691284334, "grad_norm": 9.9042387008667, "learning_rate": 2.1518565648533676e-06, "loss": 0.42, "step": 13759 }, { "epoch": 0.6960392533764985, "grad_norm": 8.88602352142334, "learning_rate": 2.1511969311753843e-06, "loss": 0.4116, "step": 13760 }, { "epoch": 0.6960898376245637, "grad_norm": 7.348939895629883, "learning_rate": 2.150537370904177e-06, "loss": 0.2758, "step": 13761 }, { "epoch": 0.6961404218726288, "grad_norm": 11.416491508483887, "learning_rate": 2.149877884056739e-06, "loss": 0.4098, "step": 13762 }, { "epoch": 0.6961910061206941, "grad_norm": 8.207162857055664, "learning_rate": 2.149218470650062e-06, "loss": 0.2441, "step": 13763 }, { "epoch": 0.6962415903687592, "grad_norm": 8.271225929260254, "learning_rate": 2.1485591307011415e-06, "loss": 0.3572, "step": 13764 }, { "epoch": 0.6962921746168244, "grad_norm": 5.580825328826904, "learning_rate": 2.1478998642269643e-06, "loss": 0.3553, "step": 13765 }, { "epoch": 0.6963427588648895, "grad_norm": 7.694788455963135, "learning_rate": 2.147240671244517e-06, "loss": 0.4187, "step": 13766 }, { "epoch": 0.6963933431129546, "grad_norm": 11.698039054870605, "learning_rate": 2.1465815517707907e-06, "loss": 0.4327, "step": 13767 }, { "epoch": 0.6964439273610198, "grad_norm": 7.4096598625183105, "learning_rate": 2.145922505822762e-06, "loss": 0.4347, "step": 13768 }, { "epoch": 0.6964945116090849, "grad_norm": 9.993677139282227, "learning_rate": 2.145263533417418e-06, "loss": 0.4607, "step": 13769 }, { "epoch": 0.6965450958571501, "grad_norm": 5.608082294464111, "learning_rate": 2.144604634571738e-06, "loss": 0.345, "step": 13770 }, { "epoch": 0.6965956801052152, "grad_norm": 6.62237024307251, "learning_rate": 2.1439458093026978e-06, "loss": 0.4492, "step": 13771 }, { "epoch": 0.6966462643532804, "grad_norm": 5.176621913909912, "learning_rate": 2.143287057627277e-06, "loss": 0.4183, "step": 13772 }, { "epoch": 0.6966968486013455, "grad_norm": 10.472588539123535, "learning_rate": 2.1426283795624487e-06, "loss": 0.4331, "step": 13773 }, { "epoch": 0.6967474328494107, "grad_norm": 10.423406600952148, "learning_rate": 2.141969775125184e-06, "loss": 0.3533, "step": 13774 }, { "epoch": 0.6967980170974758, "grad_norm": 7.647872447967529, "learning_rate": 2.1413112443324565e-06, "loss": 0.3629, "step": 13775 }, { "epoch": 0.696848601345541, "grad_norm": 6.4190826416015625, "learning_rate": 2.1406527872012326e-06, "loss": 0.3666, "step": 13776 }, { "epoch": 0.6968991855936062, "grad_norm": 7.246828079223633, "learning_rate": 2.139994403748479e-06, "loss": 0.3339, "step": 13777 }, { "epoch": 0.6969497698416713, "grad_norm": 7.907158374786377, "learning_rate": 2.1393360939911628e-06, "loss": 0.4118, "step": 13778 }, { "epoch": 0.6970003540897365, "grad_norm": 9.589282035827637, "learning_rate": 2.1386778579462463e-06, "loss": 0.3646, "step": 13779 }, { "epoch": 0.6970509383378016, "grad_norm": 13.626745223999023, "learning_rate": 2.13801969563069e-06, "loss": 0.4506, "step": 13780 }, { "epoch": 0.6971015225858668, "grad_norm": 6.846450328826904, "learning_rate": 2.1373616070614533e-06, "loss": 0.4277, "step": 13781 }, { "epoch": 0.6971521068339319, "grad_norm": 10.752411842346191, "learning_rate": 2.1367035922554936e-06, "loss": 0.5419, "step": 13782 }, { "epoch": 0.6972026910819971, "grad_norm": 11.61069393157959, "learning_rate": 2.1360456512297645e-06, "loss": 0.4418, "step": 13783 }, { "epoch": 0.6972532753300622, "grad_norm": 5.976126670837402, "learning_rate": 2.1353877840012228e-06, "loss": 0.3542, "step": 13784 }, { "epoch": 0.6973038595781273, "grad_norm": 12.012892723083496, "learning_rate": 2.134729990586819e-06, "loss": 0.3856, "step": 13785 }, { "epoch": 0.6973544438261925, "grad_norm": 9.293628692626953, "learning_rate": 2.134072271003501e-06, "loss": 0.3863, "step": 13786 }, { "epoch": 0.6974050280742576, "grad_norm": 8.236562728881836, "learning_rate": 2.1334146252682195e-06, "loss": 0.4628, "step": 13787 }, { "epoch": 0.6974556123223228, "grad_norm": 7.315888404846191, "learning_rate": 2.1327570533979197e-06, "loss": 0.3752, "step": 13788 }, { "epoch": 0.6975061965703879, "grad_norm": 8.506684303283691, "learning_rate": 2.132099555409543e-06, "loss": 0.4307, "step": 13789 }, { "epoch": 0.6975567808184532, "grad_norm": 8.301593780517578, "learning_rate": 2.131442131320036e-06, "loss": 0.4782, "step": 13790 }, { "epoch": 0.6976073650665183, "grad_norm": 5.661647796630859, "learning_rate": 2.1307847811463357e-06, "loss": 0.4004, "step": 13791 }, { "epoch": 0.6976579493145835, "grad_norm": 8.43317699432373, "learning_rate": 2.13012750490538e-06, "loss": 0.3935, "step": 13792 }, { "epoch": 0.6977085335626486, "grad_norm": 13.27114200592041, "learning_rate": 2.1294703026141096e-06, "loss": 0.5211, "step": 13793 }, { "epoch": 0.6977591178107138, "grad_norm": 13.255928993225098, "learning_rate": 2.128813174289455e-06, "loss": 0.3613, "step": 13794 }, { "epoch": 0.6978097020587789, "grad_norm": 9.605669975280762, "learning_rate": 2.1281561199483474e-06, "loss": 0.4037, "step": 13795 }, { "epoch": 0.697860286306844, "grad_norm": 10.31498908996582, "learning_rate": 2.1274991396077223e-06, "loss": 0.2733, "step": 13796 }, { "epoch": 0.6979108705549092, "grad_norm": 14.800557136535645, "learning_rate": 2.1268422332845056e-06, "loss": 0.4695, "step": 13797 }, { "epoch": 0.6979614548029743, "grad_norm": 12.192391395568848, "learning_rate": 2.126185400995623e-06, "loss": 0.3667, "step": 13798 }, { "epoch": 0.6980120390510395, "grad_norm": 5.931667804718018, "learning_rate": 2.1255286427580036e-06, "loss": 0.4021, "step": 13799 }, { "epoch": 0.6980626232991046, "grad_norm": 6.932436466217041, "learning_rate": 2.1248719585885675e-06, "loss": 0.324, "step": 13800 }, { "epoch": 0.6981132075471698, "grad_norm": 9.980682373046875, "learning_rate": 2.1242153485042346e-06, "loss": 0.3404, "step": 13801 }, { "epoch": 0.6981637917952349, "grad_norm": 8.72800350189209, "learning_rate": 2.123558812521928e-06, "loss": 0.3758, "step": 13802 }, { "epoch": 0.6982143760433002, "grad_norm": 6.327418804168701, "learning_rate": 2.122902350658563e-06, "loss": 0.4089, "step": 13803 }, { "epoch": 0.6982649602913653, "grad_norm": 6.191221237182617, "learning_rate": 2.1222459629310534e-06, "loss": 0.3092, "step": 13804 }, { "epoch": 0.6983155445394305, "grad_norm": 6.760879993438721, "learning_rate": 2.121589649356316e-06, "loss": 0.3978, "step": 13805 }, { "epoch": 0.6983661287874956, "grad_norm": 7.781081199645996, "learning_rate": 2.1209334099512606e-06, "loss": 0.4007, "step": 13806 }, { "epoch": 0.6984167130355607, "grad_norm": 10.150930404663086, "learning_rate": 2.120277244732798e-06, "loss": 0.562, "step": 13807 }, { "epoch": 0.6984672972836259, "grad_norm": 9.054265022277832, "learning_rate": 2.1196211537178346e-06, "loss": 0.4677, "step": 13808 }, { "epoch": 0.698517881531691, "grad_norm": 8.078764915466309, "learning_rate": 2.118965136923275e-06, "loss": 0.3806, "step": 13809 }, { "epoch": 0.6985684657797562, "grad_norm": 5.109955310821533, "learning_rate": 2.118309194366027e-06, "loss": 0.3255, "step": 13810 }, { "epoch": 0.6986190500278213, "grad_norm": 7.02401065826416, "learning_rate": 2.11765332606299e-06, "loss": 0.4176, "step": 13811 }, { "epoch": 0.6986696342758865, "grad_norm": 7.354607105255127, "learning_rate": 2.116997532031063e-06, "loss": 0.4141, "step": 13812 }, { "epoch": 0.6987202185239516, "grad_norm": 6.764859199523926, "learning_rate": 2.116341812287148e-06, "loss": 0.3737, "step": 13813 }, { "epoch": 0.6987708027720168, "grad_norm": 10.400176048278809, "learning_rate": 2.11568616684814e-06, "loss": 0.4069, "step": 13814 }, { "epoch": 0.6988213870200819, "grad_norm": 7.054020404815674, "learning_rate": 2.115030595730932e-06, "loss": 0.477, "step": 13815 }, { "epoch": 0.6988719712681472, "grad_norm": 21.855484008789062, "learning_rate": 2.1143750989524154e-06, "loss": 0.5058, "step": 13816 }, { "epoch": 0.6989225555162123, "grad_norm": 5.731629371643066, "learning_rate": 2.113719676529484e-06, "loss": 0.3861, "step": 13817 }, { "epoch": 0.6989731397642774, "grad_norm": 11.344635963439941, "learning_rate": 2.113064328479026e-06, "loss": 0.3189, "step": 13818 }, { "epoch": 0.6990237240123426, "grad_norm": 18.953372955322266, "learning_rate": 2.1124090548179267e-06, "loss": 0.4198, "step": 13819 }, { "epoch": 0.6990743082604077, "grad_norm": 11.501243591308594, "learning_rate": 2.1117538555630713e-06, "loss": 0.552, "step": 13820 }, { "epoch": 0.6991248925084729, "grad_norm": 9.599861145019531, "learning_rate": 2.111098730731341e-06, "loss": 0.5442, "step": 13821 }, { "epoch": 0.699175476756538, "grad_norm": 12.910414695739746, "learning_rate": 2.11044368033962e-06, "loss": 0.432, "step": 13822 }, { "epoch": 0.6992260610046032, "grad_norm": 5.489579200744629, "learning_rate": 2.1097887044047867e-06, "loss": 0.3895, "step": 13823 }, { "epoch": 0.6992766452526683, "grad_norm": 18.66731071472168, "learning_rate": 2.109133802943715e-06, "loss": 0.4207, "step": 13824 }, { "epoch": 0.6993272295007334, "grad_norm": 7.311601161956787, "learning_rate": 2.1084789759732844e-06, "loss": 0.4677, "step": 13825 }, { "epoch": 0.6993778137487986, "grad_norm": 6.339890003204346, "learning_rate": 2.107824223510367e-06, "loss": 0.4331, "step": 13826 }, { "epoch": 0.6994283979968637, "grad_norm": 8.162309646606445, "learning_rate": 2.107169545571831e-06, "loss": 0.2856, "step": 13827 }, { "epoch": 0.6994789822449289, "grad_norm": 11.33485221862793, "learning_rate": 2.1065149421745505e-06, "loss": 0.4029, "step": 13828 }, { "epoch": 0.6995295664929941, "grad_norm": 9.962603569030762, "learning_rate": 2.1058604133353915e-06, "loss": 0.4543, "step": 13829 }, { "epoch": 0.6995801507410593, "grad_norm": 6.3004021644592285, "learning_rate": 2.105205959071217e-06, "loss": 0.4205, "step": 13830 }, { "epoch": 0.6996307349891244, "grad_norm": 8.055712699890137, "learning_rate": 2.104551579398896e-06, "loss": 0.5046, "step": 13831 }, { "epoch": 0.6996813192371896, "grad_norm": 6.16830587387085, "learning_rate": 2.1038972743352865e-06, "loss": 0.3682, "step": 13832 }, { "epoch": 0.6997319034852547, "grad_norm": 15.46104621887207, "learning_rate": 2.1032430438972463e-06, "loss": 0.3944, "step": 13833 }, { "epoch": 0.6997824877333199, "grad_norm": 16.6039981842041, "learning_rate": 2.1025888881016384e-06, "loss": 0.5419, "step": 13834 }, { "epoch": 0.699833071981385, "grad_norm": 9.639996528625488, "learning_rate": 2.1019348069653156e-06, "loss": 0.4356, "step": 13835 }, { "epoch": 0.6998836562294501, "grad_norm": 7.036224842071533, "learning_rate": 2.1012808005051314e-06, "loss": 0.4599, "step": 13836 }, { "epoch": 0.6999342404775153, "grad_norm": 8.954756736755371, "learning_rate": 2.1006268687379418e-06, "loss": 0.4381, "step": 13837 }, { "epoch": 0.6999848247255804, "grad_norm": 6.127631187438965, "learning_rate": 2.0999730116805938e-06, "loss": 0.4048, "step": 13838 }, { "epoch": 0.7000354089736456, "grad_norm": 6.0337653160095215, "learning_rate": 2.099319229349935e-06, "loss": 0.3775, "step": 13839 }, { "epoch": 0.7000354089736456, "eval_loss": 0.39474156498908997, "eval_runtime": 34.9063, "eval_samples_per_second": 45.78, "eval_steps_per_second": 5.73, "step": 13839 }, { "epoch": 0.7000859932217107, "grad_norm": 7.240963935852051, "learning_rate": 2.0986655217628155e-06, "loss": 0.3937, "step": 13840 }, { "epoch": 0.7001365774697759, "grad_norm": 13.623001098632812, "learning_rate": 2.0980118889360773e-06, "loss": 0.5995, "step": 13841 }, { "epoch": 0.700187161717841, "grad_norm": 7.995843887329102, "learning_rate": 2.0973583308865613e-06, "loss": 0.3428, "step": 13842 }, { "epoch": 0.7002377459659063, "grad_norm": 8.44603443145752, "learning_rate": 2.0967048476311125e-06, "loss": 0.3972, "step": 13843 }, { "epoch": 0.7002883302139714, "grad_norm": 6.584010124206543, "learning_rate": 2.0960514391865665e-06, "loss": 0.4561, "step": 13844 }, { "epoch": 0.7003389144620366, "grad_norm": 6.870177745819092, "learning_rate": 2.095398105569761e-06, "loss": 0.3763, "step": 13845 }, { "epoch": 0.7003894987101017, "grad_norm": 7.061786651611328, "learning_rate": 2.0947448467975305e-06, "loss": 0.3384, "step": 13846 }, { "epoch": 0.7004400829581668, "grad_norm": 7.793879508972168, "learning_rate": 2.0940916628867063e-06, "loss": 0.4603, "step": 13847 }, { "epoch": 0.700490667206232, "grad_norm": 4.951472759246826, "learning_rate": 2.0934385538541223e-06, "loss": 0.3232, "step": 13848 }, { "epoch": 0.7005412514542971, "grad_norm": 5.597084999084473, "learning_rate": 2.092785519716607e-06, "loss": 0.3211, "step": 13849 }, { "epoch": 0.7005918357023623, "grad_norm": 6.736191749572754, "learning_rate": 2.092132560490984e-06, "loss": 0.3287, "step": 13850 }, { "epoch": 0.7006424199504274, "grad_norm": 5.471627235412598, "learning_rate": 2.0914796761940842e-06, "loss": 0.3596, "step": 13851 }, { "epoch": 0.7006930041984926, "grad_norm": 3.8317019939422607, "learning_rate": 2.0908268668427273e-06, "loss": 0.2776, "step": 13852 }, { "epoch": 0.7007435884465577, "grad_norm": 11.637433052062988, "learning_rate": 2.090174132453735e-06, "loss": 0.5622, "step": 13853 }, { "epoch": 0.7007941726946229, "grad_norm": 16.975505828857422, "learning_rate": 2.0895214730439258e-06, "loss": 0.4737, "step": 13854 }, { "epoch": 0.700844756942688, "grad_norm": 11.26988697052002, "learning_rate": 2.0888688886301195e-06, "loss": 0.5635, "step": 13855 }, { "epoch": 0.7008953411907533, "grad_norm": 6.148943901062012, "learning_rate": 2.0882163792291304e-06, "loss": 0.3972, "step": 13856 }, { "epoch": 0.7009459254388184, "grad_norm": 7.559780597686768, "learning_rate": 2.0875639448577723e-06, "loss": 0.3692, "step": 13857 }, { "epoch": 0.7009965096868835, "grad_norm": 6.303500175476074, "learning_rate": 2.0869115855328564e-06, "loss": 0.4105, "step": 13858 }, { "epoch": 0.7010470939349487, "grad_norm": 6.542943477630615, "learning_rate": 2.086259301271191e-06, "loss": 0.3726, "step": 13859 }, { "epoch": 0.7010976781830138, "grad_norm": 6.146594524383545, "learning_rate": 2.0856070920895876e-06, "loss": 0.4327, "step": 13860 }, { "epoch": 0.701148262431079, "grad_norm": 6.153018951416016, "learning_rate": 2.0849549580048495e-06, "loss": 0.3223, "step": 13861 }, { "epoch": 0.7011988466791441, "grad_norm": 8.114384651184082, "learning_rate": 2.08430289903378e-06, "loss": 0.3714, "step": 13862 }, { "epoch": 0.7012494309272093, "grad_norm": 10.065958976745605, "learning_rate": 2.0836509151931832e-06, "loss": 0.4231, "step": 13863 }, { "epoch": 0.7013000151752744, "grad_norm": 8.686763763427734, "learning_rate": 2.082999006499859e-06, "loss": 0.4964, "step": 13864 }, { "epoch": 0.7013505994233395, "grad_norm": 9.023338317871094, "learning_rate": 2.0823471729706023e-06, "loss": 0.4138, "step": 13865 }, { "epoch": 0.7014011836714047, "grad_norm": 8.172601699829102, "learning_rate": 2.0816954146222135e-06, "loss": 0.4553, "step": 13866 }, { "epoch": 0.7014517679194698, "grad_norm": 10.019065856933594, "learning_rate": 2.0810437314714848e-06, "loss": 0.4268, "step": 13867 }, { "epoch": 0.701502352167535, "grad_norm": 6.971786022186279, "learning_rate": 2.0803921235352074e-06, "loss": 0.3826, "step": 13868 }, { "epoch": 0.7015529364156002, "grad_norm": 6.273543357849121, "learning_rate": 2.0797405908301747e-06, "loss": 0.3304, "step": 13869 }, { "epoch": 0.7016035206636654, "grad_norm": 7.596099853515625, "learning_rate": 2.079089133373175e-06, "loss": 0.4396, "step": 13870 }, { "epoch": 0.7016541049117305, "grad_norm": 7.555072784423828, "learning_rate": 2.0784377511809895e-06, "loss": 0.3484, "step": 13871 }, { "epoch": 0.7017046891597957, "grad_norm": 5.108703136444092, "learning_rate": 2.0777864442704083e-06, "loss": 0.3971, "step": 13872 }, { "epoch": 0.7017552734078608, "grad_norm": 7.415889739990234, "learning_rate": 2.077135212658212e-06, "loss": 0.4214, "step": 13873 }, { "epoch": 0.701805857655926, "grad_norm": 5.632522106170654, "learning_rate": 2.07648405636118e-06, "loss": 0.4114, "step": 13874 }, { "epoch": 0.7018564419039911, "grad_norm": 9.650215148925781, "learning_rate": 2.0758329753960937e-06, "loss": 0.419, "step": 13875 }, { "epoch": 0.7019070261520562, "grad_norm": 5.226459980010986, "learning_rate": 2.0751819697797287e-06, "loss": 0.3183, "step": 13876 }, { "epoch": 0.7019576104001214, "grad_norm": 9.78796100616455, "learning_rate": 2.074531039528858e-06, "loss": 0.3259, "step": 13877 }, { "epoch": 0.7020081946481865, "grad_norm": 14.689399719238281, "learning_rate": 2.073880184660258e-06, "loss": 0.545, "step": 13878 }, { "epoch": 0.7020587788962517, "grad_norm": 11.375218391418457, "learning_rate": 2.073229405190698e-06, "loss": 0.4255, "step": 13879 }, { "epoch": 0.7021093631443168, "grad_norm": 5.987071514129639, "learning_rate": 2.0725787011369453e-06, "loss": 0.3557, "step": 13880 }, { "epoch": 0.702159947392382, "grad_norm": 8.473838806152344, "learning_rate": 2.0719280725157696e-06, "loss": 0.363, "step": 13881 }, { "epoch": 0.7022105316404471, "grad_norm": 10.612407684326172, "learning_rate": 2.071277519343936e-06, "loss": 0.3789, "step": 13882 }, { "epoch": 0.7022611158885124, "grad_norm": 10.269258499145508, "learning_rate": 2.0706270416382057e-06, "loss": 0.3595, "step": 13883 }, { "epoch": 0.7023117001365775, "grad_norm": 9.962778091430664, "learning_rate": 2.0699766394153415e-06, "loss": 0.4485, "step": 13884 }, { "epoch": 0.7023622843846427, "grad_norm": 9.746152877807617, "learning_rate": 2.0693263126921004e-06, "loss": 0.4078, "step": 13885 }, { "epoch": 0.7024128686327078, "grad_norm": 7.4444684982299805, "learning_rate": 2.0686760614852432e-06, "loss": 0.4362, "step": 13886 }, { "epoch": 0.7024634528807729, "grad_norm": 7.326236248016357, "learning_rate": 2.0680258858115233e-06, "loss": 0.3812, "step": 13887 }, { "epoch": 0.7025140371288381, "grad_norm": 7.780640125274658, "learning_rate": 2.0673757856876947e-06, "loss": 0.3157, "step": 13888 }, { "epoch": 0.7025646213769032, "grad_norm": 7.9801764488220215, "learning_rate": 2.066725761130507e-06, "loss": 0.4041, "step": 13889 }, { "epoch": 0.7026152056249684, "grad_norm": 5.398169994354248, "learning_rate": 2.0660758121567125e-06, "loss": 0.3313, "step": 13890 }, { "epoch": 0.7026657898730335, "grad_norm": 6.586843013763428, "learning_rate": 2.0654259387830583e-06, "loss": 0.3765, "step": 13891 }, { "epoch": 0.7027163741210987, "grad_norm": 7.537006855010986, "learning_rate": 2.064776141026288e-06, "loss": 0.3828, "step": 13892 }, { "epoch": 0.7027669583691638, "grad_norm": 10.755000114440918, "learning_rate": 2.064126418903148e-06, "loss": 0.5562, "step": 13893 }, { "epoch": 0.702817542617229, "grad_norm": 5.311598777770996, "learning_rate": 2.063476772430379e-06, "loss": 0.3376, "step": 13894 }, { "epoch": 0.7028681268652941, "grad_norm": 16.303180694580078, "learning_rate": 2.0628272016247184e-06, "loss": 0.5366, "step": 13895 }, { "epoch": 0.7029187111133594, "grad_norm": 6.871525287628174, "learning_rate": 2.0621777065029107e-06, "loss": 0.4467, "step": 13896 }, { "epoch": 0.7029692953614245, "grad_norm": 6.744163990020752, "learning_rate": 2.061528287081684e-06, "loss": 0.3852, "step": 13897 }, { "epoch": 0.7030198796094896, "grad_norm": 9.969684600830078, "learning_rate": 2.060878943377777e-06, "loss": 0.4623, "step": 13898 }, { "epoch": 0.7030704638575548, "grad_norm": 8.866228103637695, "learning_rate": 2.060229675407921e-06, "loss": 0.4034, "step": 13899 }, { "epoch": 0.7031210481056199, "grad_norm": 12.183638572692871, "learning_rate": 2.0595804831888432e-06, "loss": 0.5201, "step": 13900 }, { "epoch": 0.7031716323536851, "grad_norm": 9.586026191711426, "learning_rate": 2.058931366737276e-06, "loss": 0.5002, "step": 13901 }, { "epoch": 0.7032222166017502, "grad_norm": 10.012895584106445, "learning_rate": 2.0582823260699436e-06, "loss": 0.4261, "step": 13902 }, { "epoch": 0.7032728008498154, "grad_norm": 10.135383605957031, "learning_rate": 2.057633361203569e-06, "loss": 0.409, "step": 13903 }, { "epoch": 0.7033233850978805, "grad_norm": 9.951855659484863, "learning_rate": 2.0569844721548765e-06, "loss": 0.3758, "step": 13904 }, { "epoch": 0.7033739693459456, "grad_norm": 6.119856834411621, "learning_rate": 2.0563356589405865e-06, "loss": 0.291, "step": 13905 }, { "epoch": 0.7034245535940108, "grad_norm": 8.469856262207031, "learning_rate": 2.055686921577414e-06, "loss": 0.3406, "step": 13906 }, { "epoch": 0.7034751378420759, "grad_norm": 4.538480281829834, "learning_rate": 2.05503826008208e-06, "loss": 0.3109, "step": 13907 }, { "epoch": 0.7035257220901411, "grad_norm": 6.924924850463867, "learning_rate": 2.0543896744712982e-06, "loss": 0.3235, "step": 13908 }, { "epoch": 0.7035763063382063, "grad_norm": 7.148812294006348, "learning_rate": 2.0537411647617767e-06, "loss": 0.3329, "step": 13909 }, { "epoch": 0.7036268905862715, "grad_norm": 13.22742748260498, "learning_rate": 2.0530927309702305e-06, "loss": 0.4368, "step": 13910 }, { "epoch": 0.7036774748343366, "grad_norm": 11.52745532989502, "learning_rate": 2.0524443731133664e-06, "loss": 0.4998, "step": 13911 }, { "epoch": 0.7037280590824018, "grad_norm": 8.073746681213379, "learning_rate": 2.0517960912078897e-06, "loss": 0.4598, "step": 13912 }, { "epoch": 0.7037786433304669, "grad_norm": 6.175858974456787, "learning_rate": 2.0511478852705085e-06, "loss": 0.3455, "step": 13913 }, { "epoch": 0.7038292275785321, "grad_norm": 9.273910522460938, "learning_rate": 2.0504997553179234e-06, "loss": 0.3433, "step": 13914 }, { "epoch": 0.7038798118265972, "grad_norm": 12.746734619140625, "learning_rate": 2.0498517013668335e-06, "loss": 0.472, "step": 13915 }, { "epoch": 0.7039303960746623, "grad_norm": 5.5999298095703125, "learning_rate": 2.0492037234339416e-06, "loss": 0.2408, "step": 13916 }, { "epoch": 0.7039809803227275, "grad_norm": 6.01600980758667, "learning_rate": 2.0485558215359426e-06, "loss": 0.4465, "step": 13917 }, { "epoch": 0.7040315645707926, "grad_norm": 5.08497953414917, "learning_rate": 2.0479079956895287e-06, "loss": 0.334, "step": 13918 }, { "epoch": 0.7040821488188578, "grad_norm": 7.495347499847412, "learning_rate": 2.0472602459113973e-06, "loss": 0.4104, "step": 13919 }, { "epoch": 0.7041327330669229, "grad_norm": 6.925504684448242, "learning_rate": 2.046612572218237e-06, "loss": 0.4388, "step": 13920 }, { "epoch": 0.7041833173149881, "grad_norm": 6.177640914916992, "learning_rate": 2.045964974626738e-06, "loss": 0.4016, "step": 13921 }, { "epoch": 0.7042339015630532, "grad_norm": 9.861183166503906, "learning_rate": 2.045317453153586e-06, "loss": 0.5242, "step": 13922 }, { "epoch": 0.7042844858111185, "grad_norm": 9.839924812316895, "learning_rate": 2.0446700078154647e-06, "loss": 0.3962, "step": 13923 }, { "epoch": 0.7043350700591836, "grad_norm": 6.533062934875488, "learning_rate": 2.04402263862906e-06, "loss": 0.3986, "step": 13924 }, { "epoch": 0.7043856543072488, "grad_norm": 10.519274711608887, "learning_rate": 2.0433753456110534e-06, "loss": 0.4144, "step": 13925 }, { "epoch": 0.7044362385553139, "grad_norm": 5.203338623046875, "learning_rate": 2.042728128778122e-06, "loss": 0.3529, "step": 13926 }, { "epoch": 0.704486822803379, "grad_norm": 8.251020431518555, "learning_rate": 2.0420809881469415e-06, "loss": 0.4648, "step": 13927 }, { "epoch": 0.7045374070514442, "grad_norm": 6.879315376281738, "learning_rate": 2.0414339237341916e-06, "loss": 0.4471, "step": 13928 }, { "epoch": 0.7045879912995093, "grad_norm": 5.936745643615723, "learning_rate": 2.0407869355565436e-06, "loss": 0.3426, "step": 13929 }, { "epoch": 0.7046385755475745, "grad_norm": 10.110798835754395, "learning_rate": 2.0401400236306666e-06, "loss": 0.5589, "step": 13930 }, { "epoch": 0.7046891597956396, "grad_norm": 6.683226108551025, "learning_rate": 2.0394931879732337e-06, "loss": 0.4389, "step": 13931 }, { "epoch": 0.7047397440437048, "grad_norm": 9.494150161743164, "learning_rate": 2.0388464286009107e-06, "loss": 0.4002, "step": 13932 }, { "epoch": 0.7047903282917699, "grad_norm": 9.930190086364746, "learning_rate": 2.038199745530361e-06, "loss": 0.4031, "step": 13933 }, { "epoch": 0.704840912539835, "grad_norm": 6.254626274108887, "learning_rate": 2.0375531387782533e-06, "loss": 0.3584, "step": 13934 }, { "epoch": 0.7048914967879002, "grad_norm": 15.934465408325195, "learning_rate": 2.036906608361242e-06, "loss": 0.5071, "step": 13935 }, { "epoch": 0.7049420810359655, "grad_norm": 12.891039848327637, "learning_rate": 2.0362601542959927e-06, "loss": 0.5579, "step": 13936 }, { "epoch": 0.7049926652840306, "grad_norm": 12.555034637451172, "learning_rate": 2.0356137765991602e-06, "loss": 0.4387, "step": 13937 }, { "epoch": 0.7050432495320957, "grad_norm": 14.923542976379395, "learning_rate": 2.034967475287399e-06, "loss": 0.4625, "step": 13938 }, { "epoch": 0.7050938337801609, "grad_norm": 7.8640522956848145, "learning_rate": 2.034321250377366e-06, "loss": 0.4525, "step": 13939 }, { "epoch": 0.705144418028226, "grad_norm": 6.495423316955566, "learning_rate": 2.0336751018857105e-06, "loss": 0.3577, "step": 13940 }, { "epoch": 0.7051950022762912, "grad_norm": 10.710257530212402, "learning_rate": 2.033029029829081e-06, "loss": 0.4525, "step": 13941 }, { "epoch": 0.7052455865243563, "grad_norm": 7.883151531219482, "learning_rate": 2.0323830342241295e-06, "loss": 0.3128, "step": 13942 }, { "epoch": 0.7052961707724215, "grad_norm": 8.613018989562988, "learning_rate": 2.0317371150874985e-06, "loss": 0.4067, "step": 13943 }, { "epoch": 0.7053467550204866, "grad_norm": 6.077701091766357, "learning_rate": 2.031091272435831e-06, "loss": 0.3547, "step": 13944 }, { "epoch": 0.7053973392685517, "grad_norm": 11.609472274780273, "learning_rate": 2.0304455062857726e-06, "loss": 0.4893, "step": 13945 }, { "epoch": 0.7054479235166169, "grad_norm": 6.292057514190674, "learning_rate": 2.02979981665396e-06, "loss": 0.2756, "step": 13946 }, { "epoch": 0.705498507764682, "grad_norm": 8.502209663391113, "learning_rate": 2.0291542035570326e-06, "loss": 0.5151, "step": 13947 }, { "epoch": 0.7055490920127472, "grad_norm": 8.121343612670898, "learning_rate": 2.0285086670116255e-06, "loss": 0.4544, "step": 13948 }, { "epoch": 0.7055996762608124, "grad_norm": 8.318493843078613, "learning_rate": 2.0278632070343723e-06, "loss": 0.391, "step": 13949 }, { "epoch": 0.7056502605088776, "grad_norm": 5.830516815185547, "learning_rate": 2.027217823641904e-06, "loss": 0.3415, "step": 13950 }, { "epoch": 0.7057008447569427, "grad_norm": 6.0044074058532715, "learning_rate": 2.026572516850853e-06, "loss": 0.3291, "step": 13951 }, { "epoch": 0.7057514290050079, "grad_norm": 4.934459209442139, "learning_rate": 2.0259272866778466e-06, "loss": 0.3976, "step": 13952 }, { "epoch": 0.705802013253073, "grad_norm": 12.204010963439941, "learning_rate": 2.025282133139509e-06, "loss": 0.4381, "step": 13953 }, { "epoch": 0.7058525975011382, "grad_norm": 11.143355369567871, "learning_rate": 2.024637056252467e-06, "loss": 0.5177, "step": 13954 }, { "epoch": 0.7059031817492033, "grad_norm": 9.282377243041992, "learning_rate": 2.023992056033342e-06, "loss": 0.4144, "step": 13955 }, { "epoch": 0.7059537659972684, "grad_norm": 6.04133415222168, "learning_rate": 2.023347132498751e-06, "loss": 0.3645, "step": 13956 }, { "epoch": 0.7060043502453336, "grad_norm": 7.396640777587891, "learning_rate": 2.0227022856653167e-06, "loss": 0.4214, "step": 13957 }, { "epoch": 0.7060549344933987, "grad_norm": 7.389886379241943, "learning_rate": 2.022057515549653e-06, "loss": 0.4573, "step": 13958 }, { "epoch": 0.7061055187414639, "grad_norm": 4.576076030731201, "learning_rate": 2.0214128221683727e-06, "loss": 0.3043, "step": 13959 }, { "epoch": 0.706156102989529, "grad_norm": 9.34876823425293, "learning_rate": 2.020768205538093e-06, "loss": 0.4517, "step": 13960 }, { "epoch": 0.7062066872375942, "grad_norm": 7.589937210083008, "learning_rate": 2.0201236656754166e-06, "loss": 0.4885, "step": 13961 }, { "epoch": 0.7062572714856593, "grad_norm": 5.226612567901611, "learning_rate": 2.019479202596958e-06, "loss": 0.4455, "step": 13962 }, { "epoch": 0.7063078557337246, "grad_norm": 10.20436954498291, "learning_rate": 2.0188348163193205e-06, "loss": 0.4138, "step": 13963 }, { "epoch": 0.7063584399817897, "grad_norm": 13.022562980651855, "learning_rate": 2.0181905068591096e-06, "loss": 0.4842, "step": 13964 }, { "epoch": 0.7064090242298549, "grad_norm": 7.638881206512451, "learning_rate": 2.0175462742329257e-06, "loss": 0.396, "step": 13965 }, { "epoch": 0.70645960847792, "grad_norm": 5.959121227264404, "learning_rate": 2.016902118457372e-06, "loss": 0.4525, "step": 13966 }, { "epoch": 0.7065101927259851, "grad_norm": 4.978411674499512, "learning_rate": 2.0162580395490442e-06, "loss": 0.3362, "step": 13967 }, { "epoch": 0.7065607769740503, "grad_norm": 6.048893928527832, "learning_rate": 2.0156140375245394e-06, "loss": 0.3616, "step": 13968 }, { "epoch": 0.7066113612221154, "grad_norm": 8.360559463500977, "learning_rate": 2.014970112400454e-06, "loss": 0.4288, "step": 13969 }, { "epoch": 0.7066619454701806, "grad_norm": 6.558760166168213, "learning_rate": 2.014326264193378e-06, "loss": 0.3244, "step": 13970 }, { "epoch": 0.7067125297182457, "grad_norm": 5.398958206176758, "learning_rate": 2.013682492919902e-06, "loss": 0.4039, "step": 13971 }, { "epoch": 0.7067631139663109, "grad_norm": 8.649978637695312, "learning_rate": 2.0130387985966152e-06, "loss": 0.3583, "step": 13972 }, { "epoch": 0.706813698214376, "grad_norm": 7.291287899017334, "learning_rate": 2.012395181240105e-06, "loss": 0.2993, "step": 13973 }, { "epoch": 0.7068642824624412, "grad_norm": 8.876628875732422, "learning_rate": 2.0117516408669546e-06, "loss": 0.4465, "step": 13974 }, { "epoch": 0.7069148667105063, "grad_norm": 6.985763072967529, "learning_rate": 2.011108177493746e-06, "loss": 0.4068, "step": 13975 }, { "epoch": 0.7069654509585716, "grad_norm": 7.667336463928223, "learning_rate": 2.0104647911370588e-06, "loss": 0.4125, "step": 13976 }, { "epoch": 0.7070160352066367, "grad_norm": 6.672095775604248, "learning_rate": 2.0098214818134753e-06, "loss": 0.44, "step": 13977 }, { "epoch": 0.7070666194547018, "grad_norm": 6.618905544281006, "learning_rate": 2.009178249539569e-06, "loss": 0.3823, "step": 13978 }, { "epoch": 0.707117203702767, "grad_norm": 6.971121311187744, "learning_rate": 2.0085350943319136e-06, "loss": 0.2509, "step": 13979 }, { "epoch": 0.7071677879508321, "grad_norm": 9.94212818145752, "learning_rate": 2.0078920162070853e-06, "loss": 0.4757, "step": 13980 }, { "epoch": 0.7072183721988973, "grad_norm": 12.285396575927734, "learning_rate": 2.0072490151816526e-06, "loss": 0.409, "step": 13981 }, { "epoch": 0.7072689564469624, "grad_norm": 6.4383111000061035, "learning_rate": 2.0066060912721823e-06, "loss": 0.4156, "step": 13982 }, { "epoch": 0.7073195406950276, "grad_norm": 5.970061779022217, "learning_rate": 2.005963244495244e-06, "loss": 0.3996, "step": 13983 }, { "epoch": 0.7073701249430927, "grad_norm": 10.091899871826172, "learning_rate": 2.0053204748674016e-06, "loss": 0.4056, "step": 13984 }, { "epoch": 0.7074207091911578, "grad_norm": 12.807757377624512, "learning_rate": 2.0046777824052156e-06, "loss": 0.4232, "step": 13985 }, { "epoch": 0.707471293439223, "grad_norm": 6.880552291870117, "learning_rate": 2.0040351671252516e-06, "loss": 0.3581, "step": 13986 }, { "epoch": 0.7075218776872881, "grad_norm": 7.18998908996582, "learning_rate": 2.003392629044063e-06, "loss": 0.3659, "step": 13987 }, { "epoch": 0.7075724619353533, "grad_norm": 7.972249984741211, "learning_rate": 2.002750168178206e-06, "loss": 0.3498, "step": 13988 }, { "epoch": 0.7076230461834185, "grad_norm": 6.970620155334473, "learning_rate": 2.0021077845442404e-06, "loss": 0.3725, "step": 13989 }, { "epoch": 0.7076736304314837, "grad_norm": 9.453008651733398, "learning_rate": 2.0014654781587155e-06, "loss": 0.4049, "step": 13990 }, { "epoch": 0.7077242146795488, "grad_norm": 14.278307914733887, "learning_rate": 2.000823249038181e-06, "loss": 0.3693, "step": 13991 }, { "epoch": 0.707774798927614, "grad_norm": 7.695545673370361, "learning_rate": 2.000181097199189e-06, "loss": 0.4294, "step": 13992 }, { "epoch": 0.7078253831756791, "grad_norm": 7.497433185577393, "learning_rate": 1.9995390226582843e-06, "loss": 0.3865, "step": 13993 }, { "epoch": 0.7078759674237443, "grad_norm": 10.176424980163574, "learning_rate": 1.99889702543201e-06, "loss": 0.4401, "step": 13994 }, { "epoch": 0.7079265516718094, "grad_norm": 13.852700233459473, "learning_rate": 1.9982551055369122e-06, "loss": 0.5865, "step": 13995 }, { "epoch": 0.7079771359198745, "grad_norm": 19.070308685302734, "learning_rate": 1.9976132629895295e-06, "loss": 0.4099, "step": 13996 }, { "epoch": 0.7080277201679397, "grad_norm": 7.006189346313477, "learning_rate": 1.9969714978064e-06, "loss": 0.4338, "step": 13997 }, { "epoch": 0.7080783044160048, "grad_norm": 6.302427291870117, "learning_rate": 1.996329810004062e-06, "loss": 0.3781, "step": 13998 }, { "epoch": 0.70812888866407, "grad_norm": 5.648698329925537, "learning_rate": 1.995688199599053e-06, "loss": 0.4521, "step": 13999 }, { "epoch": 0.7081794729121351, "grad_norm": 12.184345245361328, "learning_rate": 1.9950466666078982e-06, "loss": 0.4623, "step": 14000 }, { "epoch": 0.7082300571602003, "grad_norm": 8.457483291625977, "learning_rate": 1.9944052110471347e-06, "loss": 0.3885, "step": 14001 }, { "epoch": 0.7082806414082655, "grad_norm": 9.237181663513184, "learning_rate": 1.993763832933289e-06, "loss": 0.3249, "step": 14002 }, { "epoch": 0.7083312256563307, "grad_norm": 7.3234710693359375, "learning_rate": 1.993122532282886e-06, "loss": 0.3902, "step": 14003 }, { "epoch": 0.7083818099043958, "grad_norm": 9.129989624023438, "learning_rate": 1.9924813091124545e-06, "loss": 0.3691, "step": 14004 }, { "epoch": 0.708432394152461, "grad_norm": 12.107268333435059, "learning_rate": 1.9918401634385156e-06, "loss": 0.4341, "step": 14005 }, { "epoch": 0.7084829784005261, "grad_norm": 6.984358310699463, "learning_rate": 1.991199095277588e-06, "loss": 0.3559, "step": 14006 }, { "epoch": 0.7085335626485912, "grad_norm": 19.65762710571289, "learning_rate": 1.9905581046461935e-06, "loss": 0.6842, "step": 14007 }, { "epoch": 0.7085841468966564, "grad_norm": 6.808338642120361, "learning_rate": 1.989917191560848e-06, "loss": 0.4612, "step": 14008 }, { "epoch": 0.7086347311447215, "grad_norm": 12.68513011932373, "learning_rate": 1.989276356038064e-06, "loss": 0.5086, "step": 14009 }, { "epoch": 0.7086853153927867, "grad_norm": 7.559553623199463, "learning_rate": 1.988635598094358e-06, "loss": 0.3909, "step": 14010 }, { "epoch": 0.7087358996408518, "grad_norm": 4.799829006195068, "learning_rate": 1.987994917746239e-06, "loss": 0.3973, "step": 14011 }, { "epoch": 0.708786483888917, "grad_norm": 8.639490127563477, "learning_rate": 1.987354315010216e-06, "loss": 0.4095, "step": 14012 }, { "epoch": 0.7088370681369821, "grad_norm": 7.904318332672119, "learning_rate": 1.986713789902795e-06, "loss": 0.3369, "step": 14013 }, { "epoch": 0.7088876523850473, "grad_norm": 9.323270797729492, "learning_rate": 1.98607334244048e-06, "loss": 0.3945, "step": 14014 }, { "epoch": 0.7089382366331124, "grad_norm": 16.13337516784668, "learning_rate": 1.985432972639777e-06, "loss": 0.4437, "step": 14015 }, { "epoch": 0.7089888208811777, "grad_norm": 10.33155632019043, "learning_rate": 1.9847926805171845e-06, "loss": 0.4386, "step": 14016 }, { "epoch": 0.7090394051292428, "grad_norm": 5.3834357261657715, "learning_rate": 1.9841524660892003e-06, "loss": 0.3909, "step": 14017 }, { "epoch": 0.7090899893773079, "grad_norm": 9.424201965332031, "learning_rate": 1.9835123293723242e-06, "loss": 0.4566, "step": 14018 }, { "epoch": 0.7091405736253731, "grad_norm": 9.179667472839355, "learning_rate": 1.9828722703830496e-06, "loss": 0.4433, "step": 14019 }, { "epoch": 0.7091911578734382, "grad_norm": 9.220367431640625, "learning_rate": 1.9822322891378664e-06, "loss": 0.446, "step": 14020 }, { "epoch": 0.7092417421215034, "grad_norm": 4.580914497375488, "learning_rate": 1.9815923856532704e-06, "loss": 0.3569, "step": 14021 }, { "epoch": 0.7092923263695685, "grad_norm": 12.536128044128418, "learning_rate": 1.980952559945748e-06, "loss": 0.4832, "step": 14022 }, { "epoch": 0.7093429106176337, "grad_norm": 5.746030330657959, "learning_rate": 1.980312812031786e-06, "loss": 0.3824, "step": 14023 }, { "epoch": 0.7093934948656988, "grad_norm": 8.259885787963867, "learning_rate": 1.9796731419278684e-06, "loss": 0.3566, "step": 14024 }, { "epoch": 0.709444079113764, "grad_norm": 4.082775115966797, "learning_rate": 1.979033549650479e-06, "loss": 0.2325, "step": 14025 }, { "epoch": 0.7094946633618291, "grad_norm": 10.473633766174316, "learning_rate": 1.9783940352160968e-06, "loss": 0.3377, "step": 14026 }, { "epoch": 0.7095452476098942, "grad_norm": 8.43924617767334, "learning_rate": 1.9777545986412033e-06, "loss": 0.35, "step": 14027 }, { "epoch": 0.7095958318579594, "grad_norm": 26.154769897460938, "learning_rate": 1.977115239942274e-06, "loss": 0.4812, "step": 14028 }, { "epoch": 0.7096464161060246, "grad_norm": 6.2244486808776855, "learning_rate": 1.9764759591357817e-06, "loss": 0.4773, "step": 14029 }, { "epoch": 0.7096970003540898, "grad_norm": 9.86608600616455, "learning_rate": 1.975836756238203e-06, "loss": 0.3098, "step": 14030 }, { "epoch": 0.7097475846021549, "grad_norm": 4.652653217315674, "learning_rate": 1.9751976312660058e-06, "loss": 0.3619, "step": 14031 }, { "epoch": 0.7097981688502201, "grad_norm": 6.155359745025635, "learning_rate": 1.9745585842356583e-06, "loss": 0.4536, "step": 14032 }, { "epoch": 0.7098487530982852, "grad_norm": 5.309908390045166, "learning_rate": 1.97391961516363e-06, "loss": 0.3875, "step": 14033 }, { "epoch": 0.7098993373463504, "grad_norm": 11.34347915649414, "learning_rate": 1.9732807240663843e-06, "loss": 0.4797, "step": 14034 }, { "epoch": 0.7099499215944155, "grad_norm": 7.171929359436035, "learning_rate": 1.972641910960382e-06, "loss": 0.3934, "step": 14035 }, { "epoch": 0.7100005058424806, "grad_norm": 6.675329208374023, "learning_rate": 1.9720031758620865e-06, "loss": 0.4212, "step": 14036 }, { "epoch": 0.7100510900905458, "grad_norm": 9.47982120513916, "learning_rate": 1.971364518787957e-06, "loss": 0.4934, "step": 14037 }, { "epoch": 0.7101016743386109, "grad_norm": 6.120007514953613, "learning_rate": 1.9707259397544453e-06, "loss": 0.3532, "step": 14038 }, { "epoch": 0.7101522585866761, "grad_norm": 13.773612022399902, "learning_rate": 1.970087438778011e-06, "loss": 0.5168, "step": 14039 }, { "epoch": 0.7102028428347412, "grad_norm": 5.676473140716553, "learning_rate": 1.969449015875105e-06, "loss": 0.4072, "step": 14040 }, { "epoch": 0.7102534270828064, "grad_norm": 10.306158065795898, "learning_rate": 1.9688106710621764e-06, "loss": 0.556, "step": 14041 }, { "epoch": 0.7103040113308716, "grad_norm": 7.618062496185303, "learning_rate": 1.9681724043556767e-06, "loss": 0.4396, "step": 14042 }, { "epoch": 0.7103545955789368, "grad_norm": 7.749906063079834, "learning_rate": 1.9675342157720506e-06, "loss": 0.3605, "step": 14043 }, { "epoch": 0.7104051798270019, "grad_norm": 5.531370639801025, "learning_rate": 1.966896105327742e-06, "loss": 0.3868, "step": 14044 }, { "epoch": 0.710455764075067, "grad_norm": 7.088109493255615, "learning_rate": 1.966258073039196e-06, "loss": 0.429, "step": 14045 }, { "epoch": 0.7105063483231322, "grad_norm": 5.63020133972168, "learning_rate": 1.965620118922852e-06, "loss": 0.3248, "step": 14046 }, { "epoch": 0.7105569325711973, "grad_norm": 8.912158012390137, "learning_rate": 1.9649822429951464e-06, "loss": 0.4157, "step": 14047 }, { "epoch": 0.7106075168192625, "grad_norm": 4.848527908325195, "learning_rate": 1.9643444452725195e-06, "loss": 0.3114, "step": 14048 }, { "epoch": 0.7106581010673276, "grad_norm": 6.263546466827393, "learning_rate": 1.9637067257714036e-06, "loss": 0.3947, "step": 14049 }, { "epoch": 0.7107086853153928, "grad_norm": 10.12048625946045, "learning_rate": 1.9630690845082324e-06, "loss": 0.4366, "step": 14050 }, { "epoch": 0.7107592695634579, "grad_norm": 9.577895164489746, "learning_rate": 1.9624315214994343e-06, "loss": 0.4078, "step": 14051 }, { "epoch": 0.7108098538115231, "grad_norm": 6.937655448913574, "learning_rate": 1.961794036761438e-06, "loss": 0.3596, "step": 14052 }, { "epoch": 0.7108604380595882, "grad_norm": 6.699718952178955, "learning_rate": 1.961156630310672e-06, "loss": 0.3895, "step": 14053 }, { "epoch": 0.7109110223076534, "grad_norm": 8.51618766784668, "learning_rate": 1.9605193021635604e-06, "loss": 0.436, "step": 14054 }, { "epoch": 0.7109616065557185, "grad_norm": 7.954690933227539, "learning_rate": 1.9598820523365226e-06, "loss": 0.4874, "step": 14055 }, { "epoch": 0.7110121908037838, "grad_norm": 8.354486465454102, "learning_rate": 1.959244880845983e-06, "loss": 0.3836, "step": 14056 }, { "epoch": 0.7110627750518489, "grad_norm": 10.15731143951416, "learning_rate": 1.958607787708358e-06, "loss": 0.3937, "step": 14057 }, { "epoch": 0.711113359299914, "grad_norm": 7.209853172302246, "learning_rate": 1.9579707729400626e-06, "loss": 0.3778, "step": 14058 }, { "epoch": 0.7111639435479792, "grad_norm": 5.7113752365112305, "learning_rate": 1.9573338365575143e-06, "loss": 0.3964, "step": 14059 }, { "epoch": 0.7112145277960443, "grad_norm": 5.968541622161865, "learning_rate": 1.956696978577124e-06, "loss": 0.3582, "step": 14060 }, { "epoch": 0.7112651120441095, "grad_norm": 11.867600440979004, "learning_rate": 1.9560601990153016e-06, "loss": 0.4911, "step": 14061 }, { "epoch": 0.7113156962921746, "grad_norm": 8.046000480651855, "learning_rate": 1.955423497888454e-06, "loss": 0.4564, "step": 14062 }, { "epoch": 0.7113662805402398, "grad_norm": 18.296775817871094, "learning_rate": 1.954786875212992e-06, "loss": 0.429, "step": 14063 }, { "epoch": 0.7114168647883049, "grad_norm": 5.1145100593566895, "learning_rate": 1.954150331005314e-06, "loss": 0.3927, "step": 14064 }, { "epoch": 0.71146744903637, "grad_norm": 7.191720485687256, "learning_rate": 1.953513865281826e-06, "loss": 0.4391, "step": 14065 }, { "epoch": 0.7115180332844352, "grad_norm": 11.733932495117188, "learning_rate": 1.952877478058928e-06, "loss": 0.4586, "step": 14066 }, { "epoch": 0.7115686175325003, "grad_norm": 6.044886112213135, "learning_rate": 1.952241169353015e-06, "loss": 0.3947, "step": 14067 }, { "epoch": 0.7116192017805655, "grad_norm": 5.683352470397949, "learning_rate": 1.9516049391804874e-06, "loss": 0.4449, "step": 14068 }, { "epoch": 0.7116697860286307, "grad_norm": 10.791365623474121, "learning_rate": 1.9509687875577377e-06, "loss": 0.4557, "step": 14069 }, { "epoch": 0.7117203702766959, "grad_norm": 10.408873558044434, "learning_rate": 1.950332714501155e-06, "loss": 0.344, "step": 14070 }, { "epoch": 0.711770954524761, "grad_norm": 7.130898475646973, "learning_rate": 1.949696720027135e-06, "loss": 0.3326, "step": 14071 }, { "epoch": 0.7118215387728262, "grad_norm": 9.18786334991455, "learning_rate": 1.9490608041520625e-06, "loss": 0.4304, "step": 14072 }, { "epoch": 0.7118721230208913, "grad_norm": 4.930026531219482, "learning_rate": 1.9484249668923217e-06, "loss": 0.3691, "step": 14073 }, { "epoch": 0.7119227072689565, "grad_norm": 9.145397186279297, "learning_rate": 1.9477892082643006e-06, "loss": 0.4315, "step": 14074 }, { "epoch": 0.7119732915170216, "grad_norm": 10.931325912475586, "learning_rate": 1.9471535282843796e-06, "loss": 0.3567, "step": 14075 }, { "epoch": 0.7120238757650867, "grad_norm": 7.736468315124512, "learning_rate": 1.9465179269689383e-06, "loss": 0.3919, "step": 14076 }, { "epoch": 0.7120744600131519, "grad_norm": 8.391190528869629, "learning_rate": 1.9458824043343544e-06, "loss": 0.3979, "step": 14077 }, { "epoch": 0.712125044261217, "grad_norm": 7.380347728729248, "learning_rate": 1.945246960397004e-06, "loss": 0.2868, "step": 14078 }, { "epoch": 0.7121756285092822, "grad_norm": 12.57701587677002, "learning_rate": 1.9446115951732595e-06, "loss": 0.5449, "step": 14079 }, { "epoch": 0.7122262127573473, "grad_norm": 5.317625999450684, "learning_rate": 1.943976308679496e-06, "loss": 0.3951, "step": 14080 }, { "epoch": 0.7122767970054125, "grad_norm": 14.25572681427002, "learning_rate": 1.943341100932081e-06, "loss": 0.5085, "step": 14081 }, { "epoch": 0.7123273812534777, "grad_norm": 12.311138153076172, "learning_rate": 1.942705971947381e-06, "loss": 0.4432, "step": 14082 }, { "epoch": 0.7123779655015429, "grad_norm": 10.12783145904541, "learning_rate": 1.942070921741766e-06, "loss": 0.4682, "step": 14083 }, { "epoch": 0.712428549749608, "grad_norm": 4.7591729164123535, "learning_rate": 1.941435950331597e-06, "loss": 0.342, "step": 14084 }, { "epoch": 0.7124791339976732, "grad_norm": 18.50387954711914, "learning_rate": 1.9408010577332337e-06, "loss": 0.5136, "step": 14085 }, { "epoch": 0.7125297182457383, "grad_norm": 7.876088619232178, "learning_rate": 1.94016624396304e-06, "loss": 0.4032, "step": 14086 }, { "epoch": 0.7125803024938034, "grad_norm": 8.55419921875, "learning_rate": 1.9395315090373716e-06, "loss": 0.4033, "step": 14087 }, { "epoch": 0.7126308867418686, "grad_norm": 14.084444046020508, "learning_rate": 1.938896852972582e-06, "loss": 0.5446, "step": 14088 }, { "epoch": 0.7126814709899337, "grad_norm": 8.647863388061523, "learning_rate": 1.93826227578503e-06, "loss": 0.4451, "step": 14089 }, { "epoch": 0.7127320552379989, "grad_norm": 10.761575698852539, "learning_rate": 1.9376277774910607e-06, "loss": 0.4102, "step": 14090 }, { "epoch": 0.712782639486064, "grad_norm": 6.303272247314453, "learning_rate": 1.9369933581070282e-06, "loss": 0.3839, "step": 14091 }, { "epoch": 0.7128332237341292, "grad_norm": 10.79237174987793, "learning_rate": 1.9363590176492784e-06, "loss": 0.4709, "step": 14092 }, { "epoch": 0.7128838079821943, "grad_norm": 6.560828685760498, "learning_rate": 1.9357247561341556e-06, "loss": 0.4705, "step": 14093 }, { "epoch": 0.7129343922302595, "grad_norm": 7.927711009979248, "learning_rate": 1.935090573578006e-06, "loss": 0.3718, "step": 14094 }, { "epoch": 0.7129849764783246, "grad_norm": 6.5119404792785645, "learning_rate": 1.934456469997169e-06, "loss": 0.4104, "step": 14095 }, { "epoch": 0.7130355607263898, "grad_norm": 4.919114112854004, "learning_rate": 1.9338224454079823e-06, "loss": 0.3304, "step": 14096 }, { "epoch": 0.713086144974455, "grad_norm": 5.026751518249512, "learning_rate": 1.9331884998267875e-06, "loss": 0.3322, "step": 14097 }, { "epoch": 0.7131367292225201, "grad_norm": 7.766443252563477, "learning_rate": 1.932554633269917e-06, "loss": 0.4344, "step": 14098 }, { "epoch": 0.7131873134705853, "grad_norm": 5.958529949188232, "learning_rate": 1.931920845753705e-06, "loss": 0.3545, "step": 14099 }, { "epoch": 0.7132378977186504, "grad_norm": 12.994328498840332, "learning_rate": 1.9312871372944798e-06, "loss": 0.4215, "step": 14100 }, { "epoch": 0.7132884819667156, "grad_norm": 9.451725006103516, "learning_rate": 1.9306535079085746e-06, "loss": 0.442, "step": 14101 }, { "epoch": 0.7133390662147807, "grad_norm": 9.5132474899292, "learning_rate": 1.930019957612315e-06, "loss": 0.4004, "step": 14102 }, { "epoch": 0.7133896504628459, "grad_norm": 12.66768741607666, "learning_rate": 1.929386486422026e-06, "loss": 0.4926, "step": 14103 }, { "epoch": 0.713440234710911, "grad_norm": 16.033370971679688, "learning_rate": 1.928753094354031e-06, "loss": 0.5269, "step": 14104 }, { "epoch": 0.7134908189589761, "grad_norm": 10.99968433380127, "learning_rate": 1.9281197814246484e-06, "loss": 0.4478, "step": 14105 }, { "epoch": 0.7135414032070413, "grad_norm": 9.151865005493164, "learning_rate": 1.9274865476502005e-06, "loss": 0.37, "step": 14106 }, { "epoch": 0.7135919874551064, "grad_norm": 4.376491069793701, "learning_rate": 1.926853393047003e-06, "loss": 0.3627, "step": 14107 }, { "epoch": 0.7136425717031716, "grad_norm": 8.795355796813965, "learning_rate": 1.9262203176313692e-06, "loss": 0.4885, "step": 14108 }, { "epoch": 0.7136931559512368, "grad_norm": 7.199184894561768, "learning_rate": 1.9255873214196154e-06, "loss": 0.4291, "step": 14109 }, { "epoch": 0.713743740199302, "grad_norm": 10.983907699584961, "learning_rate": 1.9249544044280493e-06, "loss": 0.5075, "step": 14110 }, { "epoch": 0.7137943244473671, "grad_norm": 7.678073406219482, "learning_rate": 1.9243215666729804e-06, "loss": 0.3738, "step": 14111 }, { "epoch": 0.7138449086954323, "grad_norm": 6.782176971435547, "learning_rate": 1.9236888081707165e-06, "loss": 0.3517, "step": 14112 }, { "epoch": 0.7138954929434974, "grad_norm": 9.433399200439453, "learning_rate": 1.9230561289375617e-06, "loss": 0.3561, "step": 14113 }, { "epoch": 0.7139460771915626, "grad_norm": 6.689326763153076, "learning_rate": 1.9224235289898184e-06, "loss": 0.3213, "step": 14114 }, { "epoch": 0.7139966614396277, "grad_norm": 8.06222915649414, "learning_rate": 1.9217910083437873e-06, "loss": 0.4917, "step": 14115 }, { "epoch": 0.7140472456876928, "grad_norm": 10.588805198669434, "learning_rate": 1.9211585670157667e-06, "loss": 0.44, "step": 14116 }, { "epoch": 0.714097829935758, "grad_norm": 15.370142936706543, "learning_rate": 1.9205262050220514e-06, "loss": 0.5399, "step": 14117 }, { "epoch": 0.7141484141838231, "grad_norm": 8.270075798034668, "learning_rate": 1.91989392237894e-06, "loss": 0.3916, "step": 14118 }, { "epoch": 0.7141989984318883, "grad_norm": 5.382627010345459, "learning_rate": 1.919261719102722e-06, "loss": 0.3467, "step": 14119 }, { "epoch": 0.7142495826799534, "grad_norm": 12.473326683044434, "learning_rate": 1.918629595209686e-06, "loss": 0.5846, "step": 14120 }, { "epoch": 0.7143001669280186, "grad_norm": 5.7491068840026855, "learning_rate": 1.917997550716125e-06, "loss": 0.4111, "step": 14121 }, { "epoch": 0.7143507511760838, "grad_norm": 9.83119010925293, "learning_rate": 1.917365585638323e-06, "loss": 0.4777, "step": 14122 }, { "epoch": 0.714401335424149, "grad_norm": 6.90224027633667, "learning_rate": 1.916733699992562e-06, "loss": 0.4114, "step": 14123 }, { "epoch": 0.7144519196722141, "grad_norm": 4.471869945526123, "learning_rate": 1.916101893795128e-06, "loss": 0.3159, "step": 14124 }, { "epoch": 0.7145025039202793, "grad_norm": 5.642030239105225, "learning_rate": 1.9154701670622993e-06, "loss": 0.3218, "step": 14125 }, { "epoch": 0.7145530881683444, "grad_norm": 14.174864768981934, "learning_rate": 1.9148385198103525e-06, "loss": 0.4326, "step": 14126 }, { "epoch": 0.7146036724164095, "grad_norm": 5.387797832489014, "learning_rate": 1.9142069520555684e-06, "loss": 0.4593, "step": 14127 }, { "epoch": 0.7146542566644747, "grad_norm": 5.83986234664917, "learning_rate": 1.913575463814214e-06, "loss": 0.3962, "step": 14128 }, { "epoch": 0.7147048409125398, "grad_norm": 6.385584831237793, "learning_rate": 1.912944055102567e-06, "loss": 0.3191, "step": 14129 }, { "epoch": 0.714755425160605, "grad_norm": 5.130874156951904, "learning_rate": 1.9123127259368953e-06, "loss": 0.3765, "step": 14130 }, { "epoch": 0.7148060094086701, "grad_norm": 8.26893424987793, "learning_rate": 1.9116814763334646e-06, "loss": 0.3581, "step": 14131 }, { "epoch": 0.7148565936567353, "grad_norm": 8.016777992248535, "learning_rate": 1.911050306308545e-06, "loss": 0.4671, "step": 14132 }, { "epoch": 0.7149071779048004, "grad_norm": 4.533873558044434, "learning_rate": 1.9104192158783976e-06, "loss": 0.3901, "step": 14133 }, { "epoch": 0.7149577621528656, "grad_norm": 4.835855007171631, "learning_rate": 1.909788205059284e-06, "loss": 0.3297, "step": 14134 }, { "epoch": 0.7150083464009308, "grad_norm": 9.479826927185059, "learning_rate": 1.9091572738674636e-06, "loss": 0.4902, "step": 14135 }, { "epoch": 0.715058930648996, "grad_norm": 10.251829147338867, "learning_rate": 1.908526422319196e-06, "loss": 0.4424, "step": 14136 }, { "epoch": 0.7151095148970611, "grad_norm": 7.979663848876953, "learning_rate": 1.9078956504307345e-06, "loss": 0.3732, "step": 14137 }, { "epoch": 0.7151600991451262, "grad_norm": 5.723272800445557, "learning_rate": 1.9072649582183321e-06, "loss": 0.3463, "step": 14138 }, { "epoch": 0.7152106833931914, "grad_norm": 5.884732723236084, "learning_rate": 1.9066343456982438e-06, "loss": 0.4104, "step": 14139 }, { "epoch": 0.7152612676412565, "grad_norm": 17.157407760620117, "learning_rate": 1.9060038128867163e-06, "loss": 0.4732, "step": 14140 }, { "epoch": 0.7153118518893217, "grad_norm": 9.682984352111816, "learning_rate": 1.9053733597999974e-06, "loss": 0.4173, "step": 14141 }, { "epoch": 0.7153624361373868, "grad_norm": 5.841881275177002, "learning_rate": 1.904742986454332e-06, "loss": 0.3573, "step": 14142 }, { "epoch": 0.715413020385452, "grad_norm": 7.501591682434082, "learning_rate": 1.904112692865962e-06, "loss": 0.3497, "step": 14143 }, { "epoch": 0.7154636046335171, "grad_norm": 7.485074996948242, "learning_rate": 1.903482479051132e-06, "loss": 0.3832, "step": 14144 }, { "epoch": 0.7155141888815822, "grad_norm": 11.873478889465332, "learning_rate": 1.902852345026079e-06, "loss": 0.3201, "step": 14145 }, { "epoch": 0.7155647731296474, "grad_norm": 7.533287525177002, "learning_rate": 1.9022222908070381e-06, "loss": 0.4593, "step": 14146 }, { "epoch": 0.7156153573777125, "grad_norm": 7.905376434326172, "learning_rate": 1.901592316410249e-06, "loss": 0.3701, "step": 14147 }, { "epoch": 0.7156659416257777, "grad_norm": 13.788257598876953, "learning_rate": 1.9009624218519413e-06, "loss": 0.4691, "step": 14148 }, { "epoch": 0.7157165258738429, "grad_norm": 7.424788951873779, "learning_rate": 1.9003326071483452e-06, "loss": 0.368, "step": 14149 }, { "epoch": 0.7157671101219081, "grad_norm": 7.968512535095215, "learning_rate": 1.8997028723156924e-06, "loss": 0.4883, "step": 14150 }, { "epoch": 0.7158176943699732, "grad_norm": 7.005064964294434, "learning_rate": 1.8990732173702081e-06, "loss": 0.2988, "step": 14151 }, { "epoch": 0.7158682786180384, "grad_norm": 8.655496597290039, "learning_rate": 1.8984436423281156e-06, "loss": 0.4395, "step": 14152 }, { "epoch": 0.7159188628661035, "grad_norm": 6.518816947937012, "learning_rate": 1.8978141472056423e-06, "loss": 0.3567, "step": 14153 }, { "epoch": 0.7159694471141687, "grad_norm": 13.012141227722168, "learning_rate": 1.8971847320190035e-06, "loss": 0.4348, "step": 14154 }, { "epoch": 0.7160200313622338, "grad_norm": 6.859098434448242, "learning_rate": 1.8965553967844175e-06, "loss": 0.3682, "step": 14155 }, { "epoch": 0.7160706156102989, "grad_norm": 4.787881851196289, "learning_rate": 1.8959261415181053e-06, "loss": 0.433, "step": 14156 }, { "epoch": 0.7161211998583641, "grad_norm": 9.2479248046875, "learning_rate": 1.8952969662362787e-06, "loss": 0.3853, "step": 14157 }, { "epoch": 0.7161717841064292, "grad_norm": 8.31730842590332, "learning_rate": 1.894667870955148e-06, "loss": 0.3623, "step": 14158 }, { "epoch": 0.7162223683544944, "grad_norm": 7.5239481925964355, "learning_rate": 1.8940388556909278e-06, "loss": 0.3401, "step": 14159 }, { "epoch": 0.7162729526025595, "grad_norm": 17.20505142211914, "learning_rate": 1.8934099204598238e-06, "loss": 0.4904, "step": 14160 }, { "epoch": 0.7163235368506247, "grad_norm": 15.35977840423584, "learning_rate": 1.892781065278041e-06, "loss": 0.5776, "step": 14161 }, { "epoch": 0.7163741210986899, "grad_norm": 12.75378131866455, "learning_rate": 1.8921522901617867e-06, "loss": 0.6147, "step": 14162 }, { "epoch": 0.7164247053467551, "grad_norm": 9.79680061340332, "learning_rate": 1.8915235951272616e-06, "loss": 0.4302, "step": 14163 }, { "epoch": 0.7164752895948202, "grad_norm": 8.40662956237793, "learning_rate": 1.8908949801906629e-06, "loss": 0.4391, "step": 14164 }, { "epoch": 0.7165258738428854, "grad_norm": 19.146137237548828, "learning_rate": 1.8902664453681924e-06, "loss": 0.4415, "step": 14165 }, { "epoch": 0.7165764580909505, "grad_norm": 13.67324161529541, "learning_rate": 1.8896379906760448e-06, "loss": 0.4415, "step": 14166 }, { "epoch": 0.7166270423390156, "grad_norm": 16.67983055114746, "learning_rate": 1.8890096161304134e-06, "loss": 0.5898, "step": 14167 }, { "epoch": 0.7166776265870808, "grad_norm": 7.3486785888671875, "learning_rate": 1.88838132174749e-06, "loss": 0.3983, "step": 14168 }, { "epoch": 0.7167282108351459, "grad_norm": 8.191642761230469, "learning_rate": 1.8877531075434618e-06, "loss": 0.3707, "step": 14169 }, { "epoch": 0.7167787950832111, "grad_norm": 4.966704845428467, "learning_rate": 1.887124973534521e-06, "loss": 0.3905, "step": 14170 }, { "epoch": 0.7168293793312762, "grad_norm": 6.541646480560303, "learning_rate": 1.8864969197368505e-06, "loss": 0.3185, "step": 14171 }, { "epoch": 0.7168799635793414, "grad_norm": 7.330568313598633, "learning_rate": 1.885868946166634e-06, "loss": 0.4001, "step": 14172 }, { "epoch": 0.7169305478274065, "grad_norm": 7.5562357902526855, "learning_rate": 1.8852410528400517e-06, "loss": 0.4623, "step": 14173 }, { "epoch": 0.7169811320754716, "grad_norm": 8.677567481994629, "learning_rate": 1.8846132397732852e-06, "loss": 0.453, "step": 14174 }, { "epoch": 0.7170317163235369, "grad_norm": 7.228092670440674, "learning_rate": 1.8839855069825108e-06, "loss": 0.4175, "step": 14175 }, { "epoch": 0.717082300571602, "grad_norm": 6.967248916625977, "learning_rate": 1.8833578544839015e-06, "loss": 0.319, "step": 14176 }, { "epoch": 0.7171328848196672, "grad_norm": 10.427350044250488, "learning_rate": 1.8827302822936345e-06, "loss": 0.4945, "step": 14177 }, { "epoch": 0.7171834690677323, "grad_norm": 6.082094192504883, "learning_rate": 1.8821027904278782e-06, "loss": 0.4012, "step": 14178 }, { "epoch": 0.7172340533157975, "grad_norm": 5.107120513916016, "learning_rate": 1.8814753789028024e-06, "loss": 0.4529, "step": 14179 }, { "epoch": 0.7172846375638626, "grad_norm": 5.974267959594727, "learning_rate": 1.8808480477345736e-06, "loss": 0.2954, "step": 14180 }, { "epoch": 0.7173352218119278, "grad_norm": 13.370941162109375, "learning_rate": 1.880220796939355e-06, "loss": 0.5392, "step": 14181 }, { "epoch": 0.7173858060599929, "grad_norm": 11.509696960449219, "learning_rate": 1.879593626533312e-06, "loss": 0.4067, "step": 14182 }, { "epoch": 0.7174363903080581, "grad_norm": 9.725523948669434, "learning_rate": 1.8789665365326049e-06, "loss": 0.4326, "step": 14183 }, { "epoch": 0.7174869745561232, "grad_norm": 5.619847297668457, "learning_rate": 1.8783395269533895e-06, "loss": 0.3322, "step": 14184 }, { "epoch": 0.7175375588041883, "grad_norm": 13.913416862487793, "learning_rate": 1.8777125978118261e-06, "loss": 0.4254, "step": 14185 }, { "epoch": 0.7175881430522535, "grad_norm": 8.238642692565918, "learning_rate": 1.877085749124068e-06, "loss": 0.429, "step": 14186 }, { "epoch": 0.7176387273003186, "grad_norm": 6.062273979187012, "learning_rate": 1.876458980906265e-06, "loss": 0.3905, "step": 14187 }, { "epoch": 0.7176893115483838, "grad_norm": 8.877370834350586, "learning_rate": 1.875832293174571e-06, "loss": 0.4542, "step": 14188 }, { "epoch": 0.717739895796449, "grad_norm": 8.640244483947754, "learning_rate": 1.875205685945133e-06, "loss": 0.484, "step": 14189 }, { "epoch": 0.7177904800445142, "grad_norm": 14.54221248626709, "learning_rate": 1.874579159234095e-06, "loss": 0.458, "step": 14190 }, { "epoch": 0.7178410642925793, "grad_norm": 12.557154655456543, "learning_rate": 1.8739527130576047e-06, "loss": 0.3372, "step": 14191 }, { "epoch": 0.7178916485406445, "grad_norm": 6.008732795715332, "learning_rate": 1.8733263474318037e-06, "loss": 0.3855, "step": 14192 }, { "epoch": 0.7179422327887096, "grad_norm": 11.641096115112305, "learning_rate": 1.8727000623728276e-06, "loss": 0.3962, "step": 14193 }, { "epoch": 0.7179928170367748, "grad_norm": 13.35881519317627, "learning_rate": 1.8720738578968183e-06, "loss": 0.5548, "step": 14194 }, { "epoch": 0.7180434012848399, "grad_norm": 6.985569000244141, "learning_rate": 1.8714477340199111e-06, "loss": 0.2691, "step": 14195 }, { "epoch": 0.718093985532905, "grad_norm": 7.812152862548828, "learning_rate": 1.870821690758237e-06, "loss": 0.3595, "step": 14196 }, { "epoch": 0.7181445697809702, "grad_norm": 8.555781364440918, "learning_rate": 1.8701957281279315e-06, "loss": 0.3817, "step": 14197 }, { "epoch": 0.7181951540290353, "grad_norm": 14.101346015930176, "learning_rate": 1.8695698461451227e-06, "loss": 0.4164, "step": 14198 }, { "epoch": 0.7182457382771005, "grad_norm": 15.982898712158203, "learning_rate": 1.8689440448259356e-06, "loss": 0.5357, "step": 14199 }, { "epoch": 0.7182963225251656, "grad_norm": 11.45044994354248, "learning_rate": 1.8683183241864999e-06, "loss": 0.5813, "step": 14200 }, { "epoch": 0.7183469067732308, "grad_norm": 6.395557880401611, "learning_rate": 1.867692684242936e-06, "loss": 0.3487, "step": 14201 }, { "epoch": 0.718397491021296, "grad_norm": 5.805137634277344, "learning_rate": 1.8670671250113648e-06, "loss": 0.2851, "step": 14202 }, { "epoch": 0.7184480752693612, "grad_norm": 8.360185623168945, "learning_rate": 1.8664416465079078e-06, "loss": 0.3803, "step": 14203 }, { "epoch": 0.7184986595174263, "grad_norm": 13.501962661743164, "learning_rate": 1.8658162487486808e-06, "loss": 0.4033, "step": 14204 }, { "epoch": 0.7185492437654915, "grad_norm": 4.464951515197754, "learning_rate": 1.865190931749799e-06, "loss": 0.3102, "step": 14205 }, { "epoch": 0.7185998280135566, "grad_norm": 10.179174423217773, "learning_rate": 1.864565695527375e-06, "loss": 0.4583, "step": 14206 }, { "epoch": 0.7186504122616217, "grad_norm": 7.1946587562561035, "learning_rate": 1.8639405400975197e-06, "loss": 0.4518, "step": 14207 }, { "epoch": 0.7187009965096869, "grad_norm": 9.492666244506836, "learning_rate": 1.8633154654763397e-06, "loss": 0.4885, "step": 14208 }, { "epoch": 0.718751580757752, "grad_norm": 6.118971824645996, "learning_rate": 1.862690471679946e-06, "loss": 0.4124, "step": 14209 }, { "epoch": 0.7188021650058172, "grad_norm": 5.753445148468018, "learning_rate": 1.8620655587244407e-06, "loss": 0.4009, "step": 14210 }, { "epoch": 0.7188527492538823, "grad_norm": 9.573990821838379, "learning_rate": 1.8614407266259254e-06, "loss": 0.4573, "step": 14211 }, { "epoch": 0.7189033335019475, "grad_norm": 6.372369766235352, "learning_rate": 1.8608159754005029e-06, "loss": 0.3369, "step": 14212 }, { "epoch": 0.7189539177500126, "grad_norm": 8.240056991577148, "learning_rate": 1.8601913050642706e-06, "loss": 0.3739, "step": 14213 }, { "epoch": 0.7190045019980777, "grad_norm": 10.0286865234375, "learning_rate": 1.8595667156333225e-06, "loss": 0.4332, "step": 14214 }, { "epoch": 0.719055086246143, "grad_norm": 7.018738269805908, "learning_rate": 1.8589422071237568e-06, "loss": 0.4282, "step": 14215 }, { "epoch": 0.7191056704942081, "grad_norm": 8.41097640991211, "learning_rate": 1.8583177795516633e-06, "loss": 0.3793, "step": 14216 }, { "epoch": 0.7191562547422733, "grad_norm": 9.917623519897461, "learning_rate": 1.857693432933132e-06, "loss": 0.3617, "step": 14217 }, { "epoch": 0.7192068389903384, "grad_norm": 5.6556243896484375, "learning_rate": 1.8570691672842512e-06, "loss": 0.3813, "step": 14218 }, { "epoch": 0.7192574232384036, "grad_norm": 6.661533832550049, "learning_rate": 1.8564449826211045e-06, "loss": 0.4055, "step": 14219 }, { "epoch": 0.7193080074864687, "grad_norm": 7.571382522583008, "learning_rate": 1.8558208789597798e-06, "loss": 0.3763, "step": 14220 }, { "epoch": 0.7193585917345339, "grad_norm": 6.364603519439697, "learning_rate": 1.8551968563163563e-06, "loss": 0.4233, "step": 14221 }, { "epoch": 0.719409175982599, "grad_norm": 7.025442123413086, "learning_rate": 1.8545729147069125e-06, "loss": 0.4474, "step": 14222 }, { "epoch": 0.7194597602306642, "grad_norm": 5.614529132843018, "learning_rate": 1.8539490541475286e-06, "loss": 0.3876, "step": 14223 }, { "epoch": 0.7195103444787293, "grad_norm": 11.991796493530273, "learning_rate": 1.8533252746542784e-06, "loss": 0.4911, "step": 14224 }, { "epoch": 0.7195609287267944, "grad_norm": 8.868677139282227, "learning_rate": 1.8527015762432343e-06, "loss": 0.393, "step": 14225 }, { "epoch": 0.7196115129748596, "grad_norm": 9.840702056884766, "learning_rate": 1.85207795893047e-06, "loss": 0.334, "step": 14226 }, { "epoch": 0.7196620972229247, "grad_norm": 10.362430572509766, "learning_rate": 1.851454422732053e-06, "loss": 0.4382, "step": 14227 }, { "epoch": 0.7197126814709899, "grad_norm": 7.356903076171875, "learning_rate": 1.850830967664049e-06, "loss": 0.3975, "step": 14228 }, { "epoch": 0.7197632657190551, "grad_norm": 8.382261276245117, "learning_rate": 1.8502075937425262e-06, "loss": 0.4332, "step": 14229 }, { "epoch": 0.7198138499671203, "grad_norm": 7.818394184112549, "learning_rate": 1.8495843009835474e-06, "loss": 0.4081, "step": 14230 }, { "epoch": 0.7198644342151854, "grad_norm": 12.213562965393066, "learning_rate": 1.8489610894031684e-06, "loss": 0.4386, "step": 14231 }, { "epoch": 0.7199150184632506, "grad_norm": 11.21574878692627, "learning_rate": 1.8483379590174528e-06, "loss": 0.5539, "step": 14232 }, { "epoch": 0.7199656027113157, "grad_norm": 6.887301921844482, "learning_rate": 1.8477149098424552e-06, "loss": 0.3479, "step": 14233 }, { "epoch": 0.7200161869593809, "grad_norm": 6.512619495391846, "learning_rate": 1.8470919418942285e-06, "loss": 0.4525, "step": 14234 }, { "epoch": 0.720066771207446, "grad_norm": 5.344446182250977, "learning_rate": 1.8464690551888287e-06, "loss": 0.348, "step": 14235 }, { "epoch": 0.7201173554555111, "grad_norm": 7.493354320526123, "learning_rate": 1.8458462497423047e-06, "loss": 0.355, "step": 14236 }, { "epoch": 0.7201679397035763, "grad_norm": 12.027066230773926, "learning_rate": 1.8452235255707025e-06, "loss": 0.4223, "step": 14237 }, { "epoch": 0.7202185239516414, "grad_norm": 7.830648899078369, "learning_rate": 1.8446008826900713e-06, "loss": 0.3474, "step": 14238 }, { "epoch": 0.7202691081997066, "grad_norm": 8.070343971252441, "learning_rate": 1.8439783211164542e-06, "loss": 0.5051, "step": 14239 }, { "epoch": 0.7203196924477717, "grad_norm": 5.7749810218811035, "learning_rate": 1.8433558408658908e-06, "loss": 0.3179, "step": 14240 }, { "epoch": 0.7203702766958369, "grad_norm": 6.111720561981201, "learning_rate": 1.8427334419544247e-06, "loss": 0.3487, "step": 14241 }, { "epoch": 0.7204208609439021, "grad_norm": 4.3531365394592285, "learning_rate": 1.8421111243980915e-06, "loss": 0.3875, "step": 14242 }, { "epoch": 0.7204714451919673, "grad_norm": 4.435239791870117, "learning_rate": 1.8414888882129274e-06, "loss": 0.3605, "step": 14243 }, { "epoch": 0.7205220294400324, "grad_norm": 7.778005599975586, "learning_rate": 1.840866733414965e-06, "loss": 0.446, "step": 14244 }, { "epoch": 0.7205726136880976, "grad_norm": 5.623983860015869, "learning_rate": 1.8402446600202367e-06, "loss": 0.351, "step": 14245 }, { "epoch": 0.7206231979361627, "grad_norm": 9.172778129577637, "learning_rate": 1.83962266804477e-06, "loss": 0.4011, "step": 14246 }, { "epoch": 0.7206737821842278, "grad_norm": 25.78107261657715, "learning_rate": 1.8390007575045949e-06, "loss": 0.4393, "step": 14247 }, { "epoch": 0.720724366432293, "grad_norm": 5.418166160583496, "learning_rate": 1.838378928415735e-06, "loss": 0.2923, "step": 14248 }, { "epoch": 0.7207749506803581, "grad_norm": 5.3978753089904785, "learning_rate": 1.8377571807942123e-06, "loss": 0.327, "step": 14249 }, { "epoch": 0.7208255349284233, "grad_norm": 7.0737714767456055, "learning_rate": 1.8371355146560503e-06, "loss": 0.3762, "step": 14250 }, { "epoch": 0.7208761191764884, "grad_norm": 8.716477394104004, "learning_rate": 1.8365139300172668e-06, "loss": 0.4072, "step": 14251 }, { "epoch": 0.7209267034245536, "grad_norm": 8.590581893920898, "learning_rate": 1.8358924268938761e-06, "loss": 0.3905, "step": 14252 }, { "epoch": 0.7209772876726187, "grad_norm": 4.71034574508667, "learning_rate": 1.835271005301897e-06, "loss": 0.3184, "step": 14253 }, { "epoch": 0.7210278719206838, "grad_norm": 13.02677059173584, "learning_rate": 1.8346496652573403e-06, "loss": 0.4421, "step": 14254 }, { "epoch": 0.7210784561687491, "grad_norm": 5.961181163787842, "learning_rate": 1.8340284067762136e-06, "loss": 0.4112, "step": 14255 }, { "epoch": 0.7211290404168142, "grad_norm": 7.676661014556885, "learning_rate": 1.833407229874532e-06, "loss": 0.4246, "step": 14256 }, { "epoch": 0.7211796246648794, "grad_norm": 5.1487579345703125, "learning_rate": 1.8327861345682934e-06, "loss": 0.3471, "step": 14257 }, { "epoch": 0.7212302089129445, "grad_norm": 8.460304260253906, "learning_rate": 1.8321651208735086e-06, "loss": 0.4607, "step": 14258 }, { "epoch": 0.7212807931610097, "grad_norm": 8.624923706054688, "learning_rate": 1.8315441888061763e-06, "loss": 0.4276, "step": 14259 }, { "epoch": 0.7213313774090748, "grad_norm": 6.029715061187744, "learning_rate": 1.8309233383822955e-06, "loss": 0.5183, "step": 14260 }, { "epoch": 0.72138196165714, "grad_norm": 9.319482803344727, "learning_rate": 1.8303025696178677e-06, "loss": 0.4807, "step": 14261 }, { "epoch": 0.7214325459052051, "grad_norm": 6.323192119598389, "learning_rate": 1.8296818825288865e-06, "loss": 0.3992, "step": 14262 }, { "epoch": 0.7214831301532703, "grad_norm": 17.624971389770508, "learning_rate": 1.8290612771313437e-06, "loss": 0.6378, "step": 14263 }, { "epoch": 0.7215337144013354, "grad_norm": 33.36893844604492, "learning_rate": 1.8284407534412346e-06, "loss": 0.5992, "step": 14264 }, { "epoch": 0.7215842986494005, "grad_norm": 7.998394012451172, "learning_rate": 1.8278203114745468e-06, "loss": 0.4525, "step": 14265 }, { "epoch": 0.7216348828974657, "grad_norm": 13.200952529907227, "learning_rate": 1.8271999512472654e-06, "loss": 0.4643, "step": 14266 }, { "epoch": 0.7216854671455308, "grad_norm": 8.494025230407715, "learning_rate": 1.8265796727753799e-06, "loss": 0.3809, "step": 14267 }, { "epoch": 0.721736051393596, "grad_norm": 8.544189453125, "learning_rate": 1.8259594760748707e-06, "loss": 0.443, "step": 14268 }, { "epoch": 0.7217866356416612, "grad_norm": 8.990434646606445, "learning_rate": 1.8253393611617192e-06, "loss": 0.4806, "step": 14269 }, { "epoch": 0.7218372198897264, "grad_norm": 15.21423625946045, "learning_rate": 1.8247193280519048e-06, "loss": 0.4625, "step": 14270 }, { "epoch": 0.7218878041377915, "grad_norm": 7.231312274932861, "learning_rate": 1.8240993767614035e-06, "loss": 0.4519, "step": 14271 }, { "epoch": 0.7219383883858567, "grad_norm": 14.535001754760742, "learning_rate": 1.8234795073061883e-06, "loss": 0.5043, "step": 14272 }, { "epoch": 0.7219889726339218, "grad_norm": 9.294251441955566, "learning_rate": 1.8228597197022353e-06, "loss": 0.4546, "step": 14273 }, { "epoch": 0.722039556881987, "grad_norm": 12.168302536010742, "learning_rate": 1.8222400139655133e-06, "loss": 0.4092, "step": 14274 }, { "epoch": 0.7220901411300521, "grad_norm": 5.635855197906494, "learning_rate": 1.8216203901119889e-06, "loss": 0.3577, "step": 14275 }, { "epoch": 0.7221407253781172, "grad_norm": 5.321691989898682, "learning_rate": 1.821000848157632e-06, "loss": 0.3288, "step": 14276 }, { "epoch": 0.7221913096261824, "grad_norm": 9.35621452331543, "learning_rate": 1.820381388118404e-06, "loss": 0.3405, "step": 14277 }, { "epoch": 0.7222418938742475, "grad_norm": 6.733516693115234, "learning_rate": 1.8197620100102664e-06, "loss": 0.3562, "step": 14278 }, { "epoch": 0.7222924781223127, "grad_norm": 11.548593521118164, "learning_rate": 1.8191427138491818e-06, "loss": 0.478, "step": 14279 }, { "epoch": 0.7223430623703778, "grad_norm": 8.551168441772461, "learning_rate": 1.8185234996511064e-06, "loss": 0.3741, "step": 14280 }, { "epoch": 0.722393646618443, "grad_norm": 8.119057655334473, "learning_rate": 1.8179043674319947e-06, "loss": 0.4063, "step": 14281 }, { "epoch": 0.7224442308665082, "grad_norm": 7.662750244140625, "learning_rate": 1.8172853172078042e-06, "loss": 0.4291, "step": 14282 }, { "epoch": 0.7224948151145734, "grad_norm": 7.468766689300537, "learning_rate": 1.816666348994482e-06, "loss": 0.2972, "step": 14283 }, { "epoch": 0.7225453993626385, "grad_norm": 8.250276565551758, "learning_rate": 1.8160474628079772e-06, "loss": 0.3645, "step": 14284 }, { "epoch": 0.7225959836107037, "grad_norm": 5.436610698699951, "learning_rate": 1.8154286586642405e-06, "loss": 0.3216, "step": 14285 }, { "epoch": 0.7226465678587688, "grad_norm": 15.426586151123047, "learning_rate": 1.8148099365792155e-06, "loss": 0.5007, "step": 14286 }, { "epoch": 0.7226971521068339, "grad_norm": 12.288186073303223, "learning_rate": 1.8141912965688425e-06, "loss": 0.4323, "step": 14287 }, { "epoch": 0.7227477363548991, "grad_norm": 7.019211769104004, "learning_rate": 1.813572738649067e-06, "loss": 0.3241, "step": 14288 }, { "epoch": 0.7227983206029642, "grad_norm": 4.644792079925537, "learning_rate": 1.8129542628358254e-06, "loss": 0.3198, "step": 14289 }, { "epoch": 0.7228489048510294, "grad_norm": 21.440271377563477, "learning_rate": 1.8123358691450533e-06, "loss": 0.405, "step": 14290 }, { "epoch": 0.7228994890990945, "grad_norm": 7.039101600646973, "learning_rate": 1.8117175575926876e-06, "loss": 0.3763, "step": 14291 }, { "epoch": 0.7229500733471597, "grad_norm": 14.579792022705078, "learning_rate": 1.8110993281946592e-06, "loss": 0.5237, "step": 14292 }, { "epoch": 0.7230006575952248, "grad_norm": 7.495621204376221, "learning_rate": 1.810481180966897e-06, "loss": 0.4096, "step": 14293 }, { "epoch": 0.72305124184329, "grad_norm": 10.72490406036377, "learning_rate": 1.8098631159253326e-06, "loss": 0.3612, "step": 14294 }, { "epoch": 0.7231018260913552, "grad_norm": 6.810201644897461, "learning_rate": 1.8092451330858907e-06, "loss": 0.3904, "step": 14295 }, { "epoch": 0.7231524103394203, "grad_norm": 5.64500093460083, "learning_rate": 1.8086272324644943e-06, "loss": 0.4438, "step": 14296 }, { "epoch": 0.7232029945874855, "grad_norm": 9.875439643859863, "learning_rate": 1.808009414077066e-06, "loss": 0.4419, "step": 14297 }, { "epoch": 0.7232535788355506, "grad_norm": 6.578048229217529, "learning_rate": 1.8073916779395228e-06, "loss": 0.4202, "step": 14298 }, { "epoch": 0.7233041630836158, "grad_norm": 9.423917770385742, "learning_rate": 1.8067740240677872e-06, "loss": 0.4231, "step": 14299 }, { "epoch": 0.7233547473316809, "grad_norm": 9.987892150878906, "learning_rate": 1.8061564524777714e-06, "loss": 0.3372, "step": 14300 }, { "epoch": 0.7234053315797461, "grad_norm": 6.0948486328125, "learning_rate": 1.805538963185388e-06, "loss": 0.3522, "step": 14301 }, { "epoch": 0.7234559158278112, "grad_norm": 7.881754398345947, "learning_rate": 1.8049215562065514e-06, "loss": 0.3462, "step": 14302 }, { "epoch": 0.7235065000758764, "grad_norm": 7.556715488433838, "learning_rate": 1.804304231557169e-06, "loss": 0.3721, "step": 14303 }, { "epoch": 0.7235570843239415, "grad_norm": 14.004087448120117, "learning_rate": 1.8036869892531457e-06, "loss": 0.4921, "step": 14304 }, { "epoch": 0.7236076685720066, "grad_norm": 9.149518013000488, "learning_rate": 1.8030698293103898e-06, "loss": 0.3828, "step": 14305 }, { "epoch": 0.7236582528200718, "grad_norm": 13.144356727600098, "learning_rate": 1.8024527517448026e-06, "loss": 0.3948, "step": 14306 }, { "epoch": 0.7237088370681369, "grad_norm": 6.462482452392578, "learning_rate": 1.8018357565722848e-06, "loss": 0.3821, "step": 14307 }, { "epoch": 0.7237594213162022, "grad_norm": 5.943284034729004, "learning_rate": 1.8012188438087353e-06, "loss": 0.2596, "step": 14308 }, { "epoch": 0.7238100055642673, "grad_norm": 5.789168834686279, "learning_rate": 1.8006020134700492e-06, "loss": 0.3823, "step": 14309 }, { "epoch": 0.7238605898123325, "grad_norm": 10.752177238464355, "learning_rate": 1.7999852655721194e-06, "loss": 0.501, "step": 14310 }, { "epoch": 0.7239111740603976, "grad_norm": 6.419435977935791, "learning_rate": 1.799368600130842e-06, "loss": 0.3856, "step": 14311 }, { "epoch": 0.7239617583084628, "grad_norm": 5.493288993835449, "learning_rate": 1.7987520171621053e-06, "loss": 0.3328, "step": 14312 }, { "epoch": 0.7240123425565279, "grad_norm": 21.147499084472656, "learning_rate": 1.7981355166817949e-06, "loss": 0.4413, "step": 14313 }, { "epoch": 0.724062926804593, "grad_norm": 5.334944248199463, "learning_rate": 1.7975190987058e-06, "loss": 0.3545, "step": 14314 }, { "epoch": 0.7241135110526582, "grad_norm": 11.476444244384766, "learning_rate": 1.7969027632500025e-06, "loss": 0.4245, "step": 14315 }, { "epoch": 0.7241640953007233, "grad_norm": 5.4080047607421875, "learning_rate": 1.796286510330283e-06, "loss": 0.3843, "step": 14316 }, { "epoch": 0.7242146795487885, "grad_norm": 6.172294616699219, "learning_rate": 1.795670339962523e-06, "loss": 0.3612, "step": 14317 }, { "epoch": 0.7242652637968536, "grad_norm": 9.287345886230469, "learning_rate": 1.7950542521625986e-06, "loss": 0.3696, "step": 14318 }, { "epoch": 0.7243158480449188, "grad_norm": 9.54747200012207, "learning_rate": 1.7944382469463833e-06, "loss": 0.3674, "step": 14319 }, { "epoch": 0.7243664322929839, "grad_norm": 8.677617073059082, "learning_rate": 1.7938223243297548e-06, "loss": 0.3721, "step": 14320 }, { "epoch": 0.7244170165410491, "grad_norm": 6.355371952056885, "learning_rate": 1.7932064843285796e-06, "loss": 0.4365, "step": 14321 }, { "epoch": 0.7244676007891143, "grad_norm": 9.528072357177734, "learning_rate": 1.7925907269587257e-06, "loss": 0.4899, "step": 14322 }, { "epoch": 0.7245181850371795, "grad_norm": 6.88436222076416, "learning_rate": 1.7919750522360635e-06, "loss": 0.3858, "step": 14323 }, { "epoch": 0.7245687692852446, "grad_norm": 11.00652027130127, "learning_rate": 1.791359460176455e-06, "loss": 0.469, "step": 14324 }, { "epoch": 0.7246193535333098, "grad_norm": 6.100318431854248, "learning_rate": 1.7907439507957619e-06, "loss": 0.4561, "step": 14325 }, { "epoch": 0.7246699377813749, "grad_norm": 6.497692584991455, "learning_rate": 1.790128524109847e-06, "loss": 0.2608, "step": 14326 }, { "epoch": 0.72472052202944, "grad_norm": 7.1219024658203125, "learning_rate": 1.7895131801345666e-06, "loss": 0.4219, "step": 14327 }, { "epoch": 0.7247711062775052, "grad_norm": 7.09271764755249, "learning_rate": 1.7888979188857758e-06, "loss": 0.363, "step": 14328 }, { "epoch": 0.7248216905255703, "grad_norm": 10.510546684265137, "learning_rate": 1.7882827403793307e-06, "loss": 0.3807, "step": 14329 }, { "epoch": 0.7248722747736355, "grad_norm": 7.922118663787842, "learning_rate": 1.787667644631082e-06, "loss": 0.3921, "step": 14330 }, { "epoch": 0.7249228590217006, "grad_norm": 11.176143646240234, "learning_rate": 1.7870526316568777e-06, "loss": 0.4613, "step": 14331 }, { "epoch": 0.7249734432697658, "grad_norm": 7.133904457092285, "learning_rate": 1.7864377014725682e-06, "loss": 0.4286, "step": 14332 }, { "epoch": 0.7250240275178309, "grad_norm": 6.005550384521484, "learning_rate": 1.7858228540939965e-06, "loss": 0.3106, "step": 14333 }, { "epoch": 0.725074611765896, "grad_norm": 10.47003173828125, "learning_rate": 1.7852080895370067e-06, "loss": 0.3176, "step": 14334 }, { "epoch": 0.7251251960139613, "grad_norm": 8.551270484924316, "learning_rate": 1.7845934078174392e-06, "loss": 0.3304, "step": 14335 }, { "epoch": 0.7251757802620264, "grad_norm": 14.526080131530762, "learning_rate": 1.7839788089511318e-06, "loss": 0.5769, "step": 14336 }, { "epoch": 0.7252263645100916, "grad_norm": 9.704232215881348, "learning_rate": 1.783364292953924e-06, "loss": 0.4011, "step": 14337 }, { "epoch": 0.7252769487581567, "grad_norm": 9.434101104736328, "learning_rate": 1.7827498598416488e-06, "loss": 0.4167, "step": 14338 }, { "epoch": 0.7253275330062219, "grad_norm": 8.037201881408691, "learning_rate": 1.7821355096301368e-06, "loss": 0.3357, "step": 14339 }, { "epoch": 0.725378117254287, "grad_norm": 11.864997863769531, "learning_rate": 1.7815212423352224e-06, "loss": 0.3377, "step": 14340 }, { "epoch": 0.7254287015023522, "grad_norm": 4.973169803619385, "learning_rate": 1.7809070579727316e-06, "loss": 0.3392, "step": 14341 }, { "epoch": 0.7254792857504173, "grad_norm": 5.451945781707764, "learning_rate": 1.780292956558491e-06, "loss": 0.3908, "step": 14342 }, { "epoch": 0.7255298699984825, "grad_norm": 7.99407434463501, "learning_rate": 1.7796789381083212e-06, "loss": 0.3516, "step": 14343 }, { "epoch": 0.7255804542465476, "grad_norm": 10.111175537109375, "learning_rate": 1.7790650026380495e-06, "loss": 0.411, "step": 14344 }, { "epoch": 0.7256310384946127, "grad_norm": 7.538082122802734, "learning_rate": 1.7784511501634922e-06, "loss": 0.46, "step": 14345 }, { "epoch": 0.7256816227426779, "grad_norm": 8.353706359863281, "learning_rate": 1.7778373807004679e-06, "loss": 0.3044, "step": 14346 }, { "epoch": 0.725732206990743, "grad_norm": 7.89246940612793, "learning_rate": 1.7772236942647913e-06, "loss": 0.3105, "step": 14347 }, { "epoch": 0.7257827912388083, "grad_norm": 15.782793045043945, "learning_rate": 1.776610090872274e-06, "loss": 0.5241, "step": 14348 }, { "epoch": 0.7258333754868734, "grad_norm": 5.258164405822754, "learning_rate": 1.7759965705387306e-06, "loss": 0.4013, "step": 14349 }, { "epoch": 0.7258839597349386, "grad_norm": 7.588324546813965, "learning_rate": 1.7753831332799682e-06, "loss": 0.4812, "step": 14350 }, { "epoch": 0.7259345439830037, "grad_norm": 13.773008346557617, "learning_rate": 1.7747697791117924e-06, "loss": 0.4048, "step": 14351 }, { "epoch": 0.7259851282310689, "grad_norm": 6.250624179840088, "learning_rate": 1.7741565080500106e-06, "loss": 0.4054, "step": 14352 }, { "epoch": 0.726035712479134, "grad_norm": 7.966902732849121, "learning_rate": 1.7735433201104241e-06, "loss": 0.4548, "step": 14353 }, { "epoch": 0.7260862967271992, "grad_norm": 9.48215389251709, "learning_rate": 1.7729302153088312e-06, "loss": 0.3897, "step": 14354 }, { "epoch": 0.7261368809752643, "grad_norm": 8.063735008239746, "learning_rate": 1.7723171936610334e-06, "loss": 0.3869, "step": 14355 }, { "epoch": 0.7261874652233294, "grad_norm": 6.859611988067627, "learning_rate": 1.7717042551828257e-06, "loss": 0.3163, "step": 14356 }, { "epoch": 0.7262380494713946, "grad_norm": 8.262529373168945, "learning_rate": 1.7710913998899999e-06, "loss": 0.3747, "step": 14357 }, { "epoch": 0.7262886337194597, "grad_norm": 7.646380424499512, "learning_rate": 1.7704786277983515e-06, "loss": 0.4441, "step": 14358 }, { "epoch": 0.7263392179675249, "grad_norm": 18.439472198486328, "learning_rate": 1.76986593892367e-06, "loss": 0.5721, "step": 14359 }, { "epoch": 0.72638980221559, "grad_norm": 9.655396461486816, "learning_rate": 1.7692533332817375e-06, "loss": 0.342, "step": 14360 }, { "epoch": 0.7264403864636552, "grad_norm": 5.23211145401001, "learning_rate": 1.7686408108883446e-06, "loss": 0.3698, "step": 14361 }, { "epoch": 0.7264909707117204, "grad_norm": 10.198629379272461, "learning_rate": 1.7680283717592733e-06, "loss": 0.3897, "step": 14362 }, { "epoch": 0.7265415549597856, "grad_norm": 21.925180435180664, "learning_rate": 1.7674160159103027e-06, "loss": 0.669, "step": 14363 }, { "epoch": 0.7265921392078507, "grad_norm": 7.707814693450928, "learning_rate": 1.7668037433572149e-06, "loss": 0.3923, "step": 14364 }, { "epoch": 0.7266427234559159, "grad_norm": 7.022498607635498, "learning_rate": 1.7661915541157854e-06, "loss": 0.4809, "step": 14365 }, { "epoch": 0.726693307703981, "grad_norm": 6.904867649078369, "learning_rate": 1.7655794482017868e-06, "loss": 0.3858, "step": 14366 }, { "epoch": 0.7267438919520461, "grad_norm": 7.263354778289795, "learning_rate": 1.764967425630995e-06, "loss": 0.3656, "step": 14367 }, { "epoch": 0.7267944762001113, "grad_norm": 11.737849235534668, "learning_rate": 1.7643554864191787e-06, "loss": 0.4553, "step": 14368 }, { "epoch": 0.7268450604481764, "grad_norm": 8.237850189208984, "learning_rate": 1.7637436305821044e-06, "loss": 0.4182, "step": 14369 }, { "epoch": 0.7268956446962416, "grad_norm": 7.865851402282715, "learning_rate": 1.7631318581355416e-06, "loss": 0.3344, "step": 14370 }, { "epoch": 0.7269462289443067, "grad_norm": 6.900636672973633, "learning_rate": 1.7625201690952532e-06, "loss": 0.4198, "step": 14371 }, { "epoch": 0.7269968131923719, "grad_norm": 15.814957618713379, "learning_rate": 1.7619085634769995e-06, "loss": 0.3959, "step": 14372 }, { "epoch": 0.727047397440437, "grad_norm": 6.5815582275390625, "learning_rate": 1.7612970412965409e-06, "loss": 0.3677, "step": 14373 }, { "epoch": 0.7270979816885021, "grad_norm": 5.392838001251221, "learning_rate": 1.7606856025696328e-06, "loss": 0.3819, "step": 14374 }, { "epoch": 0.7271485659365674, "grad_norm": 10.73840045928955, "learning_rate": 1.7600742473120346e-06, "loss": 0.41, "step": 14375 }, { "epoch": 0.7271991501846325, "grad_norm": 6.911414623260498, "learning_rate": 1.759462975539497e-06, "loss": 0.4782, "step": 14376 }, { "epoch": 0.7272497344326977, "grad_norm": 7.828457355499268, "learning_rate": 1.7588517872677701e-06, "loss": 0.4048, "step": 14377 }, { "epoch": 0.7273003186807628, "grad_norm": 5.911346435546875, "learning_rate": 1.7582406825126052e-06, "loss": 0.4334, "step": 14378 }, { "epoch": 0.727350902928828, "grad_norm": 8.512442588806152, "learning_rate": 1.7576296612897482e-06, "loss": 0.4183, "step": 14379 }, { "epoch": 0.7274014871768931, "grad_norm": 10.0623197555542, "learning_rate": 1.7570187236149433e-06, "loss": 0.3606, "step": 14380 }, { "epoch": 0.7274520714249583, "grad_norm": 6.433591365814209, "learning_rate": 1.7564078695039305e-06, "loss": 0.3955, "step": 14381 }, { "epoch": 0.7275026556730234, "grad_norm": 9.378649711608887, "learning_rate": 1.7557970989724542e-06, "loss": 0.3421, "step": 14382 }, { "epoch": 0.7275532399210886, "grad_norm": 9.040247917175293, "learning_rate": 1.755186412036251e-06, "loss": 0.3789, "step": 14383 }, { "epoch": 0.7276038241691537, "grad_norm": 5.143237113952637, "learning_rate": 1.7545758087110543e-06, "loss": 0.4107, "step": 14384 }, { "epoch": 0.7276544084172188, "grad_norm": 10.68850326538086, "learning_rate": 1.7539652890126035e-06, "loss": 0.4594, "step": 14385 }, { "epoch": 0.727704992665284, "grad_norm": 13.478657722473145, "learning_rate": 1.7533548529566235e-06, "loss": 0.4691, "step": 14386 }, { "epoch": 0.7277555769133491, "grad_norm": 4.903193950653076, "learning_rate": 1.7527445005588489e-06, "loss": 0.3855, "step": 14387 }, { "epoch": 0.7278061611614144, "grad_norm": 11.62090015411377, "learning_rate": 1.7521342318350043e-06, "loss": 0.403, "step": 14388 }, { "epoch": 0.7278567454094795, "grad_norm": 9.033814430236816, "learning_rate": 1.7515240468008149e-06, "loss": 0.4036, "step": 14389 }, { "epoch": 0.7279073296575447, "grad_norm": 13.618182182312012, "learning_rate": 1.7509139454720054e-06, "loss": 0.419, "step": 14390 }, { "epoch": 0.7279579139056098, "grad_norm": 11.450016021728516, "learning_rate": 1.7503039278642959e-06, "loss": 0.4392, "step": 14391 }, { "epoch": 0.728008498153675, "grad_norm": 11.119841575622559, "learning_rate": 1.7496939939934033e-06, "loss": 0.4373, "step": 14392 }, { "epoch": 0.7280590824017401, "grad_norm": 8.50649356842041, "learning_rate": 1.7490841438750472e-06, "loss": 0.4966, "step": 14393 }, { "epoch": 0.7281096666498053, "grad_norm": 7.919085502624512, "learning_rate": 1.7484743775249408e-06, "loss": 0.4151, "step": 14394 }, { "epoch": 0.7281602508978704, "grad_norm": 8.799962997436523, "learning_rate": 1.7478646949587941e-06, "loss": 0.3055, "step": 14395 }, { "epoch": 0.7282108351459355, "grad_norm": 12.108227729797363, "learning_rate": 1.7472550961923202e-06, "loss": 0.4697, "step": 14396 }, { "epoch": 0.7282614193940007, "grad_norm": 11.14084529876709, "learning_rate": 1.746645581241228e-06, "loss": 0.4744, "step": 14397 }, { "epoch": 0.7283120036420658, "grad_norm": 9.520276069641113, "learning_rate": 1.7460361501212176e-06, "loss": 0.3508, "step": 14398 }, { "epoch": 0.728362587890131, "grad_norm": 10.86020565032959, "learning_rate": 1.7454268028479976e-06, "loss": 0.4669, "step": 14399 }, { "epoch": 0.7284131721381961, "grad_norm": 9.919413566589355, "learning_rate": 1.7448175394372674e-06, "loss": 0.5072, "step": 14400 }, { "epoch": 0.7284637563862613, "grad_norm": 8.221639633178711, "learning_rate": 1.7442083599047255e-06, "loss": 0.3604, "step": 14401 }, { "epoch": 0.7285143406343265, "grad_norm": 9.42409896850586, "learning_rate": 1.7435992642660715e-06, "loss": 0.4453, "step": 14402 }, { "epoch": 0.7285649248823917, "grad_norm": 10.356549263000488, "learning_rate": 1.742990252536999e-06, "loss": 0.493, "step": 14403 }, { "epoch": 0.7286155091304568, "grad_norm": 12.591512680053711, "learning_rate": 1.7423813247331994e-06, "loss": 0.3652, "step": 14404 }, { "epoch": 0.728666093378522, "grad_norm": 8.792369842529297, "learning_rate": 1.7417724808703663e-06, "loss": 0.3804, "step": 14405 }, { "epoch": 0.7287166776265871, "grad_norm": 13.252952575683594, "learning_rate": 1.7411637209641857e-06, "loss": 0.4962, "step": 14406 }, { "epoch": 0.7287672618746522, "grad_norm": 5.1227498054504395, "learning_rate": 1.7405550450303432e-06, "loss": 0.3819, "step": 14407 }, { "epoch": 0.7288178461227174, "grad_norm": 6.394099712371826, "learning_rate": 1.7399464530845262e-06, "loss": 0.3734, "step": 14408 }, { "epoch": 0.7288684303707825, "grad_norm": 7.3249053955078125, "learning_rate": 1.7393379451424147e-06, "loss": 0.3593, "step": 14409 }, { "epoch": 0.7289190146188477, "grad_norm": 8.25605583190918, "learning_rate": 1.7387295212196886e-06, "loss": 0.2983, "step": 14410 }, { "epoch": 0.7289695988669128, "grad_norm": 7.761007785797119, "learning_rate": 1.738121181332026e-06, "loss": 0.3751, "step": 14411 }, { "epoch": 0.729020183114978, "grad_norm": 5.227047920227051, "learning_rate": 1.7375129254950989e-06, "loss": 0.3537, "step": 14412 }, { "epoch": 0.7290707673630431, "grad_norm": 6.892746448516846, "learning_rate": 1.736904753724586e-06, "loss": 0.369, "step": 14413 }, { "epoch": 0.7291213516111082, "grad_norm": 5.895086765289307, "learning_rate": 1.7362966660361558e-06, "loss": 0.3119, "step": 14414 }, { "epoch": 0.7291719358591735, "grad_norm": 7.278832912445068, "learning_rate": 1.7356886624454772e-06, "loss": 0.4619, "step": 14415 }, { "epoch": 0.7292225201072386, "grad_norm": 7.212531089782715, "learning_rate": 1.735080742968216e-06, "loss": 0.2767, "step": 14416 }, { "epoch": 0.7292731043553038, "grad_norm": 13.406872749328613, "learning_rate": 1.7344729076200389e-06, "loss": 0.4851, "step": 14417 }, { "epoch": 0.7293236886033689, "grad_norm": 7.583347797393799, "learning_rate": 1.7338651564166081e-06, "loss": 0.3572, "step": 14418 }, { "epoch": 0.7293742728514341, "grad_norm": 5.412703514099121, "learning_rate": 1.733257489373581e-06, "loss": 0.3445, "step": 14419 }, { "epoch": 0.7294248570994992, "grad_norm": 6.879626750946045, "learning_rate": 1.7326499065066199e-06, "loss": 0.3688, "step": 14420 }, { "epoch": 0.7294754413475644, "grad_norm": 14.830973625183105, "learning_rate": 1.7320424078313786e-06, "loss": 0.4934, "step": 14421 }, { "epoch": 0.7295260255956295, "grad_norm": 20.4927978515625, "learning_rate": 1.7314349933635093e-06, "loss": 0.4358, "step": 14422 }, { "epoch": 0.7295766098436947, "grad_norm": 7.105311393737793, "learning_rate": 1.7308276631186688e-06, "loss": 0.3301, "step": 14423 }, { "epoch": 0.7296271940917598, "grad_norm": 7.550529956817627, "learning_rate": 1.7302204171124991e-06, "loss": 0.309, "step": 14424 }, { "epoch": 0.729677778339825, "grad_norm": 9.525511741638184, "learning_rate": 1.7296132553606538e-06, "loss": 0.424, "step": 14425 }, { "epoch": 0.7297283625878901, "grad_norm": 6.063149929046631, "learning_rate": 1.7290061778787752e-06, "loss": 0.3494, "step": 14426 }, { "epoch": 0.7297789468359552, "grad_norm": 11.56153678894043, "learning_rate": 1.7283991846825055e-06, "loss": 0.4103, "step": 14427 }, { "epoch": 0.7298295310840205, "grad_norm": 8.750285148620605, "learning_rate": 1.7277922757874883e-06, "loss": 0.4219, "step": 14428 }, { "epoch": 0.7298801153320856, "grad_norm": 7.1536030769348145, "learning_rate": 1.72718545120936e-06, "loss": 0.3034, "step": 14429 }, { "epoch": 0.7299306995801508, "grad_norm": 14.712141990661621, "learning_rate": 1.7265787109637555e-06, "loss": 0.3884, "step": 14430 }, { "epoch": 0.7299812838282159, "grad_norm": 5.72285270690918, "learning_rate": 1.7259720550663134e-06, "loss": 0.392, "step": 14431 }, { "epoch": 0.7300318680762811, "grad_norm": 9.466344833374023, "learning_rate": 1.7253654835326627e-06, "loss": 0.4514, "step": 14432 }, { "epoch": 0.7300824523243462, "grad_norm": 6.969888687133789, "learning_rate": 1.7247589963784328e-06, "loss": 0.4242, "step": 14433 }, { "epoch": 0.7301330365724114, "grad_norm": 7.7284650802612305, "learning_rate": 1.7241525936192537e-06, "loss": 0.5474, "step": 14434 }, { "epoch": 0.7301836208204765, "grad_norm": 8.263630867004395, "learning_rate": 1.7235462752707504e-06, "loss": 0.3115, "step": 14435 }, { "epoch": 0.7302342050685416, "grad_norm": 7.368926525115967, "learning_rate": 1.7229400413485453e-06, "loss": 0.3244, "step": 14436 }, { "epoch": 0.7302847893166068, "grad_norm": 9.369369506835938, "learning_rate": 1.7223338918682596e-06, "loss": 0.4168, "step": 14437 }, { "epoch": 0.7303353735646719, "grad_norm": 8.230803489685059, "learning_rate": 1.7217278268455135e-06, "loss": 0.2893, "step": 14438 }, { "epoch": 0.7303859578127371, "grad_norm": 9.408367156982422, "learning_rate": 1.7211218462959207e-06, "loss": 0.3002, "step": 14439 }, { "epoch": 0.7304365420608022, "grad_norm": 5.917478084564209, "learning_rate": 1.7205159502350994e-06, "loss": 0.3114, "step": 14440 }, { "epoch": 0.7304871263088675, "grad_norm": 4.804558277130127, "learning_rate": 1.7199101386786616e-06, "loss": 0.2683, "step": 14441 }, { "epoch": 0.7305377105569326, "grad_norm": 13.947741508483887, "learning_rate": 1.7193044116422147e-06, "loss": 0.4676, "step": 14442 }, { "epoch": 0.7305882948049978, "grad_norm": 19.144269943237305, "learning_rate": 1.7186987691413703e-06, "loss": 0.5338, "step": 14443 }, { "epoch": 0.7306388790530629, "grad_norm": 11.11783218383789, "learning_rate": 1.7180932111917337e-06, "loss": 0.3255, "step": 14444 }, { "epoch": 0.730689463301128, "grad_norm": 5.808338165283203, "learning_rate": 1.7174877378089056e-06, "loss": 0.3277, "step": 14445 }, { "epoch": 0.7307400475491932, "grad_norm": 9.655485153198242, "learning_rate": 1.7168823490084918e-06, "loss": 0.5108, "step": 14446 }, { "epoch": 0.7307906317972583, "grad_norm": 6.854563236236572, "learning_rate": 1.71627704480609e-06, "loss": 0.4009, "step": 14447 }, { "epoch": 0.7308412160453235, "grad_norm": 10.879353523254395, "learning_rate": 1.7156718252172954e-06, "loss": 0.4256, "step": 14448 }, { "epoch": 0.7308918002933886, "grad_norm": 8.134496688842773, "learning_rate": 1.7150666902577085e-06, "loss": 0.3805, "step": 14449 }, { "epoch": 0.7309423845414538, "grad_norm": 6.695046901702881, "learning_rate": 1.7144616399429149e-06, "loss": 0.3241, "step": 14450 }, { "epoch": 0.7309929687895189, "grad_norm": 6.131373882293701, "learning_rate": 1.7138566742885104e-06, "loss": 0.3712, "step": 14451 }, { "epoch": 0.7310435530375841, "grad_norm": 7.170956611633301, "learning_rate": 1.7132517933100824e-06, "loss": 0.3284, "step": 14452 }, { "epoch": 0.7310941372856492, "grad_norm": 8.116169929504395, "learning_rate": 1.7126469970232168e-06, "loss": 0.4194, "step": 14453 }, { "epoch": 0.7311447215337143, "grad_norm": 6.081243515014648, "learning_rate": 1.7120422854434966e-06, "loss": 0.3453, "step": 14454 }, { "epoch": 0.7311953057817796, "grad_norm": 8.401253700256348, "learning_rate": 1.7114376585865062e-06, "loss": 0.4577, "step": 14455 }, { "epoch": 0.7312458900298447, "grad_norm": 13.861177444458008, "learning_rate": 1.7108331164678238e-06, "loss": 0.674, "step": 14456 }, { "epoch": 0.7312964742779099, "grad_norm": 5.662721157073975, "learning_rate": 1.710228659103026e-06, "loss": 0.3779, "step": 14457 }, { "epoch": 0.731347058525975, "grad_norm": 10.105830192565918, "learning_rate": 1.7096242865076911e-06, "loss": 0.3871, "step": 14458 }, { "epoch": 0.7313976427740402, "grad_norm": 6.373669624328613, "learning_rate": 1.709019998697391e-06, "loss": 0.364, "step": 14459 }, { "epoch": 0.7314482270221053, "grad_norm": 5.506974697113037, "learning_rate": 1.7084157956876947e-06, "loss": 0.3724, "step": 14460 }, { "epoch": 0.7314988112701705, "grad_norm": 5.188329219818115, "learning_rate": 1.7078116774941744e-06, "loss": 0.437, "step": 14461 }, { "epoch": 0.7315493955182356, "grad_norm": 5.631145477294922, "learning_rate": 1.7072076441323954e-06, "loss": 0.2631, "step": 14462 }, { "epoch": 0.7315999797663008, "grad_norm": 13.535175323486328, "learning_rate": 1.7066036956179222e-06, "loss": 0.5165, "step": 14463 }, { "epoch": 0.7316505640143659, "grad_norm": 11.61540699005127, "learning_rate": 1.7059998319663163e-06, "loss": 0.4619, "step": 14464 }, { "epoch": 0.731701148262431, "grad_norm": 12.55966854095459, "learning_rate": 1.7053960531931375e-06, "loss": 0.4215, "step": 14465 }, { "epoch": 0.7317517325104962, "grad_norm": 5.158379077911377, "learning_rate": 1.7047923593139454e-06, "loss": 0.3476, "step": 14466 }, { "epoch": 0.7318023167585613, "grad_norm": 6.183887004852295, "learning_rate": 1.7041887503442956e-06, "loss": 0.4222, "step": 14467 }, { "epoch": 0.7318529010066266, "grad_norm": 7.143652439117432, "learning_rate": 1.7035852262997388e-06, "loss": 0.3492, "step": 14468 }, { "epoch": 0.7319034852546917, "grad_norm": 9.611871719360352, "learning_rate": 1.702981787195831e-06, "loss": 0.4362, "step": 14469 }, { "epoch": 0.7319540695027569, "grad_norm": 5.34113883972168, "learning_rate": 1.702378433048118e-06, "loss": 0.3843, "step": 14470 }, { "epoch": 0.732004653750822, "grad_norm": 5.842563152313232, "learning_rate": 1.701775163872147e-06, "loss": 0.3642, "step": 14471 }, { "epoch": 0.7320552379988872, "grad_norm": 6.725620269775391, "learning_rate": 1.7011719796834641e-06, "loss": 0.3849, "step": 14472 }, { "epoch": 0.7321058222469523, "grad_norm": 6.824778079986572, "learning_rate": 1.7005688804976122e-06, "loss": 0.3424, "step": 14473 }, { "epoch": 0.7321564064950175, "grad_norm": 11.33047103881836, "learning_rate": 1.6999658663301288e-06, "loss": 0.5329, "step": 14474 }, { "epoch": 0.7322069907430826, "grad_norm": 6.5306925773620605, "learning_rate": 1.6993629371965575e-06, "loss": 0.4649, "step": 14475 }, { "epoch": 0.7322575749911477, "grad_norm": 9.323655128479004, "learning_rate": 1.6987600931124293e-06, "loss": 0.5152, "step": 14476 }, { "epoch": 0.7323081592392129, "grad_norm": 10.650982856750488, "learning_rate": 1.6981573340932777e-06, "loss": 0.4176, "step": 14477 }, { "epoch": 0.732358743487278, "grad_norm": 5.787348747253418, "learning_rate": 1.6975546601546383e-06, "loss": 0.3518, "step": 14478 }, { "epoch": 0.7324093277353432, "grad_norm": 5.122910499572754, "learning_rate": 1.6969520713120386e-06, "loss": 0.3043, "step": 14479 }, { "epoch": 0.7324599119834083, "grad_norm": 6.266517162322998, "learning_rate": 1.6963495675810037e-06, "loss": 0.4024, "step": 14480 }, { "epoch": 0.7325104962314736, "grad_norm": 6.309969902038574, "learning_rate": 1.6957471489770634e-06, "loss": 0.3189, "step": 14481 }, { "epoch": 0.7325610804795387, "grad_norm": 6.55487585067749, "learning_rate": 1.6951448155157369e-06, "loss": 0.4107, "step": 14482 }, { "epoch": 0.7326116647276039, "grad_norm": 9.895248413085938, "learning_rate": 1.6945425672125449e-06, "loss": 0.4059, "step": 14483 }, { "epoch": 0.732662248975669, "grad_norm": 7.310704231262207, "learning_rate": 1.6939404040830088e-06, "loss": 0.4283, "step": 14484 }, { "epoch": 0.7327128332237342, "grad_norm": 5.14169979095459, "learning_rate": 1.6933383261426423e-06, "loss": 0.4027, "step": 14485 }, { "epoch": 0.7327634174717993, "grad_norm": 9.903534889221191, "learning_rate": 1.6927363334069592e-06, "loss": 0.4103, "step": 14486 }, { "epoch": 0.7328140017198644, "grad_norm": 4.704837322235107, "learning_rate": 1.6921344258914741e-06, "loss": 0.4037, "step": 14487 }, { "epoch": 0.7328645859679296, "grad_norm": 10.701775550842285, "learning_rate": 1.6915326036116963e-06, "loss": 0.4638, "step": 14488 }, { "epoch": 0.7329151702159947, "grad_norm": 24.62384796142578, "learning_rate": 1.6909308665831287e-06, "loss": 0.5097, "step": 14489 }, { "epoch": 0.7329657544640599, "grad_norm": 6.781605243682861, "learning_rate": 1.690329214821282e-06, "loss": 0.4187, "step": 14490 }, { "epoch": 0.733016338712125, "grad_norm": 8.523119926452637, "learning_rate": 1.689727648341657e-06, "loss": 0.394, "step": 14491 }, { "epoch": 0.7330669229601902, "grad_norm": 5.59604549407959, "learning_rate": 1.6891261671597532e-06, "loss": 0.3984, "step": 14492 }, { "epoch": 0.7331175072082553, "grad_norm": 6.433204650878906, "learning_rate": 1.6885247712910725e-06, "loss": 0.319, "step": 14493 }, { "epoch": 0.7331680914563204, "grad_norm": 18.33443260192871, "learning_rate": 1.6879234607511102e-06, "loss": 0.4744, "step": 14494 }, { "epoch": 0.7332186757043857, "grad_norm": 9.588894844055176, "learning_rate": 1.6873222355553586e-06, "loss": 0.4327, "step": 14495 }, { "epoch": 0.7332692599524508, "grad_norm": 8.92648983001709, "learning_rate": 1.6867210957193131e-06, "loss": 0.2913, "step": 14496 }, { "epoch": 0.733319844200516, "grad_norm": 9.78563404083252, "learning_rate": 1.6861200412584622e-06, "loss": 0.4716, "step": 14497 }, { "epoch": 0.7333704284485811, "grad_norm": 6.117613792419434, "learning_rate": 1.6855190721882913e-06, "loss": 0.3686, "step": 14498 }, { "epoch": 0.7334210126966463, "grad_norm": 11.877668380737305, "learning_rate": 1.6849181885242904e-06, "loss": 0.4161, "step": 14499 }, { "epoch": 0.7334715969447114, "grad_norm": 9.154073715209961, "learning_rate": 1.6843173902819403e-06, "loss": 0.4069, "step": 14500 }, { "epoch": 0.7335221811927766, "grad_norm": 8.826112747192383, "learning_rate": 1.6837166774767221e-06, "loss": 0.3872, "step": 14501 }, { "epoch": 0.7335727654408417, "grad_norm": 8.24448299407959, "learning_rate": 1.6831160501241155e-06, "loss": 0.4158, "step": 14502 }, { "epoch": 0.7336233496889069, "grad_norm": 7.991009712219238, "learning_rate": 1.682515508239595e-06, "loss": 0.4645, "step": 14503 }, { "epoch": 0.733673933936972, "grad_norm": 13.689190864562988, "learning_rate": 1.6819150518386378e-06, "loss": 0.6986, "step": 14504 }, { "epoch": 0.7337245181850371, "grad_norm": 22.550140380859375, "learning_rate": 1.681314680936716e-06, "loss": 0.46, "step": 14505 }, { "epoch": 0.7337751024331023, "grad_norm": 10.462212562561035, "learning_rate": 1.6807143955492967e-06, "loss": 0.3596, "step": 14506 }, { "epoch": 0.7338256866811674, "grad_norm": 9.702890396118164, "learning_rate": 1.680114195691852e-06, "loss": 0.339, "step": 14507 }, { "epoch": 0.7338762709292327, "grad_norm": 6.920598983764648, "learning_rate": 1.6795140813798455e-06, "loss": 0.2859, "step": 14508 }, { "epoch": 0.7339268551772978, "grad_norm": 5.620450019836426, "learning_rate": 1.6789140526287389e-06, "loss": 0.317, "step": 14509 }, { "epoch": 0.733977439425363, "grad_norm": 10.571985244750977, "learning_rate": 1.6783141094539973e-06, "loss": 0.4661, "step": 14510 }, { "epoch": 0.7340280236734281, "grad_norm": 5.030009746551514, "learning_rate": 1.6777142518710783e-06, "loss": 0.3342, "step": 14511 }, { "epoch": 0.7340786079214933, "grad_norm": 4.842199802398682, "learning_rate": 1.6771144798954363e-06, "loss": 0.3895, "step": 14512 }, { "epoch": 0.7341291921695584, "grad_norm": 14.32168960571289, "learning_rate": 1.6765147935425313e-06, "loss": 0.4949, "step": 14513 }, { "epoch": 0.7341797764176236, "grad_norm": 5.508584499359131, "learning_rate": 1.6759151928278112e-06, "loss": 0.3672, "step": 14514 }, { "epoch": 0.7342303606656887, "grad_norm": 7.6965556144714355, "learning_rate": 1.6753156777667256e-06, "loss": 0.3504, "step": 14515 }, { "epoch": 0.7342809449137538, "grad_norm": 7.588629245758057, "learning_rate": 1.674716248374726e-06, "loss": 0.363, "step": 14516 }, { "epoch": 0.734331529161819, "grad_norm": 9.609392166137695, "learning_rate": 1.6741169046672573e-06, "loss": 0.4406, "step": 14517 }, { "epoch": 0.7343821134098841, "grad_norm": 7.2840256690979, "learning_rate": 1.6735176466597607e-06, "loss": 0.3037, "step": 14518 }, { "epoch": 0.7344326976579493, "grad_norm": 7.137681007385254, "learning_rate": 1.6729184743676813e-06, "loss": 0.3753, "step": 14519 }, { "epoch": 0.7344832819060144, "grad_norm": 10.996420860290527, "learning_rate": 1.6723193878064565e-06, "loss": 0.3592, "step": 14520 }, { "epoch": 0.7345338661540797, "grad_norm": 6.620541095733643, "learning_rate": 1.6717203869915216e-06, "loss": 0.4455, "step": 14521 }, { "epoch": 0.7345844504021448, "grad_norm": 6.6660003662109375, "learning_rate": 1.6711214719383152e-06, "loss": 0.3613, "step": 14522 }, { "epoch": 0.73463503465021, "grad_norm": 5.919199466705322, "learning_rate": 1.6705226426622679e-06, "loss": 0.271, "step": 14523 }, { "epoch": 0.7346856188982751, "grad_norm": 8.907822608947754, "learning_rate": 1.669923899178808e-06, "loss": 0.305, "step": 14524 }, { "epoch": 0.7347362031463402, "grad_norm": 7.704441547393799, "learning_rate": 1.6693252415033678e-06, "loss": 0.3303, "step": 14525 }, { "epoch": 0.7347867873944054, "grad_norm": 16.950313568115234, "learning_rate": 1.6687266696513727e-06, "loss": 0.35, "step": 14526 }, { "epoch": 0.7348373716424705, "grad_norm": 10.513731956481934, "learning_rate": 1.6681281836382413e-06, "loss": 0.3457, "step": 14527 }, { "epoch": 0.7348879558905357, "grad_norm": 6.552773475646973, "learning_rate": 1.6675297834794008e-06, "loss": 0.398, "step": 14528 }, { "epoch": 0.7349385401386008, "grad_norm": 5.06476354598999, "learning_rate": 1.6669314691902688e-06, "loss": 0.3897, "step": 14529 }, { "epoch": 0.734989124386666, "grad_norm": 4.37266731262207, "learning_rate": 1.6663332407862604e-06, "loss": 0.275, "step": 14530 }, { "epoch": 0.7350397086347311, "grad_norm": 6.064672946929932, "learning_rate": 1.6657350982827936e-06, "loss": 0.3443, "step": 14531 }, { "epoch": 0.7350902928827963, "grad_norm": 7.404038906097412, "learning_rate": 1.6651370416952795e-06, "loss": 0.419, "step": 14532 }, { "epoch": 0.7351408771308614, "grad_norm": 8.297676086425781, "learning_rate": 1.664539071039128e-06, "loss": 0.4355, "step": 14533 }, { "epoch": 0.7351914613789265, "grad_norm": 5.642621994018555, "learning_rate": 1.6639411863297488e-06, "loss": 0.4507, "step": 14534 }, { "epoch": 0.7352420456269918, "grad_norm": 6.263350963592529, "learning_rate": 1.6633433875825484e-06, "loss": 0.3872, "step": 14535 }, { "epoch": 0.735292629875057, "grad_norm": 7.23914098739624, "learning_rate": 1.6627456748129272e-06, "loss": 0.4166, "step": 14536 }, { "epoch": 0.7353432141231221, "grad_norm": 6.90286111831665, "learning_rate": 1.6621480480362907e-06, "loss": 0.3294, "step": 14537 }, { "epoch": 0.7353937983711872, "grad_norm": 6.781986236572266, "learning_rate": 1.661550507268037e-06, "loss": 0.3357, "step": 14538 }, { "epoch": 0.7354443826192524, "grad_norm": 8.96498966217041, "learning_rate": 1.660953052523563e-06, "loss": 0.4165, "step": 14539 }, { "epoch": 0.7354949668673175, "grad_norm": 6.190533638000488, "learning_rate": 1.6603556838182638e-06, "loss": 0.3642, "step": 14540 }, { "epoch": 0.7355455511153827, "grad_norm": 9.557721138000488, "learning_rate": 1.65975840116753e-06, "loss": 0.3934, "step": 14541 }, { "epoch": 0.7355961353634478, "grad_norm": 7.342596054077148, "learning_rate": 1.659161204586756e-06, "loss": 0.4508, "step": 14542 }, { "epoch": 0.735646719611513, "grad_norm": 6.49454402923584, "learning_rate": 1.6585640940913289e-06, "loss": 0.4154, "step": 14543 }, { "epoch": 0.7356973038595781, "grad_norm": 4.47280740737915, "learning_rate": 1.6579670696966322e-06, "loss": 0.3415, "step": 14544 }, { "epoch": 0.7357478881076432, "grad_norm": 8.813613891601562, "learning_rate": 1.6573701314180529e-06, "loss": 0.3781, "step": 14545 }, { "epoch": 0.7357984723557084, "grad_norm": 8.72944450378418, "learning_rate": 1.6567732792709717e-06, "loss": 0.3811, "step": 14546 }, { "epoch": 0.7358490566037735, "grad_norm": 6.197651386260986, "learning_rate": 1.656176513270767e-06, "loss": 0.303, "step": 14547 }, { "epoch": 0.7358996408518388, "grad_norm": 11.652624130249023, "learning_rate": 1.6555798334328176e-06, "loss": 0.5391, "step": 14548 }, { "epoch": 0.7359502250999039, "grad_norm": 9.708283424377441, "learning_rate": 1.654983239772498e-06, "loss": 0.4321, "step": 14549 }, { "epoch": 0.7360008093479691, "grad_norm": 12.24022388458252, "learning_rate": 1.654386732305181e-06, "loss": 0.4954, "step": 14550 }, { "epoch": 0.7360513935960342, "grad_norm": 6.92678165435791, "learning_rate": 1.653790311046235e-06, "loss": 0.2477, "step": 14551 }, { "epoch": 0.7361019778440994, "grad_norm": 10.740541458129883, "learning_rate": 1.6531939760110337e-06, "loss": 0.4021, "step": 14552 }, { "epoch": 0.7361525620921645, "grad_norm": 14.18486499786377, "learning_rate": 1.6525977272149363e-06, "loss": 0.5904, "step": 14553 }, { "epoch": 0.7362031463402297, "grad_norm": 7.319807052612305, "learning_rate": 1.6520015646733124e-06, "loss": 0.4459, "step": 14554 }, { "epoch": 0.7362537305882948, "grad_norm": 5.108232498168945, "learning_rate": 1.6514054884015207e-06, "loss": 0.3414, "step": 14555 }, { "epoch": 0.7363043148363599, "grad_norm": 8.229715347290039, "learning_rate": 1.6508094984149198e-06, "loss": 0.3736, "step": 14556 }, { "epoch": 0.7363548990844251, "grad_norm": 10.844783782958984, "learning_rate": 1.6502135947288707e-06, "loss": 0.5196, "step": 14557 }, { "epoch": 0.7364054833324902, "grad_norm": 8.221830368041992, "learning_rate": 1.6496177773587257e-06, "loss": 0.4455, "step": 14558 }, { "epoch": 0.7364560675805554, "grad_norm": 6.11800479888916, "learning_rate": 1.6490220463198364e-06, "loss": 0.3709, "step": 14559 }, { "epoch": 0.7365066518286205, "grad_norm": 17.43178367614746, "learning_rate": 1.6484264016275565e-06, "loss": 0.6024, "step": 14560 }, { "epoch": 0.7365572360766858, "grad_norm": 8.907979011535645, "learning_rate": 1.6478308432972335e-06, "loss": 0.4758, "step": 14561 }, { "epoch": 0.7366078203247509, "grad_norm": 8.120367050170898, "learning_rate": 1.6472353713442106e-06, "loss": 0.2423, "step": 14562 }, { "epoch": 0.7366584045728161, "grad_norm": 8.167755126953125, "learning_rate": 1.6466399857838355e-06, "loss": 0.403, "step": 14563 }, { "epoch": 0.7367089888208812, "grad_norm": 4.36688232421875, "learning_rate": 1.646044686631449e-06, "loss": 0.3179, "step": 14564 }, { "epoch": 0.7367595730689463, "grad_norm": 10.532463073730469, "learning_rate": 1.645449473902389e-06, "loss": 0.3929, "step": 14565 }, { "epoch": 0.7368101573170115, "grad_norm": 8.84545612335205, "learning_rate": 1.6448543476119938e-06, "loss": 0.4069, "step": 14566 }, { "epoch": 0.7368607415650766, "grad_norm": 13.244722366333008, "learning_rate": 1.644259307775598e-06, "loss": 0.5668, "step": 14567 }, { "epoch": 0.7369113258131418, "grad_norm": 15.381518363952637, "learning_rate": 1.6436643544085324e-06, "loss": 0.3975, "step": 14568 }, { "epoch": 0.7369619100612069, "grad_norm": 9.835583686828613, "learning_rate": 1.6430694875261315e-06, "loss": 0.3292, "step": 14569 }, { "epoch": 0.7370124943092721, "grad_norm": 21.20448875427246, "learning_rate": 1.642474707143722e-06, "loss": 0.4523, "step": 14570 }, { "epoch": 0.7370630785573372, "grad_norm": 7.892068862915039, "learning_rate": 1.6418800132766271e-06, "loss": 0.4672, "step": 14571 }, { "epoch": 0.7371136628054024, "grad_norm": 7.081827640533447, "learning_rate": 1.641285405940175e-06, "loss": 0.418, "step": 14572 }, { "epoch": 0.7371642470534675, "grad_norm": 8.91871166229248, "learning_rate": 1.6406908851496856e-06, "loss": 0.4348, "step": 14573 }, { "epoch": 0.7372148313015328, "grad_norm": 10.687898635864258, "learning_rate": 1.6400964509204758e-06, "loss": 0.3947, "step": 14574 }, { "epoch": 0.7372654155495979, "grad_norm": 6.351239204406738, "learning_rate": 1.639502103267867e-06, "loss": 0.3145, "step": 14575 }, { "epoch": 0.737315999797663, "grad_norm": 4.5583977699279785, "learning_rate": 1.638907842207172e-06, "loss": 0.3816, "step": 14576 }, { "epoch": 0.7373665840457282, "grad_norm": 6.373723030090332, "learning_rate": 1.6383136677537015e-06, "loss": 0.4102, "step": 14577 }, { "epoch": 0.7374171682937933, "grad_norm": 5.8634033203125, "learning_rate": 1.6377195799227719e-06, "loss": 0.4967, "step": 14578 }, { "epoch": 0.7374677525418585, "grad_norm": 9.643155097961426, "learning_rate": 1.6371255787296835e-06, "loss": 0.4886, "step": 14579 }, { "epoch": 0.7375183367899236, "grad_norm": 6.321534633636475, "learning_rate": 1.6365316641897483e-06, "loss": 0.4282, "step": 14580 }, { "epoch": 0.7375689210379888, "grad_norm": 11.588648796081543, "learning_rate": 1.635937836318268e-06, "loss": 0.4491, "step": 14581 }, { "epoch": 0.7376195052860539, "grad_norm": 5.98829460144043, "learning_rate": 1.6353440951305426e-06, "loss": 0.325, "step": 14582 }, { "epoch": 0.7376700895341191, "grad_norm": 7.635589599609375, "learning_rate": 1.634750440641874e-06, "loss": 0.4477, "step": 14583 }, { "epoch": 0.7377206737821842, "grad_norm": 8.637646675109863, "learning_rate": 1.6341568728675589e-06, "loss": 0.387, "step": 14584 }, { "epoch": 0.7377712580302493, "grad_norm": 11.413394927978516, "learning_rate": 1.633563391822889e-06, "loss": 0.3573, "step": 14585 }, { "epoch": 0.7378218422783145, "grad_norm": 5.906710624694824, "learning_rate": 1.6329699975231612e-06, "loss": 0.4004, "step": 14586 }, { "epoch": 0.7378724265263796, "grad_norm": 6.08806848526001, "learning_rate": 1.6323766899836635e-06, "loss": 0.3881, "step": 14587 }, { "epoch": 0.7379230107744449, "grad_norm": 7.356945037841797, "learning_rate": 1.6317834692196848e-06, "loss": 0.4431, "step": 14588 }, { "epoch": 0.73797359502251, "grad_norm": 9.038126945495605, "learning_rate": 1.6311903352465085e-06, "loss": 0.3675, "step": 14589 }, { "epoch": 0.7380241792705752, "grad_norm": 8.196378707885742, "learning_rate": 1.6305972880794235e-06, "loss": 0.4462, "step": 14590 }, { "epoch": 0.7380747635186403, "grad_norm": 9.545202255249023, "learning_rate": 1.6300043277337046e-06, "loss": 0.3585, "step": 14591 }, { "epoch": 0.7381253477667055, "grad_norm": 6.4549055099487305, "learning_rate": 1.6294114542246364e-06, "loss": 0.227, "step": 14592 }, { "epoch": 0.7381759320147706, "grad_norm": 9.36943244934082, "learning_rate": 1.6288186675674933e-06, "loss": 0.4511, "step": 14593 }, { "epoch": 0.7382265162628358, "grad_norm": 9.75307846069336, "learning_rate": 1.6282259677775485e-06, "loss": 0.4863, "step": 14594 }, { "epoch": 0.7382771005109009, "grad_norm": 8.9131498336792, "learning_rate": 1.6276333548700786e-06, "loss": 0.3441, "step": 14595 }, { "epoch": 0.738327684758966, "grad_norm": 8.664779663085938, "learning_rate": 1.6270408288603512e-06, "loss": 0.4394, "step": 14596 }, { "epoch": 0.7383782690070312, "grad_norm": 6.595965385437012, "learning_rate": 1.626448389763633e-06, "loss": 0.4687, "step": 14597 }, { "epoch": 0.7384288532550963, "grad_norm": 5.591704845428467, "learning_rate": 1.6258560375951925e-06, "loss": 0.3349, "step": 14598 }, { "epoch": 0.7384794375031615, "grad_norm": 17.352712631225586, "learning_rate": 1.625263772370293e-06, "loss": 0.5714, "step": 14599 }, { "epoch": 0.7385300217512266, "grad_norm": 6.843233585357666, "learning_rate": 1.6246715941041924e-06, "loss": 0.3541, "step": 14600 }, { "epoch": 0.7385806059992919, "grad_norm": 8.849565505981445, "learning_rate": 1.6240795028121536e-06, "loss": 0.4003, "step": 14601 }, { "epoch": 0.738631190247357, "grad_norm": 12.235780715942383, "learning_rate": 1.6234874985094323e-06, "loss": 0.4168, "step": 14602 }, { "epoch": 0.7386817744954222, "grad_norm": 7.4411187171936035, "learning_rate": 1.6228955812112823e-06, "loss": 0.3874, "step": 14603 }, { "epoch": 0.7387323587434873, "grad_norm": 6.404621601104736, "learning_rate": 1.622303750932956e-06, "loss": 0.2996, "step": 14604 }, { "epoch": 0.7387829429915524, "grad_norm": 7.601948261260986, "learning_rate": 1.6217120076897037e-06, "loss": 0.421, "step": 14605 }, { "epoch": 0.7388335272396176, "grad_norm": 7.767133712768555, "learning_rate": 1.6211203514967711e-06, "loss": 0.315, "step": 14606 }, { "epoch": 0.7388841114876827, "grad_norm": 12.306358337402344, "learning_rate": 1.6205287823694078e-06, "loss": 0.467, "step": 14607 }, { "epoch": 0.7389346957357479, "grad_norm": 5.36394739151001, "learning_rate": 1.6199373003228546e-06, "loss": 0.4069, "step": 14608 }, { "epoch": 0.738985279983813, "grad_norm": 11.83922004699707, "learning_rate": 1.6193459053723515e-06, "loss": 0.4088, "step": 14609 }, { "epoch": 0.7390358642318782, "grad_norm": 17.070833206176758, "learning_rate": 1.6187545975331403e-06, "loss": 0.4607, "step": 14610 }, { "epoch": 0.7390864484799433, "grad_norm": 8.379984855651855, "learning_rate": 1.6181633768204564e-06, "loss": 0.4094, "step": 14611 }, { "epoch": 0.7391370327280085, "grad_norm": 7.509436130523682, "learning_rate": 1.6175722432495317e-06, "loss": 0.3428, "step": 14612 }, { "epoch": 0.7391876169760736, "grad_norm": 7.323918342590332, "learning_rate": 1.616981196835602e-06, "loss": 0.412, "step": 14613 }, { "epoch": 0.7392382012241389, "grad_norm": 8.80016803741455, "learning_rate": 1.6163902375938956e-06, "loss": 0.4282, "step": 14614 }, { "epoch": 0.739288785472204, "grad_norm": 6.382527828216553, "learning_rate": 1.6157993655396387e-06, "loss": 0.3597, "step": 14615 }, { "epoch": 0.7393393697202691, "grad_norm": 7.327175140380859, "learning_rate": 1.6152085806880614e-06, "loss": 0.3819, "step": 14616 }, { "epoch": 0.7393899539683343, "grad_norm": 11.342037200927734, "learning_rate": 1.614617883054379e-06, "loss": 0.422, "step": 14617 }, { "epoch": 0.7394405382163994, "grad_norm": 8.802923202514648, "learning_rate": 1.6140272726538192e-06, "loss": 0.4833, "step": 14618 }, { "epoch": 0.7394911224644646, "grad_norm": 6.573287010192871, "learning_rate": 1.613436749501598e-06, "loss": 0.3893, "step": 14619 }, { "epoch": 0.7395417067125297, "grad_norm": 7.6380839347839355, "learning_rate": 1.6128463136129302e-06, "loss": 0.3417, "step": 14620 }, { "epoch": 0.7395922909605949, "grad_norm": 6.626235485076904, "learning_rate": 1.6122559650030335e-06, "loss": 0.3943, "step": 14621 }, { "epoch": 0.73964287520866, "grad_norm": 7.6753106117248535, "learning_rate": 1.6116657036871169e-06, "loss": 0.4203, "step": 14622 }, { "epoch": 0.7396934594567252, "grad_norm": 6.647587299346924, "learning_rate": 1.6110755296803915e-06, "loss": 0.385, "step": 14623 }, { "epoch": 0.7397440437047903, "grad_norm": 11.052197456359863, "learning_rate": 1.6104854429980627e-06, "loss": 0.3988, "step": 14624 }, { "epoch": 0.7397946279528554, "grad_norm": 8.908587455749512, "learning_rate": 1.6098954436553383e-06, "loss": 0.4634, "step": 14625 }, { "epoch": 0.7398452122009206, "grad_norm": 4.3297600746154785, "learning_rate": 1.60930553166742e-06, "loss": 0.359, "step": 14626 }, { "epoch": 0.7398957964489857, "grad_norm": 9.750628471374512, "learning_rate": 1.6087157070495064e-06, "loss": 0.4578, "step": 14627 }, { "epoch": 0.739946380697051, "grad_norm": 8.987832069396973, "learning_rate": 1.6081259698167996e-06, "loss": 0.4141, "step": 14628 }, { "epoch": 0.7399969649451161, "grad_norm": 7.942048072814941, "learning_rate": 1.6075363199844935e-06, "loss": 0.3634, "step": 14629 }, { "epoch": 0.7400475491931813, "grad_norm": 12.884207725524902, "learning_rate": 1.6069467575677828e-06, "loss": 0.5242, "step": 14630 }, { "epoch": 0.7400981334412464, "grad_norm": 7.722708225250244, "learning_rate": 1.6063572825818579e-06, "loss": 0.4944, "step": 14631 }, { "epoch": 0.7401487176893116, "grad_norm": 10.056706428527832, "learning_rate": 1.605767895041908e-06, "loss": 0.4172, "step": 14632 }, { "epoch": 0.7401993019373767, "grad_norm": 4.880821228027344, "learning_rate": 1.6051785949631222e-06, "loss": 0.3067, "step": 14633 }, { "epoch": 0.7402498861854419, "grad_norm": 6.067667007446289, "learning_rate": 1.6045893823606839e-06, "loss": 0.3268, "step": 14634 }, { "epoch": 0.740300470433507, "grad_norm": 8.6678466796875, "learning_rate": 1.6040002572497748e-06, "loss": 0.4053, "step": 14635 }, { "epoch": 0.7403510546815721, "grad_norm": 8.015766143798828, "learning_rate": 1.6034112196455775e-06, "loss": 0.4114, "step": 14636 }, { "epoch": 0.7404016389296373, "grad_norm": 4.473784923553467, "learning_rate": 1.6028222695632694e-06, "loss": 0.3764, "step": 14637 }, { "epoch": 0.7404522231777024, "grad_norm": 7.016688346862793, "learning_rate": 1.602233407018024e-06, "loss": 0.3979, "step": 14638 }, { "epoch": 0.7405028074257676, "grad_norm": 5.3315935134887695, "learning_rate": 1.6016446320250183e-06, "loss": 0.3689, "step": 14639 }, { "epoch": 0.7405533916738327, "grad_norm": 6.6311421394348145, "learning_rate": 1.6010559445994222e-06, "loss": 0.3049, "step": 14640 }, { "epoch": 0.740603975921898, "grad_norm": 7.826929092407227, "learning_rate": 1.6004673447564028e-06, "loss": 0.4126, "step": 14641 }, { "epoch": 0.7406545601699631, "grad_norm": 6.120717525482178, "learning_rate": 1.5998788325111325e-06, "loss": 0.4089, "step": 14642 }, { "epoch": 0.7407051444180283, "grad_norm": 6.883181571960449, "learning_rate": 1.5992904078787697e-06, "loss": 0.3518, "step": 14643 }, { "epoch": 0.7407557286660934, "grad_norm": 7.055516242980957, "learning_rate": 1.5987020708744778e-06, "loss": 0.3707, "step": 14644 }, { "epoch": 0.7408063129141585, "grad_norm": 6.258710861206055, "learning_rate": 1.5981138215134195e-06, "loss": 0.2725, "step": 14645 }, { "epoch": 0.7408568971622237, "grad_norm": 9.3309907913208, "learning_rate": 1.5975256598107509e-06, "loss": 0.4543, "step": 14646 }, { "epoch": 0.7409074814102888, "grad_norm": 10.46469497680664, "learning_rate": 1.5969375857816265e-06, "loss": 0.4371, "step": 14647 }, { "epoch": 0.740958065658354, "grad_norm": 5.990042209625244, "learning_rate": 1.596349599441202e-06, "loss": 0.3816, "step": 14648 }, { "epoch": 0.7410086499064191, "grad_norm": 13.20274829864502, "learning_rate": 1.5957617008046273e-06, "loss": 0.3668, "step": 14649 }, { "epoch": 0.7410592341544843, "grad_norm": 7.322551250457764, "learning_rate": 1.5951738898870494e-06, "loss": 0.391, "step": 14650 }, { "epoch": 0.7411098184025494, "grad_norm": 8.593748092651367, "learning_rate": 1.5945861667036177e-06, "loss": 0.4131, "step": 14651 }, { "epoch": 0.7411604026506146, "grad_norm": 4.633668899536133, "learning_rate": 1.5939985312694757e-06, "loss": 0.3706, "step": 14652 }, { "epoch": 0.7412109868986797, "grad_norm": 11.50998592376709, "learning_rate": 1.5934109835997624e-06, "loss": 0.4485, "step": 14653 }, { "epoch": 0.741261571146745, "grad_norm": 12.123479843139648, "learning_rate": 1.5928235237096219e-06, "loss": 0.5458, "step": 14654 }, { "epoch": 0.7413121553948101, "grad_norm": 9.383923530578613, "learning_rate": 1.5922361516141892e-06, "loss": 0.356, "step": 14655 }, { "epoch": 0.7413627396428752, "grad_norm": 8.559577941894531, "learning_rate": 1.5916488673285995e-06, "loss": 0.4318, "step": 14656 }, { "epoch": 0.7414133238909404, "grad_norm": 8.611639976501465, "learning_rate": 1.5910616708679866e-06, "loss": 0.3588, "step": 14657 }, { "epoch": 0.7414639081390055, "grad_norm": 17.03389549255371, "learning_rate": 1.5904745622474782e-06, "loss": 0.6373, "step": 14658 }, { "epoch": 0.7415144923870707, "grad_norm": 6.213283061981201, "learning_rate": 1.5898875414822068e-06, "loss": 0.3921, "step": 14659 }, { "epoch": 0.7415650766351358, "grad_norm": 13.373357772827148, "learning_rate": 1.589300608587297e-06, "loss": 0.5018, "step": 14660 }, { "epoch": 0.741615660883201, "grad_norm": 5.592657089233398, "learning_rate": 1.5887137635778721e-06, "loss": 0.3568, "step": 14661 }, { "epoch": 0.7416662451312661, "grad_norm": 7.476309299468994, "learning_rate": 1.5881270064690523e-06, "loss": 0.4521, "step": 14662 }, { "epoch": 0.7417168293793313, "grad_norm": 6.571254253387451, "learning_rate": 1.58754033727596e-06, "loss": 0.3381, "step": 14663 }, { "epoch": 0.7417674136273964, "grad_norm": 7.354515552520752, "learning_rate": 1.5869537560137105e-06, "loss": 0.3367, "step": 14664 }, { "epoch": 0.7418179978754615, "grad_norm": 7.290426254272461, "learning_rate": 1.5863672626974175e-06, "loss": 0.3926, "step": 14665 }, { "epoch": 0.7418685821235267, "grad_norm": 8.799710273742676, "learning_rate": 1.5857808573421967e-06, "loss": 0.5268, "step": 14666 }, { "epoch": 0.7419191663715918, "grad_norm": 6.351482391357422, "learning_rate": 1.5851945399631564e-06, "loss": 0.3786, "step": 14667 }, { "epoch": 0.7419697506196571, "grad_norm": 10.43108081817627, "learning_rate": 1.5846083105754046e-06, "loss": 0.5638, "step": 14668 }, { "epoch": 0.7420203348677222, "grad_norm": 9.452704429626465, "learning_rate": 1.5840221691940466e-06, "loss": 0.4226, "step": 14669 }, { "epoch": 0.7420709191157874, "grad_norm": 11.13649845123291, "learning_rate": 1.5834361158341853e-06, "loss": 0.4254, "step": 14670 }, { "epoch": 0.7421215033638525, "grad_norm": 8.918313026428223, "learning_rate": 1.5828501505109245e-06, "loss": 0.4242, "step": 14671 }, { "epoch": 0.7421720876119177, "grad_norm": 6.446921348571777, "learning_rate": 1.5822642732393617e-06, "loss": 0.3501, "step": 14672 }, { "epoch": 0.7422226718599828, "grad_norm": 10.899911880493164, "learning_rate": 1.581678484034591e-06, "loss": 0.3738, "step": 14673 }, { "epoch": 0.742273256108048, "grad_norm": 6.570088863372803, "learning_rate": 1.5810927829117118e-06, "loss": 0.3788, "step": 14674 }, { "epoch": 0.7423238403561131, "grad_norm": 7.532925128936768, "learning_rate": 1.5805071698858127e-06, "loss": 0.3555, "step": 14675 }, { "epoch": 0.7423744246041782, "grad_norm": 6.523129940032959, "learning_rate": 1.5799216449719835e-06, "loss": 0.3708, "step": 14676 }, { "epoch": 0.7424250088522434, "grad_norm": 7.583657741546631, "learning_rate": 1.5793362081853136e-06, "loss": 0.433, "step": 14677 }, { "epoch": 0.7424755931003085, "grad_norm": 8.981867790222168, "learning_rate": 1.5787508595408875e-06, "loss": 0.4789, "step": 14678 }, { "epoch": 0.7425261773483737, "grad_norm": 6.3999128341674805, "learning_rate": 1.5781655990537864e-06, "loss": 0.3851, "step": 14679 }, { "epoch": 0.7425767615964388, "grad_norm": 9.708691596984863, "learning_rate": 1.5775804267390948e-06, "loss": 0.3631, "step": 14680 }, { "epoch": 0.7426273458445041, "grad_norm": 11.012067794799805, "learning_rate": 1.5769953426118906e-06, "loss": 0.4218, "step": 14681 }, { "epoch": 0.7426779300925692, "grad_norm": 9.789505958557129, "learning_rate": 1.576410346687245e-06, "loss": 0.3242, "step": 14682 }, { "epoch": 0.7427285143406344, "grad_norm": 7.339754581451416, "learning_rate": 1.5758254389802369e-06, "loss": 0.4005, "step": 14683 }, { "epoch": 0.7427790985886995, "grad_norm": 12.7216157913208, "learning_rate": 1.5752406195059366e-06, "loss": 0.4858, "step": 14684 }, { "epoch": 0.7428296828367646, "grad_norm": 10.748266220092773, "learning_rate": 1.5746558882794116e-06, "loss": 0.4156, "step": 14685 }, { "epoch": 0.7428802670848298, "grad_norm": 8.961479187011719, "learning_rate": 1.574071245315733e-06, "loss": 0.4672, "step": 14686 }, { "epoch": 0.7429308513328949, "grad_norm": 6.738979816436768, "learning_rate": 1.5734866906299628e-06, "loss": 0.3373, "step": 14687 }, { "epoch": 0.7429814355809601, "grad_norm": 6.789628028869629, "learning_rate": 1.5729022242371627e-06, "loss": 0.4011, "step": 14688 }, { "epoch": 0.7430320198290252, "grad_norm": 4.529475212097168, "learning_rate": 1.5723178461523964e-06, "loss": 0.2948, "step": 14689 }, { "epoch": 0.7430826040770904, "grad_norm": 6.483086585998535, "learning_rate": 1.5717335563907199e-06, "loss": 0.2943, "step": 14690 }, { "epoch": 0.7431331883251555, "grad_norm": 17.466604232788086, "learning_rate": 1.5711493549671869e-06, "loss": 0.5076, "step": 14691 }, { "epoch": 0.7431837725732207, "grad_norm": 11.99715518951416, "learning_rate": 1.5705652418968549e-06, "loss": 0.4829, "step": 14692 }, { "epoch": 0.7432343568212858, "grad_norm": 8.671194076538086, "learning_rate": 1.569981217194773e-06, "loss": 0.3281, "step": 14693 }, { "epoch": 0.7432849410693511, "grad_norm": 8.239076614379883, "learning_rate": 1.5693972808759905e-06, "loss": 0.5112, "step": 14694 }, { "epoch": 0.7433355253174162, "grad_norm": 11.500557899475098, "learning_rate": 1.5688134329555537e-06, "loss": 0.5468, "step": 14695 }, { "epoch": 0.7433861095654813, "grad_norm": 16.99420738220215, "learning_rate": 1.5682296734485076e-06, "loss": 0.3915, "step": 14696 }, { "epoch": 0.7434366938135465, "grad_norm": 6.8192267417907715, "learning_rate": 1.5676460023698915e-06, "loss": 0.3346, "step": 14697 }, { "epoch": 0.7434872780616116, "grad_norm": 10.440855979919434, "learning_rate": 1.567062419734749e-06, "loss": 0.4439, "step": 14698 }, { "epoch": 0.7435378623096768, "grad_norm": 8.334537506103516, "learning_rate": 1.5664789255581158e-06, "loss": 0.3387, "step": 14699 }, { "epoch": 0.7435884465577419, "grad_norm": 11.77558708190918, "learning_rate": 1.5658955198550262e-06, "loss": 0.5012, "step": 14700 }, { "epoch": 0.7436390308058071, "grad_norm": 8.6191987991333, "learning_rate": 1.565312202640516e-06, "loss": 0.4665, "step": 14701 }, { "epoch": 0.7436896150538722, "grad_norm": 7.615171909332275, "learning_rate": 1.564728973929614e-06, "loss": 0.3839, "step": 14702 }, { "epoch": 0.7437401993019374, "grad_norm": 7.170152187347412, "learning_rate": 1.5641458337373472e-06, "loss": 0.4547, "step": 14703 }, { "epoch": 0.7437907835500025, "grad_norm": 11.359394073486328, "learning_rate": 1.5635627820787447e-06, "loss": 0.397, "step": 14704 }, { "epoch": 0.7438413677980676, "grad_norm": 10.240310668945312, "learning_rate": 1.5629798189688295e-06, "loss": 0.3805, "step": 14705 }, { "epoch": 0.7438919520461328, "grad_norm": 8.30016040802002, "learning_rate": 1.5623969444226218e-06, "loss": 0.5011, "step": 14706 }, { "epoch": 0.7439425362941979, "grad_norm": 6.411681175231934, "learning_rate": 1.561814158455142e-06, "loss": 0.4166, "step": 14707 }, { "epoch": 0.7439931205422632, "grad_norm": 6.071496486663818, "learning_rate": 1.561231461081405e-06, "loss": 0.3617, "step": 14708 }, { "epoch": 0.7440437047903283, "grad_norm": 9.296004295349121, "learning_rate": 1.5606488523164292e-06, "loss": 0.4373, "step": 14709 }, { "epoch": 0.7440942890383935, "grad_norm": 7.4425435066223145, "learning_rate": 1.5600663321752252e-06, "loss": 0.3295, "step": 14710 }, { "epoch": 0.7441448732864586, "grad_norm": 9.294925689697266, "learning_rate": 1.5594839006728007e-06, "loss": 0.4339, "step": 14711 }, { "epoch": 0.7441954575345238, "grad_norm": 9.016031265258789, "learning_rate": 1.5589015578241678e-06, "loss": 0.4881, "step": 14712 }, { "epoch": 0.7442460417825889, "grad_norm": 41.77742004394531, "learning_rate": 1.5583193036443306e-06, "loss": 0.6545, "step": 14713 }, { "epoch": 0.744296626030654, "grad_norm": 9.840521812438965, "learning_rate": 1.5577371381482898e-06, "loss": 0.4586, "step": 14714 }, { "epoch": 0.7443472102787192, "grad_norm": 10.443181991577148, "learning_rate": 1.55715506135105e-06, "loss": 0.5058, "step": 14715 }, { "epoch": 0.7443977945267843, "grad_norm": 9.721104621887207, "learning_rate": 1.5565730732676087e-06, "loss": 0.4363, "step": 14716 }, { "epoch": 0.7444483787748495, "grad_norm": 8.056685447692871, "learning_rate": 1.5559911739129601e-06, "loss": 0.331, "step": 14717 }, { "epoch": 0.7444989630229146, "grad_norm": 8.5669527053833, "learning_rate": 1.5554093633021023e-06, "loss": 0.4581, "step": 14718 }, { "epoch": 0.7445495472709798, "grad_norm": 18.177947998046875, "learning_rate": 1.5548276414500262e-06, "loss": 0.5834, "step": 14719 }, { "epoch": 0.7446001315190449, "grad_norm": 6.660064220428467, "learning_rate": 1.5542460083717176e-06, "loss": 0.3609, "step": 14720 }, { "epoch": 0.7446507157671102, "grad_norm": 5.6985015869140625, "learning_rate": 1.553664464082168e-06, "loss": 0.3656, "step": 14721 }, { "epoch": 0.7447013000151753, "grad_norm": 7.870020866394043, "learning_rate": 1.5530830085963612e-06, "loss": 0.4114, "step": 14722 }, { "epoch": 0.7447518842632405, "grad_norm": 10.238566398620605, "learning_rate": 1.5525016419292772e-06, "loss": 0.3143, "step": 14723 }, { "epoch": 0.7448024685113056, "grad_norm": 7.9847917556762695, "learning_rate": 1.5519203640959007e-06, "loss": 0.3599, "step": 14724 }, { "epoch": 0.7448530527593707, "grad_norm": 6.843548774719238, "learning_rate": 1.5513391751112084e-06, "loss": 0.387, "step": 14725 }, { "epoch": 0.7449036370074359, "grad_norm": 13.134807586669922, "learning_rate": 1.5507580749901735e-06, "loss": 0.4106, "step": 14726 }, { "epoch": 0.744954221255501, "grad_norm": 14.527103424072266, "learning_rate": 1.5501770637477737e-06, "loss": 0.581, "step": 14727 }, { "epoch": 0.7450048055035662, "grad_norm": 6.7379231452941895, "learning_rate": 1.549596141398978e-06, "loss": 0.4177, "step": 14728 }, { "epoch": 0.7450553897516313, "grad_norm": 9.6477689743042, "learning_rate": 1.549015307958754e-06, "loss": 0.313, "step": 14729 }, { "epoch": 0.7451059739996965, "grad_norm": 9.058414459228516, "learning_rate": 1.5484345634420716e-06, "loss": 0.3333, "step": 14730 }, { "epoch": 0.7451565582477616, "grad_norm": 5.914481163024902, "learning_rate": 1.5478539078638938e-06, "loss": 0.3293, "step": 14731 }, { "epoch": 0.7452071424958268, "grad_norm": 4.227816104888916, "learning_rate": 1.5472733412391822e-06, "loss": 0.3688, "step": 14732 }, { "epoch": 0.7452577267438919, "grad_norm": 9.138033866882324, "learning_rate": 1.5466928635828972e-06, "loss": 0.3584, "step": 14733 }, { "epoch": 0.7453083109919572, "grad_norm": 9.884501457214355, "learning_rate": 1.5461124749099953e-06, "loss": 0.3413, "step": 14734 }, { "epoch": 0.7453588952400223, "grad_norm": 5.972199440002441, "learning_rate": 1.5455321752354313e-06, "loss": 0.3285, "step": 14735 }, { "epoch": 0.7454094794880874, "grad_norm": 6.870606422424316, "learning_rate": 1.5449519645741611e-06, "loss": 0.3984, "step": 14736 }, { "epoch": 0.7454600637361526, "grad_norm": 6.477025508880615, "learning_rate": 1.5443718429411331e-06, "loss": 0.4296, "step": 14737 }, { "epoch": 0.7455106479842177, "grad_norm": 4.706218719482422, "learning_rate": 1.5437918103512945e-06, "loss": 0.4032, "step": 14738 }, { "epoch": 0.7455612322322829, "grad_norm": 9.564148902893066, "learning_rate": 1.5432118668195945e-06, "loss": 0.4445, "step": 14739 }, { "epoch": 0.745611816480348, "grad_norm": 6.587109088897705, "learning_rate": 1.542632012360975e-06, "loss": 0.344, "step": 14740 }, { "epoch": 0.7456624007284132, "grad_norm": 7.097555160522461, "learning_rate": 1.542052246990376e-06, "loss": 0.4299, "step": 14741 }, { "epoch": 0.7457129849764783, "grad_norm": 11.063331604003906, "learning_rate": 1.5414725707227395e-06, "loss": 0.4556, "step": 14742 }, { "epoch": 0.7457635692245435, "grad_norm": 13.29554271697998, "learning_rate": 1.540892983573002e-06, "loss": 0.4727, "step": 14743 }, { "epoch": 0.7458141534726086, "grad_norm": 5.8054986000061035, "learning_rate": 1.5403134855560947e-06, "loss": 0.2656, "step": 14744 }, { "epoch": 0.7458647377206737, "grad_norm": 12.604016304016113, "learning_rate": 1.5397340766869562e-06, "loss": 0.5314, "step": 14745 }, { "epoch": 0.7459153219687389, "grad_norm": 6.004695892333984, "learning_rate": 1.5391547569805093e-06, "loss": 0.3792, "step": 14746 }, { "epoch": 0.7459659062168041, "grad_norm": 6.198154449462891, "learning_rate": 1.5385755264516862e-06, "loss": 0.3985, "step": 14747 }, { "epoch": 0.7460164904648693, "grad_norm": 6.952739715576172, "learning_rate": 1.5379963851154113e-06, "loss": 0.3739, "step": 14748 }, { "epoch": 0.7460670747129344, "grad_norm": 5.995763778686523, "learning_rate": 1.537417332986606e-06, "loss": 0.3281, "step": 14749 }, { "epoch": 0.7461176589609996, "grad_norm": 7.6855597496032715, "learning_rate": 1.5368383700801936e-06, "loss": 0.4084, "step": 14750 }, { "epoch": 0.7461682432090647, "grad_norm": 12.156920433044434, "learning_rate": 1.5362594964110922e-06, "loss": 0.4947, "step": 14751 }, { "epoch": 0.7462188274571299, "grad_norm": 10.324061393737793, "learning_rate": 1.5356807119942153e-06, "loss": 0.4345, "step": 14752 }, { "epoch": 0.746269411705195, "grad_norm": 7.115224361419678, "learning_rate": 1.5351020168444802e-06, "loss": 0.3459, "step": 14753 }, { "epoch": 0.7463199959532602, "grad_norm": 5.5712456703186035, "learning_rate": 1.534523410976797e-06, "loss": 0.2895, "step": 14754 }, { "epoch": 0.7463705802013253, "grad_norm": 11.661256790161133, "learning_rate": 1.5339448944060736e-06, "loss": 0.4406, "step": 14755 }, { "epoch": 0.7464211644493904, "grad_norm": 12.402344703674316, "learning_rate": 1.5333664671472199e-06, "loss": 0.3746, "step": 14756 }, { "epoch": 0.7464717486974556, "grad_norm": 7.456275463104248, "learning_rate": 1.5327881292151387e-06, "loss": 0.3007, "step": 14757 }, { "epoch": 0.7465223329455207, "grad_norm": 5.732017993927002, "learning_rate": 1.5322098806247331e-06, "loss": 0.3348, "step": 14758 }, { "epoch": 0.7465729171935859, "grad_norm": 6.0543389320373535, "learning_rate": 1.5316317213909026e-06, "loss": 0.4141, "step": 14759 }, { "epoch": 0.746623501441651, "grad_norm": 7.127002716064453, "learning_rate": 1.5310536515285451e-06, "loss": 0.4025, "step": 14760 }, { "epoch": 0.7466740856897163, "grad_norm": 13.650650024414062, "learning_rate": 1.5304756710525547e-06, "loss": 0.4338, "step": 14761 }, { "epoch": 0.7467246699377814, "grad_norm": 9.987975120544434, "learning_rate": 1.5298977799778276e-06, "loss": 0.4524, "step": 14762 }, { "epoch": 0.7467752541858466, "grad_norm": 12.122760772705078, "learning_rate": 1.529319978319253e-06, "loss": 0.4629, "step": 14763 }, { "epoch": 0.7468258384339117, "grad_norm": 13.674741744995117, "learning_rate": 1.5287422660917179e-06, "loss": 0.4652, "step": 14764 }, { "epoch": 0.7468764226819768, "grad_norm": 11.25993824005127, "learning_rate": 1.528164643310111e-06, "loss": 0.3536, "step": 14765 }, { "epoch": 0.746927006930042, "grad_norm": 6.792492389678955, "learning_rate": 1.5275871099893164e-06, "loss": 0.3741, "step": 14766 }, { "epoch": 0.7469775911781071, "grad_norm": 4.949759006500244, "learning_rate": 1.5270096661442125e-06, "loss": 0.3467, "step": 14767 }, { "epoch": 0.7470281754261723, "grad_norm": 11.128680229187012, "learning_rate": 1.5264323117896824e-06, "loss": 0.4468, "step": 14768 }, { "epoch": 0.7470787596742374, "grad_norm": 7.180875301361084, "learning_rate": 1.525855046940602e-06, "loss": 0.3137, "step": 14769 }, { "epoch": 0.7471293439223026, "grad_norm": 6.842592716217041, "learning_rate": 1.5252778716118438e-06, "loss": 0.4042, "step": 14770 }, { "epoch": 0.7471799281703677, "grad_norm": 16.43895721435547, "learning_rate": 1.524700785818285e-06, "loss": 0.4198, "step": 14771 }, { "epoch": 0.7472305124184329, "grad_norm": 11.293438911437988, "learning_rate": 1.5241237895747912e-06, "loss": 0.4425, "step": 14772 }, { "epoch": 0.747281096666498, "grad_norm": 6.05204439163208, "learning_rate": 1.5235468828962302e-06, "loss": 0.368, "step": 14773 }, { "epoch": 0.7473316809145633, "grad_norm": 14.674843788146973, "learning_rate": 1.5229700657974705e-06, "loss": 0.3976, "step": 14774 }, { "epoch": 0.7473822651626284, "grad_norm": 8.874537467956543, "learning_rate": 1.5223933382933743e-06, "loss": 0.3797, "step": 14775 }, { "epoch": 0.7474328494106935, "grad_norm": 5.946469306945801, "learning_rate": 1.5218167003987994e-06, "loss": 0.3789, "step": 14776 }, { "epoch": 0.7474834336587587, "grad_norm": 7.272408485412598, "learning_rate": 1.5212401521286091e-06, "loss": 0.3755, "step": 14777 }, { "epoch": 0.7475340179068238, "grad_norm": 8.474185943603516, "learning_rate": 1.5206636934976577e-06, "loss": 0.4314, "step": 14778 }, { "epoch": 0.747584602154889, "grad_norm": 6.105806350708008, "learning_rate": 1.520087324520797e-06, "loss": 0.3426, "step": 14779 }, { "epoch": 0.7476351864029541, "grad_norm": 6.584578990936279, "learning_rate": 1.519511045212882e-06, "loss": 0.4407, "step": 14780 }, { "epoch": 0.7476857706510193, "grad_norm": 5.687323093414307, "learning_rate": 1.5189348555887606e-06, "loss": 0.363, "step": 14781 }, { "epoch": 0.7477363548990844, "grad_norm": 6.25565242767334, "learning_rate": 1.5183587556632784e-06, "loss": 0.3117, "step": 14782 }, { "epoch": 0.7477869391471496, "grad_norm": 6.620989799499512, "learning_rate": 1.517782745451285e-06, "loss": 0.3035, "step": 14783 }, { "epoch": 0.7478375233952147, "grad_norm": 5.680984020233154, "learning_rate": 1.5172068249676157e-06, "loss": 0.3999, "step": 14784 }, { "epoch": 0.7478881076432798, "grad_norm": 8.368342399597168, "learning_rate": 1.5166309942271156e-06, "loss": 0.3651, "step": 14785 }, { "epoch": 0.747938691891345, "grad_norm": 7.747830867767334, "learning_rate": 1.5160552532446216e-06, "loss": 0.3832, "step": 14786 }, { "epoch": 0.7479892761394102, "grad_norm": 10.439229965209961, "learning_rate": 1.5154796020349666e-06, "loss": 0.459, "step": 14787 }, { "epoch": 0.7480398603874754, "grad_norm": 23.280385971069336, "learning_rate": 1.5149040406129873e-06, "loss": 0.541, "step": 14788 }, { "epoch": 0.7480904446355405, "grad_norm": 7.396115303039551, "learning_rate": 1.5143285689935132e-06, "loss": 0.3159, "step": 14789 }, { "epoch": 0.7481410288836057, "grad_norm": 13.12529468536377, "learning_rate": 1.5137531871913707e-06, "loss": 0.5128, "step": 14790 }, { "epoch": 0.7481916131316708, "grad_norm": 5.099736213684082, "learning_rate": 1.51317789522139e-06, "loss": 0.4184, "step": 14791 }, { "epoch": 0.748242197379736, "grad_norm": 10.790873527526855, "learning_rate": 1.5126026930983929e-06, "loss": 0.5268, "step": 14792 }, { "epoch": 0.7482927816278011, "grad_norm": 12.342096328735352, "learning_rate": 1.5120275808371987e-06, "loss": 0.4297, "step": 14793 }, { "epoch": 0.7483433658758663, "grad_norm": 8.16118335723877, "learning_rate": 1.5114525584526308e-06, "loss": 0.4127, "step": 14794 }, { "epoch": 0.7483939501239314, "grad_norm": 9.900983810424805, "learning_rate": 1.5108776259595044e-06, "loss": 0.4653, "step": 14795 }, { "epoch": 0.7484445343719965, "grad_norm": 7.437358379364014, "learning_rate": 1.5103027833726335e-06, "loss": 0.3563, "step": 14796 }, { "epoch": 0.7484951186200617, "grad_norm": 14.202230453491211, "learning_rate": 1.5097280307068314e-06, "loss": 0.4753, "step": 14797 }, { "epoch": 0.7485457028681268, "grad_norm": 9.905183792114258, "learning_rate": 1.509153367976907e-06, "loss": 0.41, "step": 14798 }, { "epoch": 0.748596287116192, "grad_norm": 14.887848854064941, "learning_rate": 1.508578795197667e-06, "loss": 0.4077, "step": 14799 }, { "epoch": 0.7486468713642571, "grad_norm": 10.320124626159668, "learning_rate": 1.50800431238392e-06, "loss": 0.4471, "step": 14800 }, { "epoch": 0.7486974556123224, "grad_norm": 7.974280834197998, "learning_rate": 1.5074299195504676e-06, "loss": 0.3907, "step": 14801 }, { "epoch": 0.7487480398603875, "grad_norm": 9.143147468566895, "learning_rate": 1.5068556167121085e-06, "loss": 0.3266, "step": 14802 }, { "epoch": 0.7487986241084527, "grad_norm": 6.701456069946289, "learning_rate": 1.5062814038836437e-06, "loss": 0.3606, "step": 14803 }, { "epoch": 0.7488492083565178, "grad_norm": 8.30303955078125, "learning_rate": 1.505707281079869e-06, "loss": 0.4219, "step": 14804 }, { "epoch": 0.748899792604583, "grad_norm": 10.956780433654785, "learning_rate": 1.5051332483155762e-06, "loss": 0.3701, "step": 14805 }, { "epoch": 0.7489503768526481, "grad_norm": 6.82133674621582, "learning_rate": 1.5045593056055591e-06, "loss": 0.3724, "step": 14806 }, { "epoch": 0.7490009611007132, "grad_norm": 6.420109748840332, "learning_rate": 1.5039854529646065e-06, "loss": 0.2547, "step": 14807 }, { "epoch": 0.7490515453487784, "grad_norm": 7.32736349105835, "learning_rate": 1.5034116904075024e-06, "loss": 0.3746, "step": 14808 }, { "epoch": 0.7491021295968435, "grad_norm": 4.911613941192627, "learning_rate": 1.5028380179490364e-06, "loss": 0.4052, "step": 14809 }, { "epoch": 0.7491527138449087, "grad_norm": 6.673050403594971, "learning_rate": 1.5022644356039862e-06, "loss": 0.4789, "step": 14810 }, { "epoch": 0.7492032980929738, "grad_norm": 7.471603870391846, "learning_rate": 1.5016909433871318e-06, "loss": 0.4257, "step": 14811 }, { "epoch": 0.749253882341039, "grad_norm": 11.17791748046875, "learning_rate": 1.5011175413132534e-06, "loss": 0.3927, "step": 14812 }, { "epoch": 0.7493044665891041, "grad_norm": 6.638622760772705, "learning_rate": 1.5005442293971245e-06, "loss": 0.3081, "step": 14813 }, { "epoch": 0.7493550508371694, "grad_norm": 7.647754669189453, "learning_rate": 1.4999710076535168e-06, "loss": 0.413, "step": 14814 }, { "epoch": 0.7494056350852345, "grad_norm": 13.400897026062012, "learning_rate": 1.499397876097204e-06, "loss": 0.6869, "step": 14815 }, { "epoch": 0.7494562193332996, "grad_norm": 10.192375183105469, "learning_rate": 1.4988248347429523e-06, "loss": 0.463, "step": 14816 }, { "epoch": 0.7495068035813648, "grad_norm": 6.52561092376709, "learning_rate": 1.4982518836055265e-06, "loss": 0.3586, "step": 14817 }, { "epoch": 0.7495573878294299, "grad_norm": 6.689467430114746, "learning_rate": 1.4976790226996928e-06, "loss": 0.4765, "step": 14818 }, { "epoch": 0.7496079720774951, "grad_norm": 7.744055271148682, "learning_rate": 1.4971062520402112e-06, "loss": 0.3644, "step": 14819 }, { "epoch": 0.7496585563255602, "grad_norm": 12.27020263671875, "learning_rate": 1.4965335716418395e-06, "loss": 0.4662, "step": 14820 }, { "epoch": 0.7497091405736254, "grad_norm": 12.731691360473633, "learning_rate": 1.4959609815193366e-06, "loss": 0.4705, "step": 14821 }, { "epoch": 0.7497597248216905, "grad_norm": 7.35333251953125, "learning_rate": 1.495388481687456e-06, "loss": 0.3852, "step": 14822 }, { "epoch": 0.7498103090697557, "grad_norm": 8.669623374938965, "learning_rate": 1.4948160721609485e-06, "loss": 0.3549, "step": 14823 }, { "epoch": 0.7498608933178208, "grad_norm": 10.104142189025879, "learning_rate": 1.494243752954565e-06, "loss": 0.3087, "step": 14824 }, { "epoch": 0.7499114775658859, "grad_norm": 11.10565185546875, "learning_rate": 1.4936715240830501e-06, "loss": 0.5304, "step": 14825 }, { "epoch": 0.7499620618139511, "grad_norm": 7.880829811096191, "learning_rate": 1.4930993855611525e-06, "loss": 0.3056, "step": 14826 }, { "epoch": 0.7500126460620163, "grad_norm": 19.222034454345703, "learning_rate": 1.4925273374036132e-06, "loss": 0.5859, "step": 14827 }, { "epoch": 0.7500632303100815, "grad_norm": 6.759103775024414, "learning_rate": 1.4919553796251712e-06, "loss": 0.3511, "step": 14828 }, { "epoch": 0.7501138145581466, "grad_norm": 12.850446701049805, "learning_rate": 1.4913835122405668e-06, "loss": 0.4426, "step": 14829 }, { "epoch": 0.7501643988062118, "grad_norm": 10.113645553588867, "learning_rate": 1.4908117352645353e-06, "loss": 0.4957, "step": 14830 }, { "epoch": 0.7502149830542769, "grad_norm": 9.592724800109863, "learning_rate": 1.4902400487118085e-06, "loss": 0.4525, "step": 14831 }, { "epoch": 0.7502655673023421, "grad_norm": 8.781320571899414, "learning_rate": 1.4896684525971172e-06, "loss": 0.3839, "step": 14832 }, { "epoch": 0.7503161515504072, "grad_norm": 17.30116081237793, "learning_rate": 1.4890969469351923e-06, "loss": 0.3784, "step": 14833 }, { "epoch": 0.7503667357984724, "grad_norm": 9.676228523254395, "learning_rate": 1.4885255317407587e-06, "loss": 0.4169, "step": 14834 }, { "epoch": 0.7504173200465375, "grad_norm": 8.517115592956543, "learning_rate": 1.4879542070285408e-06, "loss": 0.3335, "step": 14835 }, { "epoch": 0.7504679042946026, "grad_norm": 11.146038055419922, "learning_rate": 1.4873829728132598e-06, "loss": 0.3877, "step": 14836 }, { "epoch": 0.7505184885426678, "grad_norm": 14.890772819519043, "learning_rate": 1.4868118291096333e-06, "loss": 0.5144, "step": 14837 }, { "epoch": 0.7505690727907329, "grad_norm": 10.095544815063477, "learning_rate": 1.4862407759323828e-06, "loss": 0.3591, "step": 14838 }, { "epoch": 0.7506196570387981, "grad_norm": 16.592653274536133, "learning_rate": 1.4856698132962194e-06, "loss": 0.6885, "step": 14839 }, { "epoch": 0.7506702412868632, "grad_norm": 11.6250638961792, "learning_rate": 1.4850989412158552e-06, "loss": 0.4256, "step": 14840 }, { "epoch": 0.7507208255349285, "grad_norm": 11.23167896270752, "learning_rate": 1.484528159706003e-06, "loss": 0.416, "step": 14841 }, { "epoch": 0.7507714097829936, "grad_norm": 11.549622535705566, "learning_rate": 1.4839574687813684e-06, "loss": 0.3499, "step": 14842 }, { "epoch": 0.7508219940310588, "grad_norm": 6.407125473022461, "learning_rate": 1.483386868456656e-06, "loss": 0.3905, "step": 14843 }, { "epoch": 0.7508725782791239, "grad_norm": 8.571065902709961, "learning_rate": 1.4828163587465715e-06, "loss": 0.401, "step": 14844 }, { "epoch": 0.750923162527189, "grad_norm": 10.036276817321777, "learning_rate": 1.4822459396658146e-06, "loss": 0.3785, "step": 14845 }, { "epoch": 0.7509737467752542, "grad_norm": 4.7865095138549805, "learning_rate": 1.4816756112290808e-06, "loss": 0.4626, "step": 14846 }, { "epoch": 0.7510243310233193, "grad_norm": 5.559704303741455, "learning_rate": 1.48110537345107e-06, "loss": 0.3297, "step": 14847 }, { "epoch": 0.7510749152713845, "grad_norm": 6.736371040344238, "learning_rate": 1.4805352263464756e-06, "loss": 0.438, "step": 14848 }, { "epoch": 0.7511254995194496, "grad_norm": 12.23424243927002, "learning_rate": 1.4799651699299843e-06, "loss": 0.477, "step": 14849 }, { "epoch": 0.7511760837675148, "grad_norm": 5.010958671569824, "learning_rate": 1.4793952042162906e-06, "loss": 0.307, "step": 14850 }, { "epoch": 0.7512266680155799, "grad_norm": 5.51995849609375, "learning_rate": 1.4788253292200782e-06, "loss": 0.3852, "step": 14851 }, { "epoch": 0.7512772522636451, "grad_norm": 12.779106140136719, "learning_rate": 1.4782555449560304e-06, "loss": 0.457, "step": 14852 }, { "epoch": 0.7513278365117102, "grad_norm": 6.342163562774658, "learning_rate": 1.4776858514388326e-06, "loss": 0.3202, "step": 14853 }, { "epoch": 0.7513784207597755, "grad_norm": 5.716601848602295, "learning_rate": 1.4771162486831624e-06, "loss": 0.3903, "step": 14854 }, { "epoch": 0.7514290050078406, "grad_norm": 5.7942094802856445, "learning_rate": 1.4765467367036962e-06, "loss": 0.3499, "step": 14855 }, { "epoch": 0.7514795892559057, "grad_norm": 6.429262161254883, "learning_rate": 1.4759773155151107e-06, "loss": 0.4114, "step": 14856 }, { "epoch": 0.7515301735039709, "grad_norm": 8.473152160644531, "learning_rate": 1.4754079851320784e-06, "loss": 0.4825, "step": 14857 }, { "epoch": 0.751580757752036, "grad_norm": 7.613724231719971, "learning_rate": 1.4748387455692674e-06, "loss": 0.3289, "step": 14858 }, { "epoch": 0.7516313420001012, "grad_norm": 28.329025268554688, "learning_rate": 1.474269596841349e-06, "loss": 0.2817, "step": 14859 }, { "epoch": 0.7516819262481663, "grad_norm": 4.885599136352539, "learning_rate": 1.4737005389629871e-06, "loss": 0.241, "step": 14860 }, { "epoch": 0.7517325104962315, "grad_norm": 7.6955695152282715, "learning_rate": 1.4731315719488448e-06, "loss": 0.4147, "step": 14861 }, { "epoch": 0.7517830947442966, "grad_norm": 10.910896301269531, "learning_rate": 1.472562695813583e-06, "loss": 0.4045, "step": 14862 }, { "epoch": 0.7518336789923618, "grad_norm": 9.10681438446045, "learning_rate": 1.4719939105718588e-06, "loss": 0.4159, "step": 14863 }, { "epoch": 0.7518842632404269, "grad_norm": 13.726490020751953, "learning_rate": 1.4714252162383314e-06, "loss": 0.4159, "step": 14864 }, { "epoch": 0.751934847488492, "grad_norm": 9.780487060546875, "learning_rate": 1.4708566128276536e-06, "loss": 0.3987, "step": 14865 }, { "epoch": 0.7519854317365572, "grad_norm": 10.748156547546387, "learning_rate": 1.4702881003544744e-06, "loss": 0.4365, "step": 14866 }, { "epoch": 0.7520360159846224, "grad_norm": 9.025887489318848, "learning_rate": 1.4697196788334473e-06, "loss": 0.4346, "step": 14867 }, { "epoch": 0.7520866002326876, "grad_norm": 4.969483375549316, "learning_rate": 1.469151348279217e-06, "loss": 0.3457, "step": 14868 }, { "epoch": 0.7521371844807527, "grad_norm": 5.868284225463867, "learning_rate": 1.4685831087064277e-06, "loss": 0.3504, "step": 14869 }, { "epoch": 0.7521877687288179, "grad_norm": 9.907990455627441, "learning_rate": 1.4680149601297206e-06, "loss": 0.3405, "step": 14870 }, { "epoch": 0.752238352976883, "grad_norm": 8.073403358459473, "learning_rate": 1.4674469025637383e-06, "loss": 0.3207, "step": 14871 }, { "epoch": 0.7522889372249482, "grad_norm": 6.007373809814453, "learning_rate": 1.4668789360231167e-06, "loss": 0.4417, "step": 14872 }, { "epoch": 0.7523395214730133, "grad_norm": 6.745015621185303, "learning_rate": 1.4663110605224895e-06, "loss": 0.3699, "step": 14873 }, { "epoch": 0.7523901057210785, "grad_norm": 5.438542366027832, "learning_rate": 1.4657432760764946e-06, "loss": 0.3033, "step": 14874 }, { "epoch": 0.7524406899691436, "grad_norm": 10.328701972961426, "learning_rate": 1.465175582699755e-06, "loss": 0.3736, "step": 14875 }, { "epoch": 0.7524912742172087, "grad_norm": 9.477069854736328, "learning_rate": 1.4646079804069047e-06, "loss": 0.3798, "step": 14876 }, { "epoch": 0.7525418584652739, "grad_norm": 11.884367942810059, "learning_rate": 1.464040469212567e-06, "loss": 0.4942, "step": 14877 }, { "epoch": 0.752592442713339, "grad_norm": 8.313353538513184, "learning_rate": 1.4634730491313642e-06, "loss": 0.2966, "step": 14878 }, { "epoch": 0.7526430269614042, "grad_norm": 8.974392890930176, "learning_rate": 1.4629057201779201e-06, "loss": 0.3999, "step": 14879 }, { "epoch": 0.7526936112094694, "grad_norm": 10.214669227600098, "learning_rate": 1.4623384823668523e-06, "loss": 0.4306, "step": 14880 }, { "epoch": 0.7527441954575346, "grad_norm": 10.890959739685059, "learning_rate": 1.4617713357127755e-06, "loss": 0.5643, "step": 14881 }, { "epoch": 0.7527947797055997, "grad_norm": 5.8232340812683105, "learning_rate": 1.4612042802303067e-06, "loss": 0.3585, "step": 14882 }, { "epoch": 0.7528453639536649, "grad_norm": 8.552274703979492, "learning_rate": 1.4606373159340558e-06, "loss": 0.2905, "step": 14883 }, { "epoch": 0.75289594820173, "grad_norm": 5.872816562652588, "learning_rate": 1.460070442838631e-06, "loss": 0.4259, "step": 14884 }, { "epoch": 0.7529465324497951, "grad_norm": 7.785974025726318, "learning_rate": 1.4595036609586421e-06, "loss": 0.3805, "step": 14885 }, { "epoch": 0.7529971166978603, "grad_norm": 10.154622077941895, "learning_rate": 1.4589369703086937e-06, "loss": 0.4953, "step": 14886 }, { "epoch": 0.7530477009459254, "grad_norm": 8.022967338562012, "learning_rate": 1.458370370903383e-06, "loss": 0.4185, "step": 14887 }, { "epoch": 0.7530982851939906, "grad_norm": 13.370048522949219, "learning_rate": 1.4578038627573149e-06, "loss": 0.4855, "step": 14888 }, { "epoch": 0.7531488694420557, "grad_norm": 7.088362693786621, "learning_rate": 1.4572374458850857e-06, "loss": 0.298, "step": 14889 }, { "epoch": 0.7531994536901209, "grad_norm": 13.980711936950684, "learning_rate": 1.456671120301289e-06, "loss": 0.4461, "step": 14890 }, { "epoch": 0.753250037938186, "grad_norm": 11.150559425354004, "learning_rate": 1.45610488602052e-06, "loss": 0.353, "step": 14891 }, { "epoch": 0.7533006221862512, "grad_norm": 6.072051048278809, "learning_rate": 1.4555387430573686e-06, "loss": 0.3861, "step": 14892 }, { "epoch": 0.7533512064343163, "grad_norm": 8.046246528625488, "learning_rate": 1.4549726914264207e-06, "loss": 0.2985, "step": 14893 }, { "epoch": 0.7534017906823816, "grad_norm": 12.779648780822754, "learning_rate": 1.4544067311422655e-06, "loss": 0.5021, "step": 14894 }, { "epoch": 0.7534523749304467, "grad_norm": 7.90360689163208, "learning_rate": 1.453840862219485e-06, "loss": 0.354, "step": 14895 }, { "epoch": 0.7535029591785118, "grad_norm": 5.107559680938721, "learning_rate": 1.4532750846726585e-06, "loss": 0.3847, "step": 14896 }, { "epoch": 0.753553543426577, "grad_norm": 12.725069046020508, "learning_rate": 1.4527093985163677e-06, "loss": 0.5187, "step": 14897 }, { "epoch": 0.7536041276746421, "grad_norm": 5.6300764083862305, "learning_rate": 1.4521438037651876e-06, "loss": 0.3173, "step": 14898 }, { "epoch": 0.7536547119227073, "grad_norm": 11.081826210021973, "learning_rate": 1.451578300433692e-06, "loss": 0.4398, "step": 14899 }, { "epoch": 0.7537052961707724, "grad_norm": 7.10498571395874, "learning_rate": 1.4510128885364528e-06, "loss": 0.3876, "step": 14900 }, { "epoch": 0.7537558804188376, "grad_norm": 7.627249240875244, "learning_rate": 1.4504475680880375e-06, "loss": 0.3083, "step": 14901 }, { "epoch": 0.7538064646669027, "grad_norm": 9.697725296020508, "learning_rate": 1.449882339103017e-06, "loss": 0.3809, "step": 14902 }, { "epoch": 0.7538570489149679, "grad_norm": 6.491043567657471, "learning_rate": 1.4493172015959534e-06, "loss": 0.4079, "step": 14903 }, { "epoch": 0.753907633163033, "grad_norm": 6.7275285720825195, "learning_rate": 1.448752155581409e-06, "loss": 0.3689, "step": 14904 }, { "epoch": 0.7539582174110981, "grad_norm": 5.142379283905029, "learning_rate": 1.4481872010739424e-06, "loss": 0.3645, "step": 14905 }, { "epoch": 0.7540088016591633, "grad_norm": 5.778349876403809, "learning_rate": 1.4476223380881137e-06, "loss": 0.3398, "step": 14906 }, { "epoch": 0.7540593859072285, "grad_norm": 12.403534889221191, "learning_rate": 1.4470575666384772e-06, "loss": 0.3152, "step": 14907 }, { "epoch": 0.7541099701552937, "grad_norm": 6.700560569763184, "learning_rate": 1.4464928867395843e-06, "loss": 0.3405, "step": 14908 }, { "epoch": 0.7541605544033588, "grad_norm": 6.767527103424072, "learning_rate": 1.4459282984059881e-06, "loss": 0.3091, "step": 14909 }, { "epoch": 0.754211138651424, "grad_norm": 5.339169025421143, "learning_rate": 1.4453638016522347e-06, "loss": 0.235, "step": 14910 }, { "epoch": 0.7542617228994891, "grad_norm": 9.383837699890137, "learning_rate": 1.4447993964928686e-06, "loss": 0.4762, "step": 14911 }, { "epoch": 0.7543123071475543, "grad_norm": 9.072469711303711, "learning_rate": 1.4442350829424384e-06, "loss": 0.4367, "step": 14912 }, { "epoch": 0.7543628913956194, "grad_norm": 7.627745151519775, "learning_rate": 1.4436708610154782e-06, "loss": 0.4745, "step": 14913 }, { "epoch": 0.7544134756436846, "grad_norm": 14.123945236206055, "learning_rate": 1.4431067307265318e-06, "loss": 0.4608, "step": 14914 }, { "epoch": 0.7544640598917497, "grad_norm": 8.835165977478027, "learning_rate": 1.442542692090133e-06, "loss": 0.4502, "step": 14915 }, { "epoch": 0.7545146441398148, "grad_norm": 5.9742817878723145, "learning_rate": 1.441978745120815e-06, "loss": 0.309, "step": 14916 }, { "epoch": 0.75456522838788, "grad_norm": 11.651106834411621, "learning_rate": 1.4414148898331126e-06, "loss": 0.5337, "step": 14917 }, { "epoch": 0.7546158126359451, "grad_norm": 17.651165008544922, "learning_rate": 1.4408511262415526e-06, "loss": 0.362, "step": 14918 }, { "epoch": 0.7546663968840103, "grad_norm": 5.871435165405273, "learning_rate": 1.440287454360661e-06, "loss": 0.397, "step": 14919 }, { "epoch": 0.7547169811320755, "grad_norm": 18.13224983215332, "learning_rate": 1.4397238742049652e-06, "loss": 0.4301, "step": 14920 }, { "epoch": 0.7547675653801407, "grad_norm": 5.6581244468688965, "learning_rate": 1.4391603857889857e-06, "loss": 0.4032, "step": 14921 }, { "epoch": 0.7548181496282058, "grad_norm": 12.737980842590332, "learning_rate": 1.4385969891272405e-06, "loss": 0.4806, "step": 14922 }, { "epoch": 0.754868733876271, "grad_norm": 8.692070007324219, "learning_rate": 1.4380336842342502e-06, "loss": 0.4366, "step": 14923 }, { "epoch": 0.7549193181243361, "grad_norm": 7.126385688781738, "learning_rate": 1.4374704711245286e-06, "loss": 0.3271, "step": 14924 }, { "epoch": 0.7549699023724012, "grad_norm": 6.320101261138916, "learning_rate": 1.4369073498125873e-06, "loss": 0.39, "step": 14925 }, { "epoch": 0.7550204866204664, "grad_norm": 4.873156547546387, "learning_rate": 1.436344320312938e-06, "loss": 0.3343, "step": 14926 }, { "epoch": 0.7550710708685315, "grad_norm": 7.791199207305908, "learning_rate": 1.4357813826400874e-06, "loss": 0.4203, "step": 14927 }, { "epoch": 0.7551216551165967, "grad_norm": 10.310543060302734, "learning_rate": 1.4352185368085403e-06, "loss": 0.4235, "step": 14928 }, { "epoch": 0.7551722393646618, "grad_norm": 12.240446090698242, "learning_rate": 1.434655782832803e-06, "loss": 0.4929, "step": 14929 }, { "epoch": 0.755222823612727, "grad_norm": 7.632538318634033, "learning_rate": 1.4340931207273739e-06, "loss": 0.2854, "step": 14930 }, { "epoch": 0.7552734078607921, "grad_norm": 10.434699058532715, "learning_rate": 1.4335305505067504e-06, "loss": 0.3914, "step": 14931 }, { "epoch": 0.7553239921088573, "grad_norm": 8.690917015075684, "learning_rate": 1.4329680721854321e-06, "loss": 0.3695, "step": 14932 }, { "epoch": 0.7553745763569224, "grad_norm": 7.089659690856934, "learning_rate": 1.4324056857779106e-06, "loss": 0.3804, "step": 14933 }, { "epoch": 0.7554251606049877, "grad_norm": 11.556652069091797, "learning_rate": 1.4318433912986756e-06, "loss": 0.4586, "step": 14934 }, { "epoch": 0.7554757448530528, "grad_norm": 6.31752872467041, "learning_rate": 1.4312811887622203e-06, "loss": 0.4075, "step": 14935 }, { "epoch": 0.7555263291011179, "grad_norm": 9.367860794067383, "learning_rate": 1.4307190781830276e-06, "loss": 0.42, "step": 14936 }, { "epoch": 0.7555769133491831, "grad_norm": 8.029258728027344, "learning_rate": 1.4301570595755827e-06, "loss": 0.4829, "step": 14937 }, { "epoch": 0.7556274975972482, "grad_norm": 12.161231994628906, "learning_rate": 1.42959513295437e-06, "loss": 0.4843, "step": 14938 }, { "epoch": 0.7556780818453134, "grad_norm": 7.194753170013428, "learning_rate": 1.429033298333864e-06, "loss": 0.3823, "step": 14939 }, { "epoch": 0.7557286660933785, "grad_norm": 8.928918838500977, "learning_rate": 1.4284715557285462e-06, "loss": 0.4121, "step": 14940 }, { "epoch": 0.7557792503414437, "grad_norm": 7.1181721687316895, "learning_rate": 1.4279099051528894e-06, "loss": 0.3609, "step": 14941 }, { "epoch": 0.7558298345895088, "grad_norm": 11.276713371276855, "learning_rate": 1.4273483466213668e-06, "loss": 0.4378, "step": 14942 }, { "epoch": 0.755880418837574, "grad_norm": 6.995193958282471, "learning_rate": 1.4267868801484458e-06, "loss": 0.419, "step": 14943 }, { "epoch": 0.7559310030856391, "grad_norm": 10.288825988769531, "learning_rate": 1.426225505748598e-06, "loss": 0.4082, "step": 14944 }, { "epoch": 0.7559815873337042, "grad_norm": 8.385077476501465, "learning_rate": 1.4256642234362873e-06, "loss": 0.4031, "step": 14945 }, { "epoch": 0.7560321715817694, "grad_norm": 5.424323558807373, "learning_rate": 1.4251030332259736e-06, "loss": 0.3286, "step": 14946 }, { "epoch": 0.7560827558298346, "grad_norm": 9.46384334564209, "learning_rate": 1.4245419351321222e-06, "loss": 0.4619, "step": 14947 }, { "epoch": 0.7561333400778998, "grad_norm": 7.161284446716309, "learning_rate": 1.4239809291691887e-06, "loss": 0.3493, "step": 14948 }, { "epoch": 0.7561839243259649, "grad_norm": 9.924628257751465, "learning_rate": 1.4234200153516276e-06, "loss": 0.369, "step": 14949 }, { "epoch": 0.7562345085740301, "grad_norm": 8.437727928161621, "learning_rate": 1.4228591936938956e-06, "loss": 0.3842, "step": 14950 }, { "epoch": 0.7562850928220952, "grad_norm": 5.969032287597656, "learning_rate": 1.4222984642104416e-06, "loss": 0.3096, "step": 14951 }, { "epoch": 0.7563356770701604, "grad_norm": 7.931235313415527, "learning_rate": 1.4217378269157145e-06, "loss": 0.4035, "step": 14952 }, { "epoch": 0.7563862613182255, "grad_norm": 6.7977495193481445, "learning_rate": 1.4211772818241608e-06, "loss": 0.3452, "step": 14953 }, { "epoch": 0.7564368455662906, "grad_norm": 9.913314819335938, "learning_rate": 1.4206168289502232e-06, "loss": 0.417, "step": 14954 }, { "epoch": 0.7564874298143558, "grad_norm": 4.334290027618408, "learning_rate": 1.420056468308345e-06, "loss": 0.3079, "step": 14955 }, { "epoch": 0.7565380140624209, "grad_norm": 11.406370162963867, "learning_rate": 1.4194961999129654e-06, "loss": 0.4814, "step": 14956 }, { "epoch": 0.7565885983104861, "grad_norm": 9.030753135681152, "learning_rate": 1.4189360237785181e-06, "loss": 0.3914, "step": 14957 }, { "epoch": 0.7566391825585512, "grad_norm": 7.617579936981201, "learning_rate": 1.4183759399194419e-06, "loss": 0.3461, "step": 14958 }, { "epoch": 0.7566897668066164, "grad_norm": 12.007124900817871, "learning_rate": 1.4178159483501664e-06, "loss": 0.4247, "step": 14959 }, { "epoch": 0.7567403510546816, "grad_norm": 9.637225151062012, "learning_rate": 1.4172560490851201e-06, "loss": 0.428, "step": 14960 }, { "epoch": 0.7567909353027468, "grad_norm": 9.864255905151367, "learning_rate": 1.416696242138733e-06, "loss": 0.4225, "step": 14961 }, { "epoch": 0.7568415195508119, "grad_norm": 6.224785804748535, "learning_rate": 1.416136527525429e-06, "loss": 0.3387, "step": 14962 }, { "epoch": 0.7568921037988771, "grad_norm": 11.712081909179688, "learning_rate": 1.415576905259628e-06, "loss": 0.4905, "step": 14963 }, { "epoch": 0.7569426880469422, "grad_norm": 6.128419399261475, "learning_rate": 1.4150173753557562e-06, "loss": 0.3101, "step": 14964 }, { "epoch": 0.7569932722950073, "grad_norm": 5.94162130355835, "learning_rate": 1.4144579378282252e-06, "loss": 0.3871, "step": 14965 }, { "epoch": 0.7570438565430725, "grad_norm": 6.560313701629639, "learning_rate": 1.4138985926914517e-06, "loss": 0.3466, "step": 14966 }, { "epoch": 0.7570944407911376, "grad_norm": 7.383056163787842, "learning_rate": 1.4133393399598505e-06, "loss": 0.3798, "step": 14967 }, { "epoch": 0.7571450250392028, "grad_norm": 8.978482246398926, "learning_rate": 1.4127801796478318e-06, "loss": 0.313, "step": 14968 }, { "epoch": 0.7571956092872679, "grad_norm": 5.374279022216797, "learning_rate": 1.4122211117698016e-06, "loss": 0.3127, "step": 14969 }, { "epoch": 0.7572461935353331, "grad_norm": 6.119334697723389, "learning_rate": 1.411662136340169e-06, "loss": 0.3659, "step": 14970 }, { "epoch": 0.7572967777833982, "grad_norm": 7.912184715270996, "learning_rate": 1.4111032533733359e-06, "loss": 0.425, "step": 14971 }, { "epoch": 0.7573473620314634, "grad_norm": 7.587843418121338, "learning_rate": 1.4105444628837017e-06, "loss": 0.3528, "step": 14972 }, { "epoch": 0.7573979462795285, "grad_norm": 8.423054695129395, "learning_rate": 1.4099857648856685e-06, "loss": 0.4011, "step": 14973 }, { "epoch": 0.7574485305275938, "grad_norm": 10.793651580810547, "learning_rate": 1.4094271593936304e-06, "loss": 0.5039, "step": 14974 }, { "epoch": 0.7574991147756589, "grad_norm": 4.9332756996154785, "learning_rate": 1.4088686464219803e-06, "loss": 0.2874, "step": 14975 }, { "epoch": 0.757549699023724, "grad_norm": 8.568146705627441, "learning_rate": 1.4083102259851144e-06, "loss": 0.4317, "step": 14976 }, { "epoch": 0.7576002832717892, "grad_norm": 14.924084663391113, "learning_rate": 1.407751898097417e-06, "loss": 0.4866, "step": 14977 }, { "epoch": 0.7576508675198543, "grad_norm": 9.369763374328613, "learning_rate": 1.4071936627732745e-06, "loss": 0.4053, "step": 14978 }, { "epoch": 0.7577014517679195, "grad_norm": 5.796175956726074, "learning_rate": 1.4066355200270753e-06, "loss": 0.3546, "step": 14979 }, { "epoch": 0.7577520360159846, "grad_norm": 11.325761795043945, "learning_rate": 1.4060774698731994e-06, "loss": 0.4286, "step": 14980 }, { "epoch": 0.7578026202640498, "grad_norm": 8.626540184020996, "learning_rate": 1.4055195123260247e-06, "loss": 0.3529, "step": 14981 }, { "epoch": 0.7578532045121149, "grad_norm": 17.827056884765625, "learning_rate": 1.4049616473999317e-06, "loss": 0.5271, "step": 14982 }, { "epoch": 0.75790378876018, "grad_norm": 8.358205795288086, "learning_rate": 1.4044038751092931e-06, "loss": 0.3657, "step": 14983 }, { "epoch": 0.7579543730082452, "grad_norm": 7.896731853485107, "learning_rate": 1.4038461954684808e-06, "loss": 0.4125, "step": 14984 }, { "epoch": 0.7580049572563103, "grad_norm": 6.703960418701172, "learning_rate": 1.4032886084918673e-06, "loss": 0.4173, "step": 14985 }, { "epoch": 0.7580555415043755, "grad_norm": 18.70965576171875, "learning_rate": 1.4027311141938189e-06, "loss": 0.4626, "step": 14986 }, { "epoch": 0.7581061257524407, "grad_norm": 5.69752836227417, "learning_rate": 1.4021737125886992e-06, "loss": 0.4121, "step": 14987 }, { "epoch": 0.7581567100005059, "grad_norm": 12.5890474319458, "learning_rate": 1.4016164036908741e-06, "loss": 0.4412, "step": 14988 }, { "epoch": 0.758207294248571, "grad_norm": 6.916454315185547, "learning_rate": 1.4010591875147028e-06, "loss": 0.4564, "step": 14989 }, { "epoch": 0.7582578784966362, "grad_norm": 4.683608531951904, "learning_rate": 1.4005020640745436e-06, "loss": 0.3689, "step": 14990 }, { "epoch": 0.7583084627447013, "grad_norm": 18.018774032592773, "learning_rate": 1.3999450333847515e-06, "loss": 0.4603, "step": 14991 }, { "epoch": 0.7583590469927665, "grad_norm": 7.377313613891602, "learning_rate": 1.3993880954596795e-06, "loss": 0.4149, "step": 14992 }, { "epoch": 0.7584096312408316, "grad_norm": 8.904309272766113, "learning_rate": 1.3988312503136804e-06, "loss": 0.3726, "step": 14993 }, { "epoch": 0.7584602154888967, "grad_norm": 9.778801918029785, "learning_rate": 1.3982744979611018e-06, "loss": 0.4465, "step": 14994 }, { "epoch": 0.7585107997369619, "grad_norm": 10.210646629333496, "learning_rate": 1.3977178384162882e-06, "loss": 0.3938, "step": 14995 }, { "epoch": 0.758561383985027, "grad_norm": 5.389896392822266, "learning_rate": 1.3971612716935867e-06, "loss": 0.3191, "step": 14996 }, { "epoch": 0.7586119682330922, "grad_norm": 8.133292198181152, "learning_rate": 1.3966047978073366e-06, "loss": 0.389, "step": 14997 }, { "epoch": 0.7586625524811573, "grad_norm": 7.340292453765869, "learning_rate": 1.3960484167718756e-06, "loss": 0.3641, "step": 14998 }, { "epoch": 0.7587131367292225, "grad_norm": 6.154655933380127, "learning_rate": 1.3954921286015438e-06, "loss": 0.4148, "step": 14999 }, { "epoch": 0.7587637209772877, "grad_norm": 4.766892433166504, "learning_rate": 1.3949359333106732e-06, "loss": 0.3467, "step": 15000 }, { "epoch": 0.7588143052253529, "grad_norm": 4.036284923553467, "learning_rate": 1.3943798309135947e-06, "loss": 0.2435, "step": 15001 }, { "epoch": 0.758864889473418, "grad_norm": 13.542011260986328, "learning_rate": 1.393823821424642e-06, "loss": 0.3935, "step": 15002 }, { "epoch": 0.7589154737214832, "grad_norm": 7.3205718994140625, "learning_rate": 1.3932679048581377e-06, "loss": 0.3883, "step": 15003 }, { "epoch": 0.7589660579695483, "grad_norm": 5.171292304992676, "learning_rate": 1.3927120812284055e-06, "loss": 0.2049, "step": 15004 }, { "epoch": 0.7590166422176134, "grad_norm": 7.873561382293701, "learning_rate": 1.392156350549772e-06, "loss": 0.3641, "step": 15005 }, { "epoch": 0.7590672264656786, "grad_norm": 6.288478851318359, "learning_rate": 1.391600712836555e-06, "loss": 0.347, "step": 15006 }, { "epoch": 0.7591178107137437, "grad_norm": 13.166367530822754, "learning_rate": 1.3910451681030702e-06, "loss": 0.4463, "step": 15007 }, { "epoch": 0.7591683949618089, "grad_norm": 10.21530532836914, "learning_rate": 1.3904897163636361e-06, "loss": 0.4054, "step": 15008 }, { "epoch": 0.759218979209874, "grad_norm": 8.364840507507324, "learning_rate": 1.3899343576325636e-06, "loss": 0.3822, "step": 15009 }, { "epoch": 0.7592695634579392, "grad_norm": 10.658342361450195, "learning_rate": 1.3893790919241613e-06, "loss": 0.3235, "step": 15010 }, { "epoch": 0.7593201477060043, "grad_norm": 9.132390022277832, "learning_rate": 1.3888239192527398e-06, "loss": 0.4033, "step": 15011 }, { "epoch": 0.7593707319540695, "grad_norm": 8.90821647644043, "learning_rate": 1.3882688396326039e-06, "loss": 0.39, "step": 15012 }, { "epoch": 0.7594213162021347, "grad_norm": 13.874796867370605, "learning_rate": 1.3877138530780548e-06, "loss": 0.4853, "step": 15013 }, { "epoch": 0.7594719004501999, "grad_norm": 5.642544746398926, "learning_rate": 1.3871589596033957e-06, "loss": 0.3733, "step": 15014 }, { "epoch": 0.759522484698265, "grad_norm": 7.799587249755859, "learning_rate": 1.3866041592229257e-06, "loss": 0.4232, "step": 15015 }, { "epoch": 0.7595730689463301, "grad_norm": 5.615656852722168, "learning_rate": 1.3860494519509354e-06, "loss": 0.3418, "step": 15016 }, { "epoch": 0.7596236531943953, "grad_norm": 8.097926139831543, "learning_rate": 1.385494837801723e-06, "loss": 0.3899, "step": 15017 }, { "epoch": 0.7596742374424604, "grad_norm": 32.74551010131836, "learning_rate": 1.3849403167895787e-06, "loss": 0.5285, "step": 15018 }, { "epoch": 0.7597248216905256, "grad_norm": 5.8478498458862305, "learning_rate": 1.3843858889287887e-06, "loss": 0.3881, "step": 15019 }, { "epoch": 0.7597754059385907, "grad_norm": 10.871435165405273, "learning_rate": 1.3838315542336427e-06, "loss": 0.5233, "step": 15020 }, { "epoch": 0.7598259901866559, "grad_norm": 12.35649585723877, "learning_rate": 1.3832773127184229e-06, "loss": 0.5906, "step": 15021 }, { "epoch": 0.759876574434721, "grad_norm": 9.20333480834961, "learning_rate": 1.38272316439741e-06, "loss": 0.4733, "step": 15022 }, { "epoch": 0.7599271586827862, "grad_norm": 8.693550109863281, "learning_rate": 1.3821691092848849e-06, "loss": 0.4156, "step": 15023 }, { "epoch": 0.7599777429308513, "grad_norm": 9.466565132141113, "learning_rate": 1.3816151473951233e-06, "loss": 0.3881, "step": 15024 }, { "epoch": 0.7600283271789164, "grad_norm": 6.628236770629883, "learning_rate": 1.3810612787423983e-06, "loss": 0.3094, "step": 15025 }, { "epoch": 0.7600789114269816, "grad_norm": 6.0313005447387695, "learning_rate": 1.3805075033409843e-06, "loss": 0.3739, "step": 15026 }, { "epoch": 0.7601294956750468, "grad_norm": 7.761792182922363, "learning_rate": 1.3799538212051495e-06, "loss": 0.4319, "step": 15027 }, { "epoch": 0.760180079923112, "grad_norm": 12.95032024383545, "learning_rate": 1.3794002323491612e-06, "loss": 0.617, "step": 15028 }, { "epoch": 0.7602306641711771, "grad_norm": 9.347360610961914, "learning_rate": 1.3788467367872832e-06, "loss": 0.3884, "step": 15029 }, { "epoch": 0.7602812484192423, "grad_norm": 11.003752708435059, "learning_rate": 1.3782933345337768e-06, "loss": 0.4811, "step": 15030 }, { "epoch": 0.7603318326673074, "grad_norm": 11.874184608459473, "learning_rate": 1.377740025602905e-06, "loss": 0.4416, "step": 15031 }, { "epoch": 0.7603824169153726, "grad_norm": 6.209583759307861, "learning_rate": 1.3771868100089236e-06, "loss": 0.3967, "step": 15032 }, { "epoch": 0.7604330011634377, "grad_norm": 8.100687026977539, "learning_rate": 1.3766336877660858e-06, "loss": 0.433, "step": 15033 }, { "epoch": 0.7604835854115028, "grad_norm": 12.58523941040039, "learning_rate": 1.3760806588886477e-06, "loss": 0.4864, "step": 15034 }, { "epoch": 0.760534169659568, "grad_norm": 7.295474529266357, "learning_rate": 1.3755277233908575e-06, "loss": 0.4346, "step": 15035 }, { "epoch": 0.7605847539076331, "grad_norm": 9.579859733581543, "learning_rate": 1.3749748812869623e-06, "loss": 0.4227, "step": 15036 }, { "epoch": 0.7606353381556983, "grad_norm": 10.464656829833984, "learning_rate": 1.3744221325912095e-06, "loss": 0.3095, "step": 15037 }, { "epoch": 0.7606859224037634, "grad_norm": 11.217713356018066, "learning_rate": 1.3738694773178413e-06, "loss": 0.4061, "step": 15038 }, { "epoch": 0.7607365066518286, "grad_norm": 6.471160888671875, "learning_rate": 1.3733169154810983e-06, "loss": 0.4432, "step": 15039 }, { "epoch": 0.7607870908998938, "grad_norm": 7.372366428375244, "learning_rate": 1.3727644470952168e-06, "loss": 0.4496, "step": 15040 }, { "epoch": 0.760837675147959, "grad_norm": 5.188950538635254, "learning_rate": 1.3722120721744375e-06, "loss": 0.3089, "step": 15041 }, { "epoch": 0.7608882593960241, "grad_norm": 25.137100219726562, "learning_rate": 1.3716597907329871e-06, "loss": 0.54, "step": 15042 }, { "epoch": 0.7609388436440893, "grad_norm": 9.428749084472656, "learning_rate": 1.3711076027851016e-06, "loss": 0.3371, "step": 15043 }, { "epoch": 0.7609894278921544, "grad_norm": 7.840385913848877, "learning_rate": 1.3705555083450084e-06, "loss": 0.3346, "step": 15044 }, { "epoch": 0.7610400121402195, "grad_norm": 11.102513313293457, "learning_rate": 1.3700035074269314e-06, "loss": 0.4208, "step": 15045 }, { "epoch": 0.7610905963882847, "grad_norm": 16.009784698486328, "learning_rate": 1.3694516000450975e-06, "loss": 0.4455, "step": 15046 }, { "epoch": 0.7611411806363498, "grad_norm": 10.810518264770508, "learning_rate": 1.3688997862137272e-06, "loss": 0.3832, "step": 15047 }, { "epoch": 0.761191764884415, "grad_norm": 9.183967590332031, "learning_rate": 1.3683480659470367e-06, "loss": 0.4659, "step": 15048 }, { "epoch": 0.7612423491324801, "grad_norm": 7.871180057525635, "learning_rate": 1.3677964392592462e-06, "loss": 0.4033, "step": 15049 }, { "epoch": 0.7612929333805453, "grad_norm": 11.125831604003906, "learning_rate": 1.3672449061645682e-06, "loss": 0.5127, "step": 15050 }, { "epoch": 0.7613435176286104, "grad_norm": 9.937795639038086, "learning_rate": 1.366693466677213e-06, "loss": 0.3854, "step": 15051 }, { "epoch": 0.7613941018766756, "grad_norm": 10.106324195861816, "learning_rate": 1.3661421208113928e-06, "loss": 0.4248, "step": 15052 }, { "epoch": 0.7614446861247408, "grad_norm": 13.369110107421875, "learning_rate": 1.3655908685813134e-06, "loss": 0.4347, "step": 15053 }, { "epoch": 0.761495270372806, "grad_norm": 9.01835823059082, "learning_rate": 1.3650397100011781e-06, "loss": 0.3537, "step": 15054 }, { "epoch": 0.7615458546208711, "grad_norm": 8.849299430847168, "learning_rate": 1.3644886450851901e-06, "loss": 0.3065, "step": 15055 }, { "epoch": 0.7615964388689362, "grad_norm": 7.525153636932373, "learning_rate": 1.3639376738475479e-06, "loss": 0.3894, "step": 15056 }, { "epoch": 0.7616470231170014, "grad_norm": 4.6479902267456055, "learning_rate": 1.3633867963024483e-06, "loss": 0.4549, "step": 15057 }, { "epoch": 0.7616976073650665, "grad_norm": 9.986743927001953, "learning_rate": 1.3628360124640882e-06, "loss": 0.4034, "step": 15058 }, { "epoch": 0.7617481916131317, "grad_norm": 18.917619705200195, "learning_rate": 1.3622853223466592e-06, "loss": 0.6185, "step": 15059 }, { "epoch": 0.7617987758611968, "grad_norm": 8.133234024047852, "learning_rate": 1.3617347259643492e-06, "loss": 0.4047, "step": 15060 }, { "epoch": 0.761849360109262, "grad_norm": 8.787738800048828, "learning_rate": 1.3611842233313487e-06, "loss": 0.4678, "step": 15061 }, { "epoch": 0.7618999443573271, "grad_norm": 7.798722267150879, "learning_rate": 1.360633814461842e-06, "loss": 0.3863, "step": 15062 }, { "epoch": 0.7619505286053923, "grad_norm": 15.628806114196777, "learning_rate": 1.3600834993700097e-06, "loss": 0.4209, "step": 15063 }, { "epoch": 0.7620011128534574, "grad_norm": 6.6562371253967285, "learning_rate": 1.3595332780700355e-06, "loss": 0.3918, "step": 15064 }, { "epoch": 0.7620516971015225, "grad_norm": 11.289750099182129, "learning_rate": 1.3589831505760954e-06, "loss": 0.4766, "step": 15065 }, { "epoch": 0.7621022813495877, "grad_norm": 11.032139778137207, "learning_rate": 1.3584331169023635e-06, "loss": 0.4146, "step": 15066 }, { "epoch": 0.7621528655976529, "grad_norm": 9.582428932189941, "learning_rate": 1.357883177063017e-06, "loss": 0.411, "step": 15067 }, { "epoch": 0.7622034498457181, "grad_norm": 4.714700698852539, "learning_rate": 1.3573333310722208e-06, "loss": 0.3713, "step": 15068 }, { "epoch": 0.7622540340937832, "grad_norm": 7.480051517486572, "learning_rate": 1.3567835789441487e-06, "loss": 0.3744, "step": 15069 }, { "epoch": 0.7623046183418484, "grad_norm": 5.4670305252075195, "learning_rate": 1.3562339206929626e-06, "loss": 0.3584, "step": 15070 }, { "epoch": 0.7623552025899135, "grad_norm": 9.11711597442627, "learning_rate": 1.3556843563328266e-06, "loss": 0.4641, "step": 15071 }, { "epoch": 0.7624057868379787, "grad_norm": 8.697843551635742, "learning_rate": 1.3551348858779035e-06, "loss": 0.3752, "step": 15072 }, { "epoch": 0.7624563710860438, "grad_norm": 10.520230293273926, "learning_rate": 1.3545855093423504e-06, "loss": 0.4649, "step": 15073 }, { "epoch": 0.762506955334109, "grad_norm": 5.360499858856201, "learning_rate": 1.3540362267403222e-06, "loss": 0.3781, "step": 15074 }, { "epoch": 0.7625575395821741, "grad_norm": 5.199623107910156, "learning_rate": 1.3534870380859756e-06, "loss": 0.3272, "step": 15075 }, { "epoch": 0.7626081238302392, "grad_norm": 9.548141479492188, "learning_rate": 1.3529379433934602e-06, "loss": 0.3842, "step": 15076 }, { "epoch": 0.7626587080783044, "grad_norm": 9.652058601379395, "learning_rate": 1.3523889426769249e-06, "loss": 0.4703, "step": 15077 }, { "epoch": 0.7627092923263695, "grad_norm": 7.067525386810303, "learning_rate": 1.3518400359505145e-06, "loss": 0.2597, "step": 15078 }, { "epoch": 0.7627598765744347, "grad_norm": 7.633057594299316, "learning_rate": 1.3512912232283776e-06, "loss": 0.438, "step": 15079 }, { "epoch": 0.7628104608224999, "grad_norm": 5.208908557891846, "learning_rate": 1.3507425045246497e-06, "loss": 0.3288, "step": 15080 }, { "epoch": 0.7628610450705651, "grad_norm": 5.457091331481934, "learning_rate": 1.3501938798534748e-06, "loss": 0.2999, "step": 15081 }, { "epoch": 0.7629116293186302, "grad_norm": 10.794702529907227, "learning_rate": 1.3496453492289873e-06, "loss": 0.4697, "step": 15082 }, { "epoch": 0.7629622135666954, "grad_norm": 13.12646770477295, "learning_rate": 1.3490969126653215e-06, "loss": 0.5125, "step": 15083 }, { "epoch": 0.7630127978147605, "grad_norm": 10.1783447265625, "learning_rate": 1.3485485701766105e-06, "loss": 0.3948, "step": 15084 }, { "epoch": 0.7630633820628256, "grad_norm": 6.089788913726807, "learning_rate": 1.3480003217769838e-06, "loss": 0.366, "step": 15085 }, { "epoch": 0.7631139663108908, "grad_norm": 10.705410957336426, "learning_rate": 1.3474521674805657e-06, "loss": 0.3589, "step": 15086 }, { "epoch": 0.7631645505589559, "grad_norm": 7.156097888946533, "learning_rate": 1.346904107301485e-06, "loss": 0.2804, "step": 15087 }, { "epoch": 0.7632151348070211, "grad_norm": 4.494011402130127, "learning_rate": 1.3463561412538613e-06, "loss": 0.3326, "step": 15088 }, { "epoch": 0.7632657190550862, "grad_norm": 5.91640043258667, "learning_rate": 1.3458082693518131e-06, "loss": 0.3931, "step": 15089 }, { "epoch": 0.7633163033031514, "grad_norm": 10.086298942565918, "learning_rate": 1.3452604916094614e-06, "loss": 0.3788, "step": 15090 }, { "epoch": 0.7633668875512165, "grad_norm": 6.918492317199707, "learning_rate": 1.3447128080409188e-06, "loss": 0.3526, "step": 15091 }, { "epoch": 0.7634174717992817, "grad_norm": 6.803122520446777, "learning_rate": 1.3441652186602988e-06, "loss": 0.4205, "step": 15092 }, { "epoch": 0.7634680560473469, "grad_norm": 10.340448379516602, "learning_rate": 1.3436177234817106e-06, "loss": 0.3717, "step": 15093 }, { "epoch": 0.763518640295412, "grad_norm": 13.575384140014648, "learning_rate": 1.3430703225192615e-06, "loss": 0.4502, "step": 15094 }, { "epoch": 0.7635692245434772, "grad_norm": 8.057694435119629, "learning_rate": 1.342523015787056e-06, "loss": 0.4304, "step": 15095 }, { "epoch": 0.7636198087915423, "grad_norm": 11.221478462219238, "learning_rate": 1.3419758032991998e-06, "loss": 0.4262, "step": 15096 }, { "epoch": 0.7636703930396075, "grad_norm": 4.5363945960998535, "learning_rate": 1.3414286850697917e-06, "loss": 0.3335, "step": 15097 }, { "epoch": 0.7637209772876726, "grad_norm": 5.640843868255615, "learning_rate": 1.340881661112927e-06, "loss": 0.3261, "step": 15098 }, { "epoch": 0.7637715615357378, "grad_norm": 9.742980003356934, "learning_rate": 1.3403347314427062e-06, "loss": 0.4441, "step": 15099 }, { "epoch": 0.7638221457838029, "grad_norm": 16.519102096557617, "learning_rate": 1.3397878960732196e-06, "loss": 0.6273, "step": 15100 }, { "epoch": 0.7638727300318681, "grad_norm": 7.799116611480713, "learning_rate": 1.3392411550185563e-06, "loss": 0.388, "step": 15101 }, { "epoch": 0.7639233142799332, "grad_norm": 16.664175033569336, "learning_rate": 1.3386945082928076e-06, "loss": 0.4325, "step": 15102 }, { "epoch": 0.7639738985279984, "grad_norm": 6.4323625564575195, "learning_rate": 1.3381479559100585e-06, "loss": 0.3984, "step": 15103 }, { "epoch": 0.7640244827760635, "grad_norm": 6.961854457855225, "learning_rate": 1.3376014978843898e-06, "loss": 0.3142, "step": 15104 }, { "epoch": 0.7640750670241286, "grad_norm": 4.960893630981445, "learning_rate": 1.3370551342298877e-06, "loss": 0.3809, "step": 15105 }, { "epoch": 0.7641256512721938, "grad_norm": 7.669372081756592, "learning_rate": 1.3365088649606239e-06, "loss": 0.3179, "step": 15106 }, { "epoch": 0.764176235520259, "grad_norm": 12.831276893615723, "learning_rate": 1.3359626900906797e-06, "loss": 0.4871, "step": 15107 }, { "epoch": 0.7642268197683242, "grad_norm": 14.903634071350098, "learning_rate": 1.3354166096341265e-06, "loss": 0.5488, "step": 15108 }, { "epoch": 0.7642774040163893, "grad_norm": 4.483879089355469, "learning_rate": 1.3348706236050341e-06, "loss": 0.3824, "step": 15109 }, { "epoch": 0.7643279882644545, "grad_norm": 7.480718612670898, "learning_rate": 1.3343247320174751e-06, "loss": 0.393, "step": 15110 }, { "epoch": 0.7643785725125196, "grad_norm": 6.747853755950928, "learning_rate": 1.3337789348855135e-06, "loss": 0.3901, "step": 15111 }, { "epoch": 0.7644291567605848, "grad_norm": 12.557388305664062, "learning_rate": 1.3332332322232134e-06, "loss": 0.4951, "step": 15112 }, { "epoch": 0.7644797410086499, "grad_norm": 13.356884002685547, "learning_rate": 1.3326876240446346e-06, "loss": 0.5068, "step": 15113 }, { "epoch": 0.764530325256715, "grad_norm": 14.326774597167969, "learning_rate": 1.3321421103638388e-06, "loss": 0.4248, "step": 15114 }, { "epoch": 0.7645809095047802, "grad_norm": 7.585657596588135, "learning_rate": 1.3315966911948818e-06, "loss": 0.4167, "step": 15115 }, { "epoch": 0.7646314937528453, "grad_norm": 7.250566482543945, "learning_rate": 1.331051366551816e-06, "loss": 0.3182, "step": 15116 }, { "epoch": 0.7646820780009105, "grad_norm": 7.327362060546875, "learning_rate": 1.330506136448696e-06, "loss": 0.3503, "step": 15117 }, { "epoch": 0.7647326622489756, "grad_norm": 17.689193725585938, "learning_rate": 1.3299610008995695e-06, "loss": 0.4373, "step": 15118 }, { "epoch": 0.7647832464970408, "grad_norm": 6.7595720291137695, "learning_rate": 1.3294159599184837e-06, "loss": 0.3513, "step": 15119 }, { "epoch": 0.764833830745106, "grad_norm": 8.135649681091309, "learning_rate": 1.328871013519482e-06, "loss": 0.3971, "step": 15120 }, { "epoch": 0.7648844149931712, "grad_norm": 4.792632579803467, "learning_rate": 1.328326161716606e-06, "loss": 0.3162, "step": 15121 }, { "epoch": 0.7649349992412363, "grad_norm": 6.944751262664795, "learning_rate": 1.3277814045238978e-06, "loss": 0.3902, "step": 15122 }, { "epoch": 0.7649855834893015, "grad_norm": 8.386624336242676, "learning_rate": 1.3272367419553928e-06, "loss": 0.3285, "step": 15123 }, { "epoch": 0.7650361677373666, "grad_norm": 14.924320220947266, "learning_rate": 1.3266921740251242e-06, "loss": 0.4089, "step": 15124 }, { "epoch": 0.7650867519854317, "grad_norm": 8.141056060791016, "learning_rate": 1.3261477007471274e-06, "loss": 0.3576, "step": 15125 }, { "epoch": 0.7651373362334969, "grad_norm": 6.414693355560303, "learning_rate": 1.3256033221354302e-06, "loss": 0.378, "step": 15126 }, { "epoch": 0.765187920481562, "grad_norm": 6.963554382324219, "learning_rate": 1.325059038204059e-06, "loss": 0.3784, "step": 15127 }, { "epoch": 0.7652385047296272, "grad_norm": 7.802412033081055, "learning_rate": 1.3245148489670417e-06, "loss": 0.4237, "step": 15128 }, { "epoch": 0.7652890889776923, "grad_norm": 6.924469470977783, "learning_rate": 1.3239707544383984e-06, "loss": 0.4159, "step": 15129 }, { "epoch": 0.7653396732257575, "grad_norm": 8.558237075805664, "learning_rate": 1.3234267546321483e-06, "loss": 0.388, "step": 15130 }, { "epoch": 0.7653902574738226, "grad_norm": 6.675323486328125, "learning_rate": 1.3228828495623136e-06, "loss": 0.3512, "step": 15131 }, { "epoch": 0.7654408417218878, "grad_norm": 9.28010082244873, "learning_rate": 1.3223390392429048e-06, "loss": 0.5674, "step": 15132 }, { "epoch": 0.765491425969953, "grad_norm": 13.131092071533203, "learning_rate": 1.3217953236879338e-06, "loss": 0.5582, "step": 15133 }, { "epoch": 0.7655420102180182, "grad_norm": 7.570291042327881, "learning_rate": 1.3212517029114152e-06, "loss": 0.4366, "step": 15134 }, { "epoch": 0.7655925944660833, "grad_norm": 7.406688690185547, "learning_rate": 1.3207081769273538e-06, "loss": 0.3979, "step": 15135 }, { "epoch": 0.7656431787141484, "grad_norm": 8.780911445617676, "learning_rate": 1.3201647457497552e-06, "loss": 0.3822, "step": 15136 }, { "epoch": 0.7656937629622136, "grad_norm": 8.332941055297852, "learning_rate": 1.3196214093926236e-06, "loss": 0.4756, "step": 15137 }, { "epoch": 0.7657443472102787, "grad_norm": 10.477208137512207, "learning_rate": 1.3190781678699588e-06, "loss": 0.3877, "step": 15138 }, { "epoch": 0.7657949314583439, "grad_norm": 9.916157722473145, "learning_rate": 1.318535021195757e-06, "loss": 0.4943, "step": 15139 }, { "epoch": 0.765845515706409, "grad_norm": 6.078400611877441, "learning_rate": 1.3179919693840175e-06, "loss": 0.4067, "step": 15140 }, { "epoch": 0.7658960999544742, "grad_norm": 13.416290283203125, "learning_rate": 1.317449012448731e-06, "loss": 0.4101, "step": 15141 }, { "epoch": 0.7659466842025393, "grad_norm": 12.629631996154785, "learning_rate": 1.3169061504038872e-06, "loss": 0.4249, "step": 15142 }, { "epoch": 0.7659972684506045, "grad_norm": 6.64370059967041, "learning_rate": 1.3163633832634775e-06, "loss": 0.4327, "step": 15143 }, { "epoch": 0.7660478526986696, "grad_norm": 21.489028930664062, "learning_rate": 1.3158207110414856e-06, "loss": 0.4077, "step": 15144 }, { "epoch": 0.7660984369467347, "grad_norm": 10.404350280761719, "learning_rate": 1.3152781337518956e-06, "loss": 0.4039, "step": 15145 }, { "epoch": 0.7661490211947999, "grad_norm": 5.310420513153076, "learning_rate": 1.3147356514086878e-06, "loss": 0.3563, "step": 15146 }, { "epoch": 0.7661996054428651, "grad_norm": 6.581665992736816, "learning_rate": 1.3141932640258393e-06, "loss": 0.4051, "step": 15147 }, { "epoch": 0.7662501896909303, "grad_norm": 8.649230003356934, "learning_rate": 1.313650971617329e-06, "loss": 0.3456, "step": 15148 }, { "epoch": 0.7663007739389954, "grad_norm": 10.392671585083008, "learning_rate": 1.3131087741971298e-06, "loss": 0.4344, "step": 15149 }, { "epoch": 0.7663513581870606, "grad_norm": 7.498449325561523, "learning_rate": 1.312566671779212e-06, "loss": 0.4798, "step": 15150 }, { "epoch": 0.7664019424351257, "grad_norm": 5.9659271240234375, "learning_rate": 1.3120246643775425e-06, "loss": 0.3099, "step": 15151 }, { "epoch": 0.7664525266831909, "grad_norm": 19.95838165283203, "learning_rate": 1.3114827520060908e-06, "loss": 0.3911, "step": 15152 }, { "epoch": 0.766503110931256, "grad_norm": 6.682919025421143, "learning_rate": 1.3109409346788199e-06, "loss": 0.2852, "step": 15153 }, { "epoch": 0.7665536951793211, "grad_norm": 19.059097290039062, "learning_rate": 1.3103992124096887e-06, "loss": 0.445, "step": 15154 }, { "epoch": 0.7666042794273863, "grad_norm": 14.168395042419434, "learning_rate": 1.3098575852126598e-06, "loss": 0.5952, "step": 15155 }, { "epoch": 0.7666548636754514, "grad_norm": 7.401108741760254, "learning_rate": 1.3093160531016874e-06, "loss": 0.4126, "step": 15156 }, { "epoch": 0.7667054479235166, "grad_norm": 11.333815574645996, "learning_rate": 1.308774616090726e-06, "loss": 0.443, "step": 15157 }, { "epoch": 0.7667560321715817, "grad_norm": 7.8437581062316895, "learning_rate": 1.308233274193727e-06, "loss": 0.395, "step": 15158 }, { "epoch": 0.7668066164196469, "grad_norm": 9.359270095825195, "learning_rate": 1.307692027424638e-06, "loss": 0.4122, "step": 15159 }, { "epoch": 0.7668572006677121, "grad_norm": 10.313189506530762, "learning_rate": 1.3071508757974082e-06, "loss": 0.4683, "step": 15160 }, { "epoch": 0.7669077849157773, "grad_norm": 11.773839950561523, "learning_rate": 1.3066098193259807e-06, "loss": 0.4151, "step": 15161 }, { "epoch": 0.7669583691638424, "grad_norm": 10.967419624328613, "learning_rate": 1.3060688580242959e-06, "loss": 0.3561, "step": 15162 }, { "epoch": 0.7670089534119076, "grad_norm": 13.243101119995117, "learning_rate": 1.3055279919062959e-06, "loss": 0.516, "step": 15163 }, { "epoch": 0.7670595376599727, "grad_norm": 8.506142616271973, "learning_rate": 1.3049872209859154e-06, "loss": 0.3458, "step": 15164 }, { "epoch": 0.7671101219080378, "grad_norm": 6.10151481628418, "learning_rate": 1.3044465452770883e-06, "loss": 0.2775, "step": 15165 }, { "epoch": 0.767160706156103, "grad_norm": 5.233381271362305, "learning_rate": 1.3039059647937486e-06, "loss": 0.3207, "step": 15166 }, { "epoch": 0.7672112904041681, "grad_norm": 16.059423446655273, "learning_rate": 1.3033654795498247e-06, "loss": 0.5339, "step": 15167 }, { "epoch": 0.7672618746522333, "grad_norm": 7.624317169189453, "learning_rate": 1.302825089559242e-06, "loss": 0.4363, "step": 15168 }, { "epoch": 0.7673124589002984, "grad_norm": 11.48382568359375, "learning_rate": 1.3022847948359302e-06, "loss": 0.3931, "step": 15169 }, { "epoch": 0.7673630431483636, "grad_norm": 6.165412902832031, "learning_rate": 1.3017445953938058e-06, "loss": 0.3421, "step": 15170 }, { "epoch": 0.7674136273964287, "grad_norm": 5.45672607421875, "learning_rate": 1.3012044912467886e-06, "loss": 0.3621, "step": 15171 }, { "epoch": 0.7674642116444939, "grad_norm": 21.980628967285156, "learning_rate": 1.3006644824087993e-06, "loss": 0.5861, "step": 15172 }, { "epoch": 0.7675147958925591, "grad_norm": 10.585972785949707, "learning_rate": 1.3001245688937509e-06, "loss": 0.516, "step": 15173 }, { "epoch": 0.7675653801406243, "grad_norm": 22.879066467285156, "learning_rate": 1.2995847507155546e-06, "loss": 0.5621, "step": 15174 }, { "epoch": 0.7676159643886894, "grad_norm": 6.5587663650512695, "learning_rate": 1.2990450278881222e-06, "loss": 0.4283, "step": 15175 }, { "epoch": 0.7676665486367545, "grad_norm": 8.057040214538574, "learning_rate": 1.2985054004253605e-06, "loss": 0.4058, "step": 15176 }, { "epoch": 0.7677171328848197, "grad_norm": 10.200668334960938, "learning_rate": 1.2979658683411721e-06, "loss": 0.3066, "step": 15177 }, { "epoch": 0.7677677171328848, "grad_norm": 8.438502311706543, "learning_rate": 1.2974264316494633e-06, "loss": 0.3706, "step": 15178 }, { "epoch": 0.76781830138095, "grad_norm": 12.553958892822266, "learning_rate": 1.2968870903641323e-06, "loss": 0.4687, "step": 15179 }, { "epoch": 0.7678688856290151, "grad_norm": 10.077652931213379, "learning_rate": 1.2963478444990747e-06, "loss": 0.3423, "step": 15180 }, { "epoch": 0.7679194698770803, "grad_norm": 13.115939140319824, "learning_rate": 1.2958086940681886e-06, "loss": 0.4626, "step": 15181 }, { "epoch": 0.7679700541251454, "grad_norm": 10.303215980529785, "learning_rate": 1.2952696390853659e-06, "loss": 0.4409, "step": 15182 }, { "epoch": 0.7680206383732106, "grad_norm": 10.658330917358398, "learning_rate": 1.2947306795644959e-06, "loss": 0.4835, "step": 15183 }, { "epoch": 0.7680712226212757, "grad_norm": 8.837122917175293, "learning_rate": 1.2941918155194666e-06, "loss": 0.3517, "step": 15184 }, { "epoch": 0.7681218068693408, "grad_norm": 8.493464469909668, "learning_rate": 1.2936530469641616e-06, "loss": 0.4241, "step": 15185 }, { "epoch": 0.7681723911174061, "grad_norm": 6.9211554527282715, "learning_rate": 1.2931143739124668e-06, "loss": 0.3568, "step": 15186 }, { "epoch": 0.7682229753654712, "grad_norm": 5.484762191772461, "learning_rate": 1.292575796378261e-06, "loss": 0.326, "step": 15187 }, { "epoch": 0.7682735596135364, "grad_norm": 6.042430400848389, "learning_rate": 1.2920373143754223e-06, "loss": 0.368, "step": 15188 }, { "epoch": 0.7683241438616015, "grad_norm": 5.936187267303467, "learning_rate": 1.2914989279178236e-06, "loss": 0.3622, "step": 15189 }, { "epoch": 0.7683747281096667, "grad_norm": 20.678157806396484, "learning_rate": 1.2909606370193423e-06, "loss": 0.5197, "step": 15190 }, { "epoch": 0.7684253123577318, "grad_norm": 5.998834609985352, "learning_rate": 1.290422441693846e-06, "loss": 0.3947, "step": 15191 }, { "epoch": 0.768475896605797, "grad_norm": 4.667958736419678, "learning_rate": 1.289884341955202e-06, "loss": 0.3419, "step": 15192 }, { "epoch": 0.7685264808538621, "grad_norm": 8.634632110595703, "learning_rate": 1.2893463378172783e-06, "loss": 0.4436, "step": 15193 }, { "epoch": 0.7685770651019272, "grad_norm": 5.26629114151001, "learning_rate": 1.2888084292939374e-06, "loss": 0.3603, "step": 15194 }, { "epoch": 0.7686276493499924, "grad_norm": 6.48960542678833, "learning_rate": 1.2882706163990389e-06, "loss": 0.4016, "step": 15195 }, { "epoch": 0.7686782335980575, "grad_norm": 14.849303245544434, "learning_rate": 1.287732899146441e-06, "loss": 0.4608, "step": 15196 }, { "epoch": 0.7687288178461227, "grad_norm": 7.5672807693481445, "learning_rate": 1.2871952775499985e-06, "loss": 0.345, "step": 15197 }, { "epoch": 0.7687794020941878, "grad_norm": 9.454764366149902, "learning_rate": 1.2866577516235673e-06, "loss": 0.4204, "step": 15198 }, { "epoch": 0.768829986342253, "grad_norm": 9.009150505065918, "learning_rate": 1.2861203213809965e-06, "loss": 0.4354, "step": 15199 }, { "epoch": 0.7688805705903182, "grad_norm": 6.431634426116943, "learning_rate": 1.285582986836133e-06, "loss": 0.356, "step": 15200 }, { "epoch": 0.7689311548383834, "grad_norm": 7.03129768371582, "learning_rate": 1.2850457480028256e-06, "loss": 0.3995, "step": 15201 }, { "epoch": 0.7689817390864485, "grad_norm": 16.819135665893555, "learning_rate": 1.2845086048949162e-06, "loss": 0.5649, "step": 15202 }, { "epoch": 0.7690323233345137, "grad_norm": 7.589794635772705, "learning_rate": 1.2839715575262446e-06, "loss": 0.3946, "step": 15203 }, { "epoch": 0.7690829075825788, "grad_norm": 6.512434959411621, "learning_rate": 1.2834346059106512e-06, "loss": 0.4424, "step": 15204 }, { "epoch": 0.769133491830644, "grad_norm": 9.241302490234375, "learning_rate": 1.282897750061971e-06, "loss": 0.3513, "step": 15205 }, { "epoch": 0.7691840760787091, "grad_norm": 4.597738742828369, "learning_rate": 1.2823609899940364e-06, "loss": 0.4355, "step": 15206 }, { "epoch": 0.7692346603267742, "grad_norm": 6.119915008544922, "learning_rate": 1.2818243257206808e-06, "loss": 0.4413, "step": 15207 }, { "epoch": 0.7692852445748394, "grad_norm": 4.927823066711426, "learning_rate": 1.2812877572557325e-06, "loss": 0.3115, "step": 15208 }, { "epoch": 0.7693358288229045, "grad_norm": 8.236883163452148, "learning_rate": 1.2807512846130144e-06, "loss": 0.4077, "step": 15209 }, { "epoch": 0.7693864130709697, "grad_norm": 4.845740795135498, "learning_rate": 1.2802149078063531e-06, "loss": 0.2658, "step": 15210 }, { "epoch": 0.7694369973190348, "grad_norm": 6.6536359786987305, "learning_rate": 1.2796786268495687e-06, "loss": 0.308, "step": 15211 }, { "epoch": 0.7694875815671, "grad_norm": 5.077649116516113, "learning_rate": 1.279142441756479e-06, "loss": 0.3897, "step": 15212 }, { "epoch": 0.7695381658151652, "grad_norm": 6.364086151123047, "learning_rate": 1.2786063525409026e-06, "loss": 0.4186, "step": 15213 }, { "epoch": 0.7695887500632304, "grad_norm": 14.117876052856445, "learning_rate": 1.2780703592166516e-06, "loss": 0.3962, "step": 15214 }, { "epoch": 0.7696393343112955, "grad_norm": 6.557659149169922, "learning_rate": 1.277534461797536e-06, "loss": 0.4184, "step": 15215 }, { "epoch": 0.7696899185593606, "grad_norm": 6.466202735900879, "learning_rate": 1.2769986602973683e-06, "loss": 0.4164, "step": 15216 }, { "epoch": 0.7697405028074258, "grad_norm": 6.07158088684082, "learning_rate": 1.2764629547299517e-06, "loss": 0.4079, "step": 15217 }, { "epoch": 0.7697910870554909, "grad_norm": 9.257575988769531, "learning_rate": 1.2759273451090897e-06, "loss": 0.4159, "step": 15218 }, { "epoch": 0.7698416713035561, "grad_norm": 4.059061527252197, "learning_rate": 1.2753918314485863e-06, "loss": 0.268, "step": 15219 }, { "epoch": 0.7698922555516212, "grad_norm": 6.5752739906311035, "learning_rate": 1.2748564137622393e-06, "loss": 0.3911, "step": 15220 }, { "epoch": 0.7699428397996864, "grad_norm": 10.353604316711426, "learning_rate": 1.2743210920638443e-06, "loss": 0.4108, "step": 15221 }, { "epoch": 0.7699934240477515, "grad_norm": 4.853091716766357, "learning_rate": 1.2737858663671966e-06, "loss": 0.2936, "step": 15222 }, { "epoch": 0.7700440082958167, "grad_norm": 4.565465450286865, "learning_rate": 1.2732507366860859e-06, "loss": 0.2818, "step": 15223 }, { "epoch": 0.7700945925438818, "grad_norm": 9.511603355407715, "learning_rate": 1.2727157030343012e-06, "loss": 0.3382, "step": 15224 }, { "epoch": 0.7701451767919469, "grad_norm": 8.077251434326172, "learning_rate": 1.2721807654256314e-06, "loss": 0.3829, "step": 15225 }, { "epoch": 0.7701957610400122, "grad_norm": 6.580447673797607, "learning_rate": 1.2716459238738588e-06, "loss": 0.3964, "step": 15226 }, { "epoch": 0.7702463452880773, "grad_norm": 8.270194053649902, "learning_rate": 1.2711111783927638e-06, "loss": 0.284, "step": 15227 }, { "epoch": 0.7702969295361425, "grad_norm": 10.388955116271973, "learning_rate": 1.2705765289961287e-06, "loss": 0.373, "step": 15228 }, { "epoch": 0.7703475137842076, "grad_norm": 5.691619396209717, "learning_rate": 1.2700419756977284e-06, "loss": 0.4124, "step": 15229 }, { "epoch": 0.7703980980322728, "grad_norm": 7.347615718841553, "learning_rate": 1.2695075185113354e-06, "loss": 0.219, "step": 15230 }, { "epoch": 0.7704486822803379, "grad_norm": 5.829145908355713, "learning_rate": 1.2689731574507242e-06, "loss": 0.3251, "step": 15231 }, { "epoch": 0.7704992665284031, "grad_norm": 7.034029006958008, "learning_rate": 1.268438892529663e-06, "loss": 0.3202, "step": 15232 }, { "epoch": 0.7705498507764682, "grad_norm": 9.284274101257324, "learning_rate": 1.2679047237619168e-06, "loss": 0.383, "step": 15233 }, { "epoch": 0.7706004350245333, "grad_norm": 6.8087077140808105, "learning_rate": 1.2673706511612548e-06, "loss": 0.3836, "step": 15234 }, { "epoch": 0.7706510192725985, "grad_norm": 4.59658145904541, "learning_rate": 1.2668366747414324e-06, "loss": 0.4188, "step": 15235 }, { "epoch": 0.7707016035206636, "grad_norm": 8.728375434875488, "learning_rate": 1.2663027945162126e-06, "loss": 0.3661, "step": 15236 }, { "epoch": 0.7707521877687288, "grad_norm": 14.77270793914795, "learning_rate": 1.265769010499352e-06, "loss": 0.4405, "step": 15237 }, { "epoch": 0.7708027720167939, "grad_norm": 14.282989501953125, "learning_rate": 1.2652353227046022e-06, "loss": 0.4229, "step": 15238 }, { "epoch": 0.7708533562648591, "grad_norm": 5.94038200378418, "learning_rate": 1.2647017311457193e-06, "loss": 0.4112, "step": 15239 }, { "epoch": 0.7709039405129243, "grad_norm": 7.083755016326904, "learning_rate": 1.2641682358364505e-06, "loss": 0.4079, "step": 15240 }, { "epoch": 0.7709545247609895, "grad_norm": 9.156221389770508, "learning_rate": 1.263634836790541e-06, "loss": 0.3854, "step": 15241 }, { "epoch": 0.7710051090090546, "grad_norm": 12.630029678344727, "learning_rate": 1.2631015340217383e-06, "loss": 0.3347, "step": 15242 }, { "epoch": 0.7710556932571198, "grad_norm": 9.40141773223877, "learning_rate": 1.2625683275437833e-06, "loss": 0.4419, "step": 15243 }, { "epoch": 0.7711062775051849, "grad_norm": 9.027447700500488, "learning_rate": 1.2620352173704131e-06, "loss": 0.4672, "step": 15244 }, { "epoch": 0.77115686175325, "grad_norm": 12.966643333435059, "learning_rate": 1.2615022035153685e-06, "loss": 0.41, "step": 15245 }, { "epoch": 0.7712074460013152, "grad_norm": 7.424588680267334, "learning_rate": 1.2609692859923823e-06, "loss": 0.2853, "step": 15246 }, { "epoch": 0.7712580302493803, "grad_norm": 8.378579139709473, "learning_rate": 1.2604364648151862e-06, "loss": 0.3924, "step": 15247 }, { "epoch": 0.7713086144974455, "grad_norm": 8.589715957641602, "learning_rate": 1.25990373999751e-06, "loss": 0.3943, "step": 15248 }, { "epoch": 0.7713591987455106, "grad_norm": 9.015697479248047, "learning_rate": 1.2593711115530804e-06, "loss": 0.4165, "step": 15249 }, { "epoch": 0.7714097829935758, "grad_norm": 19.518199920654297, "learning_rate": 1.2588385794956204e-06, "loss": 0.4089, "step": 15250 }, { "epoch": 0.7714603672416409, "grad_norm": 13.683637619018555, "learning_rate": 1.2583061438388555e-06, "loss": 0.5378, "step": 15251 }, { "epoch": 0.771510951489706, "grad_norm": 7.547211647033691, "learning_rate": 1.2577738045965038e-06, "loss": 0.3929, "step": 15252 }, { "epoch": 0.7715615357377713, "grad_norm": 7.48226261138916, "learning_rate": 1.2572415617822808e-06, "loss": 0.363, "step": 15253 }, { "epoch": 0.7716121199858365, "grad_norm": 10.726359367370605, "learning_rate": 1.2567094154099035e-06, "loss": 0.4511, "step": 15254 }, { "epoch": 0.7716627042339016, "grad_norm": 8.510965347290039, "learning_rate": 1.2561773654930832e-06, "loss": 0.4719, "step": 15255 }, { "epoch": 0.7717132884819667, "grad_norm": 8.081652641296387, "learning_rate": 1.255645412045528e-06, "loss": 0.3609, "step": 15256 }, { "epoch": 0.7717638727300319, "grad_norm": 6.340044021606445, "learning_rate": 1.255113555080948e-06, "loss": 0.3982, "step": 15257 }, { "epoch": 0.771814456978097, "grad_norm": 9.469043731689453, "learning_rate": 1.2545817946130457e-06, "loss": 0.3944, "step": 15258 }, { "epoch": 0.7718650412261622, "grad_norm": 10.324148178100586, "learning_rate": 1.254050130655523e-06, "loss": 0.3666, "step": 15259 }, { "epoch": 0.7719156254742273, "grad_norm": 15.335208892822266, "learning_rate": 1.2535185632220832e-06, "loss": 0.4525, "step": 15260 }, { "epoch": 0.7719662097222925, "grad_norm": 7.522150039672852, "learning_rate": 1.2529870923264193e-06, "loss": 0.4433, "step": 15261 }, { "epoch": 0.7720167939703576, "grad_norm": 15.853378295898438, "learning_rate": 1.252455717982226e-06, "loss": 0.4601, "step": 15262 }, { "epoch": 0.7720673782184228, "grad_norm": 9.762152671813965, "learning_rate": 1.2519244402031983e-06, "loss": 0.4143, "step": 15263 }, { "epoch": 0.7721179624664879, "grad_norm": 8.123900413513184, "learning_rate": 1.2513932590030248e-06, "loss": 0.4594, "step": 15264 }, { "epoch": 0.772168546714553, "grad_norm": 9.25961685180664, "learning_rate": 1.2508621743953913e-06, "loss": 0.4037, "step": 15265 }, { "epoch": 0.7722191309626183, "grad_norm": 4.845399856567383, "learning_rate": 1.2503311863939855e-06, "loss": 0.4009, "step": 15266 }, { "epoch": 0.7722697152106834, "grad_norm": 6.7091288566589355, "learning_rate": 1.2498002950124876e-06, "loss": 0.3015, "step": 15267 }, { "epoch": 0.7723202994587486, "grad_norm": 9.116556167602539, "learning_rate": 1.2492695002645766e-06, "loss": 0.3657, "step": 15268 }, { "epoch": 0.7723708837068137, "grad_norm": 7.525629997253418, "learning_rate": 1.248738802163933e-06, "loss": 0.4307, "step": 15269 }, { "epoch": 0.7724214679548789, "grad_norm": 6.155100345611572, "learning_rate": 1.2482082007242291e-06, "loss": 0.3013, "step": 15270 }, { "epoch": 0.772472052202944, "grad_norm": 10.646230697631836, "learning_rate": 1.2476776959591363e-06, "loss": 0.373, "step": 15271 }, { "epoch": 0.7725226364510092, "grad_norm": 6.198665142059326, "learning_rate": 1.2471472878823292e-06, "loss": 0.4105, "step": 15272 }, { "epoch": 0.7725732206990743, "grad_norm": 5.427029132843018, "learning_rate": 1.2466169765074687e-06, "loss": 0.3852, "step": 15273 }, { "epoch": 0.7726238049471394, "grad_norm": 10.38679313659668, "learning_rate": 1.2460867618482243e-06, "loss": 0.3701, "step": 15274 }, { "epoch": 0.7726743891952046, "grad_norm": 5.476876735687256, "learning_rate": 1.2455566439182565e-06, "loss": 0.3964, "step": 15275 }, { "epoch": 0.7727249734432697, "grad_norm": 12.222904205322266, "learning_rate": 1.2450266227312246e-06, "loss": 0.4291, "step": 15276 }, { "epoch": 0.7727755576913349, "grad_norm": 7.550114154815674, "learning_rate": 1.2444966983007884e-06, "loss": 0.4298, "step": 15277 }, { "epoch": 0.7728261419394, "grad_norm": 5.4019012451171875, "learning_rate": 1.2439668706406006e-06, "loss": 0.3583, "step": 15278 }, { "epoch": 0.7728767261874652, "grad_norm": 7.594094753265381, "learning_rate": 1.243437139764313e-06, "loss": 0.4422, "step": 15279 }, { "epoch": 0.7729273104355304, "grad_norm": 33.06927490234375, "learning_rate": 1.2429075056855783e-06, "loss": 0.6732, "step": 15280 }, { "epoch": 0.7729778946835956, "grad_norm": 8.15115737915039, "learning_rate": 1.242377968418042e-06, "loss": 0.3644, "step": 15281 }, { "epoch": 0.7730284789316607, "grad_norm": 7.353174686431885, "learning_rate": 1.2418485279753484e-06, "loss": 0.3414, "step": 15282 }, { "epoch": 0.7730790631797259, "grad_norm": 9.271461486816406, "learning_rate": 1.2413191843711419e-06, "loss": 0.5071, "step": 15283 }, { "epoch": 0.773129647427791, "grad_norm": 6.256274223327637, "learning_rate": 1.2407899376190613e-06, "loss": 0.3236, "step": 15284 }, { "epoch": 0.7731802316758561, "grad_norm": 5.653400421142578, "learning_rate": 1.2402607877327438e-06, "loss": 0.3018, "step": 15285 }, { "epoch": 0.7732308159239213, "grad_norm": 7.2026472091674805, "learning_rate": 1.2397317347258247e-06, "loss": 0.4592, "step": 15286 }, { "epoch": 0.7732814001719864, "grad_norm": 8.865629196166992, "learning_rate": 1.2392027786119365e-06, "loss": 0.3642, "step": 15287 }, { "epoch": 0.7733319844200516, "grad_norm": 6.306398868560791, "learning_rate": 1.238673919404707e-06, "loss": 0.3772, "step": 15288 }, { "epoch": 0.7733825686681167, "grad_norm": 6.513829231262207, "learning_rate": 1.238145157117767e-06, "loss": 0.4219, "step": 15289 }, { "epoch": 0.7734331529161819, "grad_norm": 5.751699447631836, "learning_rate": 1.2376164917647403e-06, "loss": 0.3832, "step": 15290 }, { "epoch": 0.773483737164247, "grad_norm": 5.945235729217529, "learning_rate": 1.2370879233592475e-06, "loss": 0.3115, "step": 15291 }, { "epoch": 0.7735343214123122, "grad_norm": 5.961421966552734, "learning_rate": 1.236559451914911e-06, "loss": 0.2715, "step": 15292 }, { "epoch": 0.7735849056603774, "grad_norm": 15.054109573364258, "learning_rate": 1.2360310774453478e-06, "loss": 0.6, "step": 15293 }, { "epoch": 0.7736354899084426, "grad_norm": 5.924994468688965, "learning_rate": 1.23550279996417e-06, "loss": 0.2298, "step": 15294 }, { "epoch": 0.7736860741565077, "grad_norm": 6.828291893005371, "learning_rate": 1.2349746194849943e-06, "loss": 0.3385, "step": 15295 }, { "epoch": 0.7737366584045728, "grad_norm": 5.228508472442627, "learning_rate": 1.234446536021428e-06, "loss": 0.407, "step": 15296 }, { "epoch": 0.773787242652638, "grad_norm": 9.55984878540039, "learning_rate": 1.2339185495870781e-06, "loss": 0.4327, "step": 15297 }, { "epoch": 0.7738378269007031, "grad_norm": 6.577387809753418, "learning_rate": 1.2333906601955536e-06, "loss": 0.3965, "step": 15298 }, { "epoch": 0.7738884111487683, "grad_norm": 7.0161333084106445, "learning_rate": 1.232862867860452e-06, "loss": 0.4277, "step": 15299 }, { "epoch": 0.7739389953968334, "grad_norm": 10.119929313659668, "learning_rate": 1.232335172595374e-06, "loss": 0.4723, "step": 15300 }, { "epoch": 0.7739895796448986, "grad_norm": 8.769207954406738, "learning_rate": 1.2318075744139202e-06, "loss": 0.3442, "step": 15301 }, { "epoch": 0.7740401638929637, "grad_norm": 10.550341606140137, "learning_rate": 1.2312800733296826e-06, "loss": 0.4209, "step": 15302 }, { "epoch": 0.7740907481410289, "grad_norm": 13.768372535705566, "learning_rate": 1.2307526693562539e-06, "loss": 0.5141, "step": 15303 }, { "epoch": 0.774141332389094, "grad_norm": 5.7700018882751465, "learning_rate": 1.2302253625072253e-06, "loss": 0.391, "step": 15304 }, { "epoch": 0.7741919166371591, "grad_norm": 10.04134750366211, "learning_rate": 1.2296981527961844e-06, "loss": 0.4275, "step": 15305 }, { "epoch": 0.7742425008852244, "grad_norm": 10.345035552978516, "learning_rate": 1.2291710402367129e-06, "loss": 0.3928, "step": 15306 }, { "epoch": 0.7742930851332895, "grad_norm": 12.226479530334473, "learning_rate": 1.228644024842398e-06, "loss": 0.4825, "step": 15307 }, { "epoch": 0.7743436693813547, "grad_norm": 10.845982551574707, "learning_rate": 1.2281171066268166e-06, "loss": 0.4385, "step": 15308 }, { "epoch": 0.7743942536294198, "grad_norm": 6.574452877044678, "learning_rate": 1.2275902856035454e-06, "loss": 0.4489, "step": 15309 }, { "epoch": 0.774444837877485, "grad_norm": 9.346010208129883, "learning_rate": 1.2270635617861625e-06, "loss": 0.3952, "step": 15310 }, { "epoch": 0.7744954221255501, "grad_norm": 11.076653480529785, "learning_rate": 1.2265369351882383e-06, "loss": 0.3744, "step": 15311 }, { "epoch": 0.7745460063736153, "grad_norm": 5.451648235321045, "learning_rate": 1.226010405823343e-06, "loss": 0.3657, "step": 15312 }, { "epoch": 0.7745965906216804, "grad_norm": 10.901994705200195, "learning_rate": 1.2254839737050433e-06, "loss": 0.504, "step": 15313 }, { "epoch": 0.7746471748697455, "grad_norm": 6.837553024291992, "learning_rate": 1.2249576388469038e-06, "loss": 0.3439, "step": 15314 }, { "epoch": 0.7746977591178107, "grad_norm": 10.890066146850586, "learning_rate": 1.2244314012624887e-06, "loss": 0.4972, "step": 15315 }, { "epoch": 0.7747483433658758, "grad_norm": 13.079233169555664, "learning_rate": 1.2239052609653567e-06, "loss": 0.4306, "step": 15316 }, { "epoch": 0.774798927613941, "grad_norm": 9.499443054199219, "learning_rate": 1.2233792179690645e-06, "loss": 0.4321, "step": 15317 }, { "epoch": 0.7748495118620061, "grad_norm": 5.911390781402588, "learning_rate": 1.2228532722871688e-06, "loss": 0.3611, "step": 15318 }, { "epoch": 0.7749000961100714, "grad_norm": 5.26800012588501, "learning_rate": 1.222327423933221e-06, "loss": 0.4143, "step": 15319 }, { "epoch": 0.7749506803581365, "grad_norm": 19.658109664916992, "learning_rate": 1.22180167292077e-06, "loss": 0.5663, "step": 15320 }, { "epoch": 0.7750012646062017, "grad_norm": 6.315598011016846, "learning_rate": 1.221276019263365e-06, "loss": 0.3755, "step": 15321 }, { "epoch": 0.7750518488542668, "grad_norm": 9.069881439208984, "learning_rate": 1.2207504629745498e-06, "loss": 0.3273, "step": 15322 }, { "epoch": 0.775102433102332, "grad_norm": 6.048861026763916, "learning_rate": 1.2202250040678664e-06, "loss": 0.2689, "step": 15323 }, { "epoch": 0.7751530173503971, "grad_norm": 14.740697860717773, "learning_rate": 1.2196996425568559e-06, "loss": 0.549, "step": 15324 }, { "epoch": 0.7752036015984622, "grad_norm": 5.499725818634033, "learning_rate": 1.2191743784550542e-06, "loss": 0.3728, "step": 15325 }, { "epoch": 0.7752541858465274, "grad_norm": 7.879622459411621, "learning_rate": 1.2186492117759952e-06, "loss": 0.4302, "step": 15326 }, { "epoch": 0.7753047700945925, "grad_norm": 8.734084129333496, "learning_rate": 1.2181241425332135e-06, "loss": 0.3865, "step": 15327 }, { "epoch": 0.7753553543426577, "grad_norm": 11.783748626708984, "learning_rate": 1.2175991707402386e-06, "loss": 0.4341, "step": 15328 }, { "epoch": 0.7754059385907228, "grad_norm": 7.438547134399414, "learning_rate": 1.2170742964105953e-06, "loss": 0.3938, "step": 15329 }, { "epoch": 0.775456522838788, "grad_norm": 5.664579391479492, "learning_rate": 1.2165495195578115e-06, "loss": 0.3826, "step": 15330 }, { "epoch": 0.7755071070868531, "grad_norm": 15.830951690673828, "learning_rate": 1.2160248401954077e-06, "loss": 0.4869, "step": 15331 }, { "epoch": 0.7755576913349183, "grad_norm": 5.144665241241455, "learning_rate": 1.215500258336903e-06, "loss": 0.3589, "step": 15332 }, { "epoch": 0.7756082755829835, "grad_norm": 7.2198591232299805, "learning_rate": 1.2149757739958167e-06, "loss": 0.3813, "step": 15333 }, { "epoch": 0.7756588598310487, "grad_norm": 7.2390265464782715, "learning_rate": 1.2144513871856628e-06, "loss": 0.3805, "step": 15334 }, { "epoch": 0.7757094440791138, "grad_norm": 10.222493171691895, "learning_rate": 1.2139270979199513e-06, "loss": 0.4917, "step": 15335 }, { "epoch": 0.7757600283271789, "grad_norm": 6.117743015289307, "learning_rate": 1.213402906212195e-06, "loss": 0.3047, "step": 15336 }, { "epoch": 0.7758106125752441, "grad_norm": 8.232308387756348, "learning_rate": 1.212878812075901e-06, "loss": 0.4385, "step": 15337 }, { "epoch": 0.7758611968233092, "grad_norm": 7.2662248611450195, "learning_rate": 1.2123548155245702e-06, "loss": 0.3692, "step": 15338 }, { "epoch": 0.7759117810713744, "grad_norm": 7.650074005126953, "learning_rate": 1.211830916571708e-06, "loss": 0.3453, "step": 15339 }, { "epoch": 0.7759623653194395, "grad_norm": 8.0014066696167, "learning_rate": 1.2113071152308138e-06, "loss": 0.4613, "step": 15340 }, { "epoch": 0.7760129495675047, "grad_norm": 8.288905143737793, "learning_rate": 1.2107834115153816e-06, "loss": 0.3648, "step": 15341 }, { "epoch": 0.7760635338155698, "grad_norm": 15.963502883911133, "learning_rate": 1.2102598054389103e-06, "loss": 0.4116, "step": 15342 }, { "epoch": 0.776114118063635, "grad_norm": 8.629494667053223, "learning_rate": 1.2097362970148907e-06, "loss": 0.5232, "step": 15343 }, { "epoch": 0.7761647023117001, "grad_norm": 17.54763412475586, "learning_rate": 1.2092128862568097e-06, "loss": 0.4852, "step": 15344 }, { "epoch": 0.7762152865597652, "grad_norm": 7.008020877838135, "learning_rate": 1.2086895731781579e-06, "loss": 0.4317, "step": 15345 }, { "epoch": 0.7762658708078305, "grad_norm": 6.946962356567383, "learning_rate": 1.208166357792418e-06, "loss": 0.4414, "step": 15346 }, { "epoch": 0.7763164550558956, "grad_norm": 10.128371238708496, "learning_rate": 1.207643240113071e-06, "loss": 0.4434, "step": 15347 }, { "epoch": 0.7763670393039608, "grad_norm": 7.939689636230469, "learning_rate": 1.2071202201535992e-06, "loss": 0.3712, "step": 15348 }, { "epoch": 0.7764176235520259, "grad_norm": 9.715721130371094, "learning_rate": 1.2065972979274776e-06, "loss": 0.4106, "step": 15349 }, { "epoch": 0.7764682078000911, "grad_norm": 7.586317539215088, "learning_rate": 1.2060744734481817e-06, "loss": 0.2527, "step": 15350 }, { "epoch": 0.7765187920481562, "grad_norm": 7.361781597137451, "learning_rate": 1.205551746729182e-06, "loss": 0.401, "step": 15351 }, { "epoch": 0.7765693762962214, "grad_norm": 16.43509864807129, "learning_rate": 1.205029117783948e-06, "loss": 0.542, "step": 15352 }, { "epoch": 0.7766199605442865, "grad_norm": 7.431994915008545, "learning_rate": 1.204506586625948e-06, "loss": 0.332, "step": 15353 }, { "epoch": 0.7766705447923516, "grad_norm": 10.0007905960083, "learning_rate": 1.2039841532686463e-06, "loss": 0.5147, "step": 15354 }, { "epoch": 0.7767211290404168, "grad_norm": 8.579285621643066, "learning_rate": 1.203461817725502e-06, "loss": 0.4358, "step": 15355 }, { "epoch": 0.7767717132884819, "grad_norm": 11.075141906738281, "learning_rate": 1.2029395800099786e-06, "loss": 0.4729, "step": 15356 }, { "epoch": 0.7768222975365471, "grad_norm": 9.235674858093262, "learning_rate": 1.2024174401355304e-06, "loss": 0.3439, "step": 15357 }, { "epoch": 0.7768728817846122, "grad_norm": 9.378275871276855, "learning_rate": 1.2018953981156123e-06, "loss": 0.2939, "step": 15358 }, { "epoch": 0.7769234660326775, "grad_norm": 6.118217945098877, "learning_rate": 1.201373453963674e-06, "loss": 0.4036, "step": 15359 }, { "epoch": 0.7769740502807426, "grad_norm": 5.225594520568848, "learning_rate": 1.200851607693168e-06, "loss": 0.4096, "step": 15360 }, { "epoch": 0.7770246345288078, "grad_norm": 11.724374771118164, "learning_rate": 1.20032985931754e-06, "loss": 0.6249, "step": 15361 }, { "epoch": 0.7770752187768729, "grad_norm": 10.290969848632812, "learning_rate": 1.199808208850234e-06, "loss": 0.3962, "step": 15362 }, { "epoch": 0.777125803024938, "grad_norm": 11.70599365234375, "learning_rate": 1.1992866563046906e-06, "loss": 0.6056, "step": 15363 }, { "epoch": 0.7771763872730032, "grad_norm": 9.243637084960938, "learning_rate": 1.198765201694349e-06, "loss": 0.4517, "step": 15364 }, { "epoch": 0.7772269715210683, "grad_norm": 6.757528781890869, "learning_rate": 1.1982438450326484e-06, "loss": 0.4864, "step": 15365 }, { "epoch": 0.7772775557691335, "grad_norm": 5.616950511932373, "learning_rate": 1.1977225863330205e-06, "loss": 0.3549, "step": 15366 }, { "epoch": 0.7773281400171986, "grad_norm": 7.437155723571777, "learning_rate": 1.1972014256088966e-06, "loss": 0.3624, "step": 15367 }, { "epoch": 0.7773787242652638, "grad_norm": 11.073457717895508, "learning_rate": 1.1966803628737073e-06, "loss": 0.367, "step": 15368 }, { "epoch": 0.7774293085133289, "grad_norm": 16.11949920654297, "learning_rate": 1.1961593981408791e-06, "loss": 0.4321, "step": 15369 }, { "epoch": 0.7774798927613941, "grad_norm": 13.511091232299805, "learning_rate": 1.1956385314238338e-06, "loss": 0.4306, "step": 15370 }, { "epoch": 0.7775304770094592, "grad_norm": 6.824562072753906, "learning_rate": 1.195117762735996e-06, "loss": 0.3772, "step": 15371 }, { "epoch": 0.7775810612575244, "grad_norm": 11.008227348327637, "learning_rate": 1.1945970920907835e-06, "loss": 0.3691, "step": 15372 }, { "epoch": 0.7776316455055896, "grad_norm": 7.5152363777160645, "learning_rate": 1.1940765195016108e-06, "loss": 0.3011, "step": 15373 }, { "epoch": 0.7776822297536548, "grad_norm": 5.479282379150391, "learning_rate": 1.1935560449818945e-06, "loss": 0.4156, "step": 15374 }, { "epoch": 0.7777328140017199, "grad_norm": 12.326203346252441, "learning_rate": 1.193035668545046e-06, "loss": 0.5507, "step": 15375 }, { "epoch": 0.777783398249785, "grad_norm": 8.004988670349121, "learning_rate": 1.1925153902044705e-06, "loss": 0.3808, "step": 15376 }, { "epoch": 0.7778339824978502, "grad_norm": 9.706332206726074, "learning_rate": 1.1919952099735783e-06, "loss": 0.4665, "step": 15377 }, { "epoch": 0.7778845667459153, "grad_norm": 7.0231099128723145, "learning_rate": 1.1914751278657717e-06, "loss": 0.3471, "step": 15378 }, { "epoch": 0.7779351509939805, "grad_norm": 13.234574317932129, "learning_rate": 1.1909551438944506e-06, "loss": 0.4537, "step": 15379 }, { "epoch": 0.7779857352420456, "grad_norm": 8.274046897888184, "learning_rate": 1.190435258073016e-06, "loss": 0.3672, "step": 15380 }, { "epoch": 0.7780363194901108, "grad_norm": 6.194300651550293, "learning_rate": 1.1899154704148636e-06, "loss": 0.2952, "step": 15381 }, { "epoch": 0.7780869037381759, "grad_norm": 7.42079496383667, "learning_rate": 1.1893957809333851e-06, "loss": 0.398, "step": 15382 }, { "epoch": 0.778137487986241, "grad_norm": 11.539898872375488, "learning_rate": 1.1888761896419742e-06, "loss": 0.5074, "step": 15383 }, { "epoch": 0.7781880722343062, "grad_norm": 10.531304359436035, "learning_rate": 1.1883566965540194e-06, "loss": 0.31, "step": 15384 }, { "epoch": 0.7782386564823713, "grad_norm": 7.151528358459473, "learning_rate": 1.1878373016829038e-06, "loss": 0.4451, "step": 15385 }, { "epoch": 0.7782892407304366, "grad_norm": 7.7189106941223145, "learning_rate": 1.1873180050420146e-06, "loss": 0.3476, "step": 15386 }, { "epoch": 0.7783398249785017, "grad_norm": 7.546360969543457, "learning_rate": 1.1867988066447311e-06, "loss": 0.3786, "step": 15387 }, { "epoch": 0.7783904092265669, "grad_norm": 7.773584842681885, "learning_rate": 1.1862797065044323e-06, "loss": 0.3149, "step": 15388 }, { "epoch": 0.778440993474632, "grad_norm": 12.697466850280762, "learning_rate": 1.1857607046344939e-06, "loss": 0.5122, "step": 15389 }, { "epoch": 0.7784915777226972, "grad_norm": 17.55600357055664, "learning_rate": 1.1852418010482868e-06, "loss": 0.473, "step": 15390 }, { "epoch": 0.7785421619707623, "grad_norm": 12.71448802947998, "learning_rate": 1.1847229957591867e-06, "loss": 0.5308, "step": 15391 }, { "epoch": 0.7785927462188275, "grad_norm": 11.95506763458252, "learning_rate": 1.1842042887805594e-06, "loss": 0.5022, "step": 15392 }, { "epoch": 0.7786433304668926, "grad_norm": 8.232932090759277, "learning_rate": 1.1836856801257695e-06, "loss": 0.3398, "step": 15393 }, { "epoch": 0.7786939147149577, "grad_norm": 7.761188507080078, "learning_rate": 1.1831671698081831e-06, "loss": 0.4011, "step": 15394 }, { "epoch": 0.7787444989630229, "grad_norm": 11.48531723022461, "learning_rate": 1.1826487578411595e-06, "loss": 0.5054, "step": 15395 }, { "epoch": 0.778795083211088, "grad_norm": 6.135137557983398, "learning_rate": 1.1821304442380565e-06, "loss": 0.3658, "step": 15396 }, { "epoch": 0.7788456674591532, "grad_norm": 12.602964401245117, "learning_rate": 1.1816122290122295e-06, "loss": 0.4699, "step": 15397 }, { "epoch": 0.7788962517072183, "grad_norm": 7.660275936126709, "learning_rate": 1.1810941121770335e-06, "loss": 0.4011, "step": 15398 }, { "epoch": 0.7789468359552836, "grad_norm": 7.899633407592773, "learning_rate": 1.1805760937458182e-06, "loss": 0.3667, "step": 15399 }, { "epoch": 0.7789974202033487, "grad_norm": 8.462803840637207, "learning_rate": 1.18005817373193e-06, "loss": 0.4799, "step": 15400 }, { "epoch": 0.7790480044514139, "grad_norm": 5.915566921234131, "learning_rate": 1.1795403521487192e-06, "loss": 0.3536, "step": 15401 }, { "epoch": 0.779098588699479, "grad_norm": 8.032379150390625, "learning_rate": 1.1790226290095224e-06, "loss": 0.4047, "step": 15402 }, { "epoch": 0.7791491729475442, "grad_norm": 22.53523063659668, "learning_rate": 1.1785050043276846e-06, "loss": 0.4398, "step": 15403 }, { "epoch": 0.7791997571956093, "grad_norm": 5.522464275360107, "learning_rate": 1.1779874781165423e-06, "loss": 0.3427, "step": 15404 }, { "epoch": 0.7792503414436744, "grad_norm": 8.185322761535645, "learning_rate": 1.17747005038943e-06, "loss": 0.3658, "step": 15405 }, { "epoch": 0.7793009256917396, "grad_norm": 6.041956424713135, "learning_rate": 1.1769527211596827e-06, "loss": 0.3669, "step": 15406 }, { "epoch": 0.7793515099398047, "grad_norm": 9.431355476379395, "learning_rate": 1.1764354904406294e-06, "loss": 0.3795, "step": 15407 }, { "epoch": 0.7794020941878699, "grad_norm": 5.931646823883057, "learning_rate": 1.1759183582455973e-06, "loss": 0.3562, "step": 15408 }, { "epoch": 0.779452678435935, "grad_norm": 7.335468292236328, "learning_rate": 1.1754013245879132e-06, "loss": 0.4112, "step": 15409 }, { "epoch": 0.7795032626840002, "grad_norm": 6.765885353088379, "learning_rate": 1.1748843894808991e-06, "loss": 0.4739, "step": 15410 }, { "epoch": 0.7795538469320653, "grad_norm": 11.452282905578613, "learning_rate": 1.174367552937874e-06, "loss": 0.4973, "step": 15411 }, { "epoch": 0.7796044311801305, "grad_norm": 11.107786178588867, "learning_rate": 1.173850814972158e-06, "loss": 0.2985, "step": 15412 }, { "epoch": 0.7796550154281957, "grad_norm": 7.658503532409668, "learning_rate": 1.1733341755970645e-06, "loss": 0.4091, "step": 15413 }, { "epoch": 0.7797055996762609, "grad_norm": 4.150890827178955, "learning_rate": 1.172817634825907e-06, "loss": 0.3618, "step": 15414 }, { "epoch": 0.779756183924326, "grad_norm": 6.730943202972412, "learning_rate": 1.172301192671994e-06, "loss": 0.3943, "step": 15415 }, { "epoch": 0.7798067681723911, "grad_norm": 7.14423131942749, "learning_rate": 1.1717848491486345e-06, "loss": 0.3886, "step": 15416 }, { "epoch": 0.7798573524204563, "grad_norm": 9.191537857055664, "learning_rate": 1.1712686042691307e-06, "loss": 0.3356, "step": 15417 }, { "epoch": 0.7799079366685214, "grad_norm": 21.99808692932129, "learning_rate": 1.1707524580467888e-06, "loss": 0.6401, "step": 15418 }, { "epoch": 0.7799585209165866, "grad_norm": 14.114009857177734, "learning_rate": 1.1702364104949071e-06, "loss": 0.4597, "step": 15419 }, { "epoch": 0.7800091051646517, "grad_norm": 7.240645408630371, "learning_rate": 1.169720461626781e-06, "loss": 0.4142, "step": 15420 }, { "epoch": 0.7800596894127169, "grad_norm": 6.630956172943115, "learning_rate": 1.1692046114557087e-06, "loss": 0.3927, "step": 15421 }, { "epoch": 0.780110273660782, "grad_norm": 5.448000431060791, "learning_rate": 1.1686888599949802e-06, "loss": 0.3115, "step": 15422 }, { "epoch": 0.7801608579088471, "grad_norm": 10.399136543273926, "learning_rate": 1.168173207257885e-06, "loss": 0.4636, "step": 15423 }, { "epoch": 0.7802114421569123, "grad_norm": 5.13521146774292, "learning_rate": 1.1676576532577115e-06, "loss": 0.329, "step": 15424 }, { "epoch": 0.7802620264049774, "grad_norm": 7.27059268951416, "learning_rate": 1.1671421980077436e-06, "loss": 0.4046, "step": 15425 }, { "epoch": 0.7803126106530427, "grad_norm": 6.823334693908691, "learning_rate": 1.1666268415212622e-06, "loss": 0.4154, "step": 15426 }, { "epoch": 0.7803631949011078, "grad_norm": 6.523692607879639, "learning_rate": 1.166111583811551e-06, "loss": 0.4005, "step": 15427 }, { "epoch": 0.780413779149173, "grad_norm": 7.721045970916748, "learning_rate": 1.165596424891881e-06, "loss": 0.5123, "step": 15428 }, { "epoch": 0.7804643633972381, "grad_norm": 6.325479507446289, "learning_rate": 1.1650813647755305e-06, "loss": 0.417, "step": 15429 }, { "epoch": 0.7805149476453033, "grad_norm": 13.901174545288086, "learning_rate": 1.1645664034757708e-06, "loss": 0.3985, "step": 15430 }, { "epoch": 0.7805655318933684, "grad_norm": 3.9137673377990723, "learning_rate": 1.1640515410058707e-06, "loss": 0.332, "step": 15431 }, { "epoch": 0.7806161161414336, "grad_norm": 6.519336223602295, "learning_rate": 1.1635367773790956e-06, "loss": 0.382, "step": 15432 }, { "epoch": 0.7806667003894987, "grad_norm": 6.244394302368164, "learning_rate": 1.163022112608712e-06, "loss": 0.4069, "step": 15433 }, { "epoch": 0.7807172846375638, "grad_norm": 9.678749084472656, "learning_rate": 1.1625075467079816e-06, "loss": 0.3903, "step": 15434 }, { "epoch": 0.780767868885629, "grad_norm": 12.31906795501709, "learning_rate": 1.1619930796901608e-06, "loss": 0.4547, "step": 15435 }, { "epoch": 0.7808184531336941, "grad_norm": 11.368358612060547, "learning_rate": 1.1614787115685094e-06, "loss": 0.4438, "step": 15436 }, { "epoch": 0.7808690373817593, "grad_norm": 7.040857791900635, "learning_rate": 1.16096444235628e-06, "loss": 0.2558, "step": 15437 }, { "epoch": 0.7809196216298244, "grad_norm": 13.877686500549316, "learning_rate": 1.1604502720667227e-06, "loss": 0.5148, "step": 15438 }, { "epoch": 0.7809702058778897, "grad_norm": 8.763984680175781, "learning_rate": 1.1599362007130892e-06, "loss": 0.4931, "step": 15439 }, { "epoch": 0.7810207901259548, "grad_norm": 9.286123275756836, "learning_rate": 1.1594222283086247e-06, "loss": 0.3599, "step": 15440 }, { "epoch": 0.78107137437402, "grad_norm": 12.370085716247559, "learning_rate": 1.158908354866573e-06, "loss": 0.5422, "step": 15441 }, { "epoch": 0.7811219586220851, "grad_norm": 9.929964065551758, "learning_rate": 1.158394580400175e-06, "loss": 0.3461, "step": 15442 }, { "epoch": 0.7811725428701503, "grad_norm": 7.508256912231445, "learning_rate": 1.1578809049226685e-06, "loss": 0.2731, "step": 15443 }, { "epoch": 0.7812231271182154, "grad_norm": 7.824328422546387, "learning_rate": 1.157367328447292e-06, "loss": 0.3671, "step": 15444 }, { "epoch": 0.7812737113662805, "grad_norm": 9.013185501098633, "learning_rate": 1.1568538509872779e-06, "loss": 0.4288, "step": 15445 }, { "epoch": 0.7813242956143457, "grad_norm": 12.117829322814941, "learning_rate": 1.1563404725558557e-06, "loss": 0.4532, "step": 15446 }, { "epoch": 0.7813748798624108, "grad_norm": 6.671668529510498, "learning_rate": 1.1558271931662574e-06, "loss": 0.4224, "step": 15447 }, { "epoch": 0.781425464110476, "grad_norm": 6.4562296867370605, "learning_rate": 1.155314012831707e-06, "loss": 0.3169, "step": 15448 }, { "epoch": 0.7814760483585411, "grad_norm": 6.31792688369751, "learning_rate": 1.1548009315654263e-06, "loss": 0.3688, "step": 15449 }, { "epoch": 0.7815266326066063, "grad_norm": 6.520472526550293, "learning_rate": 1.1542879493806398e-06, "loss": 0.4969, "step": 15450 }, { "epoch": 0.7815772168546714, "grad_norm": 9.281855583190918, "learning_rate": 1.1537750662905633e-06, "loss": 0.3584, "step": 15451 }, { "epoch": 0.7816278011027366, "grad_norm": 6.950819492340088, "learning_rate": 1.1532622823084121e-06, "loss": 0.325, "step": 15452 }, { "epoch": 0.7816783853508018, "grad_norm": 7.720101833343506, "learning_rate": 1.1527495974474035e-06, "loss": 0.38, "step": 15453 }, { "epoch": 0.781728969598867, "grad_norm": 8.725993156433105, "learning_rate": 1.1522370117207438e-06, "loss": 0.4578, "step": 15454 }, { "epoch": 0.7817795538469321, "grad_norm": 7.060192584991455, "learning_rate": 1.1517245251416403e-06, "loss": 0.3445, "step": 15455 }, { "epoch": 0.7818301380949972, "grad_norm": 6.533290386199951, "learning_rate": 1.1512121377233026e-06, "loss": 0.3675, "step": 15456 }, { "epoch": 0.7818807223430624, "grad_norm": 14.026949882507324, "learning_rate": 1.1506998494789317e-06, "loss": 0.4136, "step": 15457 }, { "epoch": 0.7819313065911275, "grad_norm": 10.093303680419922, "learning_rate": 1.1501876604217266e-06, "loss": 0.3447, "step": 15458 }, { "epoch": 0.7819818908391927, "grad_norm": 8.551136016845703, "learning_rate": 1.1496755705648883e-06, "loss": 0.4434, "step": 15459 }, { "epoch": 0.7820324750872578, "grad_norm": 5.239050388336182, "learning_rate": 1.14916357992161e-06, "loss": 0.3578, "step": 15460 }, { "epoch": 0.782083059335323, "grad_norm": 6.178897380828857, "learning_rate": 1.1486516885050836e-06, "loss": 0.3083, "step": 15461 }, { "epoch": 0.7821336435833881, "grad_norm": 5.099050045013428, "learning_rate": 1.1481398963285024e-06, "loss": 0.3581, "step": 15462 }, { "epoch": 0.7821842278314532, "grad_norm": 8.634602546691895, "learning_rate": 1.1476282034050517e-06, "loss": 0.3413, "step": 15463 }, { "epoch": 0.7822348120795184, "grad_norm": 4.7653279304504395, "learning_rate": 1.1471166097479163e-06, "loss": 0.3776, "step": 15464 }, { "epoch": 0.7822853963275835, "grad_norm": 7.188238620758057, "learning_rate": 1.1466051153702823e-06, "loss": 0.2965, "step": 15465 }, { "epoch": 0.7823359805756488, "grad_norm": 18.78823471069336, "learning_rate": 1.146093720285324e-06, "loss": 0.5934, "step": 15466 }, { "epoch": 0.7823865648237139, "grad_norm": 7.124218940734863, "learning_rate": 1.1455824245062236e-06, "loss": 0.4395, "step": 15467 }, { "epoch": 0.7824371490717791, "grad_norm": 4.41345739364624, "learning_rate": 1.1450712280461534e-06, "loss": 0.3204, "step": 15468 }, { "epoch": 0.7824877333198442, "grad_norm": 9.388640403747559, "learning_rate": 1.1445601309182875e-06, "loss": 0.5129, "step": 15469 }, { "epoch": 0.7825383175679094, "grad_norm": 14.542915344238281, "learning_rate": 1.1440491331357923e-06, "loss": 0.5876, "step": 15470 }, { "epoch": 0.7825889018159745, "grad_norm": 12.371553421020508, "learning_rate": 1.143538234711839e-06, "loss": 0.3792, "step": 15471 }, { "epoch": 0.7826394860640397, "grad_norm": 7.723186492919922, "learning_rate": 1.1430274356595905e-06, "loss": 0.4333, "step": 15472 }, { "epoch": 0.7826900703121048, "grad_norm": 6.569398403167725, "learning_rate": 1.1425167359922073e-06, "loss": 0.3893, "step": 15473 }, { "epoch": 0.78274065456017, "grad_norm": 5.102715015411377, "learning_rate": 1.1420061357228524e-06, "loss": 0.3528, "step": 15474 }, { "epoch": 0.7827912388082351, "grad_norm": 11.972508430480957, "learning_rate": 1.1414956348646799e-06, "loss": 0.4375, "step": 15475 }, { "epoch": 0.7828418230563002, "grad_norm": 5.840868949890137, "learning_rate": 1.1409852334308447e-06, "loss": 0.2844, "step": 15476 }, { "epoch": 0.7828924073043654, "grad_norm": 10.386811256408691, "learning_rate": 1.1404749314344992e-06, "loss": 0.5194, "step": 15477 }, { "epoch": 0.7829429915524305, "grad_norm": 10.142738342285156, "learning_rate": 1.1399647288887932e-06, "loss": 0.4052, "step": 15478 }, { "epoch": 0.7829935758004958, "grad_norm": 7.334458827972412, "learning_rate": 1.1394546258068723e-06, "loss": 0.346, "step": 15479 }, { "epoch": 0.7830441600485609, "grad_norm": 11.43995189666748, "learning_rate": 1.1389446222018813e-06, "loss": 0.4152, "step": 15480 }, { "epoch": 0.7830947442966261, "grad_norm": 6.461334228515625, "learning_rate": 1.13843471808696e-06, "loss": 0.3184, "step": 15481 }, { "epoch": 0.7831453285446912, "grad_norm": 6.804673194885254, "learning_rate": 1.1379249134752496e-06, "loss": 0.4064, "step": 15482 }, { "epoch": 0.7831959127927564, "grad_norm": 8.070745468139648, "learning_rate": 1.1374152083798862e-06, "loss": 0.3918, "step": 15483 }, { "epoch": 0.7832464970408215, "grad_norm": 5.599001884460449, "learning_rate": 1.1369056028140013e-06, "loss": 0.3584, "step": 15484 }, { "epoch": 0.7832970812888866, "grad_norm": 11.191390991210938, "learning_rate": 1.13639609679073e-06, "loss": 0.4302, "step": 15485 }, { "epoch": 0.7833476655369518, "grad_norm": 6.7560930252075195, "learning_rate": 1.1358866903231992e-06, "loss": 0.2927, "step": 15486 }, { "epoch": 0.7833982497850169, "grad_norm": 10.655773162841797, "learning_rate": 1.1353773834245329e-06, "loss": 0.3745, "step": 15487 }, { "epoch": 0.7834488340330821, "grad_norm": 7.443762302398682, "learning_rate": 1.134868176107859e-06, "loss": 0.4073, "step": 15488 }, { "epoch": 0.7834994182811472, "grad_norm": 14.198912620544434, "learning_rate": 1.134359068386296e-06, "loss": 0.4531, "step": 15489 }, { "epoch": 0.7835500025292124, "grad_norm": 11.468592643737793, "learning_rate": 1.1338500602729612e-06, "loss": 0.4559, "step": 15490 }, { "epoch": 0.7836005867772775, "grad_norm": 7.747256755828857, "learning_rate": 1.133341151780975e-06, "loss": 0.3522, "step": 15491 }, { "epoch": 0.7836511710253428, "grad_norm": 7.7620110511779785, "learning_rate": 1.132832342923446e-06, "loss": 0.3377, "step": 15492 }, { "epoch": 0.7837017552734079, "grad_norm": 5.377219200134277, "learning_rate": 1.1323236337134864e-06, "loss": 0.2824, "step": 15493 }, { "epoch": 0.783752339521473, "grad_norm": 20.755615234375, "learning_rate": 1.1318150241642052e-06, "loss": 0.4528, "step": 15494 }, { "epoch": 0.7838029237695382, "grad_norm": 7.842251300811768, "learning_rate": 1.131306514288708e-06, "loss": 0.3749, "step": 15495 }, { "epoch": 0.7838535080176033, "grad_norm": 15.078347206115723, "learning_rate": 1.130798104100096e-06, "loss": 0.5889, "step": 15496 }, { "epoch": 0.7839040922656685, "grad_norm": 7.98208475112915, "learning_rate": 1.1302897936114726e-06, "loss": 0.4115, "step": 15497 }, { "epoch": 0.7839546765137336, "grad_norm": 7.115753173828125, "learning_rate": 1.1297815828359343e-06, "loss": 0.366, "step": 15498 }, { "epoch": 0.7840052607617988, "grad_norm": 22.1810245513916, "learning_rate": 1.129273471786575e-06, "loss": 0.4581, "step": 15499 }, { "epoch": 0.7840558450098639, "grad_norm": 7.166258811950684, "learning_rate": 1.1287654604764903e-06, "loss": 0.34, "step": 15500 }, { "epoch": 0.7841064292579291, "grad_norm": 6.147963047027588, "learning_rate": 1.1282575489187697e-06, "loss": 0.3544, "step": 15501 }, { "epoch": 0.7841570135059942, "grad_norm": 9.08897876739502, "learning_rate": 1.1277497371264984e-06, "loss": 0.5397, "step": 15502 }, { "epoch": 0.7842075977540593, "grad_norm": 15.801030158996582, "learning_rate": 1.1272420251127647e-06, "loss": 0.6178, "step": 15503 }, { "epoch": 0.7842581820021245, "grad_norm": 10.758077621459961, "learning_rate": 1.126734412890651e-06, "loss": 0.4883, "step": 15504 }, { "epoch": 0.7843087662501896, "grad_norm": 12.871987342834473, "learning_rate": 1.1262269004732334e-06, "loss": 0.4255, "step": 15505 }, { "epoch": 0.7843593504982549, "grad_norm": 10.229352951049805, "learning_rate": 1.1257194878735933e-06, "loss": 0.4135, "step": 15506 }, { "epoch": 0.78440993474632, "grad_norm": 7.608980178833008, "learning_rate": 1.1252121751048039e-06, "loss": 0.3162, "step": 15507 }, { "epoch": 0.7844605189943852, "grad_norm": 6.176605701446533, "learning_rate": 1.1247049621799367e-06, "loss": 0.4039, "step": 15508 }, { "epoch": 0.7845111032424503, "grad_norm": 7.024835586547852, "learning_rate": 1.124197849112063e-06, "loss": 0.3178, "step": 15509 }, { "epoch": 0.7845616874905155, "grad_norm": 10.858680725097656, "learning_rate": 1.1236908359142495e-06, "loss": 0.4066, "step": 15510 }, { "epoch": 0.7846122717385806, "grad_norm": 7.123605728149414, "learning_rate": 1.1231839225995583e-06, "loss": 0.332, "step": 15511 }, { "epoch": 0.7846628559866458, "grad_norm": 5.33030891418457, "learning_rate": 1.1226771091810556e-06, "loss": 0.3187, "step": 15512 }, { "epoch": 0.7847134402347109, "grad_norm": 6.1072845458984375, "learning_rate": 1.1221703956717978e-06, "loss": 0.375, "step": 15513 }, { "epoch": 0.784764024482776, "grad_norm": 8.248699188232422, "learning_rate": 1.1216637820848415e-06, "loss": 0.4564, "step": 15514 }, { "epoch": 0.7848146087308412, "grad_norm": 6.141493320465088, "learning_rate": 1.1211572684332427e-06, "loss": 0.385, "step": 15515 }, { "epoch": 0.7848651929789063, "grad_norm": 5.972603797912598, "learning_rate": 1.1206508547300526e-06, "loss": 0.4089, "step": 15516 }, { "epoch": 0.7849157772269715, "grad_norm": 5.4479475021362305, "learning_rate": 1.12014454098832e-06, "loss": 0.3443, "step": 15517 }, { "epoch": 0.7849663614750366, "grad_norm": 4.445266246795654, "learning_rate": 1.1196383272210903e-06, "loss": 0.2952, "step": 15518 }, { "epoch": 0.7850169457231019, "grad_norm": 8.051872253417969, "learning_rate": 1.1191322134414074e-06, "loss": 0.4963, "step": 15519 }, { "epoch": 0.785067529971167, "grad_norm": 15.227453231811523, "learning_rate": 1.1186261996623148e-06, "loss": 0.4409, "step": 15520 }, { "epoch": 0.7851181142192322, "grad_norm": 4.898178577423096, "learning_rate": 1.1181202858968505e-06, "loss": 0.3589, "step": 15521 }, { "epoch": 0.7851686984672973, "grad_norm": 8.010039329528809, "learning_rate": 1.1176144721580478e-06, "loss": 0.3998, "step": 15522 }, { "epoch": 0.7852192827153625, "grad_norm": 11.487075805664062, "learning_rate": 1.1171087584589445e-06, "loss": 0.3581, "step": 15523 }, { "epoch": 0.7852698669634276, "grad_norm": 5.593970775604248, "learning_rate": 1.1166031448125704e-06, "loss": 0.2836, "step": 15524 }, { "epoch": 0.7853204512114927, "grad_norm": 4.4574294090271, "learning_rate": 1.116097631231951e-06, "loss": 0.3025, "step": 15525 }, { "epoch": 0.7853710354595579, "grad_norm": 18.954805374145508, "learning_rate": 1.115592217730116e-06, "loss": 0.5694, "step": 15526 }, { "epoch": 0.785421619707623, "grad_norm": 7.2549519538879395, "learning_rate": 1.1150869043200875e-06, "loss": 0.3865, "step": 15527 }, { "epoch": 0.7854722039556882, "grad_norm": 20.26054573059082, "learning_rate": 1.1145816910148837e-06, "loss": 0.5136, "step": 15528 }, { "epoch": 0.7855227882037533, "grad_norm": 10.28435230255127, "learning_rate": 1.114076577827527e-06, "loss": 0.3516, "step": 15529 }, { "epoch": 0.7855733724518185, "grad_norm": 7.494565010070801, "learning_rate": 1.113571564771032e-06, "loss": 0.3427, "step": 15530 }, { "epoch": 0.7856239566998836, "grad_norm": 8.754380226135254, "learning_rate": 1.1130666518584078e-06, "loss": 0.4761, "step": 15531 }, { "epoch": 0.7856745409479489, "grad_norm": 12.396943092346191, "learning_rate": 1.1125618391026688e-06, "loss": 0.4273, "step": 15532 }, { "epoch": 0.785725125196014, "grad_norm": 4.876513957977295, "learning_rate": 1.1120571265168218e-06, "loss": 0.3557, "step": 15533 }, { "epoch": 0.7857757094440792, "grad_norm": 13.250691413879395, "learning_rate": 1.1115525141138694e-06, "loss": 0.3688, "step": 15534 }, { "epoch": 0.7858262936921443, "grad_norm": 6.73215913772583, "learning_rate": 1.1110480019068186e-06, "loss": 0.3638, "step": 15535 }, { "epoch": 0.7858768779402094, "grad_norm": 4.770301342010498, "learning_rate": 1.1105435899086676e-06, "loss": 0.4059, "step": 15536 }, { "epoch": 0.7859274621882746, "grad_norm": 4.953297138214111, "learning_rate": 1.1100392781324121e-06, "loss": 0.3983, "step": 15537 }, { "epoch": 0.7859780464363397, "grad_norm": 9.790197372436523, "learning_rate": 1.10953506659105e-06, "loss": 0.3872, "step": 15538 }, { "epoch": 0.7860286306844049, "grad_norm": 13.723664283752441, "learning_rate": 1.109030955297572e-06, "loss": 0.4275, "step": 15539 }, { "epoch": 0.78607921493247, "grad_norm": 9.288661003112793, "learning_rate": 1.1085269442649665e-06, "loss": 0.4032, "step": 15540 }, { "epoch": 0.7861297991805352, "grad_norm": 19.88842010498047, "learning_rate": 1.1080230335062242e-06, "loss": 0.4362, "step": 15541 }, { "epoch": 0.7861803834286003, "grad_norm": 12.125676155090332, "learning_rate": 1.1075192230343274e-06, "loss": 0.4195, "step": 15542 }, { "epoch": 0.7862309676766654, "grad_norm": 16.005111694335938, "learning_rate": 1.1070155128622578e-06, "loss": 0.477, "step": 15543 }, { "epoch": 0.7862815519247306, "grad_norm": 8.76097297668457, "learning_rate": 1.1065119030029957e-06, "loss": 0.4285, "step": 15544 }, { "epoch": 0.7863321361727957, "grad_norm": 10.133594512939453, "learning_rate": 1.1060083934695177e-06, "loss": 0.3931, "step": 15545 }, { "epoch": 0.786382720420861, "grad_norm": 11.185964584350586, "learning_rate": 1.105504984274796e-06, "loss": 0.4283, "step": 15546 }, { "epoch": 0.7864333046689261, "grad_norm": 10.266950607299805, "learning_rate": 1.105001675431806e-06, "loss": 0.3987, "step": 15547 }, { "epoch": 0.7864838889169913, "grad_norm": 6.335598945617676, "learning_rate": 1.1044984669535146e-06, "loss": 0.3761, "step": 15548 }, { "epoch": 0.7865344731650564, "grad_norm": 11.418804168701172, "learning_rate": 1.1039953588528873e-06, "loss": 0.4254, "step": 15549 }, { "epoch": 0.7865850574131216, "grad_norm": 13.977877616882324, "learning_rate": 1.1034923511428902e-06, "loss": 0.4417, "step": 15550 }, { "epoch": 0.7866356416611867, "grad_norm": 7.85977840423584, "learning_rate": 1.102989443836484e-06, "loss": 0.3856, "step": 15551 }, { "epoch": 0.7866862259092519, "grad_norm": 6.458669185638428, "learning_rate": 1.1024866369466252e-06, "loss": 0.4011, "step": 15552 }, { "epoch": 0.786736810157317, "grad_norm": 8.432364463806152, "learning_rate": 1.1019839304862733e-06, "loss": 0.3792, "step": 15553 }, { "epoch": 0.7867873944053821, "grad_norm": 8.781761169433594, "learning_rate": 1.1014813244683804e-06, "loss": 0.2954, "step": 15554 }, { "epoch": 0.7868379786534473, "grad_norm": 7.62980318069458, "learning_rate": 1.1009788189058968e-06, "loss": 0.3232, "step": 15555 }, { "epoch": 0.7868885629015124, "grad_norm": 16.007970809936523, "learning_rate": 1.1004764138117718e-06, "loss": 0.4796, "step": 15556 }, { "epoch": 0.7869391471495776, "grad_norm": 9.301579475402832, "learning_rate": 1.099974109198949e-06, "loss": 0.4814, "step": 15557 }, { "epoch": 0.7869897313976427, "grad_norm": 6.174143314361572, "learning_rate": 1.0994719050803748e-06, "loss": 0.3879, "step": 15558 }, { "epoch": 0.787040315645708, "grad_norm": 8.176871299743652, "learning_rate": 1.0989698014689886e-06, "loss": 0.2461, "step": 15559 }, { "epoch": 0.7870908998937731, "grad_norm": 5.767384052276611, "learning_rate": 1.0984677983777265e-06, "loss": 0.4048, "step": 15560 }, { "epoch": 0.7871414841418383, "grad_norm": 8.065726280212402, "learning_rate": 1.0979658958195266e-06, "loss": 0.3513, "step": 15561 }, { "epoch": 0.7871920683899034, "grad_norm": 9.952152252197266, "learning_rate": 1.0974640938073206e-06, "loss": 0.4979, "step": 15562 }, { "epoch": 0.7872426526379686, "grad_norm": 4.658092021942139, "learning_rate": 1.0969623923540373e-06, "loss": 0.32, "step": 15563 }, { "epoch": 0.7872932368860337, "grad_norm": 8.935212135314941, "learning_rate": 1.0964607914726077e-06, "loss": 0.4628, "step": 15564 }, { "epoch": 0.7873438211340988, "grad_norm": 8.6639986038208, "learning_rate": 1.0959592911759542e-06, "loss": 0.3034, "step": 15565 }, { "epoch": 0.787394405382164, "grad_norm": 4.43015193939209, "learning_rate": 1.0954578914770004e-06, "loss": 0.3169, "step": 15566 }, { "epoch": 0.7874449896302291, "grad_norm": 4.937568187713623, "learning_rate": 1.094956592388664e-06, "loss": 0.3096, "step": 15567 }, { "epoch": 0.7874955738782943, "grad_norm": 4.48396635055542, "learning_rate": 1.0944553939238666e-06, "loss": 0.2627, "step": 15568 }, { "epoch": 0.7875461581263594, "grad_norm": 9.08266544342041, "learning_rate": 1.093954296095518e-06, "loss": 0.3324, "step": 15569 }, { "epoch": 0.7875967423744246, "grad_norm": 7.653183460235596, "learning_rate": 1.0934532989165337e-06, "loss": 0.3599, "step": 15570 }, { "epoch": 0.7876473266224897, "grad_norm": 6.924614429473877, "learning_rate": 1.0929524023998217e-06, "loss": 0.368, "step": 15571 }, { "epoch": 0.787697910870555, "grad_norm": 8.734223365783691, "learning_rate": 1.0924516065582884e-06, "loss": 0.4574, "step": 15572 }, { "epoch": 0.7877484951186201, "grad_norm": 5.5541582107543945, "learning_rate": 1.0919509114048398e-06, "loss": 0.354, "step": 15573 }, { "epoch": 0.7877990793666853, "grad_norm": 6.141481399536133, "learning_rate": 1.0914503169523772e-06, "loss": 0.3286, "step": 15574 }, { "epoch": 0.7878496636147504, "grad_norm": 7.476155757904053, "learning_rate": 1.0909498232137973e-06, "loss": 0.2695, "step": 15575 }, { "epoch": 0.7879002478628155, "grad_norm": 7.598654747009277, "learning_rate": 1.0904494302020001e-06, "loss": 0.3901, "step": 15576 }, { "epoch": 0.7879508321108807, "grad_norm": 13.35448932647705, "learning_rate": 1.0899491379298776e-06, "loss": 0.3706, "step": 15577 }, { "epoch": 0.7880014163589458, "grad_norm": 8.865843772888184, "learning_rate": 1.0894489464103208e-06, "loss": 0.4492, "step": 15578 }, { "epoch": 0.788052000607011, "grad_norm": 5.7146196365356445, "learning_rate": 1.08894885565622e-06, "loss": 0.3173, "step": 15579 }, { "epoch": 0.7881025848550761, "grad_norm": 7.005025386810303, "learning_rate": 1.088448865680461e-06, "loss": 0.3372, "step": 15580 }, { "epoch": 0.7881531691031413, "grad_norm": 13.875234603881836, "learning_rate": 1.0879489764959261e-06, "loss": 0.5254, "step": 15581 }, { "epoch": 0.7882037533512064, "grad_norm": 15.012320518493652, "learning_rate": 1.0874491881154968e-06, "loss": 0.4856, "step": 15582 }, { "epoch": 0.7882543375992715, "grad_norm": 9.040302276611328, "learning_rate": 1.086949500552052e-06, "loss": 0.3774, "step": 15583 }, { "epoch": 0.7883049218473367, "grad_norm": 11.825727462768555, "learning_rate": 1.086449913818466e-06, "loss": 0.4679, "step": 15584 }, { "epoch": 0.7883555060954018, "grad_norm": 16.676206588745117, "learning_rate": 1.0859504279276134e-06, "loss": 0.5506, "step": 15585 }, { "epoch": 0.7884060903434671, "grad_norm": 8.416471481323242, "learning_rate": 1.0854510428923648e-06, "loss": 0.4984, "step": 15586 }, { "epoch": 0.7884566745915322, "grad_norm": 9.497982025146484, "learning_rate": 1.084951758725586e-06, "loss": 0.3972, "step": 15587 }, { "epoch": 0.7885072588395974, "grad_norm": 6.020726680755615, "learning_rate": 1.0844525754401452e-06, "loss": 0.3414, "step": 15588 }, { "epoch": 0.7885578430876625, "grad_norm": 4.087277412414551, "learning_rate": 1.0839534930489044e-06, "loss": 0.2568, "step": 15589 }, { "epoch": 0.7886084273357277, "grad_norm": 5.40888786315918, "learning_rate": 1.0834545115647217e-06, "loss": 0.3747, "step": 15590 }, { "epoch": 0.7886590115837928, "grad_norm": 9.624235153198242, "learning_rate": 1.0829556310004573e-06, "loss": 0.4564, "step": 15591 }, { "epoch": 0.788709595831858, "grad_norm": 7.595410346984863, "learning_rate": 1.0824568513689653e-06, "loss": 0.4277, "step": 15592 }, { "epoch": 0.7887601800799231, "grad_norm": 7.460726737976074, "learning_rate": 1.0819581726830968e-06, "loss": 0.4714, "step": 15593 }, { "epoch": 0.7888107643279882, "grad_norm": 8.121222496032715, "learning_rate": 1.081459594955705e-06, "loss": 0.4132, "step": 15594 }, { "epoch": 0.7888613485760534, "grad_norm": 12.931602478027344, "learning_rate": 1.0809611181996316e-06, "loss": 0.5004, "step": 15595 }, { "epoch": 0.7889119328241185, "grad_norm": 6.107661724090576, "learning_rate": 1.0804627424277258e-06, "loss": 0.3954, "step": 15596 }, { "epoch": 0.7889625170721837, "grad_norm": 4.226810932159424, "learning_rate": 1.0799644676528276e-06, "loss": 0.3246, "step": 15597 }, { "epoch": 0.7890131013202488, "grad_norm": 6.930354118347168, "learning_rate": 1.0794662938877752e-06, "loss": 0.4421, "step": 15598 }, { "epoch": 0.7890636855683141, "grad_norm": 8.951630592346191, "learning_rate": 1.0789682211454084e-06, "loss": 0.4039, "step": 15599 }, { "epoch": 0.7891142698163792, "grad_norm": 6.583468914031982, "learning_rate": 1.0784702494385597e-06, "loss": 0.3933, "step": 15600 }, { "epoch": 0.7891648540644444, "grad_norm": 11.227416038513184, "learning_rate": 1.077972378780059e-06, "loss": 0.4661, "step": 15601 }, { "epoch": 0.7892154383125095, "grad_norm": 9.638161659240723, "learning_rate": 1.0774746091827382e-06, "loss": 0.4424, "step": 15602 }, { "epoch": 0.7892660225605747, "grad_norm": 7.487802505493164, "learning_rate": 1.0769769406594228e-06, "loss": 0.3328, "step": 15603 }, { "epoch": 0.7893166068086398, "grad_norm": 5.571170806884766, "learning_rate": 1.076479373222935e-06, "loss": 0.3049, "step": 15604 }, { "epoch": 0.7893671910567049, "grad_norm": 5.310480117797852, "learning_rate": 1.0759819068860966e-06, "loss": 0.31, "step": 15605 }, { "epoch": 0.7894177753047701, "grad_norm": 11.57763385772705, "learning_rate": 1.0754845416617272e-06, "loss": 0.3846, "step": 15606 }, { "epoch": 0.7894683595528352, "grad_norm": 9.315032958984375, "learning_rate": 1.0749872775626418e-06, "loss": 0.4311, "step": 15607 }, { "epoch": 0.7895189438009004, "grad_norm": 6.626203536987305, "learning_rate": 1.0744901146016534e-06, "loss": 0.4063, "step": 15608 }, { "epoch": 0.7895695280489655, "grad_norm": 4.221859455108643, "learning_rate": 1.0739930527915737e-06, "loss": 0.3112, "step": 15609 }, { "epoch": 0.7896201122970307, "grad_norm": 9.68107795715332, "learning_rate": 1.0734960921452087e-06, "loss": 0.4174, "step": 15610 }, { "epoch": 0.7896706965450958, "grad_norm": 10.066790580749512, "learning_rate": 1.0729992326753664e-06, "loss": 0.4602, "step": 15611 }, { "epoch": 0.7897212807931611, "grad_norm": 16.88155746459961, "learning_rate": 1.0725024743948486e-06, "loss": 0.5486, "step": 15612 }, { "epoch": 0.7897718650412262, "grad_norm": 11.3246488571167, "learning_rate": 1.072005817316455e-06, "loss": 0.3567, "step": 15613 }, { "epoch": 0.7898224492892914, "grad_norm": 11.445433616638184, "learning_rate": 1.0715092614529844e-06, "loss": 0.397, "step": 15614 }, { "epoch": 0.7898730335373565, "grad_norm": 6.778287410736084, "learning_rate": 1.071012806817231e-06, "loss": 0.4444, "step": 15615 }, { "epoch": 0.7899236177854216, "grad_norm": 5.5849809646606445, "learning_rate": 1.070516453421987e-06, "loss": 0.4014, "step": 15616 }, { "epoch": 0.7899742020334868, "grad_norm": 4.651922702789307, "learning_rate": 1.070020201280043e-06, "loss": 0.2631, "step": 15617 }, { "epoch": 0.7900247862815519, "grad_norm": 8.05250072479248, "learning_rate": 1.069524050404187e-06, "loss": 0.4594, "step": 15618 }, { "epoch": 0.7900753705296171, "grad_norm": 6.421453952789307, "learning_rate": 1.0690280008072008e-06, "loss": 0.3529, "step": 15619 }, { "epoch": 0.7901259547776822, "grad_norm": 7.946651458740234, "learning_rate": 1.0685320525018704e-06, "loss": 0.3679, "step": 15620 }, { "epoch": 0.7901765390257474, "grad_norm": 10.434048652648926, "learning_rate": 1.0680362055009718e-06, "loss": 0.4578, "step": 15621 }, { "epoch": 0.7902271232738125, "grad_norm": 12.50633716583252, "learning_rate": 1.0675404598172816e-06, "loss": 0.5291, "step": 15622 }, { "epoch": 0.7902777075218776, "grad_norm": 7.882421970367432, "learning_rate": 1.067044815463577e-06, "loss": 0.3074, "step": 15623 }, { "epoch": 0.7903282917699428, "grad_norm": 9.433060646057129, "learning_rate": 1.0665492724526278e-06, "loss": 0.4244, "step": 15624 }, { "epoch": 0.790378876018008, "grad_norm": 10.112707138061523, "learning_rate": 1.0660538307972012e-06, "loss": 0.5188, "step": 15625 }, { "epoch": 0.7904294602660732, "grad_norm": 7.272545337677002, "learning_rate": 1.065558490510067e-06, "loss": 0.3822, "step": 15626 }, { "epoch": 0.7904800445141383, "grad_norm": 13.580300331115723, "learning_rate": 1.0650632516039872e-06, "loss": 0.4107, "step": 15627 }, { "epoch": 0.7905306287622035, "grad_norm": 9.522998809814453, "learning_rate": 1.0645681140917213e-06, "loss": 0.4146, "step": 15628 }, { "epoch": 0.7905812130102686, "grad_norm": 8.383299827575684, "learning_rate": 1.0640730779860314e-06, "loss": 0.4087, "step": 15629 }, { "epoch": 0.7906317972583338, "grad_norm": 4.534384250640869, "learning_rate": 1.063578143299671e-06, "loss": 0.2726, "step": 15630 }, { "epoch": 0.7906823815063989, "grad_norm": 5.948877811431885, "learning_rate": 1.0630833100453926e-06, "loss": 0.3868, "step": 15631 }, { "epoch": 0.7907329657544641, "grad_norm": 7.565035820007324, "learning_rate": 1.0625885782359491e-06, "loss": 0.3548, "step": 15632 }, { "epoch": 0.7907835500025292, "grad_norm": 8.384724617004395, "learning_rate": 1.062093947884088e-06, "loss": 0.3506, "step": 15633 }, { "epoch": 0.7908341342505943, "grad_norm": 9.774924278259277, "learning_rate": 1.061599419002554e-06, "loss": 0.2478, "step": 15634 }, { "epoch": 0.7908847184986595, "grad_norm": 10.746954917907715, "learning_rate": 1.06110499160409e-06, "loss": 0.3923, "step": 15635 }, { "epoch": 0.7909353027467246, "grad_norm": 6.334446430206299, "learning_rate": 1.0606106657014353e-06, "loss": 0.3557, "step": 15636 }, { "epoch": 0.7909858869947898, "grad_norm": 10.360201835632324, "learning_rate": 1.0601164413073294e-06, "loss": 0.4169, "step": 15637 }, { "epoch": 0.7910364712428549, "grad_norm": 6.842092990875244, "learning_rate": 1.0596223184345072e-06, "loss": 0.3346, "step": 15638 }, { "epoch": 0.7910870554909202, "grad_norm": 6.342990398406982, "learning_rate": 1.0591282970956995e-06, "loss": 0.3737, "step": 15639 }, { "epoch": 0.7911376397389853, "grad_norm": 8.18863582611084, "learning_rate": 1.058634377303636e-06, "loss": 0.3881, "step": 15640 }, { "epoch": 0.7911882239870505, "grad_norm": 5.51008939743042, "learning_rate": 1.0581405590710459e-06, "loss": 0.334, "step": 15641 }, { "epoch": 0.7912388082351156, "grad_norm": 13.798744201660156, "learning_rate": 1.0576468424106523e-06, "loss": 0.4303, "step": 15642 }, { "epoch": 0.7912893924831808, "grad_norm": 8.15261173248291, "learning_rate": 1.0571532273351759e-06, "loss": 0.4261, "step": 15643 }, { "epoch": 0.7913399767312459, "grad_norm": 7.660931587219238, "learning_rate": 1.0566597138573386e-06, "loss": 0.3035, "step": 15644 }, { "epoch": 0.791390560979311, "grad_norm": 8.901419639587402, "learning_rate": 1.0561663019898554e-06, "loss": 0.4039, "step": 15645 }, { "epoch": 0.7914411452273762, "grad_norm": 8.257152557373047, "learning_rate": 1.0556729917454412e-06, "loss": 0.2994, "step": 15646 }, { "epoch": 0.7914917294754413, "grad_norm": 5.65915060043335, "learning_rate": 1.0551797831368065e-06, "loss": 0.4369, "step": 15647 }, { "epoch": 0.7915423137235065, "grad_norm": 8.283935546875, "learning_rate": 1.0546866761766595e-06, "loss": 0.4652, "step": 15648 }, { "epoch": 0.7915928979715716, "grad_norm": 7.331096172332764, "learning_rate": 1.054193670877708e-06, "loss": 0.4285, "step": 15649 }, { "epoch": 0.7916434822196368, "grad_norm": 8.780972480773926, "learning_rate": 1.0537007672526556e-06, "loss": 0.3662, "step": 15650 }, { "epoch": 0.7916940664677019, "grad_norm": 11.52271842956543, "learning_rate": 1.0532079653142009e-06, "loss": 0.4368, "step": 15651 }, { "epoch": 0.7917446507157672, "grad_norm": 7.160409927368164, "learning_rate": 1.0527152650750449e-06, "loss": 0.3535, "step": 15652 }, { "epoch": 0.7917952349638323, "grad_norm": 8.728760719299316, "learning_rate": 1.0522226665478824e-06, "loss": 0.3742, "step": 15653 }, { "epoch": 0.7918458192118975, "grad_norm": 5.090276718139648, "learning_rate": 1.0517301697454052e-06, "loss": 0.3157, "step": 15654 }, { "epoch": 0.7918964034599626, "grad_norm": 5.845663070678711, "learning_rate": 1.0512377746803066e-06, "loss": 0.3367, "step": 15655 }, { "epoch": 0.7919469877080277, "grad_norm": 9.697869300842285, "learning_rate": 1.0507454813652718e-06, "loss": 0.4573, "step": 15656 }, { "epoch": 0.7919975719560929, "grad_norm": 5.2184529304504395, "learning_rate": 1.0502532898129864e-06, "loss": 0.3515, "step": 15657 }, { "epoch": 0.792048156204158, "grad_norm": 9.255483627319336, "learning_rate": 1.0497612000361357e-06, "loss": 0.5039, "step": 15658 }, { "epoch": 0.7920987404522232, "grad_norm": 9.853261947631836, "learning_rate": 1.0492692120473963e-06, "loss": 0.4656, "step": 15659 }, { "epoch": 0.7921493247002883, "grad_norm": 8.877370834350586, "learning_rate": 1.0487773258594448e-06, "loss": 0.4837, "step": 15660 }, { "epoch": 0.7921999089483535, "grad_norm": 10.19897174835205, "learning_rate": 1.0482855414849596e-06, "loss": 0.4075, "step": 15661 }, { "epoch": 0.7922504931964186, "grad_norm": 5.975977897644043, "learning_rate": 1.047793858936611e-06, "loss": 0.4156, "step": 15662 }, { "epoch": 0.7923010774444837, "grad_norm": 7.822544097900391, "learning_rate": 1.0473022782270676e-06, "loss": 0.3447, "step": 15663 }, { "epoch": 0.7923516616925489, "grad_norm": 7.797693729400635, "learning_rate": 1.046810799368998e-06, "loss": 0.3498, "step": 15664 }, { "epoch": 0.7924022459406141, "grad_norm": 6.367042064666748, "learning_rate": 1.0463194223750655e-06, "loss": 0.4264, "step": 15665 }, { "epoch": 0.7924528301886793, "grad_norm": 10.614691734313965, "learning_rate": 1.0458281472579308e-06, "loss": 0.5047, "step": 15666 }, { "epoch": 0.7925034144367444, "grad_norm": 7.028533458709717, "learning_rate": 1.045336974030255e-06, "loss": 0.3076, "step": 15667 }, { "epoch": 0.7925539986848096, "grad_norm": 9.165911674499512, "learning_rate": 1.044845902704693e-06, "loss": 0.3849, "step": 15668 }, { "epoch": 0.7926045829328747, "grad_norm": 6.947567939758301, "learning_rate": 1.044354933293898e-06, "loss": 0.3703, "step": 15669 }, { "epoch": 0.7926551671809399, "grad_norm": 9.715924263000488, "learning_rate": 1.0438640658105225e-06, "loss": 0.2839, "step": 15670 }, { "epoch": 0.792705751429005, "grad_norm": 6.8574700355529785, "learning_rate": 1.0433733002672147e-06, "loss": 0.3994, "step": 15671 }, { "epoch": 0.7927563356770702, "grad_norm": 11.617928504943848, "learning_rate": 1.0428826366766204e-06, "loss": 0.4546, "step": 15672 }, { "epoch": 0.7928069199251353, "grad_norm": 7.798035621643066, "learning_rate": 1.0423920750513822e-06, "loss": 0.3681, "step": 15673 }, { "epoch": 0.7928575041732004, "grad_norm": 6.6354193687438965, "learning_rate": 1.0419016154041401e-06, "loss": 0.3793, "step": 15674 }, { "epoch": 0.7929080884212656, "grad_norm": 9.938346862792969, "learning_rate": 1.0414112577475339e-06, "loss": 0.519, "step": 15675 }, { "epoch": 0.7929586726693307, "grad_norm": 6.959335803985596, "learning_rate": 1.0409210020941983e-06, "loss": 0.3251, "step": 15676 }, { "epoch": 0.7930092569173959, "grad_norm": 12.4376220703125, "learning_rate": 1.0404308484567654e-06, "loss": 0.4502, "step": 15677 }, { "epoch": 0.793059841165461, "grad_norm": 8.942608833312988, "learning_rate": 1.0399407968478643e-06, "loss": 0.4768, "step": 15678 }, { "epoch": 0.7931104254135263, "grad_norm": 13.330221176147461, "learning_rate": 1.039450847280125e-06, "loss": 0.4917, "step": 15679 }, { "epoch": 0.7931610096615914, "grad_norm": 11.180760383605957, "learning_rate": 1.0389609997661716e-06, "loss": 0.4537, "step": 15680 }, { "epoch": 0.7932115939096566, "grad_norm": 7.704334735870361, "learning_rate": 1.0384712543186237e-06, "loss": 0.3879, "step": 15681 }, { "epoch": 0.7932621781577217, "grad_norm": 11.85258674621582, "learning_rate": 1.037981610950105e-06, "loss": 0.5121, "step": 15682 }, { "epoch": 0.7933127624057869, "grad_norm": 7.309989929199219, "learning_rate": 1.03749206967323e-06, "loss": 0.4281, "step": 15683 }, { "epoch": 0.793363346653852, "grad_norm": 6.609946250915527, "learning_rate": 1.0370026305006132e-06, "loss": 0.3437, "step": 15684 }, { "epoch": 0.7934139309019171, "grad_norm": 4.655862331390381, "learning_rate": 1.0365132934448664e-06, "loss": 0.4168, "step": 15685 }, { "epoch": 0.7934645151499823, "grad_norm": 6.6487956047058105, "learning_rate": 1.0360240585185971e-06, "loss": 0.3983, "step": 15686 }, { "epoch": 0.7935150993980474, "grad_norm": 7.427637100219727, "learning_rate": 1.0355349257344144e-06, "loss": 0.454, "step": 15687 }, { "epoch": 0.7935656836461126, "grad_norm": 6.140048027038574, "learning_rate": 1.0350458951049213e-06, "loss": 0.4086, "step": 15688 }, { "epoch": 0.7936162678941777, "grad_norm": 10.505054473876953, "learning_rate": 1.0345569666427174e-06, "loss": 0.4853, "step": 15689 }, { "epoch": 0.7936668521422429, "grad_norm": 12.035998344421387, "learning_rate": 1.034068140360403e-06, "loss": 0.5378, "step": 15690 }, { "epoch": 0.793717436390308, "grad_norm": 7.499611854553223, "learning_rate": 1.0335794162705737e-06, "loss": 0.4391, "step": 15691 }, { "epoch": 0.7937680206383733, "grad_norm": 5.439865589141846, "learning_rate": 1.033090794385821e-06, "loss": 0.3507, "step": 15692 }, { "epoch": 0.7938186048864384, "grad_norm": 11.405228614807129, "learning_rate": 1.032602274718738e-06, "loss": 0.4637, "step": 15693 }, { "epoch": 0.7938691891345036, "grad_norm": 9.034577369689941, "learning_rate": 1.0321138572819123e-06, "loss": 0.433, "step": 15694 }, { "epoch": 0.7939197733825687, "grad_norm": 9.125139236450195, "learning_rate": 1.0316255420879262e-06, "loss": 0.3953, "step": 15695 }, { "epoch": 0.7939703576306338, "grad_norm": 9.416107177734375, "learning_rate": 1.0311373291493664e-06, "loss": 0.3433, "step": 15696 }, { "epoch": 0.794020941878699, "grad_norm": 10.145435333251953, "learning_rate": 1.0306492184788124e-06, "loss": 0.4156, "step": 15697 }, { "epoch": 0.7940715261267641, "grad_norm": 8.982666969299316, "learning_rate": 1.030161210088838e-06, "loss": 0.3613, "step": 15698 }, { "epoch": 0.7941221103748293, "grad_norm": 7.903008937835693, "learning_rate": 1.0296733039920214e-06, "loss": 0.3991, "step": 15699 }, { "epoch": 0.7941726946228944, "grad_norm": 6.442052841186523, "learning_rate": 1.0291855002009344e-06, "loss": 0.4107, "step": 15700 }, { "epoch": 0.7942232788709596, "grad_norm": 9.13680362701416, "learning_rate": 1.0286977987281443e-06, "loss": 0.4516, "step": 15701 }, { "epoch": 0.7942738631190247, "grad_norm": 8.893287658691406, "learning_rate": 1.028210199586221e-06, "loss": 0.4703, "step": 15702 }, { "epoch": 0.7943244473670898, "grad_norm": 8.473480224609375, "learning_rate": 1.0277227027877274e-06, "loss": 0.3898, "step": 15703 }, { "epoch": 0.794375031615155, "grad_norm": 4.880237102508545, "learning_rate": 1.0272353083452236e-06, "loss": 0.4024, "step": 15704 }, { "epoch": 0.7944256158632202, "grad_norm": 4.471296787261963, "learning_rate": 1.0267480162712718e-06, "loss": 0.3743, "step": 15705 }, { "epoch": 0.7944762001112854, "grad_norm": 6.684561252593994, "learning_rate": 1.0262608265784263e-06, "loss": 0.3822, "step": 15706 }, { "epoch": 0.7945267843593505, "grad_norm": 7.355399131774902, "learning_rate": 1.02577373927924e-06, "loss": 0.3211, "step": 15707 }, { "epoch": 0.7945773686074157, "grad_norm": 6.116745948791504, "learning_rate": 1.0252867543862665e-06, "loss": 0.4467, "step": 15708 }, { "epoch": 0.7946279528554808, "grad_norm": 9.446953773498535, "learning_rate": 1.0247998719120522e-06, "loss": 0.3976, "step": 15709 }, { "epoch": 0.794678537103546, "grad_norm": 6.597794532775879, "learning_rate": 1.0243130918691441e-06, "loss": 0.3934, "step": 15710 }, { "epoch": 0.7947291213516111, "grad_norm": 12.065515518188477, "learning_rate": 1.0238264142700849e-06, "loss": 0.361, "step": 15711 }, { "epoch": 0.7947797055996763, "grad_norm": 11.516335487365723, "learning_rate": 1.0233398391274147e-06, "loss": 0.4599, "step": 15712 }, { "epoch": 0.7948302898477414, "grad_norm": 9.664791107177734, "learning_rate": 1.02285336645367e-06, "loss": 0.386, "step": 15713 }, { "epoch": 0.7948808740958065, "grad_norm": 5.456119537353516, "learning_rate": 1.0223669962613892e-06, "loss": 0.288, "step": 15714 }, { "epoch": 0.7949314583438717, "grad_norm": 8.192972183227539, "learning_rate": 1.0218807285631034e-06, "loss": 0.4152, "step": 15715 }, { "epoch": 0.7949820425919368, "grad_norm": 9.050740242004395, "learning_rate": 1.0213945633713407e-06, "loss": 0.348, "step": 15716 }, { "epoch": 0.795032626840002, "grad_norm": 5.961493492126465, "learning_rate": 1.0209085006986319e-06, "loss": 0.3266, "step": 15717 }, { "epoch": 0.7950832110880671, "grad_norm": 10.964614868164062, "learning_rate": 1.0204225405575002e-06, "loss": 0.4768, "step": 15718 }, { "epoch": 0.7951337953361324, "grad_norm": 7.138030529022217, "learning_rate": 1.0199366829604657e-06, "loss": 0.3868, "step": 15719 }, { "epoch": 0.7951843795841975, "grad_norm": 5.870479583740234, "learning_rate": 1.0194509279200505e-06, "loss": 0.3962, "step": 15720 }, { "epoch": 0.7952349638322627, "grad_norm": 18.730066299438477, "learning_rate": 1.0189652754487706e-06, "loss": 0.3995, "step": 15721 }, { "epoch": 0.7952855480803278, "grad_norm": 8.252126693725586, "learning_rate": 1.0184797255591384e-06, "loss": 0.4519, "step": 15722 }, { "epoch": 0.795336132328393, "grad_norm": 17.714839935302734, "learning_rate": 1.0179942782636693e-06, "loss": 0.3429, "step": 15723 }, { "epoch": 0.7953867165764581, "grad_norm": 10.219074249267578, "learning_rate": 1.0175089335748672e-06, "loss": 0.4115, "step": 15724 }, { "epoch": 0.7954373008245232, "grad_norm": 7.943115711212158, "learning_rate": 1.0170236915052412e-06, "loss": 0.3805, "step": 15725 }, { "epoch": 0.7954878850725884, "grad_norm": 11.345410346984863, "learning_rate": 1.0165385520672949e-06, "loss": 0.4411, "step": 15726 }, { "epoch": 0.7955384693206535, "grad_norm": 6.293512344360352, "learning_rate": 1.0160535152735268e-06, "loss": 0.3681, "step": 15727 }, { "epoch": 0.7955890535687187, "grad_norm": 6.667635917663574, "learning_rate": 1.0155685811364379e-06, "loss": 0.3356, "step": 15728 }, { "epoch": 0.7956396378167838, "grad_norm": 6.028746128082275, "learning_rate": 1.015083749668523e-06, "loss": 0.3854, "step": 15729 }, { "epoch": 0.795690222064849, "grad_norm": 13.872159004211426, "learning_rate": 1.0145990208822732e-06, "loss": 0.4126, "step": 15730 }, { "epoch": 0.7957408063129141, "grad_norm": 8.021350860595703, "learning_rate": 1.0141143947901817e-06, "loss": 0.3595, "step": 15731 }, { "epoch": 0.7957913905609794, "grad_norm": 5.6351189613342285, "learning_rate": 1.0136298714047343e-06, "loss": 0.3713, "step": 15732 }, { "epoch": 0.7958419748090445, "grad_norm": 9.193802833557129, "learning_rate": 1.0131454507384158e-06, "loss": 0.4154, "step": 15733 }, { "epoch": 0.7958925590571096, "grad_norm": 10.862261772155762, "learning_rate": 1.01266113280371e-06, "loss": 0.4523, "step": 15734 }, { "epoch": 0.7959431433051748, "grad_norm": 4.639469146728516, "learning_rate": 1.0121769176130963e-06, "loss": 0.3819, "step": 15735 }, { "epoch": 0.7959937275532399, "grad_norm": 10.910043716430664, "learning_rate": 1.011692805179051e-06, "loss": 0.4206, "step": 15736 }, { "epoch": 0.7960443118013051, "grad_norm": 12.172980308532715, "learning_rate": 1.0112087955140486e-06, "loss": 0.3672, "step": 15737 }, { "epoch": 0.7960948960493702, "grad_norm": 8.847412109375, "learning_rate": 1.010724888630561e-06, "loss": 0.4477, "step": 15738 }, { "epoch": 0.7961454802974354, "grad_norm": 9.604646682739258, "learning_rate": 1.0102410845410559e-06, "loss": 0.4154, "step": 15739 }, { "epoch": 0.7961960645455005, "grad_norm": 6.7813029289245605, "learning_rate": 1.0097573832580026e-06, "loss": 0.4046, "step": 15740 }, { "epoch": 0.7962466487935657, "grad_norm": 6.937527656555176, "learning_rate": 1.0092737847938638e-06, "loss": 0.3587, "step": 15741 }, { "epoch": 0.7962972330416308, "grad_norm": 10.263577461242676, "learning_rate": 1.008790289161099e-06, "loss": 0.4861, "step": 15742 }, { "epoch": 0.796347817289696, "grad_norm": 7.150486946105957, "learning_rate": 1.0083068963721693e-06, "loss": 0.418, "step": 15743 }, { "epoch": 0.7963984015377611, "grad_norm": 11.46538257598877, "learning_rate": 1.0078236064395292e-06, "loss": 0.4456, "step": 15744 }, { "epoch": 0.7964489857858263, "grad_norm": 7.683984279632568, "learning_rate": 1.0073404193756308e-06, "loss": 0.4041, "step": 15745 }, { "epoch": 0.7964995700338915, "grad_norm": 7.920049667358398, "learning_rate": 1.0068573351929272e-06, "loss": 0.3307, "step": 15746 }, { "epoch": 0.7965501542819566, "grad_norm": 14.322455406188965, "learning_rate": 1.006374353903865e-06, "loss": 0.485, "step": 15747 }, { "epoch": 0.7966007385300218, "grad_norm": 8.96564769744873, "learning_rate": 1.0058914755208883e-06, "loss": 0.4977, "step": 15748 }, { "epoch": 0.7966513227780869, "grad_norm": 12.757031440734863, "learning_rate": 1.0054087000564433e-06, "loss": 0.4873, "step": 15749 }, { "epoch": 0.7967019070261521, "grad_norm": 7.137920379638672, "learning_rate": 1.0049260275229666e-06, "loss": 0.3307, "step": 15750 }, { "epoch": 0.7967524912742172, "grad_norm": 5.913434028625488, "learning_rate": 1.0044434579328948e-06, "loss": 0.3244, "step": 15751 }, { "epoch": 0.7968030755222824, "grad_norm": 7.720519065856934, "learning_rate": 1.0039609912986658e-06, "loss": 0.3592, "step": 15752 }, { "epoch": 0.7968536597703475, "grad_norm": 9.004145622253418, "learning_rate": 1.0034786276327102e-06, "loss": 0.4397, "step": 15753 }, { "epoch": 0.7969042440184126, "grad_norm": 9.073833465576172, "learning_rate": 1.0029963669474552e-06, "loss": 0.3603, "step": 15754 }, { "epoch": 0.7969548282664778, "grad_norm": 7.040329456329346, "learning_rate": 1.0025142092553314e-06, "loss": 0.3138, "step": 15755 }, { "epoch": 0.7970054125145429, "grad_norm": 11.754379272460938, "learning_rate": 1.0020321545687607e-06, "loss": 0.4589, "step": 15756 }, { "epoch": 0.7970559967626081, "grad_norm": 5.087051868438721, "learning_rate": 1.0015502029001634e-06, "loss": 0.3165, "step": 15757 }, { "epoch": 0.7971065810106733, "grad_norm": 10.324105262756348, "learning_rate": 1.0010683542619604e-06, "loss": 0.4986, "step": 15758 }, { "epoch": 0.7971571652587385, "grad_norm": 6.474886417388916, "learning_rate": 1.0005866086665672e-06, "loss": 0.3599, "step": 15759 }, { "epoch": 0.7972077495068036, "grad_norm": 9.282665252685547, "learning_rate": 1.0001049661263957e-06, "loss": 0.4715, "step": 15760 }, { "epoch": 0.7972583337548688, "grad_norm": 5.60308837890625, "learning_rate": 9.996234266538607e-07, "loss": 0.3906, "step": 15761 }, { "epoch": 0.7973089180029339, "grad_norm": 5.624305725097656, "learning_rate": 9.991419902613642e-07, "loss": 0.4739, "step": 15762 }, { "epoch": 0.797359502250999, "grad_norm": 9.282377243041992, "learning_rate": 9.986606569613166e-07, "loss": 0.3955, "step": 15763 }, { "epoch": 0.7974100864990642, "grad_norm": 6.578638076782227, "learning_rate": 9.981794267661194e-07, "loss": 0.3636, "step": 15764 }, { "epoch": 0.7974606707471293, "grad_norm": 9.572660446166992, "learning_rate": 9.976982996881707e-07, "loss": 0.4605, "step": 15765 }, { "epoch": 0.7975112549951945, "grad_norm": 9.14567756652832, "learning_rate": 9.972172757398706e-07, "loss": 0.38, "step": 15766 }, { "epoch": 0.7975618392432596, "grad_norm": 6.150049209594727, "learning_rate": 9.967363549336133e-07, "loss": 0.4312, "step": 15767 }, { "epoch": 0.7976124234913248, "grad_norm": 5.884740829467773, "learning_rate": 9.962555372817895e-07, "loss": 0.3168, "step": 15768 }, { "epoch": 0.7976630077393899, "grad_norm": 5.926427841186523, "learning_rate": 9.957748227967912e-07, "loss": 0.2499, "step": 15769 }, { "epoch": 0.7977135919874551, "grad_norm": 7.77595329284668, "learning_rate": 9.952942114910035e-07, "loss": 0.4639, "step": 15770 }, { "epoch": 0.7977641762355202, "grad_norm": 22.816246032714844, "learning_rate": 9.9481370337681e-07, "loss": 0.4216, "step": 15771 }, { "epoch": 0.7978147604835855, "grad_norm": 5.082882404327393, "learning_rate": 9.94333298466595e-07, "loss": 0.4093, "step": 15772 }, { "epoch": 0.7978653447316506, "grad_norm": 6.992105960845947, "learning_rate": 9.938529967727351e-07, "loss": 0.3531, "step": 15773 }, { "epoch": 0.7979159289797157, "grad_norm": 8.386886596679688, "learning_rate": 9.933727983076069e-07, "loss": 0.4715, "step": 15774 }, { "epoch": 0.7979665132277809, "grad_norm": 11.29236888885498, "learning_rate": 9.928927030835839e-07, "loss": 0.4793, "step": 15775 }, { "epoch": 0.798017097475846, "grad_norm": 9.560555458068848, "learning_rate": 9.924127111130377e-07, "loss": 0.3805, "step": 15776 }, { "epoch": 0.7980676817239112, "grad_norm": 6.455967903137207, "learning_rate": 9.91932822408334e-07, "loss": 0.4204, "step": 15777 }, { "epoch": 0.7981182659719763, "grad_norm": 4.983173370361328, "learning_rate": 9.914530369818425e-07, "loss": 0.3473, "step": 15778 }, { "epoch": 0.7981688502200415, "grad_norm": 5.951547145843506, "learning_rate": 9.909733548459238e-07, "loss": 0.299, "step": 15779 }, { "epoch": 0.7982194344681066, "grad_norm": 5.715020179748535, "learning_rate": 9.90493776012937e-07, "loss": 0.3699, "step": 15780 }, { "epoch": 0.7982700187161718, "grad_norm": 7.990172386169434, "learning_rate": 9.900143004952418e-07, "loss": 0.4611, "step": 15781 }, { "epoch": 0.7983206029642369, "grad_norm": 8.539639472961426, "learning_rate": 9.89534928305193e-07, "loss": 0.262, "step": 15782 }, { "epoch": 0.798371187212302, "grad_norm": 5.719292163848877, "learning_rate": 9.890556594551404e-07, "loss": 0.3769, "step": 15783 }, { "epoch": 0.7984217714603672, "grad_norm": 8.09982967376709, "learning_rate": 9.885764939574372e-07, "loss": 0.3954, "step": 15784 }, { "epoch": 0.7984723557084324, "grad_norm": 9.427643775939941, "learning_rate": 9.88097431824428e-07, "loss": 0.4655, "step": 15785 }, { "epoch": 0.7985229399564976, "grad_norm": 8.856195449829102, "learning_rate": 9.876184730684569e-07, "loss": 0.2957, "step": 15786 }, { "epoch": 0.7985735242045627, "grad_norm": 8.891749382019043, "learning_rate": 9.87139617701869e-07, "loss": 0.3488, "step": 15787 }, { "epoch": 0.7986241084526279, "grad_norm": 6.792837142944336, "learning_rate": 9.866608657369985e-07, "loss": 0.396, "step": 15788 }, { "epoch": 0.798674692700693, "grad_norm": 11.006852149963379, "learning_rate": 9.861822171861824e-07, "loss": 0.4013, "step": 15789 }, { "epoch": 0.7987252769487582, "grad_norm": 8.002738952636719, "learning_rate": 9.857036720617575e-07, "loss": 0.3943, "step": 15790 }, { "epoch": 0.7987758611968233, "grad_norm": 6.3958845138549805, "learning_rate": 9.852252303760523e-07, "loss": 0.3745, "step": 15791 }, { "epoch": 0.7988264454448885, "grad_norm": 8.877546310424805, "learning_rate": 9.847468921413945e-07, "loss": 0.4456, "step": 15792 }, { "epoch": 0.7988770296929536, "grad_norm": 5.189027786254883, "learning_rate": 9.84268657370112e-07, "loss": 0.2813, "step": 15793 }, { "epoch": 0.7989276139410187, "grad_norm": 9.582947731018066, "learning_rate": 9.837905260745268e-07, "loss": 0.3845, "step": 15794 }, { "epoch": 0.7989781981890839, "grad_norm": 5.086752891540527, "learning_rate": 9.833124982669579e-07, "loss": 0.3339, "step": 15795 }, { "epoch": 0.799028782437149, "grad_norm": 10.244014739990234, "learning_rate": 9.82834573959725e-07, "loss": 0.4188, "step": 15796 }, { "epoch": 0.7990793666852142, "grad_norm": 10.099125862121582, "learning_rate": 9.823567531651418e-07, "loss": 0.359, "step": 15797 }, { "epoch": 0.7991299509332794, "grad_norm": 4.839349269866943, "learning_rate": 9.8187903589552e-07, "loss": 0.3838, "step": 15798 }, { "epoch": 0.7991805351813446, "grad_norm": 8.181710243225098, "learning_rate": 9.814014221631713e-07, "loss": 0.4237, "step": 15799 }, { "epoch": 0.7992311194294097, "grad_norm": 7.213600158691406, "learning_rate": 9.809239119804008e-07, "loss": 0.3089, "step": 15800 }, { "epoch": 0.7992817036774749, "grad_norm": 7.830071449279785, "learning_rate": 9.804465053595142e-07, "loss": 0.3584, "step": 15801 }, { "epoch": 0.79933228792554, "grad_norm": 7.29524564743042, "learning_rate": 9.79969202312812e-07, "loss": 0.482, "step": 15802 }, { "epoch": 0.7993828721736052, "grad_norm": 11.006773948669434, "learning_rate": 9.794920028525923e-07, "loss": 0.4324, "step": 15803 }, { "epoch": 0.7994334564216703, "grad_norm": 9.81144905090332, "learning_rate": 9.790149069911543e-07, "loss": 0.3906, "step": 15804 }, { "epoch": 0.7994840406697354, "grad_norm": 6.904689788818359, "learning_rate": 9.78537914740789e-07, "loss": 0.3791, "step": 15805 }, { "epoch": 0.7995346249178006, "grad_norm": 5.776566028594971, "learning_rate": 9.780610261137873e-07, "loss": 0.3366, "step": 15806 }, { "epoch": 0.7995852091658657, "grad_norm": 9.165425300598145, "learning_rate": 9.775842411224395e-07, "loss": 0.4252, "step": 15807 }, { "epoch": 0.7996357934139309, "grad_norm": 9.49890422821045, "learning_rate": 9.771075597790303e-07, "loss": 0.4309, "step": 15808 }, { "epoch": 0.799686377661996, "grad_norm": 17.760421752929688, "learning_rate": 9.766309820958408e-07, "loss": 0.6459, "step": 15809 }, { "epoch": 0.7997369619100612, "grad_norm": 5.849024295806885, "learning_rate": 9.761545080851541e-07, "loss": 0.2949, "step": 15810 }, { "epoch": 0.7997875461581263, "grad_norm": 5.065160751342773, "learning_rate": 9.756781377592467e-07, "loss": 0.3516, "step": 15811 }, { "epoch": 0.7998381304061916, "grad_norm": 11.452486038208008, "learning_rate": 9.752018711303934e-07, "loss": 0.4559, "step": 15812 }, { "epoch": 0.7998887146542567, "grad_norm": 8.783889770507812, "learning_rate": 9.747257082108664e-07, "loss": 0.3643, "step": 15813 }, { "epoch": 0.7999392989023218, "grad_norm": 9.644010543823242, "learning_rate": 9.742496490129356e-07, "loss": 0.3483, "step": 15814 }, { "epoch": 0.799989883150387, "grad_norm": 7.5443115234375, "learning_rate": 9.737736935488656e-07, "loss": 0.4797, "step": 15815 }, { "epoch": 0.8000404673984521, "grad_norm": 5.366528034210205, "learning_rate": 9.73297841830924e-07, "loss": 0.3629, "step": 15816 }, { "epoch": 0.8000404673984521, "eval_loss": 0.3897736668586731, "eval_runtime": 34.9151, "eval_samples_per_second": 45.768, "eval_steps_per_second": 5.728, "step": 15816 }, { "epoch": 0.8000910516465173, "grad_norm": 8.326913833618164, "learning_rate": 9.728220938713712e-07, "loss": 0.3609, "step": 15817 }, { "epoch": 0.8001416358945824, "grad_norm": 7.543397903442383, "learning_rate": 9.723464496824642e-07, "loss": 0.3184, "step": 15818 }, { "epoch": 0.8001922201426476, "grad_norm": 9.47301197052002, "learning_rate": 9.718709092764622e-07, "loss": 0.4066, "step": 15819 }, { "epoch": 0.8002428043907127, "grad_norm": 13.37517261505127, "learning_rate": 9.713954726656171e-07, "loss": 0.3973, "step": 15820 }, { "epoch": 0.8002933886387779, "grad_norm": 7.879464149475098, "learning_rate": 9.709201398621787e-07, "loss": 0.3614, "step": 15821 }, { "epoch": 0.800343972886843, "grad_norm": 9.768147468566895, "learning_rate": 9.70444910878398e-07, "loss": 0.4399, "step": 15822 }, { "epoch": 0.8003945571349081, "grad_norm": 5.607625484466553, "learning_rate": 9.699697857265188e-07, "loss": 0.313, "step": 15823 }, { "epoch": 0.8004451413829733, "grad_norm": 7.138385772705078, "learning_rate": 9.694947644187829e-07, "loss": 0.4179, "step": 15824 }, { "epoch": 0.8004957256310385, "grad_norm": 5.487553596496582, "learning_rate": 9.690198469674333e-07, "loss": 0.3726, "step": 15825 }, { "epoch": 0.8005463098791037, "grad_norm": 7.877519607543945, "learning_rate": 9.685450333847064e-07, "loss": 0.3108, "step": 15826 }, { "epoch": 0.8005968941271688, "grad_norm": 4.99550724029541, "learning_rate": 9.680703236828343e-07, "loss": 0.3782, "step": 15827 }, { "epoch": 0.800647478375234, "grad_norm": 6.934349060058594, "learning_rate": 9.675957178740525e-07, "loss": 0.4052, "step": 15828 }, { "epoch": 0.8006980626232991, "grad_norm": 7.206246852874756, "learning_rate": 9.67121215970589e-07, "loss": 0.3639, "step": 15829 }, { "epoch": 0.8007486468713643, "grad_norm": 8.801737785339355, "learning_rate": 9.666468179846693e-07, "loss": 0.3989, "step": 15830 }, { "epoch": 0.8007992311194294, "grad_norm": 8.608778953552246, "learning_rate": 9.661725239285202e-07, "loss": 0.3455, "step": 15831 }, { "epoch": 0.8008498153674946, "grad_norm": 6.8725266456604, "learning_rate": 9.656983338143622e-07, "loss": 0.3465, "step": 15832 }, { "epoch": 0.8009003996155597, "grad_norm": 7.130638122558594, "learning_rate": 9.652242476544122e-07, "loss": 0.341, "step": 15833 }, { "epoch": 0.8009509838636248, "grad_norm": 7.785436630249023, "learning_rate": 9.647502654608892e-07, "loss": 0.3114, "step": 15834 }, { "epoch": 0.80100156811169, "grad_norm": 7.004732608795166, "learning_rate": 9.642763872460049e-07, "loss": 0.4577, "step": 15835 }, { "epoch": 0.8010521523597551, "grad_norm": 6.855075359344482, "learning_rate": 9.638026130219691e-07, "loss": 0.3877, "step": 15836 }, { "epoch": 0.8011027366078203, "grad_norm": 7.582952976226807, "learning_rate": 9.633289428009924e-07, "loss": 0.36, "step": 15837 }, { "epoch": 0.8011533208558855, "grad_norm": 11.053055763244629, "learning_rate": 9.62855376595278e-07, "loss": 0.3992, "step": 15838 }, { "epoch": 0.8012039051039507, "grad_norm": 8.191695213317871, "learning_rate": 9.623819144170298e-07, "loss": 0.4895, "step": 15839 }, { "epoch": 0.8012544893520158, "grad_norm": 8.410029411315918, "learning_rate": 9.61908556278447e-07, "loss": 0.4249, "step": 15840 }, { "epoch": 0.801305073600081, "grad_norm": 5.349323272705078, "learning_rate": 9.614353021917254e-07, "loss": 0.4175, "step": 15841 }, { "epoch": 0.8013556578481461, "grad_norm": 10.061599731445312, "learning_rate": 9.609621521690632e-07, "loss": 0.4025, "step": 15842 }, { "epoch": 0.8014062420962113, "grad_norm": 5.002224445343018, "learning_rate": 9.6048910622265e-07, "loss": 0.3761, "step": 15843 }, { "epoch": 0.8014568263442764, "grad_norm": 10.753790855407715, "learning_rate": 9.600161643646749e-07, "loss": 0.4272, "step": 15844 }, { "epoch": 0.8015074105923415, "grad_norm": 5.803974628448486, "learning_rate": 9.595433266073257e-07, "loss": 0.3695, "step": 15845 }, { "epoch": 0.8015579948404067, "grad_norm": 12.016731262207031, "learning_rate": 9.590705929627858e-07, "loss": 0.3913, "step": 15846 }, { "epoch": 0.8016085790884718, "grad_norm": 6.043999195098877, "learning_rate": 9.585979634432364e-07, "loss": 0.4018, "step": 15847 }, { "epoch": 0.801659163336537, "grad_norm": 8.806185722351074, "learning_rate": 9.581254380608546e-07, "loss": 0.43, "step": 15848 }, { "epoch": 0.8017097475846021, "grad_norm": 5.630605697631836, "learning_rate": 9.576530168278186e-07, "loss": 0.4076, "step": 15849 }, { "epoch": 0.8017603318326673, "grad_norm": 10.889808654785156, "learning_rate": 9.571806997563004e-07, "loss": 0.2955, "step": 15850 }, { "epoch": 0.8018109160807324, "grad_norm": 7.158220291137695, "learning_rate": 9.567084868584708e-07, "loss": 0.4261, "step": 15851 }, { "epoch": 0.8018615003287977, "grad_norm": 8.551106452941895, "learning_rate": 9.562363781464972e-07, "loss": 0.4187, "step": 15852 }, { "epoch": 0.8019120845768628, "grad_norm": 11.114508628845215, "learning_rate": 9.55764373632544e-07, "loss": 0.4682, "step": 15853 }, { "epoch": 0.801962668824928, "grad_norm": 7.181393623352051, "learning_rate": 9.552924733287749e-07, "loss": 0.3239, "step": 15854 }, { "epoch": 0.8020132530729931, "grad_norm": 10.517694473266602, "learning_rate": 9.548206772473496e-07, "loss": 0.393, "step": 15855 }, { "epoch": 0.8020638373210582, "grad_norm": 8.102365493774414, "learning_rate": 9.543489854004234e-07, "loss": 0.3703, "step": 15856 }, { "epoch": 0.8021144215691234, "grad_norm": 7.433495044708252, "learning_rate": 9.538773978001536e-07, "loss": 0.3888, "step": 15857 }, { "epoch": 0.8021650058171885, "grad_norm": 7.732998371124268, "learning_rate": 9.5340591445869e-07, "loss": 0.5043, "step": 15858 }, { "epoch": 0.8022155900652537, "grad_norm": 17.767332077026367, "learning_rate": 9.529345353881803e-07, "loss": 0.551, "step": 15859 }, { "epoch": 0.8022661743133188, "grad_norm": 6.02392053604126, "learning_rate": 9.524632606007739e-07, "loss": 0.3269, "step": 15860 }, { "epoch": 0.802316758561384, "grad_norm": 4.910304069519043, "learning_rate": 9.519920901086128e-07, "loss": 0.2427, "step": 15861 }, { "epoch": 0.8023673428094491, "grad_norm": 11.394262313842773, "learning_rate": 9.515210239238365e-07, "loss": 0.5043, "step": 15862 }, { "epoch": 0.8024179270575142, "grad_norm": 10.944201469421387, "learning_rate": 9.510500620585861e-07, "loss": 0.4818, "step": 15863 }, { "epoch": 0.8024685113055794, "grad_norm": 7.110426902770996, "learning_rate": 9.505792045249967e-07, "loss": 0.3508, "step": 15864 }, { "epoch": 0.8025190955536446, "grad_norm": 8.219587326049805, "learning_rate": 9.501084513351982e-07, "loss": 0.2863, "step": 15865 }, { "epoch": 0.8025696798017098, "grad_norm": 7.295135021209717, "learning_rate": 9.496378025013236e-07, "loss": 0.3738, "step": 15866 }, { "epoch": 0.8026202640497749, "grad_norm": 6.1084675788879395, "learning_rate": 9.491672580354999e-07, "loss": 0.415, "step": 15867 }, { "epoch": 0.8026708482978401, "grad_norm": 8.772168159484863, "learning_rate": 9.486968179498507e-07, "loss": 0.3484, "step": 15868 }, { "epoch": 0.8027214325459052, "grad_norm": 10.48918342590332, "learning_rate": 9.482264822564996e-07, "loss": 0.4192, "step": 15869 }, { "epoch": 0.8027720167939704, "grad_norm": 4.258415699005127, "learning_rate": 9.477562509675658e-07, "loss": 0.3211, "step": 15870 }, { "epoch": 0.8028226010420355, "grad_norm": 12.631203651428223, "learning_rate": 9.472861240951642e-07, "loss": 0.4172, "step": 15871 }, { "epoch": 0.8028731852901007, "grad_norm": 5.378754138946533, "learning_rate": 9.468161016514116e-07, "loss": 0.3707, "step": 15872 }, { "epoch": 0.8029237695381658, "grad_norm": 5.719301223754883, "learning_rate": 9.463461836484178e-07, "loss": 0.3293, "step": 15873 }, { "epoch": 0.8029743537862309, "grad_norm": 10.60483169555664, "learning_rate": 9.458763700982904e-07, "loss": 0.4591, "step": 15874 }, { "epoch": 0.8030249380342961, "grad_norm": 7.853761672973633, "learning_rate": 9.454066610131379e-07, "loss": 0.5028, "step": 15875 }, { "epoch": 0.8030755222823612, "grad_norm": 5.64663028717041, "learning_rate": 9.449370564050625e-07, "loss": 0.3438, "step": 15876 }, { "epoch": 0.8031261065304264, "grad_norm": 9.24971866607666, "learning_rate": 9.444675562861643e-07, "loss": 0.5074, "step": 15877 }, { "epoch": 0.8031766907784916, "grad_norm": 17.630165100097656, "learning_rate": 9.439981606685417e-07, "loss": 0.4668, "step": 15878 }, { "epoch": 0.8032272750265568, "grad_norm": 14.347718238830566, "learning_rate": 9.43528869564288e-07, "loss": 0.4346, "step": 15879 }, { "epoch": 0.8032778592746219, "grad_norm": 8.061474800109863, "learning_rate": 9.430596829854988e-07, "loss": 0.3391, "step": 15880 }, { "epoch": 0.8033284435226871, "grad_norm": 6.7868971824646, "learning_rate": 9.42590600944262e-07, "loss": 0.354, "step": 15881 }, { "epoch": 0.8033790277707522, "grad_norm": 9.769255638122559, "learning_rate": 9.421216234526648e-07, "loss": 0.3639, "step": 15882 }, { "epoch": 0.8034296120188174, "grad_norm": 8.912704467773438, "learning_rate": 9.416527505227924e-07, "loss": 0.3754, "step": 15883 }, { "epoch": 0.8034801962668825, "grad_norm": 5.03317928314209, "learning_rate": 9.411839821667263e-07, "loss": 0.3434, "step": 15884 }, { "epoch": 0.8035307805149476, "grad_norm": 20.32065200805664, "learning_rate": 9.407153183965456e-07, "loss": 0.419, "step": 15885 }, { "epoch": 0.8035813647630128, "grad_norm": 8.877330780029297, "learning_rate": 9.402467592243247e-07, "loss": 0.5193, "step": 15886 }, { "epoch": 0.8036319490110779, "grad_norm": 11.571555137634277, "learning_rate": 9.397783046621406e-07, "loss": 0.5174, "step": 15887 }, { "epoch": 0.8036825332591431, "grad_norm": 7.916591167449951, "learning_rate": 9.393099547220619e-07, "loss": 0.3624, "step": 15888 }, { "epoch": 0.8037331175072082, "grad_norm": 5.073122978210449, "learning_rate": 9.388417094161567e-07, "loss": 0.244, "step": 15889 }, { "epoch": 0.8037837017552734, "grad_norm": 5.7260847091674805, "learning_rate": 9.383735687564937e-07, "loss": 0.3922, "step": 15890 }, { "epoch": 0.8038342860033385, "grad_norm": 8.43960952758789, "learning_rate": 9.379055327551311e-07, "loss": 0.3533, "step": 15891 }, { "epoch": 0.8038848702514038, "grad_norm": 7.493121147155762, "learning_rate": 9.374376014241321e-07, "loss": 0.414, "step": 15892 }, { "epoch": 0.8039354544994689, "grad_norm": 7.301651477813721, "learning_rate": 9.369697747755535e-07, "loss": 0.4153, "step": 15893 }, { "epoch": 0.803986038747534, "grad_norm": 38.26504135131836, "learning_rate": 9.365020528214491e-07, "loss": 0.4535, "step": 15894 }, { "epoch": 0.8040366229955992, "grad_norm": 7.424788475036621, "learning_rate": 9.360344355738726e-07, "loss": 0.4078, "step": 15895 }, { "epoch": 0.8040872072436643, "grad_norm": 7.67678165435791, "learning_rate": 9.355669230448733e-07, "loss": 0.3568, "step": 15896 }, { "epoch": 0.8041377914917295, "grad_norm": 10.649775505065918, "learning_rate": 9.350995152464954e-07, "loss": 0.4514, "step": 15897 }, { "epoch": 0.8041883757397946, "grad_norm": 8.771139144897461, "learning_rate": 9.346322121907858e-07, "loss": 0.4218, "step": 15898 }, { "epoch": 0.8042389599878598, "grad_norm": 5.367171287536621, "learning_rate": 9.341650138897845e-07, "loss": 0.3659, "step": 15899 }, { "epoch": 0.8042895442359249, "grad_norm": 5.909328937530518, "learning_rate": 9.336979203555291e-07, "loss": 0.3825, "step": 15900 }, { "epoch": 0.8043401284839901, "grad_norm": 8.554994583129883, "learning_rate": 9.332309316000576e-07, "loss": 0.4539, "step": 15901 }, { "epoch": 0.8043907127320552, "grad_norm": 5.259663105010986, "learning_rate": 9.327640476354027e-07, "loss": 0.3641, "step": 15902 }, { "epoch": 0.8044412969801203, "grad_norm": 8.283617973327637, "learning_rate": 9.322972684735937e-07, "loss": 0.4803, "step": 15903 }, { "epoch": 0.8044918812281855, "grad_norm": 10.64992618560791, "learning_rate": 9.318305941266592e-07, "loss": 0.4113, "step": 15904 }, { "epoch": 0.8045424654762507, "grad_norm": 5.8487548828125, "learning_rate": 9.313640246066241e-07, "loss": 0.3784, "step": 15905 }, { "epoch": 0.8045930497243159, "grad_norm": 10.204477310180664, "learning_rate": 9.308975599255094e-07, "loss": 0.3952, "step": 15906 }, { "epoch": 0.804643633972381, "grad_norm": 9.031726837158203, "learning_rate": 9.30431200095337e-07, "loss": 0.3764, "step": 15907 }, { "epoch": 0.8046942182204462, "grad_norm": 7.135430812835693, "learning_rate": 9.299649451281234e-07, "loss": 0.3788, "step": 15908 }, { "epoch": 0.8047448024685113, "grad_norm": 7.566958427429199, "learning_rate": 9.294987950358808e-07, "loss": 0.4156, "step": 15909 }, { "epoch": 0.8047953867165765, "grad_norm": 8.495532035827637, "learning_rate": 9.290327498306239e-07, "loss": 0.3586, "step": 15910 }, { "epoch": 0.8048459709646416, "grad_norm": 7.173674583435059, "learning_rate": 9.2856680952436e-07, "loss": 0.3749, "step": 15911 }, { "epoch": 0.8048965552127068, "grad_norm": 10.40334701538086, "learning_rate": 9.281009741290936e-07, "loss": 0.4007, "step": 15912 }, { "epoch": 0.8049471394607719, "grad_norm": 6.874946594238281, "learning_rate": 9.276352436568315e-07, "loss": 0.321, "step": 15913 }, { "epoch": 0.804997723708837, "grad_norm": 7.6860857009887695, "learning_rate": 9.271696181195728e-07, "loss": 0.3412, "step": 15914 }, { "epoch": 0.8050483079569022, "grad_norm": 7.629764080047607, "learning_rate": 9.267040975293141e-07, "loss": 0.4156, "step": 15915 }, { "epoch": 0.8050988922049673, "grad_norm": 8.774145126342773, "learning_rate": 9.262386818980546e-07, "loss": 0.4314, "step": 15916 }, { "epoch": 0.8051494764530325, "grad_norm": 9.20797348022461, "learning_rate": 9.25773371237782e-07, "loss": 0.5018, "step": 15917 }, { "epoch": 0.8052000607010977, "grad_norm": 6.1955180168151855, "learning_rate": 9.253081655604901e-07, "loss": 0.3779, "step": 15918 }, { "epoch": 0.8052506449491629, "grad_norm": 5.964998722076416, "learning_rate": 9.248430648781648e-07, "loss": 0.4321, "step": 15919 }, { "epoch": 0.805301229197228, "grad_norm": 7.958296775817871, "learning_rate": 9.243780692027903e-07, "loss": 0.251, "step": 15920 }, { "epoch": 0.8053518134452932, "grad_norm": 6.882744789123535, "learning_rate": 9.239131785463479e-07, "loss": 0.3891, "step": 15921 }, { "epoch": 0.8054023976933583, "grad_norm": 7.64281702041626, "learning_rate": 9.234483929208182e-07, "loss": 0.3131, "step": 15922 }, { "epoch": 0.8054529819414235, "grad_norm": 7.618453025817871, "learning_rate": 9.229837123381774e-07, "loss": 0.3878, "step": 15923 }, { "epoch": 0.8055035661894886, "grad_norm": 9.837494850158691, "learning_rate": 9.22519136810397e-07, "loss": 0.3678, "step": 15924 }, { "epoch": 0.8055541504375537, "grad_norm": 10.342020988464355, "learning_rate": 9.220546663494512e-07, "loss": 0.4149, "step": 15925 }, { "epoch": 0.8056047346856189, "grad_norm": 9.483469009399414, "learning_rate": 9.215903009673071e-07, "loss": 0.426, "step": 15926 }, { "epoch": 0.805655318933684, "grad_norm": 5.5716657638549805, "learning_rate": 9.211260406759281e-07, "loss": 0.3754, "step": 15927 }, { "epoch": 0.8057059031817492, "grad_norm": 7.68283224105835, "learning_rate": 9.206618854872806e-07, "loss": 0.2342, "step": 15928 }, { "epoch": 0.8057564874298143, "grad_norm": 10.719335556030273, "learning_rate": 9.201978354133228e-07, "loss": 0.3389, "step": 15929 }, { "epoch": 0.8058070716778795, "grad_norm": 10.372443199157715, "learning_rate": 9.197338904660125e-07, "loss": 0.3779, "step": 15930 }, { "epoch": 0.8058576559259447, "grad_norm": 15.946686744689941, "learning_rate": 9.192700506573044e-07, "loss": 0.6217, "step": 15931 }, { "epoch": 0.8059082401740099, "grad_norm": 15.402636528015137, "learning_rate": 9.188063159991495e-07, "loss": 0.3834, "step": 15932 }, { "epoch": 0.805958824422075, "grad_norm": 5.719712257385254, "learning_rate": 9.183426865034989e-07, "loss": 0.323, "step": 15933 }, { "epoch": 0.8060094086701401, "grad_norm": 8.65888500213623, "learning_rate": 9.178791621822986e-07, "loss": 0.5021, "step": 15934 }, { "epoch": 0.8060599929182053, "grad_norm": 7.338206768035889, "learning_rate": 9.174157430474912e-07, "loss": 0.2921, "step": 15935 }, { "epoch": 0.8061105771662704, "grad_norm": 6.1642985343933105, "learning_rate": 9.169524291110205e-07, "loss": 0.3069, "step": 15936 }, { "epoch": 0.8061611614143356, "grad_norm": 13.256543159484863, "learning_rate": 9.164892203848231e-07, "loss": 0.5672, "step": 15937 }, { "epoch": 0.8062117456624007, "grad_norm": 6.619691848754883, "learning_rate": 9.160261168808338e-07, "loss": 0.3122, "step": 15938 }, { "epoch": 0.8062623299104659, "grad_norm": 7.857017517089844, "learning_rate": 9.155631186109887e-07, "loss": 0.3876, "step": 15939 }, { "epoch": 0.806312914158531, "grad_norm": 9.23061466217041, "learning_rate": 9.15100225587216e-07, "loss": 0.4063, "step": 15940 }, { "epoch": 0.8063634984065962, "grad_norm": 6.052432060241699, "learning_rate": 9.146374378214429e-07, "loss": 0.3716, "step": 15941 }, { "epoch": 0.8064140826546613, "grad_norm": 11.503867149353027, "learning_rate": 9.141747553255981e-07, "loss": 0.384, "step": 15942 }, { "epoch": 0.8064646669027264, "grad_norm": 5.751059055328369, "learning_rate": 9.137121781115993e-07, "loss": 0.3824, "step": 15943 }, { "epoch": 0.8065152511507916, "grad_norm": 9.777915954589844, "learning_rate": 9.132497061913659e-07, "loss": 0.3861, "step": 15944 }, { "epoch": 0.8065658353988568, "grad_norm": 7.804317474365234, "learning_rate": 9.127873395768183e-07, "loss": 0.3225, "step": 15945 }, { "epoch": 0.806616419646922, "grad_norm": 15.981307029724121, "learning_rate": 9.123250782798687e-07, "loss": 0.432, "step": 15946 }, { "epoch": 0.8066670038949871, "grad_norm": 6.622660160064697, "learning_rate": 9.11862922312427e-07, "loss": 0.4219, "step": 15947 }, { "epoch": 0.8067175881430523, "grad_norm": 7.124788284301758, "learning_rate": 9.114008716864047e-07, "loss": 0.3765, "step": 15948 }, { "epoch": 0.8067681723911174, "grad_norm": 6.480140209197998, "learning_rate": 9.109389264137065e-07, "loss": 0.2882, "step": 15949 }, { "epoch": 0.8068187566391826, "grad_norm": 4.766302585601807, "learning_rate": 9.104770865062335e-07, "loss": 0.3114, "step": 15950 }, { "epoch": 0.8068693408872477, "grad_norm": 6.796097278594971, "learning_rate": 9.100153519758903e-07, "loss": 0.2825, "step": 15951 }, { "epoch": 0.8069199251353129, "grad_norm": 8.979897499084473, "learning_rate": 9.095537228345719e-07, "loss": 0.4244, "step": 15952 }, { "epoch": 0.806970509383378, "grad_norm": 4.876439571380615, "learning_rate": 9.090921990941726e-07, "loss": 0.3744, "step": 15953 }, { "epoch": 0.8070210936314431, "grad_norm": 10.012672424316406, "learning_rate": 9.08630780766589e-07, "loss": 0.451, "step": 15954 }, { "epoch": 0.8070716778795083, "grad_norm": 6.137413501739502, "learning_rate": 9.081694678637055e-07, "loss": 0.3427, "step": 15955 }, { "epoch": 0.8071222621275734, "grad_norm": 6.09839391708374, "learning_rate": 9.077082603974124e-07, "loss": 0.3912, "step": 15956 }, { "epoch": 0.8071728463756386, "grad_norm": 6.848634243011475, "learning_rate": 9.072471583795933e-07, "loss": 0.4398, "step": 15957 }, { "epoch": 0.8072234306237038, "grad_norm": 6.6289381980896, "learning_rate": 9.06786161822129e-07, "loss": 0.4524, "step": 15958 }, { "epoch": 0.807274014871769, "grad_norm": 5.704624652862549, "learning_rate": 9.063252707368969e-07, "loss": 0.4049, "step": 15959 }, { "epoch": 0.8073245991198341, "grad_norm": 11.009737014770508, "learning_rate": 9.058644851357767e-07, "loss": 0.455, "step": 15960 }, { "epoch": 0.8073751833678993, "grad_norm": 6.987112522125244, "learning_rate": 9.054038050306391e-07, "loss": 0.3553, "step": 15961 }, { "epoch": 0.8074257676159644, "grad_norm": 8.676769256591797, "learning_rate": 9.049432304333545e-07, "loss": 0.4175, "step": 15962 }, { "epoch": 0.8074763518640296, "grad_norm": 12.172280311584473, "learning_rate": 9.044827613557927e-07, "loss": 0.3893, "step": 15963 }, { "epoch": 0.8075269361120947, "grad_norm": 13.962241172790527, "learning_rate": 9.040223978098178e-07, "loss": 0.494, "step": 15964 }, { "epoch": 0.8075775203601598, "grad_norm": 21.870492935180664, "learning_rate": 9.035621398072913e-07, "loss": 0.8289, "step": 15965 }, { "epoch": 0.807628104608225, "grad_norm": 9.632101058959961, "learning_rate": 9.03101987360075e-07, "loss": 0.3659, "step": 15966 }, { "epoch": 0.8076786888562901, "grad_norm": 10.609013557434082, "learning_rate": 9.026419404800251e-07, "loss": 0.3793, "step": 15967 }, { "epoch": 0.8077292731043553, "grad_norm": 5.637543201446533, "learning_rate": 9.021819991789954e-07, "loss": 0.3051, "step": 15968 }, { "epoch": 0.8077798573524204, "grad_norm": 6.235060691833496, "learning_rate": 9.017221634688378e-07, "loss": 0.4387, "step": 15969 }, { "epoch": 0.8078304416004856, "grad_norm": 6.175300598144531, "learning_rate": 9.012624333613995e-07, "loss": 0.3745, "step": 15970 }, { "epoch": 0.8078810258485508, "grad_norm": 8.3336763381958, "learning_rate": 9.008028088685294e-07, "loss": 0.3779, "step": 15971 }, { "epoch": 0.807931610096616, "grad_norm": 16.24797821044922, "learning_rate": 9.003432900020704e-07, "loss": 0.4413, "step": 15972 }, { "epoch": 0.8079821943446811, "grad_norm": 8.011972427368164, "learning_rate": 8.998838767738604e-07, "loss": 0.3696, "step": 15973 }, { "epoch": 0.8080327785927462, "grad_norm": 10.338500022888184, "learning_rate": 8.994245691957415e-07, "loss": 0.2968, "step": 15974 }, { "epoch": 0.8080833628408114, "grad_norm": 8.055034637451172, "learning_rate": 8.989653672795462e-07, "loss": 0.3584, "step": 15975 }, { "epoch": 0.8081339470888765, "grad_norm": 7.084917068481445, "learning_rate": 8.985062710371068e-07, "loss": 0.3921, "step": 15976 }, { "epoch": 0.8081845313369417, "grad_norm": 9.330268859863281, "learning_rate": 8.980472804802554e-07, "loss": 0.3758, "step": 15977 }, { "epoch": 0.8082351155850068, "grad_norm": 6.985451698303223, "learning_rate": 8.975883956208176e-07, "loss": 0.3791, "step": 15978 }, { "epoch": 0.808285699833072, "grad_norm": 6.214657306671143, "learning_rate": 8.971296164706167e-07, "loss": 0.4204, "step": 15979 }, { "epoch": 0.8083362840811371, "grad_norm": 14.199496269226074, "learning_rate": 8.966709430414783e-07, "loss": 0.4916, "step": 15980 }, { "epoch": 0.8083868683292023, "grad_norm": 5.1422038078308105, "learning_rate": 8.962123753452167e-07, "loss": 0.3754, "step": 15981 }, { "epoch": 0.8084374525772674, "grad_norm": 6.990800380706787, "learning_rate": 8.957539133936488e-07, "loss": 0.3948, "step": 15982 }, { "epoch": 0.8084880368253325, "grad_norm": 7.658708572387695, "learning_rate": 8.952955571985905e-07, "loss": 0.4312, "step": 15983 }, { "epoch": 0.8085386210733977, "grad_norm": 10.830671310424805, "learning_rate": 8.948373067718514e-07, "loss": 0.3716, "step": 15984 }, { "epoch": 0.808589205321463, "grad_norm": 7.209026336669922, "learning_rate": 8.943791621252379e-07, "loss": 0.4424, "step": 15985 }, { "epoch": 0.8086397895695281, "grad_norm": 9.00833797454834, "learning_rate": 8.939211232705575e-07, "loss": 0.3542, "step": 15986 }, { "epoch": 0.8086903738175932, "grad_norm": 8.304067611694336, "learning_rate": 8.934631902196122e-07, "loss": 0.4794, "step": 15987 }, { "epoch": 0.8087409580656584, "grad_norm": 8.406454086303711, "learning_rate": 8.930053629842005e-07, "loss": 0.4052, "step": 15988 }, { "epoch": 0.8087915423137235, "grad_norm": 10.815030097961426, "learning_rate": 8.925476415761219e-07, "loss": 0.4382, "step": 15989 }, { "epoch": 0.8088421265617887, "grad_norm": 4.775455474853516, "learning_rate": 8.920900260071691e-07, "loss": 0.3214, "step": 15990 }, { "epoch": 0.8088927108098538, "grad_norm": 9.467528343200684, "learning_rate": 8.916325162891332e-07, "loss": 0.3534, "step": 15991 }, { "epoch": 0.808943295057919, "grad_norm": 9.78628921508789, "learning_rate": 8.911751124338047e-07, "loss": 0.3799, "step": 15992 }, { "epoch": 0.8089938793059841, "grad_norm": 9.108529090881348, "learning_rate": 8.907178144529705e-07, "loss": 0.3248, "step": 15993 }, { "epoch": 0.8090444635540492, "grad_norm": 6.5444536209106445, "learning_rate": 8.902606223584104e-07, "loss": 0.3932, "step": 15994 }, { "epoch": 0.8090950478021144, "grad_norm": 6.996973037719727, "learning_rate": 8.898035361619084e-07, "loss": 0.3162, "step": 15995 }, { "epoch": 0.8091456320501795, "grad_norm": 9.459905624389648, "learning_rate": 8.893465558752418e-07, "loss": 0.3917, "step": 15996 }, { "epoch": 0.8091962162982447, "grad_norm": 5.817540645599365, "learning_rate": 8.888896815101838e-07, "loss": 0.3379, "step": 15997 }, { "epoch": 0.8092468005463099, "grad_norm": 10.415724754333496, "learning_rate": 8.884329130785107e-07, "loss": 0.4369, "step": 15998 }, { "epoch": 0.8092973847943751, "grad_norm": 11.66982364654541, "learning_rate": 8.879762505919898e-07, "loss": 0.4767, "step": 15999 }, { "epoch": 0.8093479690424402, "grad_norm": 10.293327331542969, "learning_rate": 8.875196940623876e-07, "loss": 0.3128, "step": 16000 }, { "epoch": 0.8093985532905054, "grad_norm": 12.397502899169922, "learning_rate": 8.870632435014708e-07, "loss": 0.4696, "step": 16001 }, { "epoch": 0.8094491375385705, "grad_norm": 6.115878582000732, "learning_rate": 8.866068989210003e-07, "loss": 0.394, "step": 16002 }, { "epoch": 0.8094997217866357, "grad_norm": 6.488662242889404, "learning_rate": 8.86150660332733e-07, "loss": 0.3293, "step": 16003 }, { "epoch": 0.8095503060347008, "grad_norm": 5.0488810539245605, "learning_rate": 8.856945277484275e-07, "loss": 0.3683, "step": 16004 }, { "epoch": 0.8096008902827659, "grad_norm": 6.609098434448242, "learning_rate": 8.852385011798364e-07, "loss": 0.3517, "step": 16005 }, { "epoch": 0.8096514745308311, "grad_norm": 11.52004623413086, "learning_rate": 8.847825806387106e-07, "loss": 0.4911, "step": 16006 }, { "epoch": 0.8097020587788962, "grad_norm": 7.422525882720947, "learning_rate": 8.843267661367977e-07, "loss": 0.3413, "step": 16007 }, { "epoch": 0.8097526430269614, "grad_norm": 6.460195064544678, "learning_rate": 8.838710576858417e-07, "loss": 0.34, "step": 16008 }, { "epoch": 0.8098032272750265, "grad_norm": 18.758018493652344, "learning_rate": 8.834154552975882e-07, "loss": 0.5016, "step": 16009 }, { "epoch": 0.8098538115230917, "grad_norm": 13.263971328735352, "learning_rate": 8.829599589837745e-07, "loss": 0.2793, "step": 16010 }, { "epoch": 0.8099043957711569, "grad_norm": 7.755841255187988, "learning_rate": 8.825045687561373e-07, "loss": 0.3595, "step": 16011 }, { "epoch": 0.8099549800192221, "grad_norm": 4.64993953704834, "learning_rate": 8.820492846264134e-07, "loss": 0.3912, "step": 16012 }, { "epoch": 0.8100055642672872, "grad_norm": 5.4960618019104, "learning_rate": 8.815941066063327e-07, "loss": 0.3266, "step": 16013 }, { "epoch": 0.8100561485153523, "grad_norm": 5.847966194152832, "learning_rate": 8.811390347076231e-07, "loss": 0.3767, "step": 16014 }, { "epoch": 0.8101067327634175, "grad_norm": 8.888920783996582, "learning_rate": 8.806840689420132e-07, "loss": 0.3703, "step": 16015 }, { "epoch": 0.8101573170114826, "grad_norm": 10.336396217346191, "learning_rate": 8.80229209321225e-07, "loss": 0.4305, "step": 16016 }, { "epoch": 0.8102079012595478, "grad_norm": 7.011561393737793, "learning_rate": 8.797744558569782e-07, "loss": 0.4137, "step": 16017 }, { "epoch": 0.8102584855076129, "grad_norm": 7.395254611968994, "learning_rate": 8.793198085609927e-07, "loss": 0.2599, "step": 16018 }, { "epoch": 0.8103090697556781, "grad_norm": 9.796700477600098, "learning_rate": 8.788652674449837e-07, "loss": 0.3397, "step": 16019 }, { "epoch": 0.8103596540037432, "grad_norm": 8.45232105255127, "learning_rate": 8.784108325206603e-07, "loss": 0.3969, "step": 16020 }, { "epoch": 0.8104102382518084, "grad_norm": 7.032276630401611, "learning_rate": 8.779565037997357e-07, "loss": 0.3102, "step": 16021 }, { "epoch": 0.8104608224998735, "grad_norm": 13.659860610961914, "learning_rate": 8.775022812939154e-07, "loss": 0.4147, "step": 16022 }, { "epoch": 0.8105114067479386, "grad_norm": 6.227611064910889, "learning_rate": 8.770481650149026e-07, "loss": 0.4552, "step": 16023 }, { "epoch": 0.8105619909960038, "grad_norm": 7.806699275970459, "learning_rate": 8.765941549744017e-07, "loss": 0.3386, "step": 16024 }, { "epoch": 0.810612575244069, "grad_norm": 9.857954978942871, "learning_rate": 8.761402511841094e-07, "loss": 0.4369, "step": 16025 }, { "epoch": 0.8106631594921342, "grad_norm": 8.22953987121582, "learning_rate": 8.756864536557208e-07, "loss": 0.3446, "step": 16026 }, { "epoch": 0.8107137437401993, "grad_norm": 8.53304386138916, "learning_rate": 8.752327624009316e-07, "loss": 0.4472, "step": 16027 }, { "epoch": 0.8107643279882645, "grad_norm": 10.582586288452148, "learning_rate": 8.747791774314313e-07, "loss": 0.4109, "step": 16028 }, { "epoch": 0.8108149122363296, "grad_norm": 8.293700218200684, "learning_rate": 8.74325698758906e-07, "loss": 0.4019, "step": 16029 }, { "epoch": 0.8108654964843948, "grad_norm": 9.91114616394043, "learning_rate": 8.738723263950438e-07, "loss": 0.3099, "step": 16030 }, { "epoch": 0.8109160807324599, "grad_norm": 7.03908634185791, "learning_rate": 8.734190603515252e-07, "loss": 0.4028, "step": 16031 }, { "epoch": 0.810966664980525, "grad_norm": 11.10208511352539, "learning_rate": 8.729659006400304e-07, "loss": 0.5121, "step": 16032 }, { "epoch": 0.8110172492285902, "grad_norm": 18.504798889160156, "learning_rate": 8.725128472722361e-07, "loss": 0.4926, "step": 16033 }, { "epoch": 0.8110678334766553, "grad_norm": 11.132573127746582, "learning_rate": 8.720599002598157e-07, "loss": 0.5047, "step": 16034 }, { "epoch": 0.8111184177247205, "grad_norm": 12.19363784790039, "learning_rate": 8.7160705961444e-07, "loss": 0.4457, "step": 16035 }, { "epoch": 0.8111690019727856, "grad_norm": 14.928046226501465, "learning_rate": 8.7115432534778e-07, "loss": 0.3748, "step": 16036 }, { "epoch": 0.8112195862208508, "grad_norm": 4.786789417266846, "learning_rate": 8.707016974714999e-07, "loss": 0.4053, "step": 16037 }, { "epoch": 0.811270170468916, "grad_norm": 9.721277236938477, "learning_rate": 8.70249175997262e-07, "loss": 0.3683, "step": 16038 }, { "epoch": 0.8113207547169812, "grad_norm": 7.139232158660889, "learning_rate": 8.697967609367292e-07, "loss": 0.3407, "step": 16039 }, { "epoch": 0.8113713389650463, "grad_norm": 9.926217079162598, "learning_rate": 8.693444523015576e-07, "loss": 0.3217, "step": 16040 }, { "epoch": 0.8114219232131115, "grad_norm": 9.6978759765625, "learning_rate": 8.688922501034014e-07, "loss": 0.4005, "step": 16041 }, { "epoch": 0.8114725074611766, "grad_norm": 9.173956871032715, "learning_rate": 8.684401543539145e-07, "loss": 0.347, "step": 16042 }, { "epoch": 0.8115230917092418, "grad_norm": 17.623125076293945, "learning_rate": 8.67988165064746e-07, "loss": 0.5988, "step": 16043 }, { "epoch": 0.8115736759573069, "grad_norm": 6.504927635192871, "learning_rate": 8.675362822475414e-07, "loss": 0.3716, "step": 16044 }, { "epoch": 0.811624260205372, "grad_norm": 8.702786445617676, "learning_rate": 8.670845059139454e-07, "loss": 0.3237, "step": 16045 }, { "epoch": 0.8116748444534372, "grad_norm": 5.006816864013672, "learning_rate": 8.666328360755977e-07, "loss": 0.3652, "step": 16046 }, { "epoch": 0.8117254287015023, "grad_norm": 8.068327903747559, "learning_rate": 8.661812727441398e-07, "loss": 0.3419, "step": 16047 }, { "epoch": 0.8117760129495675, "grad_norm": 10.295090675354004, "learning_rate": 8.657298159312056e-07, "loss": 0.5333, "step": 16048 }, { "epoch": 0.8118265971976326, "grad_norm": 7.790319442749023, "learning_rate": 8.652784656484264e-07, "loss": 0.4674, "step": 16049 }, { "epoch": 0.8118771814456978, "grad_norm": 5.978642463684082, "learning_rate": 8.648272219074355e-07, "loss": 0.3686, "step": 16050 }, { "epoch": 0.811927765693763, "grad_norm": 5.797455787658691, "learning_rate": 8.64376084719859e-07, "loss": 0.3666, "step": 16051 }, { "epoch": 0.8119783499418282, "grad_norm": 10.256429672241211, "learning_rate": 8.639250540973204e-07, "loss": 0.3991, "step": 16052 }, { "epoch": 0.8120289341898933, "grad_norm": 6.7658562660217285, "learning_rate": 8.634741300514443e-07, "loss": 0.3648, "step": 16053 }, { "epoch": 0.8120795184379584, "grad_norm": 11.02212142944336, "learning_rate": 8.630233125938487e-07, "loss": 0.454, "step": 16054 }, { "epoch": 0.8121301026860236, "grad_norm": 14.67153263092041, "learning_rate": 8.625726017361491e-07, "loss": 0.429, "step": 16055 }, { "epoch": 0.8121806869340887, "grad_norm": 7.312024116516113, "learning_rate": 8.621219974899592e-07, "loss": 0.33, "step": 16056 }, { "epoch": 0.8122312711821539, "grad_norm": 10.421051025390625, "learning_rate": 8.616714998668934e-07, "loss": 0.4405, "step": 16057 }, { "epoch": 0.812281855430219, "grad_norm": 8.336252212524414, "learning_rate": 8.612211088785549e-07, "loss": 0.3457, "step": 16058 }, { "epoch": 0.8123324396782842, "grad_norm": 7.749817848205566, "learning_rate": 8.607708245365526e-07, "loss": 0.4495, "step": 16059 }, { "epoch": 0.8123830239263493, "grad_norm": 6.485888481140137, "learning_rate": 8.603206468524882e-07, "loss": 0.3444, "step": 16060 }, { "epoch": 0.8124336081744145, "grad_norm": 5.758327484130859, "learning_rate": 8.598705758379611e-07, "loss": 0.4146, "step": 16061 }, { "epoch": 0.8124841924224796, "grad_norm": 10.905997276306152, "learning_rate": 8.594206115045695e-07, "loss": 0.4211, "step": 16062 }, { "epoch": 0.8125347766705447, "grad_norm": 15.974621772766113, "learning_rate": 8.58970753863908e-07, "loss": 0.4795, "step": 16063 }, { "epoch": 0.81258536091861, "grad_norm": 9.090929985046387, "learning_rate": 8.58521002927567e-07, "loss": 0.5274, "step": 16064 }, { "epoch": 0.8126359451666751, "grad_norm": 9.436951637268066, "learning_rate": 8.580713587071377e-07, "loss": 0.4028, "step": 16065 }, { "epoch": 0.8126865294147403, "grad_norm": 5.804656982421875, "learning_rate": 8.576218212142046e-07, "loss": 0.3555, "step": 16066 }, { "epoch": 0.8127371136628054, "grad_norm": 11.358785629272461, "learning_rate": 8.571723904603502e-07, "loss": 0.405, "step": 16067 }, { "epoch": 0.8127876979108706, "grad_norm": 8.157910346984863, "learning_rate": 8.567230664571585e-07, "loss": 0.4593, "step": 16068 }, { "epoch": 0.8128382821589357, "grad_norm": 8.63980770111084, "learning_rate": 8.562738492162053e-07, "loss": 0.4752, "step": 16069 }, { "epoch": 0.8128888664070009, "grad_norm": 9.498369216918945, "learning_rate": 8.55824738749067e-07, "loss": 0.4528, "step": 16070 }, { "epoch": 0.812939450655066, "grad_norm": 4.799923419952393, "learning_rate": 8.553757350673147e-07, "loss": 0.3083, "step": 16071 }, { "epoch": 0.8129900349031312, "grad_norm": 8.473809242248535, "learning_rate": 8.549268381825188e-07, "loss": 0.4436, "step": 16072 }, { "epoch": 0.8130406191511963, "grad_norm": 14.131196022033691, "learning_rate": 8.544780481062454e-07, "loss": 0.408, "step": 16073 }, { "epoch": 0.8130912033992614, "grad_norm": 4.886383056640625, "learning_rate": 8.540293648500602e-07, "loss": 0.4185, "step": 16074 }, { "epoch": 0.8131417876473266, "grad_norm": 8.166326522827148, "learning_rate": 8.535807884255248e-07, "loss": 0.4064, "step": 16075 }, { "epoch": 0.8131923718953917, "grad_norm": 8.970775604248047, "learning_rate": 8.531323188441953e-07, "loss": 0.3993, "step": 16076 }, { "epoch": 0.8132429561434569, "grad_norm": 7.516655921936035, "learning_rate": 8.526839561176314e-07, "loss": 0.366, "step": 16077 }, { "epoch": 0.8132935403915221, "grad_norm": 10.881338119506836, "learning_rate": 8.522357002573845e-07, "loss": 0.4795, "step": 16078 }, { "epoch": 0.8133441246395873, "grad_norm": 10.713177680969238, "learning_rate": 8.517875512750034e-07, "loss": 0.419, "step": 16079 }, { "epoch": 0.8133947088876524, "grad_norm": 12.057212829589844, "learning_rate": 8.513395091820387e-07, "loss": 0.3756, "step": 16080 }, { "epoch": 0.8134452931357176, "grad_norm": 5.765747547149658, "learning_rate": 8.508915739900348e-07, "loss": 0.3645, "step": 16081 }, { "epoch": 0.8134958773837827, "grad_norm": 11.300307273864746, "learning_rate": 8.504437457105314e-07, "loss": 0.4672, "step": 16082 }, { "epoch": 0.8135464616318479, "grad_norm": 12.041031837463379, "learning_rate": 8.499960243550726e-07, "loss": 0.4457, "step": 16083 }, { "epoch": 0.813597045879913, "grad_norm": 5.7696051597595215, "learning_rate": 8.495484099351903e-07, "loss": 0.4754, "step": 16084 }, { "epoch": 0.8136476301279781, "grad_norm": 5.569696426391602, "learning_rate": 8.491009024624208e-07, "loss": 0.3882, "step": 16085 }, { "epoch": 0.8136982143760433, "grad_norm": 9.809414863586426, "learning_rate": 8.486535019482955e-07, "loss": 0.2824, "step": 16086 }, { "epoch": 0.8137487986241084, "grad_norm": 17.174720764160156, "learning_rate": 8.482062084043413e-07, "loss": 0.4277, "step": 16087 }, { "epoch": 0.8137993828721736, "grad_norm": 6.621825218200684, "learning_rate": 8.477590218420856e-07, "loss": 0.381, "step": 16088 }, { "epoch": 0.8138499671202387, "grad_norm": 5.894648551940918, "learning_rate": 8.473119422730508e-07, "loss": 0.3808, "step": 16089 }, { "epoch": 0.8139005513683039, "grad_norm": 15.075140953063965, "learning_rate": 8.468649697087555e-07, "loss": 0.4951, "step": 16090 }, { "epoch": 0.8139511356163691, "grad_norm": 7.8199591636657715, "learning_rate": 8.464181041607195e-07, "loss": 0.3754, "step": 16091 }, { "epoch": 0.8140017198644343, "grad_norm": 8.430864334106445, "learning_rate": 8.459713456404567e-07, "loss": 0.4382, "step": 16092 }, { "epoch": 0.8140523041124994, "grad_norm": 6.210926532745361, "learning_rate": 8.455246941594786e-07, "loss": 0.3067, "step": 16093 }, { "epoch": 0.8141028883605645, "grad_norm": 11.349469184875488, "learning_rate": 8.450781497292931e-07, "loss": 0.4867, "step": 16094 }, { "epoch": 0.8141534726086297, "grad_norm": 12.332843780517578, "learning_rate": 8.446317123614095e-07, "loss": 0.3296, "step": 16095 }, { "epoch": 0.8142040568566948, "grad_norm": 9.548707008361816, "learning_rate": 8.441853820673296e-07, "loss": 0.4323, "step": 16096 }, { "epoch": 0.81425464110476, "grad_norm": 8.506701469421387, "learning_rate": 8.437391588585542e-07, "loss": 0.4126, "step": 16097 }, { "epoch": 0.8143052253528251, "grad_norm": 9.974100112915039, "learning_rate": 8.432930427465818e-07, "loss": 0.3717, "step": 16098 }, { "epoch": 0.8143558096008903, "grad_norm": 5.919950008392334, "learning_rate": 8.428470337429062e-07, "loss": 0.3819, "step": 16099 }, { "epoch": 0.8144063938489554, "grad_norm": 12.069601058959961, "learning_rate": 8.424011318590225e-07, "loss": 0.4015, "step": 16100 }, { "epoch": 0.8144569780970206, "grad_norm": 7.938879489898682, "learning_rate": 8.419553371064199e-07, "loss": 0.3904, "step": 16101 }, { "epoch": 0.8145075623450857, "grad_norm": 6.082985877990723, "learning_rate": 8.41509649496583e-07, "loss": 0.3991, "step": 16102 }, { "epoch": 0.8145581465931508, "grad_norm": 5.97597599029541, "learning_rate": 8.410640690409999e-07, "loss": 0.3753, "step": 16103 }, { "epoch": 0.8146087308412161, "grad_norm": 5.6866960525512695, "learning_rate": 8.406185957511498e-07, "loss": 0.3163, "step": 16104 }, { "epoch": 0.8146593150892812, "grad_norm": 11.077863693237305, "learning_rate": 8.401732296385107e-07, "loss": 0.3742, "step": 16105 }, { "epoch": 0.8147098993373464, "grad_norm": 7.639617443084717, "learning_rate": 8.397279707145611e-07, "loss": 0.3723, "step": 16106 }, { "epoch": 0.8147604835854115, "grad_norm": 8.733803749084473, "learning_rate": 8.392828189907732e-07, "loss": 0.4078, "step": 16107 }, { "epoch": 0.8148110678334767, "grad_norm": 12.49404525756836, "learning_rate": 8.388377744786158e-07, "loss": 0.456, "step": 16108 }, { "epoch": 0.8148616520815418, "grad_norm": 6.063817977905273, "learning_rate": 8.383928371895605e-07, "loss": 0.3953, "step": 16109 }, { "epoch": 0.814912236329607, "grad_norm": 8.04171085357666, "learning_rate": 8.379480071350688e-07, "loss": 0.4382, "step": 16110 }, { "epoch": 0.8149628205776721, "grad_norm": 10.10139274597168, "learning_rate": 8.375032843266023e-07, "loss": 0.4915, "step": 16111 }, { "epoch": 0.8150134048257373, "grad_norm": 5.523719310760498, "learning_rate": 8.370586687756238e-07, "loss": 0.3078, "step": 16112 }, { "epoch": 0.8150639890738024, "grad_norm": 9.570328712463379, "learning_rate": 8.366141604935879e-07, "loss": 0.4022, "step": 16113 }, { "epoch": 0.8151145733218675, "grad_norm": 5.862454414367676, "learning_rate": 8.361697594919477e-07, "loss": 0.4167, "step": 16114 }, { "epoch": 0.8151651575699327, "grad_norm": 8.983474731445312, "learning_rate": 8.357254657821567e-07, "loss": 0.3543, "step": 16115 }, { "epoch": 0.8152157418179978, "grad_norm": 4.932450771331787, "learning_rate": 8.352812793756621e-07, "loss": 0.3061, "step": 16116 }, { "epoch": 0.815266326066063, "grad_norm": 5.175283432006836, "learning_rate": 8.348372002839084e-07, "loss": 0.3676, "step": 16117 }, { "epoch": 0.8153169103141282, "grad_norm": 7.766755104064941, "learning_rate": 8.343932285183403e-07, "loss": 0.4319, "step": 16118 }, { "epoch": 0.8153674945621934, "grad_norm": 5.491062641143799, "learning_rate": 8.339493640903973e-07, "loss": 0.2978, "step": 16119 }, { "epoch": 0.8154180788102585, "grad_norm": 4.849546909332275, "learning_rate": 8.335056070115155e-07, "loss": 0.3866, "step": 16120 }, { "epoch": 0.8154686630583237, "grad_norm": 10.260189056396484, "learning_rate": 8.330619572931314e-07, "loss": 0.4531, "step": 16121 }, { "epoch": 0.8155192473063888, "grad_norm": 8.798748016357422, "learning_rate": 8.32618414946676e-07, "loss": 0.4272, "step": 16122 }, { "epoch": 0.815569831554454, "grad_norm": 13.053387641906738, "learning_rate": 8.321749799835782e-07, "loss": 0.57, "step": 16123 }, { "epoch": 0.8156204158025191, "grad_norm": 11.63874340057373, "learning_rate": 8.317316524152641e-07, "loss": 0.5447, "step": 16124 }, { "epoch": 0.8156710000505842, "grad_norm": 11.505465507507324, "learning_rate": 8.312884322531562e-07, "loss": 0.4866, "step": 16125 }, { "epoch": 0.8157215842986494, "grad_norm": 8.675616264343262, "learning_rate": 8.308453195086774e-07, "loss": 0.4614, "step": 16126 }, { "epoch": 0.8157721685467145, "grad_norm": 15.736894607543945, "learning_rate": 8.304023141932444e-07, "loss": 0.3994, "step": 16127 }, { "epoch": 0.8158227527947797, "grad_norm": 9.05825138092041, "learning_rate": 8.299594163182723e-07, "loss": 0.4455, "step": 16128 }, { "epoch": 0.8158733370428448, "grad_norm": 6.690247535705566, "learning_rate": 8.295166258951731e-07, "loss": 0.3776, "step": 16129 }, { "epoch": 0.81592392129091, "grad_norm": 8.31482982635498, "learning_rate": 8.290739429353578e-07, "loss": 0.4734, "step": 16130 }, { "epoch": 0.8159745055389752, "grad_norm": 10.973403930664062, "learning_rate": 8.286313674502328e-07, "loss": 0.4101, "step": 16131 }, { "epoch": 0.8160250897870404, "grad_norm": 6.2771430015563965, "learning_rate": 8.281888994512005e-07, "loss": 0.3848, "step": 16132 }, { "epoch": 0.8160756740351055, "grad_norm": 11.38721752166748, "learning_rate": 8.277465389496653e-07, "loss": 0.4428, "step": 16133 }, { "epoch": 0.8161262582831706, "grad_norm": 8.140342712402344, "learning_rate": 8.273042859570235e-07, "loss": 0.3921, "step": 16134 }, { "epoch": 0.8161768425312358, "grad_norm": 8.835726737976074, "learning_rate": 8.268621404846722e-07, "loss": 0.4096, "step": 16135 }, { "epoch": 0.8162274267793009, "grad_norm": 9.423016548156738, "learning_rate": 8.264201025440033e-07, "loss": 0.4444, "step": 16136 }, { "epoch": 0.8162780110273661, "grad_norm": 20.015308380126953, "learning_rate": 8.259781721464061e-07, "loss": 0.4199, "step": 16137 }, { "epoch": 0.8163285952754312, "grad_norm": 16.8807430267334, "learning_rate": 8.255363493032708e-07, "loss": 0.4519, "step": 16138 }, { "epoch": 0.8163791795234964, "grad_norm": 8.74327564239502, "learning_rate": 8.250946340259813e-07, "loss": 0.3746, "step": 16139 }, { "epoch": 0.8164297637715615, "grad_norm": 8.467552185058594, "learning_rate": 8.246530263259173e-07, "loss": 0.323, "step": 16140 }, { "epoch": 0.8164803480196267, "grad_norm": 6.528263092041016, "learning_rate": 8.242115262144612e-07, "loss": 0.361, "step": 16141 }, { "epoch": 0.8165309322676918, "grad_norm": 10.742984771728516, "learning_rate": 8.237701337029874e-07, "loss": 0.4649, "step": 16142 }, { "epoch": 0.8165815165157569, "grad_norm": 9.461858749389648, "learning_rate": 8.23328848802869e-07, "loss": 0.4316, "step": 16143 }, { "epoch": 0.8166321007638222, "grad_norm": 7.629924297332764, "learning_rate": 8.228876715254791e-07, "loss": 0.4484, "step": 16144 }, { "epoch": 0.8166826850118873, "grad_norm": 9.526394844055176, "learning_rate": 8.224466018821847e-07, "loss": 0.3809, "step": 16145 }, { "epoch": 0.8167332692599525, "grad_norm": 9.603530883789062, "learning_rate": 8.220056398843496e-07, "loss": 0.4643, "step": 16146 }, { "epoch": 0.8167838535080176, "grad_norm": 10.269298553466797, "learning_rate": 8.215647855433395e-07, "loss": 0.4055, "step": 16147 }, { "epoch": 0.8168344377560828, "grad_norm": 14.288402557373047, "learning_rate": 8.211240388705117e-07, "loss": 0.3622, "step": 16148 }, { "epoch": 0.8168850220041479, "grad_norm": 10.165424346923828, "learning_rate": 8.206833998772223e-07, "loss": 0.519, "step": 16149 }, { "epoch": 0.8169356062522131, "grad_norm": 5.380599498748779, "learning_rate": 8.202428685748282e-07, "loss": 0.3099, "step": 16150 }, { "epoch": 0.8169861905002782, "grad_norm": 14.430251121520996, "learning_rate": 8.198024449746794e-07, "loss": 0.6202, "step": 16151 }, { "epoch": 0.8170367747483434, "grad_norm": 14.102234840393066, "learning_rate": 8.193621290881237e-07, "loss": 0.5198, "step": 16152 }, { "epoch": 0.8170873589964085, "grad_norm": 10.249556541442871, "learning_rate": 8.189219209265092e-07, "loss": 0.4432, "step": 16153 }, { "epoch": 0.8171379432444736, "grad_norm": 6.725090503692627, "learning_rate": 8.18481820501178e-07, "loss": 0.402, "step": 16154 }, { "epoch": 0.8171885274925388, "grad_norm": 5.746644973754883, "learning_rate": 8.180418278234692e-07, "loss": 0.3469, "step": 16155 }, { "epoch": 0.8172391117406039, "grad_norm": 5.9501142501831055, "learning_rate": 8.176019429047222e-07, "loss": 0.3304, "step": 16156 }, { "epoch": 0.8172896959886691, "grad_norm": 10.50622844696045, "learning_rate": 8.171621657562712e-07, "loss": 0.3952, "step": 16157 }, { "epoch": 0.8173402802367343, "grad_norm": 4.78309965133667, "learning_rate": 8.167224963894466e-07, "loss": 0.2373, "step": 16158 }, { "epoch": 0.8173908644847995, "grad_norm": 6.090713977813721, "learning_rate": 8.162829348155798e-07, "loss": 0.3879, "step": 16159 }, { "epoch": 0.8174414487328646, "grad_norm": 7.843443870544434, "learning_rate": 8.15843481045997e-07, "loss": 0.4463, "step": 16160 }, { "epoch": 0.8174920329809298, "grad_norm": 6.687201499938965, "learning_rate": 8.15404135092021e-07, "loss": 0.3779, "step": 16161 }, { "epoch": 0.8175426172289949, "grad_norm": 7.515946865081787, "learning_rate": 8.149648969649731e-07, "loss": 0.4392, "step": 16162 }, { "epoch": 0.81759320147706, "grad_norm": 8.716008186340332, "learning_rate": 8.145257666761703e-07, "loss": 0.4455, "step": 16163 }, { "epoch": 0.8176437857251252, "grad_norm": 8.680685043334961, "learning_rate": 8.140867442369299e-07, "loss": 0.4497, "step": 16164 }, { "epoch": 0.8176943699731903, "grad_norm": 12.267598152160645, "learning_rate": 8.136478296585631e-07, "loss": 0.6371, "step": 16165 }, { "epoch": 0.8177449542212555, "grad_norm": 6.222684383392334, "learning_rate": 8.132090229523804e-07, "loss": 0.2998, "step": 16166 }, { "epoch": 0.8177955384693206, "grad_norm": 5.2559614181518555, "learning_rate": 8.127703241296869e-07, "loss": 0.3605, "step": 16167 }, { "epoch": 0.8178461227173858, "grad_norm": 7.9239091873168945, "learning_rate": 8.123317332017899e-07, "loss": 0.4879, "step": 16168 }, { "epoch": 0.8178967069654509, "grad_norm": 7.977243900299072, "learning_rate": 8.118932501799887e-07, "loss": 0.372, "step": 16169 }, { "epoch": 0.8179472912135161, "grad_norm": 7.826719760894775, "learning_rate": 8.114548750755819e-07, "loss": 0.3644, "step": 16170 }, { "epoch": 0.8179978754615813, "grad_norm": 8.172456741333008, "learning_rate": 8.110166078998666e-07, "loss": 0.3817, "step": 16171 }, { "epoch": 0.8180484597096465, "grad_norm": 7.39457368850708, "learning_rate": 8.105784486641355e-07, "loss": 0.3993, "step": 16172 }, { "epoch": 0.8180990439577116, "grad_norm": 5.0753173828125, "learning_rate": 8.10140397379679e-07, "loss": 0.3372, "step": 16173 }, { "epoch": 0.8181496282057767, "grad_norm": 13.055134773254395, "learning_rate": 8.097024540577836e-07, "loss": 0.4508, "step": 16174 }, { "epoch": 0.8182002124538419, "grad_norm": 5.680482864379883, "learning_rate": 8.092646187097336e-07, "loss": 0.3139, "step": 16175 }, { "epoch": 0.818250796701907, "grad_norm": 7.832545280456543, "learning_rate": 8.08826891346814e-07, "loss": 0.5603, "step": 16176 }, { "epoch": 0.8183013809499722, "grad_norm": 12.593559265136719, "learning_rate": 8.08389271980301e-07, "loss": 0.5829, "step": 16177 }, { "epoch": 0.8183519651980373, "grad_norm": 11.844725608825684, "learning_rate": 8.079517606214715e-07, "loss": 0.4558, "step": 16178 }, { "epoch": 0.8184025494461025, "grad_norm": 8.955246925354004, "learning_rate": 8.075143572816007e-07, "loss": 0.4845, "step": 16179 }, { "epoch": 0.8184531336941676, "grad_norm": 8.691100120544434, "learning_rate": 8.070770619719586e-07, "loss": 0.4636, "step": 16180 }, { "epoch": 0.8185037179422328, "grad_norm": 7.7748613357543945, "learning_rate": 8.066398747038118e-07, "loss": 0.3128, "step": 16181 }, { "epoch": 0.8185543021902979, "grad_norm": 4.3226494789123535, "learning_rate": 8.062027954884278e-07, "loss": 0.3247, "step": 16182 }, { "epoch": 0.818604886438363, "grad_norm": 6.904781341552734, "learning_rate": 8.057658243370675e-07, "loss": 0.2515, "step": 16183 }, { "epoch": 0.8186554706864283, "grad_norm": 5.963559150695801, "learning_rate": 8.05328961260991e-07, "loss": 0.3726, "step": 16184 }, { "epoch": 0.8187060549344934, "grad_norm": 8.204361915588379, "learning_rate": 8.04892206271456e-07, "loss": 0.3593, "step": 16185 }, { "epoch": 0.8187566391825586, "grad_norm": 7.081148624420166, "learning_rate": 8.044555593797171e-07, "loss": 0.4084, "step": 16186 }, { "epoch": 0.8188072234306237, "grad_norm": 7.254886627197266, "learning_rate": 8.040190205970222e-07, "loss": 0.2823, "step": 16187 }, { "epoch": 0.8188578076786889, "grad_norm": 10.288411140441895, "learning_rate": 8.035825899346233e-07, "loss": 0.4346, "step": 16188 }, { "epoch": 0.818908391926754, "grad_norm": 8.698538780212402, "learning_rate": 8.031462674037655e-07, "loss": 0.3238, "step": 16189 }, { "epoch": 0.8189589761748192, "grad_norm": 5.880073547363281, "learning_rate": 8.027100530156895e-07, "loss": 0.3797, "step": 16190 }, { "epoch": 0.8190095604228843, "grad_norm": 5.152932643890381, "learning_rate": 8.022739467816387e-07, "loss": 0.364, "step": 16191 }, { "epoch": 0.8190601446709495, "grad_norm": 12.837448120117188, "learning_rate": 8.018379487128492e-07, "loss": 0.4604, "step": 16192 }, { "epoch": 0.8191107289190146, "grad_norm": 10.398165702819824, "learning_rate": 8.01402058820554e-07, "loss": 0.43, "step": 16193 }, { "epoch": 0.8191613131670797, "grad_norm": 7.216859340667725, "learning_rate": 8.009662771159881e-07, "loss": 0.3867, "step": 16194 }, { "epoch": 0.8192118974151449, "grad_norm": 9.620635032653809, "learning_rate": 8.005306036103788e-07, "loss": 0.3357, "step": 16195 }, { "epoch": 0.81926248166321, "grad_norm": 6.967536926269531, "learning_rate": 8.000950383149509e-07, "loss": 0.4533, "step": 16196 }, { "epoch": 0.8193130659112752, "grad_norm": 6.884466648101807, "learning_rate": 7.996595812409308e-07, "loss": 0.3455, "step": 16197 }, { "epoch": 0.8193636501593404, "grad_norm": 7.677276134490967, "learning_rate": 7.992242323995381e-07, "loss": 0.4831, "step": 16198 }, { "epoch": 0.8194142344074056, "grad_norm": 4.976724147796631, "learning_rate": 7.987889918019898e-07, "loss": 0.3124, "step": 16199 }, { "epoch": 0.8194648186554707, "grad_norm": 7.433210849761963, "learning_rate": 7.983538594595019e-07, "loss": 0.3935, "step": 16200 }, { "epoch": 0.8195154029035359, "grad_norm": 11.93384075164795, "learning_rate": 7.979188353832867e-07, "loss": 0.4416, "step": 16201 }, { "epoch": 0.819565987151601, "grad_norm": 7.3310699462890625, "learning_rate": 7.974839195845518e-07, "loss": 0.3424, "step": 16202 }, { "epoch": 0.8196165713996661, "grad_norm": 6.296760082244873, "learning_rate": 7.970491120745066e-07, "loss": 0.3701, "step": 16203 }, { "epoch": 0.8196671556477313, "grad_norm": 5.907044887542725, "learning_rate": 7.966144128643543e-07, "loss": 0.3346, "step": 16204 }, { "epoch": 0.8197177398957964, "grad_norm": 14.047477722167969, "learning_rate": 7.961798219652944e-07, "loss": 0.5121, "step": 16205 }, { "epoch": 0.8197683241438616, "grad_norm": 6.893645286560059, "learning_rate": 7.957453393885279e-07, "loss": 0.3354, "step": 16206 }, { "epoch": 0.8198189083919267, "grad_norm": 5.851705074310303, "learning_rate": 7.953109651452484e-07, "loss": 0.3629, "step": 16207 }, { "epoch": 0.8198694926399919, "grad_norm": 13.237648010253906, "learning_rate": 7.948766992466489e-07, "loss": 0.4676, "step": 16208 }, { "epoch": 0.819920076888057, "grad_norm": 7.917253494262695, "learning_rate": 7.944425417039209e-07, "loss": 0.3541, "step": 16209 }, { "epoch": 0.8199706611361222, "grad_norm": 13.27137279510498, "learning_rate": 7.940084925282498e-07, "loss": 0.5584, "step": 16210 }, { "epoch": 0.8200212453841874, "grad_norm": 5.346341609954834, "learning_rate": 7.935745517308202e-07, "loss": 0.398, "step": 16211 }, { "epoch": 0.8200718296322526, "grad_norm": 15.089237213134766, "learning_rate": 7.931407193228163e-07, "loss": 0.4978, "step": 16212 }, { "epoch": 0.8201224138803177, "grad_norm": 21.90867042541504, "learning_rate": 7.927069953154126e-07, "loss": 0.3516, "step": 16213 }, { "epoch": 0.8201729981283828, "grad_norm": 7.352540493011475, "learning_rate": 7.922733797197885e-07, "loss": 0.2627, "step": 16214 }, { "epoch": 0.820223582376448, "grad_norm": 5.195470333099365, "learning_rate": 7.91839872547116e-07, "loss": 0.3619, "step": 16215 }, { "epoch": 0.8202741666245131, "grad_norm": 5.53196382522583, "learning_rate": 7.914064738085641e-07, "loss": 0.3238, "step": 16216 }, { "epoch": 0.8203247508725783, "grad_norm": 9.175037384033203, "learning_rate": 7.909731835153033e-07, "loss": 0.4238, "step": 16217 }, { "epoch": 0.8203753351206434, "grad_norm": 5.834953784942627, "learning_rate": 7.905400016784976e-07, "loss": 0.3309, "step": 16218 }, { "epoch": 0.8204259193687086, "grad_norm": 11.388588905334473, "learning_rate": 7.901069283093066e-07, "loss": 0.4849, "step": 16219 }, { "epoch": 0.8204765036167737, "grad_norm": 5.215647220611572, "learning_rate": 7.896739634188933e-07, "loss": 0.3928, "step": 16220 }, { "epoch": 0.8205270878648389, "grad_norm": 10.473550796508789, "learning_rate": 7.89241107018412e-07, "loss": 0.4281, "step": 16221 }, { "epoch": 0.820577672112904, "grad_norm": 6.576726913452148, "learning_rate": 7.888083591190154e-07, "loss": 0.2809, "step": 16222 }, { "epoch": 0.8206282563609691, "grad_norm": 6.307332992553711, "learning_rate": 7.88375719731857e-07, "loss": 0.4806, "step": 16223 }, { "epoch": 0.8206788406090344, "grad_norm": 7.797133922576904, "learning_rate": 7.879431888680833e-07, "loss": 0.3464, "step": 16224 }, { "epoch": 0.8207294248570995, "grad_norm": 6.275196552276611, "learning_rate": 7.875107665388404e-07, "loss": 0.3113, "step": 16225 }, { "epoch": 0.8207800091051647, "grad_norm": 13.687738418579102, "learning_rate": 7.8707845275527e-07, "loss": 0.5598, "step": 16226 }, { "epoch": 0.8208305933532298, "grad_norm": 9.741467475891113, "learning_rate": 7.866462475285114e-07, "loss": 0.3784, "step": 16227 }, { "epoch": 0.820881177601295, "grad_norm": 9.058552742004395, "learning_rate": 7.862141508697013e-07, "loss": 0.3821, "step": 16228 }, { "epoch": 0.8209317618493601, "grad_norm": 9.249876976013184, "learning_rate": 7.857821627899759e-07, "loss": 0.5068, "step": 16229 }, { "epoch": 0.8209823460974253, "grad_norm": 8.405752182006836, "learning_rate": 7.85350283300465e-07, "loss": 0.378, "step": 16230 }, { "epoch": 0.8210329303454904, "grad_norm": 7.745815277099609, "learning_rate": 7.84918512412296e-07, "loss": 0.3288, "step": 16231 }, { "epoch": 0.8210835145935556, "grad_norm": 7.273393154144287, "learning_rate": 7.844868501365971e-07, "loss": 0.4122, "step": 16232 }, { "epoch": 0.8211340988416207, "grad_norm": 5.881971836090088, "learning_rate": 7.840552964844894e-07, "loss": 0.3948, "step": 16233 }, { "epoch": 0.8211846830896858, "grad_norm": 10.350899696350098, "learning_rate": 7.83623851467093e-07, "loss": 0.4271, "step": 16234 }, { "epoch": 0.821235267337751, "grad_norm": 9.658939361572266, "learning_rate": 7.831925150955266e-07, "loss": 0.4291, "step": 16235 }, { "epoch": 0.8212858515858161, "grad_norm": 9.920923233032227, "learning_rate": 7.827612873809037e-07, "loss": 0.3883, "step": 16236 }, { "epoch": 0.8213364358338814, "grad_norm": 9.902170181274414, "learning_rate": 7.823301683343365e-07, "loss": 0.3881, "step": 16237 }, { "epoch": 0.8213870200819465, "grad_norm": 5.035055160522461, "learning_rate": 7.818991579669333e-07, "loss": 0.31, "step": 16238 }, { "epoch": 0.8214376043300117, "grad_norm": 17.481035232543945, "learning_rate": 7.814682562898001e-07, "loss": 0.5098, "step": 16239 }, { "epoch": 0.8214881885780768, "grad_norm": 12.336529731750488, "learning_rate": 7.8103746331404e-07, "loss": 0.3784, "step": 16240 }, { "epoch": 0.821538772826142, "grad_norm": 7.152500629425049, "learning_rate": 7.806067790507543e-07, "loss": 0.3984, "step": 16241 }, { "epoch": 0.8215893570742071, "grad_norm": 8.877203941345215, "learning_rate": 7.801762035110411e-07, "loss": 0.4657, "step": 16242 }, { "epoch": 0.8216399413222722, "grad_norm": 10.941366195678711, "learning_rate": 7.79745736705993e-07, "loss": 0.3709, "step": 16243 }, { "epoch": 0.8216905255703374, "grad_norm": 7.036243915557861, "learning_rate": 7.793153786467051e-07, "loss": 0.3967, "step": 16244 }, { "epoch": 0.8217411098184025, "grad_norm": 6.25759220123291, "learning_rate": 7.788851293442651e-07, "loss": 0.3876, "step": 16245 }, { "epoch": 0.8217916940664677, "grad_norm": 6.960319995880127, "learning_rate": 7.784549888097581e-07, "loss": 0.321, "step": 16246 }, { "epoch": 0.8218422783145328, "grad_norm": 5.823221683502197, "learning_rate": 7.780249570542708e-07, "loss": 0.4191, "step": 16247 }, { "epoch": 0.821892862562598, "grad_norm": 8.351774215698242, "learning_rate": 7.775950340888827e-07, "loss": 0.3365, "step": 16248 }, { "epoch": 0.8219434468106631, "grad_norm": 5.064939498901367, "learning_rate": 7.771652199246704e-07, "loss": 0.3125, "step": 16249 }, { "epoch": 0.8219940310587283, "grad_norm": 12.986598014831543, "learning_rate": 7.76735514572713e-07, "loss": 0.371, "step": 16250 }, { "epoch": 0.8220446153067935, "grad_norm": 7.752023220062256, "learning_rate": 7.763059180440785e-07, "loss": 0.4051, "step": 16251 }, { "epoch": 0.8220951995548587, "grad_norm": 5.090603351593018, "learning_rate": 7.758764303498389e-07, "loss": 0.3936, "step": 16252 }, { "epoch": 0.8221457838029238, "grad_norm": 13.007109642028809, "learning_rate": 7.754470515010615e-07, "loss": 0.3232, "step": 16253 }, { "epoch": 0.822196368050989, "grad_norm": 12.141980171203613, "learning_rate": 7.750177815088084e-07, "loss": 0.4725, "step": 16254 }, { "epoch": 0.8222469522990541, "grad_norm": 7.867948055267334, "learning_rate": 7.74588620384143e-07, "loss": 0.323, "step": 16255 }, { "epoch": 0.8222975365471192, "grad_norm": 6.793512344360352, "learning_rate": 7.741595681381225e-07, "loss": 0.4248, "step": 16256 }, { "epoch": 0.8223481207951844, "grad_norm": 7.12454080581665, "learning_rate": 7.737306247818021e-07, "loss": 0.4569, "step": 16257 }, { "epoch": 0.8223987050432495, "grad_norm": 7.2475972175598145, "learning_rate": 7.733017903262363e-07, "loss": 0.4071, "step": 16258 }, { "epoch": 0.8224492892913147, "grad_norm": 7.933985710144043, "learning_rate": 7.728730647824745e-07, "loss": 0.3624, "step": 16259 }, { "epoch": 0.8224998735393798, "grad_norm": 11.366252899169922, "learning_rate": 7.724444481615617e-07, "loss": 0.3736, "step": 16260 }, { "epoch": 0.822550457787445, "grad_norm": 10.710094451904297, "learning_rate": 7.720159404745459e-07, "loss": 0.3749, "step": 16261 }, { "epoch": 0.8226010420355101, "grad_norm": 9.810315132141113, "learning_rate": 7.715875417324676e-07, "loss": 0.4095, "step": 16262 }, { "epoch": 0.8226516262835752, "grad_norm": 9.702834129333496, "learning_rate": 7.71159251946364e-07, "loss": 0.3795, "step": 16263 }, { "epoch": 0.8227022105316405, "grad_norm": 6.4298481941223145, "learning_rate": 7.707310711272726e-07, "loss": 0.3287, "step": 16264 }, { "epoch": 0.8227527947797056, "grad_norm": 5.1884942054748535, "learning_rate": 7.703029992862265e-07, "loss": 0.3546, "step": 16265 }, { "epoch": 0.8228033790277708, "grad_norm": 7.471543788909912, "learning_rate": 7.698750364342539e-07, "loss": 0.3978, "step": 16266 }, { "epoch": 0.8228539632758359, "grad_norm": 6.561403751373291, "learning_rate": 7.694471825823857e-07, "loss": 0.3724, "step": 16267 }, { "epoch": 0.8229045475239011, "grad_norm": 8.304793357849121, "learning_rate": 7.690194377416449e-07, "loss": 0.3875, "step": 16268 }, { "epoch": 0.8229551317719662, "grad_norm": 6.757841110229492, "learning_rate": 7.685918019230526e-07, "loss": 0.3572, "step": 16269 }, { "epoch": 0.8230057160200314, "grad_norm": 5.7282257080078125, "learning_rate": 7.681642751376301e-07, "loss": 0.2408, "step": 16270 }, { "epoch": 0.8230563002680965, "grad_norm": 8.340642929077148, "learning_rate": 7.677368573963928e-07, "loss": 0.4527, "step": 16271 }, { "epoch": 0.8231068845161617, "grad_norm": 16.26469612121582, "learning_rate": 7.673095487103527e-07, "loss": 0.5397, "step": 16272 }, { "epoch": 0.8231574687642268, "grad_norm": 8.598062515258789, "learning_rate": 7.668823490905236e-07, "loss": 0.4032, "step": 16273 }, { "epoch": 0.8232080530122919, "grad_norm": 11.1759033203125, "learning_rate": 7.664552585479112e-07, "loss": 0.4187, "step": 16274 }, { "epoch": 0.8232586372603571, "grad_norm": 6.108724117279053, "learning_rate": 7.660282770935201e-07, "loss": 0.3542, "step": 16275 }, { "epoch": 0.8233092215084222, "grad_norm": 12.022480964660645, "learning_rate": 7.656014047383559e-07, "loss": 0.3833, "step": 16276 }, { "epoch": 0.8233598057564875, "grad_norm": 8.893404960632324, "learning_rate": 7.651746414934142e-07, "loss": 0.3065, "step": 16277 }, { "epoch": 0.8234103900045526, "grad_norm": 5.765213966369629, "learning_rate": 7.647479873696923e-07, "loss": 0.419, "step": 16278 }, { "epoch": 0.8234609742526178, "grad_norm": 8.694354057312012, "learning_rate": 7.643214423781858e-07, "loss": 0.2927, "step": 16279 }, { "epoch": 0.8235115585006829, "grad_norm": 6.655850887298584, "learning_rate": 7.638950065298856e-07, "loss": 0.3717, "step": 16280 }, { "epoch": 0.8235621427487481, "grad_norm": 5.42208194732666, "learning_rate": 7.634686798357771e-07, "loss": 0.3439, "step": 16281 }, { "epoch": 0.8236127269968132, "grad_norm": 10.820632934570312, "learning_rate": 7.630424623068495e-07, "loss": 0.4133, "step": 16282 }, { "epoch": 0.8236633112448783, "grad_norm": 7.037374973297119, "learning_rate": 7.626163539540838e-07, "loss": 0.3121, "step": 16283 }, { "epoch": 0.8237138954929435, "grad_norm": 5.685662269592285, "learning_rate": 7.621903547884585e-07, "loss": 0.3598, "step": 16284 }, { "epoch": 0.8237644797410086, "grad_norm": 8.885186195373535, "learning_rate": 7.617644648209527e-07, "loss": 0.3226, "step": 16285 }, { "epoch": 0.8238150639890738, "grad_norm": 5.209062576293945, "learning_rate": 7.613386840625398e-07, "loss": 0.3189, "step": 16286 }, { "epoch": 0.8238656482371389, "grad_norm": 16.693763732910156, "learning_rate": 7.609130125241893e-07, "loss": 0.4806, "step": 16287 }, { "epoch": 0.8239162324852041, "grad_norm": 7.519748687744141, "learning_rate": 7.60487450216873e-07, "loss": 0.4134, "step": 16288 }, { "epoch": 0.8239668167332692, "grad_norm": 7.5962958335876465, "learning_rate": 7.60061997151555e-07, "loss": 0.3465, "step": 16289 }, { "epoch": 0.8240174009813344, "grad_norm": 9.16055965423584, "learning_rate": 7.59636653339198e-07, "loss": 0.5208, "step": 16290 }, { "epoch": 0.8240679852293996, "grad_norm": 6.662381172180176, "learning_rate": 7.592114187907623e-07, "loss": 0.3839, "step": 16291 }, { "epoch": 0.8241185694774648, "grad_norm": 11.439634323120117, "learning_rate": 7.587862935172036e-07, "loss": 0.4168, "step": 16292 }, { "epoch": 0.8241691537255299, "grad_norm": 6.268486976623535, "learning_rate": 7.583612775294796e-07, "loss": 0.4483, "step": 16293 }, { "epoch": 0.824219737973595, "grad_norm": 9.675176620483398, "learning_rate": 7.579363708385402e-07, "loss": 0.4654, "step": 16294 }, { "epoch": 0.8242703222216602, "grad_norm": 10.945333480834961, "learning_rate": 7.575115734553328e-07, "loss": 0.5087, "step": 16295 }, { "epoch": 0.8243209064697253, "grad_norm": 12.265759468078613, "learning_rate": 7.570868853908058e-07, "loss": 0.3802, "step": 16296 }, { "epoch": 0.8243714907177905, "grad_norm": 6.365199089050293, "learning_rate": 7.56662306655902e-07, "loss": 0.4167, "step": 16297 }, { "epoch": 0.8244220749658556, "grad_norm": 5.332681179046631, "learning_rate": 7.562378372615597e-07, "loss": 0.3954, "step": 16298 }, { "epoch": 0.8244726592139208, "grad_norm": 4.905795574188232, "learning_rate": 7.558134772187193e-07, "loss": 0.3152, "step": 16299 }, { "epoch": 0.8245232434619859, "grad_norm": 4.6088690757751465, "learning_rate": 7.553892265383139e-07, "loss": 0.3855, "step": 16300 }, { "epoch": 0.824573827710051, "grad_norm": 8.544565200805664, "learning_rate": 7.549650852312763e-07, "loss": 0.3991, "step": 16301 }, { "epoch": 0.8246244119581162, "grad_norm": 7.939999580383301, "learning_rate": 7.54541053308534e-07, "loss": 0.4171, "step": 16302 }, { "epoch": 0.8246749962061813, "grad_norm": 6.989412307739258, "learning_rate": 7.541171307810152e-07, "loss": 0.3875, "step": 16303 }, { "epoch": 0.8247255804542466, "grad_norm": 11.291298866271973, "learning_rate": 7.536933176596405e-07, "loss": 0.5133, "step": 16304 }, { "epoch": 0.8247761647023117, "grad_norm": 9.194204330444336, "learning_rate": 7.532696139553342e-07, "loss": 0.3446, "step": 16305 }, { "epoch": 0.8248267489503769, "grad_norm": 11.763035774230957, "learning_rate": 7.528460196790122e-07, "loss": 0.4132, "step": 16306 }, { "epoch": 0.824877333198442, "grad_norm": 8.415726661682129, "learning_rate": 7.524225348415886e-07, "loss": 0.4384, "step": 16307 }, { "epoch": 0.8249279174465072, "grad_norm": 6.422872066497803, "learning_rate": 7.519991594539778e-07, "loss": 0.3662, "step": 16308 }, { "epoch": 0.8249785016945723, "grad_norm": 5.837835311889648, "learning_rate": 7.515758935270878e-07, "loss": 0.3493, "step": 16309 }, { "epoch": 0.8250290859426375, "grad_norm": 10.591313362121582, "learning_rate": 7.511527370718247e-07, "loss": 0.4143, "step": 16310 }, { "epoch": 0.8250796701907026, "grad_norm": 6.754969596862793, "learning_rate": 7.507296900990934e-07, "loss": 0.3826, "step": 16311 }, { "epoch": 0.8251302544387678, "grad_norm": 11.183303833007812, "learning_rate": 7.503067526197949e-07, "loss": 0.3848, "step": 16312 }, { "epoch": 0.8251808386868329, "grad_norm": 5.841280460357666, "learning_rate": 7.498839246448253e-07, "loss": 0.4073, "step": 16313 }, { "epoch": 0.825231422934898, "grad_norm": 11.336267471313477, "learning_rate": 7.494612061850826e-07, "loss": 0.4645, "step": 16314 }, { "epoch": 0.8252820071829632, "grad_norm": 12.895898818969727, "learning_rate": 7.490385972514585e-07, "loss": 0.3979, "step": 16315 }, { "epoch": 0.8253325914310283, "grad_norm": 8.79236125946045, "learning_rate": 7.486160978548401e-07, "loss": 0.318, "step": 16316 }, { "epoch": 0.8253831756790936, "grad_norm": 8.750725746154785, "learning_rate": 7.481937080061169e-07, "loss": 0.3904, "step": 16317 }, { "epoch": 0.8254337599271587, "grad_norm": 5.7603840827941895, "learning_rate": 7.477714277161724e-07, "loss": 0.3739, "step": 16318 }, { "epoch": 0.8254843441752239, "grad_norm": 5.373595714569092, "learning_rate": 7.473492569958863e-07, "loss": 0.3392, "step": 16319 }, { "epoch": 0.825534928423289, "grad_norm": 8.43531322479248, "learning_rate": 7.469271958561386e-07, "loss": 0.3386, "step": 16320 }, { "epoch": 0.8255855126713542, "grad_norm": 8.136924743652344, "learning_rate": 7.465052443078041e-07, "loss": 0.373, "step": 16321 }, { "epoch": 0.8256360969194193, "grad_norm": 17.230966567993164, "learning_rate": 7.460834023617553e-07, "loss": 0.5356, "step": 16322 }, { "epoch": 0.8256866811674844, "grad_norm": 6.95109224319458, "learning_rate": 7.456616700288627e-07, "loss": 0.4285, "step": 16323 }, { "epoch": 0.8257372654155496, "grad_norm": 5.519575595855713, "learning_rate": 7.452400473199934e-07, "loss": 0.3475, "step": 16324 }, { "epoch": 0.8257878496636147, "grad_norm": 7.650576114654541, "learning_rate": 7.448185342460095e-07, "loss": 0.2465, "step": 16325 }, { "epoch": 0.8258384339116799, "grad_norm": 8.272810935974121, "learning_rate": 7.443971308177749e-07, "loss": 0.3479, "step": 16326 }, { "epoch": 0.825889018159745, "grad_norm": 9.661664962768555, "learning_rate": 7.439758370461481e-07, "loss": 0.3992, "step": 16327 }, { "epoch": 0.8259396024078102, "grad_norm": 11.967638969421387, "learning_rate": 7.435546529419834e-07, "loss": 0.568, "step": 16328 }, { "epoch": 0.8259901866558753, "grad_norm": 8.103960990905762, "learning_rate": 7.431335785161347e-07, "loss": 0.3713, "step": 16329 }, { "epoch": 0.8260407709039405, "grad_norm": 7.10660982131958, "learning_rate": 7.427126137794499e-07, "loss": 0.3598, "step": 16330 }, { "epoch": 0.8260913551520057, "grad_norm": 7.818777084350586, "learning_rate": 7.422917587427797e-07, "loss": 0.4389, "step": 16331 }, { "epoch": 0.8261419394000709, "grad_norm": 16.269474029541016, "learning_rate": 7.418710134169665e-07, "loss": 0.4998, "step": 16332 }, { "epoch": 0.826192523648136, "grad_norm": 8.85745620727539, "learning_rate": 7.414503778128513e-07, "loss": 0.4278, "step": 16333 }, { "epoch": 0.8262431078962011, "grad_norm": 15.611037254333496, "learning_rate": 7.410298519412745e-07, "loss": 0.42, "step": 16334 }, { "epoch": 0.8262936921442663, "grad_norm": 9.838761329650879, "learning_rate": 7.406094358130717e-07, "loss": 0.4328, "step": 16335 }, { "epoch": 0.8263442763923314, "grad_norm": 15.809086799621582, "learning_rate": 7.401891294390756e-07, "loss": 0.4116, "step": 16336 }, { "epoch": 0.8263948606403966, "grad_norm": 5.938907623291016, "learning_rate": 7.39768932830115e-07, "loss": 0.41, "step": 16337 }, { "epoch": 0.8264454448884617, "grad_norm": 10.060009956359863, "learning_rate": 7.393488459970205e-07, "loss": 0.4476, "step": 16338 }, { "epoch": 0.8264960291365269, "grad_norm": 8.173108100891113, "learning_rate": 7.389288689506146e-07, "loss": 0.3841, "step": 16339 }, { "epoch": 0.826546613384592, "grad_norm": 6.8538641929626465, "learning_rate": 7.385090017017194e-07, "loss": 0.3889, "step": 16340 }, { "epoch": 0.8265971976326572, "grad_norm": 5.760900497436523, "learning_rate": 7.380892442611543e-07, "loss": 0.2508, "step": 16341 }, { "epoch": 0.8266477818807223, "grad_norm": 6.375190258026123, "learning_rate": 7.37669596639734e-07, "loss": 0.3844, "step": 16342 }, { "epoch": 0.8266983661287874, "grad_norm": 14.261260986328125, "learning_rate": 7.372500588482745e-07, "loss": 0.4614, "step": 16343 }, { "epoch": 0.8267489503768527, "grad_norm": 11.293267250061035, "learning_rate": 7.368306308975842e-07, "loss": 0.4146, "step": 16344 }, { "epoch": 0.8267995346249178, "grad_norm": 5.548030853271484, "learning_rate": 7.364113127984701e-07, "loss": 0.3519, "step": 16345 }, { "epoch": 0.826850118872983, "grad_norm": 9.917793273925781, "learning_rate": 7.359921045617396e-07, "loss": 0.4899, "step": 16346 }, { "epoch": 0.8269007031210481, "grad_norm": 6.423942565917969, "learning_rate": 7.35573006198193e-07, "loss": 0.3537, "step": 16347 }, { "epoch": 0.8269512873691133, "grad_norm": 7.805484771728516, "learning_rate": 7.351540177186289e-07, "loss": 0.3774, "step": 16348 }, { "epoch": 0.8270018716171784, "grad_norm": 15.269558906555176, "learning_rate": 7.347351391338453e-07, "loss": 0.4705, "step": 16349 }, { "epoch": 0.8270524558652436, "grad_norm": 6.711050987243652, "learning_rate": 7.343163704546352e-07, "loss": 0.4666, "step": 16350 }, { "epoch": 0.8271030401133087, "grad_norm": 7.676932334899902, "learning_rate": 7.338977116917873e-07, "loss": 0.4267, "step": 16351 }, { "epoch": 0.8271536243613739, "grad_norm": 7.880714416503906, "learning_rate": 7.334791628560928e-07, "loss": 0.3482, "step": 16352 }, { "epoch": 0.827204208609439, "grad_norm": 9.338972091674805, "learning_rate": 7.330607239583354e-07, "loss": 0.4033, "step": 16353 }, { "epoch": 0.8272547928575041, "grad_norm": 8.434593200683594, "learning_rate": 7.326423950092948e-07, "loss": 0.4272, "step": 16354 }, { "epoch": 0.8273053771055693, "grad_norm": 7.184324741363525, "learning_rate": 7.322241760197535e-07, "loss": 0.3881, "step": 16355 }, { "epoch": 0.8273559613536344, "grad_norm": 6.242915630340576, "learning_rate": 7.318060670004867e-07, "loss": 0.2912, "step": 16356 }, { "epoch": 0.8274065456016997, "grad_norm": 6.675478935241699, "learning_rate": 7.31388067962267e-07, "loss": 0.353, "step": 16357 }, { "epoch": 0.8274571298497648, "grad_norm": 7.162961483001709, "learning_rate": 7.309701789158674e-07, "loss": 0.3794, "step": 16358 }, { "epoch": 0.82750771409783, "grad_norm": 8.688911437988281, "learning_rate": 7.305523998720549e-07, "loss": 0.4809, "step": 16359 }, { "epoch": 0.8275582983458951, "grad_norm": 7.783986568450928, "learning_rate": 7.30134730841594e-07, "loss": 0.3684, "step": 16360 }, { "epoch": 0.8276088825939603, "grad_norm": 6.702226638793945, "learning_rate": 7.297171718352481e-07, "loss": 0.3046, "step": 16361 }, { "epoch": 0.8276594668420254, "grad_norm": 6.700779438018799, "learning_rate": 7.292997228637766e-07, "loss": 0.3867, "step": 16362 }, { "epoch": 0.8277100510900905, "grad_norm": 11.484940528869629, "learning_rate": 7.288823839379345e-07, "loss": 0.4304, "step": 16363 }, { "epoch": 0.8277606353381557, "grad_norm": 9.439107894897461, "learning_rate": 7.28465155068478e-07, "loss": 0.4239, "step": 16364 }, { "epoch": 0.8278112195862208, "grad_norm": 11.248734474182129, "learning_rate": 7.280480362661574e-07, "loss": 0.3127, "step": 16365 }, { "epoch": 0.827861803834286, "grad_norm": 6.524459362030029, "learning_rate": 7.276310275417197e-07, "loss": 0.2944, "step": 16366 }, { "epoch": 0.8279123880823511, "grad_norm": 26.36747932434082, "learning_rate": 7.272141289059115e-07, "loss": 0.5418, "step": 16367 }, { "epoch": 0.8279629723304163, "grad_norm": 13.418981552124023, "learning_rate": 7.267973403694734e-07, "loss": 0.4585, "step": 16368 }, { "epoch": 0.8280135565784814, "grad_norm": 7.243435382843018, "learning_rate": 7.263806619431473e-07, "loss": 0.3581, "step": 16369 }, { "epoch": 0.8280641408265467, "grad_norm": 9.633676528930664, "learning_rate": 7.259640936376694e-07, "loss": 0.3867, "step": 16370 }, { "epoch": 0.8281147250746118, "grad_norm": 4.701773166656494, "learning_rate": 7.255476354637719e-07, "loss": 0.3266, "step": 16371 }, { "epoch": 0.828165309322677, "grad_norm": 3.8057055473327637, "learning_rate": 7.251312874321887e-07, "loss": 0.3295, "step": 16372 }, { "epoch": 0.8282158935707421, "grad_norm": 6.844332695007324, "learning_rate": 7.247150495536464e-07, "loss": 0.3284, "step": 16373 }, { "epoch": 0.8282664778188072, "grad_norm": 17.964893341064453, "learning_rate": 7.242989218388713e-07, "loss": 0.3524, "step": 16374 }, { "epoch": 0.8283170620668724, "grad_norm": 9.696216583251953, "learning_rate": 7.238829042985834e-07, "loss": 0.355, "step": 16375 }, { "epoch": 0.8283676463149375, "grad_norm": 11.3316011428833, "learning_rate": 7.234669969435066e-07, "loss": 0.4917, "step": 16376 }, { "epoch": 0.8284182305630027, "grad_norm": 9.094144821166992, "learning_rate": 7.230511997843554e-07, "loss": 0.3372, "step": 16377 }, { "epoch": 0.8284688148110678, "grad_norm": 10.089142799377441, "learning_rate": 7.226355128318429e-07, "loss": 0.4075, "step": 16378 }, { "epoch": 0.828519399059133, "grad_norm": 10.240954399108887, "learning_rate": 7.222199360966842e-07, "loss": 0.5208, "step": 16379 }, { "epoch": 0.8285699833071981, "grad_norm": 11.294428825378418, "learning_rate": 7.218044695895826e-07, "loss": 0.4595, "step": 16380 }, { "epoch": 0.8286205675552633, "grad_norm": 9.025144577026367, "learning_rate": 7.213891133212475e-07, "loss": 0.505, "step": 16381 }, { "epoch": 0.8286711518033284, "grad_norm": 6.871545314788818, "learning_rate": 7.209738673023803e-07, "loss": 0.3631, "step": 16382 }, { "epoch": 0.8287217360513935, "grad_norm": 11.190837860107422, "learning_rate": 7.205587315436802e-07, "loss": 0.4677, "step": 16383 }, { "epoch": 0.8287723202994588, "grad_norm": 6.745511054992676, "learning_rate": 7.201437060558459e-07, "loss": 0.3387, "step": 16384 }, { "epoch": 0.8288229045475239, "grad_norm": 9.56396770477295, "learning_rate": 7.197287908495709e-07, "loss": 0.3486, "step": 16385 }, { "epoch": 0.8288734887955891, "grad_norm": 17.60170555114746, "learning_rate": 7.193139859355452e-07, "loss": 0.5882, "step": 16386 }, { "epoch": 0.8289240730436542, "grad_norm": 7.1867570877075195, "learning_rate": 7.188992913244597e-07, "loss": 0.3186, "step": 16387 }, { "epoch": 0.8289746572917194, "grad_norm": 10.690195083618164, "learning_rate": 7.184847070269984e-07, "loss": 0.4683, "step": 16388 }, { "epoch": 0.8290252415397845, "grad_norm": 11.112382888793945, "learning_rate": 7.180702330538442e-07, "loss": 0.4768, "step": 16389 }, { "epoch": 0.8290758257878497, "grad_norm": 5.659627437591553, "learning_rate": 7.176558694156782e-07, "loss": 0.3243, "step": 16390 }, { "epoch": 0.8291264100359148, "grad_norm": 5.6680073738098145, "learning_rate": 7.172416161231771e-07, "loss": 0.4377, "step": 16391 }, { "epoch": 0.82917699428398, "grad_norm": 8.245325088500977, "learning_rate": 7.168274731870145e-07, "loss": 0.3339, "step": 16392 }, { "epoch": 0.8292275785320451, "grad_norm": 7.3215508460998535, "learning_rate": 7.16413440617863e-07, "loss": 0.4426, "step": 16393 }, { "epoch": 0.8292781627801102, "grad_norm": 7.358370304107666, "learning_rate": 7.159995184263902e-07, "loss": 0.3123, "step": 16394 }, { "epoch": 0.8293287470281754, "grad_norm": 8.327889442443848, "learning_rate": 7.155857066232607e-07, "loss": 0.4545, "step": 16395 }, { "epoch": 0.8293793312762405, "grad_norm": 10.624395370483398, "learning_rate": 7.151720052191408e-07, "loss": 0.4082, "step": 16396 }, { "epoch": 0.8294299155243058, "grad_norm": 5.424302101135254, "learning_rate": 7.147584142246889e-07, "loss": 0.3416, "step": 16397 }, { "epoch": 0.8294804997723709, "grad_norm": 7.636286735534668, "learning_rate": 7.143449336505603e-07, "loss": 0.4583, "step": 16398 }, { "epoch": 0.8295310840204361, "grad_norm": 12.09653091430664, "learning_rate": 7.139315635074123e-07, "loss": 0.5048, "step": 16399 }, { "epoch": 0.8295816682685012, "grad_norm": 9.556876182556152, "learning_rate": 7.135183038058957e-07, "loss": 0.3813, "step": 16400 }, { "epoch": 0.8296322525165664, "grad_norm": 8.59898853302002, "learning_rate": 7.131051545566575e-07, "loss": 0.4166, "step": 16401 }, { "epoch": 0.8296828367646315, "grad_norm": 7.456920146942139, "learning_rate": 7.126921157703465e-07, "loss": 0.3909, "step": 16402 }, { "epoch": 0.8297334210126966, "grad_norm": 9.276412963867188, "learning_rate": 7.122791874576034e-07, "loss": 0.4185, "step": 16403 }, { "epoch": 0.8297840052607618, "grad_norm": 11.581838607788086, "learning_rate": 7.118663696290684e-07, "loss": 0.4535, "step": 16404 }, { "epoch": 0.8298345895088269, "grad_norm": 11.432698249816895, "learning_rate": 7.114536622953816e-07, "loss": 0.3489, "step": 16405 }, { "epoch": 0.8298851737568921, "grad_norm": 10.739239692687988, "learning_rate": 7.110410654671729e-07, "loss": 0.4171, "step": 16406 }, { "epoch": 0.8299357580049572, "grad_norm": 8.252930641174316, "learning_rate": 7.106285791550782e-07, "loss": 0.4157, "step": 16407 }, { "epoch": 0.8299863422530224, "grad_norm": 7.55070161819458, "learning_rate": 7.102162033697241e-07, "loss": 0.351, "step": 16408 }, { "epoch": 0.8300369265010875, "grad_norm": 5.625178337097168, "learning_rate": 7.098039381217353e-07, "loss": 0.24, "step": 16409 }, { "epoch": 0.8300875107491528, "grad_norm": 7.145825386047363, "learning_rate": 7.093917834217379e-07, "loss": 0.3119, "step": 16410 }, { "epoch": 0.8301380949972179, "grad_norm": 5.962277412414551, "learning_rate": 7.089797392803511e-07, "loss": 0.3781, "step": 16411 }, { "epoch": 0.8301886792452831, "grad_norm": 5.822601318359375, "learning_rate": 7.085678057081913e-07, "loss": 0.3181, "step": 16412 }, { "epoch": 0.8302392634933482, "grad_norm": 9.831860542297363, "learning_rate": 7.081559827158729e-07, "loss": 0.4476, "step": 16413 }, { "epoch": 0.8302898477414133, "grad_norm": 7.586479663848877, "learning_rate": 7.077442703140092e-07, "loss": 0.4016, "step": 16414 }, { "epoch": 0.8303404319894785, "grad_norm": 6.107140064239502, "learning_rate": 7.073326685132082e-07, "loss": 0.3675, "step": 16415 }, { "epoch": 0.8303910162375436, "grad_norm": 9.574504852294922, "learning_rate": 7.069211773240747e-07, "loss": 0.3584, "step": 16416 }, { "epoch": 0.8304416004856088, "grad_norm": 6.434962749481201, "learning_rate": 7.065097967572143e-07, "loss": 0.4418, "step": 16417 }, { "epoch": 0.8304921847336739, "grad_norm": 6.490776062011719, "learning_rate": 7.060985268232257e-07, "loss": 0.2618, "step": 16418 }, { "epoch": 0.8305427689817391, "grad_norm": 19.368711471557617, "learning_rate": 7.056873675327064e-07, "loss": 0.4444, "step": 16419 }, { "epoch": 0.8305933532298042, "grad_norm": 5.732474327087402, "learning_rate": 7.052763188962514e-07, "loss": 0.3456, "step": 16420 }, { "epoch": 0.8306439374778694, "grad_norm": 11.720614433288574, "learning_rate": 7.04865380924451e-07, "loss": 0.3914, "step": 16421 }, { "epoch": 0.8306945217259345, "grad_norm": 9.881306648254395, "learning_rate": 7.044545536278963e-07, "loss": 0.4188, "step": 16422 }, { "epoch": 0.8307451059739996, "grad_norm": 9.340938568115234, "learning_rate": 7.040438370171726e-07, "loss": 0.3432, "step": 16423 }, { "epoch": 0.8307956902220649, "grad_norm": 6.486427307128906, "learning_rate": 7.036332311028615e-07, "loss": 0.359, "step": 16424 }, { "epoch": 0.83084627447013, "grad_norm": 6.9353413581848145, "learning_rate": 7.032227358955452e-07, "loss": 0.4331, "step": 16425 }, { "epoch": 0.8308968587181952, "grad_norm": 8.060766220092773, "learning_rate": 7.028123514058011e-07, "loss": 0.4154, "step": 16426 }, { "epoch": 0.8309474429662603, "grad_norm": 18.156307220458984, "learning_rate": 7.02402077644202e-07, "loss": 0.5154, "step": 16427 }, { "epoch": 0.8309980272143255, "grad_norm": 10.417884826660156, "learning_rate": 7.019919146213222e-07, "loss": 0.4288, "step": 16428 }, { "epoch": 0.8310486114623906, "grad_norm": 5.329254627227783, "learning_rate": 7.015818623477294e-07, "loss": 0.3932, "step": 16429 }, { "epoch": 0.8310991957104558, "grad_norm": 10.315726280212402, "learning_rate": 7.01171920833989e-07, "loss": 0.3559, "step": 16430 }, { "epoch": 0.8311497799585209, "grad_norm": 7.311827659606934, "learning_rate": 7.007620900906653e-07, "loss": 0.3665, "step": 16431 }, { "epoch": 0.831200364206586, "grad_norm": 8.373505592346191, "learning_rate": 7.003523701283182e-07, "loss": 0.3606, "step": 16432 }, { "epoch": 0.8312509484546512, "grad_norm": 7.904447555541992, "learning_rate": 6.999427609575038e-07, "loss": 0.3651, "step": 16433 }, { "epoch": 0.8313015327027163, "grad_norm": 10.55013656616211, "learning_rate": 6.995332625887796e-07, "loss": 0.5367, "step": 16434 }, { "epoch": 0.8313521169507815, "grad_norm": 9.791796684265137, "learning_rate": 6.991238750326956e-07, "loss": 0.4642, "step": 16435 }, { "epoch": 0.8314027011988466, "grad_norm": 8.27023983001709, "learning_rate": 6.987145982998e-07, "loss": 0.4353, "step": 16436 }, { "epoch": 0.8314532854469119, "grad_norm": 5.90798807144165, "learning_rate": 6.983054324006405e-07, "loss": 0.3221, "step": 16437 }, { "epoch": 0.831503869694977, "grad_norm": 8.11539077758789, "learning_rate": 6.978963773457603e-07, "loss": 0.4536, "step": 16438 }, { "epoch": 0.8315544539430422, "grad_norm": 9.926497459411621, "learning_rate": 6.974874331456976e-07, "loss": 0.3886, "step": 16439 }, { "epoch": 0.8316050381911073, "grad_norm": 4.684298992156982, "learning_rate": 6.97078599810993e-07, "loss": 0.4125, "step": 16440 }, { "epoch": 0.8316556224391725, "grad_norm": 6.510919570922852, "learning_rate": 6.966698773521791e-07, "loss": 0.3305, "step": 16441 }, { "epoch": 0.8317062066872376, "grad_norm": 7.210710525512695, "learning_rate": 6.962612657797873e-07, "loss": 0.4263, "step": 16442 }, { "epoch": 0.8317567909353027, "grad_norm": 10.601472854614258, "learning_rate": 6.958527651043501e-07, "loss": 0.4548, "step": 16443 }, { "epoch": 0.8318073751833679, "grad_norm": 6.978458881378174, "learning_rate": 6.954443753363882e-07, "loss": 0.2971, "step": 16444 }, { "epoch": 0.831857959431433, "grad_norm": 8.487725257873535, "learning_rate": 6.950360964864289e-07, "loss": 0.4101, "step": 16445 }, { "epoch": 0.8319085436794982, "grad_norm": 9.264334678649902, "learning_rate": 6.94627928564991e-07, "loss": 0.3596, "step": 16446 }, { "epoch": 0.8319591279275633, "grad_norm": 5.81751823425293, "learning_rate": 6.942198715825921e-07, "loss": 0.4046, "step": 16447 }, { "epoch": 0.8320097121756285, "grad_norm": 5.388169765472412, "learning_rate": 6.938119255497455e-07, "loss": 0.4702, "step": 16448 }, { "epoch": 0.8320602964236936, "grad_norm": 6.172176837921143, "learning_rate": 6.934040904769662e-07, "loss": 0.4105, "step": 16449 }, { "epoch": 0.8321108806717589, "grad_norm": 5.811190128326416, "learning_rate": 6.929963663747608e-07, "loss": 0.338, "step": 16450 }, { "epoch": 0.832161464919824, "grad_norm": 6.292025089263916, "learning_rate": 6.925887532536351e-07, "loss": 0.4017, "step": 16451 }, { "epoch": 0.8322120491678892, "grad_norm": 6.945101261138916, "learning_rate": 6.921812511240944e-07, "loss": 0.4162, "step": 16452 }, { "epoch": 0.8322626334159543, "grad_norm": 9.578787803649902, "learning_rate": 6.917738599966373e-07, "loss": 0.4679, "step": 16453 }, { "epoch": 0.8323132176640194, "grad_norm": 11.609682083129883, "learning_rate": 6.913665798817609e-07, "loss": 0.4615, "step": 16454 }, { "epoch": 0.8323638019120846, "grad_norm": 6.4200592041015625, "learning_rate": 6.909594107899614e-07, "loss": 0.3523, "step": 16455 }, { "epoch": 0.8324143861601497, "grad_norm": 17.1112117767334, "learning_rate": 6.905523527317303e-07, "loss": 0.6376, "step": 16456 }, { "epoch": 0.8324649704082149, "grad_norm": 5.381072998046875, "learning_rate": 6.901454057175555e-07, "loss": 0.3237, "step": 16457 }, { "epoch": 0.83251555465628, "grad_norm": 5.545557022094727, "learning_rate": 6.89738569757924e-07, "loss": 0.3385, "step": 16458 }, { "epoch": 0.8325661389043452, "grad_norm": 7.159394264221191, "learning_rate": 6.893318448633174e-07, "loss": 0.3946, "step": 16459 }, { "epoch": 0.8326167231524103, "grad_norm": 9.893362045288086, "learning_rate": 6.889252310442185e-07, "loss": 0.3657, "step": 16460 }, { "epoch": 0.8326673074004755, "grad_norm": 5.020994663238525, "learning_rate": 6.885187283111028e-07, "loss": 0.3905, "step": 16461 }, { "epoch": 0.8327178916485406, "grad_norm": 7.330060958862305, "learning_rate": 6.881123366744447e-07, "loss": 0.3903, "step": 16462 }, { "epoch": 0.8327684758966057, "grad_norm": 9.429610252380371, "learning_rate": 6.87706056144718e-07, "loss": 0.3378, "step": 16463 }, { "epoch": 0.832819060144671, "grad_norm": 10.23617172241211, "learning_rate": 6.872998867323904e-07, "loss": 0.3919, "step": 16464 }, { "epoch": 0.8328696443927361, "grad_norm": 7.905243873596191, "learning_rate": 6.868938284479265e-07, "loss": 0.4098, "step": 16465 }, { "epoch": 0.8329202286408013, "grad_norm": 4.371443271636963, "learning_rate": 6.864878813017917e-07, "loss": 0.3179, "step": 16466 }, { "epoch": 0.8329708128888664, "grad_norm": 6.455629825592041, "learning_rate": 6.860820453044453e-07, "loss": 0.3684, "step": 16467 }, { "epoch": 0.8330213971369316, "grad_norm": 8.250728607177734, "learning_rate": 6.856763204663442e-07, "loss": 0.3486, "step": 16468 }, { "epoch": 0.8330719813849967, "grad_norm": 5.103704929351807, "learning_rate": 6.852707067979452e-07, "loss": 0.3871, "step": 16469 }, { "epoch": 0.8331225656330619, "grad_norm": 10.835155487060547, "learning_rate": 6.848652043096976e-07, "loss": 0.4675, "step": 16470 }, { "epoch": 0.833173149881127, "grad_norm": 4.875870227813721, "learning_rate": 6.844598130120495e-07, "loss": 0.3153, "step": 16471 }, { "epoch": 0.8332237341291922, "grad_norm": 5.170319557189941, "learning_rate": 6.840545329154491e-07, "loss": 0.2699, "step": 16472 }, { "epoch": 0.8332743183772573, "grad_norm": 8.353842735290527, "learning_rate": 6.836493640303394e-07, "loss": 0.343, "step": 16473 }, { "epoch": 0.8333249026253224, "grad_norm": 6.352659225463867, "learning_rate": 6.832443063671578e-07, "loss": 0.3856, "step": 16474 }, { "epoch": 0.8333754868733876, "grad_norm": 12.198493003845215, "learning_rate": 6.828393599363453e-07, "loss": 0.5002, "step": 16475 }, { "epoch": 0.8334260711214527, "grad_norm": 6.674702167510986, "learning_rate": 6.82434524748335e-07, "loss": 0.3716, "step": 16476 }, { "epoch": 0.833476655369518, "grad_norm": 8.493317604064941, "learning_rate": 6.820298008135573e-07, "loss": 0.4553, "step": 16477 }, { "epoch": 0.8335272396175831, "grad_norm": 6.465143203735352, "learning_rate": 6.81625188142443e-07, "loss": 0.3416, "step": 16478 }, { "epoch": 0.8335778238656483, "grad_norm": 17.4786376953125, "learning_rate": 6.812206867454168e-07, "loss": 0.5107, "step": 16479 }, { "epoch": 0.8336284081137134, "grad_norm": 9.88322639465332, "learning_rate": 6.808162966329007e-07, "loss": 0.3933, "step": 16480 }, { "epoch": 0.8336789923617786, "grad_norm": 5.602303504943848, "learning_rate": 6.804120178153179e-07, "loss": 0.3844, "step": 16481 }, { "epoch": 0.8337295766098437, "grad_norm": 19.79831314086914, "learning_rate": 6.800078503030833e-07, "loss": 0.582, "step": 16482 }, { "epoch": 0.8337801608579088, "grad_norm": 7.944885730743408, "learning_rate": 6.796037941066119e-07, "loss": 0.3719, "step": 16483 }, { "epoch": 0.833830745105974, "grad_norm": 6.942196369171143, "learning_rate": 6.79199849236315e-07, "loss": 0.3586, "step": 16484 }, { "epoch": 0.8338813293540391, "grad_norm": 6.97566032409668, "learning_rate": 6.787960157026019e-07, "loss": 0.3489, "step": 16485 }, { "epoch": 0.8339319136021043, "grad_norm": 7.0329909324646, "learning_rate": 6.783922935158771e-07, "loss": 0.4169, "step": 16486 }, { "epoch": 0.8339824978501694, "grad_norm": 10.03467845916748, "learning_rate": 6.77988682686545e-07, "loss": 0.3325, "step": 16487 }, { "epoch": 0.8340330820982346, "grad_norm": 10.68111515045166, "learning_rate": 6.775851832250057e-07, "loss": 0.4776, "step": 16488 }, { "epoch": 0.8340836663462997, "grad_norm": 8.100214004516602, "learning_rate": 6.771817951416543e-07, "loss": 0.3513, "step": 16489 }, { "epoch": 0.834134250594365, "grad_norm": 9.083365440368652, "learning_rate": 6.767785184468884e-07, "loss": 0.4013, "step": 16490 }, { "epoch": 0.8341848348424301, "grad_norm": 9.629941940307617, "learning_rate": 6.763753531510975e-07, "loss": 0.4193, "step": 16491 }, { "epoch": 0.8342354190904953, "grad_norm": 8.691530227661133, "learning_rate": 6.759722992646689e-07, "loss": 0.3564, "step": 16492 }, { "epoch": 0.8342860033385604, "grad_norm": 7.488429069519043, "learning_rate": 6.755693567979915e-07, "loss": 0.4399, "step": 16493 }, { "epoch": 0.8343365875866255, "grad_norm": 5.807605743408203, "learning_rate": 6.751665257614464e-07, "loss": 0.3269, "step": 16494 }, { "epoch": 0.8343871718346907, "grad_norm": 24.285818099975586, "learning_rate": 6.747638061654139e-07, "loss": 0.4517, "step": 16495 }, { "epoch": 0.8344377560827558, "grad_norm": 7.412105083465576, "learning_rate": 6.743611980202708e-07, "loss": 0.3563, "step": 16496 }, { "epoch": 0.834488340330821, "grad_norm": 5.525356769561768, "learning_rate": 6.739587013363902e-07, "loss": 0.2364, "step": 16497 }, { "epoch": 0.8345389245788861, "grad_norm": 11.223423957824707, "learning_rate": 6.73556316124146e-07, "loss": 0.3697, "step": 16498 }, { "epoch": 0.8345895088269513, "grad_norm": 9.011306762695312, "learning_rate": 6.731540423939054e-07, "loss": 0.4534, "step": 16499 }, { "epoch": 0.8346400930750164, "grad_norm": 9.856942176818848, "learning_rate": 6.727518801560329e-07, "loss": 0.3593, "step": 16500 }, { "epoch": 0.8346906773230816, "grad_norm": 6.871921539306641, "learning_rate": 6.723498294208936e-07, "loss": 0.4029, "step": 16501 }, { "epoch": 0.8347412615711467, "grad_norm": 7.978143692016602, "learning_rate": 6.719478901988463e-07, "loss": 0.3792, "step": 16502 }, { "epoch": 0.834791845819212, "grad_norm": 15.990482330322266, "learning_rate": 6.715460625002462e-07, "loss": 0.4672, "step": 16503 }, { "epoch": 0.8348424300672771, "grad_norm": 8.550798416137695, "learning_rate": 6.711443463354511e-07, "loss": 0.387, "step": 16504 }, { "epoch": 0.8348930143153422, "grad_norm": 10.532243728637695, "learning_rate": 6.707427417148099e-07, "loss": 0.4961, "step": 16505 }, { "epoch": 0.8349435985634074, "grad_norm": 9.062355995178223, "learning_rate": 6.703412486486699e-07, "loss": 0.3812, "step": 16506 }, { "epoch": 0.8349941828114725, "grad_norm": 7.881399631500244, "learning_rate": 6.699398671473794e-07, "loss": 0.3789, "step": 16507 }, { "epoch": 0.8350447670595377, "grad_norm": 10.149279594421387, "learning_rate": 6.695385972212809e-07, "loss": 0.4399, "step": 16508 }, { "epoch": 0.8350953513076028, "grad_norm": 6.816515922546387, "learning_rate": 6.691374388807109e-07, "loss": 0.4723, "step": 16509 }, { "epoch": 0.835145935555668, "grad_norm": 6.388936519622803, "learning_rate": 6.687363921360091e-07, "loss": 0.3501, "step": 16510 }, { "epoch": 0.8351965198037331, "grad_norm": 9.192630767822266, "learning_rate": 6.683354569975087e-07, "loss": 0.4348, "step": 16511 }, { "epoch": 0.8352471040517983, "grad_norm": 7.626812934875488, "learning_rate": 6.679346334755399e-07, "loss": 0.4205, "step": 16512 }, { "epoch": 0.8352976882998634, "grad_norm": 7.633421897888184, "learning_rate": 6.675339215804333e-07, "loss": 0.3732, "step": 16513 }, { "epoch": 0.8353482725479285, "grad_norm": 19.622426986694336, "learning_rate": 6.671333213225129e-07, "loss": 0.4076, "step": 16514 }, { "epoch": 0.8353988567959937, "grad_norm": 5.319758892059326, "learning_rate": 6.667328327120997e-07, "loss": 0.3825, "step": 16515 }, { "epoch": 0.8354494410440588, "grad_norm": 4.948048114776611, "learning_rate": 6.663324557595158e-07, "loss": 0.3852, "step": 16516 }, { "epoch": 0.8355000252921241, "grad_norm": 9.158380508422852, "learning_rate": 6.659321904750776e-07, "loss": 0.3505, "step": 16517 }, { "epoch": 0.8355506095401892, "grad_norm": 6.321831703186035, "learning_rate": 6.655320368690965e-07, "loss": 0.2835, "step": 16518 }, { "epoch": 0.8356011937882544, "grad_norm": 17.018938064575195, "learning_rate": 6.65131994951887e-07, "loss": 0.501, "step": 16519 }, { "epoch": 0.8356517780363195, "grad_norm": 5.366611957550049, "learning_rate": 6.647320647337552e-07, "loss": 0.3714, "step": 16520 }, { "epoch": 0.8357023622843847, "grad_norm": 7.745707988739014, "learning_rate": 6.643322462250068e-07, "loss": 0.2998, "step": 16521 }, { "epoch": 0.8357529465324498, "grad_norm": 7.382568836212158, "learning_rate": 6.639325394359441e-07, "loss": 0.3606, "step": 16522 }, { "epoch": 0.835803530780515, "grad_norm": 7.059650421142578, "learning_rate": 6.635329443768668e-07, "loss": 0.3052, "step": 16523 }, { "epoch": 0.8358541150285801, "grad_norm": 8.990381240844727, "learning_rate": 6.631334610580697e-07, "loss": 0.4979, "step": 16524 }, { "epoch": 0.8359046992766452, "grad_norm": 6.026479244232178, "learning_rate": 6.627340894898493e-07, "loss": 0.3885, "step": 16525 }, { "epoch": 0.8359552835247104, "grad_norm": 6.966495513916016, "learning_rate": 6.623348296824949e-07, "loss": 0.4814, "step": 16526 }, { "epoch": 0.8360058677727755, "grad_norm": 9.122772216796875, "learning_rate": 6.619356816462941e-07, "loss": 0.4409, "step": 16527 }, { "epoch": 0.8360564520208407, "grad_norm": 5.990296363830566, "learning_rate": 6.615366453915333e-07, "loss": 0.3742, "step": 16528 }, { "epoch": 0.8361070362689058, "grad_norm": 6.457449913024902, "learning_rate": 6.611377209284942e-07, "loss": 0.3927, "step": 16529 }, { "epoch": 0.8361576205169711, "grad_norm": 5.66778039932251, "learning_rate": 6.607389082674548e-07, "loss": 0.358, "step": 16530 }, { "epoch": 0.8362082047650362, "grad_norm": 17.060461044311523, "learning_rate": 6.603402074186938e-07, "loss": 0.4621, "step": 16531 }, { "epoch": 0.8362587890131014, "grad_norm": 8.441743850708008, "learning_rate": 6.59941618392484e-07, "loss": 0.3637, "step": 16532 }, { "epoch": 0.8363093732611665, "grad_norm": 11.173710823059082, "learning_rate": 6.595431411990949e-07, "loss": 0.3159, "step": 16533 }, { "epoch": 0.8363599575092316, "grad_norm": 8.119162559509277, "learning_rate": 6.591447758487957e-07, "loss": 0.3709, "step": 16534 }, { "epoch": 0.8364105417572968, "grad_norm": 6.963608741760254, "learning_rate": 6.587465223518497e-07, "loss": 0.3441, "step": 16535 }, { "epoch": 0.8364611260053619, "grad_norm": 9.157477378845215, "learning_rate": 6.58348380718521e-07, "loss": 0.3214, "step": 16536 }, { "epoch": 0.8365117102534271, "grad_norm": 8.250589370727539, "learning_rate": 6.579503509590673e-07, "loss": 0.357, "step": 16537 }, { "epoch": 0.8365622945014922, "grad_norm": 10.810259819030762, "learning_rate": 6.575524330837447e-07, "loss": 0.4478, "step": 16538 }, { "epoch": 0.8366128787495574, "grad_norm": 10.47446060180664, "learning_rate": 6.571546271028079e-07, "loss": 0.5189, "step": 16539 }, { "epoch": 0.8366634629976225, "grad_norm": 7.792666435241699, "learning_rate": 6.567569330265067e-07, "loss": 0.3509, "step": 16540 }, { "epoch": 0.8367140472456877, "grad_norm": 8.910104751586914, "learning_rate": 6.563593508650872e-07, "loss": 0.4205, "step": 16541 }, { "epoch": 0.8367646314937528, "grad_norm": 22.580032348632812, "learning_rate": 6.559618806287971e-07, "loss": 0.4286, "step": 16542 }, { "epoch": 0.836815215741818, "grad_norm": 11.582714080810547, "learning_rate": 6.555645223278767e-07, "loss": 0.5214, "step": 16543 }, { "epoch": 0.8368657999898832, "grad_norm": 9.504598617553711, "learning_rate": 6.551672759725642e-07, "loss": 0.3768, "step": 16544 }, { "epoch": 0.8369163842379483, "grad_norm": 7.643344879150391, "learning_rate": 6.547701415730973e-07, "loss": 0.3521, "step": 16545 }, { "epoch": 0.8369669684860135, "grad_norm": 15.093214988708496, "learning_rate": 6.54373119139709e-07, "loss": 0.3569, "step": 16546 }, { "epoch": 0.8370175527340786, "grad_norm": 12.489897727966309, "learning_rate": 6.539762086826273e-07, "loss": 0.4228, "step": 16547 }, { "epoch": 0.8370681369821438, "grad_norm": 10.582235336303711, "learning_rate": 6.535794102120824e-07, "loss": 0.3529, "step": 16548 }, { "epoch": 0.8371187212302089, "grad_norm": 6.648031234741211, "learning_rate": 6.531827237382976e-07, "loss": 0.3741, "step": 16549 }, { "epoch": 0.8371693054782741, "grad_norm": 12.34648323059082, "learning_rate": 6.527861492714937e-07, "loss": 0.5979, "step": 16550 }, { "epoch": 0.8372198897263392, "grad_norm": 10.940425872802734, "learning_rate": 6.523896868218921e-07, "loss": 0.4556, "step": 16551 }, { "epoch": 0.8372704739744044, "grad_norm": 6.718936920166016, "learning_rate": 6.519933363997061e-07, "loss": 0.4289, "step": 16552 }, { "epoch": 0.8373210582224695, "grad_norm": 11.590333938598633, "learning_rate": 6.515970980151493e-07, "loss": 0.4802, "step": 16553 }, { "epoch": 0.8373716424705346, "grad_norm": 6.2194037437438965, "learning_rate": 6.512009716784329e-07, "loss": 0.3956, "step": 16554 }, { "epoch": 0.8374222267185998, "grad_norm": 9.096431732177734, "learning_rate": 6.508049573997632e-07, "loss": 0.3485, "step": 16555 }, { "epoch": 0.8374728109666649, "grad_norm": 8.40640926361084, "learning_rate": 6.504090551893438e-07, "loss": 0.3848, "step": 16556 }, { "epoch": 0.8375233952147302, "grad_norm": 8.96773624420166, "learning_rate": 6.500132650573782e-07, "loss": 0.4383, "step": 16557 }, { "epoch": 0.8375739794627953, "grad_norm": 11.336982727050781, "learning_rate": 6.496175870140637e-07, "loss": 0.3708, "step": 16558 }, { "epoch": 0.8376245637108605, "grad_norm": 7.709820747375488, "learning_rate": 6.492220210695959e-07, "loss": 0.3876, "step": 16559 }, { "epoch": 0.8376751479589256, "grad_norm": 6.565842151641846, "learning_rate": 6.488265672341681e-07, "loss": 0.2856, "step": 16560 }, { "epoch": 0.8377257322069908, "grad_norm": 9.46316146850586, "learning_rate": 6.484312255179697e-07, "loss": 0.4311, "step": 16561 }, { "epoch": 0.8377763164550559, "grad_norm": 11.182144165039062, "learning_rate": 6.480359959311866e-07, "loss": 0.4045, "step": 16562 }, { "epoch": 0.837826900703121, "grad_norm": 13.755406379699707, "learning_rate": 6.476408784840054e-07, "loss": 0.554, "step": 16563 }, { "epoch": 0.8378774849511862, "grad_norm": 12.545022964477539, "learning_rate": 6.47245873186606e-07, "loss": 0.6106, "step": 16564 }, { "epoch": 0.8379280691992513, "grad_norm": 13.882065773010254, "learning_rate": 6.468509800491657e-07, "loss": 0.344, "step": 16565 }, { "epoch": 0.8379786534473165, "grad_norm": 23.90528106689453, "learning_rate": 6.464561990818625e-07, "loss": 0.4628, "step": 16566 }, { "epoch": 0.8380292376953816, "grad_norm": 10.14541244506836, "learning_rate": 6.460615302948669e-07, "loss": 0.4594, "step": 16567 }, { "epoch": 0.8380798219434468, "grad_norm": 10.29536247253418, "learning_rate": 6.456669736983485e-07, "loss": 0.4295, "step": 16568 }, { "epoch": 0.8381304061915119, "grad_norm": 11.548152923583984, "learning_rate": 6.452725293024753e-07, "loss": 0.4716, "step": 16569 }, { "epoch": 0.8381809904395772, "grad_norm": 14.779340744018555, "learning_rate": 6.448781971174112e-07, "loss": 0.4903, "step": 16570 }, { "epoch": 0.8382315746876423, "grad_norm": 9.028205871582031, "learning_rate": 6.444839771533151e-07, "loss": 0.4721, "step": 16571 }, { "epoch": 0.8382821589357075, "grad_norm": 5.877434730529785, "learning_rate": 6.44089869420349e-07, "loss": 0.3108, "step": 16572 }, { "epoch": 0.8383327431837726, "grad_norm": 5.977197647094727, "learning_rate": 6.436958739286637e-07, "loss": 0.3129, "step": 16573 }, { "epoch": 0.8383833274318377, "grad_norm": 8.759725570678711, "learning_rate": 6.433019906884141e-07, "loss": 0.4379, "step": 16574 }, { "epoch": 0.8384339116799029, "grad_norm": 5.719062328338623, "learning_rate": 6.429082197097491e-07, "loss": 0.314, "step": 16575 }, { "epoch": 0.838484495927968, "grad_norm": 10.576448440551758, "learning_rate": 6.425145610028143e-07, "loss": 0.4899, "step": 16576 }, { "epoch": 0.8385350801760332, "grad_norm": 9.882339477539062, "learning_rate": 6.421210145777551e-07, "loss": 0.4179, "step": 16577 }, { "epoch": 0.8385856644240983, "grad_norm": 14.193471908569336, "learning_rate": 6.417275804447115e-07, "loss": 0.5796, "step": 16578 }, { "epoch": 0.8386362486721635, "grad_norm": 5.831909656524658, "learning_rate": 6.4133425861382e-07, "loss": 0.4131, "step": 16579 }, { "epoch": 0.8386868329202286, "grad_norm": 7.263252258300781, "learning_rate": 6.409410490952178e-07, "loss": 0.3589, "step": 16580 }, { "epoch": 0.8387374171682938, "grad_norm": 5.531023979187012, "learning_rate": 6.405479518990359e-07, "loss": 0.3223, "step": 16581 }, { "epoch": 0.8387880014163589, "grad_norm": 6.078324317932129, "learning_rate": 6.401549670354035e-07, "loss": 0.3411, "step": 16582 }, { "epoch": 0.8388385856644242, "grad_norm": 23.44972038269043, "learning_rate": 6.397620945144456e-07, "loss": 0.4987, "step": 16583 }, { "epoch": 0.8388891699124893, "grad_norm": 7.842061996459961, "learning_rate": 6.393693343462881e-07, "loss": 0.4936, "step": 16584 }, { "epoch": 0.8389397541605544, "grad_norm": 11.49801254272461, "learning_rate": 6.389766865410502e-07, "loss": 0.4346, "step": 16585 }, { "epoch": 0.8389903384086196, "grad_norm": 8.095863342285156, "learning_rate": 6.385841511088491e-07, "loss": 0.4702, "step": 16586 }, { "epoch": 0.8390409226566847, "grad_norm": 7.508753776550293, "learning_rate": 6.381917280598005e-07, "loss": 0.4466, "step": 16587 }, { "epoch": 0.8390915069047499, "grad_norm": 11.132139205932617, "learning_rate": 6.377994174040142e-07, "loss": 0.4092, "step": 16588 }, { "epoch": 0.839142091152815, "grad_norm": 6.782603740692139, "learning_rate": 6.374072191516012e-07, "loss": 0.4283, "step": 16589 }, { "epoch": 0.8391926754008802, "grad_norm": 8.348343849182129, "learning_rate": 6.370151333126667e-07, "loss": 0.4379, "step": 16590 }, { "epoch": 0.8392432596489453, "grad_norm": 7.16678524017334, "learning_rate": 6.366231598973133e-07, "loss": 0.3476, "step": 16591 }, { "epoch": 0.8392938438970104, "grad_norm": 5.457369327545166, "learning_rate": 6.362312989156427e-07, "loss": 0.2931, "step": 16592 }, { "epoch": 0.8393444281450756, "grad_norm": 8.508040428161621, "learning_rate": 6.358395503777509e-07, "loss": 0.4146, "step": 16593 }, { "epoch": 0.8393950123931407, "grad_norm": 5.282055854797363, "learning_rate": 6.35447914293732e-07, "loss": 0.3326, "step": 16594 }, { "epoch": 0.8394455966412059, "grad_norm": 6.978943824768066, "learning_rate": 6.350563906736789e-07, "loss": 0.4081, "step": 16595 }, { "epoch": 0.839496180889271, "grad_norm": 8.545663833618164, "learning_rate": 6.346649795276799e-07, "loss": 0.3485, "step": 16596 }, { "epoch": 0.8395467651373363, "grad_norm": 15.467942237854004, "learning_rate": 6.34273680865819e-07, "loss": 0.4906, "step": 16597 }, { "epoch": 0.8395973493854014, "grad_norm": 9.785850524902344, "learning_rate": 6.338824946981831e-07, "loss": 0.3511, "step": 16598 }, { "epoch": 0.8396479336334666, "grad_norm": 7.043153762817383, "learning_rate": 6.334914210348475e-07, "loss": 0.3907, "step": 16599 }, { "epoch": 0.8396985178815317, "grad_norm": 6.929556369781494, "learning_rate": 6.331004598858903e-07, "loss": 0.3337, "step": 16600 }, { "epoch": 0.8397491021295969, "grad_norm": 11.223142623901367, "learning_rate": 6.327096112613878e-07, "loss": 0.3085, "step": 16601 }, { "epoch": 0.839799686377662, "grad_norm": 5.990708827972412, "learning_rate": 6.323188751714099e-07, "loss": 0.3801, "step": 16602 }, { "epoch": 0.8398502706257271, "grad_norm": 10.464151382446289, "learning_rate": 6.319282516260233e-07, "loss": 0.4873, "step": 16603 }, { "epoch": 0.8399008548737923, "grad_norm": 8.57839584350586, "learning_rate": 6.315377406352963e-07, "loss": 0.3781, "step": 16604 }, { "epoch": 0.8399514391218574, "grad_norm": 5.152527809143066, "learning_rate": 6.311473422092906e-07, "loss": 0.3508, "step": 16605 }, { "epoch": 0.8400020233699226, "grad_norm": 7.155653476715088, "learning_rate": 6.307570563580639e-07, "loss": 0.3874, "step": 16606 }, { "epoch": 0.8400526076179877, "grad_norm": 8.638401985168457, "learning_rate": 6.303668830916754e-07, "loss": 0.4944, "step": 16607 }, { "epoch": 0.8401031918660529, "grad_norm": 5.731385231018066, "learning_rate": 6.29976822420178e-07, "loss": 0.3048, "step": 16608 }, { "epoch": 0.840153776114118, "grad_norm": 7.895875930786133, "learning_rate": 6.29586874353621e-07, "loss": 0.4658, "step": 16609 }, { "epoch": 0.8402043603621833, "grad_norm": 15.601516723632812, "learning_rate": 6.291970389020558e-07, "loss": 0.4157, "step": 16610 }, { "epoch": 0.8402549446102484, "grad_norm": 7.792543411254883, "learning_rate": 6.288073160755254e-07, "loss": 0.3515, "step": 16611 }, { "epoch": 0.8403055288583136, "grad_norm": 10.757315635681152, "learning_rate": 6.284177058840718e-07, "loss": 0.4433, "step": 16612 }, { "epoch": 0.8403561131063787, "grad_norm": 8.193078994750977, "learning_rate": 6.280282083377353e-07, "loss": 0.4413, "step": 16613 }, { "epoch": 0.8404066973544438, "grad_norm": 10.209681510925293, "learning_rate": 6.276388234465508e-07, "loss": 0.3626, "step": 16614 }, { "epoch": 0.840457281602509, "grad_norm": 7.841816425323486, "learning_rate": 6.272495512205534e-07, "loss": 0.4841, "step": 16615 }, { "epoch": 0.8405078658505741, "grad_norm": 8.335522651672363, "learning_rate": 6.268603916697735e-07, "loss": 0.3797, "step": 16616 }, { "epoch": 0.8405584500986393, "grad_norm": 7.289363384246826, "learning_rate": 6.26471344804237e-07, "loss": 0.2784, "step": 16617 }, { "epoch": 0.8406090343467044, "grad_norm": 9.944781303405762, "learning_rate": 6.260824106339719e-07, "loss": 0.3934, "step": 16618 }, { "epoch": 0.8406596185947696, "grad_norm": 15.67619514465332, "learning_rate": 6.256935891689975e-07, "loss": 0.442, "step": 16619 }, { "epoch": 0.8407102028428347, "grad_norm": 10.523838996887207, "learning_rate": 6.25304880419334e-07, "loss": 0.5476, "step": 16620 }, { "epoch": 0.8407607870908999, "grad_norm": 5.150754928588867, "learning_rate": 6.249162843949963e-07, "loss": 0.4292, "step": 16621 }, { "epoch": 0.840811371338965, "grad_norm": 6.123840808868408, "learning_rate": 6.24527801105999e-07, "loss": 0.3269, "step": 16622 }, { "epoch": 0.8408619555870303, "grad_norm": 7.922513961791992, "learning_rate": 6.241394305623516e-07, "loss": 0.4173, "step": 16623 }, { "epoch": 0.8409125398350954, "grad_norm": 16.912704467773438, "learning_rate": 6.23751172774062e-07, "loss": 0.594, "step": 16624 }, { "epoch": 0.8409631240831605, "grad_norm": 6.943335056304932, "learning_rate": 6.233630277511344e-07, "loss": 0.2541, "step": 16625 }, { "epoch": 0.8410137083312257, "grad_norm": 5.01784610748291, "learning_rate": 6.229749955035685e-07, "loss": 0.3798, "step": 16626 }, { "epoch": 0.8410642925792908, "grad_norm": 10.372968673706055, "learning_rate": 6.225870760413665e-07, "loss": 0.4681, "step": 16627 }, { "epoch": 0.841114876827356, "grad_norm": 10.202101707458496, "learning_rate": 6.221992693745221e-07, "loss": 0.3636, "step": 16628 }, { "epoch": 0.8411654610754211, "grad_norm": 6.617698669433594, "learning_rate": 6.218115755130272e-07, "loss": 0.4124, "step": 16629 }, { "epoch": 0.8412160453234863, "grad_norm": 8.317405700683594, "learning_rate": 6.214239944668738e-07, "loss": 0.333, "step": 16630 }, { "epoch": 0.8412666295715514, "grad_norm": 7.016063690185547, "learning_rate": 6.210365262460483e-07, "loss": 0.3322, "step": 16631 }, { "epoch": 0.8413172138196165, "grad_norm": 4.374101161956787, "learning_rate": 6.206491708605339e-07, "loss": 0.367, "step": 16632 }, { "epoch": 0.8413677980676817, "grad_norm": 11.366752624511719, "learning_rate": 6.202619283203126e-07, "loss": 0.551, "step": 16633 }, { "epoch": 0.8414183823157468, "grad_norm": 14.971600532531738, "learning_rate": 6.198747986353632e-07, "loss": 0.5512, "step": 16634 }, { "epoch": 0.841468966563812, "grad_norm": 5.438135147094727, "learning_rate": 6.194877818156597e-07, "loss": 0.2789, "step": 16635 }, { "epoch": 0.8415195508118771, "grad_norm": 9.648613929748535, "learning_rate": 6.191008778711771e-07, "loss": 0.4508, "step": 16636 }, { "epoch": 0.8415701350599424, "grad_norm": 5.9659271240234375, "learning_rate": 6.187140868118818e-07, "loss": 0.3727, "step": 16637 }, { "epoch": 0.8416207193080075, "grad_norm": 6.2164082527160645, "learning_rate": 6.183274086477414e-07, "loss": 0.2629, "step": 16638 }, { "epoch": 0.8416713035560727, "grad_norm": 6.826305389404297, "learning_rate": 6.179408433887213e-07, "loss": 0.3924, "step": 16639 }, { "epoch": 0.8417218878041378, "grad_norm": 8.519277572631836, "learning_rate": 6.175543910447812e-07, "loss": 0.3944, "step": 16640 }, { "epoch": 0.841772472052203, "grad_norm": 10.264274597167969, "learning_rate": 6.171680516258777e-07, "loss": 0.3476, "step": 16641 }, { "epoch": 0.8418230563002681, "grad_norm": 5.117361545562744, "learning_rate": 6.167818251419682e-07, "loss": 0.3526, "step": 16642 }, { "epoch": 0.8418736405483332, "grad_norm": 8.448144912719727, "learning_rate": 6.163957116030045e-07, "loss": 0.4246, "step": 16643 }, { "epoch": 0.8419242247963984, "grad_norm": 6.349114418029785, "learning_rate": 6.160097110189334e-07, "loss": 0.4171, "step": 16644 }, { "epoch": 0.8419748090444635, "grad_norm": 8.239090919494629, "learning_rate": 6.156238233997042e-07, "loss": 0.4808, "step": 16645 }, { "epoch": 0.8420253932925287, "grad_norm": 7.407660961151123, "learning_rate": 6.152380487552595e-07, "loss": 0.4158, "step": 16646 }, { "epoch": 0.8420759775405938, "grad_norm": 8.217667579650879, "learning_rate": 6.148523870955375e-07, "loss": 0.3511, "step": 16647 }, { "epoch": 0.842126561788659, "grad_norm": 6.893404006958008, "learning_rate": 6.144668384304792e-07, "loss": 0.3524, "step": 16648 }, { "epoch": 0.8421771460367241, "grad_norm": 9.461283683776855, "learning_rate": 6.140814027700176e-07, "loss": 0.355, "step": 16649 }, { "epoch": 0.8422277302847894, "grad_norm": 6.203681468963623, "learning_rate": 6.136960801240843e-07, "loss": 0.3577, "step": 16650 }, { "epoch": 0.8422783145328545, "grad_norm": 8.056530952453613, "learning_rate": 6.133108705026087e-07, "loss": 0.3511, "step": 16651 }, { "epoch": 0.8423288987809197, "grad_norm": 5.652436256408691, "learning_rate": 6.129257739155148e-07, "loss": 0.2879, "step": 16652 }, { "epoch": 0.8423794830289848, "grad_norm": 5.961045742034912, "learning_rate": 6.125407903727282e-07, "loss": 0.3683, "step": 16653 }, { "epoch": 0.8424300672770499, "grad_norm": 13.659699440002441, "learning_rate": 6.121559198841682e-07, "loss": 0.5295, "step": 16654 }, { "epoch": 0.8424806515251151, "grad_norm": 5.528853893280029, "learning_rate": 6.117711624597517e-07, "loss": 0.3822, "step": 16655 }, { "epoch": 0.8425312357731802, "grad_norm": 10.655583381652832, "learning_rate": 6.11386518109392e-07, "loss": 0.45, "step": 16656 }, { "epoch": 0.8425818200212454, "grad_norm": 8.557893753051758, "learning_rate": 6.110019868430028e-07, "loss": 0.3893, "step": 16657 }, { "epoch": 0.8426324042693105, "grad_norm": 6.30987024307251, "learning_rate": 6.106175686704908e-07, "loss": 0.3472, "step": 16658 }, { "epoch": 0.8426829885173757, "grad_norm": 6.924603462219238, "learning_rate": 6.102332636017611e-07, "loss": 0.3646, "step": 16659 }, { "epoch": 0.8427335727654408, "grad_norm": 6.027360439300537, "learning_rate": 6.098490716467181e-07, "loss": 0.3294, "step": 16660 }, { "epoch": 0.842784157013506, "grad_norm": 9.214171409606934, "learning_rate": 6.094649928152608e-07, "loss": 0.4032, "step": 16661 }, { "epoch": 0.8428347412615711, "grad_norm": 12.891480445861816, "learning_rate": 6.090810271172854e-07, "loss": 0.5346, "step": 16662 }, { "epoch": 0.8428853255096364, "grad_norm": 5.929616928100586, "learning_rate": 6.086971745626857e-07, "loss": 0.4172, "step": 16663 }, { "epoch": 0.8429359097577015, "grad_norm": 5.762283802032471, "learning_rate": 6.083134351613529e-07, "loss": 0.3384, "step": 16664 }, { "epoch": 0.8429864940057666, "grad_norm": 6.742128849029541, "learning_rate": 6.079298089231756e-07, "loss": 0.4208, "step": 16665 }, { "epoch": 0.8430370782538318, "grad_norm": 4.9628143310546875, "learning_rate": 6.075462958580391e-07, "loss": 0.3514, "step": 16666 }, { "epoch": 0.8430876625018969, "grad_norm": 7.407140254974365, "learning_rate": 6.071628959758235e-07, "loss": 0.3566, "step": 16667 }, { "epoch": 0.8431382467499621, "grad_norm": 7.086180210113525, "learning_rate": 6.067796092864108e-07, "loss": 0.4113, "step": 16668 }, { "epoch": 0.8431888309980272, "grad_norm": 5.94667387008667, "learning_rate": 6.06396435799676e-07, "loss": 0.2917, "step": 16669 }, { "epoch": 0.8432394152460924, "grad_norm": 5.050527572631836, "learning_rate": 6.060133755254921e-07, "loss": 0.3423, "step": 16670 }, { "epoch": 0.8432899994941575, "grad_norm": 8.047124862670898, "learning_rate": 6.056304284737307e-07, "loss": 0.4333, "step": 16671 }, { "epoch": 0.8433405837422226, "grad_norm": 9.0227689743042, "learning_rate": 6.052475946542597e-07, "loss": 0.393, "step": 16672 }, { "epoch": 0.8433911679902878, "grad_norm": 10.29369831085205, "learning_rate": 6.048648740769414e-07, "loss": 0.4891, "step": 16673 }, { "epoch": 0.8434417522383529, "grad_norm": 13.310566902160645, "learning_rate": 6.044822667516409e-07, "loss": 0.4813, "step": 16674 }, { "epoch": 0.8434923364864181, "grad_norm": 10.335726737976074, "learning_rate": 6.040997726882158e-07, "loss": 0.422, "step": 16675 }, { "epoch": 0.8435429207344833, "grad_norm": 7.741040229797363, "learning_rate": 6.0371739189652e-07, "loss": 0.2697, "step": 16676 }, { "epoch": 0.8435935049825485, "grad_norm": 4.9450531005859375, "learning_rate": 6.033351243864094e-07, "loss": 0.2915, "step": 16677 }, { "epoch": 0.8436440892306136, "grad_norm": 9.883125305175781, "learning_rate": 6.029529701677328e-07, "loss": 0.334, "step": 16678 }, { "epoch": 0.8436946734786788, "grad_norm": 5.559130668640137, "learning_rate": 6.025709292503367e-07, "loss": 0.3634, "step": 16679 }, { "epoch": 0.8437452577267439, "grad_norm": 8.489524841308594, "learning_rate": 6.021890016440674e-07, "loss": 0.4467, "step": 16680 }, { "epoch": 0.8437958419748091, "grad_norm": 12.106051445007324, "learning_rate": 6.018071873587644e-07, "loss": 0.4605, "step": 16681 }, { "epoch": 0.8438464262228742, "grad_norm": 8.783523559570312, "learning_rate": 6.014254864042663e-07, "loss": 0.3882, "step": 16682 }, { "epoch": 0.8438970104709393, "grad_norm": 7.784831523895264, "learning_rate": 6.010438987904099e-07, "loss": 0.412, "step": 16683 }, { "epoch": 0.8439475947190045, "grad_norm": 10.613614082336426, "learning_rate": 6.006624245270265e-07, "loss": 0.3338, "step": 16684 }, { "epoch": 0.8439981789670696, "grad_norm": 7.135415077209473, "learning_rate": 6.002810636239459e-07, "loss": 0.3336, "step": 16685 }, { "epoch": 0.8440487632151348, "grad_norm": 7.2861223220825195, "learning_rate": 5.998998160909958e-07, "loss": 0.4247, "step": 16686 }, { "epoch": 0.8440993474631999, "grad_norm": 9.310626029968262, "learning_rate": 5.99518681938e-07, "loss": 0.436, "step": 16687 }, { "epoch": 0.8441499317112651, "grad_norm": 7.152557849884033, "learning_rate": 5.991376611747784e-07, "loss": 0.4167, "step": 16688 }, { "epoch": 0.8442005159593302, "grad_norm": 9.274035453796387, "learning_rate": 5.987567538111494e-07, "loss": 0.3935, "step": 16689 }, { "epoch": 0.8442511002073955, "grad_norm": 6.8617262840271, "learning_rate": 5.983759598569272e-07, "loss": 0.4936, "step": 16690 }, { "epoch": 0.8443016844554606, "grad_norm": 10.737017631530762, "learning_rate": 5.979952793219257e-07, "loss": 0.4952, "step": 16691 }, { "epoch": 0.8443522687035258, "grad_norm": 5.63232946395874, "learning_rate": 5.976147122159537e-07, "loss": 0.4215, "step": 16692 }, { "epoch": 0.8444028529515909, "grad_norm": 9.652175903320312, "learning_rate": 5.972342585488166e-07, "loss": 0.4976, "step": 16693 }, { "epoch": 0.844453437199656, "grad_norm": 8.548686981201172, "learning_rate": 5.968539183303174e-07, "loss": 0.3562, "step": 16694 }, { "epoch": 0.8445040214477212, "grad_norm": 11.380619049072266, "learning_rate": 5.964736915702585e-07, "loss": 0.3566, "step": 16695 }, { "epoch": 0.8445546056957863, "grad_norm": 9.070853233337402, "learning_rate": 5.96093578278436e-07, "loss": 0.441, "step": 16696 }, { "epoch": 0.8446051899438515, "grad_norm": 6.533844947814941, "learning_rate": 5.957135784646439e-07, "loss": 0.336, "step": 16697 }, { "epoch": 0.8446557741919166, "grad_norm": 7.696934700012207, "learning_rate": 5.953336921386755e-07, "loss": 0.3366, "step": 16698 }, { "epoch": 0.8447063584399818, "grad_norm": 9.888069152832031, "learning_rate": 5.949539193103194e-07, "loss": 0.4349, "step": 16699 }, { "epoch": 0.8447569426880469, "grad_norm": 12.579442024230957, "learning_rate": 5.945742599893589e-07, "loss": 0.494, "step": 16700 }, { "epoch": 0.844807526936112, "grad_norm": 9.760082244873047, "learning_rate": 5.941947141855819e-07, "loss": 0.4627, "step": 16701 }, { "epoch": 0.8448581111841772, "grad_norm": 8.906916618347168, "learning_rate": 5.938152819087628e-07, "loss": 0.4957, "step": 16702 }, { "epoch": 0.8449086954322425, "grad_norm": 6.096190452575684, "learning_rate": 5.93435963168682e-07, "loss": 0.3285, "step": 16703 }, { "epoch": 0.8449592796803076, "grad_norm": 14.889091491699219, "learning_rate": 5.930567579751135e-07, "loss": 0.4871, "step": 16704 }, { "epoch": 0.8450098639283727, "grad_norm": 21.73870086669922, "learning_rate": 5.926776663378259e-07, "loss": 0.4874, "step": 16705 }, { "epoch": 0.8450604481764379, "grad_norm": 6.1787519454956055, "learning_rate": 5.92298688266591e-07, "loss": 0.3753, "step": 16706 }, { "epoch": 0.845111032424503, "grad_norm": 6.62544584274292, "learning_rate": 5.919198237711726e-07, "loss": 0.2715, "step": 16707 }, { "epoch": 0.8451616166725682, "grad_norm": 8.505184173583984, "learning_rate": 5.915410728613319e-07, "loss": 0.4417, "step": 16708 }, { "epoch": 0.8452122009206333, "grad_norm": 12.439475059509277, "learning_rate": 5.911624355468304e-07, "loss": 0.4291, "step": 16709 }, { "epoch": 0.8452627851686985, "grad_norm": 6.768960475921631, "learning_rate": 5.907839118374237e-07, "loss": 0.3335, "step": 16710 }, { "epoch": 0.8453133694167636, "grad_norm": 14.256142616271973, "learning_rate": 5.904055017428651e-07, "loss": 0.566, "step": 16711 }, { "epoch": 0.8453639536648287, "grad_norm": 8.957635879516602, "learning_rate": 5.900272052729061e-07, "loss": 0.3913, "step": 16712 }, { "epoch": 0.8454145379128939, "grad_norm": 6.882838726043701, "learning_rate": 5.896490224372947e-07, "loss": 0.3248, "step": 16713 }, { "epoch": 0.845465122160959, "grad_norm": 12.081515312194824, "learning_rate": 5.892709532457752e-07, "loss": 0.4641, "step": 16714 }, { "epoch": 0.8455157064090242, "grad_norm": 5.063699245452881, "learning_rate": 5.888929977080893e-07, "loss": 0.4135, "step": 16715 }, { "epoch": 0.8455662906570894, "grad_norm": 7.574326992034912, "learning_rate": 5.885151558339763e-07, "loss": 0.4048, "step": 16716 }, { "epoch": 0.8456168749051546, "grad_norm": 7.615382671356201, "learning_rate": 5.881374276331714e-07, "loss": 0.3614, "step": 16717 }, { "epoch": 0.8456674591532197, "grad_norm": 10.008914947509766, "learning_rate": 5.877598131154094e-07, "loss": 0.465, "step": 16718 }, { "epoch": 0.8457180434012849, "grad_norm": 8.76462459564209, "learning_rate": 5.873823122904198e-07, "loss": 0.4184, "step": 16719 }, { "epoch": 0.84576862764935, "grad_norm": 14.216004371643066, "learning_rate": 5.870049251679289e-07, "loss": 0.3121, "step": 16720 }, { "epoch": 0.8458192118974152, "grad_norm": 6.477895259857178, "learning_rate": 5.86627651757663e-07, "loss": 0.3038, "step": 16721 }, { "epoch": 0.8458697961454803, "grad_norm": 6.839889049530029, "learning_rate": 5.862504920693418e-07, "loss": 0.3369, "step": 16722 }, { "epoch": 0.8459203803935454, "grad_norm": 10.046097755432129, "learning_rate": 5.858734461126842e-07, "loss": 0.43, "step": 16723 }, { "epoch": 0.8459709646416106, "grad_norm": 6.863831996917725, "learning_rate": 5.854965138974067e-07, "loss": 0.3845, "step": 16724 }, { "epoch": 0.8460215488896757, "grad_norm": 6.5997772216796875, "learning_rate": 5.851196954332211e-07, "loss": 0.3669, "step": 16725 }, { "epoch": 0.8460721331377409, "grad_norm": 11.813319206237793, "learning_rate": 5.847429907298374e-07, "loss": 0.561, "step": 16726 }, { "epoch": 0.846122717385806, "grad_norm": 4.903205871582031, "learning_rate": 5.843663997969623e-07, "loss": 0.3444, "step": 16727 }, { "epoch": 0.8461733016338712, "grad_norm": 10.739665985107422, "learning_rate": 5.839899226442991e-07, "loss": 0.4186, "step": 16728 }, { "epoch": 0.8462238858819363, "grad_norm": 5.707839488983154, "learning_rate": 5.836135592815484e-07, "loss": 0.355, "step": 16729 }, { "epoch": 0.8462744701300016, "grad_norm": 4.49154806137085, "learning_rate": 5.832373097184102e-07, "loss": 0.3565, "step": 16730 }, { "epoch": 0.8463250543780667, "grad_norm": 8.779248237609863, "learning_rate": 5.828611739645784e-07, "loss": 0.4441, "step": 16731 }, { "epoch": 0.8463756386261319, "grad_norm": 13.034272193908691, "learning_rate": 5.824851520297436e-07, "loss": 0.51, "step": 16732 }, { "epoch": 0.846426222874197, "grad_norm": 11.027411460876465, "learning_rate": 5.821092439235971e-07, "loss": 0.3536, "step": 16733 }, { "epoch": 0.8464768071222621, "grad_norm": 5.871135234832764, "learning_rate": 5.817334496558247e-07, "loss": 0.3946, "step": 16734 }, { "epoch": 0.8465273913703273, "grad_norm": 5.75632381439209, "learning_rate": 5.813577692361089e-07, "loss": 0.3915, "step": 16735 }, { "epoch": 0.8465779756183924, "grad_norm": 6.233004570007324, "learning_rate": 5.809822026741307e-07, "loss": 0.4207, "step": 16736 }, { "epoch": 0.8466285598664576, "grad_norm": 10.888264656066895, "learning_rate": 5.806067499795681e-07, "loss": 0.525, "step": 16737 }, { "epoch": 0.8466791441145227, "grad_norm": 11.326109886169434, "learning_rate": 5.802314111620938e-07, "loss": 0.4462, "step": 16738 }, { "epoch": 0.8467297283625879, "grad_norm": 7.491210460662842, "learning_rate": 5.798561862313823e-07, "loss": 0.3877, "step": 16739 }, { "epoch": 0.846780312610653, "grad_norm": 14.10849380493164, "learning_rate": 5.794810751970986e-07, "loss": 0.4828, "step": 16740 }, { "epoch": 0.8468308968587182, "grad_norm": 9.590254783630371, "learning_rate": 5.791060780689111e-07, "loss": 0.356, "step": 16741 }, { "epoch": 0.8468814811067833, "grad_norm": 9.843104362487793, "learning_rate": 5.787311948564822e-07, "loss": 0.3332, "step": 16742 }, { "epoch": 0.8469320653548486, "grad_norm": 13.010376930236816, "learning_rate": 5.783564255694702e-07, "loss": 0.4004, "step": 16743 }, { "epoch": 0.8469826496029137, "grad_norm": 10.38879108428955, "learning_rate": 5.779817702175339e-07, "loss": 0.4406, "step": 16744 }, { "epoch": 0.8470332338509788, "grad_norm": 5.535507678985596, "learning_rate": 5.776072288103263e-07, "loss": 0.3608, "step": 16745 }, { "epoch": 0.847083818099044, "grad_norm": 7.300294876098633, "learning_rate": 5.772328013574974e-07, "loss": 0.4218, "step": 16746 }, { "epoch": 0.8471344023471091, "grad_norm": 12.443697929382324, "learning_rate": 5.768584878686978e-07, "loss": 0.5126, "step": 16747 }, { "epoch": 0.8471849865951743, "grad_norm": 17.37202262878418, "learning_rate": 5.764842883535715e-07, "loss": 0.5309, "step": 16748 }, { "epoch": 0.8472355708432394, "grad_norm": 7.460236549377441, "learning_rate": 5.761102028217591e-07, "loss": 0.3729, "step": 16749 }, { "epoch": 0.8472861550913046, "grad_norm": 4.708341121673584, "learning_rate": 5.757362312829023e-07, "loss": 0.4232, "step": 16750 }, { "epoch": 0.8473367393393697, "grad_norm": 6.453994274139404, "learning_rate": 5.753623737466368e-07, "loss": 0.2902, "step": 16751 }, { "epoch": 0.8473873235874348, "grad_norm": 8.959871292114258, "learning_rate": 5.749886302225954e-07, "loss": 0.4304, "step": 16752 }, { "epoch": 0.8474379078355, "grad_norm": 7.719948768615723, "learning_rate": 5.746150007204088e-07, "loss": 0.4292, "step": 16753 }, { "epoch": 0.8474884920835651, "grad_norm": 15.400738716125488, "learning_rate": 5.742414852497047e-07, "loss": 0.5298, "step": 16754 }, { "epoch": 0.8475390763316303, "grad_norm": 11.057988166809082, "learning_rate": 5.738680838201066e-07, "loss": 0.4283, "step": 16755 }, { "epoch": 0.8475896605796955, "grad_norm": 7.839015483856201, "learning_rate": 5.73494796441238e-07, "loss": 0.3896, "step": 16756 }, { "epoch": 0.8476402448277607, "grad_norm": 7.529951572418213, "learning_rate": 5.731216231227166e-07, "loss": 0.4271, "step": 16757 }, { "epoch": 0.8476908290758258, "grad_norm": 8.627610206604004, "learning_rate": 5.72748563874157e-07, "loss": 0.3626, "step": 16758 }, { "epoch": 0.847741413323891, "grad_norm": 9.541678428649902, "learning_rate": 5.723756187051749e-07, "loss": 0.446, "step": 16759 }, { "epoch": 0.8477919975719561, "grad_norm": 9.383875846862793, "learning_rate": 5.720027876253781e-07, "loss": 0.3786, "step": 16760 }, { "epoch": 0.8478425818200213, "grad_norm": 7.1349687576293945, "learning_rate": 5.716300706443734e-07, "loss": 0.4242, "step": 16761 }, { "epoch": 0.8478931660680864, "grad_norm": 10.792597770690918, "learning_rate": 5.712574677717669e-07, "loss": 0.4174, "step": 16762 }, { "epoch": 0.8479437503161515, "grad_norm": 16.824968338012695, "learning_rate": 5.708849790171578e-07, "loss": 0.4994, "step": 16763 }, { "epoch": 0.8479943345642167, "grad_norm": 7.275301933288574, "learning_rate": 5.705126043901438e-07, "loss": 0.3592, "step": 16764 }, { "epoch": 0.8480449188122818, "grad_norm": 7.429014682769775, "learning_rate": 5.701403439003228e-07, "loss": 0.4332, "step": 16765 }, { "epoch": 0.848095503060347, "grad_norm": 7.316893577575684, "learning_rate": 5.697681975572844e-07, "loss": 0.3676, "step": 16766 }, { "epoch": 0.8481460873084121, "grad_norm": 9.09959602355957, "learning_rate": 5.693961653706176e-07, "loss": 0.4475, "step": 16767 }, { "epoch": 0.8481966715564773, "grad_norm": 12.155668258666992, "learning_rate": 5.690242473499108e-07, "loss": 0.4191, "step": 16768 }, { "epoch": 0.8482472558045424, "grad_norm": 11.300561904907227, "learning_rate": 5.68652443504747e-07, "loss": 0.4328, "step": 16769 }, { "epoch": 0.8482978400526077, "grad_norm": 7.198636531829834, "learning_rate": 5.68280753844705e-07, "loss": 0.4009, "step": 16770 }, { "epoch": 0.8483484243006728, "grad_norm": 10.034710884094238, "learning_rate": 5.67909178379365e-07, "loss": 0.4268, "step": 16771 }, { "epoch": 0.848399008548738, "grad_norm": 8.069019317626953, "learning_rate": 5.675377171183e-07, "loss": 0.3822, "step": 16772 }, { "epoch": 0.8484495927968031, "grad_norm": 11.93325424194336, "learning_rate": 5.671663700710806e-07, "loss": 0.4429, "step": 16773 }, { "epoch": 0.8485001770448682, "grad_norm": 11.85071849822998, "learning_rate": 5.66795137247278e-07, "loss": 0.5687, "step": 16774 }, { "epoch": 0.8485507612929334, "grad_norm": 13.673145294189453, "learning_rate": 5.664240186564562e-07, "loss": 0.3714, "step": 16775 }, { "epoch": 0.8486013455409985, "grad_norm": 8.072796821594238, "learning_rate": 5.660530143081783e-07, "loss": 0.4208, "step": 16776 }, { "epoch": 0.8486519297890637, "grad_norm": 6.434755325317383, "learning_rate": 5.656821242120052e-07, "loss": 0.4041, "step": 16777 }, { "epoch": 0.8487025140371288, "grad_norm": 10.260546684265137, "learning_rate": 5.65311348377493e-07, "loss": 0.4119, "step": 16778 }, { "epoch": 0.848753098285194, "grad_norm": 7.643775939941406, "learning_rate": 5.64940686814196e-07, "loss": 0.4388, "step": 16779 }, { "epoch": 0.8488036825332591, "grad_norm": 4.864748954772949, "learning_rate": 5.64570139531665e-07, "loss": 0.4031, "step": 16780 }, { "epoch": 0.8488542667813243, "grad_norm": 13.402621269226074, "learning_rate": 5.641997065394466e-07, "loss": 0.5307, "step": 16781 }, { "epoch": 0.8489048510293894, "grad_norm": 6.458608627319336, "learning_rate": 5.63829387847089e-07, "loss": 0.3933, "step": 16782 }, { "epoch": 0.8489554352774547, "grad_norm": 8.488866806030273, "learning_rate": 5.634591834641334e-07, "loss": 0.4379, "step": 16783 }, { "epoch": 0.8490060195255198, "grad_norm": 7.691282272338867, "learning_rate": 5.630890934001171e-07, "loss": 0.4471, "step": 16784 }, { "epoch": 0.8490566037735849, "grad_norm": 9.740002632141113, "learning_rate": 5.627191176645791e-07, "loss": 0.3418, "step": 16785 }, { "epoch": 0.8491071880216501, "grad_norm": 8.03759479522705, "learning_rate": 5.623492562670512e-07, "loss": 0.3979, "step": 16786 }, { "epoch": 0.8491577722697152, "grad_norm": 6.6894378662109375, "learning_rate": 5.619795092170638e-07, "loss": 0.3378, "step": 16787 }, { "epoch": 0.8492083565177804, "grad_norm": 7.796814441680908, "learning_rate": 5.616098765241457e-07, "loss": 0.3867, "step": 16788 }, { "epoch": 0.8492589407658455, "grad_norm": 12.400233268737793, "learning_rate": 5.612403581978204e-07, "loss": 0.3781, "step": 16789 }, { "epoch": 0.8493095250139107, "grad_norm": 9.116302490234375, "learning_rate": 5.608709542476098e-07, "loss": 0.4014, "step": 16790 }, { "epoch": 0.8493601092619758, "grad_norm": 9.107505798339844, "learning_rate": 5.605016646830325e-07, "loss": 0.363, "step": 16791 }, { "epoch": 0.849410693510041, "grad_norm": 9.655010223388672, "learning_rate": 5.601324895136045e-07, "loss": 0.3916, "step": 16792 }, { "epoch": 0.8494612777581061, "grad_norm": 10.597963333129883, "learning_rate": 5.597634287488369e-07, "loss": 0.3554, "step": 16793 }, { "epoch": 0.8495118620061712, "grad_norm": 5.389490127563477, "learning_rate": 5.593944823982411e-07, "loss": 0.4198, "step": 16794 }, { "epoch": 0.8495624462542364, "grad_norm": 5.722335338592529, "learning_rate": 5.590256504713243e-07, "loss": 0.3699, "step": 16795 }, { "epoch": 0.8496130305023016, "grad_norm": 9.172942161560059, "learning_rate": 5.586569329775887e-07, "loss": 0.3419, "step": 16796 }, { "epoch": 0.8496636147503668, "grad_norm": 9.875048637390137, "learning_rate": 5.582883299265374e-07, "loss": 0.3924, "step": 16797 }, { "epoch": 0.8497141989984319, "grad_norm": 36.95054626464844, "learning_rate": 5.57919841327667e-07, "loss": 0.382, "step": 16798 }, { "epoch": 0.8497647832464971, "grad_norm": 7.327733516693115, "learning_rate": 5.575514671904725e-07, "loss": 0.3682, "step": 16799 }, { "epoch": 0.8498153674945622, "grad_norm": 7.181415557861328, "learning_rate": 5.571832075244471e-07, "loss": 0.3747, "step": 16800 }, { "epoch": 0.8498659517426274, "grad_norm": 5.361332893371582, "learning_rate": 5.568150623390788e-07, "loss": 0.4156, "step": 16801 }, { "epoch": 0.8499165359906925, "grad_norm": 6.416058540344238, "learning_rate": 5.564470316438536e-07, "loss": 0.3974, "step": 16802 }, { "epoch": 0.8499671202387576, "grad_norm": 7.675139427185059, "learning_rate": 5.560791154482564e-07, "loss": 0.4156, "step": 16803 }, { "epoch": 0.8500177044868228, "grad_norm": 8.979918479919434, "learning_rate": 5.557113137617675e-07, "loss": 0.4166, "step": 16804 }, { "epoch": 0.8500682887348879, "grad_norm": 8.818029403686523, "learning_rate": 5.553436265938616e-07, "loss": 0.3506, "step": 16805 }, { "epoch": 0.8501188729829531, "grad_norm": 11.634852409362793, "learning_rate": 5.549760539540156e-07, "loss": 0.5126, "step": 16806 }, { "epoch": 0.8501694572310182, "grad_norm": 8.262001037597656, "learning_rate": 5.546085958517e-07, "loss": 0.3908, "step": 16807 }, { "epoch": 0.8502200414790834, "grad_norm": 8.294879913330078, "learning_rate": 5.54241252296383e-07, "loss": 0.3523, "step": 16808 }, { "epoch": 0.8502706257271486, "grad_norm": 8.436408996582031, "learning_rate": 5.538740232975314e-07, "loss": 0.4643, "step": 16809 }, { "epoch": 0.8503212099752138, "grad_norm": 12.261646270751953, "learning_rate": 5.535069088646067e-07, "loss": 0.5286, "step": 16810 }, { "epoch": 0.8503717942232789, "grad_norm": 9.22523021697998, "learning_rate": 5.531399090070683e-07, "loss": 0.455, "step": 16811 }, { "epoch": 0.850422378471344, "grad_norm": 9.18670654296875, "learning_rate": 5.527730237343743e-07, "loss": 0.3476, "step": 16812 }, { "epoch": 0.8504729627194092, "grad_norm": 6.162991046905518, "learning_rate": 5.524062530559776e-07, "loss": 0.3866, "step": 16813 }, { "epoch": 0.8505235469674743, "grad_norm": 7.983514308929443, "learning_rate": 5.52039596981328e-07, "loss": 0.3788, "step": 16814 }, { "epoch": 0.8505741312155395, "grad_norm": 17.507484436035156, "learning_rate": 5.516730555198757e-07, "loss": 0.4481, "step": 16815 }, { "epoch": 0.8506247154636046, "grad_norm": 7.22499418258667, "learning_rate": 5.513066286810642e-07, "loss": 0.4523, "step": 16816 }, { "epoch": 0.8506752997116698, "grad_norm": 7.158790588378906, "learning_rate": 5.509403164743354e-07, "loss": 0.3772, "step": 16817 }, { "epoch": 0.8507258839597349, "grad_norm": 6.37467098236084, "learning_rate": 5.50574118909128e-07, "loss": 0.2792, "step": 16818 }, { "epoch": 0.8507764682078001, "grad_norm": 6.237709045410156, "learning_rate": 5.502080359948781e-07, "loss": 0.3571, "step": 16819 }, { "epoch": 0.8508270524558652, "grad_norm": 4.868549823760986, "learning_rate": 5.498420677410193e-07, "loss": 0.3618, "step": 16820 }, { "epoch": 0.8508776367039304, "grad_norm": 8.059060096740723, "learning_rate": 5.494762141569821e-07, "loss": 0.3379, "step": 16821 }, { "epoch": 0.8509282209519955, "grad_norm": 8.072615623474121, "learning_rate": 5.491104752521919e-07, "loss": 0.3794, "step": 16822 }, { "epoch": 0.8509788052000608, "grad_norm": 9.10423469543457, "learning_rate": 5.487448510360749e-07, "loss": 0.3065, "step": 16823 }, { "epoch": 0.8510293894481259, "grad_norm": 8.144913673400879, "learning_rate": 5.483793415180516e-07, "loss": 0.3853, "step": 16824 }, { "epoch": 0.851079973696191, "grad_norm": 9.175518035888672, "learning_rate": 5.480139467075396e-07, "loss": 0.4501, "step": 16825 }, { "epoch": 0.8511305579442562, "grad_norm": 8.77757453918457, "learning_rate": 5.476486666139557e-07, "loss": 0.3115, "step": 16826 }, { "epoch": 0.8511811421923213, "grad_norm": 8.124011993408203, "learning_rate": 5.47283501246711e-07, "loss": 0.3886, "step": 16827 }, { "epoch": 0.8512317264403865, "grad_norm": 7.27623176574707, "learning_rate": 5.469184506152158e-07, "loss": 0.2895, "step": 16828 }, { "epoch": 0.8512823106884516, "grad_norm": 6.534088134765625, "learning_rate": 5.465535147288759e-07, "loss": 0.3572, "step": 16829 }, { "epoch": 0.8513328949365168, "grad_norm": 9.20376968383789, "learning_rate": 5.461886935970956e-07, "loss": 0.3397, "step": 16830 }, { "epoch": 0.8513834791845819, "grad_norm": 6.6714253425598145, "learning_rate": 5.458239872292731e-07, "loss": 0.3603, "step": 16831 }, { "epoch": 0.851434063432647, "grad_norm": 18.53931999206543, "learning_rate": 5.454593956348093e-07, "loss": 0.3894, "step": 16832 }, { "epoch": 0.8514846476807122, "grad_norm": 7.239065647125244, "learning_rate": 5.450949188230975e-07, "loss": 0.3719, "step": 16833 }, { "epoch": 0.8515352319287773, "grad_norm": 10.91296100616455, "learning_rate": 5.447305568035283e-07, "loss": 0.3956, "step": 16834 }, { "epoch": 0.8515858161768425, "grad_norm": 9.314107894897461, "learning_rate": 5.443663095854917e-07, "loss": 0.4946, "step": 16835 }, { "epoch": 0.8516364004249077, "grad_norm": 8.174845695495605, "learning_rate": 5.440021771783738e-07, "loss": 0.3646, "step": 16836 }, { "epoch": 0.8516869846729729, "grad_norm": 6.8749518394470215, "learning_rate": 5.436381595915557e-07, "loss": 0.3619, "step": 16837 }, { "epoch": 0.851737568921038, "grad_norm": 9.417360305786133, "learning_rate": 5.43274256834419e-07, "loss": 0.4707, "step": 16838 }, { "epoch": 0.8517881531691032, "grad_norm": 11.177107810974121, "learning_rate": 5.429104689163406e-07, "loss": 0.4105, "step": 16839 }, { "epoch": 0.8518387374171683, "grad_norm": 12.10702133178711, "learning_rate": 5.425467958466918e-07, "loss": 0.4433, "step": 16840 }, { "epoch": 0.8518893216652335, "grad_norm": 9.855498313903809, "learning_rate": 5.421832376348473e-07, "loss": 0.4804, "step": 16841 }, { "epoch": 0.8519399059132986, "grad_norm": 11.30605411529541, "learning_rate": 5.418197942901743e-07, "loss": 0.3964, "step": 16842 }, { "epoch": 0.8519904901613637, "grad_norm": 6.296659469604492, "learning_rate": 5.414564658220344e-07, "loss": 0.4036, "step": 16843 }, { "epoch": 0.8520410744094289, "grad_norm": 13.514017105102539, "learning_rate": 5.410932522397938e-07, "loss": 0.415, "step": 16844 }, { "epoch": 0.852091658657494, "grad_norm": 12.385442733764648, "learning_rate": 5.407301535528098e-07, "loss": 0.3314, "step": 16845 }, { "epoch": 0.8521422429055592, "grad_norm": 7.63971471786499, "learning_rate": 5.403671697704377e-07, "loss": 0.3459, "step": 16846 }, { "epoch": 0.8521928271536243, "grad_norm": 9.908596992492676, "learning_rate": 5.400043009020329e-07, "loss": 0.3931, "step": 16847 }, { "epoch": 0.8522434114016895, "grad_norm": 12.222952842712402, "learning_rate": 5.396415469569444e-07, "loss": 0.4709, "step": 16848 }, { "epoch": 0.8522939956497547, "grad_norm": 7.437892913818359, "learning_rate": 5.392789079445187e-07, "loss": 0.4395, "step": 16849 }, { "epoch": 0.8523445798978199, "grad_norm": 7.551398277282715, "learning_rate": 5.389163838741024e-07, "loss": 0.3812, "step": 16850 }, { "epoch": 0.852395164145885, "grad_norm": 9.490813255310059, "learning_rate": 5.385539747550356e-07, "loss": 0.5194, "step": 16851 }, { "epoch": 0.8524457483939502, "grad_norm": 7.532267093658447, "learning_rate": 5.381916805966553e-07, "loss": 0.4593, "step": 16852 }, { "epoch": 0.8524963326420153, "grad_norm": 5.121826648712158, "learning_rate": 5.378295014082996e-07, "loss": 0.2556, "step": 16853 }, { "epoch": 0.8525469168900804, "grad_norm": 10.502400398254395, "learning_rate": 5.374674371992999e-07, "loss": 0.3379, "step": 16854 }, { "epoch": 0.8525975011381456, "grad_norm": 10.643327713012695, "learning_rate": 5.371054879789856e-07, "loss": 0.4305, "step": 16855 }, { "epoch": 0.8526480853862107, "grad_norm": 6.060319900512695, "learning_rate": 5.367436537566828e-07, "loss": 0.3282, "step": 16856 }, { "epoch": 0.8526986696342759, "grad_norm": 7.714069366455078, "learning_rate": 5.363819345417143e-07, "loss": 0.3335, "step": 16857 }, { "epoch": 0.852749253882341, "grad_norm": 8.583292007446289, "learning_rate": 5.360203303434037e-07, "loss": 0.4042, "step": 16858 }, { "epoch": 0.8527998381304062, "grad_norm": 8.301015853881836, "learning_rate": 5.356588411710661e-07, "loss": 0.3604, "step": 16859 }, { "epoch": 0.8528504223784713, "grad_norm": 6.390042304992676, "learning_rate": 5.352974670340161e-07, "loss": 0.3559, "step": 16860 }, { "epoch": 0.8529010066265365, "grad_norm": 7.2381463050842285, "learning_rate": 5.349362079415677e-07, "loss": 0.3198, "step": 16861 }, { "epoch": 0.8529515908746016, "grad_norm": 13.054046630859375, "learning_rate": 5.345750639030284e-07, "loss": 0.4313, "step": 16862 }, { "epoch": 0.8530021751226669, "grad_norm": 8.768138885498047, "learning_rate": 5.342140349277036e-07, "loss": 0.4726, "step": 16863 }, { "epoch": 0.853052759370732, "grad_norm": 5.538094997406006, "learning_rate": 5.338531210248954e-07, "loss": 0.3197, "step": 16864 }, { "epoch": 0.8531033436187971, "grad_norm": 5.75369119644165, "learning_rate": 5.334923222039057e-07, "loss": 0.3697, "step": 16865 }, { "epoch": 0.8531539278668623, "grad_norm": 11.341211318969727, "learning_rate": 5.331316384740309e-07, "loss": 0.3711, "step": 16866 }, { "epoch": 0.8532045121149274, "grad_norm": 6.595973014831543, "learning_rate": 5.327710698445632e-07, "loss": 0.4112, "step": 16867 }, { "epoch": 0.8532550963629926, "grad_norm": 10.178982734680176, "learning_rate": 5.324106163247966e-07, "loss": 0.4744, "step": 16868 }, { "epoch": 0.8533056806110577, "grad_norm": 7.817275047302246, "learning_rate": 5.320502779240155e-07, "loss": 0.3894, "step": 16869 }, { "epoch": 0.8533562648591229, "grad_norm": 6.981714248657227, "learning_rate": 5.316900546515086e-07, "loss": 0.3854, "step": 16870 }, { "epoch": 0.853406849107188, "grad_norm": 6.861975193023682, "learning_rate": 5.313299465165555e-07, "loss": 0.3466, "step": 16871 }, { "epoch": 0.8534574333552531, "grad_norm": 5.296717643737793, "learning_rate": 5.309699535284352e-07, "loss": 0.3243, "step": 16872 }, { "epoch": 0.8535080176033183, "grad_norm": 22.374954223632812, "learning_rate": 5.306100756964255e-07, "loss": 0.5226, "step": 16873 }, { "epoch": 0.8535586018513834, "grad_norm": 15.79553508758545, "learning_rate": 5.302503130297993e-07, "loss": 0.424, "step": 16874 }, { "epoch": 0.8536091860994486, "grad_norm": 8.316091537475586, "learning_rate": 5.29890665537825e-07, "loss": 0.3977, "step": 16875 }, { "epoch": 0.8536597703475138, "grad_norm": 15.048314094543457, "learning_rate": 5.29531133229772e-07, "loss": 0.3786, "step": 16876 }, { "epoch": 0.853710354595579, "grad_norm": 5.591805934906006, "learning_rate": 5.291717161149035e-07, "loss": 0.4728, "step": 16877 }, { "epoch": 0.8537609388436441, "grad_norm": 9.23134994506836, "learning_rate": 5.288124142024803e-07, "loss": 0.4527, "step": 16878 }, { "epoch": 0.8538115230917093, "grad_norm": 6.712693214416504, "learning_rate": 5.284532275017629e-07, "loss": 0.3584, "step": 16879 }, { "epoch": 0.8538621073397744, "grad_norm": 12.07129192352295, "learning_rate": 5.280941560220049e-07, "loss": 0.4018, "step": 16880 }, { "epoch": 0.8539126915878396, "grad_norm": 5.988969326019287, "learning_rate": 5.277351997724589e-07, "loss": 0.3392, "step": 16881 }, { "epoch": 0.8539632758359047, "grad_norm": 9.228798866271973, "learning_rate": 5.273763587623748e-07, "loss": 0.4359, "step": 16882 }, { "epoch": 0.8540138600839698, "grad_norm": 14.718694686889648, "learning_rate": 5.270176330009986e-07, "loss": 0.4723, "step": 16883 }, { "epoch": 0.854064444332035, "grad_norm": 16.312664031982422, "learning_rate": 5.266590224975726e-07, "loss": 0.5592, "step": 16884 }, { "epoch": 0.8541150285801001, "grad_norm": 10.653366088867188, "learning_rate": 5.263005272613403e-07, "loss": 0.3854, "step": 16885 }, { "epoch": 0.8541656128281653, "grad_norm": 9.441793441772461, "learning_rate": 5.259421473015375e-07, "loss": 0.365, "step": 16886 }, { "epoch": 0.8542161970762304, "grad_norm": 6.912321090698242, "learning_rate": 5.255838826273974e-07, "loss": 0.4791, "step": 16887 }, { "epoch": 0.8542667813242956, "grad_norm": 27.659120559692383, "learning_rate": 5.252257332481547e-07, "loss": 0.6674, "step": 16888 }, { "epoch": 0.8543173655723608, "grad_norm": 10.728078842163086, "learning_rate": 5.248676991730361e-07, "loss": 0.4325, "step": 16889 }, { "epoch": 0.854367949820426, "grad_norm": 9.131285667419434, "learning_rate": 5.24509780411267e-07, "loss": 0.3891, "step": 16890 }, { "epoch": 0.8544185340684911, "grad_norm": 8.343221664428711, "learning_rate": 5.241519769720716e-07, "loss": 0.4129, "step": 16891 }, { "epoch": 0.8544691183165563, "grad_norm": 10.733686447143555, "learning_rate": 5.237942888646686e-07, "loss": 0.5564, "step": 16892 }, { "epoch": 0.8545197025646214, "grad_norm": 12.345171928405762, "learning_rate": 5.234367160982745e-07, "loss": 0.3601, "step": 16893 }, { "epoch": 0.8545702868126865, "grad_norm": 10.429450035095215, "learning_rate": 5.230792586821049e-07, "loss": 0.4065, "step": 16894 }, { "epoch": 0.8546208710607517, "grad_norm": 9.444723129272461, "learning_rate": 5.227219166253678e-07, "loss": 0.433, "step": 16895 }, { "epoch": 0.8546714553088168, "grad_norm": 6.512523651123047, "learning_rate": 5.223646899372731e-07, "loss": 0.4732, "step": 16896 }, { "epoch": 0.854722039556882, "grad_norm": 8.48633098602295, "learning_rate": 5.220075786270251e-07, "loss": 0.3713, "step": 16897 }, { "epoch": 0.8547726238049471, "grad_norm": 7.7029218673706055, "learning_rate": 5.216505827038242e-07, "loss": 0.2602, "step": 16898 }, { "epoch": 0.8548232080530123, "grad_norm": 8.793193817138672, "learning_rate": 5.212937021768727e-07, "loss": 0.3702, "step": 16899 }, { "epoch": 0.8548737923010774, "grad_norm": 5.172215938568115, "learning_rate": 5.209369370553641e-07, "loss": 0.3425, "step": 16900 }, { "epoch": 0.8549243765491426, "grad_norm": 7.930568218231201, "learning_rate": 5.205802873484922e-07, "loss": 0.4169, "step": 16901 }, { "epoch": 0.8549749607972077, "grad_norm": 6.661755084991455, "learning_rate": 5.202237530654453e-07, "loss": 0.4127, "step": 16902 }, { "epoch": 0.855025545045273, "grad_norm": 6.69028377532959, "learning_rate": 5.198673342154132e-07, "loss": 0.3432, "step": 16903 }, { "epoch": 0.8550761292933381, "grad_norm": 9.418389320373535, "learning_rate": 5.195110308075779e-07, "loss": 0.4038, "step": 16904 }, { "epoch": 0.8551267135414032, "grad_norm": 8.675164222717285, "learning_rate": 5.191548428511212e-07, "loss": 0.4162, "step": 16905 }, { "epoch": 0.8551772977894684, "grad_norm": 6.687256813049316, "learning_rate": 5.18798770355221e-07, "loss": 0.3691, "step": 16906 }, { "epoch": 0.8552278820375335, "grad_norm": 9.032795906066895, "learning_rate": 5.184428133290536e-07, "loss": 0.4543, "step": 16907 }, { "epoch": 0.8552784662855987, "grad_norm": 11.906980514526367, "learning_rate": 5.180869717817899e-07, "loss": 0.5228, "step": 16908 }, { "epoch": 0.8553290505336638, "grad_norm": 5.830605983734131, "learning_rate": 5.177312457225991e-07, "loss": 0.4181, "step": 16909 }, { "epoch": 0.855379634781729, "grad_norm": 10.914406776428223, "learning_rate": 5.173756351606468e-07, "loss": 0.5078, "step": 16910 }, { "epoch": 0.8554302190297941, "grad_norm": 9.389620780944824, "learning_rate": 5.170201401050978e-07, "loss": 0.4003, "step": 16911 }, { "epoch": 0.8554808032778592, "grad_norm": 6.491277694702148, "learning_rate": 5.166647605651115e-07, "loss": 0.3075, "step": 16912 }, { "epoch": 0.8555313875259244, "grad_norm": 7.3042192459106445, "learning_rate": 5.163094965498444e-07, "loss": 0.3398, "step": 16913 }, { "epoch": 0.8555819717739895, "grad_norm": 5.206323623657227, "learning_rate": 5.159543480684531e-07, "loss": 0.3235, "step": 16914 }, { "epoch": 0.8556325560220547, "grad_norm": 15.953940391540527, "learning_rate": 5.155993151300869e-07, "loss": 0.4695, "step": 16915 }, { "epoch": 0.8556831402701199, "grad_norm": 7.927541732788086, "learning_rate": 5.152443977438943e-07, "loss": 0.3538, "step": 16916 }, { "epoch": 0.8557337245181851, "grad_norm": 8.675949096679688, "learning_rate": 5.148895959190214e-07, "loss": 0.3717, "step": 16917 }, { "epoch": 0.8557843087662502, "grad_norm": 5.9347710609436035, "learning_rate": 5.145349096646108e-07, "loss": 0.326, "step": 16918 }, { "epoch": 0.8558348930143154, "grad_norm": 8.633487701416016, "learning_rate": 5.141803389898009e-07, "loss": 0.4232, "step": 16919 }, { "epoch": 0.8558854772623805, "grad_norm": 4.5246806144714355, "learning_rate": 5.138258839037291e-07, "loss": 0.3242, "step": 16920 }, { "epoch": 0.8559360615104457, "grad_norm": 9.515663146972656, "learning_rate": 5.13471544415528e-07, "loss": 0.3373, "step": 16921 }, { "epoch": 0.8559866457585108, "grad_norm": 9.3756685256958, "learning_rate": 5.131173205343276e-07, "loss": 0.3603, "step": 16922 }, { "epoch": 0.8560372300065759, "grad_norm": 5.869534015655518, "learning_rate": 5.127632122692577e-07, "loss": 0.4476, "step": 16923 }, { "epoch": 0.8560878142546411, "grad_norm": 5.8448615074157715, "learning_rate": 5.124092196294406e-07, "loss": 0.4517, "step": 16924 }, { "epoch": 0.8561383985027062, "grad_norm": 6.788605213165283, "learning_rate": 5.120553426239977e-07, "loss": 0.3751, "step": 16925 }, { "epoch": 0.8561889827507714, "grad_norm": 6.381039619445801, "learning_rate": 5.117015812620496e-07, "loss": 0.4014, "step": 16926 }, { "epoch": 0.8562395669988365, "grad_norm": 9.697407722473145, "learning_rate": 5.113479355527112e-07, "loss": 0.3859, "step": 16927 }, { "epoch": 0.8562901512469017, "grad_norm": 12.856443405151367, "learning_rate": 5.10994405505093e-07, "loss": 0.3739, "step": 16928 }, { "epoch": 0.8563407354949669, "grad_norm": 7.514629364013672, "learning_rate": 5.106409911283072e-07, "loss": 0.3523, "step": 16929 }, { "epoch": 0.8563913197430321, "grad_norm": 10.725987434387207, "learning_rate": 5.102876924314598e-07, "loss": 0.4844, "step": 16930 }, { "epoch": 0.8564419039910972, "grad_norm": 7.398234844207764, "learning_rate": 5.09934509423653e-07, "loss": 0.4588, "step": 16931 }, { "epoch": 0.8564924882391624, "grad_norm": 6.668502330780029, "learning_rate": 5.0958144211399e-07, "loss": 0.3577, "step": 16932 }, { "epoch": 0.8565430724872275, "grad_norm": 9.08458423614502, "learning_rate": 5.092284905115658e-07, "loss": 0.4118, "step": 16933 }, { "epoch": 0.8565936567352926, "grad_norm": 5.921488285064697, "learning_rate": 5.08875654625477e-07, "loss": 0.4084, "step": 16934 }, { "epoch": 0.8566442409833578, "grad_norm": 4.818599700927734, "learning_rate": 5.085229344648151e-07, "loss": 0.4125, "step": 16935 }, { "epoch": 0.8566948252314229, "grad_norm": 7.802971363067627, "learning_rate": 5.081703300386681e-07, "loss": 0.355, "step": 16936 }, { "epoch": 0.8567454094794881, "grad_norm": 4.801503658294678, "learning_rate": 5.078178413561213e-07, "loss": 0.3632, "step": 16937 }, { "epoch": 0.8567959937275532, "grad_norm": 8.860755920410156, "learning_rate": 5.074654684262592e-07, "loss": 0.4437, "step": 16938 }, { "epoch": 0.8568465779756184, "grad_norm": 4.305924415588379, "learning_rate": 5.071132112581605e-07, "loss": 0.2888, "step": 16939 }, { "epoch": 0.8568971622236835, "grad_norm": 7.586019515991211, "learning_rate": 5.067610698609016e-07, "loss": 0.381, "step": 16940 }, { "epoch": 0.8569477464717487, "grad_norm": 7.061229705810547, "learning_rate": 5.064090442435582e-07, "loss": 0.3182, "step": 16941 }, { "epoch": 0.8569983307198138, "grad_norm": 4.624582290649414, "learning_rate": 5.060571344151993e-07, "loss": 0.3418, "step": 16942 }, { "epoch": 0.857048914967879, "grad_norm": 7.601370334625244, "learning_rate": 5.057053403848927e-07, "loss": 0.3499, "step": 16943 }, { "epoch": 0.8570994992159442, "grad_norm": 12.350198745727539, "learning_rate": 5.053536621617051e-07, "loss": 0.4584, "step": 16944 }, { "epoch": 0.8571500834640093, "grad_norm": 15.739787101745605, "learning_rate": 5.050020997546967e-07, "loss": 0.5078, "step": 16945 }, { "epoch": 0.8572006677120745, "grad_norm": 4.68891716003418, "learning_rate": 5.046506531729272e-07, "loss": 0.4335, "step": 16946 }, { "epoch": 0.8572512519601396, "grad_norm": 10.625812530517578, "learning_rate": 5.042993224254522e-07, "loss": 0.4203, "step": 16947 }, { "epoch": 0.8573018362082048, "grad_norm": 8.980550765991211, "learning_rate": 5.039481075213243e-07, "loss": 0.3353, "step": 16948 }, { "epoch": 0.8573524204562699, "grad_norm": 5.862598419189453, "learning_rate": 5.035970084695946e-07, "loss": 0.2902, "step": 16949 }, { "epoch": 0.8574030047043351, "grad_norm": 5.465734004974365, "learning_rate": 5.032460252793092e-07, "loss": 0.3652, "step": 16950 }, { "epoch": 0.8574535889524002, "grad_norm": 6.689321517944336, "learning_rate": 5.028951579595115e-07, "loss": 0.3558, "step": 16951 }, { "epoch": 0.8575041732004653, "grad_norm": 8.474149703979492, "learning_rate": 5.025444065192437e-07, "loss": 0.5158, "step": 16952 }, { "epoch": 0.8575547574485305, "grad_norm": 8.72851276397705, "learning_rate": 5.021937709675439e-07, "loss": 0.3962, "step": 16953 }, { "epoch": 0.8576053416965956, "grad_norm": 14.633057594299316, "learning_rate": 5.018432513134458e-07, "loss": 0.4297, "step": 16954 }, { "epoch": 0.8576559259446608, "grad_norm": 12.119063377380371, "learning_rate": 5.01492847565983e-07, "loss": 0.279, "step": 16955 }, { "epoch": 0.857706510192726, "grad_norm": 7.561492919921875, "learning_rate": 5.011425597341835e-07, "loss": 0.351, "step": 16956 }, { "epoch": 0.8577570944407912, "grad_norm": 6.625784873962402, "learning_rate": 5.007923878270732e-07, "loss": 0.4938, "step": 16957 }, { "epoch": 0.8578076786888563, "grad_norm": 9.691672325134277, "learning_rate": 5.004423318536772e-07, "loss": 0.4934, "step": 16958 }, { "epoch": 0.8578582629369215, "grad_norm": 8.029278755187988, "learning_rate": 5.000923918230132e-07, "loss": 0.3345, "step": 16959 }, { "epoch": 0.8579088471849866, "grad_norm": 9.437112808227539, "learning_rate": 4.997425677440986e-07, "loss": 0.406, "step": 16960 }, { "epoch": 0.8579594314330518, "grad_norm": 5.878559112548828, "learning_rate": 4.99392859625949e-07, "loss": 0.3826, "step": 16961 }, { "epoch": 0.8580100156811169, "grad_norm": 6.398605823516846, "learning_rate": 4.990432674775742e-07, "loss": 0.2899, "step": 16962 }, { "epoch": 0.858060599929182, "grad_norm": 6.292287349700928, "learning_rate": 4.98693791307982e-07, "loss": 0.3711, "step": 16963 }, { "epoch": 0.8581111841772472, "grad_norm": 12.131219863891602, "learning_rate": 4.983444311261798e-07, "loss": 0.4795, "step": 16964 }, { "epoch": 0.8581617684253123, "grad_norm": 11.39525032043457, "learning_rate": 4.979951869411676e-07, "loss": 0.4408, "step": 16965 }, { "epoch": 0.8582123526733775, "grad_norm": 8.565305709838867, "learning_rate": 4.976460587619447e-07, "loss": 0.4179, "step": 16966 }, { "epoch": 0.8582629369214426, "grad_norm": 10.078622817993164, "learning_rate": 4.972970465975085e-07, "loss": 0.4304, "step": 16967 }, { "epoch": 0.8583135211695078, "grad_norm": 7.743676662445068, "learning_rate": 4.969481504568519e-07, "loss": 0.3854, "step": 16968 }, { "epoch": 0.858364105417573, "grad_norm": 7.450531005859375, "learning_rate": 4.965993703489641e-07, "loss": 0.455, "step": 16969 }, { "epoch": 0.8584146896656382, "grad_norm": 5.61777400970459, "learning_rate": 4.962507062828337e-07, "loss": 0.4708, "step": 16970 }, { "epoch": 0.8584652739137033, "grad_norm": 9.367107391357422, "learning_rate": 4.95902158267444e-07, "loss": 0.4338, "step": 16971 }, { "epoch": 0.8585158581617685, "grad_norm": 5.118662357330322, "learning_rate": 4.955537263117766e-07, "loss": 0.3746, "step": 16972 }, { "epoch": 0.8585664424098336, "grad_norm": 10.495731353759766, "learning_rate": 4.952054104248094e-07, "loss": 0.4087, "step": 16973 }, { "epoch": 0.8586170266578987, "grad_norm": 9.730334281921387, "learning_rate": 4.948572106155181e-07, "loss": 0.4412, "step": 16974 }, { "epoch": 0.8586676109059639, "grad_norm": 13.641057014465332, "learning_rate": 4.945091268928742e-07, "loss": 0.4291, "step": 16975 }, { "epoch": 0.858718195154029, "grad_norm": 4.598656177520752, "learning_rate": 4.94161159265848e-07, "loss": 0.3267, "step": 16976 }, { "epoch": 0.8587687794020942, "grad_norm": 5.974247455596924, "learning_rate": 4.938133077434054e-07, "loss": 0.3673, "step": 16977 }, { "epoch": 0.8588193636501593, "grad_norm": 7.291049003601074, "learning_rate": 4.934655723345088e-07, "loss": 0.3505, "step": 16978 }, { "epoch": 0.8588699478982245, "grad_norm": 12.27706241607666, "learning_rate": 4.9311795304812e-07, "loss": 0.5693, "step": 16979 }, { "epoch": 0.8589205321462896, "grad_norm": 6.7861247062683105, "learning_rate": 4.927704498931956e-07, "loss": 0.3446, "step": 16980 }, { "epoch": 0.8589711163943548, "grad_norm": 8.687263488769531, "learning_rate": 4.924230628786897e-07, "loss": 0.4328, "step": 16981 }, { "epoch": 0.85902170064242, "grad_norm": 13.598128318786621, "learning_rate": 4.920757920135544e-07, "loss": 0.5557, "step": 16982 }, { "epoch": 0.8590722848904851, "grad_norm": 5.473788738250732, "learning_rate": 4.91728637306737e-07, "loss": 0.3512, "step": 16983 }, { "epoch": 0.8591228691385503, "grad_norm": 6.002350330352783, "learning_rate": 4.913815987671838e-07, "loss": 0.3925, "step": 16984 }, { "epoch": 0.8591734533866154, "grad_norm": 12.20219612121582, "learning_rate": 4.910346764038366e-07, "loss": 0.3846, "step": 16985 }, { "epoch": 0.8592240376346806, "grad_norm": 8.685912132263184, "learning_rate": 4.90687870225634e-07, "loss": 0.4088, "step": 16986 }, { "epoch": 0.8592746218827457, "grad_norm": 4.899486064910889, "learning_rate": 4.903411802415143e-07, "loss": 0.3825, "step": 16987 }, { "epoch": 0.8593252061308109, "grad_norm": 8.758072853088379, "learning_rate": 4.899946064604089e-07, "loss": 0.3735, "step": 16988 }, { "epoch": 0.859375790378876, "grad_norm": 7.2964768409729, "learning_rate": 4.896481488912491e-07, "loss": 0.3865, "step": 16989 }, { "epoch": 0.8594263746269412, "grad_norm": 11.046282768249512, "learning_rate": 4.893018075429623e-07, "loss": 0.4555, "step": 16990 }, { "epoch": 0.8594769588750063, "grad_norm": 19.76453399658203, "learning_rate": 4.889555824244735e-07, "loss": 0.5006, "step": 16991 }, { "epoch": 0.8595275431230714, "grad_norm": 6.917823314666748, "learning_rate": 4.886094735447017e-07, "loss": 0.3672, "step": 16992 }, { "epoch": 0.8595781273711366, "grad_norm": 10.177207946777344, "learning_rate": 4.882634809125686e-07, "loss": 0.4278, "step": 16993 }, { "epoch": 0.8596287116192017, "grad_norm": 14.248214721679688, "learning_rate": 4.879176045369877e-07, "loss": 0.4251, "step": 16994 }, { "epoch": 0.8596792958672669, "grad_norm": 6.289850234985352, "learning_rate": 4.875718444268701e-07, "loss": 0.3766, "step": 16995 }, { "epoch": 0.8597298801153321, "grad_norm": 12.670814514160156, "learning_rate": 4.872262005911282e-07, "loss": 0.3589, "step": 16996 }, { "epoch": 0.8597804643633973, "grad_norm": 8.911409378051758, "learning_rate": 4.868806730386683e-07, "loss": 0.4441, "step": 16997 }, { "epoch": 0.8598310486114624, "grad_norm": 11.772281646728516, "learning_rate": 4.865352617783903e-07, "loss": 0.5768, "step": 16998 }, { "epoch": 0.8598816328595276, "grad_norm": 5.7507548332214355, "learning_rate": 4.861899668191971e-07, "loss": 0.3958, "step": 16999 }, { "epoch": 0.8599322171075927, "grad_norm": 7.515761852264404, "learning_rate": 4.858447881699868e-07, "loss": 0.4003, "step": 17000 }, { "epoch": 0.8599828013556579, "grad_norm": 12.334522247314453, "learning_rate": 4.85499725839651e-07, "loss": 0.471, "step": 17001 }, { "epoch": 0.860033385603723, "grad_norm": 9.329519271850586, "learning_rate": 4.851547798370842e-07, "loss": 0.4316, "step": 17002 }, { "epoch": 0.8600839698517881, "grad_norm": 5.642642021179199, "learning_rate": 4.848099501711739e-07, "loss": 0.2464, "step": 17003 }, { "epoch": 0.8601345540998533, "grad_norm": 7.61904764175415, "learning_rate": 4.84465236850804e-07, "loss": 0.4046, "step": 17004 }, { "epoch": 0.8601851383479184, "grad_norm": 10.774299621582031, "learning_rate": 4.841206398848585e-07, "loss": 0.555, "step": 17005 }, { "epoch": 0.8602357225959836, "grad_norm": 10.494749069213867, "learning_rate": 4.837761592822171e-07, "loss": 0.4707, "step": 17006 }, { "epoch": 0.8602863068440487, "grad_norm": 7.313995838165283, "learning_rate": 4.834317950517547e-07, "loss": 0.4177, "step": 17007 }, { "epoch": 0.8603368910921139, "grad_norm": 10.598980903625488, "learning_rate": 4.83087547202346e-07, "loss": 0.535, "step": 17008 }, { "epoch": 0.8603874753401791, "grad_norm": 8.895378112792969, "learning_rate": 4.827434157428629e-07, "loss": 0.4451, "step": 17009 }, { "epoch": 0.8604380595882443, "grad_norm": 5.121481418609619, "learning_rate": 4.82399400682168e-07, "loss": 0.3185, "step": 17010 }, { "epoch": 0.8604886438363094, "grad_norm": 6.648171424865723, "learning_rate": 4.820555020291307e-07, "loss": 0.3108, "step": 17011 }, { "epoch": 0.8605392280843746, "grad_norm": 7.112375259399414, "learning_rate": 4.817117197926096e-07, "loss": 0.4218, "step": 17012 }, { "epoch": 0.8605898123324397, "grad_norm": 5.825078010559082, "learning_rate": 4.813680539814635e-07, "loss": 0.3537, "step": 17013 }, { "epoch": 0.8606403965805048, "grad_norm": 9.93759536743164, "learning_rate": 4.810245046045492e-07, "loss": 0.3994, "step": 17014 }, { "epoch": 0.86069098082857, "grad_norm": 6.840474605560303, "learning_rate": 4.806810716707183e-07, "loss": 0.3079, "step": 17015 }, { "epoch": 0.8607415650766351, "grad_norm": 5.400481700897217, "learning_rate": 4.803377551888195e-07, "loss": 0.329, "step": 17016 }, { "epoch": 0.8607921493247003, "grad_norm": 10.373085021972656, "learning_rate": 4.799945551677005e-07, "loss": 0.4017, "step": 17017 }, { "epoch": 0.8608427335727654, "grad_norm": 7.510312557220459, "learning_rate": 4.796514716162043e-07, "loss": 0.3611, "step": 17018 }, { "epoch": 0.8608933178208306, "grad_norm": 10.687207221984863, "learning_rate": 4.793085045431706e-07, "loss": 0.4097, "step": 17019 }, { "epoch": 0.8609439020688957, "grad_norm": 7.963634967803955, "learning_rate": 4.789656539574383e-07, "loss": 0.3993, "step": 17020 }, { "epoch": 0.8609944863169608, "grad_norm": 3.800380229949951, "learning_rate": 4.786229198678405e-07, "loss": 0.2919, "step": 17021 }, { "epoch": 0.8610450705650261, "grad_norm": 8.201211929321289, "learning_rate": 4.782803022832094e-07, "loss": 0.4073, "step": 17022 }, { "epoch": 0.8610956548130912, "grad_norm": 9.745244979858398, "learning_rate": 4.779378012123736e-07, "loss": 0.4505, "step": 17023 }, { "epoch": 0.8611462390611564, "grad_norm": 6.514120101928711, "learning_rate": 4.775954166641566e-07, "loss": 0.312, "step": 17024 }, { "epoch": 0.8611968233092215, "grad_norm": 12.266448020935059, "learning_rate": 4.77253148647383e-07, "loss": 0.5549, "step": 17025 }, { "epoch": 0.8612474075572867, "grad_norm": 5.393396377563477, "learning_rate": 4.769109971708724e-07, "loss": 0.3747, "step": 17026 }, { "epoch": 0.8612979918053518, "grad_norm": 6.330337047576904, "learning_rate": 4.7656896224343884e-07, "loss": 0.3701, "step": 17027 }, { "epoch": 0.861348576053417, "grad_norm": 10.037923812866211, "learning_rate": 4.7622704387389807e-07, "loss": 0.5316, "step": 17028 }, { "epoch": 0.8613991603014821, "grad_norm": 7.164459228515625, "learning_rate": 4.7588524207105967e-07, "loss": 0.4716, "step": 17029 }, { "epoch": 0.8614497445495473, "grad_norm": 10.82841968536377, "learning_rate": 4.7554355684373e-07, "loss": 0.4268, "step": 17030 }, { "epoch": 0.8615003287976124, "grad_norm": 6.975594997406006, "learning_rate": 4.752019882007158e-07, "loss": 0.4384, "step": 17031 }, { "epoch": 0.8615509130456775, "grad_norm": 11.840415000915527, "learning_rate": 4.7486053615081685e-07, "loss": 0.4727, "step": 17032 }, { "epoch": 0.8616014972937427, "grad_norm": 6.789058208465576, "learning_rate": 4.74519200702831e-07, "loss": 0.3173, "step": 17033 }, { "epoch": 0.8616520815418078, "grad_norm": 8.005434036254883, "learning_rate": 4.7417798186555566e-07, "loss": 0.3598, "step": 17034 }, { "epoch": 0.861702665789873, "grad_norm": 5.922966480255127, "learning_rate": 4.738368796477827e-07, "loss": 0.3266, "step": 17035 }, { "epoch": 0.8617532500379382, "grad_norm": 8.295489311218262, "learning_rate": 4.73495894058299e-07, "loss": 0.3252, "step": 17036 }, { "epoch": 0.8618038342860034, "grad_norm": 7.620449542999268, "learning_rate": 4.731550251058942e-07, "loss": 0.4956, "step": 17037 }, { "epoch": 0.8618544185340685, "grad_norm": 12.228939056396484, "learning_rate": 4.7281427279934954e-07, "loss": 0.4504, "step": 17038 }, { "epoch": 0.8619050027821337, "grad_norm": 10.256702423095703, "learning_rate": 4.724736371474453e-07, "loss": 0.3924, "step": 17039 }, { "epoch": 0.8619555870301988, "grad_norm": 8.30506706237793, "learning_rate": 4.721331181589611e-07, "loss": 0.3759, "step": 17040 }, { "epoch": 0.862006171278264, "grad_norm": 6.606619834899902, "learning_rate": 4.7179271584266937e-07, "loss": 0.4981, "step": 17041 }, { "epoch": 0.8620567555263291, "grad_norm": 7.498754501342773, "learning_rate": 4.714524302073409e-07, "loss": 0.3894, "step": 17042 }, { "epoch": 0.8621073397743942, "grad_norm": 4.070857048034668, "learning_rate": 4.7111226126174635e-07, "loss": 0.2965, "step": 17043 }, { "epoch": 0.8621579240224594, "grad_norm": 11.848323822021484, "learning_rate": 4.707722090146499e-07, "loss": 0.4274, "step": 17044 }, { "epoch": 0.8622085082705245, "grad_norm": 25.83643341064453, "learning_rate": 4.7043227347481236e-07, "loss": 0.4907, "step": 17045 }, { "epoch": 0.8622590925185897, "grad_norm": 6.326844692230225, "learning_rate": 4.700924546509955e-07, "loss": 0.3692, "step": 17046 }, { "epoch": 0.8623096767666548, "grad_norm": 6.989936828613281, "learning_rate": 4.6975275255195405e-07, "loss": 0.4213, "step": 17047 }, { "epoch": 0.86236026101472, "grad_norm": 6.58046817779541, "learning_rate": 4.694131671864427e-07, "loss": 0.3551, "step": 17048 }, { "epoch": 0.8624108452627852, "grad_norm": 6.977826118469238, "learning_rate": 4.6907369856320997e-07, "loss": 0.3799, "step": 17049 }, { "epoch": 0.8624614295108504, "grad_norm": 7.587226867675781, "learning_rate": 4.687343466910049e-07, "loss": 0.4, "step": 17050 }, { "epoch": 0.8625120137589155, "grad_norm": 5.033212184906006, "learning_rate": 4.6839511157856955e-07, "loss": 0.3693, "step": 17051 }, { "epoch": 0.8625625980069807, "grad_norm": 5.173405170440674, "learning_rate": 4.6805599323464734e-07, "loss": 0.4322, "step": 17052 }, { "epoch": 0.8626131822550458, "grad_norm": 6.44743013381958, "learning_rate": 4.677169916679758e-07, "loss": 0.4005, "step": 17053 }, { "epoch": 0.8626637665031109, "grad_norm": 5.9493560791015625, "learning_rate": 4.6737810688728904e-07, "loss": 0.3836, "step": 17054 }, { "epoch": 0.8627143507511761, "grad_norm": 6.752346515655518, "learning_rate": 4.6703933890132115e-07, "loss": 0.3399, "step": 17055 }, { "epoch": 0.8627649349992412, "grad_norm": 7.601099491119385, "learning_rate": 4.667006877188013e-07, "loss": 0.3609, "step": 17056 }, { "epoch": 0.8628155192473064, "grad_norm": 13.415862083435059, "learning_rate": 4.663621533484536e-07, "loss": 0.5083, "step": 17057 }, { "epoch": 0.8628661034953715, "grad_norm": 8.17069149017334, "learning_rate": 4.660237357990033e-07, "loss": 0.4336, "step": 17058 }, { "epoch": 0.8629166877434367, "grad_norm": 10.246391296386719, "learning_rate": 4.6568543507917063e-07, "loss": 0.4529, "step": 17059 }, { "epoch": 0.8629672719915018, "grad_norm": 10.838154792785645, "learning_rate": 4.6534725119767033e-07, "loss": 0.3072, "step": 17060 }, { "epoch": 0.863017856239567, "grad_norm": 4.307826519012451, "learning_rate": 4.6500918416322093e-07, "loss": 0.3182, "step": 17061 }, { "epoch": 0.8630684404876322, "grad_norm": 8.144980430603027, "learning_rate": 4.6467123398452886e-07, "loss": 0.3173, "step": 17062 }, { "epoch": 0.8631190247356973, "grad_norm": 8.055851936340332, "learning_rate": 4.6433340067030543e-07, "loss": 0.3935, "step": 17063 }, { "epoch": 0.8631696089837625, "grad_norm": 10.575967788696289, "learning_rate": 4.639956842292548e-07, "loss": 0.4865, "step": 17064 }, { "epoch": 0.8632201932318276, "grad_norm": 6.843729496002197, "learning_rate": 4.6365808467007834e-07, "loss": 0.3501, "step": 17065 }, { "epoch": 0.8632707774798928, "grad_norm": 7.908565998077393, "learning_rate": 4.6332060200147686e-07, "loss": 0.3378, "step": 17066 }, { "epoch": 0.8633213617279579, "grad_norm": 6.4533867835998535, "learning_rate": 4.629832362321457e-07, "loss": 0.3763, "step": 17067 }, { "epoch": 0.8633719459760231, "grad_norm": 7.995333194732666, "learning_rate": 4.626459873707767e-07, "loss": 0.4164, "step": 17068 }, { "epoch": 0.8634225302240882, "grad_norm": 12.999096870422363, "learning_rate": 4.6230885542606185e-07, "loss": 0.6, "step": 17069 }, { "epoch": 0.8634731144721534, "grad_norm": 5.04063606262207, "learning_rate": 4.619718404066875e-07, "loss": 0.3356, "step": 17070 }, { "epoch": 0.8635236987202185, "grad_norm": 11.552844047546387, "learning_rate": 4.6163494232133785e-07, "loss": 0.3654, "step": 17071 }, { "epoch": 0.8635742829682836, "grad_norm": 8.828652381896973, "learning_rate": 4.6129816117869265e-07, "loss": 0.3659, "step": 17072 }, { "epoch": 0.8636248672163488, "grad_norm": 10.371298789978027, "learning_rate": 4.609614969874321e-07, "loss": 0.4402, "step": 17073 }, { "epoch": 0.8636754514644139, "grad_norm": 7.23807430267334, "learning_rate": 4.606249497562304e-07, "loss": 0.3957, "step": 17074 }, { "epoch": 0.8637260357124791, "grad_norm": 10.34541130065918, "learning_rate": 4.60288519493759e-07, "loss": 0.3405, "step": 17075 }, { "epoch": 0.8637766199605443, "grad_norm": 7.547393321990967, "learning_rate": 4.5995220620868694e-07, "loss": 0.4153, "step": 17076 }, { "epoch": 0.8638272042086095, "grad_norm": 9.30412769317627, "learning_rate": 4.5961600990968026e-07, "loss": 0.4215, "step": 17077 }, { "epoch": 0.8638777884566746, "grad_norm": 11.425429344177246, "learning_rate": 4.5927993060540243e-07, "loss": 0.3246, "step": 17078 }, { "epoch": 0.8639283727047398, "grad_norm": 6.756385803222656, "learning_rate": 4.589439683045138e-07, "loss": 0.4719, "step": 17079 }, { "epoch": 0.8639789569528049, "grad_norm": 7.300868988037109, "learning_rate": 4.586081230156691e-07, "loss": 0.3888, "step": 17080 }, { "epoch": 0.86402954120087, "grad_norm": 6.410163402557373, "learning_rate": 4.5827239474752483e-07, "loss": 0.4581, "step": 17081 }, { "epoch": 0.8640801254489352, "grad_norm": 9.39773178100586, "learning_rate": 4.579367835087312e-07, "loss": 0.4156, "step": 17082 }, { "epoch": 0.8641307096970003, "grad_norm": 6.550952911376953, "learning_rate": 4.576012893079346e-07, "loss": 0.2561, "step": 17083 }, { "epoch": 0.8641812939450655, "grad_norm": 8.93198299407959, "learning_rate": 4.57265912153782e-07, "loss": 0.3835, "step": 17084 }, { "epoch": 0.8642318781931306, "grad_norm": 6.801373481750488, "learning_rate": 4.569306520549144e-07, "loss": 0.3398, "step": 17085 }, { "epoch": 0.8642824624411958, "grad_norm": 6.765565872192383, "learning_rate": 4.565955090199697e-07, "loss": 0.4004, "step": 17086 }, { "epoch": 0.8643330466892609, "grad_norm": 5.6540093421936035, "learning_rate": 4.562604830575862e-07, "loss": 0.3417, "step": 17087 }, { "epoch": 0.8643836309373261, "grad_norm": 9.273689270019531, "learning_rate": 4.5592557417639405e-07, "loss": 0.3931, "step": 17088 }, { "epoch": 0.8644342151853913, "grad_norm": 6.299513816833496, "learning_rate": 4.5559078238502364e-07, "loss": 0.3587, "step": 17089 }, { "epoch": 0.8644847994334565, "grad_norm": 6.710443496704102, "learning_rate": 4.55256107692103e-07, "loss": 0.3822, "step": 17090 }, { "epoch": 0.8645353836815216, "grad_norm": 13.020363807678223, "learning_rate": 4.5492155010625484e-07, "loss": 0.3821, "step": 17091 }, { "epoch": 0.8645859679295868, "grad_norm": 6.256327152252197, "learning_rate": 4.545871096360993e-07, "loss": 0.4119, "step": 17092 }, { "epoch": 0.8646365521776519, "grad_norm": 5.21919059753418, "learning_rate": 4.542527862902557e-07, "loss": 0.2505, "step": 17093 }, { "epoch": 0.864687136425717, "grad_norm": 12.043828010559082, "learning_rate": 4.5391858007733825e-07, "loss": 0.4053, "step": 17094 }, { "epoch": 0.8647377206737822, "grad_norm": 8.26706600189209, "learning_rate": 4.535844910059578e-07, "loss": 0.3931, "step": 17095 }, { "epoch": 0.8647883049218473, "grad_norm": 9.68211841583252, "learning_rate": 4.532505190847242e-07, "loss": 0.4086, "step": 17096 }, { "epoch": 0.8648388891699125, "grad_norm": 5.986710548400879, "learning_rate": 4.529166643222421e-07, "loss": 0.4191, "step": 17097 }, { "epoch": 0.8648894734179776, "grad_norm": 9.107402801513672, "learning_rate": 4.525829267271142e-07, "loss": 0.4938, "step": 17098 }, { "epoch": 0.8649400576660428, "grad_norm": 5.744517803192139, "learning_rate": 4.5224930630794085e-07, "loss": 0.4367, "step": 17099 }, { "epoch": 0.8649906419141079, "grad_norm": 7.982881546020508, "learning_rate": 4.5191580307331897e-07, "loss": 0.4717, "step": 17100 }, { "epoch": 0.865041226162173, "grad_norm": 8.972354888916016, "learning_rate": 4.5158241703184055e-07, "loss": 0.484, "step": 17101 }, { "epoch": 0.8650918104102383, "grad_norm": 6.933152198791504, "learning_rate": 4.5124914819209717e-07, "loss": 0.3792, "step": 17102 }, { "epoch": 0.8651423946583034, "grad_norm": 4.261281967163086, "learning_rate": 4.5091599656267583e-07, "loss": 0.3166, "step": 17103 }, { "epoch": 0.8651929789063686, "grad_norm": 20.085119247436523, "learning_rate": 4.505829621521618e-07, "loss": 0.6217, "step": 17104 }, { "epoch": 0.8652435631544337, "grad_norm": 8.973443984985352, "learning_rate": 4.502500449691366e-07, "loss": 0.4423, "step": 17105 }, { "epoch": 0.8652941474024989, "grad_norm": 6.275485992431641, "learning_rate": 4.499172450221767e-07, "loss": 0.3669, "step": 17106 }, { "epoch": 0.865344731650564, "grad_norm": 9.924181938171387, "learning_rate": 4.4958456231986083e-07, "loss": 0.4117, "step": 17107 }, { "epoch": 0.8653953158986292, "grad_norm": 10.773778915405273, "learning_rate": 4.4925199687075926e-07, "loss": 0.3655, "step": 17108 }, { "epoch": 0.8654459001466943, "grad_norm": 7.593896865844727, "learning_rate": 4.489195486834419e-07, "loss": 0.398, "step": 17109 }, { "epoch": 0.8654964843947595, "grad_norm": 10.633332252502441, "learning_rate": 4.4858721776647406e-07, "loss": 0.4067, "step": 17110 }, { "epoch": 0.8655470686428246, "grad_norm": 6.9033427238464355, "learning_rate": 4.4825500412842115e-07, "loss": 0.4846, "step": 17111 }, { "epoch": 0.8655976528908897, "grad_norm": 7.464876174926758, "learning_rate": 4.4792290777784244e-07, "loss": 0.3746, "step": 17112 }, { "epoch": 0.8656482371389549, "grad_norm": 8.741329193115234, "learning_rate": 4.4759092872329544e-07, "loss": 0.3175, "step": 17113 }, { "epoch": 0.86569882138702, "grad_norm": 14.78868579864502, "learning_rate": 4.4725906697333397e-07, "loss": 0.3631, "step": 17114 }, { "epoch": 0.8657494056350853, "grad_norm": 11.234638214111328, "learning_rate": 4.4692732253650894e-07, "loss": 0.4302, "step": 17115 }, { "epoch": 0.8657999898831504, "grad_norm": 7.590538501739502, "learning_rate": 4.4659569542137013e-07, "loss": 0.404, "step": 17116 }, { "epoch": 0.8658505741312156, "grad_norm": 14.501995086669922, "learning_rate": 4.462641856364625e-07, "loss": 0.5104, "step": 17117 }, { "epoch": 0.8659011583792807, "grad_norm": 13.609633445739746, "learning_rate": 4.4593279319032623e-07, "loss": 0.3472, "step": 17118 }, { "epoch": 0.8659517426273459, "grad_norm": 11.292976379394531, "learning_rate": 4.45601518091503e-07, "loss": 0.4065, "step": 17119 }, { "epoch": 0.866002326875411, "grad_norm": 8.871771812438965, "learning_rate": 4.452703603485281e-07, "loss": 0.4621, "step": 17120 }, { "epoch": 0.8660529111234762, "grad_norm": 7.962808609008789, "learning_rate": 4.4493931996993365e-07, "loss": 0.3796, "step": 17121 }, { "epoch": 0.8661034953715413, "grad_norm": 11.236876487731934, "learning_rate": 4.4460839696425115e-07, "loss": 0.4006, "step": 17122 }, { "epoch": 0.8661540796196064, "grad_norm": 7.409695625305176, "learning_rate": 4.442775913400071e-07, "loss": 0.3325, "step": 17123 }, { "epoch": 0.8662046638676716, "grad_norm": 7.606642723083496, "learning_rate": 4.439469031057253e-07, "loss": 0.4246, "step": 17124 }, { "epoch": 0.8662552481157367, "grad_norm": 9.073812484741211, "learning_rate": 4.4361633226992884e-07, "loss": 0.4284, "step": 17125 }, { "epoch": 0.8663058323638019, "grad_norm": 7.991931915283203, "learning_rate": 4.432858788411326e-07, "loss": 0.4495, "step": 17126 }, { "epoch": 0.866356416611867, "grad_norm": 6.526061534881592, "learning_rate": 4.429555428278526e-07, "loss": 0.4404, "step": 17127 }, { "epoch": 0.8664070008599322, "grad_norm": 6.538557052612305, "learning_rate": 4.4262532423860195e-07, "loss": 0.3905, "step": 17128 }, { "epoch": 0.8664575851079974, "grad_norm": 9.478323936462402, "learning_rate": 4.422952230818883e-07, "loss": 0.4917, "step": 17129 }, { "epoch": 0.8665081693560626, "grad_norm": 6.833817481994629, "learning_rate": 4.4196523936621773e-07, "loss": 0.3513, "step": 17130 }, { "epoch": 0.8665587536041277, "grad_norm": 6.491469383239746, "learning_rate": 4.416353731000944e-07, "loss": 0.3472, "step": 17131 }, { "epoch": 0.8666093378521929, "grad_norm": 16.188108444213867, "learning_rate": 4.413056242920166e-07, "loss": 0.4784, "step": 17132 }, { "epoch": 0.866659922100258, "grad_norm": 9.389883995056152, "learning_rate": 4.4097599295048144e-07, "loss": 0.3754, "step": 17133 }, { "epoch": 0.8667105063483231, "grad_norm": 6.665005683898926, "learning_rate": 4.4064647908398374e-07, "loss": 0.4427, "step": 17134 }, { "epoch": 0.8667610905963883, "grad_norm": 11.771574020385742, "learning_rate": 4.403170827010139e-07, "loss": 0.3935, "step": 17135 }, { "epoch": 0.8668116748444534, "grad_norm": 10.711617469787598, "learning_rate": 4.3998780381005804e-07, "loss": 0.4798, "step": 17136 }, { "epoch": 0.8668622590925186, "grad_norm": 5.5801262855529785, "learning_rate": 4.3965864241960366e-07, "loss": 0.2616, "step": 17137 }, { "epoch": 0.8669128433405837, "grad_norm": 6.071799278259277, "learning_rate": 4.3932959853813075e-07, "loss": 0.3986, "step": 17138 }, { "epoch": 0.8669634275886489, "grad_norm": 5.578849792480469, "learning_rate": 4.39000672174118e-07, "loss": 0.3426, "step": 17139 }, { "epoch": 0.867014011836714, "grad_norm": 12.174237251281738, "learning_rate": 4.386718633360415e-07, "loss": 0.4843, "step": 17140 }, { "epoch": 0.8670645960847791, "grad_norm": 11.90070629119873, "learning_rate": 4.3834317203237276e-07, "loss": 0.448, "step": 17141 }, { "epoch": 0.8671151803328444, "grad_norm": 7.126314640045166, "learning_rate": 4.380145982715833e-07, "loss": 0.3998, "step": 17142 }, { "epoch": 0.8671657645809095, "grad_norm": 5.639060020446777, "learning_rate": 4.376861420621381e-07, "loss": 0.3734, "step": 17143 }, { "epoch": 0.8672163488289747, "grad_norm": 9.242233276367188, "learning_rate": 4.3735780341250146e-07, "loss": 0.4277, "step": 17144 }, { "epoch": 0.8672669330770398, "grad_norm": 6.339001178741455, "learning_rate": 4.3702958233113325e-07, "loss": 0.3006, "step": 17145 }, { "epoch": 0.867317517325105, "grad_norm": 5.334659099578857, "learning_rate": 4.367014788264917e-07, "loss": 0.3578, "step": 17146 }, { "epoch": 0.8673681015731701, "grad_norm": 4.924846649169922, "learning_rate": 4.3637349290703067e-07, "loss": 0.3187, "step": 17147 }, { "epoch": 0.8674186858212353, "grad_norm": 11.022690773010254, "learning_rate": 4.3604562458120113e-07, "loss": 0.3893, "step": 17148 }, { "epoch": 0.8674692700693004, "grad_norm": 6.205205917358398, "learning_rate": 4.35717873857453e-07, "loss": 0.375, "step": 17149 }, { "epoch": 0.8675198543173656, "grad_norm": 9.789255142211914, "learning_rate": 4.3539024074423063e-07, "loss": 0.3839, "step": 17150 }, { "epoch": 0.8675704385654307, "grad_norm": 6.2587480545043945, "learning_rate": 4.3506272524997617e-07, "loss": 0.4663, "step": 17151 }, { "epoch": 0.8676210228134958, "grad_norm": 8.727021217346191, "learning_rate": 4.3473532738312897e-07, "loss": 0.3521, "step": 17152 }, { "epoch": 0.867671607061561, "grad_norm": 12.806835174560547, "learning_rate": 4.3440804715212505e-07, "loss": 0.4172, "step": 17153 }, { "epoch": 0.8677221913096261, "grad_norm": 7.486055374145508, "learning_rate": 4.3408088456539874e-07, "loss": 0.3432, "step": 17154 }, { "epoch": 0.8677727755576914, "grad_norm": 5.595376014709473, "learning_rate": 4.337538396313795e-07, "loss": 0.395, "step": 17155 }, { "epoch": 0.8678233598057565, "grad_norm": 7.845584869384766, "learning_rate": 4.334269123584939e-07, "loss": 0.3009, "step": 17156 }, { "epoch": 0.8678739440538217, "grad_norm": 9.203043937683105, "learning_rate": 4.331001027551668e-07, "loss": 0.4392, "step": 17157 }, { "epoch": 0.8679245283018868, "grad_norm": 8.040095329284668, "learning_rate": 4.327734108298198e-07, "loss": 0.4691, "step": 17158 }, { "epoch": 0.867975112549952, "grad_norm": 10.25207233428955, "learning_rate": 4.3244683659086906e-07, "loss": 0.4978, "step": 17159 }, { "epoch": 0.8680256967980171, "grad_norm": 17.963855743408203, "learning_rate": 4.3212038004673217e-07, "loss": 0.5763, "step": 17160 }, { "epoch": 0.8680762810460823, "grad_norm": 10.048582077026367, "learning_rate": 4.317940412058197e-07, "loss": 0.5061, "step": 17161 }, { "epoch": 0.8681268652941474, "grad_norm": 12.821616172790527, "learning_rate": 4.314678200765393e-07, "loss": 0.4899, "step": 17162 }, { "epoch": 0.8681774495422125, "grad_norm": 5.397998332977295, "learning_rate": 4.311417166672999e-07, "loss": 0.3153, "step": 17163 }, { "epoch": 0.8682280337902777, "grad_norm": 7.4726243019104, "learning_rate": 4.3081573098650355e-07, "loss": 0.3442, "step": 17164 }, { "epoch": 0.8682786180383428, "grad_norm": 8.691472053527832, "learning_rate": 4.304898630425475e-07, "loss": 0.2907, "step": 17165 }, { "epoch": 0.868329202286408, "grad_norm": 6.201218128204346, "learning_rate": 4.3016411284383166e-07, "loss": 0.3077, "step": 17166 }, { "epoch": 0.8683797865344731, "grad_norm": 11.130742073059082, "learning_rate": 4.298384803987482e-07, "loss": 0.3517, "step": 17167 }, { "epoch": 0.8684303707825383, "grad_norm": 7.980109691619873, "learning_rate": 4.2951296571568767e-07, "loss": 0.3539, "step": 17168 }, { "epoch": 0.8684809550306035, "grad_norm": 7.433845043182373, "learning_rate": 4.2918756880303945e-07, "loss": 0.3739, "step": 17169 }, { "epoch": 0.8685315392786687, "grad_norm": 7.656267166137695, "learning_rate": 4.288622896691869e-07, "loss": 0.4568, "step": 17170 }, { "epoch": 0.8685821235267338, "grad_norm": 14.23533821105957, "learning_rate": 4.285371283225115e-07, "loss": 0.3366, "step": 17171 }, { "epoch": 0.868632707774799, "grad_norm": 8.773951530456543, "learning_rate": 4.282120847713933e-07, "loss": 0.3479, "step": 17172 }, { "epoch": 0.8686832920228641, "grad_norm": 7.52443265914917, "learning_rate": 4.2788715902420674e-07, "loss": 0.4433, "step": 17173 }, { "epoch": 0.8687338762709292, "grad_norm": 9.38143539428711, "learning_rate": 4.2756235108932344e-07, "loss": 0.3946, "step": 17174 }, { "epoch": 0.8687844605189944, "grad_norm": 6.998348236083984, "learning_rate": 4.2723766097511554e-07, "loss": 0.4318, "step": 17175 }, { "epoch": 0.8688350447670595, "grad_norm": 9.103907585144043, "learning_rate": 4.269130886899481e-07, "loss": 0.4693, "step": 17176 }, { "epoch": 0.8688856290151247, "grad_norm": 8.956579208374023, "learning_rate": 4.265886342421838e-07, "loss": 0.4279, "step": 17177 }, { "epoch": 0.8689362132631898, "grad_norm": 8.279545783996582, "learning_rate": 4.262642976401843e-07, "loss": 0.3722, "step": 17178 }, { "epoch": 0.868986797511255, "grad_norm": 6.303274154663086, "learning_rate": 4.259400788923057e-07, "loss": 0.3191, "step": 17179 }, { "epoch": 0.8690373817593201, "grad_norm": 12.531720161437988, "learning_rate": 4.256159780069036e-07, "loss": 0.4688, "step": 17180 }, { "epoch": 0.8690879660073852, "grad_norm": 6.86570930480957, "learning_rate": 4.25291994992329e-07, "loss": 0.3278, "step": 17181 }, { "epoch": 0.8691385502554505, "grad_norm": 6.750386714935303, "learning_rate": 4.249681298569297e-07, "loss": 0.3198, "step": 17182 }, { "epoch": 0.8691891345035156, "grad_norm": 12.794964790344238, "learning_rate": 4.246443826090507e-07, "loss": 0.555, "step": 17183 }, { "epoch": 0.8692397187515808, "grad_norm": 7.6673173904418945, "learning_rate": 4.243207532570348e-07, "loss": 0.2685, "step": 17184 }, { "epoch": 0.8692903029996459, "grad_norm": 12.469158172607422, "learning_rate": 4.2399724180922196e-07, "loss": 0.5314, "step": 17185 }, { "epoch": 0.8693408872477111, "grad_norm": 6.091982364654541, "learning_rate": 4.236738482739455e-07, "loss": 0.3362, "step": 17186 }, { "epoch": 0.8693914714957762, "grad_norm": 7.327104568481445, "learning_rate": 4.233505726595416e-07, "loss": 0.4902, "step": 17187 }, { "epoch": 0.8694420557438414, "grad_norm": 11.316293716430664, "learning_rate": 4.230274149743391e-07, "loss": 0.4815, "step": 17188 }, { "epoch": 0.8694926399919065, "grad_norm": 6.058556079864502, "learning_rate": 4.227043752266641e-07, "loss": 0.3924, "step": 17189 }, { "epoch": 0.8695432242399717, "grad_norm": 15.298654556274414, "learning_rate": 4.223814534248427e-07, "loss": 0.4117, "step": 17190 }, { "epoch": 0.8695938084880368, "grad_norm": 6.970621109008789, "learning_rate": 4.220586495771928e-07, "loss": 0.3168, "step": 17191 }, { "epoch": 0.869644392736102, "grad_norm": 10.528284072875977, "learning_rate": 4.217359636920354e-07, "loss": 0.4208, "step": 17192 }, { "epoch": 0.8696949769841671, "grad_norm": 5.042309284210205, "learning_rate": 4.214133957776834e-07, "loss": 0.3144, "step": 17193 }, { "epoch": 0.8697455612322322, "grad_norm": 9.414758682250977, "learning_rate": 4.210909458424478e-07, "loss": 0.4488, "step": 17194 }, { "epoch": 0.8697961454802975, "grad_norm": 9.925834655761719, "learning_rate": 4.207686138946399e-07, "loss": 0.3448, "step": 17195 }, { "epoch": 0.8698467297283626, "grad_norm": 11.127008438110352, "learning_rate": 4.2044639994256455e-07, "loss": 0.423, "step": 17196 }, { "epoch": 0.8698973139764278, "grad_norm": 8.492471694946289, "learning_rate": 4.201243039945224e-07, "loss": 0.4499, "step": 17197 }, { "epoch": 0.8699478982244929, "grad_norm": 14.477121353149414, "learning_rate": 4.1980232605881575e-07, "loss": 0.5624, "step": 17198 }, { "epoch": 0.8699984824725581, "grad_norm": 7.74990701675415, "learning_rate": 4.1948046614374074e-07, "loss": 0.4224, "step": 17199 }, { "epoch": 0.8700490667206232, "grad_norm": 9.52055549621582, "learning_rate": 4.1915872425758895e-07, "loss": 0.4273, "step": 17200 }, { "epoch": 0.8700996509686884, "grad_norm": 7.026286602020264, "learning_rate": 4.1883710040865277e-07, "loss": 0.4536, "step": 17201 }, { "epoch": 0.8701502352167535, "grad_norm": 6.634700775146484, "learning_rate": 4.185155946052205e-07, "loss": 0.2828, "step": 17202 }, { "epoch": 0.8702008194648186, "grad_norm": 9.969075202941895, "learning_rate": 4.1819420685557387e-07, "loss": 0.4166, "step": 17203 }, { "epoch": 0.8702514037128838, "grad_norm": 9.716794967651367, "learning_rate": 4.1787293716799573e-07, "loss": 0.2741, "step": 17204 }, { "epoch": 0.8703019879609489, "grad_norm": 9.680462837219238, "learning_rate": 4.175517855507649e-07, "loss": 0.2684, "step": 17205 }, { "epoch": 0.8703525722090141, "grad_norm": 6.974255561828613, "learning_rate": 4.172307520121549e-07, "loss": 0.3813, "step": 17206 }, { "epoch": 0.8704031564570792, "grad_norm": 5.80350399017334, "learning_rate": 4.169098365604402e-07, "loss": 0.407, "step": 17207 }, { "epoch": 0.8704537407051444, "grad_norm": 4.264718055725098, "learning_rate": 4.1658903920388904e-07, "loss": 0.321, "step": 17208 }, { "epoch": 0.8705043249532096, "grad_norm": 4.988163471221924, "learning_rate": 4.162683599507666e-07, "loss": 0.3474, "step": 17209 }, { "epoch": 0.8705549092012748, "grad_norm": 6.545147895812988, "learning_rate": 4.15947798809338e-07, "loss": 0.4294, "step": 17210 }, { "epoch": 0.8706054934493399, "grad_norm": 6.700896263122559, "learning_rate": 4.1562735578786204e-07, "loss": 0.4432, "step": 17211 }, { "epoch": 0.870656077697405, "grad_norm": 4.663485527038574, "learning_rate": 4.1530703089459555e-07, "loss": 0.3904, "step": 17212 }, { "epoch": 0.8707066619454702, "grad_norm": 11.506114959716797, "learning_rate": 4.149868241377936e-07, "loss": 0.5155, "step": 17213 }, { "epoch": 0.8707572461935353, "grad_norm": 10.209059715270996, "learning_rate": 4.1466673552570667e-07, "loss": 0.376, "step": 17214 }, { "epoch": 0.8708078304416005, "grad_norm": 9.670008659362793, "learning_rate": 4.143467650665822e-07, "loss": 0.5, "step": 17215 }, { "epoch": 0.8708584146896656, "grad_norm": 4.7549214363098145, "learning_rate": 4.140269127686658e-07, "loss": 0.389, "step": 17216 }, { "epoch": 0.8709089989377308, "grad_norm": 9.230790138244629, "learning_rate": 4.137071786401986e-07, "loss": 0.3787, "step": 17217 }, { "epoch": 0.8709595831857959, "grad_norm": 6.726841449737549, "learning_rate": 4.1338756268941904e-07, "loss": 0.4202, "step": 17218 }, { "epoch": 0.8710101674338611, "grad_norm": 15.817879676818848, "learning_rate": 4.130680649245644e-07, "loss": 0.4477, "step": 17219 }, { "epoch": 0.8710607516819262, "grad_norm": 8.353079795837402, "learning_rate": 4.1274868535386594e-07, "loss": 0.4423, "step": 17220 }, { "epoch": 0.8711113359299913, "grad_norm": 7.748517990112305, "learning_rate": 4.124294239855536e-07, "loss": 0.4379, "step": 17221 }, { "epoch": 0.8711619201780566, "grad_norm": 15.79792308807373, "learning_rate": 4.1211028082785486e-07, "loss": 0.4586, "step": 17222 }, { "epoch": 0.8712125044261217, "grad_norm": 4.354946613311768, "learning_rate": 4.117912558889925e-07, "loss": 0.3642, "step": 17223 }, { "epoch": 0.8712630886741869, "grad_norm": 6.699718952178955, "learning_rate": 4.11472349177186e-07, "loss": 0.3193, "step": 17224 }, { "epoch": 0.871313672922252, "grad_norm": 9.974385261535645, "learning_rate": 4.1115356070065504e-07, "loss": 0.4196, "step": 17225 }, { "epoch": 0.8713642571703172, "grad_norm": 13.32894229888916, "learning_rate": 4.1083489046761294e-07, "loss": 0.5307, "step": 17226 }, { "epoch": 0.8714148414183823, "grad_norm": 5.749861240386963, "learning_rate": 4.105163384862698e-07, "loss": 0.2788, "step": 17227 }, { "epoch": 0.8714654256664475, "grad_norm": 9.896180152893066, "learning_rate": 4.1019790476483743e-07, "loss": 0.421, "step": 17228 }, { "epoch": 0.8715160099145126, "grad_norm": 12.192140579223633, "learning_rate": 4.0987958931151707e-07, "loss": 0.5302, "step": 17229 }, { "epoch": 0.8715665941625778, "grad_norm": 7.62156867980957, "learning_rate": 4.0956139213451316e-07, "loss": 0.3698, "step": 17230 }, { "epoch": 0.8716171784106429, "grad_norm": 16.81513214111328, "learning_rate": 4.092433132420248e-07, "loss": 0.5736, "step": 17231 }, { "epoch": 0.871667762658708, "grad_norm": 5.267112731933594, "learning_rate": 4.0892535264224656e-07, "loss": 0.2602, "step": 17232 }, { "epoch": 0.8717183469067732, "grad_norm": 15.665183067321777, "learning_rate": 4.08607510343374e-07, "loss": 0.4931, "step": 17233 }, { "epoch": 0.8717689311548383, "grad_norm": 11.034667015075684, "learning_rate": 4.0828978635359563e-07, "loss": 0.4485, "step": 17234 }, { "epoch": 0.8718195154029036, "grad_norm": 10.013874053955078, "learning_rate": 4.079721806810977e-07, "loss": 0.389, "step": 17235 }, { "epoch": 0.8718700996509687, "grad_norm": 5.622241973876953, "learning_rate": 4.0765469333406583e-07, "loss": 0.4442, "step": 17236 }, { "epoch": 0.8719206838990339, "grad_norm": 7.503256797790527, "learning_rate": 4.073373243206807e-07, "loss": 0.4168, "step": 17237 }, { "epoch": 0.871971268147099, "grad_norm": 7.175379753112793, "learning_rate": 4.070200736491181e-07, "loss": 0.4668, "step": 17238 }, { "epoch": 0.8720218523951642, "grad_norm": 8.158506393432617, "learning_rate": 4.067029413275553e-07, "loss": 0.3422, "step": 17239 }, { "epoch": 0.8720724366432293, "grad_norm": 8.163946151733398, "learning_rate": 4.0638592736416347e-07, "loss": 0.388, "step": 17240 }, { "epoch": 0.8721230208912945, "grad_norm": 8.210644721984863, "learning_rate": 4.0606903176711064e-07, "loss": 0.3642, "step": 17241 }, { "epoch": 0.8721736051393596, "grad_norm": 11.073175430297852, "learning_rate": 4.0575225454456247e-07, "loss": 0.3561, "step": 17242 }, { "epoch": 0.8722241893874247, "grad_norm": 15.160387992858887, "learning_rate": 4.054355957046818e-07, "loss": 0.4702, "step": 17243 }, { "epoch": 0.8722747736354899, "grad_norm": 4.869034290313721, "learning_rate": 4.051190552556278e-07, "loss": 0.4183, "step": 17244 }, { "epoch": 0.872325357883555, "grad_norm": 6.314666748046875, "learning_rate": 4.048026332055577e-07, "loss": 0.2573, "step": 17245 }, { "epoch": 0.8723759421316202, "grad_norm": 6.532083988189697, "learning_rate": 4.0448632956262445e-07, "loss": 0.2904, "step": 17246 }, { "epoch": 0.8724265263796853, "grad_norm": 6.195019721984863, "learning_rate": 4.041701443349771e-07, "loss": 0.3596, "step": 17247 }, { "epoch": 0.8724771106277506, "grad_norm": 8.92479133605957, "learning_rate": 4.038540775307659e-07, "loss": 0.3693, "step": 17248 }, { "epoch": 0.8725276948758157, "grad_norm": 4.716020107269287, "learning_rate": 4.035381291581336e-07, "loss": 0.4054, "step": 17249 }, { "epoch": 0.8725782791238809, "grad_norm": 6.121580123901367, "learning_rate": 4.032222992252205e-07, "loss": 0.4255, "step": 17250 }, { "epoch": 0.872628863371946, "grad_norm": 6.88446044921875, "learning_rate": 4.0290658774016667e-07, "loss": 0.3065, "step": 17251 }, { "epoch": 0.8726794476200112, "grad_norm": 8.633729934692383, "learning_rate": 4.025909947111056e-07, "loss": 0.4665, "step": 17252 }, { "epoch": 0.8727300318680763, "grad_norm": 6.745696067810059, "learning_rate": 4.0227552014616975e-07, "loss": 0.4198, "step": 17253 }, { "epoch": 0.8727806161161414, "grad_norm": 6.2110137939453125, "learning_rate": 4.019601640534898e-07, "loss": 0.3386, "step": 17254 }, { "epoch": 0.8728312003642066, "grad_norm": 8.474465370178223, "learning_rate": 4.016449264411898e-07, "loss": 0.3587, "step": 17255 }, { "epoch": 0.8728817846122717, "grad_norm": 10.2347993850708, "learning_rate": 4.0132980731739213e-07, "loss": 0.5222, "step": 17256 }, { "epoch": 0.8729323688603369, "grad_norm": 8.332024574279785, "learning_rate": 4.0101480669021865e-07, "loss": 0.3627, "step": 17257 }, { "epoch": 0.872982953108402, "grad_norm": 5.47653341293335, "learning_rate": 4.0069992456778505e-07, "loss": 0.3675, "step": 17258 }, { "epoch": 0.8730335373564672, "grad_norm": 9.586573600769043, "learning_rate": 4.0038516095820435e-07, "loss": 0.4266, "step": 17259 }, { "epoch": 0.8730841216045323, "grad_norm": 6.470241069793701, "learning_rate": 4.000705158695889e-07, "loss": 0.3562, "step": 17260 }, { "epoch": 0.8731347058525974, "grad_norm": 16.874950408935547, "learning_rate": 3.997559893100461e-07, "loss": 0.4622, "step": 17261 }, { "epoch": 0.8731852901006627, "grad_norm": 10.742387771606445, "learning_rate": 3.994415812876784e-07, "loss": 0.4834, "step": 17262 }, { "epoch": 0.8732358743487278, "grad_norm": 7.9372100830078125, "learning_rate": 3.991272918105904e-07, "loss": 0.3423, "step": 17263 }, { "epoch": 0.873286458596793, "grad_norm": 19.728107452392578, "learning_rate": 3.9881312088687896e-07, "loss": 0.542, "step": 17264 }, { "epoch": 0.8733370428448581, "grad_norm": 10.36653995513916, "learning_rate": 3.984990685246387e-07, "loss": 0.3857, "step": 17265 }, { "epoch": 0.8733876270929233, "grad_norm": 8.141005516052246, "learning_rate": 3.981851347319643e-07, "loss": 0.4248, "step": 17266 }, { "epoch": 0.8734382113409884, "grad_norm": 16.75855827331543, "learning_rate": 3.978713195169431e-07, "loss": 0.5646, "step": 17267 }, { "epoch": 0.8734887955890536, "grad_norm": 5.585817337036133, "learning_rate": 3.9755762288766265e-07, "loss": 0.3702, "step": 17268 }, { "epoch": 0.8735393798371187, "grad_norm": 8.357793807983398, "learning_rate": 3.9724404485220523e-07, "loss": 0.43, "step": 17269 }, { "epoch": 0.8735899640851839, "grad_norm": 9.363932609558105, "learning_rate": 3.9693058541865006e-07, "loss": 0.4401, "step": 17270 }, { "epoch": 0.873640548333249, "grad_norm": 6.487473011016846, "learning_rate": 3.9661724459507667e-07, "loss": 0.417, "step": 17271 }, { "epoch": 0.8736911325813141, "grad_norm": 6.221261501312256, "learning_rate": 3.9630402238955813e-07, "loss": 0.3817, "step": 17272 }, { "epoch": 0.8737417168293793, "grad_norm": 7.293449401855469, "learning_rate": 3.959909188101641e-07, "loss": 0.4382, "step": 17273 }, { "epoch": 0.8737923010774444, "grad_norm": 7.656405925750732, "learning_rate": 3.9567793386496413e-07, "loss": 0.4083, "step": 17274 }, { "epoch": 0.8738428853255097, "grad_norm": 8.094795227050781, "learning_rate": 3.953650675620224e-07, "loss": 0.405, "step": 17275 }, { "epoch": 0.8738934695735748, "grad_norm": 11.724337577819824, "learning_rate": 3.950523199094003e-07, "loss": 0.5089, "step": 17276 }, { "epoch": 0.87394405382164, "grad_norm": 11.034294128417969, "learning_rate": 3.9473969091515794e-07, "loss": 0.545, "step": 17277 }, { "epoch": 0.8739946380697051, "grad_norm": 6.364258766174316, "learning_rate": 3.9442718058735065e-07, "loss": 0.3865, "step": 17278 }, { "epoch": 0.8740452223177703, "grad_norm": 10.963336944580078, "learning_rate": 3.9411478893402967e-07, "loss": 0.4728, "step": 17279 }, { "epoch": 0.8740958065658354, "grad_norm": 7.694027900695801, "learning_rate": 3.9380251596324635e-07, "loss": 0.4529, "step": 17280 }, { "epoch": 0.8741463908139006, "grad_norm": 8.458732604980469, "learning_rate": 3.93490361683046e-07, "loss": 0.4127, "step": 17281 }, { "epoch": 0.8741969750619657, "grad_norm": 10.367938995361328, "learning_rate": 3.931783261014716e-07, "loss": 0.4574, "step": 17282 }, { "epoch": 0.8742475593100308, "grad_norm": 10.023520469665527, "learning_rate": 3.928664092265655e-07, "loss": 0.3764, "step": 17283 }, { "epoch": 0.874298143558096, "grad_norm": 11.05706787109375, "learning_rate": 3.925546110663636e-07, "loss": 0.4537, "step": 17284 }, { "epoch": 0.8743487278061611, "grad_norm": 5.161917209625244, "learning_rate": 3.9224293162890006e-07, "loss": 0.3947, "step": 17285 }, { "epoch": 0.8743993120542263, "grad_norm": 6.978332996368408, "learning_rate": 3.919313709222072e-07, "loss": 0.3411, "step": 17286 }, { "epoch": 0.8744498963022914, "grad_norm": 8.699467658996582, "learning_rate": 3.9161992895431265e-07, "loss": 0.4672, "step": 17287 }, { "epoch": 0.8745004805503567, "grad_norm": 12.408645629882812, "learning_rate": 3.9130860573324047e-07, "loss": 0.4386, "step": 17288 }, { "epoch": 0.8745510647984218, "grad_norm": 7.572598457336426, "learning_rate": 3.909974012670148e-07, "loss": 0.2897, "step": 17289 }, { "epoch": 0.874601649046487, "grad_norm": 8.247956275939941, "learning_rate": 3.906863155636531e-07, "loss": 0.2895, "step": 17290 }, { "epoch": 0.8746522332945521, "grad_norm": 6.7081708908081055, "learning_rate": 3.903753486311712e-07, "loss": 0.2909, "step": 17291 }, { "epoch": 0.8747028175426173, "grad_norm": 7.2846221923828125, "learning_rate": 3.900645004775833e-07, "loss": 0.3087, "step": 17292 }, { "epoch": 0.8747534017906824, "grad_norm": 8.556334495544434, "learning_rate": 3.89753771110899e-07, "loss": 0.4564, "step": 17293 }, { "epoch": 0.8748039860387475, "grad_norm": 7.493687629699707, "learning_rate": 3.89443160539123e-07, "loss": 0.3379, "step": 17294 }, { "epoch": 0.8748545702868127, "grad_norm": 8.013162612915039, "learning_rate": 3.8913266877026125e-07, "loss": 0.4345, "step": 17295 }, { "epoch": 0.8749051545348778, "grad_norm": 7.2372355461120605, "learning_rate": 3.8882229581231336e-07, "loss": 0.3321, "step": 17296 }, { "epoch": 0.874955738782943, "grad_norm": 10.694021224975586, "learning_rate": 3.885120416732763e-07, "loss": 0.4718, "step": 17297 }, { "epoch": 0.8750063230310081, "grad_norm": 8.8225736618042, "learning_rate": 3.882019063611464e-07, "loss": 0.3759, "step": 17298 }, { "epoch": 0.8750569072790733, "grad_norm": 7.584251403808594, "learning_rate": 3.878918898839135e-07, "loss": 0.4948, "step": 17299 }, { "epoch": 0.8751074915271384, "grad_norm": 7.553852081298828, "learning_rate": 3.8758199224956604e-07, "loss": 0.3805, "step": 17300 }, { "epoch": 0.8751580757752035, "grad_norm": 11.773480415344238, "learning_rate": 3.8727221346609055e-07, "loss": 0.3137, "step": 17301 }, { "epoch": 0.8752086600232688, "grad_norm": 6.428379535675049, "learning_rate": 3.8696255354146837e-07, "loss": 0.3711, "step": 17302 }, { "epoch": 0.875259244271334, "grad_norm": 5.7547712326049805, "learning_rate": 3.866530124836787e-07, "loss": 0.3822, "step": 17303 }, { "epoch": 0.8753098285193991, "grad_norm": 11.862622261047363, "learning_rate": 3.863435903006979e-07, "loss": 0.4525, "step": 17304 }, { "epoch": 0.8753604127674642, "grad_norm": 10.648392677307129, "learning_rate": 3.8603428700049907e-07, "loss": 0.3957, "step": 17305 }, { "epoch": 0.8754109970155294, "grad_norm": 13.886763572692871, "learning_rate": 3.8572510259105246e-07, "loss": 0.5253, "step": 17306 }, { "epoch": 0.8754615812635945, "grad_norm": 5.7243266105651855, "learning_rate": 3.854160370803245e-07, "loss": 0.4175, "step": 17307 }, { "epoch": 0.8755121655116597, "grad_norm": 11.344335556030273, "learning_rate": 3.851070904762783e-07, "loss": 0.464, "step": 17308 }, { "epoch": 0.8755627497597248, "grad_norm": 7.276806354522705, "learning_rate": 3.8479826278687624e-07, "loss": 0.3652, "step": 17309 }, { "epoch": 0.87561333400779, "grad_norm": 7.8983378410339355, "learning_rate": 3.8448955402007547e-07, "loss": 0.3507, "step": 17310 }, { "epoch": 0.8756639182558551, "grad_norm": 6.192477703094482, "learning_rate": 3.8418096418382954e-07, "loss": 0.3628, "step": 17311 }, { "epoch": 0.8757145025039202, "grad_norm": 9.478925704956055, "learning_rate": 3.838724932860921e-07, "loss": 0.4473, "step": 17312 }, { "epoch": 0.8757650867519854, "grad_norm": 13.029718399047852, "learning_rate": 3.835641413348107e-07, "loss": 0.4571, "step": 17313 }, { "epoch": 0.8758156710000505, "grad_norm": 7.534097671508789, "learning_rate": 3.8325590833793e-07, "loss": 0.3808, "step": 17314 }, { "epoch": 0.8758662552481158, "grad_norm": 5.458852767944336, "learning_rate": 3.8294779430339435e-07, "loss": 0.3496, "step": 17315 }, { "epoch": 0.8759168394961809, "grad_norm": 9.976493835449219, "learning_rate": 3.8263979923914176e-07, "loss": 0.2535, "step": 17316 }, { "epoch": 0.8759674237442461, "grad_norm": 8.40758991241455, "learning_rate": 3.823319231531092e-07, "loss": 0.4464, "step": 17317 }, { "epoch": 0.8760180079923112, "grad_norm": 6.7301788330078125, "learning_rate": 3.8202416605322866e-07, "loss": 0.3263, "step": 17318 }, { "epoch": 0.8760685922403764, "grad_norm": 11.293641090393066, "learning_rate": 3.817165279474316e-07, "loss": 0.5028, "step": 17319 }, { "epoch": 0.8761191764884415, "grad_norm": 12.19684886932373, "learning_rate": 3.8140900884364394e-07, "loss": 0.5582, "step": 17320 }, { "epoch": 0.8761697607365067, "grad_norm": 7.967745780944824, "learning_rate": 3.8110160874979097e-07, "loss": 0.4218, "step": 17321 }, { "epoch": 0.8762203449845718, "grad_norm": 6.027857303619385, "learning_rate": 3.80794327673793e-07, "loss": 0.4267, "step": 17322 }, { "epoch": 0.8762709292326369, "grad_norm": 7.311841011047363, "learning_rate": 3.8048716562356705e-07, "loss": 0.3608, "step": 17323 }, { "epoch": 0.8763215134807021, "grad_norm": 6.254961967468262, "learning_rate": 3.801801226070295e-07, "loss": 0.3871, "step": 17324 }, { "epoch": 0.8763720977287672, "grad_norm": 10.309769630432129, "learning_rate": 3.798731986320919e-07, "loss": 0.4023, "step": 17325 }, { "epoch": 0.8764226819768324, "grad_norm": 7.087488651275635, "learning_rate": 3.795663937066618e-07, "loss": 0.41, "step": 17326 }, { "epoch": 0.8764732662248975, "grad_norm": 4.812620162963867, "learning_rate": 3.792597078386462e-07, "loss": 0.292, "step": 17327 }, { "epoch": 0.8765238504729628, "grad_norm": 7.106628894805908, "learning_rate": 3.7895314103594707e-07, "loss": 0.3078, "step": 17328 }, { "epoch": 0.8765744347210279, "grad_norm": 7.424282550811768, "learning_rate": 3.7864669330646254e-07, "loss": 0.4366, "step": 17329 }, { "epoch": 0.8766250189690931, "grad_norm": 7.607439041137695, "learning_rate": 3.783403646580913e-07, "loss": 0.3201, "step": 17330 }, { "epoch": 0.8766756032171582, "grad_norm": 6.391119480133057, "learning_rate": 3.780341550987271e-07, "loss": 0.3918, "step": 17331 }, { "epoch": 0.8767261874652233, "grad_norm": 6.391892910003662, "learning_rate": 3.777280646362563e-07, "loss": 0.355, "step": 17332 }, { "epoch": 0.8767767717132885, "grad_norm": 4.783233165740967, "learning_rate": 3.7742209327856994e-07, "loss": 0.2844, "step": 17333 }, { "epoch": 0.8768273559613536, "grad_norm": 9.142526626586914, "learning_rate": 3.7711624103355103e-07, "loss": 0.4463, "step": 17334 }, { "epoch": 0.8768779402094188, "grad_norm": 7.416680812835693, "learning_rate": 3.7681050790907947e-07, "loss": 0.3332, "step": 17335 }, { "epoch": 0.8769285244574839, "grad_norm": 6.082642078399658, "learning_rate": 3.76504893913035e-07, "loss": 0.3932, "step": 17336 }, { "epoch": 0.8769791087055491, "grad_norm": 13.636752128601074, "learning_rate": 3.7619939905329196e-07, "loss": 0.442, "step": 17337 }, { "epoch": 0.8770296929536142, "grad_norm": 6.173213481903076, "learning_rate": 3.758940233377212e-07, "loss": 0.4585, "step": 17338 }, { "epoch": 0.8770802772016794, "grad_norm": 4.559053897857666, "learning_rate": 3.755887667741931e-07, "loss": 0.2393, "step": 17339 }, { "epoch": 0.8771308614497445, "grad_norm": 5.838823318481445, "learning_rate": 3.752836293705725e-07, "loss": 0.4008, "step": 17340 }, { "epoch": 0.8771814456978096, "grad_norm": 10.053775787353516, "learning_rate": 3.74978611134722e-07, "loss": 0.4804, "step": 17341 }, { "epoch": 0.8772320299458749, "grad_norm": 8.244266510009766, "learning_rate": 3.7467371207450144e-07, "loss": 0.3181, "step": 17342 }, { "epoch": 0.87728261419394, "grad_norm": 4.740468978881836, "learning_rate": 3.743689321977678e-07, "loss": 0.3445, "step": 17343 }, { "epoch": 0.8773331984420052, "grad_norm": 5.621865749359131, "learning_rate": 3.7406427151237434e-07, "loss": 0.363, "step": 17344 }, { "epoch": 0.8773837826900703, "grad_norm": 10.17827320098877, "learning_rate": 3.7375973002617025e-07, "loss": 0.3947, "step": 17345 }, { "epoch": 0.8774343669381355, "grad_norm": 7.956662178039551, "learning_rate": 3.734553077470038e-07, "loss": 0.3917, "step": 17346 }, { "epoch": 0.8774849511862006, "grad_norm": 9.154372215270996, "learning_rate": 3.7315100468271916e-07, "loss": 0.4151, "step": 17347 }, { "epoch": 0.8775355354342658, "grad_norm": 8.453309059143066, "learning_rate": 3.728468208411579e-07, "loss": 0.4974, "step": 17348 }, { "epoch": 0.8775861196823309, "grad_norm": 7.217558860778809, "learning_rate": 3.7254275623015645e-07, "loss": 0.3032, "step": 17349 }, { "epoch": 0.8776367039303961, "grad_norm": 5.531930446624756, "learning_rate": 3.7223881085755196e-07, "loss": 0.4317, "step": 17350 }, { "epoch": 0.8776872881784612, "grad_norm": 7.748384475708008, "learning_rate": 3.7193498473117594e-07, "loss": 0.2546, "step": 17351 }, { "epoch": 0.8777378724265263, "grad_norm": 7.339841365814209, "learning_rate": 3.71631277858856e-07, "loss": 0.3842, "step": 17352 }, { "epoch": 0.8777884566745915, "grad_norm": 6.917665004730225, "learning_rate": 3.713276902484181e-07, "loss": 0.3816, "step": 17353 }, { "epoch": 0.8778390409226566, "grad_norm": 6.649691581726074, "learning_rate": 3.710242219076865e-07, "loss": 0.4176, "step": 17354 }, { "epoch": 0.8778896251707219, "grad_norm": 11.183906555175781, "learning_rate": 3.7072087284448e-07, "loss": 0.4678, "step": 17355 }, { "epoch": 0.877940209418787, "grad_norm": 6.6387834548950195, "learning_rate": 3.70417643066614e-07, "loss": 0.4277, "step": 17356 }, { "epoch": 0.8779907936668522, "grad_norm": 9.20947551727295, "learning_rate": 3.7011453258190443e-07, "loss": 0.3474, "step": 17357 }, { "epoch": 0.8780413779149173, "grad_norm": 11.828289031982422, "learning_rate": 3.69811541398159e-07, "loss": 0.4676, "step": 17358 }, { "epoch": 0.8780919621629825, "grad_norm": 7.664012908935547, "learning_rate": 3.6950866952318697e-07, "loss": 0.2346, "step": 17359 }, { "epoch": 0.8781425464110476, "grad_norm": 9.4331636428833, "learning_rate": 3.6920591696479214e-07, "loss": 0.3525, "step": 17360 }, { "epoch": 0.8781931306591128, "grad_norm": 5.877285480499268, "learning_rate": 3.689032837307749e-07, "loss": 0.4727, "step": 17361 }, { "epoch": 0.8782437149071779, "grad_norm": 10.638707160949707, "learning_rate": 3.6860076982893454e-07, "loss": 0.408, "step": 17362 }, { "epoch": 0.878294299155243, "grad_norm": 10.332560539245605, "learning_rate": 3.68298375267066e-07, "loss": 0.5426, "step": 17363 }, { "epoch": 0.8783448834033082, "grad_norm": 6.731500625610352, "learning_rate": 3.6799610005296026e-07, "loss": 0.3586, "step": 17364 }, { "epoch": 0.8783954676513733, "grad_norm": 8.540342330932617, "learning_rate": 3.676939441944072e-07, "loss": 0.3876, "step": 17365 }, { "epoch": 0.8784460518994385, "grad_norm": 5.593192100524902, "learning_rate": 3.6739190769919276e-07, "loss": 0.4393, "step": 17366 }, { "epoch": 0.8784966361475036, "grad_norm": 8.302217483520508, "learning_rate": 3.6708999057509853e-07, "loss": 0.4039, "step": 17367 }, { "epoch": 0.8785472203955689, "grad_norm": 9.694916725158691, "learning_rate": 3.6678819282990496e-07, "loss": 0.4032, "step": 17368 }, { "epoch": 0.878597804643634, "grad_norm": 11.795182228088379, "learning_rate": 3.664865144713892e-07, "loss": 0.4863, "step": 17369 }, { "epoch": 0.8786483888916992, "grad_norm": 12.838200569152832, "learning_rate": 3.6618495550732434e-07, "loss": 0.3496, "step": 17370 }, { "epoch": 0.8786989731397643, "grad_norm": 9.138214111328125, "learning_rate": 3.6588351594548045e-07, "loss": 0.4069, "step": 17371 }, { "epoch": 0.8787495573878294, "grad_norm": 8.228043556213379, "learning_rate": 3.6558219579362455e-07, "loss": 0.3607, "step": 17372 }, { "epoch": 0.8788001416358946, "grad_norm": 14.22500228881836, "learning_rate": 3.6528099505952154e-07, "loss": 0.6231, "step": 17373 }, { "epoch": 0.8788507258839597, "grad_norm": 8.57900333404541, "learning_rate": 3.64979913750933e-07, "loss": 0.3213, "step": 17374 }, { "epoch": 0.8789013101320249, "grad_norm": 5.077983856201172, "learning_rate": 3.646789518756166e-07, "loss": 0.4156, "step": 17375 }, { "epoch": 0.87895189438009, "grad_norm": 6.415443420410156, "learning_rate": 3.643781094413268e-07, "loss": 0.3609, "step": 17376 }, { "epoch": 0.8790024786281552, "grad_norm": 4.956844329833984, "learning_rate": 3.640773864558167e-07, "loss": 0.2628, "step": 17377 }, { "epoch": 0.8790530628762203, "grad_norm": 6.91464376449585, "learning_rate": 3.6377678292683517e-07, "loss": 0.3709, "step": 17378 }, { "epoch": 0.8791036471242855, "grad_norm": 9.85878849029541, "learning_rate": 3.6347629886212653e-07, "loss": 0.5018, "step": 17379 }, { "epoch": 0.8791542313723506, "grad_norm": 6.75339937210083, "learning_rate": 3.631759342694352e-07, "loss": 0.3816, "step": 17380 }, { "epoch": 0.8792048156204157, "grad_norm": 7.118320941925049, "learning_rate": 3.628756891565005e-07, "loss": 0.4046, "step": 17381 }, { "epoch": 0.879255399868481, "grad_norm": 9.519082069396973, "learning_rate": 3.6257556353105795e-07, "loss": 0.4164, "step": 17382 }, { "epoch": 0.8793059841165461, "grad_norm": 8.964692115783691, "learning_rate": 3.6227555740084406e-07, "loss": 0.3619, "step": 17383 }, { "epoch": 0.8793565683646113, "grad_norm": 5.832911491394043, "learning_rate": 3.6197567077358495e-07, "loss": 0.408, "step": 17384 }, { "epoch": 0.8794071526126764, "grad_norm": 7.11760950088501, "learning_rate": 3.6167590365701054e-07, "loss": 0.3911, "step": 17385 }, { "epoch": 0.8794577368607416, "grad_norm": 10.292176246643066, "learning_rate": 3.613762560588452e-07, "loss": 0.4177, "step": 17386 }, { "epoch": 0.8795083211088067, "grad_norm": 11.324122428894043, "learning_rate": 3.6107672798680883e-07, "loss": 0.3583, "step": 17387 }, { "epoch": 0.8795589053568719, "grad_norm": 7.993025302886963, "learning_rate": 3.6077731944862083e-07, "loss": 0.3728, "step": 17388 }, { "epoch": 0.879609489604937, "grad_norm": 8.351524353027344, "learning_rate": 3.6047803045199617e-07, "loss": 0.4538, "step": 17389 }, { "epoch": 0.8796600738530022, "grad_norm": 6.683872222900391, "learning_rate": 3.6017886100464585e-07, "loss": 0.4652, "step": 17390 }, { "epoch": 0.8797106581010673, "grad_norm": 5.343685150146484, "learning_rate": 3.5987981111427874e-07, "loss": 0.3679, "step": 17391 }, { "epoch": 0.8797612423491324, "grad_norm": 7.6338300704956055, "learning_rate": 3.59580880788602e-07, "loss": 0.4226, "step": 17392 }, { "epoch": 0.8798118265971976, "grad_norm": 10.081198692321777, "learning_rate": 3.592820700353172e-07, "loss": 0.4995, "step": 17393 }, { "epoch": 0.8798624108452627, "grad_norm": 5.47899866104126, "learning_rate": 3.589833788621233e-07, "loss": 0.3263, "step": 17394 }, { "epoch": 0.879912995093328, "grad_norm": 8.643445014953613, "learning_rate": 3.5868480727671905e-07, "loss": 0.3431, "step": 17395 }, { "epoch": 0.8799635793413931, "grad_norm": 9.031965255737305, "learning_rate": 3.5838635528679666e-07, "loss": 0.3261, "step": 17396 }, { "epoch": 0.8800141635894583, "grad_norm": 12.719375610351562, "learning_rate": 3.5808802290004665e-07, "loss": 0.465, "step": 17397 }, { "epoch": 0.8800647478375234, "grad_norm": 10.810154914855957, "learning_rate": 3.577898101241556e-07, "loss": 0.466, "step": 17398 }, { "epoch": 0.8801153320855886, "grad_norm": 5.405527114868164, "learning_rate": 3.5749171696680794e-07, "loss": 0.3658, "step": 17399 }, { "epoch": 0.8801659163336537, "grad_norm": 7.179388999938965, "learning_rate": 3.571937434356865e-07, "loss": 0.2764, "step": 17400 }, { "epoch": 0.8802165005817189, "grad_norm": 8.15046215057373, "learning_rate": 3.568958895384672e-07, "loss": 0.3516, "step": 17401 }, { "epoch": 0.880267084829784, "grad_norm": 6.517200469970703, "learning_rate": 3.565981552828257e-07, "loss": 0.4003, "step": 17402 }, { "epoch": 0.8803176690778491, "grad_norm": 8.349761962890625, "learning_rate": 3.563005406764347e-07, "loss": 0.3827, "step": 17403 }, { "epoch": 0.8803682533259143, "grad_norm": 8.092041015625, "learning_rate": 3.56003045726962e-07, "loss": 0.3996, "step": 17404 }, { "epoch": 0.8804188375739794, "grad_norm": 8.684165000915527, "learning_rate": 3.55705670442073e-07, "loss": 0.3957, "step": 17405 }, { "epoch": 0.8804694218220446, "grad_norm": 6.301449775695801, "learning_rate": 3.5540841482943234e-07, "loss": 0.3734, "step": 17406 }, { "epoch": 0.8805200060701097, "grad_norm": 12.467120170593262, "learning_rate": 3.551112788966976e-07, "loss": 0.4215, "step": 17407 }, { "epoch": 0.880570590318175, "grad_norm": 15.595074653625488, "learning_rate": 3.548142626515266e-07, "loss": 0.4697, "step": 17408 }, { "epoch": 0.8806211745662401, "grad_norm": 8.174878120422363, "learning_rate": 3.5451736610157164e-07, "loss": 0.3683, "step": 17409 }, { "epoch": 0.8806717588143053, "grad_norm": 7.053830623626709, "learning_rate": 3.5422058925448366e-07, "loss": 0.3546, "step": 17410 }, { "epoch": 0.8807223430623704, "grad_norm": 7.712808609008789, "learning_rate": 3.539239321179089e-07, "loss": 0.4492, "step": 17411 }, { "epoch": 0.8807729273104355, "grad_norm": 6.97032356262207, "learning_rate": 3.536273946994928e-07, "loss": 0.433, "step": 17412 }, { "epoch": 0.8808235115585007, "grad_norm": 14.094428062438965, "learning_rate": 3.53330977006876e-07, "loss": 0.5026, "step": 17413 }, { "epoch": 0.8808740958065658, "grad_norm": 15.635045051574707, "learning_rate": 3.5303467904769575e-07, "loss": 0.471, "step": 17414 }, { "epoch": 0.880924680054631, "grad_norm": 4.85919713973999, "learning_rate": 3.5273850082958807e-07, "loss": 0.2787, "step": 17415 }, { "epoch": 0.8809752643026961, "grad_norm": 14.188074111938477, "learning_rate": 3.524424423601841e-07, "loss": 0.443, "step": 17416 }, { "epoch": 0.8810258485507613, "grad_norm": 5.568667411804199, "learning_rate": 3.521465036471122e-07, "loss": 0.4441, "step": 17417 }, { "epoch": 0.8810764327988264, "grad_norm": 14.94604778289795, "learning_rate": 3.518506846979991e-07, "loss": 0.4978, "step": 17418 }, { "epoch": 0.8811270170468916, "grad_norm": 18.8197078704834, "learning_rate": 3.5155498552046694e-07, "loss": 0.595, "step": 17419 }, { "epoch": 0.8811776012949567, "grad_norm": 7.866609573364258, "learning_rate": 3.5125940612213363e-07, "loss": 0.4106, "step": 17420 }, { "epoch": 0.881228185543022, "grad_norm": 8.343045234680176, "learning_rate": 3.50963946510619e-07, "loss": 0.4136, "step": 17421 }, { "epoch": 0.8812787697910871, "grad_norm": 7.384975910186768, "learning_rate": 3.506686066935322e-07, "loss": 0.3299, "step": 17422 }, { "epoch": 0.8813293540391522, "grad_norm": 13.541349411010742, "learning_rate": 3.503733866784864e-07, "loss": 0.4128, "step": 17423 }, { "epoch": 0.8813799382872174, "grad_norm": 8.122296333312988, "learning_rate": 3.5007828647308786e-07, "loss": 0.3688, "step": 17424 }, { "epoch": 0.8814305225352825, "grad_norm": 7.313528060913086, "learning_rate": 3.497833060849404e-07, "loss": 0.366, "step": 17425 }, { "epoch": 0.8814811067833477, "grad_norm": 5.468459606170654, "learning_rate": 3.494884455216441e-07, "loss": 0.383, "step": 17426 }, { "epoch": 0.8815316910314128, "grad_norm": 5.221230506896973, "learning_rate": 3.491937047907984e-07, "loss": 0.3884, "step": 17427 }, { "epoch": 0.881582275279478, "grad_norm": 7.08918571472168, "learning_rate": 3.4889908389999725e-07, "loss": 0.3842, "step": 17428 }, { "epoch": 0.8816328595275431, "grad_norm": 6.8057050704956055, "learning_rate": 3.4860458285683183e-07, "loss": 0.3194, "step": 17429 }, { "epoch": 0.8816834437756083, "grad_norm": 9.195235252380371, "learning_rate": 3.483102016688922e-07, "loss": 0.3554, "step": 17430 }, { "epoch": 0.8817340280236734, "grad_norm": 8.595660209655762, "learning_rate": 3.480159403437633e-07, "loss": 0.3592, "step": 17431 }, { "epoch": 0.8817846122717385, "grad_norm": 6.799525737762451, "learning_rate": 3.477217988890258e-07, "loss": 0.3344, "step": 17432 }, { "epoch": 0.8818351965198037, "grad_norm": 9.716530799865723, "learning_rate": 3.474277773122614e-07, "loss": 0.4516, "step": 17433 }, { "epoch": 0.8818857807678688, "grad_norm": 6.725574016571045, "learning_rate": 3.471338756210452e-07, "loss": 0.3908, "step": 17434 }, { "epoch": 0.8819363650159341, "grad_norm": 5.87245512008667, "learning_rate": 3.4684009382295103e-07, "loss": 0.3354, "step": 17435 }, { "epoch": 0.8819869492639992, "grad_norm": 7.362422466278076, "learning_rate": 3.4654643192554794e-07, "loss": 0.3932, "step": 17436 }, { "epoch": 0.8820375335120644, "grad_norm": 10.975787162780762, "learning_rate": 3.462528899364026e-07, "loss": 0.4284, "step": 17437 }, { "epoch": 0.8820881177601295, "grad_norm": 12.405603408813477, "learning_rate": 3.4595946786308066e-07, "loss": 0.4677, "step": 17438 }, { "epoch": 0.8821387020081947, "grad_norm": 9.14435863494873, "learning_rate": 3.456661657131416e-07, "loss": 0.4724, "step": 17439 }, { "epoch": 0.8821892862562598, "grad_norm": 9.83502197265625, "learning_rate": 3.453729834941427e-07, "loss": 0.4667, "step": 17440 }, { "epoch": 0.882239870504325, "grad_norm": 5.7320637702941895, "learning_rate": 3.4507992121363965e-07, "loss": 0.3786, "step": 17441 }, { "epoch": 0.8822904547523901, "grad_norm": 6.8035101890563965, "learning_rate": 3.447869788791841e-07, "loss": 0.3826, "step": 17442 }, { "epoch": 0.8823410390004552, "grad_norm": 12.47605037689209, "learning_rate": 3.4449415649832285e-07, "loss": 0.4126, "step": 17443 }, { "epoch": 0.8823916232485204, "grad_norm": 7.215145587921143, "learning_rate": 3.442014540786032e-07, "loss": 0.4202, "step": 17444 }, { "epoch": 0.8824422074965855, "grad_norm": 11.013257026672363, "learning_rate": 3.4390887162756623e-07, "loss": 0.2892, "step": 17445 }, { "epoch": 0.8824927917446507, "grad_norm": 6.869777679443359, "learning_rate": 3.4361640915275053e-07, "loss": 0.3318, "step": 17446 }, { "epoch": 0.8825433759927158, "grad_norm": 11.660113334655762, "learning_rate": 3.4332406666169495e-07, "loss": 0.4727, "step": 17447 }, { "epoch": 0.8825939602407811, "grad_norm": 4.673008918762207, "learning_rate": 3.43031844161929e-07, "loss": 0.3961, "step": 17448 }, { "epoch": 0.8826445444888462, "grad_norm": 10.30801010131836, "learning_rate": 3.4273974166098345e-07, "loss": 0.3851, "step": 17449 }, { "epoch": 0.8826951287369114, "grad_norm": 22.609832763671875, "learning_rate": 3.424477591663866e-07, "loss": 0.3984, "step": 17450 }, { "epoch": 0.8827457129849765, "grad_norm": 9.988581657409668, "learning_rate": 3.4215589668566087e-07, "loss": 0.3904, "step": 17451 }, { "epoch": 0.8827962972330416, "grad_norm": 4.3337483406066895, "learning_rate": 3.4186415422632625e-07, "loss": 0.3707, "step": 17452 }, { "epoch": 0.8828468814811068, "grad_norm": 7.025599479675293, "learning_rate": 3.415725317959023e-07, "loss": 0.3222, "step": 17453 }, { "epoch": 0.8828974657291719, "grad_norm": 4.3075761795043945, "learning_rate": 3.41281029401902e-07, "loss": 0.2241, "step": 17454 }, { "epoch": 0.8829480499772371, "grad_norm": 6.250221252441406, "learning_rate": 3.409896470518359e-07, "loss": 0.3822, "step": 17455 }, { "epoch": 0.8829986342253022, "grad_norm": 13.138399124145508, "learning_rate": 3.4069838475321414e-07, "loss": 0.5274, "step": 17456 }, { "epoch": 0.8830492184733674, "grad_norm": 7.5535736083984375, "learning_rate": 3.4040724251354127e-07, "loss": 0.3389, "step": 17457 }, { "epoch": 0.8830998027214325, "grad_norm": 8.918554306030273, "learning_rate": 3.40116220340318e-07, "loss": 0.5033, "step": 17458 }, { "epoch": 0.8831503869694977, "grad_norm": 12.398269653320312, "learning_rate": 3.3982531824104493e-07, "loss": 0.5076, "step": 17459 }, { "epoch": 0.8832009712175628, "grad_norm": 6.1809868812561035, "learning_rate": 3.395345362232172e-07, "loss": 0.4294, "step": 17460 }, { "epoch": 0.8832515554656281, "grad_norm": 12.420526504516602, "learning_rate": 3.392438742943277e-07, "loss": 0.3019, "step": 17461 }, { "epoch": 0.8833021397136932, "grad_norm": 5.856642723083496, "learning_rate": 3.38953332461866e-07, "loss": 0.3087, "step": 17462 }, { "epoch": 0.8833527239617583, "grad_norm": 8.621784210205078, "learning_rate": 3.3866291073331836e-07, "loss": 0.3493, "step": 17463 }, { "epoch": 0.8834033082098235, "grad_norm": 7.635552406311035, "learning_rate": 3.383726091161676e-07, "loss": 0.3515, "step": 17464 }, { "epoch": 0.8834538924578886, "grad_norm": 9.188045501708984, "learning_rate": 3.380824276178962e-07, "loss": 0.4345, "step": 17465 }, { "epoch": 0.8835044767059538, "grad_norm": 5.638255596160889, "learning_rate": 3.3779236624598034e-07, "loss": 0.3253, "step": 17466 }, { "epoch": 0.8835550609540189, "grad_norm": 4.921240329742432, "learning_rate": 3.3750242500789287e-07, "loss": 0.3671, "step": 17467 }, { "epoch": 0.8836056452020841, "grad_norm": 11.052828788757324, "learning_rate": 3.3721260391110734e-07, "loss": 0.4189, "step": 17468 }, { "epoch": 0.8836562294501492, "grad_norm": 6.600321292877197, "learning_rate": 3.3692290296308993e-07, "loss": 0.4712, "step": 17469 }, { "epoch": 0.8837068136982144, "grad_norm": 5.6392340660095215, "learning_rate": 3.3663332217130585e-07, "loss": 0.3696, "step": 17470 }, { "epoch": 0.8837573979462795, "grad_norm": 6.577728271484375, "learning_rate": 3.36343861543218e-07, "loss": 0.3764, "step": 17471 }, { "epoch": 0.8838079821943446, "grad_norm": 8.419317245483398, "learning_rate": 3.360545210862837e-07, "loss": 0.3416, "step": 17472 }, { "epoch": 0.8838585664424098, "grad_norm": 13.844208717346191, "learning_rate": 3.357653008079598e-07, "loss": 0.484, "step": 17473 }, { "epoch": 0.8839091506904749, "grad_norm": 13.584714889526367, "learning_rate": 3.3547620071569754e-07, "loss": 0.5569, "step": 17474 }, { "epoch": 0.8839597349385402, "grad_norm": 6.730055332183838, "learning_rate": 3.3518722081694597e-07, "loss": 0.4571, "step": 17475 }, { "epoch": 0.8840103191866053, "grad_norm": 6.1277360916137695, "learning_rate": 3.3489836111915366e-07, "loss": 0.3263, "step": 17476 }, { "epoch": 0.8840609034346705, "grad_norm": 9.450133323669434, "learning_rate": 3.346096216297623e-07, "loss": 0.4641, "step": 17477 }, { "epoch": 0.8841114876827356, "grad_norm": 7.626999378204346, "learning_rate": 3.34321002356211e-07, "loss": 0.3157, "step": 17478 }, { "epoch": 0.8841620719308008, "grad_norm": 7.018676280975342, "learning_rate": 3.340325033059394e-07, "loss": 0.3724, "step": 17479 }, { "epoch": 0.8842126561788659, "grad_norm": 8.83714771270752, "learning_rate": 3.337441244863798e-07, "loss": 0.3626, "step": 17480 }, { "epoch": 0.884263240426931, "grad_norm": 6.33644962310791, "learning_rate": 3.334558659049619e-07, "loss": 0.3748, "step": 17481 }, { "epoch": 0.8843138246749962, "grad_norm": 4.960544109344482, "learning_rate": 3.3316772756911643e-07, "loss": 0.3509, "step": 17482 }, { "epoch": 0.8843644089230613, "grad_norm": 7.154829978942871, "learning_rate": 3.328797094862657e-07, "loss": 0.3517, "step": 17483 }, { "epoch": 0.8844149931711265, "grad_norm": 5.922208309173584, "learning_rate": 3.3259181166383157e-07, "loss": 0.3513, "step": 17484 }, { "epoch": 0.8844655774191916, "grad_norm": 7.257073879241943, "learning_rate": 3.3230403410923316e-07, "loss": 0.2588, "step": 17485 }, { "epoch": 0.8845161616672568, "grad_norm": 7.805910587310791, "learning_rate": 3.320163768298862e-07, "loss": 0.4342, "step": 17486 }, { "epoch": 0.8845667459153219, "grad_norm": 5.868521690368652, "learning_rate": 3.3172883983320084e-07, "loss": 0.2643, "step": 17487 }, { "epoch": 0.8846173301633872, "grad_norm": 13.810747146606445, "learning_rate": 3.314414231265883e-07, "loss": 0.4975, "step": 17488 }, { "epoch": 0.8846679144114523, "grad_norm": 11.107610702514648, "learning_rate": 3.3115412671745394e-07, "loss": 0.339, "step": 17489 }, { "epoch": 0.8847184986595175, "grad_norm": 10.281706809997559, "learning_rate": 3.308669506131995e-07, "loss": 0.4349, "step": 17490 }, { "epoch": 0.8847690829075826, "grad_norm": 6.613672733306885, "learning_rate": 3.305798948212269e-07, "loss": 0.4138, "step": 17491 }, { "epoch": 0.8848196671556477, "grad_norm": 9.231104850769043, "learning_rate": 3.302929593489323e-07, "loss": 0.4386, "step": 17492 }, { "epoch": 0.8848702514037129, "grad_norm": 9.965348243713379, "learning_rate": 3.3000614420370726e-07, "loss": 0.3764, "step": 17493 }, { "epoch": 0.884920835651778, "grad_norm": 7.0465474128723145, "learning_rate": 3.2971944939294566e-07, "loss": 0.3074, "step": 17494 }, { "epoch": 0.8849714198998432, "grad_norm": 6.965689659118652, "learning_rate": 3.2943287492403274e-07, "loss": 0.3647, "step": 17495 }, { "epoch": 0.8850220041479083, "grad_norm": 5.7112202644348145, "learning_rate": 3.2914642080435264e-07, "loss": 0.3932, "step": 17496 }, { "epoch": 0.8850725883959735, "grad_norm": 11.463969230651855, "learning_rate": 3.288600870412878e-07, "loss": 0.5135, "step": 17497 }, { "epoch": 0.8851231726440386, "grad_norm": 9.953102111816406, "learning_rate": 3.2857387364221615e-07, "loss": 0.5279, "step": 17498 }, { "epoch": 0.8851737568921038, "grad_norm": 7.628682613372803, "learning_rate": 3.282877806145124e-07, "loss": 0.4677, "step": 17499 }, { "epoch": 0.8852243411401689, "grad_norm": 7.984451770782471, "learning_rate": 3.280018079655484e-07, "loss": 0.3664, "step": 17500 }, { "epoch": 0.8852749253882342, "grad_norm": 7.342822551727295, "learning_rate": 3.277159557026932e-07, "loss": 0.4191, "step": 17501 }, { "epoch": 0.8853255096362993, "grad_norm": 8.74101448059082, "learning_rate": 3.2743022383331157e-07, "loss": 0.3952, "step": 17502 }, { "epoch": 0.8853760938843644, "grad_norm": 8.741275787353516, "learning_rate": 3.271446123647676e-07, "loss": 0.4496, "step": 17503 }, { "epoch": 0.8854266781324296, "grad_norm": 8.840791702270508, "learning_rate": 3.2685912130441977e-07, "loss": 0.4168, "step": 17504 }, { "epoch": 0.8854772623804947, "grad_norm": 7.2317795753479, "learning_rate": 3.2657375065962504e-07, "loss": 0.4123, "step": 17505 }, { "epoch": 0.8855278466285599, "grad_norm": 12.555794715881348, "learning_rate": 3.262885004377364e-07, "loss": 0.5262, "step": 17506 }, { "epoch": 0.885578430876625, "grad_norm": 6.04764986038208, "learning_rate": 3.2600337064610464e-07, "loss": 0.3827, "step": 17507 }, { "epoch": 0.8856290151246902, "grad_norm": 5.964904308319092, "learning_rate": 3.2571836129207556e-07, "loss": 0.4359, "step": 17508 }, { "epoch": 0.8856795993727553, "grad_norm": 9.133825302124023, "learning_rate": 3.2543347238299496e-07, "loss": 0.4006, "step": 17509 }, { "epoch": 0.8857301836208205, "grad_norm": 8.767104148864746, "learning_rate": 3.2514870392620303e-07, "loss": 0.4219, "step": 17510 }, { "epoch": 0.8857807678688856, "grad_norm": 10.855141639709473, "learning_rate": 3.2486405592903723e-07, "loss": 0.5027, "step": 17511 }, { "epoch": 0.8858313521169507, "grad_norm": 11.20129680633545, "learning_rate": 3.2457952839883234e-07, "loss": 0.4374, "step": 17512 }, { "epoch": 0.8858819363650159, "grad_norm": 7.1495280265808105, "learning_rate": 3.2429512134291906e-07, "loss": 0.4316, "step": 17513 }, { "epoch": 0.885932520613081, "grad_norm": 6.076789379119873, "learning_rate": 3.240108347686277e-07, "loss": 0.4175, "step": 17514 }, { "epoch": 0.8859831048611463, "grad_norm": 9.281898498535156, "learning_rate": 3.237266686832824e-07, "loss": 0.431, "step": 17515 }, { "epoch": 0.8860336891092114, "grad_norm": 9.409396171569824, "learning_rate": 3.234426230942056e-07, "loss": 0.2967, "step": 17516 }, { "epoch": 0.8860842733572766, "grad_norm": 5.607451915740967, "learning_rate": 3.23158698008717e-07, "loss": 0.3904, "step": 17517 }, { "epoch": 0.8861348576053417, "grad_norm": 12.427224159240723, "learning_rate": 3.228748934341325e-07, "loss": 0.4837, "step": 17518 }, { "epoch": 0.8861854418534069, "grad_norm": 9.930842399597168, "learning_rate": 3.225912093777639e-07, "loss": 0.3557, "step": 17519 }, { "epoch": 0.886236026101472, "grad_norm": 9.587959289550781, "learning_rate": 3.2230764584692263e-07, "loss": 0.3363, "step": 17520 }, { "epoch": 0.8862866103495372, "grad_norm": 7.965482711791992, "learning_rate": 3.220242028489151e-07, "loss": 0.3132, "step": 17521 }, { "epoch": 0.8863371945976023, "grad_norm": 13.71250057220459, "learning_rate": 3.217408803910438e-07, "loss": 0.5163, "step": 17522 }, { "epoch": 0.8863877788456674, "grad_norm": 11.237529754638672, "learning_rate": 3.2145767848061117e-07, "loss": 0.4975, "step": 17523 }, { "epoch": 0.8864383630937326, "grad_norm": 5.700889587402344, "learning_rate": 3.211745971249142e-07, "loss": 0.3001, "step": 17524 }, { "epoch": 0.8864889473417977, "grad_norm": 10.01174545288086, "learning_rate": 3.208916363312453e-07, "loss": 0.4007, "step": 17525 }, { "epoch": 0.8865395315898629, "grad_norm": 5.443991661071777, "learning_rate": 3.206087961068971e-07, "loss": 0.3859, "step": 17526 }, { "epoch": 0.886590115837928, "grad_norm": 9.368130683898926, "learning_rate": 3.2032607645915815e-07, "loss": 0.472, "step": 17527 }, { "epoch": 0.8866407000859933, "grad_norm": 5.62808895111084, "learning_rate": 3.2004347739531204e-07, "loss": 0.3465, "step": 17528 }, { "epoch": 0.8866912843340584, "grad_norm": 5.769432544708252, "learning_rate": 3.1976099892264246e-07, "loss": 0.3095, "step": 17529 }, { "epoch": 0.8867418685821236, "grad_norm": 7.605626583099365, "learning_rate": 3.194786410484274e-07, "loss": 0.3744, "step": 17530 }, { "epoch": 0.8867924528301887, "grad_norm": 12.21899127960205, "learning_rate": 3.191964037799411e-07, "loss": 0.3956, "step": 17531 }, { "epoch": 0.8868430370782538, "grad_norm": 8.602060317993164, "learning_rate": 3.189142871244588e-07, "loss": 0.4995, "step": 17532 }, { "epoch": 0.886893621326319, "grad_norm": 4.990985870361328, "learning_rate": 3.1863229108924867e-07, "loss": 0.431, "step": 17533 }, { "epoch": 0.8869442055743841, "grad_norm": 16.05392074584961, "learning_rate": 3.1835041568157587e-07, "loss": 0.4988, "step": 17534 }, { "epoch": 0.8869947898224493, "grad_norm": 13.872547149658203, "learning_rate": 3.1806866090870526e-07, "loss": 0.4147, "step": 17535 }, { "epoch": 0.8870453740705144, "grad_norm": 8.815629005432129, "learning_rate": 3.1778702677789706e-07, "loss": 0.4391, "step": 17536 }, { "epoch": 0.8870959583185796, "grad_norm": 7.887054443359375, "learning_rate": 3.1750551329640776e-07, "loss": 0.3696, "step": 17537 }, { "epoch": 0.8871465425666447, "grad_norm": 12.828302383422852, "learning_rate": 3.17224120471491e-07, "loss": 0.528, "step": 17538 }, { "epoch": 0.8871971268147099, "grad_norm": 16.033660888671875, "learning_rate": 3.169428483103981e-07, "loss": 0.4538, "step": 17539 }, { "epoch": 0.887247711062775, "grad_norm": 8.93703556060791, "learning_rate": 3.1666169682037565e-07, "loss": 0.3961, "step": 17540 }, { "epoch": 0.8872982953108403, "grad_norm": 7.534784317016602, "learning_rate": 3.163806660086699e-07, "loss": 0.2865, "step": 17541 }, { "epoch": 0.8873488795589054, "grad_norm": 13.619133949279785, "learning_rate": 3.160997558825218e-07, "loss": 0.4735, "step": 17542 }, { "epoch": 0.8873994638069705, "grad_norm": 7.681865692138672, "learning_rate": 3.1581896644916843e-07, "loss": 0.4271, "step": 17543 }, { "epoch": 0.8874500480550357, "grad_norm": 7.745022296905518, "learning_rate": 3.1553829771584665e-07, "loss": 0.4241, "step": 17544 }, { "epoch": 0.8875006323031008, "grad_norm": 10.064083099365234, "learning_rate": 3.152577496897885e-07, "loss": 0.3319, "step": 17545 }, { "epoch": 0.887551216551166, "grad_norm": 6.372334003448486, "learning_rate": 3.149773223782221e-07, "loss": 0.4077, "step": 17546 }, { "epoch": 0.8876018007992311, "grad_norm": 5.8462114334106445, "learning_rate": 3.1469701578837385e-07, "loss": 0.407, "step": 17547 }, { "epoch": 0.8876523850472963, "grad_norm": 6.024744987487793, "learning_rate": 3.144168299274675e-07, "loss": 0.4186, "step": 17548 }, { "epoch": 0.8877029692953614, "grad_norm": 6.250380992889404, "learning_rate": 3.1413676480272046e-07, "loss": 0.2789, "step": 17549 }, { "epoch": 0.8877535535434266, "grad_norm": 8.626372337341309, "learning_rate": 3.1385682042135266e-07, "loss": 0.4302, "step": 17550 }, { "epoch": 0.8878041377914917, "grad_norm": 8.047857284545898, "learning_rate": 3.1357699679057385e-07, "loss": 0.4848, "step": 17551 }, { "epoch": 0.8878547220395568, "grad_norm": 10.007791519165039, "learning_rate": 3.132972939175971e-07, "loss": 0.2664, "step": 17552 }, { "epoch": 0.887905306287622, "grad_norm": 7.401401042938232, "learning_rate": 3.130177118096289e-07, "loss": 0.3568, "step": 17553 }, { "epoch": 0.8879558905356872, "grad_norm": 6.08918571472168, "learning_rate": 3.127382504738724e-07, "loss": 0.4366, "step": 17554 }, { "epoch": 0.8880064747837524, "grad_norm": 6.174192428588867, "learning_rate": 3.124589099175307e-07, "loss": 0.4257, "step": 17555 }, { "epoch": 0.8880570590318175, "grad_norm": 7.8390278816223145, "learning_rate": 3.1217969014780023e-07, "loss": 0.4457, "step": 17556 }, { "epoch": 0.8881076432798827, "grad_norm": 19.946535110473633, "learning_rate": 3.1190059117187586e-07, "loss": 0.6836, "step": 17557 }, { "epoch": 0.8881582275279478, "grad_norm": 6.082441329956055, "learning_rate": 3.1162161299695016e-07, "loss": 0.3444, "step": 17558 }, { "epoch": 0.888208811776013, "grad_norm": 5.309983730316162, "learning_rate": 3.1134275563021064e-07, "loss": 0.4042, "step": 17559 }, { "epoch": 0.8882593960240781, "grad_norm": 11.692712783813477, "learning_rate": 3.110640190788433e-07, "loss": 0.3959, "step": 17560 }, { "epoch": 0.8883099802721433, "grad_norm": 9.422577857971191, "learning_rate": 3.107854033500307e-07, "loss": 0.4448, "step": 17561 }, { "epoch": 0.8883605645202084, "grad_norm": 12.76952838897705, "learning_rate": 3.1050690845095266e-07, "loss": 0.5295, "step": 17562 }, { "epoch": 0.8884111487682735, "grad_norm": 5.45051383972168, "learning_rate": 3.1022853438878396e-07, "loss": 0.3336, "step": 17563 }, { "epoch": 0.8884617330163387, "grad_norm": 8.41480827331543, "learning_rate": 3.099502811706984e-07, "loss": 0.4369, "step": 17564 }, { "epoch": 0.8885123172644038, "grad_norm": 7.1045379638671875, "learning_rate": 3.0967214880386566e-07, "loss": 0.42, "step": 17565 }, { "epoch": 0.888562901512469, "grad_norm": 9.42802619934082, "learning_rate": 3.093941372954523e-07, "loss": 0.3992, "step": 17566 }, { "epoch": 0.8886134857605341, "grad_norm": 7.461180686950684, "learning_rate": 3.091162466526226e-07, "loss": 0.2717, "step": 17567 }, { "epoch": 0.8886640700085994, "grad_norm": 9.17998218536377, "learning_rate": 3.08838476882537e-07, "loss": 0.4209, "step": 17568 }, { "epoch": 0.8887146542566645, "grad_norm": 10.975106239318848, "learning_rate": 3.0856082799235187e-07, "loss": 0.4897, "step": 17569 }, { "epoch": 0.8887652385047297, "grad_norm": 7.20688009262085, "learning_rate": 3.082832999892238e-07, "loss": 0.5298, "step": 17570 }, { "epoch": 0.8888158227527948, "grad_norm": 5.607118606567383, "learning_rate": 3.080058928803026e-07, "loss": 0.3433, "step": 17571 }, { "epoch": 0.88886640700086, "grad_norm": 7.099032878875732, "learning_rate": 3.077286066727353e-07, "loss": 0.3857, "step": 17572 }, { "epoch": 0.8889169912489251, "grad_norm": 8.489583969116211, "learning_rate": 3.07451441373669e-07, "loss": 0.3466, "step": 17573 }, { "epoch": 0.8889675754969902, "grad_norm": 4.622477054595947, "learning_rate": 3.0717439699024465e-07, "loss": 0.4354, "step": 17574 }, { "epoch": 0.8890181597450554, "grad_norm": 7.235716342926025, "learning_rate": 3.0689747352960044e-07, "loss": 0.3594, "step": 17575 }, { "epoch": 0.8890687439931205, "grad_norm": 18.056869506835938, "learning_rate": 3.0662067099887395e-07, "loss": 0.5063, "step": 17576 }, { "epoch": 0.8891193282411857, "grad_norm": 6.2798871994018555, "learning_rate": 3.063439894051956e-07, "loss": 0.4034, "step": 17577 }, { "epoch": 0.8891699124892508, "grad_norm": 7.93687105178833, "learning_rate": 3.060674287556953e-07, "loss": 0.3835, "step": 17578 }, { "epoch": 0.889220496737316, "grad_norm": 17.601131439208984, "learning_rate": 3.0579098905750007e-07, "loss": 0.4185, "step": 17579 }, { "epoch": 0.8892710809853811, "grad_norm": 5.80485725402832, "learning_rate": 3.0551467031773253e-07, "loss": 0.3818, "step": 17580 }, { "epoch": 0.8893216652334464, "grad_norm": 10.188685417175293, "learning_rate": 3.0523847254351205e-07, "loss": 0.4331, "step": 17581 }, { "epoch": 0.8893722494815115, "grad_norm": 9.231659889221191, "learning_rate": 3.049623957419573e-07, "loss": 0.387, "step": 17582 }, { "epoch": 0.8894228337295766, "grad_norm": 9.137588500976562, "learning_rate": 3.04686439920181e-07, "loss": 0.3111, "step": 17583 }, { "epoch": 0.8894734179776418, "grad_norm": 9.3924560546875, "learning_rate": 3.044106050852935e-07, "loss": 0.3888, "step": 17584 }, { "epoch": 0.8895240022257069, "grad_norm": 13.556591033935547, "learning_rate": 3.0413489124440423e-07, "loss": 0.4569, "step": 17585 }, { "epoch": 0.8895745864737721, "grad_norm": 5.900918006896973, "learning_rate": 3.038592984046157e-07, "loss": 0.4392, "step": 17586 }, { "epoch": 0.8896251707218372, "grad_norm": 6.846848964691162, "learning_rate": 3.035838265730295e-07, "loss": 0.4587, "step": 17587 }, { "epoch": 0.8896757549699024, "grad_norm": 8.038357734680176, "learning_rate": 3.0330847575674507e-07, "loss": 0.3805, "step": 17588 }, { "epoch": 0.8897263392179675, "grad_norm": 10.392329216003418, "learning_rate": 3.030332459628571e-07, "loss": 0.3071, "step": 17589 }, { "epoch": 0.8897769234660327, "grad_norm": 10.424785614013672, "learning_rate": 3.027581371984567e-07, "loss": 0.3486, "step": 17590 }, { "epoch": 0.8898275077140978, "grad_norm": 8.815296173095703, "learning_rate": 3.024831494706337e-07, "loss": 0.343, "step": 17591 }, { "epoch": 0.8898780919621629, "grad_norm": 12.11290168762207, "learning_rate": 3.022082827864731e-07, "loss": 0.4201, "step": 17592 }, { "epoch": 0.8899286762102281, "grad_norm": 8.541825294494629, "learning_rate": 3.019335371530585e-07, "loss": 0.3711, "step": 17593 }, { "epoch": 0.8899792604582933, "grad_norm": 14.691267967224121, "learning_rate": 3.0165891257746883e-07, "loss": 0.5228, "step": 17594 }, { "epoch": 0.8900298447063585, "grad_norm": 7.178277015686035, "learning_rate": 3.0138440906678e-07, "loss": 0.3361, "step": 17595 }, { "epoch": 0.8900804289544236, "grad_norm": 4.1506476402282715, "learning_rate": 3.01110026628067e-07, "loss": 0.3401, "step": 17596 }, { "epoch": 0.8901310132024888, "grad_norm": 7.306312561035156, "learning_rate": 3.008357652683985e-07, "loss": 0.3628, "step": 17597 }, { "epoch": 0.8901815974505539, "grad_norm": 11.348926544189453, "learning_rate": 3.005616249948418e-07, "loss": 0.4926, "step": 17598 }, { "epoch": 0.8902321816986191, "grad_norm": 4.976326942443848, "learning_rate": 3.002876058144605e-07, "loss": 0.3959, "step": 17599 }, { "epoch": 0.8902827659466842, "grad_norm": 12.988900184631348, "learning_rate": 3.0001370773431684e-07, "loss": 0.5551, "step": 17600 }, { "epoch": 0.8903333501947494, "grad_norm": 7.433432579040527, "learning_rate": 2.9973993076146734e-07, "loss": 0.3807, "step": 17601 }, { "epoch": 0.8903839344428145, "grad_norm": 11.657955169677734, "learning_rate": 2.994662749029664e-07, "loss": 0.538, "step": 17602 }, { "epoch": 0.8904345186908796, "grad_norm": 8.454158782958984, "learning_rate": 2.991927401658662e-07, "loss": 0.343, "step": 17603 }, { "epoch": 0.8904851029389448, "grad_norm": 7.493853569030762, "learning_rate": 2.9891932655721435e-07, "loss": 0.4214, "step": 17604 }, { "epoch": 0.8905356871870099, "grad_norm": 10.58090877532959, "learning_rate": 2.986460340840569e-07, "loss": 0.3787, "step": 17605 }, { "epoch": 0.8905862714350751, "grad_norm": 9.579877853393555, "learning_rate": 2.983728627534355e-07, "loss": 0.501, "step": 17606 }, { "epoch": 0.8906368556831402, "grad_norm": 18.469818115234375, "learning_rate": 2.980998125723883e-07, "loss": 0.5266, "step": 17607 }, { "epoch": 0.8906874399312055, "grad_norm": 14.545122146606445, "learning_rate": 2.9782688354795306e-07, "loss": 0.5146, "step": 17608 }, { "epoch": 0.8907380241792706, "grad_norm": 23.071937561035156, "learning_rate": 2.975540756871609e-07, "loss": 0.4037, "step": 17609 }, { "epoch": 0.8907886084273358, "grad_norm": 10.961000442504883, "learning_rate": 2.972813889970416e-07, "loss": 0.4811, "step": 17610 }, { "epoch": 0.8908391926754009, "grad_norm": 9.729729652404785, "learning_rate": 2.9700882348462244e-07, "loss": 0.3844, "step": 17611 }, { "epoch": 0.890889776923466, "grad_norm": 8.38817024230957, "learning_rate": 2.9673637915692666e-07, "loss": 0.4078, "step": 17612 }, { "epoch": 0.8909403611715312, "grad_norm": 10.161128044128418, "learning_rate": 2.9646405602097306e-07, "loss": 0.4345, "step": 17613 }, { "epoch": 0.8909909454195963, "grad_norm": 6.373536586761475, "learning_rate": 2.9619185408378104e-07, "loss": 0.379, "step": 17614 }, { "epoch": 0.8910415296676615, "grad_norm": 6.10105562210083, "learning_rate": 2.959197733523633e-07, "loss": 0.3775, "step": 17615 }, { "epoch": 0.8910921139157266, "grad_norm": 7.642101764678955, "learning_rate": 2.9564781383372933e-07, "loss": 0.3272, "step": 17616 }, { "epoch": 0.8911426981637918, "grad_norm": 6.161354064941406, "learning_rate": 2.953759755348895e-07, "loss": 0.321, "step": 17617 }, { "epoch": 0.8911932824118569, "grad_norm": 5.533666133880615, "learning_rate": 2.951042584628472e-07, "loss": 0.2626, "step": 17618 }, { "epoch": 0.8912438666599221, "grad_norm": 9.417535781860352, "learning_rate": 2.9483266262460286e-07, "loss": 0.4379, "step": 17619 }, { "epoch": 0.8912944509079872, "grad_norm": 8.968408584594727, "learning_rate": 2.945611880271571e-07, "loss": 0.3776, "step": 17620 }, { "epoch": 0.8913450351560525, "grad_norm": 9.791165351867676, "learning_rate": 2.942898346775036e-07, "loss": 0.4062, "step": 17621 }, { "epoch": 0.8913956194041176, "grad_norm": 6.424278736114502, "learning_rate": 2.9401860258263415e-07, "loss": 0.2906, "step": 17622 }, { "epoch": 0.8914462036521827, "grad_norm": 4.835197925567627, "learning_rate": 2.9374749174953977e-07, "loss": 0.334, "step": 17623 }, { "epoch": 0.8914967879002479, "grad_norm": 10.420402526855469, "learning_rate": 2.934765021852043e-07, "loss": 0.423, "step": 17624 }, { "epoch": 0.891547372148313, "grad_norm": 5.233516216278076, "learning_rate": 2.9320563389661106e-07, "loss": 0.3397, "step": 17625 }, { "epoch": 0.8915979563963782, "grad_norm": 9.144271850585938, "learning_rate": 2.9293488689074054e-07, "loss": 0.4222, "step": 17626 }, { "epoch": 0.8916485406444433, "grad_norm": 8.393229484558105, "learning_rate": 2.926642611745684e-07, "loss": 0.3863, "step": 17627 }, { "epoch": 0.8916991248925085, "grad_norm": 10.134125709533691, "learning_rate": 2.923937567550678e-07, "loss": 0.4251, "step": 17628 }, { "epoch": 0.8917497091405736, "grad_norm": 7.379415988922119, "learning_rate": 2.921233736392098e-07, "loss": 0.3016, "step": 17629 }, { "epoch": 0.8918002933886388, "grad_norm": 9.129243850708008, "learning_rate": 2.918531118339601e-07, "loss": 0.449, "step": 17630 }, { "epoch": 0.8918508776367039, "grad_norm": 5.8683881759643555, "learning_rate": 2.915829713462842e-07, "loss": 0.3502, "step": 17631 }, { "epoch": 0.891901461884769, "grad_norm": 5.849881172180176, "learning_rate": 2.9131295218314313e-07, "loss": 0.4414, "step": 17632 }, { "epoch": 0.8919520461328342, "grad_norm": 39.440147399902344, "learning_rate": 2.9104305435149247e-07, "loss": 0.5132, "step": 17633 }, { "epoch": 0.8920026303808994, "grad_norm": 7.419743061065674, "learning_rate": 2.9077327785828946e-07, "loss": 0.3467, "step": 17634 }, { "epoch": 0.8920532146289646, "grad_norm": 10.390278816223145, "learning_rate": 2.9050362271048406e-07, "loss": 0.4647, "step": 17635 }, { "epoch": 0.8921037988770297, "grad_norm": 5.469418048858643, "learning_rate": 2.9023408891502516e-07, "loss": 0.395, "step": 17636 }, { "epoch": 0.8921543831250949, "grad_norm": 8.222905158996582, "learning_rate": 2.8996467647885716e-07, "loss": 0.3499, "step": 17637 }, { "epoch": 0.89220496737316, "grad_norm": 8.19165325164795, "learning_rate": 2.8969538540892294e-07, "loss": 0.3825, "step": 17638 }, { "epoch": 0.8922555516212252, "grad_norm": 6.778266429901123, "learning_rate": 2.894262157121619e-07, "loss": 0.3473, "step": 17639 }, { "epoch": 0.8923061358692903, "grad_norm": 6.691285133361816, "learning_rate": 2.8915716739550956e-07, "loss": 0.3668, "step": 17640 }, { "epoch": 0.8923567201173555, "grad_norm": 6.104289531707764, "learning_rate": 2.8888824046589825e-07, "loss": 0.3599, "step": 17641 }, { "epoch": 0.8924073043654206, "grad_norm": 7.85158109664917, "learning_rate": 2.886194349302568e-07, "loss": 0.2999, "step": 17642 }, { "epoch": 0.8924578886134857, "grad_norm": 8.80784797668457, "learning_rate": 2.883507507955136e-07, "loss": 0.4679, "step": 17643 }, { "epoch": 0.8925084728615509, "grad_norm": 7.52947998046875, "learning_rate": 2.880821880685908e-07, "loss": 0.3204, "step": 17644 }, { "epoch": 0.892559057109616, "grad_norm": 8.186750411987305, "learning_rate": 2.878137467564085e-07, "loss": 0.3187, "step": 17645 }, { "epoch": 0.8926096413576812, "grad_norm": 7.570182800292969, "learning_rate": 2.875454268658845e-07, "loss": 0.3698, "step": 17646 }, { "epoch": 0.8926602256057463, "grad_norm": 8.951333045959473, "learning_rate": 2.8727722840393267e-07, "loss": 0.4219, "step": 17647 }, { "epoch": 0.8927108098538116, "grad_norm": 7.3190107345581055, "learning_rate": 2.8700915137746257e-07, "loss": 0.469, "step": 17648 }, { "epoch": 0.8927613941018767, "grad_norm": 9.71359920501709, "learning_rate": 2.867411957933841e-07, "loss": 0.487, "step": 17649 }, { "epoch": 0.8928119783499419, "grad_norm": 6.339959621429443, "learning_rate": 2.8647336165860016e-07, "loss": 0.3913, "step": 17650 }, { "epoch": 0.892862562598007, "grad_norm": 9.994606018066406, "learning_rate": 2.862056489800119e-07, "loss": 0.3861, "step": 17651 }, { "epoch": 0.8929131468460721, "grad_norm": 7.305271148681641, "learning_rate": 2.859380577645188e-07, "loss": 0.4767, "step": 17652 }, { "epoch": 0.8929637310941373, "grad_norm": 9.081172943115234, "learning_rate": 2.856705880190169e-07, "loss": 0.4656, "step": 17653 }, { "epoch": 0.8930143153422024, "grad_norm": 10.775485038757324, "learning_rate": 2.854032397503953e-07, "loss": 0.4676, "step": 17654 }, { "epoch": 0.8930648995902676, "grad_norm": 6.259253025054932, "learning_rate": 2.8513601296554504e-07, "loss": 0.3131, "step": 17655 }, { "epoch": 0.8931154838383327, "grad_norm": 8.586827278137207, "learning_rate": 2.848689076713518e-07, "loss": 0.4458, "step": 17656 }, { "epoch": 0.8931660680863979, "grad_norm": 10.902621269226074, "learning_rate": 2.846019238746972e-07, "loss": 0.4518, "step": 17657 }, { "epoch": 0.893216652334463, "grad_norm": 9.431492805480957, "learning_rate": 2.8433506158246194e-07, "loss": 0.389, "step": 17658 }, { "epoch": 0.8932672365825282, "grad_norm": 8.896146774291992, "learning_rate": 2.8406832080152215e-07, "loss": 0.3531, "step": 17659 }, { "epoch": 0.8933178208305933, "grad_norm": 9.68135929107666, "learning_rate": 2.8380170153874955e-07, "loss": 0.3367, "step": 17660 }, { "epoch": 0.8933684050786586, "grad_norm": 5.330318450927734, "learning_rate": 2.8353520380101696e-07, "loss": 0.3466, "step": 17661 }, { "epoch": 0.8934189893267237, "grad_norm": 8.508614540100098, "learning_rate": 2.8326882759518946e-07, "loss": 0.4104, "step": 17662 }, { "epoch": 0.8934695735747888, "grad_norm": 9.602169036865234, "learning_rate": 2.8300257292813105e-07, "loss": 0.422, "step": 17663 }, { "epoch": 0.893520157822854, "grad_norm": 14.747292518615723, "learning_rate": 2.8273643980670396e-07, "loss": 0.4269, "step": 17664 }, { "epoch": 0.8935707420709191, "grad_norm": 8.220848083496094, "learning_rate": 2.824704282377644e-07, "loss": 0.3623, "step": 17665 }, { "epoch": 0.8936213263189843, "grad_norm": 12.83303451538086, "learning_rate": 2.822045382281674e-07, "loss": 0.5522, "step": 17666 }, { "epoch": 0.8936719105670494, "grad_norm": 12.798680305480957, "learning_rate": 2.8193876978476367e-07, "loss": 0.5134, "step": 17667 }, { "epoch": 0.8937224948151146, "grad_norm": 11.1115140914917, "learning_rate": 2.81673122914401e-07, "loss": 0.4724, "step": 17668 }, { "epoch": 0.8937730790631797, "grad_norm": 10.130824089050293, "learning_rate": 2.814075976239267e-07, "loss": 0.4652, "step": 17669 }, { "epoch": 0.8938236633112449, "grad_norm": 9.0424165725708, "learning_rate": 2.811421939201808e-07, "loss": 0.5271, "step": 17670 }, { "epoch": 0.89387424755931, "grad_norm": 6.612126350402832, "learning_rate": 2.8087691181000297e-07, "loss": 0.453, "step": 17671 }, { "epoch": 0.8939248318073751, "grad_norm": 5.153294563293457, "learning_rate": 2.8061175130022767e-07, "loss": 0.3915, "step": 17672 }, { "epoch": 0.8939754160554403, "grad_norm": 6.172596454620361, "learning_rate": 2.8034671239768883e-07, "loss": 0.3569, "step": 17673 }, { "epoch": 0.8940260003035055, "grad_norm": 8.201854705810547, "learning_rate": 2.8008179510921553e-07, "loss": 0.3953, "step": 17674 }, { "epoch": 0.8940765845515707, "grad_norm": 11.241788864135742, "learning_rate": 2.7981699944163334e-07, "loss": 0.3224, "step": 17675 }, { "epoch": 0.8941271687996358, "grad_norm": 7.945173740386963, "learning_rate": 2.7955232540176626e-07, "loss": 0.3661, "step": 17676 }, { "epoch": 0.894177753047701, "grad_norm": 7.027385234832764, "learning_rate": 2.792877729964344e-07, "loss": 0.3645, "step": 17677 }, { "epoch": 0.8942283372957661, "grad_norm": 7.745353698730469, "learning_rate": 2.7902334223245286e-07, "loss": 0.4248, "step": 17678 }, { "epoch": 0.8942789215438313, "grad_norm": 19.999797821044922, "learning_rate": 2.78759033116639e-07, "loss": 0.4015, "step": 17679 }, { "epoch": 0.8943295057918964, "grad_norm": 7.81105375289917, "learning_rate": 2.78494845655799e-07, "loss": 0.36, "step": 17680 }, { "epoch": 0.8943800900399616, "grad_norm": 6.309897422790527, "learning_rate": 2.7823077985674406e-07, "loss": 0.4, "step": 17681 }, { "epoch": 0.8944306742880267, "grad_norm": 9.062811851501465, "learning_rate": 2.77966835726276e-07, "loss": 0.3495, "step": 17682 }, { "epoch": 0.8944812585360918, "grad_norm": 5.882638454437256, "learning_rate": 2.7770301327119663e-07, "loss": 0.342, "step": 17683 }, { "epoch": 0.894531842784157, "grad_norm": 8.400668144226074, "learning_rate": 2.774393124983055e-07, "loss": 0.4753, "step": 17684 }, { "epoch": 0.8945824270322221, "grad_norm": 8.915263175964355, "learning_rate": 2.7717573341439597e-07, "loss": 0.3557, "step": 17685 }, { "epoch": 0.8946330112802873, "grad_norm": 6.115019798278809, "learning_rate": 2.7691227602625993e-07, "loss": 0.4135, "step": 17686 }, { "epoch": 0.8946835955283524, "grad_norm": 13.226327896118164, "learning_rate": 2.766489403406869e-07, "loss": 0.4136, "step": 17687 }, { "epoch": 0.8947341797764177, "grad_norm": 13.264886856079102, "learning_rate": 2.763857263644615e-07, "loss": 0.4315, "step": 17688 }, { "epoch": 0.8947847640244828, "grad_norm": 7.386028289794922, "learning_rate": 2.7612263410436604e-07, "loss": 0.4465, "step": 17689 }, { "epoch": 0.894835348272548, "grad_norm": 8.039379119873047, "learning_rate": 2.758596635671812e-07, "loss": 0.3881, "step": 17690 }, { "epoch": 0.8948859325206131, "grad_norm": 5.358969688415527, "learning_rate": 2.755968147596827e-07, "loss": 0.4049, "step": 17691 }, { "epoch": 0.8949365167686782, "grad_norm": 8.520343780517578, "learning_rate": 2.753340876886418e-07, "loss": 0.3589, "step": 17692 }, { "epoch": 0.8949871010167434, "grad_norm": 10.432472229003906, "learning_rate": 2.750714823608297e-07, "loss": 0.4085, "step": 17693 }, { "epoch": 0.8950376852648085, "grad_norm": 8.094757080078125, "learning_rate": 2.7480899878301327e-07, "loss": 0.4836, "step": 17694 }, { "epoch": 0.8950882695128737, "grad_norm": 11.850485801696777, "learning_rate": 2.7454663696195484e-07, "loss": 0.5111, "step": 17695 }, { "epoch": 0.8951388537609388, "grad_norm": 9.16458511352539, "learning_rate": 2.742843969044162e-07, "loss": 0.4272, "step": 17696 }, { "epoch": 0.895189438009004, "grad_norm": 13.372824668884277, "learning_rate": 2.7402227861715426e-07, "loss": 0.4101, "step": 17697 }, { "epoch": 0.8952400222570691, "grad_norm": 7.968034267425537, "learning_rate": 2.73760282106923e-07, "loss": 0.3675, "step": 17698 }, { "epoch": 0.8952906065051343, "grad_norm": 11.552037239074707, "learning_rate": 2.734984073804736e-07, "loss": 0.3275, "step": 17699 }, { "epoch": 0.8953411907531994, "grad_norm": 6.2694172859191895, "learning_rate": 2.732366544445536e-07, "loss": 0.3381, "step": 17700 }, { "epoch": 0.8953917750012647, "grad_norm": 9.428531646728516, "learning_rate": 2.7297502330590806e-07, "loss": 0.3547, "step": 17701 }, { "epoch": 0.8954423592493298, "grad_norm": 7.779450416564941, "learning_rate": 2.7271351397127833e-07, "loss": 0.3932, "step": 17702 }, { "epoch": 0.8954929434973949, "grad_norm": 9.763192176818848, "learning_rate": 2.7245212644740393e-07, "loss": 0.4444, "step": 17703 }, { "epoch": 0.8955435277454601, "grad_norm": 9.298966407775879, "learning_rate": 2.72190860741019e-07, "loss": 0.4194, "step": 17704 }, { "epoch": 0.8955941119935252, "grad_norm": 5.308453559875488, "learning_rate": 2.7192971685885584e-07, "loss": 0.3614, "step": 17705 }, { "epoch": 0.8956446962415904, "grad_norm": 13.266030311584473, "learning_rate": 2.7166869480764303e-07, "loss": 0.4082, "step": 17706 }, { "epoch": 0.8956952804896555, "grad_norm": 5.7426066398620605, "learning_rate": 2.714077945941079e-07, "loss": 0.4057, "step": 17707 }, { "epoch": 0.8957458647377207, "grad_norm": 10.472270011901855, "learning_rate": 2.711470162249724e-07, "loss": 0.4544, "step": 17708 }, { "epoch": 0.8957964489857858, "grad_norm": 7.31559944152832, "learning_rate": 2.708863597069561e-07, "loss": 0.3577, "step": 17709 }, { "epoch": 0.895847033233851, "grad_norm": 11.771980285644531, "learning_rate": 2.7062582504677526e-07, "loss": 0.5143, "step": 17710 }, { "epoch": 0.8958976174819161, "grad_norm": 7.3578782081604, "learning_rate": 2.70365412251144e-07, "loss": 0.3177, "step": 17711 }, { "epoch": 0.8959482017299812, "grad_norm": 8.653778076171875, "learning_rate": 2.7010512132677193e-07, "loss": 0.4709, "step": 17712 }, { "epoch": 0.8959987859780464, "grad_norm": 8.998332023620605, "learning_rate": 2.698449522803659e-07, "loss": 0.4453, "step": 17713 }, { "epoch": 0.8960493702261116, "grad_norm": 10.347599029541016, "learning_rate": 2.695849051186311e-07, "loss": 0.438, "step": 17714 }, { "epoch": 0.8960999544741768, "grad_norm": 6.429070472717285, "learning_rate": 2.693249798482672e-07, "loss": 0.3413, "step": 17715 }, { "epoch": 0.8961505387222419, "grad_norm": 7.361609935760498, "learning_rate": 2.690651764759711e-07, "loss": 0.4038, "step": 17716 }, { "epoch": 0.8962011229703071, "grad_norm": 6.283020496368408, "learning_rate": 2.688054950084401e-07, "loss": 0.3195, "step": 17717 }, { "epoch": 0.8962517072183722, "grad_norm": 4.978771209716797, "learning_rate": 2.685459354523623e-07, "loss": 0.3012, "step": 17718 }, { "epoch": 0.8963022914664374, "grad_norm": 6.319355487823486, "learning_rate": 2.682864978144278e-07, "loss": 0.2625, "step": 17719 }, { "epoch": 0.8963528757145025, "grad_norm": 8.363709449768066, "learning_rate": 2.6802718210132197e-07, "loss": 0.3751, "step": 17720 }, { "epoch": 0.8964034599625677, "grad_norm": 7.512559413909912, "learning_rate": 2.6776798831972484e-07, "loss": 0.44, "step": 17721 }, { "epoch": 0.8964540442106328, "grad_norm": 6.935210227966309, "learning_rate": 2.675089164763167e-07, "loss": 0.3608, "step": 17722 }, { "epoch": 0.8965046284586979, "grad_norm": 6.066154479980469, "learning_rate": 2.6724996657777335e-07, "loss": 0.3652, "step": 17723 }, { "epoch": 0.8965552127067631, "grad_norm": 8.204095840454102, "learning_rate": 2.669911386307666e-07, "loss": 0.3497, "step": 17724 }, { "epoch": 0.8966057969548282, "grad_norm": 9.289819717407227, "learning_rate": 2.6673243264196626e-07, "loss": 0.4956, "step": 17725 }, { "epoch": 0.8966563812028934, "grad_norm": 12.525917053222656, "learning_rate": 2.664738486180385e-07, "loss": 0.4383, "step": 17726 }, { "epoch": 0.8967069654509586, "grad_norm": 10.015745162963867, "learning_rate": 2.6621538656564585e-07, "loss": 0.3481, "step": 17727 }, { "epoch": 0.8967575496990238, "grad_norm": 11.78255844116211, "learning_rate": 2.6595704649144906e-07, "loss": 0.5507, "step": 17728 }, { "epoch": 0.8968081339470889, "grad_norm": 7.422756195068359, "learning_rate": 2.656988284021045e-07, "loss": 0.3323, "step": 17729 }, { "epoch": 0.8968587181951541, "grad_norm": 7.645326614379883, "learning_rate": 2.6544073230426623e-07, "loss": 0.3094, "step": 17730 }, { "epoch": 0.8969093024432192, "grad_norm": 5.752372741699219, "learning_rate": 2.651827582045841e-07, "loss": 0.2995, "step": 17731 }, { "epoch": 0.8969598866912843, "grad_norm": 7.5938520431518555, "learning_rate": 2.649249061097059e-07, "loss": 0.3233, "step": 17732 }, { "epoch": 0.8970104709393495, "grad_norm": 9.392892837524414, "learning_rate": 2.6466717602627536e-07, "loss": 0.395, "step": 17733 }, { "epoch": 0.8970610551874146, "grad_norm": 6.252382278442383, "learning_rate": 2.644095679609343e-07, "loss": 0.3552, "step": 17734 }, { "epoch": 0.8971116394354798, "grad_norm": 11.029792785644531, "learning_rate": 2.641520819203203e-07, "loss": 0.4222, "step": 17735 }, { "epoch": 0.8971622236835449, "grad_norm": 5.122488021850586, "learning_rate": 2.638947179110679e-07, "loss": 0.2692, "step": 17736 }, { "epoch": 0.8972128079316101, "grad_norm": 9.91781997680664, "learning_rate": 2.6363747593980915e-07, "loss": 0.4623, "step": 17737 }, { "epoch": 0.8972633921796752, "grad_norm": 17.77665138244629, "learning_rate": 2.6338035601317315e-07, "loss": 0.4136, "step": 17738 }, { "epoch": 0.8973139764277404, "grad_norm": 8.555163383483887, "learning_rate": 2.6312335813778346e-07, "loss": 0.4017, "step": 17739 }, { "epoch": 0.8973645606758055, "grad_norm": 7.435749530792236, "learning_rate": 2.628664823202637e-07, "loss": 0.432, "step": 17740 }, { "epoch": 0.8974151449238708, "grad_norm": 4.325038909912109, "learning_rate": 2.62609728567233e-07, "loss": 0.3436, "step": 17741 }, { "epoch": 0.8974657291719359, "grad_norm": 17.044631958007812, "learning_rate": 2.623530968853061e-07, "loss": 0.4978, "step": 17742 }, { "epoch": 0.897516313420001, "grad_norm": 27.7334041595459, "learning_rate": 2.620965872810982e-07, "loss": 0.4826, "step": 17743 }, { "epoch": 0.8975668976680662, "grad_norm": 6.61631965637207, "learning_rate": 2.6184019976121633e-07, "loss": 0.3208, "step": 17744 }, { "epoch": 0.8976174819161313, "grad_norm": 7.99052619934082, "learning_rate": 2.615839343322674e-07, "loss": 0.4406, "step": 17745 }, { "epoch": 0.8976680661641965, "grad_norm": 12.43206787109375, "learning_rate": 2.6132779100085604e-07, "loss": 0.4558, "step": 17746 }, { "epoch": 0.8977186504122616, "grad_norm": 7.189762592315674, "learning_rate": 2.610717697735815e-07, "loss": 0.3282, "step": 17747 }, { "epoch": 0.8977692346603268, "grad_norm": 8.861347198486328, "learning_rate": 2.6081587065704073e-07, "loss": 0.4125, "step": 17748 }, { "epoch": 0.8978198189083919, "grad_norm": 7.091182708740234, "learning_rate": 2.605600936578284e-07, "loss": 0.4322, "step": 17749 }, { "epoch": 0.897870403156457, "grad_norm": 5.272587776184082, "learning_rate": 2.6030443878253486e-07, "loss": 0.3471, "step": 17750 }, { "epoch": 0.8979209874045222, "grad_norm": 9.266958236694336, "learning_rate": 2.60048906037747e-07, "loss": 0.3937, "step": 17751 }, { "epoch": 0.8979715716525873, "grad_norm": 5.947242736816406, "learning_rate": 2.5979349543005075e-07, "loss": 0.3772, "step": 17752 }, { "epoch": 0.8980221559006525, "grad_norm": 10.279091835021973, "learning_rate": 2.5953820696602626e-07, "loss": 0.4028, "step": 17753 }, { "epoch": 0.8980727401487177, "grad_norm": 7.443718910217285, "learning_rate": 2.5928304065225173e-07, "loss": 0.4705, "step": 17754 }, { "epoch": 0.8981233243967829, "grad_norm": 5.622101783752441, "learning_rate": 2.59027996495303e-07, "loss": 0.4284, "step": 17755 }, { "epoch": 0.898173908644848, "grad_norm": 7.519608497619629, "learning_rate": 2.5877307450175136e-07, "loss": 0.3033, "step": 17756 }, { "epoch": 0.8982244928929132, "grad_norm": 5.712183952331543, "learning_rate": 2.585182746781656e-07, "loss": 0.3627, "step": 17757 }, { "epoch": 0.8982750771409783, "grad_norm": 8.219978332519531, "learning_rate": 2.582635970311115e-07, "loss": 0.4484, "step": 17758 }, { "epoch": 0.8983256613890435, "grad_norm": 9.828580856323242, "learning_rate": 2.580090415671505e-07, "loss": 0.4158, "step": 17759 }, { "epoch": 0.8983762456371086, "grad_norm": 12.730303764343262, "learning_rate": 2.5775460829284335e-07, "loss": 0.5033, "step": 17760 }, { "epoch": 0.8984268298851737, "grad_norm": 11.080390930175781, "learning_rate": 2.5750029721474613e-07, "loss": 0.401, "step": 17761 }, { "epoch": 0.8984774141332389, "grad_norm": 7.210488796234131, "learning_rate": 2.5724610833940956e-07, "loss": 0.3973, "step": 17762 }, { "epoch": 0.898527998381304, "grad_norm": 14.682306289672852, "learning_rate": 2.569920416733862e-07, "loss": 0.5132, "step": 17763 }, { "epoch": 0.8985785826293692, "grad_norm": 7.33586311340332, "learning_rate": 2.5673809722322197e-07, "loss": 0.3582, "step": 17764 }, { "epoch": 0.8986291668774343, "grad_norm": 8.296931266784668, "learning_rate": 2.5648427499545935e-07, "loss": 0.3277, "step": 17765 }, { "epoch": 0.8986797511254995, "grad_norm": 14.700042724609375, "learning_rate": 2.5623057499663985e-07, "loss": 0.414, "step": 17766 }, { "epoch": 0.8987303353735647, "grad_norm": 5.268319606781006, "learning_rate": 2.559769972333004e-07, "loss": 0.3189, "step": 17767 }, { "epoch": 0.8987809196216299, "grad_norm": 9.854555130004883, "learning_rate": 2.5572354171197465e-07, "loss": 0.4763, "step": 17768 }, { "epoch": 0.898831503869695, "grad_norm": 6.019801616668701, "learning_rate": 2.554702084391947e-07, "loss": 0.3352, "step": 17769 }, { "epoch": 0.8988820881177602, "grad_norm": 7.031501770019531, "learning_rate": 2.552169974214874e-07, "loss": 0.2454, "step": 17770 }, { "epoch": 0.8989326723658253, "grad_norm": 11.704604148864746, "learning_rate": 2.5496390866537656e-07, "loss": 0.3668, "step": 17771 }, { "epoch": 0.8989832566138904, "grad_norm": 7.148955821990967, "learning_rate": 2.5471094217738523e-07, "loss": 0.4148, "step": 17772 }, { "epoch": 0.8990338408619556, "grad_norm": 8.217548370361328, "learning_rate": 2.5445809796403154e-07, "loss": 0.3709, "step": 17773 }, { "epoch": 0.8990844251100207, "grad_norm": 6.353656768798828, "learning_rate": 2.542053760318292e-07, "loss": 0.2428, "step": 17774 }, { "epoch": 0.8991350093580859, "grad_norm": 5.754801273345947, "learning_rate": 2.5395277638729177e-07, "loss": 0.2923, "step": 17775 }, { "epoch": 0.899185593606151, "grad_norm": 5.546868324279785, "learning_rate": 2.5370029903692807e-07, "loss": 0.4811, "step": 17776 }, { "epoch": 0.8992361778542162, "grad_norm": 7.9767842292785645, "learning_rate": 2.534479439872428e-07, "loss": 0.4473, "step": 17777 }, { "epoch": 0.8992867621022813, "grad_norm": 5.416449546813965, "learning_rate": 2.5319571124473976e-07, "loss": 0.4218, "step": 17778 }, { "epoch": 0.8993373463503465, "grad_norm": 19.224853515625, "learning_rate": 2.5294360081591806e-07, "loss": 0.4552, "step": 17779 }, { "epoch": 0.8993879305984116, "grad_norm": 20.150882720947266, "learning_rate": 2.526916127072726e-07, "loss": 0.4368, "step": 17780 }, { "epoch": 0.8994385148464769, "grad_norm": 6.5711541175842285, "learning_rate": 2.524397469252987e-07, "loss": 0.4413, "step": 17781 }, { "epoch": 0.899489099094542, "grad_norm": 12.194010734558105, "learning_rate": 2.521880034764856e-07, "loss": 0.43, "step": 17782 }, { "epoch": 0.8995396833426071, "grad_norm": 6.587823867797852, "learning_rate": 2.5193638236731877e-07, "loss": 0.3047, "step": 17783 }, { "epoch": 0.8995902675906723, "grad_norm": 5.2835307121276855, "learning_rate": 2.5168488360428344e-07, "loss": 0.3759, "step": 17784 }, { "epoch": 0.8996408518387374, "grad_norm": 6.056638240814209, "learning_rate": 2.514335071938595e-07, "loss": 0.386, "step": 17785 }, { "epoch": 0.8996914360868026, "grad_norm": 11.16028881072998, "learning_rate": 2.5118225314252343e-07, "loss": 0.5261, "step": 17786 }, { "epoch": 0.8997420203348677, "grad_norm": 5.839613914489746, "learning_rate": 2.509311214567517e-07, "loss": 0.4186, "step": 17787 }, { "epoch": 0.8997926045829329, "grad_norm": 5.428593158721924, "learning_rate": 2.506801121430136e-07, "loss": 0.3818, "step": 17788 }, { "epoch": 0.899843188830998, "grad_norm": 7.425753116607666, "learning_rate": 2.5042922520777723e-07, "loss": 0.4272, "step": 17789 }, { "epoch": 0.8998937730790632, "grad_norm": 9.522093772888184, "learning_rate": 2.5017846065750806e-07, "loss": 0.4386, "step": 17790 }, { "epoch": 0.8999443573271283, "grad_norm": 5.67985725402832, "learning_rate": 2.4992781849866697e-07, "loss": 0.2751, "step": 17791 }, { "epoch": 0.8999949415751934, "grad_norm": 8.55524730682373, "learning_rate": 2.4967729873771216e-07, "loss": 0.3589, "step": 17792 }, { "epoch": 0.9000455258232586, "grad_norm": 21.914609909057617, "learning_rate": 2.4942690138110015e-07, "loss": 0.5186, "step": 17793 }, { "epoch": 0.9000455258232586, "eval_loss": 0.38721156120300293, "eval_runtime": 34.9097, "eval_samples_per_second": 45.775, "eval_steps_per_second": 5.729, "step": 17793 }, { "epoch": 0.9000961100713238, "grad_norm": 7.812315940856934, "learning_rate": 2.491766264352824e-07, "loss": 0.4016, "step": 17794 }, { "epoch": 0.900146694319389, "grad_norm": 4.510722637176514, "learning_rate": 2.4892647390670765e-07, "loss": 0.3176, "step": 17795 }, { "epoch": 0.9001972785674541, "grad_norm": 12.211214065551758, "learning_rate": 2.4867644380182245e-07, "loss": 0.3843, "step": 17796 }, { "epoch": 0.9002478628155193, "grad_norm": 7.39359712600708, "learning_rate": 2.484265361270677e-07, "loss": 0.3562, "step": 17797 }, { "epoch": 0.9002984470635844, "grad_norm": 7.480024814605713, "learning_rate": 2.4817675088888496e-07, "loss": 0.3287, "step": 17798 }, { "epoch": 0.9003490313116496, "grad_norm": 14.625948905944824, "learning_rate": 2.479270880937101e-07, "loss": 0.5432, "step": 17799 }, { "epoch": 0.9003996155597147, "grad_norm": 5.905305862426758, "learning_rate": 2.4767754774797483e-07, "loss": 0.3511, "step": 17800 }, { "epoch": 0.9004501998077798, "grad_norm": 9.640791893005371, "learning_rate": 2.474281298581116e-07, "loss": 0.5246, "step": 17801 }, { "epoch": 0.900500784055845, "grad_norm": 5.662335395812988, "learning_rate": 2.471788344305459e-07, "loss": 0.3582, "step": 17802 }, { "epoch": 0.9005513683039101, "grad_norm": 5.019369602203369, "learning_rate": 2.469296614717004e-07, "loss": 0.3546, "step": 17803 }, { "epoch": 0.9006019525519753, "grad_norm": 11.771208763122559, "learning_rate": 2.466806109879982e-07, "loss": 0.4718, "step": 17804 }, { "epoch": 0.9006525368000404, "grad_norm": 11.219270706176758, "learning_rate": 2.464316829858554e-07, "loss": 0.4628, "step": 17805 }, { "epoch": 0.9007031210481056, "grad_norm": 7.4976043701171875, "learning_rate": 2.4618287747168677e-07, "loss": 0.3707, "step": 17806 }, { "epoch": 0.9007537052961708, "grad_norm": 9.949105262756348, "learning_rate": 2.4593419445190224e-07, "loss": 0.4997, "step": 17807 }, { "epoch": 0.900804289544236, "grad_norm": 7.993260383605957, "learning_rate": 2.4568563393291114e-07, "loss": 0.307, "step": 17808 }, { "epoch": 0.9008548737923011, "grad_norm": 8.69392204284668, "learning_rate": 2.454371959211166e-07, "loss": 0.4332, "step": 17809 }, { "epoch": 0.9009054580403663, "grad_norm": 12.270647048950195, "learning_rate": 2.4518888042292244e-07, "loss": 0.493, "step": 17810 }, { "epoch": 0.9009560422884314, "grad_norm": 9.431700706481934, "learning_rate": 2.449406874447258e-07, "loss": 0.3427, "step": 17811 }, { "epoch": 0.9010066265364965, "grad_norm": 7.135338306427002, "learning_rate": 2.4469261699292147e-07, "loss": 0.3943, "step": 17812 }, { "epoch": 0.9010572107845617, "grad_norm": 11.712813377380371, "learning_rate": 2.444446690739033e-07, "loss": 0.4603, "step": 17813 }, { "epoch": 0.9011077950326268, "grad_norm": 8.935976028442383, "learning_rate": 2.441968436940595e-07, "loss": 0.2888, "step": 17814 }, { "epoch": 0.901158379280692, "grad_norm": 10.998862266540527, "learning_rate": 2.4394914085977493e-07, "loss": 0.4158, "step": 17815 }, { "epoch": 0.9012089635287571, "grad_norm": 9.128280639648438, "learning_rate": 2.43701560577434e-07, "loss": 0.4473, "step": 17816 }, { "epoch": 0.9012595477768223, "grad_norm": 8.886988639831543, "learning_rate": 2.4345410285341595e-07, "loss": 0.3154, "step": 17817 }, { "epoch": 0.9013101320248874, "grad_norm": 8.825963973999023, "learning_rate": 2.432067676940952e-07, "loss": 0.4877, "step": 17818 }, { "epoch": 0.9013607162729526, "grad_norm": 7.950741291046143, "learning_rate": 2.429595551058478e-07, "loss": 0.5105, "step": 17819 }, { "epoch": 0.9014113005210177, "grad_norm": 6.699279308319092, "learning_rate": 2.42712465095043e-07, "loss": 0.3864, "step": 17820 }, { "epoch": 0.901461884769083, "grad_norm": 11.511491775512695, "learning_rate": 2.424654976680457e-07, "loss": 0.4316, "step": 17821 }, { "epoch": 0.9015124690171481, "grad_norm": 13.782022476196289, "learning_rate": 2.4221865283122206e-07, "loss": 0.4319, "step": 17822 }, { "epoch": 0.9015630532652132, "grad_norm": 13.348176002502441, "learning_rate": 2.4197193059093185e-07, "loss": 0.5113, "step": 17823 }, { "epoch": 0.9016136375132784, "grad_norm": 15.376391410827637, "learning_rate": 2.4172533095353165e-07, "loss": 0.4848, "step": 17824 }, { "epoch": 0.9016642217613435, "grad_norm": 21.827442169189453, "learning_rate": 2.4147885392537707e-07, "loss": 0.4352, "step": 17825 }, { "epoch": 0.9017148060094087, "grad_norm": 8.725470542907715, "learning_rate": 2.412324995128185e-07, "loss": 0.4329, "step": 17826 }, { "epoch": 0.9017653902574738, "grad_norm": 8.824043273925781, "learning_rate": 2.409862677222041e-07, "loss": 0.3769, "step": 17827 }, { "epoch": 0.901815974505539, "grad_norm": 14.190709114074707, "learning_rate": 2.40740158559879e-07, "loss": 0.4038, "step": 17828 }, { "epoch": 0.9018665587536041, "grad_norm": 8.150365829467773, "learning_rate": 2.404941720321841e-07, "loss": 0.4156, "step": 17829 }, { "epoch": 0.9019171430016693, "grad_norm": 13.016549110412598, "learning_rate": 2.4024830814545816e-07, "loss": 0.4494, "step": 17830 }, { "epoch": 0.9019677272497344, "grad_norm": 10.042250633239746, "learning_rate": 2.4000256690603686e-07, "loss": 0.4387, "step": 17831 }, { "epoch": 0.9020183114977995, "grad_norm": 9.594364166259766, "learning_rate": 2.397569483202522e-07, "loss": 0.5081, "step": 17832 }, { "epoch": 0.9020688957458647, "grad_norm": 7.502291202545166, "learning_rate": 2.395114523944331e-07, "loss": 0.4427, "step": 17833 }, { "epoch": 0.9021194799939299, "grad_norm": 11.97223949432373, "learning_rate": 2.392660791349055e-07, "loss": 0.499, "step": 17834 }, { "epoch": 0.9021700642419951, "grad_norm": 5.267307281494141, "learning_rate": 2.3902082854799116e-07, "loss": 0.377, "step": 17835 }, { "epoch": 0.9022206484900602, "grad_norm": 7.996082782745361, "learning_rate": 2.387757006400115e-07, "loss": 0.4153, "step": 17836 }, { "epoch": 0.9022712327381254, "grad_norm": 6.1684250831604, "learning_rate": 2.3853069541728113e-07, "loss": 0.4116, "step": 17837 }, { "epoch": 0.9023218169861905, "grad_norm": 19.791248321533203, "learning_rate": 2.3828581288611375e-07, "loss": 0.3279, "step": 17838 }, { "epoch": 0.9023724012342557, "grad_norm": 8.564045906066895, "learning_rate": 2.3804105305281988e-07, "loss": 0.3462, "step": 17839 }, { "epoch": 0.9024229854823208, "grad_norm": 10.541901588439941, "learning_rate": 2.3779641592370618e-07, "loss": 0.3858, "step": 17840 }, { "epoch": 0.902473569730386, "grad_norm": 11.825549125671387, "learning_rate": 2.3755190150507534e-07, "loss": 0.5056, "step": 17841 }, { "epoch": 0.9025241539784511, "grad_norm": 8.907440185546875, "learning_rate": 2.3730750980322958e-07, "loss": 0.3949, "step": 17842 }, { "epoch": 0.9025747382265162, "grad_norm": 6.246325969696045, "learning_rate": 2.3706324082446553e-07, "loss": 0.3407, "step": 17843 }, { "epoch": 0.9026253224745814, "grad_norm": 10.690455436706543, "learning_rate": 2.36819094575077e-07, "loss": 0.4528, "step": 17844 }, { "epoch": 0.9026759067226465, "grad_norm": 10.924117088317871, "learning_rate": 2.3657507106135513e-07, "loss": 0.3889, "step": 17845 }, { "epoch": 0.9027264909707117, "grad_norm": 12.081451416015625, "learning_rate": 2.363311702895893e-07, "loss": 0.4022, "step": 17846 }, { "epoch": 0.9027770752187769, "grad_norm": 6.071834564208984, "learning_rate": 2.360873922660617e-07, "loss": 0.3364, "step": 17847 }, { "epoch": 0.9028276594668421, "grad_norm": 12.022193908691406, "learning_rate": 2.358437369970562e-07, "loss": 0.5888, "step": 17848 }, { "epoch": 0.9028782437149072, "grad_norm": 11.63740348815918, "learning_rate": 2.3560020448885002e-07, "loss": 0.3286, "step": 17849 }, { "epoch": 0.9029288279629724, "grad_norm": 18.339006423950195, "learning_rate": 2.353567947477181e-07, "loss": 0.516, "step": 17850 }, { "epoch": 0.9029794122110375, "grad_norm": 7.133640289306641, "learning_rate": 2.3511350777993326e-07, "loss": 0.3538, "step": 17851 }, { "epoch": 0.9030299964591026, "grad_norm": 6.496919631958008, "learning_rate": 2.348703435917643e-07, "loss": 0.3503, "step": 17852 }, { "epoch": 0.9030805807071678, "grad_norm": 14.222729682922363, "learning_rate": 2.3462730218947683e-07, "loss": 0.4371, "step": 17853 }, { "epoch": 0.9031311649552329, "grad_norm": 12.615667343139648, "learning_rate": 2.3438438357933359e-07, "loss": 0.3581, "step": 17854 }, { "epoch": 0.9031817492032981, "grad_norm": 9.156888961791992, "learning_rate": 2.3414158776759398e-07, "loss": 0.471, "step": 17855 }, { "epoch": 0.9032323334513632, "grad_norm": 6.671463966369629, "learning_rate": 2.3389891476051364e-07, "loss": 0.4218, "step": 17856 }, { "epoch": 0.9032829176994284, "grad_norm": 10.474609375, "learning_rate": 2.3365636456434693e-07, "loss": 0.3784, "step": 17857 }, { "epoch": 0.9033335019474935, "grad_norm": 5.759758472442627, "learning_rate": 2.3341393718534334e-07, "loss": 0.3677, "step": 17858 }, { "epoch": 0.9033840861955587, "grad_norm": 6.3170857429504395, "learning_rate": 2.3317163262974897e-07, "loss": 0.3723, "step": 17859 }, { "epoch": 0.9034346704436239, "grad_norm": 8.267130851745605, "learning_rate": 2.3292945090380769e-07, "loss": 0.3192, "step": 17860 }, { "epoch": 0.903485254691689, "grad_norm": 13.241406440734863, "learning_rate": 2.3268739201376068e-07, "loss": 0.3821, "step": 17861 }, { "epoch": 0.9035358389397542, "grad_norm": 6.604836940765381, "learning_rate": 2.3244545596584345e-07, "loss": 0.4194, "step": 17862 }, { "epoch": 0.9035864231878193, "grad_norm": 7.551913261413574, "learning_rate": 2.3220364276629216e-07, "loss": 0.4967, "step": 17863 }, { "epoch": 0.9036370074358845, "grad_norm": 5.233462333679199, "learning_rate": 2.319619524213368e-07, "loss": 0.2961, "step": 17864 }, { "epoch": 0.9036875916839496, "grad_norm": 8.071817398071289, "learning_rate": 2.3172038493720405e-07, "loss": 0.3992, "step": 17865 }, { "epoch": 0.9037381759320148, "grad_norm": 9.14697551727295, "learning_rate": 2.3147894032012064e-07, "loss": 0.4279, "step": 17866 }, { "epoch": 0.9037887601800799, "grad_norm": 10.233606338500977, "learning_rate": 2.3123761857630768e-07, "loss": 0.3324, "step": 17867 }, { "epoch": 0.9038393444281451, "grad_norm": 7.853555202484131, "learning_rate": 2.3099641971198127e-07, "loss": 0.3162, "step": 17868 }, { "epoch": 0.9038899286762102, "grad_norm": 6.486359119415283, "learning_rate": 2.3075534373335928e-07, "loss": 0.3872, "step": 17869 }, { "epoch": 0.9039405129242754, "grad_norm": 8.011341094970703, "learning_rate": 2.3051439064665228e-07, "loss": 0.44, "step": 17870 }, { "epoch": 0.9039910971723405, "grad_norm": 8.777803421020508, "learning_rate": 2.302735604580686e-07, "loss": 0.4417, "step": 17871 }, { "epoch": 0.9040416814204056, "grad_norm": 9.351136207580566, "learning_rate": 2.3003285317381553e-07, "loss": 0.467, "step": 17872 }, { "epoch": 0.9040922656684708, "grad_norm": 8.792867660522461, "learning_rate": 2.2979226880009365e-07, "loss": 0.4652, "step": 17873 }, { "epoch": 0.904142849916536, "grad_norm": 8.873394966125488, "learning_rate": 2.29551807343103e-07, "loss": 0.3817, "step": 17874 }, { "epoch": 0.9041934341646012, "grad_norm": 7.306870937347412, "learning_rate": 2.2931146880904033e-07, "loss": 0.3105, "step": 17875 }, { "epoch": 0.9042440184126663, "grad_norm": 4.802034854888916, "learning_rate": 2.2907125320409674e-07, "loss": 0.3376, "step": 17876 }, { "epoch": 0.9042946026607315, "grad_norm": 10.312411308288574, "learning_rate": 2.2883116053446452e-07, "loss": 0.4331, "step": 17877 }, { "epoch": 0.9043451869087966, "grad_norm": 11.208078384399414, "learning_rate": 2.2859119080632874e-07, "loss": 0.3181, "step": 17878 }, { "epoch": 0.9043957711568618, "grad_norm": 10.242791175842285, "learning_rate": 2.2835134402587334e-07, "loss": 0.3555, "step": 17879 }, { "epoch": 0.9044463554049269, "grad_norm": 29.953868865966797, "learning_rate": 2.2811162019927723e-07, "loss": 0.4024, "step": 17880 }, { "epoch": 0.904496939652992, "grad_norm": 6.655920028686523, "learning_rate": 2.2787201933271942e-07, "loss": 0.4529, "step": 17881 }, { "epoch": 0.9045475239010572, "grad_norm": 4.468623161315918, "learning_rate": 2.2763254143237322e-07, "loss": 0.3823, "step": 17882 }, { "epoch": 0.9045981081491223, "grad_norm": 10.090559005737305, "learning_rate": 2.2739318650440877e-07, "loss": 0.3579, "step": 17883 }, { "epoch": 0.9046486923971875, "grad_norm": 12.927085876464844, "learning_rate": 2.27153954554995e-07, "loss": 0.5732, "step": 17884 }, { "epoch": 0.9046992766452526, "grad_norm": 7.419013500213623, "learning_rate": 2.2691484559029476e-07, "loss": 0.3801, "step": 17885 }, { "epoch": 0.9047498608933178, "grad_norm": 11.051295280456543, "learning_rate": 2.2667585961647032e-07, "loss": 0.3156, "step": 17886 }, { "epoch": 0.904800445141383, "grad_norm": 5.673203945159912, "learning_rate": 2.2643699663967956e-07, "loss": 0.3705, "step": 17887 }, { "epoch": 0.9048510293894482, "grad_norm": 12.604878425598145, "learning_rate": 2.2619825666607643e-07, "loss": 0.433, "step": 17888 }, { "epoch": 0.9049016136375133, "grad_norm": 14.173360824584961, "learning_rate": 2.2595963970181434e-07, "loss": 0.4202, "step": 17889 }, { "epoch": 0.9049521978855785, "grad_norm": 7.308042049407959, "learning_rate": 2.2572114575304115e-07, "loss": 0.3655, "step": 17890 }, { "epoch": 0.9050027821336436, "grad_norm": 6.465177059173584, "learning_rate": 2.2548277482590197e-07, "loss": 0.3272, "step": 17891 }, { "epoch": 0.9050533663817087, "grad_norm": 6.177292346954346, "learning_rate": 2.2524452692653965e-07, "loss": 0.337, "step": 17892 }, { "epoch": 0.9051039506297739, "grad_norm": 11.567218780517578, "learning_rate": 2.250064020610926e-07, "loss": 0.5313, "step": 17893 }, { "epoch": 0.905154534877839, "grad_norm": 5.923598289489746, "learning_rate": 2.2476840023569702e-07, "loss": 0.2674, "step": 17894 }, { "epoch": 0.9052051191259042, "grad_norm": 5.490896224975586, "learning_rate": 2.245305214564858e-07, "loss": 0.3953, "step": 17895 }, { "epoch": 0.9052557033739693, "grad_norm": 5.962423801422119, "learning_rate": 2.2429276572958847e-07, "loss": 0.4177, "step": 17896 }, { "epoch": 0.9053062876220345, "grad_norm": 7.463687419891357, "learning_rate": 2.2405513306113125e-07, "loss": 0.3309, "step": 17897 }, { "epoch": 0.9053568718700996, "grad_norm": 4.559808731079102, "learning_rate": 2.238176234572381e-07, "loss": 0.376, "step": 17898 }, { "epoch": 0.9054074561181648, "grad_norm": 7.520847320556641, "learning_rate": 2.2358023692402753e-07, "loss": 0.4301, "step": 17899 }, { "epoch": 0.90545804036623, "grad_norm": 7.998363018035889, "learning_rate": 2.2334297346761735e-07, "loss": 0.4073, "step": 17900 }, { "epoch": 0.9055086246142952, "grad_norm": 7.680800437927246, "learning_rate": 2.2310583309412105e-07, "loss": 0.3684, "step": 17901 }, { "epoch": 0.9055592088623603, "grad_norm": 7.033201694488525, "learning_rate": 2.228688158096498e-07, "loss": 0.2495, "step": 17902 }, { "epoch": 0.9056097931104254, "grad_norm": 5.733092784881592, "learning_rate": 2.226319216203099e-07, "loss": 0.3331, "step": 17903 }, { "epoch": 0.9056603773584906, "grad_norm": 5.010790824890137, "learning_rate": 2.2239515053220694e-07, "loss": 0.3626, "step": 17904 }, { "epoch": 0.9057109616065557, "grad_norm": 13.668703079223633, "learning_rate": 2.2215850255144057e-07, "loss": 0.4916, "step": 17905 }, { "epoch": 0.9057615458546209, "grad_norm": 6.01014518737793, "learning_rate": 2.219219776841086e-07, "loss": 0.3694, "step": 17906 }, { "epoch": 0.905812130102686, "grad_norm": 8.384641647338867, "learning_rate": 2.2168557593630735e-07, "loss": 0.3781, "step": 17907 }, { "epoch": 0.9058627143507512, "grad_norm": 8.623351097106934, "learning_rate": 2.214492973141269e-07, "loss": 0.3587, "step": 17908 }, { "epoch": 0.9059132985988163, "grad_norm": 9.241313934326172, "learning_rate": 2.2121314182365573e-07, "loss": 0.4511, "step": 17909 }, { "epoch": 0.9059638828468815, "grad_norm": 11.077753067016602, "learning_rate": 2.2097710947098005e-07, "loss": 0.5758, "step": 17910 }, { "epoch": 0.9060144670949466, "grad_norm": 9.592082977294922, "learning_rate": 2.2074120026218003e-07, "loss": 0.4053, "step": 17911 }, { "epoch": 0.9060650513430117, "grad_norm": 11.699898719787598, "learning_rate": 2.205054142033358e-07, "loss": 0.4308, "step": 17912 }, { "epoch": 0.9061156355910769, "grad_norm": 14.09052562713623, "learning_rate": 2.2026975130052308e-07, "loss": 0.5149, "step": 17913 }, { "epoch": 0.9061662198391421, "grad_norm": 7.428478717803955, "learning_rate": 2.2003421155981254e-07, "loss": 0.3668, "step": 17914 }, { "epoch": 0.9062168040872073, "grad_norm": 5.362180233001709, "learning_rate": 2.1979879498727597e-07, "loss": 0.3178, "step": 17915 }, { "epoch": 0.9062673883352724, "grad_norm": 15.677010536193848, "learning_rate": 2.1956350158897855e-07, "loss": 0.4421, "step": 17916 }, { "epoch": 0.9063179725833376, "grad_norm": 9.551774024963379, "learning_rate": 2.1932833137098264e-07, "loss": 0.3411, "step": 17917 }, { "epoch": 0.9063685568314027, "grad_norm": 7.963179588317871, "learning_rate": 2.1909328433934785e-07, "loss": 0.3496, "step": 17918 }, { "epoch": 0.9064191410794679, "grad_norm": 12.069939613342285, "learning_rate": 2.1885836050013208e-07, "loss": 0.5499, "step": 17919 }, { "epoch": 0.906469725327533, "grad_norm": 6.859005928039551, "learning_rate": 2.1862355985938777e-07, "loss": 0.3263, "step": 17920 }, { "epoch": 0.9065203095755981, "grad_norm": 7.371728897094727, "learning_rate": 2.1838888242316501e-07, "loss": 0.3603, "step": 17921 }, { "epoch": 0.9065708938236633, "grad_norm": 6.448383331298828, "learning_rate": 2.181543281975118e-07, "loss": 0.3286, "step": 17922 }, { "epoch": 0.9066214780717284, "grad_norm": 9.595866203308105, "learning_rate": 2.1791989718847106e-07, "loss": 0.4955, "step": 17923 }, { "epoch": 0.9066720623197936, "grad_norm": 12.064123153686523, "learning_rate": 2.176855894020846e-07, "loss": 0.5071, "step": 17924 }, { "epoch": 0.9067226465678587, "grad_norm": 6.137102127075195, "learning_rate": 2.1745140484438876e-07, "loss": 0.3939, "step": 17925 }, { "epoch": 0.9067732308159239, "grad_norm": 4.859837055206299, "learning_rate": 2.172173435214181e-07, "loss": 0.2812, "step": 17926 }, { "epoch": 0.9068238150639891, "grad_norm": 9.402637481689453, "learning_rate": 2.169834054392045e-07, "loss": 0.5252, "step": 17927 }, { "epoch": 0.9068743993120543, "grad_norm": 7.641845226287842, "learning_rate": 2.1674959060377588e-07, "loss": 0.4679, "step": 17928 }, { "epoch": 0.9069249835601194, "grad_norm": 8.574248313903809, "learning_rate": 2.1651589902115632e-07, "loss": 0.4605, "step": 17929 }, { "epoch": 0.9069755678081846, "grad_norm": 10.358867645263672, "learning_rate": 2.1628233069736883e-07, "loss": 0.3376, "step": 17930 }, { "epoch": 0.9070261520562497, "grad_norm": 9.192286491394043, "learning_rate": 2.1604888563843074e-07, "loss": 0.19, "step": 17931 }, { "epoch": 0.9070767363043148, "grad_norm": 11.03645133972168, "learning_rate": 2.1581556385035674e-07, "loss": 0.4617, "step": 17932 }, { "epoch": 0.90712732055238, "grad_norm": 9.387890815734863, "learning_rate": 2.155823653391609e-07, "loss": 0.3466, "step": 17933 }, { "epoch": 0.9071779048004451, "grad_norm": 6.019166946411133, "learning_rate": 2.1534929011085115e-07, "loss": 0.3355, "step": 17934 }, { "epoch": 0.9072284890485103, "grad_norm": 8.236824989318848, "learning_rate": 2.151163381714333e-07, "loss": 0.4048, "step": 17935 }, { "epoch": 0.9072790732965754, "grad_norm": 12.451743125915527, "learning_rate": 2.148835095269103e-07, "loss": 0.3451, "step": 17936 }, { "epoch": 0.9073296575446406, "grad_norm": 10.751908302307129, "learning_rate": 2.146508041832812e-07, "loss": 0.4676, "step": 17937 }, { "epoch": 0.9073802417927057, "grad_norm": 6.848008155822754, "learning_rate": 2.1441822214654185e-07, "loss": 0.3673, "step": 17938 }, { "epoch": 0.9074308260407709, "grad_norm": 9.10427474975586, "learning_rate": 2.1418576342268627e-07, "loss": 0.4376, "step": 17939 }, { "epoch": 0.9074814102888361, "grad_norm": 10.666888236999512, "learning_rate": 2.1395342801770357e-07, "loss": 0.4019, "step": 17940 }, { "epoch": 0.9075319945369013, "grad_norm": 6.287429332733154, "learning_rate": 2.1372121593758066e-07, "loss": 0.2712, "step": 17941 }, { "epoch": 0.9075825787849664, "grad_norm": 7.628593921661377, "learning_rate": 2.1348912718830162e-07, "loss": 0.3503, "step": 17942 }, { "epoch": 0.9076331630330315, "grad_norm": 7.3453874588012695, "learning_rate": 2.1325716177584665e-07, "loss": 0.388, "step": 17943 }, { "epoch": 0.9076837472810967, "grad_norm": 5.384586334228516, "learning_rate": 2.1302531970619155e-07, "loss": 0.2843, "step": 17944 }, { "epoch": 0.9077343315291618, "grad_norm": 6.047828674316406, "learning_rate": 2.1279360098531266e-07, "loss": 0.3711, "step": 17945 }, { "epoch": 0.907784915777227, "grad_norm": 7.409852027893066, "learning_rate": 2.1256200561917961e-07, "loss": 0.4272, "step": 17946 }, { "epoch": 0.9078355000252921, "grad_norm": 6.1961236000061035, "learning_rate": 2.1233053361375933e-07, "loss": 0.3704, "step": 17947 }, { "epoch": 0.9078860842733573, "grad_norm": 5.165879726409912, "learning_rate": 2.120991849750176e-07, "loss": 0.315, "step": 17948 }, { "epoch": 0.9079366685214224, "grad_norm": 7.69558572769165, "learning_rate": 2.1186795970891516e-07, "loss": 0.3968, "step": 17949 }, { "epoch": 0.9079872527694876, "grad_norm": 22.21387481689453, "learning_rate": 2.1163685782141008e-07, "loss": 0.4296, "step": 17950 }, { "epoch": 0.9080378370175527, "grad_norm": 9.044183731079102, "learning_rate": 2.1140587931845757e-07, "loss": 0.3416, "step": 17951 }, { "epoch": 0.9080884212656178, "grad_norm": 7.476261615753174, "learning_rate": 2.11175024206009e-07, "loss": 0.3487, "step": 17952 }, { "epoch": 0.908139005513683, "grad_norm": 12.610543251037598, "learning_rate": 2.1094429249001291e-07, "loss": 0.3732, "step": 17953 }, { "epoch": 0.9081895897617482, "grad_norm": 9.584221839904785, "learning_rate": 2.1071368417641513e-07, "loss": 0.3453, "step": 17954 }, { "epoch": 0.9082401740098134, "grad_norm": 7.10233736038208, "learning_rate": 2.1048319927115756e-07, "loss": 0.3819, "step": 17955 }, { "epoch": 0.9082907582578785, "grad_norm": 6.9406328201293945, "learning_rate": 2.1025283778017935e-07, "loss": 0.4128, "step": 17956 }, { "epoch": 0.9083413425059437, "grad_norm": 9.627140998840332, "learning_rate": 2.100225997094163e-07, "loss": 0.418, "step": 17957 }, { "epoch": 0.9083919267540088, "grad_norm": 9.378839492797852, "learning_rate": 2.0979248506480142e-07, "loss": 0.3982, "step": 17958 }, { "epoch": 0.908442511002074, "grad_norm": 9.663347244262695, "learning_rate": 2.0956249385226334e-07, "loss": 0.3934, "step": 17959 }, { "epoch": 0.9084930952501391, "grad_norm": 6.084871292114258, "learning_rate": 2.0933262607772952e-07, "loss": 0.3957, "step": 17960 }, { "epoch": 0.9085436794982042, "grad_norm": 5.247586727142334, "learning_rate": 2.0910288174712245e-07, "loss": 0.4156, "step": 17961 }, { "epoch": 0.9085942637462694, "grad_norm": 5.741183757781982, "learning_rate": 2.0887326086636238e-07, "loss": 0.4097, "step": 17962 }, { "epoch": 0.9086448479943345, "grad_norm": 8.190406799316406, "learning_rate": 2.0864376344136572e-07, "loss": 0.4581, "step": 17963 }, { "epoch": 0.9086954322423997, "grad_norm": 12.184845924377441, "learning_rate": 2.0841438947804547e-07, "loss": 0.4548, "step": 17964 }, { "epoch": 0.9087460164904648, "grad_norm": 14.559155464172363, "learning_rate": 2.0818513898231417e-07, "loss": 0.6091, "step": 17965 }, { "epoch": 0.90879660073853, "grad_norm": 23.947975158691406, "learning_rate": 2.079560119600771e-07, "loss": 0.4999, "step": 17966 }, { "epoch": 0.9088471849865952, "grad_norm": 6.948502540588379, "learning_rate": 2.0772700841723837e-07, "loss": 0.4088, "step": 17967 }, { "epoch": 0.9088977692346604, "grad_norm": 10.859220504760742, "learning_rate": 2.074981283596994e-07, "loss": 0.305, "step": 17968 }, { "epoch": 0.9089483534827255, "grad_norm": 5.31929874420166, "learning_rate": 2.0726937179335882e-07, "loss": 0.3414, "step": 17969 }, { "epoch": 0.9089989377307907, "grad_norm": 5.959412097930908, "learning_rate": 2.0704073872410913e-07, "loss": 0.3144, "step": 17970 }, { "epoch": 0.9090495219788558, "grad_norm": 7.071030139923096, "learning_rate": 2.068122291578428e-07, "loss": 0.405, "step": 17971 }, { "epoch": 0.909100106226921, "grad_norm": 6.182740211486816, "learning_rate": 2.0658384310044856e-07, "loss": 0.3447, "step": 17972 }, { "epoch": 0.9091506904749861, "grad_norm": 12.130846977233887, "learning_rate": 2.063555805578099e-07, "loss": 0.4497, "step": 17973 }, { "epoch": 0.9092012747230512, "grad_norm": 4.0009918212890625, "learning_rate": 2.0612744153581e-07, "loss": 0.353, "step": 17974 }, { "epoch": 0.9092518589711164, "grad_norm": 7.1172261238098145, "learning_rate": 2.0589942604032742e-07, "loss": 0.3797, "step": 17975 }, { "epoch": 0.9093024432191815, "grad_norm": 11.577281951904297, "learning_rate": 2.056715340772353e-07, "loss": 0.481, "step": 17976 }, { "epoch": 0.9093530274672467, "grad_norm": 7.641755104064941, "learning_rate": 2.054437656524083e-07, "loss": 0.395, "step": 17977 }, { "epoch": 0.9094036117153118, "grad_norm": 8.122292518615723, "learning_rate": 2.0521612077171516e-07, "loss": 0.3789, "step": 17978 }, { "epoch": 0.909454195963377, "grad_norm": 10.758186340332031, "learning_rate": 2.0498859944102e-07, "loss": 0.4861, "step": 17979 }, { "epoch": 0.9095047802114422, "grad_norm": 10.27393913269043, "learning_rate": 2.0476120166618763e-07, "loss": 0.4773, "step": 17980 }, { "epoch": 0.9095553644595074, "grad_norm": 7.379271030426025, "learning_rate": 2.045339274530772e-07, "loss": 0.3912, "step": 17981 }, { "epoch": 0.9096059487075725, "grad_norm": 6.918298244476318, "learning_rate": 2.0430677680754296e-07, "loss": 0.2875, "step": 17982 }, { "epoch": 0.9096565329556376, "grad_norm": 8.230429649353027, "learning_rate": 2.0407974973544075e-07, "loss": 0.4423, "step": 17983 }, { "epoch": 0.9097071172037028, "grad_norm": 8.16307544708252, "learning_rate": 2.0385284624261926e-07, "loss": 0.3786, "step": 17984 }, { "epoch": 0.9097577014517679, "grad_norm": 7.490428447723389, "learning_rate": 2.0362606633492488e-07, "loss": 0.481, "step": 17985 }, { "epoch": 0.9098082856998331, "grad_norm": 7.117717742919922, "learning_rate": 2.0339941001820185e-07, "loss": 0.4639, "step": 17986 }, { "epoch": 0.9098588699478982, "grad_norm": 9.609834671020508, "learning_rate": 2.031728772982905e-07, "loss": 0.4306, "step": 17987 }, { "epoch": 0.9099094541959634, "grad_norm": 4.708652973175049, "learning_rate": 2.0294646818102725e-07, "loss": 0.3516, "step": 17988 }, { "epoch": 0.9099600384440285, "grad_norm": 12.159859657287598, "learning_rate": 2.0272018267224746e-07, "loss": 0.3951, "step": 17989 }, { "epoch": 0.9100106226920937, "grad_norm": 7.305516719818115, "learning_rate": 2.0249402077778092e-07, "loss": 0.3565, "step": 17990 }, { "epoch": 0.9100612069401588, "grad_norm": 8.168045043945312, "learning_rate": 2.0226798250345514e-07, "loss": 0.3104, "step": 17991 }, { "epoch": 0.9101117911882239, "grad_norm": 7.432464122772217, "learning_rate": 2.020420678550955e-07, "loss": 0.354, "step": 17992 }, { "epoch": 0.9101623754362892, "grad_norm": 14.15560531616211, "learning_rate": 2.0181627683852234e-07, "loss": 0.4542, "step": 17993 }, { "epoch": 0.9102129596843543, "grad_norm": 7.247044563293457, "learning_rate": 2.0159060945955434e-07, "loss": 0.4346, "step": 17994 }, { "epoch": 0.9102635439324195, "grad_norm": 24.13172721862793, "learning_rate": 2.0136506572400628e-07, "loss": 0.6132, "step": 17995 }, { "epoch": 0.9103141281804846, "grad_norm": 7.86276388168335, "learning_rate": 2.0113964563769018e-07, "loss": 0.3492, "step": 17996 }, { "epoch": 0.9103647124285498, "grad_norm": 6.672776699066162, "learning_rate": 2.009143492064136e-07, "loss": 0.3477, "step": 17997 }, { "epoch": 0.9104152966766149, "grad_norm": 6.802432060241699, "learning_rate": 2.0068917643598306e-07, "loss": 0.299, "step": 17998 }, { "epoch": 0.9104658809246801, "grad_norm": 6.999709606170654, "learning_rate": 2.004641273322e-07, "loss": 0.4522, "step": 17999 }, { "epoch": 0.9105164651727452, "grad_norm": 20.546491622924805, "learning_rate": 2.0023920190086365e-07, "loss": 0.4256, "step": 18000 }, { "epoch": 0.9105670494208103, "grad_norm": 6.511049270629883, "learning_rate": 2.0001440014776996e-07, "loss": 0.4181, "step": 18001 }, { "epoch": 0.9106176336688755, "grad_norm": 7.251214027404785, "learning_rate": 1.997897220787104e-07, "loss": 0.3466, "step": 18002 }, { "epoch": 0.9106682179169406, "grad_norm": 8.320406913757324, "learning_rate": 1.9956516769947642e-07, "loss": 0.3871, "step": 18003 }, { "epoch": 0.9107188021650058, "grad_norm": 15.272439956665039, "learning_rate": 1.9934073701585234e-07, "loss": 0.4399, "step": 18004 }, { "epoch": 0.9107693864130709, "grad_norm": 7.888855457305908, "learning_rate": 1.991164300336218e-07, "loss": 0.3066, "step": 18005 }, { "epoch": 0.9108199706611361, "grad_norm": 6.932697296142578, "learning_rate": 1.9889224675856522e-07, "loss": 0.358, "step": 18006 }, { "epoch": 0.9108705549092013, "grad_norm": 8.646011352539062, "learning_rate": 1.9866818719645908e-07, "loss": 0.3147, "step": 18007 }, { "epoch": 0.9109211391572665, "grad_norm": 13.001143455505371, "learning_rate": 1.9844425135307598e-07, "loss": 0.6259, "step": 18008 }, { "epoch": 0.9109717234053316, "grad_norm": 4.2752766609191895, "learning_rate": 1.982204392341869e-07, "loss": 0.2695, "step": 18009 }, { "epoch": 0.9110223076533968, "grad_norm": 8.44071102142334, "learning_rate": 1.9799675084555936e-07, "loss": 0.4042, "step": 18010 }, { "epoch": 0.9110728919014619, "grad_norm": 7.746998310089111, "learning_rate": 1.9777318619295606e-07, "loss": 0.533, "step": 18011 }, { "epoch": 0.911123476149527, "grad_norm": 6.167059421539307, "learning_rate": 1.9754974528213955e-07, "loss": 0.3846, "step": 18012 }, { "epoch": 0.9111740603975922, "grad_norm": 10.91153335571289, "learning_rate": 1.9732642811886638e-07, "loss": 0.4464, "step": 18013 }, { "epoch": 0.9112246446456573, "grad_norm": 6.4340128898620605, "learning_rate": 1.9710323470888914e-07, "loss": 0.3521, "step": 18014 }, { "epoch": 0.9112752288937225, "grad_norm": 7.646162509918213, "learning_rate": 1.9688016505796214e-07, "loss": 0.37, "step": 18015 }, { "epoch": 0.9113258131417876, "grad_norm": 5.667272090911865, "learning_rate": 1.9665721917183134e-07, "loss": 0.372, "step": 18016 }, { "epoch": 0.9113763973898528, "grad_norm": 13.289728164672852, "learning_rate": 1.96434397056241e-07, "loss": 0.3812, "step": 18017 }, { "epoch": 0.9114269816379179, "grad_norm": 11.109976768493652, "learning_rate": 1.962116987169349e-07, "loss": 0.4119, "step": 18018 }, { "epoch": 0.911477565885983, "grad_norm": 13.602479934692383, "learning_rate": 1.959891241596501e-07, "loss": 0.5543, "step": 18019 }, { "epoch": 0.9115281501340483, "grad_norm": 5.396666526794434, "learning_rate": 1.9576667339012145e-07, "loss": 0.35, "step": 18020 }, { "epoch": 0.9115787343821135, "grad_norm": 5.955693244934082, "learning_rate": 1.9554434641408215e-07, "loss": 0.3293, "step": 18021 }, { "epoch": 0.9116293186301786, "grad_norm": 8.965350151062012, "learning_rate": 1.9532214323725985e-07, "loss": 0.4184, "step": 18022 }, { "epoch": 0.9116799028782437, "grad_norm": 9.74561595916748, "learning_rate": 1.9510006386538049e-07, "loss": 0.3635, "step": 18023 }, { "epoch": 0.9117304871263089, "grad_norm": 11.053055763244629, "learning_rate": 1.948781083041673e-07, "loss": 0.5219, "step": 18024 }, { "epoch": 0.911781071374374, "grad_norm": 6.613461971282959, "learning_rate": 1.9465627655933904e-07, "loss": 0.3876, "step": 18025 }, { "epoch": 0.9118316556224392, "grad_norm": 7.164747714996338, "learning_rate": 1.944345686366117e-07, "loss": 0.3969, "step": 18026 }, { "epoch": 0.9118822398705043, "grad_norm": 4.446811199188232, "learning_rate": 1.9421298454169791e-07, "loss": 0.4432, "step": 18027 }, { "epoch": 0.9119328241185695, "grad_norm": 5.832756519317627, "learning_rate": 1.9399152428030753e-07, "loss": 0.3108, "step": 18028 }, { "epoch": 0.9119834083666346, "grad_norm": 13.716157913208008, "learning_rate": 1.9377018785814717e-07, "loss": 0.4624, "step": 18029 }, { "epoch": 0.9120339926146998, "grad_norm": 10.371926307678223, "learning_rate": 1.9354897528091998e-07, "loss": 0.3588, "step": 18030 }, { "epoch": 0.9120845768627649, "grad_norm": 7.252865791320801, "learning_rate": 1.9332788655432645e-07, "loss": 0.403, "step": 18031 }, { "epoch": 0.91213516111083, "grad_norm": 12.09527587890625, "learning_rate": 1.9310692168406253e-07, "loss": 0.4927, "step": 18032 }, { "epoch": 0.9121857453588953, "grad_norm": 6.157557487487793, "learning_rate": 1.928860806758237e-07, "loss": 0.3793, "step": 18033 }, { "epoch": 0.9122363296069604, "grad_norm": 7.385778427124023, "learning_rate": 1.926653635352993e-07, "loss": 0.3301, "step": 18034 }, { "epoch": 0.9122869138550256, "grad_norm": 10.51004695892334, "learning_rate": 1.9244477026817589e-07, "loss": 0.4558, "step": 18035 }, { "epoch": 0.9123374981030907, "grad_norm": 12.832417488098145, "learning_rate": 1.9222430088013943e-07, "loss": 0.4518, "step": 18036 }, { "epoch": 0.9123880823511559, "grad_norm": 18.12163734436035, "learning_rate": 1.920039553768699e-07, "loss": 0.6566, "step": 18037 }, { "epoch": 0.912438666599221, "grad_norm": 10.917858123779297, "learning_rate": 1.9178373376404492e-07, "loss": 0.4455, "step": 18038 }, { "epoch": 0.9124892508472862, "grad_norm": 11.125896453857422, "learning_rate": 1.9156363604733997e-07, "loss": 0.4381, "step": 18039 }, { "epoch": 0.9125398350953513, "grad_norm": 9.151320457458496, "learning_rate": 1.9134366223242495e-07, "loss": 0.3282, "step": 18040 }, { "epoch": 0.9125904193434164, "grad_norm": 9.417546272277832, "learning_rate": 1.911238123249698e-07, "loss": 0.3467, "step": 18041 }, { "epoch": 0.9126410035914816, "grad_norm": 6.595815181732178, "learning_rate": 1.9090408633063827e-07, "loss": 0.3507, "step": 18042 }, { "epoch": 0.9126915878395467, "grad_norm": 8.649494171142578, "learning_rate": 1.9068448425509202e-07, "loss": 0.3936, "step": 18043 }, { "epoch": 0.9127421720876119, "grad_norm": 6.7849555015563965, "learning_rate": 1.9046500610399087e-07, "loss": 0.2516, "step": 18044 }, { "epoch": 0.912792756335677, "grad_norm": 8.761757850646973, "learning_rate": 1.9024565188298926e-07, "loss": 0.4175, "step": 18045 }, { "epoch": 0.9128433405837422, "grad_norm": 6.925809383392334, "learning_rate": 1.9002642159773988e-07, "loss": 0.4015, "step": 18046 }, { "epoch": 0.9128939248318074, "grad_norm": 6.555146217346191, "learning_rate": 1.898073152538915e-07, "loss": 0.2997, "step": 18047 }, { "epoch": 0.9129445090798726, "grad_norm": 7.3047661781311035, "learning_rate": 1.8958833285709022e-07, "loss": 0.3938, "step": 18048 }, { "epoch": 0.9129950933279377, "grad_norm": 8.176226615905762, "learning_rate": 1.8936947441297815e-07, "loss": 0.4316, "step": 18049 }, { "epoch": 0.9130456775760029, "grad_norm": 7.302428245544434, "learning_rate": 1.8915073992719525e-07, "loss": 0.3426, "step": 18050 }, { "epoch": 0.913096261824068, "grad_norm": 6.027010440826416, "learning_rate": 1.889321294053781e-07, "loss": 0.3918, "step": 18051 }, { "epoch": 0.9131468460721331, "grad_norm": 11.875405311584473, "learning_rate": 1.8871364285315885e-07, "loss": 0.5125, "step": 18052 }, { "epoch": 0.9131974303201983, "grad_norm": 12.955118179321289, "learning_rate": 1.8849528027616803e-07, "loss": 0.4128, "step": 18053 }, { "epoch": 0.9132480145682634, "grad_norm": 8.260509490966797, "learning_rate": 1.8827704168003225e-07, "loss": 0.3942, "step": 18054 }, { "epoch": 0.9132985988163286, "grad_norm": 5.691478252410889, "learning_rate": 1.880589270703742e-07, "loss": 0.3334, "step": 18055 }, { "epoch": 0.9133491830643937, "grad_norm": 4.976758003234863, "learning_rate": 1.8784093645281553e-07, "loss": 0.3872, "step": 18056 }, { "epoch": 0.9133997673124589, "grad_norm": 7.135361194610596, "learning_rate": 1.8762306983297286e-07, "loss": 0.1883, "step": 18057 }, { "epoch": 0.913450351560524, "grad_norm": 4.645650386810303, "learning_rate": 1.8740532721645887e-07, "loss": 0.3816, "step": 18058 }, { "epoch": 0.9135009358085892, "grad_norm": 6.725902080535889, "learning_rate": 1.871877086088858e-07, "loss": 0.3783, "step": 18059 }, { "epoch": 0.9135515200566544, "grad_norm": 10.445920944213867, "learning_rate": 1.869702140158608e-07, "loss": 0.3799, "step": 18060 }, { "epoch": 0.9136021043047196, "grad_norm": 7.212690830230713, "learning_rate": 1.8675284344298771e-07, "loss": 0.3546, "step": 18061 }, { "epoch": 0.9136526885527847, "grad_norm": 6.569987773895264, "learning_rate": 1.865355968958682e-07, "loss": 0.3801, "step": 18062 }, { "epoch": 0.9137032728008498, "grad_norm": 8.701798439025879, "learning_rate": 1.8631847438009943e-07, "loss": 0.493, "step": 18063 }, { "epoch": 0.913753857048915, "grad_norm": 5.56714391708374, "learning_rate": 1.8610147590127636e-07, "loss": 0.4073, "step": 18064 }, { "epoch": 0.9138044412969801, "grad_norm": 8.19229793548584, "learning_rate": 1.8588460146499232e-07, "loss": 0.2601, "step": 18065 }, { "epoch": 0.9138550255450453, "grad_norm": 8.207348823547363, "learning_rate": 1.8566785107683339e-07, "loss": 0.4067, "step": 18066 }, { "epoch": 0.9139056097931104, "grad_norm": 9.194639205932617, "learning_rate": 1.8545122474238452e-07, "loss": 0.4099, "step": 18067 }, { "epoch": 0.9139561940411756, "grad_norm": 11.100072860717773, "learning_rate": 1.8523472246722907e-07, "loss": 0.4043, "step": 18068 }, { "epoch": 0.9140067782892407, "grad_norm": 6.5509538650512695, "learning_rate": 1.8501834425694533e-07, "loss": 0.4105, "step": 18069 }, { "epoch": 0.9140573625373059, "grad_norm": 8.616522789001465, "learning_rate": 1.8480209011710826e-07, "loss": 0.494, "step": 18070 }, { "epoch": 0.914107946785371, "grad_norm": 8.166199684143066, "learning_rate": 1.845859600532912e-07, "loss": 0.3793, "step": 18071 }, { "epoch": 0.9141585310334361, "grad_norm": 8.433119773864746, "learning_rate": 1.8436995407106307e-07, "loss": 0.4511, "step": 18072 }, { "epoch": 0.9142091152815014, "grad_norm": 8.344623565673828, "learning_rate": 1.8415407217598824e-07, "loss": 0.4748, "step": 18073 }, { "epoch": 0.9142596995295665, "grad_norm": 10.827736854553223, "learning_rate": 1.839383143736323e-07, "loss": 0.4186, "step": 18074 }, { "epoch": 0.9143102837776317, "grad_norm": 5.015081405639648, "learning_rate": 1.8372268066955245e-07, "loss": 0.2404, "step": 18075 }, { "epoch": 0.9143608680256968, "grad_norm": 10.388976097106934, "learning_rate": 1.8350717106930594e-07, "loss": 0.3678, "step": 18076 }, { "epoch": 0.914411452273762, "grad_norm": 7.817788600921631, "learning_rate": 1.8329178557844662e-07, "loss": 0.4634, "step": 18077 }, { "epoch": 0.9144620365218271, "grad_norm": 7.119024276733398, "learning_rate": 1.8307652420252286e-07, "loss": 0.4371, "step": 18078 }, { "epoch": 0.9145126207698923, "grad_norm": 10.729769706726074, "learning_rate": 1.8286138694708245e-07, "loss": 0.4966, "step": 18079 }, { "epoch": 0.9145632050179574, "grad_norm": 9.014479637145996, "learning_rate": 1.826463738176687e-07, "loss": 0.3949, "step": 18080 }, { "epoch": 0.9146137892660225, "grad_norm": 4.65847110748291, "learning_rate": 1.824314848198222e-07, "loss": 0.3439, "step": 18081 }, { "epoch": 0.9146643735140877, "grad_norm": 14.488327980041504, "learning_rate": 1.8221671995907964e-07, "loss": 0.6291, "step": 18082 }, { "epoch": 0.9147149577621528, "grad_norm": 13.27291488647461, "learning_rate": 1.8200207924097602e-07, "loss": 0.4406, "step": 18083 }, { "epoch": 0.914765542010218, "grad_norm": 10.217008590698242, "learning_rate": 1.817875626710408e-07, "loss": 0.4669, "step": 18084 }, { "epoch": 0.9148161262582831, "grad_norm": 13.03242301940918, "learning_rate": 1.8157317025480238e-07, "loss": 0.4699, "step": 18085 }, { "epoch": 0.9148667105063483, "grad_norm": 5.696788311004639, "learning_rate": 1.8135890199778516e-07, "loss": 0.2357, "step": 18086 }, { "epoch": 0.9149172947544135, "grad_norm": 8.9689359664917, "learning_rate": 1.8114475790551035e-07, "loss": 0.4575, "step": 18087 }, { "epoch": 0.9149678790024787, "grad_norm": 12.122095108032227, "learning_rate": 1.8093073798349458e-07, "loss": 0.4347, "step": 18088 }, { "epoch": 0.9150184632505438, "grad_norm": 12.600303649902344, "learning_rate": 1.8071684223725462e-07, "loss": 0.4675, "step": 18089 }, { "epoch": 0.915069047498609, "grad_norm": 5.211336135864258, "learning_rate": 1.8050307067230098e-07, "loss": 0.2947, "step": 18090 }, { "epoch": 0.9151196317466741, "grad_norm": 9.265625953674316, "learning_rate": 1.8028942329414212e-07, "loss": 0.4311, "step": 18091 }, { "epoch": 0.9151702159947392, "grad_norm": 6.082961082458496, "learning_rate": 1.80075900108283e-07, "loss": 0.4558, "step": 18092 }, { "epoch": 0.9152208002428044, "grad_norm": 5.114018440246582, "learning_rate": 1.7986250112022597e-07, "loss": 0.3193, "step": 18093 }, { "epoch": 0.9152713844908695, "grad_norm": 6.435672760009766, "learning_rate": 1.7964922633546988e-07, "loss": 0.2832, "step": 18094 }, { "epoch": 0.9153219687389347, "grad_norm": 10.657188415527344, "learning_rate": 1.7943607575951038e-07, "loss": 0.5325, "step": 18095 }, { "epoch": 0.9153725529869998, "grad_norm": 5.711996555328369, "learning_rate": 1.7922304939783918e-07, "loss": 0.4171, "step": 18096 }, { "epoch": 0.915423137235065, "grad_norm": 9.95085620880127, "learning_rate": 1.790101472559458e-07, "loss": 0.4638, "step": 18097 }, { "epoch": 0.9154737214831301, "grad_norm": 7.78725004196167, "learning_rate": 1.787973693393169e-07, "loss": 0.2901, "step": 18098 }, { "epoch": 0.9155243057311953, "grad_norm": 8.587822914123535, "learning_rate": 1.7858471565343373e-07, "loss": 0.4766, "step": 18099 }, { "epoch": 0.9155748899792605, "grad_norm": 6.374377250671387, "learning_rate": 1.7837218620377793e-07, "loss": 0.342, "step": 18100 }, { "epoch": 0.9156254742273257, "grad_norm": 7.46852445602417, "learning_rate": 1.781597809958241e-07, "loss": 0.3852, "step": 18101 }, { "epoch": 0.9156760584753908, "grad_norm": 5.730083465576172, "learning_rate": 1.7794750003504558e-07, "loss": 0.3711, "step": 18102 }, { "epoch": 0.9157266427234559, "grad_norm": 11.533530235290527, "learning_rate": 1.777353433269141e-07, "loss": 0.5533, "step": 18103 }, { "epoch": 0.9157772269715211, "grad_norm": 11.221522331237793, "learning_rate": 1.7752331087689478e-07, "loss": 0.456, "step": 18104 }, { "epoch": 0.9158278112195862, "grad_norm": 7.970518589019775, "learning_rate": 1.7731140269045043e-07, "loss": 0.3759, "step": 18105 }, { "epoch": 0.9158783954676514, "grad_norm": 7.307186603546143, "learning_rate": 1.7709961877304338e-07, "loss": 0.4497, "step": 18106 }, { "epoch": 0.9159289797157165, "grad_norm": 16.199399948120117, "learning_rate": 1.768879591301298e-07, "loss": 0.5045, "step": 18107 }, { "epoch": 0.9159795639637817, "grad_norm": 8.74221420288086, "learning_rate": 1.766764237671631e-07, "loss": 0.3081, "step": 18108 }, { "epoch": 0.9160301482118468, "grad_norm": 10.499842643737793, "learning_rate": 1.764650126895956e-07, "loss": 0.3151, "step": 18109 }, { "epoch": 0.916080732459912, "grad_norm": 4.970210075378418, "learning_rate": 1.7625372590287348e-07, "loss": 0.3891, "step": 18110 }, { "epoch": 0.9161313167079771, "grad_norm": 9.250924110412598, "learning_rate": 1.760425634124413e-07, "loss": 0.4151, "step": 18111 }, { "epoch": 0.9161819009560422, "grad_norm": 8.391253471374512, "learning_rate": 1.7583152522374135e-07, "loss": 0.3999, "step": 18112 }, { "epoch": 0.9162324852041075, "grad_norm": 5.525409698486328, "learning_rate": 1.7562061134221042e-07, "loss": 0.392, "step": 18113 }, { "epoch": 0.9162830694521726, "grad_norm": 11.71478271484375, "learning_rate": 1.7540982177328247e-07, "loss": 0.4109, "step": 18114 }, { "epoch": 0.9163336537002378, "grad_norm": 10.845447540283203, "learning_rate": 1.751991565223915e-07, "loss": 0.3934, "step": 18115 }, { "epoch": 0.9163842379483029, "grad_norm": 13.666065216064453, "learning_rate": 1.749886155949637e-07, "loss": 0.5374, "step": 18116 }, { "epoch": 0.9164348221963681, "grad_norm": 11.963687896728516, "learning_rate": 1.7477819899642535e-07, "loss": 0.4379, "step": 18117 }, { "epoch": 0.9164854064444332, "grad_norm": 15.182648658752441, "learning_rate": 1.745679067321976e-07, "loss": 0.5734, "step": 18118 }, { "epoch": 0.9165359906924984, "grad_norm": 9.23523998260498, "learning_rate": 1.7435773880769947e-07, "loss": 0.4881, "step": 18119 }, { "epoch": 0.9165865749405635, "grad_norm": 7.804490566253662, "learning_rate": 1.7414769522834718e-07, "loss": 0.3917, "step": 18120 }, { "epoch": 0.9166371591886286, "grad_norm": 6.372288227081299, "learning_rate": 1.7393777599955196e-07, "loss": 0.2887, "step": 18121 }, { "epoch": 0.9166877434366938, "grad_norm": 6.167909622192383, "learning_rate": 1.7372798112672284e-07, "loss": 0.3542, "step": 18122 }, { "epoch": 0.9167383276847589, "grad_norm": 7.76754093170166, "learning_rate": 1.7351831061526715e-07, "loss": 0.4724, "step": 18123 }, { "epoch": 0.9167889119328241, "grad_norm": 7.207449436187744, "learning_rate": 1.7330876447058665e-07, "loss": 0.3165, "step": 18124 }, { "epoch": 0.9168394961808892, "grad_norm": 5.1361823081970215, "learning_rate": 1.7309934269808038e-07, "loss": 0.285, "step": 18125 }, { "epoch": 0.9168900804289544, "grad_norm": 8.531923294067383, "learning_rate": 1.7289004530314514e-07, "loss": 0.4399, "step": 18126 }, { "epoch": 0.9169406646770196, "grad_norm": 9.278826713562012, "learning_rate": 1.7268087229117437e-07, "loss": 0.4069, "step": 18127 }, { "epoch": 0.9169912489250848, "grad_norm": 10.18551254272461, "learning_rate": 1.7247182366755765e-07, "loss": 0.4114, "step": 18128 }, { "epoch": 0.9170418331731499, "grad_norm": 9.939041137695312, "learning_rate": 1.7226289943768182e-07, "loss": 0.3697, "step": 18129 }, { "epoch": 0.9170924174212151, "grad_norm": 11.66992473602295, "learning_rate": 1.7205409960692976e-07, "loss": 0.4647, "step": 18130 }, { "epoch": 0.9171430016692802, "grad_norm": 7.5615644454956055, "learning_rate": 1.7184542418068162e-07, "loss": 0.3362, "step": 18131 }, { "epoch": 0.9171935859173453, "grad_norm": 6.2392048835754395, "learning_rate": 1.7163687316431587e-07, "loss": 0.3337, "step": 18132 }, { "epoch": 0.9172441701654105, "grad_norm": 6.064826965332031, "learning_rate": 1.7142844656320546e-07, "loss": 0.4265, "step": 18133 }, { "epoch": 0.9172947544134756, "grad_norm": 9.300244331359863, "learning_rate": 1.7122014438271994e-07, "loss": 0.4109, "step": 18134 }, { "epoch": 0.9173453386615408, "grad_norm": 11.876538276672363, "learning_rate": 1.7101196662822894e-07, "loss": 0.4456, "step": 18135 }, { "epoch": 0.9173959229096059, "grad_norm": 8.907413482666016, "learning_rate": 1.708039133050954e-07, "loss": 0.4053, "step": 18136 }, { "epoch": 0.9174465071576711, "grad_norm": 7.3484015464782715, "learning_rate": 1.7059598441867996e-07, "loss": 0.4298, "step": 18137 }, { "epoch": 0.9174970914057362, "grad_norm": 8.75196647644043, "learning_rate": 1.7038817997434175e-07, "loss": 0.3744, "step": 18138 }, { "epoch": 0.9175476756538014, "grad_norm": 11.920894622802734, "learning_rate": 1.701804999774348e-07, "loss": 0.4302, "step": 18139 }, { "epoch": 0.9175982599018666, "grad_norm": 12.674782752990723, "learning_rate": 1.699729444333098e-07, "loss": 0.3986, "step": 18140 }, { "epoch": 0.9176488441499318, "grad_norm": 5.998384475708008, "learning_rate": 1.6976551334731585e-07, "loss": 0.3412, "step": 18141 }, { "epoch": 0.9176994283979969, "grad_norm": 14.489373207092285, "learning_rate": 1.6955820672479862e-07, "loss": 0.3879, "step": 18142 }, { "epoch": 0.917750012646062, "grad_norm": 14.90967845916748, "learning_rate": 1.693510245710972e-07, "loss": 0.3908, "step": 18143 }, { "epoch": 0.9178005968941272, "grad_norm": 16.018856048583984, "learning_rate": 1.6914396689155343e-07, "loss": 0.4041, "step": 18144 }, { "epoch": 0.9178511811421923, "grad_norm": 6.631443023681641, "learning_rate": 1.6893703369150027e-07, "loss": 0.2831, "step": 18145 }, { "epoch": 0.9179017653902575, "grad_norm": 6.05592155456543, "learning_rate": 1.6873022497627066e-07, "loss": 0.3853, "step": 18146 }, { "epoch": 0.9179523496383226, "grad_norm": 6.791196346282959, "learning_rate": 1.6852354075119425e-07, "loss": 0.4506, "step": 18147 }, { "epoch": 0.9180029338863878, "grad_norm": 5.465024471282959, "learning_rate": 1.683169810215962e-07, "loss": 0.4098, "step": 18148 }, { "epoch": 0.9180535181344529, "grad_norm": 10.834578514099121, "learning_rate": 1.6811054579279839e-07, "loss": 0.3048, "step": 18149 }, { "epoch": 0.918104102382518, "grad_norm": 8.87258529663086, "learning_rate": 1.679042350701221e-07, "loss": 0.5184, "step": 18150 }, { "epoch": 0.9181546866305832, "grad_norm": 9.836018562316895, "learning_rate": 1.6769804885888142e-07, "loss": 0.4882, "step": 18151 }, { "epoch": 0.9182052708786483, "grad_norm": 7.950829029083252, "learning_rate": 1.6749198716438986e-07, "loss": 0.325, "step": 18152 }, { "epoch": 0.9182558551267136, "grad_norm": 47.423160552978516, "learning_rate": 1.6728604999195764e-07, "loss": 0.4499, "step": 18153 }, { "epoch": 0.9183064393747787, "grad_norm": 14.1379976272583, "learning_rate": 1.670802373468916e-07, "loss": 0.3638, "step": 18154 }, { "epoch": 0.9183570236228439, "grad_norm": 13.274833679199219, "learning_rate": 1.6687454923449366e-07, "loss": 0.4185, "step": 18155 }, { "epoch": 0.918407607870909, "grad_norm": 13.674722671508789, "learning_rate": 1.6666898566006506e-07, "loss": 0.4461, "step": 18156 }, { "epoch": 0.9184581921189742, "grad_norm": 7.821558475494385, "learning_rate": 1.6646354662890163e-07, "loss": 0.3811, "step": 18157 }, { "epoch": 0.9185087763670393, "grad_norm": 13.413227081298828, "learning_rate": 1.66258232146298e-07, "loss": 0.4285, "step": 18158 }, { "epoch": 0.9185593606151045, "grad_norm": 6.724432945251465, "learning_rate": 1.6605304221754493e-07, "loss": 0.3943, "step": 18159 }, { "epoch": 0.9186099448631696, "grad_norm": 9.46111011505127, "learning_rate": 1.658479768479282e-07, "loss": 0.445, "step": 18160 }, { "epoch": 0.9186605291112347, "grad_norm": 7.161109447479248, "learning_rate": 1.6564303604273302e-07, "loss": 0.3282, "step": 18161 }, { "epoch": 0.9187111133592999, "grad_norm": 7.1661834716796875, "learning_rate": 1.654382198072396e-07, "loss": 0.4153, "step": 18162 }, { "epoch": 0.918761697607365, "grad_norm": 6.647676467895508, "learning_rate": 1.6523352814672656e-07, "loss": 0.4093, "step": 18163 }, { "epoch": 0.9188122818554302, "grad_norm": 7.708488941192627, "learning_rate": 1.6502896106646681e-07, "loss": 0.4359, "step": 18164 }, { "epoch": 0.9188628661034953, "grad_norm": 8.143437385559082, "learning_rate": 1.648245185717323e-07, "loss": 0.3997, "step": 18165 }, { "epoch": 0.9189134503515606, "grad_norm": 7.870071887969971, "learning_rate": 1.646202006677916e-07, "loss": 0.4238, "step": 18166 }, { "epoch": 0.9189640345996257, "grad_norm": 9.133613586425781, "learning_rate": 1.644160073599088e-07, "loss": 0.4196, "step": 18167 }, { "epoch": 0.9190146188476909, "grad_norm": 12.446087837219238, "learning_rate": 1.6421193865334584e-07, "loss": 0.537, "step": 18168 }, { "epoch": 0.919065203095756, "grad_norm": 11.99877643585205, "learning_rate": 1.640079945533607e-07, "loss": 0.5179, "step": 18169 }, { "epoch": 0.9191157873438212, "grad_norm": 9.10815143585205, "learning_rate": 1.6380417506520862e-07, "loss": 0.4318, "step": 18170 }, { "epoch": 0.9191663715918863, "grad_norm": 10.170339584350586, "learning_rate": 1.6360048019414154e-07, "loss": 0.4406, "step": 18171 }, { "epoch": 0.9192169558399514, "grad_norm": 13.217164993286133, "learning_rate": 1.6339690994540802e-07, "loss": 0.4636, "step": 18172 }, { "epoch": 0.9192675400880166, "grad_norm": 13.006644248962402, "learning_rate": 1.6319346432425386e-07, "loss": 0.4247, "step": 18173 }, { "epoch": 0.9193181243360817, "grad_norm": 7.711638450622559, "learning_rate": 1.6299014333592156e-07, "loss": 0.3575, "step": 18174 }, { "epoch": 0.9193687085841469, "grad_norm": 5.566353797912598, "learning_rate": 1.6278694698564968e-07, "loss": 0.3346, "step": 18175 }, { "epoch": 0.919419292832212, "grad_norm": 10.800376892089844, "learning_rate": 1.6258387527867457e-07, "loss": 0.3759, "step": 18176 }, { "epoch": 0.9194698770802772, "grad_norm": 4.840705394744873, "learning_rate": 1.6238092822022878e-07, "loss": 0.4065, "step": 18177 }, { "epoch": 0.9195204613283423, "grad_norm": 6.1718831062316895, "learning_rate": 1.621781058155414e-07, "loss": 0.3799, "step": 18178 }, { "epoch": 0.9195710455764075, "grad_norm": 5.382175922393799, "learning_rate": 1.6197540806983937e-07, "loss": 0.384, "step": 18179 }, { "epoch": 0.9196216298244727, "grad_norm": 5.318973064422607, "learning_rate": 1.6177283498834574e-07, "loss": 0.3571, "step": 18180 }, { "epoch": 0.9196722140725379, "grad_norm": 9.00788688659668, "learning_rate": 1.6157038657627856e-07, "loss": 0.3183, "step": 18181 }, { "epoch": 0.919722798320603, "grad_norm": 9.692037582397461, "learning_rate": 1.613680628388564e-07, "loss": 0.4106, "step": 18182 }, { "epoch": 0.9197733825686681, "grad_norm": 25.355440139770508, "learning_rate": 1.611658637812924e-07, "loss": 0.6522, "step": 18183 }, { "epoch": 0.9198239668167333, "grad_norm": 8.130667686462402, "learning_rate": 1.6096378940879564e-07, "loss": 0.4436, "step": 18184 }, { "epoch": 0.9198745510647984, "grad_norm": 6.43588924407959, "learning_rate": 1.6076183972657421e-07, "loss": 0.3765, "step": 18185 }, { "epoch": 0.9199251353128636, "grad_norm": 9.717053413391113, "learning_rate": 1.6056001473983174e-07, "loss": 0.348, "step": 18186 }, { "epoch": 0.9199757195609287, "grad_norm": 5.4417500495910645, "learning_rate": 1.6035831445376792e-07, "loss": 0.3565, "step": 18187 }, { "epoch": 0.9200263038089939, "grad_norm": 6.938635349273682, "learning_rate": 1.6015673887358086e-07, "loss": 0.252, "step": 18188 }, { "epoch": 0.920076888057059, "grad_norm": 10.978331565856934, "learning_rate": 1.599552880044647e-07, "loss": 0.3753, "step": 18189 }, { "epoch": 0.9201274723051241, "grad_norm": 8.206560134887695, "learning_rate": 1.5975396185160974e-07, "loss": 0.3297, "step": 18190 }, { "epoch": 0.9201780565531893, "grad_norm": 8.687435150146484, "learning_rate": 1.5955276042020407e-07, "loss": 0.4488, "step": 18191 }, { "epoch": 0.9202286408012544, "grad_norm": 4.80674409866333, "learning_rate": 1.5935168371543296e-07, "loss": 0.2378, "step": 18192 }, { "epoch": 0.9202792250493197, "grad_norm": 10.089887619018555, "learning_rate": 1.591507317424762e-07, "loss": 0.5193, "step": 18193 }, { "epoch": 0.9203298092973848, "grad_norm": 11.097654342651367, "learning_rate": 1.5894990450651294e-07, "loss": 0.4887, "step": 18194 }, { "epoch": 0.92038039354545, "grad_norm": 11.78904914855957, "learning_rate": 1.5874920201271627e-07, "loss": 0.487, "step": 18195 }, { "epoch": 0.9204309777935151, "grad_norm": 7.633514404296875, "learning_rate": 1.585486242662604e-07, "loss": 0.3875, "step": 18196 }, { "epoch": 0.9204815620415803, "grad_norm": 10.606926918029785, "learning_rate": 1.5834817127231227e-07, "loss": 0.3348, "step": 18197 }, { "epoch": 0.9205321462896454, "grad_norm": 17.457815170288086, "learning_rate": 1.5814784303603726e-07, "loss": 0.5348, "step": 18198 }, { "epoch": 0.9205827305377106, "grad_norm": 9.441176414489746, "learning_rate": 1.5794763956259618e-07, "loss": 0.4122, "step": 18199 }, { "epoch": 0.9206333147857757, "grad_norm": 5.734821796417236, "learning_rate": 1.5774756085715047e-07, "loss": 0.3572, "step": 18200 }, { "epoch": 0.9206838990338408, "grad_norm": 5.338158130645752, "learning_rate": 1.5754760692485328e-07, "loss": 0.3138, "step": 18201 }, { "epoch": 0.920734483281906, "grad_norm": 5.923393249511719, "learning_rate": 1.5734777777085763e-07, "loss": 0.3483, "step": 18202 }, { "epoch": 0.9207850675299711, "grad_norm": 19.814655303955078, "learning_rate": 1.5714807340031335e-07, "loss": 0.4572, "step": 18203 }, { "epoch": 0.9208356517780363, "grad_norm": 10.500731468200684, "learning_rate": 1.5694849381836575e-07, "loss": 0.4229, "step": 18204 }, { "epoch": 0.9208862360261014, "grad_norm": 4.811825275421143, "learning_rate": 1.5674903903015681e-07, "loss": 0.3076, "step": 18205 }, { "epoch": 0.9209368202741667, "grad_norm": 10.22235107421875, "learning_rate": 1.56549709040828e-07, "loss": 0.4458, "step": 18206 }, { "epoch": 0.9209874045222318, "grad_norm": 6.292056083679199, "learning_rate": 1.5635050385551354e-07, "loss": 0.4241, "step": 18207 }, { "epoch": 0.921037988770297, "grad_norm": 5.186081409454346, "learning_rate": 1.5615142347934765e-07, "loss": 0.3569, "step": 18208 }, { "epoch": 0.9210885730183621, "grad_norm": 7.538778781890869, "learning_rate": 1.5595246791745955e-07, "loss": 0.3638, "step": 18209 }, { "epoch": 0.9211391572664273, "grad_norm": 5.762359142303467, "learning_rate": 1.5575363717497516e-07, "loss": 0.4576, "step": 18210 }, { "epoch": 0.9211897415144924, "grad_norm": 5.696913242340088, "learning_rate": 1.5555493125701982e-07, "loss": 0.2309, "step": 18211 }, { "epoch": 0.9212403257625575, "grad_norm": 9.844914436340332, "learning_rate": 1.553563501687122e-07, "loss": 0.4661, "step": 18212 }, { "epoch": 0.9212909100106227, "grad_norm": 9.237236022949219, "learning_rate": 1.5515789391516934e-07, "loss": 0.3601, "step": 18213 }, { "epoch": 0.9213414942586878, "grad_norm": 13.561058044433594, "learning_rate": 1.54959562501506e-07, "loss": 0.5689, "step": 18214 }, { "epoch": 0.921392078506753, "grad_norm": 9.789212226867676, "learning_rate": 1.5476135593283205e-07, "loss": 0.3915, "step": 18215 }, { "epoch": 0.9214426627548181, "grad_norm": 7.516702651977539, "learning_rate": 1.5456327421425388e-07, "loss": 0.4722, "step": 18216 }, { "epoch": 0.9214932470028833, "grad_norm": 6.734705924987793, "learning_rate": 1.5436531735087745e-07, "loss": 0.3626, "step": 18217 }, { "epoch": 0.9215438312509484, "grad_norm": 6.188374042510986, "learning_rate": 1.5416748534780257e-07, "loss": 0.3248, "step": 18218 }, { "epoch": 0.9215944154990136, "grad_norm": 8.859004020690918, "learning_rate": 1.539697782101268e-07, "loss": 0.3998, "step": 18219 }, { "epoch": 0.9216449997470788, "grad_norm": 9.181190490722656, "learning_rate": 1.5377219594294445e-07, "loss": 0.3959, "step": 18220 }, { "epoch": 0.921695583995144, "grad_norm": 6.295042037963867, "learning_rate": 1.5357473855134754e-07, "loss": 0.3765, "step": 18221 }, { "epoch": 0.9217461682432091, "grad_norm": 6.319025993347168, "learning_rate": 1.533774060404225e-07, "loss": 0.3358, "step": 18222 }, { "epoch": 0.9217967524912742, "grad_norm": 9.121735572814941, "learning_rate": 1.5318019841525645e-07, "loss": 0.4673, "step": 18223 }, { "epoch": 0.9218473367393394, "grad_norm": 18.637231826782227, "learning_rate": 1.5298311568092916e-07, "loss": 0.3171, "step": 18224 }, { "epoch": 0.9218979209874045, "grad_norm": 7.822846412658691, "learning_rate": 1.5278615784251883e-07, "loss": 0.3805, "step": 18225 }, { "epoch": 0.9219485052354697, "grad_norm": 8.90677547454834, "learning_rate": 1.5258932490510193e-07, "loss": 0.3095, "step": 18226 }, { "epoch": 0.9219990894835348, "grad_norm": 9.52242374420166, "learning_rate": 1.5239261687374994e-07, "loss": 0.4087, "step": 18227 }, { "epoch": 0.9220496737316, "grad_norm": 9.68030071258545, "learning_rate": 1.5219603375353052e-07, "loss": 0.3575, "step": 18228 }, { "epoch": 0.9221002579796651, "grad_norm": 6.516642093658447, "learning_rate": 1.5199957554951017e-07, "loss": 0.4589, "step": 18229 }, { "epoch": 0.9221508422277302, "grad_norm": 7.025172233581543, "learning_rate": 1.5180324226675148e-07, "loss": 0.3619, "step": 18230 }, { "epoch": 0.9222014264757954, "grad_norm": 9.34997844696045, "learning_rate": 1.5160703391031206e-07, "loss": 0.3829, "step": 18231 }, { "epoch": 0.9222520107238605, "grad_norm": 9.915432929992676, "learning_rate": 1.5141095048524957e-07, "loss": 0.4872, "step": 18232 }, { "epoch": 0.9223025949719258, "grad_norm": 10.284164428710938, "learning_rate": 1.5121499199661493e-07, "loss": 0.4738, "step": 18233 }, { "epoch": 0.9223531792199909, "grad_norm": 8.24554443359375, "learning_rate": 1.5101915844945802e-07, "loss": 0.3825, "step": 18234 }, { "epoch": 0.9224037634680561, "grad_norm": 6.656525135040283, "learning_rate": 1.5082344984882536e-07, "loss": 0.3796, "step": 18235 }, { "epoch": 0.9224543477161212, "grad_norm": 8.040014266967773, "learning_rate": 1.506278661997601e-07, "loss": 0.4213, "step": 18236 }, { "epoch": 0.9225049319641864, "grad_norm": 9.378387451171875, "learning_rate": 1.5043240750730103e-07, "loss": 0.4244, "step": 18237 }, { "epoch": 0.9225555162122515, "grad_norm": 7.368129253387451, "learning_rate": 1.502370737764852e-07, "loss": 0.451, "step": 18238 }, { "epoch": 0.9226061004603167, "grad_norm": 8.773724555969238, "learning_rate": 1.500418650123464e-07, "loss": 0.4328, "step": 18239 }, { "epoch": 0.9226566847083818, "grad_norm": 9.478652000427246, "learning_rate": 1.4984678121991337e-07, "loss": 0.3858, "step": 18240 }, { "epoch": 0.922707268956447, "grad_norm": 5.894940376281738, "learning_rate": 1.4965182240421427e-07, "loss": 0.3964, "step": 18241 }, { "epoch": 0.9227578532045121, "grad_norm": 7.894411087036133, "learning_rate": 1.494569885702718e-07, "loss": 0.3037, "step": 18242 }, { "epoch": 0.9228084374525772, "grad_norm": 10.902976036071777, "learning_rate": 1.4926227972310637e-07, "loss": 0.4117, "step": 18243 }, { "epoch": 0.9228590217006424, "grad_norm": 4.532740116119385, "learning_rate": 1.4906769586773618e-07, "loss": 0.3649, "step": 18244 }, { "epoch": 0.9229096059487075, "grad_norm": 6.576323509216309, "learning_rate": 1.48873237009175e-07, "loss": 0.4089, "step": 18245 }, { "epoch": 0.9229601901967728, "grad_norm": 8.389917373657227, "learning_rate": 1.4867890315243215e-07, "loss": 0.4359, "step": 18246 }, { "epoch": 0.9230107744448379, "grad_norm": 9.063044548034668, "learning_rate": 1.48484694302517e-07, "loss": 0.3771, "step": 18247 }, { "epoch": 0.9230613586929031, "grad_norm": 9.2915620803833, "learning_rate": 1.482906104644316e-07, "loss": 0.4864, "step": 18248 }, { "epoch": 0.9231119429409682, "grad_norm": 4.6077046394348145, "learning_rate": 1.480966516431792e-07, "loss": 0.255, "step": 18249 }, { "epoch": 0.9231625271890334, "grad_norm": 7.8242506980896, "learning_rate": 1.4790281784375694e-07, "loss": 0.2892, "step": 18250 }, { "epoch": 0.9232131114370985, "grad_norm": 9.950349807739258, "learning_rate": 1.4770910907115854e-07, "loss": 0.2804, "step": 18251 }, { "epoch": 0.9232636956851636, "grad_norm": 8.766918182373047, "learning_rate": 1.475155253303767e-07, "loss": 0.4224, "step": 18252 }, { "epoch": 0.9233142799332288, "grad_norm": 10.773347854614258, "learning_rate": 1.4732206662639969e-07, "loss": 0.4052, "step": 18253 }, { "epoch": 0.9233648641812939, "grad_norm": 11.415409088134766, "learning_rate": 1.4712873296421072e-07, "loss": 0.4449, "step": 18254 }, { "epoch": 0.9234154484293591, "grad_norm": 11.165006637573242, "learning_rate": 1.4693552434879298e-07, "loss": 0.4552, "step": 18255 }, { "epoch": 0.9234660326774242, "grad_norm": 7.43748664855957, "learning_rate": 1.4674244078512533e-07, "loss": 0.3372, "step": 18256 }, { "epoch": 0.9235166169254894, "grad_norm": 6.4748711585998535, "learning_rate": 1.4654948227818155e-07, "loss": 0.4225, "step": 18257 }, { "epoch": 0.9235672011735545, "grad_norm": 5.256566047668457, "learning_rate": 1.4635664883293545e-07, "loss": 0.2429, "step": 18258 }, { "epoch": 0.9236177854216197, "grad_norm": 14.53936767578125, "learning_rate": 1.4616394045435467e-07, "loss": 0.5661, "step": 18259 }, { "epoch": 0.9236683696696849, "grad_norm": 7.897899627685547, "learning_rate": 1.4597135714740473e-07, "loss": 0.4515, "step": 18260 }, { "epoch": 0.92371895391775, "grad_norm": 8.398945808410645, "learning_rate": 1.4577889891704887e-07, "loss": 0.3956, "step": 18261 }, { "epoch": 0.9237695381658152, "grad_norm": 7.328078269958496, "learning_rate": 1.4558656576824592e-07, "loss": 0.3188, "step": 18262 }, { "epoch": 0.9238201224138803, "grad_norm": 13.968993186950684, "learning_rate": 1.4539435770595133e-07, "loss": 0.3604, "step": 18263 }, { "epoch": 0.9238707066619455, "grad_norm": 7.075917720794678, "learning_rate": 1.4520227473511895e-07, "loss": 0.4131, "step": 18264 }, { "epoch": 0.9239212909100106, "grad_norm": 8.416753768920898, "learning_rate": 1.4501031686069756e-07, "loss": 0.3878, "step": 18265 }, { "epoch": 0.9239718751580758, "grad_norm": 11.584301948547363, "learning_rate": 1.448184840876332e-07, "loss": 0.5625, "step": 18266 }, { "epoch": 0.9240224594061409, "grad_norm": 6.524552345275879, "learning_rate": 1.4462677642086976e-07, "loss": 0.3851, "step": 18267 }, { "epoch": 0.9240730436542061, "grad_norm": 7.784486293792725, "learning_rate": 1.4443519386534654e-07, "loss": 0.4731, "step": 18268 }, { "epoch": 0.9241236279022712, "grad_norm": 4.762118816375732, "learning_rate": 1.4424373642599966e-07, "loss": 0.3155, "step": 18269 }, { "epoch": 0.9241742121503363, "grad_norm": 10.275644302368164, "learning_rate": 1.4405240410776399e-07, "loss": 0.4037, "step": 18270 }, { "epoch": 0.9242247963984015, "grad_norm": 7.474394798278809, "learning_rate": 1.4386119691556844e-07, "loss": 0.3857, "step": 18271 }, { "epoch": 0.9242753806464666, "grad_norm": 6.424368858337402, "learning_rate": 1.4367011485434012e-07, "loss": 0.3749, "step": 18272 }, { "epoch": 0.9243259648945319, "grad_norm": 8.166987419128418, "learning_rate": 1.4347915792900292e-07, "loss": 0.3725, "step": 18273 }, { "epoch": 0.924376549142597, "grad_norm": 3.944105863571167, "learning_rate": 1.4328832614447785e-07, "loss": 0.3644, "step": 18274 }, { "epoch": 0.9244271333906622, "grad_norm": 7.785332679748535, "learning_rate": 1.4309761950568102e-07, "loss": 0.3798, "step": 18275 }, { "epoch": 0.9244777176387273, "grad_norm": 6.8038716316223145, "learning_rate": 1.4290703801752792e-07, "loss": 0.3735, "step": 18276 }, { "epoch": 0.9245283018867925, "grad_norm": 7.686531066894531, "learning_rate": 1.4271658168492798e-07, "loss": 0.3912, "step": 18277 }, { "epoch": 0.9245788861348576, "grad_norm": 14.94681453704834, "learning_rate": 1.4252625051278945e-07, "loss": 0.467, "step": 18278 }, { "epoch": 0.9246294703829228, "grad_norm": 6.633467197418213, "learning_rate": 1.4233604450601734e-07, "loss": 0.2647, "step": 18279 }, { "epoch": 0.9246800546309879, "grad_norm": 9.440596580505371, "learning_rate": 1.421459636695116e-07, "loss": 0.4297, "step": 18280 }, { "epoch": 0.924730638879053, "grad_norm": 9.417665481567383, "learning_rate": 1.4195600800817057e-07, "loss": 0.4657, "step": 18281 }, { "epoch": 0.9247812231271182, "grad_norm": 11.657400131225586, "learning_rate": 1.417661775268897e-07, "loss": 0.3945, "step": 18282 }, { "epoch": 0.9248318073751833, "grad_norm": 24.676359176635742, "learning_rate": 1.415764722305596e-07, "loss": 0.6378, "step": 18283 }, { "epoch": 0.9248823916232485, "grad_norm": 10.696775436401367, "learning_rate": 1.4138689212406854e-07, "loss": 0.4482, "step": 18284 }, { "epoch": 0.9249329758713136, "grad_norm": 6.3318257331848145, "learning_rate": 1.4119743721230205e-07, "loss": 0.3745, "step": 18285 }, { "epoch": 0.9249835601193789, "grad_norm": 9.86526107788086, "learning_rate": 1.4100810750014128e-07, "loss": 0.4577, "step": 18286 }, { "epoch": 0.925034144367444, "grad_norm": 7.550030708312988, "learning_rate": 1.408189029924656e-07, "loss": 0.3786, "step": 18287 }, { "epoch": 0.9250847286155092, "grad_norm": 16.41625213623047, "learning_rate": 1.4062982369415001e-07, "loss": 0.7224, "step": 18288 }, { "epoch": 0.9251353128635743, "grad_norm": 9.407434463500977, "learning_rate": 1.4044086961006564e-07, "loss": 0.429, "step": 18289 }, { "epoch": 0.9251858971116395, "grad_norm": 7.316087245941162, "learning_rate": 1.4025204074508302e-07, "loss": 0.4097, "step": 18290 }, { "epoch": 0.9252364813597046, "grad_norm": 5.76793909072876, "learning_rate": 1.4006333710406716e-07, "loss": 0.3163, "step": 18291 }, { "epoch": 0.9252870656077697, "grad_norm": 14.432303428649902, "learning_rate": 1.3987475869187972e-07, "loss": 0.3568, "step": 18292 }, { "epoch": 0.9253376498558349, "grad_norm": 16.63138771057129, "learning_rate": 1.3968630551338124e-07, "loss": 0.5656, "step": 18293 }, { "epoch": 0.9253882341039, "grad_norm": 5.745511054992676, "learning_rate": 1.3949797757342732e-07, "loss": 0.3398, "step": 18294 }, { "epoch": 0.9254388183519652, "grad_norm": 9.176982879638672, "learning_rate": 1.3930977487687013e-07, "loss": 0.4954, "step": 18295 }, { "epoch": 0.9254894026000303, "grad_norm": 5.071663856506348, "learning_rate": 1.3912169742855918e-07, "loss": 0.4016, "step": 18296 }, { "epoch": 0.9255399868480955, "grad_norm": 7.229979038238525, "learning_rate": 1.389337452333417e-07, "loss": 0.2812, "step": 18297 }, { "epoch": 0.9255905710961606, "grad_norm": 18.703855514526367, "learning_rate": 1.3874591829605933e-07, "loss": 0.4327, "step": 18298 }, { "epoch": 0.9256411553442259, "grad_norm": 7.462826251983643, "learning_rate": 1.385582166215538e-07, "loss": 0.3014, "step": 18299 }, { "epoch": 0.925691739592291, "grad_norm": 6.040946006774902, "learning_rate": 1.3837064021466007e-07, "loss": 0.4029, "step": 18300 }, { "epoch": 0.9257423238403562, "grad_norm": 9.773420333862305, "learning_rate": 1.381831890802121e-07, "loss": 0.2731, "step": 18301 }, { "epoch": 0.9257929080884213, "grad_norm": 7.409374713897705, "learning_rate": 1.3799586322304047e-07, "loss": 0.4791, "step": 18302 }, { "epoch": 0.9258434923364864, "grad_norm": 9.807045936584473, "learning_rate": 1.3780866264797132e-07, "loss": 0.4251, "step": 18303 }, { "epoch": 0.9258940765845516, "grad_norm": 5.986662864685059, "learning_rate": 1.3762158735982857e-07, "loss": 0.3434, "step": 18304 }, { "epoch": 0.9259446608326167, "grad_norm": 6.439162254333496, "learning_rate": 1.3743463736343388e-07, "loss": 0.3542, "step": 18305 }, { "epoch": 0.9259952450806819, "grad_norm": 6.182503700256348, "learning_rate": 1.3724781266360287e-07, "loss": 0.4603, "step": 18306 }, { "epoch": 0.926045829328747, "grad_norm": 10.6813383102417, "learning_rate": 1.3706111326515003e-07, "loss": 0.4495, "step": 18307 }, { "epoch": 0.9260964135768122, "grad_norm": 7.047734260559082, "learning_rate": 1.3687453917288706e-07, "loss": 0.4017, "step": 18308 }, { "epoch": 0.9261469978248773, "grad_norm": 4.185029029846191, "learning_rate": 1.3668809039162123e-07, "loss": 0.3858, "step": 18309 }, { "epoch": 0.9261975820729424, "grad_norm": 7.576103687286377, "learning_rate": 1.3650176692615537e-07, "loss": 0.4225, "step": 18310 }, { "epoch": 0.9262481663210076, "grad_norm": 4.4759368896484375, "learning_rate": 1.363155687812917e-07, "loss": 0.3209, "step": 18311 }, { "epoch": 0.9262987505690727, "grad_norm": 8.707371711730957, "learning_rate": 1.3612949596182869e-07, "loss": 0.281, "step": 18312 }, { "epoch": 0.926349334817138, "grad_norm": 8.695093154907227, "learning_rate": 1.3594354847255965e-07, "loss": 0.2915, "step": 18313 }, { "epoch": 0.9263999190652031, "grad_norm": 9.152972221374512, "learning_rate": 1.3575772631827745e-07, "loss": 0.3262, "step": 18314 }, { "epoch": 0.9264505033132683, "grad_norm": 6.716559410095215, "learning_rate": 1.3557202950376934e-07, "loss": 0.3615, "step": 18315 }, { "epoch": 0.9265010875613334, "grad_norm": 6.233284950256348, "learning_rate": 1.353864580338199e-07, "loss": 0.3877, "step": 18316 }, { "epoch": 0.9265516718093986, "grad_norm": 11.142141342163086, "learning_rate": 1.3520101191321243e-07, "loss": 0.3676, "step": 18317 }, { "epoch": 0.9266022560574637, "grad_norm": 10.130121231079102, "learning_rate": 1.3501569114672375e-07, "loss": 0.5581, "step": 18318 }, { "epoch": 0.9266528403055289, "grad_norm": 9.33895492553711, "learning_rate": 1.3483049573913e-07, "loss": 0.4513, "step": 18319 }, { "epoch": 0.926703424553594, "grad_norm": 12.97418212890625, "learning_rate": 1.3464542569520288e-07, "loss": 0.462, "step": 18320 }, { "epoch": 0.9267540088016591, "grad_norm": 8.578197479248047, "learning_rate": 1.3446048101971143e-07, "loss": 0.4045, "step": 18321 }, { "epoch": 0.9268045930497243, "grad_norm": 6.836690902709961, "learning_rate": 1.3427566171742178e-07, "loss": 0.4231, "step": 18322 }, { "epoch": 0.9268551772977894, "grad_norm": 7.382940769195557, "learning_rate": 1.3409096779309515e-07, "loss": 0.4053, "step": 18323 }, { "epoch": 0.9269057615458546, "grad_norm": 7.851980209350586, "learning_rate": 1.3390639925149051e-07, "loss": 0.3772, "step": 18324 }, { "epoch": 0.9269563457939197, "grad_norm": 7.100743293762207, "learning_rate": 1.3372195609736515e-07, "loss": 0.4182, "step": 18325 }, { "epoch": 0.927006930041985, "grad_norm": 6.581226825714111, "learning_rate": 1.335376383354714e-07, "loss": 0.3723, "step": 18326 }, { "epoch": 0.9270575142900501, "grad_norm": 6.7881693840026855, "learning_rate": 1.333534459705571e-07, "loss": 0.4065, "step": 18327 }, { "epoch": 0.9271080985381153, "grad_norm": 9.128324508666992, "learning_rate": 1.3316937900737016e-07, "loss": 0.3553, "step": 18328 }, { "epoch": 0.9271586827861804, "grad_norm": 19.45547866821289, "learning_rate": 1.329854374506534e-07, "loss": 0.4397, "step": 18329 }, { "epoch": 0.9272092670342456, "grad_norm": 9.157153129577637, "learning_rate": 1.328016213051453e-07, "loss": 0.4663, "step": 18330 }, { "epoch": 0.9272598512823107, "grad_norm": 7.0897216796875, "learning_rate": 1.326179305755837e-07, "loss": 0.4429, "step": 18331 }, { "epoch": 0.9273104355303758, "grad_norm": 6.713512897491455, "learning_rate": 1.324343652667015e-07, "loss": 0.432, "step": 18332 }, { "epoch": 0.927361019778441, "grad_norm": 9.888492584228516, "learning_rate": 1.3225092538322825e-07, "loss": 0.4318, "step": 18333 }, { "epoch": 0.9274116040265061, "grad_norm": 9.01003360748291, "learning_rate": 1.3206761092989075e-07, "loss": 0.3697, "step": 18334 }, { "epoch": 0.9274621882745713, "grad_norm": 8.783235549926758, "learning_rate": 1.3188442191141405e-07, "loss": 0.4359, "step": 18335 }, { "epoch": 0.9275127725226364, "grad_norm": 8.792187690734863, "learning_rate": 1.3170135833251607e-07, "loss": 0.4345, "step": 18336 }, { "epoch": 0.9275633567707016, "grad_norm": 6.807309150695801, "learning_rate": 1.3151842019791638e-07, "loss": 0.4712, "step": 18337 }, { "epoch": 0.9276139410187667, "grad_norm": 6.389904022216797, "learning_rate": 1.3133560751232678e-07, "loss": 0.4182, "step": 18338 }, { "epoch": 0.927664525266832, "grad_norm": 5.536910533905029, "learning_rate": 1.31152920280459e-07, "loss": 0.4491, "step": 18339 }, { "epoch": 0.9277151095148971, "grad_norm": 8.17470645904541, "learning_rate": 1.3097035850702044e-07, "loss": 0.4538, "step": 18340 }, { "epoch": 0.9277656937629623, "grad_norm": 7.058501720428467, "learning_rate": 1.307879221967151e-07, "loss": 0.3027, "step": 18341 }, { "epoch": 0.9278162780110274, "grad_norm": 4.636490821838379, "learning_rate": 1.306056113542431e-07, "loss": 0.3551, "step": 18342 }, { "epoch": 0.9278668622590925, "grad_norm": 6.391381740570068, "learning_rate": 1.3042342598430402e-07, "loss": 0.3768, "step": 18343 }, { "epoch": 0.9279174465071577, "grad_norm": 5.088301181793213, "learning_rate": 1.3024136609159077e-07, "loss": 0.3825, "step": 18344 }, { "epoch": 0.9279680307552228, "grad_norm": 5.345361709594727, "learning_rate": 1.300594316807946e-07, "loss": 0.3451, "step": 18345 }, { "epoch": 0.928018615003288, "grad_norm": 6.996176719665527, "learning_rate": 1.2987762275660508e-07, "loss": 0.4305, "step": 18346 }, { "epoch": 0.9280691992513531, "grad_norm": 5.783626079559326, "learning_rate": 1.2969593932370517e-07, "loss": 0.3637, "step": 18347 }, { "epoch": 0.9281197834994183, "grad_norm": 8.388498306274414, "learning_rate": 1.2951438138677775e-07, "loss": 0.3775, "step": 18348 }, { "epoch": 0.9281703677474834, "grad_norm": 9.697739601135254, "learning_rate": 1.293329489505002e-07, "loss": 0.4415, "step": 18349 }, { "epoch": 0.9282209519955485, "grad_norm": 5.701276779174805, "learning_rate": 1.291516420195482e-07, "loss": 0.3385, "step": 18350 }, { "epoch": 0.9282715362436137, "grad_norm": 11.111974716186523, "learning_rate": 1.2897046059859307e-07, "loss": 0.3985, "step": 18351 }, { "epoch": 0.9283221204916788, "grad_norm": 11.26558780670166, "learning_rate": 1.2878940469230384e-07, "loss": 0.3829, "step": 18352 }, { "epoch": 0.9283727047397441, "grad_norm": 4.840790271759033, "learning_rate": 1.2860847430534562e-07, "loss": 0.2801, "step": 18353 }, { "epoch": 0.9284232889878092, "grad_norm": 13.142230033874512, "learning_rate": 1.2842766944238028e-07, "loss": 0.3722, "step": 18354 }, { "epoch": 0.9284738732358744, "grad_norm": 7.632421493530273, "learning_rate": 1.2824699010806742e-07, "loss": 0.3274, "step": 18355 }, { "epoch": 0.9285244574839395, "grad_norm": 11.031502723693848, "learning_rate": 1.2806643630706273e-07, "loss": 0.4063, "step": 18356 }, { "epoch": 0.9285750417320047, "grad_norm": 10.216753959655762, "learning_rate": 1.278860080440175e-07, "loss": 0.2585, "step": 18357 }, { "epoch": 0.9286256259800698, "grad_norm": 7.961162090301514, "learning_rate": 1.277057053235825e-07, "loss": 0.392, "step": 18358 }, { "epoch": 0.928676210228135, "grad_norm": 9.510900497436523, "learning_rate": 1.275255281504023e-07, "loss": 0.3664, "step": 18359 }, { "epoch": 0.9287267944762001, "grad_norm": 9.509469985961914, "learning_rate": 1.2734547652912042e-07, "loss": 0.2986, "step": 18360 }, { "epoch": 0.9287773787242652, "grad_norm": 8.360669136047363, "learning_rate": 1.2716555046437708e-07, "loss": 0.4141, "step": 18361 }, { "epoch": 0.9288279629723304, "grad_norm": 9.960993766784668, "learning_rate": 1.2698574996080682e-07, "loss": 0.3576, "step": 18362 }, { "epoch": 0.9288785472203955, "grad_norm": 9.344809532165527, "learning_rate": 1.2680607502304375e-07, "loss": 0.4708, "step": 18363 }, { "epoch": 0.9289291314684607, "grad_norm": 6.621085166931152, "learning_rate": 1.26626525655717e-07, "loss": 0.3716, "step": 18364 }, { "epoch": 0.9289797157165258, "grad_norm": 6.087156295776367, "learning_rate": 1.264471018634539e-07, "loss": 0.3529, "step": 18365 }, { "epoch": 0.9290302999645911, "grad_norm": 6.685498237609863, "learning_rate": 1.2626780365087743e-07, "loss": 0.3855, "step": 18366 }, { "epoch": 0.9290808842126562, "grad_norm": 8.194879531860352, "learning_rate": 1.2608863102260728e-07, "loss": 0.3434, "step": 18367 }, { "epoch": 0.9291314684607214, "grad_norm": 5.826079845428467, "learning_rate": 1.2590958398326135e-07, "loss": 0.2574, "step": 18368 }, { "epoch": 0.9291820527087865, "grad_norm": 8.850590705871582, "learning_rate": 1.2573066253745158e-07, "loss": 0.4296, "step": 18369 }, { "epoch": 0.9292326369568517, "grad_norm": 11.9863862991333, "learning_rate": 1.2555186668979036e-07, "loss": 0.3463, "step": 18370 }, { "epoch": 0.9292832212049168, "grad_norm": 14.657661437988281, "learning_rate": 1.2537319644488343e-07, "loss": 0.4785, "step": 18371 }, { "epoch": 0.9293338054529819, "grad_norm": 6.498118877410889, "learning_rate": 1.2519465180733437e-07, "loss": 0.3552, "step": 18372 }, { "epoch": 0.9293843897010471, "grad_norm": 6.225161075592041, "learning_rate": 1.2501623278174556e-07, "loss": 0.3318, "step": 18373 }, { "epoch": 0.9294349739491122, "grad_norm": 6.030930995941162, "learning_rate": 1.2483793937271283e-07, "loss": 0.4176, "step": 18374 }, { "epoch": 0.9294855581971774, "grad_norm": 5.2142014503479, "learning_rate": 1.2465977158483077e-07, "loss": 0.3064, "step": 18375 }, { "epoch": 0.9295361424452425, "grad_norm": 6.512903690338135, "learning_rate": 1.2448172942269076e-07, "loss": 0.3934, "step": 18376 }, { "epoch": 0.9295867266933077, "grad_norm": 11.305177688598633, "learning_rate": 1.2430381289088022e-07, "loss": 0.3862, "step": 18377 }, { "epoch": 0.9296373109413728, "grad_norm": 8.659537315368652, "learning_rate": 1.2412602199398383e-07, "loss": 0.4296, "step": 18378 }, { "epoch": 0.9296878951894381, "grad_norm": 7.946200847625732, "learning_rate": 1.2394835673658235e-07, "loss": 0.4155, "step": 18379 }, { "epoch": 0.9297384794375032, "grad_norm": 7.4673004150390625, "learning_rate": 1.2377081712325378e-07, "loss": 0.4862, "step": 18380 }, { "epoch": 0.9297890636855684, "grad_norm": 10.051627159118652, "learning_rate": 1.2359340315857393e-07, "loss": 0.4425, "step": 18381 }, { "epoch": 0.9298396479336335, "grad_norm": 8.038512229919434, "learning_rate": 1.234161148471136e-07, "loss": 0.3151, "step": 18382 }, { "epoch": 0.9298902321816986, "grad_norm": 5.735762596130371, "learning_rate": 1.2323895219344016e-07, "loss": 0.3596, "step": 18383 }, { "epoch": 0.9299408164297638, "grad_norm": 7.736927509307861, "learning_rate": 1.2306191520212062e-07, "loss": 0.3697, "step": 18384 }, { "epoch": 0.9299914006778289, "grad_norm": 13.619749069213867, "learning_rate": 1.2288500387771517e-07, "loss": 0.4526, "step": 18385 }, { "epoch": 0.9300419849258941, "grad_norm": 6.201078414916992, "learning_rate": 1.2270821822478353e-07, "loss": 0.3215, "step": 18386 }, { "epoch": 0.9300925691739592, "grad_norm": 8.258742332458496, "learning_rate": 1.2253155824788031e-07, "loss": 0.3793, "step": 18387 }, { "epoch": 0.9301431534220244, "grad_norm": 8.691869735717773, "learning_rate": 1.2235502395155807e-07, "loss": 0.3704, "step": 18388 }, { "epoch": 0.9301937376700895, "grad_norm": 5.83672571182251, "learning_rate": 1.2217861534036423e-07, "loss": 0.3994, "step": 18389 }, { "epoch": 0.9302443219181546, "grad_norm": 9.837441444396973, "learning_rate": 1.2200233241884686e-07, "loss": 0.3819, "step": 18390 }, { "epoch": 0.9302949061662198, "grad_norm": 7.383142948150635, "learning_rate": 1.2182617519154615e-07, "loss": 0.3825, "step": 18391 }, { "epoch": 0.9303454904142849, "grad_norm": 16.716041564941406, "learning_rate": 1.2165014366300244e-07, "loss": 0.5445, "step": 18392 }, { "epoch": 0.9303960746623502, "grad_norm": 7.32015323638916, "learning_rate": 1.2147423783775148e-07, "loss": 0.3391, "step": 18393 }, { "epoch": 0.9304466589104153, "grad_norm": 7.673548698425293, "learning_rate": 1.212984577203258e-07, "loss": 0.373, "step": 18394 }, { "epoch": 0.9304972431584805, "grad_norm": 7.873368263244629, "learning_rate": 1.211228033152545e-07, "loss": 0.4731, "step": 18395 }, { "epoch": 0.9305478274065456, "grad_norm": 7.20873498916626, "learning_rate": 1.2094727462706457e-07, "loss": 0.3996, "step": 18396 }, { "epoch": 0.9305984116546108, "grad_norm": 14.364683151245117, "learning_rate": 1.207718716602785e-07, "loss": 0.4809, "step": 18397 }, { "epoch": 0.9306489959026759, "grad_norm": 6.415250301361084, "learning_rate": 1.205965944194154e-07, "loss": 0.3854, "step": 18398 }, { "epoch": 0.9306995801507411, "grad_norm": 7.474261283874512, "learning_rate": 1.2042144290899339e-07, "loss": 0.3648, "step": 18399 }, { "epoch": 0.9307501643988062, "grad_norm": 7.130216598510742, "learning_rate": 1.2024641713352382e-07, "loss": 0.3794, "step": 18400 }, { "epoch": 0.9308007486468713, "grad_norm": 7.8043646812438965, "learning_rate": 1.2007151709751807e-07, "loss": 0.4344, "step": 18401 }, { "epoch": 0.9308513328949365, "grad_norm": 6.507930755615234, "learning_rate": 1.1989674280548203e-07, "loss": 0.3894, "step": 18402 }, { "epoch": 0.9309019171430016, "grad_norm": 7.367580413818359, "learning_rate": 1.1972209426191928e-07, "loss": 0.3794, "step": 18403 }, { "epoch": 0.9309525013910668, "grad_norm": 8.38666820526123, "learning_rate": 1.1954757147133067e-07, "loss": 0.3568, "step": 18404 }, { "epoch": 0.9310030856391319, "grad_norm": 12.988810539245605, "learning_rate": 1.1937317443821317e-07, "loss": 0.4164, "step": 18405 }, { "epoch": 0.9310536698871972, "grad_norm": 5.8368096351623535, "learning_rate": 1.1919890316705984e-07, "loss": 0.3643, "step": 18406 }, { "epoch": 0.9311042541352623, "grad_norm": 7.271370887756348, "learning_rate": 1.1902475766236154e-07, "loss": 0.4181, "step": 18407 }, { "epoch": 0.9311548383833275, "grad_norm": 6.216193675994873, "learning_rate": 1.188507379286058e-07, "loss": 0.3769, "step": 18408 }, { "epoch": 0.9312054226313926, "grad_norm": 7.773625373840332, "learning_rate": 1.1867684397027735e-07, "loss": 0.299, "step": 18409 }, { "epoch": 0.9312560068794578, "grad_norm": 11.708968162536621, "learning_rate": 1.1850307579185539e-07, "loss": 0.4064, "step": 18410 }, { "epoch": 0.9313065911275229, "grad_norm": 6.46118688583374, "learning_rate": 1.1832943339781855e-07, "loss": 0.3305, "step": 18411 }, { "epoch": 0.931357175375588, "grad_norm": 8.201140403747559, "learning_rate": 1.181559167926416e-07, "loss": 0.4437, "step": 18412 }, { "epoch": 0.9314077596236532, "grad_norm": 8.116024017333984, "learning_rate": 1.1798252598079429e-07, "loss": 0.335, "step": 18413 }, { "epoch": 0.9314583438717183, "grad_norm": 7.243833541870117, "learning_rate": 1.1780926096674583e-07, "loss": 0.4435, "step": 18414 }, { "epoch": 0.9315089281197835, "grad_norm": 7.2003045082092285, "learning_rate": 1.1763612175495987e-07, "loss": 0.3901, "step": 18415 }, { "epoch": 0.9315595123678486, "grad_norm": 6.42866325378418, "learning_rate": 1.1746310834989838e-07, "loss": 0.3343, "step": 18416 }, { "epoch": 0.9316100966159138, "grad_norm": 10.329623222351074, "learning_rate": 1.1729022075601947e-07, "loss": 0.4029, "step": 18417 }, { "epoch": 0.9316606808639789, "grad_norm": 6.140257358551025, "learning_rate": 1.1711745897777738e-07, "loss": 0.3722, "step": 18418 }, { "epoch": 0.9317112651120442, "grad_norm": 8.46384334564209, "learning_rate": 1.1694482301962461e-07, "loss": 0.3544, "step": 18419 }, { "epoch": 0.9317618493601093, "grad_norm": 6.4459123611450195, "learning_rate": 1.1677231288600932e-07, "loss": 0.4205, "step": 18420 }, { "epoch": 0.9318124336081745, "grad_norm": 6.336606979370117, "learning_rate": 1.1659992858137626e-07, "loss": 0.4065, "step": 18421 }, { "epoch": 0.9318630178562396, "grad_norm": 8.279826164245605, "learning_rate": 1.1642767011016797e-07, "loss": 0.5178, "step": 18422 }, { "epoch": 0.9319136021043047, "grad_norm": 8.989335060119629, "learning_rate": 1.162555374768226e-07, "loss": 0.3515, "step": 18423 }, { "epoch": 0.9319641863523699, "grad_norm": 8.875426292419434, "learning_rate": 1.1608353068577549e-07, "loss": 0.3556, "step": 18424 }, { "epoch": 0.932014770600435, "grad_norm": 7.869764804840088, "learning_rate": 1.1591164974146029e-07, "loss": 0.3639, "step": 18425 }, { "epoch": 0.9320653548485002, "grad_norm": 6.08624792098999, "learning_rate": 1.1573989464830404e-07, "loss": 0.3366, "step": 18426 }, { "epoch": 0.9321159390965653, "grad_norm": 10.029022216796875, "learning_rate": 1.1556826541073319e-07, "loss": 0.4385, "step": 18427 }, { "epoch": 0.9321665233446305, "grad_norm": 8.73609733581543, "learning_rate": 1.1539676203317085e-07, "loss": 0.2979, "step": 18428 }, { "epoch": 0.9322171075926956, "grad_norm": 10.069304466247559, "learning_rate": 1.1522538452003518e-07, "loss": 0.5183, "step": 18429 }, { "epoch": 0.9322676918407607, "grad_norm": 7.891778469085693, "learning_rate": 1.1505413287574263e-07, "loss": 0.3456, "step": 18430 }, { "epoch": 0.9323182760888259, "grad_norm": 11.69028091430664, "learning_rate": 1.1488300710470579e-07, "loss": 0.4099, "step": 18431 }, { "epoch": 0.9323688603368911, "grad_norm": 8.453563690185547, "learning_rate": 1.1471200721133503e-07, "loss": 0.3432, "step": 18432 }, { "epoch": 0.9324194445849563, "grad_norm": 5.620534420013428, "learning_rate": 1.1454113320003513e-07, "loss": 0.3809, "step": 18433 }, { "epoch": 0.9324700288330214, "grad_norm": 6.254651069641113, "learning_rate": 1.1437038507521036e-07, "loss": 0.3674, "step": 18434 }, { "epoch": 0.9325206130810866, "grad_norm": 10.147610664367676, "learning_rate": 1.1419976284126e-07, "loss": 0.3972, "step": 18435 }, { "epoch": 0.9325711973291517, "grad_norm": 12.134927749633789, "learning_rate": 1.1402926650258051e-07, "loss": 0.3971, "step": 18436 }, { "epoch": 0.9326217815772169, "grad_norm": 10.353034973144531, "learning_rate": 1.1385889606356504e-07, "loss": 0.4664, "step": 18437 }, { "epoch": 0.932672365825282, "grad_norm": 10.2277250289917, "learning_rate": 1.1368865152860453e-07, "loss": 0.4005, "step": 18438 }, { "epoch": 0.9327229500733472, "grad_norm": 7.7354044914245605, "learning_rate": 1.1351853290208492e-07, "loss": 0.3553, "step": 18439 }, { "epoch": 0.9327735343214123, "grad_norm": 5.3233747482299805, "learning_rate": 1.1334854018838937e-07, "loss": 0.3758, "step": 18440 }, { "epoch": 0.9328241185694774, "grad_norm": 8.750788688659668, "learning_rate": 1.1317867339189936e-07, "loss": 0.4864, "step": 18441 }, { "epoch": 0.9328747028175426, "grad_norm": 10.458943367004395, "learning_rate": 1.130089325169903e-07, "loss": 0.3608, "step": 18442 }, { "epoch": 0.9329252870656077, "grad_norm": 7.3175249099731445, "learning_rate": 1.1283931756803756e-07, "loss": 0.4783, "step": 18443 }, { "epoch": 0.9329758713136729, "grad_norm": 8.95181941986084, "learning_rate": 1.1266982854941155e-07, "loss": 0.4519, "step": 18444 }, { "epoch": 0.933026455561738, "grad_norm": 8.675617218017578, "learning_rate": 1.1250046546547821e-07, "loss": 0.543, "step": 18445 }, { "epoch": 0.9330770398098033, "grad_norm": 7.234497547149658, "learning_rate": 1.123312283206035e-07, "loss": 0.3622, "step": 18446 }, { "epoch": 0.9331276240578684, "grad_norm": 5.990774154663086, "learning_rate": 1.1216211711914726e-07, "loss": 0.3901, "step": 18447 }, { "epoch": 0.9331782083059336, "grad_norm": 9.191970825195312, "learning_rate": 1.1199313186546657e-07, "loss": 0.3817, "step": 18448 }, { "epoch": 0.9332287925539987, "grad_norm": 9.399251937866211, "learning_rate": 1.1182427256391682e-07, "loss": 0.4452, "step": 18449 }, { "epoch": 0.9332793768020639, "grad_norm": 9.306167602539062, "learning_rate": 1.1165553921884897e-07, "loss": 0.3715, "step": 18450 }, { "epoch": 0.933329961050129, "grad_norm": 7.3046369552612305, "learning_rate": 1.1148693183461012e-07, "loss": 0.3098, "step": 18451 }, { "epoch": 0.9333805452981941, "grad_norm": 7.412264823913574, "learning_rate": 1.1131845041554568e-07, "loss": 0.4147, "step": 18452 }, { "epoch": 0.9334311295462593, "grad_norm": 11.048317909240723, "learning_rate": 1.1115009496599605e-07, "loss": 0.3583, "step": 18453 }, { "epoch": 0.9334817137943244, "grad_norm": 18.33297348022461, "learning_rate": 1.1098186549030054e-07, "loss": 0.4913, "step": 18454 }, { "epoch": 0.9335322980423896, "grad_norm": 6.7162065505981445, "learning_rate": 1.1081376199279293e-07, "loss": 0.3606, "step": 18455 }, { "epoch": 0.9335828822904547, "grad_norm": 12.159140586853027, "learning_rate": 1.1064578447780527e-07, "loss": 0.5236, "step": 18456 }, { "epoch": 0.9336334665385199, "grad_norm": 7.838155746459961, "learning_rate": 1.1047793294966635e-07, "loss": 0.3884, "step": 18457 }, { "epoch": 0.933684050786585, "grad_norm": 8.426359176635742, "learning_rate": 1.1031020741270104e-07, "loss": 0.3391, "step": 18458 }, { "epoch": 0.9337346350346503, "grad_norm": 17.297754287719727, "learning_rate": 1.1014260787123088e-07, "loss": 0.396, "step": 18459 }, { "epoch": 0.9337852192827154, "grad_norm": 5.460080623626709, "learning_rate": 1.0997513432957463e-07, "loss": 0.463, "step": 18460 }, { "epoch": 0.9338358035307806, "grad_norm": 7.8267598152160645, "learning_rate": 1.0980778679204828e-07, "loss": 0.4451, "step": 18461 }, { "epoch": 0.9338863877788457, "grad_norm": 6.290886878967285, "learning_rate": 1.0964056526296285e-07, "loss": 0.3807, "step": 18462 }, { "epoch": 0.9339369720269108, "grad_norm": 6.526186466217041, "learning_rate": 1.0947346974662876e-07, "loss": 0.3712, "step": 18463 }, { "epoch": 0.933987556274976, "grad_norm": 8.137518882751465, "learning_rate": 1.0930650024735035e-07, "loss": 0.3354, "step": 18464 }, { "epoch": 0.9340381405230411, "grad_norm": 12.512811660766602, "learning_rate": 1.0913965676942973e-07, "loss": 0.4644, "step": 18465 }, { "epoch": 0.9340887247711063, "grad_norm": 8.96028995513916, "learning_rate": 1.0897293931716735e-07, "loss": 0.4284, "step": 18466 }, { "epoch": 0.9341393090191714, "grad_norm": 15.140830993652344, "learning_rate": 1.0880634789485867e-07, "loss": 0.4592, "step": 18467 }, { "epoch": 0.9341898932672366, "grad_norm": 20.623483657836914, "learning_rate": 1.0863988250679581e-07, "loss": 0.35, "step": 18468 }, { "epoch": 0.9342404775153017, "grad_norm": 12.22756576538086, "learning_rate": 1.0847354315726921e-07, "loss": 0.3934, "step": 18469 }, { "epoch": 0.9342910617633668, "grad_norm": 8.026515007019043, "learning_rate": 1.083073298505638e-07, "loss": 0.3763, "step": 18470 }, { "epoch": 0.934341646011432, "grad_norm": 6.118260860443115, "learning_rate": 1.0814124259096337e-07, "loss": 0.4492, "step": 18471 }, { "epoch": 0.9343922302594972, "grad_norm": 6.953976154327393, "learning_rate": 1.0797528138274727e-07, "loss": 0.2608, "step": 18472 }, { "epoch": 0.9344428145075624, "grad_norm": 7.309528350830078, "learning_rate": 1.0780944623019207e-07, "loss": 0.4097, "step": 18473 }, { "epoch": 0.9344933987556275, "grad_norm": 7.88832426071167, "learning_rate": 1.0764373713757048e-07, "loss": 0.4628, "step": 18474 }, { "epoch": 0.9345439830036927, "grad_norm": 7.205978870391846, "learning_rate": 1.0747815410915296e-07, "loss": 0.456, "step": 18475 }, { "epoch": 0.9345945672517578, "grad_norm": 9.2481689453125, "learning_rate": 1.0731269714920667e-07, "loss": 0.4116, "step": 18476 }, { "epoch": 0.934645151499823, "grad_norm": 11.704205513000488, "learning_rate": 1.0714736626199374e-07, "loss": 0.4601, "step": 18477 }, { "epoch": 0.9346957357478881, "grad_norm": 5.581136226654053, "learning_rate": 1.069821614517752e-07, "loss": 0.3683, "step": 18478 }, { "epoch": 0.9347463199959533, "grad_norm": 8.916178703308105, "learning_rate": 1.0681708272280767e-07, "loss": 0.3926, "step": 18479 }, { "epoch": 0.9347969042440184, "grad_norm": 11.66818618774414, "learning_rate": 1.0665213007934438e-07, "loss": 0.3636, "step": 18480 }, { "epoch": 0.9348474884920835, "grad_norm": 9.865792274475098, "learning_rate": 1.0648730352563697e-07, "loss": 0.4204, "step": 18481 }, { "epoch": 0.9348980727401487, "grad_norm": 5.162074565887451, "learning_rate": 1.0632260306593201e-07, "loss": 0.4133, "step": 18482 }, { "epoch": 0.9349486569882138, "grad_norm": 9.01852035522461, "learning_rate": 1.0615802870447279e-07, "loss": 0.3243, "step": 18483 }, { "epoch": 0.934999241236279, "grad_norm": 10.576485633850098, "learning_rate": 1.059935804455009e-07, "loss": 0.4537, "step": 18484 }, { "epoch": 0.9350498254843441, "grad_norm": 9.000179290771484, "learning_rate": 1.0582925829325352e-07, "loss": 0.4579, "step": 18485 }, { "epoch": 0.9351004097324094, "grad_norm": 10.441417694091797, "learning_rate": 1.056650622519645e-07, "loss": 0.3571, "step": 18486 }, { "epoch": 0.9351509939804745, "grad_norm": 5.028536319732666, "learning_rate": 1.0550099232586542e-07, "loss": 0.4264, "step": 18487 }, { "epoch": 0.9352015782285397, "grad_norm": 6.625286102294922, "learning_rate": 1.0533704851918347e-07, "loss": 0.3989, "step": 18488 }, { "epoch": 0.9352521624766048, "grad_norm": 8.643978118896484, "learning_rate": 1.0517323083614305e-07, "loss": 0.3858, "step": 18489 }, { "epoch": 0.93530274672467, "grad_norm": 14.37142562866211, "learning_rate": 1.050095392809658e-07, "loss": 0.6306, "step": 18490 }, { "epoch": 0.9353533309727351, "grad_norm": 7.205775737762451, "learning_rate": 1.0484597385786887e-07, "loss": 0.2868, "step": 18491 }, { "epoch": 0.9354039152208002, "grad_norm": 14.477373123168945, "learning_rate": 1.046825345710678e-07, "loss": 0.5262, "step": 18492 }, { "epoch": 0.9354544994688654, "grad_norm": 10.708293914794922, "learning_rate": 1.0451922142477366e-07, "loss": 0.4762, "step": 18493 }, { "epoch": 0.9355050837169305, "grad_norm": 9.205348014831543, "learning_rate": 1.0435603442319419e-07, "loss": 0.3261, "step": 18494 }, { "epoch": 0.9355556679649957, "grad_norm": 11.94617748260498, "learning_rate": 1.0419297357053549e-07, "loss": 0.4215, "step": 18495 }, { "epoch": 0.9356062522130608, "grad_norm": 6.730029106140137, "learning_rate": 1.0403003887099806e-07, "loss": 0.4053, "step": 18496 }, { "epoch": 0.935656836461126, "grad_norm": 10.423761367797852, "learning_rate": 1.0386723032878076e-07, "loss": 0.5136, "step": 18497 }, { "epoch": 0.9357074207091911, "grad_norm": 8.153413772583008, "learning_rate": 1.0370454794807916e-07, "loss": 0.4386, "step": 18498 }, { "epoch": 0.9357580049572564, "grad_norm": 7.716214656829834, "learning_rate": 1.0354199173308488e-07, "loss": 0.3693, "step": 18499 }, { "epoch": 0.9358085892053215, "grad_norm": 8.838174819946289, "learning_rate": 1.0337956168798569e-07, "loss": 0.4227, "step": 18500 }, { "epoch": 0.9358591734533867, "grad_norm": 8.195528030395508, "learning_rate": 1.0321725781696878e-07, "loss": 0.258, "step": 18501 }, { "epoch": 0.9359097577014518, "grad_norm": 5.486400127410889, "learning_rate": 1.0305508012421584e-07, "loss": 0.4052, "step": 18502 }, { "epoch": 0.9359603419495169, "grad_norm": 6.642298698425293, "learning_rate": 1.028930286139046e-07, "loss": 0.3767, "step": 18503 }, { "epoch": 0.9360109261975821, "grad_norm": 6.223590850830078, "learning_rate": 1.0273110329021174e-07, "loss": 0.3045, "step": 18504 }, { "epoch": 0.9360615104456472, "grad_norm": 8.126324653625488, "learning_rate": 1.0256930415730948e-07, "loss": 0.3753, "step": 18505 }, { "epoch": 0.9361120946937124, "grad_norm": 6.6952667236328125, "learning_rate": 1.024076312193667e-07, "loss": 0.3499, "step": 18506 }, { "epoch": 0.9361626789417775, "grad_norm": 8.933806419372559, "learning_rate": 1.022460844805495e-07, "loss": 0.415, "step": 18507 }, { "epoch": 0.9362132631898427, "grad_norm": 9.215776443481445, "learning_rate": 1.0208466394502125e-07, "loss": 0.4278, "step": 18508 }, { "epoch": 0.9362638474379078, "grad_norm": 9.198783874511719, "learning_rate": 1.0192336961694027e-07, "loss": 0.4329, "step": 18509 }, { "epoch": 0.936314431685973, "grad_norm": 7.680948257446289, "learning_rate": 1.0176220150046379e-07, "loss": 0.4172, "step": 18510 }, { "epoch": 0.9363650159340381, "grad_norm": 13.412816047668457, "learning_rate": 1.0160115959974404e-07, "loss": 0.4534, "step": 18511 }, { "epoch": 0.9364156001821033, "grad_norm": 12.038260459899902, "learning_rate": 1.0144024391893048e-07, "loss": 0.4893, "step": 18512 }, { "epoch": 0.9364661844301685, "grad_norm": 6.0864105224609375, "learning_rate": 1.012794544621698e-07, "loss": 0.351, "step": 18513 }, { "epoch": 0.9365167686782336, "grad_norm": 7.114111423492432, "learning_rate": 1.011187912336059e-07, "loss": 0.2951, "step": 18514 }, { "epoch": 0.9365673529262988, "grad_norm": 6.304169178009033, "learning_rate": 1.0095825423737715e-07, "loss": 0.3887, "step": 18515 }, { "epoch": 0.9366179371743639, "grad_norm": 7.727136135101318, "learning_rate": 1.007978434776219e-07, "loss": 0.442, "step": 18516 }, { "epoch": 0.9366685214224291, "grad_norm": 9.161576271057129, "learning_rate": 1.0063755895847238e-07, "loss": 0.3505, "step": 18517 }, { "epoch": 0.9367191056704942, "grad_norm": 14.914972305297852, "learning_rate": 1.0047740068405809e-07, "loss": 0.3772, "step": 18518 }, { "epoch": 0.9367696899185594, "grad_norm": 8.570711135864258, "learning_rate": 1.0031736865850795e-07, "loss": 0.3542, "step": 18519 }, { "epoch": 0.9368202741666245, "grad_norm": 13.208691596984863, "learning_rate": 1.001574628859442e-07, "loss": 0.3278, "step": 18520 }, { "epoch": 0.9368708584146896, "grad_norm": 14.440109252929688, "learning_rate": 9.999768337048743e-08, "loss": 0.5266, "step": 18521 }, { "epoch": 0.9369214426627548, "grad_norm": 6.041179656982422, "learning_rate": 9.983803011625493e-08, "loss": 0.2599, "step": 18522 }, { "epoch": 0.9369720269108199, "grad_norm": 6.3292646408081055, "learning_rate": 9.967850312736116e-08, "loss": 0.3811, "step": 18523 }, { "epoch": 0.9370226111588851, "grad_norm": 8.287989616394043, "learning_rate": 9.951910240791507e-08, "loss": 0.3763, "step": 18524 }, { "epoch": 0.9370731954069502, "grad_norm": 5.512783527374268, "learning_rate": 9.935982796202559e-08, "loss": 0.4165, "step": 18525 }, { "epoch": 0.9371237796550155, "grad_norm": 6.935277462005615, "learning_rate": 9.920067979379666e-08, "loss": 0.4577, "step": 18526 }, { "epoch": 0.9371743639030806, "grad_norm": 9.130887031555176, "learning_rate": 9.904165790732778e-08, "loss": 0.3664, "step": 18527 }, { "epoch": 0.9372249481511458, "grad_norm": 6.249090671539307, "learning_rate": 9.8882762306719e-08, "loss": 0.324, "step": 18528 }, { "epoch": 0.9372755323992109, "grad_norm": 12.517516136169434, "learning_rate": 9.872399299606261e-08, "loss": 0.4992, "step": 18529 }, { "epoch": 0.937326116647276, "grad_norm": 5.985019207000732, "learning_rate": 9.856534997945033e-08, "loss": 0.3407, "step": 18530 }, { "epoch": 0.9373767008953412, "grad_norm": 9.316792488098145, "learning_rate": 9.840683326097001e-08, "loss": 0.4598, "step": 18531 }, { "epoch": 0.9374272851434063, "grad_norm": 7.034946441650391, "learning_rate": 9.824844284470614e-08, "loss": 0.4785, "step": 18532 }, { "epoch": 0.9374778693914715, "grad_norm": 4.609511852264404, "learning_rate": 9.809017873474048e-08, "loss": 0.3068, "step": 18533 }, { "epoch": 0.9375284536395366, "grad_norm": 8.801365852355957, "learning_rate": 9.793204093515085e-08, "loss": 0.4121, "step": 18534 }, { "epoch": 0.9375790378876018, "grad_norm": 6.971888065338135, "learning_rate": 9.77740294500118e-08, "loss": 0.3809, "step": 18535 }, { "epoch": 0.9376296221356669, "grad_norm": 5.4604902267456055, "learning_rate": 9.761614428339561e-08, "loss": 0.3677, "step": 18536 }, { "epoch": 0.9376802063837321, "grad_norm": 7.533256530761719, "learning_rate": 9.745838543936959e-08, "loss": 0.2304, "step": 18537 }, { "epoch": 0.9377307906317972, "grad_norm": 8.328495025634766, "learning_rate": 9.730075292199937e-08, "loss": 0.3939, "step": 18538 }, { "epoch": 0.9377813748798625, "grad_norm": 8.575386047363281, "learning_rate": 9.714324673534725e-08, "loss": 0.4297, "step": 18539 }, { "epoch": 0.9378319591279276, "grad_norm": 8.755304336547852, "learning_rate": 9.698586688347111e-08, "loss": 0.3606, "step": 18540 }, { "epoch": 0.9378825433759927, "grad_norm": 10.8473482131958, "learning_rate": 9.682861337042659e-08, "loss": 0.3879, "step": 18541 }, { "epoch": 0.9379331276240579, "grad_norm": 10.029560089111328, "learning_rate": 9.667148620026545e-08, "loss": 0.4369, "step": 18542 }, { "epoch": 0.937983711872123, "grad_norm": 7.771979808807373, "learning_rate": 9.651448537703666e-08, "loss": 0.4035, "step": 18543 }, { "epoch": 0.9380342961201882, "grad_norm": 10.47727108001709, "learning_rate": 9.635761090478534e-08, "loss": 0.3406, "step": 18544 }, { "epoch": 0.9380848803682533, "grad_norm": 13.109075546264648, "learning_rate": 9.620086278755492e-08, "loss": 0.3644, "step": 18545 }, { "epoch": 0.9381354646163185, "grad_norm": 10.108074188232422, "learning_rate": 9.604424102938325e-08, "loss": 0.3823, "step": 18546 }, { "epoch": 0.9381860488643836, "grad_norm": 20.964065551757812, "learning_rate": 9.588774563430603e-08, "loss": 0.4103, "step": 18547 }, { "epoch": 0.9382366331124488, "grad_norm": 10.775022506713867, "learning_rate": 9.573137660635612e-08, "loss": 0.4972, "step": 18548 }, { "epoch": 0.9382872173605139, "grad_norm": 8.579817771911621, "learning_rate": 9.557513394956364e-08, "loss": 0.3392, "step": 18549 }, { "epoch": 0.938337801608579, "grad_norm": 11.142271995544434, "learning_rate": 9.541901766795313e-08, "loss": 0.4216, "step": 18550 }, { "epoch": 0.9383883858566442, "grad_norm": 9.368863105773926, "learning_rate": 9.526302776554808e-08, "loss": 0.3473, "step": 18551 }, { "epoch": 0.9384389701047094, "grad_norm": 10.153229713439941, "learning_rate": 9.510716424636856e-08, "loss": 0.4212, "step": 18552 }, { "epoch": 0.9384895543527746, "grad_norm": 13.761255264282227, "learning_rate": 9.495142711442918e-08, "loss": 0.6894, "step": 18553 }, { "epoch": 0.9385401386008397, "grad_norm": 8.76474666595459, "learning_rate": 9.479581637374502e-08, "loss": 0.4004, "step": 18554 }, { "epoch": 0.9385907228489049, "grad_norm": 5.9343647956848145, "learning_rate": 9.4640332028324e-08, "loss": 0.4195, "step": 18555 }, { "epoch": 0.93864130709697, "grad_norm": 7.135230541229248, "learning_rate": 9.448497408217294e-08, "loss": 0.3551, "step": 18556 }, { "epoch": 0.9386918913450352, "grad_norm": 7.0781145095825195, "learning_rate": 9.432974253929583e-08, "loss": 0.396, "step": 18557 }, { "epoch": 0.9387424755931003, "grad_norm": 5.666287422180176, "learning_rate": 9.417463740369171e-08, "loss": 0.3801, "step": 18558 }, { "epoch": 0.9387930598411655, "grad_norm": 8.711403846740723, "learning_rate": 9.401965867935736e-08, "loss": 0.3502, "step": 18559 }, { "epoch": 0.9388436440892306, "grad_norm": 5.9261016845703125, "learning_rate": 9.386480637028739e-08, "loss": 0.4215, "step": 18560 }, { "epoch": 0.9388942283372957, "grad_norm": 7.615523338317871, "learning_rate": 9.371008048047025e-08, "loss": 0.4609, "step": 18561 }, { "epoch": 0.9389448125853609, "grad_norm": 7.506028652191162, "learning_rate": 9.355548101389389e-08, "loss": 0.3304, "step": 18562 }, { "epoch": 0.938995396833426, "grad_norm": 9.811142921447754, "learning_rate": 9.340100797454177e-08, "loss": 0.4115, "step": 18563 }, { "epoch": 0.9390459810814912, "grad_norm": 7.4648308753967285, "learning_rate": 9.32466613663946e-08, "loss": 0.3651, "step": 18564 }, { "epoch": 0.9390965653295563, "grad_norm": 7.419430255889893, "learning_rate": 9.309244119342864e-08, "loss": 0.3063, "step": 18565 }, { "epoch": 0.9391471495776216, "grad_norm": 10.044570922851562, "learning_rate": 9.293834745961905e-08, "loss": 0.4294, "step": 18566 }, { "epoch": 0.9391977338256867, "grad_norm": 11.353018760681152, "learning_rate": 9.278438016893543e-08, "loss": 0.4594, "step": 18567 }, { "epoch": 0.9392483180737519, "grad_norm": 11.756184577941895, "learning_rate": 9.263053932534516e-08, "loss": 0.5755, "step": 18568 }, { "epoch": 0.939298902321817, "grad_norm": 7.042170524597168, "learning_rate": 9.247682493281285e-08, "loss": 0.3328, "step": 18569 }, { "epoch": 0.9393494865698822, "grad_norm": 11.438068389892578, "learning_rate": 9.232323699529866e-08, "loss": 0.4241, "step": 18570 }, { "epoch": 0.9394000708179473, "grad_norm": 5.76611852645874, "learning_rate": 9.216977551676165e-08, "loss": 0.343, "step": 18571 }, { "epoch": 0.9394506550660124, "grad_norm": 7.64417028427124, "learning_rate": 9.201644050115422e-08, "loss": 0.4469, "step": 18572 }, { "epoch": 0.9395012393140776, "grad_norm": 7.454230785369873, "learning_rate": 9.186323195242874e-08, "loss": 0.3854, "step": 18573 }, { "epoch": 0.9395518235621427, "grad_norm": 6.988358497619629, "learning_rate": 9.171014987453264e-08, "loss": 0.3599, "step": 18574 }, { "epoch": 0.9396024078102079, "grad_norm": 14.987787246704102, "learning_rate": 9.155719427141053e-08, "loss": 0.4215, "step": 18575 }, { "epoch": 0.939652992058273, "grad_norm": 9.277048110961914, "learning_rate": 9.140436514700368e-08, "loss": 0.432, "step": 18576 }, { "epoch": 0.9397035763063382, "grad_norm": 16.33689308166504, "learning_rate": 9.125166250524952e-08, "loss": 0.6616, "step": 18577 }, { "epoch": 0.9397541605544033, "grad_norm": 6.679338455200195, "learning_rate": 9.109908635008435e-08, "loss": 0.3574, "step": 18578 }, { "epoch": 0.9398047448024686, "grad_norm": 12.297162055969238, "learning_rate": 9.094663668543835e-08, "loss": 0.4693, "step": 18579 }, { "epoch": 0.9398553290505337, "grad_norm": 16.63976287841797, "learning_rate": 9.079431351524004e-08, "loss": 0.4271, "step": 18580 }, { "epoch": 0.9399059132985988, "grad_norm": 6.237412452697754, "learning_rate": 9.064211684341517e-08, "loss": 0.4386, "step": 18581 }, { "epoch": 0.939956497546664, "grad_norm": 10.25076675415039, "learning_rate": 9.049004667388395e-08, "loss": 0.4322, "step": 18582 }, { "epoch": 0.9400070817947291, "grad_norm": 7.560945987701416, "learning_rate": 9.033810301056655e-08, "loss": 0.3545, "step": 18583 }, { "epoch": 0.9400576660427943, "grad_norm": 8.462507247924805, "learning_rate": 9.01862858573771e-08, "loss": 0.3658, "step": 18584 }, { "epoch": 0.9401082502908594, "grad_norm": 6.855162143707275, "learning_rate": 9.003459521822744e-08, "loss": 0.3689, "step": 18585 }, { "epoch": 0.9401588345389246, "grad_norm": 5.5068678855896, "learning_rate": 8.988303109702779e-08, "loss": 0.2855, "step": 18586 }, { "epoch": 0.9402094187869897, "grad_norm": 11.3052339553833, "learning_rate": 8.973159349768168e-08, "loss": 0.2464, "step": 18587 }, { "epoch": 0.9402600030350549, "grad_norm": 5.905853271484375, "learning_rate": 8.958028242409211e-08, "loss": 0.3739, "step": 18588 }, { "epoch": 0.94031058728312, "grad_norm": 7.9116902351379395, "learning_rate": 8.942909788015875e-08, "loss": 0.5402, "step": 18589 }, { "epoch": 0.9403611715311851, "grad_norm": 9.355834007263184, "learning_rate": 8.927803986977623e-08, "loss": 0.3743, "step": 18590 }, { "epoch": 0.9404117557792503, "grad_norm": 7.434454917907715, "learning_rate": 8.912710839683702e-08, "loss": 0.3508, "step": 18591 }, { "epoch": 0.9404623400273155, "grad_norm": 11.558551788330078, "learning_rate": 8.897630346523078e-08, "loss": 0.5167, "step": 18592 }, { "epoch": 0.9405129242753807, "grad_norm": 6.0091657638549805, "learning_rate": 8.882562507884329e-08, "loss": 0.2967, "step": 18593 }, { "epoch": 0.9405635085234458, "grad_norm": 6.847412109375, "learning_rate": 8.867507324155644e-08, "loss": 0.3172, "step": 18594 }, { "epoch": 0.940614092771511, "grad_norm": 6.941685676574707, "learning_rate": 8.852464795725046e-08, "loss": 0.3352, "step": 18595 }, { "epoch": 0.9406646770195761, "grad_norm": 9.079080581665039, "learning_rate": 8.837434922980115e-08, "loss": 0.3986, "step": 18596 }, { "epoch": 0.9407152612676413, "grad_norm": 13.753517150878906, "learning_rate": 8.822417706308096e-08, "loss": 0.4085, "step": 18597 }, { "epoch": 0.9407658455157064, "grad_norm": 9.756038665771484, "learning_rate": 8.80741314609601e-08, "loss": 0.4554, "step": 18598 }, { "epoch": 0.9408164297637716, "grad_norm": 5.525967121124268, "learning_rate": 8.792421242730442e-08, "loss": 0.2941, "step": 18599 }, { "epoch": 0.9408670140118367, "grad_norm": 28.933834075927734, "learning_rate": 8.777441996597691e-08, "loss": 0.4802, "step": 18600 }, { "epoch": 0.9409175982599018, "grad_norm": 9.865110397338867, "learning_rate": 8.762475408083838e-08, "loss": 0.4569, "step": 18601 }, { "epoch": 0.940968182507967, "grad_norm": 8.177359580993652, "learning_rate": 8.747521477574405e-08, "loss": 0.3818, "step": 18602 }, { "epoch": 0.9410187667560321, "grad_norm": 7.828469753265381, "learning_rate": 8.732580205454755e-08, "loss": 0.3987, "step": 18603 }, { "epoch": 0.9410693510040973, "grad_norm": 10.407084465026855, "learning_rate": 8.717651592109965e-08, "loss": 0.5588, "step": 18604 }, { "epoch": 0.9411199352521625, "grad_norm": 10.712135314941406, "learning_rate": 8.70273563792462e-08, "loss": 0.3484, "step": 18605 }, { "epoch": 0.9411705195002277, "grad_norm": 11.422195434570312, "learning_rate": 8.687832343283075e-08, "loss": 0.3939, "step": 18606 }, { "epoch": 0.9412211037482928, "grad_norm": 10.924432754516602, "learning_rate": 8.672941708569415e-08, "loss": 0.3665, "step": 18607 }, { "epoch": 0.941271687996358, "grad_norm": 9.438128471374512, "learning_rate": 8.658063734167277e-08, "loss": 0.4356, "step": 18608 }, { "epoch": 0.9413222722444231, "grad_norm": 18.7841854095459, "learning_rate": 8.64319842046002e-08, "loss": 0.4262, "step": 18609 }, { "epoch": 0.9413728564924883, "grad_norm": 5.489501953125, "learning_rate": 8.628345767830782e-08, "loss": 0.3844, "step": 18610 }, { "epoch": 0.9414234407405534, "grad_norm": 7.328555583953857, "learning_rate": 8.613505776662145e-08, "loss": 0.3777, "step": 18611 }, { "epoch": 0.9414740249886185, "grad_norm": 10.664081573486328, "learning_rate": 8.598678447336583e-08, "loss": 0.3904, "step": 18612 }, { "epoch": 0.9415246092366837, "grad_norm": 11.706425666809082, "learning_rate": 8.583863780236179e-08, "loss": 0.4407, "step": 18613 }, { "epoch": 0.9415751934847488, "grad_norm": 7.04701042175293, "learning_rate": 8.569061775742627e-08, "loss": 0.3374, "step": 18614 }, { "epoch": 0.941625777732814, "grad_norm": 7.766056060791016, "learning_rate": 8.554272434237342e-08, "loss": 0.3149, "step": 18615 }, { "epoch": 0.9416763619808791, "grad_norm": 17.754257202148438, "learning_rate": 8.539495756101412e-08, "loss": 0.5799, "step": 18616 }, { "epoch": 0.9417269462289443, "grad_norm": 11.754110336303711, "learning_rate": 8.524731741715641e-08, "loss": 0.4933, "step": 18617 }, { "epoch": 0.9417775304770094, "grad_norm": 15.276450157165527, "learning_rate": 8.509980391460393e-08, "loss": 0.4525, "step": 18618 }, { "epoch": 0.9418281147250747, "grad_norm": 6.120708465576172, "learning_rate": 8.495241705715807e-08, "loss": 0.3102, "step": 18619 }, { "epoch": 0.9418786989731398, "grad_norm": 9.636394500732422, "learning_rate": 8.480515684861634e-08, "loss": 0.4294, "step": 18620 }, { "epoch": 0.941929283221205, "grad_norm": 7.2681708335876465, "learning_rate": 8.465802329277406e-08, "loss": 0.397, "step": 18621 }, { "epoch": 0.9419798674692701, "grad_norm": 6.763967037200928, "learning_rate": 8.451101639342153e-08, "loss": 0.4129, "step": 18622 }, { "epoch": 0.9420304517173352, "grad_norm": 7.37451171875, "learning_rate": 8.436413615434735e-08, "loss": 0.2471, "step": 18623 }, { "epoch": 0.9420810359654004, "grad_norm": 7.450808048248291, "learning_rate": 8.421738257933631e-08, "loss": 0.4218, "step": 18624 }, { "epoch": 0.9421316202134655, "grad_norm": 7.4123311042785645, "learning_rate": 8.407075567216982e-08, "loss": 0.3498, "step": 18625 }, { "epoch": 0.9421822044615307, "grad_norm": 10.95557689666748, "learning_rate": 8.392425543662541e-08, "loss": 0.3871, "step": 18626 }, { "epoch": 0.9422327887095958, "grad_norm": 5.091943740844727, "learning_rate": 8.377788187647951e-08, "loss": 0.3115, "step": 18627 }, { "epoch": 0.942283372957661, "grad_norm": 9.991083145141602, "learning_rate": 8.363163499550242e-08, "loss": 0.3904, "step": 18628 }, { "epoch": 0.9423339572057261, "grad_norm": 17.20365333557129, "learning_rate": 8.348551479746281e-08, "loss": 0.323, "step": 18629 }, { "epoch": 0.9423845414537912, "grad_norm": 16.236583709716797, "learning_rate": 8.333952128612655e-08, "loss": 0.4655, "step": 18630 }, { "epoch": 0.9424351257018564, "grad_norm": 6.688578128814697, "learning_rate": 8.319365446525562e-08, "loss": 0.4036, "step": 18631 }, { "epoch": 0.9424857099499216, "grad_norm": 9.557785987854004, "learning_rate": 8.30479143386076e-08, "loss": 0.3185, "step": 18632 }, { "epoch": 0.9425362941979868, "grad_norm": 5.38953971862793, "learning_rate": 8.290230090993834e-08, "loss": 0.389, "step": 18633 }, { "epoch": 0.9425868784460519, "grad_norm": 8.808662414550781, "learning_rate": 8.275681418300041e-08, "loss": 0.5039, "step": 18634 }, { "epoch": 0.9426374626941171, "grad_norm": 5.80823278427124, "learning_rate": 8.261145416154138e-08, "loss": 0.3121, "step": 18635 }, { "epoch": 0.9426880469421822, "grad_norm": 5.103146553039551, "learning_rate": 8.246622084930878e-08, "loss": 0.3359, "step": 18636 }, { "epoch": 0.9427386311902474, "grad_norm": 7.172638416290283, "learning_rate": 8.232111425004352e-08, "loss": 0.433, "step": 18637 }, { "epoch": 0.9427892154383125, "grad_norm": 10.946596145629883, "learning_rate": 8.217613436748428e-08, "loss": 0.3057, "step": 18638 }, { "epoch": 0.9428397996863777, "grad_norm": 6.098752498626709, "learning_rate": 8.203128120536808e-08, "loss": 0.3271, "step": 18639 }, { "epoch": 0.9428903839344428, "grad_norm": 7.375319957733154, "learning_rate": 8.188655476742746e-08, "loss": 0.391, "step": 18640 }, { "epoch": 0.9429409681825079, "grad_norm": 9.205078125, "learning_rate": 8.174195505739002e-08, "loss": 0.4166, "step": 18641 }, { "epoch": 0.9429915524305731, "grad_norm": 4.818864345550537, "learning_rate": 8.159748207898332e-08, "loss": 0.4136, "step": 18642 }, { "epoch": 0.9430421366786382, "grad_norm": 10.478828430175781, "learning_rate": 8.145313583592995e-08, "loss": 0.3603, "step": 18643 }, { "epoch": 0.9430927209267034, "grad_norm": 29.878787994384766, "learning_rate": 8.130891633194859e-08, "loss": 0.5646, "step": 18644 }, { "epoch": 0.9431433051747686, "grad_norm": 7.961199760437012, "learning_rate": 8.116482357075628e-08, "loss": 0.3648, "step": 18645 }, { "epoch": 0.9431938894228338, "grad_norm": 8.578532218933105, "learning_rate": 8.102085755606504e-08, "loss": 0.3823, "step": 18646 }, { "epoch": 0.9432444736708989, "grad_norm": 13.476628303527832, "learning_rate": 8.087701829158523e-08, "loss": 0.3947, "step": 18647 }, { "epoch": 0.9432950579189641, "grad_norm": 5.970231056213379, "learning_rate": 8.073330578102279e-08, "loss": 0.4598, "step": 18648 }, { "epoch": 0.9433456421670292, "grad_norm": 7.285467624664307, "learning_rate": 8.058972002808085e-08, "loss": 0.3917, "step": 18649 }, { "epoch": 0.9433962264150944, "grad_norm": 4.3442206382751465, "learning_rate": 8.04462610364598e-08, "loss": 0.3055, "step": 18650 }, { "epoch": 0.9434468106631595, "grad_norm": 5.3844404220581055, "learning_rate": 8.030292880985613e-08, "loss": 0.3749, "step": 18651 }, { "epoch": 0.9434973949112246, "grad_norm": 13.08355712890625, "learning_rate": 8.015972335196298e-08, "loss": 0.479, "step": 18652 }, { "epoch": 0.9435479791592898, "grad_norm": 10.776165008544922, "learning_rate": 8.001664466646964e-08, "loss": 0.5599, "step": 18653 }, { "epoch": 0.9435985634073549, "grad_norm": 6.2379841804504395, "learning_rate": 7.987369275706424e-08, "loss": 0.3823, "step": 18654 }, { "epoch": 0.9436491476554201, "grad_norm": 29.52375602722168, "learning_rate": 7.973086762742999e-08, "loss": 0.6054, "step": 18655 }, { "epoch": 0.9436997319034852, "grad_norm": 5.39810848236084, "learning_rate": 7.958816928124614e-08, "loss": 0.3203, "step": 18656 }, { "epoch": 0.9437503161515504, "grad_norm": 4.786204814910889, "learning_rate": 7.944559772219084e-08, "loss": 0.2987, "step": 18657 }, { "epoch": 0.9438009003996155, "grad_norm": 8.171807289123535, "learning_rate": 7.930315295393731e-08, "loss": 0.3826, "step": 18658 }, { "epoch": 0.9438514846476808, "grad_norm": 9.494099617004395, "learning_rate": 7.916083498015592e-08, "loss": 0.3177, "step": 18659 }, { "epoch": 0.9439020688957459, "grad_norm": 20.17845916748047, "learning_rate": 7.901864380451429e-08, "loss": 0.5558, "step": 18660 }, { "epoch": 0.943952653143811, "grad_norm": 14.272171974182129, "learning_rate": 7.887657943067562e-08, "loss": 0.526, "step": 18661 }, { "epoch": 0.9440032373918762, "grad_norm": 6.059298992156982, "learning_rate": 7.87346418623014e-08, "loss": 0.3003, "step": 18662 }, { "epoch": 0.9440538216399413, "grad_norm": 10.421932220458984, "learning_rate": 7.859283110304817e-08, "loss": 0.3813, "step": 18663 }, { "epoch": 0.9441044058880065, "grad_norm": 5.3295817375183105, "learning_rate": 7.845114715657077e-08, "loss": 0.4154, "step": 18664 }, { "epoch": 0.9441549901360716, "grad_norm": 4.668020725250244, "learning_rate": 7.830959002652016e-08, "loss": 0.2857, "step": 18665 }, { "epoch": 0.9442055743841368, "grad_norm": 9.784781455993652, "learning_rate": 7.81681597165429e-08, "loss": 0.3033, "step": 18666 }, { "epoch": 0.9442561586322019, "grad_norm": 5.876142501831055, "learning_rate": 7.802685623028439e-08, "loss": 0.4278, "step": 18667 }, { "epoch": 0.9443067428802671, "grad_norm": 9.203264236450195, "learning_rate": 7.788567957138504e-08, "loss": 0.3906, "step": 18668 }, { "epoch": 0.9443573271283322, "grad_norm": 7.335984706878662, "learning_rate": 7.774462974348362e-08, "loss": 0.3671, "step": 18669 }, { "epoch": 0.9444079113763973, "grad_norm": 8.12793254852295, "learning_rate": 7.760370675021278e-08, "loss": 0.3909, "step": 18670 }, { "epoch": 0.9444584956244625, "grad_norm": 8.92744255065918, "learning_rate": 7.746291059520573e-08, "loss": 0.3846, "step": 18671 }, { "epoch": 0.9445090798725277, "grad_norm": 11.856094360351562, "learning_rate": 7.732224128208898e-08, "loss": 0.5433, "step": 18672 }, { "epoch": 0.9445596641205929, "grad_norm": 6.734011173248291, "learning_rate": 7.718169881448801e-08, "loss": 0.3568, "step": 18673 }, { "epoch": 0.944610248368658, "grad_norm": 14.779322624206543, "learning_rate": 7.704128319602378e-08, "loss": 0.6121, "step": 18674 }, { "epoch": 0.9446608326167232, "grad_norm": 5.359069347381592, "learning_rate": 7.690099443031562e-08, "loss": 0.422, "step": 18675 }, { "epoch": 0.9447114168647883, "grad_norm": 7.591206073760986, "learning_rate": 7.676083252097677e-08, "loss": 0.4018, "step": 18676 }, { "epoch": 0.9447620011128535, "grad_norm": 6.843005180358887, "learning_rate": 7.662079747161988e-08, "loss": 0.315, "step": 18677 }, { "epoch": 0.9448125853609186, "grad_norm": 7.267238140106201, "learning_rate": 7.648088928585317e-08, "loss": 0.3912, "step": 18678 }, { "epoch": 0.9448631696089838, "grad_norm": 8.468755722045898, "learning_rate": 7.634110796728156e-08, "loss": 0.4345, "step": 18679 }, { "epoch": 0.9449137538570489, "grad_norm": 6.292484760284424, "learning_rate": 7.620145351950714e-08, "loss": 0.4085, "step": 18680 }, { "epoch": 0.944964338105114, "grad_norm": 6.346924781799316, "learning_rate": 7.606192594612815e-08, "loss": 0.36, "step": 18681 }, { "epoch": 0.9450149223531792, "grad_norm": 10.211682319641113, "learning_rate": 7.592252525074006e-08, "loss": 0.3935, "step": 18682 }, { "epoch": 0.9450655066012443, "grad_norm": 8.38452434539795, "learning_rate": 7.578325143693499e-08, "loss": 0.3732, "step": 18683 }, { "epoch": 0.9451160908493095, "grad_norm": 8.311457633972168, "learning_rate": 7.564410450830117e-08, "loss": 0.4957, "step": 18684 }, { "epoch": 0.9451666750973747, "grad_norm": 10.306647300720215, "learning_rate": 7.55050844684252e-08, "loss": 0.4586, "step": 18685 }, { "epoch": 0.9452172593454399, "grad_norm": 10.34877872467041, "learning_rate": 7.536619132088808e-08, "loss": 0.4627, "step": 18686 }, { "epoch": 0.945267843593505, "grad_norm": 9.030985832214355, "learning_rate": 7.522742506926973e-08, "loss": 0.4228, "step": 18687 }, { "epoch": 0.9453184278415702, "grad_norm": 6.6306023597717285, "learning_rate": 7.508878571714451e-08, "loss": 0.3373, "step": 18688 }, { "epoch": 0.9453690120896353, "grad_norm": 8.543844223022461, "learning_rate": 7.495027326808624e-08, "loss": 0.3048, "step": 18689 }, { "epoch": 0.9454195963377005, "grad_norm": 6.866329669952393, "learning_rate": 7.481188772566372e-08, "loss": 0.4119, "step": 18690 }, { "epoch": 0.9454701805857656, "grad_norm": 5.864758491516113, "learning_rate": 7.46736290934419e-08, "loss": 0.3456, "step": 18691 }, { "epoch": 0.9455207648338307, "grad_norm": 11.380112648010254, "learning_rate": 7.4535497374984e-08, "loss": 0.4211, "step": 18692 }, { "epoch": 0.9455713490818959, "grad_norm": 10.630448341369629, "learning_rate": 7.439749257384998e-08, "loss": 0.3517, "step": 18693 }, { "epoch": 0.945621933329961, "grad_norm": 11.536186218261719, "learning_rate": 7.425961469359478e-08, "loss": 0.444, "step": 18694 }, { "epoch": 0.9456725175780262, "grad_norm": 9.665511131286621, "learning_rate": 7.41218637377722e-08, "loss": 0.3726, "step": 18695 }, { "epoch": 0.9457231018260913, "grad_norm": 12.259818077087402, "learning_rate": 7.398423970993052e-08, "loss": 0.4378, "step": 18696 }, { "epoch": 0.9457736860741565, "grad_norm": 7.115213394165039, "learning_rate": 7.38467426136169e-08, "loss": 0.3076, "step": 18697 }, { "epoch": 0.9458242703222216, "grad_norm": 7.375149250030518, "learning_rate": 7.370937245237464e-08, "loss": 0.3133, "step": 18698 }, { "epoch": 0.9458748545702869, "grad_norm": 14.760591506958008, "learning_rate": 7.3572129229742e-08, "loss": 0.4437, "step": 18699 }, { "epoch": 0.945925438818352, "grad_norm": 6.553481101989746, "learning_rate": 7.343501294925671e-08, "loss": 0.3455, "step": 18700 }, { "epoch": 0.9459760230664171, "grad_norm": 4.9344868659973145, "learning_rate": 7.329802361445204e-08, "loss": 0.3528, "step": 18701 }, { "epoch": 0.9460266073144823, "grad_norm": 8.594598770141602, "learning_rate": 7.31611612288563e-08, "loss": 0.4992, "step": 18702 }, { "epoch": 0.9460771915625474, "grad_norm": 10.169405937194824, "learning_rate": 7.302442579599778e-08, "loss": 0.5129, "step": 18703 }, { "epoch": 0.9461277758106126, "grad_norm": 9.234993934631348, "learning_rate": 7.28878173193992e-08, "loss": 0.352, "step": 18704 }, { "epoch": 0.9461783600586777, "grad_norm": 9.542488098144531, "learning_rate": 7.275133580257998e-08, "loss": 0.4404, "step": 18705 }, { "epoch": 0.9462289443067429, "grad_norm": 9.673604965209961, "learning_rate": 7.261498124905841e-08, "loss": 0.408, "step": 18706 }, { "epoch": 0.946279528554808, "grad_norm": 6.048247337341309, "learning_rate": 7.247875366234669e-08, "loss": 0.4324, "step": 18707 }, { "epoch": 0.9463301128028732, "grad_norm": 6.741060256958008, "learning_rate": 7.234265304595589e-08, "loss": 0.3412, "step": 18708 }, { "epoch": 0.9463806970509383, "grad_norm": 5.632361888885498, "learning_rate": 7.220667940339266e-08, "loss": 0.2366, "step": 18709 }, { "epoch": 0.9464312812990034, "grad_norm": 19.701351165771484, "learning_rate": 7.207083273816029e-08, "loss": 0.4557, "step": 18710 }, { "epoch": 0.9464818655470686, "grad_norm": 8.054839134216309, "learning_rate": 7.193511305375989e-08, "loss": 0.3749, "step": 18711 }, { "epoch": 0.9465324497951338, "grad_norm": 6.003195762634277, "learning_rate": 7.17995203536881e-08, "loss": 0.3619, "step": 18712 }, { "epoch": 0.946583034043199, "grad_norm": 10.208989143371582, "learning_rate": 7.166405464143989e-08, "loss": 0.4108, "step": 18713 }, { "epoch": 0.9466336182912641, "grad_norm": 8.264172554016113, "learning_rate": 7.152871592050414e-08, "loss": 0.3125, "step": 18714 }, { "epoch": 0.9466842025393293, "grad_norm": 8.563299179077148, "learning_rate": 7.139350419436919e-08, "loss": 0.3879, "step": 18715 }, { "epoch": 0.9467347867873944, "grad_norm": 8.747906684875488, "learning_rate": 7.125841946651946e-08, "loss": 0.4497, "step": 18716 }, { "epoch": 0.9467853710354596, "grad_norm": 7.858222961425781, "learning_rate": 7.112346174043549e-08, "loss": 0.3812, "step": 18717 }, { "epoch": 0.9468359552835247, "grad_norm": 9.909679412841797, "learning_rate": 7.098863101959452e-08, "loss": 0.3817, "step": 18718 }, { "epoch": 0.9468865395315899, "grad_norm": 8.405405044555664, "learning_rate": 7.085392730747098e-08, "loss": 0.4475, "step": 18719 }, { "epoch": 0.946937123779655, "grad_norm": 8.882883071899414, "learning_rate": 7.071935060753543e-08, "loss": 0.3688, "step": 18720 }, { "epoch": 0.9469877080277201, "grad_norm": 20.798385620117188, "learning_rate": 7.058490092325731e-08, "loss": 0.5376, "step": 18721 }, { "epoch": 0.9470382922757853, "grad_norm": 7.1407318115234375, "learning_rate": 7.045057825809832e-08, "loss": 0.3861, "step": 18722 }, { "epoch": 0.9470888765238504, "grad_norm": 10.56372356414795, "learning_rate": 7.031638261552176e-08, "loss": 0.408, "step": 18723 }, { "epoch": 0.9471394607719156, "grad_norm": 8.281481742858887, "learning_rate": 7.01823139989849e-08, "loss": 0.442, "step": 18724 }, { "epoch": 0.9471900450199808, "grad_norm": 6.2357177734375, "learning_rate": 7.004837241194218e-08, "loss": 0.4057, "step": 18725 }, { "epoch": 0.947240629268046, "grad_norm": 8.83726978302002, "learning_rate": 6.991455785784529e-08, "loss": 0.398, "step": 18726 }, { "epoch": 0.9472912135161111, "grad_norm": 11.111944198608398, "learning_rate": 6.978087034014203e-08, "loss": 0.3443, "step": 18727 }, { "epoch": 0.9473417977641763, "grad_norm": 5.238409519195557, "learning_rate": 6.964730986227685e-08, "loss": 0.282, "step": 18728 }, { "epoch": 0.9473923820122414, "grad_norm": 9.327630996704102, "learning_rate": 6.951387642769202e-08, "loss": 0.3551, "step": 18729 }, { "epoch": 0.9474429662603066, "grad_norm": 6.453692436218262, "learning_rate": 6.938057003982591e-08, "loss": 0.435, "step": 18730 }, { "epoch": 0.9474935505083717, "grad_norm": 6.756459712982178, "learning_rate": 6.92473907021124e-08, "loss": 0.4034, "step": 18731 }, { "epoch": 0.9475441347564368, "grad_norm": 6.207891941070557, "learning_rate": 6.911433841798432e-08, "loss": 0.3929, "step": 18732 }, { "epoch": 0.947594719004502, "grad_norm": 9.16525936126709, "learning_rate": 6.898141319086948e-08, "loss": 0.3844, "step": 18733 }, { "epoch": 0.9476453032525671, "grad_norm": 14.103151321411133, "learning_rate": 6.884861502419349e-08, "loss": 0.4875, "step": 18734 }, { "epoch": 0.9476958875006323, "grad_norm": 7.444729328155518, "learning_rate": 6.871594392137803e-08, "loss": 0.4295, "step": 18735 }, { "epoch": 0.9477464717486974, "grad_norm": 6.271996974945068, "learning_rate": 6.85833998858415e-08, "loss": 0.3562, "step": 18736 }, { "epoch": 0.9477970559967626, "grad_norm": 9.475500106811523, "learning_rate": 6.845098292099894e-08, "loss": 0.3792, "step": 18737 }, { "epoch": 0.9478476402448278, "grad_norm": 5.894111156463623, "learning_rate": 6.831869303026317e-08, "loss": 0.4099, "step": 18738 }, { "epoch": 0.947898224492893, "grad_norm": 7.864254474639893, "learning_rate": 6.818653021704313e-08, "loss": 0.4094, "step": 18739 }, { "epoch": 0.9479488087409581, "grad_norm": 6.037469387054443, "learning_rate": 6.805449448474332e-08, "loss": 0.3623, "step": 18740 }, { "epoch": 0.9479993929890232, "grad_norm": 5.521114826202393, "learning_rate": 6.792258583676659e-08, "loss": 0.3294, "step": 18741 }, { "epoch": 0.9480499772370884, "grad_norm": 5.947604179382324, "learning_rate": 6.779080427651242e-08, "loss": 0.2764, "step": 18742 }, { "epoch": 0.9481005614851535, "grad_norm": 11.808145523071289, "learning_rate": 6.765914980737531e-08, "loss": 0.4906, "step": 18743 }, { "epoch": 0.9481511457332187, "grad_norm": 8.246686935424805, "learning_rate": 6.752762243274868e-08, "loss": 0.4262, "step": 18744 }, { "epoch": 0.9482017299812838, "grad_norm": 8.051230430603027, "learning_rate": 6.739622215602149e-08, "loss": 0.4457, "step": 18745 }, { "epoch": 0.948252314229349, "grad_norm": 12.249296188354492, "learning_rate": 6.726494898057934e-08, "loss": 0.3375, "step": 18746 }, { "epoch": 0.9483028984774141, "grad_norm": 7.2810564041137695, "learning_rate": 6.713380290980508e-08, "loss": 0.3035, "step": 18747 }, { "epoch": 0.9483534827254793, "grad_norm": 6.165768623352051, "learning_rate": 6.70027839470777e-08, "loss": 0.2879, "step": 18748 }, { "epoch": 0.9484040669735444, "grad_norm": 5.0184831619262695, "learning_rate": 6.687189209577283e-08, "loss": 0.3969, "step": 18749 }, { "epoch": 0.9484546512216095, "grad_norm": 6.387574672698975, "learning_rate": 6.674112735926442e-08, "loss": 0.401, "step": 18750 }, { "epoch": 0.9485052354696747, "grad_norm": 19.509084701538086, "learning_rate": 6.661048974092144e-08, "loss": 0.7762, "step": 18751 }, { "epoch": 0.94855581971774, "grad_norm": 7.041800022125244, "learning_rate": 6.647997924410954e-08, "loss": 0.3838, "step": 18752 }, { "epoch": 0.9486064039658051, "grad_norm": 5.882001876831055, "learning_rate": 6.634959587219269e-08, "loss": 0.3922, "step": 18753 }, { "epoch": 0.9486569882138702, "grad_norm": 4.046785354614258, "learning_rate": 6.621933962852988e-08, "loss": 0.2805, "step": 18754 }, { "epoch": 0.9487075724619354, "grad_norm": 5.841108798980713, "learning_rate": 6.608921051647732e-08, "loss": 0.3439, "step": 18755 }, { "epoch": 0.9487581567100005, "grad_norm": 6.996910572052002, "learning_rate": 6.595920853938897e-08, "loss": 0.4292, "step": 18756 }, { "epoch": 0.9488087409580657, "grad_norm": 5.851128578186035, "learning_rate": 6.582933370061384e-08, "loss": 0.3247, "step": 18757 }, { "epoch": 0.9488593252061308, "grad_norm": 8.31985855102539, "learning_rate": 6.569958600349813e-08, "loss": 0.4369, "step": 18758 }, { "epoch": 0.948909909454196, "grad_norm": 9.816922187805176, "learning_rate": 6.556996545138694e-08, "loss": 0.4604, "step": 18759 }, { "epoch": 0.9489604937022611, "grad_norm": 8.844131469726562, "learning_rate": 6.544047204761872e-08, "loss": 0.4409, "step": 18760 }, { "epoch": 0.9490110779503262, "grad_norm": 10.427660942077637, "learning_rate": 6.531110579553024e-08, "loss": 0.4832, "step": 18761 }, { "epoch": 0.9490616621983914, "grad_norm": 8.633746147155762, "learning_rate": 6.51818666984555e-08, "loss": 0.4392, "step": 18762 }, { "epoch": 0.9491122464464565, "grad_norm": 7.004398822784424, "learning_rate": 6.505275475972462e-08, "loss": 0.3642, "step": 18763 }, { "epoch": 0.9491628306945217, "grad_norm": 8.355682373046875, "learning_rate": 6.492376998266381e-08, "loss": 0.3873, "step": 18764 }, { "epoch": 0.9492134149425869, "grad_norm": 7.04317569732666, "learning_rate": 6.479491237059766e-08, "loss": 0.3567, "step": 18765 }, { "epoch": 0.9492639991906521, "grad_norm": 9.615860939025879, "learning_rate": 6.466618192684627e-08, "loss": 0.3028, "step": 18766 }, { "epoch": 0.9493145834387172, "grad_norm": 12.116744041442871, "learning_rate": 6.453757865472643e-08, "loss": 0.4662, "step": 18767 }, { "epoch": 0.9493651676867824, "grad_norm": 7.395170211791992, "learning_rate": 6.440910255755218e-08, "loss": 0.4348, "step": 18768 }, { "epoch": 0.9494157519348475, "grad_norm": 8.033798217773438, "learning_rate": 6.428075363863362e-08, "loss": 0.3905, "step": 18769 }, { "epoch": 0.9494663361829127, "grad_norm": 13.651740074157715, "learning_rate": 6.415253190127868e-08, "loss": 0.4236, "step": 18770 }, { "epoch": 0.9495169204309778, "grad_norm": 9.343905448913574, "learning_rate": 6.402443734879082e-08, "loss": 0.3962, "step": 18771 }, { "epoch": 0.9495675046790429, "grad_norm": 7.775148391723633, "learning_rate": 6.389646998447074e-08, "loss": 0.3764, "step": 18772 }, { "epoch": 0.9496180889271081, "grad_norm": 12.41009521484375, "learning_rate": 6.37686298116158e-08, "loss": 0.5365, "step": 18773 }, { "epoch": 0.9496686731751732, "grad_norm": 8.800039291381836, "learning_rate": 6.364091683352058e-08, "loss": 0.3847, "step": 18774 }, { "epoch": 0.9497192574232384, "grad_norm": 6.701277256011963, "learning_rate": 6.351333105347579e-08, "loss": 0.3283, "step": 18775 }, { "epoch": 0.9497698416713035, "grad_norm": 12.124506950378418, "learning_rate": 6.338587247476824e-08, "loss": 0.5221, "step": 18776 }, { "epoch": 0.9498204259193687, "grad_norm": 12.271927833557129, "learning_rate": 6.325854110068363e-08, "loss": 0.4162, "step": 18777 }, { "epoch": 0.9498710101674339, "grad_norm": 7.434778690338135, "learning_rate": 6.313133693450158e-08, "loss": 0.438, "step": 18778 }, { "epoch": 0.9499215944154991, "grad_norm": 6.231997489929199, "learning_rate": 6.300425997950055e-08, "loss": 0.4184, "step": 18779 }, { "epoch": 0.9499721786635642, "grad_norm": 9.891489028930664, "learning_rate": 6.287731023895516e-08, "loss": 0.4217, "step": 18780 }, { "epoch": 0.9500227629116293, "grad_norm": 9.175618171691895, "learning_rate": 6.275048771613612e-08, "loss": 0.3784, "step": 18781 }, { "epoch": 0.9500733471596945, "grad_norm": 4.070508003234863, "learning_rate": 6.262379241431193e-08, "loss": 0.2754, "step": 18782 }, { "epoch": 0.9501239314077596, "grad_norm": 8.28628921508789, "learning_rate": 6.249722433674666e-08, "loss": 0.4215, "step": 18783 }, { "epoch": 0.9501745156558248, "grad_norm": 12.00312614440918, "learning_rate": 6.237078348670156e-08, "loss": 0.3111, "step": 18784 }, { "epoch": 0.9502250999038899, "grad_norm": 7.61575174331665, "learning_rate": 6.224446986743571e-08, "loss": 0.4572, "step": 18785 }, { "epoch": 0.9502756841519551, "grad_norm": 5.650637626647949, "learning_rate": 6.211828348220317e-08, "loss": 0.4021, "step": 18786 }, { "epoch": 0.9503262684000202, "grad_norm": 8.004692077636719, "learning_rate": 6.199222433425467e-08, "loss": 0.3301, "step": 18787 }, { "epoch": 0.9503768526480854, "grad_norm": 9.217269897460938, "learning_rate": 6.186629242683984e-08, "loss": 0.3805, "step": 18788 }, { "epoch": 0.9504274368961505, "grad_norm": 6.0739827156066895, "learning_rate": 6.174048776320329e-08, "loss": 0.4351, "step": 18789 }, { "epoch": 0.9504780211442156, "grad_norm": 7.451237201690674, "learning_rate": 6.161481034658579e-08, "loss": 0.3297, "step": 18790 }, { "epoch": 0.9505286053922808, "grad_norm": 6.6393256187438965, "learning_rate": 6.148926018022694e-08, "loss": 0.4388, "step": 18791 }, { "epoch": 0.950579189640346, "grad_norm": 7.914337158203125, "learning_rate": 6.13638372673614e-08, "loss": 0.3101, "step": 18792 }, { "epoch": 0.9506297738884112, "grad_norm": 9.177831649780273, "learning_rate": 6.1238541611221e-08, "loss": 0.3168, "step": 18793 }, { "epoch": 0.9506803581364763, "grad_norm": 10.921704292297363, "learning_rate": 6.111337321503374e-08, "loss": 0.4346, "step": 18794 }, { "epoch": 0.9507309423845415, "grad_norm": 10.174351692199707, "learning_rate": 6.098833208202648e-08, "loss": 0.4273, "step": 18795 }, { "epoch": 0.9507815266326066, "grad_norm": 7.725271224975586, "learning_rate": 6.08634182154194e-08, "loss": 0.3985, "step": 18796 }, { "epoch": 0.9508321108806718, "grad_norm": 6.248782634735107, "learning_rate": 6.073863161843219e-08, "loss": 0.3584, "step": 18797 }, { "epoch": 0.9508826951287369, "grad_norm": 7.619668006896973, "learning_rate": 6.061397229428056e-08, "loss": 0.3894, "step": 18798 }, { "epoch": 0.950933279376802, "grad_norm": 6.345199108123779, "learning_rate": 6.048944024617587e-08, "loss": 0.3437, "step": 18799 }, { "epoch": 0.9509838636248672, "grad_norm": 6.732173919677734, "learning_rate": 6.03650354773272e-08, "loss": 0.3652, "step": 18800 }, { "epoch": 0.9510344478729323, "grad_norm": 11.416534423828125, "learning_rate": 6.024075799094086e-08, "loss": 0.3835, "step": 18801 }, { "epoch": 0.9510850321209975, "grad_norm": 11.395992279052734, "learning_rate": 6.011660779021822e-08, "loss": 0.3676, "step": 18802 }, { "epoch": 0.9511356163690626, "grad_norm": 8.351953506469727, "learning_rate": 5.999258487835891e-08, "loss": 0.4431, "step": 18803 }, { "epoch": 0.9511862006171278, "grad_norm": 14.26352596282959, "learning_rate": 5.986868925855871e-08, "loss": 0.4438, "step": 18804 }, { "epoch": 0.951236784865193, "grad_norm": 7.93228816986084, "learning_rate": 5.974492093401007e-08, "loss": 0.3668, "step": 18805 }, { "epoch": 0.9512873691132582, "grad_norm": 19.223379135131836, "learning_rate": 5.962127990790212e-08, "loss": 0.5564, "step": 18806 }, { "epoch": 0.9513379533613233, "grad_norm": 9.661872863769531, "learning_rate": 5.949776618342062e-08, "loss": 0.4945, "step": 18807 }, { "epoch": 0.9513885376093885, "grad_norm": 7.132061958312988, "learning_rate": 5.937437976374805e-08, "loss": 0.3738, "step": 18808 }, { "epoch": 0.9514391218574536, "grad_norm": 9.777615547180176, "learning_rate": 5.925112065206518e-08, "loss": 0.31, "step": 18809 }, { "epoch": 0.9514897061055188, "grad_norm": 6.291184902191162, "learning_rate": 5.912798885154614e-08, "loss": 0.375, "step": 18810 }, { "epoch": 0.9515402903535839, "grad_norm": 11.001951217651367, "learning_rate": 5.900498436536506e-08, "loss": 0.3523, "step": 18811 }, { "epoch": 0.951590874601649, "grad_norm": 9.306901931762695, "learning_rate": 5.888210719669107e-08, "loss": 0.4617, "step": 18812 }, { "epoch": 0.9516414588497142, "grad_norm": 10.432278633117676, "learning_rate": 5.875935734868998e-08, "loss": 0.3424, "step": 18813 }, { "epoch": 0.9516920430977793, "grad_norm": 5.730647563934326, "learning_rate": 5.863673482452537e-08, "loss": 0.4101, "step": 18814 }, { "epoch": 0.9517426273458445, "grad_norm": 9.475268363952637, "learning_rate": 5.8514239627356915e-08, "loss": 0.356, "step": 18815 }, { "epoch": 0.9517932115939096, "grad_norm": 5.4636149406433105, "learning_rate": 5.839187176034045e-08, "loss": 0.3922, "step": 18816 }, { "epoch": 0.9518437958419748, "grad_norm": 6.353960990905762, "learning_rate": 5.8269631226630094e-08, "loss": 0.4164, "step": 18817 }, { "epoch": 0.95189438009004, "grad_norm": 6.978785514831543, "learning_rate": 5.814751802937446e-08, "loss": 0.3916, "step": 18818 }, { "epoch": 0.9519449643381052, "grad_norm": 7.223616600036621, "learning_rate": 5.8025532171721024e-08, "loss": 0.4389, "step": 18819 }, { "epoch": 0.9519955485861703, "grad_norm": 11.028972625732422, "learning_rate": 5.790367365681282e-08, "loss": 0.5045, "step": 18820 }, { "epoch": 0.9520461328342354, "grad_norm": 7.28682804107666, "learning_rate": 5.778194248778957e-08, "loss": 0.4595, "step": 18821 }, { "epoch": 0.9520967170823006, "grad_norm": 4.530180931091309, "learning_rate": 5.766033866778875e-08, "loss": 0.4041, "step": 18822 }, { "epoch": 0.9521473013303657, "grad_norm": 7.074329853057861, "learning_rate": 5.753886219994231e-08, "loss": 0.3818, "step": 18823 }, { "epoch": 0.9521978855784309, "grad_norm": 8.936803817749023, "learning_rate": 5.7417513087382194e-08, "loss": 0.4126, "step": 18824 }, { "epoch": 0.952248469826496, "grad_norm": 6.133699417114258, "learning_rate": 5.729629133323422e-08, "loss": 0.3111, "step": 18825 }, { "epoch": 0.9522990540745612, "grad_norm": 5.262271881103516, "learning_rate": 5.717519694062201e-08, "loss": 0.3783, "step": 18826 }, { "epoch": 0.9523496383226263, "grad_norm": 9.759604454040527, "learning_rate": 5.7054229912666406e-08, "loss": 0.3513, "step": 18827 }, { "epoch": 0.9524002225706915, "grad_norm": 4.8395562171936035, "learning_rate": 5.693339025248379e-08, "loss": 0.4071, "step": 18828 }, { "epoch": 0.9524508068187566, "grad_norm": 11.878873825073242, "learning_rate": 5.681267796318834e-08, "loss": 0.5087, "step": 18829 }, { "epoch": 0.9525013910668217, "grad_norm": 10.313671112060547, "learning_rate": 5.6692093047890914e-08, "loss": 0.3811, "step": 18830 }, { "epoch": 0.9525519753148869, "grad_norm": 5.05621337890625, "learning_rate": 5.6571635509697356e-08, "loss": 0.2856, "step": 18831 }, { "epoch": 0.9526025595629521, "grad_norm": 9.009151458740234, "learning_rate": 5.645130535171295e-08, "loss": 0.5045, "step": 18832 }, { "epoch": 0.9526531438110173, "grad_norm": 6.163105487823486, "learning_rate": 5.6331102577037446e-08, "loss": 0.2507, "step": 18833 }, { "epoch": 0.9527037280590824, "grad_norm": 9.341010093688965, "learning_rate": 5.6211027188768365e-08, "loss": 0.4824, "step": 18834 }, { "epoch": 0.9527543123071476, "grad_norm": 8.561776161193848, "learning_rate": 5.60910791899999e-08, "loss": 0.3691, "step": 18835 }, { "epoch": 0.9528048965552127, "grad_norm": 4.538204193115234, "learning_rate": 5.597125858382346e-08, "loss": 0.2848, "step": 18836 }, { "epoch": 0.9528554808032779, "grad_norm": 8.46016788482666, "learning_rate": 5.5851565373325477e-08, "loss": 0.4085, "step": 18837 }, { "epoch": 0.952906065051343, "grad_norm": 11.967281341552734, "learning_rate": 5.5731999561590145e-08, "loss": 0.4874, "step": 18838 }, { "epoch": 0.9529566492994082, "grad_norm": 10.53048324584961, "learning_rate": 5.561256115169944e-08, "loss": 0.4364, "step": 18839 }, { "epoch": 0.9530072335474733, "grad_norm": 10.422163009643555, "learning_rate": 5.5493250146729797e-08, "loss": 0.5449, "step": 18840 }, { "epoch": 0.9530578177955384, "grad_norm": 5.062925338745117, "learning_rate": 5.537406654975652e-08, "loss": 0.4262, "step": 18841 }, { "epoch": 0.9531084020436036, "grad_norm": 5.139739990234375, "learning_rate": 5.525501036384995e-08, "loss": 0.3551, "step": 18842 }, { "epoch": 0.9531589862916687, "grad_norm": 6.41104793548584, "learning_rate": 5.5136081592078175e-08, "loss": 0.3829, "step": 18843 }, { "epoch": 0.9532095705397339, "grad_norm": 12.046306610107422, "learning_rate": 5.501728023750652e-08, "loss": 0.5102, "step": 18844 }, { "epoch": 0.9532601547877991, "grad_norm": 8.211812019348145, "learning_rate": 5.489860630319477e-08, "loss": 0.4128, "step": 18845 }, { "epoch": 0.9533107390358643, "grad_norm": 6.918422222137451, "learning_rate": 5.478005979220158e-08, "loss": 0.3928, "step": 18846 }, { "epoch": 0.9533613232839294, "grad_norm": 7.964101791381836, "learning_rate": 5.466164070758173e-08, "loss": 0.3263, "step": 18847 }, { "epoch": 0.9534119075319946, "grad_norm": 7.928385257720947, "learning_rate": 5.454334905238612e-08, "loss": 0.4036, "step": 18848 }, { "epoch": 0.9534624917800597, "grad_norm": 5.572155952453613, "learning_rate": 5.442518482966341e-08, "loss": 0.4095, "step": 18849 }, { "epoch": 0.9535130760281249, "grad_norm": 8.032866477966309, "learning_rate": 5.43071480424584e-08, "loss": 0.3401, "step": 18850 }, { "epoch": 0.95356366027619, "grad_norm": 7.750641822814941, "learning_rate": 5.4189238693811986e-08, "loss": 0.4724, "step": 18851 }, { "epoch": 0.9536142445242551, "grad_norm": 7.2299957275390625, "learning_rate": 5.407145678676284e-08, "loss": 0.3597, "step": 18852 }, { "epoch": 0.9536648287723203, "grad_norm": 11.855775833129883, "learning_rate": 5.3953802324345774e-08, "loss": 0.5462, "step": 18853 }, { "epoch": 0.9537154130203854, "grad_norm": 5.272653102874756, "learning_rate": 5.3836275309592234e-08, "loss": 0.3532, "step": 18854 }, { "epoch": 0.9537659972684506, "grad_norm": 6.100613594055176, "learning_rate": 5.3718875745531476e-08, "loss": 0.2912, "step": 18855 }, { "epoch": 0.9538165815165157, "grad_norm": 5.4122490882873535, "learning_rate": 5.3601603635188295e-08, "loss": 0.2703, "step": 18856 }, { "epoch": 0.9538671657645809, "grad_norm": 5.574666976928711, "learning_rate": 5.348445898158306e-08, "loss": 0.4233, "step": 18857 }, { "epoch": 0.9539177500126461, "grad_norm": 6.4106221199035645, "learning_rate": 5.336744178773667e-08, "loss": 0.3719, "step": 18858 }, { "epoch": 0.9539683342607113, "grad_norm": 5.8703718185424805, "learning_rate": 5.325055205666285e-08, "loss": 0.3215, "step": 18859 }, { "epoch": 0.9540189185087764, "grad_norm": 10.757436752319336, "learning_rate": 5.3133789791373625e-08, "loss": 0.4708, "step": 18860 }, { "epoch": 0.9540695027568415, "grad_norm": 10.032953262329102, "learning_rate": 5.301715499487825e-08, "loss": 0.3283, "step": 18861 }, { "epoch": 0.9541200870049067, "grad_norm": 6.226153373718262, "learning_rate": 5.2900647670182105e-08, "loss": 0.353, "step": 18862 }, { "epoch": 0.9541706712529718, "grad_norm": 10.248757362365723, "learning_rate": 5.278426782028667e-08, "loss": 0.3375, "step": 18863 }, { "epoch": 0.954221255501037, "grad_norm": 6.168508052825928, "learning_rate": 5.2668015448191225e-08, "loss": 0.3346, "step": 18864 }, { "epoch": 0.9542718397491021, "grad_norm": 7.449167251586914, "learning_rate": 5.255189055689114e-08, "loss": 0.4389, "step": 18865 }, { "epoch": 0.9543224239971673, "grad_norm": 8.346453666687012, "learning_rate": 5.243589314937847e-08, "loss": 0.4047, "step": 18866 }, { "epoch": 0.9543730082452324, "grad_norm": 12.848648071289062, "learning_rate": 5.232002322864304e-08, "loss": 0.4985, "step": 18867 }, { "epoch": 0.9544235924932976, "grad_norm": 5.590774059295654, "learning_rate": 5.2204280797669705e-08, "loss": 0.3373, "step": 18868 }, { "epoch": 0.9544741767413627, "grad_norm": 11.223382949829102, "learning_rate": 5.2088665859441053e-08, "loss": 0.341, "step": 18869 }, { "epoch": 0.9545247609894278, "grad_norm": 7.280882835388184, "learning_rate": 5.197317841693694e-08, "loss": 0.3155, "step": 18870 }, { "epoch": 0.954575345237493, "grad_norm": 7.160662651062012, "learning_rate": 5.185781847313165e-08, "loss": 0.4215, "step": 18871 }, { "epoch": 0.9546259294855582, "grad_norm": 7.849632740020752, "learning_rate": 5.174258603099891e-08, "loss": 0.3075, "step": 18872 }, { "epoch": 0.9546765137336234, "grad_norm": 9.024943351745605, "learning_rate": 5.162748109350801e-08, "loss": 0.3642, "step": 18873 }, { "epoch": 0.9547270979816885, "grad_norm": 7.714616775512695, "learning_rate": 5.151250366362437e-08, "loss": 0.3615, "step": 18874 }, { "epoch": 0.9547776822297537, "grad_norm": 7.236831188201904, "learning_rate": 5.139765374431116e-08, "loss": 0.3353, "step": 18875 }, { "epoch": 0.9548282664778188, "grad_norm": 9.134896278381348, "learning_rate": 5.128293133852769e-08, "loss": 0.4788, "step": 18876 }, { "epoch": 0.954878850725884, "grad_norm": 4.986384391784668, "learning_rate": 5.116833644922992e-08, "loss": 0.3021, "step": 18877 }, { "epoch": 0.9549294349739491, "grad_norm": 8.298068046569824, "learning_rate": 5.1053869079369935e-08, "loss": 0.3705, "step": 18878 }, { "epoch": 0.9549800192220143, "grad_norm": 10.376185417175293, "learning_rate": 5.093952923189871e-08, "loss": 0.3324, "step": 18879 }, { "epoch": 0.9550306034700794, "grad_norm": 5.978463649749756, "learning_rate": 5.082531690976167e-08, "loss": 0.3166, "step": 18880 }, { "epoch": 0.9550811877181445, "grad_norm": 6.045015811920166, "learning_rate": 5.071123211590201e-08, "loss": 0.3239, "step": 18881 }, { "epoch": 0.9551317719662097, "grad_norm": 10.046049118041992, "learning_rate": 5.059727485325905e-08, "loss": 0.3015, "step": 18882 }, { "epoch": 0.9551823562142748, "grad_norm": 5.936420440673828, "learning_rate": 5.048344512476988e-08, "loss": 0.4071, "step": 18883 }, { "epoch": 0.95523294046234, "grad_norm": 9.383691787719727, "learning_rate": 5.036974293336716e-08, "loss": 0.4224, "step": 18884 }, { "epoch": 0.9552835247104052, "grad_norm": 6.500050067901611, "learning_rate": 5.025616828198132e-08, "loss": 0.3374, "step": 18885 }, { "epoch": 0.9553341089584704, "grad_norm": 9.09225082397461, "learning_rate": 5.01427211735378e-08, "loss": 0.3889, "step": 18886 }, { "epoch": 0.9553846932065355, "grad_norm": 18.525625228881836, "learning_rate": 5.002940161096037e-08, "loss": 0.4367, "step": 18887 }, { "epoch": 0.9554352774546007, "grad_norm": 11.389047622680664, "learning_rate": 4.9916209597170026e-08, "loss": 0.3922, "step": 18888 }, { "epoch": 0.9554858617026658, "grad_norm": 6.079934597015381, "learning_rate": 4.980314513508167e-08, "loss": 0.3745, "step": 18889 }, { "epoch": 0.955536445950731, "grad_norm": 5.558909893035889, "learning_rate": 4.9690208227609615e-08, "loss": 0.311, "step": 18890 }, { "epoch": 0.9555870301987961, "grad_norm": 7.147679805755615, "learning_rate": 4.957739887766433e-08, "loss": 0.3888, "step": 18891 }, { "epoch": 0.9556376144468612, "grad_norm": 11.993587493896484, "learning_rate": 4.946471708815126e-08, "loss": 0.3719, "step": 18892 }, { "epoch": 0.9556881986949264, "grad_norm": 9.468420028686523, "learning_rate": 4.935216286197586e-08, "loss": 0.4532, "step": 18893 }, { "epoch": 0.9557387829429915, "grad_norm": 9.308518409729004, "learning_rate": 4.923973620203748e-08, "loss": 0.511, "step": 18894 }, { "epoch": 0.9557893671910567, "grad_norm": 5.121946811676025, "learning_rate": 4.912743711123269e-08, "loss": 0.3324, "step": 18895 }, { "epoch": 0.9558399514391218, "grad_norm": 13.071130752563477, "learning_rate": 4.9015265592455283e-08, "loss": 0.503, "step": 18896 }, { "epoch": 0.955890535687187, "grad_norm": 8.129947662353516, "learning_rate": 4.890322164859573e-08, "loss": 0.3736, "step": 18897 }, { "epoch": 0.9559411199352522, "grad_norm": 6.80755090713501, "learning_rate": 4.879130528254172e-08, "loss": 0.3407, "step": 18898 }, { "epoch": 0.9559917041833174, "grad_norm": 15.507399559020996, "learning_rate": 4.8679516497176505e-08, "loss": 0.4005, "step": 18899 }, { "epoch": 0.9560422884313825, "grad_norm": 6.361189365386963, "learning_rate": 4.856785529538055e-08, "loss": 0.3024, "step": 18900 }, { "epoch": 0.9560928726794476, "grad_norm": 6.581584453582764, "learning_rate": 4.845632168003156e-08, "loss": 0.3665, "step": 18901 }, { "epoch": 0.9561434569275128, "grad_norm": 8.675909042358398, "learning_rate": 4.8344915654003347e-08, "loss": 0.3669, "step": 18902 }, { "epoch": 0.9561940411755779, "grad_norm": 7.719410419464111, "learning_rate": 4.823363722016638e-08, "loss": 0.35, "step": 18903 }, { "epoch": 0.9562446254236431, "grad_norm": 8.355485916137695, "learning_rate": 4.8122486381387836e-08, "loss": 0.2793, "step": 18904 }, { "epoch": 0.9562952096717082, "grad_norm": 7.356461524963379, "learning_rate": 4.801146314053207e-08, "loss": 0.3811, "step": 18905 }, { "epoch": 0.9563457939197734, "grad_norm": 10.100090026855469, "learning_rate": 4.7900567500459574e-08, "loss": 0.3103, "step": 18906 }, { "epoch": 0.9563963781678385, "grad_norm": 8.210988998413086, "learning_rate": 4.778979946402862e-08, "loss": 0.4442, "step": 18907 }, { "epoch": 0.9564469624159037, "grad_norm": 8.995534896850586, "learning_rate": 4.767915903409248e-08, "loss": 0.4022, "step": 18908 }, { "epoch": 0.9564975466639688, "grad_norm": 12.30401611328125, "learning_rate": 4.7568646213503324e-08, "loss": 0.469, "step": 18909 }, { "epoch": 0.956548130912034, "grad_norm": 11.830490112304688, "learning_rate": 4.74582610051072e-08, "loss": 0.4314, "step": 18910 }, { "epoch": 0.9565987151600992, "grad_norm": 8.090099334716797, "learning_rate": 4.734800341174961e-08, "loss": 0.4429, "step": 18911 }, { "epoch": 0.9566492994081643, "grad_norm": 20.625408172607422, "learning_rate": 4.723787343627162e-08, "loss": 0.4743, "step": 18912 }, { "epoch": 0.9566998836562295, "grad_norm": 8.813874244689941, "learning_rate": 4.71278710815104e-08, "loss": 0.3906, "step": 18913 }, { "epoch": 0.9567504679042946, "grad_norm": 9.474687576293945, "learning_rate": 4.7017996350301464e-08, "loss": 0.477, "step": 18914 }, { "epoch": 0.9568010521523598, "grad_norm": 8.44813346862793, "learning_rate": 4.690824924547477e-08, "loss": 0.4478, "step": 18915 }, { "epoch": 0.9568516364004249, "grad_norm": 6.507174015045166, "learning_rate": 4.67986297698586e-08, "loss": 0.383, "step": 18916 }, { "epoch": 0.9569022206484901, "grad_norm": 6.401829719543457, "learning_rate": 4.668913792627794e-08, "loss": 0.3463, "step": 18917 }, { "epoch": 0.9569528048965552, "grad_norm": 7.864218235015869, "learning_rate": 4.65797737175544e-08, "loss": 0.39, "step": 18918 }, { "epoch": 0.9570033891446204, "grad_norm": 5.960483551025391, "learning_rate": 4.6470537146504626e-08, "loss": 0.3259, "step": 18919 }, { "epoch": 0.9570539733926855, "grad_norm": 11.099241256713867, "learning_rate": 4.6361428215945245e-08, "loss": 0.4319, "step": 18920 }, { "epoch": 0.9571045576407506, "grad_norm": 7.004403591156006, "learning_rate": 4.625244692868625e-08, "loss": 0.385, "step": 18921 }, { "epoch": 0.9571551418888158, "grad_norm": 4.8625640869140625, "learning_rate": 4.614359328753648e-08, "loss": 0.3909, "step": 18922 }, { "epoch": 0.9572057261368809, "grad_norm": 23.12616539001465, "learning_rate": 4.603486729530093e-08, "loss": 0.4142, "step": 18923 }, { "epoch": 0.9572563103849461, "grad_norm": 13.386950492858887, "learning_rate": 4.592626895478125e-08, "loss": 0.359, "step": 18924 }, { "epoch": 0.9573068946330113, "grad_norm": 7.879252910614014, "learning_rate": 4.5817798268775196e-08, "loss": 0.3962, "step": 18925 }, { "epoch": 0.9573574788810765, "grad_norm": 5.313308238983154, "learning_rate": 4.5709455240078306e-08, "loss": 0.3474, "step": 18926 }, { "epoch": 0.9574080631291416, "grad_norm": 7.5642499923706055, "learning_rate": 4.560123987148224e-08, "loss": 0.4073, "step": 18927 }, { "epoch": 0.9574586473772068, "grad_norm": 11.613131523132324, "learning_rate": 4.5493152165775324e-08, "loss": 0.4089, "step": 18928 }, { "epoch": 0.9575092316252719, "grad_norm": 7.282976150512695, "learning_rate": 4.538519212574311e-08, "loss": 0.3505, "step": 18929 }, { "epoch": 0.957559815873337, "grad_norm": 13.196073532104492, "learning_rate": 4.527735975416614e-08, "loss": 0.5595, "step": 18930 }, { "epoch": 0.9576104001214022, "grad_norm": 11.288016319274902, "learning_rate": 4.5169655053824423e-08, "loss": 0.4213, "step": 18931 }, { "epoch": 0.9576609843694673, "grad_norm": 5.881819248199463, "learning_rate": 4.5062078027492964e-08, "loss": 0.3444, "step": 18932 }, { "epoch": 0.9577115686175325, "grad_norm": 5.771313190460205, "learning_rate": 4.495462867794398e-08, "loss": 0.3765, "step": 18933 }, { "epoch": 0.9577621528655976, "grad_norm": 7.928450107574463, "learning_rate": 4.484730700794526e-08, "loss": 0.386, "step": 18934 }, { "epoch": 0.9578127371136628, "grad_norm": 8.048818588256836, "learning_rate": 4.474011302026238e-08, "loss": 0.4456, "step": 18935 }, { "epoch": 0.9578633213617279, "grad_norm": 9.701778411865234, "learning_rate": 4.4633046717658665e-08, "loss": 0.5324, "step": 18936 }, { "epoch": 0.9579139056097931, "grad_norm": 6.889627933502197, "learning_rate": 4.4526108102891374e-08, "loss": 0.3992, "step": 18937 }, { "epoch": 0.9579644898578583, "grad_norm": 21.934507369995117, "learning_rate": 4.441929717871773e-08, "loss": 0.511, "step": 18938 }, { "epoch": 0.9580150741059235, "grad_norm": 8.655407905578613, "learning_rate": 4.4312613947888327e-08, "loss": 0.427, "step": 18939 }, { "epoch": 0.9580656583539886, "grad_norm": 6.456470489501953, "learning_rate": 4.4206058413153175e-08, "loss": 0.3859, "step": 18940 }, { "epoch": 0.9581162426020537, "grad_norm": 6.596839427947998, "learning_rate": 4.4099630577257856e-08, "loss": 0.3497, "step": 18941 }, { "epoch": 0.9581668268501189, "grad_norm": 6.631457805633545, "learning_rate": 4.399333044294407e-08, "loss": 0.3491, "step": 18942 }, { "epoch": 0.958217411098184, "grad_norm": 10.87261962890625, "learning_rate": 4.388715801295185e-08, "loss": 0.375, "step": 18943 }, { "epoch": 0.9582679953462492, "grad_norm": 8.494499206542969, "learning_rate": 4.378111329001622e-08, "loss": 0.3906, "step": 18944 }, { "epoch": 0.9583185795943143, "grad_norm": 6.9454450607299805, "learning_rate": 4.367519627687e-08, "loss": 0.4445, "step": 18945 }, { "epoch": 0.9583691638423795, "grad_norm": 8.339387893676758, "learning_rate": 4.3569406976242677e-08, "loss": 0.3779, "step": 18946 }, { "epoch": 0.9584197480904446, "grad_norm": 10.249438285827637, "learning_rate": 4.34637453908604e-08, "loss": 0.4738, "step": 18947 }, { "epoch": 0.9584703323385098, "grad_norm": 7.5067057609558105, "learning_rate": 4.335821152344488e-08, "loss": 0.4268, "step": 18948 }, { "epoch": 0.9585209165865749, "grad_norm": 15.360703468322754, "learning_rate": 4.325280537671617e-08, "loss": 0.4414, "step": 18949 }, { "epoch": 0.95857150083464, "grad_norm": 12.7007417678833, "learning_rate": 4.314752695338986e-08, "loss": 0.4264, "step": 18950 }, { "epoch": 0.9586220850827053, "grad_norm": 8.765022277832031, "learning_rate": 4.30423762561788e-08, "loss": 0.3211, "step": 18951 }, { "epoch": 0.9586726693307704, "grad_norm": 6.720461845397949, "learning_rate": 4.2937353287793585e-08, "loss": 0.3433, "step": 18952 }, { "epoch": 0.9587232535788356, "grad_norm": 5.009425163269043, "learning_rate": 4.283245805093872e-08, "loss": 0.2823, "step": 18953 }, { "epoch": 0.9587738378269007, "grad_norm": 8.368459701538086, "learning_rate": 4.272769054831816e-08, "loss": 0.396, "step": 18954 }, { "epoch": 0.9588244220749659, "grad_norm": 5.552108287811279, "learning_rate": 4.262305078263085e-08, "loss": 0.4626, "step": 18955 }, { "epoch": 0.958875006323031, "grad_norm": 8.07498836517334, "learning_rate": 4.251853875657408e-08, "loss": 0.36, "step": 18956 }, { "epoch": 0.9589255905710962, "grad_norm": 11.083707809448242, "learning_rate": 4.241415447283959e-08, "loss": 0.5023, "step": 18957 }, { "epoch": 0.9589761748191613, "grad_norm": 7.139494895935059, "learning_rate": 4.2309897934117996e-08, "loss": 0.3835, "step": 18958 }, { "epoch": 0.9590267590672265, "grad_norm": 9.011066436767578, "learning_rate": 4.220576914309604e-08, "loss": 0.3642, "step": 18959 }, { "epoch": 0.9590773433152916, "grad_norm": 7.862205982208252, "learning_rate": 4.210176810245548e-08, "loss": 0.4205, "step": 18960 }, { "epoch": 0.9591279275633567, "grad_norm": 6.182086944580078, "learning_rate": 4.199789481487748e-08, "loss": 0.3161, "step": 18961 }, { "epoch": 0.9591785118114219, "grad_norm": 8.307351112365723, "learning_rate": 4.189414928303825e-08, "loss": 0.4087, "step": 18962 }, { "epoch": 0.959229096059487, "grad_norm": 6.987990856170654, "learning_rate": 4.1790531509610656e-08, "loss": 0.2932, "step": 18963 }, { "epoch": 0.9592796803075522, "grad_norm": 6.054808616638184, "learning_rate": 4.168704149726532e-08, "loss": 0.2802, "step": 18964 }, { "epoch": 0.9593302645556174, "grad_norm": 5.5672926902771, "learning_rate": 4.158367924866902e-08, "loss": 0.363, "step": 18965 }, { "epoch": 0.9593808488036826, "grad_norm": 12.710614204406738, "learning_rate": 4.148044476648405e-08, "loss": 0.5897, "step": 18966 }, { "epoch": 0.9594314330517477, "grad_norm": 7.24582576751709, "learning_rate": 4.137733805337163e-08, "loss": 0.3567, "step": 18967 }, { "epoch": 0.9594820172998129, "grad_norm": 9.641030311584473, "learning_rate": 4.127435911198796e-08, "loss": 0.4364, "step": 18968 }, { "epoch": 0.959532601547878, "grad_norm": 6.033563613891602, "learning_rate": 4.1171507944986476e-08, "loss": 0.4118, "step": 18969 }, { "epoch": 0.9595831857959431, "grad_norm": 9.42166805267334, "learning_rate": 4.1068784555017835e-08, "loss": 0.4141, "step": 18970 }, { "epoch": 0.9596337700440083, "grad_norm": 8.582991600036621, "learning_rate": 4.096618894472937e-08, "loss": 0.3702, "step": 18971 }, { "epoch": 0.9596843542920734, "grad_norm": 8.290804862976074, "learning_rate": 4.08637211167634e-08, "loss": 0.2321, "step": 18972 }, { "epoch": 0.9597349385401386, "grad_norm": 5.958768844604492, "learning_rate": 4.0761381073761154e-08, "loss": 0.3988, "step": 18973 }, { "epoch": 0.9597855227882037, "grad_norm": 14.292464256286621, "learning_rate": 4.065916881835996e-08, "loss": 0.4908, "step": 18974 }, { "epoch": 0.9598361070362689, "grad_norm": 6.007587432861328, "learning_rate": 4.055708435319272e-08, "loss": 0.3226, "step": 18975 }, { "epoch": 0.959886691284334, "grad_norm": 5.7870378494262695, "learning_rate": 4.045512768089066e-08, "loss": 0.2787, "step": 18976 }, { "epoch": 0.9599372755323992, "grad_norm": 5.725113391876221, "learning_rate": 4.035329880408057e-08, "loss": 0.4073, "step": 18977 }, { "epoch": 0.9599878597804644, "grad_norm": 6.052412986755371, "learning_rate": 4.0251597725387e-08, "loss": 0.3505, "step": 18978 }, { "epoch": 0.9600384440285296, "grad_norm": 4.947795867919922, "learning_rate": 4.015002444742955e-08, "loss": 0.3324, "step": 18979 }, { "epoch": 0.9600890282765947, "grad_norm": 12.04092025756836, "learning_rate": 4.004857897282555e-08, "loss": 0.5007, "step": 18980 }, { "epoch": 0.9601396125246598, "grad_norm": 12.148441314697266, "learning_rate": 3.9947261304189576e-08, "loss": 0.4786, "step": 18981 }, { "epoch": 0.960190196772725, "grad_norm": 7.50092077255249, "learning_rate": 3.984607144413233e-08, "loss": 0.3605, "step": 18982 }, { "epoch": 0.9602407810207901, "grad_norm": 6.619065284729004, "learning_rate": 3.97450093952606e-08, "loss": 0.3598, "step": 18983 }, { "epoch": 0.9602913652688553, "grad_norm": 7.420776844024658, "learning_rate": 3.9644075160179543e-08, "loss": 0.3555, "step": 18984 }, { "epoch": 0.9603419495169204, "grad_norm": 5.379364967346191, "learning_rate": 3.9543268741489285e-08, "loss": 0.3976, "step": 18985 }, { "epoch": 0.9603925337649856, "grad_norm": 7.167992115020752, "learning_rate": 3.944259014178775e-08, "loss": 0.4342, "step": 18986 }, { "epoch": 0.9604431180130507, "grad_norm": 7.891541481018066, "learning_rate": 3.9342039363668426e-08, "loss": 0.3972, "step": 18987 }, { "epoch": 0.9604937022611159, "grad_norm": 6.29986572265625, "learning_rate": 3.924161640972313e-08, "loss": 0.3335, "step": 18988 }, { "epoch": 0.960544286509181, "grad_norm": 5.1507344245910645, "learning_rate": 3.914132128253867e-08, "loss": 0.2742, "step": 18989 }, { "epoch": 0.9605948707572461, "grad_norm": 5.052850246429443, "learning_rate": 3.9041153984700767e-08, "loss": 0.3247, "step": 18990 }, { "epoch": 0.9606454550053114, "grad_norm": 9.641561508178711, "learning_rate": 3.8941114518789013e-08, "loss": 0.3944, "step": 18991 }, { "epoch": 0.9606960392533765, "grad_norm": 4.364185810089111, "learning_rate": 3.884120288738191e-08, "loss": 0.3483, "step": 18992 }, { "epoch": 0.9607466235014417, "grad_norm": 6.31867790222168, "learning_rate": 3.874141909305407e-08, "loss": 0.4269, "step": 18993 }, { "epoch": 0.9607972077495068, "grad_norm": 9.452144622802734, "learning_rate": 3.8641763138376196e-08, "loss": 0.4014, "step": 18994 }, { "epoch": 0.960847791997572, "grad_norm": 8.377400398254395, "learning_rate": 3.854223502591625e-08, "loss": 0.4255, "step": 18995 }, { "epoch": 0.9608983762456371, "grad_norm": 20.42564582824707, "learning_rate": 3.844283475823884e-08, "loss": 0.5321, "step": 18996 }, { "epoch": 0.9609489604937023, "grad_norm": 13.989636421203613, "learning_rate": 3.8343562337905816e-08, "loss": 0.4395, "step": 18997 }, { "epoch": 0.9609995447417674, "grad_norm": 13.012517929077148, "learning_rate": 3.824441776747456e-08, "loss": 0.4872, "step": 18998 }, { "epoch": 0.9610501289898326, "grad_norm": 7.935750484466553, "learning_rate": 3.814540104950026e-08, "loss": 0.4165, "step": 18999 }, { "epoch": 0.9611007132378977, "grad_norm": 8.204289436340332, "learning_rate": 3.804651218653421e-08, "loss": 0.3563, "step": 19000 }, { "epoch": 0.9611512974859628, "grad_norm": 6.274234294891357, "learning_rate": 3.7947751181123815e-08, "loss": 0.4236, "step": 19001 }, { "epoch": 0.961201881734028, "grad_norm": 8.07421588897705, "learning_rate": 3.784911803581537e-08, "loss": 0.3399, "step": 19002 }, { "epoch": 0.9612524659820931, "grad_norm": 7.325018882751465, "learning_rate": 3.775061275314906e-08, "loss": 0.386, "step": 19003 }, { "epoch": 0.9613030502301583, "grad_norm": 7.860820293426514, "learning_rate": 3.765223533566398e-08, "loss": 0.3225, "step": 19004 }, { "epoch": 0.9613536344782235, "grad_norm": 15.270920753479004, "learning_rate": 3.75539857858942e-08, "loss": 0.423, "step": 19005 }, { "epoch": 0.9614042187262887, "grad_norm": 6.459991455078125, "learning_rate": 3.74558641063727e-08, "loss": 0.321, "step": 19006 }, { "epoch": 0.9614548029743538, "grad_norm": 10.40376091003418, "learning_rate": 3.735787029962634e-08, "loss": 0.2893, "step": 19007 }, { "epoch": 0.961505387222419, "grad_norm": 6.428775310516357, "learning_rate": 3.726000436818089e-08, "loss": 0.411, "step": 19008 }, { "epoch": 0.9615559714704841, "grad_norm": 5.779244899749756, "learning_rate": 3.7162266314558216e-08, "loss": 0.3877, "step": 19009 }, { "epoch": 0.9616065557185492, "grad_norm": 5.377289772033691, "learning_rate": 3.7064656141276854e-08, "loss": 0.3461, "step": 19010 }, { "epoch": 0.9616571399666144, "grad_norm": 5.913250923156738, "learning_rate": 3.696717385085202e-08, "loss": 0.2795, "step": 19011 }, { "epoch": 0.9617077242146795, "grad_norm": 11.516925811767578, "learning_rate": 3.686981944579504e-08, "loss": 0.3657, "step": 19012 }, { "epoch": 0.9617583084627447, "grad_norm": 20.350744247436523, "learning_rate": 3.6772592928615015e-08, "loss": 0.6118, "step": 19013 }, { "epoch": 0.9618088927108098, "grad_norm": 14.050886154174805, "learning_rate": 3.667549430181661e-08, "loss": 0.5783, "step": 19014 }, { "epoch": 0.961859476958875, "grad_norm": 8.962928771972656, "learning_rate": 3.657852356790281e-08, "loss": 0.291, "step": 19015 }, { "epoch": 0.9619100612069401, "grad_norm": 5.837489604949951, "learning_rate": 3.648168072937164e-08, "loss": 0.3066, "step": 19016 }, { "epoch": 0.9619606454550053, "grad_norm": 9.412426948547363, "learning_rate": 3.638496578871942e-08, "loss": 0.4005, "step": 19017 }, { "epoch": 0.9620112297030705, "grad_norm": 6.282042026519775, "learning_rate": 3.628837874843638e-08, "loss": 0.3614, "step": 19018 }, { "epoch": 0.9620618139511357, "grad_norm": 7.983880996704102, "learning_rate": 3.619191961101331e-08, "loss": 0.4446, "step": 19019 }, { "epoch": 0.9621123981992008, "grad_norm": 11.580306053161621, "learning_rate": 3.6095588378934346e-08, "loss": 0.4605, "step": 19020 }, { "epoch": 0.962162982447266, "grad_norm": 11.541092872619629, "learning_rate": 3.599938505468248e-08, "loss": 0.5435, "step": 19021 }, { "epoch": 0.9622135666953311, "grad_norm": 11.661956787109375, "learning_rate": 3.5903309640736296e-08, "loss": 0.4574, "step": 19022 }, { "epoch": 0.9622641509433962, "grad_norm": 8.176274299621582, "learning_rate": 3.580736213957159e-08, "loss": 0.3083, "step": 19023 }, { "epoch": 0.9623147351914614, "grad_norm": 7.739297389984131, "learning_rate": 3.571154255366027e-08, "loss": 0.4259, "step": 19024 }, { "epoch": 0.9623653194395265, "grad_norm": 6.350265026092529, "learning_rate": 3.5615850885472035e-08, "loss": 0.3402, "step": 19025 }, { "epoch": 0.9624159036875917, "grad_norm": 8.183656692504883, "learning_rate": 3.5520287137472685e-08, "loss": 0.3096, "step": 19026 }, { "epoch": 0.9624664879356568, "grad_norm": 6.889374732971191, "learning_rate": 3.542485131212414e-08, "loss": 0.4073, "step": 19027 }, { "epoch": 0.962517072183722, "grad_norm": 7.287812232971191, "learning_rate": 3.532954341188555e-08, "loss": 0.3869, "step": 19028 }, { "epoch": 0.9625676564317871, "grad_norm": 10.027825355529785, "learning_rate": 3.523436343921327e-08, "loss": 0.3662, "step": 19029 }, { "epoch": 0.9626182406798522, "grad_norm": 5.399219512939453, "learning_rate": 3.513931139655924e-08, "loss": 0.3797, "step": 19030 }, { "epoch": 0.9626688249279175, "grad_norm": 10.911816596984863, "learning_rate": 3.504438728637316e-08, "loss": 0.4319, "step": 19031 }, { "epoch": 0.9627194091759826, "grad_norm": 9.295711517333984, "learning_rate": 3.4949591111100855e-08, "loss": 0.3693, "step": 19032 }, { "epoch": 0.9627699934240478, "grad_norm": 8.0272855758667, "learning_rate": 3.4854922873184817e-08, "loss": 0.3907, "step": 19033 }, { "epoch": 0.9628205776721129, "grad_norm": 6.997623920440674, "learning_rate": 3.4760382575064754e-08, "loss": 0.3531, "step": 19034 }, { "epoch": 0.9628711619201781, "grad_norm": 7.646657943725586, "learning_rate": 3.466597021917706e-08, "loss": 0.3092, "step": 19035 }, { "epoch": 0.9629217461682432, "grad_norm": 16.383630752563477, "learning_rate": 3.457168580795311e-08, "loss": 0.4376, "step": 19036 }, { "epoch": 0.9629723304163084, "grad_norm": 4.836324214935303, "learning_rate": 3.4477529343823754e-08, "loss": 0.3956, "step": 19037 }, { "epoch": 0.9630229146643735, "grad_norm": 7.363566875457764, "learning_rate": 3.438350082921482e-08, "loss": 0.332, "step": 19038 }, { "epoch": 0.9630734989124387, "grad_norm": 6.572153091430664, "learning_rate": 3.428960026654882e-08, "loss": 0.4356, "step": 19039 }, { "epoch": 0.9631240831605038, "grad_norm": 7.125833034515381, "learning_rate": 3.419582765824603e-08, "loss": 0.4126, "step": 19040 }, { "epoch": 0.9631746674085689, "grad_norm": 18.04319953918457, "learning_rate": 3.4102183006722304e-08, "loss": 0.4348, "step": 19041 }, { "epoch": 0.9632252516566341, "grad_norm": 6.439100742340088, "learning_rate": 3.40086663143907e-08, "loss": 0.3557, "step": 19042 }, { "epoch": 0.9632758359046992, "grad_norm": 15.03370189666748, "learning_rate": 3.391527758366098e-08, "loss": 0.4433, "step": 19043 }, { "epoch": 0.9633264201527645, "grad_norm": 8.57493782043457, "learning_rate": 3.382201681693953e-08, "loss": 0.4293, "step": 19044 }, { "epoch": 0.9633770044008296, "grad_norm": 4.418814659118652, "learning_rate": 3.3728884016628885e-08, "loss": 0.3516, "step": 19045 }, { "epoch": 0.9634275886488948, "grad_norm": 6.603053092956543, "learning_rate": 3.3635879185129336e-08, "loss": 0.4008, "step": 19046 }, { "epoch": 0.9634781728969599, "grad_norm": 6.076325416564941, "learning_rate": 3.354300232483787e-08, "loss": 0.3094, "step": 19047 }, { "epoch": 0.9635287571450251, "grad_norm": 10.266119956970215, "learning_rate": 3.3450253438147005e-08, "loss": 0.4716, "step": 19048 }, { "epoch": 0.9635793413930902, "grad_norm": 5.813227653503418, "learning_rate": 3.335763252744706e-08, "loss": 0.3532, "step": 19049 }, { "epoch": 0.9636299256411553, "grad_norm": 7.031227111816406, "learning_rate": 3.326513959512445e-08, "loss": 0.4796, "step": 19050 }, { "epoch": 0.9636805098892205, "grad_norm": 5.793607711791992, "learning_rate": 3.3172774643562275e-08, "loss": 0.2516, "step": 19051 }, { "epoch": 0.9637310941372856, "grad_norm": 6.343530178070068, "learning_rate": 3.308053767514086e-08, "loss": 0.2985, "step": 19052 }, { "epoch": 0.9637816783853508, "grad_norm": 7.881117820739746, "learning_rate": 3.2988428692237175e-08, "loss": 0.3991, "step": 19053 }, { "epoch": 0.9638322626334159, "grad_norm": 6.493616580963135, "learning_rate": 3.289644769722433e-08, "loss": 0.3145, "step": 19054 }, { "epoch": 0.9638828468814811, "grad_norm": 9.121353149414062, "learning_rate": 3.280459469247266e-08, "loss": 0.3101, "step": 19055 }, { "epoch": 0.9639334311295462, "grad_norm": 13.055095672607422, "learning_rate": 3.2712869680348594e-08, "loss": 0.5846, "step": 19056 }, { "epoch": 0.9639840153776114, "grad_norm": 12.43072509765625, "learning_rate": 3.26212726632158e-08, "loss": 0.5012, "step": 19057 }, { "epoch": 0.9640345996256766, "grad_norm": 6.005405426025391, "learning_rate": 3.252980364343461e-08, "loss": 0.2997, "step": 19058 }, { "epoch": 0.9640851838737418, "grad_norm": 8.339025497436523, "learning_rate": 3.243846262336148e-08, "loss": 0.4058, "step": 19059 }, { "epoch": 0.9641357681218069, "grad_norm": 9.181503295898438, "learning_rate": 3.234724960535118e-08, "loss": 0.3693, "step": 19060 }, { "epoch": 0.964186352369872, "grad_norm": 6.745359897613525, "learning_rate": 3.2256164591753516e-08, "loss": 0.3505, "step": 19061 }, { "epoch": 0.9642369366179372, "grad_norm": 7.266261100769043, "learning_rate": 3.216520758491492e-08, "loss": 0.3235, "step": 19062 }, { "epoch": 0.9642875208660023, "grad_norm": 14.191193580627441, "learning_rate": 3.207437858717966e-08, "loss": 0.4155, "step": 19063 }, { "epoch": 0.9643381051140675, "grad_norm": 6.403942108154297, "learning_rate": 3.198367760088861e-08, "loss": 0.4086, "step": 19064 }, { "epoch": 0.9643886893621326, "grad_norm": 6.8074822425842285, "learning_rate": 3.189310462837769e-08, "loss": 0.3846, "step": 19065 }, { "epoch": 0.9644392736101978, "grad_norm": 10.165514945983887, "learning_rate": 3.1802659671982264e-08, "loss": 0.452, "step": 19066 }, { "epoch": 0.9644898578582629, "grad_norm": 10.717333793640137, "learning_rate": 3.171234273403156e-08, "loss": 0.4357, "step": 19067 }, { "epoch": 0.964540442106328, "grad_norm": 14.791045188903809, "learning_rate": 3.162215381685374e-08, "loss": 0.4286, "step": 19068 }, { "epoch": 0.9645910263543932, "grad_norm": 4.786934852600098, "learning_rate": 3.153209292277193e-08, "loss": 0.3633, "step": 19069 }, { "epoch": 0.9646416106024583, "grad_norm": 7.746939182281494, "learning_rate": 3.144216005410761e-08, "loss": 0.3871, "step": 19070 }, { "epoch": 0.9646921948505236, "grad_norm": 8.260857582092285, "learning_rate": 3.135235521317781e-08, "loss": 0.3754, "step": 19071 }, { "epoch": 0.9647427790985887, "grad_norm": 6.949330806732178, "learning_rate": 3.1262678402296244e-08, "loss": 0.3806, "step": 19072 }, { "epoch": 0.9647933633466539, "grad_norm": 5.968415260314941, "learning_rate": 3.117312962377439e-08, "loss": 0.4727, "step": 19073 }, { "epoch": 0.964843947594719, "grad_norm": 14.475059509277344, "learning_rate": 3.108370887991874e-08, "loss": 0.4923, "step": 19074 }, { "epoch": 0.9648945318427842, "grad_norm": 7.0823163986206055, "learning_rate": 3.0994416173034114e-08, "loss": 0.2982, "step": 19075 }, { "epoch": 0.9649451160908493, "grad_norm": 6.522983551025391, "learning_rate": 3.090525150542145e-08, "loss": 0.4228, "step": 19076 }, { "epoch": 0.9649957003389145, "grad_norm": 6.88469934463501, "learning_rate": 3.0816214879377805e-08, "loss": 0.3507, "step": 19077 }, { "epoch": 0.9650462845869796, "grad_norm": 4.720477104187012, "learning_rate": 3.0727306297198e-08, "loss": 0.4075, "step": 19078 }, { "epoch": 0.9650968688350448, "grad_norm": 10.713680267333984, "learning_rate": 3.063852576117299e-08, "loss": 0.588, "step": 19079 }, { "epoch": 0.9651474530831099, "grad_norm": 8.721006393432617, "learning_rate": 3.0549873273589824e-08, "loss": 0.4214, "step": 19080 }, { "epoch": 0.965198037331175, "grad_norm": 13.032215118408203, "learning_rate": 3.046134883673335e-08, "loss": 0.4542, "step": 19081 }, { "epoch": 0.9652486215792402, "grad_norm": 13.035181999206543, "learning_rate": 3.037295245288452e-08, "loss": 0.3913, "step": 19082 }, { "epoch": 0.9652992058273053, "grad_norm": 5.755316257476807, "learning_rate": 3.02846841243204e-08, "loss": 0.3947, "step": 19083 }, { "epoch": 0.9653497900753706, "grad_norm": 12.18867301940918, "learning_rate": 3.019654385331694e-08, "loss": 0.4621, "step": 19084 }, { "epoch": 0.9654003743234357, "grad_norm": 5.611472129821777, "learning_rate": 3.0108531642144e-08, "loss": 0.3466, "step": 19085 }, { "epoch": 0.9654509585715009, "grad_norm": 8.529081344604492, "learning_rate": 3.0020647493070324e-08, "loss": 0.2983, "step": 19086 }, { "epoch": 0.965501542819566, "grad_norm": 11.05032730102539, "learning_rate": 2.993289140835964e-08, "loss": 0.4741, "step": 19087 }, { "epoch": 0.9655521270676312, "grad_norm": 6.551280498504639, "learning_rate": 2.984526339027405e-08, "loss": 0.3895, "step": 19088 }, { "epoch": 0.9656027113156963, "grad_norm": 7.613336563110352, "learning_rate": 2.975776344107062e-08, "loss": 0.3737, "step": 19089 }, { "epoch": 0.9656532955637614, "grad_norm": 8.591367721557617, "learning_rate": 2.9670391563004773e-08, "loss": 0.3665, "step": 19090 }, { "epoch": 0.9657038798118266, "grad_norm": 5.497409343719482, "learning_rate": 2.9583147758328045e-08, "loss": 0.34, "step": 19091 }, { "epoch": 0.9657544640598917, "grad_norm": 7.180057048797607, "learning_rate": 2.949603202928697e-08, "loss": 0.4532, "step": 19092 }, { "epoch": 0.9658050483079569, "grad_norm": 6.912944793701172, "learning_rate": 2.9409044378128084e-08, "loss": 0.3779, "step": 19093 }, { "epoch": 0.965855632556022, "grad_norm": 9.840842247009277, "learning_rate": 2.9322184807092368e-08, "loss": 0.4914, "step": 19094 }, { "epoch": 0.9659062168040872, "grad_norm": 9.716115951538086, "learning_rate": 2.9235453318417485e-08, "loss": 0.5584, "step": 19095 }, { "epoch": 0.9659568010521523, "grad_norm": 9.790253639221191, "learning_rate": 2.914884991433886e-08, "loss": 0.502, "step": 19096 }, { "epoch": 0.9660073853002175, "grad_norm": 7.030400276184082, "learning_rate": 2.906237459708694e-08, "loss": 0.4285, "step": 19097 }, { "epoch": 0.9660579695482827, "grad_norm": 6.028654098510742, "learning_rate": 2.8976027368891603e-08, "loss": 0.3298, "step": 19098 }, { "epoch": 0.9661085537963479, "grad_norm": 7.9542059898376465, "learning_rate": 2.8889808231976623e-08, "loss": 0.4233, "step": 19099 }, { "epoch": 0.966159138044413, "grad_norm": 7.004793643951416, "learning_rate": 2.880371718856412e-08, "loss": 0.429, "step": 19100 }, { "epoch": 0.9662097222924781, "grad_norm": 11.492164611816406, "learning_rate": 2.8717754240872864e-08, "loss": 0.5151, "step": 19101 }, { "epoch": 0.9662603065405433, "grad_norm": 5.878258228302002, "learning_rate": 2.86319193911172e-08, "loss": 0.3757, "step": 19102 }, { "epoch": 0.9663108907886084, "grad_norm": 13.319278717041016, "learning_rate": 2.85462126415087e-08, "loss": 0.3494, "step": 19103 }, { "epoch": 0.9663614750366736, "grad_norm": 16.456819534301758, "learning_rate": 2.846063399425669e-08, "loss": 0.4716, "step": 19104 }, { "epoch": 0.9664120592847387, "grad_norm": 6.904541969299316, "learning_rate": 2.8375183451565536e-08, "loss": 0.2941, "step": 19105 }, { "epoch": 0.9664626435328039, "grad_norm": 6.68666934967041, "learning_rate": 2.8289861015637908e-08, "loss": 0.3469, "step": 19106 }, { "epoch": 0.966513227780869, "grad_norm": 5.130353927612305, "learning_rate": 2.82046666886715e-08, "loss": 0.3195, "step": 19107 }, { "epoch": 0.9665638120289342, "grad_norm": 7.929477691650391, "learning_rate": 2.8119600472862328e-08, "loss": 0.4323, "step": 19108 }, { "epoch": 0.9666143962769993, "grad_norm": 7.7264180183410645, "learning_rate": 2.803466237040142e-08, "loss": 0.4311, "step": 19109 }, { "epoch": 0.9666649805250644, "grad_norm": 6.420342445373535, "learning_rate": 2.7949852383478137e-08, "loss": 0.3334, "step": 19110 }, { "epoch": 0.9667155647731297, "grad_norm": 11.869489669799805, "learning_rate": 2.786517051427795e-08, "loss": 0.4967, "step": 19111 }, { "epoch": 0.9667661490211948, "grad_norm": 5.408483982086182, "learning_rate": 2.778061676498245e-08, "loss": 0.4096, "step": 19112 }, { "epoch": 0.96681673326926, "grad_norm": 4.9268083572387695, "learning_rate": 2.7696191137770445e-08, "loss": 0.3371, "step": 19113 }, { "epoch": 0.9668673175173251, "grad_norm": 6.483836650848389, "learning_rate": 2.7611893634817976e-08, "loss": 0.3662, "step": 19114 }, { "epoch": 0.9669179017653903, "grad_norm": 6.494565963745117, "learning_rate": 2.7527724258296085e-08, "loss": 0.3607, "step": 19115 }, { "epoch": 0.9669684860134554, "grad_norm": 8.94465446472168, "learning_rate": 2.74436830103747e-08, "loss": 0.4646, "step": 19116 }, { "epoch": 0.9670190702615206, "grad_norm": 8.934183120727539, "learning_rate": 2.7359769893218203e-08, "loss": 0.4064, "step": 19117 }, { "epoch": 0.9670696545095857, "grad_norm": 8.16831111907959, "learning_rate": 2.727598490898986e-08, "loss": 0.4101, "step": 19118 }, { "epoch": 0.9671202387576509, "grad_norm": 11.1720552444458, "learning_rate": 2.7192328059848504e-08, "loss": 0.4149, "step": 19119 }, { "epoch": 0.967170823005716, "grad_norm": 9.248506546020508, "learning_rate": 2.7108799347949634e-08, "loss": 0.3595, "step": 19120 }, { "epoch": 0.9672214072537811, "grad_norm": 7.812902927398682, "learning_rate": 2.7025398775445412e-08, "loss": 0.3186, "step": 19121 }, { "epoch": 0.9672719915018463, "grad_norm": 9.080743789672852, "learning_rate": 2.6942126344484674e-08, "loss": 0.5099, "step": 19122 }, { "epoch": 0.9673225757499114, "grad_norm": 5.489114284515381, "learning_rate": 2.685898205721349e-08, "loss": 0.3671, "step": 19123 }, { "epoch": 0.9673731599979767, "grad_norm": 6.774485111236572, "learning_rate": 2.6775965915774028e-08, "loss": 0.4197, "step": 19124 }, { "epoch": 0.9674237442460418, "grad_norm": 5.687948703765869, "learning_rate": 2.669307792230569e-08, "loss": 0.3088, "step": 19125 }, { "epoch": 0.967474328494107, "grad_norm": 5.404738426208496, "learning_rate": 2.6610318078944543e-08, "loss": 0.3442, "step": 19126 }, { "epoch": 0.9675249127421721, "grad_norm": 6.8606696128845215, "learning_rate": 2.6527686387822214e-08, "loss": 0.2982, "step": 19127 }, { "epoch": 0.9675754969902373, "grad_norm": 5.374120235443115, "learning_rate": 2.644518285106923e-08, "loss": 0.4275, "step": 19128 }, { "epoch": 0.9676260812383024, "grad_norm": 7.6411542892456055, "learning_rate": 2.6362807470809993e-08, "loss": 0.3323, "step": 19129 }, { "epoch": 0.9676766654863675, "grad_norm": 11.228497505187988, "learning_rate": 2.6280560249168364e-08, "loss": 0.5189, "step": 19130 }, { "epoch": 0.9677272497344327, "grad_norm": 20.56583023071289, "learning_rate": 2.6198441188262647e-08, "loss": 0.3788, "step": 19131 }, { "epoch": 0.9677778339824978, "grad_norm": 8.55732536315918, "learning_rate": 2.611645029021004e-08, "loss": 0.4032, "step": 19132 }, { "epoch": 0.967828418230563, "grad_norm": 6.899807929992676, "learning_rate": 2.6034587557122182e-08, "loss": 0.4427, "step": 19133 }, { "epoch": 0.9678790024786281, "grad_norm": 6.1247453689575195, "learning_rate": 2.5952852991109056e-08, "loss": 0.2997, "step": 19134 }, { "epoch": 0.9679295867266933, "grad_norm": 5.733847618103027, "learning_rate": 2.58712465942762e-08, "loss": 0.2559, "step": 19135 }, { "epoch": 0.9679801709747584, "grad_norm": 8.963194847106934, "learning_rate": 2.5789768368727484e-08, "loss": 0.3912, "step": 19136 }, { "epoch": 0.9680307552228236, "grad_norm": 4.57608699798584, "learning_rate": 2.5708418316561235e-08, "loss": 0.3973, "step": 19137 }, { "epoch": 0.9680813394708888, "grad_norm": 13.299722671508789, "learning_rate": 2.5627196439874104e-08, "loss": 0.4862, "step": 19138 }, { "epoch": 0.968131923718954, "grad_norm": 5.656167507171631, "learning_rate": 2.5546102740759415e-08, "loss": 0.3635, "step": 19139 }, { "epoch": 0.9681825079670191, "grad_norm": 7.143362998962402, "learning_rate": 2.546513722130606e-08, "loss": 0.4175, "step": 19140 }, { "epoch": 0.9682330922150842, "grad_norm": 5.852505207061768, "learning_rate": 2.5384299883600693e-08, "loss": 0.4163, "step": 19141 }, { "epoch": 0.9682836764631494, "grad_norm": 9.152562141418457, "learning_rate": 2.53035907297261e-08, "loss": 0.5186, "step": 19142 }, { "epoch": 0.9683342607112145, "grad_norm": 8.390766143798828, "learning_rate": 2.5223009761762286e-08, "loss": 0.3761, "step": 19143 }, { "epoch": 0.9683848449592797, "grad_norm": 9.656712532043457, "learning_rate": 2.514255698178536e-08, "loss": 0.3298, "step": 19144 }, { "epoch": 0.9684354292073448, "grad_norm": 9.343782424926758, "learning_rate": 2.5062232391868667e-08, "loss": 0.416, "step": 19145 }, { "epoch": 0.96848601345541, "grad_norm": 7.723908424377441, "learning_rate": 2.4982035994081112e-08, "loss": 0.4117, "step": 19146 }, { "epoch": 0.9685365977034751, "grad_norm": 8.369925498962402, "learning_rate": 2.4901967790490477e-08, "loss": 0.3627, "step": 19147 }, { "epoch": 0.9685871819515403, "grad_norm": 14.448040008544922, "learning_rate": 2.4822027783159007e-08, "loss": 0.4393, "step": 19148 }, { "epoch": 0.9686377661996054, "grad_norm": 5.158434867858887, "learning_rate": 2.474221597414672e-08, "loss": 0.364, "step": 19149 }, { "epoch": 0.9686883504476705, "grad_norm": 6.797394275665283, "learning_rate": 2.4662532365510304e-08, "loss": 0.4211, "step": 19150 }, { "epoch": 0.9687389346957358, "grad_norm": 11.13834285736084, "learning_rate": 2.4582976959303116e-08, "loss": 0.4671, "step": 19151 }, { "epoch": 0.9687895189438009, "grad_norm": 9.350984573364258, "learning_rate": 2.4503549757575185e-08, "loss": 0.3605, "step": 19152 }, { "epoch": 0.9688401031918661, "grad_norm": 6.243455410003662, "learning_rate": 2.4424250762372647e-08, "loss": 0.3567, "step": 19153 }, { "epoch": 0.9688906874399312, "grad_norm": 12.984911918640137, "learning_rate": 2.4345079975739428e-08, "loss": 0.5328, "step": 19154 }, { "epoch": 0.9689412716879964, "grad_norm": 18.483213424682617, "learning_rate": 2.4266037399715002e-08, "loss": 0.5063, "step": 19155 }, { "epoch": 0.9689918559360615, "grad_norm": 5.60987663269043, "learning_rate": 2.4187123036336634e-08, "loss": 0.3791, "step": 19156 }, { "epoch": 0.9690424401841267, "grad_norm": 7.206557273864746, "learning_rate": 2.410833688763714e-08, "loss": 0.3655, "step": 19157 }, { "epoch": 0.9690930244321918, "grad_norm": 10.518209457397461, "learning_rate": 2.4029678955647672e-08, "loss": 0.4515, "step": 19158 }, { "epoch": 0.969143608680257, "grad_norm": 8.720449447631836, "learning_rate": 2.3951149242393835e-08, "loss": 0.5342, "step": 19159 }, { "epoch": 0.9691941929283221, "grad_norm": 11.150731086730957, "learning_rate": 2.3872747749900116e-08, "loss": 0.4642, "step": 19160 }, { "epoch": 0.9692447771763872, "grad_norm": 16.2115478515625, "learning_rate": 2.3794474480186568e-08, "loss": 0.4458, "step": 19161 }, { "epoch": 0.9692953614244524, "grad_norm": 9.196189880371094, "learning_rate": 2.3716329435269358e-08, "loss": 0.3467, "step": 19162 }, { "epoch": 0.9693459456725175, "grad_norm": 8.785021781921387, "learning_rate": 2.363831261716243e-08, "loss": 0.415, "step": 19163 }, { "epoch": 0.9693965299205828, "grad_norm": 7.69280481338501, "learning_rate": 2.356042402787695e-08, "loss": 0.3753, "step": 19164 }, { "epoch": 0.9694471141686479, "grad_norm": 8.931581497192383, "learning_rate": 2.3482663669418537e-08, "loss": 0.4309, "step": 19165 }, { "epoch": 0.9694976984167131, "grad_norm": 6.419468879699707, "learning_rate": 2.3405031543792257e-08, "loss": 0.3389, "step": 19166 }, { "epoch": 0.9695482826647782, "grad_norm": 13.144156455993652, "learning_rate": 2.3327527652997618e-08, "loss": 0.4437, "step": 19167 }, { "epoch": 0.9695988669128434, "grad_norm": 8.476690292358398, "learning_rate": 2.3250151999031357e-08, "loss": 0.3883, "step": 19168 }, { "epoch": 0.9696494511609085, "grad_norm": 10.247334480285645, "learning_rate": 2.3172904583888544e-08, "loss": 0.5641, "step": 19169 }, { "epoch": 0.9697000354089736, "grad_norm": 6.102267265319824, "learning_rate": 2.3095785409558702e-08, "loss": 0.4197, "step": 19170 }, { "epoch": 0.9697506196570388, "grad_norm": 9.742809295654297, "learning_rate": 2.301879447802968e-08, "loss": 0.3521, "step": 19171 }, { "epoch": 0.9698012039051039, "grad_norm": 6.006000518798828, "learning_rate": 2.2941931791284345e-08, "loss": 0.3564, "step": 19172 }, { "epoch": 0.9698517881531691, "grad_norm": 10.382554054260254, "learning_rate": 2.286519735130388e-08, "loss": 0.3855, "step": 19173 }, { "epoch": 0.9699023724012342, "grad_norm": 6.433651924133301, "learning_rate": 2.2788591160065598e-08, "loss": 0.3445, "step": 19174 }, { "epoch": 0.9699529566492994, "grad_norm": 9.01120662689209, "learning_rate": 2.2712113219543474e-08, "loss": 0.379, "step": 19175 }, { "epoch": 0.9700035408973645, "grad_norm": 7.012601375579834, "learning_rate": 2.2635763531708156e-08, "loss": 0.4131, "step": 19176 }, { "epoch": 0.9700541251454298, "grad_norm": 6.412844657897949, "learning_rate": 2.25595420985264e-08, "loss": 0.3559, "step": 19177 }, { "epoch": 0.9701047093934949, "grad_norm": 5.18344259262085, "learning_rate": 2.2483448921963302e-08, "loss": 0.3326, "step": 19178 }, { "epoch": 0.9701552936415601, "grad_norm": 6.273770809173584, "learning_rate": 2.2407484003978407e-08, "loss": 0.3967, "step": 19179 }, { "epoch": 0.9702058778896252, "grad_norm": 6.9114089012146, "learning_rate": 2.233164734653015e-08, "loss": 0.403, "step": 19180 }, { "epoch": 0.9702564621376903, "grad_norm": 13.862628936767578, "learning_rate": 2.225593895157252e-08, "loss": 0.4415, "step": 19181 }, { "epoch": 0.9703070463857555, "grad_norm": 11.594823837280273, "learning_rate": 2.2180358821055624e-08, "loss": 0.4364, "step": 19182 }, { "epoch": 0.9703576306338206, "grad_norm": 7.206468105316162, "learning_rate": 2.210490695692735e-08, "loss": 0.244, "step": 19183 }, { "epoch": 0.9704082148818858, "grad_norm": 7.6826372146606445, "learning_rate": 2.202958336113281e-08, "loss": 0.356, "step": 19184 }, { "epoch": 0.9704587991299509, "grad_norm": 11.990702629089355, "learning_rate": 2.1954388035611562e-08, "loss": 0.4599, "step": 19185 }, { "epoch": 0.9705093833780161, "grad_norm": 8.617731094360352, "learning_rate": 2.1879320982301498e-08, "loss": 0.4133, "step": 19186 }, { "epoch": 0.9705599676260812, "grad_norm": 6.69360876083374, "learning_rate": 2.180438220313774e-08, "loss": 0.3049, "step": 19187 }, { "epoch": 0.9706105518741464, "grad_norm": 11.781083106994629, "learning_rate": 2.1729571700050413e-08, "loss": 0.4548, "step": 19188 }, { "epoch": 0.9706611361222115, "grad_norm": 6.97681999206543, "learning_rate": 2.165488947496741e-08, "loss": 0.3067, "step": 19189 }, { "epoch": 0.9707117203702766, "grad_norm": 10.907273292541504, "learning_rate": 2.158033552981331e-08, "loss": 0.5151, "step": 19190 }, { "epoch": 0.9707623046183419, "grad_norm": 8.239314079284668, "learning_rate": 2.15059098665088e-08, "loss": 0.3548, "step": 19191 }, { "epoch": 0.970812888866407, "grad_norm": 6.301379203796387, "learning_rate": 2.1431612486972342e-08, "loss": 0.4108, "step": 19192 }, { "epoch": 0.9708634731144722, "grad_norm": 11.244921684265137, "learning_rate": 2.135744339311796e-08, "loss": 0.4741, "step": 19193 }, { "epoch": 0.9709140573625373, "grad_norm": 8.803017616271973, "learning_rate": 2.1283402586856906e-08, "loss": 0.4297, "step": 19194 }, { "epoch": 0.9709646416106025, "grad_norm": 10.654804229736328, "learning_rate": 2.1209490070096538e-08, "loss": 0.4811, "step": 19195 }, { "epoch": 0.9710152258586676, "grad_norm": 7.978821277618408, "learning_rate": 2.113570584474256e-08, "loss": 0.2745, "step": 19196 }, { "epoch": 0.9710658101067328, "grad_norm": 7.55889368057251, "learning_rate": 2.1062049912695116e-08, "loss": 0.3733, "step": 19197 }, { "epoch": 0.9711163943547979, "grad_norm": 5.9427056312561035, "learning_rate": 2.098852227585213e-08, "loss": 0.339, "step": 19198 }, { "epoch": 0.971166978602863, "grad_norm": 5.555510520935059, "learning_rate": 2.0915122936109312e-08, "loss": 0.4482, "step": 19199 }, { "epoch": 0.9712175628509282, "grad_norm": 9.326208114624023, "learning_rate": 2.0841851895356813e-08, "loss": 0.3054, "step": 19200 }, { "epoch": 0.9712681470989933, "grad_norm": 7.11533260345459, "learning_rate": 2.0768709155483124e-08, "loss": 0.3467, "step": 19201 }, { "epoch": 0.9713187313470585, "grad_norm": 5.466614246368408, "learning_rate": 2.0695694718373404e-08, "loss": 0.2661, "step": 19202 }, { "epoch": 0.9713693155951236, "grad_norm": 5.909693717956543, "learning_rate": 2.0622808585907818e-08, "loss": 0.2413, "step": 19203 }, { "epoch": 0.9714198998431889, "grad_norm": 9.780470848083496, "learning_rate": 2.0550050759965966e-08, "loss": 0.4175, "step": 19204 }, { "epoch": 0.971470484091254, "grad_norm": 5.035067081451416, "learning_rate": 2.047742124242136e-08, "loss": 0.3236, "step": 19205 }, { "epoch": 0.9715210683393192, "grad_norm": 9.695819854736328, "learning_rate": 2.0404920035146382e-08, "loss": 0.3893, "step": 19206 }, { "epoch": 0.9715716525873843, "grad_norm": 6.075021743774414, "learning_rate": 2.0332547140008986e-08, "loss": 0.3391, "step": 19207 }, { "epoch": 0.9716222368354495, "grad_norm": 6.197580337524414, "learning_rate": 2.026030255887379e-08, "loss": 0.3894, "step": 19208 }, { "epoch": 0.9716728210835146, "grad_norm": 10.048173904418945, "learning_rate": 2.018818629360264e-08, "loss": 0.4555, "step": 19209 }, { "epoch": 0.9717234053315797, "grad_norm": 6.794014930725098, "learning_rate": 2.011619834605405e-08, "loss": 0.4, "step": 19210 }, { "epoch": 0.9717739895796449, "grad_norm": 7.2641987800598145, "learning_rate": 2.004433871808209e-08, "loss": 0.3177, "step": 19211 }, { "epoch": 0.97182457382771, "grad_norm": 7.5334391593933105, "learning_rate": 1.997260741153917e-08, "loss": 0.5192, "step": 19212 }, { "epoch": 0.9718751580757752, "grad_norm": 6.497470855712891, "learning_rate": 1.990100442827325e-08, "loss": 0.3533, "step": 19213 }, { "epoch": 0.9719257423238403, "grad_norm": 11.6253662109375, "learning_rate": 1.982952977012953e-08, "loss": 0.4118, "step": 19214 }, { "epoch": 0.9719763265719055, "grad_norm": 7.380658149719238, "learning_rate": 1.9758183438949863e-08, "loss": 0.3844, "step": 19215 }, { "epoch": 0.9720269108199706, "grad_norm": 14.331914901733398, "learning_rate": 1.9686965436572224e-08, "loss": 0.4913, "step": 19216 }, { "epoch": 0.9720774950680359, "grad_norm": 8.2938871383667, "learning_rate": 1.9615875764832372e-08, "loss": 0.3335, "step": 19217 }, { "epoch": 0.972128079316101, "grad_norm": 8.004751205444336, "learning_rate": 1.954491442556161e-08, "loss": 0.4434, "step": 19218 }, { "epoch": 0.9721786635641662, "grad_norm": 5.703256607055664, "learning_rate": 1.9474081420588488e-08, "loss": 0.3516, "step": 19219 }, { "epoch": 0.9722292478122313, "grad_norm": 9.648547172546387, "learning_rate": 1.94033767517382e-08, "loss": 0.4293, "step": 19220 }, { "epoch": 0.9722798320602964, "grad_norm": 9.097136497497559, "learning_rate": 1.9332800420833187e-08, "loss": 0.3846, "step": 19221 }, { "epoch": 0.9723304163083616, "grad_norm": 8.081413269042969, "learning_rate": 1.9262352429691434e-08, "loss": 0.42, "step": 19222 }, { "epoch": 0.9723810005564267, "grad_norm": 11.356318473815918, "learning_rate": 1.9192032780128155e-08, "loss": 0.554, "step": 19223 }, { "epoch": 0.9724315848044919, "grad_norm": 10.242371559143066, "learning_rate": 1.9121841473955793e-08, "loss": 0.3183, "step": 19224 }, { "epoch": 0.972482169052557, "grad_norm": 6.825793743133545, "learning_rate": 1.90517785129829e-08, "loss": 0.3994, "step": 19225 }, { "epoch": 0.9725327533006222, "grad_norm": 12.249797821044922, "learning_rate": 1.8981843899014142e-08, "loss": 0.3604, "step": 19226 }, { "epoch": 0.9725833375486873, "grad_norm": 5.382823944091797, "learning_rate": 1.8912037633852522e-08, "loss": 0.3513, "step": 19227 }, { "epoch": 0.9726339217967525, "grad_norm": 9.047396659851074, "learning_rate": 1.88423597192966e-08, "loss": 0.4691, "step": 19228 }, { "epoch": 0.9726845060448176, "grad_norm": 8.565035820007324, "learning_rate": 1.8772810157141054e-08, "loss": 0.3506, "step": 19229 }, { "epoch": 0.9727350902928827, "grad_norm": 5.26038122177124, "learning_rate": 1.870338894917889e-08, "loss": 0.3995, "step": 19230 }, { "epoch": 0.972785674540948, "grad_norm": 9.824443817138672, "learning_rate": 1.8634096097198683e-08, "loss": 0.4019, "step": 19231 }, { "epoch": 0.9728362587890131, "grad_norm": 9.607089042663574, "learning_rate": 1.8564931602985115e-08, "loss": 0.398, "step": 19232 }, { "epoch": 0.9728868430370783, "grad_norm": 11.492511749267578, "learning_rate": 1.8495895468321756e-08, "loss": 0.5014, "step": 19233 }, { "epoch": 0.9729374272851434, "grad_norm": 6.748183727264404, "learning_rate": 1.842698769498663e-08, "loss": 0.4703, "step": 19234 }, { "epoch": 0.9729880115332086, "grad_norm": 12.421038627624512, "learning_rate": 1.835820828475554e-08, "loss": 0.6104, "step": 19235 }, { "epoch": 0.9730385957812737, "grad_norm": 7.447902202606201, "learning_rate": 1.8289557239400957e-08, "loss": 0.3807, "step": 19236 }, { "epoch": 0.9730891800293389, "grad_norm": 5.732038497924805, "learning_rate": 1.8221034560691463e-08, "loss": 0.3604, "step": 19237 }, { "epoch": 0.973139764277404, "grad_norm": 5.303795337677002, "learning_rate": 1.8152640250392874e-08, "loss": 0.329, "step": 19238 }, { "epoch": 0.9731903485254692, "grad_norm": 7.648007869720459, "learning_rate": 1.808437431026766e-08, "loss": 0.3424, "step": 19239 }, { "epoch": 0.9732409327735343, "grad_norm": 9.599041938781738, "learning_rate": 1.8016236742074977e-08, "loss": 0.3249, "step": 19240 }, { "epoch": 0.9732915170215994, "grad_norm": 13.371330261230469, "learning_rate": 1.794822754756953e-08, "loss": 0.3363, "step": 19241 }, { "epoch": 0.9733421012696646, "grad_norm": 9.04373836517334, "learning_rate": 1.788034672850547e-08, "loss": 0.4036, "step": 19242 }, { "epoch": 0.9733926855177297, "grad_norm": 16.417701721191406, "learning_rate": 1.7812594286630846e-08, "loss": 0.3575, "step": 19243 }, { "epoch": 0.973443269765795, "grad_norm": 12.902902603149414, "learning_rate": 1.7744970223691483e-08, "loss": 0.4302, "step": 19244 }, { "epoch": 0.9734938540138601, "grad_norm": 8.801055908203125, "learning_rate": 1.7677474541430427e-08, "loss": 0.3487, "step": 19245 }, { "epoch": 0.9735444382619253, "grad_norm": 4.989299774169922, "learning_rate": 1.7610107241586295e-08, "loss": 0.4034, "step": 19246 }, { "epoch": 0.9735950225099904, "grad_norm": 9.472009658813477, "learning_rate": 1.7542868325895468e-08, "loss": 0.4866, "step": 19247 }, { "epoch": 0.9736456067580556, "grad_norm": 5.968089580535889, "learning_rate": 1.7475757796090453e-08, "loss": 0.3263, "step": 19248 }, { "epoch": 0.9736961910061207, "grad_norm": 7.482591152191162, "learning_rate": 1.7408775653899868e-08, "loss": 0.3625, "step": 19249 }, { "epoch": 0.9737467752541858, "grad_norm": 11.5912504196167, "learning_rate": 1.7341921901050106e-08, "loss": 0.4047, "step": 19250 }, { "epoch": 0.973797359502251, "grad_norm": 5.394631385803223, "learning_rate": 1.7275196539264238e-08, "loss": 0.4478, "step": 19251 }, { "epoch": 0.9738479437503161, "grad_norm": 8.2085599899292, "learning_rate": 1.720859957026144e-08, "loss": 0.3771, "step": 19252 }, { "epoch": 0.9738985279983813, "grad_norm": 4.827025890350342, "learning_rate": 1.7142130995757567e-08, "loss": 0.341, "step": 19253 }, { "epoch": 0.9739491122464464, "grad_norm": 16.962928771972656, "learning_rate": 1.707579081746513e-08, "loss": 0.5648, "step": 19254 }, { "epoch": 0.9739996964945116, "grad_norm": 5.618841648101807, "learning_rate": 1.7009579037093882e-08, "loss": 0.4031, "step": 19255 }, { "epoch": 0.9740502807425767, "grad_norm": 7.76732873916626, "learning_rate": 1.6943495656349675e-08, "loss": 0.4194, "step": 19256 }, { "epoch": 0.974100864990642, "grad_norm": 10.442781448364258, "learning_rate": 1.687754067693559e-08, "loss": 0.374, "step": 19257 }, { "epoch": 0.9741514492387071, "grad_norm": 7.609795570373535, "learning_rate": 1.681171410055138e-08, "loss": 0.4137, "step": 19258 }, { "epoch": 0.9742020334867723, "grad_norm": 15.028616905212402, "learning_rate": 1.6746015928892357e-08, "loss": 0.4896, "step": 19259 }, { "epoch": 0.9742526177348374, "grad_norm": 7.630886077880859, "learning_rate": 1.6680446163652163e-08, "loss": 0.4496, "step": 19260 }, { "epoch": 0.9743032019829025, "grad_norm": 10.259356498718262, "learning_rate": 1.6615004806520008e-08, "loss": 0.4202, "step": 19261 }, { "epoch": 0.9743537862309677, "grad_norm": 10.70117473602295, "learning_rate": 1.654969185918287e-08, "loss": 0.4497, "step": 19262 }, { "epoch": 0.9744043704790328, "grad_norm": 13.015063285827637, "learning_rate": 1.648450732332274e-08, "loss": 0.3853, "step": 19263 }, { "epoch": 0.974454954727098, "grad_norm": 9.374778747558594, "learning_rate": 1.6419451200619384e-08, "loss": 0.4567, "step": 19264 }, { "epoch": 0.9745055389751631, "grad_norm": 5.375091552734375, "learning_rate": 1.6354523492749797e-08, "loss": 0.3507, "step": 19265 }, { "epoch": 0.9745561232232283, "grad_norm": 5.353598594665527, "learning_rate": 1.628972420138708e-08, "loss": 0.3031, "step": 19266 }, { "epoch": 0.9746067074712934, "grad_norm": 9.880282402038574, "learning_rate": 1.62250533281999e-08, "loss": 0.392, "step": 19267 }, { "epoch": 0.9746572917193586, "grad_norm": 10.637388229370117, "learning_rate": 1.616051087485526e-08, "loss": 0.4879, "step": 19268 }, { "epoch": 0.9747078759674237, "grad_norm": 6.377999782562256, "learning_rate": 1.6096096843016272e-08, "loss": 0.3745, "step": 19269 }, { "epoch": 0.9747584602154888, "grad_norm": 9.939652442932129, "learning_rate": 1.603181123434272e-08, "loss": 0.4573, "step": 19270 }, { "epoch": 0.9748090444635541, "grad_norm": 5.349751949310303, "learning_rate": 1.596765405049161e-08, "loss": 0.3557, "step": 19271 }, { "epoch": 0.9748596287116192, "grad_norm": 7.686723709106445, "learning_rate": 1.590362529311551e-08, "loss": 0.4599, "step": 19272 }, { "epoch": 0.9749102129596844, "grad_norm": 9.660411834716797, "learning_rate": 1.5839724963863655e-08, "loss": 0.4428, "step": 19273 }, { "epoch": 0.9749607972077495, "grad_norm": 8.304319381713867, "learning_rate": 1.5775953064384175e-08, "loss": 0.468, "step": 19274 }, { "epoch": 0.9750113814558147, "grad_norm": 6.141412258148193, "learning_rate": 1.5712309596319087e-08, "loss": 0.3294, "step": 19275 }, { "epoch": 0.9750619657038798, "grad_norm": 14.215465545654297, "learning_rate": 1.5648794561308743e-08, "loss": 0.4854, "step": 19276 }, { "epoch": 0.975112549951945, "grad_norm": 7.10707426071167, "learning_rate": 1.5585407960990174e-08, "loss": 0.4539, "step": 19277 }, { "epoch": 0.9751631342000101, "grad_norm": 9.924158096313477, "learning_rate": 1.5522149796995954e-08, "loss": 0.4754, "step": 19278 }, { "epoch": 0.9752137184480753, "grad_norm": 12.213037490844727, "learning_rate": 1.54590200709559e-08, "loss": 0.3602, "step": 19279 }, { "epoch": 0.9752643026961404, "grad_norm": 6.9438018798828125, "learning_rate": 1.5396018784498146e-08, "loss": 0.4214, "step": 19280 }, { "epoch": 0.9753148869442055, "grad_norm": 8.362837791442871, "learning_rate": 1.533314593924473e-08, "loss": 0.3899, "step": 19281 }, { "epoch": 0.9753654711922707, "grad_norm": 9.581123352050781, "learning_rate": 1.5270401536816027e-08, "loss": 0.3113, "step": 19282 }, { "epoch": 0.9754160554403358, "grad_norm": 5.663834571838379, "learning_rate": 1.520778557882907e-08, "loss": 0.3235, "step": 19283 }, { "epoch": 0.9754666396884011, "grad_norm": 11.254725456237793, "learning_rate": 1.5145298066897574e-08, "loss": 0.3886, "step": 19284 }, { "epoch": 0.9755172239364662, "grad_norm": 12.937041282653809, "learning_rate": 1.5082939002630802e-08, "loss": 0.4952, "step": 19285 }, { "epoch": 0.9755678081845314, "grad_norm": 7.674630165100098, "learning_rate": 1.5020708387636362e-08, "loss": 0.3647, "step": 19286 }, { "epoch": 0.9756183924325965, "grad_norm": 7.071813106536865, "learning_rate": 1.4958606223517968e-08, "loss": 0.3093, "step": 19287 }, { "epoch": 0.9756689766806617, "grad_norm": 5.564594268798828, "learning_rate": 1.4896632511874343e-08, "loss": 0.2451, "step": 19288 }, { "epoch": 0.9757195609287268, "grad_norm": 8.688708305358887, "learning_rate": 1.4834787254304206e-08, "loss": 0.4975, "step": 19289 }, { "epoch": 0.975770145176792, "grad_norm": 8.784627914428711, "learning_rate": 1.4773070452400729e-08, "loss": 0.4167, "step": 19290 }, { "epoch": 0.9758207294248571, "grad_norm": 8.855379104614258, "learning_rate": 1.4711482107753194e-08, "loss": 0.3926, "step": 19291 }, { "epoch": 0.9758713136729222, "grad_norm": 5.768211841583252, "learning_rate": 1.4650022221949778e-08, "loss": 0.3774, "step": 19292 }, { "epoch": 0.9759218979209874, "grad_norm": 10.039579391479492, "learning_rate": 1.4588690796573101e-08, "loss": 0.3895, "step": 19293 }, { "epoch": 0.9759724821690525, "grad_norm": 14.432644844055176, "learning_rate": 1.4527487833204679e-08, "loss": 0.3892, "step": 19294 }, { "epoch": 0.9760230664171177, "grad_norm": 5.976930141448975, "learning_rate": 1.4466413333421026e-08, "loss": 0.3297, "step": 19295 }, { "epoch": 0.9760736506651828, "grad_norm": 5.446269512176514, "learning_rate": 1.4405467298795328e-08, "loss": 0.3728, "step": 19296 }, { "epoch": 0.9761242349132481, "grad_norm": 6.174165725708008, "learning_rate": 1.4344649730898552e-08, "loss": 0.3596, "step": 19297 }, { "epoch": 0.9761748191613132, "grad_norm": 8.675310134887695, "learning_rate": 1.4283960631298332e-08, "loss": 0.4534, "step": 19298 }, { "epoch": 0.9762254034093784, "grad_norm": 13.660090446472168, "learning_rate": 1.4223400001557863e-08, "loss": 0.4782, "step": 19299 }, { "epoch": 0.9762759876574435, "grad_norm": 9.555516242980957, "learning_rate": 1.4162967843237007e-08, "loss": 0.4951, "step": 19300 }, { "epoch": 0.9763265719055086, "grad_norm": 7.098830699920654, "learning_rate": 1.410266415789452e-08, "loss": 0.4348, "step": 19301 }, { "epoch": 0.9763771561535738, "grad_norm": 8.06128978729248, "learning_rate": 1.4042488947083044e-08, "loss": 0.4332, "step": 19302 }, { "epoch": 0.9764277404016389, "grad_norm": 8.695982933044434, "learning_rate": 1.3982442212353565e-08, "loss": 0.4437, "step": 19303 }, { "epoch": 0.9764783246497041, "grad_norm": 6.485604286193848, "learning_rate": 1.3922523955253176e-08, "loss": 0.406, "step": 19304 }, { "epoch": 0.9765289088977692, "grad_norm": 11.067604064941406, "learning_rate": 1.3862734177326198e-08, "loss": 0.3159, "step": 19305 }, { "epoch": 0.9765794931458344, "grad_norm": 7.810334205627441, "learning_rate": 1.3803072880113066e-08, "loss": 0.4192, "step": 19306 }, { "epoch": 0.9766300773938995, "grad_norm": 9.501264572143555, "learning_rate": 1.3743540065150884e-08, "loss": 0.4394, "step": 19307 }, { "epoch": 0.9766806616419647, "grad_norm": 8.013266563415527, "learning_rate": 1.368413573397398e-08, "loss": 0.4075, "step": 19308 }, { "epoch": 0.9767312458900298, "grad_norm": 9.98521900177002, "learning_rate": 1.3624859888112795e-08, "loss": 0.3829, "step": 19309 }, { "epoch": 0.9767818301380949, "grad_norm": 4.073935031890869, "learning_rate": 1.3565712529095e-08, "loss": 0.2878, "step": 19310 }, { "epoch": 0.9768324143861602, "grad_norm": 7.771218299865723, "learning_rate": 1.3506693658444926e-08, "loss": 0.4164, "step": 19311 }, { "epoch": 0.9768829986342253, "grad_norm": 7.840000152587891, "learning_rate": 1.344780327768247e-08, "loss": 0.3199, "step": 19312 }, { "epoch": 0.9769335828822905, "grad_norm": 11.289810180664062, "learning_rate": 1.3389041388325863e-08, "loss": 0.4617, "step": 19313 }, { "epoch": 0.9769841671303556, "grad_norm": 7.57644510269165, "learning_rate": 1.3330407991888894e-08, "loss": 0.3431, "step": 19314 }, { "epoch": 0.9770347513784208, "grad_norm": 6.328084468841553, "learning_rate": 1.3271903089882575e-08, "loss": 0.4438, "step": 19315 }, { "epoch": 0.9770853356264859, "grad_norm": 15.801247596740723, "learning_rate": 1.3213526683814037e-08, "loss": 0.3854, "step": 19316 }, { "epoch": 0.9771359198745511, "grad_norm": 5.96509313583374, "learning_rate": 1.3155278775188185e-08, "loss": 0.4304, "step": 19317 }, { "epoch": 0.9771865041226162, "grad_norm": 14.276298522949219, "learning_rate": 1.3097159365505486e-08, "loss": 0.4731, "step": 19318 }, { "epoch": 0.9772370883706814, "grad_norm": 11.77142333984375, "learning_rate": 1.3039168456263628e-08, "loss": 0.4217, "step": 19319 }, { "epoch": 0.9772876726187465, "grad_norm": 9.424978256225586, "learning_rate": 1.2981306048956422e-08, "loss": 0.5402, "step": 19320 }, { "epoch": 0.9773382568668116, "grad_norm": 10.881070137023926, "learning_rate": 1.2923572145076003e-08, "loss": 0.5043, "step": 19321 }, { "epoch": 0.9773888411148768, "grad_norm": 34.521690368652344, "learning_rate": 1.286596674610896e-08, "loss": 0.4609, "step": 19322 }, { "epoch": 0.9774394253629419, "grad_norm": 4.990634918212891, "learning_rate": 1.2808489853540218e-08, "loss": 0.302, "step": 19323 }, { "epoch": 0.9774900096110072, "grad_norm": 7.727692604064941, "learning_rate": 1.2751141468850259e-08, "loss": 0.3884, "step": 19324 }, { "epoch": 0.9775405938590723, "grad_norm": 9.250515937805176, "learning_rate": 1.26939215935179e-08, "loss": 0.3121, "step": 19325 }, { "epoch": 0.9775911781071375, "grad_norm": 24.83652687072754, "learning_rate": 1.2636830229016405e-08, "loss": 0.5112, "step": 19326 }, { "epoch": 0.9776417623552026, "grad_norm": 9.286627769470215, "learning_rate": 1.2579867376817378e-08, "loss": 0.4163, "step": 19327 }, { "epoch": 0.9776923466032678, "grad_norm": 15.654725074768066, "learning_rate": 1.2523033038388532e-08, "loss": 0.4101, "step": 19328 }, { "epoch": 0.9777429308513329, "grad_norm": 8.798094749450684, "learning_rate": 1.2466327215194251e-08, "loss": 0.4281, "step": 19329 }, { "epoch": 0.977793515099398, "grad_norm": 7.014035701751709, "learning_rate": 1.240974990869559e-08, "loss": 0.3898, "step": 19330 }, { "epoch": 0.9778440993474632, "grad_norm": 6.768829345703125, "learning_rate": 1.2353301120350825e-08, "loss": 0.3094, "step": 19331 }, { "epoch": 0.9778946835955283, "grad_norm": 6.022800922393799, "learning_rate": 1.2296980851614348e-08, "loss": 0.3648, "step": 19332 }, { "epoch": 0.9779452678435935, "grad_norm": 6.091453552246094, "learning_rate": 1.2240789103937778e-08, "loss": 0.3371, "step": 19333 }, { "epoch": 0.9779958520916586, "grad_norm": 5.500362396240234, "learning_rate": 1.2184725878768289e-08, "loss": 0.3445, "step": 19334 }, { "epoch": 0.9780464363397238, "grad_norm": 7.194159984588623, "learning_rate": 1.2128791177550836e-08, "loss": 0.3823, "step": 19335 }, { "epoch": 0.9780970205877889, "grad_norm": 9.269172668457031, "learning_rate": 1.207298500172649e-08, "loss": 0.5009, "step": 19336 }, { "epoch": 0.9781476048358542, "grad_norm": 4.886490345001221, "learning_rate": 1.2017307352734097e-08, "loss": 0.3087, "step": 19337 }, { "epoch": 0.9781981890839193, "grad_norm": 11.110687255859375, "learning_rate": 1.1961758232007514e-08, "loss": 0.3338, "step": 19338 }, { "epoch": 0.9782487733319845, "grad_norm": 6.64414119720459, "learning_rate": 1.1906337640977817e-08, "loss": 0.3471, "step": 19339 }, { "epoch": 0.9782993575800496, "grad_norm": 6.226258754730225, "learning_rate": 1.185104558107386e-08, "loss": 0.4149, "step": 19340 }, { "epoch": 0.9783499418281147, "grad_norm": 9.450882911682129, "learning_rate": 1.1795882053720064e-08, "loss": 0.4101, "step": 19341 }, { "epoch": 0.9784005260761799, "grad_norm": 7.332769870758057, "learning_rate": 1.1740847060338068e-08, "loss": 0.3303, "step": 19342 }, { "epoch": 0.978451110324245, "grad_norm": 12.755120277404785, "learning_rate": 1.1685940602345625e-08, "loss": 0.4819, "step": 19343 }, { "epoch": 0.9785016945723102, "grad_norm": 10.114134788513184, "learning_rate": 1.1631162681157715e-08, "loss": 0.4118, "step": 19344 }, { "epoch": 0.9785522788203753, "grad_norm": 6.436670780181885, "learning_rate": 1.1576513298185987e-08, "loss": 0.3213, "step": 19345 }, { "epoch": 0.9786028630684405, "grad_norm": 6.423996925354004, "learning_rate": 1.1521992454838205e-08, "loss": 0.3826, "step": 19346 }, { "epoch": 0.9786534473165056, "grad_norm": 6.997116565704346, "learning_rate": 1.1467600152519354e-08, "loss": 0.3699, "step": 19347 }, { "epoch": 0.9787040315645708, "grad_norm": 6.177683353424072, "learning_rate": 1.1413336392631646e-08, "loss": 0.4256, "step": 19348 }, { "epoch": 0.9787546158126359, "grad_norm": 6.557377338409424, "learning_rate": 1.1359201176572854e-08, "loss": 0.341, "step": 19349 }, { "epoch": 0.9788052000607012, "grad_norm": 5.635694980621338, "learning_rate": 1.1305194505737416e-08, "loss": 0.349, "step": 19350 }, { "epoch": 0.9788557843087663, "grad_norm": 5.080268383026123, "learning_rate": 1.1251316381517552e-08, "loss": 0.3103, "step": 19351 }, { "epoch": 0.9789063685568314, "grad_norm": 7.884476661682129, "learning_rate": 1.1197566805301597e-08, "loss": 0.3501, "step": 19352 }, { "epoch": 0.9789569528048966, "grad_norm": 8.229637145996094, "learning_rate": 1.1143945778474552e-08, "loss": 0.4073, "step": 19353 }, { "epoch": 0.9790075370529617, "grad_norm": 5.968844890594482, "learning_rate": 1.1090453302417537e-08, "loss": 0.4418, "step": 19354 }, { "epoch": 0.9790581213010269, "grad_norm": 21.0139217376709, "learning_rate": 1.1037089378509448e-08, "loss": 0.454, "step": 19355 }, { "epoch": 0.979108705549092, "grad_norm": 8.284078598022461, "learning_rate": 1.0983854008125295e-08, "loss": 0.3656, "step": 19356 }, { "epoch": 0.9791592897971572, "grad_norm": 4.3460693359375, "learning_rate": 1.093074719263676e-08, "loss": 0.285, "step": 19357 }, { "epoch": 0.9792098740452223, "grad_norm": 12.89553165435791, "learning_rate": 1.0877768933411637e-08, "loss": 0.5376, "step": 19358 }, { "epoch": 0.9792604582932875, "grad_norm": 13.824451446533203, "learning_rate": 1.082491923181661e-08, "loss": 0.548, "step": 19359 }, { "epoch": 0.9793110425413526, "grad_norm": 7.595553874969482, "learning_rate": 1.077219808921226e-08, "loss": 0.3396, "step": 19360 }, { "epoch": 0.9793616267894177, "grad_norm": 6.605570316314697, "learning_rate": 1.0719605506956942e-08, "loss": 0.4359, "step": 19361 }, { "epoch": 0.9794122110374829, "grad_norm": 8.652205467224121, "learning_rate": 1.0667141486406796e-08, "loss": 0.3803, "step": 19362 }, { "epoch": 0.979462795285548, "grad_norm": 7.225289821624756, "learning_rate": 1.061480602891296e-08, "loss": 0.3885, "step": 19363 }, { "epoch": 0.9795133795336133, "grad_norm": 9.775436401367188, "learning_rate": 1.0562599135824358e-08, "loss": 0.4647, "step": 19364 }, { "epoch": 0.9795639637816784, "grad_norm": 8.337601661682129, "learning_rate": 1.0510520808486024e-08, "loss": 0.3054, "step": 19365 }, { "epoch": 0.9796145480297436, "grad_norm": 12.231637001037598, "learning_rate": 1.0458571048240219e-08, "loss": 0.3186, "step": 19366 }, { "epoch": 0.9796651322778087, "grad_norm": 10.421731948852539, "learning_rate": 1.0406749856424759e-08, "loss": 0.4218, "step": 19367 }, { "epoch": 0.9797157165258739, "grad_norm": 9.241477966308594, "learning_rate": 1.0355057234375799e-08, "loss": 0.3914, "step": 19368 }, { "epoch": 0.979766300773939, "grad_norm": 14.680205345153809, "learning_rate": 1.030349318342505e-08, "loss": 0.454, "step": 19369 }, { "epoch": 0.9798168850220041, "grad_norm": 9.551363945007324, "learning_rate": 1.0252057704900898e-08, "loss": 0.4231, "step": 19370 }, { "epoch": 0.9798674692700693, "grad_norm": 10.760418891906738, "learning_rate": 1.0200750800128944e-08, "loss": 0.3824, "step": 19371 }, { "epoch": 0.9799180535181344, "grad_norm": 11.594204902648926, "learning_rate": 1.0149572470431469e-08, "loss": 0.4891, "step": 19372 }, { "epoch": 0.9799686377661996, "grad_norm": 11.2518892288208, "learning_rate": 1.0098522717126857e-08, "loss": 0.4529, "step": 19373 }, { "epoch": 0.9800192220142647, "grad_norm": 7.199626922607422, "learning_rate": 1.0047601541531281e-08, "loss": 0.3395, "step": 19374 }, { "epoch": 0.9800698062623299, "grad_norm": 5.05342435836792, "learning_rate": 9.996808944955361e-09, "loss": 0.3022, "step": 19375 }, { "epoch": 0.980120390510395, "grad_norm": 12.510676383972168, "learning_rate": 9.946144928709712e-09, "loss": 0.3026, "step": 19376 }, { "epoch": 0.9801709747584603, "grad_norm": 8.325069427490234, "learning_rate": 9.89560949409829e-09, "loss": 0.3445, "step": 19377 }, { "epoch": 0.9802215590065254, "grad_norm": 7.73240327835083, "learning_rate": 9.845202642423946e-09, "loss": 0.2677, "step": 19378 }, { "epoch": 0.9802721432545906, "grad_norm": 9.43742561340332, "learning_rate": 9.794924374986192e-09, "loss": 0.3493, "step": 19379 }, { "epoch": 0.9803227275026557, "grad_norm": 6.341720104217529, "learning_rate": 9.744774693078996e-09, "loss": 0.4067, "step": 19380 }, { "epoch": 0.9803733117507208, "grad_norm": 5.82464075088501, "learning_rate": 9.694753597995766e-09, "loss": 0.3808, "step": 19381 }, { "epoch": 0.980423895998786, "grad_norm": 6.470578193664551, "learning_rate": 9.644861091025471e-09, "loss": 0.3222, "step": 19382 }, { "epoch": 0.9804744802468511, "grad_norm": 10.652944564819336, "learning_rate": 9.595097173453194e-09, "loss": 0.4202, "step": 19383 }, { "epoch": 0.9805250644949163, "grad_norm": 5.702183246612549, "learning_rate": 9.545461846561799e-09, "loss": 0.299, "step": 19384 }, { "epoch": 0.9805756487429814, "grad_norm": 6.03941011428833, "learning_rate": 9.495955111629152e-09, "loss": 0.3143, "step": 19385 }, { "epoch": 0.9806262329910466, "grad_norm": 6.857906818389893, "learning_rate": 9.446576969932009e-09, "loss": 0.293, "step": 19386 }, { "epoch": 0.9806768172391117, "grad_norm": 8.681889533996582, "learning_rate": 9.397327422742686e-09, "loss": 0.3514, "step": 19387 }, { "epoch": 0.9807274014871769, "grad_norm": 6.797067165374756, "learning_rate": 9.348206471329612e-09, "loss": 0.3976, "step": 19388 }, { "epoch": 0.980777985735242, "grad_norm": 9.936490058898926, "learning_rate": 9.299214116958999e-09, "loss": 0.4004, "step": 19389 }, { "epoch": 0.9808285699833073, "grad_norm": 11.641531944274902, "learning_rate": 9.250350360893168e-09, "loss": 0.5374, "step": 19390 }, { "epoch": 0.9808791542313724, "grad_norm": 11.475953102111816, "learning_rate": 9.201615204390557e-09, "loss": 0.5371, "step": 19391 }, { "epoch": 0.9809297384794375, "grad_norm": 6.017721176147461, "learning_rate": 9.15300864870794e-09, "loss": 0.3479, "step": 19392 }, { "epoch": 0.9809803227275027, "grad_norm": 10.858750343322754, "learning_rate": 9.104530695097647e-09, "loss": 0.3522, "step": 19393 }, { "epoch": 0.9810309069755678, "grad_norm": 12.45260238647461, "learning_rate": 9.056181344808124e-09, "loss": 0.4894, "step": 19394 }, { "epoch": 0.981081491223633, "grad_norm": 6.210757732391357, "learning_rate": 9.007960599086152e-09, "loss": 0.3947, "step": 19395 }, { "epoch": 0.9811320754716981, "grad_norm": 7.631178379058838, "learning_rate": 8.959868459173515e-09, "loss": 0.332, "step": 19396 }, { "epoch": 0.9811826597197633, "grad_norm": 7.070681571960449, "learning_rate": 8.911904926309778e-09, "loss": 0.427, "step": 19397 }, { "epoch": 0.9812332439678284, "grad_norm": 9.500732421875, "learning_rate": 8.864070001730617e-09, "loss": 0.4781, "step": 19398 }, { "epoch": 0.9812838282158935, "grad_norm": 8.106389999389648, "learning_rate": 8.816363686668938e-09, "loss": 0.3547, "step": 19399 }, { "epoch": 0.9813344124639587, "grad_norm": 9.182343482971191, "learning_rate": 8.768785982353756e-09, "loss": 0.4092, "step": 19400 }, { "epoch": 0.9813849967120238, "grad_norm": 9.340065956115723, "learning_rate": 8.721336890011312e-09, "loss": 0.3757, "step": 19401 }, { "epoch": 0.981435580960089, "grad_norm": 5.639338493347168, "learning_rate": 8.674016410863961e-09, "loss": 0.3823, "step": 19402 }, { "epoch": 0.9814861652081541, "grad_norm": 11.356459617614746, "learning_rate": 8.626824546131284e-09, "loss": 0.5051, "step": 19403 }, { "epoch": 0.9815367494562194, "grad_norm": 6.490971088409424, "learning_rate": 8.579761297028977e-09, "loss": 0.3551, "step": 19404 }, { "epoch": 0.9815873337042845, "grad_norm": 7.087993621826172, "learning_rate": 8.532826664769955e-09, "loss": 0.4293, "step": 19405 }, { "epoch": 0.9816379179523497, "grad_norm": 7.027667045593262, "learning_rate": 8.486020650563254e-09, "loss": 0.3211, "step": 19406 }, { "epoch": 0.9816885022004148, "grad_norm": 8.355303764343262, "learning_rate": 8.439343255615684e-09, "loss": 0.3814, "step": 19407 }, { "epoch": 0.98173908644848, "grad_norm": 9.153580665588379, "learning_rate": 8.392794481129618e-09, "loss": 0.3512, "step": 19408 }, { "epoch": 0.9817896706965451, "grad_norm": 8.575617790222168, "learning_rate": 8.346374328304652e-09, "loss": 0.428, "step": 19409 }, { "epoch": 0.9818402549446102, "grad_norm": 8.351275444030762, "learning_rate": 8.300082798336495e-09, "loss": 0.3771, "step": 19410 }, { "epoch": 0.9818908391926754, "grad_norm": 6.569543838500977, "learning_rate": 8.253919892418082e-09, "loss": 0.3925, "step": 19411 }, { "epoch": 0.9819414234407405, "grad_norm": 11.016520500183105, "learning_rate": 8.207885611739574e-09, "loss": 0.3457, "step": 19412 }, { "epoch": 0.9819920076888057, "grad_norm": 8.831002235412598, "learning_rate": 8.161979957486688e-09, "loss": 0.4093, "step": 19413 }, { "epoch": 0.9820425919368708, "grad_norm": 6.961845874786377, "learning_rate": 8.11620293084181e-09, "loss": 0.3316, "step": 19414 }, { "epoch": 0.982093176184936, "grad_norm": 8.468172073364258, "learning_rate": 8.07055453298511e-09, "loss": 0.3613, "step": 19415 }, { "epoch": 0.9821437604330011, "grad_norm": 7.667567253112793, "learning_rate": 8.025034765092864e-09, "loss": 0.349, "step": 19416 }, { "epoch": 0.9821943446810664, "grad_norm": 15.094903945922852, "learning_rate": 7.97964362833803e-09, "loss": 0.4782, "step": 19417 }, { "epoch": 0.9822449289291315, "grad_norm": 11.681539535522461, "learning_rate": 7.934381123889667e-09, "loss": 0.4542, "step": 19418 }, { "epoch": 0.9822955131771967, "grad_norm": 10.278837203979492, "learning_rate": 7.889247252914623e-09, "loss": 0.3818, "step": 19419 }, { "epoch": 0.9823460974252618, "grad_norm": 6.065086364746094, "learning_rate": 7.844242016575854e-09, "loss": 0.2943, "step": 19420 }, { "epoch": 0.9823966816733269, "grad_norm": 5.949140548706055, "learning_rate": 7.799365416032989e-09, "loss": 0.4038, "step": 19421 }, { "epoch": 0.9824472659213921, "grad_norm": 5.955976486206055, "learning_rate": 7.754617452441771e-09, "loss": 0.366, "step": 19422 }, { "epoch": 0.9824978501694572, "grad_norm": 6.594294548034668, "learning_rate": 7.709998126956275e-09, "loss": 0.3405, "step": 19423 }, { "epoch": 0.9825484344175224, "grad_norm": 15.69587516784668, "learning_rate": 7.665507440725583e-09, "loss": 0.5225, "step": 19424 }, { "epoch": 0.9825990186655875, "grad_norm": 9.60334587097168, "learning_rate": 7.621145394896556e-09, "loss": 0.3572, "step": 19425 }, { "epoch": 0.9826496029136527, "grad_norm": 15.093338966369629, "learning_rate": 7.576911990611613e-09, "loss": 0.4445, "step": 19426 }, { "epoch": 0.9827001871617178, "grad_norm": 7.931965351104736, "learning_rate": 7.532807229010953e-09, "loss": 0.3608, "step": 19427 }, { "epoch": 0.982750771409783, "grad_norm": 8.238665580749512, "learning_rate": 7.488831111231443e-09, "loss": 0.4049, "step": 19428 }, { "epoch": 0.9828013556578481, "grad_norm": 12.913366317749023, "learning_rate": 7.4449836384055115e-09, "loss": 0.4598, "step": 19429 }, { "epoch": 0.9828519399059134, "grad_norm": 5.872611045837402, "learning_rate": 7.401264811663367e-09, "loss": 0.383, "step": 19430 }, { "epoch": 0.9829025241539785, "grad_norm": 12.165543556213379, "learning_rate": 7.357674632131329e-09, "loss": 0.4421, "step": 19431 }, { "epoch": 0.9829531084020436, "grad_norm": 7.764157295227051, "learning_rate": 7.314213100932943e-09, "loss": 0.3184, "step": 19432 }, { "epoch": 0.9830036926501088, "grad_norm": 8.898752212524414, "learning_rate": 7.2708802191878705e-09, "loss": 0.3833, "step": 19433 }, { "epoch": 0.9830542768981739, "grad_norm": 9.036046028137207, "learning_rate": 7.2276759880129944e-09, "loss": 0.3717, "step": 19434 }, { "epoch": 0.9831048611462391, "grad_norm": 14.535428047180176, "learning_rate": 7.184600408521314e-09, "loss": 0.4103, "step": 19435 }, { "epoch": 0.9831554453943042, "grad_norm": 11.845947265625, "learning_rate": 7.141653481822497e-09, "loss": 0.3479, "step": 19436 }, { "epoch": 0.9832060296423694, "grad_norm": 8.904085159301758, "learning_rate": 7.098835209023991e-09, "loss": 0.3468, "step": 19437 }, { "epoch": 0.9832566138904345, "grad_norm": 11.083992004394531, "learning_rate": 7.056145591228247e-09, "loss": 0.4958, "step": 19438 }, { "epoch": 0.9833071981384996, "grad_norm": 13.326553344726562, "learning_rate": 7.013584629536052e-09, "loss": 0.4535, "step": 19439 }, { "epoch": 0.9833577823865648, "grad_norm": 4.291232109069824, "learning_rate": 6.9711523250431956e-09, "loss": 0.2282, "step": 19440 }, { "epoch": 0.9834083666346299, "grad_norm": 7.2743024826049805, "learning_rate": 6.928848678844358e-09, "loss": 0.4186, "step": 19441 }, { "epoch": 0.9834589508826951, "grad_norm": 6.636331081390381, "learning_rate": 6.8866736920281115e-09, "loss": 0.3894, "step": 19442 }, { "epoch": 0.9835095351307602, "grad_norm": 10.848221778869629, "learning_rate": 6.844627365681921e-09, "loss": 0.3356, "step": 19443 }, { "epoch": 0.9835601193788255, "grad_norm": 15.900792121887207, "learning_rate": 6.8027097008899205e-09, "loss": 0.5521, "step": 19444 }, { "epoch": 0.9836107036268906, "grad_norm": 8.744128227233887, "learning_rate": 6.7609206987306885e-09, "loss": 0.4805, "step": 19445 }, { "epoch": 0.9836612878749558, "grad_norm": 5.799249649047852, "learning_rate": 6.719260360282254e-09, "loss": 0.4067, "step": 19446 }, { "epoch": 0.9837118721230209, "grad_norm": 9.682963371276855, "learning_rate": 6.677728686617646e-09, "loss": 0.471, "step": 19447 }, { "epoch": 0.9837624563710861, "grad_norm": 8.908953666687012, "learning_rate": 6.636325678807121e-09, "loss": 0.4167, "step": 19448 }, { "epoch": 0.9838130406191512, "grad_norm": 6.8121209144592285, "learning_rate": 6.595051337917601e-09, "loss": 0.3399, "step": 19449 }, { "epoch": 0.9838636248672163, "grad_norm": 7.115569591522217, "learning_rate": 6.553905665012128e-09, "loss": 0.3582, "step": 19450 }, { "epoch": 0.9839142091152815, "grad_norm": 6.968055248260498, "learning_rate": 6.512888661152072e-09, "loss": 0.3633, "step": 19451 }, { "epoch": 0.9839647933633466, "grad_norm": 8.813202857971191, "learning_rate": 6.472000327392702e-09, "loss": 0.4884, "step": 19452 }, { "epoch": 0.9840153776114118, "grad_norm": 8.744154930114746, "learning_rate": 6.431240664789284e-09, "loss": 0.3982, "step": 19453 }, { "epoch": 0.9840659618594769, "grad_norm": 8.360174179077148, "learning_rate": 6.390609674390979e-09, "loss": 0.4503, "step": 19454 }, { "epoch": 0.9841165461075421, "grad_norm": 8.492742538452148, "learning_rate": 6.350107357245283e-09, "loss": 0.4503, "step": 19455 }, { "epoch": 0.9841671303556072, "grad_norm": 4.417510032653809, "learning_rate": 6.30973371439525e-09, "loss": 0.2814, "step": 19456 }, { "epoch": 0.9842177146036725, "grad_norm": 6.373264312744141, "learning_rate": 6.269488746881714e-09, "loss": 0.3792, "step": 19457 }, { "epoch": 0.9842682988517376, "grad_norm": 11.590383529663086, "learning_rate": 6.229372455741622e-09, "loss": 0.4579, "step": 19458 }, { "epoch": 0.9843188830998028, "grad_norm": 10.583419799804688, "learning_rate": 6.189384842009149e-09, "loss": 0.3434, "step": 19459 }, { "epoch": 0.9843694673478679, "grad_norm": 10.95360279083252, "learning_rate": 6.149525906713471e-09, "loss": 0.304, "step": 19460 }, { "epoch": 0.984420051595933, "grad_norm": 11.211058616638184, "learning_rate": 6.109795650882655e-09, "loss": 0.3907, "step": 19461 }, { "epoch": 0.9844706358439982, "grad_norm": 9.2714262008667, "learning_rate": 6.070194075540325e-09, "loss": 0.4323, "step": 19462 }, { "epoch": 0.9845212200920633, "grad_norm": 11.456787109375, "learning_rate": 6.0307211817067775e-09, "loss": 0.4661, "step": 19463 }, { "epoch": 0.9845718043401285, "grad_norm": 7.5648956298828125, "learning_rate": 5.991376970398422e-09, "loss": 0.4065, "step": 19464 }, { "epoch": 0.9846223885881936, "grad_norm": 8.460882186889648, "learning_rate": 5.9521614426305556e-09, "loss": 0.314, "step": 19465 }, { "epoch": 0.9846729728362588, "grad_norm": 11.766792297363281, "learning_rate": 5.913074599412372e-09, "loss": 0.4243, "step": 19466 }, { "epoch": 0.9847235570843239, "grad_norm": 7.981015682220459, "learning_rate": 5.874116441751954e-09, "loss": 0.2911, "step": 19467 }, { "epoch": 0.984774141332389, "grad_norm": 9.686995506286621, "learning_rate": 5.835286970652387e-09, "loss": 0.3275, "step": 19468 }, { "epoch": 0.9848247255804542, "grad_norm": 19.250383377075195, "learning_rate": 5.796586187114539e-09, "loss": 0.3978, "step": 19469 }, { "epoch": 0.9848753098285195, "grad_norm": 8.504829406738281, "learning_rate": 5.758014092135944e-09, "loss": 0.427, "step": 19470 }, { "epoch": 0.9849258940765846, "grad_norm": 4.622516632080078, "learning_rate": 5.719570686710252e-09, "loss": 0.2919, "step": 19471 }, { "epoch": 0.9849764783246497, "grad_norm": 8.172926902770996, "learning_rate": 5.681255971827781e-09, "loss": 0.4468, "step": 19472 }, { "epoch": 0.9850270625727149, "grad_norm": 6.580698490142822, "learning_rate": 5.6430699484760765e-09, "loss": 0.3103, "step": 19473 }, { "epoch": 0.98507764682078, "grad_norm": 8.676130294799805, "learning_rate": 5.605012617638794e-09, "loss": 0.4444, "step": 19474 }, { "epoch": 0.9851282310688452, "grad_norm": 8.711555480957031, "learning_rate": 5.5670839802973716e-09, "loss": 0.5239, "step": 19475 }, { "epoch": 0.9851788153169103, "grad_norm": 8.122750282287598, "learning_rate": 5.529284037428806e-09, "loss": 0.3819, "step": 19476 }, { "epoch": 0.9852293995649755, "grad_norm": 5.264587879180908, "learning_rate": 5.491612790006207e-09, "loss": 0.3864, "step": 19477 }, { "epoch": 0.9852799838130406, "grad_norm": 5.655960559844971, "learning_rate": 5.454070239001575e-09, "loss": 0.3534, "step": 19478 }, { "epoch": 0.9853305680611057, "grad_norm": 4.462329864501953, "learning_rate": 5.416656385381358e-09, "loss": 0.2509, "step": 19479 }, { "epoch": 0.9853811523091709, "grad_norm": 6.493712902069092, "learning_rate": 5.379371230110342e-09, "loss": 0.3242, "step": 19480 }, { "epoch": 0.985431736557236, "grad_norm": 8.640116691589355, "learning_rate": 5.3422147741488686e-09, "loss": 0.4163, "step": 19481 }, { "epoch": 0.9854823208053012, "grad_norm": 8.215129852294922, "learning_rate": 5.305187018454505e-09, "loss": 0.4387, "step": 19482 }, { "epoch": 0.9855329050533664, "grad_norm": 9.744939804077148, "learning_rate": 5.2682879639809335e-09, "loss": 0.3592, "step": 19483 }, { "epoch": 0.9855834893014316, "grad_norm": 7.259494304656982, "learning_rate": 5.231517611679615e-09, "loss": 0.4539, "step": 19484 }, { "epoch": 0.9856340735494967, "grad_norm": 5.209656715393066, "learning_rate": 5.19487596249757e-09, "loss": 0.3815, "step": 19485 }, { "epoch": 0.9856846577975619, "grad_norm": 5.7865986824035645, "learning_rate": 5.158363017379042e-09, "loss": 0.269, "step": 19486 }, { "epoch": 0.985735242045627, "grad_norm": 11.244316101074219, "learning_rate": 5.121978777264947e-09, "loss": 0.4801, "step": 19487 }, { "epoch": 0.9857858262936922, "grad_norm": 14.564942359924316, "learning_rate": 5.085723243092866e-09, "loss": 0.4139, "step": 19488 }, { "epoch": 0.9858364105417573, "grad_norm": 5.432196617126465, "learning_rate": 5.049596415797054e-09, "loss": 0.366, "step": 19489 }, { "epoch": 0.9858869947898224, "grad_norm": 7.278010845184326, "learning_rate": 5.013598296307876e-09, "loss": 0.3518, "step": 19490 }, { "epoch": 0.9859375790378876, "grad_norm": 10.137811660766602, "learning_rate": 4.977728885554034e-09, "loss": 0.5054, "step": 19491 }, { "epoch": 0.9859881632859527, "grad_norm": 6.322772979736328, "learning_rate": 4.941988184458679e-09, "loss": 0.3643, "step": 19492 }, { "epoch": 0.9860387475340179, "grad_norm": 23.003002166748047, "learning_rate": 4.906376193943851e-09, "loss": 0.4974, "step": 19493 }, { "epoch": 0.986089331782083, "grad_norm": 8.898008346557617, "learning_rate": 4.87089291492604e-09, "loss": 0.4219, "step": 19494 }, { "epoch": 0.9861399160301482, "grad_norm": 8.017460823059082, "learning_rate": 4.835538348320068e-09, "loss": 0.4051, "step": 19495 }, { "epoch": 0.9861905002782133, "grad_norm": 9.201733589172363, "learning_rate": 4.800312495036874e-09, "loss": 0.4339, "step": 19496 }, { "epoch": 0.9862410845262786, "grad_norm": 10.737010955810547, "learning_rate": 4.76521535598462e-09, "loss": 0.4793, "step": 19497 }, { "epoch": 0.9862916687743437, "grad_norm": 10.587983131408691, "learning_rate": 4.730246932067029e-09, "loss": 0.4791, "step": 19498 }, { "epoch": 0.9863422530224089, "grad_norm": 11.44684886932373, "learning_rate": 4.6954072241856e-09, "loss": 0.3444, "step": 19499 }, { "epoch": 0.986392837270474, "grad_norm": 6.609813690185547, "learning_rate": 4.660696233237394e-09, "loss": 0.3668, "step": 19500 }, { "epoch": 0.9864434215185391, "grad_norm": 7.788243293762207, "learning_rate": 4.626113960117806e-09, "loss": 0.4672, "step": 19501 }, { "epoch": 0.9864940057666043, "grad_norm": 8.15283203125, "learning_rate": 4.591660405717235e-09, "loss": 0.3256, "step": 19502 }, { "epoch": 0.9865445900146694, "grad_norm": 5.512598514556885, "learning_rate": 4.557335570923305e-09, "loss": 0.4184, "step": 19503 }, { "epoch": 0.9865951742627346, "grad_norm": 5.042743682861328, "learning_rate": 4.523139456621417e-09, "loss": 0.3774, "step": 19504 }, { "epoch": 0.9866457585107997, "grad_norm": 9.787410736083984, "learning_rate": 4.489072063691979e-09, "loss": 0.3866, "step": 19505 }, { "epoch": 0.9866963427588649, "grad_norm": 9.550613403320312, "learning_rate": 4.455133393012623e-09, "loss": 0.4792, "step": 19506 }, { "epoch": 0.98674692700693, "grad_norm": 5.1561174392700195, "learning_rate": 4.421323445458759e-09, "loss": 0.3128, "step": 19507 }, { "epoch": 0.9867975112549952, "grad_norm": 8.70837116241455, "learning_rate": 4.3876422219008006e-09, "loss": 0.4677, "step": 19508 }, { "epoch": 0.9868480955030603, "grad_norm": 5.460212707519531, "learning_rate": 4.354089723206944e-09, "loss": 0.3455, "step": 19509 }, { "epoch": 0.9868986797511256, "grad_norm": 4.881412506103516, "learning_rate": 4.320665950241498e-09, "loss": 0.3748, "step": 19510 }, { "epoch": 0.9869492639991907, "grad_norm": 5.121449947357178, "learning_rate": 4.2873709038659946e-09, "loss": 0.2836, "step": 19511 }, { "epoch": 0.9869998482472558, "grad_norm": 17.35517120361328, "learning_rate": 4.254204584938637e-09, "loss": 0.4928, "step": 19512 }, { "epoch": 0.987050432495321, "grad_norm": 5.947054862976074, "learning_rate": 4.221166994313186e-09, "loss": 0.3851, "step": 19513 }, { "epoch": 0.9871010167433861, "grad_norm": 9.586146354675293, "learning_rate": 4.188258132841738e-09, "loss": 0.5484, "step": 19514 }, { "epoch": 0.9871516009914513, "grad_norm": 9.581872940063477, "learning_rate": 4.1554780013719485e-09, "loss": 0.4118, "step": 19515 }, { "epoch": 0.9872021852395164, "grad_norm": 5.560905456542969, "learning_rate": 4.122826600748697e-09, "loss": 0.2884, "step": 19516 }, { "epoch": 0.9872527694875816, "grad_norm": 14.087910652160645, "learning_rate": 4.090303931812977e-09, "loss": 0.4166, "step": 19517 }, { "epoch": 0.9873033537356467, "grad_norm": 13.187736511230469, "learning_rate": 4.057909995403009e-09, "loss": 0.4458, "step": 19518 }, { "epoch": 0.9873539379837118, "grad_norm": 8.071462631225586, "learning_rate": 4.025644792353678e-09, "loss": 0.3572, "step": 19519 }, { "epoch": 0.987404522231777, "grad_norm": 11.943544387817383, "learning_rate": 3.993508323495987e-09, "loss": 0.4489, "step": 19520 }, { "epoch": 0.9874551064798421, "grad_norm": 8.064495086669922, "learning_rate": 3.961500589658163e-09, "loss": 0.3009, "step": 19521 }, { "epoch": 0.9875056907279073, "grad_norm": 5.715054035186768, "learning_rate": 3.9296215916651e-09, "loss": 0.323, "step": 19522 }, { "epoch": 0.9875562749759725, "grad_norm": 9.41757869720459, "learning_rate": 3.897871330338365e-09, "loss": 0.4195, "step": 19523 }, { "epoch": 0.9876068592240377, "grad_norm": 9.411458015441895, "learning_rate": 3.866249806495637e-09, "loss": 0.3769, "step": 19524 }, { "epoch": 0.9876574434721028, "grad_norm": 4.88530158996582, "learning_rate": 3.834757020952374e-09, "loss": 0.3892, "step": 19525 }, { "epoch": 0.987708027720168, "grad_norm": 6.609396457672119, "learning_rate": 3.803392974519038e-09, "loss": 0.339, "step": 19526 }, { "epoch": 0.9877586119682331, "grad_norm": 7.580861568450928, "learning_rate": 3.7721576680049836e-09, "loss": 0.343, "step": 19527 }, { "epoch": 0.9878091962162983, "grad_norm": 8.877069473266602, "learning_rate": 3.741051102214566e-09, "loss": 0.4001, "step": 19528 }, { "epoch": 0.9878597804643634, "grad_norm": 5.485137462615967, "learning_rate": 3.710073277948811e-09, "loss": 0.2932, "step": 19529 }, { "epoch": 0.9879103647124285, "grad_norm": 5.110895156860352, "learning_rate": 3.6792241960065256e-09, "loss": 0.298, "step": 19530 }, { "epoch": 0.9879609489604937, "grad_norm": 6.762024879455566, "learning_rate": 3.648503857183183e-09, "loss": 0.403, "step": 19531 }, { "epoch": 0.9880115332085588, "grad_norm": 8.350312232971191, "learning_rate": 3.617912262269263e-09, "loss": 0.4032, "step": 19532 }, { "epoch": 0.988062117456624, "grad_norm": 7.533241271972656, "learning_rate": 3.5874494120530235e-09, "loss": 0.3649, "step": 19533 }, { "epoch": 0.9881127017046891, "grad_norm": 10.597450256347656, "learning_rate": 3.557115307320502e-09, "loss": 0.4377, "step": 19534 }, { "epoch": 0.9881632859527543, "grad_norm": 10.271239280700684, "learning_rate": 3.526909948852186e-09, "loss": 0.3113, "step": 19535 }, { "epoch": 0.9882138702008194, "grad_norm": 7.534955978393555, "learning_rate": 3.496833337427452e-09, "loss": 0.4015, "step": 19536 }, { "epoch": 0.9882644544488847, "grad_norm": 7.354894161224365, "learning_rate": 3.4668854738206803e-09, "loss": 0.3414, "step": 19537 }, { "epoch": 0.9883150386969498, "grad_norm": 5.212898254394531, "learning_rate": 3.4370663588034757e-09, "loss": 0.2786, "step": 19538 }, { "epoch": 0.988365622945015, "grad_norm": 7.0650787353515625, "learning_rate": 3.407375993144113e-09, "loss": 0.386, "step": 19539 }, { "epoch": 0.9884162071930801, "grad_norm": 7.009572982788086, "learning_rate": 3.3778143776080906e-09, "loss": 0.3976, "step": 19540 }, { "epoch": 0.9884667914411452, "grad_norm": 7.583675384521484, "learning_rate": 3.3483815129570217e-09, "loss": 0.3515, "step": 19541 }, { "epoch": 0.9885173756892104, "grad_norm": 6.857216835021973, "learning_rate": 3.319077399949189e-09, "loss": 0.3614, "step": 19542 }, { "epoch": 0.9885679599372755, "grad_norm": 8.87979507446289, "learning_rate": 3.289902039340098e-09, "loss": 0.4727, "step": 19543 }, { "epoch": 0.9886185441853407, "grad_norm": 9.359600067138672, "learning_rate": 3.260855431880816e-09, "loss": 0.4303, "step": 19544 }, { "epoch": 0.9886691284334058, "grad_norm": 4.6564040184021, "learning_rate": 3.231937578320188e-09, "loss": 0.2712, "step": 19545 }, { "epoch": 0.988719712681471, "grad_norm": 11.749418258666992, "learning_rate": 3.203148479403173e-09, "loss": 0.3923, "step": 19546 }, { "epoch": 0.9887702969295361, "grad_norm": 4.773341655731201, "learning_rate": 3.174488135871956e-09, "loss": 0.4042, "step": 19547 }, { "epoch": 0.9888208811776013, "grad_norm": 5.982208728790283, "learning_rate": 3.1459565484648346e-09, "loss": 0.2992, "step": 19548 }, { "epoch": 0.9888714654256664, "grad_norm": 9.351850509643555, "learning_rate": 3.117553717916777e-09, "loss": 0.3929, "step": 19549 }, { "epoch": 0.9889220496737317, "grad_norm": 6.70554780960083, "learning_rate": 3.08927964496053e-09, "loss": 0.3805, "step": 19550 }, { "epoch": 0.9889726339217968, "grad_norm": 8.490482330322266, "learning_rate": 3.0611343303232897e-09, "loss": 0.3837, "step": 19551 }, { "epoch": 0.9890232181698619, "grad_norm": 8.14871883392334, "learning_rate": 3.0331177747311424e-09, "loss": 0.4036, "step": 19552 }, { "epoch": 0.9890738024179271, "grad_norm": 9.032148361206055, "learning_rate": 3.0052299789062876e-09, "loss": 0.4021, "step": 19553 }, { "epoch": 0.9891243866659922, "grad_norm": 9.118709564208984, "learning_rate": 2.977470943566485e-09, "loss": 0.3698, "step": 19554 }, { "epoch": 0.9891749709140574, "grad_norm": 4.309442043304443, "learning_rate": 2.949840669427273e-09, "loss": 0.346, "step": 19555 }, { "epoch": 0.9892255551621225, "grad_norm": 6.517138957977295, "learning_rate": 2.9223391572014148e-09, "loss": 0.3074, "step": 19556 }, { "epoch": 0.9892761394101877, "grad_norm": 12.784382820129395, "learning_rate": 2.8949664075966775e-09, "loss": 0.5654, "step": 19557 }, { "epoch": 0.9893267236582528, "grad_norm": 6.444453239440918, "learning_rate": 2.8677224213186083e-09, "loss": 0.3624, "step": 19558 }, { "epoch": 0.989377307906318, "grad_norm": 13.664124488830566, "learning_rate": 2.8406071990694228e-09, "loss": 0.4545, "step": 19559 }, { "epoch": 0.9894278921543831, "grad_norm": 7.471401691436768, "learning_rate": 2.8136207415468964e-09, "loss": 0.4511, "step": 19560 }, { "epoch": 0.9894784764024482, "grad_norm": 7.132508277893066, "learning_rate": 2.7867630494476937e-09, "loss": 0.3135, "step": 19561 }, { "epoch": 0.9895290606505134, "grad_norm": 9.598444938659668, "learning_rate": 2.7600341234634844e-09, "loss": 0.4225, "step": 19562 }, { "epoch": 0.9895796448985786, "grad_norm": 7.573831558227539, "learning_rate": 2.7334339642826056e-09, "loss": 0.4273, "step": 19563 }, { "epoch": 0.9896302291466438, "grad_norm": 5.061697483062744, "learning_rate": 2.7069625725911764e-09, "loss": 0.3613, "step": 19564 }, { "epoch": 0.9896808133947089, "grad_norm": 10.181730270385742, "learning_rate": 2.680619949070318e-09, "loss": 0.4451, "step": 19565 }, { "epoch": 0.9897313976427741, "grad_norm": 9.201386451721191, "learning_rate": 2.6544060943994865e-09, "loss": 0.427, "step": 19566 }, { "epoch": 0.9897819818908392, "grad_norm": 6.723944187164307, "learning_rate": 2.6283210092542533e-09, "loss": 0.3654, "step": 19567 }, { "epoch": 0.9898325661389044, "grad_norm": 9.304980278015137, "learning_rate": 2.602364694306303e-09, "loss": 0.4435, "step": 19568 }, { "epoch": 0.9898831503869695, "grad_norm": 8.222046852111816, "learning_rate": 2.576537150224545e-09, "loss": 0.344, "step": 19569 }, { "epoch": 0.9899337346350346, "grad_norm": 11.183884620666504, "learning_rate": 2.5508383776751134e-09, "loss": 0.4034, "step": 19570 }, { "epoch": 0.9899843188830998, "grad_norm": 7.953573703765869, "learning_rate": 2.5252683773197004e-09, "loss": 0.2639, "step": 19571 }, { "epoch": 0.9900349031311649, "grad_norm": 7.327794551849365, "learning_rate": 2.4998271498166694e-09, "loss": 0.4014, "step": 19572 }, { "epoch": 0.9900854873792301, "grad_norm": 7.299095630645752, "learning_rate": 2.4745146958227162e-09, "loss": 0.4213, "step": 19573 }, { "epoch": 0.9901360716272952, "grad_norm": 4.372272491455078, "learning_rate": 2.449331015988987e-09, "loss": 0.3052, "step": 19574 }, { "epoch": 0.9901866558753604, "grad_norm": 10.180352210998535, "learning_rate": 2.4242761109649625e-09, "loss": 0.4147, "step": 19575 }, { "epoch": 0.9902372401234255, "grad_norm": 8.364418029785156, "learning_rate": 2.3993499813967924e-09, "loss": 0.3218, "step": 19576 }, { "epoch": 0.9902878243714908, "grad_norm": 5.886218547821045, "learning_rate": 2.3745526279256305e-09, "loss": 0.3212, "step": 19577 }, { "epoch": 0.9903384086195559, "grad_norm": 18.524940490722656, "learning_rate": 2.349884051190965e-09, "loss": 0.378, "step": 19578 }, { "epoch": 0.990388992867621, "grad_norm": 9.299140930175781, "learning_rate": 2.3253442518289537e-09, "loss": 0.3841, "step": 19579 }, { "epoch": 0.9904395771156862, "grad_norm": 10.41025161743164, "learning_rate": 2.3009332304707588e-09, "loss": 0.3964, "step": 19580 }, { "epoch": 0.9904901613637513, "grad_norm": 8.245359420776367, "learning_rate": 2.2766509877464317e-09, "loss": 0.327, "step": 19581 }, { "epoch": 0.9905407456118165, "grad_norm": 10.387393951416016, "learning_rate": 2.2524975242810277e-09, "loss": 0.3482, "step": 19582 }, { "epoch": 0.9905913298598816, "grad_norm": 8.552586555480957, "learning_rate": 2.228472840697382e-09, "loss": 0.4732, "step": 19583 }, { "epoch": 0.9906419141079468, "grad_norm": 6.597099304199219, "learning_rate": 2.2045769376144445e-09, "loss": 0.4587, "step": 19584 }, { "epoch": 0.9906924983560119, "grad_norm": 12.094186782836914, "learning_rate": 2.180809815647833e-09, "loss": 0.3446, "step": 19585 }, { "epoch": 0.9907430826040771, "grad_norm": 13.46458625793457, "learning_rate": 2.157171475410391e-09, "loss": 0.4385, "step": 19586 }, { "epoch": 0.9907936668521422, "grad_norm": 8.43636703491211, "learning_rate": 2.133661917510521e-09, "loss": 0.3452, "step": 19587 }, { "epoch": 0.9908442511002074, "grad_norm": 6.249320030212402, "learning_rate": 2.1102811425538494e-09, "loss": 0.3974, "step": 19588 }, { "epoch": 0.9908948353482725, "grad_norm": 5.941747665405273, "learning_rate": 2.0870291511437825e-09, "loss": 0.357, "step": 19589 }, { "epoch": 0.9909454195963378, "grad_norm": 5.999048709869385, "learning_rate": 2.063905943879285e-09, "loss": 0.3965, "step": 19590 }, { "epoch": 0.9909960038444029, "grad_norm": 5.4865193367004395, "learning_rate": 2.0409115213554375e-09, "loss": 0.3454, "step": 19591 }, { "epoch": 0.991046588092468, "grad_norm": 10.62409496307373, "learning_rate": 2.0180458841656538e-09, "loss": 0.4304, "step": 19592 }, { "epoch": 0.9910971723405332, "grad_norm": 6.919830322265625, "learning_rate": 1.9953090328983514e-09, "loss": 0.438, "step": 19593 }, { "epoch": 0.9911477565885983, "grad_norm": 8.696751594543457, "learning_rate": 1.972700968139729e-09, "loss": 0.436, "step": 19594 }, { "epoch": 0.9911983408366635, "grad_norm": 10.35924243927002, "learning_rate": 1.9502216904726533e-09, "loss": 0.3178, "step": 19595 }, { "epoch": 0.9912489250847286, "grad_norm": 7.680621147155762, "learning_rate": 1.9278712004755505e-09, "loss": 0.4085, "step": 19596 }, { "epoch": 0.9912995093327938, "grad_norm": 7.307011604309082, "learning_rate": 1.9056494987251817e-09, "loss": 0.3063, "step": 19597 }, { "epoch": 0.9913500935808589, "grad_norm": 4.766814231872559, "learning_rate": 1.8835565857938664e-09, "loss": 0.4121, "step": 19598 }, { "epoch": 0.991400677828924, "grad_norm": 7.867628574371338, "learning_rate": 1.8615924622511495e-09, "loss": 0.3991, "step": 19599 }, { "epoch": 0.9914512620769892, "grad_norm": 7.907273769378662, "learning_rate": 1.839757128662134e-09, "loss": 0.5303, "step": 19600 }, { "epoch": 0.9915018463250543, "grad_norm": 6.917148113250732, "learning_rate": 1.8180505855902587e-09, "loss": 0.432, "step": 19601 }, { "epoch": 0.9915524305731195, "grad_norm": 13.657177925109863, "learning_rate": 1.7964728335945203e-09, "loss": 0.5551, "step": 19602 }, { "epoch": 0.9916030148211847, "grad_norm": 16.519474029541016, "learning_rate": 1.775023873231141e-09, "loss": 0.562, "step": 19603 }, { "epoch": 0.9916535990692499, "grad_norm": 7.731197357177734, "learning_rate": 1.7537037050524564e-09, "loss": 0.4582, "step": 19604 }, { "epoch": 0.991704183317315, "grad_norm": 9.607821464538574, "learning_rate": 1.7325123296080271e-09, "loss": 0.4408, "step": 19605 }, { "epoch": 0.9917547675653802, "grad_norm": 9.0523099899292, "learning_rate": 1.7114497474440827e-09, "loss": 0.4285, "step": 19606 }, { "epoch": 0.9918053518134453, "grad_norm": 6.354797840118408, "learning_rate": 1.6905159591035226e-09, "loss": 0.456, "step": 19607 }, { "epoch": 0.9918559360615105, "grad_norm": 7.340194225311279, "learning_rate": 1.6697109651253595e-09, "loss": 0.3756, "step": 19608 }, { "epoch": 0.9919065203095756, "grad_norm": 8.959179878234863, "learning_rate": 1.6490347660452766e-09, "loss": 0.3159, "step": 19609 }, { "epoch": 0.9919571045576407, "grad_norm": 6.310252666473389, "learning_rate": 1.6284873623967356e-09, "loss": 0.3871, "step": 19610 }, { "epoch": 0.9920076888057059, "grad_norm": 5.176197528839111, "learning_rate": 1.6080687547093132e-09, "loss": 0.2617, "step": 19611 }, { "epoch": 0.992058273053771, "grad_norm": 7.151811599731445, "learning_rate": 1.5877789435086999e-09, "loss": 0.423, "step": 19612 }, { "epoch": 0.9921088573018362, "grad_norm": 13.397876739501953, "learning_rate": 1.5676179293178106e-09, "loss": 0.5439, "step": 19613 }, { "epoch": 0.9921594415499013, "grad_norm": 10.345965385437012, "learning_rate": 1.54758571265623e-09, "loss": 0.4875, "step": 19614 }, { "epoch": 0.9922100257979665, "grad_norm": 6.183649063110352, "learning_rate": 1.5276822940396564e-09, "loss": 0.2995, "step": 19615 }, { "epoch": 0.9922606100460316, "grad_norm": 6.579170227050781, "learning_rate": 1.507907673982123e-09, "loss": 0.3986, "step": 19616 }, { "epoch": 0.9923111942940969, "grad_norm": 8.550459861755371, "learning_rate": 1.4882618529921122e-09, "loss": 0.4251, "step": 19617 }, { "epoch": 0.992361778542162, "grad_norm": 4.374020099639893, "learning_rate": 1.4687448315758856e-09, "loss": 0.3836, "step": 19618 }, { "epoch": 0.9924123627902272, "grad_norm": 5.23560094833374, "learning_rate": 1.4493566102369294e-09, "loss": 0.3841, "step": 19619 }, { "epoch": 0.9924629470382923, "grad_norm": 7.867752552032471, "learning_rate": 1.4300971894748439e-09, "loss": 0.378, "step": 19620 }, { "epoch": 0.9925135312863574, "grad_norm": 12.427191734313965, "learning_rate": 1.4109665697853437e-09, "loss": 0.3641, "step": 19621 }, { "epoch": 0.9925641155344226, "grad_norm": 14.968421936035156, "learning_rate": 1.3919647516613677e-09, "loss": 0.4101, "step": 19622 }, { "epoch": 0.9926146997824877, "grad_norm": 6.641011714935303, "learning_rate": 1.3730917355930794e-09, "loss": 0.2212, "step": 19623 }, { "epoch": 0.9926652840305529, "grad_norm": 4.569472789764404, "learning_rate": 1.3543475220662017e-09, "loss": 0.2887, "step": 19624 }, { "epoch": 0.992715868278618, "grad_norm": 9.879034042358398, "learning_rate": 1.3357321115647914e-09, "loss": 0.4273, "step": 19625 }, { "epoch": 0.9927664525266832, "grad_norm": 7.589504718780518, "learning_rate": 1.3172455045673549e-09, "loss": 0.3495, "step": 19626 }, { "epoch": 0.9928170367747483, "grad_norm": 11.188285827636719, "learning_rate": 1.2988877015501778e-09, "loss": 0.3889, "step": 19627 }, { "epoch": 0.9928676210228135, "grad_norm": 7.000720977783203, "learning_rate": 1.2806587029873251e-09, "loss": 0.3003, "step": 19628 }, { "epoch": 0.9929182052708786, "grad_norm": 8.596169471740723, "learning_rate": 1.2625585093484216e-09, "loss": 0.4506, "step": 19629 }, { "epoch": 0.9929687895189439, "grad_norm": 14.342334747314453, "learning_rate": 1.2445871210986505e-09, "loss": 0.3476, "step": 19630 }, { "epoch": 0.993019373767009, "grad_norm": 5.01922082901001, "learning_rate": 1.226744538702085e-09, "loss": 0.2545, "step": 19631 }, { "epoch": 0.9930699580150741, "grad_norm": 5.181753635406494, "learning_rate": 1.2090307626183573e-09, "loss": 0.3728, "step": 19632 }, { "epoch": 0.9931205422631393, "grad_norm": 8.674505233764648, "learning_rate": 1.1914457933037692e-09, "loss": 0.3842, "step": 19633 }, { "epoch": 0.9931711265112044, "grad_norm": 4.064835548400879, "learning_rate": 1.173989631211847e-09, "loss": 0.3044, "step": 19634 }, { "epoch": 0.9932217107592696, "grad_norm": 6.375316619873047, "learning_rate": 1.1566622767916758e-09, "loss": 0.3384, "step": 19635 }, { "epoch": 0.9932722950073347, "grad_norm": 12.481748580932617, "learning_rate": 1.1394637304901202e-09, "loss": 0.5948, "step": 19636 }, { "epoch": 0.9933228792553999, "grad_norm": 11.802810668945312, "learning_rate": 1.1223939927507143e-09, "loss": 0.5742, "step": 19637 }, { "epoch": 0.993373463503465, "grad_norm": 6.880537033081055, "learning_rate": 1.1054530640125515e-09, "loss": 0.3891, "step": 19638 }, { "epoch": 0.9934240477515301, "grad_norm": 8.18400764465332, "learning_rate": 1.0886409447125045e-09, "loss": 0.442, "step": 19639 }, { "epoch": 0.9934746319995953, "grad_norm": 8.570788383483887, "learning_rate": 1.0719576352835604e-09, "loss": 0.388, "step": 19640 }, { "epoch": 0.9935252162476604, "grad_norm": 12.335351943969727, "learning_rate": 1.0554031361564855e-09, "loss": 0.4006, "step": 19641 }, { "epoch": 0.9935758004957256, "grad_norm": 9.801029205322266, "learning_rate": 1.0389774477564952e-09, "loss": 0.4982, "step": 19642 }, { "epoch": 0.9936263847437908, "grad_norm": 19.031124114990234, "learning_rate": 1.0226805705082498e-09, "loss": 0.3607, "step": 19643 }, { "epoch": 0.993676968991856, "grad_norm": 10.190988540649414, "learning_rate": 1.006512504830859e-09, "loss": 0.3949, "step": 19644 }, { "epoch": 0.9937275532399211, "grad_norm": 13.147748947143555, "learning_rate": 9.904732511406556e-10, "loss": 0.4071, "step": 19645 }, { "epoch": 0.9937781374879863, "grad_norm": 8.021328926086426, "learning_rate": 9.745628098517534e-10, "loss": 0.2999, "step": 19646 }, { "epoch": 0.9938287217360514, "grad_norm": 18.988811492919922, "learning_rate": 9.587811813732695e-10, "loss": 0.5305, "step": 19647 }, { "epoch": 0.9938793059841166, "grad_norm": 9.999281883239746, "learning_rate": 9.431283661126556e-10, "loss": 0.5075, "step": 19648 }, { "epoch": 0.9939298902321817, "grad_norm": 9.250595092773438, "learning_rate": 9.276043644729227e-10, "loss": 0.4317, "step": 19649 }, { "epoch": 0.9939804744802468, "grad_norm": 12.995043754577637, "learning_rate": 9.122091768537511e-10, "loss": 0.5284, "step": 19650 }, { "epoch": 0.994031058728312, "grad_norm": 8.087748527526855, "learning_rate": 8.969428036526006e-10, "loss": 0.4316, "step": 19651 }, { "epoch": 0.9940816429763771, "grad_norm": 19.26307487487793, "learning_rate": 8.818052452619352e-10, "loss": 0.4195, "step": 19652 }, { "epoch": 0.9941322272244423, "grad_norm": 13.78926944732666, "learning_rate": 8.667965020725533e-10, "loss": 0.5971, "step": 19653 }, { "epoch": 0.9941828114725074, "grad_norm": 8.025328636169434, "learning_rate": 8.519165744708125e-10, "loss": 0.3788, "step": 19654 }, { "epoch": 0.9942333957205726, "grad_norm": 11.515413284301758, "learning_rate": 8.371654628402948e-10, "loss": 0.5021, "step": 19655 }, { "epoch": 0.9942839799686378, "grad_norm": 10.479616165161133, "learning_rate": 8.225431675612516e-10, "loss": 0.3887, "step": 19656 }, { "epoch": 0.994334564216703, "grad_norm": 15.624188423156738, "learning_rate": 8.080496890100486e-10, "loss": 0.5818, "step": 19657 }, { "epoch": 0.9943851484647681, "grad_norm": 4.829556465148926, "learning_rate": 7.936850275602758e-10, "loss": 0.264, "step": 19658 }, { "epoch": 0.9944357327128333, "grad_norm": 12.140303611755371, "learning_rate": 7.794491835821927e-10, "loss": 0.5216, "step": 19659 }, { "epoch": 0.9944863169608984, "grad_norm": 6.423495769500732, "learning_rate": 7.653421574427278e-10, "loss": 0.3645, "step": 19660 }, { "epoch": 0.9945369012089635, "grad_norm": 6.564516067504883, "learning_rate": 7.513639495049241e-10, "loss": 0.4129, "step": 19661 }, { "epoch": 0.9945874854570287, "grad_norm": 6.135956764221191, "learning_rate": 7.375145601296041e-10, "loss": 0.3793, "step": 19662 }, { "epoch": 0.9946380697050938, "grad_norm": 8.867734909057617, "learning_rate": 7.237939896731494e-10, "loss": 0.4199, "step": 19663 }, { "epoch": 0.994688653953159, "grad_norm": 7.52063512802124, "learning_rate": 7.10202238489166e-10, "loss": 0.2708, "step": 19664 }, { "epoch": 0.9947392382012241, "grad_norm": 14.31255054473877, "learning_rate": 6.967393069279293e-10, "loss": 0.582, "step": 19665 }, { "epoch": 0.9947898224492893, "grad_norm": 12.69304084777832, "learning_rate": 6.834051953369391e-10, "loss": 0.5198, "step": 19666 }, { "epoch": 0.9948404066973544, "grad_norm": 9.223296165466309, "learning_rate": 6.701999040586993e-10, "loss": 0.5036, "step": 19667 }, { "epoch": 0.9948909909454196, "grad_norm": 19.238693237304688, "learning_rate": 6.571234334340481e-10, "loss": 0.5569, "step": 19668 }, { "epoch": 0.9949415751934847, "grad_norm": 6.994840145111084, "learning_rate": 6.441757838004936e-10, "loss": 0.2866, "step": 19669 }, { "epoch": 0.99499215944155, "grad_norm": 7.549431800842285, "learning_rate": 6.313569554905474e-10, "loss": 0.3684, "step": 19670 }, { "epoch": 0.9950427436896151, "grad_norm": 6.870878219604492, "learning_rate": 6.186669488350561e-10, "loss": 0.3561, "step": 19671 }, { "epoch": 0.9950933279376802, "grad_norm": 6.214236259460449, "learning_rate": 6.061057641609802e-10, "loss": 0.3214, "step": 19672 }, { "epoch": 0.9951439121857454, "grad_norm": 8.468488693237305, "learning_rate": 5.93673401792505e-10, "loss": 0.3618, "step": 19673 }, { "epoch": 0.9951944964338105, "grad_norm": 7.711973190307617, "learning_rate": 5.813698620488195e-10, "loss": 0.3609, "step": 19674 }, { "epoch": 0.9952450806818757, "grad_norm": 7.799794673919678, "learning_rate": 5.691951452480026e-10, "loss": 0.4322, "step": 19675 }, { "epoch": 0.9952956649299408, "grad_norm": 11.040660858154297, "learning_rate": 5.571492517031374e-10, "loss": 0.406, "step": 19676 }, { "epoch": 0.995346249178006, "grad_norm": 13.0048828125, "learning_rate": 5.452321817250861e-10, "loss": 0.4373, "step": 19677 }, { "epoch": 0.9953968334260711, "grad_norm": 5.307572364807129, "learning_rate": 5.334439356208254e-10, "loss": 0.4713, "step": 19678 }, { "epoch": 0.9954474176741362, "grad_norm": 9.646868705749512, "learning_rate": 5.217845136940014e-10, "loss": 0.3844, "step": 19679 }, { "epoch": 0.9954980019222014, "grad_norm": 10.252823829650879, "learning_rate": 5.102539162449294e-10, "loss": 0.6198, "step": 19680 }, { "epoch": 0.9955485861702665, "grad_norm": 7.699992656707764, "learning_rate": 4.98852143570594e-10, "loss": 0.4602, "step": 19681 }, { "epoch": 0.9955991704183317, "grad_norm": 8.814010620117188, "learning_rate": 4.875791959652043e-10, "loss": 0.4141, "step": 19682 }, { "epoch": 0.9956497546663969, "grad_norm": 7.395798683166504, "learning_rate": 4.764350737190837e-10, "loss": 0.4248, "step": 19683 }, { "epoch": 0.9957003389144621, "grad_norm": 5.9326090812683105, "learning_rate": 4.6541977711977994e-10, "loss": 0.289, "step": 19684 }, { "epoch": 0.9957509231625272, "grad_norm": 6.922445297241211, "learning_rate": 4.5453330645039986e-10, "loss": 0.4464, "step": 19685 }, { "epoch": 0.9958015074105924, "grad_norm": 5.482725143432617, "learning_rate": 4.4377566199182987e-10, "loss": 0.4186, "step": 19686 }, { "epoch": 0.9958520916586575, "grad_norm": 6.8731160163879395, "learning_rate": 4.331468440210707e-10, "loss": 0.3121, "step": 19687 }, { "epoch": 0.9959026759067227, "grad_norm": 11.363448143005371, "learning_rate": 4.226468528123473e-10, "loss": 0.4333, "step": 19688 }, { "epoch": 0.9959532601547878, "grad_norm": 7.114771366119385, "learning_rate": 4.1227568863544396e-10, "loss": 0.386, "step": 19689 }, { "epoch": 0.996003844402853, "grad_norm": 7.316120624542236, "learning_rate": 4.0203335175903466e-10, "loss": 0.3371, "step": 19690 }, { "epoch": 0.9960544286509181, "grad_norm": 13.603958129882812, "learning_rate": 3.9191984244568716e-10, "loss": 0.3183, "step": 19691 }, { "epoch": 0.9961050128989832, "grad_norm": 9.470294952392578, "learning_rate": 3.819351609568589e-10, "loss": 0.272, "step": 19692 }, { "epoch": 0.9961555971470484, "grad_norm": 7.368398666381836, "learning_rate": 3.7207930754901147e-10, "loss": 0.3563, "step": 19693 }, { "epoch": 0.9962061813951135, "grad_norm": 6.80376672744751, "learning_rate": 3.6235228247694097e-10, "loss": 0.4272, "step": 19694 }, { "epoch": 0.9962567656431787, "grad_norm": 8.786437034606934, "learning_rate": 3.527540859910028e-10, "loss": 0.3416, "step": 19695 }, { "epoch": 0.9963073498912439, "grad_norm": 9.59249210357666, "learning_rate": 3.432847183382215e-10, "loss": 0.4364, "step": 19696 }, { "epoch": 0.9963579341393091, "grad_norm": 6.345753192901611, "learning_rate": 3.3394417976284623e-10, "loss": 0.334, "step": 19697 }, { "epoch": 0.9964085183873742, "grad_norm": 8.498372077941895, "learning_rate": 3.247324705052401e-10, "loss": 0.3821, "step": 19698 }, { "epoch": 0.9964591026354394, "grad_norm": 8.998209953308105, "learning_rate": 3.156495908035462e-10, "loss": 0.4441, "step": 19699 }, { "epoch": 0.9965096868835045, "grad_norm": 7.230479717254639, "learning_rate": 3.0669554089091113e-10, "loss": 0.3963, "step": 19700 }, { "epoch": 0.9965602711315696, "grad_norm": 6.506391525268555, "learning_rate": 2.978703209988165e-10, "loss": 0.3407, "step": 19701 }, { "epoch": 0.9966108553796348, "grad_norm": 8.92983627319336, "learning_rate": 2.8917393135430293e-10, "loss": 0.3305, "step": 19702 }, { "epoch": 0.9966614396276999, "grad_norm": 9.390610694885254, "learning_rate": 2.8060637218108036e-10, "loss": 0.4365, "step": 19703 }, { "epoch": 0.9967120238757651, "grad_norm": 5.622677803039551, "learning_rate": 2.721676437006382e-10, "loss": 0.299, "step": 19704 }, { "epoch": 0.9967626081238302, "grad_norm": 13.255183219909668, "learning_rate": 2.6385774613002513e-10, "loss": 0.3948, "step": 19705 }, { "epoch": 0.9968131923718954, "grad_norm": 9.030228614807129, "learning_rate": 2.5567667968351417e-10, "loss": 0.4893, "step": 19706 }, { "epoch": 0.9968637766199605, "grad_norm": 9.51986312866211, "learning_rate": 2.476244445714926e-10, "loss": 0.471, "step": 19707 }, { "epoch": 0.9969143608680257, "grad_norm": 5.876556873321533, "learning_rate": 2.3970104100212723e-10, "loss": 0.3174, "step": 19708 }, { "epoch": 0.9969649451160908, "grad_norm": 5.91949462890625, "learning_rate": 2.3190646917914395e-10, "loss": 0.3516, "step": 19709 }, { "epoch": 0.997015529364156, "grad_norm": 7.064004421234131, "learning_rate": 2.2424072930349315e-10, "loss": 0.3632, "step": 19710 }, { "epoch": 0.9970661136122212, "grad_norm": 6.846835613250732, "learning_rate": 2.1670382157279457e-10, "loss": 0.3589, "step": 19711 }, { "epoch": 0.9971166978602863, "grad_norm": 10.777107238769531, "learning_rate": 2.092957461813372e-10, "loss": 0.4004, "step": 19712 }, { "epoch": 0.9971672821083515, "grad_norm": 6.466316223144531, "learning_rate": 2.0201650331952428e-10, "loss": 0.4106, "step": 19713 }, { "epoch": 0.9972178663564166, "grad_norm": 7.380489349365234, "learning_rate": 1.948660931749835e-10, "loss": 0.4126, "step": 19714 }, { "epoch": 0.9972684506044818, "grad_norm": 5.960371017456055, "learning_rate": 1.8784451593256704e-10, "loss": 0.3454, "step": 19715 }, { "epoch": 0.9973190348525469, "grad_norm": 7.602784156799316, "learning_rate": 1.8095177177268608e-10, "loss": 0.3908, "step": 19716 }, { "epoch": 0.9973696191006121, "grad_norm": 8.236384391784668, "learning_rate": 1.7418786087297636e-10, "loss": 0.3214, "step": 19717 }, { "epoch": 0.9974202033486772, "grad_norm": 7.919238090515137, "learning_rate": 1.6755278340829794e-10, "loss": 0.3852, "step": 19718 }, { "epoch": 0.9974707875967423, "grad_norm": 7.528511047363281, "learning_rate": 1.6104653954907014e-10, "loss": 0.3094, "step": 19719 }, { "epoch": 0.9975213718448075, "grad_norm": 21.51213264465332, "learning_rate": 1.5466912946293654e-10, "loss": 0.4086, "step": 19720 }, { "epoch": 0.9975719560928726, "grad_norm": 5.94663143157959, "learning_rate": 1.4842055331421024e-10, "loss": 0.3973, "step": 19721 }, { "epoch": 0.9976225403409378, "grad_norm": 5.667537212371826, "learning_rate": 1.423008112638735e-10, "loss": 0.3377, "step": 19722 }, { "epoch": 0.997673124589003, "grad_norm": 6.645705699920654, "learning_rate": 1.363099034701332e-10, "loss": 0.3837, "step": 19723 }, { "epoch": 0.9977237088370682, "grad_norm": 5.943778038024902, "learning_rate": 1.304478300873102e-10, "loss": 0.3986, "step": 19724 }, { "epoch": 0.9977742930851333, "grad_norm": 7.4239959716796875, "learning_rate": 1.2471459126583984e-10, "loss": 0.3862, "step": 19725 }, { "epoch": 0.9978248773331985, "grad_norm": 9.696239471435547, "learning_rate": 1.1911018715338175e-10, "loss": 0.4833, "step": 19726 }, { "epoch": 0.9978754615812636, "grad_norm": 7.518631935119629, "learning_rate": 1.1363461789537511e-10, "loss": 0.3095, "step": 19727 }, { "epoch": 0.9979260458293288, "grad_norm": 7.031556606292725, "learning_rate": 1.0828788363170806e-10, "loss": 0.3317, "step": 19728 }, { "epoch": 0.9979766300773939, "grad_norm": 12.327831268310547, "learning_rate": 1.0306998450115845e-10, "loss": 0.3872, "step": 19729 }, { "epoch": 0.998027214325459, "grad_norm": 4.588497638702393, "learning_rate": 9.79809206375082e-11, "loss": 0.2822, "step": 19730 }, { "epoch": 0.9980777985735242, "grad_norm": 12.172142028808594, "learning_rate": 9.302069217231868e-11, "loss": 0.4237, "step": 19731 }, { "epoch": 0.9981283828215893, "grad_norm": 8.390605926513672, "learning_rate": 8.818929923271047e-11, "loss": 0.3796, "step": 19732 }, { "epoch": 0.9981789670696545, "grad_norm": 8.75407600402832, "learning_rate": 8.348674194413876e-11, "loss": 0.3883, "step": 19733 }, { "epoch": 0.9982295513177196, "grad_norm": 10.31982707977295, "learning_rate": 7.891302042761784e-11, "loss": 0.3436, "step": 19734 }, { "epoch": 0.9982801355657848, "grad_norm": 11.592142105102539, "learning_rate": 7.446813480027626e-11, "loss": 0.3748, "step": 19735 }, { "epoch": 0.99833071981385, "grad_norm": 6.973923683166504, "learning_rate": 7.01520851770221e-11, "loss": 0.4145, "step": 19736 }, { "epoch": 0.9983813040619152, "grad_norm": 8.031963348388672, "learning_rate": 6.596487166887767e-11, "loss": 0.4832, "step": 19737 }, { "epoch": 0.9984318883099803, "grad_norm": 7.354263782501221, "learning_rate": 6.190649438408969e-11, "loss": 0.3367, "step": 19738 }, { "epoch": 0.9984824725580455, "grad_norm": 11.429455757141113, "learning_rate": 5.7976953427574255e-11, "loss": 0.4008, "step": 19739 }, { "epoch": 0.9985330568061106, "grad_norm": 9.427903175354004, "learning_rate": 5.4176248899806546e-11, "loss": 0.3934, "step": 19740 }, { "epoch": 0.9985836410541757, "grad_norm": 9.556221961975098, "learning_rate": 5.05043808990413e-11, "loss": 0.4742, "step": 19741 }, { "epoch": 0.9986342253022409, "grad_norm": 9.356616020202637, "learning_rate": 4.696134952020259e-11, "loss": 0.42, "step": 19742 }, { "epoch": 0.998684809550306, "grad_norm": 12.163960456848145, "learning_rate": 4.354715485377359e-11, "loss": 0.4061, "step": 19743 }, { "epoch": 0.9987353937983712, "grad_norm": 9.059205055236816, "learning_rate": 4.0261796988572133e-11, "loss": 0.4124, "step": 19744 }, { "epoch": 0.9987859780464363, "grad_norm": 5.682053565979004, "learning_rate": 3.710527600842007e-11, "loss": 0.3469, "step": 19745 }, { "epoch": 0.9988365622945015, "grad_norm": 7.393243312835693, "learning_rate": 3.4077591996029e-11, "loss": 0.3495, "step": 19746 }, { "epoch": 0.9988871465425666, "grad_norm": 7.018219470977783, "learning_rate": 3.117874502800433e-11, "loss": 0.2792, "step": 19747 }, { "epoch": 0.9989377307906318, "grad_norm": 8.945494651794434, "learning_rate": 2.8408735179286107e-11, "loss": 0.4339, "step": 19748 }, { "epoch": 0.9989883150386969, "grad_norm": 7.589117050170898, "learning_rate": 2.5767562522038825e-11, "loss": 0.4259, "step": 19749 }, { "epoch": 0.9990388992867621, "grad_norm": 6.3488922119140625, "learning_rate": 2.325522712343098e-11, "loss": 0.4018, "step": 19750 }, { "epoch": 0.9990894835348273, "grad_norm": 5.530544281005859, "learning_rate": 2.0871729048965726e-11, "loss": 0.3525, "step": 19751 }, { "epoch": 0.9991400677828924, "grad_norm": 8.33525276184082, "learning_rate": 1.8617068359705338e-11, "loss": 0.4651, "step": 19752 }, { "epoch": 0.9991906520309576, "grad_norm": 7.112458229064941, "learning_rate": 1.6491245113381405e-11, "loss": 0.3408, "step": 19753 }, { "epoch": 0.9992412362790227, "grad_norm": 13.034390449523926, "learning_rate": 1.4494259365505082e-11, "loss": 0.4382, "step": 19754 }, { "epoch": 0.9992918205270879, "grad_norm": 7.286001682281494, "learning_rate": 1.2626111166591514e-11, "loss": 0.3615, "step": 19755 }, { "epoch": 0.999342404775153, "grad_norm": 9.333253860473633, "learning_rate": 1.0886800565490518e-11, "loss": 0.4314, "step": 19756 }, { "epoch": 0.9993929890232182, "grad_norm": 3.922921895980835, "learning_rate": 9.276327607166125e-12, "loss": 0.2862, "step": 19757 }, { "epoch": 0.9994435732712833, "grad_norm": 12.959921836853027, "learning_rate": 7.794692332696584e-12, "loss": 0.4574, "step": 19758 }, { "epoch": 0.9994941575193484, "grad_norm": 7.012989044189453, "learning_rate": 6.4418947803845945e-12, "loss": 0.3118, "step": 19759 }, { "epoch": 0.9995447417674136, "grad_norm": 11.619281768798828, "learning_rate": 5.217934985202178e-12, "loss": 0.4415, "step": 19760 }, { "epoch": 0.9995953260154787, "grad_norm": 7.091416835784912, "learning_rate": 4.1228129782355795e-12, "loss": 0.4137, "step": 19761 }, { "epoch": 0.9996459102635439, "grad_norm": 6.963791370391846, "learning_rate": 3.1565287877954874e-12, "loss": 0.4008, "step": 19762 }, { "epoch": 0.9996964945116091, "grad_norm": 7.77681303024292, "learning_rate": 2.3190824394170307e-12, "loss": 0.4032, "step": 19763 }, { "epoch": 0.9997470787596743, "grad_norm": 8.42581844329834, "learning_rate": 1.610473954194447e-12, "loss": 0.3855, "step": 19764 }, { "epoch": 0.9997976630077394, "grad_norm": 5.425844192504883, "learning_rate": 1.0307033510015275e-12, "loss": 0.2941, "step": 19765 }, { "epoch": 0.9998482472558046, "grad_norm": 7.033786296844482, "learning_rate": 5.7977064371606e-13, "loss": 0.4182, "step": 19766 }, { "epoch": 0.9998988315038697, "grad_norm": 5.40846586227417, "learning_rate": 2.5767584455049787e-13, "loss": 0.3702, "step": 19767 }, { "epoch": 0.9999494157519349, "grad_norm": 6.95597505569458, "learning_rate": 6.441896127640234e-14, "loss": 0.3401, "step": 19768 }, { "epoch": 1.0, "grad_norm": 10.48205280303955, "learning_rate": 0.0, "loss": 0.4768, "step": 19769 }, { "epoch": 1.0, "step": 19769, "total_flos": 2.7821163581779476e+18, "train_loss": 0.43958344443523745, "train_runtime": 19412.8643, "train_samples_per_second": 8.147, "train_steps_per_second": 1.018 } ], "logging_steps": 1, "max_steps": 19769, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 19769, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2.7821163581779476e+18, "train_batch_size": 1, "trial_name": null, "trial_params": null }