diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,4767 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 2.995008319467554, + "eval_steps": 500, + "global_step": 675, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.004437049362174155, + "grad_norm": 0.0, + "learning_rate": 0, + "loss": 10.6311, + "step": 1 + }, + { + "epoch": 0.00887409872434831, + "grad_norm": 0.0, + "learning_rate": 0, + "loss": 10.6677, + "step": 2 + }, + { + "epoch": 0.013311148086522463, + "grad_norm": 0.0, + "learning_rate": 0, + "loss": 10.6365, + "step": 3 + }, + { + "epoch": 0.01774819744869662, + "grad_norm": 0.0, + "learning_rate": 0, + "loss": 10.7214, + "step": 4 + }, + { + "epoch": 0.022185246810870772, + "grad_norm": 0.0, + "learning_rate": 0, + "loss": 10.5715, + "step": 5 + }, + { + "epoch": 0.026622296173044926, + "grad_norm": 0.0, + "learning_rate": 0, + "loss": 10.6467, + "step": 6 + }, + { + "epoch": 0.03105934553521908, + "grad_norm": 0.0, + "learning_rate": 0, + "loss": 10.5833, + "step": 7 + }, + { + "epoch": 0.03549639489739324, + "grad_norm": 0.0, + "learning_rate": 0, + "loss": 10.8142, + "step": 8 + }, + { + "epoch": 0.03993344425956739, + "grad_norm": 0.0, + "learning_rate": 0, + "loss": 10.7166, + "step": 9 + }, + { + "epoch": 0.044370493621741544, + "grad_norm": 0.0, + "learning_rate": 0, + "loss": 10.6414, + "step": 10 + }, + { + "epoch": 0.048807542983915694, + "grad_norm": 0.0, + "learning_rate": 0, + "loss": 10.571, + "step": 11 + }, + { + "epoch": 0.05324459234608985, + "grad_norm": 21158.767578125, + "learning_rate": 0.0, + "loss": 10.5605, + "step": 12 + }, + { + "epoch": 0.057681641708264, + "grad_norm": 5464.2412109375, + "learning_rate": 7.352941176470589e-07, + "loss": 9.6161, + "step": 13 + }, + { + "epoch": 0.06211869107043816, + "grad_norm": 10006.7900390625, + "learning_rate": 1.4705882352941177e-06, + "loss": 9.7633, + "step": 14 + }, + { + "epoch": 0.06655574043261231, + "grad_norm": 3737.953125, + "learning_rate": 2.2058823529411767e-06, + "loss": 9.7728, + "step": 15 + }, + { + "epoch": 0.07099278979478647, + "grad_norm": 3249.61279296875, + "learning_rate": 2.9411764705882355e-06, + "loss": 8.9384, + "step": 16 + }, + { + "epoch": 0.07542983915696062, + "grad_norm": 4874.1328125, + "learning_rate": 3.6764705882352942e-06, + "loss": 8.4148, + "step": 17 + }, + { + "epoch": 0.07986688851913477, + "grad_norm": 7288.25732421875, + "learning_rate": 4.411764705882353e-06, + "loss": 7.8928, + "step": 18 + }, + { + "epoch": 0.08430393788130892, + "grad_norm": 2073.17822265625, + "learning_rate": 5.147058823529412e-06, + "loss": 8.2863, + "step": 19 + }, + { + "epoch": 0.08874098724348309, + "grad_norm": 1633.4429931640625, + "learning_rate": 5.882352941176471e-06, + "loss": 7.6505, + "step": 20 + }, + { + "epoch": 0.09317803660565724, + "grad_norm": 2719.80078125, + "learning_rate": 6.61764705882353e-06, + "loss": 7.7203, + "step": 21 + }, + { + "epoch": 0.09761508596783139, + "grad_norm": 2726.412353515625, + "learning_rate": 7.3529411764705884e-06, + "loss": 7.3999, + "step": 22 + }, + { + "epoch": 0.10205213533000555, + "grad_norm": 1712.63720703125, + "learning_rate": 8.088235294117648e-06, + "loss": 7.1295, + "step": 23 + }, + { + "epoch": 0.1064891846921797, + "grad_norm": 1713.2330322265625, + "learning_rate": 8.823529411764707e-06, + "loss": 6.8872, + "step": 24 + }, + { + "epoch": 0.11092623405435385, + "grad_norm": 3012.1806640625, + "learning_rate": 9.558823529411764e-06, + "loss": 6.8458, + "step": 25 + }, + { + "epoch": 0.115363283416528, + "grad_norm": 1599.78857421875, + "learning_rate": 1.0294117647058824e-05, + "loss": 6.7284, + "step": 26 + }, + { + "epoch": 0.11980033277870217, + "grad_norm": 2026.845947265625, + "learning_rate": 1.1029411764705883e-05, + "loss": 6.8857, + "step": 27 + }, + { + "epoch": 0.12423738214087632, + "grad_norm": 2789.587158203125, + "learning_rate": 1.1764705882352942e-05, + "loss": 6.5732, + "step": 28 + }, + { + "epoch": 0.12867443150305047, + "grad_norm": 1993.5609130859375, + "learning_rate": 1.25e-05, + "loss": 6.5739, + "step": 29 + }, + { + "epoch": 0.13311148086522462, + "grad_norm": 2042.6405029296875, + "learning_rate": 1.323529411764706e-05, + "loss": 6.3652, + "step": 30 + }, + { + "epoch": 0.13754853022739877, + "grad_norm": 1727.5252685546875, + "learning_rate": 1.3970588235294118e-05, + "loss": 6.5057, + "step": 31 + }, + { + "epoch": 0.14198557958957295, + "grad_norm": 3566.0869140625, + "learning_rate": 1.4705882352941177e-05, + "loss": 6.4298, + "step": 32 + }, + { + "epoch": 0.1464226289517471, + "grad_norm": 1188.673583984375, + "learning_rate": 1.5441176470588237e-05, + "loss": 6.4531, + "step": 33 + }, + { + "epoch": 0.15085967831392125, + "grad_norm": 3024.59033203125, + "learning_rate": 1.6176470588235296e-05, + "loss": 6.3073, + "step": 34 + }, + { + "epoch": 0.1552967276760954, + "grad_norm": 3680.47412109375, + "learning_rate": 1.6911764705882355e-05, + "loss": 6.2053, + "step": 35 + }, + { + "epoch": 0.15973377703826955, + "grad_norm": 1899.4468994140625, + "learning_rate": 1.7647058823529414e-05, + "loss": 6.2939, + "step": 36 + }, + { + "epoch": 0.1641708264004437, + "grad_norm": 2961.06005859375, + "learning_rate": 1.8382352941176472e-05, + "loss": 6.3336, + "step": 37 + }, + { + "epoch": 0.16860787576261785, + "grad_norm": 3534.0048828125, + "learning_rate": 1.9117647058823528e-05, + "loss": 6.3052, + "step": 38 + }, + { + "epoch": 0.17304492512479203, + "grad_norm": 3085.997802734375, + "learning_rate": 1.9852941176470586e-05, + "loss": 6.2321, + "step": 39 + }, + { + "epoch": 0.17748197448696618, + "grad_norm": 2578.245361328125, + "learning_rate": 2.058823529411765e-05, + "loss": 6.2852, + "step": 40 + }, + { + "epoch": 0.18191902384914033, + "grad_norm": 2776.39599609375, + "learning_rate": 2.1323529411764707e-05, + "loss": 6.2452, + "step": 41 + }, + { + "epoch": 0.18635607321131448, + "grad_norm": 1784.9794921875, + "learning_rate": 2.2058823529411766e-05, + "loss": 6.151, + "step": 42 + }, + { + "epoch": 0.19079312257348863, + "grad_norm": 1190.688232421875, + "learning_rate": 2.2794117647058825e-05, + "loss": 6.0287, + "step": 43 + }, + { + "epoch": 0.19523017193566278, + "grad_norm": 1785.841796875, + "learning_rate": 2.3529411764705884e-05, + "loss": 6.0673, + "step": 44 + }, + { + "epoch": 0.19966722129783693, + "grad_norm": 4042.77001953125, + "learning_rate": 2.4264705882352942e-05, + "loss": 6.0552, + "step": 45 + }, + { + "epoch": 0.2041042706600111, + "grad_norm": 4556.90087890625, + "learning_rate": 2.5e-05, + "loss": 5.9113, + "step": 46 + }, + { + "epoch": 0.20854132002218526, + "grad_norm": 3234.686767578125, + "learning_rate": 2.5735294117647057e-05, + "loss": 6.0116, + "step": 47 + }, + { + "epoch": 0.2129783693843594, + "grad_norm": 4687.42724609375, + "learning_rate": 2.647058823529412e-05, + "loss": 5.8229, + "step": 48 + }, + { + "epoch": 0.21741541874653356, + "grad_norm": 2991.71923828125, + "learning_rate": 2.7205882352941174e-05, + "loss": 5.6743, + "step": 49 + }, + { + "epoch": 0.2218524681087077, + "grad_norm": 4608.99267578125, + "learning_rate": 2.7941176470588236e-05, + "loss": 5.6744, + "step": 50 + }, + { + "epoch": 0.22628951747088186, + "grad_norm": 10194.2744140625, + "learning_rate": 2.8676470588235295e-05, + "loss": 5.1944, + "step": 51 + }, + { + "epoch": 0.230726566833056, + "grad_norm": 10194.2744140625, + "learning_rate": 2.8676470588235295e-05, + "loss": 5.3622, + "step": 52 + }, + { + "epoch": 0.23516361619523019, + "grad_norm": 6200.15087890625, + "learning_rate": 2.9411764705882354e-05, + "loss": 5.5519, + "step": 53 + }, + { + "epoch": 0.23960066555740434, + "grad_norm": 18477.029296875, + "learning_rate": 3.0147058823529413e-05, + "loss": 5.4745, + "step": 54 + }, + { + "epoch": 0.24403771491957849, + "grad_norm": 15055.43359375, + "learning_rate": 3.0882352941176475e-05, + "loss": 5.4999, + "step": 55 + }, + { + "epoch": 0.24847476428175264, + "grad_norm": 5037.6669921875, + "learning_rate": 3.161764705882353e-05, + "loss": 5.2596, + "step": 56 + }, + { + "epoch": 0.2529118136439268, + "grad_norm": 32858.02734375, + "learning_rate": 3.235294117647059e-05, + "loss": 5.7379, + "step": 57 + }, + { + "epoch": 0.25734886300610094, + "grad_norm": 15704.654296875, + "learning_rate": 3.308823529411765e-05, + "loss": 5.7607, + "step": 58 + }, + { + "epoch": 0.2617859123682751, + "grad_norm": 10505.4189453125, + "learning_rate": 3.382352941176471e-05, + "loss": 5.3833, + "step": 59 + }, + { + "epoch": 0.26622296173044924, + "grad_norm": 12585.240234375, + "learning_rate": 3.455882352941177e-05, + "loss": 5.589, + "step": 60 + }, + { + "epoch": 0.2706600110926234, + "grad_norm": 10240.61328125, + "learning_rate": 3.529411764705883e-05, + "loss": 5.3518, + "step": 61 + }, + { + "epoch": 0.27509706045479754, + "grad_norm": 7670.6865234375, + "learning_rate": 3.6029411764705886e-05, + "loss": 5.1818, + "step": 62 + }, + { + "epoch": 0.2795341098169717, + "grad_norm": 5819.0947265625, + "learning_rate": 3.6764705882352945e-05, + "loss": 5.4395, + "step": 63 + }, + { + "epoch": 0.2839711591791459, + "grad_norm": 12336.7509765625, + "learning_rate": 3.7500000000000003e-05, + "loss": 5.4409, + "step": 64 + }, + { + "epoch": 0.28840820854132004, + "grad_norm": 10684.744140625, + "learning_rate": 3.8235294117647055e-05, + "loss": 5.5206, + "step": 65 + }, + { + "epoch": 0.2928452579034942, + "grad_norm": 3992.02978515625, + "learning_rate": 3.897058823529412e-05, + "loss": 5.1411, + "step": 66 + }, + { + "epoch": 0.29728230726566834, + "grad_norm": 5053.01416015625, + "learning_rate": 3.970588235294117e-05, + "loss": 5.1205, + "step": 67 + }, + { + "epoch": 0.3017193566278425, + "grad_norm": 12579.82421875, + "learning_rate": 4.044117647058824e-05, + "loss": 4.9203, + "step": 68 + }, + { + "epoch": 0.30615640599001664, + "grad_norm": 11802.810546875, + "learning_rate": 4.11764705882353e-05, + "loss": 5.1967, + "step": 69 + }, + { + "epoch": 0.3105934553521908, + "grad_norm": 11713.0302734375, + "learning_rate": 4.1911764705882356e-05, + "loss": 5.0146, + "step": 70 + }, + { + "epoch": 0.31503050471436495, + "grad_norm": 8684.224609375, + "learning_rate": 4.2647058823529415e-05, + "loss": 5.024, + "step": 71 + }, + { + "epoch": 0.3194675540765391, + "grad_norm": 4284.21533203125, + "learning_rate": 4.3382352941176474e-05, + "loss": 4.9771, + "step": 72 + }, + { + "epoch": 0.32390460343871325, + "grad_norm": 12809.7451171875, + "learning_rate": 4.411764705882353e-05, + "loss": 4.9245, + "step": 73 + }, + { + "epoch": 0.3283416528008874, + "grad_norm": 7180.11328125, + "learning_rate": 4.485294117647059e-05, + "loss": 4.8355, + "step": 74 + }, + { + "epoch": 0.33277870216306155, + "grad_norm": 6717.80810546875, + "learning_rate": 4.558823529411765e-05, + "loss": 4.9867, + "step": 75 + }, + { + "epoch": 0.3372157515252357, + "grad_norm": 7746.44970703125, + "learning_rate": 4.632352941176471e-05, + "loss": 4.934, + "step": 76 + }, + { + "epoch": 0.34165280088740985, + "grad_norm": 9719.7236328125, + "learning_rate": 4.705882352941177e-05, + "loss": 4.8041, + "step": 77 + }, + { + "epoch": 0.34608985024958405, + "grad_norm": 7071.28125, + "learning_rate": 4.7794117647058826e-05, + "loss": 4.8701, + "step": 78 + }, + { + "epoch": 0.3505268996117582, + "grad_norm": 4220.2353515625, + "learning_rate": 4.8529411764705885e-05, + "loss": 4.9001, + "step": 79 + }, + { + "epoch": 0.35496394897393235, + "grad_norm": 14180.1337890625, + "learning_rate": 4.9264705882352944e-05, + "loss": 4.5012, + "step": 80 + }, + { + "epoch": 0.3594009983361065, + "grad_norm": 9157.7890625, + "learning_rate": 5e-05, + "loss": 5.1653, + "step": 81 + }, + { + "epoch": 0.36383804769828065, + "grad_norm": 15036.89453125, + "learning_rate": 4.99176276771005e-05, + "loss": 4.8085, + "step": 82 + }, + { + "epoch": 0.3682750970604548, + "grad_norm": 7805.9140625, + "learning_rate": 4.9835255354200987e-05, + "loss": 4.9078, + "step": 83 + }, + { + "epoch": 0.37271214642262895, + "grad_norm": 5143.04052734375, + "learning_rate": 4.975288303130148e-05, + "loss": 4.4639, + "step": 84 + }, + { + "epoch": 0.3771491957848031, + "grad_norm": 14249.8447265625, + "learning_rate": 4.967051070840198e-05, + "loss": 4.4569, + "step": 85 + }, + { + "epoch": 0.38158624514697725, + "grad_norm": 9558.1298828125, + "learning_rate": 4.958813838550247e-05, + "loss": 4.6703, + "step": 86 + }, + { + "epoch": 0.3860232945091514, + "grad_norm": 7875.93603515625, + "learning_rate": 4.950576606260297e-05, + "loss": 4.4127, + "step": 87 + }, + { + "epoch": 0.39046034387132555, + "grad_norm": 9563.3369140625, + "learning_rate": 4.9423393739703464e-05, + "loss": 4.5327, + "step": 88 + }, + { + "epoch": 0.3948973932334997, + "grad_norm": 4175.3662109375, + "learning_rate": 4.934102141680396e-05, + "loss": 4.3455, + "step": 89 + }, + { + "epoch": 0.39933444259567386, + "grad_norm": 4653.71826171875, + "learning_rate": 4.9258649093904455e-05, + "loss": 4.608, + "step": 90 + }, + { + "epoch": 0.403771491957848, + "grad_norm": 4601.45263671875, + "learning_rate": 4.9176276771004943e-05, + "loss": 4.3481, + "step": 91 + }, + { + "epoch": 0.4082085413200222, + "grad_norm": 5737.9658203125, + "learning_rate": 4.909390444810544e-05, + "loss": 4.336, + "step": 92 + }, + { + "epoch": 0.41264559068219636, + "grad_norm": 5043.4365234375, + "learning_rate": 4.9011532125205934e-05, + "loss": 4.3632, + "step": 93 + }, + { + "epoch": 0.4170826400443705, + "grad_norm": 5378.66357421875, + "learning_rate": 4.892915980230643e-05, + "loss": 4.2322, + "step": 94 + }, + { + "epoch": 0.42151968940654466, + "grad_norm": 4413.9052734375, + "learning_rate": 4.884678747940692e-05, + "loss": 4.2327, + "step": 95 + }, + { + "epoch": 0.4259567387687188, + "grad_norm": 3674.305908203125, + "learning_rate": 4.8764415156507414e-05, + "loss": 4.3429, + "step": 96 + }, + { + "epoch": 0.43039378813089296, + "grad_norm": 4777.6318359375, + "learning_rate": 4.868204283360791e-05, + "loss": 4.2922, + "step": 97 + }, + { + "epoch": 0.4348308374930671, + "grad_norm": 4025.486572265625, + "learning_rate": 4.8599670510708405e-05, + "loss": 4.3369, + "step": 98 + }, + { + "epoch": 0.43926788685524126, + "grad_norm": 3471.095458984375, + "learning_rate": 4.8517298187808894e-05, + "loss": 4.391, + "step": 99 + }, + { + "epoch": 0.4437049362174154, + "grad_norm": 4713.88671875, + "learning_rate": 4.843492586490939e-05, + "loss": 4.176, + "step": 100 + }, + { + "epoch": 0.44814198557958956, + "grad_norm": 3923.6435546875, + "learning_rate": 4.8352553542009884e-05, + "loss": 4.2135, + "step": 101 + }, + { + "epoch": 0.4525790349417637, + "grad_norm": 6652.15380859375, + "learning_rate": 4.827018121911038e-05, + "loss": 4.4733, + "step": 102 + }, + { + "epoch": 0.45701608430393786, + "grad_norm": 8517.3251953125, + "learning_rate": 4.8187808896210875e-05, + "loss": 4.2726, + "step": 103 + }, + { + "epoch": 0.461453133666112, + "grad_norm": 6766.1650390625, + "learning_rate": 4.810543657331137e-05, + "loss": 4.0555, + "step": 104 + }, + { + "epoch": 0.46589018302828616, + "grad_norm": 2664.34814453125, + "learning_rate": 4.8023064250411866e-05, + "loss": 4.1152, + "step": 105 + }, + { + "epoch": 0.47032723239046037, + "grad_norm": 3303.956787109375, + "learning_rate": 4.794069192751236e-05, + "loss": 4.0515, + "step": 106 + }, + { + "epoch": 0.4747642817526345, + "grad_norm": 5097.65185546875, + "learning_rate": 4.785831960461285e-05, + "loss": 4.1409, + "step": 107 + }, + { + "epoch": 0.47920133111480867, + "grad_norm": 10082.732421875, + "learning_rate": 4.7775947281713346e-05, + "loss": 4.1343, + "step": 108 + }, + { + "epoch": 0.4836383804769828, + "grad_norm": 3584.152587890625, + "learning_rate": 4.769357495881384e-05, + "loss": 4.0524, + "step": 109 + }, + { + "epoch": 0.48807542983915697, + "grad_norm": 5950.09765625, + "learning_rate": 4.761120263591434e-05, + "loss": 4.075, + "step": 110 + }, + { + "epoch": 0.4925124792013311, + "grad_norm": 6068.30126953125, + "learning_rate": 4.7528830313014825e-05, + "loss": 4.2378, + "step": 111 + }, + { + "epoch": 0.49694952856350527, + "grad_norm": 6259.79541015625, + "learning_rate": 4.744645799011532e-05, + "loss": 4.1694, + "step": 112 + }, + { + "epoch": 0.5013865779256794, + "grad_norm": 2736.747802734375, + "learning_rate": 4.7364085667215816e-05, + "loss": 4.0974, + "step": 113 + }, + { + "epoch": 0.5058236272878536, + "grad_norm": 3057.271728515625, + "learning_rate": 4.728171334431631e-05, + "loss": 4.2289, + "step": 114 + }, + { + "epoch": 0.5102606766500277, + "grad_norm": 3766.69287109375, + "learning_rate": 4.719934102141681e-05, + "loss": 4.1914, + "step": 115 + }, + { + "epoch": 0.5146977260122019, + "grad_norm": 2897.723388671875, + "learning_rate": 4.71169686985173e-05, + "loss": 4.0673, + "step": 116 + }, + { + "epoch": 0.519134775374376, + "grad_norm": 3029.3095703125, + "learning_rate": 4.70345963756178e-05, + "loss": 3.9072, + "step": 117 + }, + { + "epoch": 0.5235718247365502, + "grad_norm": 3487.687255859375, + "learning_rate": 4.6952224052718294e-05, + "loss": 4.0699, + "step": 118 + }, + { + "epoch": 0.5280088740987243, + "grad_norm": 8178.2412109375, + "learning_rate": 4.686985172981878e-05, + "loss": 3.9756, + "step": 119 + }, + { + "epoch": 0.5324459234608985, + "grad_norm": 2972.537109375, + "learning_rate": 4.678747940691928e-05, + "loss": 3.9783, + "step": 120 + }, + { + "epoch": 0.5368829728230726, + "grad_norm": 3646.947998046875, + "learning_rate": 4.670510708401977e-05, + "loss": 3.813, + "step": 121 + }, + { + "epoch": 0.5413200221852468, + "grad_norm": 3125.714111328125, + "learning_rate": 4.662273476112027e-05, + "loss": 3.9225, + "step": 122 + }, + { + "epoch": 0.5457570715474209, + "grad_norm": 4453.3525390625, + "learning_rate": 4.654036243822076e-05, + "loss": 3.9049, + "step": 123 + }, + { + "epoch": 0.5501941209095951, + "grad_norm": 2585.39111328125, + "learning_rate": 4.645799011532125e-05, + "loss": 3.9358, + "step": 124 + }, + { + "epoch": 0.5546311702717692, + "grad_norm": 3983.77197265625, + "learning_rate": 4.637561779242175e-05, + "loss": 3.8939, + "step": 125 + }, + { + "epoch": 0.5590682196339434, + "grad_norm": 4210.779296875, + "learning_rate": 4.6293245469522244e-05, + "loss": 4.0253, + "step": 126 + }, + { + "epoch": 0.5635052689961176, + "grad_norm": 5125.97265625, + "learning_rate": 4.621087314662273e-05, + "loss": 4.2263, + "step": 127 + }, + { + "epoch": 0.5679423183582918, + "grad_norm": 3248.380126953125, + "learning_rate": 4.612850082372323e-05, + "loss": 4.0164, + "step": 128 + }, + { + "epoch": 0.5723793677204659, + "grad_norm": 4200.4560546875, + "learning_rate": 4.6046128500823723e-05, + "loss": 3.9244, + "step": 129 + }, + { + "epoch": 0.5768164170826401, + "grad_norm": 4828.31103515625, + "learning_rate": 4.596375617792422e-05, + "loss": 4.217, + "step": 130 + }, + { + "epoch": 0.5812534664448142, + "grad_norm": 2958.339111328125, + "learning_rate": 4.5881383855024714e-05, + "loss": 3.881, + "step": 131 + }, + { + "epoch": 0.5856905158069884, + "grad_norm": 3084.78759765625, + "learning_rate": 4.579901153212521e-05, + "loss": 3.8977, + "step": 132 + }, + { + "epoch": 0.5901275651691625, + "grad_norm": 5373.04052734375, + "learning_rate": 4.5716639209225705e-05, + "loss": 4.0824, + "step": 133 + }, + { + "epoch": 0.5945646145313367, + "grad_norm": 2756.171142578125, + "learning_rate": 4.56342668863262e-05, + "loss": 4.0205, + "step": 134 + }, + { + "epoch": 0.5990016638935108, + "grad_norm": 2161.1416015625, + "learning_rate": 4.555189456342669e-05, + "loss": 4.1388, + "step": 135 + }, + { + "epoch": 0.603438713255685, + "grad_norm": 3423.720458984375, + "learning_rate": 4.5469522240527185e-05, + "loss": 3.818, + "step": 136 + }, + { + "epoch": 0.6078757626178591, + "grad_norm": 3028.1962890625, + "learning_rate": 4.538714991762768e-05, + "loss": 3.9562, + "step": 137 + }, + { + "epoch": 0.6123128119800333, + "grad_norm": 2784.5029296875, + "learning_rate": 4.5304777594728176e-05, + "loss": 4.0464, + "step": 138 + }, + { + "epoch": 0.6167498613422074, + "grad_norm": 4660.537109375, + "learning_rate": 4.5222405271828664e-05, + "loss": 3.911, + "step": 139 + }, + { + "epoch": 0.6211869107043816, + "grad_norm": 3196.64990234375, + "learning_rate": 4.514003294892916e-05, + "loss": 3.9998, + "step": 140 + }, + { + "epoch": 0.6256239600665557, + "grad_norm": 4897.91162109375, + "learning_rate": 4.5057660626029655e-05, + "loss": 3.7104, + "step": 141 + }, + { + "epoch": 0.6300610094287299, + "grad_norm": 4805.7783203125, + "learning_rate": 4.497528830313015e-05, + "loss": 4.0719, + "step": 142 + }, + { + "epoch": 0.634498058790904, + "grad_norm": 3324.002197265625, + "learning_rate": 4.4892915980230646e-05, + "loss": 3.9034, + "step": 143 + }, + { + "epoch": 0.6389351081530782, + "grad_norm": 2894.477294921875, + "learning_rate": 4.481054365733114e-05, + "loss": 3.8019, + "step": 144 + }, + { + "epoch": 0.6433721575152523, + "grad_norm": 2887.784423828125, + "learning_rate": 4.472817133443164e-05, + "loss": 3.83, + "step": 145 + }, + { + "epoch": 0.6478092068774265, + "grad_norm": 2073.251953125, + "learning_rate": 4.464579901153213e-05, + "loss": 4.0059, + "step": 146 + }, + { + "epoch": 0.6522462562396006, + "grad_norm": 3794.48779296875, + "learning_rate": 4.456342668863262e-05, + "loss": 4.1077, + "step": 147 + }, + { + "epoch": 0.6566833056017748, + "grad_norm": 2190.56884765625, + "learning_rate": 4.448105436573312e-05, + "loss": 3.944, + "step": 148 + }, + { + "epoch": 0.6611203549639489, + "grad_norm": 6018.3173828125, + "learning_rate": 4.439868204283361e-05, + "loss": 3.8172, + "step": 149 + }, + { + "epoch": 0.6655574043261231, + "grad_norm": 3491.772705078125, + "learning_rate": 4.43163097199341e-05, + "loss": 4.1328, + "step": 150 + }, + { + "epoch": 0.6699944536882972, + "grad_norm": 5739.01513671875, + "learning_rate": 4.4233937397034596e-05, + "loss": 3.7836, + "step": 151 + }, + { + "epoch": 0.6744315030504714, + "grad_norm": 6953.44873046875, + "learning_rate": 4.415156507413509e-05, + "loss": 4.0114, + "step": 152 + }, + { + "epoch": 0.6788685524126455, + "grad_norm": 2166.23193359375, + "learning_rate": 4.406919275123559e-05, + "loss": 3.9442, + "step": 153 + }, + { + "epoch": 0.6833056017748197, + "grad_norm": 2447.5927734375, + "learning_rate": 4.3986820428336076e-05, + "loss": 3.8227, + "step": 154 + }, + { + "epoch": 0.687742651136994, + "grad_norm": 4474.73193359375, + "learning_rate": 4.390444810543657e-05, + "loss": 3.6613, + "step": 155 + }, + { + "epoch": 0.6921797004991681, + "grad_norm": 7070.38037109375, + "learning_rate": 4.382207578253707e-05, + "loss": 3.9871, + "step": 156 + }, + { + "epoch": 0.6966167498613423, + "grad_norm": 3446.41552734375, + "learning_rate": 4.373970345963756e-05, + "loss": 3.8815, + "step": 157 + }, + { + "epoch": 0.7010537992235164, + "grad_norm": 2785.808349609375, + "learning_rate": 4.365733113673806e-05, + "loss": 3.7899, + "step": 158 + }, + { + "epoch": 0.7054908485856906, + "grad_norm": 4404.37060546875, + "learning_rate": 4.357495881383855e-05, + "loss": 3.88, + "step": 159 + }, + { + "epoch": 0.7099278979478647, + "grad_norm": 3911.151611328125, + "learning_rate": 4.349258649093905e-05, + "loss": 3.8406, + "step": 160 + }, + { + "epoch": 0.7143649473100389, + "grad_norm": 3653.130859375, + "learning_rate": 4.3410214168039544e-05, + "loss": 3.9048, + "step": 161 + }, + { + "epoch": 0.718801996672213, + "grad_norm": 6438.55517578125, + "learning_rate": 4.332784184514003e-05, + "loss": 3.8763, + "step": 162 + }, + { + "epoch": 0.7232390460343872, + "grad_norm": 2494.703369140625, + "learning_rate": 4.324546952224053e-05, + "loss": 3.8052, + "step": 163 + }, + { + "epoch": 0.7276760953965613, + "grad_norm": 3728.2138671875, + "learning_rate": 4.3163097199341024e-05, + "loss": 3.9276, + "step": 164 + }, + { + "epoch": 0.7321131447587355, + "grad_norm": 3183.252685546875, + "learning_rate": 4.308072487644152e-05, + "loss": 3.8512, + "step": 165 + }, + { + "epoch": 0.7365501941209096, + "grad_norm": 7026.63916015625, + "learning_rate": 4.299835255354201e-05, + "loss": 4.0595, + "step": 166 + }, + { + "epoch": 0.7409872434830838, + "grad_norm": 4156.95068359375, + "learning_rate": 4.29159802306425e-05, + "loss": 3.7067, + "step": 167 + }, + { + "epoch": 0.7454242928452579, + "grad_norm": 3471.735595703125, + "learning_rate": 4.2833607907743e-05, + "loss": 3.6877, + "step": 168 + }, + { + "epoch": 0.7498613422074321, + "grad_norm": 2331.02587890625, + "learning_rate": 4.2751235584843494e-05, + "loss": 3.9505, + "step": 169 + }, + { + "epoch": 0.7542983915696062, + "grad_norm": 9993.0654296875, + "learning_rate": 4.266886326194399e-05, + "loss": 4.058, + "step": 170 + }, + { + "epoch": 0.7587354409317804, + "grad_norm": 8045.14453125, + "learning_rate": 4.2586490939044485e-05, + "loss": 3.6175, + "step": 171 + }, + { + "epoch": 0.7631724902939545, + "grad_norm": 4391.12646484375, + "learning_rate": 4.250411861614498e-05, + "loss": 3.9386, + "step": 172 + }, + { + "epoch": 0.7676095396561287, + "grad_norm": 6815.49169921875, + "learning_rate": 4.2421746293245476e-05, + "loss": 4.011, + "step": 173 + }, + { + "epoch": 0.7720465890183028, + "grad_norm": 2788.70068359375, + "learning_rate": 4.2339373970345965e-05, + "loss": 4.0139, + "step": 174 + }, + { + "epoch": 0.776483638380477, + "grad_norm": 2901.850341796875, + "learning_rate": 4.225700164744646e-05, + "loss": 3.7802, + "step": 175 + }, + { + "epoch": 0.7809206877426511, + "grad_norm": 3142.524169921875, + "learning_rate": 4.2174629324546956e-05, + "loss": 3.7496, + "step": 176 + }, + { + "epoch": 0.7853577371048253, + "grad_norm": 2731.97607421875, + "learning_rate": 4.209225700164745e-05, + "loss": 3.834, + "step": 177 + }, + { + "epoch": 0.7897947864669994, + "grad_norm": 3077.71484375, + "learning_rate": 4.200988467874794e-05, + "loss": 3.9149, + "step": 178 + }, + { + "epoch": 0.7942318358291736, + "grad_norm": 3848.865966796875, + "learning_rate": 4.1927512355848435e-05, + "loss": 3.703, + "step": 179 + }, + { + "epoch": 0.7986688851913477, + "grad_norm": 5960.97021484375, + "learning_rate": 4.184514003294893e-05, + "loss": 3.7943, + "step": 180 + }, + { + "epoch": 0.8031059345535219, + "grad_norm": 4298.31640625, + "learning_rate": 4.1762767710049426e-05, + "loss": 3.6959, + "step": 181 + }, + { + "epoch": 0.807542983915696, + "grad_norm": 5129.8994140625, + "learning_rate": 4.1680395387149915e-05, + "loss": 3.9331, + "step": 182 + }, + { + "epoch": 0.8119800332778702, + "grad_norm": 2263.238525390625, + "learning_rate": 4.159802306425041e-05, + "loss": 3.6608, + "step": 183 + }, + { + "epoch": 0.8164170826400444, + "grad_norm": 2310.1015625, + "learning_rate": 4.1515650741350906e-05, + "loss": 3.8823, + "step": 184 + }, + { + "epoch": 0.8208541320022186, + "grad_norm": 2116.949951171875, + "learning_rate": 4.14332784184514e-05, + "loss": 3.7838, + "step": 185 + }, + { + "epoch": 0.8252911813643927, + "grad_norm": 3240.662353515625, + "learning_rate": 4.13509060955519e-05, + "loss": 3.8331, + "step": 186 + }, + { + "epoch": 0.8297282307265669, + "grad_norm": 5462.232421875, + "learning_rate": 4.126853377265239e-05, + "loss": 3.779, + "step": 187 + }, + { + "epoch": 0.834165280088741, + "grad_norm": 3005.918701171875, + "learning_rate": 4.118616144975289e-05, + "loss": 3.9813, + "step": 188 + }, + { + "epoch": 0.8386023294509152, + "grad_norm": 2229.3271484375, + "learning_rate": 4.110378912685338e-05, + "loss": 3.926, + "step": 189 + }, + { + "epoch": 0.8430393788130893, + "grad_norm": 5987.62060546875, + "learning_rate": 4.102141680395387e-05, + "loss": 3.7089, + "step": 190 + }, + { + "epoch": 0.8474764281752635, + "grad_norm": 2193.736572265625, + "learning_rate": 4.093904448105437e-05, + "loss": 3.7976, + "step": 191 + }, + { + "epoch": 0.8519134775374376, + "grad_norm": 2657.3505859375, + "learning_rate": 4.085667215815486e-05, + "loss": 3.5693, + "step": 192 + }, + { + "epoch": 0.8563505268996118, + "grad_norm": 4433.97509765625, + "learning_rate": 4.077429983525536e-05, + "loss": 3.7755, + "step": 193 + }, + { + "epoch": 0.8607875762617859, + "grad_norm": 5366.97900390625, + "learning_rate": 4.069192751235585e-05, + "loss": 3.9723, + "step": 194 + }, + { + "epoch": 0.8652246256239601, + "grad_norm": 2963.74072265625, + "learning_rate": 4.060955518945634e-05, + "loss": 3.835, + "step": 195 + }, + { + "epoch": 0.8696616749861342, + "grad_norm": 2591.837158203125, + "learning_rate": 4.052718286655684e-05, + "loss": 3.7441, + "step": 196 + }, + { + "epoch": 0.8740987243483084, + "grad_norm": 5419.08154296875, + "learning_rate": 4.044481054365733e-05, + "loss": 3.9721, + "step": 197 + }, + { + "epoch": 0.8785357737104825, + "grad_norm": 3774.27880859375, + "learning_rate": 4.036243822075783e-05, + "loss": 3.7249, + "step": 198 + }, + { + "epoch": 0.8829728230726567, + "grad_norm": 5378.41064453125, + "learning_rate": 4.0280065897858324e-05, + "loss": 3.8584, + "step": 199 + }, + { + "epoch": 0.8874098724348308, + "grad_norm": 4372.23046875, + "learning_rate": 4.019769357495882e-05, + "loss": 4.0649, + "step": 200 + }, + { + "epoch": 0.891846921797005, + "grad_norm": 3496.69091796875, + "learning_rate": 4.0115321252059315e-05, + "loss": 3.6083, + "step": 201 + }, + { + "epoch": 0.8962839711591791, + "grad_norm": 5939.7666015625, + "learning_rate": 4.0032948929159804e-05, + "loss": 3.8072, + "step": 202 + }, + { + "epoch": 0.9007210205213533, + "grad_norm": 3119.9375, + "learning_rate": 3.99505766062603e-05, + "loss": 3.7089, + "step": 203 + }, + { + "epoch": 0.9051580698835274, + "grad_norm": 3708.004150390625, + "learning_rate": 3.9868204283360795e-05, + "loss": 3.8108, + "step": 204 + }, + { + "epoch": 0.9095951192457016, + "grad_norm": 2537.582275390625, + "learning_rate": 3.978583196046129e-05, + "loss": 3.7416, + "step": 205 + }, + { + "epoch": 0.9140321686078757, + "grad_norm": 4762.97119140625, + "learning_rate": 3.970345963756178e-05, + "loss": 3.6366, + "step": 206 + }, + { + "epoch": 0.9184692179700499, + "grad_norm": 3252.349365234375, + "learning_rate": 3.9621087314662274e-05, + "loss": 3.6453, + "step": 207 + }, + { + "epoch": 0.922906267332224, + "grad_norm": 3314.591796875, + "learning_rate": 3.953871499176277e-05, + "loss": 3.5574, + "step": 208 + }, + { + "epoch": 0.9273433166943982, + "grad_norm": 5309.65087890625, + "learning_rate": 3.9456342668863265e-05, + "loss": 3.8529, + "step": 209 + }, + { + "epoch": 0.9317803660565723, + "grad_norm": 8192.8037109375, + "learning_rate": 3.9373970345963754e-05, + "loss": 3.9192, + "step": 210 + }, + { + "epoch": 0.9362174154187465, + "grad_norm": 10323.5654296875, + "learning_rate": 3.929159802306425e-05, + "loss": 3.7856, + "step": 211 + }, + { + "epoch": 0.9406544647809207, + "grad_norm": 3609.074951171875, + "learning_rate": 3.9209225700164745e-05, + "loss": 3.8713, + "step": 212 + }, + { + "epoch": 0.9450915141430949, + "grad_norm": 2638.693603515625, + "learning_rate": 3.912685337726524e-05, + "loss": 3.8446, + "step": 213 + }, + { + "epoch": 0.949528563505269, + "grad_norm": 6705.25048828125, + "learning_rate": 3.9044481054365736e-05, + "loss": 3.7553, + "step": 214 + }, + { + "epoch": 0.9539656128674432, + "grad_norm": 6157.7880859375, + "learning_rate": 3.896210873146623e-05, + "loss": 3.7287, + "step": 215 + }, + { + "epoch": 0.9584026622296173, + "grad_norm": 3325.8642578125, + "learning_rate": 3.8879736408566727e-05, + "loss": 3.7986, + "step": 216 + }, + { + "epoch": 0.9628397115917915, + "grad_norm": 2881.577880859375, + "learning_rate": 3.8797364085667215e-05, + "loss": 3.8484, + "step": 217 + }, + { + "epoch": 0.9672767609539656, + "grad_norm": 4911.40673828125, + "learning_rate": 3.871499176276771e-05, + "loss": 3.6729, + "step": 218 + }, + { + "epoch": 0.9717138103161398, + "grad_norm": 7369.07958984375, + "learning_rate": 3.8632619439868206e-05, + "loss": 3.5919, + "step": 219 + }, + { + "epoch": 0.9761508596783139, + "grad_norm": 7372.93115234375, + "learning_rate": 3.85502471169687e-05, + "loss": 3.7682, + "step": 220 + }, + { + "epoch": 0.9805879090404881, + "grad_norm": 4944.72802734375, + "learning_rate": 3.846787479406919e-05, + "loss": 3.5557, + "step": 221 + }, + { + "epoch": 0.9850249584026622, + "grad_norm": 2801.068603515625, + "learning_rate": 3.8385502471169686e-05, + "loss": 3.8395, + "step": 222 + }, + { + "epoch": 0.9894620077648364, + "grad_norm": 3235.59521484375, + "learning_rate": 3.830313014827018e-05, + "loss": 3.6656, + "step": 223 + }, + { + "epoch": 0.9938990571270105, + "grad_norm": 2795.662109375, + "learning_rate": 3.822075782537068e-05, + "loss": 3.6425, + "step": 224 + }, + { + "epoch": 0.9983361064891847, + "grad_norm": 5339.10302734375, + "learning_rate": 3.813838550247117e-05, + "loss": 3.5502, + "step": 225 + }, + { + "epoch": 1.0027731558513588, + "grad_norm": 3347.005615234375, + "learning_rate": 3.805601317957167e-05, + "loss": 3.6816, + "step": 226 + }, + { + "epoch": 1.007210205213533, + "grad_norm": 4426.99853515625, + "learning_rate": 3.797364085667216e-05, + "loss": 3.6512, + "step": 227 + }, + { + "epoch": 1.0116472545757071, + "grad_norm": 3261.813232421875, + "learning_rate": 3.789126853377266e-05, + "loss": 3.5252, + "step": 228 + }, + { + "epoch": 1.0160843039378813, + "grad_norm": 2609.848388671875, + "learning_rate": 3.780889621087315e-05, + "loss": 3.3021, + "step": 229 + }, + { + "epoch": 1.0205213533000554, + "grad_norm": 8878.7431640625, + "learning_rate": 3.772652388797364e-05, + "loss": 3.2879, + "step": 230 + }, + { + "epoch": 1.0249584026622296, + "grad_norm": 4302.8759765625, + "learning_rate": 3.764415156507414e-05, + "loss": 3.6508, + "step": 231 + }, + { + "epoch": 1.0293954520244037, + "grad_norm": 4386.603515625, + "learning_rate": 3.7561779242174634e-05, + "loss": 3.3892, + "step": 232 + }, + { + "epoch": 1.033832501386578, + "grad_norm": 3775.420654296875, + "learning_rate": 3.747940691927512e-05, + "loss": 3.543, + "step": 233 + }, + { + "epoch": 1.038269550748752, + "grad_norm": 4135.2509765625, + "learning_rate": 3.739703459637562e-05, + "loss": 3.5028, + "step": 234 + }, + { + "epoch": 1.0427066001109262, + "grad_norm": 2807.273681640625, + "learning_rate": 3.731466227347611e-05, + "loss": 3.4554, + "step": 235 + }, + { + "epoch": 1.0471436494731003, + "grad_norm": 5263.39306640625, + "learning_rate": 3.723228995057661e-05, + "loss": 3.4504, + "step": 236 + }, + { + "epoch": 1.0515806988352745, + "grad_norm": 2661.557861328125, + "learning_rate": 3.71499176276771e-05, + "loss": 3.5334, + "step": 237 + }, + { + "epoch": 1.0560177481974486, + "grad_norm": 8299.2822265625, + "learning_rate": 3.706754530477759e-05, + "loss": 3.3712, + "step": 238 + }, + { + "epoch": 1.0604547975596228, + "grad_norm": 2858.3662109375, + "learning_rate": 3.698517298187809e-05, + "loss": 3.147, + "step": 239 + }, + { + "epoch": 1.064891846921797, + "grad_norm": 3577.862060546875, + "learning_rate": 3.6902800658978584e-05, + "loss": 3.6677, + "step": 240 + }, + { + "epoch": 1.069328896283971, + "grad_norm": 4315.35986328125, + "learning_rate": 3.682042833607908e-05, + "loss": 3.343, + "step": 241 + }, + { + "epoch": 1.0737659456461452, + "grad_norm": 3357.373046875, + "learning_rate": 3.6738056013179575e-05, + "loss": 3.6462, + "step": 242 + }, + { + "epoch": 1.0782029950083194, + "grad_norm": 2771.050537109375, + "learning_rate": 3.665568369028007e-05, + "loss": 3.5208, + "step": 243 + }, + { + "epoch": 1.0826400443704935, + "grad_norm": 2507.241455078125, + "learning_rate": 3.6573311367380566e-05, + "loss": 3.4599, + "step": 244 + }, + { + "epoch": 1.0870770937326677, + "grad_norm": 4087.06103515625, + "learning_rate": 3.6490939044481054e-05, + "loss": 3.6829, + "step": 245 + }, + { + "epoch": 1.0915141430948418, + "grad_norm": 8269.115234375, + "learning_rate": 3.640856672158155e-05, + "loss": 3.5641, + "step": 246 + }, + { + "epoch": 1.095951192457016, + "grad_norm": 7861.7080078125, + "learning_rate": 3.6326194398682045e-05, + "loss": 3.4072, + "step": 247 + }, + { + "epoch": 1.1003882418191901, + "grad_norm": 4300.45654296875, + "learning_rate": 3.624382207578254e-05, + "loss": 3.5897, + "step": 248 + }, + { + "epoch": 1.1048252911813643, + "grad_norm": 7326.00732421875, + "learning_rate": 3.616144975288303e-05, + "loss": 3.5901, + "step": 249 + }, + { + "epoch": 1.1092623405435384, + "grad_norm": 3843.257568359375, + "learning_rate": 3.6079077429983525e-05, + "loss": 3.5679, + "step": 250 + }, + { + "epoch": 1.1136993899057126, + "grad_norm": 2784.050537109375, + "learning_rate": 3.599670510708402e-05, + "loss": 3.4599, + "step": 251 + }, + { + "epoch": 1.1181364392678868, + "grad_norm": 1591.23583984375, + "learning_rate": 3.5914332784184516e-05, + "loss": 3.6669, + "step": 252 + }, + { + "epoch": 1.122573488630061, + "grad_norm": 4273.7880859375, + "learning_rate": 3.583196046128501e-05, + "loss": 3.4427, + "step": 253 + }, + { + "epoch": 1.127010537992235, + "grad_norm": 4322.25732421875, + "learning_rate": 3.5749588138385507e-05, + "loss": 3.3652, + "step": 254 + }, + { + "epoch": 1.1314475873544092, + "grad_norm": 3580.967529296875, + "learning_rate": 3.5667215815486e-05, + "loss": 3.4907, + "step": 255 + }, + { + "epoch": 1.1358846367165834, + "grad_norm": 3707.540771484375, + "learning_rate": 3.55848434925865e-05, + "loss": 3.6951, + "step": 256 + }, + { + "epoch": 1.1403216860787575, + "grad_norm": 3017.675048828125, + "learning_rate": 3.5502471169686986e-05, + "loss": 3.2977, + "step": 257 + }, + { + "epoch": 1.1447587354409319, + "grad_norm": 4870.4072265625, + "learning_rate": 3.542009884678748e-05, + "loss": 3.4824, + "step": 258 + }, + { + "epoch": 1.149195784803106, + "grad_norm": 3981.42822265625, + "learning_rate": 3.533772652388798e-05, + "loss": 3.5768, + "step": 259 + }, + { + "epoch": 1.1536328341652802, + "grad_norm": 4044.103515625, + "learning_rate": 3.525535420098847e-05, + "loss": 3.3853, + "step": 260 + }, + { + "epoch": 1.1580698835274543, + "grad_norm": 1712.18798828125, + "learning_rate": 3.517298187808896e-05, + "loss": 3.6393, + "step": 261 + }, + { + "epoch": 1.1625069328896285, + "grad_norm": 10580.7451171875, + "learning_rate": 3.509060955518946e-05, + "loss": 3.4974, + "step": 262 + }, + { + "epoch": 1.1669439822518026, + "grad_norm": 2071.3369140625, + "learning_rate": 3.500823723228995e-05, + "loss": 3.4609, + "step": 263 + }, + { + "epoch": 1.1713810316139768, + "grad_norm": 6024.3095703125, + "learning_rate": 3.492586490939045e-05, + "loss": 3.364, + "step": 264 + }, + { + "epoch": 1.175818080976151, + "grad_norm": 5767.52685546875, + "learning_rate": 3.4843492586490936e-05, + "loss": 3.3503, + "step": 265 + }, + { + "epoch": 1.180255130338325, + "grad_norm": 6747.88330078125, + "learning_rate": 3.476112026359143e-05, + "loss": 3.639, + "step": 266 + }, + { + "epoch": 1.1846921797004992, + "grad_norm": 4800.87158203125, + "learning_rate": 3.467874794069193e-05, + "loss": 3.1528, + "step": 267 + }, + { + "epoch": 1.1891292290626734, + "grad_norm": 5317.23681640625, + "learning_rate": 3.459637561779242e-05, + "loss": 3.4173, + "step": 268 + }, + { + "epoch": 1.1935662784248475, + "grad_norm": 7454.5, + "learning_rate": 3.451400329489292e-05, + "loss": 3.3721, + "step": 269 + }, + { + "epoch": 1.1980033277870217, + "grad_norm": 2274.439453125, + "learning_rate": 3.4431630971993414e-05, + "loss": 3.5645, + "step": 270 + }, + { + "epoch": 1.2024403771491958, + "grad_norm": 3757.284912109375, + "learning_rate": 3.434925864909391e-05, + "loss": 3.3543, + "step": 271 + }, + { + "epoch": 1.20687742651137, + "grad_norm": 2109.77880859375, + "learning_rate": 3.4266886326194404e-05, + "loss": 3.3646, + "step": 272 + }, + { + "epoch": 1.2113144758735441, + "grad_norm": 3320.041015625, + "learning_rate": 3.418451400329489e-05, + "loss": 3.4072, + "step": 273 + }, + { + "epoch": 1.2157515252357183, + "grad_norm": 4438.30126953125, + "learning_rate": 3.410214168039539e-05, + "loss": 3.3151, + "step": 274 + }, + { + "epoch": 1.2201885745978924, + "grad_norm": 2422.821044921875, + "learning_rate": 3.4019769357495884e-05, + "loss": 3.5002, + "step": 275 + }, + { + "epoch": 1.2246256239600666, + "grad_norm": 2783.8408203125, + "learning_rate": 3.393739703459638e-05, + "loss": 3.4783, + "step": 276 + }, + { + "epoch": 1.2290626733222407, + "grad_norm": 3811.885986328125, + "learning_rate": 3.385502471169687e-05, + "loss": 3.4199, + "step": 277 + }, + { + "epoch": 1.2334997226844149, + "grad_norm": 3138.676513671875, + "learning_rate": 3.3772652388797364e-05, + "loss": 3.3431, + "step": 278 + }, + { + "epoch": 1.237936772046589, + "grad_norm": 7387.8515625, + "learning_rate": 3.369028006589786e-05, + "loss": 3.2002, + "step": 279 + }, + { + "epoch": 1.2423738214087632, + "grad_norm": 2484.769775390625, + "learning_rate": 3.3607907742998355e-05, + "loss": 3.5147, + "step": 280 + }, + { + "epoch": 1.2468108707709373, + "grad_norm": 2683.162109375, + "learning_rate": 3.352553542009885e-05, + "loss": 3.3471, + "step": 281 + }, + { + "epoch": 1.2512479201331115, + "grad_norm": 6036.736328125, + "learning_rate": 3.3443163097199346e-05, + "loss": 3.354, + "step": 282 + }, + { + "epoch": 1.2556849694952856, + "grad_norm": 8086.67041015625, + "learning_rate": 3.336079077429984e-05, + "loss": 3.4199, + "step": 283 + }, + { + "epoch": 1.2601220188574598, + "grad_norm": 3705.58447265625, + "learning_rate": 3.327841845140033e-05, + "loss": 3.5757, + "step": 284 + }, + { + "epoch": 1.264559068219634, + "grad_norm": 3918.829833984375, + "learning_rate": 3.3196046128500825e-05, + "loss": 3.446, + "step": 285 + }, + { + "epoch": 1.268996117581808, + "grad_norm": 2083.683349609375, + "learning_rate": 3.311367380560132e-05, + "loss": 3.3041, + "step": 286 + }, + { + "epoch": 1.2734331669439822, + "grad_norm": 2724.96826171875, + "learning_rate": 3.3031301482701816e-05, + "loss": 3.3938, + "step": 287 + }, + { + "epoch": 1.2778702163061564, + "grad_norm": 4340.44140625, + "learning_rate": 3.2948929159802305e-05, + "loss": 3.3784, + "step": 288 + }, + { + "epoch": 1.2823072656683305, + "grad_norm": 4921.4140625, + "learning_rate": 3.28665568369028e-05, + "loss": 3.4825, + "step": 289 + }, + { + "epoch": 1.2867443150305047, + "grad_norm": 5665.88525390625, + "learning_rate": 3.2784184514003296e-05, + "loss": 3.4291, + "step": 290 + }, + { + "epoch": 1.2911813643926788, + "grad_norm": 4413.6669921875, + "learning_rate": 3.270181219110379e-05, + "loss": 3.5536, + "step": 291 + }, + { + "epoch": 1.295618413754853, + "grad_norm": 4070.07470703125, + "learning_rate": 3.261943986820428e-05, + "loss": 3.1241, + "step": 292 + }, + { + "epoch": 1.3000554631170271, + "grad_norm": 2463.054931640625, + "learning_rate": 3.2537067545304775e-05, + "loss": 3.4662, + "step": 293 + }, + { + "epoch": 1.3044925124792013, + "grad_norm": 2534.056396484375, + "learning_rate": 3.245469522240527e-05, + "loss": 3.4663, + "step": 294 + }, + { + "epoch": 1.3089295618413754, + "grad_norm": 6652.2802734375, + "learning_rate": 3.2372322899505766e-05, + "loss": 3.4475, + "step": 295 + }, + { + "epoch": 1.3133666112035496, + "grad_norm": 3905.2666015625, + "learning_rate": 3.228995057660626e-05, + "loss": 3.3675, + "step": 296 + }, + { + "epoch": 1.3178036605657237, + "grad_norm": 2461.313720703125, + "learning_rate": 3.220757825370676e-05, + "loss": 3.587, + "step": 297 + }, + { + "epoch": 1.3222407099278979, + "grad_norm": 3086.779296875, + "learning_rate": 3.212520593080725e-05, + "loss": 3.3603, + "step": 298 + }, + { + "epoch": 1.326677759290072, + "grad_norm": 2486.352783203125, + "learning_rate": 3.204283360790775e-05, + "loss": 3.2709, + "step": 299 + }, + { + "epoch": 1.3311148086522462, + "grad_norm": 2938.1455078125, + "learning_rate": 3.196046128500824e-05, + "loss": 3.3718, + "step": 300 + }, + { + "epoch": 1.3355518580144203, + "grad_norm": 5892.052734375, + "learning_rate": 3.187808896210873e-05, + "loss": 3.4811, + "step": 301 + }, + { + "epoch": 1.3399889073765945, + "grad_norm": 6452.23388671875, + "learning_rate": 3.179571663920923e-05, + "loss": 3.5559, + "step": 302 + }, + { + "epoch": 1.3444259567387689, + "grad_norm": 5348.853515625, + "learning_rate": 3.171334431630972e-05, + "loss": 3.5227, + "step": 303 + }, + { + "epoch": 1.348863006100943, + "grad_norm": 3456.395263671875, + "learning_rate": 3.163097199341021e-05, + "loss": 3.4475, + "step": 304 + }, + { + "epoch": 1.3533000554631172, + "grad_norm": 1416.5257568359375, + "learning_rate": 3.154859967051071e-05, + "loss": 3.4591, + "step": 305 + }, + { + "epoch": 1.3577371048252913, + "grad_norm": 3570.713623046875, + "learning_rate": 3.14662273476112e-05, + "loss": 3.4421, + "step": 306 + }, + { + "epoch": 1.3621741541874655, + "grad_norm": 2852.215087890625, + "learning_rate": 3.13838550247117e-05, + "loss": 3.1724, + "step": 307 + }, + { + "epoch": 1.3666112035496396, + "grad_norm": 2625.452392578125, + "learning_rate": 3.1301482701812194e-05, + "loss": 3.4433, + "step": 308 + }, + { + "epoch": 1.3710482529118138, + "grad_norm": 3905.065673828125, + "learning_rate": 3.121911037891269e-05, + "loss": 3.6359, + "step": 309 + }, + { + "epoch": 1.375485302273988, + "grad_norm": 4539.224609375, + "learning_rate": 3.1136738056013184e-05, + "loss": 3.2964, + "step": 310 + }, + { + "epoch": 1.379922351636162, + "grad_norm": 3368.270263671875, + "learning_rate": 3.105436573311368e-05, + "loss": 3.342, + "step": 311 + }, + { + "epoch": 1.3843594009983362, + "grad_norm": 5875.1650390625, + "learning_rate": 3.097199341021417e-05, + "loss": 3.2992, + "step": 312 + }, + { + "epoch": 1.3887964503605104, + "grad_norm": 11779.1884765625, + "learning_rate": 3.0889621087314664e-05, + "loss": 3.2502, + "step": 313 + }, + { + "epoch": 1.3932334997226845, + "grad_norm": 2390.0234375, + "learning_rate": 3.080724876441516e-05, + "loss": 3.1378, + "step": 314 + }, + { + "epoch": 1.3976705490848587, + "grad_norm": 3221.599853515625, + "learning_rate": 3.0724876441515655e-05, + "loss": 3.2542, + "step": 315 + }, + { + "epoch": 1.4021075984470328, + "grad_norm": 8320.6494140625, + "learning_rate": 3.0642504118616144e-05, + "loss": 3.3523, + "step": 316 + }, + { + "epoch": 1.406544647809207, + "grad_norm": 1748.6888427734375, + "learning_rate": 3.056013179571664e-05, + "loss": 3.3391, + "step": 317 + }, + { + "epoch": 1.410981697171381, + "grad_norm": 2458.482421875, + "learning_rate": 3.0477759472817135e-05, + "loss": 3.1674, + "step": 318 + }, + { + "epoch": 1.4154187465335553, + "grad_norm": 6219.91748046875, + "learning_rate": 3.039538714991763e-05, + "loss": 3.3686, + "step": 319 + }, + { + "epoch": 1.4198557958957294, + "grad_norm": 11286.7216796875, + "learning_rate": 3.0313014827018122e-05, + "loss": 3.3044, + "step": 320 + }, + { + "epoch": 1.4242928452579036, + "grad_norm": 4691.47021484375, + "learning_rate": 3.0230642504118618e-05, + "loss": 3.4039, + "step": 321 + }, + { + "epoch": 1.4287298946200777, + "grad_norm": 6158.076171875, + "learning_rate": 3.0148270181219113e-05, + "loss": 3.3732, + "step": 322 + }, + { + "epoch": 1.4331669439822519, + "grad_norm": 3103.380615234375, + "learning_rate": 3.006589785831961e-05, + "loss": 3.319, + "step": 323 + }, + { + "epoch": 1.437603993344426, + "grad_norm": 8175.55712890625, + "learning_rate": 2.9983525535420097e-05, + "loss": 3.1852, + "step": 324 + }, + { + "epoch": 1.4420410427066002, + "grad_norm": 3382.044921875, + "learning_rate": 2.9901153212520593e-05, + "loss": 3.3421, + "step": 325 + }, + { + "epoch": 1.4464780920687743, + "grad_norm": 6218.29052734375, + "learning_rate": 2.9818780889621088e-05, + "loss": 3.1956, + "step": 326 + }, + { + "epoch": 1.4509151414309485, + "grad_norm": 3371.7529296875, + "learning_rate": 2.9736408566721587e-05, + "loss": 3.136, + "step": 327 + }, + { + "epoch": 1.4553521907931226, + "grad_norm": 1970.310302734375, + "learning_rate": 2.9654036243822076e-05, + "loss": 2.9911, + "step": 328 + }, + { + "epoch": 1.4597892401552968, + "grad_norm": 2749.70458984375, + "learning_rate": 2.957166392092257e-05, + "loss": 3.2593, + "step": 329 + }, + { + "epoch": 1.464226289517471, + "grad_norm": 4199.37109375, + "learning_rate": 2.9489291598023067e-05, + "loss": 3.0715, + "step": 330 + }, + { + "epoch": 1.468663338879645, + "grad_norm": 2983.9072265625, + "learning_rate": 2.9406919275123562e-05, + "loss": 3.3807, + "step": 331 + }, + { + "epoch": 1.4731003882418192, + "grad_norm": 3894.69384765625, + "learning_rate": 2.9324546952224054e-05, + "loss": 3.3204, + "step": 332 + }, + { + "epoch": 1.4775374376039934, + "grad_norm": 6819.61572265625, + "learning_rate": 2.924217462932455e-05, + "loss": 3.5706, + "step": 333 + }, + { + "epoch": 1.4819744869661675, + "grad_norm": 10483.4296875, + "learning_rate": 2.9159802306425045e-05, + "loss": 3.4653, + "step": 334 + }, + { + "epoch": 1.4864115363283417, + "grad_norm": 3186.267333984375, + "learning_rate": 2.907742998352554e-05, + "loss": 3.4286, + "step": 335 + }, + { + "epoch": 1.4908485856905158, + "grad_norm": 3522.5712890625, + "learning_rate": 2.899505766062603e-05, + "loss": 3.2671, + "step": 336 + }, + { + "epoch": 1.49528563505269, + "grad_norm": 4962.36767578125, + "learning_rate": 2.8912685337726525e-05, + "loss": 3.341, + "step": 337 + }, + { + "epoch": 1.4997226844148641, + "grad_norm": 2928.841796875, + "learning_rate": 2.883031301482702e-05, + "loss": 3.3078, + "step": 338 + }, + { + "epoch": 1.5041597337770383, + "grad_norm": 2143.45068359375, + "learning_rate": 2.8747940691927515e-05, + "loss": 3.1212, + "step": 339 + }, + { + "epoch": 1.5085967831392124, + "grad_norm": 5597.478515625, + "learning_rate": 2.8665568369028008e-05, + "loss": 3.2542, + "step": 340 + }, + { + "epoch": 1.5130338325013866, + "grad_norm": 2670.3974609375, + "learning_rate": 2.8583196046128503e-05, + "loss": 3.3586, + "step": 341 + }, + { + "epoch": 1.5174708818635607, + "grad_norm": 2938.1201171875, + "learning_rate": 2.8500823723229e-05, + "loss": 3.1828, + "step": 342 + }, + { + "epoch": 1.5219079312257349, + "grad_norm": 3200.698974609375, + "learning_rate": 2.8418451400329494e-05, + "loss": 3.3413, + "step": 343 + }, + { + "epoch": 1.526344980587909, + "grad_norm": 2602.446044921875, + "learning_rate": 2.8336079077429983e-05, + "loss": 3.2226, + "step": 344 + }, + { + "epoch": 1.5307820299500832, + "grad_norm": 5301.88525390625, + "learning_rate": 2.8253706754530478e-05, + "loss": 3.2617, + "step": 345 + }, + { + "epoch": 1.5352190793122573, + "grad_norm": 3105.766845703125, + "learning_rate": 2.8171334431630973e-05, + "loss": 3.4838, + "step": 346 + }, + { + "epoch": 1.5396561286744315, + "grad_norm": 2592.593994140625, + "learning_rate": 2.808896210873147e-05, + "loss": 3.3851, + "step": 347 + }, + { + "epoch": 1.5440931780366056, + "grad_norm": 2639.238037109375, + "learning_rate": 2.800658978583196e-05, + "loss": 3.1511, + "step": 348 + }, + { + "epoch": 1.5485302273987798, + "grad_norm": 5876.7822265625, + "learning_rate": 2.7924217462932456e-05, + "loss": 3.1924, + "step": 349 + }, + { + "epoch": 1.552967276760954, + "grad_norm": 6174.7626953125, + "learning_rate": 2.7841845140032952e-05, + "loss": 3.2, + "step": 350 + }, + { + "epoch": 1.557404326123128, + "grad_norm": 7660.24853515625, + "learning_rate": 2.775947281713344e-05, + "loss": 3.1312, + "step": 351 + }, + { + "epoch": 1.5618413754853022, + "grad_norm": 3936.479736328125, + "learning_rate": 2.7677100494233936e-05, + "loss": 3.4326, + "step": 352 + }, + { + "epoch": 1.5662784248474764, + "grad_norm": 3265.771484375, + "learning_rate": 2.759472817133443e-05, + "loss": 3.0412, + "step": 353 + }, + { + "epoch": 1.5707154742096505, + "grad_norm": 5142.13427734375, + "learning_rate": 2.7512355848434927e-05, + "loss": 3.1123, + "step": 354 + }, + { + "epoch": 1.5751525235718247, + "grad_norm": 6845.93212890625, + "learning_rate": 2.742998352553542e-05, + "loss": 3.3672, + "step": 355 + }, + { + "epoch": 1.5795895729339988, + "grad_norm": 2055.831787109375, + "learning_rate": 2.7347611202635915e-05, + "loss": 3.3697, + "step": 356 + }, + { + "epoch": 1.584026622296173, + "grad_norm": 1842.26513671875, + "learning_rate": 2.726523887973641e-05, + "loss": 3.1402, + "step": 357 + }, + { + "epoch": 1.5884636716583471, + "grad_norm": 4246.33544921875, + "learning_rate": 2.7182866556836905e-05, + "loss": 3.3053, + "step": 358 + }, + { + "epoch": 1.5929007210205213, + "grad_norm": 2916.94677734375, + "learning_rate": 2.7100494233937394e-05, + "loss": 3.2462, + "step": 359 + }, + { + "epoch": 1.5973377703826954, + "grad_norm": 3905.638427734375, + "learning_rate": 2.7018121911037893e-05, + "loss": 3.1733, + "step": 360 + }, + { + "epoch": 1.6017748197448696, + "grad_norm": 2354.46484375, + "learning_rate": 2.693574958813839e-05, + "loss": 3.3482, + "step": 361 + }, + { + "epoch": 1.6062118691070437, + "grad_norm": 3828.93115234375, + "learning_rate": 2.6853377265238884e-05, + "loss": 3.1651, + "step": 362 + }, + { + "epoch": 1.6106489184692179, + "grad_norm": 2451.734375, + "learning_rate": 2.6771004942339373e-05, + "loss": 3.2279, + "step": 363 + }, + { + "epoch": 1.615085967831392, + "grad_norm": 2418.510498046875, + "learning_rate": 2.6688632619439868e-05, + "loss": 3.0485, + "step": 364 + }, + { + "epoch": 1.6195230171935662, + "grad_norm": 2874.52685546875, + "learning_rate": 2.6606260296540363e-05, + "loss": 3.2669, + "step": 365 + }, + { + "epoch": 1.6239600665557403, + "grad_norm": 3557.26708984375, + "learning_rate": 2.652388797364086e-05, + "loss": 3.0794, + "step": 366 + }, + { + "epoch": 1.6283971159179145, + "grad_norm": 3598.58642578125, + "learning_rate": 2.644151565074135e-05, + "loss": 3.0051, + "step": 367 + }, + { + "epoch": 1.6328341652800886, + "grad_norm": 3217.324951171875, + "learning_rate": 2.6359143327841846e-05, + "loss": 3.2616, + "step": 368 + }, + { + "epoch": 1.6372712146422628, + "grad_norm": 4292.02099609375, + "learning_rate": 2.6276771004942342e-05, + "loss": 2.9456, + "step": 369 + }, + { + "epoch": 1.641708264004437, + "grad_norm": 1767.0902099609375, + "learning_rate": 2.6194398682042837e-05, + "loss": 3.3488, + "step": 370 + }, + { + "epoch": 1.646145313366611, + "grad_norm": 4550.89501953125, + "learning_rate": 2.6112026359143326e-05, + "loss": 3.2275, + "step": 371 + }, + { + "epoch": 1.6505823627287852, + "grad_norm": 1941.79443359375, + "learning_rate": 2.602965403624382e-05, + "loss": 3.3585, + "step": 372 + }, + { + "epoch": 1.6550194120909594, + "grad_norm": 4232.7255859375, + "learning_rate": 2.5947281713344317e-05, + "loss": 3.2172, + "step": 373 + }, + { + "epoch": 1.6594564614531335, + "grad_norm": 3752.64208984375, + "learning_rate": 2.5864909390444812e-05, + "loss": 3.3491, + "step": 374 + }, + { + "epoch": 1.6638935108153077, + "grad_norm": 5631.474609375, + "learning_rate": 2.5782537067545305e-05, + "loss": 3.1454, + "step": 375 + }, + { + "epoch": 1.6683305601774818, + "grad_norm": 1958.1658935546875, + "learning_rate": 2.57001647446458e-05, + "loss": 2.9934, + "step": 376 + }, + { + "epoch": 1.672767609539656, + "grad_norm": 1635.100341796875, + "learning_rate": 2.5617792421746295e-05, + "loss": 3.1686, + "step": 377 + }, + { + "epoch": 1.6772046589018301, + "grad_norm": 4911.3125, + "learning_rate": 2.553542009884679e-05, + "loss": 3.4657, + "step": 378 + }, + { + "epoch": 1.6816417082640043, + "grad_norm": 3827.627197265625, + "learning_rate": 2.545304777594728e-05, + "loss": 3.2032, + "step": 379 + }, + { + "epoch": 1.6860787576261784, + "grad_norm": 2993.051025390625, + "learning_rate": 2.5370675453047775e-05, + "loss": 2.9632, + "step": 380 + }, + { + "epoch": 1.6905158069883528, + "grad_norm": 8272.6513671875, + "learning_rate": 2.528830313014827e-05, + "loss": 3.0626, + "step": 381 + }, + { + "epoch": 1.694952856350527, + "grad_norm": 2200.14892578125, + "learning_rate": 2.520593080724877e-05, + "loss": 3.2736, + "step": 382 + }, + { + "epoch": 1.699389905712701, + "grad_norm": 7213.10107421875, + "learning_rate": 2.5123558484349258e-05, + "loss": 3.1779, + "step": 383 + }, + { + "epoch": 1.7038269550748752, + "grad_norm": 2109.79150390625, + "learning_rate": 2.5041186161449753e-05, + "loss": 3.2537, + "step": 384 + }, + { + "epoch": 1.7082640044370494, + "grad_norm": 7044.64208984375, + "learning_rate": 2.495881383855025e-05, + "loss": 3.1927, + "step": 385 + }, + { + "epoch": 1.7127010537992235, + "grad_norm": 1790.706787109375, + "learning_rate": 2.487644151565074e-05, + "loss": 3.158, + "step": 386 + }, + { + "epoch": 1.7171381031613977, + "grad_norm": 6801.89697265625, + "learning_rate": 2.4794069192751236e-05, + "loss": 3.192, + "step": 387 + }, + { + "epoch": 1.7215751525235718, + "grad_norm": 4925.21923828125, + "learning_rate": 2.4711696869851732e-05, + "loss": 3.3486, + "step": 388 + }, + { + "epoch": 1.726012201885746, + "grad_norm": 1920.9197998046875, + "learning_rate": 2.4629324546952227e-05, + "loss": 3.281, + "step": 389 + }, + { + "epoch": 1.7304492512479202, + "grad_norm": 4378.8037109375, + "learning_rate": 2.454695222405272e-05, + "loss": 3.4103, + "step": 390 + }, + { + "epoch": 1.7348863006100943, + "grad_norm": 4085.0537109375, + "learning_rate": 2.4464579901153215e-05, + "loss": 3.0463, + "step": 391 + }, + { + "epoch": 1.7393233499722685, + "grad_norm": 3358.34375, + "learning_rate": 2.4382207578253707e-05, + "loss": 3.1587, + "step": 392 + }, + { + "epoch": 1.7437603993344426, + "grad_norm": 5024.78955078125, + "learning_rate": 2.4299835255354202e-05, + "loss": 3.0783, + "step": 393 + }, + { + "epoch": 1.7481974486966168, + "grad_norm": 4263.76806640625, + "learning_rate": 2.4217462932454694e-05, + "loss": 3.2889, + "step": 394 + }, + { + "epoch": 1.752634498058791, + "grad_norm": 2913.8828125, + "learning_rate": 2.413509060955519e-05, + "loss": 2.9302, + "step": 395 + }, + { + "epoch": 1.757071547420965, + "grad_norm": 2896.53759765625, + "learning_rate": 2.4052718286655685e-05, + "loss": 3.0902, + "step": 396 + }, + { + "epoch": 1.7615085967831392, + "grad_norm": 8661.619140625, + "learning_rate": 2.397034596375618e-05, + "loss": 3.0539, + "step": 397 + }, + { + "epoch": 1.7659456461453134, + "grad_norm": 1757.168212890625, + "learning_rate": 2.3887973640856673e-05, + "loss": 3.1075, + "step": 398 + }, + { + "epoch": 1.7703826955074875, + "grad_norm": 3529.929931640625, + "learning_rate": 2.380560131795717e-05, + "loss": 3.3339, + "step": 399 + }, + { + "epoch": 1.7748197448696617, + "grad_norm": 2815.612548828125, + "learning_rate": 2.372322899505766e-05, + "loss": 3.0102, + "step": 400 + }, + { + "epoch": 1.7792567942318358, + "grad_norm": 3648.347412109375, + "learning_rate": 2.3640856672158156e-05, + "loss": 3.3809, + "step": 401 + }, + { + "epoch": 1.78369384359401, + "grad_norm": 2135.781982421875, + "learning_rate": 2.355848434925865e-05, + "loss": 3.2105, + "step": 402 + }, + { + "epoch": 1.788130892956184, + "grad_norm": 4210.21142578125, + "learning_rate": 2.3476112026359147e-05, + "loss": 3.0471, + "step": 403 + }, + { + "epoch": 1.7925679423183583, + "grad_norm": 5141.791015625, + "learning_rate": 2.339373970345964e-05, + "loss": 3.0019, + "step": 404 + }, + { + "epoch": 1.7970049916805324, + "grad_norm": 3927.765625, + "learning_rate": 2.3311367380560134e-05, + "loss": 2.989, + "step": 405 + }, + { + "epoch": 1.8014420410427066, + "grad_norm": 6483.11669921875, + "learning_rate": 2.3228995057660626e-05, + "loss": 3.1778, + "step": 406 + }, + { + "epoch": 1.8058790904048807, + "grad_norm": 4240.06689453125, + "learning_rate": 2.3146622734761122e-05, + "loss": 3.0266, + "step": 407 + }, + { + "epoch": 1.8103161397670549, + "grad_norm": 3383.435546875, + "learning_rate": 2.3064250411861614e-05, + "loss": 3.2427, + "step": 408 + }, + { + "epoch": 1.8147531891292292, + "grad_norm": 5996.46875, + "learning_rate": 2.298187808896211e-05, + "loss": 3.0609, + "step": 409 + }, + { + "epoch": 1.8191902384914034, + "grad_norm": 3249.54150390625, + "learning_rate": 2.2899505766062605e-05, + "loss": 2.858, + "step": 410 + }, + { + "epoch": 1.8236272878535775, + "grad_norm": 7141.77099609375, + "learning_rate": 2.28171334431631e-05, + "loss": 2.9937, + "step": 411 + }, + { + "epoch": 1.8280643372157517, + "grad_norm": 4017.89404296875, + "learning_rate": 2.2734761120263592e-05, + "loss": 3.0683, + "step": 412 + }, + { + "epoch": 1.8325013865779258, + "grad_norm": 2962.71923828125, + "learning_rate": 2.2652388797364088e-05, + "loss": 3.0787, + "step": 413 + }, + { + "epoch": 1.8369384359401, + "grad_norm": 4861.8515625, + "learning_rate": 2.257001647446458e-05, + "loss": 3.1757, + "step": 414 + }, + { + "epoch": 1.8413754853022741, + "grad_norm": 7352.44775390625, + "learning_rate": 2.2487644151565075e-05, + "loss": 3.0528, + "step": 415 + }, + { + "epoch": 1.8458125346644483, + "grad_norm": 5794.9111328125, + "learning_rate": 2.240527182866557e-05, + "loss": 3.2362, + "step": 416 + }, + { + "epoch": 1.8502495840266224, + "grad_norm": 2827.435791015625, + "learning_rate": 2.2322899505766066e-05, + "loss": 2.9597, + "step": 417 + }, + { + "epoch": 1.8546866333887966, + "grad_norm": 3458.3828125, + "learning_rate": 2.224052718286656e-05, + "loss": 3.1392, + "step": 418 + }, + { + "epoch": 1.8591236827509707, + "grad_norm": 2623.705322265625, + "learning_rate": 2.215815485996705e-05, + "loss": 3.0754, + "step": 419 + }, + { + "epoch": 1.8635607321131449, + "grad_norm": 3555.89697265625, + "learning_rate": 2.2075782537067546e-05, + "loss": 2.9074, + "step": 420 + }, + { + "epoch": 1.867997781475319, + "grad_norm": 2072.844482421875, + "learning_rate": 2.1993410214168038e-05, + "loss": 3.1539, + "step": 421 + }, + { + "epoch": 1.8724348308374932, + "grad_norm": 3401.2421875, + "learning_rate": 2.1911037891268533e-05, + "loss": 3.1378, + "step": 422 + }, + { + "epoch": 1.8768718801996673, + "grad_norm": 2346.244873046875, + "learning_rate": 2.182866556836903e-05, + "loss": 3.2617, + "step": 423 + }, + { + "epoch": 1.8813089295618415, + "grad_norm": 2616.072509765625, + "learning_rate": 2.1746293245469524e-05, + "loss": 2.9015, + "step": 424 + }, + { + "epoch": 1.8857459789240156, + "grad_norm": 1976.274169921875, + "learning_rate": 2.1663920922570016e-05, + "loss": 3.219, + "step": 425 + }, + { + "epoch": 1.8901830282861898, + "grad_norm": 2609.365234375, + "learning_rate": 2.1581548599670512e-05, + "loss": 3.0268, + "step": 426 + }, + { + "epoch": 1.894620077648364, + "grad_norm": 9145.541015625, + "learning_rate": 2.1499176276771004e-05, + "loss": 3.2253, + "step": 427 + }, + { + "epoch": 1.899057127010538, + "grad_norm": 4822.47119140625, + "learning_rate": 2.14168039538715e-05, + "loss": 3.278, + "step": 428 + }, + { + "epoch": 1.9034941763727122, + "grad_norm": 2341.47900390625, + "learning_rate": 2.1334431630971995e-05, + "loss": 2.9406, + "step": 429 + }, + { + "epoch": 1.9079312257348864, + "grad_norm": 3344.80322265625, + "learning_rate": 2.125205930807249e-05, + "loss": 3.164, + "step": 430 + }, + { + "epoch": 1.9123682750970605, + "grad_norm": 1989.66015625, + "learning_rate": 2.1169686985172982e-05, + "loss": 3.0597, + "step": 431 + }, + { + "epoch": 1.9168053244592347, + "grad_norm": 4817.58642578125, + "learning_rate": 2.1087314662273478e-05, + "loss": 2.8597, + "step": 432 + }, + { + "epoch": 1.9212423738214088, + "grad_norm": 1732.5284423828125, + "learning_rate": 2.100494233937397e-05, + "loss": 3.2083, + "step": 433 + }, + { + "epoch": 1.925679423183583, + "grad_norm": 7493.603515625, + "learning_rate": 2.0922570016474465e-05, + "loss": 3.0343, + "step": 434 + }, + { + "epoch": 1.9301164725457571, + "grad_norm": 2923.343017578125, + "learning_rate": 2.0840197693574957e-05, + "loss": 3.3091, + "step": 435 + }, + { + "epoch": 1.9345535219079313, + "grad_norm": 1870.4757080078125, + "learning_rate": 2.0757825370675453e-05, + "loss": 3.0618, + "step": 436 + }, + { + "epoch": 1.9389905712701054, + "grad_norm": 2664.408203125, + "learning_rate": 2.067545304777595e-05, + "loss": 3.0118, + "step": 437 + }, + { + "epoch": 1.9434276206322796, + "grad_norm": 1833.4034423828125, + "learning_rate": 2.0593080724876444e-05, + "loss": 3.0631, + "step": 438 + }, + { + "epoch": 1.9478646699944537, + "grad_norm": 4199.58740234375, + "learning_rate": 2.0510708401976936e-05, + "loss": 2.9171, + "step": 439 + }, + { + "epoch": 1.9523017193566279, + "grad_norm": 6822.140625, + "learning_rate": 2.042833607907743e-05, + "loss": 3.1339, + "step": 440 + }, + { + "epoch": 1.956738768718802, + "grad_norm": 4385.99169921875, + "learning_rate": 2.0345963756177923e-05, + "loss": 3.1063, + "step": 441 + }, + { + "epoch": 1.9611758180809762, + "grad_norm": 4038.91015625, + "learning_rate": 2.026359143327842e-05, + "loss": 3.2365, + "step": 442 + }, + { + "epoch": 1.9656128674431503, + "grad_norm": 5223.08544921875, + "learning_rate": 2.0181219110378914e-05, + "loss": 2.824, + "step": 443 + }, + { + "epoch": 1.9700499168053245, + "grad_norm": 2778.4501953125, + "learning_rate": 2.009884678747941e-05, + "loss": 3.2645, + "step": 444 + }, + { + "epoch": 1.9744869661674986, + "grad_norm": 3180.86962890625, + "learning_rate": 2.0016474464579902e-05, + "loss": 3.0103, + "step": 445 + }, + { + "epoch": 1.9789240155296728, + "grad_norm": 5859.29345703125, + "learning_rate": 1.9934102141680397e-05, + "loss": 3.1216, + "step": 446 + }, + { + "epoch": 1.983361064891847, + "grad_norm": 4432.77392578125, + "learning_rate": 1.985172981878089e-05, + "loss": 3.1566, + "step": 447 + }, + { + "epoch": 1.987798114254021, + "grad_norm": 2626.3388671875, + "learning_rate": 1.9769357495881385e-05, + "loss": 3.1475, + "step": 448 + }, + { + "epoch": 1.9922351636161952, + "grad_norm": 4782.11572265625, + "learning_rate": 1.9686985172981877e-05, + "loss": 2.7534, + "step": 449 + }, + { + "epoch": 1.9966722129783694, + "grad_norm": 2261.256103515625, + "learning_rate": 1.9604612850082372e-05, + "loss": 3.0563, + "step": 450 + }, + { + "epoch": 2.0011092623405435, + "grad_norm": 2105.462158203125, + "learning_rate": 1.9522240527182868e-05, + "loss": 2.8962, + "step": 451 + }, + { + "epoch": 2.0055463117027177, + "grad_norm": 5077.935546875, + "learning_rate": 1.9439868204283363e-05, + "loss": 2.5199, + "step": 452 + }, + { + "epoch": 2.009983361064892, + "grad_norm": 2899.4248046875, + "learning_rate": 1.9357495881383855e-05, + "loss": 2.6137, + "step": 453 + }, + { + "epoch": 2.014420410427066, + "grad_norm": 2370.063232421875, + "learning_rate": 1.927512355848435e-05, + "loss": 2.7414, + "step": 454 + }, + { + "epoch": 2.01885745978924, + "grad_norm": 2607.39794921875, + "learning_rate": 1.9192751235584843e-05, + "loss": 2.6142, + "step": 455 + }, + { + "epoch": 2.0232945091514143, + "grad_norm": 4069.059814453125, + "learning_rate": 1.911037891268534e-05, + "loss": 2.5696, + "step": 456 + }, + { + "epoch": 2.0277315585135884, + "grad_norm": 5161.96240234375, + "learning_rate": 1.9028006589785834e-05, + "loss": 2.5722, + "step": 457 + }, + { + "epoch": 2.0321686078757626, + "grad_norm": 4082.345703125, + "learning_rate": 1.894563426688633e-05, + "loss": 2.5632, + "step": 458 + }, + { + "epoch": 2.0366056572379367, + "grad_norm": 6766.630859375, + "learning_rate": 1.886326194398682e-05, + "loss": 2.3776, + "step": 459 + }, + { + "epoch": 2.041042706600111, + "grad_norm": 8708.931640625, + "learning_rate": 1.8780889621087317e-05, + "loss": 2.6656, + "step": 460 + }, + { + "epoch": 2.045479755962285, + "grad_norm": 3315.59521484375, + "learning_rate": 1.869851729818781e-05, + "loss": 2.648, + "step": 461 + }, + { + "epoch": 2.049916805324459, + "grad_norm": 3886.526611328125, + "learning_rate": 1.8616144975288304e-05, + "loss": 2.7488, + "step": 462 + }, + { + "epoch": 2.0543538546866333, + "grad_norm": 3450.153564453125, + "learning_rate": 1.8533772652388796e-05, + "loss": 2.8399, + "step": 463 + }, + { + "epoch": 2.0587909040488075, + "grad_norm": 3993.089599609375, + "learning_rate": 1.8451400329489292e-05, + "loss": 2.6095, + "step": 464 + }, + { + "epoch": 2.0632279534109816, + "grad_norm": 3846.003662109375, + "learning_rate": 1.8369028006589787e-05, + "loss": 2.7819, + "step": 465 + }, + { + "epoch": 2.067665002773156, + "grad_norm": 5647.2919921875, + "learning_rate": 1.8286655683690283e-05, + "loss": 2.707, + "step": 466 + }, + { + "epoch": 2.07210205213533, + "grad_norm": 2751.223388671875, + "learning_rate": 1.8204283360790775e-05, + "loss": 2.8927, + "step": 467 + }, + { + "epoch": 2.076539101497504, + "grad_norm": 3433.56689453125, + "learning_rate": 1.812191103789127e-05, + "loss": 2.5743, + "step": 468 + }, + { + "epoch": 2.0809761508596782, + "grad_norm": 3154.43505859375, + "learning_rate": 1.8039538714991762e-05, + "loss": 2.4878, + "step": 469 + }, + { + "epoch": 2.0854132002218524, + "grad_norm": 4113.84326171875, + "learning_rate": 1.7957166392092258e-05, + "loss": 2.5541, + "step": 470 + }, + { + "epoch": 2.0898502495840265, + "grad_norm": 4650.24169921875, + "learning_rate": 1.7874794069192753e-05, + "loss": 2.6416, + "step": 471 + }, + { + "epoch": 2.0942872989462007, + "grad_norm": 6101.0302734375, + "learning_rate": 1.779242174629325e-05, + "loss": 2.5171, + "step": 472 + }, + { + "epoch": 2.098724348308375, + "grad_norm": 1755.3096923828125, + "learning_rate": 1.771004942339374e-05, + "loss": 2.7233, + "step": 473 + }, + { + "epoch": 2.103161397670549, + "grad_norm": 2609.34716796875, + "learning_rate": 1.7627677100494236e-05, + "loss": 2.6155, + "step": 474 + }, + { + "epoch": 2.107598447032723, + "grad_norm": 4076.533447265625, + "learning_rate": 1.754530477759473e-05, + "loss": 2.6176, + "step": 475 + }, + { + "epoch": 2.1120354963948973, + "grad_norm": 4886.240234375, + "learning_rate": 1.7462932454695224e-05, + "loss": 2.4924, + "step": 476 + }, + { + "epoch": 2.1164725457570714, + "grad_norm": 4825.35107421875, + "learning_rate": 1.7380560131795716e-05, + "loss": 2.7718, + "step": 477 + }, + { + "epoch": 2.1209095951192456, + "grad_norm": 5086.2119140625, + "learning_rate": 1.729818780889621e-05, + "loss": 2.6362, + "step": 478 + }, + { + "epoch": 2.1253466444814197, + "grad_norm": 2445.340087890625, + "learning_rate": 1.7215815485996707e-05, + "loss": 2.6348, + "step": 479 + }, + { + "epoch": 2.129783693843594, + "grad_norm": 4575.83349609375, + "learning_rate": 1.7133443163097202e-05, + "loss": 2.7851, + "step": 480 + }, + { + "epoch": 2.134220743205768, + "grad_norm": 3662.310791015625, + "learning_rate": 1.7051070840197694e-05, + "loss": 2.7279, + "step": 481 + }, + { + "epoch": 2.138657792567942, + "grad_norm": 2444.625244140625, + "learning_rate": 1.696869851729819e-05, + "loss": 2.6522, + "step": 482 + }, + { + "epoch": 2.1430948419301163, + "grad_norm": 1929.787841796875, + "learning_rate": 1.6886326194398682e-05, + "loss": 2.811, + "step": 483 + }, + { + "epoch": 2.1475318912922905, + "grad_norm": 4528.90869140625, + "learning_rate": 1.6803953871499177e-05, + "loss": 2.6628, + "step": 484 + }, + { + "epoch": 2.1519689406544646, + "grad_norm": 3910.12890625, + "learning_rate": 1.6721581548599673e-05, + "loss": 2.5904, + "step": 485 + }, + { + "epoch": 2.156405990016639, + "grad_norm": 2755.846923828125, + "learning_rate": 1.6639209225700165e-05, + "loss": 2.3434, + "step": 486 + }, + { + "epoch": 2.160843039378813, + "grad_norm": 3852.95458984375, + "learning_rate": 1.655683690280066e-05, + "loss": 2.5502, + "step": 487 + }, + { + "epoch": 2.165280088740987, + "grad_norm": 2147.770751953125, + "learning_rate": 1.6474464579901152e-05, + "loss": 2.7044, + "step": 488 + }, + { + "epoch": 2.1697171381031612, + "grad_norm": 3990.876708984375, + "learning_rate": 1.6392092257001648e-05, + "loss": 2.5242, + "step": 489 + }, + { + "epoch": 2.1741541874653354, + "grad_norm": 2413.63623046875, + "learning_rate": 1.630971993410214e-05, + "loss": 2.6773, + "step": 490 + }, + { + "epoch": 2.1785912368275095, + "grad_norm": 4493.80322265625, + "learning_rate": 1.6227347611202635e-05, + "loss": 2.2612, + "step": 491 + }, + { + "epoch": 2.1830282861896837, + "grad_norm": 3538.88427734375, + "learning_rate": 1.614497528830313e-05, + "loss": 2.72, + "step": 492 + }, + { + "epoch": 2.187465335551858, + "grad_norm": 10678.70703125, + "learning_rate": 1.6062602965403626e-05, + "loss": 2.6394, + "step": 493 + }, + { + "epoch": 2.191902384914032, + "grad_norm": 3382.633544921875, + "learning_rate": 1.598023064250412e-05, + "loss": 2.0755, + "step": 494 + }, + { + "epoch": 2.196339434276206, + "grad_norm": 5031.93408203125, + "learning_rate": 1.5897858319604614e-05, + "loss": 2.3705, + "step": 495 + }, + { + "epoch": 2.2007764836383803, + "grad_norm": 1957.3800048828125, + "learning_rate": 1.5815485996705106e-05, + "loss": 2.498, + "step": 496 + }, + { + "epoch": 2.2052135330005544, + "grad_norm": 4049.7880859375, + "learning_rate": 1.57331136738056e-05, + "loss": 2.7986, + "step": 497 + }, + { + "epoch": 2.2096505823627286, + "grad_norm": 2852.196533203125, + "learning_rate": 1.5650741350906097e-05, + "loss": 2.4517, + "step": 498 + }, + { + "epoch": 2.2140876317249027, + "grad_norm": 2434.55615234375, + "learning_rate": 1.5568369028006592e-05, + "loss": 2.6352, + "step": 499 + }, + { + "epoch": 2.218524681087077, + "grad_norm": 3371.0390625, + "learning_rate": 1.5485996705107084e-05, + "loss": 2.2291, + "step": 500 + }, + { + "epoch": 2.222961730449251, + "grad_norm": 4629.447265625, + "learning_rate": 1.540362438220758e-05, + "loss": 2.4771, + "step": 501 + }, + { + "epoch": 2.227398779811425, + "grad_norm": 3995.16796875, + "learning_rate": 1.5321252059308072e-05, + "loss": 2.536, + "step": 502 + }, + { + "epoch": 2.2318358291735994, + "grad_norm": 3601.61669921875, + "learning_rate": 1.5238879736408567e-05, + "loss": 2.6146, + "step": 503 + }, + { + "epoch": 2.2362728785357735, + "grad_norm": 5928.857421875, + "learning_rate": 1.5156507413509061e-05, + "loss": 2.8034, + "step": 504 + }, + { + "epoch": 2.2407099278979477, + "grad_norm": 3383.321044921875, + "learning_rate": 1.5074135090609557e-05, + "loss": 2.8632, + "step": 505 + }, + { + "epoch": 2.245146977260122, + "grad_norm": 14181.6611328125, + "learning_rate": 1.4991762767710049e-05, + "loss": 2.5724, + "step": 506 + }, + { + "epoch": 2.249584026622296, + "grad_norm": 8067.658203125, + "learning_rate": 1.4909390444810544e-05, + "loss": 2.6613, + "step": 507 + }, + { + "epoch": 2.25402107598447, + "grad_norm": 2069.533935546875, + "learning_rate": 1.4827018121911038e-05, + "loss": 2.9278, + "step": 508 + }, + { + "epoch": 2.2584581253466443, + "grad_norm": 3548.154541015625, + "learning_rate": 1.4744645799011533e-05, + "loss": 2.8747, + "step": 509 + }, + { + "epoch": 2.2628951747088184, + "grad_norm": 2253.08740234375, + "learning_rate": 1.4662273476112027e-05, + "loss": 2.6971, + "step": 510 + }, + { + "epoch": 2.2673322240709926, + "grad_norm": 7061.1533203125, + "learning_rate": 1.4579901153212522e-05, + "loss": 2.4785, + "step": 511 + }, + { + "epoch": 2.2717692734331667, + "grad_norm": 2889.0419921875, + "learning_rate": 1.4497528830313015e-05, + "loss": 2.4674, + "step": 512 + }, + { + "epoch": 2.276206322795341, + "grad_norm": 6647.66455078125, + "learning_rate": 1.441515650741351e-05, + "loss": 2.6839, + "step": 513 + }, + { + "epoch": 2.280643372157515, + "grad_norm": 3707.956298828125, + "learning_rate": 1.4332784184514004e-05, + "loss": 2.4873, + "step": 514 + }, + { + "epoch": 2.2850804215196896, + "grad_norm": 7174.60302734375, + "learning_rate": 1.42504118616145e-05, + "loss": 2.6425, + "step": 515 + }, + { + "epoch": 2.2895174708818637, + "grad_norm": 2339.060302734375, + "learning_rate": 1.4168039538714991e-05, + "loss": 2.7469, + "step": 516 + }, + { + "epoch": 2.293954520244038, + "grad_norm": 2994.836181640625, + "learning_rate": 1.4085667215815487e-05, + "loss": 2.4227, + "step": 517 + }, + { + "epoch": 2.298391569606212, + "grad_norm": 5566.3896484375, + "learning_rate": 1.400329489291598e-05, + "loss": 2.6391, + "step": 518 + }, + { + "epoch": 2.302828618968386, + "grad_norm": 2693.664794921875, + "learning_rate": 1.3920922570016476e-05, + "loss": 2.5954, + "step": 519 + }, + { + "epoch": 2.3072656683305603, + "grad_norm": 3443.02734375, + "learning_rate": 1.3838550247116968e-05, + "loss": 2.484, + "step": 520 + }, + { + "epoch": 2.3117027176927345, + "grad_norm": 5970.4775390625, + "learning_rate": 1.3756177924217463e-05, + "loss": 2.7316, + "step": 521 + }, + { + "epoch": 2.3161397670549086, + "grad_norm": 2575.7236328125, + "learning_rate": 1.3673805601317957e-05, + "loss": 2.484, + "step": 522 + }, + { + "epoch": 2.320576816417083, + "grad_norm": 6025.35107421875, + "learning_rate": 1.3591433278418453e-05, + "loss": 2.6442, + "step": 523 + }, + { + "epoch": 2.325013865779257, + "grad_norm": 4172.57177734375, + "learning_rate": 1.3509060955518946e-05, + "loss": 2.5493, + "step": 524 + }, + { + "epoch": 2.329450915141431, + "grad_norm": 2729.989013671875, + "learning_rate": 1.3426688632619442e-05, + "loss": 2.452, + "step": 525 + }, + { + "epoch": 2.3338879645036053, + "grad_norm": 5065.46435546875, + "learning_rate": 1.3344316309719934e-05, + "loss": 2.4446, + "step": 526 + }, + { + "epoch": 2.3383250138657794, + "grad_norm": 5649.47314453125, + "learning_rate": 1.326194398682043e-05, + "loss": 2.6647, + "step": 527 + }, + { + "epoch": 2.3427620632279536, + "grad_norm": 2708.152587890625, + "learning_rate": 1.3179571663920923e-05, + "loss": 2.601, + "step": 528 + }, + { + "epoch": 2.3471991125901277, + "grad_norm": 6233.21923828125, + "learning_rate": 1.3097199341021419e-05, + "loss": 2.5738, + "step": 529 + }, + { + "epoch": 2.351636161952302, + "grad_norm": 2998.24169921875, + "learning_rate": 1.301482701812191e-05, + "loss": 2.4114, + "step": 530 + }, + { + "epoch": 2.356073211314476, + "grad_norm": 2275.528076171875, + "learning_rate": 1.2932454695222406e-05, + "loss": 2.5962, + "step": 531 + }, + { + "epoch": 2.36051026067665, + "grad_norm": 6748.6865234375, + "learning_rate": 1.28500823723229e-05, + "loss": 2.5733, + "step": 532 + }, + { + "epoch": 2.3649473100388243, + "grad_norm": 3575.512939453125, + "learning_rate": 1.2767710049423395e-05, + "loss": 2.3462, + "step": 533 + }, + { + "epoch": 2.3693843594009985, + "grad_norm": 6557.375, + "learning_rate": 1.2685337726523888e-05, + "loss": 2.6404, + "step": 534 + }, + { + "epoch": 2.3738214087631726, + "grad_norm": 2687.523681640625, + "learning_rate": 1.2602965403624385e-05, + "loss": 2.4523, + "step": 535 + }, + { + "epoch": 2.3782584581253468, + "grad_norm": 7146.51220703125, + "learning_rate": 1.2520593080724877e-05, + "loss": 2.7674, + "step": 536 + }, + { + "epoch": 2.382695507487521, + "grad_norm": 4911.45556640625, + "learning_rate": 1.243822075782537e-05, + "loss": 2.7356, + "step": 537 + }, + { + "epoch": 2.387132556849695, + "grad_norm": 3032.93994140625, + "learning_rate": 1.2355848434925866e-05, + "loss": 2.2574, + "step": 538 + }, + { + "epoch": 2.391569606211869, + "grad_norm": 3129.145751953125, + "learning_rate": 1.227347611202636e-05, + "loss": 2.907, + "step": 539 + }, + { + "epoch": 2.3960066555740434, + "grad_norm": 4832.2109375, + "learning_rate": 1.2191103789126853e-05, + "loss": 2.488, + "step": 540 + }, + { + "epoch": 2.4004437049362175, + "grad_norm": 5117.1552734375, + "learning_rate": 1.2108731466227347e-05, + "loss": 2.5948, + "step": 541 + }, + { + "epoch": 2.4048807542983917, + "grad_norm": 3100.830322265625, + "learning_rate": 1.2026359143327843e-05, + "loss": 2.6073, + "step": 542 + }, + { + "epoch": 2.409317803660566, + "grad_norm": 2604.384521484375, + "learning_rate": 1.1943986820428336e-05, + "loss": 2.3613, + "step": 543 + }, + { + "epoch": 2.41375485302274, + "grad_norm": 6259.60595703125, + "learning_rate": 1.186161449752883e-05, + "loss": 2.3132, + "step": 544 + }, + { + "epoch": 2.418191902384914, + "grad_norm": 10345.0673828125, + "learning_rate": 1.1779242174629326e-05, + "loss": 2.5327, + "step": 545 + }, + { + "epoch": 2.4226289517470883, + "grad_norm": 8641.677734375, + "learning_rate": 1.169686985172982e-05, + "loss": 2.5945, + "step": 546 + }, + { + "epoch": 2.4270660011092624, + "grad_norm": 2760.537353515625, + "learning_rate": 1.1614497528830313e-05, + "loss": 2.5316, + "step": 547 + }, + { + "epoch": 2.4315030504714366, + "grad_norm": 2161.694580078125, + "learning_rate": 1.1532125205930807e-05, + "loss": 2.6453, + "step": 548 + }, + { + "epoch": 2.4359400998336107, + "grad_norm": 3030.161376953125, + "learning_rate": 1.1449752883031302e-05, + "loss": 2.3972, + "step": 549 + }, + { + "epoch": 2.440377149195785, + "grad_norm": 4903.71484375, + "learning_rate": 1.1367380560131796e-05, + "loss": 2.4047, + "step": 550 + }, + { + "epoch": 2.444814198557959, + "grad_norm": 3894.473876953125, + "learning_rate": 1.128500823723229e-05, + "loss": 2.5555, + "step": 551 + }, + { + "epoch": 2.449251247920133, + "grad_norm": 2659.10888671875, + "learning_rate": 1.1202635914332785e-05, + "loss": 2.4425, + "step": 552 + }, + { + "epoch": 2.4536882972823073, + "grad_norm": 3052.740966796875, + "learning_rate": 1.112026359143328e-05, + "loss": 2.3797, + "step": 553 + }, + { + "epoch": 2.4581253466444815, + "grad_norm": 2440.5185546875, + "learning_rate": 1.1037891268533773e-05, + "loss": 2.4072, + "step": 554 + }, + { + "epoch": 2.4625623960066556, + "grad_norm": 8493.0703125, + "learning_rate": 1.0955518945634267e-05, + "loss": 2.2283, + "step": 555 + }, + { + "epoch": 2.4669994453688298, + "grad_norm": 2350.975830078125, + "learning_rate": 1.0873146622734762e-05, + "loss": 2.3807, + "step": 556 + }, + { + "epoch": 2.471436494731004, + "grad_norm": 2622.6875, + "learning_rate": 1.0790774299835256e-05, + "loss": 2.7101, + "step": 557 + }, + { + "epoch": 2.475873544093178, + "grad_norm": 3066.93603515625, + "learning_rate": 1.070840197693575e-05, + "loss": 2.3996, + "step": 558 + }, + { + "epoch": 2.480310593455352, + "grad_norm": 2955.510498046875, + "learning_rate": 1.0626029654036245e-05, + "loss": 2.4769, + "step": 559 + }, + { + "epoch": 2.4847476428175264, + "grad_norm": 2765.87109375, + "learning_rate": 1.0543657331136739e-05, + "loss": 2.6356, + "step": 560 + }, + { + "epoch": 2.4891846921797005, + "grad_norm": 5913.759765625, + "learning_rate": 1.0461285008237233e-05, + "loss": 2.5294, + "step": 561 + }, + { + "epoch": 2.4936217415418747, + "grad_norm": 2482.247314453125, + "learning_rate": 1.0378912685337726e-05, + "loss": 2.3878, + "step": 562 + }, + { + "epoch": 2.498058790904049, + "grad_norm": 2827.131103515625, + "learning_rate": 1.0296540362438222e-05, + "loss": 2.8051, + "step": 563 + }, + { + "epoch": 2.502495840266223, + "grad_norm": 1647.5404052734375, + "learning_rate": 1.0214168039538716e-05, + "loss": 2.3309, + "step": 564 + }, + { + "epoch": 2.506932889628397, + "grad_norm": 5386.03125, + "learning_rate": 1.013179571663921e-05, + "loss": 2.5166, + "step": 565 + }, + { + "epoch": 2.5113699389905713, + "grad_norm": 9112.6005859375, + "learning_rate": 1.0049423393739705e-05, + "loss": 2.6083, + "step": 566 + }, + { + "epoch": 2.5158069883527454, + "grad_norm": 3959.880126953125, + "learning_rate": 9.967051070840199e-06, + "loss": 2.518, + "step": 567 + }, + { + "epoch": 2.5202440377149196, + "grad_norm": 4645.970703125, + "learning_rate": 9.884678747940692e-06, + "loss": 2.5157, + "step": 568 + }, + { + "epoch": 2.5246810870770937, + "grad_norm": 3434.185302734375, + "learning_rate": 9.802306425041186e-06, + "loss": 2.5485, + "step": 569 + }, + { + "epoch": 2.529118136439268, + "grad_norm": 2623.41943359375, + "learning_rate": 9.719934102141682e-06, + "loss": 2.4926, + "step": 570 + }, + { + "epoch": 2.533555185801442, + "grad_norm": 6182.4814453125, + "learning_rate": 9.637561779242175e-06, + "loss": 2.2396, + "step": 571 + }, + { + "epoch": 2.537992235163616, + "grad_norm": 2492.56787109375, + "learning_rate": 9.55518945634267e-06, + "loss": 2.781, + "step": 572 + }, + { + "epoch": 2.5424292845257903, + "grad_norm": 3153.014892578125, + "learning_rate": 9.472817133443165e-06, + "loss": 2.5434, + "step": 573 + }, + { + "epoch": 2.5468663338879645, + "grad_norm": 3527.29931640625, + "learning_rate": 9.390444810543658e-06, + "loss": 2.7198, + "step": 574 + }, + { + "epoch": 2.5513033832501386, + "grad_norm": 10354.1708984375, + "learning_rate": 9.308072487644152e-06, + "loss": 2.4005, + "step": 575 + }, + { + "epoch": 2.5557404326123128, + "grad_norm": 2184.916015625, + "learning_rate": 9.225700164744646e-06, + "loss": 2.3881, + "step": 576 + }, + { + "epoch": 2.560177481974487, + "grad_norm": 3189.3427734375, + "learning_rate": 9.143327841845141e-06, + "loss": 2.5018, + "step": 577 + }, + { + "epoch": 2.564614531336661, + "grad_norm": 5138.06787109375, + "learning_rate": 9.060955518945635e-06, + "loss": 2.4031, + "step": 578 + }, + { + "epoch": 2.569051580698835, + "grad_norm": 2659.936279296875, + "learning_rate": 8.978583196046129e-06, + "loss": 2.5182, + "step": 579 + }, + { + "epoch": 2.5734886300610094, + "grad_norm": 2315.149658203125, + "learning_rate": 8.896210873146624e-06, + "loss": 2.3928, + "step": 580 + }, + { + "epoch": 2.5779256794231835, + "grad_norm": 1863.8740234375, + "learning_rate": 8.813838550247118e-06, + "loss": 2.6421, + "step": 581 + }, + { + "epoch": 2.5823627287853577, + "grad_norm": 4491.3642578125, + "learning_rate": 8.731466227347612e-06, + "loss": 2.5438, + "step": 582 + }, + { + "epoch": 2.586799778147532, + "grad_norm": 9824.6298828125, + "learning_rate": 8.649093904448106e-06, + "loss": 2.6321, + "step": 583 + }, + { + "epoch": 2.591236827509706, + "grad_norm": 3020.216796875, + "learning_rate": 8.566721581548601e-06, + "loss": 2.4633, + "step": 584 + }, + { + "epoch": 2.59567387687188, + "grad_norm": 1789.668212890625, + "learning_rate": 8.484349258649095e-06, + "loss": 2.3656, + "step": 585 + }, + { + "epoch": 2.6001109262340543, + "grad_norm": 2035.4766845703125, + "learning_rate": 8.401976935749589e-06, + "loss": 2.5788, + "step": 586 + }, + { + "epoch": 2.6045479755962284, + "grad_norm": 7690.67578125, + "learning_rate": 8.319604612850082e-06, + "loss": 2.312, + "step": 587 + }, + { + "epoch": 2.6089850249584026, + "grad_norm": 7675.25146484375, + "learning_rate": 8.237232289950576e-06, + "loss": 2.5116, + "step": 588 + }, + { + "epoch": 2.6134220743205767, + "grad_norm": 5110.587890625, + "learning_rate": 8.15485996705107e-06, + "loss": 2.4912, + "step": 589 + }, + { + "epoch": 2.617859123682751, + "grad_norm": 3645.31298828125, + "learning_rate": 8.072487644151565e-06, + "loss": 2.6615, + "step": 590 + }, + { + "epoch": 2.622296173044925, + "grad_norm": 3086.28076171875, + "learning_rate": 7.99011532125206e-06, + "loss": 2.681, + "step": 591 + }, + { + "epoch": 2.626733222407099, + "grad_norm": 6240.58349609375, + "learning_rate": 7.907742998352553e-06, + "loss": 2.3262, + "step": 592 + }, + { + "epoch": 2.6311702717692733, + "grad_norm": 3572.874755859375, + "learning_rate": 7.825370675453048e-06, + "loss": 2.5625, + "step": 593 + }, + { + "epoch": 2.6356073211314475, + "grad_norm": 2021.5216064453125, + "learning_rate": 7.742998352553542e-06, + "loss": 2.5823, + "step": 594 + }, + { + "epoch": 2.6400443704936216, + "grad_norm": 1322.3262939453125, + "learning_rate": 7.660626029654036e-06, + "loss": 2.6832, + "step": 595 + }, + { + "epoch": 2.6444814198557958, + "grad_norm": 2490.99658203125, + "learning_rate": 7.5782537067545305e-06, + "loss": 2.379, + "step": 596 + }, + { + "epoch": 2.64891846921797, + "grad_norm": 4689.4189453125, + "learning_rate": 7.495881383855024e-06, + "loss": 2.5766, + "step": 597 + }, + { + "epoch": 2.653355518580144, + "grad_norm": 7085.703125, + "learning_rate": 7.413509060955519e-06, + "loss": 2.3803, + "step": 598 + }, + { + "epoch": 2.657792567942318, + "grad_norm": 9480.5029296875, + "learning_rate": 7.3311367380560135e-06, + "loss": 2.2704, + "step": 599 + }, + { + "epoch": 2.6622296173044924, + "grad_norm": 2268.37353515625, + "learning_rate": 7.248764415156507e-06, + "loss": 2.4718, + "step": 600 + }, + { + "epoch": 2.6666666666666665, + "grad_norm": 3479.536376953125, + "learning_rate": 7.166392092257002e-06, + "loss": 2.6211, + "step": 601 + }, + { + "epoch": 2.6711037160288407, + "grad_norm": 5927.5771484375, + "learning_rate": 7.084019769357496e-06, + "loss": 2.7763, + "step": 602 + }, + { + "epoch": 2.675540765391015, + "grad_norm": 3893.068115234375, + "learning_rate": 7.00164744645799e-06, + "loss": 2.4849, + "step": 603 + }, + { + "epoch": 2.679977814753189, + "grad_norm": 1726.7548828125, + "learning_rate": 6.919275123558484e-06, + "loss": 2.6186, + "step": 604 + }, + { + "epoch": 2.684414864115363, + "grad_norm": 1768.3822021484375, + "learning_rate": 6.836902800658979e-06, + "loss": 2.3795, + "step": 605 + }, + { + "epoch": 2.6888519134775377, + "grad_norm": 2237.384521484375, + "learning_rate": 6.754530477759473e-06, + "loss": 2.5742, + "step": 606 + }, + { + "epoch": 2.693288962839712, + "grad_norm": 3616.79931640625, + "learning_rate": 6.672158154859967e-06, + "loss": 2.5871, + "step": 607 + }, + { + "epoch": 2.697726012201886, + "grad_norm": 5400.53369140625, + "learning_rate": 6.589785831960462e-06, + "loss": 2.5162, + "step": 608 + }, + { + "epoch": 2.70216306156406, + "grad_norm": 4567.24853515625, + "learning_rate": 6.507413509060955e-06, + "loss": 2.3241, + "step": 609 + }, + { + "epoch": 2.7066001109262343, + "grad_norm": 3562.23486328125, + "learning_rate": 6.42504118616145e-06, + "loss": 2.578, + "step": 610 + }, + { + "epoch": 2.7110371602884085, + "grad_norm": 7854.82568359375, + "learning_rate": 6.342668863261944e-06, + "loss": 2.6906, + "step": 611 + }, + { + "epoch": 2.7154742096505826, + "grad_norm": 8927.4384765625, + "learning_rate": 6.260296540362438e-06, + "loss": 2.6397, + "step": 612 + }, + { + "epoch": 2.7199112590127568, + "grad_norm": 9978.86328125, + "learning_rate": 6.177924217462933e-06, + "loss": 2.4238, + "step": 613 + }, + { + "epoch": 2.724348308374931, + "grad_norm": 4883.0107421875, + "learning_rate": 6.095551894563427e-06, + "loss": 2.5803, + "step": 614 + }, + { + "epoch": 2.728785357737105, + "grad_norm": 7069.74169921875, + "learning_rate": 6.013179571663921e-06, + "loss": 2.5891, + "step": 615 + }, + { + "epoch": 2.733222407099279, + "grad_norm": 2908.210693359375, + "learning_rate": 5.930807248764415e-06, + "loss": 2.4699, + "step": 616 + }, + { + "epoch": 2.7376594564614534, + "grad_norm": 5147.314453125, + "learning_rate": 5.84843492586491e-06, + "loss": 2.2404, + "step": 617 + }, + { + "epoch": 2.7420965058236275, + "grad_norm": 9169.95703125, + "learning_rate": 5.7660626029654035e-06, + "loss": 2.5569, + "step": 618 + }, + { + "epoch": 2.7465335551858017, + "grad_norm": 3353.639892578125, + "learning_rate": 5.683690280065898e-06, + "loss": 2.6296, + "step": 619 + }, + { + "epoch": 2.750970604547976, + "grad_norm": 6089.81982421875, + "learning_rate": 5.601317957166393e-06, + "loss": 2.4212, + "step": 620 + }, + { + "epoch": 2.75540765391015, + "grad_norm": 2166.770263671875, + "learning_rate": 5.5189456342668865e-06, + "loss": 2.6596, + "step": 621 + }, + { + "epoch": 2.759844703272324, + "grad_norm": 2501.936767578125, + "learning_rate": 5.436573311367381e-06, + "loss": 2.5837, + "step": 622 + }, + { + "epoch": 2.7642817526344983, + "grad_norm": 9360.7138671875, + "learning_rate": 5.354200988467875e-06, + "loss": 2.7385, + "step": 623 + }, + { + "epoch": 2.7687188019966724, + "grad_norm": 2987.460205078125, + "learning_rate": 5.2718286655683695e-06, + "loss": 2.5007, + "step": 624 + }, + { + "epoch": 2.7731558513588466, + "grad_norm": 2465.62841796875, + "learning_rate": 5.189456342668863e-06, + "loss": 2.4872, + "step": 625 + }, + { + "epoch": 2.7775929007210207, + "grad_norm": 6314.24462890625, + "learning_rate": 5.107084019769358e-06, + "loss": 2.4645, + "step": 626 + }, + { + "epoch": 2.782029950083195, + "grad_norm": 4423.83642578125, + "learning_rate": 5.0247116968698524e-06, + "loss": 2.4219, + "step": 627 + }, + { + "epoch": 2.786466999445369, + "grad_norm": 3497.298095703125, + "learning_rate": 4.942339373970346e-06, + "loss": 2.5329, + "step": 628 + }, + { + "epoch": 2.790904048807543, + "grad_norm": 2176.44287109375, + "learning_rate": 4.859967051070841e-06, + "loss": 2.4151, + "step": 629 + }, + { + "epoch": 2.7953410981697173, + "grad_norm": 4388.32568359375, + "learning_rate": 4.777594728171335e-06, + "loss": 2.6517, + "step": 630 + }, + { + "epoch": 2.7997781475318915, + "grad_norm": 2107.5380859375, + "learning_rate": 4.695222405271829e-06, + "loss": 2.4082, + "step": 631 + }, + { + "epoch": 2.8042151968940656, + "grad_norm": 3466.89208984375, + "learning_rate": 4.612850082372323e-06, + "loss": 2.3583, + "step": 632 + }, + { + "epoch": 2.8086522462562398, + "grad_norm": 2463.3896484375, + "learning_rate": 4.5304777594728176e-06, + "loss": 2.4841, + "step": 633 + }, + { + "epoch": 2.813089295618414, + "grad_norm": 6578.353515625, + "learning_rate": 4.448105436573312e-06, + "loss": 2.5942, + "step": 634 + }, + { + "epoch": 2.817526344980588, + "grad_norm": 4899.65771484375, + "learning_rate": 4.365733113673806e-06, + "loss": 2.4171, + "step": 635 + }, + { + "epoch": 2.821963394342762, + "grad_norm": 2415.186767578125, + "learning_rate": 4.2833607907743006e-06, + "loss": 2.522, + "step": 636 + }, + { + "epoch": 2.8264004437049364, + "grad_norm": 3585.32080078125, + "learning_rate": 4.200988467874794e-06, + "loss": 2.404, + "step": 637 + }, + { + "epoch": 2.8308374930671105, + "grad_norm": 4646.6572265625, + "learning_rate": 4.118616144975288e-06, + "loss": 2.5074, + "step": 638 + }, + { + "epoch": 2.8352745424292847, + "grad_norm": 2135.860107421875, + "learning_rate": 4.036243822075783e-06, + "loss": 2.4504, + "step": 639 + }, + { + "epoch": 2.839711591791459, + "grad_norm": 2136.07666015625, + "learning_rate": 3.9538714991762765e-06, + "loss": 2.4772, + "step": 640 + }, + { + "epoch": 2.844148641153633, + "grad_norm": 3418.1103515625, + "learning_rate": 3.871499176276771e-06, + "loss": 2.4926, + "step": 641 + }, + { + "epoch": 2.848585690515807, + "grad_norm": 2338.78466796875, + "learning_rate": 3.7891268533772653e-06, + "loss": 2.4929, + "step": 642 + }, + { + "epoch": 2.8530227398779813, + "grad_norm": 2875.1689453125, + "learning_rate": 3.7067545304777594e-06, + "loss": 2.3233, + "step": 643 + }, + { + "epoch": 2.8574597892401554, + "grad_norm": 7241.99560546875, + "learning_rate": 3.6243822075782536e-06, + "loss": 2.327, + "step": 644 + }, + { + "epoch": 2.8618968386023296, + "grad_norm": 4670.65625, + "learning_rate": 3.542009884678748e-06, + "loss": 2.3962, + "step": 645 + }, + { + "epoch": 2.8663338879645037, + "grad_norm": 3531.360595703125, + "learning_rate": 3.459637561779242e-06, + "loss": 2.3598, + "step": 646 + }, + { + "epoch": 2.870770937326678, + "grad_norm": 3573.98779296875, + "learning_rate": 3.3772652388797366e-06, + "loss": 2.6329, + "step": 647 + }, + { + "epoch": 2.875207986688852, + "grad_norm": 5320.9638671875, + "learning_rate": 3.294892915980231e-06, + "loss": 2.5497, + "step": 648 + }, + { + "epoch": 2.879645036051026, + "grad_norm": 4809.13818359375, + "learning_rate": 3.212520593080725e-06, + "loss": 2.4882, + "step": 649 + }, + { + "epoch": 2.8840820854132003, + "grad_norm": 3600.4560546875, + "learning_rate": 3.130148270181219e-06, + "loss": 2.5207, + "step": 650 + }, + { + "epoch": 2.8885191347753745, + "grad_norm": 3297.944580078125, + "learning_rate": 3.0477759472817134e-06, + "loss": 2.3507, + "step": 651 + }, + { + "epoch": 2.8929561841375486, + "grad_norm": 4598.22314453125, + "learning_rate": 2.9654036243822076e-06, + "loss": 2.3764, + "step": 652 + }, + { + "epoch": 2.8973932334997228, + "grad_norm": 6843.4775390625, + "learning_rate": 2.8830313014827017e-06, + "loss": 2.3216, + "step": 653 + }, + { + "epoch": 2.901830282861897, + "grad_norm": 3002.115234375, + "learning_rate": 2.8006589785831964e-06, + "loss": 2.5343, + "step": 654 + }, + { + "epoch": 2.906267332224071, + "grad_norm": 2805.712646484375, + "learning_rate": 2.7182866556836905e-06, + "loss": 2.2594, + "step": 655 + }, + { + "epoch": 2.9107043815862452, + "grad_norm": 1720.480224609375, + "learning_rate": 2.6359143327841847e-06, + "loss": 2.5233, + "step": 656 + }, + { + "epoch": 2.9151414309484194, + "grad_norm": 6596.97998046875, + "learning_rate": 2.553542009884679e-06, + "loss": 2.1236, + "step": 657 + }, + { + "epoch": 2.9195784803105935, + "grad_norm": 2605.771728515625, + "learning_rate": 2.471169686985173e-06, + "loss": 2.5335, + "step": 658 + }, + { + "epoch": 2.9240155296727677, + "grad_norm": 6402.65380859375, + "learning_rate": 2.3887973640856673e-06, + "loss": 2.5684, + "step": 659 + }, + { + "epoch": 2.928452579034942, + "grad_norm": 3226.62939453125, + "learning_rate": 2.3064250411861615e-06, + "loss": 2.628, + "step": 660 + }, + { + "epoch": 2.932889628397116, + "grad_norm": 1788.3135986328125, + "learning_rate": 2.224052718286656e-06, + "loss": 2.5518, + "step": 661 + }, + { + "epoch": 2.93732667775929, + "grad_norm": 3982.594482421875, + "learning_rate": 2.1416803953871503e-06, + "loss": 2.3978, + "step": 662 + }, + { + "epoch": 2.9417637271214643, + "grad_norm": 2510.217529296875, + "learning_rate": 2.059308072487644e-06, + "loss": 2.4756, + "step": 663 + }, + { + "epoch": 2.9462007764836384, + "grad_norm": 3606.79296875, + "learning_rate": 1.9769357495881382e-06, + "loss": 2.567, + "step": 664 + }, + { + "epoch": 2.9506378258458126, + "grad_norm": 4923.14453125, + "learning_rate": 1.8945634266886326e-06, + "loss": 2.2632, + "step": 665 + }, + { + "epoch": 2.9550748752079867, + "grad_norm": 2678.0048828125, + "learning_rate": 1.8121911037891268e-06, + "loss": 2.5776, + "step": 666 + }, + { + "epoch": 2.959511924570161, + "grad_norm": 3245.629638671875, + "learning_rate": 1.729818780889621e-06, + "loss": 2.3996, + "step": 667 + }, + { + "epoch": 2.963948973932335, + "grad_norm": 2143.85986328125, + "learning_rate": 1.6474464579901154e-06, + "loss": 2.5184, + "step": 668 + }, + { + "epoch": 2.968386023294509, + "grad_norm": 3475.8603515625, + "learning_rate": 1.5650741350906096e-06, + "loss": 2.441, + "step": 669 + }, + { + "epoch": 2.9728230726566833, + "grad_norm": 3787.5869140625, + "learning_rate": 1.4827018121911038e-06, + "loss": 2.4782, + "step": 670 + }, + { + "epoch": 2.9772601220188575, + "grad_norm": 3680.81982421875, + "learning_rate": 1.4003294892915982e-06, + "loss": 2.3911, + "step": 671 + }, + { + "epoch": 2.9816971713810316, + "grad_norm": 1910.5115966796875, + "learning_rate": 1.3179571663920924e-06, + "loss": 2.4402, + "step": 672 + }, + { + "epoch": 2.986134220743206, + "grad_norm": 1568.6151123046875, + "learning_rate": 1.2355848434925866e-06, + "loss": 2.7283, + "step": 673 + }, + { + "epoch": 2.99057127010538, + "grad_norm": 6588.1650390625, + "learning_rate": 1.1532125205930807e-06, + "loss": 2.3555, + "step": 674 + }, + { + "epoch": 2.995008319467554, + "grad_norm": 3463.969482421875, + "learning_rate": 1.0708401976935751e-06, + "loss": 2.471, + "step": 675 + }, + { + "epoch": 2.995008319467554, + "step": 675, + "total_flos": 2.621312257396572e+17, + "train_loss": 3.602762575502749, + "train_runtime": 19564.7512, + "train_samples_per_second": 1.106, + "train_steps_per_second": 0.035 + } + ], + "logging_steps": 1.0, + "max_steps": 675, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 2.621312257396572e+17, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}