| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.9952038369304557, | |
| "eval_steps": 500, | |
| "global_step": 208, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.009592326139088728, | |
| "grad_norm": 2.6591761112213135, | |
| "learning_rate": 9.523809523809525e-07, | |
| "loss": 0.7408, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.019184652278177457, | |
| "grad_norm": 2.5914506912231445, | |
| "learning_rate": 1.904761904761905e-06, | |
| "loss": 0.7838, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.02877697841726619, | |
| "grad_norm": 2.4312615394592285, | |
| "learning_rate": 2.8571428571428573e-06, | |
| "loss": 0.7863, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.03836930455635491, | |
| "grad_norm": 2.2820498943328857, | |
| "learning_rate": 3.80952380952381e-06, | |
| "loss": 0.775, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.047961630695443645, | |
| "grad_norm": 1.923662543296814, | |
| "learning_rate": 4.761904761904762e-06, | |
| "loss": 0.7394, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.05755395683453238, | |
| "grad_norm": 1.4304736852645874, | |
| "learning_rate": 5.7142857142857145e-06, | |
| "loss": 0.7485, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0671462829736211, | |
| "grad_norm": 1.3127758502960205, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.7022, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.07673860911270983, | |
| "grad_norm": 1.3539044857025146, | |
| "learning_rate": 7.61904761904762e-06, | |
| "loss": 0.6913, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.08633093525179857, | |
| "grad_norm": 1.3196394443511963, | |
| "learning_rate": 8.571428571428571e-06, | |
| "loss": 0.7041, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.09592326139088729, | |
| "grad_norm": 1.1037274599075317, | |
| "learning_rate": 9.523809523809525e-06, | |
| "loss": 0.6186, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.10551558752997602, | |
| "grad_norm": 1.3987900018692017, | |
| "learning_rate": 1.0476190476190477e-05, | |
| "loss": 0.6306, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.11510791366906475, | |
| "grad_norm": 0.9192888736724854, | |
| "learning_rate": 1.1428571428571429e-05, | |
| "loss": 0.6073, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.12470023980815348, | |
| "grad_norm": 0.7664029002189636, | |
| "learning_rate": 1.2380952380952383e-05, | |
| "loss": 0.6209, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.1342925659472422, | |
| "grad_norm": 1.2668217420578003, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.6077, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.14388489208633093, | |
| "grad_norm": 1.1923428773880005, | |
| "learning_rate": 1.4285714285714287e-05, | |
| "loss": 0.5864, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.15347721822541965, | |
| "grad_norm": 0.8919103145599365, | |
| "learning_rate": 1.523809523809524e-05, | |
| "loss": 0.5837, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.1630695443645084, | |
| "grad_norm": 0.8827866911888123, | |
| "learning_rate": 1.6190476190476193e-05, | |
| "loss": 0.5701, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.17266187050359713, | |
| "grad_norm": 0.9107728600502014, | |
| "learning_rate": 1.7142857142857142e-05, | |
| "loss": 0.5569, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.18225419664268586, | |
| "grad_norm": 0.8279157876968384, | |
| "learning_rate": 1.8095238095238097e-05, | |
| "loss": 0.5824, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.19184652278177458, | |
| "grad_norm": 0.5704746842384338, | |
| "learning_rate": 1.904761904761905e-05, | |
| "loss": 0.5305, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.2014388489208633, | |
| "grad_norm": 0.788601815700531, | |
| "learning_rate": 2e-05, | |
| "loss": 0.5451, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.21103117505995203, | |
| "grad_norm": 0.9100444912910461, | |
| "learning_rate": 1.9998588839790777e-05, | |
| "loss": 0.5611, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.22062350119904076, | |
| "grad_norm": 0.529228925704956, | |
| "learning_rate": 1.999435575743774e-05, | |
| "loss": 0.5204, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.2302158273381295, | |
| "grad_norm": 0.5239225625991821, | |
| "learning_rate": 1.9987301947652354e-05, | |
| "loss": 0.5459, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.23980815347721823, | |
| "grad_norm": 0.6502991914749146, | |
| "learning_rate": 1.9977429401245764e-05, | |
| "loss": 0.5501, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.24940047961630696, | |
| "grad_norm": 0.5559574365615845, | |
| "learning_rate": 1.9964740904566903e-05, | |
| "loss": 0.5219, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.2589928057553957, | |
| "grad_norm": 0.5296733379364014, | |
| "learning_rate": 1.9949240038716092e-05, | |
| "loss": 0.5043, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.2685851318944844, | |
| "grad_norm": 0.4866749942302704, | |
| "learning_rate": 1.9930931178534353e-05, | |
| "loss": 0.5106, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.27817745803357313, | |
| "grad_norm": 1.6436631679534912, | |
| "learning_rate": 1.9909819491368677e-05, | |
| "loss": 0.5059, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.28776978417266186, | |
| "grad_norm": 0.5561839938163757, | |
| "learning_rate": 1.988591093561364e-05, | |
| "loss": 0.5209, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.2973621103117506, | |
| "grad_norm": 0.4857185482978821, | |
| "learning_rate": 1.985921225902975e-05, | |
| "loss": 0.5173, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.3069544364508393, | |
| "grad_norm": 0.41804176568984985, | |
| "learning_rate": 1.982973099683902e-05, | |
| "loss": 0.4928, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.31654676258992803, | |
| "grad_norm": 0.49216267466545105, | |
| "learning_rate": 1.9797475469598267e-05, | |
| "loss": 0.4993, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.3261390887290168, | |
| "grad_norm": 0.4702392518520355, | |
| "learning_rate": 1.9762454780850807e-05, | |
| "loss": 0.4934, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.33573141486810554, | |
| "grad_norm": 0.4079640507698059, | |
| "learning_rate": 1.972467881455713e-05, | |
| "loss": 0.5058, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.34532374100719426, | |
| "grad_norm": 0.46476152539253235, | |
| "learning_rate": 1.968415823230534e-05, | |
| "loss": 0.4971, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.354916067146283, | |
| "grad_norm": 0.43748095631599426, | |
| "learning_rate": 1.96409044703021e-05, | |
| "loss": 0.4743, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.3645083932853717, | |
| "grad_norm": 0.46373656392097473, | |
| "learning_rate": 1.9594929736144978e-05, | |
| "loss": 0.5153, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.37410071942446044, | |
| "grad_norm": 0.42808473110198975, | |
| "learning_rate": 1.9546247005377065e-05, | |
| "loss": 0.4818, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.38369304556354916, | |
| "grad_norm": 0.4117816388607025, | |
| "learning_rate": 1.9494870017824877e-05, | |
| "loss": 0.4673, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.3932853717026379, | |
| "grad_norm": 0.43123477697372437, | |
| "learning_rate": 1.9440813273720504e-05, | |
| "loss": 0.4622, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.4028776978417266, | |
| "grad_norm": 0.36057785153388977, | |
| "learning_rate": 1.938409202960922e-05, | |
| "loss": 0.459, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.41247002398081534, | |
| "grad_norm": 0.3770412802696228, | |
| "learning_rate": 1.932472229404356e-05, | |
| "loss": 0.4919, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.42206235011990406, | |
| "grad_norm": 0.4013001620769501, | |
| "learning_rate": 1.9262720823065217e-05, | |
| "loss": 0.4957, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.4316546762589928, | |
| "grad_norm": 0.3589963912963867, | |
| "learning_rate": 1.9198105115475946e-05, | |
| "loss": 0.4775, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.4412470023980815, | |
| "grad_norm": 0.3630225658416748, | |
| "learning_rate": 1.9130893407898834e-05, | |
| "loss": 0.4828, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.45083932853717024, | |
| "grad_norm": 0.32825136184692383, | |
| "learning_rate": 1.9061104669631343e-05, | |
| "loss": 0.4686, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.460431654676259, | |
| "grad_norm": 0.34000617265701294, | |
| "learning_rate": 1.8988758597291577e-05, | |
| "loss": 0.4886, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.47002398081534774, | |
| "grad_norm": 0.3201892673969269, | |
| "learning_rate": 1.8913875609259246e-05, | |
| "loss": 0.4587, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.47961630695443647, | |
| "grad_norm": 0.3151942789554596, | |
| "learning_rate": 1.8836476839912967e-05, | |
| "loss": 0.4727, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.4892086330935252, | |
| "grad_norm": 0.3298013210296631, | |
| "learning_rate": 1.8756584133665447e-05, | |
| "loss": 0.4888, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.4988009592326139, | |
| "grad_norm": 0.3365343511104584, | |
| "learning_rate": 1.86742200387983e-05, | |
| "loss": 0.4842, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.5083932853717026, | |
| "grad_norm": 0.3066977560520172, | |
| "learning_rate": 1.8589407801098192e-05, | |
| "loss": 0.4366, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.5179856115107914, | |
| "grad_norm": 0.3741927444934845, | |
| "learning_rate": 1.8502171357296144e-05, | |
| "loss": 0.47, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.5275779376498801, | |
| "grad_norm": 0.32977259159088135, | |
| "learning_rate": 1.8412535328311813e-05, | |
| "loss": 0.4816, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.5371702637889688, | |
| "grad_norm": 0.313314288854599, | |
| "learning_rate": 1.8320525012304685e-05, | |
| "loss": 0.4529, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.5467625899280576, | |
| "grad_norm": 0.3619432747364044, | |
| "learning_rate": 1.8226166377534113e-05, | |
| "loss": 0.4583, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.5563549160671463, | |
| "grad_norm": 0.2958439886569977, | |
| "learning_rate": 1.8129486055030255e-05, | |
| "loss": 0.4861, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.565947242206235, | |
| "grad_norm": 0.3760446012020111, | |
| "learning_rate": 1.8030511331077945e-05, | |
| "loss": 0.4245, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.5755395683453237, | |
| "grad_norm": 0.3982815146446228, | |
| "learning_rate": 1.7929270139515606e-05, | |
| "loss": 0.458, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.5851318944844125, | |
| "grad_norm": 0.31815674901008606, | |
| "learning_rate": 1.782579105385145e-05, | |
| "loss": 0.4383, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.5947242206235012, | |
| "grad_norm": 0.3115115463733673, | |
| "learning_rate": 1.772010327919912e-05, | |
| "loss": 0.4809, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.60431654676259, | |
| "grad_norm": 0.3637670576572418, | |
| "learning_rate": 1.761223664403505e-05, | |
| "loss": 0.4685, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.6139088729016786, | |
| "grad_norm": 0.3954397141933441, | |
| "learning_rate": 1.7502221591779932e-05, | |
| "loss": 0.4591, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.6235011990407674, | |
| "grad_norm": 0.30222615599632263, | |
| "learning_rate": 1.7390089172206594e-05, | |
| "loss": 0.4747, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.6330935251798561, | |
| "grad_norm": 0.3727622926235199, | |
| "learning_rate": 1.727587103267677e-05, | |
| "loss": 0.4706, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.6426858513189448, | |
| "grad_norm": 0.35147586464881897, | |
| "learning_rate": 1.7159599409209194e-05, | |
| "loss": 0.5017, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.6522781774580336, | |
| "grad_norm": 0.3155067265033722, | |
| "learning_rate": 1.704130711738157e-05, | |
| "loss": 0.4193, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.6618705035971223, | |
| "grad_norm": 0.328908234834671, | |
| "learning_rate": 1.692102754306895e-05, | |
| "loss": 0.4606, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.6714628297362111, | |
| "grad_norm": 0.3115673363208771, | |
| "learning_rate": 1.6798794633021192e-05, | |
| "loss": 0.4637, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.6810551558752997, | |
| "grad_norm": 0.34590572118759155, | |
| "learning_rate": 1.667464288528207e-05, | |
| "loss": 0.4672, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.6906474820143885, | |
| "grad_norm": 0.32452234625816345, | |
| "learning_rate": 1.6548607339452853e-05, | |
| "loss": 0.4525, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.7002398081534772, | |
| "grad_norm": 0.35198819637298584, | |
| "learning_rate": 1.6420723566802982e-05, | |
| "loss": 0.4512, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.709832134292566, | |
| "grad_norm": 0.3182852268218994, | |
| "learning_rate": 1.6291027660230735e-05, | |
| "loss": 0.4645, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.7194244604316546, | |
| "grad_norm": 0.34308016300201416, | |
| "learning_rate": 1.6159556224076637e-05, | |
| "loss": 0.4418, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.7290167865707434, | |
| "grad_norm": 0.36044055223464966, | |
| "learning_rate": 1.6026346363792565e-05, | |
| "loss": 0.4761, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.7386091127098321, | |
| "grad_norm": 0.3420026898384094, | |
| "learning_rate": 1.5891435675469376e-05, | |
| "loss": 0.4759, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.7482014388489209, | |
| "grad_norm": 0.3534936010837555, | |
| "learning_rate": 1.57548622352261e-05, | |
| "loss": 0.4734, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.7577937649880095, | |
| "grad_norm": 0.3484748601913452, | |
| "learning_rate": 1.561666458846365e-05, | |
| "loss": 0.4592, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.7673860911270983, | |
| "grad_norm": 0.32589343190193176, | |
| "learning_rate": 1.5476881738986037e-05, | |
| "loss": 0.4715, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.7769784172661871, | |
| "grad_norm": 0.30610567331314087, | |
| "learning_rate": 1.5335553137992286e-05, | |
| "loss": 0.425, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.7865707434052758, | |
| "grad_norm": 0.3320309817790985, | |
| "learning_rate": 1.519271867294203e-05, | |
| "loss": 0.4441, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.7961630695443646, | |
| "grad_norm": 0.31670084595680237, | |
| "learning_rate": 1.504841865629799e-05, | |
| "loss": 0.4902, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.8057553956834532, | |
| "grad_norm": 0.3276488184928894, | |
| "learning_rate": 1.490269381414849e-05, | |
| "loss": 0.4918, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.815347721822542, | |
| "grad_norm": 0.35845503211021423, | |
| "learning_rate": 1.4755585274713289e-05, | |
| "loss": 0.4449, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.8249400479616307, | |
| "grad_norm": 0.30946382880210876, | |
| "learning_rate": 1.4607134556735836e-05, | |
| "loss": 0.4239, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.8345323741007195, | |
| "grad_norm": 0.35152146220207214, | |
| "learning_rate": 1.4457383557765385e-05, | |
| "loss": 0.4795, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.8441247002398081, | |
| "grad_norm": 0.3176645338535309, | |
| "learning_rate": 1.4306374542332141e-05, | |
| "loss": 0.459, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.8537170263788969, | |
| "grad_norm": 0.31331124901771545, | |
| "learning_rate": 1.4154150130018867e-05, | |
| "loss": 0.4471, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.8633093525179856, | |
| "grad_norm": 0.32980385422706604, | |
| "learning_rate": 1.4000753283432267e-05, | |
| "loss": 0.4765, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.8729016786570744, | |
| "grad_norm": 0.3267308175563812, | |
| "learning_rate": 1.3846227296077568e-05, | |
| "loss": 0.464, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.882494004796163, | |
| "grad_norm": 0.30493155121803284, | |
| "learning_rate": 1.3690615780139703e-05, | |
| "loss": 0.4371, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.8920863309352518, | |
| "grad_norm": 0.28876054286956787, | |
| "learning_rate": 1.3533962654174542e-05, | |
| "loss": 0.4609, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.9016786570743405, | |
| "grad_norm": 0.2889139652252197, | |
| "learning_rate": 1.337631213071369e-05, | |
| "loss": 0.4466, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.9112709832134293, | |
| "grad_norm": 0.3536216914653778, | |
| "learning_rate": 1.321770870378628e-05, | |
| "loss": 0.4531, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.920863309352518, | |
| "grad_norm": 0.3439409136772156, | |
| "learning_rate": 1.3058197136361344e-05, | |
| "loss": 0.4522, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.9304556354916067, | |
| "grad_norm": 0.3083202540874481, | |
| "learning_rate": 1.2897822447714247e-05, | |
| "loss": 0.4605, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.9400479616306955, | |
| "grad_norm": 0.3410809636116028, | |
| "learning_rate": 1.2736629900720832e-05, | |
| "loss": 0.4389, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.9496402877697842, | |
| "grad_norm": 0.28464362025260925, | |
| "learning_rate": 1.257466498908276e-05, | |
| "loss": 0.4509, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.9592326139088729, | |
| "grad_norm": 0.279031366109848, | |
| "learning_rate": 1.2411973424487751e-05, | |
| "loss": 0.4728, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9688249400479616, | |
| "grad_norm": 0.29481759667396545, | |
| "learning_rate": 1.2248601123708279e-05, | |
| "loss": 0.4602, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.9784172661870504, | |
| "grad_norm": 0.30976414680480957, | |
| "learning_rate": 1.2084594195642367e-05, | |
| "loss": 0.4558, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.988009592326139, | |
| "grad_norm": 0.31227511167526245, | |
| "learning_rate": 1.1919998928300203e-05, | |
| "loss": 0.4542, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.9976019184652278, | |
| "grad_norm": 0.30893808603286743, | |
| "learning_rate": 1.1754861775740163e-05, | |
| "loss": 0.4797, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.0071942446043165, | |
| "grad_norm": 0.6754066348075867, | |
| "learning_rate": 1.1589229344958e-05, | |
| "loss": 0.7123, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.0167865707434052, | |
| "grad_norm": 0.2993621528148651, | |
| "learning_rate": 1.1423148382732854e-05, | |
| "loss": 0.3439, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.026378896882494, | |
| "grad_norm": 0.37703534960746765, | |
| "learning_rate": 1.1256665762433798e-05, | |
| "loss": 0.3908, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.0359712230215827, | |
| "grad_norm": 0.3754074275493622, | |
| "learning_rate": 1.1089828470790694e-05, | |
| "loss": 0.3997, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.0455635491606714, | |
| "grad_norm": 0.3877393901348114, | |
| "learning_rate": 1.092268359463302e-05, | |
| "loss": 0.4586, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.0551558752997603, | |
| "grad_norm": 0.40832141041755676, | |
| "learning_rate": 1.0755278307600459e-05, | |
| "loss": 0.4046, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.064748201438849, | |
| "grad_norm": 0.3649183213710785, | |
| "learning_rate": 1.058765985682898e-05, | |
| "loss": 0.374, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.0743405275779376, | |
| "grad_norm": 0.3570377826690674, | |
| "learning_rate": 1.0419875549616196e-05, | |
| "loss": 0.3692, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.0839328537170263, | |
| "grad_norm": 0.4568242132663727, | |
| "learning_rate": 1.0251972740069724e-05, | |
| "loss": 0.4327, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.0935251798561152, | |
| "grad_norm": 0.3146553039550781, | |
| "learning_rate": 1.0083998815742335e-05, | |
| "loss": 0.3766, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.1031175059952039, | |
| "grad_norm": 0.29733189940452576, | |
| "learning_rate": 9.916001184257668e-06, | |
| "loss": 0.3503, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.1127098321342925, | |
| "grad_norm": 0.3751276433467865, | |
| "learning_rate": 9.748027259930276e-06, | |
| "loss": 0.4089, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.1223021582733812, | |
| "grad_norm": 0.38529419898986816, | |
| "learning_rate": 9.580124450383804e-06, | |
| "loss": 0.4489, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.13189448441247, | |
| "grad_norm": 0.32772964239120483, | |
| "learning_rate": 9.412340143171025e-06, | |
| "loss": 0.3736, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.1414868105515588, | |
| "grad_norm": 0.32223913073539734, | |
| "learning_rate": 9.244721692399545e-06, | |
| "loss": 0.3961, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.1510791366906474, | |
| "grad_norm": 0.3014141917228699, | |
| "learning_rate": 9.07731640536698e-06, | |
| "loss": 0.3863, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.160671462829736, | |
| "grad_norm": 0.3261420726776123, | |
| "learning_rate": 8.910171529209306e-06, | |
| "loss": 0.3797, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.170263788968825, | |
| "grad_norm": 0.30444827675819397, | |
| "learning_rate": 8.743334237566202e-06, | |
| "loss": 0.4264, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.1798561151079137, | |
| "grad_norm": 0.26329320669174194, | |
| "learning_rate": 8.576851617267151e-06, | |
| "loss": 0.388, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.1894484412470023, | |
| "grad_norm": 0.28534045815467834, | |
| "learning_rate": 8.410770655042003e-06, | |
| "loss": 0.3899, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.1990407673860912, | |
| "grad_norm": 0.30171987414360046, | |
| "learning_rate": 8.24513822425984e-06, | |
| "loss": 0.431, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.20863309352518, | |
| "grad_norm": 0.27923017740249634, | |
| "learning_rate": 8.0800010716998e-06, | |
| "loss": 0.383, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.2182254196642686, | |
| "grad_norm": 0.24267299473285675, | |
| "learning_rate": 7.915405804357632e-06, | |
| "loss": 0.3746, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.2278177458033572, | |
| "grad_norm": 0.2610694169998169, | |
| "learning_rate": 7.751398876291725e-06, | |
| "loss": 0.3792, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.2374100719424461, | |
| "grad_norm": 0.29287707805633545, | |
| "learning_rate": 7.58802657551225e-06, | |
| "loss": 0.4404, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.2470023980815348, | |
| "grad_norm": 0.33182862401008606, | |
| "learning_rate": 7.425335010917244e-06, | |
| "loss": 0.3757, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.2565947242206235, | |
| "grad_norm": 0.3336862325668335, | |
| "learning_rate": 7.263370099279173e-06, | |
| "loss": 0.474, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.2661870503597124, | |
| "grad_norm": 0.26169589161872864, | |
| "learning_rate": 7.102177552285753e-06, | |
| "loss": 0.3462, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.275779376498801, | |
| "grad_norm": 0.29609155654907227, | |
| "learning_rate": 6.9418028636386595e-06, | |
| "loss": 0.3856, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.2853717026378897, | |
| "grad_norm": 0.293792188167572, | |
| "learning_rate": 6.7822912962137225e-06, | |
| "loss": 0.3923, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.2949640287769784, | |
| "grad_norm": 0.2877916395664215, | |
| "learning_rate": 6.623687869286314e-06, | |
| "loss": 0.4054, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.304556354916067, | |
| "grad_norm": 0.26534056663513184, | |
| "learning_rate": 6.466037345825462e-06, | |
| "loss": 0.3854, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.314148681055156, | |
| "grad_norm": 0.2668323218822479, | |
| "learning_rate": 6.3093842198603014e-06, | |
| "loss": 0.3847, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.3237410071942446, | |
| "grad_norm": 0.2445860207080841, | |
| "learning_rate": 6.153772703922434e-06, | |
| "loss": 0.3721, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.27340319752693176, | |
| "learning_rate": 5.999246716567737e-06, | |
| "loss": 0.3854, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.3429256594724222, | |
| "grad_norm": 0.2549520432949066, | |
| "learning_rate": 5.845849869981137e-06, | |
| "loss": 0.3819, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.3525179856115108, | |
| "grad_norm": 0.21786105632781982, | |
| "learning_rate": 5.693625457667862e-06, | |
| "loss": 0.3562, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.3621103117505995, | |
| "grad_norm": 0.2585865259170532, | |
| "learning_rate": 5.542616442234618e-06, | |
| "loss": 0.4148, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.3717026378896882, | |
| "grad_norm": 0.25335797667503357, | |
| "learning_rate": 5.392865443264164e-06, | |
| "loss": 0.4044, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.381294964028777, | |
| "grad_norm": 0.23658272624015808, | |
| "learning_rate": 5.244414725286717e-06, | |
| "loss": 0.404, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.3908872901678657, | |
| "grad_norm": 0.2405097484588623, | |
| "learning_rate": 5.097306185851515e-06, | |
| "loss": 0.405, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.4004796163069544, | |
| "grad_norm": 0.22226805984973907, | |
| "learning_rate": 4.951581343702014e-06, | |
| "loss": 0.3787, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.4100719424460433, | |
| "grad_norm": 0.2628573179244995, | |
| "learning_rate": 4.807281327057972e-06, | |
| "loss": 0.3969, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.419664268585132, | |
| "grad_norm": 0.2247915416955948, | |
| "learning_rate": 4.664446862007718e-06, | |
| "loss": 0.3686, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.4292565947242206, | |
| "grad_norm": 0.2676323652267456, | |
| "learning_rate": 4.523118261013969e-06, | |
| "loss": 0.4435, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.4388489208633093, | |
| "grad_norm": 0.25832346081733704, | |
| "learning_rate": 4.383335411536357e-06, | |
| "loss": 0.3731, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.448441247002398, | |
| "grad_norm": 0.24996225535869598, | |
| "learning_rate": 4.245137764773899e-06, | |
| "loss": 0.3882, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.4580335731414868, | |
| "grad_norm": 0.21546132862567902, | |
| "learning_rate": 4.108564324530626e-06, | |
| "loss": 0.3697, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.4676258992805755, | |
| "grad_norm": 0.25346457958221436, | |
| "learning_rate": 3.973653636207437e-06, | |
| "loss": 0.4264, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.4772182254196642, | |
| "grad_norm": 0.22166943550109863, | |
| "learning_rate": 3.840443775923365e-06, | |
| "loss": 0.3594, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.486810551558753, | |
| "grad_norm": 0.22984443604946136, | |
| "learning_rate": 3.70897233976927e-06, | |
| "loss": 0.3914, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.4964028776978417, | |
| "grad_norm": 0.24873270094394684, | |
| "learning_rate": 3.5792764331970187e-06, | |
| "loss": 0.3922, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.5059952038369304, | |
| "grad_norm": 0.23538856208324432, | |
| "learning_rate": 3.4513926605471504e-06, | |
| "loss": 0.3665, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.515587529976019, | |
| "grad_norm": 0.2180040031671524, | |
| "learning_rate": 3.3253571147179333e-06, | |
| "loss": 0.3381, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.5251798561151078, | |
| "grad_norm": 0.24928021430969238, | |
| "learning_rate": 3.2012053669788136e-06, | |
| "loss": 0.4613, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.5347721822541966, | |
| "grad_norm": 0.2552678883075714, | |
| "learning_rate": 3.0789724569310532e-06, | |
| "loss": 0.4428, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.5443645083932853, | |
| "grad_norm": 0.19574448466300964, | |
| "learning_rate": 2.9586928826184323e-06, | |
| "loss": 0.3357, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.5539568345323742, | |
| "grad_norm": 0.22724831104278564, | |
| "learning_rate": 2.8404005907908083e-06, | |
| "loss": 0.4203, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.5635491606714629, | |
| "grad_norm": 0.21780751645565033, | |
| "learning_rate": 2.724128967323234e-06, | |
| "loss": 0.3881, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.5731414868105515, | |
| "grad_norm": 0.24764929711818695, | |
| "learning_rate": 2.6099108277934105e-06, | |
| "loss": 0.4334, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.5827338129496402, | |
| "grad_norm": 0.23750008642673492, | |
| "learning_rate": 2.4977784082200728e-06, | |
| "loss": 0.3885, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.5923261390887289, | |
| "grad_norm": 0.24234673380851746, | |
| "learning_rate": 2.3877633559649505e-06, | |
| "loss": 0.3901, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.6019184652278178, | |
| "grad_norm": 0.22565028071403503, | |
| "learning_rate": 2.2798967208008806e-06, | |
| "loss": 0.3769, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.6115107913669064, | |
| "grad_norm": 0.2320602983236313, | |
| "learning_rate": 2.1742089461485504e-06, | |
| "loss": 0.404, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.6211031175059953, | |
| "grad_norm": 0.21093355119228363, | |
| "learning_rate": 2.0707298604843964e-06, | |
| "loss": 0.3879, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.630695443645084, | |
| "grad_norm": 0.23324111104011536, | |
| "learning_rate": 1.9694886689220592e-06, | |
| "loss": 0.3728, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.6402877697841727, | |
| "grad_norm": 0.23098240792751312, | |
| "learning_rate": 1.870513944969743e-06, | |
| "loss": 0.4245, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.6498800959232613, | |
| "grad_norm": 0.21313826739788055, | |
| "learning_rate": 1.773833622465888e-06, | |
| "loss": 0.3685, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.65947242206235, | |
| "grad_norm": 0.2202357053756714, | |
| "learning_rate": 1.6794749876953187e-06, | |
| "loss": 0.385, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.6690647482014387, | |
| "grad_norm": 0.24201858043670654, | |
| "learning_rate": 1.587464671688187e-06, | |
| "loss": 0.4342, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.6786570743405276, | |
| "grad_norm": 0.22546477615833282, | |
| "learning_rate": 1.4978286427038602e-06, | |
| "loss": 0.3845, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.6882494004796165, | |
| "grad_norm": 0.2069396823644638, | |
| "learning_rate": 1.4105921989018112e-06, | |
| "loss": 0.3671, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.6978417266187051, | |
| "grad_norm": 0.22818851470947266, | |
| "learning_rate": 1.325779961201703e-06, | |
| "loss": 0.4138, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.7074340527577938, | |
| "grad_norm": 0.1912079006433487, | |
| "learning_rate": 1.2434158663345553e-06, | |
| "loss": 0.3297, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.7170263788968825, | |
| "grad_norm": 0.2530308961868286, | |
| "learning_rate": 1.1635231600870334e-06, | |
| "loss": 0.4104, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.7266187050359711, | |
| "grad_norm": 0.21219122409820557, | |
| "learning_rate": 1.086124390740757e-06, | |
| "loss": 0.3553, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.7362110311750598, | |
| "grad_norm": 0.23752814531326294, | |
| "learning_rate": 1.0112414027084262e-06, | |
| "loss": 0.3862, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.7458033573141487, | |
| "grad_norm": 0.2403864711523056, | |
| "learning_rate": 9.388953303686587e-07, | |
| "loss": 0.4316, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.7553956834532374, | |
| "grad_norm": 0.20274099707603455, | |
| "learning_rate": 8.691065921011687e-07, | |
| "loss": 0.3422, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.7649880095923263, | |
| "grad_norm": 0.21107341349124908, | |
| "learning_rate": 8.018948845240538e-07, | |
| "loss": 0.3584, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.774580335731415, | |
| "grad_norm": 0.21734651923179626, | |
| "learning_rate": 7.372791769347843e-07, | |
| "loss": 0.4099, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.7841726618705036, | |
| "grad_norm": 0.2155366688966751, | |
| "learning_rate": 6.752777059564431e-07, | |
| "loss": 0.3968, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.7937649880095923, | |
| "grad_norm": 0.20627352595329285, | |
| "learning_rate": 6.159079703907823e-07, | |
| "loss": 0.34, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.803357314148681, | |
| "grad_norm": 0.2195385843515396, | |
| "learning_rate": 5.591867262794969e-07, | |
| "loss": 0.3729, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.8129496402877698, | |
| "grad_norm": 0.23355402052402496, | |
| "learning_rate": 5.051299821751254e-07, | |
| "loss": 0.3928, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.8225419664268585, | |
| "grad_norm": 0.22403055429458618, | |
| "learning_rate": 4.537529946229369e-07, | |
| "loss": 0.3818, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.8321342925659474, | |
| "grad_norm": 0.21041440963745117, | |
| "learning_rate": 4.0507026385502747e-07, | |
| "loss": 0.3448, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.841726618705036, | |
| "grad_norm": 0.22846421599388123, | |
| "learning_rate": 3.5909552969790376e-07, | |
| "loss": 0.4311, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.8513189448441247, | |
| "grad_norm": 0.22553186118602753, | |
| "learning_rate": 3.158417676946635e-07, | |
| "loss": 0.3701, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.8609112709832134, | |
| "grad_norm": 0.20845209062099457, | |
| "learning_rate": 2.753211854428728e-07, | |
| "loss": 0.3833, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.870503597122302, | |
| "grad_norm": 0.23942551016807556, | |
| "learning_rate": 2.375452191491967e-07, | |
| "loss": 0.4594, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.8800959232613907, | |
| "grad_norm": 0.21887589991092682, | |
| "learning_rate": 2.0252453040173646e-07, | |
| "loss": 0.3805, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.8896882494004796, | |
| "grad_norm": 0.21494026482105255, | |
| "learning_rate": 1.7026900316098217e-07, | |
| "loss": 0.3504, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.8992805755395683, | |
| "grad_norm": 0.21897700428962708, | |
| "learning_rate": 1.407877409702496e-07, | |
| "loss": 0.3396, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.9088729016786572, | |
| "grad_norm": 0.2428821474313736, | |
| "learning_rate": 1.1408906438636236e-07, | |
| "loss": 0.4681, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.9184652278177459, | |
| "grad_norm": 0.20344102382659912, | |
| "learning_rate": 9.018050863132566e-08, | |
| "loss": 0.3943, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.9280575539568345, | |
| "grad_norm": 0.19937625527381897, | |
| "learning_rate": 6.906882146565097e-08, | |
| "loss": 0.3736, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 1.9376498800959232, | |
| "grad_norm": 0.3866899311542511, | |
| "learning_rate": 5.0759961283911584e-08, | |
| "loss": 0.4146, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 1.9472422062350119, | |
| "grad_norm": 0.1998918056488037, | |
| "learning_rate": 3.525909543310002e-08, | |
| "loss": 0.3537, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 1.9568345323741008, | |
| "grad_norm": 0.19235286116600037, | |
| "learning_rate": 2.257059875423795e-08, | |
| "loss": 0.3812, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 1.9664268585131894, | |
| "grad_norm": 0.21340519189834595, | |
| "learning_rate": 1.2698052347649426e-08, | |
| "loss": 0.3976, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.9760191846522783, | |
| "grad_norm": 0.21041817963123322, | |
| "learning_rate": 5.644242562264923e-09, | |
| "loss": 0.3976, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 1.985611510791367, | |
| "grad_norm": 0.22794745862483978, | |
| "learning_rate": 1.4111602092226062e-09, | |
| "loss": 0.3929, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 1.9952038369304557, | |
| "grad_norm": 0.22788451611995697, | |
| "learning_rate": 0.0, | |
| "loss": 0.4285, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 1.9952038369304557, | |
| "step": 208, | |
| "total_flos": 716120051679232.0, | |
| "train_loss": 0.4522540926073606, | |
| "train_runtime": 71993.7308, | |
| "train_samples_per_second": 0.278, | |
| "train_steps_per_second": 0.003 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 208, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 716120051679232.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |