{ "best_metric": null, "best_model_checkpoint": null, "epoch": 177.5147928994083, "eval_steps": 2000, "global_step": 30000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.05917159763313609, "grad_norm": 11.824136734008789, "learning_rate": 5.999999999999999e-06, "loss": 0.6299, "step": 10 }, { "epoch": 0.11834319526627218, "grad_norm": 4.339348316192627, "learning_rate": 1.1999999999999999e-05, "loss": 0.3218, "step": 20 }, { "epoch": 0.17751479289940827, "grad_norm": 2.588338851928711, "learning_rate": 1.7999999999999997e-05, "loss": 0.2173, "step": 30 }, { "epoch": 0.23668639053254437, "grad_norm": 1.5682891607284546, "learning_rate": 2.3999999999999997e-05, "loss": 0.1711, "step": 40 }, { "epoch": 0.2958579881656805, "grad_norm": 1.4503942728042603, "learning_rate": 2.9999999999999997e-05, "loss": 0.1514, "step": 50 }, { "epoch": 0.35502958579881655, "grad_norm": 1.2580463886260986, "learning_rate": 3.5999999999999994e-05, "loss": 0.1433, "step": 60 }, { "epoch": 0.41420118343195267, "grad_norm": 1.34689199924469, "learning_rate": 4.2e-05, "loss": 0.1361, "step": 70 }, { "epoch": 0.47337278106508873, "grad_norm": 1.379999041557312, "learning_rate": 4.7999999999999994e-05, "loss": 0.1314, "step": 80 }, { "epoch": 0.5325443786982249, "grad_norm": 0.929875373840332, "learning_rate": 5.399999999999999e-05, "loss": 0.1325, "step": 90 }, { "epoch": 0.591715976331361, "grad_norm": 1.1812986135482788, "learning_rate": 5.9999999999999995e-05, "loss": 0.1191, "step": 100 }, { "epoch": 0.650887573964497, "grad_norm": 0.5973801612854004, "learning_rate": 6.599999999999999e-05, "loss": 0.1168, "step": 110 }, { "epoch": 0.7100591715976331, "grad_norm": 0.9734118580818176, "learning_rate": 7.199999999999999e-05, "loss": 0.127, "step": 120 }, { "epoch": 0.7692307692307693, "grad_norm": 1.5700788497924805, "learning_rate": 7.8e-05, "loss": 0.1152, "step": 130 }, { "epoch": 0.8284023668639053, "grad_norm": 1.9392459392547607, "learning_rate": 8.4e-05, "loss": 0.1188, "step": 140 }, { "epoch": 0.8875739644970414, "grad_norm": 0.7793346047401428, "learning_rate": 8.999999999999999e-05, "loss": 0.1133, "step": 150 }, { "epoch": 0.9467455621301775, "grad_norm": 0.6193998456001282, "learning_rate": 9.599999999999999e-05, "loss": 0.1109, "step": 160 }, { "epoch": 1.0059171597633136, "grad_norm": 1.673331379890442, "learning_rate": 0.000102, "loss": 0.1072, "step": 170 }, { "epoch": 1.0650887573964498, "grad_norm": 1.3658688068389893, "learning_rate": 0.00010799999999999998, "loss": 0.1086, "step": 180 }, { "epoch": 1.1242603550295858, "grad_norm": 0.7868382930755615, "learning_rate": 0.00011399999999999999, "loss": 0.0976, "step": 190 }, { "epoch": 1.183431952662722, "grad_norm": 0.8233304619789124, "learning_rate": 0.00011999999999999999, "loss": 0.1005, "step": 200 }, { "epoch": 1.242603550295858, "grad_norm": 0.8617013692855835, "learning_rate": 0.00012599999999999997, "loss": 0.098, "step": 210 }, { "epoch": 1.301775147928994, "grad_norm": 1.1947269439697266, "learning_rate": 0.00013199999999999998, "loss": 0.1011, "step": 220 }, { "epoch": 1.3609467455621302, "grad_norm": 0.5995688438415527, "learning_rate": 0.000138, "loss": 0.0972, "step": 230 }, { "epoch": 1.4201183431952662, "grad_norm": 0.8240473866462708, "learning_rate": 0.00014399999999999998, "loss": 0.0944, "step": 240 }, { "epoch": 1.4792899408284024, "grad_norm": 0.8028063774108887, "learning_rate": 0.00015, "loss": 0.0884, "step": 250 }, { "epoch": 1.5384615384615383, "grad_norm": 0.6609148979187012, "learning_rate": 0.000156, "loss": 0.0799, "step": 260 }, { "epoch": 1.5976331360946747, "grad_norm": 0.5619239807128906, "learning_rate": 0.000162, "loss": 0.0766, "step": 270 }, { "epoch": 1.6568047337278107, "grad_norm": 0.8065875768661499, "learning_rate": 0.000168, "loss": 0.0762, "step": 280 }, { "epoch": 1.7159763313609466, "grad_norm": 0.9938980937004089, "learning_rate": 0.00017399999999999997, "loss": 0.0768, "step": 290 }, { "epoch": 1.7751479289940828, "grad_norm": 1.3605124950408936, "learning_rate": 0.00017999999999999998, "loss": 0.0747, "step": 300 }, { "epoch": 1.834319526627219, "grad_norm": 1.40558922290802, "learning_rate": 0.000186, "loss": 0.0751, "step": 310 }, { "epoch": 1.893491124260355, "grad_norm": 0.9570429921150208, "learning_rate": 0.00019199999999999998, "loss": 0.0738, "step": 320 }, { "epoch": 1.952662721893491, "grad_norm": 0.9977752566337585, "learning_rate": 0.000198, "loss": 0.0725, "step": 330 }, { "epoch": 2.0118343195266273, "grad_norm": 0.7113844156265259, "learning_rate": 0.000204, "loss": 0.0613, "step": 340 }, { "epoch": 2.0710059171597632, "grad_norm": 0.8652175664901733, "learning_rate": 0.00020999999999999998, "loss": 0.068, "step": 350 }, { "epoch": 2.1301775147928996, "grad_norm": 1.0036293268203735, "learning_rate": 0.00021599999999999996, "loss": 0.0648, "step": 360 }, { "epoch": 2.1893491124260356, "grad_norm": 1.239670991897583, "learning_rate": 0.00022199999999999998, "loss": 0.0647, "step": 370 }, { "epoch": 2.2485207100591715, "grad_norm": 0.6006696224212646, "learning_rate": 0.00022799999999999999, "loss": 0.0636, "step": 380 }, { "epoch": 2.3076923076923075, "grad_norm": 0.9947748780250549, "learning_rate": 0.000234, "loss": 0.0605, "step": 390 }, { "epoch": 2.366863905325444, "grad_norm": 0.6557269096374512, "learning_rate": 0.00023999999999999998, "loss": 0.0661, "step": 400 }, { "epoch": 2.42603550295858, "grad_norm": 0.8801721334457397, "learning_rate": 0.00024599999999999996, "loss": 0.0586, "step": 410 }, { "epoch": 2.485207100591716, "grad_norm": 0.6034063696861267, "learning_rate": 0.00025199999999999995, "loss": 0.0607, "step": 420 }, { "epoch": 2.544378698224852, "grad_norm": 0.6921690106391907, "learning_rate": 0.000258, "loss": 0.0656, "step": 430 }, { "epoch": 2.603550295857988, "grad_norm": 0.8505598902702332, "learning_rate": 0.00026399999999999997, "loss": 0.0577, "step": 440 }, { "epoch": 2.662721893491124, "grad_norm": 0.9163134098052979, "learning_rate": 0.00027, "loss": 0.0592, "step": 450 }, { "epoch": 2.7218934911242605, "grad_norm": 0.8192881941795349, "learning_rate": 0.000276, "loss": 0.0589, "step": 460 }, { "epoch": 2.7810650887573964, "grad_norm": 0.6882975101470947, "learning_rate": 0.00028199999999999997, "loss": 0.0602, "step": 470 }, { "epoch": 2.8402366863905324, "grad_norm": 0.7579637169837952, "learning_rate": 0.00028799999999999995, "loss": 0.0538, "step": 480 }, { "epoch": 2.899408284023669, "grad_norm": 0.9263840913772583, "learning_rate": 0.000294, "loss": 0.0564, "step": 490 }, { "epoch": 2.9585798816568047, "grad_norm": 0.5948222875595093, "learning_rate": 0.0003, "loss": 0.0538, "step": 500 }, { "epoch": 3.0177514792899407, "grad_norm": 0.36613577604293823, "learning_rate": 0.0002999999149416535, "loss": 0.0528, "step": 510 }, { "epoch": 3.076923076923077, "grad_norm": 0.7347477078437805, "learning_rate": 0.00029999965976671057, "loss": 0.0548, "step": 520 }, { "epoch": 3.136094674556213, "grad_norm": 0.6503780484199524, "learning_rate": 0.00029999923447546054, "loss": 0.0496, "step": 530 }, { "epoch": 3.195266272189349, "grad_norm": 0.579636812210083, "learning_rate": 0.0002999986390683858, "loss": 0.05, "step": 540 }, { "epoch": 3.2544378698224854, "grad_norm": 0.44370511174201965, "learning_rate": 0.0002999978735461616, "loss": 0.0503, "step": 550 }, { "epoch": 3.3136094674556213, "grad_norm": 1.0849943161010742, "learning_rate": 0.00029999693790965605, "loss": 0.056, "step": 560 }, { "epoch": 3.3727810650887573, "grad_norm": 0.9913458228111267, "learning_rate": 0.0002999958321599304, "loss": 0.0586, "step": 570 }, { "epoch": 3.4319526627218933, "grad_norm": 1.0716348886489868, "learning_rate": 0.00029999455629823853, "loss": 0.0533, "step": 580 }, { "epoch": 3.4911242603550297, "grad_norm": 1.0536736249923706, "learning_rate": 0.0002999931103260275, "loss": 0.0511, "step": 590 }, { "epoch": 3.5502958579881656, "grad_norm": 0.6463172435760498, "learning_rate": 0.0002999914942449372, "loss": 0.0514, "step": 600 }, { "epoch": 3.609467455621302, "grad_norm": 0.6597800850868225, "learning_rate": 0.00029998970805680044, "loss": 0.0476, "step": 610 }, { "epoch": 3.668639053254438, "grad_norm": 0.571786642074585, "learning_rate": 0.00029998775176364297, "loss": 0.0476, "step": 620 }, { "epoch": 3.727810650887574, "grad_norm": 0.7449012398719788, "learning_rate": 0.00029998562536768343, "loss": 0.0411, "step": 630 }, { "epoch": 3.78698224852071, "grad_norm": 0.5635416507720947, "learning_rate": 0.00029998332887133335, "loss": 0.0484, "step": 640 }, { "epoch": 3.8461538461538463, "grad_norm": 0.49004629254341125, "learning_rate": 0.0002999808622771973, "loss": 0.0443, "step": 650 }, { "epoch": 3.905325443786982, "grad_norm": 0.8492295145988464, "learning_rate": 0.0002999782255880726, "loss": 0.0453, "step": 660 }, { "epoch": 3.9644970414201186, "grad_norm": 1.3753185272216797, "learning_rate": 0.0002999754188069496, "loss": 0.0486, "step": 670 }, { "epoch": 4.023668639053255, "grad_norm": 0.8239424824714661, "learning_rate": 0.0002999724419370114, "loss": 0.0479, "step": 680 }, { "epoch": 4.0828402366863905, "grad_norm": 0.5803558230400085, "learning_rate": 0.0002999692949816343, "loss": 0.0463, "step": 690 }, { "epoch": 4.1420118343195265, "grad_norm": 0.8419933915138245, "learning_rate": 0.00029996597794438706, "loss": 0.0436, "step": 700 }, { "epoch": 4.201183431952662, "grad_norm": 0.8281695246696472, "learning_rate": 0.00029996249082903174, "loss": 0.0428, "step": 710 }, { "epoch": 4.260355029585799, "grad_norm": 0.5627180337905884, "learning_rate": 0.0002999588336395231, "loss": 0.0436, "step": 720 }, { "epoch": 4.319526627218935, "grad_norm": 0.3742409348487854, "learning_rate": 0.0002999550063800087, "loss": 0.0339, "step": 730 }, { "epoch": 4.378698224852071, "grad_norm": 0.44432634115219116, "learning_rate": 0.0002999510090548291, "loss": 0.0393, "step": 740 }, { "epoch": 4.437869822485207, "grad_norm": 0.5128399133682251, "learning_rate": 0.00029994684166851786, "loss": 0.0494, "step": 750 }, { "epoch": 4.497041420118343, "grad_norm": 0.4789179861545563, "learning_rate": 0.0002999425042258011, "loss": 0.0443, "step": 760 }, { "epoch": 4.556213017751479, "grad_norm": 0.46073096990585327, "learning_rate": 0.00029993799673159804, "loss": 0.0376, "step": 770 }, { "epoch": 4.615384615384615, "grad_norm": 0.5272684693336487, "learning_rate": 0.0002999333191910206, "loss": 0.0382, "step": 780 }, { "epoch": 4.674556213017752, "grad_norm": 0.4965602159500122, "learning_rate": 0.0002999284716093737, "loss": 0.0393, "step": 790 }, { "epoch": 4.733727810650888, "grad_norm": 0.3911011219024658, "learning_rate": 0.000299923453992155, "loss": 0.0345, "step": 800 }, { "epoch": 4.792899408284024, "grad_norm": 0.41893330216407776, "learning_rate": 0.0002999182663450551, "loss": 0.0356, "step": 810 }, { "epoch": 4.85207100591716, "grad_norm": 0.4948401153087616, "learning_rate": 0.00029991290867395736, "loss": 0.0402, "step": 820 }, { "epoch": 4.911242603550296, "grad_norm": 0.577849268913269, "learning_rate": 0.0002999073809849379, "loss": 0.0332, "step": 830 }, { "epoch": 4.970414201183432, "grad_norm": 0.5438259243965149, "learning_rate": 0.00029990168328426574, "loss": 0.0344, "step": 840 }, { "epoch": 5.029585798816568, "grad_norm": 0.4858870804309845, "learning_rate": 0.00029989581557840283, "loss": 0.0338, "step": 850 }, { "epoch": 5.088757396449704, "grad_norm": 0.3909776210784912, "learning_rate": 0.00029988977787400367, "loss": 0.0345, "step": 860 }, { "epoch": 5.14792899408284, "grad_norm": 0.6723620891571045, "learning_rate": 0.0002998835701779157, "loss": 0.0372, "step": 870 }, { "epoch": 5.207100591715976, "grad_norm": 0.7856037616729736, "learning_rate": 0.0002998771924971792, "loss": 0.0371, "step": 880 }, { "epoch": 5.266272189349112, "grad_norm": 0.9887698888778687, "learning_rate": 0.00029987064483902716, "loss": 0.0408, "step": 890 }, { "epoch": 5.325443786982248, "grad_norm": 0.4961603283882141, "learning_rate": 0.00029986392721088537, "loss": 0.035, "step": 900 }, { "epoch": 5.384615384615385, "grad_norm": 0.6605113744735718, "learning_rate": 0.00029985703962037225, "loss": 0.0358, "step": 910 }, { "epoch": 5.443786982248521, "grad_norm": 0.6382588744163513, "learning_rate": 0.0002998499820752992, "loss": 0.0395, "step": 920 }, { "epoch": 5.502958579881657, "grad_norm": 0.545966386795044, "learning_rate": 0.00029984275458367027, "loss": 0.0317, "step": 930 }, { "epoch": 5.562130177514793, "grad_norm": 0.3946206271648407, "learning_rate": 0.0002998353571536822, "loss": 0.0383, "step": 940 }, { "epoch": 5.621301775147929, "grad_norm": 0.41466793417930603, "learning_rate": 0.00029982778979372447, "loss": 0.0354, "step": 950 }, { "epoch": 5.680473372781065, "grad_norm": 0.803202748298645, "learning_rate": 0.0002998200525123794, "loss": 0.0352, "step": 960 }, { "epoch": 5.739644970414201, "grad_norm": 0.5700823068618774, "learning_rate": 0.00029981214531842184, "loss": 0.0348, "step": 970 }, { "epoch": 5.798816568047338, "grad_norm": 0.597564697265625, "learning_rate": 0.0002998040682208195, "loss": 0.0359, "step": 980 }, { "epoch": 5.8579881656804735, "grad_norm": 0.687793493270874, "learning_rate": 0.00029979582122873264, "loss": 0.0376, "step": 990 }, { "epoch": 5.9171597633136095, "grad_norm": 0.37600284814834595, "learning_rate": 0.0002997874043515143, "loss": 0.0345, "step": 1000 }, { "epoch": 5.976331360946745, "grad_norm": 0.5915971398353577, "learning_rate": 0.00029977881759871014, "loss": 0.035, "step": 1010 }, { "epoch": 6.035502958579881, "grad_norm": 0.6107649803161621, "learning_rate": 0.00029977006098005844, "loss": 0.0323, "step": 1020 }, { "epoch": 6.094674556213017, "grad_norm": 0.7518269419670105, "learning_rate": 0.00029976113450549036, "loss": 0.0336, "step": 1030 }, { "epoch": 6.153846153846154, "grad_norm": 0.48567840456962585, "learning_rate": 0.00029975203818512935, "loss": 0.0311, "step": 1040 }, { "epoch": 6.21301775147929, "grad_norm": 0.5740759968757629, "learning_rate": 0.0002997427720292917, "loss": 0.0343, "step": 1050 }, { "epoch": 6.272189349112426, "grad_norm": 0.4097563922405243, "learning_rate": 0.00029973333604848624, "loss": 0.0309, "step": 1060 }, { "epoch": 6.331360946745562, "grad_norm": 0.37523114681243896, "learning_rate": 0.0002997237302534145, "loss": 0.028, "step": 1070 }, { "epoch": 6.390532544378698, "grad_norm": 0.33917033672332764, "learning_rate": 0.0002997139546549704, "loss": 0.0303, "step": 1080 }, { "epoch": 6.449704142011834, "grad_norm": 0.40523475408554077, "learning_rate": 0.0002997040092642407, "loss": 0.0312, "step": 1090 }, { "epoch": 6.508875739644971, "grad_norm": 0.3181855380535126, "learning_rate": 0.0002996938940925045, "loss": 0.0309, "step": 1100 }, { "epoch": 6.568047337278107, "grad_norm": 0.4907320737838745, "learning_rate": 0.0002996836091512335, "loss": 0.0301, "step": 1110 }, { "epoch": 6.627218934911243, "grad_norm": 0.4967445433139801, "learning_rate": 0.00029967315445209193, "loss": 0.0274, "step": 1120 }, { "epoch": 6.686390532544379, "grad_norm": 0.42770835757255554, "learning_rate": 0.00029966253000693674, "loss": 0.0286, "step": 1130 }, { "epoch": 6.745562130177515, "grad_norm": 0.682017982006073, "learning_rate": 0.0002996517358278171, "loss": 0.0334, "step": 1140 }, { "epoch": 6.804733727810651, "grad_norm": 0.4393368065357208, "learning_rate": 0.00029964077192697487, "loss": 0.0304, "step": 1150 }, { "epoch": 6.8639053254437865, "grad_norm": 0.4872368276119232, "learning_rate": 0.0002996296383168443, "loss": 0.0307, "step": 1160 }, { "epoch": 6.923076923076923, "grad_norm": 0.39712002873420715, "learning_rate": 0.00029961833501005223, "loss": 0.0294, "step": 1170 }, { "epoch": 6.982248520710059, "grad_norm": 0.6960808634757996, "learning_rate": 0.0002996068620194177, "loss": 0.036, "step": 1180 }, { "epoch": 7.041420118343195, "grad_norm": 0.6480140686035156, "learning_rate": 0.00029959521935795253, "loss": 0.0323, "step": 1190 }, { "epoch": 7.100591715976331, "grad_norm": 0.7224964499473572, "learning_rate": 0.0002995834070388607, "loss": 0.0325, "step": 1200 }, { "epoch": 7.159763313609467, "grad_norm": 0.3601278066635132, "learning_rate": 0.00029957142507553873, "loss": 0.0291, "step": 1210 }, { "epoch": 7.218934911242604, "grad_norm": 0.3797994554042816, "learning_rate": 0.0002995592734815755, "loss": 0.0293, "step": 1220 }, { "epoch": 7.27810650887574, "grad_norm": 0.44617795944213867, "learning_rate": 0.00029954695227075225, "loss": 0.0276, "step": 1230 }, { "epoch": 7.337278106508876, "grad_norm": 0.4666217267513275, "learning_rate": 0.00029953446145704256, "loss": 0.025, "step": 1240 }, { "epoch": 7.396449704142012, "grad_norm": 0.3226659297943115, "learning_rate": 0.0002995218010546125, "loss": 0.0221, "step": 1250 }, { "epoch": 7.455621301775148, "grad_norm": 0.29016244411468506, "learning_rate": 0.0002995089710778203, "loss": 0.0246, "step": 1260 }, { "epoch": 7.514792899408284, "grad_norm": 0.3193188011646271, "learning_rate": 0.00029949597154121666, "loss": 0.025, "step": 1270 }, { "epoch": 7.57396449704142, "grad_norm": 0.4275120198726654, "learning_rate": 0.0002994828024595444, "loss": 0.0254, "step": 1280 }, { "epoch": 7.633136094674557, "grad_norm": 0.5296067595481873, "learning_rate": 0.00029946946384773877, "loss": 0.0261, "step": 1290 }, { "epoch": 7.6923076923076925, "grad_norm": 0.5340242981910706, "learning_rate": 0.00029945595572092727, "loss": 0.0275, "step": 1300 }, { "epoch": 7.7514792899408285, "grad_norm": 0.375640332698822, "learning_rate": 0.0002994422780944296, "loss": 0.025, "step": 1310 }, { "epoch": 7.810650887573964, "grad_norm": 0.5609472393989563, "learning_rate": 0.00029942843098375765, "loss": 0.0255, "step": 1320 }, { "epoch": 7.8698224852071, "grad_norm": 0.33895713090896606, "learning_rate": 0.0002994144144046157, "loss": 0.0251, "step": 1330 }, { "epoch": 7.928994082840236, "grad_norm": 0.31891149282455444, "learning_rate": 0.00029940022837290003, "loss": 0.0249, "step": 1340 }, { "epoch": 7.988165680473373, "grad_norm": 0.4565357267856598, "learning_rate": 0.0002993858729046992, "loss": 0.0234, "step": 1350 }, { "epoch": 8.04733727810651, "grad_norm": 0.3872339427471161, "learning_rate": 0.0002993713480162939, "loss": 0.0233, "step": 1360 }, { "epoch": 8.106508875739644, "grad_norm": 0.25777459144592285, "learning_rate": 0.000299356653724157, "loss": 0.0231, "step": 1370 }, { "epoch": 8.165680473372781, "grad_norm": 0.43085747957229614, "learning_rate": 0.00029934179004495345, "loss": 0.0214, "step": 1380 }, { "epoch": 8.224852071005918, "grad_norm": 0.46794596314430237, "learning_rate": 0.0002993267569955403, "loss": 0.0227, "step": 1390 }, { "epoch": 8.284023668639053, "grad_norm": 0.38305190205574036, "learning_rate": 0.0002993115545929667, "loss": 0.0218, "step": 1400 }, { "epoch": 8.34319526627219, "grad_norm": 0.600847065448761, "learning_rate": 0.0002992961828544738, "loss": 0.0248, "step": 1410 }, { "epoch": 8.402366863905325, "grad_norm": 0.6282093524932861, "learning_rate": 0.000299280641797495, "loss": 0.0249, "step": 1420 }, { "epoch": 8.461538461538462, "grad_norm": 0.29690682888031006, "learning_rate": 0.00029926493143965553, "loss": 0.022, "step": 1430 }, { "epoch": 8.520710059171599, "grad_norm": 0.3577359914779663, "learning_rate": 0.00029924905179877265, "loss": 0.0216, "step": 1440 }, { "epoch": 8.579881656804734, "grad_norm": 0.44497552514076233, "learning_rate": 0.0002992330028928557, "loss": 0.0229, "step": 1450 }, { "epoch": 8.63905325443787, "grad_norm": 0.21880848705768585, "learning_rate": 0.00029921678474010584, "loss": 0.021, "step": 1460 }, { "epoch": 8.698224852071005, "grad_norm": 0.35989904403686523, "learning_rate": 0.00029920039735891634, "loss": 0.0228, "step": 1470 }, { "epoch": 8.757396449704142, "grad_norm": 0.32641860842704773, "learning_rate": 0.00029918384076787233, "loss": 0.0212, "step": 1480 }, { "epoch": 8.816568047337277, "grad_norm": 0.3977866470813751, "learning_rate": 0.00029916711498575077, "loss": 0.0228, "step": 1490 }, { "epoch": 8.875739644970414, "grad_norm": 0.3235122859477997, "learning_rate": 0.00029915022003152054, "loss": 0.0214, "step": 1500 }, { "epoch": 8.934911242603551, "grad_norm": 0.6049141883850098, "learning_rate": 0.0002991331559243425, "loss": 0.0248, "step": 1510 }, { "epoch": 8.994082840236686, "grad_norm": 0.5603741407394409, "learning_rate": 0.00029911592268356906, "loss": 0.0229, "step": 1520 }, { "epoch": 9.053254437869823, "grad_norm": 0.5064825415611267, "learning_rate": 0.0002990985203287448, "loss": 0.0209, "step": 1530 }, { "epoch": 9.112426035502958, "grad_norm": 0.5022265315055847, "learning_rate": 0.0002990809488796059, "loss": 0.0241, "step": 1540 }, { "epoch": 9.171597633136095, "grad_norm": 0.45856526494026184, "learning_rate": 0.0002990632083560803, "loss": 0.0226, "step": 1550 }, { "epoch": 9.23076923076923, "grad_norm": 0.49211156368255615, "learning_rate": 0.00029904529877828776, "loss": 0.0267, "step": 1560 }, { "epoch": 9.289940828402367, "grad_norm": 0.44354385137557983, "learning_rate": 0.00029902722016653967, "loss": 0.0216, "step": 1570 }, { "epoch": 9.349112426035504, "grad_norm": 0.3159914016723633, "learning_rate": 0.00029900897254133926, "loss": 0.0202, "step": 1580 }, { "epoch": 9.408284023668639, "grad_norm": 0.27622562646865845, "learning_rate": 0.0002989905559233814, "loss": 0.0202, "step": 1590 }, { "epoch": 9.467455621301776, "grad_norm": 0.22425854206085205, "learning_rate": 0.0002989719703335525, "loss": 0.0187, "step": 1600 }, { "epoch": 9.52662721893491, "grad_norm": 0.33633822202682495, "learning_rate": 0.00029895321579293064, "loss": 0.0187, "step": 1610 }, { "epoch": 9.585798816568047, "grad_norm": 0.24696113169193268, "learning_rate": 0.0002989342923227857, "loss": 0.0199, "step": 1620 }, { "epoch": 9.644970414201183, "grad_norm": 0.2491302192211151, "learning_rate": 0.00029891519994457887, "loss": 0.0197, "step": 1630 }, { "epoch": 9.70414201183432, "grad_norm": 0.20413026213645935, "learning_rate": 0.00029889593867996316, "loss": 0.0203, "step": 1640 }, { "epoch": 9.763313609467456, "grad_norm": 0.293801486492157, "learning_rate": 0.00029887650855078287, "loss": 0.02, "step": 1650 }, { "epoch": 9.822485207100591, "grad_norm": 0.311535507440567, "learning_rate": 0.0002988569095790741, "loss": 0.02, "step": 1660 }, { "epoch": 9.881656804733728, "grad_norm": 0.3253050148487091, "learning_rate": 0.000298837141787064, "loss": 0.019, "step": 1670 }, { "epoch": 9.940828402366863, "grad_norm": 0.48964983224868774, "learning_rate": 0.0002988172051971717, "loss": 0.0202, "step": 1680 }, { "epoch": 10.0, "grad_norm": 0.41640275716781616, "learning_rate": 0.0002987970998320073, "loss": 0.0192, "step": 1690 }, { "epoch": 10.059171597633137, "grad_norm": 0.3516252338886261, "learning_rate": 0.00029877682571437264, "loss": 0.0177, "step": 1700 }, { "epoch": 10.118343195266272, "grad_norm": 0.3049193322658539, "learning_rate": 0.00029875638286726086, "loss": 0.0183, "step": 1710 }, { "epoch": 10.177514792899409, "grad_norm": 0.277535617351532, "learning_rate": 0.0002987357713138564, "loss": 0.0166, "step": 1720 }, { "epoch": 10.236686390532544, "grad_norm": 0.3309115767478943, "learning_rate": 0.0002987149910775349, "loss": 0.0178, "step": 1730 }, { "epoch": 10.29585798816568, "grad_norm": 0.3174135684967041, "learning_rate": 0.0002986940421818637, "loss": 0.0176, "step": 1740 }, { "epoch": 10.355029585798816, "grad_norm": 0.27874478697776794, "learning_rate": 0.00029867292465060103, "loss": 0.0179, "step": 1750 }, { "epoch": 10.414201183431953, "grad_norm": 0.31313538551330566, "learning_rate": 0.00029865163850769657, "loss": 0.0176, "step": 1760 }, { "epoch": 10.47337278106509, "grad_norm": 0.45908409357070923, "learning_rate": 0.00029863018377729116, "loss": 0.0171, "step": 1770 }, { "epoch": 10.532544378698224, "grad_norm": 0.3590826094150543, "learning_rate": 0.0002986085604837169, "loss": 0.0201, "step": 1780 }, { "epoch": 10.591715976331361, "grad_norm": 0.21082349121570587, "learning_rate": 0.0002985867686514969, "loss": 0.0181, "step": 1790 }, { "epoch": 10.650887573964496, "grad_norm": 0.22552594542503357, "learning_rate": 0.00029856480830534564, "loss": 0.0161, "step": 1800 }, { "epoch": 10.710059171597633, "grad_norm": 0.3588227331638336, "learning_rate": 0.0002985426794701685, "loss": 0.0185, "step": 1810 }, { "epoch": 10.76923076923077, "grad_norm": 0.322765976190567, "learning_rate": 0.0002985203821710621, "loss": 0.0178, "step": 1820 }, { "epoch": 10.828402366863905, "grad_norm": 0.3384231925010681, "learning_rate": 0.000298497916433314, "loss": 0.0181, "step": 1830 }, { "epoch": 10.887573964497042, "grad_norm": 0.2912094295024872, "learning_rate": 0.000298475282282403, "loss": 0.0201, "step": 1840 }, { "epoch": 10.946745562130177, "grad_norm": 0.2750729024410248, "learning_rate": 0.00029845247974399854, "loss": 0.0198, "step": 1850 }, { "epoch": 11.005917159763314, "grad_norm": 0.4589415192604065, "learning_rate": 0.0002984295088439614, "loss": 0.02, "step": 1860 }, { "epoch": 11.065088757396449, "grad_norm": 0.38744813203811646, "learning_rate": 0.00029840636960834304, "loss": 0.0179, "step": 1870 }, { "epoch": 11.124260355029586, "grad_norm": 0.3215080499649048, "learning_rate": 0.000298383062063386, "loss": 0.017, "step": 1880 }, { "epoch": 11.183431952662723, "grad_norm": 0.30425411462783813, "learning_rate": 0.00029835958623552355, "loss": 0.0164, "step": 1890 }, { "epoch": 11.242603550295858, "grad_norm": 0.30013507604599, "learning_rate": 0.00029833594215137997, "loss": 0.0187, "step": 1900 }, { "epoch": 11.301775147928995, "grad_norm": 0.4685440957546234, "learning_rate": 0.00029831212983777016, "loss": 0.0193, "step": 1910 }, { "epoch": 11.36094674556213, "grad_norm": 0.18723516166210175, "learning_rate": 0.0002982881493217001, "loss": 0.0167, "step": 1920 }, { "epoch": 11.420118343195266, "grad_norm": 0.44742441177368164, "learning_rate": 0.00029826400063036627, "loss": 0.0168, "step": 1930 }, { "epoch": 11.479289940828401, "grad_norm": 0.3105863034725189, "learning_rate": 0.00029823968379115597, "loss": 0.0165, "step": 1940 }, { "epoch": 11.538461538461538, "grad_norm": 0.5177462697029114, "learning_rate": 0.0002982151988316472, "loss": 0.0175, "step": 1950 }, { "epoch": 11.597633136094675, "grad_norm": 0.4105543792247772, "learning_rate": 0.00029819054577960867, "loss": 0.0179, "step": 1960 }, { "epoch": 11.65680473372781, "grad_norm": 0.24997715651988983, "learning_rate": 0.00029816572466299967, "loss": 0.0176, "step": 1970 }, { "epoch": 11.715976331360947, "grad_norm": 0.28764787316322327, "learning_rate": 0.00029814073550997006, "loss": 0.0164, "step": 1980 }, { "epoch": 11.775147928994082, "grad_norm": 0.4386701285839081, "learning_rate": 0.0002981155783488604, "loss": 0.0171, "step": 1990 }, { "epoch": 11.834319526627219, "grad_norm": 0.5953875780105591, "learning_rate": 0.0002980902532082017, "loss": 0.0183, "step": 2000 }, { "epoch": 11.893491124260356, "grad_norm": 0.4514220356941223, "learning_rate": 0.0002980647601167154, "loss": 0.0187, "step": 2010 }, { "epoch": 11.95266272189349, "grad_norm": 0.4364643394947052, "learning_rate": 0.00029803909910331355, "loss": 0.017, "step": 2020 }, { "epoch": 12.011834319526628, "grad_norm": 0.4322860836982727, "learning_rate": 0.00029801327019709866, "loss": 0.0181, "step": 2030 }, { "epoch": 12.071005917159763, "grad_norm": 0.4648142457008362, "learning_rate": 0.0002979872734273635, "loss": 0.0156, "step": 2040 }, { "epoch": 12.1301775147929, "grad_norm": 0.26488634943962097, "learning_rate": 0.00029796110882359127, "loss": 0.0149, "step": 2050 }, { "epoch": 12.189349112426035, "grad_norm": 0.22131893038749695, "learning_rate": 0.0002979347764154557, "loss": 0.0147, "step": 2060 }, { "epoch": 12.248520710059172, "grad_norm": 0.42954498529434204, "learning_rate": 0.00029790827623282047, "loss": 0.0165, "step": 2070 }, { "epoch": 12.307692307692308, "grad_norm": 0.2694082260131836, "learning_rate": 0.00029788160830573993, "loss": 0.0144, "step": 2080 }, { "epoch": 12.366863905325443, "grad_norm": 0.2838428020477295, "learning_rate": 0.00029785477266445836, "loss": 0.0153, "step": 2090 }, { "epoch": 12.42603550295858, "grad_norm": 0.27781984210014343, "learning_rate": 0.0002978277693394104, "loss": 0.0143, "step": 2100 }, { "epoch": 12.485207100591715, "grad_norm": 0.36948660016059875, "learning_rate": 0.0002978005983612208, "loss": 0.0171, "step": 2110 }, { "epoch": 12.544378698224852, "grad_norm": 0.5213725566864014, "learning_rate": 0.00029777325976070447, "loss": 0.0177, "step": 2120 }, { "epoch": 12.603550295857989, "grad_norm": 0.22739000618457794, "learning_rate": 0.00029774575356886646, "loss": 0.016, "step": 2130 }, { "epoch": 12.662721893491124, "grad_norm": 0.23930171132087708, "learning_rate": 0.0002977180798169018, "loss": 0.0154, "step": 2140 }, { "epoch": 12.721893491124261, "grad_norm": 0.2857344448566437, "learning_rate": 0.0002976902385361957, "loss": 0.0159, "step": 2150 }, { "epoch": 12.781065088757396, "grad_norm": 0.3301510810852051, "learning_rate": 0.00029766222975832323, "loss": 0.0165, "step": 2160 }, { "epoch": 12.840236686390533, "grad_norm": 0.34791597723960876, "learning_rate": 0.0002976340535150494, "loss": 0.016, "step": 2170 }, { "epoch": 12.899408284023668, "grad_norm": 0.30873966217041016, "learning_rate": 0.0002976057098383293, "loss": 0.0155, "step": 2180 }, { "epoch": 12.958579881656805, "grad_norm": 0.266801655292511, "learning_rate": 0.0002975771987603076, "loss": 0.0161, "step": 2190 }, { "epoch": 13.017751479289942, "grad_norm": 0.31169557571411133, "learning_rate": 0.0002975485203133194, "loss": 0.0165, "step": 2200 }, { "epoch": 13.076923076923077, "grad_norm": 0.31138914823532104, "learning_rate": 0.0002975196745298889, "loss": 0.0154, "step": 2210 }, { "epoch": 13.136094674556213, "grad_norm": 0.21030279994010925, "learning_rate": 0.0002974906614427307, "loss": 0.0136, "step": 2220 }, { "epoch": 13.195266272189349, "grad_norm": 0.3075583279132843, "learning_rate": 0.00029746148108474865, "loss": 0.0136, "step": 2230 }, { "epoch": 13.254437869822485, "grad_norm": 0.359365850687027, "learning_rate": 0.00029743213348903665, "loss": 0.0133, "step": 2240 }, { "epoch": 13.31360946745562, "grad_norm": 0.27614426612854004, "learning_rate": 0.0002974026186888781, "loss": 0.0132, "step": 2250 }, { "epoch": 13.372781065088757, "grad_norm": 0.3228969871997833, "learning_rate": 0.00029737293671774613, "loss": 0.0166, "step": 2260 }, { "epoch": 13.431952662721894, "grad_norm": 0.241877481341362, "learning_rate": 0.0002973430876093033, "loss": 0.0145, "step": 2270 }, { "epoch": 13.49112426035503, "grad_norm": 0.20202961564064026, "learning_rate": 0.00029731307139740183, "loss": 0.0158, "step": 2280 }, { "epoch": 13.550295857988166, "grad_norm": 0.2550893723964691, "learning_rate": 0.00029728288811608355, "loss": 0.0149, "step": 2290 }, { "epoch": 13.609467455621301, "grad_norm": 0.25099050998687744, "learning_rate": 0.0002972525377995796, "loss": 0.0132, "step": 2300 }, { "epoch": 13.668639053254438, "grad_norm": 0.2423638105392456, "learning_rate": 0.0002972220204823105, "loss": 0.0146, "step": 2310 }, { "epoch": 13.727810650887575, "grad_norm": 0.2631838917732239, "learning_rate": 0.0002971913361988865, "loss": 0.016, "step": 2320 }, { "epoch": 13.78698224852071, "grad_norm": 0.47881340980529785, "learning_rate": 0.00029716048498410684, "loss": 0.0163, "step": 2330 }, { "epoch": 13.846153846153847, "grad_norm": 0.30391496419906616, "learning_rate": 0.00029712946687296025, "loss": 0.0159, "step": 2340 }, { "epoch": 13.905325443786982, "grad_norm": 0.35713884234428406, "learning_rate": 0.0002970982819006248, "loss": 0.0169, "step": 2350 }, { "epoch": 13.964497041420119, "grad_norm": 0.24002473056316376, "learning_rate": 0.00029706693010246756, "loss": 0.0158, "step": 2360 }, { "epoch": 14.023668639053254, "grad_norm": 0.29464787244796753, "learning_rate": 0.0002970354115140452, "loss": 0.0141, "step": 2370 }, { "epoch": 14.08284023668639, "grad_norm": 0.2377527803182602, "learning_rate": 0.00029700372617110307, "loss": 0.0146, "step": 2380 }, { "epoch": 14.142011834319527, "grad_norm": 0.20373527705669403, "learning_rate": 0.00029697187410957605, "loss": 0.016, "step": 2390 }, { "epoch": 14.201183431952662, "grad_norm": 0.2681509852409363, "learning_rate": 0.00029693985536558774, "loss": 0.0159, "step": 2400 }, { "epoch": 14.2603550295858, "grad_norm": 0.24753984808921814, "learning_rate": 0.00029690766997545116, "loss": 0.0154, "step": 2410 }, { "epoch": 14.319526627218934, "grad_norm": 0.3421126902103424, "learning_rate": 0.000296875317975668, "loss": 0.0139, "step": 2420 }, { "epoch": 14.378698224852071, "grad_norm": 0.32215237617492676, "learning_rate": 0.00029684279940292907, "loss": 0.016, "step": 2430 }, { "epoch": 14.437869822485208, "grad_norm": 0.3105561137199402, "learning_rate": 0.00029681011429411394, "loss": 0.0145, "step": 2440 }, { "epoch": 14.497041420118343, "grad_norm": 0.34773972630500793, "learning_rate": 0.00029677726268629136, "loss": 0.0148, "step": 2450 }, { "epoch": 14.55621301775148, "grad_norm": 0.39935749769210815, "learning_rate": 0.0002967442446167186, "loss": 0.0136, "step": 2460 }, { "epoch": 14.615384615384615, "grad_norm": 0.24097128212451935, "learning_rate": 0.00029671106012284183, "loss": 0.0146, "step": 2470 }, { "epoch": 14.674556213017752, "grad_norm": 0.2576833963394165, "learning_rate": 0.00029667770924229593, "loss": 0.0146, "step": 2480 }, { "epoch": 14.733727810650887, "grad_norm": 0.24078258872032166, "learning_rate": 0.00029664419201290453, "loss": 0.0138, "step": 2490 }, { "epoch": 14.792899408284024, "grad_norm": 0.24760310351848602, "learning_rate": 0.00029661050847267997, "loss": 0.0132, "step": 2500 }, { "epoch": 14.85207100591716, "grad_norm": 0.33584150671958923, "learning_rate": 0.00029657665865982303, "loss": 0.0145, "step": 2510 }, { "epoch": 14.911242603550296, "grad_norm": 0.3172397315502167, "learning_rate": 0.00029654264261272324, "loss": 0.015, "step": 2520 }, { "epoch": 14.970414201183432, "grad_norm": 0.3141225278377533, "learning_rate": 0.0002965084603699585, "loss": 0.0163, "step": 2530 }, { "epoch": 15.029585798816568, "grad_norm": 0.20409420132637024, "learning_rate": 0.00029647411197029536, "loss": 0.0153, "step": 2540 }, { "epoch": 15.088757396449704, "grad_norm": 0.22532908618450165, "learning_rate": 0.00029643959745268873, "loss": 0.0149, "step": 2550 }, { "epoch": 15.14792899408284, "grad_norm": 0.39849212765693665, "learning_rate": 0.0002964049168562818, "loss": 0.0138, "step": 2560 }, { "epoch": 15.207100591715976, "grad_norm": 0.3364091217517853, "learning_rate": 0.0002963700702204064, "loss": 0.0149, "step": 2570 }, { "epoch": 15.266272189349113, "grad_norm": 0.2697981894016266, "learning_rate": 0.00029633505758458234, "loss": 0.0147, "step": 2580 }, { "epoch": 15.325443786982248, "grad_norm": 0.2546292841434479, "learning_rate": 0.0002962998789885179, "loss": 0.0136, "step": 2590 }, { "epoch": 15.384615384615385, "grad_norm": 0.1518646627664566, "learning_rate": 0.0002962645344721095, "loss": 0.0133, "step": 2600 }, { "epoch": 15.44378698224852, "grad_norm": 0.20934759080410004, "learning_rate": 0.00029622902407544185, "loss": 0.0135, "step": 2610 }, { "epoch": 15.502958579881657, "grad_norm": 0.15325278043746948, "learning_rate": 0.0002961933478387876, "loss": 0.0134, "step": 2620 }, { "epoch": 15.562130177514792, "grad_norm": 0.2582017183303833, "learning_rate": 0.00029615750580260755, "loss": 0.0128, "step": 2630 }, { "epoch": 15.621301775147929, "grad_norm": 0.21875867247581482, "learning_rate": 0.00029612149800755066, "loss": 0.0123, "step": 2640 }, { "epoch": 15.680473372781066, "grad_norm": 0.18626302480697632, "learning_rate": 0.00029608532449445374, "loss": 0.0118, "step": 2650 }, { "epoch": 15.7396449704142, "grad_norm": 0.26508718729019165, "learning_rate": 0.0002960489853043415, "loss": 0.0148, "step": 2660 }, { "epoch": 15.798816568047338, "grad_norm": 0.3234412372112274, "learning_rate": 0.0002960124804784267, "loss": 0.0149, "step": 2670 }, { "epoch": 15.857988165680473, "grad_norm": 0.27694663405418396, "learning_rate": 0.00029597581005810986, "loss": 0.014, "step": 2680 }, { "epoch": 15.91715976331361, "grad_norm": 0.19466620683670044, "learning_rate": 0.00029593897408497933, "loss": 0.0143, "step": 2690 }, { "epoch": 15.976331360946746, "grad_norm": 0.210581436753273, "learning_rate": 0.0002959019726008112, "loss": 0.013, "step": 2700 }, { "epoch": 16.035502958579883, "grad_norm": 0.3872358500957489, "learning_rate": 0.00029586480564756923, "loss": 0.0136, "step": 2710 }, { "epoch": 16.09467455621302, "grad_norm": 0.2559163272380829, "learning_rate": 0.00029582747326740495, "loss": 0.0129, "step": 2720 }, { "epoch": 16.153846153846153, "grad_norm": 0.1259068101644516, "learning_rate": 0.00029578997550265735, "loss": 0.0117, "step": 2730 }, { "epoch": 16.21301775147929, "grad_norm": 0.308110773563385, "learning_rate": 0.0002957523123958532, "loss": 0.0129, "step": 2740 }, { "epoch": 16.272189349112427, "grad_norm": 0.2767903804779053, "learning_rate": 0.0002957144839897065, "loss": 0.0128, "step": 2750 }, { "epoch": 16.331360946745562, "grad_norm": 0.24480606615543365, "learning_rate": 0.00029567649032711895, "loss": 0.0122, "step": 2760 }, { "epoch": 16.390532544378697, "grad_norm": 0.27840766310691833, "learning_rate": 0.0002956383314511795, "loss": 0.0124, "step": 2770 }, { "epoch": 16.449704142011836, "grad_norm": 0.37049365043640137, "learning_rate": 0.0002956000074051648, "loss": 0.0134, "step": 2780 }, { "epoch": 16.50887573964497, "grad_norm": 0.2843593657016754, "learning_rate": 0.00029556151823253833, "loss": 0.0146, "step": 2790 }, { "epoch": 16.568047337278106, "grad_norm": 0.4352964460849762, "learning_rate": 0.0002955228639769513, "loss": 0.0143, "step": 2800 }, { "epoch": 16.62721893491124, "grad_norm": 0.3345726430416107, "learning_rate": 0.00029548404468224173, "loss": 0.0144, "step": 2810 }, { "epoch": 16.68639053254438, "grad_norm": 0.3143850266933441, "learning_rate": 0.0002954450603924351, "loss": 0.0133, "step": 2820 }, { "epoch": 16.745562130177515, "grad_norm": 0.28779667615890503, "learning_rate": 0.00029540591115174406, "loss": 0.0119, "step": 2830 }, { "epoch": 16.80473372781065, "grad_norm": 0.27524858713150024, "learning_rate": 0.000295366597004568, "loss": 0.0125, "step": 2840 }, { "epoch": 16.86390532544379, "grad_norm": 0.30466359853744507, "learning_rate": 0.00029532711799549365, "loss": 0.0128, "step": 2850 }, { "epoch": 16.923076923076923, "grad_norm": 0.29718518257141113, "learning_rate": 0.00029528747416929463, "loss": 0.0118, "step": 2860 }, { "epoch": 16.98224852071006, "grad_norm": 0.2482769787311554, "learning_rate": 0.00029524766557093143, "loss": 0.0133, "step": 2870 }, { "epoch": 17.041420118343197, "grad_norm": 0.20062217116355896, "learning_rate": 0.0002952076922455514, "loss": 0.0134, "step": 2880 }, { "epoch": 17.100591715976332, "grad_norm": 0.22595195472240448, "learning_rate": 0.00029516755423848864, "loss": 0.0114, "step": 2890 }, { "epoch": 17.159763313609467, "grad_norm": 0.18651042878627777, "learning_rate": 0.0002951272515952643, "loss": 0.0114, "step": 2900 }, { "epoch": 17.218934911242602, "grad_norm": 0.27541065216064453, "learning_rate": 0.00029508678436158606, "loss": 0.0125, "step": 2910 }, { "epoch": 17.27810650887574, "grad_norm": 0.3520645797252655, "learning_rate": 0.0002950461525833481, "loss": 0.0131, "step": 2920 }, { "epoch": 17.337278106508876, "grad_norm": 0.249802827835083, "learning_rate": 0.0002950053563066315, "loss": 0.0113, "step": 2930 }, { "epoch": 17.39644970414201, "grad_norm": 0.2817287743091583, "learning_rate": 0.0002949643955777038, "loss": 0.0144, "step": 2940 }, { "epoch": 17.45562130177515, "grad_norm": 0.2627015709877014, "learning_rate": 0.0002949232704430189, "loss": 0.012, "step": 2950 }, { "epoch": 17.514792899408285, "grad_norm": 0.19721731543540955, "learning_rate": 0.00029488198094921735, "loss": 0.013, "step": 2960 }, { "epoch": 17.57396449704142, "grad_norm": 0.22856706380844116, "learning_rate": 0.00029484052714312607, "loss": 0.012, "step": 2970 }, { "epoch": 17.633136094674555, "grad_norm": 0.257629930973053, "learning_rate": 0.0002947989090717583, "loss": 0.0132, "step": 2980 }, { "epoch": 17.692307692307693, "grad_norm": 0.20652718842029572, "learning_rate": 0.00029475712678231347, "loss": 0.0113, "step": 2990 }, { "epoch": 17.75147928994083, "grad_norm": 0.22036686539649963, "learning_rate": 0.00029471518032217736, "loss": 0.0141, "step": 3000 }, { "epoch": 17.810650887573964, "grad_norm": 0.2276190221309662, "learning_rate": 0.000294673069738922, "loss": 0.0148, "step": 3010 }, { "epoch": 17.869822485207102, "grad_norm": 0.32997214794158936, "learning_rate": 0.0002946307950803054, "loss": 0.0128, "step": 3020 }, { "epoch": 17.928994082840237, "grad_norm": 0.2728573977947235, "learning_rate": 0.0002945883563942719, "loss": 0.0134, "step": 3030 }, { "epoch": 17.988165680473372, "grad_norm": 0.26944658160209656, "learning_rate": 0.0002945457537289514, "loss": 0.013, "step": 3040 }, { "epoch": 18.047337278106507, "grad_norm": 0.19238656759262085, "learning_rate": 0.00029450298713266026, "loss": 0.0122, "step": 3050 }, { "epoch": 18.106508875739646, "grad_norm": 0.280457079410553, "learning_rate": 0.00029446005665390056, "loss": 0.0113, "step": 3060 }, { "epoch": 18.16568047337278, "grad_norm": 0.3050542175769806, "learning_rate": 0.00029441696234136017, "loss": 0.0113, "step": 3070 }, { "epoch": 18.224852071005916, "grad_norm": 0.20029912889003754, "learning_rate": 0.00029437370424391287, "loss": 0.0122, "step": 3080 }, { "epoch": 18.284023668639055, "grad_norm": 0.2736499011516571, "learning_rate": 0.00029433028241061815, "loss": 0.0115, "step": 3090 }, { "epoch": 18.34319526627219, "grad_norm": 0.1953633576631546, "learning_rate": 0.00029428669689072123, "loss": 0.0128, "step": 3100 }, { "epoch": 18.402366863905325, "grad_norm": 0.23280967772006989, "learning_rate": 0.0002942429477336529, "loss": 0.0124, "step": 3110 }, { "epoch": 18.46153846153846, "grad_norm": 0.2591375708580017, "learning_rate": 0.0002941990349890296, "loss": 0.0124, "step": 3120 }, { "epoch": 18.5207100591716, "grad_norm": 0.23849175870418549, "learning_rate": 0.00029415495870665327, "loss": 0.0134, "step": 3130 }, { "epoch": 18.579881656804734, "grad_norm": 0.30432355403900146, "learning_rate": 0.0002941107189365112, "loss": 0.0125, "step": 3140 }, { "epoch": 18.63905325443787, "grad_norm": 0.25189709663391113, "learning_rate": 0.00029406631572877646, "loss": 0.0121, "step": 3150 }, { "epoch": 18.698224852071007, "grad_norm": 0.2803211212158203, "learning_rate": 0.000294021749133807, "loss": 0.0118, "step": 3160 }, { "epoch": 18.757396449704142, "grad_norm": 0.2129519134759903, "learning_rate": 0.00029397701920214645, "loss": 0.0113, "step": 3170 }, { "epoch": 18.816568047337277, "grad_norm": 0.29296520352363586, "learning_rate": 0.0002939321259845234, "loss": 0.0122, "step": 3180 }, { "epoch": 18.875739644970416, "grad_norm": 0.250861257314682, "learning_rate": 0.00029388706953185185, "loss": 0.0118, "step": 3190 }, { "epoch": 18.93491124260355, "grad_norm": 0.20555175840854645, "learning_rate": 0.00029384184989523083, "loss": 0.0113, "step": 3200 }, { "epoch": 18.994082840236686, "grad_norm": 0.2816445231437683, "learning_rate": 0.00029379646712594435, "loss": 0.0117, "step": 3210 }, { "epoch": 19.05325443786982, "grad_norm": 0.18015918135643005, "learning_rate": 0.00029375092127546164, "loss": 0.0102, "step": 3220 }, { "epoch": 19.11242603550296, "grad_norm": 0.2792223393917084, "learning_rate": 0.0002937052123954367, "loss": 0.0113, "step": 3230 }, { "epoch": 19.171597633136095, "grad_norm": 0.18722820281982422, "learning_rate": 0.0002936593405377085, "loss": 0.0117, "step": 3240 }, { "epoch": 19.23076923076923, "grad_norm": 0.19241203367710114, "learning_rate": 0.00029361330575430075, "loss": 0.0106, "step": 3250 }, { "epoch": 19.28994082840237, "grad_norm": 0.19137434661388397, "learning_rate": 0.0002935671080974222, "loss": 0.0103, "step": 3260 }, { "epoch": 19.349112426035504, "grad_norm": 0.2352636456489563, "learning_rate": 0.000293520747619466, "loss": 0.0101, "step": 3270 }, { "epoch": 19.40828402366864, "grad_norm": 0.19469325244426727, "learning_rate": 0.0002934742243730101, "loss": 0.0112, "step": 3280 }, { "epoch": 19.467455621301774, "grad_norm": 0.18880976736545563, "learning_rate": 0.000293427538410817, "loss": 0.0099, "step": 3290 }, { "epoch": 19.526627218934912, "grad_norm": 0.22865886986255646, "learning_rate": 0.0002933806897858339, "loss": 0.0112, "step": 3300 }, { "epoch": 19.585798816568047, "grad_norm": 0.2087821513414383, "learning_rate": 0.0002933336785511923, "loss": 0.0104, "step": 3310 }, { "epoch": 19.644970414201183, "grad_norm": 0.21453672647476196, "learning_rate": 0.0002932865047602081, "loss": 0.0113, "step": 3320 }, { "epoch": 19.70414201183432, "grad_norm": 0.2295752912759781, "learning_rate": 0.0002932391684663818, "loss": 0.0116, "step": 3330 }, { "epoch": 19.763313609467456, "grad_norm": 0.20551998913288116, "learning_rate": 0.0002931916697233978, "loss": 0.0109, "step": 3340 }, { "epoch": 19.82248520710059, "grad_norm": 0.17892783880233765, "learning_rate": 0.0002931440085851251, "loss": 0.0108, "step": 3350 }, { "epoch": 19.881656804733726, "grad_norm": 0.13540925085544586, "learning_rate": 0.0002930961851056167, "loss": 0.0103, "step": 3360 }, { "epoch": 19.940828402366865, "grad_norm": 0.2291262000799179, "learning_rate": 0.0002930481993391098, "loss": 0.0112, "step": 3370 }, { "epoch": 20.0, "grad_norm": 0.20277675986289978, "learning_rate": 0.00029300005134002557, "loss": 0.0109, "step": 3380 }, { "epoch": 20.059171597633135, "grad_norm": 0.15098540484905243, "learning_rate": 0.0002929517411629691, "loss": 0.0116, "step": 3390 }, { "epoch": 20.118343195266274, "grad_norm": 0.2044897824525833, "learning_rate": 0.00029290326886272967, "loss": 0.0111, "step": 3400 }, { "epoch": 20.17751479289941, "grad_norm": 0.21725504100322723, "learning_rate": 0.0002928546344942802, "loss": 0.0097, "step": 3410 }, { "epoch": 20.236686390532544, "grad_norm": 0.16622619330883026, "learning_rate": 0.00029280583811277735, "loss": 0.0097, "step": 3420 }, { "epoch": 20.29585798816568, "grad_norm": 0.20357896387577057, "learning_rate": 0.00029275687977356186, "loss": 0.0098, "step": 3430 }, { "epoch": 20.355029585798817, "grad_norm": 0.13365399837493896, "learning_rate": 0.0002927077595321578, "loss": 0.0105, "step": 3440 }, { "epoch": 20.414201183431953, "grad_norm": 0.21504147350788116, "learning_rate": 0.00029265847744427303, "loss": 0.0112, "step": 3450 }, { "epoch": 20.473372781065088, "grad_norm": 0.25173717737197876, "learning_rate": 0.0002926090335657989, "loss": 0.0092, "step": 3460 }, { "epoch": 20.532544378698226, "grad_norm": 0.2055933177471161, "learning_rate": 0.0002925594279528103, "loss": 0.0109, "step": 3470 }, { "epoch": 20.59171597633136, "grad_norm": 0.2620813846588135, "learning_rate": 0.0002925096606615655, "loss": 0.0114, "step": 3480 }, { "epoch": 20.650887573964496, "grad_norm": 0.3535516560077667, "learning_rate": 0.0002924597317485061, "loss": 0.0103, "step": 3490 }, { "epoch": 20.71005917159763, "grad_norm": 0.3035389184951782, "learning_rate": 0.00029240964127025715, "loss": 0.0117, "step": 3500 }, { "epoch": 20.76923076923077, "grad_norm": 0.2327113002538681, "learning_rate": 0.0002923593892836268, "loss": 0.0114, "step": 3510 }, { "epoch": 20.828402366863905, "grad_norm": 0.24227726459503174, "learning_rate": 0.0002923089758456063, "loss": 0.0121, "step": 3520 }, { "epoch": 20.88757396449704, "grad_norm": 0.17488418519496918, "learning_rate": 0.0002922584010133702, "loss": 0.0121, "step": 3530 }, { "epoch": 20.94674556213018, "grad_norm": 0.23604446649551392, "learning_rate": 0.00029220766484427594, "loss": 0.0121, "step": 3540 }, { "epoch": 21.005917159763314, "grad_norm": 0.21806344389915466, "learning_rate": 0.000292156767395864, "loss": 0.0112, "step": 3550 }, { "epoch": 21.06508875739645, "grad_norm": 0.16898788511753082, "learning_rate": 0.0002921057087258577, "loss": 0.0126, "step": 3560 }, { "epoch": 21.124260355029588, "grad_norm": 0.21747981011867523, "learning_rate": 0.00029205448889216333, "loss": 0.0103, "step": 3570 }, { "epoch": 21.183431952662723, "grad_norm": 0.20911720395088196, "learning_rate": 0.00029200310795286983, "loss": 0.0108, "step": 3580 }, { "epoch": 21.242603550295858, "grad_norm": 0.21870778501033783, "learning_rate": 0.00029195156596624895, "loss": 0.0115, "step": 3590 }, { "epoch": 21.301775147928993, "grad_norm": 0.25113946199417114, "learning_rate": 0.000291899862990755, "loss": 0.0106, "step": 3600 }, { "epoch": 21.36094674556213, "grad_norm": 0.1931365728378296, "learning_rate": 0.00029184799908502497, "loss": 0.0101, "step": 3610 }, { "epoch": 21.420118343195266, "grad_norm": 0.12647129595279694, "learning_rate": 0.0002917959743078782, "loss": 0.01, "step": 3620 }, { "epoch": 21.4792899408284, "grad_norm": 0.2020137757062912, "learning_rate": 0.00029174378871831664, "loss": 0.01, "step": 3630 }, { "epoch": 21.53846153846154, "grad_norm": 0.17673088610172272, "learning_rate": 0.0002916914423755245, "loss": 0.0098, "step": 3640 }, { "epoch": 21.597633136094675, "grad_norm": 0.2463626265525818, "learning_rate": 0.00029163893533886847, "loss": 0.011, "step": 3650 }, { "epoch": 21.65680473372781, "grad_norm": 0.20313824713230133, "learning_rate": 0.00029158626766789727, "loss": 0.0103, "step": 3660 }, { "epoch": 21.715976331360945, "grad_norm": 0.13793836534023285, "learning_rate": 0.0002915334394223419, "loss": 0.0109, "step": 3670 }, { "epoch": 21.775147928994084, "grad_norm": 0.20752137899398804, "learning_rate": 0.00029148045066211553, "loss": 0.011, "step": 3680 }, { "epoch": 21.83431952662722, "grad_norm": 0.21220724284648895, "learning_rate": 0.0002914273014473133, "loss": 0.0114, "step": 3690 }, { "epoch": 21.893491124260354, "grad_norm": 0.2710641026496887, "learning_rate": 0.0002913739918382123, "loss": 0.012, "step": 3700 }, { "epoch": 21.952662721893493, "grad_norm": 0.1975807398557663, "learning_rate": 0.0002913205218952716, "loss": 0.011, "step": 3710 }, { "epoch": 22.011834319526628, "grad_norm": 0.3515750467777252, "learning_rate": 0.00029126689167913205, "loss": 0.0121, "step": 3720 }, { "epoch": 22.071005917159763, "grad_norm": 0.21551652252674103, "learning_rate": 0.00029121310125061625, "loss": 0.0104, "step": 3730 }, { "epoch": 22.130177514792898, "grad_norm": 0.18497437238693237, "learning_rate": 0.0002911591506707286, "loss": 0.0113, "step": 3740 }, { "epoch": 22.189349112426036, "grad_norm": 0.20644636452198029, "learning_rate": 0.00029110504000065497, "loss": 0.011, "step": 3750 }, { "epoch": 22.24852071005917, "grad_norm": 0.2143058329820633, "learning_rate": 0.00029105076930176297, "loss": 0.0099, "step": 3760 }, { "epoch": 22.307692307692307, "grad_norm": 0.1615307331085205, "learning_rate": 0.00029099633863560157, "loss": 0.0094, "step": 3770 }, { "epoch": 22.366863905325445, "grad_norm": 0.19662319123744965, "learning_rate": 0.00029094174806390115, "loss": 0.0102, "step": 3780 }, { "epoch": 22.42603550295858, "grad_norm": 0.23011863231658936, "learning_rate": 0.00029088699764857364, "loss": 0.0104, "step": 3790 }, { "epoch": 22.485207100591715, "grad_norm": 0.22892245650291443, "learning_rate": 0.0002908320874517119, "loss": 0.0111, "step": 3800 }, { "epoch": 22.54437869822485, "grad_norm": 0.18399584293365479, "learning_rate": 0.0002907770175355905, "loss": 0.0098, "step": 3810 }, { "epoch": 22.60355029585799, "grad_norm": 0.1904982626438141, "learning_rate": 0.00029072178796266454, "loss": 0.0104, "step": 3820 }, { "epoch": 22.662721893491124, "grad_norm": 0.16625170409679413, "learning_rate": 0.00029066639879557065, "loss": 0.009, "step": 3830 }, { "epoch": 22.72189349112426, "grad_norm": 0.2138693630695343, "learning_rate": 0.0002906108500971263, "loss": 0.0078, "step": 3840 }, { "epoch": 22.781065088757398, "grad_norm": 0.18212242424488068, "learning_rate": 0.00029055514193032986, "loss": 0.0098, "step": 3850 }, { "epoch": 22.840236686390533, "grad_norm": 0.261758416891098, "learning_rate": 0.00029049927435836074, "loss": 0.0108, "step": 3860 }, { "epoch": 22.899408284023668, "grad_norm": 0.17313551902770996, "learning_rate": 0.00029044324744457875, "loss": 0.0112, "step": 3870 }, { "epoch": 22.958579881656803, "grad_norm": 0.17304284870624542, "learning_rate": 0.0002903870612525248, "loss": 0.0105, "step": 3880 }, { "epoch": 23.01775147928994, "grad_norm": 0.1755586415529251, "learning_rate": 0.0002903307158459202, "loss": 0.011, "step": 3890 }, { "epoch": 23.076923076923077, "grad_norm": 0.2648956775665283, "learning_rate": 0.000290274211288667, "loss": 0.0109, "step": 3900 }, { "epoch": 23.13609467455621, "grad_norm": 0.1714528203010559, "learning_rate": 0.0002902175476448476, "loss": 0.0092, "step": 3910 }, { "epoch": 23.19526627218935, "grad_norm": 0.17324870824813843, "learning_rate": 0.0002901607249787249, "loss": 0.0097, "step": 3920 }, { "epoch": 23.254437869822485, "grad_norm": 0.1310024857521057, "learning_rate": 0.0002901037433547421, "loss": 0.009, "step": 3930 }, { "epoch": 23.31360946745562, "grad_norm": 0.158621147274971, "learning_rate": 0.00029004660283752276, "loss": 0.0092, "step": 3940 }, { "epoch": 23.37278106508876, "grad_norm": 0.2727070152759552, "learning_rate": 0.0002899893034918705, "loss": 0.0103, "step": 3950 }, { "epoch": 23.431952662721894, "grad_norm": 0.23777635395526886, "learning_rate": 0.0002899318453827692, "loss": 0.0095, "step": 3960 }, { "epoch": 23.49112426035503, "grad_norm": 0.12102575600147247, "learning_rate": 0.00028987422857538277, "loss": 0.0098, "step": 3970 }, { "epoch": 23.550295857988164, "grad_norm": 0.2163602113723755, "learning_rate": 0.00028981645313505506, "loss": 0.0117, "step": 3980 }, { "epoch": 23.609467455621303, "grad_norm": 0.224243625998497, "learning_rate": 0.0002897585191273098, "loss": 0.011, "step": 3990 }, { "epoch": 23.668639053254438, "grad_norm": 0.23942671716213226, "learning_rate": 0.00028970042661785074, "loss": 0.0102, "step": 4000 }, { "epoch": 23.727810650887573, "grad_norm": 0.3103281557559967, "learning_rate": 0.00028964217567256114, "loss": 0.0106, "step": 4010 }, { "epoch": 23.78698224852071, "grad_norm": 0.17370326817035675, "learning_rate": 0.0002895837663575041, "loss": 0.0114, "step": 4020 }, { "epoch": 23.846153846153847, "grad_norm": 0.21732768416404724, "learning_rate": 0.00028952519873892213, "loss": 0.0112, "step": 4030 }, { "epoch": 23.90532544378698, "grad_norm": 0.3384005129337311, "learning_rate": 0.00028946647288323766, "loss": 0.0118, "step": 4040 }, { "epoch": 23.964497041420117, "grad_norm": 0.2532654106616974, "learning_rate": 0.00028940758885705225, "loss": 0.0116, "step": 4050 }, { "epoch": 24.023668639053255, "grad_norm": 0.27346479892730713, "learning_rate": 0.00028934854672714693, "loss": 0.0111, "step": 4060 }, { "epoch": 24.08284023668639, "grad_norm": 0.20225337147712708, "learning_rate": 0.000289289346560482, "loss": 0.01, "step": 4070 }, { "epoch": 24.142011834319526, "grad_norm": 0.21119368076324463, "learning_rate": 0.00028922998842419715, "loss": 0.01, "step": 4080 }, { "epoch": 24.201183431952664, "grad_norm": 0.3065793812274933, "learning_rate": 0.000289170472385611, "loss": 0.0108, "step": 4090 }, { "epoch": 24.2603550295858, "grad_norm": 0.2502423822879791, "learning_rate": 0.00028911079851222143, "loss": 0.01, "step": 4100 }, { "epoch": 24.319526627218934, "grad_norm": 0.28017526865005493, "learning_rate": 0.00028905096687170526, "loss": 0.0104, "step": 4110 }, { "epoch": 24.37869822485207, "grad_norm": 0.2425481081008911, "learning_rate": 0.0002889909775319182, "loss": 0.0097, "step": 4120 }, { "epoch": 24.437869822485208, "grad_norm": 0.18837901949882507, "learning_rate": 0.0002889308305608948, "loss": 0.0098, "step": 4130 }, { "epoch": 24.497041420118343, "grad_norm": 0.3388248085975647, "learning_rate": 0.0002888705260268484, "loss": 0.0102, "step": 4140 }, { "epoch": 24.556213017751478, "grad_norm": 0.3918682038784027, "learning_rate": 0.0002888100639981712, "loss": 0.0111, "step": 4150 }, { "epoch": 24.615384615384617, "grad_norm": 0.32467445731163025, "learning_rate": 0.0002887494445434338, "loss": 0.0103, "step": 4160 }, { "epoch": 24.674556213017752, "grad_norm": 0.19867871701717377, "learning_rate": 0.0002886886677313853, "loss": 0.0109, "step": 4170 }, { "epoch": 24.733727810650887, "grad_norm": 0.26291367411613464, "learning_rate": 0.0002886277336309535, "loss": 0.0122, "step": 4180 }, { "epoch": 24.792899408284022, "grad_norm": 0.2905069589614868, "learning_rate": 0.0002885666423112444, "loss": 0.0114, "step": 4190 }, { "epoch": 24.85207100591716, "grad_norm": 0.2822631597518921, "learning_rate": 0.0002885053938415424, "loss": 0.011, "step": 4200 }, { "epoch": 24.911242603550296, "grad_norm": 0.2982347011566162, "learning_rate": 0.00028844398829131, "loss": 0.0112, "step": 4210 }, { "epoch": 24.97041420118343, "grad_norm": 0.22791704535484314, "learning_rate": 0.000288382425730188, "loss": 0.011, "step": 4220 }, { "epoch": 25.02958579881657, "grad_norm": 0.20972786843776703, "learning_rate": 0.0002883207062279951, "loss": 0.0096, "step": 4230 }, { "epoch": 25.088757396449704, "grad_norm": 0.17601874470710754, "learning_rate": 0.0002882588298547282, "loss": 0.0098, "step": 4240 }, { "epoch": 25.14792899408284, "grad_norm": 0.19682249426841736, "learning_rate": 0.0002881967966805619, "loss": 0.0096, "step": 4250 }, { "epoch": 25.207100591715978, "grad_norm": 0.1817520260810852, "learning_rate": 0.0002881346067758488, "loss": 0.0089, "step": 4260 }, { "epoch": 25.266272189349113, "grad_norm": 0.21619878709316254, "learning_rate": 0.00028807226021111915, "loss": 0.0087, "step": 4270 }, { "epoch": 25.325443786982248, "grad_norm": 0.18786078691482544, "learning_rate": 0.00028800975705708086, "loss": 0.0089, "step": 4280 }, { "epoch": 25.384615384615383, "grad_norm": 0.13016484677791595, "learning_rate": 0.0002879470973846195, "loss": 0.0085, "step": 4290 }, { "epoch": 25.443786982248522, "grad_norm": 0.16964244842529297, "learning_rate": 0.0002878842812647981, "loss": 0.0085, "step": 4300 }, { "epoch": 25.502958579881657, "grad_norm": 0.17897994816303253, "learning_rate": 0.0002878213087688571, "loss": 0.0087, "step": 4310 }, { "epoch": 25.562130177514792, "grad_norm": 0.22136935591697693, "learning_rate": 0.00028775817996821437, "loss": 0.0095, "step": 4320 }, { "epoch": 25.62130177514793, "grad_norm": 0.3095250129699707, "learning_rate": 0.000287694894934465, "loss": 0.0105, "step": 4330 }, { "epoch": 25.680473372781066, "grad_norm": 0.2547779977321625, "learning_rate": 0.00028763145373938113, "loss": 0.0095, "step": 4340 }, { "epoch": 25.7396449704142, "grad_norm": 0.1995244026184082, "learning_rate": 0.0002875678564549123, "loss": 0.009, "step": 4350 }, { "epoch": 25.798816568047336, "grad_norm": 0.22017918527126312, "learning_rate": 0.00028750410315318484, "loss": 0.0101, "step": 4360 }, { "epoch": 25.857988165680474, "grad_norm": 0.20015573501586914, "learning_rate": 0.0002874401939065021, "loss": 0.0092, "step": 4370 }, { "epoch": 25.91715976331361, "grad_norm": 0.1900600790977478, "learning_rate": 0.0002873761287873442, "loss": 0.0083, "step": 4380 }, { "epoch": 25.976331360946745, "grad_norm": 0.1540488302707672, "learning_rate": 0.00028731190786836827, "loss": 0.0089, "step": 4390 }, { "epoch": 26.035502958579883, "grad_norm": 0.13939014077186584, "learning_rate": 0.0002872475312224079, "loss": 0.0081, "step": 4400 }, { "epoch": 26.09467455621302, "grad_norm": 0.19682931900024414, "learning_rate": 0.00028718299892247325, "loss": 0.0092, "step": 4410 }, { "epoch": 26.153846153846153, "grad_norm": 0.16200615465641022, "learning_rate": 0.0002871183110417514, "loss": 0.0087, "step": 4420 }, { "epoch": 26.21301775147929, "grad_norm": 0.22560974955558777, "learning_rate": 0.00028705346765360535, "loss": 0.009, "step": 4430 }, { "epoch": 26.272189349112427, "grad_norm": 0.31287717819213867, "learning_rate": 0.0002869884688315748, "loss": 0.0101, "step": 4440 }, { "epoch": 26.331360946745562, "grad_norm": 0.2592962086200714, "learning_rate": 0.00028692331464937575, "loss": 0.0093, "step": 4450 }, { "epoch": 26.390532544378697, "grad_norm": 0.14459161460399628, "learning_rate": 0.00028685800518090017, "loss": 0.0099, "step": 4460 }, { "epoch": 26.449704142011836, "grad_norm": 0.21127337217330933, "learning_rate": 0.00028679254050021627, "loss": 0.0094, "step": 4470 }, { "epoch": 26.50887573964497, "grad_norm": 0.15213431417942047, "learning_rate": 0.00028672692068156837, "loss": 0.0088, "step": 4480 }, { "epoch": 26.568047337278106, "grad_norm": 0.167250856757164, "learning_rate": 0.00028666114579937654, "loss": 0.0097, "step": 4490 }, { "epoch": 26.62721893491124, "grad_norm": 0.19437062740325928, "learning_rate": 0.000286595215928237, "loss": 0.0087, "step": 4500 }, { "epoch": 26.68639053254438, "grad_norm": 0.17050544917583466, "learning_rate": 0.00028652913114292133, "loss": 0.0094, "step": 4510 }, { "epoch": 26.745562130177515, "grad_norm": 0.30288705229759216, "learning_rate": 0.00028646289151837716, "loss": 0.0091, "step": 4520 }, { "epoch": 26.80473372781065, "grad_norm": 0.27088767290115356, "learning_rate": 0.0002863964971297275, "loss": 0.0104, "step": 4530 }, { "epoch": 26.86390532544379, "grad_norm": 0.2539435625076294, "learning_rate": 0.0002863299480522711, "loss": 0.0105, "step": 4540 }, { "epoch": 26.923076923076923, "grad_norm": 0.18370479345321655, "learning_rate": 0.000286263244361482, "loss": 0.0093, "step": 4550 }, { "epoch": 26.98224852071006, "grad_norm": 0.1516735851764679, "learning_rate": 0.00028619638613300953, "loss": 0.0098, "step": 4560 }, { "epoch": 27.041420118343197, "grad_norm": 0.1660565882921219, "learning_rate": 0.0002861293734426784, "loss": 0.0086, "step": 4570 }, { "epoch": 27.100591715976332, "grad_norm": 0.24182651937007904, "learning_rate": 0.0002860622063664884, "loss": 0.0091, "step": 4580 }, { "epoch": 27.159763313609467, "grad_norm": 0.1703873872756958, "learning_rate": 0.0002859948849806147, "loss": 0.0085, "step": 4590 }, { "epoch": 27.218934911242602, "grad_norm": 0.18562506139278412, "learning_rate": 0.00028592740936140697, "loss": 0.0105, "step": 4600 }, { "epoch": 27.27810650887574, "grad_norm": 0.23853589594364166, "learning_rate": 0.0002858597795853902, "loss": 0.009, "step": 4610 }, { "epoch": 27.337278106508876, "grad_norm": 0.19216947257518768, "learning_rate": 0.000285791995729264, "loss": 0.0096, "step": 4620 }, { "epoch": 27.39644970414201, "grad_norm": 0.14883655309677124, "learning_rate": 0.00028572405786990294, "loss": 0.0087, "step": 4630 }, { "epoch": 27.45562130177515, "grad_norm": 0.2883383631706238, "learning_rate": 0.00028565596608435596, "loss": 0.0094, "step": 4640 }, { "epoch": 27.514792899408285, "grad_norm": 0.10501532256603241, "learning_rate": 0.0002855877204498469, "loss": 0.0081, "step": 4650 }, { "epoch": 27.57396449704142, "grad_norm": 0.16976429522037506, "learning_rate": 0.00028551932104377373, "loss": 0.0089, "step": 4660 }, { "epoch": 27.633136094674555, "grad_norm": 0.16562263667583466, "learning_rate": 0.0002854507679437091, "loss": 0.0076, "step": 4670 }, { "epoch": 27.692307692307693, "grad_norm": 0.21052220463752747, "learning_rate": 0.00028538206122739976, "loss": 0.0089, "step": 4680 }, { "epoch": 27.75147928994083, "grad_norm": 0.15177708864212036, "learning_rate": 0.0002853132009727669, "loss": 0.008, "step": 4690 }, { "epoch": 27.810650887573964, "grad_norm": 0.15329276025295258, "learning_rate": 0.0002852441872579056, "loss": 0.009, "step": 4700 }, { "epoch": 27.869822485207102, "grad_norm": 0.2416708618402481, "learning_rate": 0.0002851750201610852, "loss": 0.0083, "step": 4710 }, { "epoch": 27.928994082840237, "grad_norm": 0.16569587588310242, "learning_rate": 0.0002851056997607487, "loss": 0.0098, "step": 4720 }, { "epoch": 27.988165680473372, "grad_norm": 0.18744386732578278, "learning_rate": 0.0002850362261355134, "loss": 0.0076, "step": 4730 }, { "epoch": 28.047337278106507, "grad_norm": 0.20107322931289673, "learning_rate": 0.0002849665993641699, "loss": 0.008, "step": 4740 }, { "epoch": 28.106508875739646, "grad_norm": 0.14973485469818115, "learning_rate": 0.00028489681952568286, "loss": 0.0077, "step": 4750 }, { "epoch": 28.16568047337278, "grad_norm": 0.160791277885437, "learning_rate": 0.00028482688669919027, "loss": 0.0093, "step": 4760 }, { "epoch": 28.224852071005916, "grad_norm": 0.2237534373998642, "learning_rate": 0.00028475680096400383, "loss": 0.0091, "step": 4770 }, { "epoch": 28.284023668639055, "grad_norm": 0.20066937804222107, "learning_rate": 0.0002846865623996085, "loss": 0.0086, "step": 4780 }, { "epoch": 28.34319526627219, "grad_norm": 0.2653585970401764, "learning_rate": 0.0002846161710856627, "loss": 0.0107, "step": 4790 }, { "epoch": 28.402366863905325, "grad_norm": 0.22519032657146454, "learning_rate": 0.0002845456271019979, "loss": 0.0101, "step": 4800 }, { "epoch": 28.46153846153846, "grad_norm": 0.2456853687763214, "learning_rate": 0.0002844749305286189, "loss": 0.0093, "step": 4810 }, { "epoch": 28.5207100591716, "grad_norm": 0.2507430613040924, "learning_rate": 0.0002844040814457035, "loss": 0.01, "step": 4820 }, { "epoch": 28.579881656804734, "grad_norm": 0.15306398272514343, "learning_rate": 0.0002843330799336024, "loss": 0.0088, "step": 4830 }, { "epoch": 28.63905325443787, "grad_norm": 0.2555418908596039, "learning_rate": 0.00028426192607283924, "loss": 0.0089, "step": 4840 }, { "epoch": 28.698224852071007, "grad_norm": 0.13397648930549622, "learning_rate": 0.0002841906199441104, "loss": 0.0089, "step": 4850 }, { "epoch": 28.757396449704142, "grad_norm": 0.16031193733215332, "learning_rate": 0.000284119161628285, "loss": 0.0082, "step": 4860 }, { "epoch": 28.816568047337277, "grad_norm": 0.1991170197725296, "learning_rate": 0.0002840475512064047, "loss": 0.0081, "step": 4870 }, { "epoch": 28.875739644970416, "grad_norm": 0.14624036848545074, "learning_rate": 0.00028397578875968366, "loss": 0.0084, "step": 4880 }, { "epoch": 28.93491124260355, "grad_norm": 0.1487826704978943, "learning_rate": 0.0002839038743695085, "loss": 0.0083, "step": 4890 }, { "epoch": 28.994082840236686, "grad_norm": 0.09791311621665955, "learning_rate": 0.0002838318081174382, "loss": 0.0076, "step": 4900 }, { "epoch": 29.05325443786982, "grad_norm": 0.18728315830230713, "learning_rate": 0.0002837595900852038, "loss": 0.0088, "step": 4910 }, { "epoch": 29.11242603550296, "grad_norm": 0.23431192338466644, "learning_rate": 0.0002836872203547087, "loss": 0.0085, "step": 4920 }, { "epoch": 29.171597633136095, "grad_norm": 0.19844567775726318, "learning_rate": 0.0002836146990080281, "loss": 0.0084, "step": 4930 }, { "epoch": 29.23076923076923, "grad_norm": 0.1210494413971901, "learning_rate": 0.00028354202612740944, "loss": 0.0078, "step": 4940 }, { "epoch": 29.28994082840237, "grad_norm": 0.19138078391551971, "learning_rate": 0.0002834692017952717, "loss": 0.0077, "step": 4950 }, { "epoch": 29.349112426035504, "grad_norm": 0.14302250742912292, "learning_rate": 0.00028339622609420585, "loss": 0.0085, "step": 4960 }, { "epoch": 29.40828402366864, "grad_norm": 0.19774751365184784, "learning_rate": 0.00028332309910697446, "loss": 0.0091, "step": 4970 }, { "epoch": 29.467455621301774, "grad_norm": 0.17938676476478577, "learning_rate": 0.0002832498209165117, "loss": 0.0091, "step": 4980 }, { "epoch": 29.526627218934912, "grad_norm": 0.15326610207557678, "learning_rate": 0.0002831763916059231, "loss": 0.0099, "step": 4990 }, { "epoch": 29.585798816568047, "grad_norm": 0.21555714309215546, "learning_rate": 0.0002831028112584857, "loss": 0.0094, "step": 5000 }, { "epoch": 29.644970414201183, "grad_norm": 0.21250919997692108, "learning_rate": 0.0002830290799576479, "loss": 0.0082, "step": 5010 }, { "epoch": 29.70414201183432, "grad_norm": 0.14400653541088104, "learning_rate": 0.00028295519778702904, "loss": 0.0097, "step": 5020 }, { "epoch": 29.763313609467456, "grad_norm": 0.12629567086696625, "learning_rate": 0.00028288116483041984, "loss": 0.0094, "step": 5030 }, { "epoch": 29.82248520710059, "grad_norm": 0.1341751217842102, "learning_rate": 0.0002828069811717819, "loss": 0.0089, "step": 5040 }, { "epoch": 29.881656804733726, "grad_norm": 0.1643371731042862, "learning_rate": 0.0002827326468952477, "loss": 0.0082, "step": 5050 }, { "epoch": 29.940828402366865, "grad_norm": 0.16587167978286743, "learning_rate": 0.0002826581620851207, "loss": 0.0074, "step": 5060 }, { "epoch": 30.0, "grad_norm": 0.14516565203666687, "learning_rate": 0.00028258352682587474, "loss": 0.0086, "step": 5070 }, { "epoch": 30.059171597633135, "grad_norm": 0.15625259280204773, "learning_rate": 0.00028250874120215467, "loss": 0.0083, "step": 5080 }, { "epoch": 30.118343195266274, "grad_norm": 0.11483515799045563, "learning_rate": 0.00028243380529877575, "loss": 0.0095, "step": 5090 }, { "epoch": 30.17751479289941, "grad_norm": 0.20352621376514435, "learning_rate": 0.00028235871920072347, "loss": 0.0093, "step": 5100 }, { "epoch": 30.236686390532544, "grad_norm": 0.1736299693584442, "learning_rate": 0.00028228348299315397, "loss": 0.008, "step": 5110 }, { "epoch": 30.29585798816568, "grad_norm": 0.16963917016983032, "learning_rate": 0.00028220809676139343, "loss": 0.008, "step": 5120 }, { "epoch": 30.355029585798817, "grad_norm": 0.16833923757076263, "learning_rate": 0.0002821325605909382, "loss": 0.0075, "step": 5130 }, { "epoch": 30.414201183431953, "grad_norm": 0.14498332142829895, "learning_rate": 0.00028205687456745474, "loss": 0.0088, "step": 5140 }, { "epoch": 30.473372781065088, "grad_norm": 0.1730290651321411, "learning_rate": 0.00028198103877677936, "loss": 0.0087, "step": 5150 }, { "epoch": 30.532544378698226, "grad_norm": 0.17101500928401947, "learning_rate": 0.0002819050533049184, "loss": 0.0077, "step": 5160 }, { "epoch": 30.59171597633136, "grad_norm": 0.11132028698921204, "learning_rate": 0.00028182891823804774, "loss": 0.0068, "step": 5170 }, { "epoch": 30.650887573964496, "grad_norm": 0.17907655239105225, "learning_rate": 0.0002817526336625131, "loss": 0.0079, "step": 5180 }, { "epoch": 30.71005917159763, "grad_norm": 0.1366843730211258, "learning_rate": 0.00028167619966482966, "loss": 0.0084, "step": 5190 }, { "epoch": 30.76923076923077, "grad_norm": 0.17823350429534912, "learning_rate": 0.000281599616331682, "loss": 0.009, "step": 5200 }, { "epoch": 30.828402366863905, "grad_norm": 0.2255176454782486, "learning_rate": 0.0002815228837499242, "loss": 0.0089, "step": 5210 }, { "epoch": 30.88757396449704, "grad_norm": 0.13070261478424072, "learning_rate": 0.0002814460020065795, "loss": 0.0091, "step": 5220 }, { "epoch": 30.94674556213018, "grad_norm": 0.18219444155693054, "learning_rate": 0.00028136897118884044, "loss": 0.009, "step": 5230 }, { "epoch": 31.005917159763314, "grad_norm": 0.13457664847373962, "learning_rate": 0.00028129179138406855, "loss": 0.0088, "step": 5240 }, { "epoch": 31.06508875739645, "grad_norm": 0.1826532781124115, "learning_rate": 0.00028121446267979417, "loss": 0.0078, "step": 5250 }, { "epoch": 31.124260355029588, "grad_norm": 0.18808114528656006, "learning_rate": 0.00028113698516371674, "loss": 0.0073, "step": 5260 }, { "epoch": 31.183431952662723, "grad_norm": 0.15909230709075928, "learning_rate": 0.00028105935892370446, "loss": 0.0077, "step": 5270 }, { "epoch": 31.242603550295858, "grad_norm": 0.11690393835306168, "learning_rate": 0.0002809815840477941, "loss": 0.0076, "step": 5280 }, { "epoch": 31.301775147928993, "grad_norm": 0.12304792553186417, "learning_rate": 0.0002809036606241909, "loss": 0.0074, "step": 5290 }, { "epoch": 31.36094674556213, "grad_norm": 0.17740017175674438, "learning_rate": 0.0002808255887412688, "loss": 0.0076, "step": 5300 }, { "epoch": 31.420118343195266, "grad_norm": 0.2053842693567276, "learning_rate": 0.00028074736848757, "loss": 0.0079, "step": 5310 }, { "epoch": 31.4792899408284, "grad_norm": 0.1719513237476349, "learning_rate": 0.00028066899995180493, "loss": 0.0084, "step": 5320 }, { "epoch": 31.53846153846154, "grad_norm": 0.15273942053318024, "learning_rate": 0.0002805904832228523, "loss": 0.008, "step": 5330 }, { "epoch": 31.597633136094675, "grad_norm": 0.11745889484882355, "learning_rate": 0.00028051181838975874, "loss": 0.0078, "step": 5340 }, { "epoch": 31.65680473372781, "grad_norm": 0.16922739148139954, "learning_rate": 0.00028043300554173896, "loss": 0.0084, "step": 5350 }, { "epoch": 31.715976331360945, "grad_norm": 0.14080595970153809, "learning_rate": 0.0002803540447681755, "loss": 0.0079, "step": 5360 }, { "epoch": 31.775147928994084, "grad_norm": 0.1561710238456726, "learning_rate": 0.0002802749361586187, "loss": 0.0091, "step": 5370 }, { "epoch": 31.83431952662722, "grad_norm": 0.15054816007614136, "learning_rate": 0.00028019567980278645, "loss": 0.0082, "step": 5380 }, { "epoch": 31.893491124260354, "grad_norm": 0.1631651222705841, "learning_rate": 0.0002801162757905643, "loss": 0.0088, "step": 5390 }, { "epoch": 31.952662721893493, "grad_norm": 0.18153376877307892, "learning_rate": 0.0002800367242120053, "loss": 0.0088, "step": 5400 }, { "epoch": 32.01183431952663, "grad_norm": 0.2477361261844635, "learning_rate": 0.00027995702515732973, "loss": 0.0088, "step": 5410 }, { "epoch": 32.071005917159766, "grad_norm": 0.1771838515996933, "learning_rate": 0.00027987717871692515, "loss": 0.008, "step": 5420 }, { "epoch": 32.1301775147929, "grad_norm": 0.1930997669696808, "learning_rate": 0.00027979718498134646, "loss": 0.0079, "step": 5430 }, { "epoch": 32.18934911242604, "grad_norm": 0.144026517868042, "learning_rate": 0.0002797170440413153, "loss": 0.009, "step": 5440 }, { "epoch": 32.248520710059175, "grad_norm": 0.2305654138326645, "learning_rate": 0.0002796367559877205, "loss": 0.0089, "step": 5450 }, { "epoch": 32.30769230769231, "grad_norm": 0.22362545132637024, "learning_rate": 0.0002795563209116176, "loss": 0.0087, "step": 5460 }, { "epoch": 32.366863905325445, "grad_norm": 0.1584501564502716, "learning_rate": 0.000279475738904229, "loss": 0.0092, "step": 5470 }, { "epoch": 32.42603550295858, "grad_norm": 0.13952863216400146, "learning_rate": 0.0002793950100569436, "loss": 0.0097, "step": 5480 }, { "epoch": 32.485207100591715, "grad_norm": 0.19525058567523956, "learning_rate": 0.000279314134461317, "loss": 0.0092, "step": 5490 }, { "epoch": 32.544378698224854, "grad_norm": 0.15328921377658844, "learning_rate": 0.0002792331122090709, "loss": 0.0087, "step": 5500 }, { "epoch": 32.603550295857985, "grad_norm": 0.1792757511138916, "learning_rate": 0.0002791519433920937, "loss": 0.0093, "step": 5510 }, { "epoch": 32.662721893491124, "grad_norm": 0.1653023660182953, "learning_rate": 0.00027907062810243993, "loss": 0.0087, "step": 5520 }, { "epoch": 32.72189349112426, "grad_norm": 0.1530531495809555, "learning_rate": 0.0002789891664323301, "loss": 0.0082, "step": 5530 }, { "epoch": 32.781065088757394, "grad_norm": 0.10256236046552658, "learning_rate": 0.0002789075584741508, "loss": 0.0078, "step": 5540 }, { "epoch": 32.84023668639053, "grad_norm": 0.1302512139081955, "learning_rate": 0.00027882580432045455, "loss": 0.0077, "step": 5550 }, { "epoch": 32.89940828402367, "grad_norm": 0.17576274275779724, "learning_rate": 0.0002787439040639597, "loss": 0.0081, "step": 5560 }, { "epoch": 32.9585798816568, "grad_norm": 0.14793868362903595, "learning_rate": 0.0002786618577975502, "loss": 0.0086, "step": 5570 }, { "epoch": 33.01775147928994, "grad_norm": 0.19908052682876587, "learning_rate": 0.0002785796656142756, "loss": 0.0082, "step": 5580 }, { "epoch": 33.07692307692308, "grad_norm": 0.13024331629276276, "learning_rate": 0.00027849732760735113, "loss": 0.0074, "step": 5590 }, { "epoch": 33.13609467455621, "grad_norm": 0.14129486680030823, "learning_rate": 0.0002784148438701571, "loss": 0.0079, "step": 5600 }, { "epoch": 33.19526627218935, "grad_norm": 0.20796798169612885, "learning_rate": 0.0002783322144962394, "loss": 0.0091, "step": 5610 }, { "epoch": 33.25443786982248, "grad_norm": 0.19022347033023834, "learning_rate": 0.0002782494395793089, "loss": 0.0075, "step": 5620 }, { "epoch": 33.31360946745562, "grad_norm": 0.13695156574249268, "learning_rate": 0.00027816651921324135, "loss": 0.0076, "step": 5630 }, { "epoch": 33.37278106508876, "grad_norm": 0.18900611996650696, "learning_rate": 0.00027808345349207797, "loss": 0.0082, "step": 5640 }, { "epoch": 33.43195266272189, "grad_norm": 0.16459904611110687, "learning_rate": 0.00027800024251002436, "loss": 0.0073, "step": 5650 }, { "epoch": 33.49112426035503, "grad_norm": 0.1725461483001709, "learning_rate": 0.0002779168863614511, "loss": 0.0088, "step": 5660 }, { "epoch": 33.55029585798817, "grad_norm": 0.19459688663482666, "learning_rate": 0.0002778333851408933, "loss": 0.009, "step": 5670 }, { "epoch": 33.6094674556213, "grad_norm": 0.15084145963191986, "learning_rate": 0.0002777497389430507, "loss": 0.0081, "step": 5680 }, { "epoch": 33.66863905325444, "grad_norm": 0.12242581695318222, "learning_rate": 0.00027766594786278736, "loss": 0.0083, "step": 5690 }, { "epoch": 33.72781065088758, "grad_norm": 0.13216590881347656, "learning_rate": 0.0002775820119951316, "loss": 0.0078, "step": 5700 }, { "epoch": 33.78698224852071, "grad_norm": 0.1540583223104477, "learning_rate": 0.00027749793143527616, "loss": 0.0079, "step": 5710 }, { "epoch": 33.84615384615385, "grad_norm": 0.16777904331684113, "learning_rate": 0.00027741370627857773, "loss": 0.0087, "step": 5720 }, { "epoch": 33.905325443786985, "grad_norm": 0.16339737176895142, "learning_rate": 0.00027732933662055694, "loss": 0.0083, "step": 5730 }, { "epoch": 33.96449704142012, "grad_norm": 0.25951746106147766, "learning_rate": 0.00027724482255689847, "loss": 0.0088, "step": 5740 }, { "epoch": 34.023668639053255, "grad_norm": 0.2580641806125641, "learning_rate": 0.0002771601641834506, "loss": 0.0082, "step": 5750 }, { "epoch": 34.082840236686394, "grad_norm": 0.21895548701286316, "learning_rate": 0.00027707536159622545, "loss": 0.0081, "step": 5760 }, { "epoch": 34.142011834319526, "grad_norm": 0.17190933227539062, "learning_rate": 0.00027699041489139843, "loss": 0.0081, "step": 5770 }, { "epoch": 34.201183431952664, "grad_norm": 0.14771768450737, "learning_rate": 0.0002769053241653086, "loss": 0.0073, "step": 5780 }, { "epoch": 34.260355029585796, "grad_norm": 0.17130407691001892, "learning_rate": 0.00027682008951445845, "loss": 0.0073, "step": 5790 }, { "epoch": 34.319526627218934, "grad_norm": 0.15506872534751892, "learning_rate": 0.00027673471103551345, "loss": 0.0083, "step": 5800 }, { "epoch": 34.37869822485207, "grad_norm": 0.1424255073070526, "learning_rate": 0.00027664918882530225, "loss": 0.0078, "step": 5810 }, { "epoch": 34.437869822485204, "grad_norm": 0.2087360918521881, "learning_rate": 0.00027656352298081665, "loss": 0.0073, "step": 5820 }, { "epoch": 34.49704142011834, "grad_norm": 0.20817337930202484, "learning_rate": 0.0002764777135992112, "loss": 0.0088, "step": 5830 }, { "epoch": 34.55621301775148, "grad_norm": 0.22153311967849731, "learning_rate": 0.0002763917607778033, "loss": 0.0076, "step": 5840 }, { "epoch": 34.61538461538461, "grad_norm": 0.152023583650589, "learning_rate": 0.00027630566461407305, "loss": 0.0082, "step": 5850 }, { "epoch": 34.67455621301775, "grad_norm": 0.1600451022386551, "learning_rate": 0.000276219425205663, "loss": 0.0068, "step": 5860 }, { "epoch": 34.73372781065089, "grad_norm": 0.16682718694210052, "learning_rate": 0.0002761330426503783, "loss": 0.0072, "step": 5870 }, { "epoch": 34.79289940828402, "grad_norm": 0.11200512945652008, "learning_rate": 0.00027604651704618636, "loss": 0.0075, "step": 5880 }, { "epoch": 34.85207100591716, "grad_norm": 0.24536605179309845, "learning_rate": 0.0002759598484912169, "loss": 0.0082, "step": 5890 }, { "epoch": 34.9112426035503, "grad_norm": 0.15980571508407593, "learning_rate": 0.00027587303708376156, "loss": 0.0082, "step": 5900 }, { "epoch": 34.97041420118343, "grad_norm": 0.1491709053516388, "learning_rate": 0.00027578608292227433, "loss": 0.0091, "step": 5910 }, { "epoch": 35.02958579881657, "grad_norm": 0.18626244366168976, "learning_rate": 0.0002756989861053708, "loss": 0.0089, "step": 5920 }, { "epoch": 35.0887573964497, "grad_norm": 0.2415395975112915, "learning_rate": 0.00027561174673182843, "loss": 0.0081, "step": 5930 }, { "epoch": 35.14792899408284, "grad_norm": 0.14779822528362274, "learning_rate": 0.0002755243649005864, "loss": 0.008, "step": 5940 }, { "epoch": 35.20710059171598, "grad_norm": 0.2152884602546692, "learning_rate": 0.00027543684071074543, "loss": 0.0075, "step": 5950 }, { "epoch": 35.26627218934911, "grad_norm": 0.13887131214141846, "learning_rate": 0.0002753491742615677, "loss": 0.0089, "step": 5960 }, { "epoch": 35.32544378698225, "grad_norm": 0.21310639381408691, "learning_rate": 0.0002752613656524768, "loss": 0.0079, "step": 5970 }, { "epoch": 35.38461538461539, "grad_norm": 0.12175067514181137, "learning_rate": 0.00027517341498305733, "loss": 0.0081, "step": 5980 }, { "epoch": 35.44378698224852, "grad_norm": 0.24045783281326294, "learning_rate": 0.00027508532235305516, "loss": 0.0073, "step": 5990 }, { "epoch": 35.50295857988166, "grad_norm": 0.1525317281484604, "learning_rate": 0.00027499708786237723, "loss": 0.0078, "step": 6000 }, { "epoch": 35.562130177514796, "grad_norm": 0.17472219467163086, "learning_rate": 0.00027490871161109116, "loss": 0.0076, "step": 6010 }, { "epoch": 35.62130177514793, "grad_norm": 0.10962396115064621, "learning_rate": 0.00027482019369942553, "loss": 0.0079, "step": 6020 }, { "epoch": 35.680473372781066, "grad_norm": 0.12683483958244324, "learning_rate": 0.0002747315342277695, "loss": 0.0083, "step": 6030 }, { "epoch": 35.739644970414204, "grad_norm": 0.17404355108737946, "learning_rate": 0.00027464273329667275, "loss": 0.0075, "step": 6040 }, { "epoch": 35.798816568047336, "grad_norm": 0.2012220174074173, "learning_rate": 0.00027455379100684534, "loss": 0.0071, "step": 6050 }, { "epoch": 35.857988165680474, "grad_norm": 0.21382932364940643, "learning_rate": 0.0002744647074591579, "loss": 0.0074, "step": 6060 }, { "epoch": 35.917159763313606, "grad_norm": 0.24250242114067078, "learning_rate": 0.000274375482754641, "loss": 0.0084, "step": 6070 }, { "epoch": 35.976331360946745, "grad_norm": 0.15444986522197723, "learning_rate": 0.00027428611699448533, "loss": 0.0077, "step": 6080 }, { "epoch": 36.03550295857988, "grad_norm": 0.15509358048439026, "learning_rate": 0.0002741966102800417, "loss": 0.0085, "step": 6090 }, { "epoch": 36.094674556213015, "grad_norm": 0.19573189318180084, "learning_rate": 0.0002741069627128206, "loss": 0.0072, "step": 6100 }, { "epoch": 36.15384615384615, "grad_norm": 0.14883369207382202, "learning_rate": 0.0002740171743944924, "loss": 0.007, "step": 6110 }, { "epoch": 36.21301775147929, "grad_norm": 0.132828027009964, "learning_rate": 0.000273927245426887, "loss": 0.0076, "step": 6120 }, { "epoch": 36.27218934911242, "grad_norm": 0.10226385295391083, "learning_rate": 0.00027383717591199393, "loss": 0.0084, "step": 6130 }, { "epoch": 36.33136094674556, "grad_norm": 0.18194487690925598, "learning_rate": 0.00027374696595196203, "loss": 0.008, "step": 6140 }, { "epoch": 36.3905325443787, "grad_norm": 0.12073666602373123, "learning_rate": 0.0002736566156490994, "loss": 0.0081, "step": 6150 }, { "epoch": 36.44970414201183, "grad_norm": 0.1514735370874405, "learning_rate": 0.00027356612510587335, "loss": 0.0074, "step": 6160 }, { "epoch": 36.50887573964497, "grad_norm": 0.07807518541812897, "learning_rate": 0.00027347549442491016, "loss": 0.0076, "step": 6170 }, { "epoch": 36.56804733727811, "grad_norm": 0.15961013734340668, "learning_rate": 0.00027338472370899523, "loss": 0.0068, "step": 6180 }, { "epoch": 36.62721893491124, "grad_norm": 0.14148485660552979, "learning_rate": 0.0002732938130610726, "loss": 0.0071, "step": 6190 }, { "epoch": 36.68639053254438, "grad_norm": 0.14862705767154694, "learning_rate": 0.00027320276258424505, "loss": 0.0067, "step": 6200 }, { "epoch": 36.74556213017752, "grad_norm": 0.18952898681163788, "learning_rate": 0.0002731115723817739, "loss": 0.0072, "step": 6210 }, { "epoch": 36.80473372781065, "grad_norm": 0.11368771642446518, "learning_rate": 0.0002730202425570791, "loss": 0.0068, "step": 6220 }, { "epoch": 36.86390532544379, "grad_norm": 0.15045009553432465, "learning_rate": 0.0002729287732137388, "loss": 0.0069, "step": 6230 }, { "epoch": 36.92307692307692, "grad_norm": 0.14315468072891235, "learning_rate": 0.0002728371644554893, "loss": 0.0069, "step": 6240 }, { "epoch": 36.98224852071006, "grad_norm": 0.13757099211215973, "learning_rate": 0.0002727454163862253, "loss": 0.0077, "step": 6250 }, { "epoch": 37.0414201183432, "grad_norm": 0.14400826394557953, "learning_rate": 0.0002726535291099993, "loss": 0.0078, "step": 6260 }, { "epoch": 37.10059171597633, "grad_norm": 0.17688797414302826, "learning_rate": 0.0002725615027310216, "loss": 0.0072, "step": 6270 }, { "epoch": 37.15976331360947, "grad_norm": 0.14576929807662964, "learning_rate": 0.00027246933735366037, "loss": 0.0071, "step": 6280 }, { "epoch": 37.218934911242606, "grad_norm": 0.1656545251607895, "learning_rate": 0.0002723770330824414, "loss": 0.007, "step": 6290 }, { "epoch": 37.27810650887574, "grad_norm": 0.09736321866512299, "learning_rate": 0.00027228459002204806, "loss": 0.0074, "step": 6300 }, { "epoch": 37.337278106508876, "grad_norm": 0.0985318273305893, "learning_rate": 0.0002721920082773211, "loss": 0.0078, "step": 6310 }, { "epoch": 37.396449704142015, "grad_norm": 0.1677810102701187, "learning_rate": 0.0002720992879532584, "loss": 0.0073, "step": 6320 }, { "epoch": 37.455621301775146, "grad_norm": 0.18557986617088318, "learning_rate": 0.0002720064291550152, "loss": 0.0069, "step": 6330 }, { "epoch": 37.514792899408285, "grad_norm": 0.16365817189216614, "learning_rate": 0.00027191343198790377, "loss": 0.0069, "step": 6340 }, { "epoch": 37.57396449704142, "grad_norm": 0.13991214334964752, "learning_rate": 0.0002718202965573931, "loss": 0.0083, "step": 6350 }, { "epoch": 37.633136094674555, "grad_norm": 0.09990856796503067, "learning_rate": 0.0002717270229691093, "loss": 0.0072, "step": 6360 }, { "epoch": 37.69230769230769, "grad_norm": 0.16608808934688568, "learning_rate": 0.00027163361132883485, "loss": 0.007, "step": 6370 }, { "epoch": 37.751479289940825, "grad_norm": 0.1694093495607376, "learning_rate": 0.000271540061742509, "loss": 0.0074, "step": 6380 }, { "epoch": 37.81065088757396, "grad_norm": 0.17610028386116028, "learning_rate": 0.00027144637431622743, "loss": 0.0075, "step": 6390 }, { "epoch": 37.8698224852071, "grad_norm": 0.14397823810577393, "learning_rate": 0.0002713525491562421, "loss": 0.0069, "step": 6400 }, { "epoch": 37.928994082840234, "grad_norm": 0.14364123344421387, "learning_rate": 0.0002712585863689611, "loss": 0.0077, "step": 6410 }, { "epoch": 37.98816568047337, "grad_norm": 0.24272942543029785, "learning_rate": 0.0002711644860609488, "loss": 0.0092, "step": 6420 }, { "epoch": 38.04733727810651, "grad_norm": 0.16437174379825592, "learning_rate": 0.00027107024833892537, "loss": 0.0079, "step": 6430 }, { "epoch": 38.10650887573964, "grad_norm": 0.16555601358413696, "learning_rate": 0.00027097587330976686, "loss": 0.0071, "step": 6440 }, { "epoch": 38.16568047337278, "grad_norm": 0.1324462890625, "learning_rate": 0.000270881361080505, "loss": 0.0074, "step": 6450 }, { "epoch": 38.22485207100592, "grad_norm": 0.160003200173378, "learning_rate": 0.0002707867117583273, "loss": 0.0073, "step": 6460 }, { "epoch": 38.28402366863905, "grad_norm": 0.18849870562553406, "learning_rate": 0.0002706919254505766, "loss": 0.0087, "step": 6470 }, { "epoch": 38.34319526627219, "grad_norm": 0.14654429256916046, "learning_rate": 0.0002705970022647511, "loss": 0.0069, "step": 6480 }, { "epoch": 38.40236686390533, "grad_norm": 0.15086832642555237, "learning_rate": 0.0002705019423085042, "loss": 0.0071, "step": 6490 }, { "epoch": 38.46153846153846, "grad_norm": 0.12601831555366516, "learning_rate": 0.0002704067456896445, "loss": 0.0068, "step": 6500 }, { "epoch": 38.5207100591716, "grad_norm": 0.19676926732063293, "learning_rate": 0.00027031141251613564, "loss": 0.0074, "step": 6510 }, { "epoch": 38.57988165680474, "grad_norm": 0.18023961782455444, "learning_rate": 0.00027021594289609597, "loss": 0.0075, "step": 6520 }, { "epoch": 38.63905325443787, "grad_norm": 0.13811413943767548, "learning_rate": 0.0002701203369377986, "loss": 0.0075, "step": 6530 }, { "epoch": 38.69822485207101, "grad_norm": 0.16718992590904236, "learning_rate": 0.0002700245947496715, "loss": 0.0094, "step": 6540 }, { "epoch": 38.75739644970414, "grad_norm": 0.13854017853736877, "learning_rate": 0.00026992871644029685, "loss": 0.008, "step": 6550 }, { "epoch": 38.81656804733728, "grad_norm": 0.15033909678459167, "learning_rate": 0.00026983270211841133, "loss": 0.0071, "step": 6560 }, { "epoch": 38.875739644970416, "grad_norm": 0.3030054569244385, "learning_rate": 0.00026973655189290585, "loss": 0.0073, "step": 6570 }, { "epoch": 38.93491124260355, "grad_norm": 0.182002454996109, "learning_rate": 0.0002696402658728255, "loss": 0.0071, "step": 6580 }, { "epoch": 38.994082840236686, "grad_norm": 0.4055502712726593, "learning_rate": 0.0002695438441673694, "loss": 0.0082, "step": 6590 }, { "epoch": 39.053254437869825, "grad_norm": 0.17021793127059937, "learning_rate": 0.0002694472868858904, "loss": 0.0084, "step": 6600 }, { "epoch": 39.112426035502956, "grad_norm": 0.24419549107551575, "learning_rate": 0.0002693505941378952, "loss": 0.0078, "step": 6610 }, { "epoch": 39.171597633136095, "grad_norm": 0.20646415650844574, "learning_rate": 0.00026925376603304424, "loss": 0.0083, "step": 6620 }, { "epoch": 39.23076923076923, "grad_norm": 0.16070152819156647, "learning_rate": 0.00026915680268115125, "loss": 0.0071, "step": 6630 }, { "epoch": 39.289940828402365, "grad_norm": 0.2781049907207489, "learning_rate": 0.0002690597041921835, "loss": 0.0074, "step": 6640 }, { "epoch": 39.349112426035504, "grad_norm": 0.17088855803012848, "learning_rate": 0.0002689624706762615, "loss": 0.0083, "step": 6650 }, { "epoch": 39.40828402366864, "grad_norm": 0.15961846709251404, "learning_rate": 0.0002688651022436589, "loss": 0.007, "step": 6660 }, { "epoch": 39.467455621301774, "grad_norm": 0.1655288189649582, "learning_rate": 0.00026876759900480225, "loss": 0.0086, "step": 6670 }, { "epoch": 39.52662721893491, "grad_norm": 0.1559499353170395, "learning_rate": 0.0002686699610702712, "loss": 0.0074, "step": 6680 }, { "epoch": 39.585798816568044, "grad_norm": 0.15787792205810547, "learning_rate": 0.0002685721885507979, "loss": 0.0072, "step": 6690 }, { "epoch": 39.64497041420118, "grad_norm": 0.16029103100299835, "learning_rate": 0.0002684742815572674, "loss": 0.0075, "step": 6700 }, { "epoch": 39.70414201183432, "grad_norm": 0.145809143781662, "learning_rate": 0.00026837624020071703, "loss": 0.0069, "step": 6710 }, { "epoch": 39.76331360946745, "grad_norm": 0.1692129671573639, "learning_rate": 0.00026827806459233663, "loss": 0.0083, "step": 6720 }, { "epoch": 39.82248520710059, "grad_norm": 0.21021364629268646, "learning_rate": 0.00026817975484346823, "loss": 0.0069, "step": 6730 }, { "epoch": 39.88165680473373, "grad_norm": 0.2297717034816742, "learning_rate": 0.0002680813110656061, "loss": 0.0073, "step": 6740 }, { "epoch": 39.94082840236686, "grad_norm": 0.2134036123752594, "learning_rate": 0.00026798273337039636, "loss": 0.0077, "step": 6750 }, { "epoch": 40.0, "grad_norm": 0.1276015341281891, "learning_rate": 0.00026788402186963715, "loss": 0.0072, "step": 6760 }, { "epoch": 40.05917159763314, "grad_norm": 0.20515044033527374, "learning_rate": 0.00026778517667527823, "loss": 0.0088, "step": 6770 }, { "epoch": 40.11834319526627, "grad_norm": 0.24117036163806915, "learning_rate": 0.0002676861978994212, "loss": 0.0076, "step": 6780 }, { "epoch": 40.17751479289941, "grad_norm": 0.20714038610458374, "learning_rate": 0.00026758708565431883, "loss": 0.0076, "step": 6790 }, { "epoch": 40.23668639053255, "grad_norm": 0.22502024471759796, "learning_rate": 0.0002674878400523755, "loss": 0.0072, "step": 6800 }, { "epoch": 40.29585798816568, "grad_norm": 0.11787409335374832, "learning_rate": 0.00026738846120614676, "loss": 0.0077, "step": 6810 }, { "epoch": 40.35502958579882, "grad_norm": 0.2713890075683594, "learning_rate": 0.0002672889492283393, "loss": 0.0078, "step": 6820 }, { "epoch": 40.414201183431956, "grad_norm": 0.13489395380020142, "learning_rate": 0.0002671893042318108, "loss": 0.0066, "step": 6830 }, { "epoch": 40.47337278106509, "grad_norm": 0.1438378095626831, "learning_rate": 0.00026708952632956975, "loss": 0.0072, "step": 6840 }, { "epoch": 40.532544378698226, "grad_norm": 0.1832195520401001, "learning_rate": 0.0002669896156347754, "loss": 0.0059, "step": 6850 }, { "epoch": 40.59171597633136, "grad_norm": 0.12184067070484161, "learning_rate": 0.0002668895722607376, "loss": 0.0068, "step": 6860 }, { "epoch": 40.650887573964496, "grad_norm": 0.11286898702383041, "learning_rate": 0.0002667893963209166, "loss": 0.0065, "step": 6870 }, { "epoch": 40.710059171597635, "grad_norm": 0.15613873302936554, "learning_rate": 0.0002666890879289231, "loss": 0.0078, "step": 6880 }, { "epoch": 40.76923076923077, "grad_norm": 0.11430808901786804, "learning_rate": 0.00026658864719851803, "loss": 0.0064, "step": 6890 }, { "epoch": 40.828402366863905, "grad_norm": 0.1637515276670456, "learning_rate": 0.0002664880742436124, "loss": 0.0072, "step": 6900 }, { "epoch": 40.887573964497044, "grad_norm": 0.147795632481575, "learning_rate": 0.000266387369178267, "loss": 0.0072, "step": 6910 }, { "epoch": 40.946745562130175, "grad_norm": 0.14483684301376343, "learning_rate": 0.00026628653211669263, "loss": 0.0072, "step": 6920 }, { "epoch": 41.005917159763314, "grad_norm": 0.14209692180156708, "learning_rate": 0.0002661855631732498, "loss": 0.0079, "step": 6930 }, { "epoch": 41.06508875739645, "grad_norm": 0.09784065932035446, "learning_rate": 0.0002660844624624484, "loss": 0.0079, "step": 6940 }, { "epoch": 41.124260355029584, "grad_norm": 0.1715306043624878, "learning_rate": 0.00026598323009894805, "loss": 0.0076, "step": 6950 }, { "epoch": 41.18343195266272, "grad_norm": 0.16671815514564514, "learning_rate": 0.00026588186619755735, "loss": 0.0069, "step": 6960 }, { "epoch": 41.24260355029586, "grad_norm": 0.10903116315603256, "learning_rate": 0.0002657803708732344, "loss": 0.0071, "step": 6970 }, { "epoch": 41.30177514792899, "grad_norm": 0.12225215137004852, "learning_rate": 0.0002656787442410861, "loss": 0.0062, "step": 6980 }, { "epoch": 41.36094674556213, "grad_norm": 0.1238550990819931, "learning_rate": 0.00026557698641636835, "loss": 0.0065, "step": 6990 }, { "epoch": 41.42011834319526, "grad_norm": 0.2194967269897461, "learning_rate": 0.0002654750975144859, "loss": 0.0067, "step": 7000 }, { "epoch": 41.4792899408284, "grad_norm": 0.11007628589868546, "learning_rate": 0.0002653730776509921, "loss": 0.0069, "step": 7010 }, { "epoch": 41.53846153846154, "grad_norm": 0.15134446322917938, "learning_rate": 0.00026527092694158877, "loss": 0.0068, "step": 7020 }, { "epoch": 41.59763313609467, "grad_norm": 0.12832187116146088, "learning_rate": 0.0002651686455021263, "loss": 0.0072, "step": 7030 }, { "epoch": 41.65680473372781, "grad_norm": 0.1664343923330307, "learning_rate": 0.00026506623344860306, "loss": 0.0085, "step": 7040 }, { "epoch": 41.71597633136095, "grad_norm": 0.209603413939476, "learning_rate": 0.0002649636908971658, "loss": 0.0076, "step": 7050 }, { "epoch": 41.77514792899408, "grad_norm": 0.18023131787776947, "learning_rate": 0.0002648610179641093, "loss": 0.0071, "step": 7060 }, { "epoch": 41.83431952662722, "grad_norm": 0.16284997761249542, "learning_rate": 0.00026475821476587585, "loss": 0.0078, "step": 7070 }, { "epoch": 41.89349112426036, "grad_norm": 0.14467649161815643, "learning_rate": 0.00026465528141905595, "loss": 0.0072, "step": 7080 }, { "epoch": 41.95266272189349, "grad_norm": 0.1876014918088913, "learning_rate": 0.0002645522180403873, "loss": 0.0075, "step": 7090 }, { "epoch": 42.01183431952663, "grad_norm": 0.17539453506469727, "learning_rate": 0.00026444902474675536, "loss": 0.0072, "step": 7100 }, { "epoch": 42.071005917159766, "grad_norm": 0.16788706183433533, "learning_rate": 0.0002643457016551927, "loss": 0.0075, "step": 7110 }, { "epoch": 42.1301775147929, "grad_norm": 0.186580628156662, "learning_rate": 0.0002642422488828793, "loss": 0.0074, "step": 7120 }, { "epoch": 42.18934911242604, "grad_norm": 0.141093447804451, "learning_rate": 0.00026413866654714205, "loss": 0.0063, "step": 7130 }, { "epoch": 42.248520710059175, "grad_norm": 0.10165922343730927, "learning_rate": 0.0002640349547654549, "loss": 0.0071, "step": 7140 }, { "epoch": 42.30769230769231, "grad_norm": 0.15089169144630432, "learning_rate": 0.00026393111365543854, "loss": 0.0067, "step": 7150 }, { "epoch": 42.366863905325445, "grad_norm": 0.11481019854545593, "learning_rate": 0.0002638271433348603, "loss": 0.007, "step": 7160 }, { "epoch": 42.42603550295858, "grad_norm": 0.16681741178035736, "learning_rate": 0.0002637230439216341, "loss": 0.007, "step": 7170 }, { "epoch": 42.485207100591715, "grad_norm": 0.08017527312040329, "learning_rate": 0.00026361881553382035, "loss": 0.0066, "step": 7180 }, { "epoch": 42.544378698224854, "grad_norm": 0.13770990073680878, "learning_rate": 0.00026351445828962555, "loss": 0.0062, "step": 7190 }, { "epoch": 42.603550295857985, "grad_norm": 0.12251169979572296, "learning_rate": 0.00026340997230740244, "loss": 0.0055, "step": 7200 }, { "epoch": 42.662721893491124, "grad_norm": 0.17005927860736847, "learning_rate": 0.0002633053577056498, "loss": 0.006, "step": 7210 }, { "epoch": 42.72189349112426, "grad_norm": 0.09639030694961548, "learning_rate": 0.0002632006146030122, "loss": 0.0053, "step": 7220 }, { "epoch": 42.781065088757394, "grad_norm": 0.10558243840932846, "learning_rate": 0.00026309574311827994, "loss": 0.0067, "step": 7230 }, { "epoch": 42.84023668639053, "grad_norm": 0.12186397612094879, "learning_rate": 0.000262990743370389, "loss": 0.0063, "step": 7240 }, { "epoch": 42.89940828402367, "grad_norm": 0.10730966180562973, "learning_rate": 0.00026288561547842075, "loss": 0.0071, "step": 7250 }, { "epoch": 42.9585798816568, "grad_norm": 0.14032189548015594, "learning_rate": 0.00026278035956160196, "loss": 0.0066, "step": 7260 }, { "epoch": 43.01775147928994, "grad_norm": 0.11599422991275787, "learning_rate": 0.0002626749757393045, "loss": 0.0069, "step": 7270 }, { "epoch": 43.07692307692308, "grad_norm": 0.1487927883863449, "learning_rate": 0.0002625694641310454, "loss": 0.0062, "step": 7280 }, { "epoch": 43.13609467455621, "grad_norm": 0.12871229648590088, "learning_rate": 0.0002624638248564865, "loss": 0.0063, "step": 7290 }, { "epoch": 43.19526627218935, "grad_norm": 0.1397814154624939, "learning_rate": 0.00026235805803543466, "loss": 0.0052, "step": 7300 }, { "epoch": 43.25443786982248, "grad_norm": 0.17789584398269653, "learning_rate": 0.00026225216378784097, "loss": 0.0064, "step": 7310 }, { "epoch": 43.31360946745562, "grad_norm": 0.1381426602602005, "learning_rate": 0.00026214614223380144, "loss": 0.0069, "step": 7320 }, { "epoch": 43.37278106508876, "grad_norm": 0.1102122887969017, "learning_rate": 0.0002620399934935564, "loss": 0.0059, "step": 7330 }, { "epoch": 43.43195266272189, "grad_norm": 0.14397363364696503, "learning_rate": 0.00026193371768749017, "loss": 0.0066, "step": 7340 }, { "epoch": 43.49112426035503, "grad_norm": 0.155499666929245, "learning_rate": 0.0002618273149361314, "loss": 0.0064, "step": 7350 }, { "epoch": 43.55029585798817, "grad_norm": 0.12074228376150131, "learning_rate": 0.0002617207853601526, "loss": 0.0067, "step": 7360 }, { "epoch": 43.6094674556213, "grad_norm": 0.17261241376399994, "learning_rate": 0.00026161412908037027, "loss": 0.0071, "step": 7370 }, { "epoch": 43.66863905325444, "grad_norm": 0.09106358140707016, "learning_rate": 0.0002615073462177445, "loss": 0.0078, "step": 7380 }, { "epoch": 43.72781065088758, "grad_norm": 0.14331908524036407, "learning_rate": 0.0002614004368933788, "loss": 0.0062, "step": 7390 }, { "epoch": 43.78698224852071, "grad_norm": 0.1712951958179474, "learning_rate": 0.00026129340122852036, "loss": 0.0073, "step": 7400 }, { "epoch": 43.84615384615385, "grad_norm": 0.15222331881523132, "learning_rate": 0.00026118623934455953, "loss": 0.0061, "step": 7410 }, { "epoch": 43.905325443786985, "grad_norm": 0.1366274058818817, "learning_rate": 0.00026107895136302973, "loss": 0.0065, "step": 7420 }, { "epoch": 43.96449704142012, "grad_norm": 0.15441443026065826, "learning_rate": 0.0002609715374056076, "loss": 0.0077, "step": 7430 }, { "epoch": 44.023668639053255, "grad_norm": 0.18454864621162415, "learning_rate": 0.0002608639975941124, "loss": 0.0073, "step": 7440 }, { "epoch": 44.082840236686394, "grad_norm": 0.14270299673080444, "learning_rate": 0.0002607563320505063, "loss": 0.0064, "step": 7450 }, { "epoch": 44.142011834319526, "grad_norm": 0.18707430362701416, "learning_rate": 0.00026064854089689405, "loss": 0.0066, "step": 7460 }, { "epoch": 44.201183431952664, "grad_norm": 0.21995829045772552, "learning_rate": 0.0002605406242555227, "loss": 0.007, "step": 7470 }, { "epoch": 44.260355029585796, "grad_norm": 0.16009274125099182, "learning_rate": 0.0002604325822487818, "loss": 0.0076, "step": 7480 }, { "epoch": 44.319526627218934, "grad_norm": 0.150262713432312, "learning_rate": 0.00026032441499920306, "loss": 0.006, "step": 7490 }, { "epoch": 44.37869822485207, "grad_norm": 0.16023032367229462, "learning_rate": 0.00026021612262946004, "loss": 0.0067, "step": 7500 }, { "epoch": 44.437869822485204, "grad_norm": 0.16475118696689606, "learning_rate": 0.0002601077052623685, "loss": 0.0066, "step": 7510 }, { "epoch": 44.49704142011834, "grad_norm": 0.21810655295848846, "learning_rate": 0.0002599991630208857, "loss": 0.0076, "step": 7520 }, { "epoch": 44.55621301775148, "grad_norm": 0.19164445996284485, "learning_rate": 0.00025989049602811066, "loss": 0.0071, "step": 7530 }, { "epoch": 44.61538461538461, "grad_norm": 0.13087108731269836, "learning_rate": 0.0002597817044072838, "loss": 0.0063, "step": 7540 }, { "epoch": 44.67455621301775, "grad_norm": 0.12378490716218948, "learning_rate": 0.0002596727882817869, "loss": 0.0068, "step": 7550 }, { "epoch": 44.73372781065089, "grad_norm": 0.14039717614650726, "learning_rate": 0.0002595637477751431, "loss": 0.0074, "step": 7560 }, { "epoch": 44.79289940828402, "grad_norm": 0.15705673396587372, "learning_rate": 0.00025945458301101636, "loss": 0.0059, "step": 7570 }, { "epoch": 44.85207100591716, "grad_norm": 0.17123958468437195, "learning_rate": 0.0002593452941132117, "loss": 0.0067, "step": 7580 }, { "epoch": 44.9112426035503, "grad_norm": 0.18548083305358887, "learning_rate": 0.0002592358812056749, "loss": 0.0067, "step": 7590 }, { "epoch": 44.97041420118343, "grad_norm": 0.14425033330917358, "learning_rate": 0.0002591263444124923, "loss": 0.007, "step": 7600 }, { "epoch": 45.02958579881657, "grad_norm": 0.15528208017349243, "learning_rate": 0.000259016683857891, "loss": 0.0066, "step": 7610 }, { "epoch": 45.0887573964497, "grad_norm": 0.10904689133167267, "learning_rate": 0.00025890689966623804, "loss": 0.0061, "step": 7620 }, { "epoch": 45.14792899408284, "grad_norm": 0.14506186544895172, "learning_rate": 0.0002587969919620411, "loss": 0.0068, "step": 7630 }, { "epoch": 45.20710059171598, "grad_norm": 0.11365960538387299, "learning_rate": 0.0002586869608699476, "loss": 0.006, "step": 7640 }, { "epoch": 45.26627218934911, "grad_norm": 0.19490359723567963, "learning_rate": 0.0002585768065147452, "loss": 0.0066, "step": 7650 }, { "epoch": 45.32544378698225, "grad_norm": 0.1425994485616684, "learning_rate": 0.00025846652902136106, "loss": 0.0072, "step": 7660 }, { "epoch": 45.38461538461539, "grad_norm": 0.1438591331243515, "learning_rate": 0.00025835612851486217, "loss": 0.0071, "step": 7670 }, { "epoch": 45.44378698224852, "grad_norm": 0.11776860058307648, "learning_rate": 0.000258245605120455, "loss": 0.0069, "step": 7680 }, { "epoch": 45.50295857988166, "grad_norm": 0.11806171387434006, "learning_rate": 0.00025813495896348537, "loss": 0.0072, "step": 7690 }, { "epoch": 45.562130177514796, "grad_norm": 0.1420697271823883, "learning_rate": 0.00025802419016943834, "loss": 0.0069, "step": 7700 }, { "epoch": 45.62130177514793, "grad_norm": 0.10232560336589813, "learning_rate": 0.0002579132988639381, "loss": 0.0066, "step": 7710 }, { "epoch": 45.680473372781066, "grad_norm": 0.25152677297592163, "learning_rate": 0.0002578022851727477, "loss": 0.0068, "step": 7720 }, { "epoch": 45.739644970414204, "grad_norm": 0.1349257081747055, "learning_rate": 0.00025769114922176894, "loss": 0.0065, "step": 7730 }, { "epoch": 45.798816568047336, "grad_norm": 0.1278434544801712, "learning_rate": 0.0002575798911370425, "loss": 0.0066, "step": 7740 }, { "epoch": 45.857988165680474, "grad_norm": 0.12062302231788635, "learning_rate": 0.0002574685110447472, "loss": 0.0061, "step": 7750 }, { "epoch": 45.917159763313606, "grad_norm": 0.15775305032730103, "learning_rate": 0.00025735700907120076, "loss": 0.0067, "step": 7760 }, { "epoch": 45.976331360946745, "grad_norm": 0.11925400793552399, "learning_rate": 0.0002572453853428586, "loss": 0.006, "step": 7770 }, { "epoch": 46.03550295857988, "grad_norm": 0.13718797266483307, "learning_rate": 0.0002571336399863146, "loss": 0.0064, "step": 7780 }, { "epoch": 46.094674556213015, "grad_norm": 0.11600884050130844, "learning_rate": 0.0002570217731283003, "loss": 0.0062, "step": 7790 }, { "epoch": 46.15384615384615, "grad_norm": 0.159467414021492, "learning_rate": 0.0002569097848956852, "loss": 0.0069, "step": 7800 }, { "epoch": 46.21301775147929, "grad_norm": 0.18556825816631317, "learning_rate": 0.00025679767541547655, "loss": 0.0072, "step": 7810 }, { "epoch": 46.27218934911242, "grad_norm": 0.10627380013465881, "learning_rate": 0.00025668544481481875, "loss": 0.0065, "step": 7820 }, { "epoch": 46.33136094674556, "grad_norm": 0.15030601620674133, "learning_rate": 0.000256573093220994, "loss": 0.0064, "step": 7830 }, { "epoch": 46.3905325443787, "grad_norm": 0.18055811524391174, "learning_rate": 0.0002564606207614214, "loss": 0.0066, "step": 7840 }, { "epoch": 46.44970414201183, "grad_norm": 0.18057303130626678, "learning_rate": 0.0002563480275636572, "loss": 0.0059, "step": 7850 }, { "epoch": 46.50887573964497, "grad_norm": 0.14212895929813385, "learning_rate": 0.0002562353137553947, "loss": 0.0058, "step": 7860 }, { "epoch": 46.56804733727811, "grad_norm": 0.21400298178195953, "learning_rate": 0.0002561224794644639, "loss": 0.0064, "step": 7870 }, { "epoch": 46.62721893491124, "grad_norm": 0.1850673258304596, "learning_rate": 0.00025600952481883143, "loss": 0.0066, "step": 7880 }, { "epoch": 46.68639053254438, "grad_norm": 0.1543768048286438, "learning_rate": 0.00025589644994660035, "loss": 0.006, "step": 7890 }, { "epoch": 46.74556213017752, "grad_norm": 0.1423298418521881, "learning_rate": 0.0002557832549760103, "loss": 0.0061, "step": 7900 }, { "epoch": 46.80473372781065, "grad_norm": 0.21422745287418365, "learning_rate": 0.00025566994003543684, "loss": 0.0076, "step": 7910 }, { "epoch": 46.86390532544379, "grad_norm": 0.16427884995937347, "learning_rate": 0.00025555650525339174, "loss": 0.0074, "step": 7920 }, { "epoch": 46.92307692307692, "grad_norm": 0.23799477517604828, "learning_rate": 0.00025544295075852273, "loss": 0.0072, "step": 7930 }, { "epoch": 46.98224852071006, "grad_norm": 0.13945484161376953, "learning_rate": 0.0002553292766796132, "loss": 0.007, "step": 7940 }, { "epoch": 47.0414201183432, "grad_norm": 0.17251916229724884, "learning_rate": 0.00025521548314558223, "loss": 0.0069, "step": 7950 }, { "epoch": 47.10059171597633, "grad_norm": 0.1697927862405777, "learning_rate": 0.00025510157028548435, "loss": 0.0069, "step": 7960 }, { "epoch": 47.15976331360947, "grad_norm": 0.14655089378356934, "learning_rate": 0.0002549875382285094, "loss": 0.0065, "step": 7970 }, { "epoch": 47.218934911242606, "grad_norm": 0.1486804336309433, "learning_rate": 0.00025487338710398247, "loss": 0.0063, "step": 7980 }, { "epoch": 47.27810650887574, "grad_norm": 0.17284373939037323, "learning_rate": 0.00025475911704136354, "loss": 0.0067, "step": 7990 }, { "epoch": 47.337278106508876, "grad_norm": 0.13078799843788147, "learning_rate": 0.00025464472817024773, "loss": 0.0066, "step": 8000 }, { "epoch": 47.396449704142015, "grad_norm": 0.15447209775447845, "learning_rate": 0.0002545302206203646, "loss": 0.006, "step": 8010 }, { "epoch": 47.455621301775146, "grad_norm": 0.09062041342258453, "learning_rate": 0.0002544155945215785, "loss": 0.0064, "step": 8020 }, { "epoch": 47.514792899408285, "grad_norm": 0.1300264596939087, "learning_rate": 0.00025430085000388813, "loss": 0.0059, "step": 8030 }, { "epoch": 47.57396449704142, "grad_norm": 0.11742731928825378, "learning_rate": 0.00025418598719742676, "loss": 0.0062, "step": 8040 }, { "epoch": 47.633136094674555, "grad_norm": 0.11943674832582474, "learning_rate": 0.00025407100623246135, "loss": 0.0068, "step": 8050 }, { "epoch": 47.69230769230769, "grad_norm": 0.10319584608078003, "learning_rate": 0.0002539559072393932, "loss": 0.0063, "step": 8060 }, { "epoch": 47.751479289940825, "grad_norm": 0.12383740395307541, "learning_rate": 0.0002538406903487574, "loss": 0.0064, "step": 8070 }, { "epoch": 47.81065088757396, "grad_norm": 0.15425772964954376, "learning_rate": 0.00025372535569122273, "loss": 0.0064, "step": 8080 }, { "epoch": 47.8698224852071, "grad_norm": 0.10986463725566864, "learning_rate": 0.0002536099033975915, "loss": 0.0061, "step": 8090 }, { "epoch": 47.928994082840234, "grad_norm": 0.12153414636850357, "learning_rate": 0.0002534943335987994, "loss": 0.0059, "step": 8100 }, { "epoch": 47.98816568047337, "grad_norm": 0.14560841023921967, "learning_rate": 0.0002533786464259156, "loss": 0.0066, "step": 8110 }, { "epoch": 48.04733727810651, "grad_norm": 0.10993405431509018, "learning_rate": 0.00025326284201014203, "loss": 0.0067, "step": 8120 }, { "epoch": 48.10650887573964, "grad_norm": 0.1460290253162384, "learning_rate": 0.00025314692048281393, "loss": 0.0069, "step": 8130 }, { "epoch": 48.16568047337278, "grad_norm": 0.10955538600683212, "learning_rate": 0.0002530308819753992, "loss": 0.006, "step": 8140 }, { "epoch": 48.22485207100592, "grad_norm": 0.14039412140846252, "learning_rate": 0.0002529147266194984, "loss": 0.0058, "step": 8150 }, { "epoch": 48.28402366863905, "grad_norm": 0.146128848195076, "learning_rate": 0.0002527984545468446, "loss": 0.0059, "step": 8160 }, { "epoch": 48.34319526627219, "grad_norm": 0.1363309919834137, "learning_rate": 0.0002526820658893033, "loss": 0.0062, "step": 8170 }, { "epoch": 48.40236686390533, "grad_norm": 0.16651533544063568, "learning_rate": 0.0002525655607788722, "loss": 0.0054, "step": 8180 }, { "epoch": 48.46153846153846, "grad_norm": 0.22147195041179657, "learning_rate": 0.00025244893934768097, "loss": 0.0068, "step": 8190 }, { "epoch": 48.5207100591716, "grad_norm": 0.19092677533626556, "learning_rate": 0.00025233220172799143, "loss": 0.0056, "step": 8200 }, { "epoch": 48.57988165680474, "grad_norm": 0.14065715670585632, "learning_rate": 0.00025221534805219695, "loss": 0.0063, "step": 8210 }, { "epoch": 48.63905325443787, "grad_norm": 0.14423319697380066, "learning_rate": 0.0002520983784528226, "loss": 0.0067, "step": 8220 }, { "epoch": 48.69822485207101, "grad_norm": 0.1487833559513092, "learning_rate": 0.000251981293062525, "loss": 0.0062, "step": 8230 }, { "epoch": 48.75739644970414, "grad_norm": 0.11509937793016434, "learning_rate": 0.0002518640920140919, "loss": 0.006, "step": 8240 }, { "epoch": 48.81656804733728, "grad_norm": 0.12177855521440506, "learning_rate": 0.0002517467754404424, "loss": 0.0057, "step": 8250 }, { "epoch": 48.875739644970416, "grad_norm": 0.13854354619979858, "learning_rate": 0.0002516293434746265, "loss": 0.0057, "step": 8260 }, { "epoch": 48.93491124260355, "grad_norm": 0.15529486536979675, "learning_rate": 0.0002515117962498252, "loss": 0.006, "step": 8270 }, { "epoch": 48.994082840236686, "grad_norm": 0.10648947209119797, "learning_rate": 0.0002513941338993501, "loss": 0.0063, "step": 8280 }, { "epoch": 49.053254437869825, "grad_norm": 0.12991592288017273, "learning_rate": 0.0002512763565566434, "loss": 0.0062, "step": 8290 }, { "epoch": 49.112426035502956, "grad_norm": 0.16192428767681122, "learning_rate": 0.00025115846435527767, "loss": 0.0065, "step": 8300 }, { "epoch": 49.171597633136095, "grad_norm": 0.11252468079328537, "learning_rate": 0.00025104045742895583, "loss": 0.0056, "step": 8310 }, { "epoch": 49.23076923076923, "grad_norm": 0.11284900456666946, "learning_rate": 0.0002509223359115108, "loss": 0.0061, "step": 8320 }, { "epoch": 49.289940828402365, "grad_norm": 0.12242665141820908, "learning_rate": 0.0002508040999369056, "loss": 0.0061, "step": 8330 }, { "epoch": 49.349112426035504, "grad_norm": 0.14561255276203156, "learning_rate": 0.00025068574963923307, "loss": 0.0058, "step": 8340 }, { "epoch": 49.40828402366864, "grad_norm": 0.09667627513408661, "learning_rate": 0.0002505672851527154, "loss": 0.0061, "step": 8350 }, { "epoch": 49.467455621301774, "grad_norm": 0.17537660896778107, "learning_rate": 0.00025044870661170465, "loss": 0.0064, "step": 8360 }, { "epoch": 49.52662721893491, "grad_norm": 0.11456836760044098, "learning_rate": 0.00025033001415068203, "loss": 0.0061, "step": 8370 }, { "epoch": 49.585798816568044, "grad_norm": 0.1197870671749115, "learning_rate": 0.00025021120790425807, "loss": 0.0061, "step": 8380 }, { "epoch": 49.64497041420118, "grad_norm": 0.10899191349744797, "learning_rate": 0.00025009228800717214, "loss": 0.005, "step": 8390 }, { "epoch": 49.70414201183432, "grad_norm": 0.15418791770935059, "learning_rate": 0.00024997325459429276, "loss": 0.0059, "step": 8400 }, { "epoch": 49.76331360946745, "grad_norm": 0.12581446766853333, "learning_rate": 0.00024985410780061705, "loss": 0.0057, "step": 8410 }, { "epoch": 49.82248520710059, "grad_norm": 0.137274369597435, "learning_rate": 0.0002497348477612707, "loss": 0.0055, "step": 8420 }, { "epoch": 49.88165680473373, "grad_norm": 0.18826399743556976, "learning_rate": 0.0002496154746115078, "loss": 0.0067, "step": 8430 }, { "epoch": 49.94082840236686, "grad_norm": 0.1152419000864029, "learning_rate": 0.0002494959884867109, "loss": 0.0058, "step": 8440 }, { "epoch": 50.0, "grad_norm": 0.12623143196105957, "learning_rate": 0.0002493763895223905, "loss": 0.0062, "step": 8450 }, { "epoch": 50.05917159763314, "grad_norm": 0.12379975616931915, "learning_rate": 0.00024925667785418527, "loss": 0.0062, "step": 8460 }, { "epoch": 50.11834319526627, "grad_norm": 0.14767824113368988, "learning_rate": 0.0002491368536178614, "loss": 0.0064, "step": 8470 }, { "epoch": 50.17751479289941, "grad_norm": 0.18615718185901642, "learning_rate": 0.00024901691694931294, "loss": 0.0059, "step": 8480 }, { "epoch": 50.23668639053255, "grad_norm": 0.1271020919084549, "learning_rate": 0.0002488968679845614, "loss": 0.0058, "step": 8490 }, { "epoch": 50.29585798816568, "grad_norm": 0.11452808231115341, "learning_rate": 0.0002487767068597558, "loss": 0.0064, "step": 8500 }, { "epoch": 50.35502958579882, "grad_norm": 0.10692799836397171, "learning_rate": 0.00024865643371117205, "loss": 0.0059, "step": 8510 }, { "epoch": 50.414201183431956, "grad_norm": 0.10128989815711975, "learning_rate": 0.00024853604867521343, "loss": 0.0055, "step": 8520 }, { "epoch": 50.47337278106509, "grad_norm": 0.12310726195573807, "learning_rate": 0.00024841555188840985, "loss": 0.0052, "step": 8530 }, { "epoch": 50.532544378698226, "grad_norm": 0.0794949010014534, "learning_rate": 0.0002482949434874182, "loss": 0.0061, "step": 8540 }, { "epoch": 50.59171597633136, "grad_norm": 0.1046450138092041, "learning_rate": 0.00024817422360902175, "loss": 0.006, "step": 8550 }, { "epoch": 50.650887573964496, "grad_norm": 0.11111565679311752, "learning_rate": 0.00024805339239013024, "loss": 0.0056, "step": 8560 }, { "epoch": 50.710059171597635, "grad_norm": 0.15196217596530914, "learning_rate": 0.00024793244996777975, "loss": 0.006, "step": 8570 }, { "epoch": 50.76923076923077, "grad_norm": 0.17131759226322174, "learning_rate": 0.0002478113964791326, "loss": 0.0063, "step": 8580 }, { "epoch": 50.828402366863905, "grad_norm": 0.10632088780403137, "learning_rate": 0.00024769023206147665, "loss": 0.0063, "step": 8590 }, { "epoch": 50.887573964497044, "grad_norm": 0.1733628213405609, "learning_rate": 0.000247568956852226, "loss": 0.0065, "step": 8600 }, { "epoch": 50.946745562130175, "grad_norm": 0.11857525259256363, "learning_rate": 0.00024744757098892024, "loss": 0.0066, "step": 8610 }, { "epoch": 51.005917159763314, "grad_norm": 0.13555370271205902, "learning_rate": 0.00024732607460922437, "loss": 0.0058, "step": 8620 }, { "epoch": 51.06508875739645, "grad_norm": 0.19389945268630981, "learning_rate": 0.00024720446785092884, "loss": 0.0067, "step": 8630 }, { "epoch": 51.124260355029584, "grad_norm": 0.10990369319915771, "learning_rate": 0.0002470827508519492, "loss": 0.0068, "step": 8640 }, { "epoch": 51.18343195266272, "grad_norm": 0.19940565526485443, "learning_rate": 0.00024696092375032624, "loss": 0.007, "step": 8650 }, { "epoch": 51.24260355029586, "grad_norm": 0.10518334805965424, "learning_rate": 0.0002468389866842253, "loss": 0.0057, "step": 8660 }, { "epoch": 51.30177514792899, "grad_norm": 0.1331917941570282, "learning_rate": 0.00024671693979193656, "loss": 0.0069, "step": 8670 }, { "epoch": 51.36094674556213, "grad_norm": 0.09846493601799011, "learning_rate": 0.00024659478321187485, "loss": 0.006, "step": 8680 }, { "epoch": 51.42011834319526, "grad_norm": 0.14205743372440338, "learning_rate": 0.0002464725170825793, "loss": 0.0054, "step": 8690 }, { "epoch": 51.4792899408284, "grad_norm": 0.0984041839838028, "learning_rate": 0.00024635014154271337, "loss": 0.0058, "step": 8700 }, { "epoch": 51.53846153846154, "grad_norm": 0.1110229566693306, "learning_rate": 0.00024622765673106445, "loss": 0.0051, "step": 8710 }, { "epoch": 51.59763313609467, "grad_norm": 0.17806372046470642, "learning_rate": 0.00024610506278654397, "loss": 0.0055, "step": 8720 }, { "epoch": 51.65680473372781, "grad_norm": 0.16424840688705444, "learning_rate": 0.00024598235984818717, "loss": 0.0057, "step": 8730 }, { "epoch": 51.71597633136095, "grad_norm": 0.1300266981124878, "learning_rate": 0.00024585954805515277, "loss": 0.0056, "step": 8740 }, { "epoch": 51.77514792899408, "grad_norm": 0.0748288631439209, "learning_rate": 0.000245736627546723, "loss": 0.0053, "step": 8750 }, { "epoch": 51.83431952662722, "grad_norm": 0.11978717893362045, "learning_rate": 0.0002456135984623034, "loss": 0.0055, "step": 8760 }, { "epoch": 51.89349112426036, "grad_norm": 0.09913886338472366, "learning_rate": 0.00024549046094142274, "loss": 0.0058, "step": 8770 }, { "epoch": 51.95266272189349, "grad_norm": 0.11639928817749023, "learning_rate": 0.0002453672151237326, "loss": 0.0053, "step": 8780 }, { "epoch": 52.01183431952663, "grad_norm": 0.16336441040039062, "learning_rate": 0.0002452438611490074, "loss": 0.0063, "step": 8790 }, { "epoch": 52.071005917159766, "grad_norm": 0.1398307830095291, "learning_rate": 0.0002451203991571444, "loss": 0.0067, "step": 8800 }, { "epoch": 52.1301775147929, "grad_norm": 0.19877853989601135, "learning_rate": 0.0002449968292881631, "loss": 0.0057, "step": 8810 }, { "epoch": 52.18934911242604, "grad_norm": 0.1644536554813385, "learning_rate": 0.00024487315168220553, "loss": 0.0067, "step": 8820 }, { "epoch": 52.248520710059175, "grad_norm": 0.11039262264966965, "learning_rate": 0.0002447493664795359, "loss": 0.0063, "step": 8830 }, { "epoch": 52.30769230769231, "grad_norm": 0.1324692964553833, "learning_rate": 0.0002446254738205404, "loss": 0.0059, "step": 8840 }, { "epoch": 52.366863905325445, "grad_norm": 0.12639237940311432, "learning_rate": 0.0002445014738457271, "loss": 0.0061, "step": 8850 }, { "epoch": 52.42603550295858, "grad_norm": 0.15594525635242462, "learning_rate": 0.00024437736669572575, "loss": 0.0059, "step": 8860 }, { "epoch": 52.485207100591715, "grad_norm": 0.11067938804626465, "learning_rate": 0.0002442531525112876, "loss": 0.0064, "step": 8870 }, { "epoch": 52.544378698224854, "grad_norm": 0.15557019412517548, "learning_rate": 0.00024412883143328548, "loss": 0.0062, "step": 8880 }, { "epoch": 52.603550295857985, "grad_norm": 0.15376019477844238, "learning_rate": 0.00024400440360271324, "loss": 0.0062, "step": 8890 }, { "epoch": 52.662721893491124, "grad_norm": 0.18752418458461761, "learning_rate": 0.00024387986916068596, "loss": 0.0061, "step": 8900 }, { "epoch": 52.72189349112426, "grad_norm": 0.1110595092177391, "learning_rate": 0.0002437552282484395, "loss": 0.0058, "step": 8910 }, { "epoch": 52.781065088757394, "grad_norm": 0.1901315301656723, "learning_rate": 0.00024363048100733048, "loss": 0.0057, "step": 8920 }, { "epoch": 52.84023668639053, "grad_norm": 0.1274600625038147, "learning_rate": 0.0002435056275788363, "loss": 0.006, "step": 8930 }, { "epoch": 52.89940828402367, "grad_norm": 0.08939477801322937, "learning_rate": 0.00024338066810455453, "loss": 0.0055, "step": 8940 }, { "epoch": 52.9585798816568, "grad_norm": 0.07620944827795029, "learning_rate": 0.00024325560272620315, "loss": 0.0054, "step": 8950 }, { "epoch": 53.01775147928994, "grad_norm": 0.18162639439105988, "learning_rate": 0.00024313043158562023, "loss": 0.0058, "step": 8960 }, { "epoch": 53.07692307692308, "grad_norm": 0.11854640394449234, "learning_rate": 0.0002430051548247637, "loss": 0.0055, "step": 8970 }, { "epoch": 53.13609467455621, "grad_norm": 0.10900015383958817, "learning_rate": 0.0002428797725857115, "loss": 0.0063, "step": 8980 }, { "epoch": 53.19526627218935, "grad_norm": 0.0938253328204155, "learning_rate": 0.00024275428501066085, "loss": 0.0052, "step": 8990 }, { "epoch": 53.25443786982248, "grad_norm": 0.13831236958503723, "learning_rate": 0.00024262869224192876, "loss": 0.0058, "step": 9000 }, { "epoch": 53.31360946745562, "grad_norm": 0.17751143872737885, "learning_rate": 0.00024250299442195138, "loss": 0.0056, "step": 9010 }, { "epoch": 53.37278106508876, "grad_norm": 0.1306639462709427, "learning_rate": 0.00024237719169328402, "loss": 0.0059, "step": 9020 }, { "epoch": 53.43195266272189, "grad_norm": 0.14803485572338104, "learning_rate": 0.00024225128419860098, "loss": 0.0064, "step": 9030 }, { "epoch": 53.49112426035503, "grad_norm": 0.16005265712738037, "learning_rate": 0.00024212527208069537, "loss": 0.0057, "step": 9040 }, { "epoch": 53.55029585798817, "grad_norm": 0.14088213443756104, "learning_rate": 0.00024199915548247893, "loss": 0.0058, "step": 9050 }, { "epoch": 53.6094674556213, "grad_norm": 0.11445368081331253, "learning_rate": 0.00024187293454698195, "loss": 0.0065, "step": 9060 }, { "epoch": 53.66863905325444, "grad_norm": 0.09796857833862305, "learning_rate": 0.00024174660941735297, "loss": 0.0058, "step": 9070 }, { "epoch": 53.72781065088758, "grad_norm": 0.1569296270608902, "learning_rate": 0.00024162018023685876, "loss": 0.0055, "step": 9080 }, { "epoch": 53.78698224852071, "grad_norm": 0.14203175902366638, "learning_rate": 0.00024149364714888414, "loss": 0.0062, "step": 9090 }, { "epoch": 53.84615384615385, "grad_norm": 0.08406265079975128, "learning_rate": 0.00024136701029693166, "loss": 0.0052, "step": 9100 }, { "epoch": 53.905325443786985, "grad_norm": 0.09060053527355194, "learning_rate": 0.0002412402698246216, "loss": 0.0054, "step": 9110 }, { "epoch": 53.96449704142012, "grad_norm": 0.10884205251932144, "learning_rate": 0.0002411134258756918, "loss": 0.0058, "step": 9120 }, { "epoch": 54.023668639053255, "grad_norm": 0.11572816222906113, "learning_rate": 0.00024098647859399736, "loss": 0.0056, "step": 9130 }, { "epoch": 54.082840236686394, "grad_norm": 0.07934154570102692, "learning_rate": 0.0002408594281235107, "loss": 0.0055, "step": 9140 }, { "epoch": 54.142011834319526, "grad_norm": 0.13114017248153687, "learning_rate": 0.00024073227460832112, "loss": 0.0057, "step": 9150 }, { "epoch": 54.201183431952664, "grad_norm": 0.1401418298482895, "learning_rate": 0.0002406050181926349, "loss": 0.0058, "step": 9160 }, { "epoch": 54.260355029585796, "grad_norm": 0.12979751825332642, "learning_rate": 0.00024047765902077492, "loss": 0.0061, "step": 9170 }, { "epoch": 54.319526627218934, "grad_norm": 0.13069404661655426, "learning_rate": 0.00024035019723718068, "loss": 0.0056, "step": 9180 }, { "epoch": 54.37869822485207, "grad_norm": 0.09373171627521515, "learning_rate": 0.00024022263298640806, "loss": 0.0055, "step": 9190 }, { "epoch": 54.437869822485204, "grad_norm": 0.12898816168308258, "learning_rate": 0.00024009496641312912, "loss": 0.005, "step": 9200 }, { "epoch": 54.49704142011834, "grad_norm": 0.15127195417881012, "learning_rate": 0.00023996719766213194, "loss": 0.0056, "step": 9210 }, { "epoch": 54.55621301775148, "grad_norm": 0.15513941645622253, "learning_rate": 0.0002398393268783205, "loss": 0.0056, "step": 9220 }, { "epoch": 54.61538461538461, "grad_norm": 0.1041690856218338, "learning_rate": 0.00023971135420671448, "loss": 0.0054, "step": 9230 }, { "epoch": 54.67455621301775, "grad_norm": 0.09264548867940903, "learning_rate": 0.0002395832797924492, "loss": 0.0055, "step": 9240 }, { "epoch": 54.73372781065089, "grad_norm": 0.17218950390815735, "learning_rate": 0.0002394551037807752, "loss": 0.006, "step": 9250 }, { "epoch": 54.79289940828402, "grad_norm": 0.12399612367153168, "learning_rate": 0.00023932682631705842, "loss": 0.0062, "step": 9260 }, { "epoch": 54.85207100591716, "grad_norm": 0.12354462593793869, "learning_rate": 0.00023919844754677973, "loss": 0.0058, "step": 9270 }, { "epoch": 54.9112426035503, "grad_norm": 0.0991169810295105, "learning_rate": 0.00023906996761553496, "loss": 0.006, "step": 9280 }, { "epoch": 54.97041420118343, "grad_norm": 0.2093593031167984, "learning_rate": 0.00023894138666903468, "loss": 0.0067, "step": 9290 }, { "epoch": 55.02958579881657, "grad_norm": 0.1812223196029663, "learning_rate": 0.00023881270485310395, "loss": 0.0064, "step": 9300 }, { "epoch": 55.0887573964497, "grad_norm": 0.14644643664360046, "learning_rate": 0.00023868392231368227, "loss": 0.0065, "step": 9310 }, { "epoch": 55.14792899408284, "grad_norm": 0.12764909863471985, "learning_rate": 0.00023855503919682344, "loss": 0.0062, "step": 9320 }, { "epoch": 55.20710059171598, "grad_norm": 0.19421128928661346, "learning_rate": 0.0002384260556486952, "loss": 0.0063, "step": 9330 }, { "epoch": 55.26627218934911, "grad_norm": 0.14501290023326874, "learning_rate": 0.00023829697181557922, "loss": 0.0055, "step": 9340 }, { "epoch": 55.32544378698225, "grad_norm": 0.12068941444158554, "learning_rate": 0.00023816778784387094, "loss": 0.0056, "step": 9350 }, { "epoch": 55.38461538461539, "grad_norm": 0.13225489854812622, "learning_rate": 0.00023803850388007947, "loss": 0.0056, "step": 9360 }, { "epoch": 55.44378698224852, "grad_norm": 0.15738902986049652, "learning_rate": 0.00023790912007082702, "loss": 0.0056, "step": 9370 }, { "epoch": 55.50295857988166, "grad_norm": 0.09873766452074051, "learning_rate": 0.0002377796365628494, "loss": 0.0055, "step": 9380 }, { "epoch": 55.562130177514796, "grad_norm": 0.12731672823429108, "learning_rate": 0.00023765005350299525, "loss": 0.0059, "step": 9390 }, { "epoch": 55.62130177514793, "grad_norm": 0.15705305337905884, "learning_rate": 0.00023752037103822617, "loss": 0.0054, "step": 9400 }, { "epoch": 55.680473372781066, "grad_norm": 0.10690966993570328, "learning_rate": 0.00023739058931561654, "loss": 0.0052, "step": 9410 }, { "epoch": 55.739644970414204, "grad_norm": 0.12323465198278427, "learning_rate": 0.00023726070848235328, "loss": 0.0048, "step": 9420 }, { "epoch": 55.798816568047336, "grad_norm": 0.15907087922096252, "learning_rate": 0.0002371307286857356, "loss": 0.0063, "step": 9430 }, { "epoch": 55.857988165680474, "grad_norm": 0.1658487170934677, "learning_rate": 0.0002370006500731752, "loss": 0.0047, "step": 9440 }, { "epoch": 55.917159763313606, "grad_norm": 0.10039862245321274, "learning_rate": 0.00023687047279219567, "loss": 0.006, "step": 9450 }, { "epoch": 55.976331360946745, "grad_norm": 0.10623469203710556, "learning_rate": 0.00023674019699043252, "loss": 0.0057, "step": 9460 }, { "epoch": 56.03550295857988, "grad_norm": 0.1172390878200531, "learning_rate": 0.0002366098228156329, "loss": 0.0059, "step": 9470 }, { "epoch": 56.094674556213015, "grad_norm": 0.12300989031791687, "learning_rate": 0.0002364793504156558, "loss": 0.0047, "step": 9480 }, { "epoch": 56.15384615384615, "grad_norm": 0.10742344707250595, "learning_rate": 0.00023634877993847133, "loss": 0.005, "step": 9490 }, { "epoch": 56.21301775147929, "grad_norm": 0.13516148924827576, "learning_rate": 0.00023621811153216104, "loss": 0.0064, "step": 9500 }, { "epoch": 56.27218934911242, "grad_norm": 0.08080453425645828, "learning_rate": 0.00023608734534491733, "loss": 0.0055, "step": 9510 }, { "epoch": 56.33136094674556, "grad_norm": 0.08332648128271103, "learning_rate": 0.00023595648152504366, "loss": 0.0052, "step": 9520 }, { "epoch": 56.3905325443787, "grad_norm": 0.1893482655286789, "learning_rate": 0.0002358255202209542, "loss": 0.0051, "step": 9530 }, { "epoch": 56.44970414201183, "grad_norm": 0.10477206110954285, "learning_rate": 0.0002356944615811736, "loss": 0.0056, "step": 9540 }, { "epoch": 56.50887573964497, "grad_norm": 0.13092711567878723, "learning_rate": 0.00023556330575433696, "loss": 0.0056, "step": 9550 }, { "epoch": 56.56804733727811, "grad_norm": 0.10823517292737961, "learning_rate": 0.00023543205288918957, "loss": 0.005, "step": 9560 }, { "epoch": 56.62721893491124, "grad_norm": 0.09542914479970932, "learning_rate": 0.0002353007031345868, "loss": 0.0059, "step": 9570 }, { "epoch": 56.68639053254438, "grad_norm": 0.090452641248703, "learning_rate": 0.00023516925663949383, "loss": 0.0054, "step": 9580 }, { "epoch": 56.74556213017752, "grad_norm": 0.14848846197128296, "learning_rate": 0.00023503771355298575, "loss": 0.0058, "step": 9590 }, { "epoch": 56.80473372781065, "grad_norm": 0.1453218162059784, "learning_rate": 0.00023490607402424691, "loss": 0.005, "step": 9600 }, { "epoch": 56.86390532544379, "grad_norm": 0.13269157707691193, "learning_rate": 0.00023477433820257127, "loss": 0.0048, "step": 9610 }, { "epoch": 56.92307692307692, "grad_norm": 0.1185234934091568, "learning_rate": 0.00023464250623736192, "loss": 0.0063, "step": 9620 }, { "epoch": 56.98224852071006, "grad_norm": 0.15270432829856873, "learning_rate": 0.00023451057827813095, "loss": 0.0061, "step": 9630 }, { "epoch": 57.0414201183432, "grad_norm": 0.17734074592590332, "learning_rate": 0.0002343785544744993, "loss": 0.0053, "step": 9640 }, { "epoch": 57.10059171597633, "grad_norm": 0.14712078869342804, "learning_rate": 0.0002342464349761968, "loss": 0.0058, "step": 9650 }, { "epoch": 57.15976331360947, "grad_norm": 0.10419931262731552, "learning_rate": 0.0002341142199330615, "loss": 0.006, "step": 9660 }, { "epoch": 57.218934911242606, "grad_norm": 0.1801144927740097, "learning_rate": 0.00023398190949504004, "loss": 0.0061, "step": 9670 }, { "epoch": 57.27810650887574, "grad_norm": 0.17091859877109528, "learning_rate": 0.0002338495038121872, "loss": 0.006, "step": 9680 }, { "epoch": 57.337278106508876, "grad_norm": 0.1402900665998459, "learning_rate": 0.00023371700303466574, "loss": 0.0052, "step": 9690 }, { "epoch": 57.396449704142015, "grad_norm": 0.1354796141386032, "learning_rate": 0.00023358440731274625, "loss": 0.0058, "step": 9700 }, { "epoch": 57.455621301775146, "grad_norm": 0.1757587492465973, "learning_rate": 0.00023345171679680713, "loss": 0.0061, "step": 9710 }, { "epoch": 57.514792899408285, "grad_norm": 0.14160510897636414, "learning_rate": 0.00023331893163733414, "loss": 0.0061, "step": 9720 }, { "epoch": 57.57396449704142, "grad_norm": 0.1328752040863037, "learning_rate": 0.00023318605198492038, "loss": 0.0056, "step": 9730 }, { "epoch": 57.633136094674555, "grad_norm": 0.09240368753671646, "learning_rate": 0.00023305307799026623, "loss": 0.0053, "step": 9740 }, { "epoch": 57.69230769230769, "grad_norm": 0.11275546252727509, "learning_rate": 0.00023292000980417897, "loss": 0.0053, "step": 9750 }, { "epoch": 57.751479289940825, "grad_norm": 0.08604111522436142, "learning_rate": 0.00023278684757757276, "loss": 0.0054, "step": 9760 }, { "epoch": 57.81065088757396, "grad_norm": 0.0793604627251625, "learning_rate": 0.00023265359146146835, "loss": 0.0046, "step": 9770 }, { "epoch": 57.8698224852071, "grad_norm": 0.11562957614660263, "learning_rate": 0.00023252024160699304, "loss": 0.0051, "step": 9780 }, { "epoch": 57.928994082840234, "grad_norm": 0.13342006504535675, "learning_rate": 0.00023238679816538034, "loss": 0.0056, "step": 9790 }, { "epoch": 57.98816568047337, "grad_norm": 0.10928186029195786, "learning_rate": 0.00023225326128797, "loss": 0.0056, "step": 9800 }, { "epoch": 58.04733727810651, "grad_norm": 0.1521582305431366, "learning_rate": 0.00023211963112620775, "loss": 0.0054, "step": 9810 }, { "epoch": 58.10650887573964, "grad_norm": 0.17638911306858063, "learning_rate": 0.00023198590783164496, "loss": 0.0056, "step": 9820 }, { "epoch": 58.16568047337278, "grad_norm": 0.08938296884298325, "learning_rate": 0.00023185209155593883, "loss": 0.0056, "step": 9830 }, { "epoch": 58.22485207100592, "grad_norm": 0.1084131971001625, "learning_rate": 0.00023171818245085185, "loss": 0.0053, "step": 9840 }, { "epoch": 58.28402366863905, "grad_norm": 0.10997042804956436, "learning_rate": 0.0002315841806682519, "loss": 0.0056, "step": 9850 }, { "epoch": 58.34319526627219, "grad_norm": 0.12015027552843094, "learning_rate": 0.00023145008636011187, "loss": 0.0063, "step": 9860 }, { "epoch": 58.40236686390533, "grad_norm": 0.08944279700517654, "learning_rate": 0.00023131589967850962, "loss": 0.0052, "step": 9870 }, { "epoch": 58.46153846153846, "grad_norm": 0.1189395859837532, "learning_rate": 0.00023118162077562777, "loss": 0.0055, "step": 9880 }, { "epoch": 58.5207100591716, "grad_norm": 0.10460904240608215, "learning_rate": 0.00023104724980375357, "loss": 0.0058, "step": 9890 }, { "epoch": 58.57988165680474, "grad_norm": 0.10382593423128128, "learning_rate": 0.00023091278691527868, "loss": 0.0061, "step": 9900 }, { "epoch": 58.63905325443787, "grad_norm": 0.09118136763572693, "learning_rate": 0.00023077823226269893, "loss": 0.0056, "step": 9910 }, { "epoch": 58.69822485207101, "grad_norm": 0.09945083409547806, "learning_rate": 0.00023064358599861432, "loss": 0.0058, "step": 9920 }, { "epoch": 58.75739644970414, "grad_norm": 0.12513305246829987, "learning_rate": 0.00023050884827572861, "loss": 0.0054, "step": 9930 }, { "epoch": 58.81656804733728, "grad_norm": 0.10897884517908096, "learning_rate": 0.00023037401924684946, "loss": 0.0049, "step": 9940 }, { "epoch": 58.875739644970416, "grad_norm": 0.10397490859031677, "learning_rate": 0.000230239099064888, "loss": 0.0055, "step": 9950 }, { "epoch": 58.93491124260355, "grad_norm": 0.13942644000053406, "learning_rate": 0.00023010408788285864, "loss": 0.0053, "step": 9960 }, { "epoch": 58.994082840236686, "grad_norm": 0.11463487148284912, "learning_rate": 0.00022996898585387915, "loss": 0.006, "step": 9970 }, { "epoch": 59.053254437869825, "grad_norm": 0.09575356543064117, "learning_rate": 0.00022983379313117028, "loss": 0.0059, "step": 9980 }, { "epoch": 59.112426035502956, "grad_norm": 0.09891669452190399, "learning_rate": 0.0002296985098680556, "loss": 0.0056, "step": 9990 }, { "epoch": 59.171597633136095, "grad_norm": 0.11182832717895508, "learning_rate": 0.00022956313621796135, "loss": 0.0051, "step": 10000 }, { "epoch": 59.23076923076923, "grad_norm": 0.09778702259063721, "learning_rate": 0.00022942767233441634, "loss": 0.0054, "step": 10010 }, { "epoch": 59.289940828402365, "grad_norm": 0.10476340353488922, "learning_rate": 0.00022929211837105176, "loss": 0.0046, "step": 10020 }, { "epoch": 59.349112426035504, "grad_norm": 0.08807504922151566, "learning_rate": 0.00022915647448160076, "loss": 0.0054, "step": 10030 }, { "epoch": 59.40828402366864, "grad_norm": 0.20205192267894745, "learning_rate": 0.00022902074081989872, "loss": 0.0056, "step": 10040 }, { "epoch": 59.467455621301774, "grad_norm": 0.13148006796836853, "learning_rate": 0.00022888491753988264, "loss": 0.0061, "step": 10050 }, { "epoch": 59.52662721893491, "grad_norm": 0.11109978705644608, "learning_rate": 0.00022874900479559132, "loss": 0.0066, "step": 10060 }, { "epoch": 59.585798816568044, "grad_norm": 0.1791214495897293, "learning_rate": 0.00022861300274116484, "loss": 0.0059, "step": 10070 }, { "epoch": 59.64497041420118, "grad_norm": 0.09131525456905365, "learning_rate": 0.0002284769115308447, "loss": 0.0051, "step": 10080 }, { "epoch": 59.70414201183432, "grad_norm": 0.1286274790763855, "learning_rate": 0.00022834073131897353, "loss": 0.0054, "step": 10090 }, { "epoch": 59.76331360946745, "grad_norm": 0.12043604254722595, "learning_rate": 0.00022820446225999474, "loss": 0.0059, "step": 10100 }, { "epoch": 59.82248520710059, "grad_norm": 0.1140250712633133, "learning_rate": 0.00022806810450845273, "loss": 0.0065, "step": 10110 }, { "epoch": 59.88165680473373, "grad_norm": 0.15171197056770325, "learning_rate": 0.00022793165821899223, "loss": 0.0061, "step": 10120 }, { "epoch": 59.94082840236686, "grad_norm": 0.16151919960975647, "learning_rate": 0.00022779512354635862, "loss": 0.0063, "step": 10130 }, { "epoch": 60.0, "grad_norm": 0.11245162785053253, "learning_rate": 0.00022765850064539742, "loss": 0.006, "step": 10140 }, { "epoch": 60.05917159763314, "grad_norm": 0.09942913055419922, "learning_rate": 0.0002275217896710541, "loss": 0.0056, "step": 10150 }, { "epoch": 60.11834319526627, "grad_norm": 0.13480602204799652, "learning_rate": 0.0002273849907783743, "loss": 0.0056, "step": 10160 }, { "epoch": 60.17751479289941, "grad_norm": 0.13860592246055603, "learning_rate": 0.00022724810412250293, "loss": 0.0059, "step": 10170 }, { "epoch": 60.23668639053255, "grad_norm": 0.08309619128704071, "learning_rate": 0.00022711112985868492, "loss": 0.0048, "step": 10180 }, { "epoch": 60.29585798816568, "grad_norm": 0.09474894404411316, "learning_rate": 0.00022697406814226425, "loss": 0.0051, "step": 10190 }, { "epoch": 60.35502958579882, "grad_norm": 0.1314166784286499, "learning_rate": 0.00022683691912868411, "loss": 0.0055, "step": 10200 }, { "epoch": 60.414201183431956, "grad_norm": 0.08419807255268097, "learning_rate": 0.0002266996829734868, "loss": 0.0049, "step": 10210 }, { "epoch": 60.47337278106509, "grad_norm": 0.09326400607824326, "learning_rate": 0.00022656235983231344, "loss": 0.0049, "step": 10220 }, { "epoch": 60.532544378698226, "grad_norm": 0.10830884426832199, "learning_rate": 0.0002264249498609036, "loss": 0.0049, "step": 10230 }, { "epoch": 60.59171597633136, "grad_norm": 0.06510218977928162, "learning_rate": 0.00022628745321509567, "loss": 0.0046, "step": 10240 }, { "epoch": 60.650887573964496, "grad_norm": 0.12155796587467194, "learning_rate": 0.00022614987005082596, "loss": 0.0048, "step": 10250 }, { "epoch": 60.710059171597635, "grad_norm": 0.08757446706295013, "learning_rate": 0.00022601220052412925, "loss": 0.0047, "step": 10260 }, { "epoch": 60.76923076923077, "grad_norm": 0.1349210888147354, "learning_rate": 0.00022587444479113804, "loss": 0.0049, "step": 10270 }, { "epoch": 60.828402366863905, "grad_norm": 0.13045701384544373, "learning_rate": 0.0002257366030080826, "loss": 0.0054, "step": 10280 }, { "epoch": 60.887573964497044, "grad_norm": 0.14072254300117493, "learning_rate": 0.00022559867533129092, "loss": 0.0055, "step": 10290 }, { "epoch": 60.946745562130175, "grad_norm": 0.1041608452796936, "learning_rate": 0.00022546066191718835, "loss": 0.0054, "step": 10300 }, { "epoch": 61.005917159763314, "grad_norm": 0.12972873449325562, "learning_rate": 0.0002253225629222974, "loss": 0.0049, "step": 10310 }, { "epoch": 61.06508875739645, "grad_norm": 0.07683192193508148, "learning_rate": 0.00022518437850323778, "loss": 0.0049, "step": 10320 }, { "epoch": 61.124260355029584, "grad_norm": 0.1246119886636734, "learning_rate": 0.0002250461088167259, "loss": 0.0053, "step": 10330 }, { "epoch": 61.18343195266272, "grad_norm": 0.09923265129327774, "learning_rate": 0.00022490775401957504, "loss": 0.0055, "step": 10340 }, { "epoch": 61.24260355029586, "grad_norm": 0.08040069788694382, "learning_rate": 0.00022476931426869496, "loss": 0.0053, "step": 10350 }, { "epoch": 61.30177514792899, "grad_norm": 0.09950467199087143, "learning_rate": 0.00022463078972109168, "loss": 0.0052, "step": 10360 }, { "epoch": 61.36094674556213, "grad_norm": 0.07206843793392181, "learning_rate": 0.00022449218053386747, "loss": 0.0047, "step": 10370 }, { "epoch": 61.42011834319526, "grad_norm": 0.09949488937854767, "learning_rate": 0.00022435348686422062, "loss": 0.0048, "step": 10380 }, { "epoch": 61.4792899408284, "grad_norm": 0.06664257496595383, "learning_rate": 0.0002242147088694451, "loss": 0.0052, "step": 10390 }, { "epoch": 61.53846153846154, "grad_norm": 0.14982745051383972, "learning_rate": 0.00022407584670693063, "loss": 0.0051, "step": 10400 }, { "epoch": 61.59763313609467, "grad_norm": 0.14219392836093903, "learning_rate": 0.00022393690053416245, "loss": 0.0051, "step": 10410 }, { "epoch": 61.65680473372781, "grad_norm": 0.12524275481700897, "learning_rate": 0.0002237978705087208, "loss": 0.0059, "step": 10420 }, { "epoch": 61.71597633136095, "grad_norm": 0.11116066575050354, "learning_rate": 0.00022365875678828138, "loss": 0.0051, "step": 10430 }, { "epoch": 61.77514792899408, "grad_norm": 0.10525581240653992, "learning_rate": 0.00022351955953061453, "loss": 0.0045, "step": 10440 }, { "epoch": 61.83431952662722, "grad_norm": 0.13578540086746216, "learning_rate": 0.0002233802788935855, "loss": 0.0053, "step": 10450 }, { "epoch": 61.89349112426036, "grad_norm": 0.10322466492652893, "learning_rate": 0.000223240915035154, "loss": 0.0048, "step": 10460 }, { "epoch": 61.95266272189349, "grad_norm": 0.11120184510946274, "learning_rate": 0.0002231014681133741, "loss": 0.005, "step": 10470 }, { "epoch": 62.01183431952663, "grad_norm": 0.10388902574777603, "learning_rate": 0.00022296193828639415, "loss": 0.0046, "step": 10480 }, { "epoch": 62.071005917159766, "grad_norm": 0.08874069154262543, "learning_rate": 0.00022282232571245656, "loss": 0.0049, "step": 10490 }, { "epoch": 62.1301775147929, "grad_norm": 0.12969839572906494, "learning_rate": 0.0002226826305498975, "loss": 0.005, "step": 10500 }, { "epoch": 62.18934911242604, "grad_norm": 0.09024021774530411, "learning_rate": 0.00022254285295714683, "loss": 0.0053, "step": 10510 }, { "epoch": 62.248520710059175, "grad_norm": 0.11695891618728638, "learning_rate": 0.00022240299309272786, "loss": 0.0053, "step": 10520 }, { "epoch": 62.30769230769231, "grad_norm": 0.1076703891158104, "learning_rate": 0.00022226305111525726, "loss": 0.0048, "step": 10530 }, { "epoch": 62.366863905325445, "grad_norm": 0.09709338843822479, "learning_rate": 0.00022212302718344485, "loss": 0.0048, "step": 10540 }, { "epoch": 62.42603550295858, "grad_norm": 0.06689807027578354, "learning_rate": 0.00022198292145609328, "loss": 0.0056, "step": 10550 }, { "epoch": 62.485207100591715, "grad_norm": 0.07929633557796478, "learning_rate": 0.00022184273409209813, "loss": 0.0057, "step": 10560 }, { "epoch": 62.544378698224854, "grad_norm": 0.10822342336177826, "learning_rate": 0.00022170246525044733, "loss": 0.0046, "step": 10570 }, { "epoch": 62.603550295857985, "grad_norm": 0.0983365997672081, "learning_rate": 0.0002215621150902215, "loss": 0.005, "step": 10580 }, { "epoch": 62.662721893491124, "grad_norm": 0.08429259806871414, "learning_rate": 0.00022142168377059325, "loss": 0.0051, "step": 10590 }, { "epoch": 62.72189349112426, "grad_norm": 0.11078337579965591, "learning_rate": 0.00022128117145082737, "loss": 0.0046, "step": 10600 }, { "epoch": 62.781065088757394, "grad_norm": 0.11052476614713669, "learning_rate": 0.00022114057829028042, "loss": 0.0046, "step": 10610 }, { "epoch": 62.84023668639053, "grad_norm": 0.12641099095344543, "learning_rate": 0.0002209999044484008, "loss": 0.0048, "step": 10620 }, { "epoch": 62.89940828402367, "grad_norm": 0.12342973053455353, "learning_rate": 0.00022085915008472815, "loss": 0.0043, "step": 10630 }, { "epoch": 62.9585798816568, "grad_norm": 0.0906987264752388, "learning_rate": 0.00022071831535889366, "loss": 0.0044, "step": 10640 }, { "epoch": 63.01775147928994, "grad_norm": 0.09953322261571884, "learning_rate": 0.0002205774004306196, "loss": 0.0048, "step": 10650 }, { "epoch": 63.07692307692308, "grad_norm": 0.09597181528806686, "learning_rate": 0.00022043640545971915, "loss": 0.0048, "step": 10660 }, { "epoch": 63.13609467455621, "grad_norm": 0.13103216886520386, "learning_rate": 0.00022029533060609636, "loss": 0.0044, "step": 10670 }, { "epoch": 63.19526627218935, "grad_norm": 0.10450630635023117, "learning_rate": 0.00022015417602974573, "loss": 0.0051, "step": 10680 }, { "epoch": 63.25443786982248, "grad_norm": 0.1062573567032814, "learning_rate": 0.00022001294189075225, "loss": 0.0051, "step": 10690 }, { "epoch": 63.31360946745562, "grad_norm": 0.09980282187461853, "learning_rate": 0.00021987162834929123, "loss": 0.0046, "step": 10700 }, { "epoch": 63.37278106508876, "grad_norm": 0.09774870425462723, "learning_rate": 0.0002197302355656279, "loss": 0.0045, "step": 10710 }, { "epoch": 63.43195266272189, "grad_norm": 0.0561474934220314, "learning_rate": 0.0002195887637001174, "loss": 0.0044, "step": 10720 }, { "epoch": 63.49112426035503, "grad_norm": 0.07443531602621078, "learning_rate": 0.00021944721291320465, "loss": 0.0057, "step": 10730 }, { "epoch": 63.55029585798817, "grad_norm": 0.09168548882007599, "learning_rate": 0.00021930558336542388, "loss": 0.0054, "step": 10740 }, { "epoch": 63.6094674556213, "grad_norm": 0.09272204339504242, "learning_rate": 0.00021916387521739886, "loss": 0.0046, "step": 10750 }, { "epoch": 63.66863905325444, "grad_norm": 0.15195167064666748, "learning_rate": 0.00021902208862984235, "loss": 0.0049, "step": 10760 }, { "epoch": 63.72781065088758, "grad_norm": 0.11747525632381439, "learning_rate": 0.0002188802237635561, "loss": 0.0044, "step": 10770 }, { "epoch": 63.78698224852071, "grad_norm": 0.09321767091751099, "learning_rate": 0.00021873828077943072, "loss": 0.0046, "step": 10780 }, { "epoch": 63.84615384615385, "grad_norm": 0.16438768804073334, "learning_rate": 0.0002185962598384453, "loss": 0.0055, "step": 10790 }, { "epoch": 63.905325443786985, "grad_norm": 0.17596951127052307, "learning_rate": 0.0002184541611016674, "loss": 0.0052, "step": 10800 }, { "epoch": 63.96449704142012, "grad_norm": 0.14449362456798553, "learning_rate": 0.0002183119847302528, "loss": 0.0055, "step": 10810 }, { "epoch": 64.02366863905326, "grad_norm": 0.1366434097290039, "learning_rate": 0.00021816973088544536, "loss": 0.0054, "step": 10820 }, { "epoch": 64.0828402366864, "grad_norm": 0.18237988650798798, "learning_rate": 0.00021802739972857671, "loss": 0.0055, "step": 10830 }, { "epoch": 64.14201183431953, "grad_norm": 0.11491499841213226, "learning_rate": 0.00021788499142106623, "loss": 0.0051, "step": 10840 }, { "epoch": 64.20118343195266, "grad_norm": 0.11116617172956467, "learning_rate": 0.00021774250612442087, "loss": 0.005, "step": 10850 }, { "epoch": 64.2603550295858, "grad_norm": 0.11360373347997665, "learning_rate": 0.00021759994400023477, "loss": 0.005, "step": 10860 }, { "epoch": 64.31952662721893, "grad_norm": 0.11691198498010635, "learning_rate": 0.00021745730521018918, "loss": 0.0062, "step": 10870 }, { "epoch": 64.37869822485207, "grad_norm": 0.10430306941270828, "learning_rate": 0.00021731458991605242, "loss": 0.0054, "step": 10880 }, { "epoch": 64.43786982248521, "grad_norm": 0.10094352066516876, "learning_rate": 0.00021717179827967955, "loss": 0.0055, "step": 10890 }, { "epoch": 64.49704142011835, "grad_norm": 0.1418985277414322, "learning_rate": 0.00021702893046301208, "loss": 0.0058, "step": 10900 }, { "epoch": 64.55621301775147, "grad_norm": 0.11383172869682312, "learning_rate": 0.00021688598662807814, "loss": 0.0058, "step": 10910 }, { "epoch": 64.61538461538461, "grad_norm": 0.14190992712974548, "learning_rate": 0.0002167429669369918, "loss": 0.0055, "step": 10920 }, { "epoch": 64.67455621301775, "grad_norm": 0.12414297461509705, "learning_rate": 0.00021659987155195343, "loss": 0.0054, "step": 10930 }, { "epoch": 64.73372781065089, "grad_norm": 0.06932941824197769, "learning_rate": 0.00021645670063524905, "loss": 0.0054, "step": 10940 }, { "epoch": 64.79289940828403, "grad_norm": 0.0994962602853775, "learning_rate": 0.00021631345434925046, "loss": 0.0049, "step": 10950 }, { "epoch": 64.85207100591715, "grad_norm": 0.060435157269239426, "learning_rate": 0.00021617013285641482, "loss": 0.0054, "step": 10960 }, { "epoch": 64.91124260355029, "grad_norm": 0.09248533099889755, "learning_rate": 0.00021602673631928475, "loss": 0.005, "step": 10970 }, { "epoch": 64.97041420118343, "grad_norm": 0.12803848087787628, "learning_rate": 0.0002158832649004878, "loss": 0.0049, "step": 10980 }, { "epoch": 65.02958579881657, "grad_norm": 0.14398261904716492, "learning_rate": 0.00021573971876273656, "loss": 0.0049, "step": 10990 }, { "epoch": 65.08875739644971, "grad_norm": 0.10333438962697983, "learning_rate": 0.00021559609806882831, "loss": 0.005, "step": 11000 }, { "epoch": 65.14792899408285, "grad_norm": 0.0988832414150238, "learning_rate": 0.0002154524029816449, "loss": 0.0042, "step": 11010 }, { "epoch": 65.20710059171597, "grad_norm": 0.13144488632678986, "learning_rate": 0.00021530863366415258, "loss": 0.0051, "step": 11020 }, { "epoch": 65.26627218934911, "grad_norm": 0.09724471718072891, "learning_rate": 0.0002151647902794017, "loss": 0.0049, "step": 11030 }, { "epoch": 65.32544378698225, "grad_norm": 0.07109753787517548, "learning_rate": 0.00021502087299052675, "loss": 0.0054, "step": 11040 }, { "epoch": 65.38461538461539, "grad_norm": 0.1248779296875, "learning_rate": 0.00021487688196074583, "loss": 0.0053, "step": 11050 }, { "epoch": 65.44378698224853, "grad_norm": 0.17886269092559814, "learning_rate": 0.00021473281735336091, "loss": 0.0046, "step": 11060 }, { "epoch": 65.50295857988165, "grad_norm": 0.15200591087341309, "learning_rate": 0.00021458867933175726, "loss": 0.0044, "step": 11070 }, { "epoch": 65.56213017751479, "grad_norm": 0.1116427555680275, "learning_rate": 0.0002144444680594034, "loss": 0.0044, "step": 11080 }, { "epoch": 65.62130177514793, "grad_norm": 0.0769992470741272, "learning_rate": 0.00021430018369985096, "loss": 0.0052, "step": 11090 }, { "epoch": 65.68047337278107, "grad_norm": 0.09893304109573364, "learning_rate": 0.00021415582641673452, "loss": 0.0048, "step": 11100 }, { "epoch": 65.7396449704142, "grad_norm": 0.1570497304201126, "learning_rate": 0.00021401139637377124, "loss": 0.0051, "step": 11110 }, { "epoch": 65.79881656804734, "grad_norm": 0.07855653017759323, "learning_rate": 0.00021386689373476087, "loss": 0.0057, "step": 11120 }, { "epoch": 65.85798816568047, "grad_norm": 0.1615055799484253, "learning_rate": 0.00021372231866358555, "loss": 0.0058, "step": 11130 }, { "epoch": 65.9171597633136, "grad_norm": 0.13559557497501373, "learning_rate": 0.00021357767132420942, "loss": 0.0052, "step": 11140 }, { "epoch": 65.97633136094674, "grad_norm": 0.12476927787065506, "learning_rate": 0.0002134329518806787, "loss": 0.0058, "step": 11150 }, { "epoch": 66.03550295857988, "grad_norm": 0.10318192094564438, "learning_rate": 0.0002132881604971213, "loss": 0.0054, "step": 11160 }, { "epoch": 66.09467455621302, "grad_norm": 0.10121892392635345, "learning_rate": 0.0002131432973377468, "loss": 0.0045, "step": 11170 }, { "epoch": 66.15384615384616, "grad_norm": 0.11128852516412735, "learning_rate": 0.00021299836256684617, "loss": 0.0049, "step": 11180 }, { "epoch": 66.21301775147928, "grad_norm": 0.10839638113975525, "learning_rate": 0.0002128533563487915, "loss": 0.0049, "step": 11190 }, { "epoch": 66.27218934911242, "grad_norm": 0.11899107694625854, "learning_rate": 0.00021270827884803606, "loss": 0.0052, "step": 11200 }, { "epoch": 66.33136094674556, "grad_norm": 0.09700806438922882, "learning_rate": 0.00021256313022911382, "loss": 0.0042, "step": 11210 }, { "epoch": 66.3905325443787, "grad_norm": 0.10070978850126266, "learning_rate": 0.0002124179106566395, "loss": 0.0046, "step": 11220 }, { "epoch": 66.44970414201184, "grad_norm": 0.10981670022010803, "learning_rate": 0.0002122726202953082, "loss": 0.0048, "step": 11230 }, { "epoch": 66.50887573964496, "grad_norm": 0.09213441610336304, "learning_rate": 0.00021212725930989547, "loss": 0.0041, "step": 11240 }, { "epoch": 66.5680473372781, "grad_norm": 0.13286958634853363, "learning_rate": 0.0002119818278652567, "loss": 0.0043, "step": 11250 }, { "epoch": 66.62721893491124, "grad_norm": 0.11223473399877548, "learning_rate": 0.00021183632612632744, "loss": 0.0047, "step": 11260 }, { "epoch": 66.68639053254438, "grad_norm": 0.07051069289445877, "learning_rate": 0.00021169075425812284, "loss": 0.0044, "step": 11270 }, { "epoch": 66.74556213017752, "grad_norm": 0.11758240312337875, "learning_rate": 0.00021154511242573755, "loss": 0.0048, "step": 11280 }, { "epoch": 66.80473372781066, "grad_norm": 0.07350611686706543, "learning_rate": 0.00021139940079434568, "loss": 0.0048, "step": 11290 }, { "epoch": 66.86390532544378, "grad_norm": 0.11270706355571747, "learning_rate": 0.00021125361952920038, "loss": 0.0045, "step": 11300 }, { "epoch": 66.92307692307692, "grad_norm": 0.09271367639303207, "learning_rate": 0.00021110776879563388, "loss": 0.0041, "step": 11310 }, { "epoch": 66.98224852071006, "grad_norm": 0.09856130182743073, "learning_rate": 0.00021096184875905708, "loss": 0.004, "step": 11320 }, { "epoch": 67.0414201183432, "grad_norm": 0.08040589094161987, "learning_rate": 0.0002108158595849596, "loss": 0.0046, "step": 11330 }, { "epoch": 67.10059171597634, "grad_norm": 0.09425821900367737, "learning_rate": 0.00021066980143890935, "loss": 0.0054, "step": 11340 }, { "epoch": 67.15976331360947, "grad_norm": 0.09704465419054031, "learning_rate": 0.00021052367448655258, "loss": 0.0041, "step": 11350 }, { "epoch": 67.2189349112426, "grad_norm": 0.1260397732257843, "learning_rate": 0.00021037747889361347, "loss": 0.0043, "step": 11360 }, { "epoch": 67.27810650887574, "grad_norm": 0.1114208921790123, "learning_rate": 0.00021023121482589412, "loss": 0.0054, "step": 11370 }, { "epoch": 67.33727810650888, "grad_norm": 0.1333940029144287, "learning_rate": 0.00021008488244927425, "loss": 0.0051, "step": 11380 }, { "epoch": 67.39644970414201, "grad_norm": 0.09731826186180115, "learning_rate": 0.0002099384819297111, "loss": 0.0055, "step": 11390 }, { "epoch": 67.45562130177515, "grad_norm": 0.11376205086708069, "learning_rate": 0.00020979201343323907, "loss": 0.0048, "step": 11400 }, { "epoch": 67.51479289940828, "grad_norm": 0.09715139865875244, "learning_rate": 0.00020964547712596979, "loss": 0.005, "step": 11410 }, { "epoch": 67.57396449704142, "grad_norm": 0.1319868564605713, "learning_rate": 0.00020949887317409175, "loss": 0.0047, "step": 11420 }, { "epoch": 67.63313609467455, "grad_norm": 0.08470645546913147, "learning_rate": 0.0002093522017438701, "loss": 0.0041, "step": 11430 }, { "epoch": 67.6923076923077, "grad_norm": 0.059222348034381866, "learning_rate": 0.00020920546300164658, "loss": 0.0043, "step": 11440 }, { "epoch": 67.75147928994083, "grad_norm": 0.08248409628868103, "learning_rate": 0.00020905865711383934, "loss": 0.0051, "step": 11450 }, { "epoch": 67.81065088757397, "grad_norm": 0.07004176825284958, "learning_rate": 0.00020891178424694242, "loss": 0.0051, "step": 11460 }, { "epoch": 67.8698224852071, "grad_norm": 0.10224168747663498, "learning_rate": 0.00020876484456752614, "loss": 0.0046, "step": 11470 }, { "epoch": 67.92899408284023, "grad_norm": 0.115745909512043, "learning_rate": 0.0002086178382422364, "loss": 0.0047, "step": 11480 }, { "epoch": 67.98816568047337, "grad_norm": 0.11628622561693192, "learning_rate": 0.0002084707654377947, "loss": 0.0044, "step": 11490 }, { "epoch": 68.04733727810651, "grad_norm": 0.09396225214004517, "learning_rate": 0.0002083236263209981, "loss": 0.0041, "step": 11500 }, { "epoch": 68.10650887573965, "grad_norm": 0.0953371524810791, "learning_rate": 0.00020817642105871857, "loss": 0.0042, "step": 11510 }, { "epoch": 68.16568047337279, "grad_norm": 0.09053921699523926, "learning_rate": 0.00020802914981790332, "loss": 0.0048, "step": 11520 }, { "epoch": 68.22485207100591, "grad_norm": 0.06426618993282318, "learning_rate": 0.00020788181276557436, "loss": 0.0049, "step": 11530 }, { "epoch": 68.28402366863905, "grad_norm": 0.0800623819231987, "learning_rate": 0.00020773441006882823, "loss": 0.0051, "step": 11540 }, { "epoch": 68.34319526627219, "grad_norm": 0.10386860370635986, "learning_rate": 0.00020758694189483608, "loss": 0.0047, "step": 11550 }, { "epoch": 68.40236686390533, "grad_norm": 0.06779385358095169, "learning_rate": 0.0002074394084108432, "loss": 0.0043, "step": 11560 }, { "epoch": 68.46153846153847, "grad_norm": 0.11138040572404861, "learning_rate": 0.00020729180978416888, "loss": 0.0043, "step": 11570 }, { "epoch": 68.52071005917159, "grad_norm": 0.10565588623285294, "learning_rate": 0.0002071441461822066, "loss": 0.0051, "step": 11580 }, { "epoch": 68.57988165680473, "grad_norm": 0.1390363723039627, "learning_rate": 0.00020699641777242308, "loss": 0.0042, "step": 11590 }, { "epoch": 68.63905325443787, "grad_norm": 0.09192502498626709, "learning_rate": 0.00020684862472235895, "loss": 0.0043, "step": 11600 }, { "epoch": 68.69822485207101, "grad_norm": 0.08883720636367798, "learning_rate": 0.00020670076719962795, "loss": 0.0038, "step": 11610 }, { "epoch": 68.75739644970415, "grad_norm": 0.08408467471599579, "learning_rate": 0.00020655284537191688, "loss": 0.0043, "step": 11620 }, { "epoch": 68.81656804733728, "grad_norm": 0.12630638480186462, "learning_rate": 0.0002064048594069856, "loss": 0.0039, "step": 11630 }, { "epoch": 68.87573964497041, "grad_norm": 0.08925037086009979, "learning_rate": 0.00020625680947266672, "loss": 0.0049, "step": 11640 }, { "epoch": 68.93491124260355, "grad_norm": 0.10686878859996796, "learning_rate": 0.00020610869573686524, "loss": 0.0047, "step": 11650 }, { "epoch": 68.99408284023669, "grad_norm": 0.09164133667945862, "learning_rate": 0.0002059605183675587, "loss": 0.0043, "step": 11660 }, { "epoch": 69.05325443786982, "grad_norm": 0.07789511978626251, "learning_rate": 0.00020581227753279664, "loss": 0.005, "step": 11670 }, { "epoch": 69.11242603550296, "grad_norm": 0.10957998782396317, "learning_rate": 0.00020566397340070078, "loss": 0.0044, "step": 11680 }, { "epoch": 69.17159763313609, "grad_norm": 0.09730472415685654, "learning_rate": 0.00020551560613946444, "loss": 0.0049, "step": 11690 }, { "epoch": 69.23076923076923, "grad_norm": 0.10071387141942978, "learning_rate": 0.0002053671759173526, "loss": 0.0047, "step": 11700 }, { "epoch": 69.28994082840237, "grad_norm": 0.08281943947076797, "learning_rate": 0.0002052186829027017, "loss": 0.0042, "step": 11710 }, { "epoch": 69.3491124260355, "grad_norm": 0.08098609745502472, "learning_rate": 0.00020507012726391928, "loss": 0.0048, "step": 11720 }, { "epoch": 69.40828402366864, "grad_norm": 0.07320935279130936, "learning_rate": 0.00020492150916948397, "loss": 0.0041, "step": 11730 }, { "epoch": 69.46745562130178, "grad_norm": 0.09622322767972946, "learning_rate": 0.00020477282878794532, "loss": 0.0045, "step": 11740 }, { "epoch": 69.5266272189349, "grad_norm": 0.09563605487346649, "learning_rate": 0.00020462408628792332, "loss": 0.0042, "step": 11750 }, { "epoch": 69.58579881656804, "grad_norm": 0.09917383641004562, "learning_rate": 0.00020447528183810857, "loss": 0.0037, "step": 11760 }, { "epoch": 69.64497041420118, "grad_norm": 0.08661330491304398, "learning_rate": 0.0002043264156072619, "loss": 0.0042, "step": 11770 }, { "epoch": 69.70414201183432, "grad_norm": 0.08223852515220642, "learning_rate": 0.00020417748776421408, "loss": 0.0045, "step": 11780 }, { "epoch": 69.76331360946746, "grad_norm": 0.0950063169002533, "learning_rate": 0.00020402849847786598, "loss": 0.0043, "step": 11790 }, { "epoch": 69.8224852071006, "grad_norm": 0.08665835857391357, "learning_rate": 0.00020387944791718795, "loss": 0.0053, "step": 11800 }, { "epoch": 69.88165680473372, "grad_norm": 0.16144852340221405, "learning_rate": 0.00020373033625121996, "loss": 0.0046, "step": 11810 }, { "epoch": 69.94082840236686, "grad_norm": 0.08683206140995026, "learning_rate": 0.00020358116364907125, "loss": 0.0039, "step": 11820 }, { "epoch": 70.0, "grad_norm": 0.0774921104311943, "learning_rate": 0.00020343193027992006, "loss": 0.0044, "step": 11830 }, { "epoch": 70.05917159763314, "grad_norm": 0.08366979658603668, "learning_rate": 0.0002032826363130137, "loss": 0.0046, "step": 11840 }, { "epoch": 70.11834319526628, "grad_norm": 0.09230758994817734, "learning_rate": 0.00020313328191766815, "loss": 0.0041, "step": 11850 }, { "epoch": 70.1775147928994, "grad_norm": 0.08692984282970428, "learning_rate": 0.0002029838672632679, "loss": 0.0043, "step": 11860 }, { "epoch": 70.23668639053254, "grad_norm": 0.09858576953411102, "learning_rate": 0.0002028343925192658, "loss": 0.0049, "step": 11870 }, { "epoch": 70.29585798816568, "grad_norm": 0.09896378219127655, "learning_rate": 0.00020268485785518287, "loss": 0.0046, "step": 11880 }, { "epoch": 70.35502958579882, "grad_norm": 0.09193608909845352, "learning_rate": 0.00020253526344060798, "loss": 0.0045, "step": 11890 }, { "epoch": 70.41420118343196, "grad_norm": 0.09095265716314316, "learning_rate": 0.00020238560944519794, "loss": 0.0052, "step": 11900 }, { "epoch": 70.4733727810651, "grad_norm": 0.08170129358768463, "learning_rate": 0.00020223589603867698, "loss": 0.0047, "step": 11910 }, { "epoch": 70.53254437869822, "grad_norm": 0.08285816758871078, "learning_rate": 0.00020208612339083678, "loss": 0.0046, "step": 11920 }, { "epoch": 70.59171597633136, "grad_norm": 0.08765839040279388, "learning_rate": 0.00020193629167153623, "loss": 0.0044, "step": 11930 }, { "epoch": 70.6508875739645, "grad_norm": 0.0872284546494484, "learning_rate": 0.00020178640105070103, "loss": 0.0037, "step": 11940 }, { "epoch": 70.71005917159763, "grad_norm": 0.11650737375020981, "learning_rate": 0.000201636451698324, "loss": 0.0042, "step": 11950 }, { "epoch": 70.76923076923077, "grad_norm": 0.08129605650901794, "learning_rate": 0.00020148644378446432, "loss": 0.0051, "step": 11960 }, { "epoch": 70.82840236686391, "grad_norm": 0.13300298154354095, "learning_rate": 0.00020133637747924763, "loss": 0.0041, "step": 11970 }, { "epoch": 70.88757396449704, "grad_norm": 0.11248444020748138, "learning_rate": 0.00020118625295286583, "loss": 0.0043, "step": 11980 }, { "epoch": 70.94674556213018, "grad_norm": 0.10725320130586624, "learning_rate": 0.0002010360703755769, "loss": 0.0046, "step": 11990 }, { "epoch": 71.00591715976331, "grad_norm": 0.10561610758304596, "learning_rate": 0.0002008858299177045, "loss": 0.004, "step": 12000 }, { "epoch": 71.06508875739645, "grad_norm": 0.10917618125677109, "learning_rate": 0.00020073553174963815, "loss": 0.0035, "step": 12010 }, { "epoch": 71.12426035502959, "grad_norm": 0.08929703384637833, "learning_rate": 0.0002005851760418326, "loss": 0.004, "step": 12020 }, { "epoch": 71.18343195266272, "grad_norm": 0.10458441078662872, "learning_rate": 0.000200434762964808, "loss": 0.0043, "step": 12030 }, { "epoch": 71.24260355029585, "grad_norm": 0.07207003980875015, "learning_rate": 0.00020028429268914946, "loss": 0.004, "step": 12040 }, { "epoch": 71.30177514792899, "grad_norm": 0.11665560305118561, "learning_rate": 0.00020013376538550708, "loss": 0.0039, "step": 12050 }, { "epoch": 71.36094674556213, "grad_norm": 0.10994545370340347, "learning_rate": 0.0001999831812245956, "loss": 0.0039, "step": 12060 }, { "epoch": 71.42011834319527, "grad_norm": 0.07065438479185104, "learning_rate": 0.00019983254037719413, "loss": 0.0036, "step": 12070 }, { "epoch": 71.47928994082841, "grad_norm": 0.09861519187688828, "learning_rate": 0.0001996818430141462, "loss": 0.0045, "step": 12080 }, { "epoch": 71.53846153846153, "grad_norm": 0.08415296673774719, "learning_rate": 0.00019953108930635937, "loss": 0.0049, "step": 12090 }, { "epoch": 71.59763313609467, "grad_norm": 0.07995586842298508, "learning_rate": 0.00019938027942480516, "loss": 0.0044, "step": 12100 }, { "epoch": 71.65680473372781, "grad_norm": 0.11432268470525742, "learning_rate": 0.00019922941354051874, "loss": 0.0044, "step": 12110 }, { "epoch": 71.71597633136095, "grad_norm": 0.14294248819351196, "learning_rate": 0.0001990784918245988, "loss": 0.0048, "step": 12120 }, { "epoch": 71.77514792899409, "grad_norm": 0.1520303636789322, "learning_rate": 0.00019892751444820737, "loss": 0.0046, "step": 12130 }, { "epoch": 71.83431952662721, "grad_norm": 0.0920422226190567, "learning_rate": 0.00019877648158256965, "loss": 0.0046, "step": 12140 }, { "epoch": 71.89349112426035, "grad_norm": 0.06503323465585709, "learning_rate": 0.0001986253933989736, "loss": 0.0047, "step": 12150 }, { "epoch": 71.95266272189349, "grad_norm": 0.0932987630367279, "learning_rate": 0.00019847425006877008, "loss": 0.0049, "step": 12160 }, { "epoch": 72.01183431952663, "grad_norm": 0.0998694896697998, "learning_rate": 0.00019832305176337256, "loss": 0.0058, "step": 12170 }, { "epoch": 72.07100591715977, "grad_norm": 0.054366882890462875, "learning_rate": 0.00019817179865425658, "loss": 0.0044, "step": 12180 }, { "epoch": 72.1301775147929, "grad_norm": 0.11939561367034912, "learning_rate": 0.00019802049091296008, "loss": 0.0051, "step": 12190 }, { "epoch": 72.18934911242603, "grad_norm": 0.08802975714206696, "learning_rate": 0.00019786912871108285, "loss": 0.0039, "step": 12200 }, { "epoch": 72.24852071005917, "grad_norm": 0.10408179461956024, "learning_rate": 0.00019771771222028652, "loss": 0.0054, "step": 12210 }, { "epoch": 72.3076923076923, "grad_norm": 0.09961645305156708, "learning_rate": 0.00019756624161229423, "loss": 0.0048, "step": 12220 }, { "epoch": 72.36686390532545, "grad_norm": 0.08281514793634415, "learning_rate": 0.0001974147170588904, "loss": 0.0048, "step": 12230 }, { "epoch": 72.42603550295858, "grad_norm": 0.11361424624919891, "learning_rate": 0.00019726313873192087, "loss": 0.0049, "step": 12240 }, { "epoch": 72.48520710059172, "grad_norm": 0.09931691735982895, "learning_rate": 0.00019711150680329232, "loss": 0.0048, "step": 12250 }, { "epoch": 72.54437869822485, "grad_norm": 0.09998705238103867, "learning_rate": 0.00019695982144497215, "loss": 0.0045, "step": 12260 }, { "epoch": 72.60355029585799, "grad_norm": 0.09552233666181564, "learning_rate": 0.0001968080828289885, "loss": 0.0059, "step": 12270 }, { "epoch": 72.66272189349112, "grad_norm": 0.13415199518203735, "learning_rate": 0.0001966562911274298, "loss": 0.0045, "step": 12280 }, { "epoch": 72.72189349112426, "grad_norm": 0.08609278500080109, "learning_rate": 0.00019650444651244476, "loss": 0.0049, "step": 12290 }, { "epoch": 72.7810650887574, "grad_norm": 0.09496130049228668, "learning_rate": 0.0001963525491562421, "loss": 0.0038, "step": 12300 }, { "epoch": 72.84023668639053, "grad_norm": 0.07464519888162613, "learning_rate": 0.00019620059923109028, "loss": 0.004, "step": 12310 }, { "epoch": 72.89940828402366, "grad_norm": 0.08590399473905563, "learning_rate": 0.00019604859690931743, "loss": 0.0046, "step": 12320 }, { "epoch": 72.9585798816568, "grad_norm": 0.06719814240932465, "learning_rate": 0.00019589654236331117, "loss": 0.0037, "step": 12330 }, { "epoch": 73.01775147928994, "grad_norm": 0.09863806515932083, "learning_rate": 0.00019574443576551813, "loss": 0.0039, "step": 12340 }, { "epoch": 73.07692307692308, "grad_norm": 0.14548611640930176, "learning_rate": 0.00019559227728844423, "loss": 0.004, "step": 12350 }, { "epoch": 73.13609467455622, "grad_norm": 0.12124160677194595, "learning_rate": 0.00019544006710465415, "loss": 0.0044, "step": 12360 }, { "epoch": 73.19526627218934, "grad_norm": 0.08555400371551514, "learning_rate": 0.00019528780538677107, "loss": 0.004, "step": 12370 }, { "epoch": 73.25443786982248, "grad_norm": 0.10846718400716782, "learning_rate": 0.00019513549230747683, "loss": 0.0038, "step": 12380 }, { "epoch": 73.31360946745562, "grad_norm": 0.10693398863077164, "learning_rate": 0.0001949831280395113, "loss": 0.0042, "step": 12390 }, { "epoch": 73.37278106508876, "grad_norm": 0.0899900272488594, "learning_rate": 0.00019483071275567252, "loss": 0.0049, "step": 12400 }, { "epoch": 73.4319526627219, "grad_norm": 0.09955711662769318, "learning_rate": 0.00019467824662881654, "loss": 0.004, "step": 12410 }, { "epoch": 73.49112426035504, "grad_norm": 0.06656165421009064, "learning_rate": 0.00019452572983185673, "loss": 0.004, "step": 12420 }, { "epoch": 73.55029585798816, "grad_norm": 0.07927776128053665, "learning_rate": 0.00019437316253776422, "loss": 0.0052, "step": 12430 }, { "epoch": 73.6094674556213, "grad_norm": 0.07552718371152878, "learning_rate": 0.0001942205449195673, "loss": 0.0044, "step": 12440 }, { "epoch": 73.66863905325444, "grad_norm": 0.09998009353876114, "learning_rate": 0.0001940678771503513, "loss": 0.0041, "step": 12450 }, { "epoch": 73.72781065088758, "grad_norm": 0.13299888372421265, "learning_rate": 0.00019391515940325843, "loss": 0.0045, "step": 12460 }, { "epoch": 73.78698224852072, "grad_norm": 0.08864020556211472, "learning_rate": 0.00019376239185148767, "loss": 0.0046, "step": 12470 }, { "epoch": 73.84615384615384, "grad_norm": 0.15690624713897705, "learning_rate": 0.0001936095746682944, "loss": 0.005, "step": 12480 }, { "epoch": 73.90532544378698, "grad_norm": 0.09052576124668121, "learning_rate": 0.0001934567080269903, "loss": 0.0048, "step": 12490 }, { "epoch": 73.96449704142012, "grad_norm": 0.09654016047716141, "learning_rate": 0.00019330379210094313, "loss": 0.0043, "step": 12500 }, { "epoch": 74.02366863905326, "grad_norm": 0.062103163450956345, "learning_rate": 0.00019315082706357665, "loss": 0.0038, "step": 12510 }, { "epoch": 74.0828402366864, "grad_norm": 0.12143028527498245, "learning_rate": 0.00019299781308837016, "loss": 0.0042, "step": 12520 }, { "epoch": 74.14201183431953, "grad_norm": 0.08449026942253113, "learning_rate": 0.00019284475034885858, "loss": 0.004, "step": 12530 }, { "epoch": 74.20118343195266, "grad_norm": 0.08362516015768051, "learning_rate": 0.0001926916390186321, "loss": 0.0049, "step": 12540 }, { "epoch": 74.2603550295858, "grad_norm": 0.0817326083779335, "learning_rate": 0.00019253847927133593, "loss": 0.0046, "step": 12550 }, { "epoch": 74.31952662721893, "grad_norm": 0.09577713906764984, "learning_rate": 0.00019238527128067034, "loss": 0.0049, "step": 12560 }, { "epoch": 74.37869822485207, "grad_norm": 0.07543902844190598, "learning_rate": 0.00019223201522039027, "loss": 0.0045, "step": 12570 }, { "epoch": 74.43786982248521, "grad_norm": 0.18848590552806854, "learning_rate": 0.000192078711264305, "loss": 0.0052, "step": 12580 }, { "epoch": 74.49704142011835, "grad_norm": 0.08781205117702484, "learning_rate": 0.00019192535958627844, "loss": 0.0042, "step": 12590 }, { "epoch": 74.55621301775147, "grad_norm": 0.0669671893119812, "learning_rate": 0.00019177196036022838, "loss": 0.0045, "step": 12600 }, { "epoch": 74.61538461538461, "grad_norm": 0.10263246297836304, "learning_rate": 0.00019161851376012663, "loss": 0.0054, "step": 12610 }, { "epoch": 74.67455621301775, "grad_norm": 0.08271636068820953, "learning_rate": 0.00019146501995999873, "loss": 0.0045, "step": 12620 }, { "epoch": 74.73372781065089, "grad_norm": 0.08123993873596191, "learning_rate": 0.00019131147913392367, "loss": 0.0047, "step": 12630 }, { "epoch": 74.79289940828403, "grad_norm": 0.112729512155056, "learning_rate": 0.0001911578914560339, "loss": 0.0044, "step": 12640 }, { "epoch": 74.85207100591715, "grad_norm": 0.09449648857116699, "learning_rate": 0.0001910042571005149, "loss": 0.0039, "step": 12650 }, { "epoch": 74.91124260355029, "grad_norm": 0.0861658975481987, "learning_rate": 0.0001908505762416052, "loss": 0.0042, "step": 12660 }, { "epoch": 74.97041420118343, "grad_norm": 0.07750587165355682, "learning_rate": 0.00019069684905359584, "loss": 0.0046, "step": 12670 }, { "epoch": 75.02958579881657, "grad_norm": 0.076036736369133, "learning_rate": 0.0001905430757108307, "loss": 0.0039, "step": 12680 }, { "epoch": 75.08875739644971, "grad_norm": 0.08722876012325287, "learning_rate": 0.00019038925638770583, "loss": 0.0044, "step": 12690 }, { "epoch": 75.14792899408285, "grad_norm": 0.06700120121240616, "learning_rate": 0.00019023539125866942, "loss": 0.0041, "step": 12700 }, { "epoch": 75.20710059171597, "grad_norm": 0.08348595350980759, "learning_rate": 0.00019008148049822168, "loss": 0.0037, "step": 12710 }, { "epoch": 75.26627218934911, "grad_norm": 0.08631450682878494, "learning_rate": 0.00018992752428091455, "loss": 0.0041, "step": 12720 }, { "epoch": 75.32544378698225, "grad_norm": 0.08918267488479614, "learning_rate": 0.0001897735227813515, "loss": 0.004, "step": 12730 }, { "epoch": 75.38461538461539, "grad_norm": 0.11389092355966568, "learning_rate": 0.0001896194761741874, "loss": 0.0043, "step": 12740 }, { "epoch": 75.44378698224853, "grad_norm": 0.08278380334377289, "learning_rate": 0.00018946538463412814, "loss": 0.0042, "step": 12750 }, { "epoch": 75.50295857988165, "grad_norm": 0.061492182314395905, "learning_rate": 0.00018931124833593084, "loss": 0.0037, "step": 12760 }, { "epoch": 75.56213017751479, "grad_norm": 0.06527629494667053, "learning_rate": 0.00018915706745440308, "loss": 0.0037, "step": 12770 }, { "epoch": 75.62130177514793, "grad_norm": 0.07585921883583069, "learning_rate": 0.00018900284216440325, "loss": 0.0038, "step": 12780 }, { "epoch": 75.68047337278107, "grad_norm": 0.05799873545765877, "learning_rate": 0.00018884857264083984, "loss": 0.0041, "step": 12790 }, { "epoch": 75.7396449704142, "grad_norm": 0.0605347715318203, "learning_rate": 0.00018869425905867175, "loss": 0.0037, "step": 12800 }, { "epoch": 75.79881656804734, "grad_norm": 0.06910619884729385, "learning_rate": 0.0001885399015929078, "loss": 0.0038, "step": 12810 }, { "epoch": 75.85798816568047, "grad_norm": 0.08011303842067719, "learning_rate": 0.00018838550041860643, "loss": 0.0044, "step": 12820 }, { "epoch": 75.9171597633136, "grad_norm": 0.09228195995092392, "learning_rate": 0.00018823105571087582, "loss": 0.0037, "step": 12830 }, { "epoch": 75.97633136094674, "grad_norm": 0.10285245627164841, "learning_rate": 0.0001880765676448735, "loss": 0.0046, "step": 12840 }, { "epoch": 76.03550295857988, "grad_norm": 0.07767597585916519, "learning_rate": 0.00018792203639580603, "loss": 0.0041, "step": 12850 }, { "epoch": 76.09467455621302, "grad_norm": 0.09341380000114441, "learning_rate": 0.0001877674621389291, "loss": 0.0035, "step": 12860 }, { "epoch": 76.15384615384616, "grad_norm": 0.07405874878168106, "learning_rate": 0.00018761284504954708, "loss": 0.0038, "step": 12870 }, { "epoch": 76.21301775147928, "grad_norm": 0.08551403135061264, "learning_rate": 0.000187458185303013, "loss": 0.004, "step": 12880 }, { "epoch": 76.27218934911242, "grad_norm": 0.09447237849235535, "learning_rate": 0.00018730348307472824, "loss": 0.0037, "step": 12890 }, { "epoch": 76.33136094674556, "grad_norm": 0.08931015431880951, "learning_rate": 0.00018714873854014225, "loss": 0.004, "step": 12900 }, { "epoch": 76.3905325443787, "grad_norm": 0.08087235689163208, "learning_rate": 0.00018699395187475265, "loss": 0.0037, "step": 12910 }, { "epoch": 76.44970414201184, "grad_norm": 0.06926842033863068, "learning_rate": 0.00018683912325410468, "loss": 0.0041, "step": 12920 }, { "epoch": 76.50887573964496, "grad_norm": 0.08471207320690155, "learning_rate": 0.00018668425285379127, "loss": 0.0036, "step": 12930 }, { "epoch": 76.5680473372781, "grad_norm": 0.07987215369939804, "learning_rate": 0.00018652934084945267, "loss": 0.0039, "step": 12940 }, { "epoch": 76.62721893491124, "grad_norm": 0.09502380341291428, "learning_rate": 0.00018637438741677631, "loss": 0.0035, "step": 12950 }, { "epoch": 76.68639053254438, "grad_norm": 0.09775391966104507, "learning_rate": 0.0001862193927314967, "loss": 0.004, "step": 12960 }, { "epoch": 76.74556213017752, "grad_norm": 0.08906898647546768, "learning_rate": 0.00018606435696939501, "loss": 0.0042, "step": 12970 }, { "epoch": 76.80473372781066, "grad_norm": 0.10455282777547836, "learning_rate": 0.00018590928030629903, "loss": 0.0045, "step": 12980 }, { "epoch": 76.86390532544378, "grad_norm": 0.1704987734556198, "learning_rate": 0.00018575416291808297, "loss": 0.005, "step": 12990 }, { "epoch": 76.92307692307692, "grad_norm": 0.09462543576955795, "learning_rate": 0.00018559900498066724, "loss": 0.0047, "step": 13000 }, { "epoch": 76.98224852071006, "grad_norm": 0.08508682250976562, "learning_rate": 0.00018544380667001814, "loss": 0.0047, "step": 13010 }, { "epoch": 77.0414201183432, "grad_norm": 0.10631991177797318, "learning_rate": 0.00018528856816214787, "loss": 0.0044, "step": 13020 }, { "epoch": 77.10059171597634, "grad_norm": 0.1298869401216507, "learning_rate": 0.00018513328963311424, "loss": 0.004, "step": 13030 }, { "epoch": 77.15976331360947, "grad_norm": 0.10753747820854187, "learning_rate": 0.00018497797125902022, "loss": 0.004, "step": 13040 }, { "epoch": 77.2189349112426, "grad_norm": 0.07578105479478836, "learning_rate": 0.00018482261321601433, "loss": 0.0039, "step": 13050 }, { "epoch": 77.27810650887574, "grad_norm": 0.06981521844863892, "learning_rate": 0.0001846672156802897, "loss": 0.004, "step": 13060 }, { "epoch": 77.33727810650888, "grad_norm": 0.09316109120845795, "learning_rate": 0.0001845117788280845, "loss": 0.0048, "step": 13070 }, { "epoch": 77.39644970414201, "grad_norm": 0.0852401927113533, "learning_rate": 0.0001843563028356815, "loss": 0.0047, "step": 13080 }, { "epoch": 77.45562130177515, "grad_norm": 0.10565965622663498, "learning_rate": 0.00018420078787940764, "loss": 0.0045, "step": 13090 }, { "epoch": 77.51479289940828, "grad_norm": 0.09836021810770035, "learning_rate": 0.00018404523413563425, "loss": 0.0042, "step": 13100 }, { "epoch": 77.57396449704142, "grad_norm": 0.19807825982570648, "learning_rate": 0.00018388964178077662, "loss": 0.0044, "step": 13110 }, { "epoch": 77.63313609467455, "grad_norm": 0.11896628141403198, "learning_rate": 0.00018373401099129374, "loss": 0.0043, "step": 13120 }, { "epoch": 77.6923076923077, "grad_norm": 0.07994246482849121, "learning_rate": 0.00018357834194368823, "loss": 0.0046, "step": 13130 }, { "epoch": 77.75147928994083, "grad_norm": 0.07171639055013657, "learning_rate": 0.00018342263481450617, "loss": 0.0044, "step": 13140 }, { "epoch": 77.81065088757397, "grad_norm": 0.09270693361759186, "learning_rate": 0.00018326688978033678, "loss": 0.0042, "step": 13150 }, { "epoch": 77.8698224852071, "grad_norm": 0.06270068883895874, "learning_rate": 0.00018311110701781222, "loss": 0.0036, "step": 13160 }, { "epoch": 77.92899408284023, "grad_norm": 0.09300383180379868, "learning_rate": 0.00018295528670360747, "loss": 0.0045, "step": 13170 }, { "epoch": 77.98816568047337, "grad_norm": 0.06823378056287766, "learning_rate": 0.0001827994290144402, "loss": 0.0033, "step": 13180 }, { "epoch": 78.04733727810651, "grad_norm": 0.09153536707162857, "learning_rate": 0.00018264353412707026, "loss": 0.0039, "step": 13190 }, { "epoch": 78.10650887573965, "grad_norm": 0.1081523597240448, "learning_rate": 0.0001824876022182998, "loss": 0.0037, "step": 13200 }, { "epoch": 78.16568047337279, "grad_norm": 0.08492309600114822, "learning_rate": 0.00018233163346497304, "loss": 0.0039, "step": 13210 }, { "epoch": 78.22485207100591, "grad_norm": 0.09686946868896484, "learning_rate": 0.00018217562804397587, "loss": 0.0042, "step": 13220 }, { "epoch": 78.28402366863905, "grad_norm": 0.10443340241909027, "learning_rate": 0.00018201958613223572, "loss": 0.0036, "step": 13230 }, { "epoch": 78.34319526627219, "grad_norm": 0.12021367996931076, "learning_rate": 0.00018186350790672166, "loss": 0.0036, "step": 13240 }, { "epoch": 78.40236686390533, "grad_norm": 0.08438213914632797, "learning_rate": 0.00018170739354444364, "loss": 0.004, "step": 13250 }, { "epoch": 78.46153846153847, "grad_norm": 0.08048516511917114, "learning_rate": 0.00018155124322245277, "loss": 0.0041, "step": 13260 }, { "epoch": 78.52071005917159, "grad_norm": 0.09707218408584595, "learning_rate": 0.00018139505711784078, "loss": 0.0036, "step": 13270 }, { "epoch": 78.57988165680473, "grad_norm": 0.056621816009283066, "learning_rate": 0.00018123883540774023, "loss": 0.0039, "step": 13280 }, { "epoch": 78.63905325443787, "grad_norm": 0.0652017667889595, "learning_rate": 0.0001810825782693239, "loss": 0.004, "step": 13290 }, { "epoch": 78.69822485207101, "grad_norm": 0.11160093545913696, "learning_rate": 0.0001809262858798047, "loss": 0.004, "step": 13300 }, { "epoch": 78.75739644970415, "grad_norm": 0.06413235515356064, "learning_rate": 0.00018076995841643565, "loss": 0.0039, "step": 13310 }, { "epoch": 78.81656804733728, "grad_norm": 0.11700308322906494, "learning_rate": 0.00018061359605650956, "loss": 0.0045, "step": 13320 }, { "epoch": 78.87573964497041, "grad_norm": 0.06952869892120361, "learning_rate": 0.00018045719897735857, "loss": 0.0038, "step": 13330 }, { "epoch": 78.93491124260355, "grad_norm": 0.07781681418418884, "learning_rate": 0.00018030076735635453, "loss": 0.0042, "step": 13340 }, { "epoch": 78.99408284023669, "grad_norm": 0.06501255929470062, "learning_rate": 0.00018014430137090824, "loss": 0.0038, "step": 13350 }, { "epoch": 79.05325443786982, "grad_norm": 0.09328047931194305, "learning_rate": 0.00017998780119846952, "loss": 0.0039, "step": 13360 }, { "epoch": 79.11242603550296, "grad_norm": 0.06799747049808502, "learning_rate": 0.0001798312670165271, "loss": 0.0037, "step": 13370 }, { "epoch": 79.17159763313609, "grad_norm": 0.04706399142742157, "learning_rate": 0.000179674699002608, "loss": 0.0033, "step": 13380 }, { "epoch": 79.23076923076923, "grad_norm": 0.06519832462072372, "learning_rate": 0.00017951809733427783, "loss": 0.0036, "step": 13390 }, { "epoch": 79.28994082840237, "grad_norm": 0.08606372773647308, "learning_rate": 0.00017936146218914037, "loss": 0.0039, "step": 13400 }, { "epoch": 79.3491124260355, "grad_norm": 0.07776906341314316, "learning_rate": 0.00017920479374483725, "loss": 0.0038, "step": 13410 }, { "epoch": 79.40828402366864, "grad_norm": 0.07003175467252731, "learning_rate": 0.00017904809217904793, "loss": 0.0035, "step": 13420 }, { "epoch": 79.46745562130178, "grad_norm": 0.06181245669722557, "learning_rate": 0.00017889135766948944, "loss": 0.0038, "step": 13430 }, { "epoch": 79.5266272189349, "grad_norm": 0.07188297808170319, "learning_rate": 0.00017873459039391607, "loss": 0.0034, "step": 13440 }, { "epoch": 79.58579881656804, "grad_norm": 0.062446191906929016, "learning_rate": 0.00017857779053011952, "loss": 0.0038, "step": 13450 }, { "epoch": 79.64497041420118, "grad_norm": 0.06526285409927368, "learning_rate": 0.00017842095825592816, "loss": 0.0033, "step": 13460 }, { "epoch": 79.70414201183432, "grad_norm": 0.09937091171741486, "learning_rate": 0.00017826409374920726, "loss": 0.0045, "step": 13470 }, { "epoch": 79.76331360946746, "grad_norm": 0.10404625535011292, "learning_rate": 0.0001781071971878587, "loss": 0.0042, "step": 13480 }, { "epoch": 79.8224852071006, "grad_norm": 0.08594011515378952, "learning_rate": 0.00017795026874982054, "loss": 0.0037, "step": 13490 }, { "epoch": 79.88165680473372, "grad_norm": 0.06377685815095901, "learning_rate": 0.00017779330861306713, "loss": 0.0043, "step": 13500 }, { "epoch": 79.94082840236686, "grad_norm": 0.13347353041172028, "learning_rate": 0.00017763631695560876, "loss": 0.0041, "step": 13510 }, { "epoch": 80.0, "grad_norm": 0.11772621423006058, "learning_rate": 0.00017747929395549143, "loss": 0.0043, "step": 13520 }, { "epoch": 80.05917159763314, "grad_norm": 0.09572026878595352, "learning_rate": 0.0001773222397907967, "loss": 0.0044, "step": 13530 }, { "epoch": 80.11834319526628, "grad_norm": 0.059374939650297165, "learning_rate": 0.00017716515463964139, "loss": 0.0042, "step": 13540 }, { "epoch": 80.1775147928994, "grad_norm": 0.10720463842153549, "learning_rate": 0.00017700803868017764, "loss": 0.0042, "step": 13550 }, { "epoch": 80.23668639053254, "grad_norm": 0.06390975415706635, "learning_rate": 0.00017685089209059247, "loss": 0.0035, "step": 13560 }, { "epoch": 80.29585798816568, "grad_norm": 0.10361231118440628, "learning_rate": 0.00017669371504910745, "loss": 0.0037, "step": 13570 }, { "epoch": 80.35502958579882, "grad_norm": 0.06970731914043427, "learning_rate": 0.000176536507733979, "loss": 0.0036, "step": 13580 }, { "epoch": 80.41420118343196, "grad_norm": 0.07721058279275894, "learning_rate": 0.00017637927032349755, "loss": 0.0039, "step": 13590 }, { "epoch": 80.4733727810651, "grad_norm": 0.10870929062366486, "learning_rate": 0.00017622200299598792, "loss": 0.0043, "step": 13600 }, { "epoch": 80.53254437869822, "grad_norm": 0.09446538984775543, "learning_rate": 0.0001760647059298088, "loss": 0.0036, "step": 13610 }, { "epoch": 80.59171597633136, "grad_norm": 0.11219915002584457, "learning_rate": 0.0001759073793033525, "loss": 0.004, "step": 13620 }, { "epoch": 80.6508875739645, "grad_norm": 0.07490500062704086, "learning_rate": 0.00017575002329504492, "loss": 0.004, "step": 13630 }, { "epoch": 80.71005917159763, "grad_norm": 0.08645602315664291, "learning_rate": 0.00017559263808334527, "loss": 0.0044, "step": 13640 }, { "epoch": 80.76923076923077, "grad_norm": 0.07122586667537689, "learning_rate": 0.00017543522384674595, "loss": 0.0044, "step": 13650 }, { "epoch": 80.82840236686391, "grad_norm": 0.16270536184310913, "learning_rate": 0.0001752777807637722, "loss": 0.0044, "step": 13660 }, { "epoch": 80.88757396449704, "grad_norm": 0.12158726155757904, "learning_rate": 0.000175120309012982, "loss": 0.0041, "step": 13670 }, { "epoch": 80.94674556213018, "grad_norm": 0.13042522966861725, "learning_rate": 0.0001749628087729658, "loss": 0.0038, "step": 13680 }, { "epoch": 81.00591715976331, "grad_norm": 0.08451945334672928, "learning_rate": 0.0001748052802223465, "loss": 0.0041, "step": 13690 }, { "epoch": 81.06508875739645, "grad_norm": 0.06821294873952866, "learning_rate": 0.0001746477235397789, "loss": 0.0037, "step": 13700 }, { "epoch": 81.12426035502959, "grad_norm": 0.0800672322511673, "learning_rate": 0.00017449013890394986, "loss": 0.0038, "step": 13710 }, { "epoch": 81.18343195266272, "grad_norm": 0.10490042716264725, "learning_rate": 0.00017433252649357793, "loss": 0.0041, "step": 13720 }, { "epoch": 81.24260355029585, "grad_norm": 0.10653967410326004, "learning_rate": 0.00017417488648741302, "loss": 0.004, "step": 13730 }, { "epoch": 81.30177514792899, "grad_norm": 0.10760916769504547, "learning_rate": 0.0001740172190642365, "loss": 0.0037, "step": 13740 }, { "epoch": 81.36094674556213, "grad_norm": 0.11505173146724701, "learning_rate": 0.0001738595244028608, "loss": 0.004, "step": 13750 }, { "epoch": 81.42011834319527, "grad_norm": 0.0937047228217125, "learning_rate": 0.0001737018026821292, "loss": 0.0035, "step": 13760 }, { "epoch": 81.47928994082841, "grad_norm": 0.11170636862516403, "learning_rate": 0.00017354405408091567, "loss": 0.0041, "step": 13770 }, { "epoch": 81.53846153846153, "grad_norm": 0.15958480536937714, "learning_rate": 0.00017338627877812467, "loss": 0.004, "step": 13780 }, { "epoch": 81.59763313609467, "grad_norm": 0.16305002570152283, "learning_rate": 0.00017322847695269095, "loss": 0.0049, "step": 13790 }, { "epoch": 81.65680473372781, "grad_norm": 0.10061042755842209, "learning_rate": 0.0001730706487835794, "loss": 0.0044, "step": 13800 }, { "epoch": 81.71597633136095, "grad_norm": 0.09760911762714386, "learning_rate": 0.00017291279444978466, "loss": 0.0042, "step": 13810 }, { "epoch": 81.77514792899409, "grad_norm": 0.09667602181434631, "learning_rate": 0.00017275491413033114, "loss": 0.0042, "step": 13820 }, { "epoch": 81.83431952662721, "grad_norm": 0.09997069835662842, "learning_rate": 0.00017259700800427267, "loss": 0.0045, "step": 13830 }, { "epoch": 81.89349112426035, "grad_norm": 0.07012118399143219, "learning_rate": 0.0001724390762506924, "loss": 0.0036, "step": 13840 }, { "epoch": 81.95266272189349, "grad_norm": 0.07439204305410385, "learning_rate": 0.0001722811190487025, "loss": 0.0035, "step": 13850 }, { "epoch": 82.01183431952663, "grad_norm": 0.07580581307411194, "learning_rate": 0.000172123136577444, "loss": 0.0043, "step": 13860 }, { "epoch": 82.07100591715977, "grad_norm": 0.06705798208713531, "learning_rate": 0.00017196512901608666, "loss": 0.0041, "step": 13870 }, { "epoch": 82.1301775147929, "grad_norm": 0.06895974278450012, "learning_rate": 0.0001718070965438286, "loss": 0.0038, "step": 13880 }, { "epoch": 82.18934911242603, "grad_norm": 0.07057873159646988, "learning_rate": 0.0001716490393398962, "loss": 0.0034, "step": 13890 }, { "epoch": 82.24852071005917, "grad_norm": 0.07738516479730606, "learning_rate": 0.00017149095758354395, "loss": 0.0033, "step": 13900 }, { "epoch": 82.3076923076923, "grad_norm": 0.04670868441462517, "learning_rate": 0.00017133285145405421, "loss": 0.0036, "step": 13910 }, { "epoch": 82.36686390532545, "grad_norm": 0.09428457170724869, "learning_rate": 0.00017117472113073681, "loss": 0.0038, "step": 13920 }, { "epoch": 82.42603550295858, "grad_norm": 0.09265027940273285, "learning_rate": 0.00017101656679292923, "loss": 0.0036, "step": 13930 }, { "epoch": 82.48520710059172, "grad_norm": 0.1075807437300682, "learning_rate": 0.00017085838861999602, "loss": 0.0035, "step": 13940 }, { "epoch": 82.54437869822485, "grad_norm": 0.09126709401607513, "learning_rate": 0.00017070018679132885, "loss": 0.0039, "step": 13950 }, { "epoch": 82.60355029585799, "grad_norm": 0.08693982660770416, "learning_rate": 0.00017054196148634623, "loss": 0.0042, "step": 13960 }, { "epoch": 82.66272189349112, "grad_norm": 0.10599178820848465, "learning_rate": 0.00017038371288449327, "loss": 0.0043, "step": 13970 }, { "epoch": 82.72189349112426, "grad_norm": 0.09200841933488846, "learning_rate": 0.00017022544116524147, "loss": 0.0036, "step": 13980 }, { "epoch": 82.7810650887574, "grad_norm": 0.08926296979188919, "learning_rate": 0.00017006714650808855, "loss": 0.0045, "step": 13990 }, { "epoch": 82.84023668639053, "grad_norm": 0.07983355224132538, "learning_rate": 0.00016990882909255828, "loss": 0.0041, "step": 14000 }, { "epoch": 82.89940828402366, "grad_norm": 0.07399842888116837, "learning_rate": 0.0001697504890982003, "loss": 0.0037, "step": 14010 }, { "epoch": 82.9585798816568, "grad_norm": 0.11194520443677902, "learning_rate": 0.00016959212670458965, "loss": 0.004, "step": 14020 }, { "epoch": 83.01775147928994, "grad_norm": 0.13375751674175262, "learning_rate": 0.00016943374209132706, "loss": 0.0039, "step": 14030 }, { "epoch": 83.07692307692308, "grad_norm": 0.09113836288452148, "learning_rate": 0.0001692753354380382, "loss": 0.0044, "step": 14040 }, { "epoch": 83.13609467455622, "grad_norm": 0.06816516816616058, "learning_rate": 0.00016911690692437384, "loss": 0.0034, "step": 14050 }, { "epoch": 83.19526627218934, "grad_norm": 0.08450283855199814, "learning_rate": 0.00016895845673000964, "loss": 0.0033, "step": 14060 }, { "epoch": 83.25443786982248, "grad_norm": 0.07242479175329208, "learning_rate": 0.00016879998503464561, "loss": 0.0034, "step": 14070 }, { "epoch": 83.31360946745562, "grad_norm": 0.08749764412641525, "learning_rate": 0.00016864149201800644, "loss": 0.0036, "step": 14080 }, { "epoch": 83.37278106508876, "grad_norm": 0.0878598541021347, "learning_rate": 0.00016848297785984075, "loss": 0.0041, "step": 14090 }, { "epoch": 83.4319526627219, "grad_norm": 0.061229582875967026, "learning_rate": 0.00016832444273992127, "loss": 0.0043, "step": 14100 }, { "epoch": 83.49112426035504, "grad_norm": 0.10403190553188324, "learning_rate": 0.00016816588683804447, "loss": 0.0039, "step": 14110 }, { "epoch": 83.55029585798816, "grad_norm": 0.08923132717609406, "learning_rate": 0.0001680073103340304, "loss": 0.0038, "step": 14120 }, { "epoch": 83.6094674556213, "grad_norm": 0.11593616008758545, "learning_rate": 0.00016784871340772247, "loss": 0.0043, "step": 14130 }, { "epoch": 83.66863905325444, "grad_norm": 0.08716775476932526, "learning_rate": 0.00016769009623898715, "loss": 0.004, "step": 14140 }, { "epoch": 83.72781065088758, "grad_norm": 0.09229385852813721, "learning_rate": 0.00016753145900771409, "loss": 0.0038, "step": 14150 }, { "epoch": 83.78698224852072, "grad_norm": 0.08212786167860031, "learning_rate": 0.00016737280189381542, "loss": 0.0038, "step": 14160 }, { "epoch": 83.84615384615384, "grad_norm": 0.07167596369981766, "learning_rate": 0.00016721412507722617, "loss": 0.0031, "step": 14170 }, { "epoch": 83.90532544378698, "grad_norm": 0.11478660255670547, "learning_rate": 0.00016705542873790327, "loss": 0.0037, "step": 14180 }, { "epoch": 83.96449704142012, "grad_norm": 0.10585403442382812, "learning_rate": 0.00016689671305582623, "loss": 0.0033, "step": 14190 }, { "epoch": 84.02366863905326, "grad_norm": 0.07320217788219452, "learning_rate": 0.0001667379782109962, "loss": 0.0035, "step": 14200 }, { "epoch": 84.0828402366864, "grad_norm": 0.06419624388217926, "learning_rate": 0.00016657922438343615, "loss": 0.004, "step": 14210 }, { "epoch": 84.14201183431953, "grad_norm": 0.06397725641727448, "learning_rate": 0.0001664204517531906, "loss": 0.0038, "step": 14220 }, { "epoch": 84.20118343195266, "grad_norm": 0.1265418976545334, "learning_rate": 0.00016626166050032543, "loss": 0.0038, "step": 14230 }, { "epoch": 84.2603550295858, "grad_norm": 0.06702183932065964, "learning_rate": 0.00016610285080492754, "loss": 0.0031, "step": 14240 }, { "epoch": 84.31952662721893, "grad_norm": 0.05925055965781212, "learning_rate": 0.0001659440228471048, "loss": 0.0033, "step": 14250 }, { "epoch": 84.37869822485207, "grad_norm": 0.054463211447000504, "learning_rate": 0.00016578517680698583, "loss": 0.0037, "step": 14260 }, { "epoch": 84.43786982248521, "grad_norm": 0.08211854845285416, "learning_rate": 0.00016562631286471964, "loss": 0.0033, "step": 14270 }, { "epoch": 84.49704142011835, "grad_norm": 0.06943093985319138, "learning_rate": 0.0001654674312004757, "loss": 0.003, "step": 14280 }, { "epoch": 84.55621301775147, "grad_norm": 0.058948930352926254, "learning_rate": 0.00016530853199444345, "loss": 0.0032, "step": 14290 }, { "epoch": 84.61538461538461, "grad_norm": 0.06443957984447479, "learning_rate": 0.0001651496154268323, "loss": 0.0033, "step": 14300 }, { "epoch": 84.67455621301775, "grad_norm": 0.07743290811777115, "learning_rate": 0.00016499068167787133, "loss": 0.0034, "step": 14310 }, { "epoch": 84.73372781065089, "grad_norm": 0.07515314966440201, "learning_rate": 0.00016483173092780908, "loss": 0.0044, "step": 14320 }, { "epoch": 84.79289940828403, "grad_norm": 0.08253543823957443, "learning_rate": 0.00016467276335691336, "loss": 0.0037, "step": 14330 }, { "epoch": 84.85207100591715, "grad_norm": 0.1280813068151474, "learning_rate": 0.00016451377914547115, "loss": 0.0041, "step": 14340 }, { "epoch": 84.91124260355029, "grad_norm": 0.10183922946453094, "learning_rate": 0.00016435477847378816, "loss": 0.0037, "step": 14350 }, { "epoch": 84.97041420118343, "grad_norm": 0.10940897464752197, "learning_rate": 0.00016419576152218896, "loss": 0.0039, "step": 14360 }, { "epoch": 85.02958579881657, "grad_norm": 0.07498513907194138, "learning_rate": 0.00016403672847101628, "loss": 0.0032, "step": 14370 }, { "epoch": 85.08875739644971, "grad_norm": 0.07587834447622299, "learning_rate": 0.00016387767950063153, "loss": 0.0036, "step": 14380 }, { "epoch": 85.14792899408285, "grad_norm": 0.07439226657152176, "learning_rate": 0.0001637186147914138, "loss": 0.004, "step": 14390 }, { "epoch": 85.20710059171597, "grad_norm": 0.059176117181777954, "learning_rate": 0.0001635595345237602, "loss": 0.0036, "step": 14400 }, { "epoch": 85.26627218934911, "grad_norm": 0.0659589096903801, "learning_rate": 0.00016340043887808547, "loss": 0.0037, "step": 14410 }, { "epoch": 85.32544378698225, "grad_norm": 0.07196737825870514, "learning_rate": 0.00016324132803482178, "loss": 0.0032, "step": 14420 }, { "epoch": 85.38461538461539, "grad_norm": 0.067389577627182, "learning_rate": 0.0001630822021744185, "loss": 0.0033, "step": 14430 }, { "epoch": 85.44378698224853, "grad_norm": 0.07085523754358292, "learning_rate": 0.0001629230614773421, "loss": 0.0042, "step": 14440 }, { "epoch": 85.50295857988165, "grad_norm": 0.06972184777259827, "learning_rate": 0.0001627639061240758, "loss": 0.0038, "step": 14450 }, { "epoch": 85.56213017751479, "grad_norm": 0.10725734382867813, "learning_rate": 0.00016260473629511952, "loss": 0.004, "step": 14460 }, { "epoch": 85.62130177514793, "grad_norm": 0.0789913684129715, "learning_rate": 0.0001624455521709896, "loss": 0.0037, "step": 14470 }, { "epoch": 85.68047337278107, "grad_norm": 0.10825314372777939, "learning_rate": 0.0001622863539322184, "loss": 0.0041, "step": 14480 }, { "epoch": 85.7396449704142, "grad_norm": 0.10939929634332657, "learning_rate": 0.0001621271417593546, "loss": 0.0041, "step": 14490 }, { "epoch": 85.79881656804734, "grad_norm": 0.0923149436712265, "learning_rate": 0.00016196791583296245, "loss": 0.0038, "step": 14500 }, { "epoch": 85.85798816568047, "grad_norm": 0.07777465134859085, "learning_rate": 0.00016180867633362189, "loss": 0.0032, "step": 14510 }, { "epoch": 85.9171597633136, "grad_norm": 0.09768331050872803, "learning_rate": 0.0001616494234419282, "loss": 0.0034, "step": 14520 }, { "epoch": 85.97633136094674, "grad_norm": 0.09837043285369873, "learning_rate": 0.00016149015733849195, "loss": 0.0041, "step": 14530 }, { "epoch": 86.03550295857988, "grad_norm": 0.08500006049871445, "learning_rate": 0.00016133087820393853, "loss": 0.0033, "step": 14540 }, { "epoch": 86.09467455621302, "grad_norm": 0.08054184913635254, "learning_rate": 0.00016117158621890832, "loss": 0.004, "step": 14550 }, { "epoch": 86.15384615384616, "grad_norm": 0.047334302216768265, "learning_rate": 0.00016101228156405604, "loss": 0.0031, "step": 14560 }, { "epoch": 86.21301775147928, "grad_norm": 0.11555678397417068, "learning_rate": 0.00016085296442005096, "loss": 0.0037, "step": 14570 }, { "epoch": 86.27218934911242, "grad_norm": 0.07728418707847595, "learning_rate": 0.00016069363496757646, "loss": 0.0037, "step": 14580 }, { "epoch": 86.33136094674556, "grad_norm": 0.08080079406499863, "learning_rate": 0.0001605342933873298, "loss": 0.0037, "step": 14590 }, { "epoch": 86.3905325443787, "grad_norm": 0.06410390138626099, "learning_rate": 0.0001603749398600222, "loss": 0.0036, "step": 14600 }, { "epoch": 86.44970414201184, "grad_norm": 0.07220210880041122, "learning_rate": 0.00016021557456637816, "loss": 0.0033, "step": 14610 }, { "epoch": 86.50887573964496, "grad_norm": 0.06744199991226196, "learning_rate": 0.0001600561976871357, "loss": 0.0035, "step": 14620 }, { "epoch": 86.5680473372781, "grad_norm": 0.07012992352247238, "learning_rate": 0.00015989680940304603, "loss": 0.0035, "step": 14630 }, { "epoch": 86.62721893491124, "grad_norm": 0.08255230635404587, "learning_rate": 0.00015973740989487303, "loss": 0.003, "step": 14640 }, { "epoch": 86.68639053254438, "grad_norm": 0.07717838138341904, "learning_rate": 0.0001595779993433936, "loss": 0.0031, "step": 14650 }, { "epoch": 86.74556213017752, "grad_norm": 0.08995717018842697, "learning_rate": 0.000159418577929397, "loss": 0.0037, "step": 14660 }, { "epoch": 86.80473372781066, "grad_norm": 0.08537930250167847, "learning_rate": 0.0001592591458336849, "loss": 0.003, "step": 14670 }, { "epoch": 86.86390532544378, "grad_norm": 0.08255050331354141, "learning_rate": 0.00015909970323707095, "loss": 0.0039, "step": 14680 }, { "epoch": 86.92307692307692, "grad_norm": 0.0879998430609703, "learning_rate": 0.0001589402503203809, "loss": 0.0039, "step": 14690 }, { "epoch": 86.98224852071006, "grad_norm": 0.08746574819087982, "learning_rate": 0.00015878078726445206, "loss": 0.0034, "step": 14700 }, { "epoch": 87.0414201183432, "grad_norm": 0.07903344184160233, "learning_rate": 0.00015862131425013328, "loss": 0.0032, "step": 14710 }, { "epoch": 87.10059171597634, "grad_norm": 0.08910506218671799, "learning_rate": 0.00015846183145828472, "loss": 0.0036, "step": 14720 }, { "epoch": 87.15976331360947, "grad_norm": 0.08841601759195328, "learning_rate": 0.0001583023390697776, "loss": 0.0035, "step": 14730 }, { "epoch": 87.2189349112426, "grad_norm": 0.06852956861257553, "learning_rate": 0.000158142837265494, "loss": 0.0032, "step": 14740 }, { "epoch": 87.27810650887574, "grad_norm": 0.06677240878343582, "learning_rate": 0.0001579833262263268, "loss": 0.0035, "step": 14750 }, { "epoch": 87.33727810650888, "grad_norm": 0.08654573559761047, "learning_rate": 0.00015782380613317922, "loss": 0.0034, "step": 14760 }, { "epoch": 87.39644970414201, "grad_norm": 0.07378820329904556, "learning_rate": 0.00015766427716696477, "loss": 0.004, "step": 14770 }, { "epoch": 87.45562130177515, "grad_norm": 0.07043015211820602, "learning_rate": 0.00015750473950860712, "loss": 0.0034, "step": 14780 }, { "epoch": 87.51479289940828, "grad_norm": 0.0662299171090126, "learning_rate": 0.00015734519333903966, "loss": 0.0032, "step": 14790 }, { "epoch": 87.57396449704142, "grad_norm": 0.059803176671266556, "learning_rate": 0.00015718563883920553, "loss": 0.0032, "step": 14800 }, { "epoch": 87.63313609467455, "grad_norm": 0.07029089331626892, "learning_rate": 0.00015702607619005735, "loss": 0.0034, "step": 14810 }, { "epoch": 87.6923076923077, "grad_norm": 0.0549018532037735, "learning_rate": 0.00015686650557255683, "loss": 0.0035, "step": 14820 }, { "epoch": 87.75147928994083, "grad_norm": 0.06415469199419022, "learning_rate": 0.00015670692716767483, "loss": 0.0029, "step": 14830 }, { "epoch": 87.81065088757397, "grad_norm": 0.05397758632898331, "learning_rate": 0.00015654734115639105, "loss": 0.0035, "step": 14840 }, { "epoch": 87.8698224852071, "grad_norm": 0.0704231709241867, "learning_rate": 0.00015638774771969377, "loss": 0.0036, "step": 14850 }, { "epoch": 87.92899408284023, "grad_norm": 0.0580257810652256, "learning_rate": 0.0001562281470385797, "loss": 0.0035, "step": 14860 }, { "epoch": 87.98816568047337, "grad_norm": 0.06937891244888306, "learning_rate": 0.0001560685392940538, "loss": 0.0032, "step": 14870 }, { "epoch": 88.04733727810651, "grad_norm": 0.08922914415597916, "learning_rate": 0.00015590892466712898, "loss": 0.0032, "step": 14880 }, { "epoch": 88.10650887573965, "grad_norm": 0.0695345476269722, "learning_rate": 0.000155749303338826, "loss": 0.0032, "step": 14890 }, { "epoch": 88.16568047337279, "grad_norm": 0.06459105014801025, "learning_rate": 0.00015558967549017319, "loss": 0.0031, "step": 14900 }, { "epoch": 88.22485207100591, "grad_norm": 0.08302813023328781, "learning_rate": 0.00015543004130220633, "loss": 0.0038, "step": 14910 }, { "epoch": 88.28402366863905, "grad_norm": 0.08772355318069458, "learning_rate": 0.00015527040095596843, "loss": 0.0027, "step": 14920 }, { "epoch": 88.34319526627219, "grad_norm": 0.07848253101110458, "learning_rate": 0.00015511075463250923, "loss": 0.0033, "step": 14930 }, { "epoch": 88.40236686390533, "grad_norm": 0.06596899777650833, "learning_rate": 0.00015495110251288556, "loss": 0.0041, "step": 14940 }, { "epoch": 88.46153846153847, "grad_norm": 0.07627546042203903, "learning_rate": 0.00015479144477816066, "loss": 0.0038, "step": 14950 }, { "epoch": 88.52071005917159, "grad_norm": 0.06927487999200821, "learning_rate": 0.00015463178160940412, "loss": 0.0033, "step": 14960 }, { "epoch": 88.57988165680473, "grad_norm": 0.06320945173501968, "learning_rate": 0.00015447211318769178, "loss": 0.0032, "step": 14970 }, { "epoch": 88.63905325443787, "grad_norm": 0.06300037354230881, "learning_rate": 0.0001543124396941055, "loss": 0.0027, "step": 14980 }, { "epoch": 88.69822485207101, "grad_norm": 0.05382521450519562, "learning_rate": 0.00015415276130973255, "loss": 0.0035, "step": 14990 }, { "epoch": 88.75739644970415, "grad_norm": 0.0833495706319809, "learning_rate": 0.0001539930782156662, "loss": 0.0041, "step": 15000 }, { "epoch": 88.81656804733728, "grad_norm": 0.10417543351650238, "learning_rate": 0.00015383339059300475, "loss": 0.0038, "step": 15010 }, { "epoch": 88.87573964497041, "grad_norm": 0.11443319171667099, "learning_rate": 0.0001536736986228518, "loss": 0.0036, "step": 15020 }, { "epoch": 88.93491124260355, "grad_norm": 0.10413292050361633, "learning_rate": 0.00015351400248631578, "loss": 0.0035, "step": 15030 }, { "epoch": 88.99408284023669, "grad_norm": 0.09794535487890244, "learning_rate": 0.00015335430236450986, "loss": 0.0031, "step": 15040 }, { "epoch": 89.05325443786982, "grad_norm": 0.07894806563854218, "learning_rate": 0.00015319459843855178, "loss": 0.0034, "step": 15050 }, { "epoch": 89.11242603550296, "grad_norm": 0.05350076034665108, "learning_rate": 0.00015303489088956356, "loss": 0.0032, "step": 15060 }, { "epoch": 89.17159763313609, "grad_norm": 0.06892331689596176, "learning_rate": 0.00015287517989867133, "loss": 0.0034, "step": 15070 }, { "epoch": 89.23076923076923, "grad_norm": 0.08888150006532669, "learning_rate": 0.00015271546564700513, "loss": 0.0031, "step": 15080 }, { "epoch": 89.28994082840237, "grad_norm": 0.10927338898181915, "learning_rate": 0.00015255574831569868, "loss": 0.003, "step": 15090 }, { "epoch": 89.3491124260355, "grad_norm": 0.11526995897293091, "learning_rate": 0.0001523960280858892, "loss": 0.0033, "step": 15100 }, { "epoch": 89.40828402366864, "grad_norm": 0.07279419898986816, "learning_rate": 0.00015223630513871728, "loss": 0.003, "step": 15110 }, { "epoch": 89.46745562130178, "grad_norm": 0.08035297691822052, "learning_rate": 0.0001520765796553264, "loss": 0.0039, "step": 15120 }, { "epoch": 89.5266272189349, "grad_norm": 0.08346544206142426, "learning_rate": 0.00015191685181686315, "loss": 0.004, "step": 15130 }, { "epoch": 89.58579881656804, "grad_norm": 0.08961953222751617, "learning_rate": 0.0001517571218044766, "loss": 0.0041, "step": 15140 }, { "epoch": 89.64497041420118, "grad_norm": 0.08738288283348083, "learning_rate": 0.00015159738979931837, "loss": 0.004, "step": 15150 }, { "epoch": 89.70414201183432, "grad_norm": 0.08520068228244781, "learning_rate": 0.00015143765598254235, "loss": 0.0036, "step": 15160 }, { "epoch": 89.76331360946746, "grad_norm": 0.08609659969806671, "learning_rate": 0.00015127792053530449, "loss": 0.0034, "step": 15170 }, { "epoch": 89.8224852071006, "grad_norm": 0.06341042369604111, "learning_rate": 0.00015111818363876248, "loss": 0.0035, "step": 15180 }, { "epoch": 89.88165680473372, "grad_norm": 0.04961755499243736, "learning_rate": 0.0001509584454740758, "loss": 0.003, "step": 15190 }, { "epoch": 89.94082840236686, "grad_norm": 0.06581571698188782, "learning_rate": 0.00015079870622240524, "loss": 0.0032, "step": 15200 }, { "epoch": 90.0, "grad_norm": 0.09232275933027267, "learning_rate": 0.00015063896606491305, "loss": 0.0039, "step": 15210 }, { "epoch": 90.05917159763314, "grad_norm": 0.11049706488847733, "learning_rate": 0.00015047922518276213, "loss": 0.0032, "step": 15220 }, { "epoch": 90.11834319526628, "grad_norm": 0.0976155698299408, "learning_rate": 0.00015031948375711655, "loss": 0.0032, "step": 15230 }, { "epoch": 90.1775147928994, "grad_norm": 0.06418026983737946, "learning_rate": 0.00015015974196914088, "loss": 0.003, "step": 15240 }, { "epoch": 90.23668639053254, "grad_norm": 0.06961499154567719, "learning_rate": 0.00015, "loss": 0.003, "step": 15250 }, { "epoch": 90.29585798816568, "grad_norm": 0.09508829563856125, "learning_rate": 0.00014984025803085912, "loss": 0.0034, "step": 15260 }, { "epoch": 90.35502958579882, "grad_norm": 0.07861330360174179, "learning_rate": 0.00014968051624288342, "loss": 0.0037, "step": 15270 }, { "epoch": 90.41420118343196, "grad_norm": 0.07645835727453232, "learning_rate": 0.00014952077481723787, "loss": 0.0034, "step": 15280 }, { "epoch": 90.4733727810651, "grad_norm": 0.11310037970542908, "learning_rate": 0.00014936103393508698, "loss": 0.0038, "step": 15290 }, { "epoch": 90.53254437869822, "grad_norm": 0.08834660053253174, "learning_rate": 0.00014920129377759474, "loss": 0.0035, "step": 15300 }, { "epoch": 90.59171597633136, "grad_norm": 0.06680396944284439, "learning_rate": 0.00014904155452592424, "loss": 0.0037, "step": 15310 }, { "epoch": 90.6508875739645, "grad_norm": 0.07265255600214005, "learning_rate": 0.00014888181636123752, "loss": 0.0038, "step": 15320 }, { "epoch": 90.71005917159763, "grad_norm": 0.0713018998503685, "learning_rate": 0.00014872207946469554, "loss": 0.0035, "step": 15330 }, { "epoch": 90.76923076923077, "grad_norm": 0.0424182303249836, "learning_rate": 0.00014856234401745765, "loss": 0.0035, "step": 15340 }, { "epoch": 90.82840236686391, "grad_norm": 0.07963784039020538, "learning_rate": 0.00014840261020068166, "loss": 0.0034, "step": 15350 }, { "epoch": 90.88757396449704, "grad_norm": 0.09492604434490204, "learning_rate": 0.00014824287819552336, "loss": 0.0031, "step": 15360 }, { "epoch": 90.94674556213018, "grad_norm": 0.08144377171993256, "learning_rate": 0.00014808314818313682, "loss": 0.0031, "step": 15370 }, { "epoch": 91.00591715976331, "grad_norm": 0.04987722262740135, "learning_rate": 0.00014792342034467356, "loss": 0.0031, "step": 15380 }, { "epoch": 91.06508875739645, "grad_norm": 0.07695046067237854, "learning_rate": 0.0001477636948612827, "loss": 0.0036, "step": 15390 }, { "epoch": 91.12426035502959, "grad_norm": 0.05834494158625603, "learning_rate": 0.00014760397191411077, "loss": 0.003, "step": 15400 }, { "epoch": 91.18343195266272, "grad_norm": 0.06730993092060089, "learning_rate": 0.0001474442516843013, "loss": 0.003, "step": 15410 }, { "epoch": 91.24260355029585, "grad_norm": 0.06383945047855377, "learning_rate": 0.00014728453435299484, "loss": 0.0033, "step": 15420 }, { "epoch": 91.30177514792899, "grad_norm": 0.07671685516834259, "learning_rate": 0.00014712482010132867, "loss": 0.0031, "step": 15430 }, { "epoch": 91.36094674556213, "grad_norm": 0.06165686622262001, "learning_rate": 0.0001469651091104364, "loss": 0.0032, "step": 15440 }, { "epoch": 91.42011834319527, "grad_norm": 0.0775546208024025, "learning_rate": 0.00014680540156144823, "loss": 0.003, "step": 15450 }, { "epoch": 91.47928994082841, "grad_norm": 0.057458553463220596, "learning_rate": 0.0001466456976354901, "loss": 0.0034, "step": 15460 }, { "epoch": 91.53846153846153, "grad_norm": 0.054757121950387955, "learning_rate": 0.0001464859975136842, "loss": 0.003, "step": 15470 }, { "epoch": 91.59763313609467, "grad_norm": 0.07618381083011627, "learning_rate": 0.00014632630137714818, "loss": 0.0038, "step": 15480 }, { "epoch": 91.65680473372781, "grad_norm": 0.08369701355695724, "learning_rate": 0.00014616660940699523, "loss": 0.0032, "step": 15490 }, { "epoch": 91.71597633136095, "grad_norm": 0.10760829597711563, "learning_rate": 0.00014600692178433377, "loss": 0.0033, "step": 15500 }, { "epoch": 91.77514792899409, "grad_norm": 0.07359318435192108, "learning_rate": 0.00014584723869026745, "loss": 0.0035, "step": 15510 }, { "epoch": 91.83431952662721, "grad_norm": 0.10211027413606644, "learning_rate": 0.00014568756030589457, "loss": 0.0033, "step": 15520 }, { "epoch": 91.89349112426035, "grad_norm": 0.11060106754302979, "learning_rate": 0.0001455278868123082, "loss": 0.0033, "step": 15530 }, { "epoch": 91.95266272189349, "grad_norm": 0.10445372015237808, "learning_rate": 0.00014536821839059589, "loss": 0.0031, "step": 15540 }, { "epoch": 92.01183431952663, "grad_norm": 0.09539007395505905, "learning_rate": 0.00014520855522183937, "loss": 0.0032, "step": 15550 }, { "epoch": 92.07100591715977, "grad_norm": 0.0763893648982048, "learning_rate": 0.00014504889748711444, "loss": 0.0029, "step": 15560 }, { "epoch": 92.1301775147929, "grad_norm": 0.04711667820811272, "learning_rate": 0.00014488924536749072, "loss": 0.0034, "step": 15570 }, { "epoch": 92.18934911242603, "grad_norm": 0.07404505461454391, "learning_rate": 0.00014472959904403157, "loss": 0.0029, "step": 15580 }, { "epoch": 92.24852071005917, "grad_norm": 0.07405752688646317, "learning_rate": 0.0001445699586977936, "loss": 0.0029, "step": 15590 }, { "epoch": 92.3076923076923, "grad_norm": 0.05947965383529663, "learning_rate": 0.00014441032450982676, "loss": 0.0031, "step": 15600 }, { "epoch": 92.36686390532545, "grad_norm": 0.05913503095507622, "learning_rate": 0.00014425069666117402, "loss": 0.003, "step": 15610 }, { "epoch": 92.42603550295858, "grad_norm": 0.04918340966105461, "learning_rate": 0.00014409107533287102, "loss": 0.0029, "step": 15620 }, { "epoch": 92.48520710059172, "grad_norm": 0.05653808265924454, "learning_rate": 0.00014393146070594619, "loss": 0.0032, "step": 15630 }, { "epoch": 92.54437869822485, "grad_norm": 0.05068068206310272, "learning_rate": 0.0001437718529614203, "loss": 0.003, "step": 15640 }, { "epoch": 92.60355029585799, "grad_norm": 0.12025908380746841, "learning_rate": 0.0001436122522803062, "loss": 0.0033, "step": 15650 }, { "epoch": 92.66272189349112, "grad_norm": 0.06658908724784851, "learning_rate": 0.00014345265884360892, "loss": 0.0033, "step": 15660 }, { "epoch": 92.72189349112426, "grad_norm": 0.06315654516220093, "learning_rate": 0.00014329307283232517, "loss": 0.0031, "step": 15670 }, { "epoch": 92.7810650887574, "grad_norm": 0.05144256725907326, "learning_rate": 0.00014313349442744317, "loss": 0.0034, "step": 15680 }, { "epoch": 92.84023668639053, "grad_norm": 0.07033531367778778, "learning_rate": 0.00014297392380994265, "loss": 0.0031, "step": 15690 }, { "epoch": 92.89940828402366, "grad_norm": 0.06715501844882965, "learning_rate": 0.00014281436116079447, "loss": 0.0035, "step": 15700 }, { "epoch": 92.9585798816568, "grad_norm": 0.07233394682407379, "learning_rate": 0.00014265480666096034, "loss": 0.0029, "step": 15710 }, { "epoch": 93.01775147928994, "grad_norm": 0.09698658436536789, "learning_rate": 0.00014249526049139288, "loss": 0.0038, "step": 15720 }, { "epoch": 93.07692307692308, "grad_norm": 0.08217937499284744, "learning_rate": 0.00014233572283303523, "loss": 0.003, "step": 15730 }, { "epoch": 93.13609467455622, "grad_norm": 0.10114498436450958, "learning_rate": 0.0001421761938668208, "loss": 0.0035, "step": 15740 }, { "epoch": 93.19526627218934, "grad_norm": 0.06852945685386658, "learning_rate": 0.00014201667377367324, "loss": 0.0032, "step": 15750 }, { "epoch": 93.25443786982248, "grad_norm": 0.08286581188440323, "learning_rate": 0.000141857162734506, "loss": 0.0039, "step": 15760 }, { "epoch": 93.31360946745562, "grad_norm": 0.08890713006258011, "learning_rate": 0.00014169766093022237, "loss": 0.0033, "step": 15770 }, { "epoch": 93.37278106508876, "grad_norm": 0.08296231925487518, "learning_rate": 0.00014153816854171525, "loss": 0.0034, "step": 15780 }, { "epoch": 93.4319526627219, "grad_norm": 0.08223425596952438, "learning_rate": 0.00014137868574986667, "loss": 0.0035, "step": 15790 }, { "epoch": 93.49112426035504, "grad_norm": 0.056878525763750076, "learning_rate": 0.00014121921273554792, "loss": 0.0031, "step": 15800 }, { "epoch": 93.55029585798816, "grad_norm": 0.06905633956193924, "learning_rate": 0.00014105974967961905, "loss": 0.0033, "step": 15810 }, { "epoch": 93.6094674556213, "grad_norm": 0.04844438657164574, "learning_rate": 0.000140900296762929, "loss": 0.0031, "step": 15820 }, { "epoch": 93.66863905325444, "grad_norm": 0.09498096257448196, "learning_rate": 0.0001407408541663151, "loss": 0.0036, "step": 15830 }, { "epoch": 93.72781065088758, "grad_norm": 0.06686652451753616, "learning_rate": 0.000140581422070603, "loss": 0.0035, "step": 15840 }, { "epoch": 93.78698224852072, "grad_norm": 0.07518838346004486, "learning_rate": 0.0001404220006566064, "loss": 0.0033, "step": 15850 }, { "epoch": 93.84615384615384, "grad_norm": 0.08922500908374786, "learning_rate": 0.00014026259010512697, "loss": 0.0031, "step": 15860 }, { "epoch": 93.90532544378698, "grad_norm": 0.06725657731294632, "learning_rate": 0.00014010319059695397, "loss": 0.0033, "step": 15870 }, { "epoch": 93.96449704142012, "grad_norm": 0.06253711879253387, "learning_rate": 0.00013994380231286427, "loss": 0.0029, "step": 15880 }, { "epoch": 94.02366863905326, "grad_norm": 0.08374454826116562, "learning_rate": 0.00013978442543362182, "loss": 0.0032, "step": 15890 }, { "epoch": 94.0828402366864, "grad_norm": 0.07125059515237808, "learning_rate": 0.0001396250601399778, "loss": 0.0031, "step": 15900 }, { "epoch": 94.14201183431953, "grad_norm": 0.05445906147360802, "learning_rate": 0.00013946570661267017, "loss": 0.0034, "step": 15910 }, { "epoch": 94.20118343195266, "grad_norm": 0.08904863148927689, "learning_rate": 0.00013930636503242354, "loss": 0.0032, "step": 15920 }, { "epoch": 94.2603550295858, "grad_norm": 0.06531428545713425, "learning_rate": 0.000139147035579949, "loss": 0.0033, "step": 15930 }, { "epoch": 94.31952662721893, "grad_norm": 0.09573723375797272, "learning_rate": 0.00013898771843594399, "loss": 0.0032, "step": 15940 }, { "epoch": 94.37869822485207, "grad_norm": 0.07509324699640274, "learning_rate": 0.0001388284137810917, "loss": 0.0044, "step": 15950 }, { "epoch": 94.43786982248521, "grad_norm": 0.06253281980752945, "learning_rate": 0.00013866912179606145, "loss": 0.0031, "step": 15960 }, { "epoch": 94.49704142011835, "grad_norm": 0.0771140307188034, "learning_rate": 0.00013850984266150807, "loss": 0.0031, "step": 15970 }, { "epoch": 94.55621301775147, "grad_norm": 0.06187194958329201, "learning_rate": 0.00013835057655807177, "loss": 0.0031, "step": 15980 }, { "epoch": 94.61538461538461, "grad_norm": 0.08055388182401657, "learning_rate": 0.0001381913236663781, "loss": 0.0032, "step": 15990 }, { "epoch": 94.67455621301775, "grad_norm": 0.06004798412322998, "learning_rate": 0.0001380320841670375, "loss": 0.0036, "step": 16000 }, { "epoch": 94.73372781065089, "grad_norm": 0.058052510023117065, "learning_rate": 0.0001378728582406454, "loss": 0.003, "step": 16010 }, { "epoch": 94.79289940828403, "grad_norm": 0.10025151073932648, "learning_rate": 0.00013771364606778158, "loss": 0.0032, "step": 16020 }, { "epoch": 94.85207100591715, "grad_norm": 0.0729788988828659, "learning_rate": 0.0001375544478290104, "loss": 0.0033, "step": 16030 }, { "epoch": 94.91124260355029, "grad_norm": 0.10715433210134506, "learning_rate": 0.00013739526370488045, "loss": 0.0033, "step": 16040 }, { "epoch": 94.97041420118343, "grad_norm": 0.09647718816995621, "learning_rate": 0.00013723609387592418, "loss": 0.0031, "step": 16050 }, { "epoch": 95.02958579881657, "grad_norm": 0.06684408336877823, "learning_rate": 0.0001370769385226579, "loss": 0.0032, "step": 16060 }, { "epoch": 95.08875739644971, "grad_norm": 0.04664446786046028, "learning_rate": 0.0001369177978255815, "loss": 0.0028, "step": 16070 }, { "epoch": 95.14792899408285, "grad_norm": 0.05208440497517586, "learning_rate": 0.00013675867196517822, "loss": 0.0032, "step": 16080 }, { "epoch": 95.20710059171597, "grad_norm": 0.08110474050045013, "learning_rate": 0.0001365995611219145, "loss": 0.0031, "step": 16090 }, { "epoch": 95.26627218934911, "grad_norm": 0.06770195066928864, "learning_rate": 0.0001364404654762398, "loss": 0.0031, "step": 16100 }, { "epoch": 95.32544378698225, "grad_norm": 0.068801648914814, "learning_rate": 0.0001362813852085862, "loss": 0.0032, "step": 16110 }, { "epoch": 95.38461538461539, "grad_norm": 0.06413090974092484, "learning_rate": 0.00013612232049936844, "loss": 0.0028, "step": 16120 }, { "epoch": 95.44378698224853, "grad_norm": 0.0517580471932888, "learning_rate": 0.0001359632715289837, "loss": 0.0028, "step": 16130 }, { "epoch": 95.50295857988165, "grad_norm": 0.05974963679909706, "learning_rate": 0.00013580423847781107, "loss": 0.0032, "step": 16140 }, { "epoch": 95.56213017751479, "grad_norm": 0.08221578598022461, "learning_rate": 0.00013564522152621186, "loss": 0.0031, "step": 16150 }, { "epoch": 95.62130177514793, "grad_norm": 0.0999283716082573, "learning_rate": 0.0001354862208545289, "loss": 0.003, "step": 16160 }, { "epoch": 95.68047337278107, "grad_norm": 0.10367809236049652, "learning_rate": 0.00013532723664308664, "loss": 0.0033, "step": 16170 }, { "epoch": 95.7396449704142, "grad_norm": 0.08125555515289307, "learning_rate": 0.00013516826907219098, "loss": 0.0035, "step": 16180 }, { "epoch": 95.79881656804734, "grad_norm": 0.07147395610809326, "learning_rate": 0.00013500931832212862, "loss": 0.0033, "step": 16190 }, { "epoch": 95.85798816568047, "grad_norm": 0.07763400673866272, "learning_rate": 0.00013485038457316766, "loss": 0.0031, "step": 16200 }, { "epoch": 95.9171597633136, "grad_norm": 0.06180090084671974, "learning_rate": 0.0001346914680055565, "loss": 0.0027, "step": 16210 }, { "epoch": 95.97633136094674, "grad_norm": 0.05595600977540016, "learning_rate": 0.00013453256879952425, "loss": 0.0037, "step": 16220 }, { "epoch": 96.03550295857988, "grad_norm": 0.08516224473714828, "learning_rate": 0.00013437368713528034, "loss": 0.003, "step": 16230 }, { "epoch": 96.09467455621302, "grad_norm": 0.056127242743968964, "learning_rate": 0.00013421482319301415, "loss": 0.0032, "step": 16240 }, { "epoch": 96.15384615384616, "grad_norm": 0.05663565918803215, "learning_rate": 0.0001340559771528952, "loss": 0.0031, "step": 16250 }, { "epoch": 96.21301775147928, "grad_norm": 0.07301441580057144, "learning_rate": 0.00013389714919507246, "loss": 0.0031, "step": 16260 }, { "epoch": 96.27218934911242, "grad_norm": 0.051993269473314285, "learning_rate": 0.00013373833949967455, "loss": 0.0027, "step": 16270 }, { "epoch": 96.33136094674556, "grad_norm": 0.048337094485759735, "learning_rate": 0.00013357954824680938, "loss": 0.0034, "step": 16280 }, { "epoch": 96.3905325443787, "grad_norm": 0.08531758189201355, "learning_rate": 0.00013342077561656385, "loss": 0.0031, "step": 16290 }, { "epoch": 96.44970414201184, "grad_norm": 0.06239122152328491, "learning_rate": 0.00013326202178900377, "loss": 0.0029, "step": 16300 }, { "epoch": 96.50887573964496, "grad_norm": 0.04881719499826431, "learning_rate": 0.00013310328694417375, "loss": 0.003, "step": 16310 }, { "epoch": 96.5680473372781, "grad_norm": 0.0538143515586853, "learning_rate": 0.0001329445712620967, "loss": 0.0029, "step": 16320 }, { "epoch": 96.62721893491124, "grad_norm": 0.10291869938373566, "learning_rate": 0.00013278587492277383, "loss": 0.003, "step": 16330 }, { "epoch": 96.68639053254438, "grad_norm": 0.057994112372398376, "learning_rate": 0.00013262719810618455, "loss": 0.0027, "step": 16340 }, { "epoch": 96.74556213017752, "grad_norm": 0.0838884711265564, "learning_rate": 0.00013246854099228591, "loss": 0.0033, "step": 16350 }, { "epoch": 96.80473372781066, "grad_norm": 0.07527537643909454, "learning_rate": 0.00013230990376101283, "loss": 0.0037, "step": 16360 }, { "epoch": 96.86390532544378, "grad_norm": 0.07233381271362305, "learning_rate": 0.00013215128659227758, "loss": 0.0038, "step": 16370 }, { "epoch": 96.92307692307692, "grad_norm": 0.0833866149187088, "learning_rate": 0.0001319926896659696, "loss": 0.0032, "step": 16380 }, { "epoch": 96.98224852071006, "grad_norm": 0.055972445756196976, "learning_rate": 0.0001318341131619555, "loss": 0.0035, "step": 16390 }, { "epoch": 97.0414201183432, "grad_norm": 0.06956689059734344, "learning_rate": 0.00013167555726007865, "loss": 0.0034, "step": 16400 }, { "epoch": 97.10059171597634, "grad_norm": 0.06966190040111542, "learning_rate": 0.00013151702214015922, "loss": 0.0029, "step": 16410 }, { "epoch": 97.15976331360947, "grad_norm": 0.0627070888876915, "learning_rate": 0.00013135850798199353, "loss": 0.0037, "step": 16420 }, { "epoch": 97.2189349112426, "grad_norm": 0.09476328641176224, "learning_rate": 0.00013120001496535433, "loss": 0.0037, "step": 16430 }, { "epoch": 97.27810650887574, "grad_norm": 0.07222788035869598, "learning_rate": 0.0001310415432699904, "loss": 0.0037, "step": 16440 }, { "epoch": 97.33727810650888, "grad_norm": 0.0784858986735344, "learning_rate": 0.00013088309307562613, "loss": 0.0028, "step": 16450 }, { "epoch": 97.39644970414201, "grad_norm": 0.07490145415067673, "learning_rate": 0.0001307246645619618, "loss": 0.0031, "step": 16460 }, { "epoch": 97.45562130177515, "grad_norm": 0.07202551513910294, "learning_rate": 0.00013056625790867294, "loss": 0.0031, "step": 16470 }, { "epoch": 97.51479289940828, "grad_norm": 0.0696314349770546, "learning_rate": 0.00013040787329541032, "loss": 0.0039, "step": 16480 }, { "epoch": 97.57396449704142, "grad_norm": 0.10154827684164047, "learning_rate": 0.00013024951090179969, "loss": 0.0036, "step": 16490 }, { "epoch": 97.63313609467455, "grad_norm": 0.11780398339033127, "learning_rate": 0.0001300911709074417, "loss": 0.0032, "step": 16500 }, { "epoch": 97.6923076923077, "grad_norm": 0.08720283210277557, "learning_rate": 0.00012993285349191145, "loss": 0.0032, "step": 16510 }, { "epoch": 97.75147928994083, "grad_norm": 0.09580790996551514, "learning_rate": 0.00012977455883475853, "loss": 0.003, "step": 16520 }, { "epoch": 97.81065088757397, "grad_norm": 0.07841911166906357, "learning_rate": 0.00012961628711550676, "loss": 0.0031, "step": 16530 }, { "epoch": 97.8698224852071, "grad_norm": 0.08514852076768875, "learning_rate": 0.00012945803851365377, "loss": 0.003, "step": 16540 }, { "epoch": 97.92899408284023, "grad_norm": 0.08114304393529892, "learning_rate": 0.00012929981320867117, "loss": 0.0029, "step": 16550 }, { "epoch": 97.98816568047337, "grad_norm": 0.06607823073863983, "learning_rate": 0.000129141611380004, "loss": 0.0039, "step": 16560 }, { "epoch": 98.04733727810651, "grad_norm": 0.07988603413105011, "learning_rate": 0.0001289834332070708, "loss": 0.0032, "step": 16570 }, { "epoch": 98.10650887573965, "grad_norm": 0.05780590698122978, "learning_rate": 0.0001288252788692632, "loss": 0.0029, "step": 16580 }, { "epoch": 98.16568047337279, "grad_norm": 0.08401712030172348, "learning_rate": 0.00012866714854594584, "loss": 0.0035, "step": 16590 }, { "epoch": 98.22485207100591, "grad_norm": 0.05960412696003914, "learning_rate": 0.00012850904241645602, "loss": 0.0032, "step": 16600 }, { "epoch": 98.28402366863905, "grad_norm": 0.045081958174705505, "learning_rate": 0.00012835096066010377, "loss": 0.0028, "step": 16610 }, { "epoch": 98.34319526627219, "grad_norm": 0.056573785841464996, "learning_rate": 0.00012819290345617135, "loss": 0.0028, "step": 16620 }, { "epoch": 98.40236686390533, "grad_norm": 0.06382756680250168, "learning_rate": 0.0001280348709839133, "loss": 0.0028, "step": 16630 }, { "epoch": 98.46153846153847, "grad_norm": 0.04654393345117569, "learning_rate": 0.00012787686342255594, "loss": 0.0023, "step": 16640 }, { "epoch": 98.52071005917159, "grad_norm": 0.06569322198629379, "learning_rate": 0.00012771888095129748, "loss": 0.0025, "step": 16650 }, { "epoch": 98.57988165680473, "grad_norm": 0.07429029792547226, "learning_rate": 0.00012756092374930757, "loss": 0.003, "step": 16660 }, { "epoch": 98.63905325443787, "grad_norm": 0.06590225547552109, "learning_rate": 0.0001274029919957273, "loss": 0.0031, "step": 16670 }, { "epoch": 98.69822485207101, "grad_norm": 0.07079692929983139, "learning_rate": 0.00012724508586966888, "loss": 0.0032, "step": 16680 }, { "epoch": 98.75739644970415, "grad_norm": 0.0851840153336525, "learning_rate": 0.00012708720555021534, "loss": 0.0031, "step": 16690 }, { "epoch": 98.81656804733728, "grad_norm": 0.09271086752414703, "learning_rate": 0.0001269293512164206, "loss": 0.0033, "step": 16700 }, { "epoch": 98.87573964497041, "grad_norm": 0.052189115434885025, "learning_rate": 0.00012677152304730905, "loss": 0.0031, "step": 16710 }, { "epoch": 98.93491124260355, "grad_norm": 0.050209544599056244, "learning_rate": 0.00012661372122187533, "loss": 0.0028, "step": 16720 }, { "epoch": 98.99408284023669, "grad_norm": 0.06322485208511353, "learning_rate": 0.00012645594591908433, "loss": 0.0029, "step": 16730 }, { "epoch": 99.05325443786982, "grad_norm": 0.06419099867343903, "learning_rate": 0.00012629819731787084, "loss": 0.0029, "step": 16740 }, { "epoch": 99.11242603550296, "grad_norm": 0.0595984049141407, "learning_rate": 0.0001261404755971392, "loss": 0.0033, "step": 16750 }, { "epoch": 99.17159763313609, "grad_norm": 0.056447647511959076, "learning_rate": 0.0001259827809357635, "loss": 0.0031, "step": 16760 }, { "epoch": 99.23076923076923, "grad_norm": 0.034934189170598984, "learning_rate": 0.000125825113512587, "loss": 0.0028, "step": 16770 }, { "epoch": 99.28994082840237, "grad_norm": 0.07702696323394775, "learning_rate": 0.00012566747350642213, "loss": 0.0026, "step": 16780 }, { "epoch": 99.3491124260355, "grad_norm": 0.06824219226837158, "learning_rate": 0.00012550986109605014, "loss": 0.0028, "step": 16790 }, { "epoch": 99.40828402366864, "grad_norm": 0.0638587549328804, "learning_rate": 0.0001253522764602211, "loss": 0.003, "step": 16800 }, { "epoch": 99.46745562130178, "grad_norm": 0.06793996691703796, "learning_rate": 0.0001251947197776535, "loss": 0.0029, "step": 16810 }, { "epoch": 99.5266272189349, "grad_norm": 0.059136439114809036, "learning_rate": 0.00012503719122703414, "loss": 0.0032, "step": 16820 }, { "epoch": 99.58579881656804, "grad_norm": 0.07387898117303848, "learning_rate": 0.00012487969098701795, "loss": 0.0028, "step": 16830 }, { "epoch": 99.64497041420118, "grad_norm": 0.05197450518608093, "learning_rate": 0.00012472221923622777, "loss": 0.0027, "step": 16840 }, { "epoch": 99.70414201183432, "grad_norm": 0.07094729691743851, "learning_rate": 0.00012456477615325402, "loss": 0.0026, "step": 16850 }, { "epoch": 99.76331360946746, "grad_norm": 0.05620044097304344, "learning_rate": 0.0001244073619166547, "loss": 0.0037, "step": 16860 }, { "epoch": 99.8224852071006, "grad_norm": 0.06962645053863525, "learning_rate": 0.0001242499767049551, "loss": 0.0032, "step": 16870 }, { "epoch": 99.88165680473372, "grad_norm": 0.06103204935789108, "learning_rate": 0.0001240926206966475, "loss": 0.0028, "step": 16880 }, { "epoch": 99.94082840236686, "grad_norm": 0.07809354364871979, "learning_rate": 0.00012393529407019117, "loss": 0.0032, "step": 16890 }, { "epoch": 100.0, "grad_norm": 0.0923081561923027, "learning_rate": 0.00012377799700401205, "loss": 0.003, "step": 16900 }, { "epoch": 100.05917159763314, "grad_norm": 0.07746690511703491, "learning_rate": 0.00012362072967650245, "loss": 0.003, "step": 16910 }, { "epoch": 100.11834319526628, "grad_norm": 0.08745256811380386, "learning_rate": 0.000123463492266021, "loss": 0.0033, "step": 16920 }, { "epoch": 100.1775147928994, "grad_norm": 0.052244480699300766, "learning_rate": 0.00012330628495089253, "loss": 0.0033, "step": 16930 }, { "epoch": 100.23668639053254, "grad_norm": 0.04832267761230469, "learning_rate": 0.00012314910790940756, "loss": 0.003, "step": 16940 }, { "epoch": 100.29585798816568, "grad_norm": 0.0965530127286911, "learning_rate": 0.00012299196131982236, "loss": 0.0032, "step": 16950 }, { "epoch": 100.35502958579882, "grad_norm": 0.060458019375801086, "learning_rate": 0.00012283484536035862, "loss": 0.003, "step": 16960 }, { "epoch": 100.41420118343196, "grad_norm": 0.04015009477734566, "learning_rate": 0.00012267776020920333, "loss": 0.0033, "step": 16970 }, { "epoch": 100.4733727810651, "grad_norm": 0.03670589625835419, "learning_rate": 0.0001225207060445086, "loss": 0.0029, "step": 16980 }, { "epoch": 100.53254437869822, "grad_norm": 0.0644877701997757, "learning_rate": 0.00012236368304439124, "loss": 0.0027, "step": 16990 }, { "epoch": 100.59171597633136, "grad_norm": 0.06839296221733093, "learning_rate": 0.00012220669138693287, "loss": 0.0027, "step": 17000 }, { "epoch": 100.6508875739645, "grad_norm": 0.08051011711359024, "learning_rate": 0.00012204973125017949, "loss": 0.0028, "step": 17010 }, { "epoch": 100.71005917159763, "grad_norm": 0.06955340504646301, "learning_rate": 0.00012189280281214126, "loss": 0.0027, "step": 17020 }, { "epoch": 100.76923076923077, "grad_norm": 0.053874481469392776, "learning_rate": 0.0001217359062507927, "loss": 0.0025, "step": 17030 }, { "epoch": 100.82840236686391, "grad_norm": 0.052282918244600296, "learning_rate": 0.0001215790417440718, "loss": 0.0024, "step": 17040 }, { "epoch": 100.88757396449704, "grad_norm": 0.05500173941254616, "learning_rate": 0.00012142220946988046, "loss": 0.0024, "step": 17050 }, { "epoch": 100.94674556213018, "grad_norm": 0.05717858299612999, "learning_rate": 0.00012126540960608387, "loss": 0.0028, "step": 17060 }, { "epoch": 101.00591715976331, "grad_norm": 0.06670291721820831, "learning_rate": 0.00012110864233051056, "loss": 0.0024, "step": 17070 }, { "epoch": 101.06508875739645, "grad_norm": 0.07082294672727585, "learning_rate": 0.00012095190782095208, "loss": 0.0027, "step": 17080 }, { "epoch": 101.12426035502959, "grad_norm": 0.08749767392873764, "learning_rate": 0.00012079520625516274, "loss": 0.0036, "step": 17090 }, { "epoch": 101.18343195266272, "grad_norm": 0.10349269956350327, "learning_rate": 0.00012063853781085961, "loss": 0.0028, "step": 17100 }, { "epoch": 101.24260355029585, "grad_norm": 0.0887615829706192, "learning_rate": 0.00012048190266572216, "loss": 0.0026, "step": 17110 }, { "epoch": 101.30177514792899, "grad_norm": 0.07681963592767715, "learning_rate": 0.000120325300997392, "loss": 0.0029, "step": 17120 }, { "epoch": 101.36094674556213, "grad_norm": 0.08133795112371445, "learning_rate": 0.00012016873298347291, "loss": 0.0031, "step": 17130 }, { "epoch": 101.42011834319527, "grad_norm": 0.06870647519826889, "learning_rate": 0.00012001219880153044, "loss": 0.0027, "step": 17140 }, { "epoch": 101.47928994082841, "grad_norm": 0.0543903224170208, "learning_rate": 0.00011985569862909176, "loss": 0.003, "step": 17150 }, { "epoch": 101.53846153846153, "grad_norm": 0.057152554392814636, "learning_rate": 0.00011969923264364546, "loss": 0.0032, "step": 17160 }, { "epoch": 101.59763313609467, "grad_norm": 0.05013659968972206, "learning_rate": 0.00011954280102264141, "loss": 0.0029, "step": 17170 }, { "epoch": 101.65680473372781, "grad_norm": 0.06736499071121216, "learning_rate": 0.00011938640394349046, "loss": 0.0029, "step": 17180 }, { "epoch": 101.71597633136095, "grad_norm": 0.0791635736823082, "learning_rate": 0.0001192300415835643, "loss": 0.003, "step": 17190 }, { "epoch": 101.77514792899409, "grad_norm": 0.05466848239302635, "learning_rate": 0.0001190737141201953, "loss": 0.0027, "step": 17200 }, { "epoch": 101.83431952662721, "grad_norm": 0.05938201770186424, "learning_rate": 0.00011891742173067611, "loss": 0.0028, "step": 17210 }, { "epoch": 101.89349112426035, "grad_norm": 0.04944825917482376, "learning_rate": 0.0001187611645922598, "loss": 0.003, "step": 17220 }, { "epoch": 101.95266272189349, "grad_norm": 0.07817132771015167, "learning_rate": 0.00011860494288215915, "loss": 0.0027, "step": 17230 }, { "epoch": 102.01183431952663, "grad_norm": 0.08585783839225769, "learning_rate": 0.00011844875677754725, "loss": 0.0031, "step": 17240 }, { "epoch": 102.07100591715977, "grad_norm": 0.0744713768362999, "learning_rate": 0.00011829260645555634, "loss": 0.0029, "step": 17250 }, { "epoch": 102.1301775147929, "grad_norm": 0.06758081912994385, "learning_rate": 0.0001181364920932783, "loss": 0.0028, "step": 17260 }, { "epoch": 102.18934911242603, "grad_norm": 0.062260132282972336, "learning_rate": 0.00011798041386776422, "loss": 0.0028, "step": 17270 }, { "epoch": 102.24852071005917, "grad_norm": 0.08649986982345581, "learning_rate": 0.00011782437195602413, "loss": 0.0027, "step": 17280 }, { "epoch": 102.3076923076923, "grad_norm": 0.06951587647199631, "learning_rate": 0.00011766836653502694, "loss": 0.0026, "step": 17290 }, { "epoch": 102.36686390532545, "grad_norm": 0.08441554009914398, "learning_rate": 0.0001175123977817002, "loss": 0.0028, "step": 17300 }, { "epoch": 102.42603550295858, "grad_norm": 0.0817386582493782, "learning_rate": 0.00011735646587292976, "loss": 0.0033, "step": 17310 }, { "epoch": 102.48520710059172, "grad_norm": 0.07846654951572418, "learning_rate": 0.00011720057098555981, "loss": 0.0031, "step": 17320 }, { "epoch": 102.54437869822485, "grad_norm": 0.07076361775398254, "learning_rate": 0.0001170447132963925, "loss": 0.0027, "step": 17330 }, { "epoch": 102.60355029585799, "grad_norm": 0.057905878871679306, "learning_rate": 0.00011688889298218776, "loss": 0.0028, "step": 17340 }, { "epoch": 102.66272189349112, "grad_norm": 0.052147459238767624, "learning_rate": 0.00011673311021966323, "loss": 0.0025, "step": 17350 }, { "epoch": 102.72189349112426, "grad_norm": 0.07898368686437607, "learning_rate": 0.0001165773651854938, "loss": 0.0026, "step": 17360 }, { "epoch": 102.7810650887574, "grad_norm": 0.088932566344738, "learning_rate": 0.00011642165805631176, "loss": 0.0028, "step": 17370 }, { "epoch": 102.84023668639053, "grad_norm": 0.07817097008228302, "learning_rate": 0.00011626598900870631, "loss": 0.0026, "step": 17380 }, { "epoch": 102.89940828402366, "grad_norm": 0.06759978830814362, "learning_rate": 0.0001161103582192234, "loss": 0.0032, "step": 17390 }, { "epoch": 102.9585798816568, "grad_norm": 0.10146518051624298, "learning_rate": 0.00011595476586436575, "loss": 0.0032, "step": 17400 }, { "epoch": 103.01775147928994, "grad_norm": 0.10470709949731827, "learning_rate": 0.00011579921212059239, "loss": 0.0031, "step": 17410 }, { "epoch": 103.07692307692308, "grad_norm": 0.0961015522480011, "learning_rate": 0.00011564369716431853, "loss": 0.0029, "step": 17420 }, { "epoch": 103.13609467455622, "grad_norm": 0.05426270514726639, "learning_rate": 0.00011548822117191547, "loss": 0.003, "step": 17430 }, { "epoch": 103.19526627218934, "grad_norm": 0.06467846781015396, "learning_rate": 0.00011533278431971027, "loss": 0.0027, "step": 17440 }, { "epoch": 103.25443786982248, "grad_norm": 0.048804111778736115, "learning_rate": 0.00011517738678398568, "loss": 0.0024, "step": 17450 }, { "epoch": 103.31360946745562, "grad_norm": 0.04225113242864609, "learning_rate": 0.00011502202874097973, "loss": 0.0022, "step": 17460 }, { "epoch": 103.37278106508876, "grad_norm": 0.041439589112997055, "learning_rate": 0.00011486671036688575, "loss": 0.0028, "step": 17470 }, { "epoch": 103.4319526627219, "grad_norm": 0.06795652955770493, "learning_rate": 0.00011471143183785209, "loss": 0.0024, "step": 17480 }, { "epoch": 103.49112426035504, "grad_norm": 0.10068017244338989, "learning_rate": 0.00011455619332998184, "loss": 0.0031, "step": 17490 }, { "epoch": 103.55029585798816, "grad_norm": 0.07180853188037872, "learning_rate": 0.00011440099501933276, "loss": 0.0031, "step": 17500 }, { "epoch": 103.6094674556213, "grad_norm": 0.07564003765583038, "learning_rate": 0.00011424583708191704, "loss": 0.0031, "step": 17510 }, { "epoch": 103.66863905325444, "grad_norm": 0.07826991379261017, "learning_rate": 0.00011409071969370097, "loss": 0.003, "step": 17520 }, { "epoch": 103.72781065088758, "grad_norm": 0.08455196022987366, "learning_rate": 0.00011393564303060497, "loss": 0.0031, "step": 17530 }, { "epoch": 103.78698224852072, "grad_norm": 0.07603083550930023, "learning_rate": 0.0001137806072685033, "loss": 0.0027, "step": 17540 }, { "epoch": 103.84615384615384, "grad_norm": 0.06942234933376312, "learning_rate": 0.00011362561258322365, "loss": 0.0031, "step": 17550 }, { "epoch": 103.90532544378698, "grad_norm": 0.1017143726348877, "learning_rate": 0.0001134706591505473, "loss": 0.0027, "step": 17560 }, { "epoch": 103.96449704142012, "grad_norm": 0.05941416323184967, "learning_rate": 0.00011331574714620873, "loss": 0.0024, "step": 17570 }, { "epoch": 104.02366863905326, "grad_norm": 0.06880386173725128, "learning_rate": 0.00011316087674589532, "loss": 0.0031, "step": 17580 }, { "epoch": 104.0828402366864, "grad_norm": 0.04496857523918152, "learning_rate": 0.00011300604812524737, "loss": 0.0026, "step": 17590 }, { "epoch": 104.14201183431953, "grad_norm": 0.056964240968227386, "learning_rate": 0.00011285126145985777, "loss": 0.0026, "step": 17600 }, { "epoch": 104.20118343195266, "grad_norm": 0.07439038157463074, "learning_rate": 0.0001126965169252718, "loss": 0.0025, "step": 17610 }, { "epoch": 104.2603550295858, "grad_norm": 0.061217207461595535, "learning_rate": 0.00011254181469698703, "loss": 0.0024, "step": 17620 }, { "epoch": 104.31952662721893, "grad_norm": 0.07257065176963806, "learning_rate": 0.00011238715495045293, "loss": 0.0024, "step": 17630 }, { "epoch": 104.37869822485207, "grad_norm": 0.07212720811367035, "learning_rate": 0.00011223253786107089, "loss": 0.0024, "step": 17640 }, { "epoch": 104.43786982248521, "grad_norm": 0.06603878736495972, "learning_rate": 0.00011207796360419396, "loss": 0.0032, "step": 17650 }, { "epoch": 104.49704142011835, "grad_norm": 0.08302470296621323, "learning_rate": 0.00011192343235512648, "loss": 0.0024, "step": 17660 }, { "epoch": 104.55621301775147, "grad_norm": 0.07776890695095062, "learning_rate": 0.00011176894428912414, "loss": 0.0026, "step": 17670 }, { "epoch": 104.61538461538461, "grad_norm": 0.07839533686637878, "learning_rate": 0.00011161449958139354, "loss": 0.003, "step": 17680 }, { "epoch": 104.67455621301775, "grad_norm": 0.06615793704986572, "learning_rate": 0.00011146009840709217, "loss": 0.0025, "step": 17690 }, { "epoch": 104.73372781065089, "grad_norm": 0.0626286044716835, "learning_rate": 0.00011130574094132824, "loss": 0.0027, "step": 17700 }, { "epoch": 104.79289940828403, "grad_norm": 0.08753439784049988, "learning_rate": 0.00011115142735916015, "loss": 0.0031, "step": 17710 }, { "epoch": 104.85207100591715, "grad_norm": 0.08646012097597122, "learning_rate": 0.00011099715783559676, "loss": 0.0026, "step": 17720 }, { "epoch": 104.91124260355029, "grad_norm": 0.0456191822886467, "learning_rate": 0.0001108429325455969, "loss": 0.0025, "step": 17730 }, { "epoch": 104.97041420118343, "grad_norm": 0.08422698080539703, "learning_rate": 0.00011068875166406914, "loss": 0.0034, "step": 17740 }, { "epoch": 105.02958579881657, "grad_norm": 0.04490303993225098, "learning_rate": 0.00011053461536587182, "loss": 0.0025, "step": 17750 }, { "epoch": 105.08875739644971, "grad_norm": 0.0882582738995552, "learning_rate": 0.00011038052382581261, "loss": 0.0032, "step": 17760 }, { "epoch": 105.14792899408285, "grad_norm": 0.06793341785669327, "learning_rate": 0.00011022647721864848, "loss": 0.0029, "step": 17770 }, { "epoch": 105.20710059171597, "grad_norm": 0.06345638632774353, "learning_rate": 0.00011007247571908546, "loss": 0.0029, "step": 17780 }, { "epoch": 105.26627218934911, "grad_norm": 0.06590575724840164, "learning_rate": 0.00010991851950177832, "loss": 0.0029, "step": 17790 }, { "epoch": 105.32544378698225, "grad_norm": 0.05301280319690704, "learning_rate": 0.00010976460874133058, "loss": 0.0026, "step": 17800 }, { "epoch": 105.38461538461539, "grad_norm": 0.053538717329502106, "learning_rate": 0.00010961074361229422, "loss": 0.0027, "step": 17810 }, { "epoch": 105.44378698224853, "grad_norm": 0.053069911897182465, "learning_rate": 0.00010945692428916931, "loss": 0.0028, "step": 17820 }, { "epoch": 105.50295857988165, "grad_norm": 0.04695619270205498, "learning_rate": 0.00010930315094640416, "loss": 0.0024, "step": 17830 }, { "epoch": 105.56213017751479, "grad_norm": 0.048895977437496185, "learning_rate": 0.00010914942375839486, "loss": 0.0032, "step": 17840 }, { "epoch": 105.62130177514793, "grad_norm": 0.07576733082532883, "learning_rate": 0.00010899574289948508, "loss": 0.0027, "step": 17850 }, { "epoch": 105.68047337278107, "grad_norm": 0.06771194189786911, "learning_rate": 0.00010884210854396605, "loss": 0.0027, "step": 17860 }, { "epoch": 105.7396449704142, "grad_norm": 0.06766222417354584, "learning_rate": 0.00010868852086607629, "loss": 0.0026, "step": 17870 }, { "epoch": 105.79881656804734, "grad_norm": 0.06393452733755112, "learning_rate": 0.00010853498004000128, "loss": 0.0024, "step": 17880 }, { "epoch": 105.85798816568047, "grad_norm": 0.04189496487379074, "learning_rate": 0.00010838148623987335, "loss": 0.0024, "step": 17890 }, { "epoch": 105.9171597633136, "grad_norm": 0.046655360609292984, "learning_rate": 0.00010822803963977158, "loss": 0.0029, "step": 17900 }, { "epoch": 105.97633136094674, "grad_norm": 0.05019436776638031, "learning_rate": 0.00010807464041372156, "loss": 0.0023, "step": 17910 }, { "epoch": 106.03550295857988, "grad_norm": 0.05936266854405403, "learning_rate": 0.00010792128873569495, "loss": 0.0029, "step": 17920 }, { "epoch": 106.09467455621302, "grad_norm": 0.05131862312555313, "learning_rate": 0.00010776798477960975, "loss": 0.0024, "step": 17930 }, { "epoch": 106.15384615384616, "grad_norm": 0.06670314818620682, "learning_rate": 0.00010761472871932965, "loss": 0.0026, "step": 17940 }, { "epoch": 106.21301775147928, "grad_norm": 0.07007740437984467, "learning_rate": 0.00010746152072866405, "loss": 0.0027, "step": 17950 }, { "epoch": 106.27218934911242, "grad_norm": 0.05677367001771927, "learning_rate": 0.0001073083609813679, "loss": 0.0025, "step": 17960 }, { "epoch": 106.33136094674556, "grad_norm": 0.06829215586185455, "learning_rate": 0.00010715524965114141, "loss": 0.0025, "step": 17970 }, { "epoch": 106.3905325443787, "grad_norm": 0.06737221032381058, "learning_rate": 0.00010700218691162981, "loss": 0.0024, "step": 17980 }, { "epoch": 106.44970414201184, "grad_norm": 0.0680999755859375, "learning_rate": 0.00010684917293642334, "loss": 0.0026, "step": 17990 }, { "epoch": 106.50887573964496, "grad_norm": 0.06342760473489761, "learning_rate": 0.00010669620789905688, "loss": 0.0026, "step": 18000 }, { "epoch": 106.5680473372781, "grad_norm": 0.061195600777864456, "learning_rate": 0.00010654329197300972, "loss": 0.0025, "step": 18010 }, { "epoch": 106.62721893491124, "grad_norm": 0.038281943649053574, "learning_rate": 0.00010639042533170565, "loss": 0.0028, "step": 18020 }, { "epoch": 106.68639053254438, "grad_norm": 0.05988399311900139, "learning_rate": 0.00010623760814851235, "loss": 0.0029, "step": 18030 }, { "epoch": 106.74556213017752, "grad_norm": 0.054994627833366394, "learning_rate": 0.00010608484059674157, "loss": 0.0027, "step": 18040 }, { "epoch": 106.80473372781066, "grad_norm": 0.07298792153596878, "learning_rate": 0.00010593212284964875, "loss": 0.0028, "step": 18050 }, { "epoch": 106.86390532544378, "grad_norm": 0.08129103481769562, "learning_rate": 0.00010577945508043265, "loss": 0.0026, "step": 18060 }, { "epoch": 106.92307692307692, "grad_norm": 0.062041401863098145, "learning_rate": 0.00010562683746223573, "loss": 0.0026, "step": 18070 }, { "epoch": 106.98224852071006, "grad_norm": 0.06809581071138382, "learning_rate": 0.00010547427016814321, "loss": 0.0023, "step": 18080 }, { "epoch": 107.0414201183432, "grad_norm": 0.08028557151556015, "learning_rate": 0.00010532175337118342, "loss": 0.003, "step": 18090 }, { "epoch": 107.10059171597634, "grad_norm": 0.06411972641944885, "learning_rate": 0.00010516928724432743, "loss": 0.0028, "step": 18100 }, { "epoch": 107.15976331360947, "grad_norm": 0.07101446390151978, "learning_rate": 0.0001050168719604887, "loss": 0.0028, "step": 18110 }, { "epoch": 107.2189349112426, "grad_norm": 0.06922531127929688, "learning_rate": 0.00010486450769252318, "loss": 0.0034, "step": 18120 }, { "epoch": 107.27810650887574, "grad_norm": 0.07396738231182098, "learning_rate": 0.00010471219461322892, "loss": 0.0025, "step": 18130 }, { "epoch": 107.33727810650888, "grad_norm": 0.06487666070461273, "learning_rate": 0.00010455993289534584, "loss": 0.0031, "step": 18140 }, { "epoch": 107.39644970414201, "grad_norm": 0.06941091269254684, "learning_rate": 0.00010440772271155573, "loss": 0.0029, "step": 18150 }, { "epoch": 107.45562130177515, "grad_norm": 0.05103975161910057, "learning_rate": 0.00010425556423448183, "loss": 0.0026, "step": 18160 }, { "epoch": 107.51479289940828, "grad_norm": 0.09330017864704132, "learning_rate": 0.00010410345763668884, "loss": 0.0026, "step": 18170 }, { "epoch": 107.57396449704142, "grad_norm": 0.05627771466970444, "learning_rate": 0.00010395140309068256, "loss": 0.0027, "step": 18180 }, { "epoch": 107.63313609467455, "grad_norm": 0.05622909590601921, "learning_rate": 0.0001037994007689097, "loss": 0.0027, "step": 18190 }, { "epoch": 107.6923076923077, "grad_norm": 0.04602759703993797, "learning_rate": 0.0001036474508437579, "loss": 0.0027, "step": 18200 }, { "epoch": 107.75147928994083, "grad_norm": 0.05317976325750351, "learning_rate": 0.00010349555348755524, "loss": 0.0028, "step": 18210 }, { "epoch": 107.81065088757397, "grad_norm": 0.06165151298046112, "learning_rate": 0.0001033437088725702, "loss": 0.0027, "step": 18220 }, { "epoch": 107.8698224852071, "grad_norm": 0.061673298478126526, "learning_rate": 0.00010319191717101151, "loss": 0.0024, "step": 18230 }, { "epoch": 107.92899408284023, "grad_norm": 0.04989318177103996, "learning_rate": 0.00010304017855502786, "loss": 0.0022, "step": 18240 }, { "epoch": 107.98816568047337, "grad_norm": 0.07069017738103867, "learning_rate": 0.00010288849319670771, "loss": 0.0028, "step": 18250 }, { "epoch": 108.04733727810651, "grad_norm": 0.06672410666942596, "learning_rate": 0.0001027368612680791, "loss": 0.0027, "step": 18260 }, { "epoch": 108.10650887573965, "grad_norm": 0.05315764993429184, "learning_rate": 0.00010258528294110953, "loss": 0.0025, "step": 18270 }, { "epoch": 108.16568047337279, "grad_norm": 0.05432679131627083, "learning_rate": 0.00010243375838770578, "loss": 0.0026, "step": 18280 }, { "epoch": 108.22485207100591, "grad_norm": 0.0373397171497345, "learning_rate": 0.00010228228777971345, "loss": 0.0021, "step": 18290 }, { "epoch": 108.28402366863905, "grad_norm": 0.06335088610649109, "learning_rate": 0.0001021308712889171, "loss": 0.0027, "step": 18300 }, { "epoch": 108.34319526627219, "grad_norm": 0.03840327635407448, "learning_rate": 0.0001019795090870399, "loss": 0.0023, "step": 18310 }, { "epoch": 108.40236686390533, "grad_norm": 0.058608636260032654, "learning_rate": 0.0001018282013457434, "loss": 0.0027, "step": 18320 }, { "epoch": 108.46153846153847, "grad_norm": 0.044155191630125046, "learning_rate": 0.00010167694823662744, "loss": 0.0022, "step": 18330 }, { "epoch": 108.52071005917159, "grad_norm": 0.05117563530802727, "learning_rate": 0.00010152574993122988, "loss": 0.0025, "step": 18340 }, { "epoch": 108.57988165680473, "grad_norm": 0.04563913866877556, "learning_rate": 0.00010137460660102639, "loss": 0.0028, "step": 18350 }, { "epoch": 108.63905325443787, "grad_norm": 0.06338081508874893, "learning_rate": 0.00010122351841743035, "loss": 0.0023, "step": 18360 }, { "epoch": 108.69822485207101, "grad_norm": 0.062326688319444656, "learning_rate": 0.0001010724855517926, "loss": 0.0025, "step": 18370 }, { "epoch": 108.75739644970415, "grad_norm": 0.048502106219530106, "learning_rate": 0.00010092150817540118, "loss": 0.0028, "step": 18380 }, { "epoch": 108.81656804733728, "grad_norm": 0.05026056617498398, "learning_rate": 0.00010077058645948125, "loss": 0.0024, "step": 18390 }, { "epoch": 108.87573964497041, "grad_norm": 0.06371040642261505, "learning_rate": 0.00010061972057519484, "loss": 0.0027, "step": 18400 }, { "epoch": 108.93491124260355, "grad_norm": 0.06802099943161011, "learning_rate": 0.00010046891069364062, "loss": 0.003, "step": 18410 }, { "epoch": 108.99408284023669, "grad_norm": 0.061795614659786224, "learning_rate": 0.00010031815698585384, "loss": 0.0024, "step": 18420 }, { "epoch": 109.05325443786982, "grad_norm": 0.052875835448503494, "learning_rate": 0.0001001674596228059, "loss": 0.0024, "step": 18430 }, { "epoch": 109.11242603550296, "grad_norm": 0.05745892971754074, "learning_rate": 0.00010001681877540442, "loss": 0.0023, "step": 18440 }, { "epoch": 109.17159763313609, "grad_norm": 0.06251855194568634, "learning_rate": 9.986623461449292e-05, "loss": 0.0022, "step": 18450 }, { "epoch": 109.23076923076923, "grad_norm": 0.06061991676688194, "learning_rate": 9.971570731085056e-05, "loss": 0.0023, "step": 18460 }, { "epoch": 109.28994082840237, "grad_norm": 0.0549188032746315, "learning_rate": 9.956523703519202e-05, "loss": 0.0022, "step": 18470 }, { "epoch": 109.3491124260355, "grad_norm": 0.041007574647665024, "learning_rate": 9.941482395816737e-05, "loss": 0.0028, "step": 18480 }, { "epoch": 109.40828402366864, "grad_norm": 0.056563470512628555, "learning_rate": 9.926446825036181e-05, "loss": 0.0029, "step": 18490 }, { "epoch": 109.46745562130178, "grad_norm": 0.05450426787137985, "learning_rate": 9.911417008229544e-05, "loss": 0.0031, "step": 18500 }, { "epoch": 109.5266272189349, "grad_norm": 0.0733078271150589, "learning_rate": 9.896392962442308e-05, "loss": 0.0028, "step": 18510 }, { "epoch": 109.58579881656804, "grad_norm": 0.06381194293498993, "learning_rate": 9.88137470471341e-05, "loss": 0.0027, "step": 18520 }, { "epoch": 109.64497041420118, "grad_norm": 0.04395580664277077, "learning_rate": 9.866362252075237e-05, "loss": 0.0023, "step": 18530 }, { "epoch": 109.70414201183432, "grad_norm": 0.04626953601837158, "learning_rate": 9.851355621553568e-05, "loss": 0.0026, "step": 18540 }, { "epoch": 109.76331360946746, "grad_norm": 0.058947548270225525, "learning_rate": 9.8363548301676e-05, "loss": 0.002, "step": 18550 }, { "epoch": 109.8224852071006, "grad_norm": 0.04248568043112755, "learning_rate": 9.821359894929895e-05, "loss": 0.002, "step": 18560 }, { "epoch": 109.88165680473372, "grad_norm": 0.06139054521918297, "learning_rate": 9.806370832846378e-05, "loss": 0.0029, "step": 18570 }, { "epoch": 109.94082840236686, "grad_norm": 0.04515659064054489, "learning_rate": 9.791387660916321e-05, "loss": 0.0031, "step": 18580 }, { "epoch": 110.0, "grad_norm": 0.05436938628554344, "learning_rate": 9.776410396132301e-05, "loss": 0.0028, "step": 18590 }, { "epoch": 110.05917159763314, "grad_norm": 0.09074869006872177, "learning_rate": 9.761439055480204e-05, "loss": 0.0028, "step": 18600 }, { "epoch": 110.11834319526628, "grad_norm": 0.06270001828670502, "learning_rate": 9.746473655939202e-05, "loss": 0.0025, "step": 18610 }, { "epoch": 110.1775147928994, "grad_norm": 0.04278375208377838, "learning_rate": 9.731514214481714e-05, "loss": 0.0026, "step": 18620 }, { "epoch": 110.23668639053254, "grad_norm": 0.07645968347787857, "learning_rate": 9.716560748073418e-05, "loss": 0.0029, "step": 18630 }, { "epoch": 110.29585798816568, "grad_norm": 0.055337268859148026, "learning_rate": 9.701613273673211e-05, "loss": 0.0024, "step": 18640 }, { "epoch": 110.35502958579882, "grad_norm": 0.10408639907836914, "learning_rate": 9.686671808233185e-05, "loss": 0.0028, "step": 18650 }, { "epoch": 110.41420118343196, "grad_norm": 0.08904475718736649, "learning_rate": 9.67173636869863e-05, "loss": 0.0025, "step": 18660 }, { "epoch": 110.4733727810651, "grad_norm": 0.08776527643203735, "learning_rate": 9.656806972007997e-05, "loss": 0.0026, "step": 18670 }, { "epoch": 110.53254437869822, "grad_norm": 0.07002352923154831, "learning_rate": 9.641883635092876e-05, "loss": 0.0027, "step": 18680 }, { "epoch": 110.59171597633136, "grad_norm": 0.07340274006128311, "learning_rate": 9.626966374878002e-05, "loss": 0.0031, "step": 18690 }, { "epoch": 110.6508875739645, "grad_norm": 0.0754048228263855, "learning_rate": 9.6120552082812e-05, "loss": 0.0025, "step": 18700 }, { "epoch": 110.71005917159763, "grad_norm": 0.05978790298104286, "learning_rate": 9.597150152213401e-05, "loss": 0.0025, "step": 18710 }, { "epoch": 110.76923076923077, "grad_norm": 0.050629932433366776, "learning_rate": 9.58225122357859e-05, "loss": 0.0024, "step": 18720 }, { "epoch": 110.82840236686391, "grad_norm": 0.08626433461904526, "learning_rate": 9.56735843927381e-05, "loss": 0.0027, "step": 18730 }, { "epoch": 110.88757396449704, "grad_norm": 0.05098642036318779, "learning_rate": 9.552471816189141e-05, "loss": 0.0023, "step": 18740 }, { "epoch": 110.94674556213018, "grad_norm": 0.06246672198176384, "learning_rate": 9.537591371207666e-05, "loss": 0.0025, "step": 18750 }, { "epoch": 111.00591715976331, "grad_norm": 0.06689068675041199, "learning_rate": 9.522717121205468e-05, "loss": 0.0023, "step": 18760 }, { "epoch": 111.06508875739645, "grad_norm": 0.04545015096664429, "learning_rate": 9.5078490830516e-05, "loss": 0.0019, "step": 18770 }, { "epoch": 111.12426035502959, "grad_norm": 0.06168226897716522, "learning_rate": 9.492987273608072e-05, "loss": 0.0023, "step": 18780 }, { "epoch": 111.18343195266272, "grad_norm": 0.06753230839967728, "learning_rate": 9.47813170972983e-05, "loss": 0.0026, "step": 18790 }, { "epoch": 111.24260355029585, "grad_norm": 0.05918526649475098, "learning_rate": 9.46328240826474e-05, "loss": 0.0026, "step": 18800 }, { "epoch": 111.30177514792899, "grad_norm": 0.05961300805211067, "learning_rate": 9.448439386053556e-05, "loss": 0.0026, "step": 18810 }, { "epoch": 111.36094674556213, "grad_norm": 0.052472203969955444, "learning_rate": 9.433602659929923e-05, "loss": 0.0023, "step": 18820 }, { "epoch": 111.42011834319527, "grad_norm": 0.06282800436019897, "learning_rate": 9.418772246720336e-05, "loss": 0.0028, "step": 18830 }, { "epoch": 111.47928994082841, "grad_norm": 0.08033928275108337, "learning_rate": 9.403948163244131e-05, "loss": 0.003, "step": 18840 }, { "epoch": 111.53846153846153, "grad_norm": 0.06436646729707718, "learning_rate": 9.389130426313479e-05, "loss": 0.0024, "step": 18850 }, { "epoch": 111.59763313609467, "grad_norm": 0.08813685178756714, "learning_rate": 9.374319052733332e-05, "loss": 0.0032, "step": 18860 }, { "epoch": 111.65680473372781, "grad_norm": 0.07302023470401764, "learning_rate": 9.359514059301439e-05, "loss": 0.0025, "step": 18870 }, { "epoch": 111.71597633136095, "grad_norm": 0.05610670894384384, "learning_rate": 9.344715462808314e-05, "loss": 0.0028, "step": 18880 }, { "epoch": 111.77514792899409, "grad_norm": 0.05756024271249771, "learning_rate": 9.329923280037201e-05, "loss": 0.0023, "step": 18890 }, { "epoch": 111.83431952662721, "grad_norm": 0.05802343785762787, "learning_rate": 9.3151375277641e-05, "loss": 0.0022, "step": 18900 }, { "epoch": 111.89349112426035, "grad_norm": 0.08354628086090088, "learning_rate": 9.300358222757685e-05, "loss": 0.0033, "step": 18910 }, { "epoch": 111.95266272189349, "grad_norm": 0.07619790732860565, "learning_rate": 9.285585381779337e-05, "loss": 0.0025, "step": 18920 }, { "epoch": 112.01183431952663, "grad_norm": 0.06650859117507935, "learning_rate": 9.270819021583106e-05, "loss": 0.0024, "step": 18930 }, { "epoch": 112.07100591715977, "grad_norm": 0.09362515807151794, "learning_rate": 9.256059158915678e-05, "loss": 0.0024, "step": 18940 }, { "epoch": 112.1301775147929, "grad_norm": 0.051160778850317, "learning_rate": 9.241305810516391e-05, "loss": 0.0024, "step": 18950 }, { "epoch": 112.18934911242603, "grad_norm": 0.06877075880765915, "learning_rate": 9.226558993117174e-05, "loss": 0.0023, "step": 18960 }, { "epoch": 112.24852071005917, "grad_norm": 0.046519991010427475, "learning_rate": 9.211818723442562e-05, "loss": 0.0021, "step": 18970 }, { "epoch": 112.3076923076923, "grad_norm": 0.03509017080068588, "learning_rate": 9.197085018209668e-05, "loss": 0.0023, "step": 18980 }, { "epoch": 112.36686390532545, "grad_norm": 0.04458954930305481, "learning_rate": 9.182357894128144e-05, "loss": 0.0022, "step": 18990 }, { "epoch": 112.42603550295858, "grad_norm": 0.04951770231127739, "learning_rate": 9.167637367900191e-05, "loss": 0.0021, "step": 19000 }, { "epoch": 112.48520710059172, "grad_norm": 0.06115206331014633, "learning_rate": 9.152923456220525e-05, "loss": 0.0026, "step": 19010 }, { "epoch": 112.54437869822485, "grad_norm": 0.05037323012948036, "learning_rate": 9.138216175776358e-05, "loss": 0.002, "step": 19020 }, { "epoch": 112.60355029585799, "grad_norm": 0.07220515608787537, "learning_rate": 9.123515543247385e-05, "loss": 0.0026, "step": 19030 }, { "epoch": 112.66272189349112, "grad_norm": 0.06670856475830078, "learning_rate": 9.10882157530576e-05, "loss": 0.003, "step": 19040 }, { "epoch": 112.72189349112426, "grad_norm": 0.06525550782680511, "learning_rate": 9.09413428861607e-05, "loss": 0.0026, "step": 19050 }, { "epoch": 112.7810650887574, "grad_norm": 0.08630586415529251, "learning_rate": 9.079453699835339e-05, "loss": 0.0026, "step": 19060 }, { "epoch": 112.84023668639053, "grad_norm": 0.07929232716560364, "learning_rate": 9.064779825612991e-05, "loss": 0.0025, "step": 19070 }, { "epoch": 112.89940828402366, "grad_norm": 0.05846282094717026, "learning_rate": 9.050112682590826e-05, "loss": 0.0025, "step": 19080 }, { "epoch": 112.9585798816568, "grad_norm": 0.04993211105465889, "learning_rate": 9.03545228740302e-05, "loss": 0.0028, "step": 19090 }, { "epoch": 113.01775147928994, "grad_norm": 0.055006518959999084, "learning_rate": 9.02079865667609e-05, "loss": 0.0025, "step": 19100 }, { "epoch": 113.07692307692308, "grad_norm": 0.052738748490810394, "learning_rate": 9.00615180702889e-05, "loss": 0.0025, "step": 19110 }, { "epoch": 113.13609467455622, "grad_norm": 0.047239750623703, "learning_rate": 8.991511755072571e-05, "loss": 0.0021, "step": 19120 }, { "epoch": 113.19526627218934, "grad_norm": 0.05097019299864769, "learning_rate": 8.976878517410585e-05, "loss": 0.0025, "step": 19130 }, { "epoch": 113.25443786982248, "grad_norm": 0.07167719304561615, "learning_rate": 8.962252110638653e-05, "loss": 0.0029, "step": 19140 }, { "epoch": 113.31360946745562, "grad_norm": 0.0793907642364502, "learning_rate": 8.947632551344742e-05, "loss": 0.0025, "step": 19150 }, { "epoch": 113.37278106508876, "grad_norm": 0.07700729370117188, "learning_rate": 8.933019856109064e-05, "loss": 0.0025, "step": 19160 }, { "epoch": 113.4319526627219, "grad_norm": 0.05379784107208252, "learning_rate": 8.918414041504041e-05, "loss": 0.0023, "step": 19170 }, { "epoch": 113.49112426035504, "grad_norm": 0.07810360193252563, "learning_rate": 8.90381512409429e-05, "loss": 0.0027, "step": 19180 }, { "epoch": 113.55029585798816, "grad_norm": 0.07285702228546143, "learning_rate": 8.88922312043661e-05, "loss": 0.0027, "step": 19190 }, { "epoch": 113.6094674556213, "grad_norm": 0.08032391965389252, "learning_rate": 8.874638047079962e-05, "loss": 0.0029, "step": 19200 }, { "epoch": 113.66863905325444, "grad_norm": 0.0798841044306755, "learning_rate": 8.86005992056543e-05, "loss": 0.0024, "step": 19210 }, { "epoch": 113.72781065088758, "grad_norm": 0.05505992844700813, "learning_rate": 8.845488757426244e-05, "loss": 0.0021, "step": 19220 }, { "epoch": 113.78698224852072, "grad_norm": 0.049613721668720245, "learning_rate": 8.830924574187716e-05, "loss": 0.0023, "step": 19230 }, { "epoch": 113.84615384615384, "grad_norm": 0.07750458270311356, "learning_rate": 8.816367387367252e-05, "loss": 0.0026, "step": 19240 }, { "epoch": 113.90532544378698, "grad_norm": 0.05711774155497551, "learning_rate": 8.80181721347433e-05, "loss": 0.0025, "step": 19250 }, { "epoch": 113.96449704142012, "grad_norm": 0.05277301371097565, "learning_rate": 8.787274069010458e-05, "loss": 0.0025, "step": 19260 }, { "epoch": 114.02366863905326, "grad_norm": 0.05178957059979439, "learning_rate": 8.77273797046918e-05, "loss": 0.0021, "step": 19270 }, { "epoch": 114.0828402366864, "grad_norm": 0.08065932989120483, "learning_rate": 8.758208934336053e-05, "loss": 0.0022, "step": 19280 }, { "epoch": 114.14201183431953, "grad_norm": 0.06286032497882843, "learning_rate": 8.743686977088622e-05, "loss": 0.0024, "step": 19290 }, { "epoch": 114.20118343195266, "grad_norm": 0.05176641792058945, "learning_rate": 8.729172115196399e-05, "loss": 0.0027, "step": 19300 }, { "epoch": 114.2603550295858, "grad_norm": 0.05532832443714142, "learning_rate": 8.714664365120846e-05, "loss": 0.0022, "step": 19310 }, { "epoch": 114.31952662721893, "grad_norm": 0.06789843738079071, "learning_rate": 8.70016374331538e-05, "loss": 0.0027, "step": 19320 }, { "epoch": 114.37869822485207, "grad_norm": 0.07758024334907532, "learning_rate": 8.685670266225314e-05, "loss": 0.002, "step": 19330 }, { "epoch": 114.43786982248521, "grad_norm": 0.054560963064432144, "learning_rate": 8.671183950287872e-05, "loss": 0.0025, "step": 19340 }, { "epoch": 114.49704142011835, "grad_norm": 0.06734860688447952, "learning_rate": 8.656704811932133e-05, "loss": 0.0021, "step": 19350 }, { "epoch": 114.55621301775147, "grad_norm": 0.07300372421741486, "learning_rate": 8.64223286757906e-05, "loss": 0.0022, "step": 19360 }, { "epoch": 114.61538461538461, "grad_norm": 0.07138649374246597, "learning_rate": 8.627768133641445e-05, "loss": 0.0025, "step": 19370 }, { "epoch": 114.67455621301775, "grad_norm": 0.07813030481338501, "learning_rate": 8.613310626523909e-05, "loss": 0.0022, "step": 19380 }, { "epoch": 114.73372781065089, "grad_norm": 0.06277472525835037, "learning_rate": 8.598860362622872e-05, "loss": 0.0024, "step": 19390 }, { "epoch": 114.79289940828403, "grad_norm": 0.0790487602353096, "learning_rate": 8.584417358326549e-05, "loss": 0.0024, "step": 19400 }, { "epoch": 114.85207100591715, "grad_norm": 0.06708064675331116, "learning_rate": 8.569981630014901e-05, "loss": 0.0025, "step": 19410 }, { "epoch": 114.91124260355029, "grad_norm": 0.06758009642362595, "learning_rate": 8.55555319405966e-05, "loss": 0.0024, "step": 19420 }, { "epoch": 114.97041420118343, "grad_norm": 0.04525083303451538, "learning_rate": 8.54113206682427e-05, "loss": 0.0024, "step": 19430 }, { "epoch": 115.02958579881657, "grad_norm": 0.05347958952188492, "learning_rate": 8.526718264663903e-05, "loss": 0.0022, "step": 19440 }, { "epoch": 115.08875739644971, "grad_norm": 0.036771900951862335, "learning_rate": 8.512311803925417e-05, "loss": 0.0021, "step": 19450 }, { "epoch": 115.14792899408285, "grad_norm": 0.06748124212026596, "learning_rate": 8.497912700947329e-05, "loss": 0.0022, "step": 19460 }, { "epoch": 115.20710059171597, "grad_norm": 0.0545913390815258, "learning_rate": 8.483520972059829e-05, "loss": 0.0022, "step": 19470 }, { "epoch": 115.26627218934911, "grad_norm": 0.04988021403551102, "learning_rate": 8.469136633584743e-05, "loss": 0.0023, "step": 19480 }, { "epoch": 115.32544378698225, "grad_norm": 0.05765072628855705, "learning_rate": 8.454759701835511e-05, "loss": 0.0024, "step": 19490 }, { "epoch": 115.38461538461539, "grad_norm": 0.050064925104379654, "learning_rate": 8.440390193117169e-05, "loss": 0.0022, "step": 19500 }, { "epoch": 115.44378698224853, "grad_norm": 0.043676815927028656, "learning_rate": 8.426028123726343e-05, "loss": 0.002, "step": 19510 }, { "epoch": 115.50295857988165, "grad_norm": 0.05446246638894081, "learning_rate": 8.411673509951219e-05, "loss": 0.0025, "step": 19520 }, { "epoch": 115.56213017751479, "grad_norm": 0.07113617658615112, "learning_rate": 8.397326368071522e-05, "loss": 0.0026, "step": 19530 }, { "epoch": 115.62130177514793, "grad_norm": 0.09931206703186035, "learning_rate": 8.382986714358511e-05, "loss": 0.0025, "step": 19540 }, { "epoch": 115.68047337278107, "grad_norm": 0.06679368764162064, "learning_rate": 8.368654565074956e-05, "loss": 0.002, "step": 19550 }, { "epoch": 115.7396449704142, "grad_norm": 0.06699498742818832, "learning_rate": 8.354329936475095e-05, "loss": 0.0024, "step": 19560 }, { "epoch": 115.79881656804734, "grad_norm": 0.04202995449304581, "learning_rate": 8.340012844804656e-05, "loss": 0.0026, "step": 19570 }, { "epoch": 115.85798816568047, "grad_norm": 0.059622831642627716, "learning_rate": 8.325703306300816e-05, "loss": 0.0021, "step": 19580 }, { "epoch": 115.9171597633136, "grad_norm": 0.0478627048432827, "learning_rate": 8.311401337192186e-05, "loss": 0.0025, "step": 19590 }, { "epoch": 115.97633136094674, "grad_norm": 0.10457063466310501, "learning_rate": 8.297106953698785e-05, "loss": 0.0027, "step": 19600 }, { "epoch": 116.03550295857988, "grad_norm": 0.042714547365903854, "learning_rate": 8.282820172032046e-05, "loss": 0.0024, "step": 19610 }, { "epoch": 116.09467455621302, "grad_norm": 0.062375057488679886, "learning_rate": 8.268541008394758e-05, "loss": 0.002, "step": 19620 }, { "epoch": 116.15384615384616, "grad_norm": 0.057086631655693054, "learning_rate": 8.254269478981083e-05, "loss": 0.0018, "step": 19630 }, { "epoch": 116.21301775147928, "grad_norm": 0.04875011742115021, "learning_rate": 8.240005599976523e-05, "loss": 0.0023, "step": 19640 }, { "epoch": 116.27218934911242, "grad_norm": 0.0585559718310833, "learning_rate": 8.22574938755791e-05, "loss": 0.0028, "step": 19650 }, { "epoch": 116.33136094674556, "grad_norm": 0.06148819997906685, "learning_rate": 8.21150085789337e-05, "loss": 0.0022, "step": 19660 }, { "epoch": 116.3905325443787, "grad_norm": 0.09322823584079742, "learning_rate": 8.197260027142333e-05, "loss": 0.0023, "step": 19670 }, { "epoch": 116.44970414201184, "grad_norm": 0.05716734007000923, "learning_rate": 8.183026911455468e-05, "loss": 0.0021, "step": 19680 }, { "epoch": 116.50887573964496, "grad_norm": 0.06607849895954132, "learning_rate": 8.168801526974721e-05, "loss": 0.0022, "step": 19690 }, { "epoch": 116.5680473372781, "grad_norm": 0.0759645327925682, "learning_rate": 8.15458388983326e-05, "loss": 0.0026, "step": 19700 }, { "epoch": 116.62721893491124, "grad_norm": 0.0665799006819725, "learning_rate": 8.140374016155469e-05, "loss": 0.0023, "step": 19710 }, { "epoch": 116.68639053254438, "grad_norm": 0.062154825776815414, "learning_rate": 8.126171922056927e-05, "loss": 0.0023, "step": 19720 }, { "epoch": 116.74556213017752, "grad_norm": 0.0640532448887825, "learning_rate": 8.111977623644386e-05, "loss": 0.0023, "step": 19730 }, { "epoch": 116.80473372781066, "grad_norm": 0.04813823103904724, "learning_rate": 8.097791137015763e-05, "loss": 0.0027, "step": 19740 }, { "epoch": 116.86390532544378, "grad_norm": 0.04024325683712959, "learning_rate": 8.083612478260109e-05, "loss": 0.0024, "step": 19750 }, { "epoch": 116.92307692307692, "grad_norm": 0.0699412152171135, "learning_rate": 8.069441663457605e-05, "loss": 0.0025, "step": 19760 }, { "epoch": 116.98224852071006, "grad_norm": 0.05680805444717407, "learning_rate": 8.055278708679537e-05, "loss": 0.0023, "step": 19770 }, { "epoch": 117.0414201183432, "grad_norm": 0.056761033833026886, "learning_rate": 8.041123629988256e-05, "loss": 0.0026, "step": 19780 }, { "epoch": 117.10059171597634, "grad_norm": 0.049943726509809494, "learning_rate": 8.026976443437209e-05, "loss": 0.0025, "step": 19790 }, { "epoch": 117.15976331360947, "grad_norm": 0.08048746734857559, "learning_rate": 8.012837165070876e-05, "loss": 0.0022, "step": 19800 }, { "epoch": 117.2189349112426, "grad_norm": 0.04605983942747116, "learning_rate": 7.99870581092477e-05, "loss": 0.0019, "step": 19810 }, { "epoch": 117.27810650887574, "grad_norm": 0.07436847686767578, "learning_rate": 7.984582397025432e-05, "loss": 0.0022, "step": 19820 }, { "epoch": 117.33727810650888, "grad_norm": 0.05329681560397148, "learning_rate": 7.970466939390368e-05, "loss": 0.0023, "step": 19830 }, { "epoch": 117.39644970414201, "grad_norm": 0.051842473447322845, "learning_rate": 7.956359454028081e-05, "loss": 0.0023, "step": 19840 }, { "epoch": 117.45562130177515, "grad_norm": 0.06262028962373734, "learning_rate": 7.942259956938038e-05, "loss": 0.0023, "step": 19850 }, { "epoch": 117.51479289940828, "grad_norm": 0.0489448718726635, "learning_rate": 7.92816846411063e-05, "loss": 0.0024, "step": 19860 }, { "epoch": 117.57396449704142, "grad_norm": 0.047768015414476395, "learning_rate": 7.914084991527183e-05, "loss": 0.0022, "step": 19870 }, { "epoch": 117.63313609467455, "grad_norm": 0.06836022436618805, "learning_rate": 7.900009555159925e-05, "loss": 0.0028, "step": 19880 }, { "epoch": 117.6923076923077, "grad_norm": 0.07999991625547409, "learning_rate": 7.885942170971958e-05, "loss": 0.0022, "step": 19890 }, { "epoch": 117.75147928994083, "grad_norm": 0.08517464995384216, "learning_rate": 7.871882854917264e-05, "loss": 0.0023, "step": 19900 }, { "epoch": 117.81065088757397, "grad_norm": 0.0909566879272461, "learning_rate": 7.857831622940675e-05, "loss": 0.0021, "step": 19910 }, { "epoch": 117.8698224852071, "grad_norm": 0.05761289224028587, "learning_rate": 7.843788490977852e-05, "loss": 0.0023, "step": 19920 }, { "epoch": 117.92899408284023, "grad_norm": 0.05503533035516739, "learning_rate": 7.829753474955265e-05, "loss": 0.0021, "step": 19930 }, { "epoch": 117.98816568047337, "grad_norm": 0.04418548569083214, "learning_rate": 7.815726590790189e-05, "loss": 0.0024, "step": 19940 }, { "epoch": 118.04733727810651, "grad_norm": 0.04227496311068535, "learning_rate": 7.801707854390668e-05, "loss": 0.0024, "step": 19950 }, { "epoch": 118.10650887573965, "grad_norm": 0.06459862738847733, "learning_rate": 7.787697281655513e-05, "loss": 0.0025, "step": 19960 }, { "epoch": 118.16568047337279, "grad_norm": 0.04367069900035858, "learning_rate": 7.773694888474267e-05, "loss": 0.002, "step": 19970 }, { "epoch": 118.22485207100591, "grad_norm": 0.052368223667144775, "learning_rate": 7.759700690727214e-05, "loss": 0.0019, "step": 19980 }, { "epoch": 118.28402366863905, "grad_norm": 0.06318464875221252, "learning_rate": 7.745714704285317e-05, "loss": 0.002, "step": 19990 }, { "epoch": 118.34319526627219, "grad_norm": 0.06477081775665283, "learning_rate": 7.731736945010247e-05, "loss": 0.0024, "step": 20000 }, { "epoch": 118.40236686390533, "grad_norm": 0.08060213178396225, "learning_rate": 7.71776742875434e-05, "loss": 0.0019, "step": 20010 }, { "epoch": 118.46153846153847, "grad_norm": 0.04900382086634636, "learning_rate": 7.703806171360581e-05, "loss": 0.0024, "step": 20020 }, { "epoch": 118.52071005917159, "grad_norm": 0.042557116597890854, "learning_rate": 7.689853188662588e-05, "loss": 0.0024, "step": 20030 }, { "epoch": 118.57988165680473, "grad_norm": 0.051863089203834534, "learning_rate": 7.675908496484605e-05, "loss": 0.0025, "step": 20040 }, { "epoch": 118.63905325443787, "grad_norm": 0.057899147272109985, "learning_rate": 7.661972110641448e-05, "loss": 0.0025, "step": 20050 }, { "epoch": 118.69822485207101, "grad_norm": 0.06672780215740204, "learning_rate": 7.648044046938543e-05, "loss": 0.0023, "step": 20060 }, { "epoch": 118.75739644970415, "grad_norm": 0.057830560952425, "learning_rate": 7.634124321171858e-05, "loss": 0.002, "step": 20070 }, { "epoch": 118.81656804733728, "grad_norm": 0.0423782542347908, "learning_rate": 7.620212949127916e-05, "loss": 0.0019, "step": 20080 }, { "epoch": 118.87573964497041, "grad_norm": 0.06666363775730133, "learning_rate": 7.606309946583762e-05, "loss": 0.0021, "step": 20090 }, { "epoch": 118.93491124260355, "grad_norm": 0.05924544855952263, "learning_rate": 7.592415329306939e-05, "loss": 0.0019, "step": 20100 }, { "epoch": 118.99408284023669, "grad_norm": 0.059937357902526855, "learning_rate": 7.578529113055493e-05, "loss": 0.0021, "step": 20110 }, { "epoch": 119.05325443786982, "grad_norm": 0.05257750302553177, "learning_rate": 7.564651313577943e-05, "loss": 0.0022, "step": 20120 }, { "epoch": 119.11242603550296, "grad_norm": 0.061471130698919296, "learning_rate": 7.550781946613253e-05, "loss": 0.0024, "step": 20130 }, { "epoch": 119.17159763313609, "grad_norm": 0.04978213086724281, "learning_rate": 7.536921027890832e-05, "loss": 0.0021, "step": 20140 }, { "epoch": 119.23076923076923, "grad_norm": 0.05107851326465607, "learning_rate": 7.523068573130503e-05, "loss": 0.0019, "step": 20150 }, { "epoch": 119.28994082840237, "grad_norm": 0.040333978831768036, "learning_rate": 7.50922459804249e-05, "loss": 0.0019, "step": 20160 }, { "epoch": 119.3491124260355, "grad_norm": 0.053422730416059494, "learning_rate": 7.495389118327407e-05, "loss": 0.0025, "step": 20170 }, { "epoch": 119.40828402366864, "grad_norm": 0.054649364203214645, "learning_rate": 7.48156214967622e-05, "loss": 0.0022, "step": 20180 }, { "epoch": 119.46745562130178, "grad_norm": 0.06534082442522049, "learning_rate": 7.467743707770254e-05, "loss": 0.0024, "step": 20190 }, { "epoch": 119.5266272189349, "grad_norm": 0.036834247410297394, "learning_rate": 7.453933808281166e-05, "loss": 0.0019, "step": 20200 }, { "epoch": 119.58579881656804, "grad_norm": 0.05441750958561897, "learning_rate": 7.440132466870907e-05, "loss": 0.0023, "step": 20210 }, { "epoch": 119.64497041420118, "grad_norm": 0.08107088506221771, "learning_rate": 7.42633969919174e-05, "loss": 0.0023, "step": 20220 }, { "epoch": 119.70414201183432, "grad_norm": 0.09455548226833344, "learning_rate": 7.412555520886197e-05, "loss": 0.0021, "step": 20230 }, { "epoch": 119.76331360946746, "grad_norm": 0.06748022884130478, "learning_rate": 7.398779947587073e-05, "loss": 0.0024, "step": 20240 }, { "epoch": 119.8224852071006, "grad_norm": 0.07552973926067352, "learning_rate": 7.385012994917404e-05, "loss": 0.0023, "step": 20250 }, { "epoch": 119.88165680473372, "grad_norm": 0.05396028235554695, "learning_rate": 7.371254678490437e-05, "loss": 0.002, "step": 20260 }, { "epoch": 119.94082840236686, "grad_norm": 0.05989701300859451, "learning_rate": 7.357505013909638e-05, "loss": 0.0021, "step": 20270 }, { "epoch": 120.0, "grad_norm": 0.06365564465522766, "learning_rate": 7.343764016768657e-05, "loss": 0.0018, "step": 20280 }, { "epoch": 120.05917159763314, "grad_norm": 0.05739565193653107, "learning_rate": 7.330031702651316e-05, "loss": 0.0021, "step": 20290 }, { "epoch": 120.11834319526628, "grad_norm": 0.05884160101413727, "learning_rate": 7.316308087131586e-05, "loss": 0.0023, "step": 20300 }, { "epoch": 120.1775147928994, "grad_norm": 0.052971139550209045, "learning_rate": 7.302593185773579e-05, "loss": 0.0018, "step": 20310 }, { "epoch": 120.23668639053254, "grad_norm": 0.08232437819242477, "learning_rate": 7.288887014131508e-05, "loss": 0.0027, "step": 20320 }, { "epoch": 120.29585798816568, "grad_norm": 0.06641289591789246, "learning_rate": 7.275189587749705e-05, "loss": 0.0021, "step": 20330 }, { "epoch": 120.35502958579882, "grad_norm": 0.06591779738664627, "learning_rate": 7.261500922162568e-05, "loss": 0.002, "step": 20340 }, { "epoch": 120.41420118343196, "grad_norm": 0.05154116451740265, "learning_rate": 7.247821032894586e-05, "loss": 0.0019, "step": 20350 }, { "epoch": 120.4733727810651, "grad_norm": 0.07095394283533096, "learning_rate": 7.234149935460258e-05, "loss": 0.0022, "step": 20360 }, { "epoch": 120.53254437869822, "grad_norm": 0.04992489516735077, "learning_rate": 7.220487645364134e-05, "loss": 0.0017, "step": 20370 }, { "epoch": 120.59171597633136, "grad_norm": 0.05789138004183769, "learning_rate": 7.206834178100773e-05, "loss": 0.0021, "step": 20380 }, { "epoch": 120.6508875739645, "grad_norm": 0.07732339203357697, "learning_rate": 7.193189549154727e-05, "loss": 0.0022, "step": 20390 }, { "epoch": 120.71005917159763, "grad_norm": 0.03773341327905655, "learning_rate": 7.17955377400052e-05, "loss": 0.0018, "step": 20400 }, { "epoch": 120.76923076923077, "grad_norm": 0.05603834241628647, "learning_rate": 7.16592686810265e-05, "loss": 0.0023, "step": 20410 }, { "epoch": 120.82840236686391, "grad_norm": 0.057538144290447235, "learning_rate": 7.152308846915529e-05, "loss": 0.0023, "step": 20420 }, { "epoch": 120.88757396449704, "grad_norm": 0.05249110981822014, "learning_rate": 7.138699725883515e-05, "loss": 0.0021, "step": 20430 }, { "epoch": 120.94674556213018, "grad_norm": 0.06490004807710648, "learning_rate": 7.125099520440867e-05, "loss": 0.0021, "step": 20440 }, { "epoch": 121.00591715976331, "grad_norm": 0.0694546028971672, "learning_rate": 7.11150824601173e-05, "loss": 0.0022, "step": 20450 }, { "epoch": 121.06508875739645, "grad_norm": 0.05743798241019249, "learning_rate": 7.097925918010122e-05, "loss": 0.002, "step": 20460 }, { "epoch": 121.12426035502959, "grad_norm": 0.041834987699985504, "learning_rate": 7.084352551839922e-05, "loss": 0.0019, "step": 20470 }, { "epoch": 121.18343195266272, "grad_norm": 0.056854862719774246, "learning_rate": 7.070788162894824e-05, "loss": 0.0018, "step": 20480 }, { "epoch": 121.24260355029585, "grad_norm": 0.05964423716068268, "learning_rate": 7.057232766558363e-05, "loss": 0.0022, "step": 20490 }, { "epoch": 121.30177514792899, "grad_norm": 0.04872424155473709, "learning_rate": 7.043686378203863e-05, "loss": 0.002, "step": 20500 }, { "epoch": 121.36094674556213, "grad_norm": 0.04790770635008812, "learning_rate": 7.030149013194439e-05, "loss": 0.0019, "step": 20510 }, { "epoch": 121.42011834319527, "grad_norm": 0.05787442997097969, "learning_rate": 7.016620686882974e-05, "loss": 0.0025, "step": 20520 }, { "epoch": 121.47928994082841, "grad_norm": 0.07746332138776779, "learning_rate": 7.003101414612086e-05, "loss": 0.0022, "step": 20530 }, { "epoch": 121.53846153846153, "grad_norm": 0.0508282445371151, "learning_rate": 6.989591211714137e-05, "loss": 0.002, "step": 20540 }, { "epoch": 121.59763313609467, "grad_norm": 0.04009231552481651, "learning_rate": 6.976090093511196e-05, "loss": 0.0022, "step": 20550 }, { "epoch": 121.65680473372781, "grad_norm": 0.06438944488763809, "learning_rate": 6.962598075315046e-05, "loss": 0.0019, "step": 20560 }, { "epoch": 121.71597633136095, "grad_norm": 0.06793520599603653, "learning_rate": 6.949115172427137e-05, "loss": 0.0021, "step": 20570 }, { "epoch": 121.77514792899409, "grad_norm": 0.042215052992105484, "learning_rate": 6.935641400138568e-05, "loss": 0.0023, "step": 20580 }, { "epoch": 121.83431952662721, "grad_norm": 0.0475960448384285, "learning_rate": 6.922176773730104e-05, "loss": 0.0025, "step": 20590 }, { "epoch": 121.89349112426035, "grad_norm": 0.08834762126207352, "learning_rate": 6.908721308472129e-05, "loss": 0.0025, "step": 20600 }, { "epoch": 121.95266272189349, "grad_norm": 0.04307479411363602, "learning_rate": 6.895275019624638e-05, "loss": 0.0019, "step": 20610 }, { "epoch": 122.01183431952663, "grad_norm": 0.07866130024194717, "learning_rate": 6.881837922437225e-05, "loss": 0.0023, "step": 20620 }, { "epoch": 122.07100591715977, "grad_norm": 0.06273554265499115, "learning_rate": 6.86841003214904e-05, "loss": 0.002, "step": 20630 }, { "epoch": 122.1301775147929, "grad_norm": 0.08036698400974274, "learning_rate": 6.854991363988815e-05, "loss": 0.0022, "step": 20640 }, { "epoch": 122.18934911242603, "grad_norm": 0.04816180095076561, "learning_rate": 6.841581933174808e-05, "loss": 0.002, "step": 20650 }, { "epoch": 122.24852071005917, "grad_norm": 0.046667296439409256, "learning_rate": 6.828181754914811e-05, "loss": 0.0023, "step": 20660 }, { "epoch": 122.3076923076923, "grad_norm": 0.04655566066503525, "learning_rate": 6.814790844406112e-05, "loss": 0.0021, "step": 20670 }, { "epoch": 122.36686390532545, "grad_norm": 0.04648800194263458, "learning_rate": 6.801409216835502e-05, "loss": 0.0024, "step": 20680 }, { "epoch": 122.42603550295858, "grad_norm": 0.04814242944121361, "learning_rate": 6.788036887379228e-05, "loss": 0.0018, "step": 20690 }, { "epoch": 122.48520710059172, "grad_norm": 0.04993079975247383, "learning_rate": 6.774673871202998e-05, "loss": 0.0021, "step": 20700 }, { "epoch": 122.54437869822485, "grad_norm": 0.045801181346178055, "learning_rate": 6.761320183461966e-05, "loss": 0.0018, "step": 20710 }, { "epoch": 122.60355029585799, "grad_norm": 0.039245691150426865, "learning_rate": 6.747975839300697e-05, "loss": 0.002, "step": 20720 }, { "epoch": 122.66272189349112, "grad_norm": 0.042196858674287796, "learning_rate": 6.734640853853163e-05, "loss": 0.0022, "step": 20730 }, { "epoch": 122.72189349112426, "grad_norm": 0.054433487355709076, "learning_rate": 6.721315242242727e-05, "loss": 0.0025, "step": 20740 }, { "epoch": 122.7810650887574, "grad_norm": 0.03174488991498947, "learning_rate": 6.707999019582103e-05, "loss": 0.0017, "step": 20750 }, { "epoch": 122.84023668639053, "grad_norm": 0.061033979058265686, "learning_rate": 6.694692200973372e-05, "loss": 0.002, "step": 20760 }, { "epoch": 122.89940828402366, "grad_norm": 0.052621934562921524, "learning_rate": 6.681394801507955e-05, "loss": 0.0019, "step": 20770 }, { "epoch": 122.9585798816568, "grad_norm": 0.06991100311279297, "learning_rate": 6.668106836266586e-05, "loss": 0.0021, "step": 20780 }, { "epoch": 123.01775147928994, "grad_norm": 0.08175712078809738, "learning_rate": 6.654828320319284e-05, "loss": 0.0021, "step": 20790 }, { "epoch": 123.07692307692308, "grad_norm": 0.0638357400894165, "learning_rate": 6.64155926872537e-05, "loss": 0.0018, "step": 20800 }, { "epoch": 123.13609467455622, "grad_norm": 0.06675494462251663, "learning_rate": 6.628299696533424e-05, "loss": 0.0019, "step": 20810 }, { "epoch": 123.19526627218934, "grad_norm": 0.0647672563791275, "learning_rate": 6.615049618781277e-05, "loss": 0.0023, "step": 20820 }, { "epoch": 123.25443786982248, "grad_norm": 0.04116766154766083, "learning_rate": 6.601809050495993e-05, "loss": 0.0016, "step": 20830 }, { "epoch": 123.31360946745562, "grad_norm": 0.0910431370139122, "learning_rate": 6.588578006693853e-05, "loss": 0.0024, "step": 20840 }, { "epoch": 123.37278106508876, "grad_norm": 0.03758574277162552, "learning_rate": 6.575356502380322e-05, "loss": 0.0025, "step": 20850 }, { "epoch": 123.4319526627219, "grad_norm": 0.04570990055799484, "learning_rate": 6.562144552550066e-05, "loss": 0.0017, "step": 20860 }, { "epoch": 123.49112426035504, "grad_norm": 0.056459520012140274, "learning_rate": 6.548942172186902e-05, "loss": 0.0019, "step": 20870 }, { "epoch": 123.55029585798816, "grad_norm": 0.060542356222867966, "learning_rate": 6.535749376263804e-05, "loss": 0.0018, "step": 20880 }, { "epoch": 123.6094674556213, "grad_norm": 0.06097676232457161, "learning_rate": 6.522566179742871e-05, "loss": 0.0019, "step": 20890 }, { "epoch": 123.66863905325444, "grad_norm": 0.03903541713953018, "learning_rate": 6.509392597575309e-05, "loss": 0.002, "step": 20900 }, { "epoch": 123.72781065088758, "grad_norm": 0.04549414664506912, "learning_rate": 6.496228644701425e-05, "loss": 0.002, "step": 20910 }, { "epoch": 123.78698224852072, "grad_norm": 0.08326653391122818, "learning_rate": 6.483074336050613e-05, "loss": 0.002, "step": 20920 }, { "epoch": 123.84615384615384, "grad_norm": 0.05142775923013687, "learning_rate": 6.46992968654132e-05, "loss": 0.0019, "step": 20930 }, { "epoch": 123.90532544378698, "grad_norm": 0.04355413466691971, "learning_rate": 6.456794711081041e-05, "loss": 0.0021, "step": 20940 }, { "epoch": 123.96449704142012, "grad_norm": 0.08736597001552582, "learning_rate": 6.443669424566309e-05, "loss": 0.0022, "step": 20950 }, { "epoch": 124.02366863905326, "grad_norm": 0.039797667413949966, "learning_rate": 6.430553841882643e-05, "loss": 0.0019, "step": 20960 }, { "epoch": 124.0828402366864, "grad_norm": 0.07359644770622253, "learning_rate": 6.417447977904577e-05, "loss": 0.002, "step": 20970 }, { "epoch": 124.14201183431953, "grad_norm": 0.052719295024871826, "learning_rate": 6.40435184749563e-05, "loss": 0.0021, "step": 20980 }, { "epoch": 124.20118343195266, "grad_norm": 0.050620656460523605, "learning_rate": 6.391265465508262e-05, "loss": 0.0021, "step": 20990 }, { "epoch": 124.2603550295858, "grad_norm": 0.0474432036280632, "learning_rate": 6.378188846783898e-05, "loss": 0.0021, "step": 21000 }, { "epoch": 124.31952662721893, "grad_norm": 0.0593186691403389, "learning_rate": 6.365122006152863e-05, "loss": 0.002, "step": 21010 }, { "epoch": 124.37869822485207, "grad_norm": 0.05178886651992798, "learning_rate": 6.352064958434417e-05, "loss": 0.0019, "step": 21020 }, { "epoch": 124.43786982248521, "grad_norm": 0.06171875819563866, "learning_rate": 6.339017718436705e-05, "loss": 0.0024, "step": 21030 }, { "epoch": 124.49704142011835, "grad_norm": 0.04756481945514679, "learning_rate": 6.325980300956748e-05, "loss": 0.0017, "step": 21040 }, { "epoch": 124.55621301775147, "grad_norm": 0.04722828045487404, "learning_rate": 6.312952720780432e-05, "loss": 0.0018, "step": 21050 }, { "epoch": 124.61538461538461, "grad_norm": 0.04636172950267792, "learning_rate": 6.299934992682476e-05, "loss": 0.0022, "step": 21060 }, { "epoch": 124.67455621301775, "grad_norm": 0.05540924891829491, "learning_rate": 6.286927131426436e-05, "loss": 0.0019, "step": 21070 }, { "epoch": 124.73372781065089, "grad_norm": 0.05665341764688492, "learning_rate": 6.273929151764673e-05, "loss": 0.0025, "step": 21080 }, { "epoch": 124.79289940828403, "grad_norm": 0.10203830897808075, "learning_rate": 6.260941068438344e-05, "loss": 0.0022, "step": 21090 }, { "epoch": 124.85207100591715, "grad_norm": 0.07808943092823029, "learning_rate": 6.247962896177379e-05, "loss": 0.002, "step": 21100 }, { "epoch": 124.91124260355029, "grad_norm": 0.05048368498682976, "learning_rate": 6.234994649700475e-05, "loss": 0.0021, "step": 21110 }, { "epoch": 124.97041420118343, "grad_norm": 0.09007931500673294, "learning_rate": 6.22203634371506e-05, "loss": 0.0022, "step": 21120 }, { "epoch": 125.02958579881657, "grad_norm": 0.035282328724861145, "learning_rate": 6.209087992917297e-05, "loss": 0.0018, "step": 21130 }, { "epoch": 125.08875739644971, "grad_norm": 0.04296840727329254, "learning_rate": 6.196149611992056e-05, "loss": 0.0021, "step": 21140 }, { "epoch": 125.14792899408285, "grad_norm": 0.04896401986479759, "learning_rate": 6.183221215612904e-05, "loss": 0.002, "step": 21150 }, { "epoch": 125.20710059171597, "grad_norm": 0.04426858201622963, "learning_rate": 6.170302818442078e-05, "loss": 0.0019, "step": 21160 }, { "epoch": 125.26627218934911, "grad_norm": 0.06298718601465225, "learning_rate": 6.157394435130485e-05, "loss": 0.0018, "step": 21170 }, { "epoch": 125.32544378698225, "grad_norm": 0.04755575209856033, "learning_rate": 6.144496080317654e-05, "loss": 0.002, "step": 21180 }, { "epoch": 125.38461538461539, "grad_norm": 0.03439943492412567, "learning_rate": 6.131607768631766e-05, "loss": 0.0018, "step": 21190 }, { "epoch": 125.44378698224853, "grad_norm": 0.07258612662553787, "learning_rate": 6.1187295146896e-05, "loss": 0.0018, "step": 21200 }, { "epoch": 125.50295857988165, "grad_norm": 0.051501236855983734, "learning_rate": 6.105861333096532e-05, "loss": 0.0018, "step": 21210 }, { "epoch": 125.56213017751479, "grad_norm": 0.04718717187643051, "learning_rate": 6.0930032384465024e-05, "loss": 0.0017, "step": 21220 }, { "epoch": 125.62130177514793, "grad_norm": 0.05589725449681282, "learning_rate": 6.080155245322025e-05, "loss": 0.0021, "step": 21230 }, { "epoch": 125.68047337278107, "grad_norm": 0.05301092937588692, "learning_rate": 6.067317368294157e-05, "loss": 0.0021, "step": 21240 }, { "epoch": 125.7396449704142, "grad_norm": 0.05186552181839943, "learning_rate": 6.0544896219224764e-05, "loss": 0.002, "step": 21250 }, { "epoch": 125.79881656804734, "grad_norm": 0.042757000774145126, "learning_rate": 6.041672020755079e-05, "loss": 0.0024, "step": 21260 }, { "epoch": 125.85798816568047, "grad_norm": 0.04730259254574776, "learning_rate": 6.0288645793285516e-05, "loss": 0.002, "step": 21270 }, { "epoch": 125.9171597633136, "grad_norm": 0.08994083851575851, "learning_rate": 6.016067312167949e-05, "loss": 0.0021, "step": 21280 }, { "epoch": 125.97633136094674, "grad_norm": 0.04578619450330734, "learning_rate": 6.003280233786805e-05, "loss": 0.0017, "step": 21290 }, { "epoch": 126.03550295857988, "grad_norm": 0.046632345765829086, "learning_rate": 5.9905033586870843e-05, "loss": 0.0018, "step": 21300 }, { "epoch": 126.09467455621302, "grad_norm": 0.045381736010313034, "learning_rate": 5.977736701359189e-05, "loss": 0.0017, "step": 21310 }, { "epoch": 126.15384615384616, "grad_norm": 0.039885830134153366, "learning_rate": 5.964980276281933e-05, "loss": 0.0017, "step": 21320 }, { "epoch": 126.21301775147928, "grad_norm": 0.06753651797771454, "learning_rate": 5.952234097922512e-05, "loss": 0.0019, "step": 21330 }, { "epoch": 126.27218934911242, "grad_norm": 0.048053693026304245, "learning_rate": 5.939498180736515e-05, "loss": 0.0017, "step": 21340 }, { "epoch": 126.33136094674556, "grad_norm": 0.054615817964076996, "learning_rate": 5.92677253916789e-05, "loss": 0.0021, "step": 21350 }, { "epoch": 126.3905325443787, "grad_norm": 0.05804522708058357, "learning_rate": 5.9140571876489306e-05, "loss": 0.002, "step": 21360 }, { "epoch": 126.44970414201184, "grad_norm": 0.035092778503894806, "learning_rate": 5.9013521406002615e-05, "loss": 0.0021, "step": 21370 }, { "epoch": 126.50887573964496, "grad_norm": 0.033655837178230286, "learning_rate": 5.888657412430819e-05, "loss": 0.0018, "step": 21380 }, { "epoch": 126.5680473372781, "grad_norm": 0.053497105836868286, "learning_rate": 5.875973017537837e-05, "loss": 0.0019, "step": 21390 }, { "epoch": 126.62721893491124, "grad_norm": 0.10461793094873428, "learning_rate": 5.863298970306831e-05, "loss": 0.002, "step": 21400 }, { "epoch": 126.68639053254438, "grad_norm": 0.054711222648620605, "learning_rate": 5.850635285111583e-05, "loss": 0.0021, "step": 21410 }, { "epoch": 126.74556213017752, "grad_norm": 0.05242713540792465, "learning_rate": 5.837981976314124e-05, "loss": 0.0026, "step": 21420 }, { "epoch": 126.80473372781066, "grad_norm": 0.03273192048072815, "learning_rate": 5.8253390582647055e-05, "loss": 0.0018, "step": 21430 }, { "epoch": 126.86390532544378, "grad_norm": 0.04216664284467697, "learning_rate": 5.812706545301809e-05, "loss": 0.002, "step": 21440 }, { "epoch": 126.92307692307692, "grad_norm": 0.03163101151585579, "learning_rate": 5.800084451752108e-05, "loss": 0.0017, "step": 21450 }, { "epoch": 126.98224852071006, "grad_norm": 0.04427748918533325, "learning_rate": 5.787472791930463e-05, "loss": 0.002, "step": 21460 }, { "epoch": 127.0414201183432, "grad_norm": 0.07135216146707535, "learning_rate": 5.7748715801398985e-05, "loss": 0.0022, "step": 21470 }, { "epoch": 127.10059171597634, "grad_norm": 0.06630060821771622, "learning_rate": 5.7622808306715985e-05, "loss": 0.002, "step": 21480 }, { "epoch": 127.15976331360947, "grad_norm": 0.043073151260614395, "learning_rate": 5.749700557804861e-05, "loss": 0.0017, "step": 21490 }, { "epoch": 127.2189349112426, "grad_norm": 0.05420130118727684, "learning_rate": 5.737130775807122e-05, "loss": 0.0021, "step": 21500 }, { "epoch": 127.27810650887574, "grad_norm": 0.04543440043926239, "learning_rate": 5.7245714989339136e-05, "loss": 0.002, "step": 21510 }, { "epoch": 127.33727810650888, "grad_norm": 0.04893835633993149, "learning_rate": 5.7120227414288513e-05, "loss": 0.002, "step": 21520 }, { "epoch": 127.39644970414201, "grad_norm": 0.06165354326367378, "learning_rate": 5.699484517523627e-05, "loss": 0.0021, "step": 21530 }, { "epoch": 127.45562130177515, "grad_norm": 0.06340143829584122, "learning_rate": 5.6869568414379815e-05, "loss": 0.0021, "step": 21540 }, { "epoch": 127.51479289940828, "grad_norm": 0.03612833842635155, "learning_rate": 5.674439727379688e-05, "loss": 0.0018, "step": 21550 }, { "epoch": 127.57396449704142, "grad_norm": 0.035359643399715424, "learning_rate": 5.6619331895445474e-05, "loss": 0.0016, "step": 21560 }, { "epoch": 127.63313609467455, "grad_norm": 0.05916391313076019, "learning_rate": 5.64943724211637e-05, "loss": 0.0021, "step": 21570 }, { "epoch": 127.6923076923077, "grad_norm": 0.0923391655087471, "learning_rate": 5.636951899266948e-05, "loss": 0.0019, "step": 21580 }, { "epoch": 127.75147928994083, "grad_norm": 0.057559527456760406, "learning_rate": 5.624477175156051e-05, "loss": 0.0022, "step": 21590 }, { "epoch": 127.81065088757397, "grad_norm": 0.05210212245583534, "learning_rate": 5.6120130839314024e-05, "loss": 0.0019, "step": 21600 }, { "epoch": 127.8698224852071, "grad_norm": 0.045807477086782455, "learning_rate": 5.5995596397286715e-05, "loss": 0.0019, "step": 21610 }, { "epoch": 127.92899408284023, "grad_norm": 0.05312110483646393, "learning_rate": 5.5871168566714495e-05, "loss": 0.0019, "step": 21620 }, { "epoch": 127.98816568047337, "grad_norm": 0.05028782784938812, "learning_rate": 5.574684748871235e-05, "loss": 0.0021, "step": 21630 }, { "epoch": 128.0473372781065, "grad_norm": 0.07222411781549454, "learning_rate": 5.562263330427428e-05, "loss": 0.0019, "step": 21640 }, { "epoch": 128.10650887573965, "grad_norm": 0.0346946157515049, "learning_rate": 5.54985261542729e-05, "loss": 0.0021, "step": 21650 }, { "epoch": 128.1656804733728, "grad_norm": 0.027660422027111053, "learning_rate": 5.537452617945956e-05, "loss": 0.0015, "step": 21660 }, { "epoch": 128.22485207100593, "grad_norm": 0.051665809005498886, "learning_rate": 5.525063352046406e-05, "loss": 0.0022, "step": 21670 }, { "epoch": 128.28402366863907, "grad_norm": 0.05486876517534256, "learning_rate": 5.5126848317794436e-05, "loss": 0.0021, "step": 21680 }, { "epoch": 128.34319526627218, "grad_norm": 0.04863801226019859, "learning_rate": 5.5003170711836934e-05, "loss": 0.0017, "step": 21690 }, { "epoch": 128.4023668639053, "grad_norm": 0.05758386105298996, "learning_rate": 5.487960084285566e-05, "loss": 0.0016, "step": 21700 }, { "epoch": 128.46153846153845, "grad_norm": 0.04844481125473976, "learning_rate": 5.47561388509926e-05, "loss": 0.0021, "step": 21710 }, { "epoch": 128.5207100591716, "grad_norm": 0.046950407326221466, "learning_rate": 5.4632784876267415e-05, "loss": 0.0025, "step": 21720 }, { "epoch": 128.57988165680473, "grad_norm": 0.05732760205864906, "learning_rate": 5.450953905857724e-05, "loss": 0.0017, "step": 21730 }, { "epoch": 128.63905325443787, "grad_norm": 0.051452841609716415, "learning_rate": 5.4386401537696536e-05, "loss": 0.002, "step": 21740 }, { "epoch": 128.698224852071, "grad_norm": 0.05886024981737137, "learning_rate": 5.426337245327702e-05, "loss": 0.0019, "step": 21750 }, { "epoch": 128.75739644970415, "grad_norm": 0.044475190341472626, "learning_rate": 5.4140451944847276e-05, "loss": 0.0025, "step": 21760 }, { "epoch": 128.81656804733728, "grad_norm": 0.06444964557886124, "learning_rate": 5.4017640151812846e-05, "loss": 0.0019, "step": 21770 }, { "epoch": 128.87573964497042, "grad_norm": 0.0469147190451622, "learning_rate": 5.389493721345601e-05, "loss": 0.0018, "step": 21780 }, { "epoch": 128.93491124260356, "grad_norm": 0.04141249507665634, "learning_rate": 5.377234326893554e-05, "loss": 0.0015, "step": 21790 }, { "epoch": 128.9940828402367, "grad_norm": 0.0665765181183815, "learning_rate": 5.3649858457286615e-05, "loss": 0.0028, "step": 21800 }, { "epoch": 129.0532544378698, "grad_norm": 0.0558018833398819, "learning_rate": 5.352748291742064e-05, "loss": 0.002, "step": 21810 }, { "epoch": 129.11242603550295, "grad_norm": 0.039439499378204346, "learning_rate": 5.340521678812511e-05, "loss": 0.0021, "step": 21820 }, { "epoch": 129.1715976331361, "grad_norm": 0.04224696755409241, "learning_rate": 5.3283060208063413e-05, "loss": 0.002, "step": 21830 }, { "epoch": 129.23076923076923, "grad_norm": 0.062385112047195435, "learning_rate": 5.316101331577469e-05, "loss": 0.002, "step": 21840 }, { "epoch": 129.28994082840237, "grad_norm": 0.03680998086929321, "learning_rate": 5.303907624967377e-05, "loss": 0.0018, "step": 21850 }, { "epoch": 129.3491124260355, "grad_norm": 0.04078696668148041, "learning_rate": 5.291724914805073e-05, "loss": 0.0018, "step": 21860 }, { "epoch": 129.40828402366864, "grad_norm": 0.04901675507426262, "learning_rate": 5.279553214907115e-05, "loss": 0.0019, "step": 21870 }, { "epoch": 129.46745562130178, "grad_norm": 0.03571197763085365, "learning_rate": 5.267392539077562e-05, "loss": 0.0017, "step": 21880 }, { "epoch": 129.52662721893492, "grad_norm": 0.035126589238643646, "learning_rate": 5.255242901107974e-05, "loss": 0.0019, "step": 21890 }, { "epoch": 129.58579881656806, "grad_norm": 0.055164359509944916, "learning_rate": 5.243104314777394e-05, "loss": 0.0017, "step": 21900 }, { "epoch": 129.6449704142012, "grad_norm": 0.0466490313410759, "learning_rate": 5.230976793852336e-05, "loss": 0.0017, "step": 21910 }, { "epoch": 129.7041420118343, "grad_norm": 0.055237140506505966, "learning_rate": 5.2188603520867446e-05, "loss": 0.0018, "step": 21920 }, { "epoch": 129.76331360946745, "grad_norm": 0.03638710081577301, "learning_rate": 5.206755003222019e-05, "loss": 0.0018, "step": 21930 }, { "epoch": 129.82248520710058, "grad_norm": 0.03715963289141655, "learning_rate": 5.194660760986975e-05, "loss": 0.0017, "step": 21940 }, { "epoch": 129.88165680473372, "grad_norm": 0.0461251400411129, "learning_rate": 5.182577639097826e-05, "loss": 0.0017, "step": 21950 }, { "epoch": 129.94082840236686, "grad_norm": 0.04308957979083061, "learning_rate": 5.170505651258178e-05, "loss": 0.0015, "step": 21960 }, { "epoch": 130.0, "grad_norm": 0.06164119392633438, "learning_rate": 5.158444811159014e-05, "loss": 0.0021, "step": 21970 }, { "epoch": 130.05917159763314, "grad_norm": 0.04718993231654167, "learning_rate": 5.146395132478659e-05, "loss": 0.0015, "step": 21980 }, { "epoch": 130.11834319526628, "grad_norm": 0.04754805192351341, "learning_rate": 5.1343566288827936e-05, "loss": 0.0018, "step": 21990 }, { "epoch": 130.17751479289942, "grad_norm": 0.030281290411949158, "learning_rate": 5.1223293140244216e-05, "loss": 0.002, "step": 22000 }, { "epoch": 130.23668639053255, "grad_norm": 0.02968057431280613, "learning_rate": 5.110313201543856e-05, "loss": 0.0019, "step": 22010 }, { "epoch": 130.2958579881657, "grad_norm": 0.04921477288007736, "learning_rate": 5.098308305068706e-05, "loss": 0.0017, "step": 22020 }, { "epoch": 130.3550295857988, "grad_norm": 0.03919064253568649, "learning_rate": 5.0863146382138586e-05, "loss": 0.0017, "step": 22030 }, { "epoch": 130.41420118343194, "grad_norm": 0.03507036715745926, "learning_rate": 5.0743322145814695e-05, "loss": 0.0016, "step": 22040 }, { "epoch": 130.47337278106508, "grad_norm": 0.04633102938532829, "learning_rate": 5.062361047760938e-05, "loss": 0.002, "step": 22050 }, { "epoch": 130.53254437869822, "grad_norm": 0.046705082058906555, "learning_rate": 5.050401151328903e-05, "loss": 0.0018, "step": 22060 }, { "epoch": 130.59171597633136, "grad_norm": 0.05435183271765709, "learning_rate": 5.038452538849218e-05, "loss": 0.0016, "step": 22070 }, { "epoch": 130.6508875739645, "grad_norm": 0.03656042739748955, "learning_rate": 5.026515223872933e-05, "loss": 0.0017, "step": 22080 }, { "epoch": 130.71005917159763, "grad_norm": 0.0602632611989975, "learning_rate": 5.0145892199382945e-05, "loss": 0.0022, "step": 22090 }, { "epoch": 130.76923076923077, "grad_norm": 0.06365057826042175, "learning_rate": 5.0026745405707196e-05, "loss": 0.0016, "step": 22100 }, { "epoch": 130.8284023668639, "grad_norm": 0.10914278030395508, "learning_rate": 4.990771199282782e-05, "loss": 0.002, "step": 22110 }, { "epoch": 130.88757396449705, "grad_norm": 0.03729139268398285, "learning_rate": 4.9788792095741955e-05, "loss": 0.0017, "step": 22120 }, { "epoch": 130.9467455621302, "grad_norm": 0.07558633387088776, "learning_rate": 4.966998584931796e-05, "loss": 0.0018, "step": 22130 }, { "epoch": 131.0059171597633, "grad_norm": 0.04769191890954971, "learning_rate": 4.955129338829535e-05, "loss": 0.0017, "step": 22140 }, { "epoch": 131.06508875739644, "grad_norm": 0.04565104842185974, "learning_rate": 4.9432714847284594e-05, "loss": 0.0016, "step": 22150 }, { "epoch": 131.12426035502958, "grad_norm": 0.06653694808483124, "learning_rate": 4.931425036076694e-05, "loss": 0.0018, "step": 22160 }, { "epoch": 131.18343195266272, "grad_norm": 0.04879417642951012, "learning_rate": 4.9195900063094327e-05, "loss": 0.0018, "step": 22170 }, { "epoch": 131.24260355029585, "grad_norm": 0.0590350404381752, "learning_rate": 4.90776640884892e-05, "loss": 0.0019, "step": 22180 }, { "epoch": 131.301775147929, "grad_norm": 0.03915472701191902, "learning_rate": 4.89595425710442e-05, "loss": 0.0018, "step": 22190 }, { "epoch": 131.36094674556213, "grad_norm": 0.05775580555200577, "learning_rate": 4.8841535644722345e-05, "loss": 0.0016, "step": 22200 }, { "epoch": 131.42011834319527, "grad_norm": 0.09066729992628098, "learning_rate": 4.8723643443356574e-05, "loss": 0.0019, "step": 22210 }, { "epoch": 131.4792899408284, "grad_norm": 0.04501336067914963, "learning_rate": 4.8605866100649885e-05, "loss": 0.0015, "step": 22220 }, { "epoch": 131.53846153846155, "grad_norm": 0.04995611310005188, "learning_rate": 4.848820375017475e-05, "loss": 0.0015, "step": 22230 }, { "epoch": 131.5976331360947, "grad_norm": 0.05692601576447487, "learning_rate": 4.8370656525373446e-05, "loss": 0.0016, "step": 22240 }, { "epoch": 131.65680473372782, "grad_norm": 0.03752797469496727, "learning_rate": 4.825322455955758e-05, "loss": 0.0016, "step": 22250 }, { "epoch": 131.71597633136093, "grad_norm": 0.031230047345161438, "learning_rate": 4.8135907985908065e-05, "loss": 0.0021, "step": 22260 }, { "epoch": 131.77514792899407, "grad_norm": 0.04020587354898453, "learning_rate": 4.801870693747498e-05, "loss": 0.0019, "step": 22270 }, { "epoch": 131.8343195266272, "grad_norm": 0.04440361633896828, "learning_rate": 4.790162154717738e-05, "loss": 0.0019, "step": 22280 }, { "epoch": 131.89349112426035, "grad_norm": 0.050527893006801605, "learning_rate": 4.778465194780305e-05, "loss": 0.0017, "step": 22290 }, { "epoch": 131.9526627218935, "grad_norm": 0.03868500515818596, "learning_rate": 4.766779827200857e-05, "loss": 0.002, "step": 22300 }, { "epoch": 132.01183431952663, "grad_norm": 0.05174349620938301, "learning_rate": 4.7551060652318995e-05, "loss": 0.0018, "step": 22310 }, { "epoch": 132.07100591715977, "grad_norm": 0.027957985177636147, "learning_rate": 4.743443922112781e-05, "loss": 0.0017, "step": 22320 }, { "epoch": 132.1301775147929, "grad_norm": 0.06282690912485123, "learning_rate": 4.7317934110696685e-05, "loss": 0.0019, "step": 22330 }, { "epoch": 132.18934911242604, "grad_norm": 0.0404733307659626, "learning_rate": 4.7201545453155425e-05, "loss": 0.0017, "step": 22340 }, { "epoch": 132.24852071005918, "grad_norm": 0.03588014468550682, "learning_rate": 4.708527338050161e-05, "loss": 0.0018, "step": 22350 }, { "epoch": 132.30769230769232, "grad_norm": 0.042847614735364914, "learning_rate": 4.696911802460079e-05, "loss": 0.0017, "step": 22360 }, { "epoch": 132.36686390532543, "grad_norm": 0.048524077981710434, "learning_rate": 4.6853079517186035e-05, "loss": 0.0016, "step": 22370 }, { "epoch": 132.42603550295857, "grad_norm": 0.05757267028093338, "learning_rate": 4.6737157989857935e-05, "loss": 0.0016, "step": 22380 }, { "epoch": 132.4852071005917, "grad_norm": 0.04221310466527939, "learning_rate": 4.662135357408447e-05, "loss": 0.0019, "step": 22390 }, { "epoch": 132.54437869822485, "grad_norm": 0.04761277884244919, "learning_rate": 4.650566640120061e-05, "loss": 0.0019, "step": 22400 }, { "epoch": 132.60355029585799, "grad_norm": 0.06231418251991272, "learning_rate": 4.6390096602408555e-05, "loss": 0.0018, "step": 22410 }, { "epoch": 132.66272189349112, "grad_norm": 0.04004287347197533, "learning_rate": 4.6274644308777255e-05, "loss": 0.0015, "step": 22420 }, { "epoch": 132.72189349112426, "grad_norm": 0.047905970364809036, "learning_rate": 4.615930965124255e-05, "loss": 0.002, "step": 22430 }, { "epoch": 132.7810650887574, "grad_norm": 0.042004186660051346, "learning_rate": 4.6044092760606794e-05, "loss": 0.0015, "step": 22440 }, { "epoch": 132.84023668639054, "grad_norm": 0.05522860214114189, "learning_rate": 4.592899376753865e-05, "loss": 0.0021, "step": 22450 }, { "epoch": 132.89940828402368, "grad_norm": 0.06543130427598953, "learning_rate": 4.581401280257326e-05, "loss": 0.0017, "step": 22460 }, { "epoch": 132.95857988165682, "grad_norm": 0.048504047095775604, "learning_rate": 4.56991499961118e-05, "loss": 0.0015, "step": 22470 }, { "epoch": 133.01775147928993, "grad_norm": 0.08238343149423599, "learning_rate": 4.5584405478421506e-05, "loss": 0.0018, "step": 22480 }, { "epoch": 133.07692307692307, "grad_norm": 0.05043059214949608, "learning_rate": 4.5469779379635456e-05, "loss": 0.0018, "step": 22490 }, { "epoch": 133.1360946745562, "grad_norm": 0.03330621495842934, "learning_rate": 4.5355271829752306e-05, "loss": 0.002, "step": 22500 }, { "epoch": 133.19526627218934, "grad_norm": 0.051777031272649765, "learning_rate": 4.524088295863643e-05, "loss": 0.0019, "step": 22510 }, { "epoch": 133.25443786982248, "grad_norm": 0.04920438677072525, "learning_rate": 4.5126612896017515e-05, "loss": 0.0022, "step": 22520 }, { "epoch": 133.31360946745562, "grad_norm": 0.05067408084869385, "learning_rate": 4.501246177149056e-05, "loss": 0.0017, "step": 22530 }, { "epoch": 133.37278106508876, "grad_norm": 0.04233356937766075, "learning_rate": 4.489842971451561e-05, "loss": 0.0014, "step": 22540 }, { "epoch": 133.4319526627219, "grad_norm": 0.04471738636493683, "learning_rate": 4.478451685441775e-05, "loss": 0.0017, "step": 22550 }, { "epoch": 133.49112426035504, "grad_norm": 0.03088744729757309, "learning_rate": 4.467072332038678e-05, "loss": 0.0017, "step": 22560 }, { "epoch": 133.55029585798817, "grad_norm": 0.05071621015667915, "learning_rate": 4.455704924147726e-05, "loss": 0.0015, "step": 22570 }, { "epoch": 133.6094674556213, "grad_norm": 0.03717409819364548, "learning_rate": 4.4443494746608246e-05, "loss": 0.0018, "step": 22580 }, { "epoch": 133.66863905325442, "grad_norm": 0.047507788985967636, "learning_rate": 4.433005996456317e-05, "loss": 0.0016, "step": 22590 }, { "epoch": 133.72781065088756, "grad_norm": 0.043986424803733826, "learning_rate": 4.421674502398971e-05, "loss": 0.0015, "step": 22600 }, { "epoch": 133.7869822485207, "grad_norm": 0.039460182189941406, "learning_rate": 4.4103550053399666e-05, "loss": 0.0017, "step": 22610 }, { "epoch": 133.84615384615384, "grad_norm": 0.051089633256196976, "learning_rate": 4.399047518116861e-05, "loss": 0.0017, "step": 22620 }, { "epoch": 133.90532544378698, "grad_norm": 0.04627181962132454, "learning_rate": 4.387752053553606e-05, "loss": 0.0017, "step": 22630 }, { "epoch": 133.96449704142012, "grad_norm": 0.05010220408439636, "learning_rate": 4.376468624460524e-05, "loss": 0.0015, "step": 22640 }, { "epoch": 134.02366863905326, "grad_norm": 0.047386184334754944, "learning_rate": 4.3651972436342794e-05, "loss": 0.0019, "step": 22650 }, { "epoch": 134.0828402366864, "grad_norm": 0.04099619388580322, "learning_rate": 4.353937923857863e-05, "loss": 0.0017, "step": 22660 }, { "epoch": 134.14201183431953, "grad_norm": 0.04102177545428276, "learning_rate": 4.3426906779005995e-05, "loss": 0.0019, "step": 22670 }, { "epoch": 134.20118343195267, "grad_norm": 0.07624711841344833, "learning_rate": 4.331455518518121e-05, "loss": 0.0016, "step": 22680 }, { "epoch": 134.2603550295858, "grad_norm": 0.055269140750169754, "learning_rate": 4.320232458452344e-05, "loss": 0.0016, "step": 22690 }, { "epoch": 134.31952662721895, "grad_norm": 0.03876530006527901, "learning_rate": 4.309021510431473e-05, "loss": 0.0016, "step": 22700 }, { "epoch": 134.37869822485206, "grad_norm": 0.03506304323673248, "learning_rate": 4.2978226871699704e-05, "loss": 0.0015, "step": 22710 }, { "epoch": 134.4378698224852, "grad_norm": 0.04303513839840889, "learning_rate": 4.286636001368541e-05, "loss": 0.0017, "step": 22720 }, { "epoch": 134.49704142011834, "grad_norm": 0.03270938619971275, "learning_rate": 4.2754614657141365e-05, "loss": 0.0018, "step": 22730 }, { "epoch": 134.55621301775147, "grad_norm": 0.05647130310535431, "learning_rate": 4.2642990928799224e-05, "loss": 0.0017, "step": 22740 }, { "epoch": 134.6153846153846, "grad_norm": 0.06523936986923218, "learning_rate": 4.253148895525272e-05, "loss": 0.0016, "step": 22750 }, { "epoch": 134.67455621301775, "grad_norm": 0.028950780630111694, "learning_rate": 4.242010886295752e-05, "loss": 0.0018, "step": 22760 }, { "epoch": 134.7337278106509, "grad_norm": 0.039933815598487854, "learning_rate": 4.230885077823107e-05, "loss": 0.0015, "step": 22770 }, { "epoch": 134.79289940828403, "grad_norm": 0.08145977556705475, "learning_rate": 4.219771482725233e-05, "loss": 0.0017, "step": 22780 }, { "epoch": 134.85207100591717, "grad_norm": 0.033432673662900925, "learning_rate": 4.208670113606189e-05, "loss": 0.0017, "step": 22790 }, { "epoch": 134.9112426035503, "grad_norm": 0.060280799865722656, "learning_rate": 4.1975809830561616e-05, "loss": 0.0019, "step": 22800 }, { "epoch": 134.97041420118344, "grad_norm": 0.03735312819480896, "learning_rate": 4.18650410365146e-05, "loss": 0.0016, "step": 22810 }, { "epoch": 135.02958579881656, "grad_norm": 0.042667172849178314, "learning_rate": 4.175439487954502e-05, "loss": 0.0016, "step": 22820 }, { "epoch": 135.0887573964497, "grad_norm": 0.046687036752700806, "learning_rate": 4.164387148513787e-05, "loss": 0.0018, "step": 22830 }, { "epoch": 135.14792899408283, "grad_norm": 0.03855249658226967, "learning_rate": 4.1533470978638925e-05, "loss": 0.0015, "step": 22840 }, { "epoch": 135.20710059171597, "grad_norm": 0.040881406515836716, "learning_rate": 4.142319348525477e-05, "loss": 0.0014, "step": 22850 }, { "epoch": 135.2662721893491, "grad_norm": 0.033845093101263046, "learning_rate": 4.131303913005231e-05, "loss": 0.0016, "step": 22860 }, { "epoch": 135.32544378698225, "grad_norm": 0.041687749326229095, "learning_rate": 4.120300803795889e-05, "loss": 0.0013, "step": 22870 }, { "epoch": 135.3846153846154, "grad_norm": 0.046170759946107864, "learning_rate": 4.10931003337619e-05, "loss": 0.0015, "step": 22880 }, { "epoch": 135.44378698224853, "grad_norm": 0.0328582301735878, "learning_rate": 4.098331614210899e-05, "loss": 0.0017, "step": 22890 }, { "epoch": 135.50295857988166, "grad_norm": 0.054373323917388916, "learning_rate": 4.087365558750761e-05, "loss": 0.0016, "step": 22900 }, { "epoch": 135.5621301775148, "grad_norm": 0.0374244749546051, "learning_rate": 4.0764118794325056e-05, "loss": 0.0016, "step": 22910 }, { "epoch": 135.62130177514794, "grad_norm": 0.06135968118906021, "learning_rate": 4.06547058867883e-05, "loss": 0.0019, "step": 22920 }, { "epoch": 135.68047337278105, "grad_norm": 0.04899599030613899, "learning_rate": 4.054541698898364e-05, "loss": 0.0017, "step": 22930 }, { "epoch": 135.7396449704142, "grad_norm": 0.03836223855614662, "learning_rate": 4.043625222485688e-05, "loss": 0.0014, "step": 22940 }, { "epoch": 135.79881656804733, "grad_norm": 0.04901210963726044, "learning_rate": 4.032721171821306e-05, "loss": 0.0019, "step": 22950 }, { "epoch": 135.85798816568047, "grad_norm": 0.06058196723461151, "learning_rate": 4.0218295592716205e-05, "loss": 0.0017, "step": 22960 }, { "epoch": 135.9171597633136, "grad_norm": 0.04351752623915672, "learning_rate": 4.010950397188934e-05, "loss": 0.0015, "step": 22970 }, { "epoch": 135.97633136094674, "grad_norm": 0.0304864551872015, "learning_rate": 4.000083697911432e-05, "loss": 0.0018, "step": 22980 }, { "epoch": 136.03550295857988, "grad_norm": 0.03557174280285835, "learning_rate": 3.9892294737631494e-05, "loss": 0.0016, "step": 22990 }, { "epoch": 136.09467455621302, "grad_norm": 0.02683299407362938, "learning_rate": 3.978387737053994e-05, "loss": 0.0016, "step": 23000 }, { "epoch": 136.15384615384616, "grad_norm": 0.03646858409047127, "learning_rate": 3.9675585000796945e-05, "loss": 0.0016, "step": 23010 }, { "epoch": 136.2130177514793, "grad_norm": 0.042090706527233124, "learning_rate": 3.956741775121817e-05, "loss": 0.0014, "step": 23020 }, { "epoch": 136.27218934911244, "grad_norm": 0.054491233080625534, "learning_rate": 3.945937574447729e-05, "loss": 0.0016, "step": 23030 }, { "epoch": 136.33136094674558, "grad_norm": 0.0491589792072773, "learning_rate": 3.9351459103106e-05, "loss": 0.0019, "step": 23040 }, { "epoch": 136.3905325443787, "grad_norm": 0.06502412259578705, "learning_rate": 3.924366794949366e-05, "loss": 0.0019, "step": 23050 }, { "epoch": 136.44970414201183, "grad_norm": 0.054066140204668045, "learning_rate": 3.913600240588756e-05, "loss": 0.0016, "step": 23060 }, { "epoch": 136.50887573964496, "grad_norm": 0.1907099485397339, "learning_rate": 3.902846259439236e-05, "loss": 0.0016, "step": 23070 }, { "epoch": 136.5680473372781, "grad_norm": 0.032864365726709366, "learning_rate": 3.8921048636970233e-05, "loss": 0.0014, "step": 23080 }, { "epoch": 136.62721893491124, "grad_norm": 0.1045742854475975, "learning_rate": 3.881376065544047e-05, "loss": 0.0017, "step": 23090 }, { "epoch": 136.68639053254438, "grad_norm": 0.03032064437866211, "learning_rate": 3.870659877147962e-05, "loss": 0.0019, "step": 23100 }, { "epoch": 136.74556213017752, "grad_norm": 0.041283804923295975, "learning_rate": 3.859956310662117e-05, "loss": 0.0018, "step": 23110 }, { "epoch": 136.80473372781066, "grad_norm": 0.057955943048000336, "learning_rate": 3.84926537822555e-05, "loss": 0.0015, "step": 23120 }, { "epoch": 136.8639053254438, "grad_norm": 0.039511799812316895, "learning_rate": 3.838587091962966e-05, "loss": 0.0016, "step": 23130 }, { "epoch": 136.92307692307693, "grad_norm": 0.03427623584866524, "learning_rate": 3.827921463984738e-05, "loss": 0.0016, "step": 23140 }, { "epoch": 136.98224852071007, "grad_norm": 0.04554404690861702, "learning_rate": 3.8172685063868616e-05, "loss": 0.0017, "step": 23150 }, { "epoch": 137.04142011834318, "grad_norm": 0.05487753450870514, "learning_rate": 3.806628231250984e-05, "loss": 0.0018, "step": 23160 }, { "epoch": 137.10059171597632, "grad_norm": 0.05357057973742485, "learning_rate": 3.796000650644359e-05, "loss": 0.0014, "step": 23170 }, { "epoch": 137.15976331360946, "grad_norm": 0.029112862423062325, "learning_rate": 3.7853857766198495e-05, "loss": 0.0017, "step": 23180 }, { "epoch": 137.2189349112426, "grad_norm": 0.036127179861068726, "learning_rate": 3.7747836212159036e-05, "loss": 0.0019, "step": 23190 }, { "epoch": 137.27810650887574, "grad_norm": 0.05729967728257179, "learning_rate": 3.7641941964565384e-05, "loss": 0.0015, "step": 23200 }, { "epoch": 137.33727810650888, "grad_norm": 0.05086370185017586, "learning_rate": 3.7536175143513456e-05, "loss": 0.0015, "step": 23210 }, { "epoch": 137.39644970414201, "grad_norm": 0.04554115980863571, "learning_rate": 3.7430535868954596e-05, "loss": 0.0019, "step": 23220 }, { "epoch": 137.45562130177515, "grad_norm": 0.07729918509721756, "learning_rate": 3.732502426069548e-05, "loss": 0.0017, "step": 23230 }, { "epoch": 137.5147928994083, "grad_norm": 0.043563734740018845, "learning_rate": 3.721964043839804e-05, "loss": 0.0016, "step": 23240 }, { "epoch": 137.57396449704143, "grad_norm": 0.035153329372406006, "learning_rate": 3.711438452157923e-05, "loss": 0.0017, "step": 23250 }, { "epoch": 137.63313609467457, "grad_norm": 0.029928619042038918, "learning_rate": 3.700925662961097e-05, "loss": 0.0018, "step": 23260 }, { "epoch": 137.69230769230768, "grad_norm": 0.06686288863420486, "learning_rate": 3.6904256881720035e-05, "loss": 0.0017, "step": 23270 }, { "epoch": 137.75147928994082, "grad_norm": 0.03642739728093147, "learning_rate": 3.679938539698777e-05, "loss": 0.0015, "step": 23280 }, { "epoch": 137.81065088757396, "grad_norm": 0.035041797906160355, "learning_rate": 3.6694642294350184e-05, "loss": 0.0016, "step": 23290 }, { "epoch": 137.8698224852071, "grad_norm": 0.054321322590112686, "learning_rate": 3.659002769259753e-05, "loss": 0.0015, "step": 23300 }, { "epoch": 137.92899408284023, "grad_norm": 0.050900623202323914, "learning_rate": 3.6485541710374434e-05, "loss": 0.0015, "step": 23310 }, { "epoch": 137.98816568047337, "grad_norm": 0.047637950628995895, "learning_rate": 3.6381184466179625e-05, "loss": 0.0015, "step": 23320 }, { "epoch": 138.0473372781065, "grad_norm": 0.05428234487771988, "learning_rate": 3.627695607836583e-05, "loss": 0.0015, "step": 23330 }, { "epoch": 138.10650887573965, "grad_norm": 0.048454053699970245, "learning_rate": 3.617285666513966e-05, "loss": 0.0015, "step": 23340 }, { "epoch": 138.1656804733728, "grad_norm": 0.07702865451574326, "learning_rate": 3.606888634456149e-05, "loss": 0.0016, "step": 23350 }, { "epoch": 138.22485207100593, "grad_norm": 0.028316933661699295, "learning_rate": 3.596504523454509e-05, "loss": 0.0016, "step": 23360 }, { "epoch": 138.28402366863907, "grad_norm": 0.0628032237291336, "learning_rate": 3.5861333452857944e-05, "loss": 0.0016, "step": 23370 }, { "epoch": 138.34319526627218, "grad_norm": 0.05567818507552147, "learning_rate": 3.57577511171207e-05, "loss": 0.0014, "step": 23380 }, { "epoch": 138.4023668639053, "grad_norm": 0.03601379320025444, "learning_rate": 3.565429834480729e-05, "loss": 0.0015, "step": 23390 }, { "epoch": 138.46153846153845, "grad_norm": 0.03924316167831421, "learning_rate": 3.555097525324464e-05, "loss": 0.0014, "step": 23400 }, { "epoch": 138.5207100591716, "grad_norm": 0.047452062368392944, "learning_rate": 3.544778195961272e-05, "loss": 0.0015, "step": 23410 }, { "epoch": 138.57988165680473, "grad_norm": 0.04767373576760292, "learning_rate": 3.5344718580944076e-05, "loss": 0.0016, "step": 23420 }, { "epoch": 138.63905325443787, "grad_norm": 0.20169080793857574, "learning_rate": 3.5241785234124126e-05, "loss": 0.0017, "step": 23430 }, { "epoch": 138.698224852071, "grad_norm": 0.0416417121887207, "learning_rate": 3.513898203589073e-05, "loss": 0.0016, "step": 23440 }, { "epoch": 138.75739644970415, "grad_norm": 0.11574725806713104, "learning_rate": 3.503630910283416e-05, "loss": 0.0021, "step": 23450 }, { "epoch": 138.81656804733728, "grad_norm": 0.050227534025907516, "learning_rate": 3.493376655139692e-05, "loss": 0.0015, "step": 23460 }, { "epoch": 138.87573964497042, "grad_norm": 0.056714244186878204, "learning_rate": 3.48313544978737e-05, "loss": 0.0018, "step": 23470 }, { "epoch": 138.93491124260356, "grad_norm": 0.054489172995090485, "learning_rate": 3.472907305841118e-05, "loss": 0.0014, "step": 23480 }, { "epoch": 138.9940828402367, "grad_norm": 0.04498706012964249, "learning_rate": 3.4626922349007856e-05, "loss": 0.0015, "step": 23490 }, { "epoch": 139.0532544378698, "grad_norm": 0.04432012513279915, "learning_rate": 3.452490248551404e-05, "loss": 0.0018, "step": 23500 }, { "epoch": 139.11242603550295, "grad_norm": 0.06371647119522095, "learning_rate": 3.442301358363163e-05, "loss": 0.0015, "step": 23510 }, { "epoch": 139.1715976331361, "grad_norm": 0.09160295873880386, "learning_rate": 3.4321255758913904e-05, "loss": 0.0016, "step": 23520 }, { "epoch": 139.23076923076923, "grad_norm": 0.06643464416265488, "learning_rate": 3.421962912676559e-05, "loss": 0.002, "step": 23530 }, { "epoch": 139.28994082840237, "grad_norm": 0.04723633825778961, "learning_rate": 3.41181338024426e-05, "loss": 0.0014, "step": 23540 }, { "epoch": 139.3491124260355, "grad_norm": 0.04251872003078461, "learning_rate": 3.401676990105195e-05, "loss": 0.0013, "step": 23550 }, { "epoch": 139.40828402366864, "grad_norm": 0.043246444314718246, "learning_rate": 3.391553753755155e-05, "loss": 0.0018, "step": 23560 }, { "epoch": 139.46745562130178, "grad_norm": 0.027693504467606544, "learning_rate": 3.381443682675023e-05, "loss": 0.0016, "step": 23570 }, { "epoch": 139.52662721893492, "grad_norm": 0.042332958430051804, "learning_rate": 3.371346788330737e-05, "loss": 0.0015, "step": 23580 }, { "epoch": 139.58579881656806, "grad_norm": 0.04451478645205498, "learning_rate": 3.361263082173301e-05, "loss": 0.0014, "step": 23590 }, { "epoch": 139.6449704142012, "grad_norm": 0.035043612122535706, "learning_rate": 3.351192575638761e-05, "loss": 0.0017, "step": 23600 }, { "epoch": 139.7041420118343, "grad_norm": 0.03595069795846939, "learning_rate": 3.341135280148192e-05, "loss": 0.0016, "step": 23610 }, { "epoch": 139.76331360946745, "grad_norm": 0.03871545195579529, "learning_rate": 3.331091207107689e-05, "loss": 0.0017, "step": 23620 }, { "epoch": 139.82248520710058, "grad_norm": 0.036257848143577576, "learning_rate": 3.321060367908344e-05, "loss": 0.0015, "step": 23630 }, { "epoch": 139.88165680473372, "grad_norm": 0.05807751044631004, "learning_rate": 3.311042773926246e-05, "loss": 0.0014, "step": 23640 }, { "epoch": 139.94082840236686, "grad_norm": 0.054509054869413376, "learning_rate": 3.301038436522461e-05, "loss": 0.0017, "step": 23650 }, { "epoch": 140.0, "grad_norm": 0.03677540272474289, "learning_rate": 3.291047367043023e-05, "loss": 0.0013, "step": 23660 }, { "epoch": 140.05917159763314, "grad_norm": 0.03812675550580025, "learning_rate": 3.281069576818916e-05, "loss": 0.0014, "step": 23670 }, { "epoch": 140.11834319526628, "grad_norm": 0.03363603726029396, "learning_rate": 3.271105077166066e-05, "loss": 0.0013, "step": 23680 }, { "epoch": 140.17751479289942, "grad_norm": 0.038866303861141205, "learning_rate": 3.261153879385321e-05, "loss": 0.0017, "step": 23690 }, { "epoch": 140.23668639053255, "grad_norm": 0.03082023188471794, "learning_rate": 3.251215994762449e-05, "loss": 0.0015, "step": 23700 }, { "epoch": 140.2958579881657, "grad_norm": 0.03384172543883324, "learning_rate": 3.241291434568116e-05, "loss": 0.0016, "step": 23710 }, { "epoch": 140.3550295857988, "grad_norm": 0.0785943791270256, "learning_rate": 3.231380210057883e-05, "loss": 0.0016, "step": 23720 }, { "epoch": 140.41420118343194, "grad_norm": 0.06820673495531082, "learning_rate": 3.221482332472172e-05, "loss": 0.0016, "step": 23730 }, { "epoch": 140.47337278106508, "grad_norm": 0.04017048329114914, "learning_rate": 3.2115978130362845e-05, "loss": 0.0015, "step": 23740 }, { "epoch": 140.53254437869822, "grad_norm": 0.04338286817073822, "learning_rate": 3.2017266629603625e-05, "loss": 0.0015, "step": 23750 }, { "epoch": 140.59171597633136, "grad_norm": 0.056586854159832, "learning_rate": 3.1918688934393896e-05, "loss": 0.0015, "step": 23760 }, { "epoch": 140.6508875739645, "grad_norm": 0.04240056127309799, "learning_rate": 3.182024515653175e-05, "loss": 0.0013, "step": 23770 }, { "epoch": 140.71005917159763, "grad_norm": 0.0700332522392273, "learning_rate": 3.172193540766342e-05, "loss": 0.0019, "step": 23780 }, { "epoch": 140.76923076923077, "grad_norm": 0.044883519411087036, "learning_rate": 3.1623759799283005e-05, "loss": 0.0014, "step": 23790 }, { "epoch": 140.8284023668639, "grad_norm": 0.04239135608077049, "learning_rate": 3.152571844273261e-05, "loss": 0.0014, "step": 23800 }, { "epoch": 140.88757396449705, "grad_norm": 0.03745837137103081, "learning_rate": 3.142781144920208e-05, "loss": 0.0017, "step": 23810 }, { "epoch": 140.9467455621302, "grad_norm": 0.028152555227279663, "learning_rate": 3.13300389297288e-05, "loss": 0.0016, "step": 23820 }, { "epoch": 141.0059171597633, "grad_norm": 0.0450936034321785, "learning_rate": 3.1232400995197715e-05, "loss": 0.0016, "step": 23830 }, { "epoch": 141.06508875739644, "grad_norm": 0.04892288148403168, "learning_rate": 3.113489775634113e-05, "loss": 0.0016, "step": 23840 }, { "epoch": 141.12426035502958, "grad_norm": 0.06677325069904327, "learning_rate": 3.1037529323738494e-05, "loss": 0.0016, "step": 23850 }, { "epoch": 141.18343195266272, "grad_norm": 0.07013927400112152, "learning_rate": 3.094029580781651e-05, "loss": 0.0016, "step": 23860 }, { "epoch": 141.24260355029585, "grad_norm": 0.05410480499267578, "learning_rate": 3.084319731884876e-05, "loss": 0.0017, "step": 23870 }, { "epoch": 141.301775147929, "grad_norm": 0.029494119808077812, "learning_rate": 3.0746233966955774e-05, "loss": 0.0014, "step": 23880 }, { "epoch": 141.36094674556213, "grad_norm": 0.030343269929289818, "learning_rate": 3.064940586210477e-05, "loss": 0.0016, "step": 23890 }, { "epoch": 141.42011834319527, "grad_norm": 0.05925062298774719, "learning_rate": 3.055271311410959e-05, "loss": 0.0016, "step": 23900 }, { "epoch": 141.4792899408284, "grad_norm": 0.043458547443151474, "learning_rate": 3.045615583263058e-05, "loss": 0.0016, "step": 23910 }, { "epoch": 141.53846153846155, "grad_norm": 0.028422722592949867, "learning_rate": 3.0359734127174428e-05, "loss": 0.0019, "step": 23920 }, { "epoch": 141.5976331360947, "grad_norm": 0.03674282878637314, "learning_rate": 3.0263448107094096e-05, "loss": 0.0016, "step": 23930 }, { "epoch": 141.65680473372782, "grad_norm": 0.035029709339141846, "learning_rate": 3.016729788158868e-05, "loss": 0.0013, "step": 23940 }, { "epoch": 141.71597633136093, "grad_norm": 0.034392330795526505, "learning_rate": 3.0071283559703157e-05, "loss": 0.0011, "step": 23950 }, { "epoch": 141.77514792899407, "grad_norm": 0.050392139703035355, "learning_rate": 2.9975405250328487e-05, "loss": 0.0014, "step": 23960 }, { "epoch": 141.8343195266272, "grad_norm": 0.04367827624082565, "learning_rate": 2.9879663062201343e-05, "loss": 0.0016, "step": 23970 }, { "epoch": 141.89349112426035, "grad_norm": 0.0427592433989048, "learning_rate": 2.978405710390403e-05, "loss": 0.0014, "step": 23980 }, { "epoch": 141.9526627218935, "grad_norm": 0.03900064155459404, "learning_rate": 2.968858748386437e-05, "loss": 0.0015, "step": 23990 }, { "epoch": 142.01183431952663, "grad_norm": 0.04696662724018097, "learning_rate": 2.9593254310355478e-05, "loss": 0.0018, "step": 24000 }, { "epoch": 142.07100591715977, "grad_norm": 0.048943884670734406, "learning_rate": 2.9498057691495798e-05, "loss": 0.0014, "step": 24010 }, { "epoch": 142.1301775147929, "grad_norm": 0.05523626133799553, "learning_rate": 2.9402997735248906e-05, "loss": 0.0013, "step": 24020 }, { "epoch": 142.18934911242604, "grad_norm": 0.038781315088272095, "learning_rate": 2.930807454942337e-05, "loss": 0.0015, "step": 24030 }, { "epoch": 142.24852071005918, "grad_norm": 0.049544304609298706, "learning_rate": 2.9213288241672643e-05, "loss": 0.0017, "step": 24040 }, { "epoch": 142.30769230769232, "grad_norm": 0.03796366602182388, "learning_rate": 2.9118638919494986e-05, "loss": 0.0019, "step": 24050 }, { "epoch": 142.36686390532543, "grad_norm": 0.058190733194351196, "learning_rate": 2.902412669023318e-05, "loss": 0.0017, "step": 24060 }, { "epoch": 142.42603550295857, "grad_norm": 0.043346863240003586, "learning_rate": 2.892975166107465e-05, "loss": 0.0016, "step": 24070 }, { "epoch": 142.4852071005917, "grad_norm": 0.038047656416893005, "learning_rate": 2.883551393905115e-05, "loss": 0.0015, "step": 24080 }, { "epoch": 142.54437869822485, "grad_norm": 0.036478232592344284, "learning_rate": 2.8741413631038873e-05, "loss": 0.0016, "step": 24090 }, { "epoch": 142.60355029585799, "grad_norm": 0.05325707048177719, "learning_rate": 2.8647450843757897e-05, "loss": 0.0015, "step": 24100 }, { "epoch": 142.66272189349112, "grad_norm": 0.03878138214349747, "learning_rate": 2.855362568377253e-05, "loss": 0.0014, "step": 24110 }, { "epoch": 142.72189349112426, "grad_norm": 0.077543705701828, "learning_rate": 2.8459938257490937e-05, "loss": 0.0017, "step": 24120 }, { "epoch": 142.7810650887574, "grad_norm": 0.035193122923374176, "learning_rate": 2.8366388671165113e-05, "loss": 0.0013, "step": 24130 }, { "epoch": 142.84023668639054, "grad_norm": 0.03956323489546776, "learning_rate": 2.827297703089067e-05, "loss": 0.0013, "step": 24140 }, { "epoch": 142.89940828402368, "grad_norm": 0.045537542551755905, "learning_rate": 2.817970344260685e-05, "loss": 0.0015, "step": 24150 }, { "epoch": 142.95857988165682, "grad_norm": 0.04570288211107254, "learning_rate": 2.8086568012096215e-05, "loss": 0.0016, "step": 24160 }, { "epoch": 143.01775147928993, "grad_norm": 0.04016353562474251, "learning_rate": 2.7993570844984747e-05, "loss": 0.0014, "step": 24170 }, { "epoch": 143.07692307692307, "grad_norm": 0.0486067533493042, "learning_rate": 2.7900712046741556e-05, "loss": 0.0013, "step": 24180 }, { "epoch": 143.1360946745562, "grad_norm": 0.04290614649653435, "learning_rate": 2.780799172267889e-05, "loss": 0.0015, "step": 24190 }, { "epoch": 143.19526627218934, "grad_norm": 0.054710183292627335, "learning_rate": 2.771540997795188e-05, "loss": 0.0017, "step": 24200 }, { "epoch": 143.25443786982248, "grad_norm": 0.06712804734706879, "learning_rate": 2.7622966917558588e-05, "loss": 0.0015, "step": 24210 }, { "epoch": 143.31360946745562, "grad_norm": 0.040056586265563965, "learning_rate": 2.7530662646339664e-05, "loss": 0.0014, "step": 24220 }, { "epoch": 143.37278106508876, "grad_norm": 0.05920960009098053, "learning_rate": 2.7438497268978436e-05, "loss": 0.0014, "step": 24230 }, { "epoch": 143.4319526627219, "grad_norm": 0.046027231961488724, "learning_rate": 2.7346470890000717e-05, "loss": 0.0014, "step": 24240 }, { "epoch": 143.49112426035504, "grad_norm": 0.03860506787896156, "learning_rate": 2.7254583613774644e-05, "loss": 0.0016, "step": 24250 }, { "epoch": 143.55029585798817, "grad_norm": 0.03582184016704559, "learning_rate": 2.716283554451068e-05, "loss": 0.0014, "step": 24260 }, { "epoch": 143.6094674556213, "grad_norm": 0.034403279423713684, "learning_rate": 2.707122678626124e-05, "loss": 0.0016, "step": 24270 }, { "epoch": 143.66863905325442, "grad_norm": 0.05069539695978165, "learning_rate": 2.6979757442920902e-05, "loss": 0.0014, "step": 24280 }, { "epoch": 143.72781065088756, "grad_norm": 0.037437256425619125, "learning_rate": 2.6888427618226067e-05, "loss": 0.0012, "step": 24290 }, { "epoch": 143.7869822485207, "grad_norm": 0.04621151089668274, "learning_rate": 2.6797237415754947e-05, "loss": 0.0014, "step": 24300 }, { "epoch": 143.84615384615384, "grad_norm": 0.06180931627750397, "learning_rate": 2.6706186938927416e-05, "loss": 0.0014, "step": 24310 }, { "epoch": 143.90532544378698, "grad_norm": 0.03829403594136238, "learning_rate": 2.6615276291004756e-05, "loss": 0.0015, "step": 24320 }, { "epoch": 143.96449704142012, "grad_norm": 0.05876125767827034, "learning_rate": 2.652450557508981e-05, "loss": 0.0016, "step": 24330 }, { "epoch": 144.02366863905326, "grad_norm": 0.03768197447061539, "learning_rate": 2.643387489412665e-05, "loss": 0.0015, "step": 24340 }, { "epoch": 144.0828402366864, "grad_norm": 0.06700043380260468, "learning_rate": 2.634338435090057e-05, "loss": 0.0015, "step": 24350 }, { "epoch": 144.14201183431953, "grad_norm": 0.04718206450343132, "learning_rate": 2.625303404803793e-05, "loss": 0.0015, "step": 24360 }, { "epoch": 144.20118343195267, "grad_norm": 0.06892269849777222, "learning_rate": 2.616282408800603e-05, "loss": 0.0013, "step": 24370 }, { "epoch": 144.2603550295858, "grad_norm": 0.04448983445763588, "learning_rate": 2.6072754573112948e-05, "loss": 0.0017, "step": 24380 }, { "epoch": 144.31952662721895, "grad_norm": 0.05776520073413849, "learning_rate": 2.598282560550759e-05, "loss": 0.0016, "step": 24390 }, { "epoch": 144.37869822485206, "grad_norm": 0.04107668995857239, "learning_rate": 2.589303728717939e-05, "loss": 0.0014, "step": 24400 }, { "epoch": 144.4378698224852, "grad_norm": 0.02667148783802986, "learning_rate": 2.5803389719958294e-05, "loss": 0.0014, "step": 24410 }, { "epoch": 144.49704142011834, "grad_norm": 0.028222938999533653, "learning_rate": 2.5713883005514674e-05, "loss": 0.0015, "step": 24420 }, { "epoch": 144.55621301775147, "grad_norm": 0.052589815109968185, "learning_rate": 2.5624517245359012e-05, "loss": 0.0014, "step": 24430 }, { "epoch": 144.6153846153846, "grad_norm": 0.042072851210832596, "learning_rate": 2.5535292540842067e-05, "loss": 0.0014, "step": 24440 }, { "epoch": 144.67455621301775, "grad_norm": 0.04237598925828934, "learning_rate": 2.54462089931546e-05, "loss": 0.0014, "step": 24450 }, { "epoch": 144.7337278106509, "grad_norm": 0.0364345982670784, "learning_rate": 2.5357266703327257e-05, "loss": 0.0015, "step": 24460 }, { "epoch": 144.79289940828403, "grad_norm": 0.027481412515044212, "learning_rate": 2.5268465772230483e-05, "loss": 0.0012, "step": 24470 }, { "epoch": 144.85207100591717, "grad_norm": 0.0861203745007515, "learning_rate": 2.5179806300574458e-05, "loss": 0.0016, "step": 24480 }, { "epoch": 144.9112426035503, "grad_norm": 0.025080731138586998, "learning_rate": 2.5091288388908827e-05, "loss": 0.0014, "step": 24490 }, { "epoch": 144.97041420118344, "grad_norm": 0.03325602412223816, "learning_rate": 2.500291213762274e-05, "loss": 0.0014, "step": 24500 }, { "epoch": 145.02958579881656, "grad_norm": 0.09459739178419113, "learning_rate": 2.4914677646944786e-05, "loss": 0.0014, "step": 24510 }, { "epoch": 145.0887573964497, "grad_norm": 0.04953049123287201, "learning_rate": 2.4826585016942674e-05, "loss": 0.0016, "step": 24520 }, { "epoch": 145.14792899408283, "grad_norm": 0.046108875423669815, "learning_rate": 2.4738634347523202e-05, "loss": 0.0014, "step": 24530 }, { "epoch": 145.20710059171597, "grad_norm": 0.0897352322936058, "learning_rate": 2.4650825738432244e-05, "loss": 0.0016, "step": 24540 }, { "epoch": 145.2662721893491, "grad_norm": 0.029082974418997765, "learning_rate": 2.456315928925453e-05, "loss": 0.0015, "step": 24550 }, { "epoch": 145.32544378698225, "grad_norm": 0.03660481050610542, "learning_rate": 2.4475635099413576e-05, "loss": 0.0015, "step": 24560 }, { "epoch": 145.3846153846154, "grad_norm": 0.030457468703389168, "learning_rate": 2.438825326817155e-05, "loss": 0.0013, "step": 24570 }, { "epoch": 145.44378698224853, "grad_norm": 0.047888342291116714, "learning_rate": 2.430101389462922e-05, "loss": 0.0013, "step": 24580 }, { "epoch": 145.50295857988166, "grad_norm": 0.03452422097325325, "learning_rate": 2.421391707772566e-05, "loss": 0.0014, "step": 24590 }, { "epoch": 145.5621301775148, "grad_norm": 0.02266758680343628, "learning_rate": 2.4126962916238372e-05, "loss": 0.0015, "step": 24600 }, { "epoch": 145.62130177514794, "grad_norm": 0.042134761810302734, "learning_rate": 2.40401515087831e-05, "loss": 0.0015, "step": 24610 }, { "epoch": 145.68047337278105, "grad_norm": 0.03807304799556732, "learning_rate": 2.3953482953813584e-05, "loss": 0.0012, "step": 24620 }, { "epoch": 145.7396449704142, "grad_norm": 0.02669641561806202, "learning_rate": 2.3866957349621668e-05, "loss": 0.0014, "step": 24630 }, { "epoch": 145.79881656804733, "grad_norm": 0.026557402685284615, "learning_rate": 2.3780574794337004e-05, "loss": 0.0014, "step": 24640 }, { "epoch": 145.85798816568047, "grad_norm": 0.0750805214047432, "learning_rate": 2.3694335385926978e-05, "loss": 0.0013, "step": 24650 }, { "epoch": 145.9171597633136, "grad_norm": 0.03822938725352287, "learning_rate": 2.3608239222196683e-05, "loss": 0.0014, "step": 24660 }, { "epoch": 145.97633136094674, "grad_norm": 0.05233919993042946, "learning_rate": 2.3522286400788803e-05, "loss": 0.0014, "step": 24670 }, { "epoch": 146.03550295857988, "grad_norm": 0.0376155711710453, "learning_rate": 2.3436477019183343e-05, "loss": 0.0012, "step": 24680 }, { "epoch": 146.09467455621302, "grad_norm": 0.03995096683502197, "learning_rate": 2.335081117469777e-05, "loss": 0.0013, "step": 24690 }, { "epoch": 146.15384615384616, "grad_norm": 0.04144365340471268, "learning_rate": 2.3265288964486604e-05, "loss": 0.0013, "step": 24700 }, { "epoch": 146.2130177514793, "grad_norm": 0.10305865854024887, "learning_rate": 2.317991048554153e-05, "loss": 0.0012, "step": 24710 }, { "epoch": 146.27218934911244, "grad_norm": 0.04172438755631447, "learning_rate": 2.309467583469134e-05, "loss": 0.0014, "step": 24720 }, { "epoch": 146.33136094674558, "grad_norm": 0.023234328255057335, "learning_rate": 2.3009585108601557e-05, "loss": 0.0015, "step": 24730 }, { "epoch": 146.3905325443787, "grad_norm": 0.02782975323498249, "learning_rate": 2.2924638403774593e-05, "loss": 0.0012, "step": 24740 }, { "epoch": 146.44970414201183, "grad_norm": 0.028037171810865402, "learning_rate": 2.2839835816549363e-05, "loss": 0.0013, "step": 24750 }, { "epoch": 146.50887573964496, "grad_norm": 0.15575388073921204, "learning_rate": 2.27551774431015e-05, "loss": 0.0017, "step": 24760 }, { "epoch": 146.5680473372781, "grad_norm": 0.03408673778176308, "learning_rate": 2.267066337944301e-05, "loss": 0.0014, "step": 24770 }, { "epoch": 146.62721893491124, "grad_norm": 0.04123811051249504, "learning_rate": 2.258629372142225e-05, "loss": 0.0013, "step": 24780 }, { "epoch": 146.68639053254438, "grad_norm": 0.05267418548464775, "learning_rate": 2.250206856472383e-05, "loss": 0.0014, "step": 24790 }, { "epoch": 146.74556213017752, "grad_norm": 0.05652269721031189, "learning_rate": 2.241798800486839e-05, "loss": 0.0013, "step": 24800 }, { "epoch": 146.80473372781066, "grad_norm": 0.06049343943595886, "learning_rate": 2.2334052137212675e-05, "loss": 0.0016, "step": 24810 }, { "epoch": 146.8639053254438, "grad_norm": 0.03743109479546547, "learning_rate": 2.2250261056949304e-05, "loss": 0.0013, "step": 24820 }, { "epoch": 146.92307692307693, "grad_norm": 0.03863425925374031, "learning_rate": 2.2166614859106657e-05, "loss": 0.0015, "step": 24830 }, { "epoch": 146.98224852071007, "grad_norm": 0.04131210967898369, "learning_rate": 2.2083113638548873e-05, "loss": 0.0015, "step": 24840 }, { "epoch": 147.04142011834318, "grad_norm": 0.04003197327256203, "learning_rate": 2.199975748997563e-05, "loss": 0.0013, "step": 24850 }, { "epoch": 147.10059171597632, "grad_norm": 0.04247964546084404, "learning_rate": 2.1916546507922038e-05, "loss": 0.0014, "step": 24860 }, { "epoch": 147.15976331360946, "grad_norm": 0.06679178774356842, "learning_rate": 2.1833480786758617e-05, "loss": 0.0014, "step": 24870 }, { "epoch": 147.2189349112426, "grad_norm": 0.03274863213300705, "learning_rate": 2.1750560420691153e-05, "loss": 0.0015, "step": 24880 }, { "epoch": 147.27810650887574, "grad_norm": 0.03979913890361786, "learning_rate": 2.1667785503760575e-05, "loss": 0.0014, "step": 24890 }, { "epoch": 147.33727810650888, "grad_norm": 0.031148483976721764, "learning_rate": 2.1585156129842823e-05, "loss": 0.0013, "step": 24900 }, { "epoch": 147.39644970414201, "grad_norm": 0.04563828185200691, "learning_rate": 2.1502672392648885e-05, "loss": 0.0014, "step": 24910 }, { "epoch": 147.45562130177515, "grad_norm": 0.04440358281135559, "learning_rate": 2.1420334385724346e-05, "loss": 0.0013, "step": 24920 }, { "epoch": 147.5147928994083, "grad_norm": 0.08844203501939774, "learning_rate": 2.13381422024498e-05, "loss": 0.0015, "step": 24930 }, { "epoch": 147.57396449704143, "grad_norm": 0.04759542644023895, "learning_rate": 2.1256095936040292e-05, "loss": 0.0011, "step": 24940 }, { "epoch": 147.63313609467457, "grad_norm": 0.052551303058862686, "learning_rate": 2.117419567954544e-05, "loss": 0.0016, "step": 24950 }, { "epoch": 147.69230769230768, "grad_norm": 0.025891216471791267, "learning_rate": 2.1092441525849184e-05, "loss": 0.0014, "step": 24960 }, { "epoch": 147.75147928994082, "grad_norm": 0.16796089708805084, "learning_rate": 2.101083356766989e-05, "loss": 0.0015, "step": 24970 }, { "epoch": 147.81065088757396, "grad_norm": 0.03896087780594826, "learning_rate": 2.0929371897560032e-05, "loss": 0.0013, "step": 24980 }, { "epoch": 147.8698224852071, "grad_norm": 0.04049122333526611, "learning_rate": 2.0848056607906233e-05, "loss": 0.0014, "step": 24990 }, { "epoch": 147.92899408284023, "grad_norm": 0.048446204513311386, "learning_rate": 2.076688779092907e-05, "loss": 0.0015, "step": 25000 }, { "epoch": 147.98816568047337, "grad_norm": 0.03650864586234093, "learning_rate": 2.068586553868306e-05, "loss": 0.0013, "step": 25010 }, { "epoch": 148.0473372781065, "grad_norm": 0.05722380056977272, "learning_rate": 2.0604989943056366e-05, "loss": 0.0014, "step": 25020 }, { "epoch": 148.10650887573965, "grad_norm": 0.029346689581871033, "learning_rate": 2.052426109577095e-05, "loss": 0.0013, "step": 25030 }, { "epoch": 148.1656804733728, "grad_norm": 0.04032112658023834, "learning_rate": 2.0443679088382326e-05, "loss": 0.0013, "step": 25040 }, { "epoch": 148.22485207100593, "grad_norm": 0.03339872881770134, "learning_rate": 2.0363244012279456e-05, "loss": 0.0013, "step": 25050 }, { "epoch": 148.28402366863907, "grad_norm": 0.031031938269734383, "learning_rate": 2.028295595868469e-05, "loss": 0.0012, "step": 25060 }, { "epoch": 148.34319526627218, "grad_norm": 0.044606778770685196, "learning_rate": 2.0202815018653536e-05, "loss": 0.0013, "step": 25070 }, { "epoch": 148.4023668639053, "grad_norm": 0.06529086828231812, "learning_rate": 2.0122821283074807e-05, "loss": 0.0017, "step": 25080 }, { "epoch": 148.46153846153845, "grad_norm": 0.03264711797237396, "learning_rate": 2.0042974842670273e-05, "loss": 0.0012, "step": 25090 }, { "epoch": 148.5207100591716, "grad_norm": 0.15558867156505585, "learning_rate": 1.9963275787994708e-05, "loss": 0.0017, "step": 25100 }, { "epoch": 148.57988165680473, "grad_norm": 0.04085687920451164, "learning_rate": 1.9883724209435682e-05, "loss": 0.0015, "step": 25110 }, { "epoch": 148.63905325443787, "grad_norm": 0.031350027769804, "learning_rate": 1.9804320197213554e-05, "loss": 0.0013, "step": 25120 }, { "epoch": 148.698224852071, "grad_norm": 0.04099244624376297, "learning_rate": 1.9725063841381307e-05, "loss": 0.0016, "step": 25130 }, { "epoch": 148.75739644970415, "grad_norm": 0.03688952699303627, "learning_rate": 1.9645955231824473e-05, "loss": 0.0014, "step": 25140 }, { "epoch": 148.81656804733728, "grad_norm": 0.03549177199602127, "learning_rate": 1.9566994458261004e-05, "loss": 0.0012, "step": 25150 }, { "epoch": 148.87573964497042, "grad_norm": 0.03775453194975853, "learning_rate": 1.9488181610241233e-05, "loss": 0.0012, "step": 25160 }, { "epoch": 148.93491124260356, "grad_norm": 0.03271486982703209, "learning_rate": 1.9409516777147693e-05, "loss": 0.0015, "step": 25170 }, { "epoch": 148.9940828402367, "grad_norm": 0.08235767483711243, "learning_rate": 1.9331000048195038e-05, "loss": 0.0015, "step": 25180 }, { "epoch": 149.0532544378698, "grad_norm": 0.031390756368637085, "learning_rate": 1.9252631512429973e-05, "loss": 0.0014, "step": 25190 }, { "epoch": 149.11242603550295, "grad_norm": 0.04107058793306351, "learning_rate": 1.9174411258731164e-05, "loss": 0.0013, "step": 25200 }, { "epoch": 149.1715976331361, "grad_norm": 0.09086292237043381, "learning_rate": 1.9096339375809073e-05, "loss": 0.0013, "step": 25210 }, { "epoch": 149.23076923076923, "grad_norm": 0.0598088800907135, "learning_rate": 1.9018415952205916e-05, "loss": 0.0014, "step": 25220 }, { "epoch": 149.28994082840237, "grad_norm": 0.02449220046401024, "learning_rate": 1.894064107629549e-05, "loss": 0.0014, "step": 25230 }, { "epoch": 149.3491124260355, "grad_norm": 0.04340168833732605, "learning_rate": 1.886301483628319e-05, "loss": 0.0015, "step": 25240 }, { "epoch": 149.40828402366864, "grad_norm": 0.04635768383741379, "learning_rate": 1.8785537320205806e-05, "loss": 0.0013, "step": 25250 }, { "epoch": 149.46745562130178, "grad_norm": 0.05125417560338974, "learning_rate": 1.870820861593147e-05, "loss": 0.0013, "step": 25260 }, { "epoch": 149.52662721893492, "grad_norm": 0.061036814004182816, "learning_rate": 1.8631028811159525e-05, "loss": 0.0013, "step": 25270 }, { "epoch": 149.58579881656806, "grad_norm": 0.028564726933836937, "learning_rate": 1.8553997993420495e-05, "loss": 0.0012, "step": 25280 }, { "epoch": 149.6449704142012, "grad_norm": 0.04170204699039459, "learning_rate": 1.8477116250075825e-05, "loss": 0.0014, "step": 25290 }, { "epoch": 149.7041420118343, "grad_norm": 0.05653214454650879, "learning_rate": 1.840038366831802e-05, "loss": 0.0013, "step": 25300 }, { "epoch": 149.76331360946745, "grad_norm": 0.05192108824849129, "learning_rate": 1.832380033517037e-05, "loss": 0.0013, "step": 25310 }, { "epoch": 149.82248520710058, "grad_norm": 0.04365210235118866, "learning_rate": 1.8247366337486862e-05, "loss": 0.0013, "step": 25320 }, { "epoch": 149.88165680473372, "grad_norm": 0.053283825516700745, "learning_rate": 1.8171081761952195e-05, "loss": 0.0012, "step": 25330 }, { "epoch": 149.94082840236686, "grad_norm": 0.03430003300309181, "learning_rate": 1.8094946695081556e-05, "loss": 0.0014, "step": 25340 }, { "epoch": 150.0, "grad_norm": 0.04553251713514328, "learning_rate": 1.801896122322057e-05, "loss": 0.0013, "step": 25350 }, { "epoch": 150.05917159763314, "grad_norm": 0.055055174976587296, "learning_rate": 1.794312543254524e-05, "loss": 0.0013, "step": 25360 }, { "epoch": 150.11834319526628, "grad_norm": 0.044157736003398895, "learning_rate": 1.7867439409061763e-05, "loss": 0.0014, "step": 25370 }, { "epoch": 150.17751479289942, "grad_norm": 0.02306864783167839, "learning_rate": 1.779190323860657e-05, "loss": 0.0012, "step": 25380 }, { "epoch": 150.23668639053255, "grad_norm": 0.06826592981815338, "learning_rate": 1.771651700684602e-05, "loss": 0.0017, "step": 25390 }, { "epoch": 150.2958579881657, "grad_norm": 0.04648706689476967, "learning_rate": 1.7641280799276507e-05, "loss": 0.0015, "step": 25400 }, { "epoch": 150.3550295857988, "grad_norm": 0.043412238359451294, "learning_rate": 1.7566194701224264e-05, "loss": 0.0014, "step": 25410 }, { "epoch": 150.41420118343194, "grad_norm": 0.051488909870386124, "learning_rate": 1.749125879784528e-05, "loss": 0.0014, "step": 25420 }, { "epoch": 150.47337278106508, "grad_norm": 0.03539945185184479, "learning_rate": 1.7416473174125235e-05, "loss": 0.0014, "step": 25430 }, { "epoch": 150.53254437869822, "grad_norm": 0.029886607080698013, "learning_rate": 1.734183791487936e-05, "loss": 0.0014, "step": 25440 }, { "epoch": 150.59171597633136, "grad_norm": 0.04329317435622215, "learning_rate": 1.7267353104752265e-05, "loss": 0.0014, "step": 25450 }, { "epoch": 150.6508875739645, "grad_norm": 0.043170519173145294, "learning_rate": 1.7193018828218074e-05, "loss": 0.0013, "step": 25460 }, { "epoch": 150.71005917159763, "grad_norm": 0.02162070944905281, "learning_rate": 1.7118835169580125e-05, "loss": 0.0012, "step": 25470 }, { "epoch": 150.76923076923077, "grad_norm": 0.03195548057556152, "learning_rate": 1.7044802212970933e-05, "loss": 0.0014, "step": 25480 }, { "epoch": 150.8284023668639, "grad_norm": 0.03165261074900627, "learning_rate": 1.697092004235213e-05, "loss": 0.0012, "step": 25490 }, { "epoch": 150.88757396449705, "grad_norm": 0.03636062517762184, "learning_rate": 1.6897188741514284e-05, "loss": 0.0011, "step": 25500 }, { "epoch": 150.9467455621302, "grad_norm": 0.034507445991039276, "learning_rate": 1.6823608394076916e-05, "loss": 0.0013, "step": 25510 }, { "epoch": 151.0059171597633, "grad_norm": 0.0717768520116806, "learning_rate": 1.6750179083488324e-05, "loss": 0.0013, "step": 25520 }, { "epoch": 151.06508875739644, "grad_norm": 0.04117452725768089, "learning_rate": 1.6676900893025514e-05, "loss": 0.0013, "step": 25530 }, { "epoch": 151.12426035502958, "grad_norm": 0.05038346350193024, "learning_rate": 1.660377390579412e-05, "loss": 0.0014, "step": 25540 }, { "epoch": 151.18343195266272, "grad_norm": 0.029629334807395935, "learning_rate": 1.6530798204728276e-05, "loss": 0.0012, "step": 25550 }, { "epoch": 151.24260355029585, "grad_norm": 0.03666435554623604, "learning_rate": 1.645797387259055e-05, "loss": 0.0012, "step": 25560 }, { "epoch": 151.301775147929, "grad_norm": 0.06350334733724594, "learning_rate": 1.638530099197184e-05, "loss": 0.0014, "step": 25570 }, { "epoch": 151.36094674556213, "grad_norm": 0.02825303003191948, "learning_rate": 1.6312779645291292e-05, "loss": 0.0012, "step": 25580 }, { "epoch": 151.42011834319527, "grad_norm": 0.04274047911167145, "learning_rate": 1.6240409914796182e-05, "loss": 0.0014, "step": 25590 }, { "epoch": 151.4792899408284, "grad_norm": 0.05604611337184906, "learning_rate": 1.616819188256181e-05, "loss": 0.0014, "step": 25600 }, { "epoch": 151.53846153846155, "grad_norm": 0.05206841602921486, "learning_rate": 1.6096125630491473e-05, "loss": 0.0014, "step": 25610 }, { "epoch": 151.5976331360947, "grad_norm": 0.028722750023007393, "learning_rate": 1.602421124031633e-05, "loss": 0.0015, "step": 25620 }, { "epoch": 151.65680473372782, "grad_norm": 0.026704641059041023, "learning_rate": 1.5952448793595307e-05, "loss": 0.0013, "step": 25630 }, { "epoch": 151.71597633136093, "grad_norm": 0.024947648867964745, "learning_rate": 1.5880838371714984e-05, "loss": 0.0013, "step": 25640 }, { "epoch": 151.77514792899407, "grad_norm": 0.04667273908853531, "learning_rate": 1.5809380055889592e-05, "loss": 0.0012, "step": 25650 }, { "epoch": 151.8343195266272, "grad_norm": 0.03861982375383377, "learning_rate": 1.573807392716076e-05, "loss": 0.0013, "step": 25660 }, { "epoch": 151.89349112426035, "grad_norm": 0.043233972042798996, "learning_rate": 1.5666920066397597e-05, "loss": 0.0013, "step": 25670 }, { "epoch": 151.9526627218935, "grad_norm": 0.11952227354049683, "learning_rate": 1.5595918554296504e-05, "loss": 0.0013, "step": 25680 }, { "epoch": 152.01183431952663, "grad_norm": 0.03493724390864372, "learning_rate": 1.5525069471381086e-05, "loss": 0.0012, "step": 25690 }, { "epoch": 152.07100591715977, "grad_norm": 0.03335055336356163, "learning_rate": 1.5454372898002083e-05, "loss": 0.0013, "step": 25700 }, { "epoch": 152.1301775147929, "grad_norm": 0.046823352575302124, "learning_rate": 1.5383828914337325e-05, "loss": 0.0013, "step": 25710 }, { "epoch": 152.18934911242604, "grad_norm": 0.0341467522084713, "learning_rate": 1.5313437600391487e-05, "loss": 0.0012, "step": 25720 }, { "epoch": 152.24852071005918, "grad_norm": 0.04489339515566826, "learning_rate": 1.5243199035996156e-05, "loss": 0.0015, "step": 25730 }, { "epoch": 152.30769230769232, "grad_norm": 0.03856031596660614, "learning_rate": 1.5173113300809697e-05, "loss": 0.0012, "step": 25740 }, { "epoch": 152.36686390532543, "grad_norm": 0.04078624024987221, "learning_rate": 1.5103180474317129e-05, "loss": 0.0012, "step": 25750 }, { "epoch": 152.42603550295857, "grad_norm": 0.047217484563589096, "learning_rate": 1.5033400635830057e-05, "loss": 0.0012, "step": 25760 }, { "epoch": 152.4852071005917, "grad_norm": 0.03245517984032631, "learning_rate": 1.4963773864486578e-05, "loss": 0.0014, "step": 25770 }, { "epoch": 152.54437869822485, "grad_norm": 0.056216828525066376, "learning_rate": 1.4894300239251228e-05, "loss": 0.0013, "step": 25780 }, { "epoch": 152.60355029585799, "grad_norm": 0.05322558432817459, "learning_rate": 1.4824979838914796e-05, "loss": 0.0013, "step": 25790 }, { "epoch": 152.66272189349112, "grad_norm": 0.0310557559132576, "learning_rate": 1.4755812742094347e-05, "loss": 0.0013, "step": 25800 }, { "epoch": 152.72189349112426, "grad_norm": 0.05720161274075508, "learning_rate": 1.4686799027233082e-05, "loss": 0.0014, "step": 25810 }, { "epoch": 152.7810650887574, "grad_norm": 0.15224012732505798, "learning_rate": 1.4617938772600207e-05, "loss": 0.0013, "step": 25820 }, { "epoch": 152.84023668639054, "grad_norm": 0.1701873540878296, "learning_rate": 1.4549232056290905e-05, "loss": 0.0012, "step": 25830 }, { "epoch": 152.89940828402368, "grad_norm": 0.03744759410619736, "learning_rate": 1.448067895622626e-05, "loss": 0.0015, "step": 25840 }, { "epoch": 152.95857988165682, "grad_norm": 0.03024367429316044, "learning_rate": 1.4412279550153117e-05, "loss": 0.0014, "step": 25850 }, { "epoch": 153.01775147928993, "grad_norm": 0.045115597546100616, "learning_rate": 1.4344033915644021e-05, "loss": 0.0012, "step": 25860 }, { "epoch": 153.07692307692307, "grad_norm": 0.02580956555902958, "learning_rate": 1.4275942130097096e-05, "loss": 0.0012, "step": 25870 }, { "epoch": 153.1360946745562, "grad_norm": 0.03325715288519859, "learning_rate": 1.4208004270736e-05, "loss": 0.0014, "step": 25880 }, { "epoch": 153.19526627218934, "grad_norm": 0.06842022389173508, "learning_rate": 1.4140220414609838e-05, "loss": 0.0013, "step": 25890 }, { "epoch": 153.25443786982248, "grad_norm": 0.03664945065975189, "learning_rate": 1.4072590638593046e-05, "loss": 0.0012, "step": 25900 }, { "epoch": 153.31360946745562, "grad_norm": 0.02509402669966221, "learning_rate": 1.4005115019385322e-05, "loss": 0.0012, "step": 25910 }, { "epoch": 153.37278106508876, "grad_norm": 0.03667319193482399, "learning_rate": 1.3937793633511552e-05, "loss": 0.0013, "step": 25920 }, { "epoch": 153.4319526627219, "grad_norm": 0.04354497045278549, "learning_rate": 1.3870626557321613e-05, "loss": 0.0012, "step": 25930 }, { "epoch": 153.49112426035504, "grad_norm": 0.022944655269384384, "learning_rate": 1.3803613866990487e-05, "loss": 0.0014, "step": 25940 }, { "epoch": 153.55029585798817, "grad_norm": 0.04404069483280182, "learning_rate": 1.3736755638517998e-05, "loss": 0.0013, "step": 25950 }, { "epoch": 153.6094674556213, "grad_norm": 0.04967816174030304, "learning_rate": 1.367005194772886e-05, "loss": 0.0011, "step": 25960 }, { "epoch": 153.66863905325442, "grad_norm": 0.038186948746442795, "learning_rate": 1.3603502870272475e-05, "loss": 0.0013, "step": 25970 }, { "epoch": 153.72781065088756, "grad_norm": 0.037097230553627014, "learning_rate": 1.3537108481622871e-05, "loss": 0.0013, "step": 25980 }, { "epoch": 153.7869822485207, "grad_norm": 0.046362221240997314, "learning_rate": 1.3470868857078681e-05, "loss": 0.0013, "step": 25990 }, { "epoch": 153.84615384615384, "grad_norm": 0.026690667495131493, "learning_rate": 1.3404784071763014e-05, "loss": 0.0014, "step": 26000 }, { "epoch": 153.90532544378698, "grad_norm": 0.03415238857269287, "learning_rate": 1.3338854200623383e-05, "loss": 0.0011, "step": 26010 }, { "epoch": 153.96449704142012, "grad_norm": 0.028005467727780342, "learning_rate": 1.3273079318431612e-05, "loss": 0.0013, "step": 26020 }, { "epoch": 154.02366863905326, "grad_norm": 0.0207949411123991, "learning_rate": 1.3207459499783696e-05, "loss": 0.0012, "step": 26030 }, { "epoch": 154.0828402366864, "grad_norm": 0.058484997600317, "learning_rate": 1.314199481909982e-05, "loss": 0.0013, "step": 26040 }, { "epoch": 154.14201183431953, "grad_norm": 0.04155866429209709, "learning_rate": 1.3076685350624234e-05, "loss": 0.0013, "step": 26050 }, { "epoch": 154.20118343195267, "grad_norm": 0.030199173837900162, "learning_rate": 1.3011531168425148e-05, "loss": 0.0011, "step": 26060 }, { "epoch": 154.2603550295858, "grad_norm": 0.02280605584383011, "learning_rate": 1.2946532346394628e-05, "loss": 0.0012, "step": 26070 }, { "epoch": 154.31952662721895, "grad_norm": 0.03838810697197914, "learning_rate": 1.2881688958248637e-05, "loss": 0.0012, "step": 26080 }, { "epoch": 154.37869822485206, "grad_norm": 0.024796191602945328, "learning_rate": 1.28170010775267e-05, "loss": 0.0013, "step": 26090 }, { "epoch": 154.4378698224852, "grad_norm": 0.06685041636228561, "learning_rate": 1.2752468777592134e-05, "loss": 0.0013, "step": 26100 }, { "epoch": 154.49704142011834, "grad_norm": 0.02946637198328972, "learning_rate": 1.2688092131631728e-05, "loss": 0.0012, "step": 26110 }, { "epoch": 154.55621301775147, "grad_norm": 0.05207180976867676, "learning_rate": 1.2623871212655762e-05, "loss": 0.0012, "step": 26120 }, { "epoch": 154.6153846153846, "grad_norm": 0.03242901340126991, "learning_rate": 1.2559806093497942e-05, "loss": 0.0013, "step": 26130 }, { "epoch": 154.67455621301775, "grad_norm": 0.06875277310609818, "learning_rate": 1.2495896846815184e-05, "loss": 0.0013, "step": 26140 }, { "epoch": 154.7337278106509, "grad_norm": 0.022622661665081978, "learning_rate": 1.2432143545087708e-05, "loss": 0.0013, "step": 26150 }, { "epoch": 154.79289940828403, "grad_norm": 0.0310062188655138, "learning_rate": 1.2368546260618844e-05, "loss": 0.0014, "step": 26160 }, { "epoch": 154.85207100591717, "grad_norm": 0.041263166815042496, "learning_rate": 1.2305105065535014e-05, "loss": 0.0013, "step": 26170 }, { "epoch": 154.9112426035503, "grad_norm": 0.05648425966501236, "learning_rate": 1.2241820031785631e-05, "loss": 0.0012, "step": 26180 }, { "epoch": 154.97041420118344, "grad_norm": 0.02647496946156025, "learning_rate": 1.2178691231142883e-05, "loss": 0.0012, "step": 26190 }, { "epoch": 155.02958579881656, "grad_norm": 0.03605038672685623, "learning_rate": 1.2115718735201885e-05, "loss": 0.0013, "step": 26200 }, { "epoch": 155.0887573964497, "grad_norm": 0.043883442878723145, "learning_rate": 1.2052902615380471e-05, "loss": 0.0012, "step": 26210 }, { "epoch": 155.14792899408283, "grad_norm": 0.030350634828209877, "learning_rate": 1.1990242942919094e-05, "loss": 0.0011, "step": 26220 }, { "epoch": 155.20710059171597, "grad_norm": 0.032107770442962646, "learning_rate": 1.1927739788880803e-05, "loss": 0.0013, "step": 26230 }, { "epoch": 155.2662721893491, "grad_norm": 0.03682319074869156, "learning_rate": 1.1865393224151165e-05, "loss": 0.0011, "step": 26240 }, { "epoch": 155.32544378698225, "grad_norm": 0.04213235154747963, "learning_rate": 1.1803203319438053e-05, "loss": 0.0011, "step": 26250 }, { "epoch": 155.3846153846154, "grad_norm": 0.04526222124695778, "learning_rate": 1.1741170145271783e-05, "loss": 0.0016, "step": 26260 }, { "epoch": 155.44378698224853, "grad_norm": 0.03374585136771202, "learning_rate": 1.1679293772004877e-05, "loss": 0.0012, "step": 26270 }, { "epoch": 155.50295857988166, "grad_norm": 0.047580886632204056, "learning_rate": 1.161757426981202e-05, "loss": 0.0011, "step": 26280 }, { "epoch": 155.5621301775148, "grad_norm": 0.042107582092285156, "learning_rate": 1.1556011708690016e-05, "loss": 0.0014, "step": 26290 }, { "epoch": 155.62130177514794, "grad_norm": 0.03279139846563339, "learning_rate": 1.149460615845762e-05, "loss": 0.0013, "step": 26300 }, { "epoch": 155.68047337278105, "grad_norm": 0.03339548036456108, "learning_rate": 1.1433357688755573e-05, "loss": 0.0011, "step": 26310 }, { "epoch": 155.7396449704142, "grad_norm": 0.03925257548689842, "learning_rate": 1.1372266369046478e-05, "loss": 0.0013, "step": 26320 }, { "epoch": 155.79881656804733, "grad_norm": 0.055180974304676056, "learning_rate": 1.1311332268614665e-05, "loss": 0.0011, "step": 26330 }, { "epoch": 155.85798816568047, "grad_norm": 0.025932343676686287, "learning_rate": 1.1250555456566217e-05, "loss": 0.0012, "step": 26340 }, { "epoch": 155.9171597633136, "grad_norm": 0.03850554674863815, "learning_rate": 1.1189936001828786e-05, "loss": 0.0014, "step": 26350 }, { "epoch": 155.97633136094674, "grad_norm": 0.030772235244512558, "learning_rate": 1.1129473973151559e-05, "loss": 0.0014, "step": 26360 }, { "epoch": 156.03550295857988, "grad_norm": 0.02884221263229847, "learning_rate": 1.1069169439105203e-05, "loss": 0.0013, "step": 26370 }, { "epoch": 156.09467455621302, "grad_norm": 0.1432925909757614, "learning_rate": 1.1009022468081807e-05, "loss": 0.0013, "step": 26380 }, { "epoch": 156.15384615384616, "grad_norm": 0.03569291904568672, "learning_rate": 1.0949033128294742e-05, "loss": 0.0012, "step": 26390 }, { "epoch": 156.2130177514793, "grad_norm": 0.03674834966659546, "learning_rate": 1.0889201487778537e-05, "loss": 0.0011, "step": 26400 }, { "epoch": 156.27218934911244, "grad_norm": 0.04746526479721069, "learning_rate": 1.082952761438896e-05, "loss": 0.0013, "step": 26410 }, { "epoch": 156.33136094674558, "grad_norm": 0.037254754453897476, "learning_rate": 1.0770011575802828e-05, "loss": 0.001, "step": 26420 }, { "epoch": 156.3905325443787, "grad_norm": 0.03585275635123253, "learning_rate": 1.0710653439517946e-05, "loss": 0.0012, "step": 26430 }, { "epoch": 156.44970414201183, "grad_norm": 0.024626847356557846, "learning_rate": 1.0651453272853034e-05, "loss": 0.0012, "step": 26440 }, { "epoch": 156.50887573964496, "grad_norm": 0.07576484978199005, "learning_rate": 1.0592411142947732e-05, "loss": 0.0013, "step": 26450 }, { "epoch": 156.5680473372781, "grad_norm": 0.07810043543577194, "learning_rate": 1.0533527116762296e-05, "loss": 0.0012, "step": 26460 }, { "epoch": 156.62721893491124, "grad_norm": 0.05251089110970497, "learning_rate": 1.0474801261077809e-05, "loss": 0.0011, "step": 26470 }, { "epoch": 156.68639053254438, "grad_norm": 0.04540502279996872, "learning_rate": 1.0416233642495941e-05, "loss": 0.0012, "step": 26480 }, { "epoch": 156.74556213017752, "grad_norm": 0.04170619323849678, "learning_rate": 1.0357824327438846e-05, "loss": 0.0014, "step": 26490 }, { "epoch": 156.80473372781066, "grad_norm": 0.04136528819799423, "learning_rate": 1.0299573382149234e-05, "loss": 0.0012, "step": 26500 }, { "epoch": 156.8639053254438, "grad_norm": 0.051558613777160645, "learning_rate": 1.0241480872690145e-05, "loss": 0.0015, "step": 26510 }, { "epoch": 156.92307692307693, "grad_norm": 0.02400320954620838, "learning_rate": 1.0183546864944942e-05, "loss": 0.0011, "step": 26520 }, { "epoch": 156.98224852071007, "grad_norm": 0.03051701933145523, "learning_rate": 1.0125771424617218e-05, "loss": 0.0012, "step": 26530 }, { "epoch": 157.04142011834318, "grad_norm": 0.024886932224035263, "learning_rate": 1.0068154617230778e-05, "loss": 0.0012, "step": 26540 }, { "epoch": 157.10059171597632, "grad_norm": 0.043721653521060944, "learning_rate": 1.0010696508129501e-05, "loss": 0.0011, "step": 26550 }, { "epoch": 157.15976331360946, "grad_norm": 0.028568129986524582, "learning_rate": 9.95339716247726e-06, "loss": 0.0013, "step": 26560 }, { "epoch": 157.2189349112426, "grad_norm": 0.04601225629448891, "learning_rate": 9.896256645257893e-06, "loss": 0.0012, "step": 26570 }, { "epoch": 157.27810650887574, "grad_norm": 0.027265029028058052, "learning_rate": 9.839275021275078e-06, "loss": 0.0012, "step": 26580 }, { "epoch": 157.33727810650888, "grad_norm": 0.03827735409140587, "learning_rate": 9.78245235515237e-06, "loss": 0.0013, "step": 26590 }, { "epoch": 157.39644970414201, "grad_norm": 0.03305191174149513, "learning_rate": 9.725788711332972e-06, "loss": 0.0011, "step": 26600 }, { "epoch": 157.45562130177515, "grad_norm": 0.03454270213842392, "learning_rate": 9.669284154079776e-06, "loss": 0.0013, "step": 26610 }, { "epoch": 157.5147928994083, "grad_norm": 0.04892861843109131, "learning_rate": 9.612938747475207e-06, "loss": 0.0012, "step": 26620 }, { "epoch": 157.57396449704143, "grad_norm": 0.033085353672504425, "learning_rate": 9.556752555421249e-06, "loss": 0.0013, "step": 26630 }, { "epoch": 157.63313609467457, "grad_norm": 0.03161030635237694, "learning_rate": 9.500725641639294e-06, "loss": 0.0013, "step": 26640 }, { "epoch": 157.69230769230768, "grad_norm": 0.036526285111904144, "learning_rate": 9.444858069670086e-06, "loss": 0.0013, "step": 26650 }, { "epoch": 157.75147928994082, "grad_norm": 0.036702241748571396, "learning_rate": 9.389149902873688e-06, "loss": 0.0011, "step": 26660 }, { "epoch": 157.81065088757396, "grad_norm": 0.057086046785116196, "learning_rate": 9.333601204429348e-06, "loss": 0.0014, "step": 26670 }, { "epoch": 157.8698224852071, "grad_norm": 0.038744356483221054, "learning_rate": 9.278212037335469e-06, "loss": 0.0012, "step": 26680 }, { "epoch": 157.92899408284023, "grad_norm": 0.03798561915755272, "learning_rate": 9.222982464409522e-06, "loss": 0.0013, "step": 26690 }, { "epoch": 157.98816568047337, "grad_norm": 0.052692972123622894, "learning_rate": 9.16791254828802e-06, "loss": 0.0012, "step": 26700 }, { "epoch": 158.0473372781065, "grad_norm": 0.03424789756536484, "learning_rate": 9.113002351426351e-06, "loss": 0.0011, "step": 26710 }, { "epoch": 158.10650887573965, "grad_norm": 0.04021783173084259, "learning_rate": 9.058251936098821e-06, "loss": 0.0012, "step": 26720 }, { "epoch": 158.1656804733728, "grad_norm": 0.06665361672639847, "learning_rate": 9.00366136439844e-06, "loss": 0.0013, "step": 26730 }, { "epoch": 158.22485207100593, "grad_norm": 0.029061278328299522, "learning_rate": 8.949230698237037e-06, "loss": 0.001, "step": 26740 }, { "epoch": 158.28402366863907, "grad_norm": 0.03984074294567108, "learning_rate": 8.894959999345014e-06, "loss": 0.0013, "step": 26750 }, { "epoch": 158.34319526627218, "grad_norm": 0.12394066154956818, "learning_rate": 8.840849329271404e-06, "loss": 0.0013, "step": 26760 }, { "epoch": 158.4023668639053, "grad_norm": 0.035199932754039764, "learning_rate": 8.78689874938372e-06, "loss": 0.0012, "step": 26770 }, { "epoch": 158.46153846153845, "grad_norm": 0.019400447607040405, "learning_rate": 8.733108320867932e-06, "loss": 0.0011, "step": 26780 }, { "epoch": 158.5207100591716, "grad_norm": 0.03279200568795204, "learning_rate": 8.67947810472836e-06, "loss": 0.0011, "step": 26790 }, { "epoch": 158.57988165680473, "grad_norm": 0.03154528886079788, "learning_rate": 8.62600816178765e-06, "loss": 0.0014, "step": 26800 }, { "epoch": 158.63905325443787, "grad_norm": 0.026305489242076874, "learning_rate": 8.572698552686663e-06, "loss": 0.0011, "step": 26810 }, { "epoch": 158.698224852071, "grad_norm": 0.03105168603360653, "learning_rate": 8.519549337884435e-06, "loss": 0.0012, "step": 26820 }, { "epoch": 158.75739644970415, "grad_norm": 0.03772702068090439, "learning_rate": 8.466560577658083e-06, "loss": 0.0011, "step": 26830 }, { "epoch": 158.81656804733728, "grad_norm": 0.048646971583366394, "learning_rate": 8.413732332102734e-06, "loss": 0.0012, "step": 26840 }, { "epoch": 158.87573964497042, "grad_norm": 0.05092639476060867, "learning_rate": 8.361064661131533e-06, "loss": 0.0012, "step": 26850 }, { "epoch": 158.93491124260356, "grad_norm": 0.03235522285103798, "learning_rate": 8.308557624475465e-06, "loss": 0.0012, "step": 26860 }, { "epoch": 158.9940828402367, "grad_norm": 0.034863464534282684, "learning_rate": 8.256211281683362e-06, "loss": 0.0013, "step": 26870 }, { "epoch": 159.0532544378698, "grad_norm": 0.042543087154626846, "learning_rate": 8.204025692121802e-06, "loss": 0.0013, "step": 26880 }, { "epoch": 159.11242603550295, "grad_norm": 0.03880879655480385, "learning_rate": 8.152000914975043e-06, "loss": 0.0013, "step": 26890 }, { "epoch": 159.1715976331361, "grad_norm": 0.02811448834836483, "learning_rate": 8.100137009244956e-06, "loss": 0.0011, "step": 26900 }, { "epoch": 159.23076923076923, "grad_norm": 0.043374091386795044, "learning_rate": 8.048434033751006e-06, "loss": 0.0012, "step": 26910 }, { "epoch": 159.28994082840237, "grad_norm": 0.030857743695378304, "learning_rate": 7.99689204713012e-06, "loss": 0.0012, "step": 26920 }, { "epoch": 159.3491124260355, "grad_norm": 0.03842831403017044, "learning_rate": 7.945511107836661e-06, "loss": 0.0013, "step": 26930 }, { "epoch": 159.40828402366864, "grad_norm": 0.0298906397074461, "learning_rate": 7.894291274142295e-06, "loss": 0.0012, "step": 26940 }, { "epoch": 159.46745562130178, "grad_norm": 0.052575938403606415, "learning_rate": 7.843232604136023e-06, "loss": 0.001, "step": 26950 }, { "epoch": 159.52662721893492, "grad_norm": 0.029613886028528214, "learning_rate": 7.79233515572406e-06, "loss": 0.0012, "step": 26960 }, { "epoch": 159.58579881656806, "grad_norm": 0.05499646067619324, "learning_rate": 7.7415989866298e-06, "loss": 0.0014, "step": 26970 }, { "epoch": 159.6449704142012, "grad_norm": 0.029826173558831215, "learning_rate": 7.691024154393676e-06, "loss": 0.0011, "step": 26980 }, { "epoch": 159.7041420118343, "grad_norm": 0.07376501709222794, "learning_rate": 7.640610716373197e-06, "loss": 0.0013, "step": 26990 }, { "epoch": 159.76331360946745, "grad_norm": 0.0928957462310791, "learning_rate": 7.590358729742807e-06, "loss": 0.0014, "step": 27000 }, { "epoch": 159.82248520710058, "grad_norm": 0.03913281485438347, "learning_rate": 7.540268251493836e-06, "loss": 0.0012, "step": 27010 }, { "epoch": 159.88165680473372, "grad_norm": 0.024088582023978233, "learning_rate": 7.490339338434476e-06, "loss": 0.001, "step": 27020 }, { "epoch": 159.94082840236686, "grad_norm": 0.04149623215198517, "learning_rate": 7.44057204718968e-06, "loss": 0.0011, "step": 27030 }, { "epoch": 160.0, "grad_norm": 0.02368379943072796, "learning_rate": 7.390966434201084e-06, "loss": 0.0012, "step": 27040 }, { "epoch": 160.05917159763314, "grad_norm": 0.02380284108221531, "learning_rate": 7.34152255572697e-06, "loss": 0.0012, "step": 27050 }, { "epoch": 160.11834319526628, "grad_norm": 0.023096323013305664, "learning_rate": 7.292240467842198e-06, "loss": 0.0013, "step": 27060 }, { "epoch": 160.17751479289942, "grad_norm": 0.031155068427324295, "learning_rate": 7.243120226438126e-06, "loss": 0.0011, "step": 27070 }, { "epoch": 160.23668639053255, "grad_norm": 0.18808527290821075, "learning_rate": 7.1941618872226104e-06, "loss": 0.0013, "step": 27080 }, { "epoch": 160.2958579881657, "grad_norm": 0.03246503323316574, "learning_rate": 7.145365505719836e-06, "loss": 0.0011, "step": 27090 }, { "epoch": 160.3550295857988, "grad_norm": 0.034578174352645874, "learning_rate": 7.096731137270317e-06, "loss": 0.0011, "step": 27100 }, { "epoch": 160.41420118343194, "grad_norm": 0.04010450839996338, "learning_rate": 7.048258837030868e-06, "loss": 0.0011, "step": 27110 }, { "epoch": 160.47337278106508, "grad_norm": 0.03561859950423241, "learning_rate": 6.999948659974447e-06, "loss": 0.0011, "step": 27120 }, { "epoch": 160.53254437869822, "grad_norm": 0.05371801555156708, "learning_rate": 6.951800660890178e-06, "loss": 0.0015, "step": 27130 }, { "epoch": 160.59171597633136, "grad_norm": 0.0330117866396904, "learning_rate": 6.903814894383248e-06, "loss": 0.0011, "step": 27140 }, { "epoch": 160.6508875739645, "grad_norm": 0.04348457604646683, "learning_rate": 6.85599141487489e-06, "loss": 0.0014, "step": 27150 }, { "epoch": 160.71005917159763, "grad_norm": 0.04798068478703499, "learning_rate": 6.808330276602203e-06, "loss": 0.0012, "step": 27160 }, { "epoch": 160.76923076923077, "grad_norm": 0.02860642410814762, "learning_rate": 6.760831533618233e-06, "loss": 0.0011, "step": 27170 }, { "epoch": 160.8284023668639, "grad_norm": 0.03213057667016983, "learning_rate": 6.713495239791855e-06, "loss": 0.001, "step": 27180 }, { "epoch": 160.88757396449705, "grad_norm": 0.025131290778517723, "learning_rate": 6.666321448807693e-06, "loss": 0.0012, "step": 27190 }, { "epoch": 160.9467455621302, "grad_norm": 0.026930002495646477, "learning_rate": 6.619310214166052e-06, "loss": 0.0011, "step": 27200 }, { "epoch": 161.0059171597633, "grad_norm": 0.02498222514986992, "learning_rate": 6.572461589182948e-06, "loss": 0.0012, "step": 27210 }, { "epoch": 161.06508875739644, "grad_norm": 0.024940498173236847, "learning_rate": 6.525775626989898e-06, "loss": 0.0012, "step": 27220 }, { "epoch": 161.12426035502958, "grad_norm": 0.031163157895207405, "learning_rate": 6.4792523805339836e-06, "loss": 0.0012, "step": 27230 }, { "epoch": 161.18343195266272, "grad_norm": 0.03419974446296692, "learning_rate": 6.432891902577764e-06, "loss": 0.0011, "step": 27240 }, { "epoch": 161.24260355029585, "grad_norm": 0.030373118817806244, "learning_rate": 6.386694245699181e-06, "loss": 0.0012, "step": 27250 }, { "epoch": 161.301775147929, "grad_norm": 0.020116034895181656, "learning_rate": 6.340659462291492e-06, "loss": 0.001, "step": 27260 }, { "epoch": 161.36094674556213, "grad_norm": 0.06580699980258942, "learning_rate": 6.294787604563267e-06, "loss": 0.0011, "step": 27270 }, { "epoch": 161.42011834319527, "grad_norm": 0.042916521430015564, "learning_rate": 6.249078724538325e-06, "loss": 0.0014, "step": 27280 }, { "epoch": 161.4792899408284, "grad_norm": 0.04087061062455177, "learning_rate": 6.2035328740556e-06, "loss": 0.0012, "step": 27290 }, { "epoch": 161.53846153846155, "grad_norm": 0.031898025423288345, "learning_rate": 6.158150104769155e-06, "loss": 0.0012, "step": 27300 }, { "epoch": 161.5976331360947, "grad_norm": 0.027494361624121666, "learning_rate": 6.112930468148119e-06, "loss": 0.0011, "step": 27310 }, { "epoch": 161.65680473372782, "grad_norm": 0.03009280562400818, "learning_rate": 6.067874015476571e-06, "loss": 0.0014, "step": 27320 }, { "epoch": 161.71597633136093, "grad_norm": 0.03593838959932327, "learning_rate": 6.0229807978535495e-06, "loss": 0.0011, "step": 27330 }, { "epoch": 161.77514792899407, "grad_norm": 0.0337238572537899, "learning_rate": 5.978250866192963e-06, "loss": 0.0011, "step": 27340 }, { "epoch": 161.8343195266272, "grad_norm": 0.026686813682317734, "learning_rate": 5.933684271223532e-06, "loss": 0.0014, "step": 27350 }, { "epoch": 161.89349112426035, "grad_norm": 0.03195948526263237, "learning_rate": 5.889281063488743e-06, "loss": 0.0012, "step": 27360 }, { "epoch": 161.9526627218935, "grad_norm": 0.03291475772857666, "learning_rate": 5.845041293346747e-06, "loss": 0.001, "step": 27370 }, { "epoch": 162.01183431952663, "grad_norm": 0.023576272651553154, "learning_rate": 5.800965010970393e-06, "loss": 0.0011, "step": 27380 }, { "epoch": 162.07100591715977, "grad_norm": 0.023930279538035393, "learning_rate": 5.757052266347078e-06, "loss": 0.0011, "step": 27390 }, { "epoch": 162.1301775147929, "grad_norm": 0.032238878309726715, "learning_rate": 5.713303109278749e-06, "loss": 0.0015, "step": 27400 }, { "epoch": 162.18934911242604, "grad_norm": 0.03261534869670868, "learning_rate": 5.669717589381817e-06, "loss": 0.0012, "step": 27410 }, { "epoch": 162.24852071005918, "grad_norm": 0.041401080787181854, "learning_rate": 5.626295756087107e-06, "loss": 0.0013, "step": 27420 }, { "epoch": 162.30769230769232, "grad_norm": 0.04900214821100235, "learning_rate": 5.58303765863981e-06, "loss": 0.0014, "step": 27430 }, { "epoch": 162.36686390532543, "grad_norm": 0.026281192898750305, "learning_rate": 5.539943346099418e-06, "loss": 0.0011, "step": 27440 }, { "epoch": 162.42603550295857, "grad_norm": 0.038955751806497574, "learning_rate": 5.497012867339701e-06, "loss": 0.0011, "step": 27450 }, { "epoch": 162.4852071005917, "grad_norm": 0.02418040856719017, "learning_rate": 5.454246271048596e-06, "loss": 0.0011, "step": 27460 }, { "epoch": 162.54437869822485, "grad_norm": 0.03453129529953003, "learning_rate": 5.411643605728139e-06, "loss": 0.0012, "step": 27470 }, { "epoch": 162.60355029585799, "grad_norm": 0.039931997656822205, "learning_rate": 5.3692049196945475e-06, "loss": 0.0011, "step": 27480 }, { "epoch": 162.66272189349112, "grad_norm": 0.04695951193571091, "learning_rate": 5.326930261077972e-06, "loss": 0.0012, "step": 27490 }, { "epoch": 162.72189349112426, "grad_norm": 0.038931939750909805, "learning_rate": 5.28481967782261e-06, "loss": 0.0012, "step": 27500 }, { "epoch": 162.7810650887574, "grad_norm": 0.06568073481321335, "learning_rate": 5.242873217686527e-06, "loss": 0.0013, "step": 27510 }, { "epoch": 162.84023668639054, "grad_norm": 0.04580662399530411, "learning_rate": 5.201090928241719e-06, "loss": 0.0013, "step": 27520 }, { "epoch": 162.89940828402368, "grad_norm": 0.03353426232933998, "learning_rate": 5.1594728568738976e-06, "loss": 0.0011, "step": 27530 }, { "epoch": 162.95857988165682, "grad_norm": 0.025611143559217453, "learning_rate": 5.118019050782624e-06, "loss": 0.0011, "step": 27540 }, { "epoch": 163.01775147928993, "grad_norm": 0.029811209067702293, "learning_rate": 5.07672955698109e-06, "loss": 0.0011, "step": 27550 }, { "epoch": 163.07692307692307, "grad_norm": 0.03415427356958389, "learning_rate": 5.035604422296224e-06, "loss": 0.0011, "step": 27560 }, { "epoch": 163.1360946745562, "grad_norm": 0.04504134878516197, "learning_rate": 4.9946436933684665e-06, "loss": 0.0012, "step": 27570 }, { "epoch": 163.19526627218934, "grad_norm": 0.030713722109794617, "learning_rate": 4.95384741665189e-06, "loss": 0.0013, "step": 27580 }, { "epoch": 163.25443786982248, "grad_norm": 0.03824758529663086, "learning_rate": 4.9132156384139535e-06, "loss": 0.0013, "step": 27590 }, { "epoch": 163.31360946745562, "grad_norm": 0.03903384134173393, "learning_rate": 4.872748404735644e-06, "loss": 0.001, "step": 27600 }, { "epoch": 163.37278106508876, "grad_norm": 0.03940550982952118, "learning_rate": 4.8324457615113165e-06, "loss": 0.0012, "step": 27610 }, { "epoch": 163.4319526627219, "grad_norm": 0.037155210971832275, "learning_rate": 4.7923077544486266e-06, "loss": 0.0014, "step": 27620 }, { "epoch": 163.49112426035504, "grad_norm": 0.027334239333868027, "learning_rate": 4.752334429068577e-06, "loss": 0.0012, "step": 27630 }, { "epoch": 163.55029585798817, "grad_norm": 0.03018031269311905, "learning_rate": 4.712525830705338e-06, "loss": 0.0013, "step": 27640 }, { "epoch": 163.6094674556213, "grad_norm": 0.038768451660871506, "learning_rate": 4.6728820045062954e-06, "loss": 0.0012, "step": 27650 }, { "epoch": 163.66863905325442, "grad_norm": 0.04208901524543762, "learning_rate": 4.633402995431968e-06, "loss": 0.0011, "step": 27660 }, { "epoch": 163.72781065088756, "grad_norm": 0.025525158271193504, "learning_rate": 4.5940888482559555e-06, "loss": 0.0011, "step": 27670 }, { "epoch": 163.7869822485207, "grad_norm": 0.03954416513442993, "learning_rate": 4.554939607564861e-06, "loss": 0.0011, "step": 27680 }, { "epoch": 163.84615384615384, "grad_norm": 0.03924742713570595, "learning_rate": 4.515955317758285e-06, "loss": 0.0013, "step": 27690 }, { "epoch": 163.90532544378698, "grad_norm": 0.039566706866025925, "learning_rate": 4.477136023048727e-06, "loss": 0.0011, "step": 27700 }, { "epoch": 163.96449704142012, "grad_norm": 0.07899515330791473, "learning_rate": 4.4384817674616215e-06, "loss": 0.001, "step": 27710 }, { "epoch": 164.02366863905326, "grad_norm": 0.034207798540592194, "learning_rate": 4.399992594835183e-06, "loss": 0.001, "step": 27720 }, { "epoch": 164.0828402366864, "grad_norm": 0.0223845262080431, "learning_rate": 4.361668548820429e-06, "loss": 0.0012, "step": 27730 }, { "epoch": 164.14201183431953, "grad_norm": 0.04006556421518326, "learning_rate": 4.323509672881059e-06, "loss": 0.0012, "step": 27740 }, { "epoch": 164.20118343195267, "grad_norm": 0.04200362041592598, "learning_rate": 4.285516010293522e-06, "loss": 0.0012, "step": 27750 }, { "epoch": 164.2603550295858, "grad_norm": 0.029694829136133194, "learning_rate": 4.247687604146816e-06, "loss": 0.0011, "step": 27760 }, { "epoch": 164.31952662721895, "grad_norm": 0.02380089834332466, "learning_rate": 4.210024497342607e-06, "loss": 0.0012, "step": 27770 }, { "epoch": 164.37869822485206, "grad_norm": 0.04105116426944733, "learning_rate": 4.172526732595044e-06, "loss": 0.0012, "step": 27780 }, { "epoch": 164.4378698224852, "grad_norm": 0.05023727938532829, "learning_rate": 4.13519435243076e-06, "loss": 0.0013, "step": 27790 }, { "epoch": 164.49704142011834, "grad_norm": 0.041416533291339874, "learning_rate": 4.098027399188802e-06, "loss": 0.0011, "step": 27800 }, { "epoch": 164.55621301775147, "grad_norm": 0.05635926499962807, "learning_rate": 4.061025915020655e-06, "loss": 0.0012, "step": 27810 }, { "epoch": 164.6153846153846, "grad_norm": 0.024403348565101624, "learning_rate": 4.024189941890099e-06, "loss": 0.0011, "step": 27820 }, { "epoch": 164.67455621301775, "grad_norm": 0.037694044411182404, "learning_rate": 3.987519521573268e-06, "loss": 0.0011, "step": 27830 }, { "epoch": 164.7337278106509, "grad_norm": 0.03576852008700371, "learning_rate": 3.951014695658494e-06, "loss": 0.0012, "step": 27840 }, { "epoch": 164.79289940828403, "grad_norm": 0.04381725192070007, "learning_rate": 3.914675505546277e-06, "loss": 0.0013, "step": 27850 }, { "epoch": 164.85207100591717, "grad_norm": 0.03800586983561516, "learning_rate": 3.878501992449318e-06, "loss": 0.0011, "step": 27860 }, { "epoch": 164.9112426035503, "grad_norm": 0.022799724712967873, "learning_rate": 3.842494197392398e-06, "loss": 0.001, "step": 27870 }, { "epoch": 164.97041420118344, "grad_norm": 0.05211751535534859, "learning_rate": 3.8066521612123866e-06, "loss": 0.0013, "step": 27880 }, { "epoch": 165.02958579881656, "grad_norm": 0.038634926080703735, "learning_rate": 3.770975924558134e-06, "loss": 0.0012, "step": 27890 }, { "epoch": 165.0887573964497, "grad_norm": 0.027406936511397362, "learning_rate": 3.735465527890458e-06, "loss": 0.0011, "step": 27900 }, { "epoch": 165.14792899408283, "grad_norm": 0.02461623214185238, "learning_rate": 3.7001210114820935e-06, "loss": 0.001, "step": 27910 }, { "epoch": 165.20710059171597, "grad_norm": 0.047361742705106735, "learning_rate": 3.6649424154176597e-06, "loss": 0.0011, "step": 27920 }, { "epoch": 165.2662721893491, "grad_norm": 0.0367387980222702, "learning_rate": 3.62992977959361e-06, "loss": 0.0011, "step": 27930 }, { "epoch": 165.32544378698225, "grad_norm": 0.026743974536657333, "learning_rate": 3.595083143718147e-06, "loss": 0.0011, "step": 27940 }, { "epoch": 165.3846153846154, "grad_norm": 0.03134612366557121, "learning_rate": 3.560402547311275e-06, "loss": 0.0012, "step": 27950 }, { "epoch": 165.44378698224853, "grad_norm": 0.04097175970673561, "learning_rate": 3.525888029704599e-06, "loss": 0.0013, "step": 27960 }, { "epoch": 165.50295857988166, "grad_norm": 0.031156200915575027, "learning_rate": 3.4915396300414567e-06, "loss": 0.0011, "step": 27970 }, { "epoch": 165.5621301775148, "grad_norm": 0.02552892081439495, "learning_rate": 3.4573573872767547e-06, "loss": 0.0011, "step": 27980 }, { "epoch": 165.62130177514794, "grad_norm": 0.029852449893951416, "learning_rate": 3.423341340176933e-06, "loss": 0.0013, "step": 27990 }, { "epoch": 165.68047337278105, "grad_norm": 0.058200061321258545, "learning_rate": 3.3894915273199987e-06, "loss": 0.0011, "step": 28000 }, { "epoch": 165.7396449704142, "grad_norm": 0.029159149155020714, "learning_rate": 3.3558079870954267e-06, "loss": 0.0011, "step": 28010 }, { "epoch": 165.79881656804733, "grad_norm": 0.03717142716050148, "learning_rate": 3.322290757704044e-06, "loss": 0.0011, "step": 28020 }, { "epoch": 165.85798816568047, "grad_norm": 0.05999898537993431, "learning_rate": 3.288939877158159e-06, "loss": 0.0014, "step": 28030 }, { "epoch": 165.9171597633136, "grad_norm": 0.07101496309041977, "learning_rate": 3.2557553832813664e-06, "loss": 0.0012, "step": 28040 }, { "epoch": 165.97633136094674, "grad_norm": 0.030664626508951187, "learning_rate": 3.2227373137085954e-06, "loss": 0.0011, "step": 28050 }, { "epoch": 166.03550295857988, "grad_norm": 0.03373856469988823, "learning_rate": 3.1898857058859917e-06, "loss": 0.0011, "step": 28060 }, { "epoch": 166.09467455621302, "grad_norm": 0.026868822053074837, "learning_rate": 3.1572005970709356e-06, "loss": 0.0012, "step": 28070 }, { "epoch": 166.15384615384616, "grad_norm": 0.024885820224881172, "learning_rate": 3.1246820243319924e-06, "loss": 0.0012, "step": 28080 }, { "epoch": 166.2130177514793, "grad_norm": 0.025530753657221794, "learning_rate": 3.092330024548828e-06, "loss": 0.0012, "step": 28090 }, { "epoch": 166.27218934911244, "grad_norm": 0.03638302534818649, "learning_rate": 3.0601446344122095e-06, "loss": 0.0012, "step": 28100 }, { "epoch": 166.33136094674558, "grad_norm": 0.07561128586530685, "learning_rate": 3.0281258904239713e-06, "loss": 0.0011, "step": 28110 }, { "epoch": 166.3905325443787, "grad_norm": 0.022333651781082153, "learning_rate": 2.9962738288969003e-06, "loss": 0.0011, "step": 28120 }, { "epoch": 166.44970414201183, "grad_norm": 0.0347466841340065, "learning_rate": 2.9645884859547997e-06, "loss": 0.0012, "step": 28130 }, { "epoch": 166.50887573964496, "grad_norm": 0.04344884678721428, "learning_rate": 2.9330698975323753e-06, "loss": 0.0013, "step": 28140 }, { "epoch": 166.5680473372781, "grad_norm": 0.05664549022912979, "learning_rate": 2.9017180993752e-06, "loss": 0.0011, "step": 28150 }, { "epoch": 166.62721893491124, "grad_norm": 0.045898281037807465, "learning_rate": 2.8705331270397312e-06, "loss": 0.0011, "step": 28160 }, { "epoch": 166.68639053254438, "grad_norm": 0.031003886833786964, "learning_rate": 2.8395150158931624e-06, "loss": 0.001, "step": 28170 }, { "epoch": 166.74556213017752, "grad_norm": 0.037636131048202515, "learning_rate": 2.808663801113503e-06, "loss": 0.0011, "step": 28180 }, { "epoch": 166.80473372781066, "grad_norm": 0.028789738193154335, "learning_rate": 2.777979517689466e-06, "loss": 0.0011, "step": 28190 }, { "epoch": 166.8639053254438, "grad_norm": 0.03726493939757347, "learning_rate": 2.7474622004204304e-06, "loss": 0.0011, "step": 28200 }, { "epoch": 166.92307692307693, "grad_norm": 0.06477080285549164, "learning_rate": 2.717111883916445e-06, "loss": 0.0011, "step": 28210 }, { "epoch": 166.98224852071007, "grad_norm": 0.046521998941898346, "learning_rate": 2.6869286025981427e-06, "loss": 0.0012, "step": 28220 }, { "epoch": 167.04142011834318, "grad_norm": 0.013005734421312809, "learning_rate": 2.656912390696708e-06, "loss": 0.0011, "step": 28230 }, { "epoch": 167.10059171597632, "grad_norm": 0.18033915758132935, "learning_rate": 2.6270632822538607e-06, "loss": 0.0013, "step": 28240 }, { "epoch": 167.15976331360946, "grad_norm": 0.05131607502698898, "learning_rate": 2.5973813111218546e-06, "loss": 0.0012, "step": 28250 }, { "epoch": 167.2189349112426, "grad_norm": 0.02848578803241253, "learning_rate": 2.567866510963329e-06, "loss": 0.0012, "step": 28260 }, { "epoch": 167.27810650887574, "grad_norm": 0.03416970744729042, "learning_rate": 2.5385189152513253e-06, "loss": 0.0011, "step": 28270 }, { "epoch": 167.33727810650888, "grad_norm": 0.030682988464832306, "learning_rate": 2.5093385572693025e-06, "loss": 0.0011, "step": 28280 }, { "epoch": 167.39644970414201, "grad_norm": 0.049017295241355896, "learning_rate": 2.4803254701110385e-06, "loss": 0.0012, "step": 28290 }, { "epoch": 167.45562130177515, "grad_norm": 0.0346975140273571, "learning_rate": 2.4514796866805964e-06, "loss": 0.0009, "step": 28300 }, { "epoch": 167.5147928994083, "grad_norm": 0.02974826656281948, "learning_rate": 2.422801239692307e-06, "loss": 0.0011, "step": 28310 }, { "epoch": 167.57396449704143, "grad_norm": 0.04890916123986244, "learning_rate": 2.3942901616707365e-06, "loss": 0.0012, "step": 28320 }, { "epoch": 167.63313609467457, "grad_norm": 0.04378668591380119, "learning_rate": 2.365946484950587e-06, "loss": 0.0011, "step": 28330 }, { "epoch": 167.69230769230768, "grad_norm": 0.0260001327842474, "learning_rate": 2.3377702416767618e-06, "loss": 0.0009, "step": 28340 }, { "epoch": 167.75147928994082, "grad_norm": 0.03159381449222565, "learning_rate": 2.3097614638042493e-06, "loss": 0.0011, "step": 28350 }, { "epoch": 167.81065088757396, "grad_norm": 0.0502978190779686, "learning_rate": 2.2819201830981404e-06, "loss": 0.0013, "step": 28360 }, { "epoch": 167.8698224852071, "grad_norm": 0.149572491645813, "learning_rate": 2.2542464311335105e-06, "loss": 0.0012, "step": 28370 }, { "epoch": 167.92899408284023, "grad_norm": 0.0356571227312088, "learning_rate": 2.2267402392955215e-06, "loss": 0.0012, "step": 28380 }, { "epoch": 167.98816568047337, "grad_norm": 0.01900549791753292, "learning_rate": 2.1994016387792024e-06, "loss": 0.0011, "step": 28390 }, { "epoch": 168.0473372781065, "grad_norm": 0.025296606123447418, "learning_rate": 2.1722306605896022e-06, "loss": 0.0011, "step": 28400 }, { "epoch": 168.10650887573965, "grad_norm": 0.040859922766685486, "learning_rate": 2.1452273355416216e-06, "loss": 0.0011, "step": 28410 }, { "epoch": 168.1656804733728, "grad_norm": 0.02517537586390972, "learning_rate": 2.1183916942600464e-06, "loss": 0.001, "step": 28420 }, { "epoch": 168.22485207100593, "grad_norm": 0.024912096560001373, "learning_rate": 2.0917237671794983e-06, "loss": 0.0011, "step": 28430 }, { "epoch": 168.28402366863907, "grad_norm": 0.03008958138525486, "learning_rate": 2.0652235845443175e-06, "loss": 0.0011, "step": 28440 }, { "epoch": 168.34319526627218, "grad_norm": 0.03842540457844734, "learning_rate": 2.0388911764086966e-06, "loss": 0.0012, "step": 28450 }, { "epoch": 168.4023668639053, "grad_norm": 0.03645533695816994, "learning_rate": 2.012726572636514e-06, "loss": 0.0012, "step": 28460 }, { "epoch": 168.46153846153845, "grad_norm": 0.03187400847673416, "learning_rate": 1.986729802901349e-06, "loss": 0.0011, "step": 28470 }, { "epoch": 168.5207100591716, "grad_norm": 0.03296166658401489, "learning_rate": 1.9609008966864347e-06, "loss": 0.0012, "step": 28480 }, { "epoch": 168.57988165680473, "grad_norm": 0.02847478538751602, "learning_rate": 1.935239883284606e-06, "loss": 0.0011, "step": 28490 }, { "epoch": 168.63905325443787, "grad_norm": 0.02259470336139202, "learning_rate": 1.909746791798317e-06, "loss": 0.001, "step": 28500 }, { "epoch": 168.698224852071, "grad_norm": 0.02781587280333042, "learning_rate": 1.8844216511395726e-06, "loss": 0.0011, "step": 28510 }, { "epoch": 168.75739644970415, "grad_norm": 0.03142133355140686, "learning_rate": 1.8592644900298826e-06, "loss": 0.0013, "step": 28520 }, { "epoch": 168.81656804733728, "grad_norm": 0.02661747857928276, "learning_rate": 1.8342753370003072e-06, "loss": 0.0013, "step": 28530 }, { "epoch": 168.87573964497042, "grad_norm": 0.040934402495622635, "learning_rate": 1.8094542203912932e-06, "loss": 0.0011, "step": 28540 }, { "epoch": 168.93491124260356, "grad_norm": 0.018753712996840477, "learning_rate": 1.7848011683527562e-06, "loss": 0.0012, "step": 28550 }, { "epoch": 168.9940828402367, "grad_norm": 0.03834070637822151, "learning_rate": 1.7603162088440148e-06, "loss": 0.0012, "step": 28560 }, { "epoch": 169.0532544378698, "grad_norm": 0.024939823895692825, "learning_rate": 1.735999369633706e-06, "loss": 0.0011, "step": 28570 }, { "epoch": 169.11242603550295, "grad_norm": 0.03428959846496582, "learning_rate": 1.7118506782998699e-06, "loss": 0.0012, "step": 28580 }, { "epoch": 169.1715976331361, "grad_norm": 0.041967105120420456, "learning_rate": 1.687870162229782e-06, "loss": 0.0011, "step": 28590 }, { "epoch": 169.23076923076923, "grad_norm": 0.024704884737730026, "learning_rate": 1.6640578486200373e-06, "loss": 0.0011, "step": 28600 }, { "epoch": 169.28994082840237, "grad_norm": 0.04156314209103584, "learning_rate": 1.6404137644764338e-06, "loss": 0.0013, "step": 28610 }, { "epoch": 169.3491124260355, "grad_norm": 0.08002779632806778, "learning_rate": 1.6169379366139878e-06, "loss": 0.0012, "step": 28620 }, { "epoch": 169.40828402366864, "grad_norm": 0.04607382044196129, "learning_rate": 1.5936303916569193e-06, "loss": 0.001, "step": 28630 }, { "epoch": 169.46745562130178, "grad_norm": 0.0175462793558836, "learning_rate": 1.5704911560385668e-06, "loss": 0.0011, "step": 28640 }, { "epoch": 169.52662721893492, "grad_norm": 0.02522234432399273, "learning_rate": 1.5475202560014054e-06, "loss": 0.0011, "step": 28650 }, { "epoch": 169.58579881656806, "grad_norm": 0.026574043557047844, "learning_rate": 1.5247177175969793e-06, "loss": 0.0011, "step": 28660 }, { "epoch": 169.6449704142012, "grad_norm": 0.0984821543097496, "learning_rate": 1.5020835666859354e-06, "loss": 0.0012, "step": 28670 }, { "epoch": 169.7041420118343, "grad_norm": 0.031106052920222282, "learning_rate": 1.4796178289378735e-06, "loss": 0.0013, "step": 28680 }, { "epoch": 169.76331360946745, "grad_norm": 0.02022092416882515, "learning_rate": 1.457320529831496e-06, "loss": 0.001, "step": 28690 }, { "epoch": 169.82248520710058, "grad_norm": 0.19717851281166077, "learning_rate": 1.435191694654375e-06, "loss": 0.0013, "step": 28700 }, { "epoch": 169.88165680473372, "grad_norm": 0.03641102835536003, "learning_rate": 1.4132313485030856e-06, "loss": 0.0011, "step": 28710 }, { "epoch": 169.94082840236686, "grad_norm": 0.03270215913653374, "learning_rate": 1.391439516283105e-06, "loss": 0.0012, "step": 28720 }, { "epoch": 170.0, "grad_norm": 0.05220576003193855, "learning_rate": 1.3698162227087971e-06, "loss": 0.0011, "step": 28730 }, { "epoch": 170.05917159763314, "grad_norm": 0.03141125291585922, "learning_rate": 1.3483614923033792e-06, "loss": 0.0011, "step": 28740 }, { "epoch": 170.11834319526628, "grad_norm": 0.03559339419007301, "learning_rate": 1.3270753493989373e-06, "loss": 0.0012, "step": 28750 }, { "epoch": 170.17751479289942, "grad_norm": 0.03524290770292282, "learning_rate": 1.3059578181362607e-06, "loss": 0.0011, "step": 28760 }, { "epoch": 170.23668639053255, "grad_norm": 0.03194878622889519, "learning_rate": 1.2850089224650418e-06, "loss": 0.001, "step": 28770 }, { "epoch": 170.2958579881657, "grad_norm": 0.03815016150474548, "learning_rate": 1.2642286861436256e-06, "loss": 0.0012, "step": 28780 }, { "epoch": 170.3550295857988, "grad_norm": 0.04413263499736786, "learning_rate": 1.243617132739111e-06, "loss": 0.0012, "step": 28790 }, { "epoch": 170.41420118343194, "grad_norm": 0.029154270887374878, "learning_rate": 1.2231742856273151e-06, "loss": 0.0011, "step": 28800 }, { "epoch": 170.47337278106508, "grad_norm": 0.027199868112802505, "learning_rate": 1.202900167992693e-06, "loss": 0.0011, "step": 28810 }, { "epoch": 170.53254437869822, "grad_norm": 0.051586396992206573, "learning_rate": 1.1827948028283352e-06, "loss": 0.0011, "step": 28820 }, { "epoch": 170.59171597633136, "grad_norm": 0.17643381655216217, "learning_rate": 1.1628582129359686e-06, "loss": 0.001, "step": 28830 }, { "epoch": 170.6508875739645, "grad_norm": 0.04468314349651337, "learning_rate": 1.143090420925924e-06, "loss": 0.0011, "step": 28840 }, { "epoch": 170.71005917159763, "grad_norm": 0.023284245282411575, "learning_rate": 1.1234914492170678e-06, "loss": 0.0011, "step": 28850 }, { "epoch": 170.76923076923077, "grad_norm": 0.020489685237407684, "learning_rate": 1.1040613200368032e-06, "loss": 0.0011, "step": 28860 }, { "epoch": 170.8284023668639, "grad_norm": 0.0356673002243042, "learning_rate": 1.08480005542107e-06, "loss": 0.001, "step": 28870 }, { "epoch": 170.88757396449705, "grad_norm": 0.021655822172760963, "learning_rate": 1.0657076772142782e-06, "loss": 0.0011, "step": 28880 }, { "epoch": 170.9467455621302, "grad_norm": 0.043531160801649094, "learning_rate": 1.0467842070693234e-06, "loss": 0.0012, "step": 28890 }, { "epoch": 171.0059171597633, "grad_norm": 0.029459869489073753, "learning_rate": 1.0280296664475218e-06, "loss": 0.0012, "step": 28900 }, { "epoch": 171.06508875739644, "grad_norm": 0.030519504100084305, "learning_rate": 1.0094440766185929e-06, "loss": 0.0011, "step": 28910 }, { "epoch": 171.12426035502958, "grad_norm": 0.017614765092730522, "learning_rate": 9.910274586606925e-07, "loss": 0.001, "step": 28920 }, { "epoch": 171.18343195266272, "grad_norm": 0.026530291885137558, "learning_rate": 9.727798334602799e-07, "loss": 0.0011, "step": 28930 }, { "epoch": 171.24260355029585, "grad_norm": 0.023359982296824455, "learning_rate": 9.54701221712234e-07, "loss": 0.0011, "step": 28940 }, { "epoch": 171.301775147929, "grad_norm": 0.025930028408765793, "learning_rate": 9.367916439196709e-07, "loss": 0.0011, "step": 28950 }, { "epoch": 171.36094674556213, "grad_norm": 0.0235223900526762, "learning_rate": 9.190511203940765e-07, "loss": 0.0011, "step": 28960 }, { "epoch": 171.42011834319527, "grad_norm": 0.05453072115778923, "learning_rate": 9.014796712551564e-07, "loss": 0.0012, "step": 28970 }, { "epoch": 171.4792899408284, "grad_norm": 0.031990569084882736, "learning_rate": 8.840773164309034e-07, "loss": 0.0011, "step": 28980 }, { "epoch": 171.53846153846155, "grad_norm": 0.025053737685084343, "learning_rate": 8.668440756575301e-07, "loss": 0.0012, "step": 28990 }, { "epoch": 171.5976331360947, "grad_norm": 0.04143049195408821, "learning_rate": 8.49779968479436e-07, "loss": 0.0011, "step": 29000 }, { "epoch": 171.65680473372782, "grad_norm": 0.05615515261888504, "learning_rate": 8.328850142492072e-07, "loss": 0.0011, "step": 29010 }, { "epoch": 171.71597633136093, "grad_norm": 0.05463775247335434, "learning_rate": 8.161592321276333e-07, "loss": 0.0012, "step": 29020 }, { "epoch": 171.77514792899407, "grad_norm": 0.02986869588494301, "learning_rate": 7.996026410835909e-07, "loss": 0.0012, "step": 29030 }, { "epoch": 171.8343195266272, "grad_norm": 0.03598490357398987, "learning_rate": 7.832152598941099e-07, "loss": 0.001, "step": 29040 }, { "epoch": 171.89349112426035, "grad_norm": 0.03100016340613365, "learning_rate": 7.669971071442738e-07, "loss": 0.0011, "step": 29050 }, { "epoch": 171.9526627218935, "grad_norm": 0.05158131569623947, "learning_rate": 7.509482012273193e-07, "loss": 0.0011, "step": 29060 }, { "epoch": 172.01183431952663, "grad_norm": 0.0227607823908329, "learning_rate": 7.350685603444539e-07, "loss": 0.0013, "step": 29070 }, { "epoch": 172.07100591715977, "grad_norm": 0.025681480765342712, "learning_rate": 7.193582025049715e-07, "loss": 0.001, "step": 29080 }, { "epoch": 172.1301775147929, "grad_norm": 0.040998201817274094, "learning_rate": 7.038171455261699e-07, "loss": 0.0011, "step": 29090 }, { "epoch": 172.18934911242604, "grad_norm": 0.03212762624025345, "learning_rate": 6.884454070333168e-07, "loss": 0.001, "step": 29100 }, { "epoch": 172.24852071005918, "grad_norm": 0.026856329292058945, "learning_rate": 6.73243004459717e-07, "loss": 0.0011, "step": 29110 }, { "epoch": 172.30769230769232, "grad_norm": 0.038926128298044205, "learning_rate": 6.582099550465458e-07, "loss": 0.0011, "step": 29120 }, { "epoch": 172.36686390532543, "grad_norm": 0.05420876666903496, "learning_rate": 6.433462758429653e-07, "loss": 0.0013, "step": 29130 }, { "epoch": 172.42603550295857, "grad_norm": 0.055635180324316025, "learning_rate": 6.286519837060578e-07, "loss": 0.0013, "step": 29140 }, { "epoch": 172.4852071005917, "grad_norm": 0.03050680086016655, "learning_rate": 6.141270953007593e-07, "loss": 0.0011, "step": 29150 }, { "epoch": 172.54437869822485, "grad_norm": 0.028521088883280754, "learning_rate": 5.997716270999264e-07, "loss": 0.0011, "step": 29160 }, { "epoch": 172.60355029585799, "grad_norm": 0.03263622522354126, "learning_rate": 5.855855953842692e-07, "loss": 0.0013, "step": 29170 }, { "epoch": 172.66272189349112, "grad_norm": 0.02938295714557171, "learning_rate": 5.715690162423014e-07, "loss": 0.001, "step": 29180 }, { "epoch": 172.72189349112426, "grad_norm": 0.04246923327445984, "learning_rate": 5.577219055703907e-07, "loss": 0.0012, "step": 29190 }, { "epoch": 172.7810650887574, "grad_norm": 0.03278544917702675, "learning_rate": 5.440442790727085e-07, "loss": 0.0011, "step": 29200 }, { "epoch": 172.84023668639054, "grad_norm": 0.04122905060648918, "learning_rate": 5.305361522611962e-07, "loss": 0.001, "step": 29210 }, { "epoch": 172.89940828402368, "grad_norm": 0.022461548447608948, "learning_rate": 5.171975404555828e-07, "loss": 0.0009, "step": 29220 }, { "epoch": 172.95857988165682, "grad_norm": 0.0344880148768425, "learning_rate": 5.04028458783351e-07, "loss": 0.0012, "step": 29230 }, { "epoch": 173.01775147928993, "grad_norm": 0.019624844193458557, "learning_rate": 4.910289221796703e-07, "loss": 0.0011, "step": 29240 }, { "epoch": 173.07692307692307, "grad_norm": 0.01999909244477749, "learning_rate": 4.781989453874813e-07, "loss": 0.001, "step": 29250 }, { "epoch": 173.1360946745562, "grad_norm": 0.021129610016942024, "learning_rate": 4.65538542957411e-07, "loss": 0.0012, "step": 29260 }, { "epoch": 173.19526627218934, "grad_norm": 0.0225373562425375, "learning_rate": 4.530477292477408e-07, "loss": 0.0011, "step": 29270 }, { "epoch": 173.25443786982248, "grad_norm": 0.025503888726234436, "learning_rate": 4.407265184244724e-07, "loss": 0.0012, "step": 29280 }, { "epoch": 173.31360946745562, "grad_norm": 0.022432079538702965, "learning_rate": 4.2857492446122797e-07, "loss": 0.001, "step": 29290 }, { "epoch": 173.37278106508876, "grad_norm": 0.026211801916360855, "learning_rate": 4.1659296113925046e-07, "loss": 0.0012, "step": 29300 }, { "epoch": 173.4319526627219, "grad_norm": 0.034175410866737366, "learning_rate": 4.0478064204741976e-07, "loss": 0.0011, "step": 29310 }, { "epoch": 173.49112426035504, "grad_norm": 0.03532750532031059, "learning_rate": 3.931379805822199e-07, "loss": 0.0011, "step": 29320 }, { "epoch": 173.55029585798817, "grad_norm": 0.02554503083229065, "learning_rate": 3.816649899477553e-07, "loss": 0.0012, "step": 29330 }, { "epoch": 173.6094674556213, "grad_norm": 0.02328430488705635, "learning_rate": 3.703616831556344e-07, "loss": 0.001, "step": 29340 }, { "epoch": 173.66863905325442, "grad_norm": 0.02745955064892769, "learning_rate": 3.592280730250863e-07, "loss": 0.0012, "step": 29350 }, { "epoch": 173.72781065088756, "grad_norm": 0.02243616245687008, "learning_rate": 3.482641721828605e-07, "loss": 0.0011, "step": 29360 }, { "epoch": 173.7869822485207, "grad_norm": 0.05147891864180565, "learning_rate": 3.374699930632441e-07, "loss": 0.0012, "step": 29370 }, { "epoch": 173.84615384615384, "grad_norm": 0.03717498108744621, "learning_rate": 3.268455479080279e-07, "loss": 0.0012, "step": 29380 }, { "epoch": 173.90532544378698, "grad_norm": 0.04734199121594429, "learning_rate": 3.163908487665234e-07, "loss": 0.0012, "step": 29390 }, { "epoch": 173.96449704142012, "grad_norm": 0.05027680844068527, "learning_rate": 3.06105907495513e-07, "loss": 0.001, "step": 29400 }, { "epoch": 174.02366863905326, "grad_norm": 0.034389618784189224, "learning_rate": 2.959907357592661e-07, "loss": 0.0011, "step": 29410 }, { "epoch": 174.0828402366864, "grad_norm": 0.02802363969385624, "learning_rate": 2.86045345029523e-07, "loss": 0.0011, "step": 29420 }, { "epoch": 174.14201183431953, "grad_norm": 0.030743997544050217, "learning_rate": 2.762697465854613e-07, "loss": 0.0011, "step": 29430 }, { "epoch": 174.20118343195267, "grad_norm": 0.0424170047044754, "learning_rate": 2.666639515136959e-07, "loss": 0.0011, "step": 29440 }, { "epoch": 174.2603550295858, "grad_norm": 0.0379612110555172, "learning_rate": 2.572279707082625e-07, "loss": 0.0011, "step": 29450 }, { "epoch": 174.31952662721895, "grad_norm": 0.03199760988354683, "learning_rate": 2.479618148706175e-07, "loss": 0.0012, "step": 29460 }, { "epoch": 174.37869822485206, "grad_norm": 0.03436044976115227, "learning_rate": 2.388654945096213e-07, "loss": 0.0012, "step": 29470 }, { "epoch": 174.4378698224852, "grad_norm": 0.05076673626899719, "learning_rate": 2.2993901994148855e-07, "loss": 0.0011, "step": 29480 }, { "epoch": 174.49704142011834, "grad_norm": 0.03640078380703926, "learning_rate": 2.2118240128985444e-07, "loss": 0.0011, "step": 29490 }, { "epoch": 174.55621301775147, "grad_norm": 0.019864922389388084, "learning_rate": 2.125956484857083e-07, "loss": 0.0011, "step": 29500 }, { "epoch": 174.6153846153846, "grad_norm": 0.047587841749191284, "learning_rate": 2.0417877126736037e-07, "loss": 0.0012, "step": 29510 }, { "epoch": 174.67455621301775, "grad_norm": 0.027537493035197258, "learning_rate": 1.9593177918050818e-07, "loss": 0.0012, "step": 29520 }, { "epoch": 174.7337278106509, "grad_norm": 0.027940399944782257, "learning_rate": 1.8785468157813677e-07, "loss": 0.0009, "step": 29530 }, { "epoch": 174.79289940828403, "grad_norm": 0.04204829782247543, "learning_rate": 1.7994748762058532e-07, "loss": 0.0012, "step": 29540 }, { "epoch": 174.85207100591717, "grad_norm": 0.028614308685064316, "learning_rate": 1.7221020627548043e-07, "loss": 0.0011, "step": 29550 }, { "epoch": 174.9112426035503, "grad_norm": 0.03440850228071213, "learning_rate": 1.6464284631778625e-07, "loss": 0.001, "step": 29560 }, { "epoch": 174.97041420118344, "grad_norm": 0.03683191165328026, "learning_rate": 1.5724541632970432e-07, "loss": 0.0014, "step": 29570 }, { "epoch": 175.02958579881656, "grad_norm": 0.031064806506037712, "learning_rate": 1.5001792470075714e-07, "loss": 0.0012, "step": 29580 }, { "epoch": 175.0887573964497, "grad_norm": 0.031109299510717392, "learning_rate": 1.4296037962772122e-07, "loss": 0.0013, "step": 29590 }, { "epoch": 175.14792899408283, "grad_norm": 0.038249120116233826, "learning_rate": 1.3607278911462739e-07, "loss": 0.0009, "step": 29600 }, { "epoch": 175.20710059171597, "grad_norm": 0.029489681124687195, "learning_rate": 1.2935516097279387e-07, "loss": 0.0011, "step": 29610 }, { "epoch": 175.2662721893491, "grad_norm": 0.027515605092048645, "learning_rate": 1.2280750282074315e-07, "loss": 0.0011, "step": 29620 }, { "epoch": 175.32544378698225, "grad_norm": 0.025731978937983513, "learning_rate": 1.1642982208425189e-07, "loss": 0.0011, "step": 29630 }, { "epoch": 175.3846153846154, "grad_norm": 0.034823931753635406, "learning_rate": 1.1022212599631763e-07, "loss": 0.0012, "step": 29640 }, { "epoch": 175.44378698224853, "grad_norm": 0.04121694341301918, "learning_rate": 1.0418442159715879e-07, "loss": 0.0012, "step": 29650 }, { "epoch": 175.50295857988166, "grad_norm": 0.048361875116825104, "learning_rate": 9.831671573419797e-08, "loss": 0.0011, "step": 29660 }, { "epoch": 175.5621301775148, "grad_norm": 0.03612625226378441, "learning_rate": 9.261901506207869e-08, "loss": 0.0012, "step": 29670 }, { "epoch": 175.62130177514794, "grad_norm": 0.020030546933412552, "learning_rate": 8.709132604261537e-08, "loss": 0.0011, "step": 29680 }, { "epoch": 175.68047337278105, "grad_norm": 0.0327129028737545, "learning_rate": 8.17336549448433e-08, "loss": 0.0011, "step": 29690 }, { "epoch": 175.7396449704142, "grad_norm": 0.03432333096861839, "learning_rate": 7.654600784493536e-08, "loss": 0.0011, "step": 29700 }, { "epoch": 175.79881656804733, "grad_norm": 0.025099564343690872, "learning_rate": 7.15283906262687e-08, "loss": 0.0011, "step": 29710 }, { "epoch": 175.85798816568047, "grad_norm": 0.06272748857736588, "learning_rate": 6.668080897937466e-08, "loss": 0.0011, "step": 29720 }, { "epoch": 175.9171597633136, "grad_norm": 0.025692664086818695, "learning_rate": 6.200326840195558e-08, "loss": 0.001, "step": 29730 }, { "epoch": 175.97633136094674, "grad_norm": 0.03621475398540497, "learning_rate": 5.749577419886797e-08, "loss": 0.0012, "step": 29740 }, { "epoch": 176.03550295857988, "grad_norm": 0.021905768662691116, "learning_rate": 5.315833148210602e-08, "loss": 0.0012, "step": 29750 }, { "epoch": 176.09467455621302, "grad_norm": 0.03639153391122818, "learning_rate": 4.899094517081814e-08, "loss": 0.001, "step": 29760 }, { "epoch": 176.15384615384616, "grad_norm": 0.02187158353626728, "learning_rate": 4.499361999127371e-08, "loss": 0.0011, "step": 29770 }, { "epoch": 176.2130177514793, "grad_norm": 0.02241458371281624, "learning_rate": 4.1166360476896366e-08, "loss": 0.0012, "step": 29780 }, { "epoch": 176.27218934911244, "grad_norm": 0.02091522142291069, "learning_rate": 3.750917096821404e-08, "loss": 0.0011, "step": 29790 }, { "epoch": 176.33136094674558, "grad_norm": 0.05495670437812805, "learning_rate": 3.402205561289229e-08, "loss": 0.001, "step": 29800 }, { "epoch": 176.3905325443787, "grad_norm": 0.04948543757200241, "learning_rate": 3.070501836570094e-08, "loss": 0.0011, "step": 29810 }, { "epoch": 176.44970414201183, "grad_norm": 0.04039497673511505, "learning_rate": 2.7558062988530805e-08, "loss": 0.0012, "step": 29820 }, { "epoch": 176.50887573964496, "grad_norm": 0.028666887432336807, "learning_rate": 2.4581193050393632e-08, "loss": 0.0011, "step": 29830 }, { "epoch": 176.5680473372781, "grad_norm": 0.05701587721705437, "learning_rate": 2.1774411927372172e-08, "loss": 0.0011, "step": 29840 }, { "epoch": 176.62721893491124, "grad_norm": 0.026706289499998093, "learning_rate": 1.9137722802686772e-08, "loss": 0.001, "step": 29850 }, { "epoch": 176.68639053254438, "grad_norm": 0.03359087556600571, "learning_rate": 1.6671128666612132e-08, "loss": 0.0011, "step": 29860 }, { "epoch": 176.74556213017752, "grad_norm": 0.02895643375813961, "learning_rate": 1.4374632316560553e-08, "loss": 0.0012, "step": 29870 }, { "epoch": 176.80473372781066, "grad_norm": 0.04767026752233505, "learning_rate": 1.2248236357015329e-08, "loss": 0.0012, "step": 29880 }, { "epoch": 176.8639053254438, "grad_norm": 0.02612658590078354, "learning_rate": 1.0291943199547404e-08, "loss": 0.0011, "step": 29890 }, { "epoch": 176.92307692307693, "grad_norm": 0.030389180406928062, "learning_rate": 8.50575506278206e-09, "loss": 0.0011, "step": 29900 }, { "epoch": 176.98224852071007, "grad_norm": 0.040550533682107925, "learning_rate": 6.889673972498844e-09, "loss": 0.0012, "step": 29910 }, { "epoch": 177.04142011834318, "grad_norm": 0.02581489086151123, "learning_rate": 5.443701761465025e-09, "loss": 0.0011, "step": 29920 }, { "epoch": 177.10059171597632, "grad_norm": 0.02498718909919262, "learning_rate": 4.1678400696187885e-09, "loss": 0.0011, "step": 29930 }, { "epoch": 177.15976331360946, "grad_norm": 0.04194368049502373, "learning_rate": 3.0620903439193546e-09, "loss": 0.0011, "step": 29940 }, { "epoch": 177.2189349112426, "grad_norm": 0.054265301674604416, "learning_rate": 2.126453838380282e-09, "loss": 0.0012, "step": 29950 }, { "epoch": 177.27810650887574, "grad_norm": 0.044539887458086014, "learning_rate": 1.3609316141527382e-09, "loss": 0.0012, "step": 29960 }, { "epoch": 177.33727810650888, "grad_norm": 0.024687524884939194, "learning_rate": 7.655245394089237e-10, "loss": 0.0009, "step": 29970 }, { "epoch": 177.39644970414201, "grad_norm": 0.045133329927921295, "learning_rate": 3.402332893920334e-10, "loss": 0.0012, "step": 29980 }, { "epoch": 177.45562130177515, "grad_norm": 0.0320991612970829, "learning_rate": 8.505834646621579e-11, "loss": 0.001, "step": 29990 }, { "epoch": 177.5147928994083, "grad_norm": 0.03418470546603203, "learning_rate": 0.0, "loss": 0.001, "step": 30000 }, { "epoch": 177.5147928994083, "step": 30000, "total_flos": 0.0, "train_loss": 0.007248312342415253, "train_runtime": 6793.6703, "train_samples_per_second": 282.616, "train_steps_per_second": 4.416 } ], "logging_steps": 10, "max_steps": 30000, "num_input_tokens_seen": 0, "num_train_epochs": 178, "save_steps": 10000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 64, "trial_name": null, "trial_params": null }