{ "best_metric": null, "best_model_checkpoint": null, "epoch": 229.00763358778627, "eval_steps": 2000, "global_step": 30000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.07633587786259542, "grad_norm": 13.297364234924316, "learning_rate": 5.999999999999999e-06, "loss": 0.6993, "step": 10 }, { "epoch": 0.15267175572519084, "grad_norm": 6.46409273147583, "learning_rate": 1.1999999999999999e-05, "loss": 0.3274, "step": 20 }, { "epoch": 0.22900763358778625, "grad_norm": 1.9159376621246338, "learning_rate": 1.7999999999999997e-05, "loss": 0.224, "step": 30 }, { "epoch": 0.3053435114503817, "grad_norm": 1.0794246196746826, "learning_rate": 2.3999999999999997e-05, "loss": 0.1785, "step": 40 }, { "epoch": 0.3816793893129771, "grad_norm": 0.9618160128593445, "learning_rate": 2.9999999999999997e-05, "loss": 0.1615, "step": 50 }, { "epoch": 0.4580152671755725, "grad_norm": 1.5802503824234009, "learning_rate": 3.5999999999999994e-05, "loss": 0.1565, "step": 60 }, { "epoch": 0.5343511450381679, "grad_norm": 0.961887776851654, "learning_rate": 4.2e-05, "loss": 0.1521, "step": 70 }, { "epoch": 0.6106870229007634, "grad_norm": 1.4655838012695312, "learning_rate": 4.7999999999999994e-05, "loss": 0.148, "step": 80 }, { "epoch": 0.6870229007633588, "grad_norm": 1.9632705450057983, "learning_rate": 5.399999999999999e-05, "loss": 0.1491, "step": 90 }, { "epoch": 0.7633587786259542, "grad_norm": 0.7829811573028564, "learning_rate": 5.9999999999999995e-05, "loss": 0.1505, "step": 100 }, { "epoch": 0.8396946564885496, "grad_norm": 0.8938019871711731, "learning_rate": 6.599999999999999e-05, "loss": 0.1472, "step": 110 }, { "epoch": 0.916030534351145, "grad_norm": 0.6061606407165527, "learning_rate": 7.199999999999999e-05, "loss": 0.1396, "step": 120 }, { "epoch": 0.9923664122137404, "grad_norm": 0.825457751750946, "learning_rate": 7.8e-05, "loss": 0.1336, "step": 130 }, { "epoch": 1.0687022900763359, "grad_norm": 1.2506942749023438, "learning_rate": 8.4e-05, "loss": 0.1333, "step": 140 }, { "epoch": 1.1450381679389312, "grad_norm": 0.48072028160095215, "learning_rate": 8.999999999999999e-05, "loss": 0.13, "step": 150 }, { "epoch": 1.2213740458015268, "grad_norm": 0.7247728705406189, "learning_rate": 9.599999999999999e-05, "loss": 0.1313, "step": 160 }, { "epoch": 1.297709923664122, "grad_norm": 0.6793253421783447, "learning_rate": 0.000102, "loss": 0.1314, "step": 170 }, { "epoch": 1.3740458015267176, "grad_norm": 0.5833969712257385, "learning_rate": 0.00010799999999999998, "loss": 0.1291, "step": 180 }, { "epoch": 1.450381679389313, "grad_norm": 0.876356303691864, "learning_rate": 0.00011399999999999999, "loss": 0.127, "step": 190 }, { "epoch": 1.5267175572519083, "grad_norm": 0.610722005367279, "learning_rate": 0.00011999999999999999, "loss": 0.125, "step": 200 }, { "epoch": 1.6030534351145038, "grad_norm": 1.0098367929458618, "learning_rate": 0.00012599999999999997, "loss": 0.1226, "step": 210 }, { "epoch": 1.6793893129770994, "grad_norm": 2.2321853637695312, "learning_rate": 0.00013199999999999998, "loss": 0.117, "step": 220 }, { "epoch": 1.7557251908396947, "grad_norm": 0.8984502553939819, "learning_rate": 0.000138, "loss": 0.1169, "step": 230 }, { "epoch": 1.83206106870229, "grad_norm": 1.1197595596313477, "learning_rate": 0.00014399999999999998, "loss": 0.1063, "step": 240 }, { "epoch": 1.9083969465648853, "grad_norm": 1.5231528282165527, "learning_rate": 0.00015, "loss": 0.1078, "step": 250 }, { "epoch": 1.984732824427481, "grad_norm": 0.7078485488891602, "learning_rate": 0.000156, "loss": 0.0998, "step": 260 }, { "epoch": 2.0610687022900764, "grad_norm": 0.3521653711795807, "learning_rate": 0.000162, "loss": 0.0912, "step": 270 }, { "epoch": 2.1374045801526718, "grad_norm": 0.4744652807712555, "learning_rate": 0.000168, "loss": 0.0894, "step": 280 }, { "epoch": 2.213740458015267, "grad_norm": 0.6162226796150208, "learning_rate": 0.00017399999999999997, "loss": 0.0914, "step": 290 }, { "epoch": 2.2900763358778624, "grad_norm": 1.1632736921310425, "learning_rate": 0.00017999999999999998, "loss": 0.0912, "step": 300 }, { "epoch": 2.366412213740458, "grad_norm": 0.5588046312332153, "learning_rate": 0.000186, "loss": 0.0879, "step": 310 }, { "epoch": 2.4427480916030535, "grad_norm": 1.1539413928985596, "learning_rate": 0.00019199999999999998, "loss": 0.084, "step": 320 }, { "epoch": 2.519083969465649, "grad_norm": 0.844003438949585, "learning_rate": 0.000198, "loss": 0.0797, "step": 330 }, { "epoch": 2.595419847328244, "grad_norm": 1.2032201290130615, "learning_rate": 0.000204, "loss": 0.079, "step": 340 }, { "epoch": 2.67175572519084, "grad_norm": 1.01964271068573, "learning_rate": 0.00020999999999999998, "loss": 0.0772, "step": 350 }, { "epoch": 2.7480916030534353, "grad_norm": 1.700699806213379, "learning_rate": 0.00021599999999999996, "loss": 0.0836, "step": 360 }, { "epoch": 2.8244274809160306, "grad_norm": 0.6162527799606323, "learning_rate": 0.00022199999999999998, "loss": 0.083, "step": 370 }, { "epoch": 2.900763358778626, "grad_norm": 0.8297500014305115, "learning_rate": 0.00022799999999999999, "loss": 0.0762, "step": 380 }, { "epoch": 2.9770992366412212, "grad_norm": 0.4928501844406128, "learning_rate": 0.000234, "loss": 0.0761, "step": 390 }, { "epoch": 3.053435114503817, "grad_norm": 0.7875099182128906, "learning_rate": 0.00023999999999999998, "loss": 0.0727, "step": 400 }, { "epoch": 3.1297709923664123, "grad_norm": 0.6808378100395203, "learning_rate": 0.00024599999999999996, "loss": 0.0724, "step": 410 }, { "epoch": 3.2061068702290076, "grad_norm": 0.572036862373352, "learning_rate": 0.00025199999999999995, "loss": 0.0698, "step": 420 }, { "epoch": 3.282442748091603, "grad_norm": 1.1315292119979858, "learning_rate": 0.000258, "loss": 0.0731, "step": 430 }, { "epoch": 3.3587786259541983, "grad_norm": 0.8437827825546265, "learning_rate": 0.00026399999999999997, "loss": 0.073, "step": 440 }, { "epoch": 3.435114503816794, "grad_norm": 1.0106201171875, "learning_rate": 0.00027, "loss": 0.072, "step": 450 }, { "epoch": 3.5114503816793894, "grad_norm": 0.7645736932754517, "learning_rate": 0.000276, "loss": 0.0658, "step": 460 }, { "epoch": 3.5877862595419847, "grad_norm": 0.7549853920936584, "learning_rate": 0.00028199999999999997, "loss": 0.0718, "step": 470 }, { "epoch": 3.66412213740458, "grad_norm": 0.8444331288337708, "learning_rate": 0.00028799999999999995, "loss": 0.068, "step": 480 }, { "epoch": 3.7404580152671754, "grad_norm": 0.6928085684776306, "learning_rate": 0.000294, "loss": 0.0658, "step": 490 }, { "epoch": 3.816793893129771, "grad_norm": 0.8107240200042725, "learning_rate": 0.0003, "loss": 0.0616, "step": 500 }, { "epoch": 3.8931297709923665, "grad_norm": 0.7328493595123291, "learning_rate": 0.0002999999149416535, "loss": 0.0642, "step": 510 }, { "epoch": 3.969465648854962, "grad_norm": 1.0192691087722778, "learning_rate": 0.00029999965976671057, "loss": 0.0663, "step": 520 }, { "epoch": 4.0458015267175576, "grad_norm": 0.562080442905426, "learning_rate": 0.00029999923447546054, "loss": 0.064, "step": 530 }, { "epoch": 4.122137404580153, "grad_norm": 0.6314799189567566, "learning_rate": 0.0002999986390683858, "loss": 0.0572, "step": 540 }, { "epoch": 4.198473282442748, "grad_norm": 0.8278489112854004, "learning_rate": 0.0002999978735461616, "loss": 0.061, "step": 550 }, { "epoch": 4.2748091603053435, "grad_norm": 0.6625831127166748, "learning_rate": 0.00029999693790965605, "loss": 0.0573, "step": 560 }, { "epoch": 4.351145038167939, "grad_norm": 1.1145844459533691, "learning_rate": 0.0002999958321599304, "loss": 0.0635, "step": 570 }, { "epoch": 4.427480916030534, "grad_norm": 1.1414052248001099, "learning_rate": 0.00029999455629823853, "loss": 0.0573, "step": 580 }, { "epoch": 4.5038167938931295, "grad_norm": 0.7339691519737244, "learning_rate": 0.0002999931103260275, "loss": 0.0567, "step": 590 }, { "epoch": 4.580152671755725, "grad_norm": 0.8177664875984192, "learning_rate": 0.0002999914942449372, "loss": 0.0618, "step": 600 }, { "epoch": 4.65648854961832, "grad_norm": 0.8224189877510071, "learning_rate": 0.00029998970805680044, "loss": 0.056, "step": 610 }, { "epoch": 4.732824427480916, "grad_norm": 0.8218141794204712, "learning_rate": 0.00029998775176364297, "loss": 0.0518, "step": 620 }, { "epoch": 4.809160305343512, "grad_norm": 0.6046106815338135, "learning_rate": 0.00029998562536768343, "loss": 0.0495, "step": 630 }, { "epoch": 4.885496183206107, "grad_norm": 0.7052053213119507, "learning_rate": 0.00029998332887133335, "loss": 0.0497, "step": 640 }, { "epoch": 4.961832061068702, "grad_norm": 0.4326968193054199, "learning_rate": 0.0002999808622771973, "loss": 0.0498, "step": 650 }, { "epoch": 5.038167938931298, "grad_norm": 0.8576260805130005, "learning_rate": 0.0002999782255880726, "loss": 0.0515, "step": 660 }, { "epoch": 5.114503816793893, "grad_norm": 0.5418573617935181, "learning_rate": 0.0002999754188069496, "loss": 0.0478, "step": 670 }, { "epoch": 5.190839694656488, "grad_norm": 0.5370392203330994, "learning_rate": 0.0002999724419370114, "loss": 0.0491, "step": 680 }, { "epoch": 5.267175572519084, "grad_norm": 0.41614219546318054, "learning_rate": 0.0002999692949816343, "loss": 0.0462, "step": 690 }, { "epoch": 5.34351145038168, "grad_norm": 0.42778339982032776, "learning_rate": 0.00029996597794438706, "loss": 0.0476, "step": 700 }, { "epoch": 5.419847328244275, "grad_norm": 0.6431729793548584, "learning_rate": 0.00029996249082903174, "loss": 0.0455, "step": 710 }, { "epoch": 5.4961832061068705, "grad_norm": 0.4207027554512024, "learning_rate": 0.0002999588336395231, "loss": 0.0426, "step": 720 }, { "epoch": 5.572519083969466, "grad_norm": 0.4943631589412689, "learning_rate": 0.0002999550063800087, "loss": 0.0428, "step": 730 }, { "epoch": 5.648854961832061, "grad_norm": 0.5425825119018555, "learning_rate": 0.0002999510090548291, "loss": 0.0437, "step": 740 }, { "epoch": 5.7251908396946565, "grad_norm": 0.438178151845932, "learning_rate": 0.00029994684166851786, "loss": 0.0415, "step": 750 }, { "epoch": 5.801526717557252, "grad_norm": 0.5176675319671631, "learning_rate": 0.0002999425042258011, "loss": 0.0452, "step": 760 }, { "epoch": 5.877862595419847, "grad_norm": 0.9786661863327026, "learning_rate": 0.00029993799673159804, "loss": 0.0468, "step": 770 }, { "epoch": 5.9541984732824424, "grad_norm": 1.1573694944381714, "learning_rate": 0.0002999333191910206, "loss": 0.0472, "step": 780 }, { "epoch": 6.030534351145038, "grad_norm": 0.4467450678348541, "learning_rate": 0.0002999284716093737, "loss": 0.0467, "step": 790 }, { "epoch": 6.106870229007634, "grad_norm": 0.6254957914352417, "learning_rate": 0.000299923453992155, "loss": 0.0446, "step": 800 }, { "epoch": 6.183206106870229, "grad_norm": 0.8170360922813416, "learning_rate": 0.0002999182663450551, "loss": 0.042, "step": 810 }, { "epoch": 6.259541984732825, "grad_norm": 0.6274629235267639, "learning_rate": 0.00029991290867395736, "loss": 0.0412, "step": 820 }, { "epoch": 6.33587786259542, "grad_norm": 0.5773537158966064, "learning_rate": 0.0002999073809849379, "loss": 0.0416, "step": 830 }, { "epoch": 6.412213740458015, "grad_norm": 1.238549828529358, "learning_rate": 0.00029990168328426574, "loss": 0.044, "step": 840 }, { "epoch": 6.488549618320611, "grad_norm": 0.4407813549041748, "learning_rate": 0.00029989581557840283, "loss": 0.0462, "step": 850 }, { "epoch": 6.564885496183206, "grad_norm": 0.6357528567314148, "learning_rate": 0.00029988977787400367, "loss": 0.0402, "step": 860 }, { "epoch": 6.641221374045801, "grad_norm": 0.6091104745864868, "learning_rate": 0.0002998835701779157, "loss": 0.0397, "step": 870 }, { "epoch": 6.717557251908397, "grad_norm": 0.4143543243408203, "learning_rate": 0.0002998771924971792, "loss": 0.042, "step": 880 }, { "epoch": 6.793893129770993, "grad_norm": 0.5831260085105896, "learning_rate": 0.00029987064483902716, "loss": 0.0383, "step": 890 }, { "epoch": 6.870229007633588, "grad_norm": 0.3744438886642456, "learning_rate": 0.00029986392721088537, "loss": 0.0402, "step": 900 }, { "epoch": 6.9465648854961835, "grad_norm": 0.5642991065979004, "learning_rate": 0.00029985703962037225, "loss": 0.0369, "step": 910 }, { "epoch": 7.022900763358779, "grad_norm": 0.452793151140213, "learning_rate": 0.0002998499820752992, "loss": 0.034, "step": 920 }, { "epoch": 7.099236641221374, "grad_norm": 0.3975425660610199, "learning_rate": 0.00029984275458367027, "loss": 0.0377, "step": 930 }, { "epoch": 7.175572519083969, "grad_norm": 0.43458667397499084, "learning_rate": 0.0002998353571536822, "loss": 0.0362, "step": 940 }, { "epoch": 7.251908396946565, "grad_norm": 0.32390084862709045, "learning_rate": 0.00029982778979372447, "loss": 0.0335, "step": 950 }, { "epoch": 7.32824427480916, "grad_norm": 0.3514920175075531, "learning_rate": 0.0002998200525123794, "loss": 0.0335, "step": 960 }, { "epoch": 7.404580152671755, "grad_norm": 0.25919872522354126, "learning_rate": 0.00029981214531842184, "loss": 0.0325, "step": 970 }, { "epoch": 7.480916030534351, "grad_norm": 0.5165295004844666, "learning_rate": 0.0002998040682208195, "loss": 0.0309, "step": 980 }, { "epoch": 7.557251908396947, "grad_norm": 0.4125653803348541, "learning_rate": 0.00029979582122873264, "loss": 0.0312, "step": 990 }, { "epoch": 7.633587786259542, "grad_norm": 0.5286191701889038, "learning_rate": 0.0002997874043515143, "loss": 0.0319, "step": 1000 }, { "epoch": 7.709923664122138, "grad_norm": 0.6198322772979736, "learning_rate": 0.00029977881759871014, "loss": 0.0347, "step": 1010 }, { "epoch": 7.786259541984733, "grad_norm": 0.9348865747451782, "learning_rate": 0.00029977006098005844, "loss": 0.0383, "step": 1020 }, { "epoch": 7.862595419847328, "grad_norm": 0.6678290367126465, "learning_rate": 0.00029976113450549036, "loss": 0.0428, "step": 1030 }, { "epoch": 7.938931297709924, "grad_norm": 0.506940484046936, "learning_rate": 0.00029975203818512935, "loss": 0.0423, "step": 1040 }, { "epoch": 8.01526717557252, "grad_norm": 0.4260866343975067, "learning_rate": 0.0002997427720292917, "loss": 0.0406, "step": 1050 }, { "epoch": 8.091603053435115, "grad_norm": 0.5327318906784058, "learning_rate": 0.00029973333604848624, "loss": 0.0364, "step": 1060 }, { "epoch": 8.16793893129771, "grad_norm": 0.7507275938987732, "learning_rate": 0.0002997237302534145, "loss": 0.0365, "step": 1070 }, { "epoch": 8.244274809160306, "grad_norm": 0.6194325089454651, "learning_rate": 0.0002997139546549704, "loss": 0.0377, "step": 1080 }, { "epoch": 8.320610687022901, "grad_norm": 0.4790773093700409, "learning_rate": 0.0002997040092642407, "loss": 0.0345, "step": 1090 }, { "epoch": 8.396946564885496, "grad_norm": 0.3612281084060669, "learning_rate": 0.0002996938940925045, "loss": 0.0335, "step": 1100 }, { "epoch": 8.473282442748092, "grad_norm": 0.3161233067512512, "learning_rate": 0.0002996836091512335, "loss": 0.0335, "step": 1110 }, { "epoch": 8.549618320610687, "grad_norm": 0.3041672706604004, "learning_rate": 0.00029967315445209193, "loss": 0.032, "step": 1120 }, { "epoch": 8.625954198473282, "grad_norm": 0.42357876896858215, "learning_rate": 0.00029966253000693674, "loss": 0.0313, "step": 1130 }, { "epoch": 8.702290076335878, "grad_norm": 0.6217378377914429, "learning_rate": 0.0002996517358278171, "loss": 0.0301, "step": 1140 }, { "epoch": 8.778625954198473, "grad_norm": 0.7184749245643616, "learning_rate": 0.00029964077192697487, "loss": 0.0305, "step": 1150 }, { "epoch": 8.854961832061068, "grad_norm": 0.31486737728118896, "learning_rate": 0.0002996296383168443, "loss": 0.0292, "step": 1160 }, { "epoch": 8.931297709923664, "grad_norm": 0.31588199734687805, "learning_rate": 0.00029961833501005223, "loss": 0.0294, "step": 1170 }, { "epoch": 9.007633587786259, "grad_norm": 0.339009165763855, "learning_rate": 0.0002996068620194177, "loss": 0.0301, "step": 1180 }, { "epoch": 9.083969465648854, "grad_norm": 0.35241368412971497, "learning_rate": 0.00029959521935795253, "loss": 0.0283, "step": 1190 }, { "epoch": 9.16030534351145, "grad_norm": 0.3480391800403595, "learning_rate": 0.0002995834070388607, "loss": 0.0278, "step": 1200 }, { "epoch": 9.236641221374045, "grad_norm": 0.2943292558193207, "learning_rate": 0.00029957142507553873, "loss": 0.0301, "step": 1210 }, { "epoch": 9.312977099236642, "grad_norm": 0.3550558388233185, "learning_rate": 0.0002995592734815755, "loss": 0.0285, "step": 1220 }, { "epoch": 9.389312977099237, "grad_norm": 0.3985001742839813, "learning_rate": 0.00029954695227075225, "loss": 0.0284, "step": 1230 }, { "epoch": 9.465648854961833, "grad_norm": 0.48568612337112427, "learning_rate": 0.00029953446145704256, "loss": 0.0291, "step": 1240 }, { "epoch": 9.541984732824428, "grad_norm": 0.3974258601665497, "learning_rate": 0.0002995218010546125, "loss": 0.0254, "step": 1250 }, { "epoch": 9.618320610687023, "grad_norm": 0.42430728673934937, "learning_rate": 0.0002995089710778203, "loss": 0.0274, "step": 1260 }, { "epoch": 9.694656488549619, "grad_norm": 0.4209881126880646, "learning_rate": 0.00029949597154121666, "loss": 0.0287, "step": 1270 }, { "epoch": 9.770992366412214, "grad_norm": 0.329190194606781, "learning_rate": 0.0002994828024595444, "loss": 0.0286, "step": 1280 }, { "epoch": 9.84732824427481, "grad_norm": 0.3262768089771271, "learning_rate": 0.00029946946384773877, "loss": 0.0288, "step": 1290 }, { "epoch": 9.923664122137405, "grad_norm": 0.3811452388763428, "learning_rate": 0.00029945595572092727, "loss": 0.0267, "step": 1300 }, { "epoch": 10.0, "grad_norm": 0.3417709171772003, "learning_rate": 0.0002994422780944296, "loss": 0.0251, "step": 1310 }, { "epoch": 10.076335877862595, "grad_norm": 0.26550206542015076, "learning_rate": 0.00029942843098375765, "loss": 0.025, "step": 1320 }, { "epoch": 10.15267175572519, "grad_norm": 0.31315726041793823, "learning_rate": 0.0002994144144046157, "loss": 0.0246, "step": 1330 }, { "epoch": 10.229007633587786, "grad_norm": 0.3284418284893036, "learning_rate": 0.00029940022837290003, "loss": 0.0253, "step": 1340 }, { "epoch": 10.305343511450381, "grad_norm": 0.4335983097553253, "learning_rate": 0.0002993858729046992, "loss": 0.0236, "step": 1350 }, { "epoch": 10.381679389312977, "grad_norm": 0.5847964286804199, "learning_rate": 0.0002993713480162939, "loss": 0.0255, "step": 1360 }, { "epoch": 10.458015267175572, "grad_norm": 0.46290674805641174, "learning_rate": 0.000299356653724157, "loss": 0.027, "step": 1370 }, { "epoch": 10.534351145038167, "grad_norm": 0.32573679089546204, "learning_rate": 0.00029934179004495345, "loss": 0.0256, "step": 1380 }, { "epoch": 10.610687022900763, "grad_norm": 0.4240122437477112, "learning_rate": 0.0002993267569955403, "loss": 0.028, "step": 1390 }, { "epoch": 10.68702290076336, "grad_norm": 0.4480084180831909, "learning_rate": 0.0002993115545929667, "loss": 0.0253, "step": 1400 }, { "epoch": 10.763358778625955, "grad_norm": 0.22174416482448578, "learning_rate": 0.0002992961828544738, "loss": 0.0237, "step": 1410 }, { "epoch": 10.83969465648855, "grad_norm": 0.25368282198905945, "learning_rate": 0.000299280641797495, "loss": 0.0267, "step": 1420 }, { "epoch": 10.916030534351146, "grad_norm": 0.21080680191516876, "learning_rate": 0.00029926493143965553, "loss": 0.0242, "step": 1430 }, { "epoch": 10.992366412213741, "grad_norm": 0.26196160912513733, "learning_rate": 0.00029924905179877265, "loss": 0.024, "step": 1440 }, { "epoch": 11.068702290076336, "grad_norm": 0.36000731587409973, "learning_rate": 0.0002992330028928557, "loss": 0.0227, "step": 1450 }, { "epoch": 11.145038167938932, "grad_norm": 0.24752654135227203, "learning_rate": 0.00029921678474010584, "loss": 0.0243, "step": 1460 }, { "epoch": 11.221374045801527, "grad_norm": 0.4830677807331085, "learning_rate": 0.00029920039735891634, "loss": 0.0255, "step": 1470 }, { "epoch": 11.297709923664122, "grad_norm": 0.2941085398197174, "learning_rate": 0.00029918384076787233, "loss": 0.0243, "step": 1480 }, { "epoch": 11.374045801526718, "grad_norm": 0.2960546314716339, "learning_rate": 0.00029916711498575077, "loss": 0.0231, "step": 1490 }, { "epoch": 11.450381679389313, "grad_norm": 0.4960053563117981, "learning_rate": 0.00029915022003152054, "loss": 0.0238, "step": 1500 }, { "epoch": 11.526717557251908, "grad_norm": 0.3530866205692291, "learning_rate": 0.0002991331559243425, "loss": 0.023, "step": 1510 }, { "epoch": 11.603053435114504, "grad_norm": 0.3872201442718506, "learning_rate": 0.00029911592268356906, "loss": 0.024, "step": 1520 }, { "epoch": 11.679389312977099, "grad_norm": 0.33585992455482483, "learning_rate": 0.0002990985203287448, "loss": 0.0235, "step": 1530 }, { "epoch": 11.755725190839694, "grad_norm": 0.2851446568965912, "learning_rate": 0.0002990809488796059, "loss": 0.0227, "step": 1540 }, { "epoch": 11.83206106870229, "grad_norm": 0.4621485471725464, "learning_rate": 0.0002990632083560803, "loss": 0.0232, "step": 1550 }, { "epoch": 11.908396946564885, "grad_norm": 0.4396305978298187, "learning_rate": 0.00029904529877828776, "loss": 0.023, "step": 1560 }, { "epoch": 11.98473282442748, "grad_norm": 0.37867969274520874, "learning_rate": 0.00029902722016653967, "loss": 0.0235, "step": 1570 }, { "epoch": 12.061068702290076, "grad_norm": 0.5124136209487915, "learning_rate": 0.00029900897254133926, "loss": 0.0233, "step": 1580 }, { "epoch": 12.137404580152673, "grad_norm": 0.47947779297828674, "learning_rate": 0.0002989905559233814, "loss": 0.0244, "step": 1590 }, { "epoch": 12.213740458015268, "grad_norm": 0.48070523142814636, "learning_rate": 0.0002989719703335525, "loss": 0.0222, "step": 1600 }, { "epoch": 12.290076335877863, "grad_norm": 0.3365572690963745, "learning_rate": 0.00029895321579293064, "loss": 0.0224, "step": 1610 }, { "epoch": 12.366412213740459, "grad_norm": 0.2986162304878235, "learning_rate": 0.0002989342923227857, "loss": 0.0232, "step": 1620 }, { "epoch": 12.442748091603054, "grad_norm": 0.2770279347896576, "learning_rate": 0.00029891519994457887, "loss": 0.0224, "step": 1630 }, { "epoch": 12.51908396946565, "grad_norm": 0.32686689496040344, "learning_rate": 0.00029889593867996316, "loss": 0.0222, "step": 1640 }, { "epoch": 12.595419847328245, "grad_norm": 0.39268556237220764, "learning_rate": 0.00029887650855078287, "loss": 0.0206, "step": 1650 }, { "epoch": 12.67175572519084, "grad_norm": 0.26340553164482117, "learning_rate": 0.0002988569095790741, "loss": 0.023, "step": 1660 }, { "epoch": 12.748091603053435, "grad_norm": 0.2524992525577545, "learning_rate": 0.000298837141787064, "loss": 0.0229, "step": 1670 }, { "epoch": 12.82442748091603, "grad_norm": 0.27329736948013306, "learning_rate": 0.0002988172051971717, "loss": 0.0218, "step": 1680 }, { "epoch": 12.900763358778626, "grad_norm": 0.38466787338256836, "learning_rate": 0.0002987970998320073, "loss": 0.0207, "step": 1690 }, { "epoch": 12.977099236641221, "grad_norm": 0.26403993368148804, "learning_rate": 0.00029877682571437264, "loss": 0.0209, "step": 1700 }, { "epoch": 13.053435114503817, "grad_norm": 0.24106334149837494, "learning_rate": 0.00029875638286726086, "loss": 0.0218, "step": 1710 }, { "epoch": 13.129770992366412, "grad_norm": 0.34761911630630493, "learning_rate": 0.0002987357713138564, "loss": 0.0217, "step": 1720 }, { "epoch": 13.206106870229007, "grad_norm": 0.36723530292510986, "learning_rate": 0.0002987149910775349, "loss": 0.0221, "step": 1730 }, { "epoch": 13.282442748091603, "grad_norm": 0.2713863253593445, "learning_rate": 0.0002986940421818637, "loss": 0.0213, "step": 1740 }, { "epoch": 13.358778625954198, "grad_norm": 0.3182651698589325, "learning_rate": 0.00029867292465060103, "loss": 0.021, "step": 1750 }, { "epoch": 13.435114503816793, "grad_norm": 0.29176831245422363, "learning_rate": 0.00029865163850769657, "loss": 0.0206, "step": 1760 }, { "epoch": 13.511450381679388, "grad_norm": 0.4259887933731079, "learning_rate": 0.00029863018377729116, "loss": 0.0221, "step": 1770 }, { "epoch": 13.587786259541986, "grad_norm": 0.5326780676841736, "learning_rate": 0.0002986085604837169, "loss": 0.021, "step": 1780 }, { "epoch": 13.664122137404581, "grad_norm": 0.2907402813434601, "learning_rate": 0.0002985867686514969, "loss": 0.0249, "step": 1790 }, { "epoch": 13.740458015267176, "grad_norm": 0.2631639540195465, "learning_rate": 0.00029856480830534564, "loss": 0.0244, "step": 1800 }, { "epoch": 13.816793893129772, "grad_norm": 0.4345681667327881, "learning_rate": 0.0002985426794701685, "loss": 0.0228, "step": 1810 }, { "epoch": 13.893129770992367, "grad_norm": 0.36692410707473755, "learning_rate": 0.0002985203821710621, "loss": 0.0228, "step": 1820 }, { "epoch": 13.969465648854962, "grad_norm": 0.5500999093055725, "learning_rate": 0.000298497916433314, "loss": 0.023, "step": 1830 }, { "epoch": 14.045801526717558, "grad_norm": 0.3562085032463074, "learning_rate": 0.000298475282282403, "loss": 0.0244, "step": 1840 }, { "epoch": 14.122137404580153, "grad_norm": 0.3216492831707001, "learning_rate": 0.00029845247974399854, "loss": 0.0241, "step": 1850 }, { "epoch": 14.198473282442748, "grad_norm": 0.3303261995315552, "learning_rate": 0.0002984295088439614, "loss": 0.0219, "step": 1860 }, { "epoch": 14.274809160305344, "grad_norm": 0.17324160039424896, "learning_rate": 0.00029840636960834304, "loss": 0.0224, "step": 1870 }, { "epoch": 14.351145038167939, "grad_norm": 0.3529289960861206, "learning_rate": 0.000298383062063386, "loss": 0.0211, "step": 1880 }, { "epoch": 14.427480916030534, "grad_norm": 0.36948323249816895, "learning_rate": 0.00029835958623552355, "loss": 0.023, "step": 1890 }, { "epoch": 14.50381679389313, "grad_norm": 0.5195873975753784, "learning_rate": 0.00029833594215137997, "loss": 0.0195, "step": 1900 }, { "epoch": 14.580152671755725, "grad_norm": 0.4122161269187927, "learning_rate": 0.00029831212983777016, "loss": 0.0209, "step": 1910 }, { "epoch": 14.65648854961832, "grad_norm": 0.41419556736946106, "learning_rate": 0.0002982881493217001, "loss": 0.023, "step": 1920 }, { "epoch": 14.732824427480915, "grad_norm": 0.703037440776825, "learning_rate": 0.00029826400063036627, "loss": 0.0227, "step": 1930 }, { "epoch": 14.80916030534351, "grad_norm": 0.6086626648902893, "learning_rate": 0.00029823968379115597, "loss": 0.025, "step": 1940 }, { "epoch": 14.885496183206106, "grad_norm": 0.4609583914279938, "learning_rate": 0.0002982151988316472, "loss": 0.0229, "step": 1950 }, { "epoch": 14.961832061068701, "grad_norm": 0.31456810235977173, "learning_rate": 0.00029819054577960867, "loss": 0.0206, "step": 1960 }, { "epoch": 15.038167938931299, "grad_norm": 0.2893737852573395, "learning_rate": 0.00029816572466299967, "loss": 0.0211, "step": 1970 }, { "epoch": 15.114503816793894, "grad_norm": 0.2887732982635498, "learning_rate": 0.00029814073550997006, "loss": 0.0191, "step": 1980 }, { "epoch": 15.19083969465649, "grad_norm": 0.2718079388141632, "learning_rate": 0.0002981155783488604, "loss": 0.0189, "step": 1990 }, { "epoch": 15.267175572519085, "grad_norm": 0.3429242670536041, "learning_rate": 0.0002980902532082017, "loss": 0.0181, "step": 2000 }, { "epoch": 15.34351145038168, "grad_norm": 0.22392407059669495, "learning_rate": 0.0002980647601167154, "loss": 0.018, "step": 2010 }, { "epoch": 15.419847328244275, "grad_norm": 0.23777836561203003, "learning_rate": 0.00029803909910331355, "loss": 0.0182, "step": 2020 }, { "epoch": 15.49618320610687, "grad_norm": 0.307410329580307, "learning_rate": 0.00029801327019709866, "loss": 0.0183, "step": 2030 }, { "epoch": 15.572519083969466, "grad_norm": 0.23634618520736694, "learning_rate": 0.0002979872734273635, "loss": 0.0192, "step": 2040 }, { "epoch": 15.648854961832061, "grad_norm": 0.1744905710220337, "learning_rate": 0.00029796110882359127, "loss": 0.0175, "step": 2050 }, { "epoch": 15.725190839694656, "grad_norm": 0.26253461837768555, "learning_rate": 0.0002979347764154557, "loss": 0.0174, "step": 2060 }, { "epoch": 15.801526717557252, "grad_norm": 0.3329990804195404, "learning_rate": 0.00029790827623282047, "loss": 0.0178, "step": 2070 }, { "epoch": 15.877862595419847, "grad_norm": 0.336376428604126, "learning_rate": 0.00029788160830573993, "loss": 0.0192, "step": 2080 }, { "epoch": 15.954198473282442, "grad_norm": 0.2983839511871338, "learning_rate": 0.00029785477266445836, "loss": 0.0181, "step": 2090 }, { "epoch": 16.03053435114504, "grad_norm": 0.296197772026062, "learning_rate": 0.0002978277693394104, "loss": 0.0169, "step": 2100 }, { "epoch": 16.106870229007633, "grad_norm": 0.2077913135290146, "learning_rate": 0.0002978005983612208, "loss": 0.0192, "step": 2110 }, { "epoch": 16.18320610687023, "grad_norm": 0.24259240925312042, "learning_rate": 0.00029777325976070447, "loss": 0.0179, "step": 2120 }, { "epoch": 16.259541984732824, "grad_norm": 0.2814135253429413, "learning_rate": 0.00029774575356886646, "loss": 0.0173, "step": 2130 }, { "epoch": 16.33587786259542, "grad_norm": 0.321178138256073, "learning_rate": 0.0002977180798169018, "loss": 0.0187, "step": 2140 }, { "epoch": 16.412213740458014, "grad_norm": 0.26111677289009094, "learning_rate": 0.0002976902385361957, "loss": 0.0186, "step": 2150 }, { "epoch": 16.48854961832061, "grad_norm": 0.32644879817962646, "learning_rate": 0.00029766222975832323, "loss": 0.0199, "step": 2160 }, { "epoch": 16.564885496183205, "grad_norm": 0.3182311952114105, "learning_rate": 0.0002976340535150494, "loss": 0.0184, "step": 2170 }, { "epoch": 16.641221374045802, "grad_norm": 0.46414169669151306, "learning_rate": 0.0002976057098383293, "loss": 0.0189, "step": 2180 }, { "epoch": 16.717557251908396, "grad_norm": 0.2640193998813629, "learning_rate": 0.0002975771987603076, "loss": 0.0194, "step": 2190 }, { "epoch": 16.793893129770993, "grad_norm": 0.29700085520744324, "learning_rate": 0.0002975485203133194, "loss": 0.0184, "step": 2200 }, { "epoch": 16.870229007633586, "grad_norm": 0.2691374123096466, "learning_rate": 0.0002975196745298889, "loss": 0.0171, "step": 2210 }, { "epoch": 16.946564885496183, "grad_norm": 0.2633708119392395, "learning_rate": 0.0002974906614427307, "loss": 0.0178, "step": 2220 }, { "epoch": 17.022900763358777, "grad_norm": 0.20800796151161194, "learning_rate": 0.00029746148108474865, "loss": 0.0196, "step": 2230 }, { "epoch": 17.099236641221374, "grad_norm": 0.3060513436794281, "learning_rate": 0.00029743213348903665, "loss": 0.0171, "step": 2240 }, { "epoch": 17.17557251908397, "grad_norm": 0.28063687682151794, "learning_rate": 0.0002974026186888781, "loss": 0.0182, "step": 2250 }, { "epoch": 17.251908396946565, "grad_norm": 0.24804764986038208, "learning_rate": 0.00029737293671774613, "loss": 0.0192, "step": 2260 }, { "epoch": 17.328244274809162, "grad_norm": 0.3719424605369568, "learning_rate": 0.0002973430876093033, "loss": 0.018, "step": 2270 }, { "epoch": 17.404580152671755, "grad_norm": 0.2215825915336609, "learning_rate": 0.00029731307139740183, "loss": 0.0168, "step": 2280 }, { "epoch": 17.480916030534353, "grad_norm": 0.22199778258800507, "learning_rate": 0.00029728288811608355, "loss": 0.0166, "step": 2290 }, { "epoch": 17.557251908396946, "grad_norm": 0.22484183311462402, "learning_rate": 0.0002972525377995796, "loss": 0.016, "step": 2300 }, { "epoch": 17.633587786259543, "grad_norm": 0.2126537412405014, "learning_rate": 0.0002972220204823105, "loss": 0.0179, "step": 2310 }, { "epoch": 17.709923664122137, "grad_norm": 0.3392879366874695, "learning_rate": 0.0002971913361988865, "loss": 0.0174, "step": 2320 }, { "epoch": 17.786259541984734, "grad_norm": 0.3461496829986572, "learning_rate": 0.00029716048498410684, "loss": 0.0175, "step": 2330 }, { "epoch": 17.862595419847327, "grad_norm": 0.4419268071651459, "learning_rate": 0.00029712946687296025, "loss": 0.0172, "step": 2340 }, { "epoch": 17.938931297709924, "grad_norm": 0.262058287858963, "learning_rate": 0.0002970982819006248, "loss": 0.0175, "step": 2350 }, { "epoch": 18.015267175572518, "grad_norm": 0.48134708404541016, "learning_rate": 0.00029706693010246756, "loss": 0.0185, "step": 2360 }, { "epoch": 18.091603053435115, "grad_norm": 0.31412458419799805, "learning_rate": 0.0002970354115140452, "loss": 0.0189, "step": 2370 }, { "epoch": 18.16793893129771, "grad_norm": 0.22777320444583893, "learning_rate": 0.00029700372617110307, "loss": 0.0174, "step": 2380 }, { "epoch": 18.244274809160306, "grad_norm": 0.31756728887557983, "learning_rate": 0.00029697187410957605, "loss": 0.0207, "step": 2390 }, { "epoch": 18.3206106870229, "grad_norm": 0.36132949590682983, "learning_rate": 0.00029693985536558774, "loss": 0.018, "step": 2400 }, { "epoch": 18.396946564885496, "grad_norm": 0.3123205304145813, "learning_rate": 0.00029690766997545116, "loss": 0.0191, "step": 2410 }, { "epoch": 18.47328244274809, "grad_norm": 0.23600128293037415, "learning_rate": 0.000296875317975668, "loss": 0.0171, "step": 2420 }, { "epoch": 18.549618320610687, "grad_norm": 0.3467651605606079, "learning_rate": 0.00029684279940292907, "loss": 0.018, "step": 2430 }, { "epoch": 18.625954198473284, "grad_norm": 0.3375345766544342, "learning_rate": 0.00029681011429411394, "loss": 0.0174, "step": 2440 }, { "epoch": 18.702290076335878, "grad_norm": 0.2862045466899872, "learning_rate": 0.00029677726268629136, "loss": 0.018, "step": 2450 }, { "epoch": 18.778625954198475, "grad_norm": 0.23653537034988403, "learning_rate": 0.0002967442446167186, "loss": 0.018, "step": 2460 }, { "epoch": 18.85496183206107, "grad_norm": 0.2839006781578064, "learning_rate": 0.00029671106012284183, "loss": 0.0187, "step": 2470 }, { "epoch": 18.931297709923665, "grad_norm": 0.19230085611343384, "learning_rate": 0.00029667770924229593, "loss": 0.0177, "step": 2480 }, { "epoch": 19.00763358778626, "grad_norm": 0.2190081775188446, "learning_rate": 0.00029664419201290453, "loss": 0.0156, "step": 2490 }, { "epoch": 19.083969465648856, "grad_norm": 0.1796020120382309, "learning_rate": 0.00029661050847267997, "loss": 0.0164, "step": 2500 }, { "epoch": 19.16030534351145, "grad_norm": 0.32525503635406494, "learning_rate": 0.00029657665865982303, "loss": 0.0157, "step": 2510 }, { "epoch": 19.236641221374047, "grad_norm": 0.18527977168560028, "learning_rate": 0.00029654264261272324, "loss": 0.0156, "step": 2520 }, { "epoch": 19.31297709923664, "grad_norm": 0.30235493183135986, "learning_rate": 0.0002965084603699585, "loss": 0.016, "step": 2530 }, { "epoch": 19.389312977099237, "grad_norm": 0.26955267786979675, "learning_rate": 0.00029647411197029536, "loss": 0.0168, "step": 2540 }, { "epoch": 19.46564885496183, "grad_norm": 0.2033633291721344, "learning_rate": 0.00029643959745268873, "loss": 0.0156, "step": 2550 }, { "epoch": 19.541984732824428, "grad_norm": 0.3382219970226288, "learning_rate": 0.0002964049168562818, "loss": 0.0163, "step": 2560 }, { "epoch": 19.61832061068702, "grad_norm": 0.27191078662872314, "learning_rate": 0.0002963700702204064, "loss": 0.0178, "step": 2570 }, { "epoch": 19.69465648854962, "grad_norm": 0.2028510719537735, "learning_rate": 0.00029633505758458234, "loss": 0.0182, "step": 2580 }, { "epoch": 19.770992366412212, "grad_norm": 0.27970293164253235, "learning_rate": 0.0002962998789885179, "loss": 0.0164, "step": 2590 }, { "epoch": 19.84732824427481, "grad_norm": 0.2538454830646515, "learning_rate": 0.0002962645344721095, "loss": 0.0167, "step": 2600 }, { "epoch": 19.923664122137403, "grad_norm": 0.37499645352363586, "learning_rate": 0.00029622902407544185, "loss": 0.0154, "step": 2610 }, { "epoch": 20.0, "grad_norm": 0.24018356204032898, "learning_rate": 0.0002961933478387876, "loss": 0.0168, "step": 2620 }, { "epoch": 20.076335877862597, "grad_norm": 0.22372686862945557, "learning_rate": 0.00029615750580260755, "loss": 0.0156, "step": 2630 }, { "epoch": 20.15267175572519, "grad_norm": 0.30653923749923706, "learning_rate": 0.00029612149800755066, "loss": 0.0161, "step": 2640 }, { "epoch": 20.229007633587788, "grad_norm": 0.30025187134742737, "learning_rate": 0.00029608532449445374, "loss": 0.0152, "step": 2650 }, { "epoch": 20.30534351145038, "grad_norm": 0.2449946105480194, "learning_rate": 0.0002960489853043415, "loss": 0.0164, "step": 2660 }, { "epoch": 20.38167938931298, "grad_norm": 0.23487500846385956, "learning_rate": 0.0002960124804784267, "loss": 0.0152, "step": 2670 }, { "epoch": 20.458015267175572, "grad_norm": 0.23396098613739014, "learning_rate": 0.00029597581005810986, "loss": 0.0145, "step": 2680 }, { "epoch": 20.53435114503817, "grad_norm": 0.25128626823425293, "learning_rate": 0.00029593897408497933, "loss": 0.0148, "step": 2690 }, { "epoch": 20.610687022900763, "grad_norm": 0.21609310805797577, "learning_rate": 0.0002959019726008112, "loss": 0.0155, "step": 2700 }, { "epoch": 20.68702290076336, "grad_norm": 0.26793137192726135, "learning_rate": 0.00029586480564756923, "loss": 0.0156, "step": 2710 }, { "epoch": 20.763358778625953, "grad_norm": 0.2092086374759674, "learning_rate": 0.00029582747326740495, "loss": 0.0141, "step": 2720 }, { "epoch": 20.83969465648855, "grad_norm": 0.23329131305217743, "learning_rate": 0.00029578997550265735, "loss": 0.0151, "step": 2730 }, { "epoch": 20.916030534351144, "grad_norm": 0.1923300325870514, "learning_rate": 0.0002957523123958532, "loss": 0.0154, "step": 2740 }, { "epoch": 20.99236641221374, "grad_norm": 0.2909185290336609, "learning_rate": 0.0002957144839897065, "loss": 0.0177, "step": 2750 }, { "epoch": 21.068702290076335, "grad_norm": 0.26453688740730286, "learning_rate": 0.00029567649032711895, "loss": 0.0163, "step": 2760 }, { "epoch": 21.14503816793893, "grad_norm": 0.3507690727710724, "learning_rate": 0.0002956383314511795, "loss": 0.0181, "step": 2770 }, { "epoch": 21.221374045801525, "grad_norm": 0.18822893500328064, "learning_rate": 0.0002956000074051648, "loss": 0.0186, "step": 2780 }, { "epoch": 21.297709923664122, "grad_norm": 0.40644538402557373, "learning_rate": 0.00029556151823253833, "loss": 0.0199, "step": 2790 }, { "epoch": 21.374045801526716, "grad_norm": 0.3049580454826355, "learning_rate": 0.0002955228639769513, "loss": 0.0183, "step": 2800 }, { "epoch": 21.450381679389313, "grad_norm": 0.3607460558414459, "learning_rate": 0.00029548404468224173, "loss": 0.0174, "step": 2810 }, { "epoch": 21.52671755725191, "grad_norm": 0.30051371455192566, "learning_rate": 0.0002954450603924351, "loss": 0.0151, "step": 2820 }, { "epoch": 21.603053435114504, "grad_norm": 0.26490694284439087, "learning_rate": 0.00029540591115174406, "loss": 0.0156, "step": 2830 }, { "epoch": 21.6793893129771, "grad_norm": 0.2580675482749939, "learning_rate": 0.000295366597004568, "loss": 0.017, "step": 2840 }, { "epoch": 21.755725190839694, "grad_norm": 0.3435441851615906, "learning_rate": 0.00029532711799549365, "loss": 0.017, "step": 2850 }, { "epoch": 21.83206106870229, "grad_norm": 0.4150358736515045, "learning_rate": 0.00029528747416929463, "loss": 0.0174, "step": 2860 }, { "epoch": 21.908396946564885, "grad_norm": 0.25315651297569275, "learning_rate": 0.00029524766557093143, "loss": 0.017, "step": 2870 }, { "epoch": 21.984732824427482, "grad_norm": 0.31266653537750244, "learning_rate": 0.0002952076922455514, "loss": 0.0157, "step": 2880 }, { "epoch": 22.061068702290076, "grad_norm": 0.22555911540985107, "learning_rate": 0.00029516755423848864, "loss": 0.0144, "step": 2890 }, { "epoch": 22.137404580152673, "grad_norm": 0.25117605924606323, "learning_rate": 0.0002951272515952643, "loss": 0.0152, "step": 2900 }, { "epoch": 22.213740458015266, "grad_norm": 0.2168475240468979, "learning_rate": 0.00029508678436158606, "loss": 0.0138, "step": 2910 }, { "epoch": 22.290076335877863, "grad_norm": 0.23664060235023499, "learning_rate": 0.0002950461525833481, "loss": 0.0161, "step": 2920 }, { "epoch": 22.366412213740457, "grad_norm": 0.27318814396858215, "learning_rate": 0.0002950053563066315, "loss": 0.0161, "step": 2930 }, { "epoch": 22.442748091603054, "grad_norm": 0.2026703953742981, "learning_rate": 0.0002949643955777038, "loss": 0.0162, "step": 2940 }, { "epoch": 22.519083969465647, "grad_norm": 0.19126379489898682, "learning_rate": 0.0002949232704430189, "loss": 0.0167, "step": 2950 }, { "epoch": 22.595419847328245, "grad_norm": 0.1500840187072754, "learning_rate": 0.00029488198094921735, "loss": 0.0153, "step": 2960 }, { "epoch": 22.671755725190838, "grad_norm": 0.2941581606864929, "learning_rate": 0.00029484052714312607, "loss": 0.0156, "step": 2970 }, { "epoch": 22.748091603053435, "grad_norm": 0.25173816084861755, "learning_rate": 0.0002947989090717583, "loss": 0.016, "step": 2980 }, { "epoch": 22.82442748091603, "grad_norm": 0.24619977176189423, "learning_rate": 0.00029475712678231347, "loss": 0.0154, "step": 2990 }, { "epoch": 22.900763358778626, "grad_norm": 0.3533461093902588, "learning_rate": 0.00029471518032217736, "loss": 0.016, "step": 3000 }, { "epoch": 22.977099236641223, "grad_norm": 0.33427536487579346, "learning_rate": 0.000294673069738922, "loss": 0.0163, "step": 3010 }, { "epoch": 23.053435114503817, "grad_norm": 0.3168981075286865, "learning_rate": 0.0002946307950803054, "loss": 0.0162, "step": 3020 }, { "epoch": 23.129770992366414, "grad_norm": 0.21112599968910217, "learning_rate": 0.0002945883563942719, "loss": 0.0156, "step": 3030 }, { "epoch": 23.206106870229007, "grad_norm": 0.29890426993370056, "learning_rate": 0.0002945457537289514, "loss": 0.0155, "step": 3040 }, { "epoch": 23.282442748091604, "grad_norm": 0.33570462465286255, "learning_rate": 0.00029450298713266026, "loss": 0.016, "step": 3050 }, { "epoch": 23.358778625954198, "grad_norm": 0.31405994296073914, "learning_rate": 0.00029446005665390056, "loss": 0.0165, "step": 3060 }, { "epoch": 23.435114503816795, "grad_norm": 0.36926284432411194, "learning_rate": 0.00029441696234136017, "loss": 0.0177, "step": 3070 }, { "epoch": 23.51145038167939, "grad_norm": 0.16154274344444275, "learning_rate": 0.00029437370424391287, "loss": 0.0159, "step": 3080 }, { "epoch": 23.587786259541986, "grad_norm": 0.2112104743719101, "learning_rate": 0.00029433028241061815, "loss": 0.0159, "step": 3090 }, { "epoch": 23.66412213740458, "grad_norm": 0.22717849910259247, "learning_rate": 0.00029428669689072123, "loss": 0.015, "step": 3100 }, { "epoch": 23.740458015267176, "grad_norm": 0.24502483010292053, "learning_rate": 0.0002942429477336529, "loss": 0.0146, "step": 3110 }, { "epoch": 23.81679389312977, "grad_norm": 0.27032968401908875, "learning_rate": 0.0002941990349890296, "loss": 0.0153, "step": 3120 }, { "epoch": 23.893129770992367, "grad_norm": 0.20628239214420319, "learning_rate": 0.00029415495870665327, "loss": 0.014, "step": 3130 }, { "epoch": 23.96946564885496, "grad_norm": 0.18944168090820312, "learning_rate": 0.0002941107189365112, "loss": 0.0146, "step": 3140 }, { "epoch": 24.045801526717558, "grad_norm": 0.2052525132894516, "learning_rate": 0.00029406631572877646, "loss": 0.0138, "step": 3150 }, { "epoch": 24.12213740458015, "grad_norm": 0.23070399463176727, "learning_rate": 0.000294021749133807, "loss": 0.013, "step": 3160 }, { "epoch": 24.198473282442748, "grad_norm": 0.23603607714176178, "learning_rate": 0.00029397701920214645, "loss": 0.0129, "step": 3170 }, { "epoch": 24.274809160305345, "grad_norm": 0.29337644577026367, "learning_rate": 0.0002939321259845234, "loss": 0.013, "step": 3180 }, { "epoch": 24.35114503816794, "grad_norm": 0.25555551052093506, "learning_rate": 0.00029388706953185185, "loss": 0.0138, "step": 3190 }, { "epoch": 24.427480916030536, "grad_norm": 0.1829186975955963, "learning_rate": 0.00029384184989523083, "loss": 0.0134, "step": 3200 }, { "epoch": 24.50381679389313, "grad_norm": 0.313329815864563, "learning_rate": 0.00029379646712594435, "loss": 0.0135, "step": 3210 }, { "epoch": 24.580152671755727, "grad_norm": 0.16749639809131622, "learning_rate": 0.00029375092127546164, "loss": 0.0137, "step": 3220 }, { "epoch": 24.65648854961832, "grad_norm": 0.17289191484451294, "learning_rate": 0.0002937052123954367, "loss": 0.0138, "step": 3230 }, { "epoch": 24.732824427480917, "grad_norm": 0.2209453284740448, "learning_rate": 0.0002936593405377085, "loss": 0.0128, "step": 3240 }, { "epoch": 24.80916030534351, "grad_norm": 0.22406715154647827, "learning_rate": 0.00029361330575430075, "loss": 0.0146, "step": 3250 }, { "epoch": 24.885496183206108, "grad_norm": 0.19136761128902435, "learning_rate": 0.0002935671080974222, "loss": 0.0145, "step": 3260 }, { "epoch": 24.9618320610687, "grad_norm": 0.22183312475681305, "learning_rate": 0.000293520747619466, "loss": 0.0153, "step": 3270 }, { "epoch": 25.0381679389313, "grad_norm": 0.25630468130111694, "learning_rate": 0.0002934742243730101, "loss": 0.0149, "step": 3280 }, { "epoch": 25.114503816793892, "grad_norm": 0.30678021907806396, "learning_rate": 0.000293427538410817, "loss": 0.0156, "step": 3290 }, { "epoch": 25.19083969465649, "grad_norm": 0.21654050052165985, "learning_rate": 0.0002933806897858339, "loss": 0.0147, "step": 3300 }, { "epoch": 25.267175572519083, "grad_norm": 0.30847570300102234, "learning_rate": 0.0002933336785511923, "loss": 0.0147, "step": 3310 }, { "epoch": 25.34351145038168, "grad_norm": 0.21043366193771362, "learning_rate": 0.0002932865047602081, "loss": 0.0138, "step": 3320 }, { "epoch": 25.419847328244273, "grad_norm": 0.3074447512626648, "learning_rate": 0.0002932391684663818, "loss": 0.0156, "step": 3330 }, { "epoch": 25.49618320610687, "grad_norm": 0.32755857706069946, "learning_rate": 0.0002931916697233978, "loss": 0.0149, "step": 3340 }, { "epoch": 25.572519083969464, "grad_norm": 0.3372158110141754, "learning_rate": 0.0002931440085851251, "loss": 0.0146, "step": 3350 }, { "epoch": 25.64885496183206, "grad_norm": 0.2669939398765564, "learning_rate": 0.0002930961851056167, "loss": 0.0142, "step": 3360 }, { "epoch": 25.725190839694655, "grad_norm": 0.2929087281227112, "learning_rate": 0.0002930481993391098, "loss": 0.0147, "step": 3370 }, { "epoch": 25.801526717557252, "grad_norm": 0.22061574459075928, "learning_rate": 0.00029300005134002557, "loss": 0.0131, "step": 3380 }, { "epoch": 25.87786259541985, "grad_norm": 0.3028363585472107, "learning_rate": 0.0002929517411629691, "loss": 0.0134, "step": 3390 }, { "epoch": 25.954198473282442, "grad_norm": 0.2300546020269394, "learning_rate": 0.00029290326886272967, "loss": 0.0137, "step": 3400 }, { "epoch": 26.03053435114504, "grad_norm": 0.21855516731739044, "learning_rate": 0.0002928546344942802, "loss": 0.0141, "step": 3410 }, { "epoch": 26.106870229007633, "grad_norm": 0.18946591019630432, "learning_rate": 0.00029280583811277735, "loss": 0.0139, "step": 3420 }, { "epoch": 26.18320610687023, "grad_norm": 0.19487009942531586, "learning_rate": 0.00029275687977356186, "loss": 0.0137, "step": 3430 }, { "epoch": 26.259541984732824, "grad_norm": 0.20209208130836487, "learning_rate": 0.0002927077595321578, "loss": 0.0132, "step": 3440 }, { "epoch": 26.33587786259542, "grad_norm": 0.2209881693124771, "learning_rate": 0.00029265847744427303, "loss": 0.0137, "step": 3450 }, { "epoch": 26.412213740458014, "grad_norm": 0.21442578732967377, "learning_rate": 0.0002926090335657989, "loss": 0.0136, "step": 3460 }, { "epoch": 26.48854961832061, "grad_norm": 0.22246500849723816, "learning_rate": 0.0002925594279528103, "loss": 0.0135, "step": 3470 }, { "epoch": 26.564885496183205, "grad_norm": 0.22301892936229706, "learning_rate": 0.0002925096606615655, "loss": 0.0133, "step": 3480 }, { "epoch": 26.641221374045802, "grad_norm": 0.21544921398162842, "learning_rate": 0.0002924597317485061, "loss": 0.0128, "step": 3490 }, { "epoch": 26.717557251908396, "grad_norm": 0.23722727596759796, "learning_rate": 0.00029240964127025715, "loss": 0.0147, "step": 3500 }, { "epoch": 26.793893129770993, "grad_norm": 0.27501773834228516, "learning_rate": 0.0002923593892836268, "loss": 0.0132, "step": 3510 }, { "epoch": 26.870229007633586, "grad_norm": 0.25271084904670715, "learning_rate": 0.0002923089758456063, "loss": 0.0123, "step": 3520 }, { "epoch": 26.946564885496183, "grad_norm": 0.21191899478435516, "learning_rate": 0.0002922584010133702, "loss": 0.0133, "step": 3530 }, { "epoch": 27.022900763358777, "grad_norm": 0.22761109471321106, "learning_rate": 0.00029220766484427594, "loss": 0.0136, "step": 3540 }, { "epoch": 27.099236641221374, "grad_norm": 0.21930576860904694, "learning_rate": 0.000292156767395864, "loss": 0.0141, "step": 3550 }, { "epoch": 27.17557251908397, "grad_norm": 0.3116973340511322, "learning_rate": 0.0002921057087258577, "loss": 0.0138, "step": 3560 }, { "epoch": 27.251908396946565, "grad_norm": 0.28197211027145386, "learning_rate": 0.00029205448889216333, "loss": 0.0138, "step": 3570 }, { "epoch": 27.328244274809162, "grad_norm": 0.20362894237041473, "learning_rate": 0.00029200310795286983, "loss": 0.0125, "step": 3580 }, { "epoch": 27.404580152671755, "grad_norm": 0.23354914784431458, "learning_rate": 0.00029195156596624895, "loss": 0.0131, "step": 3590 }, { "epoch": 27.480916030534353, "grad_norm": 0.2576790452003479, "learning_rate": 0.000291899862990755, "loss": 0.0137, "step": 3600 }, { "epoch": 27.557251908396946, "grad_norm": 0.15316276252269745, "learning_rate": 0.00029184799908502497, "loss": 0.0125, "step": 3610 }, { "epoch": 27.633587786259543, "grad_norm": 0.16182218492031097, "learning_rate": 0.0002917959743078782, "loss": 0.0114, "step": 3620 }, { "epoch": 27.709923664122137, "grad_norm": 0.20993201434612274, "learning_rate": 0.00029174378871831664, "loss": 0.0128, "step": 3630 }, { "epoch": 27.786259541984734, "grad_norm": 0.2623291015625, "learning_rate": 0.0002916914423755245, "loss": 0.013, "step": 3640 }, { "epoch": 27.862595419847327, "grad_norm": 0.2712068557739258, "learning_rate": 0.00029163893533886847, "loss": 0.0131, "step": 3650 }, { "epoch": 27.938931297709924, "grad_norm": 0.1644788682460785, "learning_rate": 0.00029158626766789727, "loss": 0.0133, "step": 3660 }, { "epoch": 28.015267175572518, "grad_norm": 0.2264963686466217, "learning_rate": 0.0002915334394223419, "loss": 0.0131, "step": 3670 }, { "epoch": 28.091603053435115, "grad_norm": 0.24077154695987701, "learning_rate": 0.00029148045066211553, "loss": 0.0128, "step": 3680 }, { "epoch": 28.16793893129771, "grad_norm": 0.1684757024049759, "learning_rate": 0.0002914273014473133, "loss": 0.0125, "step": 3690 }, { "epoch": 28.244274809160306, "grad_norm": 0.23623353242874146, "learning_rate": 0.0002913739918382123, "loss": 0.0124, "step": 3700 }, { "epoch": 28.3206106870229, "grad_norm": 0.32589009404182434, "learning_rate": 0.0002913205218952716, "loss": 0.0128, "step": 3710 }, { "epoch": 28.396946564885496, "grad_norm": 0.17773936688899994, "learning_rate": 0.00029126689167913205, "loss": 0.0137, "step": 3720 }, { "epoch": 28.47328244274809, "grad_norm": 0.25648069381713867, "learning_rate": 0.00029121310125061625, "loss": 0.0149, "step": 3730 }, { "epoch": 28.549618320610687, "grad_norm": 0.21268154680728912, "learning_rate": 0.0002911591506707286, "loss": 0.014, "step": 3740 }, { "epoch": 28.625954198473284, "grad_norm": 0.2599167823791504, "learning_rate": 0.00029110504000065497, "loss": 0.0131, "step": 3750 }, { "epoch": 28.702290076335878, "grad_norm": 0.19912320375442505, "learning_rate": 0.00029105076930176297, "loss": 0.013, "step": 3760 }, { "epoch": 28.778625954198475, "grad_norm": 0.131193608045578, "learning_rate": 0.00029099633863560157, "loss": 0.0109, "step": 3770 }, { "epoch": 28.85496183206107, "grad_norm": 0.17518292367458344, "learning_rate": 0.00029094174806390115, "loss": 0.0123, "step": 3780 }, { "epoch": 28.931297709923665, "grad_norm": 0.22402244806289673, "learning_rate": 0.00029088699764857364, "loss": 0.0114, "step": 3790 }, { "epoch": 29.00763358778626, "grad_norm": 0.12520021200180054, "learning_rate": 0.0002908320874517119, "loss": 0.0124, "step": 3800 }, { "epoch": 29.083969465648856, "grad_norm": 0.2125815749168396, "learning_rate": 0.0002907770175355905, "loss": 0.0125, "step": 3810 }, { "epoch": 29.16030534351145, "grad_norm": 0.23710201680660248, "learning_rate": 0.00029072178796266454, "loss": 0.0114, "step": 3820 }, { "epoch": 29.236641221374047, "grad_norm": 0.17837679386138916, "learning_rate": 0.00029066639879557065, "loss": 0.0117, "step": 3830 }, { "epoch": 29.31297709923664, "grad_norm": 0.1656421571969986, "learning_rate": 0.0002906108500971263, "loss": 0.0133, "step": 3840 }, { "epoch": 29.389312977099237, "grad_norm": 0.31068938970565796, "learning_rate": 0.00029055514193032986, "loss": 0.0113, "step": 3850 }, { "epoch": 29.46564885496183, "grad_norm": 0.15090344846248627, "learning_rate": 0.00029049927435836074, "loss": 0.0132, "step": 3860 }, { "epoch": 29.541984732824428, "grad_norm": 0.21045300364494324, "learning_rate": 0.00029044324744457875, "loss": 0.0114, "step": 3870 }, { "epoch": 29.61832061068702, "grad_norm": 0.1623835563659668, "learning_rate": 0.0002903870612525248, "loss": 0.0117, "step": 3880 }, { "epoch": 29.69465648854962, "grad_norm": 0.16781829297542572, "learning_rate": 0.0002903307158459202, "loss": 0.0123, "step": 3890 }, { "epoch": 29.770992366412212, "grad_norm": 0.16055527329444885, "learning_rate": 0.000290274211288667, "loss": 0.0116, "step": 3900 }, { "epoch": 29.84732824427481, "grad_norm": 0.17405042052268982, "learning_rate": 0.0002902175476448476, "loss": 0.0124, "step": 3910 }, { "epoch": 29.923664122137403, "grad_norm": 0.23891368508338928, "learning_rate": 0.0002901607249787249, "loss": 0.0141, "step": 3920 }, { "epoch": 30.0, "grad_norm": 0.18947049975395203, "learning_rate": 0.0002901037433547421, "loss": 0.0135, "step": 3930 }, { "epoch": 30.076335877862597, "grad_norm": 0.2228708267211914, "learning_rate": 0.00029004660283752276, "loss": 0.0134, "step": 3940 }, { "epoch": 30.15267175572519, "grad_norm": 0.2544506788253784, "learning_rate": 0.0002899893034918705, "loss": 0.013, "step": 3950 }, { "epoch": 30.229007633587788, "grad_norm": 0.20885154604911804, "learning_rate": 0.0002899318453827692, "loss": 0.0128, "step": 3960 }, { "epoch": 30.30534351145038, "grad_norm": 0.26366451382637024, "learning_rate": 0.00028987422857538277, "loss": 0.0128, "step": 3970 }, { "epoch": 30.38167938931298, "grad_norm": 0.22556272149085999, "learning_rate": 0.00028981645313505506, "loss": 0.0132, "step": 3980 }, { "epoch": 30.458015267175572, "grad_norm": 0.1817709058523178, "learning_rate": 0.0002897585191273098, "loss": 0.0124, "step": 3990 }, { "epoch": 30.53435114503817, "grad_norm": 0.23004992306232452, "learning_rate": 0.00028970042661785074, "loss": 0.0128, "step": 4000 }, { "epoch": 30.610687022900763, "grad_norm": 0.24111934006214142, "learning_rate": 0.00028964217567256114, "loss": 0.0134, "step": 4010 }, { "epoch": 30.68702290076336, "grad_norm": 0.25717976689338684, "learning_rate": 0.0002895837663575041, "loss": 0.0144, "step": 4020 }, { "epoch": 30.763358778625953, "grad_norm": 0.2161615490913391, "learning_rate": 0.00028952519873892213, "loss": 0.013, "step": 4030 }, { "epoch": 30.83969465648855, "grad_norm": 0.21292878687381744, "learning_rate": 0.00028946647288323766, "loss": 0.0142, "step": 4040 }, { "epoch": 30.916030534351144, "grad_norm": 0.28133440017700195, "learning_rate": 0.00028940758885705225, "loss": 0.0155, "step": 4050 }, { "epoch": 30.99236641221374, "grad_norm": 0.31102636456489563, "learning_rate": 0.00028934854672714693, "loss": 0.0142, "step": 4060 }, { "epoch": 31.068702290076335, "grad_norm": 0.29843416810035706, "learning_rate": 0.000289289346560482, "loss": 0.0164, "step": 4070 }, { "epoch": 31.14503816793893, "grad_norm": 0.1796748787164688, "learning_rate": 0.00028922998842419715, "loss": 0.0139, "step": 4080 }, { "epoch": 31.221374045801525, "grad_norm": 0.21050024032592773, "learning_rate": 0.000289170472385611, "loss": 0.0138, "step": 4090 }, { "epoch": 31.297709923664122, "grad_norm": 0.24995115399360657, "learning_rate": 0.00028911079851222143, "loss": 0.0126, "step": 4100 }, { "epoch": 31.374045801526716, "grad_norm": 0.319960355758667, "learning_rate": 0.00028905096687170526, "loss": 0.0128, "step": 4110 }, { "epoch": 31.450381679389313, "grad_norm": 0.1276302933692932, "learning_rate": 0.0002889909775319182, "loss": 0.0124, "step": 4120 }, { "epoch": 31.52671755725191, "grad_norm": 0.1773679554462433, "learning_rate": 0.0002889308305608948, "loss": 0.0121, "step": 4130 }, { "epoch": 31.603053435114504, "grad_norm": 0.17289985716342926, "learning_rate": 0.0002888705260268484, "loss": 0.013, "step": 4140 }, { "epoch": 31.6793893129771, "grad_norm": 0.25654909014701843, "learning_rate": 0.0002888100639981712, "loss": 0.0114, "step": 4150 }, { "epoch": 31.755725190839694, "grad_norm": 0.21556445956230164, "learning_rate": 0.0002887494445434338, "loss": 0.0131, "step": 4160 }, { "epoch": 31.83206106870229, "grad_norm": 0.3145628869533539, "learning_rate": 0.0002886886677313853, "loss": 0.0138, "step": 4170 }, { "epoch": 31.908396946564885, "grad_norm": 0.24232634902000427, "learning_rate": 0.0002886277336309535, "loss": 0.0137, "step": 4180 }, { "epoch": 31.984732824427482, "grad_norm": 0.20032744109630585, "learning_rate": 0.0002885666423112444, "loss": 0.0125, "step": 4190 }, { "epoch": 32.06106870229008, "grad_norm": 0.24124976992607117, "learning_rate": 0.0002885053938415424, "loss": 0.0131, "step": 4200 }, { "epoch": 32.13740458015267, "grad_norm": 0.17066846787929535, "learning_rate": 0.00028844398829131, "loss": 0.0127, "step": 4210 }, { "epoch": 32.213740458015266, "grad_norm": 0.23181477189064026, "learning_rate": 0.000288382425730188, "loss": 0.0124, "step": 4220 }, { "epoch": 32.29007633587786, "grad_norm": 0.2650354504585266, "learning_rate": 0.0002883207062279951, "loss": 0.0135, "step": 4230 }, { "epoch": 32.36641221374046, "grad_norm": 0.2508505582809448, "learning_rate": 0.0002882588298547282, "loss": 0.0118, "step": 4240 }, { "epoch": 32.44274809160305, "grad_norm": 0.20936551690101624, "learning_rate": 0.0002881967966805619, "loss": 0.0117, "step": 4250 }, { "epoch": 32.51908396946565, "grad_norm": 0.20056526362895966, "learning_rate": 0.0002881346067758488, "loss": 0.012, "step": 4260 }, { "epoch": 32.595419847328245, "grad_norm": 0.16372551023960114, "learning_rate": 0.00028807226021111915, "loss": 0.0122, "step": 4270 }, { "epoch": 32.67175572519084, "grad_norm": 0.18403998017311096, "learning_rate": 0.00028800975705708086, "loss": 0.0117, "step": 4280 }, { "epoch": 32.74809160305343, "grad_norm": 0.17488311231136322, "learning_rate": 0.0002879470973846195, "loss": 0.0114, "step": 4290 }, { "epoch": 32.82442748091603, "grad_norm": 0.13971993327140808, "learning_rate": 0.0002878842812647981, "loss": 0.0111, "step": 4300 }, { "epoch": 32.900763358778626, "grad_norm": 0.17583777010440826, "learning_rate": 0.0002878213087688571, "loss": 0.0116, "step": 4310 }, { "epoch": 32.97709923664122, "grad_norm": 0.1820119023323059, "learning_rate": 0.00028775817996821437, "loss": 0.0117, "step": 4320 }, { "epoch": 33.05343511450382, "grad_norm": 0.2015954703092575, "learning_rate": 0.000287694894934465, "loss": 0.0113, "step": 4330 }, { "epoch": 33.12977099236641, "grad_norm": 0.16993926465511322, "learning_rate": 0.00028763145373938113, "loss": 0.012, "step": 4340 }, { "epoch": 33.20610687022901, "grad_norm": 0.21269874274730682, "learning_rate": 0.0002875678564549123, "loss": 0.0122, "step": 4350 }, { "epoch": 33.282442748091604, "grad_norm": 0.23238570988178253, "learning_rate": 0.00028750410315318484, "loss": 0.0123, "step": 4360 }, { "epoch": 33.3587786259542, "grad_norm": 0.10783619433641434, "learning_rate": 0.0002874401939065021, "loss": 0.013, "step": 4370 }, { "epoch": 33.43511450381679, "grad_norm": 0.25055089592933655, "learning_rate": 0.0002873761287873442, "loss": 0.0117, "step": 4380 }, { "epoch": 33.51145038167939, "grad_norm": 0.1564796268939972, "learning_rate": 0.00028731190786836827, "loss": 0.0104, "step": 4390 }, { "epoch": 33.587786259541986, "grad_norm": 0.1760939657688141, "learning_rate": 0.0002872475312224079, "loss": 0.0104, "step": 4400 }, { "epoch": 33.66412213740458, "grad_norm": 0.2031378448009491, "learning_rate": 0.00028718299892247325, "loss": 0.0116, "step": 4410 }, { "epoch": 33.74045801526717, "grad_norm": 0.20807945728302002, "learning_rate": 0.0002871183110417514, "loss": 0.0113, "step": 4420 }, { "epoch": 33.81679389312977, "grad_norm": 0.24740906059741974, "learning_rate": 0.00028705346765360535, "loss": 0.0116, "step": 4430 }, { "epoch": 33.89312977099237, "grad_norm": 0.17369431257247925, "learning_rate": 0.0002869884688315748, "loss": 0.0117, "step": 4440 }, { "epoch": 33.969465648854964, "grad_norm": 0.16990123689174652, "learning_rate": 0.00028692331464937575, "loss": 0.0125, "step": 4450 }, { "epoch": 34.045801526717554, "grad_norm": 0.17801208794116974, "learning_rate": 0.00028685800518090017, "loss": 0.0123, "step": 4460 }, { "epoch": 34.12213740458015, "grad_norm": 0.19329388439655304, "learning_rate": 0.00028679254050021627, "loss": 0.0117, "step": 4470 }, { "epoch": 34.19847328244275, "grad_norm": 0.23113957047462463, "learning_rate": 0.00028672692068156837, "loss": 0.0112, "step": 4480 }, { "epoch": 34.274809160305345, "grad_norm": 0.17523393034934998, "learning_rate": 0.00028666114579937654, "loss": 0.0107, "step": 4490 }, { "epoch": 34.35114503816794, "grad_norm": 0.16607479751110077, "learning_rate": 0.000286595215928237, "loss": 0.0111, "step": 4500 }, { "epoch": 34.42748091603053, "grad_norm": 0.23250840604305267, "learning_rate": 0.00028652913114292133, "loss": 0.0111, "step": 4510 }, { "epoch": 34.50381679389313, "grad_norm": 0.3611747622489929, "learning_rate": 0.00028646289151837716, "loss": 0.0134, "step": 4520 }, { "epoch": 34.58015267175573, "grad_norm": 0.2558183968067169, "learning_rate": 0.0002863964971297275, "loss": 0.0126, "step": 4530 }, { "epoch": 34.656488549618324, "grad_norm": 0.2714023292064667, "learning_rate": 0.0002863299480522711, "loss": 0.0132, "step": 4540 }, { "epoch": 34.732824427480914, "grad_norm": 0.15352556109428406, "learning_rate": 0.000286263244361482, "loss": 0.0134, "step": 4550 }, { "epoch": 34.80916030534351, "grad_norm": 0.2544955909252167, "learning_rate": 0.00028619638613300953, "loss": 0.0124, "step": 4560 }, { "epoch": 34.88549618320611, "grad_norm": 0.24464571475982666, "learning_rate": 0.0002861293734426784, "loss": 0.0116, "step": 4570 }, { "epoch": 34.961832061068705, "grad_norm": 0.19236060976982117, "learning_rate": 0.0002860622063664884, "loss": 0.0104, "step": 4580 }, { "epoch": 35.038167938931295, "grad_norm": 0.19736382365226746, "learning_rate": 0.0002859948849806147, "loss": 0.0106, "step": 4590 }, { "epoch": 35.11450381679389, "grad_norm": 0.16228386759757996, "learning_rate": 0.00028592740936140697, "loss": 0.0107, "step": 4600 }, { "epoch": 35.19083969465649, "grad_norm": 0.17170938849449158, "learning_rate": 0.0002858597795853902, "loss": 0.0111, "step": 4610 }, { "epoch": 35.267175572519086, "grad_norm": 0.24485325813293457, "learning_rate": 0.000285791995729264, "loss": 0.0102, "step": 4620 }, { "epoch": 35.343511450381676, "grad_norm": 0.1755574494600296, "learning_rate": 0.00028572405786990294, "loss": 0.011, "step": 4630 }, { "epoch": 35.41984732824427, "grad_norm": 0.19914552569389343, "learning_rate": 0.00028565596608435596, "loss": 0.0103, "step": 4640 }, { "epoch": 35.49618320610687, "grad_norm": 0.13760389387607574, "learning_rate": 0.0002855877204498469, "loss": 0.011, "step": 4650 }, { "epoch": 35.57251908396947, "grad_norm": 0.2346106618642807, "learning_rate": 0.00028551932104377373, "loss": 0.0107, "step": 4660 }, { "epoch": 35.64885496183206, "grad_norm": 0.19085685908794403, "learning_rate": 0.0002854507679437091, "loss": 0.0098, "step": 4670 }, { "epoch": 35.725190839694655, "grad_norm": 0.14811711013317108, "learning_rate": 0.00028538206122739976, "loss": 0.0114, "step": 4680 }, { "epoch": 35.80152671755725, "grad_norm": 0.12879614531993866, "learning_rate": 0.0002853132009727669, "loss": 0.0103, "step": 4690 }, { "epoch": 35.87786259541985, "grad_norm": 0.1895889937877655, "learning_rate": 0.0002852441872579056, "loss": 0.0106, "step": 4700 }, { "epoch": 35.954198473282446, "grad_norm": 0.19250351190567017, "learning_rate": 0.0002851750201610852, "loss": 0.0111, "step": 4710 }, { "epoch": 36.030534351145036, "grad_norm": 0.14889097213745117, "learning_rate": 0.0002851056997607487, "loss": 0.011, "step": 4720 }, { "epoch": 36.10687022900763, "grad_norm": 0.26144668459892273, "learning_rate": 0.0002850362261355134, "loss": 0.0113, "step": 4730 }, { "epoch": 36.18320610687023, "grad_norm": 0.23588454723358154, "learning_rate": 0.0002849665993641699, "loss": 0.0117, "step": 4740 }, { "epoch": 36.25954198473283, "grad_norm": 0.1864294558763504, "learning_rate": 0.00028489681952568286, "loss": 0.0113, "step": 4750 }, { "epoch": 36.33587786259542, "grad_norm": 0.1824018508195877, "learning_rate": 0.00028482688669919027, "loss": 0.0128, "step": 4760 }, { "epoch": 36.412213740458014, "grad_norm": 0.22160357236862183, "learning_rate": 0.00028475680096400383, "loss": 0.012, "step": 4770 }, { "epoch": 36.48854961832061, "grad_norm": 0.19882750511169434, "learning_rate": 0.0002846865623996085, "loss": 0.0125, "step": 4780 }, { "epoch": 36.56488549618321, "grad_norm": 0.24023662507534027, "learning_rate": 0.0002846161710856627, "loss": 0.0119, "step": 4790 }, { "epoch": 36.6412213740458, "grad_norm": 0.20313464105129242, "learning_rate": 0.0002845456271019979, "loss": 0.0108, "step": 4800 }, { "epoch": 36.717557251908396, "grad_norm": 0.2312474250793457, "learning_rate": 0.0002844749305286189, "loss": 0.0117, "step": 4810 }, { "epoch": 36.79389312977099, "grad_norm": 0.1883460432291031, "learning_rate": 0.0002844040814457035, "loss": 0.0115, "step": 4820 }, { "epoch": 36.87022900763359, "grad_norm": 0.20207954943180084, "learning_rate": 0.0002843330799336024, "loss": 0.0102, "step": 4830 }, { "epoch": 36.94656488549618, "grad_norm": 0.2089201956987381, "learning_rate": 0.00028426192607283924, "loss": 0.0111, "step": 4840 }, { "epoch": 37.02290076335878, "grad_norm": 0.19836455583572388, "learning_rate": 0.0002841906199441104, "loss": 0.01, "step": 4850 }, { "epoch": 37.099236641221374, "grad_norm": 0.13542115688323975, "learning_rate": 0.000284119161628285, "loss": 0.0103, "step": 4860 }, { "epoch": 37.17557251908397, "grad_norm": 0.1740223467350006, "learning_rate": 0.0002840475512064047, "loss": 0.0106, "step": 4870 }, { "epoch": 37.25190839694657, "grad_norm": 0.13962678611278534, "learning_rate": 0.00028397578875968366, "loss": 0.0102, "step": 4880 }, { "epoch": 37.32824427480916, "grad_norm": 0.1193007230758667, "learning_rate": 0.0002839038743695085, "loss": 0.01, "step": 4890 }, { "epoch": 37.404580152671755, "grad_norm": 0.2286120355129242, "learning_rate": 0.0002838318081174382, "loss": 0.0101, "step": 4900 }, { "epoch": 37.48091603053435, "grad_norm": 0.19597183167934418, "learning_rate": 0.0002837595900852038, "loss": 0.0108, "step": 4910 }, { "epoch": 37.55725190839695, "grad_norm": 0.16181227564811707, "learning_rate": 0.0002836872203547087, "loss": 0.0113, "step": 4920 }, { "epoch": 37.63358778625954, "grad_norm": 0.25190484523773193, "learning_rate": 0.0002836146990080281, "loss": 0.0107, "step": 4930 }, { "epoch": 37.70992366412214, "grad_norm": 0.20482443273067474, "learning_rate": 0.00028354202612740944, "loss": 0.0119, "step": 4940 }, { "epoch": 37.786259541984734, "grad_norm": 0.1700560599565506, "learning_rate": 0.0002834692017952717, "loss": 0.0114, "step": 4950 }, { "epoch": 37.86259541984733, "grad_norm": 0.18750186264514923, "learning_rate": 0.00028339622609420585, "loss": 0.0112, "step": 4960 }, { "epoch": 37.93893129770992, "grad_norm": 0.10341199487447739, "learning_rate": 0.00028332309910697446, "loss": 0.0103, "step": 4970 }, { "epoch": 38.01526717557252, "grad_norm": 0.1980419009923935, "learning_rate": 0.0002832498209165117, "loss": 0.01, "step": 4980 }, { "epoch": 38.091603053435115, "grad_norm": 0.13724146783351898, "learning_rate": 0.0002831763916059231, "loss": 0.0096, "step": 4990 }, { "epoch": 38.16793893129771, "grad_norm": 0.18822596967220306, "learning_rate": 0.0002831028112584857, "loss": 0.0102, "step": 5000 }, { "epoch": 38.2442748091603, "grad_norm": 0.16042549908161163, "learning_rate": 0.0002830290799576479, "loss": 0.0111, "step": 5010 }, { "epoch": 38.3206106870229, "grad_norm": 0.23966854810714722, "learning_rate": 0.00028295519778702904, "loss": 0.0118, "step": 5020 }, { "epoch": 38.396946564885496, "grad_norm": 0.17393776774406433, "learning_rate": 0.00028288116483041984, "loss": 0.0115, "step": 5030 }, { "epoch": 38.47328244274809, "grad_norm": 0.22507421672344208, "learning_rate": 0.0002828069811717819, "loss": 0.0112, "step": 5040 }, { "epoch": 38.54961832061069, "grad_norm": 0.21440955996513367, "learning_rate": 0.0002827326468952477, "loss": 0.0115, "step": 5050 }, { "epoch": 38.62595419847328, "grad_norm": 0.16700772941112518, "learning_rate": 0.0002826581620851207, "loss": 0.0116, "step": 5060 }, { "epoch": 38.70229007633588, "grad_norm": 0.14702317118644714, "learning_rate": 0.00028258352682587474, "loss": 0.0107, "step": 5070 }, { "epoch": 38.778625954198475, "grad_norm": 0.12157769501209259, "learning_rate": 0.00028250874120215467, "loss": 0.0092, "step": 5080 }, { "epoch": 38.85496183206107, "grad_norm": 0.14037643373012543, "learning_rate": 0.00028243380529877575, "loss": 0.0097, "step": 5090 }, { "epoch": 38.93129770992366, "grad_norm": 0.11986864358186722, "learning_rate": 0.00028235871920072347, "loss": 0.0097, "step": 5100 }, { "epoch": 39.00763358778626, "grad_norm": 0.20346175134181976, "learning_rate": 0.00028228348299315397, "loss": 0.0114, "step": 5110 }, { "epoch": 39.083969465648856, "grad_norm": 0.17947809398174286, "learning_rate": 0.00028220809676139343, "loss": 0.0112, "step": 5120 }, { "epoch": 39.16030534351145, "grad_norm": 0.20891578495502472, "learning_rate": 0.0002821325605909382, "loss": 0.0103, "step": 5130 }, { "epoch": 39.23664122137404, "grad_norm": 0.20238623023033142, "learning_rate": 0.00028205687456745474, "loss": 0.0116, "step": 5140 }, { "epoch": 39.31297709923664, "grad_norm": 0.22119246423244476, "learning_rate": 0.00028198103877677936, "loss": 0.0101, "step": 5150 }, { "epoch": 39.38931297709924, "grad_norm": 0.17157317698001862, "learning_rate": 0.0002819050533049184, "loss": 0.0101, "step": 5160 }, { "epoch": 39.465648854961835, "grad_norm": 0.2048521488904953, "learning_rate": 0.00028182891823804774, "loss": 0.0106, "step": 5170 }, { "epoch": 39.541984732824424, "grad_norm": 0.13306495547294617, "learning_rate": 0.0002817526336625131, "loss": 0.0091, "step": 5180 }, { "epoch": 39.61832061068702, "grad_norm": 0.17353276908397675, "learning_rate": 0.00028167619966482966, "loss": 0.0093, "step": 5190 }, { "epoch": 39.69465648854962, "grad_norm": 0.17307431995868683, "learning_rate": 0.000281599616331682, "loss": 0.0106, "step": 5200 }, { "epoch": 39.770992366412216, "grad_norm": 0.14908336102962494, "learning_rate": 0.0002815228837499242, "loss": 0.0105, "step": 5210 }, { "epoch": 39.847328244274806, "grad_norm": 0.21360327303409576, "learning_rate": 0.0002814460020065795, "loss": 0.011, "step": 5220 }, { "epoch": 39.9236641221374, "grad_norm": 0.16842179000377655, "learning_rate": 0.00028136897118884044, "loss": 0.0104, "step": 5230 }, { "epoch": 40.0, "grad_norm": 0.1888081133365631, "learning_rate": 0.00028129179138406855, "loss": 0.0105, "step": 5240 }, { "epoch": 40.0763358778626, "grad_norm": 0.1671408712863922, "learning_rate": 0.00028121446267979417, "loss": 0.0097, "step": 5250 }, { "epoch": 40.152671755725194, "grad_norm": 0.18054036796092987, "learning_rate": 0.00028113698516371674, "loss": 0.0091, "step": 5260 }, { "epoch": 40.229007633587784, "grad_norm": 0.20723305642604828, "learning_rate": 0.00028105935892370446, "loss": 0.0107, "step": 5270 }, { "epoch": 40.30534351145038, "grad_norm": 0.18004123866558075, "learning_rate": 0.0002809815840477941, "loss": 0.0114, "step": 5280 }, { "epoch": 40.38167938931298, "grad_norm": 0.21794173121452332, "learning_rate": 0.0002809036606241909, "loss": 0.0103, "step": 5290 }, { "epoch": 40.458015267175576, "grad_norm": 0.28489673137664795, "learning_rate": 0.0002808255887412688, "loss": 0.0112, "step": 5300 }, { "epoch": 40.534351145038165, "grad_norm": 0.1481126844882965, "learning_rate": 0.00028074736848757, "loss": 0.0097, "step": 5310 }, { "epoch": 40.61068702290076, "grad_norm": 0.15996508300304413, "learning_rate": 0.00028066899995180493, "loss": 0.0082, "step": 5320 }, { "epoch": 40.68702290076336, "grad_norm": 0.16353878378868103, "learning_rate": 0.0002805904832228523, "loss": 0.009, "step": 5330 }, { "epoch": 40.76335877862596, "grad_norm": 0.2459944188594818, "learning_rate": 0.00028051181838975874, "loss": 0.0092, "step": 5340 }, { "epoch": 40.83969465648855, "grad_norm": 0.10940029472112656, "learning_rate": 0.00028043300554173896, "loss": 0.0097, "step": 5350 }, { "epoch": 40.916030534351144, "grad_norm": 0.16796660423278809, "learning_rate": 0.0002803540447681755, "loss": 0.0094, "step": 5360 }, { "epoch": 40.99236641221374, "grad_norm": 0.16863645613193512, "learning_rate": 0.0002802749361586187, "loss": 0.0099, "step": 5370 }, { "epoch": 41.06870229007634, "grad_norm": 0.13783586025238037, "learning_rate": 0.00028019567980278645, "loss": 0.0091, "step": 5380 }, { "epoch": 41.14503816793893, "grad_norm": 0.11061608791351318, "learning_rate": 0.0002801162757905643, "loss": 0.0092, "step": 5390 }, { "epoch": 41.221374045801525, "grad_norm": 0.136240154504776, "learning_rate": 0.0002800367242120053, "loss": 0.0097, "step": 5400 }, { "epoch": 41.29770992366412, "grad_norm": 0.1539125293493271, "learning_rate": 0.00027995702515732973, "loss": 0.0087, "step": 5410 }, { "epoch": 41.37404580152672, "grad_norm": 0.15205824375152588, "learning_rate": 0.00027987717871692515, "loss": 0.0092, "step": 5420 }, { "epoch": 41.45038167938931, "grad_norm": 0.1631704717874527, "learning_rate": 0.00027979718498134646, "loss": 0.0085, "step": 5430 }, { "epoch": 41.52671755725191, "grad_norm": 0.14011022448539734, "learning_rate": 0.0002797170440413153, "loss": 0.0098, "step": 5440 }, { "epoch": 41.603053435114504, "grad_norm": 0.32856714725494385, "learning_rate": 0.0002796367559877205, "loss": 0.0098, "step": 5450 }, { "epoch": 41.6793893129771, "grad_norm": 0.20607997477054596, "learning_rate": 0.0002795563209116176, "loss": 0.011, "step": 5460 }, { "epoch": 41.7557251908397, "grad_norm": 0.22206304967403412, "learning_rate": 0.000279475738904229, "loss": 0.0129, "step": 5470 }, { "epoch": 41.83206106870229, "grad_norm": 0.18307673931121826, "learning_rate": 0.0002793950100569436, "loss": 0.012, "step": 5480 }, { "epoch": 41.908396946564885, "grad_norm": 0.1957813948392868, "learning_rate": 0.000279314134461317, "loss": 0.011, "step": 5490 }, { "epoch": 41.98473282442748, "grad_norm": 0.1835593432188034, "learning_rate": 0.0002792331122090709, "loss": 0.0122, "step": 5500 }, { "epoch": 42.06106870229008, "grad_norm": 0.1622667759656906, "learning_rate": 0.0002791519433920937, "loss": 0.0117, "step": 5510 }, { "epoch": 42.13740458015267, "grad_norm": 0.2333444207906723, "learning_rate": 0.00027907062810243993, "loss": 0.0113, "step": 5520 }, { "epoch": 42.213740458015266, "grad_norm": 0.16240788996219635, "learning_rate": 0.0002789891664323301, "loss": 0.0096, "step": 5530 }, { "epoch": 42.29007633587786, "grad_norm": 0.18940560519695282, "learning_rate": 0.0002789075584741508, "loss": 0.0105, "step": 5540 }, { "epoch": 42.36641221374046, "grad_norm": 0.18209503591060638, "learning_rate": 0.00027882580432045455, "loss": 0.0091, "step": 5550 }, { "epoch": 42.44274809160305, "grad_norm": 0.21052083373069763, "learning_rate": 0.0002787439040639597, "loss": 0.0095, "step": 5560 }, { "epoch": 42.51908396946565, "grad_norm": 0.11451749503612518, "learning_rate": 0.0002786618577975502, "loss": 0.009, "step": 5570 }, { "epoch": 42.595419847328245, "grad_norm": 0.17762811481952667, "learning_rate": 0.0002785796656142756, "loss": 0.0087, "step": 5580 }, { "epoch": 42.67175572519084, "grad_norm": 0.2944685220718384, "learning_rate": 0.00027849732760735113, "loss": 0.0094, "step": 5590 }, { "epoch": 42.74809160305343, "grad_norm": 0.15770643949508667, "learning_rate": 0.0002784148438701571, "loss": 0.0096, "step": 5600 }, { "epoch": 42.82442748091603, "grad_norm": 0.15594616532325745, "learning_rate": 0.0002783322144962394, "loss": 0.0105, "step": 5610 }, { "epoch": 42.900763358778626, "grad_norm": 0.15631425380706787, "learning_rate": 0.0002782494395793089, "loss": 0.0099, "step": 5620 }, { "epoch": 42.97709923664122, "grad_norm": 0.19879840314388275, "learning_rate": 0.00027816651921324135, "loss": 0.0092, "step": 5630 }, { "epoch": 43.05343511450382, "grad_norm": 0.17070774734020233, "learning_rate": 0.00027808345349207797, "loss": 0.0101, "step": 5640 }, { "epoch": 43.12977099236641, "grad_norm": 0.16994762420654297, "learning_rate": 0.00027800024251002436, "loss": 0.009, "step": 5650 }, { "epoch": 43.20610687022901, "grad_norm": 0.18364204466342926, "learning_rate": 0.0002779168863614511, "loss": 0.01, "step": 5660 }, { "epoch": 43.282442748091604, "grad_norm": 0.1228414997458458, "learning_rate": 0.0002778333851408933, "loss": 0.0094, "step": 5670 }, { "epoch": 43.3587786259542, "grad_norm": 0.15428395569324493, "learning_rate": 0.0002777497389430507, "loss": 0.0091, "step": 5680 }, { "epoch": 43.43511450381679, "grad_norm": 0.16349372267723083, "learning_rate": 0.00027766594786278736, "loss": 0.0092, "step": 5690 }, { "epoch": 43.51145038167939, "grad_norm": 0.19047121703624725, "learning_rate": 0.0002775820119951316, "loss": 0.0095, "step": 5700 }, { "epoch": 43.587786259541986, "grad_norm": 0.14982996881008148, "learning_rate": 0.00027749793143527616, "loss": 0.0099, "step": 5710 }, { "epoch": 43.66412213740458, "grad_norm": 0.18341393768787384, "learning_rate": 0.00027741370627857773, "loss": 0.0103, "step": 5720 }, { "epoch": 43.74045801526717, "grad_norm": 0.14886951446533203, "learning_rate": 0.00027732933662055694, "loss": 0.01, "step": 5730 }, { "epoch": 43.81679389312977, "grad_norm": 0.13199369609355927, "learning_rate": 0.00027724482255689847, "loss": 0.009, "step": 5740 }, { "epoch": 43.89312977099237, "grad_norm": 0.1397256702184677, "learning_rate": 0.0002771601641834506, "loss": 0.0099, "step": 5750 }, { "epoch": 43.969465648854964, "grad_norm": 0.142601877450943, "learning_rate": 0.00027707536159622545, "loss": 0.0109, "step": 5760 }, { "epoch": 44.045801526717554, "grad_norm": 0.17512831091880798, "learning_rate": 0.00027699041489139843, "loss": 0.0103, "step": 5770 }, { "epoch": 44.12213740458015, "grad_norm": 0.22144131362438202, "learning_rate": 0.0002769053241653086, "loss": 0.0097, "step": 5780 }, { "epoch": 44.19847328244275, "grad_norm": 0.20264875888824463, "learning_rate": 0.00027682008951445845, "loss": 0.0102, "step": 5790 }, { "epoch": 44.274809160305345, "grad_norm": 0.20429769158363342, "learning_rate": 0.00027673471103551345, "loss": 0.0101, "step": 5800 }, { "epoch": 44.35114503816794, "grad_norm": 0.18966354429721832, "learning_rate": 0.00027664918882530225, "loss": 0.0094, "step": 5810 }, { "epoch": 44.42748091603053, "grad_norm": 0.13264338672161102, "learning_rate": 0.00027656352298081665, "loss": 0.0101, "step": 5820 }, { "epoch": 44.50381679389313, "grad_norm": 0.18010492622852325, "learning_rate": 0.0002764777135992112, "loss": 0.0095, "step": 5830 }, { "epoch": 44.58015267175573, "grad_norm": 0.20886415243148804, "learning_rate": 0.0002763917607778033, "loss": 0.0089, "step": 5840 }, { "epoch": 44.656488549618324, "grad_norm": 0.16522078216075897, "learning_rate": 0.00027630566461407305, "loss": 0.0101, "step": 5850 }, { "epoch": 44.732824427480914, "grad_norm": 0.13648925721645355, "learning_rate": 0.000276219425205663, "loss": 0.0101, "step": 5860 }, { "epoch": 44.80916030534351, "grad_norm": 0.19120123982429504, "learning_rate": 0.0002761330426503783, "loss": 0.0094, "step": 5870 }, { "epoch": 44.88549618320611, "grad_norm": 0.18119847774505615, "learning_rate": 0.00027604651704618636, "loss": 0.0088, "step": 5880 }, { "epoch": 44.961832061068705, "grad_norm": 0.18839700520038605, "learning_rate": 0.0002759598484912169, "loss": 0.0098, "step": 5890 }, { "epoch": 45.038167938931295, "grad_norm": 0.15616732835769653, "learning_rate": 0.00027587303708376156, "loss": 0.0096, "step": 5900 }, { "epoch": 45.11450381679389, "grad_norm": 0.140389084815979, "learning_rate": 0.00027578608292227433, "loss": 0.0093, "step": 5910 }, { "epoch": 45.19083969465649, "grad_norm": 0.18531467020511627, "learning_rate": 0.0002756989861053708, "loss": 0.0102, "step": 5920 }, { "epoch": 45.267175572519086, "grad_norm": 0.21750186383724213, "learning_rate": 0.00027561174673182843, "loss": 0.0097, "step": 5930 }, { "epoch": 45.343511450381676, "grad_norm": 0.143514484167099, "learning_rate": 0.0002755243649005864, "loss": 0.01, "step": 5940 }, { "epoch": 45.41984732824427, "grad_norm": 0.15143033862113953, "learning_rate": 0.00027543684071074543, "loss": 0.0098, "step": 5950 }, { "epoch": 45.49618320610687, "grad_norm": 0.18131192028522491, "learning_rate": 0.0002753491742615677, "loss": 0.01, "step": 5960 }, { "epoch": 45.57251908396947, "grad_norm": 0.11990111321210861, "learning_rate": 0.0002752613656524768, "loss": 0.0099, "step": 5970 }, { "epoch": 45.64885496183206, "grad_norm": 0.12245701253414154, "learning_rate": 0.00027517341498305733, "loss": 0.0105, "step": 5980 }, { "epoch": 45.725190839694655, "grad_norm": 0.1404580920934677, "learning_rate": 0.00027508532235305516, "loss": 0.0101, "step": 5990 }, { "epoch": 45.80152671755725, "grad_norm": 0.13683819770812988, "learning_rate": 0.00027499708786237723, "loss": 0.0098, "step": 6000 }, { "epoch": 45.87786259541985, "grad_norm": 0.13641896843910217, "learning_rate": 0.00027490871161109116, "loss": 0.0093, "step": 6010 }, { "epoch": 45.954198473282446, "grad_norm": 0.20045650005340576, "learning_rate": 0.00027482019369942553, "loss": 0.0095, "step": 6020 }, { "epoch": 46.030534351145036, "grad_norm": 0.21987608075141907, "learning_rate": 0.0002747315342277695, "loss": 0.0091, "step": 6030 }, { "epoch": 46.10687022900763, "grad_norm": 0.21844890713691711, "learning_rate": 0.00027464273329667275, "loss": 0.0109, "step": 6040 }, { "epoch": 46.18320610687023, "grad_norm": 0.19249963760375977, "learning_rate": 0.00027455379100684534, "loss": 0.0091, "step": 6050 }, { "epoch": 46.25954198473283, "grad_norm": 0.22168408334255219, "learning_rate": 0.0002744647074591579, "loss": 0.0102, "step": 6060 }, { "epoch": 46.33587786259542, "grad_norm": 0.1422118842601776, "learning_rate": 0.000274375482754641, "loss": 0.0112, "step": 6070 }, { "epoch": 46.412213740458014, "grad_norm": 0.21650353074073792, "learning_rate": 0.00027428611699448533, "loss": 0.0107, "step": 6080 }, { "epoch": 46.48854961832061, "grad_norm": 0.17264889180660248, "learning_rate": 0.0002741966102800417, "loss": 0.0102, "step": 6090 }, { "epoch": 46.56488549618321, "grad_norm": 0.1596447229385376, "learning_rate": 0.0002741069627128206, "loss": 0.0093, "step": 6100 }, { "epoch": 46.6412213740458, "grad_norm": 0.172441303730011, "learning_rate": 0.0002740171743944924, "loss": 0.009, "step": 6110 }, { "epoch": 46.717557251908396, "grad_norm": 0.18215081095695496, "learning_rate": 0.000273927245426887, "loss": 0.0103, "step": 6120 }, { "epoch": 46.79389312977099, "grad_norm": 0.13303907215595245, "learning_rate": 0.00027383717591199393, "loss": 0.0083, "step": 6130 }, { "epoch": 46.87022900763359, "grad_norm": 0.1901869922876358, "learning_rate": 0.00027374696595196203, "loss": 0.0096, "step": 6140 }, { "epoch": 46.94656488549618, "grad_norm": 0.10262227058410645, "learning_rate": 0.0002736566156490994, "loss": 0.0095, "step": 6150 }, { "epoch": 47.02290076335878, "grad_norm": 0.14739590883255005, "learning_rate": 0.00027356612510587335, "loss": 0.0085, "step": 6160 }, { "epoch": 47.099236641221374, "grad_norm": 0.14651551842689514, "learning_rate": 0.00027347549442491016, "loss": 0.0094, "step": 6170 }, { "epoch": 47.17557251908397, "grad_norm": 0.1927761435508728, "learning_rate": 0.00027338472370899523, "loss": 0.0091, "step": 6180 }, { "epoch": 47.25190839694657, "grad_norm": 0.16236604750156403, "learning_rate": 0.0002732938130610726, "loss": 0.0105, "step": 6190 }, { "epoch": 47.32824427480916, "grad_norm": 0.19460533559322357, "learning_rate": 0.00027320276258424505, "loss": 0.0104, "step": 6200 }, { "epoch": 47.404580152671755, "grad_norm": 0.15930339694023132, "learning_rate": 0.0002731115723817739, "loss": 0.0104, "step": 6210 }, { "epoch": 47.48091603053435, "grad_norm": 0.13027901947498322, "learning_rate": 0.0002730202425570791, "loss": 0.0101, "step": 6220 }, { "epoch": 47.55725190839695, "grad_norm": 0.18544086813926697, "learning_rate": 0.0002729287732137388, "loss": 0.0123, "step": 6230 }, { "epoch": 47.63358778625954, "grad_norm": 0.17071180045604706, "learning_rate": 0.0002728371644554893, "loss": 0.0108, "step": 6240 }, { "epoch": 47.70992366412214, "grad_norm": 0.27330538630485535, "learning_rate": 0.0002727454163862253, "loss": 0.0097, "step": 6250 }, { "epoch": 47.786259541984734, "grad_norm": 0.18766507506370544, "learning_rate": 0.0002726535291099993, "loss": 0.0105, "step": 6260 }, { "epoch": 47.86259541984733, "grad_norm": 0.12009459733963013, "learning_rate": 0.0002725615027310216, "loss": 0.0101, "step": 6270 }, { "epoch": 47.93893129770992, "grad_norm": 0.11593929678201675, "learning_rate": 0.00027246933735366037, "loss": 0.0086, "step": 6280 }, { "epoch": 48.01526717557252, "grad_norm": 0.13773618638515472, "learning_rate": 0.0002723770330824414, "loss": 0.0101, "step": 6290 }, { "epoch": 48.091603053435115, "grad_norm": 0.17289403080940247, "learning_rate": 0.00027228459002204806, "loss": 0.0091, "step": 6300 }, { "epoch": 48.16793893129771, "grad_norm": 0.1684638261795044, "learning_rate": 0.0002721920082773211, "loss": 0.0093, "step": 6310 }, { "epoch": 48.2442748091603, "grad_norm": 0.1450728476047516, "learning_rate": 0.0002720992879532584, "loss": 0.0094, "step": 6320 }, { "epoch": 48.3206106870229, "grad_norm": 0.15592429041862488, "learning_rate": 0.0002720064291550152, "loss": 0.0092, "step": 6330 }, { "epoch": 48.396946564885496, "grad_norm": 0.1633678376674652, "learning_rate": 0.00027191343198790377, "loss": 0.01, "step": 6340 }, { "epoch": 48.47328244274809, "grad_norm": 0.13475382328033447, "learning_rate": 0.0002718202965573931, "loss": 0.0091, "step": 6350 }, { "epoch": 48.54961832061069, "grad_norm": 0.14100494980812073, "learning_rate": 0.0002717270229691093, "loss": 0.0095, "step": 6360 }, { "epoch": 48.62595419847328, "grad_norm": 0.15145069360733032, "learning_rate": 0.00027163361132883485, "loss": 0.0095, "step": 6370 }, { "epoch": 48.70229007633588, "grad_norm": 0.1817992627620697, "learning_rate": 0.000271540061742509, "loss": 0.0095, "step": 6380 }, { "epoch": 48.778625954198475, "grad_norm": 0.15830519795417786, "learning_rate": 0.00027144637431622743, "loss": 0.0095, "step": 6390 }, { "epoch": 48.85496183206107, "grad_norm": 0.211346834897995, "learning_rate": 0.0002713525491562421, "loss": 0.0092, "step": 6400 }, { "epoch": 48.93129770992366, "grad_norm": 0.1773797869682312, "learning_rate": 0.0002712585863689611, "loss": 0.0096, "step": 6410 }, { "epoch": 49.00763358778626, "grad_norm": 0.15111806988716125, "learning_rate": 0.0002711644860609488, "loss": 0.0094, "step": 6420 }, { "epoch": 49.083969465648856, "grad_norm": 0.1786964237689972, "learning_rate": 0.00027107024833892537, "loss": 0.0086, "step": 6430 }, { "epoch": 49.16030534351145, "grad_norm": 0.12940526008605957, "learning_rate": 0.00027097587330976686, "loss": 0.009, "step": 6440 }, { "epoch": 49.23664122137404, "grad_norm": 0.12582065165042877, "learning_rate": 0.000270881361080505, "loss": 0.0091, "step": 6450 }, { "epoch": 49.31297709923664, "grad_norm": 0.10780740529298782, "learning_rate": 0.0002707867117583273, "loss": 0.0095, "step": 6460 }, { "epoch": 49.38931297709924, "grad_norm": 0.19650834798812866, "learning_rate": 0.0002706919254505766, "loss": 0.0104, "step": 6470 }, { "epoch": 49.465648854961835, "grad_norm": 0.19569462537765503, "learning_rate": 0.0002705970022647511, "loss": 0.0095, "step": 6480 }, { "epoch": 49.541984732824424, "grad_norm": 0.15975622832775116, "learning_rate": 0.0002705019423085042, "loss": 0.0093, "step": 6490 }, { "epoch": 49.61832061068702, "grad_norm": 0.09325077384710312, "learning_rate": 0.0002704067456896445, "loss": 0.0089, "step": 6500 }, { "epoch": 49.69465648854962, "grad_norm": 0.1230226382613182, "learning_rate": 0.00027031141251613564, "loss": 0.0077, "step": 6510 }, { "epoch": 49.770992366412216, "grad_norm": 0.17234501242637634, "learning_rate": 0.00027021594289609597, "loss": 0.0087, "step": 6520 }, { "epoch": 49.847328244274806, "grad_norm": 0.1347607672214508, "learning_rate": 0.0002701203369377986, "loss": 0.0085, "step": 6530 }, { "epoch": 49.9236641221374, "grad_norm": 0.14832928776741028, "learning_rate": 0.0002700245947496715, "loss": 0.0087, "step": 6540 }, { "epoch": 50.0, "grad_norm": 0.09738212078809738, "learning_rate": 0.00026992871644029685, "loss": 0.0079, "step": 6550 }, { "epoch": 50.0763358778626, "grad_norm": 0.1227264553308487, "learning_rate": 0.00026983270211841133, "loss": 0.0082, "step": 6560 }, { "epoch": 50.152671755725194, "grad_norm": 0.1594538688659668, "learning_rate": 0.00026973655189290585, "loss": 0.0099, "step": 6570 }, { "epoch": 50.229007633587784, "grad_norm": 0.1378728151321411, "learning_rate": 0.0002696402658728255, "loss": 0.0093, "step": 6580 }, { "epoch": 50.30534351145038, "grad_norm": 0.15824854373931885, "learning_rate": 0.0002695438441673694, "loss": 0.011, "step": 6590 }, { "epoch": 50.38167938931298, "grad_norm": 0.136995330452919, "learning_rate": 0.0002694472868858904, "loss": 0.0102, "step": 6600 }, { "epoch": 50.458015267175576, "grad_norm": 0.11527341604232788, "learning_rate": 0.0002693505941378952, "loss": 0.009, "step": 6610 }, { "epoch": 50.534351145038165, "grad_norm": 0.1296534687280655, "learning_rate": 0.00026925376603304424, "loss": 0.0101, "step": 6620 }, { "epoch": 50.61068702290076, "grad_norm": 0.1343197077512741, "learning_rate": 0.00026915680268115125, "loss": 0.0103, "step": 6630 }, { "epoch": 50.68702290076336, "grad_norm": 0.12560690939426422, "learning_rate": 0.0002690597041921835, "loss": 0.0091, "step": 6640 }, { "epoch": 50.76335877862596, "grad_norm": 0.17722758650779724, "learning_rate": 0.0002689624706762615, "loss": 0.0091, "step": 6650 }, { "epoch": 50.83969465648855, "grad_norm": 0.1450289487838745, "learning_rate": 0.0002688651022436589, "loss": 0.0087, "step": 6660 }, { "epoch": 50.916030534351144, "grad_norm": 0.14958162605762482, "learning_rate": 0.00026876759900480225, "loss": 0.008, "step": 6670 }, { "epoch": 50.99236641221374, "grad_norm": 0.1601370871067047, "learning_rate": 0.0002686699610702712, "loss": 0.0085, "step": 6680 }, { "epoch": 51.06870229007634, "grad_norm": 0.11638301610946655, "learning_rate": 0.0002685721885507979, "loss": 0.0087, "step": 6690 }, { "epoch": 51.14503816793893, "grad_norm": 0.132828027009964, "learning_rate": 0.0002684742815572674, "loss": 0.0075, "step": 6700 }, { "epoch": 51.221374045801525, "grad_norm": 0.18796780705451965, "learning_rate": 0.00026837624020071703, "loss": 0.0094, "step": 6710 }, { "epoch": 51.29770992366412, "grad_norm": 0.12449473142623901, "learning_rate": 0.00026827806459233663, "loss": 0.0084, "step": 6720 }, { "epoch": 51.37404580152672, "grad_norm": 0.11377227306365967, "learning_rate": 0.00026817975484346823, "loss": 0.0091, "step": 6730 }, { "epoch": 51.45038167938931, "grad_norm": 0.1484663188457489, "learning_rate": 0.0002680813110656061, "loss": 0.0093, "step": 6740 }, { "epoch": 51.52671755725191, "grad_norm": 0.14351825416088104, "learning_rate": 0.00026798273337039636, "loss": 0.0095, "step": 6750 }, { "epoch": 51.603053435114504, "grad_norm": 0.14768099784851074, "learning_rate": 0.00026788402186963715, "loss": 0.009, "step": 6760 }, { "epoch": 51.6793893129771, "grad_norm": 0.14263977110385895, "learning_rate": 0.00026778517667527823, "loss": 0.0089, "step": 6770 }, { "epoch": 51.7557251908397, "grad_norm": 0.1426563411951065, "learning_rate": 0.0002676861978994212, "loss": 0.0083, "step": 6780 }, { "epoch": 51.83206106870229, "grad_norm": 0.16189096868038177, "learning_rate": 0.00026758708565431883, "loss": 0.0094, "step": 6790 }, { "epoch": 51.908396946564885, "grad_norm": 0.14942017197608948, "learning_rate": 0.0002674878400523755, "loss": 0.009, "step": 6800 }, { "epoch": 51.98473282442748, "grad_norm": 0.1483214795589447, "learning_rate": 0.00026738846120614676, "loss": 0.0089, "step": 6810 }, { "epoch": 52.06106870229008, "grad_norm": 0.17906005680561066, "learning_rate": 0.0002672889492283393, "loss": 0.0107, "step": 6820 }, { "epoch": 52.13740458015267, "grad_norm": 0.1387445032596588, "learning_rate": 0.0002671893042318108, "loss": 0.0091, "step": 6830 }, { "epoch": 52.213740458015266, "grad_norm": 0.23211362957954407, "learning_rate": 0.00026708952632956975, "loss": 0.0101, "step": 6840 }, { "epoch": 52.29007633587786, "grad_norm": 0.23178142309188843, "learning_rate": 0.0002669896156347754, "loss": 0.0092, "step": 6850 }, { "epoch": 52.36641221374046, "grad_norm": 0.21561317145824432, "learning_rate": 0.0002668895722607376, "loss": 0.0091, "step": 6860 }, { "epoch": 52.44274809160305, "grad_norm": 0.1640937626361847, "learning_rate": 0.0002667893963209166, "loss": 0.0092, "step": 6870 }, { "epoch": 52.51908396946565, "grad_norm": 0.1372833251953125, "learning_rate": 0.0002666890879289231, "loss": 0.0097, "step": 6880 }, { "epoch": 52.595419847328245, "grad_norm": 0.1488010734319687, "learning_rate": 0.00026658864719851803, "loss": 0.0103, "step": 6890 }, { "epoch": 52.67175572519084, "grad_norm": 0.14813227951526642, "learning_rate": 0.0002664880742436124, "loss": 0.0098, "step": 6900 }, { "epoch": 52.74809160305343, "grad_norm": 0.15854477882385254, "learning_rate": 0.000266387369178267, "loss": 0.0092, "step": 6910 }, { "epoch": 52.82442748091603, "grad_norm": 0.11721304059028625, "learning_rate": 0.00026628653211669263, "loss": 0.0088, "step": 6920 }, { "epoch": 52.900763358778626, "grad_norm": 0.1929912120103836, "learning_rate": 0.0002661855631732498, "loss": 0.0084, "step": 6930 }, { "epoch": 52.97709923664122, "grad_norm": 0.10498005896806717, "learning_rate": 0.0002660844624624484, "loss": 0.0079, "step": 6940 }, { "epoch": 53.05343511450382, "grad_norm": 0.10716800391674042, "learning_rate": 0.00026598323009894805, "loss": 0.0083, "step": 6950 }, { "epoch": 53.12977099236641, "grad_norm": 0.11678940802812576, "learning_rate": 0.00026588186619755735, "loss": 0.0082, "step": 6960 }, { "epoch": 53.20610687022901, "grad_norm": 0.1717982292175293, "learning_rate": 0.0002657803708732344, "loss": 0.0091, "step": 6970 }, { "epoch": 53.282442748091604, "grad_norm": 0.1686500608921051, "learning_rate": 0.0002656787442410861, "loss": 0.0085, "step": 6980 }, { "epoch": 53.3587786259542, "grad_norm": 0.1396321952342987, "learning_rate": 0.00026557698641636835, "loss": 0.0085, "step": 6990 }, { "epoch": 53.43511450381679, "grad_norm": 0.1341378539800644, "learning_rate": 0.0002654750975144859, "loss": 0.0079, "step": 7000 }, { "epoch": 53.51145038167939, "grad_norm": 0.124003104865551, "learning_rate": 0.0002653730776509921, "loss": 0.0095, "step": 7010 }, { "epoch": 53.587786259541986, "grad_norm": 0.18334771692752838, "learning_rate": 0.00026527092694158877, "loss": 0.0096, "step": 7020 }, { "epoch": 53.66412213740458, "grad_norm": 0.1416471153497696, "learning_rate": 0.0002651686455021263, "loss": 0.0098, "step": 7030 }, { "epoch": 53.74045801526717, "grad_norm": 0.14335116744041443, "learning_rate": 0.00026506623344860306, "loss": 0.0099, "step": 7040 }, { "epoch": 53.81679389312977, "grad_norm": 0.15584681928157806, "learning_rate": 0.0002649636908971658, "loss": 0.0121, "step": 7050 }, { "epoch": 53.89312977099237, "grad_norm": 0.19893914461135864, "learning_rate": 0.0002648610179641093, "loss": 0.0113, "step": 7060 }, { "epoch": 53.969465648854964, "grad_norm": 0.1664087325334549, "learning_rate": 0.00026475821476587585, "loss": 0.0115, "step": 7070 }, { "epoch": 54.045801526717554, "grad_norm": 0.18171508610248566, "learning_rate": 0.00026465528141905595, "loss": 0.0102, "step": 7080 }, { "epoch": 54.12213740458015, "grad_norm": 0.17161455750465393, "learning_rate": 0.0002645522180403873, "loss": 0.01, "step": 7090 }, { "epoch": 54.19847328244275, "grad_norm": 0.12637189030647278, "learning_rate": 0.00026444902474675536, "loss": 0.0092, "step": 7100 }, { "epoch": 54.274809160305345, "grad_norm": 0.15204553306102753, "learning_rate": 0.0002643457016551927, "loss": 0.0092, "step": 7110 }, { "epoch": 54.35114503816794, "grad_norm": 0.16782180964946747, "learning_rate": 0.0002642422488828793, "loss": 0.0094, "step": 7120 }, { "epoch": 54.42748091603053, "grad_norm": 0.13394570350646973, "learning_rate": 0.00026413866654714205, "loss": 0.0089, "step": 7130 }, { "epoch": 54.50381679389313, "grad_norm": 0.08902191370725632, "learning_rate": 0.0002640349547654549, "loss": 0.0088, "step": 7140 }, { "epoch": 54.58015267175573, "grad_norm": 0.10665497928857803, "learning_rate": 0.00026393111365543854, "loss": 0.0091, "step": 7150 }, { "epoch": 54.656488549618324, "grad_norm": 0.0987040102481842, "learning_rate": 0.0002638271433348603, "loss": 0.0082, "step": 7160 }, { "epoch": 54.732824427480914, "grad_norm": 0.13163644075393677, "learning_rate": 0.0002637230439216341, "loss": 0.0092, "step": 7170 }, { "epoch": 54.80916030534351, "grad_norm": 0.14364294707775116, "learning_rate": 0.00026361881553382035, "loss": 0.0107, "step": 7180 }, { "epoch": 54.88549618320611, "grad_norm": 0.11596263200044632, "learning_rate": 0.00026351445828962555, "loss": 0.0085, "step": 7190 }, { "epoch": 54.961832061068705, "grad_norm": 0.12828397750854492, "learning_rate": 0.00026340997230740244, "loss": 0.0087, "step": 7200 }, { "epoch": 55.038167938931295, "grad_norm": 0.11460773646831512, "learning_rate": 0.0002633053577056498, "loss": 0.0083, "step": 7210 }, { "epoch": 55.11450381679389, "grad_norm": 0.11488089710474014, "learning_rate": 0.0002632006146030122, "loss": 0.0094, "step": 7220 }, { "epoch": 55.19083969465649, "grad_norm": 0.17392957210540771, "learning_rate": 0.00026309574311827994, "loss": 0.0088, "step": 7230 }, { "epoch": 55.267175572519086, "grad_norm": 0.17735671997070312, "learning_rate": 0.000262990743370389, "loss": 0.0091, "step": 7240 }, { "epoch": 55.343511450381676, "grad_norm": 0.14501376450061798, "learning_rate": 0.00026288561547842075, "loss": 0.0086, "step": 7250 }, { "epoch": 55.41984732824427, "grad_norm": 0.1427486687898636, "learning_rate": 0.00026278035956160196, "loss": 0.0082, "step": 7260 }, { "epoch": 55.49618320610687, "grad_norm": 0.16225934028625488, "learning_rate": 0.0002626749757393045, "loss": 0.0084, "step": 7270 }, { "epoch": 55.57251908396947, "grad_norm": 0.1675959676504135, "learning_rate": 0.0002625694641310454, "loss": 0.0086, "step": 7280 }, { "epoch": 55.64885496183206, "grad_norm": 0.17598648369312286, "learning_rate": 0.0002624638248564865, "loss": 0.0088, "step": 7290 }, { "epoch": 55.725190839694655, "grad_norm": 0.10684394091367722, "learning_rate": 0.00026235805803543466, "loss": 0.008, "step": 7300 }, { "epoch": 55.80152671755725, "grad_norm": 0.14399273693561554, "learning_rate": 0.00026225216378784097, "loss": 0.0083, "step": 7310 }, { "epoch": 55.87786259541985, "grad_norm": 0.13222572207450867, "learning_rate": 0.00026214614223380144, "loss": 0.0079, "step": 7320 }, { "epoch": 55.954198473282446, "grad_norm": 0.15194641053676605, "learning_rate": 0.0002620399934935564, "loss": 0.0085, "step": 7330 }, { "epoch": 56.030534351145036, "grad_norm": 0.1031711995601654, "learning_rate": 0.00026193371768749017, "loss": 0.008, "step": 7340 }, { "epoch": 56.10687022900763, "grad_norm": 0.18336021900177002, "learning_rate": 0.0002618273149361314, "loss": 0.0089, "step": 7350 }, { "epoch": 56.18320610687023, "grad_norm": 0.15615668892860413, "learning_rate": 0.0002617207853601526, "loss": 0.0088, "step": 7360 }, { "epoch": 56.25954198473283, "grad_norm": 0.1989274024963379, "learning_rate": 0.00026161412908037027, "loss": 0.008, "step": 7370 }, { "epoch": 56.33587786259542, "grad_norm": 0.2012815773487091, "learning_rate": 0.0002615073462177445, "loss": 0.0097, "step": 7380 }, { "epoch": 56.412213740458014, "grad_norm": 0.25554099678993225, "learning_rate": 0.0002614004368933788, "loss": 0.0094, "step": 7390 }, { "epoch": 56.48854961832061, "grad_norm": 0.14058351516723633, "learning_rate": 0.00026129340122852036, "loss": 0.009, "step": 7400 }, { "epoch": 56.56488549618321, "grad_norm": 0.15234065055847168, "learning_rate": 0.00026118623934455953, "loss": 0.0091, "step": 7410 }, { "epoch": 56.6412213740458, "grad_norm": 0.15382155776023865, "learning_rate": 0.00026107895136302973, "loss": 0.0101, "step": 7420 }, { "epoch": 56.717557251908396, "grad_norm": 0.1484890580177307, "learning_rate": 0.0002609715374056076, "loss": 0.0092, "step": 7430 }, { "epoch": 56.79389312977099, "grad_norm": 0.1558588594198227, "learning_rate": 0.0002608639975941124, "loss": 0.0089, "step": 7440 }, { "epoch": 56.87022900763359, "grad_norm": 0.13958413898944855, "learning_rate": 0.0002607563320505063, "loss": 0.0087, "step": 7450 }, { "epoch": 56.94656488549618, "grad_norm": 0.1325370818376541, "learning_rate": 0.00026064854089689405, "loss": 0.0086, "step": 7460 }, { "epoch": 57.02290076335878, "grad_norm": 0.16879047453403473, "learning_rate": 0.0002605406242555227, "loss": 0.0091, "step": 7470 }, { "epoch": 57.099236641221374, "grad_norm": 0.10713186860084534, "learning_rate": 0.0002604325822487818, "loss": 0.0082, "step": 7480 }, { "epoch": 57.17557251908397, "grad_norm": 0.18872936069965363, "learning_rate": 0.00026032441499920306, "loss": 0.0097, "step": 7490 }, { "epoch": 57.25190839694657, "grad_norm": 0.17626366019248962, "learning_rate": 0.00026021612262946004, "loss": 0.0085, "step": 7500 }, { "epoch": 57.32824427480916, "grad_norm": 0.20342500507831573, "learning_rate": 0.0002601077052623685, "loss": 0.0083, "step": 7510 }, { "epoch": 57.404580152671755, "grad_norm": 0.1067720279097557, "learning_rate": 0.0002599991630208857, "loss": 0.0084, "step": 7520 }, { "epoch": 57.48091603053435, "grad_norm": 0.20952217280864716, "learning_rate": 0.00025989049602811066, "loss": 0.009, "step": 7530 }, { "epoch": 57.55725190839695, "grad_norm": 0.15441496670246124, "learning_rate": 0.0002597817044072838, "loss": 0.0095, "step": 7540 }, { "epoch": 57.63358778625954, "grad_norm": 0.10185621678829193, "learning_rate": 0.0002596727882817869, "loss": 0.0074, "step": 7550 }, { "epoch": 57.70992366412214, "grad_norm": 0.16355207562446594, "learning_rate": 0.0002595637477751431, "loss": 0.008, "step": 7560 }, { "epoch": 57.786259541984734, "grad_norm": 0.14642255008220673, "learning_rate": 0.00025945458301101636, "loss": 0.0082, "step": 7570 }, { "epoch": 57.86259541984733, "grad_norm": 0.19516520202159882, "learning_rate": 0.0002593452941132117, "loss": 0.0083, "step": 7580 }, { "epoch": 57.93893129770992, "grad_norm": 0.15463177859783173, "learning_rate": 0.0002592358812056749, "loss": 0.0088, "step": 7590 }, { "epoch": 58.01526717557252, "grad_norm": 0.14167249202728271, "learning_rate": 0.0002591263444124923, "loss": 0.0079, "step": 7600 }, { "epoch": 58.091603053435115, "grad_norm": 0.12588582932949066, "learning_rate": 0.000259016683857891, "loss": 0.0084, "step": 7610 }, { "epoch": 58.16793893129771, "grad_norm": 0.13345034420490265, "learning_rate": 0.00025890689966623804, "loss": 0.0084, "step": 7620 }, { "epoch": 58.2442748091603, "grad_norm": 0.12898240983486176, "learning_rate": 0.0002587969919620411, "loss": 0.009, "step": 7630 }, { "epoch": 58.3206106870229, "grad_norm": 0.17836321890354156, "learning_rate": 0.0002586869608699476, "loss": 0.0092, "step": 7640 }, { "epoch": 58.396946564885496, "grad_norm": 0.1953604519367218, "learning_rate": 0.0002585768065147452, "loss": 0.0082, "step": 7650 }, { "epoch": 58.47328244274809, "grad_norm": 0.16018278896808624, "learning_rate": 0.00025846652902136106, "loss": 0.0083, "step": 7660 }, { "epoch": 58.54961832061069, "grad_norm": 0.1338210105895996, "learning_rate": 0.00025835612851486217, "loss": 0.0078, "step": 7670 }, { "epoch": 58.62595419847328, "grad_norm": 0.134304478764534, "learning_rate": 0.000258245605120455, "loss": 0.0078, "step": 7680 }, { "epoch": 58.70229007633588, "grad_norm": 0.1631697714328766, "learning_rate": 0.00025813495896348537, "loss": 0.0081, "step": 7690 }, { "epoch": 58.778625954198475, "grad_norm": 0.12952634692192078, "learning_rate": 0.00025802419016943834, "loss": 0.008, "step": 7700 }, { "epoch": 58.85496183206107, "grad_norm": 0.17917485535144806, "learning_rate": 0.0002579132988639381, "loss": 0.0069, "step": 7710 }, { "epoch": 58.93129770992366, "grad_norm": 0.10362667590379715, "learning_rate": 0.0002578022851727477, "loss": 0.0078, "step": 7720 }, { "epoch": 59.00763358778626, "grad_norm": 0.2149238884449005, "learning_rate": 0.00025769114922176894, "loss": 0.0091, "step": 7730 }, { "epoch": 59.083969465648856, "grad_norm": 0.10587195307016373, "learning_rate": 0.0002575798911370425, "loss": 0.0082, "step": 7740 }, { "epoch": 59.16030534351145, "grad_norm": 0.11803413182497025, "learning_rate": 0.0002574685110447472, "loss": 0.0082, "step": 7750 }, { "epoch": 59.23664122137404, "grad_norm": 0.19588199257850647, "learning_rate": 0.00025735700907120076, "loss": 0.0071, "step": 7760 }, { "epoch": 59.31297709923664, "grad_norm": 0.1292734146118164, "learning_rate": 0.0002572453853428586, "loss": 0.0079, "step": 7770 }, { "epoch": 59.38931297709924, "grad_norm": 0.12188327312469482, "learning_rate": 0.0002571336399863146, "loss": 0.0072, "step": 7780 }, { "epoch": 59.465648854961835, "grad_norm": 0.13246265053749084, "learning_rate": 0.0002570217731283003, "loss": 0.0073, "step": 7790 }, { "epoch": 59.541984732824424, "grad_norm": 0.10716478526592255, "learning_rate": 0.0002569097848956852, "loss": 0.0072, "step": 7800 }, { "epoch": 59.61832061068702, "grad_norm": 0.15210285782814026, "learning_rate": 0.00025679767541547655, "loss": 0.0075, "step": 7810 }, { "epoch": 59.69465648854962, "grad_norm": 0.15280434489250183, "learning_rate": 0.00025668544481481875, "loss": 0.0085, "step": 7820 }, { "epoch": 59.770992366412216, "grad_norm": 0.14775653183460236, "learning_rate": 0.000256573093220994, "loss": 0.0084, "step": 7830 }, { "epoch": 59.847328244274806, "grad_norm": 0.14930316805839539, "learning_rate": 0.0002564606207614214, "loss": 0.0078, "step": 7840 }, { "epoch": 59.9236641221374, "grad_norm": 0.10046887397766113, "learning_rate": 0.0002563480275636572, "loss": 0.0079, "step": 7850 }, { "epoch": 60.0, "grad_norm": 0.1417957842350006, "learning_rate": 0.0002562353137553947, "loss": 0.0078, "step": 7860 }, { "epoch": 60.0763358778626, "grad_norm": 0.10676194727420807, "learning_rate": 0.0002561224794644639, "loss": 0.0076, "step": 7870 }, { "epoch": 60.152671755725194, "grad_norm": 0.1361350268125534, "learning_rate": 0.00025600952481883143, "loss": 0.0079, "step": 7880 }, { "epoch": 60.229007633587784, "grad_norm": 0.11955258995294571, "learning_rate": 0.00025589644994660035, "loss": 0.0075, "step": 7890 }, { "epoch": 60.30534351145038, "grad_norm": 0.10744364559650421, "learning_rate": 0.0002557832549760103, "loss": 0.0074, "step": 7900 }, { "epoch": 60.38167938931298, "grad_norm": 0.17523334920406342, "learning_rate": 0.00025566994003543684, "loss": 0.0083, "step": 7910 }, { "epoch": 60.458015267175576, "grad_norm": 0.17258870601654053, "learning_rate": 0.00025555650525339174, "loss": 0.0079, "step": 7920 }, { "epoch": 60.534351145038165, "grad_norm": 0.14048689603805542, "learning_rate": 0.00025544295075852273, "loss": 0.0076, "step": 7930 }, { "epoch": 60.61068702290076, "grad_norm": 0.11171811074018478, "learning_rate": 0.0002553292766796132, "loss": 0.0074, "step": 7940 }, { "epoch": 60.68702290076336, "grad_norm": 0.15726244449615479, "learning_rate": 0.00025521548314558223, "loss": 0.008, "step": 7950 }, { "epoch": 60.76335877862596, "grad_norm": 0.1703062802553177, "learning_rate": 0.00025510157028548435, "loss": 0.0079, "step": 7960 }, { "epoch": 60.83969465648855, "grad_norm": 0.1165282279253006, "learning_rate": 0.0002549875382285094, "loss": 0.0079, "step": 7970 }, { "epoch": 60.916030534351144, "grad_norm": 0.15920227766036987, "learning_rate": 0.00025487338710398247, "loss": 0.0076, "step": 7980 }, { "epoch": 60.99236641221374, "grad_norm": 0.1461462527513504, "learning_rate": 0.00025475911704136354, "loss": 0.0074, "step": 7990 }, { "epoch": 61.06870229007634, "grad_norm": 0.11665527522563934, "learning_rate": 0.00025464472817024773, "loss": 0.0072, "step": 8000 }, { "epoch": 61.14503816793893, "grad_norm": 0.14942559599876404, "learning_rate": 0.0002545302206203646, "loss": 0.0079, "step": 8010 }, { "epoch": 61.221374045801525, "grad_norm": 0.2221251279115677, "learning_rate": 0.0002544155945215785, "loss": 0.0082, "step": 8020 }, { "epoch": 61.29770992366412, "grad_norm": 0.15947479009628296, "learning_rate": 0.00025430085000388813, "loss": 0.0078, "step": 8030 }, { "epoch": 61.37404580152672, "grad_norm": 0.14518491923809052, "learning_rate": 0.00025418598719742676, "loss": 0.008, "step": 8040 }, { "epoch": 61.45038167938931, "grad_norm": 0.12558750808238983, "learning_rate": 0.00025407100623246135, "loss": 0.0076, "step": 8050 }, { "epoch": 61.52671755725191, "grad_norm": 0.15346883237361908, "learning_rate": 0.0002539559072393932, "loss": 0.0082, "step": 8060 }, { "epoch": 61.603053435114504, "grad_norm": 0.15478919446468353, "learning_rate": 0.0002538406903487574, "loss": 0.0075, "step": 8070 }, { "epoch": 61.6793893129771, "grad_norm": 0.13551078736782074, "learning_rate": 0.00025372535569122273, "loss": 0.0081, "step": 8080 }, { "epoch": 61.7557251908397, "grad_norm": 0.13524523377418518, "learning_rate": 0.0002536099033975915, "loss": 0.0079, "step": 8090 }, { "epoch": 61.83206106870229, "grad_norm": 0.16157351434230804, "learning_rate": 0.0002534943335987994, "loss": 0.0078, "step": 8100 }, { "epoch": 61.908396946564885, "grad_norm": 0.16973881423473358, "learning_rate": 0.0002533786464259156, "loss": 0.0075, "step": 8110 }, { "epoch": 61.98473282442748, "grad_norm": 0.07974033802747726, "learning_rate": 0.00025326284201014203, "loss": 0.007, "step": 8120 }, { "epoch": 62.06106870229008, "grad_norm": 0.11081836372613907, "learning_rate": 0.00025314692048281393, "loss": 0.0075, "step": 8130 }, { "epoch": 62.13740458015267, "grad_norm": 0.13641496002674103, "learning_rate": 0.0002530308819753992, "loss": 0.0081, "step": 8140 }, { "epoch": 62.213740458015266, "grad_norm": 0.12006375193595886, "learning_rate": 0.0002529147266194984, "loss": 0.0075, "step": 8150 }, { "epoch": 62.29007633587786, "grad_norm": 0.13716377317905426, "learning_rate": 0.0002527984545468446, "loss": 0.0073, "step": 8160 }, { "epoch": 62.36641221374046, "grad_norm": 0.14244121313095093, "learning_rate": 0.0002526820658893033, "loss": 0.0071, "step": 8170 }, { "epoch": 62.44274809160305, "grad_norm": 0.12264581024646759, "learning_rate": 0.0002525655607788722, "loss": 0.0072, "step": 8180 }, { "epoch": 62.51908396946565, "grad_norm": 0.12790954113006592, "learning_rate": 0.00025244893934768097, "loss": 0.0069, "step": 8190 }, { "epoch": 62.595419847328245, "grad_norm": 0.11812767386436462, "learning_rate": 0.00025233220172799143, "loss": 0.0077, "step": 8200 }, { "epoch": 62.67175572519084, "grad_norm": 0.10782524943351746, "learning_rate": 0.00025221534805219695, "loss": 0.0084, "step": 8210 }, { "epoch": 62.74809160305343, "grad_norm": 0.12704947590827942, "learning_rate": 0.0002520983784528226, "loss": 0.0076, "step": 8220 }, { "epoch": 62.82442748091603, "grad_norm": 0.1217181608080864, "learning_rate": 0.000251981293062525, "loss": 0.0081, "step": 8230 }, { "epoch": 62.900763358778626, "grad_norm": 0.14119642972946167, "learning_rate": 0.0002518640920140919, "loss": 0.0075, "step": 8240 }, { "epoch": 62.97709923664122, "grad_norm": 0.09174049645662308, "learning_rate": 0.0002517467754404424, "loss": 0.0063, "step": 8250 }, { "epoch": 63.05343511450382, "grad_norm": 0.15839794278144836, "learning_rate": 0.0002516293434746265, "loss": 0.0073, "step": 8260 }, { "epoch": 63.12977099236641, "grad_norm": 0.10843265801668167, "learning_rate": 0.0002515117962498252, "loss": 0.0071, "step": 8270 }, { "epoch": 63.20610687022901, "grad_norm": 0.1443076878786087, "learning_rate": 0.0002513941338993501, "loss": 0.007, "step": 8280 }, { "epoch": 63.282442748091604, "grad_norm": 0.12063659727573395, "learning_rate": 0.0002512763565566434, "loss": 0.0071, "step": 8290 }, { "epoch": 63.3587786259542, "grad_norm": 0.09028756618499756, "learning_rate": 0.00025115846435527767, "loss": 0.0066, "step": 8300 }, { "epoch": 63.43511450381679, "grad_norm": 0.07062724977731705, "learning_rate": 0.00025104045742895583, "loss": 0.0072, "step": 8310 }, { "epoch": 63.51145038167939, "grad_norm": 0.21143276989459991, "learning_rate": 0.0002509223359115108, "loss": 0.0085, "step": 8320 }, { "epoch": 63.587786259541986, "grad_norm": 0.0979008674621582, "learning_rate": 0.0002508040999369056, "loss": 0.0072, "step": 8330 }, { "epoch": 63.66412213740458, "grad_norm": 0.14583511650562286, "learning_rate": 0.00025068574963923307, "loss": 0.0078, "step": 8340 }, { "epoch": 63.74045801526717, "grad_norm": 0.1913616806268692, "learning_rate": 0.0002505672851527154, "loss": 0.0081, "step": 8350 }, { "epoch": 63.81679389312977, "grad_norm": 0.17316405475139618, "learning_rate": 0.00025044870661170465, "loss": 0.0074, "step": 8360 }, { "epoch": 63.89312977099237, "grad_norm": 0.16366834938526154, "learning_rate": 0.00025033001415068203, "loss": 0.0069, "step": 8370 }, { "epoch": 63.969465648854964, "grad_norm": 0.12490269541740417, "learning_rate": 0.00025021120790425807, "loss": 0.0073, "step": 8380 }, { "epoch": 64.04580152671755, "grad_norm": 0.11639098823070526, "learning_rate": 0.00025009228800717214, "loss": 0.0073, "step": 8390 }, { "epoch": 64.12213740458016, "grad_norm": 0.14254304766654968, "learning_rate": 0.00024997325459429276, "loss": 0.0087, "step": 8400 }, { "epoch": 64.19847328244275, "grad_norm": 0.08445347845554352, "learning_rate": 0.00024985410780061705, "loss": 0.0075, "step": 8410 }, { "epoch": 64.27480916030534, "grad_norm": 0.12115830928087234, "learning_rate": 0.0002497348477612707, "loss": 0.0067, "step": 8420 }, { "epoch": 64.35114503816794, "grad_norm": 0.13861225545406342, "learning_rate": 0.0002496154746115078, "loss": 0.0072, "step": 8430 }, { "epoch": 64.42748091603053, "grad_norm": 0.10297079384326935, "learning_rate": 0.0002494959884867109, "loss": 0.007, "step": 8440 }, { "epoch": 64.50381679389314, "grad_norm": 0.09681098163127899, "learning_rate": 0.0002493763895223905, "loss": 0.0066, "step": 8450 }, { "epoch": 64.58015267175573, "grad_norm": 0.1353456825017929, "learning_rate": 0.00024925667785418527, "loss": 0.0073, "step": 8460 }, { "epoch": 64.65648854961832, "grad_norm": 0.14676009118556976, "learning_rate": 0.0002491368536178614, "loss": 0.0083, "step": 8470 }, { "epoch": 64.73282442748092, "grad_norm": 0.1663789451122284, "learning_rate": 0.00024901691694931294, "loss": 0.0078, "step": 8480 }, { "epoch": 64.80916030534351, "grad_norm": 0.1637004017829895, "learning_rate": 0.0002488968679845614, "loss": 0.0079, "step": 8490 }, { "epoch": 64.8854961832061, "grad_norm": 0.13884173333644867, "learning_rate": 0.0002487767068597558, "loss": 0.0075, "step": 8500 }, { "epoch": 64.9618320610687, "grad_norm": 0.09634540975093842, "learning_rate": 0.00024865643371117205, "loss": 0.0077, "step": 8510 }, { "epoch": 65.0381679389313, "grad_norm": 0.15140649676322937, "learning_rate": 0.00024853604867521343, "loss": 0.0082, "step": 8520 }, { "epoch": 65.1145038167939, "grad_norm": 0.11218662559986115, "learning_rate": 0.00024841555188840985, "loss": 0.0076, "step": 8530 }, { "epoch": 65.19083969465649, "grad_norm": 0.11331967264413834, "learning_rate": 0.0002482949434874182, "loss": 0.0075, "step": 8540 }, { "epoch": 65.26717557251908, "grad_norm": 0.10223134607076645, "learning_rate": 0.00024817422360902175, "loss": 0.0066, "step": 8550 }, { "epoch": 65.34351145038168, "grad_norm": 0.09511267393827438, "learning_rate": 0.00024805339239013024, "loss": 0.0073, "step": 8560 }, { "epoch": 65.41984732824427, "grad_norm": 0.1313227117061615, "learning_rate": 0.00024793244996777975, "loss": 0.0073, "step": 8570 }, { "epoch": 65.49618320610686, "grad_norm": 0.1288388967514038, "learning_rate": 0.0002478113964791326, "loss": 0.0071, "step": 8580 }, { "epoch": 65.57251908396947, "grad_norm": 0.11828352510929108, "learning_rate": 0.00024769023206147665, "loss": 0.007, "step": 8590 }, { "epoch": 65.64885496183206, "grad_norm": 0.11540471762418747, "learning_rate": 0.000247568956852226, "loss": 0.0073, "step": 8600 }, { "epoch": 65.72519083969466, "grad_norm": 0.22252339124679565, "learning_rate": 0.00024744757098892024, "loss": 0.008, "step": 8610 }, { "epoch": 65.80152671755725, "grad_norm": 0.12820450961589813, "learning_rate": 0.00024732607460922437, "loss": 0.0077, "step": 8620 }, { "epoch": 65.87786259541984, "grad_norm": 0.14160895347595215, "learning_rate": 0.00024720446785092884, "loss": 0.0095, "step": 8630 }, { "epoch": 65.95419847328245, "grad_norm": 0.14882704615592957, "learning_rate": 0.0002470827508519492, "loss": 0.0126, "step": 8640 }, { "epoch": 66.03053435114504, "grad_norm": 0.2030603140592575, "learning_rate": 0.00024696092375032624, "loss": 0.0132, "step": 8650 }, { "epoch": 66.10687022900764, "grad_norm": 0.22969640791416168, "learning_rate": 0.0002468389866842253, "loss": 0.0121, "step": 8660 }, { "epoch": 66.18320610687023, "grad_norm": 0.24858173727989197, "learning_rate": 0.00024671693979193656, "loss": 0.0124, "step": 8670 }, { "epoch": 66.25954198473282, "grad_norm": 0.1559084951877594, "learning_rate": 0.00024659478321187485, "loss": 0.0105, "step": 8680 }, { "epoch": 66.33587786259542, "grad_norm": 0.17520537972450256, "learning_rate": 0.0002464725170825793, "loss": 0.0101, "step": 8690 }, { "epoch": 66.41221374045801, "grad_norm": 0.13466548919677734, "learning_rate": 0.00024635014154271337, "loss": 0.0088, "step": 8700 }, { "epoch": 66.4885496183206, "grad_norm": 0.1535608023405075, "learning_rate": 0.00024622765673106445, "loss": 0.0089, "step": 8710 }, { "epoch": 66.56488549618321, "grad_norm": 0.11714678257703781, "learning_rate": 0.00024610506278654397, "loss": 0.0077, "step": 8720 }, { "epoch": 66.6412213740458, "grad_norm": 0.09966038167476654, "learning_rate": 0.00024598235984818717, "loss": 0.0072, "step": 8730 }, { "epoch": 66.7175572519084, "grad_norm": 0.12011579424142838, "learning_rate": 0.00024585954805515277, "loss": 0.0087, "step": 8740 }, { "epoch": 66.79389312977099, "grad_norm": 0.13791148364543915, "learning_rate": 0.000245736627546723, "loss": 0.0076, "step": 8750 }, { "epoch": 66.87022900763358, "grad_norm": 0.10188236832618713, "learning_rate": 0.0002456135984623034, "loss": 0.0083, "step": 8760 }, { "epoch": 66.94656488549619, "grad_norm": 0.1264120489358902, "learning_rate": 0.00024549046094142274, "loss": 0.0077, "step": 8770 }, { "epoch": 67.02290076335878, "grad_norm": 0.10441657155752182, "learning_rate": 0.0002453672151237326, "loss": 0.0069, "step": 8780 }, { "epoch": 67.09923664122137, "grad_norm": 0.10505234450101852, "learning_rate": 0.0002452438611490074, "loss": 0.0069, "step": 8790 }, { "epoch": 67.17557251908397, "grad_norm": 0.1266278177499771, "learning_rate": 0.0002451203991571444, "loss": 0.0076, "step": 8800 }, { "epoch": 67.25190839694656, "grad_norm": 0.11309097707271576, "learning_rate": 0.0002449968292881631, "loss": 0.0069, "step": 8810 }, { "epoch": 67.32824427480917, "grad_norm": 0.16106292605400085, "learning_rate": 0.00024487315168220553, "loss": 0.0075, "step": 8820 }, { "epoch": 67.40458015267176, "grad_norm": 0.16443900763988495, "learning_rate": 0.0002447493664795359, "loss": 0.0085, "step": 8830 }, { "epoch": 67.48091603053435, "grad_norm": 0.13377918303012848, "learning_rate": 0.0002446254738205404, "loss": 0.0077, "step": 8840 }, { "epoch": 67.55725190839695, "grad_norm": 0.14221568405628204, "learning_rate": 0.0002445014738457271, "loss": 0.007, "step": 8850 }, { "epoch": 67.63358778625954, "grad_norm": 0.16514648497104645, "learning_rate": 0.00024437736669572575, "loss": 0.0079, "step": 8860 }, { "epoch": 67.70992366412214, "grad_norm": 0.11693326383829117, "learning_rate": 0.0002442531525112876, "loss": 0.0066, "step": 8870 }, { "epoch": 67.78625954198473, "grad_norm": 0.09760645776987076, "learning_rate": 0.00024412883143328548, "loss": 0.0069, "step": 8880 }, { "epoch": 67.86259541984732, "grad_norm": 0.10211591422557831, "learning_rate": 0.00024400440360271324, "loss": 0.0066, "step": 8890 }, { "epoch": 67.93893129770993, "grad_norm": 0.11075524240732193, "learning_rate": 0.00024387986916068596, "loss": 0.0073, "step": 8900 }, { "epoch": 68.01526717557252, "grad_norm": 0.11959671974182129, "learning_rate": 0.0002437552282484395, "loss": 0.007, "step": 8910 }, { "epoch": 68.09160305343511, "grad_norm": 0.11660947650671005, "learning_rate": 0.00024363048100733048, "loss": 0.0071, "step": 8920 }, { "epoch": 68.16793893129771, "grad_norm": 0.12005265057086945, "learning_rate": 0.0002435056275788363, "loss": 0.0072, "step": 8930 }, { "epoch": 68.2442748091603, "grad_norm": 0.21102257072925568, "learning_rate": 0.00024338066810455453, "loss": 0.0079, "step": 8940 }, { "epoch": 68.3206106870229, "grad_norm": 0.12384557723999023, "learning_rate": 0.00024325560272620315, "loss": 0.0079, "step": 8950 }, { "epoch": 68.3969465648855, "grad_norm": 0.1804552525281906, "learning_rate": 0.00024313043158562023, "loss": 0.0076, "step": 8960 }, { "epoch": 68.47328244274809, "grad_norm": 0.11951126158237457, "learning_rate": 0.0002430051548247637, "loss": 0.007, "step": 8970 }, { "epoch": 68.54961832061069, "grad_norm": 0.0840354785323143, "learning_rate": 0.0002428797725857115, "loss": 0.0077, "step": 8980 }, { "epoch": 68.62595419847328, "grad_norm": 0.16780419647693634, "learning_rate": 0.00024275428501066085, "loss": 0.0068, "step": 8990 }, { "epoch": 68.70229007633588, "grad_norm": 0.12358251959085464, "learning_rate": 0.00024262869224192876, "loss": 0.0074, "step": 9000 }, { "epoch": 68.77862595419847, "grad_norm": 0.09860498458147049, "learning_rate": 0.00024250299442195138, "loss": 0.0071, "step": 9010 }, { "epoch": 68.85496183206106, "grad_norm": 0.11712931841611862, "learning_rate": 0.00024237719169328402, "loss": 0.0074, "step": 9020 }, { "epoch": 68.93129770992367, "grad_norm": 0.13205499947071075, "learning_rate": 0.00024225128419860098, "loss": 0.0069, "step": 9030 }, { "epoch": 69.00763358778626, "grad_norm": 0.11505349725484848, "learning_rate": 0.00024212527208069537, "loss": 0.008, "step": 9040 }, { "epoch": 69.08396946564885, "grad_norm": 0.12649579346179962, "learning_rate": 0.00024199915548247893, "loss": 0.0071, "step": 9050 }, { "epoch": 69.16030534351145, "grad_norm": 0.1279064118862152, "learning_rate": 0.00024187293454698195, "loss": 0.0077, "step": 9060 }, { "epoch": 69.23664122137404, "grad_norm": 0.12645374238491058, "learning_rate": 0.00024174660941735297, "loss": 0.007, "step": 9070 }, { "epoch": 69.31297709923665, "grad_norm": 0.08574175089597702, "learning_rate": 0.00024162018023685876, "loss": 0.0078, "step": 9080 }, { "epoch": 69.38931297709924, "grad_norm": 0.13259302079677582, "learning_rate": 0.00024149364714888414, "loss": 0.0074, "step": 9090 }, { "epoch": 69.46564885496183, "grad_norm": 0.10573895275592804, "learning_rate": 0.00024136701029693166, "loss": 0.0074, "step": 9100 }, { "epoch": 69.54198473282443, "grad_norm": 0.12685395777225494, "learning_rate": 0.0002412402698246216, "loss": 0.0069, "step": 9110 }, { "epoch": 69.61832061068702, "grad_norm": 0.12513259053230286, "learning_rate": 0.0002411134258756918, "loss": 0.0075, "step": 9120 }, { "epoch": 69.69465648854961, "grad_norm": 0.1311226785182953, "learning_rate": 0.00024098647859399736, "loss": 0.009, "step": 9130 }, { "epoch": 69.77099236641222, "grad_norm": 0.1788034439086914, "learning_rate": 0.0002408594281235107, "loss": 0.009, "step": 9140 }, { "epoch": 69.8473282442748, "grad_norm": 0.09440135210752487, "learning_rate": 0.00024073227460832112, "loss": 0.007, "step": 9150 }, { "epoch": 69.92366412213741, "grad_norm": 0.1342211365699768, "learning_rate": 0.0002406050181926349, "loss": 0.0075, "step": 9160 }, { "epoch": 70.0, "grad_norm": 0.10581208020448685, "learning_rate": 0.00024047765902077492, "loss": 0.0069, "step": 9170 }, { "epoch": 70.07633587786259, "grad_norm": 0.08533486723899841, "learning_rate": 0.00024035019723718068, "loss": 0.0064, "step": 9180 }, { "epoch": 70.1526717557252, "grad_norm": 0.09097129106521606, "learning_rate": 0.00024022263298640806, "loss": 0.0076, "step": 9190 }, { "epoch": 70.22900763358778, "grad_norm": 0.16256183385849, "learning_rate": 0.00024009496641312912, "loss": 0.0075, "step": 9200 }, { "epoch": 70.30534351145039, "grad_norm": 0.12632857263088226, "learning_rate": 0.00023996719766213194, "loss": 0.0077, "step": 9210 }, { "epoch": 70.38167938931298, "grad_norm": 0.14249728620052338, "learning_rate": 0.0002398393268783205, "loss": 0.0071, "step": 9220 }, { "epoch": 70.45801526717557, "grad_norm": 0.11664295941591263, "learning_rate": 0.00023971135420671448, "loss": 0.0069, "step": 9230 }, { "epoch": 70.53435114503817, "grad_norm": 0.08995619416236877, "learning_rate": 0.0002395832797924492, "loss": 0.0065, "step": 9240 }, { "epoch": 70.61068702290076, "grad_norm": 0.06652569770812988, "learning_rate": 0.0002394551037807752, "loss": 0.0064, "step": 9250 }, { "epoch": 70.68702290076335, "grad_norm": 0.09533814340829849, "learning_rate": 0.00023932682631705842, "loss": 0.0066, "step": 9260 }, { "epoch": 70.76335877862596, "grad_norm": 0.12985511124134064, "learning_rate": 0.00023919844754677973, "loss": 0.0071, "step": 9270 }, { "epoch": 70.83969465648855, "grad_norm": 0.1681094467639923, "learning_rate": 0.00023906996761553496, "loss": 0.0074, "step": 9280 }, { "epoch": 70.91603053435115, "grad_norm": 0.1476244181394577, "learning_rate": 0.00023894138666903468, "loss": 0.0081, "step": 9290 }, { "epoch": 70.99236641221374, "grad_norm": 0.09503567963838577, "learning_rate": 0.00023881270485310395, "loss": 0.0082, "step": 9300 }, { "epoch": 71.06870229007633, "grad_norm": 0.14320001006126404, "learning_rate": 0.00023868392231368227, "loss": 0.0066, "step": 9310 }, { "epoch": 71.14503816793894, "grad_norm": 0.09209337085485458, "learning_rate": 0.00023855503919682344, "loss": 0.0069, "step": 9320 }, { "epoch": 71.22137404580153, "grad_norm": 0.12488540261983871, "learning_rate": 0.0002384260556486952, "loss": 0.0076, "step": 9330 }, { "epoch": 71.29770992366412, "grad_norm": 0.13360756635665894, "learning_rate": 0.00023829697181557922, "loss": 0.007, "step": 9340 }, { "epoch": 71.37404580152672, "grad_norm": 0.12585008144378662, "learning_rate": 0.00023816778784387094, "loss": 0.0066, "step": 9350 }, { "epoch": 71.45038167938931, "grad_norm": 0.09323146939277649, "learning_rate": 0.00023803850388007947, "loss": 0.0066, "step": 9360 }, { "epoch": 71.52671755725191, "grad_norm": 0.1230412945151329, "learning_rate": 0.00023790912007082702, "loss": 0.0066, "step": 9370 }, { "epoch": 71.6030534351145, "grad_norm": 0.10536643862724304, "learning_rate": 0.0002377796365628494, "loss": 0.0068, "step": 9380 }, { "epoch": 71.6793893129771, "grad_norm": 0.10268345475196838, "learning_rate": 0.00023765005350299525, "loss": 0.0071, "step": 9390 }, { "epoch": 71.7557251908397, "grad_norm": 0.1162390187382698, "learning_rate": 0.00023752037103822617, "loss": 0.0066, "step": 9400 }, { "epoch": 71.83206106870229, "grad_norm": 0.11484206467866898, "learning_rate": 0.00023739058931561654, "loss": 0.0066, "step": 9410 }, { "epoch": 71.90839694656489, "grad_norm": 0.1431216150522232, "learning_rate": 0.00023726070848235328, "loss": 0.0068, "step": 9420 }, { "epoch": 71.98473282442748, "grad_norm": 0.10196653008460999, "learning_rate": 0.0002371307286857356, "loss": 0.007, "step": 9430 }, { "epoch": 72.06106870229007, "grad_norm": 0.11013298481702805, "learning_rate": 0.0002370006500731752, "loss": 0.0064, "step": 9440 }, { "epoch": 72.13740458015268, "grad_norm": 0.13434675335884094, "learning_rate": 0.00023687047279219567, "loss": 0.0061, "step": 9450 }, { "epoch": 72.21374045801527, "grad_norm": 0.10884860903024673, "learning_rate": 0.00023674019699043252, "loss": 0.0065, "step": 9460 }, { "epoch": 72.29007633587786, "grad_norm": 0.1040448248386383, "learning_rate": 0.0002366098228156329, "loss": 0.0059, "step": 9470 }, { "epoch": 72.36641221374046, "grad_norm": 0.20671886205673218, "learning_rate": 0.0002364793504156558, "loss": 0.0075, "step": 9480 }, { "epoch": 72.44274809160305, "grad_norm": 0.1904536485671997, "learning_rate": 0.00023634877993847133, "loss": 0.0074, "step": 9490 }, { "epoch": 72.51908396946565, "grad_norm": 0.11901413649320602, "learning_rate": 0.00023621811153216104, "loss": 0.0074, "step": 9500 }, { "epoch": 72.59541984732824, "grad_norm": 0.10868986696004868, "learning_rate": 0.00023608734534491733, "loss": 0.0066, "step": 9510 }, { "epoch": 72.67175572519083, "grad_norm": 0.09788993746042252, "learning_rate": 0.00023595648152504366, "loss": 0.0068, "step": 9520 }, { "epoch": 72.74809160305344, "grad_norm": 0.09648136794567108, "learning_rate": 0.0002358255202209542, "loss": 0.0073, "step": 9530 }, { "epoch": 72.82442748091603, "grad_norm": 0.11376792937517166, "learning_rate": 0.0002356944615811736, "loss": 0.0073, "step": 9540 }, { "epoch": 72.90076335877862, "grad_norm": 0.10669957101345062, "learning_rate": 0.00023556330575433696, "loss": 0.0075, "step": 9550 }, { "epoch": 72.97709923664122, "grad_norm": 0.12257901579141617, "learning_rate": 0.00023543205288918957, "loss": 0.0067, "step": 9560 }, { "epoch": 73.05343511450381, "grad_norm": 0.1717688888311386, "learning_rate": 0.0002353007031345868, "loss": 0.0071, "step": 9570 }, { "epoch": 73.12977099236642, "grad_norm": 0.09956348687410355, "learning_rate": 0.00023516925663949383, "loss": 0.007, "step": 9580 }, { "epoch": 73.20610687022901, "grad_norm": 0.10239382088184357, "learning_rate": 0.00023503771355298575, "loss": 0.0066, "step": 9590 }, { "epoch": 73.2824427480916, "grad_norm": 0.09500628709793091, "learning_rate": 0.00023490607402424691, "loss": 0.0069, "step": 9600 }, { "epoch": 73.3587786259542, "grad_norm": 0.09801182150840759, "learning_rate": 0.00023477433820257127, "loss": 0.0058, "step": 9610 }, { "epoch": 73.43511450381679, "grad_norm": 0.0816919282078743, "learning_rate": 0.00023464250623736192, "loss": 0.0069, "step": 9620 }, { "epoch": 73.5114503816794, "grad_norm": 0.1430516391992569, "learning_rate": 0.00023451057827813095, "loss": 0.0071, "step": 9630 }, { "epoch": 73.58778625954199, "grad_norm": 0.13846184313297272, "learning_rate": 0.0002343785544744993, "loss": 0.0064, "step": 9640 }, { "epoch": 73.66412213740458, "grad_norm": 0.10020057111978531, "learning_rate": 0.0002342464349761968, "loss": 0.0072, "step": 9650 }, { "epoch": 73.74045801526718, "grad_norm": 0.3463461697101593, "learning_rate": 0.0002341142199330615, "loss": 0.007, "step": 9660 }, { "epoch": 73.81679389312977, "grad_norm": 0.11461147665977478, "learning_rate": 0.00023398190949504004, "loss": 0.0074, "step": 9670 }, { "epoch": 73.89312977099236, "grad_norm": 0.11157041788101196, "learning_rate": 0.0002338495038121872, "loss": 0.0083, "step": 9680 }, { "epoch": 73.96946564885496, "grad_norm": 0.10317132622003555, "learning_rate": 0.00023371700303466574, "loss": 0.0086, "step": 9690 }, { "epoch": 74.04580152671755, "grad_norm": 0.13943658769130707, "learning_rate": 0.00023358440731274625, "loss": 0.0086, "step": 9700 }, { "epoch": 74.12213740458016, "grad_norm": 0.10233092308044434, "learning_rate": 0.00023345171679680713, "loss": 0.0078, "step": 9710 }, { "epoch": 74.19847328244275, "grad_norm": 0.13965290784835815, "learning_rate": 0.00023331893163733414, "loss": 0.0077, "step": 9720 }, { "epoch": 74.27480916030534, "grad_norm": 0.10408375412225723, "learning_rate": 0.00023318605198492038, "loss": 0.0071, "step": 9730 }, { "epoch": 74.35114503816794, "grad_norm": 0.14331607520580292, "learning_rate": 0.00023305307799026623, "loss": 0.0077, "step": 9740 }, { "epoch": 74.42748091603053, "grad_norm": 0.18536069989204407, "learning_rate": 0.00023292000980417897, "loss": 0.0073, "step": 9750 }, { "epoch": 74.50381679389314, "grad_norm": 0.14269468188285828, "learning_rate": 0.00023278684757757276, "loss": 0.0071, "step": 9760 }, { "epoch": 74.58015267175573, "grad_norm": 0.1237950325012207, "learning_rate": 0.00023265359146146835, "loss": 0.0067, "step": 9770 }, { "epoch": 74.65648854961832, "grad_norm": 0.1006525382399559, "learning_rate": 0.00023252024160699304, "loss": 0.0072, "step": 9780 }, { "epoch": 74.73282442748092, "grad_norm": 0.08319064974784851, "learning_rate": 0.00023238679816538034, "loss": 0.0066, "step": 9790 }, { "epoch": 74.80916030534351, "grad_norm": 0.12816523015499115, "learning_rate": 0.00023225326128797, "loss": 0.0064, "step": 9800 }, { "epoch": 74.8854961832061, "grad_norm": 0.10767759382724762, "learning_rate": 0.00023211963112620775, "loss": 0.0063, "step": 9810 }, { "epoch": 74.9618320610687, "grad_norm": 0.11587604135274887, "learning_rate": 0.00023198590783164496, "loss": 0.007, "step": 9820 }, { "epoch": 75.0381679389313, "grad_norm": 0.08855404704809189, "learning_rate": 0.00023185209155593883, "loss": 0.0066, "step": 9830 }, { "epoch": 75.1145038167939, "grad_norm": 0.11267802864313126, "learning_rate": 0.00023171818245085185, "loss": 0.0074, "step": 9840 }, { "epoch": 75.19083969465649, "grad_norm": 0.11265883594751358, "learning_rate": 0.0002315841806682519, "loss": 0.0063, "step": 9850 }, { "epoch": 75.26717557251908, "grad_norm": 0.11180579662322998, "learning_rate": 0.00023145008636011187, "loss": 0.0066, "step": 9860 }, { "epoch": 75.34351145038168, "grad_norm": 0.34904715418815613, "learning_rate": 0.00023131589967850962, "loss": 0.0067, "step": 9870 }, { "epoch": 75.41984732824427, "grad_norm": 0.10252287238836288, "learning_rate": 0.00023118162077562777, "loss": 0.0062, "step": 9880 }, { "epoch": 75.49618320610686, "grad_norm": 0.2283155769109726, "learning_rate": 0.00023104724980375357, "loss": 0.0064, "step": 9890 }, { "epoch": 75.57251908396947, "grad_norm": 0.08799088001251221, "learning_rate": 0.00023091278691527868, "loss": 0.0063, "step": 9900 }, { "epoch": 75.64885496183206, "grad_norm": 0.2242794632911682, "learning_rate": 0.00023077823226269893, "loss": 0.0068, "step": 9910 }, { "epoch": 75.72519083969466, "grad_norm": 0.109209805727005, "learning_rate": 0.00023064358599861432, "loss": 0.007, "step": 9920 }, { "epoch": 75.80152671755725, "grad_norm": 0.11765850335359573, "learning_rate": 0.00023050884827572861, "loss": 0.0079, "step": 9930 }, { "epoch": 75.87786259541984, "grad_norm": 0.11611426621675491, "learning_rate": 0.00023037401924684946, "loss": 0.0084, "step": 9940 }, { "epoch": 75.95419847328245, "grad_norm": 0.14325708150863647, "learning_rate": 0.000230239099064888, "loss": 0.0074, "step": 9950 }, { "epoch": 76.03053435114504, "grad_norm": 0.13660018146038055, "learning_rate": 0.00023010408788285864, "loss": 0.0092, "step": 9960 }, { "epoch": 76.10687022900764, "grad_norm": 0.14066056907176971, "learning_rate": 0.00022996898585387915, "loss": 0.0079, "step": 9970 }, { "epoch": 76.18320610687023, "grad_norm": 0.10456259548664093, "learning_rate": 0.00022983379313117028, "loss": 0.007, "step": 9980 }, { "epoch": 76.25954198473282, "grad_norm": 0.12047851085662842, "learning_rate": 0.0002296985098680556, "loss": 0.0073, "step": 9990 }, { "epoch": 76.33587786259542, "grad_norm": 0.08484099805355072, "learning_rate": 0.00022956313621796135, "loss": 0.0066, "step": 10000 }, { "epoch": 76.41221374045801, "grad_norm": 0.1270473748445511, "learning_rate": 0.00022942767233441634, "loss": 0.0074, "step": 10010 }, { "epoch": 76.4885496183206, "grad_norm": 0.1563781499862671, "learning_rate": 0.00022929211837105176, "loss": 0.0078, "step": 10020 }, { "epoch": 76.56488549618321, "grad_norm": 0.09306219220161438, "learning_rate": 0.00022915647448160076, "loss": 0.0063, "step": 10030 }, { "epoch": 76.6412213740458, "grad_norm": 0.11702046543359756, "learning_rate": 0.00022902074081989872, "loss": 0.0083, "step": 10040 }, { "epoch": 76.7175572519084, "grad_norm": 0.12244874984025955, "learning_rate": 0.00022888491753988264, "loss": 0.0085, "step": 10050 }, { "epoch": 76.79389312977099, "grad_norm": 0.09440808743238449, "learning_rate": 0.00022874900479559132, "loss": 0.0082, "step": 10060 }, { "epoch": 76.87022900763358, "grad_norm": 0.124620720744133, "learning_rate": 0.00022861300274116484, "loss": 0.0075, "step": 10070 }, { "epoch": 76.94656488549619, "grad_norm": 0.11184998601675034, "learning_rate": 0.0002284769115308447, "loss": 0.0071, "step": 10080 }, { "epoch": 77.02290076335878, "grad_norm": 0.12385670095682144, "learning_rate": 0.00022834073131897353, "loss": 0.0076, "step": 10090 }, { "epoch": 77.09923664122137, "grad_norm": 0.09280102699995041, "learning_rate": 0.00022820446225999474, "loss": 0.0063, "step": 10100 }, { "epoch": 77.17557251908397, "grad_norm": 0.13645803928375244, "learning_rate": 0.00022806810450845273, "loss": 0.0064, "step": 10110 }, { "epoch": 77.25190839694656, "grad_norm": 0.10379781574010849, "learning_rate": 0.00022793165821899223, "loss": 0.0063, "step": 10120 }, { "epoch": 77.32824427480917, "grad_norm": 0.10674098134040833, "learning_rate": 0.00022779512354635862, "loss": 0.0067, "step": 10130 }, { "epoch": 77.40458015267176, "grad_norm": 0.10530492663383484, "learning_rate": 0.00022765850064539742, "loss": 0.0064, "step": 10140 }, { "epoch": 77.48091603053435, "grad_norm": 0.07551411539316177, "learning_rate": 0.0002275217896710541, "loss": 0.006, "step": 10150 }, { "epoch": 77.55725190839695, "grad_norm": 0.07631093263626099, "learning_rate": 0.0002273849907783743, "loss": 0.0062, "step": 10160 }, { "epoch": 77.63358778625954, "grad_norm": 0.09773093461990356, "learning_rate": 0.00022724810412250293, "loss": 0.0061, "step": 10170 }, { "epoch": 77.70992366412214, "grad_norm": 0.18397502601146698, "learning_rate": 0.00022711112985868492, "loss": 0.0068, "step": 10180 }, { "epoch": 77.78625954198473, "grad_norm": 0.1481582224369049, "learning_rate": 0.00022697406814226425, "loss": 0.0075, "step": 10190 }, { "epoch": 77.86259541984732, "grad_norm": 0.1704908311367035, "learning_rate": 0.00022683691912868411, "loss": 0.0077, "step": 10200 }, { "epoch": 77.93893129770993, "grad_norm": 0.1949082314968109, "learning_rate": 0.0002266996829734868, "loss": 0.0077, "step": 10210 }, { "epoch": 78.01526717557252, "grad_norm": 0.16796641051769257, "learning_rate": 0.00022656235983231344, "loss": 0.0075, "step": 10220 }, { "epoch": 78.09160305343511, "grad_norm": 0.13302011787891388, "learning_rate": 0.0002264249498609036, "loss": 0.0067, "step": 10230 }, { "epoch": 78.16793893129771, "grad_norm": 0.15392795205116272, "learning_rate": 0.00022628745321509567, "loss": 0.0063, "step": 10240 }, { "epoch": 78.2442748091603, "grad_norm": 0.08158693462610245, "learning_rate": 0.00022614987005082596, "loss": 0.0068, "step": 10250 }, { "epoch": 78.3206106870229, "grad_norm": 0.08815085142850876, "learning_rate": 0.00022601220052412925, "loss": 0.0068, "step": 10260 }, { "epoch": 78.3969465648855, "grad_norm": 0.08642593026161194, "learning_rate": 0.00022587444479113804, "loss": 0.006, "step": 10270 }, { "epoch": 78.47328244274809, "grad_norm": 0.11750423163175583, "learning_rate": 0.0002257366030080826, "loss": 0.007, "step": 10280 }, { "epoch": 78.54961832061069, "grad_norm": 0.09136500209569931, "learning_rate": 0.00022559867533129092, "loss": 0.0063, "step": 10290 }, { "epoch": 78.62595419847328, "grad_norm": 0.19787956774234772, "learning_rate": 0.00022546066191718835, "loss": 0.0062, "step": 10300 }, { "epoch": 78.70229007633588, "grad_norm": 0.16334418952465057, "learning_rate": 0.0002253225629222974, "loss": 0.0062, "step": 10310 }, { "epoch": 78.77862595419847, "grad_norm": 0.10427317768335342, "learning_rate": 0.00022518437850323778, "loss": 0.0064, "step": 10320 }, { "epoch": 78.85496183206106, "grad_norm": 0.14173778891563416, "learning_rate": 0.0002250461088167259, "loss": 0.0068, "step": 10330 }, { "epoch": 78.93129770992367, "grad_norm": 0.11260110139846802, "learning_rate": 0.00022490775401957504, "loss": 0.0059, "step": 10340 }, { "epoch": 79.00763358778626, "grad_norm": 0.10782694071531296, "learning_rate": 0.00022476931426869496, "loss": 0.006, "step": 10350 }, { "epoch": 79.08396946564885, "grad_norm": 0.07656214386224747, "learning_rate": 0.00022463078972109168, "loss": 0.0058, "step": 10360 }, { "epoch": 79.16030534351145, "grad_norm": 0.11111703515052795, "learning_rate": 0.00022449218053386747, "loss": 0.0059, "step": 10370 }, { "epoch": 79.23664122137404, "grad_norm": 0.07023119181394577, "learning_rate": 0.00022435348686422062, "loss": 0.0062, "step": 10380 }, { "epoch": 79.31297709923665, "grad_norm": 0.08782238513231277, "learning_rate": 0.0002242147088694451, "loss": 0.0064, "step": 10390 }, { "epoch": 79.38931297709924, "grad_norm": 0.10301970690488815, "learning_rate": 0.00022407584670693063, "loss": 0.0058, "step": 10400 }, { "epoch": 79.46564885496183, "grad_norm": 0.09970909357070923, "learning_rate": 0.00022393690053416245, "loss": 0.0068, "step": 10410 }, { "epoch": 79.54198473282443, "grad_norm": 0.12351123243570328, "learning_rate": 0.0002237978705087208, "loss": 0.0059, "step": 10420 }, { "epoch": 79.61832061068702, "grad_norm": 0.08311138302087784, "learning_rate": 0.00022365875678828138, "loss": 0.0056, "step": 10430 }, { "epoch": 79.69465648854961, "grad_norm": 0.10216003656387329, "learning_rate": 0.00022351955953061453, "loss": 0.006, "step": 10440 }, { "epoch": 79.77099236641222, "grad_norm": 0.09009886533021927, "learning_rate": 0.0002233802788935855, "loss": 0.0062, "step": 10450 }, { "epoch": 79.8473282442748, "grad_norm": 0.12239541858434677, "learning_rate": 0.000223240915035154, "loss": 0.006, "step": 10460 }, { "epoch": 79.92366412213741, "grad_norm": 0.0852673351764679, "learning_rate": 0.0002231014681133741, "loss": 0.0057, "step": 10470 }, { "epoch": 80.0, "grad_norm": 0.08977757394313812, "learning_rate": 0.00022296193828639415, "loss": 0.006, "step": 10480 }, { "epoch": 80.07633587786259, "grad_norm": 0.11681103706359863, "learning_rate": 0.00022282232571245656, "loss": 0.0061, "step": 10490 }, { "epoch": 80.1526717557252, "grad_norm": 0.0850805938243866, "learning_rate": 0.0002226826305498975, "loss": 0.0064, "step": 10500 }, { "epoch": 80.22900763358778, "grad_norm": 0.10747648775577545, "learning_rate": 0.00022254285295714683, "loss": 0.0059, "step": 10510 }, { "epoch": 80.30534351145039, "grad_norm": 0.10625876486301422, "learning_rate": 0.00022240299309272786, "loss": 0.0068, "step": 10520 }, { "epoch": 80.38167938931298, "grad_norm": 0.0895535796880722, "learning_rate": 0.00022226305111525726, "loss": 0.0058, "step": 10530 }, { "epoch": 80.45801526717557, "grad_norm": 0.10632060468196869, "learning_rate": 0.00022212302718344485, "loss": 0.0064, "step": 10540 }, { "epoch": 80.53435114503817, "grad_norm": 0.08401358872652054, "learning_rate": 0.00022198292145609328, "loss": 0.0062, "step": 10550 }, { "epoch": 80.61068702290076, "grad_norm": 0.10626497864723206, "learning_rate": 0.00022184273409209813, "loss": 0.0065, "step": 10560 }, { "epoch": 80.68702290076335, "grad_norm": 0.11306319385766983, "learning_rate": 0.00022170246525044733, "loss": 0.0061, "step": 10570 }, { "epoch": 80.76335877862596, "grad_norm": 0.10385667532682419, "learning_rate": 0.0002215621150902215, "loss": 0.0061, "step": 10580 }, { "epoch": 80.83969465648855, "grad_norm": 0.11929717659950256, "learning_rate": 0.00022142168377059325, "loss": 0.0059, "step": 10590 }, { "epoch": 80.91603053435115, "grad_norm": 0.09675750881433487, "learning_rate": 0.00022128117145082737, "loss": 0.0058, "step": 10600 }, { "epoch": 80.99236641221374, "grad_norm": 0.1226838007569313, "learning_rate": 0.00022114057829028042, "loss": 0.006, "step": 10610 }, { "epoch": 81.06870229007633, "grad_norm": 0.1159643903374672, "learning_rate": 0.0002209999044484008, "loss": 0.0056, "step": 10620 }, { "epoch": 81.14503816793894, "grad_norm": 0.09449452906847, "learning_rate": 0.00022085915008472815, "loss": 0.0054, "step": 10630 }, { "epoch": 81.22137404580153, "grad_norm": 0.08593685179948807, "learning_rate": 0.00022071831535889366, "loss": 0.0059, "step": 10640 }, { "epoch": 81.29770992366412, "grad_norm": 0.09778854995965958, "learning_rate": 0.0002205774004306196, "loss": 0.0055, "step": 10650 }, { "epoch": 81.37404580152672, "grad_norm": 0.11133436858654022, "learning_rate": 0.00022043640545971915, "loss": 0.0057, "step": 10660 }, { "epoch": 81.45038167938931, "grad_norm": 0.15398620069026947, "learning_rate": 0.00022029533060609636, "loss": 0.006, "step": 10670 }, { "epoch": 81.52671755725191, "grad_norm": 0.09706629067659378, "learning_rate": 0.00022015417602974573, "loss": 0.0059, "step": 10680 }, { "epoch": 81.6030534351145, "grad_norm": 0.09307242929935455, "learning_rate": 0.00022001294189075225, "loss": 0.0058, "step": 10690 }, { "epoch": 81.6793893129771, "grad_norm": 0.13905352354049683, "learning_rate": 0.00021987162834929123, "loss": 0.0064, "step": 10700 }, { "epoch": 81.7557251908397, "grad_norm": 0.0904931053519249, "learning_rate": 0.0002197302355656279, "loss": 0.0058, "step": 10710 }, { "epoch": 81.83206106870229, "grad_norm": 0.07986491173505783, "learning_rate": 0.0002195887637001174, "loss": 0.0061, "step": 10720 }, { "epoch": 81.90839694656489, "grad_norm": 0.12921205163002014, "learning_rate": 0.00021944721291320465, "loss": 0.0065, "step": 10730 }, { "epoch": 81.98473282442748, "grad_norm": 0.09055145829916, "learning_rate": 0.00021930558336542388, "loss": 0.0064, "step": 10740 }, { "epoch": 82.06106870229007, "grad_norm": 0.10757362842559814, "learning_rate": 0.00021916387521739886, "loss": 0.0067, "step": 10750 }, { "epoch": 82.13740458015268, "grad_norm": 0.16961802542209625, "learning_rate": 0.00021902208862984235, "loss": 0.0067, "step": 10760 }, { "epoch": 82.21374045801527, "grad_norm": 0.12471865862607956, "learning_rate": 0.0002188802237635561, "loss": 0.0072, "step": 10770 }, { "epoch": 82.29007633587786, "grad_norm": 0.09455563127994537, "learning_rate": 0.00021873828077943072, "loss": 0.0063, "step": 10780 }, { "epoch": 82.36641221374046, "grad_norm": 0.12464411556720734, "learning_rate": 0.0002185962598384453, "loss": 0.0067, "step": 10790 }, { "epoch": 82.44274809160305, "grad_norm": 0.10916679352521896, "learning_rate": 0.0002184541611016674, "loss": 0.0068, "step": 10800 }, { "epoch": 82.51908396946565, "grad_norm": 0.12759728729724884, "learning_rate": 0.0002183119847302528, "loss": 0.006, "step": 10810 }, { "epoch": 82.59541984732824, "grad_norm": 0.12059081345796585, "learning_rate": 0.00021816973088544536, "loss": 0.0053, "step": 10820 }, { "epoch": 82.67175572519083, "grad_norm": 0.12541471421718597, "learning_rate": 0.00021802739972857671, "loss": 0.0064, "step": 10830 }, { "epoch": 82.74809160305344, "grad_norm": 0.11338577419519424, "learning_rate": 0.00021788499142106623, "loss": 0.0058, "step": 10840 }, { "epoch": 82.82442748091603, "grad_norm": 0.09921527653932571, "learning_rate": 0.00021774250612442087, "loss": 0.0065, "step": 10850 }, { "epoch": 82.90076335877862, "grad_norm": 0.10148105025291443, "learning_rate": 0.00021759994400023477, "loss": 0.0063, "step": 10860 }, { "epoch": 82.97709923664122, "grad_norm": 0.11092088371515274, "learning_rate": 0.00021745730521018918, "loss": 0.0059, "step": 10870 }, { "epoch": 83.05343511450381, "grad_norm": 0.08844443410634995, "learning_rate": 0.00021731458991605242, "loss": 0.0064, "step": 10880 }, { "epoch": 83.12977099236642, "grad_norm": 0.10405127704143524, "learning_rate": 0.00021717179827967955, "loss": 0.0057, "step": 10890 }, { "epoch": 83.20610687022901, "grad_norm": 0.10621622204780579, "learning_rate": 0.00021702893046301208, "loss": 0.0052, "step": 10900 }, { "epoch": 83.2824427480916, "grad_norm": 0.15451808273792267, "learning_rate": 0.00021688598662807814, "loss": 0.0069, "step": 10910 }, { "epoch": 83.3587786259542, "grad_norm": 0.08546767383813858, "learning_rate": 0.0002167429669369918, "loss": 0.007, "step": 10920 }, { "epoch": 83.43511450381679, "grad_norm": 0.17478047311306, "learning_rate": 0.00021659987155195343, "loss": 0.0064, "step": 10930 }, { "epoch": 83.5114503816794, "grad_norm": 0.08732287585735321, "learning_rate": 0.00021645670063524905, "loss": 0.0064, "step": 10940 }, { "epoch": 83.58778625954199, "grad_norm": 0.10999668389558792, "learning_rate": 0.00021631345434925046, "loss": 0.0067, "step": 10950 }, { "epoch": 83.66412213740458, "grad_norm": 0.10966475307941437, "learning_rate": 0.00021617013285641482, "loss": 0.0065, "step": 10960 }, { "epoch": 83.74045801526718, "grad_norm": 0.09313569217920303, "learning_rate": 0.00021602673631928475, "loss": 0.0062, "step": 10970 }, { "epoch": 83.81679389312977, "grad_norm": 0.12145984917879105, "learning_rate": 0.0002158832649004878, "loss": 0.007, "step": 10980 }, { "epoch": 83.89312977099236, "grad_norm": 0.13802117109298706, "learning_rate": 0.00021573971876273656, "loss": 0.006, "step": 10990 }, { "epoch": 83.96946564885496, "grad_norm": 0.13931679725646973, "learning_rate": 0.00021559609806882831, "loss": 0.0062, "step": 11000 }, { "epoch": 84.04580152671755, "grad_norm": 0.07513763755559921, "learning_rate": 0.0002154524029816449, "loss": 0.0058, "step": 11010 }, { "epoch": 84.12213740458016, "grad_norm": 0.17049621045589447, "learning_rate": 0.00021530863366415258, "loss": 0.0064, "step": 11020 }, { "epoch": 84.19847328244275, "grad_norm": 0.10217233747243881, "learning_rate": 0.0002151647902794017, "loss": 0.0057, "step": 11030 }, { "epoch": 84.27480916030534, "grad_norm": 0.15845787525177002, "learning_rate": 0.00021502087299052675, "loss": 0.0063, "step": 11040 }, { "epoch": 84.35114503816794, "grad_norm": 0.11185991764068604, "learning_rate": 0.00021487688196074583, "loss": 0.0065, "step": 11050 }, { "epoch": 84.42748091603053, "grad_norm": 0.10472971946001053, "learning_rate": 0.00021473281735336091, "loss": 0.0059, "step": 11060 }, { "epoch": 84.50381679389314, "grad_norm": 0.09645886719226837, "learning_rate": 0.00021458867933175726, "loss": 0.0059, "step": 11070 }, { "epoch": 84.58015267175573, "grad_norm": 0.12530089914798737, "learning_rate": 0.0002144444680594034, "loss": 0.0057, "step": 11080 }, { "epoch": 84.65648854961832, "grad_norm": 0.09190116077661514, "learning_rate": 0.00021430018369985096, "loss": 0.0058, "step": 11090 }, { "epoch": 84.73282442748092, "grad_norm": 0.10703671723604202, "learning_rate": 0.00021415582641673452, "loss": 0.0061, "step": 11100 }, { "epoch": 84.80916030534351, "grad_norm": 0.08808312565088272, "learning_rate": 0.00021401139637377124, "loss": 0.0059, "step": 11110 }, { "epoch": 84.8854961832061, "grad_norm": 0.09161626547574997, "learning_rate": 0.00021386689373476087, "loss": 0.0058, "step": 11120 }, { "epoch": 84.9618320610687, "grad_norm": 0.15287847816944122, "learning_rate": 0.00021372231866358555, "loss": 0.0061, "step": 11130 }, { "epoch": 85.0381679389313, "grad_norm": 0.10912524908781052, "learning_rate": 0.00021357767132420942, "loss": 0.006, "step": 11140 }, { "epoch": 85.1145038167939, "grad_norm": 0.12298277020454407, "learning_rate": 0.0002134329518806787, "loss": 0.0059, "step": 11150 }, { "epoch": 85.19083969465649, "grad_norm": 0.09772927314043045, "learning_rate": 0.0002132881604971213, "loss": 0.0057, "step": 11160 }, { "epoch": 85.26717557251908, "grad_norm": 0.11979064345359802, "learning_rate": 0.0002131432973377468, "loss": 0.006, "step": 11170 }, { "epoch": 85.34351145038168, "grad_norm": 0.11322556436061859, "learning_rate": 0.00021299836256684617, "loss": 0.0059, "step": 11180 }, { "epoch": 85.41984732824427, "grad_norm": 0.10876648128032684, "learning_rate": 0.0002128533563487915, "loss": 0.0063, "step": 11190 }, { "epoch": 85.49618320610686, "grad_norm": 0.09072581678628922, "learning_rate": 0.00021270827884803606, "loss": 0.0055, "step": 11200 }, { "epoch": 85.57251908396947, "grad_norm": 0.09980268776416779, "learning_rate": 0.00021256313022911382, "loss": 0.0069, "step": 11210 }, { "epoch": 85.64885496183206, "grad_norm": 0.14092281460762024, "learning_rate": 0.0002124179106566395, "loss": 0.0093, "step": 11220 }, { "epoch": 85.72519083969466, "grad_norm": 0.23660187423229218, "learning_rate": 0.0002122726202953082, "loss": 0.0089, "step": 11230 }, { "epoch": 85.80152671755725, "grad_norm": 0.19298437237739563, "learning_rate": 0.00021212725930989547, "loss": 0.0121, "step": 11240 }, { "epoch": 85.87786259541984, "grad_norm": 0.19646772742271423, "learning_rate": 0.0002119818278652567, "loss": 0.0102, "step": 11250 }, { "epoch": 85.95419847328245, "grad_norm": 0.1100725531578064, "learning_rate": 0.00021183632612632744, "loss": 0.0089, "step": 11260 }, { "epoch": 86.03053435114504, "grad_norm": 0.12599967420101166, "learning_rate": 0.00021169075425812284, "loss": 0.0076, "step": 11270 }, { "epoch": 86.10687022900764, "grad_norm": 0.08668056130409241, "learning_rate": 0.00021154511242573755, "loss": 0.0073, "step": 11280 }, { "epoch": 86.18320610687023, "grad_norm": 0.22840383648872375, "learning_rate": 0.00021139940079434568, "loss": 0.008, "step": 11290 }, { "epoch": 86.25954198473282, "grad_norm": 0.14157161116600037, "learning_rate": 0.00021125361952920038, "loss": 0.0062, "step": 11300 }, { "epoch": 86.33587786259542, "grad_norm": 0.08708589524030685, "learning_rate": 0.00021110776879563388, "loss": 0.0065, "step": 11310 }, { "epoch": 86.41221374045801, "grad_norm": 0.10831278562545776, "learning_rate": 0.00021096184875905708, "loss": 0.0064, "step": 11320 }, { "epoch": 86.4885496183206, "grad_norm": 0.14114585518836975, "learning_rate": 0.0002108158595849596, "loss": 0.0074, "step": 11330 }, { "epoch": 86.56488549618321, "grad_norm": 0.08164858818054199, "learning_rate": 0.00021066980143890935, "loss": 0.0068, "step": 11340 }, { "epoch": 86.6412213740458, "grad_norm": 0.1258406788110733, "learning_rate": 0.00021052367448655258, "loss": 0.0067, "step": 11350 }, { "epoch": 86.7175572519084, "grad_norm": 0.11088953167200089, "learning_rate": 0.00021037747889361347, "loss": 0.0062, "step": 11360 }, { "epoch": 86.79389312977099, "grad_norm": 0.12168974429368973, "learning_rate": 0.00021023121482589412, "loss": 0.0067, "step": 11370 }, { "epoch": 86.87022900763358, "grad_norm": 0.11877045780420303, "learning_rate": 0.00021008488244927425, "loss": 0.0068, "step": 11380 }, { "epoch": 86.94656488549619, "grad_norm": 0.12265896052122116, "learning_rate": 0.0002099384819297111, "loss": 0.0066, "step": 11390 }, { "epoch": 87.02290076335878, "grad_norm": 0.10145389288663864, "learning_rate": 0.00020979201343323907, "loss": 0.0064, "step": 11400 }, { "epoch": 87.09923664122137, "grad_norm": 0.15085172653198242, "learning_rate": 0.00020964547712596979, "loss": 0.0065, "step": 11410 }, { "epoch": 87.17557251908397, "grad_norm": 0.08728236705064774, "learning_rate": 0.00020949887317409175, "loss": 0.0059, "step": 11420 }, { "epoch": 87.25190839694656, "grad_norm": 0.10847751796245575, "learning_rate": 0.0002093522017438701, "loss": 0.0059, "step": 11430 }, { "epoch": 87.32824427480917, "grad_norm": 0.11960544437170029, "learning_rate": 0.00020920546300164658, "loss": 0.0057, "step": 11440 }, { "epoch": 87.40458015267176, "grad_norm": 0.09862975031137466, "learning_rate": 0.00020905865711383934, "loss": 0.0057, "step": 11450 }, { "epoch": 87.48091603053435, "grad_norm": 0.08974260091781616, "learning_rate": 0.00020891178424694242, "loss": 0.006, "step": 11460 }, { "epoch": 87.55725190839695, "grad_norm": 0.1050775796175003, "learning_rate": 0.00020876484456752614, "loss": 0.0066, "step": 11470 }, { "epoch": 87.63358778625954, "grad_norm": 0.11310622841119766, "learning_rate": 0.0002086178382422364, "loss": 0.006, "step": 11480 }, { "epoch": 87.70992366412214, "grad_norm": 0.0920935794711113, "learning_rate": 0.0002084707654377947, "loss": 0.0059, "step": 11490 }, { "epoch": 87.78625954198473, "grad_norm": 0.10890083014965057, "learning_rate": 0.0002083236263209981, "loss": 0.0051, "step": 11500 }, { "epoch": 87.86259541984732, "grad_norm": 0.13331131637096405, "learning_rate": 0.00020817642105871857, "loss": 0.0058, "step": 11510 }, { "epoch": 87.93893129770993, "grad_norm": 0.09147053956985474, "learning_rate": 0.00020802914981790332, "loss": 0.0057, "step": 11520 }, { "epoch": 88.01526717557252, "grad_norm": 0.13239166140556335, "learning_rate": 0.00020788181276557436, "loss": 0.006, "step": 11530 }, { "epoch": 88.09160305343511, "grad_norm": 0.1031903624534607, "learning_rate": 0.00020773441006882823, "loss": 0.0056, "step": 11540 }, { "epoch": 88.16793893129771, "grad_norm": 0.08591253310441971, "learning_rate": 0.00020758694189483608, "loss": 0.0051, "step": 11550 }, { "epoch": 88.2442748091603, "grad_norm": 0.13625198602676392, "learning_rate": 0.0002074394084108432, "loss": 0.0056, "step": 11560 }, { "epoch": 88.3206106870229, "grad_norm": 0.12332779914140701, "learning_rate": 0.00020729180978416888, "loss": 0.0059, "step": 11570 }, { "epoch": 88.3969465648855, "grad_norm": 0.11999931186437607, "learning_rate": 0.0002071441461822066, "loss": 0.0059, "step": 11580 }, { "epoch": 88.47328244274809, "grad_norm": 0.10281587392091751, "learning_rate": 0.00020699641777242308, "loss": 0.0059, "step": 11590 }, { "epoch": 88.54961832061069, "grad_norm": 0.0972411036491394, "learning_rate": 0.00020684862472235895, "loss": 0.0056, "step": 11600 }, { "epoch": 88.62595419847328, "grad_norm": 0.11088306456804276, "learning_rate": 0.00020670076719962795, "loss": 0.0057, "step": 11610 }, { "epoch": 88.70229007633588, "grad_norm": 0.11124849319458008, "learning_rate": 0.00020655284537191688, "loss": 0.0054, "step": 11620 }, { "epoch": 88.77862595419847, "grad_norm": 0.10634355247020721, "learning_rate": 0.0002064048594069856, "loss": 0.0055, "step": 11630 }, { "epoch": 88.85496183206106, "grad_norm": 0.08958704769611359, "learning_rate": 0.00020625680947266672, "loss": 0.0055, "step": 11640 }, { "epoch": 88.93129770992367, "grad_norm": 0.12612074613571167, "learning_rate": 0.00020610869573686524, "loss": 0.0061, "step": 11650 }, { "epoch": 89.00763358778626, "grad_norm": 0.11995506286621094, "learning_rate": 0.0002059605183675587, "loss": 0.006, "step": 11660 }, { "epoch": 89.08396946564885, "grad_norm": 0.09883546084165573, "learning_rate": 0.00020581227753279664, "loss": 0.0057, "step": 11670 }, { "epoch": 89.16030534351145, "grad_norm": 0.19075235724449158, "learning_rate": 0.00020566397340070078, "loss": 0.0066, "step": 11680 }, { "epoch": 89.23664122137404, "grad_norm": 0.13648636639118195, "learning_rate": 0.00020551560613946444, "loss": 0.0065, "step": 11690 }, { "epoch": 89.31297709923665, "grad_norm": 0.09704729169607162, "learning_rate": 0.0002053671759173526, "loss": 0.0074, "step": 11700 }, { "epoch": 89.38931297709924, "grad_norm": 0.12507158517837524, "learning_rate": 0.0002052186829027017, "loss": 0.0072, "step": 11710 }, { "epoch": 89.46564885496183, "grad_norm": 0.10420115292072296, "learning_rate": 0.00020507012726391928, "loss": 0.0068, "step": 11720 }, { "epoch": 89.54198473282443, "grad_norm": 0.1240818127989769, "learning_rate": 0.00020492150916948397, "loss": 0.0069, "step": 11730 }, { "epoch": 89.61832061068702, "grad_norm": 0.09022738039493561, "learning_rate": 0.00020477282878794532, "loss": 0.006, "step": 11740 }, { "epoch": 89.69465648854961, "grad_norm": 0.1307733952999115, "learning_rate": 0.00020462408628792332, "loss": 0.0057, "step": 11750 }, { "epoch": 89.77099236641222, "grad_norm": 0.1689925193786621, "learning_rate": 0.00020447528183810857, "loss": 0.0058, "step": 11760 }, { "epoch": 89.8473282442748, "grad_norm": 0.09602520614862442, "learning_rate": 0.0002043264156072619, "loss": 0.0059, "step": 11770 }, { "epoch": 89.92366412213741, "grad_norm": 0.09149222820997238, "learning_rate": 0.00020417748776421408, "loss": 0.0054, "step": 11780 }, { "epoch": 90.0, "grad_norm": 0.07817134261131287, "learning_rate": 0.00020402849847786598, "loss": 0.0057, "step": 11790 }, { "epoch": 90.07633587786259, "grad_norm": 0.09808716177940369, "learning_rate": 0.00020387944791718795, "loss": 0.0051, "step": 11800 }, { "epoch": 90.1526717557252, "grad_norm": 0.12219888716936111, "learning_rate": 0.00020373033625121996, "loss": 0.0052, "step": 11810 }, { "epoch": 90.22900763358778, "grad_norm": 0.12695854902267456, "learning_rate": 0.00020358116364907125, "loss": 0.0056, "step": 11820 }, { "epoch": 90.30534351145039, "grad_norm": 0.1263623982667923, "learning_rate": 0.00020343193027992006, "loss": 0.0053, "step": 11830 }, { "epoch": 90.38167938931298, "grad_norm": 0.08486003428697586, "learning_rate": 0.0002032826363130137, "loss": 0.0055, "step": 11840 }, { "epoch": 90.45801526717557, "grad_norm": 0.11597206443548203, "learning_rate": 0.00020313328191766815, "loss": 0.0055, "step": 11850 }, { "epoch": 90.53435114503817, "grad_norm": 0.12867087125778198, "learning_rate": 0.0002029838672632679, "loss": 0.0058, "step": 11860 }, { "epoch": 90.61068702290076, "grad_norm": 0.08184795081615448, "learning_rate": 0.0002028343925192658, "loss": 0.0052, "step": 11870 }, { "epoch": 90.68702290076335, "grad_norm": 0.10319215804338455, "learning_rate": 0.00020268485785518287, "loss": 0.0058, "step": 11880 }, { "epoch": 90.76335877862596, "grad_norm": 0.11330152302980423, "learning_rate": 0.00020253526344060798, "loss": 0.0051, "step": 11890 }, { "epoch": 90.83969465648855, "grad_norm": 0.14841118454933167, "learning_rate": 0.00020238560944519794, "loss": 0.0054, "step": 11900 }, { "epoch": 90.91603053435115, "grad_norm": 0.11827138811349869, "learning_rate": 0.00020223589603867698, "loss": 0.0055, "step": 11910 }, { "epoch": 90.99236641221374, "grad_norm": 0.07645179331302643, "learning_rate": 0.00020208612339083678, "loss": 0.005, "step": 11920 }, { "epoch": 91.06870229007633, "grad_norm": 0.08791926503181458, "learning_rate": 0.00020193629167153623, "loss": 0.0049, "step": 11930 }, { "epoch": 91.14503816793894, "grad_norm": 0.07964352518320084, "learning_rate": 0.00020178640105070103, "loss": 0.0059, "step": 11940 }, { "epoch": 91.22137404580153, "grad_norm": 0.10535834729671478, "learning_rate": 0.000201636451698324, "loss": 0.0054, "step": 11950 }, { "epoch": 91.29770992366412, "grad_norm": 0.09553395211696625, "learning_rate": 0.00020148644378446432, "loss": 0.0049, "step": 11960 }, { "epoch": 91.37404580152672, "grad_norm": 0.09020697325468063, "learning_rate": 0.00020133637747924763, "loss": 0.0057, "step": 11970 }, { "epoch": 91.45038167938931, "grad_norm": 0.08033432066440582, "learning_rate": 0.00020118625295286583, "loss": 0.005, "step": 11980 }, { "epoch": 91.52671755725191, "grad_norm": 0.07954462617635727, "learning_rate": 0.0002010360703755769, "loss": 0.0051, "step": 11990 }, { "epoch": 91.6030534351145, "grad_norm": 0.08556050807237625, "learning_rate": 0.0002008858299177045, "loss": 0.0047, "step": 12000 }, { "epoch": 91.6793893129771, "grad_norm": 0.11900793015956879, "learning_rate": 0.00020073553174963815, "loss": 0.0054, "step": 12010 }, { "epoch": 91.7557251908397, "grad_norm": 0.09307467937469482, "learning_rate": 0.0002005851760418326, "loss": 0.0053, "step": 12020 }, { "epoch": 91.83206106870229, "grad_norm": 0.10792802274227142, "learning_rate": 0.000200434762964808, "loss": 0.0056, "step": 12030 }, { "epoch": 91.90839694656489, "grad_norm": 0.11820261925458908, "learning_rate": 0.00020028429268914946, "loss": 0.005, "step": 12040 }, { "epoch": 91.98473282442748, "grad_norm": 0.08214404433965683, "learning_rate": 0.00020013376538550708, "loss": 0.0053, "step": 12050 }, { "epoch": 92.06106870229007, "grad_norm": 0.11356298625469208, "learning_rate": 0.0001999831812245956, "loss": 0.0055, "step": 12060 }, { "epoch": 92.13740458015268, "grad_norm": 0.12504775822162628, "learning_rate": 0.00019983254037719413, "loss": 0.0058, "step": 12070 }, { "epoch": 92.21374045801527, "grad_norm": 0.08103538304567337, "learning_rate": 0.0001996818430141462, "loss": 0.005, "step": 12080 }, { "epoch": 92.29007633587786, "grad_norm": 0.09302625805139542, "learning_rate": 0.00019953108930635937, "loss": 0.0051, "step": 12090 }, { "epoch": 92.36641221374046, "grad_norm": 0.08917225897312164, "learning_rate": 0.00019938027942480516, "loss": 0.0047, "step": 12100 }, { "epoch": 92.44274809160305, "grad_norm": 0.0824621394276619, "learning_rate": 0.00019922941354051874, "loss": 0.0048, "step": 12110 }, { "epoch": 92.51908396946565, "grad_norm": 0.09149263799190521, "learning_rate": 0.0001990784918245988, "loss": 0.005, "step": 12120 }, { "epoch": 92.59541984732824, "grad_norm": 0.09338316321372986, "learning_rate": 0.00019892751444820737, "loss": 0.005, "step": 12130 }, { "epoch": 92.67175572519083, "grad_norm": 0.1277763694524765, "learning_rate": 0.00019877648158256965, "loss": 0.0054, "step": 12140 }, { "epoch": 92.74809160305344, "grad_norm": 0.07491037249565125, "learning_rate": 0.0001986253933989736, "loss": 0.0053, "step": 12150 }, { "epoch": 92.82442748091603, "grad_norm": 0.09429208189249039, "learning_rate": 0.00019847425006877008, "loss": 0.0057, "step": 12160 }, { "epoch": 92.90076335877862, "grad_norm": 0.10837208479642868, "learning_rate": 0.00019832305176337256, "loss": 0.0057, "step": 12170 }, { "epoch": 92.97709923664122, "grad_norm": 0.09884150326251984, "learning_rate": 0.00019817179865425658, "loss": 0.0055, "step": 12180 }, { "epoch": 93.05343511450381, "grad_norm": 0.10374869406223297, "learning_rate": 0.00019802049091296008, "loss": 0.0055, "step": 12190 }, { "epoch": 93.12977099236642, "grad_norm": 0.09618736803531647, "learning_rate": 0.00019786912871108285, "loss": 0.0051, "step": 12200 }, { "epoch": 93.20610687022901, "grad_norm": 0.11668981611728668, "learning_rate": 0.00019771771222028652, "loss": 0.0054, "step": 12210 }, { "epoch": 93.2824427480916, "grad_norm": 0.1342141181230545, "learning_rate": 0.00019756624161229423, "loss": 0.0055, "step": 12220 }, { "epoch": 93.3587786259542, "grad_norm": 0.10235777497291565, "learning_rate": 0.0001974147170588904, "loss": 0.0056, "step": 12230 }, { "epoch": 93.43511450381679, "grad_norm": 0.112867072224617, "learning_rate": 0.00019726313873192087, "loss": 0.0058, "step": 12240 }, { "epoch": 93.5114503816794, "grad_norm": 0.12745636701583862, "learning_rate": 0.00019711150680329232, "loss": 0.0058, "step": 12250 }, { "epoch": 93.58778625954199, "grad_norm": 0.09834025800228119, "learning_rate": 0.00019695982144497215, "loss": 0.0063, "step": 12260 }, { "epoch": 93.66412213740458, "grad_norm": 0.058787744492292404, "learning_rate": 0.0001968080828289885, "loss": 0.006, "step": 12270 }, { "epoch": 93.74045801526718, "grad_norm": 0.10420133173465729, "learning_rate": 0.0001966562911274298, "loss": 0.006, "step": 12280 }, { "epoch": 93.81679389312977, "grad_norm": 0.09973949193954468, "learning_rate": 0.00019650444651244476, "loss": 0.006, "step": 12290 }, { "epoch": 93.89312977099236, "grad_norm": 0.19366873800754547, "learning_rate": 0.0001963525491562421, "loss": 0.0067, "step": 12300 }, { "epoch": 93.96946564885496, "grad_norm": 0.12973175942897797, "learning_rate": 0.00019620059923109028, "loss": 0.0064, "step": 12310 }, { "epoch": 94.04580152671755, "grad_norm": 0.12365761399269104, "learning_rate": 0.00019604859690931743, "loss": 0.0056, "step": 12320 }, { "epoch": 94.12213740458016, "grad_norm": 0.11434603482484818, "learning_rate": 0.00019589654236331117, "loss": 0.006, "step": 12330 }, { "epoch": 94.19847328244275, "grad_norm": 0.14723211526870728, "learning_rate": 0.00019574443576551813, "loss": 0.0059, "step": 12340 }, { "epoch": 94.27480916030534, "grad_norm": 0.14706934988498688, "learning_rate": 0.00019559227728844423, "loss": 0.0063, "step": 12350 }, { "epoch": 94.35114503816794, "grad_norm": 0.13763180375099182, "learning_rate": 0.00019544006710465415, "loss": 0.0065, "step": 12360 }, { "epoch": 94.42748091603053, "grad_norm": 0.11383082717657089, "learning_rate": 0.00019528780538677107, "loss": 0.0056, "step": 12370 }, { "epoch": 94.50381679389314, "grad_norm": 0.08900892734527588, "learning_rate": 0.00019513549230747683, "loss": 0.0059, "step": 12380 }, { "epoch": 94.58015267175573, "grad_norm": 0.08069955557584763, "learning_rate": 0.0001949831280395113, "loss": 0.0055, "step": 12390 }, { "epoch": 94.65648854961832, "grad_norm": 0.1035219207406044, "learning_rate": 0.00019483071275567252, "loss": 0.0052, "step": 12400 }, { "epoch": 94.73282442748092, "grad_norm": 0.08631502836942673, "learning_rate": 0.00019467824662881654, "loss": 0.0051, "step": 12410 }, { "epoch": 94.80916030534351, "grad_norm": 0.09132368862628937, "learning_rate": 0.00019452572983185673, "loss": 0.0054, "step": 12420 }, { "epoch": 94.8854961832061, "grad_norm": 0.08428333699703217, "learning_rate": 0.00019437316253776422, "loss": 0.0055, "step": 12430 }, { "epoch": 94.9618320610687, "grad_norm": 0.1992415189743042, "learning_rate": 0.0001942205449195673, "loss": 0.0069, "step": 12440 }, { "epoch": 95.0381679389313, "grad_norm": 0.10432060807943344, "learning_rate": 0.0001940678771503513, "loss": 0.006, "step": 12450 }, { "epoch": 95.1145038167939, "grad_norm": 0.08059830218553543, "learning_rate": 0.00019391515940325843, "loss": 0.0059, "step": 12460 }, { "epoch": 95.19083969465649, "grad_norm": 0.09467202425003052, "learning_rate": 0.00019376239185148767, "loss": 0.0053, "step": 12470 }, { "epoch": 95.26717557251908, "grad_norm": 0.10419633984565735, "learning_rate": 0.0001936095746682944, "loss": 0.0054, "step": 12480 }, { "epoch": 95.34351145038168, "grad_norm": 0.09132517874240875, "learning_rate": 0.0001934567080269903, "loss": 0.0058, "step": 12490 }, { "epoch": 95.41984732824427, "grad_norm": 0.09152195602655411, "learning_rate": 0.00019330379210094313, "loss": 0.0051, "step": 12500 }, { "epoch": 95.49618320610686, "grad_norm": 0.08304169774055481, "learning_rate": 0.00019315082706357665, "loss": 0.0053, "step": 12510 }, { "epoch": 95.57251908396947, "grad_norm": 0.09586511552333832, "learning_rate": 0.00019299781308837016, "loss": 0.0055, "step": 12520 }, { "epoch": 95.64885496183206, "grad_norm": 0.12610852718353271, "learning_rate": 0.00019284475034885858, "loss": 0.006, "step": 12530 }, { "epoch": 95.72519083969466, "grad_norm": 0.10778668522834778, "learning_rate": 0.0001926916390186321, "loss": 0.0053, "step": 12540 }, { "epoch": 95.80152671755725, "grad_norm": 0.08345305919647217, "learning_rate": 0.00019253847927133593, "loss": 0.0055, "step": 12550 }, { "epoch": 95.87786259541984, "grad_norm": 0.085738405585289, "learning_rate": 0.00019238527128067034, "loss": 0.0049, "step": 12560 }, { "epoch": 95.95419847328245, "grad_norm": 0.08968999236822128, "learning_rate": 0.00019223201522039027, "loss": 0.0059, "step": 12570 }, { "epoch": 96.03053435114504, "grad_norm": 0.09141431003808975, "learning_rate": 0.000192078711264305, "loss": 0.0053, "step": 12580 }, { "epoch": 96.10687022900764, "grad_norm": 0.12021718919277191, "learning_rate": 0.00019192535958627844, "loss": 0.0057, "step": 12590 }, { "epoch": 96.18320610687023, "grad_norm": 0.1008194163441658, "learning_rate": 0.00019177196036022838, "loss": 0.0054, "step": 12600 }, { "epoch": 96.25954198473282, "grad_norm": 0.07361122220754623, "learning_rate": 0.00019161851376012663, "loss": 0.0055, "step": 12610 }, { "epoch": 96.33587786259542, "grad_norm": 0.06284360587596893, "learning_rate": 0.00019146501995999873, "loss": 0.0058, "step": 12620 }, { "epoch": 96.41221374045801, "grad_norm": 0.09501846134662628, "learning_rate": 0.00019131147913392367, "loss": 0.005, "step": 12630 }, { "epoch": 96.4885496183206, "grad_norm": 0.09642339497804642, "learning_rate": 0.0001911578914560339, "loss": 0.0049, "step": 12640 }, { "epoch": 96.56488549618321, "grad_norm": 0.10918520390987396, "learning_rate": 0.0001910042571005149, "loss": 0.0061, "step": 12650 }, { "epoch": 96.6412213740458, "grad_norm": 0.19244182109832764, "learning_rate": 0.0001908505762416052, "loss": 0.0054, "step": 12660 }, { "epoch": 96.7175572519084, "grad_norm": 0.10888776183128357, "learning_rate": 0.00019069684905359584, "loss": 0.0051, "step": 12670 }, { "epoch": 96.79389312977099, "grad_norm": 0.09605462104082108, "learning_rate": 0.0001905430757108307, "loss": 0.0051, "step": 12680 }, { "epoch": 96.87022900763358, "grad_norm": 0.09049655497074127, "learning_rate": 0.00019038925638770583, "loss": 0.0051, "step": 12690 }, { "epoch": 96.94656488549619, "grad_norm": 0.08315491676330566, "learning_rate": 0.00019023539125866942, "loss": 0.005, "step": 12700 }, { "epoch": 97.02290076335878, "grad_norm": 0.14329379796981812, "learning_rate": 0.00019008148049822168, "loss": 0.0057, "step": 12710 }, { "epoch": 97.09923664122137, "grad_norm": 0.14050407707691193, "learning_rate": 0.00018992752428091455, "loss": 0.0066, "step": 12720 }, { "epoch": 97.17557251908397, "grad_norm": 0.07346025109291077, "learning_rate": 0.0001897735227813515, "loss": 0.0053, "step": 12730 }, { "epoch": 97.25190839694656, "grad_norm": 0.23710666596889496, "learning_rate": 0.0001896194761741874, "loss": 0.0069, "step": 12740 }, { "epoch": 97.32824427480917, "grad_norm": 0.0782107561826706, "learning_rate": 0.00018946538463412814, "loss": 0.0056, "step": 12750 }, { "epoch": 97.40458015267176, "grad_norm": 0.09000726044178009, "learning_rate": 0.00018931124833593084, "loss": 0.0052, "step": 12760 }, { "epoch": 97.48091603053435, "grad_norm": 0.07588670402765274, "learning_rate": 0.00018915706745440308, "loss": 0.0054, "step": 12770 }, { "epoch": 97.55725190839695, "grad_norm": 0.09921785444021225, "learning_rate": 0.00018900284216440325, "loss": 0.005, "step": 12780 }, { "epoch": 97.63358778625954, "grad_norm": 0.07487939298152924, "learning_rate": 0.00018884857264083984, "loss": 0.0056, "step": 12790 }, { "epoch": 97.70992366412214, "grad_norm": 0.08710671216249466, "learning_rate": 0.00018869425905867175, "loss": 0.0052, "step": 12800 }, { "epoch": 97.78625954198473, "grad_norm": 0.07325290888547897, "learning_rate": 0.0001885399015929078, "loss": 0.0053, "step": 12810 }, { "epoch": 97.86259541984732, "grad_norm": 0.06731563806533813, "learning_rate": 0.00018838550041860643, "loss": 0.0049, "step": 12820 }, { "epoch": 97.93893129770993, "grad_norm": 0.1069914773106575, "learning_rate": 0.00018823105571087582, "loss": 0.0048, "step": 12830 }, { "epoch": 98.01526717557252, "grad_norm": 0.08385618776082993, "learning_rate": 0.0001880765676448735, "loss": 0.0044, "step": 12840 }, { "epoch": 98.09160305343511, "grad_norm": 0.09988831728696823, "learning_rate": 0.00018792203639580603, "loss": 0.0043, "step": 12850 }, { "epoch": 98.16793893129771, "grad_norm": 0.09479862451553345, "learning_rate": 0.0001877674621389291, "loss": 0.0051, "step": 12860 }, { "epoch": 98.2442748091603, "grad_norm": 0.09942471235990524, "learning_rate": 0.00018761284504954708, "loss": 0.0042, "step": 12870 }, { "epoch": 98.3206106870229, "grad_norm": 0.10942773520946503, "learning_rate": 0.000187458185303013, "loss": 0.0049, "step": 12880 }, { "epoch": 98.3969465648855, "grad_norm": 0.10972045361995697, "learning_rate": 0.00018730348307472824, "loss": 0.0052, "step": 12890 }, { "epoch": 98.47328244274809, "grad_norm": 0.07719683647155762, "learning_rate": 0.00018714873854014225, "loss": 0.0056, "step": 12900 }, { "epoch": 98.54961832061069, "grad_norm": 0.09306437522172928, "learning_rate": 0.00018699395187475265, "loss": 0.0049, "step": 12910 }, { "epoch": 98.62595419847328, "grad_norm": 0.10391296446323395, "learning_rate": 0.00018683912325410468, "loss": 0.0055, "step": 12920 }, { "epoch": 98.70229007633588, "grad_norm": 0.08146488666534424, "learning_rate": 0.00018668425285379127, "loss": 0.0054, "step": 12930 }, { "epoch": 98.77862595419847, "grad_norm": 0.08431102335453033, "learning_rate": 0.00018652934084945267, "loss": 0.0053, "step": 12940 }, { "epoch": 98.85496183206106, "grad_norm": 0.0783485546708107, "learning_rate": 0.00018637438741677631, "loss": 0.005, "step": 12950 }, { "epoch": 98.93129770992367, "grad_norm": 0.09908328950405121, "learning_rate": 0.0001862193927314967, "loss": 0.0051, "step": 12960 }, { "epoch": 99.00763358778626, "grad_norm": 0.07146568596363068, "learning_rate": 0.00018606435696939501, "loss": 0.0053, "step": 12970 }, { "epoch": 99.08396946564885, "grad_norm": 0.10720167309045792, "learning_rate": 0.00018590928030629903, "loss": 0.0047, "step": 12980 }, { "epoch": 99.16030534351145, "grad_norm": 0.07774188369512558, "learning_rate": 0.00018575416291808297, "loss": 0.0043, "step": 12990 }, { "epoch": 99.23664122137404, "grad_norm": 0.08632880449295044, "learning_rate": 0.00018559900498066724, "loss": 0.0044, "step": 13000 }, { "epoch": 99.31297709923665, "grad_norm": 0.12080992013216019, "learning_rate": 0.00018544380667001814, "loss": 0.0047, "step": 13010 }, { "epoch": 99.38931297709924, "grad_norm": 0.11937975138425827, "learning_rate": 0.00018528856816214787, "loss": 0.005, "step": 13020 }, { "epoch": 99.46564885496183, "grad_norm": 0.08136864751577377, "learning_rate": 0.00018513328963311424, "loss": 0.0049, "step": 13030 }, { "epoch": 99.54198473282443, "grad_norm": 0.08067765831947327, "learning_rate": 0.00018497797125902022, "loss": 0.0051, "step": 13040 }, { "epoch": 99.61832061068702, "grad_norm": 0.09699923545122147, "learning_rate": 0.00018482261321601433, "loss": 0.0049, "step": 13050 }, { "epoch": 99.69465648854961, "grad_norm": 0.07891429215669632, "learning_rate": 0.0001846672156802897, "loss": 0.0053, "step": 13060 }, { "epoch": 99.77099236641222, "grad_norm": 0.08347102254629135, "learning_rate": 0.0001845117788280845, "loss": 0.0057, "step": 13070 }, { "epoch": 99.8473282442748, "grad_norm": 0.1251756101846695, "learning_rate": 0.0001843563028356815, "loss": 0.0057, "step": 13080 }, { "epoch": 99.92366412213741, "grad_norm": 0.08790383487939835, "learning_rate": 0.00018420078787940764, "loss": 0.0055, "step": 13090 }, { "epoch": 100.0, "grad_norm": 0.11510046571493149, "learning_rate": 0.00018404523413563425, "loss": 0.0053, "step": 13100 }, { "epoch": 100.07633587786259, "grad_norm": 0.1126776710152626, "learning_rate": 0.00018388964178077662, "loss": 0.0056, "step": 13110 }, { "epoch": 100.1526717557252, "grad_norm": 0.10751194506883621, "learning_rate": 0.00018373401099129374, "loss": 0.0051, "step": 13120 }, { "epoch": 100.22900763358778, "grad_norm": 0.09632817655801773, "learning_rate": 0.00018357834194368823, "loss": 0.0053, "step": 13130 }, { "epoch": 100.30534351145039, "grad_norm": 0.10124155133962631, "learning_rate": 0.00018342263481450617, "loss": 0.005, "step": 13140 }, { "epoch": 100.38167938931298, "grad_norm": 0.0920691192150116, "learning_rate": 0.00018326688978033678, "loss": 0.0054, "step": 13150 }, { "epoch": 100.45801526717557, "grad_norm": 0.06691168993711472, "learning_rate": 0.00018311110701781222, "loss": 0.0051, "step": 13160 }, { "epoch": 100.53435114503817, "grad_norm": 0.08223991096019745, "learning_rate": 0.00018295528670360747, "loss": 0.0052, "step": 13170 }, { "epoch": 100.61068702290076, "grad_norm": 0.07944302260875702, "learning_rate": 0.0001827994290144402, "loss": 0.0053, "step": 13180 }, { "epoch": 100.68702290076335, "grad_norm": 0.07284484803676605, "learning_rate": 0.00018264353412707026, "loss": 0.0051, "step": 13190 }, { "epoch": 100.76335877862596, "grad_norm": 0.10603784024715424, "learning_rate": 0.0001824876022182998, "loss": 0.0048, "step": 13200 }, { "epoch": 100.83969465648855, "grad_norm": 0.10490042716264725, "learning_rate": 0.00018233163346497304, "loss": 0.005, "step": 13210 }, { "epoch": 100.91603053435115, "grad_norm": 0.11381945759057999, "learning_rate": 0.00018217562804397587, "loss": 0.0049, "step": 13220 }, { "epoch": 100.99236641221374, "grad_norm": 0.10784310102462769, "learning_rate": 0.00018201958613223572, "loss": 0.0052, "step": 13230 }, { "epoch": 101.06870229007633, "grad_norm": 0.11253736913204193, "learning_rate": 0.00018186350790672166, "loss": 0.0049, "step": 13240 }, { "epoch": 101.14503816793894, "grad_norm": 0.09969451278448105, "learning_rate": 0.00018170739354444364, "loss": 0.0045, "step": 13250 }, { "epoch": 101.22137404580153, "grad_norm": 0.08545141667127609, "learning_rate": 0.00018155124322245277, "loss": 0.005, "step": 13260 }, { "epoch": 101.29770992366412, "grad_norm": 0.0801873654127121, "learning_rate": 0.00018139505711784078, "loss": 0.0043, "step": 13270 }, { "epoch": 101.37404580152672, "grad_norm": 0.06917334347963333, "learning_rate": 0.00018123883540774023, "loss": 0.0043, "step": 13280 }, { "epoch": 101.45038167938931, "grad_norm": 0.10405208170413971, "learning_rate": 0.0001810825782693239, "loss": 0.0044, "step": 13290 }, { "epoch": 101.52671755725191, "grad_norm": 0.065947987139225, "learning_rate": 0.0001809262858798047, "loss": 0.0046, "step": 13300 }, { "epoch": 101.6030534351145, "grad_norm": 0.07857424765825272, "learning_rate": 0.00018076995841643565, "loss": 0.0045, "step": 13310 }, { "epoch": 101.6793893129771, "grad_norm": 0.09352642297744751, "learning_rate": 0.00018061359605650956, "loss": 0.0045, "step": 13320 }, { "epoch": 101.7557251908397, "grad_norm": 0.09057946503162384, "learning_rate": 0.00018045719897735857, "loss": 0.0044, "step": 13330 }, { "epoch": 101.83206106870229, "grad_norm": 0.07546000182628632, "learning_rate": 0.00018030076735635453, "loss": 0.0047, "step": 13340 }, { "epoch": 101.90839694656489, "grad_norm": 0.07946012914180756, "learning_rate": 0.00018014430137090824, "loss": 0.005, "step": 13350 }, { "epoch": 101.98473282442748, "grad_norm": 0.08771426975727081, "learning_rate": 0.00017998780119846952, "loss": 0.005, "step": 13360 }, { "epoch": 102.06106870229007, "grad_norm": 0.08349328488111496, "learning_rate": 0.0001798312670165271, "loss": 0.0052, "step": 13370 }, { "epoch": 102.13740458015268, "grad_norm": 0.099086232483387, "learning_rate": 0.000179674699002608, "loss": 0.0047, "step": 13380 }, { "epoch": 102.21374045801527, "grad_norm": 0.07345457375049591, "learning_rate": 0.00017951809733427783, "loss": 0.0041, "step": 13390 }, { "epoch": 102.29007633587786, "grad_norm": 0.11011980473995209, "learning_rate": 0.00017936146218914037, "loss": 0.0047, "step": 13400 }, { "epoch": 102.36641221374046, "grad_norm": 0.10335379838943481, "learning_rate": 0.00017920479374483725, "loss": 0.0047, "step": 13410 }, { "epoch": 102.44274809160305, "grad_norm": 0.11337413638830185, "learning_rate": 0.00017904809217904793, "loss": 0.0051, "step": 13420 }, { "epoch": 102.51908396946565, "grad_norm": 0.10762869566679001, "learning_rate": 0.00017889135766948944, "loss": 0.0047, "step": 13430 }, { "epoch": 102.59541984732824, "grad_norm": 0.09810832887887955, "learning_rate": 0.00017873459039391607, "loss": 0.006, "step": 13440 }, { "epoch": 102.67175572519083, "grad_norm": 0.08014798909425735, "learning_rate": 0.00017857779053011952, "loss": 0.0047, "step": 13450 }, { "epoch": 102.74809160305344, "grad_norm": 0.08650603145360947, "learning_rate": 0.00017842095825592816, "loss": 0.005, "step": 13460 }, { "epoch": 102.82442748091603, "grad_norm": 0.11080346256494522, "learning_rate": 0.00017826409374920726, "loss": 0.0047, "step": 13470 }, { "epoch": 102.90076335877862, "grad_norm": 0.09018059819936752, "learning_rate": 0.0001781071971878587, "loss": 0.0052, "step": 13480 }, { "epoch": 102.97709923664122, "grad_norm": 0.08413974940776825, "learning_rate": 0.00017795026874982054, "loss": 0.0051, "step": 13490 }, { "epoch": 103.05343511450381, "grad_norm": 0.08660850673913956, "learning_rate": 0.00017779330861306713, "loss": 0.0046, "step": 13500 }, { "epoch": 103.12977099236642, "grad_norm": 0.11580168455839157, "learning_rate": 0.00017763631695560876, "loss": 0.0046, "step": 13510 }, { "epoch": 103.20610687022901, "grad_norm": 0.12119609117507935, "learning_rate": 0.00017747929395549143, "loss": 0.0046, "step": 13520 }, { "epoch": 103.2824427480916, "grad_norm": 0.10104073584079742, "learning_rate": 0.0001773222397907967, "loss": 0.0049, "step": 13530 }, { "epoch": 103.3587786259542, "grad_norm": 0.09768396615982056, "learning_rate": 0.00017716515463964139, "loss": 0.0048, "step": 13540 }, { "epoch": 103.43511450381679, "grad_norm": 0.08208809792995453, "learning_rate": 0.00017700803868017764, "loss": 0.0049, "step": 13550 }, { "epoch": 103.5114503816794, "grad_norm": 0.07479066401720047, "learning_rate": 0.00017685089209059247, "loss": 0.0053, "step": 13560 }, { "epoch": 103.58778625954199, "grad_norm": 0.1105550229549408, "learning_rate": 0.00017669371504910745, "loss": 0.005, "step": 13570 }, { "epoch": 103.66412213740458, "grad_norm": 0.09717659652233124, "learning_rate": 0.000176536507733979, "loss": 0.0052, "step": 13580 }, { "epoch": 103.74045801526718, "grad_norm": 0.09880314022302628, "learning_rate": 0.00017637927032349755, "loss": 0.0049, "step": 13590 }, { "epoch": 103.81679389312977, "grad_norm": 0.1178559884428978, "learning_rate": 0.00017622200299598792, "loss": 0.0051, "step": 13600 }, { "epoch": 103.89312977099236, "grad_norm": 0.13308820128440857, "learning_rate": 0.0001760647059298088, "loss": 0.0053, "step": 13610 }, { "epoch": 103.96946564885496, "grad_norm": 0.1300085335969925, "learning_rate": 0.0001759073793033525, "loss": 0.0055, "step": 13620 }, { "epoch": 104.04580152671755, "grad_norm": 0.11449213325977325, "learning_rate": 0.00017575002329504492, "loss": 0.0053, "step": 13630 }, { "epoch": 104.12213740458016, "grad_norm": 0.07634349912405014, "learning_rate": 0.00017559263808334527, "loss": 0.0048, "step": 13640 }, { "epoch": 104.19847328244275, "grad_norm": 0.0753246620297432, "learning_rate": 0.00017543522384674595, "loss": 0.0048, "step": 13650 }, { "epoch": 104.27480916030534, "grad_norm": 0.0673556700348854, "learning_rate": 0.0001752777807637722, "loss": 0.0045, "step": 13660 }, { "epoch": 104.35114503816794, "grad_norm": 0.09730527549982071, "learning_rate": 0.000175120309012982, "loss": 0.0046, "step": 13670 }, { "epoch": 104.42748091603053, "grad_norm": 0.06148066744208336, "learning_rate": 0.0001749628087729658, "loss": 0.0047, "step": 13680 }, { "epoch": 104.50381679389314, "grad_norm": 0.11222966760396957, "learning_rate": 0.0001748052802223465, "loss": 0.005, "step": 13690 }, { "epoch": 104.58015267175573, "grad_norm": 0.09568056464195251, "learning_rate": 0.0001746477235397789, "loss": 0.0049, "step": 13700 }, { "epoch": 104.65648854961832, "grad_norm": 0.0656394362449646, "learning_rate": 0.00017449013890394986, "loss": 0.0044, "step": 13710 }, { "epoch": 104.73282442748092, "grad_norm": 0.09454227238893509, "learning_rate": 0.00017433252649357793, "loss": 0.005, "step": 13720 }, { "epoch": 104.80916030534351, "grad_norm": 0.06431503593921661, "learning_rate": 0.00017417488648741302, "loss": 0.0044, "step": 13730 }, { "epoch": 104.8854961832061, "grad_norm": 0.07720022648572922, "learning_rate": 0.0001740172190642365, "loss": 0.0049, "step": 13740 }, { "epoch": 104.9618320610687, "grad_norm": 0.09384473413228989, "learning_rate": 0.0001738595244028608, "loss": 0.0047, "step": 13750 }, { "epoch": 105.0381679389313, "grad_norm": 0.08099418133497238, "learning_rate": 0.0001737018026821292, "loss": 0.0046, "step": 13760 }, { "epoch": 105.1145038167939, "grad_norm": 0.07922247797250748, "learning_rate": 0.00017354405408091567, "loss": 0.0046, "step": 13770 }, { "epoch": 105.19083969465649, "grad_norm": 0.10829923301935196, "learning_rate": 0.00017338627877812467, "loss": 0.0045, "step": 13780 }, { "epoch": 105.26717557251908, "grad_norm": 0.0789497047662735, "learning_rate": 0.00017322847695269095, "loss": 0.0049, "step": 13790 }, { "epoch": 105.34351145038168, "grad_norm": 0.1096484363079071, "learning_rate": 0.0001730706487835794, "loss": 0.0047, "step": 13800 }, { "epoch": 105.41984732824427, "grad_norm": 0.07817990332841873, "learning_rate": 0.00017291279444978466, "loss": 0.0053, "step": 13810 }, { "epoch": 105.49618320610686, "grad_norm": 0.09676060080528259, "learning_rate": 0.00017275491413033114, "loss": 0.005, "step": 13820 }, { "epoch": 105.57251908396947, "grad_norm": 0.09888887405395508, "learning_rate": 0.00017259700800427267, "loss": 0.0048, "step": 13830 }, { "epoch": 105.64885496183206, "grad_norm": 0.07110144197940826, "learning_rate": 0.0001724390762506924, "loss": 0.0045, "step": 13840 }, { "epoch": 105.72519083969466, "grad_norm": 0.09008005261421204, "learning_rate": 0.0001722811190487025, "loss": 0.0041, "step": 13850 }, { "epoch": 105.80152671755725, "grad_norm": 0.06796430051326752, "learning_rate": 0.000172123136577444, "loss": 0.0045, "step": 13860 }, { "epoch": 105.87786259541984, "grad_norm": 0.050181902945041656, "learning_rate": 0.00017196512901608666, "loss": 0.0045, "step": 13870 }, { "epoch": 105.95419847328245, "grad_norm": 0.06742695719003677, "learning_rate": 0.0001718070965438286, "loss": 0.0044, "step": 13880 }, { "epoch": 106.03053435114504, "grad_norm": 0.08093133568763733, "learning_rate": 0.0001716490393398962, "loss": 0.0045, "step": 13890 }, { "epoch": 106.10687022900764, "grad_norm": 0.06437555700540543, "learning_rate": 0.00017149095758354395, "loss": 0.0046, "step": 13900 }, { "epoch": 106.18320610687023, "grad_norm": 0.105768121778965, "learning_rate": 0.00017133285145405421, "loss": 0.0043, "step": 13910 }, { "epoch": 106.25954198473282, "grad_norm": 0.08882465213537216, "learning_rate": 0.00017117472113073681, "loss": 0.0039, "step": 13920 }, { "epoch": 106.33587786259542, "grad_norm": 0.07190801948308945, "learning_rate": 0.00017101656679292923, "loss": 0.0047, "step": 13930 }, { "epoch": 106.41221374045801, "grad_norm": 0.08083082735538483, "learning_rate": 0.00017085838861999602, "loss": 0.0047, "step": 13940 }, { "epoch": 106.4885496183206, "grad_norm": 0.07864580303430557, "learning_rate": 0.00017070018679132885, "loss": 0.0051, "step": 13950 }, { "epoch": 106.56488549618321, "grad_norm": 0.09398052096366882, "learning_rate": 0.00017054196148634623, "loss": 0.0047, "step": 13960 }, { "epoch": 106.6412213740458, "grad_norm": 0.09532858431339264, "learning_rate": 0.00017038371288449327, "loss": 0.005, "step": 13970 }, { "epoch": 106.7175572519084, "grad_norm": 0.11149079352617264, "learning_rate": 0.00017022544116524147, "loss": 0.0049, "step": 13980 }, { "epoch": 106.79389312977099, "grad_norm": 0.09700220823287964, "learning_rate": 0.00017006714650808855, "loss": 0.0051, "step": 13990 }, { "epoch": 106.87022900763358, "grad_norm": 0.0707506537437439, "learning_rate": 0.00016990882909255828, "loss": 0.005, "step": 14000 }, { "epoch": 106.94656488549619, "grad_norm": 0.0743270069360733, "learning_rate": 0.0001697504890982003, "loss": 0.0048, "step": 14010 }, { "epoch": 107.02290076335878, "grad_norm": 0.07250791043043137, "learning_rate": 0.00016959212670458965, "loss": 0.005, "step": 14020 }, { "epoch": 107.09923664122137, "grad_norm": 0.09110283851623535, "learning_rate": 0.00016943374209132706, "loss": 0.005, "step": 14030 }, { "epoch": 107.17557251908397, "grad_norm": 0.07070577889680862, "learning_rate": 0.0001692753354380382, "loss": 0.0046, "step": 14040 }, { "epoch": 107.25190839694656, "grad_norm": 0.13567492365837097, "learning_rate": 0.00016911690692437384, "loss": 0.0049, "step": 14050 }, { "epoch": 107.32824427480917, "grad_norm": 0.22503182291984558, "learning_rate": 0.00016895845673000964, "loss": 0.0051, "step": 14060 }, { "epoch": 107.40458015267176, "grad_norm": 0.08834319561719894, "learning_rate": 0.00016879998503464561, "loss": 0.0049, "step": 14070 }, { "epoch": 107.48091603053435, "grad_norm": 0.11455649882555008, "learning_rate": 0.00016864149201800644, "loss": 0.0055, "step": 14080 }, { "epoch": 107.55725190839695, "grad_norm": 0.14482314884662628, "learning_rate": 0.00016848297785984075, "loss": 0.0064, "step": 14090 }, { "epoch": 107.63358778625954, "grad_norm": 0.08357861638069153, "learning_rate": 0.00016832444273992127, "loss": 0.0053, "step": 14100 }, { "epoch": 107.70992366412214, "grad_norm": 0.1424528956413269, "learning_rate": 0.00016816588683804447, "loss": 0.0054, "step": 14110 }, { "epoch": 107.78625954198473, "grad_norm": 0.08314794301986694, "learning_rate": 0.0001680073103340304, "loss": 0.0046, "step": 14120 }, { "epoch": 107.86259541984732, "grad_norm": 0.08626110106706619, "learning_rate": 0.00016784871340772247, "loss": 0.0048, "step": 14130 }, { "epoch": 107.93893129770993, "grad_norm": 0.06999144703149796, "learning_rate": 0.00016769009623898715, "loss": 0.006, "step": 14140 }, { "epoch": 108.01526717557252, "grad_norm": 0.11077379435300827, "learning_rate": 0.00016753145900771409, "loss": 0.0054, "step": 14150 }, { "epoch": 108.09160305343511, "grad_norm": 0.08547571301460266, "learning_rate": 0.00016737280189381542, "loss": 0.0061, "step": 14160 }, { "epoch": 108.16793893129771, "grad_norm": 0.07285712659358978, "learning_rate": 0.00016721412507722617, "loss": 0.0052, "step": 14170 }, { "epoch": 108.2442748091603, "grad_norm": 0.11476943641901016, "learning_rate": 0.00016705542873790327, "loss": 0.0057, "step": 14180 }, { "epoch": 108.3206106870229, "grad_norm": 0.10240926593542099, "learning_rate": 0.00016689671305582623, "loss": 0.0054, "step": 14190 }, { "epoch": 108.3969465648855, "grad_norm": 0.07563871890306473, "learning_rate": 0.0001667379782109962, "loss": 0.0046, "step": 14200 }, { "epoch": 108.47328244274809, "grad_norm": 0.11773407459259033, "learning_rate": 0.00016657922438343615, "loss": 0.0048, "step": 14210 }, { "epoch": 108.54961832061069, "grad_norm": 0.09807853400707245, "learning_rate": 0.0001664204517531906, "loss": 0.006, "step": 14220 }, { "epoch": 108.62595419847328, "grad_norm": 0.09840322285890579, "learning_rate": 0.00016626166050032543, "loss": 0.0048, "step": 14230 }, { "epoch": 108.70229007633588, "grad_norm": 0.13296011090278625, "learning_rate": 0.00016610285080492754, "loss": 0.0049, "step": 14240 }, { "epoch": 108.77862595419847, "grad_norm": 0.10428967326879501, "learning_rate": 0.0001659440228471048, "loss": 0.0046, "step": 14250 }, { "epoch": 108.85496183206106, "grad_norm": 0.08326748758554459, "learning_rate": 0.00016578517680698583, "loss": 0.0051, "step": 14260 }, { "epoch": 108.93129770992367, "grad_norm": 0.08438200503587723, "learning_rate": 0.00016562631286471964, "loss": 0.005, "step": 14270 }, { "epoch": 109.00763358778626, "grad_norm": 0.10506819933652878, "learning_rate": 0.0001654674312004757, "loss": 0.0048, "step": 14280 }, { "epoch": 109.08396946564885, "grad_norm": 0.12549395859241486, "learning_rate": 0.00016530853199444345, "loss": 0.0045, "step": 14290 }, { "epoch": 109.16030534351145, "grad_norm": 0.10266952216625214, "learning_rate": 0.0001651496154268323, "loss": 0.0052, "step": 14300 }, { "epoch": 109.23664122137404, "grad_norm": 0.13090965151786804, "learning_rate": 0.00016499068167787133, "loss": 0.0051, "step": 14310 }, { "epoch": 109.31297709923665, "grad_norm": 0.09972165524959564, "learning_rate": 0.00016483173092780908, "loss": 0.0054, "step": 14320 }, { "epoch": 109.38931297709924, "grad_norm": 0.11910360306501389, "learning_rate": 0.00016467276335691336, "loss": 0.0055, "step": 14330 }, { "epoch": 109.46564885496183, "grad_norm": 0.1191544458270073, "learning_rate": 0.00016451377914547115, "loss": 0.0071, "step": 14340 }, { "epoch": 109.54198473282443, "grad_norm": 0.11750300228595734, "learning_rate": 0.00016435477847378816, "loss": 0.0055, "step": 14350 }, { "epoch": 109.61832061068702, "grad_norm": 0.07722923159599304, "learning_rate": 0.00016419576152218896, "loss": 0.0061, "step": 14360 }, { "epoch": 109.69465648854961, "grad_norm": 0.11351567506790161, "learning_rate": 0.00016403672847101628, "loss": 0.0057, "step": 14370 }, { "epoch": 109.77099236641222, "grad_norm": 0.12854823470115662, "learning_rate": 0.00016387767950063153, "loss": 0.0052, "step": 14380 }, { "epoch": 109.8473282442748, "grad_norm": 0.10249361395835876, "learning_rate": 0.0001637186147914138, "loss": 0.0059, "step": 14390 }, { "epoch": 109.92366412213741, "grad_norm": 0.08682294189929962, "learning_rate": 0.0001635595345237602, "loss": 0.0053, "step": 14400 }, { "epoch": 110.0, "grad_norm": 0.07835379242897034, "learning_rate": 0.00016340043887808547, "loss": 0.0051, "step": 14410 }, { "epoch": 110.07633587786259, "grad_norm": 0.0707458108663559, "learning_rate": 0.00016324132803482178, "loss": 0.0046, "step": 14420 }, { "epoch": 110.1526717557252, "grad_norm": 0.11742016673088074, "learning_rate": 0.0001630822021744185, "loss": 0.0051, "step": 14430 }, { "epoch": 110.22900763358778, "grad_norm": 0.08114538341760635, "learning_rate": 0.0001629230614773421, "loss": 0.0052, "step": 14440 }, { "epoch": 110.30534351145039, "grad_norm": 0.10613620281219482, "learning_rate": 0.0001627639061240758, "loss": 0.0046, "step": 14450 }, { "epoch": 110.38167938931298, "grad_norm": 0.08354470878839493, "learning_rate": 0.00016260473629511952, "loss": 0.0045, "step": 14460 }, { "epoch": 110.45801526717557, "grad_norm": 0.07735593616962433, "learning_rate": 0.0001624455521709896, "loss": 0.0045, "step": 14470 }, { "epoch": 110.53435114503817, "grad_norm": 0.06251084059476852, "learning_rate": 0.0001622863539322184, "loss": 0.0043, "step": 14480 }, { "epoch": 110.61068702290076, "grad_norm": 0.06657090038061142, "learning_rate": 0.0001621271417593546, "loss": 0.0041, "step": 14490 }, { "epoch": 110.68702290076335, "grad_norm": 0.058634210377931595, "learning_rate": 0.00016196791583296245, "loss": 0.0045, "step": 14500 }, { "epoch": 110.76335877862596, "grad_norm": 0.06588494032621384, "learning_rate": 0.00016180867633362189, "loss": 0.0041, "step": 14510 }, { "epoch": 110.83969465648855, "grad_norm": 0.10228168964385986, "learning_rate": 0.0001616494234419282, "loss": 0.0045, "step": 14520 }, { "epoch": 110.91603053435115, "grad_norm": 0.06296516954898834, "learning_rate": 0.00016149015733849195, "loss": 0.0041, "step": 14530 }, { "epoch": 110.99236641221374, "grad_norm": 0.10837358981370926, "learning_rate": 0.00016133087820393853, "loss": 0.0051, "step": 14540 }, { "epoch": 111.06870229007633, "grad_norm": 0.08700711280107498, "learning_rate": 0.00016117158621890832, "loss": 0.0046, "step": 14550 }, { "epoch": 111.14503816793894, "grad_norm": 0.08561043441295624, "learning_rate": 0.00016101228156405604, "loss": 0.0045, "step": 14560 }, { "epoch": 111.22137404580153, "grad_norm": 0.09594454616308212, "learning_rate": 0.00016085296442005096, "loss": 0.0047, "step": 14570 }, { "epoch": 111.29770992366412, "grad_norm": 0.0743875503540039, "learning_rate": 0.00016069363496757646, "loss": 0.0045, "step": 14580 }, { "epoch": 111.37404580152672, "grad_norm": 0.11883817613124847, "learning_rate": 0.0001605342933873298, "loss": 0.004, "step": 14590 }, { "epoch": 111.45038167938931, "grad_norm": 0.06604287028312683, "learning_rate": 0.0001603749398600222, "loss": 0.0046, "step": 14600 }, { "epoch": 111.52671755725191, "grad_norm": 0.06978397816419601, "learning_rate": 0.00016021557456637816, "loss": 0.0044, "step": 14610 }, { "epoch": 111.6030534351145, "grad_norm": 0.10603348910808563, "learning_rate": 0.0001600561976871357, "loss": 0.0047, "step": 14620 }, { "epoch": 111.6793893129771, "grad_norm": 0.0919542908668518, "learning_rate": 0.00015989680940304603, "loss": 0.0044, "step": 14630 }, { "epoch": 111.7557251908397, "grad_norm": 0.0813869833946228, "learning_rate": 0.00015973740989487303, "loss": 0.0045, "step": 14640 }, { "epoch": 111.83206106870229, "grad_norm": 0.08310044556856155, "learning_rate": 0.0001595779993433936, "loss": 0.0047, "step": 14650 }, { "epoch": 111.90839694656489, "grad_norm": 0.09320095181465149, "learning_rate": 0.000159418577929397, "loss": 0.0043, "step": 14660 }, { "epoch": 111.98473282442748, "grad_norm": 0.09022139012813568, "learning_rate": 0.0001592591458336849, "loss": 0.0045, "step": 14670 }, { "epoch": 112.06106870229007, "grad_norm": 0.08379196375608444, "learning_rate": 0.00015909970323707095, "loss": 0.0047, "step": 14680 }, { "epoch": 112.13740458015268, "grad_norm": 0.08576789498329163, "learning_rate": 0.0001589402503203809, "loss": 0.0045, "step": 14690 }, { "epoch": 112.21374045801527, "grad_norm": 0.1074470728635788, "learning_rate": 0.00015878078726445206, "loss": 0.0041, "step": 14700 }, { "epoch": 112.29007633587786, "grad_norm": 0.06470780819654465, "learning_rate": 0.00015862131425013328, "loss": 0.0044, "step": 14710 }, { "epoch": 112.36641221374046, "grad_norm": 0.08336208760738373, "learning_rate": 0.00015846183145828472, "loss": 0.0041, "step": 14720 }, { "epoch": 112.44274809160305, "grad_norm": 0.11156567186117172, "learning_rate": 0.0001583023390697776, "loss": 0.0045, "step": 14730 }, { "epoch": 112.51908396946565, "grad_norm": 0.062195971608161926, "learning_rate": 0.000158142837265494, "loss": 0.0042, "step": 14740 }, { "epoch": 112.59541984732824, "grad_norm": 0.08501240611076355, "learning_rate": 0.0001579833262263268, "loss": 0.0044, "step": 14750 }, { "epoch": 112.67175572519083, "grad_norm": 0.06555528938770294, "learning_rate": 0.00015782380613317922, "loss": 0.0044, "step": 14760 }, { "epoch": 112.74809160305344, "grad_norm": 0.09308788180351257, "learning_rate": 0.00015766427716696477, "loss": 0.0044, "step": 14770 }, { "epoch": 112.82442748091603, "grad_norm": 0.0831044465303421, "learning_rate": 0.00015750473950860712, "loss": 0.0041, "step": 14780 }, { "epoch": 112.90076335877862, "grad_norm": 0.0747564435005188, "learning_rate": 0.00015734519333903966, "loss": 0.0042, "step": 14790 }, { "epoch": 112.97709923664122, "grad_norm": 0.09457650780677795, "learning_rate": 0.00015718563883920553, "loss": 0.0045, "step": 14800 }, { "epoch": 113.05343511450381, "grad_norm": 0.08566130697727203, "learning_rate": 0.00015702607619005735, "loss": 0.004, "step": 14810 }, { "epoch": 113.12977099236642, "grad_norm": 0.10495772957801819, "learning_rate": 0.00015686650557255683, "loss": 0.0049, "step": 14820 }, { "epoch": 113.20610687022901, "grad_norm": 0.09908543527126312, "learning_rate": 0.00015670692716767483, "loss": 0.0047, "step": 14830 }, { "epoch": 113.2824427480916, "grad_norm": 0.10702883452177048, "learning_rate": 0.00015654734115639105, "loss": 0.0047, "step": 14840 }, { "epoch": 113.3587786259542, "grad_norm": 0.09336569160223007, "learning_rate": 0.00015638774771969377, "loss": 0.0046, "step": 14850 }, { "epoch": 113.43511450381679, "grad_norm": 0.06786751747131348, "learning_rate": 0.0001562281470385797, "loss": 0.0044, "step": 14860 }, { "epoch": 113.5114503816794, "grad_norm": 0.08758401870727539, "learning_rate": 0.0001560685392940538, "loss": 0.0042, "step": 14870 }, { "epoch": 113.58778625954199, "grad_norm": 0.07034429162740707, "learning_rate": 0.00015590892466712898, "loss": 0.0041, "step": 14880 }, { "epoch": 113.66412213740458, "grad_norm": 0.09382674843072891, "learning_rate": 0.000155749303338826, "loss": 0.004, "step": 14890 }, { "epoch": 113.74045801526718, "grad_norm": 0.09130297601222992, "learning_rate": 0.00015558967549017319, "loss": 0.0039, "step": 14900 }, { "epoch": 113.81679389312977, "grad_norm": 0.07922729104757309, "learning_rate": 0.00015543004130220633, "loss": 0.0042, "step": 14910 }, { "epoch": 113.89312977099236, "grad_norm": 0.06698399037122726, "learning_rate": 0.00015527040095596843, "loss": 0.0041, "step": 14920 }, { "epoch": 113.96946564885496, "grad_norm": 0.081124447286129, "learning_rate": 0.00015511075463250923, "loss": 0.0044, "step": 14930 }, { "epoch": 114.04580152671755, "grad_norm": 0.09107506275177002, "learning_rate": 0.00015495110251288556, "loss": 0.004, "step": 14940 }, { "epoch": 114.12213740458016, "grad_norm": 0.09215667843818665, "learning_rate": 0.00015479144477816066, "loss": 0.0046, "step": 14950 }, { "epoch": 114.19847328244275, "grad_norm": 0.1279340386390686, "learning_rate": 0.00015463178160940412, "loss": 0.0047, "step": 14960 }, { "epoch": 114.27480916030534, "grad_norm": 0.06789569556713104, "learning_rate": 0.00015447211318769178, "loss": 0.0046, "step": 14970 }, { "epoch": 114.35114503816794, "grad_norm": 0.0736195296049118, "learning_rate": 0.0001543124396941055, "loss": 0.0042, "step": 14980 }, { "epoch": 114.42748091603053, "grad_norm": 0.07486481964588165, "learning_rate": 0.00015415276130973255, "loss": 0.0042, "step": 14990 }, { "epoch": 114.50381679389314, "grad_norm": 0.08140617609024048, "learning_rate": 0.0001539930782156662, "loss": 0.0041, "step": 15000 }, { "epoch": 114.58015267175573, "grad_norm": 0.12177591025829315, "learning_rate": 0.00015383339059300475, "loss": 0.0059, "step": 15010 }, { "epoch": 114.65648854961832, "grad_norm": 0.06530344486236572, "learning_rate": 0.0001536736986228518, "loss": 0.0045, "step": 15020 }, { "epoch": 114.73282442748092, "grad_norm": 0.07920712977647781, "learning_rate": 0.00015351400248631578, "loss": 0.0049, "step": 15030 }, { "epoch": 114.80916030534351, "grad_norm": 0.0965488851070404, "learning_rate": 0.00015335430236450986, "loss": 0.0045, "step": 15040 }, { "epoch": 114.8854961832061, "grad_norm": 0.09541898220777512, "learning_rate": 0.00015319459843855178, "loss": 0.0047, "step": 15050 }, { "epoch": 114.9618320610687, "grad_norm": 0.08147253096103668, "learning_rate": 0.00015303489088956356, "loss": 0.0047, "step": 15060 }, { "epoch": 115.0381679389313, "grad_norm": 0.08355256170034409, "learning_rate": 0.00015287517989867133, "loss": 0.0042, "step": 15070 }, { "epoch": 115.1145038167939, "grad_norm": 0.09481508284807205, "learning_rate": 0.00015271546564700513, "loss": 0.0046, "step": 15080 }, { "epoch": 115.19083969465649, "grad_norm": 0.07805559039115906, "learning_rate": 0.00015255574831569868, "loss": 0.0044, "step": 15090 }, { "epoch": 115.26717557251908, "grad_norm": 0.07555066794157028, "learning_rate": 0.0001523960280858892, "loss": 0.0038, "step": 15100 }, { "epoch": 115.34351145038168, "grad_norm": 0.10559952259063721, "learning_rate": 0.00015223630513871728, "loss": 0.0044, "step": 15110 }, { "epoch": 115.41984732824427, "grad_norm": 0.11491971462965012, "learning_rate": 0.0001520765796553264, "loss": 0.0044, "step": 15120 }, { "epoch": 115.49618320610686, "grad_norm": 0.06694521754980087, "learning_rate": 0.00015191685181686315, "loss": 0.0047, "step": 15130 }, { "epoch": 115.57251908396947, "grad_norm": 0.09743628650903702, "learning_rate": 0.0001517571218044766, "loss": 0.0044, "step": 15140 }, { "epoch": 115.64885496183206, "grad_norm": 0.08871922641992569, "learning_rate": 0.00015159738979931837, "loss": 0.0041, "step": 15150 }, { "epoch": 115.72519083969466, "grad_norm": 0.09946122765541077, "learning_rate": 0.00015143765598254235, "loss": 0.0045, "step": 15160 }, { "epoch": 115.80152671755725, "grad_norm": 0.0793207511305809, "learning_rate": 0.00015127792053530449, "loss": 0.0047, "step": 15170 }, { "epoch": 115.87786259541984, "grad_norm": 0.07743946462869644, "learning_rate": 0.00015111818363876248, "loss": 0.0041, "step": 15180 }, { "epoch": 115.95419847328245, "grad_norm": 0.06318088620901108, "learning_rate": 0.0001509584454740758, "loss": 0.004, "step": 15190 }, { "epoch": 116.03053435114504, "grad_norm": 0.07280120253562927, "learning_rate": 0.00015079870622240524, "loss": 0.0041, "step": 15200 }, { "epoch": 116.10687022900764, "grad_norm": 0.078289695084095, "learning_rate": 0.00015063896606491305, "loss": 0.0039, "step": 15210 }, { "epoch": 116.18320610687023, "grad_norm": 0.05697508901357651, "learning_rate": 0.00015047922518276213, "loss": 0.0042, "step": 15220 }, { "epoch": 116.25954198473282, "grad_norm": 0.10051528364419937, "learning_rate": 0.00015031948375711655, "loss": 0.0041, "step": 15230 }, { "epoch": 116.33587786259542, "grad_norm": 0.049469418823719025, "learning_rate": 0.00015015974196914088, "loss": 0.0039, "step": 15240 }, { "epoch": 116.41221374045801, "grad_norm": 0.0707528367638588, "learning_rate": 0.00015, "loss": 0.004, "step": 15250 }, { "epoch": 116.4885496183206, "grad_norm": 0.08235233277082443, "learning_rate": 0.00014984025803085912, "loss": 0.0042, "step": 15260 }, { "epoch": 116.56488549618321, "grad_norm": 0.06415761262178421, "learning_rate": 0.00014968051624288342, "loss": 0.0037, "step": 15270 }, { "epoch": 116.6412213740458, "grad_norm": 0.09433627128601074, "learning_rate": 0.00014952077481723787, "loss": 0.0035, "step": 15280 }, { "epoch": 116.7175572519084, "grad_norm": 0.082395538687706, "learning_rate": 0.00014936103393508698, "loss": 0.0041, "step": 15290 }, { "epoch": 116.79389312977099, "grad_norm": 0.09782354533672333, "learning_rate": 0.00014920129377759474, "loss": 0.0041, "step": 15300 }, { "epoch": 116.87022900763358, "grad_norm": 0.08240281045436859, "learning_rate": 0.00014904155452592424, "loss": 0.004, "step": 15310 }, { "epoch": 116.94656488549619, "grad_norm": 0.09049045294523239, "learning_rate": 0.00014888181636123752, "loss": 0.0046, "step": 15320 }, { "epoch": 117.02290076335878, "grad_norm": 0.09104426205158234, "learning_rate": 0.00014872207946469554, "loss": 0.0045, "step": 15330 }, { "epoch": 117.09923664122137, "grad_norm": 0.08435387164354324, "learning_rate": 0.00014856234401745765, "loss": 0.0043, "step": 15340 }, { "epoch": 117.17557251908397, "grad_norm": 0.0721377283334732, "learning_rate": 0.00014840261020068166, "loss": 0.0043, "step": 15350 }, { "epoch": 117.25190839694656, "grad_norm": 0.09002400934696198, "learning_rate": 0.00014824287819552336, "loss": 0.004, "step": 15360 }, { "epoch": 117.32824427480917, "grad_norm": 0.07745300978422165, "learning_rate": 0.00014808314818313682, "loss": 0.0038, "step": 15370 }, { "epoch": 117.40458015267176, "grad_norm": 0.0895882174372673, "learning_rate": 0.00014792342034467356, "loss": 0.0038, "step": 15380 }, { "epoch": 117.48091603053435, "grad_norm": 0.06370102614164352, "learning_rate": 0.0001477636948612827, "loss": 0.0038, "step": 15390 }, { "epoch": 117.55725190839695, "grad_norm": 0.07594361156225204, "learning_rate": 0.00014760397191411077, "loss": 0.004, "step": 15400 }, { "epoch": 117.63358778625954, "grad_norm": 0.0625503733754158, "learning_rate": 0.0001474442516843013, "loss": 0.004, "step": 15410 }, { "epoch": 117.70992366412214, "grad_norm": 0.10104270279407501, "learning_rate": 0.00014728453435299484, "loss": 0.0043, "step": 15420 }, { "epoch": 117.78625954198473, "grad_norm": 0.0861218273639679, "learning_rate": 0.00014712482010132867, "loss": 0.0044, "step": 15430 }, { "epoch": 117.86259541984732, "grad_norm": 0.0880136713385582, "learning_rate": 0.0001469651091104364, "loss": 0.0045, "step": 15440 }, { "epoch": 117.93893129770993, "grad_norm": 0.08169922977685928, "learning_rate": 0.00014680540156144823, "loss": 0.0045, "step": 15450 }, { "epoch": 118.01526717557252, "grad_norm": 0.06039072945713997, "learning_rate": 0.0001466456976354901, "loss": 0.0042, "step": 15460 }, { "epoch": 118.09160305343511, "grad_norm": 0.08045228570699692, "learning_rate": 0.0001464859975136842, "loss": 0.0042, "step": 15470 }, { "epoch": 118.16793893129771, "grad_norm": 0.05844773352146149, "learning_rate": 0.00014632630137714818, "loss": 0.0043, "step": 15480 }, { "epoch": 118.2442748091603, "grad_norm": 0.07517963647842407, "learning_rate": 0.00014616660940699523, "loss": 0.0042, "step": 15490 }, { "epoch": 118.3206106870229, "grad_norm": 0.08174688369035721, "learning_rate": 0.00014600692178433377, "loss": 0.0043, "step": 15500 }, { "epoch": 118.3969465648855, "grad_norm": 0.06705062091350555, "learning_rate": 0.00014584723869026745, "loss": 0.004, "step": 15510 }, { "epoch": 118.47328244274809, "grad_norm": 0.0883251503109932, "learning_rate": 0.00014568756030589457, "loss": 0.004, "step": 15520 }, { "epoch": 118.54961832061069, "grad_norm": 0.054249655455350876, "learning_rate": 0.0001455278868123082, "loss": 0.0038, "step": 15530 }, { "epoch": 118.62595419847328, "grad_norm": 0.06506066024303436, "learning_rate": 0.00014536821839059589, "loss": 0.0043, "step": 15540 }, { "epoch": 118.70229007633588, "grad_norm": 0.09535917639732361, "learning_rate": 0.00014520855522183937, "loss": 0.0042, "step": 15550 }, { "epoch": 118.77862595419847, "grad_norm": 0.07628203928470612, "learning_rate": 0.00014504889748711444, "loss": 0.0036, "step": 15560 }, { "epoch": 118.85496183206106, "grad_norm": 0.08024118095636368, "learning_rate": 0.00014488924536749072, "loss": 0.0038, "step": 15570 }, { "epoch": 118.93129770992367, "grad_norm": 0.0971929132938385, "learning_rate": 0.00014472959904403157, "loss": 0.0043, "step": 15580 }, { "epoch": 119.00763358778626, "grad_norm": 0.08755987882614136, "learning_rate": 0.0001445699586977936, "loss": 0.0043, "step": 15590 }, { "epoch": 119.08396946564885, "grad_norm": 0.07786548882722855, "learning_rate": 0.00014441032450982676, "loss": 0.004, "step": 15600 }, { "epoch": 119.16030534351145, "grad_norm": 0.08550527691841125, "learning_rate": 0.00014425069666117402, "loss": 0.0044, "step": 15610 }, { "epoch": 119.23664122137404, "grad_norm": 0.07173260301351547, "learning_rate": 0.00014409107533287102, "loss": 0.0037, "step": 15620 }, { "epoch": 119.31297709923665, "grad_norm": 0.08647353202104568, "learning_rate": 0.00014393146070594619, "loss": 0.0042, "step": 15630 }, { "epoch": 119.38931297709924, "grad_norm": 0.06750532984733582, "learning_rate": 0.0001437718529614203, "loss": 0.0043, "step": 15640 }, { "epoch": 119.46564885496183, "grad_norm": 0.1489580124616623, "learning_rate": 0.0001436122522803062, "loss": 0.0041, "step": 15650 }, { "epoch": 119.54198473282443, "grad_norm": 0.07561802119016647, "learning_rate": 0.00014345265884360892, "loss": 0.0045, "step": 15660 }, { "epoch": 119.61832061068702, "grad_norm": 0.08979002386331558, "learning_rate": 0.00014329307283232517, "loss": 0.0047, "step": 15670 }, { "epoch": 119.69465648854961, "grad_norm": 0.11433997005224228, "learning_rate": 0.00014313349442744317, "loss": 0.0048, "step": 15680 }, { "epoch": 119.77099236641222, "grad_norm": 0.06561678647994995, "learning_rate": 0.00014297392380994265, "loss": 0.0039, "step": 15690 }, { "epoch": 119.8473282442748, "grad_norm": 0.06163293868303299, "learning_rate": 0.00014281436116079447, "loss": 0.0042, "step": 15700 }, { "epoch": 119.92366412213741, "grad_norm": 0.10638786107301712, "learning_rate": 0.00014265480666096034, "loss": 0.0045, "step": 15710 }, { "epoch": 120.0, "grad_norm": 0.06625767797231674, "learning_rate": 0.00014249526049139288, "loss": 0.0039, "step": 15720 }, { "epoch": 120.07633587786259, "grad_norm": 0.10135330259799957, "learning_rate": 0.00014233572283303523, "loss": 0.0046, "step": 15730 }, { "epoch": 120.1526717557252, "grad_norm": 0.08309479802846909, "learning_rate": 0.0001421761938668208, "loss": 0.0041, "step": 15740 }, { "epoch": 120.22900763358778, "grad_norm": 0.06800645589828491, "learning_rate": 0.00014201667377367324, "loss": 0.0041, "step": 15750 }, { "epoch": 120.30534351145039, "grad_norm": 0.06658114492893219, "learning_rate": 0.000141857162734506, "loss": 0.0047, "step": 15760 }, { "epoch": 120.38167938931298, "grad_norm": 0.08722653239965439, "learning_rate": 0.00014169766093022237, "loss": 0.004, "step": 15770 }, { "epoch": 120.45801526717557, "grad_norm": 0.07419612258672714, "learning_rate": 0.00014153816854171525, "loss": 0.0045, "step": 15780 }, { "epoch": 120.53435114503817, "grad_norm": 0.09253362566232681, "learning_rate": 0.00014137868574986667, "loss": 0.0041, "step": 15790 }, { "epoch": 120.61068702290076, "grad_norm": 0.08433306217193604, "learning_rate": 0.00014121921273554792, "loss": 0.0037, "step": 15800 }, { "epoch": 120.68702290076335, "grad_norm": 0.07484055310487747, "learning_rate": 0.00014105974967961905, "loss": 0.004, "step": 15810 }, { "epoch": 120.76335877862596, "grad_norm": 0.06360892951488495, "learning_rate": 0.000140900296762929, "loss": 0.0037, "step": 15820 }, { "epoch": 120.83969465648855, "grad_norm": 0.08628654479980469, "learning_rate": 0.0001407408541663151, "loss": 0.0039, "step": 15830 }, { "epoch": 120.91603053435115, "grad_norm": 0.08160670101642609, "learning_rate": 0.000140581422070603, "loss": 0.0035, "step": 15840 }, { "epoch": 120.99236641221374, "grad_norm": 0.0720876082777977, "learning_rate": 0.0001404220006566064, "loss": 0.004, "step": 15850 }, { "epoch": 121.06870229007633, "grad_norm": 0.07908736914396286, "learning_rate": 0.00014026259010512697, "loss": 0.0043, "step": 15860 }, { "epoch": 121.14503816793894, "grad_norm": 0.07022912800312042, "learning_rate": 0.00014010319059695397, "loss": 0.004, "step": 15870 }, { "epoch": 121.22137404580153, "grad_norm": 0.06441128998994827, "learning_rate": 0.00013994380231286427, "loss": 0.0041, "step": 15880 }, { "epoch": 121.29770992366412, "grad_norm": 0.07958530634641647, "learning_rate": 0.00013978442543362182, "loss": 0.0041, "step": 15890 }, { "epoch": 121.37404580152672, "grad_norm": 0.0759674459695816, "learning_rate": 0.0001396250601399778, "loss": 0.004, "step": 15900 }, { "epoch": 121.45038167938931, "grad_norm": 0.07271786034107208, "learning_rate": 0.00013946570661267017, "loss": 0.0042, "step": 15910 }, { "epoch": 121.52671755725191, "grad_norm": 0.11647696793079376, "learning_rate": 0.00013930636503242354, "loss": 0.0043, "step": 15920 }, { "epoch": 121.6030534351145, "grad_norm": 0.06785374879837036, "learning_rate": 0.000139147035579949, "loss": 0.0035, "step": 15930 }, { "epoch": 121.6793893129771, "grad_norm": 0.06078290939331055, "learning_rate": 0.00013898771843594399, "loss": 0.0038, "step": 15940 }, { "epoch": 121.7557251908397, "grad_norm": 0.06830593943595886, "learning_rate": 0.0001388284137810917, "loss": 0.0037, "step": 15950 }, { "epoch": 121.83206106870229, "grad_norm": 0.09083689004182816, "learning_rate": 0.00013866912179606145, "loss": 0.0041, "step": 15960 }, { "epoch": 121.90839694656489, "grad_norm": 0.06018128991127014, "learning_rate": 0.00013850984266150807, "loss": 0.0036, "step": 15970 }, { "epoch": 121.98473282442748, "grad_norm": 0.08897878229618073, "learning_rate": 0.00013835057655807177, "loss": 0.0039, "step": 15980 }, { "epoch": 122.06106870229007, "grad_norm": 0.09196311980485916, "learning_rate": 0.0001381913236663781, "loss": 0.0038, "step": 15990 }, { "epoch": 122.13740458015268, "grad_norm": 0.07686488330364227, "learning_rate": 0.0001380320841670375, "loss": 0.0035, "step": 16000 }, { "epoch": 122.21374045801527, "grad_norm": 0.10203896462917328, "learning_rate": 0.0001378728582406454, "loss": 0.0036, "step": 16010 }, { "epoch": 122.29007633587786, "grad_norm": 0.1012977585196495, "learning_rate": 0.00013771364606778158, "loss": 0.0041, "step": 16020 }, { "epoch": 122.36641221374046, "grad_norm": 0.09086854010820389, "learning_rate": 0.0001375544478290104, "loss": 0.0042, "step": 16030 }, { "epoch": 122.44274809160305, "grad_norm": 0.12291844189167023, "learning_rate": 0.00013739526370488045, "loss": 0.0039, "step": 16040 }, { "epoch": 122.51908396946565, "grad_norm": 0.07931002974510193, "learning_rate": 0.00013723609387592418, "loss": 0.0042, "step": 16050 }, { "epoch": 122.59541984732824, "grad_norm": 0.07123789191246033, "learning_rate": 0.0001370769385226579, "loss": 0.0036, "step": 16060 }, { "epoch": 122.67175572519083, "grad_norm": 0.07079660892486572, "learning_rate": 0.0001369177978255815, "loss": 0.0032, "step": 16070 }, { "epoch": 122.74809160305344, "grad_norm": 0.06851878762245178, "learning_rate": 0.00013675867196517822, "loss": 0.0038, "step": 16080 }, { "epoch": 122.82442748091603, "grad_norm": 0.09032273292541504, "learning_rate": 0.0001365995611219145, "loss": 0.004, "step": 16090 }, { "epoch": 122.90076335877862, "grad_norm": 0.06766428053379059, "learning_rate": 0.0001364404654762398, "loss": 0.0041, "step": 16100 }, { "epoch": 122.97709923664122, "grad_norm": 0.07547888159751892, "learning_rate": 0.0001362813852085862, "loss": 0.004, "step": 16110 }, { "epoch": 123.05343511450381, "grad_norm": 0.0732823982834816, "learning_rate": 0.00013612232049936844, "loss": 0.0041, "step": 16120 }, { "epoch": 123.12977099236642, "grad_norm": 0.06699147075414658, "learning_rate": 0.0001359632715289837, "loss": 0.004, "step": 16130 }, { "epoch": 123.20610687022901, "grad_norm": 0.06599820405244827, "learning_rate": 0.00013580423847781107, "loss": 0.0038, "step": 16140 }, { "epoch": 123.2824427480916, "grad_norm": 0.056410279124975204, "learning_rate": 0.00013564522152621186, "loss": 0.0036, "step": 16150 }, { "epoch": 123.3587786259542, "grad_norm": 0.07993070781230927, "learning_rate": 0.0001354862208545289, "loss": 0.0035, "step": 16160 }, { "epoch": 123.43511450381679, "grad_norm": 0.08269251883029938, "learning_rate": 0.00013532723664308664, "loss": 0.0042, "step": 16170 }, { "epoch": 123.5114503816794, "grad_norm": 0.06854764372110367, "learning_rate": 0.00013516826907219098, "loss": 0.0039, "step": 16180 }, { "epoch": 123.58778625954199, "grad_norm": 0.06519988179206848, "learning_rate": 0.00013500931832212862, "loss": 0.0041, "step": 16190 }, { "epoch": 123.66412213740458, "grad_norm": 0.09018848091363907, "learning_rate": 0.00013485038457316766, "loss": 0.0043, "step": 16200 }, { "epoch": 123.74045801526718, "grad_norm": 0.07244815677404404, "learning_rate": 0.0001346914680055565, "loss": 0.0037, "step": 16210 }, { "epoch": 123.81679389312977, "grad_norm": 0.08038055896759033, "learning_rate": 0.00013453256879952425, "loss": 0.0045, "step": 16220 }, { "epoch": 123.89312977099236, "grad_norm": 0.09511376172304153, "learning_rate": 0.00013437368713528034, "loss": 0.004, "step": 16230 }, { "epoch": 123.96946564885496, "grad_norm": 0.06873805820941925, "learning_rate": 0.00013421482319301415, "loss": 0.0044, "step": 16240 }, { "epoch": 124.04580152671755, "grad_norm": 0.08453937619924545, "learning_rate": 0.0001340559771528952, "loss": 0.0039, "step": 16250 }, { "epoch": 124.12213740458016, "grad_norm": 0.09666585177183151, "learning_rate": 0.00013389714919507246, "loss": 0.004, "step": 16260 }, { "epoch": 124.19847328244275, "grad_norm": 0.12770633399486542, "learning_rate": 0.00013373833949967455, "loss": 0.0041, "step": 16270 }, { "epoch": 124.27480916030534, "grad_norm": 0.10505737364292145, "learning_rate": 0.00013357954824680938, "loss": 0.0043, "step": 16280 }, { "epoch": 124.35114503816794, "grad_norm": 0.0701487734913826, "learning_rate": 0.00013342077561656385, "loss": 0.0038, "step": 16290 }, { "epoch": 124.42748091603053, "grad_norm": 0.07603242248296738, "learning_rate": 0.00013326202178900377, "loss": 0.0034, "step": 16300 }, { "epoch": 124.50381679389314, "grad_norm": 0.06709441542625427, "learning_rate": 0.00013310328694417375, "loss": 0.0035, "step": 16310 }, { "epoch": 124.58015267175573, "grad_norm": 0.09169193357229233, "learning_rate": 0.0001329445712620967, "loss": 0.0041, "step": 16320 }, { "epoch": 124.65648854961832, "grad_norm": 0.07088664174079895, "learning_rate": 0.00013278587492277383, "loss": 0.0042, "step": 16330 }, { "epoch": 124.73282442748092, "grad_norm": 0.09668882191181183, "learning_rate": 0.00013262719810618455, "loss": 0.0039, "step": 16340 }, { "epoch": 124.80916030534351, "grad_norm": 0.07583287358283997, "learning_rate": 0.00013246854099228591, "loss": 0.0039, "step": 16350 }, { "epoch": 124.8854961832061, "grad_norm": 0.0997387170791626, "learning_rate": 0.00013230990376101283, "loss": 0.0043, "step": 16360 }, { "epoch": 124.9618320610687, "grad_norm": 0.07344502955675125, "learning_rate": 0.00013215128659227758, "loss": 0.0042, "step": 16370 }, { "epoch": 125.0381679389313, "grad_norm": 0.0726001188158989, "learning_rate": 0.0001319926896659696, "loss": 0.0043, "step": 16380 }, { "epoch": 125.1145038167939, "grad_norm": 0.08041812479496002, "learning_rate": 0.0001318341131619555, "loss": 0.0037, "step": 16390 }, { "epoch": 125.19083969465649, "grad_norm": 0.08161003887653351, "learning_rate": 0.00013167555726007865, "loss": 0.0038, "step": 16400 }, { "epoch": 125.26717557251908, "grad_norm": 0.07059843093156815, "learning_rate": 0.00013151702214015922, "loss": 0.0036, "step": 16410 }, { "epoch": 125.34351145038168, "grad_norm": 0.09790705144405365, "learning_rate": 0.00013135850798199353, "loss": 0.0037, "step": 16420 }, { "epoch": 125.41984732824427, "grad_norm": 0.07946792989969254, "learning_rate": 0.00013120001496535433, "loss": 0.0037, "step": 16430 }, { "epoch": 125.49618320610686, "grad_norm": 0.06541581451892853, "learning_rate": 0.0001310415432699904, "loss": 0.0039, "step": 16440 }, { "epoch": 125.57251908396947, "grad_norm": 0.11630236357450485, "learning_rate": 0.00013088309307562613, "loss": 0.0037, "step": 16450 }, { "epoch": 125.64885496183206, "grad_norm": 0.08801501989364624, "learning_rate": 0.0001307246645619618, "loss": 0.0039, "step": 16460 }, { "epoch": 125.72519083969466, "grad_norm": 0.08378341048955917, "learning_rate": 0.00013056625790867294, "loss": 0.0039, "step": 16470 }, { "epoch": 125.80152671755725, "grad_norm": 0.13479085266590118, "learning_rate": 0.00013040787329541032, "loss": 0.0048, "step": 16480 }, { "epoch": 125.87786259541984, "grad_norm": 0.10700758546590805, "learning_rate": 0.00013024951090179969, "loss": 0.0046, "step": 16490 }, { "epoch": 125.95419847328245, "grad_norm": 0.1004539355635643, "learning_rate": 0.0001300911709074417, "loss": 0.0047, "step": 16500 }, { "epoch": 126.03053435114504, "grad_norm": 0.1030244305729866, "learning_rate": 0.00012993285349191145, "loss": 0.0047, "step": 16510 }, { "epoch": 126.10687022900764, "grad_norm": 0.061489276587963104, "learning_rate": 0.00012977455883475853, "loss": 0.0041, "step": 16520 }, { "epoch": 126.18320610687023, "grad_norm": 0.08434920012950897, "learning_rate": 0.00012961628711550676, "loss": 0.0042, "step": 16530 }, { "epoch": 126.25954198473282, "grad_norm": 0.07435864210128784, "learning_rate": 0.00012945803851365377, "loss": 0.0039, "step": 16540 }, { "epoch": 126.33587786259542, "grad_norm": 0.07408982515335083, "learning_rate": 0.00012929981320867117, "loss": 0.0042, "step": 16550 }, { "epoch": 126.41221374045801, "grad_norm": 0.08683385699987411, "learning_rate": 0.000129141611380004, "loss": 0.0038, "step": 16560 }, { "epoch": 126.4885496183206, "grad_norm": 0.07510925084352493, "learning_rate": 0.0001289834332070708, "loss": 0.0036, "step": 16570 }, { "epoch": 126.56488549618321, "grad_norm": 0.045522525906562805, "learning_rate": 0.0001288252788692632, "loss": 0.0038, "step": 16580 }, { "epoch": 126.6412213740458, "grad_norm": 0.07399231195449829, "learning_rate": 0.00012866714854594584, "loss": 0.0036, "step": 16590 }, { "epoch": 126.7175572519084, "grad_norm": 0.06991526484489441, "learning_rate": 0.00012850904241645602, "loss": 0.0038, "step": 16600 }, { "epoch": 126.79389312977099, "grad_norm": 0.07447106391191483, "learning_rate": 0.00012835096066010377, "loss": 0.0037, "step": 16610 }, { "epoch": 126.87022900763358, "grad_norm": 0.07078109681606293, "learning_rate": 0.00012819290345617135, "loss": 0.0036, "step": 16620 }, { "epoch": 126.94656488549619, "grad_norm": 0.06123663857579231, "learning_rate": 0.0001280348709839133, "loss": 0.004, "step": 16630 }, { "epoch": 127.02290076335878, "grad_norm": 0.06271295994520187, "learning_rate": 0.00012787686342255594, "loss": 0.0039, "step": 16640 }, { "epoch": 127.09923664122137, "grad_norm": 0.05625668913125992, "learning_rate": 0.00012771888095129748, "loss": 0.0037, "step": 16650 }, { "epoch": 127.17557251908397, "grad_norm": 0.05835595726966858, "learning_rate": 0.00012756092374930757, "loss": 0.0034, "step": 16660 }, { "epoch": 127.25190839694656, "grad_norm": 0.06823009997606277, "learning_rate": 0.0001274029919957273, "loss": 0.0034, "step": 16670 }, { "epoch": 127.32824427480917, "grad_norm": 0.07502990961074829, "learning_rate": 0.00012724508586966888, "loss": 0.0038, "step": 16680 }, { "epoch": 127.40458015267176, "grad_norm": 0.06605974584817886, "learning_rate": 0.00012708720555021534, "loss": 0.0035, "step": 16690 }, { "epoch": 127.48091603053435, "grad_norm": 0.06436695158481598, "learning_rate": 0.0001269293512164206, "loss": 0.004, "step": 16700 }, { "epoch": 127.55725190839695, "grad_norm": 0.06582081317901611, "learning_rate": 0.00012677152304730905, "loss": 0.0036, "step": 16710 }, { "epoch": 127.63358778625954, "grad_norm": 0.07707630097866058, "learning_rate": 0.00012661372122187533, "loss": 0.0036, "step": 16720 }, { "epoch": 127.70992366412214, "grad_norm": 0.06870464235544205, "learning_rate": 0.00012645594591908433, "loss": 0.0035, "step": 16730 }, { "epoch": 127.78625954198473, "grad_norm": 0.09410656243562698, "learning_rate": 0.00012629819731787084, "loss": 0.0042, "step": 16740 }, { "epoch": 127.86259541984732, "grad_norm": 0.07334926724433899, "learning_rate": 0.0001261404755971392, "loss": 0.0034, "step": 16750 }, { "epoch": 127.93893129770993, "grad_norm": 0.07354626059532166, "learning_rate": 0.0001259827809357635, "loss": 0.0037, "step": 16760 }, { "epoch": 128.01526717557252, "grad_norm": 0.08415473997592926, "learning_rate": 0.000125825113512587, "loss": 0.0035, "step": 16770 }, { "epoch": 128.0916030534351, "grad_norm": 0.08924662321805954, "learning_rate": 0.00012566747350642213, "loss": 0.0035, "step": 16780 }, { "epoch": 128.1679389312977, "grad_norm": 0.0663425549864769, "learning_rate": 0.00012550986109605014, "loss": 0.0037, "step": 16790 }, { "epoch": 128.24427480916032, "grad_norm": 0.08386814594268799, "learning_rate": 0.0001253522764602211, "loss": 0.0041, "step": 16800 }, { "epoch": 128.3206106870229, "grad_norm": 0.05901413783431053, "learning_rate": 0.0001251947197776535, "loss": 0.0036, "step": 16810 }, { "epoch": 128.3969465648855, "grad_norm": 0.1449218988418579, "learning_rate": 0.00012503719122703414, "loss": 0.004, "step": 16820 }, { "epoch": 128.4732824427481, "grad_norm": 0.10901849716901779, "learning_rate": 0.00012487969098701795, "loss": 0.0037, "step": 16830 }, { "epoch": 128.54961832061068, "grad_norm": 0.08588291704654694, "learning_rate": 0.00012472221923622777, "loss": 0.0039, "step": 16840 }, { "epoch": 128.6259541984733, "grad_norm": 0.06304645538330078, "learning_rate": 0.00012456477615325402, "loss": 0.0041, "step": 16850 }, { "epoch": 128.70229007633588, "grad_norm": 0.07843796163797379, "learning_rate": 0.0001244073619166547, "loss": 0.0039, "step": 16860 }, { "epoch": 128.77862595419847, "grad_norm": 0.07134220004081726, "learning_rate": 0.0001242499767049551, "loss": 0.0036, "step": 16870 }, { "epoch": 128.85496183206106, "grad_norm": 0.08122045546770096, "learning_rate": 0.0001240926206966475, "loss": 0.0037, "step": 16880 }, { "epoch": 128.93129770992365, "grad_norm": 0.06688018143177032, "learning_rate": 0.00012393529407019117, "loss": 0.0036, "step": 16890 }, { "epoch": 129.00763358778627, "grad_norm": 0.06648776680231094, "learning_rate": 0.00012377799700401205, "loss": 0.0037, "step": 16900 }, { "epoch": 129.08396946564886, "grad_norm": 0.07223814725875854, "learning_rate": 0.00012362072967650245, "loss": 0.0034, "step": 16910 }, { "epoch": 129.16030534351145, "grad_norm": 0.06312404572963715, "learning_rate": 0.000123463492266021, "loss": 0.0034, "step": 16920 }, { "epoch": 129.23664122137404, "grad_norm": 0.1309927999973297, "learning_rate": 0.00012330628495089253, "loss": 0.0036, "step": 16930 }, { "epoch": 129.31297709923663, "grad_norm": 0.06880945712327957, "learning_rate": 0.00012314910790940756, "loss": 0.0039, "step": 16940 }, { "epoch": 129.38931297709922, "grad_norm": 0.08583508431911469, "learning_rate": 0.00012299196131982236, "loss": 0.0036, "step": 16950 }, { "epoch": 129.46564885496184, "grad_norm": 0.07069321721792221, "learning_rate": 0.00012283484536035862, "loss": 0.0034, "step": 16960 }, { "epoch": 129.54198473282443, "grad_norm": 0.10379092395305634, "learning_rate": 0.00012267776020920333, "loss": 0.0038, "step": 16970 }, { "epoch": 129.61832061068702, "grad_norm": 0.10026632994413376, "learning_rate": 0.0001225207060445086, "loss": 0.0034, "step": 16980 }, { "epoch": 129.6946564885496, "grad_norm": 0.08621808141469955, "learning_rate": 0.00012236368304439124, "loss": 0.0037, "step": 16990 }, { "epoch": 129.7709923664122, "grad_norm": 0.08666995912790298, "learning_rate": 0.00012220669138693287, "loss": 0.0036, "step": 17000 }, { "epoch": 129.84732824427482, "grad_norm": 0.06064877286553383, "learning_rate": 0.00012204973125017949, "loss": 0.0032, "step": 17010 }, { "epoch": 129.9236641221374, "grad_norm": 0.08156301081180573, "learning_rate": 0.00012189280281214126, "loss": 0.0035, "step": 17020 }, { "epoch": 130.0, "grad_norm": 0.09948518872261047, "learning_rate": 0.0001217359062507927, "loss": 0.0034, "step": 17030 }, { "epoch": 130.0763358778626, "grad_norm": 0.07379215955734253, "learning_rate": 0.0001215790417440718, "loss": 0.0033, "step": 17040 }, { "epoch": 130.15267175572518, "grad_norm": 0.06242477521300316, "learning_rate": 0.00012142220946988046, "loss": 0.0033, "step": 17050 }, { "epoch": 130.2290076335878, "grad_norm": 0.06902344524860382, "learning_rate": 0.00012126540960608387, "loss": 0.0034, "step": 17060 }, { "epoch": 130.3053435114504, "grad_norm": 0.07440471649169922, "learning_rate": 0.00012110864233051056, "loss": 0.0035, "step": 17070 }, { "epoch": 130.38167938931298, "grad_norm": 0.07727916538715363, "learning_rate": 0.00012095190782095208, "loss": 0.0035, "step": 17080 }, { "epoch": 130.45801526717557, "grad_norm": 0.06237182393670082, "learning_rate": 0.00012079520625516274, "loss": 0.0032, "step": 17090 }, { "epoch": 130.53435114503816, "grad_norm": 0.07339644432067871, "learning_rate": 0.00012063853781085961, "loss": 0.0037, "step": 17100 }, { "epoch": 130.61068702290078, "grad_norm": 0.05747790262103081, "learning_rate": 0.00012048190266572216, "loss": 0.0031, "step": 17110 }, { "epoch": 130.68702290076337, "grad_norm": 0.07737213373184204, "learning_rate": 0.000120325300997392, "loss": 0.0039, "step": 17120 }, { "epoch": 130.76335877862596, "grad_norm": 0.09122870862483978, "learning_rate": 0.00012016873298347291, "loss": 0.0039, "step": 17130 }, { "epoch": 130.83969465648855, "grad_norm": 0.09049054980278015, "learning_rate": 0.00012001219880153044, "loss": 0.0038, "step": 17140 }, { "epoch": 130.91603053435114, "grad_norm": 0.09401526302099228, "learning_rate": 0.00011985569862909176, "loss": 0.0035, "step": 17150 }, { "epoch": 130.99236641221373, "grad_norm": 0.08840252459049225, "learning_rate": 0.00011969923264364546, "loss": 0.004, "step": 17160 }, { "epoch": 131.06870229007635, "grad_norm": 0.07141760736703873, "learning_rate": 0.00011954280102264141, "loss": 0.0036, "step": 17170 }, { "epoch": 131.14503816793894, "grad_norm": 0.10286658257246017, "learning_rate": 0.00011938640394349046, "loss": 0.0035, "step": 17180 }, { "epoch": 131.22137404580153, "grad_norm": 0.08482830226421356, "learning_rate": 0.0001192300415835643, "loss": 0.0037, "step": 17190 }, { "epoch": 131.29770992366412, "grad_norm": 0.06905300915241241, "learning_rate": 0.0001190737141201953, "loss": 0.0038, "step": 17200 }, { "epoch": 131.3740458015267, "grad_norm": 0.06460582464933395, "learning_rate": 0.00011891742173067611, "loss": 0.0035, "step": 17210 }, { "epoch": 131.45038167938932, "grad_norm": 0.07347679883241653, "learning_rate": 0.0001187611645922598, "loss": 0.004, "step": 17220 }, { "epoch": 131.5267175572519, "grad_norm": 0.05899829789996147, "learning_rate": 0.00011860494288215915, "loss": 0.0037, "step": 17230 }, { "epoch": 131.6030534351145, "grad_norm": 0.08011378347873688, "learning_rate": 0.00011844875677754725, "loss": 0.0034, "step": 17240 }, { "epoch": 131.6793893129771, "grad_norm": 0.0741611197590828, "learning_rate": 0.00011829260645555634, "loss": 0.0036, "step": 17250 }, { "epoch": 131.75572519083968, "grad_norm": 0.07178223133087158, "learning_rate": 0.0001181364920932783, "loss": 0.0038, "step": 17260 }, { "epoch": 131.8320610687023, "grad_norm": 0.07192444056272507, "learning_rate": 0.00011798041386776422, "loss": 0.0037, "step": 17270 }, { "epoch": 131.9083969465649, "grad_norm": 0.06405699253082275, "learning_rate": 0.00011782437195602413, "loss": 0.0043, "step": 17280 }, { "epoch": 131.98473282442748, "grad_norm": 0.07182040810585022, "learning_rate": 0.00011766836653502694, "loss": 0.0034, "step": 17290 }, { "epoch": 132.06106870229007, "grad_norm": 0.07316207140684128, "learning_rate": 0.0001175123977817002, "loss": 0.0038, "step": 17300 }, { "epoch": 132.13740458015266, "grad_norm": 0.0631694570183754, "learning_rate": 0.00011735646587292976, "loss": 0.0031, "step": 17310 }, { "epoch": 132.21374045801528, "grad_norm": 0.0679394081234932, "learning_rate": 0.00011720057098555981, "loss": 0.0035, "step": 17320 }, { "epoch": 132.29007633587787, "grad_norm": 0.062247756868600845, "learning_rate": 0.0001170447132963925, "loss": 0.0032, "step": 17330 }, { "epoch": 132.36641221374046, "grad_norm": 0.08750530332326889, "learning_rate": 0.00011688889298218776, "loss": 0.0034, "step": 17340 }, { "epoch": 132.44274809160305, "grad_norm": 0.09421888738870621, "learning_rate": 0.00011673311021966323, "loss": 0.0032, "step": 17350 }, { "epoch": 132.51908396946564, "grad_norm": 0.07298508286476135, "learning_rate": 0.0001165773651854938, "loss": 0.0034, "step": 17360 }, { "epoch": 132.59541984732823, "grad_norm": 0.08681797981262207, "learning_rate": 0.00011642165805631176, "loss": 0.0035, "step": 17370 }, { "epoch": 132.67175572519085, "grad_norm": 0.0828031376004219, "learning_rate": 0.00011626598900870631, "loss": 0.0034, "step": 17380 }, { "epoch": 132.74809160305344, "grad_norm": 0.08460311591625214, "learning_rate": 0.0001161103582192234, "loss": 0.0042, "step": 17390 }, { "epoch": 132.82442748091603, "grad_norm": 0.08951039612293243, "learning_rate": 0.00011595476586436575, "loss": 0.0042, "step": 17400 }, { "epoch": 132.90076335877862, "grad_norm": 0.09603152424097061, "learning_rate": 0.00011579921212059239, "loss": 0.0036, "step": 17410 }, { "epoch": 132.9770992366412, "grad_norm": 0.07565827667713165, "learning_rate": 0.00011564369716431853, "loss": 0.0038, "step": 17420 }, { "epoch": 133.05343511450383, "grad_norm": 0.07499822229146957, "learning_rate": 0.00011548822117191547, "loss": 0.0033, "step": 17430 }, { "epoch": 133.12977099236642, "grad_norm": 0.067399762570858, "learning_rate": 0.00011533278431971027, "loss": 0.0034, "step": 17440 }, { "epoch": 133.206106870229, "grad_norm": 0.07530099898576736, "learning_rate": 0.00011517738678398568, "loss": 0.0037, "step": 17450 }, { "epoch": 133.2824427480916, "grad_norm": 0.0675441324710846, "learning_rate": 0.00011502202874097973, "loss": 0.0033, "step": 17460 }, { "epoch": 133.3587786259542, "grad_norm": 0.052385348826646805, "learning_rate": 0.00011486671036688575, "loss": 0.0035, "step": 17470 }, { "epoch": 133.4351145038168, "grad_norm": 0.06632131338119507, "learning_rate": 0.00011471143183785209, "loss": 0.0033, "step": 17480 }, { "epoch": 133.5114503816794, "grad_norm": 0.06676006317138672, "learning_rate": 0.00011455619332998184, "loss": 0.0035, "step": 17490 }, { "epoch": 133.58778625954199, "grad_norm": 0.09083187580108643, "learning_rate": 0.00011440099501933276, "loss": 0.0036, "step": 17500 }, { "epoch": 133.66412213740458, "grad_norm": 0.0753694474697113, "learning_rate": 0.00011424583708191704, "loss": 0.0036, "step": 17510 }, { "epoch": 133.74045801526717, "grad_norm": 0.07600123435258865, "learning_rate": 0.00011409071969370097, "loss": 0.0035, "step": 17520 }, { "epoch": 133.81679389312978, "grad_norm": 0.0703793615102768, "learning_rate": 0.00011393564303060497, "loss": 0.0033, "step": 17530 }, { "epoch": 133.89312977099237, "grad_norm": 0.08673941344022751, "learning_rate": 0.0001137806072685033, "loss": 0.0035, "step": 17540 }, { "epoch": 133.96946564885496, "grad_norm": 0.0684499442577362, "learning_rate": 0.00011362561258322365, "loss": 0.0034, "step": 17550 }, { "epoch": 134.04580152671755, "grad_norm": 0.08876513689756393, "learning_rate": 0.0001134706591505473, "loss": 0.0033, "step": 17560 }, { "epoch": 134.12213740458014, "grad_norm": 0.07127581536769867, "learning_rate": 0.00011331574714620873, "loss": 0.0032, "step": 17570 }, { "epoch": 134.19847328244273, "grad_norm": 0.07703084498643875, "learning_rate": 0.00011316087674589532, "loss": 0.0035, "step": 17580 }, { "epoch": 134.27480916030535, "grad_norm": 0.13223016262054443, "learning_rate": 0.00011300604812524737, "loss": 0.0039, "step": 17590 }, { "epoch": 134.35114503816794, "grad_norm": 0.09591227024793625, "learning_rate": 0.00011285126145985777, "loss": 0.0035, "step": 17600 }, { "epoch": 134.42748091603053, "grad_norm": 0.06411512941122055, "learning_rate": 0.0001126965169252718, "loss": 0.0034, "step": 17610 }, { "epoch": 134.50381679389312, "grad_norm": 0.08299662172794342, "learning_rate": 0.00011254181469698703, "loss": 0.0036, "step": 17620 }, { "epoch": 134.5801526717557, "grad_norm": 0.0761709064245224, "learning_rate": 0.00011238715495045293, "loss": 0.0035, "step": 17630 }, { "epoch": 134.65648854961833, "grad_norm": 0.07936061918735504, "learning_rate": 0.00011223253786107089, "loss": 0.0037, "step": 17640 }, { "epoch": 134.73282442748092, "grad_norm": 0.0955384224653244, "learning_rate": 0.00011207796360419396, "loss": 0.0037, "step": 17650 }, { "epoch": 134.8091603053435, "grad_norm": 0.07221472263336182, "learning_rate": 0.00011192343235512648, "loss": 0.0036, "step": 17660 }, { "epoch": 134.8854961832061, "grad_norm": 0.09242245554924011, "learning_rate": 0.00011176894428912414, "loss": 0.0036, "step": 17670 }, { "epoch": 134.9618320610687, "grad_norm": 0.08328031748533249, "learning_rate": 0.00011161449958139354, "loss": 0.0035, "step": 17680 }, { "epoch": 135.0381679389313, "grad_norm": 0.08879037946462631, "learning_rate": 0.00011146009840709217, "loss": 0.0034, "step": 17690 }, { "epoch": 135.1145038167939, "grad_norm": 0.08053122460842133, "learning_rate": 0.00011130574094132824, "loss": 0.0035, "step": 17700 }, { "epoch": 135.1908396946565, "grad_norm": 0.06158880144357681, "learning_rate": 0.00011115142735916015, "loss": 0.0036, "step": 17710 }, { "epoch": 135.26717557251908, "grad_norm": 0.07538311928510666, "learning_rate": 0.00011099715783559676, "loss": 0.0032, "step": 17720 }, { "epoch": 135.34351145038167, "grad_norm": 0.07238037884235382, "learning_rate": 0.0001108429325455969, "loss": 0.0035, "step": 17730 }, { "epoch": 135.4198473282443, "grad_norm": 0.09602409601211548, "learning_rate": 0.00011068875166406914, "loss": 0.0041, "step": 17740 }, { "epoch": 135.49618320610688, "grad_norm": 0.10277482122182846, "learning_rate": 0.00011053461536587182, "loss": 0.0034, "step": 17750 }, { "epoch": 135.57251908396947, "grad_norm": 0.07063233852386475, "learning_rate": 0.00011038052382581261, "loss": 0.0033, "step": 17760 }, { "epoch": 135.64885496183206, "grad_norm": 0.08515869826078415, "learning_rate": 0.00011022647721864848, "loss": 0.0033, "step": 17770 }, { "epoch": 135.72519083969465, "grad_norm": 0.0889614149928093, "learning_rate": 0.00011007247571908546, "loss": 0.0038, "step": 17780 }, { "epoch": 135.80152671755727, "grad_norm": 0.10914881527423859, "learning_rate": 0.00010991851950177832, "loss": 0.004, "step": 17790 }, { "epoch": 135.87786259541986, "grad_norm": 0.05489872395992279, "learning_rate": 0.00010976460874133058, "loss": 0.0035, "step": 17800 }, { "epoch": 135.95419847328245, "grad_norm": 0.08776187151670456, "learning_rate": 0.00010961074361229422, "loss": 0.0035, "step": 17810 }, { "epoch": 136.03053435114504, "grad_norm": 0.0801672637462616, "learning_rate": 0.00010945692428916931, "loss": 0.0035, "step": 17820 }, { "epoch": 136.10687022900763, "grad_norm": 0.07157515734434128, "learning_rate": 0.00010930315094640416, "loss": 0.0035, "step": 17830 }, { "epoch": 136.18320610687022, "grad_norm": 0.07541216909885406, "learning_rate": 0.00010914942375839486, "loss": 0.0034, "step": 17840 }, { "epoch": 136.25954198473283, "grad_norm": 0.07464343309402466, "learning_rate": 0.00010899574289948508, "loss": 0.0033, "step": 17850 }, { "epoch": 136.33587786259542, "grad_norm": 0.08599980175495148, "learning_rate": 0.00010884210854396605, "loss": 0.0033, "step": 17860 }, { "epoch": 136.41221374045801, "grad_norm": 0.08143074810504913, "learning_rate": 0.00010868852086607629, "loss": 0.0034, "step": 17870 }, { "epoch": 136.4885496183206, "grad_norm": 0.08908207714557648, "learning_rate": 0.00010853498004000128, "loss": 0.0033, "step": 17880 }, { "epoch": 136.5648854961832, "grad_norm": 0.06897306442260742, "learning_rate": 0.00010838148623987335, "loss": 0.0032, "step": 17890 }, { "epoch": 136.6412213740458, "grad_norm": 0.08334552496671677, "learning_rate": 0.00010822803963977158, "loss": 0.0033, "step": 17900 }, { "epoch": 136.7175572519084, "grad_norm": 0.07511241734027863, "learning_rate": 0.00010807464041372156, "loss": 0.0033, "step": 17910 }, { "epoch": 136.793893129771, "grad_norm": 0.06025998294353485, "learning_rate": 0.00010792128873569495, "loss": 0.0035, "step": 17920 }, { "epoch": 136.87022900763358, "grad_norm": 0.059994667768478394, "learning_rate": 0.00010776798477960975, "loss": 0.0034, "step": 17930 }, { "epoch": 136.94656488549617, "grad_norm": 0.10059051215648651, "learning_rate": 0.00010761472871932965, "loss": 0.0032, "step": 17940 }, { "epoch": 137.0229007633588, "grad_norm": 0.06325399875640869, "learning_rate": 0.00010746152072866405, "loss": 0.0035, "step": 17950 }, { "epoch": 137.09923664122138, "grad_norm": 0.07879184931516647, "learning_rate": 0.0001073083609813679, "loss": 0.0033, "step": 17960 }, { "epoch": 137.17557251908397, "grad_norm": 0.0698622316122055, "learning_rate": 0.00010715524965114141, "loss": 0.0031, "step": 17970 }, { "epoch": 137.25190839694656, "grad_norm": 0.06485337764024734, "learning_rate": 0.00010700218691162981, "loss": 0.0032, "step": 17980 }, { "epoch": 137.32824427480915, "grad_norm": 0.07517604529857635, "learning_rate": 0.00010684917293642334, "loss": 0.0036, "step": 17990 }, { "epoch": 137.40458015267177, "grad_norm": 0.10851731896400452, "learning_rate": 0.00010669620789905688, "loss": 0.0034, "step": 18000 }, { "epoch": 137.48091603053436, "grad_norm": 0.07510973513126373, "learning_rate": 0.00010654329197300972, "loss": 0.0035, "step": 18010 }, { "epoch": 137.55725190839695, "grad_norm": 0.06691794842481613, "learning_rate": 0.00010639042533170565, "loss": 0.0034, "step": 18020 }, { "epoch": 137.63358778625954, "grad_norm": 0.056504182517528534, "learning_rate": 0.00010623760814851235, "loss": 0.0032, "step": 18030 }, { "epoch": 137.70992366412213, "grad_norm": 0.09167207032442093, "learning_rate": 0.00010608484059674157, "loss": 0.0036, "step": 18040 }, { "epoch": 137.78625954198472, "grad_norm": 0.06533283740282059, "learning_rate": 0.00010593212284964875, "loss": 0.0032, "step": 18050 }, { "epoch": 137.86259541984734, "grad_norm": 0.09848375618457794, "learning_rate": 0.00010577945508043265, "loss": 0.0034, "step": 18060 }, { "epoch": 137.93893129770993, "grad_norm": 0.08985587954521179, "learning_rate": 0.00010562683746223573, "loss": 0.0036, "step": 18070 }, { "epoch": 138.01526717557252, "grad_norm": 0.054671499878168106, "learning_rate": 0.00010547427016814321, "loss": 0.0035, "step": 18080 }, { "epoch": 138.0916030534351, "grad_norm": 0.06228265538811684, "learning_rate": 0.00010532175337118342, "loss": 0.0031, "step": 18090 }, { "epoch": 138.1679389312977, "grad_norm": 0.07576092332601547, "learning_rate": 0.00010516928724432743, "loss": 0.0031, "step": 18100 }, { "epoch": 138.24427480916032, "grad_norm": 0.06521950662136078, "learning_rate": 0.0001050168719604887, "loss": 0.0032, "step": 18110 }, { "epoch": 138.3206106870229, "grad_norm": 0.06634350121021271, "learning_rate": 0.00010486450769252318, "loss": 0.0032, "step": 18120 }, { "epoch": 138.3969465648855, "grad_norm": 0.06394030153751373, "learning_rate": 0.00010471219461322892, "loss": 0.0032, "step": 18130 }, { "epoch": 138.4732824427481, "grad_norm": 0.08032191544771194, "learning_rate": 0.00010455993289534584, "loss": 0.0031, "step": 18140 }, { "epoch": 138.54961832061068, "grad_norm": 0.10912154614925385, "learning_rate": 0.00010440772271155573, "loss": 0.0033, "step": 18150 }, { "epoch": 138.6259541984733, "grad_norm": 0.07789362967014313, "learning_rate": 0.00010425556423448183, "loss": 0.0032, "step": 18160 }, { "epoch": 138.70229007633588, "grad_norm": 0.07243809849023819, "learning_rate": 0.00010410345763668884, "loss": 0.0032, "step": 18170 }, { "epoch": 138.77862595419847, "grad_norm": 0.06689304858446121, "learning_rate": 0.00010395140309068256, "loss": 0.0032, "step": 18180 }, { "epoch": 138.85496183206106, "grad_norm": 0.08166931569576263, "learning_rate": 0.0001037994007689097, "loss": 0.0031, "step": 18190 }, { "epoch": 138.93129770992365, "grad_norm": 0.06704624742269516, "learning_rate": 0.0001036474508437579, "loss": 0.0029, "step": 18200 }, { "epoch": 139.00763358778627, "grad_norm": 0.08336254209280014, "learning_rate": 0.00010349555348755524, "loss": 0.0034, "step": 18210 }, { "epoch": 139.08396946564886, "grad_norm": 0.08899963647127151, "learning_rate": 0.0001033437088725702, "loss": 0.0029, "step": 18220 }, { "epoch": 139.16030534351145, "grad_norm": 0.05495620146393776, "learning_rate": 0.00010319191717101151, "loss": 0.0029, "step": 18230 }, { "epoch": 139.23664122137404, "grad_norm": 0.06985968351364136, "learning_rate": 0.00010304017855502786, "loss": 0.0028, "step": 18240 }, { "epoch": 139.31297709923663, "grad_norm": 0.08164716511964798, "learning_rate": 0.00010288849319670771, "loss": 0.0029, "step": 18250 }, { "epoch": 139.38931297709922, "grad_norm": 0.06735731661319733, "learning_rate": 0.0001027368612680791, "loss": 0.0035, "step": 18260 }, { "epoch": 139.46564885496184, "grad_norm": 0.06297899037599564, "learning_rate": 0.00010258528294110953, "loss": 0.0033, "step": 18270 }, { "epoch": 139.54198473282443, "grad_norm": 0.0892508402466774, "learning_rate": 0.00010243375838770578, "loss": 0.0033, "step": 18280 }, { "epoch": 139.61832061068702, "grad_norm": 0.10426025092601776, "learning_rate": 0.00010228228777971345, "loss": 0.0036, "step": 18290 }, { "epoch": 139.6946564885496, "grad_norm": 0.08848349750041962, "learning_rate": 0.0001021308712889171, "loss": 0.0038, "step": 18300 }, { "epoch": 139.7709923664122, "grad_norm": 0.07589084655046463, "learning_rate": 0.0001019795090870399, "loss": 0.0037, "step": 18310 }, { "epoch": 139.84732824427482, "grad_norm": 0.07513239979743958, "learning_rate": 0.0001018282013457434, "loss": 0.0032, "step": 18320 }, { "epoch": 139.9236641221374, "grad_norm": 0.09330571442842484, "learning_rate": 0.00010167694823662744, "loss": 0.0034, "step": 18330 }, { "epoch": 140.0, "grad_norm": 0.08535508066415787, "learning_rate": 0.00010152574993122988, "loss": 0.0034, "step": 18340 }, { "epoch": 140.0763358778626, "grad_norm": 0.1113654300570488, "learning_rate": 0.00010137460660102639, "loss": 0.0037, "step": 18350 }, { "epoch": 140.15267175572518, "grad_norm": 0.0736532136797905, "learning_rate": 0.00010122351841743035, "loss": 0.0035, "step": 18360 }, { "epoch": 140.2290076335878, "grad_norm": 0.08860868960618973, "learning_rate": 0.0001010724855517926, "loss": 0.0031, "step": 18370 }, { "epoch": 140.3053435114504, "grad_norm": 0.06160381808876991, "learning_rate": 0.00010092150817540118, "loss": 0.003, "step": 18380 }, { "epoch": 140.38167938931298, "grad_norm": 0.07011482119560242, "learning_rate": 0.00010077058645948125, "loss": 0.0031, "step": 18390 }, { "epoch": 140.45801526717557, "grad_norm": 0.06074855104088783, "learning_rate": 0.00010061972057519484, "loss": 0.0032, "step": 18400 }, { "epoch": 140.53435114503816, "grad_norm": 0.08197187632322311, "learning_rate": 0.00010046891069364062, "loss": 0.0035, "step": 18410 }, { "epoch": 140.61068702290078, "grad_norm": 0.06935855001211166, "learning_rate": 0.00010031815698585384, "loss": 0.003, "step": 18420 }, { "epoch": 140.68702290076337, "grad_norm": 0.05934326350688934, "learning_rate": 0.0001001674596228059, "loss": 0.0031, "step": 18430 }, { "epoch": 140.76335877862596, "grad_norm": 0.08809772878885269, "learning_rate": 0.00010001681877540442, "loss": 0.0033, "step": 18440 }, { "epoch": 140.83969465648855, "grad_norm": 0.09381341934204102, "learning_rate": 9.986623461449292e-05, "loss": 0.0035, "step": 18450 }, { "epoch": 140.91603053435114, "grad_norm": 0.09422459453344345, "learning_rate": 9.971570731085056e-05, "loss": 0.0033, "step": 18460 }, { "epoch": 140.99236641221373, "grad_norm": 0.08201088756322861, "learning_rate": 9.956523703519202e-05, "loss": 0.0031, "step": 18470 }, { "epoch": 141.06870229007635, "grad_norm": 0.06516539305448532, "learning_rate": 9.941482395816737e-05, "loss": 0.0031, "step": 18480 }, { "epoch": 141.14503816793894, "grad_norm": 0.06602111458778381, "learning_rate": 9.926446825036181e-05, "loss": 0.003, "step": 18490 }, { "epoch": 141.22137404580153, "grad_norm": 0.07777220755815506, "learning_rate": 9.911417008229544e-05, "loss": 0.003, "step": 18500 }, { "epoch": 141.29770992366412, "grad_norm": 0.08189195394515991, "learning_rate": 9.896392962442308e-05, "loss": 0.0029, "step": 18510 }, { "epoch": 141.3740458015267, "grad_norm": 0.08109882473945618, "learning_rate": 9.88137470471341e-05, "loss": 0.0029, "step": 18520 }, { "epoch": 141.45038167938932, "grad_norm": 0.07632119208574295, "learning_rate": 9.866362252075237e-05, "loss": 0.0028, "step": 18530 }, { "epoch": 141.5267175572519, "grad_norm": 0.06572234630584717, "learning_rate": 9.851355621553568e-05, "loss": 0.0029, "step": 18540 }, { "epoch": 141.6030534351145, "grad_norm": 0.08660091459751129, "learning_rate": 9.8363548301676e-05, "loss": 0.0031, "step": 18550 }, { "epoch": 141.6793893129771, "grad_norm": 0.0523909367620945, "learning_rate": 9.821359894929895e-05, "loss": 0.0031, "step": 18560 }, { "epoch": 141.75572519083968, "grad_norm": 0.07985446602106094, "learning_rate": 9.806370832846378e-05, "loss": 0.0032, "step": 18570 }, { "epoch": 141.8320610687023, "grad_norm": 0.07571123540401459, "learning_rate": 9.791387660916321e-05, "loss": 0.0029, "step": 18580 }, { "epoch": 141.9083969465649, "grad_norm": 0.09397567808628082, "learning_rate": 9.776410396132301e-05, "loss": 0.0035, "step": 18590 }, { "epoch": 141.98473282442748, "grad_norm": 0.056722674518823624, "learning_rate": 9.761439055480204e-05, "loss": 0.003, "step": 18600 }, { "epoch": 142.06106870229007, "grad_norm": 0.07240104675292969, "learning_rate": 9.746473655939202e-05, "loss": 0.0033, "step": 18610 }, { "epoch": 142.13740458015266, "grad_norm": 0.1086646243929863, "learning_rate": 9.731514214481714e-05, "loss": 0.0037, "step": 18620 }, { "epoch": 142.21374045801528, "grad_norm": 0.0638151690363884, "learning_rate": 9.716560748073418e-05, "loss": 0.0032, "step": 18630 }, { "epoch": 142.29007633587787, "grad_norm": 0.06882358342409134, "learning_rate": 9.701613273673211e-05, "loss": 0.0035, "step": 18640 }, { "epoch": 142.36641221374046, "grad_norm": 0.06252092123031616, "learning_rate": 9.686671808233185e-05, "loss": 0.0028, "step": 18650 }, { "epoch": 142.44274809160305, "grad_norm": 0.08045171946287155, "learning_rate": 9.67173636869863e-05, "loss": 0.0031, "step": 18660 }, { "epoch": 142.51908396946564, "grad_norm": 0.07197963446378708, "learning_rate": 9.656806972007997e-05, "loss": 0.0032, "step": 18670 }, { "epoch": 142.59541984732823, "grad_norm": 0.06809152662754059, "learning_rate": 9.641883635092876e-05, "loss": 0.0031, "step": 18680 }, { "epoch": 142.67175572519085, "grad_norm": 0.06739526242017746, "learning_rate": 9.626966374878002e-05, "loss": 0.0032, "step": 18690 }, { "epoch": 142.74809160305344, "grad_norm": 0.07248133420944214, "learning_rate": 9.6120552082812e-05, "loss": 0.0032, "step": 18700 }, { "epoch": 142.82442748091603, "grad_norm": 0.0871865451335907, "learning_rate": 9.597150152213401e-05, "loss": 0.0029, "step": 18710 }, { "epoch": 142.90076335877862, "grad_norm": 0.08785657584667206, "learning_rate": 9.58225122357859e-05, "loss": 0.0033, "step": 18720 }, { "epoch": 142.9770992366412, "grad_norm": 0.0891893059015274, "learning_rate": 9.56735843927381e-05, "loss": 0.0029, "step": 18730 }, { "epoch": 143.05343511450383, "grad_norm": 0.06258516013622284, "learning_rate": 9.552471816189141e-05, "loss": 0.003, "step": 18740 }, { "epoch": 143.12977099236642, "grad_norm": 0.073508620262146, "learning_rate": 9.537591371207666e-05, "loss": 0.003, "step": 18750 }, { "epoch": 143.206106870229, "grad_norm": 0.0547512024641037, "learning_rate": 9.522717121205468e-05, "loss": 0.0029, "step": 18760 }, { "epoch": 143.2824427480916, "grad_norm": 0.06893076002597809, "learning_rate": 9.5078490830516e-05, "loss": 0.0028, "step": 18770 }, { "epoch": 143.3587786259542, "grad_norm": 0.07158563286066055, "learning_rate": 9.492987273608072e-05, "loss": 0.0027, "step": 18780 }, { "epoch": 143.4351145038168, "grad_norm": 0.06421563774347305, "learning_rate": 9.47813170972983e-05, "loss": 0.0032, "step": 18790 }, { "epoch": 143.5114503816794, "grad_norm": 0.09061388671398163, "learning_rate": 9.46328240826474e-05, "loss": 0.0029, "step": 18800 }, { "epoch": 143.58778625954199, "grad_norm": 0.08171732723712921, "learning_rate": 9.448439386053556e-05, "loss": 0.003, "step": 18810 }, { "epoch": 143.66412213740458, "grad_norm": 0.11272929608821869, "learning_rate": 9.433602659929923e-05, "loss": 0.0032, "step": 18820 }, { "epoch": 143.74045801526717, "grad_norm": 0.07037962228059769, "learning_rate": 9.418772246720336e-05, "loss": 0.0038, "step": 18830 }, { "epoch": 143.81679389312978, "grad_norm": 0.07233633100986481, "learning_rate": 9.403948163244131e-05, "loss": 0.0031, "step": 18840 }, { "epoch": 143.89312977099237, "grad_norm": 0.07079844176769257, "learning_rate": 9.389130426313479e-05, "loss": 0.003, "step": 18850 }, { "epoch": 143.96946564885496, "grad_norm": 0.08107413351535797, "learning_rate": 9.374319052733332e-05, "loss": 0.0032, "step": 18860 }, { "epoch": 144.04580152671755, "grad_norm": 0.07768434286117554, "learning_rate": 9.359514059301439e-05, "loss": 0.0031, "step": 18870 }, { "epoch": 144.12213740458014, "grad_norm": 0.05583273619413376, "learning_rate": 9.344715462808314e-05, "loss": 0.0026, "step": 18880 }, { "epoch": 144.19847328244273, "grad_norm": 0.05909670889377594, "learning_rate": 9.329923280037201e-05, "loss": 0.0032, "step": 18890 }, { "epoch": 144.27480916030535, "grad_norm": 0.07353813201189041, "learning_rate": 9.3151375277641e-05, "loss": 0.003, "step": 18900 }, { "epoch": 144.35114503816794, "grad_norm": 0.0820692628622055, "learning_rate": 9.300358222757685e-05, "loss": 0.0026, "step": 18910 }, { "epoch": 144.42748091603053, "grad_norm": 0.07494720071554184, "learning_rate": 9.285585381779337e-05, "loss": 0.003, "step": 18920 }, { "epoch": 144.50381679389312, "grad_norm": 0.08090074360370636, "learning_rate": 9.270819021583106e-05, "loss": 0.0029, "step": 18930 }, { "epoch": 144.5801526717557, "grad_norm": 0.12172301858663559, "learning_rate": 9.256059158915678e-05, "loss": 0.0031, "step": 18940 }, { "epoch": 144.65648854961833, "grad_norm": 0.10968704521656036, "learning_rate": 9.241305810516391e-05, "loss": 0.0029, "step": 18950 }, { "epoch": 144.73282442748092, "grad_norm": 0.08783160150051117, "learning_rate": 9.226558993117174e-05, "loss": 0.0031, "step": 18960 }, { "epoch": 144.8091603053435, "grad_norm": 0.06607697904109955, "learning_rate": 9.211818723442562e-05, "loss": 0.0031, "step": 18970 }, { "epoch": 144.8854961832061, "grad_norm": 0.05663434416055679, "learning_rate": 9.197085018209668e-05, "loss": 0.0031, "step": 18980 }, { "epoch": 144.9618320610687, "grad_norm": 0.06583233922719955, "learning_rate": 9.182357894128144e-05, "loss": 0.0032, "step": 18990 }, { "epoch": 145.0381679389313, "grad_norm": 0.07183825224637985, "learning_rate": 9.167637367900191e-05, "loss": 0.0029, "step": 19000 }, { "epoch": 145.1145038167939, "grad_norm": 0.07823116332292557, "learning_rate": 9.152923456220525e-05, "loss": 0.0031, "step": 19010 }, { "epoch": 145.1908396946565, "grad_norm": 0.07032425701618195, "learning_rate": 9.138216175776358e-05, "loss": 0.0029, "step": 19020 }, { "epoch": 145.26717557251908, "grad_norm": 0.07179781049489975, "learning_rate": 9.123515543247385e-05, "loss": 0.0032, "step": 19030 }, { "epoch": 145.34351145038167, "grad_norm": 0.06336041539907455, "learning_rate": 9.10882157530576e-05, "loss": 0.003, "step": 19040 }, { "epoch": 145.4198473282443, "grad_norm": 0.06709475070238113, "learning_rate": 9.09413428861607e-05, "loss": 0.0028, "step": 19050 }, { "epoch": 145.49618320610688, "grad_norm": 0.06345654278993607, "learning_rate": 9.079453699835339e-05, "loss": 0.0028, "step": 19060 }, { "epoch": 145.57251908396947, "grad_norm": 0.05199535936117172, "learning_rate": 9.064779825612991e-05, "loss": 0.0031, "step": 19070 }, { "epoch": 145.64885496183206, "grad_norm": 0.056889478117227554, "learning_rate": 9.050112682590826e-05, "loss": 0.0029, "step": 19080 }, { "epoch": 145.72519083969465, "grad_norm": 0.07553195208311081, "learning_rate": 9.03545228740302e-05, "loss": 0.0035, "step": 19090 }, { "epoch": 145.80152671755727, "grad_norm": 0.06325870007276535, "learning_rate": 9.02079865667609e-05, "loss": 0.0026, "step": 19100 }, { "epoch": 145.87786259541986, "grad_norm": 0.06529030948877335, "learning_rate": 9.00615180702889e-05, "loss": 0.003, "step": 19110 }, { "epoch": 145.95419847328245, "grad_norm": 0.0553898848593235, "learning_rate": 8.991511755072571e-05, "loss": 0.0031, "step": 19120 }, { "epoch": 146.03053435114504, "grad_norm": 0.06495235860347748, "learning_rate": 8.976878517410585e-05, "loss": 0.0031, "step": 19130 }, { "epoch": 146.10687022900763, "grad_norm": 0.08064547926187515, "learning_rate": 8.962252110638653e-05, "loss": 0.0029, "step": 19140 }, { "epoch": 146.18320610687022, "grad_norm": 0.05380216985940933, "learning_rate": 8.947632551344742e-05, "loss": 0.0029, "step": 19150 }, { "epoch": 146.25954198473283, "grad_norm": 0.08094267547130585, "learning_rate": 8.933019856109064e-05, "loss": 0.0031, "step": 19160 }, { "epoch": 146.33587786259542, "grad_norm": 0.081087626516819, "learning_rate": 8.918414041504041e-05, "loss": 0.0032, "step": 19170 }, { "epoch": 146.41221374045801, "grad_norm": 0.08447154611349106, "learning_rate": 8.90381512409429e-05, "loss": 0.0028, "step": 19180 }, { "epoch": 146.4885496183206, "grad_norm": 0.07587023079395294, "learning_rate": 8.88922312043661e-05, "loss": 0.0027, "step": 19190 }, { "epoch": 146.5648854961832, "grad_norm": 0.07812447845935822, "learning_rate": 8.874638047079962e-05, "loss": 0.0031, "step": 19200 }, { "epoch": 146.6412213740458, "grad_norm": 0.06533341854810715, "learning_rate": 8.86005992056543e-05, "loss": 0.0027, "step": 19210 }, { "epoch": 146.7175572519084, "grad_norm": 0.06456413865089417, "learning_rate": 8.845488757426244e-05, "loss": 0.0025, "step": 19220 }, { "epoch": 146.793893129771, "grad_norm": 0.09169802814722061, "learning_rate": 8.830924574187716e-05, "loss": 0.0031, "step": 19230 }, { "epoch": 146.87022900763358, "grad_norm": 0.06284616887569427, "learning_rate": 8.816367387367252e-05, "loss": 0.003, "step": 19240 }, { "epoch": 146.94656488549617, "grad_norm": 0.09204675257205963, "learning_rate": 8.80181721347433e-05, "loss": 0.0028, "step": 19250 }, { "epoch": 147.0229007633588, "grad_norm": 0.07699024677276611, "learning_rate": 8.787274069010458e-05, "loss": 0.0025, "step": 19260 }, { "epoch": 147.09923664122138, "grad_norm": 0.09933602809906006, "learning_rate": 8.77273797046918e-05, "loss": 0.0029, "step": 19270 }, { "epoch": 147.17557251908397, "grad_norm": 0.08356738835573196, "learning_rate": 8.758208934336053e-05, "loss": 0.0029, "step": 19280 }, { "epoch": 147.25190839694656, "grad_norm": 0.09409868717193604, "learning_rate": 8.743686977088622e-05, "loss": 0.0026, "step": 19290 }, { "epoch": 147.32824427480915, "grad_norm": 0.05249028280377388, "learning_rate": 8.729172115196399e-05, "loss": 0.0027, "step": 19300 }, { "epoch": 147.40458015267177, "grad_norm": 0.090017169713974, "learning_rate": 8.714664365120846e-05, "loss": 0.0028, "step": 19310 }, { "epoch": 147.48091603053436, "grad_norm": 0.06579986214637756, "learning_rate": 8.70016374331538e-05, "loss": 0.0028, "step": 19320 }, { "epoch": 147.55725190839695, "grad_norm": 0.06821981817483902, "learning_rate": 8.685670266225314e-05, "loss": 0.0031, "step": 19330 }, { "epoch": 147.63358778625954, "grad_norm": 0.08105731010437012, "learning_rate": 8.671183950287872e-05, "loss": 0.0028, "step": 19340 }, { "epoch": 147.70992366412213, "grad_norm": 0.07285632193088531, "learning_rate": 8.656704811932133e-05, "loss": 0.003, "step": 19350 }, { "epoch": 147.78625954198472, "grad_norm": 0.08571650832891464, "learning_rate": 8.64223286757906e-05, "loss": 0.003, "step": 19360 }, { "epoch": 147.86259541984734, "grad_norm": 0.08256779611110687, "learning_rate": 8.627768133641445e-05, "loss": 0.0031, "step": 19370 }, { "epoch": 147.93893129770993, "grad_norm": 0.05241544917225838, "learning_rate": 8.613310626523909e-05, "loss": 0.0033, "step": 19380 }, { "epoch": 148.01526717557252, "grad_norm": 0.07321493327617645, "learning_rate": 8.598860362622872e-05, "loss": 0.003, "step": 19390 }, { "epoch": 148.0916030534351, "grad_norm": 0.07568832486867905, "learning_rate": 8.584417358326549e-05, "loss": 0.0031, "step": 19400 }, { "epoch": 148.1679389312977, "grad_norm": 0.06682102382183075, "learning_rate": 8.569981630014901e-05, "loss": 0.003, "step": 19410 }, { "epoch": 148.24427480916032, "grad_norm": 0.09373755753040314, "learning_rate": 8.55555319405966e-05, "loss": 0.0031, "step": 19420 }, { "epoch": 148.3206106870229, "grad_norm": 0.080172099173069, "learning_rate": 8.54113206682427e-05, "loss": 0.003, "step": 19430 }, { "epoch": 148.3969465648855, "grad_norm": 0.07044123858213425, "learning_rate": 8.526718264663903e-05, "loss": 0.0027, "step": 19440 }, { "epoch": 148.4732824427481, "grad_norm": 0.07569611072540283, "learning_rate": 8.512311803925417e-05, "loss": 0.0027, "step": 19450 }, { "epoch": 148.54961832061068, "grad_norm": 0.06499068439006805, "learning_rate": 8.497912700947329e-05, "loss": 0.0031, "step": 19460 }, { "epoch": 148.6259541984733, "grad_norm": 0.07677905261516571, "learning_rate": 8.483520972059829e-05, "loss": 0.0031, "step": 19470 }, { "epoch": 148.70229007633588, "grad_norm": 0.09200398623943329, "learning_rate": 8.469136633584743e-05, "loss": 0.003, "step": 19480 }, { "epoch": 148.77862595419847, "grad_norm": 0.08276019245386124, "learning_rate": 8.454759701835511e-05, "loss": 0.0029, "step": 19490 }, { "epoch": 148.85496183206106, "grad_norm": 0.07673590630292892, "learning_rate": 8.440390193117169e-05, "loss": 0.0025, "step": 19500 }, { "epoch": 148.93129770992365, "grad_norm": 0.08123070001602173, "learning_rate": 8.426028123726343e-05, "loss": 0.0026, "step": 19510 }, { "epoch": 149.00763358778627, "grad_norm": 0.09721755981445312, "learning_rate": 8.411673509951219e-05, "loss": 0.0028, "step": 19520 }, { "epoch": 149.08396946564886, "grad_norm": 0.06246260181069374, "learning_rate": 8.397326368071522e-05, "loss": 0.0026, "step": 19530 }, { "epoch": 149.16030534351145, "grad_norm": 0.08201941847801208, "learning_rate": 8.382986714358511e-05, "loss": 0.0028, "step": 19540 }, { "epoch": 149.23664122137404, "grad_norm": 0.05413436517119408, "learning_rate": 8.368654565074956e-05, "loss": 0.0024, "step": 19550 }, { "epoch": 149.31297709923663, "grad_norm": 0.06429063528776169, "learning_rate": 8.354329936475095e-05, "loss": 0.0026, "step": 19560 }, { "epoch": 149.38931297709922, "grad_norm": 0.07416065782308578, "learning_rate": 8.340012844804656e-05, "loss": 0.0025, "step": 19570 }, { "epoch": 149.46564885496184, "grad_norm": 0.11003588140010834, "learning_rate": 8.325703306300816e-05, "loss": 0.003, "step": 19580 }, { "epoch": 149.54198473282443, "grad_norm": 0.08371774852275848, "learning_rate": 8.311401337192186e-05, "loss": 0.0027, "step": 19590 }, { "epoch": 149.61832061068702, "grad_norm": 0.08815993368625641, "learning_rate": 8.297106953698785e-05, "loss": 0.0033, "step": 19600 }, { "epoch": 149.6946564885496, "grad_norm": 0.08714954555034637, "learning_rate": 8.282820172032046e-05, "loss": 0.0028, "step": 19610 }, { "epoch": 149.7709923664122, "grad_norm": 0.0706057995557785, "learning_rate": 8.268541008394758e-05, "loss": 0.0027, "step": 19620 }, { "epoch": 149.84732824427482, "grad_norm": 0.07057519257068634, "learning_rate": 8.254269478981083e-05, "loss": 0.0027, "step": 19630 }, { "epoch": 149.9236641221374, "grad_norm": 0.0842265859246254, "learning_rate": 8.240005599976523e-05, "loss": 0.0028, "step": 19640 }, { "epoch": 150.0, "grad_norm": 0.06233743205666542, "learning_rate": 8.22574938755791e-05, "loss": 0.0028, "step": 19650 }, { "epoch": 150.0763358778626, "grad_norm": 0.07128167152404785, "learning_rate": 8.21150085789337e-05, "loss": 0.0027, "step": 19660 }, { "epoch": 150.15267175572518, "grad_norm": 0.06901545077562332, "learning_rate": 8.197260027142333e-05, "loss": 0.0028, "step": 19670 }, { "epoch": 150.2290076335878, "grad_norm": 0.051990777254104614, "learning_rate": 8.183026911455468e-05, "loss": 0.003, "step": 19680 }, { "epoch": 150.3053435114504, "grad_norm": 0.07578859478235245, "learning_rate": 8.168801526974721e-05, "loss": 0.0027, "step": 19690 }, { "epoch": 150.38167938931298, "grad_norm": 0.09771639108657837, "learning_rate": 8.15458388983326e-05, "loss": 0.0029, "step": 19700 }, { "epoch": 150.45801526717557, "grad_norm": 0.07489226013422012, "learning_rate": 8.140374016155469e-05, "loss": 0.0029, "step": 19710 }, { "epoch": 150.53435114503816, "grad_norm": 0.07546740770339966, "learning_rate": 8.126171922056927e-05, "loss": 0.0031, "step": 19720 }, { "epoch": 150.61068702290078, "grad_norm": 0.08109984546899796, "learning_rate": 8.111977623644386e-05, "loss": 0.0027, "step": 19730 }, { "epoch": 150.68702290076337, "grad_norm": 0.09849539399147034, "learning_rate": 8.097791137015763e-05, "loss": 0.0028, "step": 19740 }, { "epoch": 150.76335877862596, "grad_norm": 0.06832431256771088, "learning_rate": 8.083612478260109e-05, "loss": 0.0029, "step": 19750 }, { "epoch": 150.83969465648855, "grad_norm": 0.06837894022464752, "learning_rate": 8.069441663457605e-05, "loss": 0.0026, "step": 19760 }, { "epoch": 150.91603053435114, "grad_norm": 0.10627114027738571, "learning_rate": 8.055278708679537e-05, "loss": 0.0024, "step": 19770 }, { "epoch": 150.99236641221373, "grad_norm": 0.07059499621391296, "learning_rate": 8.041123629988256e-05, "loss": 0.0026, "step": 19780 }, { "epoch": 151.06870229007635, "grad_norm": 0.06606269627809525, "learning_rate": 8.026976443437209e-05, "loss": 0.0028, "step": 19790 }, { "epoch": 151.14503816793894, "grad_norm": 0.07408957183361053, "learning_rate": 8.012837165070876e-05, "loss": 0.0026, "step": 19800 }, { "epoch": 151.22137404580153, "grad_norm": 0.08273787796497345, "learning_rate": 7.99870581092477e-05, "loss": 0.0029, "step": 19810 }, { "epoch": 151.29770992366412, "grad_norm": 0.08021242171525955, "learning_rate": 7.984582397025432e-05, "loss": 0.0028, "step": 19820 }, { "epoch": 151.3740458015267, "grad_norm": 0.10280722379684448, "learning_rate": 7.970466939390368e-05, "loss": 0.0029, "step": 19830 }, { "epoch": 151.45038167938932, "grad_norm": 0.09210003167390823, "learning_rate": 7.956359454028081e-05, "loss": 0.003, "step": 19840 }, { "epoch": 151.5267175572519, "grad_norm": 0.08739759027957916, "learning_rate": 7.942259956938038e-05, "loss": 0.0027, "step": 19850 }, { "epoch": 151.6030534351145, "grad_norm": 0.0719342976808548, "learning_rate": 7.92816846411063e-05, "loss": 0.0027, "step": 19860 }, { "epoch": 151.6793893129771, "grad_norm": 0.07027608901262283, "learning_rate": 7.914084991527183e-05, "loss": 0.0026, "step": 19870 }, { "epoch": 151.75572519083968, "grad_norm": 0.06904811412096024, "learning_rate": 7.900009555159925e-05, "loss": 0.0026, "step": 19880 }, { "epoch": 151.8320610687023, "grad_norm": 0.08667945861816406, "learning_rate": 7.885942170971958e-05, "loss": 0.0028, "step": 19890 }, { "epoch": 151.9083969465649, "grad_norm": 0.08470647782087326, "learning_rate": 7.871882854917264e-05, "loss": 0.0032, "step": 19900 }, { "epoch": 151.98473282442748, "grad_norm": 0.07111072540283203, "learning_rate": 7.857831622940675e-05, "loss": 0.0029, "step": 19910 }, { "epoch": 152.06106870229007, "grad_norm": 0.08362358063459396, "learning_rate": 7.843788490977852e-05, "loss": 0.0028, "step": 19920 }, { "epoch": 152.13740458015266, "grad_norm": 0.07762838155031204, "learning_rate": 7.829753474955265e-05, "loss": 0.0026, "step": 19930 }, { "epoch": 152.21374045801528, "grad_norm": 0.05775099992752075, "learning_rate": 7.815726590790189e-05, "loss": 0.0026, "step": 19940 }, { "epoch": 152.29007633587787, "grad_norm": 0.06592975556850433, "learning_rate": 7.801707854390668e-05, "loss": 0.0031, "step": 19950 }, { "epoch": 152.36641221374046, "grad_norm": 0.07081467658281326, "learning_rate": 7.787697281655513e-05, "loss": 0.0027, "step": 19960 }, { "epoch": 152.44274809160305, "grad_norm": 0.04643511772155762, "learning_rate": 7.773694888474267e-05, "loss": 0.0027, "step": 19970 }, { "epoch": 152.51908396946564, "grad_norm": 0.07057954370975494, "learning_rate": 7.759700690727214e-05, "loss": 0.0024, "step": 19980 }, { "epoch": 152.59541984732823, "grad_norm": 0.05838805437088013, "learning_rate": 7.745714704285317e-05, "loss": 0.0022, "step": 19990 }, { "epoch": 152.67175572519085, "grad_norm": 0.05929241701960564, "learning_rate": 7.731736945010247e-05, "loss": 0.003, "step": 20000 }, { "epoch": 152.74809160305344, "grad_norm": 0.08759178221225739, "learning_rate": 7.71776742875434e-05, "loss": 0.0024, "step": 20010 }, { "epoch": 152.82442748091603, "grad_norm": 0.09800530970096588, "learning_rate": 7.703806171360581e-05, "loss": 0.0028, "step": 20020 }, { "epoch": 152.90076335877862, "grad_norm": 0.05581768974661827, "learning_rate": 7.689853188662588e-05, "loss": 0.0026, "step": 20030 }, { "epoch": 152.9770992366412, "grad_norm": 0.07298070192337036, "learning_rate": 7.675908496484605e-05, "loss": 0.0026, "step": 20040 }, { "epoch": 153.05343511450383, "grad_norm": 0.07643323391675949, "learning_rate": 7.661972110641448e-05, "loss": 0.0024, "step": 20050 }, { "epoch": 153.12977099236642, "grad_norm": 0.102878637611866, "learning_rate": 7.648044046938543e-05, "loss": 0.0026, "step": 20060 }, { "epoch": 153.206106870229, "grad_norm": 0.05298960953950882, "learning_rate": 7.634124321171858e-05, "loss": 0.0028, "step": 20070 }, { "epoch": 153.2824427480916, "grad_norm": 0.08185269683599472, "learning_rate": 7.620212949127916e-05, "loss": 0.0027, "step": 20080 }, { "epoch": 153.3587786259542, "grad_norm": 0.1582772582769394, "learning_rate": 7.606309946583762e-05, "loss": 0.0029, "step": 20090 }, { "epoch": 153.4351145038168, "grad_norm": 0.06945278495550156, "learning_rate": 7.592415329306939e-05, "loss": 0.0026, "step": 20100 }, { "epoch": 153.5114503816794, "grad_norm": 0.08417531102895737, "learning_rate": 7.578529113055493e-05, "loss": 0.0032, "step": 20110 }, { "epoch": 153.58778625954199, "grad_norm": 0.08206114172935486, "learning_rate": 7.564651313577943e-05, "loss": 0.0028, "step": 20120 }, { "epoch": 153.66412213740458, "grad_norm": 0.06993961334228516, "learning_rate": 7.550781946613253e-05, "loss": 0.0027, "step": 20130 }, { "epoch": 153.74045801526717, "grad_norm": 0.0562141127884388, "learning_rate": 7.536921027890832e-05, "loss": 0.0026, "step": 20140 }, { "epoch": 153.81679389312978, "grad_norm": 0.06560752540826797, "learning_rate": 7.523068573130503e-05, "loss": 0.0026, "step": 20150 }, { "epoch": 153.89312977099237, "grad_norm": 0.06448066979646683, "learning_rate": 7.50922459804249e-05, "loss": 0.0024, "step": 20160 }, { "epoch": 153.96946564885496, "grad_norm": 0.0739925354719162, "learning_rate": 7.495389118327407e-05, "loss": 0.0029, "step": 20170 }, { "epoch": 154.04580152671755, "grad_norm": 0.07964913547039032, "learning_rate": 7.48156214967622e-05, "loss": 0.0027, "step": 20180 }, { "epoch": 154.12213740458014, "grad_norm": 0.06940018385648727, "learning_rate": 7.467743707770254e-05, "loss": 0.0023, "step": 20190 }, { "epoch": 154.19847328244273, "grad_norm": 0.0707584097981453, "learning_rate": 7.453933808281166e-05, "loss": 0.0026, "step": 20200 }, { "epoch": 154.27480916030535, "grad_norm": 0.07698853313922882, "learning_rate": 7.440132466870907e-05, "loss": 0.0031, "step": 20210 }, { "epoch": 154.35114503816794, "grad_norm": 0.06870463490486145, "learning_rate": 7.42633969919174e-05, "loss": 0.0026, "step": 20220 }, { "epoch": 154.42748091603053, "grad_norm": 0.08482933044433594, "learning_rate": 7.412555520886197e-05, "loss": 0.0025, "step": 20230 }, { "epoch": 154.50381679389312, "grad_norm": 0.09738239645957947, "learning_rate": 7.398779947587073e-05, "loss": 0.0026, "step": 20240 }, { "epoch": 154.5801526717557, "grad_norm": 0.08029542863368988, "learning_rate": 7.385012994917404e-05, "loss": 0.0023, "step": 20250 }, { "epoch": 154.65648854961833, "grad_norm": 0.06089054048061371, "learning_rate": 7.371254678490437e-05, "loss": 0.0027, "step": 20260 }, { "epoch": 154.73282442748092, "grad_norm": 0.09448499977588654, "learning_rate": 7.357505013909638e-05, "loss": 0.0028, "step": 20270 }, { "epoch": 154.8091603053435, "grad_norm": 0.0953303873538971, "learning_rate": 7.343764016768657e-05, "loss": 0.0026, "step": 20280 }, { "epoch": 154.8854961832061, "grad_norm": 0.10695008188486099, "learning_rate": 7.330031702651316e-05, "loss": 0.0026, "step": 20290 }, { "epoch": 154.9618320610687, "grad_norm": 0.09159107506275177, "learning_rate": 7.316308087131586e-05, "loss": 0.0029, "step": 20300 }, { "epoch": 155.0381679389313, "grad_norm": 0.0897974967956543, "learning_rate": 7.302593185773579e-05, "loss": 0.0025, "step": 20310 }, { "epoch": 155.1145038167939, "grad_norm": 0.07279577851295471, "learning_rate": 7.288887014131508e-05, "loss": 0.0026, "step": 20320 }, { "epoch": 155.1908396946565, "grad_norm": 0.0747234970331192, "learning_rate": 7.275189587749705e-05, "loss": 0.0027, "step": 20330 }, { "epoch": 155.26717557251908, "grad_norm": 0.06692302972078323, "learning_rate": 7.261500922162568e-05, "loss": 0.0025, "step": 20340 }, { "epoch": 155.34351145038167, "grad_norm": 0.05474691838026047, "learning_rate": 7.247821032894586e-05, "loss": 0.0024, "step": 20350 }, { "epoch": 155.4198473282443, "grad_norm": 0.06611966341733932, "learning_rate": 7.234149935460258e-05, "loss": 0.0026, "step": 20360 }, { "epoch": 155.49618320610688, "grad_norm": 0.057101741433143616, "learning_rate": 7.220487645364134e-05, "loss": 0.0027, "step": 20370 }, { "epoch": 155.57251908396947, "grad_norm": 0.07845055311918259, "learning_rate": 7.206834178100773e-05, "loss": 0.0026, "step": 20380 }, { "epoch": 155.64885496183206, "grad_norm": 0.07105549424886703, "learning_rate": 7.193189549154727e-05, "loss": 0.0024, "step": 20390 }, { "epoch": 155.72519083969465, "grad_norm": 0.07200198620557785, "learning_rate": 7.17955377400052e-05, "loss": 0.0024, "step": 20400 }, { "epoch": 155.80152671755727, "grad_norm": 0.13774210214614868, "learning_rate": 7.16592686810265e-05, "loss": 0.0027, "step": 20410 }, { "epoch": 155.87786259541986, "grad_norm": 0.0656709149479866, "learning_rate": 7.152308846915529e-05, "loss": 0.0028, "step": 20420 }, { "epoch": 155.95419847328245, "grad_norm": 0.10810025781393051, "learning_rate": 7.138699725883515e-05, "loss": 0.0028, "step": 20430 }, { "epoch": 156.03053435114504, "grad_norm": 0.0687631368637085, "learning_rate": 7.125099520440867e-05, "loss": 0.0024, "step": 20440 }, { "epoch": 156.10687022900763, "grad_norm": 0.09200803190469742, "learning_rate": 7.11150824601173e-05, "loss": 0.0027, "step": 20450 }, { "epoch": 156.18320610687022, "grad_norm": 0.09717967361211777, "learning_rate": 7.097925918010122e-05, "loss": 0.0027, "step": 20460 }, { "epoch": 156.25954198473283, "grad_norm": 0.06093241274356842, "learning_rate": 7.084352551839922e-05, "loss": 0.003, "step": 20470 }, { "epoch": 156.33587786259542, "grad_norm": 0.07521457225084305, "learning_rate": 7.070788162894824e-05, "loss": 0.0026, "step": 20480 }, { "epoch": 156.41221374045801, "grad_norm": 0.06158873066306114, "learning_rate": 7.057232766558363e-05, "loss": 0.0029, "step": 20490 }, { "epoch": 156.4885496183206, "grad_norm": 0.07519913464784622, "learning_rate": 7.043686378203863e-05, "loss": 0.0023, "step": 20500 }, { "epoch": 156.5648854961832, "grad_norm": 0.06330856680870056, "learning_rate": 7.030149013194439e-05, "loss": 0.0027, "step": 20510 }, { "epoch": 156.6412213740458, "grad_norm": 0.05133967474102974, "learning_rate": 7.016620686882974e-05, "loss": 0.0024, "step": 20520 }, { "epoch": 156.7175572519084, "grad_norm": 0.09252005070447922, "learning_rate": 7.003101414612086e-05, "loss": 0.0024, "step": 20530 }, { "epoch": 156.793893129771, "grad_norm": 0.07333563268184662, "learning_rate": 6.989591211714137e-05, "loss": 0.0025, "step": 20540 }, { "epoch": 156.87022900763358, "grad_norm": 0.09405118227005005, "learning_rate": 6.976090093511196e-05, "loss": 0.0029, "step": 20550 }, { "epoch": 156.94656488549617, "grad_norm": 0.07460989058017731, "learning_rate": 6.962598075315046e-05, "loss": 0.0027, "step": 20560 }, { "epoch": 157.0229007633588, "grad_norm": 0.07728581130504608, "learning_rate": 6.949115172427137e-05, "loss": 0.0029, "step": 20570 }, { "epoch": 157.09923664122138, "grad_norm": 0.07974912971258163, "learning_rate": 6.935641400138568e-05, "loss": 0.0027, "step": 20580 }, { "epoch": 157.17557251908397, "grad_norm": 0.06969688832759857, "learning_rate": 6.922176773730104e-05, "loss": 0.0026, "step": 20590 }, { "epoch": 157.25190839694656, "grad_norm": 0.19186438620090485, "learning_rate": 6.908721308472129e-05, "loss": 0.0029, "step": 20600 }, { "epoch": 157.32824427480915, "grad_norm": 0.07386868447065353, "learning_rate": 6.895275019624638e-05, "loss": 0.0027, "step": 20610 }, { "epoch": 157.40458015267177, "grad_norm": 0.07287906110286713, "learning_rate": 6.881837922437225e-05, "loss": 0.0028, "step": 20620 }, { "epoch": 157.48091603053436, "grad_norm": 0.05664809048175812, "learning_rate": 6.86841003214904e-05, "loss": 0.0028, "step": 20630 }, { "epoch": 157.55725190839695, "grad_norm": 0.0952431708574295, "learning_rate": 6.854991363988815e-05, "loss": 0.0028, "step": 20640 }, { "epoch": 157.63358778625954, "grad_norm": 0.0727730542421341, "learning_rate": 6.841581933174808e-05, "loss": 0.0028, "step": 20650 }, { "epoch": 157.70992366412213, "grad_norm": 0.05990252271294594, "learning_rate": 6.828181754914811e-05, "loss": 0.0023, "step": 20660 }, { "epoch": 157.78625954198472, "grad_norm": 0.07227367162704468, "learning_rate": 6.814790844406112e-05, "loss": 0.0029, "step": 20670 }, { "epoch": 157.86259541984734, "grad_norm": 0.051172178238630295, "learning_rate": 6.801409216835502e-05, "loss": 0.0026, "step": 20680 }, { "epoch": 157.93893129770993, "grad_norm": 0.05831562727689743, "learning_rate": 6.788036887379228e-05, "loss": 0.0024, "step": 20690 }, { "epoch": 158.01526717557252, "grad_norm": 0.07706089317798615, "learning_rate": 6.774673871202998e-05, "loss": 0.0025, "step": 20700 }, { "epoch": 158.0916030534351, "grad_norm": 0.07770222425460815, "learning_rate": 6.761320183461966e-05, "loss": 0.0028, "step": 20710 }, { "epoch": 158.1679389312977, "grad_norm": 0.07851127535104752, "learning_rate": 6.747975839300697e-05, "loss": 0.0025, "step": 20720 }, { "epoch": 158.24427480916032, "grad_norm": 0.11021630465984344, "learning_rate": 6.734640853853163e-05, "loss": 0.0026, "step": 20730 }, { "epoch": 158.3206106870229, "grad_norm": 0.08283508569002151, "learning_rate": 6.721315242242727e-05, "loss": 0.0029, "step": 20740 }, { "epoch": 158.3969465648855, "grad_norm": 0.06768762320280075, "learning_rate": 6.707999019582103e-05, "loss": 0.0027, "step": 20750 }, { "epoch": 158.4732824427481, "grad_norm": 0.09211503714323044, "learning_rate": 6.694692200973372e-05, "loss": 0.0026, "step": 20760 }, { "epoch": 158.54961832061068, "grad_norm": 0.0939657986164093, "learning_rate": 6.681394801507955e-05, "loss": 0.0026, "step": 20770 }, { "epoch": 158.6259541984733, "grad_norm": 0.07526792585849762, "learning_rate": 6.668106836266586e-05, "loss": 0.0024, "step": 20780 }, { "epoch": 158.70229007633588, "grad_norm": 0.059897441416978836, "learning_rate": 6.654828320319284e-05, "loss": 0.0024, "step": 20790 }, { "epoch": 158.77862595419847, "grad_norm": 0.058402128517627716, "learning_rate": 6.64155926872537e-05, "loss": 0.0026, "step": 20800 }, { "epoch": 158.85496183206106, "grad_norm": 0.06308060139417648, "learning_rate": 6.628299696533424e-05, "loss": 0.0028, "step": 20810 }, { "epoch": 158.93129770992365, "grad_norm": 0.07480074465274811, "learning_rate": 6.615049618781277e-05, "loss": 0.0022, "step": 20820 }, { "epoch": 159.00763358778627, "grad_norm": 0.0777663141489029, "learning_rate": 6.601809050495993e-05, "loss": 0.0024, "step": 20830 }, { "epoch": 159.08396946564886, "grad_norm": 0.07570364326238632, "learning_rate": 6.588578006693853e-05, "loss": 0.0025, "step": 20840 }, { "epoch": 159.16030534351145, "grad_norm": 0.06377175450325012, "learning_rate": 6.575356502380322e-05, "loss": 0.0026, "step": 20850 }, { "epoch": 159.23664122137404, "grad_norm": 0.07236968725919724, "learning_rate": 6.562144552550066e-05, "loss": 0.0024, "step": 20860 }, { "epoch": 159.31297709923663, "grad_norm": 0.07538765668869019, "learning_rate": 6.548942172186902e-05, "loss": 0.0022, "step": 20870 }, { "epoch": 159.38931297709922, "grad_norm": 0.07262559980154037, "learning_rate": 6.535749376263804e-05, "loss": 0.0026, "step": 20880 }, { "epoch": 159.46564885496184, "grad_norm": 0.08544399589300156, "learning_rate": 6.522566179742871e-05, "loss": 0.0027, "step": 20890 }, { "epoch": 159.54198473282443, "grad_norm": 0.07203151285648346, "learning_rate": 6.509392597575309e-05, "loss": 0.0023, "step": 20900 }, { "epoch": 159.61832061068702, "grad_norm": 0.06400939077138901, "learning_rate": 6.496228644701425e-05, "loss": 0.0023, "step": 20910 }, { "epoch": 159.6946564885496, "grad_norm": 0.06140679493546486, "learning_rate": 6.483074336050613e-05, "loss": 0.003, "step": 20920 }, { "epoch": 159.7709923664122, "grad_norm": 0.05835077539086342, "learning_rate": 6.46992968654132e-05, "loss": 0.0025, "step": 20930 }, { "epoch": 159.84732824427482, "grad_norm": 0.07998442649841309, "learning_rate": 6.456794711081041e-05, "loss": 0.0024, "step": 20940 }, { "epoch": 159.9236641221374, "grad_norm": 0.06980960816144943, "learning_rate": 6.443669424566309e-05, "loss": 0.0025, "step": 20950 }, { "epoch": 160.0, "grad_norm": 0.07643865793943405, "learning_rate": 6.430553841882643e-05, "loss": 0.0024, "step": 20960 }, { "epoch": 160.0763358778626, "grad_norm": 0.07785706222057343, "learning_rate": 6.417447977904577e-05, "loss": 0.0023, "step": 20970 }, { "epoch": 160.15267175572518, "grad_norm": 0.0766030102968216, "learning_rate": 6.40435184749563e-05, "loss": 0.0023, "step": 20980 }, { "epoch": 160.2290076335878, "grad_norm": 0.07036742568016052, "learning_rate": 6.391265465508262e-05, "loss": 0.0022, "step": 20990 }, { "epoch": 160.3053435114504, "grad_norm": 0.08511533588171005, "learning_rate": 6.378188846783898e-05, "loss": 0.0022, "step": 21000 }, { "epoch": 160.38167938931298, "grad_norm": 0.08204297721385956, "learning_rate": 6.365122006152863e-05, "loss": 0.0025, "step": 21010 }, { "epoch": 160.45801526717557, "grad_norm": 0.06618582457304001, "learning_rate": 6.352064958434417e-05, "loss": 0.0024, "step": 21020 }, { "epoch": 160.53435114503816, "grad_norm": 0.07678372412919998, "learning_rate": 6.339017718436705e-05, "loss": 0.0024, "step": 21030 }, { "epoch": 160.61068702290078, "grad_norm": 0.08943220227956772, "learning_rate": 6.325980300956748e-05, "loss": 0.0025, "step": 21040 }, { "epoch": 160.68702290076337, "grad_norm": 0.09826239943504333, "learning_rate": 6.312952720780432e-05, "loss": 0.0026, "step": 21050 }, { "epoch": 160.76335877862596, "grad_norm": 0.1021861732006073, "learning_rate": 6.299934992682476e-05, "loss": 0.0027, "step": 21060 }, { "epoch": 160.83969465648855, "grad_norm": 0.08960767835378647, "learning_rate": 6.286927131426436e-05, "loss": 0.0025, "step": 21070 }, { "epoch": 160.91603053435114, "grad_norm": 0.09621064364910126, "learning_rate": 6.273929151764673e-05, "loss": 0.0027, "step": 21080 }, { "epoch": 160.99236641221373, "grad_norm": 0.10396171361207962, "learning_rate": 6.260941068438344e-05, "loss": 0.0025, "step": 21090 }, { "epoch": 161.06870229007635, "grad_norm": 0.09560813009738922, "learning_rate": 6.247962896177379e-05, "loss": 0.0027, "step": 21100 }, { "epoch": 161.14503816793894, "grad_norm": 0.07029492408037186, "learning_rate": 6.234994649700475e-05, "loss": 0.0027, "step": 21110 }, { "epoch": 161.22137404580153, "grad_norm": 0.06526323407888412, "learning_rate": 6.22203634371506e-05, "loss": 0.0027, "step": 21120 }, { "epoch": 161.29770992366412, "grad_norm": 0.10436400771141052, "learning_rate": 6.209087992917297e-05, "loss": 0.0024, "step": 21130 }, { "epoch": 161.3740458015267, "grad_norm": 0.0836491659283638, "learning_rate": 6.196149611992056e-05, "loss": 0.0026, "step": 21140 }, { "epoch": 161.45038167938932, "grad_norm": 0.07142173498868942, "learning_rate": 6.183221215612904e-05, "loss": 0.0023, "step": 21150 }, { "epoch": 161.5267175572519, "grad_norm": 0.05662307143211365, "learning_rate": 6.170302818442078e-05, "loss": 0.0025, "step": 21160 }, { "epoch": 161.6030534351145, "grad_norm": 0.06735645979642868, "learning_rate": 6.157394435130485e-05, "loss": 0.0024, "step": 21170 }, { "epoch": 161.6793893129771, "grad_norm": 0.07151156663894653, "learning_rate": 6.144496080317654e-05, "loss": 0.0024, "step": 21180 }, { "epoch": 161.75572519083968, "grad_norm": 0.06480149924755096, "learning_rate": 6.131607768631766e-05, "loss": 0.0025, "step": 21190 }, { "epoch": 161.8320610687023, "grad_norm": 0.07937781512737274, "learning_rate": 6.1187295146896e-05, "loss": 0.0022, "step": 21200 }, { "epoch": 161.9083969465649, "grad_norm": 0.07652852684259415, "learning_rate": 6.105861333096532e-05, "loss": 0.0024, "step": 21210 }, { "epoch": 161.98473282442748, "grad_norm": 0.06353870034217834, "learning_rate": 6.0930032384465024e-05, "loss": 0.0026, "step": 21220 }, { "epoch": 162.06106870229007, "grad_norm": 0.08272892236709595, "learning_rate": 6.080155245322025e-05, "loss": 0.0025, "step": 21230 }, { "epoch": 162.13740458015266, "grad_norm": 0.06450434029102325, "learning_rate": 6.067317368294157e-05, "loss": 0.0025, "step": 21240 }, { "epoch": 162.21374045801528, "grad_norm": 0.0729118064045906, "learning_rate": 6.0544896219224764e-05, "loss": 0.0022, "step": 21250 }, { "epoch": 162.29007633587787, "grad_norm": 0.04290701821446419, "learning_rate": 6.041672020755079e-05, "loss": 0.0023, "step": 21260 }, { "epoch": 162.36641221374046, "grad_norm": 0.06472191959619522, "learning_rate": 6.0288645793285516e-05, "loss": 0.0023, "step": 21270 }, { "epoch": 162.44274809160305, "grad_norm": 0.057678356766700745, "learning_rate": 6.016067312167949e-05, "loss": 0.0026, "step": 21280 }, { "epoch": 162.51908396946564, "grad_norm": 0.07438462972640991, "learning_rate": 6.003280233786805e-05, "loss": 0.0023, "step": 21290 }, { "epoch": 162.59541984732823, "grad_norm": 0.06738650798797607, "learning_rate": 5.9905033586870843e-05, "loss": 0.0024, "step": 21300 }, { "epoch": 162.67175572519085, "grad_norm": 0.09184187650680542, "learning_rate": 5.977736701359189e-05, "loss": 0.0023, "step": 21310 }, { "epoch": 162.74809160305344, "grad_norm": 0.07407114654779434, "learning_rate": 5.964980276281933e-05, "loss": 0.0025, "step": 21320 }, { "epoch": 162.82442748091603, "grad_norm": 0.11909207701683044, "learning_rate": 5.952234097922512e-05, "loss": 0.0024, "step": 21330 }, { "epoch": 162.90076335877862, "grad_norm": 0.11075392365455627, "learning_rate": 5.939498180736515e-05, "loss": 0.0026, "step": 21340 }, { "epoch": 162.9770992366412, "grad_norm": 0.08858765661716461, "learning_rate": 5.92677253916789e-05, "loss": 0.0024, "step": 21350 }, { "epoch": 163.05343511450383, "grad_norm": 0.07468192279338837, "learning_rate": 5.9140571876489306e-05, "loss": 0.0022, "step": 21360 }, { "epoch": 163.12977099236642, "grad_norm": 0.06716360151767731, "learning_rate": 5.9013521406002615e-05, "loss": 0.0024, "step": 21370 }, { "epoch": 163.206106870229, "grad_norm": 0.054314784705638885, "learning_rate": 5.888657412430819e-05, "loss": 0.0021, "step": 21380 }, { "epoch": 163.2824427480916, "grad_norm": 0.08088497072458267, "learning_rate": 5.875973017537837e-05, "loss": 0.0023, "step": 21390 }, { "epoch": 163.3587786259542, "grad_norm": 0.0653010681271553, "learning_rate": 5.863298970306831e-05, "loss": 0.0023, "step": 21400 }, { "epoch": 163.4351145038168, "grad_norm": 0.07018575072288513, "learning_rate": 5.850635285111583e-05, "loss": 0.002, "step": 21410 }, { "epoch": 163.5114503816794, "grad_norm": 0.0771377682685852, "learning_rate": 5.837981976314124e-05, "loss": 0.0027, "step": 21420 }, { "epoch": 163.58778625954199, "grad_norm": 0.08034101128578186, "learning_rate": 5.8253390582647055e-05, "loss": 0.0022, "step": 21430 }, { "epoch": 163.66412213740458, "grad_norm": 0.061767689883708954, "learning_rate": 5.812706545301809e-05, "loss": 0.0023, "step": 21440 }, { "epoch": 163.74045801526717, "grad_norm": 0.07505184412002563, "learning_rate": 5.800084451752108e-05, "loss": 0.0021, "step": 21450 }, { "epoch": 163.81679389312978, "grad_norm": 0.05059756711125374, "learning_rate": 5.787472791930463e-05, "loss": 0.002, "step": 21460 }, { "epoch": 163.89312977099237, "grad_norm": 0.08349630236625671, "learning_rate": 5.7748715801398985e-05, "loss": 0.0021, "step": 21470 }, { "epoch": 163.96946564885496, "grad_norm": 0.06029805541038513, "learning_rate": 5.7622808306715985e-05, "loss": 0.0022, "step": 21480 }, { "epoch": 164.04580152671755, "grad_norm": 0.08859965950250626, "learning_rate": 5.749700557804861e-05, "loss": 0.0026, "step": 21490 }, { "epoch": 164.12213740458014, "grad_norm": 0.10337599366903305, "learning_rate": 5.737130775807122e-05, "loss": 0.0024, "step": 21500 }, { "epoch": 164.19847328244273, "grad_norm": 0.07644496858119965, "learning_rate": 5.7245714989339136e-05, "loss": 0.0022, "step": 21510 }, { "epoch": 164.27480916030535, "grad_norm": 0.07437508553266525, "learning_rate": 5.7120227414288513e-05, "loss": 0.0021, "step": 21520 }, { "epoch": 164.35114503816794, "grad_norm": 0.08450701087713242, "learning_rate": 5.699484517523627e-05, "loss": 0.0025, "step": 21530 }, { "epoch": 164.42748091603053, "grad_norm": 0.057751741260290146, "learning_rate": 5.6869568414379815e-05, "loss": 0.0021, "step": 21540 }, { "epoch": 164.50381679389312, "grad_norm": 0.06817489117383957, "learning_rate": 5.674439727379688e-05, "loss": 0.0022, "step": 21550 }, { "epoch": 164.5801526717557, "grad_norm": 0.09481434524059296, "learning_rate": 5.6619331895445474e-05, "loss": 0.0026, "step": 21560 }, { "epoch": 164.65648854961833, "grad_norm": 0.07433342188596725, "learning_rate": 5.64943724211637e-05, "loss": 0.0025, "step": 21570 }, { "epoch": 164.73282442748092, "grad_norm": 0.05995732918381691, "learning_rate": 5.636951899266948e-05, "loss": 0.0023, "step": 21580 }, { "epoch": 164.8091603053435, "grad_norm": 0.08888203650712967, "learning_rate": 5.624477175156051e-05, "loss": 0.0024, "step": 21590 }, { "epoch": 164.8854961832061, "grad_norm": 0.07344658672809601, "learning_rate": 5.6120130839314024e-05, "loss": 0.0026, "step": 21600 }, { "epoch": 164.9618320610687, "grad_norm": 0.06551302969455719, "learning_rate": 5.5995596397286715e-05, "loss": 0.0024, "step": 21610 }, { "epoch": 165.0381679389313, "grad_norm": 0.08344949036836624, "learning_rate": 5.5871168566714495e-05, "loss": 0.0022, "step": 21620 }, { "epoch": 165.1145038167939, "grad_norm": 0.06675373017787933, "learning_rate": 5.574684748871235e-05, "loss": 0.0024, "step": 21630 }, { "epoch": 165.1908396946565, "grad_norm": 0.08589081466197968, "learning_rate": 5.562263330427428e-05, "loss": 0.0025, "step": 21640 }, { "epoch": 165.26717557251908, "grad_norm": 0.07087681442499161, "learning_rate": 5.54985261542729e-05, "loss": 0.0021, "step": 21650 }, { "epoch": 165.34351145038167, "grad_norm": 0.06803197413682938, "learning_rate": 5.537452617945956e-05, "loss": 0.0023, "step": 21660 }, { "epoch": 165.4198473282443, "grad_norm": 0.08389036357402802, "learning_rate": 5.525063352046406e-05, "loss": 0.0024, "step": 21670 }, { "epoch": 165.49618320610688, "grad_norm": 0.07285240292549133, "learning_rate": 5.5126848317794436e-05, "loss": 0.0022, "step": 21680 }, { "epoch": 165.57251908396947, "grad_norm": 0.09471844881772995, "learning_rate": 5.5003170711836934e-05, "loss": 0.0024, "step": 21690 }, { "epoch": 165.64885496183206, "grad_norm": 0.07826350629329681, "learning_rate": 5.487960084285566e-05, "loss": 0.0025, "step": 21700 }, { "epoch": 165.72519083969465, "grad_norm": 0.07638572156429291, "learning_rate": 5.47561388509926e-05, "loss": 0.0023, "step": 21710 }, { "epoch": 165.80152671755727, "grad_norm": 0.07695648074150085, "learning_rate": 5.4632784876267415e-05, "loss": 0.0025, "step": 21720 }, { "epoch": 165.87786259541986, "grad_norm": 0.0597250871360302, "learning_rate": 5.450953905857724e-05, "loss": 0.0025, "step": 21730 }, { "epoch": 165.95419847328245, "grad_norm": 0.0530141219496727, "learning_rate": 5.4386401537696536e-05, "loss": 0.0021, "step": 21740 }, { "epoch": 166.03053435114504, "grad_norm": 0.07000498473644257, "learning_rate": 5.426337245327702e-05, "loss": 0.0023, "step": 21750 }, { "epoch": 166.10687022900763, "grad_norm": 0.07704681158065796, "learning_rate": 5.4140451944847276e-05, "loss": 0.0022, "step": 21760 }, { "epoch": 166.18320610687022, "grad_norm": 0.052257757633924484, "learning_rate": 5.4017640151812846e-05, "loss": 0.0023, "step": 21770 }, { "epoch": 166.25954198473283, "grad_norm": 0.09458326548337936, "learning_rate": 5.389493721345601e-05, "loss": 0.0025, "step": 21780 }, { "epoch": 166.33587786259542, "grad_norm": 0.0783146470785141, "learning_rate": 5.377234326893554e-05, "loss": 0.0022, "step": 21790 }, { "epoch": 166.41221374045801, "grad_norm": 0.055303722620010376, "learning_rate": 5.3649858457286615e-05, "loss": 0.0021, "step": 21800 }, { "epoch": 166.4885496183206, "grad_norm": 0.05343861132860184, "learning_rate": 5.352748291742064e-05, "loss": 0.0022, "step": 21810 }, { "epoch": 166.5648854961832, "grad_norm": 0.07040229439735413, "learning_rate": 5.340521678812511e-05, "loss": 0.0021, "step": 21820 }, { "epoch": 166.6412213740458, "grad_norm": 0.06380795687437057, "learning_rate": 5.3283060208063413e-05, "loss": 0.0021, "step": 21830 }, { "epoch": 166.7175572519084, "grad_norm": 0.07188495993614197, "learning_rate": 5.316101331577469e-05, "loss": 0.0022, "step": 21840 }, { "epoch": 166.793893129771, "grad_norm": 0.06299508363008499, "learning_rate": 5.303907624967377e-05, "loss": 0.0019, "step": 21850 }, { "epoch": 166.87022900763358, "grad_norm": 0.09739651530981064, "learning_rate": 5.291724914805073e-05, "loss": 0.002, "step": 21860 }, { "epoch": 166.94656488549617, "grad_norm": 0.060103341937065125, "learning_rate": 5.279553214907115e-05, "loss": 0.002, "step": 21870 }, { "epoch": 167.0229007633588, "grad_norm": 0.08588945120573044, "learning_rate": 5.267392539077562e-05, "loss": 0.0023, "step": 21880 }, { "epoch": 167.09923664122138, "grad_norm": 0.08422274887561798, "learning_rate": 5.255242901107974e-05, "loss": 0.0021, "step": 21890 }, { "epoch": 167.17557251908397, "grad_norm": 0.07573724538087845, "learning_rate": 5.243104314777394e-05, "loss": 0.0022, "step": 21900 }, { "epoch": 167.25190839694656, "grad_norm": 0.0714113861322403, "learning_rate": 5.230976793852336e-05, "loss": 0.0024, "step": 21910 }, { "epoch": 167.32824427480915, "grad_norm": 0.06494367122650146, "learning_rate": 5.2188603520867446e-05, "loss": 0.002, "step": 21920 }, { "epoch": 167.40458015267177, "grad_norm": 0.0657128319144249, "learning_rate": 5.206755003222019e-05, "loss": 0.0019, "step": 21930 }, { "epoch": 167.48091603053436, "grad_norm": 0.06323283910751343, "learning_rate": 5.194660760986975e-05, "loss": 0.0019, "step": 21940 }, { "epoch": 167.55725190839695, "grad_norm": 0.12619160115718842, "learning_rate": 5.182577639097826e-05, "loss": 0.0022, "step": 21950 }, { "epoch": 167.63358778625954, "grad_norm": 0.07944892346858978, "learning_rate": 5.170505651258178e-05, "loss": 0.0021, "step": 21960 }, { "epoch": 167.70992366412213, "grad_norm": 0.05804096534848213, "learning_rate": 5.158444811159014e-05, "loss": 0.0021, "step": 21970 }, { "epoch": 167.78625954198472, "grad_norm": 0.07673601806163788, "learning_rate": 5.146395132478659e-05, "loss": 0.0021, "step": 21980 }, { "epoch": 167.86259541984734, "grad_norm": 0.05939161777496338, "learning_rate": 5.1343566288827936e-05, "loss": 0.0023, "step": 21990 }, { "epoch": 167.93893129770993, "grad_norm": 0.10250108689069748, "learning_rate": 5.1223293140244216e-05, "loss": 0.0024, "step": 22000 }, { "epoch": 168.01526717557252, "grad_norm": 0.099357008934021, "learning_rate": 5.110313201543856e-05, "loss": 0.002, "step": 22010 }, { "epoch": 168.0916030534351, "grad_norm": 0.05040944740176201, "learning_rate": 5.098308305068706e-05, "loss": 0.0022, "step": 22020 }, { "epoch": 168.1679389312977, "grad_norm": 0.05677852779626846, "learning_rate": 5.0863146382138586e-05, "loss": 0.002, "step": 22030 }, { "epoch": 168.24427480916032, "grad_norm": 0.08553130179643631, "learning_rate": 5.0743322145814695e-05, "loss": 0.0021, "step": 22040 }, { "epoch": 168.3206106870229, "grad_norm": 0.056743260473012924, "learning_rate": 5.062361047760938e-05, "loss": 0.0021, "step": 22050 }, { "epoch": 168.3969465648855, "grad_norm": 0.05666061118245125, "learning_rate": 5.050401151328903e-05, "loss": 0.002, "step": 22060 }, { "epoch": 168.4732824427481, "grad_norm": 0.07207838445901871, "learning_rate": 5.038452538849218e-05, "loss": 0.0021, "step": 22070 }, { "epoch": 168.54961832061068, "grad_norm": 0.0611736886203289, "learning_rate": 5.026515223872933e-05, "loss": 0.0025, "step": 22080 }, { "epoch": 168.6259541984733, "grad_norm": 0.12450186908245087, "learning_rate": 5.0145892199382945e-05, "loss": 0.0023, "step": 22090 }, { "epoch": 168.70229007633588, "grad_norm": 0.06082407757639885, "learning_rate": 5.0026745405707196e-05, "loss": 0.0019, "step": 22100 }, { "epoch": 168.77862595419847, "grad_norm": 0.09416545927524567, "learning_rate": 4.990771199282782e-05, "loss": 0.0022, "step": 22110 }, { "epoch": 168.85496183206106, "grad_norm": 0.06591419875621796, "learning_rate": 4.9788792095741955e-05, "loss": 0.002, "step": 22120 }, { "epoch": 168.93129770992365, "grad_norm": 0.0689043402671814, "learning_rate": 4.966998584931796e-05, "loss": 0.0022, "step": 22130 }, { "epoch": 169.00763358778627, "grad_norm": 0.09316141158342361, "learning_rate": 4.955129338829535e-05, "loss": 0.0022, "step": 22140 }, { "epoch": 169.08396946564886, "grad_norm": 0.0928068682551384, "learning_rate": 4.9432714847284594e-05, "loss": 0.0021, "step": 22150 }, { "epoch": 169.16030534351145, "grad_norm": 0.07600428909063339, "learning_rate": 4.931425036076694e-05, "loss": 0.002, "step": 22160 }, { "epoch": 169.23664122137404, "grad_norm": 0.06812296062707901, "learning_rate": 4.9195900063094327e-05, "loss": 0.0021, "step": 22170 }, { "epoch": 169.31297709923663, "grad_norm": 0.08311454951763153, "learning_rate": 4.90776640884892e-05, "loss": 0.0021, "step": 22180 }, { "epoch": 169.38931297709922, "grad_norm": 0.06359574943780899, "learning_rate": 4.89595425710442e-05, "loss": 0.0024, "step": 22190 }, { "epoch": 169.46564885496184, "grad_norm": 0.06462552398443222, "learning_rate": 4.8841535644722345e-05, "loss": 0.0021, "step": 22200 }, { "epoch": 169.54198473282443, "grad_norm": 0.11824774742126465, "learning_rate": 4.8723643443356574e-05, "loss": 0.0021, "step": 22210 }, { "epoch": 169.61832061068702, "grad_norm": 0.07499135285615921, "learning_rate": 4.8605866100649885e-05, "loss": 0.0021, "step": 22220 }, { "epoch": 169.6946564885496, "grad_norm": 0.09763187915086746, "learning_rate": 4.848820375017475e-05, "loss": 0.0021, "step": 22230 }, { "epoch": 169.7709923664122, "grad_norm": 0.06190437451004982, "learning_rate": 4.8370656525373446e-05, "loss": 0.0023, "step": 22240 }, { "epoch": 169.84732824427482, "grad_norm": 0.06098948046565056, "learning_rate": 4.825322455955758e-05, "loss": 0.0021, "step": 22250 }, { "epoch": 169.9236641221374, "grad_norm": 0.06307119876146317, "learning_rate": 4.8135907985908065e-05, "loss": 0.0021, "step": 22260 }, { "epoch": 170.0, "grad_norm": 0.054521843791007996, "learning_rate": 4.801870693747498e-05, "loss": 0.0022, "step": 22270 }, { "epoch": 170.0763358778626, "grad_norm": 0.07490163296461105, "learning_rate": 4.790162154717738e-05, "loss": 0.0022, "step": 22280 }, { "epoch": 170.15267175572518, "grad_norm": 0.0922338142991066, "learning_rate": 4.778465194780305e-05, "loss": 0.002, "step": 22290 }, { "epoch": 170.2290076335878, "grad_norm": 0.09634072333574295, "learning_rate": 4.766779827200857e-05, "loss": 0.002, "step": 22300 }, { "epoch": 170.3053435114504, "grad_norm": 0.09144623577594757, "learning_rate": 4.7551060652318995e-05, "loss": 0.0022, "step": 22310 }, { "epoch": 170.38167938931298, "grad_norm": 0.083775594830513, "learning_rate": 4.743443922112781e-05, "loss": 0.0022, "step": 22320 }, { "epoch": 170.45801526717557, "grad_norm": 0.048577770590782166, "learning_rate": 4.7317934110696685e-05, "loss": 0.0022, "step": 22330 }, { "epoch": 170.53435114503816, "grad_norm": 0.06618029624223709, "learning_rate": 4.7201545453155425e-05, "loss": 0.002, "step": 22340 }, { "epoch": 170.61068702290078, "grad_norm": 0.05640489608049393, "learning_rate": 4.708527338050161e-05, "loss": 0.0019, "step": 22350 }, { "epoch": 170.68702290076337, "grad_norm": 0.07909304648637772, "learning_rate": 4.696911802460079e-05, "loss": 0.0023, "step": 22360 }, { "epoch": 170.76335877862596, "grad_norm": 0.08773373812437057, "learning_rate": 4.6853079517186035e-05, "loss": 0.0022, "step": 22370 }, { "epoch": 170.83969465648855, "grad_norm": 0.11187729984521866, "learning_rate": 4.6737157989857935e-05, "loss": 0.002, "step": 22380 }, { "epoch": 170.91603053435114, "grad_norm": 0.07338199019432068, "learning_rate": 4.662135357408447e-05, "loss": 0.0022, "step": 22390 }, { "epoch": 170.99236641221373, "grad_norm": 0.06875409930944443, "learning_rate": 4.650566640120061e-05, "loss": 0.0019, "step": 22400 }, { "epoch": 171.06870229007635, "grad_norm": 0.051236592233181, "learning_rate": 4.6390096602408555e-05, "loss": 0.0019, "step": 22410 }, { "epoch": 171.14503816793894, "grad_norm": 0.09550406783819199, "learning_rate": 4.6274644308777255e-05, "loss": 0.0022, "step": 22420 }, { "epoch": 171.22137404580153, "grad_norm": 0.06691096723079681, "learning_rate": 4.615930965124255e-05, "loss": 0.0019, "step": 22430 }, { "epoch": 171.29770992366412, "grad_norm": 0.14559143781661987, "learning_rate": 4.6044092760606794e-05, "loss": 0.0023, "step": 22440 }, { "epoch": 171.3740458015267, "grad_norm": 0.06982150673866272, "learning_rate": 4.592899376753865e-05, "loss": 0.0021, "step": 22450 }, { "epoch": 171.45038167938932, "grad_norm": 0.09084024280309677, "learning_rate": 4.581401280257326e-05, "loss": 0.0022, "step": 22460 }, { "epoch": 171.5267175572519, "grad_norm": 0.09372306615114212, "learning_rate": 4.56991499961118e-05, "loss": 0.0021, "step": 22470 }, { "epoch": 171.6030534351145, "grad_norm": 0.07847164571285248, "learning_rate": 4.5584405478421506e-05, "loss": 0.0018, "step": 22480 }, { "epoch": 171.6793893129771, "grad_norm": 0.08807926625013351, "learning_rate": 4.5469779379635456e-05, "loss": 0.0018, "step": 22490 }, { "epoch": 171.75572519083968, "grad_norm": 0.06494437903165817, "learning_rate": 4.5355271829752306e-05, "loss": 0.0018, "step": 22500 }, { "epoch": 171.8320610687023, "grad_norm": 0.07939145714044571, "learning_rate": 4.524088295863643e-05, "loss": 0.0019, "step": 22510 }, { "epoch": 171.9083969465649, "grad_norm": 0.07586106657981873, "learning_rate": 4.5126612896017515e-05, "loss": 0.0023, "step": 22520 }, { "epoch": 171.98473282442748, "grad_norm": 0.0948263630270958, "learning_rate": 4.501246177149056e-05, "loss": 0.0021, "step": 22530 }, { "epoch": 172.06106870229007, "grad_norm": 0.07565697282552719, "learning_rate": 4.489842971451561e-05, "loss": 0.0022, "step": 22540 }, { "epoch": 172.13740458015266, "grad_norm": 0.05728227272629738, "learning_rate": 4.478451685441775e-05, "loss": 0.0021, "step": 22550 }, { "epoch": 172.21374045801528, "grad_norm": 0.10066434741020203, "learning_rate": 4.467072332038678e-05, "loss": 0.0019, "step": 22560 }, { "epoch": 172.29007633587787, "grad_norm": 0.08130485564470291, "learning_rate": 4.455704924147726e-05, "loss": 0.0021, "step": 22570 }, { "epoch": 172.36641221374046, "grad_norm": 0.07097799330949783, "learning_rate": 4.4443494746608246e-05, "loss": 0.0021, "step": 22580 }, { "epoch": 172.44274809160305, "grad_norm": 0.08592285960912704, "learning_rate": 4.433005996456317e-05, "loss": 0.002, "step": 22590 }, { "epoch": 172.51908396946564, "grad_norm": 0.08530633896589279, "learning_rate": 4.421674502398971e-05, "loss": 0.002, "step": 22600 }, { "epoch": 172.59541984732823, "grad_norm": 0.07332337647676468, "learning_rate": 4.4103550053399666e-05, "loss": 0.0017, "step": 22610 }, { "epoch": 172.67175572519085, "grad_norm": 0.07563639432191849, "learning_rate": 4.399047518116861e-05, "loss": 0.0019, "step": 22620 }, { "epoch": 172.74809160305344, "grad_norm": 0.07271081954240799, "learning_rate": 4.387752053553606e-05, "loss": 0.0021, "step": 22630 }, { "epoch": 172.82442748091603, "grad_norm": 0.062323957681655884, "learning_rate": 4.376468624460524e-05, "loss": 0.002, "step": 22640 }, { "epoch": 172.90076335877862, "grad_norm": 0.060121435672044754, "learning_rate": 4.3651972436342794e-05, "loss": 0.002, "step": 22650 }, { "epoch": 172.9770992366412, "grad_norm": 0.10620911419391632, "learning_rate": 4.353937923857863e-05, "loss": 0.002, "step": 22660 }, { "epoch": 173.05343511450383, "grad_norm": 0.0925116091966629, "learning_rate": 4.3426906779005995e-05, "loss": 0.0019, "step": 22670 }, { "epoch": 173.12977099236642, "grad_norm": 0.06619174033403397, "learning_rate": 4.331455518518121e-05, "loss": 0.0021, "step": 22680 }, { "epoch": 173.206106870229, "grad_norm": 0.06605743616819382, "learning_rate": 4.320232458452344e-05, "loss": 0.0018, "step": 22690 }, { "epoch": 173.2824427480916, "grad_norm": 0.05487534776329994, "learning_rate": 4.309021510431473e-05, "loss": 0.0019, "step": 22700 }, { "epoch": 173.3587786259542, "grad_norm": 0.08186230808496475, "learning_rate": 4.2978226871699704e-05, "loss": 0.0023, "step": 22710 }, { "epoch": 173.4351145038168, "grad_norm": 0.08233430236577988, "learning_rate": 4.286636001368541e-05, "loss": 0.002, "step": 22720 }, { "epoch": 173.5114503816794, "grad_norm": 0.08447281271219254, "learning_rate": 4.2754614657141365e-05, "loss": 0.0022, "step": 22730 }, { "epoch": 173.58778625954199, "grad_norm": 0.09632053226232529, "learning_rate": 4.2642990928799224e-05, "loss": 0.0022, "step": 22740 }, { "epoch": 173.66412213740458, "grad_norm": 0.06393273919820786, "learning_rate": 4.253148895525272e-05, "loss": 0.0023, "step": 22750 }, { "epoch": 173.74045801526717, "grad_norm": 0.06360380351543427, "learning_rate": 4.242010886295752e-05, "loss": 0.0019, "step": 22760 }, { "epoch": 173.81679389312978, "grad_norm": 0.09053414314985275, "learning_rate": 4.230885077823107e-05, "loss": 0.0019, "step": 22770 }, { "epoch": 173.89312977099237, "grad_norm": 0.09916870296001434, "learning_rate": 4.219771482725233e-05, "loss": 0.0022, "step": 22780 }, { "epoch": 173.96946564885496, "grad_norm": 0.08432137966156006, "learning_rate": 4.208670113606189e-05, "loss": 0.002, "step": 22790 }, { "epoch": 174.04580152671755, "grad_norm": 0.08115588873624802, "learning_rate": 4.1975809830561616e-05, "loss": 0.0021, "step": 22800 }, { "epoch": 174.12213740458014, "grad_norm": 0.06385581195354462, "learning_rate": 4.18650410365146e-05, "loss": 0.0019, "step": 22810 }, { "epoch": 174.19847328244273, "grad_norm": 0.06175491586327553, "learning_rate": 4.175439487954502e-05, "loss": 0.0022, "step": 22820 }, { "epoch": 174.27480916030535, "grad_norm": 0.07850757986307144, "learning_rate": 4.164387148513787e-05, "loss": 0.0019, "step": 22830 }, { "epoch": 174.35114503816794, "grad_norm": 0.06935087591409683, "learning_rate": 4.1533470978638925e-05, "loss": 0.0019, "step": 22840 }, { "epoch": 174.42748091603053, "grad_norm": 0.060245662927627563, "learning_rate": 4.142319348525477e-05, "loss": 0.002, "step": 22850 }, { "epoch": 174.50381679389312, "grad_norm": 0.0843137800693512, "learning_rate": 4.131303913005231e-05, "loss": 0.0019, "step": 22860 }, { "epoch": 174.5801526717557, "grad_norm": 0.06442392617464066, "learning_rate": 4.120300803795889e-05, "loss": 0.0023, "step": 22870 }, { "epoch": 174.65648854961833, "grad_norm": 0.06778953969478607, "learning_rate": 4.10931003337619e-05, "loss": 0.002, "step": 22880 }, { "epoch": 174.73282442748092, "grad_norm": 0.06305519491434097, "learning_rate": 4.098331614210899e-05, "loss": 0.0019, "step": 22890 }, { "epoch": 174.8091603053435, "grad_norm": 0.056172944605350494, "learning_rate": 4.087365558750761e-05, "loss": 0.002, "step": 22900 }, { "epoch": 174.8854961832061, "grad_norm": 0.08328600227832794, "learning_rate": 4.0764118794325056e-05, "loss": 0.0019, "step": 22910 }, { "epoch": 174.9618320610687, "grad_norm": 0.072299063205719, "learning_rate": 4.06547058867883e-05, "loss": 0.002, "step": 22920 }, { "epoch": 175.0381679389313, "grad_norm": 0.06204748898744583, "learning_rate": 4.054541698898364e-05, "loss": 0.0018, "step": 22930 }, { "epoch": 175.1145038167939, "grad_norm": 0.08830232173204422, "learning_rate": 4.043625222485688e-05, "loss": 0.0019, "step": 22940 }, { "epoch": 175.1908396946565, "grad_norm": 0.050838302820920944, "learning_rate": 4.032721171821306e-05, "loss": 0.0021, "step": 22950 }, { "epoch": 175.26717557251908, "grad_norm": 0.07141420990228653, "learning_rate": 4.0218295592716205e-05, "loss": 0.002, "step": 22960 }, { "epoch": 175.34351145038167, "grad_norm": 0.05812468379735947, "learning_rate": 4.010950397188934e-05, "loss": 0.0019, "step": 22970 }, { "epoch": 175.4198473282443, "grad_norm": 0.07011732459068298, "learning_rate": 4.000083697911432e-05, "loss": 0.0019, "step": 22980 }, { "epoch": 175.49618320610688, "grad_norm": 0.08934054523706436, "learning_rate": 3.9892294737631494e-05, "loss": 0.002, "step": 22990 }, { "epoch": 175.57251908396947, "grad_norm": 0.06754131615161896, "learning_rate": 3.978387737053994e-05, "loss": 0.0021, "step": 23000 }, { "epoch": 175.64885496183206, "grad_norm": 0.09552391618490219, "learning_rate": 3.9675585000796945e-05, "loss": 0.002, "step": 23010 }, { "epoch": 175.72519083969465, "grad_norm": 0.07913973927497864, "learning_rate": 3.956741775121817e-05, "loss": 0.002, "step": 23020 }, { "epoch": 175.80152671755727, "grad_norm": 0.09618212282657623, "learning_rate": 3.945937574447729e-05, "loss": 0.0019, "step": 23030 }, { "epoch": 175.87786259541986, "grad_norm": 0.05516286566853523, "learning_rate": 3.9351459103106e-05, "loss": 0.0019, "step": 23040 }, { "epoch": 175.95419847328245, "grad_norm": 0.07715852558612823, "learning_rate": 3.924366794949366e-05, "loss": 0.002, "step": 23050 }, { "epoch": 176.03053435114504, "grad_norm": 0.055619798600673676, "learning_rate": 3.913600240588756e-05, "loss": 0.0019, "step": 23060 }, { "epoch": 176.10687022900763, "grad_norm": 0.07984987646341324, "learning_rate": 3.902846259439236e-05, "loss": 0.0018, "step": 23070 }, { "epoch": 176.18320610687022, "grad_norm": 0.08399543911218643, "learning_rate": 3.8921048636970233e-05, "loss": 0.0018, "step": 23080 }, { "epoch": 176.25954198473283, "grad_norm": 0.07985786348581314, "learning_rate": 3.881376065544047e-05, "loss": 0.0019, "step": 23090 }, { "epoch": 176.33587786259542, "grad_norm": 0.0752643346786499, "learning_rate": 3.870659877147962e-05, "loss": 0.0019, "step": 23100 }, { "epoch": 176.41221374045801, "grad_norm": 0.07299000024795532, "learning_rate": 3.859956310662117e-05, "loss": 0.0017, "step": 23110 }, { "epoch": 176.4885496183206, "grad_norm": 0.06956378370523453, "learning_rate": 3.84926537822555e-05, "loss": 0.0018, "step": 23120 }, { "epoch": 176.5648854961832, "grad_norm": 0.0971088856458664, "learning_rate": 3.838587091962966e-05, "loss": 0.002, "step": 23130 }, { "epoch": 176.6412213740458, "grad_norm": 0.04359539970755577, "learning_rate": 3.827921463984738e-05, "loss": 0.0017, "step": 23140 }, { "epoch": 176.7175572519084, "grad_norm": 0.08573538064956665, "learning_rate": 3.8172685063868616e-05, "loss": 0.0019, "step": 23150 }, { "epoch": 176.793893129771, "grad_norm": 0.08720584958791733, "learning_rate": 3.806628231250984e-05, "loss": 0.0021, "step": 23160 }, { "epoch": 176.87022900763358, "grad_norm": 0.054023418575525284, "learning_rate": 3.796000650644359e-05, "loss": 0.0018, "step": 23170 }, { "epoch": 176.94656488549617, "grad_norm": 0.08907061070203781, "learning_rate": 3.7853857766198495e-05, "loss": 0.0019, "step": 23180 }, { "epoch": 177.0229007633588, "grad_norm": 0.05550255998969078, "learning_rate": 3.7747836212159036e-05, "loss": 0.002, "step": 23190 }, { "epoch": 177.09923664122138, "grad_norm": 0.07670805603265762, "learning_rate": 3.7641941964565384e-05, "loss": 0.0017, "step": 23200 }, { "epoch": 177.17557251908397, "grad_norm": 0.08743257820606232, "learning_rate": 3.7536175143513456e-05, "loss": 0.0017, "step": 23210 }, { "epoch": 177.25190839694656, "grad_norm": 0.06264103949069977, "learning_rate": 3.7430535868954596e-05, "loss": 0.0018, "step": 23220 }, { "epoch": 177.32824427480915, "grad_norm": 0.06606828421354294, "learning_rate": 3.732502426069548e-05, "loss": 0.0021, "step": 23230 }, { "epoch": 177.40458015267177, "grad_norm": 0.057837873697280884, "learning_rate": 3.721964043839804e-05, "loss": 0.0017, "step": 23240 }, { "epoch": 177.48091603053436, "grad_norm": 0.07130409777164459, "learning_rate": 3.711438452157923e-05, "loss": 0.0017, "step": 23250 }, { "epoch": 177.55725190839695, "grad_norm": 0.04247087985277176, "learning_rate": 3.700925662961097e-05, "loss": 0.0018, "step": 23260 }, { "epoch": 177.63358778625954, "grad_norm": 0.10796550661325455, "learning_rate": 3.6904256881720035e-05, "loss": 0.0019, "step": 23270 }, { "epoch": 177.70992366412213, "grad_norm": 0.11317647993564606, "learning_rate": 3.679938539698777e-05, "loss": 0.0019, "step": 23280 }, { "epoch": 177.78625954198472, "grad_norm": 0.06834660470485687, "learning_rate": 3.6694642294350184e-05, "loss": 0.002, "step": 23290 }, { "epoch": 177.86259541984734, "grad_norm": 0.09301381558179855, "learning_rate": 3.659002769259753e-05, "loss": 0.0019, "step": 23300 }, { "epoch": 177.93893129770993, "grad_norm": 0.06996477395296097, "learning_rate": 3.6485541710374434e-05, "loss": 0.0019, "step": 23310 }, { "epoch": 178.01526717557252, "grad_norm": 0.06243106722831726, "learning_rate": 3.6381184466179625e-05, "loss": 0.0018, "step": 23320 }, { "epoch": 178.0916030534351, "grad_norm": 0.048459943383932114, "learning_rate": 3.627695607836583e-05, "loss": 0.0015, "step": 23330 }, { "epoch": 178.1679389312977, "grad_norm": 0.09365159273147583, "learning_rate": 3.617285666513966e-05, "loss": 0.0019, "step": 23340 }, { "epoch": 178.24427480916032, "grad_norm": 0.09426851570606232, "learning_rate": 3.606888634456149e-05, "loss": 0.0017, "step": 23350 }, { "epoch": 178.3206106870229, "grad_norm": 0.07546114921569824, "learning_rate": 3.596504523454509e-05, "loss": 0.0018, "step": 23360 }, { "epoch": 178.3969465648855, "grad_norm": 0.09224723279476166, "learning_rate": 3.5861333452857944e-05, "loss": 0.0019, "step": 23370 }, { "epoch": 178.4732824427481, "grad_norm": 0.05902371183037758, "learning_rate": 3.57577511171207e-05, "loss": 0.0016, "step": 23380 }, { "epoch": 178.54961832061068, "grad_norm": 0.08460257947444916, "learning_rate": 3.565429834480729e-05, "loss": 0.0017, "step": 23390 }, { "epoch": 178.6259541984733, "grad_norm": 0.05433852598071098, "learning_rate": 3.555097525324464e-05, "loss": 0.0019, "step": 23400 }, { "epoch": 178.70229007633588, "grad_norm": 0.09826377779245377, "learning_rate": 3.544778195961272e-05, "loss": 0.0016, "step": 23410 }, { "epoch": 178.77862595419847, "grad_norm": 0.06174956634640694, "learning_rate": 3.5344718580944076e-05, "loss": 0.0019, "step": 23420 }, { "epoch": 178.85496183206106, "grad_norm": 0.07483351230621338, "learning_rate": 3.5241785234124126e-05, "loss": 0.0016, "step": 23430 }, { "epoch": 178.93129770992365, "grad_norm": 0.08738655596971512, "learning_rate": 3.513898203589073e-05, "loss": 0.0019, "step": 23440 }, { "epoch": 179.00763358778627, "grad_norm": 0.05621917545795441, "learning_rate": 3.503630910283416e-05, "loss": 0.0017, "step": 23450 }, { "epoch": 179.08396946564886, "grad_norm": 0.07926781475543976, "learning_rate": 3.493376655139692e-05, "loss": 0.0018, "step": 23460 }, { "epoch": 179.16030534351145, "grad_norm": 0.1674116849899292, "learning_rate": 3.48313544978737e-05, "loss": 0.0018, "step": 23470 }, { "epoch": 179.23664122137404, "grad_norm": 0.13762518763542175, "learning_rate": 3.472907305841118e-05, "loss": 0.0018, "step": 23480 }, { "epoch": 179.31297709923663, "grad_norm": 0.0829978883266449, "learning_rate": 3.4626922349007856e-05, "loss": 0.0017, "step": 23490 }, { "epoch": 179.38931297709922, "grad_norm": 0.10244785249233246, "learning_rate": 3.452490248551404e-05, "loss": 0.0019, "step": 23500 }, { "epoch": 179.46564885496184, "grad_norm": 0.08845517039299011, "learning_rate": 3.442301358363163e-05, "loss": 0.0018, "step": 23510 }, { "epoch": 179.54198473282443, "grad_norm": 0.07827398926019669, "learning_rate": 3.4321255758913904e-05, "loss": 0.0017, "step": 23520 }, { "epoch": 179.61832061068702, "grad_norm": 0.10066317766904831, "learning_rate": 3.421962912676559e-05, "loss": 0.0018, "step": 23530 }, { "epoch": 179.6946564885496, "grad_norm": 0.07378624379634857, "learning_rate": 3.41181338024426e-05, "loss": 0.0018, "step": 23540 }, { "epoch": 179.7709923664122, "grad_norm": 0.09591282159090042, "learning_rate": 3.401676990105195e-05, "loss": 0.0015, "step": 23550 }, { "epoch": 179.84732824427482, "grad_norm": 0.05835454910993576, "learning_rate": 3.391553753755155e-05, "loss": 0.0019, "step": 23560 }, { "epoch": 179.9236641221374, "grad_norm": 0.11041402816772461, "learning_rate": 3.381443682675023e-05, "loss": 0.0019, "step": 23570 }, { "epoch": 180.0, "grad_norm": 0.06344657391309738, "learning_rate": 3.371346788330737e-05, "loss": 0.0017, "step": 23580 }, { "epoch": 180.0763358778626, "grad_norm": 0.058059703558683395, "learning_rate": 3.361263082173301e-05, "loss": 0.0017, "step": 23590 }, { "epoch": 180.15267175572518, "grad_norm": 0.09026490151882172, "learning_rate": 3.351192575638761e-05, "loss": 0.0014, "step": 23600 }, { "epoch": 180.2290076335878, "grad_norm": 0.08687600493431091, "learning_rate": 3.341135280148192e-05, "loss": 0.0021, "step": 23610 }, { "epoch": 180.3053435114504, "grad_norm": 0.07239003479480743, "learning_rate": 3.331091207107689e-05, "loss": 0.0019, "step": 23620 }, { "epoch": 180.38167938931298, "grad_norm": 0.0853022113442421, "learning_rate": 3.321060367908344e-05, "loss": 0.0015, "step": 23630 }, { "epoch": 180.45801526717557, "grad_norm": 0.0760430321097374, "learning_rate": 3.311042773926246e-05, "loss": 0.0017, "step": 23640 }, { "epoch": 180.53435114503816, "grad_norm": 0.08154309540987015, "learning_rate": 3.301038436522461e-05, "loss": 0.0019, "step": 23650 }, { "epoch": 180.61068702290078, "grad_norm": 0.10787450522184372, "learning_rate": 3.291047367043023e-05, "loss": 0.0017, "step": 23660 }, { "epoch": 180.68702290076337, "grad_norm": 0.07328809797763824, "learning_rate": 3.281069576818916e-05, "loss": 0.0017, "step": 23670 }, { "epoch": 180.76335877862596, "grad_norm": 0.08556774258613586, "learning_rate": 3.271105077166066e-05, "loss": 0.0018, "step": 23680 }, { "epoch": 180.83969465648855, "grad_norm": 0.059831179678440094, "learning_rate": 3.261153879385321e-05, "loss": 0.0018, "step": 23690 }, { "epoch": 180.91603053435114, "grad_norm": 0.0790327861905098, "learning_rate": 3.251215994762449e-05, "loss": 0.0017, "step": 23700 }, { "epoch": 180.99236641221373, "grad_norm": 0.06402378529310226, "learning_rate": 3.241291434568116e-05, "loss": 0.0017, "step": 23710 }, { "epoch": 181.06870229007635, "grad_norm": 0.0466703362762928, "learning_rate": 3.231380210057883e-05, "loss": 0.0017, "step": 23720 }, { "epoch": 181.14503816793894, "grad_norm": 0.07791215926408768, "learning_rate": 3.221482332472172e-05, "loss": 0.0016, "step": 23730 }, { "epoch": 181.22137404580153, "grad_norm": 0.04788624867796898, "learning_rate": 3.2115978130362845e-05, "loss": 0.0017, "step": 23740 }, { "epoch": 181.29770992366412, "grad_norm": 0.07351028174161911, "learning_rate": 3.2017266629603625e-05, "loss": 0.0017, "step": 23750 }, { "epoch": 181.3740458015267, "grad_norm": 0.08274664729833603, "learning_rate": 3.1918688934393896e-05, "loss": 0.0018, "step": 23760 }, { "epoch": 181.45038167938932, "grad_norm": 0.05913885310292244, "learning_rate": 3.182024515653175e-05, "loss": 0.0018, "step": 23770 }, { "epoch": 181.5267175572519, "grad_norm": 0.07082172483205795, "learning_rate": 3.172193540766342e-05, "loss": 0.0018, "step": 23780 }, { "epoch": 181.6030534351145, "grad_norm": 0.07528842985630035, "learning_rate": 3.1623759799283005e-05, "loss": 0.0017, "step": 23790 }, { "epoch": 181.6793893129771, "grad_norm": 0.09175938367843628, "learning_rate": 3.152571844273261e-05, "loss": 0.0017, "step": 23800 }, { "epoch": 181.75572519083968, "grad_norm": 0.062117286026477814, "learning_rate": 3.142781144920208e-05, "loss": 0.0017, "step": 23810 }, { "epoch": 181.8320610687023, "grad_norm": 0.0748920887708664, "learning_rate": 3.13300389297288e-05, "loss": 0.0016, "step": 23820 }, { "epoch": 181.9083969465649, "grad_norm": 0.085489921271801, "learning_rate": 3.1232400995197715e-05, "loss": 0.0017, "step": 23830 }, { "epoch": 181.98473282442748, "grad_norm": 0.0940658375620842, "learning_rate": 3.113489775634113e-05, "loss": 0.0016, "step": 23840 }, { "epoch": 182.06106870229007, "grad_norm": 0.08191808313131332, "learning_rate": 3.1037529323738494e-05, "loss": 0.0016, "step": 23850 }, { "epoch": 182.13740458015266, "grad_norm": 0.08981950581073761, "learning_rate": 3.094029580781651e-05, "loss": 0.0019, "step": 23860 }, { "epoch": 182.21374045801528, "grad_norm": 0.11262917518615723, "learning_rate": 3.084319731884876e-05, "loss": 0.0019, "step": 23870 }, { "epoch": 182.29007633587787, "grad_norm": 0.07363846898078918, "learning_rate": 3.0746233966955774e-05, "loss": 0.002, "step": 23880 }, { "epoch": 182.36641221374046, "grad_norm": 0.10349009931087494, "learning_rate": 3.064940586210477e-05, "loss": 0.0018, "step": 23890 }, { "epoch": 182.44274809160305, "grad_norm": 0.1284685730934143, "learning_rate": 3.055271311410959e-05, "loss": 0.0018, "step": 23900 }, { "epoch": 182.51908396946564, "grad_norm": 0.07555246353149414, "learning_rate": 3.045615583263058e-05, "loss": 0.0018, "step": 23910 }, { "epoch": 182.59541984732823, "grad_norm": 0.10001549124717712, "learning_rate": 3.0359734127174428e-05, "loss": 0.0019, "step": 23920 }, { "epoch": 182.67175572519085, "grad_norm": 0.07043018192052841, "learning_rate": 3.0263448107094096e-05, "loss": 0.0017, "step": 23930 }, { "epoch": 182.74809160305344, "grad_norm": 0.05772029235959053, "learning_rate": 3.016729788158868e-05, "loss": 0.0016, "step": 23940 }, { "epoch": 182.82442748091603, "grad_norm": 0.061876680701971054, "learning_rate": 3.0071283559703157e-05, "loss": 0.0017, "step": 23950 }, { "epoch": 182.90076335877862, "grad_norm": 0.06596852838993073, "learning_rate": 2.9975405250328487e-05, "loss": 0.0017, "step": 23960 }, { "epoch": 182.9770992366412, "grad_norm": 0.09211495518684387, "learning_rate": 2.9879663062201343e-05, "loss": 0.0015, "step": 23970 }, { "epoch": 183.05343511450383, "grad_norm": 0.055858220905065536, "learning_rate": 2.978405710390403e-05, "loss": 0.0015, "step": 23980 }, { "epoch": 183.12977099236642, "grad_norm": 0.06461702287197113, "learning_rate": 2.968858748386437e-05, "loss": 0.0017, "step": 23990 }, { "epoch": 183.206106870229, "grad_norm": 0.06024142727255821, "learning_rate": 2.9593254310355478e-05, "loss": 0.0016, "step": 24000 }, { "epoch": 183.2824427480916, "grad_norm": 0.05453919246792793, "learning_rate": 2.9498057691495798e-05, "loss": 0.0017, "step": 24010 }, { "epoch": 183.3587786259542, "grad_norm": 0.0708857923746109, "learning_rate": 2.9402997735248906e-05, "loss": 0.0016, "step": 24020 }, { "epoch": 183.4351145038168, "grad_norm": 0.0873054787516594, "learning_rate": 2.930807454942337e-05, "loss": 0.0017, "step": 24030 }, { "epoch": 183.5114503816794, "grad_norm": 0.08687259256839752, "learning_rate": 2.9213288241672643e-05, "loss": 0.0017, "step": 24040 }, { "epoch": 183.58778625954199, "grad_norm": 0.07108956575393677, "learning_rate": 2.9118638919494986e-05, "loss": 0.0018, "step": 24050 }, { "epoch": 183.66412213740458, "grad_norm": 0.05077968165278435, "learning_rate": 2.902412669023318e-05, "loss": 0.0017, "step": 24060 }, { "epoch": 183.74045801526717, "grad_norm": 0.06156648322939873, "learning_rate": 2.892975166107465e-05, "loss": 0.0017, "step": 24070 }, { "epoch": 183.81679389312978, "grad_norm": 0.052934497594833374, "learning_rate": 2.883551393905115e-05, "loss": 0.0018, "step": 24080 }, { "epoch": 183.89312977099237, "grad_norm": 0.0803590640425682, "learning_rate": 2.8741413631038873e-05, "loss": 0.0019, "step": 24090 }, { "epoch": 183.96946564885496, "grad_norm": 0.08565166592597961, "learning_rate": 2.8647450843757897e-05, "loss": 0.0018, "step": 24100 }, { "epoch": 184.04580152671755, "grad_norm": 0.08395157009363174, "learning_rate": 2.855362568377253e-05, "loss": 0.0018, "step": 24110 }, { "epoch": 184.12213740458014, "grad_norm": 0.06767348200082779, "learning_rate": 2.8459938257490937e-05, "loss": 0.0015, "step": 24120 }, { "epoch": 184.19847328244273, "grad_norm": 0.05654019117355347, "learning_rate": 2.8366388671165113e-05, "loss": 0.0018, "step": 24130 }, { "epoch": 184.27480916030535, "grad_norm": 0.10666227340698242, "learning_rate": 2.827297703089067e-05, "loss": 0.0017, "step": 24140 }, { "epoch": 184.35114503816794, "grad_norm": 0.07185996323823929, "learning_rate": 2.817970344260685e-05, "loss": 0.0017, "step": 24150 }, { "epoch": 184.42748091603053, "grad_norm": 0.07732955366373062, "learning_rate": 2.8086568012096215e-05, "loss": 0.0017, "step": 24160 }, { "epoch": 184.50381679389312, "grad_norm": 0.0503440760076046, "learning_rate": 2.7993570844984747e-05, "loss": 0.0015, "step": 24170 }, { "epoch": 184.5801526717557, "grad_norm": 0.09067167341709137, "learning_rate": 2.7900712046741556e-05, "loss": 0.0018, "step": 24180 }, { "epoch": 184.65648854961833, "grad_norm": 0.0919594019651413, "learning_rate": 2.780799172267889e-05, "loss": 0.0016, "step": 24190 }, { "epoch": 184.73282442748092, "grad_norm": 0.06016545370221138, "learning_rate": 2.771540997795188e-05, "loss": 0.0018, "step": 24200 }, { "epoch": 184.8091603053435, "grad_norm": 0.07643895596265793, "learning_rate": 2.7622966917558588e-05, "loss": 0.0015, "step": 24210 }, { "epoch": 184.8854961832061, "grad_norm": 0.09329942613840103, "learning_rate": 2.7530662646339664e-05, "loss": 0.0014, "step": 24220 }, { "epoch": 184.9618320610687, "grad_norm": 0.06850417703390121, "learning_rate": 2.7438497268978436e-05, "loss": 0.0017, "step": 24230 }, { "epoch": 185.0381679389313, "grad_norm": 0.07056055217981339, "learning_rate": 2.7346470890000717e-05, "loss": 0.0016, "step": 24240 }, { "epoch": 185.1145038167939, "grad_norm": 0.0516885407269001, "learning_rate": 2.7254583613774644e-05, "loss": 0.0014, "step": 24250 }, { "epoch": 185.1908396946565, "grad_norm": 0.07472877949476242, "learning_rate": 2.716283554451068e-05, "loss": 0.0015, "step": 24260 }, { "epoch": 185.26717557251908, "grad_norm": 0.06652415543794632, "learning_rate": 2.707122678626124e-05, "loss": 0.0019, "step": 24270 }, { "epoch": 185.34351145038167, "grad_norm": 0.06315628439188004, "learning_rate": 2.6979757442920902e-05, "loss": 0.0017, "step": 24280 }, { "epoch": 185.4198473282443, "grad_norm": 0.04097567871212959, "learning_rate": 2.6888427618226067e-05, "loss": 0.0015, "step": 24290 }, { "epoch": 185.49618320610688, "grad_norm": 0.053865499794483185, "learning_rate": 2.6797237415754947e-05, "loss": 0.0018, "step": 24300 }, { "epoch": 185.57251908396947, "grad_norm": 0.06757107377052307, "learning_rate": 2.6706186938927416e-05, "loss": 0.0016, "step": 24310 }, { "epoch": 185.64885496183206, "grad_norm": 0.08699755370616913, "learning_rate": 2.6615276291004756e-05, "loss": 0.0015, "step": 24320 }, { "epoch": 185.72519083969465, "grad_norm": 0.06345250457525253, "learning_rate": 2.652450557508981e-05, "loss": 0.0015, "step": 24330 }, { "epoch": 185.80152671755727, "grad_norm": 0.06574654579162598, "learning_rate": 2.643387489412665e-05, "loss": 0.0018, "step": 24340 }, { "epoch": 185.87786259541986, "grad_norm": 0.107233427464962, "learning_rate": 2.634338435090057e-05, "loss": 0.0018, "step": 24350 }, { "epoch": 185.95419847328245, "grad_norm": 0.05704833194613457, "learning_rate": 2.625303404803793e-05, "loss": 0.0015, "step": 24360 }, { "epoch": 186.03053435114504, "grad_norm": 0.04696066677570343, "learning_rate": 2.616282408800603e-05, "loss": 0.0016, "step": 24370 }, { "epoch": 186.10687022900763, "grad_norm": 0.06226532533764839, "learning_rate": 2.6072754573112948e-05, "loss": 0.0017, "step": 24380 }, { "epoch": 186.18320610687022, "grad_norm": 0.10011943429708481, "learning_rate": 2.598282560550759e-05, "loss": 0.0018, "step": 24390 }, { "epoch": 186.25954198473283, "grad_norm": 0.050568193197250366, "learning_rate": 2.589303728717939e-05, "loss": 0.0017, "step": 24400 }, { "epoch": 186.33587786259542, "grad_norm": 0.09756170958280563, "learning_rate": 2.5803389719958294e-05, "loss": 0.0015, "step": 24410 }, { "epoch": 186.41221374045801, "grad_norm": 0.08217103034257889, "learning_rate": 2.5713883005514674e-05, "loss": 0.0015, "step": 24420 }, { "epoch": 186.4885496183206, "grad_norm": 0.15761341154575348, "learning_rate": 2.5624517245359012e-05, "loss": 0.002, "step": 24430 }, { "epoch": 186.5648854961832, "grad_norm": 0.060979973524808884, "learning_rate": 2.5535292540842067e-05, "loss": 0.0017, "step": 24440 }, { "epoch": 186.6412213740458, "grad_norm": 0.068170465528965, "learning_rate": 2.54462089931546e-05, "loss": 0.0014, "step": 24450 }, { "epoch": 186.7175572519084, "grad_norm": 0.05390310660004616, "learning_rate": 2.5357266703327257e-05, "loss": 0.0015, "step": 24460 }, { "epoch": 186.793893129771, "grad_norm": 0.06509419530630112, "learning_rate": 2.5268465772230483e-05, "loss": 0.0014, "step": 24470 }, { "epoch": 186.87022900763358, "grad_norm": 0.06074541062116623, "learning_rate": 2.5179806300574458e-05, "loss": 0.0016, "step": 24480 }, { "epoch": 186.94656488549617, "grad_norm": 0.08062497526407242, "learning_rate": 2.5091288388908827e-05, "loss": 0.0017, "step": 24490 }, { "epoch": 187.0229007633588, "grad_norm": 0.05213091894984245, "learning_rate": 2.500291213762274e-05, "loss": 0.0015, "step": 24500 }, { "epoch": 187.09923664122138, "grad_norm": 0.06698432564735413, "learning_rate": 2.4914677646944786e-05, "loss": 0.0016, "step": 24510 }, { "epoch": 187.17557251908397, "grad_norm": 0.08073922246694565, "learning_rate": 2.4826585016942674e-05, "loss": 0.0023, "step": 24520 }, { "epoch": 187.25190839694656, "grad_norm": 0.08055303245782852, "learning_rate": 2.4738634347523202e-05, "loss": 0.0016, "step": 24530 }, { "epoch": 187.32824427480915, "grad_norm": 0.07889450341463089, "learning_rate": 2.4650825738432244e-05, "loss": 0.0017, "step": 24540 }, { "epoch": 187.40458015267177, "grad_norm": 0.07702290266752243, "learning_rate": 2.456315928925453e-05, "loss": 0.0019, "step": 24550 }, { "epoch": 187.48091603053436, "grad_norm": 0.06622006744146347, "learning_rate": 2.4475635099413576e-05, "loss": 0.0016, "step": 24560 }, { "epoch": 187.55725190839695, "grad_norm": 0.044217534363269806, "learning_rate": 2.438825326817155e-05, "loss": 0.0016, "step": 24570 }, { "epoch": 187.63358778625954, "grad_norm": 0.057145971804857254, "learning_rate": 2.430101389462922e-05, "loss": 0.0017, "step": 24580 }, { "epoch": 187.70992366412213, "grad_norm": 0.05084016174077988, "learning_rate": 2.421391707772566e-05, "loss": 0.0015, "step": 24590 }, { "epoch": 187.78625954198472, "grad_norm": 0.06321080029010773, "learning_rate": 2.4126962916238372e-05, "loss": 0.0016, "step": 24600 }, { "epoch": 187.86259541984734, "grad_norm": 0.060190025717020035, "learning_rate": 2.40401515087831e-05, "loss": 0.0014, "step": 24610 }, { "epoch": 187.93893129770993, "grad_norm": 0.10487782955169678, "learning_rate": 2.3953482953813584e-05, "loss": 0.0016, "step": 24620 }, { "epoch": 188.01526717557252, "grad_norm": 0.15454520285129547, "learning_rate": 2.3866957349621668e-05, "loss": 0.0016, "step": 24630 }, { "epoch": 188.0916030534351, "grad_norm": 0.0815773457288742, "learning_rate": 2.3780574794337004e-05, "loss": 0.0014, "step": 24640 }, { "epoch": 188.1679389312977, "grad_norm": 0.06713855266571045, "learning_rate": 2.3694335385926978e-05, "loss": 0.0016, "step": 24650 }, { "epoch": 188.24427480916032, "grad_norm": 0.11670026183128357, "learning_rate": 2.3608239222196683e-05, "loss": 0.0017, "step": 24660 }, { "epoch": 188.3206106870229, "grad_norm": 0.08409931510686874, "learning_rate": 2.3522286400788803e-05, "loss": 0.0016, "step": 24670 }, { "epoch": 188.3969465648855, "grad_norm": 0.06676127761602402, "learning_rate": 2.3436477019183343e-05, "loss": 0.0018, "step": 24680 }, { "epoch": 188.4732824427481, "grad_norm": 0.070936419069767, "learning_rate": 2.335081117469777e-05, "loss": 0.0015, "step": 24690 }, { "epoch": 188.54961832061068, "grad_norm": 0.11781471222639084, "learning_rate": 2.3265288964486604e-05, "loss": 0.0015, "step": 24700 }, { "epoch": 188.6259541984733, "grad_norm": 0.07344403117895126, "learning_rate": 2.317991048554153e-05, "loss": 0.0014, "step": 24710 }, { "epoch": 188.70229007633588, "grad_norm": 0.10494732111692429, "learning_rate": 2.309467583469134e-05, "loss": 0.0016, "step": 24720 }, { "epoch": 188.77862595419847, "grad_norm": 0.1278795599937439, "learning_rate": 2.3009585108601557e-05, "loss": 0.0017, "step": 24730 }, { "epoch": 188.85496183206106, "grad_norm": 0.10972313582897186, "learning_rate": 2.2924638403774593e-05, "loss": 0.0017, "step": 24740 }, { "epoch": 188.93129770992365, "grad_norm": 0.09441270679235458, "learning_rate": 2.2839835816549363e-05, "loss": 0.0016, "step": 24750 }, { "epoch": 189.00763358778627, "grad_norm": 0.07510344684123993, "learning_rate": 2.27551774431015e-05, "loss": 0.0014, "step": 24760 }, { "epoch": 189.08396946564886, "grad_norm": 0.0891740471124649, "learning_rate": 2.267066337944301e-05, "loss": 0.0016, "step": 24770 }, { "epoch": 189.16030534351145, "grad_norm": 0.06708107888698578, "learning_rate": 2.258629372142225e-05, "loss": 0.0017, "step": 24780 }, { "epoch": 189.23664122137404, "grad_norm": 0.07057501375675201, "learning_rate": 2.250206856472383e-05, "loss": 0.0016, "step": 24790 }, { "epoch": 189.31297709923663, "grad_norm": 0.05878053605556488, "learning_rate": 2.241798800486839e-05, "loss": 0.0015, "step": 24800 }, { "epoch": 189.38931297709922, "grad_norm": 0.05496065318584442, "learning_rate": 2.2334052137212675e-05, "loss": 0.0015, "step": 24810 }, { "epoch": 189.46564885496184, "grad_norm": 0.06391538679599762, "learning_rate": 2.2250261056949304e-05, "loss": 0.0018, "step": 24820 }, { "epoch": 189.54198473282443, "grad_norm": 0.08336050808429718, "learning_rate": 2.2166614859106657e-05, "loss": 0.0015, "step": 24830 }, { "epoch": 189.61832061068702, "grad_norm": 0.06072665750980377, "learning_rate": 2.2083113638548873e-05, "loss": 0.0015, "step": 24840 }, { "epoch": 189.6946564885496, "grad_norm": 0.07213082164525986, "learning_rate": 2.199975748997563e-05, "loss": 0.0017, "step": 24850 }, { "epoch": 189.7709923664122, "grad_norm": 0.0628577470779419, "learning_rate": 2.1916546507922038e-05, "loss": 0.0015, "step": 24860 }, { "epoch": 189.84732824427482, "grad_norm": 0.056142013520002365, "learning_rate": 2.1833480786758617e-05, "loss": 0.0015, "step": 24870 }, { "epoch": 189.9236641221374, "grad_norm": 0.08306145668029785, "learning_rate": 2.1750560420691153e-05, "loss": 0.0014, "step": 24880 }, { "epoch": 190.0, "grad_norm": 0.1032290831208229, "learning_rate": 2.1667785503760575e-05, "loss": 0.0019, "step": 24890 }, { "epoch": 190.0763358778626, "grad_norm": 0.10549236088991165, "learning_rate": 2.1585156129842823e-05, "loss": 0.0016, "step": 24900 }, { "epoch": 190.15267175572518, "grad_norm": 0.07933816313743591, "learning_rate": 2.1502672392648885e-05, "loss": 0.0017, "step": 24910 }, { "epoch": 190.2290076335878, "grad_norm": 0.06975159049034119, "learning_rate": 2.1420334385724346e-05, "loss": 0.0016, "step": 24920 }, { "epoch": 190.3053435114504, "grad_norm": 0.06995077431201935, "learning_rate": 2.13381422024498e-05, "loss": 0.0017, "step": 24930 }, { "epoch": 190.38167938931298, "grad_norm": 0.17419399321079254, "learning_rate": 2.1256095936040292e-05, "loss": 0.0017, "step": 24940 }, { "epoch": 190.45801526717557, "grad_norm": 0.05566052719950676, "learning_rate": 2.117419567954544e-05, "loss": 0.0016, "step": 24950 }, { "epoch": 190.53435114503816, "grad_norm": 0.04833453521132469, "learning_rate": 2.1092441525849184e-05, "loss": 0.0013, "step": 24960 }, { "epoch": 190.61068702290078, "grad_norm": 0.09262041002511978, "learning_rate": 2.101083356766989e-05, "loss": 0.0016, "step": 24970 }, { "epoch": 190.68702290076337, "grad_norm": 0.05508921295404434, "learning_rate": 2.0929371897560032e-05, "loss": 0.0016, "step": 24980 }, { "epoch": 190.76335877862596, "grad_norm": 0.0655728131532669, "learning_rate": 2.0848056607906233e-05, "loss": 0.0014, "step": 24990 }, { "epoch": 190.83969465648855, "grad_norm": 0.09373540431261063, "learning_rate": 2.076688779092907e-05, "loss": 0.0013, "step": 25000 }, { "epoch": 190.91603053435114, "grad_norm": 0.07756651937961578, "learning_rate": 2.068586553868306e-05, "loss": 0.0015, "step": 25010 }, { "epoch": 190.99236641221373, "grad_norm": 0.06997334212064743, "learning_rate": 2.0604989943056366e-05, "loss": 0.0015, "step": 25020 }, { "epoch": 191.06870229007635, "grad_norm": 0.07587023824453354, "learning_rate": 2.052426109577095e-05, "loss": 0.0014, "step": 25030 }, { "epoch": 191.14503816793894, "grad_norm": 0.06477432698011398, "learning_rate": 2.0443679088382326e-05, "loss": 0.0014, "step": 25040 }, { "epoch": 191.22137404580153, "grad_norm": 0.06418919563293457, "learning_rate": 2.0363244012279456e-05, "loss": 0.0015, "step": 25050 }, { "epoch": 191.29770992366412, "grad_norm": 0.07221043109893799, "learning_rate": 2.028295595868469e-05, "loss": 0.0014, "step": 25060 }, { "epoch": 191.3740458015267, "grad_norm": 0.04291442781686783, "learning_rate": 2.0202815018653536e-05, "loss": 0.0015, "step": 25070 }, { "epoch": 191.45038167938932, "grad_norm": 0.0676242783665657, "learning_rate": 2.0122821283074807e-05, "loss": 0.0015, "step": 25080 }, { "epoch": 191.5267175572519, "grad_norm": 0.07291562110185623, "learning_rate": 2.0042974842670273e-05, "loss": 0.0015, "step": 25090 }, { "epoch": 191.6030534351145, "grad_norm": 0.07170842587947845, "learning_rate": 1.9963275787994708e-05, "loss": 0.0015, "step": 25100 }, { "epoch": 191.6793893129771, "grad_norm": 0.043904952704906464, "learning_rate": 1.9883724209435682e-05, "loss": 0.0015, "step": 25110 }, { "epoch": 191.75572519083968, "grad_norm": 0.057069532573223114, "learning_rate": 1.9804320197213554e-05, "loss": 0.0017, "step": 25120 }, { "epoch": 191.8320610687023, "grad_norm": 0.056121546775102615, "learning_rate": 1.9725063841381307e-05, "loss": 0.0014, "step": 25130 }, { "epoch": 191.9083969465649, "grad_norm": 0.056225087493658066, "learning_rate": 1.9645955231824473e-05, "loss": 0.0016, "step": 25140 }, { "epoch": 191.98473282442748, "grad_norm": 0.06080973520874977, "learning_rate": 1.9566994458261004e-05, "loss": 0.0014, "step": 25150 }, { "epoch": 192.06106870229007, "grad_norm": 0.07437245547771454, "learning_rate": 1.9488181610241233e-05, "loss": 0.0015, "step": 25160 }, { "epoch": 192.13740458015266, "grad_norm": 0.06528355926275253, "learning_rate": 1.9409516777147693e-05, "loss": 0.0014, "step": 25170 }, { "epoch": 192.21374045801528, "grad_norm": 0.07784280180931091, "learning_rate": 1.9331000048195038e-05, "loss": 0.0015, "step": 25180 }, { "epoch": 192.29007633587787, "grad_norm": 0.05951184406876564, "learning_rate": 1.9252631512429973e-05, "loss": 0.0014, "step": 25190 }, { "epoch": 192.36641221374046, "grad_norm": 0.0781291127204895, "learning_rate": 1.9174411258731164e-05, "loss": 0.0013, "step": 25200 }, { "epoch": 192.44274809160305, "grad_norm": 0.050289757549762726, "learning_rate": 1.9096339375809073e-05, "loss": 0.0015, "step": 25210 }, { "epoch": 192.51908396946564, "grad_norm": 0.09034400433301926, "learning_rate": 1.9018415952205916e-05, "loss": 0.0015, "step": 25220 }, { "epoch": 192.59541984732823, "grad_norm": 0.0622110478579998, "learning_rate": 1.894064107629549e-05, "loss": 0.0015, "step": 25230 }, { "epoch": 192.67175572519085, "grad_norm": 0.0816025659441948, "learning_rate": 1.886301483628319e-05, "loss": 0.0015, "step": 25240 }, { "epoch": 192.74809160305344, "grad_norm": 0.05815637484192848, "learning_rate": 1.8785537320205806e-05, "loss": 0.0014, "step": 25250 }, { "epoch": 192.82442748091603, "grad_norm": 0.06899966299533844, "learning_rate": 1.870820861593147e-05, "loss": 0.0016, "step": 25260 }, { "epoch": 192.90076335877862, "grad_norm": 0.07410213351249695, "learning_rate": 1.8631028811159525e-05, "loss": 0.0014, "step": 25270 }, { "epoch": 192.9770992366412, "grad_norm": 0.06716839224100113, "learning_rate": 1.8553997993420495e-05, "loss": 0.0014, "step": 25280 }, { "epoch": 193.05343511450383, "grad_norm": 0.050671614706516266, "learning_rate": 1.8477116250075825e-05, "loss": 0.0014, "step": 25290 }, { "epoch": 193.12977099236642, "grad_norm": 0.07199980318546295, "learning_rate": 1.840038366831802e-05, "loss": 0.0015, "step": 25300 }, { "epoch": 193.206106870229, "grad_norm": 0.07542963325977325, "learning_rate": 1.832380033517037e-05, "loss": 0.0014, "step": 25310 }, { "epoch": 193.2824427480916, "grad_norm": 0.07237185537815094, "learning_rate": 1.8247366337486862e-05, "loss": 0.0013, "step": 25320 }, { "epoch": 193.3587786259542, "grad_norm": 0.06254756450653076, "learning_rate": 1.8171081761952195e-05, "loss": 0.0015, "step": 25330 }, { "epoch": 193.4351145038168, "grad_norm": 0.09891267865896225, "learning_rate": 1.8094946695081556e-05, "loss": 0.0014, "step": 25340 }, { "epoch": 193.5114503816794, "grad_norm": 0.060233667492866516, "learning_rate": 1.801896122322057e-05, "loss": 0.0014, "step": 25350 }, { "epoch": 193.58778625954199, "grad_norm": 0.05953383818268776, "learning_rate": 1.794312543254524e-05, "loss": 0.0016, "step": 25360 }, { "epoch": 193.66412213740458, "grad_norm": 0.04748088866472244, "learning_rate": 1.7867439409061763e-05, "loss": 0.0015, "step": 25370 }, { "epoch": 193.74045801526717, "grad_norm": 0.05607225000858307, "learning_rate": 1.779190323860657e-05, "loss": 0.0013, "step": 25380 }, { "epoch": 193.81679389312978, "grad_norm": 0.05008069425821304, "learning_rate": 1.771651700684602e-05, "loss": 0.0015, "step": 25390 }, { "epoch": 193.89312977099237, "grad_norm": 0.07543632388114929, "learning_rate": 1.7641280799276507e-05, "loss": 0.0014, "step": 25400 }, { "epoch": 193.96946564885496, "grad_norm": 0.09924454241991043, "learning_rate": 1.7566194701224264e-05, "loss": 0.0015, "step": 25410 }, { "epoch": 194.04580152671755, "grad_norm": 0.04703288897871971, "learning_rate": 1.749125879784528e-05, "loss": 0.0012, "step": 25420 }, { "epoch": 194.12213740458014, "grad_norm": 0.057214852422475815, "learning_rate": 1.7416473174125235e-05, "loss": 0.0013, "step": 25430 }, { "epoch": 194.19847328244273, "grad_norm": 0.07140440493822098, "learning_rate": 1.734183791487936e-05, "loss": 0.0016, "step": 25440 }, { "epoch": 194.27480916030535, "grad_norm": 0.05039164796471596, "learning_rate": 1.7267353104752265e-05, "loss": 0.0016, "step": 25450 }, { "epoch": 194.35114503816794, "grad_norm": 0.06975285708904266, "learning_rate": 1.7193018828218074e-05, "loss": 0.0015, "step": 25460 }, { "epoch": 194.42748091603053, "grad_norm": 0.055973030626773834, "learning_rate": 1.7118835169580125e-05, "loss": 0.0014, "step": 25470 }, { "epoch": 194.50381679389312, "grad_norm": 0.05024188011884689, "learning_rate": 1.7044802212970933e-05, "loss": 0.0016, "step": 25480 }, { "epoch": 194.5801526717557, "grad_norm": 0.032858505845069885, "learning_rate": 1.697092004235213e-05, "loss": 0.0016, "step": 25490 }, { "epoch": 194.65648854961833, "grad_norm": 0.06839431822299957, "learning_rate": 1.6897188741514284e-05, "loss": 0.0015, "step": 25500 }, { "epoch": 194.73282442748092, "grad_norm": 0.05322925001382828, "learning_rate": 1.6823608394076916e-05, "loss": 0.0013, "step": 25510 }, { "epoch": 194.8091603053435, "grad_norm": 0.0730646625161171, "learning_rate": 1.6750179083488324e-05, "loss": 0.0014, "step": 25520 }, { "epoch": 194.8854961832061, "grad_norm": 0.08897680044174194, "learning_rate": 1.6676900893025514e-05, "loss": 0.0014, "step": 25530 }, { "epoch": 194.9618320610687, "grad_norm": 0.06531178206205368, "learning_rate": 1.660377390579412e-05, "loss": 0.0014, "step": 25540 }, { "epoch": 195.0381679389313, "grad_norm": 0.05939594283699989, "learning_rate": 1.6530798204728276e-05, "loss": 0.0013, "step": 25550 }, { "epoch": 195.1145038167939, "grad_norm": 0.05738574266433716, "learning_rate": 1.645797387259055e-05, "loss": 0.0017, "step": 25560 }, { "epoch": 195.1908396946565, "grad_norm": 0.07366278022527695, "learning_rate": 1.638530099197184e-05, "loss": 0.0013, "step": 25570 }, { "epoch": 195.26717557251908, "grad_norm": 0.05478206276893616, "learning_rate": 1.6312779645291292e-05, "loss": 0.0015, "step": 25580 }, { "epoch": 195.34351145038167, "grad_norm": 0.07147829234600067, "learning_rate": 1.6240409914796182e-05, "loss": 0.0014, "step": 25590 }, { "epoch": 195.4198473282443, "grad_norm": 0.06170695647597313, "learning_rate": 1.616819188256181e-05, "loss": 0.0014, "step": 25600 }, { "epoch": 195.49618320610688, "grad_norm": 0.07565584033727646, "learning_rate": 1.6096125630491473e-05, "loss": 0.0014, "step": 25610 }, { "epoch": 195.57251908396947, "grad_norm": 0.05489473417401314, "learning_rate": 1.602421124031633e-05, "loss": 0.0014, "step": 25620 }, { "epoch": 195.64885496183206, "grad_norm": 0.06307155638933182, "learning_rate": 1.5952448793595307e-05, "loss": 0.0015, "step": 25630 }, { "epoch": 195.72519083969465, "grad_norm": 0.07451443374156952, "learning_rate": 1.5880838371714984e-05, "loss": 0.0014, "step": 25640 }, { "epoch": 195.80152671755727, "grad_norm": 0.08102034777402878, "learning_rate": 1.5809380055889592e-05, "loss": 0.0015, "step": 25650 }, { "epoch": 195.87786259541986, "grad_norm": 0.05768885463476181, "learning_rate": 1.573807392716076e-05, "loss": 0.0015, "step": 25660 }, { "epoch": 195.95419847328245, "grad_norm": 0.09588508307933807, "learning_rate": 1.5666920066397597e-05, "loss": 0.0015, "step": 25670 }, { "epoch": 196.03053435114504, "grad_norm": 0.036776695400476456, "learning_rate": 1.5595918554296504e-05, "loss": 0.0015, "step": 25680 }, { "epoch": 196.10687022900763, "grad_norm": 0.06880450248718262, "learning_rate": 1.5525069471381086e-05, "loss": 0.0013, "step": 25690 }, { "epoch": 196.18320610687022, "grad_norm": 0.06806475669145584, "learning_rate": 1.5454372898002083e-05, "loss": 0.0014, "step": 25700 }, { "epoch": 196.25954198473283, "grad_norm": 0.05585639178752899, "learning_rate": 1.5383828914337325e-05, "loss": 0.0013, "step": 25710 }, { "epoch": 196.33587786259542, "grad_norm": 0.04133257269859314, "learning_rate": 1.5313437600391487e-05, "loss": 0.0014, "step": 25720 }, { "epoch": 196.41221374045801, "grad_norm": 0.05767213925719261, "learning_rate": 1.5243199035996156e-05, "loss": 0.0013, "step": 25730 }, { "epoch": 196.4885496183206, "grad_norm": 0.0726919174194336, "learning_rate": 1.5173113300809697e-05, "loss": 0.0014, "step": 25740 }, { "epoch": 196.5648854961832, "grad_norm": 0.06086377054452896, "learning_rate": 1.5103180474317129e-05, "loss": 0.0013, "step": 25750 }, { "epoch": 196.6412213740458, "grad_norm": 0.07495767623186111, "learning_rate": 1.5033400635830057e-05, "loss": 0.0016, "step": 25760 }, { "epoch": 196.7175572519084, "grad_norm": 0.07106615602970123, "learning_rate": 1.4963773864486578e-05, "loss": 0.0013, "step": 25770 }, { "epoch": 196.793893129771, "grad_norm": 0.06342902034521103, "learning_rate": 1.4894300239251228e-05, "loss": 0.0013, "step": 25780 }, { "epoch": 196.87022900763358, "grad_norm": 0.050683245062828064, "learning_rate": 1.4824979838914796e-05, "loss": 0.0013, "step": 25790 }, { "epoch": 196.94656488549617, "grad_norm": 0.05290408805012703, "learning_rate": 1.4755812742094347e-05, "loss": 0.0016, "step": 25800 }, { "epoch": 197.0229007633588, "grad_norm": 0.05192754417657852, "learning_rate": 1.4686799027233082e-05, "loss": 0.0015, "step": 25810 }, { "epoch": 197.09923664122138, "grad_norm": 0.10283289104700089, "learning_rate": 1.4617938772600207e-05, "loss": 0.0016, "step": 25820 }, { "epoch": 197.17557251908397, "grad_norm": 0.05121411010622978, "learning_rate": 1.4549232056290905e-05, "loss": 0.0013, "step": 25830 }, { "epoch": 197.25190839694656, "grad_norm": 0.06308673322200775, "learning_rate": 1.448067895622626e-05, "loss": 0.0014, "step": 25840 }, { "epoch": 197.32824427480915, "grad_norm": 0.05108404532074928, "learning_rate": 1.4412279550153117e-05, "loss": 0.0012, "step": 25850 }, { "epoch": 197.40458015267177, "grad_norm": 0.07968974113464355, "learning_rate": 1.4344033915644021e-05, "loss": 0.0015, "step": 25860 }, { "epoch": 197.48091603053436, "grad_norm": 0.07406777143478394, "learning_rate": 1.4275942130097096e-05, "loss": 0.0013, "step": 25870 }, { "epoch": 197.55725190839695, "grad_norm": 0.09649431705474854, "learning_rate": 1.4208004270736e-05, "loss": 0.0017, "step": 25880 }, { "epoch": 197.63358778625954, "grad_norm": 0.11769559234380722, "learning_rate": 1.4140220414609838e-05, "loss": 0.0017, "step": 25890 }, { "epoch": 197.70992366412213, "grad_norm": 0.0932316854596138, "learning_rate": 1.4072590638593046e-05, "loss": 0.0015, "step": 25900 }, { "epoch": 197.78625954198472, "grad_norm": 0.059026967734098434, "learning_rate": 1.4005115019385322e-05, "loss": 0.0015, "step": 25910 }, { "epoch": 197.86259541984734, "grad_norm": 0.090200275182724, "learning_rate": 1.3937793633511552e-05, "loss": 0.0014, "step": 25920 }, { "epoch": 197.93893129770993, "grad_norm": 0.06659440696239471, "learning_rate": 1.3870626557321613e-05, "loss": 0.0014, "step": 25930 }, { "epoch": 198.01526717557252, "grad_norm": 0.07234080880880356, "learning_rate": 1.3803613866990487e-05, "loss": 0.0013, "step": 25940 }, { "epoch": 198.0916030534351, "grad_norm": 0.06536644697189331, "learning_rate": 1.3736755638517998e-05, "loss": 0.0014, "step": 25950 }, { "epoch": 198.1679389312977, "grad_norm": 0.04618670791387558, "learning_rate": 1.367005194772886e-05, "loss": 0.0012, "step": 25960 }, { "epoch": 198.24427480916032, "grad_norm": 0.047835927456617355, "learning_rate": 1.3603502870272475e-05, "loss": 0.0013, "step": 25970 }, { "epoch": 198.3206106870229, "grad_norm": 0.06228303909301758, "learning_rate": 1.3537108481622871e-05, "loss": 0.0015, "step": 25980 }, { "epoch": 198.3969465648855, "grad_norm": 0.05015227571129799, "learning_rate": 1.3470868857078681e-05, "loss": 0.0014, "step": 25990 }, { "epoch": 198.4732824427481, "grad_norm": 0.07118888944387436, "learning_rate": 1.3404784071763014e-05, "loss": 0.0014, "step": 26000 }, { "epoch": 198.54961832061068, "grad_norm": 0.04789242520928383, "learning_rate": 1.3338854200623383e-05, "loss": 0.0015, "step": 26010 }, { "epoch": 198.6259541984733, "grad_norm": 0.06826778501272202, "learning_rate": 1.3273079318431612e-05, "loss": 0.0015, "step": 26020 }, { "epoch": 198.70229007633588, "grad_norm": 0.09107141941785812, "learning_rate": 1.3207459499783696e-05, "loss": 0.0017, "step": 26030 }, { "epoch": 198.77862595419847, "grad_norm": 0.04822579026222229, "learning_rate": 1.314199481909982e-05, "loss": 0.0014, "step": 26040 }, { "epoch": 198.85496183206106, "grad_norm": 0.1108035296201706, "learning_rate": 1.3076685350624234e-05, "loss": 0.0015, "step": 26050 }, { "epoch": 198.93129770992365, "grad_norm": 0.06311202794313431, "learning_rate": 1.3011531168425148e-05, "loss": 0.0013, "step": 26060 }, { "epoch": 199.00763358778627, "grad_norm": 0.054660454392433167, "learning_rate": 1.2946532346394628e-05, "loss": 0.0012, "step": 26070 }, { "epoch": 199.08396946564886, "grad_norm": 0.07628846168518066, "learning_rate": 1.2881688958248637e-05, "loss": 0.0013, "step": 26080 }, { "epoch": 199.16030534351145, "grad_norm": 0.06221949681639671, "learning_rate": 1.28170010775267e-05, "loss": 0.0013, "step": 26090 }, { "epoch": 199.23664122137404, "grad_norm": 0.17054525017738342, "learning_rate": 1.2752468777592134e-05, "loss": 0.0013, "step": 26100 }, { "epoch": 199.31297709923663, "grad_norm": 0.08260475844144821, "learning_rate": 1.2688092131631728e-05, "loss": 0.0013, "step": 26110 }, { "epoch": 199.38931297709922, "grad_norm": 0.0830073133111, "learning_rate": 1.2623871212655762e-05, "loss": 0.0016, "step": 26120 }, { "epoch": 199.46564885496184, "grad_norm": 0.03772320970892906, "learning_rate": 1.2559806093497942e-05, "loss": 0.0011, "step": 26130 }, { "epoch": 199.54198473282443, "grad_norm": 0.03715347871184349, "learning_rate": 1.2495896846815184e-05, "loss": 0.0014, "step": 26140 }, { "epoch": 199.61832061068702, "grad_norm": 0.05999510735273361, "learning_rate": 1.2432143545087708e-05, "loss": 0.0013, "step": 26150 }, { "epoch": 199.6946564885496, "grad_norm": 0.047915324568748474, "learning_rate": 1.2368546260618844e-05, "loss": 0.0013, "step": 26160 }, { "epoch": 199.7709923664122, "grad_norm": 0.0654478371143341, "learning_rate": 1.2305105065535014e-05, "loss": 0.0015, "step": 26170 }, { "epoch": 199.84732824427482, "grad_norm": 0.03670601546764374, "learning_rate": 1.2241820031785631e-05, "loss": 0.0013, "step": 26180 }, { "epoch": 199.9236641221374, "grad_norm": 0.06793992966413498, "learning_rate": 1.2178691231142883e-05, "loss": 0.0014, "step": 26190 }, { "epoch": 200.0, "grad_norm": 0.06024183705449104, "learning_rate": 1.2115718735201885e-05, "loss": 0.0014, "step": 26200 }, { "epoch": 200.0763358778626, "grad_norm": 0.057208430022001266, "learning_rate": 1.2052902615380471e-05, "loss": 0.0013, "step": 26210 }, { "epoch": 200.15267175572518, "grad_norm": 0.06247391551733017, "learning_rate": 1.1990242942919094e-05, "loss": 0.0013, "step": 26220 }, { "epoch": 200.2290076335878, "grad_norm": 0.05275759473443031, "learning_rate": 1.1927739788880803e-05, "loss": 0.0011, "step": 26230 }, { "epoch": 200.3053435114504, "grad_norm": 0.04802360385656357, "learning_rate": 1.1865393224151165e-05, "loss": 0.0014, "step": 26240 }, { "epoch": 200.38167938931298, "grad_norm": 0.0634264275431633, "learning_rate": 1.1803203319438053e-05, "loss": 0.0014, "step": 26250 }, { "epoch": 200.45801526717557, "grad_norm": 0.03752947226166725, "learning_rate": 1.1741170145271783e-05, "loss": 0.0012, "step": 26260 }, { "epoch": 200.53435114503816, "grad_norm": 0.06397858262062073, "learning_rate": 1.1679293772004877e-05, "loss": 0.0012, "step": 26270 }, { "epoch": 200.61068702290078, "grad_norm": 0.0492403544485569, "learning_rate": 1.161757426981202e-05, "loss": 0.0013, "step": 26280 }, { "epoch": 200.68702290076337, "grad_norm": 0.06053004413843155, "learning_rate": 1.1556011708690016e-05, "loss": 0.0015, "step": 26290 }, { "epoch": 200.76335877862596, "grad_norm": 0.034203507006168365, "learning_rate": 1.149460615845762e-05, "loss": 0.0013, "step": 26300 }, { "epoch": 200.83969465648855, "grad_norm": 0.0549517385661602, "learning_rate": 1.1433357688755573e-05, "loss": 0.0013, "step": 26310 }, { "epoch": 200.91603053435114, "grad_norm": 0.08094070851802826, "learning_rate": 1.1372266369046478e-05, "loss": 0.0012, "step": 26320 }, { "epoch": 200.99236641221373, "grad_norm": 0.06414365768432617, "learning_rate": 1.1311332268614665e-05, "loss": 0.0014, "step": 26330 }, { "epoch": 201.06870229007635, "grad_norm": 0.038974303752183914, "learning_rate": 1.1250555456566217e-05, "loss": 0.0013, "step": 26340 }, { "epoch": 201.14503816793894, "grad_norm": 0.05227290093898773, "learning_rate": 1.1189936001828786e-05, "loss": 0.0013, "step": 26350 }, { "epoch": 201.22137404580153, "grad_norm": 0.07375174760818481, "learning_rate": 1.1129473973151559e-05, "loss": 0.0014, "step": 26360 }, { "epoch": 201.29770992366412, "grad_norm": 0.07465815544128418, "learning_rate": 1.1069169439105203e-05, "loss": 0.0015, "step": 26370 }, { "epoch": 201.3740458015267, "grad_norm": 0.05255136266350746, "learning_rate": 1.1009022468081807e-05, "loss": 0.0012, "step": 26380 }, { "epoch": 201.45038167938932, "grad_norm": 0.07022616267204285, "learning_rate": 1.0949033128294742e-05, "loss": 0.0016, "step": 26390 }, { "epoch": 201.5267175572519, "grad_norm": 0.07827412337064743, "learning_rate": 1.0889201487778537e-05, "loss": 0.0013, "step": 26400 }, { "epoch": 201.6030534351145, "grad_norm": 0.06714841723442078, "learning_rate": 1.082952761438896e-05, "loss": 0.0013, "step": 26410 }, { "epoch": 201.6793893129771, "grad_norm": 0.062312230467796326, "learning_rate": 1.0770011575802828e-05, "loss": 0.0013, "step": 26420 }, { "epoch": 201.75572519083968, "grad_norm": 0.05162837728857994, "learning_rate": 1.0710653439517946e-05, "loss": 0.0012, "step": 26430 }, { "epoch": 201.8320610687023, "grad_norm": 0.06423509865999222, "learning_rate": 1.0651453272853034e-05, "loss": 0.0013, "step": 26440 }, { "epoch": 201.9083969465649, "grad_norm": 0.0746331512928009, "learning_rate": 1.0592411142947732e-05, "loss": 0.0013, "step": 26450 }, { "epoch": 201.98473282442748, "grad_norm": 0.16105543076992035, "learning_rate": 1.0533527116762296e-05, "loss": 0.0013, "step": 26460 }, { "epoch": 202.06106870229007, "grad_norm": 0.06791657954454422, "learning_rate": 1.0474801261077809e-05, "loss": 0.0013, "step": 26470 }, { "epoch": 202.13740458015266, "grad_norm": 0.07498209178447723, "learning_rate": 1.0416233642495941e-05, "loss": 0.0013, "step": 26480 }, { "epoch": 202.21374045801528, "grad_norm": 0.08263704180717468, "learning_rate": 1.0357824327438846e-05, "loss": 0.0013, "step": 26490 }, { "epoch": 202.29007633587787, "grad_norm": 0.11453201621770859, "learning_rate": 1.0299573382149234e-05, "loss": 0.0013, "step": 26500 }, { "epoch": 202.36641221374046, "grad_norm": 0.043959155678749084, "learning_rate": 1.0241480872690145e-05, "loss": 0.0011, "step": 26510 }, { "epoch": 202.44274809160305, "grad_norm": 0.11462806910276413, "learning_rate": 1.0183546864944942e-05, "loss": 0.0013, "step": 26520 }, { "epoch": 202.51908396946564, "grad_norm": 0.07191414386034012, "learning_rate": 1.0125771424617218e-05, "loss": 0.0014, "step": 26530 }, { "epoch": 202.59541984732823, "grad_norm": 0.07046619057655334, "learning_rate": 1.0068154617230778e-05, "loss": 0.0013, "step": 26540 }, { "epoch": 202.67175572519085, "grad_norm": 0.050990547984838486, "learning_rate": 1.0010696508129501e-05, "loss": 0.0013, "step": 26550 }, { "epoch": 202.74809160305344, "grad_norm": 0.05201224982738495, "learning_rate": 9.95339716247726e-06, "loss": 0.0013, "step": 26560 }, { "epoch": 202.82442748091603, "grad_norm": 0.10305797308683395, "learning_rate": 9.896256645257893e-06, "loss": 0.0014, "step": 26570 }, { "epoch": 202.90076335877862, "grad_norm": 0.07364854961633682, "learning_rate": 9.839275021275078e-06, "loss": 0.0014, "step": 26580 }, { "epoch": 202.9770992366412, "grad_norm": 0.07126438617706299, "learning_rate": 9.78245235515237e-06, "loss": 0.0015, "step": 26590 }, { "epoch": 203.05343511450383, "grad_norm": 0.06116129830479622, "learning_rate": 9.725788711332972e-06, "loss": 0.0014, "step": 26600 }, { "epoch": 203.12977099236642, "grad_norm": 0.1144738420844078, "learning_rate": 9.669284154079776e-06, "loss": 0.0014, "step": 26610 }, { "epoch": 203.206106870229, "grad_norm": 0.07525362074375153, "learning_rate": 9.612938747475207e-06, "loss": 0.0014, "step": 26620 }, { "epoch": 203.2824427480916, "grad_norm": 0.05583926662802696, "learning_rate": 9.556752555421249e-06, "loss": 0.0013, "step": 26630 }, { "epoch": 203.3587786259542, "grad_norm": 0.0491446852684021, "learning_rate": 9.500725641639294e-06, "loss": 0.0013, "step": 26640 }, { "epoch": 203.4351145038168, "grad_norm": 0.08112872391939163, "learning_rate": 9.444858069670086e-06, "loss": 0.0014, "step": 26650 }, { "epoch": 203.5114503816794, "grad_norm": 0.06544926762580872, "learning_rate": 9.389149902873688e-06, "loss": 0.0012, "step": 26660 }, { "epoch": 203.58778625954199, "grad_norm": 0.05709274858236313, "learning_rate": 9.333601204429348e-06, "loss": 0.0012, "step": 26670 }, { "epoch": 203.66412213740458, "grad_norm": 0.03269326686859131, "learning_rate": 9.278212037335469e-06, "loss": 0.0013, "step": 26680 }, { "epoch": 203.74045801526717, "grad_norm": 0.09314236044883728, "learning_rate": 9.222982464409522e-06, "loss": 0.0014, "step": 26690 }, { "epoch": 203.81679389312978, "grad_norm": 0.07231587171554565, "learning_rate": 9.16791254828802e-06, "loss": 0.0011, "step": 26700 }, { "epoch": 203.89312977099237, "grad_norm": 0.04576656594872475, "learning_rate": 9.113002351426351e-06, "loss": 0.0013, "step": 26710 }, { "epoch": 203.96946564885496, "grad_norm": 0.06238718703389168, "learning_rate": 9.058251936098821e-06, "loss": 0.0012, "step": 26720 }, { "epoch": 204.04580152671755, "grad_norm": 0.09597546607255936, "learning_rate": 9.00366136439844e-06, "loss": 0.0014, "step": 26730 }, { "epoch": 204.12213740458014, "grad_norm": 0.09496791660785675, "learning_rate": 8.949230698237037e-06, "loss": 0.0012, "step": 26740 }, { "epoch": 204.19847328244273, "grad_norm": 0.08133654296398163, "learning_rate": 8.894959999345014e-06, "loss": 0.0013, "step": 26750 }, { "epoch": 204.27480916030535, "grad_norm": 0.1431337147951126, "learning_rate": 8.840849329271404e-06, "loss": 0.0013, "step": 26760 }, { "epoch": 204.35114503816794, "grad_norm": 0.07239469885826111, "learning_rate": 8.78689874938372e-06, "loss": 0.0012, "step": 26770 }, { "epoch": 204.42748091603053, "grad_norm": 0.06027946248650551, "learning_rate": 8.733108320867932e-06, "loss": 0.0012, "step": 26780 }, { "epoch": 204.50381679389312, "grad_norm": 0.14697539806365967, "learning_rate": 8.67947810472836e-06, "loss": 0.0014, "step": 26790 }, { "epoch": 204.5801526717557, "grad_norm": 0.07038645446300507, "learning_rate": 8.62600816178765e-06, "loss": 0.0012, "step": 26800 }, { "epoch": 204.65648854961833, "grad_norm": 0.07756838947534561, "learning_rate": 8.572698552686663e-06, "loss": 0.0012, "step": 26810 }, { "epoch": 204.73282442748092, "grad_norm": 0.05653005838394165, "learning_rate": 8.519549337884435e-06, "loss": 0.0013, "step": 26820 }, { "epoch": 204.8091603053435, "grad_norm": 0.07117787003517151, "learning_rate": 8.466560577658083e-06, "loss": 0.0015, "step": 26830 }, { "epoch": 204.8854961832061, "grad_norm": 0.06036967784166336, "learning_rate": 8.413732332102734e-06, "loss": 0.0012, "step": 26840 }, { "epoch": 204.9618320610687, "grad_norm": 0.1452104151248932, "learning_rate": 8.361064661131533e-06, "loss": 0.0014, "step": 26850 }, { "epoch": 205.0381679389313, "grad_norm": 0.09159868210554123, "learning_rate": 8.308557624475465e-06, "loss": 0.0013, "step": 26860 }, { "epoch": 205.1145038167939, "grad_norm": 0.05109237879514694, "learning_rate": 8.256211281683362e-06, "loss": 0.0013, "step": 26870 }, { "epoch": 205.1908396946565, "grad_norm": 0.06664712727069855, "learning_rate": 8.204025692121802e-06, "loss": 0.0012, "step": 26880 }, { "epoch": 205.26717557251908, "grad_norm": 0.056709252297878265, "learning_rate": 8.152000914975043e-06, "loss": 0.0014, "step": 26890 }, { "epoch": 205.34351145038167, "grad_norm": 0.10278716683387756, "learning_rate": 8.100137009244956e-06, "loss": 0.0012, "step": 26900 }, { "epoch": 205.4198473282443, "grad_norm": 0.08502020686864853, "learning_rate": 8.048434033751006e-06, "loss": 0.0013, "step": 26910 }, { "epoch": 205.49618320610688, "grad_norm": 0.0658947005867958, "learning_rate": 7.99689204713012e-06, "loss": 0.0011, "step": 26920 }, { "epoch": 205.57251908396947, "grad_norm": 0.0905240848660469, "learning_rate": 7.945511107836661e-06, "loss": 0.0013, "step": 26930 }, { "epoch": 205.64885496183206, "grad_norm": 0.04238907992839813, "learning_rate": 7.894291274142295e-06, "loss": 0.0011, "step": 26940 }, { "epoch": 205.72519083969465, "grad_norm": 0.10271405428647995, "learning_rate": 7.843232604136023e-06, "loss": 0.0014, "step": 26950 }, { "epoch": 205.80152671755727, "grad_norm": 0.09067624807357788, "learning_rate": 7.79233515572406e-06, "loss": 0.0013, "step": 26960 }, { "epoch": 205.87786259541986, "grad_norm": 0.07940857857465744, "learning_rate": 7.7415989866298e-06, "loss": 0.0015, "step": 26970 }, { "epoch": 205.95419847328245, "grad_norm": 0.08306518942117691, "learning_rate": 7.691024154393676e-06, "loss": 0.0014, "step": 26980 }, { "epoch": 206.03053435114504, "grad_norm": 0.052101049572229385, "learning_rate": 7.640610716373197e-06, "loss": 0.0015, "step": 26990 }, { "epoch": 206.10687022900763, "grad_norm": 0.06758704781532288, "learning_rate": 7.590358729742807e-06, "loss": 0.0013, "step": 27000 }, { "epoch": 206.18320610687022, "grad_norm": 0.05620907247066498, "learning_rate": 7.540268251493836e-06, "loss": 0.0012, "step": 27010 }, { "epoch": 206.25954198473283, "grad_norm": 0.05611012876033783, "learning_rate": 7.490339338434476e-06, "loss": 0.0013, "step": 27020 }, { "epoch": 206.33587786259542, "grad_norm": 0.07398800551891327, "learning_rate": 7.44057204718968e-06, "loss": 0.0012, "step": 27030 }, { "epoch": 206.41221374045801, "grad_norm": 0.04951023682951927, "learning_rate": 7.390966434201084e-06, "loss": 0.0012, "step": 27040 }, { "epoch": 206.4885496183206, "grad_norm": 0.07077933102846146, "learning_rate": 7.34152255572697e-06, "loss": 0.0014, "step": 27050 }, { "epoch": 206.5648854961832, "grad_norm": 0.0559135340154171, "learning_rate": 7.292240467842198e-06, "loss": 0.0012, "step": 27060 }, { "epoch": 206.6412213740458, "grad_norm": 0.036825258284807205, "learning_rate": 7.243120226438126e-06, "loss": 0.0013, "step": 27070 }, { "epoch": 206.7175572519084, "grad_norm": 0.07349317520856857, "learning_rate": 7.1941618872226104e-06, "loss": 0.0012, "step": 27080 }, { "epoch": 206.793893129771, "grad_norm": 0.07397443056106567, "learning_rate": 7.145365505719836e-06, "loss": 0.0014, "step": 27090 }, { "epoch": 206.87022900763358, "grad_norm": 0.06549029052257538, "learning_rate": 7.096731137270317e-06, "loss": 0.0013, "step": 27100 }, { "epoch": 206.94656488549617, "grad_norm": 0.03941896930336952, "learning_rate": 7.048258837030868e-06, "loss": 0.0011, "step": 27110 }, { "epoch": 207.0229007633588, "grad_norm": 0.06150936707854271, "learning_rate": 6.999948659974447e-06, "loss": 0.0013, "step": 27120 }, { "epoch": 207.09923664122138, "grad_norm": 0.06528383493423462, "learning_rate": 6.951800660890178e-06, "loss": 0.0014, "step": 27130 }, { "epoch": 207.17557251908397, "grad_norm": 0.05457814782857895, "learning_rate": 6.903814894383248e-06, "loss": 0.0014, "step": 27140 }, { "epoch": 207.25190839694656, "grad_norm": 0.05375095084309578, "learning_rate": 6.85599141487489e-06, "loss": 0.0014, "step": 27150 }, { "epoch": 207.32824427480915, "grad_norm": 0.07261896133422852, "learning_rate": 6.808330276602203e-06, "loss": 0.0012, "step": 27160 }, { "epoch": 207.40458015267177, "grad_norm": 0.04184341803193092, "learning_rate": 6.760831533618233e-06, "loss": 0.0011, "step": 27170 }, { "epoch": 207.48091603053436, "grad_norm": 0.0689283162355423, "learning_rate": 6.713495239791855e-06, "loss": 0.0014, "step": 27180 }, { "epoch": 207.55725190839695, "grad_norm": 0.06138835474848747, "learning_rate": 6.666321448807693e-06, "loss": 0.0014, "step": 27190 }, { "epoch": 207.63358778625954, "grad_norm": 0.04956292733550072, "learning_rate": 6.619310214166052e-06, "loss": 0.0012, "step": 27200 }, { "epoch": 207.70992366412213, "grad_norm": 0.07732786983251572, "learning_rate": 6.572461589182948e-06, "loss": 0.0012, "step": 27210 }, { "epoch": 207.78625954198472, "grad_norm": 0.04907825216650963, "learning_rate": 6.525775626989898e-06, "loss": 0.0013, "step": 27220 }, { "epoch": 207.86259541984734, "grad_norm": 0.07578302174806595, "learning_rate": 6.4792523805339836e-06, "loss": 0.0012, "step": 27230 }, { "epoch": 207.93893129770993, "grad_norm": 0.046786025166511536, "learning_rate": 6.432891902577764e-06, "loss": 0.0012, "step": 27240 }, { "epoch": 208.01526717557252, "grad_norm": 0.057731471955776215, "learning_rate": 6.386694245699181e-06, "loss": 0.0012, "step": 27250 }, { "epoch": 208.0916030534351, "grad_norm": 0.05093731731176376, "learning_rate": 6.340659462291492e-06, "loss": 0.0014, "step": 27260 }, { "epoch": 208.1679389312977, "grad_norm": 0.04305582493543625, "learning_rate": 6.294787604563267e-06, "loss": 0.0012, "step": 27270 }, { "epoch": 208.24427480916032, "grad_norm": 0.054172057658433914, "learning_rate": 6.249078724538325e-06, "loss": 0.0012, "step": 27280 }, { "epoch": 208.3206106870229, "grad_norm": 0.05862890183925629, "learning_rate": 6.2035328740556e-06, "loss": 0.0012, "step": 27290 }, { "epoch": 208.3969465648855, "grad_norm": 0.0827888771891594, "learning_rate": 6.158150104769155e-06, "loss": 0.0013, "step": 27300 }, { "epoch": 208.4732824427481, "grad_norm": 0.10330336540937424, "learning_rate": 6.112930468148119e-06, "loss": 0.0012, "step": 27310 }, { "epoch": 208.54961832061068, "grad_norm": 0.041781824082136154, "learning_rate": 6.067874015476571e-06, "loss": 0.0012, "step": 27320 }, { "epoch": 208.6259541984733, "grad_norm": 0.05017821490764618, "learning_rate": 6.0229807978535495e-06, "loss": 0.0012, "step": 27330 }, { "epoch": 208.70229007633588, "grad_norm": 0.04558820649981499, "learning_rate": 5.978250866192963e-06, "loss": 0.0012, "step": 27340 }, { "epoch": 208.77862595419847, "grad_norm": 0.038026366382837296, "learning_rate": 5.933684271223532e-06, "loss": 0.0012, "step": 27350 }, { "epoch": 208.85496183206106, "grad_norm": 0.06364163011312485, "learning_rate": 5.889281063488743e-06, "loss": 0.0013, "step": 27360 }, { "epoch": 208.93129770992365, "grad_norm": 0.04445384070277214, "learning_rate": 5.845041293346747e-06, "loss": 0.0013, "step": 27370 }, { "epoch": 209.00763358778627, "grad_norm": 0.06833741813898087, "learning_rate": 5.800965010970393e-06, "loss": 0.0013, "step": 27380 }, { "epoch": 209.08396946564886, "grad_norm": 0.05848439782857895, "learning_rate": 5.757052266347078e-06, "loss": 0.0015, "step": 27390 }, { "epoch": 209.16030534351145, "grad_norm": 0.08115576207637787, "learning_rate": 5.713303109278749e-06, "loss": 0.0013, "step": 27400 }, { "epoch": 209.23664122137404, "grad_norm": 0.05861181765794754, "learning_rate": 5.669717589381817e-06, "loss": 0.0013, "step": 27410 }, { "epoch": 209.31297709923663, "grad_norm": 0.11657141894102097, "learning_rate": 5.626295756087107e-06, "loss": 0.0012, "step": 27420 }, { "epoch": 209.38931297709922, "grad_norm": 0.06871122121810913, "learning_rate": 5.58303765863981e-06, "loss": 0.0013, "step": 27430 }, { "epoch": 209.46564885496184, "grad_norm": 0.05315982922911644, "learning_rate": 5.539943346099418e-06, "loss": 0.0012, "step": 27440 }, { "epoch": 209.54198473282443, "grad_norm": 0.04639442637562752, "learning_rate": 5.497012867339701e-06, "loss": 0.0012, "step": 27450 }, { "epoch": 209.61832061068702, "grad_norm": 0.07584572583436966, "learning_rate": 5.454246271048596e-06, "loss": 0.0013, "step": 27460 }, { "epoch": 209.6946564885496, "grad_norm": 0.051477257162332535, "learning_rate": 5.411643605728139e-06, "loss": 0.0013, "step": 27470 }, { "epoch": 209.7709923664122, "grad_norm": 0.04693716764450073, "learning_rate": 5.3692049196945475e-06, "loss": 0.0011, "step": 27480 }, { "epoch": 209.84732824427482, "grad_norm": 0.06328834593296051, "learning_rate": 5.326930261077972e-06, "loss": 0.0013, "step": 27490 }, { "epoch": 209.9236641221374, "grad_norm": 0.0690498873591423, "learning_rate": 5.28481967782261e-06, "loss": 0.0011, "step": 27500 }, { "epoch": 210.0, "grad_norm": 0.03060213476419449, "learning_rate": 5.242873217686527e-06, "loss": 0.0012, "step": 27510 }, { "epoch": 210.0763358778626, "grad_norm": 0.05056371912360191, "learning_rate": 5.201090928241719e-06, "loss": 0.0013, "step": 27520 }, { "epoch": 210.15267175572518, "grad_norm": 0.0742863193154335, "learning_rate": 5.1594728568738976e-06, "loss": 0.0013, "step": 27530 }, { "epoch": 210.2290076335878, "grad_norm": 0.08828605711460114, "learning_rate": 5.118019050782624e-06, "loss": 0.0013, "step": 27540 }, { "epoch": 210.3053435114504, "grad_norm": 0.06699583679437637, "learning_rate": 5.07672955698109e-06, "loss": 0.0013, "step": 27550 }, { "epoch": 210.38167938931298, "grad_norm": 0.03990701958537102, "learning_rate": 5.035604422296224e-06, "loss": 0.0011, "step": 27560 }, { "epoch": 210.45801526717557, "grad_norm": 0.05989127233624458, "learning_rate": 4.9946436933684665e-06, "loss": 0.0012, "step": 27570 }, { "epoch": 210.53435114503816, "grad_norm": 0.04652866721153259, "learning_rate": 4.95384741665189e-06, "loss": 0.0012, "step": 27580 }, { "epoch": 210.61068702290078, "grad_norm": 0.11713515967130661, "learning_rate": 4.9132156384139535e-06, "loss": 0.0012, "step": 27590 }, { "epoch": 210.68702290076337, "grad_norm": 0.06640242040157318, "learning_rate": 4.872748404735644e-06, "loss": 0.0012, "step": 27600 }, { "epoch": 210.76335877862596, "grad_norm": 0.0418890118598938, "learning_rate": 4.8324457615113165e-06, "loss": 0.0013, "step": 27610 }, { "epoch": 210.83969465648855, "grad_norm": 0.04723799601197243, "learning_rate": 4.7923077544486266e-06, "loss": 0.0012, "step": 27620 }, { "epoch": 210.91603053435114, "grad_norm": 0.05541541799902916, "learning_rate": 4.752334429068577e-06, "loss": 0.0011, "step": 27630 }, { "epoch": 210.99236641221373, "grad_norm": 0.1043303981423378, "learning_rate": 4.712525830705338e-06, "loss": 0.0012, "step": 27640 }, { "epoch": 211.06870229007635, "grad_norm": 0.06761635094881058, "learning_rate": 4.6728820045062954e-06, "loss": 0.0012, "step": 27650 }, { "epoch": 211.14503816793894, "grad_norm": 0.08528868108987808, "learning_rate": 4.633402995431968e-06, "loss": 0.0013, "step": 27660 }, { "epoch": 211.22137404580153, "grad_norm": 0.09717858582735062, "learning_rate": 4.5940888482559555e-06, "loss": 0.0013, "step": 27670 }, { "epoch": 211.29770992366412, "grad_norm": 0.05620357766747475, "learning_rate": 4.554939607564861e-06, "loss": 0.0012, "step": 27680 }, { "epoch": 211.3740458015267, "grad_norm": 0.06806053221225739, "learning_rate": 4.515955317758285e-06, "loss": 0.001, "step": 27690 }, { "epoch": 211.45038167938932, "grad_norm": 0.04358511045575142, "learning_rate": 4.477136023048727e-06, "loss": 0.0013, "step": 27700 }, { "epoch": 211.5267175572519, "grad_norm": 0.05117904022336006, "learning_rate": 4.4384817674616215e-06, "loss": 0.0014, "step": 27710 }, { "epoch": 211.6030534351145, "grad_norm": 0.038691144436597824, "learning_rate": 4.399992594835183e-06, "loss": 0.001, "step": 27720 }, { "epoch": 211.6793893129771, "grad_norm": 0.07051052898168564, "learning_rate": 4.361668548820429e-06, "loss": 0.0012, "step": 27730 }, { "epoch": 211.75572519083968, "grad_norm": 0.0854918584227562, "learning_rate": 4.323509672881059e-06, "loss": 0.0013, "step": 27740 }, { "epoch": 211.8320610687023, "grad_norm": 0.04021378234028816, "learning_rate": 4.285516010293522e-06, "loss": 0.0012, "step": 27750 }, { "epoch": 211.9083969465649, "grad_norm": 0.047155678272247314, "learning_rate": 4.247687604146816e-06, "loss": 0.0014, "step": 27760 }, { "epoch": 211.98473282442748, "grad_norm": 0.06576815992593765, "learning_rate": 4.210024497342607e-06, "loss": 0.0012, "step": 27770 }, { "epoch": 212.06106870229007, "grad_norm": 0.04660751298069954, "learning_rate": 4.172526732595044e-06, "loss": 0.0012, "step": 27780 }, { "epoch": 212.13740458015266, "grad_norm": 0.03399916738271713, "learning_rate": 4.13519435243076e-06, "loss": 0.001, "step": 27790 }, { "epoch": 212.21374045801528, "grad_norm": 0.04355301707983017, "learning_rate": 4.098027399188802e-06, "loss": 0.001, "step": 27800 }, { "epoch": 212.29007633587787, "grad_norm": 0.08043128997087479, "learning_rate": 4.061025915020655e-06, "loss": 0.0013, "step": 27810 }, { "epoch": 212.36641221374046, "grad_norm": 0.04774537682533264, "learning_rate": 4.024189941890099e-06, "loss": 0.0012, "step": 27820 }, { "epoch": 212.44274809160305, "grad_norm": 0.04703238606452942, "learning_rate": 3.987519521573268e-06, "loss": 0.0013, "step": 27830 }, { "epoch": 212.51908396946564, "grad_norm": 0.050585031509399414, "learning_rate": 3.951014695658494e-06, "loss": 0.0012, "step": 27840 }, { "epoch": 212.59541984732823, "grad_norm": 0.05005447939038277, "learning_rate": 3.914675505546277e-06, "loss": 0.0014, "step": 27850 }, { "epoch": 212.67175572519085, "grad_norm": 0.04051746800541878, "learning_rate": 3.878501992449318e-06, "loss": 0.0014, "step": 27860 }, { "epoch": 212.74809160305344, "grad_norm": 0.08794652670621872, "learning_rate": 3.842494197392398e-06, "loss": 0.0012, "step": 27870 }, { "epoch": 212.82442748091603, "grad_norm": 0.07232579588890076, "learning_rate": 3.8066521612123866e-06, "loss": 0.0013, "step": 27880 }, { "epoch": 212.90076335877862, "grad_norm": 0.07157439738512039, "learning_rate": 3.770975924558134e-06, "loss": 0.0011, "step": 27890 }, { "epoch": 212.9770992366412, "grad_norm": 0.09016720205545425, "learning_rate": 3.735465527890458e-06, "loss": 0.0012, "step": 27900 }, { "epoch": 213.05343511450383, "grad_norm": 0.048123881220817566, "learning_rate": 3.7001210114820935e-06, "loss": 0.0012, "step": 27910 }, { "epoch": 213.12977099236642, "grad_norm": 0.03819369524717331, "learning_rate": 3.6649424154176597e-06, "loss": 0.0011, "step": 27920 }, { "epoch": 213.206106870229, "grad_norm": 0.051055099815130234, "learning_rate": 3.62992977959361e-06, "loss": 0.0011, "step": 27930 }, { "epoch": 213.2824427480916, "grad_norm": 0.03426053002476692, "learning_rate": 3.595083143718147e-06, "loss": 0.0012, "step": 27940 }, { "epoch": 213.3587786259542, "grad_norm": 0.05659709870815277, "learning_rate": 3.560402547311275e-06, "loss": 0.0012, "step": 27950 }, { "epoch": 213.4351145038168, "grad_norm": 0.06023978069424629, "learning_rate": 3.525888029704599e-06, "loss": 0.0014, "step": 27960 }, { "epoch": 213.5114503816794, "grad_norm": 0.04443872720003128, "learning_rate": 3.4915396300414567e-06, "loss": 0.0011, "step": 27970 }, { "epoch": 213.58778625954199, "grad_norm": 0.09381512552499771, "learning_rate": 3.4573573872767547e-06, "loss": 0.0014, "step": 27980 }, { "epoch": 213.66412213740458, "grad_norm": 0.0653676763176918, "learning_rate": 3.423341340176933e-06, "loss": 0.0012, "step": 27990 }, { "epoch": 213.74045801526717, "grad_norm": 0.058748070150613785, "learning_rate": 3.3894915273199987e-06, "loss": 0.001, "step": 28000 }, { "epoch": 213.81679389312978, "grad_norm": 0.052936721593141556, "learning_rate": 3.3558079870954267e-06, "loss": 0.0012, "step": 28010 }, { "epoch": 213.89312977099237, "grad_norm": 0.052125271409749985, "learning_rate": 3.322290757704044e-06, "loss": 0.0011, "step": 28020 }, { "epoch": 213.96946564885496, "grad_norm": 0.048019032925367355, "learning_rate": 3.288939877158159e-06, "loss": 0.0011, "step": 28030 }, { "epoch": 214.04580152671755, "grad_norm": 0.02028910256922245, "learning_rate": 3.2557553832813664e-06, "loss": 0.0013, "step": 28040 }, { "epoch": 214.12213740458014, "grad_norm": 0.06618207693099976, "learning_rate": 3.2227373137085954e-06, "loss": 0.0012, "step": 28050 }, { "epoch": 214.19847328244273, "grad_norm": 0.10718299448490143, "learning_rate": 3.1898857058859917e-06, "loss": 0.0014, "step": 28060 }, { "epoch": 214.27480916030535, "grad_norm": 0.034376729279756546, "learning_rate": 3.1572005970709356e-06, "loss": 0.0012, "step": 28070 }, { "epoch": 214.35114503816794, "grad_norm": 0.07495838403701782, "learning_rate": 3.1246820243319924e-06, "loss": 0.0013, "step": 28080 }, { "epoch": 214.42748091603053, "grad_norm": 0.05478425323963165, "learning_rate": 3.092330024548828e-06, "loss": 0.0013, "step": 28090 }, { "epoch": 214.50381679389312, "grad_norm": 0.05403842031955719, "learning_rate": 3.0601446344122095e-06, "loss": 0.001, "step": 28100 }, { "epoch": 214.5801526717557, "grad_norm": 0.06174512952566147, "learning_rate": 3.0281258904239713e-06, "loss": 0.001, "step": 28110 }, { "epoch": 214.65648854961833, "grad_norm": 0.03383449465036392, "learning_rate": 2.9962738288969003e-06, "loss": 0.0011, "step": 28120 }, { "epoch": 214.73282442748092, "grad_norm": 0.038432877510786057, "learning_rate": 2.9645884859547997e-06, "loss": 0.0012, "step": 28130 }, { "epoch": 214.8091603053435, "grad_norm": 0.0396750271320343, "learning_rate": 2.9330698975323753e-06, "loss": 0.0012, "step": 28140 }, { "epoch": 214.8854961832061, "grad_norm": 0.09042259305715561, "learning_rate": 2.9017180993752e-06, "loss": 0.0011, "step": 28150 }, { "epoch": 214.9618320610687, "grad_norm": 0.03561563044786453, "learning_rate": 2.8705331270397312e-06, "loss": 0.0011, "step": 28160 }, { "epoch": 215.0381679389313, "grad_norm": 0.05805269628763199, "learning_rate": 2.8395150158931624e-06, "loss": 0.0012, "step": 28170 }, { "epoch": 215.1145038167939, "grad_norm": 0.03985278308391571, "learning_rate": 2.808663801113503e-06, "loss": 0.0012, "step": 28180 }, { "epoch": 215.1908396946565, "grad_norm": 0.07294709980487823, "learning_rate": 2.777979517689466e-06, "loss": 0.0012, "step": 28190 }, { "epoch": 215.26717557251908, "grad_norm": 0.07162711769342422, "learning_rate": 2.7474622004204304e-06, "loss": 0.0012, "step": 28200 }, { "epoch": 215.34351145038167, "grad_norm": 0.048122454434633255, "learning_rate": 2.717111883916445e-06, "loss": 0.0012, "step": 28210 }, { "epoch": 215.4198473282443, "grad_norm": 0.0401887483894825, "learning_rate": 2.6869286025981427e-06, "loss": 0.0012, "step": 28220 }, { "epoch": 215.49618320610688, "grad_norm": 0.08610865473747253, "learning_rate": 2.656912390696708e-06, "loss": 0.0012, "step": 28230 }, { "epoch": 215.57251908396947, "grad_norm": 0.03488088771700859, "learning_rate": 2.6270632822538607e-06, "loss": 0.0012, "step": 28240 }, { "epoch": 215.64885496183206, "grad_norm": 0.09417955577373505, "learning_rate": 2.5973813111218546e-06, "loss": 0.0013, "step": 28250 }, { "epoch": 215.72519083969465, "grad_norm": 0.037277400493621826, "learning_rate": 2.567866510963329e-06, "loss": 0.001, "step": 28260 }, { "epoch": 215.80152671755727, "grad_norm": 0.07364153116941452, "learning_rate": 2.5385189152513253e-06, "loss": 0.0012, "step": 28270 }, { "epoch": 215.87786259541986, "grad_norm": 0.05835796520113945, "learning_rate": 2.5093385572693025e-06, "loss": 0.0013, "step": 28280 }, { "epoch": 215.95419847328245, "grad_norm": 0.03694463521242142, "learning_rate": 2.4803254701110385e-06, "loss": 0.0013, "step": 28290 }, { "epoch": 216.03053435114504, "grad_norm": 0.05013750120997429, "learning_rate": 2.4514796866805964e-06, "loss": 0.0012, "step": 28300 }, { "epoch": 216.10687022900763, "grad_norm": 0.06350730359554291, "learning_rate": 2.422801239692307e-06, "loss": 0.0011, "step": 28310 }, { "epoch": 216.18320610687022, "grad_norm": 0.09364441782236099, "learning_rate": 2.3942901616707365e-06, "loss": 0.0013, "step": 28320 }, { "epoch": 216.25954198473283, "grad_norm": 0.05382562056183815, "learning_rate": 2.365946484950587e-06, "loss": 0.0012, "step": 28330 }, { "epoch": 216.33587786259542, "grad_norm": 0.06417946517467499, "learning_rate": 2.3377702416767618e-06, "loss": 0.0012, "step": 28340 }, { "epoch": 216.41221374045801, "grad_norm": 0.05037758871912956, "learning_rate": 2.3097614638042493e-06, "loss": 0.001, "step": 28350 }, { "epoch": 216.4885496183206, "grad_norm": 0.08553322404623032, "learning_rate": 2.2819201830981404e-06, "loss": 0.001, "step": 28360 }, { "epoch": 216.5648854961832, "grad_norm": 0.05735957995057106, "learning_rate": 2.2542464311335105e-06, "loss": 0.0012, "step": 28370 }, { "epoch": 216.6412213740458, "grad_norm": 0.07428009063005447, "learning_rate": 2.2267402392955215e-06, "loss": 0.0012, "step": 28380 }, { "epoch": 216.7175572519084, "grad_norm": 0.06259562820196152, "learning_rate": 2.1994016387792024e-06, "loss": 0.0012, "step": 28390 }, { "epoch": 216.793893129771, "grad_norm": 0.06594822555780411, "learning_rate": 2.1722306605896022e-06, "loss": 0.0011, "step": 28400 }, { "epoch": 216.87022900763358, "grad_norm": 0.06514959782361984, "learning_rate": 2.1452273355416216e-06, "loss": 0.0012, "step": 28410 }, { "epoch": 216.94656488549617, "grad_norm": 0.05773065984249115, "learning_rate": 2.1183916942600464e-06, "loss": 0.0011, "step": 28420 }, { "epoch": 217.0229007633588, "grad_norm": 0.062470655888319016, "learning_rate": 2.0917237671794983e-06, "loss": 0.0012, "step": 28430 }, { "epoch": 217.09923664122138, "grad_norm": 0.04735656455159187, "learning_rate": 2.0652235845443175e-06, "loss": 0.0011, "step": 28440 }, { "epoch": 217.17557251908397, "grad_norm": 0.07425522059202194, "learning_rate": 2.0388911764086966e-06, "loss": 0.0012, "step": 28450 }, { "epoch": 217.25190839694656, "grad_norm": 0.04955790191888809, "learning_rate": 2.012726572636514e-06, "loss": 0.0012, "step": 28460 }, { "epoch": 217.32824427480915, "grad_norm": 0.04772379249334335, "learning_rate": 1.986729802901349e-06, "loss": 0.0011, "step": 28470 }, { "epoch": 217.40458015267177, "grad_norm": 0.057304855436086655, "learning_rate": 1.9609008966864347e-06, "loss": 0.0013, "step": 28480 }, { "epoch": 217.48091603053436, "grad_norm": 0.12455258518457413, "learning_rate": 1.935239883284606e-06, "loss": 0.0011, "step": 28490 }, { "epoch": 217.55725190839695, "grad_norm": 0.06902036815881729, "learning_rate": 1.909746791798317e-06, "loss": 0.0012, "step": 28500 }, { "epoch": 217.63358778625954, "grad_norm": 0.06051567941904068, "learning_rate": 1.8844216511395726e-06, "loss": 0.0013, "step": 28510 }, { "epoch": 217.70992366412213, "grad_norm": 0.05806569755077362, "learning_rate": 1.8592644900298826e-06, "loss": 0.0012, "step": 28520 }, { "epoch": 217.78625954198472, "grad_norm": 0.05155491083860397, "learning_rate": 1.8342753370003072e-06, "loss": 0.0012, "step": 28530 }, { "epoch": 217.86259541984734, "grad_norm": 0.04644613340497017, "learning_rate": 1.8094542203912932e-06, "loss": 0.0012, "step": 28540 }, { "epoch": 217.93893129770993, "grad_norm": 0.07163870334625244, "learning_rate": 1.7848011683527562e-06, "loss": 0.0012, "step": 28550 }, { "epoch": 218.01526717557252, "grad_norm": 0.03413549065589905, "learning_rate": 1.7603162088440148e-06, "loss": 0.0011, "step": 28560 }, { "epoch": 218.0916030534351, "grad_norm": 0.054278697818517685, "learning_rate": 1.735999369633706e-06, "loss": 0.0011, "step": 28570 }, { "epoch": 218.1679389312977, "grad_norm": 0.055673837661743164, "learning_rate": 1.7118506782998699e-06, "loss": 0.0013, "step": 28580 }, { "epoch": 218.24427480916032, "grad_norm": 0.048833996057510376, "learning_rate": 1.687870162229782e-06, "loss": 0.0011, "step": 28590 }, { "epoch": 218.3206106870229, "grad_norm": 0.0601934939622879, "learning_rate": 1.6640578486200373e-06, "loss": 0.0012, "step": 28600 }, { "epoch": 218.3969465648855, "grad_norm": 0.07330451905727386, "learning_rate": 1.6404137644764338e-06, "loss": 0.0012, "step": 28610 }, { "epoch": 218.4732824427481, "grad_norm": 0.05386408418416977, "learning_rate": 1.6169379366139878e-06, "loss": 0.0011, "step": 28620 }, { "epoch": 218.54961832061068, "grad_norm": 0.06385169923305511, "learning_rate": 1.5936303916569193e-06, "loss": 0.0011, "step": 28630 }, { "epoch": 218.6259541984733, "grad_norm": 0.05264114961028099, "learning_rate": 1.5704911560385668e-06, "loss": 0.0012, "step": 28640 }, { "epoch": 218.70229007633588, "grad_norm": 0.07434290647506714, "learning_rate": 1.5475202560014054e-06, "loss": 0.0013, "step": 28650 }, { "epoch": 218.77862595419847, "grad_norm": 0.06376872956752777, "learning_rate": 1.5247177175969793e-06, "loss": 0.0011, "step": 28660 }, { "epoch": 218.85496183206106, "grad_norm": 0.09929394721984863, "learning_rate": 1.5020835666859354e-06, "loss": 0.0012, "step": 28670 }, { "epoch": 218.93129770992365, "grad_norm": 0.056733351200819016, "learning_rate": 1.4796178289378735e-06, "loss": 0.0011, "step": 28680 }, { "epoch": 219.00763358778627, "grad_norm": 0.0697406604886055, "learning_rate": 1.457320529831496e-06, "loss": 0.0011, "step": 28690 }, { "epoch": 219.08396946564886, "grad_norm": 0.03538890182971954, "learning_rate": 1.435191694654375e-06, "loss": 0.0012, "step": 28700 }, { "epoch": 219.16030534351145, "grad_norm": 0.06297555565834045, "learning_rate": 1.4132313485030856e-06, "loss": 0.0013, "step": 28710 }, { "epoch": 219.23664122137404, "grad_norm": 0.05963689088821411, "learning_rate": 1.391439516283105e-06, "loss": 0.0011, "step": 28720 }, { "epoch": 219.31297709923663, "grad_norm": 0.04802251607179642, "learning_rate": 1.3698162227087971e-06, "loss": 0.0011, "step": 28730 }, { "epoch": 219.38931297709922, "grad_norm": 0.04815133288502693, "learning_rate": 1.3483614923033792e-06, "loss": 0.0013, "step": 28740 }, { "epoch": 219.46564885496184, "grad_norm": 0.05833139270544052, "learning_rate": 1.3270753493989373e-06, "loss": 0.0011, "step": 28750 }, { "epoch": 219.54198473282443, "grad_norm": 0.039056017994880676, "learning_rate": 1.3059578181362607e-06, "loss": 0.0011, "step": 28760 }, { "epoch": 219.61832061068702, "grad_norm": 0.06857652217149734, "learning_rate": 1.2850089224650418e-06, "loss": 0.001, "step": 28770 }, { "epoch": 219.6946564885496, "grad_norm": 0.05409243702888489, "learning_rate": 1.2642286861436256e-06, "loss": 0.0011, "step": 28780 }, { "epoch": 219.7709923664122, "grad_norm": 0.048067983239889145, "learning_rate": 1.243617132739111e-06, "loss": 0.0011, "step": 28790 }, { "epoch": 219.84732824427482, "grad_norm": 0.057755809277296066, "learning_rate": 1.2231742856273151e-06, "loss": 0.0013, "step": 28800 }, { "epoch": 219.9236641221374, "grad_norm": 0.0708526149392128, "learning_rate": 1.202900167992693e-06, "loss": 0.0011, "step": 28810 }, { "epoch": 220.0, "grad_norm": 0.07495452463626862, "learning_rate": 1.1827948028283352e-06, "loss": 0.0012, "step": 28820 }, { "epoch": 220.0763358778626, "grad_norm": 0.06641015410423279, "learning_rate": 1.1628582129359686e-06, "loss": 0.0011, "step": 28830 }, { "epoch": 220.15267175572518, "grad_norm": 0.05163255333900452, "learning_rate": 1.143090420925924e-06, "loss": 0.001, "step": 28840 }, { "epoch": 220.2290076335878, "grad_norm": 0.061515387147665024, "learning_rate": 1.1234914492170678e-06, "loss": 0.0012, "step": 28850 }, { "epoch": 220.3053435114504, "grad_norm": 0.0743178129196167, "learning_rate": 1.1040613200368032e-06, "loss": 0.0013, "step": 28860 }, { "epoch": 220.38167938931298, "grad_norm": 0.05529727786779404, "learning_rate": 1.08480005542107e-06, "loss": 0.0012, "step": 28870 }, { "epoch": 220.45801526717557, "grad_norm": 0.05987564101815224, "learning_rate": 1.0657076772142782e-06, "loss": 0.0012, "step": 28880 }, { "epoch": 220.53435114503816, "grad_norm": 0.060307178646326065, "learning_rate": 1.0467842070693234e-06, "loss": 0.0012, "step": 28890 }, { "epoch": 220.61068702290078, "grad_norm": 0.06236105039715767, "learning_rate": 1.0280296664475218e-06, "loss": 0.0011, "step": 28900 }, { "epoch": 220.68702290076337, "grad_norm": 0.06035061180591583, "learning_rate": 1.0094440766185929e-06, "loss": 0.0011, "step": 28910 }, { "epoch": 220.76335877862596, "grad_norm": 0.1589188277721405, "learning_rate": 9.910274586606925e-07, "loss": 0.0013, "step": 28920 }, { "epoch": 220.83969465648855, "grad_norm": 0.06512752175331116, "learning_rate": 9.727798334602799e-07, "loss": 0.0011, "step": 28930 }, { "epoch": 220.91603053435114, "grad_norm": 0.11055031418800354, "learning_rate": 9.54701221712234e-07, "loss": 0.0012, "step": 28940 }, { "epoch": 220.99236641221373, "grad_norm": 0.06782139092683792, "learning_rate": 9.367916439196709e-07, "loss": 0.0011, "step": 28950 }, { "epoch": 221.06870229007635, "grad_norm": 0.060518600046634674, "learning_rate": 9.190511203940765e-07, "loss": 0.0012, "step": 28960 }, { "epoch": 221.14503816793894, "grad_norm": 0.07487376034259796, "learning_rate": 9.014796712551564e-07, "loss": 0.0011, "step": 28970 }, { "epoch": 221.22137404580153, "grad_norm": 0.07516837865114212, "learning_rate": 8.840773164309034e-07, "loss": 0.0011, "step": 28980 }, { "epoch": 221.29770992366412, "grad_norm": 0.06557362526655197, "learning_rate": 8.668440756575301e-07, "loss": 0.0012, "step": 28990 }, { "epoch": 221.3740458015267, "grad_norm": 0.04982545226812363, "learning_rate": 8.49779968479436e-07, "loss": 0.0013, "step": 29000 }, { "epoch": 221.45038167938932, "grad_norm": 0.04154108092188835, "learning_rate": 8.328850142492072e-07, "loss": 0.0011, "step": 29010 }, { "epoch": 221.5267175572519, "grad_norm": 0.08118744939565659, "learning_rate": 8.161592321276333e-07, "loss": 0.0012, "step": 29020 }, { "epoch": 221.6030534351145, "grad_norm": 0.04348035156726837, "learning_rate": 7.996026410835909e-07, "loss": 0.0012, "step": 29030 }, { "epoch": 221.6793893129771, "grad_norm": 0.04675734043121338, "learning_rate": 7.832152598941099e-07, "loss": 0.0011, "step": 29040 }, { "epoch": 221.75572519083968, "grad_norm": 0.0485297255218029, "learning_rate": 7.669971071442738e-07, "loss": 0.0012, "step": 29050 }, { "epoch": 221.8320610687023, "grad_norm": 0.05670668184757233, "learning_rate": 7.509482012273193e-07, "loss": 0.0012, "step": 29060 }, { "epoch": 221.9083969465649, "grad_norm": 0.04381979629397392, "learning_rate": 7.350685603444539e-07, "loss": 0.001, "step": 29070 }, { "epoch": 221.98473282442748, "grad_norm": 0.05890791863203049, "learning_rate": 7.193582025049715e-07, "loss": 0.0011, "step": 29080 }, { "epoch": 222.06106870229007, "grad_norm": 0.048356786370277405, "learning_rate": 7.038171455261699e-07, "loss": 0.0012, "step": 29090 }, { "epoch": 222.13740458015266, "grad_norm": 0.05283192917704582, "learning_rate": 6.884454070333168e-07, "loss": 0.0012, "step": 29100 }, { "epoch": 222.21374045801528, "grad_norm": 0.05877598002552986, "learning_rate": 6.73243004459717e-07, "loss": 0.0012, "step": 29110 }, { "epoch": 222.29007633587787, "grad_norm": 0.043742451816797256, "learning_rate": 6.582099550465458e-07, "loss": 0.0012, "step": 29120 }, { "epoch": 222.36641221374046, "grad_norm": 0.05259019881486893, "learning_rate": 6.433462758429653e-07, "loss": 0.0011, "step": 29130 }, { "epoch": 222.44274809160305, "grad_norm": 0.06895053386688232, "learning_rate": 6.286519837060578e-07, "loss": 0.0013, "step": 29140 }, { "epoch": 222.51908396946564, "grad_norm": 0.05642092600464821, "learning_rate": 6.141270953007593e-07, "loss": 0.0011, "step": 29150 }, { "epoch": 222.59541984732823, "grad_norm": 0.07199185341596603, "learning_rate": 5.997716270999264e-07, "loss": 0.0011, "step": 29160 }, { "epoch": 222.67175572519085, "grad_norm": 0.05231864005327225, "learning_rate": 5.855855953842692e-07, "loss": 0.0011, "step": 29170 }, { "epoch": 222.74809160305344, "grad_norm": 0.09026332199573517, "learning_rate": 5.715690162423014e-07, "loss": 0.0012, "step": 29180 }, { "epoch": 222.82442748091603, "grad_norm": 0.04320318624377251, "learning_rate": 5.577219055703907e-07, "loss": 0.0011, "step": 29190 }, { "epoch": 222.90076335877862, "grad_norm": 0.04865435138344765, "learning_rate": 5.440442790727085e-07, "loss": 0.0011, "step": 29200 }, { "epoch": 222.9770992366412, "grad_norm": 0.044444888830184937, "learning_rate": 5.305361522611962e-07, "loss": 0.0011, "step": 29210 }, { "epoch": 223.05343511450383, "grad_norm": 0.0530143566429615, "learning_rate": 5.171975404555828e-07, "loss": 0.0012, "step": 29220 }, { "epoch": 223.12977099236642, "grad_norm": 0.04259978607296944, "learning_rate": 5.04028458783351e-07, "loss": 0.0011, "step": 29230 }, { "epoch": 223.206106870229, "grad_norm": 0.059064097702503204, "learning_rate": 4.910289221796703e-07, "loss": 0.0012, "step": 29240 }, { "epoch": 223.2824427480916, "grad_norm": 0.035399582237005234, "learning_rate": 4.781989453874813e-07, "loss": 0.0011, "step": 29250 }, { "epoch": 223.3587786259542, "grad_norm": 0.07363159954547882, "learning_rate": 4.65538542957411e-07, "loss": 0.0014, "step": 29260 }, { "epoch": 223.4351145038168, "grad_norm": 0.0497397854924202, "learning_rate": 4.530477292477408e-07, "loss": 0.0012, "step": 29270 }, { "epoch": 223.5114503816794, "grad_norm": 0.046827282756567, "learning_rate": 4.407265184244724e-07, "loss": 0.0012, "step": 29280 }, { "epoch": 223.58778625954199, "grad_norm": 0.0644400417804718, "learning_rate": 4.2857492446122797e-07, "loss": 0.001, "step": 29290 }, { "epoch": 223.66412213740458, "grad_norm": 0.06662241369485855, "learning_rate": 4.1659296113925046e-07, "loss": 0.0012, "step": 29300 }, { "epoch": 223.74045801526717, "grad_norm": 0.041658274829387665, "learning_rate": 4.0478064204741976e-07, "loss": 0.0011, "step": 29310 }, { "epoch": 223.81679389312978, "grad_norm": 0.06456856429576874, "learning_rate": 3.931379805822199e-07, "loss": 0.0012, "step": 29320 }, { "epoch": 223.89312977099237, "grad_norm": 0.03828177973628044, "learning_rate": 3.816649899477553e-07, "loss": 0.0012, "step": 29330 }, { "epoch": 223.96946564885496, "grad_norm": 0.05183401703834534, "learning_rate": 3.703616831556344e-07, "loss": 0.0011, "step": 29340 }, { "epoch": 224.04580152671755, "grad_norm": 0.06533772498369217, "learning_rate": 3.592280730250863e-07, "loss": 0.0012, "step": 29350 }, { "epoch": 224.12213740458014, "grad_norm": 0.05212431028485298, "learning_rate": 3.482641721828605e-07, "loss": 0.001, "step": 29360 }, { "epoch": 224.19847328244273, "grad_norm": 0.0451519601047039, "learning_rate": 3.374699930632441e-07, "loss": 0.0011, "step": 29370 }, { "epoch": 224.27480916030535, "grad_norm": 0.05775589868426323, "learning_rate": 3.268455479080279e-07, "loss": 0.0011, "step": 29380 }, { "epoch": 224.35114503816794, "grad_norm": 0.0626661404967308, "learning_rate": 3.163908487665234e-07, "loss": 0.0012, "step": 29390 }, { "epoch": 224.42748091603053, "grad_norm": 0.03611939400434494, "learning_rate": 3.06105907495513e-07, "loss": 0.0011, "step": 29400 }, { "epoch": 224.50381679389312, "grad_norm": 0.05288393795490265, "learning_rate": 2.959907357592661e-07, "loss": 0.0011, "step": 29410 }, { "epoch": 224.5801526717557, "grad_norm": 0.06242530420422554, "learning_rate": 2.86045345029523e-07, "loss": 0.0011, "step": 29420 }, { "epoch": 224.65648854961833, "grad_norm": 0.04657497629523277, "learning_rate": 2.762697465854613e-07, "loss": 0.0012, "step": 29430 }, { "epoch": 224.73282442748092, "grad_norm": 0.0472450889647007, "learning_rate": 2.666639515136959e-07, "loss": 0.001, "step": 29440 }, { "epoch": 224.8091603053435, "grad_norm": 0.06530901044607162, "learning_rate": 2.572279707082625e-07, "loss": 0.0013, "step": 29450 }, { "epoch": 224.8854961832061, "grad_norm": 0.07491778582334518, "learning_rate": 2.479618148706175e-07, "loss": 0.0012, "step": 29460 }, { "epoch": 224.9618320610687, "grad_norm": 0.058362483978271484, "learning_rate": 2.388654945096213e-07, "loss": 0.0013, "step": 29470 }, { "epoch": 225.0381679389313, "grad_norm": 0.04691874608397484, "learning_rate": 2.2993901994148855e-07, "loss": 0.0011, "step": 29480 }, { "epoch": 225.1145038167939, "grad_norm": 0.053337059915065765, "learning_rate": 2.2118240128985444e-07, "loss": 0.0012, "step": 29490 }, { "epoch": 225.1908396946565, "grad_norm": 0.04230809584259987, "learning_rate": 2.125956484857083e-07, "loss": 0.0011, "step": 29500 }, { "epoch": 225.26717557251908, "grad_norm": 0.06206056848168373, "learning_rate": 2.0417877126736037e-07, "loss": 0.0012, "step": 29510 }, { "epoch": 225.34351145038167, "grad_norm": 0.06817013025283813, "learning_rate": 1.9593177918050818e-07, "loss": 0.0013, "step": 29520 }, { "epoch": 225.4198473282443, "grad_norm": 0.0697116106748581, "learning_rate": 1.8785468157813677e-07, "loss": 0.0012, "step": 29530 }, { "epoch": 225.49618320610688, "grad_norm": 0.04865439608693123, "learning_rate": 1.7994748762058532e-07, "loss": 0.001, "step": 29540 }, { "epoch": 225.57251908396947, "grad_norm": 0.03147342428565025, "learning_rate": 1.7221020627548043e-07, "loss": 0.001, "step": 29550 }, { "epoch": 225.64885496183206, "grad_norm": 0.09071359038352966, "learning_rate": 1.6464284631778625e-07, "loss": 0.0012, "step": 29560 }, { "epoch": 225.72519083969465, "grad_norm": 0.09303729236125946, "learning_rate": 1.5724541632970432e-07, "loss": 0.0013, "step": 29570 }, { "epoch": 225.80152671755727, "grad_norm": 0.05324946716427803, "learning_rate": 1.5001792470075714e-07, "loss": 0.0012, "step": 29580 }, { "epoch": 225.87786259541986, "grad_norm": 0.051077406853437424, "learning_rate": 1.4296037962772122e-07, "loss": 0.0012, "step": 29590 }, { "epoch": 225.95419847328245, "grad_norm": 0.05060196667909622, "learning_rate": 1.3607278911462739e-07, "loss": 0.0011, "step": 29600 }, { "epoch": 226.03053435114504, "grad_norm": 0.0551789291203022, "learning_rate": 1.2935516097279387e-07, "loss": 0.0012, "step": 29610 }, { "epoch": 226.10687022900763, "grad_norm": 0.055997200310230255, "learning_rate": 1.2280750282074315e-07, "loss": 0.0011, "step": 29620 }, { "epoch": 226.18320610687022, "grad_norm": 0.06044953688979149, "learning_rate": 1.1642982208425189e-07, "loss": 0.0011, "step": 29630 }, { "epoch": 226.25954198473283, "grad_norm": 0.034183502197265625, "learning_rate": 1.1022212599631763e-07, "loss": 0.0011, "step": 29640 }, { "epoch": 226.33587786259542, "grad_norm": 0.07105205953121185, "learning_rate": 1.0418442159715879e-07, "loss": 0.0012, "step": 29650 }, { "epoch": 226.41221374045801, "grad_norm": 0.054589904844760895, "learning_rate": 9.831671573419797e-08, "loss": 0.0012, "step": 29660 }, { "epoch": 226.4885496183206, "grad_norm": 0.05667176470160484, "learning_rate": 9.261901506207869e-08, "loss": 0.0011, "step": 29670 }, { "epoch": 226.5648854961832, "grad_norm": 0.05608929693698883, "learning_rate": 8.709132604261537e-08, "loss": 0.0012, "step": 29680 }, { "epoch": 226.6412213740458, "grad_norm": 0.03606346622109413, "learning_rate": 8.17336549448433e-08, "loss": 0.0011, "step": 29690 }, { "epoch": 226.7175572519084, "grad_norm": 0.0540076345205307, "learning_rate": 7.654600784493536e-08, "loss": 0.0014, "step": 29700 }, { "epoch": 226.793893129771, "grad_norm": 0.07126715779304504, "learning_rate": 7.15283906262687e-08, "loss": 0.0011, "step": 29710 }, { "epoch": 226.87022900763358, "grad_norm": 0.07249713689088821, "learning_rate": 6.668080897937466e-08, "loss": 0.0011, "step": 29720 }, { "epoch": 226.94656488549617, "grad_norm": 0.05324969068169594, "learning_rate": 6.200326840195558e-08, "loss": 0.0012, "step": 29730 }, { "epoch": 227.0229007633588, "grad_norm": 0.03399122133851051, "learning_rate": 5.749577419886797e-08, "loss": 0.001, "step": 29740 }, { "epoch": 227.09923664122138, "grad_norm": 0.048613205552101135, "learning_rate": 5.315833148210602e-08, "loss": 0.0011, "step": 29750 }, { "epoch": 227.17557251908397, "grad_norm": 0.05525076761841774, "learning_rate": 4.899094517081814e-08, "loss": 0.0011, "step": 29760 }, { "epoch": 227.25190839694656, "grad_norm": 0.07438613474369049, "learning_rate": 4.499361999127371e-08, "loss": 0.0011, "step": 29770 }, { "epoch": 227.32824427480915, "grad_norm": 0.0654211938381195, "learning_rate": 4.1166360476896366e-08, "loss": 0.0012, "step": 29780 }, { "epoch": 227.40458015267177, "grad_norm": 0.06107145547866821, "learning_rate": 3.750917096821404e-08, "loss": 0.0011, "step": 29790 }, { "epoch": 227.48091603053436, "grad_norm": 0.07306665182113647, "learning_rate": 3.402205561289229e-08, "loss": 0.0012, "step": 29800 }, { "epoch": 227.55725190839695, "grad_norm": 0.05526094138622284, "learning_rate": 3.070501836570094e-08, "loss": 0.0012, "step": 29810 }, { "epoch": 227.63358778625954, "grad_norm": 0.06135351210832596, "learning_rate": 2.7558062988530805e-08, "loss": 0.0011, "step": 29820 }, { "epoch": 227.70992366412213, "grad_norm": 0.040081024169921875, "learning_rate": 2.4581193050393632e-08, "loss": 0.0013, "step": 29830 }, { "epoch": 227.78625954198472, "grad_norm": 0.03797725588083267, "learning_rate": 2.1774411927372172e-08, "loss": 0.0011, "step": 29840 }, { "epoch": 227.86259541984734, "grad_norm": 0.05748415365815163, "learning_rate": 1.9137722802686772e-08, "loss": 0.0012, "step": 29850 }, { "epoch": 227.93893129770993, "grad_norm": 0.06156569719314575, "learning_rate": 1.6671128666612132e-08, "loss": 0.0012, "step": 29860 }, { "epoch": 228.01526717557252, "grad_norm": 0.050229959189891815, "learning_rate": 1.4374632316560553e-08, "loss": 0.001, "step": 29870 }, { "epoch": 228.0916030534351, "grad_norm": 0.04377325251698494, "learning_rate": 1.2248236357015329e-08, "loss": 0.0012, "step": 29880 }, { "epoch": 228.1679389312977, "grad_norm": 0.0403478667140007, "learning_rate": 1.0291943199547404e-08, "loss": 0.0014, "step": 29890 }, { "epoch": 228.24427480916032, "grad_norm": 0.051895685493946075, "learning_rate": 8.50575506278206e-09, "loss": 0.0013, "step": 29900 }, { "epoch": 228.3206106870229, "grad_norm": 0.048128969967365265, "learning_rate": 6.889673972498844e-09, "loss": 0.0011, "step": 29910 }, { "epoch": 228.3969465648855, "grad_norm": 0.03783611208200455, "learning_rate": 5.443701761465025e-09, "loss": 0.0011, "step": 29920 }, { "epoch": 228.4732824427481, "grad_norm": 0.05991312116384506, "learning_rate": 4.1678400696187885e-09, "loss": 0.0011, "step": 29930 }, { "epoch": 228.54961832061068, "grad_norm": 0.08384490758180618, "learning_rate": 3.0620903439193546e-09, "loss": 0.0012, "step": 29940 }, { "epoch": 228.6259541984733, "grad_norm": 0.04013016074895859, "learning_rate": 2.126453838380282e-09, "loss": 0.0012, "step": 29950 }, { "epoch": 228.70229007633588, "grad_norm": 0.05379505828022957, "learning_rate": 1.3609316141527382e-09, "loss": 0.0011, "step": 29960 }, { "epoch": 228.77862595419847, "grad_norm": 0.05796055495738983, "learning_rate": 7.655245394089237e-10, "loss": 0.0012, "step": 29970 }, { "epoch": 228.85496183206106, "grad_norm": 0.05037011206150055, "learning_rate": 3.402332893920334e-10, "loss": 0.001, "step": 29980 }, { "epoch": 228.93129770992365, "grad_norm": 0.028419513255357742, "learning_rate": 8.505834646621579e-11, "loss": 0.0011, "step": 29990 }, { "epoch": 229.00763358778627, "grad_norm": 0.04819118231534958, "learning_rate": 0.0, "loss": 0.0011, "step": 30000 }, { "epoch": 229.00763358778627, "step": 30000, "total_flos": 0.0, "train_loss": 0.00860016386192292, "train_runtime": 6955.2438, "train_samples_per_second": 276.051, "train_steps_per_second": 4.313 } ], "logging_steps": 10, "max_steps": 30000, "num_input_tokens_seen": 0, "num_train_epochs": 230, "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 }