| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 9480, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0010548523206751054, | |
| "grad_norm": 1.224403738975525, | |
| "learning_rate": 0.00015789473684210527, | |
| "loss": 7.6018, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.002109704641350211, | |
| "grad_norm": 1.0830174684524536, | |
| "learning_rate": 0.00031578947368421053, | |
| "loss": 7.0526, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0031645569620253164, | |
| "grad_norm": 0.8968034982681274, | |
| "learning_rate": 0.00047368421052631577, | |
| "loss": 6.4082, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.004219409282700422, | |
| "grad_norm": 0.6159543395042419, | |
| "learning_rate": 0.0006315789473684211, | |
| "loss": 5.8731, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.005274261603375527, | |
| "grad_norm": 0.39616236090660095, | |
| "learning_rate": 0.0007894736842105263, | |
| "loss": 5.4938, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.006329113924050633, | |
| "grad_norm": 0.6134156584739685, | |
| "learning_rate": 0.0009473684210526315, | |
| "loss": 5.0902, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.007383966244725738, | |
| "grad_norm": 0.33243003487586975, | |
| "learning_rate": 0.0011052631578947368, | |
| "loss": 4.6618, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.008438818565400843, | |
| "grad_norm": 0.23324701189994812, | |
| "learning_rate": 0.0012631578947368421, | |
| "loss": 4.339, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.00949367088607595, | |
| "grad_norm": 0.4921824336051941, | |
| "learning_rate": 0.0014210526315789472, | |
| "loss": 4.1184, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.010548523206751054, | |
| "grad_norm": 0.3369714319705963, | |
| "learning_rate": 0.0014999989494847376, | |
| "loss": 3.9508, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.011603375527426161, | |
| "grad_norm": 0.27228906750679016, | |
| "learning_rate": 0.0014999905453802946, | |
| "loss": 3.8005, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.012658227848101266, | |
| "grad_norm": 0.2569299340248108, | |
| "learning_rate": 0.0014999737372655805, | |
| "loss": 3.6694, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.013713080168776372, | |
| "grad_norm": 0.6321362257003784, | |
| "learning_rate": 0.0014999485253289388, | |
| "loss": 3.5732, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.014767932489451477, | |
| "grad_norm": 0.265931636095047, | |
| "learning_rate": 0.0014999149098528814, | |
| "loss": 3.4612, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.015822784810126583, | |
| "grad_norm": 0.36976251006126404, | |
| "learning_rate": 0.0014998728912140862, | |
| "loss": 3.3647, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.016877637130801686, | |
| "grad_norm": 0.29034847021102905, | |
| "learning_rate": 0.0014998224698833922, | |
| "loss": 3.2909, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.017932489451476793, | |
| "grad_norm": 0.31680262088775635, | |
| "learning_rate": 0.0014997636464257956, | |
| "loss": 3.2053, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0189873417721519, | |
| "grad_norm": 0.3954222500324249, | |
| "learning_rate": 0.0014996964215004416, | |
| "loss": 3.1432, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.020042194092827006, | |
| "grad_norm": 0.40973424911499023, | |
| "learning_rate": 0.0014996207958606182, | |
| "loss": 3.0757, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.02109704641350211, | |
| "grad_norm": 0.457225501537323, | |
| "learning_rate": 0.001499536770353748, | |
| "loss": 3.0013, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.022151898734177215, | |
| "grad_norm": 0.5260195732116699, | |
| "learning_rate": 0.0014994443459213774, | |
| "loss": 2.9466, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.023206751054852322, | |
| "grad_norm": 0.33434826135635376, | |
| "learning_rate": 0.001499343523599168, | |
| "loss": 2.8918, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.024261603375527425, | |
| "grad_norm": 0.3714075982570648, | |
| "learning_rate": 0.0014992343045168823, | |
| "loss": 2.8235, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.02531645569620253, | |
| "grad_norm": 0.3708628714084625, | |
| "learning_rate": 0.0014991166898983739, | |
| "loss": 2.769, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.026371308016877638, | |
| "grad_norm": 0.37826642394065857, | |
| "learning_rate": 0.001498990681061572, | |
| "loss": 2.7176, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.027426160337552744, | |
| "grad_norm": 0.3962521255016327, | |
| "learning_rate": 0.001498856279418467, | |
| "loss": 2.6607, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.028481012658227847, | |
| "grad_norm": 0.4650373160839081, | |
| "learning_rate": 0.0014987134864750948, | |
| "loss": 2.6222, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.029535864978902954, | |
| "grad_norm": 0.36165574193000793, | |
| "learning_rate": 0.0014985623038315206, | |
| "loss": 2.571, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.03059071729957806, | |
| "grad_norm": 0.4171614944934845, | |
| "learning_rate": 0.0014984027331818193, | |
| "loss": 2.5323, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.03164556962025317, | |
| "grad_norm": 0.5617461204528809, | |
| "learning_rate": 0.0014982347763140584, | |
| "loss": 2.4971, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03270042194092827, | |
| "grad_norm": 0.6129226684570312, | |
| "learning_rate": 0.0014980584351102762, | |
| "loss": 2.4632, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.03375527426160337, | |
| "grad_norm": 0.46741676330566406, | |
| "learning_rate": 0.001497873711546462, | |
| "loss": 2.4185, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.03481012658227848, | |
| "grad_norm": 0.451256662607193, | |
| "learning_rate": 0.0014976806076925334, | |
| "loss": 2.3998, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.035864978902953586, | |
| "grad_norm": 0.43502041697502136, | |
| "learning_rate": 0.0014974791257123137, | |
| "loss": 2.367, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.03691983122362869, | |
| "grad_norm": 0.3662364184856415, | |
| "learning_rate": 0.001497269267863507, | |
| "loss": 2.3158, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.0379746835443038, | |
| "grad_norm": 0.4172323942184448, | |
| "learning_rate": 0.0014970510364976724, | |
| "loss": 2.2855, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.039029535864978905, | |
| "grad_norm": 0.3845485746860504, | |
| "learning_rate": 0.0014968244340601996, | |
| "loss": 2.2709, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.04008438818565401, | |
| "grad_norm": 0.5006797313690186, | |
| "learning_rate": 0.001496589463090279, | |
| "loss": 2.2583, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.04113924050632911, | |
| "grad_norm": 0.6061115264892578, | |
| "learning_rate": 0.001496346126220875, | |
| "loss": 2.2285, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.04219409282700422, | |
| "grad_norm": 0.36179161071777344, | |
| "learning_rate": 0.0014960944261786966, | |
| "loss": 2.1955, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.043248945147679324, | |
| "grad_norm": 0.3911934792995453, | |
| "learning_rate": 0.0014958343657841655, | |
| "loss": 2.1831, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.04430379746835443, | |
| "grad_norm": 0.4641297459602356, | |
| "learning_rate": 0.001495565947951385, | |
| "loss": 2.1568, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.04535864978902954, | |
| "grad_norm": 0.3561594486236572, | |
| "learning_rate": 0.0014952891756881085, | |
| "loss": 2.1319, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.046413502109704644, | |
| "grad_norm": 0.3676946759223938, | |
| "learning_rate": 0.0014950040520957037, | |
| "loss": 2.1113, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.04746835443037975, | |
| "grad_norm": 0.397060364484787, | |
| "learning_rate": 0.0014947105803691204, | |
| "loss": 2.1096, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.04852320675105485, | |
| "grad_norm": 0.5825448036193848, | |
| "learning_rate": 0.0014944087637968522, | |
| "loss": 2.0875, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.049578059071729956, | |
| "grad_norm": 0.3827154040336609, | |
| "learning_rate": 0.0014940986057609012, | |
| "loss": 2.0607, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.05063291139240506, | |
| "grad_norm": 0.44185370206832886, | |
| "learning_rate": 0.0014937801097367396, | |
| "loss": 2.0567, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.05168776371308017, | |
| "grad_norm": 0.37032192945480347, | |
| "learning_rate": 0.001493453279293271, | |
| "loss": 2.0288, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.052742616033755275, | |
| "grad_norm": 0.4027710556983948, | |
| "learning_rate": 0.0014931181180927902, | |
| "loss": 2.0196, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05379746835443038, | |
| "grad_norm": 0.4590783715248108, | |
| "learning_rate": 0.001492774629890942, | |
| "loss": 2.0054, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.05485232067510549, | |
| "grad_norm": 0.43571019172668457, | |
| "learning_rate": 0.001492422818536679, | |
| "loss": 1.9931, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.05590717299578059, | |
| "grad_norm": 0.36802706122398376, | |
| "learning_rate": 0.00149206268797222, | |
| "loss": 1.9784, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.056962025316455694, | |
| "grad_norm": 0.42650750279426575, | |
| "learning_rate": 0.0014916942422330032, | |
| "loss": 1.9588, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.0580168776371308, | |
| "grad_norm": 0.3881192207336426, | |
| "learning_rate": 0.001491317485447643, | |
| "loss": 1.9576, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.05907172995780591, | |
| "grad_norm": 0.37164539098739624, | |
| "learning_rate": 0.0014909324218378838, | |
| "loss": 1.9199, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.060126582278481014, | |
| "grad_norm": 0.4603891372680664, | |
| "learning_rate": 0.0014905390557185508, | |
| "loss": 1.9272, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.06118143459915612, | |
| "grad_norm": 0.4279107451438904, | |
| "learning_rate": 0.0014901373914975036, | |
| "loss": 1.9275, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.06223628691983123, | |
| "grad_norm": 0.4203340411186218, | |
| "learning_rate": 0.0014897274336755856, | |
| "loss": 1.9022, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.06329113924050633, | |
| "grad_norm": 0.47206705808639526, | |
| "learning_rate": 0.001489309186846575, | |
| "loss": 1.8864, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.06434599156118144, | |
| "grad_norm": 0.46364250779151917, | |
| "learning_rate": 0.0014888826556971313, | |
| "loss": 1.8886, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.06540084388185655, | |
| "grad_norm": 0.41355302929878235, | |
| "learning_rate": 0.0014884478450067444, | |
| "loss": 1.8716, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.06645569620253164, | |
| "grad_norm": 0.5133219957351685, | |
| "learning_rate": 0.0014880047596476807, | |
| "loss": 1.8572, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.06751054852320675, | |
| "grad_norm": 0.5945013761520386, | |
| "learning_rate": 0.0014875534045849274, | |
| "loss": 1.8732, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.06856540084388185, | |
| "grad_norm": 0.37314435839653015, | |
| "learning_rate": 0.0014870937848761388, | |
| "loss": 1.8569, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.06962025316455696, | |
| "grad_norm": 0.36437222361564636, | |
| "learning_rate": 0.001486625905671578, | |
| "loss": 1.8432, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.07067510548523206, | |
| "grad_norm": 0.4192541539669037, | |
| "learning_rate": 0.00148614977221406, | |
| "loss": 1.8279, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.07172995780590717, | |
| "grad_norm": 0.35485246777534485, | |
| "learning_rate": 0.0014856653898388927, | |
| "loss": 1.8186, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.07278481012658228, | |
| "grad_norm": 0.43379876017570496, | |
| "learning_rate": 0.001485172763973817, | |
| "loss": 1.8302, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.07383966244725738, | |
| "grad_norm": 0.36517855525016785, | |
| "learning_rate": 0.0014846719001389466, | |
| "loss": 1.8109, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.07489451476793249, | |
| "grad_norm": 0.3900572955608368, | |
| "learning_rate": 0.001484162803946705, | |
| "loss": 1.8012, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.0759493670886076, | |
| "grad_norm": 0.5315179824829102, | |
| "learning_rate": 0.0014836454811017635, | |
| "loss": 1.787, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.0770042194092827, | |
| "grad_norm": 0.3824789524078369, | |
| "learning_rate": 0.0014831199374009778, | |
| "loss": 1.7798, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.07805907172995781, | |
| "grad_norm": 0.3980269730091095, | |
| "learning_rate": 0.0014825861787333208, | |
| "loss": 1.7912, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.07911392405063292, | |
| "grad_norm": 0.4327452778816223, | |
| "learning_rate": 0.0014820442110798197, | |
| "loss": 1.7699, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.08016877637130802, | |
| "grad_norm": 0.4212438762187958, | |
| "learning_rate": 0.0014814940405134865, | |
| "loss": 1.7656, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.08122362869198312, | |
| "grad_norm": 0.44190293550491333, | |
| "learning_rate": 0.001480935673199251, | |
| "loss": 1.7593, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.08227848101265822, | |
| "grad_norm": 0.45607924461364746, | |
| "learning_rate": 0.0014803691153938915, | |
| "loss": 1.7574, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.08333333333333333, | |
| "grad_norm": 0.37207379937171936, | |
| "learning_rate": 0.0014797943734459653, | |
| "loss": 1.758, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.08438818565400844, | |
| "grad_norm": 0.49604687094688416, | |
| "learning_rate": 0.001479211453795736, | |
| "loss": 1.7343, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.08544303797468354, | |
| "grad_norm": 0.41798657178878784, | |
| "learning_rate": 0.0014786203629751033, | |
| "loss": 1.72, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.08649789029535865, | |
| "grad_norm": 0.5231151580810547, | |
| "learning_rate": 0.0014780211076075279, | |
| "loss": 1.7314, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.08755274261603375, | |
| "grad_norm": 0.3442583978176117, | |
| "learning_rate": 0.0014774136944079594, | |
| "loss": 1.7333, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.08860759493670886, | |
| "grad_norm": 0.3477618992328644, | |
| "learning_rate": 0.0014767981301827592, | |
| "loss": 1.7091, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.08966244725738397, | |
| "grad_norm": 0.3565273582935333, | |
| "learning_rate": 0.0014761744218296249, | |
| "loss": 1.7073, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.09071729957805907, | |
| "grad_norm": 0.3818325102329254, | |
| "learning_rate": 0.001475542576337513, | |
| "loss": 1.7096, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.09177215189873418, | |
| "grad_norm": 0.3839077055454254, | |
| "learning_rate": 0.001474902600786561, | |
| "loss": 1.7062, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.09282700421940929, | |
| "grad_norm": 0.5717847347259521, | |
| "learning_rate": 0.0014742545023480075, | |
| "loss": 1.7008, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.0938818565400844, | |
| "grad_norm": 0.361653596162796, | |
| "learning_rate": 0.0014735982882841117, | |
| "loss": 1.6898, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.0949367088607595, | |
| "grad_norm": 0.34179946780204773, | |
| "learning_rate": 0.0014729339659480727, | |
| "loss": 1.6894, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.09599156118143459, | |
| "grad_norm": 0.37158748507499695, | |
| "learning_rate": 0.0014722615427839468, | |
| "loss": 1.6956, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.0970464135021097, | |
| "grad_norm": 0.36093300580978394, | |
| "learning_rate": 0.0014715810263265633, | |
| "loss": 1.6844, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.0981012658227848, | |
| "grad_norm": 0.3625962436199188, | |
| "learning_rate": 0.0014708924242014423, | |
| "loss": 1.6677, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.09915611814345991, | |
| "grad_norm": 0.39460405707359314, | |
| "learning_rate": 0.0014701957441247064, | |
| "loss": 1.6769, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.10021097046413502, | |
| "grad_norm": 0.4288382828235626, | |
| "learning_rate": 0.0014694909939029959, | |
| "loss": 1.6593, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.10126582278481013, | |
| "grad_norm": 0.4599696397781372, | |
| "learning_rate": 0.0014687781814333814, | |
| "loss": 1.6695, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.10232067510548523, | |
| "grad_norm": 0.34559813141822815, | |
| "learning_rate": 0.0014680573147032746, | |
| "loss": 1.6685, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.10337552742616034, | |
| "grad_norm": 0.4324914515018463, | |
| "learning_rate": 0.0014673284017903392, | |
| "loss": 1.647, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.10443037974683544, | |
| "grad_norm": 0.3713063895702362, | |
| "learning_rate": 0.0014665914508624, | |
| "loss": 1.6466, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.10548523206751055, | |
| "grad_norm": 0.36441338062286377, | |
| "learning_rate": 0.0014658464701773526, | |
| "loss": 1.6539, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.10654008438818566, | |
| "grad_norm": 0.3947252929210663, | |
| "learning_rate": 0.0014650934680830688, | |
| "loss": 1.644, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.10759493670886076, | |
| "grad_norm": 0.39086753129959106, | |
| "learning_rate": 0.0014643324530173051, | |
| "loss": 1.6356, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.10864978902953587, | |
| "grad_norm": 0.4416680932044983, | |
| "learning_rate": 0.0014635634335076067, | |
| "loss": 1.6387, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.10970464135021098, | |
| "grad_norm": 0.39951401948928833, | |
| "learning_rate": 0.001462786418171213, | |
| "loss": 1.6317, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.11075949367088607, | |
| "grad_norm": 0.33045339584350586, | |
| "learning_rate": 0.0014620014157149597, | |
| "loss": 1.6345, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.11181434599156118, | |
| "grad_norm": 0.3590207099914551, | |
| "learning_rate": 0.001461208434935183, | |
| "loss": 1.6198, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.11286919831223628, | |
| "grad_norm": 0.38089054822921753, | |
| "learning_rate": 0.0014604074847176197, | |
| "loss": 1.6115, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.11392405063291139, | |
| "grad_norm": 0.3784695863723755, | |
| "learning_rate": 0.0014595985740373082, | |
| "loss": 1.6124, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.1149789029535865, | |
| "grad_norm": 0.46172553300857544, | |
| "learning_rate": 0.0014587817119584873, | |
| "loss": 1.6223, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.1160337552742616, | |
| "grad_norm": 0.41999849677085876, | |
| "learning_rate": 0.001457956907634496, | |
| "loss": 1.6097, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.11708860759493671, | |
| "grad_norm": 0.32970911264419556, | |
| "learning_rate": 0.0014571241703076692, | |
| "loss": 1.6125, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.11814345991561181, | |
| "grad_norm": 0.34392091631889343, | |
| "learning_rate": 0.0014562835093092348, | |
| "loss": 1.6047, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.11919831223628692, | |
| "grad_norm": 0.4005463421344757, | |
| "learning_rate": 0.0014554349340592104, | |
| "loss": 1.5932, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.12025316455696203, | |
| "grad_norm": 0.4000810980796814, | |
| "learning_rate": 0.001454578454066296, | |
| "loss": 1.6066, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.12130801687763713, | |
| "grad_norm": 0.5581340789794922, | |
| "learning_rate": 0.0014537140789277678, | |
| "loss": 1.5937, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.12236286919831224, | |
| "grad_norm": 0.3587999641895294, | |
| "learning_rate": 0.0014528418183293716, | |
| "loss": 1.5998, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.12341772151898735, | |
| "grad_norm": 0.3587309718132019, | |
| "learning_rate": 0.001451961682045213, | |
| "loss": 1.5899, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.12447257383966245, | |
| "grad_norm": 0.38655397295951843, | |
| "learning_rate": 0.001451073679937649, | |
| "loss": 1.5741, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.12552742616033755, | |
| "grad_norm": 0.36412402987480164, | |
| "learning_rate": 0.0014501778219571766, | |
| "loss": 1.5749, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.12658227848101267, | |
| "grad_norm": 0.3607521653175354, | |
| "learning_rate": 0.0014492741181423225, | |
| "loss": 1.5859, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.12763713080168776, | |
| "grad_norm": 0.3544326722621918, | |
| "learning_rate": 0.0014483625786195285, | |
| "loss": 1.5767, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.12869198312236288, | |
| "grad_norm": 0.3618502616882324, | |
| "learning_rate": 0.0014474432136030405, | |
| "loss": 1.5659, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.12974683544303797, | |
| "grad_norm": 0.3665081858634949, | |
| "learning_rate": 0.0014465160333947923, | |
| "loss": 1.5577, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.1308016877637131, | |
| "grad_norm": 0.3543892502784729, | |
| "learning_rate": 0.0014455810483842908, | |
| "loss": 1.5737, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.13185654008438819, | |
| "grad_norm": 0.435255229473114, | |
| "learning_rate": 0.0014446382690484997, | |
| "loss": 1.5761, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.13291139240506328, | |
| "grad_norm": 0.3356614410877228, | |
| "learning_rate": 0.0014436877059517215, | |
| "loss": 1.5561, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.1339662447257384, | |
| "grad_norm": 0.32167506217956543, | |
| "learning_rate": 0.0014427293697454803, | |
| "loss": 1.5594, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.1350210970464135, | |
| "grad_norm": 0.34562185406684875, | |
| "learning_rate": 0.001441763271168401, | |
| "loss": 1.5668, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.1360759493670886, | |
| "grad_norm": 0.3838872015476227, | |
| "learning_rate": 0.00144078942104609, | |
| "loss": 1.5561, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.1371308016877637, | |
| "grad_norm": 0.3552131652832031, | |
| "learning_rate": 0.001439807830291013, | |
| "loss": 1.5432, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.13818565400843882, | |
| "grad_norm": 0.3455946743488312, | |
| "learning_rate": 0.0014388185099023744, | |
| "loss": 1.55, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.13924050632911392, | |
| "grad_norm": 0.3956320583820343, | |
| "learning_rate": 0.0014378214709659916, | |
| "loss": 1.5514, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.14029535864978904, | |
| "grad_norm": 0.4718681573867798, | |
| "learning_rate": 0.0014368167246541733, | |
| "loss": 1.5419, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.14135021097046413, | |
| "grad_norm": 0.37479954957962036, | |
| "learning_rate": 0.0014358042822255918, | |
| "loss": 1.5427, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.14240506329113925, | |
| "grad_norm": 0.384405255317688, | |
| "learning_rate": 0.0014347841550251597, | |
| "loss": 1.5463, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.14345991561181434, | |
| "grad_norm": 0.41001036763191223, | |
| "learning_rate": 0.0014337563544838997, | |
| "loss": 1.5386, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.14451476793248946, | |
| "grad_norm": 0.4328341782093048, | |
| "learning_rate": 0.001432720892118819, | |
| "loss": 1.5383, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.14556962025316456, | |
| "grad_norm": 0.4074663519859314, | |
| "learning_rate": 0.0014316777795327794, | |
| "loss": 1.5296, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.14662447257383968, | |
| "grad_norm": 0.45945727825164795, | |
| "learning_rate": 0.001430627028414366, | |
| "loss": 1.5394, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.14767932489451477, | |
| "grad_norm": 0.582301676273346, | |
| "learning_rate": 0.0014295686505377586, | |
| "loss": 1.5195, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.14873417721518986, | |
| "grad_norm": 0.42094552516937256, | |
| "learning_rate": 0.0014285026577625982, | |
| "loss": 1.5244, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.14978902953586498, | |
| "grad_norm": 0.36885884404182434, | |
| "learning_rate": 0.0014274290620338542, | |
| "loss": 1.5288, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.15084388185654007, | |
| "grad_norm": 0.3604186773300171, | |
| "learning_rate": 0.0014263478753816906, | |
| "loss": 1.521, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.1518987341772152, | |
| "grad_norm": 0.35973232984542847, | |
| "learning_rate": 0.0014252591099213326, | |
| "loss": 1.5165, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.1529535864978903, | |
| "grad_norm": 0.3624361455440521, | |
| "learning_rate": 0.001424162777852928, | |
| "loss": 1.5175, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.1540084388185654, | |
| "grad_norm": 0.35657012462615967, | |
| "learning_rate": 0.0014230588914614134, | |
| "loss": 1.5148, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.1550632911392405, | |
| "grad_norm": 0.44882655143737793, | |
| "learning_rate": 0.0014219474631163745, | |
| "loss": 1.5112, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.15611814345991562, | |
| "grad_norm": 0.4358622133731842, | |
| "learning_rate": 0.001420828505271909, | |
| "loss": 1.5153, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.1571729957805907, | |
| "grad_norm": 0.42174819111824036, | |
| "learning_rate": 0.0014197020304664856, | |
| "loss": 1.5155, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.15822784810126583, | |
| "grad_norm": 0.40337657928466797, | |
| "learning_rate": 0.0014185680513228048, | |
| "loss": 1.5093, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.15928270042194093, | |
| "grad_norm": 0.43946516513824463, | |
| "learning_rate": 0.0014174265805476564, | |
| "loss": 1.5124, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.16033755274261605, | |
| "grad_norm": 0.5411512851715088, | |
| "learning_rate": 0.0014162776309317778, | |
| "loss": 1.5142, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.16139240506329114, | |
| "grad_norm": 0.3460575044155121, | |
| "learning_rate": 0.0014151212153497108, | |
| "loss": 1.4959, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.16244725738396623, | |
| "grad_norm": 0.4668763279914856, | |
| "learning_rate": 0.0014139573467596561, | |
| "loss": 1.4874, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.16350210970464135, | |
| "grad_norm": 0.3798237144947052, | |
| "learning_rate": 0.00141278603820333, | |
| "loss": 1.4891, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.16455696202531644, | |
| "grad_norm": 0.3989150822162628, | |
| "learning_rate": 0.0014116073028058165, | |
| "loss": 1.4922, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.16561181434599156, | |
| "grad_norm": 0.3972000181674957, | |
| "learning_rate": 0.0014104211537754217, | |
| "loss": 1.4833, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 0.4881563186645508, | |
| "learning_rate": 0.001409227604403524, | |
| "loss": 1.4954, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.16772151898734178, | |
| "grad_norm": 0.40641921758651733, | |
| "learning_rate": 0.0014080266680644277, | |
| "loss": 1.49, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.16877637130801687, | |
| "grad_norm": 0.3936208188533783, | |
| "learning_rate": 0.0014068183582152103, | |
| "loss": 1.4916, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.169831223628692, | |
| "grad_norm": 0.3647351562976837, | |
| "learning_rate": 0.001405602688395574, | |
| "loss": 1.4895, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.17088607594936708, | |
| "grad_norm": 0.4019123613834381, | |
| "learning_rate": 0.0014043796722276924, | |
| "loss": 1.4603, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.1719409282700422, | |
| "grad_norm": 0.39883333444595337, | |
| "learning_rate": 0.0014031493234160591, | |
| "loss": 1.4766, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.1729957805907173, | |
| "grad_norm": 0.36587414145469666, | |
| "learning_rate": 0.0014019116557473332, | |
| "loss": 1.4798, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.17405063291139242, | |
| "grad_norm": 0.44427385926246643, | |
| "learning_rate": 0.0014006666830901854, | |
| "loss": 1.4702, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.1751054852320675, | |
| "grad_norm": 0.34642136096954346, | |
| "learning_rate": 0.001399414419395142, | |
| "loss": 1.4708, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.17616033755274263, | |
| "grad_norm": 0.4188520014286041, | |
| "learning_rate": 0.0013981548786944293, | |
| "loss": 1.4838, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.17721518987341772, | |
| "grad_norm": 0.4440024495124817, | |
| "learning_rate": 0.0013968880751018158, | |
| "loss": 1.4674, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.17827004219409281, | |
| "grad_norm": 0.5161822438240051, | |
| "learning_rate": 0.0013956140228124545, | |
| "loss": 1.4627, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.17932489451476794, | |
| "grad_norm": 0.43483030796051025, | |
| "learning_rate": 0.0013943327361027231, | |
| "loss": 1.482, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.18037974683544303, | |
| "grad_norm": 0.3826151490211487, | |
| "learning_rate": 0.0013930442293300649, | |
| "loss": 1.4594, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.18143459915611815, | |
| "grad_norm": 0.351190984249115, | |
| "learning_rate": 0.0013917485169328279, | |
| "loss": 1.4576, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.18248945147679324, | |
| "grad_norm": 0.38680753111839294, | |
| "learning_rate": 0.0013904456134301016, | |
| "loss": 1.4569, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.18354430379746836, | |
| "grad_norm": 0.34136396646499634, | |
| "learning_rate": 0.0013891355334215562, | |
| "loss": 1.459, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.18459915611814345, | |
| "grad_norm": 0.35594695806503296, | |
| "learning_rate": 0.0013878182915872776, | |
| "loss": 1.4808, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.18565400843881857, | |
| "grad_norm": 0.3368285596370697, | |
| "learning_rate": 0.001386493902687604, | |
| "loss": 1.4578, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.18670886075949367, | |
| "grad_norm": 0.33792853355407715, | |
| "learning_rate": 0.00138516238156296, | |
| "loss": 1.4599, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.1877637130801688, | |
| "grad_norm": 0.35535696148872375, | |
| "learning_rate": 0.0013838237431336895, | |
| "loss": 1.4613, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.18881856540084388, | |
| "grad_norm": 0.3883817791938782, | |
| "learning_rate": 0.0013824780023998899, | |
| "loss": 1.463, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.189873417721519, | |
| "grad_norm": 0.3484611511230469, | |
| "learning_rate": 0.0013811251744412431, | |
| "loss": 1.4543, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.1909282700421941, | |
| "grad_norm": 0.3812905550003052, | |
| "learning_rate": 0.0013797652744168473, | |
| "loss": 1.4457, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.19198312236286919, | |
| "grad_norm": 0.3451510965824127, | |
| "learning_rate": 0.0013783983175650457, | |
| "loss": 1.4416, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.1930379746835443, | |
| "grad_norm": 0.391886442899704, | |
| "learning_rate": 0.0013770243192032581, | |
| "loss": 1.4458, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.1940928270042194, | |
| "grad_norm": 0.3714632987976074, | |
| "learning_rate": 0.0013756432947278064, | |
| "loss": 1.447, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.19514767932489452, | |
| "grad_norm": 0.35794293880462646, | |
| "learning_rate": 0.0013742552596137444, | |
| "loss": 1.4432, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.1962025316455696, | |
| "grad_norm": 0.350976824760437, | |
| "learning_rate": 0.0013728602294146833, | |
| "loss": 1.4472, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.19725738396624473, | |
| "grad_norm": 0.40909984707832336, | |
| "learning_rate": 0.0013714582197626175, | |
| "loss": 1.4506, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.19831223628691982, | |
| "grad_norm": 0.3741796612739563, | |
| "learning_rate": 0.0013700492463677501, | |
| "loss": 1.4352, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.19936708860759494, | |
| "grad_norm": 0.36106088757514954, | |
| "learning_rate": 0.0013686333250183154, | |
| "loss": 1.4426, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.20042194092827004, | |
| "grad_norm": 0.3645462095737457, | |
| "learning_rate": 0.001367210471580404, | |
| "loss": 1.437, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.20147679324894516, | |
| "grad_norm": 0.39476481080055237, | |
| "learning_rate": 0.0013657807019977835, | |
| "loss": 1.4379, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.20253164556962025, | |
| "grad_norm": 0.3888195753097534, | |
| "learning_rate": 0.0013643440322917198, | |
| "loss": 1.4382, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.20358649789029537, | |
| "grad_norm": 0.37703272700309753, | |
| "learning_rate": 0.0013629004785607989, | |
| "loss": 1.4363, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.20464135021097046, | |
| "grad_norm": 0.3618707060813904, | |
| "learning_rate": 0.0013614500569807445, | |
| "loss": 1.4306, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.20569620253164558, | |
| "grad_norm": 0.39237603545188904, | |
| "learning_rate": 0.0013599927838042394, | |
| "loss": 1.4269, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.20675105485232068, | |
| "grad_norm": 0.3404817283153534, | |
| "learning_rate": 0.0013585286753607408, | |
| "loss": 1.4289, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.20780590717299577, | |
| "grad_norm": 0.37156736850738525, | |
| "learning_rate": 0.0013570577480562986, | |
| "loss": 1.44, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.2088607594936709, | |
| "grad_norm": 0.3893108069896698, | |
| "learning_rate": 0.0013555800183733717, | |
| "loss": 1.4195, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.20991561181434598, | |
| "grad_norm": 0.3751927316188812, | |
| "learning_rate": 0.0013540955028706425, | |
| "loss": 1.418, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.2109704641350211, | |
| "grad_norm": 0.35814914107322693, | |
| "learning_rate": 0.0013526042181828324, | |
| "loss": 1.4193, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2120253164556962, | |
| "grad_norm": 0.36299997568130493, | |
| "learning_rate": 0.0013511061810205143, | |
| "loss": 1.4257, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.21308016877637131, | |
| "grad_norm": 0.4335196614265442, | |
| "learning_rate": 0.001349601408169926, | |
| "loss": 1.4377, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.2141350210970464, | |
| "grad_norm": 0.41183096170425415, | |
| "learning_rate": 0.0013480899164927823, | |
| "loss": 1.4227, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.21518987341772153, | |
| "grad_norm": 0.3383892774581909, | |
| "learning_rate": 0.0013465717229260853, | |
| "loss": 1.4217, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.21624472573839662, | |
| "grad_norm": 0.4018089175224304, | |
| "learning_rate": 0.001345046844481935, | |
| "loss": 1.418, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.21729957805907174, | |
| "grad_norm": 0.3578946590423584, | |
| "learning_rate": 0.0013435152982473396, | |
| "loss": 1.4104, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.21835443037974683, | |
| "grad_norm": 0.35767829418182373, | |
| "learning_rate": 0.0013419771013840217, | |
| "loss": 1.4171, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.21940928270042195, | |
| "grad_norm": 0.37678778171539307, | |
| "learning_rate": 0.001340432271128229, | |
| "loss": 1.421, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.22046413502109705, | |
| "grad_norm": 0.3800208866596222, | |
| "learning_rate": 0.0013388808247905381, | |
| "loss": 1.4097, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.22151898734177214, | |
| "grad_norm": 0.41718804836273193, | |
| "learning_rate": 0.0013373227797556634, | |
| "loss": 1.4164, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.22257383966244726, | |
| "grad_norm": 0.39602622389793396, | |
| "learning_rate": 0.00133575815348226, | |
| "loss": 1.4052, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.22362869198312235, | |
| "grad_norm": 0.33915984630584717, | |
| "learning_rate": 0.0013341869635027292, | |
| "loss": 1.405, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.22468354430379747, | |
| "grad_norm": 0.387729674577713, | |
| "learning_rate": 0.001332609227423022, | |
| "loss": 1.4054, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.22573839662447256, | |
| "grad_norm": 0.37279507517814636, | |
| "learning_rate": 0.0013310249629224417, | |
| "loss": 1.4021, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.22679324894514769, | |
| "grad_norm": 0.43018755316734314, | |
| "learning_rate": 0.0013294341877534454, | |
| "loss": 1.4107, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.22784810126582278, | |
| "grad_norm": 0.36134418845176697, | |
| "learning_rate": 0.0013278369197414458, | |
| "loss": 1.4214, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.2289029535864979, | |
| "grad_norm": 0.34999746084213257, | |
| "learning_rate": 0.0013262331767846104, | |
| "loss": 1.4059, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.229957805907173, | |
| "grad_norm": 0.3411415219306946, | |
| "learning_rate": 0.0013246229768536628, | |
| "loss": 1.3894, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.2310126582278481, | |
| "grad_norm": 0.48818546533584595, | |
| "learning_rate": 0.001323006337991679, | |
| "loss": 1.4056, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.2320675105485232, | |
| "grad_norm": 0.3689866364002228, | |
| "learning_rate": 0.0013213832783138873, | |
| "loss": 1.4079, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.23312236286919832, | |
| "grad_norm": 0.47169798612594604, | |
| "learning_rate": 0.0013197538160074633, | |
| "loss": 1.3925, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.23417721518987342, | |
| "grad_norm": 0.36988016963005066, | |
| "learning_rate": 0.0013181179693313283, | |
| "loss": 1.4046, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.23523206751054854, | |
| "grad_norm": 0.3319711983203888, | |
| "learning_rate": 0.0013164757566159428, | |
| "loss": 1.397, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.23628691983122363, | |
| "grad_norm": 0.3849964141845703, | |
| "learning_rate": 0.001314827196263102, | |
| "loss": 1.3931, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.23734177215189872, | |
| "grad_norm": 0.355008602142334, | |
| "learning_rate": 0.0013131723067457302, | |
| "loss": 1.3916, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.23839662447257384, | |
| "grad_norm": 0.40703532099723816, | |
| "learning_rate": 0.0013115111066076721, | |
| "loss": 1.3824, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.23945147679324894, | |
| "grad_norm": 0.36470040678977966, | |
| "learning_rate": 0.0013098436144634862, | |
| "loss": 1.4164, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.24050632911392406, | |
| "grad_norm": 0.3571237623691559, | |
| "learning_rate": 0.0013081698489982364, | |
| "loss": 1.402, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.24156118143459915, | |
| "grad_norm": 0.39680442214012146, | |
| "learning_rate": 0.001306489828967282, | |
| "loss": 1.3905, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.24261603375527427, | |
| "grad_norm": 0.39895832538604736, | |
| "learning_rate": 0.0013048035731960679, | |
| "loss": 1.3893, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.24367088607594936, | |
| "grad_norm": 0.373119980096817, | |
| "learning_rate": 0.0013031111005799133, | |
| "loss": 1.3882, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.24472573839662448, | |
| "grad_norm": 0.3598041534423828, | |
| "learning_rate": 0.0013014124300838004, | |
| "loss": 1.3977, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.24578059071729957, | |
| "grad_norm": 0.3711779713630676, | |
| "learning_rate": 0.0012997075807421612, | |
| "loss": 1.3823, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.2468354430379747, | |
| "grad_norm": 0.34823834896087646, | |
| "learning_rate": 0.0012979965716586653, | |
| "loss": 1.3766, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.2478902953586498, | |
| "grad_norm": 0.3646671175956726, | |
| "learning_rate": 0.0012962794220060048, | |
| "loss": 1.3777, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.2489451476793249, | |
| "grad_norm": 0.43606895208358765, | |
| "learning_rate": 0.0012945561510256801, | |
| "loss": 1.3853, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.3553369641304016, | |
| "learning_rate": 0.001292826778027784, | |
| "loss": 1.3832, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.2510548523206751, | |
| "grad_norm": 0.33519911766052246, | |
| "learning_rate": 0.0012910913223907856, | |
| "loss": 1.3786, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.2521097046413502, | |
| "grad_norm": 0.3543609082698822, | |
| "learning_rate": 0.0012893498035613123, | |
| "loss": 1.3813, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.25316455696202533, | |
| "grad_norm": 0.35886815190315247, | |
| "learning_rate": 0.001287602241053933, | |
| "loss": 1.381, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.2542194092827004, | |
| "grad_norm": 0.36419469118118286, | |
| "learning_rate": 0.0012858486544509392, | |
| "loss": 1.3838, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.2552742616033755, | |
| "grad_norm": 0.377615749835968, | |
| "learning_rate": 0.0012840890634021249, | |
| "loss": 1.377, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.2563291139240506, | |
| "grad_norm": 0.34815406799316406, | |
| "learning_rate": 0.0012823234876245667, | |
| "loss": 1.3734, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.25738396624472576, | |
| "grad_norm": 0.3714103698730469, | |
| "learning_rate": 0.0012805519469024035, | |
| "loss": 1.3823, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.25843881856540085, | |
| "grad_norm": 0.395673930644989, | |
| "learning_rate": 0.0012787744610866143, | |
| "loss": 1.3688, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.25949367088607594, | |
| "grad_norm": 0.34790366888046265, | |
| "learning_rate": 0.0012769910500947954, | |
| "loss": 1.3829, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.26054852320675104, | |
| "grad_norm": 0.34407252073287964, | |
| "learning_rate": 0.0012752017339109376, | |
| "loss": 1.3743, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.2616033755274262, | |
| "grad_norm": 0.3564300835132599, | |
| "learning_rate": 0.0012734065325852029, | |
| "loss": 1.3773, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.2626582278481013, | |
| "grad_norm": 0.3561383783817291, | |
| "learning_rate": 0.0012716054662336987, | |
| "loss": 1.3661, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.26371308016877637, | |
| "grad_norm": 0.46398332715034485, | |
| "learning_rate": 0.001269798555038252, | |
| "loss": 1.3706, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.26476793248945146, | |
| "grad_norm": 0.34344884753227234, | |
| "learning_rate": 0.0012679858192461864, | |
| "loss": 1.3699, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.26582278481012656, | |
| "grad_norm": 0.40484264492988586, | |
| "learning_rate": 0.0012661672791700906, | |
| "loss": 1.3746, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.2668776371308017, | |
| "grad_norm": 0.3783828318119049, | |
| "learning_rate": 0.0012643429551875945, | |
| "loss": 1.3701, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.2679324894514768, | |
| "grad_norm": 0.4477307200431824, | |
| "learning_rate": 0.0012625128677411388, | |
| "loss": 1.3695, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.2689873417721519, | |
| "grad_norm": 0.47914379835128784, | |
| "learning_rate": 0.0012606770373377475, | |
| "loss": 1.3659, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.270042194092827, | |
| "grad_norm": 0.35758259892463684, | |
| "learning_rate": 0.0012588354845487959, | |
| "loss": 1.3724, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.27109704641350213, | |
| "grad_norm": 0.3747115135192871, | |
| "learning_rate": 0.001256988230009783, | |
| "loss": 1.3646, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.2721518987341772, | |
| "grad_norm": 0.3435615301132202, | |
| "learning_rate": 0.0012551352944200976, | |
| "loss": 1.3652, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.2732067510548523, | |
| "grad_norm": 0.3976442217826843, | |
| "learning_rate": 0.0012532766985427874, | |
| "loss": 1.3688, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.2742616033755274, | |
| "grad_norm": 0.3415212631225586, | |
| "learning_rate": 0.0012514124632043272, | |
| "loss": 1.3703, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.27531645569620256, | |
| "grad_norm": 0.3382601737976074, | |
| "learning_rate": 0.0012495426092943842, | |
| "loss": 1.3649, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.27637130801687765, | |
| "grad_norm": 0.35768750309944153, | |
| "learning_rate": 0.0012476671577655845, | |
| "loss": 1.3642, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.27742616033755274, | |
| "grad_norm": 0.3365379869937897, | |
| "learning_rate": 0.0012457861296332774, | |
| "loss": 1.3542, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.27848101265822783, | |
| "grad_norm": 0.363199383020401, | |
| "learning_rate": 0.001243899545975303, | |
| "loss": 1.3624, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.2795358649789029, | |
| "grad_norm": 0.3533865213394165, | |
| "learning_rate": 0.0012420074279317515, | |
| "loss": 1.3562, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.2805907172995781, | |
| "grad_norm": 0.44423454999923706, | |
| "learning_rate": 0.0012401097967047298, | |
| "loss": 1.3517, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.28164556962025317, | |
| "grad_norm": 0.36125123500823975, | |
| "learning_rate": 0.001238206673558122, | |
| "loss": 1.3569, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.28270042194092826, | |
| "grad_norm": 0.35034412145614624, | |
| "learning_rate": 0.0012362980798173526, | |
| "loss": 1.3548, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.28375527426160335, | |
| "grad_norm": 0.3607058823108673, | |
| "learning_rate": 0.0012343840368691462, | |
| "loss": 1.3552, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.2848101265822785, | |
| "grad_norm": 0.36174681782722473, | |
| "learning_rate": 0.0012324645661612886, | |
| "loss": 1.3468, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.2858649789029536, | |
| "grad_norm": 0.37004148960113525, | |
| "learning_rate": 0.0012305396892023867, | |
| "loss": 1.356, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.2869198312236287, | |
| "grad_norm": 0.34047403931617737, | |
| "learning_rate": 0.0012286094275616264, | |
| "loss": 1.3549, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.2879746835443038, | |
| "grad_norm": 0.3488231301307678, | |
| "learning_rate": 0.0012266738028685318, | |
| "loss": 1.3395, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.2890295358649789, | |
| "grad_norm": 0.3499659597873688, | |
| "learning_rate": 0.001224732836812723, | |
| "loss": 1.3459, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.290084388185654, | |
| "grad_norm": 0.3290250599384308, | |
| "learning_rate": 0.0012227865511436724, | |
| "loss": 1.3501, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.2911392405063291, | |
| "grad_norm": 0.4917152523994446, | |
| "learning_rate": 0.001220834967670461, | |
| "loss": 1.3623, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.2921940928270042, | |
| "grad_norm": 0.3639252185821533, | |
| "learning_rate": 0.0012188781082615346, | |
| "loss": 1.3574, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.29324894514767935, | |
| "grad_norm": 0.34025079011917114, | |
| "learning_rate": 0.0012169159948444588, | |
| "loss": 1.3537, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.29430379746835444, | |
| "grad_norm": 0.342074453830719, | |
| "learning_rate": 0.001214948649405672, | |
| "loss": 1.3491, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.29535864978902954, | |
| "grad_norm": 0.3434010446071625, | |
| "learning_rate": 0.0012129760939902407, | |
| "loss": 1.3473, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.29641350210970463, | |
| "grad_norm": 0.37486496567726135, | |
| "learning_rate": 0.0012109983507016114, | |
| "loss": 1.3492, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.2974683544303797, | |
| "grad_norm": 0.3642338812351227, | |
| "learning_rate": 0.0012090154417013636, | |
| "loss": 1.3532, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.29852320675105487, | |
| "grad_norm": 0.31444185972213745, | |
| "learning_rate": 0.0012070273892089605, | |
| "loss": 1.3255, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.29957805907172996, | |
| "grad_norm": 0.3436901271343231, | |
| "learning_rate": 0.0012050342155015012, | |
| "loss": 1.3333, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.30063291139240506, | |
| "grad_norm": 0.3226953148841858, | |
| "learning_rate": 0.0012030359429134707, | |
| "loss": 1.3377, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.30168776371308015, | |
| "grad_norm": 0.32987070083618164, | |
| "learning_rate": 0.0012010325938364883, | |
| "loss": 1.3423, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.3027426160337553, | |
| "grad_norm": 0.3697974979877472, | |
| "learning_rate": 0.0011990241907190592, | |
| "loss": 1.3434, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.3037974683544304, | |
| "grad_norm": 0.35540589690208435, | |
| "learning_rate": 0.001197010756066321, | |
| "loss": 1.3316, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.3048523206751055, | |
| "grad_norm": 0.3480130434036255, | |
| "learning_rate": 0.0011949923124397917, | |
| "loss": 1.3328, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.3059071729957806, | |
| "grad_norm": 0.3540296256542206, | |
| "learning_rate": 0.001192968882457118, | |
| "loss": 1.3406, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.3069620253164557, | |
| "grad_norm": 0.351339727640152, | |
| "learning_rate": 0.001190940488791821, | |
| "loss": 1.3406, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.3080168776371308, | |
| "grad_norm": 0.393935889005661, | |
| "learning_rate": 0.0011889071541730419, | |
| "loss": 1.3353, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.3090717299578059, | |
| "grad_norm": 0.34540510177612305, | |
| "learning_rate": 0.001186868901385288, | |
| "loss": 1.3329, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.310126582278481, | |
| "grad_norm": 0.36295583844184875, | |
| "learning_rate": 0.001184825753268177, | |
| "loss": 1.3359, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.3111814345991561, | |
| "grad_norm": 0.33286401629447937, | |
| "learning_rate": 0.0011827777327161814, | |
| "loss": 1.3453, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.31223628691983124, | |
| "grad_norm": 0.36974620819091797, | |
| "learning_rate": 0.0011807248626783714, | |
| "loss": 1.3248, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.31329113924050633, | |
| "grad_norm": 0.3501949608325958, | |
| "learning_rate": 0.0011786671661581584, | |
| "loss": 1.3218, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.3143459915611814, | |
| "grad_norm": 0.3529147207736969, | |
| "learning_rate": 0.001176604666213036, | |
| "loss": 1.3278, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.3154008438818565, | |
| "grad_norm": 0.33370137214660645, | |
| "learning_rate": 0.0011745373859543236, | |
| "loss": 1.3411, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.31645569620253167, | |
| "grad_norm": 0.3584390878677368, | |
| "learning_rate": 0.0011724653485469063, | |
| "loss": 1.3221, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.31751054852320676, | |
| "grad_norm": 0.3395592272281647, | |
| "learning_rate": 0.0011703885772089743, | |
| "loss": 1.3285, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.31856540084388185, | |
| "grad_norm": 0.33367520570755005, | |
| "learning_rate": 0.0011683070952117646, | |
| "loss": 1.3291, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.31962025316455694, | |
| "grad_norm": 0.4275783896446228, | |
| "learning_rate": 0.0011662209258792998, | |
| "loss": 1.3179, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.3206751054852321, | |
| "grad_norm": 0.385898619890213, | |
| "learning_rate": 0.0011641300925881257, | |
| "loss": 1.3256, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.3217299578059072, | |
| "grad_norm": 0.3597271144390106, | |
| "learning_rate": 0.0011620346187670501, | |
| "loss": 1.3219, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.3227848101265823, | |
| "grad_norm": 0.34854650497436523, | |
| "learning_rate": 0.0011599345278968806, | |
| "loss": 1.3354, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.32383966244725737, | |
| "grad_norm": 0.34655341506004333, | |
| "learning_rate": 0.0011578298435101604, | |
| "loss": 1.3218, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.32489451476793246, | |
| "grad_norm": 0.34645092487335205, | |
| "learning_rate": 0.0011557205891909062, | |
| "loss": 1.3141, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.3259493670886076, | |
| "grad_norm": 0.338767945766449, | |
| "learning_rate": 0.0011536067885743423, | |
| "loss": 1.3299, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.3270042194092827, | |
| "grad_norm": 0.34491071105003357, | |
| "learning_rate": 0.001151488465346637, | |
| "loss": 1.3169, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.3280590717299578, | |
| "grad_norm": 0.3524955213069916, | |
| "learning_rate": 0.0011493656432446362, | |
| "loss": 1.3186, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.3291139240506329, | |
| "grad_norm": 0.37061384320259094, | |
| "learning_rate": 0.0011472383460555983, | |
| "loss": 1.3234, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.33016877637130804, | |
| "grad_norm": 0.36625683307647705, | |
| "learning_rate": 0.001145106597616927, | |
| "loss": 1.3338, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.33122362869198313, | |
| "grad_norm": 0.384627103805542, | |
| "learning_rate": 0.001142970421815904, | |
| "loss": 1.3098, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.3322784810126582, | |
| "grad_norm": 0.3396291136741638, | |
| "learning_rate": 0.0011408298425894226, | |
| "loss": 1.3219, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.3925064206123352, | |
| "learning_rate": 0.0011386848839237186, | |
| "loss": 1.3212, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.33438818565400846, | |
| "grad_norm": 0.3743514120578766, | |
| "learning_rate": 0.0011365355698541005, | |
| "loss": 1.3204, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.33544303797468356, | |
| "grad_norm": 0.37474945187568665, | |
| "learning_rate": 0.0011343819244646824, | |
| "loss": 1.3147, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.33649789029535865, | |
| "grad_norm": 0.41234973073005676, | |
| "learning_rate": 0.001132223971888112, | |
| "loss": 1.3149, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.33755274261603374, | |
| "grad_norm": 0.3496325612068176, | |
| "learning_rate": 0.0011300617363053024, | |
| "loss": 1.31, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.33860759493670883, | |
| "grad_norm": 0.3794001042842865, | |
| "learning_rate": 0.0011278952419451586, | |
| "loss": 1.3232, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 0.339662447257384, | |
| "grad_norm": 0.38375622034072876, | |
| "learning_rate": 0.0011257245130843077, | |
| "loss": 1.3148, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.3407172995780591, | |
| "grad_norm": 0.3591216802597046, | |
| "learning_rate": 0.0011235495740468265, | |
| "loss": 1.3092, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 0.34177215189873417, | |
| "grad_norm": 0.3375595211982727, | |
| "learning_rate": 0.0011213704492039694, | |
| "loss": 1.29, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.34282700421940926, | |
| "grad_norm": 0.37955981492996216, | |
| "learning_rate": 0.001119187162973894, | |
| "loss": 1.3161, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.3438818565400844, | |
| "grad_norm": 0.35981568694114685, | |
| "learning_rate": 0.001116999739821388, | |
| "loss": 1.3048, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.3449367088607595, | |
| "grad_norm": 0.35022369027137756, | |
| "learning_rate": 0.0011148082042575968, | |
| "loss": 1.3239, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 0.3459915611814346, | |
| "grad_norm": 0.3633371889591217, | |
| "learning_rate": 0.0011126125808397461, | |
| "loss": 1.322, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.3470464135021097, | |
| "grad_norm": 0.4322781264781952, | |
| "learning_rate": 0.0011104128941708683, | |
| "loss": 1.3041, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 0.34810126582278483, | |
| "grad_norm": 0.34436267614364624, | |
| "learning_rate": 0.001108209168899527, | |
| "loss": 1.3112, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.3491561181434599, | |
| "grad_norm": 0.3594895601272583, | |
| "learning_rate": 0.0011060014297195396, | |
| "loss": 1.3094, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 0.350210970464135, | |
| "grad_norm": 0.35558730363845825, | |
| "learning_rate": 0.0011037897013697015, | |
| "loss": 1.3153, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.3512658227848101, | |
| "grad_norm": 0.3509814441204071, | |
| "learning_rate": 0.0011015740086335092, | |
| "loss": 1.303, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 0.35232067510548526, | |
| "grad_norm": 0.39170390367507935, | |
| "learning_rate": 0.0010993543763388814, | |
| "loss": 1.3075, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.35337552742616035, | |
| "grad_norm": 0.38959237933158875, | |
| "learning_rate": 0.0010971308293578814, | |
| "loss": 1.3028, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.35443037974683544, | |
| "grad_norm": 0.4082445204257965, | |
| "learning_rate": 0.0010949033926064397, | |
| "loss": 1.3028, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.35548523206751054, | |
| "grad_norm": 0.45328179001808167, | |
| "learning_rate": 0.0010926720910440725, | |
| "loss": 1.3113, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 0.35654008438818563, | |
| "grad_norm": 0.40795767307281494, | |
| "learning_rate": 0.001090436949673603, | |
| "loss": 1.3048, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.3575949367088608, | |
| "grad_norm": 0.34642016887664795, | |
| "learning_rate": 0.0010881979935408815, | |
| "loss": 1.2998, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 0.35864978902953587, | |
| "grad_norm": 0.4289262592792511, | |
| "learning_rate": 0.0010859552477345052, | |
| "loss": 1.3076, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.35970464135021096, | |
| "grad_norm": 0.3674197494983673, | |
| "learning_rate": 0.001083708737385536, | |
| "loss": 1.3056, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 0.36075949367088606, | |
| "grad_norm": 0.3692392110824585, | |
| "learning_rate": 0.0010814584876672187, | |
| "loss": 1.2879, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.3618143459915612, | |
| "grad_norm": 0.3641003966331482, | |
| "learning_rate": 0.0010792045237947008, | |
| "loss": 1.2997, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 0.3628691983122363, | |
| "grad_norm": 0.3449475169181824, | |
| "learning_rate": 0.0010769468710247478, | |
| "loss": 1.3001, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.3639240506329114, | |
| "grad_norm": 0.3534505367279053, | |
| "learning_rate": 0.0010746855546554612, | |
| "loss": 1.2934, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.3649789029535865, | |
| "grad_norm": 0.383791983127594, | |
| "learning_rate": 0.0010724206000259954, | |
| "loss": 1.2894, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.36603375527426163, | |
| "grad_norm": 0.3717612028121948, | |
| "learning_rate": 0.0010701520325162727, | |
| "loss": 1.3072, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 0.3670886075949367, | |
| "grad_norm": 0.3450316786766052, | |
| "learning_rate": 0.0010678798775467001, | |
| "loss": 1.3092, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.3681434599156118, | |
| "grad_norm": 0.3202390670776367, | |
| "learning_rate": 0.0010656041605778832, | |
| "loss": 1.3014, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 0.3691983122362869, | |
| "grad_norm": 0.3394745886325836, | |
| "learning_rate": 0.001063324907110342, | |
| "loss": 1.2814, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.370253164556962, | |
| "grad_norm": 0.3754570782184601, | |
| "learning_rate": 0.0010610421426842241, | |
| "loss": 1.3015, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 0.37130801687763715, | |
| "grad_norm": 0.367051899433136, | |
| "learning_rate": 0.00105875589287902, | |
| "loss": 1.2982, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.37236286919831224, | |
| "grad_norm": 0.39652103185653687, | |
| "learning_rate": 0.0010564661833132752, | |
| "loss": 1.3068, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 0.37341772151898733, | |
| "grad_norm": 0.35056713223457336, | |
| "learning_rate": 0.001054173039644303, | |
| "loss": 1.2984, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 0.3744725738396624, | |
| "grad_norm": 0.37702634930610657, | |
| "learning_rate": 0.0010518764875678981, | |
| "loss": 1.2922, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.3755274261603376, | |
| "grad_norm": 0.35276398062705994, | |
| "learning_rate": 0.001049576552818048, | |
| "loss": 1.2941, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 0.37658227848101267, | |
| "grad_norm": 0.3581322729587555, | |
| "learning_rate": 0.0010472732611666448, | |
| "loss": 1.2877, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 0.37763713080168776, | |
| "grad_norm": 0.4018959701061249, | |
| "learning_rate": 0.0010449666384231954, | |
| "loss": 1.2892, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 0.37869198312236285, | |
| "grad_norm": 0.382752925157547, | |
| "learning_rate": 0.0010426567104345346, | |
| "loss": 1.3005, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 0.379746835443038, | |
| "grad_norm": 0.37618669867515564, | |
| "learning_rate": 0.0010403435030845332, | |
| "loss": 1.2861, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.3808016877637131, | |
| "grad_norm": 0.35512155294418335, | |
| "learning_rate": 0.0010380270422938093, | |
| "loss": 1.284, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 0.3818565400843882, | |
| "grad_norm": 0.47898605465888977, | |
| "learning_rate": 0.0010357073540194362, | |
| "loss": 1.2887, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 0.3829113924050633, | |
| "grad_norm": 0.3588807284832001, | |
| "learning_rate": 0.001033384464254655, | |
| "loss": 1.2857, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 0.38396624472573837, | |
| "grad_norm": 0.3950936198234558, | |
| "learning_rate": 0.001031058399028579, | |
| "loss": 1.2873, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 0.3850210970464135, | |
| "grad_norm": 0.3477298319339752, | |
| "learning_rate": 0.001028729184405905, | |
| "loss": 1.2954, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.3860759493670886, | |
| "grad_norm": 0.36207327246665955, | |
| "learning_rate": 0.0010263968464866201, | |
| "loss": 1.2892, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 0.3871308016877637, | |
| "grad_norm": 0.44805607199668884, | |
| "learning_rate": 0.0010240614114057098, | |
| "loss": 1.2815, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 0.3881856540084388, | |
| "grad_norm": 0.43384575843811035, | |
| "learning_rate": 0.001021722905332864, | |
| "loss": 1.2851, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.38924050632911394, | |
| "grad_norm": 0.35028716921806335, | |
| "learning_rate": 0.0010193813544721855, | |
| "loss": 1.2895, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 0.39029535864978904, | |
| "grad_norm": 0.3653584122657776, | |
| "learning_rate": 0.001017036785061895, | |
| "loss": 1.2941, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.39135021097046413, | |
| "grad_norm": 0.34270143508911133, | |
| "learning_rate": 0.0010146892233740376, | |
| "loss": 1.2841, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 0.3924050632911392, | |
| "grad_norm": 0.41713616251945496, | |
| "learning_rate": 0.0010123386957141883, | |
| "loss": 1.2662, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.39345991561181437, | |
| "grad_norm": 0.3545433282852173, | |
| "learning_rate": 0.0010099852284211573, | |
| "loss": 1.2843, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 0.39451476793248946, | |
| "grad_norm": 0.3524881601333618, | |
| "learning_rate": 0.0010076288478666944, | |
| "loss": 1.2817, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 0.39556962025316456, | |
| "grad_norm": 0.3451279103755951, | |
| "learning_rate": 0.0010052695804551946, | |
| "loss": 1.2877, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.39662447257383965, | |
| "grad_norm": 0.36001715064048767, | |
| "learning_rate": 0.0010029074526234014, | |
| "loss": 1.275, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 0.39767932489451474, | |
| "grad_norm": 0.3992428183555603, | |
| "learning_rate": 0.0010005424908401104, | |
| "loss": 1.2909, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 0.3987341772151899, | |
| "grad_norm": 0.3687851130962372, | |
| "learning_rate": 0.0009981747216058728, | |
| "loss": 1.2761, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 0.399789029535865, | |
| "grad_norm": 0.39275357127189636, | |
| "learning_rate": 0.0009958041714526998, | |
| "loss": 1.2806, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 0.4008438818565401, | |
| "grad_norm": 0.3487364947795868, | |
| "learning_rate": 0.0009934308669437627, | |
| "loss": 1.2866, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.40189873417721517, | |
| "grad_norm": 0.3634994924068451, | |
| "learning_rate": 0.0009910548346730972, | |
| "loss": 1.2745, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 0.4029535864978903, | |
| "grad_norm": 0.35668596625328064, | |
| "learning_rate": 0.0009886761012653062, | |
| "loss": 1.2684, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 0.4040084388185654, | |
| "grad_norm": 0.3821624517440796, | |
| "learning_rate": 0.000986294693375258, | |
| "loss": 1.2716, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 0.4050632911392405, | |
| "grad_norm": 0.3544701039791107, | |
| "learning_rate": 0.000983910637687791, | |
| "loss": 1.2849, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.4061181434599156, | |
| "grad_norm": 0.3593977987766266, | |
| "learning_rate": 0.0009815239609174138, | |
| "loss": 1.268, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.40717299578059074, | |
| "grad_norm": 0.3400956988334656, | |
| "learning_rate": 0.0009791346898080043, | |
| "loss": 1.28, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 0.40822784810126583, | |
| "grad_norm": 0.3652053475379944, | |
| "learning_rate": 0.0009767428511325122, | |
| "loss": 1.2733, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 0.4092827004219409, | |
| "grad_norm": 0.37829872965812683, | |
| "learning_rate": 0.0009743484716926576, | |
| "loss": 1.2622, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 0.410337552742616, | |
| "grad_norm": 0.36608776450157166, | |
| "learning_rate": 0.0009719515783186319, | |
| "loss": 1.2642, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 0.41139240506329117, | |
| "grad_norm": 0.3877875506877899, | |
| "learning_rate": 0.0009695521978687951, | |
| "loss": 1.2656, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.41244725738396626, | |
| "grad_norm": 0.37717291712760925, | |
| "learning_rate": 0.0009671503572293767, | |
| "loss": 1.2705, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 0.41350210970464135, | |
| "grad_norm": 0.3739968538284302, | |
| "learning_rate": 0.0009647460833141742, | |
| "loss": 1.2633, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 0.41455696202531644, | |
| "grad_norm": 0.3548559546470642, | |
| "learning_rate": 0.0009623394030642507, | |
| "loss": 1.2697, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 0.41561181434599154, | |
| "grad_norm": 0.3345405161380768, | |
| "learning_rate": 0.0009599303434476334, | |
| "loss": 1.2639, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 0.4166666666666667, | |
| "grad_norm": 0.3435943126678467, | |
| "learning_rate": 0.0009575189314590118, | |
| "loss": 1.2698, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.4177215189873418, | |
| "grad_norm": 0.3582841157913208, | |
| "learning_rate": 0.0009551051941194346, | |
| "loss": 1.2699, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 0.41877637130801687, | |
| "grad_norm": 0.39052221179008484, | |
| "learning_rate": 0.0009526891584760071, | |
| "loss": 1.2617, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 0.41983122362869196, | |
| "grad_norm": 0.34528857469558716, | |
| "learning_rate": 0.0009502708516015889, | |
| "loss": 1.2704, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 0.4208860759493671, | |
| "grad_norm": 0.35940083861351013, | |
| "learning_rate": 0.0009478503005944888, | |
| "loss": 1.2612, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 0.4219409282700422, | |
| "grad_norm": 0.4698091447353363, | |
| "learning_rate": 0.0009454275325781632, | |
| "loss": 1.2752, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4229957805907173, | |
| "grad_norm": 0.35269781947135925, | |
| "learning_rate": 0.0009430025747009104, | |
| "loss": 1.2627, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 0.4240506329113924, | |
| "grad_norm": 0.35480797290802, | |
| "learning_rate": 0.0009405754541355677, | |
| "loss": 1.2644, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 0.42510548523206754, | |
| "grad_norm": 0.35654541850090027, | |
| "learning_rate": 0.0009381461980792061, | |
| "loss": 1.2581, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 0.42616033755274263, | |
| "grad_norm": 0.4350382387638092, | |
| "learning_rate": 0.0009357148337528256, | |
| "loss": 1.265, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 0.4272151898734177, | |
| "grad_norm": 0.43560877442359924, | |
| "learning_rate": 0.0009332813884010511, | |
| "loss": 1.2698, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.4282700421940928, | |
| "grad_norm": 0.3371978998184204, | |
| "learning_rate": 0.0009308458892918259, | |
| "loss": 1.271, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 0.4293248945147679, | |
| "grad_norm": 0.3576991856098175, | |
| "learning_rate": 0.0009284083637161064, | |
| "loss": 1.2678, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 0.43037974683544306, | |
| "grad_norm": 0.39786258339881897, | |
| "learning_rate": 0.0009259688389875574, | |
| "loss": 1.2712, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 0.43143459915611815, | |
| "grad_norm": 0.40100061893463135, | |
| "learning_rate": 0.0009235273424422442, | |
| "loss": 1.264, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 0.43248945147679324, | |
| "grad_norm": 0.3555314242839813, | |
| "learning_rate": 0.0009210839014383282, | |
| "loss": 1.2598, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.43354430379746833, | |
| "grad_norm": 0.38855957984924316, | |
| "learning_rate": 0.0009186385433557584, | |
| "loss": 1.2577, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 0.4345991561181435, | |
| "grad_norm": 0.42936259508132935, | |
| "learning_rate": 0.0009161912955959668, | |
| "loss": 1.2618, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 0.4356540084388186, | |
| "grad_norm": 0.3469286561012268, | |
| "learning_rate": 0.000913742185581559, | |
| "loss": 1.2507, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 0.43670886075949367, | |
| "grad_norm": 0.338794082403183, | |
| "learning_rate": 0.0009112912407560086, | |
| "loss": 1.2559, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 0.43776371308016876, | |
| "grad_norm": 0.38378769159317017, | |
| "learning_rate": 0.0009088384885833495, | |
| "loss": 1.2606, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.4388185654008439, | |
| "grad_norm": 0.37784048914909363, | |
| "learning_rate": 0.000906383956547867, | |
| "loss": 1.2539, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 0.439873417721519, | |
| "grad_norm": 0.3505966067314148, | |
| "learning_rate": 0.0009039276721537915, | |
| "loss": 1.2546, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 0.4409282700421941, | |
| "grad_norm": 0.37530627846717834, | |
| "learning_rate": 0.0009014696629249886, | |
| "loss": 1.253, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 0.4419831223628692, | |
| "grad_norm": 0.37926673889160156, | |
| "learning_rate": 0.0008990099564046522, | |
| "loss": 1.258, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 0.4430379746835443, | |
| "grad_norm": 0.3719365894794464, | |
| "learning_rate": 0.0008965485801549946, | |
| "loss": 1.2585, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.4440928270042194, | |
| "grad_norm": 0.34479984641075134, | |
| "learning_rate": 0.000894085561756939, | |
| "loss": 1.2487, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 0.4451476793248945, | |
| "grad_norm": 0.3615456521511078, | |
| "learning_rate": 0.0008916209288098088, | |
| "loss": 1.257, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 0.4462025316455696, | |
| "grad_norm": 0.3427673280239105, | |
| "learning_rate": 0.0008891547089310198, | |
| "loss": 1.2533, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 0.4472573839662447, | |
| "grad_norm": 0.34341564774513245, | |
| "learning_rate": 0.0008866869297557699, | |
| "loss": 1.2475, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 0.44831223628691985, | |
| "grad_norm": 0.34663861989974976, | |
| "learning_rate": 0.0008842176189367299, | |
| "loss": 1.254, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.44936708860759494, | |
| "grad_norm": 0.35745692253112793, | |
| "learning_rate": 0.0008817468041437329, | |
| "loss": 1.2599, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 0.45042194092827004, | |
| "grad_norm": 0.35569432377815247, | |
| "learning_rate": 0.0008792745130634654, | |
| "loss": 1.2498, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 0.45147679324894513, | |
| "grad_norm": 0.34108027815818787, | |
| "learning_rate": 0.0008768007733991561, | |
| "loss": 1.2439, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 0.4525316455696203, | |
| "grad_norm": 0.3217419385910034, | |
| "learning_rate": 0.0008743256128702658, | |
| "loss": 1.255, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 0.45358649789029537, | |
| "grad_norm": 0.39865434169769287, | |
| "learning_rate": 0.0008718490592121768, | |
| "loss": 1.2542, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.45464135021097046, | |
| "grad_norm": 0.3799431025981903, | |
| "learning_rate": 0.0008693711401758822, | |
| "loss": 1.2478, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 0.45569620253164556, | |
| "grad_norm": 0.3823153078556061, | |
| "learning_rate": 0.0008668918835276747, | |
| "loss": 1.2481, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 0.45675105485232065, | |
| "grad_norm": 0.3550722599029541, | |
| "learning_rate": 0.0008644113170488355, | |
| "loss": 1.2545, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 0.4578059071729958, | |
| "grad_norm": 0.3509629964828491, | |
| "learning_rate": 0.0008619294685353235, | |
| "loss": 1.2467, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 0.4588607594936709, | |
| "grad_norm": 0.36587169766426086, | |
| "learning_rate": 0.0008594463657974627, | |
| "loss": 1.2447, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.459915611814346, | |
| "grad_norm": 0.3689899146556854, | |
| "learning_rate": 0.0008569620366596322, | |
| "loss": 1.2618, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 0.4609704641350211, | |
| "grad_norm": 0.3530426621437073, | |
| "learning_rate": 0.000854476508959953, | |
| "loss": 1.2376, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 0.4620253164556962, | |
| "grad_norm": 0.35614463686943054, | |
| "learning_rate": 0.0008519898105499762, | |
| "loss": 1.2429, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 0.4630801687763713, | |
| "grad_norm": 0.351806104183197, | |
| "learning_rate": 0.0008495019692943721, | |
| "loss": 1.2407, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 0.4641350210970464, | |
| "grad_norm": 0.37533193826675415, | |
| "learning_rate": 0.0008470130130706166, | |
| "loss": 1.2446, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.4651898734177215, | |
| "grad_norm": 0.3407420814037323, | |
| "learning_rate": 0.0008445229697686795, | |
| "loss": 1.259, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.46624472573839665, | |
| "grad_norm": 0.3381921648979187, | |
| "learning_rate": 0.0008420318672907119, | |
| "loss": 1.2444, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 0.46729957805907174, | |
| "grad_norm": 0.33693447709083557, | |
| "learning_rate": 0.0008395397335507334, | |
| "loss": 1.2468, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 0.46835443037974683, | |
| "grad_norm": 0.39592495560646057, | |
| "learning_rate": 0.0008370465964743196, | |
| "loss": 1.2399, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 0.4694092827004219, | |
| "grad_norm": 0.3851742446422577, | |
| "learning_rate": 0.0008345524839982886, | |
| "loss": 1.2503, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.4704641350210971, | |
| "grad_norm": 0.3582525849342346, | |
| "learning_rate": 0.0008320574240703886, | |
| "loss": 1.2354, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 0.47151898734177217, | |
| "grad_norm": 0.3568393290042877, | |
| "learning_rate": 0.0008295614446489842, | |
| "loss": 1.2437, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 0.47257383966244726, | |
| "grad_norm": 0.3440687656402588, | |
| "learning_rate": 0.0008270645737027441, | |
| "loss": 1.2495, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 0.47362869198312235, | |
| "grad_norm": 0.3949049711227417, | |
| "learning_rate": 0.0008245668392103259, | |
| "loss": 1.2386, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 0.47468354430379744, | |
| "grad_norm": 0.33692407608032227, | |
| "learning_rate": 0.0008220682691600645, | |
| "loss": 1.2374, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.4757383966244726, | |
| "grad_norm": 0.3463355600833893, | |
| "learning_rate": 0.0008195688915496571, | |
| "loss": 1.2433, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 0.4767932489451477, | |
| "grad_norm": 0.34808549284935, | |
| "learning_rate": 0.0008170687343858506, | |
| "loss": 1.2243, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 0.4778481012658228, | |
| "grad_norm": 0.49410006403923035, | |
| "learning_rate": 0.0008145678256841265, | |
| "loss": 1.2453, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 0.47890295358649787, | |
| "grad_norm": 0.3992741107940674, | |
| "learning_rate": 0.0008120661934683879, | |
| "loss": 1.2427, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 0.479957805907173, | |
| "grad_norm": 0.34049564599990845, | |
| "learning_rate": 0.0008095638657706456, | |
| "loss": 1.2424, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.4810126582278481, | |
| "grad_norm": 0.33464476466178894, | |
| "learning_rate": 0.000807060870630703, | |
| "loss": 1.2317, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 0.4820675105485232, | |
| "grad_norm": 0.356582909822464, | |
| "learning_rate": 0.000804557236095843, | |
| "loss": 1.2142, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 0.4831223628691983, | |
| "grad_norm": 0.39746856689453125, | |
| "learning_rate": 0.0008020529902205129, | |
| "loss": 1.2384, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 0.48417721518987344, | |
| "grad_norm": 0.36983558535575867, | |
| "learning_rate": 0.0007995481610660108, | |
| "loss": 1.2247, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 0.48523206751054854, | |
| "grad_norm": 0.3777245581150055, | |
| "learning_rate": 0.0007970427767001702, | |
| "loss": 1.241, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.48628691983122363, | |
| "grad_norm": 0.35135287046432495, | |
| "learning_rate": 0.0007945368651970464, | |
| "loss": 1.2422, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 0.4873417721518987, | |
| "grad_norm": 0.3400181829929352, | |
| "learning_rate": 0.0007920304546366013, | |
| "loss": 1.2234, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 0.4883966244725738, | |
| "grad_norm": 0.41734570264816284, | |
| "learning_rate": 0.000789523573104389, | |
| "loss": 1.2398, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 0.48945147679324896, | |
| "grad_norm": 0.3581525385379791, | |
| "learning_rate": 0.0007870162486912414, | |
| "loss": 1.2394, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 0.49050632911392406, | |
| "grad_norm": 0.35023796558380127, | |
| "learning_rate": 0.0007845085094929527, | |
| "loss": 1.2366, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.49156118143459915, | |
| "grad_norm": 0.39396238327026367, | |
| "learning_rate": 0.0007820003836099649, | |
| "loss": 1.232, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 0.49261603375527424, | |
| "grad_norm": 0.38713324069976807, | |
| "learning_rate": 0.0007794918991470537, | |
| "loss": 1.2324, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 0.4936708860759494, | |
| "grad_norm": 0.3926469385623932, | |
| "learning_rate": 0.0007769830842130119, | |
| "loss": 1.2275, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 0.4947257383966245, | |
| "grad_norm": 0.32987460494041443, | |
| "learning_rate": 0.0007744739669203361, | |
| "loss": 1.2215, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 0.4957805907172996, | |
| "grad_norm": 0.37924712896347046, | |
| "learning_rate": 0.0007719645753849108, | |
| "loss": 1.2297, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.49683544303797467, | |
| "grad_norm": 0.3345460593700409, | |
| "learning_rate": 0.0007694549377256932, | |
| "loss": 1.2321, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 0.4978902953586498, | |
| "grad_norm": 0.4350821077823639, | |
| "learning_rate": 0.0007669450820643987, | |
| "loss": 1.2302, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 0.4989451476793249, | |
| "grad_norm": 0.36744949221611023, | |
| "learning_rate": 0.0007644350365251855, | |
| "loss": 1.2244, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.35269591212272644, | |
| "learning_rate": 0.0007619248292343399, | |
| "loss": 1.2237, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 0.5010548523206751, | |
| "grad_norm": 0.3423689603805542, | |
| "learning_rate": 0.0007594144883199599, | |
| "loss": 1.2441, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.5021097046413502, | |
| "grad_norm": 0.33250823616981506, | |
| "learning_rate": 0.0007569040419116413, | |
| "loss": 1.2274, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 0.5031645569620253, | |
| "grad_norm": 0.4062730073928833, | |
| "learning_rate": 0.000754393518140162, | |
| "loss": 1.2284, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 0.5042194092827004, | |
| "grad_norm": 0.33738499879837036, | |
| "learning_rate": 0.0007518829451371665, | |
| "loss": 1.2336, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 0.5052742616033755, | |
| "grad_norm": 0.3440134525299072, | |
| "learning_rate": 0.0007493723510348516, | |
| "loss": 1.22, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 0.35827070474624634, | |
| "learning_rate": 0.0007468617639656496, | |
| "loss": 1.2188, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.5073839662447257, | |
| "grad_norm": 0.3508043885231018, | |
| "learning_rate": 0.0007443512120619144, | |
| "loss": 1.2208, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 0.5084388185654009, | |
| "grad_norm": 0.3430309593677521, | |
| "learning_rate": 0.0007418407234556067, | |
| "loss": 1.2432, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 0.509493670886076, | |
| "grad_norm": 0.43787845969200134, | |
| "learning_rate": 0.0007393303262779767, | |
| "loss": 1.2253, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 0.510548523206751, | |
| "grad_norm": 0.39787721633911133, | |
| "learning_rate": 0.0007368200486592507, | |
| "loss": 1.2201, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 0.5116033755274262, | |
| "grad_norm": 0.40354400873184204, | |
| "learning_rate": 0.0007343099187283149, | |
| "loss": 1.2322, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.5126582278481012, | |
| "grad_norm": 0.4470100402832031, | |
| "learning_rate": 0.0007317999646124011, | |
| "loss": 1.2214, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 0.5137130801687764, | |
| "grad_norm": 0.3791539967060089, | |
| "learning_rate": 0.0007292902144367704, | |
| "loss": 1.2157, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 0.5147679324894515, | |
| "grad_norm": 0.3646821081638336, | |
| "learning_rate": 0.0007267806963243995, | |
| "loss": 1.2323, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 0.5158227848101266, | |
| "grad_norm": 0.34422731399536133, | |
| "learning_rate": 0.0007242714383956639, | |
| "loss": 1.2224, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 0.5168776371308017, | |
| "grad_norm": 0.3907710313796997, | |
| "learning_rate": 0.000721762468768024, | |
| "loss": 1.2291, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.5179324894514767, | |
| "grad_norm": 0.37104958295822144, | |
| "learning_rate": 0.0007192538155557094, | |
| "loss": 1.2241, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 0.5189873417721519, | |
| "grad_norm": 0.4097077250480652, | |
| "learning_rate": 0.0007167455068694046, | |
| "loss": 1.2126, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 0.520042194092827, | |
| "grad_norm": 0.3646121919155121, | |
| "learning_rate": 0.000714237570815933, | |
| "loss": 1.2246, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 0.5210970464135021, | |
| "grad_norm": 0.39202240109443665, | |
| "learning_rate": 0.0007117300354979423, | |
| "loss": 1.2217, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 0.5221518987341772, | |
| "grad_norm": 0.35184553265571594, | |
| "learning_rate": 0.000709222929013591, | |
| "loss": 1.2289, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.5232067510548524, | |
| "grad_norm": 0.39552363753318787, | |
| "learning_rate": 0.0007067162794562309, | |
| "loss": 1.212, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 0.5242616033755274, | |
| "grad_norm": 0.3939037024974823, | |
| "learning_rate": 0.0007042101149140943, | |
| "loss": 1.219, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 0.5253164556962026, | |
| "grad_norm": 0.3670158088207245, | |
| "learning_rate": 0.0007017044634699787, | |
| "loss": 1.2112, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 0.5263713080168776, | |
| "grad_norm": 0.3672795295715332, | |
| "learning_rate": 0.0006991993532009319, | |
| "loss": 1.2155, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 0.5274261603375527, | |
| "grad_norm": 0.37238597869873047, | |
| "learning_rate": 0.0006966948121779378, | |
| "loss": 1.2237, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5284810126582279, | |
| "grad_norm": 0.34321847558021545, | |
| "learning_rate": 0.000694190868465601, | |
| "loss": 1.2167, | |
| "step": 5010 | |
| }, | |
| { | |
| "epoch": 0.5295358649789029, | |
| "grad_norm": 0.35932210087776184, | |
| "learning_rate": 0.0006916875501218343, | |
| "loss": 1.2118, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 0.5305907172995781, | |
| "grad_norm": 0.33948570489883423, | |
| "learning_rate": 0.0006891848851975416, | |
| "loss": 1.197, | |
| "step": 5030 | |
| }, | |
| { | |
| "epoch": 0.5316455696202531, | |
| "grad_norm": 0.3529501259326935, | |
| "learning_rate": 0.0006866829017363054, | |
| "loss": 1.2252, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 0.5327004219409283, | |
| "grad_norm": 0.3476143479347229, | |
| "learning_rate": 0.0006841816277740722, | |
| "loss": 1.2125, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.5337552742616034, | |
| "grad_norm": 0.38103732466697693, | |
| "learning_rate": 0.0006816810913388379, | |
| "loss": 1.2257, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 0.5348101265822784, | |
| "grad_norm": 0.3743418753147125, | |
| "learning_rate": 0.0006791813204503342, | |
| "loss": 1.2119, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 0.5358649789029536, | |
| "grad_norm": 0.3913949131965637, | |
| "learning_rate": 0.0006766823431197147, | |
| "loss": 1.2239, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 0.5369198312236287, | |
| "grad_norm": 0.36954477429389954, | |
| "learning_rate": 0.0006741841873492406, | |
| "loss": 1.215, | |
| "step": 5090 | |
| }, | |
| { | |
| "epoch": 0.5379746835443038, | |
| "grad_norm": 0.3766242563724518, | |
| "learning_rate": 0.0006716868811319671, | |
| "loss": 1.2004, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.5390295358649789, | |
| "grad_norm": 0.36319437623023987, | |
| "learning_rate": 0.0006691904524514297, | |
| "loss": 1.2125, | |
| "step": 5110 | |
| }, | |
| { | |
| "epoch": 0.540084388185654, | |
| "grad_norm": 0.35376471281051636, | |
| "learning_rate": 0.0006666949292813306, | |
| "loss": 1.2082, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.5411392405063291, | |
| "grad_norm": 0.3592052757740021, | |
| "learning_rate": 0.0006642003395852258, | |
| "loss": 1.2081, | |
| "step": 5130 | |
| }, | |
| { | |
| "epoch": 0.5421940928270043, | |
| "grad_norm": 0.3607555031776428, | |
| "learning_rate": 0.0006617067113162103, | |
| "loss": 1.2217, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 0.5432489451476793, | |
| "grad_norm": 0.34854963421821594, | |
| "learning_rate": 0.0006592140724166073, | |
| "loss": 1.2167, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.5443037974683544, | |
| "grad_norm": 0.3794786036014557, | |
| "learning_rate": 0.0006567224508176523, | |
| "loss": 1.2116, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 0.5453586497890295, | |
| "grad_norm": 0.352797269821167, | |
| "learning_rate": 0.0006542318744391821, | |
| "loss": 1.2106, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 0.5464135021097046, | |
| "grad_norm": 0.3626922369003296, | |
| "learning_rate": 0.0006517423711893209, | |
| "loss": 1.2181, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 0.5474683544303798, | |
| "grad_norm": 0.3714168667793274, | |
| "learning_rate": 0.0006492539689641685, | |
| "loss": 1.2022, | |
| "step": 5190 | |
| }, | |
| { | |
| "epoch": 0.5485232067510548, | |
| "grad_norm": 0.39416635036468506, | |
| "learning_rate": 0.0006467666956474865, | |
| "loss": 1.2142, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.54957805907173, | |
| "grad_norm": 0.36706307530403137, | |
| "learning_rate": 0.0006442805791103873, | |
| "loss": 1.2016, | |
| "step": 5210 | |
| }, | |
| { | |
| "epoch": 0.5506329113924051, | |
| "grad_norm": 0.3452375829219818, | |
| "learning_rate": 0.0006417956472110205, | |
| "loss": 1.2108, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 0.5516877637130801, | |
| "grad_norm": 0.4015010893344879, | |
| "learning_rate": 0.0006393119277942614, | |
| "loss": 1.2049, | |
| "step": 5230 | |
| }, | |
| { | |
| "epoch": 0.5527426160337553, | |
| "grad_norm": 0.42108821868896484, | |
| "learning_rate": 0.0006368294486913987, | |
| "loss": 1.209, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 0.5537974683544303, | |
| "grad_norm": 0.3588517904281616, | |
| "learning_rate": 0.0006343482377198232, | |
| "loss": 1.2027, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.5548523206751055, | |
| "grad_norm": 0.3523201048374176, | |
| "learning_rate": 0.0006318683226827151, | |
| "loss": 1.2083, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 0.5559071729957806, | |
| "grad_norm": 0.38106194138526917, | |
| "learning_rate": 0.0006293897313687331, | |
| "loss": 1.2161, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 0.5569620253164557, | |
| "grad_norm": 0.38883480429649353, | |
| "learning_rate": 0.0006269124915517037, | |
| "loss": 1.2054, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 0.5580168776371308, | |
| "grad_norm": 0.3482867181301117, | |
| "learning_rate": 0.0006244366309903084, | |
| "loss": 1.2067, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 0.5590717299578059, | |
| "grad_norm": 0.3491329848766327, | |
| "learning_rate": 0.0006219621774277737, | |
| "loss": 1.2112, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.560126582278481, | |
| "grad_norm": 0.35099756717681885, | |
| "learning_rate": 0.00061948915859156, | |
| "loss": 1.1956, | |
| "step": 5310 | |
| }, | |
| { | |
| "epoch": 0.5611814345991561, | |
| "grad_norm": 0.3603163957595825, | |
| "learning_rate": 0.0006170176021930509, | |
| "loss": 1.2005, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 0.5622362869198312, | |
| "grad_norm": 0.3842976689338684, | |
| "learning_rate": 0.0006145475359272424, | |
| "loss": 1.207, | |
| "step": 5330 | |
| }, | |
| { | |
| "epoch": 0.5632911392405063, | |
| "grad_norm": 0.35818201303482056, | |
| "learning_rate": 0.0006120789874724336, | |
| "loss": 1.1974, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 0.5643459915611815, | |
| "grad_norm": 0.34876447916030884, | |
| "learning_rate": 0.0006096119844899151, | |
| "loss": 1.1943, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.5654008438818565, | |
| "grad_norm": 0.3554922938346863, | |
| "learning_rate": 0.0006071465546236601, | |
| "loss": 1.2042, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 0.5664556962025317, | |
| "grad_norm": 0.3837421238422394, | |
| "learning_rate": 0.0006046827255000135, | |
| "loss": 1.2043, | |
| "step": 5370 | |
| }, | |
| { | |
| "epoch": 0.5675105485232067, | |
| "grad_norm": 0.4092616140842438, | |
| "learning_rate": 0.0006022205247273845, | |
| "loss": 1.1952, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 0.5685654008438819, | |
| "grad_norm": 0.37640655040740967, | |
| "learning_rate": 0.0005997599798959343, | |
| "loss": 1.1963, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 0.569620253164557, | |
| "grad_norm": 0.35962364077568054, | |
| "learning_rate": 0.0005973011185772694, | |
| "loss": 1.1974, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.570675105485232, | |
| "grad_norm": 0.37594228982925415, | |
| "learning_rate": 0.0005948439683241318, | |
| "loss": 1.2018, | |
| "step": 5410 | |
| }, | |
| { | |
| "epoch": 0.5717299578059072, | |
| "grad_norm": 0.36223894357681274, | |
| "learning_rate": 0.0005923885566700896, | |
| "loss": 1.1931, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 0.5727848101265823, | |
| "grad_norm": 0.3633806109428406, | |
| "learning_rate": 0.0005899349111292293, | |
| "loss": 1.1905, | |
| "step": 5430 | |
| }, | |
| { | |
| "epoch": 0.5738396624472574, | |
| "grad_norm": 0.38978320360183716, | |
| "learning_rate": 0.0005874830591958474, | |
| "loss": 1.1868, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 0.5748945147679325, | |
| "grad_norm": 0.35502946376800537, | |
| "learning_rate": 0.000585033028344142, | |
| "loss": 1.1953, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.5759493670886076, | |
| "grad_norm": 0.3547123670578003, | |
| "learning_rate": 0.0005825848460279048, | |
| "loss": 1.1962, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 0.5770042194092827, | |
| "grad_norm": 0.3606116473674774, | |
| "learning_rate": 0.0005801385396802146, | |
| "loss": 1.1962, | |
| "step": 5470 | |
| }, | |
| { | |
| "epoch": 0.5780590717299579, | |
| "grad_norm": 0.3635554313659668, | |
| "learning_rate": 0.0005776941367131282, | |
| "loss": 1.1984, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 0.5791139240506329, | |
| "grad_norm": 0.4081866145133972, | |
| "learning_rate": 0.0005752516645173745, | |
| "loss": 1.2017, | |
| "step": 5490 | |
| }, | |
| { | |
| "epoch": 0.580168776371308, | |
| "grad_norm": 0.3537876009941101, | |
| "learning_rate": 0.0005728111504620472, | |
| "loss": 1.1939, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.5812236286919831, | |
| "grad_norm": 0.40251171588897705, | |
| "learning_rate": 0.0005703726218942976, | |
| "loss": 1.1996, | |
| "step": 5510 | |
| }, | |
| { | |
| "epoch": 0.5822784810126582, | |
| "grad_norm": 0.4112827777862549, | |
| "learning_rate": 0.0005679361061390295, | |
| "loss": 1.192, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 0.5833333333333334, | |
| "grad_norm": 0.39286091923713684, | |
| "learning_rate": 0.0005655016304985908, | |
| "loss": 1.1874, | |
| "step": 5530 | |
| }, | |
| { | |
| "epoch": 0.5843881856540084, | |
| "grad_norm": 0.3851916193962097, | |
| "learning_rate": 0.0005630692222524709, | |
| "loss": 1.1848, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 0.5854430379746836, | |
| "grad_norm": 0.3481987416744232, | |
| "learning_rate": 0.0005606389086569911, | |
| "loss": 1.1994, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.5864978902953587, | |
| "grad_norm": 0.36745893955230713, | |
| "learning_rate": 0.0005582107169450023, | |
| "loss": 1.2026, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 0.5875527426160337, | |
| "grad_norm": 0.3689424991607666, | |
| "learning_rate": 0.0005557846743255783, | |
| "loss": 1.1887, | |
| "step": 5570 | |
| }, | |
| { | |
| "epoch": 0.5886075949367089, | |
| "grad_norm": 0.3701568841934204, | |
| "learning_rate": 0.0005533608079837109, | |
| "loss": 1.1889, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 0.5896624472573839, | |
| "grad_norm": 0.35756173729896545, | |
| "learning_rate": 0.0005509391450800061, | |
| "loss": 1.1906, | |
| "step": 5590 | |
| }, | |
| { | |
| "epoch": 0.5907172995780591, | |
| "grad_norm": 0.38948529958724976, | |
| "learning_rate": 0.0005485197127503795, | |
| "loss": 1.1877, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.5917721518987342, | |
| "grad_norm": 0.401199072599411, | |
| "learning_rate": 0.0005461025381057516, | |
| "loss": 1.1901, | |
| "step": 5610 | |
| }, | |
| { | |
| "epoch": 0.5928270042194093, | |
| "grad_norm": 0.3774757385253906, | |
| "learning_rate": 0.0005436876482317444, | |
| "loss": 1.1974, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 0.5938818565400844, | |
| "grad_norm": 0.3717878460884094, | |
| "learning_rate": 0.0005412750701883782, | |
| "loss": 1.1923, | |
| "step": 5630 | |
| }, | |
| { | |
| "epoch": 0.5949367088607594, | |
| "grad_norm": 0.4294511675834656, | |
| "learning_rate": 0.0005388648310097682, | |
| "loss": 1.2008, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 0.5959915611814346, | |
| "grad_norm": 0.36430466175079346, | |
| "learning_rate": 0.000536456957703821, | |
| "loss": 1.1945, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.5970464135021097, | |
| "grad_norm": 0.3893060088157654, | |
| "learning_rate": 0.0005340514772519324, | |
| "loss": 1.1896, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 0.5981012658227848, | |
| "grad_norm": 0.3386535942554474, | |
| "learning_rate": 0.0005316484166086863, | |
| "loss": 1.1962, | |
| "step": 5670 | |
| }, | |
| { | |
| "epoch": 0.5991561181434599, | |
| "grad_norm": 0.372712105512619, | |
| "learning_rate": 0.00052924780270155, | |
| "loss": 1.1963, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 0.6002109704641351, | |
| "grad_norm": 0.36947306990623474, | |
| "learning_rate": 0.0005268496624305747, | |
| "loss": 1.1764, | |
| "step": 5690 | |
| }, | |
| { | |
| "epoch": 0.6012658227848101, | |
| "grad_norm": 0.35860103368759155, | |
| "learning_rate": 0.0005244540226680931, | |
| "loss": 1.182, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.6023206751054853, | |
| "grad_norm": 0.3623606562614441, | |
| "learning_rate": 0.0005220609102584185, | |
| "loss": 1.1873, | |
| "step": 5710 | |
| }, | |
| { | |
| "epoch": 0.6033755274261603, | |
| "grad_norm": 0.3721083998680115, | |
| "learning_rate": 0.0005196703520175437, | |
| "loss": 1.1913, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 0.6044303797468354, | |
| "grad_norm": 0.40760260820388794, | |
| "learning_rate": 0.0005172823747328415, | |
| "loss": 1.1886, | |
| "step": 5730 | |
| }, | |
| { | |
| "epoch": 0.6054852320675106, | |
| "grad_norm": 0.35366445779800415, | |
| "learning_rate": 0.0005148970051627632, | |
| "loss": 1.185, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 0.6065400843881856, | |
| "grad_norm": 0.3632625937461853, | |
| "learning_rate": 0.0005125142700365394, | |
| "loss": 1.159, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.6075949367088608, | |
| "grad_norm": 0.40271735191345215, | |
| "learning_rate": 0.000510134196053881, | |
| "loss": 1.1697, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 0.6086497890295358, | |
| "grad_norm": 0.37240713834762573, | |
| "learning_rate": 0.0005077568098846789, | |
| "loss": 1.1709, | |
| "step": 5770 | |
| }, | |
| { | |
| "epoch": 0.609704641350211, | |
| "grad_norm": 0.40513280034065247, | |
| "learning_rate": 0.000505382138168706, | |
| "loss": 1.1806, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 0.6107594936708861, | |
| "grad_norm": 0.3707410991191864, | |
| "learning_rate": 0.0005030102075153181, | |
| "loss": 1.1779, | |
| "step": 5790 | |
| }, | |
| { | |
| "epoch": 0.6118143459915611, | |
| "grad_norm": 0.39285629987716675, | |
| "learning_rate": 0.0005006410445031569, | |
| "loss": 1.1814, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.6128691983122363, | |
| "grad_norm": 0.36199751496315, | |
| "learning_rate": 0.0004982746756798507, | |
| "loss": 1.184, | |
| "step": 5810 | |
| }, | |
| { | |
| "epoch": 0.6139240506329114, | |
| "grad_norm": 0.36564263701438904, | |
| "learning_rate": 0.0004959111275617174, | |
| "loss": 1.1968, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 0.6149789029535865, | |
| "grad_norm": 0.3686883747577667, | |
| "learning_rate": 0.0004935504266334677, | |
| "loss": 1.1806, | |
| "step": 5830 | |
| }, | |
| { | |
| "epoch": 0.6160337552742616, | |
| "grad_norm": 0.3595767617225647, | |
| "learning_rate": 0.0004911925993479085, | |
| "loss": 1.1911, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 0.6170886075949367, | |
| "grad_norm": 0.380021870136261, | |
| "learning_rate": 0.0004888376721256456, | |
| "loss": 1.1966, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.6181434599156118, | |
| "grad_norm": 0.3993105888366699, | |
| "learning_rate": 0.00048648567135478805, | |
| "loss": 1.1838, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 0.619198312236287, | |
| "grad_norm": 0.3877832889556885, | |
| "learning_rate": 0.0004841366233906538, | |
| "loss": 1.1741, | |
| "step": 5870 | |
| }, | |
| { | |
| "epoch": 0.620253164556962, | |
| "grad_norm": 0.3675520122051239, | |
| "learning_rate": 0.0004817905545554717, | |
| "loss": 1.1869, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 0.6213080168776371, | |
| "grad_norm": 0.365359365940094, | |
| "learning_rate": 0.00047944749113808884, | |
| "loss": 1.1756, | |
| "step": 5890 | |
| }, | |
| { | |
| "epoch": 0.6223628691983122, | |
| "grad_norm": 0.34352290630340576, | |
| "learning_rate": 0.00047710745939367474, | |
| "loss": 1.1786, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.6234177215189873, | |
| "grad_norm": 0.3574180006980896, | |
| "learning_rate": 0.0004747704855434278, | |
| "loss": 1.1714, | |
| "step": 5910 | |
| }, | |
| { | |
| "epoch": 0.6244725738396625, | |
| "grad_norm": 0.3392818868160248, | |
| "learning_rate": 0.0004724365957742809, | |
| "loss": 1.1683, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 0.6255274261603375, | |
| "grad_norm": 0.33924534916877747, | |
| "learning_rate": 0.00047010581623860883, | |
| "loss": 1.1744, | |
| "step": 5930 | |
| }, | |
| { | |
| "epoch": 0.6265822784810127, | |
| "grad_norm": 0.36496856808662415, | |
| "learning_rate": 0.0004677781730539342, | |
| "loss": 1.1808, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 0.6276371308016878, | |
| "grad_norm": 0.35622742772102356, | |
| "learning_rate": 0.0004654536923026356, | |
| "loss": 1.172, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.6286919831223629, | |
| "grad_norm": 0.35895973443984985, | |
| "learning_rate": 0.00046313240003165466, | |
| "loss": 1.1776, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 0.629746835443038, | |
| "grad_norm": 0.3521154224872589, | |
| "learning_rate": 0.0004608143222522048, | |
| "loss": 1.1763, | |
| "step": 5970 | |
| }, | |
| { | |
| "epoch": 0.630801687763713, | |
| "grad_norm": 0.34627509117126465, | |
| "learning_rate": 0.0004584994849394795, | |
| "loss": 1.1611, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 0.6318565400843882, | |
| "grad_norm": 0.38166508078575134, | |
| "learning_rate": 0.0004561879140323607, | |
| "loss": 1.1793, | |
| "step": 5990 | |
| }, | |
| { | |
| "epoch": 0.6329113924050633, | |
| "grad_norm": 0.37901800870895386, | |
| "learning_rate": 0.0004538796354331298, | |
| "loss": 1.1864, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6339662447257384, | |
| "grad_norm": 0.4082631766796112, | |
| "learning_rate": 0.0004515746750071754, | |
| "loss": 1.1753, | |
| "step": 6010 | |
| }, | |
| { | |
| "epoch": 0.6350210970464135, | |
| "grad_norm": 0.3545146882534027, | |
| "learning_rate": 0.0004492730585827046, | |
| "loss": 1.1744, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 0.6360759493670886, | |
| "grad_norm": 0.3988998830318451, | |
| "learning_rate": 0.0004469748119504529, | |
| "loss": 1.1594, | |
| "step": 6030 | |
| }, | |
| { | |
| "epoch": 0.6371308016877637, | |
| "grad_norm": 0.38194578886032104, | |
| "learning_rate": 0.0004446799608633964, | |
| "loss": 1.1786, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 0.6381856540084389, | |
| "grad_norm": 0.37773770093917847, | |
| "learning_rate": 0.00044238853103646154, | |
| "loss": 1.1644, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.6392405063291139, | |
| "grad_norm": 0.39295461773872375, | |
| "learning_rate": 0.00044010054814623925, | |
| "loss": 1.18, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 0.640295358649789, | |
| "grad_norm": 0.3779560327529907, | |
| "learning_rate": 0.0004378160378306944, | |
| "loss": 1.171, | |
| "step": 6070 | |
| }, | |
| { | |
| "epoch": 0.6413502109704642, | |
| "grad_norm": 0.36832717061042786, | |
| "learning_rate": 0.00043553502568888095, | |
| "loss": 1.1697, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 0.6424050632911392, | |
| "grad_norm": 0.3654510974884033, | |
| "learning_rate": 0.0004332575372806534, | |
| "loss": 1.1664, | |
| "step": 6090 | |
| }, | |
| { | |
| "epoch": 0.6434599156118144, | |
| "grad_norm": 0.4271853268146515, | |
| "learning_rate": 0.00043098359812638145, | |
| "loss": 1.1625, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.6445147679324894, | |
| "grad_norm": 0.3969768285751343, | |
| "learning_rate": 0.00042871323370666383, | |
| "loss": 1.1747, | |
| "step": 6110 | |
| }, | |
| { | |
| "epoch": 0.6455696202531646, | |
| "grad_norm": 0.3841383755207062, | |
| "learning_rate": 0.0004264464694620421, | |
| "loss": 1.1684, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 0.6466244725738397, | |
| "grad_norm": 0.3560936450958252, | |
| "learning_rate": 0.000424183330792717, | |
| "loss": 1.1723, | |
| "step": 6130 | |
| }, | |
| { | |
| "epoch": 0.6476793248945147, | |
| "grad_norm": 0.48002129793167114, | |
| "learning_rate": 0.0004219238430582621, | |
| "loss": 1.1676, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 0.6487341772151899, | |
| "grad_norm": 0.3844591975212097, | |
| "learning_rate": 0.0004196680315773408, | |
| "loss": 1.1788, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.6497890295358649, | |
| "grad_norm": 0.3658396899700165, | |
| "learning_rate": 0.00041741592162742214, | |
| "loss": 1.1715, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 0.6508438818565401, | |
| "grad_norm": 0.34545275568962097, | |
| "learning_rate": 0.0004151675384444978, | |
| "loss": 1.1503, | |
| "step": 6170 | |
| }, | |
| { | |
| "epoch": 0.6518987341772152, | |
| "grad_norm": 0.3446422219276428, | |
| "learning_rate": 0.00041292290722279914, | |
| "loss": 1.1725, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 0.6529535864978903, | |
| "grad_norm": 0.3521484434604645, | |
| "learning_rate": 0.00041068205311451517, | |
| "loss": 1.174, | |
| "step": 6190 | |
| }, | |
| { | |
| "epoch": 0.6540084388185654, | |
| "grad_norm": 0.3657844662666321, | |
| "learning_rate": 0.00040844500122951026, | |
| "loss": 1.1691, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.6550632911392406, | |
| "grad_norm": 0.38760536909103394, | |
| "learning_rate": 0.00040621177663504313, | |
| "loss": 1.1634, | |
| "step": 6210 | |
| }, | |
| { | |
| "epoch": 0.6561181434599156, | |
| "grad_norm": 0.4080621898174286, | |
| "learning_rate": 0.00040398240435548583, | |
| "loss": 1.1728, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 0.6571729957805907, | |
| "grad_norm": 0.37887656688690186, | |
| "learning_rate": 0.00040175690937204324, | |
| "loss": 1.1616, | |
| "step": 6230 | |
| }, | |
| { | |
| "epoch": 0.6582278481012658, | |
| "grad_norm": 0.3961949050426483, | |
| "learning_rate": 0.00039953531662247343, | |
| "loss": 1.1635, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 0.6592827004219409, | |
| "grad_norm": 0.36524301767349243, | |
| "learning_rate": 0.0003973176510008075, | |
| "loss": 1.1672, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.6603375527426161, | |
| "grad_norm": 0.393848717212677, | |
| "learning_rate": 0.00039510393735707233, | |
| "loss": 1.1622, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 0.6613924050632911, | |
| "grad_norm": 0.36277323961257935, | |
| "learning_rate": 0.00039289420049700986, | |
| "loss": 1.1707, | |
| "step": 6270 | |
| }, | |
| { | |
| "epoch": 0.6624472573839663, | |
| "grad_norm": 0.3623575270175934, | |
| "learning_rate": 0.0003906884651818006, | |
| "loss": 1.1691, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 0.6635021097046413, | |
| "grad_norm": 0.4982074797153473, | |
| "learning_rate": 0.00038848675612778577, | |
| "loss": 1.1671, | |
| "step": 6290 | |
| }, | |
| { | |
| "epoch": 0.6645569620253164, | |
| "grad_norm": 0.35465696454048157, | |
| "learning_rate": 0.00038628909800619046, | |
| "loss": 1.1594, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.6656118143459916, | |
| "grad_norm": 0.37563908100128174, | |
| "learning_rate": 0.0003840955154428467, | |
| "loss": 1.1604, | |
| "step": 6310 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.3676312565803528, | |
| "learning_rate": 0.00038190603301791864, | |
| "loss": 1.1569, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 0.6677215189873418, | |
| "grad_norm": 0.42015132308006287, | |
| "learning_rate": 0.0003797206752656258, | |
| "loss": 1.1656, | |
| "step": 6330 | |
| }, | |
| { | |
| "epoch": 0.6687763713080169, | |
| "grad_norm": 0.36727169156074524, | |
| "learning_rate": 0.0003775394666739688, | |
| "loss": 1.1577, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 0.669831223628692, | |
| "grad_norm": 0.37604963779449463, | |
| "learning_rate": 0.00037536243168445507, | |
| "loss": 1.1656, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.6708860759493671, | |
| "grad_norm": 0.3591715097427368, | |
| "learning_rate": 0.0003731895946918246, | |
| "loss": 1.1637, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 0.6719409282700421, | |
| "grad_norm": 0.4027197062969208, | |
| "learning_rate": 0.0003710209800437769, | |
| "loss": 1.1454, | |
| "step": 6370 | |
| }, | |
| { | |
| "epoch": 0.6729957805907173, | |
| "grad_norm": 0.34489601850509644, | |
| "learning_rate": 0.00036885661204069767, | |
| "loss": 1.1495, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 0.6740506329113924, | |
| "grad_norm": 0.3564709722995758, | |
| "learning_rate": 0.0003666965149353878, | |
| "loss": 1.1643, | |
| "step": 6390 | |
| }, | |
| { | |
| "epoch": 0.6751054852320675, | |
| "grad_norm": 0.4152984917163849, | |
| "learning_rate": 0.0003645407129327898, | |
| "loss": 1.1621, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.6761603375527426, | |
| "grad_norm": 0.407699316740036, | |
| "learning_rate": 0.00036238923018971783, | |
| "loss": 1.1588, | |
| "step": 6410 | |
| }, | |
| { | |
| "epoch": 0.6772151898734177, | |
| "grad_norm": 0.3636753261089325, | |
| "learning_rate": 0.0003602420908145865, | |
| "loss": 1.1499, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 0.6782700421940928, | |
| "grad_norm": 0.36881223320961, | |
| "learning_rate": 0.00035809931886714093, | |
| "loss": 1.1594, | |
| "step": 6430 | |
| }, | |
| { | |
| "epoch": 0.679324894514768, | |
| "grad_norm": 0.36326852440834045, | |
| "learning_rate": 0.00035596093835818683, | |
| "loss": 1.1439, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 0.680379746835443, | |
| "grad_norm": 0.35837841033935547, | |
| "learning_rate": 0.00035382697324932245, | |
| "loss": 1.1491, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.6814345991561181, | |
| "grad_norm": 0.43497446179389954, | |
| "learning_rate": 0.00035169744745266866, | |
| "loss": 1.1627, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 0.6824894514767933, | |
| "grad_norm": 0.379119336605072, | |
| "learning_rate": 0.0003495723848306017, | |
| "loss": 1.1577, | |
| "step": 6470 | |
| }, | |
| { | |
| "epoch": 0.6835443037974683, | |
| "grad_norm": 0.3709147274494171, | |
| "learning_rate": 0.0003474518091954859, | |
| "loss": 1.1617, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 0.6845991561181435, | |
| "grad_norm": 0.36736902594566345, | |
| "learning_rate": 0.0003453357443094068, | |
| "loss": 1.1582, | |
| "step": 6490 | |
| }, | |
| { | |
| "epoch": 0.6856540084388185, | |
| "grad_norm": 0.391516774892807, | |
| "learning_rate": 0.00034322421388390456, | |
| "loss": 1.171, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.6867088607594937, | |
| "grad_norm": 0.3488520681858063, | |
| "learning_rate": 0.0003411172415797087, | |
| "loss": 1.1706, | |
| "step": 6510 | |
| }, | |
| { | |
| "epoch": 0.6877637130801688, | |
| "grad_norm": 0.36531731486320496, | |
| "learning_rate": 0.0003390148510064727, | |
| "loss": 1.1581, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 0.6888185654008439, | |
| "grad_norm": 0.37665581703186035, | |
| "learning_rate": 0.0003369170657225094, | |
| "loss": 1.1484, | |
| "step": 6530 | |
| }, | |
| { | |
| "epoch": 0.689873417721519, | |
| "grad_norm": 0.37849798798561096, | |
| "learning_rate": 0.0003348239092345275, | |
| "loss": 1.1588, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 0.6909282700421941, | |
| "grad_norm": 0.38719886541366577, | |
| "learning_rate": 0.0003327354049973672, | |
| "loss": 1.1545, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.6919831223628692, | |
| "grad_norm": 0.41207027435302734, | |
| "learning_rate": 0.00033065157641373847, | |
| "loss": 1.1541, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 0.6930379746835443, | |
| "grad_norm": 0.42637428641319275, | |
| "learning_rate": 0.0003285724468339576, | |
| "loss": 1.1508, | |
| "step": 6570 | |
| }, | |
| { | |
| "epoch": 0.6940928270042194, | |
| "grad_norm": 0.3876138925552368, | |
| "learning_rate": 0.00032649803955568755, | |
| "loss": 1.1502, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 0.6951476793248945, | |
| "grad_norm": 0.3820565938949585, | |
| "learning_rate": 0.00032442837782367434, | |
| "loss": 1.1524, | |
| "step": 6590 | |
| }, | |
| { | |
| "epoch": 0.6962025316455697, | |
| "grad_norm": 0.3652574419975281, | |
| "learning_rate": 0.0003223634848294883, | |
| "loss": 1.1583, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.6972573839662447, | |
| "grad_norm": 0.4173029363155365, | |
| "learning_rate": 0.00032030338371126374, | |
| "loss": 1.1439, | |
| "step": 6610 | |
| }, | |
| { | |
| "epoch": 0.6983122362869199, | |
| "grad_norm": 0.362628310918808, | |
| "learning_rate": 0.0003182480975534395, | |
| "loss": 1.1546, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 0.6993670886075949, | |
| "grad_norm": 0.4287622272968292, | |
| "learning_rate": 0.00031619764938650057, | |
| "loss": 1.1467, | |
| "step": 6630 | |
| }, | |
| { | |
| "epoch": 0.70042194092827, | |
| "grad_norm": 0.3689073324203491, | |
| "learning_rate": 0.0003141520621867197, | |
| "loss": 1.1522, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 0.7014767932489452, | |
| "grad_norm": 0.38702285289764404, | |
| "learning_rate": 0.00031211135887590074, | |
| "loss": 1.1454, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.7025316455696202, | |
| "grad_norm": 0.3564895689487457, | |
| "learning_rate": 0.0003100755623211205, | |
| "loss": 1.1472, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 0.7035864978902954, | |
| "grad_norm": 0.36778292059898376, | |
| "learning_rate": 0.0003080446953344735, | |
| "loss": 1.1505, | |
| "step": 6670 | |
| }, | |
| { | |
| "epoch": 0.7046413502109705, | |
| "grad_norm": 0.35331690311431885, | |
| "learning_rate": 0.00030601878067281575, | |
| "loss": 1.1511, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 0.7056962025316456, | |
| "grad_norm": 0.36919105052948, | |
| "learning_rate": 0.00030399784103751044, | |
| "loss": 1.144, | |
| "step": 6690 | |
| }, | |
| { | |
| "epoch": 0.7067510548523207, | |
| "grad_norm": 0.35663625597953796, | |
| "learning_rate": 0.000301981899074173, | |
| "loss": 1.1471, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.7078059071729957, | |
| "grad_norm": 0.3756558299064636, | |
| "learning_rate": 0.0002999709773724171, | |
| "loss": 1.1485, | |
| "step": 6710 | |
| }, | |
| { | |
| "epoch": 0.7088607594936709, | |
| "grad_norm": 0.3560740053653717, | |
| "learning_rate": 0.00029796509846560294, | |
| "loss": 1.1377, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 0.709915611814346, | |
| "grad_norm": 0.3501749336719513, | |
| "learning_rate": 0.0002959642848305828, | |
| "loss": 1.1435, | |
| "step": 6730 | |
| }, | |
| { | |
| "epoch": 0.7109704641350211, | |
| "grad_norm": 0.36983656883239746, | |
| "learning_rate": 0.00029396855888745045, | |
| "loss": 1.15, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 0.7120253164556962, | |
| "grad_norm": 0.4085451662540436, | |
| "learning_rate": 0.0002919779429992895, | |
| "loss": 1.1543, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.7130801687763713, | |
| "grad_norm": 0.37220215797424316, | |
| "learning_rate": 0.0002899924594719231, | |
| "loss": 1.1488, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 0.7141350210970464, | |
| "grad_norm": 0.3557329475879669, | |
| "learning_rate": 0.00028801213055366335, | |
| "loss": 1.1492, | |
| "step": 6770 | |
| }, | |
| { | |
| "epoch": 0.7151898734177216, | |
| "grad_norm": 0.37931346893310547, | |
| "learning_rate": 0.00028603697843506315, | |
| "loss": 1.1465, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 0.7162447257383966, | |
| "grad_norm": 0.37470945715904236, | |
| "learning_rate": 0.0002840670252486662, | |
| "loss": 1.1506, | |
| "step": 6790 | |
| }, | |
| { | |
| "epoch": 0.7172995780590717, | |
| "grad_norm": 0.37293675541877747, | |
| "learning_rate": 0.00028210229306876, | |
| "loss": 1.1384, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.7183544303797469, | |
| "grad_norm": 0.3550741374492645, | |
| "learning_rate": 0.0002801428039111279, | |
| "loss": 1.1451, | |
| "step": 6810 | |
| }, | |
| { | |
| "epoch": 0.7194092827004219, | |
| "grad_norm": 0.37655240297317505, | |
| "learning_rate": 0.00027818857973280274, | |
| "loss": 1.1438, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 0.7204641350210971, | |
| "grad_norm": 0.3859023451805115, | |
| "learning_rate": 0.0002762396424318206, | |
| "loss": 1.1437, | |
| "step": 6830 | |
| }, | |
| { | |
| "epoch": 0.7215189873417721, | |
| "grad_norm": 0.36085501313209534, | |
| "learning_rate": 0.00027429601384697526, | |
| "loss": 1.1343, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 0.7225738396624473, | |
| "grad_norm": 0.37223494052886963, | |
| "learning_rate": 0.00027235771575757466, | |
| "loss": 1.1437, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.7236286919831224, | |
| "grad_norm": 0.38057178258895874, | |
| "learning_rate": 0.0002704247698831951, | |
| "loss": 1.1365, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 0.7246835443037974, | |
| "grad_norm": 0.36014029383659363, | |
| "learning_rate": 0.0002684971978834389, | |
| "loss": 1.135, | |
| "step": 6870 | |
| }, | |
| { | |
| "epoch": 0.7257383966244726, | |
| "grad_norm": 0.34835484623908997, | |
| "learning_rate": 0.0002665750213576914, | |
| "loss": 1.1451, | |
| "step": 6880 | |
| }, | |
| { | |
| "epoch": 0.7267932489451476, | |
| "grad_norm": 0.36837393045425415, | |
| "learning_rate": 0.0002646582618448794, | |
| "loss": 1.1347, | |
| "step": 6890 | |
| }, | |
| { | |
| "epoch": 0.7278481012658228, | |
| "grad_norm": 0.37352582812309265, | |
| "learning_rate": 0.00026274694082322896, | |
| "loss": 1.1332, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.7289029535864979, | |
| "grad_norm": 0.42627763748168945, | |
| "learning_rate": 0.0002608410797100255, | |
| "loss": 1.1583, | |
| "step": 6910 | |
| }, | |
| { | |
| "epoch": 0.729957805907173, | |
| "grad_norm": 0.41890236735343933, | |
| "learning_rate": 0.0002589406998613733, | |
| "loss": 1.1393, | |
| "step": 6920 | |
| }, | |
| { | |
| "epoch": 0.7310126582278481, | |
| "grad_norm": 0.3742920756340027, | |
| "learning_rate": 0.0002570458225719567, | |
| "loss": 1.1431, | |
| "step": 6930 | |
| }, | |
| { | |
| "epoch": 0.7320675105485233, | |
| "grad_norm": 0.38046887516975403, | |
| "learning_rate": 0.00025515646907480074, | |
| "loss": 1.1469, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 0.7331223628691983, | |
| "grad_norm": 0.3562321364879608, | |
| "learning_rate": 0.00025327266054103395, | |
| "loss": 1.1443, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.7341772151898734, | |
| "grad_norm": 0.35831841826438904, | |
| "learning_rate": 0.0002513944180796509, | |
| "loss": 1.1379, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 0.7352320675105485, | |
| "grad_norm": 0.3924483358860016, | |
| "learning_rate": 0.0002495217627372752, | |
| "loss": 1.1327, | |
| "step": 6970 | |
| }, | |
| { | |
| "epoch": 0.7362869198312236, | |
| "grad_norm": 0.36096253991127014, | |
| "learning_rate": 0.0002476547154979248, | |
| "loss": 1.1359, | |
| "step": 6980 | |
| }, | |
| { | |
| "epoch": 0.7373417721518988, | |
| "grad_norm": 0.37537628412246704, | |
| "learning_rate": 0.00024579329728277534, | |
| "loss": 1.1359, | |
| "step": 6990 | |
| }, | |
| { | |
| "epoch": 0.7383966244725738, | |
| "grad_norm": 0.37565308809280396, | |
| "learning_rate": 0.00024393752894992708, | |
| "loss": 1.1476, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.739451476793249, | |
| "grad_norm": 0.35901403427124023, | |
| "learning_rate": 0.00024208743129417004, | |
| "loss": 1.1208, | |
| "step": 7010 | |
| }, | |
| { | |
| "epoch": 0.740506329113924, | |
| "grad_norm": 0.39581504464149475, | |
| "learning_rate": 0.00024024302504675206, | |
| "loss": 1.1346, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 0.7415611814345991, | |
| "grad_norm": 0.3442055881023407, | |
| "learning_rate": 0.0002384043308751454, | |
| "loss": 1.1373, | |
| "step": 7030 | |
| }, | |
| { | |
| "epoch": 0.7426160337552743, | |
| "grad_norm": 0.3714121878147125, | |
| "learning_rate": 0.00023657136938281653, | |
| "loss": 1.1447, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 0.7436708860759493, | |
| "grad_norm": 0.38861745595932007, | |
| "learning_rate": 0.00023474416110899377, | |
| "loss": 1.1352, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.7447257383966245, | |
| "grad_norm": 0.36330005526542664, | |
| "learning_rate": 0.00023292272652843807, | |
| "loss": 1.1367, | |
| "step": 7060 | |
| }, | |
| { | |
| "epoch": 0.7457805907172996, | |
| "grad_norm": 0.3554536998271942, | |
| "learning_rate": 0.00023110708605121317, | |
| "loss": 1.143, | |
| "step": 7070 | |
| }, | |
| { | |
| "epoch": 0.7468354430379747, | |
| "grad_norm": 0.37097129225730896, | |
| "learning_rate": 0.00022929726002245728, | |
| "loss": 1.1393, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 0.7478902953586498, | |
| "grad_norm": 0.38714399933815, | |
| "learning_rate": 0.00022749326872215472, | |
| "loss": 1.1365, | |
| "step": 7090 | |
| }, | |
| { | |
| "epoch": 0.7489451476793249, | |
| "grad_norm": 0.3776571452617645, | |
| "learning_rate": 0.0002256951323649087, | |
| "loss": 1.1327, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.3625503182411194, | |
| "learning_rate": 0.00022390287109971547, | |
| "loss": 1.145, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 0.7510548523206751, | |
| "grad_norm": 0.41759389638900757, | |
| "learning_rate": 0.00022211650500973746, | |
| "loss": 1.1381, | |
| "step": 7120 | |
| }, | |
| { | |
| "epoch": 0.7521097046413502, | |
| "grad_norm": 0.3729211390018463, | |
| "learning_rate": 0.0002203360541120789, | |
| "loss": 1.1454, | |
| "step": 7130 | |
| }, | |
| { | |
| "epoch": 0.7531645569620253, | |
| "grad_norm": 0.3729153573513031, | |
| "learning_rate": 0.00021856153835756164, | |
| "loss": 1.1291, | |
| "step": 7140 | |
| }, | |
| { | |
| "epoch": 0.7542194092827004, | |
| "grad_norm": 0.39670905470848083, | |
| "learning_rate": 0.00021679297763050104, | |
| "loss": 1.1275, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.7552742616033755, | |
| "grad_norm": 0.3594678044319153, | |
| "learning_rate": 0.0002150303917484834, | |
| "loss": 1.1357, | |
| "step": 7160 | |
| }, | |
| { | |
| "epoch": 0.7563291139240507, | |
| "grad_norm": 0.40017199516296387, | |
| "learning_rate": 0.0002132738004621446, | |
| "loss": 1.1439, | |
| "step": 7170 | |
| }, | |
| { | |
| "epoch": 0.7573839662447257, | |
| "grad_norm": 0.3732672333717346, | |
| "learning_rate": 0.00021152322345494763, | |
| "loss": 1.1231, | |
| "step": 7180 | |
| }, | |
| { | |
| "epoch": 0.7584388185654009, | |
| "grad_norm": 0.39866897463798523, | |
| "learning_rate": 0.00020977868034296253, | |
| "loss": 1.1264, | |
| "step": 7190 | |
| }, | |
| { | |
| "epoch": 0.759493670886076, | |
| "grad_norm": 0.35106202960014343, | |
| "learning_rate": 0.00020804019067464667, | |
| "loss": 1.1298, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.760548523206751, | |
| "grad_norm": 0.4421609342098236, | |
| "learning_rate": 0.00020630777393062575, | |
| "loss": 1.134, | |
| "step": 7210 | |
| }, | |
| { | |
| "epoch": 0.7616033755274262, | |
| "grad_norm": 0.40509048104286194, | |
| "learning_rate": 0.00020458144952347523, | |
| "loss": 1.1451, | |
| "step": 7220 | |
| }, | |
| { | |
| "epoch": 0.7626582278481012, | |
| "grad_norm": 0.3772280514240265, | |
| "learning_rate": 0.00020286123679750314, | |
| "loss": 1.1437, | |
| "step": 7230 | |
| }, | |
| { | |
| "epoch": 0.7637130801687764, | |
| "grad_norm": 0.3716863691806793, | |
| "learning_rate": 0.00020114715502853292, | |
| "loss": 1.1409, | |
| "step": 7240 | |
| }, | |
| { | |
| "epoch": 0.7647679324894515, | |
| "grad_norm": 0.41956421732902527, | |
| "learning_rate": 0.0001994392234236878, | |
| "loss": 1.1332, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.7658227848101266, | |
| "grad_norm": 0.3653375804424286, | |
| "learning_rate": 0.0001977374611211754, | |
| "loss": 1.1337, | |
| "step": 7260 | |
| }, | |
| { | |
| "epoch": 0.7668776371308017, | |
| "grad_norm": 0.3775090277194977, | |
| "learning_rate": 0.00019604188719007313, | |
| "loss": 1.1317, | |
| "step": 7270 | |
| }, | |
| { | |
| "epoch": 0.7679324894514767, | |
| "grad_norm": 0.3718429505825043, | |
| "learning_rate": 0.00019435252063011504, | |
| "loss": 1.133, | |
| "step": 7280 | |
| }, | |
| { | |
| "epoch": 0.7689873417721519, | |
| "grad_norm": 0.37331250309944153, | |
| "learning_rate": 0.0001926693803714779, | |
| "loss": 1.1447, | |
| "step": 7290 | |
| }, | |
| { | |
| "epoch": 0.770042194092827, | |
| "grad_norm": 0.3802056312561035, | |
| "learning_rate": 0.00019099248527457068, | |
| "loss": 1.1459, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.7710970464135021, | |
| "grad_norm": 0.36767441034317017, | |
| "learning_rate": 0.0001893218541298216, | |
| "loss": 1.1268, | |
| "step": 7310 | |
| }, | |
| { | |
| "epoch": 0.7721518987341772, | |
| "grad_norm": 0.34967729449272156, | |
| "learning_rate": 0.00018765750565746827, | |
| "loss": 1.1214, | |
| "step": 7320 | |
| }, | |
| { | |
| "epoch": 0.7732067510548524, | |
| "grad_norm": 0.3683644235134125, | |
| "learning_rate": 0.00018599945850734812, | |
| "loss": 1.1291, | |
| "step": 7330 | |
| }, | |
| { | |
| "epoch": 0.7742616033755274, | |
| "grad_norm": 0.3712475001811981, | |
| "learning_rate": 0.00018434773125868895, | |
| "loss": 1.1222, | |
| "step": 7340 | |
| }, | |
| { | |
| "epoch": 0.7753164556962026, | |
| "grad_norm": 0.3514195680618286, | |
| "learning_rate": 0.00018270234241990108, | |
| "loss": 1.1235, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.7763713080168776, | |
| "grad_norm": 0.37643975019454956, | |
| "learning_rate": 0.0001810633104283698, | |
| "loss": 1.1306, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 0.7774261603375527, | |
| "grad_norm": 0.38709020614624023, | |
| "learning_rate": 0.0001794306536502492, | |
| "loss": 1.1333, | |
| "step": 7370 | |
| }, | |
| { | |
| "epoch": 0.7784810126582279, | |
| "grad_norm": 0.3852018117904663, | |
| "learning_rate": 0.0001778043903802555, | |
| "loss": 1.1191, | |
| "step": 7380 | |
| }, | |
| { | |
| "epoch": 0.7795358649789029, | |
| "grad_norm": 0.36321789026260376, | |
| "learning_rate": 0.0001761845388414627, | |
| "loss": 1.128, | |
| "step": 7390 | |
| }, | |
| { | |
| "epoch": 0.7805907172995781, | |
| "grad_norm": 0.3746873140335083, | |
| "learning_rate": 0.00017457111718509831, | |
| "loss": 1.1372, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.7816455696202531, | |
| "grad_norm": 0.3483794033527374, | |
| "learning_rate": 0.00017296414349033976, | |
| "loss": 1.1182, | |
| "step": 7410 | |
| }, | |
| { | |
| "epoch": 0.7827004219409283, | |
| "grad_norm": 0.35910797119140625, | |
| "learning_rate": 0.00017136363576411172, | |
| "loss": 1.1232, | |
| "step": 7420 | |
| }, | |
| { | |
| "epoch": 0.7837552742616034, | |
| "grad_norm": 0.3550094664096832, | |
| "learning_rate": 0.00016976961194088526, | |
| "loss": 1.1108, | |
| "step": 7430 | |
| }, | |
| { | |
| "epoch": 0.7848101265822784, | |
| "grad_norm": 0.3682705760002136, | |
| "learning_rate": 0.00016818208988247533, | |
| "loss": 1.1147, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 0.7858649789029536, | |
| "grad_norm": 0.35992518067359924, | |
| "learning_rate": 0.0001666010873778419, | |
| "loss": 1.1181, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.7869198312236287, | |
| "grad_norm": 0.3707635998725891, | |
| "learning_rate": 0.00016502662214289, | |
| "loss": 1.1287, | |
| "step": 7460 | |
| }, | |
| { | |
| "epoch": 0.7879746835443038, | |
| "grad_norm": 0.3830517530441284, | |
| "learning_rate": 0.00016345871182027124, | |
| "loss": 1.1204, | |
| "step": 7470 | |
| }, | |
| { | |
| "epoch": 0.7890295358649789, | |
| "grad_norm": 0.36266350746154785, | |
| "learning_rate": 0.00016189737397918653, | |
| "loss": 1.1258, | |
| "step": 7480 | |
| }, | |
| { | |
| "epoch": 0.790084388185654, | |
| "grad_norm": 0.363208144903183, | |
| "learning_rate": 0.0001603426261151884, | |
| "loss": 1.1323, | |
| "step": 7490 | |
| }, | |
| { | |
| "epoch": 0.7911392405063291, | |
| "grad_norm": 0.3821084201335907, | |
| "learning_rate": 0.00015879448564998648, | |
| "loss": 1.1273, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.7921940928270043, | |
| "grad_norm": 0.3608359396457672, | |
| "learning_rate": 0.0001572529699312501, | |
| "loss": 1.133, | |
| "step": 7510 | |
| }, | |
| { | |
| "epoch": 0.7932489451476793, | |
| "grad_norm": 0.3614010810852051, | |
| "learning_rate": 0.0001557180962324158, | |
| "loss": 1.1172, | |
| "step": 7520 | |
| }, | |
| { | |
| "epoch": 0.7943037974683544, | |
| "grad_norm": 0.3900236487388611, | |
| "learning_rate": 0.00015418988175249282, | |
| "loss": 1.1198, | |
| "step": 7530 | |
| }, | |
| { | |
| "epoch": 0.7953586497890295, | |
| "grad_norm": 0.40939193964004517, | |
| "learning_rate": 0.00015266834361587063, | |
| "loss": 1.1163, | |
| "step": 7540 | |
| }, | |
| { | |
| "epoch": 0.7964135021097046, | |
| "grad_norm": 0.3722957670688629, | |
| "learning_rate": 0.00015115349887212678, | |
| "loss": 1.1139, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.7974683544303798, | |
| "grad_norm": 0.3719434440135956, | |
| "learning_rate": 0.00014964536449583657, | |
| "loss": 1.119, | |
| "step": 7560 | |
| }, | |
| { | |
| "epoch": 0.7985232067510548, | |
| "grad_norm": 0.37369686365127563, | |
| "learning_rate": 0.00014814395738638195, | |
| "loss": 1.1259, | |
| "step": 7570 | |
| }, | |
| { | |
| "epoch": 0.79957805907173, | |
| "grad_norm": 0.3760931193828583, | |
| "learning_rate": 0.00014664929436776278, | |
| "loss": 1.1248, | |
| "step": 7580 | |
| }, | |
| { | |
| "epoch": 0.8006329113924051, | |
| "grad_norm": 0.3629474937915802, | |
| "learning_rate": 0.00014516139218840788, | |
| "loss": 1.1123, | |
| "step": 7590 | |
| }, | |
| { | |
| "epoch": 0.8016877637130801, | |
| "grad_norm": 0.3516145348548889, | |
| "learning_rate": 0.00014368026752098782, | |
| "loss": 1.1198, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.8027426160337553, | |
| "grad_norm": 0.36271005868911743, | |
| "learning_rate": 0.00014220593696222768, | |
| "loss": 1.132, | |
| "step": 7610 | |
| }, | |
| { | |
| "epoch": 0.8037974683544303, | |
| "grad_norm": 0.34938865900039673, | |
| "learning_rate": 0.00014073841703272092, | |
| "loss": 1.1069, | |
| "step": 7620 | |
| }, | |
| { | |
| "epoch": 0.8048523206751055, | |
| "grad_norm": 0.3703489601612091, | |
| "learning_rate": 0.00013927772417674558, | |
| "loss": 1.1308, | |
| "step": 7630 | |
| }, | |
| { | |
| "epoch": 0.8059071729957806, | |
| "grad_norm": 0.3835078477859497, | |
| "learning_rate": 0.00013782387476207788, | |
| "loss": 1.1225, | |
| "step": 7640 | |
| }, | |
| { | |
| "epoch": 0.8069620253164557, | |
| "grad_norm": 0.3790573477745056, | |
| "learning_rate": 0.00013637688507981064, | |
| "loss": 1.1379, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.8080168776371308, | |
| "grad_norm": 0.3718129098415375, | |
| "learning_rate": 0.0001349367713441697, | |
| "loss": 1.1152, | |
| "step": 7660 | |
| }, | |
| { | |
| "epoch": 0.8090717299578059, | |
| "grad_norm": 0.36849159002304077, | |
| "learning_rate": 0.0001335035496923326, | |
| "loss": 1.122, | |
| "step": 7670 | |
| }, | |
| { | |
| "epoch": 0.810126582278481, | |
| "grad_norm": 0.3801332116127014, | |
| "learning_rate": 0.0001320772361842478, | |
| "loss": 1.1326, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.8111814345991561, | |
| "grad_norm": 0.3597279191017151, | |
| "learning_rate": 0.00013065784680245442, | |
| "loss": 1.1092, | |
| "step": 7690 | |
| }, | |
| { | |
| "epoch": 0.8122362869198312, | |
| "grad_norm": 0.3624008893966675, | |
| "learning_rate": 0.00012924539745190402, | |
| "loss": 1.1181, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.8132911392405063, | |
| "grad_norm": 0.38765400648117065, | |
| "learning_rate": 0.0001278399039597809, | |
| "loss": 1.1156, | |
| "step": 7710 | |
| }, | |
| { | |
| "epoch": 0.8143459915611815, | |
| "grad_norm": 0.3619614243507385, | |
| "learning_rate": 0.0001264413820753261, | |
| "loss": 1.1205, | |
| "step": 7720 | |
| }, | |
| { | |
| "epoch": 0.8154008438818565, | |
| "grad_norm": 0.36417073011398315, | |
| "learning_rate": 0.00012504984746966003, | |
| "loss": 1.126, | |
| "step": 7730 | |
| }, | |
| { | |
| "epoch": 0.8164556962025317, | |
| "grad_norm": 0.35633671283721924, | |
| "learning_rate": 0.00012366531573560754, | |
| "loss": 1.1255, | |
| "step": 7740 | |
| }, | |
| { | |
| "epoch": 0.8175105485232067, | |
| "grad_norm": 0.3816451132297516, | |
| "learning_rate": 0.00012228780238752264, | |
| "loss": 1.1127, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.8185654008438819, | |
| "grad_norm": 0.3993898630142212, | |
| "learning_rate": 0.00012091732286111514, | |
| "loss": 1.1123, | |
| "step": 7760 | |
| }, | |
| { | |
| "epoch": 0.819620253164557, | |
| "grad_norm": 0.36348336935043335, | |
| "learning_rate": 0.00011955389251327737, | |
| "loss": 1.1179, | |
| "step": 7770 | |
| }, | |
| { | |
| "epoch": 0.820675105485232, | |
| "grad_norm": 0.3756120502948761, | |
| "learning_rate": 0.00011819752662191197, | |
| "loss": 1.1125, | |
| "step": 7780 | |
| }, | |
| { | |
| "epoch": 0.8217299578059072, | |
| "grad_norm": 0.3802875876426697, | |
| "learning_rate": 0.00011684824038576115, | |
| "loss": 1.1294, | |
| "step": 7790 | |
| }, | |
| { | |
| "epoch": 0.8227848101265823, | |
| "grad_norm": 0.3661904036998749, | |
| "learning_rate": 0.00011550604892423593, | |
| "loss": 1.1176, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.8238396624472574, | |
| "grad_norm": 0.3550266623497009, | |
| "learning_rate": 0.0001141709672772471, | |
| "loss": 1.1166, | |
| "step": 7810 | |
| }, | |
| { | |
| "epoch": 0.8248945147679325, | |
| "grad_norm": 0.36318251490592957, | |
| "learning_rate": 0.00011284301040503625, | |
| "loss": 1.1232, | |
| "step": 7820 | |
| }, | |
| { | |
| "epoch": 0.8259493670886076, | |
| "grad_norm": 0.35563451051712036, | |
| "learning_rate": 0.0001115221931880088, | |
| "loss": 1.116, | |
| "step": 7830 | |
| }, | |
| { | |
| "epoch": 0.8270042194092827, | |
| "grad_norm": 0.3556043803691864, | |
| "learning_rate": 0.00011020853042656648, | |
| "loss": 1.0999, | |
| "step": 7840 | |
| }, | |
| { | |
| "epoch": 0.8280590717299579, | |
| "grad_norm": 0.36785390973091125, | |
| "learning_rate": 0.000108902036840942, | |
| "loss": 1.1088, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.8291139240506329, | |
| "grad_norm": 0.34999290108680725, | |
| "learning_rate": 0.00010760272707103389, | |
| "loss": 1.0979, | |
| "step": 7860 | |
| }, | |
| { | |
| "epoch": 0.830168776371308, | |
| "grad_norm": 0.3790420889854431, | |
| "learning_rate": 0.00010631061567624259, | |
| "loss": 1.1117, | |
| "step": 7870 | |
| }, | |
| { | |
| "epoch": 0.8312236286919831, | |
| "grad_norm": 0.3649984896183014, | |
| "learning_rate": 0.00010502571713530706, | |
| "loss": 1.1302, | |
| "step": 7880 | |
| }, | |
| { | |
| "epoch": 0.8322784810126582, | |
| "grad_norm": 0.3923533260822296, | |
| "learning_rate": 0.00010374804584614308, | |
| "loss": 1.1276, | |
| "step": 7890 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 0.38636764883995056, | |
| "learning_rate": 0.00010247761612568129, | |
| "loss": 1.1056, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.8343881856540084, | |
| "grad_norm": 0.38178226351737976, | |
| "learning_rate": 0.0001012144422097069, | |
| "loss": 1.1166, | |
| "step": 7910 | |
| }, | |
| { | |
| "epoch": 0.8354430379746836, | |
| "grad_norm": 0.38185352087020874, | |
| "learning_rate": 9.995853825270052e-05, | |
| "loss": 1.1147, | |
| "step": 7920 | |
| }, | |
| { | |
| "epoch": 0.8364978902953587, | |
| "grad_norm": 0.3619385063648224, | |
| "learning_rate": 9.870991832767919e-05, | |
| "loss": 1.1225, | |
| "step": 7930 | |
| }, | |
| { | |
| "epoch": 0.8375527426160337, | |
| "grad_norm": 0.3500303328037262, | |
| "learning_rate": 9.746859642603884e-05, | |
| "loss": 1.1175, | |
| "step": 7940 | |
| }, | |
| { | |
| "epoch": 0.8386075949367089, | |
| "grad_norm": 0.3563097417354584, | |
| "learning_rate": 9.623458645739755e-05, | |
| "loss": 1.1219, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.8396624472573839, | |
| "grad_norm": 0.35280612111091614, | |
| "learning_rate": 9.50079022494395e-05, | |
| "loss": 1.1071, | |
| "step": 7960 | |
| }, | |
| { | |
| "epoch": 0.8407172995780591, | |
| "grad_norm": 0.3701493442058563, | |
| "learning_rate": 9.378855754776028e-05, | |
| "loss": 1.1076, | |
| "step": 7970 | |
| }, | |
| { | |
| "epoch": 0.8417721518987342, | |
| "grad_norm": 0.36128586530685425, | |
| "learning_rate": 9.257656601571266e-05, | |
| "loss": 1.1075, | |
| "step": 7980 | |
| }, | |
| { | |
| "epoch": 0.8428270042194093, | |
| "grad_norm": 0.3913003206253052, | |
| "learning_rate": 9.137194123425349e-05, | |
| "loss": 1.1176, | |
| "step": 7990 | |
| }, | |
| { | |
| "epoch": 0.8438818565400844, | |
| "grad_norm": 0.3562246859073639, | |
| "learning_rate": 9.017469670179168e-05, | |
| "loss": 1.1125, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8449367088607594, | |
| "grad_norm": 0.38052770495414734, | |
| "learning_rate": 8.898484583403668e-05, | |
| "loss": 1.1189, | |
| "step": 8010 | |
| }, | |
| { | |
| "epoch": 0.8459915611814346, | |
| "grad_norm": 0.36378705501556396, | |
| "learning_rate": 8.780240196384873e-05, | |
| "loss": 1.1116, | |
| "step": 8020 | |
| }, | |
| { | |
| "epoch": 0.8470464135021097, | |
| "grad_norm": 0.3562518358230591, | |
| "learning_rate": 8.662737834108861e-05, | |
| "loss": 1.1104, | |
| "step": 8030 | |
| }, | |
| { | |
| "epoch": 0.8481012658227848, | |
| "grad_norm": 0.37795576453208923, | |
| "learning_rate": 8.545978813246987e-05, | |
| "loss": 1.1219, | |
| "step": 8040 | |
| }, | |
| { | |
| "epoch": 0.8491561181434599, | |
| "grad_norm": 0.3587080240249634, | |
| "learning_rate": 8.429964442141072e-05, | |
| "loss": 1.1039, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.8502109704641351, | |
| "grad_norm": 0.35426101088523865, | |
| "learning_rate": 8.314696020788806e-05, | |
| "loss": 1.1127, | |
| "step": 8060 | |
| }, | |
| { | |
| "epoch": 0.8512658227848101, | |
| "grad_norm": 0.3638628423213959, | |
| "learning_rate": 8.200174840829136e-05, | |
| "loss": 1.1169, | |
| "step": 8070 | |
| }, | |
| { | |
| "epoch": 0.8523206751054853, | |
| "grad_norm": 0.3730154037475586, | |
| "learning_rate": 8.08640218552778e-05, | |
| "loss": 1.119, | |
| "step": 8080 | |
| }, | |
| { | |
| "epoch": 0.8533755274261603, | |
| "grad_norm": 0.3520881235599518, | |
| "learning_rate": 7.973379329762925e-05, | |
| "loss": 1.1072, | |
| "step": 8090 | |
| }, | |
| { | |
| "epoch": 0.8544303797468354, | |
| "grad_norm": 0.3606489300727844, | |
| "learning_rate": 7.861107540010845e-05, | |
| "loss": 1.1025, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.8554852320675106, | |
| "grad_norm": 0.40547868609428406, | |
| "learning_rate": 7.749588074331762e-05, | |
| "loss": 1.118, | |
| "step": 8110 | |
| }, | |
| { | |
| "epoch": 0.8565400843881856, | |
| "grad_norm": 0.36077240109443665, | |
| "learning_rate": 7.63882218235575e-05, | |
| "loss": 1.1092, | |
| "step": 8120 | |
| }, | |
| { | |
| "epoch": 0.8575949367088608, | |
| "grad_norm": 0.3992495834827423, | |
| "learning_rate": 7.528811105268699e-05, | |
| "loss": 1.1092, | |
| "step": 8130 | |
| }, | |
| { | |
| "epoch": 0.8586497890295358, | |
| "grad_norm": 0.3654575049877167, | |
| "learning_rate": 7.41955607579845e-05, | |
| "loss": 1.108, | |
| "step": 8140 | |
| }, | |
| { | |
| "epoch": 0.859704641350211, | |
| "grad_norm": 0.3554239869117737, | |
| "learning_rate": 7.311058318200969e-05, | |
| "loss": 1.1055, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.8607594936708861, | |
| "grad_norm": 0.37257564067840576, | |
| "learning_rate": 7.203319048246599e-05, | |
| "loss": 1.1156, | |
| "step": 8160 | |
| }, | |
| { | |
| "epoch": 0.8618143459915611, | |
| "grad_norm": 0.36394527554512024, | |
| "learning_rate": 7.096339473206471e-05, | |
| "loss": 1.1073, | |
| "step": 8170 | |
| }, | |
| { | |
| "epoch": 0.8628691983122363, | |
| "grad_norm": 0.37236538529396057, | |
| "learning_rate": 6.990120791838953e-05, | |
| "loss": 1.1135, | |
| "step": 8180 | |
| }, | |
| { | |
| "epoch": 0.8639240506329114, | |
| "grad_norm": 0.38898542523384094, | |
| "learning_rate": 6.884664194376233e-05, | |
| "loss": 1.106, | |
| "step": 8190 | |
| }, | |
| { | |
| "epoch": 0.8649789029535865, | |
| "grad_norm": 0.36607620120048523, | |
| "learning_rate": 6.779970862510989e-05, | |
| "loss": 1.1171, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.8660337552742616, | |
| "grad_norm": 0.36071106791496277, | |
| "learning_rate": 6.676041969383107e-05, | |
| "loss": 1.1126, | |
| "step": 8210 | |
| }, | |
| { | |
| "epoch": 0.8670886075949367, | |
| "grad_norm": 0.35975003242492676, | |
| "learning_rate": 6.572878679566605e-05, | |
| "loss": 1.1197, | |
| "step": 8220 | |
| }, | |
| { | |
| "epoch": 0.8681434599156118, | |
| "grad_norm": 0.3584916889667511, | |
| "learning_rate": 6.470482149056509e-05, | |
| "loss": 1.1115, | |
| "step": 8230 | |
| }, | |
| { | |
| "epoch": 0.869198312236287, | |
| "grad_norm": 0.3623376488685608, | |
| "learning_rate": 6.368853525255942e-05, | |
| "loss": 1.1084, | |
| "step": 8240 | |
| }, | |
| { | |
| "epoch": 0.870253164556962, | |
| "grad_norm": 0.36445266008377075, | |
| "learning_rate": 6.267993946963249e-05, | |
| "loss": 1.1233, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.8713080168776371, | |
| "grad_norm": 0.37702205777168274, | |
| "learning_rate": 6.167904544359265e-05, | |
| "loss": 1.1153, | |
| "step": 8260 | |
| }, | |
| { | |
| "epoch": 0.8723628691983122, | |
| "grad_norm": 0.36750108003616333, | |
| "learning_rate": 6.068586438994617e-05, | |
| "loss": 1.1131, | |
| "step": 8270 | |
| }, | |
| { | |
| "epoch": 0.8734177215189873, | |
| "grad_norm": 0.3449283242225647, | |
| "learning_rate": 5.970040743777161e-05, | |
| "loss": 1.0963, | |
| "step": 8280 | |
| }, | |
| { | |
| "epoch": 0.8744725738396625, | |
| "grad_norm": 0.36028075218200684, | |
| "learning_rate": 5.8722685629595454e-05, | |
| "loss": 1.0979, | |
| "step": 8290 | |
| }, | |
| { | |
| "epoch": 0.8755274261603375, | |
| "grad_norm": 0.35771775245666504, | |
| "learning_rate": 5.7752709921267855e-05, | |
| "loss": 1.1218, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.8765822784810127, | |
| "grad_norm": 0.3550027012825012, | |
| "learning_rate": 5.6790491181840294e-05, | |
| "loss": 1.1001, | |
| "step": 8310 | |
| }, | |
| { | |
| "epoch": 0.8776371308016878, | |
| "grad_norm": 0.36124446988105774, | |
| "learning_rate": 5.583604019344354e-05, | |
| "loss": 1.1182, | |
| "step": 8320 | |
| }, | |
| { | |
| "epoch": 0.8786919831223629, | |
| "grad_norm": 0.3725493252277374, | |
| "learning_rate": 5.4889367651167007e-05, | |
| "loss": 1.1114, | |
| "step": 8330 | |
| }, | |
| { | |
| "epoch": 0.879746835443038, | |
| "grad_norm": 0.34998440742492676, | |
| "learning_rate": 5.3950484162938714e-05, | |
| "loss": 1.101, | |
| "step": 8340 | |
| }, | |
| { | |
| "epoch": 0.880801687763713, | |
| "grad_norm": 0.36161261796951294, | |
| "learning_rate": 5.3019400249406686e-05, | |
| "loss": 1.1022, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.8818565400843882, | |
| "grad_norm": 0.36241891980171204, | |
| "learning_rate": 5.209612634382077e-05, | |
| "loss": 1.1013, | |
| "step": 8360 | |
| }, | |
| { | |
| "epoch": 0.8829113924050633, | |
| "grad_norm": 0.35523760318756104, | |
| "learning_rate": 5.118067279191599e-05, | |
| "loss": 1.1102, | |
| "step": 8370 | |
| }, | |
| { | |
| "epoch": 0.8839662447257384, | |
| "grad_norm": 0.3642718195915222, | |
| "learning_rate": 5.0273049851796205e-05, | |
| "loss": 1.1132, | |
| "step": 8380 | |
| }, | |
| { | |
| "epoch": 0.8850210970464135, | |
| "grad_norm": 0.36435121297836304, | |
| "learning_rate": 4.9373267693819805e-05, | |
| "loss": 1.105, | |
| "step": 8390 | |
| }, | |
| { | |
| "epoch": 0.8860759493670886, | |
| "grad_norm": 0.3773235082626343, | |
| "learning_rate": 4.848133640048513e-05, | |
| "loss": 1.1051, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.8871308016877637, | |
| "grad_norm": 0.3574071228504181, | |
| "learning_rate": 4.75972659663178e-05, | |
| "loss": 1.1247, | |
| "step": 8410 | |
| }, | |
| { | |
| "epoch": 0.8881856540084389, | |
| "grad_norm": 0.3590472638607025, | |
| "learning_rate": 4.672106629775882e-05, | |
| "loss": 1.1096, | |
| "step": 8420 | |
| }, | |
| { | |
| "epoch": 0.8892405063291139, | |
| "grad_norm": 0.3647381365299225, | |
| "learning_rate": 4.585274721305333e-05, | |
| "loss": 1.1088, | |
| "step": 8430 | |
| }, | |
| { | |
| "epoch": 0.890295358649789, | |
| "grad_norm": 0.3700782060623169, | |
| "learning_rate": 4.4992318442140575e-05, | |
| "loss": 1.1025, | |
| "step": 8440 | |
| }, | |
| { | |
| "epoch": 0.8913502109704642, | |
| "grad_norm": 0.3659191429615021, | |
| "learning_rate": 4.413978962654508e-05, | |
| "loss": 1.1094, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.8924050632911392, | |
| "grad_norm": 0.3675118684768677, | |
| "learning_rate": 4.3295170319268554e-05, | |
| "loss": 1.1051, | |
| "step": 8460 | |
| }, | |
| { | |
| "epoch": 0.8934599156118144, | |
| "grad_norm": 0.3624016344547272, | |
| "learning_rate": 4.245846998468261e-05, | |
| "loss": 1.1058, | |
| "step": 8470 | |
| }, | |
| { | |
| "epoch": 0.8945147679324894, | |
| "grad_norm": 0.3543236553668976, | |
| "learning_rate": 4.16296979984232e-05, | |
| "loss": 1.1013, | |
| "step": 8480 | |
| }, | |
| { | |
| "epoch": 0.8955696202531646, | |
| "grad_norm": 0.34951353073120117, | |
| "learning_rate": 4.080886364728506e-05, | |
| "loss": 1.1069, | |
| "step": 8490 | |
| }, | |
| { | |
| "epoch": 0.8966244725738397, | |
| "grad_norm": 0.38264498114585876, | |
| "learning_rate": 3.999597612911793e-05, | |
| "loss": 1.0967, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.8976793248945147, | |
| "grad_norm": 0.37308940291404724, | |
| "learning_rate": 3.9191044552723345e-05, | |
| "loss": 1.1087, | |
| "step": 8510 | |
| }, | |
| { | |
| "epoch": 0.8987341772151899, | |
| "grad_norm": 0.38366225361824036, | |
| "learning_rate": 3.839407793775268e-05, | |
| "loss": 1.1049, | |
| "step": 8520 | |
| }, | |
| { | |
| "epoch": 0.8997890295358649, | |
| "grad_norm": 0.36603066325187683, | |
| "learning_rate": 3.760508521460584e-05, | |
| "loss": 1.1171, | |
| "step": 8530 | |
| }, | |
| { | |
| "epoch": 0.9008438818565401, | |
| "grad_norm": 0.35585013031959534, | |
| "learning_rate": 3.682407522433173e-05, | |
| "loss": 1.1129, | |
| "step": 8540 | |
| }, | |
| { | |
| "epoch": 0.9018987341772152, | |
| "grad_norm": 0.38468706607818604, | |
| "learning_rate": 3.605105671852854e-05, | |
| "loss": 1.105, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.9029535864978903, | |
| "grad_norm": 0.3588978350162506, | |
| "learning_rate": 3.528603835924626e-05, | |
| "loss": 1.0985, | |
| "step": 8560 | |
| }, | |
| { | |
| "epoch": 0.9040084388185654, | |
| "grad_norm": 0.39159154891967773, | |
| "learning_rate": 3.4529028718888935e-05, | |
| "loss": 1.1147, | |
| "step": 8570 | |
| }, | |
| { | |
| "epoch": 0.9050632911392406, | |
| "grad_norm": 0.3527233898639679, | |
| "learning_rate": 3.378003628011938e-05, | |
| "loss": 1.1063, | |
| "step": 8580 | |
| }, | |
| { | |
| "epoch": 0.9061181434599156, | |
| "grad_norm": 0.3757486343383789, | |
| "learning_rate": 3.303906943576346e-05, | |
| "loss": 1.1108, | |
| "step": 8590 | |
| }, | |
| { | |
| "epoch": 0.9071729957805907, | |
| "grad_norm": 0.3887219727039337, | |
| "learning_rate": 3.230613648871661e-05, | |
| "loss": 1.1016, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.9082278481012658, | |
| "grad_norm": 0.3607241213321686, | |
| "learning_rate": 3.158124565185022e-05, | |
| "loss": 1.107, | |
| "step": 8610 | |
| }, | |
| { | |
| "epoch": 0.9092827004219409, | |
| "grad_norm": 0.3600178360939026, | |
| "learning_rate": 3.086440504792026e-05, | |
| "loss": 1.1031, | |
| "step": 8620 | |
| }, | |
| { | |
| "epoch": 0.9103375527426161, | |
| "grad_norm": 0.38059672713279724, | |
| "learning_rate": 3.015562270947553e-05, | |
| "loss": 1.1108, | |
| "step": 8630 | |
| }, | |
| { | |
| "epoch": 0.9113924050632911, | |
| "grad_norm": 0.3696359097957611, | |
| "learning_rate": 2.945490657876837e-05, | |
| "loss": 1.0924, | |
| "step": 8640 | |
| }, | |
| { | |
| "epoch": 0.9124472573839663, | |
| "grad_norm": 0.36051735281944275, | |
| "learning_rate": 2.8762264507665113e-05, | |
| "loss": 1.098, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.9135021097046413, | |
| "grad_norm": 0.37566593289375305, | |
| "learning_rate": 2.807770425755829e-05, | |
| "loss": 1.0973, | |
| "step": 8660 | |
| }, | |
| { | |
| "epoch": 0.9145569620253164, | |
| "grad_norm": 0.35185009241104126, | |
| "learning_rate": 2.7401233499279866e-05, | |
| "loss": 1.1077, | |
| "step": 8670 | |
| }, | |
| { | |
| "epoch": 0.9156118143459916, | |
| "grad_norm": 0.36478352546691895, | |
| "learning_rate": 2.6732859813014987e-05, | |
| "loss": 1.1157, | |
| "step": 8680 | |
| }, | |
| { | |
| "epoch": 0.9166666666666666, | |
| "grad_norm": 0.3797094225883484, | |
| "learning_rate": 2.607259068821721e-05, | |
| "loss": 1.1092, | |
| "step": 8690 | |
| }, | |
| { | |
| "epoch": 0.9177215189873418, | |
| "grad_norm": 0.3498828113079071, | |
| "learning_rate": 2.5420433523524493e-05, | |
| "loss": 1.1054, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.9187763713080169, | |
| "grad_norm": 0.36742252111434937, | |
| "learning_rate": 2.4776395626676162e-05, | |
| "loss": 1.1059, | |
| "step": 8710 | |
| }, | |
| { | |
| "epoch": 0.919831223628692, | |
| "grad_norm": 0.3539731204509735, | |
| "learning_rate": 2.414048421443141e-05, | |
| "loss": 1.0994, | |
| "step": 8720 | |
| }, | |
| { | |
| "epoch": 0.9208860759493671, | |
| "grad_norm": 0.36758720874786377, | |
| "learning_rate": 2.3512706412488012e-05, | |
| "loss": 1.1128, | |
| "step": 8730 | |
| }, | |
| { | |
| "epoch": 0.9219409282700421, | |
| "grad_norm": 0.3628343343734741, | |
| "learning_rate": 2.2893069255402993e-05, | |
| "loss": 1.1013, | |
| "step": 8740 | |
| }, | |
| { | |
| "epoch": 0.9229957805907173, | |
| "grad_norm": 0.3503541052341461, | |
| "learning_rate": 2.2281579686513176e-05, | |
| "loss": 1.0978, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.9240506329113924, | |
| "grad_norm": 0.3604075014591217, | |
| "learning_rate": 2.1678244557857663e-05, | |
| "loss": 1.088, | |
| "step": 8760 | |
| }, | |
| { | |
| "epoch": 0.9251054852320675, | |
| "grad_norm": 0.34871819615364075, | |
| "learning_rate": 2.1083070630101232e-05, | |
| "loss": 1.0962, | |
| "step": 8770 | |
| }, | |
| { | |
| "epoch": 0.9261603375527426, | |
| "grad_norm": 0.35838931798934937, | |
| "learning_rate": 2.0496064572458395e-05, | |
| "loss": 1.1052, | |
| "step": 8780 | |
| }, | |
| { | |
| "epoch": 0.9272151898734177, | |
| "grad_norm": 0.3570609986782074, | |
| "learning_rate": 1.991723296261863e-05, | |
| "loss": 1.0995, | |
| "step": 8790 | |
| }, | |
| { | |
| "epoch": 0.9282700421940928, | |
| "grad_norm": 0.35785019397735596, | |
| "learning_rate": 1.9346582286672686e-05, | |
| "loss": 1.0957, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.929324894514768, | |
| "grad_norm": 0.3643856644630432, | |
| "learning_rate": 1.878411893904014e-05, | |
| "loss": 1.1149, | |
| "step": 8810 | |
| }, | |
| { | |
| "epoch": 0.930379746835443, | |
| "grad_norm": 0.3485681712627411, | |
| "learning_rate": 1.822984922239737e-05, | |
| "loss": 1.1, | |
| "step": 8820 | |
| }, | |
| { | |
| "epoch": 0.9314345991561181, | |
| "grad_norm": 0.3704765737056732, | |
| "learning_rate": 1.7683779347607286e-05, | |
| "loss": 1.1168, | |
| "step": 8830 | |
| }, | |
| { | |
| "epoch": 0.9324894514767933, | |
| "grad_norm": 0.35022515058517456, | |
| "learning_rate": 1.714591543364938e-05, | |
| "loss": 1.1045, | |
| "step": 8840 | |
| }, | |
| { | |
| "epoch": 0.9335443037974683, | |
| "grad_norm": 0.37648147344589233, | |
| "learning_rate": 1.6616263507551437e-05, | |
| "loss": 1.1046, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.9345991561181435, | |
| "grad_norm": 0.3764314651489258, | |
| "learning_rate": 1.609482950432195e-05, | |
| "loss": 1.1033, | |
| "step": 8860 | |
| }, | |
| { | |
| "epoch": 0.9356540084388185, | |
| "grad_norm": 0.36434733867645264, | |
| "learning_rate": 1.5581619266883563e-05, | |
| "loss": 1.1028, | |
| "step": 8870 | |
| }, | |
| { | |
| "epoch": 0.9367088607594937, | |
| "grad_norm": 0.37274083495140076, | |
| "learning_rate": 1.5076638546007548e-05, | |
| "loss": 1.1092, | |
| "step": 8880 | |
| }, | |
| { | |
| "epoch": 0.9377637130801688, | |
| "grad_norm": 0.361558198928833, | |
| "learning_rate": 1.457989300024945e-05, | |
| "loss": 1.1047, | |
| "step": 8890 | |
| }, | |
| { | |
| "epoch": 0.9388185654008439, | |
| "grad_norm": 0.3575560748577118, | |
| "learning_rate": 1.4091388195885625e-05, | |
| "loss": 1.0988, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.939873417721519, | |
| "grad_norm": 0.36193081736564636, | |
| "learning_rate": 1.3611129606851041e-05, | |
| "loss": 1.098, | |
| "step": 8910 | |
| }, | |
| { | |
| "epoch": 0.9409282700421941, | |
| "grad_norm": 0.3471304476261139, | |
| "learning_rate": 1.313912261467759e-05, | |
| "loss": 1.1043, | |
| "step": 8920 | |
| }, | |
| { | |
| "epoch": 0.9419831223628692, | |
| "grad_norm": 0.35767582058906555, | |
| "learning_rate": 1.267537250843412e-05, | |
| "loss": 1.1029, | |
| "step": 8930 | |
| }, | |
| { | |
| "epoch": 0.9430379746835443, | |
| "grad_norm": 0.3535594046115875, | |
| "learning_rate": 1.2219884484667071e-05, | |
| "loss": 1.1013, | |
| "step": 8940 | |
| }, | |
| { | |
| "epoch": 0.9440928270042194, | |
| "grad_norm": 0.35520821809768677, | |
| "learning_rate": 1.1772663647341947e-05, | |
| "loss": 1.1031, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.9451476793248945, | |
| "grad_norm": 0.35638466477394104, | |
| "learning_rate": 1.1333715007786932e-05, | |
| "loss": 1.1059, | |
| "step": 8960 | |
| }, | |
| { | |
| "epoch": 0.9462025316455697, | |
| "grad_norm": 0.36158987879753113, | |
| "learning_rate": 1.0903043484635694e-05, | |
| "loss": 1.1043, | |
| "step": 8970 | |
| }, | |
| { | |
| "epoch": 0.9472573839662447, | |
| "grad_norm": 0.37424609065055847, | |
| "learning_rate": 1.0480653903772924e-05, | |
| "loss": 1.0999, | |
| "step": 8980 | |
| }, | |
| { | |
| "epoch": 0.9483122362869199, | |
| "grad_norm": 0.35378211736679077, | |
| "learning_rate": 1.0066550998280132e-05, | |
| "loss": 1.1059, | |
| "step": 8990 | |
| }, | |
| { | |
| "epoch": 0.9493670886075949, | |
| "grad_norm": 0.3605569303035736, | |
| "learning_rate": 9.660739408382608e-06, | |
| "loss": 1.1022, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.95042194092827, | |
| "grad_norm": 0.3631739020347595, | |
| "learning_rate": 9.26322368139737e-06, | |
| "loss": 1.0971, | |
| "step": 9010 | |
| }, | |
| { | |
| "epoch": 0.9514767932489452, | |
| "grad_norm": 0.3738535940647125, | |
| "learning_rate": 8.874008271682222e-06, | |
| "loss": 1.0982, | |
| "step": 9020 | |
| }, | |
| { | |
| "epoch": 0.9525316455696202, | |
| "grad_norm": 0.3611426055431366, | |
| "learning_rate": 8.493097540585775e-06, | |
| "loss": 1.117, | |
| "step": 9030 | |
| }, | |
| { | |
| "epoch": 0.9535864978902954, | |
| "grad_norm": 0.36006152629852295, | |
| "learning_rate": 8.120495756399005e-06, | |
| "loss": 1.0943, | |
| "step": 9040 | |
| }, | |
| { | |
| "epoch": 0.9546413502109705, | |
| "grad_norm": 0.3576846718788147, | |
| "learning_rate": 7.756207094306605e-06, | |
| "loss": 1.1044, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.9556962025316456, | |
| "grad_norm": 0.36890554428100586, | |
| "learning_rate": 7.400235636340957e-06, | |
| "loss": 1.1062, | |
| "step": 9060 | |
| }, | |
| { | |
| "epoch": 0.9567510548523207, | |
| "grad_norm": 0.3569638729095459, | |
| "learning_rate": 7.0525853713362395e-06, | |
| "loss": 1.1074, | |
| "step": 9070 | |
| }, | |
| { | |
| "epoch": 0.9578059071729957, | |
| "grad_norm": 0.35618263483047485, | |
| "learning_rate": 6.71326019488322e-06, | |
| "loss": 1.1086, | |
| "step": 9080 | |
| }, | |
| { | |
| "epoch": 0.9588607594936709, | |
| "grad_norm": 0.37716180086135864, | |
| "learning_rate": 6.3822639092862846e-06, | |
| "loss": 1.1082, | |
| "step": 9090 | |
| }, | |
| { | |
| "epoch": 0.959915611814346, | |
| "grad_norm": 0.3503626883029938, | |
| "learning_rate": 6.059600223520478e-06, | |
| "loss": 1.0881, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.9609704641350211, | |
| "grad_norm": 0.36392760276794434, | |
| "learning_rate": 5.745272753189784e-06, | |
| "loss": 1.0974, | |
| "step": 9110 | |
| }, | |
| { | |
| "epoch": 0.9620253164556962, | |
| "grad_norm": 0.35463055968284607, | |
| "learning_rate": 5.439285020487156e-06, | |
| "loss": 1.1097, | |
| "step": 9120 | |
| }, | |
| { | |
| "epoch": 0.9630801687763713, | |
| "grad_norm": 0.35742634534835815, | |
| "learning_rate": 5.141640454154467e-06, | |
| "loss": 1.0939, | |
| "step": 9130 | |
| }, | |
| { | |
| "epoch": 0.9641350210970464, | |
| "grad_norm": 0.36812102794647217, | |
| "learning_rate": 4.852342389444458e-06, | |
| "loss": 1.1132, | |
| "step": 9140 | |
| }, | |
| { | |
| "epoch": 0.9651898734177216, | |
| "grad_norm": 0.37311768531799316, | |
| "learning_rate": 4.571394068083185e-06, | |
| "loss": 1.1018, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.9662447257383966, | |
| "grad_norm": 0.3573930263519287, | |
| "learning_rate": 4.298798638233709e-06, | |
| "loss": 1.1089, | |
| "step": 9160 | |
| }, | |
| { | |
| "epoch": 0.9672995780590717, | |
| "grad_norm": 0.3686988949775696, | |
| "learning_rate": 4.034559154461049e-06, | |
| "loss": 1.1077, | |
| "step": 9170 | |
| }, | |
| { | |
| "epoch": 0.9683544303797469, | |
| "grad_norm": 0.34888580441474915, | |
| "learning_rate": 3.7786785776976198e-06, | |
| "loss": 1.0993, | |
| "step": 9180 | |
| }, | |
| { | |
| "epoch": 0.9694092827004219, | |
| "grad_norm": 0.371014803647995, | |
| "learning_rate": 3.5311597752100964e-06, | |
| "loss": 1.0974, | |
| "step": 9190 | |
| }, | |
| { | |
| "epoch": 0.9704641350210971, | |
| "grad_norm": 0.36439818143844604, | |
| "learning_rate": 3.2920055205676867e-06, | |
| "loss": 1.1026, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.9715189873417721, | |
| "grad_norm": 0.34828996658325195, | |
| "learning_rate": 3.06121849361049e-06, | |
| "loss": 1.1013, | |
| "step": 9210 | |
| }, | |
| { | |
| "epoch": 0.9725738396624473, | |
| "grad_norm": 0.3600952625274658, | |
| "learning_rate": 2.838801280419856e-06, | |
| "loss": 1.1016, | |
| "step": 9220 | |
| }, | |
| { | |
| "epoch": 0.9736286919831224, | |
| "grad_norm": 0.35362669825553894, | |
| "learning_rate": 2.624756373289322e-06, | |
| "loss": 1.0932, | |
| "step": 9230 | |
| }, | |
| { | |
| "epoch": 0.9746835443037974, | |
| "grad_norm": 0.35408636927604675, | |
| "learning_rate": 2.419086170696472e-06, | |
| "loss": 1.0874, | |
| "step": 9240 | |
| }, | |
| { | |
| "epoch": 0.9757383966244726, | |
| "grad_norm": 0.35020846128463745, | |
| "learning_rate": 2.2217929772764545e-06, | |
| "loss": 1.0993, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.9767932489451476, | |
| "grad_norm": 0.3536199629306793, | |
| "learning_rate": 2.0328790037957568e-06, | |
| "loss": 1.0937, | |
| "step": 9260 | |
| }, | |
| { | |
| "epoch": 0.9778481012658228, | |
| "grad_norm": 0.35778024792671204, | |
| "learning_rate": 1.8523463671278052e-06, | |
| "loss": 1.1041, | |
| "step": 9270 | |
| }, | |
| { | |
| "epoch": 0.9789029535864979, | |
| "grad_norm": 0.34945109486579895, | |
| "learning_rate": 1.6801970902288188e-06, | |
| "loss": 1.0929, | |
| "step": 9280 | |
| }, | |
| { | |
| "epoch": 0.979957805907173, | |
| "grad_norm": 0.36028343439102173, | |
| "learning_rate": 1.5164331021155774e-06, | |
| "loss": 1.1024, | |
| "step": 9290 | |
| }, | |
| { | |
| "epoch": 0.9810126582278481, | |
| "grad_norm": 0.3585197329521179, | |
| "learning_rate": 1.3610562378435221e-06, | |
| "loss": 1.0997, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.9820675105485233, | |
| "grad_norm": 0.3567984998226166, | |
| "learning_rate": 1.2140682384862712e-06, | |
| "loss": 1.1006, | |
| "step": 9310 | |
| }, | |
| { | |
| "epoch": 0.9831223628691983, | |
| "grad_norm": 0.3587944209575653, | |
| "learning_rate": 1.0754707511161365e-06, | |
| "loss": 1.0956, | |
| "step": 9320 | |
| }, | |
| { | |
| "epoch": 0.9841772151898734, | |
| "grad_norm": 0.360120564699173, | |
| "learning_rate": 9.452653287856383e-07, | |
| "loss": 1.1082, | |
| "step": 9330 | |
| }, | |
| { | |
| "epoch": 0.9852320675105485, | |
| "grad_norm": 0.36455395817756653, | |
| "learning_rate": 8.234534305101015e-07, | |
| "loss": 1.105, | |
| "step": 9340 | |
| }, | |
| { | |
| "epoch": 0.9862869198312236, | |
| "grad_norm": 0.3427172303199768, | |
| "learning_rate": 7.100364212513367e-07, | |
| "loss": 1.116, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.9873417721518988, | |
| "grad_norm": 0.3511948585510254, | |
| "learning_rate": 6.050155719023176e-07, | |
| "loss": 1.1072, | |
| "step": 9360 | |
| }, | |
| { | |
| "epoch": 0.9883966244725738, | |
| "grad_norm": 0.3564155399799347, | |
| "learning_rate": 5.08392059272944e-07, | |
| "loss": 1.1053, | |
| "step": 9370 | |
| }, | |
| { | |
| "epoch": 0.989451476793249, | |
| "grad_norm": 0.3686405420303345, | |
| "learning_rate": 4.2016696607680147e-07, | |
| "loss": 1.0926, | |
| "step": 9380 | |
| }, | |
| { | |
| "epoch": 0.990506329113924, | |
| "grad_norm": 0.3639363646507263, | |
| "learning_rate": 3.4034128091917085e-07, | |
| "loss": 1.0972, | |
| "step": 9390 | |
| }, | |
| { | |
| "epoch": 0.9915611814345991, | |
| "grad_norm": 0.35043707489967346, | |
| "learning_rate": 2.689158982859541e-07, | |
| "loss": 1.1099, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.9926160337552743, | |
| "grad_norm": 0.35883840918540955, | |
| "learning_rate": 2.05891618533266e-07, | |
| "loss": 1.0861, | |
| "step": 9410 | |
| }, | |
| { | |
| "epoch": 0.9936708860759493, | |
| "grad_norm": 0.3705896735191345, | |
| "learning_rate": 1.5126914787894074e-07, | |
| "loss": 1.1049, | |
| "step": 9420 | |
| }, | |
| { | |
| "epoch": 0.9947257383966245, | |
| "grad_norm": 0.34956392645835876, | |
| "learning_rate": 1.0504909839462173e-07, | |
| "loss": 1.1043, | |
| "step": 9430 | |
| }, | |
| { | |
| "epoch": 0.9957805907172996, | |
| "grad_norm": 0.3487434983253479, | |
| "learning_rate": 6.723198799826746e-08, | |
| "loss": 1.1014, | |
| "step": 9440 | |
| }, | |
| { | |
| "epoch": 0.9968354430379747, | |
| "grad_norm": 0.34953734278678894, | |
| "learning_rate": 3.781824044932214e-08, | |
| "loss": 1.1097, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.9978902953586498, | |
| "grad_norm": 0.3557943105697632, | |
| "learning_rate": 1.6808185342970238e-08, | |
| "loss": 1.0945, | |
| "step": 9460 | |
| }, | |
| { | |
| "epoch": 0.9989451476793249, | |
| "grad_norm": 0.3526889979839325, | |
| "learning_rate": 4.202058107305451e-09, | |
| "loss": 1.1073, | |
| "step": 9470 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 1.046021819114685, | |
| "learning_rate": 0.0, | |
| "loss": 1.0947, | |
| "step": 9480 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 9480, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.036240179760947e+16, | |
| "train_batch_size": 1024, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |