| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 9940, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.005030181086519115, | |
| "grad_norm": 7.163974285125732, | |
| "learning_rate": 4.024144869215292e-06, | |
| "loss": 0.8528, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01006036217303823, | |
| "grad_norm": 5.244224548339844, | |
| "learning_rate": 8.048289738430584e-06, | |
| "loss": 0.7448, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.015090543259557344, | |
| "grad_norm": 2.7584030628204346, | |
| "learning_rate": 1.2072434607645876e-05, | |
| "loss": 0.5215, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02012072434607646, | |
| "grad_norm": 2.765652656555176, | |
| "learning_rate": 1.609657947686117e-05, | |
| "loss": 0.3683, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.025150905432595575, | |
| "grad_norm": 2.0063560009002686, | |
| "learning_rate": 2.012072434607646e-05, | |
| "loss": 0.2833, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.030181086519114688, | |
| "grad_norm": 3.22554087638855, | |
| "learning_rate": 2.4144869215291752e-05, | |
| "loss": 0.2435, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.035211267605633804, | |
| "grad_norm": 1.5418182611465454, | |
| "learning_rate": 2.8169014084507046e-05, | |
| "loss": 0.188, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04024144869215292, | |
| "grad_norm": 7.7757697105407715, | |
| "learning_rate": 3.219315895372234e-05, | |
| "loss": 0.1598, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04527162977867203, | |
| "grad_norm": 2.0921378135681152, | |
| "learning_rate": 3.621730382293763e-05, | |
| "loss": 0.1485, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05030181086519115, | |
| "grad_norm": 1.6707618236541748, | |
| "learning_rate": 4.024144869215292e-05, | |
| "loss": 0.1427, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05533199195171026, | |
| "grad_norm": 2.270603895187378, | |
| "learning_rate": 4.426559356136821e-05, | |
| "loss": 0.1314, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.060362173038229376, | |
| "grad_norm": 1.4634528160095215, | |
| "learning_rate": 4.8289738430583503e-05, | |
| "loss": 0.1273, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.06539235412474849, | |
| "grad_norm": 1.9672114849090576, | |
| "learning_rate": 5.2313883299798795e-05, | |
| "loss": 0.0889, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07042253521126761, | |
| "grad_norm": 1.466834545135498, | |
| "learning_rate": 5.633802816901409e-05, | |
| "loss": 0.1037, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.07545271629778671, | |
| "grad_norm": 1.9027611017227173, | |
| "learning_rate": 6.036217303822938e-05, | |
| "loss": 0.1027, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08048289738430583, | |
| "grad_norm": 0.8662084937095642, | |
| "learning_rate": 6.438631790744468e-05, | |
| "loss": 0.0917, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.08551307847082495, | |
| "grad_norm": 2.0497639179229736, | |
| "learning_rate": 6.841046277665996e-05, | |
| "loss": 0.0964, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.09054325955734406, | |
| "grad_norm": 1.6987979412078857, | |
| "learning_rate": 7.243460764587526e-05, | |
| "loss": 0.0868, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.09557344064386318, | |
| "grad_norm": 0.845194935798645, | |
| "learning_rate": 7.645875251509054e-05, | |
| "loss": 0.1019, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1006036217303823, | |
| "grad_norm": 1.4700064659118652, | |
| "learning_rate": 8.048289738430584e-05, | |
| "loss": 0.1007, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1056338028169014, | |
| "grad_norm": 1.8693934679031372, | |
| "learning_rate": 8.450704225352113e-05, | |
| "loss": 0.0897, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.11066398390342053, | |
| "grad_norm": 1.1521185636520386, | |
| "learning_rate": 8.853118712273642e-05, | |
| "loss": 0.1064, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.11569416498993963, | |
| "grad_norm": 1.094040036201477, | |
| "learning_rate": 9.255533199195171e-05, | |
| "loss": 0.0706, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.12072434607645875, | |
| "grad_norm": 1.2210173606872559, | |
| "learning_rate": 9.657947686116701e-05, | |
| "loss": 0.078, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.12575452716297786, | |
| "grad_norm": 0.688697874546051, | |
| "learning_rate": 0.00010060362173038229, | |
| "loss": 0.0685, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.13078470824949698, | |
| "grad_norm": 0.9590277075767517, | |
| "learning_rate": 0.00010462776659959759, | |
| "loss": 0.0799, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.1358148893360161, | |
| "grad_norm": 0.8572778105735779, | |
| "learning_rate": 0.00010865191146881289, | |
| "loss": 0.0705, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.14084507042253522, | |
| "grad_norm": 0.8274833559989929, | |
| "learning_rate": 0.00011267605633802819, | |
| "loss": 0.0887, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.14587525150905434, | |
| "grad_norm": 0.9543363451957703, | |
| "learning_rate": 0.00011670020120724347, | |
| "loss": 0.0618, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.15090543259557343, | |
| "grad_norm": 0.7259362936019897, | |
| "learning_rate": 0.00012072434607645876, | |
| "loss": 0.075, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.15593561368209255, | |
| "grad_norm": 0.8904047012329102, | |
| "learning_rate": 0.00012474849094567405, | |
| "loss": 0.07, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.16096579476861167, | |
| "grad_norm": 0.7555816173553467, | |
| "learning_rate": 0.00012877263581488935, | |
| "loss": 0.0677, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.1659959758551308, | |
| "grad_norm": 0.7622585892677307, | |
| "learning_rate": 0.00013279678068410465, | |
| "loss": 0.0762, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.1710261569416499, | |
| "grad_norm": 1.1940929889678955, | |
| "learning_rate": 0.00013682092555331992, | |
| "loss": 0.059, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.176056338028169, | |
| "grad_norm": 0.582521915435791, | |
| "learning_rate": 0.00014084507042253522, | |
| "loss": 0.0732, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.18108651911468812, | |
| "grad_norm": 0.858220636844635, | |
| "learning_rate": 0.00014486921529175052, | |
| "loss": 0.06, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.18611670020120724, | |
| "grad_norm": 0.5738099217414856, | |
| "learning_rate": 0.0001488933601609658, | |
| "loss": 0.0568, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.19114688128772636, | |
| "grad_norm": 0.581499457359314, | |
| "learning_rate": 0.00015291750503018109, | |
| "loss": 0.062, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.19617706237424548, | |
| "grad_norm": 0.5559177994728088, | |
| "learning_rate": 0.00015694164989939638, | |
| "loss": 0.0774, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.2012072434607646, | |
| "grad_norm": 0.6640293598175049, | |
| "learning_rate": 0.00016096579476861168, | |
| "loss": 0.058, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2062374245472837, | |
| "grad_norm": 0.5883966088294983, | |
| "learning_rate": 0.00016498993963782695, | |
| "loss": 0.0625, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.2112676056338028, | |
| "grad_norm": 0.5281286835670471, | |
| "learning_rate": 0.00016901408450704225, | |
| "loss": 0.0733, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.21629778672032193, | |
| "grad_norm": 0.46224042773246765, | |
| "learning_rate": 0.00017303822937625755, | |
| "loss": 0.0644, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.22132796780684105, | |
| "grad_norm": 0.4865374267101288, | |
| "learning_rate": 0.00017706237424547285, | |
| "loss": 0.0677, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.22635814889336017, | |
| "grad_norm": 1.1236791610717773, | |
| "learning_rate": 0.00018108651911468815, | |
| "loss": 0.0719, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.23138832997987926, | |
| "grad_norm": 1.1186903715133667, | |
| "learning_rate": 0.00018511066398390342, | |
| "loss": 0.064, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.23641851106639838, | |
| "grad_norm": 0.6385508179664612, | |
| "learning_rate": 0.00018913480885311872, | |
| "loss": 0.074, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.2414486921529175, | |
| "grad_norm": 1.3683863878250122, | |
| "learning_rate": 0.00019315895372233401, | |
| "loss": 0.0626, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.24647887323943662, | |
| "grad_norm": 0.5213209390640259, | |
| "learning_rate": 0.0001971830985915493, | |
| "loss": 0.0585, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.2515090543259557, | |
| "grad_norm": 0.988146960735321, | |
| "learning_rate": 0.00019999995019278672, | |
| "loss": 0.0479, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.25653923541247486, | |
| "grad_norm": 0.707419216632843, | |
| "learning_rate": 0.0001999990647325972, | |
| "loss": 0.0558, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.26156941649899396, | |
| "grad_norm": 0.5720793604850769, | |
| "learning_rate": 0.0001999970724567263, | |
| "loss": 0.0383, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.2665995975855131, | |
| "grad_norm": 0.53882896900177, | |
| "learning_rate": 0.00019999397338722502, | |
| "loss": 0.0565, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.2716297786720322, | |
| "grad_norm": 0.8112373352050781, | |
| "learning_rate": 0.00019998976755839472, | |
| "loss": 0.0647, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.2766599597585513, | |
| "grad_norm": 0.7053709626197815, | |
| "learning_rate": 0.00019998445501678657, | |
| "loss": 0.0533, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.28169014084507044, | |
| "grad_norm": 0.40134745836257935, | |
| "learning_rate": 0.0001999780358212011, | |
| "loss": 0.0621, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.28672032193158953, | |
| "grad_norm": 0.9213513135910034, | |
| "learning_rate": 0.00019997051004268777, | |
| "loss": 0.0729, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.2917505030181087, | |
| "grad_norm": 0.6868863105773926, | |
| "learning_rate": 0.00019996187776454374, | |
| "loss": 0.0636, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.29678068410462777, | |
| "grad_norm": 1.0402840375900269, | |
| "learning_rate": 0.0001999521390823134, | |
| "loss": 0.0561, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.30181086519114686, | |
| "grad_norm": 0.6680485010147095, | |
| "learning_rate": 0.00019994129410378695, | |
| "loss": 0.0551, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.306841046277666, | |
| "grad_norm": 0.6051433682441711, | |
| "learning_rate": 0.00019992934294899944, | |
| "loss": 0.0596, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.3118712273641851, | |
| "grad_norm": 0.5517066121101379, | |
| "learning_rate": 0.00019991628575022946, | |
| "loss": 0.057, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.31690140845070425, | |
| "grad_norm": 0.4259048104286194, | |
| "learning_rate": 0.00019990212265199738, | |
| "loss": 0.0451, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.32193158953722334, | |
| "grad_norm": 0.4159329831600189, | |
| "learning_rate": 0.0001998868538110641, | |
| "loss": 0.0493, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.32696177062374243, | |
| "grad_norm": 0.5596720576286316, | |
| "learning_rate": 0.0001998704793964291, | |
| "loss": 0.0556, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.3319919517102616, | |
| "grad_norm": 0.821570098400116, | |
| "learning_rate": 0.00019985299958932866, | |
| "loss": 0.0517, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.33702213279678067, | |
| "grad_norm": 0.6846859455108643, | |
| "learning_rate": 0.0001998344145832339, | |
| "loss": 0.0367, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.3420523138832998, | |
| "grad_norm": 0.6848050951957703, | |
| "learning_rate": 0.00019981472458384844, | |
| "loss": 0.0536, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.3470824949698189, | |
| "grad_norm": 2.0011439323425293, | |
| "learning_rate": 0.00019979392980910637, | |
| "loss": 0.0505, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.352112676056338, | |
| "grad_norm": 0.7617112994194031, | |
| "learning_rate": 0.00019977203048916962, | |
| "loss": 0.0518, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.35714285714285715, | |
| "grad_norm": 0.360905259847641, | |
| "learning_rate": 0.00019974902686642558, | |
| "loss": 0.0545, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.36217303822937624, | |
| "grad_norm": 0.8060867190361023, | |
| "learning_rate": 0.00019972491919548438, | |
| "loss": 0.0481, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.3672032193158954, | |
| "grad_norm": 0.7293612360954285, | |
| "learning_rate": 0.00019969970774317593, | |
| "loss": 0.0676, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.3722334004024145, | |
| "grad_norm": 0.5890267491340637, | |
| "learning_rate": 0.00019967339278854714, | |
| "loss": 0.0571, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.3772635814889336, | |
| "grad_norm": 0.7763799428939819, | |
| "learning_rate": 0.00019964597462285888, | |
| "loss": 0.057, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.3822937625754527, | |
| "grad_norm": 0.5891818404197693, | |
| "learning_rate": 0.00019961745354958246, | |
| "loss": 0.0473, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.3873239436619718, | |
| "grad_norm": 0.41080498695373535, | |
| "learning_rate": 0.00019958782988439654, | |
| "loss": 0.0506, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.39235412474849096, | |
| "grad_norm": 0.46914100646972656, | |
| "learning_rate": 0.00019955710395518363, | |
| "loss": 0.053, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.39738430583501005, | |
| "grad_norm": 0.8423359990119934, | |
| "learning_rate": 0.00019952527610202624, | |
| "loss": 0.0429, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.4024144869215292, | |
| "grad_norm": 0.5299778580665588, | |
| "learning_rate": 0.00019949234667720336, | |
| "loss": 0.0499, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4074446680080483, | |
| "grad_norm": 0.5854944586753845, | |
| "learning_rate": 0.00019945831604518645, | |
| "loss": 0.0473, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.4124748490945674, | |
| "grad_norm": 0.6145129203796387, | |
| "learning_rate": 0.0001994231845826354, | |
| "loss": 0.0645, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.41750503018108653, | |
| "grad_norm": 0.6725190877914429, | |
| "learning_rate": 0.00019938695267839436, | |
| "loss": 0.0511, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.4225352112676056, | |
| "grad_norm": 0.5802650451660156, | |
| "learning_rate": 0.0001993496207334875, | |
| "loss": 0.0388, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.4275653923541248, | |
| "grad_norm": 0.6143836975097656, | |
| "learning_rate": 0.00019931118916111448, | |
| "loss": 0.043, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.43259557344064387, | |
| "grad_norm": 0.6360625624656677, | |
| "learning_rate": 0.00019927165838664598, | |
| "loss": 0.0541, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.43762575452716296, | |
| "grad_norm": 0.49990031123161316, | |
| "learning_rate": 0.00019923102884761892, | |
| "loss": 0.0412, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.4426559356136821, | |
| "grad_norm": 0.6556900143623352, | |
| "learning_rate": 0.00019918930099373157, | |
| "loss": 0.055, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.4476861167002012, | |
| "grad_norm": 0.42863693833351135, | |
| "learning_rate": 0.00019914647528683865, | |
| "loss": 0.046, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.45271629778672035, | |
| "grad_norm": 0.31682249903678894, | |
| "learning_rate": 0.00019910255220094634, | |
| "loss": 0.0477, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.45774647887323944, | |
| "grad_norm": 0.5968728065490723, | |
| "learning_rate": 0.0001990575322222067, | |
| "loss": 0.0469, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.46277665995975853, | |
| "grad_norm": 0.4193131923675537, | |
| "learning_rate": 0.00019901141584891262, | |
| "loss": 0.0584, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.4678068410462777, | |
| "grad_norm": 0.5087130069732666, | |
| "learning_rate": 0.00019896420359149207, | |
| "loss": 0.0595, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.47283702213279677, | |
| "grad_norm": 0.3542875349521637, | |
| "learning_rate": 0.00019891589597250265, | |
| "loss": 0.0384, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.4778672032193159, | |
| "grad_norm": 0.3090812563896179, | |
| "learning_rate": 0.00019886649352662567, | |
| "loss": 0.0427, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.482897384305835, | |
| "grad_norm": 0.6682177782058716, | |
| "learning_rate": 0.00019881599680066024, | |
| "loss": 0.0565, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.4879275653923541, | |
| "grad_norm": 0.4167076349258423, | |
| "learning_rate": 0.0001987644063535173, | |
| "loss": 0.0392, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.49295774647887325, | |
| "grad_norm": 0.4293626546859741, | |
| "learning_rate": 0.00019871172275621332, | |
| "loss": 0.0409, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.49798792756539234, | |
| "grad_norm": 0.6032306551933289, | |
| "learning_rate": 0.00019865794659186406, | |
| "loss": 0.0453, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.5030181086519114, | |
| "grad_norm": 0.5772558450698853, | |
| "learning_rate": 0.00019860307845567815, | |
| "loss": 0.0413, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5080482897384306, | |
| "grad_norm": 0.5984110236167908, | |
| "learning_rate": 0.00019854711895495036, | |
| "loss": 0.0551, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.5130784708249497, | |
| "grad_norm": 0.46663838624954224, | |
| "learning_rate": 0.00019849006870905503, | |
| "loss": 0.0369, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.5181086519114688, | |
| "grad_norm": 0.5539775490760803, | |
| "learning_rate": 0.00019843192834943912, | |
| "loss": 0.0378, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.5231388329979879, | |
| "grad_norm": 0.5635536313056946, | |
| "learning_rate": 0.0001983726985196153, | |
| "loss": 0.0394, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.528169014084507, | |
| "grad_norm": 0.6095893979072571, | |
| "learning_rate": 0.00019831237987515474, | |
| "loss": 0.0354, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.5331991951710262, | |
| "grad_norm": 0.4370132386684418, | |
| "learning_rate": 0.00019825097308367987, | |
| "loss": 0.0477, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.5382293762575453, | |
| "grad_norm": 0.7766854763031006, | |
| "learning_rate": 0.00019818847882485704, | |
| "loss": 0.0368, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.5432595573440644, | |
| "grad_norm": 0.8621564507484436, | |
| "learning_rate": 0.00019812489779038903, | |
| "loss": 0.0493, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.5482897384305835, | |
| "grad_norm": 0.41698649525642395, | |
| "learning_rate": 0.00019806023068400723, | |
| "loss": 0.047, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.5533199195171026, | |
| "grad_norm": 0.334310382604599, | |
| "learning_rate": 0.00019799447822146403, | |
| "loss": 0.0352, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.5583501006036218, | |
| "grad_norm": 0.36509206891059875, | |
| "learning_rate": 0.0001979276411305248, | |
| "loss": 0.0387, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.5633802816901409, | |
| "grad_norm": 0.49226775765419006, | |
| "learning_rate": 0.00019785972015095988, | |
| "loss": 0.0412, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.56841046277666, | |
| "grad_norm": 0.48167258501052856, | |
| "learning_rate": 0.0001977907160345363, | |
| "loss": 0.0454, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.5734406438631791, | |
| "grad_norm": 0.43359094858169556, | |
| "learning_rate": 0.00019772062954500965, | |
| "loss": 0.0355, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.5784708249496981, | |
| "grad_norm": 0.4004541039466858, | |
| "learning_rate": 0.00019764946145811542, | |
| "loss": 0.0401, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.5835010060362174, | |
| "grad_norm": 0.4662073254585266, | |
| "learning_rate": 0.00019757721256156047, | |
| "loss": 0.0382, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.5885311871227364, | |
| "grad_norm": 0.3257390260696411, | |
| "learning_rate": 0.00019750388365501447, | |
| "loss": 0.042, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.5935613682092555, | |
| "grad_norm": 0.5180749297142029, | |
| "learning_rate": 0.0001974294755501008, | |
| "loss": 0.0428, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.5985915492957746, | |
| "grad_norm": 0.4941740334033966, | |
| "learning_rate": 0.00019735398907038779, | |
| "loss": 0.04, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.6036217303822937, | |
| "grad_norm": 0.3722670078277588, | |
| "learning_rate": 0.00019727742505137936, | |
| "loss": 0.0493, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.6086519114688129, | |
| "grad_norm": 0.32677406072616577, | |
| "learning_rate": 0.0001971997843405061, | |
| "loss": 0.0462, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.613682092555332, | |
| "grad_norm": 0.48656272888183594, | |
| "learning_rate": 0.00019712106779711555, | |
| "loss": 0.0451, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.6187122736418511, | |
| "grad_norm": 0.4287594258785248, | |
| "learning_rate": 0.00019704127629246293, | |
| "loss": 0.0294, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.6237424547283702, | |
| "grad_norm": 0.46294787526130676, | |
| "learning_rate": 0.0001969604107097014, | |
| "loss": 0.0462, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.6287726358148893, | |
| "grad_norm": 0.32905468344688416, | |
| "learning_rate": 0.00019687847194387221, | |
| "loss": 0.0324, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.6338028169014085, | |
| "grad_norm": 0.4024835228919983, | |
| "learning_rate": 0.00019679546090189503, | |
| "loss": 0.0407, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.6388329979879276, | |
| "grad_norm": 0.539588451385498, | |
| "learning_rate": 0.00019671137850255766, | |
| "loss": 0.0368, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.6438631790744467, | |
| "grad_norm": 0.32176288962364197, | |
| "learning_rate": 0.00019662622567650595, | |
| "loss": 0.0344, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.6488933601609658, | |
| "grad_norm": 0.45465490221977234, | |
| "learning_rate": 0.0001965400033662336, | |
| "loss": 0.0361, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.6539235412474849, | |
| "grad_norm": 0.47942492365837097, | |
| "learning_rate": 0.00019645271252607155, | |
| "loss": 0.0408, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.6589537223340041, | |
| "grad_norm": 0.2223718762397766, | |
| "learning_rate": 0.00019636435412217758, | |
| "loss": 0.0347, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.6639839034205232, | |
| "grad_norm": 0.550037682056427, | |
| "learning_rate": 0.00019627492913252547, | |
| "loss": 0.0381, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.6690140845070423, | |
| "grad_norm": 0.39280569553375244, | |
| "learning_rate": 0.0001961844385468943, | |
| "loss": 0.0455, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.6740442655935613, | |
| "grad_norm": 0.64002925157547, | |
| "learning_rate": 0.00019609288336685742, | |
| "loss": 0.0495, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.6790744466800804, | |
| "grad_norm": 0.34422725439071655, | |
| "learning_rate": 0.00019600026460577142, | |
| "loss": 0.0309, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.6841046277665996, | |
| "grad_norm": 0.42025962471961975, | |
| "learning_rate": 0.00019590658328876484, | |
| "loss": 0.0479, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.6891348088531187, | |
| "grad_norm": 0.48690280318260193, | |
| "learning_rate": 0.0001958118404527269, | |
| "loss": 0.0352, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.6941649899396378, | |
| "grad_norm": 0.4300714433193207, | |
| "learning_rate": 0.0001957160371462959, | |
| "loss": 0.0367, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.6991951710261569, | |
| "grad_norm": 0.3296925723552704, | |
| "learning_rate": 0.00019561917442984788, | |
| "loss": 0.0399, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.704225352112676, | |
| "grad_norm": 0.5653759837150574, | |
| "learning_rate": 0.00019552125337548462, | |
| "loss": 0.0354, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.7092555331991952, | |
| "grad_norm": 0.39718741178512573, | |
| "learning_rate": 0.00019542227506702173, | |
| "loss": 0.0514, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 0.5332531332969666, | |
| "learning_rate": 0.00019532224059997692, | |
| "loss": 0.0395, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.7193158953722334, | |
| "grad_norm": 0.41001367568969727, | |
| "learning_rate": 0.0001952211510815578, | |
| "loss": 0.0465, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.7243460764587525, | |
| "grad_norm": 0.4375529885292053, | |
| "learning_rate": 0.0001951190076306494, | |
| "loss": 0.0332, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.7293762575452716, | |
| "grad_norm": 0.3857330083847046, | |
| "learning_rate": 0.00019501581137780204, | |
| "loss": 0.0372, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.7344064386317908, | |
| "grad_norm": 0.2521601915359497, | |
| "learning_rate": 0.0001949115634652187, | |
| "loss": 0.0342, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.7394366197183099, | |
| "grad_norm": 0.32540401816368103, | |
| "learning_rate": 0.00019480626504674245, | |
| "loss": 0.0301, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.744466800804829, | |
| "grad_norm": 0.27026352286338806, | |
| "learning_rate": 0.00019469991728784356, | |
| "loss": 0.0354, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.7494969818913481, | |
| "grad_norm": 0.5376250147819519, | |
| "learning_rate": 0.00019459252136560674, | |
| "loss": 0.0429, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.7545271629778671, | |
| "grad_norm": 0.36513984203338623, | |
| "learning_rate": 0.00019448407846871804, | |
| "loss": 0.0396, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.7595573440643864, | |
| "grad_norm": 0.16466562449932098, | |
| "learning_rate": 0.0001943745897974516, | |
| "loss": 0.0252, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.7645875251509054, | |
| "grad_norm": 0.2507416307926178, | |
| "learning_rate": 0.0001942640565636566, | |
| "loss": 0.0338, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.7696177062374245, | |
| "grad_norm": 0.27079200744628906, | |
| "learning_rate": 0.0001941524799907436, | |
| "loss": 0.0331, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.7746478873239436, | |
| "grad_norm": 0.5537086129188538, | |
| "learning_rate": 0.00019403986131367123, | |
| "loss": 0.0312, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.7796780684104627, | |
| "grad_norm": 0.37343883514404297, | |
| "learning_rate": 0.00019392620177893224, | |
| "loss": 0.0447, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.7847082494969819, | |
| "grad_norm": 0.32680392265319824, | |
| "learning_rate": 0.00019381150264454, | |
| "loss": 0.0275, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.789738430583501, | |
| "grad_norm": 0.64634770154953, | |
| "learning_rate": 0.00019369576518001437, | |
| "loss": 0.034, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.7947686116700201, | |
| "grad_norm": 0.2528303861618042, | |
| "learning_rate": 0.00019357899066636773, | |
| "loss": 0.0298, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.7997987927565392, | |
| "grad_norm": 0.4246571660041809, | |
| "learning_rate": 0.00019346118039609086, | |
| "loss": 0.0446, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.8048289738430584, | |
| "grad_norm": 0.3372986316680908, | |
| "learning_rate": 0.0001933423356731384, | |
| "loss": 0.0363, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.8098591549295775, | |
| "grad_norm": 0.49657967686653137, | |
| "learning_rate": 0.00019322245781291475, | |
| "loss": 0.039, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.8148893360160966, | |
| "grad_norm": 0.24667495489120483, | |
| "learning_rate": 0.00019310154814225925, | |
| "loss": 0.0326, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.8199195171026157, | |
| "grad_norm": 0.43383368849754333, | |
| "learning_rate": 0.00019297960799943161, | |
| "loss": 0.0402, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.8249496981891348, | |
| "grad_norm": 0.513130247592926, | |
| "learning_rate": 0.00019285663873409715, | |
| "loss": 0.0357, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.829979879275654, | |
| "grad_norm": 0.4496906101703644, | |
| "learning_rate": 0.00019273264170731157, | |
| "loss": 0.0287, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.8350100603621731, | |
| "grad_norm": 0.467529296875, | |
| "learning_rate": 0.00019260761829150637, | |
| "loss": 0.0437, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.8400402414486922, | |
| "grad_norm": 0.402126282453537, | |
| "learning_rate": 0.0001924815698704732, | |
| "loss": 0.0336, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.8450704225352113, | |
| "grad_norm": 0.41219964623451233, | |
| "learning_rate": 0.00019235449783934881, | |
| "loss": 0.0353, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.8501006036217303, | |
| "grad_norm": 0.6443662047386169, | |
| "learning_rate": 0.00019222640360459954, | |
| "loss": 0.0515, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.8551307847082495, | |
| "grad_norm": 0.3046891391277313, | |
| "learning_rate": 0.0001920972885840057, | |
| "loss": 0.0339, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.8601609657947686, | |
| "grad_norm": 0.5683744549751282, | |
| "learning_rate": 0.00019196715420664596, | |
| "loss": 0.0377, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.8651911468812877, | |
| "grad_norm": 0.46794387698173523, | |
| "learning_rate": 0.0001918360019128815, | |
| "loss": 0.0414, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.8702213279678068, | |
| "grad_norm": 0.4453373849391937, | |
| "learning_rate": 0.00019170383315434002, | |
| "loss": 0.0289, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.8752515090543259, | |
| "grad_norm": 0.44001007080078125, | |
| "learning_rate": 0.00019157064939389978, | |
| "loss": 0.0433, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.8802816901408451, | |
| "grad_norm": 0.28705525398254395, | |
| "learning_rate": 0.00019143645210567328, | |
| "loss": 0.0327, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.8853118712273642, | |
| "grad_norm": 0.49537310004234314, | |
| "learning_rate": 0.00019130124277499109, | |
| "loss": 0.0375, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.8903420523138833, | |
| "grad_norm": 0.5446850657463074, | |
| "learning_rate": 0.00019116502289838523, | |
| "loss": 0.0357, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.8953722334004024, | |
| "grad_norm": 0.49629637598991394, | |
| "learning_rate": 0.0001910277939835728, | |
| "loss": 0.0291, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.9004024144869215, | |
| "grad_norm": 0.3879849910736084, | |
| "learning_rate": 0.00019088955754943912, | |
| "loss": 0.0397, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.9054325955734407, | |
| "grad_norm": 0.4239087402820587, | |
| "learning_rate": 0.00019075031512602104, | |
| "loss": 0.0417, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.9104627766599598, | |
| "grad_norm": 0.5208977460861206, | |
| "learning_rate": 0.00019061006825448997, | |
| "loss": 0.0345, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.9154929577464789, | |
| "grad_norm": 0.28360894322395325, | |
| "learning_rate": 0.0001904688184871348, | |
| "loss": 0.0302, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.920523138832998, | |
| "grad_norm": 0.41871729493141174, | |
| "learning_rate": 0.00019032656738734467, | |
| "loss": 0.0381, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.9255533199195171, | |
| "grad_norm": 0.3086782693862915, | |
| "learning_rate": 0.00019018331652959184, | |
| "loss": 0.0397, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.9305835010060363, | |
| "grad_norm": 0.45297491550445557, | |
| "learning_rate": 0.00019003906749941405, | |
| "loss": 0.0347, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.9356136820925554, | |
| "grad_norm": 0.576064944267273, | |
| "learning_rate": 0.00018989382189339718, | |
| "loss": 0.033, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.9406438631790744, | |
| "grad_norm": 0.4694220721721649, | |
| "learning_rate": 0.00018974758131915732, | |
| "loss": 0.0279, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.9456740442655935, | |
| "grad_norm": 0.3088204264640808, | |
| "learning_rate": 0.00018960034739532336, | |
| "loss": 0.0344, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.9507042253521126, | |
| "grad_norm": 0.3804841637611389, | |
| "learning_rate": 0.00018945212175151856, | |
| "loss": 0.0421, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.9557344064386318, | |
| "grad_norm": 0.4735386073589325, | |
| "learning_rate": 0.00018930290602834298, | |
| "loss": 0.0434, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.9607645875251509, | |
| "grad_norm": 0.6512228846549988, | |
| "learning_rate": 0.0001891527018773551, | |
| "loss": 0.0332, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.96579476861167, | |
| "grad_norm": 0.3166397213935852, | |
| "learning_rate": 0.00018900151096105358, | |
| "loss": 0.0271, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.9708249496981891, | |
| "grad_norm": 0.36923766136169434, | |
| "learning_rate": 0.00018884933495285882, | |
| "loss": 0.0339, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.9758551307847082, | |
| "grad_norm": 0.43903985619544983, | |
| "learning_rate": 0.0001886961755370945, | |
| "loss": 0.0289, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.9808853118712274, | |
| "grad_norm": 0.4627651274204254, | |
| "learning_rate": 0.00018854203440896888, | |
| "loss": 0.0356, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.9859154929577465, | |
| "grad_norm": 0.42779436707496643, | |
| "learning_rate": 0.0001883869132745561, | |
| "loss": 0.0346, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.9909456740442656, | |
| "grad_norm": 0.4050188660621643, | |
| "learning_rate": 0.00018823081385077733, | |
| "loss": 0.0365, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.9959758551307847, | |
| "grad_norm": 0.28512129187583923, | |
| "learning_rate": 0.00018807373786538153, | |
| "loss": 0.0399, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.0010060362173039, | |
| "grad_norm": 0.3914794623851776, | |
| "learning_rate": 0.00018791568705692668, | |
| "loss": 0.034, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 1.0060362173038229, | |
| "grad_norm": 0.4772788882255554, | |
| "learning_rate": 0.00018775666317476022, | |
| "loss": 0.0308, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.011066398390342, | |
| "grad_norm": 0.2963551878929138, | |
| "learning_rate": 0.0001875966679789999, | |
| "loss": 0.0309, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 1.0160965794768613, | |
| "grad_norm": 0.17126326262950897, | |
| "learning_rate": 0.0001874357032405142, | |
| "loss": 0.0264, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 1.0211267605633803, | |
| "grad_norm": 0.3092329800128937, | |
| "learning_rate": 0.00018727377074090272, | |
| "loss": 0.0325, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 1.0261569416498995, | |
| "grad_norm": 0.23635587096214294, | |
| "learning_rate": 0.00018711087227247657, | |
| "loss": 0.0318, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 1.0311871227364184, | |
| "grad_norm": 0.4015599489212036, | |
| "learning_rate": 0.00018694700963823837, | |
| "loss": 0.0276, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.0362173038229376, | |
| "grad_norm": 0.3345971405506134, | |
| "learning_rate": 0.00018678218465186243, | |
| "loss": 0.0327, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 1.0412474849094568, | |
| "grad_norm": 0.5647363066673279, | |
| "learning_rate": 0.0001866163991376746, | |
| "loss": 0.0331, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 1.0462776659959758, | |
| "grad_norm": 0.3128892481327057, | |
| "learning_rate": 0.0001864496549306321, | |
| "loss": 0.0377, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 1.051307847082495, | |
| "grad_norm": 0.45699113607406616, | |
| "learning_rate": 0.00018628195387630325, | |
| "loss": 0.0285, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 1.056338028169014, | |
| "grad_norm": 0.2944326400756836, | |
| "learning_rate": 0.00018611329783084697, | |
| "loss": 0.0337, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.0613682092555332, | |
| "grad_norm": 0.5147525668144226, | |
| "learning_rate": 0.00018594368866099226, | |
| "loss": 0.0403, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 1.0663983903420524, | |
| "grad_norm": 0.3949408233165741, | |
| "learning_rate": 0.0001857731282440176, | |
| "loss": 0.0398, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 1.0714285714285714, | |
| "grad_norm": 0.6203738451004028, | |
| "learning_rate": 0.00018560161846773002, | |
| "loss": 0.0374, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 1.0764587525150906, | |
| "grad_norm": 0.5248531699180603, | |
| "learning_rate": 0.00018542916123044444, | |
| "loss": 0.035, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 1.0814889336016096, | |
| "grad_norm": 0.25072234869003296, | |
| "learning_rate": 0.00018525575844096243, | |
| "loss": 0.0315, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.0865191146881288, | |
| "grad_norm": 0.46970999240875244, | |
| "learning_rate": 0.00018508141201855125, | |
| "loss": 0.0473, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 1.091549295774648, | |
| "grad_norm": 0.4322168231010437, | |
| "learning_rate": 0.00018490612389292243, | |
| "loss": 0.0322, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 1.096579476861167, | |
| "grad_norm": 0.31416237354278564, | |
| "learning_rate": 0.0001847298960042106, | |
| "loss": 0.026, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 1.1016096579476862, | |
| "grad_norm": 0.4525179862976074, | |
| "learning_rate": 0.0001845527303029519, | |
| "loss": 0.0341, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 1.1066398390342052, | |
| "grad_norm": 0.3020547926425934, | |
| "learning_rate": 0.00018437462875006247, | |
| "loss": 0.0362, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.1116700201207244, | |
| "grad_norm": 0.24597640335559845, | |
| "learning_rate": 0.0001841955933168166, | |
| "loss": 0.0319, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 1.1167002012072436, | |
| "grad_norm": 0.5903005003929138, | |
| "learning_rate": 0.00018401562598482517, | |
| "loss": 0.0302, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 1.1217303822937625, | |
| "grad_norm": 0.3023013174533844, | |
| "learning_rate": 0.00018383472874601334, | |
| "loss": 0.0372, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 1.1267605633802817, | |
| "grad_norm": 0.39982765913009644, | |
| "learning_rate": 0.00018365290360259894, | |
| "loss": 0.0285, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 1.1317907444668007, | |
| "grad_norm": 0.32488811016082764, | |
| "learning_rate": 0.00018347015256706998, | |
| "loss": 0.0301, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.13682092555332, | |
| "grad_norm": 0.3721481263637543, | |
| "learning_rate": 0.00018328647766216246, | |
| "loss": 0.0466, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 1.1418511066398391, | |
| "grad_norm": 0.37539154291152954, | |
| "learning_rate": 0.00018310188092083803, | |
| "loss": 0.0305, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 1.1468812877263581, | |
| "grad_norm": 0.4233405590057373, | |
| "learning_rate": 0.00018291636438626152, | |
| "loss": 0.0383, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 1.1519114688128773, | |
| "grad_norm": 0.3756147623062134, | |
| "learning_rate": 0.00018272993011177822, | |
| "loss": 0.0339, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 1.1569416498993963, | |
| "grad_norm": 0.2825983762741089, | |
| "learning_rate": 0.00018254258016089123, | |
| "loss": 0.0323, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.1619718309859155, | |
| "grad_norm": 0.3585676848888397, | |
| "learning_rate": 0.00018235431660723855, | |
| "loss": 0.0257, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 1.1670020120724347, | |
| "grad_norm": 0.34615543484687805, | |
| "learning_rate": 0.00018216514153457025, | |
| "loss": 0.031, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 1.1720321931589537, | |
| "grad_norm": 0.3647988438606262, | |
| "learning_rate": 0.00018197505703672522, | |
| "loss": 0.0298, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 1.1770623742454729, | |
| "grad_norm": 0.2951864004135132, | |
| "learning_rate": 0.0001817840652176082, | |
| "loss": 0.0295, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 1.1820925553319919, | |
| "grad_norm": 0.24966521561145782, | |
| "learning_rate": 0.00018159216819116635, | |
| "loss": 0.0331, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.187122736418511, | |
| "grad_norm": 0.31395167112350464, | |
| "learning_rate": 0.00018139936808136593, | |
| "loss": 0.0299, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 1.1921529175050303, | |
| "grad_norm": 0.2532835006713867, | |
| "learning_rate": 0.00018120566702216877, | |
| "loss": 0.0327, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 1.1971830985915493, | |
| "grad_norm": 0.30898261070251465, | |
| "learning_rate": 0.00018101106715750855, | |
| "loss": 0.026, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 1.2022132796780685, | |
| "grad_norm": 0.34393829107284546, | |
| "learning_rate": 0.0001808155706412673, | |
| "loss": 0.0413, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 1.2072434607645874, | |
| "grad_norm": 0.32124215364456177, | |
| "learning_rate": 0.00018061917963725134, | |
| "loss": 0.0282, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.2122736418511066, | |
| "grad_norm": 0.4290277063846588, | |
| "learning_rate": 0.0001804218963191674, | |
| "loss": 0.0316, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 1.2173038229376258, | |
| "grad_norm": 0.43430641293525696, | |
| "learning_rate": 0.00018022372287059866, | |
| "loss": 0.031, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 1.2223340040241448, | |
| "grad_norm": 0.2853562831878662, | |
| "learning_rate": 0.00018002466148498035, | |
| "loss": 0.028, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 1.227364185110664, | |
| "grad_norm": 0.40154245495796204, | |
| "learning_rate": 0.00017982471436557579, | |
| "loss": 0.0309, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 1.232394366197183, | |
| "grad_norm": 0.45144811272621155, | |
| "learning_rate": 0.00017962388372545177, | |
| "loss": 0.0313, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.2374245472837022, | |
| "grad_norm": 0.310092031955719, | |
| "learning_rate": 0.00017942217178745396, | |
| "loss": 0.0257, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 1.2424547283702214, | |
| "grad_norm": 0.31381756067276, | |
| "learning_rate": 0.00017921958078418278, | |
| "loss": 0.0289, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 1.2474849094567404, | |
| "grad_norm": 0.2533479630947113, | |
| "learning_rate": 0.00017901611295796806, | |
| "loss": 0.0307, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 1.2525150905432596, | |
| "grad_norm": 0.2752733826637268, | |
| "learning_rate": 0.0001788117705608447, | |
| "loss": 0.0358, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 1.2575452716297786, | |
| "grad_norm": 0.40528246760368347, | |
| "learning_rate": 0.00017860655585452763, | |
| "loss": 0.0358, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.2625754527162978, | |
| "grad_norm": 0.37113526463508606, | |
| "learning_rate": 0.00017840047111038651, | |
| "loss": 0.0319, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 1.267605633802817, | |
| "grad_norm": 0.37338030338287354, | |
| "learning_rate": 0.00017819351860942105, | |
| "loss": 0.0292, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 1.272635814889336, | |
| "grad_norm": 0.4197677671909332, | |
| "learning_rate": 0.00017798570064223533, | |
| "loss": 0.0335, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 1.2776659959758552, | |
| "grad_norm": 0.3801611363887787, | |
| "learning_rate": 0.0001777770195090128, | |
| "loss": 0.0364, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 1.2826961770623742, | |
| "grad_norm": 0.49140316247940063, | |
| "learning_rate": 0.00017756747751949045, | |
| "loss": 0.0244, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.2877263581488934, | |
| "grad_norm": 0.37291839718818665, | |
| "learning_rate": 0.00017735707699293368, | |
| "loss": 0.0375, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 1.2927565392354126, | |
| "grad_norm": 0.27540504932403564, | |
| "learning_rate": 0.0001771458202581102, | |
| "loss": 0.036, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 1.2977867203219315, | |
| "grad_norm": 0.3908439576625824, | |
| "learning_rate": 0.00017693370965326467, | |
| "loss": 0.0343, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 1.3028169014084507, | |
| "grad_norm": 0.46150150895118713, | |
| "learning_rate": 0.00017672074752609235, | |
| "loss": 0.0277, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 1.3078470824949697, | |
| "grad_norm": 0.3451060950756073, | |
| "learning_rate": 0.0001765069362337136, | |
| "loss": 0.0366, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.312877263581489, | |
| "grad_norm": 0.5258057117462158, | |
| "learning_rate": 0.00017629227814264738, | |
| "loss": 0.0305, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 1.3179074446680081, | |
| "grad_norm": 0.3369615972042084, | |
| "learning_rate": 0.00017607677562878543, | |
| "loss": 0.0303, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 1.3229376257545271, | |
| "grad_norm": 0.48052117228507996, | |
| "learning_rate": 0.00017586043107736558, | |
| "loss": 0.0353, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 1.3279678068410463, | |
| "grad_norm": 0.44058990478515625, | |
| "learning_rate": 0.0001756432468829457, | |
| "loss": 0.0357, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 1.3329979879275653, | |
| "grad_norm": 0.3025529980659485, | |
| "learning_rate": 0.000175425225449377, | |
| "loss": 0.03, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.3380281690140845, | |
| "grad_norm": 0.3708013892173767, | |
| "learning_rate": 0.00017520636918977743, | |
| "loss": 0.0263, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 1.3430583501006037, | |
| "grad_norm": 0.41252729296684265, | |
| "learning_rate": 0.00017498668052650502, | |
| "loss": 0.0376, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 1.3480885311871227, | |
| "grad_norm": 0.4583453834056854, | |
| "learning_rate": 0.0001747661618911312, | |
| "loss": 0.035, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 1.3531187122736419, | |
| "grad_norm": 0.26764222979545593, | |
| "learning_rate": 0.00017454481572441353, | |
| "loss": 0.0258, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 1.3581488933601609, | |
| "grad_norm": 0.36722269654273987, | |
| "learning_rate": 0.000174322644476269, | |
| "loss": 0.025, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.36317907444668, | |
| "grad_norm": 0.2594809830188751, | |
| "learning_rate": 0.0001740996506057469, | |
| "loss": 0.0271, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 1.3682092555331993, | |
| "grad_norm": 0.36106789112091064, | |
| "learning_rate": 0.00017387583658100142, | |
| "loss": 0.0293, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 1.3732394366197183, | |
| "grad_norm": 0.3363661468029022, | |
| "learning_rate": 0.00017365120487926441, | |
| "loss": 0.0249, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 1.3782696177062375, | |
| "grad_norm": 0.275418758392334, | |
| "learning_rate": 0.000173425757986818, | |
| "loss": 0.028, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 1.3832997987927564, | |
| "grad_norm": 0.26659509539604187, | |
| "learning_rate": 0.00017319949839896722, | |
| "loss": 0.023, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.3883299798792756, | |
| "grad_norm": 0.461508184671402, | |
| "learning_rate": 0.00017297242862001198, | |
| "loss": 0.0302, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 1.3933601609657948, | |
| "grad_norm": 0.30697330832481384, | |
| "learning_rate": 0.0001727445511632197, | |
| "loss": 0.0252, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 1.3983903420523138, | |
| "grad_norm": 0.35531529784202576, | |
| "learning_rate": 0.00017251586855079746, | |
| "loss": 0.0251, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 1.403420523138833, | |
| "grad_norm": 0.4393905997276306, | |
| "learning_rate": 0.0001722863833138639, | |
| "loss": 0.0339, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 1.408450704225352, | |
| "grad_norm": 0.59965980052948, | |
| "learning_rate": 0.00017205609799242138, | |
| "loss": 0.0344, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.4134808853118712, | |
| "grad_norm": 0.2698121964931488, | |
| "learning_rate": 0.0001718250151353278, | |
| "loss": 0.0309, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 1.4185110663983904, | |
| "grad_norm": 0.5065796375274658, | |
| "learning_rate": 0.00017159313730026837, | |
| "loss": 0.0271, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 1.4235412474849094, | |
| "grad_norm": 0.39271390438079834, | |
| "learning_rate": 0.0001713604670537273, | |
| "loss": 0.0362, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 1.4285714285714286, | |
| "grad_norm": 0.2914319932460785, | |
| "learning_rate": 0.00017112700697095954, | |
| "loss": 0.036, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 1.4336016096579476, | |
| "grad_norm": 0.33919307589530945, | |
| "learning_rate": 0.00017089275963596195, | |
| "loss": 0.0303, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.4386317907444668, | |
| "grad_norm": 0.38898414373397827, | |
| "learning_rate": 0.0001706577276414451, | |
| "loss": 0.0336, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 1.443661971830986, | |
| "grad_norm": 0.3567386567592621, | |
| "learning_rate": 0.00017042191358880424, | |
| "loss": 0.0315, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 1.448692152917505, | |
| "grad_norm": 0.5339713096618652, | |
| "learning_rate": 0.00017018532008809074, | |
| "loss": 0.0299, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 1.4537223340040242, | |
| "grad_norm": 0.4281061887741089, | |
| "learning_rate": 0.00016994794975798305, | |
| "loss": 0.0327, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 1.4587525150905432, | |
| "grad_norm": 0.3700096905231476, | |
| "learning_rate": 0.00016970980522575775, | |
| "loss": 0.0299, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.4637826961770624, | |
| "grad_norm": 0.41299402713775635, | |
| "learning_rate": 0.00016947088912726052, | |
| "loss": 0.0347, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 1.4688128772635816, | |
| "grad_norm": 0.2901936173439026, | |
| "learning_rate": 0.00016923120410687695, | |
| "loss": 0.0297, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 1.4738430583501005, | |
| "grad_norm": 0.4472912549972534, | |
| "learning_rate": 0.00016899075281750326, | |
| "loss": 0.0274, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 1.4788732394366197, | |
| "grad_norm": 0.3437786102294922, | |
| "learning_rate": 0.00016874953792051693, | |
| "loss": 0.0353, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 1.4839034205231387, | |
| "grad_norm": 0.31530702114105225, | |
| "learning_rate": 0.00016850756208574717, | |
| "loss": 0.0369, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.488933601609658, | |
| "grad_norm": 0.37885403633117676, | |
| "learning_rate": 0.00016826482799144556, | |
| "loss": 0.0326, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 1.4939637826961771, | |
| "grad_norm": 0.3617626428604126, | |
| "learning_rate": 0.00016802133832425625, | |
| "loss": 0.0294, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 1.4989939637826961, | |
| "grad_norm": 0.34277963638305664, | |
| "learning_rate": 0.0001677770957791862, | |
| "loss": 0.0269, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 1.5040241448692153, | |
| "grad_norm": 0.46021127700805664, | |
| "learning_rate": 0.00016753210305957557, | |
| "loss": 0.0307, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 1.5090543259557343, | |
| "grad_norm": 0.5008649230003357, | |
| "learning_rate": 0.0001672863628770675, | |
| "loss": 0.0434, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.5140845070422535, | |
| "grad_norm": 0.2918837368488312, | |
| "learning_rate": 0.0001670398779515784, | |
| "loss": 0.0297, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 1.5191146881287727, | |
| "grad_norm": 0.30590662360191345, | |
| "learning_rate": 0.00016679265101126743, | |
| "loss": 0.03, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 1.524144869215292, | |
| "grad_norm": 0.29896557331085205, | |
| "learning_rate": 0.00016654468479250688, | |
| "loss": 0.0332, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 1.529175050301811, | |
| "grad_norm": 0.5691266059875488, | |
| "learning_rate": 0.00016629598203985135, | |
| "loss": 0.0312, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 1.5342052313883299, | |
| "grad_norm": 0.5821362733840942, | |
| "learning_rate": 0.00016604654550600762, | |
| "loss": 0.0329, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.539235412474849, | |
| "grad_norm": 0.3751659691333771, | |
| "learning_rate": 0.00016579637795180425, | |
| "loss": 0.029, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 1.5442655935613683, | |
| "grad_norm": 0.30066096782684326, | |
| "learning_rate": 0.0001655454821461608, | |
| "loss": 0.0207, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 1.5492957746478875, | |
| "grad_norm": 0.5700194239616394, | |
| "learning_rate": 0.00016529386086605737, | |
| "loss": 0.0464, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 1.5543259557344065, | |
| "grad_norm": 0.3803764581680298, | |
| "learning_rate": 0.00016504151689650386, | |
| "loss": 0.0295, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 1.5593561368209254, | |
| "grad_norm": 0.28992944955825806, | |
| "learning_rate": 0.0001647884530305089, | |
| "loss": 0.0339, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.5643863179074446, | |
| "grad_norm": 0.1968836933374405, | |
| "learning_rate": 0.00016453467206904935, | |
| "loss": 0.027, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 1.5694164989939638, | |
| "grad_norm": 0.4015430510044098, | |
| "learning_rate": 0.00016428017682103892, | |
| "loss": 0.0308, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 1.574446680080483, | |
| "grad_norm": 0.2642582356929779, | |
| "learning_rate": 0.00016402497010329725, | |
| "loss": 0.0293, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 1.579476861167002, | |
| "grad_norm": 0.446898877620697, | |
| "learning_rate": 0.00016376905474051873, | |
| "loss": 0.0274, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 1.584507042253521, | |
| "grad_norm": 0.4257184565067291, | |
| "learning_rate": 0.00016351243356524123, | |
| "loss": 0.032, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.5895372233400402, | |
| "grad_norm": 0.3191612660884857, | |
| "learning_rate": 0.00016325510941781474, | |
| "loss": 0.0286, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 1.5945674044265594, | |
| "grad_norm": 0.2913196086883545, | |
| "learning_rate": 0.00016299708514636992, | |
| "loss": 0.0331, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 1.5995975855130786, | |
| "grad_norm": 0.35634562373161316, | |
| "learning_rate": 0.00016273836360678652, | |
| "loss": 0.0289, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 1.6046277665995976, | |
| "grad_norm": 0.4977535009384155, | |
| "learning_rate": 0.00016247894766266196, | |
| "loss": 0.0308, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 1.6096579476861166, | |
| "grad_norm": 0.32202795147895813, | |
| "learning_rate": 0.0001622188401852794, | |
| "loss": 0.0297, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.6146881287726358, | |
| "grad_norm": 0.18511469662189484, | |
| "learning_rate": 0.00016195804405357613, | |
| "loss": 0.0243, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 1.619718309859155, | |
| "grad_norm": 0.2580728530883789, | |
| "learning_rate": 0.00016169656215411164, | |
| "loss": 0.0247, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 1.6247484909456742, | |
| "grad_norm": 0.46906596422195435, | |
| "learning_rate": 0.00016143439738103564, | |
| "loss": 0.0245, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 1.6297786720321932, | |
| "grad_norm": 0.39322683215141296, | |
| "learning_rate": 0.00016117155263605608, | |
| "loss": 0.0282, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 1.6348088531187122, | |
| "grad_norm": 0.2526596188545227, | |
| "learning_rate": 0.00016090803082840707, | |
| "loss": 0.025, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.6398390342052314, | |
| "grad_norm": 0.3361895978450775, | |
| "learning_rate": 0.00016064383487481655, | |
| "loss": 0.0208, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 1.6448692152917506, | |
| "grad_norm": 0.30806657671928406, | |
| "learning_rate": 0.0001603789676994741, | |
| "loss": 0.0289, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 1.6498993963782698, | |
| "grad_norm": 0.47636348009109497, | |
| "learning_rate": 0.00016011343223399865, | |
| "loss": 0.0232, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 1.6549295774647887, | |
| "grad_norm": 0.3957202732563019, | |
| "learning_rate": 0.00015984723141740576, | |
| "loss": 0.0338, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 1.6599597585513077, | |
| "grad_norm": 0.3279256820678711, | |
| "learning_rate": 0.0001595803681960754, | |
| "loss": 0.032, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.664989939637827, | |
| "grad_norm": 0.1826469749212265, | |
| "learning_rate": 0.00015931284552371918, | |
| "loss": 0.0202, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 1.6700201207243461, | |
| "grad_norm": 0.37201932072639465, | |
| "learning_rate": 0.00015904466636134772, | |
| "loss": 0.0272, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 1.6750503018108653, | |
| "grad_norm": 0.3043290078639984, | |
| "learning_rate": 0.00015877583367723773, | |
| "loss": 0.0313, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 1.6800804828973843, | |
| "grad_norm": 0.31204384565353394, | |
| "learning_rate": 0.00015850635044689938, | |
| "loss": 0.0275, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 1.6851106639839033, | |
| "grad_norm": 0.22221866250038147, | |
| "learning_rate": 0.00015823621965304325, | |
| "loss": 0.0254, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.6901408450704225, | |
| "grad_norm": 0.22048987448215485, | |
| "learning_rate": 0.00015796544428554725, | |
| "loss": 0.026, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 1.6951710261569417, | |
| "grad_norm": 0.34224534034729004, | |
| "learning_rate": 0.00015769402734142367, | |
| "loss": 0.0329, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 1.700201207243461, | |
| "grad_norm": 0.24496717751026154, | |
| "learning_rate": 0.00015742197182478586, | |
| "loss": 0.0233, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 1.70523138832998, | |
| "grad_norm": 0.3360805809497833, | |
| "learning_rate": 0.00015714928074681515, | |
| "loss": 0.0272, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 1.7102615694164989, | |
| "grad_norm": 0.21322770416736603, | |
| "learning_rate": 0.00015687595712572737, | |
| "loss": 0.0245, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.715291750503018, | |
| "grad_norm": 0.31249675154685974, | |
| "learning_rate": 0.00015660200398673948, | |
| "loss": 0.0254, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 1.7203219315895373, | |
| "grad_norm": 0.37619125843048096, | |
| "learning_rate": 0.00015632742436203615, | |
| "loss": 0.0292, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 1.7253521126760565, | |
| "grad_norm": 0.29316648840904236, | |
| "learning_rate": 0.00015605222129073617, | |
| "loss": 0.0264, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 1.7303822937625755, | |
| "grad_norm": 0.31073886156082153, | |
| "learning_rate": 0.0001557763978188588, | |
| "loss": 0.0322, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 1.7354124748490944, | |
| "grad_norm": 0.3250234127044678, | |
| "learning_rate": 0.00015549995699929, | |
| "loss": 0.0338, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.7404426559356136, | |
| "grad_norm": 0.3571024537086487, | |
| "learning_rate": 0.00015522290189174867, | |
| "loss": 0.028, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 1.7454728370221329, | |
| "grad_norm": 0.3457786440849304, | |
| "learning_rate": 0.000154945235562753, | |
| "loss": 0.0289, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 1.750503018108652, | |
| "grad_norm": 0.3116254508495331, | |
| "learning_rate": 0.00015466696108558611, | |
| "loss": 0.024, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 1.755533199195171, | |
| "grad_norm": 0.43166661262512207, | |
| "learning_rate": 0.00015438808154026238, | |
| "loss": 0.0305, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 1.76056338028169, | |
| "grad_norm": 0.3314637541770935, | |
| "learning_rate": 0.0001541086000134932, | |
| "loss": 0.023, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.7655935613682092, | |
| "grad_norm": 0.407803475856781, | |
| "learning_rate": 0.0001538285195986529, | |
| "loss": 0.029, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 1.7706237424547284, | |
| "grad_norm": 0.28635403513908386, | |
| "learning_rate": 0.0001535478433957444, | |
| "loss": 0.0302, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 1.7756539235412476, | |
| "grad_norm": 0.43160927295684814, | |
| "learning_rate": 0.000153266574511365, | |
| "loss": 0.0271, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 1.7806841046277666, | |
| "grad_norm": 0.4747959077358246, | |
| "learning_rate": 0.0001529847160586719, | |
| "loss": 0.034, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 1.7857142857142856, | |
| "grad_norm": 0.2701944410800934, | |
| "learning_rate": 0.0001527022711573479, | |
| "loss": 0.028, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.7907444668008048, | |
| "grad_norm": 0.20618535578250885, | |
| "learning_rate": 0.00015241924293356665, | |
| "loss": 0.0311, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 1.795774647887324, | |
| "grad_norm": 0.3267618417739868, | |
| "learning_rate": 0.0001521356345199582, | |
| "loss": 0.0358, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 1.8008048289738432, | |
| "grad_norm": 0.24747027456760406, | |
| "learning_rate": 0.0001518514490555743, | |
| "loss": 0.0296, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 1.8058350100603622, | |
| "grad_norm": 0.23700428009033203, | |
| "learning_rate": 0.00015156668968585368, | |
| "loss": 0.0313, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 1.8108651911468812, | |
| "grad_norm": 0.29978063702583313, | |
| "learning_rate": 0.0001512813595625872, | |
| "loss": 0.0321, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.8158953722334004, | |
| "grad_norm": 0.33228403329849243, | |
| "learning_rate": 0.00015099546184388287, | |
| "loss": 0.0305, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 1.8209255533199196, | |
| "grad_norm": 0.3012440502643585, | |
| "learning_rate": 0.00015070899969413115, | |
| "loss": 0.0264, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 1.8259557344064388, | |
| "grad_norm": 0.31137219071388245, | |
| "learning_rate": 0.00015042197628396966, | |
| "loss": 0.0276, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 1.8309859154929577, | |
| "grad_norm": 0.2960556745529175, | |
| "learning_rate": 0.0001501343947902482, | |
| "loss": 0.0244, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 1.8360160965794767, | |
| "grad_norm": 0.41761451959609985, | |
| "learning_rate": 0.0001498462583959936, | |
| "loss": 0.0271, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.841046277665996, | |
| "grad_norm": 0.2327389270067215, | |
| "learning_rate": 0.00014955757029037455, | |
| "loss": 0.0209, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 1.8460764587525151, | |
| "grad_norm": 0.4204856753349304, | |
| "learning_rate": 0.0001492683336686661, | |
| "loss": 0.0301, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 1.8511066398390343, | |
| "grad_norm": 0.31292521953582764, | |
| "learning_rate": 0.00014897855173221452, | |
| "loss": 0.0248, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 1.8561368209255533, | |
| "grad_norm": 0.304541677236557, | |
| "learning_rate": 0.00014868822768840162, | |
| "loss": 0.0237, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 1.8611670020120723, | |
| "grad_norm": 0.41588783264160156, | |
| "learning_rate": 0.00014839736475060966, | |
| "loss": 0.0268, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.8661971830985915, | |
| "grad_norm": 0.376002699136734, | |
| "learning_rate": 0.0001481059661381852, | |
| "loss": 0.0287, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 1.8712273641851107, | |
| "grad_norm": 0.3036446273326874, | |
| "learning_rate": 0.00014781403507640404, | |
| "loss": 0.0313, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 1.87625754527163, | |
| "grad_norm": 0.15947189927101135, | |
| "learning_rate": 0.00014752157479643515, | |
| "loss": 0.0269, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 1.881287726358149, | |
| "grad_norm": 0.3008073568344116, | |
| "learning_rate": 0.00014722858853530502, | |
| "loss": 0.031, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 1.8863179074446679, | |
| "grad_norm": 0.34803229570388794, | |
| "learning_rate": 0.00014693507953586192, | |
| "loss": 0.0289, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.891348088531187, | |
| "grad_norm": 0.47900477051734924, | |
| "learning_rate": 0.0001466410510467398, | |
| "loss": 0.0239, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 1.8963782696177063, | |
| "grad_norm": 0.3730434775352478, | |
| "learning_rate": 0.00014634650632232255, | |
| "loss": 0.0247, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 1.9014084507042255, | |
| "grad_norm": 0.2846371829509735, | |
| "learning_rate": 0.00014605144862270782, | |
| "loss": 0.0274, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 1.9064386317907445, | |
| "grad_norm": 0.27134642004966736, | |
| "learning_rate": 0.00014575588121367111, | |
| "loss": 0.0295, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 1.9114688128772634, | |
| "grad_norm": 0.2962862253189087, | |
| "learning_rate": 0.00014545980736662943, | |
| "loss": 0.029, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.9164989939637826, | |
| "grad_norm": 0.4870172441005707, | |
| "learning_rate": 0.00014516323035860515, | |
| "loss": 0.0334, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 1.9215291750503019, | |
| "grad_norm": 0.4422418475151062, | |
| "learning_rate": 0.00014486615347218985, | |
| "loss": 0.0236, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 1.926559356136821, | |
| "grad_norm": 0.15602879226207733, | |
| "learning_rate": 0.00014456857999550787, | |
| "loss": 0.025, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 1.93158953722334, | |
| "grad_norm": 0.3513808846473694, | |
| "learning_rate": 0.00014427051322217987, | |
| "loss": 0.0294, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 1.936619718309859, | |
| "grad_norm": 0.40983107686042786, | |
| "learning_rate": 0.00014397195645128645, | |
| "loss": 0.0249, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.9416498993963782, | |
| "grad_norm": 0.22102007269859314, | |
| "learning_rate": 0.00014367291298733178, | |
| "loss": 0.0296, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 1.9466800804828974, | |
| "grad_norm": 0.2670123279094696, | |
| "learning_rate": 0.00014337338614020672, | |
| "loss": 0.028, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 1.9517102615694166, | |
| "grad_norm": 0.3241349160671234, | |
| "learning_rate": 0.00014307337922515238, | |
| "loss": 0.0295, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 1.9567404426559356, | |
| "grad_norm": 0.2833192050457001, | |
| "learning_rate": 0.00014277289556272342, | |
| "loss": 0.0354, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 1.9617706237424546, | |
| "grad_norm": 0.31631726026535034, | |
| "learning_rate": 0.0001424719384787512, | |
| "loss": 0.0279, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.9668008048289738, | |
| "grad_norm": 0.38899293541908264, | |
| "learning_rate": 0.0001421705113043071, | |
| "loss": 0.0223, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 1.971830985915493, | |
| "grad_norm": 0.33592867851257324, | |
| "learning_rate": 0.00014186861737566558, | |
| "loss": 0.0256, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 1.9768611670020122, | |
| "grad_norm": 0.29194048047065735, | |
| "learning_rate": 0.00014156626003426724, | |
| "loss": 0.0285, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 1.9818913480885312, | |
| "grad_norm": 0.5011776089668274, | |
| "learning_rate": 0.00014126344262668176, | |
| "loss": 0.0367, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 1.9869215291750502, | |
| "grad_norm": 0.29997947812080383, | |
| "learning_rate": 0.00014096016850457117, | |
| "loss": 0.0283, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.9919517102615694, | |
| "grad_norm": 0.26518872380256653, | |
| "learning_rate": 0.0001406564410246523, | |
| "loss": 0.0367, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 1.9969818913480886, | |
| "grad_norm": 0.23170867562294006, | |
| "learning_rate": 0.00014035226354866, | |
| "loss": 0.0278, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 2.0020120724346078, | |
| "grad_norm": 0.25075894594192505, | |
| "learning_rate": 0.0001400476394433098, | |
| "loss": 0.0248, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 2.007042253521127, | |
| "grad_norm": 0.31776684522628784, | |
| "learning_rate": 0.00013974257208026054, | |
| "loss": 0.0239, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 2.0120724346076457, | |
| "grad_norm": 0.33745163679122925, | |
| "learning_rate": 0.00013943706483607725, | |
| "loss": 0.0221, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.017102615694165, | |
| "grad_norm": 0.21726344525814056, | |
| "learning_rate": 0.00013913112109219364, | |
| "loss": 0.0262, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 2.022132796780684, | |
| "grad_norm": 0.3767722249031067, | |
| "learning_rate": 0.0001388247442348747, | |
| "loss": 0.023, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 2.0271629778672033, | |
| "grad_norm": 0.4051326513290405, | |
| "learning_rate": 0.00013851793765517925, | |
| "loss": 0.0248, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 2.0321931589537225, | |
| "grad_norm": 0.2341754138469696, | |
| "learning_rate": 0.00013821070474892238, | |
| "loss": 0.0229, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 2.0372233400402413, | |
| "grad_norm": 0.30645281076431274, | |
| "learning_rate": 0.00013790304891663792, | |
| "loss": 0.0336, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.0422535211267605, | |
| "grad_norm": 0.37985026836395264, | |
| "learning_rate": 0.00013759497356354068, | |
| "loss": 0.025, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 2.0472837022132797, | |
| "grad_norm": 0.35980600118637085, | |
| "learning_rate": 0.00013728648209948886, | |
| "loss": 0.0344, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 2.052313883299799, | |
| "grad_norm": 0.23638148605823517, | |
| "learning_rate": 0.0001369775779389463, | |
| "loss": 0.029, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 2.057344064386318, | |
| "grad_norm": 0.21310564875602722, | |
| "learning_rate": 0.00013666826450094467, | |
| "loss": 0.0236, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 2.062374245472837, | |
| "grad_norm": 0.256998747587204, | |
| "learning_rate": 0.00013635854520904563, | |
| "loss": 0.023, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.067404426559356, | |
| "grad_norm": 0.23511482775211334, | |
| "learning_rate": 0.00013604842349130295, | |
| "loss": 0.0202, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 2.0724346076458753, | |
| "grad_norm": 0.3412628769874573, | |
| "learning_rate": 0.00013573790278022452, | |
| "loss": 0.023, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 2.0774647887323945, | |
| "grad_norm": 0.3615216314792633, | |
| "learning_rate": 0.0001354269865127344, | |
| "loss": 0.0217, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 2.0824949698189137, | |
| "grad_norm": 0.17151637375354767, | |
| "learning_rate": 0.0001351156781301348, | |
| "loss": 0.019, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 2.0875251509054324, | |
| "grad_norm": 0.29277974367141724, | |
| "learning_rate": 0.0001348039810780679, | |
| "loss": 0.0294, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.0925553319919517, | |
| "grad_norm": 0.2807733714580536, | |
| "learning_rate": 0.00013449189880647782, | |
| "loss": 0.0195, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 2.097585513078471, | |
| "grad_norm": 0.31441447138786316, | |
| "learning_rate": 0.00013417943476957248, | |
| "loss": 0.0203, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 2.10261569416499, | |
| "grad_norm": 0.3053465783596039, | |
| "learning_rate": 0.0001338665924257851, | |
| "loss": 0.0243, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 2.1076458752515093, | |
| "grad_norm": 0.22241270542144775, | |
| "learning_rate": 0.0001335533752377362, | |
| "loss": 0.0233, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 2.112676056338028, | |
| "grad_norm": 0.2186601608991623, | |
| "learning_rate": 0.00013323978667219513, | |
| "loss": 0.0215, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.1177062374245472, | |
| "grad_norm": 0.2129117101430893, | |
| "learning_rate": 0.00013292583020004184, | |
| "loss": 0.0236, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 2.1227364185110664, | |
| "grad_norm": 0.40424126386642456, | |
| "learning_rate": 0.00013261150929622822, | |
| "loss": 0.0286, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 2.1277665995975856, | |
| "grad_norm": 0.27489861845970154, | |
| "learning_rate": 0.00013229682743973992, | |
| "loss": 0.0241, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 2.132796780684105, | |
| "grad_norm": 0.20953650772571564, | |
| "learning_rate": 0.0001319817881135576, | |
| "loss": 0.0214, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 2.1378269617706236, | |
| "grad_norm": 0.3081333041191101, | |
| "learning_rate": 0.0001316663948046186, | |
| "loss": 0.0269, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.142857142857143, | |
| "grad_norm": 0.28377407789230347, | |
| "learning_rate": 0.00013135065100377814, | |
| "loss": 0.0207, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 2.147887323943662, | |
| "grad_norm": 0.28675928711891174, | |
| "learning_rate": 0.00013103456020577078, | |
| "loss": 0.0251, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 2.152917505030181, | |
| "grad_norm": 0.21341602504253387, | |
| "learning_rate": 0.00013071812590917175, | |
| "loss": 0.0244, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 2.1579476861167004, | |
| "grad_norm": 0.27167797088623047, | |
| "learning_rate": 0.0001304013516163583, | |
| "loss": 0.0348, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 2.162977867203219, | |
| "grad_norm": 0.2633381187915802, | |
| "learning_rate": 0.00013008424083347072, | |
| "loss": 0.0174, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.1680080482897384, | |
| "grad_norm": 0.30004456639289856, | |
| "learning_rate": 0.00012976679707037367, | |
| "loss": 0.0212, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 2.1730382293762576, | |
| "grad_norm": 0.2568664252758026, | |
| "learning_rate": 0.00012944902384061746, | |
| "loss": 0.0222, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 2.1780684104627768, | |
| "grad_norm": 0.4086577594280243, | |
| "learning_rate": 0.00012913092466139883, | |
| "loss": 0.0205, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 2.183098591549296, | |
| "grad_norm": 0.34458601474761963, | |
| "learning_rate": 0.00012881250305352233, | |
| "loss": 0.0166, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 2.1881287726358147, | |
| "grad_norm": 0.20635934174060822, | |
| "learning_rate": 0.00012849376254136125, | |
| "loss": 0.0232, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.193158953722334, | |
| "grad_norm": 0.39725396037101746, | |
| "learning_rate": 0.00012817470665281853, | |
| "loss": 0.0259, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 2.198189134808853, | |
| "grad_norm": 0.23941123485565186, | |
| "learning_rate": 0.0001278553389192878, | |
| "loss": 0.0243, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 2.2032193158953723, | |
| "grad_norm": 0.25489136576652527, | |
| "learning_rate": 0.00012753566287561428, | |
| "loss": 0.0234, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 2.2082494969818915, | |
| "grad_norm": 0.24230477213859558, | |
| "learning_rate": 0.00012721568206005562, | |
| "loss": 0.0283, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 2.2132796780684103, | |
| "grad_norm": 0.1920616328716278, | |
| "learning_rate": 0.00012689540001424284, | |
| "loss": 0.0176, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.2183098591549295, | |
| "grad_norm": 0.34041905403137207, | |
| "learning_rate": 0.00012657482028314096, | |
| "loss": 0.019, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 2.2233400402414487, | |
| "grad_norm": 0.16888068616390228, | |
| "learning_rate": 0.00012625394641500995, | |
| "loss": 0.0212, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 2.228370221327968, | |
| "grad_norm": 0.30867740511894226, | |
| "learning_rate": 0.00012593278196136525, | |
| "loss": 0.0209, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 2.233400402414487, | |
| "grad_norm": 0.2323102056980133, | |
| "learning_rate": 0.0001256113304769387, | |
| "loss": 0.0269, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 2.238430583501006, | |
| "grad_norm": 0.2643173038959503, | |
| "learning_rate": 0.00012528959551963897, | |
| "loss": 0.0204, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.243460764587525, | |
| "grad_norm": 0.26139459013938904, | |
| "learning_rate": 0.00012496758065051234, | |
| "loss": 0.0277, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 2.2484909456740443, | |
| "grad_norm": 0.5168742537498474, | |
| "learning_rate": 0.0001246452894337032, | |
| "loss": 0.0254, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 2.2535211267605635, | |
| "grad_norm": 0.4557676911354065, | |
| "learning_rate": 0.0001243227254364147, | |
| "loss": 0.0226, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 2.2585513078470827, | |
| "grad_norm": 0.3516198396682739, | |
| "learning_rate": 0.00012399989222886906, | |
| "loss": 0.0243, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 2.2635814889336014, | |
| "grad_norm": 0.20463688671588898, | |
| "learning_rate": 0.00012367679338426833, | |
| "loss": 0.0363, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.2686116700201207, | |
| "grad_norm": 0.384490042924881, | |
| "learning_rate": 0.00012335343247875456, | |
| "loss": 0.0248, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 2.27364185110664, | |
| "grad_norm": 0.24953685700893402, | |
| "learning_rate": 0.00012302981309137052, | |
| "loss": 0.0327, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 2.278672032193159, | |
| "grad_norm": 0.25503483414649963, | |
| "learning_rate": 0.00012270593880401974, | |
| "loss": 0.0289, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 2.2837022132796783, | |
| "grad_norm": 0.2341834008693695, | |
| "learning_rate": 0.00012238181320142722, | |
| "loss": 0.0222, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 2.288732394366197, | |
| "grad_norm": 0.27594631910324097, | |
| "learning_rate": 0.0001220574398710995, | |
| "loss": 0.0298, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.2937625754527162, | |
| "grad_norm": 0.26579129695892334, | |
| "learning_rate": 0.00012173282240328505, | |
| "loss": 0.0225, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 2.2987927565392354, | |
| "grad_norm": 0.2124788463115692, | |
| "learning_rate": 0.00012140796439093447, | |
| "loss": 0.0237, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 2.3038229376257546, | |
| "grad_norm": 0.2432093769311905, | |
| "learning_rate": 0.00012108286942966085, | |
| "loss": 0.0264, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 2.308853118712274, | |
| "grad_norm": 0.2657240629196167, | |
| "learning_rate": 0.00012075754111769984, | |
| "loss": 0.0193, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 2.3138832997987926, | |
| "grad_norm": 0.21074149012565613, | |
| "learning_rate": 0.00012043198305586986, | |
| "loss": 0.0176, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.318913480885312, | |
| "grad_norm": 0.29499852657318115, | |
| "learning_rate": 0.00012010619884753238, | |
| "loss": 0.0268, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 2.323943661971831, | |
| "grad_norm": 0.12746977806091309, | |
| "learning_rate": 0.00011978019209855174, | |
| "loss": 0.0241, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 2.32897384305835, | |
| "grad_norm": 0.18850328028202057, | |
| "learning_rate": 0.00011945396641725554, | |
| "loss": 0.0165, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 2.3340040241448694, | |
| "grad_norm": 0.26257720589637756, | |
| "learning_rate": 0.00011912752541439455, | |
| "loss": 0.0239, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 2.339034205231388, | |
| "grad_norm": 0.399151086807251, | |
| "learning_rate": 0.00011880087270310278, | |
| "loss": 0.0345, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.3440643863179074, | |
| "grad_norm": 0.308682382106781, | |
| "learning_rate": 0.00011847401189885745, | |
| "loss": 0.0268, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 2.3490945674044266, | |
| "grad_norm": 0.3351981043815613, | |
| "learning_rate": 0.00011814694661943906, | |
| "loss": 0.0262, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 2.3541247484909458, | |
| "grad_norm": 0.2986966371536255, | |
| "learning_rate": 0.00011781968048489127, | |
| "loss": 0.0199, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 2.359154929577465, | |
| "grad_norm": 0.18461820483207703, | |
| "learning_rate": 0.00011749221711748086, | |
| "loss": 0.0295, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 2.3641851106639837, | |
| "grad_norm": 0.33392488956451416, | |
| "learning_rate": 0.0001171645601416576, | |
| "loss": 0.0222, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.369215291750503, | |
| "grad_norm": 0.3629428446292877, | |
| "learning_rate": 0.0001168367131840142, | |
| "loss": 0.0228, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 2.374245472837022, | |
| "grad_norm": 0.3002311587333679, | |
| "learning_rate": 0.00011650867987324614, | |
| "loss": 0.0209, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 2.3792756539235413, | |
| "grad_norm": 0.24123568832874298, | |
| "learning_rate": 0.00011618046384011152, | |
| "loss": 0.0189, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 2.3843058350100605, | |
| "grad_norm": 0.3153541386127472, | |
| "learning_rate": 0.0001158520687173908, | |
| "loss": 0.0306, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 2.3893360160965793, | |
| "grad_norm": 0.36185353994369507, | |
| "learning_rate": 0.00011552349813984678, | |
| "loss": 0.0181, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.3943661971830985, | |
| "grad_norm": 0.21552976965904236, | |
| "learning_rate": 0.00011519475574418405, | |
| "loss": 0.0199, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 2.3993963782696177, | |
| "grad_norm": 0.3644528388977051, | |
| "learning_rate": 0.00011486584516900907, | |
| "loss": 0.0251, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 2.404426559356137, | |
| "grad_norm": 0.3115015923976898, | |
| "learning_rate": 0.00011453677005478968, | |
| "loss": 0.0343, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 2.409456740442656, | |
| "grad_norm": 0.41604316234588623, | |
| "learning_rate": 0.00011420753404381499, | |
| "loss": 0.03, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 2.414486921529175, | |
| "grad_norm": 0.3559543192386627, | |
| "learning_rate": 0.00011387814078015482, | |
| "loss": 0.0246, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.419517102615694, | |
| "grad_norm": 0.1890721470117569, | |
| "learning_rate": 0.00011354859390961958, | |
| "loss": 0.0209, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 2.4245472837022133, | |
| "grad_norm": 0.30657851696014404, | |
| "learning_rate": 0.00011321889707971979, | |
| "loss": 0.02, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 2.4295774647887325, | |
| "grad_norm": 0.2877327501773834, | |
| "learning_rate": 0.00011288905393962586, | |
| "loss": 0.0237, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 2.4346076458752517, | |
| "grad_norm": 0.3301834166049957, | |
| "learning_rate": 0.00011255906814012744, | |
| "loss": 0.0219, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 2.4396378269617705, | |
| "grad_norm": 0.21552981436252594, | |
| "learning_rate": 0.00011222894333359338, | |
| "loss": 0.0221, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.4446680080482897, | |
| "grad_norm": 0.33082151412963867, | |
| "learning_rate": 0.00011189868317393086, | |
| "loss": 0.021, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 2.449698189134809, | |
| "grad_norm": 0.30314934253692627, | |
| "learning_rate": 0.00011156829131654543, | |
| "loss": 0.0311, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 2.454728370221328, | |
| "grad_norm": 0.31226858496665955, | |
| "learning_rate": 0.00011123777141830008, | |
| "loss": 0.0272, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 2.4597585513078473, | |
| "grad_norm": 0.32737886905670166, | |
| "learning_rate": 0.00011090712713747514, | |
| "loss": 0.0248, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 2.464788732394366, | |
| "grad_norm": 0.2482103705406189, | |
| "learning_rate": 0.00011057636213372755, | |
| "loss": 0.0195, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.4698189134808852, | |
| "grad_norm": 0.4342176616191864, | |
| "learning_rate": 0.00011024548006805051, | |
| "loss": 0.0265, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 2.4748490945674044, | |
| "grad_norm": 0.3512166440486908, | |
| "learning_rate": 0.00010991448460273287, | |
| "loss": 0.0296, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 2.4798792756539236, | |
| "grad_norm": 0.4724549949169159, | |
| "learning_rate": 0.00010958337940131857, | |
| "loss": 0.0265, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 2.484909456740443, | |
| "grad_norm": 0.29837068915367126, | |
| "learning_rate": 0.00010925216812856621, | |
| "loss": 0.0249, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 2.4899396378269616, | |
| "grad_norm": 0.14961381256580353, | |
| "learning_rate": 0.00010892085445040836, | |
| "loss": 0.0198, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.494969818913481, | |
| "grad_norm": 0.2834800183773041, | |
| "learning_rate": 0.00010858944203391106, | |
| "loss": 0.0262, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 0.3048977255821228, | |
| "learning_rate": 0.00010825793454723325, | |
| "loss": 0.0193, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 2.505030181086519, | |
| "grad_norm": 0.42306941747665405, | |
| "learning_rate": 0.00010792633565958603, | |
| "loss": 0.0164, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 2.5100603621730384, | |
| "grad_norm": 0.29460790753364563, | |
| "learning_rate": 0.00010759464904119229, | |
| "loss": 0.0219, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 2.515090543259557, | |
| "grad_norm": 0.2361808717250824, | |
| "learning_rate": 0.00010726287836324582, | |
| "loss": 0.0229, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.5201207243460764, | |
| "grad_norm": 0.2822340130805969, | |
| "learning_rate": 0.00010693102729787088, | |
| "loss": 0.03, | |
| "step": 5010 | |
| }, | |
| { | |
| "epoch": 2.5251509054325956, | |
| "grad_norm": 0.3465881049633026, | |
| "learning_rate": 0.00010659909951808145, | |
| "loss": 0.0208, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 2.5301810865191148, | |
| "grad_norm": 0.27967023849487305, | |
| "learning_rate": 0.00010626709869774056, | |
| "loss": 0.0257, | |
| "step": 5030 | |
| }, | |
| { | |
| "epoch": 2.535211267605634, | |
| "grad_norm": 0.4234648644924164, | |
| "learning_rate": 0.00010593502851151977, | |
| "loss": 0.0245, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 2.5402414486921527, | |
| "grad_norm": 0.2795238792896271, | |
| "learning_rate": 0.00010560289263485836, | |
| "loss": 0.0201, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 2.545271629778672, | |
| "grad_norm": 0.27361616492271423, | |
| "learning_rate": 0.00010527069474392266, | |
| "loss": 0.0152, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 2.550301810865191, | |
| "grad_norm": 0.21000875532627106, | |
| "learning_rate": 0.00010493843851556539, | |
| "loss": 0.0203, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 2.5553319919517103, | |
| "grad_norm": 0.265424519777298, | |
| "learning_rate": 0.00010460612762728498, | |
| "loss": 0.0227, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 2.5603621730382295, | |
| "grad_norm": 0.2341059446334839, | |
| "learning_rate": 0.00010427376575718488, | |
| "loss": 0.0257, | |
| "step": 5090 | |
| }, | |
| { | |
| "epoch": 2.5653923541247483, | |
| "grad_norm": 0.3186612129211426, | |
| "learning_rate": 0.00010394135658393278, | |
| "loss": 0.0249, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.5704225352112675, | |
| "grad_norm": 0.22894106805324554, | |
| "learning_rate": 0.00010360890378671997, | |
| "loss": 0.0222, | |
| "step": 5110 | |
| }, | |
| { | |
| "epoch": 2.5754527162977867, | |
| "grad_norm": 0.32082125544548035, | |
| "learning_rate": 0.00010327641104522052, | |
| "loss": 0.0258, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 2.580482897384306, | |
| "grad_norm": 0.21953271329402924, | |
| "learning_rate": 0.00010294388203955067, | |
| "loss": 0.0199, | |
| "step": 5130 | |
| }, | |
| { | |
| "epoch": 2.585513078470825, | |
| "grad_norm": 0.14508609473705292, | |
| "learning_rate": 0.00010261132045022804, | |
| "loss": 0.0219, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 2.590543259557344, | |
| "grad_norm": 0.23425619304180145, | |
| "learning_rate": 0.00010227872995813083, | |
| "loss": 0.0238, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.595573440643863, | |
| "grad_norm": 0.18086619675159454, | |
| "learning_rate": 0.00010194611424445721, | |
| "loss": 0.0225, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 2.6006036217303823, | |
| "grad_norm": 0.2463824599981308, | |
| "learning_rate": 0.0001016134769906845, | |
| "loss": 0.0199, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 2.6056338028169015, | |
| "grad_norm": 0.23146170377731323, | |
| "learning_rate": 0.00010128082187852846, | |
| "loss": 0.0215, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 2.6106639839034207, | |
| "grad_norm": 0.24659287929534912, | |
| "learning_rate": 0.00010094815258990241, | |
| "loss": 0.0192, | |
| "step": 5190 | |
| }, | |
| { | |
| "epoch": 2.6156941649899395, | |
| "grad_norm": 0.18079569935798645, | |
| "learning_rate": 0.00010061547280687664, | |
| "loss": 0.0201, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.6207243460764587, | |
| "grad_norm": 0.21298180520534515, | |
| "learning_rate": 0.00010028278621163762, | |
| "loss": 0.0208, | |
| "step": 5210 | |
| }, | |
| { | |
| "epoch": 2.625754527162978, | |
| "grad_norm": 0.21694862842559814, | |
| "learning_rate": 9.99500964864472e-05, | |
| "loss": 0.0151, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 2.630784708249497, | |
| "grad_norm": 0.2606973648071289, | |
| "learning_rate": 9.961740731360184e-05, | |
| "loss": 0.0219, | |
| "step": 5230 | |
| }, | |
| { | |
| "epoch": 2.6358148893360163, | |
| "grad_norm": 0.34560972452163696, | |
| "learning_rate": 9.928472237539196e-05, | |
| "loss": 0.0188, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 2.640845070422535, | |
| "grad_norm": 0.26653993129730225, | |
| "learning_rate": 9.895204535406104e-05, | |
| "loss": 0.0161, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.6458752515090542, | |
| "grad_norm": 0.16782517731189728, | |
| "learning_rate": 9.861937993176495e-05, | |
| "loss": 0.0167, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 2.6509054325955734, | |
| "grad_norm": 0.2072647660970688, | |
| "learning_rate": 9.828672979053119e-05, | |
| "loss": 0.0185, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 2.6559356136820926, | |
| "grad_norm": 0.34195080399513245, | |
| "learning_rate": 9.795409861221813e-05, | |
| "loss": 0.0156, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 2.660965794768612, | |
| "grad_norm": 0.3683469891548157, | |
| "learning_rate": 9.762149007847424e-05, | |
| "loss": 0.0166, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 2.6659959758551306, | |
| "grad_norm": 0.2584269940853119, | |
| "learning_rate": 9.728890787069737e-05, | |
| "loss": 0.0241, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 2.67102615694165, | |
| "grad_norm": 0.1505228877067566, | |
| "learning_rate": 9.695635566999397e-05, | |
| "loss": 0.0286, | |
| "step": 5310 | |
| }, | |
| { | |
| "epoch": 2.676056338028169, | |
| "grad_norm": 0.27530187368392944, | |
| "learning_rate": 9.662383715713837e-05, | |
| "loss": 0.0243, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 2.681086519114688, | |
| "grad_norm": 0.3398008942604065, | |
| "learning_rate": 9.629135601253204e-05, | |
| "loss": 0.0286, | |
| "step": 5330 | |
| }, | |
| { | |
| "epoch": 2.6861167002012074, | |
| "grad_norm": 0.20086489617824554, | |
| "learning_rate": 9.595891591616282e-05, | |
| "loss": 0.0192, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 2.691146881287726, | |
| "grad_norm": 0.25084254145622253, | |
| "learning_rate": 9.562652054756429e-05, | |
| "loss": 0.0277, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 2.6961770623742454, | |
| "grad_norm": 0.25037065148353577, | |
| "learning_rate": 9.529417358577489e-05, | |
| "loss": 0.0214, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 2.7012072434607646, | |
| "grad_norm": 0.18306289613246918, | |
| "learning_rate": 9.496187870929732e-05, | |
| "loss": 0.0157, | |
| "step": 5370 | |
| }, | |
| { | |
| "epoch": 2.7062374245472838, | |
| "grad_norm": 0.17593130469322205, | |
| "learning_rate": 9.462963959605778e-05, | |
| "loss": 0.0206, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 2.711267605633803, | |
| "grad_norm": 0.3547375500202179, | |
| "learning_rate": 9.429745992336522e-05, | |
| "loss": 0.0217, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 2.7162977867203217, | |
| "grad_norm": 0.26571425795555115, | |
| "learning_rate": 9.396534336787081e-05, | |
| "loss": 0.0198, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 2.721327967806841, | |
| "grad_norm": 0.3025442957878113, | |
| "learning_rate": 9.363329360552703e-05, | |
| "loss": 0.021, | |
| "step": 5410 | |
| }, | |
| { | |
| "epoch": 2.72635814889336, | |
| "grad_norm": 0.30082857608795166, | |
| "learning_rate": 9.330131431154708e-05, | |
| "loss": 0.0214, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 2.7313883299798793, | |
| "grad_norm": 0.18924900889396667, | |
| "learning_rate": 9.29694091603642e-05, | |
| "loss": 0.0141, | |
| "step": 5430 | |
| }, | |
| { | |
| "epoch": 2.7364185110663986, | |
| "grad_norm": 0.28385961055755615, | |
| "learning_rate": 9.263758182559103e-05, | |
| "loss": 0.0235, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 2.7414486921529173, | |
| "grad_norm": 0.3015013635158539, | |
| "learning_rate": 9.230583597997888e-05, | |
| "loss": 0.0234, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 2.7464788732394365, | |
| "grad_norm": 0.3755188584327698, | |
| "learning_rate": 9.197417529537716e-05, | |
| "loss": 0.0206, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 2.7515090543259557, | |
| "grad_norm": 0.2895854413509369, | |
| "learning_rate": 9.16426034426926e-05, | |
| "loss": 0.0266, | |
| "step": 5470 | |
| }, | |
| { | |
| "epoch": 2.756539235412475, | |
| "grad_norm": 0.35291990637779236, | |
| "learning_rate": 9.131112409184886e-05, | |
| "loss": 0.0173, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 2.761569416498994, | |
| "grad_norm": 0.2543307840824127, | |
| "learning_rate": 9.097974091174568e-05, | |
| "loss": 0.0173, | |
| "step": 5490 | |
| }, | |
| { | |
| "epoch": 2.766599597585513, | |
| "grad_norm": 0.14231809973716736, | |
| "learning_rate": 9.064845757021833e-05, | |
| "loss": 0.0162, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.771629778672032, | |
| "grad_norm": 0.23540771007537842, | |
| "learning_rate": 9.031727773399709e-05, | |
| "loss": 0.0187, | |
| "step": 5510 | |
| }, | |
| { | |
| "epoch": 2.7766599597585513, | |
| "grad_norm": 0.2745067775249481, | |
| "learning_rate": 8.998620506866663e-05, | |
| "loss": 0.0195, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 2.7816901408450705, | |
| "grad_norm": 0.22975705564022064, | |
| "learning_rate": 8.965524323862535e-05, | |
| "loss": 0.0201, | |
| "step": 5530 | |
| }, | |
| { | |
| "epoch": 2.7867203219315897, | |
| "grad_norm": 0.2695607841014862, | |
| "learning_rate": 8.9324395907045e-05, | |
| "loss": 0.0152, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 2.7917505030181085, | |
| "grad_norm": 0.3059997260570526, | |
| "learning_rate": 8.899366673582994e-05, | |
| "loss": 0.0205, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 2.7967806841046277, | |
| "grad_norm": 0.2964552938938141, | |
| "learning_rate": 8.866305938557675e-05, | |
| "loss": 0.0216, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 2.801810865191147, | |
| "grad_norm": 0.23506313562393188, | |
| "learning_rate": 8.833257751553365e-05, | |
| "loss": 0.0223, | |
| "step": 5570 | |
| }, | |
| { | |
| "epoch": 2.806841046277666, | |
| "grad_norm": 0.2811063230037689, | |
| "learning_rate": 8.800222478356e-05, | |
| "loss": 0.0153, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 2.8118712273641853, | |
| "grad_norm": 0.1559114307165146, | |
| "learning_rate": 8.767200484608584e-05, | |
| "loss": 0.0235, | |
| "step": 5590 | |
| }, | |
| { | |
| "epoch": 2.816901408450704, | |
| "grad_norm": 0.3868955075740814, | |
| "learning_rate": 8.734192135807142e-05, | |
| "loss": 0.0238, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 2.8219315895372232, | |
| "grad_norm": 0.3004298210144043, | |
| "learning_rate": 8.701197797296671e-05, | |
| "loss": 0.0205, | |
| "step": 5610 | |
| }, | |
| { | |
| "epoch": 2.8269617706237424, | |
| "grad_norm": 0.30548134446144104, | |
| "learning_rate": 8.668217834267096e-05, | |
| "loss": 0.0213, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 2.8319919517102616, | |
| "grad_norm": 0.4484052360057831, | |
| "learning_rate": 8.635252611749239e-05, | |
| "loss": 0.015, | |
| "step": 5630 | |
| }, | |
| { | |
| "epoch": 2.837022132796781, | |
| "grad_norm": 0.2639616131782532, | |
| "learning_rate": 8.602302494610764e-05, | |
| "loss": 0.0241, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 2.8420523138832996, | |
| "grad_norm": 0.290503591299057, | |
| "learning_rate": 8.569367847552143e-05, | |
| "loss": 0.0215, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 2.847082494969819, | |
| "grad_norm": 0.2626757025718689, | |
| "learning_rate": 8.536449035102624e-05, | |
| "loss": 0.0249, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 2.852112676056338, | |
| "grad_norm": 0.18214431405067444, | |
| "learning_rate": 8.50354642161619e-05, | |
| "loss": 0.0155, | |
| "step": 5670 | |
| }, | |
| { | |
| "epoch": 2.857142857142857, | |
| "grad_norm": 0.23727771639823914, | |
| "learning_rate": 8.47066037126754e-05, | |
| "loss": 0.027, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 2.8621730382293764, | |
| "grad_norm": 0.2826462984085083, | |
| "learning_rate": 8.437791248048037e-05, | |
| "loss": 0.0189, | |
| "step": 5690 | |
| }, | |
| { | |
| "epoch": 2.867203219315895, | |
| "grad_norm": 0.2718562185764313, | |
| "learning_rate": 8.404939415761693e-05, | |
| "loss": 0.0235, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 2.8722334004024144, | |
| "grad_norm": 0.27361899614334106, | |
| "learning_rate": 8.372105238021138e-05, | |
| "loss": 0.0235, | |
| "step": 5710 | |
| }, | |
| { | |
| "epoch": 2.8772635814889336, | |
| "grad_norm": 0.18473175168037415, | |
| "learning_rate": 8.339289078243602e-05, | |
| "loss": 0.0158, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 2.8822937625754528, | |
| "grad_norm": 0.24013389647006989, | |
| "learning_rate": 8.306491299646884e-05, | |
| "loss": 0.0215, | |
| "step": 5730 | |
| }, | |
| { | |
| "epoch": 2.887323943661972, | |
| "grad_norm": 0.22806933522224426, | |
| "learning_rate": 8.273712265245336e-05, | |
| "loss": 0.0148, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 2.8923541247484907, | |
| "grad_norm": 0.23373086750507355, | |
| "learning_rate": 8.240952337845844e-05, | |
| "loss": 0.0224, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.89738430583501, | |
| "grad_norm": 0.19605842232704163, | |
| "learning_rate": 8.208211880043812e-05, | |
| "loss": 0.0146, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 2.902414486921529, | |
| "grad_norm": 0.2649703621864319, | |
| "learning_rate": 8.175491254219151e-05, | |
| "loss": 0.0168, | |
| "step": 5770 | |
| }, | |
| { | |
| "epoch": 2.9074446680080483, | |
| "grad_norm": 0.1927386075258255, | |
| "learning_rate": 8.142790822532266e-05, | |
| "loss": 0.017, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 2.9124748490945676, | |
| "grad_norm": 0.18656544387340546, | |
| "learning_rate": 8.110110946920049e-05, | |
| "loss": 0.0198, | |
| "step": 5790 | |
| }, | |
| { | |
| "epoch": 2.9175050301810863, | |
| "grad_norm": 0.1952710896730423, | |
| "learning_rate": 8.077451989091868e-05, | |
| "loss": 0.0172, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 2.9225352112676055, | |
| "grad_norm": 0.31542059779167175, | |
| "learning_rate": 8.044814310525574e-05, | |
| "loss": 0.024, | |
| "step": 5810 | |
| }, | |
| { | |
| "epoch": 2.9275653923541247, | |
| "grad_norm": 0.22629360854625702, | |
| "learning_rate": 8.012198272463486e-05, | |
| "loss": 0.0234, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 2.932595573440644, | |
| "grad_norm": 0.2527533769607544, | |
| "learning_rate": 7.979604235908406e-05, | |
| "loss": 0.0192, | |
| "step": 5830 | |
| }, | |
| { | |
| "epoch": 2.937625754527163, | |
| "grad_norm": 0.27879974246025085, | |
| "learning_rate": 7.947032561619617e-05, | |
| "loss": 0.0162, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 2.942655935613682, | |
| "grad_norm": 0.33129119873046875, | |
| "learning_rate": 7.914483610108888e-05, | |
| "loss": 0.0196, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 2.947686116700201, | |
| "grad_norm": 0.29584264755249023, | |
| "learning_rate": 7.881957741636486e-05, | |
| "loss": 0.0226, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 2.9527162977867203, | |
| "grad_norm": 0.3168732821941376, | |
| "learning_rate": 7.849455316207197e-05, | |
| "loss": 0.0178, | |
| "step": 5870 | |
| }, | |
| { | |
| "epoch": 2.9577464788732395, | |
| "grad_norm": 0.1575564593076706, | |
| "learning_rate": 7.816976693566324e-05, | |
| "loss": 0.0221, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 2.9627766599597587, | |
| "grad_norm": 0.2546273469924927, | |
| "learning_rate": 7.784522233195716e-05, | |
| "loss": 0.0167, | |
| "step": 5890 | |
| }, | |
| { | |
| "epoch": 2.9678068410462775, | |
| "grad_norm": 0.19285175204277039, | |
| "learning_rate": 7.75209229430979e-05, | |
| "loss": 0.0196, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 2.9728370221327967, | |
| "grad_norm": 0.25114384293556213, | |
| "learning_rate": 7.719687235851554e-05, | |
| "loss": 0.0224, | |
| "step": 5910 | |
| }, | |
| { | |
| "epoch": 2.977867203219316, | |
| "grad_norm": 0.2990531623363495, | |
| "learning_rate": 7.687307416488625e-05, | |
| "loss": 0.0185, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 2.982897384305835, | |
| "grad_norm": 0.19075430929660797, | |
| "learning_rate": 7.654953194609281e-05, | |
| "loss": 0.0238, | |
| "step": 5930 | |
| }, | |
| { | |
| "epoch": 2.9879275653923543, | |
| "grad_norm": 0.28794053196907043, | |
| "learning_rate": 7.62262492831847e-05, | |
| "loss": 0.0198, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 2.992957746478873, | |
| "grad_norm": 0.26047641038894653, | |
| "learning_rate": 7.590322975433857e-05, | |
| "loss": 0.014, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 2.9979879275653922, | |
| "grad_norm": 0.2440265715122223, | |
| "learning_rate": 7.558047693481868e-05, | |
| "loss": 0.017, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 3.0030181086519114, | |
| "grad_norm": 0.21606865525245667, | |
| "learning_rate": 7.525799439693723e-05, | |
| "loss": 0.0173, | |
| "step": 5970 | |
| }, | |
| { | |
| "epoch": 3.0080482897384306, | |
| "grad_norm": 0.24635550379753113, | |
| "learning_rate": 7.493578571001497e-05, | |
| "loss": 0.0117, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 3.01307847082495, | |
| "grad_norm": 0.17602556943893433, | |
| "learning_rate": 7.461385444034145e-05, | |
| "loss": 0.0149, | |
| "step": 5990 | |
| }, | |
| { | |
| "epoch": 3.0181086519114686, | |
| "grad_norm": 0.2443585842847824, | |
| "learning_rate": 7.429220415113582e-05, | |
| "loss": 0.02, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.023138832997988, | |
| "grad_norm": 0.25802722573280334, | |
| "learning_rate": 7.39708384025072e-05, | |
| "loss": 0.0185, | |
| "step": 6010 | |
| }, | |
| { | |
| "epoch": 3.028169014084507, | |
| "grad_norm": 0.1731959879398346, | |
| "learning_rate": 7.364976075141534e-05, | |
| "loss": 0.0159, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 3.033199195171026, | |
| "grad_norm": 0.2295977920293808, | |
| "learning_rate": 7.33289747516313e-05, | |
| "loss": 0.0223, | |
| "step": 6030 | |
| }, | |
| { | |
| "epoch": 3.0382293762575454, | |
| "grad_norm": 0.27841848134994507, | |
| "learning_rate": 7.300848395369801e-05, | |
| "loss": 0.0246, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 3.043259557344064, | |
| "grad_norm": 0.2819521725177765, | |
| "learning_rate": 7.268829190489111e-05, | |
| "loss": 0.0175, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 3.0482897384305834, | |
| "grad_norm": 0.32469579577445984, | |
| "learning_rate": 7.236840214917949e-05, | |
| "loss": 0.0179, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 3.0533199195171026, | |
| "grad_norm": 0.28703153133392334, | |
| "learning_rate": 7.204881822718625e-05, | |
| "loss": 0.0191, | |
| "step": 6070 | |
| }, | |
| { | |
| "epoch": 3.058350100603622, | |
| "grad_norm": 0.27047568559646606, | |
| "learning_rate": 7.172954367614945e-05, | |
| "loss": 0.0162, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 3.063380281690141, | |
| "grad_norm": 0.2722102403640747, | |
| "learning_rate": 7.141058202988293e-05, | |
| "loss": 0.0189, | |
| "step": 6090 | |
| }, | |
| { | |
| "epoch": 3.0684104627766597, | |
| "grad_norm": 0.2391163408756256, | |
| "learning_rate": 7.109193681873726e-05, | |
| "loss": 0.0175, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 3.073440643863179, | |
| "grad_norm": 0.4041613042354584, | |
| "learning_rate": 7.077361156956055e-05, | |
| "loss": 0.021, | |
| "step": 6110 | |
| }, | |
| { | |
| "epoch": 3.078470824949698, | |
| "grad_norm": 0.2219713032245636, | |
| "learning_rate": 7.045560980565957e-05, | |
| "loss": 0.0169, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 3.0835010060362174, | |
| "grad_norm": 0.21718035638332367, | |
| "learning_rate": 7.013793504676061e-05, | |
| "loss": 0.0194, | |
| "step": 6130 | |
| }, | |
| { | |
| "epoch": 3.0885311871227366, | |
| "grad_norm": 0.19979757070541382, | |
| "learning_rate": 6.982059080897059e-05, | |
| "loss": 0.0165, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 3.0935613682092553, | |
| "grad_norm": 0.2233651578426361, | |
| "learning_rate": 6.950358060473814e-05, | |
| "loss": 0.0151, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 3.0985915492957745, | |
| "grad_norm": 0.24543243646621704, | |
| "learning_rate": 6.918690794281475e-05, | |
| "loss": 0.019, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 3.1036217303822937, | |
| "grad_norm": 0.3978748917579651, | |
| "learning_rate": 6.887057632821587e-05, | |
| "loss": 0.0216, | |
| "step": 6170 | |
| }, | |
| { | |
| "epoch": 3.108651911468813, | |
| "grad_norm": 0.22909055650234222, | |
| "learning_rate": 6.855458926218219e-05, | |
| "loss": 0.0206, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 3.113682092555332, | |
| "grad_norm": 0.32850995659828186, | |
| "learning_rate": 6.823895024214077e-05, | |
| "loss": 0.0145, | |
| "step": 6190 | |
| }, | |
| { | |
| "epoch": 3.118712273641851, | |
| "grad_norm": 0.22510278224945068, | |
| "learning_rate": 6.79236627616665e-05, | |
| "loss": 0.0215, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 3.12374245472837, | |
| "grad_norm": 0.2886231541633606, | |
| "learning_rate": 6.760873031044328e-05, | |
| "loss": 0.0152, | |
| "step": 6210 | |
| }, | |
| { | |
| "epoch": 3.1287726358148893, | |
| "grad_norm": 0.17671062052249908, | |
| "learning_rate": 6.729415637422543e-05, | |
| "loss": 0.0242, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 3.1338028169014085, | |
| "grad_norm": 0.3519156873226166, | |
| "learning_rate": 6.69799444347992e-05, | |
| "loss": 0.0158, | |
| "step": 6230 | |
| }, | |
| { | |
| "epoch": 3.1388329979879277, | |
| "grad_norm": 0.2459878772497177, | |
| "learning_rate": 6.666609796994416e-05, | |
| "loss": 0.0175, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 3.1438631790744465, | |
| "grad_norm": 0.20523589849472046, | |
| "learning_rate": 6.635262045339465e-05, | |
| "loss": 0.0174, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 3.1488933601609657, | |
| "grad_norm": 0.23958848416805267, | |
| "learning_rate": 6.603951535480146e-05, | |
| "loss": 0.0215, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 3.153923541247485, | |
| "grad_norm": 0.21609632670879364, | |
| "learning_rate": 6.572678613969331e-05, | |
| "loss": 0.0205, | |
| "step": 6270 | |
| }, | |
| { | |
| "epoch": 3.158953722334004, | |
| "grad_norm": 0.20294173061847687, | |
| "learning_rate": 6.541443626943855e-05, | |
| "loss": 0.0177, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 3.1639839034205233, | |
| "grad_norm": 0.21324007213115692, | |
| "learning_rate": 6.51024692012069e-05, | |
| "loss": 0.0187, | |
| "step": 6290 | |
| }, | |
| { | |
| "epoch": 3.169014084507042, | |
| "grad_norm": 0.42404302954673767, | |
| "learning_rate": 6.479088838793106e-05, | |
| "loss": 0.0205, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 3.1740442655935612, | |
| "grad_norm": 0.2945919334888458, | |
| "learning_rate": 6.447969727826859e-05, | |
| "loss": 0.0148, | |
| "step": 6310 | |
| }, | |
| { | |
| "epoch": 3.1790744466800804, | |
| "grad_norm": 0.3685127794742584, | |
| "learning_rate": 6.41688993165637e-05, | |
| "loss": 0.0194, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 3.1841046277665996, | |
| "grad_norm": 0.295194536447525, | |
| "learning_rate": 6.385849794280915e-05, | |
| "loss": 0.0202, | |
| "step": 6330 | |
| }, | |
| { | |
| "epoch": 3.189134808853119, | |
| "grad_norm": 0.2186952531337738, | |
| "learning_rate": 6.354849659260815e-05, | |
| "loss": 0.0172, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 3.1941649899396376, | |
| "grad_norm": 0.22984078526496887, | |
| "learning_rate": 6.323889869713637e-05, | |
| "loss": 0.015, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 3.199195171026157, | |
| "grad_norm": 0.2033555507659912, | |
| "learning_rate": 6.292970768310387e-05, | |
| "loss": 0.0145, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 3.204225352112676, | |
| "grad_norm": 0.3300851583480835, | |
| "learning_rate": 6.262092697271732e-05, | |
| "loss": 0.0202, | |
| "step": 6370 | |
| }, | |
| { | |
| "epoch": 3.209255533199195, | |
| "grad_norm": 0.2983047366142273, | |
| "learning_rate": 6.2312559983642e-05, | |
| "loss": 0.0174, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 3.2142857142857144, | |
| "grad_norm": 0.1934136003255844, | |
| "learning_rate": 6.200461012896402e-05, | |
| "loss": 0.0233, | |
| "step": 6390 | |
| }, | |
| { | |
| "epoch": 3.219315895372233, | |
| "grad_norm": 0.23245446383953094, | |
| "learning_rate": 6.169708081715253e-05, | |
| "loss": 0.018, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 3.2243460764587524, | |
| "grad_norm": 0.24971647560596466, | |
| "learning_rate": 6.1389975452022e-05, | |
| "loss": 0.0167, | |
| "step": 6410 | |
| }, | |
| { | |
| "epoch": 3.2293762575452716, | |
| "grad_norm": 0.23975920677185059, | |
| "learning_rate": 6.108329743269461e-05, | |
| "loss": 0.014, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 3.234406438631791, | |
| "grad_norm": 0.14767713844776154, | |
| "learning_rate": 6.07770501535625e-05, | |
| "loss": 0.019, | |
| "step": 6430 | |
| }, | |
| { | |
| "epoch": 3.23943661971831, | |
| "grad_norm": 0.22770993411540985, | |
| "learning_rate": 6.047123700425026e-05, | |
| "loss": 0.0128, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 3.2444668008048287, | |
| "grad_norm": 0.2288396656513214, | |
| "learning_rate": 6.016586136957745e-05, | |
| "loss": 0.0173, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 3.249496981891348, | |
| "grad_norm": 0.1313876062631607, | |
| "learning_rate": 5.98609266295211e-05, | |
| "loss": 0.0131, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 3.254527162977867, | |
| "grad_norm": 0.18414969742298126, | |
| "learning_rate": 5.955643615917825e-05, | |
| "loss": 0.016, | |
| "step": 6470 | |
| }, | |
| { | |
| "epoch": 3.2595573440643864, | |
| "grad_norm": 0.19654971361160278, | |
| "learning_rate": 5.925239332872878e-05, | |
| "loss": 0.0128, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 3.2645875251509056, | |
| "grad_norm": 0.1665862649679184, | |
| "learning_rate": 5.894880150339783e-05, | |
| "loss": 0.0148, | |
| "step": 6490 | |
| }, | |
| { | |
| "epoch": 3.2696177062374243, | |
| "grad_norm": 0.1726180911064148, | |
| "learning_rate": 5.864566404341878e-05, | |
| "loss": 0.0147, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 3.2746478873239435, | |
| "grad_norm": 0.2508947253227234, | |
| "learning_rate": 5.834298430399594e-05, | |
| "loss": 0.0212, | |
| "step": 6510 | |
| }, | |
| { | |
| "epoch": 3.2796780684104627, | |
| "grad_norm": 0.18195697665214539, | |
| "learning_rate": 5.804076563526744e-05, | |
| "loss": 0.0163, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 3.284708249496982, | |
| "grad_norm": 0.38305532932281494, | |
| "learning_rate": 5.773901138226826e-05, | |
| "loss": 0.0133, | |
| "step": 6530 | |
| }, | |
| { | |
| "epoch": 3.289738430583501, | |
| "grad_norm": 0.19784590601921082, | |
| "learning_rate": 5.743772488489294e-05, | |
| "loss": 0.0191, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 3.29476861167002, | |
| "grad_norm": 0.11469227820634842, | |
| "learning_rate": 5.71369094778589e-05, | |
| "loss": 0.0188, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 3.299798792756539, | |
| "grad_norm": 0.2073812335729599, | |
| "learning_rate": 5.6836568490669384e-05, | |
| "loss": 0.0117, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 3.3048289738430583, | |
| "grad_norm": 0.12024883180856705, | |
| "learning_rate": 5.653670524757667e-05, | |
| "loss": 0.0167, | |
| "step": 6570 | |
| }, | |
| { | |
| "epoch": 3.3098591549295775, | |
| "grad_norm": 0.1882588267326355, | |
| "learning_rate": 5.623732306754511e-05, | |
| "loss": 0.0123, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 3.3148893360160967, | |
| "grad_norm": 0.13178786635398865, | |
| "learning_rate": 5.5938425264214657e-05, | |
| "loss": 0.0156, | |
| "step": 6590 | |
| }, | |
| { | |
| "epoch": 3.3199195171026155, | |
| "grad_norm": 0.1943099945783615, | |
| "learning_rate": 5.564001514586403e-05, | |
| "loss": 0.0117, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 3.3249496981891347, | |
| "grad_norm": 0.2532840073108673, | |
| "learning_rate": 5.534209601537407e-05, | |
| "loss": 0.0157, | |
| "step": 6610 | |
| }, | |
| { | |
| "epoch": 3.329979879275654, | |
| "grad_norm": 0.216691792011261, | |
| "learning_rate": 5.50446711701913e-05, | |
| "loss": 0.0195, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 3.335010060362173, | |
| "grad_norm": 0.1876860409975052, | |
| "learning_rate": 5.474774390229129e-05, | |
| "loss": 0.0152, | |
| "step": 6630 | |
| }, | |
| { | |
| "epoch": 3.3400402414486923, | |
| "grad_norm": 0.11754748225212097, | |
| "learning_rate": 5.4451317498142365e-05, | |
| "loss": 0.0159, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 3.345070422535211, | |
| "grad_norm": 0.5310174822807312, | |
| "learning_rate": 5.4155395238669185e-05, | |
| "loss": 0.0155, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 3.3501006036217302, | |
| "grad_norm": 0.3709560036659241, | |
| "learning_rate": 5.385998039921627e-05, | |
| "loss": 0.0194, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 3.3551307847082494, | |
| "grad_norm": 0.25374695658683777, | |
| "learning_rate": 5.3565076249512034e-05, | |
| "loss": 0.0184, | |
| "step": 6670 | |
| }, | |
| { | |
| "epoch": 3.3601609657947686, | |
| "grad_norm": 0.2403491884469986, | |
| "learning_rate": 5.3270686053632323e-05, | |
| "loss": 0.0153, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 3.365191146881288, | |
| "grad_norm": 0.23633185029029846, | |
| "learning_rate": 5.29768130699645e-05, | |
| "loss": 0.0161, | |
| "step": 6690 | |
| }, | |
| { | |
| "epoch": 3.3702213279678066, | |
| "grad_norm": 0.3900575637817383, | |
| "learning_rate": 5.268346055117129e-05, | |
| "loss": 0.0258, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 3.375251509054326, | |
| "grad_norm": 0.141703262925148, | |
| "learning_rate": 5.239063174415466e-05, | |
| "loss": 0.0121, | |
| "step": 6710 | |
| }, | |
| { | |
| "epoch": 3.380281690140845, | |
| "grad_norm": 0.17671038210391998, | |
| "learning_rate": 5.209832989002015e-05, | |
| "loss": 0.0149, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 3.385311871227364, | |
| "grad_norm": 0.24591059982776642, | |
| "learning_rate": 5.18065582240407e-05, | |
| "loss": 0.0234, | |
| "step": 6730 | |
| }, | |
| { | |
| "epoch": 3.3903420523138834, | |
| "grad_norm": 0.26220864057540894, | |
| "learning_rate": 5.151531997562116e-05, | |
| "loss": 0.0255, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 3.395372233400402, | |
| "grad_norm": 0.29695531725883484, | |
| "learning_rate": 5.122461836826218e-05, | |
| "loss": 0.019, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 3.4004024144869214, | |
| "grad_norm": 0.12897849082946777, | |
| "learning_rate": 5.0934456619524896e-05, | |
| "loss": 0.0191, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 3.4054325955734406, | |
| "grad_norm": 0.16902518272399902, | |
| "learning_rate": 5.064483794099508e-05, | |
| "loss": 0.0207, | |
| "step": 6770 | |
| }, | |
| { | |
| "epoch": 3.41046277665996, | |
| "grad_norm": 0.20852209627628326, | |
| "learning_rate": 5.0355765538247636e-05, | |
| "loss": 0.0146, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 3.415492957746479, | |
| "grad_norm": 0.16347025334835052, | |
| "learning_rate": 5.006724261081118e-05, | |
| "loss": 0.0186, | |
| "step": 6790 | |
| }, | |
| { | |
| "epoch": 3.4205231388329977, | |
| "grad_norm": 0.18849849700927734, | |
| "learning_rate": 4.977927235213259e-05, | |
| "loss": 0.0217, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 3.425553319919517, | |
| "grad_norm": 0.30719512701034546, | |
| "learning_rate": 4.9491857949541696e-05, | |
| "loss": 0.0176, | |
| "step": 6810 | |
| }, | |
| { | |
| "epoch": 3.430583501006036, | |
| "grad_norm": 0.21589145064353943, | |
| "learning_rate": 4.9205002584215855e-05, | |
| "loss": 0.015, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 3.4356136820925554, | |
| "grad_norm": 0.23759104311466217, | |
| "learning_rate": 4.891870943114496e-05, | |
| "loss": 0.0198, | |
| "step": 6830 | |
| }, | |
| { | |
| "epoch": 3.4406438631790746, | |
| "grad_norm": 0.26880696415901184, | |
| "learning_rate": 4.86329816590962e-05, | |
| "loss": 0.0142, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 3.4456740442655933, | |
| "grad_norm": 0.192754328250885, | |
| "learning_rate": 4.8347822430578856e-05, | |
| "loss": 0.0166, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 3.4507042253521125, | |
| "grad_norm": 0.12232445925474167, | |
| "learning_rate": 4.80632349018096e-05, | |
| "loss": 0.0175, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 3.4557344064386317, | |
| "grad_norm": 0.1528378129005432, | |
| "learning_rate": 4.777922222267721e-05, | |
| "loss": 0.0134, | |
| "step": 6870 | |
| }, | |
| { | |
| "epoch": 3.460764587525151, | |
| "grad_norm": 0.37869587540626526, | |
| "learning_rate": 4.749578753670799e-05, | |
| "loss": 0.0209, | |
| "step": 6880 | |
| }, | |
| { | |
| "epoch": 3.46579476861167, | |
| "grad_norm": 0.24779389798641205, | |
| "learning_rate": 4.721293398103086e-05, | |
| "loss": 0.0167, | |
| "step": 6890 | |
| }, | |
| { | |
| "epoch": 3.470824949698189, | |
| "grad_norm": 0.1252845823764801, | |
| "learning_rate": 4.6930664686342526e-05, | |
| "loss": 0.0118, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 3.475855130784708, | |
| "grad_norm": 0.28004470467567444, | |
| "learning_rate": 4.664898277687313e-05, | |
| "loss": 0.0229, | |
| "step": 6910 | |
| }, | |
| { | |
| "epoch": 3.4808853118712273, | |
| "grad_norm": 0.1652926206588745, | |
| "learning_rate": 4.636789137035129e-05, | |
| "loss": 0.0201, | |
| "step": 6920 | |
| }, | |
| { | |
| "epoch": 3.4859154929577465, | |
| "grad_norm": 0.12315156310796738, | |
| "learning_rate": 4.6087393577969926e-05, | |
| "loss": 0.0133, | |
| "step": 6930 | |
| }, | |
| { | |
| "epoch": 3.4909456740442657, | |
| "grad_norm": 0.2795407772064209, | |
| "learning_rate": 4.580749250435158e-05, | |
| "loss": 0.0148, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 3.4959758551307845, | |
| "grad_norm": 0.25727009773254395, | |
| "learning_rate": 4.5528191247514226e-05, | |
| "loss": 0.0161, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 3.5010060362173037, | |
| "grad_norm": 0.2040313482284546, | |
| "learning_rate": 4.524949289883692e-05, | |
| "loss": 0.0169, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 3.506036217303823, | |
| "grad_norm": 0.21510617434978485, | |
| "learning_rate": 4.497140054302548e-05, | |
| "loss": 0.019, | |
| "step": 6970 | |
| }, | |
| { | |
| "epoch": 3.511066398390342, | |
| "grad_norm": 0.22986271977424622, | |
| "learning_rate": 4.469391725807854e-05, | |
| "loss": 0.0177, | |
| "step": 6980 | |
| }, | |
| { | |
| "epoch": 3.5160965794768613, | |
| "grad_norm": 0.1664620190858841, | |
| "learning_rate": 4.4417046115253304e-05, | |
| "loss": 0.0207, | |
| "step": 6990 | |
| }, | |
| { | |
| "epoch": 3.52112676056338, | |
| "grad_norm": 0.24035824835300446, | |
| "learning_rate": 4.414079017903166e-05, | |
| "loss": 0.021, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.5261569416498992, | |
| "grad_norm": 0.2778620719909668, | |
| "learning_rate": 4.386515250708627e-05, | |
| "loss": 0.0179, | |
| "step": 7010 | |
| }, | |
| { | |
| "epoch": 3.5311871227364184, | |
| "grad_norm": 0.2031363546848297, | |
| "learning_rate": 4.3590136150246555e-05, | |
| "loss": 0.0151, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 3.5362173038229376, | |
| "grad_norm": 0.29513317346572876, | |
| "learning_rate": 4.3315744152465276e-05, | |
| "loss": 0.0166, | |
| "step": 7030 | |
| }, | |
| { | |
| "epoch": 3.541247484909457, | |
| "grad_norm": 0.2097848355770111, | |
| "learning_rate": 4.3041979550784395e-05, | |
| "loss": 0.0142, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 3.5462776659959756, | |
| "grad_norm": 0.21217374503612518, | |
| "learning_rate": 4.276884537530187e-05, | |
| "loss": 0.0131, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 3.551307847082495, | |
| "grad_norm": 0.18939641118049622, | |
| "learning_rate": 4.2496344649137776e-05, | |
| "loss": 0.0166, | |
| "step": 7060 | |
| }, | |
| { | |
| "epoch": 3.556338028169014, | |
| "grad_norm": 0.17117512226104736, | |
| "learning_rate": 4.222448038840113e-05, | |
| "loss": 0.0187, | |
| "step": 7070 | |
| }, | |
| { | |
| "epoch": 3.561368209255533, | |
| "grad_norm": 0.15176957845687866, | |
| "learning_rate": 4.1953255602156394e-05, | |
| "loss": 0.0193, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 3.5663983903420524, | |
| "grad_norm": 0.3019493818283081, | |
| "learning_rate": 4.168267329239002e-05, | |
| "loss": 0.0171, | |
| "step": 7090 | |
| }, | |
| { | |
| "epoch": 3.571428571428571, | |
| "grad_norm": 0.18399718403816223, | |
| "learning_rate": 4.141273645397754e-05, | |
| "loss": 0.018, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 3.5764587525150904, | |
| "grad_norm": 0.16979122161865234, | |
| "learning_rate": 4.114344807465007e-05, | |
| "loss": 0.0209, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 3.5814889336016096, | |
| "grad_norm": 0.2553984820842743, | |
| "learning_rate": 4.087481113496159e-05, | |
| "loss": 0.0123, | |
| "step": 7120 | |
| }, | |
| { | |
| "epoch": 3.586519114688129, | |
| "grad_norm": 0.20643579959869385, | |
| "learning_rate": 4.060682860825559e-05, | |
| "loss": 0.0159, | |
| "step": 7130 | |
| }, | |
| { | |
| "epoch": 3.591549295774648, | |
| "grad_norm": 0.20078648626804352, | |
| "learning_rate": 4.033950346063248e-05, | |
| "loss": 0.0142, | |
| "step": 7140 | |
| }, | |
| { | |
| "epoch": 3.5965794768611667, | |
| "grad_norm": 0.13958579301834106, | |
| "learning_rate": 4.007283865091662e-05, | |
| "loss": 0.0153, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 3.6016096579476864, | |
| "grad_norm": 0.21849416196346283, | |
| "learning_rate": 3.98068371306235e-05, | |
| "loss": 0.022, | |
| "step": 7160 | |
| }, | |
| { | |
| "epoch": 3.606639839034205, | |
| "grad_norm": 0.2273043692111969, | |
| "learning_rate": 3.954150184392723e-05, | |
| "loss": 0.0141, | |
| "step": 7170 | |
| }, | |
| { | |
| "epoch": 3.6116700201207244, | |
| "grad_norm": 0.18715119361877441, | |
| "learning_rate": 3.927683572762778e-05, | |
| "loss": 0.0149, | |
| "step": 7180 | |
| }, | |
| { | |
| "epoch": 3.6167002012072436, | |
| "grad_norm": 0.27733615040779114, | |
| "learning_rate": 3.9012841711118677e-05, | |
| "loss": 0.0177, | |
| "step": 7190 | |
| }, | |
| { | |
| "epoch": 3.6217303822937623, | |
| "grad_norm": 0.22101663053035736, | |
| "learning_rate": 3.874952271635444e-05, | |
| "loss": 0.0177, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 3.626760563380282, | |
| "grad_norm": 0.21932683885097504, | |
| "learning_rate": 3.848688165781819e-05, | |
| "loss": 0.0126, | |
| "step": 7210 | |
| }, | |
| { | |
| "epoch": 3.6317907444668007, | |
| "grad_norm": 0.18129754066467285, | |
| "learning_rate": 3.82249214424896e-05, | |
| "loss": 0.0133, | |
| "step": 7220 | |
| }, | |
| { | |
| "epoch": 3.63682092555332, | |
| "grad_norm": 0.2246370166540146, | |
| "learning_rate": 3.796364496981247e-05, | |
| "loss": 0.0148, | |
| "step": 7230 | |
| }, | |
| { | |
| "epoch": 3.641851106639839, | |
| "grad_norm": 0.22842873632907867, | |
| "learning_rate": 3.7703055131662854e-05, | |
| "loss": 0.0129, | |
| "step": 7240 | |
| }, | |
| { | |
| "epoch": 3.646881287726358, | |
| "grad_norm": 0.2930648922920227, | |
| "learning_rate": 3.744315481231694e-05, | |
| "loss": 0.0167, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 3.6519114688128775, | |
| "grad_norm": 0.17970743775367737, | |
| "learning_rate": 3.7183946888419066e-05, | |
| "loss": 0.0135, | |
| "step": 7260 | |
| }, | |
| { | |
| "epoch": 3.6569416498993963, | |
| "grad_norm": 0.342587411403656, | |
| "learning_rate": 3.692543422895004e-05, | |
| "loss": 0.0183, | |
| "step": 7270 | |
| }, | |
| { | |
| "epoch": 3.6619718309859155, | |
| "grad_norm": 0.17181113362312317, | |
| "learning_rate": 3.6667619695195285e-05, | |
| "loss": 0.0112, | |
| "step": 7280 | |
| }, | |
| { | |
| "epoch": 3.6670020120724347, | |
| "grad_norm": 0.14279602468013763, | |
| "learning_rate": 3.6410506140713216e-05, | |
| "loss": 0.0139, | |
| "step": 7290 | |
| }, | |
| { | |
| "epoch": 3.6720321931589535, | |
| "grad_norm": 0.15404970943927765, | |
| "learning_rate": 3.615409641130351e-05, | |
| "loss": 0.0197, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 3.677062374245473, | |
| "grad_norm": 0.21196570992469788, | |
| "learning_rate": 3.589839334497587e-05, | |
| "loss": 0.0114, | |
| "step": 7310 | |
| }, | |
| { | |
| "epoch": 3.682092555331992, | |
| "grad_norm": 0.14234349131584167, | |
| "learning_rate": 3.564339977191834e-05, | |
| "loss": 0.0122, | |
| "step": 7320 | |
| }, | |
| { | |
| "epoch": 3.687122736418511, | |
| "grad_norm": 0.3660185933113098, | |
| "learning_rate": 3.538911851446619e-05, | |
| "loss": 0.0153, | |
| "step": 7330 | |
| }, | |
| { | |
| "epoch": 3.6921529175050303, | |
| "grad_norm": 0.20586389303207397, | |
| "learning_rate": 3.5135552387070636e-05, | |
| "loss": 0.0167, | |
| "step": 7340 | |
| }, | |
| { | |
| "epoch": 3.697183098591549, | |
| "grad_norm": 0.19840209186077118, | |
| "learning_rate": 3.48827041962675e-05, | |
| "loss": 0.0131, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 3.7022132796780687, | |
| "grad_norm": 0.3007209897041321, | |
| "learning_rate": 3.463057674064646e-05, | |
| "loss": 0.0223, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 3.7072434607645874, | |
| "grad_norm": 0.20882894098758698, | |
| "learning_rate": 3.437917281081975e-05, | |
| "loss": 0.0204, | |
| "step": 7370 | |
| }, | |
| { | |
| "epoch": 3.7122736418511066, | |
| "grad_norm": 0.19728295505046844, | |
| "learning_rate": 3.412849518939155e-05, | |
| "loss": 0.018, | |
| "step": 7380 | |
| }, | |
| { | |
| "epoch": 3.717303822937626, | |
| "grad_norm": 0.2532462179660797, | |
| "learning_rate": 3.387854665092709e-05, | |
| "loss": 0.0161, | |
| "step": 7390 | |
| }, | |
| { | |
| "epoch": 3.7223340040241446, | |
| "grad_norm": 0.1997775435447693, | |
| "learning_rate": 3.3629329961921765e-05, | |
| "loss": 0.0112, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 3.7273641851106643, | |
| "grad_norm": 0.17711062729358673, | |
| "learning_rate": 3.338084788077085e-05, | |
| "loss": 0.0156, | |
| "step": 7410 | |
| }, | |
| { | |
| "epoch": 3.732394366197183, | |
| "grad_norm": 0.1401996910572052, | |
| "learning_rate": 3.313310315773864e-05, | |
| "loss": 0.0281, | |
| "step": 7420 | |
| }, | |
| { | |
| "epoch": 3.737424547283702, | |
| "grad_norm": 0.1834188550710678, | |
| "learning_rate": 3.288609853492827e-05, | |
| "loss": 0.0162, | |
| "step": 7430 | |
| }, | |
| { | |
| "epoch": 3.7424547283702214, | |
| "grad_norm": 0.19655165076255798, | |
| "learning_rate": 3.2639836746251216e-05, | |
| "loss": 0.0124, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 3.74748490945674, | |
| "grad_norm": 0.17778749763965607, | |
| "learning_rate": 3.2394320517397015e-05, | |
| "loss": 0.0153, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 3.75251509054326, | |
| "grad_norm": 0.16552262008190155, | |
| "learning_rate": 3.214955256580327e-05, | |
| "loss": 0.017, | |
| "step": 7460 | |
| }, | |
| { | |
| "epoch": 3.7575452716297786, | |
| "grad_norm": 0.19796250760555267, | |
| "learning_rate": 3.1905535600625314e-05, | |
| "loss": 0.0182, | |
| "step": 7470 | |
| }, | |
| { | |
| "epoch": 3.762575452716298, | |
| "grad_norm": 0.19010251760482788, | |
| "learning_rate": 3.166227232270651e-05, | |
| "loss": 0.0182, | |
| "step": 7480 | |
| }, | |
| { | |
| "epoch": 3.767605633802817, | |
| "grad_norm": 0.21966631710529327, | |
| "learning_rate": 3.141976542454806e-05, | |
| "loss": 0.018, | |
| "step": 7490 | |
| }, | |
| { | |
| "epoch": 3.7726358148893357, | |
| "grad_norm": 0.14321380853652954, | |
| "learning_rate": 3.117801759027959e-05, | |
| "loss": 0.0139, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.7776659959758554, | |
| "grad_norm": 0.22963641583919525, | |
| "learning_rate": 3.093703149562892e-05, | |
| "loss": 0.0112, | |
| "step": 7510 | |
| }, | |
| { | |
| "epoch": 3.782696177062374, | |
| "grad_norm": 0.15116992592811584, | |
| "learning_rate": 3.069680980789294e-05, | |
| "loss": 0.0193, | |
| "step": 7520 | |
| }, | |
| { | |
| "epoch": 3.7877263581488934, | |
| "grad_norm": 0.14807961881160736, | |
| "learning_rate": 3.0457355185907877e-05, | |
| "loss": 0.0222, | |
| "step": 7530 | |
| }, | |
| { | |
| "epoch": 3.7927565392354126, | |
| "grad_norm": 0.19666004180908203, | |
| "learning_rate": 3.0218670280019745e-05, | |
| "loss": 0.0138, | |
| "step": 7540 | |
| }, | |
| { | |
| "epoch": 3.7977867203219313, | |
| "grad_norm": 0.22170531749725342, | |
| "learning_rate": 2.9980757732055277e-05, | |
| "loss": 0.0185, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 3.802816901408451, | |
| "grad_norm": 0.1371658444404602, | |
| "learning_rate": 2.974362017529242e-05, | |
| "loss": 0.0135, | |
| "step": 7560 | |
| }, | |
| { | |
| "epoch": 3.8078470824949697, | |
| "grad_norm": 0.1677546203136444, | |
| "learning_rate": 2.9507260234431444e-05, | |
| "loss": 0.0096, | |
| "step": 7570 | |
| }, | |
| { | |
| "epoch": 3.812877263581489, | |
| "grad_norm": 0.2982484996318817, | |
| "learning_rate": 2.9271680525565724e-05, | |
| "loss": 0.0155, | |
| "step": 7580 | |
| }, | |
| { | |
| "epoch": 3.817907444668008, | |
| "grad_norm": 0.23418201506137848, | |
| "learning_rate": 2.9036883656152734e-05, | |
| "loss": 0.0151, | |
| "step": 7590 | |
| }, | |
| { | |
| "epoch": 3.822937625754527, | |
| "grad_norm": 0.30079197883605957, | |
| "learning_rate": 2.8802872224985434e-05, | |
| "loss": 0.0167, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 3.8279678068410465, | |
| "grad_norm": 0.16451403498649597, | |
| "learning_rate": 2.8569648822163185e-05, | |
| "loss": 0.0113, | |
| "step": 7610 | |
| }, | |
| { | |
| "epoch": 3.8329979879275653, | |
| "grad_norm": 0.28625065088272095, | |
| "learning_rate": 2.8337216029063395e-05, | |
| "loss": 0.0146, | |
| "step": 7620 | |
| }, | |
| { | |
| "epoch": 3.8380281690140845, | |
| "grad_norm": 0.22233416140079498, | |
| "learning_rate": 2.810557641831266e-05, | |
| "loss": 0.0201, | |
| "step": 7630 | |
| }, | |
| { | |
| "epoch": 3.8430583501006037, | |
| "grad_norm": 0.10642603039741516, | |
| "learning_rate": 2.787473255375853e-05, | |
| "loss": 0.0125, | |
| "step": 7640 | |
| }, | |
| { | |
| "epoch": 3.8480885311871225, | |
| "grad_norm": 0.2190830409526825, | |
| "learning_rate": 2.7644686990441027e-05, | |
| "loss": 0.0169, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 3.853118712273642, | |
| "grad_norm": 0.1361585557460785, | |
| "learning_rate": 2.7415442274564273e-05, | |
| "loss": 0.0144, | |
| "step": 7660 | |
| }, | |
| { | |
| "epoch": 3.858148893360161, | |
| "grad_norm": 0.27642127871513367, | |
| "learning_rate": 2.718700094346851e-05, | |
| "loss": 0.0131, | |
| "step": 7670 | |
| }, | |
| { | |
| "epoch": 3.86317907444668, | |
| "grad_norm": 0.3688669800758362, | |
| "learning_rate": 2.6959365525601822e-05, | |
| "loss": 0.0142, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 3.8682092555331993, | |
| "grad_norm": 0.06550594419240952, | |
| "learning_rate": 2.6732538540492292e-05, | |
| "loss": 0.0151, | |
| "step": 7690 | |
| }, | |
| { | |
| "epoch": 3.873239436619718, | |
| "grad_norm": 0.2921280264854431, | |
| "learning_rate": 2.6506522498720065e-05, | |
| "loss": 0.0107, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 3.8782696177062377, | |
| "grad_norm": 0.2402195930480957, | |
| "learning_rate": 2.6281319901889488e-05, | |
| "loss": 0.0162, | |
| "step": 7710 | |
| }, | |
| { | |
| "epoch": 3.8832997987927564, | |
| "grad_norm": 0.15326948463916779, | |
| "learning_rate": 2.6056933242601544e-05, | |
| "loss": 0.0171, | |
| "step": 7720 | |
| }, | |
| { | |
| "epoch": 3.8883299798792756, | |
| "grad_norm": 0.15069030225276947, | |
| "learning_rate": 2.5833365004426215e-05, | |
| "loss": 0.0137, | |
| "step": 7730 | |
| }, | |
| { | |
| "epoch": 3.893360160965795, | |
| "grad_norm": 0.2580581307411194, | |
| "learning_rate": 2.561061766187496e-05, | |
| "loss": 0.0159, | |
| "step": 7740 | |
| }, | |
| { | |
| "epoch": 3.8983903420523136, | |
| "grad_norm": 0.20609134435653687, | |
| "learning_rate": 2.538869368037332e-05, | |
| "loss": 0.0165, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 3.9034205231388333, | |
| "grad_norm": 0.12185626477003098, | |
| "learning_rate": 2.5167595516233722e-05, | |
| "loss": 0.0148, | |
| "step": 7760 | |
| }, | |
| { | |
| "epoch": 3.908450704225352, | |
| "grad_norm": 0.14162567257881165, | |
| "learning_rate": 2.4947325616628225e-05, | |
| "loss": 0.0139, | |
| "step": 7770 | |
| }, | |
| { | |
| "epoch": 3.913480885311871, | |
| "grad_norm": 0.14534728229045868, | |
| "learning_rate": 2.4727886419561374e-05, | |
| "loss": 0.0129, | |
| "step": 7780 | |
| }, | |
| { | |
| "epoch": 3.9185110663983904, | |
| "grad_norm": 0.1677662879228592, | |
| "learning_rate": 2.450928035384339e-05, | |
| "loss": 0.0177, | |
| "step": 7790 | |
| }, | |
| { | |
| "epoch": 3.9235412474849096, | |
| "grad_norm": 0.35635554790496826, | |
| "learning_rate": 2.4291509839063042e-05, | |
| "loss": 0.0138, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 3.928571428571429, | |
| "grad_norm": 0.1618734896183014, | |
| "learning_rate": 2.407457728556115e-05, | |
| "loss": 0.0133, | |
| "step": 7810 | |
| }, | |
| { | |
| "epoch": 3.9336016096579476, | |
| "grad_norm": 0.1893433928489685, | |
| "learning_rate": 2.385848509440364e-05, | |
| "loss": 0.0152, | |
| "step": 7820 | |
| }, | |
| { | |
| "epoch": 3.938631790744467, | |
| "grad_norm": 0.2308843731880188, | |
| "learning_rate": 2.3643235657355145e-05, | |
| "loss": 0.0128, | |
| "step": 7830 | |
| }, | |
| { | |
| "epoch": 3.943661971830986, | |
| "grad_norm": 0.21923568844795227, | |
| "learning_rate": 2.342883135685253e-05, | |
| "loss": 0.0147, | |
| "step": 7840 | |
| }, | |
| { | |
| "epoch": 3.948692152917505, | |
| "grad_norm": 0.24298080801963806, | |
| "learning_rate": 2.321527456597833e-05, | |
| "loss": 0.0177, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 3.9537223340040244, | |
| "grad_norm": 0.14858295023441315, | |
| "learning_rate": 2.300256764843477e-05, | |
| "loss": 0.0154, | |
| "step": 7860 | |
| }, | |
| { | |
| "epoch": 3.958752515090543, | |
| "grad_norm": 0.21552342176437378, | |
| "learning_rate": 2.2790712958517324e-05, | |
| "loss": 0.013, | |
| "step": 7870 | |
| }, | |
| { | |
| "epoch": 3.9637826961770624, | |
| "grad_norm": 0.2514561712741852, | |
| "learning_rate": 2.2579712841088873e-05, | |
| "loss": 0.0153, | |
| "step": 7880 | |
| }, | |
| { | |
| "epoch": 3.9688128772635816, | |
| "grad_norm": 0.1734912097454071, | |
| "learning_rate": 2.23695696315537e-05, | |
| "loss": 0.0198, | |
| "step": 7890 | |
| }, | |
| { | |
| "epoch": 3.9738430583501008, | |
| "grad_norm": 0.2863246500492096, | |
| "learning_rate": 2.216028565583148e-05, | |
| "loss": 0.016, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 3.97887323943662, | |
| "grad_norm": 0.2181919515132904, | |
| "learning_rate": 2.1951863230331793e-05, | |
| "loss": 0.012, | |
| "step": 7910 | |
| }, | |
| { | |
| "epoch": 3.9839034205231387, | |
| "grad_norm": 0.2490726262331009, | |
| "learning_rate": 2.174430466192826e-05, | |
| "loss": 0.0121, | |
| "step": 7920 | |
| }, | |
| { | |
| "epoch": 3.988933601609658, | |
| "grad_norm": 0.10584773123264313, | |
| "learning_rate": 2.153761224793317e-05, | |
| "loss": 0.0144, | |
| "step": 7930 | |
| }, | |
| { | |
| "epoch": 3.993963782696177, | |
| "grad_norm": 0.16053903102874756, | |
| "learning_rate": 2.1331788276072007e-05, | |
| "loss": 0.0169, | |
| "step": 7940 | |
| }, | |
| { | |
| "epoch": 3.9989939637826963, | |
| "grad_norm": 0.11051679402589798, | |
| "learning_rate": 2.1126835024458003e-05, | |
| "loss": 0.0094, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 4.0040241448692155, | |
| "grad_norm": 0.2628016173839569, | |
| "learning_rate": 2.0922754761567143e-05, | |
| "loss": 0.0132, | |
| "step": 7960 | |
| }, | |
| { | |
| "epoch": 4.009054325955734, | |
| "grad_norm": 0.14394591748714447, | |
| "learning_rate": 2.0719549746212897e-05, | |
| "loss": 0.0148, | |
| "step": 7970 | |
| }, | |
| { | |
| "epoch": 4.014084507042254, | |
| "grad_norm": 0.11262981593608856, | |
| "learning_rate": 2.0517222227521304e-05, | |
| "loss": 0.0096, | |
| "step": 7980 | |
| }, | |
| { | |
| "epoch": 4.019114688128773, | |
| "grad_norm": 0.09093613922595978, | |
| "learning_rate": 2.0315774444905965e-05, | |
| "loss": 0.0128, | |
| "step": 7990 | |
| }, | |
| { | |
| "epoch": 4.0241448692152915, | |
| "grad_norm": 0.1167258694767952, | |
| "learning_rate": 2.0115208628043436e-05, | |
| "loss": 0.0167, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 4.029175050301811, | |
| "grad_norm": 0.1440437287092209, | |
| "learning_rate": 1.9915526996848333e-05, | |
| "loss": 0.0136, | |
| "step": 8010 | |
| }, | |
| { | |
| "epoch": 4.03420523138833, | |
| "grad_norm": 0.23001736402511597, | |
| "learning_rate": 1.971673176144896e-05, | |
| "loss": 0.0133, | |
| "step": 8020 | |
| }, | |
| { | |
| "epoch": 4.0392354124748495, | |
| "grad_norm": 0.18372376263141632, | |
| "learning_rate": 1.9518825122162766e-05, | |
| "loss": 0.0121, | |
| "step": 8030 | |
| }, | |
| { | |
| "epoch": 4.044265593561368, | |
| "grad_norm": 0.16671916842460632, | |
| "learning_rate": 1.932180926947189e-05, | |
| "loss": 0.0124, | |
| "step": 8040 | |
| }, | |
| { | |
| "epoch": 4.049295774647887, | |
| "grad_norm": 0.22989486157894135, | |
| "learning_rate": 1.912568638399915e-05, | |
| "loss": 0.0142, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 4.054325955734407, | |
| "grad_norm": 0.28540411591529846, | |
| "learning_rate": 1.893045863648364e-05, | |
| "loss": 0.0162, | |
| "step": 8060 | |
| }, | |
| { | |
| "epoch": 4.059356136820925, | |
| "grad_norm": 0.21366238594055176, | |
| "learning_rate": 1.873612818775692e-05, | |
| "loss": 0.0139, | |
| "step": 8070 | |
| }, | |
| { | |
| "epoch": 4.064386317907445, | |
| "grad_norm": 0.08002448827028275, | |
| "learning_rate": 1.8542697188719005e-05, | |
| "loss": 0.0168, | |
| "step": 8080 | |
| }, | |
| { | |
| "epoch": 4.069416498993964, | |
| "grad_norm": 0.17863331735134125, | |
| "learning_rate": 1.83501677803145e-05, | |
| "loss": 0.0098, | |
| "step": 8090 | |
| }, | |
| { | |
| "epoch": 4.074446680080483, | |
| "grad_norm": 0.2530530095100403, | |
| "learning_rate": 1.815854209350908e-05, | |
| "loss": 0.013, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 4.079476861167002, | |
| "grad_norm": 0.18760378658771515, | |
| "learning_rate": 1.7967822249265677e-05, | |
| "loss": 0.0166, | |
| "step": 8110 | |
| }, | |
| { | |
| "epoch": 4.084507042253521, | |
| "grad_norm": 0.09926089644432068, | |
| "learning_rate": 1.777801035852119e-05, | |
| "loss": 0.018, | |
| "step": 8120 | |
| }, | |
| { | |
| "epoch": 4.089537223340041, | |
| "grad_norm": 0.24970552325248718, | |
| "learning_rate": 1.758910852216309e-05, | |
| "loss": 0.0157, | |
| "step": 8130 | |
| }, | |
| { | |
| "epoch": 4.094567404426559, | |
| "grad_norm": 0.12321308255195618, | |
| "learning_rate": 1.7401118831006004e-05, | |
| "loss": 0.0115, | |
| "step": 8140 | |
| }, | |
| { | |
| "epoch": 4.099597585513078, | |
| "grad_norm": 0.1537485420703888, | |
| "learning_rate": 1.721404336576884e-05, | |
| "loss": 0.0092, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 4.104627766599598, | |
| "grad_norm": 0.13459086418151855, | |
| "learning_rate": 1.702788419705148e-05, | |
| "loss": 0.0116, | |
| "step": 8160 | |
| }, | |
| { | |
| "epoch": 4.109657947686117, | |
| "grad_norm": 0.3569525480270386, | |
| "learning_rate": 1.684264338531214e-05, | |
| "loss": 0.0155, | |
| "step": 8170 | |
| }, | |
| { | |
| "epoch": 4.114688128772636, | |
| "grad_norm": 0.11241878569126129, | |
| "learning_rate": 1.6658322980844298e-05, | |
| "loss": 0.0089, | |
| "step": 8180 | |
| }, | |
| { | |
| "epoch": 4.119718309859155, | |
| "grad_norm": 0.31290173530578613, | |
| "learning_rate": 1.6474925023754174e-05, | |
| "loss": 0.0148, | |
| "step": 8190 | |
| }, | |
| { | |
| "epoch": 4.124748490945674, | |
| "grad_norm": 0.14965899288654327, | |
| "learning_rate": 1.6292451543938124e-05, | |
| "loss": 0.0131, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 4.129778672032193, | |
| "grad_norm": 0.29743674397468567, | |
| "learning_rate": 1.6110904561060126e-05, | |
| "loss": 0.016, | |
| "step": 8210 | |
| }, | |
| { | |
| "epoch": 4.134808853118712, | |
| "grad_norm": 0.26381608843803406, | |
| "learning_rate": 1.5930286084529457e-05, | |
| "loss": 0.0127, | |
| "step": 8220 | |
| }, | |
| { | |
| "epoch": 4.139839034205232, | |
| "grad_norm": 0.3434705436229706, | |
| "learning_rate": 1.57505981134784e-05, | |
| "loss": 0.0138, | |
| "step": 8230 | |
| }, | |
| { | |
| "epoch": 4.144869215291751, | |
| "grad_norm": 0.3142394423484802, | |
| "learning_rate": 1.557184263674024e-05, | |
| "loss": 0.0176, | |
| "step": 8240 | |
| }, | |
| { | |
| "epoch": 4.149899396378269, | |
| "grad_norm": 0.17862974107265472, | |
| "learning_rate": 1.5394021632827093e-05, | |
| "loss": 0.0123, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 4.154929577464789, | |
| "grad_norm": 0.2902679145336151, | |
| "learning_rate": 1.5217137069908128e-05, | |
| "loss": 0.0153, | |
| "step": 8260 | |
| }, | |
| { | |
| "epoch": 4.159959758551308, | |
| "grad_norm": 0.28409862518310547, | |
| "learning_rate": 1.5041190905787772e-05, | |
| "loss": 0.0117, | |
| "step": 8270 | |
| }, | |
| { | |
| "epoch": 4.164989939637827, | |
| "grad_norm": 0.32609495520591736, | |
| "learning_rate": 1.4866185087883933e-05, | |
| "loss": 0.0165, | |
| "step": 8280 | |
| }, | |
| { | |
| "epoch": 4.170020120724346, | |
| "grad_norm": 0.3060837686061859, | |
| "learning_rate": 1.4692121553206595e-05, | |
| "loss": 0.0153, | |
| "step": 8290 | |
| }, | |
| { | |
| "epoch": 4.175050301810865, | |
| "grad_norm": 0.25559473037719727, | |
| "learning_rate": 1.4519002228336232e-05, | |
| "loss": 0.0104, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 4.1800804828973845, | |
| "grad_norm": 0.12555906176567078, | |
| "learning_rate": 1.4346829029402654e-05, | |
| "loss": 0.0086, | |
| "step": 8310 | |
| }, | |
| { | |
| "epoch": 4.185110663983903, | |
| "grad_norm": 0.12174857407808304, | |
| "learning_rate": 1.4175603862063591e-05, | |
| "loss": 0.0088, | |
| "step": 8320 | |
| }, | |
| { | |
| "epoch": 4.190140845070423, | |
| "grad_norm": 0.11725517362356186, | |
| "learning_rate": 1.4005328621483794e-05, | |
| "loss": 0.0152, | |
| "step": 8330 | |
| }, | |
| { | |
| "epoch": 4.195171026156942, | |
| "grad_norm": 0.20686453580856323, | |
| "learning_rate": 1.3836005192313994e-05, | |
| "loss": 0.0159, | |
| "step": 8340 | |
| }, | |
| { | |
| "epoch": 4.2002012072434605, | |
| "grad_norm": 0.2010798454284668, | |
| "learning_rate": 1.3667635448669913e-05, | |
| "loss": 0.0136, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 4.20523138832998, | |
| "grad_norm": 0.18625499308109283, | |
| "learning_rate": 1.3500221254111777e-05, | |
| "loss": 0.0222, | |
| "step": 8360 | |
| }, | |
| { | |
| "epoch": 4.210261569416499, | |
| "grad_norm": 0.2409621626138687, | |
| "learning_rate": 1.3333764461623421e-05, | |
| "loss": 0.0234, | |
| "step": 8370 | |
| }, | |
| { | |
| "epoch": 4.2152917505030185, | |
| "grad_norm": 0.12151456624269485, | |
| "learning_rate": 1.3168266913591976e-05, | |
| "loss": 0.0152, | |
| "step": 8380 | |
| }, | |
| { | |
| "epoch": 4.220321931589537, | |
| "grad_norm": 0.2698768675327301, | |
| "learning_rate": 1.3003730441787399e-05, | |
| "loss": 0.0104, | |
| "step": 8390 | |
| }, | |
| { | |
| "epoch": 4.225352112676056, | |
| "grad_norm": 0.21013912558555603, | |
| "learning_rate": 1.2840156867342179e-05, | |
| "loss": 0.0095, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 4.230382293762576, | |
| "grad_norm": 0.10292834788560867, | |
| "learning_rate": 1.2677548000731243e-05, | |
| "loss": 0.0113, | |
| "step": 8410 | |
| }, | |
| { | |
| "epoch": 4.2354124748490944, | |
| "grad_norm": 0.05743186175823212, | |
| "learning_rate": 1.2515905641751824e-05, | |
| "loss": 0.0105, | |
| "step": 8420 | |
| }, | |
| { | |
| "epoch": 4.240442655935614, | |
| "grad_norm": 0.18373721837997437, | |
| "learning_rate": 1.2355231579503645e-05, | |
| "loss": 0.0156, | |
| "step": 8430 | |
| }, | |
| { | |
| "epoch": 4.245472837022133, | |
| "grad_norm": 0.18463782966136932, | |
| "learning_rate": 1.219552759236906e-05, | |
| "loss": 0.0101, | |
| "step": 8440 | |
| }, | |
| { | |
| "epoch": 4.250503018108652, | |
| "grad_norm": 0.13232889771461487, | |
| "learning_rate": 1.2036795447993387e-05, | |
| "loss": 0.0099, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 4.255533199195171, | |
| "grad_norm": 0.1151013895869255, | |
| "learning_rate": 1.1879036903265328e-05, | |
| "loss": 0.0096, | |
| "step": 8460 | |
| }, | |
| { | |
| "epoch": 4.26056338028169, | |
| "grad_norm": 0.1502464860677719, | |
| "learning_rate": 1.1722253704297492e-05, | |
| "loss": 0.0114, | |
| "step": 8470 | |
| }, | |
| { | |
| "epoch": 4.26559356136821, | |
| "grad_norm": 0.2458263784646988, | |
| "learning_rate": 1.1566447586407169e-05, | |
| "loss": 0.017, | |
| "step": 8480 | |
| }, | |
| { | |
| "epoch": 4.270623742454728, | |
| "grad_norm": 0.2860092222690582, | |
| "learning_rate": 1.1411620274097013e-05, | |
| "loss": 0.017, | |
| "step": 8490 | |
| }, | |
| { | |
| "epoch": 4.275653923541247, | |
| "grad_norm": 0.14720211923122406, | |
| "learning_rate": 1.1257773481036049e-05, | |
| "loss": 0.0135, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 4.280684104627767, | |
| "grad_norm": 0.3655484616756439, | |
| "learning_rate": 1.110490891004059e-05, | |
| "loss": 0.0228, | |
| "step": 8510 | |
| }, | |
| { | |
| "epoch": 4.285714285714286, | |
| "grad_norm": 0.08732811361551285, | |
| "learning_rate": 1.0953028253055542e-05, | |
| "loss": 0.0096, | |
| "step": 8520 | |
| }, | |
| { | |
| "epoch": 4.290744466800805, | |
| "grad_norm": 0.14107802510261536, | |
| "learning_rate": 1.0802133191135566e-05, | |
| "loss": 0.0096, | |
| "step": 8530 | |
| }, | |
| { | |
| "epoch": 4.295774647887324, | |
| "grad_norm": 0.18336671590805054, | |
| "learning_rate": 1.0652225394426441e-05, | |
| "loss": 0.0146, | |
| "step": 8540 | |
| }, | |
| { | |
| "epoch": 4.300804828973843, | |
| "grad_norm": 0.1546245664358139, | |
| "learning_rate": 1.0503306522146738e-05, | |
| "loss": 0.0113, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 4.305835010060362, | |
| "grad_norm": 0.1467217206954956, | |
| "learning_rate": 1.0355378222569256e-05, | |
| "loss": 0.0136, | |
| "step": 8560 | |
| }, | |
| { | |
| "epoch": 4.310865191146881, | |
| "grad_norm": 0.1393778920173645, | |
| "learning_rate": 1.0208442133002948e-05, | |
| "loss": 0.0107, | |
| "step": 8570 | |
| }, | |
| { | |
| "epoch": 4.315895372233401, | |
| "grad_norm": 0.3030548393726349, | |
| "learning_rate": 1.0062499879774734e-05, | |
| "loss": 0.0153, | |
| "step": 8580 | |
| }, | |
| { | |
| "epoch": 4.32092555331992, | |
| "grad_norm": 0.30646151304244995, | |
| "learning_rate": 9.917553078211417e-06, | |
| "loss": 0.0117, | |
| "step": 8590 | |
| }, | |
| { | |
| "epoch": 4.325955734406438, | |
| "grad_norm": 0.15872900187969208, | |
| "learning_rate": 9.773603332621972e-06, | |
| "loss": 0.0114, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 4.330985915492958, | |
| "grad_norm": 0.2105736881494522, | |
| "learning_rate": 9.630652236279625e-06, | |
| "loss": 0.0108, | |
| "step": 8610 | |
| }, | |
| { | |
| "epoch": 4.336016096579477, | |
| "grad_norm": 0.13897240161895752, | |
| "learning_rate": 9.488701371404329e-06, | |
| "loss": 0.0089, | |
| "step": 8620 | |
| }, | |
| { | |
| "epoch": 4.341046277665996, | |
| "grad_norm": 0.2540101408958435, | |
| "learning_rate": 9.347752309145241e-06, | |
| "loss": 0.0212, | |
| "step": 8630 | |
| }, | |
| { | |
| "epoch": 4.346076458752515, | |
| "grad_norm": 0.10785099118947983, | |
| "learning_rate": 9.20780660956324e-06, | |
| "loss": 0.0082, | |
| "step": 8640 | |
| }, | |
| { | |
| "epoch": 4.351106639839034, | |
| "grad_norm": 0.10478243976831436, | |
| "learning_rate": 9.068865821613803e-06, | |
| "loss": 0.0131, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 4.3561368209255535, | |
| "grad_norm": 0.12173474580049515, | |
| "learning_rate": 8.930931483129667e-06, | |
| "loss": 0.0107, | |
| "step": 8660 | |
| }, | |
| { | |
| "epoch": 4.361167002012072, | |
| "grad_norm": 0.19661535322666168, | |
| "learning_rate": 8.794005120804082e-06, | |
| "loss": 0.0121, | |
| "step": 8670 | |
| }, | |
| { | |
| "epoch": 4.366197183098592, | |
| "grad_norm": 0.1414082646369934, | |
| "learning_rate": 8.658088250173624e-06, | |
| "loss": 0.0106, | |
| "step": 8680 | |
| }, | |
| { | |
| "epoch": 4.371227364185111, | |
| "grad_norm": 0.07380446791648865, | |
| "learning_rate": 8.523182375601635e-06, | |
| "loss": 0.0116, | |
| "step": 8690 | |
| }, | |
| { | |
| "epoch": 4.3762575452716295, | |
| "grad_norm": 0.1488698124885559, | |
| "learning_rate": 8.389288990261413e-06, | |
| "loss": 0.0093, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 4.381287726358149, | |
| "grad_norm": 0.17949314415454865, | |
| "learning_rate": 8.256409576119827e-06, | |
| "loss": 0.0099, | |
| "step": 8710 | |
| }, | |
| { | |
| "epoch": 4.386317907444668, | |
| "grad_norm": 0.34648457169532776, | |
| "learning_rate": 8.124545603920842e-06, | |
| "loss": 0.0109, | |
| "step": 8720 | |
| }, | |
| { | |
| "epoch": 4.3913480885311875, | |
| "grad_norm": 0.16411487758159637, | |
| "learning_rate": 7.993698533169192e-06, | |
| "loss": 0.0092, | |
| "step": 8730 | |
| }, | |
| { | |
| "epoch": 4.396378269617706, | |
| "grad_norm": 0.20851458609104156, | |
| "learning_rate": 7.863869812114366e-06, | |
| "loss": 0.0125, | |
| "step": 8740 | |
| }, | |
| { | |
| "epoch": 4.401408450704225, | |
| "grad_norm": 0.13391834497451782, | |
| "learning_rate": 7.73506087773439e-06, | |
| "loss": 0.0111, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 4.406438631790745, | |
| "grad_norm": 0.2575322687625885, | |
| "learning_rate": 7.60727315572013e-06, | |
| "loss": 0.0137, | |
| "step": 8760 | |
| }, | |
| { | |
| "epoch": 4.4114688128772634, | |
| "grad_norm": 0.1566155105829239, | |
| "learning_rate": 7.480508060459346e-06, | |
| "loss": 0.0121, | |
| "step": 8770 | |
| }, | |
| { | |
| "epoch": 4.416498993963783, | |
| "grad_norm": 0.23036964237689972, | |
| "learning_rate": 7.3547669950211005e-06, | |
| "loss": 0.0189, | |
| "step": 8780 | |
| }, | |
| { | |
| "epoch": 4.421529175050302, | |
| "grad_norm": 0.16328741610050201, | |
| "learning_rate": 7.230051351140266e-06, | |
| "loss": 0.0188, | |
| "step": 8790 | |
| }, | |
| { | |
| "epoch": 4.426559356136821, | |
| "grad_norm": 0.45226040482521057, | |
| "learning_rate": 7.106362509202036e-06, | |
| "loss": 0.0175, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 4.43158953722334, | |
| "grad_norm": 0.4067038297653198, | |
| "learning_rate": 6.983701838226708e-06, | |
| "loss": 0.0134, | |
| "step": 8810 | |
| }, | |
| { | |
| "epoch": 4.436619718309859, | |
| "grad_norm": 0.1728140115737915, | |
| "learning_rate": 6.86207069585455e-06, | |
| "loss": 0.0096, | |
| "step": 8820 | |
| }, | |
| { | |
| "epoch": 4.441649899396379, | |
| "grad_norm": 0.22685836255550385, | |
| "learning_rate": 6.741470428330676e-06, | |
| "loss": 0.0135, | |
| "step": 8830 | |
| }, | |
| { | |
| "epoch": 4.446680080482897, | |
| "grad_norm": 0.18156301975250244, | |
| "learning_rate": 6.621902370490274e-06, | |
| "loss": 0.0144, | |
| "step": 8840 | |
| }, | |
| { | |
| "epoch": 4.451710261569416, | |
| "grad_norm": 0.20519036054611206, | |
| "learning_rate": 6.503367845743702e-06, | |
| "loss": 0.015, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 4.456740442655936, | |
| "grad_norm": 0.26189279556274414, | |
| "learning_rate": 6.385868166061981e-06, | |
| "loss": 0.011, | |
| "step": 8860 | |
| }, | |
| { | |
| "epoch": 4.461770623742455, | |
| "grad_norm": 0.1770821064710617, | |
| "learning_rate": 6.269404631962106e-06, | |
| "loss": 0.0112, | |
| "step": 8870 | |
| }, | |
| { | |
| "epoch": 4.466800804828974, | |
| "grad_norm": 0.31295013427734375, | |
| "learning_rate": 6.153978532492821e-06, | |
| "loss": 0.0186, | |
| "step": 8880 | |
| }, | |
| { | |
| "epoch": 4.471830985915493, | |
| "grad_norm": 0.47270047664642334, | |
| "learning_rate": 6.0395911452202355e-06, | |
| "loss": 0.0149, | |
| "step": 8890 | |
| }, | |
| { | |
| "epoch": 4.476861167002012, | |
| "grad_norm": 0.1958407759666443, | |
| "learning_rate": 5.926243736213743e-06, | |
| "loss": 0.0076, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 4.481891348088531, | |
| "grad_norm": 0.14082856476306915, | |
| "learning_rate": 5.813937560031979e-06, | |
| "loss": 0.014, | |
| "step": 8910 | |
| }, | |
| { | |
| "epoch": 4.48692152917505, | |
| "grad_norm": 0.2103276550769806, | |
| "learning_rate": 5.702673859708896e-06, | |
| "loss": 0.0121, | |
| "step": 8920 | |
| }, | |
| { | |
| "epoch": 4.49195171026157, | |
| "grad_norm": 0.10313666611909866, | |
| "learning_rate": 5.592453866740155e-06, | |
| "loss": 0.0118, | |
| "step": 8930 | |
| }, | |
| { | |
| "epoch": 4.496981891348089, | |
| "grad_norm": 0.23443832993507385, | |
| "learning_rate": 5.48327880106927e-06, | |
| "loss": 0.0123, | |
| "step": 8940 | |
| }, | |
| { | |
| "epoch": 4.502012072434607, | |
| "grad_norm": 0.08273012936115265, | |
| "learning_rate": 5.375149871074336e-06, | |
| "loss": 0.0101, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 4.507042253521127, | |
| "grad_norm": 0.13498452305793762, | |
| "learning_rate": 5.268068273554483e-06, | |
| "loss": 0.0142, | |
| "step": 8960 | |
| }, | |
| { | |
| "epoch": 4.512072434607646, | |
| "grad_norm": 0.12951943278312683, | |
| "learning_rate": 5.1620351937167076e-06, | |
| "loss": 0.0111, | |
| "step": 8970 | |
| }, | |
| { | |
| "epoch": 4.517102615694165, | |
| "grad_norm": 0.18177278339862823, | |
| "learning_rate": 5.057051805162749e-06, | |
| "loss": 0.0143, | |
| "step": 8980 | |
| }, | |
| { | |
| "epoch": 4.522132796780684, | |
| "grad_norm": 0.23840029537677765, | |
| "learning_rate": 4.953119269876061e-06, | |
| "loss": 0.0121, | |
| "step": 8990 | |
| }, | |
| { | |
| "epoch": 4.527162977867203, | |
| "grad_norm": 0.13470517098903656, | |
| "learning_rate": 4.8502387382090345e-06, | |
| "loss": 0.0136, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 4.5321931589537225, | |
| "grad_norm": 0.1814277470111847, | |
| "learning_rate": 4.748411348870141e-06, | |
| "loss": 0.0167, | |
| "step": 9010 | |
| }, | |
| { | |
| "epoch": 4.537223340040241, | |
| "grad_norm": 0.3026091158390045, | |
| "learning_rate": 4.647638228911466e-06, | |
| "loss": 0.0157, | |
| "step": 9020 | |
| }, | |
| { | |
| "epoch": 4.542253521126761, | |
| "grad_norm": 0.13517017662525177, | |
| "learning_rate": 4.547920493716118e-06, | |
| "loss": 0.0093, | |
| "step": 9030 | |
| }, | |
| { | |
| "epoch": 4.54728370221328, | |
| "grad_norm": 0.06449972838163376, | |
| "learning_rate": 4.4492592469859486e-06, | |
| "loss": 0.0101, | |
| "step": 9040 | |
| }, | |
| { | |
| "epoch": 4.5523138832997985, | |
| "grad_norm": 0.24287287890911102, | |
| "learning_rate": 4.3516555807293415e-06, | |
| "loss": 0.015, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 4.557344064386318, | |
| "grad_norm": 0.13659290969371796, | |
| "learning_rate": 4.255110575249055e-06, | |
| "loss": 0.0098, | |
| "step": 9060 | |
| }, | |
| { | |
| "epoch": 4.562374245472837, | |
| "grad_norm": 0.258847177028656, | |
| "learning_rate": 4.1596252991303655e-06, | |
| "loss": 0.0109, | |
| "step": 9070 | |
| }, | |
| { | |
| "epoch": 4.5674044265593565, | |
| "grad_norm": 0.2787615954875946, | |
| "learning_rate": 4.065200809229163e-06, | |
| "loss": 0.0116, | |
| "step": 9080 | |
| }, | |
| { | |
| "epoch": 4.572434607645875, | |
| "grad_norm": 0.1802290678024292, | |
| "learning_rate": 3.971838150660268e-06, | |
| "loss": 0.0101, | |
| "step": 9090 | |
| }, | |
| { | |
| "epoch": 4.577464788732394, | |
| "grad_norm": 0.22423778474330902, | |
| "learning_rate": 3.879538356785917e-06, | |
| "loss": 0.0103, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 4.582494969818914, | |
| "grad_norm": 0.18749003112316132, | |
| "learning_rate": 3.7883024492042286e-06, | |
| "loss": 0.0084, | |
| "step": 9110 | |
| }, | |
| { | |
| "epoch": 4.5875251509054324, | |
| "grad_norm": 0.08380598574876785, | |
| "learning_rate": 3.698131437737995e-06, | |
| "loss": 0.0092, | |
| "step": 9120 | |
| }, | |
| { | |
| "epoch": 4.592555331991952, | |
| "grad_norm": 0.11922164261341095, | |
| "learning_rate": 3.6090263204234363e-06, | |
| "loss": 0.0129, | |
| "step": 9130 | |
| }, | |
| { | |
| "epoch": 4.597585513078471, | |
| "grad_norm": 0.13963715732097626, | |
| "learning_rate": 3.520988083499199e-06, | |
| "loss": 0.0113, | |
| "step": 9140 | |
| }, | |
| { | |
| "epoch": 4.60261569416499, | |
| "grad_norm": 0.15656189620494843, | |
| "learning_rate": 3.434017701395431e-06, | |
| "loss": 0.0109, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 4.607645875251509, | |
| "grad_norm": 0.0764361172914505, | |
| "learning_rate": 3.348116136722912e-06, | |
| "loss": 0.0131, | |
| "step": 9160 | |
| }, | |
| { | |
| "epoch": 4.612676056338028, | |
| "grad_norm": 0.23831793665885925, | |
| "learning_rate": 3.2632843402625625e-06, | |
| "loss": 0.0115, | |
| "step": 9170 | |
| }, | |
| { | |
| "epoch": 4.617706237424548, | |
| "grad_norm": 0.19923029839992523, | |
| "learning_rate": 3.1795232509547633e-06, | |
| "loss": 0.0102, | |
| "step": 9180 | |
| }, | |
| { | |
| "epoch": 4.622736418511066, | |
| "grad_norm": 0.1905340999364853, | |
| "learning_rate": 3.096833795889076e-06, | |
| "loss": 0.0104, | |
| "step": 9190 | |
| }, | |
| { | |
| "epoch": 4.627766599597585, | |
| "grad_norm": 0.2911848723888397, | |
| "learning_rate": 3.015216890293904e-06, | |
| "loss": 0.0113, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 4.632796780684105, | |
| "grad_norm": 0.19489365816116333, | |
| "learning_rate": 2.9346734375264027e-06, | |
| "loss": 0.0145, | |
| "step": 9210 | |
| }, | |
| { | |
| "epoch": 4.637826961770624, | |
| "grad_norm": 0.10868648439645767, | |
| "learning_rate": 2.8552043290624997e-06, | |
| "loss": 0.0128, | |
| "step": 9220 | |
| }, | |
| { | |
| "epoch": 4.642857142857143, | |
| "grad_norm": 0.21155446767807007, | |
| "learning_rate": 2.7768104444869436e-06, | |
| "loss": 0.0127, | |
| "step": 9230 | |
| }, | |
| { | |
| "epoch": 4.647887323943662, | |
| "grad_norm": 0.20911335945129395, | |
| "learning_rate": 2.6994926514836925e-06, | |
| "loss": 0.0159, | |
| "step": 9240 | |
| }, | |
| { | |
| "epoch": 4.652917505030181, | |
| "grad_norm": 0.20620648562908173, | |
| "learning_rate": 2.6232518058261658e-06, | |
| "loss": 0.0136, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 4.6579476861167, | |
| "grad_norm": 0.0914982482790947, | |
| "learning_rate": 2.5480887513679166e-06, | |
| "loss": 0.0114, | |
| "step": 9260 | |
| }, | |
| { | |
| "epoch": 4.662977867203219, | |
| "grad_norm": 0.09038899838924408, | |
| "learning_rate": 2.4740043200332074e-06, | |
| "loss": 0.0125, | |
| "step": 9270 | |
| }, | |
| { | |
| "epoch": 4.668008048289739, | |
| "grad_norm": 0.13401928544044495, | |
| "learning_rate": 2.400999331807796e-06, | |
| "loss": 0.0155, | |
| "step": 9280 | |
| }, | |
| { | |
| "epoch": 4.673038229376258, | |
| "grad_norm": 0.18889258801937103, | |
| "learning_rate": 2.3290745947298966e-06, | |
| "loss": 0.0098, | |
| "step": 9290 | |
| }, | |
| { | |
| "epoch": 4.678068410462776, | |
| "grad_norm": 0.15076126158237457, | |
| "learning_rate": 2.258230904881231e-06, | |
| "loss": 0.0151, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 4.683098591549296, | |
| "grad_norm": 0.17581748962402344, | |
| "learning_rate": 2.1884690463781833e-06, | |
| "loss": 0.0148, | |
| "step": 9310 | |
| }, | |
| { | |
| "epoch": 4.688128772635815, | |
| "grad_norm": 0.08170731365680695, | |
| "learning_rate": 2.1197897913632026e-06, | |
| "loss": 0.0216, | |
| "step": 9320 | |
| }, | |
| { | |
| "epoch": 4.693158953722334, | |
| "grad_norm": 0.19143046438694, | |
| "learning_rate": 2.0521938999961243e-06, | |
| "loss": 0.0166, | |
| "step": 9330 | |
| }, | |
| { | |
| "epoch": 4.698189134808853, | |
| "grad_norm": 0.13908398151397705, | |
| "learning_rate": 1.9856821204458864e-06, | |
| "loss": 0.0121, | |
| "step": 9340 | |
| }, | |
| { | |
| "epoch": 4.703219315895372, | |
| "grad_norm": 0.1692146509885788, | |
| "learning_rate": 1.9202551888821807e-06, | |
| "loss": 0.0073, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 4.7082494969818915, | |
| "grad_norm": 0.11902793496847153, | |
| "learning_rate": 1.855913829467315e-06, | |
| "loss": 0.0098, | |
| "step": 9360 | |
| }, | |
| { | |
| "epoch": 4.71327967806841, | |
| "grad_norm": 0.3413757383823395, | |
| "learning_rate": 1.7926587543482088e-06, | |
| "loss": 0.0225, | |
| "step": 9370 | |
| }, | |
| { | |
| "epoch": 4.71830985915493, | |
| "grad_norm": 0.13018754124641418, | |
| "learning_rate": 1.7304906636485097e-06, | |
| "loss": 0.0097, | |
| "step": 9380 | |
| }, | |
| { | |
| "epoch": 4.723340040241449, | |
| "grad_norm": 0.17999660968780518, | |
| "learning_rate": 1.6694102454608118e-06, | |
| "loss": 0.0174, | |
| "step": 9390 | |
| }, | |
| { | |
| "epoch": 4.7283702213279675, | |
| "grad_norm": 0.1070287749171257, | |
| "learning_rate": 1.6094181758390947e-06, | |
| "loss": 0.0116, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 4.733400402414487, | |
| "grad_norm": 0.18182510137557983, | |
| "learning_rate": 1.5505151187912071e-06, | |
| "loss": 0.0109, | |
| "step": 9410 | |
| }, | |
| { | |
| "epoch": 4.738430583501006, | |
| "grad_norm": 0.1901055872440338, | |
| "learning_rate": 1.4927017262715059e-06, | |
| "loss": 0.0093, | |
| "step": 9420 | |
| }, | |
| { | |
| "epoch": 4.7434607645875255, | |
| "grad_norm": 0.1983153074979782, | |
| "learning_rate": 1.435978638173685e-06, | |
| "loss": 0.0153, | |
| "step": 9430 | |
| }, | |
| { | |
| "epoch": 4.748490945674044, | |
| "grad_norm": 0.1601441502571106, | |
| "learning_rate": 1.3803464823236356e-06, | |
| "loss": 0.0101, | |
| "step": 9440 | |
| }, | |
| { | |
| "epoch": 4.753521126760563, | |
| "grad_norm": 0.1466078907251358, | |
| "learning_rate": 1.325805874472552e-06, | |
| "loss": 0.0183, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 4.758551307847083, | |
| "grad_norm": 0.24373824894428253, | |
| "learning_rate": 1.272357418290082e-06, | |
| "loss": 0.008, | |
| "step": 9460 | |
| }, | |
| { | |
| "epoch": 4.7635814889336014, | |
| "grad_norm": 0.08891261368989944, | |
| "learning_rate": 1.2200017053576318e-06, | |
| "loss": 0.0106, | |
| "step": 9470 | |
| }, | |
| { | |
| "epoch": 4.768611670020121, | |
| "grad_norm": 0.17067329585552216, | |
| "learning_rate": 1.1687393151618931e-06, | |
| "loss": 0.0096, | |
| "step": 9480 | |
| }, | |
| { | |
| "epoch": 4.77364185110664, | |
| "grad_norm": 0.1952274590730667, | |
| "learning_rate": 1.1185708150883268e-06, | |
| "loss": 0.0122, | |
| "step": 9490 | |
| }, | |
| { | |
| "epoch": 4.778672032193159, | |
| "grad_norm": 0.18577119708061218, | |
| "learning_rate": 1.0694967604149563e-06, | |
| "loss": 0.0132, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 4.783702213279678, | |
| "grad_norm": 0.19541311264038086, | |
| "learning_rate": 1.0215176943061955e-06, | |
| "loss": 0.0135, | |
| "step": 9510 | |
| }, | |
| { | |
| "epoch": 4.788732394366197, | |
| "grad_norm": 0.14900928735733032, | |
| "learning_rate": 9.746341478068298e-07, | |
| "loss": 0.0122, | |
| "step": 9520 | |
| }, | |
| { | |
| "epoch": 4.793762575452717, | |
| "grad_norm": 0.2078647017478943, | |
| "learning_rate": 9.288466398361783e-07, | |
| "loss": 0.0098, | |
| "step": 9530 | |
| }, | |
| { | |
| "epoch": 4.798792756539235, | |
| "grad_norm": 0.137592151761055, | |
| "learning_rate": 8.841556771822746e-07, | |
| "loss": 0.0142, | |
| "step": 9540 | |
| }, | |
| { | |
| "epoch": 4.803822937625754, | |
| "grad_norm": 0.1908939927816391, | |
| "learning_rate": 8.405617544963385e-07, | |
| "loss": 0.0143, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 4.808853118712274, | |
| "grad_norm": 0.17629876732826233, | |
| "learning_rate": 7.980653542872584e-07, | |
| "loss": 0.0097, | |
| "step": 9560 | |
| }, | |
| { | |
| "epoch": 4.813883299798793, | |
| "grad_norm": 0.11937274038791656, | |
| "learning_rate": 7.566669469162513e-07, | |
| "loss": 0.0126, | |
| "step": 9570 | |
| }, | |
| { | |
| "epoch": 4.818913480885312, | |
| "grad_norm": 0.21683438122272491, | |
| "learning_rate": 7.16366990591677e-07, | |
| "loss": 0.0089, | |
| "step": 9580 | |
| }, | |
| { | |
| "epoch": 4.823943661971831, | |
| "grad_norm": 0.1359615921974182, | |
| "learning_rate": 6.771659313639212e-07, | |
| "loss": 0.01, | |
| "step": 9590 | |
| }, | |
| { | |
| "epoch": 4.82897384305835, | |
| "grad_norm": 0.23794005811214447, | |
| "learning_rate": 6.390642031205318e-07, | |
| "loss": 0.0183, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 4.834004024144869, | |
| "grad_norm": 0.10441375523805618, | |
| "learning_rate": 6.020622275813459e-07, | |
| "loss": 0.0107, | |
| "step": 9610 | |
| }, | |
| { | |
| "epoch": 4.839034205231388, | |
| "grad_norm": 0.2170896977186203, | |
| "learning_rate": 5.661604142938703e-07, | |
| "loss": 0.0144, | |
| "step": 9620 | |
| }, | |
| { | |
| "epoch": 4.844064386317908, | |
| "grad_norm": 0.1544094830751419, | |
| "learning_rate": 5.313591606287194e-07, | |
| "loss": 0.0095, | |
| "step": 9630 | |
| }, | |
| { | |
| "epoch": 4.849094567404427, | |
| "grad_norm": 0.09597515314817429, | |
| "learning_rate": 4.976588517752178e-07, | |
| "loss": 0.0078, | |
| "step": 9640 | |
| }, | |
| { | |
| "epoch": 4.854124748490945, | |
| "grad_norm": 0.18722732365131378, | |
| "learning_rate": 4.6505986073717143e-07, | |
| "loss": 0.0167, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 4.859154929577465, | |
| "grad_norm": 0.15623292326927185, | |
| "learning_rate": 4.3356254832869204e-07, | |
| "loss": 0.0122, | |
| "step": 9660 | |
| }, | |
| { | |
| "epoch": 4.864185110663984, | |
| "grad_norm": 0.19763480126857758, | |
| "learning_rate": 4.0316726317023435e-07, | |
| "loss": 0.0274, | |
| "step": 9670 | |
| }, | |
| { | |
| "epoch": 4.869215291750503, | |
| "grad_norm": 0.0955481231212616, | |
| "learning_rate": 3.7387434168473235e-07, | |
| "loss": 0.0147, | |
| "step": 9680 | |
| }, | |
| { | |
| "epoch": 4.874245472837022, | |
| "grad_norm": 0.18850040435791016, | |
| "learning_rate": 3.4568410809385774e-07, | |
| "loss": 0.0146, | |
| "step": 9690 | |
| }, | |
| { | |
| "epoch": 4.879275653923541, | |
| "grad_norm": 0.2946015000343323, | |
| "learning_rate": 3.185968744144563e-07, | |
| "loss": 0.0107, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 4.8843058350100605, | |
| "grad_norm": 0.29948338866233826, | |
| "learning_rate": 2.926129404550837e-07, | |
| "loss": 0.0088, | |
| "step": 9710 | |
| }, | |
| { | |
| "epoch": 4.889336016096579, | |
| "grad_norm": 0.23983007669448853, | |
| "learning_rate": 2.6773259381268625e-07, | |
| "loss": 0.0176, | |
| "step": 9720 | |
| }, | |
| { | |
| "epoch": 4.894366197183099, | |
| "grad_norm": 0.11162281781435013, | |
| "learning_rate": 2.439561098694254e-07, | |
| "loss": 0.0112, | |
| "step": 9730 | |
| }, | |
| { | |
| "epoch": 4.899396378269618, | |
| "grad_norm": 0.1520148515701294, | |
| "learning_rate": 2.212837517896027e-07, | |
| "loss": 0.0105, | |
| "step": 9740 | |
| }, | |
| { | |
| "epoch": 4.9044265593561365, | |
| "grad_norm": 0.1838511973619461, | |
| "learning_rate": 1.9971577051678404e-07, | |
| "loss": 0.0193, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 4.909456740442656, | |
| "grad_norm": 0.1083097979426384, | |
| "learning_rate": 1.7925240477100203e-07, | |
| "loss": 0.0142, | |
| "step": 9760 | |
| }, | |
| { | |
| "epoch": 4.914486921529175, | |
| "grad_norm": 0.16764095425605774, | |
| "learning_rate": 1.598938810461137e-07, | |
| "loss": 0.0102, | |
| "step": 9770 | |
| }, | |
| { | |
| "epoch": 4.9195171026156945, | |
| "grad_norm": 0.22907580435276031, | |
| "learning_rate": 1.416404136073024e-07, | |
| "loss": 0.0144, | |
| "step": 9780 | |
| }, | |
| { | |
| "epoch": 4.924547283702213, | |
| "grad_norm": 0.2832927703857422, | |
| "learning_rate": 1.2449220448870204e-07, | |
| "loss": 0.0134, | |
| "step": 9790 | |
| }, | |
| { | |
| "epoch": 4.929577464788732, | |
| "grad_norm": 0.10838713496923447, | |
| "learning_rate": 1.0844944349114316e-07, | |
| "loss": 0.01, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 4.934607645875252, | |
| "grad_norm": 0.0964408740401268, | |
| "learning_rate": 9.351230818008815e-08, | |
| "loss": 0.0109, | |
| "step": 9810 | |
| }, | |
| { | |
| "epoch": 4.9396378269617705, | |
| "grad_norm": 0.1444048136472702, | |
| "learning_rate": 7.968096388364377e-08, | |
| "loss": 0.0113, | |
| "step": 9820 | |
| }, | |
| { | |
| "epoch": 4.94466800804829, | |
| "grad_norm": 0.09675177931785583, | |
| "learning_rate": 6.69555636907182e-08, | |
| "loss": 0.0104, | |
| "step": 9830 | |
| }, | |
| { | |
| "epoch": 4.949698189134809, | |
| "grad_norm": 0.14633896946907043, | |
| "learning_rate": 5.533624844936691e-08, | |
| "loss": 0.009, | |
| "step": 9840 | |
| }, | |
| { | |
| "epoch": 4.954728370221328, | |
| "grad_norm": 0.27734798192977905, | |
| "learning_rate": 4.4823146765182735e-08, | |
| "loss": 0.011, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 4.959758551307847, | |
| "grad_norm": 0.2644200921058655, | |
| "learning_rate": 3.5416374999919235e-08, | |
| "loss": 0.0124, | |
| "step": 9860 | |
| }, | |
| { | |
| "epoch": 4.964788732394366, | |
| "grad_norm": 0.24501149356365204, | |
| "learning_rate": 2.7116037270169538e-08, | |
| "loss": 0.0123, | |
| "step": 9870 | |
| }, | |
| { | |
| "epoch": 4.969818913480886, | |
| "grad_norm": 0.24458244442939758, | |
| "learning_rate": 1.9922225446245e-08, | |
| "loss": 0.0146, | |
| "step": 9880 | |
| }, | |
| { | |
| "epoch": 4.974849094567404, | |
| "grad_norm": 0.06575839966535568, | |
| "learning_rate": 1.383501915112051e-08, | |
| "loss": 0.0089, | |
| "step": 9890 | |
| }, | |
| { | |
| "epoch": 4.979879275653923, | |
| "grad_norm": 0.2329985499382019, | |
| "learning_rate": 8.854485759568487e-09, | |
| "loss": 0.0136, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 4.984909456740443, | |
| "grad_norm": 0.17096395790576935, | |
| "learning_rate": 4.980680397448367e-09, | |
| "loss": 0.0118, | |
| "step": 9910 | |
| }, | |
| { | |
| "epoch": 4.989939637826962, | |
| "grad_norm": 0.16430173814296722, | |
| "learning_rate": 2.213645941029352e-09, | |
| "loss": 0.0088, | |
| "step": 9920 | |
| }, | |
| { | |
| "epoch": 4.994969818913481, | |
| "grad_norm": 0.23126807808876038, | |
| "learning_rate": 5.534130165907314e-10, | |
| "loss": 0.0184, | |
| "step": 9930 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.1565922051668167, | |
| "learning_rate": 0.0, | |
| "loss": 0.0118, | |
| "step": 9940 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 9940, | |
| "total_flos": 3.545130433511883e+17, | |
| "train_loss": 0.029606477042259105, | |
| "train_runtime": 4441.4052, | |
| "train_samples_per_second": 35.808, | |
| "train_steps_per_second": 2.238 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 9940, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.545130433511883e+17, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |