| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 9705, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0030911901081916537, | |
| "grad_norm": 0.0021730465814471245, | |
| "learning_rate": 0.0009989696032972695, | |
| "loss": 1.1462, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0061823802163833074, | |
| "grad_norm": 0.0018489831127226353, | |
| "learning_rate": 0.000997939206594539, | |
| "loss": 1.0204, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.00927357032457496, | |
| "grad_norm": 0.0019520210335031152, | |
| "learning_rate": 0.0009969088098918083, | |
| "loss": 1.0607, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.012364760432766615, | |
| "grad_norm": 0.002070036716759205, | |
| "learning_rate": 0.0009958784131890777, | |
| "loss": 1.0284, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.015455950540958269, | |
| "grad_norm": 0.0020703268237411976, | |
| "learning_rate": 0.0009948480164863472, | |
| "loss": 1.0654, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01854714064914992, | |
| "grad_norm": 0.0019948079716414213, | |
| "learning_rate": 0.0009938176197836167, | |
| "loss": 1.0034, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.021638330757341576, | |
| "grad_norm": 0.002126147970557213, | |
| "learning_rate": 0.0009927872230808862, | |
| "loss": 1.0173, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.02472952086553323, | |
| "grad_norm": 0.0017046661814674735, | |
| "learning_rate": 0.0009917568263781555, | |
| "loss": 1.0153, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.027820710973724884, | |
| "grad_norm": 0.0018810734618455172, | |
| "learning_rate": 0.000990726429675425, | |
| "loss": 0.9208, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.030911901081916538, | |
| "grad_norm": 0.0019539243075996637, | |
| "learning_rate": 0.0009896960329726944, | |
| "loss": 1.0072, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03400309119010819, | |
| "grad_norm": 0.0015656334580853581, | |
| "learning_rate": 0.000988665636269964, | |
| "loss": 0.9477, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.03709428129829984, | |
| "grad_norm": 0.0018790484173223376, | |
| "learning_rate": 0.0009876352395672334, | |
| "loss": 0.9543, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.0401854714064915, | |
| "grad_norm": 0.0014801452634856105, | |
| "learning_rate": 0.0009866048428645029, | |
| "loss": 0.9507, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.04327666151468315, | |
| "grad_norm": 0.0023686159402132034, | |
| "learning_rate": 0.0009855744461617724, | |
| "loss": 0.9471, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.04636785162287481, | |
| "grad_norm": 0.0018689304124563932, | |
| "learning_rate": 0.0009845440494590419, | |
| "loss": 0.9483, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04945904173106646, | |
| "grad_norm": 0.0017488012090325356, | |
| "learning_rate": 0.0009835136527563111, | |
| "loss": 0.954, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.05255023183925812, | |
| "grad_norm": 0.001690636039711535, | |
| "learning_rate": 0.0009824832560535806, | |
| "loss": 0.9277, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.05564142194744977, | |
| "grad_norm": 0.0017418304923921824, | |
| "learning_rate": 0.00098145285935085, | |
| "loss": 0.9258, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.05873261205564142, | |
| "grad_norm": 0.0014526835875585675, | |
| "learning_rate": 0.0009804224626481196, | |
| "loss": 0.9577, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.061823802163833076, | |
| "grad_norm": 0.002275065751746297, | |
| "learning_rate": 0.000979392065945389, | |
| "loss": 0.9241, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06491499227202473, | |
| "grad_norm": 0.0012924526818096638, | |
| "learning_rate": 0.0009783616692426585, | |
| "loss": 0.9271, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.06800618238021638, | |
| "grad_norm": 0.001761179999448359, | |
| "learning_rate": 0.000977331272539928, | |
| "loss": 0.9516, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.07109737248840804, | |
| "grad_norm": 0.001756934798322618, | |
| "learning_rate": 0.0009763008758371973, | |
| "loss": 0.9176, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.07418856259659969, | |
| "grad_norm": 0.0015818781685084105, | |
| "learning_rate": 0.0009752704791344668, | |
| "loss": 0.9549, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.07727975270479134, | |
| "grad_norm": 0.0015478008426725864, | |
| "learning_rate": 0.0009742400824317363, | |
| "loss": 0.9418, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.080370942812983, | |
| "grad_norm": 0.0014897435903549194, | |
| "learning_rate": 0.0009732096857290057, | |
| "loss": 0.9052, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.08346213292117466, | |
| "grad_norm": 0.0013134022010490298, | |
| "learning_rate": 0.0009721792890262751, | |
| "loss": 0.9479, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.0865533230293663, | |
| "grad_norm": 0.0018637892790138721, | |
| "learning_rate": 0.0009711488923235445, | |
| "loss": 0.8966, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.08964451313755796, | |
| "grad_norm": 0.0032756649889051914, | |
| "learning_rate": 0.000970118495620814, | |
| "loss": 0.9641, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.09273570324574962, | |
| "grad_norm": 0.001614488777704537, | |
| "learning_rate": 0.0009690880989180835, | |
| "loss": 0.9373, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.09582689335394126, | |
| "grad_norm": 0.0015908046625554562, | |
| "learning_rate": 0.000968057702215353, | |
| "loss": 0.953, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.09891808346213292, | |
| "grad_norm": 0.0017299671890214086, | |
| "learning_rate": 0.0009670273055126224, | |
| "loss": 0.9651, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.10200927357032458, | |
| "grad_norm": 0.0015082499012351036, | |
| "learning_rate": 0.0009659969088098919, | |
| "loss": 0.952, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.10510046367851623, | |
| "grad_norm": 0.00181696773506701, | |
| "learning_rate": 0.0009649665121071612, | |
| "loss": 0.9673, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.10819165378670788, | |
| "grad_norm": 0.001661508227698505, | |
| "learning_rate": 0.0009639361154044307, | |
| "loss": 0.9167, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.11128284389489954, | |
| "grad_norm": 0.0016009090468287468, | |
| "learning_rate": 0.0009629057187017002, | |
| "loss": 0.9363, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.1143740340030912, | |
| "grad_norm": 0.0020189261995255947, | |
| "learning_rate": 0.0009618753219989696, | |
| "loss": 0.9025, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.11746522411128284, | |
| "grad_norm": 0.0024259143974632025, | |
| "learning_rate": 0.0009608449252962391, | |
| "loss": 0.9646, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.1205564142194745, | |
| "grad_norm": 0.0015644305385649204, | |
| "learning_rate": 0.0009598145285935085, | |
| "loss": 0.9226, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.12364760432766615, | |
| "grad_norm": 0.0017478523077443242, | |
| "learning_rate": 0.000958784131890778, | |
| "loss": 0.9232, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.1267387944358578, | |
| "grad_norm": 0.002042934997007251, | |
| "learning_rate": 0.0009577537351880475, | |
| "loss": 0.9464, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.12982998454404945, | |
| "grad_norm": 0.0017407669220119715, | |
| "learning_rate": 0.0009567233384853169, | |
| "loss": 0.9258, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.13292117465224113, | |
| "grad_norm": 0.001527661457657814, | |
| "learning_rate": 0.0009556929417825863, | |
| "loss": 0.9104, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.13601236476043277, | |
| "grad_norm": 0.0014940258115530014, | |
| "learning_rate": 0.0009546625450798557, | |
| "loss": 0.9208, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.1391035548686244, | |
| "grad_norm": 0.0018555278657004237, | |
| "learning_rate": 0.0009536321483771252, | |
| "loss": 0.9441, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.14219474497681608, | |
| "grad_norm": 0.002033649478107691, | |
| "learning_rate": 0.0009526017516743947, | |
| "loss": 0.9155, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.14528593508500773, | |
| "grad_norm": 0.0015595933655276895, | |
| "learning_rate": 0.0009515713549716642, | |
| "loss": 0.9164, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.14837712519319937, | |
| "grad_norm": 0.0016203809063881636, | |
| "learning_rate": 0.0009505409582689336, | |
| "loss": 0.9267, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.15146831530139104, | |
| "grad_norm": 0.0014497883385047317, | |
| "learning_rate": 0.0009495105615662029, | |
| "loss": 0.9328, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.1545595054095827, | |
| "grad_norm": 0.0019730357453227043, | |
| "learning_rate": 0.0009484801648634724, | |
| "loss": 0.9057, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.15765069551777433, | |
| "grad_norm": 0.0020096744410693645, | |
| "learning_rate": 0.0009474497681607419, | |
| "loss": 0.9217, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.160741885625966, | |
| "grad_norm": 0.0019119798671454191, | |
| "learning_rate": 0.0009464193714580114, | |
| "loss": 0.9324, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.16383307573415765, | |
| "grad_norm": 0.0014629390789195895, | |
| "learning_rate": 0.0009453889747552809, | |
| "loss": 0.9424, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.16692426584234932, | |
| "grad_norm": 0.0016167230205610394, | |
| "learning_rate": 0.0009443585780525502, | |
| "loss": 0.8622, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.17001545595054096, | |
| "grad_norm": 0.0016633981140330434, | |
| "learning_rate": 0.0009433281813498197, | |
| "loss": 0.9366, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.1731066460587326, | |
| "grad_norm": 0.0015964192571118474, | |
| "learning_rate": 0.0009422977846470892, | |
| "loss": 0.9552, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.17619783616692428, | |
| "grad_norm": 0.0013956124894320965, | |
| "learning_rate": 0.0009412673879443586, | |
| "loss": 0.9995, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.17928902627511592, | |
| "grad_norm": 0.0014371597208082676, | |
| "learning_rate": 0.0009402369912416281, | |
| "loss": 0.8622, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.18238021638330756, | |
| "grad_norm": 0.0021855118684470654, | |
| "learning_rate": 0.0009392065945388974, | |
| "loss": 0.9862, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.18547140649149924, | |
| "grad_norm": 0.001544400816783309, | |
| "learning_rate": 0.0009381761978361669, | |
| "loss": 0.919, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.18856259659969088, | |
| "grad_norm": 0.0016864053905010223, | |
| "learning_rate": 0.0009371458011334364, | |
| "loss": 0.9405, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.19165378670788252, | |
| "grad_norm": 0.0021651415154337883, | |
| "learning_rate": 0.0009361154044307059, | |
| "loss": 0.9939, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.1947449768160742, | |
| "grad_norm": 0.001466871122829616, | |
| "learning_rate": 0.0009350850077279754, | |
| "loss": 0.8776, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.19783616692426584, | |
| "grad_norm": 0.0013067360268905759, | |
| "learning_rate": 0.0009340546110252446, | |
| "loss": 0.9684, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.2009273570324575, | |
| "grad_norm": 0.001740931300446391, | |
| "learning_rate": 0.0009330242143225141, | |
| "loss": 0.9786, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.20401854714064915, | |
| "grad_norm": 0.0017273235134780407, | |
| "learning_rate": 0.0009319938176197836, | |
| "loss": 0.9089, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.2071097372488408, | |
| "grad_norm": 0.001577013055793941, | |
| "learning_rate": 0.0009309634209170531, | |
| "loss": 0.9154, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.21020092735703247, | |
| "grad_norm": 0.001800213591195643, | |
| "learning_rate": 0.0009299330242143226, | |
| "loss": 0.9726, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.2132921174652241, | |
| "grad_norm": 0.0015475867548957467, | |
| "learning_rate": 0.000928902627511592, | |
| "loss": 0.8864, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.21638330757341576, | |
| "grad_norm": 0.002068218309432268, | |
| "learning_rate": 0.0009278722308088614, | |
| "loss": 0.8666, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.21947449768160743, | |
| "grad_norm": 0.0014894594205543399, | |
| "learning_rate": 0.0009268418341061309, | |
| "loss": 0.9178, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.22256568778979907, | |
| "grad_norm": 0.0018193743890151381, | |
| "learning_rate": 0.0009258114374034003, | |
| "loss": 0.913, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.22565687789799072, | |
| "grad_norm": 0.0019854323472827673, | |
| "learning_rate": 0.0009247810407006698, | |
| "loss": 0.9015, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.2287480680061824, | |
| "grad_norm": 0.0015401191776618361, | |
| "learning_rate": 0.0009237506439979392, | |
| "loss": 0.942, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.23183925811437403, | |
| "grad_norm": 0.0016189438756555319, | |
| "learning_rate": 0.0009227202472952086, | |
| "loss": 0.9647, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.23493044822256567, | |
| "grad_norm": 0.0018937138374894857, | |
| "learning_rate": 0.0009216898505924781, | |
| "loss": 0.9455, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.23802163833075735, | |
| "grad_norm": 0.0015417198883369565, | |
| "learning_rate": 0.0009206594538897476, | |
| "loss": 0.926, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.241112828438949, | |
| "grad_norm": 0.0014824847457930446, | |
| "learning_rate": 0.0009196290571870171, | |
| "loss": 0.8998, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.24420401854714066, | |
| "grad_norm": 0.0015080615412443876, | |
| "learning_rate": 0.0009185986604842864, | |
| "loss": 0.905, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.2472952086553323, | |
| "grad_norm": 0.0017820092616602778, | |
| "learning_rate": 0.0009175682637815559, | |
| "loss": 0.9504, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.250386398763524, | |
| "grad_norm": 0.001539805787615478, | |
| "learning_rate": 0.0009165378670788253, | |
| "loss": 0.9369, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.2534775888717156, | |
| "grad_norm": 0.001816835138015449, | |
| "learning_rate": 0.0009155074703760948, | |
| "loss": 0.8814, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.25656877897990726, | |
| "grad_norm": 0.0017855115002021194, | |
| "learning_rate": 0.0009144770736733643, | |
| "loss": 0.9066, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.2596599690880989, | |
| "grad_norm": 0.0015401588752865791, | |
| "learning_rate": 0.0009134466769706337, | |
| "loss": 0.9257, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.26275115919629055, | |
| "grad_norm": 0.0023962745908647776, | |
| "learning_rate": 0.0009124162802679032, | |
| "loss": 0.9428, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.26584234930448225, | |
| "grad_norm": 0.0014296959852799773, | |
| "learning_rate": 0.0009113858835651726, | |
| "loss": 0.9263, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.2689335394126739, | |
| "grad_norm": 0.0018063073512166739, | |
| "learning_rate": 0.000910355486862442, | |
| "loss": 0.9235, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.27202472952086554, | |
| "grad_norm": 0.0016391921089962125, | |
| "learning_rate": 0.0009093250901597115, | |
| "loss": 0.8952, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.2751159196290572, | |
| "grad_norm": 0.0023141205310821533, | |
| "learning_rate": 0.0009082946934569809, | |
| "loss": 0.9494, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.2782071097372488, | |
| "grad_norm": 0.0018004857702180743, | |
| "learning_rate": 0.0009072642967542504, | |
| "loss": 0.9346, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.28129829984544047, | |
| "grad_norm": 0.002469270955771208, | |
| "learning_rate": 0.0009062339000515199, | |
| "loss": 0.9245, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.28438948995363217, | |
| "grad_norm": 0.001746363122947514, | |
| "learning_rate": 0.0009052035033487893, | |
| "loss": 0.8899, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.2874806800618238, | |
| "grad_norm": 0.0017825034447014332, | |
| "learning_rate": 0.0009041731066460588, | |
| "loss": 0.9097, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.29057187017001546, | |
| "grad_norm": 0.0015837879618629813, | |
| "learning_rate": 0.0009031427099433283, | |
| "loss": 0.9406, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.2936630602782071, | |
| "grad_norm": 0.00214924244210124, | |
| "learning_rate": 0.0009021123132405976, | |
| "loss": 0.8884, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.29675425038639874, | |
| "grad_norm": 0.001666102441959083, | |
| "learning_rate": 0.0009010819165378671, | |
| "loss": 0.9393, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.29984544049459044, | |
| "grad_norm": 0.0013613239862024784, | |
| "learning_rate": 0.0009000515198351365, | |
| "loss": 0.9177, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.3029366306027821, | |
| "grad_norm": 0.0017140130512416363, | |
| "learning_rate": 0.000899021123132406, | |
| "loss": 0.9315, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.30602782071097373, | |
| "grad_norm": 0.001512790797278285, | |
| "learning_rate": 0.0008979907264296755, | |
| "loss": 0.9364, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.3091190108191654, | |
| "grad_norm": 0.0017727742670103908, | |
| "learning_rate": 0.0008969603297269449, | |
| "loss": 0.9175, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.312210200927357, | |
| "grad_norm": 0.0019065055530518293, | |
| "learning_rate": 0.0008959299330242144, | |
| "loss": 0.9585, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.31530139103554866, | |
| "grad_norm": 0.002561497036367655, | |
| "learning_rate": 0.0008948995363214837, | |
| "loss": 0.936, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.31839258114374036, | |
| "grad_norm": 0.0015061356825754046, | |
| "learning_rate": 0.0008938691396187532, | |
| "loss": 0.8856, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.321483771251932, | |
| "grad_norm": 0.0022356980480253696, | |
| "learning_rate": 0.0008928387429160227, | |
| "loss": 0.9152, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.32457496136012365, | |
| "grad_norm": 0.0017119398107752204, | |
| "learning_rate": 0.0008918083462132921, | |
| "loss": 0.8799, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.3276661514683153, | |
| "grad_norm": 0.0021370877511799335, | |
| "learning_rate": 0.0008907779495105616, | |
| "loss": 0.9308, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.33075734157650694, | |
| "grad_norm": 0.0019707437604665756, | |
| "learning_rate": 0.0008897475528078311, | |
| "loss": 0.9182, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.33384853168469864, | |
| "grad_norm": 0.001411254983395338, | |
| "learning_rate": 0.0008887171561051005, | |
| "loss": 0.9057, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.3369397217928903, | |
| "grad_norm": 0.0021633992437273264, | |
| "learning_rate": 0.00088768675940237, | |
| "loss": 0.9089, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.3400309119010819, | |
| "grad_norm": 0.001840689335949719, | |
| "learning_rate": 0.0008866563626996393, | |
| "loss": 0.8939, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.34312210200927357, | |
| "grad_norm": 0.0018973862752318382, | |
| "learning_rate": 0.0008856259659969088, | |
| "loss": 0.9206, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.3462132921174652, | |
| "grad_norm": 0.0015567062655463815, | |
| "learning_rate": 0.0008845955692941783, | |
| "loss": 0.9424, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.34930448222565685, | |
| "grad_norm": 0.0021733948960900307, | |
| "learning_rate": 0.0008835651725914478, | |
| "loss": 0.9965, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.35239567233384855, | |
| "grad_norm": 0.0025423939805477858, | |
| "learning_rate": 0.0008825347758887172, | |
| "loss": 0.8603, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.3554868624420402, | |
| "grad_norm": 0.0014392200391739607, | |
| "learning_rate": 0.0008815043791859866, | |
| "loss": 0.9483, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.35857805255023184, | |
| "grad_norm": 0.0015780443791300058, | |
| "learning_rate": 0.0008804739824832561, | |
| "loss": 0.8963, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.3616692426584235, | |
| "grad_norm": 0.002231495687738061, | |
| "learning_rate": 0.0008794435857805255, | |
| "loss": 0.9123, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.36476043276661513, | |
| "grad_norm": 0.0014009552542120218, | |
| "learning_rate": 0.000878413189077795, | |
| "loss": 0.9337, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.3678516228748068, | |
| "grad_norm": 0.0020310496911406517, | |
| "learning_rate": 0.0008773827923750644, | |
| "loss": 0.9137, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.37094281298299847, | |
| "grad_norm": 0.00151388393715024, | |
| "learning_rate": 0.0008763523956723338, | |
| "loss": 0.9091, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.3740340030911901, | |
| "grad_norm": 0.0018349160673096776, | |
| "learning_rate": 0.0008753219989696033, | |
| "loss": 0.9486, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.37712519319938176, | |
| "grad_norm": 0.0018245832761749625, | |
| "learning_rate": 0.0008742916022668728, | |
| "loss": 0.9163, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.3802163833075734, | |
| "grad_norm": 0.002115410752594471, | |
| "learning_rate": 0.0008732612055641423, | |
| "loss": 0.9043, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.38330757341576505, | |
| "grad_norm": 0.0019245495786890388, | |
| "learning_rate": 0.0008722308088614118, | |
| "loss": 0.9212, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.38639876352395675, | |
| "grad_norm": 0.001513317576609552, | |
| "learning_rate": 0.000871200412158681, | |
| "loss": 0.9068, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.3894899536321484, | |
| "grad_norm": 0.0017635183176025748, | |
| "learning_rate": 0.0008701700154559505, | |
| "loss": 0.9261, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.39258114374034003, | |
| "grad_norm": 0.0017686467617750168, | |
| "learning_rate": 0.00086913961875322, | |
| "loss": 0.8763, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.3956723338485317, | |
| "grad_norm": 0.0015009143389761448, | |
| "learning_rate": 0.0008681092220504895, | |
| "loss": 0.8968, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.3987635239567233, | |
| "grad_norm": 0.0014831022126600146, | |
| "learning_rate": 0.000867078825347759, | |
| "loss": 0.8927, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.401854714064915, | |
| "grad_norm": 0.0029206760227680206, | |
| "learning_rate": 0.0008660484286450283, | |
| "loss": 0.9455, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.40494590417310666, | |
| "grad_norm": 0.0014479625970125198, | |
| "learning_rate": 0.0008650180319422978, | |
| "loss": 0.9229, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.4080370942812983, | |
| "grad_norm": 0.0014661536552011967, | |
| "learning_rate": 0.0008639876352395672, | |
| "loss": 0.9088, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.41112828438948995, | |
| "grad_norm": 0.0014888847945258021, | |
| "learning_rate": 0.0008629572385368367, | |
| "loss": 0.9633, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.4142194744976816, | |
| "grad_norm": 0.0017181559232994914, | |
| "learning_rate": 0.0008619268418341062, | |
| "loss": 0.9224, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.41731066460587324, | |
| "grad_norm": 0.0015694062458351254, | |
| "learning_rate": 0.0008608964451313755, | |
| "loss": 0.8789, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.42040185471406494, | |
| "grad_norm": 0.001495172269642353, | |
| "learning_rate": 0.000859866048428645, | |
| "loss": 0.8666, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.4234930448222566, | |
| "grad_norm": 0.002242365386337042, | |
| "learning_rate": 0.0008588356517259145, | |
| "loss": 0.9108, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.4265842349304482, | |
| "grad_norm": 0.0014668918447569013, | |
| "learning_rate": 0.000857805255023184, | |
| "loss": 0.916, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.42967542503863987, | |
| "grad_norm": 0.001498398371040821, | |
| "learning_rate": 0.0008567748583204535, | |
| "loss": 0.9183, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.4327666151468315, | |
| "grad_norm": 0.001534744049422443, | |
| "learning_rate": 0.0008557444616177228, | |
| "loss": 0.9114, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.43585780525502316, | |
| "grad_norm": 0.0014011348830536008, | |
| "learning_rate": 0.0008547140649149922, | |
| "loss": 0.9394, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.43894899536321486, | |
| "grad_norm": 0.0016137160127982497, | |
| "learning_rate": 0.0008536836682122617, | |
| "loss": 0.8852, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.4420401854714065, | |
| "grad_norm": 0.0018244803650304675, | |
| "learning_rate": 0.0008526532715095312, | |
| "loss": 0.9006, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.44513137557959814, | |
| "grad_norm": 0.002001643180847168, | |
| "learning_rate": 0.0008516228748068007, | |
| "loss": 0.8877, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.4482225656877898, | |
| "grad_norm": 0.001545743434689939, | |
| "learning_rate": 0.0008505924781040701, | |
| "loss": 0.9169, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.45131375579598143, | |
| "grad_norm": 0.0015252763405442238, | |
| "learning_rate": 0.0008495620814013395, | |
| "loss": 0.9302, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.45440494590417313, | |
| "grad_norm": 0.0018486313056200743, | |
| "learning_rate": 0.0008485316846986089, | |
| "loss": 0.8845, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.4574961360123648, | |
| "grad_norm": 0.0013468407560139894, | |
| "learning_rate": 0.0008475012879958784, | |
| "loss": 0.9079, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.4605873261205564, | |
| "grad_norm": 0.0011928731109946966, | |
| "learning_rate": 0.0008464708912931479, | |
| "loss": 0.8997, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.46367851622874806, | |
| "grad_norm": 0.0018724995898082852, | |
| "learning_rate": 0.0008454404945904173, | |
| "loss": 0.944, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.4667697063369397, | |
| "grad_norm": 0.0021075448021292686, | |
| "learning_rate": 0.0008444100978876868, | |
| "loss": 0.9155, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.46986089644513135, | |
| "grad_norm": 0.0016975891776382923, | |
| "learning_rate": 0.0008433797011849562, | |
| "loss": 0.9349, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.47295208655332305, | |
| "grad_norm": 0.001552650355733931, | |
| "learning_rate": 0.0008423493044822257, | |
| "loss": 0.9145, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.4760432766615147, | |
| "grad_norm": 0.0016928149852901697, | |
| "learning_rate": 0.0008413189077794952, | |
| "loss": 0.8694, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.47913446676970634, | |
| "grad_norm": 0.0015649759443476796, | |
| "learning_rate": 0.0008402885110767645, | |
| "loss": 0.9505, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.482225656877898, | |
| "grad_norm": 0.0020843998063355684, | |
| "learning_rate": 0.000839258114374034, | |
| "loss": 0.9463, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.4853168469860896, | |
| "grad_norm": 0.0012236249167472124, | |
| "learning_rate": 0.0008382277176713034, | |
| "loss": 0.9033, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.4884080370942813, | |
| "grad_norm": 0.0017901939572766423, | |
| "learning_rate": 0.0008371973209685729, | |
| "loss": 0.911, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.49149922720247297, | |
| "grad_norm": 0.0018610935658216476, | |
| "learning_rate": 0.0008361669242658424, | |
| "loss": 0.8655, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.4945904173106646, | |
| "grad_norm": 0.001789487199857831, | |
| "learning_rate": 0.0008351365275631119, | |
| "loss": 0.9427, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.49768160741885625, | |
| "grad_norm": 0.00190592254512012, | |
| "learning_rate": 0.0008341061308603813, | |
| "loss": 0.9192, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.500772797527048, | |
| "grad_norm": 0.0016090746503323317, | |
| "learning_rate": 0.0008330757341576506, | |
| "loss": 0.9487, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.5038639876352395, | |
| "grad_norm": 0.0016335912514477968, | |
| "learning_rate": 0.0008320453374549201, | |
| "loss": 0.9051, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.5069551777434312, | |
| "grad_norm": 0.0016785924090072513, | |
| "learning_rate": 0.0008310149407521896, | |
| "loss": 0.9104, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.5100463678516228, | |
| "grad_norm": 0.0022380806040018797, | |
| "learning_rate": 0.0008299845440494591, | |
| "loss": 0.9038, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.5131375579598145, | |
| "grad_norm": 0.0016855926951393485, | |
| "learning_rate": 0.0008289541473467285, | |
| "loss": 0.9256, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.5162287480680062, | |
| "grad_norm": 0.0019196901703253388, | |
| "learning_rate": 0.000827923750643998, | |
| "loss": 0.9271, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.5193199381761978, | |
| "grad_norm": 0.001529015600681305, | |
| "learning_rate": 0.0008268933539412674, | |
| "loss": 0.9053, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.5224111282843895, | |
| "grad_norm": 0.001290348474867642, | |
| "learning_rate": 0.0008258629572385369, | |
| "loss": 0.8879, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.5255023183925811, | |
| "grad_norm": 0.0017919385572895408, | |
| "learning_rate": 0.0008248325605358063, | |
| "loss": 0.8537, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.5285935085007728, | |
| "grad_norm": 0.0017021787352859974, | |
| "learning_rate": 0.0008238021638330757, | |
| "loss": 0.9126, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.5316846986089645, | |
| "grad_norm": 0.0017202612943947315, | |
| "learning_rate": 0.0008227717671303452, | |
| "loss": 0.8977, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.5347758887171561, | |
| "grad_norm": 0.0021942094899713993, | |
| "learning_rate": 0.0008217413704276147, | |
| "loss": 0.944, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.5378670788253478, | |
| "grad_norm": 0.001882906537503004, | |
| "learning_rate": 0.0008207109737248841, | |
| "loss": 0.9704, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.5409582689335394, | |
| "grad_norm": 0.0015875013777986169, | |
| "learning_rate": 0.0008196805770221536, | |
| "loss": 0.927, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.5440494590417311, | |
| "grad_norm": 0.0017645510379225016, | |
| "learning_rate": 0.000818650180319423, | |
| "loss": 0.9719, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.5471406491499228, | |
| "grad_norm": 0.0016093801241368055, | |
| "learning_rate": 0.0008176197836166924, | |
| "loss": 0.8753, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.5502318392581144, | |
| "grad_norm": 0.0016733432421460748, | |
| "learning_rate": 0.0008165893869139619, | |
| "loss": 0.9132, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.5533230293663061, | |
| "grad_norm": 0.0014284460339695215, | |
| "learning_rate": 0.0008155589902112313, | |
| "loss": 0.9242, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.5564142194744977, | |
| "grad_norm": 0.0018177167512476444, | |
| "learning_rate": 0.0008145285935085008, | |
| "loss": 0.9469, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.5595054095826894, | |
| "grad_norm": 0.0020286261569708586, | |
| "learning_rate": 0.0008134981968057702, | |
| "loss": 0.8884, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.5625965996908809, | |
| "grad_norm": 0.0014468576991930604, | |
| "learning_rate": 0.0008124678001030397, | |
| "loss": 0.908, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.5656877897990726, | |
| "grad_norm": 0.001559574855491519, | |
| "learning_rate": 0.0008114374034003092, | |
| "loss": 0.9104, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.5687789799072643, | |
| "grad_norm": 0.0017769662663340569, | |
| "learning_rate": 0.0008104070066975787, | |
| "loss": 0.9222, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.5718701700154559, | |
| "grad_norm": 0.001862520119175315, | |
| "learning_rate": 0.000809376609994848, | |
| "loss": 0.8862, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.5749613601236476, | |
| "grad_norm": 0.0021106936037540436, | |
| "learning_rate": 0.0008083462132921174, | |
| "loss": 0.8862, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.5780525502318392, | |
| "grad_norm": 0.0013291973154991865, | |
| "learning_rate": 0.0008073158165893869, | |
| "loss": 0.9381, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.5811437403400309, | |
| "grad_norm": 0.001646311953663826, | |
| "learning_rate": 0.0008062854198866564, | |
| "loss": 0.8862, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.5842349304482226, | |
| "grad_norm": 0.0015801474219188094, | |
| "learning_rate": 0.0008052550231839259, | |
| "loss": 0.9402, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.5873261205564142, | |
| "grad_norm": 0.0015533153200522065, | |
| "learning_rate": 0.0008042246264811953, | |
| "loss": 0.9242, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.5904173106646059, | |
| "grad_norm": 0.0016890340484678745, | |
| "learning_rate": 0.0008031942297784647, | |
| "loss": 0.9234, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.5935085007727975, | |
| "grad_norm": 0.0014929634053260088, | |
| "learning_rate": 0.0008021638330757341, | |
| "loss": 0.9383, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.5965996908809892, | |
| "grad_norm": 0.001471440540626645, | |
| "learning_rate": 0.0008011334363730036, | |
| "loss": 0.908, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.5996908809891809, | |
| "grad_norm": 0.00180807092692703, | |
| "learning_rate": 0.0008001030396702731, | |
| "loss": 0.9211, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.6027820710973725, | |
| "grad_norm": 0.0016187585424631834, | |
| "learning_rate": 0.0007990726429675426, | |
| "loss": 0.9254, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.6058732612055642, | |
| "grad_norm": 0.0016198824159801006, | |
| "learning_rate": 0.0007980422462648119, | |
| "loss": 0.9118, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.6089644513137558, | |
| "grad_norm": 0.0017275193240493536, | |
| "learning_rate": 0.0007970118495620814, | |
| "loss": 0.9444, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.6120556414219475, | |
| "grad_norm": 0.002495982451364398, | |
| "learning_rate": 0.0007959814528593509, | |
| "loss": 0.8879, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.615146831530139, | |
| "grad_norm": 0.0013608982553705573, | |
| "learning_rate": 0.0007949510561566204, | |
| "loss": 0.8695, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.6182380216383307, | |
| "grad_norm": 0.0015398486284539104, | |
| "learning_rate": 0.0007939206594538898, | |
| "loss": 0.916, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.6213292117465224, | |
| "grad_norm": 0.0016108902636915445, | |
| "learning_rate": 0.0007928902627511591, | |
| "loss": 0.9244, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.624420401854714, | |
| "grad_norm": 0.0016804412007331848, | |
| "learning_rate": 0.0007918598660484286, | |
| "loss": 0.9358, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.6275115919629057, | |
| "grad_norm": 0.0018602460622787476, | |
| "learning_rate": 0.0007908294693456981, | |
| "loss": 0.8829, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.6306027820710973, | |
| "grad_norm": 0.0017209933139383793, | |
| "learning_rate": 0.0007897990726429676, | |
| "loss": 0.9365, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.633693972179289, | |
| "grad_norm": 0.002164942678064108, | |
| "learning_rate": 0.0007887686759402371, | |
| "loss": 0.9349, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.6367851622874807, | |
| "grad_norm": 0.0021378265228122473, | |
| "learning_rate": 0.0007877382792375064, | |
| "loss": 0.8884, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.6398763523956723, | |
| "grad_norm": 0.001859784359112382, | |
| "learning_rate": 0.0007867078825347758, | |
| "loss": 0.9461, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.642967542503864, | |
| "grad_norm": 0.001505703548900783, | |
| "learning_rate": 0.0007856774858320453, | |
| "loss": 0.9284, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.6460587326120556, | |
| "grad_norm": 0.0014758047182112932, | |
| "learning_rate": 0.0007846470891293148, | |
| "loss": 0.8943, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.6491499227202473, | |
| "grad_norm": 0.001482132589444518, | |
| "learning_rate": 0.0007836166924265843, | |
| "loss": 0.9309, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.652241112828439, | |
| "grad_norm": 0.0018712684977799654, | |
| "learning_rate": 0.0007825862957238537, | |
| "loss": 0.9705, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.6553323029366306, | |
| "grad_norm": 0.0025388901121914387, | |
| "learning_rate": 0.0007815558990211231, | |
| "loss": 0.9129, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.6584234930448223, | |
| "grad_norm": 0.001495323609560728, | |
| "learning_rate": 0.0007805255023183926, | |
| "loss": 0.9435, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.6615146831530139, | |
| "grad_norm": 0.0016260349657386541, | |
| "learning_rate": 0.0007794951056156621, | |
| "loss": 0.9124, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.6646058732612056, | |
| "grad_norm": 0.0019677565433084965, | |
| "learning_rate": 0.0007784647089129315, | |
| "loss": 0.9317, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.6676970633693973, | |
| "grad_norm": 0.0023265851195901632, | |
| "learning_rate": 0.0007774343122102009, | |
| "loss": 0.9538, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.6707882534775889, | |
| "grad_norm": 0.0014457949437201023, | |
| "learning_rate": 0.0007764039155074703, | |
| "loss": 0.9538, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.6738794435857806, | |
| "grad_norm": 0.0014781310455873609, | |
| "learning_rate": 0.0007753735188047398, | |
| "loss": 0.8816, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.6769706336939721, | |
| "grad_norm": 0.0017087183659896255, | |
| "learning_rate": 0.0007743431221020093, | |
| "loss": 0.9142, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.6800618238021638, | |
| "grad_norm": 0.002620991552248597, | |
| "learning_rate": 0.0007733127253992788, | |
| "loss": 0.9345, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.6831530139103554, | |
| "grad_norm": 0.001568863750435412, | |
| "learning_rate": 0.0007722823286965483, | |
| "loss": 0.9089, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.6862442040185471, | |
| "grad_norm": 0.0017851140582934022, | |
| "learning_rate": 0.0007712519319938175, | |
| "loss": 0.8943, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.6893353941267388, | |
| "grad_norm": 0.0013759072171524167, | |
| "learning_rate": 0.000770221535291087, | |
| "loss": 0.9176, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.6924265842349304, | |
| "grad_norm": 0.0015892439987510443, | |
| "learning_rate": 0.0007691911385883565, | |
| "loss": 0.8854, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.6955177743431221, | |
| "grad_norm": 0.001841311459429562, | |
| "learning_rate": 0.000768160741885626, | |
| "loss": 0.8824, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.6986089644513137, | |
| "grad_norm": 0.0016689961776137352, | |
| "learning_rate": 0.0007671303451828955, | |
| "loss": 0.9442, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.7017001545595054, | |
| "grad_norm": 0.0018330076709389687, | |
| "learning_rate": 0.0007660999484801649, | |
| "loss": 0.9398, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.7047913446676971, | |
| "grad_norm": 0.0021366167347878218, | |
| "learning_rate": 0.0007650695517774343, | |
| "loss": 0.9507, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.7078825347758887, | |
| "grad_norm": 0.0017535964725539088, | |
| "learning_rate": 0.0007640391550747038, | |
| "loss": 0.8902, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.7109737248840804, | |
| "grad_norm": 0.0015847982140257955, | |
| "learning_rate": 0.0007630087583719732, | |
| "loss": 0.9163, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.714064914992272, | |
| "grad_norm": 0.0016199509846046567, | |
| "learning_rate": 0.0007619783616692427, | |
| "loss": 0.8775, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.7171561051004637, | |
| "grad_norm": 0.001773869269527495, | |
| "learning_rate": 0.0007609479649665121, | |
| "loss": 0.9406, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.7202472952086554, | |
| "grad_norm": 0.0015184408985078335, | |
| "learning_rate": 0.0007599175682637816, | |
| "loss": 0.8945, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.723338485316847, | |
| "grad_norm": 0.005015387199819088, | |
| "learning_rate": 0.000758887171561051, | |
| "loss": 0.9407, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.7264296754250387, | |
| "grad_norm": 0.0020386965479701757, | |
| "learning_rate": 0.0007578567748583205, | |
| "loss": 0.9122, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.7295208655332303, | |
| "grad_norm": 0.002024392830207944, | |
| "learning_rate": 0.00075682637815559, | |
| "loss": 0.8902, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.732612055641422, | |
| "grad_norm": 0.0016152510652318597, | |
| "learning_rate": 0.0007557959814528593, | |
| "loss": 0.907, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.7357032457496137, | |
| "grad_norm": 0.001684612943790853, | |
| "learning_rate": 0.0007547655847501288, | |
| "loss": 0.8557, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.7387944358578052, | |
| "grad_norm": 0.0018417539540678263, | |
| "learning_rate": 0.0007537351880473982, | |
| "loss": 0.932, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.7418856259659969, | |
| "grad_norm": 0.0020610857754945755, | |
| "learning_rate": 0.0007527047913446677, | |
| "loss": 0.9125, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.7449768160741885, | |
| "grad_norm": 0.0023381186183542013, | |
| "learning_rate": 0.0007516743946419372, | |
| "loss": 0.9251, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.7480680061823802, | |
| "grad_norm": 0.002318663988262415, | |
| "learning_rate": 0.0007506439979392066, | |
| "loss": 0.9043, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.7511591962905718, | |
| "grad_norm": 0.0017375986790284514, | |
| "learning_rate": 0.0007496136012364761, | |
| "loss": 0.9061, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.7542503863987635, | |
| "grad_norm": 0.0018655112944543362, | |
| "learning_rate": 0.0007485832045337456, | |
| "loss": 1.0007, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.7573415765069552, | |
| "grad_norm": 0.0019265462178736925, | |
| "learning_rate": 0.0007475528078310149, | |
| "loss": 0.9089, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.7604327666151468, | |
| "grad_norm": 0.0018528448417782784, | |
| "learning_rate": 0.0007465224111282844, | |
| "loss": 0.8856, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.7635239567233385, | |
| "grad_norm": 0.0016357959248125553, | |
| "learning_rate": 0.0007454920144255538, | |
| "loss": 0.8712, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.7666151468315301, | |
| "grad_norm": 0.0013976657064631581, | |
| "learning_rate": 0.0007444616177228233, | |
| "loss": 0.9761, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.7697063369397218, | |
| "grad_norm": 0.002004920970648527, | |
| "learning_rate": 0.0007434312210200928, | |
| "loss": 0.9023, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.7727975270479135, | |
| "grad_norm": 0.0015114221023395658, | |
| "learning_rate": 0.0007424008243173622, | |
| "loss": 0.9048, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.7758887171561051, | |
| "grad_norm": 0.0016935282619670033, | |
| "learning_rate": 0.0007413704276146317, | |
| "loss": 0.9617, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.7789799072642968, | |
| "grad_norm": 0.0015191826969385147, | |
| "learning_rate": 0.000740340030911901, | |
| "loss": 0.927, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.7820710973724884, | |
| "grad_norm": 0.0022661250550299883, | |
| "learning_rate": 0.0007393096342091705, | |
| "loss": 0.8871, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.7851622874806801, | |
| "grad_norm": 0.002032969379797578, | |
| "learning_rate": 0.00073827923750644, | |
| "loss": 0.9149, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.7882534775888718, | |
| "grad_norm": 0.0017924593994393945, | |
| "learning_rate": 0.0007372488408037095, | |
| "loss": 0.9564, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.7913446676970634, | |
| "grad_norm": 0.0014548065373674035, | |
| "learning_rate": 0.0007362184441009789, | |
| "loss": 0.9095, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.794435857805255, | |
| "grad_norm": 0.0025248208548873663, | |
| "learning_rate": 0.0007351880473982483, | |
| "loss": 0.9337, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.7975270479134466, | |
| "grad_norm": 0.002107022562995553, | |
| "learning_rate": 0.0007341576506955178, | |
| "loss": 0.9482, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.8006182380216383, | |
| "grad_norm": 0.0016369119985029101, | |
| "learning_rate": 0.0007331272539927873, | |
| "loss": 0.9035, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.80370942812983, | |
| "grad_norm": 0.0018871091306209564, | |
| "learning_rate": 0.0007320968572900567, | |
| "loss": 0.8882, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.8068006182380216, | |
| "grad_norm": 0.0019260449334979057, | |
| "learning_rate": 0.0007310664605873261, | |
| "loss": 0.9074, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.8098918083462133, | |
| "grad_norm": 0.0018819051329046488, | |
| "learning_rate": 0.0007300360638845955, | |
| "loss": 0.9218, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.8129829984544049, | |
| "grad_norm": 0.0015719749499112368, | |
| "learning_rate": 0.000729005667181865, | |
| "loss": 0.9155, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.8160741885625966, | |
| "grad_norm": 0.0015479732537642121, | |
| "learning_rate": 0.0007279752704791345, | |
| "loss": 0.909, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.8191653786707882, | |
| "grad_norm": 0.0013954916503280401, | |
| "learning_rate": 0.000726944873776404, | |
| "loss": 0.9217, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.8222565687789799, | |
| "grad_norm": 0.0013768866192549467, | |
| "learning_rate": 0.0007259144770736735, | |
| "loss": 0.9496, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.8253477588871716, | |
| "grad_norm": 0.001571536879055202, | |
| "learning_rate": 0.0007248840803709427, | |
| "loss": 0.9305, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.8284389489953632, | |
| "grad_norm": 0.002294939709827304, | |
| "learning_rate": 0.0007238536836682122, | |
| "loss": 0.9262, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.8315301391035549, | |
| "grad_norm": 0.0016881937626749277, | |
| "learning_rate": 0.0007228232869654817, | |
| "loss": 0.8624, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.8346213292117465, | |
| "grad_norm": 0.001524370047263801, | |
| "learning_rate": 0.0007217928902627512, | |
| "loss": 0.9194, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.8377125193199382, | |
| "grad_norm": 0.0020258829463273287, | |
| "learning_rate": 0.0007207624935600207, | |
| "loss": 0.9034, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.8408037094281299, | |
| "grad_norm": 0.0019363865721970797, | |
| "learning_rate": 0.00071973209685729, | |
| "loss": 0.929, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.8438948995363215, | |
| "grad_norm": 0.00164938741363585, | |
| "learning_rate": 0.0007187017001545595, | |
| "loss": 0.9569, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.8469860896445132, | |
| "grad_norm": 0.0016705304151400924, | |
| "learning_rate": 0.000717671303451829, | |
| "loss": 0.8835, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.8500772797527048, | |
| "grad_norm": 0.0018051480874419212, | |
| "learning_rate": 0.0007166409067490984, | |
| "loss": 0.8838, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.8531684698608965, | |
| "grad_norm": 0.0015429792692884803, | |
| "learning_rate": 0.0007156105100463679, | |
| "loss": 0.9004, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.8562596599690881, | |
| "grad_norm": 0.002758611924946308, | |
| "learning_rate": 0.0007145801133436372, | |
| "loss": 0.9299, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.8593508500772797, | |
| "grad_norm": 0.002330324612557888, | |
| "learning_rate": 0.0007135497166409067, | |
| "loss": 0.9749, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.8624420401854714, | |
| "grad_norm": 0.0015028082998469472, | |
| "learning_rate": 0.0007125193199381762, | |
| "loss": 0.9128, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.865533230293663, | |
| "grad_norm": 0.0014309794642031193, | |
| "learning_rate": 0.0007114889232354457, | |
| "loss": 0.8633, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.8686244204018547, | |
| "grad_norm": 0.0013788605574518442, | |
| "learning_rate": 0.0007104585265327152, | |
| "loss": 0.921, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.8717156105100463, | |
| "grad_norm": 0.0019114799797534943, | |
| "learning_rate": 0.0007094281298299847, | |
| "loss": 0.9312, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.874806800618238, | |
| "grad_norm": 0.002490201499313116, | |
| "learning_rate": 0.0007083977331272539, | |
| "loss": 0.9027, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.8778979907264297, | |
| "grad_norm": 0.0015281651867553592, | |
| "learning_rate": 0.0007073673364245234, | |
| "loss": 0.9189, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.8809891808346213, | |
| "grad_norm": 0.0015036484692245722, | |
| "learning_rate": 0.0007063369397217929, | |
| "loss": 0.8979, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.884080370942813, | |
| "grad_norm": 0.0015251452568918467, | |
| "learning_rate": 0.0007053065430190624, | |
| "loss": 0.9028, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.8871715610510046, | |
| "grad_norm": 0.001428957679308951, | |
| "learning_rate": 0.0007042761463163319, | |
| "loss": 0.8717, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.8902627511591963, | |
| "grad_norm": 0.0015625401865690947, | |
| "learning_rate": 0.0007032457496136012, | |
| "loss": 0.9324, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.893353941267388, | |
| "grad_norm": 0.001634020241908729, | |
| "learning_rate": 0.0007022153529108707, | |
| "loss": 0.8697, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.8964451313755796, | |
| "grad_norm": 0.0015363607089966536, | |
| "learning_rate": 0.0007011849562081401, | |
| "loss": 0.8956, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.8995363214837713, | |
| "grad_norm": 0.0012279663933441043, | |
| "learning_rate": 0.0007001545595054096, | |
| "loss": 0.9616, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.9026275115919629, | |
| "grad_norm": 0.0015052827075123787, | |
| "learning_rate": 0.0006991241628026791, | |
| "loss": 0.8944, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.9057187017001546, | |
| "grad_norm": 0.001815865165553987, | |
| "learning_rate": 0.0006980937660999485, | |
| "loss": 0.8739, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.9088098918083463, | |
| "grad_norm": 0.0019412669353187084, | |
| "learning_rate": 0.0006970633693972179, | |
| "loss": 0.8708, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.9119010819165378, | |
| "grad_norm": 0.001834962284192443, | |
| "learning_rate": 0.0006960329726944874, | |
| "loss": 0.9456, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.9149922720247295, | |
| "grad_norm": 0.002000207779929042, | |
| "learning_rate": 0.0006950025759917569, | |
| "loss": 0.9188, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.9180834621329211, | |
| "grad_norm": 0.0017901693936437368, | |
| "learning_rate": 0.0006939721792890264, | |
| "loss": 0.9377, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.9211746522411128, | |
| "grad_norm": 0.002317288890480995, | |
| "learning_rate": 0.0006929417825862957, | |
| "loss": 0.9159, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.9242658423493045, | |
| "grad_norm": 0.0019505108939483762, | |
| "learning_rate": 0.0006919113858835651, | |
| "loss": 0.9171, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.9273570324574961, | |
| "grad_norm": 0.0016651200130581856, | |
| "learning_rate": 0.0006908809891808346, | |
| "loss": 0.9331, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.9304482225656878, | |
| "grad_norm": 0.0016768771456554532, | |
| "learning_rate": 0.0006898505924781041, | |
| "loss": 0.9731, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.9335394126738794, | |
| "grad_norm": 0.0020638711284846067, | |
| "learning_rate": 0.0006888201957753736, | |
| "loss": 0.9023, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.9366306027820711, | |
| "grad_norm": 0.001335518783889711, | |
| "learning_rate": 0.000687789799072643, | |
| "loss": 0.917, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.9397217928902627, | |
| "grad_norm": 0.0013942529913038015, | |
| "learning_rate": 0.0006867594023699125, | |
| "loss": 0.9273, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.9428129829984544, | |
| "grad_norm": 0.0013294010423123837, | |
| "learning_rate": 0.0006857290056671818, | |
| "loss": 0.8909, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.9459041731066461, | |
| "grad_norm": 0.0017269228119403124, | |
| "learning_rate": 0.0006846986089644513, | |
| "loss": 0.9405, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.9489953632148377, | |
| "grad_norm": 0.0018391566118225455, | |
| "learning_rate": 0.0006836682122617208, | |
| "loss": 0.9, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.9520865533230294, | |
| "grad_norm": 0.0018784053390845656, | |
| "learning_rate": 0.0006826378155589902, | |
| "loss": 0.9081, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.955177743431221, | |
| "grad_norm": 0.0017941402038559318, | |
| "learning_rate": 0.0006816074188562597, | |
| "loss": 0.9343, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.9582689335394127, | |
| "grad_norm": 0.0019338412676006556, | |
| "learning_rate": 0.0006805770221535291, | |
| "loss": 0.8904, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.9613601236476044, | |
| "grad_norm": 0.0016496024327352643, | |
| "learning_rate": 0.0006795466254507986, | |
| "loss": 0.8983, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.964451313755796, | |
| "grad_norm": 0.0015189488185569644, | |
| "learning_rate": 0.0006785162287480681, | |
| "loss": 0.9118, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.9675425038639877, | |
| "grad_norm": 0.001589839463122189, | |
| "learning_rate": 0.0006774858320453374, | |
| "loss": 0.8996, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.9706336939721792, | |
| "grad_norm": 0.00146203744225204, | |
| "learning_rate": 0.0006764554353426069, | |
| "loss": 0.9122, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.973724884080371, | |
| "grad_norm": 0.0015465939650312066, | |
| "learning_rate": 0.0006754250386398764, | |
| "loss": 0.935, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.9768160741885626, | |
| "grad_norm": 0.002063942141830921, | |
| "learning_rate": 0.0006743946419371458, | |
| "loss": 0.9088, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.9799072642967542, | |
| "grad_norm": 0.0012433248339220881, | |
| "learning_rate": 0.0006733642452344153, | |
| "loss": 0.8963, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.9829984544049459, | |
| "grad_norm": 0.001582261291332543, | |
| "learning_rate": 0.0006723338485316847, | |
| "loss": 0.9304, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.9860896445131375, | |
| "grad_norm": 0.0015674213645979762, | |
| "learning_rate": 0.0006713034518289542, | |
| "loss": 0.9299, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.9891808346213292, | |
| "grad_norm": 0.0017826062394306064, | |
| "learning_rate": 0.0006702730551262236, | |
| "loss": 0.8951, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.9922720247295209, | |
| "grad_norm": 0.002112460555508733, | |
| "learning_rate": 0.000669242658423493, | |
| "loss": 0.9393, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 0.9953632148377125, | |
| "grad_norm": 0.0014212781097739935, | |
| "learning_rate": 0.0006682122617207625, | |
| "loss": 0.916, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.9984544049459042, | |
| "grad_norm": 0.001774617237970233, | |
| "learning_rate": 0.0006671818650180319, | |
| "loss": 0.9271, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 1.001545595054096, | |
| "grad_norm": 0.0016346676275134087, | |
| "learning_rate": 0.0006661514683153014, | |
| "loss": 0.8911, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 1.0046367851622875, | |
| "grad_norm": 0.0014871886232867837, | |
| "learning_rate": 0.0006651210716125709, | |
| "loss": 0.8988, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.007727975270479, | |
| "grad_norm": 0.0014850738225504756, | |
| "learning_rate": 0.0006640906749098404, | |
| "loss": 0.8936, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 1.010819165378671, | |
| "grad_norm": 0.0015949340304359794, | |
| "learning_rate": 0.0006630602782071098, | |
| "loss": 0.8987, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 1.0139103554868625, | |
| "grad_norm": 0.0019407504005357623, | |
| "learning_rate": 0.0006620298815043791, | |
| "loss": 0.8736, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 1.017001545595054, | |
| "grad_norm": 0.0019730369094759226, | |
| "learning_rate": 0.0006609994848016486, | |
| "loss": 0.8818, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 1.0200927357032457, | |
| "grad_norm": 0.0018432583892717957, | |
| "learning_rate": 0.0006599690880989181, | |
| "loss": 0.9042, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.0231839258114375, | |
| "grad_norm": 0.0017056462820619345, | |
| "learning_rate": 0.0006589386913961876, | |
| "loss": 0.932, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 1.026275115919629, | |
| "grad_norm": 0.0015121812466531992, | |
| "learning_rate": 0.000657908294693457, | |
| "loss": 0.8639, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 1.0293663060278206, | |
| "grad_norm": 0.004580393433570862, | |
| "learning_rate": 0.0006568778979907264, | |
| "loss": 0.9044, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 1.0324574961360125, | |
| "grad_norm": 0.0015995085705071688, | |
| "learning_rate": 0.0006558475012879959, | |
| "loss": 0.9453, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 1.035548686244204, | |
| "grad_norm": 0.002061218721792102, | |
| "learning_rate": 0.0006548171045852653, | |
| "loss": 0.8802, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.0386398763523956, | |
| "grad_norm": 0.0015771668404340744, | |
| "learning_rate": 0.0006537867078825348, | |
| "loss": 0.8759, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 1.0417310664605872, | |
| "grad_norm": 0.0015714009059593081, | |
| "learning_rate": 0.0006527563111798043, | |
| "loss": 0.8915, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 1.044822256568779, | |
| "grad_norm": 0.0019235257059335709, | |
| "learning_rate": 0.0006517259144770736, | |
| "loss": 0.9143, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 1.0479134466769706, | |
| "grad_norm": 0.0017326029483228922, | |
| "learning_rate": 0.0006506955177743431, | |
| "loss": 0.8883, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 1.0510046367851622, | |
| "grad_norm": 0.001543081016279757, | |
| "learning_rate": 0.0006496651210716126, | |
| "loss": 0.9235, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.054095826893354, | |
| "grad_norm": 0.0018034736858680844, | |
| "learning_rate": 0.0006486347243688821, | |
| "loss": 0.9445, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 1.0571870170015456, | |
| "grad_norm": 0.001585203455761075, | |
| "learning_rate": 0.0006476043276661516, | |
| "loss": 0.9192, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 1.0602782071097372, | |
| "grad_norm": 0.0018698821077123284, | |
| "learning_rate": 0.0006465739309634208, | |
| "loss": 0.9463, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 1.063369397217929, | |
| "grad_norm": 0.0014568824553862214, | |
| "learning_rate": 0.0006455435342606903, | |
| "loss": 0.8907, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 1.0664605873261206, | |
| "grad_norm": 0.002080600941553712, | |
| "learning_rate": 0.0006445131375579598, | |
| "loss": 0.9314, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.0695517774343122, | |
| "grad_norm": 0.0016853328561410308, | |
| "learning_rate": 0.0006434827408552293, | |
| "loss": 0.9277, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 1.0726429675425038, | |
| "grad_norm": 0.0016854364657774568, | |
| "learning_rate": 0.0006424523441524988, | |
| "loss": 0.9045, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 1.0757341576506956, | |
| "grad_norm": 0.0020330501720309258, | |
| "learning_rate": 0.0006414219474497683, | |
| "loss": 0.9253, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 1.0788253477588872, | |
| "grad_norm": 0.001624101772904396, | |
| "learning_rate": 0.0006403915507470376, | |
| "loss": 0.9275, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 1.0819165378670788, | |
| "grad_norm": 0.0021750659216195345, | |
| "learning_rate": 0.000639361154044307, | |
| "loss": 0.9483, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.0850077279752706, | |
| "grad_norm": 0.002389618894085288, | |
| "learning_rate": 0.0006383307573415765, | |
| "loss": 0.9259, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 1.0880989180834622, | |
| "grad_norm": 0.0017334523145109415, | |
| "learning_rate": 0.000637300360638846, | |
| "loss": 0.9142, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 1.0911901081916537, | |
| "grad_norm": 0.0015356771182268858, | |
| "learning_rate": 0.0006362699639361155, | |
| "loss": 0.9139, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 1.0942812982998453, | |
| "grad_norm": 0.001495121861808002, | |
| "learning_rate": 0.0006352395672333848, | |
| "loss": 0.9382, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 1.0973724884080371, | |
| "grad_norm": 0.0018960656598210335, | |
| "learning_rate": 0.0006342091705306543, | |
| "loss": 0.9208, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.1004636785162287, | |
| "grad_norm": 0.0019199528032913804, | |
| "learning_rate": 0.0006331787738279238, | |
| "loss": 0.8903, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 1.1035548686244203, | |
| "grad_norm": 0.0016839156160131097, | |
| "learning_rate": 0.0006321483771251933, | |
| "loss": 0.8923, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 1.1066460587326121, | |
| "grad_norm": 0.0016232216730713844, | |
| "learning_rate": 0.0006311179804224627, | |
| "loss": 0.8847, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 1.1097372488408037, | |
| "grad_norm": 0.0016339183785021305, | |
| "learning_rate": 0.000630087583719732, | |
| "loss": 0.9283, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 1.1128284389489953, | |
| "grad_norm": 0.001477651298046112, | |
| "learning_rate": 0.0006290571870170015, | |
| "loss": 0.9211, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.1159196290571871, | |
| "grad_norm": 0.0016854658024385571, | |
| "learning_rate": 0.000628026790314271, | |
| "loss": 0.8857, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 1.1190108191653787, | |
| "grad_norm": 0.001720211817882955, | |
| "learning_rate": 0.0006269963936115405, | |
| "loss": 0.93, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 1.1221020092735703, | |
| "grad_norm": 0.0018675002502277493, | |
| "learning_rate": 0.00062596599690881, | |
| "loss": 0.9222, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 1.125193199381762, | |
| "grad_norm": 0.0014751511625945568, | |
| "learning_rate": 0.0006249356002060794, | |
| "loss": 0.8624, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 1.1282843894899537, | |
| "grad_norm": 0.001356113119982183, | |
| "learning_rate": 0.0006239052035033487, | |
| "loss": 0.92, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.1313755795981453, | |
| "grad_norm": 0.0012932941317558289, | |
| "learning_rate": 0.0006228748068006182, | |
| "loss": 0.9394, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 1.1344667697063369, | |
| "grad_norm": 0.0022874162532389164, | |
| "learning_rate": 0.0006218444100978877, | |
| "loss": 0.8998, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 1.1375579598145287, | |
| "grad_norm": 0.0017121599521487951, | |
| "learning_rate": 0.0006208140133951572, | |
| "loss": 0.9188, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 1.1406491499227203, | |
| "grad_norm": 0.001489990041591227, | |
| "learning_rate": 0.0006197836166924266, | |
| "loss": 0.8829, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 1.1437403400309119, | |
| "grad_norm": 0.0015349757159128785, | |
| "learning_rate": 0.000618753219989696, | |
| "loss": 0.9028, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.1468315301391034, | |
| "grad_norm": 0.0022630542516708374, | |
| "learning_rate": 0.0006177228232869655, | |
| "loss": 0.9536, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 1.1499227202472952, | |
| "grad_norm": 0.0017121895216405392, | |
| "learning_rate": 0.000616692426584235, | |
| "loss": 0.9069, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 1.1530139103554868, | |
| "grad_norm": 0.0019011534750461578, | |
| "learning_rate": 0.0006156620298815044, | |
| "loss": 0.8873, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 1.1561051004636784, | |
| "grad_norm": 0.0014921397669240832, | |
| "learning_rate": 0.0006146316331787738, | |
| "loss": 0.9354, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 1.1591962905718702, | |
| "grad_norm": 0.001888715079985559, | |
| "learning_rate": 0.0006136012364760433, | |
| "loss": 0.9054, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.1622874806800618, | |
| "grad_norm": 0.0026860106736421585, | |
| "learning_rate": 0.0006125708397733127, | |
| "loss": 0.9096, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 1.1653786707882534, | |
| "grad_norm": 0.0014799319906160235, | |
| "learning_rate": 0.0006115404430705822, | |
| "loss": 0.892, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 1.1684698608964452, | |
| "grad_norm": 0.001760624349117279, | |
| "learning_rate": 0.0006105100463678517, | |
| "loss": 0.9615, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 1.1715610510046368, | |
| "grad_norm": 0.0016477296594530344, | |
| "learning_rate": 0.0006094796496651211, | |
| "loss": 0.8632, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 1.1746522411128284, | |
| "grad_norm": 0.0015910883666947484, | |
| "learning_rate": 0.0006084492529623905, | |
| "loss": 0.9434, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.1777434312210202, | |
| "grad_norm": 0.002248365432024002, | |
| "learning_rate": 0.0006074188562596599, | |
| "loss": 0.9087, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 1.1808346213292118, | |
| "grad_norm": 0.0020230677910149097, | |
| "learning_rate": 0.0006063884595569294, | |
| "loss": 0.9356, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 1.1839258114374034, | |
| "grad_norm": 0.0016803268808871508, | |
| "learning_rate": 0.0006053580628541989, | |
| "loss": 0.9394, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 1.187017001545595, | |
| "grad_norm": 0.0012555584544315934, | |
| "learning_rate": 0.0006043276661514683, | |
| "loss": 0.8883, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 1.1901081916537868, | |
| "grad_norm": 0.0015084665501490235, | |
| "learning_rate": 0.0006032972694487378, | |
| "loss": 0.8967, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.1931993817619784, | |
| "grad_norm": 0.001311285886913538, | |
| "learning_rate": 0.0006022668727460073, | |
| "loss": 0.8849, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 1.19629057187017, | |
| "grad_norm": 0.0018462970620021224, | |
| "learning_rate": 0.0006012364760432767, | |
| "loss": 0.8442, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 1.1993817619783615, | |
| "grad_norm": 0.0016767915803939104, | |
| "learning_rate": 0.0006002060793405461, | |
| "loss": 0.9013, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 1.2024729520865534, | |
| "grad_norm": 0.0018409952754154801, | |
| "learning_rate": 0.0005991756826378155, | |
| "loss": 0.9189, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 1.205564142194745, | |
| "grad_norm": 0.0016513046575710177, | |
| "learning_rate": 0.000598145285935085, | |
| "loss": 0.935, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.2086553323029365, | |
| "grad_norm": 0.0017494558123871684, | |
| "learning_rate": 0.0005971148892323545, | |
| "loss": 0.9038, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 1.2117465224111283, | |
| "grad_norm": 0.0021330672316253185, | |
| "learning_rate": 0.0005960844925296239, | |
| "loss": 0.8762, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 1.21483771251932, | |
| "grad_norm": 0.00180672702845186, | |
| "learning_rate": 0.0005950540958268934, | |
| "loss": 0.8729, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 1.2179289026275115, | |
| "grad_norm": 0.001298164832405746, | |
| "learning_rate": 0.0005940236991241628, | |
| "loss": 0.9208, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 1.2210200927357033, | |
| "grad_norm": 0.0016548632411286235, | |
| "learning_rate": 0.0005929933024214322, | |
| "loss": 0.8658, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.224111282843895, | |
| "grad_norm": 0.0019319544080644846, | |
| "learning_rate": 0.0005919629057187017, | |
| "loss": 0.9357, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 1.2272024729520865, | |
| "grad_norm": 0.0016805862542241812, | |
| "learning_rate": 0.0005909325090159712, | |
| "loss": 0.8781, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 1.2302936630602783, | |
| "grad_norm": 0.0023612873628735542, | |
| "learning_rate": 0.0005899021123132406, | |
| "loss": 0.8979, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 1.23338485316847, | |
| "grad_norm": 0.001474004122428596, | |
| "learning_rate": 0.00058887171561051, | |
| "loss": 0.9272, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 1.2364760432766615, | |
| "grad_norm": 0.0016212017508223653, | |
| "learning_rate": 0.0005878413189077795, | |
| "loss": 0.9453, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.239567233384853, | |
| "grad_norm": 0.0017642155289649963, | |
| "learning_rate": 0.000586810922205049, | |
| "loss": 0.9517, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 1.242658423493045, | |
| "grad_norm": 0.0018736496567726135, | |
| "learning_rate": 0.0005857805255023185, | |
| "loss": 0.8862, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 1.2457496136012365, | |
| "grad_norm": 0.001532053924165666, | |
| "learning_rate": 0.0005847501287995878, | |
| "loss": 0.8866, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 1.248840803709428, | |
| "grad_norm": 0.0017100380500778556, | |
| "learning_rate": 0.0005837197320968572, | |
| "loss": 0.9678, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 1.2519319938176197, | |
| "grad_norm": 0.0018053441308438778, | |
| "learning_rate": 0.0005826893353941267, | |
| "loss": 0.9066, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.2550231839258115, | |
| "grad_norm": 0.0015492727980017662, | |
| "learning_rate": 0.0005816589386913962, | |
| "loss": 0.9649, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 1.258114374034003, | |
| "grad_norm": 0.0014449515147134662, | |
| "learning_rate": 0.0005806285419886657, | |
| "loss": 0.8973, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 1.2612055641421946, | |
| "grad_norm": 0.001734506106004119, | |
| "learning_rate": 0.0005795981452859352, | |
| "loss": 0.8801, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 1.2642967542503865, | |
| "grad_norm": 0.0013686501188203692, | |
| "learning_rate": 0.0005785677485832046, | |
| "loss": 0.8936, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 1.267387944358578, | |
| "grad_norm": 0.0016823920886963606, | |
| "learning_rate": 0.0005775373518804739, | |
| "loss": 0.9334, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.2704791344667696, | |
| "grad_norm": 0.0022253175266087055, | |
| "learning_rate": 0.0005765069551777434, | |
| "loss": 0.9328, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 1.2735703245749614, | |
| "grad_norm": 0.0017447506543248892, | |
| "learning_rate": 0.0005754765584750129, | |
| "loss": 0.884, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 1.276661514683153, | |
| "grad_norm": 0.0017223991453647614, | |
| "learning_rate": 0.0005744461617722824, | |
| "loss": 0.8914, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 1.2797527047913446, | |
| "grad_norm": 0.001338414615020156, | |
| "learning_rate": 0.0005734157650695518, | |
| "loss": 0.8588, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 1.2828438948995364, | |
| "grad_norm": 0.0020726649090647697, | |
| "learning_rate": 0.0005723853683668212, | |
| "loss": 0.8827, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.285935085007728, | |
| "grad_norm": 0.0016284299781545997, | |
| "learning_rate": 0.0005713549716640907, | |
| "loss": 0.9044, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 1.2890262751159196, | |
| "grad_norm": 0.0015132429543882608, | |
| "learning_rate": 0.0005703245749613602, | |
| "loss": 0.8997, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 1.2921174652241114, | |
| "grad_norm": 0.0019543899688869715, | |
| "learning_rate": 0.0005692941782586296, | |
| "loss": 0.8779, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 1.295208655332303, | |
| "grad_norm": 0.0016743885353207588, | |
| "learning_rate": 0.000568263781555899, | |
| "loss": 0.9378, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 1.2982998454404946, | |
| "grad_norm": 0.0015272155869752169, | |
| "learning_rate": 0.0005672333848531684, | |
| "loss": 0.9391, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.3013910355486862, | |
| "grad_norm": 0.001885525998659432, | |
| "learning_rate": 0.0005662029881504379, | |
| "loss": 0.9257, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 1.3044822256568778, | |
| "grad_norm": 0.0016695179510861635, | |
| "learning_rate": 0.0005651725914477074, | |
| "loss": 0.9012, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 1.3075734157650696, | |
| "grad_norm": 0.0013361867750063539, | |
| "learning_rate": 0.0005641421947449769, | |
| "loss": 0.8968, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 1.3106646058732612, | |
| "grad_norm": 0.0015216304454952478, | |
| "learning_rate": 0.0005631117980422464, | |
| "loss": 0.9027, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 1.3137557959814528, | |
| "grad_norm": 0.0013232153141871095, | |
| "learning_rate": 0.0005620814013395156, | |
| "loss": 0.8984, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.3168469860896446, | |
| "grad_norm": 0.0019559410866349936, | |
| "learning_rate": 0.0005610510046367851, | |
| "loss": 0.8915, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 1.3199381761978362, | |
| "grad_norm": 0.0014317093882709742, | |
| "learning_rate": 0.0005600206079340546, | |
| "loss": 0.9054, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 1.3230293663060277, | |
| "grad_norm": 0.0013388870283961296, | |
| "learning_rate": 0.0005589902112313241, | |
| "loss": 0.8892, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 1.3261205564142196, | |
| "grad_norm": 0.0015756129287183285, | |
| "learning_rate": 0.0005579598145285936, | |
| "loss": 0.8773, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 1.3292117465224111, | |
| "grad_norm": 0.0014559467090293765, | |
| "learning_rate": 0.0005569294178258629, | |
| "loss": 0.9396, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.3323029366306027, | |
| "grad_norm": 0.0015288847498595715, | |
| "learning_rate": 0.0005558990211231324, | |
| "loss": 0.8731, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 1.3353941267387945, | |
| "grad_norm": 0.0019514070590958, | |
| "learning_rate": 0.0005548686244204019, | |
| "loss": 0.9192, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 1.3384853168469861, | |
| "grad_norm": 0.001415872247889638, | |
| "learning_rate": 0.0005538382277176713, | |
| "loss": 0.8815, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 1.3415765069551777, | |
| "grad_norm": 0.0014958428218960762, | |
| "learning_rate": 0.0005528078310149408, | |
| "loss": 0.9032, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 1.3446676970633695, | |
| "grad_norm": 0.002053620759397745, | |
| "learning_rate": 0.0005517774343122102, | |
| "loss": 0.887, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.3477588871715611, | |
| "grad_norm": 0.0018398250686004758, | |
| "learning_rate": 0.0005507470376094796, | |
| "loss": 0.9225, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 1.3508500772797527, | |
| "grad_norm": 0.0017333675641566515, | |
| "learning_rate": 0.0005497166409067491, | |
| "loss": 0.8951, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 1.3539412673879443, | |
| "grad_norm": 0.002342061372473836, | |
| "learning_rate": 0.0005486862442040186, | |
| "loss": 0.8829, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 1.3570324574961359, | |
| "grad_norm": 0.00188271829392761, | |
| "learning_rate": 0.0005476558475012881, | |
| "loss": 0.8816, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 1.3601236476043277, | |
| "grad_norm": 0.0013321408769115806, | |
| "learning_rate": 0.0005466254507985575, | |
| "loss": 0.9114, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.3632148377125193, | |
| "grad_norm": 0.00140297575853765, | |
| "learning_rate": 0.0005455950540958268, | |
| "loss": 0.9209, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 1.3663060278207109, | |
| "grad_norm": 0.002004598267376423, | |
| "learning_rate": 0.0005445646573930963, | |
| "loss": 0.8764, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 1.3693972179289027, | |
| "grad_norm": 0.0019030956318601966, | |
| "learning_rate": 0.0005435342606903658, | |
| "loss": 0.9532, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 1.3724884080370943, | |
| "grad_norm": 0.002172063337638974, | |
| "learning_rate": 0.0005425038639876353, | |
| "loss": 0.9243, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 1.3755795981452859, | |
| "grad_norm": 0.0018728856230154634, | |
| "learning_rate": 0.0005414734672849047, | |
| "loss": 0.893, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.3786707882534777, | |
| "grad_norm": 0.0015217667678371072, | |
| "learning_rate": 0.0005404430705821742, | |
| "loss": 0.9209, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 1.3817619783616693, | |
| "grad_norm": 0.001939924550242722, | |
| "learning_rate": 0.0005394126738794436, | |
| "loss": 0.9407, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 1.3848531684698608, | |
| "grad_norm": 0.001418776111677289, | |
| "learning_rate": 0.000538382277176713, | |
| "loss": 0.8918, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 1.3879443585780527, | |
| "grad_norm": 0.0015887707704678178, | |
| "learning_rate": 0.0005373518804739825, | |
| "loss": 0.8835, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 1.3910355486862442, | |
| "grad_norm": 0.002404952421784401, | |
| "learning_rate": 0.0005363214837712519, | |
| "loss": 0.9215, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.3941267387944358, | |
| "grad_norm": 0.0013456016313284636, | |
| "learning_rate": 0.0005352910870685214, | |
| "loss": 0.9352, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 1.3972179289026276, | |
| "grad_norm": 0.0014747907407581806, | |
| "learning_rate": 0.0005342606903657908, | |
| "loss": 0.9033, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 1.4003091190108192, | |
| "grad_norm": 0.0016936981119215488, | |
| "learning_rate": 0.0005332302936630603, | |
| "loss": 0.8774, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 1.4034003091190108, | |
| "grad_norm": 0.00203963671810925, | |
| "learning_rate": 0.0005321998969603298, | |
| "loss": 0.941, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 1.4064914992272024, | |
| "grad_norm": 0.0018227493856102228, | |
| "learning_rate": 0.0005311695002575992, | |
| "loss": 0.8724, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.409582689335394, | |
| "grad_norm": 0.002075436757877469, | |
| "learning_rate": 0.0005301391035548686, | |
| "loss": 0.8892, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 1.4126738794435858, | |
| "grad_norm": 0.0016266778111457825, | |
| "learning_rate": 0.000529108706852138, | |
| "loss": 0.8583, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 1.4157650695517774, | |
| "grad_norm": 0.0016663891728967428, | |
| "learning_rate": 0.0005280783101494075, | |
| "loss": 0.9634, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 1.418856259659969, | |
| "grad_norm": 0.0015336048090830445, | |
| "learning_rate": 0.000527047913446677, | |
| "loss": 0.9069, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 1.4219474497681608, | |
| "grad_norm": 0.0020592466462403536, | |
| "learning_rate": 0.0005260175167439464, | |
| "loss": 0.8638, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.4250386398763524, | |
| "grad_norm": 0.0019336834084242582, | |
| "learning_rate": 0.0005249871200412159, | |
| "loss": 0.9, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 1.428129829984544, | |
| "grad_norm": 0.001620001159608364, | |
| "learning_rate": 0.0005239567233384854, | |
| "loss": 0.8783, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 1.4312210200927358, | |
| "grad_norm": 0.0018929082434624434, | |
| "learning_rate": 0.0005229263266357547, | |
| "loss": 0.8888, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 1.4343122102009274, | |
| "grad_norm": 0.00144308025483042, | |
| "learning_rate": 0.0005218959299330242, | |
| "loss": 0.9338, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 1.437403400309119, | |
| "grad_norm": 0.0015054781688377261, | |
| "learning_rate": 0.0005208655332302936, | |
| "loss": 0.8962, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.4404945904173108, | |
| "grad_norm": 0.0017711712280288339, | |
| "learning_rate": 0.0005198351365275631, | |
| "loss": 0.9417, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 1.4435857805255023, | |
| "grad_norm": 0.0019218528177589178, | |
| "learning_rate": 0.0005188047398248326, | |
| "loss": 0.9703, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 1.446676970633694, | |
| "grad_norm": 0.0019779358990490437, | |
| "learning_rate": 0.000517774343122102, | |
| "loss": 0.8804, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 1.4497681607418857, | |
| "grad_norm": 0.001458328333683312, | |
| "learning_rate": 0.0005167439464193715, | |
| "loss": 0.9228, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 1.4528593508500773, | |
| "grad_norm": 0.0017885727575048804, | |
| "learning_rate": 0.000515713549716641, | |
| "loss": 0.9256, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.455950540958269, | |
| "grad_norm": 0.0015944467158988118, | |
| "learning_rate": 0.0005146831530139103, | |
| "loss": 0.8739, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 1.4590417310664605, | |
| "grad_norm": 0.001835488947108388, | |
| "learning_rate": 0.0005136527563111798, | |
| "loss": 0.952, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 1.4621329211746523, | |
| "grad_norm": 0.0025023729540407658, | |
| "learning_rate": 0.0005126223596084493, | |
| "loss": 0.9499, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 1.465224111282844, | |
| "grad_norm": 0.0017449932638555765, | |
| "learning_rate": 0.0005115919629057187, | |
| "loss": 0.9269, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 1.4683153013910355, | |
| "grad_norm": 0.0013545970432460308, | |
| "learning_rate": 0.0005105615662029882, | |
| "loss": 0.8909, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.471406491499227, | |
| "grad_norm": 0.001730005955323577, | |
| "learning_rate": 0.0005095311695002576, | |
| "loss": 0.8975, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 1.474497681607419, | |
| "grad_norm": 0.0017201779410243034, | |
| "learning_rate": 0.0005085007727975271, | |
| "loss": 0.9217, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 1.4775888717156105, | |
| "grad_norm": 0.0020651696249842644, | |
| "learning_rate": 0.0005074703760947965, | |
| "loss": 0.8986, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 1.480680061823802, | |
| "grad_norm": 0.0016624495619907975, | |
| "learning_rate": 0.000506439979392066, | |
| "loss": 0.8983, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 1.4837712519319939, | |
| "grad_norm": 0.0014232158428058028, | |
| "learning_rate": 0.0005054095826893354, | |
| "loss": 0.8733, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.4868624420401855, | |
| "grad_norm": 0.0019593520555645227, | |
| "learning_rate": 0.0005043791859866048, | |
| "loss": 0.9165, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 1.489953632148377, | |
| "grad_norm": 0.002281294437125325, | |
| "learning_rate": 0.0005033487892838743, | |
| "loss": 0.9311, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 1.4930448222565689, | |
| "grad_norm": 0.001690705306828022, | |
| "learning_rate": 0.0005023183925811438, | |
| "loss": 0.9152, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 1.4961360123647605, | |
| "grad_norm": 0.0024157485458999872, | |
| "learning_rate": 0.0005012879958784133, | |
| "loss": 0.9276, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 1.499227202472952, | |
| "grad_norm": 0.0015235710889101028, | |
| "learning_rate": 0.0005002575991756827, | |
| "loss": 0.8763, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.5023183925811439, | |
| "grad_norm": 0.0018749197479337454, | |
| "learning_rate": 0.0004992272024729521, | |
| "loss": 0.9412, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 1.5054095826893354, | |
| "grad_norm": 0.0020271523389965296, | |
| "learning_rate": 0.0004981968057702215, | |
| "loss": 0.8985, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 1.508500772797527, | |
| "grad_norm": 0.0015005801105871797, | |
| "learning_rate": 0.000497166409067491, | |
| "loss": 0.9426, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 1.5115919629057188, | |
| "grad_norm": 0.002262561582028866, | |
| "learning_rate": 0.0004961360123647605, | |
| "loss": 0.9107, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 1.5146831530139102, | |
| "grad_norm": 0.001319503178820014, | |
| "learning_rate": 0.0004951056156620298, | |
| "loss": 0.9054, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.517774343122102, | |
| "grad_norm": 0.0014885893324390054, | |
| "learning_rate": 0.0004940752189592993, | |
| "loss": 0.9265, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 1.5208655332302936, | |
| "grad_norm": 0.0017433296889066696, | |
| "learning_rate": 0.0004930448222565688, | |
| "loss": 0.8997, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 1.5239567233384852, | |
| "grad_norm": 0.0013538316125050187, | |
| "learning_rate": 0.0004920144255538382, | |
| "loss": 0.8618, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 1.527047913446677, | |
| "grad_norm": 0.0014708703383803368, | |
| "learning_rate": 0.0004909840288511077, | |
| "loss": 0.9203, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 1.5301391035548686, | |
| "grad_norm": 0.0017004169058054686, | |
| "learning_rate": 0.0004899536321483772, | |
| "loss": 0.8936, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.5332302936630602, | |
| "grad_norm": 0.0017624979373067617, | |
| "learning_rate": 0.0004889232354456466, | |
| "loss": 0.8778, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 1.536321483771252, | |
| "grad_norm": 0.0015045328764244914, | |
| "learning_rate": 0.000487892838742916, | |
| "loss": 0.8828, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 1.5394126738794436, | |
| "grad_norm": 0.0018641521455720067, | |
| "learning_rate": 0.0004868624420401855, | |
| "loss": 0.9158, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 1.5425038639876352, | |
| "grad_norm": 0.001571865752339363, | |
| "learning_rate": 0.00048583204533745493, | |
| "loss": 0.8962, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 1.545595054095827, | |
| "grad_norm": 0.0016725645400583744, | |
| "learning_rate": 0.00048480164863472436, | |
| "loss": 0.9001, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.5486862442040186, | |
| "grad_norm": 0.001643617171794176, | |
| "learning_rate": 0.00048377125193199385, | |
| "loss": 0.9103, | |
| "step": 5010 | |
| }, | |
| { | |
| "epoch": 1.5517774343122102, | |
| "grad_norm": 0.002170222345739603, | |
| "learning_rate": 0.0004827408552292633, | |
| "loss": 0.9045, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 1.554868624420402, | |
| "grad_norm": 0.0015412492211908102, | |
| "learning_rate": 0.0004817104585265327, | |
| "loss": 0.8995, | |
| "step": 5030 | |
| }, | |
| { | |
| "epoch": 1.5579598145285936, | |
| "grad_norm": 0.0019475616281852126, | |
| "learning_rate": 0.0004806800618238022, | |
| "loss": 0.8543, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 1.5610510046367851, | |
| "grad_norm": 0.0014466085704043508, | |
| "learning_rate": 0.0004796496651210716, | |
| "loss": 0.899, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.564142194744977, | |
| "grad_norm": 0.0015615527518093586, | |
| "learning_rate": 0.00047861926841834105, | |
| "loss": 0.9303, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 1.5672333848531683, | |
| "grad_norm": 0.0016237753443419933, | |
| "learning_rate": 0.0004775888717156105, | |
| "loss": 0.914, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 1.5703245749613601, | |
| "grad_norm": 0.0015945249469950795, | |
| "learning_rate": 0.00047655847501287997, | |
| "loss": 0.899, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 1.573415765069552, | |
| "grad_norm": 0.001989311771467328, | |
| "learning_rate": 0.00047552807831014945, | |
| "loss": 0.9609, | |
| "step": 5090 | |
| }, | |
| { | |
| "epoch": 1.5765069551777433, | |
| "grad_norm": 0.0025777083355933428, | |
| "learning_rate": 0.00047449768160741883, | |
| "loss": 0.9187, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.5795981452859351, | |
| "grad_norm": 0.0016967840492725372, | |
| "learning_rate": 0.0004734672849046883, | |
| "loss": 0.9198, | |
| "step": 5110 | |
| }, | |
| { | |
| "epoch": 1.5826893353941267, | |
| "grad_norm": 0.0015623560175299644, | |
| "learning_rate": 0.00047243688820195774, | |
| "loss": 0.9066, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 1.5857805255023183, | |
| "grad_norm": 0.0014336062595248222, | |
| "learning_rate": 0.00047140649149922723, | |
| "loss": 0.8992, | |
| "step": 5130 | |
| }, | |
| { | |
| "epoch": 1.58887171561051, | |
| "grad_norm": 0.0018111519748345017, | |
| "learning_rate": 0.00047037609479649666, | |
| "loss": 0.9826, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 1.5919629057187017, | |
| "grad_norm": 0.0016681707929819822, | |
| "learning_rate": 0.0004693456980937661, | |
| "loss": 0.9236, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 1.5950540958268933, | |
| "grad_norm": 0.0015410635387524962, | |
| "learning_rate": 0.0004683153013910356, | |
| "loss": 0.9151, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 1.598145285935085, | |
| "grad_norm": 0.0017971232300624251, | |
| "learning_rate": 0.00046728490468830506, | |
| "loss": 0.9007, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 1.6012364760432767, | |
| "grad_norm": 0.0019288517069071531, | |
| "learning_rate": 0.00046625450798557443, | |
| "loss": 0.9123, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 1.6043276661514683, | |
| "grad_norm": 0.0013020862825214863, | |
| "learning_rate": 0.0004652241112828439, | |
| "loss": 0.8848, | |
| "step": 5190 | |
| }, | |
| { | |
| "epoch": 1.60741885625966, | |
| "grad_norm": 0.0015427186153829098, | |
| "learning_rate": 0.00046419371458011335, | |
| "loss": 0.948, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.6105100463678517, | |
| "grad_norm": 0.0016680203843861818, | |
| "learning_rate": 0.0004631633178773828, | |
| "loss": 0.9122, | |
| "step": 5210 | |
| }, | |
| { | |
| "epoch": 1.6136012364760433, | |
| "grad_norm": 0.0014202597085386515, | |
| "learning_rate": 0.00046213292117465226, | |
| "loss": 0.8974, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 1.616692426584235, | |
| "grad_norm": 0.0018021473661065102, | |
| "learning_rate": 0.0004611025244719217, | |
| "loss": 0.8518, | |
| "step": 5230 | |
| }, | |
| { | |
| "epoch": 1.6197836166924264, | |
| "grad_norm": 0.001819357625208795, | |
| "learning_rate": 0.0004600721277691912, | |
| "loss": 0.9143, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 1.6228748068006182, | |
| "grad_norm": 0.0018893377855420113, | |
| "learning_rate": 0.00045904173106646055, | |
| "loss": 0.8866, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.62596599690881, | |
| "grad_norm": 0.0018815461080521345, | |
| "learning_rate": 0.00045801133436373004, | |
| "loss": 0.9116, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 1.6290571870170014, | |
| "grad_norm": 0.0018397814128547907, | |
| "learning_rate": 0.0004569809376609995, | |
| "loss": 0.9021, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 1.6321483771251932, | |
| "grad_norm": 0.002361022401601076, | |
| "learning_rate": 0.00045595054095826895, | |
| "loss": 0.9044, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 1.6352395672333848, | |
| "grad_norm": 0.0016238827956840396, | |
| "learning_rate": 0.0004549201442555384, | |
| "loss": 0.9038, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 1.6383307573415764, | |
| "grad_norm": 0.0020596208050847054, | |
| "learning_rate": 0.0004538897475528078, | |
| "loss": 0.8645, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.6414219474497682, | |
| "grad_norm": 0.0016590558225288987, | |
| "learning_rate": 0.0004528593508500773, | |
| "loss": 0.8853, | |
| "step": 5310 | |
| }, | |
| { | |
| "epoch": 1.6445131375579598, | |
| "grad_norm": 0.0019768117927014828, | |
| "learning_rate": 0.0004518289541473468, | |
| "loss": 0.8622, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 1.6476043276661514, | |
| "grad_norm": 0.0016761173028498888, | |
| "learning_rate": 0.00045079855744461616, | |
| "loss": 0.9272, | |
| "step": 5330 | |
| }, | |
| { | |
| "epoch": 1.6506955177743432, | |
| "grad_norm": 0.0013793542748317122, | |
| "learning_rate": 0.00044976816074188564, | |
| "loss": 0.935, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 1.6537867078825348, | |
| "grad_norm": 0.001541083212941885, | |
| "learning_rate": 0.0004487377640391551, | |
| "loss": 0.8854, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 1.6568778979907264, | |
| "grad_norm": 0.0018007430480793118, | |
| "learning_rate": 0.00044770736733642456, | |
| "loss": 0.8571, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 1.6599690880989182, | |
| "grad_norm": 0.0015837026294320822, | |
| "learning_rate": 0.000446676970633694, | |
| "loss": 0.8731, | |
| "step": 5370 | |
| }, | |
| { | |
| "epoch": 1.6630602782071098, | |
| "grad_norm": 0.0019535787869244814, | |
| "learning_rate": 0.0004456465739309634, | |
| "loss": 0.8952, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 1.6661514683153014, | |
| "grad_norm": 0.0015085084596648812, | |
| "learning_rate": 0.0004446161772282329, | |
| "loss": 0.9039, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 1.6692426584234932, | |
| "grad_norm": 0.0016192490002140403, | |
| "learning_rate": 0.0004435857805255023, | |
| "loss": 0.8972, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.6723338485316845, | |
| "grad_norm": 0.0025146189145743847, | |
| "learning_rate": 0.00044255538382277176, | |
| "loss": 0.909, | |
| "step": 5410 | |
| }, | |
| { | |
| "epoch": 1.6754250386398764, | |
| "grad_norm": 0.001912578009068966, | |
| "learning_rate": 0.00044152498712004125, | |
| "loss": 0.909, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 1.6785162287480682, | |
| "grad_norm": 0.0015931734815239906, | |
| "learning_rate": 0.0004404945904173107, | |
| "loss": 0.9285, | |
| "step": 5430 | |
| }, | |
| { | |
| "epoch": 1.6816074188562595, | |
| "grad_norm": 0.0015723604010418057, | |
| "learning_rate": 0.0004394641937145801, | |
| "loss": 0.883, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 1.6846986089644513, | |
| "grad_norm": 0.0017312741838395596, | |
| "learning_rate": 0.00043843379701184954, | |
| "loss": 0.8687, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.687789799072643, | |
| "grad_norm": 0.0015717818168923259, | |
| "learning_rate": 0.000437403400309119, | |
| "loss": 0.8634, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 1.6908809891808345, | |
| "grad_norm": 0.0017777059692889452, | |
| "learning_rate": 0.0004363730036063885, | |
| "loss": 0.9342, | |
| "step": 5470 | |
| }, | |
| { | |
| "epoch": 1.6939721792890263, | |
| "grad_norm": 0.0014749905094504356, | |
| "learning_rate": 0.0004353426069036579, | |
| "loss": 0.8881, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 1.697063369397218, | |
| "grad_norm": 0.0015921080484986305, | |
| "learning_rate": 0.00043431221020092737, | |
| "loss": 0.8682, | |
| "step": 5490 | |
| }, | |
| { | |
| "epoch": 1.7001545595054095, | |
| "grad_norm": 0.0017204548930749297, | |
| "learning_rate": 0.00043328181349819685, | |
| "loss": 0.9315, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.7032457496136013, | |
| "grad_norm": 0.0013450038386508822, | |
| "learning_rate": 0.0004322514167954663, | |
| "loss": 0.8749, | |
| "step": 5510 | |
| }, | |
| { | |
| "epoch": 1.706336939721793, | |
| "grad_norm": 0.0020300759933888912, | |
| "learning_rate": 0.0004312210200927357, | |
| "loss": 0.9331, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 1.7094281298299845, | |
| "grad_norm": 0.0023906801361590624, | |
| "learning_rate": 0.00043019062339000514, | |
| "loss": 0.9049, | |
| "step": 5530 | |
| }, | |
| { | |
| "epoch": 1.7125193199381763, | |
| "grad_norm": 0.0016755072865635157, | |
| "learning_rate": 0.00042916022668727463, | |
| "loss": 0.8818, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 1.7156105100463679, | |
| "grad_norm": 0.001380381640046835, | |
| "learning_rate": 0.00042812982998454406, | |
| "loss": 0.9016, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.7187017001545595, | |
| "grad_norm": 0.0016399535816162825, | |
| "learning_rate": 0.0004270994332818135, | |
| "loss": 0.9137, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 1.7217928902627513, | |
| "grad_norm": 0.0018158459570258856, | |
| "learning_rate": 0.000426069036579083, | |
| "loss": 0.8935, | |
| "step": 5570 | |
| }, | |
| { | |
| "epoch": 1.7248840803709427, | |
| "grad_norm": 0.0017615389078855515, | |
| "learning_rate": 0.0004250386398763524, | |
| "loss": 0.8964, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 1.7279752704791345, | |
| "grad_norm": 0.0018352493643760681, | |
| "learning_rate": 0.00042400824317362183, | |
| "loss": 0.9344, | |
| "step": 5590 | |
| }, | |
| { | |
| "epoch": 1.7310664605873263, | |
| "grad_norm": 0.0015487250639125705, | |
| "learning_rate": 0.0004229778464708913, | |
| "loss": 0.888, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.7341576506955176, | |
| "grad_norm": 0.00184920453466475, | |
| "learning_rate": 0.00042194744976816075, | |
| "loss": 0.866, | |
| "step": 5610 | |
| }, | |
| { | |
| "epoch": 1.7372488408037094, | |
| "grad_norm": 0.0018842272693291306, | |
| "learning_rate": 0.00042091705306543023, | |
| "loss": 0.9157, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 1.740340030911901, | |
| "grad_norm": 0.001251103589311242, | |
| "learning_rate": 0.0004198866563626996, | |
| "loss": 0.8774, | |
| "step": 5630 | |
| }, | |
| { | |
| "epoch": 1.7434312210200926, | |
| "grad_norm": 0.0017979164840653539, | |
| "learning_rate": 0.0004188562596599691, | |
| "loss": 0.9327, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 1.7465224111282844, | |
| "grad_norm": 0.003756187856197357, | |
| "learning_rate": 0.0004178258629572386, | |
| "loss": 0.898, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.749613601236476, | |
| "grad_norm": 0.0025360011495649815, | |
| "learning_rate": 0.000416795466254508, | |
| "loss": 0.9021, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 1.7527047913446676, | |
| "grad_norm": 0.0015577342128381133, | |
| "learning_rate": 0.00041576506955177744, | |
| "loss": 0.8972, | |
| "step": 5670 | |
| }, | |
| { | |
| "epoch": 1.7557959814528594, | |
| "grad_norm": 0.002707903040573001, | |
| "learning_rate": 0.00041473467284904687, | |
| "loss": 0.8647, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 1.758887171561051, | |
| "grad_norm": 0.0017573883524164557, | |
| "learning_rate": 0.00041370427614631635, | |
| "loss": 0.9007, | |
| "step": 5690 | |
| }, | |
| { | |
| "epoch": 1.7619783616692426, | |
| "grad_norm": 0.002097573596984148, | |
| "learning_rate": 0.0004126738794435858, | |
| "loss": 0.865, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.7650695517774344, | |
| "grad_norm": 0.0018730917945504189, | |
| "learning_rate": 0.0004116434827408552, | |
| "loss": 0.9073, | |
| "step": 5710 | |
| }, | |
| { | |
| "epoch": 1.768160741885626, | |
| "grad_norm": 0.0017573771765455604, | |
| "learning_rate": 0.0004106130860381247, | |
| "loss": 0.9246, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 1.7712519319938176, | |
| "grad_norm": 0.0017129804473370314, | |
| "learning_rate": 0.00040958268933539413, | |
| "loss": 0.93, | |
| "step": 5730 | |
| }, | |
| { | |
| "epoch": 1.7743431221020094, | |
| "grad_norm": 0.0012526778737083077, | |
| "learning_rate": 0.00040855229263266356, | |
| "loss": 0.8911, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 1.7774343122102008, | |
| "grad_norm": 0.002290197880938649, | |
| "learning_rate": 0.00040752189592993304, | |
| "loss": 0.9049, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.7805255023183926, | |
| "grad_norm": 0.001954218838363886, | |
| "learning_rate": 0.0004064914992272025, | |
| "loss": 0.8975, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 1.7836166924265844, | |
| "grad_norm": 0.002614745870232582, | |
| "learning_rate": 0.00040546110252447196, | |
| "loss": 0.8844, | |
| "step": 5770 | |
| }, | |
| { | |
| "epoch": 1.7867078825347757, | |
| "grad_norm": 0.0014066528528928757, | |
| "learning_rate": 0.00040443070582174133, | |
| "loss": 0.8726, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 1.7897990726429676, | |
| "grad_norm": 0.0013754137326031923, | |
| "learning_rate": 0.0004034003091190108, | |
| "loss": 0.9034, | |
| "step": 5790 | |
| }, | |
| { | |
| "epoch": 1.7928902627511591, | |
| "grad_norm": 0.0014499702956527472, | |
| "learning_rate": 0.0004023699124162803, | |
| "loss": 0.8515, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.7959814528593507, | |
| "grad_norm": 0.0015088847139850259, | |
| "learning_rate": 0.00040133951571354973, | |
| "loss": 0.9027, | |
| "step": 5810 | |
| }, | |
| { | |
| "epoch": 1.7990726429675425, | |
| "grad_norm": 0.0013973376480862498, | |
| "learning_rate": 0.00040030911901081916, | |
| "loss": 0.8943, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 1.8021638330757341, | |
| "grad_norm": 0.0014548856997862458, | |
| "learning_rate": 0.00039927872230808865, | |
| "loss": 0.9106, | |
| "step": 5830 | |
| }, | |
| { | |
| "epoch": 1.8052550231839257, | |
| "grad_norm": 0.0017355557065457106, | |
| "learning_rate": 0.0003982483256053581, | |
| "loss": 0.8727, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 1.8083462132921175, | |
| "grad_norm": 0.0021262154914438725, | |
| "learning_rate": 0.0003972179289026275, | |
| "loss": 0.8824, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 1.8114374034003091, | |
| "grad_norm": 0.001427137991413474, | |
| "learning_rate": 0.00039618753219989694, | |
| "loss": 0.9374, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 1.8145285935085007, | |
| "grad_norm": 0.0016721707070246339, | |
| "learning_rate": 0.0003951571354971664, | |
| "loss": 0.919, | |
| "step": 5870 | |
| }, | |
| { | |
| "epoch": 1.8176197836166925, | |
| "grad_norm": 0.0017290489049628377, | |
| "learning_rate": 0.0003941267387944359, | |
| "loss": 0.9544, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 1.820710973724884, | |
| "grad_norm": 0.001801205798983574, | |
| "learning_rate": 0.0003930963420917053, | |
| "loss": 0.8656, | |
| "step": 5890 | |
| }, | |
| { | |
| "epoch": 1.8238021638330757, | |
| "grad_norm": 0.0016462091589346528, | |
| "learning_rate": 0.00039206594538897477, | |
| "loss": 0.8515, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.8268933539412675, | |
| "grad_norm": 0.001921969000250101, | |
| "learning_rate": 0.0003910355486862442, | |
| "loss": 0.8472, | |
| "step": 5910 | |
| }, | |
| { | |
| "epoch": 1.829984544049459, | |
| "grad_norm": 0.0016511150170117617, | |
| "learning_rate": 0.0003900051519835137, | |
| "loss": 0.8858, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 1.8330757341576507, | |
| "grad_norm": 0.0014706592774018645, | |
| "learning_rate": 0.0003889747552807831, | |
| "loss": 0.8983, | |
| "step": 5930 | |
| }, | |
| { | |
| "epoch": 1.8361669242658425, | |
| "grad_norm": 0.0024984250776469707, | |
| "learning_rate": 0.00038794435857805254, | |
| "loss": 0.9051, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 1.8392581143740339, | |
| "grad_norm": 0.001705004251562059, | |
| "learning_rate": 0.00038691396187532203, | |
| "loss": 0.902, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 1.8423493044822257, | |
| "grad_norm": 0.0019023518543690443, | |
| "learning_rate": 0.00038588356517259146, | |
| "loss": 0.8798, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 1.8454404945904173, | |
| "grad_norm": 0.0018084270413964987, | |
| "learning_rate": 0.0003848531684698609, | |
| "loss": 0.9092, | |
| "step": 5970 | |
| }, | |
| { | |
| "epoch": 1.8485316846986088, | |
| "grad_norm": 0.002029530005529523, | |
| "learning_rate": 0.00038382277176713037, | |
| "loss": 0.9116, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 1.8516228748068007, | |
| "grad_norm": 0.0018030694918707013, | |
| "learning_rate": 0.0003827923750643998, | |
| "loss": 0.9126, | |
| "step": 5990 | |
| }, | |
| { | |
| "epoch": 1.8547140649149922, | |
| "grad_norm": 0.0018464057939127088, | |
| "learning_rate": 0.00038176197836166923, | |
| "loss": 0.8679, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.8578052550231838, | |
| "grad_norm": 0.0018209113040938973, | |
| "learning_rate": 0.00038073158165893866, | |
| "loss": 0.9302, | |
| "step": 6010 | |
| }, | |
| { | |
| "epoch": 1.8608964451313756, | |
| "grad_norm": 0.001530204783193767, | |
| "learning_rate": 0.00037970118495620815, | |
| "loss": 0.8936, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 1.8639876352395672, | |
| "grad_norm": 0.0017929489258676767, | |
| "learning_rate": 0.00037867078825347763, | |
| "loss": 0.8799, | |
| "step": 6030 | |
| }, | |
| { | |
| "epoch": 1.8670788253477588, | |
| "grad_norm": 0.0020685286726802588, | |
| "learning_rate": 0.000377640391550747, | |
| "loss": 0.9022, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 1.8701700154559506, | |
| "grad_norm": 0.0016937406035140157, | |
| "learning_rate": 0.0003766099948480165, | |
| "loss": 0.922, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 1.8732612055641422, | |
| "grad_norm": 0.001672919373959303, | |
| "learning_rate": 0.0003755795981452859, | |
| "loss": 0.8984, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 1.8763523956723338, | |
| "grad_norm": 0.0020905956625938416, | |
| "learning_rate": 0.0003745492014425554, | |
| "loss": 0.9168, | |
| "step": 6070 | |
| }, | |
| { | |
| "epoch": 1.8794435857805256, | |
| "grad_norm": 0.0016705166781321168, | |
| "learning_rate": 0.00037351880473982484, | |
| "loss": 0.9095, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 1.8825347758887172, | |
| "grad_norm": 0.002511728322133422, | |
| "learning_rate": 0.00037248840803709427, | |
| "loss": 0.9591, | |
| "step": 6090 | |
| }, | |
| { | |
| "epoch": 1.8856259659969088, | |
| "grad_norm": 0.0016106483526527882, | |
| "learning_rate": 0.00037145801133436375, | |
| "loss": 0.9133, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.8887171561051006, | |
| "grad_norm": 0.0018942320020869374, | |
| "learning_rate": 0.00037042761463163324, | |
| "loss": 0.9216, | |
| "step": 6110 | |
| }, | |
| { | |
| "epoch": 1.891808346213292, | |
| "grad_norm": 0.0014648385113105178, | |
| "learning_rate": 0.0003693972179289026, | |
| "loss": 0.854, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 1.8948995363214838, | |
| "grad_norm": 0.0018195216543972492, | |
| "learning_rate": 0.0003683668212261721, | |
| "loss": 0.9546, | |
| "step": 6130 | |
| }, | |
| { | |
| "epoch": 1.8979907264296756, | |
| "grad_norm": 0.0016678489046171308, | |
| "learning_rate": 0.00036733642452344153, | |
| "loss": 0.882, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 1.901081916537867, | |
| "grad_norm": 0.0018015914829447865, | |
| "learning_rate": 0.000366306027820711, | |
| "loss": 0.9164, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 1.9041731066460588, | |
| "grad_norm": 0.0018244273960590363, | |
| "learning_rate": 0.00036527563111798044, | |
| "loss": 0.9182, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 1.9072642967542504, | |
| "grad_norm": 0.002539639361202717, | |
| "learning_rate": 0.00036424523441524987, | |
| "loss": 0.8878, | |
| "step": 6170 | |
| }, | |
| { | |
| "epoch": 1.910355486862442, | |
| "grad_norm": 0.0017704921774566174, | |
| "learning_rate": 0.00036321483771251936, | |
| "loss": 0.8888, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 1.9134466769706338, | |
| "grad_norm": 0.0023106811568140984, | |
| "learning_rate": 0.00036218444100978873, | |
| "loss": 0.9106, | |
| "step": 6190 | |
| }, | |
| { | |
| "epoch": 1.9165378670788253, | |
| "grad_norm": 0.0019237243104726076, | |
| "learning_rate": 0.0003611540443070582, | |
| "loss": 0.8905, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.919629057187017, | |
| "grad_norm": 0.0014340688940137625, | |
| "learning_rate": 0.0003601236476043277, | |
| "loss": 0.8845, | |
| "step": 6210 | |
| }, | |
| { | |
| "epoch": 1.9227202472952087, | |
| "grad_norm": 0.0015864827437326312, | |
| "learning_rate": 0.00035909325090159713, | |
| "loss": 0.8996, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 1.9258114374034003, | |
| "grad_norm": 0.0015774049097672105, | |
| "learning_rate": 0.00035806285419886656, | |
| "loss": 0.9557, | |
| "step": 6230 | |
| }, | |
| { | |
| "epoch": 1.928902627511592, | |
| "grad_norm": 0.0022414042614400387, | |
| "learning_rate": 0.000357032457496136, | |
| "loss": 0.8609, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 1.9319938176197837, | |
| "grad_norm": 0.001590002211742103, | |
| "learning_rate": 0.0003560020607934055, | |
| "loss": 0.9381, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 1.9350850077279753, | |
| "grad_norm": 0.002488743746653199, | |
| "learning_rate": 0.00035497166409067496, | |
| "loss": 0.9132, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 1.938176197836167, | |
| "grad_norm": 0.0019566200207918882, | |
| "learning_rate": 0.00035394126738794434, | |
| "loss": 0.9023, | |
| "step": 6270 | |
| }, | |
| { | |
| "epoch": 1.9412673879443587, | |
| "grad_norm": 0.0015884449239820242, | |
| "learning_rate": 0.0003529108706852138, | |
| "loss": 0.9132, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 1.94435857805255, | |
| "grad_norm": 0.002089353743940592, | |
| "learning_rate": 0.00035188047398248325, | |
| "loss": 0.8855, | |
| "step": 6290 | |
| }, | |
| { | |
| "epoch": 1.947449768160742, | |
| "grad_norm": 0.0014843277167528868, | |
| "learning_rate": 0.00035085007727975274, | |
| "loss": 0.898, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.9505409582689337, | |
| "grad_norm": 0.0015669453423470259, | |
| "learning_rate": 0.00034981968057702217, | |
| "loss": 0.9371, | |
| "step": 6310 | |
| }, | |
| { | |
| "epoch": 1.953632148377125, | |
| "grad_norm": 0.0016016702866181731, | |
| "learning_rate": 0.0003487892838742916, | |
| "loss": 0.9119, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 1.9567233384853169, | |
| "grad_norm": 0.0017695052083581686, | |
| "learning_rate": 0.0003477588871715611, | |
| "loss": 0.8987, | |
| "step": 6330 | |
| }, | |
| { | |
| "epoch": 1.9598145285935085, | |
| "grad_norm": 0.00243277451954782, | |
| "learning_rate": 0.00034672849046883046, | |
| "loss": 0.9257, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 1.9629057187017, | |
| "grad_norm": 0.0014211182715371251, | |
| "learning_rate": 0.00034569809376609994, | |
| "loss": 0.8801, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 1.9659969088098919, | |
| "grad_norm": 0.0024740241933614016, | |
| "learning_rate": 0.0003446676970633694, | |
| "loss": 0.8899, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 1.9690880989180835, | |
| "grad_norm": 0.001807063934393227, | |
| "learning_rate": 0.00034363730036063886, | |
| "loss": 0.8971, | |
| "step": 6370 | |
| }, | |
| { | |
| "epoch": 1.972179289026275, | |
| "grad_norm": 0.0013645980507135391, | |
| "learning_rate": 0.0003426069036579083, | |
| "loss": 0.9323, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 1.9752704791344669, | |
| "grad_norm": 0.0015155840665102005, | |
| "learning_rate": 0.0003415765069551777, | |
| "loss": 0.8874, | |
| "step": 6390 | |
| }, | |
| { | |
| "epoch": 1.9783616692426584, | |
| "grad_norm": 0.0016512033762410283, | |
| "learning_rate": 0.0003405461102524472, | |
| "loss": 0.8489, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.98145285935085, | |
| "grad_norm": 0.001505164080299437, | |
| "learning_rate": 0.0003395157135497167, | |
| "loss": 0.9208, | |
| "step": 6410 | |
| }, | |
| { | |
| "epoch": 1.9845440494590418, | |
| "grad_norm": 0.0018190627451986074, | |
| "learning_rate": 0.00033848531684698606, | |
| "loss": 0.9155, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 1.9876352395672334, | |
| "grad_norm": 0.0019098619231954217, | |
| "learning_rate": 0.00033745492014425555, | |
| "loss": 0.913, | |
| "step": 6430 | |
| }, | |
| { | |
| "epoch": 1.990726429675425, | |
| "grad_norm": 0.0015993445413187146, | |
| "learning_rate": 0.00033642452344152503, | |
| "loss": 0.8934, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 1.9938176197836168, | |
| "grad_norm": 0.0017613953677937388, | |
| "learning_rate": 0.00033539412673879446, | |
| "loss": 0.8946, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 1.9969088098918082, | |
| "grad_norm": 0.0015573868295177817, | |
| "learning_rate": 0.0003343637300360639, | |
| "loss": 0.9159, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.0019827033393085003, | |
| "learning_rate": 0.0003333333333333333, | |
| "loss": 0.9071, | |
| "step": 6470 | |
| }, | |
| { | |
| "epoch": 2.003091190108192, | |
| "grad_norm": 0.001513644470833242, | |
| "learning_rate": 0.0003323029366306028, | |
| "loss": 0.9439, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 2.006182380216383, | |
| "grad_norm": 0.001435752958059311, | |
| "learning_rate": 0.00033127253992787224, | |
| "loss": 0.9027, | |
| "step": 6490 | |
| }, | |
| { | |
| "epoch": 2.009273570324575, | |
| "grad_norm": 0.0016444657230749726, | |
| "learning_rate": 0.00033024214322514167, | |
| "loss": 0.8816, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.012364760432767, | |
| "grad_norm": 0.0021295547485351562, | |
| "learning_rate": 0.00032921174652241115, | |
| "loss": 0.9254, | |
| "step": 6510 | |
| }, | |
| { | |
| "epoch": 2.015455950540958, | |
| "grad_norm": 0.001806983258575201, | |
| "learning_rate": 0.0003281813498196806, | |
| "loss": 0.9273, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 2.01854714064915, | |
| "grad_norm": 0.0016168680740520358, | |
| "learning_rate": 0.00032715095311695, | |
| "loss": 0.9036, | |
| "step": 6530 | |
| }, | |
| { | |
| "epoch": 2.021638330757342, | |
| "grad_norm": 0.0014968597097322345, | |
| "learning_rate": 0.0003261205564142195, | |
| "loss": 0.8532, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 2.024729520865533, | |
| "grad_norm": 0.0019258251413702965, | |
| "learning_rate": 0.0003250901597114889, | |
| "loss": 0.9347, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 2.027820710973725, | |
| "grad_norm": 0.0017398808849975467, | |
| "learning_rate": 0.0003240597630087584, | |
| "loss": 0.8724, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 2.0309119010819163, | |
| "grad_norm": 0.001705456175841391, | |
| "learning_rate": 0.0003230293663060278, | |
| "loss": 0.9058, | |
| "step": 6570 | |
| }, | |
| { | |
| "epoch": 2.034003091190108, | |
| "grad_norm": 0.0018997775623574853, | |
| "learning_rate": 0.00032199896960329727, | |
| "loss": 0.9097, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 2.0370942812983, | |
| "grad_norm": 0.001412282115779817, | |
| "learning_rate": 0.00032096857290056676, | |
| "loss": 0.8857, | |
| "step": 6590 | |
| }, | |
| { | |
| "epoch": 2.0401854714064913, | |
| "grad_norm": 0.0014306355733424425, | |
| "learning_rate": 0.0003199381761978362, | |
| "loss": 0.91, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.043276661514683, | |
| "grad_norm": 0.001689639175310731, | |
| "learning_rate": 0.0003189077794951056, | |
| "loss": 0.9038, | |
| "step": 6610 | |
| }, | |
| { | |
| "epoch": 2.046367851622875, | |
| "grad_norm": 0.0015731025487184525, | |
| "learning_rate": 0.00031787738279237505, | |
| "loss": 0.913, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 2.0494590417310663, | |
| "grad_norm": 0.0014561581192538142, | |
| "learning_rate": 0.00031684698608964453, | |
| "loss": 0.8767, | |
| "step": 6630 | |
| }, | |
| { | |
| "epoch": 2.052550231839258, | |
| "grad_norm": 0.0017503297422081232, | |
| "learning_rate": 0.00031581658938691396, | |
| "loss": 0.917, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 2.05564142194745, | |
| "grad_norm": 0.00222258223220706, | |
| "learning_rate": 0.0003147861926841834, | |
| "loss": 0.8869, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 2.0587326120556413, | |
| "grad_norm": 0.0023604268208146095, | |
| "learning_rate": 0.0003137557959814529, | |
| "loss": 0.9149, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 2.061823802163833, | |
| "grad_norm": 0.0023476933129131794, | |
| "learning_rate": 0.0003127253992787223, | |
| "loss": 0.949, | |
| "step": 6670 | |
| }, | |
| { | |
| "epoch": 2.064914992272025, | |
| "grad_norm": 0.0021014835219830275, | |
| "learning_rate": 0.00031169500257599174, | |
| "loss": 0.9374, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 2.0680061823802163, | |
| "grad_norm": 0.001549664419144392, | |
| "learning_rate": 0.0003106646058732612, | |
| "loss": 0.9189, | |
| "step": 6690 | |
| }, | |
| { | |
| "epoch": 2.071097372488408, | |
| "grad_norm": 0.0018286675913259387, | |
| "learning_rate": 0.00030963420917053065, | |
| "loss": 0.9028, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.0741885625966, | |
| "grad_norm": 0.0018017146503552794, | |
| "learning_rate": 0.00030860381246780014, | |
| "loss": 0.8899, | |
| "step": 6710 | |
| }, | |
| { | |
| "epoch": 2.0772797527047913, | |
| "grad_norm": 0.001875316142104566, | |
| "learning_rate": 0.0003075734157650695, | |
| "loss": 0.9123, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 2.080370942812983, | |
| "grad_norm": 0.0018925000913441181, | |
| "learning_rate": 0.000306543019062339, | |
| "loss": 0.9139, | |
| "step": 6730 | |
| }, | |
| { | |
| "epoch": 2.0834621329211744, | |
| "grad_norm": 0.0015832912176847458, | |
| "learning_rate": 0.0003055126223596085, | |
| "loss": 0.9131, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 2.0865533230293662, | |
| "grad_norm": 0.0018890087958425283, | |
| "learning_rate": 0.0003044822256568779, | |
| "loss": 0.8798, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 2.089644513137558, | |
| "grad_norm": 0.0020662271417677402, | |
| "learning_rate": 0.00030345182895414734, | |
| "loss": 0.9027, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 2.0927357032457494, | |
| "grad_norm": 0.001351105165667832, | |
| "learning_rate": 0.0003024214322514168, | |
| "loss": 0.8641, | |
| "step": 6770 | |
| }, | |
| { | |
| "epoch": 2.0958268933539412, | |
| "grad_norm": 0.001607073936611414, | |
| "learning_rate": 0.00030139103554868626, | |
| "loss": 0.8716, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 2.098918083462133, | |
| "grad_norm": 0.0022712023928761482, | |
| "learning_rate": 0.0003003606388459557, | |
| "loss": 0.9286, | |
| "step": 6790 | |
| }, | |
| { | |
| "epoch": 2.1020092735703244, | |
| "grad_norm": 0.0019429827807471156, | |
| "learning_rate": 0.0002993302421432251, | |
| "loss": 0.9157, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.105100463678516, | |
| "grad_norm": 0.0023945835418999195, | |
| "learning_rate": 0.0002982998454404946, | |
| "loss": 0.9332, | |
| "step": 6810 | |
| }, | |
| { | |
| "epoch": 2.108191653786708, | |
| "grad_norm": 0.0017035908531397581, | |
| "learning_rate": 0.0002972694487377641, | |
| "loss": 0.8984, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 2.1112828438948994, | |
| "grad_norm": 0.001636037020944059, | |
| "learning_rate": 0.00029623905203503346, | |
| "loss": 0.8535, | |
| "step": 6830 | |
| }, | |
| { | |
| "epoch": 2.114374034003091, | |
| "grad_norm": 0.0015807118033990264, | |
| "learning_rate": 0.00029520865533230295, | |
| "loss": 0.8537, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 2.117465224111283, | |
| "grad_norm": 0.0019097881158813834, | |
| "learning_rate": 0.0002941782586295724, | |
| "loss": 0.9224, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 2.1205564142194744, | |
| "grad_norm": 0.002146846381947398, | |
| "learning_rate": 0.00029314786192684186, | |
| "loss": 0.9206, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 2.123647604327666, | |
| "grad_norm": 0.0016978083876892924, | |
| "learning_rate": 0.0002921174652241113, | |
| "loss": 0.9441, | |
| "step": 6870 | |
| }, | |
| { | |
| "epoch": 2.126738794435858, | |
| "grad_norm": 0.0015323269180953503, | |
| "learning_rate": 0.0002910870685213807, | |
| "loss": 0.9336, | |
| "step": 6880 | |
| }, | |
| { | |
| "epoch": 2.1298299845440494, | |
| "grad_norm": 0.0014333493309095502, | |
| "learning_rate": 0.0002900566718186502, | |
| "loss": 0.9192, | |
| "step": 6890 | |
| }, | |
| { | |
| "epoch": 2.132921174652241, | |
| "grad_norm": 0.0017516023945063353, | |
| "learning_rate": 0.00028902627511591964, | |
| "loss": 0.8886, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.1360123647604325, | |
| "grad_norm": 0.0013508939882740378, | |
| "learning_rate": 0.00028799587841318907, | |
| "loss": 0.8788, | |
| "step": 6910 | |
| }, | |
| { | |
| "epoch": 2.1391035548686244, | |
| "grad_norm": 0.002034724224358797, | |
| "learning_rate": 0.00028696548171045855, | |
| "loss": 0.8884, | |
| "step": 6920 | |
| }, | |
| { | |
| "epoch": 2.142194744976816, | |
| "grad_norm": 0.0014981675194576383, | |
| "learning_rate": 0.000285935085007728, | |
| "loss": 0.9187, | |
| "step": 6930 | |
| }, | |
| { | |
| "epoch": 2.1452859350850075, | |
| "grad_norm": 0.002125379629433155, | |
| "learning_rate": 0.00028490468830499747, | |
| "loss": 0.8873, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 2.1483771251931993, | |
| "grad_norm": 0.001538197393529117, | |
| "learning_rate": 0.00028387429160226684, | |
| "loss": 0.9144, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 2.151468315301391, | |
| "grad_norm": 0.002346182242035866, | |
| "learning_rate": 0.0002828438948995363, | |
| "loss": 0.9295, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 2.1545595054095825, | |
| "grad_norm": 0.0015058065764606, | |
| "learning_rate": 0.0002818134981968058, | |
| "loss": 0.857, | |
| "step": 6970 | |
| }, | |
| { | |
| "epoch": 2.1576506955177743, | |
| "grad_norm": 0.0017613341333344579, | |
| "learning_rate": 0.0002807831014940752, | |
| "loss": 0.8723, | |
| "step": 6980 | |
| }, | |
| { | |
| "epoch": 2.160741885625966, | |
| "grad_norm": 0.0021604488138109446, | |
| "learning_rate": 0.00027975270479134467, | |
| "loss": 0.934, | |
| "step": 6990 | |
| }, | |
| { | |
| "epoch": 2.1638330757341575, | |
| "grad_norm": 0.0017267893999814987, | |
| "learning_rate": 0.0002787223080886141, | |
| "loss": 0.8985, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.1669242658423493, | |
| "grad_norm": 0.0014519842807203531, | |
| "learning_rate": 0.0002776919113858836, | |
| "loss": 0.9185, | |
| "step": 7010 | |
| }, | |
| { | |
| "epoch": 2.170015455950541, | |
| "grad_norm": 0.0023860172368586063, | |
| "learning_rate": 0.000276661514683153, | |
| "loss": 0.9081, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 2.1731066460587325, | |
| "grad_norm": 0.0017903451807796955, | |
| "learning_rate": 0.00027563111798042245, | |
| "loss": 0.9024, | |
| "step": 7030 | |
| }, | |
| { | |
| "epoch": 2.1761978361669243, | |
| "grad_norm": 0.0015208751428872347, | |
| "learning_rate": 0.00027460072127769193, | |
| "loss": 0.8988, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 2.179289026275116, | |
| "grad_norm": 0.0019341211300343275, | |
| "learning_rate": 0.0002735703245749614, | |
| "loss": 0.9378, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 2.1823802163833075, | |
| "grad_norm": 0.0016929497942328453, | |
| "learning_rate": 0.0002725399278722308, | |
| "loss": 0.9231, | |
| "step": 7060 | |
| }, | |
| { | |
| "epoch": 2.1854714064914993, | |
| "grad_norm": 0.0015913585666567087, | |
| "learning_rate": 0.0002715095311695003, | |
| "loss": 0.8959, | |
| "step": 7070 | |
| }, | |
| { | |
| "epoch": 2.1885625965996907, | |
| "grad_norm": 0.001789154834114015, | |
| "learning_rate": 0.0002704791344667697, | |
| "loss": 0.9459, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 2.1916537867078825, | |
| "grad_norm": 0.00176356197334826, | |
| "learning_rate": 0.0002694487377640392, | |
| "loss": 0.915, | |
| "step": 7090 | |
| }, | |
| { | |
| "epoch": 2.1947449768160743, | |
| "grad_norm": 0.0034130492713302374, | |
| "learning_rate": 0.0002684183410613086, | |
| "loss": 0.8311, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 2.1978361669242656, | |
| "grad_norm": 0.001789650646969676, | |
| "learning_rate": 0.00026738794435857805, | |
| "loss": 0.9061, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 2.2009273570324575, | |
| "grad_norm": 0.0015091504901647568, | |
| "learning_rate": 0.00026635754765584754, | |
| "loss": 0.882, | |
| "step": 7120 | |
| }, | |
| { | |
| "epoch": 2.2040185471406493, | |
| "grad_norm": 0.0016386975767090917, | |
| "learning_rate": 0.0002653271509531169, | |
| "loss": 0.8652, | |
| "step": 7130 | |
| }, | |
| { | |
| "epoch": 2.2071097372488406, | |
| "grad_norm": 0.0018872515065595508, | |
| "learning_rate": 0.0002642967542503864, | |
| "loss": 0.9064, | |
| "step": 7140 | |
| }, | |
| { | |
| "epoch": 2.2102009273570324, | |
| "grad_norm": 0.0016174730844795704, | |
| "learning_rate": 0.0002632663575476559, | |
| "loss": 0.8615, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 2.2132921174652243, | |
| "grad_norm": 0.0023867471609264612, | |
| "learning_rate": 0.0002622359608449253, | |
| "loss": 0.903, | |
| "step": 7160 | |
| }, | |
| { | |
| "epoch": 2.2163833075734156, | |
| "grad_norm": 0.0018768367590382695, | |
| "learning_rate": 0.00026120556414219474, | |
| "loss": 0.8828, | |
| "step": 7170 | |
| }, | |
| { | |
| "epoch": 2.2194744976816074, | |
| "grad_norm": 0.0019271258497610688, | |
| "learning_rate": 0.00026017516743946417, | |
| "loss": 0.8861, | |
| "step": 7180 | |
| }, | |
| { | |
| "epoch": 2.2225656877897992, | |
| "grad_norm": 0.001598935341462493, | |
| "learning_rate": 0.00025914477073673366, | |
| "loss": 0.8821, | |
| "step": 7190 | |
| }, | |
| { | |
| "epoch": 2.2256568778979906, | |
| "grad_norm": 0.002534502651542425, | |
| "learning_rate": 0.00025811437403400314, | |
| "loss": 0.923, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 2.2287480680061824, | |
| "grad_norm": 0.001254307571798563, | |
| "learning_rate": 0.0002570839773312725, | |
| "loss": 0.8744, | |
| "step": 7210 | |
| }, | |
| { | |
| "epoch": 2.2318392581143742, | |
| "grad_norm": 0.0013876528246328235, | |
| "learning_rate": 0.000256053580628542, | |
| "loss": 0.909, | |
| "step": 7220 | |
| }, | |
| { | |
| "epoch": 2.2349304482225656, | |
| "grad_norm": 0.0030744040850549936, | |
| "learning_rate": 0.00025502318392581143, | |
| "loss": 0.9358, | |
| "step": 7230 | |
| }, | |
| { | |
| "epoch": 2.2380216383307574, | |
| "grad_norm": 0.001663259114138782, | |
| "learning_rate": 0.0002539927872230809, | |
| "loss": 0.9078, | |
| "step": 7240 | |
| }, | |
| { | |
| "epoch": 2.2411128284389488, | |
| "grad_norm": 0.00184043834451586, | |
| "learning_rate": 0.00025296239052035035, | |
| "loss": 0.8703, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 2.2442040185471406, | |
| "grad_norm": 0.002181377960368991, | |
| "learning_rate": 0.0002519319938176198, | |
| "loss": 0.8823, | |
| "step": 7260 | |
| }, | |
| { | |
| "epoch": 2.2472952086553324, | |
| "grad_norm": 0.0015466611366719007, | |
| "learning_rate": 0.00025090159711488926, | |
| "loss": 0.8792, | |
| "step": 7270 | |
| }, | |
| { | |
| "epoch": 2.250386398763524, | |
| "grad_norm": 0.002074003452435136, | |
| "learning_rate": 0.0002498712004121587, | |
| "loss": 0.9361, | |
| "step": 7280 | |
| }, | |
| { | |
| "epoch": 2.2534775888717156, | |
| "grad_norm": 0.0014814219903200865, | |
| "learning_rate": 0.0002488408037094281, | |
| "loss": 0.8684, | |
| "step": 7290 | |
| }, | |
| { | |
| "epoch": 2.2565687789799074, | |
| "grad_norm": 0.0022694983053952456, | |
| "learning_rate": 0.00024781040700669755, | |
| "loss": 0.8592, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 2.2596599690880987, | |
| "grad_norm": 0.0019474639557301998, | |
| "learning_rate": 0.00024678001030396704, | |
| "loss": 0.8866, | |
| "step": 7310 | |
| }, | |
| { | |
| "epoch": 2.2627511591962906, | |
| "grad_norm": 0.0018335517961531878, | |
| "learning_rate": 0.00024574961360123647, | |
| "loss": 0.9352, | |
| "step": 7320 | |
| }, | |
| { | |
| "epoch": 2.2658423493044824, | |
| "grad_norm": 0.001545943901874125, | |
| "learning_rate": 0.00024471921689850595, | |
| "loss": 0.868, | |
| "step": 7330 | |
| }, | |
| { | |
| "epoch": 2.2689335394126737, | |
| "grad_norm": 0.0019200016977265477, | |
| "learning_rate": 0.00024368882019577538, | |
| "loss": 0.9414, | |
| "step": 7340 | |
| }, | |
| { | |
| "epoch": 2.2720247295208655, | |
| "grad_norm": 0.002204937394708395, | |
| "learning_rate": 0.0002426584234930448, | |
| "loss": 0.919, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 2.2751159196290573, | |
| "grad_norm": 0.002049383707344532, | |
| "learning_rate": 0.0002416280267903143, | |
| "loss": 0.9024, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 2.2782071097372487, | |
| "grad_norm": 0.0016608345322310925, | |
| "learning_rate": 0.00024059763008758373, | |
| "loss": 0.9112, | |
| "step": 7370 | |
| }, | |
| { | |
| "epoch": 2.2812982998454405, | |
| "grad_norm": 0.0020049915183335543, | |
| "learning_rate": 0.00023956723338485318, | |
| "loss": 0.9082, | |
| "step": 7380 | |
| }, | |
| { | |
| "epoch": 2.2843894899536323, | |
| "grad_norm": 0.0017916331999003887, | |
| "learning_rate": 0.0002385368366821226, | |
| "loss": 0.9242, | |
| "step": 7390 | |
| }, | |
| { | |
| "epoch": 2.2874806800618237, | |
| "grad_norm": 0.0015080280136317015, | |
| "learning_rate": 0.00023750643997939207, | |
| "loss": 0.9115, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 2.2905718701700155, | |
| "grad_norm": 0.00174785649869591, | |
| "learning_rate": 0.00023647604327666153, | |
| "loss": 0.8643, | |
| "step": 7410 | |
| }, | |
| { | |
| "epoch": 2.293663060278207, | |
| "grad_norm": 0.002300349297001958, | |
| "learning_rate": 0.00023544564657393096, | |
| "loss": 0.9215, | |
| "step": 7420 | |
| }, | |
| { | |
| "epoch": 2.2967542503863987, | |
| "grad_norm": 0.001717501669190824, | |
| "learning_rate": 0.00023441524987120042, | |
| "loss": 0.881, | |
| "step": 7430 | |
| }, | |
| { | |
| "epoch": 2.2998454404945905, | |
| "grad_norm": 0.0015334953786805272, | |
| "learning_rate": 0.00023338485316846985, | |
| "loss": 0.9106, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 2.3029366306027823, | |
| "grad_norm": 0.00180353585164994, | |
| "learning_rate": 0.00023235445646573933, | |
| "loss": 0.9088, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 2.3060278207109737, | |
| "grad_norm": 0.0016415161080658436, | |
| "learning_rate": 0.00023132405976300876, | |
| "loss": 0.8925, | |
| "step": 7460 | |
| }, | |
| { | |
| "epoch": 2.3091190108191655, | |
| "grad_norm": 0.002030453644692898, | |
| "learning_rate": 0.00023029366306027822, | |
| "loss": 0.8819, | |
| "step": 7470 | |
| }, | |
| { | |
| "epoch": 2.312210200927357, | |
| "grad_norm": 0.0019409249071031809, | |
| "learning_rate": 0.00022926326635754765, | |
| "loss": 0.867, | |
| "step": 7480 | |
| }, | |
| { | |
| "epoch": 2.3153013910355487, | |
| "grad_norm": 0.0034679800737649202, | |
| "learning_rate": 0.0002282328696548171, | |
| "loss": 0.8954, | |
| "step": 7490 | |
| }, | |
| { | |
| "epoch": 2.3183925811437405, | |
| "grad_norm": 0.0020659409929066896, | |
| "learning_rate": 0.00022720247295208656, | |
| "loss": 0.9357, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.321483771251932, | |
| "grad_norm": 0.0021641727071255445, | |
| "learning_rate": 0.00022617207624935602, | |
| "loss": 0.8911, | |
| "step": 7510 | |
| }, | |
| { | |
| "epoch": 2.3245749613601236, | |
| "grad_norm": 0.0016261821147054434, | |
| "learning_rate": 0.00022514167954662545, | |
| "loss": 0.9102, | |
| "step": 7520 | |
| }, | |
| { | |
| "epoch": 2.3276661514683155, | |
| "grad_norm": 0.0018004965968430042, | |
| "learning_rate": 0.0002241112828438949, | |
| "loss": 0.8924, | |
| "step": 7530 | |
| }, | |
| { | |
| "epoch": 2.330757341576507, | |
| "grad_norm": 0.0016102648805826902, | |
| "learning_rate": 0.00022308088614116434, | |
| "loss": 0.8812, | |
| "step": 7540 | |
| }, | |
| { | |
| "epoch": 2.3338485316846986, | |
| "grad_norm": 0.0018460742430761456, | |
| "learning_rate": 0.00022205048943843382, | |
| "loss": 0.8935, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 2.3369397217928904, | |
| "grad_norm": 0.0018313312903046608, | |
| "learning_rate": 0.00022102009273570325, | |
| "loss": 0.8892, | |
| "step": 7560 | |
| }, | |
| { | |
| "epoch": 2.340030911901082, | |
| "grad_norm": 0.0018456524703651667, | |
| "learning_rate": 0.0002199896960329727, | |
| "loss": 0.8874, | |
| "step": 7570 | |
| }, | |
| { | |
| "epoch": 2.3431221020092736, | |
| "grad_norm": 0.0028236303478479385, | |
| "learning_rate": 0.00021895929933024214, | |
| "loss": 0.9291, | |
| "step": 7580 | |
| }, | |
| { | |
| "epoch": 2.346213292117465, | |
| "grad_norm": 0.0012284901458770037, | |
| "learning_rate": 0.00021792890262751157, | |
| "loss": 0.8771, | |
| "step": 7590 | |
| }, | |
| { | |
| "epoch": 2.349304482225657, | |
| "grad_norm": 0.0018673281883820891, | |
| "learning_rate": 0.00021689850592478106, | |
| "loss": 0.8804, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 2.3523956723338486, | |
| "grad_norm": 0.0013668534811586142, | |
| "learning_rate": 0.00021586810922205049, | |
| "loss": 0.9415, | |
| "step": 7610 | |
| }, | |
| { | |
| "epoch": 2.3554868624420404, | |
| "grad_norm": 0.0014791954308748245, | |
| "learning_rate": 0.00021483771251931994, | |
| "loss": 0.9123, | |
| "step": 7620 | |
| }, | |
| { | |
| "epoch": 2.358578052550232, | |
| "grad_norm": 0.0016408261144533753, | |
| "learning_rate": 0.00021380731581658937, | |
| "loss": 0.9296, | |
| "step": 7630 | |
| }, | |
| { | |
| "epoch": 2.3616692426584236, | |
| "grad_norm": 0.0016208128072321415, | |
| "learning_rate": 0.00021277691911385886, | |
| "loss": 0.8988, | |
| "step": 7640 | |
| }, | |
| { | |
| "epoch": 2.364760432766615, | |
| "grad_norm": 0.0016079987399280071, | |
| "learning_rate": 0.0002117465224111283, | |
| "loss": 0.8697, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 2.3678516228748068, | |
| "grad_norm": 0.0018998971208930016, | |
| "learning_rate": 0.00021071612570839775, | |
| "loss": 0.8971, | |
| "step": 7660 | |
| }, | |
| { | |
| "epoch": 2.3709428129829986, | |
| "grad_norm": 0.0013807121431455016, | |
| "learning_rate": 0.00020968572900566718, | |
| "loss": 0.9038, | |
| "step": 7670 | |
| }, | |
| { | |
| "epoch": 2.37403400309119, | |
| "grad_norm": 0.0016093014273792505, | |
| "learning_rate": 0.00020865533230293663, | |
| "loss": 0.8853, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 2.3771251931993818, | |
| "grad_norm": 0.001660670735873282, | |
| "learning_rate": 0.0002076249356002061, | |
| "loss": 0.8867, | |
| "step": 7690 | |
| }, | |
| { | |
| "epoch": 2.3802163833075736, | |
| "grad_norm": 0.001987049588933587, | |
| "learning_rate": 0.00020659453889747555, | |
| "loss": 0.9064, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 2.383307573415765, | |
| "grad_norm": 0.0014046692522242665, | |
| "learning_rate": 0.00020556414219474498, | |
| "loss": 0.9145, | |
| "step": 7710 | |
| }, | |
| { | |
| "epoch": 2.3863987635239567, | |
| "grad_norm": 0.0018706049304455519, | |
| "learning_rate": 0.00020453374549201444, | |
| "loss": 0.8975, | |
| "step": 7720 | |
| }, | |
| { | |
| "epoch": 2.3894899536321486, | |
| "grad_norm": 0.001406969386152923, | |
| "learning_rate": 0.00020350334878928387, | |
| "loss": 0.9143, | |
| "step": 7730 | |
| }, | |
| { | |
| "epoch": 2.39258114374034, | |
| "grad_norm": 0.001984959002584219, | |
| "learning_rate": 0.00020247295208655332, | |
| "loss": 0.9189, | |
| "step": 7740 | |
| }, | |
| { | |
| "epoch": 2.3956723338485317, | |
| "grad_norm": 0.0022590451408177614, | |
| "learning_rate": 0.00020144255538382278, | |
| "loss": 0.9174, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 2.398763523956723, | |
| "grad_norm": 0.0013520545326173306, | |
| "learning_rate": 0.0002004121586810922, | |
| "loss": 0.8876, | |
| "step": 7760 | |
| }, | |
| { | |
| "epoch": 2.401854714064915, | |
| "grad_norm": 0.001583244651556015, | |
| "learning_rate": 0.00019938176197836167, | |
| "loss": 0.9106, | |
| "step": 7770 | |
| }, | |
| { | |
| "epoch": 2.4049459041731067, | |
| "grad_norm": 0.0016318537527695298, | |
| "learning_rate": 0.00019835136527563113, | |
| "loss": 0.8566, | |
| "step": 7780 | |
| }, | |
| { | |
| "epoch": 2.4080370942812985, | |
| "grad_norm": 0.0015274740289896727, | |
| "learning_rate": 0.00019732096857290058, | |
| "loss": 0.9214, | |
| "step": 7790 | |
| }, | |
| { | |
| "epoch": 2.41112828438949, | |
| "grad_norm": 0.0017980411648750305, | |
| "learning_rate": 0.00019629057187017, | |
| "loss": 0.8636, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 2.4142194744976817, | |
| "grad_norm": 0.0016120801446959376, | |
| "learning_rate": 0.00019526017516743947, | |
| "loss": 0.8821, | |
| "step": 7810 | |
| }, | |
| { | |
| "epoch": 2.417310664605873, | |
| "grad_norm": 0.001640370930545032, | |
| "learning_rate": 0.0001942297784647089, | |
| "loss": 0.9068, | |
| "step": 7820 | |
| }, | |
| { | |
| "epoch": 2.420401854714065, | |
| "grad_norm": 0.0015430683270096779, | |
| "learning_rate": 0.00019319938176197838, | |
| "loss": 0.8889, | |
| "step": 7830 | |
| }, | |
| { | |
| "epoch": 2.4234930448222567, | |
| "grad_norm": 0.0015471933875232935, | |
| "learning_rate": 0.00019216898505924782, | |
| "loss": 0.8924, | |
| "step": 7840 | |
| }, | |
| { | |
| "epoch": 2.426584234930448, | |
| "grad_norm": 0.0021910767536610365, | |
| "learning_rate": 0.00019113858835651727, | |
| "loss": 0.8569, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 2.42967542503864, | |
| "grad_norm": 0.0017051781760528684, | |
| "learning_rate": 0.0001901081916537867, | |
| "loss": 0.937, | |
| "step": 7860 | |
| }, | |
| { | |
| "epoch": 2.4327666151468317, | |
| "grad_norm": 0.0018304622499272227, | |
| "learning_rate": 0.00018907779495105616, | |
| "loss": 0.9255, | |
| "step": 7870 | |
| }, | |
| { | |
| "epoch": 2.435857805255023, | |
| "grad_norm": 0.001733385375700891, | |
| "learning_rate": 0.00018804739824832562, | |
| "loss": 0.8877, | |
| "step": 7880 | |
| }, | |
| { | |
| "epoch": 2.438948995363215, | |
| "grad_norm": 0.0014930800534784794, | |
| "learning_rate": 0.00018701700154559507, | |
| "loss": 0.9079, | |
| "step": 7890 | |
| }, | |
| { | |
| "epoch": 2.4420401854714067, | |
| "grad_norm": 0.0017946161096915603, | |
| "learning_rate": 0.0001859866048428645, | |
| "loss": 0.8372, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 2.445131375579598, | |
| "grad_norm": 0.00181410217192024, | |
| "learning_rate": 0.00018495620814013394, | |
| "loss": 0.9301, | |
| "step": 7910 | |
| }, | |
| { | |
| "epoch": 2.44822256568779, | |
| "grad_norm": 0.0018277463968843222, | |
| "learning_rate": 0.00018392581143740342, | |
| "loss": 0.9406, | |
| "step": 7920 | |
| }, | |
| { | |
| "epoch": 2.451313755795981, | |
| "grad_norm": 0.0017499460373073816, | |
| "learning_rate": 0.00018289541473467285, | |
| "loss": 0.9327, | |
| "step": 7930 | |
| }, | |
| { | |
| "epoch": 2.454404945904173, | |
| "grad_norm": 0.0017523594433441758, | |
| "learning_rate": 0.0001818650180319423, | |
| "loss": 0.9541, | |
| "step": 7940 | |
| }, | |
| { | |
| "epoch": 2.457496136012365, | |
| "grad_norm": 0.001588582992553711, | |
| "learning_rate": 0.00018083462132921174, | |
| "loss": 0.9244, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 2.4605873261205566, | |
| "grad_norm": 0.0015729885781183839, | |
| "learning_rate": 0.0001798042246264812, | |
| "loss": 0.902, | |
| "step": 7960 | |
| }, | |
| { | |
| "epoch": 2.463678516228748, | |
| "grad_norm": 0.001764149172231555, | |
| "learning_rate": 0.00017877382792375065, | |
| "loss": 0.9115, | |
| "step": 7970 | |
| }, | |
| { | |
| "epoch": 2.46676970633694, | |
| "grad_norm": 0.0020394367165863514, | |
| "learning_rate": 0.0001777434312210201, | |
| "loss": 0.8913, | |
| "step": 7980 | |
| }, | |
| { | |
| "epoch": 2.469860896445131, | |
| "grad_norm": 0.001661314396187663, | |
| "learning_rate": 0.00017671303451828954, | |
| "loss": 0.8924, | |
| "step": 7990 | |
| }, | |
| { | |
| "epoch": 2.472952086553323, | |
| "grad_norm": 0.0016929521225392818, | |
| "learning_rate": 0.000175682637815559, | |
| "loss": 0.9158, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.476043276661515, | |
| "grad_norm": 0.0014037607470527291, | |
| "learning_rate": 0.00017465224111282843, | |
| "loss": 0.8929, | |
| "step": 8010 | |
| }, | |
| { | |
| "epoch": 2.479134466769706, | |
| "grad_norm": 0.0012340841349214315, | |
| "learning_rate": 0.0001736218444100979, | |
| "loss": 0.9042, | |
| "step": 8020 | |
| }, | |
| { | |
| "epoch": 2.482225656877898, | |
| "grad_norm": 0.0016911630518734455, | |
| "learning_rate": 0.00017259144770736734, | |
| "loss": 0.8805, | |
| "step": 8030 | |
| }, | |
| { | |
| "epoch": 2.48531684698609, | |
| "grad_norm": 0.0015811780467629433, | |
| "learning_rate": 0.0001715610510046368, | |
| "loss": 0.9066, | |
| "step": 8040 | |
| }, | |
| { | |
| "epoch": 2.488408037094281, | |
| "grad_norm": 0.0022526499815285206, | |
| "learning_rate": 0.00017053065430190623, | |
| "loss": 0.8729, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 2.491499227202473, | |
| "grad_norm": 0.0014964583097025752, | |
| "learning_rate": 0.00016950025759917566, | |
| "loss": 0.8867, | |
| "step": 8060 | |
| }, | |
| { | |
| "epoch": 2.4945904173106648, | |
| "grad_norm": 0.001667377888225019, | |
| "learning_rate": 0.00016846986089644514, | |
| "loss": 0.9289, | |
| "step": 8070 | |
| }, | |
| { | |
| "epoch": 2.497681607418856, | |
| "grad_norm": 0.0015655744355171919, | |
| "learning_rate": 0.00016743946419371457, | |
| "loss": 0.8989, | |
| "step": 8080 | |
| }, | |
| { | |
| "epoch": 2.500772797527048, | |
| "grad_norm": 0.0018760713282972574, | |
| "learning_rate": 0.00016640906749098403, | |
| "loss": 0.8951, | |
| "step": 8090 | |
| }, | |
| { | |
| "epoch": 2.5038639876352393, | |
| "grad_norm": 0.0018504380714148283, | |
| "learning_rate": 0.00016537867078825346, | |
| "loss": 0.9548, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 2.506955177743431, | |
| "grad_norm": 0.0015352640766650438, | |
| "learning_rate": 0.00016434827408552295, | |
| "loss": 0.8789, | |
| "step": 8110 | |
| }, | |
| { | |
| "epoch": 2.510046367851623, | |
| "grad_norm": 0.0014199953293427825, | |
| "learning_rate": 0.00016331787738279238, | |
| "loss": 0.8956, | |
| "step": 8120 | |
| }, | |
| { | |
| "epoch": 2.5131375579598147, | |
| "grad_norm": 0.0022967271506786346, | |
| "learning_rate": 0.00016228748068006183, | |
| "loss": 0.8662, | |
| "step": 8130 | |
| }, | |
| { | |
| "epoch": 2.516228748068006, | |
| "grad_norm": 0.0015619174810126424, | |
| "learning_rate": 0.00016125708397733126, | |
| "loss": 0.9188, | |
| "step": 8140 | |
| }, | |
| { | |
| "epoch": 2.519319938176198, | |
| "grad_norm": 0.00181775342207402, | |
| "learning_rate": 0.00016022668727460072, | |
| "loss": 0.9115, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 2.5224111282843893, | |
| "grad_norm": 0.001615070621483028, | |
| "learning_rate": 0.00015919629057187018, | |
| "loss": 0.8719, | |
| "step": 8160 | |
| }, | |
| { | |
| "epoch": 2.525502318392581, | |
| "grad_norm": 0.002030865289270878, | |
| "learning_rate": 0.00015816589386913964, | |
| "loss": 0.8618, | |
| "step": 8170 | |
| }, | |
| { | |
| "epoch": 2.528593508500773, | |
| "grad_norm": 0.0018763948464766145, | |
| "learning_rate": 0.00015713549716640907, | |
| "loss": 0.9493, | |
| "step": 8180 | |
| }, | |
| { | |
| "epoch": 2.5316846986089647, | |
| "grad_norm": 0.0015980995958670974, | |
| "learning_rate": 0.00015610510046367852, | |
| "loss": 0.9139, | |
| "step": 8190 | |
| }, | |
| { | |
| "epoch": 2.534775888717156, | |
| "grad_norm": 0.0017758564790710807, | |
| "learning_rate": 0.00015507470376094795, | |
| "loss": 0.8785, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 2.537867078825348, | |
| "grad_norm": 0.0018766775028780103, | |
| "learning_rate": 0.0001540443070582174, | |
| "loss": 0.9099, | |
| "step": 8210 | |
| }, | |
| { | |
| "epoch": 2.5409582689335393, | |
| "grad_norm": 0.0018965909257531166, | |
| "learning_rate": 0.00015301391035548687, | |
| "loss": 0.9314, | |
| "step": 8220 | |
| }, | |
| { | |
| "epoch": 2.544049459041731, | |
| "grad_norm": 0.0015854688826948404, | |
| "learning_rate": 0.0001519835136527563, | |
| "loss": 0.9137, | |
| "step": 8230 | |
| }, | |
| { | |
| "epoch": 2.547140649149923, | |
| "grad_norm": 0.0018873221706598997, | |
| "learning_rate": 0.00015095311695002576, | |
| "loss": 0.8828, | |
| "step": 8240 | |
| }, | |
| { | |
| "epoch": 2.5502318392581143, | |
| "grad_norm": 0.0014826676342636347, | |
| "learning_rate": 0.00014992272024729521, | |
| "loss": 0.9031, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 2.553323029366306, | |
| "grad_norm": 0.0014809026615694165, | |
| "learning_rate": 0.00014889232354456467, | |
| "loss": 0.8574, | |
| "step": 8260 | |
| }, | |
| { | |
| "epoch": 2.5564142194744974, | |
| "grad_norm": 0.0014648685464635491, | |
| "learning_rate": 0.0001478619268418341, | |
| "loss": 0.8706, | |
| "step": 8270 | |
| }, | |
| { | |
| "epoch": 2.5595054095826892, | |
| "grad_norm": 0.001307973056100309, | |
| "learning_rate": 0.00014683153013910356, | |
| "loss": 0.9271, | |
| "step": 8280 | |
| }, | |
| { | |
| "epoch": 2.562596599690881, | |
| "grad_norm": 0.0019158597569912672, | |
| "learning_rate": 0.000145801133436373, | |
| "loss": 0.8816, | |
| "step": 8290 | |
| }, | |
| { | |
| "epoch": 2.565687789799073, | |
| "grad_norm": 0.0016835506539791822, | |
| "learning_rate": 0.00014477073673364247, | |
| "loss": 0.9075, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 2.5687789799072642, | |
| "grad_norm": 0.0020169655326753855, | |
| "learning_rate": 0.0001437403400309119, | |
| "loss": 0.8927, | |
| "step": 8310 | |
| }, | |
| { | |
| "epoch": 2.571870170015456, | |
| "grad_norm": 0.001817848184145987, | |
| "learning_rate": 0.00014270994332818136, | |
| "loss": 0.8812, | |
| "step": 8320 | |
| }, | |
| { | |
| "epoch": 2.5749613601236474, | |
| "grad_norm": 0.001979761989787221, | |
| "learning_rate": 0.0001416795466254508, | |
| "loss": 0.9054, | |
| "step": 8330 | |
| }, | |
| { | |
| "epoch": 2.578052550231839, | |
| "grad_norm": 0.002088018460199237, | |
| "learning_rate": 0.00014064914992272025, | |
| "loss": 0.8691, | |
| "step": 8340 | |
| }, | |
| { | |
| "epoch": 2.581143740340031, | |
| "grad_norm": 0.0017918642843142152, | |
| "learning_rate": 0.0001396187532199897, | |
| "loss": 0.9458, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 2.584234930448223, | |
| "grad_norm": 0.0016459986800327897, | |
| "learning_rate": 0.00013858835651725916, | |
| "loss": 0.9186, | |
| "step": 8360 | |
| }, | |
| { | |
| "epoch": 2.587326120556414, | |
| "grad_norm": 0.001407464500516653, | |
| "learning_rate": 0.0001375579598145286, | |
| "loss": 0.8839, | |
| "step": 8370 | |
| }, | |
| { | |
| "epoch": 2.590417310664606, | |
| "grad_norm": 0.001718651968985796, | |
| "learning_rate": 0.00013652756311179802, | |
| "loss": 0.9203, | |
| "step": 8380 | |
| }, | |
| { | |
| "epoch": 2.5935085007727974, | |
| "grad_norm": 0.0018458360573276877, | |
| "learning_rate": 0.0001354971664090675, | |
| "loss": 0.911, | |
| "step": 8390 | |
| }, | |
| { | |
| "epoch": 2.596599690880989, | |
| "grad_norm": 0.001538407290354371, | |
| "learning_rate": 0.00013446676970633694, | |
| "loss": 0.9139, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 2.599690880989181, | |
| "grad_norm": 0.001976667670533061, | |
| "learning_rate": 0.0001334363730036064, | |
| "loss": 0.9075, | |
| "step": 8410 | |
| }, | |
| { | |
| "epoch": 2.6027820710973724, | |
| "grad_norm": 0.0021034348756074905, | |
| "learning_rate": 0.00013240597630087583, | |
| "loss": 0.9181, | |
| "step": 8420 | |
| }, | |
| { | |
| "epoch": 2.605873261205564, | |
| "grad_norm": 0.00171546614728868, | |
| "learning_rate": 0.00013137557959814528, | |
| "loss": 0.9361, | |
| "step": 8430 | |
| }, | |
| { | |
| "epoch": 2.6089644513137555, | |
| "grad_norm": 0.001742625143378973, | |
| "learning_rate": 0.00013034518289541474, | |
| "loss": 0.8895, | |
| "step": 8440 | |
| }, | |
| { | |
| "epoch": 2.6120556414219473, | |
| "grad_norm": 0.0018996672006323934, | |
| "learning_rate": 0.0001293147861926842, | |
| "loss": 0.8474, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 2.615146831530139, | |
| "grad_norm": 0.001571224071085453, | |
| "learning_rate": 0.00012828438948995363, | |
| "loss": 0.8976, | |
| "step": 8460 | |
| }, | |
| { | |
| "epoch": 2.618238021638331, | |
| "grad_norm": 0.0018322218675166368, | |
| "learning_rate": 0.0001272539927872231, | |
| "loss": 0.9207, | |
| "step": 8470 | |
| }, | |
| { | |
| "epoch": 2.6213292117465223, | |
| "grad_norm": 0.0013319810386747122, | |
| "learning_rate": 0.00012622359608449252, | |
| "loss": 0.8674, | |
| "step": 8480 | |
| }, | |
| { | |
| "epoch": 2.624420401854714, | |
| "grad_norm": 0.0014434581389650702, | |
| "learning_rate": 0.000125193199381762, | |
| "loss": 0.829, | |
| "step": 8490 | |
| }, | |
| { | |
| "epoch": 2.6275115919629055, | |
| "grad_norm": 0.0027343255933374166, | |
| "learning_rate": 0.00012416280267903143, | |
| "loss": 0.8882, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.6306027820710973, | |
| "grad_norm": 0.003409834112972021, | |
| "learning_rate": 0.0001231324059763009, | |
| "loss": 0.9302, | |
| "step": 8510 | |
| }, | |
| { | |
| "epoch": 2.633693972179289, | |
| "grad_norm": 0.0013600644888356328, | |
| "learning_rate": 0.00012210200927357035, | |
| "loss": 0.876, | |
| "step": 8520 | |
| }, | |
| { | |
| "epoch": 2.636785162287481, | |
| "grad_norm": 0.0016775266267359257, | |
| "learning_rate": 0.00012107161257083978, | |
| "loss": 0.9064, | |
| "step": 8530 | |
| }, | |
| { | |
| "epoch": 2.6398763523956723, | |
| "grad_norm": 0.0017377499025315046, | |
| "learning_rate": 0.00012004121586810922, | |
| "loss": 0.9224, | |
| "step": 8540 | |
| }, | |
| { | |
| "epoch": 2.642967542503864, | |
| "grad_norm": 0.001847295556217432, | |
| "learning_rate": 0.00011901081916537868, | |
| "loss": 0.9149, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 2.6460587326120555, | |
| "grad_norm": 0.0014812115114182234, | |
| "learning_rate": 0.00011798042246264812, | |
| "loss": 0.8929, | |
| "step": 8560 | |
| }, | |
| { | |
| "epoch": 2.6491499227202473, | |
| "grad_norm": 0.0017952779307961464, | |
| "learning_rate": 0.00011695002575991758, | |
| "loss": 0.9285, | |
| "step": 8570 | |
| }, | |
| { | |
| "epoch": 2.652241112828439, | |
| "grad_norm": 0.001425527036190033, | |
| "learning_rate": 0.00011591962905718702, | |
| "loss": 0.8703, | |
| "step": 8580 | |
| }, | |
| { | |
| "epoch": 2.6553323029366305, | |
| "grad_norm": 0.001610907376743853, | |
| "learning_rate": 0.00011488923235445648, | |
| "loss": 0.9103, | |
| "step": 8590 | |
| }, | |
| { | |
| "epoch": 2.6584234930448223, | |
| "grad_norm": 0.001997585641220212, | |
| "learning_rate": 0.00011385883565172592, | |
| "loss": 0.898, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 2.6615146831530136, | |
| "grad_norm": 0.0015265914844349027, | |
| "learning_rate": 0.00011282843894899535, | |
| "loss": 0.8225, | |
| "step": 8610 | |
| }, | |
| { | |
| "epoch": 2.6646058732612055, | |
| "grad_norm": 0.002251180587336421, | |
| "learning_rate": 0.00011179804224626481, | |
| "loss": 0.8886, | |
| "step": 8620 | |
| }, | |
| { | |
| "epoch": 2.6676970633693973, | |
| "grad_norm": 0.0016794728580862284, | |
| "learning_rate": 0.00011076764554353426, | |
| "loss": 0.8991, | |
| "step": 8630 | |
| }, | |
| { | |
| "epoch": 2.670788253477589, | |
| "grad_norm": 0.0021927470806986094, | |
| "learning_rate": 0.00010973724884080371, | |
| "loss": 0.872, | |
| "step": 8640 | |
| }, | |
| { | |
| "epoch": 2.6738794435857804, | |
| "grad_norm": 0.0016734660603106022, | |
| "learning_rate": 0.00010870685213807316, | |
| "loss": 0.9082, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 2.6769706336939723, | |
| "grad_norm": 0.0023362876381725073, | |
| "learning_rate": 0.00010767645543534261, | |
| "loss": 0.8867, | |
| "step": 8660 | |
| }, | |
| { | |
| "epoch": 2.6800618238021636, | |
| "grad_norm": 0.0015447117621079087, | |
| "learning_rate": 0.00010664605873261206, | |
| "loss": 0.8706, | |
| "step": 8670 | |
| }, | |
| { | |
| "epoch": 2.6831530139103554, | |
| "grad_norm": 0.0017007689457386732, | |
| "learning_rate": 0.0001056156620298815, | |
| "loss": 0.9457, | |
| "step": 8680 | |
| }, | |
| { | |
| "epoch": 2.6862442040185472, | |
| "grad_norm": 0.0018746848218142986, | |
| "learning_rate": 0.00010458526532715096, | |
| "loss": 0.8834, | |
| "step": 8690 | |
| }, | |
| { | |
| "epoch": 2.689335394126739, | |
| "grad_norm": 0.0016901030903682113, | |
| "learning_rate": 0.0001035548686244204, | |
| "loss": 0.9321, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 2.6924265842349304, | |
| "grad_norm": 0.0017849161522462964, | |
| "learning_rate": 0.00010252447192168986, | |
| "loss": 0.8886, | |
| "step": 8710 | |
| }, | |
| { | |
| "epoch": 2.6955177743431222, | |
| "grad_norm": 0.0016776255797594786, | |
| "learning_rate": 0.0001014940752189593, | |
| "loss": 0.8842, | |
| "step": 8720 | |
| }, | |
| { | |
| "epoch": 2.6986089644513136, | |
| "grad_norm": 0.002052722033113241, | |
| "learning_rate": 0.00010046367851622876, | |
| "loss": 0.8921, | |
| "step": 8730 | |
| }, | |
| { | |
| "epoch": 2.7017001545595054, | |
| "grad_norm": 0.0020015337504446507, | |
| "learning_rate": 9.94332818134982e-05, | |
| "loss": 0.878, | |
| "step": 8740 | |
| }, | |
| { | |
| "epoch": 2.704791344667697, | |
| "grad_norm": 0.0018238229677081108, | |
| "learning_rate": 9.840288511076765e-05, | |
| "loss": 0.9286, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 2.7078825347758886, | |
| "grad_norm": 0.0019556416664272547, | |
| "learning_rate": 9.737248840803709e-05, | |
| "loss": 0.9109, | |
| "step": 8760 | |
| }, | |
| { | |
| "epoch": 2.7109737248840804, | |
| "grad_norm": 0.001552661880850792, | |
| "learning_rate": 9.634209170530654e-05, | |
| "loss": 0.9073, | |
| "step": 8770 | |
| }, | |
| { | |
| "epoch": 2.7140649149922718, | |
| "grad_norm": 0.0018825504230335355, | |
| "learning_rate": 9.5311695002576e-05, | |
| "loss": 0.9033, | |
| "step": 8780 | |
| }, | |
| { | |
| "epoch": 2.7171561051004636, | |
| "grad_norm": 0.0018266792176291347, | |
| "learning_rate": 9.428129829984544e-05, | |
| "loss": 0.946, | |
| "step": 8790 | |
| }, | |
| { | |
| "epoch": 2.7202472952086554, | |
| "grad_norm": 0.0014049782184883952, | |
| "learning_rate": 9.325090159711488e-05, | |
| "loss": 0.8817, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 2.723338485316847, | |
| "grad_norm": 0.0015499057481065392, | |
| "learning_rate": 9.222050489438434e-05, | |
| "loss": 0.8569, | |
| "step": 8810 | |
| }, | |
| { | |
| "epoch": 2.7264296754250386, | |
| "grad_norm": 0.0018245774554088712, | |
| "learning_rate": 9.119010819165378e-05, | |
| "loss": 0.8934, | |
| "step": 8820 | |
| }, | |
| { | |
| "epoch": 2.7295208655332304, | |
| "grad_norm": 0.001886395737528801, | |
| "learning_rate": 9.015971148892324e-05, | |
| "loss": 0.9212, | |
| "step": 8830 | |
| }, | |
| { | |
| "epoch": 2.7326120556414217, | |
| "grad_norm": 0.0017441367963328958, | |
| "learning_rate": 8.912931478619268e-05, | |
| "loss": 0.9378, | |
| "step": 8840 | |
| }, | |
| { | |
| "epoch": 2.7357032457496135, | |
| "grad_norm": 0.0015955539420247078, | |
| "learning_rate": 8.809891808346214e-05, | |
| "loss": 0.8835, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 2.7387944358578054, | |
| "grad_norm": 0.001902989810332656, | |
| "learning_rate": 8.706852138073158e-05, | |
| "loss": 0.9072, | |
| "step": 8860 | |
| }, | |
| { | |
| "epoch": 2.741885625965997, | |
| "grad_norm": 0.0014242329634726048, | |
| "learning_rate": 8.603812467800103e-05, | |
| "loss": 0.8712, | |
| "step": 8870 | |
| }, | |
| { | |
| "epoch": 2.7449768160741885, | |
| "grad_norm": 0.001587534206919372, | |
| "learning_rate": 8.500772797527049e-05, | |
| "loss": 0.8655, | |
| "step": 8880 | |
| }, | |
| { | |
| "epoch": 2.7480680061823803, | |
| "grad_norm": 0.001545070088468492, | |
| "learning_rate": 8.397733127253993e-05, | |
| "loss": 0.9077, | |
| "step": 8890 | |
| }, | |
| { | |
| "epoch": 2.7511591962905717, | |
| "grad_norm": 0.0018857244867831469, | |
| "learning_rate": 8.294693456980939e-05, | |
| "loss": 0.896, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 2.7542503863987635, | |
| "grad_norm": 0.0012212925357744098, | |
| "learning_rate": 8.191653786707883e-05, | |
| "loss": 0.8993, | |
| "step": 8910 | |
| }, | |
| { | |
| "epoch": 2.7573415765069553, | |
| "grad_norm": 0.0014877498615533113, | |
| "learning_rate": 8.088614116434827e-05, | |
| "loss": 0.9214, | |
| "step": 8920 | |
| }, | |
| { | |
| "epoch": 2.7604327666151467, | |
| "grad_norm": 0.0014304481446743011, | |
| "learning_rate": 7.985574446161772e-05, | |
| "loss": 0.8926, | |
| "step": 8930 | |
| }, | |
| { | |
| "epoch": 2.7635239567233385, | |
| "grad_norm": 0.001448179711587727, | |
| "learning_rate": 7.882534775888716e-05, | |
| "loss": 0.873, | |
| "step": 8940 | |
| }, | |
| { | |
| "epoch": 2.76661514683153, | |
| "grad_norm": 0.0015107804210856557, | |
| "learning_rate": 7.779495105615662e-05, | |
| "loss": 0.8396, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 2.7697063369397217, | |
| "grad_norm": 0.001311837462708354, | |
| "learning_rate": 7.676455435342606e-05, | |
| "loss": 0.9185, | |
| "step": 8960 | |
| }, | |
| { | |
| "epoch": 2.7727975270479135, | |
| "grad_norm": 0.0017208693316206336, | |
| "learning_rate": 7.573415765069552e-05, | |
| "loss": 0.8959, | |
| "step": 8970 | |
| }, | |
| { | |
| "epoch": 2.7758887171561053, | |
| "grad_norm": 0.0016424921341240406, | |
| "learning_rate": 7.470376094796496e-05, | |
| "loss": 0.8892, | |
| "step": 8980 | |
| }, | |
| { | |
| "epoch": 2.7789799072642967, | |
| "grad_norm": 0.0014334353618323803, | |
| "learning_rate": 7.367336424523442e-05, | |
| "loss": 0.8908, | |
| "step": 8990 | |
| }, | |
| { | |
| "epoch": 2.7820710973724885, | |
| "grad_norm": 0.0019086402608081698, | |
| "learning_rate": 7.264296754250387e-05, | |
| "loss": 0.8791, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.78516228748068, | |
| "grad_norm": 0.0013547363923862576, | |
| "learning_rate": 7.161257083977331e-05, | |
| "loss": 0.9024, | |
| "step": 9010 | |
| }, | |
| { | |
| "epoch": 2.7882534775888717, | |
| "grad_norm": 0.0013833148404955864, | |
| "learning_rate": 7.058217413704277e-05, | |
| "loss": 0.9195, | |
| "step": 9020 | |
| }, | |
| { | |
| "epoch": 2.7913446676970635, | |
| "grad_norm": 0.0014539181720465422, | |
| "learning_rate": 6.955177743431221e-05, | |
| "loss": 0.8835, | |
| "step": 9030 | |
| }, | |
| { | |
| "epoch": 2.7944358578052553, | |
| "grad_norm": 0.001557295210659504, | |
| "learning_rate": 6.852138073158167e-05, | |
| "loss": 0.8959, | |
| "step": 9040 | |
| }, | |
| { | |
| "epoch": 2.7975270479134466, | |
| "grad_norm": 0.0020530694164335728, | |
| "learning_rate": 6.749098402885111e-05, | |
| "loss": 0.9193, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 2.8006182380216385, | |
| "grad_norm": 0.0016817068681120872, | |
| "learning_rate": 6.646058732612057e-05, | |
| "loss": 0.8456, | |
| "step": 9060 | |
| }, | |
| { | |
| "epoch": 2.80370942812983, | |
| "grad_norm": 0.0019015265861526132, | |
| "learning_rate": 6.543019062339001e-05, | |
| "loss": 0.8723, | |
| "step": 9070 | |
| }, | |
| { | |
| "epoch": 2.8068006182380216, | |
| "grad_norm": 0.0015511283418163657, | |
| "learning_rate": 6.439979392065944e-05, | |
| "loss": 0.9393, | |
| "step": 9080 | |
| }, | |
| { | |
| "epoch": 2.8098918083462134, | |
| "grad_norm": 0.00140860746614635, | |
| "learning_rate": 6.33693972179289e-05, | |
| "loss": 0.8914, | |
| "step": 9090 | |
| }, | |
| { | |
| "epoch": 2.812982998454405, | |
| "grad_norm": 0.00200218940153718, | |
| "learning_rate": 6.233900051519836e-05, | |
| "loss": 0.8723, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 2.8160741885625966, | |
| "grad_norm": 0.001929080463014543, | |
| "learning_rate": 6.13086038124678e-05, | |
| "loss": 0.9241, | |
| "step": 9110 | |
| }, | |
| { | |
| "epoch": 2.819165378670788, | |
| "grad_norm": 0.0016536141047254205, | |
| "learning_rate": 6.0278207109737246e-05, | |
| "loss": 0.9228, | |
| "step": 9120 | |
| }, | |
| { | |
| "epoch": 2.82225656877898, | |
| "grad_norm": 0.0014851295854896307, | |
| "learning_rate": 5.9247810407006696e-05, | |
| "loss": 0.9038, | |
| "step": 9130 | |
| }, | |
| { | |
| "epoch": 2.8253477588871716, | |
| "grad_norm": 0.0014977608807384968, | |
| "learning_rate": 5.821741370427615e-05, | |
| "loss": 0.8881, | |
| "step": 9140 | |
| }, | |
| { | |
| "epoch": 2.8284389489953634, | |
| "grad_norm": 0.002265679184347391, | |
| "learning_rate": 5.71870170015456e-05, | |
| "loss": 0.8694, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 2.8315301391035548, | |
| "grad_norm": 0.0016861397307366133, | |
| "learning_rate": 5.615662029881505e-05, | |
| "loss": 0.849, | |
| "step": 9160 | |
| }, | |
| { | |
| "epoch": 2.8346213292117466, | |
| "grad_norm": 0.001969564938917756, | |
| "learning_rate": 5.512622359608449e-05, | |
| "loss": 0.8832, | |
| "step": 9170 | |
| }, | |
| { | |
| "epoch": 2.837712519319938, | |
| "grad_norm": 0.0019365076441317797, | |
| "learning_rate": 5.409582689335394e-05, | |
| "loss": 0.8763, | |
| "step": 9180 | |
| }, | |
| { | |
| "epoch": 2.8408037094281298, | |
| "grad_norm": 0.0024066604673862457, | |
| "learning_rate": 5.306543019062339e-05, | |
| "loss": 0.9283, | |
| "step": 9190 | |
| }, | |
| { | |
| "epoch": 2.8438948995363216, | |
| "grad_norm": 0.0017798724584281445, | |
| "learning_rate": 5.203503348789284e-05, | |
| "loss": 0.9098, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 2.8469860896445134, | |
| "grad_norm": 0.0014776487369090319, | |
| "learning_rate": 5.100463678516229e-05, | |
| "loss": 0.8685, | |
| "step": 9210 | |
| }, | |
| { | |
| "epoch": 2.8500772797527048, | |
| "grad_norm": 0.0016472855350002646, | |
| "learning_rate": 4.997424008243174e-05, | |
| "loss": 0.9151, | |
| "step": 9220 | |
| }, | |
| { | |
| "epoch": 2.8531684698608966, | |
| "grad_norm": 0.001969245495274663, | |
| "learning_rate": 4.894384337970119e-05, | |
| "loss": 0.8841, | |
| "step": 9230 | |
| }, | |
| { | |
| "epoch": 2.856259659969088, | |
| "grad_norm": 0.00151623017154634, | |
| "learning_rate": 4.791344667697063e-05, | |
| "loss": 0.8764, | |
| "step": 9240 | |
| }, | |
| { | |
| "epoch": 2.8593508500772797, | |
| "grad_norm": 0.0014065582072362304, | |
| "learning_rate": 4.688304997424008e-05, | |
| "loss": 0.8737, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 2.8624420401854715, | |
| "grad_norm": 0.001735221827402711, | |
| "learning_rate": 4.5852653271509534e-05, | |
| "loss": 0.8494, | |
| "step": 9260 | |
| }, | |
| { | |
| "epoch": 2.865533230293663, | |
| "grad_norm": 0.001790928072296083, | |
| "learning_rate": 4.4822256568778984e-05, | |
| "loss": 0.9272, | |
| "step": 9270 | |
| }, | |
| { | |
| "epoch": 2.8686244204018547, | |
| "grad_norm": 0.002074885880574584, | |
| "learning_rate": 4.379185986604843e-05, | |
| "loss": 0.8767, | |
| "step": 9280 | |
| }, | |
| { | |
| "epoch": 2.871715610510046, | |
| "grad_norm": 0.002597965532913804, | |
| "learning_rate": 4.276146316331788e-05, | |
| "loss": 0.9302, | |
| "step": 9290 | |
| }, | |
| { | |
| "epoch": 2.874806800618238, | |
| "grad_norm": 0.0019227894954383373, | |
| "learning_rate": 4.173106646058733e-05, | |
| "loss": 0.931, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 2.8778979907264297, | |
| "grad_norm": 0.0014659016160294414, | |
| "learning_rate": 4.070066975785677e-05, | |
| "loss": 0.944, | |
| "step": 9310 | |
| }, | |
| { | |
| "epoch": 2.8809891808346215, | |
| "grad_norm": 0.0017797836335375905, | |
| "learning_rate": 3.9670273055126224e-05, | |
| "loss": 0.95, | |
| "step": 9320 | |
| }, | |
| { | |
| "epoch": 2.884080370942813, | |
| "grad_norm": 0.0014417979400604963, | |
| "learning_rate": 3.8639876352395674e-05, | |
| "loss": 0.8956, | |
| "step": 9330 | |
| }, | |
| { | |
| "epoch": 2.8871715610510047, | |
| "grad_norm": 0.0013740757713094354, | |
| "learning_rate": 3.7609479649665125e-05, | |
| "loss": 0.9165, | |
| "step": 9340 | |
| }, | |
| { | |
| "epoch": 2.890262751159196, | |
| "grad_norm": 0.0016021078918129206, | |
| "learning_rate": 3.6579082946934575e-05, | |
| "loss": 0.9061, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 2.893353941267388, | |
| "grad_norm": 0.0017530877375975251, | |
| "learning_rate": 3.554868624420402e-05, | |
| "loss": 0.926, | |
| "step": 9360 | |
| }, | |
| { | |
| "epoch": 2.8964451313755797, | |
| "grad_norm": 0.0014526183949783444, | |
| "learning_rate": 3.451828954147346e-05, | |
| "loss": 0.8938, | |
| "step": 9370 | |
| }, | |
| { | |
| "epoch": 2.8995363214837715, | |
| "grad_norm": 0.001970699056982994, | |
| "learning_rate": 3.3487892838742914e-05, | |
| "loss": 0.8899, | |
| "step": 9380 | |
| }, | |
| { | |
| "epoch": 2.902627511591963, | |
| "grad_norm": 0.001455257530324161, | |
| "learning_rate": 3.2457496136012364e-05, | |
| "loss": 0.8998, | |
| "step": 9390 | |
| }, | |
| { | |
| "epoch": 2.9057187017001547, | |
| "grad_norm": 0.0023150176275521517, | |
| "learning_rate": 3.1427099433281815e-05, | |
| "loss": 0.8652, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 2.908809891808346, | |
| "grad_norm": 0.0021134279668331146, | |
| "learning_rate": 3.0396702730551262e-05, | |
| "loss": 0.8573, | |
| "step": 9410 | |
| }, | |
| { | |
| "epoch": 2.911901081916538, | |
| "grad_norm": 0.0016436997102573514, | |
| "learning_rate": 2.936630602782071e-05, | |
| "loss": 0.9263, | |
| "step": 9420 | |
| }, | |
| { | |
| "epoch": 2.9149922720247297, | |
| "grad_norm": 0.001375730847939849, | |
| "learning_rate": 2.833590932509016e-05, | |
| "loss": 0.8869, | |
| "step": 9430 | |
| }, | |
| { | |
| "epoch": 2.918083462132921, | |
| "grad_norm": 0.001517411321401596, | |
| "learning_rate": 2.730551262235961e-05, | |
| "loss": 0.9652, | |
| "step": 9440 | |
| }, | |
| { | |
| "epoch": 2.921174652241113, | |
| "grad_norm": 0.0015275340992957354, | |
| "learning_rate": 2.6275115919629058e-05, | |
| "loss": 0.8882, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 2.9242658423493046, | |
| "grad_norm": 0.0019407563377171755, | |
| "learning_rate": 2.5244719216898505e-05, | |
| "loss": 0.9207, | |
| "step": 9460 | |
| }, | |
| { | |
| "epoch": 2.927357032457496, | |
| "grad_norm": 0.0013504921225830913, | |
| "learning_rate": 2.4214322514167955e-05, | |
| "loss": 0.8926, | |
| "step": 9470 | |
| }, | |
| { | |
| "epoch": 2.930448222565688, | |
| "grad_norm": 0.0017280855681747198, | |
| "learning_rate": 2.3183925811437406e-05, | |
| "loss": 0.9399, | |
| "step": 9480 | |
| }, | |
| { | |
| "epoch": 2.9335394126738796, | |
| "grad_norm": 0.0019845685455948114, | |
| "learning_rate": 2.215352910870685e-05, | |
| "loss": 0.9159, | |
| "step": 9490 | |
| }, | |
| { | |
| "epoch": 2.936630602782071, | |
| "grad_norm": 0.001844555838033557, | |
| "learning_rate": 2.11231324059763e-05, | |
| "loss": 0.9431, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.939721792890263, | |
| "grad_norm": 0.0013761295704171062, | |
| "learning_rate": 2.009273570324575e-05, | |
| "loss": 0.9008, | |
| "step": 9510 | |
| }, | |
| { | |
| "epoch": 2.942812982998454, | |
| "grad_norm": 0.001978537067770958, | |
| "learning_rate": 1.90623390005152e-05, | |
| "loss": 0.949, | |
| "step": 9520 | |
| }, | |
| { | |
| "epoch": 2.945904173106646, | |
| "grad_norm": 0.001545691047795117, | |
| "learning_rate": 1.8031942297784645e-05, | |
| "loss": 0.9337, | |
| "step": 9530 | |
| }, | |
| { | |
| "epoch": 2.948995363214838, | |
| "grad_norm": 0.0017250186065211892, | |
| "learning_rate": 1.7001545595054096e-05, | |
| "loss": 0.8819, | |
| "step": 9540 | |
| }, | |
| { | |
| "epoch": 2.9520865533230296, | |
| "grad_norm": 0.0014475194038823247, | |
| "learning_rate": 1.5971148892323546e-05, | |
| "loss": 0.9038, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 2.955177743431221, | |
| "grad_norm": 0.0015622404171153903, | |
| "learning_rate": 1.4940752189592994e-05, | |
| "loss": 0.8629, | |
| "step": 9560 | |
| }, | |
| { | |
| "epoch": 2.958268933539413, | |
| "grad_norm": 0.00176430051214993, | |
| "learning_rate": 1.3910355486862443e-05, | |
| "loss": 0.9482, | |
| "step": 9570 | |
| }, | |
| { | |
| "epoch": 2.961360123647604, | |
| "grad_norm": 0.002005284419283271, | |
| "learning_rate": 1.287995878413189e-05, | |
| "loss": 0.9089, | |
| "step": 9580 | |
| }, | |
| { | |
| "epoch": 2.964451313755796, | |
| "grad_norm": 0.0017773432191461325, | |
| "learning_rate": 1.184956208140134e-05, | |
| "loss": 0.9071, | |
| "step": 9590 | |
| }, | |
| { | |
| "epoch": 2.9675425038639878, | |
| "grad_norm": 0.00213231542147696, | |
| "learning_rate": 1.0819165378670788e-05, | |
| "loss": 0.9035, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 2.970633693972179, | |
| "grad_norm": 0.002046185778453946, | |
| "learning_rate": 9.788768675940238e-06, | |
| "loss": 0.9451, | |
| "step": 9610 | |
| }, | |
| { | |
| "epoch": 2.973724884080371, | |
| "grad_norm": 0.001680073793977499, | |
| "learning_rate": 8.758371973209685e-06, | |
| "loss": 0.8838, | |
| "step": 9620 | |
| }, | |
| { | |
| "epoch": 2.9768160741885628, | |
| "grad_norm": 0.0015061446465551853, | |
| "learning_rate": 7.727975270479134e-06, | |
| "loss": 0.8764, | |
| "step": 9630 | |
| }, | |
| { | |
| "epoch": 2.979907264296754, | |
| "grad_norm": 0.0019097643671557307, | |
| "learning_rate": 6.697578567748584e-06, | |
| "loss": 0.9369, | |
| "step": 9640 | |
| }, | |
| { | |
| "epoch": 2.982998454404946, | |
| "grad_norm": 0.0017981400014832616, | |
| "learning_rate": 5.667181865018033e-06, | |
| "loss": 0.8498, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 2.9860896445131377, | |
| "grad_norm": 0.0021447227336466312, | |
| "learning_rate": 4.636785162287481e-06, | |
| "loss": 0.9168, | |
| "step": 9660 | |
| }, | |
| { | |
| "epoch": 2.989180834621329, | |
| "grad_norm": 0.0015040052821859717, | |
| "learning_rate": 3.6063884595569293e-06, | |
| "loss": 0.9233, | |
| "step": 9670 | |
| }, | |
| { | |
| "epoch": 2.992272024729521, | |
| "grad_norm": 0.00173336046282202, | |
| "learning_rate": 2.575991756826378e-06, | |
| "loss": 0.9224, | |
| "step": 9680 | |
| }, | |
| { | |
| "epoch": 2.9953632148377123, | |
| "grad_norm": 0.00168697745539248, | |
| "learning_rate": 1.5455950540958269e-06, | |
| "loss": 0.9131, | |
| "step": 9690 | |
| }, | |
| { | |
| "epoch": 2.998454404945904, | |
| "grad_norm": 0.0016643669223412871, | |
| "learning_rate": 5.151983513652757e-07, | |
| "loss": 0.945, | |
| "step": 9700 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 9705, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "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": 2.329354151166001e+18, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |