Update results.py

results.py

@@ -421,7 +421,6 @@ llama_graph2 = fig
 #tbd
 #llama graph 4
 
-
 # Data for different buckets and years
 data = {
     "1-1": {
@@ -470,6 +469,67 @@ llama_graph4 = fig_4
 
 ##llama graph 5
 
+# Data for different years and duplication counts
+data = {
+    "2014": {
+        "duplicates": list(range(1, 88)),
+        "perplexities": [9.73915022954687, 8.920626670584653, 10.794318768046299, 8.843588010546013, 9.050843813116494, 8.930966597941875, 9.037282952268393, 8.861595199889763, 8.753115155540607, 9.750853693230052, 8.684437196297443, 9.220661575409212, 8.779997997083122, 8.395973456433415, 8.73217254093509, 8.881588084417483, 8.631494252144279, 8.819106667105284, 8.896815931063438, 8.659420224734506, 9.093486198999384, 9.696607960142106, 8.508805626033137, 8.770987058526325, 8.618185592039293, 7.915860238031309, 8.83767692043796, 8.727961757725632, 8.485070840013236, 9.764445257512499, 8.600757313466264, 9.081950235300344, 8.644217818204663, 8.55864241676558, 8.233431060928043, 8.184825012856324, 8.457157826105203, 8.306222356999731, 8.804352979539, 8.313005637251672, 9.026721732161297, 8.449083429742776, 8.733235778798473, 9.006749183049818, 8.200737283467852, 8.251897579249375, 7.787355375938481, 7.268723383407269, 7.571889349513068, 8.351626749016013, 8.400107029097354, 7.723260069861529, 8.064236740258659, 7.888157433279341, 8.148191006636575, 7.906415374915906, 7.676551458219022, 7.967158452060384, 7.932665786459153, 7.9763750026305305, 7.667885913458702, 7.839381126838969, 7.92098160290061, 8.028601269205508, 7.563111341601447, 7.6678178481341215, 7.111641341466761, 7.42738813332767, 6.926910421359465, 7.363397686520664, 7.4639497513997775, 7.970748152252673, 7.356418997104175, 7.343504223513313, 6.8766667549967355, 6.358501202955649, 6.424115736059589, 6.008071046749611, 6.639372134895001, 5.6177395770356835, 5.443864799344331, 5.6647930543792, 4.172380654474099, 5.128083927608737, 5.146055825516793, 3.8998212782432287, 2.6866750639386963],
+    },
+    "2015": {
+        "duplicates": list(range(1, 88)),
+        "perplexities": [9.219127989173247, 8.662048439524266, 8.38989056455147, 7.962841520680627, 7.9129221891669, 8.601740522408205, 7.072072698161579, 8.547935092167322, 9.032202360967963, 9.644330594119374, 8.8352494084492, 9.04389212105191, 8.887379409385023, 8.467909602302054, 8.822876185378563, 8.655601459122618, 8.73446662907839, 8.858527575243034, 8.884133792160123, 8.489950876265588, 9.093447553065644, 9.577116330804195, 8.564448599529959, 8.758142067509203, 8.56227523869569, 7.897451823056177, 8.780353272984362, 8.69589044489135, 8.469444871399634, 9.707586741035607, 8.549040139041137, 9.062827976790162, 8.645009864743498, 8.544819434495203, 8.218203612587951, 8.191690418776773, 8.42249826138689, 8.296810827637307, 8.760096580513409, 8.3252124274653, 8.992877987118824, 8.454953276821902, 8.699916032688582, 8.953715077645272, 8.181603571788644, 8.193005628373307, 7.792425191978106, 7.2744004758877745, 7.554966996453596, 8.35034561406835, 8.41512904956328, 7.684576106312188, 8.064672849239695, 7.937226299628162, 8.236727165389356, 7.824011302908751, 7.622715782362401, 7.960707630747855, 7.932794820691164, 7.929912243495168, 7.655015402069077, 7.962802955194318, 7.851307904639461, 7.964993709642837, 7.551712280758451, 7.708432510205161, 7.1666588182507835, 7.3873387045347325, 6.911323642701545, 7.47032697252429, 7.5402879073116535, 7.826497044764086, 7.269655663988105, 7.343193140755126, 6.732955100033813, 6.447813049685099, 6.444597512230532, 6.007330197959276, 6.503017736395001, 5.297028390356502, 5.425460792319903, 5.536870445249572, 4.171727106880239, 5.112414745674864, 5.144884382794889, 3.914611382312465, 2.6866750639386963],
+    },
+    "2016": {
+        "duplicates": list(range(1, 88)),
+        "perplexities": [9.224890447512712, 8.865209727234555, 7.716537450318067, 8.505567000250219, 8.231186376526843, 8.439303045267778, 7.877163015312243, 7.994466878194956, 8.658354490436203, 8.525931733072895, 9.410755179427637, 9.700628031783868, 8.99451144975165, 9.002303200677753, 9.126870672453386, 9.422070510449645, 8.936682491853515, 8.911484811267863, 8.891667216351042, 8.666177883702614, 9.155910144590026, 9.594754908402185, 8.621298759427466, 8.787852330773674, 8.7227109238452, 7.982914897395895, 8.878126149070932, 8.717610707273684, 8.500066718763687, 9.77299952938655, 8.484499188420418, 9.006429013502562, 8.589528938847574, 8.48246092046138, 8.167057989146516, 8.171040607118929, 8.37171605234296, 8.266790595091017, 8.734798864250555, 8.29240112699872, 8.963302188522936, 8.41185193834277, 8.658957651571432, 8.923242538915662, 8.078070177365753, 8.153012171180952, 7.737730977556822, 7.323638159806731, 7.59168066096202, 8.37506434888842, 8.423868629202383, 7.671655504568603, 8.065075392658766, 7.976138380650688, 8.205116203468345, 7.81671515229199, 7.552886431000103, 7.904920126191346, 7.86434557317493, 7.888281162416497, 7.619817540814656, 7.956601658437982, 7.850774863461917, 8.024547071152215, 7.517340246604877, 7.638602768798686, 7.109799072356206, 7.351570828393637, 6.931128546796388, 7.506046773490761, 7.4664617315330455, 7.669725603465176, 7.254794269540855, 7.3180585905711455, 6.736605721220725, 6.380176442621056, 6.484391243201081, 6.037335383847396, 6.5047803196868665, 5.393419865823429, 5.382493790662867, 5.530611638476127, 4.195599613009973, 5.12749648121783, 5.162191190570913, 3.916322213479375, 2.6866750639387456],
+    },
+    "2017": {
+        "duplicates": list(range(1, 88)),
+        "perplexities":[9.488743166894329, 9.898479048180542, 8.696778620360424, 8.366687823872034, 8.630884650266044, 8.676086291852428, 9.334997107239593, 9.185879201546161, 8.84595356333466, 9.393437387426701, 9.45406621593655, 9.624060249862962, 9.532091483098174, 9.28384082428477, 9.469552273926178, 9.409055052262094, 9.358522040360171, 9.297106280140776, 9.073575169188139, 9.100944640377902, 9.23197269860482, 9.245904776150217, 9.170311706541447, 8.910824090510237, 8.795205662257203, 8.552457741951807, 8.930826813105282, 8.265717221808657, 8.603004747276188, 9.612646939458466, 8.552662222672813, 8.998548290935052, 8.560421893590284, 8.474723014427948, 8.206627804901014, 8.223373542290199, 8.398403355144557, 8.287572376736062, 8.754052702949243, 8.299825024291893, 8.941167619680243, 8.415160982913156, 8.638716545065352, 8.848352711284978, 8.027616325562285, 8.215417483840307, 7.796294849683444, 7.352280888696609, 7.632599613586875, 8.212116380094994, 8.36906213664436, 7.611956268590032, 7.998619733141829, 7.856761646868557, 8.045287904958364, 7.748316066259832, 7.501006529761084, 7.741538102662183, 7.751563576545041, 7.770809347580409, 7.433360335360308, 7.521045712972479, 7.342706611396849, 7.668010856653189, 7.324083777740233, 7.408132498092624, 7.120381253454176, 7.197819855700191, 6.845740773242076, 7.434903892374942, 7.30172469589484, 7.572324587463354, 7.2429522069242855, 7.407759480504424, 6.731089619105264, 6.270051935102005, 6.3655986521155326, 6.030460569178489, 6.450442179508357, 5.330140935634091, 5.408213974705514, 5.4956074785168365, 4.202359666343794, 5.132012988189369, 5.155812750486333, 3.917511891854019, 2.686675063938749],
+    },
+    "2018": {
+        "duplicates": list(range(1, 88)),
+        "perplexities": [9.896529727642656, 9.291082874617985, 9.691247942464267, 9.726042361988922, 9.023219233278645, 9.828083681879585, 9.061982292637602, 9.222594755383295, 9.221127620511943, 9.147737783035632, 9.217942333429908, 9.539190089929702, 9.468073594196243, 9.437273531098606, 9.435014041935224, 9.273790337174349, 9.396223055633108, 9.287408522624531, 9.106775224145558, 9.116191168902963, 9.237111779685268, 9.209051051549245, 9.222483610869865, 9.135002988610983, 8.788784301843895, 8.934854040474761, 9.017725654773155, 7.556551156032443, 8.798309646970585, 9.013026921998197, 8.454361664597716, 8.74055667139578, 8.44422407689182, 8.32983201978779, 8.359235182862202, 8.353476542703998, 8.197721584268855, 8.26450576475226, 8.595793902257883, 8.225557269937037, 8.826470933221817, 8.293081795411524, 8.570012186388858, 8.770785639469723, 7.88712952313622, 8.04581707470935, 7.626361664897192, 7.1029187216659935, 7.320944236203329, 8.070993423549334, 8.27396015964389, 7.3820727875281875, 7.903671423277116, 7.7618204686483825, 7.827655269209816, 7.675624293658499, 7.264953900477738, 7.583393879771943, 7.602901960669463, 7.671533291630901, 7.276738084631234, 7.3453361167238915, 7.135545835325713, 7.485342152740761, 7.167623169217046, 7.381420434659487, 7.025242147597371, 7.069891649575291, 6.69865442435757, 7.328644984573098, 7.258232298527563, 7.535910329008589, 7.255533113180425, 7.560918656956746, 6.666615954230938, 6.221938216559355, 6.336678407141162, 6.054298848202251, 6.548775617064581, 5.321650356255492, 5.4358231429446455, 5.421178189821786, 4.244486836711896, 5.109567298910753, 5.17074725626889, 3.9038325478573555, 2.6866750639387633],
+    },
+    "2019": {
+        "duplicates": list(range(1, 88)),
+        "perplexities": [9.32068228893817, 11.395142057563769, 9.418956531524783, 10.900691920426821, 9.438262555370176, 8.884239866713687, 9.062124035437813, 8.50267614928243, 8.638876645072413, 8.686925644350039, 8.714235065140539, 9.113035752580663, 9.125236586170647, 9.163963570975344, 9.25494911293901, 8.935379208998043, 9.225725827052665, 9.033303873966254, 8.950682642906997, 8.903920653639469, 8.911229320689435, 9.015000200687188, 9.096624316542014, 9.016863768067717, 8.737328799715891, 8.788723623351723, 8.932436150017006, 7.699853163934218, 8.707767111902166, 8.95646015823901, 8.35912690813594, 8.398373212501705, 7.932236470700611, 8.161242760113705, 8.609004455587062, 8.226779586405538, 7.842118333008395, 7.9881363454553, 8.036089351408874, 7.920609938080794, 7.95212600254153, 7.579485372600602, 7.88538668960003, 7.9594135744662236, 7.514800487806593, 7.689464139444159, 7.7544565651227, 7.41897302703282, 7.214254081876507, 7.895563781767841, 8.100226164186843, 7.272920541743749, 7.915118433849716, 7.658091861348199, 7.791626057064827, 7.627133141656945, 7.251831253315811, 7.569422424576574, 7.4906231710283935, 7.624606161129255, 7.152179142613762, 7.400538018976905, 6.9780334804483495, 7.455001719017696, 7.07739603554367, 7.201520909070617, 6.8033512129548726, 6.932923527827607, 6.796367425560507, 7.349719615399371, 7.290546797526088, 7.5224745941917055, 7.234885644016566, 7.373214375228053, 6.746192846100825, 6.23400120737243, 6.33988992018498, 6.030827322485673, 6.526425922427022, 5.371781592457229, 5.52171375482226, 5.521789795055974, 4.281913900699859, 5.135628510388279, 5.198507257987441, 3.9021738986547896, 2.6866750639387633],
+    },
+    "2020": {
+        "duplicates": list(range(1, 88)),
+        "perplexities": [10.012609650486192, 7.996780616269575, 8.486989186126861, 8.589723559690643, 8.39765295502809, 8.493985763693104, 8.444565685004276, 8.521057974766807, 8.262640776205831, 8.635134709906502, 8.811211245843253, 9.235066189908858, 9.022297426838923, 9.050058439072979, 9.021682564717057, 8.873493708246608, 9.015417293624877, 8.740954222830077, 9.0380883860745, 8.889091357432216, 8.912750014641842, 8.882441443817264, 8.998337842644444, 8.921169948677084, 8.691071558938896, 8.49208784075201, 8.776939087082413, 8.594979421555715, 8.72886934455455, 8.995731720491424, 8.421636184069852, 8.47617010726572, 7.749949346989389, 8.161269940208024, 8.744952692782391, 8.32025892064496, 7.869262480191432, 8.000847753164901, 8.076273972096198, 8.032342430082904, 7.983041009058366, 7.653597631636462, 7.975156835319908, 7.737411017432282, 7.605514142199969, 7.787824887059794, 7.857457978446249, 7.633203676037199, 7.047268036460168, 7.690311434532034, 8.040577048529848, 7.389586464023044, 7.740869008676299, 7.609659150626595, 7.837672584415724, 7.612754624743091, 7.287910254959312, 7.626199197144874, 7.319186118558612, 7.7127748347332385, 6.954623517457656, 7.486793070503545, 7.0394625786365, 7.450374675494852, 7.127267484316468, 7.250906351593999, 6.786363739265897, 6.980573512727148, 6.941463773809052, 7.423486191350063, 7.318743327632096, 7.629349084770016, 7.28811046001031, 7.435335626891362, 6.787682578185395, 6.222443482377916, 6.382107628208513, 5.9634307406816855, 6.574337724456447, 5.299637438768189, 5.445760462611284, 5.536805681860065, 4.174658965854665, 5.1170397769906595, 5.103551563888967, 3.9064523489117278, 2.6866750639387456],
+    },
+    "2021": {
+        "duplicates": list(range(1, 88)),
+        "perplexities": [9.841089195168713, 8.376002975517325, 8.380380483628153, 8.188986954093957, 8.498635277059357, 8.023999615729231, 8.382569297820478, 8.547323342280805, 8.188133428925276, 8.92217442481792, 9.176245490194058, 9.243284040091787, 9.034153905594078, 9.202098403867247, 9.134743272368441, 8.987006737157861, 8.92449319257734, 8.825574314893755, 8.847995893024844, 8.941295135829169, 8.987413373621635, 8.946047635293768, 8.910627453693433, 8.788838301728244, 8.76786062380576, 8.382956738298825, 8.699642493063092, 8.653055171884096, 8.740110731986059, 8.962375284828505, 8.333692265451546, 8.538371629422324, 7.61125418543177, 8.043172182416328, 8.716508021173402, 8.108788324737684, 7.9519597656706695, 8.013683771558808, 8.146173728648037, 8.096827091707578, 8.084160146852566, 7.625363299166386, 8.20296870216023, 7.82588401311041, 7.70124082508177, 7.78873594093791, 8.037661887779432, 7.5026482593453085, 7.087004259771732, 7.513718369391617, 8.138555082388203, 7.544675643417071, 7.6985222025131135, 7.620353449057761, 7.677585788165064, 7.607561443995849, 7.675688681631018, 7.426483327070772, 7.298729035123734, 7.597011886621879, 7.054853268704371, 7.2720630088080185, 6.999536763473811, 7.4649197409435155, 7.188519614053191, 7.05059392608445, 6.8880258512600125, 7.233553654860446, 6.863558927804786, 7.576343786143615, 7.169748797692396, 7.360833458217927, 7.151198016478686, 7.209746838078912, 6.785065344185751, 6.1445251557486005, 6.316811670053038, 5.861969568678405, 6.340724036900801, 5.198467430353175, 5.459134505462628, 5.462830131530051, 4.189086813428227, 5.114653776848574, 5.111679826185254, 3.920071733342082, 2.6866750639387456],
+    },
+    "2022": {
+        "duplicates": list(range(1, 88)),
+        "perplexities": [9.483592568856565, 7.19946566373024, 6.689435696789969, 7.735695023912755, 8.891994985807424, 7.974675072464074, 8.38103445140788, 7.767845780936468, 8.394625582114237, 8.433832557729854, 8.774679671114042, 8.974690045840946, 8.970427821588581, 9.259164010333704, 9.106573684091558, 9.082142024590542, 9.00694910652822, 8.951939452627803, 8.834410968583011, 8.976270078171723, 8.9788747911979, 9.147340358822175, 8.94766754786145, 8.863791307783835, 8.876434772786448, 8.537925507030351, 8.824632718943558, 8.670890722043863, 8.713248499915446, 8.991496303395792, 8.497388391287544, 8.694429166461068, 7.660306866593211, 8.063103914688337, 8.463798472858477, 8.147225054222934, 7.98918792448372, 8.129552622440363, 8.156552658413169, 7.993772322199681, 8.077300760319622, 7.687003596840271, 8.216734207325837, 7.818779874173861, 7.784873713139422, 7.841502120166083, 8.009056410739252, 7.643507378655463, 7.027652318922246, 7.578018517117392, 8.21806831598481, 7.738109238050354, 7.657938120144026, 7.670918897910402, 7.789453032753468, 7.6784988295976015, 7.758997892938832, 7.354548111787801, 7.324563987410127, 7.542453103467507, 7.082203068171782, 7.41432116627747, 6.902549671764324, 7.489234680249967, 7.354679134462809, 6.977235447939879, 7.297608588934183, 7.439224077591953, 7.01074155919995, 7.77157588876759, 7.452660719169145, 7.206525557457038, 6.837175958650314, 6.816276605521482, 6.841636273605858, 6.4126962333247475, 6.544729510231483, 5.84044063139556, 6.384392701668377, 5.342064638285962, 5.449806194943077, 5.550638967579051, 4.188085941944831, 5.167854549955437, 5.137630179031516, 3.901881866936028, 2.686675063938701],
+    },
+    "2023": {
+        "duplicates": list(range(1, 88)),
+        "perplexities": 8.70489099557147, 7.580704259900713, 5.836272117595713, 7.267518477506005, 7.283085974734155, 5.803062744865375, 8.134525344019872, 7.557124524037357, 8.285446726695811, 8.351123312334009, 8.590647921692078, 9.005573576795152, 8.871116907846343, 9.192666716497001, 9.005592946042935, 9.143088639505718, 9.091899749285021, 8.968510745275767, 8.975056578670769, 9.091779598794488, 8.980027479169559, 9.159988763765217, 8.883393229479356, 8.874910963718058, 8.931427901889217, 8.625523963019026, 8.735000607943736, 8.63742312596692, 8.70111762921543, 9.206595504882097, 8.763642440521853, 8.911783192379705, 7.634828216241755, 8.259433628637158, 8.82001651315211, 8.107726825798297, 8.206921799311043, 8.355228813696131, 8.298238657805719, 8.033613885206083, 8.225747435920205, 7.934196000879481, 8.286077333650876, 7.980486854114626, 7.721311081423362, 7.689565789535576, 7.790611228082437, 7.624460645122273, 7.0093123271820055, 7.791169711054489, 8.14175443228974, 7.776459425768002, 7.526576191153727, 7.756713091828537, 7.801566191856336, 7.843167604084829, 7.7089466527367705, 7.409181212687685, 7.442037119594963, 7.337342474375209, 7.190330728288846, 7.254100247522111, 6.79586626781295, 7.483337535174346, 7.294262214559599, 7.030373799134058, 7.149000704995256, 7.324234704301861, 6.822226914595191, 7.792044674564248, 7.149607692275065, 7.078295043693808, 6.946740616609221, 6.700263270499871, 6.4699642011179925, 6.359739677958311, 6.494171402228005, 5.966063278273302, 6.15136761591695, 5.309723356734837, 5.00827796373236, 4.976659481606136, 4.042120043465434, 4.842002060938444, 5.114027203055052, 3.8900363946754233, 2.6866750639387074],
+    },
+}
+
+# Create figure
+fig = go.Figure()
+
+# Add traces for each year
+for year, year_data in data.items():
+    fig.add_trace(go.Scatter(x=year_data["duplicates"], y=year_data["perplexities"], mode='lines+markers', name=year))
+
+# Update layout
+fig.update_layout(
+    title="Perplexity Across Different Dump Duplication Counts (Local)",
+    xaxis_title="Number of Dumps Duplication",
+    yaxis_title="Average Perplexity",
+    legend_title="Year"
+)
+
+# Show the figure
+llama_graph5 = fig
 
 
 intro_div = Div(