You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

2-mlp_bp.ipynb 525 kB

6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279228022812282228322842285228622872288228922902291229222932294229522962297229822992300230123022303230423052306230723082309231023112312231323142315231623172318231923202321232223232324232523262327232823292330233123322333233423352336233723382339234023412342234323442345234623472348234923502351235223532354235523562357235823592360236123622363236423652366236723682369237023712372237323742375237623772378237923802381238223832384238523862387238823892390239123922393239423952396239723982399240024012402240324042405240624072408240924102411241224132414241524162417241824192420242124222423242424252426242724282429243024312432243324342435243624372438243924402441244224432444244524462447244824492450245124522453245424552456245724582459246024612462246324642465246624672468246924702471247224732474247524762477247824792480248124822483248424852486248724882489249024912492249324942495249624972498249925002501250225032504250525062507250825092510251125122513251425152516251725182519252025212522252325242525252625272528252925302531253225332534253525362537253825392540254125422543254425452546254725482549255025512552255325542555255625572558255925602561256225632564256525662567256825692570257125722573257425752576257725782579258025812582258325842585258625872588258925902591259225932594259525962597259825992600260126022603260426052606260726082609261026112612261326142615261626172618261926202621262226232624262526262627262826292630263126322633263426352636263726382639264026412642264326442645264626472648264926502651265226532654265526562657265826592660266126622663266426652666266726682669267026712672267326742675267626772678267926802681268226832684268526862687268826892690269126922693269426952696269726982699270027012702270327042705270627072708270927102711271227132714271527162717271827192720272127222723272427252726272727282729273027312732273327342735273627372738273927402741274227432744274527462747274827492750275127522753275427552756275727582759276027612762276327642765276627672768276927702771277227732774277527762777277827792780278127822783278427852786278727882789279027912792279327942795279627972798279928002801280228032804280528062807280828092810281128122813281428152816281728182819282028212822282328242825282628272828282928302831283228332834283528362837283828392840284128422843284428452846284728482849285028512852285328542855285628572858285928602861286228632864286528662867286828692870287128722873287428752876287728782879288028812882288328842885288628872888288928902891289228932894289528962897289828992900290129022903290429052906290729082909291029112912291329142915291629172918291929202921292229232924292529262927292829292930293129322933293429352936293729382939294029412942294329442945294629472948294929502951295229532954295529562957295829592960296129622963296429652966296729682969297029712972297329742975297629772978297929802981298229832984298529862987298829892990299129922993299429952996299729982999300030013002300330043005300630073008300930103011301230133014301530163017301830193020302130223023302430253026302730283029303030313032303330343035303630373038303930403041304230433044304530463047304830493050305130523053305430553056305730583059306030613062306330643065306630673068306930703071307230733074307530763077307830793080308130823083308430853086308730883089309030913092309330943095309630973098309931003101310231033104310531063107310831093110311131123113311431153116311731183119312031213122312331243125312631273128312931303131313231333134313531363137313831393140314131423143314431453146314731483149315031513152315331543155315631573158315931603161316231633164316531663167316831693170317131723173317431753176317731783179318031813182318331843185318631873188318931903191319231933194319531963197319831993200320132023203320432053206320732083209321032113212321332143215321632173218321932203221322232233224322532263227322832293230323132323233323432353236323732383239324032413242324332443245324632473248324932503251325232533254325532563257325832593260326132623263326432653266326732683269327032713272327332743275327632773278327932803281328232833284328532863287328832893290329132923293329432953296329732983299330033013302330333043305330633073308330933103311331233133314331533163317331833193320332133223323332433253326332733283329333033313332333333343335333633373338333933403341334233433344334533463347334833493350335133523353335433553356335733583359336033613362336333643365336633673368336933703371337233733374337533763377337833793380338133823383338433853386338733883389339033913392339333943395339633973398339934003401340234033404340534063407340834093410341134123413341434153416341734183419342034213422342334243425342634273428342934303431343234333434343534363437343834393440344134423443344434453446344734483449345034513452345334543455345634573458345934603461346234633464346534663467346834693470347134723473347434753476347734783479348034813482348334843485348634873488348934903491349234933494349534963497349834993500350135023503350435053506350735083509351035113512351335143515351635173518351935203521352235233524352535263527352835293530353135323533353435353536353735383539354035413542354335443545354635473548354935503551355235533554355535563557355835593560356135623563356435653566356735683569357035713572357335743575357635773578357935803581358235833584358535863587358835893590359135923593359435953596359735983599360036013602360336043605360636073608360936103611361236133614361536163617361836193620362136223623362436253626362736283629363036313632363336343635363636373638363936403641364236433644364536463647364836493650365136523653365436553656365736583659366036613662366336643665366636673668366936703671367236733674367536763677367836793680368136823683368436853686368736883689369036913692369336943695369636973698369937003701370237033704370537063707370837093710371137123713371437153716371737183719372037213722372337243725372637273728372937303731373237333734373537363737373837393740374137423743374437453746374737483749375037513752375337543755375637573758375937603761376237633764376537663767376837693770377137723773377437753776377737783779378037813782378337843785378637873788378937903791379237933794379537963797379837993800380138023803380438053806380738083809381038113812381338143815381638173818381938203821382238233824382538263827382838293830383138323833383438353836383738383839384038413842384338443845384638473848384938503851385238533854385538563857385838593860386138623863386438653866386738683869387038713872387338743875387638773878387938803881388238833884388538863887388838893890389138923893389438953896389738983899390039013902390339043905390639073908390939103911391239133914391539163917391839193920392139223923392439253926392739283929393039313932393339343935393639373938393939403941394239433944394539463947394839493950395139523953395439553956395739583959396039613962396339643965396639673968396939703971397239733974397539763977397839793980398139823983398439853986398739883989399039913992399339943995399639973998399940004001400240034004400540064007400840094010401140124013401440154016401740184019402040214022402340244025402640274028402940304031403240334034403540364037403840394040404140424043404440454046404740484049405040514052405340544055405640574058405940604061406240634064406540664067406840694070407140724073407440754076407740784079408040814082408340844085408640874088408940904091409240934094409540964097409840994100410141024103410441054106410741084109411041114112411341144115411641174118411941204121412241234124412541264127412841294130413141324133413441354136413741384139414041414142414341444145414641474148414941504151415241534154415541564157415841594160416141624163416441654166416741684169417041714172417341744175417641774178417941804181418241834184418541864187418841894190419141924193419441954196419741984199420042014202420342044205420642074208420942104211421242134214421542164217421842194220422142224223422442254226422742284229423042314232423342344235423642374238423942404241424242434244424542464247424842494250425142524253425442554256425742584259426042614262426342644265426642674268426942704271427242734274427542764277427842794280428142824283428442854286428742884289429042914292429342944295429642974298429943004301430243034304430543064307430843094310431143124313431443154316431743184319432043214322432343244325432643274328432943304331433243334334433543364337433843394340434143424343434443454346434743484349435043514352435343544355435643574358435943604361436243634364436543664367436843694370437143724373437443754376437743784379438043814382438343844385438643874388438943904391439243934394439543964397439843994400440144024403440444054406440744084409441044114412441344144415441644174418441944204421442244234424442544264427442844294430443144324433443444354436443744384439444044414442444344444445444644474448444944504451445244534454445544564457445844594460446144624463446444654466446744684469447044714472447344744475447644774478447944804481448244834484448544864487448844894490449144924493449444954496449744984499450045014502450345044505450645074508450945104511451245134514451545164517451845194520452145224523452445254526452745284529453045314532453345344535453645374538453945404541454245434544454545464547454845494550455145524553455445554556455745584559456045614562456345644565456645674568456945704571457245734574457545764577457845794580458145824583458445854586458745884589459045914592459345944595459645974598459946004601460246034604460546064607460846094610461146124613461446154616461746184619462046214622462346244625462646274628462946304631463246334634463546364637463846394640464146424643464446454646464746484649465046514652465346544655465646574658465946604661466246634664466546664667466846694670467146724673467446754676467746784679468046814682468346844685468646874688468946904691469246934694469546964697469846994700470147024703470447054706470747084709471047114712471347144715471647174718471947204721472247234724472547264727472847294730473147324733473447354736473747384739474047414742474347444745474647474748474947504751475247534754475547564757475847594760476147624763476447654766476747684769477047714772477347744775477647774778477947804781478247834784478547864787478847894790479147924793479447954796479747984799480048014802480348044805480648074808480948104811481248134814481548164817481848194820482148224823482448254826482748284829483048314832483348344835483648374838483948404841484248434844484548464847484848494850485148524853485448554856485748584859486048614862486348644865486648674868486948704871487248734874487548764877487848794880488148824883488448854886488748884889489048914892489348944895489648974898489949004901490249034904490549064907490849094910491149124913491449154916491749184919492049214922492349244925492649274928492949304931493249334934493549364937493849394940494149424943494449454946494749484949
  1. {
  2. "cells": [
  3. {
  4. "cell_type": "markdown",
  5. "metadata": {},
  6. "source": [
  7. "# 多层神经网络和反向传播\n"
  8. ]
  9. },
  10. {
  11. "cell_type": "markdown",
  12. "metadata": {},
  13. "source": [
  14. "## 1. 神经元\n",
  15. "\n",
  16. "神经元和感知器本质上是一样的,只不过我们说感知器的时候,它的激活函数是阶跃函数;而当我们说神经元时,激活函数往往选择为sigmoid函数或tanh函数。如下图所示:\n",
  17. "\n",
  18. "![neuron](images/neuron.gif)\n",
  19. "\n",
  20. "计算一个神经元的输出的方法和计算一个感知器的输出是一样的。假设神经元的输入是向量$\\vec{x}$,权重向量是$\\vec{w}$(偏置项是$w_0$),激活函数是sigmoid函数,则其输出y:\n",
  21. "$$\n",
  22. "y = sigmod(\\vec{w}^T \\cdot \\vec{x})\n",
  23. "$$\n",
  24. "\n",
  25. "sigmoid函数的定义如下:\n",
  26. "$$\n",
  27. "sigmod(x) = \\frac{1}{1+e^{-x}}\n",
  28. "$$\n",
  29. "将其带入前面的式子,得到\n",
  30. "$$\n",
  31. "y = \\frac{1}{1+e^{-\\vec{w}^T \\cdot \\vec{x}}}\n",
  32. "$$\n",
  33. "\n",
  34. "sigmoid函数是一个非线性函数,值域是(0,1)。函数图像如下图所示\n",
  35. "\n",
  36. "![sigmod_function](images/sigmod.jpg)\n",
  37. "\n",
  38. "sigmoid函数的导数是:\n",
  39. "\\begin{eqnarray}\n",
  40. "y & = & sigmod(x) \\tag{1} \\\\\n",
  41. "y' & = & y(1-y)\n",
  42. "\\end{eqnarray}\n",
  43. "\n",
  44. "可以看到,sigmoid函数的导数非常有趣,它可以用sigmoid函数自身来表示。这样,一旦计算出sigmoid函数的值,计算它的导数的值就非常方便。\n",
  45. "\n"
  46. ]
  47. },
  48. {
  49. "cell_type": "markdown",
  50. "metadata": {},
  51. "source": [
  52. "## 2. 神经网络是啥?\n",
  53. "\n",
  54. "![nn1](images/nn1.jpeg)\n",
  55. "\n",
  56. "神经网络其实就是按照一定规则连接起来的多个神经元。上图展示了一个全连接(full connected, FC)神经网络,通过观察上面的图,我们可以发现它的规则包括:\n",
  57. "\n",
  58. "* 神经元按照层来布局。最左边的层叫做输入层,负责接收输入数据;最右边的层叫输出层,我们可以从这层获取神经网络输出数据。输入层和输出层之间的层叫做隐藏层,因为它们对于外部来说是不可见的。\n",
  59. "* 同一层的神经元之间没有连接。\n",
  60. "* 第N层的每个神经元和第N-1层的所有神经元相连(这就是full connected的含义),第N-1层神经元的输出就是第N层神经元的输入。\n",
  61. "* 每个连接都有一个权值。\n",
  62. "\n",
  63. "上面这些规则定义了全连接神经网络的结构。事实上还存在很多其它结构的神经网络,比如卷积神经网络(CNN)、循环神经网络(RNN),他们都具有不同的连接规则。\n"
  64. ]
  65. },
  66. {
  67. "cell_type": "markdown",
  68. "metadata": {},
  69. "source": [
  70. "## 3. 计算神经网络的输出\n",
  71. "\n",
  72. "神经网络实际上就是一个输入向量$\\vec{x}$到输出向量$\\vec{y}$的函数,即:\n",
  73. "\n",
  74. "$$\n",
  75. "\\vec{y} = f_{network}(\\vec{x})\n",
  76. "$$\n",
  77. "根据输入计算神经网络的输出,需要首先将输入向量$\\vec{x}$的每个元素的值$x_i$赋给神经网络的输入层的对应神经元,然后根据式1依次向前计算每一层的每个神经元的值,直到最后一层输出层的所有神经元的值计算完毕。最后,将输出层每个神经元的值串在一起就得到了输出向量$\\vec{y}$。\n",
  78. "\n",
  79. "接下来举一个例子来说明这个过程,我们先给神经网络的每个单元写上编号。\n",
  80. "\n",
  81. "![nn2](images/nn2.png)\n",
  82. "\n",
  83. "如上图,输入层有三个节点,我们将其依次编号为1、2、3;隐藏层的4个节点,编号依次为4、5、6、7;最后输出层的两个节点编号为8、9。因为我们这个神经网络是全连接网络,所以可以看到每个节点都和上一层的所有节点有连接。比如,我们可以看到隐藏层的节点4,它和输入层的三个节点1、2、3之间都有连接,其连接上的权重分别为$w_{41}$,$w_{42}$,$w_{43}$。那么,我们怎样计算节点4的输出值$a_4$呢?\n",
  84. "\n",
  85. "\n",
  86. "为了计算节点4的输出值,我们必须先得到其所有上游节点(也就是节点1、2、3)的输出值。节点1、2、3是输入层的节点,所以,他们的输出值就是输入向量$\\vec{x}$本身。按照上图画出的对应关系,可以看到节点1、2、3的输出值分别是$x_1$,$x_2$,$x_3$。我们要求输入向量的维度和输入层神经元个数相同,而输入向量的某个元素对应到哪个输入节点是可以自由决定的,你偏非要把$x_1$赋值给节点2也是完全没有问题的,但这样除了把自己弄晕之外,并没有什么价值。\n",
  87. "\n",
  88. "一旦我们有了节点1、2、3的输出值,我们就可以根据式1计算节点4的输出值$a_4$:\n",
  89. "\n",
  90. "![eqn_3_4](images/eqn_3_4.png)\n",
  91. "\n",
  92. "上式的$w_{4b}$是节点4的偏置项,图中没有画出来。而$w_{41}$,$w_{42}$,$w_{43}$分别为节点1、2、3到节点4连接的权重,在给权重$w_{ji}$编号时,我们把目标节点的编号$j$放在前面,把源节点的编号$i$放在后面。\n",
  93. "\n",
  94. "同样,我们可以继续计算出节点5、6、7的输出值$a_5$,$a_6$,$a_7$。这样,隐藏层的4个节点的输出值就计算完成了,我们就可以接着计算输出层的节点8的输出值$y_1$:\n",
  95. "\n",
  96. "![eqn_5_6](images/eqn_5_6.png)\n",
  97. "\n",
  98. "同理,我们还可以计算出$y_2$的值。这样输出层所有节点的输出值计算完毕,我们就得到了在输入向量$\\vec{x} = (x_1, x_2, x_3)^T$时,神经网络的输出向量$\\vec{y} = (y_1, y_2)^T$。这里我们也看到,输出向量的维度和输出层神经元个数相同。\n",
  99. "\n"
  100. ]
  101. },
  102. {
  103. "cell_type": "markdown",
  104. "metadata": {},
  105. "source": [
  106. "## 4. 神经网络的矩阵表示\n",
  107. "\n",
  108. "神经网络的计算如果用矩阵来表示会很方便(当然逼格也更高),我们先来看看隐藏层的矩阵表示。\n",
  109. "\n",
  110. "首先我们把隐藏层4个节点的计算依次排列出来:\n",
  111. "\n",
  112. "![eqn_hidden_units](images/eqn_hidden_units.png)\n",
  113. "\n",
  114. "接着,定义网络的输入向量$\\vec{x}$和隐藏层每个节点的权重向量$\\vec{w}$。令\n",
  115. "\n",
  116. "![eqn_7_12](images/eqn_7_12.png)\n",
  117. "\n",
  118. "代入到前面的一组式子,得到:\n",
  119. "\n",
  120. "![eqn_13_16](images/eqn_13_16.png)\n",
  121. "\n",
  122. "现在,我们把上述计算$a_4$, $a_5$,$a_6$,$a_7$的四个式子写到一个矩阵里面,每个式子作为矩阵的一行,就可以利用矩阵来表示它们的计算了。令\n",
  123. "\n",
  124. "![eqn_matrix1](images/eqn_matrix1.png)\n",
  125. "\n",
  126. "带入前面的一组式子,得到\n",
  127. "\n",
  128. "![formular_2](images/formular_2.png)\n",
  129. "\n",
  130. "在式2中,$f$是激活函数,在本例中是$sigmod$函数;$W$是某一层的权重矩阵;$\\vec{x}$是某层的输入向量;$\\vec{a}$是某层的输出向量。式2说明神经网络的每一层的作用实际上就是先将输入向量左乘一个数组进行线性变换,得到一个新的向量,然后再对这个向量逐元素应用一个激活函数。\n",
  131. "\n",
  132. "每一层的算法都是一样的。比如,对于包含一个输入层,一个输出层和三个隐藏层的神经网络,我们假设其权重矩阵分别为$W_1$,$W_2$,$W_3$,$W_4$,每个隐藏层的输出分别是$\\vec{a}_1$,$\\vec{a}_2$,$\\vec{a}_3$,神经网络的输入为$\\vec{x}$,神经网络的输出为$\\vec{y}$,如下图所示:\n",
  133. "\n",
  134. "![nn_parameters_demo](images/nn_parameters_demo.png)\n",
  135. "\n",
  136. "则每一层的输出向量的计算可以表示为:\n",
  137. "\n",
  138. "![eqn_17_20](images/eqn_17_20.png)\n",
  139. "\n",
  140. "\n",
  141. "这就是神经网络输出值的矩阵计算方法。\n",
  142. "\n",
  143. "如果写成一个公式:\n",
  144. "$$\n",
  145. "\\vec{y} = f(W4 \\cdot f(W3 \\cdot f(W2 \\cdot f(W1 \\cdot \\vec{x}))))\n",
  146. "$$"
  147. ]
  148. },
  149. {
  150. "cell_type": "markdown",
  151. "metadata": {},
  152. "source": [
  153. "\n",
  154. "神经网络正向计算的过程比较简单,就是一层一层不断做运算就可以了,动态的演示如下图所示:\n",
  155. "![](images/nn-forward.gif)"
  156. ]
  157. },
  158. {
  159. "cell_type": "markdown",
  160. "metadata": {},
  161. "source": [
  162. "## 5. 神经网络的训练 - 反向传播算法\n",
  163. "\n",
  164. "现在,我们需要知道一个神经网络的每个连接上的权值是如何得到的。我们可以说神经网络是一个模型,那么这些权值就是模型的参数,也就是模型要学习的东西。然而,一个神经网络的连接方式、网络的层数、每层的节点数这些参数,则不是学习出来的,而是人为事先设置的。对于这些人为设置的参数,我们称之为超参数(Hyper-Parameters)。\n",
  165. "\n",
  166. "反向传播算法其实就是链式求导法则的应用。然而,这个如此简单且显而易见的方法,却是在Roseblatt提出感知器算法将近30年之后才被发明和普及的。对此,Bengio这样回应道:\n",
  167. "\n",
  168. "> 很多看似显而易见的想法只有在事后才变得显而易见。\n",
  169. "\n",
  170. "按照机器学习的通用套路,我们先确定神经网络的目标函数,然后用随机梯度下降优化算法去求目标函数最小值时的参数值。\n",
  171. "\n",
  172. "我们取网络所有输出层节点的误差平方和作为目标函数:\n",
  173. "\n",
  174. "![bp_loss](images/bp_loss.png)\n",
  175. "\n",
  176. "其中,$E_d$表示是样本$d$的误差, **t是样本的标签值**,**y是神经网络的输出值**。\n",
  177. "\n",
  178. "然后,使用随机梯度下降算法对目标函数进行优化:\n",
  179. "\n",
  180. "![bp_weight_update](images/bp_weight_update.png)\n",
  181. "\n",
  182. "随机梯度下降算法也就是需要求出误差$E_d$对于每个权重$w_{ji}$的偏导数(也就是梯度),怎么求呢?\n",
  183. "\n",
  184. "![nn3](images/nn3.png)\n",
  185. "\n",
  186. "观察上图,我们发现权重$w_{ji}$仅能通过影响节点$j$的输入值影响网络的其它部分,设$net_j$是节点$j$的加权输入,即\n",
  187. "\n",
  188. "![eqn_21_22](images/eqn_21_22.png)\n",
  189. "\n",
  190. "$E_d$是$net_j$的函数,而$net_j$是$w_{ji}$的函数。根据链式求导法则,可以得到:\n",
  191. "\n",
  192. "![eqn_23_25](images/eqn_23_25.png)\n",
  193. "\n",
  194. "\n",
  195. "上式中,$x_{ji}$是节点传递给节点$j$的输入值,也就是节点$i$的输出值。\n",
  196. "\n",
  197. "对于的$\\frac{\\partial E_d}{\\partial net_j}$推导,需要区分输出层和隐藏层两种情况。\n",
  198. "\n"
  199. ]
  200. },
  201. {
  202. "cell_type": "markdown",
  203. "metadata": {},
  204. "source": [
  205. "### 5.1 输出层权值训练\n",
  206. "\n",
  207. "![nn3](images/nn3.png)\n",
  208. "\n",
  209. "对于输出层来说,$net_j$仅能通过节点$j$的输出值$y_j$来影响网络其它部分,也就是说$E_d$是$y_j$的函数,而$y_j$是$net_j$的函数,其中$y_j = sigmod(net_j)$。所以我们可以再次使用链式求导法则:\n",
  210. "\n",
  211. "![eqn_26](images/eqn_26.png)\n",
  212. "\n",
  213. "考虑上式第一项:\n",
  214. "\n",
  215. "![eqn_27_29](images/eqn_27_29.png)\n",
  216. "\n",
  217. "\n",
  218. "考虑上式第二项:\n",
  219. "\n",
  220. "![eqn_30_31](images/eqn_30_31.png)\n",
  221. "\n",
  222. "将第一项和第二项带入,得到:\n",
  223. "\n",
  224. "![eqn_ed_net_j.png](images/eqn_ed_net_j.png)\n",
  225. "\n",
  226. "如果令$\\delta_j = - \\frac{\\partial E_d}{\\partial net_j}$,也就是一个节点的误差项$\\delta$是网络误差对这个节点输入的偏导数的相反数。带入上式,得到:\n",
  227. "\n",
  228. "![eqn_delta_j.png](images/eqn_delta_j.png)\n",
  229. "\n",
  230. "将上述推导带入随机梯度下降公式,得到:\n",
  231. "\n",
  232. "![eqn_32_34.png](images/eqn_32_34.png)\n"
  233. ]
  234. },
  235. {
  236. "cell_type": "markdown",
  237. "metadata": {},
  238. "source": [
  239. "### 5.2 隐藏层权值训练\n",
  240. "\n",
  241. "现在我们要推导出隐藏层的$\\frac{\\partial E_d}{\\partial net_j}$。\n",
  242. "\n",
  243. "![nn3](images/nn3.png)\n",
  244. "\n",
  245. "首先,我们需要定义节点$j$的所有直接下游节点的集合$Downstream(j)$。例如,对于节点4来说,它的直接下游节点是节点8、节点9。可以看到$net_j$只能通过影响$Downstream(j)$再影响$E_d$。设$net_k$是节点$j$的下游节点的输入,则$E_d$是$net_k$的函数,而$net_k$是$net_j$的函数。因为$net_k$有多个,我们应用全导数公式,可以做出如下推导:\n",
  246. "\n",
  247. "![eqn_35_40](images/eqn_35_40.png)\n",
  248. "\n",
  249. "因为$\\delta_j = - \\frac{\\partial E_d}{\\partial net_j}$,带入上式得到:\n",
  250. "\n",
  251. "![eqn_delta_hidden.png](images/eqn_delta_hidden.png)\n",
  252. "\n",
  253. "\n",
  254. "至此,我们已经推导出了反向传播算法。需要注意的是,我们刚刚推导出的训练规则是根据激活函数是sigmoid函数、平方和误差、全连接网络、随机梯度下降优化算法。如果激活函数不同、误差计算方式不同、网络连接结构不同、优化算法不同,则具体的训练规则也会不一样。但是无论怎样,训练规则的推导方式都是一样的,应用链式求导法则进行推导即可。\n"
  255. ]
  256. },
  257. {
  258. "cell_type": "markdown",
  259. "metadata": {},
  260. "source": [
  261. "### 5.3 具体解释\n",
  262. "\n",
  263. "我们假设每个训练样本为$(\\vec{x}, \\vec{t})$,其中向量$\\vec{x}$是训练样本的特征,而$\\vec{t}$是样本的目标值。\n",
  264. "\n",
  265. "![nn3](images/nn3.png)\n",
  266. "\n",
  267. "首先,我们根据上一节介绍的算法,用样本的特征$\\vec{x}$,计算出神经网络中每个隐藏层节点的输出$a_i$,以及输出层每个节点的输出$y_i$。\n",
  268. "\n",
  269. "然后,我们按照下面的方法计算出每个节点的误差项$\\delta_i$:\n",
  270. "\n",
  271. "* **对于输出层节点$i$**\n",
  272. "\n",
  273. "![formular_3.png](images/formular_3.png)\n",
  274. "\n",
  275. "其中,$\\delta_i$是节点$i$的误差项,$y_i$是节点$i$的输出值,$t_i$是样本对应于节点$i$的目标值。举个例子,根据上图,对于输出层节点8来说,它的输出值是$y_1$,而样本的目标值是$t_1$,带入上面的公式得到节点8的误差项应该是:\n",
  276. "\n",
  277. "![forumlar_delta8.png](images/forumlar_delta8.png)\n",
  278. "\n",
  279. "* **对于隐藏层节点**\n",
  280. "\n",
  281. "![formular_4.png](images/formular_4.png)\n",
  282. "\n",
  283. "其中,$a_i$是节点$i$的输出值,$w_{ki}$是节点$i$到它的下一层节点$k$的连接的权重,$\\delta_k$是节点$i$的下一层节点$k$的误差项。例如,对于隐藏层节点4来说,计算方法如下:\n",
  284. "\n",
  285. "![forumlar_delta4.png](images/forumlar_delta4.png)\n",
  286. "\n",
  287. "\n",
  288. "\n",
  289. "最后,更新每个连接上的权值:\n",
  290. "\n",
  291. "![formular_5.png](images/formular_5.png)\n",
  292. "\n",
  293. "其中,$w_{ji}$是节点$i$到节点$j$的权重,$\\eta$是一个成为学习速率的常数,$\\delta_j$是节点$j$的误差项,$x_{ji}$是节点$i$传递给节点$j$的输入。例如,权重$w_{84}$的更新方法如下:\n",
  294. "\n",
  295. "![eqn_w84_update.png](images/eqn_w84_update.png)\n",
  296. "\n",
  297. "类似的,权重$w_{41}$的更新方法如下:\n",
  298. "\n",
  299. "![eqn_w41_update.png](images/eqn_w41_update.png)\n",
  300. "\n",
  301. "\n",
  302. "偏置项的输入值永远为1。例如,节点4的偏置项$w_{4b}$应该按照下面的方法计算:\n",
  303. "\n",
  304. "![eqn_w4b_update.png](images/eqn_w4b_update.png)\n",
  305. "\n",
  306. "我们已经介绍了神经网络每个节点误差项的计算和权重更新方法。显然,计算一个节点的误差项,需要先计算每个与其相连的下一层节点的误差项。这就要求误差项的计算顺序必须是从输出层开始,然后反向依次计算每个隐藏层的误差项,直到与输入层相连的那个隐藏层。这就是反向传播算法的名字的含义。当所有节点的误差项计算完毕后,我们就可以根据式5来更新所有的权重。\n",
  307. "\n"
  308. ]
  309. },
  310. {
  311. "cell_type": "markdown",
  312. "metadata": {},
  313. "source": [
  314. "## 6. 为什么要使用激活函数\n",
  315. "激活函数在神经网络中非常重要,使用激活函数也是非常必要的,前面我们从人脑神经元的角度理解了激活函数,因为神经元需要通过激活才能往后传播,所以神经网络中需要激活函数,下面我们从数学的角度理解一下激活函数的必要性。\n",
  316. "\n",
  317. "比如一个两层的神经网络,使用 A 表示激活函数,那么\n",
  318. "\n",
  319. "$$\n",
  320. "y = w_2 A(w_1 x)\n",
  321. "$$\n",
  322. "\n",
  323. "如果我们不使用激活函数,那么神经网络的结果就是\n",
  324. "\n",
  325. "$$\n",
  326. "y = w_2 (w_1 x) = (w_2 w_1) x = \\bar{w} x\n",
  327. "$$\n",
  328. "\n",
  329. "可以看到,我们将两层神经网络的参数合在一起,用 $\\bar{w}$ 来表示,两层的神经网络其实就变成了一层神经网络,只不过参数变成了新的 $\\bar{w}$,所以如果不使用激活函数,那么不管多少层的神经网络,$y = w_n \\cdots w_2 w_1 x = \\bar{w} x$,就都变成了单层神经网络,所以在每一层我们都必须使用激活函数。\n",
  330. "\n",
  331. "最后我们看看激活函数对神经网络的影响(FIXME:可以找一个XOR动画的例子)\n",
  332. "\n",
  333. "![](images/nn-activation-function.gif)\n",
  334. "\n",
  335. "可以看到使用了激活函数之后,神经网络可以通过改变权重实现任意形状,越是复杂的神经网络能拟合的形状越复杂,这就是著名的神经网络万有逼近定理。神经网络使用的激活函数都是非线性的,每个激活函数都输入一个值,然后做一种特定的数学运算得到一个结果。\n",
  336. "\n",
  337. "### 6.1 sigmoid 激活函数\n",
  338. "\n",
  339. "$$\\sigma(x) = \\frac{1}{1 + e^{-x}}$$\n",
  340. "\n",
  341. "![](images/act-sigmoid.jpg)\n",
  342. "\n",
  343. "### 6.2 tanh 激活函数\n",
  344. "\n",
  345. "$$tanh(x) = 2 \\sigma(2x) - 1$$\n",
  346. "\n",
  347. "![](images/act-tanh.jpg)\n",
  348. "\n",
  349. "### 6.3 ReLU 激活函数\n",
  350. "\n",
  351. "$$ReLU(x) = max(0, x)$$\n",
  352. "\n",
  353. "![](images/act-relu.jpg)\n",
  354. "\n",
  355. "当输入 $x<0$ 时,输出为 $0$,当 $x> 0$ 时,输出为 $x$。该激活函数使网络更快速地收敛。它不会饱和,即它可以对抗梯度消失问题,至少在正区域($x> 0$ 时)可以这样,因此神经元至少在一半区域中不会把所有零进行反向传播。由于使用了简单的阈值化(thresholding),ReLU 计算效率很高。\n",
  356. "\n",
  357. "在网络中,不同的输入可能包含着大小不同关键特征,使用大小可变的数据结构去做容器,则更加灵活。假如神经元激活具有稀疏性,那么不同激活路径上:不同数量(选择性不激活)、不同功能(分布式激活)。两种可优化的结构生成的激活路径,可以更好地从有效的数据的维度上,学习到相对稀疏的特征,起到自动化解离效果。\n",
  358. "\n",
  359. "![](images/nn-sparse.png)\n",
  360. "\n",
  361. "在深度神经网络中,对非线性的依赖程度就少一些。另外,稀疏特征并不需要网络具有很强的处理线性不可分机制。因此在深度学习模型中,使用简单、速度快的线性激活函数可能更为合适。如图,一旦神经元与神经元之间改为线性激活,网络的非线性部分仅仅来自于神经元部分选择性激活。\n",
  362. "\n",
  363. "\n",
  364. "更倾向于使用线性神经激活函数的另外一个原因是,减轻梯度法训练深度网络时的Vanishing Gradient Problem。\n",
  365. "\n",
  366. "看过BP推导的人都知道,误差从输出层反向传播算梯度时,在各层都要乘当前层的输入神经元值,激活函数的一阶导数。\n",
  367. "$$\n",
  368. "grad = error ⋅ sigmoid'(x) ⋅ x\n",
  369. "$$\n",
  370. "\n",
  371. "使用双端饱和(即值域被限制)Sigmoid系函数会有两个问题:\n",
  372. "\n",
  373. "1. sigmoid'(x) ∈ (0,1) 导数缩放\n",
  374. "2. x∈(0,1)或x∈(-1,1) 饱和值缩放\n",
  375. "\n",
  376. "这样,经过每一层时,Error都是成倍的衰减,一旦进行递推式的多层的反向传播,梯度就会不停的衰减,消失,使得网络学习变慢。而校正激活函数的梯度是1,且只有一端饱和,梯度很好的在反向传播中流动,训练速度得到了很大的提高。"
  377. ]
  378. },
  379. {
  380. "cell_type": "markdown",
  381. "metadata": {},
  382. "source": [
  383. "## 算法与处理步骤\n",
  384. "\n",
  385. "```\n",
  386. "# 每次训练\n",
  387. "for k in range(epoch)\n",
  388. " # 正向计算\n",
  389. " for j in range(NN_depth):\n",
  390. " # 式2 ( a = xxx)\n",
  391. " X_j = f( W_{j, j-1} X_{j-1})\n",
  392. "\n",
  393. " # 反向误差计算\n",
  394. " for j in range(NN_depth, 0, -1):\n",
  395. " # 式3, 式4\n",
  396. " delta = y_i(1-y_i)(t_i-y_i)\n",
  397. " or \n",
  398. " delta = a_i(1-a_i) \\sum w_ki delta_k\n",
  399. "\n",
  400. " # 式5\n",
  401. " w_ji = w_j + epsilon delta_j x_ji\n",
  402. "```\n"
  403. ]
  404. },
  405. {
  406. "cell_type": "markdown",
  407. "metadata": {},
  408. "source": [
  409. "## 7. 示例程序"
  410. ]
  411. },
  412. {
  413. "cell_type": "code",
  414. "execution_count": 4,
  415. "metadata": {},
  416. "outputs": [
  417. {
  418. "data": {
  419. "image/png": "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\n",
  420. "text/plain": [
  421. "<Figure size 432x288 with 1 Axes>"
  422. ]
  423. },
  424. "metadata": {
  425. "needs_background": "light"
  426. },
  427. "output_type": "display_data"
  428. }
  429. ],
  430. "source": [
  431. "%matplotlib inline\n",
  432. "\n",
  433. "import numpy as np\n",
  434. "from sklearn import datasets, linear_model\n",
  435. "import matplotlib.pyplot as plt\n",
  436. "\n",
  437. "# generate sample data\n",
  438. "np.random.seed(0)\n",
  439. "X, y = datasets.make_moons(200, noise=0.20)\n",
  440. "\n",
  441. "# generate nn output target\n",
  442. "t = np.zeros((X.shape[0], 2))\n",
  443. "t[np.where(y==0), 0] = 1\n",
  444. "t[np.where(y==1), 1] = 1\n",
  445. "\n",
  446. "# plot data\n",
  447. "plt.scatter(X[:, 0], X[:, 1], c=y, cmap=plt.cm.Spectral)\n",
  448. "plt.show()"
  449. ]
  450. },
  451. {
  452. "cell_type": "code",
  453. "execution_count": 5,
  454. "metadata": {},
  455. "outputs": [
  456. {
  457. "data": {
  458. "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXwAAAD4CAYAAADvsV2wAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy86wFpkAAAACXBIWXMAAAsTAAALEwEAmpwYAAB5y0lEQVR4nO2ddXgUVxeH3zvrcRKSECAQ3N3dizuUUlpKW+rUXT5a6u7uCoWWIsXd3d2CBQtxX5/vjw0py05cNoF5n4eHZOye2SRn7px7zu8IWZZRUVFRUbn+kbxtgIqKiopK2aA6fBUVFZUbBNXhq6ioqNwgqA5fRUVF5QZBdfgqKioqNwhabxuQG5UrV5ajoqK8bYaKiopKhWLnzp3xsiyHKu0rtw4/KiqKHTt2eNsMFRUVlQqFEOJMbvvUkI6KiorKDYLq8FVUVFRuEFSHr6KionKDoDp8FRUVlRsE1eGrFBlzXDKX1u0j/Wyst01RUVEpAOU2S0el/CI7nWx59HOOfb8IjVGP02Kjat/W9Pzzf2h9jN42T0VFJRfUGb5KoTn02RyO/7QEp8WGLSUDh9nKhRW72DzlU2+bBoDTZifpwCkyzsd52xQVlXKF6vBVCs2hT2bjyLS4bXOYrZycsQqHxeolq1ycnLmaGeGjWdD5YWbXm8iiHo+RdTnJqzapqJQXVIevUmgsSenKO5wy9izvOfz4HUfZcPd7WJPTsadn4TBbubzlEMsHv+A1m1RUyhOqw1cpNFV6tAAhPLb71gjDnmkmcW+0V2b6Bz76G8c1DxzZ5iD58BmSD50uc3tUVMobqsNXKTTt3r0Xnb8JSeda8xcaCY1Jj76SH7Pr3s6i7o8xPXQUR75dUKZ2ZcRcBoUObpJOS+bFxDK1RUWlPKJm6agUmsD6kYzY/wMH3p9F3JZDBDasQerxcyTsOo7Tasdhds2ytz3xJQG1I6jatw3mhBQOfz6X88t24FcznCaPjyG0XcMStatav7Yk7DiWM/4VnBYbIa3qluhYKioVEVFee9q2bdtWVsXTKgYZ5+OYXfd2HBabx76q/drQ/bfnmdfqXqxJ6S5nLAQak56u3z9F7Vt6l5gdlsRU5jafjDk+BafVDoDW10jTJ8fS6pVJJTaOikp5RgixU5bltkr71JCOSrHJik1CMugU92Wej2f/O39iiU/9b+YtyzgyLWx+8BOcNnuJ2WEIDmD47m9pNGUEAQ0iCevchG4/P6s6exWVbNSQjkqxCWpUE9nh9Ngu6bRU7deGmIVbFB270+4g5dg5KjWJKjFbjKFBtH//Adq//0CJXVNF5XpBneGrFButyUCbtyaj8THkbBM6DbpAH5o9PQ5j5UDF82SbHUMlv7IyU0Xlhked4auUCI2njCSgbjX2vzeTzAsJVOvXhubPjcenamWaPD6GxL3R2DPMOccLnZbQjo3xqVrZi1arqNxYqA5fpcSoPqA91Qe099hec1Q3kg6cYv87fyIZdDhtdio1rUWvWVO9YKWKyo2LmqWjUmZYktJI3HMCU0QIQQ1reNscFZXrkryydNQZvkqZYajkT0SvVt42wwOHJTtVVK+caaSicr2gOvzrmJTj58iIiSO4ee1cF05vZNLPxLLh7ve4tHYvCIjo3Zqu3z+Fb/VQb5umolIqqA6/BIhZtJWDH/2N+XIykUM60uSJMRhDvOdgLcnprBzxP+K3H0XSa3GYrTSaMoJ2796HUNDAuRGxZ1lY0GkK5svJyE5XSunFlbtY0HkKY078Xiazfdnp5MzcjZyauRqNUU/9uwa6dIpUVEoJ1eEXk33v/sne137LyUBJORbDid+WMWLPdxiCA7xi0/o73iZuyyGXzEGWS8b46Nf/EtQ4ivp3DvCKTeWNM7PXYUvPynH2ALLDiS0lk7PzNlFrbI9SHV+WZVaNfpkLK3a5fneE4Mw/62n82GjavHZXqY6tcuOi5uEXA2tKOnte+cUt3dBpsWGOT+HQ53O9YpMlOZ3zy3bkSAtcwZ5h5tBHf3vFpvJI6okL2NOzPLbbM82knThf6uNfWL7zP2cPIMvYM8wc/OAv0k5fKvXxVW5MVIdfDBL3RCtKCjjNNs4v2eYFi8CWmoGQlH+slsS0Mram/FKpeW20fiaP7RofA5Wa1y718c/+u8ltonAFIUlcWL6z1MdXuTFRHX4xMIYFKWvBCIFPNe8s/PlWD0Uf5OuxXWg1VBugmKl1Q1JjaCd8qobkSDwDSHodfjXCqTagXamPrw/0Q2g1njs0Ap2/54NIRaUkUB1+MQhqVJOgRjU9/nC1Jj1NHhvtFZuEJNHlmyfQ+BhyZvqSQYc+0LfURMQyzsdx8OPZ7Ht7Bgl7o4lZuIUjX/9Lwu7jpTIeQOalRHa/8jPLh7zAzv/9RObFhEKdL+m0DNn0GXXvuAldoC/6ID/q3TmAQes/wRyXotgPN/NSIhvueZ8ZVUbzV+0J7H9vJk67o0j2153YD0nn6fAFEDm0U5GuqaKSH2rhVTHJik1k5cipJO6NRtJpkWWZDh8/RP07B3rVroQ9Jzj40d+knjhPRM+WNH5kJKbw4BIfJ3rGSjbe/T4ADpsdHE4kvRah0bhSHXu1os8/09xm0sUl+chZFnSagsNsxWmxIRl0aAw6Bm/8rFhCbCnHYlgz/nVSDp0BIfCrFUHP6S8S3KIO1pR05jS5i6zLycjZTl7jYyBycEd6zSxaxfCJ35ez6b6Pchy/EII+81+nSrfmRb4HFZW8Cq9Uh19CpJ26iCUhlaCmtdAa9d42p0ywJKYys/o4j4YjV6MxGWj16iSaPXlziY27pN9TXFy1x727lRBU6d6cgas/LNI17WYrf0WNxxyX4nZdfaAvY09P5/hPS9j50o8ezds1Rj3D93xLYP3IIo1rTc3g4uo9aAw6Inq1RGO4MX53VEoPVQ+/DPCvFUHltg1uGGcPcG7xNuU49FU4siwc+3ZhiY57ae0+z1aGssyl9fso6gTm7LyN2LMsHtd12hycnLGa2A0HPJw9uFRBE3afKNKYAPoAX2oO70L1Ae1VZ69S6qgOX6XUcVo9O2EVB61J2TFqjPp8C8tkp5OMc3HY0jLdtmfExOE0e9ppzzSTfjaWwAaRSHqFsJQs41czvODGq6h4EdXhqxSZ6gPb58Szc0My6Iga17NEx6175wA017xJaYx66t7RP8/zTs/ZwMzqNzO7wR1MDxvFmvGvYcvOxQ9t31DRoWv9TIR1bEyD+4Z4rEMInRb/WlUJ7dComHekolI2qA6/guCw2rBneuZtexNDcABdfngKjVGfa4tDU1glWjx/a4mO2/ate6jSozkakwFdgA8ak4Gwrk1p/959uZ4Tt/Uw6257k6xLSTiyLDgtNs7M3cia8a8DEN6tGZXbNkBj+q+Ji8aoJ6BeNaoP7oBfjXD6L3vXNdM36JD0Wqr1a82AFe+pchUqFQZ10bacY0lKY9N9H3J2/iZkh5PglnXp8t2ThLSs623Tcsg4H8eGO9/j4urdHq0OjRHB3BIzM9disOKQfPgMyYfPEtQwkqDGUXkeu3L0y5ydu9EjRq8x6hl9/Fd8q4XisFjZ/+5MDnz4F7aUDBAC3+qV6fTV40QO6pBzjjkuGY1Rj87fp8TvSUWluKiLthUUWZZZ2u9pzs7fhNNqR3Y4Sdh5jMU9Hi903nlp4lstlPSzsYp9be2pmaQeLx2pgqBGNYka1S1fZw+QFn3Bc6EXV8gpI8aVc68x6Ek9fu4/WQpZJiMmjtU3TyNu62GcDgcnflvOmlvfYPW4Vzk9e12RF4lVVLxBiSRHCyF+BIYAl2VZbqqwXwCfAIOATGCSLMu7SmLs65n4bUdIORrjoYvjsNo4+t1CWk2dWORry04n55ftIHFPNP61I6gxvHOxskS0PkblcRxOt163JYklKY397/7J6dnr0fmZaDRlBPUm9Vd8mwjv1ozkw2eQbe5rDk6LjaBGrmYs5oQUTv+9ziPN1JFlZc8bv7sygdbszZFEiF2/n5iFW+j24zOlcn8qKiVNSc3wfwbykmEcCNTL/ncv8FUJjXtdkxp9ASF5xoedFhvJB04V+brW1Azmt72f1Te/yq7//cjGe97nr9q3kX4mtsjXbPjAMLS+7k5fSBJBTWriFxlW5Ovmhi0ji3/bPcDBj2eTduI8iXtOsPWRz9l4n3IefrNnbkHna4KrHgZaHyNNnhqLPtDVSD0jJi7XTJzEPSfcnD24BOlOzVxD4r7okr05FZVSokQcvizL64DEPA4ZDvwqu9gCBAkhIkpi7OuZ4Oa1cdo9wyQaHwOhHRsX+bq7p/5M8uGz2NOzXJLAaVmYLyex/s53inzN+ncPJGp0dzQmPVo/E1p/Ez7VK9Prr1eKfM28iP5tBZmXknBa/kultGeaOfnHSkW1Sb/IMIZu/4pa43pgDK9EUNMoOn31GK2n3ZlzTECdqjhtnllHQiOhC/RVFDuTnU5XEVgBcNodnPprLevvepcdz39PahmocqqoXE1Z6eFXA2Ku+v5c9raLVx8khLgX1xsANWqoPU8rNa1FlZ4tuLRmD44sV5hBaCR0fibq3VV06YboGSvdHCW4Qi+xGw5gy8hyzYQLiZAkuv38LM1fnEDclsP4VA0holfLUlmsBbiwchcOhawlSaclbuth/KOqeOwLqFOVnn+8lOs1df4+NHlsNIc++cctI0pjMhA5qCNpJy54fG6STosh2D9fex0WK4t7P0nSvpPYM8wInYZDn/5D99+eJ2pUt3zPV1EpCcrVoq0sy9/KstxWluW2oaFqmzmAPv9Mo8kTYzGGBaEL8CFqTA+Gbv8KQ5Bf0S9aiguNgfWqU/f2flTt07rUnD2AX1Q4QkF8TJZlfKqGFPm6rV+/i3bv34dfrSro/H2o2q8Ngzd8QpNHRynejxBQc2TXfK977MclJO09mfOWINscOLIsrB47jV3TflFWXVVRKWHKaoZ/HrhabKR69jaVfNAY9LR57a4S7YJU6+ZeHPt+kXsFrCQI69i4SLN7b9Dw/mEc/Wo+9qtCMEIjYaoSTHjXZkW+rhCChvcPo+H9wzz29Z79CmvHv+7KzJFdGT595r5WoPTMU3+uUq6jkGX2vzWD5H0n6T17WpHtVlEpCGXl8OcDU4QQfwIdgBRZli/mc45KKdH69bu4uHo3GTFx2NOz0PqZ0Bj1BLeqy+yGd6Ax6Ghw31BXdakmb60cbxFQpyq9/3mV9Xe+gy01E9nhpFKLOvSaNTXfQqiMc3HYsywE1K1WqKKp6gPac0vsbOK2HEZoNYR2aFjgz+faBe2rcVptnFuyjZSjMQQ2KJoIm4pKQSiRwishxAygJ1AZiAVeBnQAsix/nZ2W+TmuTJ5M4E5ZlvOsqlILr0oXp91BzILNJO6JxrdGGAc+nEX6yUs5KYkaHyPVB7ajdyktupYE1pR0Yjfsx5pmJrxLk3yzgdLPxLJq7CskHziNkCT0lfzo/tvzRPRsWeq2npm3kXW3vam48Auu9YMu3z1JrZt7lrotKtc3qjzyDYo1JZ2sy8n41QxHo1eWPgA4+ecqNt77oUePV42PgSGbPye4WW1kWcZhthZIoKwsOPTZHHY8+y2SQYfsdGII8uempe8Q1FB5sd/pcPB33dvJjIlza1yu9TUy8uCP+NUoXQE0WZbZ9sSXHP58rmKBmtbXyIBVHxDarmGp2qFy/aNW2t5g2M1W1t7+JjOqjGF+m/uZETaKw1/Oy/X4S2v3Kjb0dtrs7H9nBnve+oMZYaP4PWAIM6uN5dhPi0vT/Hy5vPkgO57/DofZii0lA3taFhnn4lh20zNuzvxqLq3egzUxzWO/0+bg2PeLSt1mIQQdPnqIgWs+8tAdknRaAhtEUrltg1K3Q+XGRnX4ZYAsy0T/vpy5re5lZuQ41t/9Luln/ytyclhtZF1OwukoWru8a9l0/4ec+Wc9TosNe3oWttRMdjzzDWfnb1I83rdGOJKCjr9sc3Bq1lp2v/gjloRUZIeTrEtJbHn4M6JnrCoRW4vC4a/m56Sp5iDLWFPSubz5kOI5mRcSkGXPh4HTaiNhzwn2v/snhz6bU+qSFeFdmjJo7ccENY1C0mmQ9FqqD+5A/2Xvlos3J5XrGzWkUwbsfPEHDn36T078Vmgk9EF+DN/zLUe+/pdDH892SRCYDLR+7U4aPTi8yGNZUzP4M3w0Dountnvl9g0ZuuULj+2ZFxOYXX9irvFlJfzrVGXM8d+KbGdxWDboOc4v2e6xXRfgS4/pL7oJnV0h5WgM81rd6yGbILSaHEcrtK75T7dfnqPWmB6KYzusNqJ/W07078vRGA00uG8INYZ3KZKztiSnozHo0JpKR3pC5cYkr5BOWWXp3LBYktI4+NHfbo7mSnXrmnGvkrgnGnt2JyWH2cr2Z75BH+RHnVv75HpNu9nK0W/+Jfq35Ug6LQ3uHUKdif2QNBosCamuLlQKDj/znGdjbgCfiBBuWvw2aye8Scb5eMglLHI1GTGX8z2mtKg5shux6/Z7pDk6rTbCOjdRPCewQSQ1R3fj7JyNOecJnQbZ7vxPAC37I1t/xztU69cmR3Ih5/oOB0v7P0vC9qM514jdsJ+6d9xEp88fLfR95FZLEb/zGAfen0nqiQtU6d6cpk+Oxadq5UJfX0XlWtSQTimTdOCUola802rj8pbDOc7+Co5MC3te/TXX6zkdDpb2fYqdL/5Awq7jxG09zJZHPmPthDcB8I0MU2wYLiSJsDzy08O7NmPs6elU7dOqQPcVUK8aAJkX4tn96q+sm/Q2x39a4moTWMrUub0fgQ0j/xNsEwKNj4E2b03OsyCt28/P0u6D+6nUvDb+datRqVltxeOEVsM5hTeImAVbSNh51O1BY88wc/zHJaQcP1e8m8rm7L+bWNTjMU7NWkvCzmMc/mIuc5pNVpSLUFEpLKrDL2V8q4d6lOMDrhLNXMJpmRfic73euUVbSdx30q2/qj3DTMyCzSTsOYGk1dD2vfvcFCqFJKH1NdB62qQ8bRVCKIqgXYvGx0Dbt+8ldtNBZje4g31vTSf61+VseeQz5jafjCUpLc/zi4vWqGfQhk9p/9GDVOvfjtrje9N/6bs0eXR0nudJGg0N7xvKiD3fMebYr4S2y22RVFb82Zxfuh17ukLYSxJcWrO3CHdyzahOJ5vu/8j1s80e32m1Y0vNYPfUn4p9fRUV1eGXMv61Igjt1BjpmrRIjUmPIThA8ZxKTWvler2La5QzamSHk9j1+wFocPcgev/1MmGdm+BbI4yom11yDAUp6qkxrDM1RnRxOX1JuLpZ6bX4VKuMxqinUrPa9Jr1MtUHdWDd7a688isPNHuGmYyYy+x7a3q+4xQXrVFPg3sGc9Pit+nx+wuEd/FQ5c6TtJMXuLzpoKJjl+0Oqg1o77HdFBakqKYpaTQF0tPJj/SYy1iT0z3tcTi5sEJVE1cpPmoMvwzo88801k96h3NLtiNpJLT+PnT68jEcWWY23vuh22xd42Og3bu5t+rziQhGY9R7LD5Kei2m8Eo531cf2IHqAz0XL/NDSBLdf32e+O1HOb9kG7pAX2rf0gtTeLDbcelnY8m66CmQ6rTaOfXX2jzvwdvYzVYWdnmErMvJHvs0Rj1df3haMTRU947+7H9vFuCueyO0GqoP7lhke+J3HGXTAx+TsPNYrscYQpQnByoqhUF1+GWAPtCPPnNew5KUhjUlA78aYTlCXPpAP3a//DNppy5RqVkt2rw5mfBcFh4B6tzWjz3TPGP8kk5L5LDOJWKvEILQ9g0JbZ97EZDGqM+125PWVPRGKmXB2TkbsGWYPWb3kl5HmzfvpvYtvRXP868VQc8ZL7Fu4luuDbKMzt+Hvv++gVYhrbUgpJ28wOLeTyq+tV1B62ukyRNjsZutJOw8htbXSHCLOmoap0qhUR1+Ack4H8fJGauwJKRS7aa2VOnZstB/cIZK/hgqub/6Rw7uSGQhZocao56eM6ey4e73XGmUsowxrBJ95kwrstMpCqawSlRuU5+4rYfdKke1PgZF4TFvk3Ishm1PfMnF1XsRAsXFZafVhjUlM8/r1BjWmfGX/yFuy2E0Rj2V29YvsipoVmwiGyZ/kGtzeq2/CdnmoNGUEWhMev4MHwVCIDudmMKD6bfgTVV7R6VQqHn4BSBm4RZWj3sV2eHEabGh9TUS0bsVvf+ZVmbiYilHY1g38S0S97i6K1Vu35Dmz9+Kf1Q4gY1qIoTAbrZydu4G0s/EEtq+YZEeSoUhPeYyi3s8jjkhBZwyslOm+qD29JzxPyRt+RFdy7yYwJwmd2FNychTGlrrZ6LH7y9Qo4TelPIi+o8VbLznAxxWGzg9bdL6m2g19Q7q3TWAzAsJ/Nv+QRxXP6SEwKdqCGNPTy+3Ancq3kHNwy8GDouVtRPe8MiKubhqN6dnraX2eOXX/5LElp7Fwq6PYElMy3FYcZsPsfGe9xl78g/ANYNd1O0xHGYL9iyX5k1wizr0X/ZuqRX2+EWGMebEb1xctZuMc3FUbteQSk2iSmWs4nDoszmuGX1ekxsBhsoBVB9c+HWPwpJ5KdHl7K9Zh7kaR5aV6BkrubBiJ5JO6y5lDSDL2FIzubRmL1X7tC5li1WuF1SHnw+XNx1U3G7PMHPit2Vl4vBPzVrjcg5XOSzZ6cSWmsm/HR4i+eBpZKd7KqE9PYuEXcfY/+6ftHr5jlKzTUgSVfu2KbXrlwTx248op8ZejQzm2GSS9p0kpFW9UrXn7NyNoNCrOAchkGWZxF3HXd9qJEXBNZAxx6eUjpEq1yVqWmY+CI0GcpkY5he2sGdZOPXXWo5+u4DU6AtFtiHt5EVF2QN7hpmk/SddzkBh9urIsnLil2VFHvd6oVKz2giFYrRrcZit7H2z9FNKnXaHYhgHXNW/QivBVQ5e2dm7hN/CuxYuHVXlxkZ1+PkQ1rmJYu611teYZ1/ZuG1HmFl1LBsmv8fWJ75kbrO72frYF7lmtuRF5bb10frl0okqn8vlph55I9H44ZFoFKqdPZBlkg+eKnV7agztpLhd42OgSrfmyAqN1JGEWy2H1tdI40dH41tNbQWqUnBUh58PklZDnzmvovM3ofUzIRl0aEwGat3SixrDuyie43Q4WDH0RazZ0r2OTAsOs5VjPyzi3KKthbYhckgn/GqEuUk0CE3+PzqNUU+dCX0LPd71hn+tCAasfJ+Q1vUQkoTQaxU/P6GRCGldv9Tt8asZTuvX7kRj0rt0jyQJjclA4ykjCGxYQ9E2rY+R+vcMIrRTY6r1b0eP6S/S5s27S91WlesLNUungNjSMjkzZwOWxDSq9m2dZzVs7Ib9LB/8ArY0zxS/GsO70GfOq4Ue35qSzu5pv3JyxiqEJAhqXJPLmw56ygRno/UzEVCvGoPWfowut7eDGxCH1Yak1bDtqa85+u0Ct8V4ra+Rodu+JKhRzTKxJfnQaU7OWoNsdxI1uhshreqRdOAU/3Z8yM0uhMCnWghjT6kZOSr5o3a8KmMurNzFqtEvY0v1dPhVb2pL/yXvFHsMa0o6f9eb6NbUQ+i1+FYNofaEvoS2b0j1QR0QksSxHxZz8KO/sCalU7VfG1q/dmepd3gqK+xZFnY8/x0nflqCPctKlR7N6fjZI7l2vrqC7HRy8KO/OfDhX1gS0wht35D2Hz5I5TalP8PPj5N/rmLjfR8ihEB2ODFVCabfwjcJrH/95NxbElM5+MlsYhZswRgWRNPHxlCtfztvm3VdoDr8MsaeZWFG+GiP6kmtr5FOXz5G3dv7lcg4aacvsfWRzzi/dAeSXkvtW/vQ/v370fn75Byz7cmvOPrNgv8kgTUS+kA/Rhz4AZ8qwbldusKwtP8zXFqzF6ftP7kDXaAvow7/XKHv70pVrc7PRKXmta+rqlpLUhrzWt1LVmxSTvaU1sdIq2l30PTJm71sXcVHbXFYxmhNBrr99AwakyFHqljrZySsU5MSTeP0j6pC3/lvcIdlKbenLaTLN0+4OXtzfApHvprvVskpO5zYMrI49MnsErPDW8Qs2sKF5TvdnD24HrhHv57vJatKBq1RT3iXptelhMKRL+dhvpzsliprzzSza+rPWFMzvGhZ8ci8lMjZfzcRv+NokZIzygI1D7+UiBrdnZDW9Tjxy1LM8SlUH9SR6gPaFbkMvygk7T+JZNB5FPg4LTYurS2+nK+32fzQp4rbZaud+J3HcDocrkXa68xhVnRiFm5RLDqT9FoSd5+gSo8WXrCq6MiyzPanv+bwF/PQGPTIDgd+UeH0X/puuWtcozr8UsS/VgStXplU5POdDge2lAx0gb5FWqzzjQzDabV7bBeSREC96kW2qzyQcS6OrEueap1XSNh9gl/0/ZH0WupOvIn2Hz6AzrfsFq/Tz8Sy//2ZxG05TFDjmjR96maCc2m4cqPhU7WyYj8Ip82OMSzIO0YVg1MzV3P0mwU4Lbact5aUIzGsGvMKQzZ97mXr3FEdfjlElmUOvDeTvW9Nx5FlRetjoOXLE2n8yKhCzVYD6lYjtENDLm865FaaLxl1NH1ibGmYXuLIskzshv1c3nAAU5VgosZ0R+fvg9PuyPNtKeuCqxm502Ij+rdlpJ28wIDl75eJzcmHz7Cg0xTsWRZkm4PE3Sc4M3sdfea/QdXeBesoVhwyLyZw4rdlZJ6LJ6JXKyKHdipX2kZNHhvNuSXb3DKRhFZDUMMaZZYhVRjMCSmc/nsd1mRX0kPla1J3D34y26MwUnY4SdwTTXrMZfwiw8rS3DxRHX4JkHriPDtf/IFLa/ZgqBxI06fGUW9S/yKHEg59+g97Xvst55fIarWx68Uf0foaaTB5cKGu1WfOq2y46z1iFm5FaASGSv50/vZJglvUKZJtZYnTZmf50Be5vPEADrNLH2jbE1/Sf+X7hLSqh0+1ENJOKFQwS8KtktVhtnF58yGSD58pE4ey/ZlvsKVl5cxgZacTe6aFzQ98xOijubevLAkurdvH8sHPIzucOMxWjv+8lKBGNRi45qNy0yw9vGszOnz0INue/BohCZw2B5Wa1KTPvNe9bZoHF1bsZOXIqciyjNNqZ89rv1FrbA+6/vhMzt+3LUV53UFoNYqZet5EzdIpJulnY5nb4h7saVk56ZFaHyONHx1FmzeKVhgzI3w05rhkj+0+1UMZd/bPIl3TmpqBLS0Tn6qVK0xM+9Dnc9jx3HfuOemAX1QVxkT/TsLOYyzp8xROuwNHlgWtnwkhcDnba9AF+NDtl+eomUuxXEnye+BQxRoModUwIWGu28J6SSI7ncyMHOfRmEZj0tPqlUk0e3pcqYxbVOxZFpL2ncQQEkBA3WreNscDh8XKjCpjPBy61tdI999fyPld2vH8dxz8aLaHwJ2hcgC3XPy7zGsn1CydUmT/u39izzS7SRjYM80c/OhvrCme7eryQ3Y6FZ09kGfMOj/0Ab74VgutMM4e4PiPSzycPYA5LpmUozFUbtuAsaf+oO3bk2ny5Fi6//oc1QZ1cMWHr8FptRPUuGzCBfogX8XtklaDphR7FiQfPqs4o3RkWYn+Y0WpjVtUtCYDoR0alUtnDxC74YCiRpU9w8yJn5fmfN/s6XGYIoLRZL9BCY2ExsdAl++eKneFcqrDLyax6w8oap9IBh3Jh88W+npCkvCrFaG4L7B+xV5oLTR5vX1m7zMEB9D44VG0f+9+ZIeTs/M3KXayqnpTWwLLaKG6yeNj0Pq4h080Rj21J/TJSdMtDSSdJtd0QI2+AFpCKu7k8ft39edsCA5gxN7vaP3anVTr35Z6dw9i6JYvyuRtsrCoDr+YBNSrpjijdJitZMRc5tiPi0nYc6JQ12z3/v1ornUYJgPt3r+/WLaWV+yZZk7OXM3RbxeQdvK/mHzdSf09Pgdw9XcNvKaSVpZltj35FU4FqQl9JT96zfxfyRueC40fGeWy3ahHF+iLxqinWv92dPz04VIdN6BedXyrh3r8Pmp9jdS/p3BrPyoQ1rWZ4natr5G6E29y26YP8KXpE2O5afE7dPn68TylV7yJGsMvILIsE7NgMydnrELSuVL9Inq3In7HURb3esIt9CAZdK6siOzFQ1mWqdKtOX3mvorGULBX+nNLtrHrpR9JPXGewAaRtHlz8nXZ6CJ200GWD3oOWZZdMsBOmUZTRtD2nXtx2uwsG/Ac8TuOYs/IQmsyIrQS/Ze/R2g79367DouVX30HKcoOSwYdd2QtKatbysEcn0LKkbP4RVVxOeISxJ5pdoUOrvl9Sj50mkW9nsBptrkK0iRB5OCO9Jj+YrkLL1QEzi3Zxqoxr4BTxmG1oTUZiBzaiR6/v1CmNTWFQZVWKCayLLP2tjeJmb8pJ3NG62ukwX1DaP/+A5ydv4nND36MJTENWZYxVPLHHJ+CbP8v1KMx6Wn27C20mlq0ZiSyLJOw6zjpZ2IJaV0P/6gqhT7/wrIdHP91GTid1Lmtn0trx4sxfafNzoyIMVgT09y2a32N9P5nGtX6tUWWZS6u2k3shgP4RARTa1xP9IF+HteSZZnpISOwJnuum/hFVcnpDFbRSTp4mg13v0fCruMIIag2oB1dvnsSU1ilnGMcFisxC7eSdTGB8K7NKkRGVnkmKzaRUzPXYE3JoGq/NoR2aFSu18JUh19MLq3fx/JBz3vk2mpMeobv+pbABpHIsuxaVJUEf9UY71HuD0XPsjHHp7C0/zOkHjuH0Eg4rXZqjetJl+8Lvii06YGPiP59hdsDK2psD7r9+Eyh7SkpLqzazapRUxUXGqNu7kGvP6cW6np7357Ovtf/cJOS0PoY6fj5w9SbNKDY9pYFsixjSUxFazKg9TG67TMnpDC77u1YUzNz4stCqyGgblVGHvix3M44VcoWNUunmJxbtNXNiVxBluH80u0ACCHwiQjJc3Esrx6mebHu9rdIOnAKe4YZW2omDrOVU3+t5ciX8wp0fuLeaE78ttztgWXPMHNq1hrith8pkk0lgdOSR09Xcz4tCRVo/ux4mj4zDq2fCY1Rjz7Ql9Zv3lVhnP2ldfuY3eAOZla7mT+Ch7Pq5mlumV7Hf16Kw2p3b3Vpd5BxPp6Layq+VIZK6XNdO/ykg6c5PXsdyYdOF+s6ugBfJJ2nI5c0kkdOtaGSP4GNPKV5hU5DzRGFX7W3pqRzcfVuj0wgR6aFw5/PLdA1zi/drvjG4TDbOL9ke6FtKinCuzdHtnt25NL6GoskMieEoNXUiUxImMvYU38wPm4OTR4ZXRKmljopx8+xfNDzpJ04j9Nqx2m1EzN/M8uHvPjfMUfO4sjyTFOVHU7SitBC8/yyHawc/TJL+z/D8Z+WKP6OqBQep8NB8pGzZF5M8LYpHlyXDt+eaWZJv6f5t8ODbLj7Pea3f5Cl/Z/BrvDHUhDq3NoHoVGO2dUY2dVjW7efnkEX4JOTc631NeITEULr1+8q9Nj2TAsil4bXSgVGSuj8fRTTATV6rWIR0PnlO1jQeQrTw0axuM+TxObSyL246HxNdPnhKXdVUV8jVXq0IGpM9yJfV9JpMYUHl5qcgDk+hcubD5JZjLqIazn0yWwc1xTuOK02EnYfJ+ngaQBC2zVE62v0OFcIUeg4/Y4XvmfV6Jc5O2cDF5bvZMsjn7Gkr6uITaXonJ6zgT8jxvBv+wf5q/YEFvd6gqzLSd42K4cScfhCiAFCiKNCiBNCiOcU9k8SQsQJIfZk/5tcEuPmxranv3GV42daXCGQTAuX1u9nx3PfFel6fjXD6fbTs2h8DOj8fdAF+KDz96HPnFcxBHkuIIa0qsfoY7/S8uWJ1J3Un/YfPMDIgz+6LawVFFOVYIzhCrruGonIYcq9Ua8lV+cpoNa4nm6bTv+znpUjpxK35TCW+BQurd7D0n5PF0hd8/Lmg6wY9hL/NJ7E+rvfJfXE+XzPqT2uFyP2fU+z58bT8KHh9P77FfrOf71cZpQ4HQ42Pfgxs2rcwrJBz/N37QmsufV1D0ddFFKOxrgt8l9B0mpIPxMLQO0JfdAH+bnaImajMeoJaVufyu0aFHis9JjLHProb48QX8Ku45ydt7EYd3FjkXkxAXN8Ss73iXujWXf7m1jiU7GnZ+G02IjddIDlg573opXuFHvRVgihAY4B/YBzwHZgvCzLh646ZhLQVpblKQW9bnEWbX/1G6RYoanzN3FbyoIiXRNcbQ4vrt6D0GqI6N0KbSlWTV7N6lte4/SsNW7bhFbD6KO/4J9Lkda1nFu8ldXjXst5W5AdTnr88SI1hnXOOUaWZf6qdSsZZy97nF+5bQOGbvsy1+ufnb+JNeNfd61TyDJCI6H1MTBk8+cENY4qkI3lnX3vzGDPa7+5/W5pTHoa3DeUDh8+WKxr737lZ/a986ebRjy4HPqoo7/kCHBlXkpkx7Pfcnb+JjR6HXUn9afVK3cUSifnxG/L2fzQJx4NegDqTryJbj8/W6x7ud6J33mMdbe9SdrpSyBDSOt69Jj+Inte/ZXoX5e7Vd2DK3Fg8ObPykwtNa9F25Io+2sPnJBl+WT2YH8Cw4FDeZ5VSsiynOviqD2X/q8FRefv4+Ygy4KMc3HEKMy6JL2WCyt2og/yJ/N8PKEdGhLasXGu6WLVB3ZgfOxsLq7aDbLsemBdkwXisNjIPBeveH7i/pO52ijLMpsf+sQtviw7nNjSzex4/nv6lkNRrKJw6JN/PCYSjiwrR79dQPv37y9Wlkyjh0Zw+It5WO0OVz0CoPExUOvmnm5qiz5Vgun+i8dLdKHQB/kq/p4IrQZDaCCyLOc89P1qXh+tMEsKc3wKS3o/6aaVFL/9CIu6P4Z/naoezh5c63dZFxKgHMhjl4TDrwbEXPX9OaCDwnGjhRDdcb0NPC7Lcsy1Bwgh7gXuBahRI++epLkhhCC8WzNi1+13L40Wgio9mhfpmt4kduMBJL0OxzUzP0emhS2PfI5Gr8VhsSPpNIR1akLfBW/kmimkNRmIHNwx17E0Bh1af5Oi+p9PREiu51kSUt1ebXOQZZceyXWCUo4/uLKvnHYHGn3RHb4xNIhhO75m7e1vcnnTQVcBmSQR0qYesiyXaN53tf7tkPSef/qSTkt4l6b80/hOMs66wkh+NcPp+ef/CG6u5vIDnPh1mcc6h+xwYktJx69GGPFGvWLDoZDW9crSzFwpq0Xbf4EoWZabA8uBX5QOkmX5W1mW28qy3DY0tOiViZ2/eBRdgA+SweX4JKMeXYAPHT97pMjX9BbG0CBAOezmtNiwpWXhtNqwZ5iJ3XiAQ5/8U+SxhBA0fepmDzkDrY+BFi9NyPU8l0qlskMyhgYW2Z7yRmjHRorbgxrVLBGtmsS90STsOp5TLexIz2Lns99z6LOi/0yV0Oh19F/2LqYqwTlrUlofIx0/ncKGu94j9WgMjiwrjiwrKUdiWNzrCWwK4Z8bkbToC4qZUk67g+DmtTEE+7s9TLW+Rho9PDL779j7lITDPw9EXvV99extOciynCDL8pVP6XugTQmMmytBjaMYdfhnmj09jupDOtL8mXGMOvwzQQ2L9tZQVDIvJXLkm385/OU8Ms7FFfp8WZZJ2HMcW4ZnDYASjiwLx35cnO9xifui2fb012x+6BMurNzlJgTV4vlbafLYGLS+RjQmA7oAH1pNm0S9Owfmej1ttjCYxuS+pqH1NdLsmVsKZHtFoMNHD7kebhrXn42QJLQ+Rjp98WiJXH/niz94hIzsmWb2vPqbYqigOIS0qsfNMX/Sb9Fb9PrrZcZfno2QJMXUTKfVzum/15bo+BWVsC5N0fp5dk4TkkSVHi0YtusbGj00Av86VancrgFdvn2Stm/f4wVLlSmJRVstrjBNH1yOfjtwqyzLB686JkKW5YvZX48EnpVlOffYAuWr0rYoHP91KZvv/9ilpyMDskybd+6hycOjCnyNfe/+yd5Xf/Mo+pJMerA7Ff84/etUZczx33K95sFPZ7Pz+R9wWm3IDidaXyM1hneh+2/Pu83SHRYrloRUjKFBBVJ4tJutbLjrXc7M2YDGoMNpc9D0qbG0emVSuS5DLyypJ86z/90/id9+lKAmUTR/bnyJCWX95jdYscBPaDVMSJyHTsHRFIW0UxfJiImjUtMoDMEBOdv3vP47u1/+2VMlUhK0nnYnLV7M/S3vRsFhsTKv5b2knb6Us8CuMRkI79aM/kve8bJ1Lkp10VaWZbsQYgqwFNAAP8qyfFAI8SqwQ5bl+cAjQohhgB1IBCYVd9zygC09i9N/ryXzQgJhnRpTpWdLhBBkXkxg8/0fe8Tydj77HdX7tyOwfmQuV/wPp8PBvremKzqAwPrVcWRaSD3unvYoGfXUmdAn12tmXkpk53Pfu9llzzBzdt5GLq7cRdW+bXLGTj1xAa2PocByvlqjnp7TX8Icl0zG+XgC6lQttUYf3iSgbjW6fPtkqVzbv25VkvZ5Lo7rg3wV8+8LizU1g9VjphG7YT+SQYfTYqPhlBG0e+dehBCEdWyE1tfokb2j9THmGs66EbCbrdhSMzBWDkRj0DN48+fse/MPTs1cg6TXUn/yIJo8PsbbZhaIEhHnlmV5EbDomm1Tr/r6eaD8JKPmQezGA+x84XuSD51xNSF/dRLVB7T3OC5xbzSLez2B02bPab9XuW0DblryNmfmbFBuwmF3cPrvdbR4If+Zki01UzFWCJB+6hIDVn7Akj5P4rT91+0poG5VmubR1ejCsh1uOdxXsGeYOT17HVX7tuH88h2su/0t7Bmupi4B9arTe/Y0AupUzddmcK05lJd4ZUWj7VuTWTVmmtvPXetjoPW0O0vkLWnD5Pe5tH4fTost56F/9Kv5BDWsQf27BhLRuxXBLeqQsOt4jg0ak4HglnWIKINevOUNh8XKlkc/d6VayjKGID86fDKFWjf3pN2799Hu3fu8bWKhuS4rbYvKpbV7WXrTM8Su348lIZX4HUdZNeYVTl0Tv5RlmdU3T8OanO5yjA4n9gwzcduOcOizOa60OqVQmSwXuHxdH+iL1lf5FT6wQSSV29Rn7Mk/aPvOPTndnoZu+wpdLueASyYYJb8huXLm005fYtXIlzFfTsaeYcaRZSXpwCmW9HoCp0OtwCxtqg/sQM8/XyKgQSRCI+EbGUaHzx6m4QPDin1tW1omMfM3e+T52zPMHPjwL8AVh+6//D1aTr2dgAaRBDaIpOXLE+m/7L3rKixXUDbd9xHRv61wZWFZbGTFJrH+rncLVIRYXlGbmF/Ftqe+9phVOzItbH/ya6JGd8/5pU8/dZEMhXx1R5aFEz8vpe+CN9nx7Lce+yW9jpoKUgxKCEmi1bQ72Pncd9jdCn0MtHnTVahsCA6g8ZSRBb6/yEEdFPXiNQYddW6/iWM/LMJpv+aB5JSxpmRwcdVuqvVTDAuqlCA1hnamxtCSr/WwpmbkKtFhTUjN+Vpr1NP82fE0f3Z8idtQkbAkpXFq1hqPsKwj08Ke139nQI8WXrKseKgz/KtIztYsuZbMC/E4LDZST5zn4MezOfHb8lzbn8ky+EdVodWrk9CY9AitJqfHZeNHRxUqn7nxlJF0+OxhfGuGI+m0VGpWiz7/TCtyIxSdvw+9Z09D62tE529C62dEY9TR+rU7CWlZl4wzsTitnm8gstPp0RhbpfySdvoSR77+lxO/Lc9R2/SJCMEQ7O9xrNBIVO1XqklzFZKsiwlIOmV5j6u7slU01Bn+VZgigkk/dclju9bPxP53/2T/2zNcKYxCKEr7akwG6t3ZH4BmT40jcnBHTs1ag2x3UnN0N0Ja1i20TfXvHEj9PFIiC0u1m9pyy4W/iFm4BYfZSrX+7XKKqqr2bcOZORs8dP9lh5Owzk1KzAaVomHLyMJpcyjqN11h9ys/s//dmSAEQiOx+YGPXc1kbmpLh88eZu1tb7rCOk4ZSa9F62uk1at3luFdVAz8akUgK7wNC0kirFNjL1hUMqgNUK7i+M9L2DLlM7fMGI2PgXp3DuD4j0sUF1E1PgbXoqmviZBWdblp6btlprFT0tjNVv5tez9pJy/mvMqWh0YpNzrm+BTW3/kOF5btBCCgQXW6/fgMldu6C6Zd3nyQJf2e9sjl1/gYCGlZl7ithwEwhlXCFBpERL/WNH18DD5VK5fNjVQw9r0zg72v/f6fPxACra+Rodu+LPOansJQ2lo61w117+iPNTmd3dN+xWmxIbQSTR4fg8Ns9ZA2gP+0TvxrRRDWqTERfVpX6MUtrVHPkM2fc/Dj2ZyauRqtj5GGDw7zaNisUnbIssySPk+SfORsTk+E5AOnWdLnSUYd/tnNWR//ZSkOBb0oR6aFy5sP5YQhzXHJaAw62r51T6lJSF8PNHvmFnyrh7LvrelkxSYR2qkxbd+cXK6dfX6oDv8qhBA0eWwMjaaMxByfgiHYH41ex7ZnvlGM2QshCO/SlPp3D/KCtaWDzt+Hlv+7nZb/u93bppQpcduPsHvqzyTuiyagXnVavTyRiF7eT0W8vPkQaacueTbAsdo5+t1CWr38X49kp8We69rStV2yLAkpnF+yjcghBZPYvhERQlBnQl/qTOjrbVNKDHXRVgFJq8GnSnCOPkqtsT08ZAPAFduOHJJnwXCJknkhnotr9pBxvvAyDRUNW1pmkRvWFJbYjQdY3OsJzi/dTtbFRGLX7WP50Bc5M3dDmYyfF+m5LBA6LTZSjpx121ZrXM8CF2g5rHZST1TcxUeVoqE6/AIQ2q4hjR4cjsZkQGgkJJ0WjUlPh08ewqTUnKSEcdrsrL3tTf6ucxsrR05ldr2JrBn/Wp6NNzLOxZG4/2SFa1uXdOAU/7Z/kD9CRvBH0DCWDXquRDtLKbHtqa89ZY8zLWx7/Eu8vcYV3Kpejlzy1Wh8DB4L6dX6t6PG8C4e4ndKSDotlZp7X65XpWy57hZtE/dGE7f1MD5VQ6g2oH2JxigT90Vzdt4mJL02J3ZfFux88QcOfjzbbdFYYzLQ6KERtHv3Xrdjs2ITWTVmGgk7jyF0GiSNhk5fPUbtcb3KxNbiYE5IYXa9iW4yxEKrwb9WFUYd/rlYevN58avPQMUeCkKSuC19odcX4VeMeInzy3bivGKjRsIUVolBGz7h6Nf/cm7RVoxhQTR9fAzVB3dk54s/cOD9WYodtMBVgFepaS2GbvuyQq85FRVZlkk+dAZ7ehbBreqWiNJpeeKGWLR12h2sHveqqym3AEmjQedvYtC6j/GvXTBZgPwIbl7HK7rgR76a71kQlmXh6Df/ejj85YNfIHHfSdcfe3ZywYa73yOgTlWPrA5vIssyhz79hwPvz8KSkErltg0IaVvfY3FctjvIupTIhZW7Sq3wyxheiYzsNoJXo/U1olHQjS9rGj88ypWhI4Srm5gQNH36ZhZ3exRzfCpOqw0OQvzWI7T4322cmb0uV2evr+RPndv60uaNu29IZ5964jzLh75I5rm4HNXTLt8/Ra0xPbxsWdng/d/mEuLI1/M5v3R7jmN04MpbXnXzqwzf8bV3jSsmuWmR29Kz3JpjJB04RcoRz96oDrOVg5/MpsdvL5S6rQVlx7PfcuTL+Tkpb7Eb9nN5yyFFR+W0OxXrI0qKFi/cyrbHv/JIx2382OgivVVYktKI3bAffYAvYV2bFqs/rzU1g5Ujp7q9gch2Bzue+RahkVzOPht7ppk9034FjbLNkk7D2FN/oA/wLbI9FRmnw8GS3k+ScT7ebRF7/R3vUKlJFEGNauZ5vt1s5dSMVZxfvgPfyDAa3DukwBpT5YXrxuEf/fpfzz62TpmUQ2fIOBeHb/WiN1TxNqHtG7q6IF1D5XYN3GZpWZcSEUrVgU5ZsU9tWRC7YT+7Xv6ZlCNnCWoSRetpkwhqEsXhz+d6hFFkhxOhkTxi1kISBLcovTer+pMHY45PZd9b00GWkZ0yDR8YRsuphc9UOvjZP+x89jtXEwzZ9ZZw09J3itzP9Oy8TYrbZYdDuem5XodfVLii6qYxNOi6VDAtKLHr9rkqj68JYzttdo58/S8dP3G13LZlZHHs+0WcnbcRY2gQjaaMIKRlXf7t+BAZZy9jzzAjdFoOfzGX3n+/QkirulxcvQedvw9V+7Up1yGi68bhK+XJAyCJ3PdVEDp+9jCLejyOw2xFtjsQWg0ag45On7t38AppXc9DHAtcjbCr9W9XVubmcH7ZDlaOmprzIM66mMiSTQfp8PFDSHqtZ9xclkFICJ0mJw1RY9QT0rYBlds3LDU7hRC0eP5Wmj4xhsyLiZjCKxWqKfgV4rYeZufzLvnpK/dmS8tkWf9nuTnmzyLN9K0pGcrhmVyW3pw2Oy1evI31d77jysnPdm4aHwPtP3rwhgzjXMEcl4KSeqBsd5B5IQFwvTX/2+FB0k/HuqIFQhCzcAsRPVuSfupSzs9Vttlx2OysGvsKst0llnhFlqRyh0Z0//lZAhvkL4Ne1lw3WTq1b+2NxuD5ZDWFVcK/dtksrpYWIa3qMXz3t9SfPIjK7RtS766BDN/9rUdM3hAcQNOnb3ZLzZP0OgwhASWiuFhYtj72uWL2y8GPZys/hIWgap9W1L29H/pK/hjDK9H4sdHctOitMnFUGoMe/6gqRXL2AEe+/lex8MmeYSZ2/f4iXbNavzaKCqcao17xbU4X4EPUmO4MWvcx1Qe0wxQRTFiXJvSZ8yq1xvYskg3XC2FdmriFwK6g9TW6hAWBY98v/M/ZA8gyjkwL5xZvU1zYd2RYcFpsbhpU8VsPM7/Dg2ReTCidGykG180Mv9lT4zjzzwbSz1zCnm52/UFoJLr//sJ1MasJqFOVzl8+lu9xrV6ZRHCLuhz86G8sCSlEDulEs2duwVDJUzirNJFlmZSjHn3qAUg5dIZ6dw/k5IxVbg8EjUlPq1fvJLRdQ1q9Momk/Sfxr121yA64rLEkpSoXPglXf4OiENggkvqTB3P8x8U5GkdaXyMRfVtzcdVu7Db39R1bWibnFm0lcnBH+i18q0hjXq/4Vgul4ZQRHP3635zPUmPS418rglrjewNwdu5G5T4UkgBHwTMaHVkWDn85nzavlS+douvG4ev8fRi282vOzF7PpXV78YuqQt07+uNTpfTz5MsTQgiiRnUjalQ3r9thCA7AcpX07tX4161Go4dGcOTLedizLATUrUbHzx4mpHU91t/5jqubkEGH02ontGMj+s59rdzHn2uO7MbFlbs9xOecVjvh3ZoV+bodPn6I6gPbc/ynJTitdupM6IMxrBKXVu3xONaRaeHY94uIHFx2BYEViXbv3kd456Yc/nIuttRMosb2oOEDw3JSbw25NO+RtBqETiEMmQuyzUH89iMlZXaJcd04fACNXkft8b2pnf20VvEuTZ4Yy64Xf1Dcd+SzOYw7N4u2b03GabOjMbj+4Pa/N5NTf611i4Nf3nSQTQ98TI/fC5dlFLf9CCdnrEJ2ytQe15OwTqWr+Fl7fG+OfrOApP0nXU5fCDQmPW3euLtYb1hCCKoPaO/Wee3Cyl3KzWygzCqUKyJCCGqO7JprX4rGD4/k3OKt7qFISeBXI5zwns05+dsKl+S5cIVL7Rlm5RoOnaZUEw2KynVXeKVStlxau5ftz35D0oHT+FYLpeXLE6lzq6uvri3TzO/+QxTDHBqTgYkZizy2z6p5CxkxntIRkl7HbSnzcx4M+bFz6o8c/PDvnJi6xqSnwX1D6PDBg4W5vULjsNo4OWMVZ2avxxDsT4P7hpTKg8aeZWFG+GjP/rO+Rjp98agqeFcMDn46m53Pf4+k0yE7HPhUrcxNS97Gv1YEaacucnnTQUxVggnr2pQVQ1/k0uo9HpllOn8fRhz4Ab/IsDK3P6/CK9XhqxSZS+v3sWzAc+4VwD4G2r9/Pw3vdy0S/9P4Tg/NF4CI3q0YsOJ9j+2/VxqGLSXDY7vQarg17h/0gblrwV8h5VgM81rd67GAqjEZGLLps3I58yoKp2evY93Et5HtDpw2O1o/I2Edm9Bv0VtuFeauXsprOTVrDTp/H+pPHkSVbs2xJKdz4pelJO6NJqRVXepOvKlAn++NgDU1g7ithzEEBxDSul6u64BOh4PTs9aw6+WfSTt1CRxOQjs2otOXjxWp/0VJoDp8lVJhYddHFOsDDCEBjI+djZAkLq7ezfKhL7pee52yq/uXycDg9Z8oOt5VN0/j7D8bkJ3uM6bAhjUYdeinAtl14INZ7HzxR4+MDKGRaDl14nWlBJp26iInflmKOS6F6oM7Un1AO7diMafDwbKBzxG3+VBOmElr0lP/3iFE/7Yce5YFR6YFrY8BjY9L690/qooX76jiIssystNZrEK7kiAvh3/dpGWqlD1JubSEtKVnYUlKAyCiVyuGbPzMpT1UOwJD5UAQsG7iW5yZ46lG2e7te9AF+boaruOa2Wt9jHT+5vEC23UlQ+tahEZCU0Gb0+SGf60IWr0yiU5fPErkoA4elcEx8zf/5+wBZBl7poVDn/yDJSktJ1Ztz7RgTUxjy8OflvUtXDcIIbzu7PNDdfgqRcavZrjido1e5xYaCG5Rh0YPjSDzUiLm2CTsaVkk7T/F2tvf5PjPS9zO9a9dlVEHf6TpkzdTpVdLGtw7mGG7v6FKt+YFtqvm6O4oVSYJSaLWzT09tjssLumJea3vZV6b+zj85dwKpzKaG2fnbfTIGgJc6yrXtPCTnc6crlrlmYQ9J9j80Mesunka0X+syFM1VsWd6ypLR6Vsaf3qnay59XW3jAatj5EmT471UCnd8dx3ikVYO579jroTb3KbmZrCg2nz+l1FtsunSjBdf3yGDXe+i9BqXHIJDiedvnzU4yElO50sG/AccduP5Ni3/ZlviVmwhX4Ly6bgqzTRBfkpylXkhlTGYnHpMZc5/PlckvafJLR9Qxo+MCxPyfFjPy5my8Of4bTakB1Ozi/exuEv5jFw9QcFXtC/kVEdfh6kHI0h43w8wc1rY6wc6G1zyh01hnWmy7dPsP3pbzHHJaP1NdLs6XE0f268x7FJ+z21XQCsKelYUzPzbMxdFGqP60W1fm2IWbgVZJnqgzoo/gwvrNxF/M5jbg8jR6aF2PX7ubzpIOFdmpaoXWVN/bsHcuy7hR7FRFcc+9UVopJBR+3sDKuyIH7XMRb3fAKn1VWpenHNXg59NoehW79E0mtJPxNLpSZRGIIDAFdR2ZaHP3O7F3uGmaR90UT/sZL6dw0sM9sLS8b5OC5vPIgxNJDw7s29Fvq5IRz++eU72PW/H0k9foHAhpG0ef2uPNvXWRJTWTH8fyTsOo6k1+K02Gj08Ejavn1PhZ/xlTR1bu1L7fF9sGea0ZoMuapL+kaGkawQ85f0OnR+plKxzRAcQN3b+3Fu8VYW93qC1BPn8a0eSqtpk3JSR2M3HvBIbQSXNlPshv0V3uEHN6tNh08eYuujnyPpXIJukl5Lz5lT2fHsN6QcPZeTNhvUOIr2799fZrZtuu8jt8/eabZitdpY0PUR7KmZrsI7i42GDw6j3Xv3c3nTQSSdBsc1Py57poVTs9aUS4cvyzLbn/mGI5/PRdLrABl9oB8DVn1AQN1qZW7Pde/wYxZsZvW413JmBXGbD7F8yIv0+WdaroJia29/i/htR3Da7DnnHflyHpWa1qLu7f3KzPaKghACnW/eTrvVK3ew7o63PcM/j48uUpOahD0nSD50hqBGNQhpVS/X484t2caqMdNyfo5p0RfYeO8HOMxW6t81EJ8qIWh8DB7hJo1Rj+k6qdJuMHkwtcb24NLafWh9DFTp0QJJp2Xotq+4vOmgS8m0cU1COzYuswmNw2ojcfcJzx1OGcvlZNcx2QVNR79eQGD9SIKaROXas1cfVD4ln8/O3ehS8rXYcvSjbOlmlg99kVGHfirzCeR1v2i77cmvFJuHbHvqK8XjzQkpXFy122PRzp5h5uBHf5Wandc7UaO70+GTKRgqByDpdWj9TDR+YnROE+74ncdY0u8p/ggezpxmd3Nq1hrF69gysljc6wkWdX2ETQ98xKJuj7Kox2O59gzY+fz3nj//TAs7X/wBWZapNa6n4uu1pJGIGt29eDddjtAH+lFjWGeq9m3jmunjelCHd2lK/bsHEdapSZk6H6GRlKW8FbBnmjnw4V+EdWqsKK+h9THQ8L6hJW1iiXD4y7mei+ayTGZMHCmHz5S5Pde1w5dlmdTj5xX3pRxRFvaypWYqpvQBWBLSSsy2G5EGdw9i/KXZ3HJhFhMS5tLm1bsQkuSK5fZ4jIsrd2NNTif54GnW3/Uuh7+Y63GN7c98S9yWQ9gzLdjTsrBnWojbdiTXB3jKsXOK2y0JqdgzzRgq+dN/+Xv4Roah9TWi9THgVyuCAas+KLVQk4qrI13tW3rlpN/mhyUxDSFJ9Fv8NsYwl66/1t8HjVFPs+duzTNE601yE80TWglbukL2VClzXYd0hBAYQwOzdbDdMYVXUjzHr2Y4On8fj1d8odVQbWB7xXNUCo6QpJxFuCvseulH7AoZPLte+pEG9w7JmZECRP+6zENa2WmxEf3bCrp8/YTHeH41wxUrfXUBPmh9XDLSoe0bMvb0dNdxQhDYIFJdqykDOn72COmnY4nbfgRJq8lZQL5Wm0ZIEhG9WgKuNYlx52Zxac0eLEnpVOneLM+sHm8TNbYHSQdOecpmC0Fwy7Kv+L6uZ/gAzZ+fkPOHfQWtj5EWL92meLyQJLp88wQan/8WICWDDkMlf1r+T/kcleKRsOu44naHzU5WbJL7tlya2TgsVpSqxlu/fhcaH3d55Ss//6uduhCCoEY1CWpYQ3X2ZYTOz8TA1R8ydOuXdPv5WUbs+57es19x/bwk189A0mnRBfjQ5s3JOedJWg1V+7ah1tge5drZAzR8YBgBdarl9KgQWg0ak4Eu3z3plc5Y1/UMH6Dxo6OwZ5rZ/86fOG12JIOOFi9OoEEeMb8awzozeMOnHPzob9JOXiCidysaTxmJMRfp1CuknbzAvndnEr/tCEFNatLsmVuK3NquoNjNVna9+APHflyMI9NClV4t6fjpFALrl79uO7nhF1UFc/ZCnRuyS6bhaiJ6teTCil3ui3dCENGrpaKjjhrVDYfZwo7nviPzfAKGEH9avHQbjR8eVcJ3oVJUKjWJolKTKAAC6lZj8IZP2f/eTFKPnyO8azOaPjG2wrYo1fmaGLLtS07NWEXMoq34VA2h4f1D8+yfa0l2tWEsjR4WN4yWjtNmx5KUhiE4oEhZIfmRuP8ki7o+gj0ruw2hJKEx6ui74E0ierYs8fGusGzw81xavee/12Ah0Af6MurwT+V+9nOFc4u3smrsNPdmKD4GGkweTIePH3I7NuX4ORZ0fMgln5xlRWPUozHqGbL583xbyjmsNiSd9rqawdvNVqzJ6RhDA8t9Wf/1QlZsIvvenkHMgi0YQgJo8thoao3rVezfq7RTF1k38W3itx0BAZWa1ab7b88T1LBGoa6jiqeVAUtuepqLK3Z5bC+M6FdhST58hvltH/AsqjHqaf7cLbSaekepjFsanPhtGduf/gZrSgaSRqLB/UNp+/a9ig9nc3wKx75bSPyu44S0rkuDyYPzffu6mvhdxzjy1XzMsUlEDu1Mndv6VpiuWldw2uxse/Irjv2wKKdZerv376feHf1LbAxbRhZ7pv3KiV+XITuc1BzdnTZv3IUx5MYtQrQkpjKn2d1Y4lNzMvm0vkZXnc5VYaeM83Gkn7lMUMNIjzUrJRwWK3/VnoA5Nvk/4UAh0Ffy4+ZT0wvV/Ccvh3/dh3TKCiXVSIDU4+ewZ1lKxaEkHzqjWIjiNFuJ3360xMcrTerefhN1JvTFkpCKLtA3z/imsXIgzZ+/tUjjHP1xEVsf/hynxYbsdHJx9R4Ofz6HIZs/91jrKc9sefRzTvy6LGcx0GG2svmhTzCGBuX0Zy0OsiyztN/TJO6Jznl7PPHTEi6u3MXIAz/csDIGh7+chzUp3S1t255h5tDHs2n6xFi0vkbW3vYm5xdvyykca3DvENp/+ECuRYkAMf9uxpae5a4SK8s4LTZOzVxN/cmDS8T+Elm0FUIMEEIcFUKcEEI8p7DfIISYmb1/qxAiqiTGLU/oc5EGkPTaUtMnCWwQidPu8BzToCO4Ze7FSOUVIUkYQ4NKZTHLYbWxeconbJr8AY4sS84flj3DTOqJCxz73rMZS3nFlpHFiZ+XKmoT7X3ttxIZI3b9fld2yVUZM87sRfTTs9eXyBgVkQvLdyp2uJIMOhJ2H2fLI59xPrvhuS0lA4fZyrHvF3L4i3l5Xjft1CXF69ozzKSeulhi9hfb4QshNMAXwECgMTBeCNH4msPuBpJkWa4LfAS8U9xxyxtNHh3tkQ2iMempd+fAUoutVmpai7COjT1ymTUGHY0eKJ+FKErIsow1Jb1EVA8tyemkRl/weBBuuvdDjn2/WPEcR5aF6Okr2PHC9yzt/ww7nvuOjHOeXbe8TWr0BXa88D0b7nrPo1/AFdLPxpbIWAm7j+O0eU4m7OlZJOw8ViJjVET8oqrkZBBdjWyzYwwN5OQfKz0ctz3Tkm/RZkib+opvTVo/E5XbNCie0VdfrwSu0R44IcvySQAhxJ/AcODQVccMB17J/vpv4HMhhJDL6wJCEWj65FjST1/i+E9Lcl7lIod2ov0HpatN0mfea2x78iuif12Ow2ojvHMTOn7xKD5VK5fquCXF+aXb2fTgJ2Seu4zQaKh7x020/+ihnKbSBcWWkcXGye9zZu5GJK0GSa+j/YcPUO+O/pgTUjg1a41HQ5SrSdh1nMQ9J3FabVxau5cjX81n0PqPCW5e+Fzpc0u2ceizOVgSUqk5siuNHhxe7AbsZ+ZuYO1tb+K0OZBzk24WgsptS8Y5+NeOyNGRuhqtr5GAetVLZAxvY45PYcfz33Fm9nqEJFF7Qh/avH5Xnj+rJo+N5vTsdW5vV0KrIahJFP61IpAdng9JAEtiep62RPRqSaWmUSTsicaZ/cCQDDr8aoRRY2inItydMsVetBVCjAEGyLI8Ofv724EOsixPueqYA9nHnMv+Pjr7mPhrrnUvcC9AjRo12pw5U/alx8XFHJ9CytEY/GtVKVOnK8syyHKeccLyRvzOYyzq8Zh7do5RT+SwTvT6c2qhrrVq9Mucy36VzrmWj4G+817HEOzP4l5P5Fr1iCSBwow5vGszBq37uFB27H3jd/a9NQN7pquKUmPU41cznKE7vspXbyg3HBYrM8JH524/uDpZ+RgYvPHTIj2krsVpd/B3vdvJPBf3n7Ry9iLi2JN/oA8on9o1BcVhsTKnyV1kxMTlxOMlg45KTaIYuv2rPDNuTs9ex6b7P8JhseK0OQjr1JheM6diqBzI7Hq3k3bymhCMEEQO7kDf+W/kaZM908zeN/7gxC9LkZ0ytcf3puXU2wvddrLCdLySZflbWZbbyrLcNjS0YubdGisHEt6laZnPsIUQFcrZA+x7e4ZHBaLDbCVm/mYyLyUW+DrmuGRiFm31eJV2ZFrY9/YM/OtUdZMBvhohSbkKcsVuOqBYzJWrHQkp7H3jjxxnD677SY+5zPEflcNJBSF+R+4hFI2vEUNIANX6t2XQ+k9KxNmDq7hp8IZPiOjVCkmnQeg0hLZvyOANn1Z4Zw9wZs4Gsi4nuy2+Oi02Uo6d4+LqPXmeGzW6O7dc/Juh275i7KnpDFz1IcbQIIQQdPrqMbfCMaHVoPM30fade/O1SetjpM0bdzPu3CxuufAX7T94oMR7DJdESOc8cHUCdPXsbUrHnBNCaIFAIKEExlapwKQcjVF0tpJBR0bMZXyuUqu0Z5oxx6fgExHiJrUAkBWbhEav8wg/AGScjUUf4Eujh0dy5It5bs5YMugYsPJ9lg98Hlua5+xZazIUKrc6busRJL1O8cETs2BLkYu9NEY9slP5wVOla1NuWlw6S2K+1ULpv+xd7JlmZIez2GGp8kT8ruOKsthOq42kvdFU7Z23No+k1Sjmx1fr19ZVOPbun6QcjSG0QyOaPXNLuekTXBIOfztQTwhRC5djvwW4NmduPnAHsBkYA6y6nuL3KkUjrGMjUg6f8ejG5LTYcuLETpudrY9/wfGflrh6huq0tH7jbho9ODzneP86VZFlz5CM0EiEd2sGQNu378G3eigH3p+JJSGNyh0a0v79+wlpVY/6kwdx5Kv57uEgo556dw4o1P0YKwcqL6ZKAp+IkEJd62pCWtXFUMnPw0FpfY15VoyXFBUpXbWgBNWvjtbX6KFkqTHo8a9TtVjXDmlZl57TXyrWNUqLYscAZFm2A1OApcBhYJYsyweFEK8KIYZlH/YDECKEOAE8AXikbqrceDR7djxaHwNcNYvW+hhp+NDwnA5Y2576muM/LcWRZXU12k7JYMcz33B69rr/zjEZaDVtkutaV5AktH4mWrzo0j8SQtD44ZHcfOZPbk9fyMCVH+To6Ld+426q9m2Nxqh31QCY9FTp1bJAr+FXU7ldA3wigj1CaxqjnkZTRuR7viU5neRDp7FfU0gnJIm+/76BoXKgSyXS1+h6IN01kBrDuxTKxoKQHnOZmAWbc21Sfz1Q65beaEwGt4wboZHQB/tTvQTqGMoraqWtildJOniaHc9+S+yG/RhCAmj65M00fGAYQgjsZivTQ4Z7Kg0ClZrXZsSe79y2nZm7gX1vzyDzQgIRPVvQ6pU78K9d8NlayvFzpBw+S2DDyCJrEaWdvsSKIS+QfjoWoZWQnTIdP51CvUm5vy04LFY23fcRJ2euRqPXIjtlmj9/K82fv9UtpOSw2ji/dDuW+FSq9GheqHsrCE6Hg433fMCpP1e7Ms1sdkJa1aPfgjdKPJbsDRxWGzHzN5EafYGQlnXxq1WFjfd84CqaFIKqfVrT9YenKkyGW26o0goqFZLMS4n8XXuCYkGKISSAW+PmeMGq/JFlmeRDZ7CmZBDSqm6+VdZrJ77F6Vlr3BaWtT5GOn35KHUn3lTa5uZw4KO/2fW/H92ypiS9jsihHen91yu5npdxPo64zYcwhlUivGvTcpk8kB5zmYVdHsaWkoE9W4MpoG5VBq75yKWvJIkCVQ87HQ6OfDGPw1/Nx55hpubIrrT83+3lque1Kq2gUiExhgai9TUqOvySyjcvDYQQOeqPeSHLMlse/pSTv6/w2GfPNLPv7ell6vAPfzbHo3rXabUR8+9mV8/ia2L5Of1av5iH0GkBGWNwAANWvl/st4/YTQc59u0CLIlpRI3pTq1behWrAnvDXe+RdTExZ73Inp5F8uGz7PrfT3T8ZEo+Z7vuNW7LITZP+ZTkQ2dyEgSOfvMvZ+dvYuT+HypEw5zy9yhWUclG0mho9/797hXMQrjS196823uGlRAnp6/k+M9Lc92fFZtcdsaAYqbSFa5dVwBXauPRr//FYbZiT8vEnpZFesxlVgz/X7HsOPDhXyy96WlO/LacmAWb2fzQJyzp/aRH29GCYs+ycGntXsXkgJPTV+Z7vsNiZelNz7Ck71Mk7j7hlg3mtNoxx6Vw4rflRbKtrFEdvkq5pt4d/ek1cyqV2zfEGF6J6oM7MHjjp3k2Lq8oHPrkH48Z9dWEdWpUhtZAtf5tFdt7+tUIV1R8PPy5Qr9Wp0zaqYuulNsiYE5IYddLP7g+l+xwsz3DTOLe6Fz7HOdLHmHr3NJdr+bAh39xedNBxbUkAEemmdh1e4tmWxmjOnyVck/k4I4M3fIF4y/+TeevHsMcl0xaCQpKeQtrakau+zQ+Btq8dU8ZWgNt3pyMPsgPTbashdBq0Poa6fL9U4r1CLYUZfsljSbPewNXLPzgx3/zd73bmRExhg2T3yPzQjyx6/YjKYRu7Blmt8yswqD1MRLWqbGHBo6k01Lr5p75nn/sh8UeEuRu19Hr8K9TrUi2lTVqDF+lQuC0O9hw17uc/mstklGP02Kjat/W9Jw5tcJp2V+h5siuHPx4tkfBmGTQMXTrlwVaByhJ/GqEM+rQTxz+aj6x6/cT1KgGjR4eSWAu2jlRY7qTfPiM5xqLJAhpWTfPsTbc/R6n//5Pk+bEr8s5t3Arnb56DJQm3ZIokK58bnT98RkWdn4Ye5YZe7oZrb8Jn4iQAoUGZQVFWjfTdBoa3Fsy8sWljerwVSoE+96ezul/1uOw2HL62l5YsYttT3xJ568e97J1RaPZM7dwauYasmKTcGRZEBoJyaCj15//KxVnn3zkLFsf+4JLa/ei9TFQf/JgWr86yS07xRgaRKupEwt0vUYPj+TE78vJOHMZe6bZZb9eR9fvn+Liqt1sf+ZbUo7F4FutMq2mTaLOhL6AK3X19Ky1bg8K2e7AmppB8uGzaEx6j/UEjVFPg/uGFPneA+pUZczJ3zn911rSoi8Q3LIuNYZ19qjaVqL2+N6KD2aEwC8qnO6/PIdfjfAi21aWqGmZKhWCGVXGYL6c5LFdY9Rze/rCcpkKWBBsaZkc/2kJ55ftwC8qnEYPjciz32lRybyYwD+N73QJsGX/zWtMeqr2a0Pfua8X+br2LAsn/1hBzKKt+FYPpeH9Q8k4H8/KEVPdwiAaHwMdPnqQBvcM4cycDay/8x1FMbjqA9vT5q3JLOv/LPZMCwjXwmi79+6j0UMjimxncbCmZrCwyyOkn4nFnp6FxmRAaCR6zZpKtf7tyl3LTDUtU6XCk1sGicNqw2l3oNFXTIev8/eh8SOjaPxI6TZVP/xltnTEVRM8R5aVC8t3knriPAF1ixaD1ppcbwpXd2Rae/vbHjFvR6aFnS/+SP3Jg/GtGeaRMQMgdFoCGkQS3LwON5+bSez6/dhSMwnv3jyn8tob6AN8Gb7rG87O20jspoP416pCnQl9ixVi8haqw1epEIR3a8aF5Ts9Mi4qNa1VKh2y8uP88h1sf/obUo6cxRQRQquXJ+ZZTettEnYcURSXk3Q6kg6eLrLDVyL1mHKGjjU5HXt6FgH1qyu+kUk6Tc4sXtJoiOjZssRsKi6STkvUmB5EjenhbVOKRcWcFqnccHT46EF0/qacdpFXMkg6f/VYmdtycfVuVo6YStK+kzitdjLOxLJ5yqcc/OyfMreloFRqXkex1abTZiewQdFkJHLDLxdlSJ2fCa2vka0Pf+bZ3UxAnQl9CSimcJlK3qgOX6VCENSoJiP2/0CjKSMI69KU+pMHMWzXN4R1alLmtux44QfFkMWeV37BmUvHI2/TaMoID+kAjUFHWJcmijK/xaH1a3d6tPvU+hpp/sKtyA4nJ/9c7fm2IcO5hVtL1A4VT9SQjkqFwS8yjPbvP+BtM0g5clZxuz3TgjU5HWNIyemqyLJMxrk4JK2mWBLLfpFhDFz7EZsf+oS4LYfRGHTUmdiPDh8+WGK2XqHmiK50+e5Jdjz7HZnn4zEE+9P8xQk0eXR0jra+Evnl7heXlKMxnJ23EUmnpebobhUms6YkUR2+SoXGkpjKtie+4tRfa5GdTmoM60yHT6a4NU9RImHPCbY/9TVxWw6jD/an6VM30/jhkQXKuPCvXZXE3cc9tmsMuhJVlYzfdYy1E94g48xlkGWCmkTRc+bUIoc9QlrWZcjGz3A6HAhJKtXskjrj+1BnfB8cVptLnCx7LJ2vicD61Uk+dE37UsmlVlla7H71V/a/PcP1sJEEO1/8gQ6fPUyDuweV2pjlETUtU6XC4nQ4mNtsMmnRF3J0VoRWg09EMKOO/pprI/SUozHMb3u/myyA1sdIgweH0f7d+/IdN2bRVlbfPM29F6/JQHCrutjTsvCNDKXpUzcXa9HRkpjKX7UnuKcuSgJTWCXGnp7ulYXqkiJ24wGWDXgWh8WGbHcg6XVofQwM2fpFrkVexSFxXzQLOj3sEYbTGPWMOflHvpODikaF6WmrolIYzi/dQca5ODdRLdnuwJKUzpl/1ud63t43//CoDrVnmjny+VysKen5jhs5qAPdfn4W35phIAS6IF+EJEjYcZSk/Sc5t2gry4e8wLGfit7HNvqPFTht16wHOGXsGVnELNhS5OuWB8K7NGXYrm+of89gqvRoTpPHRjPywA+l4uwBTs1ag/PaRWIASRAzf1OpjFleUUM6KhWW5EMKZf24pG+TDpzK9bz47UcV48iSXktq9AUqt66f79i1xvSg1pgeOG12tjz+Bce+W4h8lYN2ZFrY9sRX1JnQt0iz8bTTsYr6LQ6rncxzcTjtDqxJaegr+SNpNciyzMk/VnDg49lYk9KIHNyRFi9OwBRePmevgfWq0/mLR4t0rj3Lwo7nvuP4T0twZFkI796cTp89TFDjqFzPKaeBjDJHneGrVEicDgfGkIAcoS83NBKnZq5m//szsWV4NqoObFTDra1izjWt9kIv5Ek6LReX73Rz9leQnU5Sj51TPC/pwCmO/7SECyt2KvbBDe/UGK2CvrrQSKScOM/00JHMrHEL0yuPYO/b09nx7LdseuBjEncdJ/3UJY5+s4B5re7DnJBSqPupCKwa9TLHvluIPT0L2eHk0pq9LOj8MJkXExSPrzW2JxqDwkPXKRM5rHMpW1u+UB2+SoUj+o8V/Bkxhk0PfeJq7H2t83Y4ST91id0v/+wSzLrmLaDFCxPQmK5JUTQZiLq5R5E6F5nCKylul612DNdcz2l3sGrMy/zb4SG2PPIZq0a/zN/1JpJxLs7tuBrDu+BfO8LNUWlMBvxqhHHipyXYUjJwWmzYUjPZ+9rvHPx4ttuahNNmx5qczpGv5hf6fsozyYfPcGndPvc3O1nGYbFx+Evlew1uUYdmz4xDY9IjdBqETovQSFRu14DLmw7izEcc7XpCdfgqFYoLq3az8b4PscSn4rzyR59Lsokjy0rayYuc+nOV2/bKberTZ86rBNSrhtBIaHwMNLh3MF2+fbJINjV9epx7A3Vc4aEqPVt4LAge+vQfzi3ZjiPLgj3DjC0ti4yzsay91V3PRtJpGbz+E5o8NQ6/WlUIqFeNli9PxJqa6aFB78iyICvk/zvMVi6s2FWkeyqvpByJQdJpPLY7LTYSdx3L9bxWL9/BsJ3fUHt8H9eviySIXb+f9ZPeZlGPx3BYlLXurzfUGL5KhWLfm394Ng1xyqCR0JoMrhn/VdgzzMQs3Oohe1CtX1tGH/0Ve6YZyaBD0ng6kYJSY2hnWk2bxK6Xf0bSanBa7YR1bkKP6S95HHv0mwUe9ssOJ3Hbj2KOT3F7w9D5+9DmtTtp89qdOdt2vfiDshEKMWohSfjXiijaTZVTAhtGei5m45KUDs5j7UWWZS6t20v0b8vdAvr2dDNJe6M59sNiGj04vFRsLk+oDl+lQpFb4xNJr1VcmRNaDT5Vcy9YurZPa1Fp+uTNNLh/KMmHzmAKr5TrWoDSIjO4+uDmtu9qAhtEeuaw47pPcNdul4w6mjw2uiDmex2nw8G5RVu5uGYvPhHB1L29H6bwYOxmK7LdkdMv1hReCZ2fyTPF0qCj0QNDc73+wU9ms+uFHxR/R+yZFk5OX6k6fBWV8kZYl6ZknL3skWUjJIE+yM8lqXvVH7Wk09Lw/twdQUmi8zUR2q5hrvvNCSmgUY4/+VSrjE+1yvmO0e79+1k1+hXFDB7Z6XT1/DXp0fqa6PztEwS3qFPwG/ASdrOVpX2eInH/SZf8sFHPnld+JbhVXeK3HUGWZSo1jaLL90+x6Z4PsSa7p84KjUTv2dPwqar8+TntDvZO+zXPB+q1UhDXK2oMX6VC0fJ/t7tm5Ve1q9P6GGn9yiQGrPqAgHrV0Poa0QX4oAv0pdsvz5aKvnxRWD74BTLPxXts1/gY6P7rcwWqfK0+oD19/32D0A6N3PrPynYHOGUkg4769w5h3IVZ1BzepUTtLy2Ofj2fhD0ncsJxDrMVe6aZyxsP4LTZke0OEvdEs6j7YyQfOevRzFxoNFxctTvX61sSU3Oa5iih9TXS8N6iN1epSKgz/EJgTUnHaXeUqFaKSuEIrFedIVu/YPfUn4hdvx9TRDDNn59ArbEu2dpRh38m+dAZ7OlZBLeqW24qUhP3nyTpwCnPdnkCao7sVigRuKq9WxG84A1mVrvZ403HabZydt7GAmnkxG0/wrrb3iQ1+gKSVkutcT3p+v1TBeoCVZKc+G15nj1jr+Cw2kEho8ZptZF82DPMdQVDJX+ETgNmhZ1CUOf2ftQc3b0wJldYVIdfADLOx7Fu4ttc3ngAgMAGNej2y7P59u1UKR2CGtag16yXFfcJIcq8F2xByIiJQ9JpcXCNY5MhK5f88byQ7Q7FWgJw1RPkR9LB0yzoOCUn/OW02oj+bTkJu44xcv+PhbanOBT4AZNH+qRv9dA8r9/sqXHsf+dP7Jlmt+2dvnqM+ncNLLCtFR01pJMPToeDRd0fJ3bdPpxWO06rnaT9J1nc83HM8ddfUYtK6RDSso5yDFlAcMvCx9lN4cGKImqS3jVTz4+N936guICZfPAMl9bvK7Q9xaHBvUPQ+ua/eC7yyKTKTYP/Ci1euo2WL09EH+zv6kVbM5weM166oZw9qA4/Xy6u2IU5Ptnz1dnm4PjPS7xklUp5QJZlbGmZipWy1+JTtTI1hnZSuAicmbOBoogYdv/teXQBPmhMrgVHrZ8J/1oRtHjp9nzPTdp3Mtd9e9+cXmhbikPdO26i+qAOaHwMSAYdWn8Tkk6LZPhv5i+0GjQ+eiSFymqtr5FKTWvlOYYQgmZPj+PWuDncYV7C2FPTiRrVrcTvpbyjhnTyIe30JUXdFUeWhdQTF7xgkUp5IPqPFWx/+hvM8SlofQw0fWocLV64Nc9m6lofoysMc41zN8elEL/tCKEdGhXKBld1cE/ithzCEBxA3Un9qXNrnwKtW2j9TR4FXFdIzyX1tbSQNBp6zZxKwu7jxG44gCm8EtX6t+XAh39z7PuFOLKsRA7pSPPnxrOw66NYLbacz1BoNZgiQqjat2DSykIIRBmvUZQnbtw7LyAhreshFEo5tX4mwjo19oJFKt7mzLyNbLzvw5wCKltqJvvfno7sdNJq6sRcz8u8lKhcKyBJhQ4PxizYzOpxr+VksWh9jWRdSCBqZNcCOfwmj41h53PfKe4LqF86qpX5EdKqHiGt6gGuNQZTWBAdPp5C5NBOOVLXgzd8wvo73yVh9wkAqt3Uhi7fP53ng1blP1SHnw+h7RoS1rkxsRsP4MhyxWAlvRZjaBC1bu7pXeNUvMLul3/2qJa1Z1o48MEsWrwwAUmrHGuOHNKR2A37Pc51WKyFmt077Q7W3/muW2aLPcNMesxlDn70N61emZTvNZo/cwv735+F9ZoHjWTU0+RR7xVryU4nG+56j1N/rQVkhFaDpNMycOUHBLeoQ1DjKIZu/RJbehYiu7q6PJB85Cwnp6/EYbZSc2RXQjs2LtUGM0VFfSwWgL7/vkGzZ8fjGxmKMbwS9ScPZui2L8rNL5tK2ZJ+OlZxu9Niz1NPv96dA/CLDHMTbtP6GmnxwoRCibYlHz6jmFfutNg4/fe6Al9n+M6vCWhYA8moQ+NnQmPU0+a1O0u181R+nJyxitOz1+HIsuDIsmJPy8KamMaK4S+5rXPo/Ezl5u/v8JfzmN/mPva9PYMDH8xiab+n2fTAx0Valylt1Bl+AdAY9LSaOjHP13WVG4egJjWJ23zIY7vO34Shkn+u5+l8TQzd9iVHvv6XM/+sRx/sT+MpI6g+sEOhxtf5+3jm82djTkhFdjoLFOLwiwxj1MEfSdp3EnN8CpXb1i+RFo2ZlxLZ+/rvnFu0BX0lf5o8NoY6t/Ut0Iz36HcLFdcWLIlpJO6NLnep0FmxiWx/6mu3DCx7poWTf6yg7m19Ce/azIvWeaI6fBWVQtL2rXtYNug59xaHPgbavHl3vo5W5+9Ds6fH0ezpcUUe3z+qCkGNa7ri2NfMIi1JaRz48G+aPXVzga4lhCiS/IKcLUmsMejcHLk5IYX5rV06/LLNAadj2fzAxyTuP1mg9pG51RAIITwqbMsD5xZtc6t4voI908KpWWvKncMvVkhHCBEshFguhDie/b+iMLgQwiGE2JP97/oS6Fa54ajSvTk3LXyLyh0aovU1ElC/Ol1/eJoG95RdeX7Pv15GSSJTtto59PHfpTauLMvsf38m0yuP5He/wcyKHEf09BU5+498OR9rcrpbQ5gr7SPNccn5Xr/ObX08pKYBJJ0mZ0G3PCHpNMoFcEIgKTVd8TLFjeE/B6yUZbkesDL7eyWyZFlumf1vWDHHVFHxOlV6tGDo5i+4PW0ho4/8Qu1xvcp0fJ8qwblW2lqS8u/LW1T2vzeTPa/8gjUpDdnpJPNCAutue4tVY1/BYbVxYeUuxQIzyaDLyazJi/p3DyK4df2cbl+SwdXgvMeMl3JdDPcm1Yd0Ukzb1hh01JnQ1wsW5U1xQzrDgZ7ZX/8CrAGeLeY1VVSuO5IPnebgx7NJOXaOKt2b02jKCExhyp2yCoLWZCCwQSQph8967AvrUnBdnsIgO53se2u6S5H0Gs78s55VY15RbiWIqwNXQdRANQY9A1d/wPnF2zi/fCemiGDqTbwpVyVMb2MI8qPH9BdZe+sbCEkgO2WQZVq+PLHcrTcAiOKsJAshkmVZDsr+WgBJV76/5jg7sAewA2/Lsjw3l+vdC9wLUKNGjTZnzuQuiKSiUlE4v3wHq0ZOxWGxITucSAYdOj8Tw3Z9g19kWJGve3HNHpYPeQGn2eZaqNVq0Bj1DN74KcHNapfgHbiwpmYwI3SkYgMSACSBRq/zmOGL7HDM0C1flLhN5QVLYipn523CYbFRfVD7QvdGLkmEEDtlWW6ruC8/hy+EWAEoCVW8CPxytYMXQiTJsuwxbRFCVJNl+bwQojawCugjy3J0XuO2bdtW3rFjR562qaiUd2RZ5q9at5Jx9rL7Do1E3Ql96fZz8V6IE/efZP87f5J88DSV2zek2TO3KGrslASy08mMKqOxxKcW6rwqPVvSa9bUIvULVik8eTn8fEM6siznGogSQsQKISJkWb4ohIgALisdJ8vy+ez/Twoh1gCtgDwdvorK9UDWpUTMl5M9dzicRE9fSfuPHswzlTM/gpvVpsfvL+R5jDkhhbToi/hFhRcrjCQkiTZv3M2mhz4Be/76QeCqSG80ZYTq7MsJxV20nQ/ckf31HcC8aw8QQlQSQhiyv64MdAE8k5hVVK5DdH4mV1xXAdnuYMPd75Xa2E6Hg80PfcLM6uNYetPTzKo5nnV3vF2s9MYG9wyhwwcPei4YS8oLyEISucb1Vcqe4jr8t4F+QojjQN/s7xFCtBVCfJ99TCNghxBiL7AaVwxfdfgqNwQ6fx+qD2yf6/5zi7a6abSXJPvfncnxX5bitNiwpWbmVOLuzK0RegFp/PBIhu34ipC29RGShKTXUrVfmxzVTncEEV6s3FVxp1iLtqWJGsO/sXE6HJyauYbo35cj6XXUv3sgkUM6lUt9kvywJKczI2yUYnWspNdyy4W/MAQHlPi4M6qMwXw5yWO71tfIbakLSuSzdFisLr0bjYbdr/zM/ndnuorPNAJk6PfvG1Tp0aLY46gUnGLF8FVUyhpZllk54n9cWrM3p8z+4spd1Jl4E52/eNTL1hUeQ5AftW7pxcnpK+Ga8I5/7aql4uwBrMlpitvtmRZkhyurp7hoDP/pArV6ZRL17hrIhWU70PqZiBzSCV12Pr1K+UAVT1Mpd1xYsZNLa/e5aarYM8yc+HlJnr1LyzNt35yMMSQwRzhN0mnR+hrp8v1TpTZmaHtlBc5KTaNKrYjJr0Y49ScPpvYtvVVnXw5RZ/gq5Y7zS7djT8/y3CHDhZW7CWpUs+yNKia+1UMZdfgnjn63kNgNBwhsGEmjh0bgn09rvuLQ/qMHWdzzcRxmq2tGL0lojHo6fv5IqY3pDSzJ6Rz7fiGX1uwhoF51Gj00goC61bxtVrlEdfgq5Q5DsD+SQYfzGglgodVgqFR8NUdvYQgOoPmz48usFr1ym/oM2/E1+96eQcLOYwQ1q0Xz58aXSFFW8pGzpJ++RHDz2l6tgs28lMj8NvdhTc7AkWVB6HZy7LtF9P33dSJ6tfKaXeUVddFWpdyRfjaWfxrd6dbgA1wZL+POz1JDBV7EmpLOiuH/I37HUSSdFqfFRp3b+9H5q8e80nVq4/0fcvzHJR4L4n5RVRgT/XuFXOQvLnkt2qoxfJVyh1+NcHrOeAmdvwldgA86fx8MIQH0W/y26uy9zIZ7PiBu62EcmRZsKRk4zFai/1jBoc/nesWemAVbFLOfsi4lknkh3gsWlW/UkI5KuaTGsM7cEvsPlzceQNJpCevcpFyqJd5I2DKyiJm/yUOz3pFp4fCn/9DkkVFlbpPOz4TCag+yLKP1VScH16LO8FXKLVqjnqp9WlOle/Prwtmnn43l4po9ZMUmetuUIuHItOQqyWxNyShja1w0mjICjYJ+PsCuF38okAb/jYTq8FVUShhZlrGmpOOwuFQj7VkWVoz4H/80nMTKkVOZFXUrG+//EKcjF9XJEsRutnJh5S4urduHM5e2iAXFUDlQeYFWkqjWXzFkXOo0enA4tcf1QtJr3eQdnBYbx75fxPy292NLy/SKbeURNaSjolKCXFq7l433fkj66YsgSUSN6Y6k1XBh2Q4cZmuOdHD07ysIqFOtWK0O8+PMvI2sn/iWa1Yug2TQ0mfu64R3LppevhCCLt89ycphL+GwWN2kntu8MbmErS+wUVe+8Chqc9rsWBLSOP7LUhpPGVn2tpVD1CwdFZUSIvnIWf5t94BbwZhk0Lli3gp/Zz7VKjMuZmap2JJrplNAdqZTMeLbyUfOcvDj2aQejSG8W7NiN3MpDqdnr2P9pHcUG59foebo7vT+6+UytMq7qNIKKiplwMGP/vZo/nFtLcHVlGbcO/r35Yqt92SnzNl5m6hza58iXzuoYQ26fP14ccwrMY79sChPZy/ptQTUU4uwrqDG8FVUSojkw2cUnayidLAQVOnerNRsMSek4bR6PmxkuwNrkrLGTkUk1+5b2Ug6LQ3vG1pG1pR/VIevolJChHVqgqT31H6XtBo0Jj1C4/pzk3RadP4m2r13f6nZUn1AO7S+Rs8dQlC1b5tSG7esqTvxJuX7BPxqVeGmxW/jV9N77QbLG6rDV1EpIZo8Ogqtj8FtRq/xMVDn9n4M2/4VdW7vR+V2DWlw/1BG7P2+VDWBqvZpTZUezd2codbXSN1J/QlsEFlq45Y1tcf3dt2nn+s+JYMOyaSn26/PMebE74R3Lb23qIqIumirolKCpJ44z47nvuXCyt3oA31p/MgoGj86CklT9nUE11NPgbyQZZmLq3ZzftkOjKGB1Lm1j1f1fbxNsZqYewvV4auoqKgUHlVLR0XlBkeWZZIPnyFx/0lkZ8EakKtcf6hpmSoq1zmJ+6JZOfJlzLFJIIHO35deM/+nxrdvQNQZvorKdYw908ziXk+Sfuoi9kwz9nQzWRcTWDboOVVn5gZEdfgqKtcxZ+dtwmmze2yX7U6ip6/0gkUq3kR1+Coq1zFZlxI95IwBHGYrmRcSvGCRijdRHb6KynVMeLdmCK3nn7nWz0SVHi28YJGKN1EdvorKdUzltg2o1r8dWp//CrA0PgaCW9TxmqSxivdQs3RUVK5zes2ayvEfl3Ds+0U47XbqTryJhvcP9UoxmIp3UQuvVFRUVK4j1MIrFRUVFRXV4auoqKjcKKgxfBUVlTIjPeYyJ35eQuaFBKr2aU2NEV2viwb1FQXV4auoqJQJ55fvYOXIqcgOJ06Ljeg/VhL03kwGrvkIrcngbfNuCNSQjoqKSqnjdDhYO+FNHJmWnLaP9vQskg6e5sjX/3rZuhsH1eGrqKiUOkn7Tnr0+wVwZFo4+ccKL1h0Y6I6fBUVlVJHY9BBLingGjWcU2aoDl9FRaXUCWxUE1OVYLim25bW10iDe4d4yaobD9Xhq6iolDpCCPrOew1D5QB0/j5ofAxoTAaiRnenzoQ+3jbvhqFYWTpCiLHAK0AjoL0sy4qlsUKIAcAngAb4Xpblt4szroqKSsUjqHEU42Jmcm7xNsyxSYR3a1aqjdxVPCluWuYBYBTwTW4HCCE0wBdAP+AcsF0IMV+W5UPFHFtFRaWCodHrqDm8i7fNuGEplsOXZfkwuF7X8qA9cEKW5ZPZx/4JDAdUh6+ioqJShpRFDL8aEHPV9+eyt3kghLhXCLFDCLEjLi6uDExTUVFRuXHId4YvhFgBVFHY9aIsy/NK0hhZlr8FvgWXWmZJXltFRUXlRidfhy/Lct9ijnEeiLzq++rZ21RUVFRUypCyCOlsB+oJIWoJIfTALcD8MhhXRUVFReUqitUARQgxEvgMCAWSgT2yLPcXQlTFlX45KPu4QcDHuNIyf5Rl+Y0CXDsOOJP9bWUgvsiGli0VxVbVzpJFtbNkqSh2QvmztaYsy6FKO8ptx6urEULsyK2DS3mjotiq2lmyqHaWLBXFTqhYtqqVtioqKio3CKrDV1FRUblBqCgO/1tvG1AIKoqtqp0li2pnyVJR7IQKZGuFiOGrqKioqBSfijLDV1FRUVEpJqrDV1FRUblBKJcOXwgxVghxUAjhFELkmu4khDgthNgvhNgjhFCUZi5tCmHrACHEUSHECSHEc2VpY/b4wUKI5UKI49n/V8rlOEf257lHCFFmBXL5fT5CCIMQYmb2/q1CiKiysu0aO/Kzc5IQIu6qz3CyF2z8UQhxWQhxIJf9QgjxafY97BNCtC5rG7PtyM/OnkKIlKs+y6llbWO2HZFCiNVCiEPZf+uPKhxTLj7TfJFludz9w6Wv3wBYA7TN47jTQOXybiuugrNooDagB/YCjcvYzneB57K/fg54J5fj0r3wGeb7+QAPAl9nf30LMLOc2jkJ+LysbbvGhu5Aa+BALvsHAYsBAXQEtpZTO3sCC7z5WWbbEQG0zv7aHzim8HMvF59pfv/K5QxfluXDsiwf9bYdBaGAtuZIRMuybAWuSESXJcOBX7K//gUYUcbj50VBPp+r7f8b6CPy0eUuBcrDzzFfZFleByTmcchw4FfZxRYgSAgRUTbW/UcB7CwXyLJ8UZblXdlfpwGH8VT8LRefaX6US4dfCGRgmRBipxDiXm8bkwcFloguRcJlWb6Y/fUlIDyX44zZEtVbhBAjysa0An0+OcfIsmwHUoCQMrFOwYZscvs5js5+rf9bCBGpsN/blIffx4LSSQixVwixWAjRxNvGZIcSWwFbr9lVIT7T4na8KjIlJLvcVZbl80KIMGC5EOJI9qyhRClLiejikJedV38jy7IshMgtH7dm9mdaG1glhNgvy3J0Sdt6HfMvMEOWZYsQ4j5cbyW9vWxTRWUXrt/H9Gw9rrlAPW8ZI4TwA2YDj8mynOotO4qD1xy+XHzZZWRZPp/9/2UhxBxcr9wl7vBLwNYykYjOy04hRKwQIkKW5YvZr5qXc7nGlc/0pBBiDa7ZTGk7/IJ8PleOOSeE0AKBQEIp23Ut+dopy/LVNn2Pa+2kvFEhJMuvdqqyLC8SQnwphKgsy3KZC5UJIXS4nP0fsiz/o3BIhfhMK2xIRwjhK4Twv/I1cBOuHrvlkfIgET0fuCP76zsAjzcTIUQlIYQh++vKQBfKphVlQT6fq+0fA6ySs1fLypB87bwmbjsMV7y3vDEfmJidWdIRSLkq3FduEEJUubJOI4Roj8tflfVDnmwbfgAOy7L8YS6HVYjP1Ourxkr/gJG4YmAWIBZYmr29KrAo++vauLIk9gIHcYVXyqWt8n+r+MdwzZbL3FZc8e6VwHFgBRCcvb0tLilrgM7A/uzPdD9wdxna5/H5AK8Cw7K/NgJ/ASeAbUBtL/2887Pzrezfx73AaqChF2ycAVwEbNm/m3cD9wP3Z+8XwBfZ97CfPDLhvGznlKs+yy1AZy/Z2RXXeuE+YE/2v0Hl8TPN758qraCioqJyg1BhQzoqKioqKoVDdfgqKioqNwiqw1dRUVG5QVAdvoqKisoNgurwVVRUVG4QVIevoqKicoOgOnwVFRWVG4T/A09TNHYQAYzGAAAAAElFTkSuQmCC\n",
  459. "text/plain": [
  460. "<Figure size 432x288 with 1 Axes>"
  461. ]
  462. },
  463. "metadata": {
  464. "needs_background": "light"
  465. },
  466. "output_type": "display_data"
  467. }
  468. ],
  469. "source": [
  470. "# generate the NN model\n",
  471. "class NN_Model:\n",
  472. " epsilon = 0.01 # learning rate\n",
  473. " n_epoch = 1000 # iterative number\n",
  474. " \n",
  475. "nn = NN_Model()\n",
  476. "nn.n_input_dim = X.shape[1] # input size\n",
  477. "nn.n_output_dim = 2 # output node size\n",
  478. "nn.n_hide_dim = 4 # hidden node size\n",
  479. "\n",
  480. "nn.X = X\n",
  481. "nn.y = y \n",
  482. "\n",
  483. "# initial weight array\n",
  484. "nn.W1 = np.random.randn(nn.n_input_dim, nn.n_hide_dim) / np.sqrt(nn.n_input_dim)\n",
  485. "nn.b1 = np.zeros((1, nn.n_hide_dim))\n",
  486. "nn.W2 = np.random.randn(nn.n_hide_dim, nn.n_output_dim) / np.sqrt(nn.n_hide_dim)\n",
  487. "nn.b2 = np.zeros((1, nn.n_output_dim))\n",
  488. "\n",
  489. "# define sigmod & its derivate function\n",
  490. "def sigmod(X):\n",
  491. " return 1.0/(1+np.exp(-X))\n",
  492. "\n",
  493. "def sigmod_derivative(X):\n",
  494. " f = sigmod(X)\n",
  495. " return f*(1-f)\n",
  496. "\n",
  497. "# network forward calculation\n",
  498. "def forward(n, X):\n",
  499. " n.z1 = sigmod(X.dot(n.W1) + n.b1)\n",
  500. " n.z2 = sigmod(n.z1.dot(n.W2) + n.b2)\n",
  501. " return n\n",
  502. "\n",
  503. "\n",
  504. "# use random weight to perdict\n",
  505. "forward(nn, X)\n",
  506. "y_pred = np.argmax(nn.z2, axis=1)\n",
  507. "\n",
  508. "# plot data\n",
  509. "plt.scatter(X[:, 0], X[:, 1], c=y_pred, cmap=plt.cm.Spectral)\n",
  510. "plt.show()"
  511. ]
  512. },
  513. {
  514. "cell_type": "code",
  515. "execution_count": 6,
  516. "metadata": {},
  517. "outputs": [
  518. {
  519. "name": "stdout",
  520. "output_type": "stream",
  521. "text": [
  522. "epoch [ 0] L = 103.723994, acc = 0.500000\n",
  523. "epoch [ 1] L = 99.572368, acc = 0.500000\n",
  524. "epoch [ 2] L = 96.298444, acc = 0.695000\n",
  525. "epoch [ 3] L = 93.755262, acc = 0.745000\n",
  526. "epoch [ 4] L = 91.723534, acc = 0.725000\n",
  527. "epoch [ 5] L = 90.013431, acc = 0.710000\n",
  528. "epoch [ 6] L = 88.493838, acc = 0.720000\n",
  529. "epoch [ 7] L = 87.084077, acc = 0.740000\n",
  530. "epoch [ 8] L = 85.737551, acc = 0.755000\n",
  531. "epoch [ 9] L = 84.428420, acc = 0.760000\n",
  532. "epoch [ 10] L = 83.142909, acc = 0.770000\n",
  533. "epoch [ 11] L = 81.874127, acc = 0.775000\n",
  534. "epoch [ 12] L = 80.619101, acc = 0.775000\n",
  535. "epoch [ 13] L = 79.377104, acc = 0.775000\n",
  536. "epoch [ 14] L = 78.148705, acc = 0.775000\n",
  537. "epoch [ 15] L = 76.935216, acc = 0.785000\n",
  538. "epoch [ 16] L = 75.738363, acc = 0.790000\n",
  539. "epoch [ 17] L = 74.560076, acc = 0.795000\n",
  540. "epoch [ 18] L = 73.402344, acc = 0.795000\n",
  541. "epoch [ 19] L = 72.267120, acc = 0.800000\n",
  542. "epoch [ 20] L = 71.156245, acc = 0.805000\n",
  543. "epoch [ 21] L = 70.071400, acc = 0.805000\n",
  544. "epoch [ 22] L = 69.014062, acc = 0.810000\n",
  545. "epoch [ 23] L = 67.985488, acc = 0.810000\n",
  546. "epoch [ 24] L = 66.986694, acc = 0.810000\n",
  547. "epoch [ 25] L = 66.018452, acc = 0.810000\n",
  548. "epoch [ 26] L = 65.081296, acc = 0.810000\n",
  549. "epoch [ 27] L = 64.175529, acc = 0.810000\n",
  550. "epoch [ 28] L = 63.301240, acc = 0.810000\n",
  551. "epoch [ 29] L = 62.458318, acc = 0.810000\n",
  552. "epoch [ 30] L = 61.646476, acc = 0.810000\n",
  553. "epoch [ 31] L = 60.865270, acc = 0.810000\n",
  554. "epoch [ 32] L = 60.114123, acc = 0.810000\n",
  555. "epoch [ 33] L = 59.392345, acc = 0.815000\n",
  556. "epoch [ 34] L = 58.699157, acc = 0.815000\n",
  557. "epoch [ 35] L = 58.033705, acc = 0.815000\n",
  558. "epoch [ 36] L = 57.395081, acc = 0.815000\n",
  559. "epoch [ 37] L = 56.782341, acc = 0.815000\n",
  560. "epoch [ 38] L = 56.194515, acc = 0.815000\n",
  561. "epoch [ 39] L = 55.630622, acc = 0.815000\n",
  562. "epoch [ 40] L = 55.089680, acc = 0.815000\n",
  563. "epoch [ 41] L = 54.570712, acc = 0.825000\n",
  564. "epoch [ 42] L = 54.072759, acc = 0.820000\n",
  565. "epoch [ 43] L = 53.594882, acc = 0.820000\n",
  566. "epoch [ 44] L = 53.136167, acc = 0.825000\n",
  567. "epoch [ 45] L = 52.695728, acc = 0.825000\n",
  568. "epoch [ 46] L = 52.272712, acc = 0.825000\n",
  569. "epoch [ 47] L = 51.866299, acc = 0.825000\n",
  570. "epoch [ 48] L = 51.475703, acc = 0.825000\n",
  571. "epoch [ 49] L = 51.100173, acc = 0.825000\n",
  572. "epoch [ 50] L = 50.738994, acc = 0.830000\n",
  573. "epoch [ 51] L = 50.391484, acc = 0.830000\n",
  574. "epoch [ 52] L = 50.056996, acc = 0.830000\n",
  575. "epoch [ 53] L = 49.734917, acc = 0.830000\n",
  576. "epoch [ 54] L = 49.424664, acc = 0.830000\n",
  577. "epoch [ 55] L = 49.125686, acc = 0.830000\n",
  578. "epoch [ 56] L = 48.837462, acc = 0.830000\n",
  579. "epoch [ 57] L = 48.559498, acc = 0.830000\n",
  580. "epoch [ 58] L = 48.291329, acc = 0.830000\n",
  581. "epoch [ 59] L = 48.032513, acc = 0.830000\n",
  582. "epoch [ 60] L = 47.782634, acc = 0.830000\n",
  583. "epoch [ 61] L = 47.541299, acc = 0.830000\n",
  584. "epoch [ 62] L = 47.308135, acc = 0.830000\n",
  585. "epoch [ 63] L = 47.082790, acc = 0.830000\n",
  586. "epoch [ 64] L = 46.864932, acc = 0.830000\n",
  587. "epoch [ 65] L = 46.654245, acc = 0.830000\n",
  588. "epoch [ 66] L = 46.450431, acc = 0.830000\n",
  589. "epoch [ 67] L = 46.253208, acc = 0.830000\n",
  590. "epoch [ 68] L = 46.062309, acc = 0.830000\n",
  591. "epoch [ 69] L = 45.877480, acc = 0.830000\n",
  592. "epoch [ 70] L = 45.698480, acc = 0.830000\n",
  593. "epoch [ 71] L = 45.525081, acc = 0.835000\n",
  594. "epoch [ 72] L = 45.357065, acc = 0.835000\n",
  595. "epoch [ 73] L = 45.194228, acc = 0.835000\n",
  596. "epoch [ 74] L = 45.036371, acc = 0.835000\n",
  597. "epoch [ 75] L = 44.883309, acc = 0.835000\n",
  598. "epoch [ 76] L = 44.734863, acc = 0.840000\n",
  599. "epoch [ 77] L = 44.590864, acc = 0.840000\n",
  600. "epoch [ 78] L = 44.451150, acc = 0.835000\n",
  601. "epoch [ 79] L = 44.315567, acc = 0.835000\n",
  602. "epoch [ 80] L = 44.183966, acc = 0.835000\n",
  603. "epoch [ 81] L = 44.056207, acc = 0.835000\n",
  604. "epoch [ 82] L = 43.932155, acc = 0.835000\n",
  605. "epoch [ 83] L = 43.811681, acc = 0.835000\n",
  606. "epoch [ 84] L = 43.694661, acc = 0.835000\n",
  607. "epoch [ 85] L = 43.580977, acc = 0.835000\n",
  608. "epoch [ 86] L = 43.470515, acc = 0.835000\n",
  609. "epoch [ 87] L = 43.363166, acc = 0.835000\n",
  610. "epoch [ 88] L = 43.258826, acc = 0.835000\n",
  611. "epoch [ 89] L = 43.157394, acc = 0.835000\n",
  612. "epoch [ 90] L = 43.058774, acc = 0.835000\n",
  613. "epoch [ 91] L = 42.962874, acc = 0.835000\n",
  614. "epoch [ 92] L = 42.869604, acc = 0.835000\n",
  615. "epoch [ 93] L = 42.778879, acc = 0.835000\n",
  616. "epoch [ 94] L = 42.690616, acc = 0.835000\n",
  617. "epoch [ 95] L = 42.604736, acc = 0.835000\n",
  618. "epoch [ 96] L = 42.521163, acc = 0.835000\n",
  619. "epoch [ 97] L = 42.439823, acc = 0.830000\n",
  620. "epoch [ 98] L = 42.360646, acc = 0.830000\n",
  621. "epoch [ 99] L = 42.283563, acc = 0.830000\n",
  622. "epoch [ 100] L = 42.208509, acc = 0.830000\n",
  623. "epoch [ 101] L = 42.135420, acc = 0.830000\n",
  624. "epoch [ 102] L = 42.064236, acc = 0.830000\n",
  625. "epoch [ 103] L = 41.994896, acc = 0.830000\n",
  626. "epoch [ 104] L = 41.927345, acc = 0.830000\n",
  627. "epoch [ 105] L = 41.861528, acc = 0.830000\n",
  628. "epoch [ 106] L = 41.797391, acc = 0.835000\n",
  629. "epoch [ 107] L = 41.734884, acc = 0.835000\n",
  630. "epoch [ 108] L = 41.673958, acc = 0.835000\n",
  631. "epoch [ 109] L = 41.614563, acc = 0.835000\n",
  632. "epoch [ 110] L = 41.556656, acc = 0.835000\n",
  633. "epoch [ 111] L = 41.500191, acc = 0.835000\n",
  634. "epoch [ 112] L = 41.445126, acc = 0.835000\n",
  635. "epoch [ 113] L = 41.391418, acc = 0.835000\n",
  636. "epoch [ 114] L = 41.339029, acc = 0.835000\n",
  637. "epoch [ 115] L = 41.287919, acc = 0.835000\n",
  638. "epoch [ 116] L = 41.238051, acc = 0.835000\n",
  639. "epoch [ 117] L = 41.189388, acc = 0.835000\n",
  640. "epoch [ 118] L = 41.141897, acc = 0.835000\n",
  641. "epoch [ 119] L = 41.095542, acc = 0.835000\n",
  642. "epoch [ 120] L = 41.050292, acc = 0.840000\n",
  643. "epoch [ 121] L = 41.006114, acc = 0.840000\n",
  644. "epoch [ 122] L = 40.962979, acc = 0.840000\n",
  645. "epoch [ 123] L = 40.920857, acc = 0.840000\n",
  646. "epoch [ 124] L = 40.879718, acc = 0.840000\n",
  647. "epoch [ 125] L = 40.839536, acc = 0.840000\n",
  648. "epoch [ 126] L = 40.800284, acc = 0.840000\n",
  649. "epoch [ 127] L = 40.761935, acc = 0.845000\n",
  650. "epoch [ 128] L = 40.724464, acc = 0.845000\n",
  651. "epoch [ 129] L = 40.687848, acc = 0.845000\n",
  652. "epoch [ 130] L = 40.652063, acc = 0.845000\n",
  653. "epoch [ 131] L = 40.617086, acc = 0.845000\n",
  654. "epoch [ 132] L = 40.582895, acc = 0.850000\n",
  655. "epoch [ 133] L = 40.549468, acc = 0.850000\n",
  656. "epoch [ 134] L = 40.516785, acc = 0.850000\n",
  657. "epoch [ 135] L = 40.484827, acc = 0.850000\n",
  658. "epoch [ 136] L = 40.453572, acc = 0.850000\n",
  659. "epoch [ 137] L = 40.423004, acc = 0.850000\n",
  660. "epoch [ 138] L = 40.393102, acc = 0.850000\n",
  661. "epoch [ 139] L = 40.363851, acc = 0.850000\n",
  662. "epoch [ 140] L = 40.335232, acc = 0.850000\n",
  663. "epoch [ 141] L = 40.307229, acc = 0.850000\n",
  664. "epoch [ 142] L = 40.279826, acc = 0.850000\n",
  665. "epoch [ 143] L = 40.253008, acc = 0.850000\n",
  666. "epoch [ 144] L = 40.226759, acc = 0.850000\n",
  667. "epoch [ 145] L = 40.201064, acc = 0.850000\n",
  668. "epoch [ 146] L = 40.175909, acc = 0.850000\n",
  669. "epoch [ 147] L = 40.151281, acc = 0.850000\n",
  670. "epoch [ 148] L = 40.127167, acc = 0.850000\n",
  671. "epoch [ 149] L = 40.103552, acc = 0.850000\n",
  672. "epoch [ 150] L = 40.080424, acc = 0.850000\n",
  673. "epoch [ 151] L = 40.057772, acc = 0.850000\n",
  674. "epoch [ 152] L = 40.035583, acc = 0.850000\n",
  675. "epoch [ 153] L = 40.013846, acc = 0.850000\n",
  676. "epoch [ 154] L = 39.992550, acc = 0.850000\n",
  677. "epoch [ 155] L = 39.971683, acc = 0.850000\n",
  678. "epoch [ 156] L = 39.951235, acc = 0.855000\n",
  679. "epoch [ 157] L = 39.931196, acc = 0.855000\n",
  680. "epoch [ 158] L = 39.911556, acc = 0.855000\n",
  681. "epoch [ 159] L = 39.892306, acc = 0.855000\n",
  682. "epoch [ 160] L = 39.873435, acc = 0.855000\n",
  683. "epoch [ 161] L = 39.854934, acc = 0.855000\n",
  684. "epoch [ 162] L = 39.836796, acc = 0.855000\n",
  685. "epoch [ 163] L = 39.819011, acc = 0.855000\n",
  686. "epoch [ 164] L = 39.801570, acc = 0.855000\n",
  687. "epoch [ 165] L = 39.784466, acc = 0.855000\n",
  688. "epoch [ 166] L = 39.767691, acc = 0.855000\n",
  689. "epoch [ 167] L = 39.751236, acc = 0.855000\n",
  690. "epoch [ 168] L = 39.735095, acc = 0.855000\n",
  691. "epoch [ 169] L = 39.719261, acc = 0.855000\n",
  692. "epoch [ 170] L = 39.703725, acc = 0.855000\n",
  693. "epoch [ 171] L = 39.688481, acc = 0.855000\n",
  694. "epoch [ 172] L = 39.673523, acc = 0.855000\n",
  695. "epoch [ 173] L = 39.658844, acc = 0.855000\n",
  696. "epoch [ 174] L = 39.644437, acc = 0.855000\n",
  697. "epoch [ 175] L = 39.630297, acc = 0.855000\n",
  698. "epoch [ 176] L = 39.616417, acc = 0.855000\n",
  699. "epoch [ 177] L = 39.602791, acc = 0.855000\n",
  700. "epoch [ 178] L = 39.589414, acc = 0.855000\n",
  701. "epoch [ 179] L = 39.576281, acc = 0.855000\n",
  702. "epoch [ 180] L = 39.563385, acc = 0.855000\n",
  703. "epoch [ 181] L = 39.550721, acc = 0.855000\n",
  704. "epoch [ 182] L = 39.538285, acc = 0.855000\n",
  705. "epoch [ 183] L = 39.526072, acc = 0.855000\n",
  706. "epoch [ 184] L = 39.514076, acc = 0.855000\n",
  707. "epoch [ 185] L = 39.502292, acc = 0.855000\n",
  708. "epoch [ 186] L = 39.490717, acc = 0.855000\n",
  709. "epoch [ 187] L = 39.479346, acc = 0.855000\n",
  710. "epoch [ 188] L = 39.468173, acc = 0.855000\n",
  711. "epoch [ 189] L = 39.457196, acc = 0.855000\n",
  712. "epoch [ 190] L = 39.446409, acc = 0.855000\n",
  713. "epoch [ 191] L = 39.435809, acc = 0.855000\n",
  714. "epoch [ 192] L = 39.425392, acc = 0.855000\n",
  715. "epoch [ 193] L = 39.415154, acc = 0.855000\n",
  716. "epoch [ 194] L = 39.405091, acc = 0.855000\n",
  717. "epoch [ 195] L = 39.395199, acc = 0.855000\n",
  718. "epoch [ 196] L = 39.385476, acc = 0.855000\n",
  719. "epoch [ 197] L = 39.375916, acc = 0.855000\n",
  720. "epoch [ 198] L = 39.366518, acc = 0.855000\n",
  721. "epoch [ 199] L = 39.357277, acc = 0.855000\n",
  722. "epoch [ 200] L = 39.348190, acc = 0.855000\n",
  723. "epoch [ 201] L = 39.339255, acc = 0.855000\n",
  724. "epoch [ 202] L = 39.330468, acc = 0.855000\n",
  725. "epoch [ 203] L = 39.321826, acc = 0.855000\n",
  726. "epoch [ 204] L = 39.313326, acc = 0.855000\n",
  727. "epoch [ 205] L = 39.304965, acc = 0.855000\n",
  728. "epoch [ 206] L = 39.296741, acc = 0.855000\n",
  729. "epoch [ 207] L = 39.288650, acc = 0.855000\n",
  730. "epoch [ 208] L = 39.280691, acc = 0.855000\n",
  731. "epoch [ 209] L = 39.272860, acc = 0.855000\n",
  732. "epoch [ 210] L = 39.265155, acc = 0.855000\n",
  733. "epoch [ 211] L = 39.257573, acc = 0.855000\n",
  734. "epoch [ 212] L = 39.250112, acc = 0.855000\n",
  735. "epoch [ 213] L = 39.242770, acc = 0.855000\n",
  736. "epoch [ 214] L = 39.235544, acc = 0.855000\n",
  737. "epoch [ 215] L = 39.228431, acc = 0.855000\n",
  738. "epoch [ 216] L = 39.221431, acc = 0.855000\n",
  739. "epoch [ 217] L = 39.214540, acc = 0.855000\n",
  740. "epoch [ 218] L = 39.207757, acc = 0.855000\n",
  741. "epoch [ 219] L = 39.201079, acc = 0.855000\n",
  742. "epoch [ 220] L = 39.194505, acc = 0.855000\n",
  743. "epoch [ 221] L = 39.188032, acc = 0.855000\n",
  744. "epoch [ 222] L = 39.181658, acc = 0.855000\n",
  745. "epoch [ 223] L = 39.175382, acc = 0.855000\n",
  746. "epoch [ 224] L = 39.169201, acc = 0.855000\n",
  747. "epoch [ 225] L = 39.163115, acc = 0.855000\n",
  748. "epoch [ 226] L = 39.157121, acc = 0.855000\n",
  749. "epoch [ 227] L = 39.151217, acc = 0.855000\n",
  750. "epoch [ 228] L = 39.145402, acc = 0.855000\n",
  751. "epoch [ 229] L = 39.139673, acc = 0.855000\n",
  752. "epoch [ 230] L = 39.134031, acc = 0.855000\n",
  753. "epoch [ 231] L = 39.128472, acc = 0.855000\n",
  754. "epoch [ 232] L = 39.122995, acc = 0.855000\n",
  755. "epoch [ 233] L = 39.117600, acc = 0.855000\n",
  756. "epoch [ 234] L = 39.112283, acc = 0.855000\n",
  757. "epoch [ 235] L = 39.107045, acc = 0.855000\n",
  758. "epoch [ 236] L = 39.101883, acc = 0.855000\n",
  759. "epoch [ 237] L = 39.096796, acc = 0.855000\n",
  760. "epoch [ 238] L = 39.091783, acc = 0.855000\n",
  761. "epoch [ 239] L = 39.086842, acc = 0.855000\n",
  762. "epoch [ 240] L = 39.081972, acc = 0.855000\n",
  763. "epoch [ 241] L = 39.077172, acc = 0.855000\n",
  764. "epoch [ 242] L = 39.072441, acc = 0.855000\n",
  765. "epoch [ 243] L = 39.067776, acc = 0.855000\n",
  766. "epoch [ 244] L = 39.063178, acc = 0.855000\n",
  767. "epoch [ 245] L = 39.058645, acc = 0.855000\n",
  768. "epoch [ 246] L = 39.054176, acc = 0.855000\n",
  769. "epoch [ 247] L = 39.049770, acc = 0.855000\n",
  770. "epoch [ 248] L = 39.045425, acc = 0.855000\n",
  771. "epoch [ 249] L = 39.041140, acc = 0.855000\n",
  772. "epoch [ 250] L = 39.036915, acc = 0.855000\n",
  773. "epoch [ 251] L = 39.032748, acc = 0.855000\n",
  774. "epoch [ 252] L = 39.028639, acc = 0.855000\n",
  775. "epoch [ 253] L = 39.024586, acc = 0.855000\n",
  776. "epoch [ 254] L = 39.020589, acc = 0.850000\n",
  777. "epoch [ 255] L = 39.016646, acc = 0.850000\n",
  778. "epoch [ 256] L = 39.012756, acc = 0.850000\n",
  779. "epoch [ 257] L = 39.008920, acc = 0.850000\n",
  780. "epoch [ 258] L = 39.005134, acc = 0.850000\n",
  781. "epoch [ 259] L = 39.001400, acc = 0.850000\n",
  782. "epoch [ 260] L = 38.997716, acc = 0.850000\n",
  783. "epoch [ 261] L = 38.994081, acc = 0.850000\n",
  784. "epoch [ 262] L = 38.990494, acc = 0.850000\n",
  785. "epoch [ 263] L = 38.986955, acc = 0.850000\n",
  786. "epoch [ 264] L = 38.983462, acc = 0.845000\n",
  787. "epoch [ 265] L = 38.980015, acc = 0.845000\n",
  788. "epoch [ 266] L = 38.976614, acc = 0.845000\n",
  789. "epoch [ 267] L = 38.973256, acc = 0.845000\n",
  790. "epoch [ 268] L = 38.969943, acc = 0.845000\n",
  791. "epoch [ 269] L = 38.966672, acc = 0.845000\n",
  792. "epoch [ 270] L = 38.963444, acc = 0.845000\n",
  793. "epoch [ 271] L = 38.960257, acc = 0.845000\n",
  794. "epoch [ 272] L = 38.957111, acc = 0.845000\n",
  795. "epoch [ 273] L = 38.954005, acc = 0.845000\n",
  796. "epoch [ 274] L = 38.950939, acc = 0.845000\n",
  797. "epoch [ 275] L = 38.947911, acc = 0.845000\n",
  798. "epoch [ 276] L = 38.944922, acc = 0.845000\n",
  799. "epoch [ 277] L = 38.941970, acc = 0.845000\n",
  800. "epoch [ 278] L = 38.939056, acc = 0.845000\n",
  801. "epoch [ 279] L = 38.936178, acc = 0.845000\n",
  802. "epoch [ 280] L = 38.933335, acc = 0.845000\n",
  803. "epoch [ 281] L = 38.930528, acc = 0.845000\n",
  804. "epoch [ 282] L = 38.927756, acc = 0.845000\n",
  805. "epoch [ 283] L = 38.925018, acc = 0.845000\n",
  806. "epoch [ 284] L = 38.922313, acc = 0.845000\n",
  807. "epoch [ 285] L = 38.919641, acc = 0.845000\n",
  808. "epoch [ 286] L = 38.917002, acc = 0.845000\n",
  809. "epoch [ 287] L = 38.914395, acc = 0.845000\n",
  810. "epoch [ 288] L = 38.911820, acc = 0.845000\n",
  811. "epoch [ 289] L = 38.909275, acc = 0.845000\n",
  812. "epoch [ 290] L = 38.906761, acc = 0.845000\n",
  813. "epoch [ 291] L = 38.904278, acc = 0.845000\n",
  814. "epoch [ 292] L = 38.901823, acc = 0.845000\n",
  815. "epoch [ 293] L = 38.899398, acc = 0.845000\n",
  816. "epoch [ 294] L = 38.897002, acc = 0.845000\n",
  817. "epoch [ 295] L = 38.894633, acc = 0.845000\n",
  818. "epoch [ 296] L = 38.892293, acc = 0.845000\n",
  819. "epoch [ 297] L = 38.889980, acc = 0.845000\n",
  820. "epoch [ 298] L = 38.887694, acc = 0.845000\n",
  821. "epoch [ 299] L = 38.885434, acc = 0.845000\n",
  822. "epoch [ 300] L = 38.883201, acc = 0.845000\n",
  823. "epoch [ 301] L = 38.880993, acc = 0.845000\n",
  824. "epoch [ 302] L = 38.878811, acc = 0.845000\n",
  825. "epoch [ 303] L = 38.876653, acc = 0.845000\n",
  826. "epoch [ 304] L = 38.874521, acc = 0.845000\n",
  827. "epoch [ 305] L = 38.872412, acc = 0.845000\n",
  828. "epoch [ 306] L = 38.870327, acc = 0.845000\n",
  829. "epoch [ 307] L = 38.868266, acc = 0.845000\n",
  830. "epoch [ 308] L = 38.866228, acc = 0.845000\n",
  831. "epoch [ 309] L = 38.864212, acc = 0.845000\n",
  832. "epoch [ 310] L = 38.862219, acc = 0.845000\n",
  833. "epoch [ 311] L = 38.860249, acc = 0.845000\n",
  834. "epoch [ 312] L = 38.858300, acc = 0.845000\n",
  835. "epoch [ 313] L = 38.856372, acc = 0.845000\n",
  836. "epoch [ 314] L = 38.854466, acc = 0.845000\n",
  837. "epoch [ 315] L = 38.852580, acc = 0.845000\n",
  838. "epoch [ 316] L = 38.850715, acc = 0.845000\n",
  839. "epoch [ 317] L = 38.848870, acc = 0.845000\n",
  840. "epoch [ 318] L = 38.847045, acc = 0.845000\n",
  841. "epoch [ 319] L = 38.845240, acc = 0.845000\n",
  842. "epoch [ 320] L = 38.843454, acc = 0.845000\n",
  843. "epoch [ 321] L = 38.841687, acc = 0.845000\n",
  844. "epoch [ 322] L = 38.839939, acc = 0.845000\n",
  845. "epoch [ 323] L = 38.838209, acc = 0.845000\n",
  846. "epoch [ 324] L = 38.836498, acc = 0.845000\n",
  847. "epoch [ 325] L = 38.834804, acc = 0.845000\n",
  848. "epoch [ 326] L = 38.833128, acc = 0.845000\n",
  849. "epoch [ 327] L = 38.831470, acc = 0.845000\n",
  850. "epoch [ 328] L = 38.829829, acc = 0.845000\n",
  851. "epoch [ 329] L = 38.828205, acc = 0.845000\n",
  852. "epoch [ 330] L = 38.826598, acc = 0.845000\n",
  853. "epoch [ 331] L = 38.825007, acc = 0.845000\n",
  854. "epoch [ 332] L = 38.823432, acc = 0.845000\n",
  855. "epoch [ 333] L = 38.821874, acc = 0.845000\n",
  856. "epoch [ 334] L = 38.820331, acc = 0.845000\n",
  857. "epoch [ 335] L = 38.818804, acc = 0.845000\n",
  858. "epoch [ 336] L = 38.817292, acc = 0.845000\n",
  859. "epoch [ 337] L = 38.815795, acc = 0.845000\n",
  860. "epoch [ 338] L = 38.814314, acc = 0.845000\n",
  861. "epoch [ 339] L = 38.812847, acc = 0.845000\n",
  862. "epoch [ 340] L = 38.811394, acc = 0.845000\n",
  863. "epoch [ 341] L = 38.809956, acc = 0.845000\n",
  864. "epoch [ 342] L = 38.808532, acc = 0.845000\n",
  865. "epoch [ 343] L = 38.807122, acc = 0.845000\n",
  866. "epoch [ 344] L = 38.805725, acc = 0.845000\n",
  867. "epoch [ 345] L = 38.804342, acc = 0.845000\n",
  868. "epoch [ 346] L = 38.802972, acc = 0.845000\n",
  869. "epoch [ 347] L = 38.801616, acc = 0.845000\n",
  870. "epoch [ 348] L = 38.800273, acc = 0.845000\n",
  871. "epoch [ 349] L = 38.798942, acc = 0.845000\n",
  872. "epoch [ 350] L = 38.797624, acc = 0.845000\n",
  873. "epoch [ 351] L = 38.796318, acc = 0.845000\n",
  874. "epoch [ 352] L = 38.795025, acc = 0.845000\n",
  875. "epoch [ 353] L = 38.793744, acc = 0.845000\n",
  876. "epoch [ 354] L = 38.792475, acc = 0.845000\n",
  877. "epoch [ 355] L = 38.791217, acc = 0.845000\n",
  878. "epoch [ 356] L = 38.789971, acc = 0.845000\n",
  879. "epoch [ 357] L = 38.788737, acc = 0.845000\n",
  880. "epoch [ 358] L = 38.787514, acc = 0.845000\n",
  881. "epoch [ 359] L = 38.786302, acc = 0.845000\n",
  882. "epoch [ 360] L = 38.785101, acc = 0.845000\n",
  883. "epoch [ 361] L = 38.783911, acc = 0.845000\n",
  884. "epoch [ 362] L = 38.782732, acc = 0.845000\n",
  885. "epoch [ 363] L = 38.781564, acc = 0.845000\n",
  886. "epoch [ 364] L = 38.780405, acc = 0.845000\n",
  887. "epoch [ 365] L = 38.779258, acc = 0.845000\n",
  888. "epoch [ 366] L = 38.778120, acc = 0.845000\n",
  889. "epoch [ 367] L = 38.776992, acc = 0.845000\n",
  890. "epoch [ 368] L = 38.775874, acc = 0.845000\n",
  891. "epoch [ 369] L = 38.774766, acc = 0.845000\n",
  892. "epoch [ 370] L = 38.773668, acc = 0.845000\n",
  893. "epoch [ 371] L = 38.772579, acc = 0.845000\n",
  894. "epoch [ 372] L = 38.771500, acc = 0.845000\n",
  895. "epoch [ 373] L = 38.770430, acc = 0.845000\n",
  896. "epoch [ 374] L = 38.769369, acc = 0.845000\n",
  897. "epoch [ 375] L = 38.768317, acc = 0.845000\n",
  898. "epoch [ 376] L = 38.767273, acc = 0.845000\n",
  899. "epoch [ 377] L = 38.766239, acc = 0.845000\n",
  900. "epoch [ 378] L = 38.765214, acc = 0.845000\n",
  901. "epoch [ 379] L = 38.764197, acc = 0.845000\n",
  902. "epoch [ 380] L = 38.763188, acc = 0.845000\n",
  903. "epoch [ 381] L = 38.762188, acc = 0.845000\n",
  904. "epoch [ 382] L = 38.761196, acc = 0.845000\n",
  905. "epoch [ 383] L = 38.760212, acc = 0.845000\n",
  906. "epoch [ 384] L = 38.759236, acc = 0.845000\n",
  907. "epoch [ 385] L = 38.758269, acc = 0.845000\n",
  908. "epoch [ 386] L = 38.757309, acc = 0.845000\n",
  909. "epoch [ 387] L = 38.756356, acc = 0.845000\n",
  910. "epoch [ 388] L = 38.755412, acc = 0.845000\n",
  911. "epoch [ 389] L = 38.754475, acc = 0.845000\n",
  912. "epoch [ 390] L = 38.753545, acc = 0.845000\n",
  913. "epoch [ 391] L = 38.752623, acc = 0.845000\n",
  914. "epoch [ 392] L = 38.751708, acc = 0.845000\n",
  915. "epoch [ 393] L = 38.750800, acc = 0.845000\n",
  916. "epoch [ 394] L = 38.749899, acc = 0.845000\n",
  917. "epoch [ 395] L = 38.749006, acc = 0.845000\n",
  918. "epoch [ 396] L = 38.748119, acc = 0.845000\n",
  919. "epoch [ 397] L = 38.747239, acc = 0.845000\n",
  920. "epoch [ 398] L = 38.746366, acc = 0.845000\n",
  921. "epoch [ 399] L = 38.745499, acc = 0.845000\n",
  922. "epoch [ 400] L = 38.744639, acc = 0.845000\n",
  923. "epoch [ 401] L = 38.743785, acc = 0.850000\n",
  924. "epoch [ 402] L = 38.742938, acc = 0.850000\n",
  925. "epoch [ 403] L = 38.742097, acc = 0.850000\n",
  926. "epoch [ 404] L = 38.741263, acc = 0.850000\n",
  927. "epoch [ 405] L = 38.740435, acc = 0.850000\n",
  928. "epoch [ 406] L = 38.739612, acc = 0.850000\n",
  929. "epoch [ 407] L = 38.738796, acc = 0.850000\n",
  930. "epoch [ 408] L = 38.737986, acc = 0.850000\n",
  931. "epoch [ 409] L = 38.737181, acc = 0.850000\n",
  932. "epoch [ 410] L = 38.736383, acc = 0.850000\n",
  933. "epoch [ 411] L = 38.735590, acc = 0.850000\n",
  934. "epoch [ 412] L = 38.734803, acc = 0.850000\n",
  935. "epoch [ 413] L = 38.734021, acc = 0.850000\n",
  936. "epoch [ 414] L = 38.733245, acc = 0.850000\n",
  937. "epoch [ 415] L = 38.732475, acc = 0.850000\n",
  938. "epoch [ 416] L = 38.731710, acc = 0.850000\n",
  939. "epoch [ 417] L = 38.730950, acc = 0.850000\n",
  940. "epoch [ 418] L = 38.730195, acc = 0.850000\n",
  941. "epoch [ 419] L = 38.729446, acc = 0.850000\n",
  942. "epoch [ 420] L = 38.728702, acc = 0.850000\n",
  943. "epoch [ 421] L = 38.727963, acc = 0.850000\n",
  944. "epoch [ 422] L = 38.727229, acc = 0.850000\n",
  945. "epoch [ 423] L = 38.726500, acc = 0.850000\n",
  946. "epoch [ 424] L = 38.725776, acc = 0.850000\n",
  947. "epoch [ 425] L = 38.725057, acc = 0.850000\n",
  948. "epoch [ 426] L = 38.724342, acc = 0.850000\n",
  949. "epoch [ 427] L = 38.723633, acc = 0.850000\n",
  950. "epoch [ 428] L = 38.722928, acc = 0.850000\n",
  951. "epoch [ 429] L = 38.722227, acc = 0.850000\n",
  952. "epoch [ 430] L = 38.721532, acc = 0.850000\n",
  953. "epoch [ 431] L = 38.720840, acc = 0.850000\n",
  954. "epoch [ 432] L = 38.720154, acc = 0.850000\n",
  955. "epoch [ 433] L = 38.719471, acc = 0.850000\n",
  956. "epoch [ 434] L = 38.718794, acc = 0.850000\n",
  957. "epoch [ 435] L = 38.718120, acc = 0.850000\n",
  958. "epoch [ 436] L = 38.717451, acc = 0.850000\n",
  959. "epoch [ 437] L = 38.716786, acc = 0.850000\n",
  960. "epoch [ 438] L = 38.716125, acc = 0.850000\n",
  961. "epoch [ 439] L = 38.715468, acc = 0.850000\n",
  962. "epoch [ 440] L = 38.714815, acc = 0.850000\n",
  963. "epoch [ 441] L = 38.714167, acc = 0.850000\n",
  964. "epoch [ 442] L = 38.713522, acc = 0.850000\n",
  965. "epoch [ 443] L = 38.712881, acc = 0.850000\n",
  966. "epoch [ 444] L = 38.712245, acc = 0.850000\n",
  967. "epoch [ 445] L = 38.711612, acc = 0.850000\n",
  968. "epoch [ 446] L = 38.710983, acc = 0.850000\n",
  969. "epoch [ 447] L = 38.710357, acc = 0.850000\n",
  970. "epoch [ 448] L = 38.709736, acc = 0.850000\n",
  971. "epoch [ 449] L = 38.709118, acc = 0.850000\n",
  972. "epoch [ 450] L = 38.708504, acc = 0.850000\n",
  973. "epoch [ 451] L = 38.707893, acc = 0.850000\n",
  974. "epoch [ 452] L = 38.707286, acc = 0.850000\n",
  975. "epoch [ 453] L = 38.706683, acc = 0.850000\n",
  976. "epoch [ 454] L = 38.706083, acc = 0.850000\n",
  977. "epoch [ 455] L = 38.705486, acc = 0.850000\n",
  978. "epoch [ 456] L = 38.704893, acc = 0.850000\n",
  979. "epoch [ 457] L = 38.704304, acc = 0.850000\n",
  980. "epoch [ 458] L = 38.703717, acc = 0.850000\n",
  981. "epoch [ 459] L = 38.703134, acc = 0.850000\n",
  982. "epoch [ 460] L = 38.702554, acc = 0.850000\n",
  983. "epoch [ 461] L = 38.701978, acc = 0.850000\n",
  984. "epoch [ 462] L = 38.701405, acc = 0.850000\n",
  985. "epoch [ 463] L = 38.700834, acc = 0.850000\n",
  986. "epoch [ 464] L = 38.700267, acc = 0.850000\n",
  987. "epoch [ 465] L = 38.699704, acc = 0.850000\n",
  988. "epoch [ 466] L = 38.699143, acc = 0.850000\n",
  989. "epoch [ 467] L = 38.698585, acc = 0.850000\n",
  990. "epoch [ 468] L = 38.698030, acc = 0.850000\n",
  991. "epoch [ 469] L = 38.697478, acc = 0.850000\n",
  992. "epoch [ 470] L = 38.696930, acc = 0.850000\n",
  993. "epoch [ 471] L = 38.696384, acc = 0.850000\n",
  994. "epoch [ 472] L = 38.695841, acc = 0.850000\n",
  995. "epoch [ 473] L = 38.695300, acc = 0.850000\n",
  996. "epoch [ 474] L = 38.694763, acc = 0.850000\n",
  997. "epoch [ 475] L = 38.694228, acc = 0.850000\n",
  998. "epoch [ 476] L = 38.693697, acc = 0.850000\n",
  999. "epoch [ 477] L = 38.693168, acc = 0.850000\n",
  1000. "epoch [ 478] L = 38.692641, acc = 0.850000\n",
  1001. "epoch [ 479] L = 38.692118, acc = 0.850000\n",
  1002. "epoch [ 480] L = 38.691597, acc = 0.850000\n",
  1003. "epoch [ 481] L = 38.691078, acc = 0.850000\n",
  1004. "epoch [ 482] L = 38.690562, acc = 0.850000\n",
  1005. "epoch [ 483] L = 38.690049, acc = 0.850000\n",
  1006. "epoch [ 484] L = 38.689538, acc = 0.850000\n",
  1007. "epoch [ 485] L = 38.689030, acc = 0.850000\n",
  1008. "epoch [ 486] L = 38.688525, acc = 0.850000\n",
  1009. "epoch [ 487] L = 38.688021, acc = 0.850000\n",
  1010. "epoch [ 488] L = 38.687521, acc = 0.850000\n",
  1011. "epoch [ 489] L = 38.687022, acc = 0.850000\n",
  1012. "epoch [ 490] L = 38.686526, acc = 0.850000\n",
  1013. "epoch [ 491] L = 38.686033, acc = 0.850000\n",
  1014. "epoch [ 492] L = 38.685542, acc = 0.850000\n",
  1015. "epoch [ 493] L = 38.685053, acc = 0.850000\n",
  1016. "epoch [ 494] L = 38.684566, acc = 0.850000\n",
  1017. "epoch [ 495] L = 38.684082, acc = 0.850000\n",
  1018. "epoch [ 496] L = 38.683600, acc = 0.850000\n",
  1019. "epoch [ 497] L = 38.683120, acc = 0.850000\n",
  1020. "epoch [ 498] L = 38.682643, acc = 0.850000\n",
  1021. "epoch [ 499] L = 38.682167, acc = 0.850000\n",
  1022. "epoch [ 500] L = 38.681694, acc = 0.850000\n",
  1023. "epoch [ 501] L = 38.681223, acc = 0.850000\n",
  1024. "epoch [ 502] L = 38.680754, acc = 0.850000\n",
  1025. "epoch [ 503] L = 38.680287, acc = 0.850000\n",
  1026. "epoch [ 504] L = 38.679823, acc = 0.850000\n",
  1027. "epoch [ 505] L = 38.679360, acc = 0.850000\n",
  1028. "epoch [ 506] L = 38.678899, acc = 0.850000\n",
  1029. "epoch [ 507] L = 38.678441, acc = 0.850000\n",
  1030. "epoch [ 508] L = 38.677984, acc = 0.850000\n",
  1031. "epoch [ 509] L = 38.677530, acc = 0.850000\n",
  1032. "epoch [ 510] L = 38.677077, acc = 0.850000\n",
  1033. "epoch [ 511] L = 38.676627, acc = 0.850000\n",
  1034. "epoch [ 512] L = 38.676178, acc = 0.850000\n",
  1035. "epoch [ 513] L = 38.675731, acc = 0.850000\n",
  1036. "epoch [ 514] L = 38.675286, acc = 0.850000\n",
  1037. "epoch [ 515] L = 38.674843, acc = 0.850000\n",
  1038. "epoch [ 516] L = 38.674402, acc = 0.850000\n",
  1039. "epoch [ 517] L = 38.673963, acc = 0.850000\n",
  1040. "epoch [ 518] L = 38.673526, acc = 0.850000\n"
  1041. ]
  1042. },
  1043. {
  1044. "name": "stdout",
  1045. "output_type": "stream",
  1046. "text": [
  1047. "epoch [ 519] L = 38.673090, acc = 0.850000\n",
  1048. "epoch [ 520] L = 38.672656, acc = 0.850000\n",
  1049. "epoch [ 521] L = 38.672224, acc = 0.850000\n",
  1050. "epoch [ 522] L = 38.671794, acc = 0.850000\n",
  1051. "epoch [ 523] L = 38.671365, acc = 0.850000\n",
  1052. "epoch [ 524] L = 38.670939, acc = 0.850000\n",
  1053. "epoch [ 525] L = 38.670514, acc = 0.850000\n",
  1054. "epoch [ 526] L = 38.670090, acc = 0.850000\n",
  1055. "epoch [ 527] L = 38.669669, acc = 0.850000\n",
  1056. "epoch [ 528] L = 38.669249, acc = 0.850000\n",
  1057. "epoch [ 529] L = 38.668830, acc = 0.850000\n",
  1058. "epoch [ 530] L = 38.668414, acc = 0.850000\n",
  1059. "epoch [ 531] L = 38.667999, acc = 0.850000\n",
  1060. "epoch [ 532] L = 38.667585, acc = 0.850000\n",
  1061. "epoch [ 533] L = 38.667173, acc = 0.850000\n",
  1062. "epoch [ 534] L = 38.666763, acc = 0.850000\n",
  1063. "epoch [ 535] L = 38.666354, acc = 0.850000\n",
  1064. "epoch [ 536] L = 38.665947, acc = 0.850000\n",
  1065. "epoch [ 537] L = 38.665542, acc = 0.850000\n",
  1066. "epoch [ 538] L = 38.665138, acc = 0.850000\n",
  1067. "epoch [ 539] L = 38.664735, acc = 0.850000\n",
  1068. "epoch [ 540] L = 38.664334, acc = 0.850000\n",
  1069. "epoch [ 541] L = 38.663935, acc = 0.850000\n",
  1070. "epoch [ 542] L = 38.663537, acc = 0.850000\n",
  1071. "epoch [ 543] L = 38.663140, acc = 0.850000\n",
  1072. "epoch [ 544] L = 38.662745, acc = 0.850000\n",
  1073. "epoch [ 545] L = 38.662351, acc = 0.850000\n",
  1074. "epoch [ 546] L = 38.661959, acc = 0.850000\n",
  1075. "epoch [ 547] L = 38.661568, acc = 0.850000\n",
  1076. "epoch [ 548] L = 38.661179, acc = 0.850000\n",
  1077. "epoch [ 549] L = 38.660791, acc = 0.850000\n",
  1078. "epoch [ 550] L = 38.660404, acc = 0.850000\n",
  1079. "epoch [ 551] L = 38.660019, acc = 0.850000\n",
  1080. "epoch [ 552] L = 38.659635, acc = 0.850000\n",
  1081. "epoch [ 553] L = 38.659253, acc = 0.850000\n",
  1082. "epoch [ 554] L = 38.658872, acc = 0.850000\n",
  1083. "epoch [ 555] L = 38.658492, acc = 0.850000\n",
  1084. "epoch [ 556] L = 38.658113, acc = 0.850000\n",
  1085. "epoch [ 557] L = 38.657736, acc = 0.850000\n",
  1086. "epoch [ 558] L = 38.657360, acc = 0.850000\n",
  1087. "epoch [ 559] L = 38.656986, acc = 0.850000\n",
  1088. "epoch [ 560] L = 38.656612, acc = 0.850000\n",
  1089. "epoch [ 561] L = 38.656240, acc = 0.850000\n",
  1090. "epoch [ 562] L = 38.655869, acc = 0.850000\n",
  1091. "epoch [ 563] L = 38.655500, acc = 0.850000\n",
  1092. "epoch [ 564] L = 38.655131, acc = 0.850000\n",
  1093. "epoch [ 565] L = 38.654764, acc = 0.850000\n",
  1094. "epoch [ 566] L = 38.654398, acc = 0.850000\n",
  1095. "epoch [ 567] L = 38.654034, acc = 0.850000\n",
  1096. "epoch [ 568] L = 38.653670, acc = 0.850000\n",
  1097. "epoch [ 569] L = 38.653308, acc = 0.850000\n",
  1098. "epoch [ 570] L = 38.652947, acc = 0.850000\n",
  1099. "epoch [ 571] L = 38.652587, acc = 0.850000\n",
  1100. "epoch [ 572] L = 38.652228, acc = 0.850000\n",
  1101. "epoch [ 573] L = 38.651870, acc = 0.850000\n",
  1102. "epoch [ 574] L = 38.651514, acc = 0.850000\n",
  1103. "epoch [ 575] L = 38.651159, acc = 0.850000\n",
  1104. "epoch [ 576] L = 38.650804, acc = 0.850000\n",
  1105. "epoch [ 577] L = 38.650451, acc = 0.850000\n",
  1106. "epoch [ 578] L = 38.650099, acc = 0.850000\n",
  1107. "epoch [ 579] L = 38.649748, acc = 0.850000\n",
  1108. "epoch [ 580] L = 38.649399, acc = 0.850000\n",
  1109. "epoch [ 581] L = 38.649050, acc = 0.850000\n",
  1110. "epoch [ 582] L = 38.648702, acc = 0.850000\n",
  1111. "epoch [ 583] L = 38.648356, acc = 0.850000\n",
  1112. "epoch [ 584] L = 38.648010, acc = 0.850000\n",
  1113. "epoch [ 585] L = 38.647666, acc = 0.850000\n",
  1114. "epoch [ 586] L = 38.647322, acc = 0.850000\n",
  1115. "epoch [ 587] L = 38.646980, acc = 0.850000\n",
  1116. "epoch [ 588] L = 38.646639, acc = 0.850000\n",
  1117. "epoch [ 589] L = 38.646299, acc = 0.850000\n",
  1118. "epoch [ 590] L = 38.645959, acc = 0.850000\n",
  1119. "epoch [ 591] L = 38.645621, acc = 0.850000\n",
  1120. "epoch [ 592] L = 38.645284, acc = 0.850000\n",
  1121. "epoch [ 593] L = 38.644948, acc = 0.850000\n",
  1122. "epoch [ 594] L = 38.644612, acc = 0.850000\n",
  1123. "epoch [ 595] L = 38.644278, acc = 0.850000\n",
  1124. "epoch [ 596] L = 38.643945, acc = 0.850000\n",
  1125. "epoch [ 597] L = 38.643612, acc = 0.850000\n",
  1126. "epoch [ 598] L = 38.643281, acc = 0.850000\n",
  1127. "epoch [ 599] L = 38.642951, acc = 0.850000\n",
  1128. "epoch [ 600] L = 38.642621, acc = 0.850000\n",
  1129. "epoch [ 601] L = 38.642293, acc = 0.850000\n",
  1130. "epoch [ 602] L = 38.641965, acc = 0.850000\n",
  1131. "epoch [ 603] L = 38.641638, acc = 0.850000\n",
  1132. "epoch [ 604] L = 38.641313, acc = 0.850000\n",
  1133. "epoch [ 605] L = 38.640988, acc = 0.850000\n",
  1134. "epoch [ 606] L = 38.640664, acc = 0.850000\n",
  1135. "epoch [ 607] L = 38.640341, acc = 0.850000\n",
  1136. "epoch [ 608] L = 38.640019, acc = 0.850000\n",
  1137. "epoch [ 609] L = 38.639698, acc = 0.850000\n",
  1138. "epoch [ 610] L = 38.639377, acc = 0.850000\n",
  1139. "epoch [ 611] L = 38.639058, acc = 0.850000\n",
  1140. "epoch [ 612] L = 38.638739, acc = 0.850000\n",
  1141. "epoch [ 613] L = 38.638422, acc = 0.850000\n",
  1142. "epoch [ 614] L = 38.638105, acc = 0.850000\n",
  1143. "epoch [ 615] L = 38.637789, acc = 0.850000\n",
  1144. "epoch [ 616] L = 38.637474, acc = 0.850000\n",
  1145. "epoch [ 617] L = 38.637160, acc = 0.850000\n",
  1146. "epoch [ 618] L = 38.636846, acc = 0.850000\n",
  1147. "epoch [ 619] L = 38.636534, acc = 0.850000\n",
  1148. "epoch [ 620] L = 38.636222, acc = 0.850000\n",
  1149. "epoch [ 621] L = 38.635911, acc = 0.850000\n",
  1150. "epoch [ 622] L = 38.635601, acc = 0.850000\n",
  1151. "epoch [ 623] L = 38.635292, acc = 0.850000\n",
  1152. "epoch [ 624] L = 38.634983, acc = 0.850000\n",
  1153. "epoch [ 625] L = 38.634676, acc = 0.850000\n",
  1154. "epoch [ 626] L = 38.634369, acc = 0.850000\n",
  1155. "epoch [ 627] L = 38.634063, acc = 0.850000\n",
  1156. "epoch [ 628] L = 38.633757, acc = 0.850000\n",
  1157. "epoch [ 629] L = 38.633453, acc = 0.850000\n",
  1158. "epoch [ 630] L = 38.633149, acc = 0.850000\n",
  1159. "epoch [ 631] L = 38.632846, acc = 0.850000\n",
  1160. "epoch [ 632] L = 38.632544, acc = 0.850000\n",
  1161. "epoch [ 633] L = 38.632243, acc = 0.850000\n",
  1162. "epoch [ 634] L = 38.631942, acc = 0.850000\n",
  1163. "epoch [ 635] L = 38.631642, acc = 0.850000\n",
  1164. "epoch [ 636] L = 38.631343, acc = 0.850000\n",
  1165. "epoch [ 637] L = 38.631045, acc = 0.850000\n",
  1166. "epoch [ 638] L = 38.630747, acc = 0.850000\n",
  1167. "epoch [ 639] L = 38.630451, acc = 0.850000\n",
  1168. "epoch [ 640] L = 38.630154, acc = 0.850000\n",
  1169. "epoch [ 641] L = 38.629859, acc = 0.850000\n",
  1170. "epoch [ 642] L = 38.629564, acc = 0.850000\n",
  1171. "epoch [ 643] L = 38.629271, acc = 0.850000\n",
  1172. "epoch [ 644] L = 38.628977, acc = 0.850000\n",
  1173. "epoch [ 645] L = 38.628685, acc = 0.850000\n",
  1174. "epoch [ 646] L = 38.628393, acc = 0.850000\n",
  1175. "epoch [ 647] L = 38.628102, acc = 0.850000\n",
  1176. "epoch [ 648] L = 38.627812, acc = 0.850000\n",
  1177. "epoch [ 649] L = 38.627522, acc = 0.850000\n",
  1178. "epoch [ 650] L = 38.627233, acc = 0.850000\n",
  1179. "epoch [ 651] L = 38.626945, acc = 0.850000\n",
  1180. "epoch [ 652] L = 38.626657, acc = 0.850000\n",
  1181. "epoch [ 653] L = 38.626371, acc = 0.850000\n",
  1182. "epoch [ 654] L = 38.626084, acc = 0.850000\n",
  1183. "epoch [ 655] L = 38.625799, acc = 0.850000\n",
  1184. "epoch [ 656] L = 38.625514, acc = 0.850000\n",
  1185. "epoch [ 657] L = 38.625230, acc = 0.850000\n",
  1186. "epoch [ 658] L = 38.624946, acc = 0.850000\n",
  1187. "epoch [ 659] L = 38.624664, acc = 0.850000\n",
  1188. "epoch [ 660] L = 38.624381, acc = 0.850000\n",
  1189. "epoch [ 661] L = 38.624100, acc = 0.850000\n",
  1190. "epoch [ 662] L = 38.623819, acc = 0.850000\n",
  1191. "epoch [ 663] L = 38.623539, acc = 0.850000\n",
  1192. "epoch [ 664] L = 38.623259, acc = 0.850000\n",
  1193. "epoch [ 665] L = 38.622980, acc = 0.850000\n",
  1194. "epoch [ 666] L = 38.622702, acc = 0.850000\n",
  1195. "epoch [ 667] L = 38.622424, acc = 0.850000\n",
  1196. "epoch [ 668] L = 38.622147, acc = 0.850000\n",
  1197. "epoch [ 669] L = 38.621871, acc = 0.850000\n",
  1198. "epoch [ 670] L = 38.621595, acc = 0.850000\n",
  1199. "epoch [ 671] L = 38.621320, acc = 0.850000\n",
  1200. "epoch [ 672] L = 38.621046, acc = 0.850000\n",
  1201. "epoch [ 673] L = 38.620772, acc = 0.850000\n",
  1202. "epoch [ 674] L = 38.620499, acc = 0.850000\n",
  1203. "epoch [ 675] L = 38.620226, acc = 0.850000\n",
  1204. "epoch [ 676] L = 38.619954, acc = 0.850000\n",
  1205. "epoch [ 677] L = 38.619682, acc = 0.850000\n",
  1206. "epoch [ 678] L = 38.619412, acc = 0.850000\n",
  1207. "epoch [ 679] L = 38.619141, acc = 0.850000\n",
  1208. "epoch [ 680] L = 38.618872, acc = 0.850000\n",
  1209. "epoch [ 681] L = 38.618603, acc = 0.850000\n",
  1210. "epoch [ 682] L = 38.618334, acc = 0.850000\n",
  1211. "epoch [ 683] L = 38.618066, acc = 0.850000\n",
  1212. "epoch [ 684] L = 38.617799, acc = 0.850000\n",
  1213. "epoch [ 685] L = 38.617532, acc = 0.850000\n",
  1214. "epoch [ 686] L = 38.617266, acc = 0.850000\n",
  1215. "epoch [ 687] L = 38.617001, acc = 0.850000\n",
  1216. "epoch [ 688] L = 38.616736, acc = 0.850000\n",
  1217. "epoch [ 689] L = 38.616471, acc = 0.850000\n",
  1218. "epoch [ 690] L = 38.616208, acc = 0.850000\n",
  1219. "epoch [ 691] L = 38.615944, acc = 0.850000\n",
  1220. "epoch [ 692] L = 38.615682, acc = 0.850000\n",
  1221. "epoch [ 693] L = 38.615420, acc = 0.850000\n",
  1222. "epoch [ 694] L = 38.615158, acc = 0.850000\n",
  1223. "epoch [ 695] L = 38.614897, acc = 0.850000\n",
  1224. "epoch [ 696] L = 38.614637, acc = 0.850000\n",
  1225. "epoch [ 697] L = 38.614377, acc = 0.850000\n",
  1226. "epoch [ 698] L = 38.614117, acc = 0.850000\n",
  1227. "epoch [ 699] L = 38.613859, acc = 0.850000\n",
  1228. "epoch [ 700] L = 38.613600, acc = 0.850000\n",
  1229. "epoch [ 701] L = 38.613343, acc = 0.850000\n",
  1230. "epoch [ 702] L = 38.613085, acc = 0.850000\n",
  1231. "epoch [ 703] L = 38.612829, acc = 0.850000\n",
  1232. "epoch [ 704] L = 38.612573, acc = 0.850000\n",
  1233. "epoch [ 705] L = 38.612317, acc = 0.850000\n",
  1234. "epoch [ 706] L = 38.612062, acc = 0.850000\n",
  1235. "epoch [ 707] L = 38.611808, acc = 0.850000\n",
  1236. "epoch [ 708] L = 38.611554, acc = 0.850000\n",
  1237. "epoch [ 709] L = 38.611300, acc = 0.850000\n",
  1238. "epoch [ 710] L = 38.611047, acc = 0.850000\n",
  1239. "epoch [ 711] L = 38.610795, acc = 0.850000\n",
  1240. "epoch [ 712] L = 38.610543, acc = 0.850000\n",
  1241. "epoch [ 713] L = 38.610291, acc = 0.850000\n",
  1242. "epoch [ 714] L = 38.610041, acc = 0.850000\n",
  1243. "epoch [ 715] L = 38.609790, acc = 0.850000\n",
  1244. "epoch [ 716] L = 38.609540, acc = 0.850000\n",
  1245. "epoch [ 717] L = 38.609291, acc = 0.850000\n",
  1246. "epoch [ 718] L = 38.609042, acc = 0.850000\n",
  1247. "epoch [ 719] L = 38.608794, acc = 0.850000\n",
  1248. "epoch [ 720] L = 38.608546, acc = 0.850000\n",
  1249. "epoch [ 721] L = 38.608299, acc = 0.850000\n",
  1250. "epoch [ 722] L = 38.608052, acc = 0.850000\n",
  1251. "epoch [ 723] L = 38.607805, acc = 0.850000\n",
  1252. "epoch [ 724] L = 38.607559, acc = 0.850000\n",
  1253. "epoch [ 725] L = 38.607314, acc = 0.850000\n",
  1254. "epoch [ 726] L = 38.607069, acc = 0.850000\n",
  1255. "epoch [ 727] L = 38.606825, acc = 0.850000\n",
  1256. "epoch [ 728] L = 38.606581, acc = 0.850000\n",
  1257. "epoch [ 729] L = 38.606337, acc = 0.850000\n",
  1258. "epoch [ 730] L = 38.606094, acc = 0.850000\n",
  1259. "epoch [ 731] L = 38.605852, acc = 0.850000\n",
  1260. "epoch [ 732] L = 38.605610, acc = 0.850000\n",
  1261. "epoch [ 733] L = 38.605368, acc = 0.850000\n",
  1262. "epoch [ 734] L = 38.605127, acc = 0.850000\n",
  1263. "epoch [ 735] L = 38.604887, acc = 0.850000\n",
  1264. "epoch [ 736] L = 38.604647, acc = 0.850000\n",
  1265. "epoch [ 737] L = 38.604407, acc = 0.850000\n",
  1266. "epoch [ 738] L = 38.604168, acc = 0.850000\n",
  1267. "epoch [ 739] L = 38.603929, acc = 0.850000\n",
  1268. "epoch [ 740] L = 38.603691, acc = 0.850000\n",
  1269. "epoch [ 741] L = 38.603453, acc = 0.850000\n",
  1270. "epoch [ 742] L = 38.603216, acc = 0.850000\n",
  1271. "epoch [ 743] L = 38.602979, acc = 0.850000\n",
  1272. "epoch [ 744] L = 38.602742, acc = 0.850000\n",
  1273. "epoch [ 745] L = 38.602506, acc = 0.850000\n",
  1274. "epoch [ 746] L = 38.602271, acc = 0.850000\n",
  1275. "epoch [ 747] L = 38.602036, acc = 0.850000\n",
  1276. "epoch [ 748] L = 38.601801, acc = 0.850000\n",
  1277. "epoch [ 749] L = 38.601567, acc = 0.850000\n",
  1278. "epoch [ 750] L = 38.601333, acc = 0.850000\n",
  1279. "epoch [ 751] L = 38.601100, acc = 0.850000\n",
  1280. "epoch [ 752] L = 38.600867, acc = 0.850000\n",
  1281. "epoch [ 753] L = 38.600635, acc = 0.850000\n",
  1282. "epoch [ 754] L = 38.600403, acc = 0.850000\n",
  1283. "epoch [ 755] L = 38.600171, acc = 0.850000\n",
  1284. "epoch [ 756] L = 38.599940, acc = 0.850000\n",
  1285. "epoch [ 757] L = 38.599709, acc = 0.850000\n",
  1286. "epoch [ 758] L = 38.599479, acc = 0.850000\n",
  1287. "epoch [ 759] L = 38.599249, acc = 0.850000\n",
  1288. "epoch [ 760] L = 38.599020, acc = 0.850000\n",
  1289. "epoch [ 761] L = 38.598791, acc = 0.850000\n",
  1290. "epoch [ 762] L = 38.598562, acc = 0.850000\n",
  1291. "epoch [ 763] L = 38.598334, acc = 0.850000\n",
  1292. "epoch [ 764] L = 38.598107, acc = 0.850000\n",
  1293. "epoch [ 765] L = 38.597879, acc = 0.850000\n",
  1294. "epoch [ 766] L = 38.597653, acc = 0.850000\n",
  1295. "epoch [ 767] L = 38.597426, acc = 0.850000\n",
  1296. "epoch [ 768] L = 38.597200, acc = 0.850000\n",
  1297. "epoch [ 769] L = 38.596975, acc = 0.850000\n",
  1298. "epoch [ 770] L = 38.596749, acc = 0.850000\n",
  1299. "epoch [ 771] L = 38.596525, acc = 0.850000\n",
  1300. "epoch [ 772] L = 38.596300, acc = 0.850000\n",
  1301. "epoch [ 773] L = 38.596076, acc = 0.850000\n",
  1302. "epoch [ 774] L = 38.595853, acc = 0.850000\n",
  1303. "epoch [ 775] L = 38.595630, acc = 0.850000\n",
  1304. "epoch [ 776] L = 38.595407, acc = 0.850000\n",
  1305. "epoch [ 777] L = 38.595185, acc = 0.850000\n",
  1306. "epoch [ 778] L = 38.594963, acc = 0.850000\n",
  1307. "epoch [ 779] L = 38.594741, acc = 0.850000\n",
  1308. "epoch [ 780] L = 38.594520, acc = 0.850000\n",
  1309. "epoch [ 781] L = 38.594299, acc = 0.850000\n",
  1310. "epoch [ 782] L = 38.594079, acc = 0.850000\n",
  1311. "epoch [ 783] L = 38.593859, acc = 0.850000\n",
  1312. "epoch [ 784] L = 38.593640, acc = 0.850000\n",
  1313. "epoch [ 785] L = 38.593421, acc = 0.850000\n",
  1314. "epoch [ 786] L = 38.593202, acc = 0.850000\n",
  1315. "epoch [ 787] L = 38.592984, acc = 0.850000\n",
  1316. "epoch [ 788] L = 38.592766, acc = 0.850000\n",
  1317. "epoch [ 789] L = 38.592548, acc = 0.850000\n",
  1318. "epoch [ 790] L = 38.592331, acc = 0.850000\n",
  1319. "epoch [ 791] L = 38.592114, acc = 0.850000\n",
  1320. "epoch [ 792] L = 38.591898, acc = 0.850000\n",
  1321. "epoch [ 793] L = 38.591682, acc = 0.850000\n",
  1322. "epoch [ 794] L = 38.591466, acc = 0.850000\n",
  1323. "epoch [ 795] L = 38.591251, acc = 0.850000\n",
  1324. "epoch [ 796] L = 38.591036, acc = 0.850000\n",
  1325. "epoch [ 797] L = 38.590821, acc = 0.850000\n",
  1326. "epoch [ 798] L = 38.590607, acc = 0.850000\n",
  1327. "epoch [ 799] L = 38.590394, acc = 0.850000\n",
  1328. "epoch [ 800] L = 38.590180, acc = 0.850000\n",
  1329. "epoch [ 801] L = 38.589967, acc = 0.850000\n",
  1330. "epoch [ 802] L = 38.589755, acc = 0.850000\n",
  1331. "epoch [ 803] L = 38.589542, acc = 0.850000\n",
  1332. "epoch [ 804] L = 38.589330, acc = 0.850000\n",
  1333. "epoch [ 805] L = 38.589119, acc = 0.850000\n",
  1334. "epoch [ 806] L = 38.588908, acc = 0.850000\n",
  1335. "epoch [ 807] L = 38.588697, acc = 0.850000\n",
  1336. "epoch [ 808] L = 38.588487, acc = 0.850000\n",
  1337. "epoch [ 809] L = 38.588277, acc = 0.850000\n",
  1338. "epoch [ 810] L = 38.588067, acc = 0.850000\n",
  1339. "epoch [ 811] L = 38.587858, acc = 0.850000\n",
  1340. "epoch [ 812] L = 38.587649, acc = 0.850000\n",
  1341. "epoch [ 813] L = 38.587440, acc = 0.850000\n",
  1342. "epoch [ 814] L = 38.587232, acc = 0.850000\n",
  1343. "epoch [ 815] L = 38.587024, acc = 0.850000\n",
  1344. "epoch [ 816] L = 38.586817, acc = 0.850000\n",
  1345. "epoch [ 817] L = 38.586609, acc = 0.850000\n",
  1346. "epoch [ 818] L = 38.586403, acc = 0.850000\n",
  1347. "epoch [ 819] L = 38.586196, acc = 0.850000\n",
  1348. "epoch [ 820] L = 38.585990, acc = 0.850000\n",
  1349. "epoch [ 821] L = 38.585784, acc = 0.850000\n",
  1350. "epoch [ 822] L = 38.585579, acc = 0.850000\n",
  1351. "epoch [ 823] L = 38.585374, acc = 0.850000\n",
  1352. "epoch [ 824] L = 38.585169, acc = 0.850000\n",
  1353. "epoch [ 825] L = 38.584965, acc = 0.850000\n",
  1354. "epoch [ 826] L = 38.584761, acc = 0.850000\n",
  1355. "epoch [ 827] L = 38.584557, acc = 0.850000\n",
  1356. "epoch [ 828] L = 38.584354, acc = 0.850000\n",
  1357. "epoch [ 829] L = 38.584151, acc = 0.850000\n",
  1358. "epoch [ 830] L = 38.583948, acc = 0.850000\n",
  1359. "epoch [ 831] L = 38.583746, acc = 0.850000\n",
  1360. "epoch [ 832] L = 38.583544, acc = 0.850000\n",
  1361. "epoch [ 833] L = 38.583342, acc = 0.850000\n",
  1362. "epoch [ 834] L = 38.583141, acc = 0.850000\n",
  1363. "epoch [ 835] L = 38.582940, acc = 0.850000\n",
  1364. "epoch [ 836] L = 38.582740, acc = 0.850000\n",
  1365. "epoch [ 837] L = 38.582539, acc = 0.850000\n",
  1366. "epoch [ 838] L = 38.582339, acc = 0.850000\n",
  1367. "epoch [ 839] L = 38.582140, acc = 0.850000\n",
  1368. "epoch [ 840] L = 38.581941, acc = 0.850000\n",
  1369. "epoch [ 841] L = 38.581742, acc = 0.850000\n",
  1370. "epoch [ 842] L = 38.581543, acc = 0.850000\n",
  1371. "epoch [ 843] L = 38.581345, acc = 0.850000\n",
  1372. "epoch [ 844] L = 38.581147, acc = 0.850000\n",
  1373. "epoch [ 845] L = 38.580949, acc = 0.850000\n",
  1374. "epoch [ 846] L = 38.580752, acc = 0.850000\n",
  1375. "epoch [ 847] L = 38.580555, acc = 0.850000\n",
  1376. "epoch [ 848] L = 38.580359, acc = 0.850000\n",
  1377. "epoch [ 849] L = 38.580162, acc = 0.850000\n",
  1378. "epoch [ 850] L = 38.579966, acc = 0.850000\n",
  1379. "epoch [ 851] L = 38.579771, acc = 0.850000\n",
  1380. "epoch [ 852] L = 38.579575, acc = 0.850000\n",
  1381. "epoch [ 853] L = 38.579380, acc = 0.850000\n",
  1382. "epoch [ 854] L = 38.579186, acc = 0.850000\n",
  1383. "epoch [ 855] L = 38.578991, acc = 0.850000\n",
  1384. "epoch [ 856] L = 38.578797, acc = 0.850000\n",
  1385. "epoch [ 857] L = 38.578603, acc = 0.850000\n",
  1386. "epoch [ 858] L = 38.578410, acc = 0.850000\n",
  1387. "epoch [ 859] L = 38.578217, acc = 0.850000\n",
  1388. "epoch [ 860] L = 38.578024, acc = 0.850000\n",
  1389. "epoch [ 861] L = 38.577832, acc = 0.850000\n",
  1390. "epoch [ 862] L = 38.577639, acc = 0.850000\n",
  1391. "epoch [ 863] L = 38.577448, acc = 0.850000\n",
  1392. "epoch [ 864] L = 38.577256, acc = 0.850000\n",
  1393. "epoch [ 865] L = 38.577065, acc = 0.850000\n",
  1394. "epoch [ 866] L = 38.576874, acc = 0.850000\n",
  1395. "epoch [ 867] L = 38.576683, acc = 0.850000\n",
  1396. "epoch [ 868] L = 38.576493, acc = 0.850000\n",
  1397. "epoch [ 869] L = 38.576303, acc = 0.850000\n",
  1398. "epoch [ 870] L = 38.576113, acc = 0.850000\n",
  1399. "epoch [ 871] L = 38.575924, acc = 0.850000\n",
  1400. "epoch [ 872] L = 38.575735, acc = 0.850000\n",
  1401. "epoch [ 873] L = 38.575546, acc = 0.850000\n",
  1402. "epoch [ 874] L = 38.575358, acc = 0.850000\n",
  1403. "epoch [ 875] L = 38.575169, acc = 0.850000\n",
  1404. "epoch [ 876] L = 38.574981, acc = 0.850000\n",
  1405. "epoch [ 877] L = 38.574794, acc = 0.850000\n",
  1406. "epoch [ 878] L = 38.574607, acc = 0.850000\n",
  1407. "epoch [ 879] L = 38.574420, acc = 0.850000\n",
  1408. "epoch [ 880] L = 38.574233, acc = 0.850000\n",
  1409. "epoch [ 881] L = 38.574047, acc = 0.850000\n",
  1410. "epoch [ 882] L = 38.573861, acc = 0.850000\n",
  1411. "epoch [ 883] L = 38.573675, acc = 0.850000\n",
  1412. "epoch [ 884] L = 38.573489, acc = 0.850000\n",
  1413. "epoch [ 885] L = 38.573304, acc = 0.850000\n",
  1414. "epoch [ 886] L = 38.573119, acc = 0.850000\n",
  1415. "epoch [ 887] L = 38.572934, acc = 0.850000\n",
  1416. "epoch [ 888] L = 38.572750, acc = 0.850000\n",
  1417. "epoch [ 889] L = 38.572566, acc = 0.850000\n",
  1418. "epoch [ 890] L = 38.572382, acc = 0.850000\n",
  1419. "epoch [ 891] L = 38.572199, acc = 0.850000\n",
  1420. "epoch [ 892] L = 38.572016, acc = 0.850000\n",
  1421. "epoch [ 893] L = 38.571833, acc = 0.850000\n",
  1422. "epoch [ 894] L = 38.571650, acc = 0.850000\n",
  1423. "epoch [ 895] L = 38.571468, acc = 0.850000\n",
  1424. "epoch [ 896] L = 38.571286, acc = 0.850000\n",
  1425. "epoch [ 897] L = 38.571104, acc = 0.850000\n",
  1426. "epoch [ 898] L = 38.570923, acc = 0.850000\n",
  1427. "epoch [ 899] L = 38.570742, acc = 0.850000\n",
  1428. "epoch [ 900] L = 38.570561, acc = 0.850000\n",
  1429. "epoch [ 901] L = 38.570380, acc = 0.850000\n",
  1430. "epoch [ 902] L = 38.570200, acc = 0.850000\n",
  1431. "epoch [ 903] L = 38.570020, acc = 0.850000\n",
  1432. "epoch [ 904] L = 38.569840, acc = 0.850000\n",
  1433. "epoch [ 905] L = 38.569660, acc = 0.850000\n",
  1434. "epoch [ 906] L = 38.569481, acc = 0.850000\n",
  1435. "epoch [ 907] L = 38.569302, acc = 0.850000\n",
  1436. "epoch [ 908] L = 38.569124, acc = 0.850000\n",
  1437. "epoch [ 909] L = 38.568945, acc = 0.850000\n",
  1438. "epoch [ 910] L = 38.568767, acc = 0.850000\n",
  1439. "epoch [ 911] L = 38.568589, acc = 0.850000\n",
  1440. "epoch [ 912] L = 38.568412, acc = 0.850000\n",
  1441. "epoch [ 913] L = 38.568234, acc = 0.850000\n",
  1442. "epoch [ 914] L = 38.568057, acc = 0.850000\n",
  1443. "epoch [ 915] L = 38.567881, acc = 0.850000\n",
  1444. "epoch [ 916] L = 38.567704, acc = 0.850000\n",
  1445. "epoch [ 917] L = 38.567528, acc = 0.850000\n",
  1446. "epoch [ 918] L = 38.567352, acc = 0.850000\n",
  1447. "epoch [ 919] L = 38.567176, acc = 0.850000\n",
  1448. "epoch [ 920] L = 38.567001, acc = 0.850000\n",
  1449. "epoch [ 921] L = 38.566826, acc = 0.850000\n",
  1450. "epoch [ 922] L = 38.566651, acc = 0.850000\n",
  1451. "epoch [ 923] L = 38.566476, acc = 0.850000\n",
  1452. "epoch [ 924] L = 38.566302, acc = 0.850000\n",
  1453. "epoch [ 925] L = 38.566128, acc = 0.850000\n",
  1454. "epoch [ 926] L = 38.565954, acc = 0.850000\n",
  1455. "epoch [ 927] L = 38.565780, acc = 0.850000\n",
  1456. "epoch [ 928] L = 38.565607, acc = 0.850000\n",
  1457. "epoch [ 929] L = 38.565434, acc = 0.850000\n",
  1458. "epoch [ 930] L = 38.565261, acc = 0.850000\n",
  1459. "epoch [ 931] L = 38.565089, acc = 0.850000\n",
  1460. "epoch [ 932] L = 38.564917, acc = 0.850000\n",
  1461. "epoch [ 933] L = 38.564745, acc = 0.850000\n",
  1462. "epoch [ 934] L = 38.564573, acc = 0.850000\n",
  1463. "epoch [ 935] L = 38.564401, acc = 0.850000\n",
  1464. "epoch [ 936] L = 38.564230, acc = 0.850000\n",
  1465. "epoch [ 937] L = 38.564059, acc = 0.850000\n",
  1466. "epoch [ 938] L = 38.563888, acc = 0.850000\n",
  1467. "epoch [ 939] L = 38.563718, acc = 0.850000\n",
  1468. "epoch [ 940] L = 38.563548, acc = 0.850000\n",
  1469. "epoch [ 941] L = 38.563378, acc = 0.850000\n",
  1470. "epoch [ 942] L = 38.563208, acc = 0.850000\n",
  1471. "epoch [ 943] L = 38.563039, acc = 0.850000\n",
  1472. "epoch [ 944] L = 38.562869, acc = 0.850000\n",
  1473. "epoch [ 945] L = 38.562701, acc = 0.850000\n",
  1474. "epoch [ 946] L = 38.562532, acc = 0.850000\n",
  1475. "epoch [ 947] L = 38.562363, acc = 0.850000\n",
  1476. "epoch [ 948] L = 38.562195, acc = 0.850000\n",
  1477. "epoch [ 949] L = 38.562027, acc = 0.850000\n",
  1478. "epoch [ 950] L = 38.561860, acc = 0.850000\n",
  1479. "epoch [ 951] L = 38.561692, acc = 0.850000\n",
  1480. "epoch [ 952] L = 38.561525, acc = 0.850000\n",
  1481. "epoch [ 953] L = 38.561358, acc = 0.850000\n",
  1482. "epoch [ 954] L = 38.561191, acc = 0.850000\n",
  1483. "epoch [ 955] L = 38.561025, acc = 0.850000\n",
  1484. "epoch [ 956] L = 38.560859, acc = 0.850000\n",
  1485. "epoch [ 957] L = 38.560693, acc = 0.850000\n",
  1486. "epoch [ 958] L = 38.560527, acc = 0.850000\n",
  1487. "epoch [ 959] L = 38.560361, acc = 0.850000\n",
  1488. "epoch [ 960] L = 38.560196, acc = 0.850000\n",
  1489. "epoch [ 961] L = 38.560031, acc = 0.850000\n",
  1490. "epoch [ 962] L = 38.559866, acc = 0.850000\n",
  1491. "epoch [ 963] L = 38.559702, acc = 0.850000\n",
  1492. "epoch [ 964] L = 38.559537, acc = 0.850000\n",
  1493. "epoch [ 965] L = 38.559373, acc = 0.850000\n",
  1494. "epoch [ 966] L = 38.559210, acc = 0.850000\n",
  1495. "epoch [ 967] L = 38.559046, acc = 0.850000\n",
  1496. "epoch [ 968] L = 38.558883, acc = 0.850000\n",
  1497. "epoch [ 969] L = 38.558719, acc = 0.850000\n",
  1498. "epoch [ 970] L = 38.558557, acc = 0.850000\n",
  1499. "epoch [ 971] L = 38.558394, acc = 0.850000\n",
  1500. "epoch [ 972] L = 38.558232, acc = 0.850000\n",
  1501. "epoch [ 973] L = 38.558069, acc = 0.850000\n",
  1502. "epoch [ 974] L = 38.557907, acc = 0.850000\n",
  1503. "epoch [ 975] L = 38.557746, acc = 0.850000\n",
  1504. "epoch [ 976] L = 38.557584, acc = 0.850000\n",
  1505. "epoch [ 977] L = 38.557423, acc = 0.850000\n",
  1506. "epoch [ 978] L = 38.557262, acc = 0.850000\n",
  1507. "epoch [ 979] L = 38.557101, acc = 0.850000\n",
  1508. "epoch [ 980] L = 38.556941, acc = 0.850000\n",
  1509. "epoch [ 981] L = 38.556780, acc = 0.850000\n",
  1510. "epoch [ 982] L = 38.556620, acc = 0.850000\n",
  1511. "epoch [ 983] L = 38.556460, acc = 0.850000\n",
  1512. "epoch [ 984] L = 38.556301, acc = 0.850000\n",
  1513. "epoch [ 985] L = 38.556141, acc = 0.850000\n",
  1514. "epoch [ 986] L = 38.555982, acc = 0.850000\n",
  1515. "epoch [ 987] L = 38.555823, acc = 0.850000\n",
  1516. "epoch [ 988] L = 38.555664, acc = 0.850000\n",
  1517. "epoch [ 989] L = 38.555506, acc = 0.850000\n",
  1518. "epoch [ 990] L = 38.555347, acc = 0.850000\n",
  1519. "epoch [ 991] L = 38.555189, acc = 0.850000\n",
  1520. "epoch [ 992] L = 38.555032, acc = 0.850000\n",
  1521. "epoch [ 993] L = 38.554874, acc = 0.850000\n",
  1522. "epoch [ 994] L = 38.554717, acc = 0.850000\n",
  1523. "epoch [ 995] L = 38.554559, acc = 0.850000\n",
  1524. "epoch [ 996] L = 38.554402, acc = 0.850000\n",
  1525. "epoch [ 997] L = 38.554246, acc = 0.850000\n",
  1526. "epoch [ 998] L = 38.554089, acc = 0.850000\n",
  1527. "epoch [ 999] L = 38.553933, acc = 0.850000\n",
  1528. "epoch [1000] L = 38.553777, acc = 0.850000\n",
  1529. "epoch [1001] L = 38.553621, acc = 0.850000\n",
  1530. "epoch [1002] L = 38.553465, acc = 0.850000\n",
  1531. "epoch [1003] L = 38.553310, acc = 0.850000\n",
  1532. "epoch [1004] L = 38.553154, acc = 0.850000\n",
  1533. "epoch [1005] L = 38.552999, acc = 0.850000\n",
  1534. "epoch [1006] L = 38.552845, acc = 0.850000\n",
  1535. "epoch [1007] L = 38.552690, acc = 0.850000\n",
  1536. "epoch [1008] L = 38.552536, acc = 0.850000\n",
  1537. "epoch [1009] L = 38.552382, acc = 0.850000\n",
  1538. "epoch [1010] L = 38.552228, acc = 0.850000\n",
  1539. "epoch [1011] L = 38.552074, acc = 0.850000\n",
  1540. "epoch [1012] L = 38.551920, acc = 0.850000\n",
  1541. "epoch [1013] L = 38.551767, acc = 0.850000\n",
  1542. "epoch [1014] L = 38.551614, acc = 0.850000\n",
  1543. "epoch [1015] L = 38.551461, acc = 0.850000\n",
  1544. "epoch [1016] L = 38.551308, acc = 0.850000\n",
  1545. "epoch [1017] L = 38.551156, acc = 0.850000\n",
  1546. "epoch [1018] L = 38.551004, acc = 0.850000\n",
  1547. "epoch [1019] L = 38.550852, acc = 0.850000\n",
  1548. "epoch [1020] L = 38.550700, acc = 0.850000\n",
  1549. "epoch [1021] L = 38.550548, acc = 0.850000\n",
  1550. "epoch [1022] L = 38.550397, acc = 0.850000\n",
  1551. "epoch [1023] L = 38.550245, acc = 0.850000\n",
  1552. "epoch [1024] L = 38.550094, acc = 0.850000\n",
  1553. "epoch [1025] L = 38.549944, acc = 0.850000\n",
  1554. "epoch [1026] L = 38.549793, acc = 0.850000\n",
  1555. "epoch [1027] L = 38.549642, acc = 0.850000\n",
  1556. "epoch [1028] L = 38.549492, acc = 0.850000\n",
  1557. "epoch [1029] L = 38.549342, acc = 0.850000\n",
  1558. "epoch [1030] L = 38.549192, acc = 0.850000\n",
  1559. "epoch [1031] L = 38.549043, acc = 0.850000\n",
  1560. "epoch [1032] L = 38.548893, acc = 0.850000\n",
  1561. "epoch [1033] L = 38.548744, acc = 0.850000\n",
  1562. "epoch [1034] L = 38.548595, acc = 0.850000\n",
  1563. "epoch [1035] L = 38.548446, acc = 0.850000\n",
  1564. "epoch [1036] L = 38.548298, acc = 0.850000\n",
  1565. "epoch [1037] L = 38.548149, acc = 0.850000\n",
  1566. "epoch [1038] L = 38.548001, acc = 0.850000\n",
  1567. "epoch [1039] L = 38.547853, acc = 0.850000\n",
  1568. "epoch [1040] L = 38.547705, acc = 0.850000\n"
  1569. ]
  1570. },
  1571. {
  1572. "name": "stdout",
  1573. "output_type": "stream",
  1574. "text": [
  1575. "epoch [1041] L = 38.547558, acc = 0.850000\n",
  1576. "epoch [1042] L = 38.547410, acc = 0.850000\n",
  1577. "epoch [1043] L = 38.547263, acc = 0.850000\n",
  1578. "epoch [1044] L = 38.547116, acc = 0.850000\n",
  1579. "epoch [1045] L = 38.546969, acc = 0.850000\n",
  1580. "epoch [1046] L = 38.546823, acc = 0.850000\n",
  1581. "epoch [1047] L = 38.546676, acc = 0.850000\n",
  1582. "epoch [1048] L = 38.546530, acc = 0.850000\n",
  1583. "epoch [1049] L = 38.546384, acc = 0.845000\n",
  1584. "epoch [1050] L = 38.546238, acc = 0.845000\n",
  1585. "epoch [1051] L = 38.546092, acc = 0.845000\n",
  1586. "epoch [1052] L = 38.545947, acc = 0.845000\n",
  1587. "epoch [1053] L = 38.545802, acc = 0.845000\n",
  1588. "epoch [1054] L = 38.545656, acc = 0.845000\n",
  1589. "epoch [1055] L = 38.545512, acc = 0.845000\n",
  1590. "epoch [1056] L = 38.545367, acc = 0.845000\n",
  1591. "epoch [1057] L = 38.545222, acc = 0.845000\n",
  1592. "epoch [1058] L = 38.545078, acc = 0.845000\n",
  1593. "epoch [1059] L = 38.544934, acc = 0.845000\n",
  1594. "epoch [1060] L = 38.544790, acc = 0.845000\n",
  1595. "epoch [1061] L = 38.544646, acc = 0.845000\n",
  1596. "epoch [1062] L = 38.544502, acc = 0.845000\n",
  1597. "epoch [1063] L = 38.544359, acc = 0.845000\n",
  1598. "epoch [1064] L = 38.544216, acc = 0.845000\n",
  1599. "epoch [1065] L = 38.544073, acc = 0.845000\n",
  1600. "epoch [1066] L = 38.543930, acc = 0.845000\n",
  1601. "epoch [1067] L = 38.543787, acc = 0.845000\n",
  1602. "epoch [1068] L = 38.543645, acc = 0.845000\n",
  1603. "epoch [1069] L = 38.543502, acc = 0.845000\n",
  1604. "epoch [1070] L = 38.543360, acc = 0.845000\n",
  1605. "epoch [1071] L = 38.543218, acc = 0.845000\n",
  1606. "epoch [1072] L = 38.543077, acc = 0.845000\n",
  1607. "epoch [1073] L = 38.542935, acc = 0.845000\n",
  1608. "epoch [1074] L = 38.542794, acc = 0.845000\n",
  1609. "epoch [1075] L = 38.542652, acc = 0.845000\n",
  1610. "epoch [1076] L = 38.542511, acc = 0.845000\n",
  1611. "epoch [1077] L = 38.542370, acc = 0.845000\n",
  1612. "epoch [1078] L = 38.542230, acc = 0.845000\n",
  1613. "epoch [1079] L = 38.542089, acc = 0.845000\n",
  1614. "epoch [1080] L = 38.541949, acc = 0.845000\n",
  1615. "epoch [1081] L = 38.541809, acc = 0.845000\n",
  1616. "epoch [1082] L = 38.541669, acc = 0.845000\n",
  1617. "epoch [1083] L = 38.541529, acc = 0.845000\n",
  1618. "epoch [1084] L = 38.541389, acc = 0.845000\n",
  1619. "epoch [1085] L = 38.541250, acc = 0.845000\n",
  1620. "epoch [1086] L = 38.541111, acc = 0.845000\n",
  1621. "epoch [1087] L = 38.540972, acc = 0.845000\n",
  1622. "epoch [1088] L = 38.540833, acc = 0.845000\n",
  1623. "epoch [1089] L = 38.540694, acc = 0.845000\n",
  1624. "epoch [1090] L = 38.540555, acc = 0.845000\n",
  1625. "epoch [1091] L = 38.540417, acc = 0.845000\n",
  1626. "epoch [1092] L = 38.540279, acc = 0.845000\n",
  1627. "epoch [1093] L = 38.540141, acc = 0.845000\n",
  1628. "epoch [1094] L = 38.540003, acc = 0.845000\n",
  1629. "epoch [1095] L = 38.539865, acc = 0.845000\n",
  1630. "epoch [1096] L = 38.539728, acc = 0.845000\n",
  1631. "epoch [1097] L = 38.539590, acc = 0.845000\n",
  1632. "epoch [1098] L = 38.539453, acc = 0.845000\n",
  1633. "epoch [1099] L = 38.539316, acc = 0.845000\n",
  1634. "epoch [1100] L = 38.539179, acc = 0.845000\n",
  1635. "epoch [1101] L = 38.539043, acc = 0.845000\n",
  1636. "epoch [1102] L = 38.538906, acc = 0.845000\n",
  1637. "epoch [1103] L = 38.538770, acc = 0.845000\n",
  1638. "epoch [1104] L = 38.538634, acc = 0.845000\n",
  1639. "epoch [1105] L = 38.538498, acc = 0.845000\n",
  1640. "epoch [1106] L = 38.538362, acc = 0.845000\n",
  1641. "epoch [1107] L = 38.538226, acc = 0.845000\n",
  1642. "epoch [1108] L = 38.538090, acc = 0.845000\n",
  1643. "epoch [1109] L = 38.537955, acc = 0.845000\n",
  1644. "epoch [1110] L = 38.537820, acc = 0.845000\n",
  1645. "epoch [1111] L = 38.537685, acc = 0.845000\n",
  1646. "epoch [1112] L = 38.537550, acc = 0.845000\n",
  1647. "epoch [1113] L = 38.537415, acc = 0.845000\n",
  1648. "epoch [1114] L = 38.537281, acc = 0.845000\n",
  1649. "epoch [1115] L = 38.537147, acc = 0.845000\n",
  1650. "epoch [1116] L = 38.537012, acc = 0.845000\n",
  1651. "epoch [1117] L = 38.536878, acc = 0.845000\n",
  1652. "epoch [1118] L = 38.536744, acc = 0.845000\n",
  1653. "epoch [1119] L = 38.536611, acc = 0.845000\n",
  1654. "epoch [1120] L = 38.536477, acc = 0.845000\n",
  1655. "epoch [1121] L = 38.536344, acc = 0.845000\n",
  1656. "epoch [1122] L = 38.536211, acc = 0.845000\n",
  1657. "epoch [1123] L = 38.536078, acc = 0.845000\n",
  1658. "epoch [1124] L = 38.535945, acc = 0.845000\n",
  1659. "epoch [1125] L = 38.535812, acc = 0.845000\n",
  1660. "epoch [1126] L = 38.535679, acc = 0.845000\n",
  1661. "epoch [1127] L = 38.535547, acc = 0.845000\n",
  1662. "epoch [1128] L = 38.535415, acc = 0.845000\n",
  1663. "epoch [1129] L = 38.535283, acc = 0.845000\n",
  1664. "epoch [1130] L = 38.535151, acc = 0.845000\n",
  1665. "epoch [1131] L = 38.535019, acc = 0.845000\n",
  1666. "epoch [1132] L = 38.534887, acc = 0.845000\n",
  1667. "epoch [1133] L = 38.534756, acc = 0.845000\n",
  1668. "epoch [1134] L = 38.534624, acc = 0.845000\n",
  1669. "epoch [1135] L = 38.534493, acc = 0.845000\n",
  1670. "epoch [1136] L = 38.534362, acc = 0.845000\n",
  1671. "epoch [1137] L = 38.534231, acc = 0.845000\n",
  1672. "epoch [1138] L = 38.534101, acc = 0.845000\n",
  1673. "epoch [1139] L = 38.533970, acc = 0.845000\n",
  1674. "epoch [1140] L = 38.533840, acc = 0.845000\n",
  1675. "epoch [1141] L = 38.533709, acc = 0.845000\n",
  1676. "epoch [1142] L = 38.533579, acc = 0.845000\n",
  1677. "epoch [1143] L = 38.533449, acc = 0.845000\n",
  1678. "epoch [1144] L = 38.533320, acc = 0.845000\n",
  1679. "epoch [1145] L = 38.533190, acc = 0.845000\n",
  1680. "epoch [1146] L = 38.533060, acc = 0.845000\n",
  1681. "epoch [1147] L = 38.532931, acc = 0.845000\n",
  1682. "epoch [1148] L = 38.532802, acc = 0.845000\n",
  1683. "epoch [1149] L = 38.532673, acc = 0.845000\n",
  1684. "epoch [1150] L = 38.532544, acc = 0.845000\n",
  1685. "epoch [1151] L = 38.532415, acc = 0.845000\n",
  1686. "epoch [1152] L = 38.532287, acc = 0.845000\n",
  1687. "epoch [1153] L = 38.532158, acc = 0.845000\n",
  1688. "epoch [1154] L = 38.532030, acc = 0.845000\n",
  1689. "epoch [1155] L = 38.531902, acc = 0.845000\n",
  1690. "epoch [1156] L = 38.531774, acc = 0.845000\n",
  1691. "epoch [1157] L = 38.531646, acc = 0.845000\n",
  1692. "epoch [1158] L = 38.531518, acc = 0.845000\n",
  1693. "epoch [1159] L = 38.531391, acc = 0.845000\n",
  1694. "epoch [1160] L = 38.531263, acc = 0.845000\n",
  1695. "epoch [1161] L = 38.531136, acc = 0.845000\n",
  1696. "epoch [1162] L = 38.531009, acc = 0.845000\n",
  1697. "epoch [1163] L = 38.530882, acc = 0.845000\n",
  1698. "epoch [1164] L = 38.530755, acc = 0.845000\n",
  1699. "epoch [1165] L = 38.530628, acc = 0.845000\n",
  1700. "epoch [1166] L = 38.530502, acc = 0.845000\n",
  1701. "epoch [1167] L = 38.530376, acc = 0.845000\n",
  1702. "epoch [1168] L = 38.530249, acc = 0.845000\n",
  1703. "epoch [1169] L = 38.530123, acc = 0.845000\n",
  1704. "epoch [1170] L = 38.529997, acc = 0.845000\n",
  1705. "epoch [1171] L = 38.529871, acc = 0.845000\n",
  1706. "epoch [1172] L = 38.529746, acc = 0.845000\n",
  1707. "epoch [1173] L = 38.529620, acc = 0.845000\n",
  1708. "epoch [1174] L = 38.529495, acc = 0.845000\n",
  1709. "epoch [1175] L = 38.529369, acc = 0.845000\n",
  1710. "epoch [1176] L = 38.529244, acc = 0.845000\n",
  1711. "epoch [1177] L = 38.529119, acc = 0.845000\n",
  1712. "epoch [1178] L = 38.528995, acc = 0.845000\n",
  1713. "epoch [1179] L = 38.528870, acc = 0.845000\n",
  1714. "epoch [1180] L = 38.528745, acc = 0.845000\n",
  1715. "epoch [1181] L = 38.528621, acc = 0.845000\n",
  1716. "epoch [1182] L = 38.528497, acc = 0.845000\n",
  1717. "epoch [1183] L = 38.528372, acc = 0.845000\n",
  1718. "epoch [1184] L = 38.528248, acc = 0.845000\n",
  1719. "epoch [1185] L = 38.528125, acc = 0.845000\n",
  1720. "epoch [1186] L = 38.528001, acc = 0.845000\n",
  1721. "epoch [1187] L = 38.527877, acc = 0.845000\n",
  1722. "epoch [1188] L = 38.527754, acc = 0.845000\n",
  1723. "epoch [1189] L = 38.527630, acc = 0.845000\n",
  1724. "epoch [1190] L = 38.527507, acc = 0.845000\n",
  1725. "epoch [1191] L = 38.527384, acc = 0.845000\n",
  1726. "epoch [1192] L = 38.527261, acc = 0.845000\n",
  1727. "epoch [1193] L = 38.527139, acc = 0.845000\n",
  1728. "epoch [1194] L = 38.527016, acc = 0.845000\n",
  1729. "epoch [1195] L = 38.526893, acc = 0.845000\n",
  1730. "epoch [1196] L = 38.526771, acc = 0.845000\n",
  1731. "epoch [1197] L = 38.526649, acc = 0.845000\n",
  1732. "epoch [1198] L = 38.526527, acc = 0.845000\n",
  1733. "epoch [1199] L = 38.526405, acc = 0.845000\n",
  1734. "epoch [1200] L = 38.526283, acc = 0.845000\n",
  1735. "epoch [1201] L = 38.526161, acc = 0.845000\n",
  1736. "epoch [1202] L = 38.526040, acc = 0.845000\n",
  1737. "epoch [1203] L = 38.525918, acc = 0.845000\n",
  1738. "epoch [1204] L = 38.525797, acc = 0.845000\n",
  1739. "epoch [1205] L = 38.525676, acc = 0.845000\n",
  1740. "epoch [1206] L = 38.525555, acc = 0.845000\n",
  1741. "epoch [1207] L = 38.525434, acc = 0.845000\n",
  1742. "epoch [1208] L = 38.525313, acc = 0.845000\n",
  1743. "epoch [1209] L = 38.525192, acc = 0.845000\n",
  1744. "epoch [1210] L = 38.525072, acc = 0.845000\n",
  1745. "epoch [1211] L = 38.524951, acc = 0.845000\n",
  1746. "epoch [1212] L = 38.524831, acc = 0.845000\n",
  1747. "epoch [1213] L = 38.524711, acc = 0.845000\n",
  1748. "epoch [1214] L = 38.524591, acc = 0.845000\n",
  1749. "epoch [1215] L = 38.524471, acc = 0.845000\n",
  1750. "epoch [1216] L = 38.524351, acc = 0.845000\n",
  1751. "epoch [1217] L = 38.524231, acc = 0.845000\n",
  1752. "epoch [1218] L = 38.524112, acc = 0.845000\n",
  1753. "epoch [1219] L = 38.523993, acc = 0.845000\n",
  1754. "epoch [1220] L = 38.523873, acc = 0.845000\n",
  1755. "epoch [1221] L = 38.523754, acc = 0.845000\n",
  1756. "epoch [1222] L = 38.523635, acc = 0.845000\n",
  1757. "epoch [1223] L = 38.523516, acc = 0.845000\n",
  1758. "epoch [1224] L = 38.523397, acc = 0.845000\n",
  1759. "epoch [1225] L = 38.523279, acc = 0.845000\n",
  1760. "epoch [1226] L = 38.523160, acc = 0.845000\n",
  1761. "epoch [1227] L = 38.523042, acc = 0.845000\n",
  1762. "epoch [1228] L = 38.522924, acc = 0.845000\n",
  1763. "epoch [1229] L = 38.522806, acc = 0.845000\n",
  1764. "epoch [1230] L = 38.522688, acc = 0.845000\n",
  1765. "epoch [1231] L = 38.522570, acc = 0.845000\n",
  1766. "epoch [1232] L = 38.522452, acc = 0.845000\n",
  1767. "epoch [1233] L = 38.522334, acc = 0.845000\n",
  1768. "epoch [1234] L = 38.522217, acc = 0.845000\n",
  1769. "epoch [1235] L = 38.522099, acc = 0.845000\n",
  1770. "epoch [1236] L = 38.521982, acc = 0.845000\n",
  1771. "epoch [1237] L = 38.521865, acc = 0.845000\n",
  1772. "epoch [1238] L = 38.521748, acc = 0.845000\n",
  1773. "epoch [1239] L = 38.521631, acc = 0.845000\n",
  1774. "epoch [1240] L = 38.521514, acc = 0.845000\n",
  1775. "epoch [1241] L = 38.521397, acc = 0.845000\n",
  1776. "epoch [1242] L = 38.521281, acc = 0.845000\n",
  1777. "epoch [1243] L = 38.521164, acc = 0.845000\n",
  1778. "epoch [1244] L = 38.521048, acc = 0.845000\n",
  1779. "epoch [1245] L = 38.520932, acc = 0.845000\n",
  1780. "epoch [1246] L = 38.520816, acc = 0.845000\n",
  1781. "epoch [1247] L = 38.520700, acc = 0.845000\n",
  1782. "epoch [1248] L = 38.520584, acc = 0.845000\n",
  1783. "epoch [1249] L = 38.520468, acc = 0.845000\n",
  1784. "epoch [1250] L = 38.520353, acc = 0.845000\n",
  1785. "epoch [1251] L = 38.520237, acc = 0.845000\n",
  1786. "epoch [1252] L = 38.520122, acc = 0.845000\n",
  1787. "epoch [1253] L = 38.520006, acc = 0.845000\n",
  1788. "epoch [1254] L = 38.519891, acc = 0.845000\n",
  1789. "epoch [1255] L = 38.519776, acc = 0.845000\n",
  1790. "epoch [1256] L = 38.519661, acc = 0.845000\n",
  1791. "epoch [1257] L = 38.519547, acc = 0.845000\n",
  1792. "epoch [1258] L = 38.519432, acc = 0.845000\n",
  1793. "epoch [1259] L = 38.519317, acc = 0.845000\n",
  1794. "epoch [1260] L = 38.519203, acc = 0.845000\n",
  1795. "epoch [1261] L = 38.519089, acc = 0.845000\n",
  1796. "epoch [1262] L = 38.518974, acc = 0.845000\n",
  1797. "epoch [1263] L = 38.518860, acc = 0.845000\n",
  1798. "epoch [1264] L = 38.518746, acc = 0.845000\n",
  1799. "epoch [1265] L = 38.518632, acc = 0.845000\n",
  1800. "epoch [1266] L = 38.518519, acc = 0.845000\n",
  1801. "epoch [1267] L = 38.518405, acc = 0.845000\n",
  1802. "epoch [1268] L = 38.518291, acc = 0.845000\n",
  1803. "epoch [1269] L = 38.518178, acc = 0.845000\n",
  1804. "epoch [1270] L = 38.518065, acc = 0.845000\n",
  1805. "epoch [1271] L = 38.517951, acc = 0.845000\n",
  1806. "epoch [1272] L = 38.517838, acc = 0.845000\n",
  1807. "epoch [1273] L = 38.517725, acc = 0.845000\n",
  1808. "epoch [1274] L = 38.517612, acc = 0.845000\n",
  1809. "epoch [1275] L = 38.517500, acc = 0.845000\n",
  1810. "epoch [1276] L = 38.517387, acc = 0.845000\n",
  1811. "epoch [1277] L = 38.517274, acc = 0.845000\n",
  1812. "epoch [1278] L = 38.517162, acc = 0.845000\n",
  1813. "epoch [1279] L = 38.517050, acc = 0.845000\n",
  1814. "epoch [1280] L = 38.516937, acc = 0.845000\n",
  1815. "epoch [1281] L = 38.516825, acc = 0.845000\n",
  1816. "epoch [1282] L = 38.516713, acc = 0.845000\n",
  1817. "epoch [1283] L = 38.516601, acc = 0.845000\n",
  1818. "epoch [1284] L = 38.516490, acc = 0.845000\n",
  1819. "epoch [1285] L = 38.516378, acc = 0.845000\n",
  1820. "epoch [1286] L = 38.516266, acc = 0.845000\n",
  1821. "epoch [1287] L = 38.516155, acc = 0.845000\n",
  1822. "epoch [1288] L = 38.516044, acc = 0.845000\n",
  1823. "epoch [1289] L = 38.515932, acc = 0.845000\n",
  1824. "epoch [1290] L = 38.515821, acc = 0.845000\n",
  1825. "epoch [1291] L = 38.515710, acc = 0.845000\n",
  1826. "epoch [1292] L = 38.515599, acc = 0.845000\n",
  1827. "epoch [1293] L = 38.515488, acc = 0.845000\n",
  1828. "epoch [1294] L = 38.515378, acc = 0.845000\n",
  1829. "epoch [1295] L = 38.515267, acc = 0.845000\n",
  1830. "epoch [1296] L = 38.515157, acc = 0.845000\n",
  1831. "epoch [1297] L = 38.515046, acc = 0.845000\n",
  1832. "epoch [1298] L = 38.514936, acc = 0.845000\n",
  1833. "epoch [1299] L = 38.514826, acc = 0.845000\n",
  1834. "epoch [1300] L = 38.514716, acc = 0.845000\n",
  1835. "epoch [1301] L = 38.514606, acc = 0.845000\n",
  1836. "epoch [1302] L = 38.514496, acc = 0.845000\n",
  1837. "epoch [1303] L = 38.514386, acc = 0.845000\n",
  1838. "epoch [1304] L = 38.514276, acc = 0.845000\n",
  1839. "epoch [1305] L = 38.514167, acc = 0.845000\n",
  1840. "epoch [1306] L = 38.514057, acc = 0.845000\n",
  1841. "epoch [1307] L = 38.513948, acc = 0.845000\n",
  1842. "epoch [1308] L = 38.513839, acc = 0.845000\n",
  1843. "epoch [1309] L = 38.513729, acc = 0.845000\n",
  1844. "epoch [1310] L = 38.513620, acc = 0.845000\n",
  1845. "epoch [1311] L = 38.513511, acc = 0.845000\n",
  1846. "epoch [1312] L = 38.513403, acc = 0.845000\n",
  1847. "epoch [1313] L = 38.513294, acc = 0.845000\n",
  1848. "epoch [1314] L = 38.513185, acc = 0.845000\n",
  1849. "epoch [1315] L = 38.513077, acc = 0.845000\n",
  1850. "epoch [1316] L = 38.512968, acc = 0.845000\n",
  1851. "epoch [1317] L = 38.512860, acc = 0.845000\n",
  1852. "epoch [1318] L = 38.512752, acc = 0.845000\n",
  1853. "epoch [1319] L = 38.512643, acc = 0.845000\n",
  1854. "epoch [1320] L = 38.512535, acc = 0.845000\n",
  1855. "epoch [1321] L = 38.512427, acc = 0.845000\n",
  1856. "epoch [1322] L = 38.512320, acc = 0.845000\n",
  1857. "epoch [1323] L = 38.512212, acc = 0.845000\n",
  1858. "epoch [1324] L = 38.512104, acc = 0.845000\n",
  1859. "epoch [1325] L = 38.511997, acc = 0.845000\n",
  1860. "epoch [1326] L = 38.511889, acc = 0.845000\n",
  1861. "epoch [1327] L = 38.511782, acc = 0.845000\n",
  1862. "epoch [1328] L = 38.511674, acc = 0.845000\n",
  1863. "epoch [1329] L = 38.511567, acc = 0.845000\n",
  1864. "epoch [1330] L = 38.511460, acc = 0.845000\n",
  1865. "epoch [1331] L = 38.511353, acc = 0.845000\n",
  1866. "epoch [1332] L = 38.511246, acc = 0.845000\n",
  1867. "epoch [1333] L = 38.511140, acc = 0.845000\n",
  1868. "epoch [1334] L = 38.511033, acc = 0.845000\n",
  1869. "epoch [1335] L = 38.510926, acc = 0.845000\n",
  1870. "epoch [1336] L = 38.510820, acc = 0.845000\n",
  1871. "epoch [1337] L = 38.510713, acc = 0.845000\n",
  1872. "epoch [1338] L = 38.510607, acc = 0.845000\n",
  1873. "epoch [1339] L = 38.510501, acc = 0.845000\n",
  1874. "epoch [1340] L = 38.510395, acc = 0.845000\n",
  1875. "epoch [1341] L = 38.510289, acc = 0.845000\n",
  1876. "epoch [1342] L = 38.510183, acc = 0.845000\n",
  1877. "epoch [1343] L = 38.510077, acc = 0.845000\n",
  1878. "epoch [1344] L = 38.509971, acc = 0.845000\n",
  1879. "epoch [1345] L = 38.509866, acc = 0.845000\n",
  1880. "epoch [1346] L = 38.509760, acc = 0.845000\n",
  1881. "epoch [1347] L = 38.509655, acc = 0.845000\n",
  1882. "epoch [1348] L = 38.509549, acc = 0.845000\n",
  1883. "epoch [1349] L = 38.509444, acc = 0.845000\n",
  1884. "epoch [1350] L = 38.509339, acc = 0.845000\n",
  1885. "epoch [1351] L = 38.509234, acc = 0.845000\n",
  1886. "epoch [1352] L = 38.509129, acc = 0.845000\n",
  1887. "epoch [1353] L = 38.509024, acc = 0.845000\n",
  1888. "epoch [1354] L = 38.508919, acc = 0.845000\n",
  1889. "epoch [1355] L = 38.508814, acc = 0.845000\n",
  1890. "epoch [1356] L = 38.508710, acc = 0.845000\n",
  1891. "epoch [1357] L = 38.508605, acc = 0.845000\n",
  1892. "epoch [1358] L = 38.508501, acc = 0.845000\n",
  1893. "epoch [1359] L = 38.508396, acc = 0.845000\n",
  1894. "epoch [1360] L = 38.508292, acc = 0.845000\n",
  1895. "epoch [1361] L = 38.508188, acc = 0.845000\n",
  1896. "epoch [1362] L = 38.508084, acc = 0.845000\n",
  1897. "epoch [1363] L = 38.507980, acc = 0.845000\n",
  1898. "epoch [1364] L = 38.507876, acc = 0.845000\n",
  1899. "epoch [1365] L = 38.507772, acc = 0.845000\n",
  1900. "epoch [1366] L = 38.507669, acc = 0.845000\n",
  1901. "epoch [1367] L = 38.507565, acc = 0.845000\n",
  1902. "epoch [1368] L = 38.507461, acc = 0.845000\n",
  1903. "epoch [1369] L = 38.507358, acc = 0.845000\n",
  1904. "epoch [1370] L = 38.507255, acc = 0.845000\n",
  1905. "epoch [1371] L = 38.507151, acc = 0.845000\n",
  1906. "epoch [1372] L = 38.507048, acc = 0.845000\n",
  1907. "epoch [1373] L = 38.506945, acc = 0.845000\n",
  1908. "epoch [1374] L = 38.506842, acc = 0.845000\n",
  1909. "epoch [1375] L = 38.506739, acc = 0.845000\n",
  1910. "epoch [1376] L = 38.506636, acc = 0.845000\n",
  1911. "epoch [1377] L = 38.506534, acc = 0.845000\n",
  1912. "epoch [1378] L = 38.506431, acc = 0.845000\n",
  1913. "epoch [1379] L = 38.506328, acc = 0.845000\n",
  1914. "epoch [1380] L = 38.506226, acc = 0.845000\n",
  1915. "epoch [1381] L = 38.506123, acc = 0.845000\n",
  1916. "epoch [1382] L = 38.506021, acc = 0.845000\n",
  1917. "epoch [1383] L = 38.505919, acc = 0.845000\n",
  1918. "epoch [1384] L = 38.505817, acc = 0.845000\n",
  1919. "epoch [1385] L = 38.505715, acc = 0.845000\n",
  1920. "epoch [1386] L = 38.505613, acc = 0.845000\n",
  1921. "epoch [1387] L = 38.505511, acc = 0.845000\n",
  1922. "epoch [1388] L = 38.505409, acc = 0.845000\n",
  1923. "epoch [1389] L = 38.505307, acc = 0.845000\n",
  1924. "epoch [1390] L = 38.505206, acc = 0.845000\n",
  1925. "epoch [1391] L = 38.505104, acc = 0.845000\n",
  1926. "epoch [1392] L = 38.505003, acc = 0.845000\n",
  1927. "epoch [1393] L = 38.504901, acc = 0.845000\n",
  1928. "epoch [1394] L = 38.504800, acc = 0.845000\n",
  1929. "epoch [1395] L = 38.504699, acc = 0.845000\n",
  1930. "epoch [1396] L = 38.504598, acc = 0.845000\n",
  1931. "epoch [1397] L = 38.504497, acc = 0.845000\n",
  1932. "epoch [1398] L = 38.504396, acc = 0.845000\n",
  1933. "epoch [1399] L = 38.504295, acc = 0.845000\n",
  1934. "epoch [1400] L = 38.504194, acc = 0.845000\n",
  1935. "epoch [1401] L = 38.504093, acc = 0.845000\n",
  1936. "epoch [1402] L = 38.503993, acc = 0.845000\n",
  1937. "epoch [1403] L = 38.503892, acc = 0.845000\n",
  1938. "epoch [1404] L = 38.503792, acc = 0.845000\n",
  1939. "epoch [1405] L = 38.503691, acc = 0.845000\n",
  1940. "epoch [1406] L = 38.503591, acc = 0.845000\n",
  1941. "epoch [1407] L = 38.503491, acc = 0.845000\n",
  1942. "epoch [1408] L = 38.503391, acc = 0.845000\n",
  1943. "epoch [1409] L = 38.503291, acc = 0.845000\n",
  1944. "epoch [1410] L = 38.503191, acc = 0.845000\n",
  1945. "epoch [1411] L = 38.503091, acc = 0.845000\n",
  1946. "epoch [1412] L = 38.502991, acc = 0.845000\n",
  1947. "epoch [1413] L = 38.502891, acc = 0.845000\n",
  1948. "epoch [1414] L = 38.502792, acc = 0.845000\n",
  1949. "epoch [1415] L = 38.502692, acc = 0.845000\n",
  1950. "epoch [1416] L = 38.502593, acc = 0.845000\n",
  1951. "epoch [1417] L = 38.502493, acc = 0.845000\n",
  1952. "epoch [1418] L = 38.502394, acc = 0.845000\n",
  1953. "epoch [1419] L = 38.502295, acc = 0.845000\n",
  1954. "epoch [1420] L = 38.502195, acc = 0.845000\n",
  1955. "epoch [1421] L = 38.502096, acc = 0.845000\n",
  1956. "epoch [1422] L = 38.501997, acc = 0.845000\n",
  1957. "epoch [1423] L = 38.501898, acc = 0.845000\n",
  1958. "epoch [1424] L = 38.501800, acc = 0.845000\n",
  1959. "epoch [1425] L = 38.501701, acc = 0.845000\n",
  1960. "epoch [1426] L = 38.501602, acc = 0.845000\n",
  1961. "epoch [1427] L = 38.501503, acc = 0.845000\n",
  1962. "epoch [1428] L = 38.501405, acc = 0.845000\n",
  1963. "epoch [1429] L = 38.501307, acc = 0.845000\n",
  1964. "epoch [1430] L = 38.501208, acc = 0.845000\n",
  1965. "epoch [1431] L = 38.501110, acc = 0.845000\n",
  1966. "epoch [1432] L = 38.501012, acc = 0.845000\n",
  1967. "epoch [1433] L = 38.500913, acc = 0.845000\n",
  1968. "epoch [1434] L = 38.500815, acc = 0.845000\n",
  1969. "epoch [1435] L = 38.500717, acc = 0.845000\n",
  1970. "epoch [1436] L = 38.500619, acc = 0.845000\n",
  1971. "epoch [1437] L = 38.500522, acc = 0.845000\n",
  1972. "epoch [1438] L = 38.500424, acc = 0.845000\n",
  1973. "epoch [1439] L = 38.500326, acc = 0.845000\n",
  1974. "epoch [1440] L = 38.500229, acc = 0.845000\n",
  1975. "epoch [1441] L = 38.500131, acc = 0.845000\n",
  1976. "epoch [1442] L = 38.500034, acc = 0.845000\n",
  1977. "epoch [1443] L = 38.499936, acc = 0.845000\n",
  1978. "epoch [1444] L = 38.499839, acc = 0.845000\n",
  1979. "epoch [1445] L = 38.499742, acc = 0.845000\n",
  1980. "epoch [1446] L = 38.499644, acc = 0.845000\n",
  1981. "epoch [1447] L = 38.499547, acc = 0.845000\n",
  1982. "epoch [1448] L = 38.499450, acc = 0.845000\n",
  1983. "epoch [1449] L = 38.499353, acc = 0.845000\n",
  1984. "epoch [1450] L = 38.499257, acc = 0.845000\n",
  1985. "epoch [1451] L = 38.499160, acc = 0.845000\n",
  1986. "epoch [1452] L = 38.499063, acc = 0.845000\n",
  1987. "epoch [1453] L = 38.498966, acc = 0.845000\n",
  1988. "epoch [1454] L = 38.498870, acc = 0.845000\n",
  1989. "epoch [1455] L = 38.498773, acc = 0.845000\n",
  1990. "epoch [1456] L = 38.498677, acc = 0.845000\n",
  1991. "epoch [1457] L = 38.498581, acc = 0.845000\n",
  1992. "epoch [1458] L = 38.498484, acc = 0.845000\n",
  1993. "epoch [1459] L = 38.498388, acc = 0.845000\n",
  1994. "epoch [1460] L = 38.498292, acc = 0.845000\n",
  1995. "epoch [1461] L = 38.498196, acc = 0.845000\n",
  1996. "epoch [1462] L = 38.498100, acc = 0.845000\n",
  1997. "epoch [1463] L = 38.498004, acc = 0.845000\n",
  1998. "epoch [1464] L = 38.497908, acc = 0.845000\n",
  1999. "epoch [1465] L = 38.497812, acc = 0.845000\n",
  2000. "epoch [1466] L = 38.497717, acc = 0.845000\n",
  2001. "epoch [1467] L = 38.497621, acc = 0.845000\n",
  2002. "epoch [1468] L = 38.497526, acc = 0.845000\n",
  2003. "epoch [1469] L = 38.497430, acc = 0.845000\n",
  2004. "epoch [1470] L = 38.497335, acc = 0.845000\n",
  2005. "epoch [1471] L = 38.497239, acc = 0.845000\n",
  2006. "epoch [1472] L = 38.497144, acc = 0.845000\n",
  2007. "epoch [1473] L = 38.497049, acc = 0.845000\n",
  2008. "epoch [1474] L = 38.496954, acc = 0.845000\n",
  2009. "epoch [1475] L = 38.496859, acc = 0.845000\n",
  2010. "epoch [1476] L = 38.496764, acc = 0.845000\n",
  2011. "epoch [1477] L = 38.496669, acc = 0.845000\n",
  2012. "epoch [1478] L = 38.496574, acc = 0.845000\n",
  2013. "epoch [1479] L = 38.496479, acc = 0.845000\n",
  2014. "epoch [1480] L = 38.496385, acc = 0.845000\n",
  2015. "epoch [1481] L = 38.496290, acc = 0.845000\n",
  2016. "epoch [1482] L = 38.496195, acc = 0.845000\n",
  2017. "epoch [1483] L = 38.496101, acc = 0.845000\n",
  2018. "epoch [1484] L = 38.496006, acc = 0.845000\n",
  2019. "epoch [1485] L = 38.495912, acc = 0.845000\n",
  2020. "epoch [1486] L = 38.495818, acc = 0.845000\n",
  2021. "epoch [1487] L = 38.495724, acc = 0.845000\n",
  2022. "epoch [1488] L = 38.495629, acc = 0.845000\n",
  2023. "epoch [1489] L = 38.495535, acc = 0.845000\n",
  2024. "epoch [1490] L = 38.495441, acc = 0.845000\n",
  2025. "epoch [1491] L = 38.495347, acc = 0.845000\n",
  2026. "epoch [1492] L = 38.495254, acc = 0.845000\n",
  2027. "epoch [1493] L = 38.495160, acc = 0.845000\n",
  2028. "epoch [1494] L = 38.495066, acc = 0.845000\n",
  2029. "epoch [1495] L = 38.494972, acc = 0.845000\n",
  2030. "epoch [1496] L = 38.494879, acc = 0.845000\n",
  2031. "epoch [1497] L = 38.494785, acc = 0.845000\n",
  2032. "epoch [1498] L = 38.494692, acc = 0.845000\n",
  2033. "epoch [1499] L = 38.494598, acc = 0.845000\n",
  2034. "epoch [1500] L = 38.494505, acc = 0.845000\n",
  2035. "epoch [1501] L = 38.494412, acc = 0.845000\n",
  2036. "epoch [1502] L = 38.494319, acc = 0.845000\n",
  2037. "epoch [1503] L = 38.494225, acc = 0.845000\n",
  2038. "epoch [1504] L = 38.494132, acc = 0.845000\n",
  2039. "epoch [1505] L = 38.494039, acc = 0.845000\n",
  2040. "epoch [1506] L = 38.493946, acc = 0.845000\n",
  2041. "epoch [1507] L = 38.493854, acc = 0.845000\n",
  2042. "epoch [1508] L = 38.493761, acc = 0.845000\n",
  2043. "epoch [1509] L = 38.493668, acc = 0.845000\n",
  2044. "epoch [1510] L = 38.493575, acc = 0.845000\n",
  2045. "epoch [1511] L = 38.493483, acc = 0.845000\n",
  2046. "epoch [1512] L = 38.493390, acc = 0.845000\n",
  2047. "epoch [1513] L = 38.493298, acc = 0.845000\n",
  2048. "epoch [1514] L = 38.493205, acc = 0.845000\n",
  2049. "epoch [1515] L = 38.493113, acc = 0.845000\n",
  2050. "epoch [1516] L = 38.493021, acc = 0.845000\n",
  2051. "epoch [1517] L = 38.492929, acc = 0.845000\n",
  2052. "epoch [1518] L = 38.492836, acc = 0.845000\n",
  2053. "epoch [1519] L = 38.492744, acc = 0.845000\n",
  2054. "epoch [1520] L = 38.492652, acc = 0.845000\n",
  2055. "epoch [1521] L = 38.492560, acc = 0.845000\n",
  2056. "epoch [1522] L = 38.492468, acc = 0.845000\n",
  2057. "epoch [1523] L = 38.492377, acc = 0.845000\n",
  2058. "epoch [1524] L = 38.492285, acc = 0.845000\n",
  2059. "epoch [1525] L = 38.492193, acc = 0.845000\n",
  2060. "epoch [1526] L = 38.492101, acc = 0.845000\n",
  2061. "epoch [1527] L = 38.492010, acc = 0.845000\n",
  2062. "epoch [1528] L = 38.491918, acc = 0.845000\n",
  2063. "epoch [1529] L = 38.491827, acc = 0.845000\n",
  2064. "epoch [1530] L = 38.491736, acc = 0.845000\n",
  2065. "epoch [1531] L = 38.491644, acc = 0.845000\n",
  2066. "epoch [1532] L = 38.491553, acc = 0.845000\n",
  2067. "epoch [1533] L = 38.491462, acc = 0.845000\n",
  2068. "epoch [1534] L = 38.491371, acc = 0.845000\n",
  2069. "epoch [1535] L = 38.491280, acc = 0.845000\n",
  2070. "epoch [1536] L = 38.491189, acc = 0.845000\n",
  2071. "epoch [1537] L = 38.491098, acc = 0.845000\n",
  2072. "epoch [1538] L = 38.491007, acc = 0.845000\n",
  2073. "epoch [1539] L = 38.490916, acc = 0.845000\n",
  2074. "epoch [1540] L = 38.490825, acc = 0.845000\n",
  2075. "epoch [1541] L = 38.490735, acc = 0.845000\n",
  2076. "epoch [1542] L = 38.490644, acc = 0.845000\n",
  2077. "epoch [1543] L = 38.490553, acc = 0.845000\n",
  2078. "epoch [1544] L = 38.490463, acc = 0.845000\n",
  2079. "epoch [1545] L = 38.490372, acc = 0.845000\n",
  2080. "epoch [1546] L = 38.490282, acc = 0.845000\n",
  2081. "epoch [1547] L = 38.490192, acc = 0.845000\n",
  2082. "epoch [1548] L = 38.490101, acc = 0.845000\n",
  2083. "epoch [1549] L = 38.490011, acc = 0.845000\n",
  2084. "epoch [1550] L = 38.489921, acc = 0.845000\n",
  2085. "epoch [1551] L = 38.489831, acc = 0.845000\n",
  2086. "epoch [1552] L = 38.489741, acc = 0.845000\n",
  2087. "epoch [1553] L = 38.489651, acc = 0.845000\n",
  2088. "epoch [1554] L = 38.489561, acc = 0.845000\n",
  2089. "epoch [1555] L = 38.489471, acc = 0.845000\n",
  2090. "epoch [1556] L = 38.489381, acc = 0.845000\n",
  2091. "epoch [1557] L = 38.489292, acc = 0.845000\n",
  2092. "epoch [1558] L = 38.489202, acc = 0.845000\n",
  2093. "epoch [1559] L = 38.489112, acc = 0.845000\n",
  2094. "epoch [1560] L = 38.489023, acc = 0.845000\n",
  2095. "epoch [1561] L = 38.488933, acc = 0.845000\n",
  2096. "epoch [1562] L = 38.488844, acc = 0.845000\n",
  2097. "epoch [1563] L = 38.488755, acc = 0.845000\n",
  2098. "epoch [1564] L = 38.488665, acc = 0.845000\n",
  2099. "epoch [1565] L = 38.488576, acc = 0.845000\n",
  2100. "epoch [1566] L = 38.488487, acc = 0.845000\n",
  2101. "epoch [1567] L = 38.488398, acc = 0.845000\n",
  2102. "epoch [1568] L = 38.488309, acc = 0.845000\n",
  2103. "epoch [1569] L = 38.488220, acc = 0.845000\n",
  2104. "epoch [1570] L = 38.488131, acc = 0.845000\n",
  2105. "epoch [1571] L = 38.488042, acc = 0.845000\n",
  2106. "epoch [1572] L = 38.487953, acc = 0.845000\n",
  2107. "epoch [1573] L = 38.487864, acc = 0.845000\n",
  2108. "epoch [1574] L = 38.487775, acc = 0.845000\n",
  2109. "epoch [1575] L = 38.487687, acc = 0.845000\n",
  2110. "epoch [1576] L = 38.487598, acc = 0.845000\n",
  2111. "epoch [1577] L = 38.487510, acc = 0.845000\n",
  2112. "epoch [1578] L = 38.487421, acc = 0.845000\n",
  2113. "epoch [1579] L = 38.487333, acc = 0.845000\n",
  2114. "epoch [1580] L = 38.487244, acc = 0.845000\n",
  2115. "epoch [1581] L = 38.487156, acc = 0.845000\n"
  2116. ]
  2117. },
  2118. {
  2119. "name": "stdout",
  2120. "output_type": "stream",
  2121. "text": [
  2122. "epoch [1582] L = 38.487068, acc = 0.845000\n",
  2123. "epoch [1583] L = 38.486979, acc = 0.845000\n",
  2124. "epoch [1584] L = 38.486891, acc = 0.845000\n",
  2125. "epoch [1585] L = 38.486803, acc = 0.845000\n",
  2126. "epoch [1586] L = 38.486715, acc = 0.845000\n",
  2127. "epoch [1587] L = 38.486627, acc = 0.845000\n",
  2128. "epoch [1588] L = 38.486539, acc = 0.845000\n",
  2129. "epoch [1589] L = 38.486451, acc = 0.845000\n",
  2130. "epoch [1590] L = 38.486364, acc = 0.845000\n",
  2131. "epoch [1591] L = 38.486276, acc = 0.845000\n",
  2132. "epoch [1592] L = 38.486188, acc = 0.845000\n",
  2133. "epoch [1593] L = 38.486101, acc = 0.845000\n",
  2134. "epoch [1594] L = 38.486013, acc = 0.845000\n",
  2135. "epoch [1595] L = 38.485925, acc = 0.845000\n",
  2136. "epoch [1596] L = 38.485838, acc = 0.845000\n",
  2137. "epoch [1597] L = 38.485750, acc = 0.845000\n",
  2138. "epoch [1598] L = 38.485663, acc = 0.845000\n",
  2139. "epoch [1599] L = 38.485576, acc = 0.845000\n",
  2140. "epoch [1600] L = 38.485489, acc = 0.845000\n",
  2141. "epoch [1601] L = 38.485401, acc = 0.845000\n",
  2142. "epoch [1602] L = 38.485314, acc = 0.845000\n",
  2143. "epoch [1603] L = 38.485227, acc = 0.845000\n",
  2144. "epoch [1604] L = 38.485140, acc = 0.845000\n",
  2145. "epoch [1605] L = 38.485053, acc = 0.845000\n",
  2146. "epoch [1606] L = 38.484966, acc = 0.845000\n",
  2147. "epoch [1607] L = 38.484879, acc = 0.845000\n",
  2148. "epoch [1608] L = 38.484792, acc = 0.845000\n",
  2149. "epoch [1609] L = 38.484706, acc = 0.845000\n",
  2150. "epoch [1610] L = 38.484619, acc = 0.845000\n",
  2151. "epoch [1611] L = 38.484532, acc = 0.845000\n",
  2152. "epoch [1612] L = 38.484446, acc = 0.845000\n",
  2153. "epoch [1613] L = 38.484359, acc = 0.845000\n",
  2154. "epoch [1614] L = 38.484273, acc = 0.845000\n",
  2155. "epoch [1615] L = 38.484186, acc = 0.845000\n",
  2156. "epoch [1616] L = 38.484100, acc = 0.845000\n",
  2157. "epoch [1617] L = 38.484014, acc = 0.845000\n",
  2158. "epoch [1618] L = 38.483927, acc = 0.845000\n",
  2159. "epoch [1619] L = 38.483841, acc = 0.845000\n",
  2160. "epoch [1620] L = 38.483755, acc = 0.845000\n",
  2161. "epoch [1621] L = 38.483669, acc = 0.845000\n",
  2162. "epoch [1622] L = 38.483583, acc = 0.845000\n",
  2163. "epoch [1623] L = 38.483497, acc = 0.845000\n",
  2164. "epoch [1624] L = 38.483411, acc = 0.845000\n",
  2165. "epoch [1625] L = 38.483325, acc = 0.845000\n",
  2166. "epoch [1626] L = 38.483239, acc = 0.845000\n",
  2167. "epoch [1627] L = 38.483153, acc = 0.845000\n",
  2168. "epoch [1628] L = 38.483067, acc = 0.845000\n",
  2169. "epoch [1629] L = 38.482982, acc = 0.845000\n",
  2170. "epoch [1630] L = 38.482896, acc = 0.845000\n",
  2171. "epoch [1631] L = 38.482810, acc = 0.845000\n",
  2172. "epoch [1632] L = 38.482725, acc = 0.845000\n",
  2173. "epoch [1633] L = 38.482639, acc = 0.845000\n",
  2174. "epoch [1634] L = 38.482554, acc = 0.845000\n",
  2175. "epoch [1635] L = 38.482469, acc = 0.845000\n",
  2176. "epoch [1636] L = 38.482383, acc = 0.845000\n",
  2177. "epoch [1637] L = 38.482298, acc = 0.845000\n",
  2178. "epoch [1638] L = 38.482213, acc = 0.845000\n",
  2179. "epoch [1639] L = 38.482127, acc = 0.845000\n",
  2180. "epoch [1640] L = 38.482042, acc = 0.845000\n",
  2181. "epoch [1641] L = 38.481957, acc = 0.845000\n",
  2182. "epoch [1642] L = 38.481872, acc = 0.845000\n",
  2183. "epoch [1643] L = 38.481787, acc = 0.845000\n",
  2184. "epoch [1644] L = 38.481702, acc = 0.845000\n",
  2185. "epoch [1645] L = 38.481617, acc = 0.845000\n",
  2186. "epoch [1646] L = 38.481533, acc = 0.845000\n",
  2187. "epoch [1647] L = 38.481448, acc = 0.845000\n",
  2188. "epoch [1648] L = 38.481363, acc = 0.845000\n",
  2189. "epoch [1649] L = 38.481278, acc = 0.845000\n",
  2190. "epoch [1650] L = 38.481194, acc = 0.845000\n",
  2191. "epoch [1651] L = 38.481109, acc = 0.845000\n",
  2192. "epoch [1652] L = 38.481025, acc = 0.845000\n",
  2193. "epoch [1653] L = 38.480940, acc = 0.845000\n",
  2194. "epoch [1654] L = 38.480856, acc = 0.845000\n",
  2195. "epoch [1655] L = 38.480771, acc = 0.845000\n",
  2196. "epoch [1656] L = 38.480687, acc = 0.845000\n",
  2197. "epoch [1657] L = 38.480603, acc = 0.845000\n",
  2198. "epoch [1658] L = 38.480519, acc = 0.845000\n",
  2199. "epoch [1659] L = 38.480434, acc = 0.845000\n",
  2200. "epoch [1660] L = 38.480350, acc = 0.845000\n",
  2201. "epoch [1661] L = 38.480266, acc = 0.845000\n",
  2202. "epoch [1662] L = 38.480182, acc = 0.845000\n",
  2203. "epoch [1663] L = 38.480098, acc = 0.845000\n",
  2204. "epoch [1664] L = 38.480014, acc = 0.845000\n",
  2205. "epoch [1665] L = 38.479930, acc = 0.845000\n",
  2206. "epoch [1666] L = 38.479847, acc = 0.845000\n",
  2207. "epoch [1667] L = 38.479763, acc = 0.845000\n",
  2208. "epoch [1668] L = 38.479679, acc = 0.845000\n",
  2209. "epoch [1669] L = 38.479595, acc = 0.845000\n",
  2210. "epoch [1670] L = 38.479512, acc = 0.845000\n",
  2211. "epoch [1671] L = 38.479428, acc = 0.845000\n",
  2212. "epoch [1672] L = 38.479345, acc = 0.845000\n",
  2213. "epoch [1673] L = 38.479261, acc = 0.845000\n",
  2214. "epoch [1674] L = 38.479178, acc = 0.845000\n",
  2215. "epoch [1675] L = 38.479094, acc = 0.845000\n",
  2216. "epoch [1676] L = 38.479011, acc = 0.845000\n",
  2217. "epoch [1677] L = 38.478928, acc = 0.845000\n",
  2218. "epoch [1678] L = 38.478844, acc = 0.845000\n",
  2219. "epoch [1679] L = 38.478761, acc = 0.845000\n",
  2220. "epoch [1680] L = 38.478678, acc = 0.845000\n",
  2221. "epoch [1681] L = 38.478595, acc = 0.845000\n",
  2222. "epoch [1682] L = 38.478512, acc = 0.845000\n",
  2223. "epoch [1683] L = 38.478429, acc = 0.845000\n",
  2224. "epoch [1684] L = 38.478346, acc = 0.845000\n",
  2225. "epoch [1685] L = 38.478263, acc = 0.845000\n",
  2226. "epoch [1686] L = 38.478180, acc = 0.845000\n",
  2227. "epoch [1687] L = 38.478097, acc = 0.845000\n",
  2228. "epoch [1688] L = 38.478014, acc = 0.845000\n",
  2229. "epoch [1689] L = 38.477932, acc = 0.845000\n",
  2230. "epoch [1690] L = 38.477849, acc = 0.845000\n",
  2231. "epoch [1691] L = 38.477766, acc = 0.845000\n",
  2232. "epoch [1692] L = 38.477684, acc = 0.845000\n",
  2233. "epoch [1693] L = 38.477601, acc = 0.845000\n",
  2234. "epoch [1694] L = 38.477519, acc = 0.845000\n",
  2235. "epoch [1695] L = 38.477436, acc = 0.845000\n",
  2236. "epoch [1696] L = 38.477354, acc = 0.845000\n",
  2237. "epoch [1697] L = 38.477271, acc = 0.845000\n",
  2238. "epoch [1698] L = 38.477189, acc = 0.845000\n",
  2239. "epoch [1699] L = 38.477107, acc = 0.845000\n",
  2240. "epoch [1700] L = 38.477025, acc = 0.845000\n",
  2241. "epoch [1701] L = 38.476942, acc = 0.845000\n",
  2242. "epoch [1702] L = 38.476860, acc = 0.845000\n",
  2243. "epoch [1703] L = 38.476778, acc = 0.845000\n",
  2244. "epoch [1704] L = 38.476696, acc = 0.845000\n",
  2245. "epoch [1705] L = 38.476614, acc = 0.845000\n",
  2246. "epoch [1706] L = 38.476532, acc = 0.845000\n",
  2247. "epoch [1707] L = 38.476450, acc = 0.845000\n",
  2248. "epoch [1708] L = 38.476368, acc = 0.845000\n",
  2249. "epoch [1709] L = 38.476287, acc = 0.845000\n",
  2250. "epoch [1710] L = 38.476205, acc = 0.845000\n",
  2251. "epoch [1711] L = 38.476123, acc = 0.845000\n",
  2252. "epoch [1712] L = 38.476041, acc = 0.845000\n",
  2253. "epoch [1713] L = 38.475960, acc = 0.845000\n",
  2254. "epoch [1714] L = 38.475878, acc = 0.845000\n",
  2255. "epoch [1715] L = 38.475797, acc = 0.845000\n",
  2256. "epoch [1716] L = 38.475715, acc = 0.845000\n",
  2257. "epoch [1717] L = 38.475634, acc = 0.845000\n",
  2258. "epoch [1718] L = 38.475552, acc = 0.845000\n",
  2259. "epoch [1719] L = 38.475471, acc = 0.845000\n",
  2260. "epoch [1720] L = 38.475390, acc = 0.845000\n",
  2261. "epoch [1721] L = 38.475308, acc = 0.845000\n",
  2262. "epoch [1722] L = 38.475227, acc = 0.845000\n",
  2263. "epoch [1723] L = 38.475146, acc = 0.845000\n",
  2264. "epoch [1724] L = 38.475065, acc = 0.845000\n",
  2265. "epoch [1725] L = 38.474984, acc = 0.845000\n",
  2266. "epoch [1726] L = 38.474903, acc = 0.845000\n",
  2267. "epoch [1727] L = 38.474822, acc = 0.845000\n",
  2268. "epoch [1728] L = 38.474741, acc = 0.845000\n",
  2269. "epoch [1729] L = 38.474660, acc = 0.845000\n",
  2270. "epoch [1730] L = 38.474579, acc = 0.845000\n",
  2271. "epoch [1731] L = 38.474498, acc = 0.845000\n",
  2272. "epoch [1732] L = 38.474417, acc = 0.845000\n",
  2273. "epoch [1733] L = 38.474337, acc = 0.845000\n",
  2274. "epoch [1734] L = 38.474256, acc = 0.845000\n",
  2275. "epoch [1735] L = 38.474175, acc = 0.845000\n",
  2276. "epoch [1736] L = 38.474095, acc = 0.845000\n",
  2277. "epoch [1737] L = 38.474014, acc = 0.845000\n",
  2278. "epoch [1738] L = 38.473934, acc = 0.845000\n",
  2279. "epoch [1739] L = 38.473853, acc = 0.845000\n",
  2280. "epoch [1740] L = 38.473773, acc = 0.845000\n",
  2281. "epoch [1741] L = 38.473692, acc = 0.845000\n",
  2282. "epoch [1742] L = 38.473612, acc = 0.845000\n",
  2283. "epoch [1743] L = 38.473532, acc = 0.845000\n",
  2284. "epoch [1744] L = 38.473451, acc = 0.845000\n",
  2285. "epoch [1745] L = 38.473371, acc = 0.845000\n",
  2286. "epoch [1746] L = 38.473291, acc = 0.845000\n",
  2287. "epoch [1747] L = 38.473211, acc = 0.845000\n",
  2288. "epoch [1748] L = 38.473131, acc = 0.845000\n",
  2289. "epoch [1749] L = 38.473051, acc = 0.845000\n",
  2290. "epoch [1750] L = 38.472970, acc = 0.845000\n",
  2291. "epoch [1751] L = 38.472891, acc = 0.845000\n",
  2292. "epoch [1752] L = 38.472811, acc = 0.845000\n",
  2293. "epoch [1753] L = 38.472731, acc = 0.845000\n",
  2294. "epoch [1754] L = 38.472651, acc = 0.845000\n",
  2295. "epoch [1755] L = 38.472571, acc = 0.845000\n",
  2296. "epoch [1756] L = 38.472491, acc = 0.845000\n",
  2297. "epoch [1757] L = 38.472412, acc = 0.845000\n",
  2298. "epoch [1758] L = 38.472332, acc = 0.845000\n",
  2299. "epoch [1759] L = 38.472252, acc = 0.845000\n",
  2300. "epoch [1760] L = 38.472173, acc = 0.845000\n",
  2301. "epoch [1761] L = 38.472093, acc = 0.845000\n",
  2302. "epoch [1762] L = 38.472014, acc = 0.845000\n",
  2303. "epoch [1763] L = 38.471934, acc = 0.845000\n",
  2304. "epoch [1764] L = 38.471855, acc = 0.845000\n",
  2305. "epoch [1765] L = 38.471775, acc = 0.845000\n",
  2306. "epoch [1766] L = 38.471696, acc = 0.845000\n",
  2307. "epoch [1767] L = 38.471617, acc = 0.845000\n",
  2308. "epoch [1768] L = 38.471537, acc = 0.845000\n",
  2309. "epoch [1769] L = 38.471458, acc = 0.845000\n",
  2310. "epoch [1770] L = 38.471379, acc = 0.845000\n",
  2311. "epoch [1771] L = 38.471300, acc = 0.845000\n",
  2312. "epoch [1772] L = 38.471221, acc = 0.845000\n",
  2313. "epoch [1773] L = 38.471142, acc = 0.845000\n",
  2314. "epoch [1774] L = 38.471063, acc = 0.845000\n",
  2315. "epoch [1775] L = 38.470984, acc = 0.845000\n",
  2316. "epoch [1776] L = 38.470905, acc = 0.845000\n",
  2317. "epoch [1777] L = 38.470826, acc = 0.845000\n",
  2318. "epoch [1778] L = 38.470747, acc = 0.845000\n",
  2319. "epoch [1779] L = 38.470668, acc = 0.845000\n",
  2320. "epoch [1780] L = 38.470589, acc = 0.845000\n",
  2321. "epoch [1781] L = 38.470511, acc = 0.845000\n",
  2322. "epoch [1782] L = 38.470432, acc = 0.845000\n",
  2323. "epoch [1783] L = 38.470353, acc = 0.845000\n",
  2324. "epoch [1784] L = 38.470275, acc = 0.845000\n",
  2325. "epoch [1785] L = 38.470196, acc = 0.845000\n",
  2326. "epoch [1786] L = 38.470118, acc = 0.845000\n",
  2327. "epoch [1787] L = 38.470039, acc = 0.845000\n",
  2328. "epoch [1788] L = 38.469961, acc = 0.845000\n",
  2329. "epoch [1789] L = 38.469882, acc = 0.845000\n",
  2330. "epoch [1790] L = 38.469804, acc = 0.845000\n",
  2331. "epoch [1791] L = 38.469725, acc = 0.845000\n",
  2332. "epoch [1792] L = 38.469647, acc = 0.845000\n",
  2333. "epoch [1793] L = 38.469569, acc = 0.845000\n",
  2334. "epoch [1794] L = 38.469491, acc = 0.845000\n",
  2335. "epoch [1795] L = 38.469413, acc = 0.845000\n",
  2336. "epoch [1796] L = 38.469334, acc = 0.845000\n",
  2337. "epoch [1797] L = 38.469256, acc = 0.845000\n",
  2338. "epoch [1798] L = 38.469178, acc = 0.845000\n",
  2339. "epoch [1799] L = 38.469100, acc = 0.845000\n",
  2340. "epoch [1800] L = 38.469022, acc = 0.845000\n",
  2341. "epoch [1801] L = 38.468944, acc = 0.845000\n",
  2342. "epoch [1802] L = 38.468866, acc = 0.845000\n",
  2343. "epoch [1803] L = 38.468789, acc = 0.845000\n",
  2344. "epoch [1804] L = 38.468711, acc = 0.845000\n",
  2345. "epoch [1805] L = 38.468633, acc = 0.845000\n",
  2346. "epoch [1806] L = 38.468555, acc = 0.845000\n",
  2347. "epoch [1807] L = 38.468477, acc = 0.845000\n",
  2348. "epoch [1808] L = 38.468400, acc = 0.845000\n",
  2349. "epoch [1809] L = 38.468322, acc = 0.845000\n",
  2350. "epoch [1810] L = 38.468245, acc = 0.845000\n",
  2351. "epoch [1811] L = 38.468167, acc = 0.845000\n",
  2352. "epoch [1812] L = 38.468089, acc = 0.845000\n",
  2353. "epoch [1813] L = 38.468012, acc = 0.845000\n",
  2354. "epoch [1814] L = 38.467935, acc = 0.845000\n",
  2355. "epoch [1815] L = 38.467857, acc = 0.845000\n",
  2356. "epoch [1816] L = 38.467780, acc = 0.845000\n",
  2357. "epoch [1817] L = 38.467702, acc = 0.845000\n",
  2358. "epoch [1818] L = 38.467625, acc = 0.845000\n",
  2359. "epoch [1819] L = 38.467548, acc = 0.845000\n",
  2360. "epoch [1820] L = 38.467471, acc = 0.845000\n",
  2361. "epoch [1821] L = 38.467394, acc = 0.845000\n",
  2362. "epoch [1822] L = 38.467316, acc = 0.845000\n",
  2363. "epoch [1823] L = 38.467239, acc = 0.845000\n",
  2364. "epoch [1824] L = 38.467162, acc = 0.845000\n",
  2365. "epoch [1825] L = 38.467085, acc = 0.845000\n",
  2366. "epoch [1826] L = 38.467008, acc = 0.845000\n",
  2367. "epoch [1827] L = 38.466931, acc = 0.845000\n",
  2368. "epoch [1828] L = 38.466854, acc = 0.845000\n",
  2369. "epoch [1829] L = 38.466777, acc = 0.845000\n",
  2370. "epoch [1830] L = 38.466701, acc = 0.845000\n",
  2371. "epoch [1831] L = 38.466624, acc = 0.845000\n",
  2372. "epoch [1832] L = 38.466547, acc = 0.845000\n",
  2373. "epoch [1833] L = 38.466470, acc = 0.845000\n",
  2374. "epoch [1834] L = 38.466394, acc = 0.845000\n",
  2375. "epoch [1835] L = 38.466317, acc = 0.845000\n",
  2376. "epoch [1836] L = 38.466240, acc = 0.845000\n",
  2377. "epoch [1837] L = 38.466164, acc = 0.845000\n",
  2378. "epoch [1838] L = 38.466087, acc = 0.845000\n",
  2379. "epoch [1839] L = 38.466011, acc = 0.845000\n",
  2380. "epoch [1840] L = 38.465934, acc = 0.845000\n",
  2381. "epoch [1841] L = 38.465858, acc = 0.845000\n",
  2382. "epoch [1842] L = 38.465781, acc = 0.845000\n",
  2383. "epoch [1843] L = 38.465705, acc = 0.845000\n",
  2384. "epoch [1844] L = 38.465629, acc = 0.845000\n",
  2385. "epoch [1845] L = 38.465553, acc = 0.845000\n",
  2386. "epoch [1846] L = 38.465476, acc = 0.845000\n",
  2387. "epoch [1847] L = 38.465400, acc = 0.845000\n",
  2388. "epoch [1848] L = 38.465324, acc = 0.845000\n",
  2389. "epoch [1849] L = 38.465248, acc = 0.845000\n",
  2390. "epoch [1850] L = 38.465172, acc = 0.845000\n",
  2391. "epoch [1851] L = 38.465096, acc = 0.845000\n",
  2392. "epoch [1852] L = 38.465020, acc = 0.845000\n",
  2393. "epoch [1853] L = 38.464944, acc = 0.845000\n",
  2394. "epoch [1854] L = 38.464868, acc = 0.845000\n",
  2395. "epoch [1855] L = 38.464792, acc = 0.845000\n",
  2396. "epoch [1856] L = 38.464716, acc = 0.845000\n",
  2397. "epoch [1857] L = 38.464640, acc = 0.845000\n",
  2398. "epoch [1858] L = 38.464564, acc = 0.845000\n",
  2399. "epoch [1859] L = 38.464488, acc = 0.845000\n",
  2400. "epoch [1860] L = 38.464413, acc = 0.845000\n",
  2401. "epoch [1861] L = 38.464337, acc = 0.845000\n",
  2402. "epoch [1862] L = 38.464261, acc = 0.845000\n",
  2403. "epoch [1863] L = 38.464186, acc = 0.845000\n",
  2404. "epoch [1864] L = 38.464110, acc = 0.845000\n",
  2405. "epoch [1865] L = 38.464034, acc = 0.845000\n",
  2406. "epoch [1866] L = 38.463959, acc = 0.845000\n",
  2407. "epoch [1867] L = 38.463883, acc = 0.845000\n",
  2408. "epoch [1868] L = 38.463808, acc = 0.845000\n",
  2409. "epoch [1869] L = 38.463733, acc = 0.845000\n",
  2410. "epoch [1870] L = 38.463657, acc = 0.845000\n",
  2411. "epoch [1871] L = 38.463582, acc = 0.845000\n",
  2412. "epoch [1872] L = 38.463506, acc = 0.845000\n",
  2413. "epoch [1873] L = 38.463431, acc = 0.845000\n",
  2414. "epoch [1874] L = 38.463356, acc = 0.845000\n",
  2415. "epoch [1875] L = 38.463281, acc = 0.845000\n",
  2416. "epoch [1876] L = 38.463206, acc = 0.845000\n",
  2417. "epoch [1877] L = 38.463130, acc = 0.845000\n",
  2418. "epoch [1878] L = 38.463055, acc = 0.845000\n",
  2419. "epoch [1879] L = 38.462980, acc = 0.845000\n",
  2420. "epoch [1880] L = 38.462905, acc = 0.845000\n",
  2421. "epoch [1881] L = 38.462830, acc = 0.845000\n",
  2422. "epoch [1882] L = 38.462755, acc = 0.845000\n",
  2423. "epoch [1883] L = 38.462680, acc = 0.845000\n",
  2424. "epoch [1884] L = 38.462605, acc = 0.845000\n",
  2425. "epoch [1885] L = 38.462530, acc = 0.845000\n",
  2426. "epoch [1886] L = 38.462456, acc = 0.845000\n",
  2427. "epoch [1887] L = 38.462381, acc = 0.845000\n",
  2428. "epoch [1888] L = 38.462306, acc = 0.845000\n",
  2429. "epoch [1889] L = 38.462231, acc = 0.845000\n",
  2430. "epoch [1890] L = 38.462157, acc = 0.845000\n",
  2431. "epoch [1891] L = 38.462082, acc = 0.845000\n",
  2432. "epoch [1892] L = 38.462007, acc = 0.845000\n",
  2433. "epoch [1893] L = 38.461933, acc = 0.845000\n",
  2434. "epoch [1894] L = 38.461858, acc = 0.845000\n",
  2435. "epoch [1895] L = 38.461784, acc = 0.845000\n",
  2436. "epoch [1896] L = 38.461709, acc = 0.845000\n",
  2437. "epoch [1897] L = 38.461635, acc = 0.845000\n",
  2438. "epoch [1898] L = 38.461560, acc = 0.845000\n",
  2439. "epoch [1899] L = 38.461486, acc = 0.845000\n",
  2440. "epoch [1900] L = 38.461412, acc = 0.845000\n",
  2441. "epoch [1901] L = 38.461337, acc = 0.845000\n",
  2442. "epoch [1902] L = 38.461263, acc = 0.845000\n",
  2443. "epoch [1903] L = 38.461189, acc = 0.845000\n",
  2444. "epoch [1904] L = 38.461114, acc = 0.845000\n",
  2445. "epoch [1905] L = 38.461040, acc = 0.845000\n",
  2446. "epoch [1906] L = 38.460966, acc = 0.845000\n",
  2447. "epoch [1907] L = 38.460892, acc = 0.845000\n",
  2448. "epoch [1908] L = 38.460818, acc = 0.845000\n",
  2449. "epoch [1909] L = 38.460744, acc = 0.845000\n",
  2450. "epoch [1910] L = 38.460670, acc = 0.845000\n",
  2451. "epoch [1911] L = 38.460596, acc = 0.845000\n",
  2452. "epoch [1912] L = 38.460522, acc = 0.845000\n",
  2453. "epoch [1913] L = 38.460448, acc = 0.845000\n",
  2454. "epoch [1914] L = 38.460374, acc = 0.845000\n",
  2455. "epoch [1915] L = 38.460300, acc = 0.845000\n",
  2456. "epoch [1916] L = 38.460226, acc = 0.845000\n",
  2457. "epoch [1917] L = 38.460153, acc = 0.845000\n",
  2458. "epoch [1918] L = 38.460079, acc = 0.845000\n",
  2459. "epoch [1919] L = 38.460005, acc = 0.845000\n",
  2460. "epoch [1920] L = 38.459931, acc = 0.845000\n",
  2461. "epoch [1921] L = 38.459858, acc = 0.845000\n",
  2462. "epoch [1922] L = 38.459784, acc = 0.845000\n",
  2463. "epoch [1923] L = 38.459711, acc = 0.845000\n",
  2464. "epoch [1924] L = 38.459637, acc = 0.845000\n",
  2465. "epoch [1925] L = 38.459563, acc = 0.845000\n",
  2466. "epoch [1926] L = 38.459490, acc = 0.845000\n",
  2467. "epoch [1927] L = 38.459416, acc = 0.845000\n",
  2468. "epoch [1928] L = 38.459343, acc = 0.845000\n",
  2469. "epoch [1929] L = 38.459270, acc = 0.845000\n",
  2470. "epoch [1930] L = 38.459196, acc = 0.845000\n",
  2471. "epoch [1931] L = 38.459123, acc = 0.845000\n",
  2472. "epoch [1932] L = 38.459050, acc = 0.845000\n",
  2473. "epoch [1933] L = 38.458976, acc = 0.845000\n",
  2474. "epoch [1934] L = 38.458903, acc = 0.845000\n",
  2475. "epoch [1935] L = 38.458830, acc = 0.845000\n",
  2476. "epoch [1936] L = 38.458757, acc = 0.845000\n",
  2477. "epoch [1937] L = 38.458684, acc = 0.845000\n",
  2478. "epoch [1938] L = 38.458610, acc = 0.845000\n",
  2479. "epoch [1939] L = 38.458537, acc = 0.845000\n",
  2480. "epoch [1940] L = 38.458464, acc = 0.845000\n",
  2481. "epoch [1941] L = 38.458391, acc = 0.845000\n",
  2482. "epoch [1942] L = 38.458318, acc = 0.845000\n",
  2483. "epoch [1943] L = 38.458245, acc = 0.845000\n",
  2484. "epoch [1944] L = 38.458172, acc = 0.845000\n",
  2485. "epoch [1945] L = 38.458100, acc = 0.845000\n",
  2486. "epoch [1946] L = 38.458027, acc = 0.845000\n",
  2487. "epoch [1947] L = 38.457954, acc = 0.845000\n",
  2488. "epoch [1948] L = 38.457881, acc = 0.845000\n",
  2489. "epoch [1949] L = 38.457808, acc = 0.845000\n",
  2490. "epoch [1950] L = 38.457736, acc = 0.845000\n",
  2491. "epoch [1951] L = 38.457663, acc = 0.845000\n",
  2492. "epoch [1952] L = 38.457590, acc = 0.845000\n",
  2493. "epoch [1953] L = 38.457518, acc = 0.845000\n",
  2494. "epoch [1954] L = 38.457445, acc = 0.845000\n",
  2495. "epoch [1955] L = 38.457372, acc = 0.845000\n",
  2496. "epoch [1956] L = 38.457300, acc = 0.845000\n",
  2497. "epoch [1957] L = 38.457227, acc = 0.845000\n",
  2498. "epoch [1958] L = 38.457155, acc = 0.845000\n",
  2499. "epoch [1959] L = 38.457082, acc = 0.845000\n",
  2500. "epoch [1960] L = 38.457010, acc = 0.845000\n",
  2501. "epoch [1961] L = 38.456938, acc = 0.845000\n",
  2502. "epoch [1962] L = 38.456865, acc = 0.845000\n",
  2503. "epoch [1963] L = 38.456793, acc = 0.845000\n",
  2504. "epoch [1964] L = 38.456721, acc = 0.845000\n",
  2505. "epoch [1965] L = 38.456648, acc = 0.845000\n",
  2506. "epoch [1966] L = 38.456576, acc = 0.845000\n",
  2507. "epoch [1967] L = 38.456504, acc = 0.845000\n",
  2508. "epoch [1968] L = 38.456432, acc = 0.845000\n",
  2509. "epoch [1969] L = 38.456360, acc = 0.845000\n",
  2510. "epoch [1970] L = 38.456287, acc = 0.845000\n",
  2511. "epoch [1971] L = 38.456215, acc = 0.845000\n",
  2512. "epoch [1972] L = 38.456143, acc = 0.845000\n",
  2513. "epoch [1973] L = 38.456071, acc = 0.845000\n",
  2514. "epoch [1974] L = 38.455999, acc = 0.845000\n",
  2515. "epoch [1975] L = 38.455927, acc = 0.845000\n",
  2516. "epoch [1976] L = 38.455855, acc = 0.845000\n",
  2517. "epoch [1977] L = 38.455784, acc = 0.845000\n",
  2518. "epoch [1978] L = 38.455712, acc = 0.845000\n",
  2519. "epoch [1979] L = 38.455640, acc = 0.845000\n",
  2520. "epoch [1980] L = 38.455568, acc = 0.845000\n",
  2521. "epoch [1981] L = 38.455496, acc = 0.845000\n",
  2522. "epoch [1982] L = 38.455425, acc = 0.845000\n",
  2523. "epoch [1983] L = 38.455353, acc = 0.845000\n",
  2524. "epoch [1984] L = 38.455281, acc = 0.845000\n",
  2525. "epoch [1985] L = 38.455209, acc = 0.845000\n",
  2526. "epoch [1986] L = 38.455138, acc = 0.845000\n",
  2527. "epoch [1987] L = 38.455066, acc = 0.845000\n",
  2528. "epoch [1988] L = 38.454995, acc = 0.845000\n",
  2529. "epoch [1989] L = 38.454923, acc = 0.845000\n",
  2530. "epoch [1990] L = 38.454852, acc = 0.845000\n",
  2531. "epoch [1991] L = 38.454780, acc = 0.845000\n",
  2532. "epoch [1992] L = 38.454709, acc = 0.845000\n",
  2533. "epoch [1993] L = 38.454637, acc = 0.845000\n",
  2534. "epoch [1994] L = 38.454566, acc = 0.845000\n",
  2535. "epoch [1995] L = 38.454494, acc = 0.845000\n",
  2536. "epoch [1996] L = 38.454423, acc = 0.845000\n",
  2537. "epoch [1997] L = 38.454352, acc = 0.845000\n",
  2538. "epoch [1998] L = 38.454281, acc = 0.845000\n",
  2539. "epoch [1999] L = 38.454209, acc = 0.845000\n"
  2540. ]
  2541. }
  2542. ],
  2543. "source": [
  2544. "# FIXME: change variable name to math\n",
  2545. "\n",
  2546. "from sklearn.metrics import accuracy_score\n",
  2547. "\n",
  2548. "y_true = np.array(nn.y).astype(float)\n",
  2549. "\n",
  2550. "# back-propagation\n",
  2551. "def backpropagation(n, X, y):\n",
  2552. " for i in range(n.n_epoch):\n",
  2553. " # forward to calculate each node's output\n",
  2554. " forward(n, X)\n",
  2555. " \n",
  2556. " # print loss, accuracy\n",
  2557. " L = np.sum((n.z2 - y)**2)\n",
  2558. " \n",
  2559. " y_pred = np.argmax(nn.z2, axis=1)\n",
  2560. " acc = accuracy_score(y_true, y_pred)\n",
  2561. " \n",
  2562. " print(\"epoch [%4d] L = %f, acc = %f\" % (i, L, acc))\n",
  2563. " \n",
  2564. " # calc weights update\n",
  2565. " d2 = n.z2*(1-n.z2)*(y - n.z2)\n",
  2566. " d1 = n.z1*(1-n.z1)*(np.dot(d2, n.W2.T))\n",
  2567. " \n",
  2568. " # update weights\n",
  2569. " n.W2 += n.epsilon * np.dot(n.z1.T, d2)\n",
  2570. " n.b2 += n.epsilon * np.sum(d2, axis=0)\n",
  2571. " n.W1 += n.epsilon * np.dot(X.T, d1)\n",
  2572. " n.b1 += n.epsilon * np.sum(d1, axis=0)\n",
  2573. "\n",
  2574. "nn.n_epoch = 2000\n",
  2575. "backpropagation(nn, X, t)\n"
  2576. ]
  2577. },
  2578. {
  2579. "cell_type": "code",
  2580. "execution_count": 7,
  2581. "metadata": {},
  2582. "outputs": [
  2583. {
  2584. "data": {
  2585. "image/png": "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\n",
  2586. "text/plain": [
  2587. "<Figure size 432x288 with 1 Axes>"
  2588. ]
  2589. },
  2590. "metadata": {
  2591. "needs_background": "light"
  2592. },
  2593. "output_type": "display_data"
  2594. },
  2595. {
  2596. "data": {
  2597. "image/png": "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\n",
  2598. "text/plain": [
  2599. "<Figure size 432x288 with 1 Axes>"
  2600. ]
  2601. },
  2602. "metadata": {
  2603. "needs_background": "light"
  2604. },
  2605. "output_type": "display_data"
  2606. }
  2607. ],
  2608. "source": [
  2609. "# plot data\n",
  2610. "y_pred = np.argmax(nn.z2, axis=1)\n",
  2611. "\n",
  2612. "plt.scatter(X[:, 0], X[:, 1], c=nn.y, cmap=plt.cm.Spectral)\n",
  2613. "plt.title(\"ground truth\")\n",
  2614. "plt.show()\n",
  2615. "\n",
  2616. "plt.scatter(X[:, 0], X[:, 1], c=y_pred, cmap=plt.cm.Spectral)\n",
  2617. "plt.title(\"predicted\")\n",
  2618. "plt.show()\n"
  2619. ]
  2620. },
  2621. {
  2622. "cell_type": "markdown",
  2623. "metadata": {},
  2624. "source": [
  2625. "## 8. 如何使用类的方法封装多层神经网络?"
  2626. ]
  2627. },
  2628. {
  2629. "cell_type": "code",
  2630. "execution_count": 4,
  2631. "metadata": {},
  2632. "outputs": [],
  2633. "source": [
  2634. "import numpy as np\n",
  2635. "from sklearn import datasets, linear_model\n",
  2636. "from sklearn.metrics import accuracy_score\n",
  2637. "import matplotlib.pyplot as plt\n",
  2638. "\n",
  2639. "\n",
  2640. "# define sigmod\n",
  2641. "def sigmod(X):\n",
  2642. " return 1.0/(1+np.exp(-X))\n",
  2643. "\n",
  2644. "\n",
  2645. "# generate the NN model\n",
  2646. "class NN_Model:\n",
  2647. " def __init__(self, nodes=None):\n",
  2648. " self.epsilon = 0.01 # learning rate\n",
  2649. " self.n_epoch = 1000 # iterative number\n",
  2650. " \n",
  2651. " if not nodes:\n",
  2652. " self.nodes = [2, 6, 2] # default nodes size (from input -> output)\n",
  2653. " else:\n",
  2654. " self.nodes = nodes\n",
  2655. " \n",
  2656. " def init_weight(self):\n",
  2657. " W = []\n",
  2658. " B = []\n",
  2659. " \n",
  2660. " n_layer = len(self.nodes)\n",
  2661. " for i in range(n_layer-1):\n",
  2662. " w = np.random.randn(self.nodes[i], self.nodes[i+1]) / np.sqrt(self.nodes[i])\n",
  2663. " b = np.random.randn(1, self.nodes[i+1])\n",
  2664. " \n",
  2665. " W.append(w)\n",
  2666. " B.append(b)\n",
  2667. " \n",
  2668. " self.W = W\n",
  2669. " self.B = B\n",
  2670. " \n",
  2671. " def forward(self, X):\n",
  2672. " Z = []\n",
  2673. " x0 = X\n",
  2674. " for i in range(len(self.nodes)-1):\n",
  2675. " z = sigmod(np.dot(x0, self.W[i]) + self.B[i])\n",
  2676. " x0 = z\n",
  2677. " \n",
  2678. " Z.append(z)\n",
  2679. " \n",
  2680. " self.Z = Z\n",
  2681. " return Z[-1]\n",
  2682. " \n",
  2683. " # back-propagation\n",
  2684. " def backpropagation(self, X, y, n_epoch=None, epsilon=None):\n",
  2685. " if not n_epoch: n_epoch = self.n_epoch\n",
  2686. " if not epsilon: epsilon = self.epsilon\n",
  2687. " \n",
  2688. " self.X = X\n",
  2689. " self.Y = y\n",
  2690. " \n",
  2691. " for i in range(n_epoch):\n",
  2692. " # forward to calculate each node's output\n",
  2693. " self.forward(X)\n",
  2694. "\n",
  2695. " self.evaluate()\n",
  2696. " \n",
  2697. " # calc weights update\n",
  2698. " W = self.W\n",
  2699. " B = self.B\n",
  2700. " Z = self.Z\n",
  2701. " \n",
  2702. " D = []\n",
  2703. " d0 = y\n",
  2704. " n_layer = len(self.nodes)\n",
  2705. " for j in range(n_layer-1, 0, -1):\n",
  2706. " jj = j - 1\n",
  2707. " z = self.Z[jj]\n",
  2708. " \n",
  2709. " if j == n_layer - 1:\n",
  2710. " d = z*(1-z)*(d0 - z)\n",
  2711. " else:\n",
  2712. " d = z*(1-z)*np.dot(d0, W[j].T)\n",
  2713. " \n",
  2714. " d0 = d\n",
  2715. " D.insert(0, d)\n",
  2716. " \n",
  2717. " # update weights\n",
  2718. " for j in range(n_layer-1, 0, -1):\n",
  2719. " jj = j - 1\n",
  2720. " \n",
  2721. " if jj != 0:\n",
  2722. " W[jj] += epsilon * np.dot(Z[jj-1].T, D[jj])\n",
  2723. " else:\n",
  2724. " W[jj] += epsilon * np.dot(X.T, D[jj])\n",
  2725. " \n",
  2726. " B[jj] += epsilon * np.sum(D[jj], axis=0)\n",
  2727. " \n",
  2728. " def evaluate(self):\n",
  2729. " z = self.Z[-1]\n",
  2730. " \n",
  2731. " # print loss, accuracy\n",
  2732. " L = np.sum((z - self.Y)**2)\n",
  2733. " \n",
  2734. " y_pred = np.argmax(z, axis=1)\n",
  2735. " y_true = np.argmax(self.Y, axis=1)\n",
  2736. " acc = accuracy_score(y_true, y_pred)\n",
  2737. " \n",
  2738. " print(\"L = %f, acc = %f\" % (L, acc))\n",
  2739. " "
  2740. ]
  2741. },
  2742. {
  2743. "cell_type": "code",
  2744. "execution_count": 5,
  2745. "metadata": {},
  2746. "outputs": [
  2747. {
  2748. "data": {
  2749. "image/png": "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\n",
  2750. "text/plain": [
  2751. "<Figure size 432x288 with 1 Axes>"
  2752. ]
  2753. },
  2754. "metadata": {
  2755. "needs_background": "light"
  2756. },
  2757. "output_type": "display_data"
  2758. }
  2759. ],
  2760. "source": [
  2761. "# generate sample data\n",
  2762. "np.random.seed(0)\n",
  2763. "X, y = datasets.make_moons(200, noise=0.20)\n",
  2764. "\n",
  2765. "# generate nn output target\n",
  2766. "t = np.zeros((X.shape[0], 2))\n",
  2767. "t[np.where(y==0), 0] = 1\n",
  2768. "t[np.where(y==1), 1] = 1\n",
  2769. "\n",
  2770. "# plot data\n",
  2771. "plt.scatter(X[:, 0], X[:, 1], c=y, cmap=plt.cm.Spectral)\n",
  2772. "plt.show()"
  2773. ]
  2774. },
  2775. {
  2776. "cell_type": "code",
  2777. "execution_count": 6,
  2778. "metadata": {},
  2779. "outputs": [
  2780. {
  2781. "name": "stdout",
  2782. "output_type": "stream",
  2783. "text": [
  2784. "L = 121.621107, acc = 0.500000\n",
  2785. "L = 115.928422, acc = 0.500000\n",
  2786. "L = 111.304997, acc = 0.500000\n",
  2787. "L = 107.789222, acc = 0.500000\n",
  2788. "L = 105.265297, acc = 0.500000\n",
  2789. "L = 103.533617, acc = 0.500000\n",
  2790. "L = 102.380546, acc = 0.500000\n",
  2791. "L = 101.622557, acc = 0.500000\n",
  2792. "L = 101.121698, acc = 0.500000\n",
  2793. "L = 100.782803, acc = 0.510000\n",
  2794. "L = 100.543751, acc = 0.530000\n",
  2795. "L = 100.365372, acc = 0.540000\n",
  2796. "L = 100.223492, acc = 0.520000\n",
  2797. "L = 100.103371, acc = 0.475000\n",
  2798. "L = 99.996073, acc = 0.460000\n",
  2799. "L = 99.896185, acc = 0.465000\n",
  2800. "L = 99.800411, acc = 0.465000\n",
  2801. "L = 99.706725, acc = 0.495000\n",
  2802. "L = 99.613854, acc = 0.515000\n",
  2803. "L = 99.520981, acc = 0.560000\n",
  2804. "L = 99.427551, acc = 0.585000\n",
  2805. "L = 99.333171, acc = 0.630000\n",
  2806. "L = 99.237541, acc = 0.660000\n",
  2807. "L = 99.140415, acc = 0.690000\n",
  2808. "L = 99.041582, acc = 0.705000\n",
  2809. "L = 98.940844, acc = 0.710000\n",
  2810. "L = 98.838015, acc = 0.720000\n",
  2811. "L = 98.732913, acc = 0.740000\n",
  2812. "L = 98.625357, acc = 0.745000\n",
  2813. "L = 98.515164, acc = 0.755000\n",
  2814. "L = 98.402148, acc = 0.785000\n",
  2815. "L = 98.286120, acc = 0.790000\n",
  2816. "L = 98.166887, acc = 0.800000\n",
  2817. "L = 98.044250, acc = 0.800000\n",
  2818. "L = 97.918005, acc = 0.805000\n",
  2819. "L = 97.787942, acc = 0.815000\n",
  2820. "L = 97.653845, acc = 0.830000\n",
  2821. "L = 97.515489, acc = 0.830000\n",
  2822. "L = 97.372644, acc = 0.830000\n",
  2823. "L = 97.225071, acc = 0.830000\n",
  2824. "L = 97.072523, acc = 0.830000\n",
  2825. "L = 96.914745, acc = 0.835000\n",
  2826. "L = 96.751472, acc = 0.835000\n",
  2827. "L = 96.582430, acc = 0.835000\n",
  2828. "L = 96.407335, acc = 0.835000\n",
  2829. "L = 96.225894, acc = 0.835000\n",
  2830. "L = 96.037800, acc = 0.835000\n",
  2831. "L = 95.842740, acc = 0.835000\n",
  2832. "L = 95.640384, acc = 0.835000\n",
  2833. "L = 95.430396, acc = 0.835000\n",
  2834. "L = 95.212423, acc = 0.835000\n",
  2835. "L = 94.986104, acc = 0.830000\n",
  2836. "L = 94.751064, acc = 0.830000\n",
  2837. "L = 94.506915, acc = 0.830000\n",
  2838. "L = 94.253259, acc = 0.830000\n",
  2839. "L = 93.989683, acc = 0.830000\n",
  2840. "L = 93.715765, acc = 0.830000\n",
  2841. "L = 93.431069, acc = 0.830000\n",
  2842. "L = 93.135151, acc = 0.830000\n",
  2843. "L = 92.827554, acc = 0.830000\n",
  2844. "L = 92.507814, acc = 0.830000\n",
  2845. "L = 92.175457, acc = 0.830000\n",
  2846. "L = 91.830004, acc = 0.835000\n",
  2847. "L = 91.470973, acc = 0.835000\n",
  2848. "L = 91.097875, acc = 0.835000\n",
  2849. "L = 90.710225, acc = 0.840000\n",
  2850. "L = 90.307539, acc = 0.845000\n",
  2851. "L = 89.889339, acc = 0.845000\n",
  2852. "L = 89.455160, acc = 0.845000\n",
  2853. "L = 89.004546, acc = 0.840000\n",
  2854. "L = 88.537066, acc = 0.840000\n",
  2855. "L = 88.052308, acc = 0.840000\n",
  2856. "L = 87.549895, acc = 0.840000\n",
  2857. "L = 87.029483, acc = 0.845000\n",
  2858. "L = 86.490773, acc = 0.845000\n",
  2859. "L = 85.933518, acc = 0.845000\n",
  2860. "L = 85.357526, acc = 0.845000\n",
  2861. "L = 84.762674, acc = 0.845000\n",
  2862. "L = 84.148911, acc = 0.845000\n",
  2863. "L = 83.516272, acc = 0.845000\n",
  2864. "L = 82.864878, acc = 0.845000\n",
  2865. "L = 82.194952, acc = 0.845000\n",
  2866. "L = 81.506820, acc = 0.840000\n",
  2867. "L = 80.800921, acc = 0.840000\n",
  2868. "L = 80.077810, acc = 0.840000\n",
  2869. "L = 79.338167, acc = 0.840000\n",
  2870. "L = 78.582791, acc = 0.840000\n",
  2871. "L = 77.812612, acc = 0.840000\n",
  2872. "L = 77.028680, acc = 0.840000\n",
  2873. "L = 76.232171, acc = 0.840000\n",
  2874. "L = 75.424374, acc = 0.840000\n",
  2875. "L = 74.606691, acc = 0.840000\n",
  2876. "L = 73.780620, acc = 0.840000\n",
  2877. "L = 72.947751, acc = 0.840000\n",
  2878. "L = 72.109745, acc = 0.840000\n",
  2879. "L = 71.268324, acc = 0.840000\n",
  2880. "L = 70.425252, acc = 0.840000\n",
  2881. "L = 69.582316, acc = 0.840000\n",
  2882. "L = 68.741307, acc = 0.840000\n",
  2883. "L = 67.904004, acc = 0.840000\n",
  2884. "L = 67.072151, acc = 0.840000\n",
  2885. "L = 66.247442, acc = 0.840000\n",
  2886. "L = 65.431502, acc = 0.840000\n",
  2887. "L = 64.625872, acc = 0.840000\n",
  2888. "L = 63.831996, acc = 0.840000\n",
  2889. "L = 63.051206, acc = 0.840000\n",
  2890. "L = 62.284717, acc = 0.840000\n",
  2891. "L = 61.533617, acc = 0.840000\n",
  2892. "L = 60.798864, acc = 0.840000\n",
  2893. "L = 60.081280, acc = 0.840000\n",
  2894. "L = 59.381556, acc = 0.840000\n",
  2895. "L = 58.700250, acc = 0.840000\n",
  2896. "L = 58.037794, acc = 0.840000\n",
  2897. "L = 57.394496, acc = 0.840000\n",
  2898. "L = 56.770551, acc = 0.840000\n",
  2899. "L = 56.166043, acc = 0.840000\n",
  2900. "L = 55.580959, acc = 0.840000\n",
  2901. "L = 55.015197, acc = 0.840000\n",
  2902. "L = 54.468573, acc = 0.840000\n",
  2903. "L = 53.940833, acc = 0.840000\n",
  2904. "L = 53.431659, acc = 0.840000\n",
  2905. "L = 52.940684, acc = 0.840000\n",
  2906. "L = 52.467494, acc = 0.840000\n",
  2907. "L = 52.011639, acc = 0.840000\n",
  2908. "L = 51.572642, acc = 0.840000\n",
  2909. "L = 51.150004, acc = 0.840000\n",
  2910. "L = 50.743209, acc = 0.840000\n",
  2911. "L = 50.351731, acc = 0.840000\n",
  2912. "L = 49.975042, acc = 0.840000\n",
  2913. "L = 49.612610, acc = 0.835000\n",
  2914. "L = 49.263906, acc = 0.835000\n",
  2915. "L = 48.928410, acc = 0.840000\n",
  2916. "L = 48.605606, acc = 0.840000\n",
  2917. "L = 48.294993, acc = 0.840000\n",
  2918. "L = 47.996079, acc = 0.840000\n",
  2919. "L = 47.708390, acc = 0.840000\n",
  2920. "L = 47.431462, acc = 0.840000\n",
  2921. "L = 47.164849, acc = 0.840000\n",
  2922. "L = 46.908123, acc = 0.840000\n",
  2923. "L = 46.660868, acc = 0.840000\n",
  2924. "L = 46.422687, acc = 0.840000\n",
  2925. "L = 46.193200, acc = 0.840000\n",
  2926. "L = 45.972040, acc = 0.840000\n",
  2927. "L = 45.758860, acc = 0.840000\n",
  2928. "L = 45.553325, acc = 0.840000\n",
  2929. "L = 45.355116, acc = 0.840000\n",
  2930. "L = 45.163929, acc = 0.835000\n",
  2931. "L = 44.979474, acc = 0.835000\n",
  2932. "L = 44.801473, acc = 0.835000\n",
  2933. "L = 44.629662, acc = 0.835000\n",
  2934. "L = 44.463789, acc = 0.835000\n",
  2935. "L = 44.303614, acc = 0.835000\n",
  2936. "L = 44.148907, acc = 0.835000\n",
  2937. "L = 43.999451, acc = 0.835000\n",
  2938. "L = 43.855036, acc = 0.835000\n",
  2939. "L = 43.715465, acc = 0.835000\n",
  2940. "L = 43.580546, acc = 0.835000\n",
  2941. "L = 43.450099, acc = 0.835000\n",
  2942. "L = 43.323950, acc = 0.835000\n",
  2943. "L = 43.201935, acc = 0.835000\n",
  2944. "L = 43.083894, acc = 0.835000\n",
  2945. "L = 42.969678, acc = 0.835000\n",
  2946. "L = 42.859141, acc = 0.835000\n",
  2947. "L = 42.752145, acc = 0.835000\n",
  2948. "L = 42.648557, acc = 0.835000\n",
  2949. "L = 42.548251, acc = 0.835000\n",
  2950. "L = 42.451106, acc = 0.835000\n",
  2951. "L = 42.357004, acc = 0.835000\n",
  2952. "L = 42.265834, acc = 0.835000\n",
  2953. "L = 42.177489, acc = 0.835000\n",
  2954. "L = 42.091866, acc = 0.845000\n",
  2955. "L = 42.008866, acc = 0.845000\n",
  2956. "L = 41.928395, acc = 0.845000\n",
  2957. "L = 41.850363, acc = 0.845000\n",
  2958. "L = 41.774680, acc = 0.845000\n",
  2959. "L = 41.701264, acc = 0.845000\n",
  2960. "L = 41.630034, acc = 0.845000\n",
  2961. "L = 41.560912, acc = 0.845000\n",
  2962. "L = 41.493823, acc = 0.845000\n",
  2963. "L = 41.428697, acc = 0.845000\n",
  2964. "L = 41.365463, acc = 0.845000\n",
  2965. "L = 41.304056, acc = 0.850000\n",
  2966. "L = 41.244412, acc = 0.850000\n",
  2967. "L = 41.186469, acc = 0.850000\n",
  2968. "L = 41.130168, acc = 0.850000\n",
  2969. "L = 41.075452, acc = 0.850000\n",
  2970. "L = 41.022266, acc = 0.850000\n",
  2971. "L = 40.970558, acc = 0.850000\n",
  2972. "L = 40.920276, acc = 0.850000\n",
  2973. "L = 40.871372, acc = 0.850000\n",
  2974. "L = 40.823798, acc = 0.850000\n",
  2975. "L = 40.777509, acc = 0.850000\n",
  2976. "L = 40.732461, acc = 0.855000\n",
  2977. "L = 40.688613, acc = 0.855000\n",
  2978. "L = 40.645922, acc = 0.855000\n",
  2979. "L = 40.604351, acc = 0.855000\n",
  2980. "L = 40.563861, acc = 0.855000\n",
  2981. "L = 40.524415, acc = 0.855000\n",
  2982. "L = 40.485980, acc = 0.855000\n",
  2983. "L = 40.448521, acc = 0.855000\n",
  2984. "L = 40.412004, acc = 0.855000\n",
  2985. "L = 40.376400, acc = 0.855000\n",
  2986. "L = 40.341678, acc = 0.855000\n",
  2987. "L = 40.307807, acc = 0.855000\n",
  2988. "L = 40.274761, acc = 0.855000\n",
  2989. "L = 40.242511, acc = 0.855000\n",
  2990. "L = 40.211032, acc = 0.855000\n",
  2991. "L = 40.180297, acc = 0.855000\n",
  2992. "L = 40.150284, acc = 0.855000\n",
  2993. "L = 40.120967, acc = 0.855000\n",
  2994. "L = 40.092325, acc = 0.855000\n",
  2995. "L = 40.064334, acc = 0.855000\n",
  2996. "L = 40.036975, acc = 0.855000\n",
  2997. "L = 40.010226, acc = 0.855000\n",
  2998. "L = 39.984068, acc = 0.855000\n",
  2999. "L = 39.958481, acc = 0.855000\n",
  3000. "L = 39.933446, acc = 0.855000\n",
  3001. "L = 39.908947, acc = 0.855000\n",
  3002. "L = 39.884966, acc = 0.855000\n",
  3003. "L = 39.861486, acc = 0.855000\n",
  3004. "L = 39.838490, acc = 0.855000\n",
  3005. "L = 39.815964, acc = 0.855000\n",
  3006. "L = 39.793892, acc = 0.855000\n",
  3007. "L = 39.772260, acc = 0.855000\n",
  3008. "L = 39.751053, acc = 0.855000\n",
  3009. "L = 39.730259, acc = 0.855000\n",
  3010. "L = 39.709863, acc = 0.855000\n",
  3011. "L = 39.689852, acc = 0.855000\n",
  3012. "L = 39.670216, acc = 0.855000\n",
  3013. "L = 39.650941, acc = 0.855000\n",
  3014. "L = 39.632017, acc = 0.855000\n",
  3015. "L = 39.613431, acc = 0.855000\n",
  3016. "L = 39.595173, acc = 0.855000\n",
  3017. "L = 39.577233, acc = 0.855000\n",
  3018. "L = 39.559600, acc = 0.855000\n",
  3019. "L = 39.542265, acc = 0.855000\n",
  3020. "L = 39.525218, acc = 0.855000\n",
  3021. "L = 39.508449, acc = 0.855000\n",
  3022. "L = 39.491950, acc = 0.855000\n",
  3023. "L = 39.475713, acc = 0.855000\n",
  3024. "L = 39.459727, acc = 0.855000\n",
  3025. "L = 39.443987, acc = 0.855000\n",
  3026. "L = 39.428483, acc = 0.855000\n",
  3027. "L = 39.413208, acc = 0.855000\n",
  3028. "L = 39.398154, acc = 0.855000\n",
  3029. "L = 39.383314, acc = 0.855000\n",
  3030. "L = 39.368682, acc = 0.855000\n",
  3031. "L = 39.354250, acc = 0.855000\n",
  3032. "L = 39.340012, acc = 0.855000\n",
  3033. "L = 39.325961, acc = 0.860000\n",
  3034. "L = 39.312091, acc = 0.860000\n",
  3035. "L = 39.298397, acc = 0.860000\n",
  3036. "L = 39.284872, acc = 0.860000\n",
  3037. "L = 39.271510, acc = 0.860000\n",
  3038. "L = 39.258306, acc = 0.860000\n",
  3039. "L = 39.245255, acc = 0.860000\n",
  3040. "L = 39.232351, acc = 0.860000\n",
  3041. "L = 39.219590, acc = 0.860000\n",
  3042. "L = 39.206966, acc = 0.860000\n",
  3043. "L = 39.194474, acc = 0.860000\n",
  3044. "L = 39.182111, acc = 0.860000\n",
  3045. "L = 39.169870, acc = 0.860000\n",
  3046. "L = 39.157749, acc = 0.860000\n",
  3047. "L = 39.145742, acc = 0.860000\n",
  3048. "L = 39.133846, acc = 0.850000\n",
  3049. "L = 39.122056, acc = 0.850000\n",
  3050. "L = 39.110369, acc = 0.850000\n",
  3051. "L = 39.098780, acc = 0.850000\n",
  3052. "L = 39.087286, acc = 0.850000\n",
  3053. "L = 39.075884, acc = 0.850000\n",
  3054. "L = 39.064569, acc = 0.850000\n",
  3055. "L = 39.053338, acc = 0.850000\n",
  3056. "L = 39.042188, acc = 0.850000\n",
  3057. "L = 39.031116, acc = 0.850000\n",
  3058. "L = 39.020118, acc = 0.850000\n",
  3059. "L = 39.009191, acc = 0.850000\n",
  3060. "L = 38.998332, acc = 0.850000\n",
  3061. "L = 38.987539, acc = 0.850000\n",
  3062. "L = 38.976808, acc = 0.850000\n",
  3063. "L = 38.966136, acc = 0.850000\n",
  3064. "L = 38.955522, acc = 0.850000\n",
  3065. "L = 38.944961, acc = 0.850000\n",
  3066. "L = 38.934453, acc = 0.850000\n",
  3067. "L = 38.923993, acc = 0.855000\n",
  3068. "L = 38.913579, acc = 0.855000\n",
  3069. "L = 38.903210, acc = 0.855000\n",
  3070. "L = 38.892883, acc = 0.855000\n",
  3071. "L = 38.882595, acc = 0.855000\n",
  3072. "L = 38.872344, acc = 0.855000\n",
  3073. "L = 38.862129, acc = 0.855000\n",
  3074. "L = 38.851946, acc = 0.855000\n",
  3075. "L = 38.841794, acc = 0.855000\n",
  3076. "L = 38.831671, acc = 0.855000\n",
  3077. "L = 38.821574, acc = 0.855000\n",
  3078. "L = 38.811503, acc = 0.855000\n",
  3079. "L = 38.801454, acc = 0.855000\n",
  3080. "L = 38.791426, acc = 0.855000\n",
  3081. "L = 38.781418, acc = 0.855000\n",
  3082. "L = 38.771427, acc = 0.855000\n",
  3083. "L = 38.761452, acc = 0.855000\n",
  3084. "L = 38.751491, acc = 0.855000\n",
  3085. "L = 38.741542, acc = 0.855000\n",
  3086. "L = 38.731604, acc = 0.855000\n",
  3087. "L = 38.721676, acc = 0.855000\n",
  3088. "L = 38.711755, acc = 0.855000\n",
  3089. "L = 38.701840, acc = 0.855000\n",
  3090. "L = 38.691929, acc = 0.855000\n",
  3091. "L = 38.682022, acc = 0.855000\n",
  3092. "L = 38.672117, acc = 0.855000\n",
  3093. "L = 38.662212, acc = 0.855000\n",
  3094. "L = 38.652306, acc = 0.855000\n",
  3095. "L = 38.642397, acc = 0.855000\n",
  3096. "L = 38.632485, acc = 0.855000\n",
  3097. "L = 38.622568, acc = 0.855000\n",
  3098. "L = 38.612645, acc = 0.855000\n",
  3099. "L = 38.602715, acc = 0.855000\n",
  3100. "L = 38.592775, acc = 0.855000\n",
  3101. "L = 38.582826, acc = 0.855000\n",
  3102. "L = 38.572866, acc = 0.855000\n",
  3103. "L = 38.562894, acc = 0.855000\n",
  3104. "L = 38.552908, acc = 0.855000\n",
  3105. "L = 38.542908, acc = 0.855000\n",
  3106. "L = 38.532892, acc = 0.855000\n",
  3107. "L = 38.522860, acc = 0.855000\n",
  3108. "L = 38.512811, acc = 0.855000\n",
  3109. "L = 38.502742, acc = 0.855000\n",
  3110. "L = 38.492655, acc = 0.855000\n",
  3111. "L = 38.482546, acc = 0.855000\n",
  3112. "L = 38.472416, acc = 0.855000\n",
  3113. "L = 38.462263, acc = 0.855000\n",
  3114. "L = 38.452087, acc = 0.855000\n",
  3115. "L = 38.441886, acc = 0.855000\n",
  3116. "L = 38.431660, acc = 0.855000\n",
  3117. "L = 38.421407, acc = 0.855000\n",
  3118. "L = 38.411128, acc = 0.855000\n",
  3119. "L = 38.400820, acc = 0.855000\n",
  3120. "L = 38.390483, acc = 0.855000\n",
  3121. "L = 38.380116, acc = 0.855000\n",
  3122. "L = 38.369719, acc = 0.855000\n",
  3123. "L = 38.359290, acc = 0.855000\n",
  3124. "L = 38.348829, acc = 0.855000\n",
  3125. "L = 38.338334, acc = 0.855000\n",
  3126. "L = 38.327806, acc = 0.855000\n",
  3127. "L = 38.317242, acc = 0.855000\n",
  3128. "L = 38.306643, acc = 0.855000\n",
  3129. "L = 38.296008, acc = 0.855000\n",
  3130. "L = 38.285335, acc = 0.855000\n",
  3131. "L = 38.274625, acc = 0.855000\n",
  3132. "L = 38.263875, acc = 0.855000\n",
  3133. "L = 38.253086, acc = 0.855000\n",
  3134. "L = 38.242257, acc = 0.855000\n",
  3135. "L = 38.231387, acc = 0.855000\n",
  3136. "L = 38.220475, acc = 0.855000\n",
  3137. "L = 38.209520, acc = 0.855000\n",
  3138. "L = 38.198523, acc = 0.855000\n",
  3139. "L = 38.187481, acc = 0.855000\n",
  3140. "L = 38.176394, acc = 0.855000\n",
  3141. "L = 38.165262, acc = 0.855000\n",
  3142. "L = 38.154084, acc = 0.855000\n"
  3143. ]
  3144. },
  3145. {
  3146. "name": "stdout",
  3147. "output_type": "stream",
  3148. "text": [
  3149. "L = 38.142859, acc = 0.855000\n",
  3150. "L = 38.131586, acc = 0.855000\n",
  3151. "L = 38.120265, acc = 0.855000\n",
  3152. "L = 38.108895, acc = 0.855000\n",
  3153. "L = 38.097475, acc = 0.855000\n",
  3154. "L = 38.086004, acc = 0.855000\n",
  3155. "L = 38.074483, acc = 0.855000\n",
  3156. "L = 38.062909, acc = 0.855000\n",
  3157. "L = 38.051283, acc = 0.855000\n",
  3158. "L = 38.039603, acc = 0.855000\n",
  3159. "L = 38.027870, acc = 0.855000\n",
  3160. "L = 38.016082, acc = 0.855000\n",
  3161. "L = 38.004238, acc = 0.855000\n",
  3162. "L = 37.992338, acc = 0.860000\n",
  3163. "L = 37.980381, acc = 0.860000\n",
  3164. "L = 37.968367, acc = 0.860000\n",
  3165. "L = 37.956295, acc = 0.860000\n",
  3166. "L = 37.944163, acc = 0.860000\n",
  3167. "L = 37.931972, acc = 0.860000\n",
  3168. "L = 37.919720, acc = 0.860000\n",
  3169. "L = 37.907408, acc = 0.860000\n",
  3170. "L = 37.895033, acc = 0.860000\n",
  3171. "L = 37.882596, acc = 0.860000\n",
  3172. "L = 37.870096, acc = 0.860000\n",
  3173. "L = 37.857532, acc = 0.860000\n",
  3174. "L = 37.844903, acc = 0.860000\n",
  3175. "L = 37.832209, acc = 0.860000\n",
  3176. "L = 37.819448, acc = 0.860000\n",
  3177. "L = 37.806621, acc = 0.860000\n",
  3178. "L = 37.793727, acc = 0.860000\n",
  3179. "L = 37.780763, acc = 0.860000\n",
  3180. "L = 37.767731, acc = 0.860000\n",
  3181. "L = 37.754630, acc = 0.860000\n",
  3182. "L = 37.741457, acc = 0.860000\n",
  3183. "L = 37.728214, acc = 0.865000\n",
  3184. "L = 37.714898, acc = 0.865000\n",
  3185. "L = 37.701510, acc = 0.865000\n",
  3186. "L = 37.688048, acc = 0.865000\n",
  3187. "L = 37.674512, acc = 0.865000\n",
  3188. "L = 37.660902, acc = 0.865000\n",
  3189. "L = 37.647215, acc = 0.865000\n",
  3190. "L = 37.633452, acc = 0.865000\n",
  3191. "L = 37.619612, acc = 0.865000\n",
  3192. "L = 37.605693, acc = 0.865000\n",
  3193. "L = 37.591696, acc = 0.865000\n",
  3194. "L = 37.577620, acc = 0.865000\n",
  3195. "L = 37.563463, acc = 0.865000\n",
  3196. "L = 37.549225, acc = 0.870000\n",
  3197. "L = 37.534905, acc = 0.865000\n",
  3198. "L = 37.520503, acc = 0.865000\n",
  3199. "L = 37.506017, acc = 0.865000\n",
  3200. "L = 37.491447, acc = 0.865000\n",
  3201. "L = 37.476792, acc = 0.865000\n",
  3202. "L = 37.462052, acc = 0.865000\n",
  3203. "L = 37.447224, acc = 0.865000\n",
  3204. "L = 37.432310, acc = 0.865000\n",
  3205. "L = 37.417307, acc = 0.865000\n",
  3206. "L = 37.402215, acc = 0.865000\n",
  3207. "L = 37.387033, acc = 0.865000\n",
  3208. "L = 37.371761, acc = 0.865000\n",
  3209. "L = 37.356398, acc = 0.865000\n",
  3210. "L = 37.340942, acc = 0.865000\n",
  3211. "L = 37.325393, acc = 0.865000\n",
  3212. "L = 37.309751, acc = 0.865000\n",
  3213. "L = 37.294013, acc = 0.865000\n",
  3214. "L = 37.278181, acc = 0.865000\n",
  3215. "L = 37.262252, acc = 0.865000\n",
  3216. "L = 37.246226, acc = 0.865000\n",
  3217. "L = 37.230101, acc = 0.865000\n",
  3218. "L = 37.213879, acc = 0.865000\n",
  3219. "L = 37.197556, acc = 0.865000\n",
  3220. "L = 37.181133, acc = 0.865000\n",
  3221. "L = 37.164609, acc = 0.865000\n",
  3222. "L = 37.147983, acc = 0.865000\n",
  3223. "L = 37.131254, acc = 0.865000\n",
  3224. "L = 37.114421, acc = 0.865000\n",
  3225. "L = 37.097483, acc = 0.865000\n",
  3226. "L = 37.080440, acc = 0.865000\n",
  3227. "L = 37.063291, acc = 0.865000\n",
  3228. "L = 37.046035, acc = 0.865000\n",
  3229. "L = 37.028670, acc = 0.865000\n",
  3230. "L = 37.011197, acc = 0.865000\n",
  3231. "L = 36.993614, acc = 0.865000\n",
  3232. "L = 36.975921, acc = 0.865000\n",
  3233. "L = 36.958116, acc = 0.865000\n",
  3234. "L = 36.940199, acc = 0.865000\n",
  3235. "L = 36.922169, acc = 0.865000\n",
  3236. "L = 36.904025, acc = 0.865000\n",
  3237. "L = 36.885767, acc = 0.865000\n",
  3238. "L = 36.867393, acc = 0.870000\n",
  3239. "L = 36.848902, acc = 0.870000\n",
  3240. "L = 36.830294, acc = 0.870000\n",
  3241. "L = 36.811568, acc = 0.870000\n",
  3242. "L = 36.792724, acc = 0.870000\n",
  3243. "L = 36.773759, acc = 0.870000\n",
  3244. "L = 36.754674, acc = 0.870000\n",
  3245. "L = 36.735467, acc = 0.870000\n",
  3246. "L = 36.716138, acc = 0.870000\n",
  3247. "L = 36.696686, acc = 0.870000\n",
  3248. "L = 36.677111, acc = 0.870000\n",
  3249. "L = 36.657410, acc = 0.870000\n",
  3250. "L = 36.637584, acc = 0.870000\n",
  3251. "L = 36.617631, acc = 0.870000\n",
  3252. "L = 36.597551, acc = 0.870000\n",
  3253. "L = 36.577343, acc = 0.870000\n",
  3254. "L = 36.557006, acc = 0.870000\n",
  3255. "L = 36.536540, acc = 0.870000\n",
  3256. "L = 36.515943, acc = 0.875000\n",
  3257. "L = 36.495215, acc = 0.875000\n",
  3258. "L = 36.474354, acc = 0.875000\n",
  3259. "L = 36.453361, acc = 0.875000\n",
  3260. "L = 36.432234, acc = 0.875000\n",
  3261. "L = 36.410972, acc = 0.875000\n",
  3262. "L = 36.389576, acc = 0.875000\n",
  3263. "L = 36.368043, acc = 0.875000\n",
  3264. "L = 36.346373, acc = 0.875000\n",
  3265. "L = 36.324566, acc = 0.875000\n",
  3266. "L = 36.302620, acc = 0.875000\n",
  3267. "L = 36.280535, acc = 0.875000\n",
  3268. "L = 36.258310, acc = 0.875000\n",
  3269. "L = 36.235944, acc = 0.875000\n",
  3270. "L = 36.213437, acc = 0.875000\n",
  3271. "L = 36.190788, acc = 0.875000\n",
  3272. "L = 36.167996, acc = 0.875000\n",
  3273. "L = 36.145060, acc = 0.875000\n",
  3274. "L = 36.121980, acc = 0.875000\n",
  3275. "L = 36.098755, acc = 0.875000\n",
  3276. "L = 36.075384, acc = 0.875000\n",
  3277. "L = 36.051866, acc = 0.875000\n",
  3278. "L = 36.028201, acc = 0.875000\n",
  3279. "L = 36.004388, acc = 0.875000\n",
  3280. "L = 35.980427, acc = 0.875000\n",
  3281. "L = 35.956316, acc = 0.875000\n",
  3282. "L = 35.932055, acc = 0.875000\n",
  3283. "L = 35.907644, acc = 0.875000\n",
  3284. "L = 35.883081, acc = 0.875000\n",
  3285. "L = 35.858366, acc = 0.875000\n",
  3286. "L = 35.833499, acc = 0.875000\n",
  3287. "L = 35.808478, acc = 0.875000\n",
  3288. "L = 35.783303, acc = 0.875000\n",
  3289. "L = 35.757974, acc = 0.875000\n",
  3290. "L = 35.732489, acc = 0.875000\n",
  3291. "L = 35.706849, acc = 0.875000\n",
  3292. "L = 35.681052, acc = 0.875000\n",
  3293. "L = 35.655099, acc = 0.875000\n",
  3294. "L = 35.628988, acc = 0.875000\n",
  3295. "L = 35.602718, acc = 0.875000\n",
  3296. "L = 35.576290, acc = 0.875000\n",
  3297. "L = 35.549703, acc = 0.875000\n",
  3298. "L = 35.522956, acc = 0.875000\n",
  3299. "L = 35.496049, acc = 0.875000\n",
  3300. "L = 35.468980, acc = 0.875000\n",
  3301. "L = 35.441751, acc = 0.875000\n",
  3302. "L = 35.414359, acc = 0.875000\n",
  3303. "L = 35.386805, acc = 0.875000\n",
  3304. "L = 35.359088, acc = 0.875000\n",
  3305. "L = 35.331208, acc = 0.875000\n",
  3306. "L = 35.303164, acc = 0.875000\n",
  3307. "L = 35.274956, acc = 0.875000\n",
  3308. "L = 35.246582, acc = 0.875000\n",
  3309. "L = 35.218044, acc = 0.875000\n",
  3310. "L = 35.189340, acc = 0.875000\n",
  3311. "L = 35.160470, acc = 0.875000\n",
  3312. "L = 35.131434, acc = 0.875000\n",
  3313. "L = 35.102230, acc = 0.875000\n",
  3314. "L = 35.072860, acc = 0.875000\n",
  3315. "L = 35.043321, acc = 0.875000\n",
  3316. "L = 35.013615, acc = 0.875000\n",
  3317. "L = 34.983741, acc = 0.880000\n",
  3318. "L = 34.953697, acc = 0.880000\n",
  3319. "L = 34.923485, acc = 0.880000\n",
  3320. "L = 34.893103, acc = 0.880000\n",
  3321. "L = 34.862552, acc = 0.880000\n",
  3322. "L = 34.831831, acc = 0.880000\n",
  3323. "L = 34.800940, acc = 0.880000\n",
  3324. "L = 34.769878, acc = 0.880000\n",
  3325. "L = 34.738645, acc = 0.880000\n",
  3326. "L = 34.707242, acc = 0.880000\n",
  3327. "L = 34.675667, acc = 0.880000\n",
  3328. "L = 34.643920, acc = 0.880000\n",
  3329. "L = 34.612002, acc = 0.880000\n",
  3330. "L = 34.579912, acc = 0.880000\n",
  3331. "L = 34.547650, acc = 0.880000\n",
  3332. "L = 34.515216, acc = 0.880000\n",
  3333. "L = 34.482609, acc = 0.880000\n",
  3334. "L = 34.449830, acc = 0.880000\n",
  3335. "L = 34.416878, acc = 0.885000\n",
  3336. "L = 34.383754, acc = 0.885000\n",
  3337. "L = 34.350456, acc = 0.885000\n",
  3338. "L = 34.316985, acc = 0.885000\n",
  3339. "L = 34.283341, acc = 0.890000\n",
  3340. "L = 34.249524, acc = 0.890000\n",
  3341. "L = 34.215534, acc = 0.890000\n",
  3342. "L = 34.181370, acc = 0.890000\n",
  3343. "L = 34.147033, acc = 0.890000\n",
  3344. "L = 34.112523, acc = 0.890000\n",
  3345. "L = 34.077839, acc = 0.890000\n",
  3346. "L = 34.042982, acc = 0.890000\n",
  3347. "L = 34.007951, acc = 0.890000\n",
  3348. "L = 33.972747, acc = 0.890000\n",
  3349. "L = 33.937370, acc = 0.890000\n",
  3350. "L = 33.901819, acc = 0.890000\n",
  3351. "L = 33.866095, acc = 0.890000\n",
  3352. "L = 33.830199, acc = 0.890000\n",
  3353. "L = 33.794129, acc = 0.890000\n",
  3354. "L = 33.757886, acc = 0.890000\n",
  3355. "L = 33.721471, acc = 0.890000\n",
  3356. "L = 33.684882, acc = 0.890000\n",
  3357. "L = 33.648122, acc = 0.890000\n",
  3358. "L = 33.611189, acc = 0.890000\n",
  3359. "L = 33.574083, acc = 0.890000\n",
  3360. "L = 33.536806, acc = 0.890000\n",
  3361. "L = 33.499357, acc = 0.890000\n",
  3362. "L = 33.461737, acc = 0.895000\n",
  3363. "L = 33.423945, acc = 0.895000\n",
  3364. "L = 33.385982, acc = 0.895000\n",
  3365. "L = 33.347848, acc = 0.895000\n",
  3366. "L = 33.309543, acc = 0.895000\n",
  3367. "L = 33.271069, acc = 0.895000\n",
  3368. "L = 33.232424, acc = 0.895000\n",
  3369. "L = 33.193610, acc = 0.895000\n",
  3370. "L = 33.154626, acc = 0.895000\n",
  3371. "L = 33.115473, acc = 0.895000\n",
  3372. "L = 33.076151, acc = 0.895000\n",
  3373. "L = 33.036662, acc = 0.895000\n",
  3374. "L = 32.997004, acc = 0.895000\n",
  3375. "L = 32.957179, acc = 0.895000\n",
  3376. "L = 32.917186, acc = 0.895000\n",
  3377. "L = 32.877027, acc = 0.895000\n",
  3378. "L = 32.836701, acc = 0.895000\n",
  3379. "L = 32.796210, acc = 0.895000\n",
  3380. "L = 32.755553, acc = 0.895000\n",
  3381. "L = 32.714731, acc = 0.895000\n",
  3382. "L = 32.673745, acc = 0.895000\n",
  3383. "L = 32.632595, acc = 0.895000\n",
  3384. "L = 32.591282, acc = 0.895000\n",
  3385. "L = 32.549805, acc = 0.895000\n",
  3386. "L = 32.508166, acc = 0.895000\n",
  3387. "L = 32.466366, acc = 0.895000\n",
  3388. "L = 32.424404, acc = 0.895000\n",
  3389. "L = 32.382281, acc = 0.895000\n",
  3390. "L = 32.339998, acc = 0.895000\n",
  3391. "L = 32.297556, acc = 0.895000\n",
  3392. "L = 32.254955, acc = 0.895000\n",
  3393. "L = 32.212196, acc = 0.900000\n",
  3394. "L = 32.169279, acc = 0.900000\n",
  3395. "L = 32.126206, acc = 0.900000\n",
  3396. "L = 32.082976, acc = 0.900000\n",
  3397. "L = 32.039590, acc = 0.900000\n",
  3398. "L = 31.996050, acc = 0.900000\n",
  3399. "L = 31.952356, acc = 0.900000\n",
  3400. "L = 31.908508, acc = 0.900000\n",
  3401. "L = 31.864507, acc = 0.900000\n",
  3402. "L = 31.820355, acc = 0.900000\n",
  3403. "L = 31.776051, acc = 0.900000\n",
  3404. "L = 31.731597, acc = 0.900000\n",
  3405. "L = 31.686994, acc = 0.900000\n",
  3406. "L = 31.642241, acc = 0.900000\n",
  3407. "L = 31.597341, acc = 0.900000\n",
  3408. "L = 31.552294, acc = 0.900000\n",
  3409. "L = 31.507100, acc = 0.900000\n",
  3410. "L = 31.461761, acc = 0.900000\n",
  3411. "L = 31.416278, acc = 0.900000\n",
  3412. "L = 31.370651, acc = 0.900000\n",
  3413. "L = 31.324881, acc = 0.900000\n",
  3414. "L = 31.278969, acc = 0.900000\n",
  3415. "L = 31.232916, acc = 0.900000\n",
  3416. "L = 31.186724, acc = 0.900000\n",
  3417. "L = 31.140392, acc = 0.900000\n",
  3418. "L = 31.093922, acc = 0.900000\n",
  3419. "L = 31.047316, acc = 0.900000\n",
  3420. "L = 31.000573, acc = 0.900000\n",
  3421. "L = 30.953695, acc = 0.900000\n",
  3422. "L = 30.906683, acc = 0.900000\n",
  3423. "L = 30.859538, acc = 0.905000\n",
  3424. "L = 30.812261, acc = 0.905000\n",
  3425. "L = 30.764853, acc = 0.905000\n",
  3426. "L = 30.717315, acc = 0.905000\n",
  3427. "L = 30.669648, acc = 0.905000\n",
  3428. "L = 30.621854, acc = 0.905000\n",
  3429. "L = 30.573933, acc = 0.905000\n",
  3430. "L = 30.525886, acc = 0.910000\n",
  3431. "L = 30.477715, acc = 0.910000\n",
  3432. "L = 30.429421, acc = 0.910000\n",
  3433. "L = 30.381005, acc = 0.910000\n",
  3434. "L = 30.332468, acc = 0.910000\n",
  3435. "L = 30.283811, acc = 0.910000\n",
  3436. "L = 30.235036, acc = 0.910000\n",
  3437. "L = 30.186143, acc = 0.910000\n",
  3438. "L = 30.137135, acc = 0.910000\n",
  3439. "L = 30.088011, acc = 0.910000\n",
  3440. "L = 30.038774, acc = 0.910000\n",
  3441. "L = 29.989424, acc = 0.910000\n",
  3442. "L = 29.939963, acc = 0.910000\n",
  3443. "L = 29.890392, acc = 0.910000\n",
  3444. "L = 29.840713, acc = 0.910000\n",
  3445. "L = 29.790926, acc = 0.910000\n",
  3446. "L = 29.741034, acc = 0.910000\n",
  3447. "L = 29.691036, acc = 0.910000\n",
  3448. "L = 29.640935, acc = 0.910000\n",
  3449. "L = 29.590733, acc = 0.910000\n",
  3450. "L = 29.540429, acc = 0.910000\n",
  3451. "L = 29.490027, acc = 0.910000\n",
  3452. "L = 29.439526, acc = 0.915000\n",
  3453. "L = 29.388929, acc = 0.915000\n",
  3454. "L = 29.338237, acc = 0.915000\n",
  3455. "L = 29.287451, acc = 0.915000\n",
  3456. "L = 29.236573, acc = 0.915000\n",
  3457. "L = 29.185604, acc = 0.915000\n",
  3458. "L = 29.134546, acc = 0.915000\n",
  3459. "L = 29.083399, acc = 0.915000\n",
  3460. "L = 29.032166, acc = 0.915000\n",
  3461. "L = 28.980848, acc = 0.915000\n",
  3462. "L = 28.929446, acc = 0.915000\n",
  3463. "L = 28.877963, acc = 0.915000\n",
  3464. "L = 28.826398, acc = 0.915000\n",
  3465. "L = 28.774755, acc = 0.915000\n",
  3466. "L = 28.723034, acc = 0.915000\n",
  3467. "L = 28.671237, acc = 0.915000\n",
  3468. "L = 28.619366, acc = 0.915000\n",
  3469. "L = 28.567421, acc = 0.915000\n",
  3470. "L = 28.515405, acc = 0.915000\n",
  3471. "L = 28.463320, acc = 0.915000\n",
  3472. "L = 28.411166, acc = 0.915000\n",
  3473. "L = 28.358945, acc = 0.915000\n",
  3474. "L = 28.306660, acc = 0.915000\n",
  3475. "L = 28.254311, acc = 0.915000\n",
  3476. "L = 28.201900, acc = 0.915000\n",
  3477. "L = 28.149428, acc = 0.920000\n",
  3478. "L = 28.096899, acc = 0.920000\n",
  3479. "L = 28.044312, acc = 0.920000\n",
  3480. "L = 27.991670, acc = 0.920000\n",
  3481. "L = 27.938974, acc = 0.920000\n",
  3482. "L = 27.886226, acc = 0.920000\n",
  3483. "L = 27.833427, acc = 0.920000\n",
  3484. "L = 27.780580, acc = 0.920000\n",
  3485. "L = 27.727686, acc = 0.920000\n",
  3486. "L = 27.674747, acc = 0.920000\n",
  3487. "L = 27.621764, acc = 0.920000\n",
  3488. "L = 27.568739, acc = 0.920000\n",
  3489. "L = 27.515673, acc = 0.920000\n",
  3490. "L = 27.462569, acc = 0.925000\n",
  3491. "L = 27.409429, acc = 0.925000\n",
  3492. "L = 27.356253, acc = 0.925000\n",
  3493. "L = 27.303043, acc = 0.925000\n",
  3494. "L = 27.249802, acc = 0.925000\n",
  3495. "L = 27.196531, acc = 0.925000\n",
  3496. "L = 27.143232, acc = 0.925000\n",
  3497. "L = 27.089906, acc = 0.925000\n",
  3498. "L = 27.036556, acc = 0.925000\n",
  3499. "L = 26.983183, acc = 0.925000\n",
  3500. "L = 26.929788, acc = 0.925000\n",
  3501. "L = 26.876374, acc = 0.925000\n",
  3502. "L = 26.822943, acc = 0.925000\n",
  3503. "L = 26.769495, acc = 0.925000\n",
  3504. "L = 26.716034, acc = 0.925000\n",
  3505. "L = 26.662560, acc = 0.925000\n",
  3506. "L = 26.609075, acc = 0.925000\n",
  3507. "L = 26.555582, acc = 0.925000\n",
  3508. "L = 26.502081, acc = 0.925000\n",
  3509. "L = 26.448576, acc = 0.925000\n",
  3510. "L = 26.395067, acc = 0.925000\n",
  3511. "L = 26.341556, acc = 0.925000\n",
  3512. "L = 26.288045, acc = 0.925000\n",
  3513. "L = 26.234536, acc = 0.925000\n",
  3514. "L = 26.181031, acc = 0.930000\n",
  3515. "L = 26.127532, acc = 0.930000\n",
  3516. "L = 26.074040, acc = 0.930000\n"
  3517. ]
  3518. },
  3519. {
  3520. "name": "stdout",
  3521. "output_type": "stream",
  3522. "text": [
  3523. "L = 26.020557, acc = 0.930000\n",
  3524. "L = 25.967084, acc = 0.930000\n",
  3525. "L = 25.913625, acc = 0.930000\n",
  3526. "L = 25.860180, acc = 0.930000\n",
  3527. "L = 25.806751, acc = 0.930000\n",
  3528. "L = 25.753340, acc = 0.930000\n",
  3529. "L = 25.699949, acc = 0.930000\n",
  3530. "L = 25.646580, acc = 0.930000\n",
  3531. "L = 25.593234, acc = 0.930000\n",
  3532. "L = 25.539913, acc = 0.930000\n",
  3533. "L = 25.486619, acc = 0.930000\n",
  3534. "L = 25.433354, acc = 0.930000\n",
  3535. "L = 25.380119, acc = 0.930000\n",
  3536. "L = 25.326917, acc = 0.930000\n",
  3537. "L = 25.273749, acc = 0.930000\n",
  3538. "L = 25.220617, acc = 0.930000\n",
  3539. "L = 25.167522, acc = 0.930000\n",
  3540. "L = 25.114466, acc = 0.930000\n",
  3541. "L = 25.061452, acc = 0.930000\n",
  3542. "L = 25.008480, acc = 0.930000\n",
  3543. "L = 24.955553, acc = 0.930000\n",
  3544. "L = 24.902673, acc = 0.930000\n",
  3545. "L = 24.849840, acc = 0.935000\n",
  3546. "L = 24.797058, acc = 0.935000\n",
  3547. "L = 24.744326, acc = 0.935000\n",
  3548. "L = 24.691648, acc = 0.935000\n",
  3549. "L = 24.639025, acc = 0.935000\n",
  3550. "L = 24.586459, acc = 0.935000\n",
  3551. "L = 24.533951, acc = 0.935000\n",
  3552. "L = 24.481502, acc = 0.935000\n",
  3553. "L = 24.429116, acc = 0.935000\n",
  3554. "L = 24.376793, acc = 0.935000\n",
  3555. "L = 24.324535, acc = 0.935000\n",
  3556. "L = 24.272343, acc = 0.940000\n",
  3557. "L = 24.220220, acc = 0.940000\n",
  3558. "L = 24.168167, acc = 0.940000\n",
  3559. "L = 24.116186, acc = 0.940000\n",
  3560. "L = 24.064277, acc = 0.940000\n",
  3561. "L = 24.012444, acc = 0.940000\n",
  3562. "L = 23.960687, acc = 0.940000\n",
  3563. "L = 23.909008, acc = 0.940000\n",
  3564. "L = 23.857408, acc = 0.940000\n",
  3565. "L = 23.805890, acc = 0.940000\n",
  3566. "L = 23.754455, acc = 0.940000\n",
  3567. "L = 23.703103, acc = 0.940000\n",
  3568. "L = 23.651838, acc = 0.940000\n",
  3569. "L = 23.600660, acc = 0.940000\n",
  3570. "L = 23.549570, acc = 0.940000\n",
  3571. "L = 23.498571, acc = 0.940000\n",
  3572. "L = 23.447664, acc = 0.940000\n",
  3573. "L = 23.396850, acc = 0.940000\n",
  3574. "L = 23.346131, acc = 0.940000\n",
  3575. "L = 23.295508, acc = 0.940000\n",
  3576. "L = 23.244983, acc = 0.940000\n",
  3577. "L = 23.194557, acc = 0.940000\n",
  3578. "L = 23.144231, acc = 0.940000\n",
  3579. "L = 23.094008, acc = 0.940000\n",
  3580. "L = 23.043887, acc = 0.940000\n",
  3581. "L = 22.993871, acc = 0.940000\n",
  3582. "L = 22.943961, acc = 0.940000\n",
  3583. "L = 22.894159, acc = 0.940000\n",
  3584. "L = 22.844465, acc = 0.940000\n",
  3585. "L = 22.794881, acc = 0.940000\n",
  3586. "L = 22.745409, acc = 0.940000\n",
  3587. "L = 22.696048, acc = 0.940000\n",
  3588. "L = 22.646802, acc = 0.940000\n",
  3589. "L = 22.597671, acc = 0.945000\n",
  3590. "L = 22.548656, acc = 0.945000\n",
  3591. "L = 22.499758, acc = 0.945000\n",
  3592. "L = 22.450980, acc = 0.945000\n",
  3593. "L = 22.402321, acc = 0.945000\n",
  3594. "L = 22.353783, acc = 0.945000\n",
  3595. "L = 22.305367, acc = 0.945000\n",
  3596. "L = 22.257075, acc = 0.945000\n",
  3597. "L = 22.208907, acc = 0.945000\n",
  3598. "L = 22.160865, acc = 0.945000\n",
  3599. "L = 22.112950, acc = 0.945000\n",
  3600. "L = 22.065162, acc = 0.945000\n",
  3601. "L = 22.017503, acc = 0.945000\n",
  3602. "L = 21.969973, acc = 0.945000\n",
  3603. "L = 21.922575, acc = 0.945000\n",
  3604. "L = 21.875308, acc = 0.945000\n",
  3605. "L = 21.828175, acc = 0.945000\n",
  3606. "L = 21.781174, acc = 0.945000\n",
  3607. "L = 21.734309, acc = 0.945000\n",
  3608. "L = 21.687579, acc = 0.945000\n",
  3609. "L = 21.640986, acc = 0.945000\n",
  3610. "L = 21.594530, acc = 0.945000\n",
  3611. "L = 21.548213, acc = 0.945000\n",
  3612. "L = 21.502034, acc = 0.945000\n",
  3613. "L = 21.455996, acc = 0.945000\n",
  3614. "L = 21.410098, acc = 0.945000\n",
  3615. "L = 21.364342, acc = 0.945000\n",
  3616. "L = 21.318729, acc = 0.945000\n",
  3617. "L = 21.273258, acc = 0.945000\n",
  3618. "L = 21.227932, acc = 0.945000\n",
  3619. "L = 21.182750, acc = 0.945000\n",
  3620. "L = 21.137714, acc = 0.945000\n",
  3621. "L = 21.092823, acc = 0.945000\n",
  3622. "L = 21.048079, acc = 0.945000\n",
  3623. "L = 21.003483, acc = 0.945000\n",
  3624. "L = 20.959034, acc = 0.945000\n",
  3625. "L = 20.914734, acc = 0.945000\n",
  3626. "L = 20.870584, acc = 0.945000\n",
  3627. "L = 20.826583, acc = 0.945000\n",
  3628. "L = 20.782732, acc = 0.945000\n",
  3629. "L = 20.739032, acc = 0.945000\n",
  3630. "L = 20.695484, acc = 0.945000\n",
  3631. "L = 20.652088, acc = 0.945000\n",
  3632. "L = 20.608844, acc = 0.945000\n",
  3633. "L = 20.565752, acc = 0.945000\n",
  3634. "L = 20.522815, acc = 0.945000\n",
  3635. "L = 20.480031, acc = 0.945000\n",
  3636. "L = 20.437401, acc = 0.945000\n",
  3637. "L = 20.394926, acc = 0.945000\n",
  3638. "L = 20.352605, acc = 0.945000\n",
  3639. "L = 20.310440, acc = 0.945000\n",
  3640. "L = 20.268431, acc = 0.945000\n",
  3641. "L = 20.226578, acc = 0.945000\n",
  3642. "L = 20.184881, acc = 0.945000\n",
  3643. "L = 20.143341, acc = 0.945000\n",
  3644. "L = 20.101957, acc = 0.945000\n",
  3645. "L = 20.060731, acc = 0.945000\n",
  3646. "L = 20.019662, acc = 0.945000\n",
  3647. "L = 19.978751, acc = 0.945000\n",
  3648. "L = 19.937997, acc = 0.945000\n",
  3649. "L = 19.897402, acc = 0.945000\n",
  3650. "L = 19.856965, acc = 0.945000\n",
  3651. "L = 19.816686, acc = 0.945000\n",
  3652. "L = 19.776566, acc = 0.945000\n",
  3653. "L = 19.736604, acc = 0.945000\n",
  3654. "L = 19.696802, acc = 0.945000\n",
  3655. "L = 19.657158, acc = 0.945000\n",
  3656. "L = 19.617673, acc = 0.945000\n",
  3657. "L = 19.578347, acc = 0.945000\n",
  3658. "L = 19.539180, acc = 0.945000\n",
  3659. "L = 19.500173, acc = 0.945000\n",
  3660. "L = 19.461324, acc = 0.945000\n",
  3661. "L = 19.422635, acc = 0.945000\n",
  3662. "L = 19.384105, acc = 0.945000\n",
  3663. "L = 19.345734, acc = 0.945000\n",
  3664. "L = 19.307522, acc = 0.945000\n",
  3665. "L = 19.269469, acc = 0.945000\n",
  3666. "L = 19.231575, acc = 0.945000\n",
  3667. "L = 19.193840, acc = 0.945000\n",
  3668. "L = 19.156264, acc = 0.945000\n",
  3669. "L = 19.118847, acc = 0.945000\n",
  3670. "L = 19.081588, acc = 0.945000\n",
  3671. "L = 19.044488, acc = 0.945000\n",
  3672. "L = 19.007546, acc = 0.945000\n",
  3673. "L = 18.970763, acc = 0.945000\n",
  3674. "L = 18.934137, acc = 0.945000\n",
  3675. "L = 18.897670, acc = 0.945000\n",
  3676. "L = 18.861360, acc = 0.945000\n",
  3677. "L = 18.825208, acc = 0.945000\n",
  3678. "L = 18.789212, acc = 0.945000\n",
  3679. "L = 18.753374, acc = 0.945000\n",
  3680. "L = 18.717693, acc = 0.945000\n",
  3681. "L = 18.682169, acc = 0.945000\n",
  3682. "L = 18.646800, acc = 0.945000\n",
  3683. "L = 18.611588, acc = 0.945000\n",
  3684. "L = 18.576532, acc = 0.945000\n",
  3685. "L = 18.541631, acc = 0.945000\n",
  3686. "L = 18.506885, acc = 0.945000\n",
  3687. "L = 18.472295, acc = 0.945000\n",
  3688. "L = 18.437858, acc = 0.945000\n",
  3689. "L = 18.403577, acc = 0.945000\n",
  3690. "L = 18.369449, acc = 0.945000\n",
  3691. "L = 18.335475, acc = 0.945000\n",
  3692. "L = 18.301654, acc = 0.945000\n",
  3693. "L = 18.267985, acc = 0.945000\n",
  3694. "L = 18.234470, acc = 0.945000\n",
  3695. "L = 18.201106, acc = 0.945000\n",
  3696. "L = 18.167895, acc = 0.945000\n",
  3697. "L = 18.134835, acc = 0.945000\n",
  3698. "L = 18.101925, acc = 0.945000\n",
  3699. "L = 18.069167, acc = 0.945000\n",
  3700. "L = 18.036558, acc = 0.945000\n",
  3701. "L = 18.004099, acc = 0.945000\n",
  3702. "L = 17.971790, acc = 0.945000\n",
  3703. "L = 17.939629, acc = 0.945000\n",
  3704. "L = 17.907617, acc = 0.945000\n",
  3705. "L = 17.875753, acc = 0.945000\n",
  3706. "L = 17.844036, acc = 0.945000\n",
  3707. "L = 17.812467, acc = 0.945000\n",
  3708. "L = 17.781044, acc = 0.945000\n",
  3709. "L = 17.749767, acc = 0.945000\n",
  3710. "L = 17.718636, acc = 0.945000\n",
  3711. "L = 17.687650, acc = 0.945000\n",
  3712. "L = 17.656809, acc = 0.945000\n",
  3713. "L = 17.626111, acc = 0.945000\n",
  3714. "L = 17.595558, acc = 0.945000\n",
  3715. "L = 17.565148, acc = 0.945000\n",
  3716. "L = 17.534880, acc = 0.945000\n",
  3717. "L = 17.504755, acc = 0.945000\n",
  3718. "L = 17.474771, acc = 0.945000\n",
  3719. "L = 17.444929, acc = 0.945000\n",
  3720. "L = 17.415227, acc = 0.945000\n",
  3721. "L = 17.385665, acc = 0.945000\n",
  3722. "L = 17.356243, acc = 0.945000\n",
  3723. "L = 17.326960, acc = 0.945000\n",
  3724. "L = 17.297815, acc = 0.945000\n",
  3725. "L = 17.268808, acc = 0.945000\n",
  3726. "L = 17.239939, acc = 0.945000\n",
  3727. "L = 17.211206, acc = 0.945000\n",
  3728. "L = 17.182610, acc = 0.950000\n",
  3729. "L = 17.154149, acc = 0.950000\n",
  3730. "L = 17.125824, acc = 0.950000\n",
  3731. "L = 17.097633, acc = 0.950000\n",
  3732. "L = 17.069577, acc = 0.950000\n",
  3733. "L = 17.041653, acc = 0.950000\n",
  3734. "L = 17.013863, acc = 0.950000\n",
  3735. "L = 16.986205, acc = 0.950000\n",
  3736. "L = 16.958679, acc = 0.950000\n",
  3737. "L = 16.931284, acc = 0.950000\n",
  3738. "L = 16.904020, acc = 0.950000\n",
  3739. "L = 16.876886, acc = 0.950000\n",
  3740. "L = 16.849881, acc = 0.950000\n",
  3741. "L = 16.823006, acc = 0.950000\n",
  3742. "L = 16.796258, acc = 0.950000\n",
  3743. "L = 16.769639, acc = 0.950000\n",
  3744. "L = 16.743146, acc = 0.950000\n",
  3745. "L = 16.716780, acc = 0.950000\n",
  3746. "L = 16.690540, acc = 0.950000\n",
  3747. "L = 16.664426, acc = 0.950000\n",
  3748. "L = 16.638436, acc = 0.950000\n",
  3749. "L = 16.612571, acc = 0.950000\n",
  3750. "L = 16.586829, acc = 0.950000\n",
  3751. "L = 16.561211, acc = 0.950000\n",
  3752. "L = 16.535715, acc = 0.950000\n",
  3753. "L = 16.510341, acc = 0.950000\n",
  3754. "L = 16.485088, acc = 0.950000\n",
  3755. "L = 16.459956, acc = 0.950000\n",
  3756. "L = 16.434944, acc = 0.950000\n",
  3757. "L = 16.410051, acc = 0.950000\n",
  3758. "L = 16.385278, acc = 0.950000\n",
  3759. "L = 16.360623, acc = 0.950000\n",
  3760. "L = 16.336085, acc = 0.950000\n",
  3761. "L = 16.311665, acc = 0.950000\n",
  3762. "L = 16.287362, acc = 0.950000\n",
  3763. "L = 16.263175, acc = 0.950000\n",
  3764. "L = 16.239103, acc = 0.955000\n",
  3765. "L = 16.215146, acc = 0.955000\n",
  3766. "L = 16.191303, acc = 0.955000\n",
  3767. "L = 16.167574, acc = 0.955000\n",
  3768. "L = 16.143958, acc = 0.955000\n",
  3769. "L = 16.120455, acc = 0.955000\n",
  3770. "L = 16.097064, acc = 0.955000\n",
  3771. "L = 16.073784, acc = 0.955000\n",
  3772. "L = 16.050615, acc = 0.955000\n",
  3773. "L = 16.027556, acc = 0.955000\n",
  3774. "L = 16.004606, acc = 0.955000\n",
  3775. "L = 15.981766, acc = 0.955000\n",
  3776. "L = 15.959035, acc = 0.955000\n",
  3777. "L = 15.936411, acc = 0.955000\n",
  3778. "L = 15.913895, acc = 0.955000\n",
  3779. "L = 15.891485, acc = 0.955000\n",
  3780. "L = 15.869182, acc = 0.955000\n",
  3781. "L = 15.846984, acc = 0.955000\n",
  3782. "L = 15.824892, acc = 0.955000\n",
  3783. "L = 15.802904, acc = 0.955000\n",
  3784. "L = 15.781020, acc = 0.955000\n",
  3785. "L = 15.759239, acc = 0.955000\n",
  3786. "L = 15.737561, acc = 0.955000\n",
  3787. "L = 15.715986, acc = 0.955000\n",
  3788. "L = 15.694512, acc = 0.955000\n",
  3789. "L = 15.673140, acc = 0.955000\n",
  3790. "L = 15.651868, acc = 0.955000\n",
  3791. "L = 15.630696, acc = 0.955000\n",
  3792. "L = 15.609624, acc = 0.955000\n",
  3793. "L = 15.588651, acc = 0.955000\n",
  3794. "L = 15.567776, acc = 0.955000\n",
  3795. "L = 15.546999, acc = 0.955000\n",
  3796. "L = 15.526320, acc = 0.955000\n",
  3797. "L = 15.505737, acc = 0.955000\n",
  3798. "L = 15.485251, acc = 0.955000\n",
  3799. "L = 15.464860, acc = 0.955000\n",
  3800. "L = 15.444565, acc = 0.955000\n",
  3801. "L = 15.424364, acc = 0.955000\n",
  3802. "L = 15.404258, acc = 0.955000\n",
  3803. "L = 15.384245, acc = 0.955000\n",
  3804. "L = 15.364325, acc = 0.955000\n",
  3805. "L = 15.344498, acc = 0.955000\n",
  3806. "L = 15.324763, acc = 0.955000\n",
  3807. "L = 15.305119, acc = 0.955000\n",
  3808. "L = 15.285567, acc = 0.955000\n",
  3809. "L = 15.266105, acc = 0.955000\n",
  3810. "L = 15.246733, acc = 0.955000\n",
  3811. "L = 15.227450, acc = 0.955000\n",
  3812. "L = 15.208257, acc = 0.955000\n",
  3813. "L = 15.189152, acc = 0.955000\n",
  3814. "L = 15.170135, acc = 0.955000\n",
  3815. "L = 15.151205, acc = 0.955000\n",
  3816. "L = 15.132362, acc = 0.955000\n",
  3817. "L = 15.113606, acc = 0.955000\n",
  3818. "L = 15.094936, acc = 0.955000\n",
  3819. "L = 15.076352, acc = 0.955000\n",
  3820. "L = 15.057852, acc = 0.955000\n",
  3821. "L = 15.039437, acc = 0.955000\n",
  3822. "L = 15.021106, acc = 0.955000\n",
  3823. "L = 15.002859, acc = 0.955000\n",
  3824. "L = 14.984695, acc = 0.955000\n",
  3825. "L = 14.966613, acc = 0.955000\n",
  3826. "L = 14.948613, acc = 0.960000\n",
  3827. "L = 14.930695, acc = 0.960000\n",
  3828. "L = 14.912859, acc = 0.960000\n",
  3829. "L = 14.895103, acc = 0.960000\n",
  3830. "L = 14.877427, acc = 0.960000\n",
  3831. "L = 14.859831, acc = 0.960000\n",
  3832. "L = 14.842314, acc = 0.960000\n",
  3833. "L = 14.824876, acc = 0.960000\n",
  3834. "L = 14.807517, acc = 0.960000\n",
  3835. "L = 14.790236, acc = 0.960000\n",
  3836. "L = 14.773032, acc = 0.960000\n",
  3837. "L = 14.755905, acc = 0.960000\n",
  3838. "L = 14.738855, acc = 0.960000\n",
  3839. "L = 14.721881, acc = 0.960000\n",
  3840. "L = 14.704982, acc = 0.960000\n",
  3841. "L = 14.688159, acc = 0.960000\n",
  3842. "L = 14.671411, acc = 0.960000\n",
  3843. "L = 14.654737, acc = 0.960000\n",
  3844. "L = 14.638137, acc = 0.960000\n",
  3845. "L = 14.621611, acc = 0.960000\n",
  3846. "L = 14.605157, acc = 0.960000\n",
  3847. "L = 14.588777, acc = 0.960000\n",
  3848. "L = 14.572468, acc = 0.960000\n",
  3849. "L = 14.556232, acc = 0.960000\n",
  3850. "L = 14.540067, acc = 0.960000\n",
  3851. "L = 14.523973, acc = 0.960000\n",
  3852. "L = 14.507949, acc = 0.960000\n",
  3853. "L = 14.491996, acc = 0.960000\n",
  3854. "L = 14.476113, acc = 0.960000\n",
  3855. "L = 14.460298, acc = 0.960000\n",
  3856. "L = 14.444553, acc = 0.960000\n",
  3857. "L = 14.428877, acc = 0.960000\n",
  3858. "L = 14.413268, acc = 0.960000\n",
  3859. "L = 14.397727, acc = 0.960000\n",
  3860. "L = 14.382254, acc = 0.960000\n",
  3861. "L = 14.366847, acc = 0.960000\n",
  3862. "L = 14.351507, acc = 0.960000\n",
  3863. "L = 14.336234, acc = 0.960000\n",
  3864. "L = 14.321026, acc = 0.960000\n",
  3865. "L = 14.305883, acc = 0.960000\n",
  3866. "L = 14.290805, acc = 0.960000\n",
  3867. "L = 14.275793, acc = 0.960000\n",
  3868. "L = 14.260844, acc = 0.960000\n",
  3869. "L = 14.245959, acc = 0.960000\n",
  3870. "L = 14.231138, acc = 0.960000\n",
  3871. "L = 14.216380, acc = 0.960000\n",
  3872. "L = 14.201684, acc = 0.960000\n",
  3873. "L = 14.187051, acc = 0.960000\n",
  3874. "L = 14.172480, acc = 0.960000\n",
  3875. "L = 14.157971, acc = 0.960000\n",
  3876. "L = 14.143523, acc = 0.960000\n",
  3877. "L = 14.129136, acc = 0.960000\n",
  3878. "L = 14.114810, acc = 0.960000\n",
  3879. "L = 14.100543, acc = 0.960000\n",
  3880. "L = 14.086337, acc = 0.960000\n",
  3881. "L = 14.072190, acc = 0.960000\n",
  3882. "L = 14.058102, acc = 0.960000\n",
  3883. "L = 14.044074, acc = 0.960000\n",
  3884. "L = 14.030103, acc = 0.960000\n",
  3885. "L = 14.016191, acc = 0.960000\n",
  3886. "L = 14.002337, acc = 0.960000\n",
  3887. "L = 13.988540, acc = 0.960000\n",
  3888. "L = 13.974800, acc = 0.960000\n",
  3889. "L = 13.961117, acc = 0.960000\n",
  3890. "L = 13.947491, acc = 0.960000\n",
  3891. "L = 13.933920, acc = 0.960000\n",
  3892. "L = 13.920406, acc = 0.960000\n",
  3893. "L = 13.906947, acc = 0.960000\n",
  3894. "L = 13.893543, acc = 0.960000\n",
  3895. "L = 13.880194, acc = 0.960000\n",
  3896. "L = 13.866899, acc = 0.960000\n",
  3897. "L = 13.853659, acc = 0.960000\n",
  3898. "L = 13.840472, acc = 0.960000\n",
  3899. "L = 13.827339, acc = 0.960000\n",
  3900. "L = 13.814260, acc = 0.960000\n",
  3901. "L = 13.801233, acc = 0.960000\n"
  3902. ]
  3903. },
  3904. {
  3905. "name": "stdout",
  3906. "output_type": "stream",
  3907. "text": [
  3908. "L = 13.788259, acc = 0.960000\n",
  3909. "L = 13.775337, acc = 0.960000\n",
  3910. "L = 13.762467, acc = 0.965000\n",
  3911. "L = 13.749649, acc = 0.965000\n",
  3912. "L = 13.736883, acc = 0.965000\n",
  3913. "L = 13.724167, acc = 0.965000\n",
  3914. "L = 13.711503, acc = 0.965000\n",
  3915. "L = 13.698888, acc = 0.965000\n",
  3916. "L = 13.686324, acc = 0.965000\n",
  3917. "L = 13.673810, acc = 0.965000\n",
  3918. "L = 13.661346, acc = 0.965000\n",
  3919. "L = 13.648931, acc = 0.965000\n",
  3920. "L = 13.636565, acc = 0.965000\n",
  3921. "L = 13.624248, acc = 0.965000\n",
  3922. "L = 13.611980, acc = 0.965000\n",
  3923. "L = 13.599759, acc = 0.965000\n",
  3924. "L = 13.587587, acc = 0.965000\n",
  3925. "L = 13.575462, acc = 0.965000\n",
  3926. "L = 13.563385, acc = 0.965000\n",
  3927. "L = 13.551354, acc = 0.965000\n",
  3928. "L = 13.539371, acc = 0.965000\n",
  3929. "L = 13.527434, acc = 0.965000\n",
  3930. "L = 13.515544, acc = 0.965000\n",
  3931. "L = 13.503699, acc = 0.965000\n",
  3932. "L = 13.491901, acc = 0.965000\n",
  3933. "L = 13.480147, acc = 0.965000\n",
  3934. "L = 13.468439, acc = 0.965000\n",
  3935. "L = 13.456776, acc = 0.965000\n",
  3936. "L = 13.445158, acc = 0.965000\n",
  3937. "L = 13.433585, acc = 0.965000\n",
  3938. "L = 13.422055, acc = 0.965000\n",
  3939. "L = 13.410569, acc = 0.965000\n",
  3940. "L = 13.399128, acc = 0.965000\n",
  3941. "L = 13.387729, acc = 0.965000\n",
  3942. "L = 13.376374, acc = 0.965000\n",
  3943. "L = 13.365062, acc = 0.965000\n",
  3944. "L = 13.353792, acc = 0.965000\n",
  3945. "L = 13.342565, acc = 0.965000\n",
  3946. "L = 13.331380, acc = 0.965000\n",
  3947. "L = 13.320237, acc = 0.965000\n",
  3948. "L = 13.309136, acc = 0.965000\n",
  3949. "L = 13.298077, acc = 0.965000\n",
  3950. "L = 13.287059, acc = 0.965000\n",
  3951. "L = 13.276081, acc = 0.965000\n",
  3952. "L = 13.265145, acc = 0.965000\n",
  3953. "L = 13.254249, acc = 0.965000\n",
  3954. "L = 13.243393, acc = 0.965000\n",
  3955. "L = 13.232578, acc = 0.965000\n",
  3956. "L = 13.221802, acc = 0.965000\n",
  3957. "L = 13.211066, acc = 0.965000\n",
  3958. "L = 13.200370, acc = 0.965000\n",
  3959. "L = 13.189712, acc = 0.965000\n",
  3960. "L = 13.179094, acc = 0.965000\n",
  3961. "L = 13.168515, acc = 0.965000\n",
  3962. "L = 13.157973, acc = 0.965000\n",
  3963. "L = 13.147471, acc = 0.965000\n",
  3964. "L = 13.137006, acc = 0.965000\n",
  3965. "L = 13.126579, acc = 0.965000\n",
  3966. "L = 13.116190, acc = 0.965000\n",
  3967. "L = 13.105839, acc = 0.965000\n",
  3968. "L = 13.095525, acc = 0.965000\n",
  3969. "L = 13.085247, acc = 0.965000\n",
  3970. "L = 13.075007, acc = 0.965000\n",
  3971. "L = 13.064803, acc = 0.965000\n",
  3972. "L = 13.054636, acc = 0.965000\n",
  3973. "L = 13.044504, acc = 0.965000\n",
  3974. "L = 13.034409, acc = 0.965000\n",
  3975. "L = 13.024350, acc = 0.965000\n",
  3976. "L = 13.014326, acc = 0.965000\n",
  3977. "L = 13.004338, acc = 0.965000\n",
  3978. "L = 12.994385, acc = 0.965000\n",
  3979. "L = 12.984467, acc = 0.965000\n",
  3980. "L = 12.974583, acc = 0.965000\n",
  3981. "L = 12.964735, acc = 0.965000\n",
  3982. "L = 12.954920, acc = 0.965000\n",
  3983. "L = 12.945140, acc = 0.965000\n",
  3984. "L = 12.935394, acc = 0.965000\n",
  3985. "L = 12.925682, acc = 0.965000\n",
  3986. "L = 12.916004, acc = 0.965000\n",
  3987. "L = 12.906359, acc = 0.965000\n",
  3988. "L = 12.896747, acc = 0.965000\n",
  3989. "L = 12.887169, acc = 0.965000\n",
  3990. "L = 12.877623, acc = 0.965000\n",
  3991. "L = 12.868110, acc = 0.965000\n",
  3992. "L = 12.858630, acc = 0.965000\n",
  3993. "L = 12.849182, acc = 0.965000\n",
  3994. "L = 12.839766, acc = 0.965000\n",
  3995. "L = 12.830383, acc = 0.965000\n",
  3996. "L = 12.821031, acc = 0.965000\n",
  3997. "L = 12.811711, acc = 0.965000\n",
  3998. "L = 12.802422, acc = 0.965000\n",
  3999. "L = 12.793165, acc = 0.965000\n",
  4000. "L = 12.783939, acc = 0.965000\n",
  4001. "L = 12.774743, acc = 0.965000\n",
  4002. "L = 12.765579, acc = 0.965000\n",
  4003. "L = 12.756445, acc = 0.965000\n",
  4004. "L = 12.747342, acc = 0.965000\n",
  4005. "L = 12.738269, acc = 0.965000\n",
  4006. "L = 12.729226, acc = 0.965000\n",
  4007. "L = 12.720214, acc = 0.965000\n",
  4008. "L = 12.711231, acc = 0.965000\n",
  4009. "L = 12.702277, acc = 0.965000\n",
  4010. "L = 12.693354, acc = 0.965000\n",
  4011. "L = 12.684459, acc = 0.965000\n",
  4012. "L = 12.675594, acc = 0.965000\n",
  4013. "L = 12.666757, acc = 0.965000\n",
  4014. "L = 12.657950, acc = 0.965000\n",
  4015. "L = 12.649171, acc = 0.965000\n",
  4016. "L = 12.640421, acc = 0.965000\n",
  4017. "L = 12.631699, acc = 0.965000\n",
  4018. "L = 12.623006, acc = 0.965000\n",
  4019. "L = 12.614340, acc = 0.965000\n",
  4020. "L = 12.605703, acc = 0.965000\n",
  4021. "L = 12.597093, acc = 0.965000\n",
  4022. "L = 12.588511, acc = 0.965000\n",
  4023. "L = 12.579956, acc = 0.965000\n",
  4024. "L = 12.571429, acc = 0.965000\n",
  4025. "L = 12.562928, acc = 0.965000\n",
  4026. "L = 12.554455, acc = 0.965000\n",
  4027. "L = 12.546009, acc = 0.965000\n",
  4028. "L = 12.537590, acc = 0.965000\n",
  4029. "L = 12.529197, acc = 0.965000\n",
  4030. "L = 12.520831, acc = 0.965000\n",
  4031. "L = 12.512491, acc = 0.965000\n",
  4032. "L = 12.504177, acc = 0.965000\n",
  4033. "L = 12.495889, acc = 0.965000\n",
  4034. "L = 12.487627, acc = 0.965000\n",
  4035. "L = 12.479391, acc = 0.965000\n",
  4036. "L = 12.471180, acc = 0.965000\n",
  4037. "L = 12.462995, acc = 0.965000\n",
  4038. "L = 12.454836, acc = 0.965000\n",
  4039. "L = 12.446701, acc = 0.965000\n",
  4040. "L = 12.438592, acc = 0.965000\n",
  4041. "L = 12.430508, acc = 0.965000\n",
  4042. "L = 12.422448, acc = 0.965000\n",
  4043. "L = 12.414413, acc = 0.965000\n",
  4044. "L = 12.406403, acc = 0.965000\n",
  4045. "L = 12.398417, acc = 0.965000\n",
  4046. "L = 12.390456, acc = 0.965000\n",
  4047. "L = 12.382519, acc = 0.965000\n",
  4048. "L = 12.374605, acc = 0.965000\n",
  4049. "L = 12.366716, acc = 0.965000\n",
  4050. "L = 12.358851, acc = 0.965000\n",
  4051. "L = 12.351009, acc = 0.965000\n",
  4052. "L = 12.343190, acc = 0.965000\n",
  4053. "L = 12.335396, acc = 0.965000\n",
  4054. "L = 12.327624, acc = 0.965000\n",
  4055. "L = 12.319876, acc = 0.965000\n",
  4056. "L = 12.312151, acc = 0.965000\n",
  4057. "L = 12.304448, acc = 0.965000\n",
  4058. "L = 12.296769, acc = 0.965000\n",
  4059. "L = 12.289112, acc = 0.965000\n",
  4060. "L = 12.281478, acc = 0.965000\n",
  4061. "L = 12.273866, acc = 0.965000\n",
  4062. "L = 12.266277, acc = 0.965000\n",
  4063. "L = 12.258710, acc = 0.965000\n",
  4064. "L = 12.251165, acc = 0.965000\n",
  4065. "L = 12.243642, acc = 0.965000\n",
  4066. "L = 12.236141, acc = 0.965000\n",
  4067. "L = 12.228662, acc = 0.965000\n",
  4068. "L = 12.221204, acc = 0.965000\n",
  4069. "L = 12.213768, acc = 0.965000\n",
  4070. "L = 12.206354, acc = 0.965000\n",
  4071. "L = 12.198961, acc = 0.965000\n",
  4072. "L = 12.191589, acc = 0.965000\n",
  4073. "L = 12.184238, acc = 0.965000\n",
  4074. "L = 12.176908, acc = 0.965000\n",
  4075. "L = 12.169599, acc = 0.965000\n",
  4076. "L = 12.162311, acc = 0.965000\n",
  4077. "L = 12.155044, acc = 0.965000\n",
  4078. "L = 12.147797, acc = 0.965000\n",
  4079. "L = 12.140571, acc = 0.965000\n",
  4080. "L = 12.133365, acc = 0.965000\n",
  4081. "L = 12.126180, acc = 0.965000\n",
  4082. "L = 12.119015, acc = 0.965000\n",
  4083. "L = 12.111869, acc = 0.965000\n",
  4084. "L = 12.104744, acc = 0.965000\n",
  4085. "L = 12.097639, acc = 0.965000\n",
  4086. "L = 12.090553, acc = 0.965000\n",
  4087. "L = 12.083487, acc = 0.965000\n",
  4088. "L = 12.076441, acc = 0.965000\n",
  4089. "L = 12.069414, acc = 0.965000\n",
  4090. "L = 12.062407, acc = 0.965000\n",
  4091. "L = 12.055419, acc = 0.965000\n",
  4092. "L = 12.048450, acc = 0.965000\n",
  4093. "L = 12.041500, acc = 0.965000\n",
  4094. "L = 12.034569, acc = 0.965000\n",
  4095. "L = 12.027657, acc = 0.965000\n",
  4096. "L = 12.020764, acc = 0.965000\n",
  4097. "L = 12.013890, acc = 0.965000\n",
  4098. "L = 12.007034, acc = 0.965000\n",
  4099. "L = 12.000197, acc = 0.965000\n",
  4100. "L = 11.993379, acc = 0.965000\n",
  4101. "L = 11.986578, acc = 0.965000\n",
  4102. "L = 11.979796, acc = 0.965000\n",
  4103. "L = 11.973033, acc = 0.965000\n",
  4104. "L = 11.966287, acc = 0.965000\n",
  4105. "L = 11.959559, acc = 0.965000\n",
  4106. "L = 11.952849, acc = 0.965000\n",
  4107. "L = 11.946157, acc = 0.965000\n",
  4108. "L = 11.939483, acc = 0.965000\n",
  4109. "L = 11.932827, acc = 0.965000\n",
  4110. "L = 11.926188, acc = 0.965000\n",
  4111. "L = 11.919566, acc = 0.965000\n",
  4112. "L = 11.912962, acc = 0.965000\n",
  4113. "L = 11.906376, acc = 0.965000\n",
  4114. "L = 11.899806, acc = 0.965000\n",
  4115. "L = 11.893254, acc = 0.965000\n",
  4116. "L = 11.886718, acc = 0.965000\n",
  4117. "L = 11.880200, acc = 0.965000\n",
  4118. "L = 11.873699, acc = 0.965000\n",
  4119. "L = 11.867214, acc = 0.965000\n",
  4120. "L = 11.860747, acc = 0.965000\n",
  4121. "L = 11.854295, acc = 0.965000\n",
  4122. "L = 11.847861, acc = 0.965000\n",
  4123. "L = 11.841443, acc = 0.965000\n",
  4124. "L = 11.835041, acc = 0.965000\n",
  4125. "L = 11.828656, acc = 0.965000\n",
  4126. "L = 11.822287, acc = 0.965000\n",
  4127. "L = 11.815935, acc = 0.965000\n",
  4128. "L = 11.809598, acc = 0.965000\n",
  4129. "L = 11.803278, acc = 0.965000\n",
  4130. "L = 11.796973, acc = 0.965000\n",
  4131. "L = 11.790684, acc = 0.965000\n",
  4132. "L = 11.784412, acc = 0.965000\n",
  4133. "L = 11.778154, acc = 0.965000\n",
  4134. "L = 11.771913, acc = 0.965000\n",
  4135. "L = 11.765687, acc = 0.965000\n",
  4136. "L = 11.759477, acc = 0.965000\n",
  4137. "L = 11.753282, acc = 0.970000\n",
  4138. "L = 11.747103, acc = 0.970000\n",
  4139. "L = 11.740939, acc = 0.970000\n",
  4140. "L = 11.734790, acc = 0.970000\n",
  4141. "L = 11.728656, acc = 0.970000\n",
  4142. "L = 11.722538, acc = 0.970000\n",
  4143. "L = 11.716434, acc = 0.970000\n",
  4144. "L = 11.710346, acc = 0.970000\n",
  4145. "L = 11.704272, acc = 0.970000\n",
  4146. "L = 11.698213, acc = 0.970000\n",
  4147. "L = 11.692169, acc = 0.970000\n",
  4148. "L = 11.686140, acc = 0.970000\n",
  4149. "L = 11.680125, acc = 0.970000\n",
  4150. "L = 11.674125, acc = 0.970000\n",
  4151. "L = 11.668140, acc = 0.970000\n",
  4152. "L = 11.662169, acc = 0.970000\n",
  4153. "L = 11.656212, acc = 0.970000\n",
  4154. "L = 11.650269, acc = 0.970000\n",
  4155. "L = 11.644341, acc = 0.970000\n",
  4156. "L = 11.638427, acc = 0.970000\n",
  4157. "L = 11.632527, acc = 0.970000\n",
  4158. "L = 11.626641, acc = 0.970000\n",
  4159. "L = 11.620769, acc = 0.970000\n",
  4160. "L = 11.614911, acc = 0.970000\n",
  4161. "L = 11.609067, acc = 0.970000\n",
  4162. "L = 11.603237, acc = 0.970000\n",
  4163. "L = 11.597420, acc = 0.970000\n",
  4164. "L = 11.591618, acc = 0.970000\n",
  4165. "L = 11.585828, acc = 0.970000\n",
  4166. "L = 11.580053, acc = 0.970000\n",
  4167. "L = 11.574291, acc = 0.970000\n",
  4168. "L = 11.568542, acc = 0.970000\n",
  4169. "L = 11.562807, acc = 0.970000\n",
  4170. "L = 11.557085, acc = 0.970000\n",
  4171. "L = 11.551376, acc = 0.970000\n",
  4172. "L = 11.545680, acc = 0.970000\n",
  4173. "L = 11.539998, acc = 0.970000\n",
  4174. "L = 11.534329, acc = 0.970000\n",
  4175. "L = 11.528673, acc = 0.970000\n",
  4176. "L = 11.523029, acc = 0.970000\n",
  4177. "L = 11.517399, acc = 0.970000\n",
  4178. "L = 11.511782, acc = 0.970000\n",
  4179. "L = 11.506177, acc = 0.970000\n",
  4180. "L = 11.500585, acc = 0.970000\n",
  4181. "L = 11.495006, acc = 0.970000\n",
  4182. "L = 11.489440, acc = 0.970000\n",
  4183. "L = 11.483886, acc = 0.970000\n",
  4184. "L = 11.478345, acc = 0.970000\n",
  4185. "L = 11.472816, acc = 0.970000\n",
  4186. "L = 11.467300, acc = 0.970000\n",
  4187. "L = 11.461796, acc = 0.970000\n",
  4188. "L = 11.456304, acc = 0.970000\n",
  4189. "L = 11.450825, acc = 0.970000\n",
  4190. "L = 11.445358, acc = 0.970000\n",
  4191. "L = 11.439903, acc = 0.970000\n",
  4192. "L = 11.434461, acc = 0.970000\n",
  4193. "L = 11.429030, acc = 0.970000\n",
  4194. "L = 11.423612, acc = 0.970000\n",
  4195. "L = 11.418205, acc = 0.970000\n",
  4196. "L = 11.412810, acc = 0.970000\n",
  4197. "L = 11.407428, acc = 0.970000\n",
  4198. "L = 11.402057, acc = 0.970000\n",
  4199. "L = 11.396698, acc = 0.970000\n",
  4200. "L = 11.391351, acc = 0.970000\n",
  4201. "L = 11.386015, acc = 0.970000\n",
  4202. "L = 11.380691, acc = 0.970000\n",
  4203. "L = 11.375379, acc = 0.970000\n",
  4204. "L = 11.370078, acc = 0.970000\n",
  4205. "L = 11.364789, acc = 0.970000\n",
  4206. "L = 11.359511, acc = 0.970000\n",
  4207. "L = 11.354245, acc = 0.970000\n",
  4208. "L = 11.348990, acc = 0.970000\n",
  4209. "L = 11.343746, acc = 0.970000\n",
  4210. "L = 11.338514, acc = 0.970000\n",
  4211. "L = 11.333293, acc = 0.970000\n",
  4212. "L = 11.328083, acc = 0.970000\n",
  4213. "L = 11.322884, acc = 0.970000\n",
  4214. "L = 11.317696, acc = 0.970000\n",
  4215. "L = 11.312520, acc = 0.970000\n",
  4216. "L = 11.307354, acc = 0.970000\n",
  4217. "L = 11.302200, acc = 0.970000\n",
  4218. "L = 11.297056, acc = 0.970000\n",
  4219. "L = 11.291923, acc = 0.970000\n",
  4220. "L = 11.286802, acc = 0.970000\n",
  4221. "L = 11.281691, acc = 0.970000\n",
  4222. "L = 11.276590, acc = 0.970000\n",
  4223. "L = 11.271501, acc = 0.970000\n",
  4224. "L = 11.266422, acc = 0.970000\n",
  4225. "L = 11.261354, acc = 0.970000\n",
  4226. "L = 11.256296, acc = 0.970000\n",
  4227. "L = 11.251249, acc = 0.970000\n",
  4228. "L = 11.246213, acc = 0.970000\n",
  4229. "L = 11.241187, acc = 0.970000\n",
  4230. "L = 11.236171, acc = 0.970000\n",
  4231. "L = 11.231166, acc = 0.970000\n",
  4232. "L = 11.226172, acc = 0.970000\n",
  4233. "L = 11.221187, acc = 0.970000\n",
  4234. "L = 11.216213, acc = 0.970000\n",
  4235. "L = 11.211249, acc = 0.970000\n",
  4236. "L = 11.206296, acc = 0.970000\n",
  4237. "L = 11.201352, acc = 0.970000\n",
  4238. "L = 11.196419, acc = 0.970000\n",
  4239. "L = 11.191496, acc = 0.970000\n",
  4240. "L = 11.186583, acc = 0.970000\n",
  4241. "L = 11.181680, acc = 0.970000\n",
  4242. "L = 11.176787, acc = 0.970000\n",
  4243. "L = 11.171904, acc = 0.970000\n",
  4244. "L = 11.167031, acc = 0.970000\n",
  4245. "L = 11.162167, acc = 0.970000\n",
  4246. "L = 11.157314, acc = 0.970000\n",
  4247. "L = 11.152470, acc = 0.970000\n",
  4248. "L = 11.147637, acc = 0.970000\n",
  4249. "L = 11.142813, acc = 0.970000\n",
  4250. "L = 11.137998, acc = 0.970000\n",
  4251. "L = 11.133194, acc = 0.970000\n",
  4252. "L = 11.128399, acc = 0.970000\n",
  4253. "L = 11.123613, acc = 0.970000\n",
  4254. "L = 11.118838, acc = 0.970000\n",
  4255. "L = 11.114072, acc = 0.970000\n",
  4256. "L = 11.109315, acc = 0.970000\n",
  4257. "L = 11.104568, acc = 0.970000\n",
  4258. "L = 11.099830, acc = 0.970000\n",
  4259. "L = 11.095102, acc = 0.970000\n",
  4260. "L = 11.090383, acc = 0.970000\n",
  4261. "L = 11.085673, acc = 0.970000\n",
  4262. "L = 11.080973, acc = 0.970000\n",
  4263. "L = 11.076282, acc = 0.970000\n",
  4264. "L = 11.071600, acc = 0.970000\n",
  4265. "L = 11.066928, acc = 0.970000\n",
  4266. "L = 11.062264, acc = 0.970000\n",
  4267. "L = 11.057610, acc = 0.970000\n",
  4268. "L = 11.052965, acc = 0.970000\n",
  4269. "L = 11.048329, acc = 0.970000\n",
  4270. "L = 11.043702, acc = 0.970000\n",
  4271. "L = 11.039085, acc = 0.970000\n",
  4272. "L = 11.034476, acc = 0.970000\n",
  4273. "L = 11.029876, acc = 0.970000\n",
  4274. "L = 11.025285, acc = 0.970000\n",
  4275. "L = 11.020703, acc = 0.970000\n",
  4276. "L = 11.016130, acc = 0.970000\n",
  4277. "L = 11.011566, acc = 0.970000\n",
  4278. "L = 11.007011, acc = 0.970000\n",
  4279. "L = 11.002464, acc = 0.970000\n",
  4280. "L = 10.997927, acc = 0.970000\n",
  4281. "L = 10.993398, acc = 0.970000\n",
  4282. "L = 10.988877, acc = 0.970000\n",
  4283. "L = 10.984366, acc = 0.970000\n",
  4284. "L = 10.979863, acc = 0.970000\n",
  4285. "L = 10.975369, acc = 0.970000\n",
  4286. "L = 10.970883, acc = 0.970000\n",
  4287. "L = 10.966406, acc = 0.970000\n",
  4288. "L = 10.961938, acc = 0.970000\n",
  4289. "L = 10.957478, acc = 0.970000\n",
  4290. "L = 10.953026, acc = 0.970000\n"
  4291. ]
  4292. },
  4293. {
  4294. "name": "stdout",
  4295. "output_type": "stream",
  4296. "text": [
  4297. "L = 10.948583, acc = 0.970000\n",
  4298. "L = 10.944149, acc = 0.970000\n",
  4299. "L = 10.939723, acc = 0.970000\n",
  4300. "L = 10.935305, acc = 0.970000\n",
  4301. "L = 10.930896, acc = 0.965000\n",
  4302. "L = 10.926495, acc = 0.965000\n",
  4303. "L = 10.922102, acc = 0.965000\n",
  4304. "L = 10.917718, acc = 0.965000\n",
  4305. "L = 10.913342, acc = 0.965000\n",
  4306. "L = 10.908974, acc = 0.965000\n",
  4307. "L = 10.904614, acc = 0.965000\n",
  4308. "L = 10.900263, acc = 0.965000\n",
  4309. "L = 10.895920, acc = 0.965000\n",
  4310. "L = 10.891585, acc = 0.965000\n",
  4311. "L = 10.887258, acc = 0.965000\n",
  4312. "L = 10.882939, acc = 0.965000\n",
  4313. "L = 10.878628, acc = 0.965000\n",
  4314. "L = 10.874325, acc = 0.965000\n",
  4315. "L = 10.870031, acc = 0.965000\n",
  4316. "L = 10.865744, acc = 0.965000\n",
  4317. "L = 10.861465, acc = 0.965000\n",
  4318. "L = 10.857195, acc = 0.965000\n",
  4319. "L = 10.852932, acc = 0.965000\n",
  4320. "L = 10.848677, acc = 0.965000\n",
  4321. "L = 10.844430, acc = 0.965000\n",
  4322. "L = 10.840191, acc = 0.965000\n",
  4323. "L = 10.835959, acc = 0.965000\n",
  4324. "L = 10.831736, acc = 0.965000\n",
  4325. "L = 10.827520, acc = 0.965000\n",
  4326. "L = 10.823312, acc = 0.965000\n",
  4327. "L = 10.819112, acc = 0.965000\n",
  4328. "L = 10.814919, acc = 0.965000\n",
  4329. "L = 10.810735, acc = 0.965000\n",
  4330. "L = 10.806558, acc = 0.965000\n",
  4331. "L = 10.802388, acc = 0.965000\n",
  4332. "L = 10.798226, acc = 0.965000\n",
  4333. "L = 10.794072, acc = 0.965000\n",
  4334. "L = 10.789926, acc = 0.965000\n",
  4335. "L = 10.785787, acc = 0.965000\n",
  4336. "L = 10.781655, acc = 0.965000\n",
  4337. "L = 10.777531, acc = 0.965000\n",
  4338. "L = 10.773415, acc = 0.965000\n",
  4339. "L = 10.769306, acc = 0.965000\n",
  4340. "L = 10.765205, acc = 0.965000\n",
  4341. "L = 10.761111, acc = 0.965000\n",
  4342. "L = 10.757024, acc = 0.965000\n",
  4343. "L = 10.752945, acc = 0.965000\n",
  4344. "L = 10.748873, acc = 0.965000\n",
  4345. "L = 10.744809, acc = 0.965000\n",
  4346. "L = 10.740752, acc = 0.965000\n",
  4347. "L = 10.736702, acc = 0.965000\n",
  4348. "L = 10.732660, acc = 0.965000\n",
  4349. "L = 10.728625, acc = 0.965000\n",
  4350. "L = 10.724597, acc = 0.965000\n",
  4351. "L = 10.720576, acc = 0.965000\n",
  4352. "L = 10.716563, acc = 0.965000\n",
  4353. "L = 10.712557, acc = 0.965000\n",
  4354. "L = 10.708558, acc = 0.965000\n",
  4355. "L = 10.704566, acc = 0.965000\n",
  4356. "L = 10.700582, acc = 0.965000\n",
  4357. "L = 10.696604, acc = 0.965000\n",
  4358. "L = 10.692634, acc = 0.965000\n",
  4359. "L = 10.688671, acc = 0.965000\n",
  4360. "L = 10.684715, acc = 0.965000\n",
  4361. "L = 10.680766, acc = 0.965000\n",
  4362. "L = 10.676824, acc = 0.965000\n",
  4363. "L = 10.672889, acc = 0.965000\n",
  4364. "L = 10.668961, acc = 0.965000\n",
  4365. "L = 10.665040, acc = 0.965000\n",
  4366. "L = 10.661126, acc = 0.965000\n",
  4367. "L = 10.657219, acc = 0.965000\n",
  4368. "L = 10.653318, acc = 0.965000\n",
  4369. "L = 10.649425, acc = 0.965000\n",
  4370. "L = 10.645539, acc = 0.965000\n",
  4371. "L = 10.641659, acc = 0.965000\n",
  4372. "L = 10.637787, acc = 0.965000\n",
  4373. "L = 10.633921, acc = 0.965000\n",
  4374. "L = 10.630062, acc = 0.965000\n",
  4375. "L = 10.626210, acc = 0.965000\n",
  4376. "L = 10.622365, acc = 0.965000\n",
  4377. "L = 10.618526, acc = 0.965000\n",
  4378. "L = 10.614695, acc = 0.965000\n",
  4379. "L = 10.610870, acc = 0.965000\n",
  4380. "L = 10.607051, acc = 0.965000\n",
  4381. "L = 10.603240, acc = 0.965000\n",
  4382. "L = 10.599435, acc = 0.965000\n",
  4383. "L = 10.595637, acc = 0.965000\n",
  4384. "L = 10.591845, acc = 0.965000\n",
  4385. "L = 10.588060, acc = 0.965000\n",
  4386. "L = 10.584282, acc = 0.965000\n",
  4387. "L = 10.580510, acc = 0.965000\n",
  4388. "L = 10.576745, acc = 0.965000\n",
  4389. "L = 10.572987, acc = 0.965000\n",
  4390. "L = 10.569235, acc = 0.965000\n",
  4391. "L = 10.565490, acc = 0.965000\n",
  4392. "L = 10.561751, acc = 0.965000\n",
  4393. "L = 10.558019, acc = 0.965000\n",
  4394. "L = 10.554293, acc = 0.965000\n",
  4395. "L = 10.550574, acc = 0.965000\n",
  4396. "L = 10.546861, acc = 0.965000\n",
  4397. "L = 10.543154, acc = 0.965000\n",
  4398. "L = 10.539454, acc = 0.965000\n",
  4399. "L = 10.535761, acc = 0.965000\n",
  4400. "L = 10.532074, acc = 0.965000\n",
  4401. "L = 10.528393, acc = 0.965000\n",
  4402. "L = 10.524719, acc = 0.965000\n",
  4403. "L = 10.521051, acc = 0.965000\n",
  4404. "L = 10.517389, acc = 0.965000\n",
  4405. "L = 10.513734, acc = 0.965000\n",
  4406. "L = 10.510085, acc = 0.965000\n",
  4407. "L = 10.506442, acc = 0.965000\n",
  4408. "L = 10.502806, acc = 0.965000\n",
  4409. "L = 10.499176, acc = 0.965000\n",
  4410. "L = 10.495552, acc = 0.965000\n",
  4411. "L = 10.491934, acc = 0.965000\n",
  4412. "L = 10.488323, acc = 0.965000\n",
  4413. "L = 10.484717, acc = 0.965000\n",
  4414. "L = 10.481118, acc = 0.965000\n",
  4415. "L = 10.477526, acc = 0.965000\n",
  4416. "L = 10.473939, acc = 0.965000\n",
  4417. "L = 10.470359, acc = 0.965000\n",
  4418. "L = 10.466784, acc = 0.965000\n",
  4419. "L = 10.463216, acc = 0.965000\n",
  4420. "L = 10.459654, acc = 0.965000\n",
  4421. "L = 10.456098, acc = 0.965000\n",
  4422. "L = 10.452548, acc = 0.965000\n",
  4423. "L = 10.449004, acc = 0.965000\n",
  4424. "L = 10.445466, acc = 0.965000\n",
  4425. "L = 10.441935, acc = 0.965000\n",
  4426. "L = 10.438409, acc = 0.965000\n",
  4427. "L = 10.434889, acc = 0.965000\n",
  4428. "L = 10.431376, acc = 0.965000\n",
  4429. "L = 10.427868, acc = 0.965000\n",
  4430. "L = 10.424366, acc = 0.965000\n",
  4431. "L = 10.420870, acc = 0.965000\n",
  4432. "L = 10.417381, acc = 0.965000\n",
  4433. "L = 10.413897, acc = 0.965000\n",
  4434. "L = 10.410419, acc = 0.965000\n",
  4435. "L = 10.406947, acc = 0.965000\n",
  4436. "L = 10.403481, acc = 0.965000\n",
  4437. "L = 10.400020, acc = 0.965000\n",
  4438. "L = 10.396566, acc = 0.965000\n",
  4439. "L = 10.393117, acc = 0.965000\n",
  4440. "L = 10.389675, acc = 0.965000\n",
  4441. "L = 10.386238, acc = 0.965000\n",
  4442. "L = 10.382807, acc = 0.965000\n",
  4443. "L = 10.379381, acc = 0.965000\n",
  4444. "L = 10.375962, acc = 0.965000\n",
  4445. "L = 10.372548, acc = 0.965000\n",
  4446. "L = 10.369140, acc = 0.965000\n",
  4447. "L = 10.365738, acc = 0.965000\n",
  4448. "L = 10.362341, acc = 0.965000\n",
  4449. "L = 10.358951, acc = 0.965000\n",
  4450. "L = 10.355566, acc = 0.965000\n",
  4451. "L = 10.352186, acc = 0.965000\n",
  4452. "L = 10.348813, acc = 0.965000\n",
  4453. "L = 10.345445, acc = 0.965000\n",
  4454. "L = 10.342082, acc = 0.965000\n",
  4455. "L = 10.338726, acc = 0.965000\n",
  4456. "L = 10.335375, acc = 0.965000\n",
  4457. "L = 10.332029, acc = 0.965000\n",
  4458. "L = 10.328690, acc = 0.965000\n",
  4459. "L = 10.325355, acc = 0.965000\n",
  4460. "L = 10.322027, acc = 0.965000\n",
  4461. "L = 10.318704, acc = 0.965000\n",
  4462. "L = 10.315386, acc = 0.965000\n",
  4463. "L = 10.312074, acc = 0.965000\n",
  4464. "L = 10.308768, acc = 0.965000\n",
  4465. "L = 10.305467, acc = 0.965000\n",
  4466. "L = 10.302172, acc = 0.965000\n",
  4467. "L = 10.298882, acc = 0.965000\n",
  4468. "L = 10.295597, acc = 0.965000\n",
  4469. "L = 10.292319, acc = 0.965000\n",
  4470. "L = 10.289045, acc = 0.965000\n",
  4471. "L = 10.285777, acc = 0.965000\n",
  4472. "L = 10.282515, acc = 0.965000\n",
  4473. "L = 10.279258, acc = 0.965000\n",
  4474. "L = 10.276006, acc = 0.965000\n",
  4475. "L = 10.272760, acc = 0.965000\n",
  4476. "L = 10.269519, acc = 0.965000\n",
  4477. "L = 10.266283, acc = 0.965000\n",
  4478. "L = 10.263053, acc = 0.965000\n",
  4479. "L = 10.259828, acc = 0.965000\n",
  4480. "L = 10.256609, acc = 0.965000\n",
  4481. "L = 10.253395, acc = 0.965000\n",
  4482. "L = 10.250186, acc = 0.965000\n",
  4483. "L = 10.246983, acc = 0.965000\n",
  4484. "L = 10.243785, acc = 0.965000\n",
  4485. "L = 10.240592, acc = 0.965000\n",
  4486. "L = 10.237404, acc = 0.965000\n",
  4487. "L = 10.234222, acc = 0.965000\n",
  4488. "L = 10.231045, acc = 0.965000\n",
  4489. "L = 10.227873, acc = 0.965000\n",
  4490. "L = 10.224707, acc = 0.965000\n",
  4491. "L = 10.221545, acc = 0.965000\n",
  4492. "L = 10.218389, acc = 0.965000\n",
  4493. "L = 10.215238, acc = 0.965000\n",
  4494. "L = 10.212092, acc = 0.965000\n",
  4495. "L = 10.208952, acc = 0.965000\n",
  4496. "L = 10.205816, acc = 0.965000\n",
  4497. "L = 10.202686, acc = 0.965000\n",
  4498. "L = 10.199561, acc = 0.965000\n",
  4499. "L = 10.196441, acc = 0.965000\n",
  4500. "L = 10.193326, acc = 0.965000\n",
  4501. "L = 10.190216, acc = 0.965000\n",
  4502. "L = 10.187112, acc = 0.965000\n",
  4503. "L = 10.184012, acc = 0.965000\n",
  4504. "L = 10.180918, acc = 0.965000\n",
  4505. "L = 10.177828, acc = 0.965000\n",
  4506. "L = 10.174744, acc = 0.965000\n",
  4507. "L = 10.171665, acc = 0.965000\n",
  4508. "L = 10.168591, acc = 0.965000\n",
  4509. "L = 10.165521, acc = 0.965000\n",
  4510. "L = 10.162457, acc = 0.965000\n",
  4511. "L = 10.159398, acc = 0.965000\n",
  4512. "L = 10.156344, acc = 0.965000\n",
  4513. "L = 10.153294, acc = 0.965000\n",
  4514. "L = 10.150250, acc = 0.965000\n",
  4515. "L = 10.147211, acc = 0.965000\n",
  4516. "L = 10.144176, acc = 0.965000\n",
  4517. "L = 10.141147, acc = 0.965000\n",
  4518. "L = 10.138122, acc = 0.965000\n",
  4519. "L = 10.135103, acc = 0.965000\n",
  4520. "L = 10.132088, acc = 0.965000\n",
  4521. "L = 10.129078, acc = 0.965000\n",
  4522. "L = 10.126073, acc = 0.965000\n",
  4523. "L = 10.123073, acc = 0.965000\n",
  4524. "L = 10.120078, acc = 0.965000\n",
  4525. "L = 10.117088, acc = 0.965000\n",
  4526. "L = 10.114103, acc = 0.965000\n",
  4527. "L = 10.111122, acc = 0.965000\n",
  4528. "L = 10.108146, acc = 0.965000\n",
  4529. "L = 10.105175, acc = 0.965000\n",
  4530. "L = 10.102209, acc = 0.965000\n",
  4531. "L = 10.099248, acc = 0.965000\n",
  4532. "L = 10.096291, acc = 0.965000\n",
  4533. "L = 10.093339, acc = 0.965000\n",
  4534. "L = 10.090392, acc = 0.965000\n",
  4535. "L = 10.087450, acc = 0.965000\n",
  4536. "L = 10.084512, acc = 0.965000\n",
  4537. "L = 10.081579, acc = 0.965000\n",
  4538. "L = 10.078651, acc = 0.965000\n",
  4539. "L = 10.075728, acc = 0.965000\n",
  4540. "L = 10.072809, acc = 0.965000\n",
  4541. "L = 10.069895, acc = 0.965000\n",
  4542. "L = 10.066986, acc = 0.965000\n",
  4543. "L = 10.064081, acc = 0.965000\n",
  4544. "L = 10.061181, acc = 0.965000\n",
  4545. "L = 10.058286, acc = 0.965000\n",
  4546. "L = 10.055395, acc = 0.965000\n",
  4547. "L = 10.052509, acc = 0.965000\n",
  4548. "L = 10.049628, acc = 0.965000\n",
  4549. "L = 10.046751, acc = 0.965000\n",
  4550. "L = 10.043879, acc = 0.965000\n",
  4551. "L = 10.041011, acc = 0.965000\n",
  4552. "L = 10.038148, acc = 0.965000\n",
  4553. "L = 10.035290, acc = 0.965000\n",
  4554. "L = 10.032436, acc = 0.965000\n",
  4555. "L = 10.029586, acc = 0.965000\n",
  4556. "L = 10.026742, acc = 0.965000\n",
  4557. "L = 10.023901, acc = 0.965000\n",
  4558. "L = 10.021066, acc = 0.965000\n",
  4559. "L = 10.018234, acc = 0.965000\n",
  4560. "L = 10.015408, acc = 0.965000\n",
  4561. "L = 10.012586, acc = 0.965000\n",
  4562. "L = 10.009768, acc = 0.965000\n",
  4563. "L = 10.006955, acc = 0.965000\n",
  4564. "L = 10.004146, acc = 0.965000\n",
  4565. "L = 10.001342, acc = 0.965000\n",
  4566. "L = 9.998542, acc = 0.965000\n",
  4567. "L = 9.995746, acc = 0.965000\n",
  4568. "L = 9.992955, acc = 0.965000\n",
  4569. "L = 9.990169, acc = 0.965000\n",
  4570. "L = 9.987387, acc = 0.965000\n",
  4571. "L = 9.984609, acc = 0.965000\n",
  4572. "L = 9.981835, acc = 0.965000\n",
  4573. "L = 9.979066, acc = 0.965000\n",
  4574. "L = 9.976302, acc = 0.965000\n",
  4575. "L = 9.973542, acc = 0.965000\n",
  4576. "L = 9.970786, acc = 0.965000\n",
  4577. "L = 9.968034, acc = 0.965000\n",
  4578. "L = 9.965287, acc = 0.965000\n",
  4579. "L = 9.962544, acc = 0.965000\n",
  4580. "L = 9.959806, acc = 0.965000\n",
  4581. "L = 9.957071, acc = 0.965000\n",
  4582. "L = 9.954342, acc = 0.965000\n",
  4583. "L = 9.951616, acc = 0.965000\n",
  4584. "L = 9.948895, acc = 0.965000\n",
  4585. "L = 9.946177, acc = 0.965000\n",
  4586. "L = 9.943465, acc = 0.970000\n",
  4587. "L = 9.940756, acc = 0.970000\n",
  4588. "L = 9.938052, acc = 0.970000\n",
  4589. "L = 9.935352, acc = 0.970000\n",
  4590. "L = 9.932656, acc = 0.970000\n",
  4591. "L = 9.929964, acc = 0.970000\n",
  4592. "L = 9.927277, acc = 0.970000\n",
  4593. "L = 9.924594, acc = 0.970000\n",
  4594. "L = 9.921915, acc = 0.970000\n",
  4595. "L = 9.919240, acc = 0.970000\n",
  4596. "L = 9.916569, acc = 0.970000\n",
  4597. "L = 9.913903, acc = 0.970000\n",
  4598. "L = 9.911240, acc = 0.970000\n",
  4599. "L = 9.908582, acc = 0.970000\n",
  4600. "L = 9.905928, acc = 0.970000\n",
  4601. "L = 9.903278, acc = 0.970000\n",
  4602. "L = 9.900632, acc = 0.970000\n",
  4603. "L = 9.897991, acc = 0.970000\n",
  4604. "L = 9.895353, acc = 0.970000\n",
  4605. "L = 9.892720, acc = 0.970000\n",
  4606. "L = 9.890090, acc = 0.970000\n",
  4607. "L = 9.887465, acc = 0.970000\n",
  4608. "L = 9.884844, acc = 0.970000\n",
  4609. "L = 9.882227, acc = 0.970000\n",
  4610. "L = 9.879614, acc = 0.970000\n",
  4611. "L = 9.877004, acc = 0.970000\n",
  4612. "L = 9.874399, acc = 0.970000\n",
  4613. "L = 9.871798, acc = 0.970000\n",
  4614. "L = 9.869201, acc = 0.970000\n",
  4615. "L = 9.866608, acc = 0.970000\n",
  4616. "L = 9.864019, acc = 0.970000\n",
  4617. "L = 9.861434, acc = 0.970000\n",
  4618. "L = 9.858853, acc = 0.970000\n",
  4619. "L = 9.856276, acc = 0.970000\n",
  4620. "L = 9.853703, acc = 0.970000\n",
  4621. "L = 9.851134, acc = 0.970000\n",
  4622. "L = 9.848569, acc = 0.970000\n",
  4623. "L = 9.846008, acc = 0.970000\n",
  4624. "L = 9.843450, acc = 0.970000\n",
  4625. "L = 9.840897, acc = 0.970000\n",
  4626. "L = 9.838348, acc = 0.970000\n",
  4627. "L = 9.835802, acc = 0.970000\n",
  4628. "L = 9.833261, acc = 0.970000\n",
  4629. "L = 9.830723, acc = 0.970000\n",
  4630. "L = 9.828189, acc = 0.970000\n",
  4631. "L = 9.825659, acc = 0.970000\n",
  4632. "L = 9.823133, acc = 0.970000\n",
  4633. "L = 9.820611, acc = 0.970000\n",
  4634. "L = 9.818092, acc = 0.970000\n",
  4635. "L = 9.815578, acc = 0.970000\n",
  4636. "L = 9.813067, acc = 0.970000\n",
  4637. "L = 9.810560, acc = 0.970000\n",
  4638. "L = 9.808057, acc = 0.970000\n",
  4639. "L = 9.805558, acc = 0.970000\n",
  4640. "L = 9.803062, acc = 0.970000\n",
  4641. "L = 9.800570, acc = 0.970000\n",
  4642. "L = 9.798082, acc = 0.970000\n",
  4643. "L = 9.795598, acc = 0.970000\n",
  4644. "L = 9.793118, acc = 0.970000\n",
  4645. "L = 9.790641, acc = 0.970000\n",
  4646. "L = 9.788168, acc = 0.970000\n",
  4647. "L = 9.785699, acc = 0.970000\n",
  4648. "L = 9.783234, acc = 0.970000\n",
  4649. "L = 9.780772, acc = 0.970000\n",
  4650. "L = 9.778314, acc = 0.970000\n",
  4651. "L = 9.775860, acc = 0.970000\n",
  4652. "L = 9.773410, acc = 0.970000\n",
  4653. "L = 9.770963, acc = 0.970000\n",
  4654. "L = 9.768520, acc = 0.970000\n",
  4655. "L = 9.766080, acc = 0.970000\n",
  4656. "L = 9.763645, acc = 0.970000\n",
  4657. "L = 9.761213, acc = 0.970000\n",
  4658. "L = 9.758784, acc = 0.970000\n",
  4659. "L = 9.756359, acc = 0.970000\n",
  4660. "L = 9.753938, acc = 0.970000\n",
  4661. "L = 9.751521, acc = 0.970000\n",
  4662. "L = 9.749107, acc = 0.970000\n",
  4663. "L = 9.746696, acc = 0.970000\n",
  4664. "L = 9.744290, acc = 0.970000\n",
  4665. "L = 9.741887, acc = 0.970000\n",
  4666. "L = 9.739487, acc = 0.970000\n",
  4667. "L = 9.737091, acc = 0.970000\n",
  4668. "L = 9.734699, acc = 0.970000\n",
  4669. "L = 9.732310, acc = 0.970000\n",
  4670. "L = 9.729925, acc = 0.970000\n",
  4671. "L = 9.727544, acc = 0.970000\n",
  4672. "L = 9.725166, acc = 0.970000\n",
  4673. "L = 9.722791, acc = 0.970000\n",
  4674. "L = 9.720420, acc = 0.970000\n",
  4675. "L = 9.718053, acc = 0.970000\n",
  4676. "L = 9.715689, acc = 0.970000\n"
  4677. ]
  4678. },
  4679. {
  4680. "name": "stdout",
  4681. "output_type": "stream",
  4682. "text": [
  4683. "L = 9.713328, acc = 0.970000\n",
  4684. "L = 9.710971, acc = 0.970000\n",
  4685. "L = 9.708618, acc = 0.970000\n",
  4686. "L = 9.706268, acc = 0.970000\n",
  4687. "L = 9.703921, acc = 0.970000\n",
  4688. "L = 9.701578, acc = 0.970000\n",
  4689. "L = 9.699239, acc = 0.970000\n",
  4690. "L = 9.696903, acc = 0.970000\n",
  4691. "L = 9.694570, acc = 0.970000\n",
  4692. "L = 9.692241, acc = 0.970000\n",
  4693. "L = 9.689915, acc = 0.970000\n",
  4694. "L = 9.687593, acc = 0.970000\n",
  4695. "L = 9.685274, acc = 0.970000\n",
  4696. "L = 9.682959, acc = 0.970000\n",
  4697. "L = 9.680647, acc = 0.970000\n",
  4698. "L = 9.678338, acc = 0.970000\n",
  4699. "L = 9.676033, acc = 0.970000\n",
  4700. "L = 9.673731, acc = 0.970000\n",
  4701. "L = 9.671432, acc = 0.970000\n",
  4702. "L = 9.669137, acc = 0.970000\n",
  4703. "L = 9.666845, acc = 0.970000\n",
  4704. "L = 9.664557, acc = 0.970000\n",
  4705. "L = 9.662272, acc = 0.970000\n",
  4706. "L = 9.659990, acc = 0.970000\n",
  4707. "L = 9.657712, acc = 0.970000\n",
  4708. "L = 9.655437, acc = 0.970000\n",
  4709. "L = 9.653165, acc = 0.970000\n",
  4710. "L = 9.650897, acc = 0.970000\n",
  4711. "L = 9.648631, acc = 0.970000\n",
  4712. "L = 9.646370, acc = 0.970000\n",
  4713. "L = 9.644111, acc = 0.970000\n",
  4714. "L = 9.641856, acc = 0.970000\n",
  4715. "L = 9.639604, acc = 0.970000\n",
  4716. "L = 9.637355, acc = 0.970000\n",
  4717. "L = 9.635110, acc = 0.970000\n",
  4718. "L = 9.632868, acc = 0.970000\n",
  4719. "L = 9.630629, acc = 0.970000\n",
  4720. "L = 9.628393, acc = 0.970000\n",
  4721. "L = 9.626160, acc = 0.970000\n",
  4722. "L = 9.623931, acc = 0.970000\n",
  4723. "L = 9.621705, acc = 0.970000\n",
  4724. "L = 9.619482, acc = 0.970000\n",
  4725. "L = 9.617263, acc = 0.970000\n",
  4726. "L = 9.615046, acc = 0.970000\n",
  4727. "L = 9.612833, acc = 0.970000\n",
  4728. "L = 9.610623, acc = 0.970000\n",
  4729. "L = 9.608416, acc = 0.970000\n",
  4730. "L = 9.606213, acc = 0.970000\n",
  4731. "L = 9.604012, acc = 0.970000\n",
  4732. "L = 9.601815, acc = 0.970000\n",
  4733. "L = 9.599621, acc = 0.970000\n",
  4734. "L = 9.597430, acc = 0.970000\n",
  4735. "L = 9.595242, acc = 0.970000\n",
  4736. "L = 9.593057, acc = 0.970000\n",
  4737. "L = 9.590876, acc = 0.970000\n",
  4738. "L = 9.588697, acc = 0.970000\n",
  4739. "L = 9.586522, acc = 0.970000\n",
  4740. "L = 9.584349, acc = 0.970000\n",
  4741. "L = 9.582180, acc = 0.970000\n",
  4742. "L = 9.580014, acc = 0.970000\n",
  4743. "L = 9.577851, acc = 0.970000\n",
  4744. "L = 9.575691, acc = 0.970000\n",
  4745. "L = 9.573534, acc = 0.970000\n",
  4746. "L = 9.571381, acc = 0.970000\n",
  4747. "L = 9.569230, acc = 0.970000\n",
  4748. "L = 9.567082, acc = 0.970000\n",
  4749. "L = 9.564937, acc = 0.970000\n",
  4750. "L = 9.562796, acc = 0.970000\n",
  4751. "L = 9.560657, acc = 0.970000\n",
  4752. "L = 9.558522, acc = 0.970000\n",
  4753. "L = 9.556389, acc = 0.970000\n",
  4754. "L = 9.554260, acc = 0.970000\n",
  4755. "L = 9.552133, acc = 0.970000\n",
  4756. "L = 9.550010, acc = 0.970000\n",
  4757. "L = 9.547889, acc = 0.970000\n",
  4758. "L = 9.545772, acc = 0.970000\n",
  4759. "L = 9.543657, acc = 0.970000\n",
  4760. "L = 9.541546, acc = 0.970000\n",
  4761. "L = 9.539437, acc = 0.970000\n",
  4762. "L = 9.537332, acc = 0.970000\n",
  4763. "L = 9.535229, acc = 0.970000\n",
  4764. "L = 9.533129, acc = 0.970000\n",
  4765. "L = 9.531032, acc = 0.970000\n",
  4766. "L = 9.528939, acc = 0.970000\n",
  4767. "L = 9.526848, acc = 0.970000\n",
  4768. "L = 9.524760, acc = 0.970000\n",
  4769. "L = 9.522675, acc = 0.970000\n",
  4770. "L = 9.520592, acc = 0.970000\n",
  4771. "L = 9.518513, acc = 0.970000\n",
  4772. "L = 9.516437, acc = 0.970000\n",
  4773. "L = 9.514363, acc = 0.970000\n",
  4774. "L = 9.512293, acc = 0.970000\n",
  4775. "L = 9.510225, acc = 0.970000\n",
  4776. "L = 9.508160, acc = 0.970000\n",
  4777. "L = 9.506098, acc = 0.970000\n",
  4778. "L = 9.504039, acc = 0.970000\n",
  4779. "L = 9.501983, acc = 0.970000\n",
  4780. "L = 9.499930, acc = 0.970000\n",
  4781. "L = 9.497879, acc = 0.970000\n",
  4782. "L = 9.495832, acc = 0.970000\n",
  4783. "L = 9.493787, acc = 0.975000\n",
  4784. "L = 9.491745, acc = 0.975000\n",
  4785. "L = 9.489706, acc = 0.975000\n",
  4786. "L = 9.487669, acc = 0.975000\n",
  4787. "L = 9.485636, acc = 0.975000\n",
  4788. "L = 9.483605, acc = 0.975000\n",
  4789. "L = 9.481577, acc = 0.975000\n",
  4790. "L = 9.479552, acc = 0.975000\n",
  4791. "L = 9.477529, acc = 0.975000\n",
  4792. "L = 9.475510, acc = 0.975000\n",
  4793. "L = 9.473493, acc = 0.975000\n",
  4794. "L = 9.471479, acc = 0.975000\n",
  4795. "L = 9.469467, acc = 0.975000\n",
  4796. "L = 9.467459, acc = 0.975000\n",
  4797. "L = 9.465453, acc = 0.975000\n",
  4798. "L = 9.463450, acc = 0.975000\n",
  4799. "L = 9.461450, acc = 0.975000\n",
  4800. "L = 9.459452, acc = 0.975000\n",
  4801. "L = 9.457457, acc = 0.975000\n",
  4802. "L = 9.455465, acc = 0.975000\n",
  4803. "L = 9.453475, acc = 0.975000\n",
  4804. "L = 9.451489, acc = 0.975000\n",
  4805. "L = 9.449505, acc = 0.975000\n",
  4806. "L = 9.447523, acc = 0.975000\n",
  4807. "L = 9.445545, acc = 0.975000\n",
  4808. "L = 9.443569, acc = 0.975000\n",
  4809. "L = 9.441595, acc = 0.975000\n",
  4810. "L = 9.439625, acc = 0.975000\n",
  4811. "L = 9.437657, acc = 0.975000\n",
  4812. "L = 9.435692, acc = 0.975000\n",
  4813. "L = 9.433729, acc = 0.975000\n"
  4814. ]
  4815. }
  4816. ],
  4817. "source": [
  4818. "# use the NN model and training\n",
  4819. "nn = NN_Model([2, 6, 4, 2])\n",
  4820. "nn.init_weight()\n",
  4821. "nn.backpropagation(X, t, 2000)\n",
  4822. "\n"
  4823. ]
  4824. },
  4825. {
  4826. "cell_type": "code",
  4827. "execution_count": 9,
  4828. "metadata": {},
  4829. "outputs": [
  4830. {
  4831. "data": {
  4832. "image/png": "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\n",
  4833. "text/plain": [
  4834. "<Figure size 432x288 with 1 Axes>"
  4835. ]
  4836. },
  4837. "metadata": {
  4838. "needs_background": "light"
  4839. },
  4840. "output_type": "display_data"
  4841. },
  4842. {
  4843. "data": {
  4844. "image/png": "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\n",
  4845. "text/plain": [
  4846. "<Figure size 432x288 with 1 Axes>"
  4847. ]
  4848. },
  4849. "metadata": {
  4850. "needs_background": "light"
  4851. },
  4852. "output_type": "display_data"
  4853. }
  4854. ],
  4855. "source": [
  4856. "# predict results & plot results\n",
  4857. "y_res = nn.forward(X)\n",
  4858. "y_pred = np.argmax(y_res, axis=1)\n",
  4859. "\n",
  4860. "# plot data\n",
  4861. "plt.scatter(X[:, 0], X[:, 1], c=y, cmap=plt.cm.Spectral)\n",
  4862. "plt.title(\"ground truth\")\n",
  4863. "plt.show()\n",
  4864. "\n",
  4865. "plt.scatter(X[:, 0], X[:, 1], c=y_pred, cmap=plt.cm.Spectral)\n",
  4866. "plt.title(\"predicted\")\n",
  4867. "plt.show()"
  4868. ]
  4869. },
  4870. {
  4871. "cell_type": "markdown",
  4872. "metadata": {},
  4873. "source": [
  4874. "## 9. 深入分析与问题"
  4875. ]
  4876. },
  4877. {
  4878. "cell_type": "code",
  4879. "execution_count": 11,
  4880. "metadata": {},
  4881. "outputs": [
  4882. {
  4883. "name": "stdout",
  4884. "output_type": "stream",
  4885. "text": [
  4886. "[[0.03154963 0.97354996]\n",
  4887. " [0.30242346 0.68475421]\n",
  4888. " [0.84429554 0.17625119]\n",
  4889. " [0.04812804 0.95826417]\n",
  4890. " [0.04183504 0.96405488]\n",
  4891. " [0.80767817 0.17874873]\n",
  4892. " [0.05463129 0.94906635]\n",
  4893. " [0.83768873 0.14807047]\n",
  4894. " [0.05043638 0.95552076]]\n"
  4895. ]
  4896. }
  4897. ],
  4898. "source": [
  4899. "# print some results\n",
  4900. "\n",
  4901. "print(y_res[1:10, :])"
  4902. ]
  4903. },
  4904. {
  4905. "cell_type": "markdown",
  4906. "metadata": {},
  4907. "source": [
  4908. "**问题**\n",
  4909. "1. 我们希望得到的每个类别的概率\n",
  4910. "2. 如何做多分类问题?\n",
  4911. "3. 如何能让神经网络更快的训练好?\n",
  4912. "4. 如何更好的构建网络的类定义,从而让神经网络的类支持更多的类型的处理层?"
  4913. ]
  4914. },
  4915. {
  4916. "cell_type": "markdown",
  4917. "metadata": {},
  4918. "source": [
  4919. "## References\n",
  4920. "* 反向传播算法\n",
  4921. " * [零基础入门深度学习(3) - 神经网络和反向传播算法](https://www.zybuluo.com/hanbingtao/note/476663)\n",
  4922. " * [Neural Network Using Python and Numpy](https://www.python-course.eu/neural_networks_with_python_numpy.php)\n",
  4923. " * http://www.cedar.buffalo.edu/%7Esrihari/CSE574/Chap5/Chap5.3-BackProp.pdf\n",
  4924. " * https://mattmazur.com/2015/03/17/a-step-by-step-backpropagation-example/\n"
  4925. ]
  4926. }
  4927. ],
  4928. "metadata": {
  4929. "kernelspec": {
  4930. "display_name": "Python 3",
  4931. "language": "python",
  4932. "name": "python3"
  4933. },
  4934. "language_info": {
  4935. "codemirror_mode": {
  4936. "name": "ipython",
  4937. "version": 3
  4938. },
  4939. "file_extension": ".py",
  4940. "mimetype": "text/x-python",
  4941. "name": "python",
  4942. "nbconvert_exporter": "python",
  4943. "pygments_lexer": "ipython3",
  4944. "version": "3.6.9"
  4945. }
  4946. },
  4947. "nbformat": 4,
  4948. "nbformat_minor": 2
  4949. }

机器学习越来越多应用到飞行器、机器人等领域,其目的是利用计算机实现类似人类的智能,从而实现装备的智能化与无人化。本课程旨在引导学生掌握机器学习的基本知识、典型方法与技术,通过具体的应用案例激发学生对该学科的兴趣,鼓励学生能够从人工智能的角度来分析、解决飞行器、机器人所面临的问题和挑战。本课程主要内容包括Python编程基础,机器学习模型,无监督学习、监督学习、深度学习基础知识与实现,并学习如何利用机器学习解决实际问题,从而全面提升自我的《综合能力》。