Scribd
Upload
13 views23 pages

NNDL Lab Record

Record work

Uploaded by

nareshamgoth04
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
13 views23 pages

NNDL Lab Record

Record work

Uploaded by

nareshamgoth04
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
pots : QIBGIAGGRA [Link]-_OL a geting OP “the Spysien MDE envionment and executing Python Progyarm ingiall golution : rheve adie 3 voays +0 — U ype System methocd -1 + UBing Anaconda Tnstal) Spyder TDE oT) —the Anaconda - = navigator 1 open he Anaconda novigatov application 2-40 to the Envivonments tal 4 Select the environment where you want tO iwSial) Spyies 10 & seosch for 'SPyelev’ ty the Seasich loo) fb click on he “ingtaiy! button newt 10 Spyder dy mSiall t+ Method-2 : TnSto\\ SPyciev TPE using Pip, the Python Package - eras) a * OPen the —teymival OY command PyOmpt Avanthi Institute of Engineering & Technology Gunthapally (V), Abdullapurmet (M) Near Ramol Filmolty, RRDist, Hyderabad Command -to ‘mn a. Tyre the —fottowing TOE’ pip tnsiall Pychev 3. Once the ISHallation %3 complete, you Spystev TOE Prom ~he -texminal oy 0 prompt by rq ‘Spydey Method -3 : Frm he of ficial Spyder PDE « Pu can also download and isiall Spyder FOE dhecily trom +he official website . 1. Go-40 +he website at WAEPS sf frocow9. Spycier -ide pag |. a click on +he dovonlvad button 3 Select your o5 trom +e dropdown menu fe I. chooSe appropriate § vesision of spysier Epe oy gow Sj a bP. Download the ‘nstallaton Pile and wun tH eestg) 2 Pysiom: program : kgo to ‘File’ >'new rile’ +0 create a new Python - - gevipt write Some code in tre Script edttoy 3 Sow the Script +2 a file wrth —he “oy * ex4enSion | tor ex, you can Gove it 8 “[Link]® . of Engineering & Technol. Fvantht intizate of Foatneering & Technology roto. + 6A 1 Exot. No. ly TO YUN —the Script, 90 to preeess Fir. this ‘Run*= “Run he file” oF Skewte he pyrron Sevipe and alispley the output rer) in the console tnchividual Js. gor can also use the console -to Yun individua hnes Of code. ro do his, type dhe code in the console ond press entey “the output wih be Aisplayect th the console Trograrnme y Rame= “vivee Nome pint CHello student 2) mint C’ yous home : 3 Name) Engineering & Technolo Avanthi Ee Gnas ne cere Gunthapaly (0 mano > 6680 Expt. No 2s anstalting Kevas, Tengorflow anc Pysorch braves ond mayen ge of hem solution ™ Stalling Kesag GER? G00 official Ine www Python. 0 SHep-8: Cate vidual environment e:\=pip Install Kexag inside he eP-3: Ackvate the envivonment by movie command "Kexasenv' foldey and type the be loo s\env\ sevipts \netivate Anstall —rensortiow. ; rensorPioo 78 one of he we need ~10 insiall before Keres “an be Instatled backend engines —thot 1; TYPE Command create a virtual environment cilomkdiy +egt C:\> cd 43 ci\> pip insta\l +ensorflow sep-a, Type command to eveote A vivtual envivin- -ment. with Python Python-am venv tf —vittual env PEP-3: “YPC commond +o actvateg -rhe environment _ source +£- virtual — env | bin] actvate Avanthi Institute of Engineering & Technology Gunthapaly (V), Abdullapurmet (M) Near Ramoj Filmeity, [Link], Hyderabad TYPE command +0 fnStall Tensorfinw without PO Support CN> PIP tnstaly - upgrade tensorflovs use the Same command for upderting rensorf low. ( Een eseictie ee A Envivoment by the +yPe Ones to +egt yous environment, open EN Python bash ; CV > Python Type command fox import ~the senor flow Packages into the. python. impovt enSorfious ag tf Hype command Know versidn Piint Cveusion of Tensorfiow ¢ 4 exsion ~) Python leis instaijoton the otowsing PyHhnon Nbra vies + ‘Avanthi Institute of Engineering & Technology Gunthapally (V), Abdullapurmet (M) Near Ramo! Fimoty, [Link]., Hyderetad proce SS > S440 Command line sv iMSlall a new Itbyaxy that you tyP2 the -fotowing os ‘y He Bersih command C:\SPHthon—m pip inslall enlame Replace name pee Ey ey) which ingial. '. ingsal na ©\ pusnon mn opip install HY [Link]\l pandas - = ndas c:\> python -m Pip ‘he Mmstal 3. Tngiall Scfiert - leosin e:\> python —ro pip ansiall Seiki —leasn , Tngtait motplotlilo c:\> python om Pip insiall mot plod egera SC ¢:\> Python —™ pip Ynstal) seipy iin) Seakon ci\= python pip insall Sealovrn ‘Avanthi Institute of Engineering & Technology Gunthapally (Vj, Abdullapurmet (Mt) Near Ramoli Filmy, [Link], Hyderabad quinoa» 668% Soto 03. ot Gj roriying “the convolution Neusal network on vision proiolems Solution: Sep-1 Import the neccesasy Nabrasies import mumnfoy ag np import esas trom e108. datasets import mnhist Prom xesia8 « mmocelgtenport rroctel. * Hepa: Creole the toain data ana est ote ist. load - Cx brain, Ytrain), Cx test Y- eg) = a's datat > Step-3 : checking data format ng ~ Tous iva) —colg= 23,88 — 4K image - cata ema! Ge eechanrels— first xeaoiin ex —-brotn reshape Cxtra%n. shape C01,1, 'mg vows, img cols) : x tege =X-test. veshape Cx test Shape COI 1, Gere nag col) input = oa 70095, emp _col, i) else. Yetrain= - X ABE fnput = ‘Avanthi Institute of Engineering & Technology Gunthapally(V) Abdulapurmet(M) Near Remoj Fimely, [Link], Hyderabad ‘Float 3a’) Leste = X-*BL. agtuny ¢ ain = x- ain. astupe ( "Float aa’) een |= ane Legd = 65 tain = Kexo$. utils. to — categurical Cy-tr0%n ) a y. ¢ on d 4eje = KeNAZ. US. -L0 ~ categorical (y-test) d- d j inp x = input Cshape =tnpx) foyer! = convaD (sa, kexnel Side =(39), achvertion - 5) Gnpx) aay Chonan ™ layess & = convad (6%, (3.3), achuation = "10 peeetey loye73 = moxpooling a0 C pool —siae = (23) Clayen a) Noyes ¢ = Dvopout Coe) Clotes) 3) joyes 5 = Hatéen() Closes: layes 6 = Dense Camo achvation =' si oD) Cay 6) oye 7 = Dense Cio achuetion - "s04trrox') Clays 6) Step -4 : cag compile and f4 Purckon | ede} — mode) =rmode) (Linput x), lays 4) J des. compile CopHinides = Keres, ophmi aed. Adacoitenc) fo: = i SS = KeHa8. logges, Ca Aeqorical —Cr0sen —y, ‘Avanthi Institute of Engineering & Technology Gunthapally (V), Atdullapurmet (M) Near Ramoii Filmeity, [Link], Hyderesten metyvics Laccusia Poceusiacy ) procel «fi Cain yetrain, epochs=!a, latch - Size function : Glep-® + Evaluate vesilo' Geox =model. evaluate Cx teste y-test- Pint C'v0ss =', score £6) Print Caceusacy =! scove fii) ) thi Institute of Engineering & Technology 1 (0), Asidlapurmet(M) Nr Ramo Fmt, RRO, Hyder page clossifrcation on MNIST dataset Conn model ery Pull connected ayes) wrth d q solution : sep Tmport the necessany librowies port Mumpy ag np tmport tensovtlow 3 : tmnpor $ Jyom tensor flow. Kesas 12, motels, eaaaes Wnport loyes SeP-B : Wad she Pre procegg the MNIST Cataget Coin. troges train labels), Cert images, 434 labels) = dotosets, mnist. Wad —datat) -tyain- tages , ees — i one images (awh.o E3t— fmnages | ant. ain. images = np. expand — dims(train- images, oxig=-1) test images =p. expand — dims (test - images, axis =-1) SP-3: Build the cnn mode) model = mdclels. sequential( ) roodel add Cloyes13. convad (sa, (33) acHvation = ‘elu, input - Shape = Cag, ag.) 2}. add Clayess. moxpooling anCCa, a) Medel. ade Cloyens. convai) (64.(33), actvertion ='7e1u’)) rset. ada Cloyers. Hattent }) ‘Avanthi Institute of Engineering & Technology Gunthapally (Vj, Abdullapurmet (M) Near Ramoli Fimcity, [Link]., Hyderabad ave 9 model: cla Clayess. Nense (oy, ackvedion mode! aca ( foyens. Dense Civ, ackvaton Limax’) §4ep-4 s compile -the mode} model. compile Cophmigesr= ‘adam! 1055 =! Spas Crosseneropy), metrics = [ ‘aceusracy’) ) sep-# “ain —the mode} model. f+ Cerin images, Aroiin- labels, epochs =m, batch Side = 64, valiclation— split = 0-1) SHep-6 : Ewluote he mpde) onthe test Set 4054-1085, fe3t Saccwsacy 2model. evaluate Cxest - *MAGS e384 _ label) print (“Test aegha aon. ert — aceusiaey ) Sep-4 : mare predictipn3 On new eee prediction’ = model. predict (-tege— images fira]_) Predicted — labels —[Link] (preclickons, oxis =i) Prin C* pyedlicled tabel4:" predicted - lavels) Print C “TwWe foloel4." tegt_labels [:m)) int (* predicted labelg:* predicted —labels’) Pink CrTvUe labels:” test images [:n')) ‘Avanthi Institute of Engineering & Technology Ginthapaly(¥),Abilanmot (a) Near Ramo) Fimdty, RRLDist, Hyderabad pegcrip tion = rma clogsiticaton mist clatoset onthe box Staorting point nleusal example and a great how to butld a convolutonal a classic undies Stancting network CenN) with a fatty connected tay ecog nrbon ‘Avanthi Institute of Engineering & Technology ‘Gunthapally (V), Abdullapurmet (M) Near Ramoji Filmeity, RLR Dist, Hyderabad pane: 66a 0% Expt Ne G) -aerlying ee eep ee ral nine eee of NUp ¢ ae oe include Degcviphon § deep leosining models see | t Recwirent Neusial Networks Cenns) . RNNS ae clesigned to handle sequent) where axé data, mang them wey guided for Ntp to3KS she wcler of words matted. P7100e33 input dot hidden Sats Step-by-step media while men to caplune — context Spl: RNNS RNNS oie ates gree to handle Sequential data, TKD then well - Suitect Por Nip to3kg whee ae the orces of words matte 18 Applications : -branglotion Cea, -qoogke!s transfor mes heZec} mode) ‘Bem’ for Nim) Tee generation (eg, openaz'S ep Eee) > Question Efrsosing Ceq. 4pm¥lely Beer tage model "BERT Joy QA) ‘Avanthi Institute of Engineeri cunt ee, for glep-A> Bidlivechonal ENco« Jey Repregentahongs rransPormen3 CSERD Beer ts a -brans-forvrres -boged langue angua pretrained ona lasige covpug of ee sTeet Completion sent geneiaton Ce gy eveative cori Hing code yenes— > lang uage translation S1ep-3: Convolutional neusja) nletwork3 boy Nup: Athvough move) Commonly used Pox Computers i530 CNNS con be adapted +0 Ntp ta8ks Applications : saeet cla9sificakon > Sentiment analysis SEP: Sequence —-to- Sequence models~ Sequence —to- Sequence models uge encodey _ de coder omchiectwies +0 handle bans forming one Sequence fato anotres;, Such 03 machind ranslehon and Summosigotion -ka3k3 Yet involve ‘Avanthi Institute of Engineering & AUT oracle mene applicakons » smaochine -bhansletoo > Tevet most aalion SHep-8 2 Attention mechorisrs: atiention mechani8m§ adie not mode in improv bat reg have been fast rument modelS performance Of vosiioug Nip ta3K3 and of the allo modelS of —focug ory specific posts © they uindesHand 9 herr cusing pwoees Sing enhancing uncexSianding and pertermance -spgcoiong > language translation > eet GEosiabon Avanthi Institute of Engineering & Technology Gunthapaly(V), Abduilapurmet (tM) Near Ramo Fimoty, RRDIst, Hyderabad TrAIN A Sentiment analygtg model On Imoe 0 ae uge ANN Layeig with Ls™|4aev Notes Solution A Vibvosie8 Step! Impoye the necessary import +ensorvfiny as +f mbd. ¥ from tensorfiow - Kesa3. datagets *mport Step-a: Load and preprocess the M06 dataget nurn - words =1000 Cerin - data, train label3), (4egb— data. +3 mE jabelS) = e bg) feb load data Crum words = nure—107 max -length = 2h0 -hain- data = pad —Sequennces (Cerain_ data, moxlen ae max - length) +egt— data =pad_ sequences (4eit- data, maxlen = max length) Hep-3; Build +he eNN model with LS™ oy Gav aes embedding _ dim=100 model - sequential ¢> ‘Avanthi Institute of Engineering & Technology Gunthapaly(V), Abdulapurmat(M) Near Ramo} Fin, RR Di Hyder output — model. add Cembe: iding Gnput -dim = num word S, ¢ aim, inpuk -length = max—length)) modal. add CqQvCunits =6u, dropout = 0-8, TECUmren ®= yo pout = 0.3) ' gi grnot model. ada (pense (unit$ =i, ackvaron= 9 Slep-u = compile ancl rain he model batch-dSiae =1a8 epochs =% wmode). -# (train. data, Yygin_labelSbatch-Si ge epochS= epochs vali dahon=split a) =b-s Step: Evaluate the mode] on the test Set 4e3t loss te3t— aeeuracy ‘=model. evaluate (test -data, tegt— label 3) Wink (-1e5L [Link]:', te5e aceusacy ) eaciption 0 train a senkment avalysis mode) onthe MDG dataset using RNN ones 8 with Lsm/qao Und, vel} use FeENSflgvd and Keegy. the IMDB dotoset Conai movie yeviewS labeled 03 positive OF hegertive Sertimend ‘Avanthi Institute of Engineering & Technology Gurthapaly(V),Abdulepumet(M) Nr Remo} Flay, RARDist, Hysrabee Applying the nutoencocdes aleorritfoeng fog encocting the qeal-worlad cata Solution : Stepas colect and prepyocess Data preparation data you he veal world Y : a ane data *5 want to encode. engusie that the ae As ee 5 sulable —Brmet and novmalae tt d awe ye He Step-a : Quild +he autoencocles mode! Cea . ines encode? model using a neusial networ WKe tensorfiow and Kees import tenSoyflow a8 tf , from tensor Piovo .kesja8. \oyesis Import Jopude, DENSE. * ‘npul- Shae = < yous, input - Side > fi a ; t_ ctimension > latent dim = modeladd (nense(ainé. inpub- dim= latent - mode}. ada Cleakyret v Calpha=o.0D) mide). add (atch novraligeton Crnomentum 0.8) model. add Coense (m3) model add Cleaxyeery Catpha=0.01)) mode}, adel (nense Cioa4)) wrode). add (Dense Cagxa grt, achvaton tanh!) model. ace (ReSpenseshape (Casas) euHN mode) dim) ‘Avanthi Institute of Engineering & Technology Gunithapally (Vj, Abdullapurmet (NM) Near Ramo} Fimely, [Link], Hyderabad deh utia — digeviminatoy Geng shaped mode | = Sequensal C y ie wondelvaclel Coden Cinput shape = fing shape)? mode), add (pense (ma) vnedels add Creakyeerv Calpha = 0:00) wuecelell Gena (ane) model ach) Crense C1, ackveton="5 yetsin mode} Bay moid')) Step-3: cefine the -cAn mode} by combinivy) the Generator] and discriminator def build _gan Coenesiator. di seriminatoa)* disoiminetes trainable = False rode] =Sequentall ) mode! x. add Cgenesiator) model: add (discriminate) fetun rvcie! Sep-y : ayain the -GAN mode) # Detne hypesipediometen3 latent dim=100 pa Shape = CAs, 28,1) gendtortor = build - generator Coden _ dim) Asia = duilde disevimimatey Cimg- shape) Avanthi Institute of Engineering & Technology Feary (V) Abdulapurmet (M) Neat Ramo Ficty, RRLDIt, Hyderabad 4Taiwing loop 0 epochs = ior batch- Sige = 6y half - botch = batch-siae Ia fey epoch in yange Cepochs): Tan he cdisoiminatoy tak [Link].-vandlint COx_-[Link] £6] balk betel “Veal inages = gen x bain Lid] ulSe = npesonclorn. netenaiNCUNMeRapp. batch, latent — aid) genwiated - *mages = pensar: predict Cnoise) 1085-70) = digeyiminarter. tevain —on- batchCrea|- images np. ones(ChalP batch, 1) d_toss- fake = discriminatiy. Erain—on-botch Cgensiated z images, npr ze08 Chal? baich.'))) d_losg= 0: * Mp. add (a-tosg_ rea el — loss—faxe) # trdin the generator 1n03S@ =np- Yandom. rormal (01, Chetch_ S1ae. lent dim)) vals ~ labelS =np.0ne3CC batch, $i Ae.1)) 91055 = gan,-tvain- on- batch Choise, valid —\akel 8) ‘Avanthi Institute of Engineering & Technology Gunopaly(V), Abana (h) Noar Remo Fie, RR Ds, Hyderabas ription ° a a pours Newoorks CaAns) ae a Powel Gendietve -Aclves Sia ful ebags of cleep leosning ynvciel 08K, Incluctng image genesrtion me aged for vosiiou$ ised AS and unsupe leosinin)) Institute of Engineering & Technology Gunthapally (V), Abdullapurmet (M) Near Ramo Filmely, [Link]., Hyderabad

You might also like