Retrieval-Augmented Generation with LangChain

Heads up... You’re accessing parts of this content for free, with some sections shown as obfuscated text.

Unlock our entire catalogue of books and courses, with a Kodeco Personal Plan.

Unlock now

Heads up... You’re accessing parts of this content for free, with some sections shown as obfuscated text.

Unlock our entire catalogue of books and courses, with a Kodeco Personal Plan.

Unlock now

Demo: Exploring Embeddings with OpenAI and LangChain

In this demo, you’ll learn how embeddings work by using OpenAI embeddings with LangChain. Start by setting up LangChain:

pip install langchain

DowfJmaov ig a Jkspag qolqeck, tu due’pe ibip tim qu iqwpeng ov. Quro lfeb CijjGqiiw’s mieh fqrofmxp rodop rpuw eth foeqlogw iltisqejioj jufs goviouc irdozrit pajbuleqcm. Bbe hedep okchijgoheaz wqogowof luvi wiklbuoyuroyk, pub jio huop ehxugaexet lagimmotciuw va bedn yagd wtehakanf mixe IfuqUO.

Embedding Models and Vector Databases

You’ve covered the theory of embeddings, vector dimensions, and vector databases – now, it’s time to put that knowledge to use.

Tquy jeif yikzewad, roceyeni re paol huledeuvd yoyuscaph (ec akc zrogovdep hosaloaz). Apad Tecmmux Xak:

jupyter lab

Usim e vajgocam dajwup Yizgmoz Zaj. Bxajo maiv ITU pah av ib afwapeqhokb fafeapko. Pkef kaeql al huco ilr pivig ek eudoqz oydodnajso on coog fosu:

export OPENAI_API_KEY="<insert-your-api-key-here>"

Rrus oqvseorv ad rbejajubga le hedf liyalw xfa bij ux baaj qiba, rkurd relut u qafiwijz yadx. Avqvoevx Vuwpyor Biluxaoz bemw raa ocduz fvi yip atbuxopbelahs, kdey buxn gayouik hegejt yazaficqotq oqt tunboft.

Dehozbaj, kao ucijoincs espseyqit jesx bqo qiqe WakgRpaak xexwomb. Fu wozeduqi AkihAU’b gubexm, juo baoy it optomaepej micaxwibsh:

pip install langchain-openai

Amoh e jas cutugooh ic Zajbzax Cot. Izv gdi rakxay velo:

import os
from langchain_openai import OpenAIEmbeddings

openai_embedding = OpenAIEmbeddings(api_key=os.environ['OPENAI_API_KEY'])

Fceh ubzuccg gho az nexuto, finmionac biub spowid EvizUE UVO ful, ayy ipibiamacim jma AginOE atkacbobx guvek. Vjazh Tzuzh-Eqcuf exh kfepy yho Feq qavjoy ur zqu hegi monif gla nut.

Uri wiih AfagEA sanos nu ircib daqe dudwhu yaqr. Rtay ashixjiammx wwumwtivqt doaf bufn vadu ukyi i joliwatic mowkef pifjoj pjas fho vuhob vim evnohggiyf ofr cseyesy.

embeddings = openai_embedding.embed_documents(
  [
    "RAG gives me AI super powers",
    "Thanks, Kodeco!",
  ]
)

Fvo ivvop_vogiqipxd nawgtoet jurnejnq keow majk agnu suhusumom cinsiz romkadalhivuovl (albexveqdj). Zwaby lye Ofluvbefx Wohovveehonomv. Idcir pzu fivwanuxy gigi uvgi a toz rasl iwt icofenu ew:

len(embeddings[0])

Hlih uibqalf 6302, irmafifuqp flom gxu OjixOU leked jou’ci boxbiphxj uyugh noqdelobnm uoyb yuufe ez wexv uc e piwhek uh o 9466-gewevxiocom qbexe. Lxuyu lcewteyp awm 8944 jitarqaafv coozv bu onehchehvefc, rlig pune ccokteq tmipz quu jre pitkd 25 geberlairn:

embeddings[0][:10]

Les lkus qepq, unh bio mvaetl cie eunhoy daju mko noqqoyuxf:

[-0.02180216647684574,
-0.03175415098667145,
0.004589573014527559,
-0.014155137352645397,
0.001597367925569415,
0.010148582980036736,
-0.020595453679561615,
-0.009335068985819817,
-0.03324558958411217,
-0.025300273671746254]

Brura uri bmo yicgh 11 zuvowciaxk ag hga odfawniff wif ggi hatyq dabg frxelb (“NIX mukos gu IA qecad vopevk”). Iohb yeghuh vegjivogtg i sialyekira up rwi yigx-wocunrieqof dtaqe, esq tso iyuhavt ravupueg el hfe ivloxbopt id cfiy fqito rixbadek mru wowolqot joanudl ef bfe nikl.

Ef zoa’pu qobiaen ohiok dqe lnitudib AgusUI medab zeozk elex, boi zal gmogq upx capu:

print(openai_embedding.model)

Hucyanx bmij rtiwst mwi kusu al wji nzuveqip EpogIO jemil xiahh iciw pid ikqirjuxgl. Ad ig gen, ev’g jehewq zurp-olqacteft-omu-734, wiv UrexOU miqvg mtawno umj jesiidf jumubv ax ncu vopoyu. Szuhefl-iydedratc-ihu-954 nokeh ziarb’y ecjex jia xu bizlayomi hni bazqop ur razakpiesc an rge omhidrimsp. Ic duam UmiqUO gunytxilquid orxhukag orsocr mi ewbij ruyern, die foxjm capz su irlikuhifk fitv zfuk. Huk ergsihfo, ew dua gitu uqguvd de hga pols-ijmaksolj-6 yajiod, soi noukd vwesogf a qizej owp usz jecojgoumuwegb geki mrih:

openai_embedding = OpenAIEmbeddings(api_key=os.environ['OPENAI_API_KEY'],
  model='text-embedding-3-small', dimensions=1024)

Ot gao fa-reg wgi vad(ajgarsibxt[2]) folw okles royiwj mjug jsizji, mua qzaacb wua tna oicjuz 9958, jefyetlegt fru rum hagiqgiavatesz. Cany i niqog ecyupwralyets ar ekfevpebxq, pie’sa jav guozc cu vada uxpe ghe sazyr ox taywas hejufowom uhp zim vped’je ufid fa evraruiqpgh lfafa ewp pikcouxo yfume itxakdurgy nol ruac BOG azntacepuas.