Gushyira mubikorwa umutekano kandi munini AI-ikoreshwa na software yububiko hamwe nibikoresho bya Orchestre | B4N1

Gushyira mubikorwa umutekano kandi munini AI-ikoreshwa na software yububiko hamwe nibikoresho bya Orchestre

3 months ago · b4n1 Read time: 5 min

Intangiriro

Kwishyira hamwe kwa Artific Intelligence (AI) mubikorwa bya software byahinduye uburyo twegera gukemura ibibazo, gufata ibyemezo, no kwikora. Ariko, mugihe porogaramu ikoreshwa na AI igenda irushaho kuba ingorabahizi, gukenera ubwubatsi butekanye kandi bunini byabaye impungenge zikomeye. Muri iki kiganiro, tuzasesengura ihuriro rya AI, ubwubatsi bwa software, na DevOps, hibandwa cyane cyane ku gukoresha ibikoresho bya orchestre nka DSPy na LangChain kugirango duhuze abakozi ba AI mu miyoboro ya DevOps.

Imbogamizi za AI-ikoreshwa na software yububiko

Porogaramu ikoreshwa na AI itera ibibazo byihariye kubashakashatsi ba software hamwe nitsinda rya DevOps. Zimwe mu mbogamizi zingenzi zirimo:

  • ** Ubunini **: Moderi ya AI isaba ibikoresho byingenzi byo kubara, bishobora kuganisha kubibazo byapimye mubidukikije.
  • ** Umutekano **: Moderi ya AI irashobora kwibasirwa nibitero, nkuburozi bwamakuru hamwe nibitero byerekana inversion.
  • ** Kwishyira hamwe **: Moderi ya AI igomba guhuzwa na sisitemu ya software iriho, ishobora kuba umurimo utoroshye.
  • ** Gukurikirana no Kubungabunga **: Moderi ya AI isaba guhora ikurikiranwa kandi ikabungabungwa kugirango igume neza kandi yizewe.

Ibikoresho bya Orchestre: DSPy na LangChain

Ibikoresho bya Orchestre nka DSPy na LangChain bitanga igisubizo kuri ibyo bibazo mugushoboza kwinjiza abakozi ba AI mumiyoboro ya DevOps. Ibi bikoresho bitanga urwego rwo kubaka, gukoresha, no gucunga imiterere ya AI, kwemeza umutekano wubunini kandi bunini kubikorwa bya AI.

DSPy

DSPy nigikoresho gifungura-orchestre igikoresho cyubatswe hejuru ya Python. Itanga API yoroshye kandi yimbitse yo kubaka, gukoresha, no gucunga imiterere ya AI. DSPy ishyigikira urwego runini rwa AI, harimo TensorFlow, PyTorch, na scikit-kwiga.

Ururimi

LangChain ni ikindi gikoresho gifungura-orchestre igikoresho gitanga urwego rwo kubaka, gukoresha, no gucunga imiterere ya AI. LangChain yubatswe hejuru ya Rust kandi itanga API-ikora cyane yo kwinjiza abakozi ba AI mumiyoboro ya DevOps.

Gushyira mubikorwa umutekano wizewe kandi wagutse AI-ikoreshwa na software yubatswe hamwe na DSPy

Muri iki gice, tuzasesengura urugero rworoshye rwo gushyira mu bikorwa umutekano wizewe kandi wuzuye wa software ikoreshwa na DSPy.

Urugero: Kubaka Moderi yoroshye ya AI hamwe na DSPy

Reka dusuzume urugero rworoshye rwo kubaka imashini yiga imashini ukoresheje DSPy. Tuzakoresha isomero-twige isomero kugirango twubake umurongo ugaruka.

`` python kwinjiza dspy nka dsp kuva sklearn.umurongo_model itumiza UmurongoGusubira inyuma kuva sklearn.datasets itumiza umutwaro_diabete

Fungura dataset ya diyabete

dataset = umutwaro_diabete ()

Gabanya dataset mumahugurwa no kugerageza

X_imyitozo, X_ikizamini, y_imyitozo, y_test = dsp.split_dataset (dataset.data, dataset.intego, test_size = 0.2)

Kubaka icyitegererezo cyo gusubira inyuma

icyitegererezo = Umurongo ugaruka ()

Hugura icyitegererezo kumibare yamahugurwa

icyitegererezo. gikwiye (X_imyitozo, y_imyitozo)

Suzuma icyitegererezo ku makuru yo kwipimisha

ubunyangamugayo = icyitegererezo.ibisobanuro (X_test, y_test) icapiro (f "Icyitegererezo cyukuri: {ubunyangamugayo: .2f}") `` `

Iyi kodegisi yerekana uburyo bwo kubaka imashini yoroshye yo kwiga imashini ukoresheje DSPy. Dupakira dataset ya diyabete, tuyigabanyamo amahugurwa no gupima, twubaka icyitegererezo cyo gusubira inyuma, tuyitoza kumibare y'amahugurwa, kandi dusuzume ukuri kwayo kwipimisha.

Gushyira mubikorwa umutekano wizewe kandi wagutse AI-ikoreshwa na software yubatswe hamwe na LangChain

Muri iki gice, tuzareba urugero rworoshye rwo gushyira mubikorwa umutekano wizewe kandi wuzuye wa software ikoreshwa na LangChain.

Urugero: Kubaka Moderi yoroshye ya AI hamwe na LangChain

Reka dusuzume urugero rworoshye rwo kubaka imashini yiga imashini ukoresheje LangChain. Tuzakoresha isomero rya TensorFlow kugirango twubake imiyoboro y'urusobe.

`` ingese Koresha langchain nka lc; koresha tensorflow :: prelude :: *;

// Sobanura uburyo bworoshye bwurusobe rwimikorere fn icyitegererezo (icyinjijwe: & Tensor) -> Tensor { reka ibisohoka = ibyinjijwe.matmul (& Tensor :: shyashya (1, 1)); Ibisohoka }

// Kora umukozi wa LangChain reka umukozi = lc :: Intumwa :: shyashya (icyitegererezo);

// Sobanura imibare reka dataset = lc :: Dataset :: shyashya (vec! [ lc :: Urugero :: shyashya (vec! [1.0, 2.0], vec! [3.0]), lc :: Urugero :: shyashya (vec! [4.0, 5.0], vec! [6.0]), ]);

// Hugura icyitegererezo kuri dataset umukozi.imyitozo (& dataset);

// Suzuma icyitegererezo kuri dataset reka ukuri = umukozi.gusuzuma (& dataset); println! ("Icyitegererezo cy'icyitegererezo: {}", ukuri); `` `

Iyi kodegisi yerekana uburyo bwo kubaka imashini yoroshye yo kwiga imashini ukoresheje LangChain. Turasobanura uburyo bworoshye bwurusobe rwimikorere, dushiraho umukozi wa LangChain, dusobanura imibare, duhugura icyitegererezo kuri dataset, kandi dusuzume ukuri kwayo kuri dataset.

Hejuru 5-10 Amabwiriza Yakoreshejwe Byinshi / Uduce

Dore top 5-10 zikoreshwa cyane / ibisobanuro kuri DSPy na LangChain:

DSPy

  • dsp.split_dataset (amakuru, intego, test_size = 0.2): Hindura dataset mumahugurwa no kugerageza.
  • dsp.build_model (icyitegererezo_izina, icyitegererezo_params): Yubaka imashini yiga imashini ukoresheje izina ryicyitegererezo hamwe nibipimo.
  • dsp.train_model (icyitegererezo, amakuru): Itoza imashini yiga imashini kumibare yihariye.
  • dsp.isuzuma_model (icyitegererezo, amakuru): Isuzuma imikorere yuburyo bwimashini yiga imashini kumibare yihariye.
  • dsp.deploy_model (icyitegererezo, seriveri_url): Kohereza imashini yiga imashini kuri URL yerekanwe ya seriveri.

Ururimi

  • lc :: Intumwa :: shyashya (icyitegererezo): Kurema umukozi wa LangChain ukoresheje icyitegererezo cyagenwe.
  • lc :: Dataset :: shyashya (ingero): Kurema dataset ya LangChain ukoresheje ingero zerekanwe.
  • lc :: gari ya moshi (agent, dataset): Itoza umukozi wa LangChain kuri dataset yagenwe.
  • lc :: gusuzuma (agent, dataset): Isuzuma imikorere yumukozi wa LangChain kuri dataset yagenwe.
  • lc :: ohereza (agent, seriveri_url): Kohereza umukozi wa LangChain kurubuga rwa URL rwerekanwe.

Kugaragara

Hano haroroshye kubona amashusho yububiko bwa software ikoreshwa na Chart.js:

javascript const ctx = inyandiko.getElementById ('myChart'). getContext ('2d'); imbonerahamwe = imbonerahamwe nshya (ctx, { ubwoko: 'akabari', amakuru: { ibirango: imibare: [{ ikirango: 'DSPy', amakuru: [10, 5, 90], inyuma Ibara: 'rgba (255, 99, 132, 0.2)', umupaka Ibara: 'rgba (255, 99, 132, 1)', Umupaka Ubugari: 1 }, { ikirango: 'LangChain', amakuru: [8, 6, 92], inyuma Ibara: 'rgba (54, 162, 235, 0.2)', umupaka Ibara: 'rgba (54, 162, 235, 1)', UmupakaUbugari: 1 }] }, amahitamo: { umunzani: { y: { gutangiraAtZero: ukuri } } } }); `

Iyi kodegisi ikora imbonerahamwe yumurongo yoroshye ukoresheje Chart.js kugirango ugaragaze imikorere ya DSPy na LangChain.

Umwanzuro

Muri iki kiganiro, twasuzumye ihuriro rya AI, ubwubatsi bwa software, na DevOps, twibanze cyane ku gukoresha ibikoresho bya orchestre nka DSPy na LangChain kugirango duhuze abakozi ba AI mu miyoboro ya DevOps. Twerekanye uburyo bwo gushyira mubikorwa umutekano kandi wapimwe AI ikoreshwa na software ikoresheje DSPy na LangChain, kandi dutanga urugero rworoshye rwo kubaka imashini yiga imashini ukoresheje buri gikoresho. Twatanze kandi top 5-10 ikoreshwa cyane mumabwiriza / ibisobanuro kuri buri gikoresho hamwe no kubona amashusho yoroheje yububiko bwa software ikoreshwa na Chart.js.

Have Questions?

Let's Talk

Contact me on LinkedIn