Kuronga AI Agents neLangChain uye DSPy yeEnhanced Software Architecture

Kuronga maAI Agents ane LangChain uye DSPy yeEnhanced Software Architecture

Introduction

Mumamiriro emazuva ano anomhanya-mhanya edhijitari, masisitimu esoftware ari kuramba achiwedzera kuoma, nekubatanidzwa kweArtificial Intelligence (AI) uye Machine Kudzidza (ML) kugona. Nekuda kweizvozvo, kugadzira scalable uye kusimba masisitimu anogona kunyatso kurongedzera maAI vamiririri kwave kunetsa kune software mainjiniya. Ichi chinyorwa chinoongorora mharadzano yeAI, software engineering, uye orchestration maturusi ekugadzira magadzirirwo esoftware anobudirira, aine tarisiro yekushandisa maAI vamiririri vachishandisa LangChain neDSPy.

**Kudiwa kweAI Orchestration **

Echinyakare software masisitimu anowanzo kugadzirwa ane monolithic architecture, apo ese maficha akanyatsobatanidzwa uye anoshanda pamwechete mumutsara fashoni. Nekudaro, nekuuya kweAI neML, masisitimu esoftware ari kuwedzera kugoverwa uye kugoverwa, nevamiriri vakawanda vanoshanda pamwechete kuti vazadzise chinangwa chakafanana. Izvi zvakakonzera kudiwa kweAI orchestration, iyo inosanganisira kugadzira masisitimu anogona kunyatso gadzirisa uye kuronga kudyidzana pakati peakawanda maAI vamiririri.

**LangChain: Chimiro chekuvaka AI Agents **

LangChain iPython framework yekuvaka AI vamiririri vanogona kupindirana pamwe chete uye nekunze masisitimu. Inopa seti yezvishandiso uye maraibhurari ayo anogonesa vanogadzira kugadzira vamiririri vanogona kufunga, kudzidza, uye kujairana nekuchinja nharaunda. LangChain yakavakirwa pamusoro peRust programming mutauro uye inopa yakakwira-level API yekuvaka AI agents.

DSPy: A Framework for Orchestration

DSPy chirongwa chePython chekuvaka masisitimu akagoverwa anogona kuronga akawanda maAI vamiririri. Inopa seti yezvishandiso uye maraibhurari ayo anogonesa vanogadzira kugadzira uye kuendesa scalable uye inotsiga masisitimu ayo anogona kugadzirisa kupindirana kwakaoma pakati pevazhinji vamiririri. DSPy yakavakirwa pamusoro peApache Kafka messaging system uye inopa yakakwirira-level API yekuvaka yakagoverwa masisitimu.

Kuita maAI Agents neLangChain uye DSPy

Kushandisa AI vamiririri vachishandisa LangChain uye DSPy, tinofanira kutevera matanho aya:

Define the Agent Architecture: Tsanangura mavakirwo emumiririri weAI, kusanganisira zvinopinda, zvinobuda, nekudyidzana nevamwe vamiririri.
** Shandisai Agent Logic **: Shandisai pfungwa yeAI mumiririri achishandisa LangChain, kusanganisira kufunga, kudzidza, uye kugadzirisa maitiro.
Kuronga Kudyidzana Kwemumiririri: Shandisa DSPy kuronga kuwirirana pakati pevamiririri veAI vakawanda, kusanganisira mitemo yekukurukurirana uye nzira dzekuchinjana data.
Dhikoira sisitimu: Tumira sisitimu uchishandisa gore papuratifomu kana kuti mudziyo wekutakura zvinhu zvakaita seDocker.

**Code Muenzaniso: Kuita Nyore AI Agent neLangChain **

Heino muenzaniso wakapfava wekodhi unoratidza maitiro ekushandisa mumiriri weAI uchishandisa LangChain:

import langchain
kubva langchain.chains import LLMChain


mumiririri = langchain.Agent(
    input_schema="text": str},
    output_schema="text": str},
    cheni=LLMChain(
        muenzaniso="davinci:instructix",
        max_length=2048,
        tembiricha=0.7,
    ),
)

# Shandisa iyo agent logic
def agent_logic(input_text):
    output_text = agent.run(input_text)
    return output_text

# Edza mumiririri
input_text = "Chii chinonzi revo yehupenyu?"
output_text = agent_logic(input_text)
dhinda(output_text)

Uyu muenzaniso wekodhi unoratidza nzira yekutsanangura mumiriri weAI uchishandisa LangChain, kusanganisira yekupinza uye kubuda schemas, uye pfungwa yemumiririri. Mumiririri anoitwa achishandisa Huru Mutauro Model (LLM) cheni, inova mhando yeAI mumiririri anogona kufunga uye kugadzira zvinyorwa.

Pamusoro 5-10 Inonyanya Kushandiswa Mirairo/Zvimedu zveLangChain uye DSPy

Heano epamusoro 5-10 anonyanya kushandiswa mirairo/zvimedu zveLangChain uye DSPy:

langchain.Agent(): Inogadzira chiitiko chitsva cheAI mumiriri.

mumiririri = langchain.Agent(
    input_schema="text": str},
    output_schema="text": str},
    cheni=LLMChain(
        muenzaniso="davinci:instructix",
        max_length=2048,
        tembiricha=0.7,
    ),
)

agent.run(): Inomhanyisa mumiriri weAI neinopihwa.

output_text = agent.run(input_text)

DSPy.create_cluster(): Inogadzira sumbu idzva remanodhi ehurongwa hwakagoverwa.

cluster = DSPy.create_cluster(
    nhamba_nodhi=5,
    node_type = "mushandi",
)

DSPy.create_agent(): Inogadzira chiitiko chitsva cheAI mumiririri wegadziriro yakagoverwa.

mumiririri = DSPy.create_agent(
    sumbu=sumbu,
    mumiririri_rudzi = "LLM",
)

DSPy.deploy_agent(): Inotumira mumiririri weAI kune yakagoverwa system.

DSPy.deploy_agent(
    mumiririri = mumiririri,
    sumbu=sumbu,
)

DSPy.get_agent_status(): Inotora mamiriro emumiririri weAI.

chinzvimbo = DSPy.get_agent_status(
    mumiririri = mumiririri,
    sumbu=sumbu,
)

DSPy.update_agent(): Inogadziridza mumiririri weAI nedata idzva kana paramita.

DSPy.update_agent(
    mumiririri = mumiririri,
    sumbu=sumbu,
    new_data={"text": "mavara matsva ekuisa"},
)

DSPy.delete_agent(): Inodzima mumiririri weAI kubva kune yakagoverwa system.

DSPy.delete_agent(
    mumiririri = mumiririri,
    sumbu=sumbu,
)

DSPy.create_chain(): Inogadzira cheni itsva yevamiririri veAI yegadziriro yakagoverwa.

cheni = DSPy.create_chain(
    vamiririri=[mumiriri1, mumiriri2, mumiriri3],
)

DSPy.deploy_chain(): Inotumira ketani yeAI agents kune yakagoverwa system.

DSPy.deploy_chain(
    cheni=cheni,
    sumbu=sumbu,
)

Kuona

Heino ratidziro yeLangChain uye DSPy architecture uchishandisa Chart.js:

{
  "type": "pieChart",
  "data": {
    "labels": ["LangChain", "DSPy"],
    "datasets": [
      {
        "label": "AI Agents",
        "data": [50, 50],
        "backgroundColor": ["#4CAF50", "#FF9800"]
      }
    ]
  },
  "sarudzo": {
    "zita": {
      "kuratidza": chokwadi,
      "zvinyorwa": "LangChain uye DSPy Architecture"
    },
    "legend": {
      "kuratidza": chokwadi
    }
  }
}

Ichi chitarisiko chinoratidza kugoverwa kweAI agents pakati peLangChain neDSPy, ne50% yevashandi vanomhanya paLangChain uye 50% vachimhanya paDSPy.

Mhedziso

LangChain neDSPy inopa hurongwa hwakasimba hwekuvaka uye kuronga maAI maajenti, zvichiita kuti vanogadzira vagadzire masisitimu anokatyamadza uye akasimba anogona kugadzirisa zvinobudirira kusangana kwakaoma pakati pevazhinji vamiririri. Nekutevera matanho anotsanangurwa muchinyorwa chino, vagadziri vanogona kushandisa maAI maajenti vachishandisa LangChain neDSPy, uye voaendesa kune yakagoverwa sisitimu vachishandisa DSPy. Mienzaniso yekodhi uye kuratidzwa kwakapihwa kunoratidza kugona kweLangChain neDSPy, uye yepamusoro 5-10 inonyanya kushandiswa mirairo / snippets inopa inokurumidza kutaurwa kune vanogadzira.

Kuronga AI Agents neLangChain uye DSPy yeEnhanced Software Architecture

Have Questions?