Gadzira LangChain mumiririri | B4N1

Gadzira LangChain mumiririri

8 months ago · b4n1 Read time: 4 min

Kubatanidza maAI Agents eOtomatiki Software Delivery: Kubatanidza LangChain uye DSPy muDevOps Pipelines

Introduction

Iyo software yekuvandudza mamiriro ari kukurumidza kushanduka, nekuwedzera kutorwa kweArtificial Intelligence (AI) uye Machine Kudzidza (ML) mune akasiyana siyana ehupenyu hwebudiriro. Imwe yenzvimbo dzinonyanya kuvimbisa yekushandiswa kweAI iri mukuita otomatiki software yekuendesa maitiro, kusanganisira telemetry uye kuenderera mberi kwekubatanidza / kuenderera mberi kwekutumira (CI/CD). Muchinyorwa chino, isu tichaongorora basa revamiririri veAI mukuita otomatiki hurongwa hwekuunza software uye kuongorora mukuita kunoshanda kwematurusi ekuimba seLangChain neDSPy.

Basa reAI Agents muSoftware Delivery

AI vamiririri mapurogiramu esoftware anogona kuita mabasa akazvimirira, kuita sarudzo zvichienderana nedata uye mitemo. Muchirevo chekuendesa software, vamiririri veAI vanogona kushandiswa kuita otomatiki maitiro akasiyana, akadai se:

  1. Telemetry: Vamiririri veAI vanogona kuunganidza uye kuongorora data kubva kwakasiyana-siyana, sematanda ekushandisa, metrics ekushanda, uye mhinduro yemushandisi, kuti vape ruzivo rwemaitiro nemafambisirwo echirongwa.
  2. ** Kuenderera Kubatanidzwa **: AI vamiririri vanogona automate kuvaka, kuyedza, uye kutumira kwekuchinja kwekodhi, kuve nechokwadi chekuti chikumbiro chinogara chiri munzvimbo inogona kutumirwa.
  3. ** Kuenderera Kunoenderera mberi **: Vamiririri veAI vanogona kushandura kutumira kwekuchinja kune kugadzirwa, kuderedza nguva uye simba rinodiwa kuti vawane zvitsva uye kugadzirisa kune vashandisi.
  4. Kutarisisa uye Kuchenjedza: Vamiririri veAI vanogona kutarisa kushanda kwechirongwa uye kuzivisa vanogadzira nevashandisi nezvezvingangoitika zvisati zvanyanya kukosha.

**LangChain: Chimiro chekuvaka AI Agents **

LangChain ndeye Python dhizaini yekuvaka AI vamiririri vanogona kudyidzana neakasiyana data masosi uye kuita mabasa vakazvimirira. LangChain inopa iri nyore uye intuitive API yevamiririri vekuvaka vanogona:

  1. ** Kudyidzana nezvinyorwa zvemashoko **: LangChain inopa huwandu hwezvinyorwa zvemashoko, kusanganisira zvinyorwa, APIs, uye mafaira emafaira, ayo vamiririri vanogona kuwirirana navo.
  2. ** Ita mabasa **: LangChain inopa huwandu hwemabasa ayo vamiririri vanogona kuita, kusanganisira kunyorwa kwemavara, kucherechedzwa kwechifananidzo, uye kushandiswa kwemutauro wechisikigo.
  3. **Dzidza kubva kune data **: LangChain inopa huwandu hwemashini ekudzidza algorithms ayo vamiririri vanogona kushandisa kudzidza kubva kune data.

DSPy: Chimiro Chekuvaka Yakaparadzirwa AI Agents

DSPy iPython framework yekuvaka yakagoverwa AI maajenti anogona kukwira kuti asangane nezvinodiwa zvehukuru-hukuru hwekushandisa. DSPy inopa huwandu hwezvinhu, kusanganisira:

  1. ** Distributed computing **: DSPy inobvumira vamiririri kuti vashandise pamichina yakawanda, zvichiita kuti ive yakanaka kune yakakura-yakakura kushandiswa.
  2. Data parallelism: DSPy inobvumira vamiririri kuti vashandise data nenzira yakafanana, zvichiita kuti zvive zvakanaka kune zvikumbiro zvinoda kushandiswa kwepamusoro-soro.
  3. Scalability: DSPy inobvumira vamiririri kuyera kuti vasangane nezvinodiwa zvezvikumbiro zvakakura.

Kubatanidza LangChain uye DSPy muDevOps Pipelines

Kubatanidza LangChain uye DSPy mumapaipi eDevOps, tinogona kushandisa matanho anotevera:

  1. ** Isa LangChain uye DSPy **: Isa LangChain uye DSPy uchishandisa pip.
pip kuisa langchain dspy
  1. **Gadzira mumiririri weLangChain **: Gadzira mumiririri weLangChain anogona kuwirirana nezvinyorwa zve data uye kuita mabasa.
import langchain


mumiririri = langchain.Agent(
    zita = "mumiririri_wangu",
    data_sources=[langchain.DataSource("database", "my_database")],
    mabasa=[langchain.Task("text_processing", "my_text_processing")]
)
  1. Gadzira mumiririri weDSPy: Gadzira mumiririri weDSPy anogona kushanda pamichina yakawanda uye kugadzirisa data nenzira yakafanana.
import dspy

# Gadzira DSPy mumiriri
mumiririri = dspy.Agent(
    zita = "mumiririri_wangu",
    data_sources=[dspy.DataSource("database", "my_database")],
    mabasa=[dspy.Task("text_processing", "my_text_processing")]
)
  1. **Kubatanidza LangChain uye DSPy vamiririri **: Batanidza LangChain uye DSPy vamiririri muDevOps pombi uchishandisa chishandiso chakaita seJenkins kana GitLab CI/CD.
import langchain
import dspy

# Gadzira LangChain mumiriri
langchain_agent = langchain.Agent(
    zita = "my_langchain_agent",
    data_sources=[langchain.DataSource("database", "my_database")],
    mabasa=[langchain.Task("text_processing", "my_text_processing")]
)

# Gadzira DSPy mumiriri
dspy_agent = dspy.Agent(
    zita = "my_dspya_agent",
    data_sources=[dspy.DataSource("database", "my_database")],
    mabasa=[dspy.Task("text_processing", "my_text_processing")]
)

# Batanidza vamiririri mupombi yeDevOps
pombi = [
    langchain_agent,
    dspy_agent
]

Pamusoro 5-10 Inonyanya Kushandiswa Mirairo/Zvidimbu

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

  1. LangChain: Gadzira LangChain mumiririri
import langchain

mumiririri = langchain.Agent(
    zita = "mumiririri_wangu",
    data_sources=[langchain.DataSource("database", "my_database")],
    mabasa=[langchain.Task("text_processing", "my_text_processing")]
)
  1. DSPy: Gadzira mumiririri weDSPy
import dspy

mumiririri = dspy.Agent(
    zita = "mumiririri_wangu",
    data_sources=[dspy.DataSource("database", "my_database")],
    mabasa=[dspy.Task("text_processing", "my_text_processing")]
)
  1. **LangChain: Bata nedata sosi **
import langchain

mumiririri = langchain.Agent(
    zita = "mumiririri_wangu",
    data_sources=[langchain.DataSource("database", "my_database")],
    mabasa=[langchain.Task("text_processing", "my_text_processing")]
)

data = agent.data_sources[0].get_data()
  1. DSPy: Bata pamwe nekwakabva data
import dspy

mumiririri = dspy.Agent(
    zita = "mumiririri_wangu",
    data_sources=[dspy.DataSource("database", "my_database")],
    mabasa=[dspy.Task("text_processing", "my_text_processing")]
)

data = agent.data_sources[0].get_data()
  1. **LangChain: Ita basa **
import langchain

mumiririri = langchain.Agent(
    zita = "mumiririri_wangu",
    data_sources=[langchain.DataSource("database", "my_database")],
    mabasa=[langchain.Task("text_processing", "my_text_processing")]
)

mhedzisiro = agent.tasks[0].perform_task(data)
  1. **DSPy: Ita basa **
import dspy

mumiririri = dspy.Agent(
    zita = "mumiririri_wangu",
    data_sources=[dspy.DataSource("database", "my_database")],
    mabasa=[dspy.Task("text_processing", "my_text_processing")]
)

mhedzisiro = agent.tasks[0].perform_task(data)
  1. **LangChain: Dzidza kubva kune data **
import langchain

mumiririri = langchain.Agent(
    zita = "mumiririri_wangu",
    data_sources=[langchain.DataSource("database", "my_database")],
    mabasa=[langchain.Task("text_processing", "my_text_processing")]
)

agent.learn_from_data(data)
  1. **DSPy: Dzidza kubva kune data **
import dspy

mumiririri = dspy.Agent(
    zita = "mumiririri_wangu",
    data_sources=[dspy.DataSource("database", "my_database")],
    mabasa=[dspy.Task("text_processing", "my_text_processing")]
)

agent.learn_from_data(data)
  1. **LangChain: Batanidza nepombi yeDevOps **
import langchain

mumiririri = langchain.Agent(
    zita = "mumiririri_wangu",
    data_sources=[langchain.DataSource("database", "my_database")],
    mabasa=[langchain.Task("text_processing", "my_text_processing")]
)

pombi = [
    mumiririri
]
  1. DSPy: Batanidza nepombi yeDevOps
import dspy

mumiririri = dspy.Agent(
    zita = "mumiririri_wangu",
    data_sources=[dspy.DataSource("database", "my_database")],
    mabasa=[dspy.Task("text_processing", "my_text_processing")]
)

pombi = [
    mumiririri
]

Kuona

Heino ratidziro yeLangChain uye DSPy vamiririri muDevOps pombi:

{
  "type": "bha",
  "data": {
    "labels": ["LangChain", "DSPy"],
    "datasets": [
      {
        "label": "Agent",
        "data": [10, 20],
        "backgroundColor": ["rgba(255, 99, 132, 0.2)", "rgba(54, 162, 235, 0.2)"],
        "borderColor": ["rgba(255, 99, 132, 1)", "rgba(54, 162, 235, 1)"],
        "borderWidth": 1
      }
    ]
  },
  "sarudzo": {
    "zita": {
      "kuratidza": chokwadi,
      "zvinyorwa": "LangChain uye DSPy Agents muDevOps Pipeline"
    },
    "zvikero": {
      "yAxes": [
        {
          "zvikwekwe": {
            "beginAtZero": chokwadi
          }
        }
      ]
    }
  }
}

Kuona uku kunoratidza vamiririri veLangChain neDSPy mupombi yeDevOps, nemumiririri weLangChain aine mabasa gumi uye mumiriri weDSPy aine mabasa makumi maviri.

Mhedziso

Muchinyorwa chino, takaongorora basa revamiririri veAI mukuita otomatiki magadzirirwo esoftware uye takaongorora mashandisirwo anoshanda ematurusi ekuimba seLangChain neDSPy. Takaona kuti LangChain uye DSPy inogona kushandiswa sei kugadzira maAI maajenti anogona kupindirana nedata masosi, kuita mabasa, uye kudzidza kubva kune data. Isu takaonawo kuti LangChain neDSPy inogona kubatanidzwa sei mumapaipi eDevOps uchishandisa maturusi akaita seJenkins kana GitLab CI/CD. Nekuwedzera kugamuchirwa kweAI neML mukugadzirwa kwesoftware, LangChain neDSPy zviri kuwedzera kukosha maturusi kune vanogadzira uye vashandisi vari kutsvaga kuita otomatiki nzira dzekutumira software.

Have Questions?

Let's Talk

Contact me on LinkedIn