I-Orchestrating AI Agents kunye neLangChain kunye ne-DSPy yoLwakhiwo lweSoftware oPhumelelweyo

Ukucwangcisa iiarhente ze-AI kunye neLangChain kunye neDSPy yoKwakha iSoftware ePhuculweyo

Intshayelelo

Kwimeko yedijithali yanamhlanje ekhawulezayo, iinkqubo zesoftware ziya zisiba nzima ngakumbi, ngokudityaniswa kweArtificial Intelligence (AI) kunye nesakhono sokuFunda ngoomatshini (ML). Ngenxa yoko, ukuyila iinkqubo ezinokwehla kunye nezomelelayo ezinokuthi zilungelelanise ngokufanelekileyo iiarhente ze-AI kuye kwaba yinkxalabo ecinezelayo kwiinjineli zesoftware. Eli nqaku liphonononga i-intersection ye-AI, ubunjineli besofthiwe, kunye nezixhobo zokuyila ukuyila i-architectures ye-software esebenzayo, ngokugxininisa ekuphunyezweni kwe-AI agents usebenzisa i-LangChain kunye ne-DSPy.

Imfuno ye-AI Orchestration

Iinkqubo zesoftware yemveli zihlala ziyilwe ngoyilo lwe-monolithic, apho onke amacandelo adityaniswe ngokuqinileyo kwaye asebenza kunye ngendlela yomgca. Nangona kunjalo, ngokufika kwe-AI kunye ne-ML, iinkqubo zesoftware ziya zisasazwa ngakumbi kwaye zisasazwe, kunye neearhente ezininzi ezisebenzisanayo ukufezekisa injongo efanayo. Oku kuye kwakhokelela kwimfuno ye-AI ye-orchestration, ebandakanya ukuyila iinkqubo ezinokulawula ngokufanelekileyo kunye nokulungelelanisa ukusebenzisana phakathi kwee-arhente ezininzi ze-AI.

LangChain: Isakhelo soKwakha iiarhente ze-AI

I-LangChain yinkqubo yePython yokwakha i-AI agents ezinokusebenzisana kunye kunye neenkqubo zangaphandle. Ibonelela ngeseti yezixhobo kunye namathala eencwadi avumela abaphuhlisi ukuba benze iiarhente ezinokuqiqa, zifunde, kwaye ziziqhelanise neemeko eziguqukayo. I-LangChain yakhiwe phezu kolwimi lweprogram ye-Rust kwaye inikezela nge-API ephezulu yokwakha i-AI agents.

DSPy: Isakhelo seOkhestra

I-DSPy yinkqubo-sikhokelo yePython yokwakha iinkqubo ezisasazwayo ezinokuququzelela ii-arhente ezininzi ze-AI. Ibonelela ngeseti yezixhobo kunye namathala eencwadi avumela abaphuhlisi ukuba bayile kwaye basebenzise iinkqubo ezikhawulezayo nezinokuqina ezinokulawula intsebenziswano enzima phakathi kweearhente ezininzi. I-DSPy yakhiwe phezulu kwenkqubo ye-Apache Kafka yokuthumela imiyalezo kwaye inikezela nge-API ephezulu yokwakhiwa kweenkqubo ezisasazwayo.

Ukuphumeza iiarhente ze-AI ngeLangChain kunye ne-DSPy

Ukuphumeza ii-arhente ze-AI usebenzisa i-LangChain kunye ne-DSPy, kufuneka silandele la manyathelo:

Chaza i-Agent Architecture: Chaza i-architecture ye-arhente ye-AI, kubandakanywa amagalelo, iziphumo, kunye nokusebenzisana nezinye ii-arhente.
**Ukuphumeza i-Agent Logic **: Ukuphumeza ingqiqo ye-arhente ye-AI usebenzisa i-LangChain, kubandakanywa nokuqiqa, ukufunda, kunye neendlela zokulungelelanisa.
I-Orchestrate i-Agent Interactions: Sebenzisa i-DSPy ukucwangcisa intsebenziswano phakathi kwee-arhente ezininzi ze-AI, kubandakanywa iiprothokholi zonxibelelwano kunye neendlela zotshintshiselwano lwedatha.
Sebenzisa iSixokelelwano: Sebenzisa isistim usebenzisa iqonga lelifu okanye isakhelo sesikhongozeli, esifana neDocker.

Umzekelo weKhowudi: Ukusebenzisa i-Agent ye-AI elula ngeLangChain

Nanku umzekelo wekhowudi olula obonisa indlela yokuphumeza i-arhente ye-AI usebenzisa iLangChain:

ngenisa i-langchain
ukusuka langchain.chains ukungenisa LLMChain


iarhente = langchain.Agent(
    input_schema="text": str},
    output_schema="text": str},
    ikhonkco=LLMChain(
        imodeli="davinci:instructix",
        ubude_ubude=2048,
        iqondo lobushushu=0.7,
    ),
)

# Sebenzisa ingqiqo ye-arhente
def agent_logic(input_text):
    output_text = agent.run(input_text)
    buyisela output_text

# Vavanya iarhente
input_text = "Yintoni intsingiselo yobomi?"
output_text = i-arhente_logic (input_text)
print(output_text)

Lo mzekelo wekhowudi ubonisa indlela yokuchaza i-arhente ye-AI usebenzisa i-LangChain, kubandakanywa i-schemas yegalelo kunye nemveliso, kunye nengqiqo ye-arhente. I-arhente iphunyezwa ngokusebenzisa i-Large Language Model (LLM) ikhonkco, luhlobo lwe-arhente ye-AI enokuthi iqiqe kwaye ivelise isicatshulwa.

Ephezulu 5-10 iMiyalelo/Iziqwengana eziSetyenzisiweyo ezininzi zeLangChain kunye neDSPy

Nantsi eyona miyalelo/iziqwengana eziphezulu ezi-5-10 zeLangChain kunye neDSPy:

langchain.Agent(): Yenza umzekelo omtsha we-AI.

iarhente = langchain.Agent(
    input_schema="text": str},
    output_schema="text": str},
    ikhonkco=LLMChain(
        imodeli="davinci:instructix",
        ubude_ubude=2048,
        iqondo lobushushu=0.7,
    ),
)

i-agent.run(): Iqhuba i-arhente ye-AI ngegalelo elinikiweyo.

output_text = agent.run(input_text)

DSPy.create_cluster(): Yenza iqoqo elitsha leendibano zesistim esasazwayo.

iqela = DSPy.create_cluster(
    inani_iindawo=5,
    uhlobo_ lwendlela = "umsebenzi",
)

DSPy.create_agent(): Yenza i-arhente entsha ye-AI yesistim esasazwayo.

iarhente = DSPy.create_agent(
    Iqela=iqela,
    uhlobo_lwearhente="LLM",
)

I-DSPy.deploy_agent (): Ihambisa i-arhente ye-AI kwisistim esisasazwayo.

DSPy.deploy_arhente(
    iarhente=iarhente,
    Iqela=iqela,
)

DSPy.get_agent_status(): Ifumana kwakhona ubume be-arhente ye-AI.

ubume = DSPy.get_agent_status(
    iarhente=iarhente,
    Iqela=iqela,
)

DSPy.update_agent(): Hlaziya i-arhente ye-AI ngedatha entsha okanye iiparamitha.

DSPy.update_arhente(
    iarhente=iarhente,
    Iqela=iqela,
    new_data="text": "umbhalo omtsha wegalelo"},
)

DSPy.delete_agent(): Icima i-arhente ye-AI kwinkqubo esasazwayo.

DSPy.cima_iarhente(
    iarhente=iarhente,
    Iqela=iqela,
)

DSPy.create_chain(): Yenza ikhonkco elitsha lee-arhente ze-AI kwinkqubo esasazwayo.

ikhonkco = DSPy.create_chain(
    ii-arhente=[i-arhente1, i-arhente2, i-arhente3],
)

DSPy.deploy_chain(): Ihambisa ikhonkco lee-arhente ze-AI kwisistim esasazwayo.

DSPy.deploy_chain(
    ikhonkco=ityathanga,
    Iqela=iqela,
)

Umbono

Nantsi imbonakalo yoyilo lweLangChain kunye neDSPy usebenzisa iChart.js:

{
  "type": "pieChart",
  "idatha": {
    "labels": ["LangChain", "DSPy"],
    "iiseti zedatha": [
      {
        "ileyibheli": "ii-AI Agents",
        "idatha": [50, 50],
        "backgroundColor": ["#4CAF50", "#FF9800"]
      }
    ]
  },
  "iinketho": {
    "isihloko": {
      "bonisa": yinyani,
      "umbhalo": "LangChain kunye neDSPy Architecture"
    },
    "legend": {
      "bonisa": yinyani
    }
  }
}

Lo mzobo ubonisa ukuhanjiswa kwee-agent ze-AI phakathi kweLangChain kunye ne-DSPy, kunye ne-50% yee-agent ezisebenza kwi-LangChain kunye ne-50% eqhuba kwi-DSPy.

Isiphelo

I-LangChain kunye ne-DSPy zibonelela ngesakhelo esinamandla sokwakha kunye nokuqulunqa ii-arhente ze-AI, ezivumela abaphuhlisi ukuba bayile iinkqubo ezinobungozi kunye nokuqina ezinokulawula ngokufanelekileyo intsebenziswano enzima phakathi kwee-arhente ezininzi. Ngokulandela amanyathelo achazwe kweli nqaku, abaphuhlisi banokusebenzisa i-AI agents usebenzisa i-LangChain kunye ne-DSPy, kwaye bayithumele kwinkqubo yokusabalalisa usebenzisa i-DSPy. Imizekelo yekhowudi kunye nokuboniswa okubonelelweyo kubonisa amandla eLangChain kunye ne-DSPy, kwaye i-5-10 ephezulu kakhulu imiyalelo / iziqwenga ezisetyenziswayo zibonelela ngokukhawuleza kubaphuhlisi.

I-Orchestrating AI Agents kunye neLangChain kunye ne-DSPy yoLwakhiwo lweSoftware oPhumelelweyo

Have Questions?