Iwuli sistemu ngwanrọ na-agwọ onwe ya na Multi-Agent AI Architectures na LangChain | B4N1

Iwuli sistemu ngwanrọ na-agwọ onwe ya na Multi-Agent AI Architectures na LangChain

5 months ago · b4n1 Read time: 5 min

** Iji wuo sistemụ sọftụwia na-agwọ onwe ya na ọtụtụ ụlọ ọrụ AI Architectures na LangChain ***

** Okwu mmalite**

N'ime mmepe ngwanrọ ngwa ngwa nke taa, iwulite sistemụ sọftụwia na-agwọ onwe ya aghọwo ihe dị oke mkpa maka ịhụ na ntụkwasị obi, scalability, na resilience nke sistemụ ngwanrọ dị mgbagwoju anya. Ihe owuwu monolithic ọdịnala ezughị ezu iji gboo ihe achọrọ nke mmepe ngwanrọ ọgbara ọhụrụ, ebe sistemụ ga-enwe ike ime mgbanwe maka mgbanwe chọrọ, jikwaa ọdịda ndị a na-atụghị anya ya, wee gbakee na mmejọ ngwa ngwa. Nke a bụ ebe ọtụtụ ụlọ ọrụ AI architectures na LangChain na-abata, na-enye ihe ngwọta dị ike maka ịmepụta usoro ngwanrọ na-agwọ onwe ya nke nwere ike ịmụta, gbanwee, na nzaghachi ọnọdụ mgbanwe.

** Gịnị bụ Multi-Agent AI Architectures?**

Multi-agent AI architectures bụ ụdị nkesa AI usoro nke mejupụtara ọtụtụ ndị nọọrọ onwe ha na-emekọrịta ihe na ibe ha na gburugburu ha iji nweta otu ebumnuche. Onye ọrụ ọ bụla nwere ikike ime mkpebi nke ya, ihe ọmụma na akụrụngwa, ha nwekwara ike ịkparịta ụka n'etiti onwe ha ka ha kesaa ozi, hazie omume, na ime mgbanwe na ọnọdụ mgbanwe.

N'ihe gbasara sistemụ ngwanrọ, enwere ike iji ụlọ ọrụ AI ọtụtụ ndị nnọchite anya iji wuo usoro ọgwụgwọ onwe nke nwere ike ịchọpụta ma meghachi omume na njehie, ọdịda, na nsogbu ndị ọzọ ozugbo. Site n'ịkesa usoro mkpebi n'ofe ọtụtụ ndị ọrụ, usoro ndị a nwere ike ịmụta site na ahụmahụ, ime mgbanwe na ọnọdụ mgbanwe, ma nwetaghachi site na njehie ngwa ngwa.

** Gịnị bụ LangChain?**

LangChain bụ ọba akwụkwọ Python mepere emepe maka iwu sistemu AI ọtụtụ ndị nnọchite anya. Ọ na-enye usoro dị mfe, nke na-agbanwe agbanwe, na nke nwere ike ịmepụta maka ịmepụta usoro AI siri ike nke nwere ike ịmụta, gbanwee, na ịzaghachi ọnọdụ mgbanwe. Emebere LangChain ka ọ rụọ ọrụ na usoro AI dị iche iche, gụnyere TensorFlow, PyTorch, na Keras, a pụkwara iji ya rụọ usoro nke jikọtara ya na isi mmalite data dị iche iche, gụnyere ọdụ data, API, na ngwaọrụ IoT.

** Iji LangChain wulite sistemu ngwanrọ ọgwụgwọ onwe onye ***

Iji wuo usoro ngwanrọ na-agwọ onwe ya na LangChain, anyị kwesịrị ịmepụta ọtụtụ ụlọ ọrụ AI nke nwere ike ịchọpụta ma zaghachi njehie, ọdịda, na ihe ndị ọzọ na-adịghị mma na oge. Nke a bụ nkọwapụta ọkwa dị elu nke ihe owuwu ụlọ:

  1. ** Agent Design ***: Anyị kwesịrị ịmepụta otu ndị ọrụ nwere onwe ha nwere ike imekọrịta ibe ha na gburugburu ha iji nweta otu ebumnuche. Onye ọrụ ọ bụla kwesịrị inwe ikike ime mkpebi nke ya, ihe ọmụma na akụrụngwa.
  2. ** Eserese Ọmụma ***: Anyị kwesịrị ịmepụta eserese ihe ọmụma nke na-anọchite anya mmekọrịta dị n'etiti ndị ọrụ, gburugburu ebe obibi, na ebumnuche nke usoro. Enwere ike iji eserese a tụgharịa uche maka ọnọdụ sistemụ ahụ wee mee mkpebi.
  3. ** Mkpebi-Mkpebi ***: Anyị kwesịrị imejuputa usoro mkpebi nke na-enye ndị ọrụ ohere iche echiche banyere ọnọdụ nke usoro ahụ ma mee mkpebi dabere na eserese ihe ọmụma.
  4. ** Action Excution ***: Anyị kwesịrị imejuputa usoro ihe omume nke na-enye ndị ọrụ ohere ime mkpebi ha ma soro gburugburu ebe obibi na-emekọrịta ihe.

Nke a bụ ọmụmaatụ dị mfe nke sistemu ngwanrọ na-agwọ onwe ya nke ejiri LangChain rụọ: "Python mbubata langchain si langchain.agents mbubata Agent si langchain.llms bubata LLM

eserese = { 'Agent': ['Agent1', 'Agent2', 'Agent3'], 'gburugburu': ['database', 'API', 'IoT ngwaọrụ'], 'ihe mgbaru ọsọ': ['chọpụta njehie', 'zaghachi ọdịda', 'naghachi na njehie'] }

Kọwaa usoro ime mkpebi

Def decision_making (onye nnọchi anya, ihe ọmụma_graph): # Ihe kpatara ọnọdụ sistemụ ahụ steeti = ihe ọmụma_graph [' agents'] [onye nnọchi anya] # Mee mkpebi dabere na steeti mkpebi = 'zaghachi ọdịda' nloghachi mkpebi

Kọwaa usoro ihe omume

def action_execution(onye nnọchi anya, mkpebi): # Mee mkpebi ahụ ọ bụrụ na mkpebi == 'zaghachi ọdịda': # Zaghachi ọdịda ebipụta ('Na-azaghachi ọdịda...') laghachi

Mepụta onye nnọchi anya LangChain

onye nnọchi anya = onye nnọchi anya ( aha = 'onye na-agwọ onwe ya', mkpebi_ime = mkpebi_ime, action_execution=action_execution, ihe ọmụma_graph=graph )

Mepụta LangChain LLM

LM = LLM ( aha = 'na-agwọ onwe-llm', model = 'davinci' )

Mepụta sistemụ LangChain

sistemu = langchain.System( onye nnọchi anya = onye nnọchi anya, lm=llm )

Mepee sistemụ ahụ

usoro.run() `` Ihe atụ a na-egosi usoro ngwanrọ na-agwọ onwe ya dị mfe nke nwere ike ịchọpụta mmejọ, zaghachi ọdịda, na weghachite site na njehie ngwa ngwa. Usoro ahụ nwere usoro ndị ọrụ kwụụrụ onwe ha na-emekọrịta ihe na gburugburu ha iji nweta otu ebumnuche. Ndị ọrụ na-eji eserese ihe ọmụma na-atụgharị uche banyere ọnọdụ nke usoro ahụ ma mee mkpebi dabere na eserese ahụ.

** Iwu 5-10 Kachasị Eji mee ihe

Nke a bụ iwu/snippets 5-10 kacha eji eme ihe maka LangChain:

  1. langchain.Agent(): Na-emepụta onye nnọchi anya LangChain ọhụrụ.
  2. `langchain.LLM()': Na-emepụta LangChain LLM ọhụrụ.
  3. `langchain.System()': Na-emepụta usoro LangChain ọhụrụ.
  4. agent.decision_making(): Na-eme usoro ime mkpebi maka onye nnọchi anya.
  5. `agent.action_execution()': Na-eme usoro mmebe omume maka onye nnọchite anya.
  6. llm.predict(): Na-eji LLM na-eme amụma.
  7. system.run(): Na-agba ọsọ usoro LangChain.
  8. graph.add_node(): Na-agbakwunye ọnụ ọhụrụ na eserese ngosi.
  9. graph.add_edge(): Na-agbakwunye ihu ọhụrụ na eserese ihe ọmụma.
  10. `graph.remove_node()': na-ewepu ọnụ na eserese ihe ọmụma.

** Nlegharị anya ***

Nke a bụ nhụta dị mfe nke eserese ihe ọmụma site na iji Chart.js: Json { "ụdị": "graph", "data": { "ọnụ": [ {"id": "agent1", "labelụ": "Onye nnọchite anya 1"}, {"id": "agent2", "label": "Agent 2"}, {"id": "agent3", "labelụ": "Onye nnọchite anya 3"}, {"id": "database", "label": "Ebe nchekwa data"}, {"id": "API", "labelụ": "API"}, {"id": "Ngwaọrụ IoT", "labelụ": "Ngwaọrụ IoT"} ], "ọnụ": [ {"source": "agent1", "target": "database"}, {"isi iyi": "agent2", "target": "API"}, {"isi iyi": "agent3", "target": "ngwaọrụ IoT"}, {"source": "agent1", "target": "agent2"}, {"isi iyi": "agent2", "target": "agent3"} ] }, "Nhọrọ": { "nhazi": { "ụdị": "circular" }, "ọnụ": { "nha": 50 }, "ọnụ": { "obosara": 2 } } } `` Ọhụụ a gosipụtara mmekọrịta dị n'etiti ndị ọrụ, gburugburu ebe obibi na ebumnuche nke sistemụ.

Mkpebi

Iwulite usoro ngwanrọ na-agwọ onwe ya na ọtụtụ ụlọ ọrụ AI architectures na LangChain na-enye ngwọta dị ike maka ijide n'aka na ntụkwasị obi, scalability, na resilience nke usoro ngwanrọ mgbagwoju anya. Site n'ịkesa usoro mkpebi n'ofe ọtụtụ ndị ọrụ, usoro ndị a nwere ike ịmụta site na ahụmahụ, ime mgbanwe na ọnọdụ mgbanwe, ma nwetaghachi site na njehie ngwa ngwa. Na LangChain, ndị mmepe nwere ike iwulite usoro AI siri ike nke jikọtara ya na isi mmalite data dị iche iche, gụnyere ọdụ data, API, na ngwaọrụ IoT. Site n'ịgbaso koodu ihe atụ na iji n'elu 5-10 iwu / snippets kacha eji eme ihe, ndị mmepe nwere ike iji LangChain wulite usoro ngwanrọ na-agwọ onwe ha.

Have Questions?

Let's Talk

Contact me on LinkedIn