Na-eji Ngwa Orchestration na-eme Nhazi Ngwanrọ echekwabara na nke nwere ike ịgbanye | B4N1

Na-eji Ngwa Orchestration na-eme Nhazi Ngwanrọ echekwabara na nke nwere ike ịgbanye

3 months ago · b4n1 Read time: 5 min

Okwu mmalite

Ntinye aka nke Artificial Intelligence (AI) n'ime ngwa ngwanrọ agbanweela ụzọ anyị si abịaru nsogbu, ime mkpebi, na akpaaka. Otú ọ dị, ka ngwa ndị AI na-akwalite na-aghọwanye mgbagwoju anya, mkpa maka ihe owuwu dị nchebe na nke nwere ike ibugharị aghọwo ihe na-achọsi ike. N'isiokwu a, anyị ga-enyocha njikọ nke AI, software engineering, na DevOps, na-emesi ike na iji ngwá ọrụ nhazi dị ka DSPy na LangChain iji jikọta ndị ọrụ AI na pipeline DevOps.

Ihe ịma aka nke ụlọ ọrụ ngwanrọ AI-chụpụrụ

Ngwa sọftụwia AI na-ebutere ndị injinia sọftụwia na ndị otu DevOps ihe ịma aka pụrụ iche. Ụfọdụ n'ime isi ihe ịma aka gụnyere:

** Scalability ***: Ụdị AI chọrọ akụrụngwa mgbakọ na mwepụ, nke nwere ike iduga nsogbu scalability na gburugburu mmepụta. ** Nchekwa ***: Ụdị AI nwere ike bụrụ ngwa ngwa na mwakpo, dị ka nsị data na mwakpo ntụgharị ihe nlereanya. ** Njikọta ***: Ekwesịrị ijikọ ụdị AI na sistemụ ngwanrọ dị ugbu a, nke nwere ike ịbụ ọrụ siri ike. ** nlekota na nlekọta ***: Ụdị AI chọrọ nleba anya na nleba anya na-aga n'ihu iji hụ na ha nọgidere na-abụ nke ziri ezi na ntụkwasị obi.

Ngwa Orchestration: DSPy na LangChain

Ngwá ọrụ nhazi dị ka DSPy na LangChain na-enye ngwọta maka ihe ịma aka ndị a site n'ịkwado njikọ nke ndị ọrụ AI na pipeline DevOps. Ngwá ọrụ ndị a na-enye usoro maka iwulite, ibugharị, na ijikwa ụdị AI, na-eme ka ụkpụrụ ụlọ dị nchebe na nke nwere ike ime ka ngwa AI na-ebugharị.

DSPy

DSPy bụ ngwá ọrụ nhazi ihe na-emepe emepe wuru n'elu Python. Ọ na-enye API dị mfe ma nwee nghọta maka iwulite, ibugharị, na ijikwa ụdị AI. DSPy na-akwado ọtụtụ usoro AI, gụnyere TensorFlow, PyTorch, na scikit-learn.

LangChain

LangChain bụ ihe ọzọ na-emepe emepe ngwa ngwa nke na-enye usoro maka iwulite, ibugharị, na ijikwa ụdị AI. A na-ewu LangChain n'elu Rust ma na-enye API arụmọrụ dị elu maka ijikọ ndị ọrụ AI na pipeline DevOps.

Na-eji DSPy na-arụ ọrụ nchekwa nchekwa na ngwa ngwa AI-chụpụrụ

N'ime ngalaba a, anyị ga-enyocha ihe atụ dị mfe nke itinye ụkpụrụ ụlọ sọftụwia AI na-ebugharị na DSPy nwere nchekwa.

Ọmụmaatụ: Iji DSPy wulite Model AI dị Mfe

Ka anyị tụlee ihe atụ dị mfe nke ịrụ ụdị mmụta igwe site na iji DSPy. Anyị ga-eji ọba akwụkwọ scikit-mụta iji wuo usoro nkwụghachi azụ ahịrị.

"Python mbubata dspy ka dsp si sklearn.linear_model mbubata LinearRegression si sklearn.datasets bubata load_diabetes

Budata ihe ndekọ nke ọrịa shuga

dataset = load_diabetes()

Kewaa dataset ka ọ bụrụ ọzụzụ na usoro nnwale

X_train, X_test, y_train, y_test = dsp.split_dataset (dataset.data, dataset.target, test_size=0.2)

Wulite ihe nlegharị anya ahịrị

nlereanya = LinearRegression()

Zụlite ihe nlereanya na data ọzụzụ

model.fit (X_train, y_train)

Nyochaa ihe nlereanya na data nnwale

izi ezi = ụdị. akara (X_test, y_test) ebipụta (f"Izizi ihe nlere anya: {nkenke:.2f}") ``

Ihe ngọngọ koodu a na-egosi otu esi arụ ụdị mmụta igwe dị mfe site na iji DSPy. Anyị na-ebu dataset nke ọrịa shuga, kewaa ya n'ọzụzụ na nhazi nnwale, wuo usoro nkwụghachi azụ ahịrị, zụọ ya na data ọzụzụ, wee nyochaa izi ezi ya na data nnwale.

Na-eji LangChain na-arụ ọrụ nchekwa ngwa ngwa na nke AI na-ebugharị.

Na ngalaba a, anyị ga-enyocha ihe atụ dị mfe nke iji LangChain na-arụ ọrụ ngwanrọ AI na-ebugharị na nchekwa.

Ọmụmaatụ: Iji LangChain wulite Model AI dị Mfe

Ka anyị tụlee ihe atụ dị mfe nke iji LangChain wuo ụdị mmụta igwe. Anyị ga-eji ọbá akwụkwọ TensorFlow wuo ụdị netwọkụ akwara.

`` nchara jiri langchain dị ka lc; jiri tensorflow:: mmalite::*;

// Kọwaa ụdị netwọkụ akwara dị mfe fn ụdị (ntinye: & Tensor) -> Tensor { ka mmepụta = input.matmul (& Tensor :: ọhụrụ (1, 1)); mmepụta }

// Mepụta onye nnọchi anya LangChain ka onye ọrụ = lc :: Agent :: ọhụrụ (ihe nlereanya);

// Kọwaa ihe ndekọ data ka dataset = lc :: Dataset :: ọhụrụ (vec![ lc:: Ihe atụ:: ọhụrụ (vec![1.0, 2.0], vec![3.0]), lc:: Ihe atụ:: ọhụrụ(vec![4.0, 5.0], vec![6.0]), ]);

// Zụọ ihe nlereanya na dataset agent.train (&dataset);

// Nyochaa ihe nlereanya na dataset ka izi ezi = agent.nyocha (&dataset); println!("Izizi ihe nlereanya: {}", izi ezi); ``

Ihe ngọngọ koodu a na-egosi otu esi arụ ụdị mmụta igwe dị mfe site na iji LangChain. Anyị na-akọwapụta ụdị netwọkụ akwara dị mfe, mepụta onye nnọchi anya LangChain, kọwapụta dataset, zụọ ihe nlere anya na dataset, wee nyochaa izi ezi ya na dataset.

Top 5-10 Iwu/Snippets kacha eji

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

DSPy

  • `dsp.split_dataset(data, target, test_size=0.2)': Na-ekewa ihe nkesa n'ime ọzụzụ na nhazi ule.
  • dsp.build_model(model_name, model_params): Na-ewulite igwe mmụta n'ịmụ igwe site na iji aha ihe atụ na ihe atụ akọwapụtara.
  • `dsp.train_model(model, data): na-azụ ụdị mmụta igwe na data akọwapụtara.
  • dsp.evaluate_model(model, data): Na-enyocha arụmọrụ nke igwe mmụta ihe na data akọwapụtara.
  • dsp.deploy_model(model, server_url): Na-ebunye ụdị mmụta igwe na URL sava akọwapụtara.

LangChain

  • `lc:: Onye nnọchi anya :: ọhụrụ (ihe nlereanya): Na-emepụta onye nnọchi anya LangChain site na iji ụdị akọwapụtara.
  • `lc:: Dataset:: ọhụrụ(atụ)': Na-eji ihe atụ akọwapụtara nke ọma na-emepụta dataset LangChain.
  • `lc :: ụgbọ oloko (onye nnọchi anya, dataset): na-azụ onye nnọchi anya LangChain na dataset akọwapụtara.
  • `lc :: nyochaa(onye nnọchi anya, dataset):': Na-enyocha arụmọrụ nke onye nnọchi anya LangChain na dataset akọwapụtara.
  • lc:: bukwasa(onye nnọchite anya, server_url): Na-ebuga onye nnọchi anya LangChain na URL nkesa akọwapụtara.

Nlere anya

Nke a bụ nleba anya dị mfe nke nhazi ngwanrọ AI na-eji Chart.js:

javascript const ctx = document.getElementById('myChart').getContext('2d'); const chart = Chart ọhụrụ (ctx, { ụdị: 'bar', data: { akara: ['Oge Ọzụzụ','Oge Ntinye','Ezizi Model'], ihe ndekọ data: [{ akara: 'DSPy', data: [10, 5, 90], ndabereAgba: 'rgba(255, 99, 132, 0.2)', oke agba: 'rgba(255, 99, 132, 1)', Obosara oke: 1 }, { akara: 'LangChain', data: [8, 6, 92], ndabereAgba: 'rgba(54, 162, 235, 0.2)', oke agba: 'rgba(54, 162, 235, 1)', Obosara oke: 1 }] }, nhọrọ: { akpịrịkpa: { y: { beginAtZero: eziokwu } } } }

Ihe ngọngọ koodu a na-emepụta eserese mmanya dị mfe site na iji Chart.js iji hụ na arụmọrụ nke DSPy na LangChain.

Mmechi

N'isiokwu a, anyị nyochara njikọ nke AI, injinia ngwanrọ, na DevOps, na-emesi ike na iji ngwá ọrụ nhazi dị ka DSPy na LangChain iji jikọta ndị ọrụ AI na pipeline DevOps. Anyị gosipụtara otu esi etinye ihe nrụnye ngwanrọ AI na-echekwa nke ọma site na iji DSPy na LangChain, wee nye ihe atụ dị mfe nke iwulite igwe mmụta mmụta igwe site na iji ngwaọrụ ọ bụla. Anyị nyekwara iwu/snippets 5-10 kacha eji eme ihe maka ngwa ọrụ ọ bụla yana nleba anya dị mfe nke nhazi ngwanrọ AI na-ebugharị site na iji Chart.js.

Have Questions?

Let's Talk

Contact me on LinkedIn