Intshayelelo
Ukuqhubela phambili ngokukhawuleza kwi-Artificial Intelligence (AI) kunye nobunjineli besoftware kukhokelele kuphuhliso lweenkqubo zesoftware ezintsonkothileyo ezidibanisa iiarhente ze-AI kunye nemodyuli yokuhamba komsebenzi. Ezi nkqubo zinamandla okuguqula indlela esisondela ngayo kwi-DevOps automation kunye ne-orchestration. Nangona kunjalo, ukuyila iinkqubo zesoftware eziqhutywa yi-AI ezinobungozi kunye nokuziphatha kufuna ukuqonda okunzulu kwenkqubela phambili yamva nje kwi-AI, ubunjineli besoftware, kunye noyilo loyilo.
Kweli nqaku, siza kuphonononga imigaqo yoyilo kunye nezona ndlela zilungileyo zokuyila iimodyuli, i-scalable, kunye neenkqubo zesoftware yeenqobo ezisesikweni ezidibanisa iiarhente ze-AI kunye noyilo lwemodyuli yokuhamba komsebenzi. Siza kugxila ekudibaniseni ukuqhubela phambili kwamva nje kwi-AI (IiModeli zoLwimi olukhulu (LLMs), i-Retrieval-Augmented Generation (RAG)), ubunjineli besoftware (Python, Rust, Go), kunye noyilo lwezakhiwo ukudala iinkqubo ezinamandla nezigcinakalayo.
Uyilo lweModyuli yokuHamba kokuSebenza
Uyilo lwemodyuli yokuhamba komsebenzi yipatheni yoyilo eqhekeza iinkqubo ezinzima kwiimodyuli ezincinci, ezizimeleyo ezinokugcinwa lula, zihlaziywe, kwaye zilinganiswe. Imodyuli nganye inoxanduva lomsebenzi othile okanye umsebenzi othile, kwaye bayanxibelelana ngojongano oluchazwe kakuhle.
Nanku umzekelo olula woyilo lwemodyuli yokuhamba komsebenzi kwiPython:
# ukuhamba komsebenzi.py
Ukuhamba kweklasi:
def __init__(self):
self.modules = []
def add_module(ukwakho, imodyuli):
self.modules.append(modyuli)
def run(self):
yemodyuli kwi-self.modules:
imodyuli.run()
# imodyuli1.py
iklasi yeModyuli1:
def run(self):
print("Imodyuli yoku-1 iyasebenza")
# imodyuli2.py
iklasi yeModyuli2:
def run(self):
print("Imodyuli yesi-2 iyasebenza")
# engundoqo.py
ukuba __igama__ == "__main__":
ukuhamba komsebenzi = ukuhamba komsebenzi ()
ukuhamba komsebenzi.yongeza_imodyuli(Imodyuli1())
ukuhamba komsebenzi.yongeza_imodyuli(Imodyuli2())
workflow.run()
Kulo mzekelo, sineklasi yokuqukuqela komsebenzi elawula uluhlu lweemodyuli. Imodyuli nganye inoxanduva lomsebenzi othile, kwaye zongezwa kwindlela yokusebenza kusetyenziswa yokongeza_imodyuli indlela. Indlela yokuqhuba isetyenziselwa ukuphumeza imodyuli nganye kuhambo lomsebenzi.
Iiarhente ze-AI
Iiarhente ze-AI zizinto zesoftware ezisebenzisa i-algorithms yokufunda koomatshini ukwenza izigqibo kunye nokuthatha amanyathelo kwinkqubo. Zingasetyenziselwa ukwenza imisebenzi, ukuphucula amava omsebenzisi, kunye nokwandisa ukusebenza kwenkqubo.
Kukho iintlobo ezininzi zeearhente ze-AI, kubandakanya:
- Iiarhente ezisekelwe kumthetho: Ezi arhente zisebenzisa imithetho echazwe kwangaphambili ukwenza izigqibo.
- Iiarhente zokufunda ngomatshini: Ezi arhente zisebenzisa i-algorithms yokufunda ngomatshini ukufunda kwiidatha kwaye zenze izigqibo.
- Iiarhente zeHybrid: Ezi arhente zidibanisa iindlela ezisekelwe kumgaqo kunye nokufunda koomatshini ukwenza izigqibo.
Nanku umzekelo we-arhente ye-AI elula kwiPython usebenzisa i-algorithm yokufunda ngomatshini:
# iarhente.py
ukusuka sklearn.ensemble yokungenisa RandomForestClassifier
ukusuka sklearn.model_selection yokungenisa train_test_split
ukusuka sklearn.datasets import load_iris
Iarhente yeklasi:
def __init__(self):
self.model = RandomForestClassifier ()
def uloliwe(uqobo, X, y):
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
self.model.fit(X_train, y_train)
def qikelela(ukwakho, X):
buyisela self.model.predict(X)
# engundoqo.py
ukuba __igama__ == "__main__":
iris = load_iris ()
X = iris.data
y = iris.ithagethi
iarhente = iarhente ()
ummeli.train(X, y)
uqikelelo = iarhente.predict(X)
print(iziqikelelo)
Kulo mzekelo, sine-arhente ye-AI elula esebenzisa i-random classifier yehlathi ukwenza uqikelelo. I-arhente iqeqeshwe kwi-iris dataset kwaye yenza uqikelelo kwidathasethi efanayo.
Ukudibanisa ii-Agents ze-AI kunye neModyuli yoLwakhiwo lokuhamba komsebenzi
Ukudibanisa ii-arhente ze-AI kunye nemodyuli yoyilo lokuhamba komsebenzi kusivumela ukuba senze iinkqubo ezintsonkothileyo ezidibanisa ukuthathwa kwezigqibo eziqhutywa yi-AI kunye nemodyuli, uyilo olunokwehla.
Nanku umzekelo wenkqubo edibanisa iiarhente ze-AI kunye nemodyuli yoyilo lokuhamba komsebenzi:
# ukuhamba komsebenzi.py
Ukuhamba kweklasi:
def __init__(self):
self.modules = []
self.agent = Akukho
def add_module(ukwakho, imodyuli):
self.modules.append(modyuli)
def add_agent(self, agent):
self.agent = iarhente
def run(self):
yemodyuli kwi-self.modules:
imodyuli.run()
ukuba uzenzele.
self.agent.predict()
# imodyuli1.py
iklasi yeModyuli1:
def run(self):
print("Imodyuli yoku-1 iyasebenza")
# imodyuli2.py
iklasi yeModyuli2:
def run(self):
print("Imodyuli yesi-2 iyasebenza")
# iarhente.py
ukusuka sklearn.ensemble yokungenisa RandomForestClassifier
ukusuka sklearn.model_selection yokungenisa train_test_split
ukusuka sklearn.datasets import load_iris
Iarhente yeklasi:
def __init__(self):
self.model = RandomForestClassifier ()
def uloliwe(uqobo, X, y):
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
self.model.fit(X_train, y_train)
def qikelela(ukwakho, X):
buyisela self.model.predict(X)
# engundoqo.py
ukuba __igama__ == "__main__":
ukuhamba komsebenzi = ukuhamba komsebenzi ()
ukuhamba komsebenzi.yongeza_imodyuli(Imodyuli1())
ukuhamba komsebenzi.yongeza_imodyuli(Imodyuli2())
iarhente = iarhente ()
ummeli.train(X, y)
ukuhamba komsebenzi.yongeza_iarhente(iarhente)
workflow.run()
Kulo mzekelo, sinenkqubo edibanisa ii-arhente ze-AI kunye ne-modular workflow architecture. Inkqubo ineemodyuli ezimbini kunye ne-arhente ye-AI esebenzisa i-random classifier yehlathi ukwenza izibikezelo. I-arhente yongezwa kwindlela yokusebenza kusetyenziswa indlela `yokongeza_iarhente, kwaye yenza uqikelelo emva kokuba iminqongo iphelile.
Top 5-10 Uninzi Imiyalelo / iziqwengana ezisetyenziswayo
Nantsi eminye yemiyalelo edla ngokusetyenziswa kunye neziqwengana ukuyila iinkqubo zesoftware eziqhutywa yi-AI ezinobunzima kunye nenkqubo yesoftware eqhutywa ngokwemodyuli kunye neearhente ze-AI:
- Python:
ngenisa iipanda njenge pd(ukungenisa ngaphandle ilayibrari yepandas) - Python:
pd.read_csv('data.csv')(ukufunda ifayile ye-CSV kwi-pandas dataframe) - Python:
agent.train(X, y)(uqeqesho lwe-arhente ye-AI kwiseti yedatha) - Python:
iarhente.predict(X)(ukwenza uqikelelo usebenzisa i-arhente ye-AI) - Umhlwa:
vumela i-arhente ye-mut = I-arhente::entsha()(ukudala i-arhente ye-AI entsha kwi-Rust) - Umhlwa:
i-agent.train(X, y)(uqeqesho lwe-arhente ye-AI kwiseti yedatha e-Rust) - Hamba:
ummeli := i-NewAgent()(ukwenza i-arhente entsha ye-AI kwi-Go) - Hamba:
ummeli.Train(X, y)(uqeqesho lwe-arhente ye-AI kwisethi yedatha kwi-Go) - Itshathi.js:
itshathi.data.iiseti zedatha[0].data.push(30)(ukongeza indawo entsha yedatha kwitshati usebenzisa iChart.js)
Umboniso
Nanku umzekelo wetshathi yebha usebenzisa iChart.js:
{
"type": "bar",
"idatha": {
"Labels": ["Ileyibhile 1", "Ileyibhile 2"],
"iiseti zedatha": [
{
"label": "Iseti yedatha 1",
"idatha": [10, 20]
}
]
}
}
Le tshati ineeleyibhile ezimbini kunye nedatha enye eneendawo ezimbini zedatha. Itshathi ingenziwa ngokwezifiso usebenzisa iinketho ezahlukeneyo kunye nezicwangciso.
Isiphelo
Ukuyila iisistim zesoftware eziqhutywayo kunye nezisesikweni eziqhutywa yi-AI ezinoyilo lwemodyuli yokuhamba komsebenzi kunye neearhente ze-AI zifuna ukuqonda okunzulu kwenkqubela phambili yamva nje kwi-AI, ubunjineli besoftware, kunye nolwakhiwo. Ngokudibanisa iiarhente ze-AI kunye nemodyuli yoyilo lokuhamba komsebenzi, sinokwenza iinkqubo ezintsonkothileyo ezidibanisa ukwenza izigqibo eziqhutywa yi-AI kunye nemodyuli, uyilo olunokwehla.
Kweli nqaku, siye saphonononga imigaqo yoyilo kunye nezona ndlela zilungileyo zokuyila iimodyuli, ezinobungozi, kunye neenkqubo zesoftware yeenqobo ezisesikweni ezidibanisa iiarhente ze-AI kunye noyilo lwemodyuli yokuhamba komsebenzi. Siphinde sanikezela ngemizekelo yee-arhente ze-AI ezilula kunye ne-modular workflow architecture kwi-Python, i-Rust, kunye ne-Go, kunye nomzekelo wokuboniswa usebenzisa i-Chart.js.
Ngokulandela izikhokelo kunye nezenzo ezilungileyo ezichazwe kweli nqaku, abaphuhlisi banokudala iisistim zesoftware eziqhutywa yi-AI ezomeleleyo kwaye zigcinwe ezihlangabezana neemfuno zezicelo zanamhlanje kunye namashishini.