Mokhoa oa ho sebelisa DeepSeek ho Visual Studio Code

DeepSeek R1 e hlahile e le mokhoa o matla le o lokolohileng ho litharollo tse ling. Thepa ea eona e ntle ka ho fetisisa ke hore e lumella bahlahisi ho ba le AI e tsoetseng pele bakeng sa thuso ea khoutu ntle le ho itšetleha ka li-server tsa leru. Sehloohong sena re u hlalosetsa Mokhoa oa ho sebelisa DeepSeek ho Visual Studio Code.

'Me ke hore, ka lebaka la ho fumaneha ha eona liphetolelong tse ntlafalitsoeng bakeng sa phethahatso ya lehae, ho kopanngoa ha eona hoa khoneha ntle le litšenyehelo tse eketsehileng. Sohle seo u lokelang ho se etsa ke ho sebelisa lisebelisoa tse joalo Ollama, LM Studio le Jan, hammoho le ho kopanngoa le li-plugins tse kang CodeGPT le Cline. Re u bolella tsohle lirapeng tse latelang:

DeepSeek R1 ke eng?

Joalokaha re se re hlalositse mona, DeepSeek R1 Ke mohlala oa puo ea mohloli o bulehileng e qothisanang lehlokoa le litharollo tsa khoebo tse kang GPT-4 mesebetsing ea ho beha mabaka ka mokhoa o utloahalang, tlhahiso ea khoutu le ho rarolla mathata a lipalo. Molemo oa eona o ka sehloohong ke hore e ka tsamaisoa sebakeng sa heno ntle le ho itšetleha ka li-server tsa kantle, e fana ka boemo bo phahameng ba lekunutu bakeng sa bahlahisi.

Ho ipapisitsoe le lisebelisoa tse fumanehang, mefuta e fapaneng ea mohlala e ka sebelisoa, ho tloha ho li-parameter tsa 1.5B (bakeng sa likhomphutha tse itekanetseng) ho isa ho li-parameter tsa 70B (bakeng sa li-PC tse sebetsang hantle tse nang le li-GPU tse tsoetseng pele).

Mekhoa ea ho Run DeepSeek ho VSCode

Ho finyella tshebetso e ntle ka ho fetisisa ka DeepSeek en Code ea Visual Studio, ho bohlokoa ho khetha tharollo e nepahetseng ea ho e tsamaisa ho sistimi ea hau. Ho na le likhetho tse tharo tse kholo:

Khetho ea 1: Ho sebelisa Ollama

Ollama Ke sethala se bobebe se u lumellang ho tsamaisa mefuta ea AI sebakeng sa heno. Latela mehato ena ho kenya le ho sebelisa DeepSeek le Ollama:

1. Khoasolla le ho kenya Ollama ho tsoa webosaeteng ea eona ea semmuso (ollama.com).
2. Ho terminal, tsamaisa: ollama pull deepseek-r1:1.5b (bakeng sa mehlala e bobebe) kapa phapang e kholoanyane haeba hardware e e lumella.
3. Ha e se e jarollotsoe, Ollama o tla amohela mohlala ho http://localhost:11434, e etsa hore e fumanehe ho VSCode.

Khetho ea 2: Ho sebelisa LM Studio

LM Studio ke mokhoa o mong oa ho jarolla le ho laola mefuta ena ea lipuo habonolo (le ho sebelisa DeepSeek ho Visual Studio Code). Mokhoa oa ho e sebelisa ke ona:

1. Taba ea pele, download LM Studio ebe o e kenya ho sistimi ea hau.
2. Batla 'me u khoasolle mohlala DeepSeek R1 ho tloha tab Discover.
3. Kenya mohlala 'me u nolofalletse seva sa lehae ho tsamaisa DeepSeek ho Visual Studio Code.

Khetho ea 3: Ho Sebelisa Jan

Khetho ea boraro eo re e khothalletsang ke Jan, mokhoa o mong o sebetsang oa ho tsamaisa mefuta ea AI sebakeng sa heno. Ho e sebelisa, o tlameha ho etsa se latelang:

• Khoasolla pele mofuta oa Jan e tsamaisanang le sistimi ea hau ea ts'ebetso.
• Ebe u khoasolla DeepSeek R1 ho Hugging Face ebe u e kenya ka Jan.
• Qetellong, qala seva ho kena http://localhost:1337 'me ue behe ho VSCode.

Haeba u batla ho ithuta haholoanyane ka mokhoa oa ho sebelisa DeepSeek libakeng tse fapaneng, ikutloe u lokolohile ho sheba tataiso ea rona ho DeepSeek ho Windows 11 tikoloho.

Ho kopanya DeepSeek le Visual Studio Code

Hang ha u se u e-na le eona DeepSeek sebetsa sebakeng sa heno, ke nako ea ho e kopanya ho Code ea Visual Studio. Ho etsa sena, o ka sebelisa li-extensions joalo ka CodeGPT o qoela.

Ho lokisa CodeGPT

1. Ho tsoa ho tabo Extensions Ho VSCode (Ctrl + Shift + X), batla le ho kenya CodeGPT.
2. Kena ho li-setting tsa katoloso ebe u khetha Ollama joalo ka mofani oa LLM.
3. Kenya URL ea seva moo e sebetsang teng DeepSeek sebakeng sa heno.
4. Khetha mofuta o jarollotsoeng oa DeepSeek 'me u o boloke.

Ho lokisa Cline

qoela Ke sesebelisoa se shebaneng le ts'ebetso e ikemetseng ea khoutu. Ho e sebelisa le DeepSeek ho Visual Studio Code, latela mehato ena:

1. Khoasolla katoloso qoela ho VSCode.
2. Bula litlhophiso ebe u khetha mofani oa API (Ollama kapa Jan).
3. Kenya URL ea seva ea lehae moo e sebetsang teng DeepSeek.
4. Khetha mofuta oa AI 'me u netefatse litlhophiso.

Bakeng sa tlhaiso-leseling e batsi mabapi le ts'ebetsong ea DeepSeek, ke khothaletsa hore u hlahlobe Kamoo Microsoft e kopanyang DeepSeek R1 ho Windows Copilot, e ka u fang pono e batsi mabapi le bokhoni ba bona.

Malebela a ho Khetha Mohlala o Nepahetseng

El Ts'ebetso ea DeepSeek ho Virtual Studio Code e tla itšetleha haholo ka mohlala o khethiloeng le bokhoni ba hardware ea hau. Bakeng sa litšupiso, tafole e latelang e loketse ho shejoa:

Haeba komporo ea hau e se na matla, u ka khetha mefuta e menyenyane kapa mefuta e lekantsoeng ho fokotsa tšebeliso ea memori.

Joalo ka ha u bona, ho sebelisa DeepSeek ho Visual Studio Code ho re fa khetho e ntle, e sa lefelloeng ho bathusi ba bang ba khoutu e lefuoang. Monyetla oa ho e tsamaisa sebakeng sa heno ka Ollama, LM Studio o Jan, e fa bahlahisi monyetla oa ho rua molemo ho sesebelisoa se tsoetseng pele ntle le ho itšetleha ka litšebeletso tse thehiloeng marung kapa litšenyehelo tsa khoeli le khoeli. Haeba u theha tikoloho ea hau hantle, u tla ba le mothusi ea ikemetseng, ea matla oa AI tlas'a taolo ea hau.

Sehlooho se amanang:

DeepSeek: Ntho e ngoe le e ngoe eo u hlokang ho e tseba ka AI ea mahala ea boqapi

Daniel Terrasa

Mohlophisi ea ikhethileng litabeng tsa theknoloji le marang-rang ea nang le boiphihlelo bo fetang lilemo tse leshome mecheng ea litaba tse fapaneng tsa dijithale. Ke sebelitse ke le mohlophisi le moetsi oa dikahare bakeng sa khoebo ea e-commerce, puisano, papatso ea marang-rang le lik'hamphani tsa papatso. Ke ngotse hape ka liwebsaete tsa moruo, tsa lichelete le tsa makala a mang. Mosebetsi oa ka le ona ke takatso ea ka. Joale, ka lingoliloeng tsa ka ho Tecnobits, Ke leka ho hlahloba litaba tsohle le menyetla e mecha eo lefatše la theknoloji le re fang eona letsatsi le leng le le leng ho ntlafatsa bophelo ba rona.