Iimodeli ze-Llama ze-Meta ezimququzeleli-yodwa kwi-NVIDIA Tesla P40 (24 GB VRAM, uyilo lwePascal). Zilungele iimodeli ze-8B ngokuchanekileyo okupheleleyo kunye neemodeli ezinkulu xa zigqityiwe. Abancedisi be-GPU abakwazi ukuthengwa; qhagamshelana nathi ukuze sikwazisa xa bekhona.
$ curl -fsSL https://ollama.com/install.sh | sh && ollama run llama3
# Umzekelo wofakelo lomyalelo
I-Llama yiMeta's family of openly available large language models. Ukusebenza kwe-Llama kumncedisi wakho we-GPU ugcina izilumkiso ne-data ifihlakeleyo.
Iqhuba iimodyuli ze-8B kwi-24 GB ye-VRAM. Iimodyuli ze-70B azifakwanga kwikhadi elinye le-24 GB kumanqanaba okuphindaphinda aqhelekileyo.
Iinkcukacha zakho zobuchule nedata zihlala kwiseva yakho, ngaphandle kokunye unxibelelwano lwe-API.
Sebenzisa i-GGUF ne-lama.cpp okanye i-Ollama, okanye i-PyTorch eqhelekileyo.
Ukubonelela ngeemodeli nge-Ollama okanye iseva ye-lama.cpp yendawo ephelileyo ehambelana ne-OpenAI.
LLaMA inokusebenzisa i NVIDIA GPU ukukhawulezisa umsebenzi wayo. VPS.org GPU abancedisi abafumanekanga ukucela ngoku; qhagamshelana nathi ukuva xa bekhona. LLaMA yi GPU- ekhawulezayo yomsebenzi oqhelekileyo. Qinisekisa ukuba udweliso lwenkqubo luxhaswe yi CUDA kwaye ukuba umqhubi wakho / uguqulelo lwexesha lokusebenza luhambelana neemfuno zomsebenzi womsebenzi we LLaMA.
Abancedisi be-GPU abafumanekanga ukucela ngoku. Namhlanje ungaqhuba i-LLaMA kwi-CPU eqhelekileyo, apho isebenza khona ngokucothayo kune-GPU: SSH ngaphakathi kwaye uqhube i-curl -fsSL https://ollama.com/install.sh | sh && ollama run llama3.
Iimfuneko ze-VRAM ze-LLaMA zixhomekeke kwimodeli kunye nobungakanani beqela; khangela ushicilelo lweprojekthi yemifanekiso yangoku.
Ixabiso le-GPU liza kupapashwa xa iiseva ze-GPU zingafunyanwa. Xhulumanisa nathi ukuba ufuna ukuva xa oku kwenzeka.
Ewe. Isincedisi ngasinye se-VPS.org sifika sinee-root access ezipheleleyo, ngoko ungaqhuba ezinye izixhobo kunye ne-LLaMA ngaphakathi kwe-RAM yeseva kunye nokugcina.
Yonke i VPS.org umncedisi ifika nge root SSH unikezelo olupheleleyo, ngoko ungafaka abaqhubi, iincwadi ze LLaMA nokuba ufuna njani.
Hayi. Akukho mncedisi ufika nexesha lokuqhuba le-CUDA okanye umqhubi we-NVIDIA ofakwe ngaphambili, kwaye abancedisi be-GPU abafumanekanga ukucela ngoku. Ufaka uguqulelo lwe-LLaMA olufunayo ngokwakho.
Ewe. Abancedisi be-VPS.org basebenza njengezimatshini ezibonakalayo eziqhubekayo, azikho iziganeko ezikhawulezayo: iimodyuli ze-LLaMA, iinkqubo zokumiselwa kwenkqubo kunye nedata zihlala kwi-SSD yokugcina yeseva de uzicime.
Gcina i-data esebenza kwi-SSD yomncedisi ukufikelela ngokukhawuleza ngexesha lokusebenza kwe-LLaMA, kwaye ukope i-artifacts egqityiweyo (isisindo, u Generations, i-embeddings) ukugcinwa ngaphandle kwe-VPS.org; iifoto ezikhawulezayo kunye nokhuseleko luhlala kwiqela elinye lokugcina.
Unganyusa nayiphi na iseva ye-VPS.org ibe yinkqubo enkulu ukusuka kwiqela lenjongo yolawulo; ubungakanani obutsha buqala kwakhona kwaye ukufaka kwe-LLaMA kugqitha. Iiseva ze-GPU azifumanekanga ngoku.
Ewe. Iindawo zokugcina zihlawulwe zidityaniswe (ngosuku +30% yexabiso lenkqubo, igcinwe iintsuku ezi-7; ngeveki +20%, igcinwe iintsuku ezi-28) kwaye iifoto ezikhawulezayo zikhululekile, eziluncedo phambi kwe LLaMA ende iqhuba. Zonke zihlala kwiqela lokugcina elifanayo njengeseva.
Nxibelelana noncedo ngaphakathi kweentsuku ezingama-30 zokuthengiswa kwakho kokuqala ukuze ucele ubuyiselo. Iiseva ze-GPU azithengwanga, ngoko akukho mfuneko yokuthumela i-GPU.
Ubungakanani be-GPU bumiselwe. Sixelele malunga nobunzima bomsebenzi wakho kwaye siza kuphendula ngokufumaneka nokuthengiswa.