Uqeqesho nokusebenza kwe-PyTorch ku-NVIDIA Tesla P40 (24 GB VRAM, Pascal architecture), nge-root access egcwele. Amaseva we-GPU angeke akwazi uku-oda; xhumana nathi ukuze uqinisekiswe uma bekhona.
$ pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124
# Isibonelo sokufaka umyalo
I-PyTorch iyisimo sokufunda okunzulu esisetshenziswa kakhulu. Isisebenzisi se-GPU siqhuba uqeqesho lwe-PyTorch kanye nokucabanga ku-GPU, okushesha kakhulu kunakwe-CPU kumamodeli amaningi.
Ufaka umqhubi we NVIDIA, CUDA kanye ne-PyTorch edingekayo. Khangela ukuthi i-PyTorch yakho ixhasa i-Pascal GPUs (ikhono lokulinganisa 6.1).
Indawo yamamodeli amancane namaphakathi, LoRA okuhlelwa kahle kanye nobukhulu obuphakathi bohlelo.
Sebenzisa ama-notebooks we-Jupyter kuseva ukuze usebenze ngokuzenzekelayo ku-GPU.
Misa isimo ngokugcwele njengoba iphrojekthi yakho idinga.
PyTorch ingasebenzisa i-NVIDIA GPU ukusheshisa umsebenzi wayo. VPS.org GPU amaseva akhona ukuthenga; xhumana nathi ukulalela lapho. PyTorch uqeqesho / ubunzima bomsebenzi obuncane. Izinhlelo zemisebenzi ehamba isikhathi eside: thola isithombe se-VPS yakho njalo futhi ugcine ikhophi yesisindo esiqeqeshiwe ngaphandle kwe-VPS.org.
Amaseva we-GPU akhona ukucela okwamanje. Namuhla ungasebenzisa i-PyTorch ku-CPU ejwayelekile, lapho isebenza khona ngokucophelela kunakwi-GPU: SSH ngaphakathi bese usebenzisa i-pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124.
Ukuqeqeshwa kwe-VRAM kulawulwa yisimo se-optimise plus activations. Ukuthuthukiswa okuphelele kwe-7B LLM kudinga cishe i-24 kuya ku-48 GB; LoRA / QLoRA ilungele i-8 kuya ku-16 GB.
Intengo ye-GPU izoshicilelwa uma izisebenzisi ze-GPU zingasatshalaliswa. Xhumana nathi uma ufuna ukulalela uma lokhu kwenzeke.
Yebo. Yonke iseva ye-VPS.org ifika nge-root access egcwele, ngakho ungasebenzisa ezinye izisetshenziswa kanye ne-PyTorch ngaphakathi kwe-RAM ne-storage yeseva.
Yonke isisebenzisi se VPS.org sifika sigcwele ukungena kwe-root SSH, ngakho ungafaka ama-driver, ama-library ne PyTorch noma ngabe ufuna.
Akukho umsekeli ofika ngesikhathi sokusebenza se-CUDA noma umqhubi we-NVIDIA ofakwe ngaphambili, futhi ama-servers we-GPU atholakalanga ukuthenga okwamanje. Ungafaka izibuyekezo PyTorch ezidingayo wena.
Yebo. Amaseva we-VPS.org asebenza ngama-machines aqhubekayo, hhayi izimo ezihambayo: amamodeli we-PyTorch, izilungiselelo nedatha zihlala ku-SSD yokugcina yeseva kuze kube ucima.
Gcina idatha esebenza kuseva ye-SSD ukuze ufinyelele ngokushesha ngesikhathi se-PyTorch, futhi ukope izitho eziqediwe (izitho, izizukulwane, izingeniso) ukugcinwa ngaphandle kwe-VPS.org; ama-snapshots nama-backup ahlala ku-storage cluster efanayo.
Ungakhuphula noma iyiphi iseva ye-VPS.org ibe yi-plan enkulu kusuka kwindawo yokulawula; ukushintsha ubukhulu kuqala kabusha futhi ukufakwa kwe-PyTorch kudlulisa. Amaseva we-GPU akhona.
Yebo. Amafayela okugcina akhokhelwa angeza-on (ngesonto +30% yentengo yephrojekthi, igcinwe 7 izinsuku; ngeviki +20%, igcinwe 28 izinsuku) futhi ama-snapshots amahhala, okusiza ngaphambi kokuhamba isikhathi eside PyTorch. Kokubili kuhlale ku-indawo yokugcina efanayo njengeseva.
Xhumana nosizo ngaphakathi kwezinsuku ezingama-30 zokuthenga kwakho okukuqala ukuze ucele ukubuyiselwa imali. Amaseva we-GPU athengiswe, ngakho-ke akukho sinqumo se-GPU esihlobene.
Ikhono le-GPU lilinganiselwe. Sitshele ngomsebenzi wakho futhi sizophendula ngokutholakala nokuthengwa.