Kudzidzisa uye kufambisa PyTorch mamodheru paNVIDIA Tesla P40 (24 GB VRAM, Pascal architecture), neruzivo rwe root. GPU servers hazvigone kupihwa; taura nesu kuti uzive kana zvavepo.
$ pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124
# Mufananidzo wekuisa command
PyTorch inoshandisa pfungwa yekuti pfungwa inofanira kuiswa panzvimbo yechirongwa, uye kuti pfungwa inofanira kuiswa panzvimbo yechirongwa. GPU server inofambisa PyTorch kudzidzisa uye kuongorora paGPU, iyo iri nyore kupfuura paCPU.
Unogona kuisa NVIDIA driver, CUDA uye PyTorch build iwe yaunoda. Cheka kuti PyTorch build yako ichiri kutsigira Pascal GPUs (compute capability 6.1).
Room for small and mid-sized models, LoRA fine-tuning and moderate batch sizes.
Kushanda Jupyter notebooks pa server ye interaction basa pa GPU.
Kugadzirisa mamiriro ekunze zvakajeka sezvako project zvinoda.
PyTorch anogona kushandisa NVIDIA GPU kuti vaite nhanho yake basa. VPS.org GPU servers hazvina kuwanikwa kuti aite; taura nesu kuti uzive kana ivo. PyTorch ndeimwe kudzidzisa / fine-tuning workload. Plan for long-running jobs: snapshot your VPS regularly and keep a copy of trained weights outside VPS.org.
GPU servers hadzina kuwanikwa kuti utore. Nhasi unogona kuisa PyTorch pane yakajairika CPU package, uko inoita basa rakakura kupfuura pane GPU: SSH in uye shandisa pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124.
Kudzidzisa VRAM kunotungamirirwa neoptimizer state plus activation. Full fine-tuning ye7B LLM inodiwa zvakapetwa kaviri 24 kusvika 48 GB; LoRA / QLoRA inokwana 8 kusvika 16 GB.
GPU mutengo achaburitswa kana GPU servers anogona kupihwa. Tibate kana iwe uchida kuziva kana ichiitika.
Yega. Every VPS.org server inouya nefull root access, saka unogona kutamba zvimwe zvinhu pamwe PyTorch mukati server's RAM uye kuchengetera.
Every VPS.org server inouya nepfupi root SSH kuwanikwa, saka unogona kuisa madhiraivha, mabhuku uye PyTorch asi iwe unofanirwa.
Hapana server inouya neCUDA runtime kana NVIDIA mutyairi akaiswa, uye GPU servers hazvina kuwanikwa kuti aite kusvika parizvino. Unogona kuisa zvinyorwa PyTorch zvinoda iwe pachako.
VPS.org servers ndezvemapurojekiti akaomarara, kwete ephemeral instances: PyTorch models, configs uye data inogara paSSD storage yeserver kusvika wabvisa.
Tenga data rekushandisa pane SSD ye server yekuwedzera kuwanikwa kwenguva pfupi panguva ye PyTorch runs, uye kopa zvakapera zviwanikwa (zviyero, zvizvarwa, embeddings) kune yekuchengetera kunze kwe VPS.org; snapshots uye backups vanogara pamwechete kuchengetera cluster.
Unogona kusimudzira chero VPS.org server kune yakakura chirongwa kubva kudzora panel; resize rinotangazve uye yako PyTorch kuiswa inotora pamusoro. GPU servers hazvina kuwanikwa kusvika parizvino.
Yeah. Backups ndivo akabhadhara add-on (zuva nezuva + 30% yemari chirongwa, akachengetwa 7 mazuva; zuva nezuva + 20%, akachengetwa 28 mazuva) uye snapshots vakasununguka, izvo zvinokosha pamberi PyTorch mazuva. Both stay on the same storage cluster as the server.
Contact support within 30 days of your first purchase to request a refund. GPU servers are not sold yet, so no GPU order is involved.
GPU kugona kuri kuwanikwa chete. Tiudzai nezvebasa rako uye tichakuudza kuti ripi rinowanikwa uye ripi risiri.