Rent and manage on-demand vast.ai GPU instances from task requirements.
npx skills add https://github.com/Shallow-W/llm-wiki --skill vast-gpu-shallow-w
Please help me install this Agent Skill. Skill: vast-gpu Source: https://github.com/Shallow-W/llm-wiki/tree/main/.claude/skills/vast-gpu Command: npx skills add https://github.com/Shallow-W/llm-wiki --skill vast-gpu-shallow-w
This Skill eliminates the time-consuming process of choosing GPU hardware by automatically estimating GPU needs, finding cost-optimized offers on vast.ai, and managing the full rent-to-destroy lifecycle.
vast-instances.json state file to support /run-experiment and /monitor-experiment workflows.Trigger GPU renting by asking for an on-demand GPU for your task, for example: "Rent a cloud GPU to run my experiment, using vast ai."
High-intent search queries and answers about installing and using this skill.
You can rent a cloud GPU by estimating VRAM, disk, and runtime needs from your experiment plan or scripts, allowing the system to find cost-optimized offers automatically. This eliminates manual GPU model selection by deriving hardware requirements directly from task descriptions.
You need vastai CLI authentication with Python ≥ 3.10 and pre-uploaded SSH keys to provision vast.ai on-demand GPU instances. These prerequisites enable automated instance provisioning, secure access, and lifecycle management.
Cost optimization for on-demand GPUs works by searching broad GPU tiers and ranking options using estimated total cost, which includes runtime hours, rather than just hourly rates. This ensures economical choices by evaluating reliability and speed alongside price.
To manage the lifecycle of a rented GPU instance and stop billing, you provision the instance, sync code, run the experiment, and then destroy the instance after downloading results. This controlled lifecycle halts billing immediately after task completion.
Yes, you can use existing scripts, experiment plans, or user descriptions to estimate GPU VRAM, number of GPUs, disk space, and CUDA/PyTorch compatibility. The system derives hardware requirements directly from these inputs for automated sizing.
dogekikiAutomate vast.ai GPU instance provisioning, configuration, and cleanup via vastai CLI.
duypham2801Rent, configure, and destroy vast.ai GPU instances for ML training workloads.
zhuyingqinProvision and manage vast.ai GPU instances for training tasks.
hve4638Rent and manage Vast.ai GPU instances for ML tasks.