remote-gpu-trainer

仓库创建 2026年6月23日最近提交 22 天前SkillHot 收录 22 天前
▸ 精选理由

专为短期租用机器设计,关注预占、计费和可恢复性。

▸ 风险提示

需远程访问与凭据,可能执行拉取/写入脚本

这个 Skill 做什么

在租用或远程 GPU 实例上部署、监控和抢救长时训练/推理作业并取回结果。

在租用或别人的远程 GPU(云平台、裸机、Slurm、K8s 等)上部署、监控并抢救长时训练或推理作业,把结果安全取回。适合训练、ablation、批量推理或遭遇抢占、OOM、断线等情况,提供 checkpoint、断点续训、上传/下载与节省成本的关机策略。核心是把工作与实例分离,让成果比机器更持久,防止计费或被抢占导致的数据丢失。

▸ 展开 SKILL.md 英文原文

Use whenever the user deploys, trains, monitors, or troubleshoots a long-running GPU job on a RENTED or remote instance they do not own — training, eval, ablation sweeps, batch inference, or large data processing — on AutoDL, RunPod, vast.ai, Lambda, Paperspace, Chinese platforms (恒源云/矩池云/Featurize/揽睿星舟), a bare SSH box, Slurm, or Kubernetes; single OR multi-instance. Triggers (multilingual): "远程 GPU 训练", "GPU 租赁", "GPU rental", "租卡", "spot 抢占", "spot preemption", "断点续训", "resumable training", "tmux 训练守护", "防 SSH 断线", "scp/rsync 上传", "多实例 ablation", "远程 GPU 监控", "省钱关机/销毁实例", "stop vs terminate billing", "checkpoint 磁盘满", "CUDA OOM/显存不足", "loss NaN/loss spike", "loss 不下降/不收敛", "overfit 单 batch", "FSDP/DeepSpeed 配置", "多卡训练 hang", "dataloader worker/数据增广 bug". NOT for purely local single-GPU training, in-instance multi-GPU DDP (use torchrun/accelerate), managed multi-cloud price-shopping (use SkyPilot's skill), or zero-ops serverless (use Modal).

开发编程GPU 管理远程训练作业编排通用
0
Stars
0
Forks
1
仓库内 Skill
+0
7 日增星
安装 / 使用
给你的 Agent 一句话(通用)
帮我安装这个 skill:https://raw.githubusercontent.com/Udalefourfold471/remote-gpu-trainer/main/SKILL.md
或 curl 直取 SKILL.md
curl -fsSL "https://raw.githubusercontent.com/Udalefourfold471/remote-gpu-trainer/main/SKILL.md"
SKILL.MD 节选查看完整文件 ↗
# remote-gpu-trainer — Remote GPU Job Orchestration

## Overview

Deploy and babysit long-running GPU jobs on **rented boxes you don't own**, across any platform, and
get the result off the box before the meter or a preemption kills it. The core insight: **you are a
short-term tenant on someone else's machine** — so the job is to *detach the work, make the result
outlive the instance, and stop the meter safely*, not to provision a cluster.

This skill is **platform-agnostic at the core, platform-specific at the edges**: a fixed set of
operating principles + a 6-phase lifecycle that hold everywhere, plus one **profile per platform**
(`profiles/<platform>.md`) that owns every concrete path, pr
via SKILL·HOT · 数据来自 GitHub 公开信息 · 原文版权归作者所有