Agent Skills › companion-inc/feynman › modal-compute

modal-compute

GitHub

用于在 Modal 无服务器基础设施上运行研究基准测试或模型复现任务，提供突发 GPU 算力支持。适用于无需持久状态、需快速启动的独立实验脚本执行场景。

skills/modal-compute/SKILL.md companion-inc/feynman

触发场景

需要运行独立的 GPU 训练或推理脚本进行模型复现或基准测试且需突发算力需要在云端快速执行无状态 Python 实验

安装

npx skills add companion-inc/feynman --skill modal-compute -g -y

更多选项

不安装直接使用

npx skills use companion-inc/feynman@modal-compute

指定 Agent (Claude Code)

npx skills add companion-inc/feynman --skill modal-compute -a claude-code -g -y

安装 repo 全部 skill

npx skills add companion-inc/feynman --all -g -y

预览 repo 内 skill

npx skills add companion-inc/feynman --list

SKILL.md

Frontmatter

{
    "name": "modal-compute",
    "description": "Run explicitly chosen research benchmark or replication jobs on Modal's serverless infrastructure. Use when a Feynman research workflow needs burst remote GPU compute and the Modal CLI is available."
}

Modal Compute

Use the modal CLI for bounded research experiments that need burst GPU compute. No pod lifecycle to manage; write a decorated Python script, run it, and save raw outputs back into the research artifact folder. Do not use this skill to deploy services or unrelated batch jobs.

Setup

pip install modal
modal setup

Commands

Command	Description
`modal run script.py`	Run one research experiment script on Modal
`modal run --detach script.py`	Run a long research experiment and record the returned app/run identifier
`modal shell --gpu a100`	Open an interactive GPU shell for research environment debugging

GPU types

T4, L4, A10G, L40S, A100, A100-80GB, H100, H200, B200

Multi-GPU: "H100:4" for 4x H100s.

Script pattern

import modal

app = modal.App("experiment")
image = modal.Image.debian_slim(python_version="3.11").pip_install("torch==2.8.0")

@app.function(gpu="A100", image=image, timeout=600)
def train():
    import torch
    # training code here

@app.local_entrypoint()
def main():
    train.remote()