Deep Learning Academy · 课时

使用 torchrun 启动任务

生成并协调工作进程

使用 torchrun 启动任务 是 CoddyKit 上的免费 Deep Learning Academy 课时。这是第 4 节课，共 4 节。你可以在下方免费阅读本课时的完整内容 — 然后在浏览器中使用内置代码编辑器和全天候 AI 导师进行实践。这是 Deep Learning Academy 学习路径的一部分，你的进度在网页和 CoddyKit 应用中同步。 Deep Learning Academy 课程共包含 4 节课。

本课时的部分内容尚未翻译，以英文显示。

Who Starts the Processes

DDP needs one process per GPU, but who spawns them? You do not launch them by hand. torchrun is the tool that does it for you.

Meet torchrun

torchrun is PyTorch's launcher. You give it your script and how many processes to start, and it handles the coordination.

torchrun --nproc_per_node=4 train.py

Processes per Node

The nproc_per_node flag sets how many processes to spawn on this machine. Usually it equals the number of GPUs you have.

torchrun --nproc_per_node=8 train.py

Environment Variables Appear

torchrun injects key values as environment variables: RANK, LOCAL_RANK, and WORLD_SIZE. Your script reads them to know who it is.

import os
local_rank = int(os.environ["LOCAL_RANK"])

No Manual Address Wiring

Older launchers made you pass ranks by hand. With torchrun, the rendezvous is automatic, so your script stays clean and portable.

Init Reads the Env

Because torchrun sets the env vars, your init_process_group call needs no arguments beyond the backend. It picks everything up automatically.

import torch.distributed as dist
dist.init_process_group(backend="nccl")

Scale to Many Machines

To go multi-node, add nnodes and a node rank. Each machine runs the same command with its own node id.

torchrun --nnodes=2 --node_rank=0 --nproc_per_node=4 train.py

Point to the Master

Across nodes, all processes must find a meeting point. The rdzv_endpoint gives the host and port they rendezvous on.

torchrun --rdzv_endpoint=host0:29500 train.py

Survive a Worker Crash

torchrun supports elastic training: set a min and max node count, and the job can recover if a worker drops out. 💪

torchrun --nnodes=1:4 --max_restarts=3 train.py

Log From One Rank

Every process prints, so logs get noisy. Gate prints behind a rank check so only rank 0 reports progress.

if int(os.environ["RANK"]) == 0:
    print("epoch done")

The Whole Recipe

The pattern is simple: write a normal DDP script, then launch it with torchrun. The launcher and DDP handle the rest together.

Quick Check

Recall what torchrun hands to your script.

Recap

You learned to launch jobs with torchrun: set processes per node, read the injected env vars, and scale to many nodes with elastic recovery.

常见问题解答

「使用 torchrun 启动任务」课时是免费的吗？

是的 — 「使用 torchrun 启动任务」的完整文本可在网页上免费阅读。要进行交互式练习（内置代码编辑器和全天候 AI 导师）并解锁 Deep Learning Academy 课程的其余内容，请升级到 CoddyKit PRO。 Deep Learning Academy 课程共包含 4 节课。

「使用 torchrun 启动任务」这节课中我会学到什么？

生成并协调工作进程你通过在浏览器中直接运行的动手代码来练习 Deep Learning Academy，全天候 AI 导师会在你学习这节课的过程中回答你的问题。

学习 Deep Learning Academy 需要有经验吗？

无需任何先前经验。CoddyKit 上的 Deep Learning Academy 课程适合初学者到高级学习者，你可以从这里开始或从头开始，按照自己的节奏学习。这是第 4 节课，共 4 节。

「使用 torchrun 启动任务」课时需要多长时间？

大多数 CoddyKit 课程大约需要 5–10 分钟。每节课都很精短且互动，所以你能稳步进步，并在网页和应用中从离开的地方继续。

我能在这节 Deep Learning Academy 课中编写并运行代码吗？

能。每节 Deep Learning Academy 课都包含内置代码编辑器，你可以在浏览器中直接编写并运行真实代码，并获得即时 AI 反馈 — 无需本地设置。

此课程中的所有课时

← 返回 Deep Learning Academy