Skip to content
Get packs

PyTorch Lightning Training

Guides an agent to write and scale PyTorch training code using Lightning's Trainer, callbacks and distributed strategies.

Source Download skill
Repository
Orchestra-Research/AI-Research-SKILLs
Stars
14,195
Licence
MIT
Files
4 files

What it does

When installed, the agent gets guidance on PyTorch Lightning, a popular framework that organizes model training code and removes repetitive boilerplate. The agent can convert plain PyTorch training loops into Lightning modules, add validation and testing, set up checkpointing and early stopping, configure learning rate schedules, and switch the same code between one GPU and many GPUs or multiple machines using DDP, FSDP or DeepSpeed. The skill is mainly reference material and code examples that the agent follows when writing your training code.

Great for

  • Machine learning engineers who want cleaner training code that scales from a laptop to multiple GPUs
  • Researchers converting an existing PyTorch training script to a more standard structure
  • Teams that want consistent, reproducible training setups

Skip it if

  • People who have never trained a model: the skill assumes working knowledge of PyTorch, datasets and GPUs
  • Projects not using PyTorch (for example TensorFlow or JAX), where the guidance does not apply
  • Production jobs run without review: generated distributed training configurations should be tested on a small run first, since mistakes can waste expensive GPU time

Try asking

  • “Convert my PyTorch training loop into a PyTorch Lightning LightningModule with validation metrics”
  • “Set up checkpointing and early stopping callbacks for my image classifier”
  • “Configure my Lightning Trainer to train on 8 GPUs using DDP”
  • “Add a cosine learning rate scheduler to my existing Lightning model”
Step-by-step guide: how to use PyTorch Lightning TrainingInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 4 original files, the MIT licence and an attribution note.

    pytorch-lightning.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip pytorch-lightning.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip pytorch-lightning.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Give the agent your existing training script and state your hardware (number of GPUs, nodes) so it picks the right Trainer settings
  • Ask the agent to run a short test with max_epochs=1 or a small data subset before launching a full distributed job
  • Check the library version, since Lightning APIs change between releases; the skill's examples may not match your installed version

What's inside

  • references/callbacks.md
  • references/distributed.md
  • references/hyperparameter-tuning.md
  • SKILL.md

Redistributed unmodified from Orchestra-Research/AI-Research-SKILLs under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.