Every premium and members-only resource for $9.99/month, as a no-commitment subscription. See the plans

PalateKit

Sign in
Menu

Request a resource

Describe the resource you are missing in one sentence. We will email you when it is published.

0 / 280

A page, a video or an image online.

Unlock PalateKit

Unlock everything

Plan

Paying means accepting our terms of sale.

Secure payment by Stripe. Receipts and invoices sent by email.

Welcome back

Sign in to install, download or save this resource. A free account is enough for free resources.

Adding FSDP2 to a PyTorch training script v1.0.0

Use this resource to add FSDP2 sharding to a training script without initialization errors

New

  • 596 downloads
  • 242 likes
  • 235 saves

Free

Install with the CLI

In a terminal, from the folder you want:

npx --yes palatekit@latest install pytorch-fsdp2-sharding-setup

With no option, the skill goes to ~/.agents/skills/; add --agent claude-code (or codex, cursor…) to target an agent.

First time: connect the CLI with

npx --yes palatekit@latest login

The agent follows a step-by-step procedure to initialize distributed computing, build the model on the meta device, apply fully_shard bottom-up, then configure mixed precision, offload and checkpoints. It also presents alternatives to FSDP2.

Mostly picked by

Updated November 2, 2026

Card for the “Adding FSDP2 to a PyTorch training script” skill

Details

Works in
Install guides
Generates with
Your agent's model
Licence
MIT

Similar resources

View all

What you get

  • SKILL.md: step-by-step procedure from initialization to optimizer
  • references/pytorch_fsdp2_tutorial.md, pytorch_fully_shard_api.md, pytorch_dcp_recipe.md and pytorch_device_mesh_tutorial.md: tutorials, fully_shard API, checkpoints and DeviceMesh

Good fit for

  • Your model does not fit in a single GPU's memory
  • You want DTensor-based sharding with DeviceMesh

Not for

    Tags

    • fsdp2
    • sharding
    • pytorch
    • distributed

    About

    You say

    “Add FSDP2 to my 13B model training script on four GPUs, with distributed checkpointing.”

    Card for the “Adding FSDP2 to a PyTorch training script” skill

    Skill published by Orchestra Research under the MIT license, delivered in its original version.

    Show more

    How to use it

    1. Install the skill in your agent's skills folder
    2. Ask your agent to add FSDP2 sharding to a training script without initialization errors
    3. Review and adjust the result

    Content preview

    Files

    • LICENSE1.1 KB
    • NOTICE-PALATEKIT.md0.5 KB
    • references/pytorch_dcp_async_recipe.md0.8 KB
    • references/pytorch_dcp_overview.md1 KB
    • references/pytorch_dcp_recipe.md1 KB
    • references/pytorch_ddp_notes.md0.5 KB
    • references/pytorch_device_mesh_tutorial.md1.2 KB
    • references/pytorch_examples_fsdp2.md0.7 KB
    • references/pytorch_fsdp1_api.md0.4 KB
    • references/pytorch_fsdp2_tutorial.md2.5 KB
    • references/pytorch_fully_shard_api.md2.9 KB
    • references/pytorch_tp_tutorial.md1 KB
    • references/ray_train_fsdp2_example.md0.6 KB
    • references/torchtitan_fsdp_notes.md0.7 KB
    • SKILL.md11 KB

    Origin and licence

    Listed unmodified

    Author
    Orchestra Research
    Source
    github.com/orchestra-research/ai-research-skills/tree/773a52944ba4747a18bd4ae9ade53fff041adcbc/08-distributed-training/pytorch-fsdp2
    Licence
    MIT

    Notice

    « pytorch-fsdp2 » (Orchestra Research, MIT) livré dans sa version d'origine : fiche en français et en anglais par PalateKit.

    User reviews

    –

    No ratings yet

    Loading reviews…

    Delete your review?

    Your rating and comment will be removed from the page.