home/categories/framework-internals/orchestra-research-ai-research-skills-08-distributed-training-pytorch-fsdp2-skill-md
framework-internalsdevelopment
pytorch-fsdp2
Adds PyTorch FSDP2 (fully_shard) to training scripts with correct init, sharding, mixed precision/offload config, and distributed checkpointing. Use when models exceed single-GPU memory or when you need DTensor-based sharding with DeviceMesh.
maintainer
Orchestra-Research
Updated 1/30/2026
Stars
6563
Forks
515
quick start
Installation and usage
Adds PyTorch FSDP2 (fully_shard) to training scripts with correct init, sharding, mixed precision/offload config, and distributed checkpointing. Use when models exceed single-GPU memory or when you need DTensor-based sharding with DeviceMesh.
Installation
$ install --globalskills.sh
Usage
Once installed, you can use this skill by running the following command in your terminal:
skills use pytorch-fsdp2