Pytorch Fsdp2 Per Parameter Sharding

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.

robertoatila Updated 9 repo stars
J.A.R.V.I.S. — Autonomous Cognitive Runtime

File contents

robertoatila/jarvis-skill-registry commit 08f311244f

Frequently asked questions

npx get-skill add robertoatila/pytorch-fsdp2-per-parameter-sharding-2