# FSDP2 (fully_shard): Per-Parameter Sharding + DTensor + 2024+ PyTorch Innovation

> Source: https://sukruyusufkaya.com/en/learn/fine-tuning-cookbook/ftc-fsdp2-per-parameter-sharding-dtensor
> Updated: 2026-05-14T14:42:52.316Z
> Category: Fine-Tuning Cookbook (Model-by-Model)
> Module: Part IV — Mid-Large Models (13B-70B+) + Distributed Internals
**TLDR:** FSDP2 (PyTorch 2.4+) — evolution of FSDP. Per-parameter sharding (FlatParameter pattern dropped), DTensor backbone, FQN-based resumable checkpointing, easier mixed precision. Llama 3.3 70B + FSDP2 + DCP recipe.