Revision History
Fully Sharded Data Parallel (FSDP) · 7 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Correction: Corrected FSDP's PyTorch 1.11 status from beta to prototype, fixed a reversed description of selective activation checkpointing, corrected what the FSDP VLDB paper and the Llama 2 paper actually report (no 1-trillion-parameter experiment; FSDP used for Llama 2's PPO stage), clarified that flattened sharding is FSDP1-only, and removed EleutherAI and Stanford CRFM as FSDP users (plus 16 minor fixes)
Version 7 · Nov 2, 2026, 03:21 PM
- Link [TorchTitan] to the new torchtitan article
Version 6 · Oct 16, 2026, 10:57 PM
- Improve internal links (2026-10-13)
Version 5 · Oct 13, 2026, 10:18 AM
- Added 6 contextual internal links
Version 4 · Jul 23, 2026, 03:28 PM