Search results

Disable DDP averaging to avoid repeated gradient averaging

bugSomething isn't workingSomething isn't working

CLA SignedThis label is managed by the Meta Open Source bot.This label is managed by the Meta Open Source bot.

fb-exported

meta-exported

Status: Open (in progress).

pytorch/torchtitannumber 2323

#2323 In pytorch/torchtitan;

· Shagun-G opened on Feb 4, 2026

code refactor : making key steps modular train_step()

CLA SignedThis label is managed by the Meta Open Source bot.This label is managed by the Meta Open Source bot.

fb-exported

Status: Open (in progress).

pytorch/torchtitannumber 1650

#1650 In pytorch/torchtitan;

· Shagun-G opened on Aug 28, 2025

Enable save plan caching

CLA SignedThis label is managed by the Meta Open Source bot.This label is managed by the Meta Open Source bot.

fb-exported

Status: Open (in progress).

pytorch/torchtitannumber 1140

#1140 In pytorch/torchtitan;

· meetv18 opened on Apr 23, 2025