NVIDIA / NVIDIA/Megatron-LM

[Determinism] torch.utils.deterministic.fill_uninitialized_memory default to false for determinism perf

Open
#6,253 0 comments 0 reactions 1 assignee Claimed by @ZhiyuLi-Nvidia View on GitHub
bug Determinism
Dominant language
Python
Stars
17.9k
Forks
4.5k
Avg merge
4d 6h
Merged PRs (30d)
271

Description

**Describe the bug**

Determinism tax reduced from ~15% to ~2%, based on optimizations proposed by Nathan Tuck US

torch.utils.deterministic.fill_uninitialized_memory defaults to true with torch determinism setup but it is too strict not necessary to ensure determinism.

**Steps/Code to reproduce bug**

Please list *minimal* steps or code snippet for us to be able to reproduce the bug.

A helpful guide on on how to craft a minimal bug report http://matthewrocklin.com/blog/work/2018/02/28/minimal-bug-reports.

**Expected behavior**

A clear and concise description of what you expected to happen.

**Additional context**

Add any other context about the problem here.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.