microsoft / microsoft/onnxruntime

Overflow detection for quantized/half/mixed precision models

Open
#10,355 2 comments 2 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

core runtime feature request
Dominant language
C++
Stars
21.9k
Forks
4.2k
Avg merge
4d 8h
Merged PRs (30d)
179

Description

Is your feature request related to a problem? Please describe.
It is common that many issues and unexpected model outputs from quantized, fp16 or mixed precision models are caused by overflowing. It would be much faster and easier to pinpoint the outstanding operator, if there is some degree of overflow detection support on computation/operator level.

Describe the solution you'd like
In case there is a concern for performance impact, the overflow detection can be set enabled only at debug build.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

The issue names no files, tests, or entry points. Start by locating computation and operator execution paths for quantized, fp16, and mixed-precision models, then investigate how a debug-only overflow check could identify the outstanding operator. Done should include operator-level overflow detection with an understood performance impact.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.