open-mmlab / open-mmlab/mmengine

Add compat_cfg to handle the compatibility issues in one place

Open
#381 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

P1 refactoring
Dominant language
Python
Stars
1.5k
Forks
464
PR merge metrics
No merged PRs in 30d

Description

As described in the title.
Should implement compat_cfg in mmengine to handle the compatibility issues in one place.
In the function, it could handle the compatibility issues of

  • compatibility issue of pytorch, like persistent_workers that only works in more recent version of pytorch
  • compatibility issue of the cfg changes in runner

Motivation
Currently there are multiple places that do some implicit conversion of cfg options, we should put them in one place.

Related resources
In MMDet 2.x, there is an implementation of compat_config.py

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

Start by reading the referenced MMDet 2.x compat_config.py implementation and locating the existing implicit cfg conversions in MMEngine's runner. Define done as centralizing the listed PyTorch and runner-config compatibility handling in a compat_cfg function, with the affected compatibility cases covered.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
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.