Lightning-AI / Lightning-AI/pytorch-lightning

Impove how argument passing via CLI and config file is handled in regards to argument linking

Open
#20,341 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

feature lightningcli
Dominant language
Python
Stars
31.4k
Forks
3.8k
Avg merge
6d 7h
Merged PRs (30d)
6

Description

### Description & Motivation

I was working on several larger projects and I noticed that the behavior or argument linking is somewhat arbitrary. I propose to make this more consistent to avoid user confusion. Furthermore I propose to write a documentation page for this, since the one from jsonargmarse is not sufficient due to some nuances in combination with lighnting CLI.

### Pitch

I would like to make the behavior of passing arguments via cli and config file more consistent in regards to argument linking. There are several aspects to this that I still need to figure out. I will start by writing test cases and update on the go.

### Alternatives

_No response_

### Additional context

_No response_

cc @lantiga @borda @mauvilsa

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 defining test cases for argument linking between CLI inputs and config-file inputs, as described in the issue, and use them to identify the inconsistent cases. The work is done when the behavior is consistent and a documentation page explains the Lightning CLI-specific nuances beyond jsonargparse.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
cli
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.