NVIDIA-NeMo / NVIDIA-NeMo/Curator

[FEA] Create intermediate representation (IR) from code source

Open
#209 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

community-request enhancement
Dominant language
Python
Stars
1.8k
Forks
328
Avg merge
4d 5h
Merged PRs (30d)
30

Description

When training an LLM on code-related tasks, it has been empirically proven that paring source code with its intermediate representation (IR) improves the LLM capabilities (https://arxiv.org/abs/2403.03894).

Adding this capability to NeMo Curator would be, IMHO, useful,

Hope it helps!
Miguel

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 reviewing NeMo Curator's existing code-processing and data-preparation capabilities, then use the linked paper to clarify the intended IR formats and supported languages. Done would require a defined way to produce IR alongside source code, but the issue does not specify acceptance criteria.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
data, 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.