NVIDIA / NVIDIA/TensorRT-LLM

[chore] AutoDeploy: Remove Patches

Open
#13,776 0 comments 0 reactions 1 assignee View on GitHub

@govind-ramnarayan is already working on this.

Since May 5, 2026.

AutoDeploy
Dominant language
Python
Stars
14.7k
Forks
2.8k
Avg merge
2d 23h
Merged PRs (30d)
489

Description

We want to remove all custom modeling patches under tensorrt_llm/_torch/auto_deploy/models/patches/

This will help unblock the sharding IR.

Many of them are removable from custom modeling code, as done in the paperclip sprint: https://github.com/NVIDIA/TensorRT-LLM/compare/main...nv-auto-deploy:TensorRT-LLM:feat/paperclip_maximizer

Most of those were removed by https://github.com/NVIDIA/TensorRT-LLM/pull/13247, but some remain:

  • Patches for models that are unused in the model registry (e.g. [granitemoe.py])
  • Patches for models where the custom modeling file required substantial infrastructure changes (e.g. gptoss-mxfp4 )
  • Patches not removed in the sprint (e.g. pixtral )

Creating this issue to track the status of this.

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.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.