NVIDIA / NVIDIA/TensorRT-LLM

[AutoDeploy] Enhance checkpoint loading pipeline

Open
#5,555 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

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

Description

More robust usage of load_checkpoint_in_model

  1. Investigate load_checkpoint_in_model's load_hook applying issue:
  2. understand the difference between set_model_state_dict and the torch-native module.load_state_dict, the first one works for the DS FP4 case while the second doesn't
  3. Add unit tests for our use case(e.g. load_hook) and guard around version updates

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 in tensorrt_llm/_torch/auto_deploy/models/hf.py at load_checkpoint_in_model and trace its load_hook behavior. Compare set_model_state_dict with torch-native module.load_state_dict, especially for the DS FP4 case. Done means the load_hook use case is covered by unit tests and version updates have appropriate guards.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
Domain
machine-learning
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
48/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.