THUDM / THUDM/slime

Regression: Garbled/Mojibake rollout output on main @46e2cd (Qwen3-4B demo); v0.1 works

Open
#572 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
8.5k
Forks
1.3k
Avg merge
5h 36m
Merged PRs (30d)
22

Description

On the main branch at commit 46e2cd4f1b3e56c3178977dba45cfde8fb53546c
running the demo from https://github.com/THUDM/slime/blob/main/docs/zh/examples/qwen3-4B.md
produced garbled (mojibake) rollout responses.

After switching to the stable 0.1 release and using the same Megatron weight-conversion command and training script, generation worked normally.

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 with docs/zh/examples/qwen3-4B.md and reproduce the rollout at commit 46e2cd4f1b3e56c3178977dba45cfde8fb53546. Compare the same Megatron weight-conversion command and training script against the stable 0.1 release, then trace the generation path to isolate the regression. Done means the demo produces readable rollout responses on main and the regression is covered by a reproducible check.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.