ROCm / ROCm/FastFlowLM

Qwen 3.5 4B versus Qwen 3 VL 4B IT in Perplexica/Vane

Open
#439 4 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C++
Stars
1.9k
Forks
152
Avg merge
4h 14m
Merged PRs (30d)
11

Description

Qwen 3 VL 4B IT performs very well in web search answers using perplexica/vane and lemonade server on arch linux. While the 3.5 4B gets stuck on "Brainstorming" but never calls the search tool. This might be a difference between the LLMs or their quants used by FLM.

Tool/Frontend: Perplexica/Vane

Server: Lemonade/FLM

Hardware: Ryzen AI 7 350, 32 GB RAM

OS: Arch linux.

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 reproducing the Perplexica/Vane web-search flow through Lemonade/FLM on Arch Linux, comparing Qwen 3.5 4B with Qwen 3 VL 4B IT on the reported hardware. Trace whether the failure to call the search tool comes from the model or its quant, and consider the issue complete when that cause is isolated and the appropriate behavior is documented or corrected.

Written by the indexing model from the issue text.

Assessment

Tech stack
arch-linux
Domain
ai, backend
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
38/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.