OpenEuroLLM / OpenEuroLLM/Taskboard

Post-flagship: explore the architecture that is optimal for deployment

Open
#357 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

T4.2 - model exploration
Dominant language
No language data
Stars
3
Forks
0
PR merge metrics
No merged PRs in 30d

Description

Goal
Raised in a model exploration meeting: what can we try to attempt to come up with the most optimal architecture. The issue is considered to be exploratory task for post-flagship.

Description
This includes

  • efficient long context handling
  • sliding window attention
  • hybrid (eg Qwen 3.5 design, gated delta net layers)
  • aim for 1M context length in final version?
  • Quantization aware (easy to quantize to lower precision, eg NVFP4); easy GGUF conversion; llama.cpp compat, etc
    TODO: add possible modifications list to the task
  • re-do weak scaling now

Deliverable scope
(What is the expected outcome for this issue to be complete)

Dependencies
(Blocking issues, PRs, or external factors)

Contributor guide

No contributing guide indexed for this repository

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 defining the missing deliverable and narrowing the listed options: long-context handling, sliding-window or hybrid attention, quantization, conversion compatibility, and weak scaling. Done should include a documented architecture choice, its evaluation criteria, and the requested weak-scaling results.

Written by the indexing model from the issue text.

Assessment

Domain
machine-learning, performance
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Quiet
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.