thu-ml / thu-ml/TurboDiffusion

Generalization of the merging operation

Open
#121 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
3.7k
Forks
277
Avg merge
2h 57m
Merged PRs (30d)
2

Description

Hi @zhengkw18 @jt-zhang Thanks for your great efforts!

Regarding turbodiffusion/scripts/merge_models.py, could it be employed as a common practice of merging step distillation weights (e.g., DMD, discrete / continuous CD, adversarial distillation, etc.) and sparse attn weights (e.g., VSA, STA, radial attn, etc.), in order to combine their strength?

Or else, it is just a nice property of rCM together with SLA? Thanks in advance.

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

Start by reading turbodiffusion/scripts/merge_models.py and the issue's references to step-distillation and sparse-attention weights. Compare how the current merge handles rCM and SLA with the listed distillation and attention methods; the issue does not define a concrete implementation or acceptance criteria, so the desired general solution needs clarification first.

Written by the indexing model from the issue text.

Assessment

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

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.