modelscope / modelscope/DiffSynth-Studio
对 FLUX2-clein 非base模型做步数蒸馏，发现结果变糊
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 13.1k
- Forks
- 1.3k
- Avg merge
- 13h 12m
- Merged PRs (30d)
- 45
Description
我对FLUX.2-clein-base-9B多图编辑任务做了lora微调，想进一步对非base做加速蒸馏， 但是发现训了一轮后编辑图变模糊，有大佬知道怎么回事吗？ 还是说非base模型不适合做进一步蒸馏训练？
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reproducing the reported one-round distillation on the FLUX.2-clein-base-9B multi-image editing task and compare the edited images before and after training. Investigate whether distilling the non-base model is supported; done means identifying the cause of the blur or establishing the model's compatibility limits.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100