modelscope / modelscope/DiffSynth-Studio
训练qwen-image-edit的lora或full微调后，推理时候完全不会对图像做修改，为什么？
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 13.1k
- Forks
- 1.3k
- Avg merge
- 13h 12m
- Merged PRs (30d)
- 45
Description
我是在做一个分割任务，数据集的设置为：原图，绘制了mask后的图像。提示词类似这样“将xxx绘制为红色”。
不训练直接做zero-shot的时候，是可以将一部分内容正确绘制为红色的。但是经过lora或者full微调后，会完全不修改图像。我的数据集数量几千张，应该也是够的吧。full微调的时候，最高跑过10epoch，用时三四天，还是无法修改图像，有人知道为什么吗？
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 reviewing the dataset format, prompts, and LoRA or full-finetuning settings described in the issue, then compare the fine-tuned inference path with the working zero-shot path. Confirm that the trained model actually modifies the image for the segmentation prompt; the issue provides no file, test, or entry point to inspect.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- ai, machine-learning
- Issue type
- Bug
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100