modelscope / modelscope/DiffSynth-Studio
I find a BUG on loss_weight.
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 13.1k
- Forks
- 1.3k
- Avg merge
- 13h 12m
- Merged PRs (30d)
- 45
Description
I find loss weight become biger when time step is smaller. But flow matching method usually train more on big time step. is it a bug or you design it？
shift as 3. Red point's time index is 900, means timestep value is small because timestep is inverse sorted.
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
No source file, test, or entry point is identified. Start by reproducing the loss_weight versus timestep behavior described in issue #1148, checking the attached plot and the inverse-sorted timestep meaning. Done means determining whether the weighting is intentional for flow matching and, if not, defining a precise correction and regression check.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100