Loss Instability and Grid Artifacts in MoGe Reproduction
Nobody has claimed this yet.
- Dominant language
- JavaScript
- Stars
- 3k
- Forks
- 234
- PR merge metrics
- No merged PRs in 30d
Description
Hi, I'm reproducing MoGe with your training code (unmodified). Two issues:
MoGe Ori
Our Reprodution
With normal loss:
Loss spikes suddenly during training (around 20k step, and visualizations (depth/normals) collapse. Any advice on stabilizing it, possibly related to GT normal computation or outlier handling?
Without normal loss:
Training stabilizes, but normals still show grid-like, unsmooth patterns (see attached: figures). Why might this happen, and why do metrics improve despite worse visuals?
Details:
Environment: bs=8, 8 H20 GPUs.
Dataset: Processed by us, datasets combination similar to your setup. Here is an visualization example of one batch.
What’s the role of normal loss here, and how should I adjust dataset processing to match your results? Thanks!
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
The report names no file, test, or training entry point; begin by locating the training configuration and normal-loss or ground-truth-normal processing used for the batch-size-8 run. Reproduce the loss spike around 20k steps and compare the normal-loss and no-normal-loss runs against the reported depth and normal artifacts; progress requires confirming a cause and matching training behavior.
Written by the indexing model from the issue text.
Assessment
- Domain
- computer-vision, machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100