-
Notifications
You must be signed in to change notification settings - Fork 78
Open
Labels
Points: 30competitiveFor competitive issues, only top PRs need to be accepted.For competitive issues, only top PRs need to be accepted.everyoneFor competitive issues, best PRs are only merged, no need to assign yourselfFor competitive issues, best PRs are only merged, no need to assign yourself
Description
Issue: Accuracy Maxing
Task
Maximize the accuracy of the Celestial Vision model.
Work Includes
- Improve data preprocessing and augmentation
- Experiment with stronger model architectures or ensembles
- Tune hyperparameters and loss functions
- Analyze errors to reduce false positives and false negatives
Constraints
- Any techniques, libraries, or models may be used
- Online / hosted APIs are strictly not allowed
Goal
Achieve the highest possible accuracy with a reproducible and stable training pipeline.
Where to Work
- All work should be done inside the
Participants/folder - You may use a new notebook/script or extend an existing one
Metadata
Metadata
Assignees
Labels
Points: 30competitiveFor competitive issues, only top PRs need to be accepted.For competitive issues, only top PRs need to be accepted.everyoneFor competitive issues, best PRs are only merged, no need to assign yourselfFor competitive issues, best PRs are only merged, no need to assign yourself