Model Architecture · Established · Advanced
Spherical Linear Interpolation Merging
Also known as: SLERP Merging, Model Fusion
A mathematical technique used to combine two fine-tuned checkpoint weights by interpolating along a spherical vector trajectory.
What Spherical Linear Interpolation Merging is
SLERP preserves directional information and non-linear feature representations better than simple weighted linear averaging.
How it works
Normalizes model weight vectors, computes the angle between them, and interpolates weights along the geometric arc.
Why it matters
Popularized in the open-source community for creating high-performing merged foundation models without retraining.
Common uses
- →Foundation model weight merging
- →Multi-capability model creation
- →Zero-cost fine-tuning fusion
More in this collection
Browse all AI Concepts