الملخص
Optimizing merging coefficients for large language models can require many costly benchmark evaluations. We propose \textbf{Simplex-Lattice Interpolation Merging (SLIM)}, which constructs a quadratic surrogate of aggregate performance on the coefficient simplex using a classical mixture design. Evaluations of individual experts and equal-weight pairs determine the surrogate with the minimum number of measurements needed to identify a general quadratic on this domain. SLIM then optimizes the surrogate without further target-metric evaluations. Experiments on two model architectures demonstrate accurate prediction of unseen multi-expert mixtures and competitive merge performance under limited evaluation budgets. Matched-budget comparisons show that structured evaluation points improve prediction fidelity over random designs, including those using regularized fitting.
الكلمات المفتاحية
الموضوع
بيانات النشر
- المجلة
- غير متاح
- وصول مفتوح
- وصول مفتوح أخضر
اقتبس هذه المقالة
APA 7
Jeong, S., Suzuki, M., & Matsuo, Y. (2026). SLIM: Simplex-Lattice Interpolation Merging. https://omanscience.com/ar/articles/slim-simplex-lattice-interpolation-merging
MLA 9
Jeong, Seongcheol, et al. "SLIM: Simplex-Lattice Interpolation Merging." https://omanscience.com/ar/articles/slim-simplex-lattice-interpolation-merging.
شيكاغو (المؤلف–التاريخ)
Jeong, Seongcheol, Masahiro Suzuki, and Yutaka Matsuo. 2026. "SLIM: Simplex-Lattice Interpolation Merging." https://omanscience.com/ar/articles/slim-simplex-lattice-interpolation-merging.
هارفارد
Jeong, S., Suzuki, M. and Matsuo, Y. (2026) 'SLIM: Simplex-Lattice Interpolation Merging', Available at: https://omanscience.com/ar/articles/slim-simplex-lattice-interpolation-merging.
فانكوفر
Jeong S, Suzuki M, Matsuo Y. SLIM: Simplex-Lattice Interpolation Merging. https://omanscience.com/ar/articles/slim-simplex-lattice-interpolation-merging
IEEE
S. Jeong, M. Suzuki, and Y. Matsuo, "SLIM: Simplex-Lattice Interpolation Merging," https://omanscience.com/ar/articles/slim-simplex-lattice-interpolation-merging.