Abstract
Knowledge distillation is an established technique for improving the capabilities of small, efficient student models by training them with the representations of larger, more capable teacher models. Much of the recent work in the distillation of large language models (LLMs) has focused on distilling abilities learned during post-training, such as instruction following, chain-of-thought reasoning, and tool usage. This has left a large research gap in general knowledge distillation for LLMs, which is essential for developing efficient and private systems suitable for deployment on edge devices. We take a first-principles approach, evaluating previous lessons from prior works and conducting new explorations to develop a distillation methodology suitable for modern LLMs. We present KDFP, a novel methodology for white-box general knowledge distillation in LLMs. We demonstrate that KDFP outperforms existing methods by 1.6% $-$ 4.9% across 9 benchmarks while increasing training efficiency by up to 99.1% through ephemeral parameter reduction.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Swift, R., & Psounis, K. (2026). KDFP: A first-principles approach to knowledge distillation in large language models. https://omanscience.com/en/articles/kdfp-a-first-principles-approach-to-knowledge-distillation-in-large-language-models
MLA 9
Swift, Ryan, and Konstantinos Psounis. "KDFP: A first-principles approach to knowledge distillation in large language models." https://omanscience.com/en/articles/kdfp-a-first-principles-approach-to-knowledge-distillation-in-large-language-models.
Chicago (author–date)
Swift, Ryan, and Konstantinos Psounis. 2026. "KDFP: A first-principles approach to knowledge distillation in large language models." https://omanscience.com/en/articles/kdfp-a-first-principles-approach-to-knowledge-distillation-in-large-language-models.
Harvard
Swift, R. and Psounis, K. (2026) 'KDFP: A first-principles approach to knowledge distillation in large language models', Available at: https://omanscience.com/en/articles/kdfp-a-first-principles-approach-to-knowledge-distillation-in-large-language-models.
Vancouver
Swift R, Psounis K. KDFP: A first-principles approach to knowledge distillation in large language models. https://omanscience.com/en/articles/kdfp-a-first-principles-approach-to-knowledge-distillation-in-large-language-models
IEEE
R. Swift, and K. Psounis, "KDFP: A first-principles approach to knowledge distillation in large language models," https://omanscience.com/en/articles/kdfp-a-first-principles-approach-to-knowledge-distillation-in-large-language-models.