UniRank: Unified Rank Allocation for Low-Rank LLM Compression
TL;DR
It is shown, both geometrically and empirically, that high input--output cosine similarity implies low effective rank, and rank-preserving fine-tuning (RPFT) is proposed, which adapts only a small subset of retained singular components so that the allocated rank stays bounded without re-decomposition.
Abstract
Low-rank decomposition is a promising compression paradigm for large language models (LLMs), yet its effectiveness hinges on rank budget allocation across weight matrices: uniform or hand-crafted rules ignore module-wise importance, while learning-based allocation incurs substantial training overhead. We formulate rank allocation as a global sorting-and-truncation pipeline that scores every singular component by combining local singular energy with global functional importance, estimated via layer-wise input--output cosine similarity on a tiny calibration set. We show, both geometrically and empirically, that high input--output cosine similarity implies low effective rank. We further propose rank-preserving fine-tuning (RPFT), which adapts only a small subset of retained singular components so that the allocated rank stays bounded without re-decomposition. Experimental results show that UniRank cuts zero-shot perplexity by up to 50\%, improves average reasoning accuracy by 3.0\% over LoRAP at 25\% sparsity, and boosts four SVD-based decomposition methods as a plug-and-play module.
