Predicting LLM Compression Degradation from Spectral Statistics

April 20, 2026 ยท Grace Period ยท + Add venue

โณ Grace Period
This paper is less than 90 days old. We give authors time to release their code before passing judgment.
Authors Mingxue, Xu arXiv ID 2604.18085 Category cs.LG: Machine Learning Citations 0
Abstract
Matrix-level low-rank compression is a promising way to reduce the cost of large language models, but running compression and evaluating the resulting models on language tasks can be prohibitively expensive. Can compression-induced degradation be predicted before committing to this compute? We systematically analyze the Qwen3 and Gemma3 model families across four representative low-rank compression methods: vanilla SVD, two ASVD variants, and SVD-LLM. We find that stable rank and information density, measured in bits per parameter, dominate performance degradation. The interaction term $ฮณ\cdot \barฯ_s$, defined as compression ratio times stable rank, is a robust predictor of accuracy degradation, achieving leave-one-out cross-validation Pearson correlations of $0.890$ for attention layers and $0.839$ for MLP layers. We provide theoretical intuition for why this predictor succeeds by connecting it to standard SVD truncation bounds and error composition mechanisms in transformer layers. These findings enable a predict-then-compress workflow: compute $ฮณ\cdot \barฯ_s$ from weights, estimate degradation, and invest compute only in desirable configurations.
Community shame:
Not yet rated
Community Contributions

Found the code? Know the venue? Think something is wrong? Let us know!

๐Ÿ“œ Similar Papers

In the same crypt โ€” Machine Learning