AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Swift-SVD: Theoretical Optimality Meets Practical Efficiency in Low-Rank LLM Compression

Ruoling Qi, Yirui Liu, Xuaner Wu, Xiangyu Wang, Ming Li, Chen Chen, Jian Chen, Yin Chen, Qizhen Weng

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Distributed LLM inference systems · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add dynamic rank allocation (loss plus layer importance with preserved ratio) (ablation)Swift-SVD · C4 PPL, LLaMA-7B ratio 0.6perplexity: 23.32 -> 21.92Improved
add dynamic rank allocation (ablation)Swift-SVD · LLaMA-7B zero-shot average accuracy ratio 0.8accuracy: 0.50 -> 0.51Improved
allocate ranks by Frobenius loss only, no preserved ratio (ablation)Swift-SVD · C4 PPL, LLaMA-7B ratio 0.8perplexity: 11.42 -> 16.04Hurt
allocate ranks by layer importance only, no preserved ratio (ablation)Swift-SVD · C4 PPL, LLaMA-7B ratio 0.8perplexity: 11.42 -> 14.88Hurt
add fixed preserved ratio 0.5 to loss-only allocation (ablation)Swift-SVD(C) · C4 PPL, LLaMA-7B ratio 0.8perplexity: 16.04 -> 11.78Improved
add fixed preserved ratio 0.5 to importance-only allocation (ablation)Swift-SVD(I) · C4 PPL, LLaMA-7B ratio 0.8perplexity: 14.88 -> 11.73Improved
combine loss and importance in allocation instead of importance alone with preserved ratio (ablation)Swift-SVD-dagger(I) · C4 PPL, OPT-6.7B ratio 0.8perplexity: 14.74 -> 13.68Improved
use Swift-SVD* dynamic allocation instead of Dobi-SVD training-based dynamic allocationDobi-SVD · C4 PPL, LLaMA-7B ratio 0.4perplexity: 190.62 -> 137.77Improved
calibrate on C4 only and evaluate on WikiText-2Swift-SVD · LLaMA-7B ratio 0.6 nsamples 256perplexity: 13.42 -> 37.00Hurt
increase calibration samples from 256 to 2048 (ablation)Swift-SVD · C4 PPL, LLaMA-7B ratio 0.6perplexity: 23.17 -> 22.37Improved
use Swift-SVD closed-form eigendecomposition instead of SVD-LLMSVD-LLM · reconstruction loss [4096x4096]^2loss: 3365.6499 -> 3308.6428Improved
use Swift-SVD instead of Dobi-SVD(w/o) for compression, 512 samplesDobi-SVD(w/o) · compression timespeedup: 1.0x -> 76.9xImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy