English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

This work studies post-training parameter quantization in large languagemodels (LLMs). We introduce quantization with incoherence processing (QuIP), anew method based on the insight that quantization benefits from$\textit{incoherent}$ weight and Hessian matrices, i.e., from the weights beingeven in magnitude and the directions in which it is important to round themaccurately being unaligned with the coordinate axes. QuIP consists of twosteps: (1) an adaptive rounding procedure minimizing a quadratic proxyobjective; (2) efficient pre- and post-processing that ensures weight andHessian incoherence via multiplication by random orthogonal matrices. Wecomplement QuIP with the first theoretical analysis for an LLM-scalequantization algorithm, and show that our theory also applies to an existingmethod, OPTQ. Empirically, we find that our incoherence preprocessing improvesseveral existing quantization algorithms and yields the first LLM quantizationmethods that produce viable results using only two bits per weight. Our codecan be found at https://github.com/Cornell-RelaxML/QuIP.

QuIP: 2-Bit Quantization of Large Language Models With Guarantees