collaborators
Showing cs.LGShow all

10 papers · 1 filter

cs.LG2026

High-Rate Quantized Matrix Multiplication II

Or Ordentlich, Yury Polyanskiy

This is the second part of the work investigating quantized matrix multiplication (MatMul). In part I we considered the case of calibration-free quantization, whereas here we discu…

cs.LG2026

WaterSIC: Information-Theoretically (Near) Optimal Linear Layer Quantization

Egor Lifar, Semyon Savkin, Or Ordentlich +1

This paper considers the problem of converting a given dense linear layer to low precision. The tradeoff between compressed length and output discrepancy is analyzed information th…

cs.LG2026

Measure-to-measure Regression with Transformers

Matthew Vandergrift, Martha White, Yury Polyanskiy +2

Many learning problems require predicting how populations evolve under an unknown transformation. A natural representation for such populations is a probability measure, with point…

cs.LG2026

Representation Alignment Rests on Linear Structure

Kiril Bangachev, Guy Bresler, Yury Polyanskiy

We investigate the Platonic Representation Hypothesis (PRH) through a tripartite statistical framework of representations: signal, bias, and noise. {1) Signal:} We propose that Pla…

cs.LG2026

Is Dimensionality a Barrier for Retrieval Models?

Kiril Bangachev, Guy Bresler, Jonathan Kogan +1

Why does the low dimensionality of representations, typically , not prevent modern embedding-based retrieval models from scaling to billions, or even trillions, of d…

cs.LG2026

Continuous First, Discrete Later: VQ-VAEs Without Dimensional Collapse

Xinyu Zhao, Nikita Karagodin, Hamed Hassani +3

While many approaches to improve VQ-VAE performance focus on codebook size and utilization, the effect of dimensional collapse, where trained VQ-VAE representations live in an extr…