Quip: 2-bit quantization of large language models with guarantees
This work studies post-training parameter quantization in large language models (LLMs).
We introduce quantization with incoherence processing (QuIP), a new method based on the …
We introduce quantization with incoherence processing (QuIP), a new method based on the …
A practical method for constructing equivariant multilayer perceptrons for arbitrary matrix groups
Symmetries and equivariance are fundamental to the generalization of neural networks on
domains such as images, graphs, and point clouds. Existing work has primarily focused on a …
domains such as images, graphs, and point clouds. Existing work has primarily focused on a …
Streaming pca and subspace tracking: The missing data case
For many modern applications in science and engineering, data are collected in a streaming
fashion carrying time-varying information, and practitioners need to process them with a …
fashion carrying time-varying information, and practitioners need to process them with a …
Scatterbrain: Unifying sparse and low-rank attention
Recent advances in efficient Transformers have exploited either the sparsity or low-rank
properties of attention matrices to reduce the computational and memory bottlenecks of …
properties of attention matrices to reduce the computational and memory bottlenecks of …
Atomo: Communication-efficient learning via atomic sparsification
Distributed model training suffers from communication overheads due to frequent gradient
updates transmitted between compute nodes. To mitigate these overheads, several studies …
updates transmitted between compute nodes. To mitigate these overheads, several studies …
Matrix completion has no spurious local minimum
Matrix completion is a basic machine learning problem that has wide applications,
especially in collaborative filtering and recommender systems. Simple non-convex …
especially in collaborative filtering and recommender systems. Simple non-convex …
A geometric analysis of phase retrieval
Can we recover a complex signal from its Fourier magnitudes? More generally, given a set
of m measurements, y_k=\left| a _k^* x\right| yk= ak∗ x for k= 1, ..., mk= 1,…, m, is it possible …
of m measurements, y_k=\left| a _k^* x\right| yk= ak∗ x for k= 1, ..., mk= 1,…, m, is it possible …
Global optimality of local search for low rank matrix recovery
S Bhojanapalli, B Neyshabur… - Advances in Neural …, 2016 - proceedings.neurips.cc
We show that there are no spurious local minima in the non-convex factorized
parametrization of low-rank matrix recovery from incoherent linear measurements. With …
parametrization of low-rank matrix recovery from incoherent linear measurements. With …
Guaranteed matrix completion via non-convex factorization
R Sun, ZQ Luo - IEEE Transactions on Information Theory, 2016 - ieeexplore.ieee.org
Matrix factorization is a popular approach for large-scale matrix completion. The optimization
formulation based on matrix factorization, even with huge size, can be solved very efficiently …
formulation based on matrix factorization, even with huge size, can be solved very efficiently …
Low-rank solutions of linear matrix equations via procrustes flow
In this paper we study the problem of recovering a low-rank matrix from linear
measurements. Our algorithm, which we call Procrustes Flow, starts from an initial estimate …
measurements. Our algorithm, which we call Procrustes Flow, starts from an initial estimate …