LGJun 4, 2024

Loki: Low-rank Keys for Efficient Sparse Attention

arXiv:2406.02542v275 citations
Originality Incremental advance
AI Analysis

This addresses efficiency issues in LLM inference for users handling long sequences, though it is an incremental improvement over existing sparse attention techniques.

The paper tackles the high compute and memory costs of self-attention in large language models (LLMs) by proposing Loki, a sparse attention method that uses low-rank key vectors to select tokens, speeding up attention computation while maintaining model efficacy better than other approximation methods.

Inference on large language models (LLMs) can be expensive in terms of the compute and memory costs involved, especially when long sequence lengths are used. In particular, the self-attention mechanism used in LLM inference contributes significantly to these costs, which has sparked an interest in approximating the self-attention computation to reduce such costs. In this work, we propose to approximate self-attention by focusing on the dimensionality of key vectors computed in the attention block. Our analysis reveals that key vectors lie in a significantly lower-dimensional space, consistently across several datasets and models. Exploiting this observation, we propose Loki, a novel sparse attention method that ranks and selects tokens in the KV-cache based on attention scores computed in low-dimensional space. Our evaluations show that Loki is able to speed up the attention computation due to reduced data movement (load/store) and compute costs while maintaining the efficacy of the models better than other popular approximation methods.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes