LOOM-Scope: a comprehensive and efficient LOng-cOntext Model evaluation framework

TL;DR

LOOM-Scope is a unified framework that standardizes and accelerates long-context model evaluation, enabling reliable and comprehensive assessment of large language models' performance in processing extended texts.

Contribution

It introduces a standardized evaluation framework that supports efficient inference and a lightweight benchmark suite for comprehensive long-context model assessment.

Findings

Standardizes evaluation settings across benchmarks

Supports deployment of inference acceleration methods

Provides a holistic, lightweight benchmark suite

Abstract

Long-context processing has become a fundamental capability for large language models~(LLMs). To assess model's long-context performance, numerous long-context evaluation benchmarks have been proposed. However, variations in evaluation settings across these benchmarks lead to inconsistent results, making it difficult to draw reliable comparisons. Besides, the high computational cost of long-context evaluation poses a significant barrier for the community to conduct comprehensive assessments of long-context models. In this paper, we propose LOOM-Scope, a comprehensive and efficient framework for long-context evaluation. LOOM-Scope standardizes evaluation settings across diverse benchmarks, supports deployment of efficient long-context inference acceleration methods, and introduces a holistic yet lightweight benchmark suite to evaluate models comprehensively. Homepage: https://loomscope.github.io