LOVM: Language-Only Vision Model Selection

TL;DR

This paper introduces LOVM, a new benchmark and task for selecting the best pre-trained vision-language models based solely on text descriptions, eliminating the need for dataset-specific evaluations.

Contribution

We propose a novel task and benchmark for zero-shot model selection using only text descriptions, enabling efficient VLM evaluation without access to downstream datasets.

Findings

Established the LOVM benchmark with evaluations of 35 VLMs across 23 datasets.

Demonstrated the effectiveness of text-based model ranking methods.

Provided insights into zero-shot performance prediction for VLMs.

Abstract

Pre-trained multi-modal vision-language models (VLMs) are becoming increasingly popular due to their exceptional performance on downstream vision applications, particularly in the few- and zero-shot settings. However, selecting the best-performing VLM for some downstream applications is non-trivial, as it is dataset and task-dependent. Meanwhile, the exhaustive evaluation of all available VLMs on a novel application is not only time and computationally demanding but also necessitates the collection of a labeled dataset for evaluation. As the number of open-source VLM variants increases, there is a need for an efficient model selection strategy that does not require access to a curated evaluation dataset. This paper proposes a novel task and benchmark for efficiently evaluating VLMs' zero-shot performance on downstream applications without access to the downstream task dataset. Specifically, we introduce a new task LOVM: Language-Only Vision Model Selection, where methods are expected to perform both model selection and performance prediction based solely on a text description of the desired downstream application. We then introduced an extensive LOVM benchmark consisting of ground-truth evaluations of 35 pre-trained VLMs and 23 datasets, where methods are expected to rank the pre-trained VLMs and predict their zero-shot performance.