LiveXiv -- A Multi-Modal Live Benchmark Based on Arxiv Papers Content

TL;DR

LiveXiv is a scalable, evolving benchmark based on arXiv papers that automatically generates visual question-answer pairs from scientific manuscripts to evaluate multi-modal models without data contamination.

Contribution

The paper introduces LiveXiv, a novel live benchmark for multi-modal models based on arXiv papers, with automatic VQA generation and an efficient evaluation method to assess model performance over time.

Findings

Benchmark is challenging for current models.

Automatic annotations closely match manual ones (<2.5% variance).

Evaluation cost is significantly reduced by subset evaluation.

Abstract

The large-scale training of multi-modal models on data scraped from the web has shown outstanding utility in infusing these models with the required world knowledge to perform effectively on multiple downstream tasks. However, one downside of scraping data from the web can be the potential sacrifice of the benchmarks on which the abilities of these models are often evaluated. To safeguard against test data contamination and to truly test the abilities of these foundation models we propose LiveXiv: A scalable evolving live benchmark based on scientific ArXiv papers. LiveXiv accesses domain-specific manuscripts at any given timestamp and proposes to automatically generate visual question-answer pairs (VQA). This is done without any human-in-the-loop, using the multi-modal content in the manuscripts, like graphs, charts, and tables. Moreover, we introduce an efficient evaluation approach that estimates the performance of all models on the evolving benchmark using evaluations of only a subset of models. This significantly reduces the overall evaluation cost. We benchmark multiple open and proprietary Large Multi-modal Models (LMMs) on the first version of our benchmark, showing its challenging nature and exposing the models true abilities, avoiding contamination. Lastly, in our commitment to high quality, we have collected and evaluated a manually verified subset. By comparing its overall results to our automatic annotations, we have found that the performance variance is indeed minimal (<2.5%). Our dataset is available online on HuggingFace, and our code will be available here.