How Well Does GPT-4V(ision) Adapt to Distribution Shifts? A Preliminary Investigation

TL;DR

This paper evaluates GPT-4V(ision)'s ability to adapt to distribution shifts across diverse datasets and perturbations, revealing its strengths and limitations in robustness and generalization in dynamic environments.

Contribution

It provides a comprehensive benchmark of GPT-4V's zero-shot and perturbed data performance, highlighting its adaptability and identifying areas for improvement.

Findings

GPT-4V shows strong zero-shot performance on certain datasets.

Performance degrades under specific data perturbations.

In-context learning can enhance GPT-4V's adaptation capabilities.

Abstract

In machine learning, generalization against distribution shifts -- where deployment conditions diverge from the training scenarios -- is crucial, particularly in fields like climate modeling, biomedicine, and autonomous driving. The emergence of foundation models, distinguished by their extensive pretraining and task versatility, has led to an increased interest in their adaptability to distribution shifts. GPT-4V(ision) acts as the most advanced publicly accessible multimodal foundation model, with extensive applications across various domains, including anomaly detection, video understanding, image generation, and medical diagnosis. However, its robustness against data distributions remains largely underexplored. Addressing this gap, this study rigorously evaluates GPT-4V's adaptability and generalization capabilities in dynamic environments, benchmarking against prominent models like CLIP, LLaVA, and Gemini. We delve into GPT-4V's zero-shot generalization across 13 diverse datasets spanning natural, medical, and molecular domains. We further investigate its adaptability to controlled data perturbations and examine the efficacy of in-context learning as a tool to enhance its adaptation. Our findings delineate GPT-4V's capability boundaries in distribution shifts, shedding light on its strengths and limitations across various scenarios. Importantly, this investigation contributes to our understanding of how AI foundation models generalize to distribution shifts, offering pivotal insights into their adaptability and robustness. The code is publicly available at https://github.com/jameszhou-gl/gpt-4v-distribution-shift.