Pith. sign in

REVIEW 8 cited by

Many-Shot In-Context Learning in Multimodal Foundation Models

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2405.09798 v2 pith:OCMIMK4N submitted 2024-05-16 cs.LG cs.AIcs.CLcs.CV

Yixing Jiang , Jeremy Irvin , Ji Hun Wang , Muhammad Ahmed Chaudhry , Jonathan H. Chen , Andrew Y. Ng This is my paper
classification cs.LGcs.AIcs.CLcs.CV
keywords modelsdatasetsfoundationmany-shotmultimodalexamplesgeminiperformance
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Large language models are effective at few-shot in-context learning (ICL). Recent advancements in multimodal foundation models have enabled unprecedentedly long context windows, presenting an opportunity to explore their capability to perform ICL with many more demonstrating examples. In this work, we evaluate the performance of multimodal foundation models scaling from few-shot to many-shot ICL. We benchmark GPT-4o and Gemini 1.5 Pro across 14 datasets spanning multiple domains (natural imagery, medical imagery, remote sensing, and molecular imagery) and tasks (image classification, visual QA, and object localization). We observe that many-shot ICL, including up to almost 2,000 demonstrating examples, leads to substantial improvements compared to few-shot (<100 examples) ICL across all of the datasets. Further, Gemini 1.5 Pro performance continues to improve log-linearly up to the maximum number of tested examples on many datasets. We also find open-weights multimodal foundation models like Llama 3.2-Vision do not benefit from the demonstrating examples, highlighting an important gap between open and closed multimodal foundation models. Given the high inference costs required for many-shot ICL, we also explore the impact of batching multiple queries in a single API call. We show that batching up to 50 queries can lead to performance improvements under zero-shot and many-shot ICL, with substantial gains in the zero-shot setting on multiple datasets, while drastically reducing per-query cost and latency. Finally, while GPT-4o and Gemini 1.5 Pro achieve similar zero-shot performance across the datasets, Gemini 1.5 Pro learns more quickly than GPT-4o on most datasets. Our results suggest that many-shot ICL could enable users to efficiently adapt multimodal foundation models to new applications and domains. Our codebase is publicly available at https://github.com/stanfordmlgroup/ManyICL .

Discussion (0). Sign in to comment.

Forward citations

Cited by 8 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. LLaVA-NeXT-Interleave: Tackling Multi-image, Video, and 3D in Large Multimodal Models

    cs.CV 2024-07 unverdicted novelty 7.0 of 10

    LLaVA-NeXT-Interleave unifies multi-image, video, and 3D capabilities in large multimodal models via a new 1.18M-sample interleaved dataset and benchmark, achieving leading results across those tasks while preserving ...

  2. MuirBench: A Comprehensive Benchmark for Robust Multi-image Understanding

    cs.CV 2024-06 conditional novelty 7.0 of 10

    MuirBench is a new benchmark showing that top multimodal LLMs struggle with robust multi-image understanding, with GPT-4o at 68% and open-source models below 33% accuracy.

  3. In-Context Learning for Wound Classification with Small Multimodal Language Models

    cs.CV 2026-07 conditional novelty 6.0 of 10

    Retrieval-based in-context learning, not zero-shot prompting, drives wound-classification gains in small multimodal models, with Qwen 3.5 27B reaching 0.872 accuracy on Kaggle and 0.678 on Medetec.

  4. When Youth Enter the Algorithmic Wild: Discovering and Understanding Potentially Harmful Teen Videos on Douyin and Kwai

    cs.CR 2026-05 unverdicted novelty 6.0 of 10

    PHTV-Scout measures 6.11% prevalence of potentially harmful teen videos on Douyin and Kwai (53.2% child sexual exploitation imagery), shows Youth Mode blocks all such content but is used by only 30-41% of teens, and a...

  5. Personal Visual Context Learning in Large Multimodal Models

    cs.CV 2026-05 unverdicted novelty 6.0 of 10

    Introduces Personal VCL formalization and benchmark revealing LMM context gaps, plus an Agentic Context Bank baseline that boosts personalized visual reasoning.

  6. How Well Does GPT-4o Understand Vision? Evaluating Multimodal Foundation Models on Standard Computer Vision Tasks

    cs.CV 2025-07 unverdicted novelty 6.0 of 10

    Multimodal foundation models achieve respectable but sub-specialist performance on semantic vision tasks and weaker results on geometric tasks when evaluated through prompt chaining on established benchmarks.

  7. Visual Species Recognition with Large Multimodal Models as Post-Hoc Correctors

    cs.LG 2025-12 conditional novelty 5.0 of 10

    LMMs underperform few-shot experts on species recognition, but re-ranking the expert's top-5 candidates with an LMM improves mean accuracy by 6.4 points across five benchmarks.

  8. Inference-Time Agentic Decision Rules Beat Longer Evolving Search for Multi-Image Medical Reasoning

    cs.CV 2026-07 conditional novelty 4.0 of 10

    On MedFrameQA, order-vote (57.89%) beats fixed prompting (52.73%) and order-rerank (55.79%), and a single 100-generation run drops final-test accuracy to 56.02%.

Pith tools