ActICL: Post-Training for Agentic Multimodal In-Context Learning

arXiv:2610.19475 · cs.CV, cs.CL · Submitted 2026-10-14 · Read on arXiv

cs.CV, cs.CL

Submitted: 2026-10-14

Updated: 2026-10-14

Comments: Project page: https://kaamava.github.io/acticl-project-page/

Code: https://github.com/lupantech/PromptPG

Project page: https://kaamava.github.io/acticl-project-page

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• In-Context Learning — 1× in this paper · explained in 7 episodes

Sources

• GRIP: Feedback-Guided Prompt Retrieval for Large Multimodal Models
• Qwen3-VL Technical Report
• BabyVision: Visual Reasoning Beyond Language
• OpenSearch-VL: An Open Recipe for Frontier Multimodal Search Agents
• True Multimodal In-Context Learning Needs Attention to the Visual Context
• Molmo2: Open Weights and Data for Vision-Language Models with Video Understanding and Grounding
• WebWatcher: Breaking New Frontier of Vision-Language Deep Research Agent
• DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
• Many-Shot In-Context Learning in Multimodal Foundation Models
• PerceptionBench: Evaluating Atomic Visual Perception in Multimodal Large Language Models
• Understanding R1-Zero-Like Training: A Critical Perspective
• Dynamic Prompt Learning via Policy Gradient for Semi-structured Mathematical Reasoning
• DeepMMSearch-R1: Empowering Multimodal LLMs in Multimodal Web Search
• Gradient Starvation in Binary-Reward GRPO: Why Group-Mean Centering Fails and Why the Simplest Fix Works
• DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
• Order Matters: Exploring Order Sensitivity in Multimodal Large Language Models
• MuirBench: A Comprehensive Benchmark for Robust Multi-image Understanding
• MindCube: Spatial Mental Modeling from Limited Views
• MME-Reasoning: A Comprehensive Benchmark for Logical Reasoning in MLLMs
• Sigmoid Loss for Language Image Pre-Training

Related papers

• Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
• AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
• Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
• MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
• TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
• A Survey on Efficient Vision-Language-Action Models