RAG in a Trenchcoat: When Minimal Memory Is Enough for Agentic Systems, and When It Isn’t

Jingyu Liu, Zongze Li, Zach Xu, Zhanhui Zhou and 3 more

Sydney Poster Session 2, Tue, Dec 8, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

Beyond 3 Million Tokens: A Multi-Modal Foundation Model for Full-Resolution Heliophysics

Sujit Roy, Johannes Schmude, Ata A Asanjan, Thorsten Kurth and 12 more

Atlanta Poster Session 4, Thu, Dec 10, 4:30 PM–7:30 PM, Hall C1 · Published 2026

MLLMs Fail to Refuse when Using Tools Agentically

Rikiya Takehi, Ryo Hachiuma, Shaona Ghosh, Dan Zhao and 2 more

Atlanta Poster Session 6, Fri, Dec 11, 4:30 PM–7:30 PM, Hall C1 · Published 2026

MulTaBench: Benchmarking Multimodal Tabular Learning with Text and Image

MulTaBench benchmarks 40 multimodal tabular datasets and shows target-aware tuning of text and image embeddings improves predictive performance over frozen embeddings.

Alan Arazi, Eilam Shapira, Shoham Grunblat, Mor Ventura and 7 more

Paris Poster Session 5, Fri, Dec 11, 11:30 AM–1:30 PM, Paris Poster Hall · Published 2026 · ▲ 142 on Hugging Face

Low-Rank Adaptation for Critic Learning in Off-Policy Reinforcement Learning

Low-rank adaptation regularizes critic learning by constraining updates to low-dimensional subspaces via frozen base weights, reducing loss and improving off-policy RL performance.

Yuan Zhuang, Yuexin Bian, Sihong He, Jie Feng and 6 more

Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Retrieval from Within: An Intrinsic Capability of Attention-Based Models

INTRA unifies retrieval and generation via decoder attention over internal encoder states, outperforming engineered RAG on question-answering benchmarks.

Elad Hoffer, Yochai Blau, Edan kinderman, Ron Banner and 2 more

Paris Poster Session 4, Thu, Dec 10, 5:30 PM–7:30 PM, Paris Poster Hall · Published 2026 · ▲ 10 on Hugging Face

Normalized Architectures are Natively 4-Bit

Normalizing weights and hidden states to the unit hypersphere makes nGPT robust to 4-bit arithmetic, enabling stable end-to-end NVFP4 training without extra scaling or transforms.

Maxim Fishman, Brian Chmiel, Ron Banner, Daniel Soudry and 1 more

Paris Poster Session 2, Wed, Dec 9, 5:00 PM–7:00 PM, Paris Poster Hall · Published 2026 · ▲ 2 on Hugging Face