Lattice AI Research

Research focus

Multimodal Models (3)Eval Frameworks & Benchmarks (2)Recommendation & Information Retrieval (1)Reasoning & Chain-of-Thought (1)

Frequent co-authors

Taku Hasegawa (1)Kyosuke Nishida (1)Haruka Kawasaki (1)Haruto Yoshida (1)

Papers (3)

Jul 7, 2026

Ryota Tanaka +26d ago

CMDR: Contextual Multimodal Document Retrieval

Context-aware embeddings can drastically improve multimodal document retrieval, revealing insights that independent page evaluations miss.

Ryota Tanaka, Taku Hasegawa, Kyosuke Nishida

Multimodal Models Recommendation & Information Retrieval

Apr 6, 2026

Haruka Kawasaki +1Apr 6, 2026

Responses Fall Short of Understanding: Revealing the Gap between Internal Representations and Responses in Visual Document Understanding

LVLMs know more than they say: critical information for visual document understanding is often better represented in intermediate layers than in the final output, suggesting a bottleneck in the generation process.

Haruka Kawasaki, Ryota Tanaka

Eval Frameworks & Benchmarks Multimodal Models Reasoning & Chain-of-Thought

Mar 3, 2026

Nodes Are Early, Edges Are Late: Probing Diagram Representations in Large Vision-Language Models

LVLMs encode nodes early but edges late, suggesting a fundamental bottleneck in how these models process relational information in diagrams.

Haruto Yoshida, Keito Kudo, Yoichi Aoki +4

Eval Frameworks & Benchmarks Interpretability & Mechanistic Interp Multimodal Models

Search

Ryota Tanaka

Publication activitypapers/week, last 8 weeks

Research focus

Frequent co-authors

Papers (3)