Who Is Left of Whom? Tracing Spatial Evidence and Role Binding in Relative-Position Reasoning

arXiv:2609.35486 · cs.CL, cs.CV · Submitted 2026-09-28 · Read on arXiv

cs.CL, cs.CV

Submitted: 2026-09-28

Updated: 2026-09-28

Project page: https://esi-bench.github.io

Terminology

• Consistency — 43× in this paper · explained in 3 episodes
• Steering — 41× in this paper · explained in 2 episodes
• Accuracy — 35× in this paper · explained in 1 episode
• Benchmark — 16× in this paper · explained in 2 episodes
• Activation Patching — 14× in this paper · explained in 2 episodes
• Benchmarking — 9× in this paper · explained in 3 episodes
• Activation Steering — 4× in this paper · explained in 4 episodes
• Interpretability — 4× in this paper · explained in 15 episodes
• Mechanistic interpretability — 3× in this paper · explained in 2 episodes
• Robustness — 3× in this paper · explained in 5 episodes
• Transformer — 3× in this paper · explained in 2 episodes
• Alignment — 2× in this paper · explained in 1 episode
• Conference on Computer Vision and Pattern Recognition — 2× in this paper
• Trace — 2× in this paper · explained in 2 episodes
• Aggregation — 1× in this paper · explained in 1 episode
• Foundation Model — 1× in this paper · explained in 3 episodes
• Foundation Models — 1× in this paper · explained in 7 episodes
• Generative AI — 1× in this paper · explained in 2 episodes
• Hallucination — 1× in this paper · explained in 3 episodes
• Large Language Models (LLMs) — 1× in this paper · explained in 6 episodes
• Quantization — 1× in this paper · explained in 12 episodes
• Temporal Reasoning — 1× in this paper · explained in 2 episodes
• World Model — 1× in this paper · explained in 7 episodes

Sources

• Sparse Autoencoders Find Highly Interpretable Features in Language Models
• MINER: Mining the Underlying Pattern of Modality-Specific Neurons in Multimodal Large Language Models
• SpatialText: A Pure-Text Cognitive Benchmark for Spatial Understanding in Large Language Models
• LatentLens: Revealing Highly Interpretable Visual Tokens in LLMs
• Beyond Semantics: Rediscovering Spatial Awareness in Vision-Language Models
• Pathways of Visual Information Flow in Vision-Language Models
• Qwen3 Technical Report
• Steering Language Models With Activation Engineering
• InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency
• Towards AI-Complete Question Answering: A Set of Prerequisite Toy Tasks

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering