LLM activations encode current and prior entities in orthogonal slots, but models only use the current slot for explicit factual retrieval despite prior-slot information being linearly decodable.
How do Language Models Bind Entities in Context?
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
roles
background 2polarities
background 2representative citing papers
Large language models encode relational bindings via a cell-based representation: a low-dimensional linear subspace in which each cell corresponds to an entity-relation index pair and attributes are retrieved from the matching cell.
Transformer hidden states contain rank-indexed orientation signatures for true r-argument relations (r=3-6) that survive surface controls and can be patched to alter model outputs on relation tasks.
LLMs exhibit the Position Curse, with backward position retrieval in lists lagging far behind forward retrieval, showing only partial gains from PosBench fine-tuning.
Activation patching provides evidence about neural network circuits when the choice of metric is aligned with the hypothesis and common interpretation errors are avoided.
The paper introduces a taxonomy of AI safety for LLMs organized into Trustworthy AI, Responsible AI, and Safe AI perspectives, accompanied by a review of state-of-the-art methods, challenges, and future directions.
citing papers explorer
-
Slot Machines: How LLMs Keep Track of Multiple Entities
LLM activations encode current and prior entities in orthogonal slots, but models only use the current slot for explicit factual retrieval despite prior-slot information being linearly decodable.
-
Cell-Based Representation of Relational Binding in Language Models
Large language models encode relational bindings via a cell-based representation: a low-dimensional linear subspace in which each cell corresponds to an entity-relation index pair and attributes are retrieved from the matching cell.
-
Relational Rank Geometry in Transformers: Detecting and Steering Hidden-State Relation Frames
Transformer hidden states contain rank-indexed orientation signatures for true r-argument relations (r=3-6) that survive surface controls and can be patched to alter model outputs on relation tasks.
-
The Position Curse: LLMs Struggle to Locate the Last Few Items in a List
LLMs exhibit the Position Curse, with backward position retrieval in lists lagging far behind forward retrieval, showing only partial gains from PosBench fine-tuning.
-
How to use and interpret activation patching
Activation patching provides evidence about neural network circuits when the choice of metric is aligned with the hypothesis and common interpretation errors are avoided.
-
AI Safety Landscape for Large Language Models: Taxonomy, State-of-the-art, and Future Directions
The paper introduces a taxonomy of AI safety for LLMs organized into Trustworthy AI, Responsible AI, and Safe AI perspectives, accompanied by a review of state-of-the-art methods, challenges, and future directions.