REVIEW 30 cited by
Dissecting Recall of Factual Associations in Auto-Regressive Language Models
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Transformer-based language models (LMs) are known to capture factual knowledge in their parameters. While previous work looked into where factual associations are stored, only little is known about how they are retrieved internally during inference. We investigate this question through the lens of information flow. Given a subject-relation query, we study how the model aggregates information about the subject and relation to predict the correct attribute. With interventions on attention edges, we first identify two critical points where information propagates to the prediction: one from the relation positions followed by another from the subject positions. Next, by analyzing the information at these points, we unveil a three-step internal mechanism for attribute extraction. First, the representation at the last-subject position goes through an enrichment process, driven by the early MLP sublayers, to encode many subject-related attributes. Second, information from the relation propagates to the prediction. Third, the prediction representation "queries" the enriched subject to extract the attribute. Perhaps surprisingly, this extraction is typically done via attention heads, which often encode subject-attribute mappings in their parameters. Overall, our findings introduce a comprehensive view of how factual associations are stored and extracted internally in LMs, facilitating future research on knowledge localization and editing.
Forward citations
Cited by 30 Pith papers
-
CATCH-ME if you RAG: a dataset of Contextually Annotated multi-Turn Counterspeech against Hate and Misinformation Exchanges
Presents a new expert-curated dataset of multi-turn counterspeech dialogues in five languages targeting hate against seven groups, with span annotations linking to verified external knowledge for RAG applications.
-
Verbalizable Representations Form a Global Workspace in Language Models
Language models represent their current reasoning in a small, readable set of verbalizable vectors (the J-space) that functions like a global workspace.
-
Turn-Averaged SAEs for Feature Discovery and Long-Context Attribution
Turn-averaged SAEs reconstruct average activations over conversation turns to represent high-level turn characteristics with a fixed number of features, simplifying long-context interpretability compared to per-token SAEs.
-
TimeROME-DLM: Temporal Causal Tracing and Low-Rank Inference-Time Knowledge Editing for Masked Diffusion Language Models
TimeROME-DLM enables training-free knowledge editing in masked diffusion language models via temporal causal tracing and low-rank residual edit memory applied at inference time.
-
PRISM: Recovering Instruction Sets from Language Model Activations
PRISM is a new activation-conditioned model that recovers full sets of simultaneous instructions from LLM hidden states via judge-guided GRPO training and outperforms prior activation-to-language methods on security-r...
-
When LLMs Stop Following Steps: A Diagnostic Study of Procedural Execution in Language Models
LLM first-answer accuracy on procedural arithmetic drops from 61% on 5-step tasks to 20% on 95-step tasks, with frequent failures including skipped steps, premature answers, and hallucinated operations.
-
V-SEAM: Visual Semantic Editing and Attention Modulating for Causal Interpretability of Vision-Language Models
V-SEAM combines concept-level visual semantic editing with attention head modulation to identify positive and negative contributors across object, attribute, and relationship levels, then uses this to improve VLM perf...
-
A Mechanistic View of Authority Hierarchy in LLM Sycophancy
Authority sycophancy in LLMs is a layer-localized erasure of correct answer representations that scales with authority level and resists simple interventions.
-
Perfect Detection, Failed Control: The Geometry of Knowing vs. Steering in Language Models
Across four language models, the linear direction detecting hallucinated entities achieves perfect separability yet lies at cosine 0.12 from the direction that induces refusal, with the gap unchanged by instruction tuning.
-
Factual Retrieval in LLMs Is a Redundant, Distributed and Non-Contiguous Process
Attribute retrieval in LLMs follows non-contiguous, redundant layer paths identified via iterative patching, implying highly distributed knowledge storage.
-
Assisted Counterspeech Writing at the Crossroads of Hate Speech and Misinformation
LLMs generate adequate counterspeech for co-occurring hate and misinformation in 40% of cases, with a mixed knowledge strategy from fact-checkers and NGOs proving most effective after expert revision.
-
Not Just RLHF: Why Alignment Alone Won't Fix Multi-Agent Sycophancy
Base LLMs show multi-agent yield to peer pressure at rates equal to or higher than aligned models, localized by activation patching to mid-layers where attention dominates, with one dissenter cutting yield by 54-73 po...
-
Not Just RLHF: Why Alignment Alone Won't Fix Multi-Agent Sycophancy
Pretrained base models exhibit higher yield to peer disagreement than RLHF instruct variants, with the effect localized to mid-layer attention and mitigated by structured dissent rather than prompt defenses.
-
Correcting Influence: Unboxing LLM Outputs with Orthogonal Latent Spaces
A latent mediation framework with sparse autoencoders enables non-additive token-level influence attribution in LLMs by learning orthogonal features and back-propagating attributions.
-
Tool Calling is Linearly Readable and Steerable in Language Models
Tool identity is linearly readable and steerable in LLMs via mean activation differences, with 77-100% switch accuracy and error prediction from activation gaps.
-
When LLMs Stop Following Steps: A Diagnostic Study of Procedural Execution in Language Models
A new benchmark shows LLM first-answer accuracy on procedural arithmetic drops from 63% (5 steps) to 20% (95 steps) due to execution failures like skipped steps and premature answers.
-
When LLMs Stop Following Steps: A Diagnostic Study of Procedural Execution in Language Models
LLM accuracy on controlled procedural arithmetic drops from 61% at 5 steps to 20% at 95 steps, with failures including skipped steps, premature answers, and hallucinated operations.
-
The Override Gap: A Magnitude Account of Knowledge Conflict Failure in Hypernetwork-Based Instant LLM Adaptation
Knowledge conflicts in hypernetwork LLM adaptation stem from constant adapter margins losing to frequency-dependent pretrained margins; selective layer boosting and conflict-aware triggering raise deep-conflict accura...
-
From Signal Degradation to Computation Collapse: Uncovering the Two Failure Modes of LLM Quantization
LLM 2-bit quantization fails via either cumulative signal degradation or early computation collapse in key components.
-
How do LLMs Compute Verbal Confidence
Mechanistic experiments on Gemma 3 27B, Qwen 2.5 7B and Magistral Small 24B show verbal confidence is cached at post-answer positions from answer tokens and captures richer answer-quality information beyond token log-...
-
Prototype Transformer: Towards Language Model Architectures Interpretable by Design
ProtoT is an autoregressive language model whose attention is replaced by learned prototype channels that are claimed to capture nameable concepts and allow targeted edits, at linear sequence cost but with slightly lo...
-
Geometry of Reason: Spectral Signatures of Valid Mathematical Reasoning
Spectral features of attention are claimed to classify proof validity with near-perfect effect sizes, but the main evaluation relabels proofs using the classifier's own outputs.
-
Attention Illuminates LLM Reasoning: The Preplan-and-Anchor Rhythm Enables Fine-Grained Policy Optimization
LLM attention maps reveal a preplan-and-anchor pattern, and reweighting RL credit toward the flagged tokens improves math/QA reasoning.
-
How Training Data Shapes the Use of Parametric and In-Context Knowledge in Language Models
Balanced parametric and in-context knowledge use in LLMs is an emergent property requiring intra-document repetition, moderate inconsistency, and skewed distributions in training data.
-
When LLMs Stop Following Steps: A Diagnostic Study of Procedural Execution in Language Models
Across 15 LLMs, first-answer accuracy on arithmetic procedural execution falls from 63% at 5 steps to 20% at 95 steps, with under-execution increasing.
-
The Override Gap: A Magnitude Account of Knowledge Conflict Failure in Hypernetwork-Based Instant LLM Adaptation
Knowledge conflicts in hypernetwork LLM adaptation stem from constant adapter margins losing to frequency-dependent pretrained margins; selective layer boosting and conflict-aware triggering close the gap.
-
Linear Relational Decoding of Morphology in Language Models
A single matrix multiplication on middle-layer word states reproduces morphological transformations in GPT-J and Llama-7b with roughly 90% faithfulness.
-
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.
-
Towards Best Practices of Activation Patching in Language Models: Metrics and Methods
Varying evaluation metrics and corruption methods in activation patching produces different localization and circuit discovery outcomes in language models, leading to recommendations for preferred practices.
-
NEAT: Concept driven Neuron Attribution in LLMs
NEAT identifies concept neurons by feeding a single mean hidden-state vector through the model and ranking neurons by their effect on concept-word probabilities.
Discussion (0). Sign in to comment.