Circuit Oracle is an off‑the‑shelf large language model that, using a multi‑agent harness and task‑specific skills, can answer natural‑language questions about a target model by reading its attribution graph via tool calls. The system was evaluated on three safety‑relevant proxy tasks—detecting spurious features in probe circuits, eliciting hidden knowledge from taboo‑finetuned models, and suppression jailbreaking via feature‑level interventions—achieving 78.2 % accuracy on spurious‑feature detection and mixed results on the other tasks. In addition to task scores, Circuit Oracle produces pruned subgraphs and context‑specific feature explanations, which the authors release for further study.
By Shariar Kabir, Hong Kiat Tan, Swastik Agrawal, Sai V R Chereddy, Sriram Balasubramanian
arXiv:2609.23892v1 Announce Type: new
Abstract: Mechanistic interpretability defines features as the fundamental units of a neural network and circuits as the weighted subgraphs that carry out its co...
By Edward G. Friedman, Xiangchen Song
arXiv:2603. 21014v2 Announce Type: replace Abstract: Mechanistic interpretability seeks to understand how Large Language Models (LLMs) represent and process information.
By Florent Draye, Vedant Palit, Abir Harrasse, Tung-Yu Wu, Jiarui Liu, Punya Syon Pandey, Roderick Wu, Chih-Hao Hsu, Terry Jingchen Zhang, Zhijing Jin, Bernhard Sch\"olkopf
The paper investigates graph soft prompting, where a graph neural network generates graph-specific tokens for large language model prompts. It identifies a problem called token collapse, in which these tokens become nearly identical across different inputs, rendering them ineffective at conveying graph information. The authors show that even simple constant prompts can match or outperform collapsed tokens, and propose a two‑stage method to produce more varied tokens while maintaining performance.
By Andrea Giuseppe Di Francesco, Rishabh Jain, Franco Maria Nardini, Nicola Tonellotto, Pietro Li\`o, Fabrizio Silvestri
arXiv:2606. 28548v1 Announce Type: cross Abstract: Sparse autoencoders (SAEs) have become a useful tool for extracting interpretable features in language models.
By Kevin Der, Harish Kamath, Ben Thompson
S^3martCirc is a self‑supervised framework that jointly discovers and interprets neural circuits in large language models, rather than treating circuit discovery and functional interpretation as separate stages. It abstracts node behavior into two general computational roles that generalize across tasks and introduces a quantitative metric for assigning these roles, enabling simultaneous identification of importance and function. Experiments demonstrate that S^3martCirc outperforms existing methods in circuit discovery.
By Wendy Zheng, Yinhan He, Liang Wu, Jundong Li
arXiv:2512. 14332v2 Announce Type: replace-cross Abstract: The field of Language Reasoning Models (LRMs) has been very active over the past few years with advances in training and inference techniques enabling LRMs to reason longer, and more accurately.
By Yannis Belkhiter, Seshu Tirupathi, Giulio Zizzo, John D. Kelleher
arXiv:2607. 22610v1 Announce Type: new Abstract: When a language model produces a response in a multi-turn conversation, which tokens from prior turns shaped that answer, and how did those dependencies propagate across prior turns?
By Jessica Tang, Shraddha Barke, Sharad Agarwal
arXiv:2601. 22594v2 Announce Type: replace-cross Abstract: The high-level concepts that a neural network uses to perform computation need not be aligned to individual neurons (Smolensky, 1986).
By Aryaman Arora, Zhengxuan Wu, Jacob Steinhardt, Sarah Schwettmann
The paper introduces Concept-Targeted Attribution (CTA), a method that trains attribution graphs to explain the emergence of internal concept representations in language models, rather than just the final token prediction. CTA produces probe-specific circuits that reveal which internal computations drive a linear probe’s accuracy, and cross-layer transcoders demonstrate that these graphs contain predictive structure across multiple concept categories. Causal ablations show that probe-targeted and logit-targeted graphs capture distinct mechanisms, with probe-relevant features affecting internal concept scores and logit-relevant features altering generated tokens.
By Vedant Palit, Florent Draye, Terry Jingchen Zhang, Bernhard Sch\"olkopf, Zhijing Jin
RAW-Explainer is a new explainability method designed for path reasoning graph neural networks used in knowledge graph completion. It produces connected, concise, and interpretable subgraph explanations for link prediction by leveraging heterogeneous information and a random walk objective. The approach also addresses distribution shift when evaluating small explanatory subgraphs with a robust evaluator, achieving a balance between explanation quality and computational efficiency on standard benchmarks.
By Ryoji Kubo, Samyam Lamichhane, Djellel Difallah
As artificial intelligence and machine learning (AI/ML) models become integral to network operations, their lack of transparency poses a significant barrier to operator trust. Existing explainable artificial intelligence (XAI) techniques often fail to bridge this gap for non-specialists, producing technical outputs that are difficult to translate into actionable insights.