CIRCUS: Circuit Consensus under Uncertainty via Stability Ensembles

• URL: http://arxiv.org/abs/2603.00523v1
• Date: Sat, 28 Feb 2026 07:44:04 GMT
• Title: CIRCUS: Circuit Consensus under Uncertainty via Stability Ensembles
• Abstract summary: We show that CIRCUS produces a threshold-robust "core" circuit while explicitly surfacing contingent alternatives.<n>On Gemma-2-2B and Llama-3.2-1B, strict consensus circuits are 40x smaller than the union of constructs all configurations while retaining comparable influence-flow explanatory power.<n>We further validate causal relevance with activation patching, where consensus-identified nodes consistently beat matched non-consensus controls.
• Score: 1.3011345529764784
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Mechanistic circuit discovery is notoriously sensitive to arbitrary analyst choices, especially pruning thresholds and feature dictionaries, often yielding brittle "one-shot" explanations with no principled notion of uncertainty. We reframe circuit discovery as an uncertainty-quantification problem over these analytic degrees of freedom. Our method, CIRCUS, constructs an ensemble of attribution graphs by pruning a single raw attribution run under multiple configurations, assigns each edge a stability score (the fraction of configurations that retain it), and extracts a strict-consensus circuit consisting only of edges that appear in all views. This produces a threshold-robust "core" circuit while explicitly surfacing contingent alternatives and enabling rejection of low-agreement structure. CIRCUS requires no retraining and adds negligible overhead, since it aggregates structure across already-computed pruned graphs. On Gemma-2-2B and Llama-3.2-1B, strict consensus circuits are ~40x smaller than the union of all configurations while retaining comparable influence-flow explanatory power, and they outperform a same-edge-budget baseline (union pruned to match the consensus size). We further validate causal relevance with activation patching, where consensus-identified nodes consistently beat matched non-consensus controls (p=0.0004). Overall, CIRCUS provides a practical, uncertainty-aware framework for reporting trustworthy, auditable mechanistic circuits with an explicit core/contingent/noise decomposition.

Related papers

• Byzantine Accountability Without Consensus: Strong Eventual Consistency for Non-Associative, Stochastic, Robust Aggregation [0.0]
  ACFA replicates a content-addressed OR-Set of signed contributions and a grow-only set of self-authenticating equivocation proofs.<n>A prototype (10 nodes, 3 Byzantine) passes 16/16 falsification checks.
  arXiv  Detail & Related papers  (2026-07-11T13:23:06Z)
• Closing the Null Space: Guidance-Aware Quantization for Classifier-Free Diffusion [0.0]
  Two-pass CFG execution pattern imposes a latency overhead that parameter-count and bit-operation metrics conceal entirely.<n>Guidance-Aware Mixed Precision (GAMP) is proposed, which calibrates directly on the guided prediction.
  arXiv  Detail & Related papers  (2026-07-09T08:42:07Z)
• C$^{2}$R: Cross-sample Consistency Regularization Mitigates Feature Splitting and Absorption in Sparse Autoencoders [56.67585330425431]
  Sparse Autoencoders (SAEs) are widely used to interpret large language models by decomposing activations into sparse, human-understandable features.<n>These issues stem from inconsistent latent assignment across samples.<n>We introduce C$2$R, which penalizes the co-activation of directionally similar latents.
  arXiv  Detail & Related papers  (2026-06-29T17:45:31Z)
• PerceptionRubrics: Calibrating Multimodal Evaluation to Human Perception [75.70273182182397]
  PerceptionRubrics is a rubric-based evaluation framework.<n>It addresses the gap between saturated benchmark scores and real-world brittleness.<n>It pairs 1,038 information-dense images with over 10,000 instance-specific rubrics.
  arXiv  Detail & Related papers  (2026-06-26T17:59:15Z)
• R-APS: Compositional Reasoning and In-Context Meta-Learning for Constrained Design via Reflective Adversarial Pareto Search [18.8323743697237]
  Large language models (LLMs) are fluent on open-ended tasks, yet in agentic settings, fluency does not ensure reliable delivery.<n>We trace this gap to three coupled structural failures: errors propagate without localization, worst-case perturbations go unevaluated, and accumulated knowledge is never invalidated.<n>We argue these share a root cause: abductive, counterfactual, meta-inductive, corrective, and inductive reasoning pull a shared context in incompatible directions.
  arXiv  Detail & Related papers  (2026-06-03T12:45:39Z)
• MIST: Reliable Streaming Decision Trees for Online Class-Incremental Learning via McDiarmid Bound [6.276425555068115]
  Streaming decision trees are natural candidates for open-world continual learning.<n>They still fail in online class-incremental learning due to two coupled miscalibrations.<n>We present MIST (McDiarmid Incremental Streaming Tree), which resolves both failures through three integrated components.
  arXiv  Detail & Related papers  (2026-05-12T06:45:00Z)
• Learning from Emptiness: De-biasing Listwise Rerankers with Content-Agnostic Probability Calibration [76.08899010904652]
  CapCal is a training-free framework that mechanically decouples positional bias from ranking decisions.<n>It achieves superior performance among training-free methods while preserving single-pass efficiency.
  arXiv  Detail & Related papers  (2026-04-11T10:47:22Z)
• GroupEnsemble: Efficient Uncertainty Estimation for DETR-based Object Detection [10.241307426902178]
  GroupEnsemble is an efficient and effective uncertainty estimation method for DETR-like models.<n>We validated our method under autonomous driving scenes and common daily scenes using the Cityscapes and COCO datasets.<n>The results show that a hybrid approach combining MC-Dropout and GroupEnsemble outperforms Deep Ensembles on several metrics at a fraction of the cost.
  arXiv  Detail & Related papers  (2026-03-02T13:26:40Z)
• Certified Circuits: Stability Guarantees for Mechanistic Circuits [80.30622018787835]
  Certified Circuits provides provable stability guarantees for circuit discovery.<n>On ImageNet and OOD datasets, certified circuits achieve up to 91% higher accuracy.
  arXiv  Detail & Related papers  (2026-02-26T13:07:31Z)
• Assigning Confidence: K-partition Ensembles [6.194291632696817]
  We introduce CAKE, a framework that evaluates each point using two complementary statistics computed over a clustering ensemble.<n>C CAKE effectively highlights ambiguous points and stable core members, providing a confidence ranking that can guide filtering or prioritization to improve clustering quality.
  arXiv  Detail & Related papers  (2026-02-20T18:59:53Z)
• Generalizing GNNs with Tokenized Mixture of Experts [75.8310720413187]
  We show that improving stability requires reducing reliance on shift-sensitive features, leaving an irreducible worst-case generalization floor.<n>We propose STEM-GNN, a pretrain-then-finetune framework with a mixture-of-experts encoder for diverse computation paths.<n>Across nine node, link, and graph benchmarks, STEM-GNN achieves a stronger three-way balance, improving robustness to degree/homophily shifts and to feature/edge corruptions while remaining competitive on clean graphs.
  arXiv  Detail & Related papers  (2026-02-09T22:48:30Z)
• CORE: Context-Robust Remasking for Diffusion Language Models [51.59514489363897]
  We propose Context-Robust Remasking (CORE), a training-free framework for inference-time revision.<n>Rather than trusting static token probabilities, CORE identifies context-brittle tokens by probing their sensitivity to targeted masked-context perturbations.<n>On LLaDA-8B-Base, CORE delivers consistent improvements across reasoning and code benchmarks, outperforming compute-matched baselines and improving MBPP by up to 9.2 percentage points.
  arXiv  Detail & Related papers  (2026-02-04T00:12:30Z)
• Unsupervised Conformal Inference: Bootstrapping and Alignment to Control LLM Uncertainty [49.19257648205146]
  We propose an unsupervised conformal inference framework for generation.<n>Our gates achieve close-to-nominal coverage and provide tighter, more stable thresholds than split UCP.<n>The result is a label-free, API-compatible gate for test-time filtering.
  arXiv  Detail & Related papers  (2025-09-26T23:40:47Z)
• SIM-CoT: Supervised Implicit Chain-of-Thought [108.30049193668083]
  Implicit Chain-of-Thought (CoT) methods offer a token-efficient alternative to explicit CoT reasoning in Large Language Models.<n>We identify a core latent instability issue when scaling the computational budget of implicit CoT.<n>We propose SIM-CoT, a plug-and-play training module that introduces step-level supervision to stabilize and enrich the latent reasoning space.
  arXiv  Detail & Related papers  (2025-09-24T17:01:32Z)
• Certified Pruning from Counterfactual Consistency: Exact Certificates and Structured SAT Families [0.0]
  We formalize epsilon-counterfactual instrumentation and epsilon-stability, capturing noisy but testable constraints in laboratory contextuality experiments.<n>We extend these constraints to networks of contexts, yielding contextuality-type inequalities that apply globally across a CNF-SAT instance.<n>We implement a propagate-and-prune solver in which every learned clause is certified by a dual Farkas certificate verified in exact arithmetic.
  arXiv  Detail & Related papers  (2025-09-04T21:13:31Z)
• Uncertainty-Aware Relational Graph Neural Network for Few-Shot Knowledge Graph Completion [12.887073684904147]
  Few-shot knowledge graph completion (FKGC) aims to query the unseen facts of a relation given its few-shot reference entity pairs. Existing FKGC works neglect such uncertainty, which leads them more susceptible to limited reference samples with noises. We propose a novel uncertainty-aware few-shot KG completion framework (UFKGC) to model uncertainty for a better understanding of the limited data.
  arXiv  Detail & Related papers  (2024-03-07T14:23:25Z)
• Corruption-Tolerant Gaussian Process Bandit Optimization [130.60115798580136]
  We consider the problem of optimizing an unknown (typically non-producing) function with a bounded norm. We introduce an algorithm based on Fast-Slow GP-UCB based on "fast but non-robust" and "slow" We argue that certain dependencies cannot be required depending on the corruption level.
  arXiv  Detail & Related papers  (2020-03-04T09:46:58Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.