Multi-component Causal Tracing in Large Language Models

• URL: http://arxiv.org/abs/2606.03085v1
• Date: Tue, 02 Jun 2026 03:15:35 GMT
• Title: Multi-component Causal Tracing in Large Language Models
• Abstract summary: Causal tracing intervenes on a large language model's internal representations.<n>This paper presents a unified framework for causally tracing multiple components simultaneously.
• Score: 36.05807217872701
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Causal tracing systematically intervenes on a large language model's (LLM's) internal representations to uncover and quantify the causal pathways linking specific inputs or computations to specific metrics of interest, quantifying the LLM's behavior. Building on previous single-component or single-layer studies, this paper presents a unified framework for causally tracing multiple components simultaneously. This framework systematically identifies the subsets of components (e.g., attention heads and multi-layer perceptron neurons) most critical to a desired target performance metric (e.g., accuracy and fairness). This is achieved by incorporating flexible interventions applied to a wide range of desired metrics. To address the combinatorial complexity of the multi-component problem, an efficient algorithm is designed that leverages soft interventions and a carefully designed metric transformation, converting the combinatorial search problem into a continuous one that can be solved efficiently under proper constraints, thereby generating proper binary decisions for selecting components. Experimental results demonstrate that the proposed method efficiently identifies subsets of the model's components that have a high impact on the target metric, outperforming existing baseline approaches. Our code is available at https://github.com/ZiruiYan/multi-component-causal-tracing.

Related papers

• Learning Multi-Indicator Weights for Data Selection: A Joint Task-Model Adaptation Framework with Efficient Proxies [50.39041754816285]
  We propose a framework for learning multi-validation weights that jointly adapts data selection to both the downstream task and the specific model.<n>Our approach achieves performance comparable to or exceeding full-dataset tuning while using only 30% of the training samples on GSM8K.
  arXiv  Detail & Related papers  (2026-05-10T17:30:16Z)
• A Learnable Fully Interacted Two-Tower Model for Pre-Ranking System [15.03225449071182]
  The two-tower model is widely used in pre-ranking systems due to a good balance between efficiency and effectiveness.<n>A novel architecture named learnable Fully Interacted Two-tower Model (FIT) is proposed, which enables rich information interactions.
  arXiv  Detail & Related papers  (2025-09-16T10:52:03Z)
• Improving Retrieval-Augmented Generation through Multi-Agent Reinforcement Learning [88.55095746156428]
  Retrieval-augmented generation (RAG) is widely utilized to incorporate external knowledge into large language models.<n>A standard RAG pipeline consists of several components, such as query rewriting, document retrieval, document filtering, and answer generation.<n>We propose treating the complex RAG pipeline with multiple components as a multi-agent cooperative task, in which each component can be regarded as an RL agent.
  arXiv  Detail & Related papers  (2025-01-25T14:24:50Z)
• FactorLLM: Factorizing Knowledge via Mixture of Experts for Large Language Models [50.331708897857574]
  We introduce FactorLLM, a novel approach that decomposes well-trained dense FFNs into sparse sub-networks without requiring any further modifications. FactorLLM achieves comparable performance to the source model securing up to 85% model performance while obtaining over a 30% increase in inference speed.
  arXiv  Detail & Related papers  (2024-08-15T16:45:16Z)
• Retrieval with Learned Similarities [2.729516456192901]
  State-of-the-art retrieval algorithms have migrated to learned similarities.<n>We show that Mixture-of-Logits (MoL) can be realized empirically to achieve superior performance on diverse retrieval scenarios.
  arXiv  Detail & Related papers  (2024-07-22T08:19:34Z)
• UCB-driven Utility Function Search for Multi-objective Reinforcement Learning [51.00436121587591]
  In Multi-objective Reinforcement Learning (MORL) agents are tasked with optimising decision-making behaviours.<n>We focus on the case of linear utility functions parametrised by weight vectors w.<n>We introduce a method based on Upper Confidence Bound to efficiently search for the most promising weight vectors during different stages of the learning process.
  arXiv  Detail & Related papers  (2024-05-01T09:34:42Z)
• Random Aggregate Beamforming for Over-the-Air Federated Learning in Large-Scale Networks [66.18765335695414]
  We consider a joint device selection and aggregate beamforming design with the objectives of minimizing the aggregate error and maximizing the number of selected devices. To tackle the problems in a cost-effective manner, we propose a random aggregate beamforming-based scheme. We additionally use analysis to study the obtained aggregate error and the number of the selected devices when the number of devices becomes large.
  arXiv  Detail & Related papers  (2024-02-20T23:59:45Z)
• Query-Efficient Correlation Clustering with Noisy Oracle [17.11782578276788]
  We introduce two novel formulations of online learning problems rooted in the paradigm of Pure Exploration in Combinatorial Multi-Armed Bandits (PE-CMAB) We design algorithms that combine a sampling strategy with a classic approximation algorithm for correlation and study their theoretical guarantees. Our results are the first examples of clustering-time algorithms that work for the case of PE-CMAB in which the underlying offline optimization problem is NP-hard.
  arXiv  Detail & Related papers  (2024-02-02T13:31:24Z)
• BOtied: Multi-objective Bayesian optimization with tied multivariate ranks [33.414682601242006]
  In this paper, we show a natural connection between non-dominated solutions and the extreme quantile of the joint cumulative distribution function. Motivated by this link, we propose the Pareto-compliant CDF indicator and the associated acquisition function, BOtied. Our experiments on a variety of synthetic and real-world problems demonstrate that BOtied outperforms state-of-the-art MOBO acquisition functions.
  arXiv  Detail & Related papers  (2023-06-01T04:50:06Z)
• Variable Importance Matching for Causal Inference [73.25504313552516]
  We describe a general framework called Model-to-Match that achieves these goals. Model-to-Match uses variable importance measurements to construct a distance metric. We operationalize the Model-to-Match framework with LASSO.
  arXiv  Detail & Related papers  (2023-02-23T00:43:03Z)
• Task-Oriented Sensing, Computation, and Communication Integration for Multi-Device Edge AI [108.08079323459822]
  This paper studies a new multi-intelligent edge artificial-latency (AI) system, which jointly exploits the AI model split inference and integrated sensing and communication (ISAC) We measure the inference accuracy by adopting an approximate but tractable metric, namely discriminant gain.
  arXiv  Detail & Related papers  (2022-07-03T06:57:07Z)
• 3D Medical Multi-modal Segmentation Network Guided by Multi-source Correlation Constraint [2.867517731896504]
  We propose a multi-modality segmentation network with a correlation constraint. Our experiment results tested on BraTS-2018 dataset for brain tumor segmentation demonstrate the effectiveness of our proposed method.
  arXiv  Detail & Related papers  (2021-02-05T11:23:12Z)
• A Deep Joint Sparse Non-negative Matrix Factorization Framework for Identifying the Common and Subject-specific Functional Units of Tongue Motion During Speech [7.870139900799612]
  We develop a new deep learning framework to identify common and subject-specific functional units of tongue motion during speech. We transform NMF with sparse and graph regularizations into modular architectures akin to deep neural networks.
  arXiv  Detail & Related papers  (2020-07-09T15:05:44Z)

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.