LLM Explainability with Counterfactual Chains and Causal Graphs

Hugging Face Daily Papers Papers

Summary

This paper proposes a four-phase method for constructing causal graphs that model LLM inference processes, using counterfactual augmentation to enable stable causal discovery and provide transparent, concept-level explainability.

Causal graphs provide a high-level language for making mechanisms transparent. Recent work uses Large Language Models (LLMs) to recover causal graphs of external-world processes. Instead, in this paper, we use causal graphs to model LLM inference itself, providing stakeholders with a transparent view of how the model perceives and organizes high-level concepts to produce a prediction. We propose a four-phase method for constructing such graphs. Given a target LLM and a set of textual examples, our method discovers class-discriminative, human-interpretable concepts and maps each input to LLM-perceived concept states. We then introduce an MCMC-inspired counterfactual augmentation procedure that expands the sparse observational data through chains of counterfactuals. This enables stable causal discovery with σ-CG, yielding informative, interpretable graphs. We apply our method to three LLMs across disease diagnosis, sentiment analysis, and LLM-as-a-judge classification tasks. We evaluate the learned graphs for predictive fidelity and structural stability, and the MCMC-inspired augmentation for convergence and downstream utility. Our results show that the discovered causal graphs capture meaningful dependencies consistent with LLMs' reasoning. Together, this paper provides a foundation for concept-level explainability of LLMs.
Original Article
View Cached Full Text

Cached at: 06/08/26, 07:14 AM

Paper page - LLM Explainability with Counterfactual Chains and Causal Graphs

Source: https://huggingface.co/papers/2606.05972

Abstract

Causal graphs are used to model large language model inference processes, enabling transparent visualization of how models perceive and organize high-level concepts for predictions through a four-phase method involving concept discovery, mapping, and MCMC-inspired counterfactual augmentation.

Causal graphsprovide a high-level language for making mechanisms transparent. Recent work usesLarge Language Models(LLMs) to recovercausal graphsof external-world processes. Instead, in this paper, we usecausal graphsto model LLM inference itself, providing stakeholders with a transparent view of how the model perceives and organizes high-level concepts to produce a prediction. We propose a four-phase method for constructing such graphs. Given a target LLM and a set of textual examples, our method discovers class-discriminative, human-interpretable concepts and maps each input to LLM-perceived concept states. We then introduce anMCMC-inspiredcounterfactual augmentationprocedure that expands the sparse observational data through chains of counterfactuals. This enables stablecausal discoverywith σ-CG, yielding informative, interpretable graphs. We apply our method to three LLMs across disease diagnosis, sentiment analysis, and LLM-as-a-judge classification tasks. We evaluate the learned graphs forpredictive fidelityandstructural stability, and theMCMC-inspiredaugmentation for convergence and downstream utility. Our results show that the discoveredcausal graphscapture meaningful dependencies consistent with LLMs’ reasoning. Together, this paper provides a foundation forconcept-level explainabilityof LLMs.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2606\.05972

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2606.05972 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2606.05972 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2606.05972 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

LLMs are not the black box you were promised

Hacker News Top

An article summarizing Anthropic's 2025 paper on mechanistic interpretability, showing that LLMs are not black boxes and that circuit tracing can reveal multi-step reasoning and human-identifiable concepts.