paper-with-me

홈 › Papers

Causal Tongue-Tie: LLMs Can Encode Causal Direction, But Their Yes/No Outputs Fail to Express

2026-05-25 · Ziyi Ding, Xiao-Ping Zhang arxiv

We find a mismatch between what large language models encode about a causal question and what they answer. On anti-commonsense CLadder items, a fixed linear probe recovers the evidence-supported answer from the model's hidden state (accuracy approximately 0.97), while the spoken Yes/No reverts to the commonsense one (accuracy approximately 0.5). We call this approximately +0.5 gap Causal Tongue-Tie: a wrong Yes/No decomposes into two separable failure modes: no internal signal versus a signal the verbal interface cannot say. The implication cuts both ways for output-only causal benchmarks: a benchmark "correct" need not mean the model has understood, and a benchmark "wrong" need not mean it cannot. Sweeping claims about whether LLMs can do causal reasoning, drawn from a single accuracy number, deserve a second look.

📄 PDF Abstract BibTeX arXiv:2605.25891

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

BidirLM: From Text to Omnimodal Bidirectional Encoders by Adapting and Composing Causal LLMs

2026-04-02 · Nicolas Boizard, Théo Deschamps-Berger, Hippolyte Gisserot-Boukhlef, Céline Hudelot 외 arxiv

Transforming causal generative language models into bidirectional encoders offers a powerful alternative to BERT-style architectures. However, current approaches remain limited: they lack consensus on optimal training ob…

Ideological Bias in LLMs' Economic Causal Reasoning

2026-04-23 · Donggyu Lee, Hyeok Yun, Jungwon Kim, Junsik Min 외 arxiv

Do large language models (LLMs) exhibit systematic ideological bias when reasoning about economic causal effects? As LLMs are increasingly used in policy analysis and economic reporting, where directionally correct causa…

Understanding Causality with Large Language Models: Feasibility and Opportunities

2023-04-11 · Cheng Zhang, Stefan Bauer, Paul Bennett, Jiangfeng Gao 외

We assess the ability of large language models (LLMs) to answer causal questions by analyzing their strengths and weaknesses against three types of causal question. We believe that current LLMs can answer causal question…

Decision Making

Causal2Vec: Improving Decoder-only LLMs as Embedding Models through a Contextual Token

2025-07-31 · Ailiang Lin, Zhuoyun Li, Yusong Wang, Kotaro Funakoshi 외 arxiv

Decoder-only large language models (LLMs) have been increasingly adopted to build embedding models for diverse tasks. To overcome the inherent limitations of causal attention in representation learning, many existing met…

Representation Learning

CausalBench: A Comprehensive Benchmark for Causal Learning Capability of LLMs

2024-04-09 · Yu Zhou, Xingyu Wu, Beicheng Huang, Jibin Wu 외

The ability to understand causality significantly impacts the competence of large language models (LLMs) in output explanation and counterfactual reasoning, as causality reveals the underlying data distribution. However,…

counterfactualCounterfactual ReasoningReading Comprehension