arXiv AI

CausalVerify: An Execution-Grounded Benchmark for LLM Causal Inference Workflows

arXiv AI
21h ago

Are Stated Reasoning Steps Causally Load-Bearing?

The study investigates whether the reasoning steps a language model writes are causally responsible for its answers. Using a causal intervention method on the activation stream, the authors find that for Qwen3-4B, about 77% of stated steps are causally load‑bearing, while behavioral tests overestimate this by roughly 11 percentage points. The faithfulness of reasoning decreases with model size and depth of reasoning, especially for the smaller Qwen3-1.7B.

By Abhiram Bhupatiraju, Rayan Nyaupane
arXiv Computation and Language
Sep 14

Can LLMs in Draft-Verify-Revise Pipelines Resolve Deictic Ambiguity?

The study investigates how deictic ambiguity—specifically the shifting reference of expressions like "previous"—affects Draft‑Verify‑Revise pipelines that use multiple large language models (LLMs). Using a synthetic dataset of 10 base examples and 21 reasoning‑effort configurations, six LLMs were evaluated for their ability to correctly resolve the ambiguous expression across the draft, verify, and revise stages. Results show wide variance in balanced accuracy, with GPT‑5.2 improving from 0.156 to 0.942 with increased reasoning effort, while Gemini 3 Pro consistently achieved high accuracy above 0.94 even at low reasoning effort, and meta‑evaluators often relied on surface cues when making errors.

By Obinna I. Ekekezie
arXiv AI
Aug 11

Reproducing and Stress-Testing Two Approaches to LLM Reasoning Reliability: Test-Time Probability Aggregation and Logic-Representation Editing

arXiv:2608. 08514v1 Announce Type: new Abstract: We independently reproduce two recent methods for making large language model (LLM) reasoning more reliable, and stress-test them across domains and models (RPC across four new task domains with Qwen3-8B, LCF across four 7-8B models).

By Minhan Cho, Jimin Kweon
arXiv AI
Aug 21

Credit Without Ground Truth: Auditing Step-Level Credit Assignment in LLM Agents Against Executed Replay

arXiv:2608. 19760v1 Announce Type: cross Abstract: Audited against causal ground truth from executed replay in a single-agent tool environment (ALFWorld), none of the step-level credit signals used to train LLM agents -- LLM-judge scores, outcome-conditioned logprob ratios, or the policy's own confidence -- identifies which steps causally matter better than chance.

By Haiyue Zhang
arXiv Machine Learning
2d ago

Counterfactual Tool Ranking under Utility, Cost, and Privilege Constraints

The paper introduces a counterfactual tool ranking framework that accounts for authority, historical support, and estimation nuances. Using eleven enterprise-inspired tools, synthetic and real-world experiments on the Berkeley Function Calling Leaderboard, the study compares direct regression and doubly robust (DR) methods, finding that DR performs better in shifted environments while direct regression excels in linear settings. The authors also evaluate Qwen2.5 models on held-out tasks, analyze policy differences under missing support, and present a falsifiable evaluation method with publicly available evidence.

By Jiapeng Li