AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Do LLMs Really Struggle at NL-FOL Translation? Revealing Their Strengths via a Novel Benchmarking Strategy

Andrea Brunello, Luca Geatti, Michele Mignani, Angelo Montanari, Nicola Saccomanno

AAAI 2026 · Large language models › Language, safety and interpretability · approach: LLM unlearning and benchmarks · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add task-specific instruction to the embedding input (ablation)Qwen3-Embedding-8B · D_FOLIO FOL most similaraccuracy: .44 -> .76Improved
add task-specific instruction to the embedding input (ablation)Qwen3-Embedding-8B · D_FOLIO FOL ranking-bothscore: .13 -> .37Improved
use o3-mini instead of GPT-4o-minilogical translation · D_Stanfordequivalence accuracy: .84 -> .94Improved
use Qwen3-30B instead of Qwen3-8Blogical translation · D_FOLIOequivalence accuracy: .72 -> .74No clear effect
enable thinking mode (ablation)Qwen3-30B · Ranking-Equivalence and Ranking-Bothaccuracy: up to a 6 point increment in accuracyImproved
enable thinking mode (ablation)Qwen3-8B · Ranking-Equivalence and Ranking-Bothaccuracy: performance differences are marginalNo clear effect
enable thinking mode (ablation)Qwen3-8B and Qwen3-30B · Logical Translationaccuracy: identical average accuracyNo clear effect
use FOL formulas instead of NL renderings as candidatesGPT-4o-mini · D_FOLIO ranking-equivalencescore: .52 -> .64Improved
use FOL formulas instead of NL renderings as candidatesQwen-Emb-plain · D_Stanford most similaraccuracy: .62 -> .46Hurt
use ranking-equivalence task instead of most similarGPT-4o-mini · D_FOLIO FOLsuccess rate: fails ranking-equivalence in 31% while opposite occurs in 4%Hurt
evaluate on D_Stanford instead of D_FOLIOdialogue models · logical translationaccuracy: .11--.14 differenceImproved
use dialogue-oriented o3-mini instead of Qwen-Emb-plainranking task · D_Stanford FOL ranking-bothscore: .89 vs .09Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy