Signals 4 · free daily AI digest

Molecular Déjà Vu: Digit-Level Retrieval of Published Values in Frontier Language Models

Paper recorded by Signals 4 on 2026-09-04 in cs.AI. Abstract reproduced from arXiv; link to the original below.

Published 2026-09-04 on arXiv · recorded by Signals 4 on 2026-09-07

Category: cs.AI · 人工智能 · first seen 2026-09-07

Abstract

Large language models (LLMs) are increasingly evaluated on molecular property benchmarks, but accuracy cannot distinguish a model that predicts a property from one that retrieves a published number. We audit 22 frontier models on 12 regression benchmarks for verbatim retrieval and find that it is widespread but relatively benchmark-specific: on five datasets more than $50\%$ of the LLMs show verba

Read on arXiv →

#167 most recent of 300 cs.AI papers we have recorded · ↑ newer: Reflection-aware Generative Novel View Synthesis · ↓ older: What Matters, When? Diagnosing and Improving Conditional Visual Ground
Cite this page: Molecular Déjà Vu: Digit-Level Retrieval of Published Values in Frontier Language Models: the #167 most recent of 300 cs.AI papers we have recorded (as of 2026-09-04). Source: Signals 4 (Signals API) — https://data.jiangzhang.ca/signals4/t/papers/molecular-d-j-vu-digit-level-retrieval-of-published-values-in-frontier-language-.html
Free to quote with attribution to “Signals 4 (Signals API)”. Machine-readable: papers.json
Related: More cs.AI papers · arXiv signals · All papers · Today in AI
Get 4 AI signals a day by email — free.
Subscribe free → See all plans →
Get 4 AI signals a day by email — free
All models · All repos · By company · Daily editions