← back
arXivMalia Barker, Bishal Lakha, Edoardo Serra, Francesco GulloTue, Jun 2, 2026, 6:09 AM PDT
score 17.1

LLMs fail simple math when numbers change slightly

Original: Testing LLM Arithmetic Reasoning Generalization with Automatic Numeric-Remapping Attacks

Source: arxiv.org

Writing ELI5 summary…