arXiv:2610.28819v1 Announce Type: new Abstract: Common instruction following (IF) benchmarks, such as IFEval, evaluate the ability of language models (LMs) to follow simple constraints that can easily be verified programmatically. These same constraints, in turn, are used as reward signals for reinforcement learning. This design may lead to reward hacking of constraints, pushing models away from the main task. Recent attempts to address this resulted in benchmarks with richer constraints, but required expert labor, remained closed-access, and cannot be easily extended. We present Lift-Bench, a
Scoped constraints offer more precise instruction-following evaluation/training, helping teams measure real capability gains and reduce prompt-hacking and spec ambiguity.
Published to Cognify News · Week 43, 2026