works on
benchmark creation 1cross-lingual evaluation 1cultural alignment 1large language models 1multilingual question answering 1

From the 1 of 3 linked papers with an AI index.

collaborators
Showing cs.AIShow all

1 paper · 1 filter

cs.AI2026

Harness-IF: Evaluating Instruction Following Across Instruction Surfaces in Coding Agents

Zining Huang, Haoran Que, Hong Zeng +8

When a coding agent obeys a rule, it may simply have been going to do that anyway. Existing instruction-following benchmarks cannot tell the difference: they concentrate rules in t…