AGIEval-en

AGIEval-en is a benchmark used to evaluate the complex reasoning abilities of large language models (LLMs) through reasoning-intensive tasks.

Papers using AGIEval-en (2)

AttentionInfluence: Adopting Attention Head Influence for Weak-to-Strong Pretraining Data Selection2025AttentionInfluence: Adopting Attention Head Influence for Weak-to-Strong Pretraining Data Selection2025