Dual-ViewEval method for efficient agent benchmarking
This research provides a more efficient way to evaluate agent models, which is crucial for developing and comparing them. The method's ability to achieve significant compression while maintaining accuracy is a notable improvement in the field.