الملخص
تمت ترجمة أجزاء من هذه الصفحة آلياً وقد تحتوي على أخطاء.
Agent benchmarks are substantially more costly to evaluate than conventional LLM benchmarks. Benchmark compression is therefore a natural solution, yet existing methods primarily model redundancy in task--model final-score distributions, which is important in agentic evaluation. To address this limitation, we analyze large-scale trajectories and identify six complementary process signals that are systematically associated with final agent performance. To disentangle agent performance redundancy from a complete perspective, we propose DualViewEval, an agent benchmark compression method that jointly exploits outcome and process relations to learn an exact-size miniset and predict the full-benchmark scores. Across five agent benchmarks and five representative baselines, DualViewEval achieves the best results in all datasets. With only 20 tasks, it achieves $24\times$--$40\times$ compression on APEX-Agents and BFCL, reducing mean absolute error (MAE) by $14.5\%$--$28.2\%$ over the strongest competitors while improving Kendall's $τ$ by up to $7.2\%$ relative to EssenceBench on SWE-bench Verified. The selected minisets further reveal capability differences among different agents, providing compact and diagnostic feedback for efficient agentic model development.
الكلمات المفتاحية
الموضوع
بيانات النشر
- المجلة
- غير متاح
- وصول مفتوح
- وصول مفتوح أخضر
اقتبس هذه المقالة
APA 7
Guo, X., Wu, J., Deng, D., Li, Y., Zheng, H., Zheng, S., & Pan, M. (2026). ما وراء النتائج: التعلم العلائقي ثنائي المنظور لقياس الوكلاء بكفاءة. https://omanscience.com/ar/articles/beyond-outcomes-dual-view-relational-learning-for-efficient-agent-benchmarking
MLA 9
Guo, Xinshuai, et al. "ما وراء النتائج: التعلم العلائقي ثنائي المنظور لقياس الوكلاء بكفاءة." https://omanscience.com/ar/articles/beyond-outcomes-dual-view-relational-learning-for-efficient-agent-benchmarking.
شيكاغو (المؤلف–التاريخ)
Guo, Xinshuai, Junjie Wu, Dolly Deng, Yinghui Li, Haitao Zheng, Suncong Zheng, and Maxm Pan. 2026. "ما وراء النتائج: التعلم العلائقي ثنائي المنظور لقياس الوكلاء بكفاءة." https://omanscience.com/ar/articles/beyond-outcomes-dual-view-relational-learning-for-efficient-agent-benchmarking.
هارفارد
Guo, X., Wu, J., Deng, D., Li, Y., Zheng, H., Zheng, S. and Pan, M. (2026) 'ما وراء النتائج: التعلم العلائقي ثنائي المنظور لقياس الوكلاء بكفاءة', Available at: https://omanscience.com/ar/articles/beyond-outcomes-dual-view-relational-learning-for-efficient-agent-benchmarking.
فانكوفر
Guo X, Wu J, Deng D, Li Y, Zheng H, Zheng S, et al. ما وراء النتائج: التعلم العلائقي ثنائي المنظور لقياس الوكلاء بكفاءة. https://omanscience.com/ar/articles/beyond-outcomes-dual-view-relational-learning-for-efficient-agent-benchmarking
IEEE
X. Guo, J. Wu, D. Deng, Y. Li, H. Zheng, S. Zheng, and M. Pan, "ما وراء النتائج: التعلم العلائقي ثنائي المنظور لقياس الوكلاء بكفاءة," https://omanscience.com/ar/articles/beyond-outcomes-dual-view-relational-learning-for-efficient-agent-benchmarking.