Beyond Outcomes: Dual-View Relational Learning for Efficient Agent Benchmarking
DualViewEval compresses agent benchmarks by jointly modeling outcome and process signals, achieving 24x-40x compression with only 20 tasks on APEX-Agents and BFCL.
DualViewEval is an agent benchmark compression method that jointly exploits outcome and process relations from trajectories to learn exact-size minisets predicting full-benchmark scores. The authors analyze large-scale trajectories and identify six process signals systematically associated with final agent performance. Across five agent benchmarks and five baselines, it achieves the best results on all datasets: with only 20 tasks it reaches 24x-40x compression on APEX-Agents and BFCL, reduces MAE by 14.5%-28.2% over the strongest competitors, and improves Kendall's tau by up to 7.2% relative to EssenceBench on SWE-bench Verified.
- Identifies six trajectory process signals correlated with final agent performance
- Learns exact-size minisets predicting full-benchmark scores from outcome and process relations
- 20-task minisets yield 24x-40x compression on APEX-Agents and BFCL
- MAE reduced 14.5%-28.2% versus strongest competitors
Full article164 words · extracted from arxiv.org · click to collapse
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.
Text extracted automatically; images, tables and formatting may be missing. Original: https://arxiv.org/abs/2609.18909