Focus on Single Solutions in Many-Objective Bayesian Optimization
Explore why targeting single high-quality solutions outperforms full Pareto front search in many-objective Bayesian optimization under limited budgets.
HiL-Bench: Evaluating AI Agents’ Help-Seeking Judgment
Discover HiL-Bench, a benchmark measuring AI agents' ability to know when to ask for help in uncertain tasks, improving decision-making and performance.
Spatial-Gym: Stepwise Evaluation of Spatial Reasoning Agents
Discover how Spatial-Gym benchmarks spatial reasoning in AI agents step-by-step, revealing key insights to improve navigation and decision-making models.
Constraint-Aware Memory Boosts Language-Based Drug Discovery
Discover how Constraint-Aware Corrective Memory (CACM) improves language-based drug discovery with precise diagnostics and 36% higher success rates.
SAGE Benchmark: Advanced Evaluation for Service Agents
Discover SAGE, a dynamic benchmark for evaluating LLMs in customer service using graph-guided SOPs and adversarial intent analysis.