Adversarial Training
HLE Adversarial Training
Expert-crafted adversarial prompts for critical reasoning dimensions.
Read case studyCode Intelligence
Code Refactoring Benchmark
Mining baselines and deep verification for code quality benchmarks.
Read case studySimulation
RL Gym Simulation
Multi-agent environments for reinforcement learning from human feedback.
Read case studyHallucination Control
Hallucination Optimization Training
A hallucination taxonomy and expert annotation pipeline.
Read case studyAgent Evaluation
Report Generation Benchmark
Seven quality dimensions for evaluating agent-generated reports.
Read case studyAgent Evaluation
Search Benchmark
A tiered framework for accuracy, coverage, and depth in search.
Read case studyHave a training or evaluation challenge?
Discuss your project