SEA-Eval: Benchmark for Evaluating Self-Evolving AI Agents
Discover SEA-Eval, a benchmark designed to evaluate self-evolving AI agents beyond episodic tasks, focusing on long-term performance and reliability.
PilotBench: Benchmarking AI Safety in General Aviation
Discover PilotBench, a benchmark evaluating AI models on safety and precision in general aviation flight predictions with real-world data.
Boost LLM Problem Solving with Tutor-Student Agents
Discover how tutor-student multi-agent interaction enhances Large Language Model problem solving with efficient, role-based collaboration.
StaRPO: Enhancing RL with Stability for Better Reasoning
StaRPO improves reinforcement learning by adding stability metrics, boosting logical consistency and accuracy in AI reasoning tasks.
SPPO: Efficient PPO for Long-Horizon Reasoning Tasks
Discover SPPO, a scalable Sequence-Level PPO algorithm improving stability and efficiency in long-horizon reasoning for Large Language Models.