CoCoA: Unsupervised Code Evaluation with LLMs
Discover CoCoA, a novel framework using LLMs for unsupervised code evaluation by separating comprehension and auditing for higher accuracy.
LAPIS-SHRED: Efficient Spatio-Temporal Dynamics Reconstruction
LAPIS-SHRED reconstructs full spatio-temporal dynamics from sparse data using shallow recurrent decoders for accurate forecasting and analysis.
AI Weather Forecasting: Key Mathematical Foundations
Discover how AI weather prediction improves by focusing on learning pipelines, loss functions, and data diversity over architecture alone.
YC-Bench: AI Benchmark for Long-Term Planning Success
YC-Bench evaluates AI agents on long-term planning and execution in complex simulations, highlighting strengths and failure modes of top models.
CliffSearch: AI-Driven Scientific Algorithm Discovery Framework
Discover CliffSearch, an AI-powered framework combining theory and code for structured, accurate, and novel scientific algorithm discovery.