Stellar Colosseum model for long-horizon math research
This is a new model-agnostic system for long-horizon research in math and computer science, achieving 71.0% accuracy on TCS-Bench and solving 218 of 222 Codeforces problems.
This is a new model-agnostic system for long-horizon research in math and computer science, achieving 71.0% accuracy on TCS-Bench and solving 218 of 222 Codeforces problems.