Reproduce “Eterna is Solved” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Tristan Cazenave
A sequence optimization algorithm called MOGNRPALR (Multi-Objective Generalized Nested Rollout Policy Adaptation with Limited Repetition) successfully solves all 100 problems in the Eterna RNA design benchmark through parallel execution of 200 runs, surpassing previous approaches by combining nested Monte Carlo search with multi-objective evaluation criteria.
Key insights
- Using biases in MOGNRPALR improves performance on difficult problems
- Parameter tuning through profile parameter search enhances algorithm effectiveness
- Multiple parallel runs help overcome local optima issues
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.