Benchmarks

• HPOBench - Containerized multi-fidelity HPO problems.
• YAHPO Gym - Fast surrogate HPO benchmarks.
• BoLT - Surrogate and tabular benchmarks for BO on expensive LLM tasks, without running real LLM training.
• HPO-B - Transfer-HPO benchmark with related tasks.
• LassoBench - High-dimensional weighted-Lasso HPO.
• COCO / BBOB - Noiseless and noisy black-box testbeds used beyond BO.
• Bayesmark - Scoring harness for comparing BO libraries on ML models.