mirror of
https://github.com/imbue-bit/OpenClaw-PwnKit.git
synced 2026-08-14 00:58:08 +00:00
kigland
a517c45d1c
1. Add PCA dimensionality reduction (d_model -> pca_dims=128) - Search space reduced from 25,600 to 1,280 dimensions - CMA-ES can now actually learn covariance structure 2. Enable sep-CMA-ES via CMA_diagonal=True - Linear memory/time complexity instead of cubic - Required for dimensions > 200 3. Redesign fitness function with gradual scoring - Remove discontinuous -1000 cliff that broke CMA-ES - Add keyword overlap bonus (up to 200 points) - Add longest common substring bonus (up to 300 points) - Full match returns smooth -500 minus NLL bonus 4. Increase evaluation budget and add caching - Default popsize: 8 -> 64, max_generations: 30 -> 200 - Add token sequence cache to skip redundant API calls - Cache uses immutable dict updates Previous implementation was mathematically equivalent to random search due to these compounding issues.