enFoldX gains accuracy as AF3 ensemble disagreement rises for non-binders
quaidmorris · x · 2026-07-22
enFoldX uses ensemble disagreement as a signal
The post says enFoldX summarizes an AlphaFold3 ensemble instead of trusting one prediction. Non-cognate pairs show larger structural disagreement — higher RMSD and higher iPAE — and performance improves as more ensemble members are added.
That behavior differs from single-structure models, which do not benefit as much from additional samples.
More from Research
- Open-source runtime lets each repo define its own AI code reviewer — ibabufrik · 2026-07-22
- OpenAI and Apollo Research introduce Contrastive SDF to measure reward-seeking — OpenAI · 2026-07-22
- NVIDIA says to tune the harness before tuning the model with LangChain — NVIDIAAI · 2026-07-22
- The Thimble and the Waterfall: AI's Data Bottleneck and Feedback Loops — dyamins · 2026-07-22
- NVIDIA shows 22 SIGGRAPH papers and Omniverse tools for robot simulation — facontidavide · 2026-07-22
- Building a Knowledge Graph Without a Graph DB: 1000x Cheaper Than GraphRAG — TheRedfather · 2026-07-22