Bolting a text embedder onto protein embeddings enables natural-language protein search
johnowhitaker · x · 2026-09-06
johnowhitaker bolted a text embedding model onto an existing protein+reaction embedding model, making it possible to search for proteins or chemical reactions with plain natural language. He finds it hilarious that this works—a counterintuitive cross-modal embedding alignment trick with real AI4Science utility.
More from Research
- One prompt is enough: distillation from a single query hits 71.5% of full-data gains — burkov · 2026-09-06
- 96-plasmid yeast transformation in 3.25 hours with open-source Pylabrobot — nlarusstone · 2026-09-06
- BioTorch launches in beta: 116 PyTorch exercises to onboard researchers into biological AI — yawnxyz · 2026-09-06
- eXCV workshop at ECCV 2026 to tackle explainability in the foundation-model era — CSProfKGD · 2026-09-06
- ETH Zurich tested 100 developers: written communication predicts vibe coding success — _akpiper · 2026-09-06
- ChatGPT's new model now rivals Claude for math animations, demo shows — PTenigma · 2026-09-06