Stanford's Evo 2 genome model learns in-context, designs proteins and phages from few examples
anshulkundaje · x · 2026-10-31
Brian Hie's Stanford team released ICL preprints for the Evo 2 genome language model on bioRxiv. Genome models can now learn tasks from a handful of prompt examples without retraining, generating proteins, de novo antibodies, CRISPR-Cas9 loci, mitochondrial genomes, and entire bacteriophages. Synthetic sequences can be chained iteratively into a chain-of-thought for harder design goals. Evo2-flash is also now the default Evo 2 backend on Proto, including the web app and MCP.
More from Research
- Karpathy's nanochat hits 59K stars: a full LLM training lab in one repo — techNmak · 2026-11-01
- Same 2.4 eV band gap, different routes: SHAP explains per-prediction feature contributions — bravo_abad · 2026-11-01
- Hands-on tutorial trains a decision model on Qwen3-0.6B from scratch — dl_weekly · 2026-11-01
- Fine-grained MoE is the easiest win: DeepSeek-V3-style 128 experts cut loss by up to 6.2% — ivan_bezdomny · 2026-11-01
- Peng Ding's 'Linear Model and Extensions' hits print; free arXiv version includes R code — lihua_lei_stat · 2026-11-01
- SAM 3 Flags 25 Cells Experts Called Clean in Malaria Smears, Revealing Label Disputes — MaziyarPanahi · 2026-11-01