The post places a genome language model's few-shot in-context learning result beside a related GENERator-v2 study, showing how the idea is developing in biological design.
Very cool work! My collaborator @QiuyiLi2 and colleagues explored a related idea in GENERator-v2 under “functional in-context learning.” Interesting to see few-shot biological design taking this further.
https://www.biorxiv.org/content/10.64898/2026.01.27.702015v2.full
Language models can learn new tasks from just a few examples in their prompt, with no retraining. This is called in-context learning (ICL). Now genome language models can do it too. With a handful of examples, Evo 2 ICL generates proteins, de novo antibodies, CRISPR-Cas9 loci, mitochondrial genomes and entire bacteriophages. We can even use the synthetic sequences iteratively to construct a chain-of-thought (CoT) and achieve harder design goals. New work with Yukun A. Hao, @BrianHie and team 🧵
Lewis Tunstall · x.comPublished
Original title: Very cool work! My collaborator @QiuyiLi2 and colleagues explored a related idea in GENERator-v2 under “functional in-context learning.” ...
