13 Oct 2025
Researchers at Wuhan University, Hong Kong Polytechnic University, and Stanford University developed the "Protein-as-Second-Language" framework, enabling general large language models (LLMs) to interpret and reason about protein sequences in a zero-shot setting. This approach, which involves dynamically generating bilingual context from a curated 79,926 QA dataset, achieved an average 7% ROUGE-L improvement and consistently outperformed fine-tuned protein language models on various protein understanding tasks.
There are no more papers matching your filters at the moment.
