🆕Biohub’s Protein World Model: ESMC-6B, ESMFold2, 6.8B proteins, 1.1B structures, antibody design, ...
TL;DR · AI 摘要
Biohub的Protein World Model通过ESMC-6B和ESMFold2处理68亿蛋白质和11亿结构，展示了生物建模可能像语言建模一样扩展，强调稀疏自编码器揭示模型内部生物学。
核心要点
- Biohub的ESMC-6B和ESMFold2处理68亿蛋白质和11亿结构。
- ESMFold2在抗体-抗原预测上击败了专用系统。
- Biohub的5亿美元虚拟生物学计划旨在构建细胞、疾病和生理学的预测模型。
结构提纲
按章节快速跳转。
思维导图
用一张图看清主题之间的关系。
查看大纲文本（无障碍 / 无 JS 友好）
- Biohub的Protein World Model
金句 / Highlights
值得收藏与分享的关键句。
ESMFold2在抗体-抗原预测上击败了专用系统。
Biohub的5亿美元虚拟生物学计划旨在构建细胞、疾病和生理学的预测模型。
稀疏自编码器揭示了模型内部的生物学信息。
@biohub Head of Science @alexrives explains why biology may scale like language modeling, how metagenomics unlocked" / X
Latent.Space on X: "🆕Biohub’s Protein World Model: ESMC-6B, ESMFold2, 6.8B proteins, 1.1B structures, antibody design, SAEs, & the bitter lesson for biology https://t.co/2PGoHjttCS @biohub Head of Science @alexrives explains why biology may scale like language modeling, how metagenomics unlocked" / X
Don’t miss what’s happening
Biohub’s Protein World Model: ESMC-6B, ESMFold2, 6.8B proteins, 1.1B structures, antibody design, SAEs, & the bitter lesson for biology https://latent.space/p/esmfold2
Head of Science
explains why biology may scale like language modeling, how metagenomics unlocked the next ESM scaling curve, why protein LMs can learn structure/function from sequence alone, how sparse autoencoders reveal biology inside the model, why ESMFold2 can beat specialized systems on antibody-antigen prediction, and how Biohub’s $500M Virtual Biology Initiative aims to build predictive models of cells, disease, and eventually physiology.
[](https://t.co/2PGoHjttCS)
latent.space 
·
3
2
4