Claude Sonnet 5.5 · Protein Design Library Ranking(Prioritization of protein designs for experimental testing) 54.8% · 附件证据(ev55-protein-ranking)

Claude Sonnet 5.5 的 Protein Design Library Ranking · Prioritization of protein designs for experimental testing 官方出处。Adaptive max with biology safeguards disabled · 54.8%。见《Claude Sonnet 5.5 System Card》§8.17.5, PDF p.144。证据 ev55-protein-ranking。

原文与定位

Claude Sonnet 5.5 · Protein Design Library Ranking(Prioritization of protein designs for experimental testing),Adaptive max with biology safeguards disabled配置,Score 54.8%(具体成绩公布时间未确认)。见《Claude Sonnet 5.5 System Card》§8.17.5, PDF p.144。 查看成绩条件

We ran only 1 attempt per problem for Sequence Generation and 3 attempts per problem for Library Ranking. Claude Sonnet 5.5 shows large improvements over Claude Sonnet 5 on both (Sequence Generation 51.0% vs. 20.2%; Library Ranking 54.8% vs. 44.0%).
官方材料
Claude Sonnet 5.5 System Card · PDF 第 144 页
定位
附件 · §8.17.5, PDF p.144
材料版本
PDF cover dated 2026-09-28; 148 pages; collected 2026-11-10
版本公布
未确认
披露时间
未确认 · 首发/后补未确认
采集时间
历史依据
暂无可靠历史版本依据
引用文本

Claude Sonnet 5.5 · Protein Design Library Ranking(Prioritization of protein designs for experimental testing),Adaptive max with biology safeguards disabled配置,Score 54.8%(具体成绩公布时间未确认)。官方出处 Claude Sonnet 5.5 System Card(附件,PDF cover dated 2026-09-28; 148 pages; collected 2026-11-10)§8.17.5, PDF p.144,https://www.anthropic.com/claude-sonnet-5-5-system-card#page=144。Benchlog https://benchlog.lekshome.cc/evidence/claude-sonnet-5-5/ev55-protein-ranking/

关联成绩与条件

Claude Sonnet 5.5 · Protein Design Library Ranking 成绩与配置
指标与成绩配置与评测条件证据关系
Score54.8%原文 54.8% · 单位 %
公布 未确认
Adaptive max with biology safeguards disabled
设置
Adaptive thinking at max effort; API evaluation with biology safeguards turned off.
未知项
Sampling settings follow each benchmark's report.
工具
工具条件未记录
推理预算
max
框架
未记录
协议
Protein Design Library Ranking; three attempts per problem.
条件
API evaluation with biology safeguards disabled; adaptive thinking and max effort.

直接支持该成绩