Event ·Multi-source

Sonnet 5 对 Opus 4.8 进行对齐训练

Anthropic 让 Sonnet 5 对 Opus 4.8 的早期检查点进行后训练。经过训练后，该模型的安全分数接近完整对齐训练的 Opus 4.8 水平。这是测试模型能否对其更强大的继任者的首次尝试。

Current read

Anthropic 让 Sonnet 5 训练 Opus 4.8，安全分数接近原版，模型自我对齐有新进展。

6 sources58° AI heatLast updated 2026/8/28 17:13:37

Evidence

6 sources

Primary sourceAnthropic

2026/8/28 17:13:37View source

Related source 1IT之家

2026/8/29 13:25:27View source

Related source 2Harrison Chase

2026/8/27 21:09:33View source

Related source 3Simon Willison’s Weblog

2026/8/27 22:50:25View source

Related source 4techcrunch

2026/8/28 19:30:38View source

Related source 5宝玉

2026/8/29 04:12:12View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story