精选65Chubby♨️产品发布/更新
Anthropic澄清Claude Code评分映射问题非降级
Anthropic Thariq says Claude Code’s “high = 10” output was a numerical mapping i…
Anthropic Thariq says Claude Code’s “high = 10” output was a numerical mapping issue, not a stealth downgrade.
Anthropic 的 Thariq 表示，Claude Code 的“high = 10”输出是一个数值映射问题，而非隐蔽的性能降级。
“High” is apparently still high, and internal evals show no performance regression.
"High" 显然仍然很高，内部评估显示没有性能回归。
Thanks for the clarification. But I still have to say that most models currently feel significantly dumber to use. Especially Opus.
感谢澄清。但我仍得说，目前大多数模型用起来明显感觉变笨了。尤其是 Opus。
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力