← 返回列表

@dwarkesh_sp: Btw a bunch of the questions were just off the cuff - nothing @reinerpope prepped for. The guy is just first principles deriving how many t...

@dwarkesh_sp 3 信息等级 3 1 噪音/剔除;2 较弱;3 普通事实;4 重要行业动态;5 极重大事件。该分数是信息显著性,不是投资建议。 发布:2026-04-30T03:14 抓取:2026-05-03 15:25
🔗 原文链接
摘要

推特消息称,有人通过第一性原理推导出GPT-5预训练的token数量、Gemini 3的KV缓存字节数以及Claude缓存命中的内存类型。

客观事实
  • 推导了GPT-5预训练的token数量
  • 推导了Gemini 3的KV缓存字节数
  • 推导了Claude缓存命中的内存类型
GPT-5 Gemini 3 Claude

原文

Btw a bunch of the questions were just off the cuff - nothing @reinerpope prepped for.

The guy is just first principles deriving how many tokens GPT 5 was pretrained on, or the bytes per token in Gemini 3's KV cache, or which kind of memory each Claude cache hit sits on.

likes: 838 | retweets: 23 | replies: 19 | views: 72180