Always in the middle of something.

Where the watermark in Claude's output can sit, and what a check on it would report. From Anthropic's 14 August announcement and the SynthID-Text paper behind it.

Anthropic 說明了未來 Claude 的文字浮水印怎麼做：做法是換一個亂數來源，用來決定下一個字挑哪一個，不加字也不多花 token。這篇是來聊聊我們目前獲得的資訊有哪些。

Google Search will phone a local business on a searcher's behalf. The Business Profile help page says what makes it dial, what stops it, where it is unavailable, and what your answer becomes afterwards.

在美國用 Google 搜尋查附近哪裡有貨，結果頁會多出一個選項，按下去後 Google 會打電話去問附近幾家店，再把摘要寄給你。這篇照著官方說明描述一次流程是怎麼進行的，也會順便看看店家那一邊的規則。

The B in a name like gpt-oss-20b counts parameters, in billions. Multiply that count by the bytes each parameter takes and you get a floor for the memory the weights need. That is also why some models now ship with two numbers instead of one.

模型的 7B、70B、235B 講的是參數量，B 就是十億。這篇說明參數量代表什麼、怎麼換算，以及 Qwen3-235B-A22B 這種寫法裡的兩個數字又各自代表什麼。

Fan three or four AI reviewers at the same diff and what comes back is a list of candidates. Agentic OS's own audit records show how many survive checking, and why its default has dispatched subagents return evidence while one primary owns the write.

開一個 agent 去看另一個 agent 做完的東西很容易，難的是後面那個看得見前面漏掉了什麼。從 Agentic OS 的 review 規定看下去：要讓第二個 agent 看得出東西，靠的是刻意不給它 session、對話記錄和實作理由。

A transformer redoes the same attention projections for every past token at each decoding step. The KV cache stores those keys and values so they get reused instead of recomputed, and the one cost it adds is memory that grows with the sequence.

把 softmax 加交叉熵對 logit 的導數一路算出來，結果剛好是預測機率減去標籤。這篇從一個三類別的小例子走進這個梯度，看它為什麼乾淨、又為什麼信心錯得越離譜就修得越用力。