neuron-explorer
Steering-first neuron explorer for Qwen3.5-4B: push every MLP neuron, keep every rollout, browse the results
翻譯準備中
尚未偵測到更新。GitHub 發布版本後會顯示在這裡。
個人 GitHub 帳號下公開儲存庫的擁有者可以登入並驗證所有權,自行管理介紹,無需管理員核准。
登入並產生驗證檔案,在 30 分鐘內將其提交至此儲存庫的預設分支。其他分支或分叉儲存庫中的檔案無效。然後在下方檢查檔案,即可無需管理員核准直接管理專案。請保留檔案以維持管理權限。
如需登入,將前往登入頁面。登入後請回到此處,再次點擊所選的驗證按鈕。
在專案資料夾中執行 /nomorevibe verify,即可驗證所有權並自行管理介紹。
一分鐘完成驗證估算單次多步驟 AI 代理執行的實際成本。模型化二次方上下文累積、工具結果大小、重試次數與提示快取,涵蓋 17 種模型。
AI 翻譯
Estimates what one multi-step AI agent run actually costs. Models quadratic context accumulation, tool-result size, retries and prompt caching across 17 models.
取自儲存庫 README
正在載入完整 README。 以下為已儲存的摘錄。
A calculator that estimates what one multi-step AI agent run actually costs.
Every other LLM cost calculator prices a single API request: input tokens times rate, plus output tokens times rate. That is correct for a chatbot.
An agent is not a chatbot. It makes dozens of calls in a loop, and because language models hold no state between calls, every step re-sends the entire conversation so far. Step 10 pays for steps 1 through 9 all over again.
The second term is quadratic. Double the step count and the input cost roughly quadruples.檢視完整 README ↗目前無法載入完整 README。請透過原文連結閱讀。 以下為已儲存的摘錄。
A calculator that estimates what one multi-step AI agent run actually costs.
Every other LLM cost calculator prices a single API request: input tokens times rate, plus output tokens times rate. That is correct for a chatbot.
An agent is not a chatbot. It makes dozens of calls in a loop, and because language models hold no state between calls, every step re-sends the entire conversation so far. Step 10 pays for steps 1 through 9 all over again.
The second term is quadratic. Double the step count and the input cost roughly quadruples.檢視完整 README ↗