neuron-explorer
Steering-first neuron explorer for Qwen3.5-4B: push every MLP neuron, keep every rollout, browse the results
翻译准备中
可读的 PyTorch 模型架构图。追踪器提供事实,代理提供抽象,覆盖检查证明没有任何内容被悄悄遗漏。
AI 翻译
Readable architecture diagrams for PyTorch models. The tracer supplies the facts, an agent supplies the abstraction, and a coverage check proves nothing was silently dropped.
开发工具Python · HTMLGitHub @syncytium2在线运营者未确认
试用过了吗? 点一下就好,无需登录。
个人 GitHub 账号下公开仓库的所有者可以登录并验证所有权,自行管理介绍,无需管理员批准。
登录并生成验证文件,在 30 分钟内将其提交到此仓库的默认分支。其他分支或分叉仓库中的文件无效。然后在下方检查文件,即可无需管理员批准直接管理项目。请保留文件以维持管理权限。
如需登录,将前往登录页面。登录后请回到此处,再次点击所选的验证按钮。
在项目文件夹中运行 /nomorevibe verify,即可验证所有权并自行管理介绍。
一分钟完成验证正在加载完整 README。 以下为已保存的摘录。
Readable architecture diagrams for PyTorch models. The tracer supplies the
facts, a person or an agent supplies the abstraction, and a coverage check proves
nothing was silently dropped.
Status: the three stages work end to end on the model that prompted them.
SPEC.md carries the design, the measurements behind it, and the
failure it exists to prevent. DECISIONS.md answers what the
spec left open and records three places building found it mistaken — read that
second, and before changing the trace layer.
Whisper tiny, drawn from the model card查看完整 README ↗目前无法加载完整 README。请通过原文链接阅读。 以下为已保存的摘录。
Readable architecture diagrams for PyTorch models. The tracer supplies the
facts, a person or an agent supplies the abstraction, and a coverage check proves
nothing was silently dropped.
Status: the three stages work end to end on the model that prompted them.
SPEC.md carries the design, the measurements behind it, and the
failure it exists to prevent. DECISIONS.md answers what the
spec left open and records three places building found it mistaken — read that
second, and before changing the trace layer.
Whisper tiny, drawn from the model card查看完整 README ↗