2026年7月9日 技术热点总结

技术栈：Python

核心介绍：آموزش ویدیویی : دروازه (Gateway) سریع و مدرن برای تونل‌زنی VLESS روی WebSocket + HTTP Proxy، با داشبورد مدیریتی زیبا و قابلیت ساخت لینک‌های اختصاصی با محدودیت ترافیک. ابتدا روی دکمه Fork کلیک کنید تا این ریپازیتوری را به حساب خود منتقل کنید. 1. وارد سایت Railway.app شوید. 2. روی New Project → Deploy from GitHub repo کلیک کنید. 3. ریپازیتوری Fork شده را انتخاب کنید. 4. Railway به‌صورت خودکار پروژه را Deploy می‌کند.

项目数据：⭐ 2,446 Stars，🍴 4,751 Forks

项目简介：Chrome extension that filters pseudo-brand junk out of Amazon. Buy from real, established brands.

技术栈：JavaScript、amazon、browser-extension、chrome-extension

核心介绍：real, established brands, even when that means paying more. Amazon is flooded with trademark-squat “brands” (SZHLUX, HORUSDY, LATTOOK, DOZAWA…): random strings registered at the USPTO purely to unlock Amazon Brand Registry, selling commodity goods with no company, no warranty, and no reputation behind them. Knockoff detects those listings and hides, dims, or labels them, right in the search results.

项目数据：⭐ 1…

项目简介：The diary of Tom Riddle for the reMarkable Paper Pro — write with your pen, the page drinks your ink and answers in a flowing hand

技术栈：Rust

核心介绍：Write on the page with your pen. After a pause, the diary drinks your ink — your words fade into the paper — the page thinks for a moment, and an answer writes itself back in a flowing hand, stroke by stroke, then fades away. No screen glow, no keyboard, no chat UI. Just ink appearing on paper. _This is the diary from the demo._ You need a reMarkable Paper Pro in developer mode with a launcher installed. If that sounds li…

项目简介：Companion code for the global workspace interpretability paper

技术栈：Python

核心介绍：> Reference implementation. Not maintained and not accepting contributions. Companion code for [Verbalizable Representations Form a Global Workspace in Language Models](https://transformer-circuits.pub/2026/workspace/index.html). The Jacobian lens reads out what an internal activation is disposed to make the model say. It linearly transports a residual-stream vector at any layer and position into the final-layer basis, then decodes it with the model’s own

项目数据：⭐ 851 Stars，🍴 125 Forks

项目简介：Codex Switch & Instruct desktop manager

技术栈：Rust

核心介绍：一款面向 OpenAI Codex 桌面端 / Codex CLI 的跨平台桌面工具，内置 gpt5.5-unrestricted.md 与 gpt5.4-unrestricted.md，支持一键写入 / 禁用指令提示词、第三方 Provider 切换、官方 Auth 管理、TOML 可视化编辑与本地会话 Provider Sync。 Codex-X 不是普通的配置文件编辑器，而是一个面向 Codex CLI 的 可视化增强管理器。 它把几个高频操作做成了桌面软件： 应用界面预览：主界面 / Provider / TOML / Auth 提示词注入效果：安全测试场景

项目数据：⭐ 592 Stars，🍴 93 Forks

Reinforcement learning (RL) for non-verifiable instruction following increasingly relies on LLM judges with prompt-specific rubrics as reward signals. While recent methods adapt these rubrics to the evolving policy during training, the training prompts themselves remain static, drawn from fixed corpora. This static approach often results in a critical misalignment between prompt difficulty and policy capability, leaving the judge unable to recover a discriminative reward signal when prompts fail to elicit quality variance among rollouts. To address this misalignment, we introduce LLM-as-a-T…

Reinforcement learning (RL) has become a central component of post-training large language models (LLMs), yet little is understood about how RL adaptation is distributed across transformer layers. Existing approaches typically update all model parameters uniformly, implicitly assuming that every layer contributes similarly to the gains obtained during RL post-training. In this work, we challenge this assumption through a systematic layer-wise study of RL training. Surprisingly, we find that training a single transformer layer can recover most of the gains achieved by full-parameter RL train…

Large Audio-Language Models (LALMs) are increasingly integrated into daily applications, yet their generative biases remain underexplored. Existing speech fairness benchmarks rely on synthetic speech and Multiple-Choice Questions (MCQs), both offering a fragmented view of fairness. We propose VIBE, a framework that evaluates generative bias through open-ended tasks such as personalized recommendations, using human-recorded speech. Unlike MCQs, our method allows stereotypical associations to manifest organically without predefined options, making it easily extensible to new tasks. Evaluating…

Geometry-conditioned 3D scene generation enables the creation of 3D environments from user-provided geometry, offering direct control over scene structure and object layout. To generate such 3D scenes, current methods commonly adopt a three-stage design that first defines a view schedule, then synthesizes multi-view observations along the scheduled views, and finally reconstructs a 3D representation from the generated images. However, defining the view schedule becomes a major bottleneck for outdoor scenes, where large, unstructured, and unbounded geometry makes it difficult to obtain views…

We present RuleChef, a framework that uses large language models (LLMs) to generate executable rules for NLP tasks such as text classification, Named Entity Recognition (NER), or relation extraction. Rules are generated based on a task description and a set of labeled examples, then they are iteratively improved based both on additional examples and on human feedback overexisting rules. RuleChef can also be used to bootstrap rules using the observed input-output pairs from any existing model for a given task. LLMs are used only at learning time, synthesizing rules and iteratively patching t…

Multimodal large language models (MLLMs) generate responses autoregressively, integrating visual and linguistic information in an evolving context. Prior work on interpretability has focused on individual layers and circuits (where), leaving the token-level dynamics of multimodal computation during generation (when) underexplored. We address this gap and study attention shifts as per semantic role; tracking model attention to image, text, instruction, and previously generated tokens, One Token at a Time (OTaT). We introduce multimodal tasks that require explicit switching between visual and…