2026年9月11日 技术热点总结

技术栈：Python

核心介绍：English | 中文 generalizes the production-proven recipe — SSD expert offload + Recover-LoRA + prerouter routing prediction — into an extensible framework. The backend is isolated by design: the current MLX backend runs on Apple Silicon, and additional platforms (CUDA, …) plug into the same core abstractions. Two model tiers ship with the framework. Each tier is an end-to-end release: the released checkpoint, the trained LoRA adapters, and the trained prerouter heads work together as one unit.

项目数据：⭐ 980 Stars，🍴 80 Forks

项目简介：Asisten crypto berbahasa Indonesia: RAG pengetahuan 267 topik + data pasar realtime (6 bursa, WebSocket, derivatif, on-chain, TVL, DeFi) + tool-calling agent + LLM synthesis

技术栈：Python

核心介绍：Asisten crypto berbahasa Indonesia yang menggabungkan RAG (Retrieval-Augmented Generation) dengan data pasar realtime. Menjawab pertanyaan konsep/edukasi dari korpus pengetahuan, sekaligus menyajikan harga, market cap, funding rate, TVL, dan sentimen langsung dari sumber live — tanpa API key untuk data pasar.

项目数据：⭐ 898 Stars，🍴 0 Forks

项目简介：Agent skill for impressive 3D visuals using Blender + image gen + subagent critic

技术栈：JavaScript

核心介绍：By @anshuc. An agent skill that builds a game, app, or scene with impressive visuals, by creating a closed loop: 1. AI “dreams” up a high-quality target screenshot using image generation 2. AI builds with this target in mind 3. A separate AI critic compares the live screenshot to the target and provides feedback 4. AI loops back to step 2 until critic is satisfied 5. Optionally, AI loops back to step 1 and dreams up an even better target based on the current state.

**项目数…

项目简介：Topic in, narrated explainer video out. A Claude Code / Codex skill that turns any topic into a black-canvas motion-graphics explainer video with TTS voiceover, subtitles and a chapter progress bar. Chinese or English; every frame drawn in code with Remotion.

技术栈：TypeScript、agent-skills、ai-agents、ai-video、claude-code、claude-code-skill、codex、edge-tts、educational-video

核心介绍：It is not a CLI. What ships here is the whole method an AI coding agent needs to finish the film: a compilable Remotion template, a primitives and lighting library, tooling for voiceover / storyboard / r…

项目简介：A practical hardware and software setup for a compact RL-powered biped robot.

技术栈：Python

核心介绍：Microduck 是一个小型双足机器人项目，包含两部分： 本 README 面向装机和第一次运行：准备硬件、接线、刷写 microduck.img.xz、初始化 Wi-Fi、SSH 登录、运行模型，以及后续更新代码。3D打印的文件在项目中。 视频教程：https://www.youtube.com/watch?v=Vep8AjoCnEM 感谢开源项目 microban 和 microduck。本项目的机械结构、部署思路、仿真训练和实机行走流程都受益于这些开源工作。 当前发布镜像已经内置： 当前预构建镜像： 下面是组装一台 Microduck 的主要物料。价格会随地区、采购渠道和批量变化，仅用于估算。 控制代码期望的 ID 映射如下。烧录镜像之前或

项目数据：⭐ 650 Stars，🍴 164 Forks

Sequential memory agents process long documents by reading chunks one after another while maintaining a compact memory state, coupling document traversal to reasoning depth. This coupling introduces sensitivity to evidence placement and ties inference latency linearly to document length. We introduce PARSER, which decouples reading from reasoning. A bank of lightweight subagents each bound to a single chunk read the entire document in parallel, while a lead agent reasons in depth through iterative scatter–gather rounds: at each round it broadcasts a query to all subagents, aggregates the r…

AI-powered search products such as ChatGPT search, Google's AI Overviews, and Perplexity provide LLM-synthesized answers grounded in live web results. We developed OreoLook (formerly lixSearch), an open-source answer engine using automated browser agents and provider-routed LLM inference. Its local search, caching, session-management, and embedding stack runs on commodity CPU hardware; answer synthesis is performed by a remote inference provider. As usage grew, sessions lost context, equivalent queries triggered redundant work, and URLs were repeatedly embedded across sessions.
We present…

We study scheming in LLM agents, in which agents covertly pursue misaligned goals. Our focus is to understand how scheming arises from the interaction of key factors, such as instrumental goals, environmental affordances, oversight conditions, and perceived consequences. Prior work examines only a small number of scenarios, limiting the ability to isolate how these conditions shape an agent's propensity or capability to scheme. This limited scale and task diversity also restrict coverage of realistic deployment settings and the range of scheming strategies that can be observed. To this end,…

We consider the problem of support recovery for sparse binary signals from noisy linear measurements. For sparse Gaussian measurement matrices we identify sufficient conditions on the minimal sample size for maximum-likelihood recovery in the high-SNR regime ds/p to infty, where p denotes the signal dimension, s the number of non-zero components of the signal, and d the expected number of non-zero components per row of measurement. Combined with known lower bounds, this yields an information-theoretic threshold of order slog(p/s) / log(ds/p), making explicit the price of measurement sparsit…

Training data attribution (TDA) aims to identify training examples that shape model behavior, but its intervention value depends on both which examples are selected and how they are modified. Influence functions (IF) estimate behavioral changes under infinitesimal reweighting, yet IF-selected examples often show limited advantages over random selection under conventional weight-based interventions. This raises the question of whether influential examples lack intervention value or whether reweighting fails to realize their behavioral leverage.We introduce influence-guided response rewriting…

Leading benchmarks for formal theorem proving with large language models are small collections drawn from competition math, such as the IMO and Putnam, that poorly represent field-specific applications. We introduce StochBench, a Lean 4 benchmark of 450 graduate stochastic-processes problems at varying abstraction levels, each paired with its natural-language source. Addressing a field underrepresented in Mathlib, it covers finite and countable Markov chains, renewal processes, random walks, martingales, stopping times, queues, Brownian motion, stochastic calculus, weak convergence, and Poi…