2026年10月4日 技术热点总结

项目简介：High-performance, token-based structured data engine for .NET. A core component of REX, the technology behind CAPCOM’s next-generation game engine.

技术栈：C#

核心介绍：RE:Dox is not just a JSON serializer. It parses structured data into a compact, fixed-size token DOM / IR, then uses that structural index for reading, editing, format conversion, serialization, deserialization, and automatic parallel deserialization. JSON / JSON5 / CBOR / MessagePack / TOML / XML / HTML / CSV / INI / DOX Compact token DOM / IR Reader / Writer / Serializer / Deserializer .NET objects, JSON, CBO…

项目简介：Open source SDK to build Muse gadgets

技术栈：C

核心介绍：Muse gadgets are open source devices you build yourself. Program an off-the-shelf ESP32 board or set up a Raspberry Pi with our device SDKs, then connect Muse to your displays, buttons, sensors, actuators, and whatever else you’ve got lying on your workbench. We open sourced the SDKs and firmware here. It’s built by hackers, for hackers, just for fun. Side effects of tinkering may include bricked boards, voided warranties, brownouts, or bankruptcies. Proceed at your own risk!

项目数据：⭐ 840 Stars，🍴 136 Forks

项目简介：Play Skyrim as a Minecraft player: Minecraft physics, inventory, blocks and combat inside Skyrim’s world (SKSE plugin + Fabric mod).

技术栈：C++

核心介绍：Play Skyrim as a Minecraft player. You move with Minecraft’s physics, carry Minecraft’s inventory and HUD, and place and break blocks in Skyrim’s world. You fight Skyrim’s NPCs with Minecraft weapons, and they fight back. Neither game is rewritten. Minecraft runs its own game logic, and Skyrim runs its world, NPCs, quests and saves. A Skyrim SKSE plugin and a Minecraft Fabric mod talk to each other through shared memory. Minecra…

项目简介：Build plugins that feel like native, first-class features of ChatGPT.

技术栈：TypeScript

核心介绍：OpenAI MCP Extensions adds ChatGPT-specific capabilities to MCP so developers can build plugins that feel like native, first-class features. All examples below use the Bits & Bolts plugin, which you can try by installing the plugin. Allow users to access your app from the sidebar. Render custom file viewers when users open a supported file type. Let users search your plugin’s resources from the composer and add references to a message. Let users select a CAD part using thumbnail choi…

Natural-language service requests can require a language-model decision before execution starts, consuming part of the request's latency budget. We integrate Jev's decision-oriented application programming interface (API) into edge service orchestration to reduce this overhead while retaining service completion. The integration extracts four to eight bounded intent fields and applies a shared validator, admission policy, and scheduler, accounting for decision waiting throughout the request timeline. We compare Jev, two self-hosted decision models, and three hosted large language models (LLM…

Video generation models are increasingly being explored as world simulators for embodied planning and learning. To do so effectively, these models must not only generate visually appealing frames, but also predict how environments dynamically evolve when executing goal-directed actions. While evaluating these capabilities is crucial, existing benchmarks focus mainly on single short actions or step-by-step instructions. This leaves multi-step physical reasoning underexplored, especially in egocentric video generation that requires planning to simulate proper execution to accomplish high-leve…

An assistant that serves the same user over a long horizon has to answer from what that user has revealed: which preferences still hold, which were revised, and which constraints apply now. Retaining that information is not the same as acting on it, and the two are usually optimized as if they were. Keeping the information as text makes the reader's input grow with the retained history, while compressing it into a fixed number of latent vectors bounds the interface but is typically trained to reconstruct text or imitate reference answers, both of which are scored on sequences the reader nev…

Video generation has rapidly progressed from short, low-quality clips to high-resolution, long-duration sequences with complex spatiotemporal dynamics. Despite strong generative priors learned through large-scale pretraining, pretrained video models often fail to reliably follow human intent, maintain temporal coherence, or satisfy physical and safety constraints. Compared with image and text generation, alignment in video generation presents unique challenges, including error accumulation over time, motion-appearance coupling, multi-objective trade-offs, and limited supervision for tempora…

An activation-steering coefficient sets intervention strength, but requesting a particular degree of persona expression requires a behavioral scale. We study persona dosing: controlling a language model through a trait description and a requested mean intensity. PersonaDose specializes a shared, description-conditioned FLAS controller on persona responses, then calibrates its flow time against measured trait expression. Training responses are not paired with requested target intensities. Across Llama-3.1-8B, Qwen3-8B, and Gemma-3-4B, PersonaDose raises core-trait expression at the Persona V…

Pretrained transformers use little of their depth to follow references in context. Thirteen base models reliably follow only 1.4-3.6 lines, and extra pretrained loops add little. A task-trained rank-8 LoRA at one early layer extends this computation with all model weights frozen. Qwen3-8B improves from 15.5% to 99% exact accuracy on 24-line chains; a longer-trained LoRA reaches 50 lines. Ouro-1.4B reaches 60 lines after four loops and at least 160 after eight. The LoRA starts a relay: program lines pass on their chain identity through a short range of middle layers. Frozen heads read progre…