2026年9月30日 技术热点总结

项目简介：一个自己找热点、自己写日报的网站框架。把信源和精选标准换成你的，它就是你的行业热点站。

技术栈：TypeScript、ai、llm、mcp、news-aggregator、rss、self-hosted

核心介绍：一个自己找热点、自己写日报的网站框架。 把信源换成你的，把精选标准换成你的 KnowHow，它就是你的行业热点站。 跑起来 · 改成你的行业 · 它是怎么工作的 · 文档 AIHOT 是我做的一个 AI 热点网站。它每天从一批信源里收资料，用大模型先筛一遍、再独立打两次分，挑出真正值得看的，写成中文标题和摘要；把不同来源说的同一件事聚成一个事件，按有多少人在说排出热点；每天早上出一份日报。 这个仓库是它的完整框架：网站、后台、精选流程、聚簇和热度算法，所有提示词的原文和入选门槛，都在这里。 这半年，很多做法律、做 HR、做金融、做贵金属的朋友问我，能不能也给他们的行业做一个。 我做不了。我不懂你们的行业，不知道哪些信源有用，也不知道什么样的消息，对你们来说才叫热点。 既然我没办法满足所有人，那就把火种交到大家自己手上。

项目数据：⭐ 3,087 Stars，🍴 883 Forks

项目简介：Fine-tunes of Qwen3.5 and Gemma 4 for zero-shot classification

技术栈：Python

核心介绍：> New: v1.1 (29 September 2026). Jeff-Qwen3.5-0.8B and Jeff-Qwen3.5-2B now choose among up to 254 options > (v1.0: 26), with better calibration. On our long-list test the 0.8B goes from 40% to 95%. The 2B’s benchmark score > dips from 83.1% to 82.0%. Details in the changelog; v1.0 stays available on Hugging Face as code, with the same request format as Jev.** You describe a situation and list the options in plain words; Jeff returns a

项目数据：⭐ 1,011 Stars，🍴 38 Forks

项目简介：Independent CarPlay receiver for compatible Android head units. Wired and wireless public preview.

技术栈：Kotlin

核心介绍：> BYD support scope： These projects focus on BYD cars. They may work on other brands, but other brands are unsupported and there are no plans to add support or fix brand-specific incompatibilities. Download & website · Release · Report a problem

项目数据：⭐ 931 Stars，🍴 158 Forks

Code generated by large language models passes security checks at a rate that has barely moved in four years. In regulated backends, the defect classes that matter most are money arithmetic, time handling, retry safety, and access control. Teams answer with instruction files, yet the largest controlled study of instruction files to date found no benefit. This paper tests a narrower idea: generated code improves when the prompt carries a specification, a fixed preamble stating what must be true of the result. We pre-registered hypotheses, refuters, analysis code, and a one-shot generation ru…

While modern text-to-speech (TTS) systems generate highly natural speech and support inline non-verbal vocalization (NVV) tags, accurate control over these events remains challenging. A key gap is the lack of established post-training methods for non-verbal control in continuous autoregressive flow-matching TTS. To this end, we present NVAlign, a direct-gradient post-training framework for NVV tag-following in this architecture. We first perform supervised fine-tuning (SFT) of TTS models and an NVV-aware automatic speech recognition (NV-ASR) model on NVV-annotated speech, then freeze the NV…

Since the advent of knowledge distillation, KL divergence has been the standard loss in distillation. Recently, on-policy distillation (OPD) has emerged as an efficient post-training paradigm for LLMs. As a distillation method, OPD naturally inherits KL divergence as its standard loss. However, in this work, we find that KL divergence may not be necessary for OPD. We show that simply preserving the update direction is sufficient for effective OPD. As long as the update direction is toward the teacher, OPD works. More precisely, it is not the direction of every token, but the direction of a …

When generating text, a Transformer produces representations of past tokens at every layer, but each layer can normally use only representations from the same depth. This restriction prevents the model from fully reusing information it has already computed. We introduce WhiteMatter, which allows every layer to draw on past-token representations from any depth. A learned mixer selects the most useful depths for the current context and combines their representations into shared key-value (KV) cache channels. Sharing these channels across layers can reduce the cache size. Given the same number…

RLVR provides reliable trajectory-level credit, while OPSD offers dense supervision for token-level credit. This exposes a fundamental coupling when updating step-level credit direction and magnitude with teacher supervision, preventing steps from receiving reliable credit directions and contribution magnitudes, while making both vulnerable to teacher judgment errors and preference variance, as supported by our theoretical analysis. To separate credit direction from its contribution magnitude, we introduce Decoupled Credit Self-Distillation (DCSD), which theoretically decouples credit direc…

Learning from experience in LLM agents has become a key paradigm for developing self-evolving agents that continuously learn and expand their capabilities. Within this paradigm, synthesizing the agent skill has emerged as a promising solution for transforming accumulated experience into reusable procedural knowledge, serving as an important layer for the harness system that supplies agents at runtime. Despite its potential, existing approaches largely abstract past experience into fixed procedural knowledge before downstream demands are known, which risks discarding knowledge that later bec…