2026年6月文章摘要
2026年6月文章摘要
6月1号
6月5号
6月6号
-
Detection Showcase. LocateAnything performs diverse localization tasks under a unified vision-language model, including document understanding, GUI grounding, dense object detection, and OCR localization.
Detection Showcase: LocateAnything能够在统一的视觉语言模型下执行多种定位任务,这些任务包括文档理解、GUI对象定位、密集对象检测以及OCR定位。 -
Gemma 4 QAT models: Optimizing model compression for mobile and laptop efficiency
Gemma 4量化感知训练模型:优化模型压缩以提升移动设备和笔记本电脑的使用效率 -
Jason Swett | My Agent Skill for Test-Driven Development
First I clue the agent in to what I call the specify-encode-fulfill loop, which is my personal alternative to red-green-refactor. Specify-encode-fulfill (SEF) goes like this:
首先,我会向代理说明我所谓的“指定-编码-执行循环”,这是我对“红绿重构”方法的个人改进方案。“指定-编码-执行循环”的具体步骤如下:- Specify: Come up with the specifications for what you want to build
要求:明确你想要构建的目标的具体规格要求。 - Encode: Encode those specifications as automated tests (executable specifications)
编码:将这些规范转换为自动化测试用例(可执行的规范)。 - Fulfill: Write the code to fulfill the specifications
实现方式: 编写符合这些规格要求的代码即可。
SEF is the high-level view of what, to me, TDD is all about. At a slightly lower level is Kent Beck's Canon TDD, which I've described below in my own words.
在我看来,SEF正是TDD的核心所在。而稍低一个层次的是Kent Beck所提出的TDD原则,下面我会用自己的话来描述它。- Write a list of the specifications within scope of the current TDD session
- Encode each item in the list as an automated test
- Change the code just barely enough to make the current test failure go away. Avoid "speculative coding" - if we write more code than necessary to make the current test failure go away, we risk having code never exercised by any test
- Optionally refactor, but not before committing the behavior change. Never mix behavior changes with refactoring
- Until the list is empty, go back to #2
In my judgment, the biggest AI productivity gains come from when AI is combined with timeless, immutable principles which were discovered decades ago, hold just as true today, and which, no matter what new technologies may arise, will never cease to be useful.
在我看来,人工智能在生产力方面的最大提升,恰恰来自于将其与那些几十年前就被发现、至今仍然适用、而且无论未来出现什么样的新技术都不会失去其价值的、永恒不变的原理相结合。 - Specify: Come up with the specifications for what you want to build
6月22号
距离上次搜集信息过去好久了,哎
6月23号
-
2 位动态量化 UD-IQ2_M 占用 239GB 磁盘空间——这可以直接安装在配备 256GB 统一内存的 Mac 上,并且在配备 1x24GB GPU 和 256GB 内存并启用 MoE 卸载功能的情况下也能良好运行。1 位量化模型可安装在 223GB 内存上,而 8 位量化模型则需要 810GB 内存。
-
Prompt Injection as Role Confusion
This is a blog-style writeup of the paper. We show prompt injections are driven by a flaw in how LLMs perceive roles. This lets us create new attacks, explain mech interp results, and predict when attacks succeed. We then discuss what roles are and why they matter, and share research ideas for a science of roles.
-
DeepSeek 多模态
- DeepSeek 杀入多模态,识图功能正式上线!
- DeepSeek识图模式全量上线×V4.1多模态发布倒计时
AIGC?由于我账户没有 多模态API,无法验证真实性?
-
VibeThinker-3B: Exploring the Frontier of Verifiable Reasoning in Small Language Models
微博团队?

浙公网安备 33010602011771号