翻出来一个 2023 年的模型,可以将 embedding 转换回文本,实现 vec2text:
他是国王
- 他是男人
+ 她是女人
= 她是女王
https://fixupx.com/MF_FOOM/status/1687219083761385475
https://github.com/MF-FOOM/wikivec2text
是在这条推特下提到的:
https://x.com/rough__sea/status/1945571619868684373
他是国王
- 他是男人
+ 她是女人
= 她是女王
https://fixupx.com/MF_FOOM/status/1687219083761385475
https://github.com/MF-FOOM/wikivec2text
是在这条推特下提到的:
https://x.com/rough__sea/status/1945571619868684373
🧵 Thread • FixupX
MF FOOM (@MF_FOOM)
mf trained a simple model to translate ada-002 embeddings back to text and found something interesting:
sentence embeddings have arithmetic properties
the classic "king - man + woman = queen" still works just like w/ token embeddings — but can go a few…
sentence embeddings have arithmetic properties
the classic "king - man + woman = queen" still works just like w/ token embeddings — but can go a few…
🤩7👌2🆒2👍1
Forwarded from Hacker News
Signs of autism could be encoded in the way you walk (Score: 150+ in 1 day)
Link: https://readhacker.news/s/6xTkr
Comments: https://readhacker.news/c/6xTkr
Link: https://readhacker.news/s/6xTkr
Comments: https://readhacker.news/c/6xTkr
ScienceAlert
Signs of Autism Could Be Encoded in The Way You Walk
It's official.
Forwarded from Twitter Picture Bot
🥰20🍓6❤2
Forwarded from 知乎大巴扎
Telegraph
孩子问「自转一圈相当于整个宇宙围着我公转一圈,那些离我很远的天体岂不是轻易超光速了」,我该如何解释?
很好的问题,而且其物理内涵实际上并不像有些人以为的那样简单。以下我尽量避免使用公式和术语,试图用大多数人能理解的方式来进行回答(提示:这在很多细节上是不严谨的,但是不影响主旨)。 首先需要知道,所谓“物体运动不能超过光速”只是一种科普式的简化。在相对论里,“速度”这个概念并不是我们日常想象的那样,简单地用“距离除以时间”。因为相对论的核心观念就在于,什么叫做“距离”,什么叫做“时间”,这在“不同人的眼里”是不一样的。用相对论的话说,在不同的“参考系”里面,“距离”和“时间”的计算方式都会不同,只有先确定…
❤12🥱3😭2👻2
#TIL Haskell 在 2021 年将字符串库 Text 的内部实现从 UTF-16 切换到了 UTF-8,通过 SIMD 把性能(尤其是编解码和搜索)拉高了数倍甚至一个数量级
Proposal: https://discourse.haskell.org/t/hf-tech-proposal-1-utf-8-encoded-text/2499
PR:
- https://github.com/haskell/text/pull/365
- https://github.com/haskell/bytestring/pull/423
性能测试结果: https://gist.github.com/Bodigrim/365e388e080b17de45e80ab50a55fb4f
Proposal: https://discourse.haskell.org/t/hf-tech-proposal-1-utf-8-encoded-text/2499
PR:
- https://github.com/haskell/text/pull/365
- https://github.com/haskell/bytestring/pull/423
性能测试结果: https://gist.github.com/Bodigrim/365e388e080b17de45e80ab50a55fb4f
Haskell Community
HF Tech Proposal #1: UTF-8 Encoded Text
This proposal outlines a project plan for the migration of the text package from its current default encoding (UTF-16) to a new default of UTF-8. The lack of UTF-8 as a default in the text package is a pain point raised by the Haskell Community and many…
👍4🤷♀3
#TIL 二岐鹿角蕨可能是目前发现的唯一一种具有真社会性的植物。在作为附生植物生长时,其具有负责收集水源和分配营养的个体和负责提供结构支撑的个体(注:不是同一植株的不同叶片类型),且两种个体的分工由位置决定。其中有约 40% 的个体完全不育。
https://en.m.wikipedia.org/wiki/Eusociality#In_plants
https://www.nytimes.com/2021/07/02/science/ferns-social.html
https://pmc.ncbi.nlm.nih.gov/articles/PMC8525959/
https://en.m.wikipedia.org/wiki/Eusociality#In_plants
https://www.nytimes.com/2021/07/02/science/ferns-social.html
https://pmc.ncbi.nlm.nih.gov/articles/PMC8525959/
Wikipedia
Eusociality
Eusociality (Greek εὖ eu 'good' and social) is the highest level of organization of sociality. It is defined by the following characteristics: cooperative brood care (including care of offspring from other individuals), overlapping generations within a colony…
🆒7🐳2❤1🔥1
Forwarded from 每日消费电子观察 (horo)
降本增效:英特尔宣布终止开发 Clear Linux OS 发行版 - IT之家
https://www.ithome.com/0/869/282.htm
https://www.ithome.com/0/869/282.htm
Ithome
降本增效:英特尔宣布终止开发 Clear Linux OS 发行版 - IT之家
过去十年间,Clear Linux 展现了其在 x86_64 硬件上不俗潜力,而且哪怕是在 AMD 平台上也表现出了出色的性能。但随着英特尔的成本削减和裁员行动,Clear Linux 即将终止开发。
😭16😁4🤔2😢1
Forwarded from Hacker News
Microsoft Office is using an artificially complex XML schema as a lock-in tool (Score: 151+ in 8 hours)
Link: https://readhacker.news/s/6y3X3
Comments: https://readhacker.news/c/6y3X3
Link: https://readhacker.news/s/6y3X3
Comments: https://readhacker.news/c/6y3X3
The Document Foundation Blog
An artificially complex XML schema as a lock-in tool - The Document Foundation Blog
A document format is a tool for sharing knowledge and, as such, should be as simple and accessible as possible in relation to the complexity of the document content itself. This remains true even when the format is based on an XML schema that is hidden from…