Forwarded from Hacker News (yahnc_bot)
matklad.github.io
Zig And Rust
This post will be a bit all over the place.
Several months ago, I wrote Hard Mode Rust, exploring an allocation-conscious style of programming.
In the ensuing discussion, @jamii name-dropped TigerBeetle, a reliable, distributed, fast, and small database written…
Several months ago, I wrote Hard Mode Rust, exploring an allocation-conscious style of programming.
In the ensuing discussion, @jamii name-dropped TigerBeetle, a reliable, distributed, fast, and small database written…
🔥1
Forwarded from Hacker News (yahnc_bot)
ChatGPT outperforms crowd-workers for text-annotation tasks https://arxiv.org/abs/2303.15056
Forwarded from Hacker News (yahnc_bot)
Procedural 3D mesh generation in a 64kB intro https://www.ctrl-alt-test.fr/2023/procedural-3d-mesh-generation-in-a-64kb-intro/
Ctrl-Alt-Test
Procedural 3D mesh generation in a 64kB intro - Ctrl-Alt-Test
In this third article of our series on the making of a 64kB intro, we present how we use and combine a few simple techniques to create a variety of procedural 3D meshes for very few bytes.
Forwarded from Lobste.rs
Q 向强大群友们提问
有没有人 Fork 了 solarized,然后把它的 Highlight colors 修了的
又遇到一个用 Highlight color 然后什么都看不见的...
有没有人 Fork 了 solarized,然后把它的 Highlight colors 修了的
又遇到一个用 Highlight color 然后什么都看不见的...
EMS 国际什么情况(
需要通关缴税,不打电话不发短信,EMS 国际物流查询没说法,必须得用他们的小程序查才能看到缴税通知
真就互联网 \subseteq 微信
需要通关缴税,不打电话不发短信,EMS 国际物流查询没说法,必须得用他们的小程序查才能看到缴税通知
真就互联网 \subseteq 微信
Meow-meow 🍓
邮局:ems国际业务不在这里呀,你打个电话,说不定得去朝阳
ems国际报关客服:你得换一个小程序,用这个(不是 ems 的小程序),你说那网站(就是他们第一方网站 int.ems.com.cn )我都没听说过
这太蠢了...
这太蠢了...
Forwarded from Hacker News (yahnc_bot)
Llama.cpp 30B runs with only 6GB of RAM now https://github.com/ggerganov/llama.cpp/pull/613
GitHub
Make loading weights 10-100x faster by jart · Pull Request #613 · ggml-org/llama.cpp
This is a breaking change that's going to give us three benefits:
Your inference commands should load 100x faster
You may be able to safely load models 2x larger
You can run many concurren...
Your inference commands should load 100x faster
You may be able to safely load models 2x larger
You can run many concurren...