The deeper you dig on how something works, the more of a miracle it becomes that anything works at all.
COMPUTING
-
Ada Lovelace’s First Computer Program from 1843
By
–
This diagram represents what's widely considered to be the world's first "non-trivial" computer program, developed in 1843 by Ada Lovelace for Charles Babbage's Analytical Engine. (v/
@fermatslibrary
) -
Ultra-Scale Playbook: Training DeepSeek-V3 with Advanced Parallelism
By
–
The Ultra-Scale Playbook (the large-scale LLM training guide from the @huggingface science team) is out now! It is a 246-page, very nicely designed PDF that walks you through learning how to train your own DeepSeek-V3 model using:
• 5D parallelism, • ZeRO,
• fast kernels, -
Moonshot Releases Faster Kimi K2 Turbo
By
–
Moonshot released kimi-k2-turbo-preview, a 4 times faster version of the earlier available Kimi K2 model.
-
Qwen 30B model achieves 60 tokens per second performance
By
–
I got 60 tokens per second from the new Qwen 30B model
-
Qwen 4bit and 6bit Model Memory Requirements Guide
By
–
The 4bit model should work comfortably on 32GB, the 6bit might just fit on that too If you have 16GB or less you're out of luck for this particular Qwen model
-
19th Cowboys 20th Flyboys 21st Codeboys Evolution
By
–
19th century: Cowboys
20th century: Flyboys
21st century: Codeboys -
Context Window Performance and RAM Requirements for AI Models
By
–
60 tokens a second, context window is supposedly up to 262,144 but that will use a ton more RAM, probably more than my Mac can handle (I've not tried though)
-
Running Qwen3-Coder-Flash 30B on M2 MacBook Pro
By
–
Here's the pelican riding a bicycle and the implementation of space invaders I got from running the 24.82GB 6bit MLX version of Qwen3-Coder-30B-A3B-Instruct – aka Qwen3-Coder-Flash – on my 64GB M2 MacBook Pro
