True Positive Weekly #75

The most important artificial intelligence and machine learning news and articles

Hey, in this issue: the annotated Mamba paper; stealing part of a production language model; fixing all Gemma bugs; hallucination is inevitable - proven; training great LLMs entirely from the ground up; GGUF, the long way around; and more.

• The annotated Mamba paper

• [ArXiv] Stealing part of a production language model

• [MIT TR] Large language models can do jaw-dropping things. But nobody knows exactly why

• Fixing all Gemma bugs

• [ArXiv] Hallucination is inevitable: An innate limitation of large language models

• Training great LLMs entirely from the ground up in the wilderness as a startup

• [Ars Technica] Matrix multiplication advancement could lead to faster, more efficient AI models

• [Explained] GGUF, the long way around

Enjoy the newsletter? Please help us make it bigger and better by sharing it with colleagues and friends.