This my personal space on the internet to share things that i enjoyed, things i think and some to document

hope you have fun

I am currently interested in knowing why and how neural networks function as they do.

• data as control : to control the data used to pretrain these model as way to clearly defined the model and its behaviour there by making it easier to study. Example training on assempy level instructions, formal languages, other synthetic yet interesting data
• how make better NN especially transformers
• how to evaluate frontier llm for true intelligence

things consumed daily