Publications

Positional encoding is rather an accidental requirement	What is internal logic behind the LLM pipeline of tensors?	If anybody wants to learn LLM by hand - AI and ML Community	Here's untold story about latent matrix space
I've found logic predicates in the LLM self-attention!	What is happening inside?	Digging the model loss in a mirror of the training process	My Transformer research is available in the public domain
How many redundant parts are there in a Transformer?	The whole Transformer Series	How small LLM can be?	Two-faced mathematics - is it real?
How to check validity of global model parameters of LLM	Self-attention in the architecture of transformer holds a mystery	Neural nets grew up from trivial Frank Rosenblatt's fully connected layers	What is this thing for?
Midnight talks with LLM	What language does LLM speak?	Apply hard scrutiny to find out hidden relationships between raw numbers	Life with fully connected layers is full of computational surprises
I've put the project on the GitHub under MIT license	Graphs of self-crossing for positional embeddings	I decided to learn LLM by hand	I've got a clear understanding of why a Transformer can comprehend
To be continued...	To be continued...	To be continued...	To be continued...
To be continued...

© Image credit to all 3^rd party artworks has been expressed in linked publications.