Pinned Loading

1. decoder-only-transformer-from-scratch decoder-only-transformer-from-scratch Public

   A decoder-only Transformer implemented from scratch in PyTorch for autoregressive next-token prediction, with inspectable intermediate states for architectural analysis.

   Python

2. Inside-Transformers Inside-Transformers Public

   This repository is a systematic exploration of Transformer architectures, focusing on how and why they work. It includes internal mechanics and data flow analysis, architecture-level breakdowns, co…

   Jupyter Notebook