#self-attention

Transformer Attention from Scratch in NumPy (Python)

Build transformer attention from scratch in NumPy with runnable code. Scaled dot-product, multi-head attention, causal masking, and heatmaps step by step.