This repository contains a PyTorch implementation of a Transformer model, inspired by the original paper Attention Is All You Need (2017). The model is trained from scratch on an English-to-German dataset using only basic PyTorch components, with custom implementations of multi-head attention, positional encoding, and encoder-decoder architecture.

• Encoder-decoder Transformer (no external libraries)
• Custom attention, positional encoding, and training loop
• Character-level tokenization
• Outputs valid German sequences

src/
├── model/
├── data/
notebooks/
outputs/