Fix Pylance type errors, add inductor compilation support

Full training run, code cleanup, mypy/ruff fixes

feat: Add FLAN-T5 compatibility with relative position bias

Style: Fix linting errors and organize imports (ruff & mypy)

Implemented the following parts for the Transformer model: poisional_encoding, feedfoward, encoder, and skeleton of decoder. As well as Test cases for each class respectively and visualizations

Implemented ScaledDotProduct Attention and Multi-Head Attention