Clean network implementation for learning temporal-logic-style scoring rules from trajectory predicates.

This repository does not include the full nuPlan pipeline. It contains the core learnable network, extraction utilities, training helpers, tests, and the project website/docs.

• flora/: network, differentiable logic operators, extraction, and training utilities
• tests/: unit tests for the core implementation
• examples/: minimal usage example
• docs/: website and documentation source

uv sync --all-extras

uv run pytest tests/