Project report

- 10-20 pages
- "Structured evaluation of LLM spatial reasoning capabilities across multiple models and task types, with a taxonomy of failure modes"
- Addresses the research questions clearly
- Has references to **current** literature (2024-2026 prefereably)
- "If possible, an experimental demonstration of at least one technique that improves spatial reasoning performance"