This is definitely not a strength for multi-modal LLM. Multi-modal capabilities are still too flaky especially when looking at a page of a PDF which can have mu

This is definitely not a strength for multi-modal LLM. Multi-modal capabilities are still too flaky especially when looking at a page of a PDF which can have multiple areas of focus.