3 ms·
Where are the computer vision people? This is the perfect type of problem for multi modal LLMs
by KRAKRISMOTT 3y ago
Where are the computer vision people? This is the perfect type of problem for multi modal LLMs
- IlliOnato 3y agoExcept that the errors made by an LLM might be harder to spot then converter errors that typically are very blatant, and don't usually alter text (perhaps just drop parts of it). Also, a bug in a converter is conceptually much easier to fix than to re-train your LLM. I am not sure that AI in it's current state is useful when "high fidelity" is required.