Two questions on (gemini) prompt design:
Two questions on (gemini) prompt design:
1) Will a prompts input-schema have value to the prompt? The output-schema is a major part in shaping the prompt, but is the inputschema known to the model at all?
2) Do we know how multiple parts in a message affects gemini? What is the genkit code for joining parts? Simply appending them? I see a value if programatically splitting prompt across parts and generate these really dynamically. But i wonder if gemini behaves differently when I do this as opposed to compiling one big text with multiple paragraphs, sections/formatting etc.
1) Will a prompts input-schema have value to the prompt? The output-schema is a major part in shaping the prompt, but is the inputschema known to the model at all?
2) Do we know how multiple parts in a message affects gemini? What is the genkit code for joining parts? Simply appending them? I see a value if programatically splitting prompt across parts and generate these really dynamically. But i wonder if gemini behaves differently when I do this as opposed to compiling one big text with multiple paragraphs, sections/formatting etc.
