JSON Output Correction
Hello,
I am sorry if this isn't the place to ask such questions; I am a newbie in LLM, Computer Vision, and HF.
I successfully ran some tests extracting text as JSON with specific document fields. I am impressed with the results, but the problem is that the value found by the model in the document is sometimes modified. I would like to have the raw value. Should I change the temperature parameter, or are there other instructions I could add to the prompt?
Role: Graphic Designer
Task: Create a vector image featuring an attractive, friendly man and woman.
Output Format: The final image should be in vector format, maintaining the same art style as a previously provided spaceship design.
Details:
- Composition: The man and woman should be the central focus of the image, positioned in the center facing forward.
- Posture: They should be standing side by side, a few feet apart, holding hands.
- Facial Expression: Both characters should have their heads slightly turned towards each other, displaying sheepish grins.
- Style Consistency: Ensure that the art style matches exactly with the spaceship element previously provided.
Tone: Formal
Constraints:
- Ensure all proportions and features are consistent with the original spaceship art style.
- The background should be minimalistic to emphasize the characters.
Example: If the spaceship was drawn in a sleek, modern style with vibrant colors, the characters must also reflect this style in their depiction.
Please provide a detailed description of the final product based on these guidelines.