Skip to content

Refine image identification prompt for clarity and specificity #6

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Aug 12, 2025
Merged

Refine image identification prompt for clarity and specificity #6

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 4 additions & 5 deletions api/memoryalpha/rag.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -328,16 +328,15 @@ def search_image(self, image_path: str, top_k: int = 5,

# 6. Pass all info to the model, attach images
prompt = (
"You are an expert Star Trek analyst. Your task is to identify the user-provided image (attached as Image 0) as specifically as possible. "
"Among the retrieved images and their metadata below, determine which image best matches Image 0. "
"Use the metadata (image name, source page, description, similarity score, and text context) of the closest match to identify the user image. "
"Do NOT mention the match number, just provide the identification. "
"You are an expert Star Trek analyst. Look at the first attached image and determine which of the retrieved images below most closely matches it. "
"Use the metadata (image name, source page, description, similarity score, and text context) of the closest match to identify what is shown. "
"Provide a direct identification without mentioning image numbers, matches, or references to user images. "
"If no close match is found, say so clearly.\n\n"
)
prompt += "\n".join(formatted_images)
if text_contexts:
prompt += "\n".join(text_contexts)
prompt += "\nRespond with one or two lines identifying the user-provided image, based on the closest match and its metadata."
prompt += "\nRespond with one or two lines directly identifying what is shown in the image, based on the closest match and its metadata."

messages = [
{"role": "system", "content": "You are an expert Star Trek analyst."},
Expand Down