apunkt/SurfSense
1
0
Fork 0
mirror of https://github.com/MODSetter/SurfSense.git synced 2026-05-01 03:46:25 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

Simplify vision LLM image description prompt

Browse source
This commit is contained in:
CREDO23 2026-04-09 14:56:18 +02:00
parent afd3c2cde2
commit caaec2e0a7
1 changed files with 3 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

9
surfsense_backend/app/etl_pipeline/parsers/vision_llm.py
View file

@ -4,12 +4,9 @@ import mimetypes
from langchain_core.messages import HumanMessage
_PROMPT = (
"Analyze this image thoroughly and produce a detailed markdown description.\n\n"
"Include:\n"
"- All visible text, transcribed verbatim\n"
"- Description of diagrams, charts, tables, or visual structures\n"
"- Key subjects, objects, or scenes depicted\n\n"
"Output only the markdown content, no preamble."
"Describe this image in markdown. "
"Transcribe any visible text verbatim. "
"Be concise but complete — let the image content guide the level of detail."
)