Hugging Face's logo

Spaces:
FoodDesert
/
Prompt_Squirrel_RAG
Running

App Files Community
Prompt_Squirrel_RAG / README.md
Food Desert
Fix docs marker placement and simplify Space README
d5a5b38
preview code
|
raw
history blame contribute delete
2.2 kB

A newer version of the Gradio SDK is available: 6.11.0

Upgrade
metadata 
title: Prompt Squirrel
colorFrom: gray
colorTo: gray
sdk: gradio
sdk_version: 5.43.1
python_version: 3.10.12
app_file: app.py
pinned: false
license: apache-2.0

Prompt Squirrel RAG

Prompt Squirrel converts rough natural-language image descriptions into a structured, editable prompt built from a fixed image-tag vocabulary.

The intent is practical: produce a strong starting prompt quickly, then let users directly refine tag choices in the UI.

What This Space Does

  • Rewrites free text into retrieval-friendly pseudo-tags.
  • Retrieves candidate tags from a fixed vocabulary.
  • Uses closed-set LLM selection (index-only) so models cannot invent arbitrary tags.
  • Expands implications and groups tags into ranked editable rows.
  • Builds the final suggested prompt from current toggle states.

How To Use

  1. Enter an image description in Enter Prompt.
  2. Click Run.
  3. Toggle tags in rows to add/remove them.
  4. Copy the final text from Suggested Prompt.

Technologies Used

  • FastText embeddings for semantic retrieval.
  • HNSW indexes for efficient nearest-neighbor search.
  • Reduced TF-IDF vectors for context-aware ranking.
  • OpenRouter instruction models for rewrite, structural inference, probe inference, and closed-set selection. Default model: mistralai/mistral-small-24b-instruct-2501 (chosen empirically on internal caption-evident evaluations; configurable).
  • Gradio for the interactive web UI.

Current Snapshot

Latest relabel-aware rescore on caption-evident n=30:

  • Micro precision/recall/F1: 0.554 / 0.750 / 0.638
  • Macro precision/recall/F1: 0.564 / 0.764 / 0.633
  • Ground-truth scale: 30 images, 440 total tag assignments, 205 unique tags

Latency from local UI timing logs (n=110 runs):

  • Median: 8.56s
  • P75: 12.56s
  • P90: 18.33s

Documentation And Contracts