File size: 2,199 Bytes
d5a5b38
ed9438a
 
 
 
 
 
 
 
 
 
 
 
 
d5a5b38
ed9438a
d5a5b38
ed9438a
d5a5b38
ed9438a
d5a5b38
 
 
 
 
ed9438a
d5a5b38
ed9438a
d5a5b38
 
 
 
ed9438a
 
 
 
 
 
 
d5a5b38
 
ed9438a
d5a5b38
ed9438a
 
 
 
 
 
 
 
 
 
 
 
 
d5a5b38
ed9438a
d5a5b38
ed9438a
 
d5a5b38
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
---
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

- Full architecture and rationale (same source used by the in-app docs accordion): [docs/space_overview.md](docs/space_overview.md)
- Retrieval contract: [docs/retrieval_contract.md](docs/retrieval_contract.md)
- Rewrite contract: [docs/rewrite_contract.md](docs/rewrite_contract.md)
- Closed-set selection contract: [docs/stage3_contract.md](docs/stage3_contract.md)