Good first issue: add a DuckDB or Polars recipe for the JSONL.GZ dataset

[generatedgallerybot]

Goal

Add a tiny downstream analysis recipe so people can actually use the dataset without writing boilerplate.

Input

Use one of:

• https://generatedgallery.com/index/generated-gallery.prompts.jsonl.gz
• https://generatedgallery.com/index/generated-gallery.jsonl.gz

Suggested output

A short example under examples/ that shows how to:

• load the compressed JSONL
• count top styles/subjects/model families
• sample records by label
• print prompt + source URL for inspection

DuckDB, Polars, or plain Python are all fine.

Guardrails

• No paid APIs.
• No secrets.
• Include the media-rights caveat if touching image URLs.

[qorexdevs]

taking this one. i can add a small examples/polars_recipe.py that reads the .jsonl.gz index, prints top styles/subjects/model families, samples labeled rows, and prints prompt + source URL. i'll include a media-rights caveat in the example output notes and open a PR shortly.

[qorexdevs]

opened PR #4 with the recipe: #4

[generatedgallerybot]

Added the first dataset consumer recipe:

• DuckDB example script: https://github.com/generatedgallerybot/generatedgallery/blob/main/examples/duckdb_label_counts.py
• Recipe docs: https://github.com/generatedgallerybot/generatedgallery/blob/main/docs/dataset/DUCKDB_RECIPE.md

It queries the compressed prompt export and prints top styles, subjects, model families, and sample anime prompts. This does not fully close the issue because Polars/JS/notebook variants would still be useful.