# apple/pico-banana-400k

Repository: https://github.com/apple/pico-banana-400k
Canonical: https://ross.abutalabs.com/products/pico-banana-400k
Language: Python
License: NOASSERTION
License Family: other
Last push: 2025-12-16T00:26:40+00:00

## Health v2 (maintenance only)
Score: 42/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 57, release rhythm 35, longevity 22
- inputs: {"age_days": 316, "days_push": 261, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1845, forks 81 (observed 2026-08-28T04:05:43.394731+00:00)

## What it is
A large-scale dataset of ~400K text-image-edit triplets for text-guided image editing research, built from Open Images using Nano-Banana edits with Gemini-based quality verification. It includes single-turn SFT data, preference-learning pairs, and multi-turn editing examples spanning 35 edit operations across 8 semantic categories.

## Use cases
- train a text-guided image editing model
- fine-tune an image editing model with instruction data
- build preference learning data for image edits
- research multi-turn image editing
- evaluate image editing model quality
- train diffusion models for instruction-based editing

## When to choose
- training or fine-tuning instruction-guided image editing models
- researching preference optimization or robustness with failure cases
- needing diverse, quality-filtered edit triplets at scale

## When to avoid
- needing fully human-annotated edits rather than model-generated ones
- requiring images outside 512-1024px resolution
- projects needing a permissive license (license is custom/unspecified)

## Facets
- artifact type: dataset
- maturity: active
- function: machine-learning, data-generation, image-processing
- domain: machine-learning, computer-vision, image-processing, artificial-intelligence
- platform: python, cross-platform
- tags: image-editing, text-guided-editing, instruction-tuning, preference-learning, multimodal, sft, open-images, nano-banana, gemini

## Member repositories
- apple/pico-banana-400k (main) score 42

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:43.394731+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:17:54.508238+00:00, confidence not recorded.
  - readme: https://github.com/apple/pico-banana-400k (fetched 2026-08-28T04:05:43.394731+00:00, sha b1a3ef8a9e17)
- Data as of 2026-08-30T08:39:29.467469+00:00.