Hugging Face's logo

Datasets:
Rootport
/
HareSkip-calibration

Modalities:
Image
Formats:
imagefolder
Size:
1K - 10K
Tags:
image
diffusion
diffusion-transformer
inference-acceleration
step-skipping
benchmark
Libraries:
Datasets
License:
Dataset card Data Studio Files
xet
 Community
1
HareSkip-calibration / README.md
Rootport's picture
Rootport
docs: add dataset card (CC-BY-4.0, provenance, license notes)
53fff18 verified
preview code
|
Raw
History Blame Contribute Delete
12.3 kB
metadata 
license: cc-by-4.0
pretty_name: HareSkip Calibration Dataset
size_categories:
  - 1K<n<10K
tags:
  - image
  - diffusion
  - diffusion-transformer
  - inference-acceleration
  - step-skipping
  - benchmark
  - anime

HareSkip Calibration Dataset

1. Summary

This dataset contains all images and measurements generated during a July 2026 recalibration experiment for HareSkip, a step-skipping inference-acceleration extension for DiT-based image diffusion, implemented on top of Forge neo (repository: Rootport-AI/forge-neo-Anima-HareSkip).

The experiment measured how much visual damage results from skipping specific denoising steps, across single-step sweeps, multi-step interaction patterns, and out-of-sample validation patterns representative of HareSkip's actual runtime behavior. The full write-up, methodology, and analysis are in the experiment report:

docs/recalibration-2026-07/REPORT.md in the forge-neo-Anima-HareSkip repository.

This dataset consists of:

  • 3,186 PNG images (1536×1536), covering reference (no-skip) images and skip-pattern candidates across 45 conditions (3 samplers × 5 prompts × 3 seeds), plus a step-count anchor set and a Shift-transfer spot check.
  • Measurements: 6 Judge comparison runs (LPIPS-VGG/AlexNet, SSIM, PSNR, etc.) as CSV, plus a frozen pre-registered prediction table.
  • Patterns: the skip-pattern text files used to drive generation, including the frozen out-of-sample holdout patterns.
  • Provenance: a fixed-commit statement, frozen-file hashes, and a full SHA-256 manifest.

The purpose of publishing this dataset is reproducibility: every image is traceable to an exact commit, an exact skip pattern, and an exact set of generation parameters, so that the report's findings can be independently re-derived or re-analyzed.

2. Dataset structure 

images/
measurements/
patterns/
provenance/

images/ (3,186 PNG files, 1536×1536)

Directory Images Content
Reference-images02/ 45 No-skip reference images, one per condition (3 samplers × 5 prompts × 3 seeds), regenerated on the fixed commit
Calibration-phase1-02/ 1,305 Stage 1: single-step skip sweep (steps 2–30, 29 positions × 45 conditions)
Calibration-stage2/ 1,260 Stage 2: interaction patterns (adjacent pairs, distance scans, cross-band pairs, chains, triples — 28 patterns × 45 conditions)
Stage3-Calibration/ 450 Stage 3: out-of-sample validation against HareSkip's actual runtime patterns (10 patterns × 45 conditions)
StepAnchor-images/ 90 ±1-step anchor: no extension, total step count varied to 29/31 only (45 conditions × 2)
ShiftSpotcheck-Reference/ 6 Reference images for the Shift-transfer spot check (Shift=1)
ShiftSpotcheck-Calibration/ 30 Shift=1 candidates at 5 characteristic curve points × 6 conditions

Step 1 is excluded throughout: Manual Skip mode rejects step 1 (its residual is not yet held, so it is physically unskippable).

measurements/

Path Content
judge-results/<run>/merged-results.csv + run-config.json (6 runs) Nz DoppelPix Judge comparison output: LPIPS-VGG (primary metric), LPIPS-AlexNet, SSIM (win 7 / win 11), PSNR, per candidate-vs-reference image pair
predictions/stage3-predictions.csv Frozen pre-registered prediction table for the 450 Stage-3 out-of-sample points (predicted before generation; SHA-256 below)
predictions/fit_params.json Frozen saturation-link fit parameters used to produce the predictions

The 6 included runs are: Calibration-phase1-02_20260718-150151, Calibration-phase1-02-remainder_20260719-160107, StepAnchor_20260727-075911, Calibration-stage2_20260727-081812, ShiftSpotcheck_20260729-084448, Stage3_20260729-085016. All completed with zero metrics errors.

An earlier measurement run (Calibration-phase1_20260716-234016) and its associated images/analysis are excluded from this dataset: they were generated before a Forge-neo version confound was discovered (see §4) and would mislead reproducibility checks if published alongside the corrected data.

patterns/

Skip-pattern definitions used to drive Manual Skip generation, plus the prompt set:

File Content
layer1-single-skip-sweep-30steps.txt Stage 1 patterns (29 lines)
stage2-interaction-patterns.txt Stage 2 patterns (28 lines)
stage3-holdout-patterns.txt Stage 3 out-of-sample holdout patterns (10 lines, frozen; SHA-256 below)
stage3-holdout-patterns-REPORT.md Explanation of how the Stage 3 patterns were derived and verified
stage3-holdout-simulate.py Offline simulator used to enumerate the Stage 3 patterns from HareSkip's probability model
prompt.txt The 5 prompts used across all conditions

provenance/

File Content
PROVENANCE.md Fixed-commit statement, frozen-file hashes, excluded-data rationale, included-run list
manifest.sha256 SHA-256 checksums for every file in the dataset (sha256sum-compatible format), 3,206 entries

3. Self-describing images

Every PNG carries a tEXt chunk keyed parameters, in Forge's standard infotext format, embedding the full generation condition — prompt, negative prompt, seed, sampler, schedule type, CFG scale, Shift, model and hash, HareSkip mode, ResRefine formula, and the realized skipped-step set (Manual skipped_steps). Example (from images/Calibration-phase1-02/ERSDE-Beta_Prompt001-schoolgirl-5193/00000-3000995193.png):

masterpiece, best quality, score_7, safe, a girl with demon horns is sitting indian style on ground, holding a cat. ...
Negative prompt: worst quality, low quality, score_1, score_2, score_3, artist name
Steps: 30, Sampler: ER SDE, Schedule type: Beta, CFG scale: 4.0, Shift: 3.0, Seed: 3000995193, Size: 1536x1536, Model: anima_baseV10, Model hash: bd43b7cffe, Module 1: qwen_image_vae, Module 2: qwen_3_06b_base, RNG: CPU, HareSkip enabled: True, HareSkip mode: Manual Skip, ResRefine formula: Reuse (residual only), Beta schedule alpha: 0.6, Beta schedule beta: 0.6, Manual skipped_steps: 2, Version: neo-2.27

Manual skipped_steps is written by the patcher from the steps it actually skipped during generation, not from the requested input — it is a realized value, not a nominal one. This makes every image self-describing on its own: the generation condition can be recovered from the file alone, with no dependency on directory naming or an external index. Dataset-wide acceptance testing cross-checked this realized value against file name and pattern canon for all 3,186 images.

4. Reproducibility & integrity

All 3,186 images were generated on a single, fixed Forge neo commit: b61642140acb7c2f1c65c5d0f2ab961b7366c02e (reported in PNG metadata as Version: neo-2.27). This fix was adopted after an earlier measurement round was found to be confounded by an undocumented Forge-neo version drift between reference and candidate images (see the source report, §2.2, for the full incident writeup); that earlier round is excluded from this dataset (§2 above).

Integrity verification

Every file in the dataset is listed in provenance/manifest.sha256. To verify the full download:

sha256sum -c provenance/manifest.sha256

Frozen hashes and pre-registration

Stage 3 (out-of-sample validation) followed a pre-registration discipline: predictions were frozen before the corresponding images were generated or measured.

File SHA-256 (first/last 8 hex)
patterns/stage3-holdout-patterns.txt 659A7445...7CD3
measurements/predictions/stage3-predictions.csv F8495DD1...FA131D

(Full 64-character hashes are recorded in provenance/manifest.sha256.) The Stage 3 holdout patterns were derived by offline simulation of HareSkip's probability model (sigmoid_band_v0.1) and independently verified against the extension's own code before generation. The prediction table's parameters were fit only on Stage 2 data; re-fitting on Stage 3 data was explicitly disallowed. This lets a third party check that the predictions in this dataset were not adjusted after seeing the outcomes they are being judged against.

5. Model and licensing

The images in this dataset were generated with Anima (Anima-Base v1.0) by CircleStone Labs & Comfy Org (https://huggingface.co/circlestone-labs/Anima), a derivative of NVIDIA Cosmos-Predict2-2B-Text2Image ("Built on NVIDIA Cosmos"). The exact model file used has SHA-256 bd43b7cffe1ed1153d9c41e7beb2f18cb1273eafbaa3af3edd6a173dc90a006e, which matches the Model hash: bd43b7cffe recorded in every image's infotext (verified against the dataset).

License note: The Anima model weights themselves are distributed under the CircleStone Labs Non-Commercial License v1.2 (non-commercial use only for the weights). However, that license's §2.e addresses Outputs (generated images) explicitly:

"We claim no ownership rights in and to the Outputs. ... You may use Outputs for any purpose (including for commercial purposes)."

Accordingly, this dataset — the collection of Outputs plus the original measurements produced by this experiment — is released under CC-BY 4.0. This license covers the images and measurement data in this repository. It does not cover the Anima model itself; anyone wishing to use the Anima model weights to generate new images must comply with CircleStone Labs' own license terms.

6. What this data shows

  • Damage from skipping a step is concentrated in the early denoising steps (Spearman ρ = −0.784 between step position and damage), with the worst position typically at skip 2–4 rather than the very first skippable step.
  • A simple predictive formula — the sum of single-step damages passed through a two-parameter saturation link — passed pre-registered out-of-sample validation on 450 held-out points (Spearman ρ = 0.930), with no need for a streak (consecutive-skip) penalty term.
  • Predicted damage carries a consistent safe-side bias (actual damage tends to be ≤ predicted), which is favorable for quality-guarantee use but means absolute-value predictions should not be taken as tight bounds.
  • The hypothesis that damage is governed purely by trajectory coordinate z (independent of the Shift parameter) did not hold cleanly: changing Shift from 3 to 1 changes the shape (steepness) of the damage curve, not just its position.

See docs/recalibration-2026-07/REPORT.md for full methodology, statistics, and discussion.

7. Citation 

@misc{rootport2026hareskipcalibration,
  author       = {Rootport},
  title        = {HareSkip Calibration Dataset},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/Rootport/HareSkip-calibration}},
}

日本語要約

本データセットは、DiTベース画像生成の高速化拡張「HareSkip」（Forge neo拡張、リポジトリ: Rootport-AI/forge-neo-Anima-HareSkip）の較正実験で生成した全画像・測定値です。固定コミット（b6164214、neo-2.27）で生成した3,186枚のPNG（1536×1536）と、LPIPS/SSIM等の測定CSV、パターン正典、出自記録一式を収録しています。各PNGのinfotextに実現スキップ位置を含む生成条件が完全に埋め込まれており、画像単体で再現性を検証できます。生成モデルはAnima（CircleStone Labs & Comfy Org、NVIDIA Cosmos派生）で、モデル本体は非商用ライセンスですが、同ライセンス§2.eにより生成画像（Outputs）の商用利用を含む自由な利用が明記されているため、本データセット（Outputs＋独自測定値）はCC-BY 4.0で公開します。詳細な分析結果（損傷の序盤集中、予測式のout-of-sample検証、Shift依存性など）は元リポジトリの実験レポートを参照してください。