datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
omnidocbench-render-compare
OmniDocBench Render-and-Compare
This dataset contains the rendered HTML reconstructions and comparison images produced
by a render-and-compare pipeline — a reference-free visual similarity evaluation
framework for OCR systems.
Overview
The pipeline processes each page of OmniDocBench through
a Qwen3.5-122B-A10B OCR model, renders the structured output back to a PNG via HTML
(reconstructed.png), and compares it against the original page scan (masked_original.png)
using… See the full description on the dataset page: https://huggingface.co/datasets/gt-free-ocr-metrics/omnidocbench-render-compare.korean-text-rendering-data
한글 텍스트 렌더링 학습 데이터
이미지 안에 정확한 한글 텍스트를 렌더링하는 능력 개선을 위해 만들어진 합성(synthetic) 이미지-프롬프트 데이터셋입니다. 2026년 5월~7월에 걸쳐 진행된 세 차례의 별도 학습 이터레이션에서 나온 데이터를 통합했습니다.
총 79,460장, 2개 config(콘텐츠 유형)로 구성. 각 config는 독립적으로 로드할 수 있습니다.
from datasets import load_dataset
ds = load_dataset("<repo_id>", name="diagram") # 유형별로 필요한 것만
이 릴리즈는 순수 한글 타이포그래피 학습에 초점을 맞춰 atomic_text(99.4% 한글)와
diagram(100% 한글) 두 유형만 포함합니다. 둘 다 코드·템플릿 기반 결정론적 생성이라
외부 생성형 서비스에 의존하지 않고, 라이선스 문제가 없습니다. "프롬프트 안 인용부호=정답
텍스트" 컨벤션은 둘 다… See the full description on the dataset page: https://huggingface.co/datasets/fasoo/korean-text-rendering-data.metrixel-character-renders
Metrixel Animated Character Renders
Multi-view renders, signed-distance-field volumes and per-view mesh tensors produced by Metrixel from a small set of rigged, animated humanoid characters.
Each character is captured from four camera angles (0°, 90°, 180°, 270°) across ~30 sampled frames of its motion clip, at 512×512. Every frame/angle carries a matching 64³ signed-distance-field volume and a per-view mesh tensor, so the geometry, the implicit surface and the image are aligned… See the full description on the dataset page: https://huggingface.co/datasets/EntVista/metrixel-character-renders.anny-render-corpus-train
anny-render-corpus
A camera-controlled render corpus from the ANNY rig, and the measurements that motivated it.
Code: weftspun/anny-render-corpus, on the 6-datasource side of the hexagon.
Everything here is produced by scripts in that repository and can be regenerated from it.
What this is for
Asked in plain language for eight camera azimuths, OmniGen2 returns a body that does not
turn. Recovered azimuth tracks the request with a slope of 0.04, where 1.00 is… See the full description on the dataset page: https://huggingface.co/datasets/chibifire/anny-render-corpus-train.Rendered_512_32_TestVoluspa_the_Seeresss_Vision_the_Ultimate_Poetic_Rendering
Völuspá the Seeress's Vision the Ultimate Poetic Rendering
Dataset Details
Dataset Description
This dataset weaves the ancient threads of Völuspá, the Seeress's profound vision from the Poetic Edda, into a modern ShareGPT JSONL format. It holds 66 sacred exchanges, each a rune-carved conversation: a mortal seeker requests the recitation of a stanza, and the divine voice responds with the poem's eternal words—from creation's dawn through Ragnarök's… See the full description on the dataset page: https://huggingface.co/datasets/RuneForgeAI/Voluspa_the_Seeresss_Vision_the_Ultimate_Poetic_Rendering.trace-render-xss-probe-0910omnidocbench-render-compare-sample
OmniDocBench Render-and-Compare — Sample
This is a 60-page stratified sample of
gt-free-ocr-metrics/omnidocbench-render-compare
(the full dataset is ~10 GB).
It is provided to help reviewers explore the data without downloading the full dataset,
as recommended by the NeurIPS 2025 Datasets & Benchmarks Track guidelines.
Sampling Methodology
Pages were selected by stratified random sampling from the full dataset:
Each page in ocr_all (1 355 pages) was assigned to one… See the full description on the dataset page: https://huggingface.co/datasets/gt-free-ocr-metrics/omnidocbench-render-compare-sample.function_calling_renderrunopsy-bench
Runopsy-Bench
Twenty labelled agent traces for measuring failure-onset localization: given a run
that went wrong, which step did it start going wrong at — not which step it stopped at.
Produced for Runopsy, an open-source causal failure
analysis engine for agent runs. pip install runopsy.
Read this first: these traces are synthetic
Every case here was generated, not recorded. They are single-fault traces written to
exercise a specific failure mode, with the onset… See the full description on the dataset page: https://huggingface.co/datasets/renderfy/runopsy-bench.DSLR-FullQueryPairs