Ploba logo

Find the right AI tool for the job: apps, agents, APIs, MCP servers and datasets, explained plainly.

Get the AI Digest

Discover

  • Explore
  • Top Charts
  • Categories
  • Agents
  • MCP servers
  • Developer tools

Learn & community

  • Learn AI
  • AI Glossary
  • Blog
  • Hackathons
  • People
  • AI Digest

For makers

  • List your tool
  • Docs
  • Claim a listing
  • Advertise
  • Write for us

Company

  • About
  • Careers
  • Changelog
  • Trust
  • Security
  • Contact

© 2026 Ploba. All rights reserved.

  • Privacy
  • Terms
  • Cookies
  • Advertising
  • All policies
llms.txt
Ploba logo
DiscoverHackathonsPeopleLearnAI Digest
Sign inGet started
HomeTop ChartsExploreCategories
TypesWeb apps1,894Agents585Plugins416APIs376Mobile apps152Datasets85
AdvertisingYour product hereReach people browsing AI tools on Ploba.Advertise hereAdvertisingYour product hereReach people browsing AI tools on Ploba.Advertise here
  1. Explore
  2. /
  3. AI
  4. /
  5. Spark 234K
AI
S2

Spark 234K

by OpenDataArena

Spark-234K is a dataset of 234K scientific reasoning question-answer pairs from OpenDataArena, built from frontier research papers. It is meant for training language models on hard, self-contained science problems.

Dataset·Free

View on Hugging Face(opens in a new tab)

About Spark 234K

Key features of Spark 234K

  • 234K science question-answer pairs
  • Built from about 370K sampled frontier papers across 10 disciplines
  • Questions come from reasoning skeletons distilled from each paper
  • Answers checked for self-containment, source consistency and reasoning grounding
  • Semantic deduplication and 13-gram decontamination against benchmarks
  • Human audit of 300 samples reports 90.3% agreement with experts

Who Spark 234K is for

  • Researchers training scientific reasoning models
  • Teams doing supervised fine-tuning for science and math tasks
  • Dataset builders comparing synthetic data pipelines

How to use Spark 234K

  1. 1

    Step 1: Open the dataset page on Hugging Face and review the card and the cited paper.

  2. 2

    Step 2: Load the data with the Hugging Face datasets library using the OpenDataArena/Spark-234K identifier.

  3. 3

    Step 3: Inspect a sample of the question-answer pairs across disciplines before training.

  4. 4

    Step 4: Use the pairs as supervised fine-tuning data for a reasoning model.

  5. 5

    Step 5: Cite the SPARK paper listed on the card if you use the data in published work.

Spark 234K FAQ

What is in Spark-234K?

It holds 234K question-answer pairs about scientific topics, generated from frontier research papers across 10 disciplines. Each item is meant to be self-contained.

How was the data checked?

Each question-answer triple is verified for self-containment, source consistency, answer coherence and reasoning grounding. A human audit of 300 samples reports 90.3% agreement with expert answers.

Does it avoid benchmark contamination?

The pipeline includes semantic deduplication per discipline and 13-gram exact-match decontamination against downstream benchmarks.

Where do the questions come from?

They come from a collection of about 3.36M papers called SCI-BASE, of which about 370K frontier papers are sampled after filtering and discipline balancing.

Details

Pricing
Free
Type
Dataset
Categories
AIDataResearch
Added
Sep 17, 2026
OpPublisherOpenDataArena

Built from public sources

We put this page together from Spark 234K’s own website and docs so you can compare it with other tools. Ploba isn’t affiliated with Spark 234K.

Work at Spark 234K? Claim the listing to manage it yourself.

Something wrong here?

Related tools

See more
FineWeb iconFineWebHugging Face · FreeFineWeb is Hugging Face's dataset of more than 18.5T tokens of cleaned and deduplicated English web text from Common Crawl. It is meant for pre-training language models.AI · FreeQoQokedasQokedas · Contact for pricingTraining-data company that turns real-world scientific signals and observations that are not normally written down into datasets for improving AI models in science.AI · Contact for pricingUnUniPhys-BenchSpatialVerse · FreeUniPhys-Bench is a human-verified benchmark of articulated 3D objects annotated with joint motion and physical properties. This repository holds the primary release of 1,473 objects.AI · Free