This is a 1k sample of the OpenThoughts-114k dataset. Open synthetic reasoning dataset with high-quality examples covering math, science, code, and puzzles! Inspect the content with rich formatting with Curator Viewer.
Dataset Card
This is a 1k sample of the OpenThoughts-114k dataset. Open synthetic reasoning dataset with high-quality examples covering math, science, code, and puzzles! Inspect the content with rich formatting with Curator Viewer. default subset containing ready-to-train data used to finetune the OpenThinker-7B and OpenThinker-32B models: metadata subset containing extra columns used in dataset construction: - problem - groundtruthsolution - deepseekreasoning - deepseeksolution - domain - source - testcases (code only) - startercode(code only) The numbers reported in the tables below are evaluated with our open-source tool Evalchemy. We are fully open-source. Our model weights, datasets, data…
Excerpt from the card by Ryan Marten.
Structure
default 1,000 rows
| Split | Rows | Size |
|---|---|---|
| train | 1,000 | 34.2 MB |
metadata 1,000 rows
| Split | Rows | Size |
|---|---|---|
| train | 1,000 | 28.0 MB |
Details
- Repository
- ryanmarten/OpenThoughts-1k-sample
- Publisher
- Ryan Marten
- Task category
- Not stated by the source
- Tags
- Not stated by the source
- Size category
- Not stated by the source
- Languages
- Not stated by the source
- Revision
- a82400884621626d41bef89b7604f8054e7e00e0
- Last updated
- 2025-08-31
SAVRN's Notes on OpenThoughts-1k-sample
A smoke-test slice, honestly labeled: OpenThoughts-1k-sample is 2,000 rows across two configs, a 1,000-row default set with system and conversation columns plus a 1,000-row metadata config carrying problem, ground-truth and DeepSeek reasoning fields. It is 28.1 MB in 7 files with 1.3M monthly downloads. The card ties it to OpenThoughts-114k and to the OpenThinker-7B and OpenThinker-32B fine-tunes, and its revision is from 2025-08-31.
The license line is the problem: this page states no license at all, so despite the parent project's open-source claims you hold no documented rights for a commercial run until you verify terms on the upstream repo. Use it to validate a reasoning-data pipeline cheaply; production work needs the full 114k set, whose terms you would have to clear separately.
Files
7 files, 28.1 MB in total.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| data/train-00000-of-00001.parquet | Data | 14.0 MB | 32c296a2fb33 |
| metadata/train-00000-of-00001.parquet | Data | 13.3 MB | 235faa2830b5 |
| README.md | Documentation | 8.5 KB | — |
| diagram.png | Other | 471.6 KB | 04699a3588cf |
| diagram_dark.png | Other | 259.6 KB | 274c015b29dc |
| open_thoughts.png | Other | 15.3 KB | 36478f501f8a |
| .gitattributes | Repository | 2.5 KB | — |
License and Download
- License
- Not stated by the source
- Access
- No access gate
Released by Ryan Marten through its official repository on Hugging Face.