SKU SFT-2307 · Sold by External
OpenThoughts-114k
Product specifications
| SKU | SFT-2307 |
|---|---|
| Data type | SFT dataset |
| Volume | 114000 rows |
| Size on disk | 3.55 GB |
| Format | Parquet (Hugging Face) |
| Access model | PUBLIC LICENSE |
| Pricing | Free · open-source license |
| Quality score | — |
| License | Apache-2.0 |
OpenThoughts-114k is a synthetic reasoning SFT dataset of ~114,000 high-quality examples covering math, science, code and puzzles, distilled from DeepSeek-R1 and verified (OpenThoughts paper arXiv:2506.04178). It ships two configs: 'default' (system + conversations) and 'metadata' (adds problem, ground_truth_solution, deepseek_reasoning/solution, domain, source, plus test_cases/starter_code for code). Chain-of-thought reasoning only; no tool/function-call traces.