| --- |
| language: |
| - id |
|
|
| license: apache-2.0 |
| task_categories: |
| - text-generation |
| - question-answering |
| size_categories: |
| - 100M<n<1B |
| tags: |
| - math |
| - reasoning |
| - multilingual |
| - cot |
| - synthetic |
| - lumina |
| - foundations |
| - grandmaster-math |
| --- |
| |
| # Lumina-Math-Foundations-200M |
|
|
| **Lumina-Math-Foundations-200M** is an industrial-scale multilingual foundational mathematical reasoning dataset engineered for high-capacity foundation model pretraining, continued pretraining, and reasoning alignment. |
|
|
| ### Supported Languages: |
| - Indonesian (`id`) |
| - English (`en`) *(Upcoming)* in v2 |
| - Korean (`ko`) *(Upcoming)* in v2 |
| - Chinese (`zh`) *(Upcoming)* in v2 |
| - Japanese (`ja`) *(Upcoming)* in v2 |
|
|
| All records are stored as Apache Parquet files in the `data/` directory. |
|
|
