---
language:
- tr
---
This datasets contains 9623924 sentences / chunks in Turkish to finetune sentence similarity models. It's combination of Altai's legal datasets as well as [this dataset](https://huggingface.co/datasets/hcsolakoglu/turkish-wikipedia-qa-4-million) which contains raw Wikipedia articles and generated question-answers in Turkish.
After all chunks were concatenated in a single list, `MarkdownTextSplitter` was applied with `chunk_size=1000` and `overlap=200`.