Med-NLI-TR dataset is obtained by translating the sentences of MedNLI into Turkish using GPT4.1.
Dataset Structure
The original dataset only had train and test splits. We applied the following splitting methodology to obtain validation and test splits:
If a train-val-test split is available, we use the existing divisions as provided.
For datasets with a train-test split only, we create a val split from the… See the full description on the dataset page: https://huggingface.co/datasets/boun-tabilab/Med-NLI-TR.
