Mixed-STS-tr / README.md
DolphyWind's picture
Add dataset splits
5a12258 verified
|
Raw History Blame Contribute Delete
688 Bytes
metadata
task_categories:
  - sentence-similarity
language:
  - tr
tags:
  - sts
  - tr
  - turkish
  - sts-tr
  - semantic-textual-similarity
pretty_name: Mixed STS
size_categories:
  - 1M<n<10M
configs:
  - config_name: default
    data_files:
      - split: train
        path: mixed-sts-train.parquet
      - split: validation
        path: mixed-sts-validation.parquet
      - split: test
        path: mixed-sts-test.parquet

This dataset is a combination of STSb-tr, CxC-sts-mt-tr (80-20 split) and all-nli-tr (pair-score split).