CoolFace
Datasetpublic

zouhar/wmt-human-all

This dataset is continuously updated and contains a compilation of human translation quality assessment from past WMT campaigns. Specifically, this dataset merges all annotation protocols (DA, MQM, ESA) on a semi-unified scale (0 to 100). The current version of the dataset includes human scores up to WMT 2025 (inclusive) and has been created with the following script: import subset2evaluate # version 1.0.20 import json import statistics data =… See the full description on the dataset page: https://huggingface.co/datasets/zouhar/wmt-human-all.

sourceHugging Faceapache-2.0updated 8mo agoView on Hugging Face
1likes109downloads
6 commits on main
dd8fa818mo ago

add WMT25

Vilém Zouhar
f5d30188mo ago

Upload folder using huggingface_hub

Vilém Zouhar
778585c1y ago

Upload folder using huggingface_hub

Vilém Zouhar
61115d71y ago

Upload folder using huggingface_hub

Vilém Zouhar
86de10c1y ago

Upload folder using huggingface_hub

Vilém Zouhar
e3b4c391y ago

initial commit

Vilém Zouhar