Views
No views yet
pip install git+https://github.com/prosho-97/guardians-mt-eval1from sentinel_metric import download_model, load_from_checkpoint
2
3model_path = download_model("Prosho/sentinel-src-25")
4model = load_from_checkpoint(model_path)
5
6data = [
7 {"src": "Please sign the form."},
8 {"src": "He spilled the beans, then backpedaled—talk about mixed signals!"}
9]
10
11output = model.predict(data, batch_size=8, gpus=1)1# Segment scores
2>>> output.scores
3[0.5604351758956909, -0.08413456380367279]
4
5# System score
6>>> output.system_score
70.238150306046009061@inproceedings{proietti-etal-2025-estimating,
2 title = "Estimating Machine Translation Difficulty",
3 author = "Proietti, Lorenzo and
4 Perrella, Stefano and
5 Zouhar, Vil{\'e}m and
6 Navigli, Roberto and
7 Kocmi, Tom",
8 editor = "Christodoulopoulos, Christos and
9 Chakraborty, Tanmoy and
10 Rose, Carolyn and
11 Peng, Violet",
12 booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
13 month = nov,
14 year = "2025",
15 address = "Suzhou, China",
16 publisher = "Association for Computational Linguistics",
17 url = "https://aclanthology.org/2025.findings-emnlp.1317/",
18 doi = "10.18653/v1/2025.findings-emnlp.1317",
19 pages = "24261--24285",
20 ISBN = "979-8-89176-335-7"
21}