This model is based on COMET-early-exit, which is a fork but not compatible with original Unbabel's COMET.
To run the model, you need to first install this version of COMET either with:
This model specifically behaves like standard quality estimation, but outputs two numbers: scores (as usual) and confidences, which is the estimated absolute error from the human score.
Thus, contrary to expectations, higher "confidence" correponds to less correct QE estimation.
python
1model = comet_early_exit.load_from_checkpoint(comet_early_exit.download_model("zouharvi/COMET-instant-confidence"))2data =[3{4"src":"Can I receive my food in 10 to 15 minutes?",5"mt":"Moh bych obdržet jídlo v 10 do 15 minut?",6},7{8"src":"Can I receive my food in 10 to 15 minutes?",9"mt":"Mohl bych dostat jídlo během 10 či 15 minut?",10}11]12model_output = model.predict(data, batch_size=8, gpus=1)13print("scores", model_output["scores"])14print("estimated errors", model_output["confidences"])1516assertlen(model_output["scores"])==2andlen(model_output["confidences"])==2
@misc{zouhar2025earlyexitinstantconfidencetranslation,
title={Early-Exit and Instant Confidence Translation Quality Estimation},
author={Vilém Zouhar and Maike Züfle and Beni Egressy and Julius Cheng and Jan Niehues},
year={2025},
eprint={2502.14429},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2502.14429},
}