A trained QE model for IMPARA, a reference-less performance measure for GEC task.

This model achieves 95.93 for Pearson's correlation and 93.01 for Spearman's, for Grundkiewicz +15's Expected Wins score (Note that bert-base-cased is used for SE model).

You can see the detail in this GitHub repository, e.g. How to use this model.

Downloads last month
82
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Collection including gotutiyan/IMPARA-QE