Transfer learning that adapts a model trained on data-rich sources to low-resource targets has been widely applied in natural language processing . When training a transfer model over multiple sources, not everysource is equally useful for the target . To better transfer a model, it isessential to understand the values of the sources . In this paper, we developSEAL-Shap, an efficient source valuation framework for quantifying theusefulness of the source . The framework is not only effective in choosing useful transfer sources but also the source valuesmatch the intuitive source-target similarity .

Author(s) : Md Rizwan Parvez, Kai-Wei Chang

Links : PDF - Abstract

Code :


Keywords : sources - transfer - source - model - target -

Leave a Reply

Your email address will not be published. Required fields are marked *