US Patent Issued to Salesforce on Jan. 13 for "Systems and methods for ensembling soft prompts in few-shot fine-tuning of language models" (California, Georgia Inventors)

ALEXANDRIA, Va., Jan. 13 -- United States Patent no. 12,524,613, issued on Jan. 13, was assigned to Salesforce Inc. (San Francisco).

"Systems and methods for ensembling soft prompts in few-shot fine-tuning of language models" was invented by Xiangyu Peng (Atlanta), Chen Xing (Palo Alto, Calif.), Prafulla Kumar Choubey (San Jose, Calif.) and CHien-Sheng Wu (Mountain View, Calif.).

According to the abstract* released by the U.S. Patent & Trademark Office: "Embodiments described herein provide a mechanism that ensembles trainable soft prompts to transfer knowledge from source tasks under few-shot learning settings. Specifically, given a source task input from a source task training dataset, a set of soft prompts may be trained using a frozen...