US 12,242,538 B2
Theme based font search
Akshay Mallipeddi, Cupertino, CA (US)
Assigned to Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed by Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed on Mar. 23, 2023, as Appl. No. 18/188,667.
Prior Publication US 2024/0320268 A1, Sep. 26, 2024
Int. Cl. G06F 16/00 (2019.01); G06F 16/9032 (2019.01); G06F 16/9038 (2019.01); G06F 16/906 (2019.01); G06F 18/2413 (2023.01)
CPC G06F 16/90328 (2019.01) [G06F 16/9038 (2019.01); G06F 16/906 (2019.01); G06F 18/2413 (2023.01)] 16 Claims
OG exemplary drawing
 
1. A method comprising:
training a compact model from a large language model using contrastive learning, wherein the contrastive learning comprises inputting domain-specific training data comprising sentences extracted from design templates to the large language model to generate, for each extracted sentence, a domain-specific sentence embedding;
inputting a font category to the compact model to generate a font tag embedding;
inputting a search query to the compact model to generate a query embedding;
determining a distance measure between the query embedding and the font tag embedding; and
responsive to the distance measure satisfying a predetermined threshold, providing the font category for the font tag embedding as a suggested font category.