US 12,332,074 B1
Machine-learned foundational models for emulating conversational interactions between points of interest
Victor Carbune, Zurich (CH); Kevin Allekotte, Zurich (CH); Haroon Baig, Zurich (CH); Paula Marques Fernandes, Zurich (CH); and Matthew Sharifi, Kilchberg (CH)
Assigned to GOOGLE LLC, Mountain View, CA (US)
Filed by Google LLC, Mountain View, CA (US)
Filed on Apr. 19, 2024, as Appl. No. 18/640,952.
Int. Cl. G01C 21/36 (2006.01); H04L 51/02 (2022.01)
CPC G01C 21/3682 (2013.01) [H04L 51/02 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
obtaining, by a computing system comprising one or more processor devices, information indicative of a particular geographic area in which a user computing device is located, wherein the particular geographic area comprises a plurality of Points of Interest (POIs);
for a first POI of the plurality of POIs:
applying, by the computing system, a tuning process to a first instance of a Large Foundational Model (LFM), wherein the first machine-learned POI-specific language model comprises the first instance of the LFM trained at least in part to perform multiple types of language tasks, and wherein the tuning process is configured to cause the first instance of the LFM to generate textual content from a perspective of the first POI; and
generating, by the computing system, a first set of textual content with the first machine-learned POI-specific language model configured to generate textual content from the perspective of the first POI;
for a second POI of the plurality of POIs:
generating, by the computing system, a second set of textual content with a second machine-learned POI-specific language model configured to generate textual content from a perspective of the second POI; and
providing, by the computing system, at least some of the second set of textual content to the user computing device.