公开(公告)号：US20220392434A1

公开(公告)日：2022-12-08

申请号：US17342490

申请日：2021-06-08

Applicant: Microsoft Technology Licensing, LLC

Inventor： Abedelkader ASI ,  Yarin KUPER ,  Royi RONEN ,  Song WANG ,  Olga GOLDENBERG ,  Shimrit Rada BEMIS ,  Erez ALTUS ,  Yi MAO ,  Weizhu CHEN

IPC: G10L15/06 ,  G06N20/00

Abstract: The disclosure herein describes reducing training bias in outputs generated by a generative language model. A communication segment associated with a communication is obtained by at least one processor of a generative language model. An output value associated with the communication segment is generated by the generative language model. The output value is mapped to a set of training bias values associated with the generative language model and based on the mapping of the output value to a training bias value of the set of training bias values, an alternative output value is generated. The alternative output value is used in a generated segment output for the communication segment. The accuracy of segment outputs generated by the generative language model is improved through reducing or eliminating its training biases.

公开(公告)号：US20240193350A1

公开(公告)日：2024-06-13

申请号：US18064218

申请日：2022-12-09

Applicant: Microsoft Technology Licensing, LLC

Inventor： Yujia XIE ,  Lesly Sadiht MICULICICH WERLEN ,  Song WANG ,  Pengcheng HE ,  Yuantao WANG ,  Wei XIONG ,  Yanling XIONG

IPC: G06F40/166 ,  G06F40/117 ,  G06F40/284 ,  G06F40/47 ,  G06N20/00

CPC classification number: G06F40/166 ,  G06F40/117 ,  G06F40/284 ,  G06F40/47 ,  G06N20/00

Abstract: The disclosure herein describes training a text processing model to generate model output text data using input text data and a sentence count. A training data entry including input text data and output text data is obtained. A sentence count of the output text data is determined, and the output text data is labeled with a sentence count label and a sentence number label. Model output text data is generated with a text processing model using the input text data and determined sentence count as input data. Loss data associated with a difference between the generated model output text data and the labeled output text data is determined and the text processing model is adjusted using the determined loss data. The use of labeled output text data enables the model to be trained to produce output text data with a target sentence count in a computationally efficient manner.

公开(公告)号：US20240365427A1

公开(公告)日：2024-10-31

申请号：US18307702

申请日：2023-04-26

Applicant: Microsoft Technology Licensing, LLC

Inventor： Manikanta KOTARU ,  Song WANG ,  Bozidar RADUNOVIC ,  Xenofon FOUKAS

IPC: H04W76/28 ,  H04L65/60 ,  H04W28/02

CPC classification number: H04W76/28 ,  H04L65/60 ,  H04W28/0247

Abstract: The present disclosure generally relates to improving energy efficiency of reality-based headsets by enabling discontinuous transmission (DRX) without significantly reducing the quality of experience (QoE) of a user of the headsets when consuming digital content presented thereon. The present disclosure includes a predictive content management system that obtains consumable content (including predictive content) to be consumed on a wearable UE. The system receives, from a radio access network (RAN), discontinuous reception (DRX) configuration information indicating active and inactive periods with which the UE and RAN may communicate. The system additionally facilitates stacking consumable content, delivering the consumable content based on the DRX configuration information, and performing certain latency masking features to avoid loss of quality of the presented content.