Efficient inner product operations
This Google patent describes a system and method designed to significantly improve the speed and accuracy of finding items within large data sets using a hybrid vector space inner-product search. read more
Hierarchical quantization for fast inner product search
The Google patent describes an efficient system and method for rapidly calculating inner products, particularly between high-dimensionality vectors, which is a core problem in Maximum Inner Product Search (MIPS). It read more
Using large generative models with improved grounding to improve image context queries
The Micorosoft patent describes an image query system that improves the accuracy and efficiency of responding to natural language queries about images. It achieves this by using a combination of read more
Real time retrieval of content items for multi-category synthesis and personalized knowledge augmentation
This Microsoft patent describes a method for providing synthesized and personalized supplemental information to a user. It involves using a Large Language Model (LLM) to determine categories and keywords from read more
Graph-R1: Towards Agentic GraphRAG Framework via End-to-end Reinforcement Learning
This research paper document introduces Graph-R1, a new framework that addresses the limitations of current Retrieval-Augmented Generation (RAG) systems. It utilizes an agentic approach with end-to-end reinforcement learning to improve read more
Quantization-based fast inner product search
This Google patent outlines an efficient system for approximating the inner product between a query and a large database of items, which are represented as vectors. The method addresses the read more
Generating content labels according to user search queries
This Google patent describes a method for generating content labels, such as titles or hashtags, for a topic based on user search queries. The process involves retrieving terms associated with read more
Management of indexed data to improve content retrieval processing
This Micorosoft patent describes a method for managing indexed data to enhance content retrieval, particularly when dealing with large datasets. It introduces techniques like generating and managing data shards to read more
Direct Reasoning Optimization: LLMs Can Reward And Refine Their Own Reasoning for Open-Ended Tasks
This document introduces a new method called Direct Reasoning Optimization (DRO), which is designed to improve how large language models (LLMs) perform on tasks that are more creative and open-ended. read more
Systems and methods for providing reliable information for queries
The patent describes systems and methods designed to provide users with reliable, expert-verified information. When a user submits a query, the system identifies the specific field of expertise, searches a read more