Accenture Global Solutions Limited

Sankar Narayan Das

Kuntal Dey

Vikrant Shyamkant Kaulgud

Sanjay Podder

Giuseppe Giordano

Kapil Singi

Luca Schiatti

Adam Patten Burden

Andrea Maria Milazzo

Andrew Francis Hickl

Swaminathan Gopalakrishnan

Sharath Haikadi Vasudeva Achar

Srushti Ajay Gajbhiye

Amal Jose

Anuj Gupta

Raman Gupta

A retraining monitoring system maintains the sustainability of a production machine learning (ML) model system that includes a production ML model retraining platform. The retraining monitoring system collects contextual data from the production ML model system and determines if one or more of a currently-selected architectural options has to be changed for sustainability. An architectural option of the production ML model retraining platform, such as, a processing location is selected from a cloud retraining platform or an on-premises retraining platform by a selection process based on a multi-armed bandit problem. An evaluation of the retraining architecture is dealt with as a reinforcement learning problem to implement one of a periodic retraining architecture or a reactive retraining architecture.

Sustainable retraining for production machine learning

The present disclosure recites methods and systems for context augmentation. The method includes generating, in accordance with a policy of a RAG system, a first set of queries for a chunk of a plurality of chunks, the plurality of chunks represents a context of the RAG system. Further, the method includes determining, based upon a set of quality indicators, a quality class of each query of the first set of queries. Thereafter, the method includes, updating, the chunk to an updated chunk to further include query data of one or more queries of the first set of queries based upon the quality class of the each query of the first set of queries. Further, the method includes storing, the updated chunk in a database as a vector for retrieval from the RAG system upon receiving a prompt.

Method and system for context augmentation in retrieval augmented generation

A method and system for cloud distributed hybrid data storage and normalization are disclosed. The method may include obtaining a data set comprising data entities. A data entity may comprise data fields each containing a data element. The method may further include determining policy constraint meta-data for each of the data elements based on the storage policy constraint. The policy constraint meta-data may include a first meta-tag indicating the storage policy constraint for the data element. The method may further include determining whether a server satisfies the storage policy constraint based on the first meta-tag for the data element. When the server satisfies the storage policy constraint, the method may further include transmitting the data element to the server to store the data element on the server. When the server fails to satisfy the storage policy constraint, the method may further include, storing the data element on the client.

Cloud distributed hybrid data storage and normalization

Growing community of inventors

Barrackpore, India

Sankar Narayan Das