Intellectual Property India Publishes Patent Application for 'Methods And Systems For Generating Task-Specific Output Using A Single Quantized Model Graph' Filed by Samsung Electronics Co. Ltd.

MUMBAI, India, April 20 -- Intellectual Property India has published a patent application (202441054180 A) filed by Samsung Electronics Co. Ltd., Gyeonggi, Republic of Korea, on Oct. 16, 2024, for 'methods and systems for generating task-specific output using a single quantized model graph.'

Inventor(s) include Miriyala, Srinivas Soumitri; Mishra, Aakansha; R, Prasanna; Anand, Payal; Hosakoppa, Manohara Krishnamurthy; Naidu, Praveen Doreswamy; and Mala, Venkappa.

The application for the patent was published on April 17, under issue no. 16/2026.

According to the abstract released by the Intellectual Property India: "Disclosed is a method (1200) for generating a task-specific output using a single quantized model graph for a plurality of Parameter Efficient Fine Tuning (PEFT) models in a generative Artificial Intelligence (AI) system. The method includes obtaining (1202) a base model pre-trained for a plurality of generative AI tasks to be performed by the plurality of PEFT models. The method includes performing (1204) a quantization sensitivity analysis across the plurality of PEFT models to obtain a Quantization Sensitivity Score (QSS) and selecting (1206) a quantization-sensitive PEFT model based on the corresponding QSS. The method further includes determining (1208) a fixed quantization configuration based on the selected quantization-sensitive PEFT model. The method includes adjusting (1210) one or more weights of each of the plurality of PEFT models, generating (1212) the single quantized inference graph incorporating the base model and enabling dynamic selection of any of the adjusted plurality of PEFT models."

Disclaimer: Curated by HT Syndication.