MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202611100654 A) filed by Ajay Kumar Garg Engineering College on August 20, 2026, for Low-Resource Language Model Adaptation System.
Inventors include Ms. Vaishali Deshwal; Aarya Shresth Kapoor; Ansh Gupta; and Aryan Mishra.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: Abstract Disclosed is a system for adapting and providing a language model for a low-resource language. The system comprises a memory storing a pretrained language model comprising quantized model parameters, trainable low-rank adapter parameters associated with the pretrained language model, target-language training data, and a task-specific evaluation dataset. A deployment interface receives language input and provides model-generated output. A processing unit coupled to the memory and the deployment interface retains the quantized model parameters without parameter updating, trains the low-rank adapter parameters using the target-language training data while maintaining the quantized model parameters unchanged to obtain an adapted language model, evaluates the adapted language model using the task-specific evaluation dataset, and processes received language input using the adapted language model to provide corresponding model-generated output through the deployment interface. Fig. 1 Dated August 18, 2026 / Mungalpara Jigneshbhai Chhaganbhai IN/PA-2640 Agent for the Applicant
Disclaimer: Curated by HT Syndication.
