Saved in:
Bibliographic Details
Main Author: Tel-Zur, Guy
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.11947
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909907511410688
author Tel-Zur, Guy
author_facet Tel-Zur, Guy
contents This project addresses a critical pedagogical need: offering students continuous, on-demand academic assistance beyond conventional reception hours. I present a domain-specific Retrieval-Augmented Generation (RAG) system powered by a quantized Mistral-7B Instruct model and deployed as a Telegram bot. The assistant enhances learning by delivering real-time, personalized responses aligned with the "Introduction to Parallel Processing" course materials. GPU acceleration significantly improves inference latency, enabling practical deployment on consumer hardware. This approach demonstrates how consumer GPUs can enable affordable, private, and effective AI tutoring for HPC education.
format Preprint
id arxiv_https___arxiv_org_abs_2509_11947
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A GPU-Accelerated RAG-Based Telegram Assistant for Supporting Parallel Processing Students
Tel-Zur, Guy
Computers and Society
Artificial Intelligence
This project addresses a critical pedagogical need: offering students continuous, on-demand academic assistance beyond conventional reception hours. I present a domain-specific Retrieval-Augmented Generation (RAG) system powered by a quantized Mistral-7B Instruct model and deployed as a Telegram bot. The assistant enhances learning by delivering real-time, personalized responses aligned with the "Introduction to Parallel Processing" course materials. GPU acceleration significantly improves inference latency, enabling practical deployment on consumer hardware. This approach demonstrates how consumer GPUs can enable affordable, private, and effective AI tutoring for HPC education.
title A GPU-Accelerated RAG-Based Telegram Assistant for Supporting Parallel Processing Students
topic Computers and Society
Artificial Intelligence
url https://arxiv.org/abs/2509.11947