Đồ án tốt nghiệp - Document processing and querying using large language models

Tên môn học: Khóa luận tốt nghiệp

Giá: 88.000đ

Tổng thanh toán:

88.000đ

Người đăng tải:

#10377

Chi tiết tài liệu :

Đồ án tốt nghiệp ngành Công nghệ thông tin - Document processing and querying using large language models - Trường Đại học Sư phạm Kỹ thuật Thành phố Hồ Chí Minh (HCMUTE)

Định dạng	PDF
Dung lượng	6.25 MB
Số trang	128
Ngành	Công nghệ thông tin

Lượt xem	113
Ngày đăng	25/12/2025

HO CHI MINH CITY UNIVERSITY OF TECHNOLOGY AND EDUCATION
FACULTY OF INTERNATIONAL EDUCATION

GRADUATION PROJECT

DOCUMENT PROCESSING AND QUERYING
USING LARGE LANGUAGE MODELS

DINH TAN DUNG
Student ID: 20110357
Major: INFORMATION TECHNOLOGY
Supervisor: TRAN NHAT QUANG, PhD.

Ho Chi Minh City, September 2024

HO CHI MINH CITY UNIVERSITY OF TECHNOLOGY AND EDUCATION
FACULTY OF INTERNATIONAL EDUCATION

GRADUATION PROJECT

DOCUMENT PROCESSING AND QUERYING
USING LARGE LANGUAGE MODELS

DINH TAN DUNG
Student ID: 20110357
Major: INFORMATION TECHNOLOGY
Supervisor: TRAN NHAT QUANG, PhD.

Ho Chi Minh City, September 2024

THE SOCIALIST REPUBLIC OF VIETNAM
Independence – Freedom– Happiness

-Ho Chi Minh City, September 14, 2024

GRADUATION PROJECT ASSIGNMENT
Student name:

Student ID: 20110357

Dinh Tan Dung
Major:

Class: 20110CLA1

Information Technology
Supervisor:

Phone number: 0856037586

Tran Nhat Quang.PhD
Date of assignment:

Date of submission:

23/09/2024

01/12/2024

1. Project title:

DOCUMENT PROCESSING AND QUERYING
USING LARGE LANGUAGE MODELS
2. Initial materials provided by supervisor: None.
3. Content of the project:

ACKNOWLEDGEMENT
LIST OF FIGURES
LIST OF TABLES
ABSTRACT
CHAPTER 1: INTRODUCTION
1.1. PROBLEM BACKGROUND
1.2. RESEARCH PROBLEM
1.3. OBJECTIVES AND CONTRIBUTIONS
1.3.1. Objectives
1.3.2. Contributions
1.3.3. Report structure

CHAPTER 2: RELATED WORK AND OTHER PLATFORMS
2.1. LocalGPT - THE BASE SOURCE CODE FOR THIS PROJECT
2.1.1. Problems
2.2. ChatGPT
2.3. GOOGLE GEMINI
2.4. LM STUDIO
CHAPTER 3: METHODOLOGY
3.1. RETRIEVAL AUGMENTED GENERATION (RAG)
3.1.1. Introduction about RAG
3.1.2. RAG Architecture
3.1.3. Advantages of RAG
3.1.4. Challenges and solutions
3.2. EMBEDDING AND VECTORIZATION
3.3. PROMPT ENGINEERING
3.4. INFORMATION RETRIEVAL (IR)
CHAPTER 4: IMPLEMENTATIONS
4.1. FUNCTIONAL DEION
4.1.1. LLM model ion function and prompt to use
4.1.2. Chat Question Function
4.1.3. Chat Data Storage, Loading and Exporting Functions
4.1.4. Ability to handle multiple file types
4.1.5. Deploy Docker
4.2. HARDWARE SYSTEM REQUIREMENT
4.2.1. Reviews
4.2.2. Detailed deion on hardware requirements
4.2.3. Hardware Performance Evaluation Criteria
4.3. DEPLOY DOCKER FOR PROJECT
4.3.1. Literature Review
4.3.2. Methodology
4.3.3. Implementation
4.4. UI/UX DEVELOPMENT
4.4.1. Literature Review
4.4.2. Methodology
4.4.3. Implementation.
4.5. CONFIGURATION METRIC FOR RAG SYSTEM
CHAPTER 5: EVALUATION

5.1. CHOOSING L
...

...
iet Physics Doklady, 10(8), 707–710.
NirDiamant/RAG_Techniques: This repository showcases various advanced techniques for
Retrieval-Augmented Generation (RAG) systems. RAG systems combine information
retrieval with generative models to provide accurate and contextually rich ... (n.d.). GitHub.
Retrieved December 20, 2024, from https://github.com/NirDiamant/RAG_Techniques

125

APPENDIX: INSTALLATION INSTRUCTIONS
Step

Action

1. Install
Python

Download
Python

- Go to the Python homepage.
- Download the latest LTS release version of Python.

Run the
Installer

- Check “Add Python to PATH variable” to enable
terminal access.
- Check “Disable path length limit” for longer file
paths.
- Complete installation and uncheck View Release
Notes.

2. Install
Anaconda

Install
Anaconda
Navigator

- Download and install Anaconda Navigator (Anaconda
3) for virtual environment management.

3. Install Tools

Visual Studio
IDE

- Download Visual Studio IDE.
- the “Desktop Development with C++” package
during installation.

CUDA Toolkit

- Download CUDA Toolkit 11.8 from the NVIDIA
website.

4. Prepare
System

Details

CuDNN
Library

- Download CuDNN version 8.9.5, which is compatible
with CUDA Toolkit 11.X.

Install
Development
IDE

- Install PyCharm (preferred) or Visual Studio Code.
PyCharm is recommended for its utilities.

Install Git

- Download and install Git for version control and
accessing the source code.

Restart
Computer

- Restart your computer after installing Git to apply
changes.

Project
Folder

- a folder on a drive other than C:/.
- Open the folder in Windows Explorer.
- Type “cmd” in the search bar to open Command
Prompt.

126

5. Clone
Source Code

Clone
Repository

- Use the command:
git clone [Repository_URL]

6. Set Up
Environment

Initialize
Anaconda
Environment

- Open Anaconda Prompt.
- Run:
conda -n localgpt_llama2_gpu python=3.10
conda activate localgpt_llama2_gpu

7. Install
Libraries

Install
Dependencies

- Install required libraries using the provided
requirements file:
pip install -r requirements.txt

8. Verify Setup

Test
Environment

- Run a test file to verify setup. Expected results:
python -c import torch
print(torch.cuda.is_available())

9. Run
Application

Prepare
Documents

- Place files to query in the SOURCE_DOCUMENT
folder.

Embed Data

- Use the command:
python ingest.py device_type cuda

Start
Application

- Run the interface with:
streamlit run_localgptUI.py

127