This repository contains a Dockerized web application that serves as a demo for the research related to the paper "Towards Better Understanding of Cybercrime: The Role of Fine-Tuned LLMs in Translation". The research is about translating Russian hacktivism messages using a Fine-Tuned GPT-3.5 model and comparing it with traditional translation services. The goal is to showcase the differences between translations and highlight the effectiveness of the LLM-based approach in real-world scenarios, such as understanding cybercrime discussions.
Clone the Repository:
:~$ git clone https://github.com/stratosphereips/demo-hacktivism-llm-translation.git demo-llm-translation
:~$ cd demo-llm-translation
:~$ docker build -t llm-translation-demo .
:~$ docker run -d -p 80:80 --name llm-translation-demo llm-translation-demoOpen your web browser and go to http://localhost.
The application displays messages originally written in Russian along with their translated versions. It automatically rotates through different messages every 5 seconds, highlighting the differences in translations from multiple sources.
This tool was developed at the Stratosphere Laboratory at the Czech Technical University in Prague. The Stratosphere Laboratory is committed to advancing research in cybersecurity, machine learning, and open-source tools for the security community.