Hate Speech Detection

Comparing performance of different Large Language Models in analysing tweets containing Hate Speech. Comparing with traditional NLP based techniques to Identify Hate Speech.

Models Used:

• deepseek-r1:1.5b

Dataset Used:

• Hate Speech and Offensive Language Dataset

Installation:

Install and run Ollama Docker container:
Follow the steps provided by the official Ollama documentation https://hub.docker.com/r/ollama/ollama It is preferable to run Ollama with a GPU since we will send thousands of requests to the container.

Test Installation:
Visit http://localhost:11434 to test if ollama is running, you should see the message, ollama is running.

Run the Model:
Find the model you want to run and run it inside the container. To enter the interactive terminal of the container use:

 docker exec -it /bin/bash

Here run the command:

ollama run <Model-Name>

This will automatically pull and run the model.

Note: Please Ensure you are using an updated version of the Ollama container, or some models may not run.

Changes:
Change the model name in test_model.py to test the chosen model. You can run the model on the larger dataset provided at: https://www.kaggle.com/datasets/mrmorj/hate-speech-and-offensive-language-dataset or on the short dataset included here.