GPT-2 TwitterBot

Description

My project facilitates the downloading of data from Twitter, enabling users to collect tweets from specific accounts. Subsequently, users can fine-tune a GPT-2 or other large language models using the collected data, allowing for personalized tweet generation. With my tool, users can effortlessly gather Twitter data, fine-tune models, and generate tweets tailored to their preferences.

Languages and Utilities Used

• Python
• PyTorch
• HuggingFace

Environments Used

Program walk-through:

1. Set your access api in auth_tw.py

2. Download tweets.

Use in get_user_tweets.py for downloading tweets. Add accounts that you want to base your generation in this file.

3. Fine-Tune LLM Model( by default GPT-2 )
Use Fine_tuning_GPT2.ipynb notebook for fine-tuning.

4. Use fine-tuned model for tweets generation
Use Content_Generator.ipynb notebook for tweets generation. Example of generated text.

5. Example tweet