This repo contains some of my personally curated projects focused on Large Language Models(LLMs), particularly related to open-source models. These projects encompass frameworks for LLM training, deployment tools, tutorials, and practical applications.
The applications represent my original ideas, which I conceive, plan, develop, and test in real-world scenarios. They serve as a feedback loop, allowing me to gain experience in creating innovative ideas and testing various LLM models. Additionally, I utilise freely available platforms like Google Colab, Kaggle, and others to develop these projects.
Mini-projects: Developed real-world applications of Generative AI. Text, speech, and images are generated using multi-modal models. These applications have been used in various real-life scenarios and are freely accessible to the public, allowing them to enjoy and share with friends and family. These are some of the projects:
- Bloomsday Audio Tour This project focuses on creating an audio-guided tour of one of the most challenging books to read, offering a unique and practical use case for Generative AI in the travel and tourism industry. The experience is designed to engage readers and travelers alike, bringing a classic and beloved book to life.
- Optimizing Data Scientist Resume for ATS: Leveraging NLP and LLMs Put together a resume for an experienced data scientist that breezes through the automatic tracking system (ATS). This resume makes use of both traditional NLP techniques and the latest LLMs to highlight your skills in a natural and engaging way.
- Whodunit Lunch Party Development of a murder mystery role-playing game that garnered a 4.7-star rating, employing LLMs for its creation. The game’s immersive plot, intriguing characters, and cleverly crafted clues were designed to enhance the gaming experience. The project uses a diverse tech stack that includes Python, ChatGPT, Dall-e, open-source models, and Gemini.
- Immersive Audio Tour: National Gallery of Ireland Adapted and optimised open-source LLMs from platforms like Hugging Face to create innovative solutions e.g., developed a guided museum tour using a generative text-to-audio model, which received a 4.7-star rating from attendees.
- ChatGPT-Driven Scavenger Hunt: Doors of South Dublin’s Georgian Architecture Developed a scavenger hunt game aimed at locating and photographing numerous famous doors of Dublin within a team setting. Utilised Google Maps, LLMs, and creative problem-solving techniques to execute the project effectively.
- Treasure Hunt: Connect with People, Socialise, Be Creative and Have Fun Created a treasure hunt game at St. Stephen’s Green Park to foster connections, encourage socialisation, stimulate creativity, and promote fun within a group. Leveraged Python, LLMs, and innovative problem-solving strategies.
- Tutorials Notebooks and short courses I completed to deepen my understanding of LLMs. These cover topics such as HuggingFace, Similarity Search, Sentence Transformers, and Generative AI models.
If you have any questions, feedback, or would like access to the code used in these projects, feel free to reach out to me. I'm here to assist and share the knowledge I've gained during this exciting journey.