for Prototype for GPT-AI with Weconomics datasets
Open Learning
Semester programme:Open Learning/Innovation
Client company:Weconomics
Marek Mitala
Radja Dalimonthee
Project description
The project "Prototype for GPT-AI with Weconomics datasets" by Weconomics Solutions focuses on developing a proprietary AI model, the Weconomics Generative Pre-trained Transformer (Dash-GPT), using Weconomics datasets. The goal is to leverage existing technologies while maintaining development independence. Key challenges include ensuring data quality, accessibility, ethical considerations, privacy, and avoiding vendor lock-in. This involves a complex process of training and fine-tuning the model with proprietary data, requiring expertise in machine learning, data processing, and deployment. The feasibility of in-house development versus external reliance is a crucial consideration. The output is the Weconomics GPT prototype in software.
Context
The project is situated within the domain of Artificial Intelligence, specifically focusing on the development of advanced language models as part of the broader "future of work" theme. Weconomics Solutions is creating a proprietary AI solution, designed to process and answer questions using their specific datasets.
Results
Resulted in working a Chat-GPT style website with each user having their own chats and accounts. The website allows user to see into the dataset by utilizing Knowledge graph that creates relationships between keywords that were extracted from a given source. Utilizing PathRAG for the most accurate responses. With the ability to switch between different modes.
About the project group
Duo of Open Learning semester 2. students interested in the world of AI with zero previous experience into building an AI before coming to the project. Work on this the 80% of a given week with Agile style of working that allowed for flexibility and efficienc