Welcome to the first-ever workshop organized by ISA! The theme of this workshop is Meltemi, the first open large language model for Greek, and how it can be used in conjunction with RAG (Retrieval-Augmented Generation) Technology.
We had the honor of hosting the Meltemi Team from the Institute for Language and Speech Processing (ILSP), including:
- Vassilis Katsouros, Director of Research and Director of ILSP
- Sokratis Sofianopoulos, Collaborating Researcher
- Giorgos Paraskeuopoulos, Collaborating Researcher
- Prokopis Prokopidis, Collaborating Researcher
- Dimitris Roussis, Collaborating Researcher
- Leon Voukoutis, Collaborating Researcher
This repository serves as a summary and resource hub for the day’s activities, including key presentations and hands-on coding sessions.
A captivating discussion about Meltemi, exploring its origins, how the idea took shape and became a reality, technical details of its development, and hints about the next steps. This engaging panel was moderated by Konstantinos Eleftheriou, Head of Software at ISA, with the entire Meltemi Team as panelists.
An introductory tutorial on Hugging Face, hosted by Ippokratis Pantelidis and Nikos Mitsakis, Heads of AI at ISA, answering foundational questions such as "What is Hugging Face?" and "How and why can we use it?". This session covered the Hugging Face user interface, explained how to leverage its resources in code, and demonstrated using Meltemi directly through Hugging Face to tackle various tasks.
The complete Jupyter Notebook from this session can be found here.
This session, presented by Giorgos Paraskeuopoulos covered both theoretical and practical aspects of RAG, including its components, the pipeline, and its usefulness. Giorgos explained how to leverage Meltemi with RAG technology to answer questions with up-to-date, precise answers, while minimizing hallucinations in the generation process. The session seamlessly transitioned from theory to practice, with code examples showing how to implement RAG with Meltemi.
The resources (Slides and Jupyter Notebook) for the session can be found here.
At the end of the workshop, participants had the chance to apply everything they had learned, using Meltemi with RAG to build their own applications. This hands-on session allowed attendees to experiment with the tools and techniques discussed earlier and develop real-world AI solutions.
The project description can be found here.
The best 3 projects from this phase are soon to be announced!
This workshop would not have been possible without the dedication and hard work of the following:
- The Meltemi Team from the Institute for Language and Speech Processing (ILSP) for their great sessions and continued support.
- Athens University of Economics and Business for their generous support and for providing the amphitheatre in the Trias Building.
A big thank you to all speakers and participants who were instrumental in creating such a memorable event!
- Follow the Information Science Association for updates on future events.
- Follow the Institute for Language and Speech Processing (ILSP).
- Use Meltemi straight from here
Read this in Greek