About Me
Hello, I'm Claudio Greco, a Member of Technical Staff at Reka AI working on world models, large multimodal foundation models and multimodal agents.
In the past, I've been a Research Engineer at Alana AI, where I worked on building multimodal foundation models, and a Research Fellow at Heriot-Watt University, where I worked on designing trustworthy conversational agents.
Before that, I obtained a PhD in Cognitive and Brain Sciences at the Center for Mind/Brain Sciences of the University of Trento working in the Language and Vision Research Group under the supervision of Raffaella Bernardi. Before enrolling in my Ph.D program, I obtained B.Sc. and M.Sc. degrees in Computer Science at the University of Bari, where I worked on social network analysis and (conversational) recommender systems, respectively.
Experience
- Designed, developed and trained from scratch a Multimodal (Vision & Language) Large Foundation Model, scaling the training process up to 7 billion parameters.
- Built an instruction-tuning dataset for Egocentric Video Understanding.
- Fine-tuned an open-source Multimodal Large Foundation Model on the Egocentric Video Understanding dataset using LoRa, showcasing competitive results compared to Gemini Pro 1.5 and GPT-4V, even surpassing the latter in spatial reasoning, and outperforming open-source models.
Education
Topics: Continual / Transfer Learning, Transformers, and Multimodal Learning.
Supervisor: Raffaella Bernardi.