Generative AI Models: Fine-Tuning & Evaluating

Hello AI Brussels members!

Join us for two exciting tech talks accompanied by delicious pizza and Belgian beers.

Topic 1: Parameter-efficient fine-tuning
Nathanael Mariaule will present PEFT techniques. Instead of laboriously updating all the weights of the original model, PEFT-based methods, inspired by LoRA, ingeniously inject new trainable matrices into the model, supercharging training efficiency.

Topic 2: Evaluating Generative AI downstream tasks
Léonardo Remondini will delve into the art of evaluating Generative AI for downstream applications. Explore cutting-edge evaluation methods, both with and without reference text, including embedding-based, language model-based, and LLM-assisted approaches.

AGENDA:
- 6 pm: Open doors & drinks
- 6:30 pm: Tech Talk 1 (be sharp to avoid interrupting the speakers)
- 7:15 pm: Tech Talk 2
- 8:00 pm: Open floor
- 8:15 pm: Informal Networking Reception + Pizzas and more drinks

Hope to see you there!