How do Llamas process multilingual text? A latent exploration through activation patching
Published:
We provide evidence for the existence of language-agnostic concept representations within LLMs
Download here
Research
Satisficing Reinforcement Learning
Published:
Inspired by satisficing, we introduce a novel concept of non-maximizing agents, א-aspiring agents, whose goal is to achieve an expected gain of א. We derive aspiration-based algorithms from Q-learning and DQN.
Download here
Local Optimization in Monte-Carlo Tree Search for the Traveling Salesman Problem
Published:
Enhancement of Monte-Carlo Tree Search for the Traveling Salesman Problem with 2-opt Optimization
Download here
Download the French version here