Metalearning Innovations
Tom discusses the motivation behind Stackx, emphasizing the importance of multiple discount factors in learning strategies. By exploring different planning horizons, the team aims to enhance agent performance through metagradients, focusing on problem understanding and fine-tuning for optimal results.In this clip
From this podcast

Machine Learning Street Talk (MLST)
#49 - Meta-Gradients in RL - Dr. Tom Zahavy (DeepMind)
Related Questions