Models, methods, and the machinery of intelligence.
Hugging Face
Verdict: RoutineLarge language model (LLM) routing distributes inference work across different models, advancing the cost-quality Pareto frontier of LLM serving.
General-purpose service robots need navigation systems that can handle diverse human requests in unfamiliar environments, combining task generality with scene generality.
Interactive virtual worlds allow agents to learn through exploration and interaction.
Reinforcement learning (RL) is the central training paradigm for advancing large foundation models towards self-improvement.
Dense correspondence matching has historically been bounded by simplifying spatio-temporal priors, such as smooth motion and rigid geometry.
Recent 3D world models generate photorealistic, explorable scenes that remain frozen in time.
Egocentric world models predict first-person observations conditioned on an agent's actions, but most focus on a single agent.
Flow policies capture rich and diverse action distributions, and fine-tuning them with off-policy RL to improve beyond the demonstrations has drawn growing interest.