Models, methods, and the machinery of intelligence.
Hugging Face
Verdict: RoutineGeneral-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.
State tracking requires an update at every input, but the depth a Transformer applies to each token is fixed regardless of sequence length.
Real-time robot control demands enough visual history to infer motion and task progress, but processing that history can delay action.
Large language model (LLM) routing distributes inference work across different models, advancing the cost-quality Pareto frontier of LLM serving.
Assistants from 2011 answered and waited, and agents from 2023 did a task and stopped.
Reinforcement learning (RL) is the central training paradigm for advancing large foundation models towards self-improvement.
General-purpose agents increasingly write code, use tools, and complete complex digital tasks, raising the question of how far these capabilities carry into the physical world.