Results for "reasoning + action"
Set of all actions available to the agent.
Stepwise reasoning patterns that can improve multi-step tasks; often handled implicitly or summarized for safety/privacy.
Interleaving reasoning and tool use.
Continuous cycle of observation, reasoning, action, and feedback.
Temporary reasoning space (often hidden).
Simple agent responding directly to inputs.
Expected cumulative reward from a state or state-action pair.
Expected return of taking action in a state.
The field of building systems that perform tasks associated with human intelligence—perception, reasoning, language, planning, and decision-making—via algori...
Predicts next state given current state and action.
Formal framework for sequential decision-making under uncertainty.
Maximum number of tokens the model can attend to in one forward pass; constrains long-document reasoning.
Framework for reasoning about cause-effect relationships beyond correlation, often using structural assumptions and experiments.
Agent reasoning about future outcomes.
AI capable of performing most intellectual tasks humans can.
What would have happened under different conditions.
Methods for breaking goals into steps; can be classical (A*, STRIPS) or LLM-driven with tool calls.
System that independently pursues goals over time.
AI systems that perceive and act in the physical world through sensors and actuators.
Strategy mapping states to actions.
Fundamental recursive relationship defining optimal value functions.
Learning only from current policy’s data.
Optimizing policies directly via gradient ascent on expected reward.
Balancing learning new behaviors vs exploiting known rewards.
Continuous loop adjusting actions based on state feedback.
Learning policies from expert demonstrations.
Learning action mapping directly from demonstrations.
Acting to minimize surprise or free energy.
Achieving task performance by providing a small number of examples inside the prompt without weight updates.
Constraining outputs to retrieved or provided sources, often with citation, to improve factual reliability.