Results for "search"
Semantic Search
IntermediateRetrieval based on embedding similarity rather than keyword overlap, capturing paraphrases and related concepts.
Semantic search is like having a smart friend who understands what you really mean, not just the words you use. Instead of looking for exact phrases, it tries to find answers that match the meaning of your question. For example, if you search for 'best places to eat Italian food,' a semantic sear...
Choosing step size along gradient direction.
Search algorithm for generation that keeps top-k partial sequences; can improve likelihood but reduce diversity.
Retrieval based on embedding similarity rather than keyword overlap, capturing paraphrases and related concepts.
Directly optimizing control policies.
A datastore optimized for similarity search over embeddings, enabling semantic retrieval at scale.
Optimal pathfinding algorithm.
Configuration choices not learned directly (or not typically learned) that govern training or architecture.
Methods for breaking goals into steps; can be classical (A*, STRIPS) or LLM-driven with tool calls.
Raw model outputs before converting to probabilities; manipulated during decoding and calibration.
Number of steps considered in planning.
Agent reasoning about future outcomes.
Direction of steepest ascent of a function.
Lowest possible loss.
Restricting updates to safe regions.
RL using learned or known environment models.
Computing collision-free trajectories.
Finding routes from start to goal.
Sampling-based motion planner.
Imagined future trajectories.
Finding mathematical equations from data.
Fast approximation of costly simulations.