Results for "compressed space"

Diffusion performed in latent space for efficiency.

Modeling environment evolution in latent space.

The internal space where learned representations live; operations here often correlate with semantics or generative factors.

Space of all possible robot configurations.

Set of all actions available to the agent.

Set of vectors closed under addition and scalar multiplication.

A continuous vector encoding of an item (word, image, user) such that semantic similarity corresponds to geometric closeness.

All possible configurations an agent may encounter.

Models time evolution via hidden states.

Model that compresses input into latent space and reconstructs it.

Computing collision-free trajectories.

Fast approximation of costly simulations.

A parameterized mapping from inputs to outputs; includes architecture + learned parameters.

A datastore optimized for similarity search over embeddings, enabling semantic retrieval at scale.

Designing input features to expose useful structure (e.g., ratios, lags, aggregations), often crucial outside deep learning.

The shape of the loss function over parameter space.

A narrow minimum often associated with poorer generalization.

A wide basin often correlated with better generalization.

Encodes positional information via rotation in embedding space.

Strategy mapping states to actions.

Formal framework for sequential decision-making under uncertainty.

Temporary reasoning space (often hidden).

Using production outcomes to improve models.

Variable whose values depend on chance.

A measurable property or attribute used as model input (raw or engineered), such as age, pixel intensity, or token ID.

Retrieval based on embedding similarity rather than keyword overlap, capturing paraphrases and related concepts.

Automatically learning useful internal features (latent variables) that capture salient structure for downstream tasks.

Expanding training data via transformations (flips, noise, paraphrases) to improve robustness.

Model-generated content that is fluent but unsupported by evidence or incorrect; mitigated by grounding and verification.

Mechanisms for retaining context across turns/sessions: scratchpads, vector memories, structured stores.