Results for "data distribution"
Scales logits before sampling; higher increases randomness/diversity, lower increases determinism.
Converts logits to probabilities by exponentiation and normalization; common in classification and LMs.
Measures how much information an observable random variable carries about unknown parameters.
Performance drop when moving from simulation to reality.
Learning policies from expert demonstrations.
Restricting distribution of powerful models.
Shift in feature distribution over time.
Inferring sensitive features of training data.
Processes and controls for data quality, access, lineage, retention, and compliance across the AI lifecycle.
Tracking where data came from and how it was transformed; key for debugging and compliance.
Learning where data arrives sequentially and the model updates continuously, often under changing distributions.
The internal space where learned representations live; operations here often correlate with semantics or generative factors.
Recovering training data from gradients.
Diffusion performed in latent space for efficiency.
Enables external computation or lookup.
A narrow minimum often associated with poorer generalization.
Built-in assumptions guiding learning efficiency and generalization.
Methods that learn training procedures or initializations so models can adapt quickly to new tasks with little data.
Techniques that discourage overly complex solutions to improve generalization (reduce overfitting).
Probabilistic energy-based neural network with hidden variables.
Model-generated content that is fluent but unsupported by evidence or incorrect; mitigated by grounding and verification.
Probabilistic graphical model for structured prediction.
Identifying abrupt changes in data generation.
Randomizing simulation parameters to improve real-world transfer.
Simultaneous Localization and Mapping for robotics.
Monte Carlo method for state estimation.
Prompt augmented with retrieved documents.
Inferring the agent’s internal state from noisy sensor data.
Fast approximation of costly simulations.
Training across many devices/silos without centralizing raw data; aggregates updates, not data.