Results for "data distribution"

A preference-based training method optimizing policies directly from pairwise comparisons without explicit RL loops.

Model trained to predict human preferences (or utility) for candidate outputs; used in RLHF-style pipelines.

Local surrogate explanation method approximating model behavior near a specific input.

A hidden variable influences both cause and effect, biasing naive estimates of causal impact.

Ordering training samples from easier to harder to improve convergence or generalization.

Standardized documentation describing intended use, performance, limitations, data, and ethical considerations.

Automated testing and deployment processes for models and data workflows, extending DevOps to ML artifacts.

A broader capability to infer internal system state from telemetry, crucial for AI services and agents.

Time from request to response; critical for real-time inference and UX.

How many requests or tokens can be processed per unit time; affects scalability and cost.

Hardware resources used for training/inference; constrained by memory bandwidth, FLOPs, and parallelism.

Exponential of average negative log-likelihood; lower means better predictive fit, not necessarily better utility.

Inputs crafted to cause model errors or unsafe behavior, often imperceptible in vision or subtle in text.

System design where humans validate or guide model outputs, especially for high-stakes decisions.

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

Coordinating tools, models, and steps (retrieval, calls, validation) to deliver reliable end-to-end behavior.

Constraining model outputs into a schema used to call external APIs/tools safely and deterministically.

AI focused on interpreting images/video: classification, detection, segmentation, tracking, and 3D understanding.

A model is PAC-learnable if it can, with high probability, learn an approximately correct hypothesis from finite samples.

AI subfield dealing with understanding and generating human language, including syntax, semantics, and pragmatics.

Error due to sensitivity to fluctuations in the training dataset.

Converting audio speech into text, often using encoder-decoder or transducer architectures.

Variability introduced by minibatch sampling during SGD.

A wide basin often correlated with better generalization.

A narrow hidden layer forcing compact representations.

The range of functions a model can represent.

Allows model to attend to information from different subspaces simultaneously.

Encodes token position explicitly, often via sinusoids.

Categorizing AI applications by impact and regulatory risk.

Models trained to decide when to call tools.