Results for "dataset documentation"

Required descriptions of model behavior and limits.

Structured dataset documentation covering collection, composition, recommended uses, biases, and maintenance.

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

Differences between training and deployed patient populations.

A structured collection of examples used to train/evaluate models; quality, bias, and coverage often dominate outcomes.

Policies and practices for approving, monitoring, auditing, and documenting models in production.

Ensuring decisions can be explained and traced.

AI used in sensitive domains requiring compliance.

Ability to inspect and verify AI decisions.

Error due to sensitivity to fluctuations in the training dataset.

Training with a small labeled dataset plus a larger unlabeled dataset, leveraging assumptions like smoothness/cluster structure.

A robust evaluation technique that trains/evaluates across multiple splits to estimate performance variability.

One complete traversal of the training dataset during training.

Halting training when validation performance stops improving to reduce overfitting.

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

A dataset + metric suite for comparing models; can be gamed or misaligned with real-world goals.

Scaling law optimizing compute vs data.

Learning action mapping directly from demonstrations.

Learning a function from input-output pairs (labeled data), optimizing performance on predicting outputs for unseen inputs.

Reusing knowledge from a source task/domain to improve learning on a target task/domain, typically via pretrained models.

A function measuring prediction error (and sometimes calibration), guiding gradient-based optimization.

Minimizing average loss on training data; can overfit when data is limited or biased.

Separating data into training (fit), validation (tune), and test (final estimate) to avoid leakage and optimism bias.

Fraction of correct predictions; can be misleading on imbalanced datasets.

When information from evaluation data improperly influences training, inflating reported performance.

Of true positives, the fraction correctly identified; sensitive to false negatives.

Of true negatives, the fraction correctly identified.

Fine-tuning on (prompt, response) pairs to align a model with instruction-following behaviors.

A gradient method using random minibatches for efficient training on large datasets.

Human or automated process of assigning targets; quality, consistency, and guidelines matter heavily.