Results for "data mismatch"

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.

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).

One complete traversal of the training dataset during training.

A parameterized function composed of interconnected units organized in layers with nonlinear activations.

Networks using convolution operations with weight sharing and locality, effective for images and signals.

Networks with recurrent connections for sequences; largely supplanted by Transformers for many tasks.

Converting text into discrete units (tokens) for modeling; subword tokenizers balance vocabulary size and coverage.

A high-capacity language model trained on massive corpora, exhibiting broad generalization and emergent behaviors.

Letting an LLM call external functions/APIs to fetch data, compute, or take actions, improving reliability.

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

Reinforcement learning from human feedback: uses preference data to train a reward model and optimize the policy.

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

Practices for operationalizing ML: versioning, CI/CD, monitoring, retraining, and reliable production management.

Central system to store model versions, metadata, approvals, and deployment state.

Ability to replicate results given same code/data; harder in distributed training and nondeterministic ops.

Observing model inputs/outputs, latency, cost, and quality over time to catch regressions and drift.

Attacks that manipulate model instructions (especially via retrieved content) to override system goals or exfiltrate data.

Models that process or generate multiple modalities, enabling vision-language tasks, speech, video understanding, etc.

Updating beliefs about parameters using observed evidence and prior distributions.

A theoretical framework analyzing what classes of functions can be learned, how efficiently, and with what guarantees.

Bayesian parameter estimation using the mode of the posterior distribution.

Using same parameters across different parts of a model.

Techniques to handle longer documents without quadratic cost.

A single attention mechanism within multi-head attention.

Capabilities that appear only beyond certain model sizes.

Chooses which experts process each token.

Probabilistic energy-based neural network with hidden variables.

Central catalog of deployed and experimental models.