small language model
SLM
A small language model is a language model with a comparatively compact architecture or resource footprint, designed to run efficiently on bounded tasks and hardware.
AI Terminology
Search clear AI terminology definitions, acronyms, related concepts and the reviewed videos that explain them.
Showing 221–240 of 705 terms
Clear filterssparse attention
Sparse attention reduces transformer computation by allowing each token to attend to a selected subset of other tokens instead of comparing with every token.
specification-driven development
Specification-driven development begins with an explicit description of required behavior and uses that specification to guide implementation and verification.
supercomputing provider
AI supercomputing providerAn AI supercomputing provider operates large accelerator clusters and sells managed access to organizations training or serving AI models.
superhuman artificial intelligence
superhuman AISuperhuman AI is AI that exceeds the best human performance within one or more defined cognitive tasks or domains.
task planning
AI task planningAI task planning converts a goal into an ordered set of actions that respects dependencies, capabilities and environmental constraints.
task-level artificial intelligence automation
AI task automationtask-level AI automationTask-level AI automation assigns a defined activity within a larger job or workflow to an AI system.
token pricing
AI token pricingAI token pricing charges for model use according to the number and type of tokens processed or generated.
tool contract
A tool contract is a validated definition of a tool's inputs, outputs, errors and side-effect behavior that an AI agent can rely on.
unattended artificial intelligence agent
An unattended AI agent performs background work without a person supervising every execution step.
user intent inference
User intent inference is the process of estimating the goal or need behind a person's words and behavior so a system can choose a relevant response or action.
video generation model
AI video generation modelAn AI video generation model creates or modifies moving visual sequences from prompts, images, video or other context.
visual benchmark
An AI visual benchmark evaluates a model on tasks that require interpreting or producing visual information.
workforce transformation
AI workforce transformationAI-driven work transformationAI workforce transformation is the change in tasks, roles, skills and organizational structures caused by the adoption of AI systems.
abstention
AI abstentionAI abstention is the deliberate choice not to provide or act on a prediction when evidence, confidence, permissions, or operating conditions are insufficient.
abstraction layer
An AI abstraction layer hides low-level model and infrastructure details behind a simpler, stable interface.
accelerator
An AI accelerator is specialized computing hardware designed to perform the matrix and tensor operations used by AI training or inference efficiently.
adaptive robot motion control
Adaptive robot motion control changes movement during execution in response to sensed objects, forces or environmental conditions.
adversarial evaluation
Adversarial evaluation tests an AI system with deliberately difficult, deceptive, or hostile inputs to reveal failures that ordinary benchmarks may miss.
agent advocacy
AI agent advocacyAI agent advocacy is the practice of making developer products understandable, discoverable, and usable by software agents while preserving trusted relationships with human developers.
Knowledge graph
Explore AI concepts
Follow general terms into specialised sub-terms. Select any node to open its definition.