EXPLORE / 06
Terms index
Specialized AI and software-development terms with concise explanations
Entity records
4517 records · 65/113
- 2561
Semantic Security
Content-meaning-based filtering preventing attacks like prompt injection, unlike keyword matching
Website pending - 2562
TensorFlow
An end-to-end open source machine learning platform developed by Google
Website pending - 2563
DDP
A training strategy that shards data across devices for parallel gradient computation and synchronization
Website pending - 2564
GPT Architecture
An autoregressive generative model structure based on stacked Transformer decoders.
Website pending - 2565
Face Swapping
Technology that transfers facial features from a source person to a target in video.
Website pending - 2566
Agent Framework
Software architecture supporting LLMs in autonomous planning, tool calling, and task execution
Website pending - 2567
Spec
Detailed documentation describing software functionality, behavior, or constraints, often used to guide AI code generation.
Website pending - 2568
Call Chain / Trace
A data structure recording all service nodes and interaction sequences a request passes through in a distributed system.
Website pending - 2569
Agent Observability
The capability to monitor and analyze an AI agent's decision paths, tool calls, and resource consumption.
Website pending - 2570
End-to-End Delivery
A mode where the entire process from requirement analysis to final product deployment is automated or completed as a unified whole.
Website pending - 2571
Runtime Sovereignty
Judgment and control over the AI execution environment, state management, and ultimate authority ownership.
Website pending - 2572
Mental Model
The internal cognitive structure developers use to predict system behavior and guide debugging.
Website pending - 2573
3D World Building
The process of creating 3D virtual environments and objects using algorithms or tools
Website pending - 2574
Object Storage
A data storage architecture that manages data as objects, suitable for large-scale access to unstructured data.
Website pending - 2575
Hot-swapping
Technique allowing dynamic loading or unloading of components without restarting the system
Website pending - 2576
Plugin Architecture
A design pattern allowing dynamic system extension via independent modules without modifying core code.
Website pending - 2577
Cross-platform Development
Building apps for multiple OSes from a single codebase.
Website pending - 2578
Over-engineering
Designing a solution to be more complex than necessary for the problem at hand
Website pending - 2579
ReAct
An agent framework combining Reasoning and Acting to enhance model decision-making capabilities
Website pending - 2580
Multi-head Latent Attention (MLA)
An attention mechanism that compresses KV cache via low-rank projection to reduce memory usage and computation.
Website pending - 2581
MLA
An attention variant that compresses key-value pairs to reduce memory usage and computation
Website pending - 2582
Regression Evaluation
Testing system changes against historical cases to ensure no performance degradation.
Website pending - 2583
Full-duplex Voice Interaction
A dialogue mode allowing users to speak simultaneously with AI and interrupt at any time
Website pending - 2584
End-to-End Voice Dialogue
Generating audio responses directly from audio input without intermediate text conversion
Website pending - 2585
N-gram Embedding
A method using continuous word sequences for lookup addressing to enhance memory efficiency.
Website pending - 2586
Model-as-a-Service
A business model delivering AI model capabilities via cloud APIs for developers.
Website pending - 2587
Distributed System
A system architecture where multiple computers collaborate over a network to complete common tasks.
Website pending - 2588
Pareto Frontier
A set of optimal solutions where improving one objective (e.g., performance) requires sacrificing another (e.g., cost)
Website pending - 2589
Graph Structure
A data structure of nodes and edges representing entity relationships
Website pending - 2590
Inverted Index
A classic data structure mapping terms to documents for fast full-text retrieval.
Website pending - 2591
Semantic ID (SID)
Discrete sequences encoding unstructured data for model generation.
Website pending - 2592
Generative Retrieval
A paradigm where models directly generate document identifiers instead of traditional matching.
Website pending - 2593
Cache Compression
Inference optimization technique that reduces KV Cache memory footprint via algorithms
Website pending - 2594
Skill Bank
A structured knowledge base storing reusable task modules or tool calls for dynamic retrieval by agents.
Website pending - 2595
Non-parametric Components
Elements like prompts, memory, and tool definitions that do not directly participate in gradient updates.
Website pending - 2596
Self-Evolution
An AI paradigm where systems continuously optimize themselves via environmental interaction and feedback.
Website pending - 2597
Agent-as-Customer
A paradigm where AI agents autonomously understand tasks, select resources, and complete transaction loops
Website pending - 2598
MaaS
A service model providing pre-trained model inference capabilities via cloud APIs.
Website pending - 2599
Falsifiability
The capacity of a scientific theory to be proven wrong by empirical evidence
Website pending - 2600
Autoformalization
Converting natural language math statements into machine-verifiable formal languages
Website pending