EXPLORE / 06
Terms index
Specialized AI and software-development terms with concise explanations
RECORDS
Entity records
3397 records · 66/85
- 2601
Capability Uplift
The extent to which a tool or model improves user performance on specific tasks compared to unaided efforts
Website pending - 2602
Spacetime Patches
Decomposing video frames into 3D blocks to process spatial and temporal information simultaneously.
Website pending - 2603
Transformer
A deep learning architecture based on self-attention, widely used for sequence modeling.
Website pending - 2604
Data Privacy Protection
Measures to prevent unauthorized access, use, or disclosure of user data.
Website pending - 2605
SOC-2 Compliance
An auditing procedure that ensures service providers securely manage data.
Website pending - 2606
Code Completion
Feature that predicts and automatically inserts subsequent code snippets based on context
Website pending - 2607
IDE
Software application that provides comprehensive facilities for software development
Website pending - 2608
Zero-shot TTS
Generating a specific speaker's voice without fine-tuning, using only a short reference audio clip
Website pending - 2609
GPT-4
A multimodal large language model by OpenAI with strong reasoning capabilities
Website pending - 2610
Instruction Tuning
Training method using instruction-response pairs to optimize model adherence to human intent
Website pending - 2611
Automated Agent
Software entities that autonomously execute task sequences, here referring to chatbots
Website pending - 2612
Custom Model
An AI model specifically optimized and trained for particular business needs or datasets
Website pending - 2613
AIG-CSAM
Illegal pornographic content involving minors generated using artificial intelligence technologies.
Website pending - 2614
Safety by Design
An approach that integrates safety considerations into every stage of product development rather than addressing them post-hoc.
Website pending - 2615
ASR
Technology that converts human speech into text, as demonstrated by models like Whisper.
Website pending - 2616
Private Link
Technology establishing dedicated network channels between cloud providers to avoid public internet transmission
Website pending - 2617
MFA
A security mechanism requiring two or more verification methods to confirm user identity
Website pending - 2618
Deprecation
The process where official support for a feature or version is phased out and discouraged.
Website pending - 2619
Flagship Model
The core large model version representing a company's highest technical level and performance.
Website pending - 2620
Real-time Inference
Ability of a model to process input and generate output within millisecond-level latency
Website pending - 2621
End-to-End
Generating final output directly from raw input without intermediate conversion steps.
Website pending - 2622
Data API
Interface allowing external programs to access and extract data from a platform
Website pending - 2623
Frontier AI Safety Commitments
Voluntary agreements among companies on safety standards for developing and deploying high-risk models
Website pending - 2624
AI Safety Guardrails
Technical constraints preventing models from generating harmful or non-compliant content
Website pending - 2625
Sparse Autoencoder
A neural network architecture that learns compact data representations by enforcing sparse activation in hidden layers.
Website pending - 2626
Evidence-Based Medicine
Clinical decision-making based on the best available research evidence.
Website pending - 2627
Text Classification
The task of assigning predefined categories to unstructured text data.
Website pending - 2628
Active Learning
An algorithmic strategy where the model selects the most informative samples for labeling.
Website pending - 2629
Iterative Sampling
A generation method that approximates the target distribution through multiple repeated computational steps.
Website pending - 2630
One-step Sampling
Generating data from noise via a single forward pass, bypassing iterative denoising.
Website pending - 2631
Bio-risk
Potential dangers arising from AI misuse in creating or enhancing biological threats
Website pending - 2632
Critique
Structured evaluation pointing out errors or suggesting improvements for model outputs
Website pending - 2633
Prover-Verifier Games
A training mechanism forcing models to generate answers that can be confirmed as correct by an independent verifier.
Website pending - 2634
Legibility
The degree to which model outputs are logically clear and easily understood or verified by external agents.
Website pending - 2635
SCIM
An open standard protocol for automating the exchange of user identity information across domains.
Website pending - 2636
Rule-Based Rewards
Defining reinforcement learning reward signals via preset logical rules rather than human preferences
Website pending - 2637
JSON Schema
Standard specification used to describe and validate JSON data structure
Website pending - 2638
Human Validation
Process where human experts check data quality to eliminate noise
Website pending - 2639
Corporate Governance
The system of rules, practices, and processes by which a company is directed and controlled
Website pending - 2640
Covert Influence Operation
Activities attempting to manipulate public opinion while hiding the true identity or intentions of actors.
Website pending