← Back to AI Factory Chat

AI Factory Glossary

3,145 technical terms and definitions

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z All
Showing page 56 of 63 (3,145 entries)

sustainable materials, environmental & sustainability

Sustainable materials for semiconductor manufacturing consider environmental impact recyclability and supply chain sustainability.

sustainable sourcing, environmental & sustainability

Sustainable sourcing selects suppliers based on environmental and social responsibility criteria.

svar, svar, time series models

Structural Vector Autoregression imposes causal ordering on VAR models for policy analysis and forecasting.

svcca, svcca, explainable ai

Compare layer representations.

svd compression, svd, model optimization

Singular value decomposition compression represents matrices through truncated SVD.

swe-bench, ai agents

SWE-Bench assesses agents on real-world software engineering tasks from GitHub.

swiglu activation, neural architecture

Swish-gated linear unit activation.

swin transformer,computer vision

Hierarchical vision transformer with shifted windows.

swinir, multimodal ai

SwinIR applies Swin Transformer to image restoration tasks including super-resolution.

swish, neural architecture

Smooth activation x*sigmoid(x).

switch transformer, llm architecture

Switch Transformer uses simplified routing to single expert per token.

switch transformer,model architecture

Sparse MoE model with one expert per token.

switchable normalization, neural architecture

Learn to select best normalization.

switching state space, time series models

Switching state space models allow dynamics to change between discrete regimes capturing non-stationary behavior.

symmetric vs asymmetric quantization,model optimization

Quantization with/without zero-point offset.

symplectic neural networks, scientific ml

Preserve Hamiltonian structure in dynamics.

symptom extraction, healthcare ai

Identify symptoms from text.

synchronized multimodal representations, multimodal ai

Align representations across modalities.

synflow proxy, neural architecture search

Synflow zero-cost proxy scores architectures using synaptic flow capturing trainability without weight updates.

syntactic heads, explainable ai

Encode grammatical structure.

synthesizer, llm architecture

Synthesizer learns attention weights directly without computing query-key products.

synthesizer, transformer

Generate attention without queries and keys.

synthetic accessibility, chemistry ai

Estimate ease of chemical synthesis.

synthetic data, training techniques

Synthetic data artificially generates training examples preserving statistical properties.

synthetic patient generation, healthcare ai

Create realistic synthetic medical data.

system design,scale,high availability

I can walk you through system design interviews or real architectures: load balancers, caching, sharding, and HA/failover.

system prompt extraction,ai safety

Attempt to recover hidden system instructions.

system reliability modeling, reliability

Predict system-level reliability.

t-closeness, training techniques

T-closeness ensures sensitive attribute distribution within groups matches overall distribution.

t0, t0, foundation model

Zero-shot instruction-following model.

t0, t0, training techniques

T0 trains models on prompted datasets converted to unified text-to-text format.

t2i-adapter, generative models

Lightweight control adapter.

t5 (text-to-text transfer transformer),t5,text-to-text transfer transformer,foundation model

Treats all NLP tasks as text-to-text problems.

tail-free sampling, tfs, text generation

Remove low-probability tail.

take-back program, environmental & sustainability

Take-back programs enable manufacturers to collect used products for recycling or proper disposal.

talking heads attention, transformer

Learn to redistribute attention across heads.

task allocation, ai agents

Task allocation assigns responsibilities to agents based on capabilities and load.

task arithmetic, model merging

Add/subtract task vectors.

task decomposition, ai agents

Task decomposition breaks complex goals into manageable subtasks.

task diversity, training techniques

Task diversity in instruction tuning exposes models to varied problem types.

task-specific pre-training, transfer learning

Pre-train for specific downstream task.

taylor expansion pruning, model optimization

Taylor expansion pruning approximates loss change from removing weights using Taylor series.

tbats, tbats, time series models

TBATS combines Box-Cox transformation Fourier seasonality ARMA errors and trend for complex seasonal time series.

tcad model parameters, tcad, simulation

Physical parameters used in device/process simulation.

tcn, tcn, time series models

Temporal Convolutional Networks use dilated causal convolutions for sequence modeling with long effective history.

te-nas, te-nas, neural architecture search

Training-free ensemble NAS combines multiple zero-cost proxies improving architecture evaluation reliability.

teacher-student cl, advanced training

Teacher-student curriculum learning uses a teacher model to assess sample difficulty and guide curriculum design for student training.

teacher-student framework, model compression

General paradigm for distillation.

teacher-student training, model optimization

Teacher-student training transfers knowledge from complex to simple models through soft targets.

teaching assistant, model compression

Intermediate model between teacher and student.