← Back to AI Factory Chat

AI Factory Glossary

186 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 3 of 4 (186 entries)

ambipolar diffusion, device physics

Coupled electron-hole diffusion.

amoebanet, neural architecture search

AmoebaNet uses regularized evolution for neural architecture search demonstrating that simple aging-based regularization improves discovered architectures.

amsaa model, amsaa, business & standards

AMSAA model provides statistical framework for reliability growth analysis.

amsaa model, amsaa, reliability

Army reliability growth model.

ancestral sampling, generative models

Stochastic diffusion sampling.

ancestral sampling,generative models

Stochastic sampling following original process.

anchors, explainable ai

High-precision rules explaining predictions.

ani, ani, chemistry ai

Neural network potential for organic molecules.

annealed langevin dynamics,generative models

Use schedule of noise levels.

anode, anode, neural architecture

Specific augmentation strategy for Neural ODEs.

anomaly detection, ai safety

Anomaly detection identifies unusual inputs warranting careful handling.

ansor, model optimization

Ansor generates optimized tensor programs through hierarchical search and learned cost models.

antibody design,healthcare ai

Create therapeutic antibodies.

antifuse repair, yield enhancement

Antifuse-based repair creates permanent connections to redundant memory elements by applying high voltage to form conductive links.

any-precision networks, model optimization

Support arbitrary bit-widths.

aot compilation, aot, model optimization

Ahead-of-time compilation generates optimized binaries before deployment reducing startup time.

api documentation generation, api, code ai

Auto-generate API docs from code.

api learning,ai agent

Train models to understand and use external APIs.

api sequence generation,code ai

Generate correct sequences of API calls.

appraisal costs, quality

Cost of inspection and testing.

appropriate refusals, ai safety

Decline genuinely harmful requests.

approximate computing, model optimization

Approximate computing trades accuracy for efficiency using lower-precision or simplified operations.

architecture crossover, neural architecture search

Architecture crossover combines parent architectures by exchanging substructures creating offspring networks.

architecture encoding, neural architecture search

Architecture encoding represents network structures as vectors graphs or sequences enabling architecture optimization.

architecture mutation, neural architecture search

Architecture mutations in evolutionary NAS modify network structures through operations like adding layers or changing connections.

argmax flows, generative models

Invertible functions for discrete data.

arima modeling, arima, statistics

Time series forecasting.

arima, arima, time series models

AutoRegressive Integrated Moving Average models time series through differencing and combining autoregression with moving average components.

assertion generation, code ai

Generate test assertions.

asymmetric loss functions, machine learning

Different penalties for different errors.

async generation, llm optimization

Asynchronous generation handles multiple requests concurrently maximizing throughput.

async,await,concurrency

Async/await enables concurrent I/O without threads. Event loop. Python asyncio, JavaScript promises.

atlas,foundation model

Retrieval-augmented model for knowledge-intensive tasks.

attention distance analysis, explainable ai

Study how far attention reaches.

attention flow, explainable ai

Trace attention patterns through network.

attention forecasting, time series models

Attention mechanisms in time series weight historical values adaptively for improved forecasting accuracy.

attention head roles, explainable ai

Different functions of attention heads.

attention head scaling, transformer

Scale attention by head dimension.

attention mask,transformer

Binary mask indicating which tokens to attend to vs ignore (for padding).

attention pooling graph, graph neural networks

Attention-based graph pooling learns soft cluster assignments through self-attention over node features.

attention rollout in vit, explainable ai

Aggregate attention across layers.

attention rollout, explainable ai

Aggregate attention across layers for interpretability.

attention sink, llm architecture

Keep first tokens for stable attention.

attention transfer, model compression

Transfer attention maps from teacher.

attention visualization in vit, explainable ai

Visualize what model attends to.

attention visualization,ai safety

Inspect attention weights to see what the model focuses on.

attention-based explain, recommendation systems

Attention-based explanations visualize which user history items or features influenced specific recommendations.

attention-based fusion, multimodal ai

Use attention to weight modalities.

attentionnas, neural architecture search

AttentionNAS discovers efficient attention mechanisms and architectures jointly through neural architecture search.

attentivenas, neural architecture search

AttentiveNAS uses attention mechanisms to predict architecture performance from training statistics without full training.