AI Factory Glossary

progressive growing,generative models

Gradually increase resolution during GAN training.

progressive neural networks, continual learning

Add capacity for new tasks.

progressive neural networks,continual learning

Add new capacity for each task.

progressive shrinking, neural architecture search

Progressive shrinking trains once-for-all networks by gradually incorporating smaller sub-networks starting from the largest architecture.

prompt caching, llm optimization

Prompt caching reuses processed prompts reducing latency and cost for repeated prefixes.

prompt chaining, prompting

Use output of one prompt as input to next.

prompt chaining, prompting techniques

Prompt chaining sequences multiple prompts passing outputs as subsequent inputs.

prompt chunking, generative models

Split long prompts.

prompt embeddings, generative models

Vector representation of prompts.

prompt injection attacks, ai safety

Adversarial inputs to subvert model.

prompt injection defense, ai safety

Techniques to prevent injection.

prompt injection, ai safety

Prompt injection inserts malicious instructions into prompts attempting to hijack model behavior.

prompt injection,ai safety

Attack where user input tricks model into ignoring instructions or leaking info.

prompt leaking,ai safety

Trick model into revealing system prompts or instructions.

prompt moderation, ai safety

Check prompts before processing.

prompt truncation, generative models

Cut off excess tokens.

prompt weighting, generative models

Emphasize prompt parts differently.

prompt-to-prompt editing,generative models

Edit images by modifying text prompts.

prompt-to-prompt, multimodal ai

Prompt-to-prompt editing modifies images by adjusting text prompts while preserving structure.

property-based test generation, code ai

Generate property tests.

prophet, time series models

Prophet is an additive time series model decomposing signals into trend seasonality and holiday components with automatic changepoint detection.

proprietary model, llm architecture

Proprietary models have restricted access to weights and implementation.

protected health information detection, phi, healthcare ai

Identify sensitive medical data.

protein design,healthcare ai

Design proteins with desired properties.

protein function prediction from text, healthcare ai

Infer protein function from descriptions.

protein structure prediction,healthcare ai

Predict 3D protein structures (AlphaFold).

protein-ligand binding, healthcare ai

Interaction between protein and small molecule.

prototype learning, explainable ai

Learn representative prototypes for each class.

proxylessnas, neural architecture

Search directly on target hardware.

proxylessnas, neural architecture search

ProxylessNAS directly learns architectures on target hardware by training over-parameterized networks with path-level binarization and latency constraints.

pruning, model optimization

Pruning removes unnecessary network components reducing size and computation.

pruning,model optimization

Remove weights or neurons that contribute little to performance.

pseudo-labeling, advanced training

Pseudo-labeling assigns predicted labels to unlabeled data treating them as ground truth for semi-supervised learning.

pseudonymization, training techniques

Pseudonymization replaces identifiers with pseudonyms allowing reversal with keys.

pubmedbert,domain,biomedical

PubMedBERT is biomedical BERT. Trained on PubMed abstracts.

pull request summarization, code ai

Summarize code changes.

purpose limitation, training techniques

Purpose limitation restricts data use to specified intended purposes.

pyraformer, time series models

Pyraformer uses pyramidal attention with multi-resolution representations for efficient long-term time series forecasting.

pyramid vision transformer (pvt),pyramid vision transformer,pvt,computer vision

Multi-scale vision transformer.

python repl integration,code ai

Execute generated Python code for answers.

pytorch mobile, model optimization

PyTorch Mobile enables optimized on-device inference for PyTorch models on mobile platforms.

qaoa, qaoa, quantum ai

Variational algorithm for optimization.

qkv bias, qkv, transformer

Whether to use bias in attention projections.

quality at source, supply chain & logistics

Quality at source ensures suppliers deliver defect-free materials eliminating incoming inspection.

quantile loss, machine learning

Optimize specific quantiles.

quantitative structure-activity relationship, qsar, chemistry ai

Relate structure to biological activity.

quantization for edge devices, edge ai

Reduce precision for edge inference.

quantization-aware training (qat),quantization-aware training,qat,model optimization

Training with quantization in mind so the model learns to work well at lower precision.

quantization-aware training, model optimization

Quantization-aware training simulates quantization during training improving low-precision accuracy.

quantization,model optimization

Reducing model weight precision (32-bit → 8-bit/4-bit) to save memory and speed up inference.