perplexity
ppl, evaluation, cross-entropy, language model, metric
Perplexity is the standard *intrinsic* measure of how well a language model predicts text, and the cleanest way to understand it is as the model's average *branching factor*…
2,262 technical terms and definitions
ppl, evaluation, cross-entropy, language model, metric
Perplexity is the standard *intrinsic* measure of how well a language model predicts text, and the cleanest way to understand it is as the model's average *branching factor*…
evaluation
Perplexity is the standard *intrinsic* measure of how well a language model predicts text, and the cleanest way to understand it is as the model's average *branching factor*…
ppl, perplexity metric, language model perplexity, bits per token, cross entropy perplexity, evaluation
Perplexity is the standard *intrinsic* measure of how well a language model predicts text, and the cleanest way to understand it is as the model's average *branching factor*…
evaluation
Perplexity-based detection uses a language model's perplexity (a measure of surprise or uncertainty) on specific text as a signal to detect whether that text was part of…
data quality
Perplexity filtering is quality filtering that removes text with abnormal language-model perplexity values - Very high perplexity often indicates corrupted or nonsensical text…
persistent data structure, persistent data, data structure
Persistent Data Structure Introduction & Motivation Persistent Data Structure addresses a central problem in data-structures process control and version-storage-logarithmic…
Persistent Homology and Topological Graph Learning Introduction & Motivation Persistent Homology and Topological Graph Learning addresses a central problem in graph network…
gpu persistent thread, persistent cuda, long running kernel, gpu polling kernel
Persistent Kernels are the GPU programming technique where a kernel is launched once and runs indefinitely, continuously polling for new work from a shared queue rather than being…
long running gpu kernel, work queue gpu, kernel launch overhead, producer consumer gpu
Persistent GPU Kernels is the programming technique where a single GPU kernel runs continuously for the lifetime of the application (or a large phase of it), consuming work items…
persistent memory latency hierarchy and access patterns, persistent memory latency, memory latency hierarchy, latency hierarchy access, hierarchy access patterns, memory latency
Persistent Memory Latency Hierarchy and Access Patterns Introduction & Motivation Persistent Memory Latency Hierarchy and Access Patterns addresses a central problem in storage…
pmem concurrency, dax programming model, byte addressable storage runtime, nv memory software
Persistent Memory Programming is the software model for using byte addressable nonvolatile memory as a durable low latency data tier. What It Covers Core concept: combines load…
persistent storage processing integration, persistent storage processing, storage processing integration, persistent storage, storage processing, processing integration
Persistent Storage Processing Integration Introduction & Motivation Persistent Storage Processing Integration addresses a central problem in storage-computing process control…
persistent kernel, warp level programming, producer consumer gpu, circular buffer gpu
Persistent Threads is a GPU programming pattern where a fixed number of threads remain alive for the entire program duration — repeatedly fetching work from a shared queue rather…
persistent volume management, persistent volume, volume management
Persistent Volume Management Introduction & Motivation Persistent Volume Management addresses a central problem in cloud process control and volume-lifecycle-creation-deletion…
person re identification reid, person re identification, re identification reid, person re, re identification, identification reid
Person Re Identification ReID Introduction & Motivation Person Re Identification ReID addresses a central problem in vision process control and person-reid-pedestrian-matching…
character, roleplay
AI Persona is the character, personality, and behavioral identity defined in a system prompt that transforms a general-purpose language model into a specific, consistent…
dialogue
Persona-based models is dialogue models that explicitly incorporate persona attributes to shape response behavior - Persona embeddings prompts or adapters steer style preferences…
dialogue
Persona Consistency is the challenge of ensuring AI dialogue systems maintain coherent personality traits, knowledge, and behavioral patterns throughout extended conversations…
dialogue
Persona consistency is the ability to maintain stable style tone and identity traits across conversation turns - Consistency mechanisms condition responses on persona constraints…
persona prompts, role play prompting, assigning a persona, expert persona, system prompt persona, in-context impersonation, LLM persona effect
Persona prompting is the practice of instructing a language model to adopt a specific identity, such as a senior tax accountant, a pirate, a skeptical reviewer or a particular…
PC, microprocessor, individual computing, software ecosystem, decentralized computing
The personal computer is a general-purpose computer sized, priced, and designed for use by an individual, a category that emerged in the 1970s and transformed computing…
user modeling, collaborative filtering, content-based filtering, filter bubble, Eli Pariser, cold start problem, recommender privacy
Personalization is the practice of adapting what a system shows or does to the individual user, based on a model of that person's preferences, behaviour and context. It governs…
personalization user modeling, personalization user, user modeling
Personalization User Modeling Introduction & Motivation Personalization User Modeling addresses a central problem in systems process control and personalization-latency-real-time…
Personalization vs Popularity Bias Introduction & Motivation Personalization vs Popularity Bias addresses a central problem in recsys systems and {target} optimization: how…
Ditto Fair and Robust Personalized Federated Learning Introduction & Motivation Ditto Fair and Robust Personalized Federated Learning addresses a central problem in federated…
federated learning
Personalized Federated Learning is an approach that learns models customized to individual clients while leveraging collective knowledge — enabling each participant to benefit…
Personalized Federated Learning Introduction & Motivation Personalized Federated Learning addresses a central problem in federated and local-adaptation-strategies architecture…
Per-FedAvg Model-Agnostic Meta-Learning for Personalization Introduction & Motivation Per-FedAvg Model-Agnostic Meta-Learning for Personalization addresses a central problem…
recommender systems
Personalized ranking orders items specifically for each user — customizing the order of search results, product listings, or content feeds based on individual preferences…
healthcare ai
Personalized treatment plans use AI to customize therapy for each individual patient — integrating patient history, genomics, biomarkers, comorbidities, preferences…
contamination
Personnel contamination is a fundamental cleanroom challenge where human operators are the largest single source of particles, chemicals, and biological contaminants — the human…
ai safety
Perspective API is a free, ML-powered API developed by Google's Jigsaw team that analyzes text and scores it for various toxicity attributes — including toxicity, insults…
ai safety
Perspective API is the text-moderation service that scores toxicity-related attributes to help detect abusive or harmful language - it is commonly used as a moderation signal…
reasoning
Perspective taking is the cognitive ability to consider situations, problems, or information from different viewpoints — including those of other individuals, stakeholders…
pert, quality & reliability
PERT is program evaluation and review technique that estimates project duration under uncertainty using three-point time estimates - It is a core method in modern semiconductor…
reinforcement learning advanced
Pessimistic MDP is an offline reinforcement-learning formulation that penalizes uncertain value estimates to avoid over-optimistic actions. - It treats out-of-distribution regions…
pets, reinforcement learning advanced
PETS is probabilistic ensembles with trajectory sampling for model-based control - Ensembles model dynamics uncertainty and planning evaluates action sequences through sampled…
pfas abatement destruction efficiency prediction, pfas abatement destruction, abatement destruction efficiency, destruction efficiency prediction, pfas abatement
PFAS Abatement Destruction Efficiency Prediction with Machine Learning Introduction & Motivation PFAS Abatement Destruction Efficiency Prediction with Machine Learning addresses…
pfas-free process transition drift prediction, pfas-free process transition, process transition drift, transition drift prediction, pfas-free process, process transition
PFAS-Free Process Transition Drift Prediction with Machine Learning Introduction & Motivation PFAS-Free Process Transition Drift Prediction with Machine Learning addresses…
pfc, environmental & sustainability
PFC abatement is reduction of perfluorinated compound emissions from semiconductor process exhaust - Combustion plasma or catalytic systems decompose high-global-warming-gas…
pfc, environmental & sustainability
PFC Destruction Efficiency is the effectiveness of abatement systems in destroying perfluorinated compound emissions - It is a critical climate-impact metric for semiconductor…
partitioned global address space, coarray parallel model, upc language model, shmem programming
PGAS Programming Model is the parallel model that presents a global memory view while preserving data locality awareness. What It Covers Core concept: enables direct remote reads…
pgd, ai safety
PGD (Projected Gradient Descent) is the standard strong adversarial attack — an iterative first-order attack that takes multiple gradient ascent steps to maximize the loss within…
pgd, interpretability
PGD Attack is an iterative projected-gradient adversarial attack that refines perturbations over multiple steps - It is a strong first-order method for stress-testing model…
postgres, extension
pgvector: Vector Similarity for PostgreSQL Overview pgvector is an open-source extension for PostgreSQL that enables storing, querying, and indexing vectors. It turns the world's…
manufacturing equipment
pH Measurement is monitoring method that measures acidity or alkalinity of process fluids using electrochemical sensors - It is a core method in modern semiconductor AI…
Pharmacology: Drug Action, Dose Response, and Medication Safety Pharmacology studies how drugs act on biological systems and how bodies absorb, distribute, transform…
healthcare ai
Pharmacophore Modeling defines a drug not by its literal atomic structure or chemical bonds, but as a three-dimensional spatial arrangement of abstract chemical interaction points…
phase alignment accuracy prediction, phase alignment accuracy, alignment accuracy prediction, phase alignment, alignment accuracy, accuracy prediction
Phase Alignment Accuracy Prediction with Machine Learning Introduction & Motivation Phase Alignment Accuracy Prediction with Machine Learning addresses a central problem in phase…
phase-change cooling material, phase-change cooling, cooling material
Phase-Change Cooling Material Introduction & Motivation Phase-Change Cooling Material addresses a central problem in cooling process control and latent-heat-capacity prediction…