multi-prompt composition
prompting
Multi-prompt composition is the technique of combining multiple prompt segments to blend concepts, styles, or constraints in one generation run - it supports structured control…
29,259 technical terms and definitions
prompting
Multi-prompt composition is the technique of combining multiple prompt segments to blend concepts, styles, or constraints in one generation run - it supports structured control…
ha system, active active, active passive, multi region, failover, availability nines, error budget
High availability designs a service to remain usable through routine failures, maintenance and demand changes within a stated availability objective. AI serving needs redundant…
rag
Multi-Query is a retrieval strategy that generates multiple reformulated queries from one user request to improve evidence coverage - It is a core method in modern RAG…
mqa, efficient
Multi-query attention (MQA) has every single query head share one single common key-and-value projection, rather than each individual query head instead using its own separate…
grouped query attention GQA, attention heads reduction, inference efficiency, KV cache
Multi-Query and Grouped Query Attention (GQA) are attention variants that share key-value representations across multiple query heads — reducing KV cache memory by 8-16x…
multi-query attention, mqa, llm architecture
Multi-head attention (MHA), multi-query attention (MQA), and grouped-query attention (GQA) are three ways to wire the key and value projections of a Transformer's attention layer.…
optimization
Multi-query KV cache is the attention design where multiple query heads share a single set of key and value heads to reduce KV cache size and memory bandwidth - it is widely used…
rag
Multi-query retrieval generates query variations to achieve broader document coverage. Mechanism: Original query → LLM generates N alternative phrasings → retrieve with each →…
rag
Multi-query retrieval is the strategy of generating multiple query variants for one information need and retrieving with each to improve coverage - it increases recall…
active active architecture, active passive failover, geo redundancy, cloud disaster recovery
Multi-Region Deployment is the architecture practice of running an application and its critical data services across two or more geographic regions so that a regional outage…
multi region deployment strategy, multi region deployment, region deployment strategy, multi region, region deployment, deployment strategy
Multi Region Deployment Strategy Introduction & Motivation Multi Region Deployment Strategy addresses a central problem in cloud process control and region-latency-user-experience…
Multi-Relational Graph Embeddings Introduction & Motivation Multi-Relational Graph Embeddings addresses a central problem in graph network representation…
multimodal ai
Multi-Resolution Hash is a coordinate encoding technique that stores learned features in hierarchical hash tables - It captures both coarse and fine spatial detail with compact…
3d vision
Multi-resolution hash tables is the stacked hashed feature grids at increasing resolutions used to represent spatial detail across scales - they are the core structure behind fast…
computer vision
Multi-Resolution Training is a training strategy that exposes the model to inputs at multiple spatial resolutions during training — enabling the model to learn features…
optimization
Multi-Response Optimization is the simultaneous optimization of multiple quality characteristics (CD, thickness, uniformity, defects) — finding process conditions that jointly…
generative models
Multi-scale discriminator is the GAN discriminator design that evaluates generated images at multiple spatial resolutions to capture both global layout and local texture quality…
Multi-Scale Discriminator Introduction & Motivation Multi-Scale Discriminator addresses a central problem in generative generation and scale-invariance-improvement synthesis: how…
multimodal ai
Multi-Scale Generation is generation strategies that model and refine content at multiple spatial scales - It supports coherent global structure with detailed local textures. What…
multiscale modeling, HMM method, level set, Knudsen number, scale bridging, hierarchical modeling, atomistic to continuum
Semiconductor Manufacturing: Multi-Scale Problems and Mathematical Modeling 1. The Multi-Scale Hierarchy Semiconductor manufacturing spans roughly 12 orders of magnitude in length…
inference
Multi-Scale Testing is a test-time technique that runs inference at multiple input resolutions and combines the results — detecting objects or segmenting scenes more accurately…
computer vision
MViT (Multi-Scale Vision Transformer) is the pyramidal transformer architecture that progressively reduces spatial resolution while increasing channel depth so the network…
multi sensor fusion, camera lidar radar fusion, imu fusion, feature fusion, state estimation
Sensor fusion combines measurements from multiple sensors or times into a more complete, accurate, and robust state estimate. Cameras provide semantics, lidar provides geometry…
multi-site probe test correlation prediction, multi-site probe test, probe test correlation, test correlation prediction, multi-site probe, probe test, test correlation
Multi-Site Probe Test Correlation Prediction with Machine Learning Introduction & Motivation Multi-Site Probe Test Correlation Prediction with Machine Learning addresses a central…
multi site test cell correlation and parallel test consistency with machine learning, multi, multi-site test cell, test cell correlation, multi-site test
Multi-Site Test Cell Correlation and Parallel Test Consistency with Machine Learning Introduction & Motivation Multi-Site Test Cell Correlation and Parallel Test Consistency…
multi-site test yield correlation prediction, multi-site test yield, test yield correlation, yield correlation prediction, multi-site test, test yield, yield correlation
Multi-Site Test Yield Correlation Prediction with Machine Learning Introduction & Motivation Multi-Site Test Yield Correlation Prediction with Machine Learning addresses a central…
advanced test & probe
Multi-Site Testing is simultaneous testing of multiple devices in parallel on automated test equipment - It increases throughput and reduces cost per device by sharing tester…
quality & reliability
Multi-Skilled Operator is an operator certified to execute multiple process areas with consistent quality performance - It is a core method in modern semiconductor operational…
transfer learning
Multi-source domain adaptation is a transfer learning approach where knowledge is transferred from multiple different source domains simultaneously to improve performance on…
Multi-Source Domain Adaptation with Dynamic Weighting Introduction & Motivation Multi-Source Domain Adaptation with Dynamic Weighting addresses a central problem in transfer…
Multi-Source Domain Learning Introduction & Motivation Multi-Source Domain Learning addresses a central problem in vision processing and source-combination-benefit prediction: how…
ai safety
Multi-stage moderation is the defense-in-depth moderation architecture that applies multiple screening layers with increasing sophistication - staged filtering improves safety…
rag
Multi-Stage Retrieval is a funnel architecture that applies progressively stronger retrieval and ranking stages - It is a core method in modern retrieval and RAG execution…
recommendation systems
Multi-stakeholder recommendation is recommendation design that balances outcomes across users providers platforms and other stakeholders - Objective functions include multiple…
recommender systems
Multi-stakeholder recommendation balances interests of users, providers, and platforms — optimizing recommendations not just for user satisfaction but also for content creator…
etch
Multi-Step Etch Recipe is the sequential combination of distinct plasma etch steps — each with independently optimized chemistry, pressure, power, and time — designed to achieve…
Multi-Step Ahead Forecasting Introduction & Motivation Multi-Step Ahead Forecasting addresses a central problem in timeseries forecasting and forecast-error-accumulation…
ai safety
Multi-step jailbreak is a sophisticated adversarial technique that specifically bypasses an LLM's own built-in safety constraints through a carefully constructed sequence…
ai safety
Defending against multi-step jailbreaks genuinely requires a model, or a separate wrapping safety system, to actually track and correctly evaluate an entire ongoing conversation's…
audio & speech
Multi-Style Training is training with diverse acoustic styles such as reverberation, noise, and channel variation - It improves generalization by covering a broad range…
domain adaptation
Multi-Target Domain Adaptation (MTDA) is a domain adaptation setting where a model trained on a single source domain must simultaneously adapt to multiple target domains, each…
Multi-Task Active Learning with Shared Acquisition Introduction & Motivation Multi-Task Active Learning with Shared Acquisition addresses a central problem in (subject) learning…
Multi-Task Active Learning Introduction & Motivation Multi-Task Active Learning addresses a central problem in (subject) learning and efficient (target) annotation: how to Achieve…
Multi-Task Attention Networks Introduction & Motivation Multi-Task Attention Networks addresses a central problem in multitask learning across several related prediction…
Multi-Task Attention Networks (Extended Study) Introduction & Motivation Multi-Task Attention Networks (Extended Study) addresses a central problem in multitask learning across…
Multi-Task Attention Networks (Extended Study) Introduction & Motivation Multi-Task Attention Networks (Extended Study) addresses a central problem in multitask learning across…
Multi-Task Bandits Introduction & Motivation Multi-Task Bandits addresses a central problem in multitask learning across several related prediction objectives: how to Jointly…
Multi-Task Bandits (Extended Study) Introduction & Motivation Multi-Task Bandits (Extended Study) addresses a central problem in multitask learning across several related…
Multi-Task Bandits (Extended Study) Introduction & Motivation Multi-Task Bandits (Extended Study) addresses a central problem in multitask learning across several related…
Multi-Task Distillation Introduction & Motivation Multi-Task Distillation addresses a central problem in multitask learning across several related prediction objectives: how…