AI Factory Glossary

chunk overlap, rag

Chunk overlap prevents splitting important information across boundaries.

chunk overlap,rag

Overlap between chunks to avoid splitting important context at boundaries.

chunk size optimization, rag

Find optimal chunk length.

chunk size optimization,rag

Tune size of text chunks to balance context and retrieval precision.

chunk size, rag

Chunk size determines granularity of text segments for retrieval.

chunked prefill,disaggregated

Chunked prefill processes long prompts in chunks. Disaggregated prefill separates from decode. Better scheduling.

chunking,text splitting,overlap

Chunking splits documents for embedding/retrieval. Use semantic boundaries, 512-1024 tokens, with overlap for continuity.

ci cd,pipeline,automate

CI/CD automates test and deploy. GitHub Actions, GitLab CI.

ci,cd,devops,pipeline

I can outline CI/CD pipelines (build, test, deploy), recommend tools, and show how to automate key checks.

circuit breaker pattern,software engineering

Stop calling failing service to prevent cascade.

circuit breaker, llm optimization

Circuit breakers halt requests to failing services allowing recovery time.

circuit breaker,fallback,degradation

Circuit breaker stops calling failing services. Fallback to cached response or simpler model. Graceful degradation.

circuit discovery, explainable ai

Find functional circuits in networks.

circuit edit,analysis

Modify circuit using FIB for debug.

circular economy, environmental & sustainability

Circular economy approaches maximize material reuse refurbishment and recycling minimizing waste in semiconductor manufacturing lifecycle.

citation accuracy, evaluation

Whether citations support claims.

citation analysis,legal ai

Analyze legal citation networks.

citation generation, rag

Generate proper citations.

citation, evaluation

Citations reference sources supporting generated information.

citation,attribution,source

Citations link generated text to source documents. Important for trust and fact-checking. Include doc IDs or quotes.

citation,reference,format

Format citations. APA, MLA, Chicago. Bibliography generation.

ckan, ckan, recommendation systems

Collaborative Knowledge-Aware Attention Network jointly learns from user-item and item-entity graphs for recommendations.

cky algorithm, cky, structured prediction

Cocke-Kasami-Younger algorithm performs bottom-up chart parsing for context-free grammars in cubic time complexity.

cl4srec, recommendation systems

Contrastive Learning for Sequential Recommendation uses augmentation and InfoNCE loss for representation learning.

claim detection,nlp

Identify factual claims in text.

claimbuster,nlp

System for identifying check-worthy claims.

clarinet, audio & speech

ClariNet combines Gaussian inverse autoregressive flow with WaveNet for parallel fast speech synthesis.

class token, cls, computer vision

Special token for classification in ViT.

class weight,imbalanced,loss

Class weights adjust loss for imbalance. Higher weight for minority.

class-balanced loss, machine learning

Weight classes by inverse frequency.

class-incremental learning,continual learning

Add new classes over time.

classical planning,ai agent

AI planning using STRIPS or PDDL.

classification for binning, data analysis

Classify dies into performance bins.

classification,multiclass,predict

Classification predicts categories. Binary or multiclass.

classifier guidance,generative models

Use classifier gradients to guide generation.

classifier-based filtering, data quality

Use classifier to select data.

classifier-free guidance, cfg, generative models

Guidance without separate classifier.

classifier-free guidance, multimodal ai

Classifier-free guidance steers diffusion models using conditional and unconditional score estimates.

classifier-free guidance,generative models

Control generation strength by mixing conditional and unconditional predictions.

classify,categorize,label

Classify text into categories. Topic, intent, type.

claude vision,foundation model

Anthropic's vision-capable model.

claude,foundation model

Anthropic's helpful honest and harmless AI assistant.

clause extraction,legal ai

Identify key clauses in contracts.

clcrec, recommendation systems

Contrastive Learning for Cold-start Recommendation uses self-supervised learning to improve initialization.

clean-label poisoning, ai safety

Poison with correctly labeled examples.

cleanlab,data quality,label

Cleanlab finds label errors in data. Data-centric AI. Improve training data.

cleanliness requirements, quality

Specifications for contamination levels.

cleanroom behavior, facility

Best practices to minimize contamination.

cleanroom classification,facility

Rating of particle count per volume (Class 1 10 100 etc).

cleanroom garment, manufacturing operations

Cleanroom garments prevent particle shedding from personnel into environment.