← Back to AI Factory Chat

AI Factory Glossary

217 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 4 of 5 (217 entries)

organic contamination, contamination

Organic residues on surfaces.

organic interposer, advanced packaging

Lower-cost organic substrate.

organic interposer, business & strategy

Organic interposers use polymer materials providing lower-cost alternative to silicon.

organosilicate glass (osg),organosilicate glass,osg,beol

Organic-inorganic hybrid low-k.

orientation imaging microscopy, oim, metrology

General term for orientation mapping.

orthogonal convolutions, ai safety

Constrain convolutions to be orthogonal.

orthogonal initialization, optimization

Initialize with orthogonal matrices.

osat (outsourced semiconductor assembly and test),osat,outsourced semiconductor assembly and test,industry

Third-party packaging and testing services.

osat, osat, business & strategy

Outsourced Semiconductor Assembly and Test companies provide backend services for packaged chips.

ostwald ripening, process

Large precipitates grow at expense of small.

otter,multimodal ai

Multi-modal instruction tuning.

out of control (ooc),out of control,ooc,spc

Process exceeding control limits.

out of distribution,ood,detect

OOD detection identifies inputs unlike training data. Model should abstain or flag. Safety critical.

out-of-control signals, spc

Indicators of special cause variation.

out-of-distribution, ai safety

Out-of-distribution inputs differ from training data potentially causing failures.

out-of-spec operation, production

Tool not meeting specifications.

out-of-vocabulary (oov),out-of-vocabulary,oov,nlp

Words not in the model's vocabulary (less common with subword tokenization).

outbound logistics, supply chain & logistics

Outbound logistics handles product distribution from factories to customers.

outgassing, contamination

Release of trapped gases.

outlier detection, data analysis

Find data points far from normal.

outlier detection, yield enhancement

Outlier detection identifies devices with abnormal parametric values indicating process excursions or latent reliability risks.

outlier,anomaly,remove

Detect and handle outliers. Remove, clip, or transform.

outlines,framework

Library for structured text generation and prompting.

outlines,structured,json

Outlines guarantees structured LLM output. JSON schema, regex. Reliable parsing.

outpainting, generative models

Extend image beyond boundaries.

outpainting, multimodal ai

Outpainting extends images beyond original boundaries generating consistent extrapolations.

outpainting,generative models

Extend image beyond its original boundaries.

output constraint, prompting techniques

Output constraints limit response characteristics like length style or content.

output filter, ai safety

Output filters block unsafe responses before delivery to users.

output filter,moderation,classifier

Filter model outputs with content classifiers. Block harmful content before returning to user.

output filtering,ai safety

Post-process model outputs to remove harmful sensitive or off-policy content.

output moderation, ai safety

Check generations before returning.

over-etch,etch

Extra etch time to ensure complete removal accounting for variations.

over-processing waste, manufacturing operations

Over-processing waste performs unnecessary operations or exceeds requirements.

over-refusal, ai safety

Decline benign requests unnecessarily.

over-sampling minority class, machine learning

Duplicate rare examples.

over-travel, advanced test & probe

Over-travel is additional probe displacement beyond touchdown ensuring stable contact and compensating for non-planarity.

overall yield,production

Composite yield from wafer to ship.

overconfidence, ai safety

Overconfidence occurs when models express higher certainty than warranted by accuracy.

overetch step,etch

Additional etch to clear residues ensuring completeness.

overfitting,underfit,generalization

Overfitting: model memorizes training data, fails on new data. Underfitting: model too simple. Goal is good generalization.

overkill,quality

Good part incorrectly rejected.

overlapping chunks, rag

Chunks with overlap for context.

overlapping process window, lithography

Region satisfying multiple constraints.

overlapping process windows, process

Find overlap of multiple constraints.

overlay error,lithography

Misalignment between layers measured and minimized.

overlay fingerprint, metrology

Characteristic overlay pattern across wafer.

overlay metrology,metrology

Measure alignment error between layers using optical or SEM.

overlay process window, metrology

Range of overlay values meeting specs.

overlay,lithography

Alignment accuracy between successive lithography layers.