← Back to AI Factory Chat

AI Factory Glossary

751 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 8 of 16 (751 entries)

mim decap, mim, signal & power integrity

Metal-Insulator-Metal decaps offer lower series resistance than MOS caps with moderate density.

min tokens, llm optimization

Min tokens ensures generation continues until minimum length.

min-p sampling, llm optimization

Min-p sampling sets minimum probability relative to top token.

min-p sampling, text generation

Minimum probability threshold.

mincut pool, graph neural networks

MinCut pooling learns cluster assignments by minimizing normalized cut objectives creating coarsened graphs with balanced communities.

minerva,google,math

Minerva is Google math reasoning model. Trained on math data.

minhash for deduplication, data quality

Efficient similarity detection.

mini-batch online learning,machine learning

Update with small batches of streaming data.

minigpt-4,multimodal ai

Vision-language model aligned with GPT-4.

minigpt,vision language,open

MiniGPT connects vision encoder to LLM. Open source multimodal. Research project.

minimum batch, manufacturing operations

Minimum batch requirements specify smallest lot size for processing.

minor nonconformance, quality & reliability

Minor nonconformances are isolated deviations not significantly impacting quality.

minor stoppage, manufacturing operations

Minor stoppages are brief interruptions under several minutes often not formally recorded.

minority carrier lifetime, device physics

Time before recombination.

mip-nerf, multimodal ai

Mip-NeRF anti-aliases NeRF by integrating over conical frustums rather than points.

mirostat sampling, text generation

Perplexity-controlled sampling.

mirostat, llm optimization

Mirostat dynamically adjusts temperature maintaining target perplexity.

misfit dislocations, defects

Dislocations from lattice mismatch.

mish, neural architecture

Smooth activation x*tanh(softplus(x)).

misinformation detection,nlp

Identify false or misleading information.

misorientation analysis, metrology

Quantify crystal orientation differences.

misr, misr, advanced test & probe

Multiple Input Signature Register compresses test responses into compact signatures for comparison with expected values in BIST.

missing modality handling, multimodal ai

Handle incomplete multimodal data.

missing values,impute,handle

Handle missing values: impute mean, median, or model-based.

mistake-proofing, quality

Prevent errors in processes.

mistake-proofing, quality & reliability

Mistake-proofing designs processes and equipment to prevent or detect errors immediately.

mistral,foundation model

Efficient open-source language model with sliding window attention.

mix-and-match, business

Combine chiplets from different sources/technologies.

mixed integer linear programming verification, milp, ai safety

Encode network as MILP for verification.

mixed model production, manufacturing operations

Mixed model production manufactures multiple products on same line enabling variety without dedicated resources.

mixed precision training,model training

Use lower precision (FP16) for some operations to speed up and save memory.

mixed precision,amp,automatic

Automatic Mixed Precision (AMP) uses FP16/BF16 where safe. 2x memory savings, faster compute. Loss scaling prevents underflow.

mixed precision,fp16,bf16,amp

Mixed precision uses FP16/BF16 for speed and memory, FP32 for stability. AMP automates this. 2x speedup on modern GPUs.

mixed-precision training, model optimization

Mixed-precision training uses different numeric precisions for different operations balancing speed and accuracy.

mixmatch, advanced training

MixMatch unifies consistency regularization entropy minimization and MixUp for semi-supervised learning with unlabeled data.

mixmatch, semi-supervised learning

Combine consistency augmentation and pseudo-labeling.

mixtral,foundation model

Mixture of Experts version of Mistral.

mixture design, doe

Optimize formulations where components sum to constant.

mixture of agents (moa),mixture of agents,moa,multi-agent

Route queries to different specialized agents based on task type.

mixture of depths (mod),mixture of depths,mod,llm architecture

Dynamic computation allocation across layers based on input complexity.

mixture of depths advanced, llm architecture

Dynamically allocate computation across transformer layers based on token importance.

mixture of depths, llm architecture

Mixture of depths dynamically allocates computation across layers per token.

mixture of experts (moe),mixture of experts,moe,model architecture

Route each input to a few specialized expert networks instead of all parameters.

mixture-of-experts for multi-task, multi-task learning

Route tasks to experts.

mixup / cutmix,data augmentation

Blend training examples together to improve robustness.

mixup for vit, computer vision

Blend entire images and labels.

mixup text, advanced training

MixUp for text combines embeddings of two examples and interpolates their labels for training with continuous semantic augmentation.

mixup, data augmentation

Linearly interpolate training examples.

mixup,blend,regularize

Mixup blends training examples. Regularization, smoother predictions.

mlc llm,universal,compile

MLC LLM provides universal LLM deployment. Compile to any device.