← Back to AI Factory Chat

AI Factory Glossary

169 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 4 (169 entries)

lora for diffusion,generative models

Efficient fine-tuning of diffusion models with low-rank adapters.

lora merging, generative models

Combine multiple LoRAs.

loss scaling,model training

Multiply loss by constant to prevent gradients from underflowing in FP16.

loss spike,instability,training

Loss spikes indicate training instability. Reduce LR, check data, add gradient clipping. May need to restart.

loss spikes, training phenomena

Sudden increases in loss during training.

lot sizing, supply chain & logistics

Lot sizing determines optimal production or order quantities balancing setup costs and inventory.

lottery ticket hypothesis, model optimization

Lottery ticket hypothesis posits that dense networks contain sparse subnetworks trainable to full accuracy.

lottery ticket hypothesis,model training

Sparse subnetworks that train from scratch.

louvain algorithm, graph algorithms

Fast community detection method.

low-angle grain boundary, defects

Small misorientation between grains.

low-precision training, optimization

Use FP16 or BF16 for training.

low-rank factorization, model optimization

Low-rank factorization decomposes weight matrices into products of smaller matrices.

low-rank tensor fusion, multimodal ai

Efficient tensor fusion.

lp norm constraints, ai safety

Bound perturbations in Lp norm.

lru cache, lru, llm optimization

Least Recently Used cache evicts oldest accessed entries maintaining frequently used items.

lstm anomaly, lstm, time series models

LSTM-based anomaly detection flags time steps with high prediction error or unusual hidden states.

lstm-vae anomaly, lstm-vae, time series models

LSTM-VAE combines variational autoencoders with LSTM networks to detect anomalies in sequential data through reconstruction probability thresholds.

lstnet, time series models

Long- and Short-term Time-series Network combines CNNs and RNNs with skip connections for multivariate forecasting.

lvi, lvi, failure analysis advanced

Laser Voltage Imaging spatially maps voltage distributions across die surface revealing shorts or voltage drops.