Meaning
An attention variant where multiple query heads share a single key-value head to compress memory usage.
Articles Featuring This Term
Related Terms
- Text-to-Speech (TTS)
- Rotary Position Embedding Relative Position Rotation
- Sensitivity and Specificity Detection vs False-Alarm Rates
- DualPipe Computation-Communication Overlap Parallelism
- Convolutional Neural Network (CNN) Visual Pattern Deep Learning
- Autoregressive Model Sequential Next-Token Generator
- Side-Channel Attack Hardware Telemetry Leakage Attack
- MuJoCo Physics Engine Multi-Joint Dynamics Simulation
- Post-Quantum Cryptography Quantum-Resistant Cryptography



