Knowledge Graph — Coursera Notes › Academic disciplines › Information Technology / Computer Science › Artificial Intelligence › Machine Learning & Data › Model Optimization
Quantization
concept · part of Model Optimization
Reducing numeric precision from 32-bit float to 16-bit or 8-bit integer for faster inference with minimal accuracy loss.
In QLoRA, quantization is applied to the pretrained model before adding low-rank adaptations.
Connections
- Used for Model Optimization
- Related to Pruning
- Related to Model Optimization
This is the text view of an interactive 3D knowledge graph — open this page with JavaScript enabled to explore it visually.