quantization

3 articles tagged “quantization”, most recent first.

Sep 10, 20261 min read

Hugging Face author maps tensor sensitivity to improve GGUF quantization

A 96-hour experiment across 1,000-plus model variants points to more precise, per-tensor bit allocation for llama.cpp quantization.

Sep 10, 20261 min read

TurboQuant brings ultra-low-bit compression to vector search

Hugging Face explains how TurboQuant cuts embedding storage while preserving more search quality than aggressive binary compression.

Sep 3, 20261 min read

A Quantized 2B Vision Model Outperformed 8B on a Quest 3 Test

Tests on Meta’s Quest 3 suggest careful quantization can matter more than parameter count for on-device vision tasks.