Open Models Almanac
Methodsoptimizationquantization-aware training

single mention · optimization

Per-Block Scalar Scaling

Introducing a scalar scale at each block to bound activation ranges for stable FP16 inference.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See quantization-aware training for the methods filed beside it.

Filed under optimization › quantization-aware training