quantized

metrale_model_layers::weight_map::model_a

Function quantized 

Source 
pub fn quantized(
    store: &WeightStore,
    prefix: &str,
    gpu: &dyn GpuBackend,
) -> Result<QuantizedWeight>
Expand description

2026-09-25: Build a QuantizedWeight from ModelOpt NVFP4 keys: weight, weight_scale, the FP32 scalar weight_scale_2 (read to the host), and an optional input_scale.