Open Models Almanac
Methodsinference & servinginference quantization

single mention · inference & serving

NVFP4 W4A4 quantization of MTP experts

MTP experts are quantized using NVFP4 with 4-bit weights and 4-bit activations.

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

Filed under inference & serving › inference quantization