NVFP4 quantization
Named by 5 kept items in this workspace.
Search every workspace for NVFP4 quantization. It matches the name as written, so it can find other things that share it.
The items that name it 5
Recovers 99.62% of median BF16 accuracy for Nemotron 3.5 and is used for Mistral Large 4 experts.
Meta's Muse Spark Open checkpoint quantizes only the experts, cutting weights from about 1.6TB to about 530GB.
All techniques in this workspace.