The bf16 layers are compressed using the Dfloat11 lossless compression mixed with nvfp4 layers.
Koston
ApacheOne
AI & ML interests
ExplORe, Comfy models for nodes
Recent Activity
updated a model about 15 hours ago
ApacheOne/HSWQ-fp8-SD15 published a model about 15 hours ago
ApacheOne/HSWQ-fp8-SD15 updated a model about 20 hours ago
ApacheOne/HSWQ-fp8-IllustriousOrganizations
The bf16 layers are compressed using the Dfloat11 lossless compression mixed with nvfp4 layers.
collection of mixed dtypes between weight layers, 3d matrix layers and 2d blacklisted layers
Group of Models for nodes or depens.
cleaning(WIP) seems like a policy update blocked even my cleaned out repos. even non-NFAA content repos were effected also.
A collection of hunyuan loras in alphabetical order. #-Z.