Antirez recipe for MoE quantization: Q2 uses IQ2_XXS gate/up + Q2_K down; Q4 uses Q4_K gate/up + MXFP4 down. Everything else unchanged.
Nicolo' D'Evangelista
Ninnix96
AI & ML interests
None yet
Recent Activity
updated a collection about 6 hours ago
Antirez's recipe updated a model about 6 hours ago
Ninnix96/Ling-3.0-tiny-gguf published a model about 6 hours ago
Ninnix96/Ling-3.0-tiny-ggufOrganizations
None yet
models 6
Ninnix96/Ling-3.0-tiny-gguf
Text Generation • 8B • Updated
Ninnix96/Qwengram-4B
Text Generation • 4B • Updated • 723 • 2
Ninnix96/Qwengram-2B
Text Generation • 2B • Updated • 983 • 4
Ninnix96/Qwengram-0.8B
Text Generation • 0.8B • Updated • 3.73k • 15
Ninnix96/KAT-Coder-V2.5-Dev-gguf
Text Generation • 35B • Updated • 92 • 1
Ninnix96/Qwen3.6-35B-A3B-gguf
Text Generation • 35B • Updated • 649 • 2
datasets 0
None public yet