Modelpublic

rdtand/LFM2.5-8B-A1B-PrismaQuant-6.5bit-vllm

sourceHugging Faceupdated 4mo agoView on Hugging Face

0likes31downloads

2 commits on main

26a9d414mo ago

PrismaQuant LFM2.5-8B-A1B mixed-precision (~6.58 bpp, NVFP4 experts), vLLM-servable; ToolEvalBench 70/100 = BF16 parity

rdtand

c2b83a94mo ago

initial commit

rdtand