Modelpublic

Avesed/Qwen3.8-27B-INT8-W8A8

sourceHugging Faceapache-2.0updated 1mo agoView on Hugging Face

9likes7.2kdownloads

3 commits on main

86b84271mo ago

model card: GSM8K 96.8% (N=250, boxed protocol)

Avesed

c264fc91mo ago

Qwen3.8-27B W8A8 (int8 channel weights + int8 dynamic acts) + bf16 MTP head

Avesed

c52aa821mo ago

initial commit

Avesed