qwen3_5_moe¶
Model type |
|
Class |
|
Task |
|
Source |
|
Description¶
Qwen3.5-MoE causal language model.
Combines the hybrid DeltaNet/full-attention architecture of Qwen3.5 with Mixture-of-Experts FFN layers that include a shared expert gated by sigmoid.
Usage¶
mobius build --model <MODEL_ID> output_dir/
from mobius import build
model = build("<MODEL_ID>")