[MoE Refactor][5/N] Isolate zero expert to LongCatFlash (#28891)

Signed-off-by: baonudesifeizhai <85092850+baonudesifeizhai@users.noreply.github.com>
Signed-off-by: Dongjie Zou <85092850+baonudesifeizhai@users.noreply.github.com>
Signed-off-by: baonudesifeizhai <baonudesifeizhai@gmail.com>
Signed-off-by: Robert Shaw <robertgshaw2@gmail.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <robertgshaw2@gmail.com>
This commit is contained in:
baonudesifeizhai
2025-12-20 18:22:04 +00:00
committed by GitHub
co-authored by Robert Shaw Robert Shaw
parent 560ae9638c
commit 54c8924384
19 changed files with 264 additions and 109 deletions
@@ -796,7 +796,7 @@ class ModelOptFp8MoEMethod(FusedMoEMethodBase):
)
# Expert selection
topk_weights, topk_ids, _ = layer.select_experts(
topk_weights, topk_ids = layer.select_experts(
hidden_states=x,
router_logits=router_logits,
)
@@ -1599,7 +1599,7 @@ class ModelOptNvFp4FusedMoE(FusedMoEMethodBase):
x_routing, _ = x
else:
x_routing = x
topk_weights, topk_ids, _ = layer.select_experts(
topk_weights, topk_ids = layer.select_experts(
hidden_states=x_routing,
router_logits=router_logits,
)