[MoE Refactor] EPLB refactoring for FusedMoE (#41055)
Signed-off-by: Bill Nell <bnell@redhat.com> Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
This commit is contained in:
@@ -940,6 +940,17 @@ class EplbLayerState:
|
||||
GPU work.
|
||||
"""
|
||||
|
||||
def set_layer_state(
|
||||
self,
|
||||
moe_layer_idx: int,
|
||||
expert_load_view: torch.Tensor,
|
||||
logical_to_physical_map: torch.Tensor,
|
||||
logical_replica_count: torch.Tensor,
|
||||
) -> None:
|
||||
self.expert_load_view = expert_load_view[moe_layer_idx]
|
||||
self.logical_to_physical_map = logical_to_physical_map[moe_layer_idx]
|
||||
self.logical_replica_count = logical_replica_count[moe_layer_idx]
|
||||
|
||||
|
||||
def _node_count_with_rank_mapping(
|
||||
pg: ProcessGroup | StatelessProcessGroup,
|
||||
|
||||
Reference in New Issue
Block a user