[Quantization][MoE] remove unused ep logic from moe marlin (#31571)
Signed-off-by: Jinzhen Lin <jinzhen.ljz@antgroup.com> Co-authored-by: Michael Goin <mgoin64@gmail.com>
This commit is contained in:
co-authored by
Michael Goin
parent
28c94770ad
commit
2f4bdee61e
@@ -2162,7 +2162,6 @@ def moe_wna16_marlin_gemm(
|
||||
moe_block_size: int,
|
||||
top_k: int,
|
||||
mul_topk_weights: bool,
|
||||
is_ep: bool,
|
||||
b_q_type: ScalarType,
|
||||
size_m: int,
|
||||
size_n: int,
|
||||
@@ -2194,7 +2193,6 @@ def moe_wna16_marlin_gemm(
|
||||
moe_block_size,
|
||||
top_k,
|
||||
mul_topk_weights,
|
||||
is_ep,
|
||||
b_q_type.id,
|
||||
size_m,
|
||||
size_n,
|
||||
@@ -2256,7 +2254,6 @@ if hasattr(torch.ops, "_moe_C") and hasattr(torch.ops._moe_C, "marlin_gemm_moe")
|
||||
moe_block_size: int,
|
||||
top_k: int,
|
||||
mul_topk_weights: bool,
|
||||
is_ep: bool,
|
||||
b_q_type: ScalarType,
|
||||
size_m: int,
|
||||
size_n: int,
|
||||
|
||||
Reference in New Issue
Block a user