[TPU] Re-enable the Pallas MoE kernel (#18025)
Signed-off-by: Michael Goin <mgoin64@gmail.com>
This commit is contained in:
@@ -50,8 +50,7 @@ if is_rocm_aiter_moe_enabled():
|
||||
else:
|
||||
from vllm.model_executor.layers.fused_moe.fused_moe import grouped_topk
|
||||
if current_platform.is_tpu():
|
||||
# the iterative moe implementation is used until the moe_pallas is fixed
|
||||
from .moe_torch_iterative import fused_moe as fused_moe_pallas
|
||||
from .moe_pallas import fused_moe as fused_moe_pallas
|
||||
else:
|
||||
fused_moe_pallas = None # type: ignore
|
||||
logger = init_logger(__name__)
|
||||
|
||||
Reference in New Issue
Block a user