syntax error fixes

Signed-off-by: Neta Zmora <[email protected]>

Author: nzmora-nvidia

tensorrt_llm/_torch/auto_deploy/transform/library/fused_moe.py

@@ -710,7 +710,7 @@ def get_param_or_buffer(target):
         # Create new node with get_attr for stacked parameters
         with graph.inserting_before(node):
             new_node = graph.call_function(
-                torch.ops.auto_deploy.trtllm_quant_fp8moe_fused,
+                torch.ops.auto_deploy.trtllm_quant_fp8_moe_fused,
                 args=(
                     hidden_states,
                     selected_experts,

tests/unittest/_torch/auto_deploy/unit/singlegpu/custom_ops/test_trtllm_moe.py

@@ -376,7 +376,7 @@ def dequantize_weights(w31_weight, w2_weight, w31_scales, w2_scales):
 
     activation_type = _activation_type_from_str(activation_func)
 
-    ad_test_output = torch.ops.auto_deploy.trtllm_quant_fp8moe_fused(
+    ad_test_output = torch.ops.auto_deploy.trtllm_quant_fp8_moe_fused(
         x,  # Note! unquantized input is expected
         selected_experts.to(torch.int),
         routing_weights,