diff --git a/tests/test_optimize.py b/tests/test_optimize.py
index 75a8f66..db153b5 100644
--- a/tests/test_optimize.py
+++ b/tests/test_optimize.py
@@ -12,17 +12,18 @@
 from .op_test_utils import check_op_type_count
 
 test_cases = [
-    ['encoder', 'transformer', 4, 256, 3, 0, False],
+    ['encoder', 'transformer', 4, 256, 'Attention', 3, False],
     # ['encoder', 'contextual_block_transformer', 4, 256, 3, 0, False],
-    ['encoder', 'transformer', 4, 256, 3, 0, True],
+    ['encoder', 'transformer', 4, 256, 'Attention', 3, True],
+    ['encoder', 'conformer_rpe_latest', 4, 256, 'RelPosAttention', 3, True],
     # ['encoder', 'contextual_block_transformer', 4, 256, 3, 0, True],
-    ['decoder', 'transformer', 4, 256, 0, 6, True],
-    ['lm', 'transformer_pe', 4, 256, 0, 3, True],
-    ['lm', 'transformer', 4, 256, 0, 2, True],
+    ['decoder', 'transformer', 4, 256, 'CrossAttention', 6, True],
+    ['lm', 'transformer_pe', 4, 256, 'CrossAttention', 3, True],
+    ['lm', 'transformer', 4, 256, 'CrossAttention', 2, True],
 ]
 
-@pytest.mark.parametrize('model_type, model_name, n_head, h_size, n_att, n_cross_att, use_custom_ort', test_cases)
-def test_optimize(model_type, model_name, n_head, h_size, n_att, n_cross_att, use_custom_ort, model_export):
+@pytest.mark.parametrize('model_type, model_name, n_head, h_size, node_name, node_num, use_custom_ort', test_cases)
+def test_optimize(model_type, model_name, n_head, h_size, node_name, node_num, use_custom_ort, model_export):
     export_dir = model_export.cache_dir / 'test' / \
         model_type / f'cache_{model_name}'
     output_dir = model_export.cache_dir / 'test' / \
@@ -45,11 +46,6 @@ def test_optimize(model_type, model_name, n_head, h_size, n_att, n_cross_att, us
     )
     
     # load the optimized model and check if the number of fused nodes is correct.
-    nodes = {}
-    if n_att > 0:
-        nodes['Attention'] = n_att
-    if n_cross_att > 0:
-        nodes['CrossAttention'] = n_cross_att
-        
+    nodes = {node_name : node_num}
     check_op_type_count(str(output_dir / model_name), **nodes)
     
\ No newline at end of file