Trilu layer

testdata/dnn/onnx/data/iput_mha_rope_1.pb

@@ -0,0 +1,3 @@
+
Biput_mha_rope_1J�^)2?J��=QA9?d��>�|?e��>`n�>�On?�m�>8:B?���>e:m?
[?#&e?=?�>�*?��)?�,?3�P?̴c?�˓>�s?�%?^0?s��>�[m?��M>�c>W=ы�=�p�>��^>��x?#�j?>�	?�>s>z��>⥫>�5?ʪ0?���>��=�I�>��>��>HS?�o`?bn?�n >+�*>Ϫ	?�
+�<k-�>[)?�V?�<�� ?x�4?�D%?�q�<��M>4-�<V�^>�\J?�{?5�=?�2?��G;���>�N.?Jl>�L�>sa=��>߭�>$4a?�9r?�>�Gg?>v>ؕ�>���>�>�^?�!�>�'?�K�>�#�>D��<cc?���=�=�?d�?_?X�{?��=$�>#u>��1?$P,?��l?7�>6��=�^=?�4?� �>v��>ggb>ue�>.o�>B�O>~^�>�&�>)�>�f?�W�>��O?8.?��>V>���>/�=.�?�	+?�pr?��I?��>��?~�?=`?�}�=�j9?��2>
WD?%b�>U�>��>��l?��w>3?9?2�;>�a?&N�>�+?>?'?B�7?�3C?��>��=���>c�(?u�
+?_�?8�/?t�>��.?�`?2��>+RL?�E�>�r<^�<>T^	?�Wa?�mG?'�=��E?�?�&�>դ#?�T?�?��Z<�Jl?��?@�?M�T?3��>�7?|	�=:�L=79K?���>�s�>)?�� ?rD->m�A?Ҡ`?8�>�@�>

testdata/dnn/onnx/data/output_mha_rope.pb

@@ -0,0 +1,2 @@
+

Boutput_mha_ropeJ�

+�<@��J@`4�@AM@9�@Ό�@�;@X�@�k=@�lK@`ȅ@��Q@|�@�ȧ@e�;@&@i<@.J@���@E�L@�@>ۦ@��:@4�@�=@7K@.�@�Q@� @Dè@��;@Ej @��=@�L@,�@d�U@�#@$-�@_<@ȶ#@

testdata/dnn/onnx/data/output_tensor.pb

@@ -0,0 +1 @@
+

Boutput_tensorJ�
G�?��b@��@ȩ7@��@��@�&5@��@-�?Q�c@��@�7@��@�@G6@1�@�-�?��_@�]@�4@o�@��@}%5@��@���?�Wd@(o@�6@��@O�@��6@2�@���?��a@��@�f7@��@.�@��4@�@

testdata/dnn/onnx/generate_data_onnx_trilu_layer.py

@@ -0,0 +1,297 @@
+from onnx import TensorProto, version_converter
+import onnx
+from onnx.helper import (
+    make_model, make_node, make_graph, make_tensor_value_info,
+)
+import numpy as np
+import onnxruntime as ort
+import os
+
+testcases = [
+  {
+    "lab":"tril",
+    "X":\
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [9, 4, 1, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "Y":\
+      [[4, 0, 0, 0, 0],
+      [1, 2, 0, 0, 0],
+      [9, 4, 1, 0, 0],
+      [4, 3, 4, 2, 0]],
+    "upper": 0,
+    "k": 0
+  },
+  {
+    "lab":"tril_neg",
+    "X":\
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [9, 4, 1, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "Y":\
+      [[0, 0, 0, 0, 0],
+      [1, 0, 0, 0, 0],
+      [9, 4, 0, 0, 0],
+      [4, 3, 4, 0, 0]],
+    "upper": 0,
+    "k": -1
+  },
+  {
+    "lab":"tril_one_row",
+    "X":\
+      [[[6, 2, 4, 1, 6]],
+      [[8, 3, 8, 7, 0]],
+      [[2, 2, 9, 5, 9]]],
+    "Y":\
+      [[[6, 0, 0, 0, 0]],
+      [[8, 0, 0, 0, 0]],
+      [[2, 0, 0, 0, 0]]],
+    "upper": 0,
+  },
+  {
+    "lab":"tril_one_row1D",
+    "X":\
+      [6, 2, 4, 1, 6],
+    "Y":\
+      [6, 0, 0, 0, 0],
+    "upper": 0,
+  },
+  {
+    "lab":"tril_out_neg",
+    "X":\
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [9, 4, 1, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "Y":\
+      [[0, 0, 0, 0, 0],
+      [0, 0, 0, 0, 0],
+      [0, 0, 0, 0, 0],
+      [0, 0, 0, 0, 0]],
+    "upper": 0,
+    "k": -7
+  },
+  {
+    "lab":"tril_out_pos",
+    "X":\
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [9, 4, 1, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "Y":\
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [9, 4, 1, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "upper": 0,
+    "k": 6
+  },
+  {
+    "lab":"tril_pos",
+    "X":\
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [9, 4, 1, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "Y":\
+      [[4, 7, 3, 0, 0],
+      [1, 2, 8, 6, 0],
+      [9, 4, 1, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "upper": 0,
+    "k": 2
+  },
+  {
+    "lab":"tril_square",
+    "X":\
+      [[[0, 4, 3],
+        [2, 0, 9],
+        [8, 2, 5]],
+      [[2, 7, 2],
+        [2, 6, 0],
+        [2, 6, 5]]],
+    "Y":\
+      [[[0, 0, 0],
+        [2, 0, 0],
+        [8, 2, 5]],
+      [[2, 0, 0],
+        [2, 6, 0],
+        [2, 6, 5]]],
+    "upper": 0,
+  },
+  {
+    "lab":"tril_square_neg",
+    "X":\
+      [[[0, 4, 3],
+        [2, 0, 9],
+        [8, 2, 5]],
+      [[2, 7, 2],
+        [2, 6, 0],
+        [2, 6, 5]]],
+    "Y":\
+      [[[0, 0, 0],
+        [2, 0, 0],
+        [8, 2, 0]],
+      [[0, 0, 0],
+        [2, 0, 0],
+        [2, 6, 0]]],
+    "upper": 0,
+    "k": -1
+  },
+  {
+    "lab":"triu",
+    "X":\
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [9, 4, 0, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "Y":\
+      [[4, 7, 3, 7, 9],
+      [0, 2, 8, 6, 9],
+      [0, 0, 0, 8, 7],
+      [0, 0, 0, 2, 4]],
+  },
+  {
+    "lab":"triu_neg",
+    "X":\
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [9, 4, 0, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "Y":\
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [0, 4, 0, 8, 7],
+      [0, 0, 4, 2, 4]],
+    "k": -1
+  },
+  {
+    "lab":"triu_one_row",
+    "X":\
+      [[[1, 4, 9, 7, 1]],
+      [[9, 2, 8, 8, 4]],
+      [[3, 9, 7, 4, 2]]],
+    "Y":\
+      [[[0, 4, 9, 7, 1]],
+      [[0, 2, 8, 8, 4]],
+      [[0, 9, 7, 4, 2]]],
+    "k": 1,
+  },
+  {
+    "lab":"triu_out_neg_out",
+    "X":\
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [9, 4, 0, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "Y":\
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [9, 4, 0, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "k":-7
+  },
+  {
+    "lab":"triu_out_pos",
+    "X":
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [9, 4, 0, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "Y":
+      [[0, 0, 0, 0, 0],
+      [0, 0, 0, 0, 0],
+      [0, 0, 0, 0, 0],
+      [0, 0, 0, 0, 0]],
+    "k": 6
+  },
+  {
+    "lab":"triu_pos",
+    "X":
+      [[4, 7, 3, 7, 9],
+      [1, 2, 8, 6, 9],
+      [9, 4, 0, 8, 7],
+      [4, 3, 4, 2, 4]],
+    "Y":
+      [[0, 0, 3, 7, 9],
+      [0, 0, 0, 6, 9],
+      [0, 0, 0, 0, 7],
+      [0, 0, 0, 0, 0]],
+    "k":2
+  },
+  {
+    "lab": "triu_square",
+    "X":
+      [[[4, 6, 9],
+        [7, 5, 4],
+        [8, 1, 2]],
+      [[1, 4, 9],
+        [9, 6, 3],
+        [8, 9, 8]]],
+    "Y":
+      [[[4, 6, 9],
+        [0, 5, 4],
+        [0, 0, 2]],
+      [[1, 4, 9],
+        [0, 6, 3],
+        [0, 0, 8]]]
+  },
+  {
+    "lab": "triu_square_neg",
+    "X":
+      [[[4, 6, 9],
+        [7, 5, 4],
+        [8, 1, 2]],
+      [[1, 4, 9],
+        [9, 6, 3],
+        [8, 9, 8]]],
+    "Y":
+      [[[4, 6, 9],
+        [7, 5, 4],
+        [0, 1, 2]],
+      [[1, 4, 9],
+        [9, 6, 3],
+        [0, 9, 8]]],
+    "k": -1
+  }
+]
+
+for i in range(len(testcases)):
+    data_case = testcases[i]
+    k = data_case.get("k", 0)
+    upper = data_case.get("upper", 1)
+    lab= data_case.get("lab", "tril")
+
+    input= np.array(data_case["X"], dtype=np.float32)
+    output= np.array(data_case["Y"], dtype=np.float32)
+
+    x = make_tensor_value_info("x", TensorProto.FLOAT,  input.shape)
+    y = make_tensor_value_info("y", TensorProto.FLOAT,  output.shape)
+
+    k_ = make_tensor_value_info("k", TensorProto.INT64, [1])
+
+    node = onnx.helper.make_node(
+        "Trilu",
+        inputs=["x","k"],
+        outputs=["y"],
+        upper=upper,
+    )
+
+
+    graph = make_graph(
+        [node],
+        "trilu_graph",
+        [x, k_],
+        [y],
+    )
+
+    model = make_model(graph, producer_name="trilu_tril", ir_version=10)
+    onnx.save_model(model, os.path.join("models",f"trilu_{lab}.onnx"))
+
+    # save x and y as npy
+    np.save(os.path.join("data",f"input_trilu_{lab}_0.npy"), input.data)
+    np.save(os.path.join("data",f"input_trilu_{lab}_1.npy"), np.array([k], dtype=np.int64).data)
+
+    np.save(os.path.join("data",f"output_trilu_{lab}.npy"), output.data)

testdata/dnn/onnx/models/trilu_triu.onnx

@@ -0,0 +1,19 @@
+
+
+trilu_tril:h
+
+
x
+
k
y"Trilu*
+upper
�
trilu_graphZ
+
x
+

+
+Z
+
k
+
+
+
b
+
y
+

+
+B

testdata/dnn/onnx/models/trilu_triu_neg.onnx

@@ -0,0 +1,19 @@
+
+
+trilu_tril:h
+
+
x
+
k
y"Trilu*
+upper
�
trilu_graphZ
+
x
+

+
+Z
+
k
+
+
+
b
+
y
+

+
+B

testdata/dnn/onnx/models/trilu_triu_one_row.onnx

@@ -0,0 +1,21 @@
+
+
+trilu_tril:p
+
+
x
+
k
y"Trilu*
+upper
�
trilu_graphZ
+
x
+

+
+

+Z
+
k
+
+
+
b
+
y
+

+
+

+B

testdata/dnn/onnx/models/trilu_triu_out_neg_out.onnx

@@ -0,0 +1,19 @@
+
+
+trilu_tril:h
+
+
x
+
k
y"Trilu*
+upper
�
trilu_graphZ
+
x
+

+
+Z
+
k
+
+
+
b
+
y
+

+
+B