added test data from LSTM layout feature

 + some refactoring of data generator

testdata/dnn/onnx/generate_onnx_models.py

@@ -1299,10 +1299,13 @@ def __init__(self, **params: Any) -> None:
         params[R] = weight_scale * np.ones(
             (1, number_of_gates * hidden_size, hidden_size)
         ).astype(np.float32)
+
         params[B] = np.ones((1, 2 * number_of_gates * hidden_size)).astype(np.float32)
 
-        params[H_0] = np.ones((1, batch_size, hidden_size)).astype(np.float32)
-        params[C_0] = np.ones((1, batch_size, hidden_size)).astype(np.float32)
+        if H_0 not in params and C_0 not in params:
+            params[H_0] = np.ones((1, batch_size, hidden_size)).astype(np.float32)
+            params[C_0] = np.ones((1, batch_size, hidden_size)).astype(np.float32)
+
         params[P] = weight_scale * np.ones((1, number_of_peepholes * hidden_size)).astype(
             np.float32)
 
@@ -1317,22 +1320,22 @@ def __init__(self, **params: Any) -> None:
             b = (
                 params[B]
                 if B in params
-                else np.ones(2 * number_of_gates * hidden_size, dtype=np.float32)
+                else np.ones(1, 2 * number_of_gates * hidden_size, dtype=np.float32)
             )
             p = (
                 params[P]
                 if P in params
-                else np.ones(number_of_peepholes * hidden_size, dtype=np.float32)
+                else np.ones(1, number_of_peepholes * hidden_size, dtype=np.float32)
             )
             h_0 = (
                 params[H_0]
                 if H_0 in params
-                else np.ones((batch_size, hidden_size), dtype=np.float32)
+                else np.ones((1, batch_size, hidden_size), dtype=np.float32)
             )
             c_0 = (
                 params[C_0]
                 if C_0 in params
-                else np.ones((batch_size, hidden_size), dtype=np.float32)
+                else np.ones((1, batch_size, hidden_size), dtype=np.float32)
             )
 
             self.X = x
@@ -1375,7 +1378,6 @@ def step(self) -> Tuple[np.ndarray, np.ndarray]:
         H_t = self.H_0
         C_t = self.C_0
         for x in np.split(self.X, self.X.shape[0], axis=0):
-            print(x.shape, self.W.shape)
             gates = (
                 np.dot(x, np.transpose(self.W))
                 + np.dot(H_t, np.transpose(self.R))
@@ -1391,7 +1393,6 @@ def step(self) -> Tuple[np.ndarray, np.ndarray]:
             h_list.append(H)
             H_t = H
             C_t = C
-            print(H.shape)
 
         concatenated = np.concatenate(h_list)
         if self.num_directions == 1:
@@ -1483,14 +1484,12 @@ def save_model_and_data_lstm_layout(lstm, layout, x, hx, cx, basename):
     onnx.checker.check_model(m)
 
     for i, data in enumerate(inputs):
-        print(data.shape)
         input_files = os.path.join("data", "input_" + basename + f"_{str(i)}.npy")
         data = data.astype(np.float32)
         np.save(input_files, np.ascontiguousarray(data.data))
 
     output_file = os.path.join("data", "output_" + basename + ".npy")
     Y_h = data.astype(np.float32)
-    print(Y_h.shape)
     np.save(output_file, np.ascontiguousarray(Y_h.data))
 
 
@@ -1510,7 +1509,14 @@ def save_model_and_data_lstm_layout(lstm, layout, x, hx, cx, basename):
     else:
         x = np.ones((seq_length, batch_size, input_size)).astype(np.float32)
 
-    lstm = LayoutLSTM(X=x, layout=layout, hidden_size=hidden_size, input_size=input_size)
+    lstm = LayoutLSTM(
+        X=x,
+        initial_h=hx,
+        initial_c=cx,
+        layout=layout,
+        hidden_size=hidden_size,
+        input_size=input_size
+    )
 
     save_model_and_data_lstm_layout(lstm, layout, x, hx, cx, f"lstm_layout_{str(layout)}")