diff --git a/MindChemistry/applications/crystalflow/models/conditioning.py b/MindChemistry/applications/crystalflow/models/conditioning.py
index 799b8b6ad056b3a09cc2893b8870c38751504783..da21f09c5a99827e5589c52080eebb9a50444946 100644
--- a/MindChemistry/applications/crystalflow/models/conditioning.py
+++ b/MindChemistry/applications/crystalflow/models/conditioning.py
@@ -26,7 +26,7 @@ class GaussianExpansion(nn.Cell):
         Args:
             prop (Tensor): property values of (N_b x 1) shape.
         Returns:
-            Tensor: layer output of (N_b x N_g) shape.
+            Tensor. layer output of (N_b x N_g) shape.
         """
         prop = prop.reshape(prop.shape[0], -1)
         coeff = -0.5 / ops.pow(self.widths, 2)[None, :]
diff --git a/MindChemistry/applications/crystalflow/models/cspnet.py b/MindChemistry/applications/crystalflow/models/cspnet.py
index c5ae68956ab53227a9925685750823ff3ea033b1..2efd7763b86b1ae86da20330e2f43e51b716cbb8 100644
--- a/MindChemistry/applications/crystalflow/models/cspnet.py
+++ b/MindChemistry/applications/crystalflow/models/cspnet.py
@@ -51,7 +51,7 @@ class SinusoidsEmbedding(nn.Cell):
             x (Tensor): Distance
 
         Returns:
-            Tensor: Fourier embedding
+            Tensor. Fourier embedding
         """
         emb = ops.ExpandDims()(x, -1) * Tensor(
             self.frequencies,
@@ -70,7 +70,7 @@ def mul_mask(features, mask):
             and value 0 specifies the corresponding dimension of input tensor to be zero.
 
     Returns:
-        Tensor: Output tensor
+        Tensor. Output tensor
     """
     return ops.mul(features, ops.reshape(mask, (-1, 1)))
 
@@ -172,7 +172,7 @@ class CSPLayer(nn.Cell):
             edge_mask (Tensor): Edge mask for padded tensor. Shape: (num_edges,)
 
         Returns:
-            Tensor: The output tensor. Shape: (num_atoms, hidden_dim)
+            Tensor. The output tensor. Shape: (num_atoms, hidden_dim)
         """
         node_input = node_features
         node_features = mul_mask(node_features, node_mask)
diff --git a/MindChemistry/applications/crystalflow/models/cspnet_condition.py b/MindChemistry/applications/crystalflow/models/cspnet_condition.py
index 9ae1c2cddeeb509bf4ace6625934f8110e7b4e51..791ea4140f296a1e7a9755f7ef3a44114f79d491 100644
--- a/MindChemistry/applications/crystalflow/models/cspnet_condition.py
+++ b/MindChemistry/applications/crystalflow/models/cspnet_condition.py
@@ -51,7 +51,7 @@ class SinusoidsEmbedding(nn.Cell):
             x (Tensor): Distance
 
         Returns:
-            Tensor: Fourier embedding
+            Tensor. Fourier embedding
         """
         emb = ops.ExpandDims()(x, -1) * Tensor(
             self.frequencies,
@@ -70,7 +70,7 @@ def mul_mask(features, mask):
             and value 0 specifies the corresponding dimension of input tensor to be zero.
 
     Returns:
-        Tensor: Output tensor
+        Tensor. Output tensor
     """
     return ops.mul(features, ops.reshape(mask, (-1, 1)))
 
@@ -173,7 +173,7 @@ class CSPLayer(nn.Cell):
             edge_mask (Tensor): Edge mask for padded tensor. Shape: (num_edges,)
 
         Returns:
-            Tensor: The output tensor. Shape: (num_atoms, hidden_dim)
+            Tensor. The output tensor. Shape: (num_atoms, hidden_dim)
         """
         node_input = node_features
         node_features = mul_mask(node_features, node_mask)
diff --git a/MindChemistry/applications/crystalflow/models/diff_utils.py b/MindChemistry/applications/crystalflow/models/diff_utils.py
index e74e20a0071700a0d85852faf0410a5d4a5fd71a..7eff1cdf040a083ad68ab06575518ee50108965d 100644
--- a/MindChemistry/applications/crystalflow/models/diff_utils.py
+++ b/MindChemistry/applications/crystalflow/models/diff_utils.py
@@ -112,7 +112,7 @@ def d_log_p_wrapped_normal_ms(x, sigma, n=10, t=1.0):
         t (int): The period of wrapped normal distribution.  Defaults to 1.0.
 
     Returns:
-        Tensor: The score for the input value x.
+        Tensor. The score for the input value x.
     """
     p_ = 0
     for i in range(-n, n + 1):
diff --git a/MindChemistry/applications/crystalflow/models/flow.py b/MindChemistry/applications/crystalflow/models/flow.py
index 821f3520b1d1d5e8e6327cda5dae22e0233b6fbc..6098c5f8da3bdf8a4b81d8479c3b814aafc087e5 100644
--- a/MindChemistry/applications/crystalflow/models/flow.py
+++ b/MindChemistry/applications/crystalflow/models/flow.py
@@ -47,7 +47,7 @@ class SinusoidalTimeEmbeddings(nn.Cell):
             time (Tensor): flow time step
 
         Returns:
-            Tensor: Time embedding
+            Tensor. Time embedding
         """
         half_dim = self.dim // 2
         embeddings = math.log(10000) / (half_dim - 1)
@@ -65,7 +65,7 @@ def lattice_params_to_matrix_mindspore(lengths, angles):
         lengths (Tensor): Tensor of shape (N, 3), unit A
         angles (Tensor):: Tensor of shape (N, 3), unit degree
     Returns:
-        Tensor: Tensor of shape (N, 3, 3)
+        Tensor. Tensor of shape (N, 3, 3)
     """
     angles_r = ops.deg2rad(angles)
     coses = ops.cos(angles_r)
diff --git a/MindChemistry/applications/crystalflow/models/flow_condition.py b/MindChemistry/applications/crystalflow/models/flow_condition.py
index 938012b7c3201f938fc6f943d009f1822172b6b6..f3731f80de81079908a375fcdf26c6c52e253de8 100644
--- a/MindChemistry/applications/crystalflow/models/flow_condition.py
+++ b/MindChemistry/applications/crystalflow/models/flow_condition.py
@@ -48,7 +48,7 @@ class SinusoidalTimeEmbeddings(nn.Cell):
             time (Tensor): flow time step
 
         Returns:
-            Tensor: Time embedding
+            Tensor. Time embedding
         """
         half_dim = self.dim // 2
         embeddings = math.log(10000) / (half_dim - 1)
@@ -66,7 +66,7 @@ def lattice_params_to_matrix_mindspore(lengths, angles):
         lengths (Tensor): Tensor of shape (N, 3), unit A
         angles (Tensor):: Tensor of shape (N, 3), unit degree
     Returns:
-        Tensor: Tensor of shape (N, 3, 3)
+        Tensor. Tensor of shape (N, 3, 3)
     """
     angles_r = ops.deg2rad(angles)
     coses = ops.cos(angles_r)
diff --git a/MindChemistry/applications/crystalflow/models/infer_utils.py b/MindChemistry/applications/crystalflow/models/infer_utils.py
index 7ff160955ca9c32e3c157a320f3e36d46a821cd0..2c72b3e071601948ffedb6bd83e11b01d5680cf4 100644
--- a/MindChemistry/applications/crystalflow/models/infer_utils.py
+++ b/MindChemistry/applications/crystalflow/models/infer_utils.py
@@ -50,7 +50,7 @@ def count_consecutive_occurrences(lst):
         lst (list): The input list
 
     Returns:
-        list: List of numbers of consecutive occurrences of each digit in the list.
+        list. List of numbers of consecutive occurrences of each digit in the list.
     """
     if not lst:
         return []
diff --git a/MindChemistry/applications/diffcsp/models/cspnet.py b/MindChemistry/applications/diffcsp/models/cspnet.py
index a12731b41aba347599b97487c7a5c76c2948eaf1..363c8d1c3f39b001f5be639b0a690c2bb3dd7ead 100644
--- a/MindChemistry/applications/diffcsp/models/cspnet.py
+++ b/MindChemistry/applications/diffcsp/models/cspnet.py
@@ -51,7 +51,7 @@ class SinusoidsEmbedding(nn.Cell):
             x (Tensor): Distance
 
         Returns:
-            Tensor: Fourier embedding
+            Tensor. Fourier embedding
         """
         emb = ops.ExpandDims()(x, -1) * Tensor(
             self.frequencies,
@@ -70,7 +70,7 @@ def mul_mask(features, mask):
             and value 0 specifies the corresponding dimension of input tensor to be zero.
 
     Returns:
-        Tensor: Output tensor
+        Tensor. Output tensor
     """
     return ops.mul(features, ops.reshape(mask, (-1, 1)))
 
@@ -175,7 +175,7 @@ class CSPLayer(nn.Cell):
             edge_mask (Tensor): Edge mask for padded tensor. Shape: (num_edges,)
 
         Returns:
-            Tensor: The output tensor. Shape: (num_atoms, hidden_dim)
+            Tensor. The output tensor. Shape: (num_atoms, hidden_dim)
         """
         node_input = node_features
         node_features = mul_mask(node_features, node_mask)
diff --git a/MindChemistry/applications/diffcsp/models/diff_utils.py b/MindChemistry/applications/diffcsp/models/diff_utils.py
index e74e20a0071700a0d85852faf0410a5d4a5fd71a..7eff1cdf040a083ad68ab06575518ee50108965d 100644
--- a/MindChemistry/applications/diffcsp/models/diff_utils.py
+++ b/MindChemistry/applications/diffcsp/models/diff_utils.py
@@ -112,7 +112,7 @@ def d_log_p_wrapped_normal_ms(x, sigma, n=10, t=1.0):
         t (int): The period of wrapped normal distribution.  Defaults to 1.0.
 
     Returns:
-        Tensor: The score for the input value x.
+        Tensor. The score for the input value x.
     """
     p_ = 0
     for i in range(-n, n + 1):
diff --git a/MindChemistry/applications/diffcsp/models/diffusion.py b/MindChemistry/applications/diffcsp/models/diffusion.py
index 21577030ce4a79cd1b2e37d02f5e0daa06a99492..7b29a861c80e96bf1112e3db632e641c6fc8a1a1 100644
--- a/MindChemistry/applications/diffcsp/models/diffusion.py
+++ b/MindChemistry/applications/diffcsp/models/diffusion.py
@@ -48,7 +48,7 @@ class SinusoidalTimeEmbeddings(nn.Cell):
             time (Tensor): diffution time step
 
         Returns:
-            Tensor: Time embedding
+            Tensor. Time embedding
         """
         half_dim = self.dim // 2
         embeddings = math.log(10000) / (half_dim - 1)
@@ -66,7 +66,7 @@ def lattice_params_to_matrix_mindspore(lengths, angles):
         lengths (Tensor): Tensor of shape (N, 3), unit A
         angles (Tensor):: Tensor of shape (N, 3), unit degree
     Returns:
-        Tensor: Tensor of shape (N, 3, 3)
+        Tensor. Tensor of shape (N, 3, 3)
     """
     angles_r = ops.deg2rad(angles)
     coses = ops.cos(angles_r)
diff --git a/MindChemistry/applications/diffcsp/models/infer_utils.py b/MindChemistry/applications/diffcsp/models/infer_utils.py
index 7ff160955ca9c32e3c157a320f3e36d46a821cd0..2c72b3e071601948ffedb6bd83e11b01d5680cf4 100644
--- a/MindChemistry/applications/diffcsp/models/infer_utils.py
+++ b/MindChemistry/applications/diffcsp/models/infer_utils.py
@@ -50,7 +50,7 @@ def count_consecutive_occurrences(lst):
         lst (list): The input list
 
     Returns:
-        list: List of numbers of consecutive occurrences of each digit in the list.
+        list. List of numbers of consecutive occurrences of each digit in the list.
     """
     if not lst:
         return []
diff --git a/MindChemistry/applications/orb/src/base.py b/MindChemistry/applications/orb/src/base.py
index 046e5950f209f46aa72b2933468211a74d9ae67d..861fd38d0f19ae35e74fa9516aaadc6a7ca32345 100644
--- a/MindChemistry/applications/orb/src/base.py
+++ b/MindChemistry/applications/orb/src/base.py
@@ -172,7 +172,7 @@ class AtomGraphs(NamedTuple):
         Any None attributes are not included in the dictionary.
 
         Returns:
-            dict: A dictionary mapping attribute_name -> tensor/scalar
+            dict. A dictionary mapping attribute_name -> tensor/scalar
         """
         ret = {}
         for key, val in self._asdict().items():
@@ -190,7 +190,7 @@ class AtomGraphs(NamedTuple):
         """Return a single dictionary mapping each AtomGraph property to a corresponding list of tensors/scalars.
 
         Returns:
-            dict: A dict mapping attribute_name -> list of length batch_size containing tensors/scalars.
+            dict. A dict mapping attribute_name -> list of length batch_size containing tensors/scalars.
         """
         batch_dict = defaultdict(list)
         for graph in self.split(self):
@@ -272,7 +272,7 @@ def batch_graphs(graphs: List[AtomGraphs]) -> AtomGraphs:
         graphs (List[AtomGraphs]): A list of AtomGraphs to be batched together.
 
     Returns:
-        AtomGraphs: A new AtomGraphs object with the concatenated nodes,
+        AtomGraphs. A new AtomGraphs object with the concatenated nodes,
         edges, and features from the input graphs, along with concatenated target,
         system ID, and other information.
     """
@@ -342,7 +342,7 @@ def refeaturize_atomgraphs(
             which considers boundary crossing for more than cell is used.
 
     Returns:
-        AtomGraphs: A refeaturized AtomGraphs object.
+        AtomGraphs. A refeaturized AtomGraphs object.
     """
     if cell is None:
         cell = atoms.cell
diff --git a/MindChemistry/applications/orb/src/featurization_utilities.py b/MindChemistry/applications/orb/src/featurization_utilities.py
index dd68d4ad8d4322bb334d9c70109ae7ce029b9050..a85f3d06e1ea03f72f33d77315b55c5c87462975 100644
--- a/MindChemistry/applications/orb/src/featurization_utilities.py
+++ b/MindChemistry/applications/orb/src/featurization_utilities.py
@@ -42,7 +42,7 @@ def gaussian_basis_function(
         scale (Tensor, optional): The width of the gaussians. Defaults to 1.
 
     Returns:
-        Tensor: A tensor of shape [num_scalars, num_bases].
+        Tensor. A tensor of shape [num_scalars, num_bases].
     """
     assert len(scalars.shape) == 1
     gaussian_means = ops.arange(
@@ -63,7 +63,7 @@ def featurize_edges(
         distance_featurization (DistanceFeaturization): A function that featurizes the distances of the vectors.
 
     Returns:
-        tensor: Edge features, shape [num_edge, num_edge_features].
+        tensor. Edge features, shape [num_edge, num_edge_features].
     """
     edge_features = []
     edge_norms = mint.linalg.norm(edge_vectors, dim=1)
@@ -86,7 +86,7 @@ def compute_edge_vectors(
         positions (tensor): Positions of each node. Shape [num_nodes, 3]
 
     Returns:
-        tensor: The vectors of each edge.
+        tensor. The vectors of each edge.
     """
     senders = edge_index[0]
     receivers = edge_index[1]
@@ -160,7 +160,7 @@ def _compute_img_positions_torch(
             for each atom, i.e this function is agnostic as to whether it is computing
             with respect to a batch or not.
     Returns:
-        Tensor: The positions of the periodic images. Shape [num_atoms, 27, 3].
+        Tensor. The positions of the periodic images. Shape [num_atoms, 27, 3].
     """
     num_positions = len(positions)
 
diff --git a/MindChemistry/mindchemistry/cell/allegro/allegro.py b/MindChemistry/mindchemistry/cell/allegro/allegro.py
index 741254e9de6e75d228851a3e5bf5842a3e0c5267..2e6fbe6460e434a9cd55b0a80adae5a290d464cc 100644
--- a/MindChemistry/mindchemistry/cell/allegro/allegro.py
+++ b/MindChemistry/mindchemistry/cell/allegro/allegro.py
@@ -34,7 +34,7 @@ def tp_path_exists(irreps_in1, irreps_in2, ir_out):
         ir_out: ir_out
 
     Returns:
-        bool: tp_path_exists or not
+        bool. tp_path_exists or not
     """
     irreps_in1 = Irreps(irreps_in1).simplify()
     irreps_in2 = Irreps(irreps_in2).simplify()
diff --git a/MindChemistry/mindchemistry/cell/cspnet.py b/MindChemistry/mindchemistry/cell/cspnet.py
index be8a8a3e91bd96f78663ad7cbe14f529d7b72a7e..41db5094857a0be5a584459ec1c965c9ba90ce52 100644
--- a/MindChemistry/mindchemistry/cell/cspnet.py
+++ b/MindChemistry/mindchemistry/cell/cspnet.py
@@ -57,7 +57,7 @@ class SinusoidsEmbedding(nn.Cell):
             x (Tensor): Distance
 
         Returns:
-            Tensor: Fourier embedding
+            Tensor. Fourier embedding
         """
         emb = self.expand_dims(x, -1) * self.frequencies_tensor
         emb = self.reshape(emb, (-1, self.n_frequencies * self.n_space))
@@ -74,7 +74,7 @@ def mul_mask(features, mask):
             and value 0 specifies the corresponding dimension of input tensor to be zero.
 
     Returns:
-        Tensor: Output tensor
+        Tensor. Output tensor
     """
     return ops.mul(features, ops.reshape(mask, (-1, 1)))
 
@@ -180,7 +180,7 @@ class CSPLayer(nn.Cell):
             edge_mask (Tensor): Edge mask for padded tensor. Shape: (num_edges,)
 
         Returns:
-            Tensor: The output tensor. Shape: (num_atoms, hidden_dim)
+            Tensor. The output tensor. Shape: (num_atoms, hidden_dim)
         """
         node_input = node_features
         node_features = mul_mask(node_features, node_mask)
diff --git a/MindChemistry/mindchemistry/cell/dimenet/preprocess.py b/MindChemistry/mindchemistry/cell/dimenet/preprocess.py
index 905ce82c1fa17826ff68b3675427b622532e21b2..48476e670ccb9292445c404f6198937625609bbd 100644
--- a/MindChemistry/mindchemistry/cell/dimenet/preprocess.py
+++ b/MindChemistry/mindchemistry/cell/dimenet/preprocess.py
@@ -146,7 +146,7 @@ class PreProcess:
             edge_index (np.ndarray): Array of edge indices.
 
         Returns:
-            Tuple: Tuple containing computed triplets.
+            Tuple. Tuple containing computed triplets.
         """
         node_1, node_2 = edge_index
         num_atoms = np.max(edge_index) + 1
diff --git a/MindChemistry/mindchemistry/cell/gemnet/gemnet_wrap.py b/MindChemistry/mindchemistry/cell/gemnet/gemnet_wrap.py
index b769d8496707bffe3b1140d4722e3bbc8e3d5fdc..57857f302545fbe064cf5bc5f9f414c1c3e2048a 100644
--- a/MindChemistry/mindchemistry/cell/gemnet/gemnet_wrap.py
+++ b/MindChemistry/mindchemistry/cell/gemnet/gemnet_wrap.py
@@ -107,7 +107,7 @@ class GemNetWrap:
             batch_size (int): The batch size.
 
         Returns:
-            tuple: A tuple containing the energy and force predictions.
+            tuple. A tuple containing the energy and force predictions.
                 - E_a (ms.Tensor): The predicted energies.
                 - F_j (ms.Tensor): The predicted forces.
         """
diff --git a/MindChemistry/mindchemistry/cell/orb/gns.py b/MindChemistry/mindchemistry/cell/orb/gns.py
index ab53083f8605dbe0463a5d9a0554d5e5990428f9..e489780b3f5dd87bdee937ceb18b5fe5dacb2964 100644
--- a/MindChemistry/mindchemistry/cell/orb/gns.py
+++ b/MindChemistry/mindchemistry/cell/orb/gns.py
@@ -59,7 +59,7 @@ def get_cutoff(p: int, r: Tensor, r_max: float) -> Tensor:
         r_max (float): Maximum distance for the cutoff.
 
     Returns:
-        Tensor: Cutoff tensor.
+        Tensor. Cutoff tensor.
     """
     envelope = 1.0 - ((p + 1.0) * (p + 2.0) / 2.0) * ops.pow(r / r_max, p) + \
         p * (p + 2.0) * ops.pow(r / r_max, p + 1) - \
@@ -84,7 +84,7 @@ def gaussian_basis_function(
         scale (Tensor, optional): The width of the gaussians. Defaults to 1.
 
     Returns:
-        Tensor: A tensor of shape [num_scalars, num_bases].
+        Tensor. A tensor of shape [num_scalars, num_bases].
     """
     assert len(scalars.shape) == 1
     gaussian_means = ops.arange(
diff --git a/MindChemistry/mindchemistry/cell/orb/orb.py b/MindChemistry/mindchemistry/cell/orb/orb.py
index 8afc55dbf5e931505ad6747b53ff9fcd4e1fbe23..404223e3633cfa2628b11bda85f60dcc0809b8f8 100644
--- a/MindChemistry/mindchemistry/cell/orb/orb.py
+++ b/MindChemistry/mindchemistry/cell/orb/orb.py
@@ -259,7 +259,7 @@ class NodeHead(ms.nn.Cell):
             n_node: Number of nodes in the graph.
 
         Returns:
-            node_pred: Node-level predictions of shape (n_nodes, target_property_dim).
+            node_pred. Node-level predictions of shape (n_nodes, target_property_dim).
         """
         out = self(node_features, n_node)
         pred = out["node_pred"]
@@ -378,7 +378,7 @@ class GraphHead(ms.nn.Cell):
             atomic_numbers: Optional atomic numbers for reference energy calculation
 
         Returns:
-            probs: Graph-level predictions of shape (n_graphs, target_property_dim).
+            probs. Graph-level predictions of shape (n_graphs, target_property_dim).
             If compute_stress is True, this will be the stress tensor.
             If compute_stress is False, this will be the graph-level property (e.g., energy).
         """
@@ -495,7 +495,7 @@ class EnergyHead(GraphHead):
             atomic_numbers: Optional atomic numbers for reference energy calculation
 
         Returns:
-            graph_pred: Energy prediction
+            graph_pred. Energy prediction
         """
         if atomic_numbers is None:
             raise ValueError("atomic_numbers is required for energy prediction")
@@ -671,7 +671,7 @@ class Orb(ms.nn.Cell):
             atomic_numbers: A tensor of atomic numbers for reference energy calculation.
 
         Returns:
-            ouput_dict: A dictionary containing the predictions:
+            ouput_dict. A dictionary containing the predictions:
             - `graph_pred`: Graph-level predictions (e.g., energy) of shape (n_graphs, graph_property_dim).
             - `stress_pred`: Stress predictions (if stress_head is provided) of shape (n_graphs, stress_dim).
             - `node_pred`: Node-level predictions of shape (n_nodes, node_property_dim).
diff --git a/MindChemistry/mindchemistry/cell/orb/utils.py b/MindChemistry/mindchemistry/cell/orb/utils.py
index 26797d9df20cc84e83ce9a20ebf63653fc2be33b..16eb538a42b348f3d2ad839e743917a6567eba0a 100644
--- a/MindChemistry/mindchemistry/cell/orb/utils.py
+++ b/MindChemistry/mindchemistry/cell/orb/utils.py
@@ -240,7 +240,7 @@ def build_mlp(
       checkpoint: Whether to use checkpointing.
 
     Returns:
-      mlp: An MLP sequential container.
+      mlp. An MLP sequential container.
     """
     # Size of each layer
     layer_sizes = [input_size] + hidden_layer_sizes
diff --git a/docs/api_python/mindchemistry/cell/mindchemistry.cell.Allegro.rst b/docs/api_python/mindchemistry/cell/mindchemistry.cell.Allegro.rst
index 1174823165ca65ce89e856719f89d47d5e3abcee..0ca03b7c550c5aacdfd72e229d1b7c900d59f49a 100644
--- a/docs/api_python/mindchemistry/cell/mindchemistry.cell.Allegro.rst
+++ b/docs/api_python/mindchemistry/cell/mindchemistry.cell.Allegro.rst
@@ -33,4 +33,16 @@ mindchemistry.cell.Allegro
         - **ValueError**: 如果 `new_tps_irreps` 的长度与 `tps_irreps` 不等。
         - **ValueError**: 如果 `tps_irreps` 的阶数不为零。
         - **ValueError**: 如果 `full_out_irreps` 的阶数不为零。
-        - **ValueError**: 如果 `out_irreps` 的阶数不为零。
\ No newline at end of file
+        - **ValueError**: 如果 `out_irreps` 的阶数不为零。
+
+    .. py:method:: construct(embedding_data, edge_index, atom_types)
+
+        构建Allegro网络的计算图。
+
+        参数：
+            - **embedding_data** (tuple(Tensor)) - 包含嵌入数据的元组，包括节点嵌入、边向量、边嵌入、边截断和边掩码等信息。
+            - **edge_index** (Tensor) - 形状为 :math:`(2, edge\_num)` 的张量，表示边的连接关系，其中edge_index[1]表示中心节点，edge_index[0]表示邻居节点。
+            - **atom_types** (Tensor) - 原子类型张量。
+
+        返回：
+            Tensor，形状为 :math:`(edge\_num, final\_latent\_out)` 的边特征张量。
\ No newline at end of file
diff --git a/docs/api_python/mindchemistry/cell/mindchemistry.cell.AutoEncoder.rst b/docs/api_python/mindchemistry/cell/mindchemistry.cell.AutoEncoder.rst
index 01496a0f2a7e39c86cfef58c932409186cd437d9..bdf8bd35a2a71146a6fbc60e52a19e8223bdb13b 100644
--- a/docs/api_python/mindchemistry/cell/mindchemistry.cell.AutoEncoder.rst
+++ b/docs/api_python/mindchemistry/cell/mindchemistry.cell.AutoEncoder.rst
@@ -25,3 +25,36 @@ mindchemistry.cell.AutoEncoder
         - **latents** (Tensor) - 形状为 :math:`(*, channels[-1])` 的张量。
         - **x_recon** (Tensor) - 形状为 :math:`(*, channels[0])` 的张量。
 
+    .. py:method:: construct(x)
+
+        构建自编码器网络的计算图。
+
+        参数：
+            - **x** (Tensor) - 输入数据，形状为 :math:`(*, channels[0])` 的张量。
+
+        返回：
+            tuple[Tensor]，包含两个张量：
+
+            - **x_recon** (Tensor) - 重构的输入数据，形状为 :math:`(*, channels[0])` 的张量。
+            - **latents** (Tensor) - 隐层编码，形状为 :math:`(*, channels[-1])` 的张量。
+
+    .. py:method:: encode(x)
+
+        将输入数据编码为隐层表示。
+
+        参数：
+            - **x** (Tensor) - 输入数据，形状为 :math:`(*, channels[0])` 的张量。
+
+        返回：
+            Tensor，隐层编码，形状为 :math:`(*, channels[-1])` 的张量。
+
+    .. py:method:: decode(z)
+
+        将隐层表示解码为重构数据。
+
+        参数：
+            - **z** (Tensor) - 隐层编码，形状为 :math:`(*, channels[-1])` 的张量。
+
+        返回：
+            Tensor，重构的输入数据，形状为 :math:`(*, channels[0])` 的张量。
+
diff --git a/docs/api_python/mindchemistry/cell/mindchemistry.cell.FCNet.rst b/docs/api_python/mindchemistry/cell/mindchemistry.cell.FCNet.rst
index 05bcc56ba735b1c8dd81e08eda251d03c0101f33..9f7823f683b77a58925643456ca82775d89f9103 100644
--- a/docs/api_python/mindchemistry/cell/mindchemistry.cell.FCNet.rst
+++ b/docs/api_python/mindchemistry/cell/mindchemistry.cell.FCNet.rst
@@ -23,3 +23,13 @@ mindchemistry.cell.FCNet
     输出：
         - **output** (Tensor) - 形状为 :math:`(*, channels[-1])` 的Tensor。
 
+    .. py:method:: construct(x)
+
+        构造函数，执行网络的前向传播。
+
+        参数：
+            - **x** (Tensor) - 输入张量，形状为 :math:`(*, channels[0])`。
+
+        返回：
+            Tensor，输出张量，形状为 :math:`(*, channels[-1])`。
+
diff --git a/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.EnergyHead.rst b/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.EnergyHead.rst
index fb549db328a7b26008669b6bbe43d7dc1a925bd9..001f02fcbcb0c7c3e5c3cc16df37e151bfd7414c 100644
--- a/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.EnergyHead.rst
+++ b/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.EnergyHead.rst
@@ -25,4 +25,19 @@ mindchemistry.cell.orb.EnergyHead
 
     异常：
         - **ValueError** - 如果 `node_features` 中缺少必需的特征键。
-        - **ValueError** - 如果 `node_aggregation` 不是支持的类型。
\ No newline at end of file
+        - **ValueError** - 如果 `node_aggregation` 不是支持的类型。
+
+    .. py:method:: predict(node_features, n_node, atom_types=None)
+
+        预测能量。
+
+        参数：
+            - **node_features** (dict) - 节点特征字典，必须包含键"feat"，形状为 :math:`(n_{nodes}, latent\_dim)`。
+            - **n_node** (Tensor) - 图中节点数量，形状为 :math:`(1,)`。
+            - **atom_types** (Tensor, 可选) - 原子类型，形状为 :math:`(n_{nodes},)`。默认值： ``None``。
+
+        返回：
+            dict，包含键"graph_pred"的字典，值的形状为 :math:`(1, target\_property\_dim)`。
+
+        异常：
+            - **ValueError** - 如果 `node_features` 中缺少必需的特征键。
\ No newline at end of file
diff --git a/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.GraphHead.rst b/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.GraphHead.rst
index 75ae5ad7c52da0a39f5b91142a8458c68fd51323..7c439e46ee95ebbe2c054fcaf2d4d50d66ca7cf4 100644
--- a/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.GraphHead.rst
+++ b/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.GraphHead.rst
@@ -22,4 +22,19 @@ mindchemistry.cell.orb.GraphHead
         - **output** (dict) - 包含键"stress_pred"的字典，值的形状为 :math:`(1, target\_property\_dim)`。
 
     异常：
-        - **ValueError** - 如果 `node_features` 中缺少必需的特征键。
\ No newline at end of file
+        - **ValueError** - 如果 `node_features` 中缺少必需的特征键。
+
+    .. py:method:: predict(node_features, n_node, atomic_numbers=None)
+
+        预测图级属性。
+
+        参数：
+            - **node_features** (dict) - 节点特征字典，必须包含键"feat"，形状为 :math:`(n_{nodes}, latent\_dim)`。
+            - **n_node** (Tensor) - 图中节点数量，形状为 :math:`(1,)`。
+            - **atomic_numbers** (Tensor, 可选) - 原子序数，用于参考能量计算。当compute_stress为False时必须提供。默认值： ``None``。
+
+        返回：
+            Tensor，图级预测结果，形状为 :math:`(n_{graphs}, target\_property\_dim)`。如果compute_stress为True，则为应力张量；如果compute_stress为False，则为图级属性（例如能量）。
+
+        异常：
+            - **AssertionError** - 如果compute_stress为False且未提供atomic_numbers。
\ No newline at end of file
diff --git a/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.MoleculeGNS.rst b/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.MoleculeGNS.rst
index c44551f329f4d6bd28dcacf08210fc60cc7a662d..7e702bd57b6ed804fe278f0bbff5cc069b42eb93 100644
--- a/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.MoleculeGNS.rst
+++ b/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.MoleculeGNS.rst
@@ -31,4 +31,24 @@ mindchemistry.cell.orb.MoleculeGNS
 
     异常：
         - **ValueError** - 如果 `edge_features` 或 `node_features` 中缺少必需的特征键。
-        - **ValueError** - 如果 `interactions` 不是支持的类型。
\ No newline at end of file
+        - **ValueError** - 如果 `interactions` 不是支持的类型。
+
+    .. py:method:: featurize_nodes(node_features)
+
+        特征化图的节点。
+
+        参数：
+            - **node_features** (Dict[str, Tensor]) - 节点特征字典。
+
+        返回：
+            Dict[str, Tensor]，带有原子嵌入的更新节点特征。
+
+    .. py:method:: featurize_edges(edge_features)
+
+        特征化图的边。
+
+        参数：
+            - **edge_features** (Dict[str, Tensor]) - 边特征字典。
+
+        返回：
+            Dict[str, Tensor]，带有径向基函数和单位向量的更新边特征。
\ No newline at end of file
diff --git a/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.NodeHead.rst b/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.NodeHead.rst
index 2e422d861892a40688faea4b8b38ca0a5848bb63..1d198d9c790edc594ac12023335a0a29ca936bd8 100644
--- a/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.NodeHead.rst
+++ b/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.NodeHead.rst
@@ -23,4 +23,15 @@ mindchemistry.cell.orb.NodeHead
         - **output** (dict) - 包含键 "node_pred" 的字典，值的形状为 :math:`(n_{nodes}, target\_property\_dim)`。
 
     异常：
-        - **ValueError** - 如果 `node_features` 中缺少必需的特征键。
\ No newline at end of file
+        - **ValueError** - 如果 `node_features` 中缺少必需的特征键。
+
+    .. py:method:: predict(node_features, n_node)
+
+        预测节点级属性。
+
+        参数：
+            - **node_features** (dict) - 节点特征字典，必须包含键 "feat"，形状为 :math:`(n_{nodes}, latent\_dim)`。
+            - **n_node** (Tensor) - 图中节点数量，形状为 :math:`(1,)`。
+
+        返回：
+            Tensor，节点级预测结果，形状为 :math:`(n_{nodes}, target\_property\_dim)`。
\ No newline at end of file
diff --git a/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.Orb.rst b/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.Orb.rst
index fe1d53a26ce54523eb47205d9a5fa5729892f8c2..253508696d1bea5d780f7ce53a467779bed769c5 100644
--- a/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.Orb.rst
+++ b/docs/api_python/mindchemistry/cell/mindchemistry.cell.orb.Orb.rst
@@ -32,4 +32,23 @@ mindchemistry.cell.orb.Orb
     异常：
         - **ValueError** - 如果既未提供node_head也未提供graph_head。
         - **ValueError** - 如果cutoff_layers超过基础模型中的消息传递步骤数。
-        - **ValueError** - 如果graph_head需要时未提供atomic_numbers。
\ No newline at end of file
+        - **ValueError** - 如果graph_head需要时未提供atomic_numbers。
+
+    .. py:method:: predict(edge_features, node_features, senders, receivers, n_node, atomic_numbers)
+
+        预测节点和/或图级属性。
+
+        参数：
+            - **edge_features** (dict) - 边特征字典（例如，`{"vectors": Tensor, "r": Tensor}`）。
+            - **node_features** (dict) - 节点特征字典（例如，`{"atomic_numbers": Tensor, "positions": Tensor, "atomic_numbers_embedding": Tensor}`）。
+            - **senders** (Tensor) - 每条边的发送节点索引。形状：:math:`(n_{edges},)`。
+            - **receivers** (Tensor) - 每条边的接收节点索引。形状：:math:`(n_{edges},)`。
+            - **n_node** (Tensor) - 批次中每个图的节点数。形状：:math:`(n_{graphs},)`。
+            - **atomic_numbers** (Tensor) - 用于参考能量计算的原子序数张量。
+
+        返回：
+            dict，包含以下预测结果：
+
+            - **graph_pred** (Tensor) - 图级预测（例如，能量）。形状：:math:`(n_{graphs}, graph\_property\_dim)`。
+            - **stress_pred** (Tensor) - 应力预测（如果提供了stress_head）。形状：:math:`(n_{graphs}, stress\_dim)`。
+            - **node_pred** (Tensor) - 节点级预测。形状：:math:`(n_{nodes}, node\_property\_dim)`。
\ No newline at end of file