Fix typos (PaddlePaddle#1150)

Author: co63oc

docs/zh/examples/drivaernetplusplus.md

@@ -345,7 +345,7 @@ export ep2=-----------------ID-----------------
 
 17. 下载数据，将 PubDAS 中的 License.txt 从 ep1 数据源传送到 ep2（自己路径下，第 8，9 步设置）。
 ``` sh
-# here is defaut path (your home path)
+# here is default path (your home path)
 globus transfer $ep1:License.txt $ep2:/~/License.txt
 Message: The transfer has been accepted and a task has been created and queued for execution
 Task ID: -----------------传送任务ID-----------------

docs/zh/examples/phygeonet.md

@@ -214,7 +214,7 @@ T=3
 ![image](https://paddle-org.bj.bcebos.com/paddlescience/docs/PhyGeoNet/heat_equation_bc_2.png)
 
 T=6
-![iamge](https://paddle-org.bj.bcebos.com/paddlescience/docs/PhyGeoNet/heat_equation_bc_3.png)
+![image](https://paddle-org.bj.bcebos.com/paddlescience/docs/PhyGeoNet/heat_equation_bc_3.png)
 
 ## 6. 总结
 

docs/zh/examples/tempoGAN.md

@@ -146,7 +146,7 @@ examples/tempoGAN/conf/tempogan.yaml:27:37
 --8<--
 ```
 
-注意到其中包含 3 个 bool 类型的变量 `use_amp`、`use_spatialdisc` 和 `use_tempodisc`，它们分别表示是否使用混合精度训练(AMP)、是否使用 Discriminator 和是否使用 Discriminator_tempo，当 `use_spatialdisc` 和 `use_tempodisc` 都被设置为 `False` 时，本问题的网络结构将会变为一个单纯的 Genrator 模型，不再是 GAN 网络了。
+注意到其中包含 3 个 bool 类型的变量 `use_amp`、`use_spatialdisc` 和 `use_tempodisc`，它们分别表示是否使用混合精度训练(AMP)、是否使用 Discriminator 和是否使用 Discriminator_tempo，当 `use_spatialdisc` 和 `use_tempodisc` 都被设置为 `False` 时，本问题的网络结构将会变为一个单纯的 Generator 模型，不再是 GAN 网络了。
 
 同时需要指定训练轮数和学习率等超参数，注意由于 GAN 网络训练流程与一般单个模型的网络不同，`EPOCHS` 的设置也有所不同。
 

examples/adr/README.md

@@ -1 +1 @@
-souce code will be available after code already orgnized in the near future.
+source code will be available after code already orgnized in the near future.

examples/moflow/optimize_moflow.py

@@ -72,7 +72,7 @@ def load_property_csv(filepath, normalize=True):
 
 
 def smiles_to_adj(mol_smiles, data_name="qm9"):
-    """Use simles to adj, atoms
+    """Use smiles to adj, atoms
 
     Args:
         mol_smiles: eg. CC(C)(C)c1ccc2occ(CC(=O)Nc3ccccc3F)c2c1

jointContribution/CFDGCN/mesh_utils.py

@@ -268,7 +268,7 @@ def delauney(x):
         face = np.array([[0, 1, 2]])
     else:
         raise ValueError(
-            f"Not enough points to contruct Delaunay triangulation, got {pos.shape[0]} "
+            f"Not enough points to construct Delaunay triangulation, got {pos.shape[0]} "
             + "but expected at least 3"
         )
 

jointContribution/PIRBN/README.md

@@ -13,13 +13,13 @@ Inspired by findings, we proposed the PIRBN, which can exhibit the local propert
 
 Numerical examples include:
 
- - 1D sine funtion (**Eq. 13** in the manuscript)
+ - 1D sine function (**Eq. 13** in the manuscript)
 
       **PDE**: $\frac{\partial^2 }{\partial x^2}u(x)-4\mu^2\pi^2 sin(2\mu\pi(x))=0, x\in[0,1]$
 
       **BC**:  $u(0)=u(1)=0.$
 
- - 1D sine funtion (**Eq. 15** in the manuscript)
+ - 1D sine function (**Eq. 15** in the manuscript)
       **PDE**: $\frac{\partial^2 }{\partial x^2}u(x-100)-4\mu^2\pi^2 sin(2\mu\pi(x-100))=0, x\in[100,101]$
 
       **BC**:  $u(100)=u(101)=0.$
@@ -32,7 +32,7 @@ For more details in terms of mathematical proofs and numerical examples, please
 
 <https://github.com/JinshuaiBai/PIRBN>
 
-## Enviornmental settings
+## Environmental settings
 
 ``` shell
 pip install -r requirements.txt

jointContribution/PIRBN/jacobian_function.py

@@ -24,9 +24,9 @@ def jacobian(y, x):
     J = paddle.zeros(J_shape)
     J_flat = flat(
         J, start_axis=0, stop_axis=len(y.shape) - 1
-    )  # partialy flatten as y_flat
+    )  # partially flatten as y_flat
     for i, y_i in enumerate(y):
-        assert y_i.size == 1, f"y_i.size shoule be 1, but got {y_i.size}"
+        assert y_i.size == 1, f"y_i.size should be 1, but got {y_i.size}"
         grad = paddle.grad(y_i, x, allow_unused=True)[
             0
         ]  # grad[i] == sum by j (dy[j] / dx[i])

jointContribution/PIRBN/rbn_net.py

@@ -133,7 +133,7 @@ class Activation(paddle.nn.Layer):
     def __init__(self, n_neu, c, n_in, activation_function="gaussian"):
         super(Activation, self).__init__()
         self.activation_function = activation_function
-        # PINN      y = w2 * (tanh(w1 * x + b1)) + b2         w,b are trainable parameters, b is bais
+        # PINN      y = w2 * (tanh(w1 * x + b1)) + b2         w,b are trainable parameters, b is bias
         # PIRBN     y = w * exp(b^2 * |x-c|^2)          w,b are trainable parameters, c is constant, b is not bias
 
         if self.activation_function == "gaussian":

jointContribution/graphcast/datasets.py

@@ -50,7 +50,7 @@ def get_day_progress(
         point at which UNIX time starts).
       longitude: 1D array of longitudes at which day progress is computed.
     Returns:
-      2D array of day progress values normalized to be in the [0, 1) inverval
+      2D array of day progress values normalized to be in the [0, 1) interval
         for each time point at each longitude.
     """
     # [0.0, 1.0) Interval.

ppsci/arch/afno.py

@@ -165,7 +165,7 @@ def __init__(
         super().__init__()
         if hidden_size % num_blocks != 0:
             raise ValueError(
-                f"hidden_size({hidden_size}) should be divisble by num_blocks({num_blocks})."
+                f"hidden_size({hidden_size}) should be divisible by num_blocks({num_blocks})."
             )
 
         self.hidden_size = hidden_size

ppsci/arch/cfdgcn.py

@@ -194,7 +194,7 @@ class CFDGCN(nn.Layer):
         config_file (str): Name of configuration file for su2 module.
         coarse_mesh (str): Path of coarse mesh file.
         fine_marker_dict (Dict[str, List[List[int]]]): Dict of fine marker.
-        process_sim (Callable, optional): Preprocess funtion. Defaults to `lambda x, y: x`.
+        process_sim (Callable, optional): Preprocess function. Defaults to `lambda x, y: x`.
         freeze_mesh (bool, optional): Whether set `stop_gradient=True` for nodes. Defaults to False.
         num_convs (int, optional): Number of conv layers. Defaults to 6.
         num_end_convs (int, optional): Number of end conv layers. Defaults to 3.

ppsci/arch/cuboid_transformer.py

@@ -491,7 +491,7 @@ class CuboidTransformer(base.Arch):
         gated_ffn (bool, optional): Whether to use gate FFN. Defaults to False.
         norm_layer (str, optional): The type of normilize. Defaults to "layer_norm".
         padding_type (str, optional): The type of padding. Defaults to "ignore".
-        pos_embed_type (str, optional): The type of pos embeding. Defaults to "t+hw".
+        pos_embed_type (str, optional): The type of pos embedding. Defaults to "t+hw".
         checkpoint_level (bool, optional): Whether to use checkpoint. Defaults to True.
         use_relative_pos (bool, optional): Whether to use relative pose. Defaults to True.
         self_attn_use_final_proj (bool, optional): Whether to use final projection. Defaults to True.

ppsci/arch/cuboid_transformer_decoder.py

@@ -13,13 +13,13 @@
 
 
 class PosEmbed(nn.Layer):
-    """Pose embeding
+    """Pose embedding
 
     Args:
-        embed_dim (int): The dimension of embeding.
-        maxT (int): The embeding max time.
-        maxH (int): The embeding max height.
-        maxW (int): The embeding max width.
+        embed_dim (int): The dimension of embedding.
+        maxT (int): The embedding max time.
+        maxH (int): The embedding max height.
+        maxW (int): The embedding max width.
         typ (str):
             The type of the positional embedding.
             - t+h+w:
@@ -180,7 +180,7 @@ class CuboidCrossAttentionLayer(nn.Layer):
     The complexity of the layer is O((T2 / n_t * Bh * Bw) * (T1 / n_t * Bh * Bw) * n_t (H / Bh) (W / Bw)) = O(T2 * T1 / n_t H W Bh Bw)
 
     Args:
-        dim (int): The dimention of input tensor.
+        dim (int): The dimension of input tensor.
         num_heads (int): The number of head.
         n_temporal (int, optional): The num of temporal. Defaults to 1.
         cuboid_hw (tuple, optional): The height and width of cuboid. Defaults to (7, 7).
@@ -880,7 +880,7 @@ class CuboidTransformerDecoder(nn.Layer):
         norm_layer (str, optional): The normalization layer. Defaults to "layer_norm".
         use_inter_ffn (bool, optional): Whether to use inter FFN. Defaults to False.
         hierarchical_pos_embed (bool, optional): Whether to use hierarchical pos_embed. Defaults to False.
-        pos_embed_type (str, optional): The type of pos embeding. Defaults to "t+hw".
+        pos_embed_type (str, optional): The type of pos embedding. Defaults to "t+hw".
         max_temporal_relative (int, optional): The max number of teemporal relative. Defaults to 50.
         padding_type (str, optional): The type of padding. Defaults to "ignore".
         checkpoint_level (bool, optional):  Whether to enable gradient checkpointing. Defaults to True.

ppsci/arch/extformer_moe_cuboid.py

@@ -496,7 +496,7 @@ class ExtFormerMoECuboid(base.Arch):
         gated_ffn (bool, optional): Whether to use gate FFN. Defaults to False.
         norm_layer (str, optional): The type of normilize. Defaults to "layer_norm".
         padding_type (str, optional): The type of padding. Defaults to "ignore".
-        pos_embed_type (str, optional): The type of pos embeding. Defaults to "t+hw".
+        pos_embed_type (str, optional): The type of pos embedding. Defaults to "t+hw".
         checkpoint_level (bool, optional): Whether to use checkpoint. Defaults to True.
         use_relative_pos (bool, optional): Whether to use relative pose. Defaults to True.
         self_attn_use_final_proj (bool, optional): Whether to use final projection. Defaults to True.

ppsci/arch/extformer_moe_cuboid_decoder.py

@@ -14,13 +14,13 @@
 
 
 class PosEmbed(nn.Layer):
-    """pose embeding
+    """pose embedding
 
     Args:
-        embed_dim (int): The dimension of embeding.
-        maxT (int): The embeding max time.
-        maxH (int): The embeding max height.
-        maxW (int): The embeding max width.
+        embed_dim (int): The dimension of embedding.
+        maxT (int): The embedding max time.
+        maxH (int): The embedding max height.
+        maxW (int): The embedding max width.
         typ (str):
             The type of the positional embedding.
             - t+h+w:
@@ -189,7 +189,7 @@ class CuboidCrossAttentionLayer(nn.Layer):
     The complexity of the layer is O((T2 / n_t * Bh * Bw) * (T1 / n_t * Bh * Bw) * n_t (H / Bh) (W / Bw)) = O(T2 * T1 / n_t H W Bh Bw)
 
     Args:
-        dim (int): The dimention of input tensor.
+        dim (int): The dimension of input tensor.
         num_heads (int): The number of head.
         n_temporal (int, optional): The num of temporal. Defaults to 1.
         cuboid_hw (tuple, optional): The height and width of cuboid. Defaults to (7, 7).
@@ -980,7 +980,7 @@ class CuboidTransformerDecoder(nn.Layer):
         norm_layer (str, optional): The normalization layer. Defaults to "layer_norm".
         use_inter_ffn (bool, optional): Whether to use inter FFN. Defaults to False.
         hierarchical_pos_embed (bool, optional): Whether to use hierarchical pos_embed. Defaults to False.
-        pos_embed_type (str, optional): The type of pos embeding. Defaults to "t+hw".
+        pos_embed_type (str, optional): The type of pos embedding. Defaults to "t+hw".
         max_temporal_relative (int, optional): The max number of teemporal relative. Defaults to 50.
         padding_type (str, optional): The type of padding. Defaults to "ignore".
         checkpoint_level (bool, optional):  Whether to enable gradient checkpointing. Defaults to True.

ppsci/arch/fno_block.py

@@ -87,7 +87,7 @@ def pad(self, x):
             # (so we must reverse the padding list)
             padding = padding[::-1]
 
-            # the F.pad(x, padding) funtion pads the tensor 'x' in reverse order of the "padding" list i.e. the last axis of tensor 'x' will be padded by the amount mention at the first position of the 'padding' vector. The details about F.pad can be found here:
+            # the F.pad(x, padding) function pads the tensor 'x' in reverse order of the "padding" list i.e. the last axis of tensor 'x' will be padded by the amount mention at the first position of the 'padding' vector. The details about F.pad can be found here:
             # https://www.paddlepaddle.org.cn/documentation/docs/zh/api/paddle/nn/functional/pad_cn.html
 
             if self.padding_mode == "symmetric":
@@ -249,7 +249,7 @@ def set_embedding(self, x):
     def forward(self, x):
         assert (
             self.embedding is not None
-        ), "AdaIN: update embeddding before running forward"
+        ), "AdaIN: update embedding before running forward"
 
         weight, bias = paddle.split(
             self.mlp(self.embedding),
@@ -484,7 +484,7 @@ def resample(x, res_scale, axis, output_shape=None):
         else:
             res_scale = [res_scale] * len(axis)
     else:
-        assert len(res_scale) == len(axis), "leght of res_scale and axis are not same"
+        assert len(res_scale) == len(axis), "length of res_scale and axis are not same"
 
     old_size = x.shape[-len(axis) :]
     if output_shape is None:

ppsci/arch/geofno.py

@@ -67,7 +67,7 @@ def compl_mul1d(self, op1, op2_real, op2_imag):
 
     def forward(self, x, output_size=None):
         batchsize = x.shape[0]
-        # Compute Fourier coeffcients up to factor of e^(- something constant)
+        # Compute Fourier coefficients up to factor of e^(- something constant)
         x_ft = paddle.fft.rfft(x)
 
         # Multiply relevant Fourier modes

ppsci/arch/ifm_mlp.py

@@ -44,8 +44,8 @@ class IFMMLP(base.Arch):
         embed_name (str): Embed name used in arch, such as "IMF", "None".
         inputs (int): Input dim.
         outputs (int): Output dim.
-        d_out (int): Embedding output dim for some architechture.
-        sigma (float): Hyper parameter for some architechture.
+        d_out (int): Embedding output dim for some architecture.
+        sigma (float): Hyper parameter for some architecture.
         dp_ratio (float): Dropout ratio.
         reg (bool): Regularization flag.
         first_omega_0 (float): Frequency factor used in first layer.
@@ -319,8 +319,8 @@ class GM_DNN(nn.Layer):
         inputs (int): Input dim.
         hidden_units (List[int]): Units num in hidden layers.
         outputs (int): Output dim.
-        d_out (int): Embedding output dim for some architechture.
-        sigma (float): Hyper parameter for some architechture.
+        d_out (int): Embedding output dim for some architecture.
+        sigma (float): Hyper parameter for some architecture.
         dp_ratio (float): Dropout ratio.
         reg (bool): Regularization flag.
     """
@@ -498,8 +498,8 @@ class SIM_DNN(nn.Layer):
         inputs (int): Input dim.
         hidden_units (List[int]): Units num in hidden layers.
         outputs (int): Output dim.
-        d_out (int): Embedding output dim for some architechture.
-        sigma (float): Hyper parameter for some architechture.
+        d_out (int): Embedding output dim for some architecture.
+        sigma (float): Hyper parameter for some architecture.
         dp_ratio (float): Dropout ratio.
         reg (bool): Regularization flag.
     """

ppsci/arch/mlp.py

@@ -151,7 +151,7 @@ class MLP(base.Arch):
         input_dim (Optional[int]): Number of input's dimension. Defaults to None.
         output_dim (Optional[int]): Number of output's dimension. Defaults to None.
         periods (Optional[Dict[int, Tuple[float, bool]]]): Period of each input key,
-            input in given channel will be period embeded if specified, each tuple of
+            input in given channel will be period embedded if specified, each tuple of
             periods list is [period, trainable]. Defaults to None.
         fourier (Optional[Dict[str, Union[float, int]]]): Random fourier feature embedding,
             e.g. {'dim': 256, 'scale': 1.0}. Defaults to None.
@@ -220,7 +220,7 @@ def __init__(
         # initialize FC layer(s)
         cur_size = len(self.input_keys) if input_dim is None else input_dim
         if input_dim is None and periods:
-            # period embeded channel(s) will be doubled automatically
+            # period embedded channel(s) will be doubled automatically
             # if input_dim is not specified
             cur_size += len(periods)
 
@@ -384,7 +384,7 @@ def __init__(
         # initialize FC layer(s)
         cur_size = len(self.input_keys) if input_dim is None else input_dim
         if input_dim is None and periods:
-            # period embeded channel(s) will be doubled automatically
+            # period embedded channel(s) will be doubled automatically
             # if input_dim is not specified
             cur_size += len(periods)
 
@@ -653,7 +653,7 @@ class PirateNet(base.Arch):
         input_dim (Optional[int]): Number of input's dimension. Defaults to None.
         output_dim (Optional[int]): Number of output's dimension. Defaults to None.
         periods (Optional[Dict[int, Tuple[float, bool]]]): Period of each input key,
-            input in given channel will be period embeded if specified, each tuple of
+            input in given channel will be period embedded if specified, each tuple of
             periods list is [period, trainable]. Defaults to None.
         fourier (Optional[Dict[str, Union[float, int]]]): Random fourier feature embedding,
             e.g. {'dim': 256, 'scale': 1.0}. Defaults to None.
@@ -712,7 +712,7 @@ def __init__(
         # initialize FC layer(s)
         cur_size = len(self.input_keys) if input_dim is None else input_dim
         if input_dim is None and periods:
-            # period embeded channel(s) will be doubled automatically
+            # period embedded channel(s) will be doubled automatically
             # if input_dim is not specified
             cur_size += len(periods)
 

ppsci/arch/sfnonet.py

@@ -408,7 +408,7 @@ class SFNONet(base.Arch):
             Defaults to None.
         non_linearity (nn.functional, optional): Non-Linearity module to use. Defaults to F.gelu.
         norm (str, optional): Normalization layer to use. Defaults to None.
-        ada_in_features (int,optional): The input channles of the adaptive normalization.Defaults to None.
+        ada_in_features (int,optional): The input channels of the adaptive normalization.Defaults to None.
         preactivation (bool, optional): Whether to use resnet-style preactivation. Defaults to False.
         fno_skip (str, optional): Type of skip connection to use,{'linear', 'identity', 'soft-gating'}.
             Defaults to "soft-gating".

ppsci/arch/tfnonet.py

@@ -31,7 +31,7 @@ class FNONet(base.Arch):
             Defaults to None.
         non_linearity (nn.functional, optional): Non-Linearity module to use. Defaults to F.gelu.
         norm (str, optional): Normalization layer to use. Defaults to None.
-        ada_in_features (int,optional): The input channles of the adaptive normalization.Defaults to None.s
+        ada_in_features (int,optional): The input channels of the adaptive normalization.Defaults to None.s
         preactivation (bool, optional): Whether to use resnet-style preactivation. Defaults to False.
         skip (str, optional): Type of skip connection to use,{'linear', 'identity', 'soft-gating'}.
             Defaults to "soft-gating".

ppsci/arch/unonet.py

@@ -45,7 +45,7 @@ class UNONet(base.Arch):
             Defaults to None.
         non_linearity (nn.functional, optional): Non-Linearity module to use. Defaults to F.gelu.
         norm (str, optional): Normalization layer to use. Defaults to None.
-        ada_in_features (Optional[int],optional): The input channles of the adaptive normalization.Defaults to
+        ada_in_features (Optional[int],optional): The input channels of the adaptive normalization.Defaults to
             None.
         preactivation (bool, optional): Whether to use resnet-style preactivation. Defaults to False.
         fno_skip (str, optional): Type of skip connection to use for fno_block. Defaults to "linear".
@@ -162,7 +162,7 @@ def __init__(
                 # example, if n_layers = 5, then 4:0, 3:1
                 self.horizontal_skips_map[n_layers - i - 1] = i
         # self.uno_scalings may be a 1d list specifying uniform scaling factor at each layer
-        # or a 2d list, where each row specifies scaling factors along each dimention.
+        # or a 2d list, where each row specifies scaling factors along each dimension.
         # To get the final (end to end) scaling factors we need to multiply
         # the scaling factors (a list) of all layer.