remove nets.py in fluid (#51717)

longranger2 · web-flow · commit 87f72107dcd1 · 2023-06-28T16:28:44.000+08:00
* remove function sequence_conv_pool

* fix test_glu.py

* remove function scaled_dot_product_attention

* remove function img_conv_group

* remove function simple_img_conv_pool

* delete the nets.py

* fix bug

* fix bug

* fix bug

* fix bug

* fix bug

* fix bug

* Update test_image_classification.py

* Update test_recognize_digits.py

* Update test_recommender_system.py

* Update test_image_classification_fp16.py

* fix bug

* fix bug

* fix bug

* remove the nets.py in fluid

* fix bug

* fix the codestyle

* fix conflict

* fix bug

* Fix TODO
diff --git a/python/paddle/fluid/__init__.py b/python/paddle/fluid/__init__.py
@@ -54,7 +54,6 @@
 from . import layers
 from . import dygraph
 from . import contrib
-from . import nets
 from . import optimizer
 from . import backward
 from .backward import gradients
@@ -112,7 +111,6 @@
         'disable_dygraph',
         'enable_imperative',
         'disable_imperative',
-        'nets',
         'optimizer',
         'backward',
         'LoDTensor',
diff --git a/test/book/notest_understand_sentiment.py b/test/book/notest_understand_sentiment.py
@@ -20,6 +20,10 @@
 
 import numpy as np
 
+# TODO: remove sys.path.append
+sys.path.append("../legacy_test")
+import nets
+
 import paddle
 from paddle import fluid
 
@@ -30,14 +34,14 @@ def convolution_net(
     emb = fluid.layers.embedding(
         input=data, size=[input_dim, emb_dim], is_sparse=True
     )
-    conv_3 = fluid.nets.sequence_conv_pool(
+    conv_3 = nets.sequence_conv_pool(
         input=emb,
         num_filters=hid_dim,
         filter_size=3,
         act="tanh",
         pool_type="sqrt",
     )
-    conv_4 = fluid.nets.sequence_conv_pool(
+    conv_4 = nets.sequence_conv_pool(
         input=emb,
         num_filters=hid_dim,
         filter_size=4,
diff --git a/test/book/test_image_classification.py b/test/book/test_image_classification.py
@@ -21,6 +21,10 @@
 
 import numpy
 
+# TODO: remove sys.path.append
+sys.path.append("../legacy_test")
+import nets
+
 import paddle
 from paddle import fluid
 
@@ -74,7 +78,7 @@ def layer_warp(block_func, input, ch_in, ch_out, count, stride):
 
 def vgg16_bn_drop(input):
     def conv_block(input, num_filter, groups, dropouts):
-        return fluid.nets.img_conv_group(
+        return nets.img_conv_group(
             input=input,
             pool_size=2,
             pool_stride=2,
diff --git a/test/book/test_recognize_digits.py b/test/book/test_recognize_digits.py
@@ -19,6 +19,10 @@
 
 import numpy
 
+# TODO: remove sys.path.append
+sys.path.append("../legacy_test")
+import nets
+
 import paddle
 from paddle import fluid
 from paddle.fluid import core
@@ -45,7 +49,7 @@ def mlp(img, label):
 
 
 def conv_net(img, label):
-    conv_pool_1 = fluid.nets.simple_img_conv_pool(
+    conv_pool_1 = nets.simple_img_conv_pool(
         input=img,
         filter_size=5,
         num_filters=20,
@@ -54,7 +58,7 @@ def conv_net(img, label):
         act="relu",
     )
     conv_pool_1 = paddle.static.nn.batch_norm(conv_pool_1)
-    conv_pool_2 = fluid.nets.simple_img_conv_pool(
+    conv_pool_2 = nets.simple_img_conv_pool(
         input=conv_pool_1,
         filter_size=5,
         num_filters=50,
diff --git a/test/book/test_recommender_system.py b/test/book/test_recommender_system.py
@@ -19,9 +19,13 @@
 
 import numpy as np
 
+# TODO: remove sys.path.append
+sys.path.append("../legacy_test")
+import nets
+
 import paddle
 from paddle import fluid
-from paddle.fluid import framework, layers, nets
+from paddle.fluid import framework, layers
 from paddle.fluid.executor import Executor
 from paddle.fluid.optimizer import SGDOptimizer
 
diff --git a/test/collective/fleet/pipeline_mnist.py b/test/collective/fleet/pipeline_mnist.py
@@ -14,6 +14,7 @@
 
 from functools import reduce
 
+from legacy_test import nets
 from legacy_test.test_dist_base import TestDistRunnerBase, runtime_main
 
 import paddle
@@ -31,7 +32,7 @@
 
 
 def cnn_model(data):
-    conv_pool_1 = fluid.nets.simple_img_conv_pool(
+    conv_pool_1 = nets.simple_img_conv_pool(
         input=data,
         filter_size=5,
         num_filters=20,
@@ -42,7 +43,7 @@ def cnn_model(data):
             initializer=paddle.nn.initializer.Constant(value=0.01)
         ),
     )
-    conv_pool_2 = fluid.nets.simple_img_conv_pool(
+    conv_pool_2 = nets.simple_img_conv_pool(
         input=conv_pool_1,
         filter_size=5,
         num_filters=50,
diff --git a/test/collective/fleet/pipeline_mnist_multi_device.py b/test/collective/fleet/pipeline_mnist_multi_device.py
@@ -14,6 +14,7 @@
 
 from functools import reduce
 
+from legacy_test import nets
 from legacy_test.test_dist_base import TestDistRunnerBase, runtime_main
 
 import paddle
@@ -31,7 +32,7 @@
 
 
 def cnn_model(data):
-    conv_pool_1 = fluid.nets.simple_img_conv_pool(
+    conv_pool_1 = nets.simple_img_conv_pool(
         input=data,
         filter_size=5,
         num_filters=20,
@@ -42,7 +43,7 @@ def cnn_model(data):
             initializer=paddle.nn.initializer.Constant(value=0.01)
         ),
     )
-    conv_pool_2 = fluid.nets.simple_img_conv_pool(
+    conv_pool_2 = nets.simple_img_conv_pool(
         input=conv_pool_1,
         filter_size=5,
         num_filters=50,
diff --git a/test/collective/fleet/pipeline_mnist_one_device.py b/test/collective/fleet/pipeline_mnist_one_device.py
@@ -14,6 +14,7 @@
 
 from functools import reduce
 
+from legacy_test import nets
 from legacy_test.test_dist_base import TestDistRunnerBase, runtime_main
 
 import paddle
@@ -31,7 +32,7 @@
 
 
 def cnn_model(data):
-    conv_pool_1 = fluid.nets.simple_img_conv_pool(
+    conv_pool_1 = nets.simple_img_conv_pool(
         input=data,
         filter_size=5,
         num_filters=20,
@@ -42,7 +43,7 @@ def cnn_model(data):
             initializer=paddle.nn.initializer.Constant(value=0.01)
         ),
     )
-    conv_pool_2 = fluid.nets.simple_img_conv_pool(
+    conv_pool_2 = nets.simple_img_conv_pool(
         input=conv_pool_1,
         filter_size=5,
         num_filters=50,
diff --git a/test/contrib/test_image_classification_fp16.py b/test/contrib/test_image_classification_fp16.py
@@ -22,6 +22,10 @@
 
 import numpy
 
+# TODO: remove sys.path.append
+sys.path.append("../legacy_test")
+import nets
+
 import paddle
 from paddle import fluid
 from paddle.static.amp import decorate
@@ -76,7 +80,7 @@ def layer_warp(block_func, input, ch_in, ch_out, count, stride):
 
 def vgg16_bn_drop(input):
     def conv_block(input, num_filter, groups, dropouts):
-        return fluid.nets.img_conv_group(
+        return nets.img_conv_group(
             input=input,
             pool_size=2,
             pool_stride=2,
diff --git a/test/legacy_test/dist_allreduce_op.py b/test/legacy_test/dist_allreduce_op.py
@@ -14,6 +14,7 @@
 
 from functools import reduce
 
+import nets
 from test_dist_base import TestDistRunnerBase, runtime_main
 
 import paddle
@@ -31,7 +32,7 @@
 
 
 def cnn_model(data):
-    conv_pool_1 = fluid.nets.simple_img_conv_pool(
+    conv_pool_1 = nets.simple_img_conv_pool(
         input=data,
         filter_size=5,
         num_filters=20,
@@ -42,7 +43,7 @@ def cnn_model(data):
             initializer=paddle.nn.initializer.Constant(value=0.01)
         ),
     )
-    conv_pool_2 = fluid.nets.simple_img_conv_pool(
+    conv_pool_2 = nets.simple_img_conv_pool(
         input=conv_pool_1,
         filter_size=5,
         num_filters=50,
diff --git a/test/legacy_test/dist_fleet_raw_program_optimizer.py b/test/legacy_test/dist_fleet_raw_program_optimizer.py
@@ -14,6 +14,7 @@
 
 from functools import reduce
 
+import nets
 from test_dist_base import TestDistRunnerBase, runtime_main
 
 import paddle
@@ -32,7 +33,7 @@
 
 
 def cnn_model(data):
-    conv_pool_1 = fluid.nets.simple_img_conv_pool(
+    conv_pool_1 = nets.simple_img_conv_pool(
         input=data,
         filter_size=5,
         num_filters=20,
@@ -43,7 +44,7 @@ def cnn_model(data):
             initializer=paddle.nn.initializer.Constant(value=0.01)
         ),
     )
-    conv_pool_2 = fluid.nets.simple_img_conv_pool(
+    conv_pool_2 = nets.simple_img_conv_pool(
         input=conv_pool_1,
         filter_size=5,
         num_filters=50,
diff --git a/test/legacy_test/dist_fleet_raw_program_optimizer_fuse_allreduce.py b/test/legacy_test/dist_fleet_raw_program_optimizer_fuse_allreduce.py
@@ -14,6 +14,7 @@
 
 from functools import reduce
 
+import nets
 from test_dist_base import TestDistRunnerBase, runtime_main
 
 import paddle
@@ -32,7 +33,7 @@
 
 
 def cnn_model(data):
-    conv_pool_1 = fluid.nets.simple_img_conv_pool(
+    conv_pool_1 = nets.simple_img_conv_pool(
         input=data,
         filter_size=5,
         num_filters=20,
@@ -43,7 +44,7 @@ def cnn_model(data):
             initializer=paddle.nn.initializer.Constant(value=0.01)
         ),
     )
-    conv_pool_2 = fluid.nets.simple_img_conv_pool(
+    conv_pool_2 = nets.simple_img_conv_pool(
         input=conv_pool_1,
         filter_size=5,
         num_filters=50,
diff --git a/test/legacy_test/dist_mnist.py b/test/legacy_test/dist_mnist.py
@@ -14,6 +14,7 @@
 
 from functools import reduce
 
+import nets
 from test_dist_base import TestDistRunnerBase, runtime_main
 
 import paddle
@@ -31,7 +32,7 @@
 
 
 def cnn_model(data):
-    conv_pool_1 = fluid.nets.simple_img_conv_pool(
+    conv_pool_1 = nets.simple_img_conv_pool(
         input=data,
         filter_size=5,
         num_filters=20,
@@ -42,7 +43,7 @@ def cnn_model(data):
             initializer=paddle.nn.initializer.Constant(value=0.01)
         ),
     )
-    conv_pool_2 = fluid.nets.simple_img_conv_pool(
+    conv_pool_2 = nets.simple_img_conv_pool(
         input=conv_pool_1,
         filter_size=5,
         num_filters=50,
diff --git a/test/legacy_test/dist_mnist_dgc.py b/test/legacy_test/dist_mnist_dgc.py
@@ -14,6 +14,7 @@
 
 from functools import reduce
 
+from legacy_test.nets import simple_img_conv_pool
 from legacy_test.test_dist_base import (
     TestDistRunnerBase,
     _insert_comm_op,
@@ -34,7 +35,7 @@
 
 
 def cnn_model(data):
-    conv_pool_1 = fluid.nets.simple_img_conv_pool(
+    conv_pool_1 = simple_img_conv_pool(
         input=data,
         filter_size=5,
         num_filters=20,
@@ -45,7 +46,7 @@ def cnn_model(data):
             initializer=paddle.nn.initializer.Constant(value=0.01)
         ),
     )
-    conv_pool_2 = fluid.nets.simple_img_conv_pool(
+    conv_pool_2 = simple_img_conv_pool(
         input=conv_pool_1,
         filter_size=5,
         num_filters=50,
diff --git a/test/legacy_test/dist_text_classification.py b/test/legacy_test/dist_text_classification.py
@@ -17,6 +17,7 @@
 import string
 import tarfile
 
+import nets
 from test_dist_base import TestDistRunnerBase, runtime_main
 
 import paddle
@@ -63,7 +64,7 @@ def conv_net(
         ),
     )
 
-    conv_3 = fluid.nets.sequence_conv_pool(
+    conv_3 = nets.sequence_conv_pool(
         input=emb,
         num_filters=num_filters,
         filter_size=window_size,
diff --git a/test/legacy_test/nets.py b/test/legacy_test/nets.py
@@ -13,10 +13,8 @@
 # limitations under the License.
 
 import paddle
-from . import layers
-from .data_feeder import check_variable_and_dtype, convert_dtype
-from ..utils import deprecated
-import paddle
+from paddle.fluid.data_feeder import check_variable_and_dtype, convert_dtype
+from paddle.utils import deprecated
 
 __all__ = [
     "simple_img_conv_pool",
@@ -494,9 +492,8 @@ def scaled_dot_product_attention(
     if not (queries.dtype == keys.dtype == values.dtype):
         raise TypeError(
             "The dtype of keys, values and queries should be the same."
-            "But received queries.dtype = %s, "
-            " keys.dtype = %s, values.dtype) = %s."
-            % (
+            "But received queries.dtype = {}, "
+            " keys.dtype = {}, values.dtype) = {}.".format(
                 convert_dtype(queries.dtype),
                 convert_dtype(keys.dtype),
                 convert_dtype(values.dtype),
diff --git a/test/legacy_test/test_desc_clone.py b/test/legacy_test/test_desc_clone.py
diff --git a/test/legacy_test/test_glu.py b/test/legacy_test/test_glu.py
diff --git a/test/legacy_test/test_image_classification_layer.py b/test/legacy_test/test_image_classification_layer.py
diff --git a/test/legacy_test/test_layers.py b/test/legacy_test/test_layers.py
diff --git a/test/legacy_test/test_load_state_dict_from_old_format.py b/test/legacy_test/test_load_state_dict_from_old_format.py
diff --git a/test/legacy_test/test_mix_precision_all_reduce_fuse.py b/test/legacy_test/test_mix_precision_all_reduce_fuse.py
diff --git a/test/legacy_test/test_multihead_attention.py b/test/legacy_test/test_multihead_attention.py
diff --git a/test/legacy_test/test_scaled_dot_product_attention.py b/test/legacy_test/test_scaled_dot_product_attention.py
