[Clean Fluid API]Remove API: unsqueeze, use paddle.unsqueeze to replace...

[Clean Fluid API]Remove API: unsqueeze, use paddle.unsqueeze to replace paddle.fluid.layers.unsqueeze (#47936)

* Remove API: unsqueeze
	Replace the funcation call.
	Use paddle.unsqueeze to replace paddle.fluid.layers.unsqueeze

* Remove API: unsqueeze
	remove unsqueeze which is not used in Paddle 2.0

* Remove API: unsqueeze
	fix the bug by adding 'import paddle' to file
tests/unittests/test_rnn_cell_api.py

* Modify the call of unsqueeze from old style to new style

* fix the call arguments of unsqueeze

* replace unqueeze with paddle.unsqueeze

* fix the error in PR-CI-Static-Check, name 'paddle' is not defined

* fix the error in static in unsqueeze

python/paddle/fluid/layers/nn.py

@@ -71,7 +71,6 @@ __all__ = [
     'spectral_norm',
     'one_hot',
     'autoincreased_step_counter',
-    'unsqueeze',
     'lod_reset',
     'clip',
     'clip_by_norm',

python/paddle/fluid/tests/unittests/dygraph_to_static/seq2seq_dygraph_model.py

@@ -189,7 +189,7 @@ class BaseModel(fluid.dygraph.Layer):
         return paddle.reshape(x, shape=(-1, self.beam_size, x.shape[1]))
 
     def _expand_to_beam_size(self, x):
-        x = fluid.layers.unsqueeze(x, [1])
+        x = paddle.unsqueeze(x, [1])
         expand_shape = [-1] * len(x.shape)
         expand_shape[1] = self.beam_size * x.shape[1]
         x = paddle.expand(x, expand_shape)
@@ -401,7 +401,7 @@ class BaseModel(fluid.dygraph.Layer):
         dec_cell = [self._expand_to_beam_size(ele) for ele in dec_cell]
 
         batch_pos = paddle.expand(
-            fluid.layers.unsqueeze(
+            paddle.unsqueeze(
                 to_variable(np.arange(0, self.batch_size, 1, dtype="int64")),
                 [1],
             ),
@@ -450,7 +450,7 @@ class BaseModel(fluid.dygraph.Layer):
 
             step_log_probs = paddle.multiply(
                 paddle.expand(
-                    fluid.layers.unsqueeze(beam_finished, [2]),
+                    paddle.unsqueeze(beam_finished, [2]),
                     [-1, -1, self.tar_vocab_size],
                 ),
                 noend_mask_tensor,
@@ -664,7 +664,7 @@ class AttentionModel(fluid.dygraph.Layer):
         return paddle.reshape(x, shape=(-1, x.shape[2]))
 
     def tile_beam_merge_with_batch(self, x):
-        x = fluid.layers.unsqueeze(x, [1])  # [batch_size, 1, ...]
+        x = paddle.unsqueeze(x, [1])  # [batch_size, 1, ...]
         expand_shape = [-1] * len(x.shape)
         expand_shape[1] = self.beam_size * x.shape[1]
         x = paddle.expand(x, expand_shape)  # [batch_size, beam_size, ...]
@@ -684,7 +684,7 @@ class AttentionModel(fluid.dygraph.Layer):
         return paddle.reshape(x, shape=(-1, self.beam_size, x.shape[1]))
 
     def _expand_to_beam_size(self, x):
-        x = fluid.layers.unsqueeze(x, [1])
+        x = paddle.unsqueeze(x, [1])
         expand_shape = [-1] * len(x.shape)
         expand_shape[1] = self.beam_size * x.shape[1]
         x = paddle.expand(x, expand_shape)
@@ -703,7 +703,7 @@ class AttentionModel(fluid.dygraph.Layer):
         return paddle.gather_nd(x, topk_coordinates)
 
     def attention(self, query, enc_output, mask=None):
-        query = fluid.layers.unsqueeze(query, [1])
+        query = paddle.unsqueeze(query, [1])
         memory = self.attn_fc(enc_output)
         attn = paddle.matmul(query, memory, transpose_y=True)
 

python/paddle/fluid/tests/unittests/dygraph_to_static/test_bmn.py

@@ -135,7 +135,7 @@ class Conv1D(fluid.dygraph.Layer):
         )
 
     def forward(self, x):
-        x = fluid.layers.unsqueeze(input=x, axes=[2])
+        x = paddle.unsqueeze(x, axis=[2])
         x = self._conv2d(x)
         x = paddle.squeeze(x, axis=[2])
         return x

python/paddle/fluid/tests/unittests/dygraph_to_static/transformer_dygraph_model.py

@@ -775,7 +775,7 @@ class Transformer(Layer):
             finished = layers.cast(finished, dtype=probs.dtype)
             probs = paddle.multiply(
                 paddle.expand(
-                    layers.unsqueeze(finished, [2]),
+                    paddle.unsqueeze(finished, [2]),
                     [-1, -1, self.trg_vocab_size],
                 ),
                 noend_mask_tensor,
@@ -805,7 +805,7 @@ class Transformer(Layer):
         noend_array[eos_id] = 0
         noend_mask_tensor = to_variable(np.array(noend_array, dtype="float32"))
         batch_pos = paddle.expand(
-            layers.unsqueeze(
+            paddle.unsqueeze(
                 to_variable(np.arange(0, batch_size, 1, dtype="int64")), [1]
             ),
             [-1, beam_size],

python/paddle/fluid/tests/unittests/ipu/test_unsqueeze_op_ipu.py

@@ -47,7 +47,7 @@ class TestBase(IPUOpTest):
         x = paddle.static.data(
             name=self.feed_list[0], shape=self.feed_shape[0], dtype='float32'
         )
-        out = paddle.fluid.layers.unsqueeze(x, **self.attrs)
+        out = paddle.unsqueeze(x, **self.attrs)
         self.fetch_list = [out.name]
 
     def run_model(self, exec_mode):

python/paddle/fluid/tests/unittests/test_dynamic_rnn_stop_gradient.py

@@ -47,7 +47,7 @@ def build_and_run_program(place, batch_size, beam_size, stop_gradient=False):
             bs = layers.cast(bs, 'int64')
         bs.stop_gradient = stop_gradient
         batch_pos = paddle.expand(
-            layers.unsqueeze(paddle.arange(0, bs, 1, dtype=bs.dtype), [1]),
+            paddle.unsqueeze(paddle.arange(0, bs, 1, dtype=bs.dtype), [1]),
             [-1, beam_size],
         )
         topk_coordinates = paddle.stack([batch_pos, indices], axis=2)

python/paddle/fluid/tests/unittests/test_eager_deletion_while_op.py

@@ -128,7 +128,7 @@ class TestEagerDeletionWhileOpBase(unittest.TestCase):
 
         sum_result = paddle.tensor.array_read(array=mem_array, i=j)
         sum_result.persistable = True
-        tmp = layers.unsqueeze(sum_result, axes=[0])
+        tmp = paddle.unsqueeze(sum_result, axis=[0])
         tmp = paddle.expand(tmp, [10, -1])
         fc = layers.fc(tmp, size=256)
         loss = paddle.mean(sum_result)

python/paddle/fluid/tests/unittests/test_layers.py

@@ -2683,7 +2683,7 @@ class TestBook(LayerTest):
         # TODO(minqiyang): dygraph do not support lod now
         with self.static_graph():
             x = layers.data(name='x', shape=[8, 2], dtype='float32')
-            out = layers.unsqueeze(input=x, axes=[1])
+            out = paddle.unsqueeze(x, axis=[1])
             return out
 
     def test_sequence_scatter(self):

python/paddle/fluid/tests/unittests/test_recurrent_op.py

@@ -632,7 +632,7 @@ class RecurrentOpSubBlockTest(RecurrentOpTest1):
             step_in = rnn.step_input(x)
             concat_in = layers.concat([step_in, pre_h], 1)
             new_h = paddle.matmul(concat_in, w2)
-            new_h = layers.unsqueeze(new_h, [1])
+            new_h = paddle.unsqueeze(new_h, [1])
             new_h, _ = dot_attention(new_h, y)
             new_h = paddle.squeeze(new_h, [1])