[remove fluid.layers.cross_entropy] remove unit tests (part 1) (#48726)

* replace layers.cross_entropy with paddle.entropy

* fix args

* fix codestyle

python/paddle/fluid/contrib/slim/tests/imperative_test_utils.py

@@ -70,7 +70,9 @@ def train_lenet(lenet, reader, optimizer):
         label = paddle.to_tensor(y_data)
 
         out = lenet(img)
-        loss = fluid.layers.cross_entropy(out, label)
+        loss = paddle.nn.functional.cross_entropy(
+            out, label, reduction='none', use_softmax=False
+        )
         avg_loss = paddle.mean(loss)
         avg_loss.backward()
 

python/paddle/fluid/contrib/slim/tests/test_graph.py

@@ -47,7 +47,9 @@ def conv_block():
         act="relu",
     )
     prediction = fluid.layers.fc(input=conv_pool_2, size=10, act='softmax')
-    loss = fluid.layers.cross_entropy(input=prediction, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=prediction, label=label, reduction='none', use_softmax=False
+    )
     avg_loss = paddle.mean(loss)
     return [img, label], avg_loss
 

python/paddle/fluid/contrib/slim/tests/test_imperative_qat.py

@@ -130,7 +130,9 @@ class TestImperativeQat(unittest.TestCase):
                     label = fluid.dygraph.to_variable(y_data)
                     out = lenet(img)
                     acc = paddle.static.accuracy(out, label)
-                    loss = fluid.layers.cross_entropy(out, label)
+                    loss = paddle.nn.functional.cross_entropy(
+                        out, label, reduction='none', use_softmax=False
+                    )
                     avg_loss = paddle.mean(loss)
                     avg_loss.backward()
                     adam.minimize(avg_loss)

python/paddle/fluid/contrib/slim/tests/test_imperative_qat_amp.py

@@ -119,7 +119,9 @@ class TestImperativeQatAmp(unittest.TestCase):
                 with paddle.amp.auto_cast():
                     out = model(img)
                     acc = paddle.static.accuracy(out, label)
-                    loss = fluid.layers.cross_entropy(out, label)
+                    loss = paddle.nn.functional.cross_entropy(
+                        out, label, reduction='none', use_softmax=False
+                    )
                     avg_loss = paddle.mean(loss)
                 scaled_loss = scaler.scale(avg_loss)
                 scaled_loss.backward()
@@ -129,7 +131,9 @@ class TestImperativeQatAmp(unittest.TestCase):
             else:
                 out = model(img)
                 acc = paddle.static.accuracy(out, label)
-                loss = fluid.layers.cross_entropy(out, label)
+                loss = paddle.nn.functional.cross_entropy(
+                    out, label, reduction='none', use_softmax=False
+                )
                 avg_loss = paddle.mean(loss)
                 avg_loss.backward()
 

python/paddle/fluid/contrib/slim/tests/test_imperative_qat_lsq.py

@@ -171,7 +171,9 @@ class TestImperativeQatLSQ(unittest.TestCase):
                 label = fluid.dygraph.to_variable(y_data)
                 out = lenet(img)
                 acc = paddle.static.accuracy(out, label)
-                loss = fluid.layers.cross_entropy(out, label)
+                loss = paddle.nn.functional.cross_entropy(
+                    out, label, reduction='none', use_softmax=False
+                )
                 avg_loss = paddle.mean(loss)
 
                 avg_loss.backward()

python/paddle/fluid/contrib/slim/tests/test_quantization_mkldnn_pass.py

@@ -47,7 +47,9 @@ def conv_net(img, label):
         act="relu",
     )
     prediction = fluid.layers.fc(input=conv_pool_2, size=10, act='softmax')
-    loss = fluid.layers.cross_entropy(input=prediction, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=prediction, label=label, reduction='none', use_softmax=False
+    )
     avg_loss = paddle.mean(loss)
     return avg_loss
 

python/paddle/fluid/contrib/slim/tests/test_quantization_pass.py

@@ -39,7 +39,9 @@ def linear_fc(num):
     hidden = data
     for _ in range(num):
         hidden = fluid.layers.fc(hidden, size=128, act='relu')
-    loss = fluid.layers.cross_entropy(input=hidden, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=hidden, label=label, reduction='none', use_softmax=False
+    )
     loss = paddle.mean(loss)
     return loss
 
@@ -87,7 +89,9 @@ def residual_block(num, quant_skip_pattern=None):
             input=hidden, pool_size=2, pool_type='avg', pool_stride=2
         )
     fc = fluid.layers.fc(input=pool, size=10)
-    loss = fluid.layers.cross_entropy(input=fc, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=fc, label=label, reduction='none', use_softmax=False
+    )
     loss = paddle.mean(loss)
     return loss
 
@@ -115,7 +119,9 @@ def conv_net(img, label, quant_skip_pattern):
     hidden = fluid.layers.fc(input=conv_pool_2, size=100, act='relu')
     with fluid.name_scope(quant_skip_pattern):
         prediction = fluid.layers.fc(input=hidden, size=10, act='softmax')
-    loss = fluid.layers.cross_entropy(input=prediction, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=prediction, label=label, reduction='none', use_softmax=False
+    )
     avg_loss = paddle.mean(loss)
     return avg_loss
 
@@ -756,7 +762,9 @@ def quant_dequant_residual_block(num, quant_skip_pattern=None):
         )
         pool_add = paddle.nn.functional.relu(paddle.add(x=pool1, y=pool2))
     fc = fluid.layers.fc(input=pool_add, size=10)
-    loss = fluid.layers.cross_entropy(input=fc, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=fc, label=label, reduction='none', use_softmax=False
+    )
     loss = paddle.mean(loss)
     return loss
 

python/paddle/fluid/contrib/slim/tests/test_quantization_scale_pass.py

@@ -55,7 +55,9 @@ def conv_net(img, label):
     )
     hidden = fluid.layers.fc(input=conv_pool_2, size=100, act='relu')
     prediction = fluid.layers.fc(input=hidden, size=10, act='softmax')
-    loss = fluid.layers.cross_entropy(input=prediction, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=prediction, label=label, reduction='none', use_softmax=False
+    )
     avg_loss = paddle.mean(loss)
     return avg_loss
 

python/paddle/fluid/contrib/slim/tests/test_quantize_transpiler_v2.py

@@ -52,7 +52,9 @@ def conv_net(img, label):
     with fluid.name_scope("skip_quant"):
         hidden = fluid.layers.fc(input=conv_pool_1, size=100, act='relu')
     prediction = fluid.layers.fc(input=hidden, size=10, act='softmax')
-    loss = fluid.layers.cross_entropy(input=prediction, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=prediction, label=label, reduction='none', use_softmax=False
+    )
     avg_loss = paddle.mean(loss)
     return avg_loss
 

python/paddle/fluid/contrib/slim/tests/test_user_defined_quantization.py

@@ -57,7 +57,9 @@ def conv_net(img, label):
     )
     hidden = fluid.layers.fc(input=conv_pool_2, size=100, act='relu')
     prediction = fluid.layers.fc(input=hidden, size=10, act='softmax')
-    loss = fluid.layers.cross_entropy(input=prediction, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=prediction, label=label, reduction='none', use_softmax=False
+    )
     avg_loss = paddle.mean(loss)
     return avg_loss
 

python/paddle/fluid/contrib/tests/test_quantize_transpiler.py

@@ -30,7 +30,9 @@ def linear_fc(num):
     hidden = data
     for _ in range(num):
         hidden = fluid.layers.fc(hidden, size=128, act='relu')
-    loss = fluid.layers.cross_entropy(input=hidden, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=hidden, label=label, reduction='none', use_softmax=False
+    )
     loss = paddle.mean(loss)
     return loss
 
@@ -58,7 +60,9 @@ def residual_block(num):
         short = conv_bn_layer(hidden, 16, 1, 1, 0, act=None)
         hidden = paddle.nn.functional.relu(paddle.add(x=conv, y=short))
     fc = fluid.layers.fc(input=hidden, size=10)
-    loss = fluid.layers.cross_entropy(input=fc, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=fc, label=label, reduction='none', use_softmax=False
+    )
     loss = paddle.mean(loss)
     return loss
 
@@ -82,7 +86,9 @@ def conv_net(img, label):
         act="relu",
     )
     prediction = fluid.layers.fc(input=conv_pool_2, size=10, act='softmax')
-    loss = fluid.layers.cross_entropy(input=prediction, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=prediction, label=label, reduction='none', use_softmax=False
+    )
     avg_loss = paddle.mean(loss)
     return avg_loss
 

python/paddle/fluid/contrib/tests/test_weight_decay_extend.py

@@ -86,7 +86,9 @@ def bow_net(
     fc_1 = fluid.layers.fc(input=bow_tanh, size=hid_dim, act="tanh")
     fc_2 = fluid.layers.fc(input=fc_1, size=hid_dim2, act="tanh")
     prediction = fluid.layers.fc(input=[fc_2], size=class_dim, act="softmax")
-    cost = fluid.layers.cross_entropy(input=prediction, label=label)
+    cost = paddle.nn.functional.cross_entropy(
+        input=prediction, label=label, reduction='none', use_softmax=False
+    )
     avg_cost = paddle.mean(x=cost)
 
     return avg_cost

python/paddle/fluid/incubate/fleet/tests/fleet_deep_ctr.py

@@ -150,7 +150,9 @@ def model():
     auc_var, batch_auc_var, auc_states = paddle.static.auc(
         input=predict, label=label
     )
-    cost = fluid.layers.cross_entropy(input=predict, label=label)
+    cost = paddle.nn.functional.cross_entropy(
+        input=predict, label=label, reduction='none', use_softmax=False
+    )
     avg_cost = paddle.mean(x=cost)
 
     return datas, avg_cost, predict, train_file_path

python/paddle/fluid/tests/book/notest_understand_sentiment.py

@@ -47,7 +47,9 @@ def convolution_net(
     prediction = fluid.layers.fc(
         input=[conv_3, conv_4], size=class_dim, act="softmax"
     )
-    cost = fluid.layers.cross_entropy(input=prediction, label=label)
+    cost = paddle.nn.functional.cross_entropy(
+        input=prediction, label=label, reduction='none', use_softmax=False
+    )
     avg_cost = paddle.mean(cost)
     accuracy = paddle.static.accuracy(input=prediction, label=label)
     return avg_cost, accuracy, prediction
@@ -82,7 +84,9 @@ def stacked_lstm_net(
     prediction = fluid.layers.fc(
         input=[fc_last, lstm_last], size=class_dim, act='softmax'
     )
-    cost = fluid.layers.cross_entropy(input=prediction, label=label)
+    cost = paddle.nn.functional.cross_entropy(
+        input=prediction, label=label, reduction='none', use_softmax=False
+    )
     avg_cost = paddle.mean(cost)
     accuracy = paddle.static.accuracy(input=prediction, label=label)
     return avg_cost, accuracy, prediction

python/paddle/fluid/tests/book/test_image_classification.py

@@ -119,7 +119,9 @@ def train(net_type, use_cuda, save_dirname, is_local):
         raise ValueError("%s network is not supported" % net_type)
 
     predict = fluid.layers.fc(input=net, size=classdim, act='softmax')
-    cost = fluid.layers.cross_entropy(input=predict, label=label)
+    cost = paddle.nn.functional.cross_entropy(
+        input=predict, label=label, reduction='none', use_softmax=False
+    )
     avg_cost = paddle.mean(cost)
     acc = paddle.static.accuracy(input=predict, label=label)
 

python/paddle/fluid/tests/book/test_recognize_digits.py

@@ -30,7 +30,9 @@ BATCH_SIZE = 64
 
 def loss_net(hidden, label):
     prediction = fluid.layers.fc(input=hidden, size=10, act='softmax')
-    loss = fluid.layers.cross_entropy(input=prediction, label=label)
+    loss = paddle.nn.functional.cross_entropy(
+        input=prediction, label=label, reduction='none', use_softmax=False
+    )
     avg_loss = paddle.mean(loss)
     acc = paddle.static.accuracy(input=prediction, label=label)
     return prediction, avg_loss, acc

python/paddle/fluid/tests/book/test_word2vec_book.py

@@ -96,7 +96,12 @@ def train(
         predict_word = fluid.layers.fc(
             input=hidden1, size=dict_size, act='softmax'
         )
-        cost = fluid.layers.cross_entropy(input=predict_word, label=words[4])
+        cost = paddle.nn.functional.cross_entropy(
+            input=predict_word,
+            label=words[4],
+            reduction='none',
+            use_softmax=False,
+        )
         avg_cost = paddle.mean(cost)
         return avg_cost, predict_word
 

python/paddle/fluid/tests/test_error_clip.py

@@ -31,7 +31,9 @@ with fluid.program_guard(main_program=prog):
 
     label = fluid.layers.data(name='y', shape=[1], dtype='int64')
 
-    cost = fluid.layers.cross_entropy(input=predict, label=label)
+    cost = paddle.nn.functional.cross_entropy(
+        input=predict, label=label, reduction='none', use_softmax=False
+    )
     avg_cost = paddle.mean(cost)
 
 prog_clip = prog.clone()