Merge pull request #242 from kaih70/master

fix demo & test cases

python/paddle_fl/mpc/examples/linear_reg_with_uci/README.md

@@ -84,19 +84,7 @@ Data owner encrypts data. Concrete operations are consistent with “Prepare Dat
 
 According to the suffix of file name, distribute encrypted data files to `/tmp ` directories of all 3 computation parties. For example, send `house_feature.part0` and `house_label.part0` to `/tmp` of party 0 with `scp` command.
 
-#### (3). Modify uci_demo.py
-
-Each computation party makes the following modifications on `uci_demo.py` according to the environment of machine.
-
-* Modify IP Information
-
-  Modify `localhost` in the following code as the IP address of the machine.
-
-  ```python
-  pfl_mpc.init("aby3", int(role), "localhost", server, int(port))
-  ```
-
-#### (4). Launch Demo on Each Party
+#### (3). Launch Demo on Each Party
 
 **Note** that Redis service is necessary for demo running. Remember to clear the cache of Redis server before launching demo on each computation party, in order to avoid any negative influences caused by the cached records in Redis. The following command can be used for clear Redis, where REDIS_BIN is the executable binary of redis-cli, SERVER and PORT represent the IP and port of Redis server respectively.
 
@@ -107,16 +95,16 @@ $REDIS_BIN -h $SERVER -p $PORT flushall
 Launch demo on each computation party with the following command,
 
 ```
-$PYTHON_EXECUTABLE uci_demo.py $PARTY_ID $SERVER $PORT
+$PYTHON_EXECUTABLE uci_demo.py $PARTY_ID $SERVER $PORT $SELF_ADDR
 ```
 
-where PYTHON_EXECUTABLE is the python which installs PaddleFL, PARTY_ID is the ID of computation party, which is 0, 1, or 2, SERVER and PORT represent the IP and port of Redis server respectively.
+where PYTHON_EXECUTABLE is the python which installs PaddleFL, PARTY_ID is the ID of computation party, which is 0, 1, or 2, SERVER and PORT represent the IP and port of Redis server respectively, SELF_ADDR represents the IP address of the machine.
 
 Similarly, training loss with cypher text format would be printed on the screen of each computation party. And at the same time, the loss and predictions would be saved in `/tmp` directory.
 
 **Note** that remember to delete the loss and prediction files in `/tmp` directory generated in last running, in case of any influence on the decrypted results of current running.
 
-#### (5). Decrypt Loss and Prediction Data
+#### (4). Decrypt Loss and Prediction Data
 
 Each computation party sends `uci_loss.part` and `uci_prediction.part` files in `/tmp` directory to the `/tmp` directory of data owner. Data owner decrypts and gets the plain text of loss and predictions with ` load_decrypt_data()` in `process_data.py`.
 

python/paddle_fl/mpc/examples/linear_reg_with_uci/README_CN.md

@@ -85,20 +85,8 @@ fi
 
 `house_feature.part0`和`house_label.part0`发送到party0的/tmp目录下。
 
-#### 3. 计算party修改uci_demo.py脚本
 
-各计算party根据自己的机器环境，对uci_demo.py做如下改动：
-
-* 修改IP信息
-
-  将脚本如下内容中的`localhost`修改为自己的IP地址：
-
-  ```python
-  pfl_mpc.init("aby3", int(role), "localhost", server, int(port))
-  ```
-
-
-#### 4. 各计算party启动demo
+#### 3. 各计算party启动demo
 
 **注意**：运行需要用到redis服务。为了确保redis中已保存的数据不会影响demo的运行，请在各计算party启动demo之前，使用如下命令清空redis。其中，REDIS_BIN表示redis-cli可执行程序，SERVER和PORT分别表示redis server的IP地址和端口号。
 
@@ -109,16 +97,16 @@ $REDIS_BIN -h $SERVER -p $PORT flushall
 在各计算party分别执行以下命令，启动demo：
 
 ```
-$PYTHON_EXECUTABLE uci_demo.py $PARTY_ID $SERVER $PORT
+$PYTHON_EXECUTABLE uci_demo.py $PARTY_ID $SERVER $PORT $SELF_ADDR
 ```
 
-其中，PYTHON_EXECUTABLE表示自己安装了PaddleFL的python，PARTY_ID表示计算party的编号，值为0、1或2，SERVER和PORT分别表示redis server的IP地址和端口号。
+其中，PYTHON_EXECUTABLE表示自己安装了PaddleFL的python，PARTY_ID表示计算party的编号，值为0、1或2，SERVER和PORT分别表示redis server的IP地址和端口号, SELF_ADDR为自己的IP地址。
 
 同样地，运行之后将在各计算party的屏幕上打印训练过程中的密文loss数据。同时，对应的密文loss和prediction数据将会保存到`/tmp`目录下的文件中，文件命名格式类似于步骤1中所述。
 
 **注意**：再次启动运行demo之前，请先将上次在`/tmp`保存的loss和prediction文件删除，以免影响本次密文数据的恢复结果。
 
-#### 5. 数据方解密loss和prediction
+#### 4. 数据方解密loss和prediction
 
 各计算party将`/tmp`目录下的`uci_loss.part`和`uci_prediction.part`文件发送到数据方的/tmp目录下。数据方使用process_data.py脚本中的load_decrypt_data()解密恢复出loss数据和prediction数据。
 

python/paddle_fl/mpc/examples/linear_reg_with_uci/process_data.py

@@ -49,7 +49,7 @@ def encrypted_housing_labels():
     mpc_du.save_shares(encrypted_housing_labels, "/tmp/house_label")
 
 
-def generate_encrypted_data_online(role, server, port):
+def generate_encrypted_data_online(role, server, port,  selfaddr="localhost"):
     """
     generate encrypted samples
     """
@@ -65,7 +65,7 @@ def encrypted_data_generator(data_location_party, sample_reader, index1, index2_
         main_program = fluid.Program()
         startup_program = fluid.Program()
         with fluid.program_guard(main_program, startup_program):
-            pfl_mpc.init("aby3", int(role), "localhost", server, int(port))
+            pfl_mpc.init("aby3", int(role), selfaddr, server, int(port))
             input = fluid.data(name='input', shape=[feature_num], dtype='float32')
             out = pfl_mpc.layers.share(input, party_id=data_location_party)
 

python/paddle_fl/mpc/examples/linear_reg_with_uci/uci_demo.py

@@ -30,7 +30,12 @@
 mpc_du = get_datautils(mpc_protocol_name)
 
 role, server, port = sys.argv[1], sys.argv[2], sys.argv[3]
-pfl_mpc.init(mpc_protocol_name, int(role), "localhost", server, int(port))
+
+selfaddr="localhost"
+if len(sys.argv) >= 5:
+    selfaddr=sys.argv[4]
+
+pfl_mpc.init(mpc_protocol_name, int(role), selfaddr, server, int(port))
 #pfl_mpc.init(mpc_protocol_name, int(role), "localhost", server, int(port), "localhost:90001;localhost:90002;localhost:90003", "gloo")
 #pfl_mpc.init(mpc_protocol_name, int(role), "localhost", server, int(port), "localhost:90001;localhost:90002;localhost:90003", "grpc")
 
@@ -41,7 +46,7 @@
 mpc_data_dir = "./mpc_data/"
 
 # generate share online
-feature_reader, label_reader = process_data.generate_encrypted_data_online(role, server, port)
+feature_reader, label_reader = process_data.generate_encrypted_data_online(role, server, port, selfaddr)
 
 """
 # load shares from file
@@ -96,7 +101,7 @@
     start_time = time.time()
     step = 0
 
-    # Method 1: feed data directly 
+    # Method 1: feed data directly
     # for feature, label in zip(batch_feature(), batch_label()):
     #     mpc_loss = exe.run(feed={"x": feature, "y": label}, fetch_list=[avg_loss])
 
@@ -113,7 +118,7 @@
         with open(loss_file, 'ab') as f:
             f.write(np.array(mpc_loss).tostring())
         step += 1
- 
+
     end_time = time.time()
     print('Mpc Training of Epoch={} Batch_size={}, epoch_cost={:.4f} s'
           .format(epoch_id, BATCH_SIZE, (end_time - start_time)))

python/paddle_fl/mpc/tests/unittests/test_op_conv.py

@@ -471,10 +471,10 @@ def init_paddings(self):
         self.pad = [0, 0, 1, 2]
         self.padding_algorithm = "EXPLICIT"
 
-    def test_check_grad(self):
-        place = core.CPUPlace()
-        self.check_grad_with_place(
-            place, {'Input', 'Filter'}, 'Output', max_relative_error=5)
+    #def test_check_grad(self):
+    #    place = core.CPUPlace()
+    #    self.check_grad_with_place(
+    #        place, {'Input', 'Filter'}, 'Output', max_relative_error=5)
 
     #def test_check_grad(self):
     #    pass