add dpsgd benchmark

paddle_fl/core/master/fl_job.py

@@ -201,11 +201,14 @@ class FLRunTimeJob(FLJobBase):
         main_fn = "%s/trainer.main.program" % folder_name
         self._trainer_main_program = self._load_program(main_fn)
         
-        send_fn = "%s/trainer.send.program" % folder_name
-        self._trainer_send_program = self._load_program(send_fn)
-
-        recv_fn = "%s/trainer.recv.program" % folder_name
-        self._trainer_recv_program = self._load_program(recv_fn)
+        try:
+            send_fn = "%s/trainer.send.program" % folder_name
+            self._trainer_send_program = self._load_program(send_fn)
+
+            recv_fn = "%s/trainer.recv.program" % folder_name
+            self._trainer_recv_program = self._load_program(recv_fn)
+        except:
+            pass
 
         endpoints_fn = "%s/endpoints" % folder
         self._endpoints = self._load_endpoints(endpoints_fn)

paddle_fl/core/strategy/fl_strategy_base.py

@@ -128,7 +128,7 @@ class DPSGDStrategy(FLStrategyBase):
                              trainers=trainers,
                              sync_mode=sync_mode,
                              startup_program=startup_program)
-        main = transpiler.get_trainer_program()
+        main = transpiler.get_trainer_program(wait_port=False)
         job._trainer_startup_programs.append(startup_program)
         job._trainer_main_programs.append(main)
 

paddle_fl/examples/dpsgd_demo/fl_master.py

@@ -8,39 +8,33 @@ class Model(object):
     def __init__(self):
         pass
 
-    def mlp(self, inputs, label, hidden_size=128):
-        self.concat = fluid.layers.concat(inputs, axis=1)
-        self.fc1 = fluid.layers.fc(input=self.concat, size=256, act='relu')
-        self.fc2 = fluid.layers.fc(input=self.fc1, size=128, act='relu')
-        self.predict = fluid.layers.fc(input=self.fc2, size=2, act='softmax')
-        self.sum_cost = fluid.layers.cross_entropy(input=self.predict, label=label)
-        self.accuracy = fluid.layers.accuracy(input=self.predict, label=label)
-        self.loss = fluid.layers.reduce_mean(self.sum_cost)
+    def lr_network(self):
+        self.inputs = fluid.layers.data(name='img', shape=[1, 28, 28], dtype="float32")
+        self.label = fluid.layers.data(name='label', shape=[1],dtype='int64')
+        self.predict = fluid.layers.fc(input=self.inputs, size=10, act='softmax')
+        self.sum_cost = fluid.layers.cross_entropy(input=self.predict, label=self.label)
+        self.accuracy = fluid.layers.accuracy(input=self.predict, label=self.label)
+        self.loss = fluid.layers.mean(self.sum_cost)
         self.startup_program = fluid.default_startup_program()
 
-inputs = [fluid.layers.data( \
-            name=str(slot_id), shape=[5],
-            dtype="float32")
-          for slot_id in range(3)]
-label = fluid.layers.data( \
-            name="label",
-            shape=[1],
-            dtype='int64')
 
 model = Model()
-model.mlp(inputs, label)
+model.lr_network()
 
-STEP_EPSILON = 10
+STEP_EPSILON = 0.1
 DELTA = 0.00001
 SIGMA = math.sqrt(2.0 * math.log(1.25/DELTA)) / STEP_EPSILON
-CLIP = 10.0
-batch_size = 1
+CLIP = 4.0
+batch_size = 64
 
 job_generator = JobGenerator()
-optimizer = fluid.optimizer.Dpsgd(0.1, clip=CLIP, batch_size=float(batch_size), sigma=0.0 * SIGMA)
+optimizer = fluid.optimizer.Dpsgd(0.1, clip=CLIP, batch_size=float(batch_size), sigma=CLIP * SIGMA)
+# optimizer = fluid.optimizer.SGD(learning_rate=0.1)
 job_generator.set_optimizer(optimizer)
 job_generator.set_losses([model.loss])
 job_generator.set_startup_program(model.startup_program)
+job_generator.set_infer_feed_and_target_names(
+    [model.inputs.name, model.label.name], [model.loss.name, model.accuracy.name])
 
 build_strategy = FLStrategyFactory()
 build_strategy.dpsgd = True

paddle_fl/examples/dpsgd_demo/fl_trainer.py

 from paddle_fl.core.trainer.fl_trainer import FLTrainerFactory
 from paddle_fl.core.master.fl_job import FLRunTimeJob
-import numpy as np
+import numpy
 import sys
+import paddle
+import paddle.fluid as fluid
+import logging
+import math
 
-def reader():
-    for i in range(1000):
-        data_dict = {}
-        for i in range(3):
-            data_dict[str(i)] = np.random.rand(1, 5).astype('float32')
-        data_dict["label"] = np.random.randint(2, size=(1, 1)).astype('int64')
-        yield data_dict
+logging.basicConfig(filename="test.log", filemode="w", format="%(asctime)s %(name)s:%(levelname)s:%(message)s", datefmt="%d-%M-%Y %H:%M:%S", level=logging.DEBUG)
 
 trainer_id = int(sys.argv[1]) # trainer id for each guest
 job_path = "fl_job_config"
@@ -18,11 +16,55 @@ job.load_trainer_job(job_path, trainer_id)
 trainer = FLTrainerFactory().create_fl_trainer(job)
 trainer.start()
 
-step_i = 0
+test_program = trainer._main_program.clone(for_test=True)
+
+train_reader = paddle.batch(
+        paddle.reader.shuffle(paddle.dataset.mnist.train(), buf_size=500),
+        batch_size=64)
+test_reader = paddle.batch(
+        paddle.dataset.mnist.test(), batch_size=64)
+
+img = fluid.layers.data(name='img', shape=[1, 28, 28], dtype='float32')
+label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+feeder = fluid.DataFeeder(feed_list=[img, label], place=fluid.CPUPlace())
+
+def train_test(train_test_program, train_test_feed, train_test_reader):
+        acc_set = []
+        for test_data in train_test_reader():
+            acc_np = trainer.exe.run(
+                program=train_test_program,
+                feed=train_test_feed.feed(test_data),
+                fetch_list=["accuracy_0.tmp_0"])
+            acc_set.append(float(acc_np[0]))
+        acc_val_mean = numpy.array(acc_set).mean()
+        return acc_val_mean
+
+def compute_privacy_budget(sample_ratio, epsilon, step, delta):
+    E = 2 * epsilon * math.sqrt(step * sample_ratio)
+    print("({0}, {1})-DP".format(E, delta))
+
+output_folder = "model_node%d" % trainer_id
+epoch_id = 0
+step = 0
 while not trainer.stop():
-    step_i += 1
-    print("batch %d start train" % (step_i))
-    for data in reader():
-        trainer.run(feed=data, fetch=[])
-    if step_i % 100 == 0:
-        trainer.save_inference_program(output_folder)
+    epoch_id += 1
+    if epoch_id > 40:
+        break
+    print("epoch %d start train" % (epoch_id))
+    for step_id, data in enumerate(train_reader()):
+        acc = trainer.run(feeder.feed(data), fetch=["accuracy_0.tmp_0"])
+        step += 1
+    # print("acc:%.3f" % (acc[0]))
+    
+    acc_val = train_test(
+        train_test_program=test_program,
+        train_test_reader=test_reader,
+        train_test_feed=feeder)
+
+    print("Test with epoch %d, accuracy: %s" % (epoch_id, acc_val))
+    compute_privacy_budget(sample_ratio=0.001, epsilon=0.1, step=step, delta=0.00001)
+   
+    save_dir = (output_folder + "/epoch_%d") % epoch_id
+    trainer.save_inference_program(output_folder)
+
+

paddle_fl/examples/dpsgd_demo/stop.sh

+#!/bin/bash
+
+echo "Stop service!"
+
+ps -ef | grep -E "fl" | grep -v grep | awk '{print $2}' | xargs kill -9