update DPBasicBlocks and timer

code/deeppoly.py

@@ -388,10 +388,14 @@ def __init__(self, nested: resnet.BasicBlock, in_feature):
         self.out_features = output_shape[0] * output_shape[1] * output_shape[2]
 
     def forward(self, x):
-        lb = []
-        up = []
+        lbs = []
+        ubs = []
+        slbs = []
+        subs = []
         for path in self.paths:
             in_feature = self.in_feature
+            slb = 1
+            sub = 1
             for layer in path:
                 if type(layer) == nn.Conv2d:
                     dp_conv = DPConv(layer, in_feature)
@@ -401,7 +405,8 @@ def forward(self, x):
                     dp_relu = DPReLU(in_feature)
                     dp_relu(x)
                     in_feature = dp_relu.out_features
-            lb
+            lbs.append(x.lb)
+            ubs.append(x.ub)
 
         return x
 

code/evaluate

@@ -1,7 +1,7 @@
 #!/bin/bash
 
 rm $1/res.txt
-for net in {1..10}
+for net in {1..7}
 do
 	echo Evaluating network net${net}...
 	for spec in `ls $1/net${net}/`

code/verifier.py

@@ -8,6 +8,8 @@
 from resnet import ResNet, BasicBlock
 from deeppoly import DeepPoly, DPReLU, DPLinear, DPConv, DPBasicBlock
 
+import time
+
 
 DEVICE = 'cpu'
 DTYPE = torch.float32
@@ -304,6 +306,7 @@ def analyze(net, inputs, eps, true_label):
 
 
 def main():
+    start = time.perf_counter_ns()
     parser = argparse.ArgumentParser(description='Neural network verification using DeepPoly relaxation')
     parser.add_argument('--net', type=str, required=True, help='Neural network architecture to be verified.')
     parser.add_argument('--spec', type=str, required=True, help='Test case to verify.')
@@ -315,6 +318,7 @@ def main():
     inputs, true_label, eps = get_spec(args.spec, dataset)
     net = get_net(args.net, net_name)
     # print([module for module in net.modules()])
+    '''
     if type(net) == NormalizedResnet:
         print(net)
     for module in net.modules():
@@ -330,16 +334,19 @@ def main():
                                 if type(mo) == nn.Sequential:
                                     print(">>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>")
                                     print(mo)
+    '''
 
     outs = net(inputs)
     pred_label = outs.max(dim=1)[1].item()
     assert pred_label == true_label
-    '''
+    # '''
     if analyze(net, inputs, eps, true_label):
-        print('verified')
+        end = time.perf_counter_ns()
+        print('', round(end - start)*0.000000001, 's, verified')
     else:
-        print('not verified')
-    '''
+        end = time.perf_counter_ns()
+        print('', round(end - start)*0.000000001, 's, not verified')
+    # '''
 
 if __name__ == '__main__':
     main()