fix issue 5

examples/multimedia/avmnist_MFM.py

@@ -27,4 +27,4 @@
 objective=MFM_objective(2.0,[sigmloss1dcentercrop(28,34),sigmloss1dcentercrop(112,130)],[1.0,1.0])
 train(encoders,fuse,head,traindata,validdata,25,decoders+intermediates,objective=objective,objective_args_dict={'decoders':decoders,'intermediates':intermediates})
 model=torch.load('best.pt')
-test(model,testdata)
+test(model,testdata,no_robust=True)

examples/multimedia/avmnist_MVAE_mixed.py

@@ -25,4 +25,4 @@
 elbo=MVAE_objective(2.0,[sigmloss1dcentercrop(28,34),sigmloss1dcentercrop(112,130)],[1.0,1.0],annealing=0.0)
 train(encoders,fuse,head,traindata,validdata,20,decoders,objective=elbo,objective_args_dict={'decoders':decoders})
 mvae = torch.load('best.pt')
-test(mvae,head,testdata)
+test(mvae,testdata,no_robust=True)

examples/multimedia/avmnist_architecture_search_test.py

@@ -11,4 +11,4 @@
 
 traindata, validdata, testdata = get_dataloader('/data/yiwei/avmnist/_MFAS/avmnist',batch_size=32)
 model = torch.load('temp/best.pt').cuda()
-test(model,testdata)
+test(model,testdata,no_robust=True)

examples/multimedia/avmnist_cca.py

@@ -27,4 +27,4 @@
 
 print("Testing:")
 model=torch.load('best_cca.pt').cuda()
-test(model,testdata)
+test(model,testdata,no_robust=True)

examples/multimedia/avmnist_gradient_blend.py

@@ -21,6 +21,6 @@
 
 print("Testing:")
 model=torch.load(filename).cuda()
-test(model,testdata)
+test(model,testdata,no_robust=True)
 
 

examples/multimedia/avmnist_low_rank_tensor.py

@@ -19,6 +19,6 @@
 
 print("Testing:")
 model=torch.load(filename).cuda()
-test(model,testdata)
+test(model,testdata,no_robust=True)
 
 

examples/multimedia/avmnist_multi_interac_matrix.py

@@ -21,6 +21,6 @@
 
 print("Testing:")
 model=torch.load(filename).cuda()
-test(model,testdata)
+test(model,testdata,no_robust=True)
 
 

examples/multimedia/avmnist_refnet.py

@@ -20,6 +20,6 @@
 
 print("Testing:")
 model=torch.load('best.pt').cuda()
-test(model,testdata)
+test(model,testdata,no_robust=True)
 
 

examples/multimedia/avmnist_simple_late_fusion.py

@@ -8,17 +8,17 @@
 from torch import nn
 import torch
 
-traindata, validdata, testdata = get_dataloader('/home/pliang/yiwei/avmnist/_MFAS/avmnist')
+traindata, validdata, testdata = get_dataloader('/home/paul/yiwei/avmnist/_MFAS/avmnist')
 channels=6
 encoders=[LeNet(1,channels,3).cuda(),LeNet(1,channels,5).cuda()]
 head=MLP(channels*40,100,10).cuda()
 
 fusion=Concat().cuda()
 
-train(encoders,fusion,head,traindata,validdata,30,optimtype=torch.optim.SGD,lr=0.1,weight_decay=0.0001)
+train(encoders,fusion,head,traindata,validdata,3,optimtype=torch.optim.SGD,lr=0.1,weight_decay=0.0001)
 
 print("Testing:")
 model=torch.load('best.pt').cuda()
-test(model,testdata)
+test(model,testdata,no_robust=True)
 
 

examples/multimedia/avmnist_tensor_matrix.py

@@ -20,6 +20,6 @@
 
 print("Testing:")
 model=torch.load('best.pt').cuda()
-test(model,testdata)
+test(model,testdata,no_robust=True)
 
 

examples/multimedia/avmnist_unimodal_0.py

@@ -20,6 +20,6 @@
 print("Testing:")
 encoder=torch.load('encoder.pt').cuda()
 head = torch.load('head.pt')
-test(encoder,head,testdata,modalnum=modalnum)
+test(encoder,head,testdata,modalnum=modalnum,no_robust=True)
 
 

examples/multimedia/avmnist_unimodal_1.py

@@ -20,6 +20,6 @@
 print("Testing:")
 encoder=torch.load('encoder.pt').cuda()
 head = torch.load('head.pt')
-test(encoder,head,testdata,modalnum=modalnum)
+test(encoder,head,testdata,modalnum=modalnum,no_robust=True)
 
 

training_structures/Supervised_Learning.py

@@ -292,7 +292,12 @@ def processinput(inp):
 # criterion: only needed for regression, put MSELoss there
 # all other arguments are same as train
 def test(
- model, test_dataloaders_all, dataset, method_name='My method', is_packed=False, criterion=nn.CrossEntropyLoss(), task="classification", auprc=False, input_to_float=True):
+ model, test_dataloaders_all, dataset='default', method_name='My method', is_packed=False, criterion=nn.CrossEntropyLoss(), task="classification", auprc=False, input_to_float=True, no_robust=False):
+ if no_robust:
+ def testprocess():
+ single_test(model,test_dataloaders_all,is_packed,criterion,task,auprc,input_to_float)
+ all_in_one_test(testprocess,[model])
+ return
  def testprocess():
  single_test(model, test_dataloaders_all[list(test_dataloaders_all.keys())[0]][0], is_packed, criterion, task, auprc, input_to_float)
  all_in_one_test(testprocess, [model])

training_structures/architecture_search.py

@@ -180,7 +180,12 @@ def single_test(model,test_dataloader,auprc=False):
  return {'Accuracy': float(corrects)/total}
 
 
-def test(model, test_dataloaders_all, dataset, method_name='My method', auprc=False):
+def test(model, test_dataloaders_all, dataset, method_name='My method', auprc=False, no_robust=False):
+ if no_robust:
+ def testprocess():
+ single_test(model,test_dataloaders_all, auprc)
+ all_in_one_test(testprocess,[model])
+ return
  def testprocess():
  single_test(model, test_dataloaders_all[list(test_dataloaders_all.keys())[0]][0], auprc)
  all_in_one_test(testprocess, [model])

training_structures/gradient_blend.py

@@ -348,7 +348,12 @@ def single_test(model, test_dataloader, auprc=False, classification=True):
  return {'MSE': (totalloss/total).item()}
 
 
-def test(model, test_dataloaders_all, dataset, method_name='My method', auprc=False, classification=True):
+def test(model, test_dataloaders_all, dataset, method_name='My method', auprc=False, classification=True,no_robust=False):
+ if no_robust:
+ def testprocess():
+ single_test(model,test_dataloaders_all,auprc,classification)
+ all_in_one_test(testprocess,[model])
+ return
  def testprocess():
  single_test(model, test_dataloaders_all[list(test_dataloaders_all.keys())[0]][0], auprc, classification)
  all_in_one_test(testprocess, [model])

training_structures/unimodal.py

@@ -154,7 +154,12 @@ def single_test(encoder, head, test_dataloader, auprc=False, modalnum=0, task='c
  return {'MSE': (totalloss / totals).item()}
 
 
-def test(encoder, head, test_dataloaders_all, dataset, method_name='My method', auprc=False, modalnum=0, task='classification', criterion=None):
+def test(encoder, head, test_dataloaders_all, dataset='default', method_name='My method', auprc=False, modalnum=0, task='classification', criterion=None,no_robust=False):
+ if no_robust:
+ def testprocess():
+ single_test(encoder,head,test_dataloaders_all,auprc,modalnum,task,criterion)
+ all_in_one_test(testprocess,[model])
+ return
  def testprocess():
  single_test(encoder, head, test_dataloaders_all[list(test_dataloaders_all.keys())[0]][0], auprc, modalnum, task, criterion)
  all_in_one_test(testprocess, [encoder, head])