Update data gen

bpp/README.md

@@ -1 +1 @@
-Run utils.py to generate test datasets.
+Run `python utils.py` to generate test datasets.

cvrp/README.md

@@ -1 +1 @@
-Run utils.py to generate test datasets.
+Run `python utils.py` to generate test datasets.

cvrp/utils.py

@@ -41,7 +41,7 @@ def load_test_dataset(problem_size, device):
 
 if __name__ == '__main__':
     import pathlib
-    pathlib.Path('./data/cvrp').mkdir(parents=False, exist_ok=True) 
+    pathlib.Path('../data/cvrp').mkdir(parents=False, exist_ok=True) 
     torch.manual_seed(123456)
     for n in [20, 100, 500]:
         inst_list = []
@@ -50,5 +50,5 @@ def load_test_dataset(problem_size, device):
             inst = torch.cat((demands.unsqueeze(0), distances), dim=0) # (n+2, n+1)
             inst_list.append(inst)
         testDataset = torch.stack(inst_list)
-        torch.save(testDataset, f'./data/cvrp/testDataset-{n}.pt')
+        torch.save(testDataset, f'../data/cvrp/testDataset-{n}.pt')
 

mkp/README.md

@@ -1 +1 @@
-Run utils.py to generate datasets.
+Run `python utils.py` to generate datasets.

mkp/utils.py

@@ -49,7 +49,9 @@ def load_test_dataset(problem_size, device):
     return val_list
 
 if __name__ == '__main__':
-    # generate test dataset
+    # generate val and test dataset
+    import pathlib
+    pathlib.Path('../data/mkp').mkdir(parents=False, exist_ok=True) 
     M = 5
     torch.manual_seed(12345)
     for problem_size in [50]:
@@ -58,7 +60,7 @@ def load_test_dataset(problem_size, device):
             prize, weight = gen_instance(problem_size, M, 'cpu')
             valDataset.append(torch.cat((prize.unsqueeze(1), weight), dim=1))
         valDataset = torch.stack(valDataset)
-        torch.save(valDataset, f'./data/mkp/valDataset-{problem_size}.pt')
+        torch.save(valDataset, f'../data/mkp/valDataset-{problem_size}.pt')
 
     torch.manual_seed(123456)
     for problem_size in [50]:
@@ -67,7 +69,7 @@ def load_test_dataset(problem_size, device):
             prize, weight = gen_instance(problem_size, M, 'cpu')
             testDataset.append(torch.cat((prize.unsqueeze(1), weight), dim=1))
         testDataset = torch.stack(testDataset)
-        torch.save(testDataset, f'./data/mkp/testDataset-{problem_size}.pt')
+        torch.save(testDataset, f'../data/mkp/testDataset-{problem_size}.pt')
 
 
 

mkp_transformer/README.md

@@ -1 +1 @@
-Run utils.py to generate datasets.
+Run `python utils.py` to generate datasets.

mkp_transformer/utils.py

@@ -46,7 +46,7 @@ def load_test_dataset(problem_size, device):
 if __name__ == '__main__':
     # generate val and test dataset
     import pathlib
-    pathlib.Path('./data/mkp_transformer').mkdir(parents=False, exist_ok=True) 
+    pathlib.Path('../data/mkp_transformer').mkdir(parents=False, exist_ok=True) 
 
     torch.manual_seed(12345)
     for problem_size in [300, 500]:
@@ -55,7 +55,7 @@ def load_test_dataset(problem_size, device):
             price, weight = gen_instance(problem_size, 5)
             testDataset.append(torch.cat((price.unsqueeze(0), weight), dim=0))
         testDataset = torch.stack(testDataset)
-        torch.save(testDataset, f'./data/mkp_transformer/valDataset-{problem_size}.pt')
+        torch.save(testDataset, f'../data/mkp_transformer/valDataset-{problem_size}.pt')
 
     torch.manual_seed(123456)
     for problem_size in [300, 500]:
@@ -64,7 +64,7 @@ def load_test_dataset(problem_size, device):
             price, weight = gen_instance(problem_size, 5)
             testDataset.append(torch.cat((price.unsqueeze(0), weight), dim=0))
         testDataset = torch.stack(testDataset)
-        torch.save(testDataset, f'./data/mkp_transformer/testDataset-{problem_size}.pt')
+        torch.save(testDataset, f'../data/mkp_transformer/testDataset-{problem_size}.pt')
 
 
 

op/README.md

@@ -0,0 +1 @@
+Run `python utils.py` to generate datasets.

op/utils.py

@@ -71,11 +71,13 @@ def load_test_dataset(n_node, k_sparse, device):
     return val_list
 
 if __name__ == "__main__":
+    import pathlib
+    pathlib.Path('../data/op').mkdir(parents=False, exist_ok=True) 
     torch.manual_seed(12345)
     for problem_size in [100, 200, 300]:
         coor = torch.rand(size=(30, problem_size, 2))
-        torch.save(coor, f"data/op/valDataset-{problem_size}.pt")
+        torch.save(coor, f"../data/op/valDataset-{problem_size}.pt")
     torch.manual_seed(123456)
     for problem_size in [100, 200, 300]:
         coor = torch.rand(size=(100, problem_size, 2))
-        torch.save(coor, f"data/op/testDataset-{problem_size}.pt")
+        torch.save(coor, f"../data/op/testDataset-{problem_size}.pt")

pctsp/README.md

@@ -1 +1 @@
-Run utils.py to generate test datasets.
+Run `python utils.py` to generate test datasets.

pctsp/utils.py

@@ -50,10 +50,10 @@ def load_test_dataset(n_node, device):
 if __name__ == "__main__":
     torch.manual_seed(123456)
     import pathlib
-    pathlib.Path('./data/pctsp').mkdir(parents=False, exist_ok=True) 
+    pathlib.Path('../data/pctsp').mkdir(parents=False, exist_ok=True) 
     for n in [20, 100, 500]:
         testDataset = []
         for _ in range(100):
             dist_mat, prizes, penalties = gen_inst(n, 'cpu')
             testDataset.append(torch.cat([dist_mat, prizes.unsqueeze(0), penalties.unsqueeze(0)], dim=0))
-        torch.save(torch.stack(testDataset), f"data/pctsp/testDataset-{n}.pt")
+        torch.save(torch.stack(testDataset), f"../data/pctsp/testDataset-{n}.pt")

smtwtp/README.md

@@ -1 +1 @@
-Run the utils.py to generate the test datasets.
+Run `python utils.py` to generate the test datasets.

smtwtp/utils.py

@@ -32,13 +32,13 @@ def load_test_dataset(n_node, device):
 if __name__ == '__main__':
     torch.manual_seed(123456)
     import pathlib
-    pathlib.Path('./data/smtwtp').mkdir(parents=False, exist_ok=True) 
+    pathlib.Path('../data/smtwtp').mkdir(parents=False, exist_ok=True) 
     problem_sizes = [50, 100, 500]
     dataset_size = 100
     for p_size in problem_sizes:
         dataset = []
         for _ in range(dataset_size):        
             pyg_data, due_time, weights, processing_time = instance_gen(p_size, 'cpu')
             dataset.append([pyg_data, due_time, weights, processing_time])
-        with open(f"data/smtwtp/test{p_size}.pkl", "wb") as f:
+        with open(f"../data/smtwtp/test{p_size}.pkl", "wb") as f:
             pickle.dump(dataset, f)

sop/README.md

@@ -0,0 +1 @@
+Run `python utils.py` to generate the test datasets.

sop/utils.py

@@ -66,6 +66,8 @@ def load_test_dataset(n_node, device):
     return loaded_list
 
 if __name__ == "__main__":
+    import pathlib
+    pathlib.Path('../data/sop').mkdir(parents=False, exist_ok=True) 
     torch.manual_seed(123456)
     problem_sizes = [20, 50, 100]
     dataset_size = 100
@@ -74,7 +76,7 @@ def load_test_dataset(n_node, device):
         for _ in range(dataset_size):        
             distances, adj_mat, mask = training_instance_gen(p_size, 'cpu')
             dataset.append([distances, adj_mat, mask])
-        with open(f"data/sop/test{p_size}.pkl", "wb") as f:
+        with open(f"../data/sop/test{p_size}.pkl", "wb") as f:
             pickle.dump(dataset, f)