Fix errors and refactoring

dongkwan-kim · Jun 29, 2024 · eddd5d5 · eddd5d5
1 parent 9cae252
commit eddd5d5
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 20 deletions.
diff --git a/WL4S/wl4s.py b/WL4S/wl4s.py
@@ -25,7 +25,7 @@
 ModelType = Union[MultiOutputClassifier, LinearSVC, SVC]
 
 DATASETS_REAL = ["PPIBP", "EMUser", "HPOMetab", "HPONeuro"]
-DATASETS_SYN = ["Component", "Density", "Coreness", "CutRatio"]
+DATASETS_SYN = ["CutRatio", "Density", "Coreness", "Component"]
 MODEL_KWARGS_KEY = ["C", "kernel", "dual"]
 
 parser = argparse.ArgumentParser()
@@ -47,9 +47,10 @@
 
 
 class WL4S(torch.nn.Module):
-    def __init__(self, stype, num_layers, norm,
+    def __init__(self, dataset_name, stype, num_layers, norm,
                  dtype="histogram", splits=None, k_to_sample=None, precompute=False):
         super(WL4S, self).__init__()
+        self.dataset_name = dataset_name
         self.stype = stype
         self.norm = norm
         self.dtype = dtype
@@ -77,7 +78,7 @@ def forward(self, x, edge_index, batch_or_sub_batch, x_to_xs=None, mask=None):
             if self.dtype == "kernel":
                 kernel_key = kk(self.k_to_sample, self.stype, self.norm, i)
                 if self.splits[-1] <= 250:  # synthetic graphs, backward compatibility
-                    kernel_key = kk(self.k_to_sample, self.stype[:3], self.norm, i, edge_index.size(1) // 100)
+                    kernel_key = kk(self.k_to_sample, self.stype[:3], self.norm, i, self.dataset_name)
 
                 h = hist_linear_kernels(hist=h, splits=self.splits, key=kernel_key)
                 if self.precompute:
@@ -89,7 +90,7 @@ def forward(self, x, edge_index, batch_or_sub_batch, x_to_xs=None, mask=None):
 
 
 def kk(k_to_sample, stype, norm, i, *args):
-    return "_".join([str(s) for s in [k_to_sample, stype, norm, i, *args]])
+    return "_".join([str(s) for s in [*args, k_to_sample, stype, norm, i]])
 
 
 @fscaches(path="../_caches", keys_to_exclude=["hist"], verbose=True)
@@ -157,7 +158,7 @@ def get_data_and_model(args, precompute=False):
         f"{splits} != [{len(train_dts), len(val_dts), len(test_dts)}]"
     all_data = Batch.from_data_list(train_dts + val_dts + test_dts)
 
-    wl = WL4S(stype=args.stype, num_layers=args.wl_layers, norm=args.hist_norm,
+    wl = WL4S(dataset_name=args.dataset_name, stype=args.stype, num_layers=args.wl_layers, norm=args.hist_norm,
               dtype=args.dtype, splits=splits, k_to_sample=args.k_to_sample, precompute=precompute)
 
     if args.stype == "connected":

diff --git a/WL4S/wl4s2v2.py b/WL4S/wl4s2v2.py
@@ -2,6 +2,8 @@
 
 
 def get_data_mixed_kernels(args):
+    assert args.dtype == "kernel"
+
     args.stype = "separated"
     k_list_s, splits_s, y_s = get_data_and_model(args)
 
@@ -45,8 +47,8 @@ def get_data_mixed_kernels(args):
         run_one(__args__)
     else:
         for _a_c, _a_s in [
-            (0.99, 0.01), (0.9, 0.1), (0.5, 0.1),
-            (0.01, 0.99), (0.1, 0.9), (0.1, 0.5),
+            (0.999, 0.001), (0.99, 0.01), (0.9, 0.1),
+            (0.001, 0.999), (0.01, 0.99), (0.1, 0.9),
         ]:
             __args__.a_c, __args__.a_s = _a_c, _a_s
 

diff --git a/WL4S/wl4s_k.py b/WL4S/wl4s_k.py
@@ -7,6 +7,9 @@
         "stype": ["separated"],
         "wl_cumcat": [False, True],
         "hist_norm": [False, True],
+        "model": ["SVC"],
+        "kernel": ["precomputed"],
+        "dtype": ["kernel"],
     }
     Cx100 = [8, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096, 8192, 16384, 32768, 65536, 131072, 262144, 524288, 1048576]
     MORE_HPARAM_SPACE = {
@@ -18,26 +21,19 @@
 
     MODE = __args__.MODE
     if MODE == "syn_k":
+        __args__.stype = "connected"
+        HPARAM_SPACE["stype"] = ["connected"]
+        kws = dict(file_dir="../_logs_wl4s_k", log_postfix=f"_inf")
+        hp_search_syn(__args__, HPARAM_SPACE, MORE_HPARAM_SPACE, **kws)
+
         __args__.stype = "separated"
-        HPARAM_SPACE = {**HPARAM_SPACE, "model": ["LinearSVC"]}
-        MORE_HPARAM_SPACE = {**MORE_HPARAM_SPACE, "dual": [True, False]}
+        HPARAM_SPACE["stype"] = ["separated"]
         for k_to_sample in [None, 1, 2]:
             __args__.k_to_sample = k_to_sample
             kws = dict(file_dir="../_logs_wl4s_k", log_postfix=f"_{k_to_sample or 0}")
             hp_search_syn(__args__, HPARAM_SPACE, MORE_HPARAM_SPACE, **kws)
 
-        __args__.stype = "connected"
-        HPARAM_SPACE["stype"] = ["connected"]
-        kws = dict(file_dir="../_logs_wl4s_k", log_postfix=f"_inf")
-        hp_search_syn(__args__, HPARAM_SPACE, MORE_HPARAM_SPACE, **kws)
-
     else:
-        HPARAM_SPACE = {
-            **HPARAM_SPACE,
-            "model": ["SVC"], "kernel": ["precomputed"], "dtype": ["kernel"],
-        }
-        __args__.dtype = "kernel"
-
         if MODE == "real_precomputation":
             for k_to_sample in [None, 1, 2]:
                 for dataset_name in ["PPIBP", "EMUser"]:
@@ -62,6 +58,7 @@
 
         elif MODE == "real_k":
             for dataset_name in ["PPIBP", "EMUser", "HPOMetab", "HPONeuro"]:
+                # OOM for HPOMetab & HPONeuro with k >= 1
                 k_to_sample_list = [None] if dataset_name in ["HPOMetab", "HPONeuro"] else [None, 1, 2]
                 for k_to_sample in k_to_sample_list:
                     __args__.k_to_sample = k_to_sample