symmetrize arxiv_2023

core/data_utils/load_data_lp.py

@@ -12,6 +12,10 @@
 from torch_geometric.datasets import Planetoid
 from torch_geometric.data import Data, InMemoryDataset, Dataset
 from torch_geometric.transforms import RandomLinkSplit
+from torch_geometric.utils import to_undirected 
+import warnings
+warnings.filterwarnings("ignore", category=UserWarning)
+
 from ogb.nodeproppred import PygNodePropPredDataset
 from sklearn.preprocessing import normalize
 from yacs.config import CfgNode as CN
@@ -28,6 +32,7 @@
 from graphgps.utility.utils import time_logger
 from typing import Dict, Tuple, List, Union
 
+
 FILE = 'core/dataset/ogbn_products_orig/ogbn-products.csv'
 FILE_PATH = get_git_repo_root_path() + '/'
 
@@ -37,8 +42,10 @@ def load_taglp_arxiv2023(cfg: CN) -> Tuple[Dict[str, Data], List[str]]:
     # add one default argument
 
     data, text = load_tag_arxiv23()
-    undirected = data.is_directed()
-
+    if data.is_directed() is True:
+        data.edge_index = to_undirected(data.edge_index)
+        undirected = True
+
     splits = get_edge_split(data,
                             undirected,
                             cfg.device,
@@ -56,7 +63,7 @@ def load_taglp_cora(cfg: CN) -> Tuple[Dict[str, Data], List[str]]:
     data, data_citeid = load_graph_cora(False)
     text = load_text_cora(data_citeid)
     # text = None
-    undirected = data.is_directed()
+    undirected = data.is_undirected()
 
     splits = get_edge_split(data,
                             undirected,
@@ -74,7 +81,7 @@ def load_taglp_ogbn_arxiv(cfg: CN) -> Tuple[Dict[str, Data], List[str]]:
 
     data = load_graph_ogbn_arxiv(False)
     text = load_text_ogbn_arxiv()
-    undirected = data.is_directed()
+    undirected = data.is_undirected()
 
     cfg = config_device(cfg)
 
@@ -115,7 +122,7 @@ def load_taglp_product(cfg: CN) -> Tuple[Dict[str, Data], List[str]]:
     # add one default argument
 
     data, text = load_tag_product()
-    undirected = data.is_directed()
+    undirected = data.is_undirected()
 
     cfg = config_device(cfg)
 
@@ -135,7 +142,7 @@ def load_taglp_pubmed(cfg: CN) -> Tuple[Dict[str, Data], List[str]]:
 
     data = load_graph_pubmed(False)
     text = load_text_pubmed()
-    undirected = data.is_directed()
+    undirected = data.is_undirected()
 
     splits = get_edge_split(data,
                             undirected,
@@ -152,7 +159,7 @@ def load_taglp_citeseer(cfg: CN) -> Tuple[Dict[str, Data], List[str]]:
 
     data = load_graph_citeseer()
     text = load_text_citeseer()
-    undirected = data.is_directed()
+    undirected = data.is_undirected()
 
     splits = get_edge_split(data,
                             undirected,
@@ -166,11 +173,15 @@ def load_taglp_citeseer(cfg: CN) -> Tuple[Dict[str, Data], List[str]]:
 
 def load_taglp_citationv8(cfg: CN) -> Tuple[Dict[str, Data], List[str]]:
     # add one default argument
-
+    
     data = load_graph_citationv8()
     text = load_text_citationv8()
-    undirected = data.is_directed()
-
+    if data.is_directed() is True:
+        data.edge_index  = to_undirected(data.edge_index)
+        undirected  = True 
+    else:
+        undirected = data.is_undirected()
+
     splits = get_edge_split(data,
                             undirected,
                             cfg.device,
@@ -185,32 +196,44 @@ def load_taglp_citationv8(cfg: CN) -> Tuple[Dict[str, Data], List[str]]:
 # TEST CODE
 if __name__ == '__main__':
     args = init_cfg_test()
-    print(args)
-    '''data, text, __  = load_taglp_arxiv2023(args.data)
+    print('arxiv2023')
+    splits, text, data  = load_taglp_arxiv2023(args.data)
+    print(f'directed: {data.is_directed()}')
     print(data)
     print(type(text))
-    data, text = load_taglp_cora(args.data)
+
+    print('citationv8')
+    splits, text, data = load_taglp_citationv8(args.data)
+    print(f'directed: {data.is_directed()}')
     print(data)
     print(type(text))
 
-    
-
-    data, text = load_taglp_product(args.data)
+    exit(-1)
+    print('cora')
+    splits, text, data = load_taglp_cora(args.data)
+    print(f'directed: {data.is_directed()}')
     print(data)
     print(type(text))
 
-    data, text = load_taglp_pubmed(args.data)
+    print('product')
+    splits, text, data = load_taglp_product(args.data)
+    print(f'directed: {data.is_directed()}')
     print(data)
-    print(type(text))'''
+    print(type(text))
 
-    splits, text, data = load_taglp_citeseer(args.data)
+    print('pubmed')
+    splits, text, data = load_taglp_pubmed(args.data)
+    print(f'directed: {data.is_directed()}')
     print(data)
     print(type(text))
 
-    splits, text, data = load_taglp_citationv8(args.data)
+    splits, text, data = load_taglp_citeseer(args.data)
+    print(f'directed: {data.is_directed()}')
     print(data)
     print(type(text))
-
+
+    print(args.data)
     splits, text, data = load_taglp_ogbn_arxiv(args.data)
+    print(f'directed: {data.is_directed()}')
     print(data)
-    print(type(text))
+    print(type(text))

core/model_finetuning/mlp.py

@@ -155,16 +155,17 @@ def project_main():
                 start = time.time()
                 clf.partial_fit(train_dataset, train_labels, classes=classes)
                 print(f'this epoch costs {time.time() - start}')
-
-        # Calculate and print metrics for test set
-        test_metrics = get_metrics(clf, test_dataset, test_labels, evaluator_hit, evaluator_mrr)
-        print(test_metrics)
-        # Calculate and print metrics for train set
-        train_metrics = get_metrics(clf, train_dataset, train_labels, evaluator_hit, evaluator_mrr)
-        print(train_metrics)
-        # Calculate and print metrics for validation set
-        val_metrics = get_metrics(clf, val_dataset, val_labels, evaluator_hit, evaluator_mrr)
-        print(val_metrics)
+
+                if i % 100 == 0:
+                    # Calculate and print metrics for test set
+                    test_metrics = get_metrics(clf, test_dataset, test_labels, evaluator_hit, evaluator_mrr)
+                    print(test_metrics)
+                    # Calculate and print metrics for train set
+                    train_metrics = get_metrics(clf, train_dataset, train_labels, evaluator_hit, evaluator_mrr)
+                    print(train_metrics)
+                    # Calculate and print metrics for validation set
+                    val_metrics = get_metrics(clf, val_dataset, val_labels, evaluator_hit, evaluator_mrr)
+                    print(val_metrics)
 
         results_rank = {
             key: (train_metrics[key], val_metrics[key], test_metrics[key])

core/model_finetuning/scripts/pubmed.sh

@@ -28,7 +28,7 @@ module load compiler/gnu/12
 cd /hkfs/work/workspace/scratch/cc7738-benchmark_tag/TAPE_chen/core/model_finetuning
 
 
-for iter in 1000 2000; do
+for iter in 1500; do
     echo "python mlp.py --data pubmed --decoder MLP --max_iter $iter"
     python mlp.py --data pubmed --decoder MLP --max_iter $iter
 done