Merge pull request #8 from iamgroot42/michael/default_load_hf

bug fix for loading neighbors
iamgroot42 · Feb 15, 2024 · 18f9085 · 18f9085
2 parents 4789787 + 19c1535
commit 18f9085
Show file tree

Hide file tree

Showing 4 changed files with 18 additions and 18 deletions.
diff --git a/mimir/config.py b/mimir/config.py
@@ -146,7 +146,7 @@ class ExperimentConfig(Serializable):
     "Dump data to cache? Exits program after dumping"
     load_from_cache: Optional[bool] = False
     """Load data from cache?"""
-    load_from_hf: Optional[bool] = False
+    load_from_hf: Optional[bool] = True
     """Load data from HuggingFace?"""
     blackbox_attacks: Optional[List[str]] = field(
         default_factory=lambda: None

diff --git a/mimir/custom_datasets.py b/mimir/custom_datasets.py
@@ -12,6 +12,20 @@
 
 DATASETS = ['writing', 'english', 'german', 'pubmed']
 
+SOURCES_UPLOADED = [
+    "arxiv",
+    "dm_mathematics",
+    "github",
+    "hackernews",
+    "pile_cc",
+    "pubmed_central",
+    "wikipedia_(en)",
+    "full_pile",
+    "c4",
+    "temporal_arxiv",
+    "temporal_wiki"
+]
+
 
 def load_pubmed(cache_dir):
     data = datasets.load_dataset('pubmed_qa', 'pqa_labeled', split='train', cache_dir=cache_dir)
@@ -41,20 +55,6 @@ def load_cached(cache_dir,
         if not filename.startswith("the_pile"):
             raise ValueError(f"HuggingFace data only available for The Pile.")
 
-        SOURCES_UPLOADED = [
-            "arxiv",
-            "dm_mathematics",
-            "github",
-            "hackernews",
-            "pile_cc",
-            "pubmed_central",
-            "wikipedia_(en)",
-            "full_pile",
-            "c4",
-            "temporal_arxiv",
-            "temporal_wiki"
-        ]
-
         for source in SOURCES_UPLOADED:
             # Got a match
             if source in filename and filename.startswith(f"the_pile_{source}"):

diff --git a/python_scripts/mimir.py b/python_scripts/mimir.py
@@ -189,6 +189,6 @@ def _generate_examples(self, file_path_dict):
                 yield id, {
                     "member": json.loads(member),
                     "nonmember": json.loads(nonmember),
-                    "member_neighbors": json.loads(member_neighbors)[0],
-                    "nonmember_neighbors": json.loads(nonmember_neighbors)[0],
+                    "member_neighbors": json.loads(member_neighbors),
+                    "nonmember_neighbors": json.loads(nonmember_neighbors),
                 }
diff --git a/run.py b/run.py
@@ -190,7 +190,7 @@ def get_mia_scores(
                                         else None
                                     ),
                                     loss=loss,
-                                    batch_siz=4,
+                                    batch_size=4,
                                     substr_neighbors=substr_neighbors,
                                 )