1 files changed, 22 insertions, 0 deletions
diff --git a/scripts/debug_personamem_hash.py b/scripts/debug_personamem_hash.py
new file mode 100644
index 0000000..7ef442d
--- /dev/null
+++ b/scripts/debug_personamem_hash.py
@@ -0,0 +1,22 @@
+import hashlib
+import json
+
+def get_line_hash(line_str: str) -> str:
+    """Compute SHA256 hash of the line content to match shared_context_id."""
+    return hashlib.sha256(line_str.strip().encode("utf-8")).hexdigest()
+
+def debug_hash():
+    jsonl_path = "data/raw_datasets/personamem/shared_contexts_32k.jsonl"
+    with open(jsonl_path, "r") as f:
+        first_line = f.readline()
+        
+    computed_hash = get_line_hash(first_line)
+    target_hash = "e898d03fec683b1cabf29f57287ff66f8a31842543ecef44b56766844c1c1301"
+    
+    print(f"Computed: {computed_hash}")
+    print(f"Target:   {target_hash}")
+    print(f"Match:    {computed_hash == target_hash}")
+
+if __name__ == "__main__":
+    debug_hash()
+