diff --git a/olmocr/data/repackage_olmocrmix.py b/olmocr/data/repackage_olmocrmix.py
index 86ebadb..512435d 100644
--- a/olmocr/data/repackage_olmocrmix.py
+++ b/olmocr/data/repackage_olmocrmix.py
@@ -54,7 +54,7 @@ def infer_doc_id(md_path: Path, processed_root: Path) -> str:
     rel = md_path.relative_to(processed_root)
     if len(rel.parts) < 2:
         stem = rel.stem
-        prefix = rel.stem[:4]
+        prefix = rel.stem
     else:
         prefix = rel.parts[0]
         stem = Path(rel.parts[-1]).stem
diff --git a/tests/sample_dataset/0000/blanktext.md b/tests/sample_dataset/empty_document/blanktext.md
similarity index 100%
rename from tests/sample_dataset/0000/blanktext.md
rename to tests/sample_dataset/empty_document/blanktext.md
diff --git a/tests/sample_dataset/0000/blanktext.pdf b/tests/sample_dataset/empty_document/blanktext.pdf
similarity index 100%
rename from tests/sample_dataset/0000/blanktext.pdf
rename to tests/sample_dataset/empty_document/blanktext.pdf
diff --git a/tests/sample_dataset/0000/edgar.md b/tests/sample_dataset/simple_document/edgar.md
similarity index 100%
rename from tests/sample_dataset/0000/edgar.md
rename to tests/sample_dataset/simple_document/edgar.md
diff --git a/tests/sample_dataset/0000/edgar.pdf b/tests/sample_dataset/simple_document/edgar.pdf
similarity index 100%
rename from tests/sample_dataset/0000/edgar.pdf
rename to tests/sample_dataset/simple_document/edgar.pdf