diff --git a/.github/workflows/export-parquet.yml b/.github/workflows/export-parquet.yml
new file mode 100644
index 0000000..6f659a6
--- /dev/null
+++ b/.github/workflows/export-parquet.yml
@@ -0,0 +1,23 @@
+name: export-parquet
+on:
+  workflow_dispatch: {}
+jobs:
+  export:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+      - name: Install deps
+        run: |
+          python -m pip install --upgrade pip
+          pip install pandas pyarrow
+      - name: Export parquet to JSONL
+        run: |
+          python scripts/export_parquet.py
+      - name: Upload artifact
+        uses: actions/upload-artifact@v4
+        with:
+          name: rows-jsonl
+          path: rows.jsonl
diff --git a/scripts/export_parquet.py b/scripts/export_parquet.py
new file mode 100644
index 0000000..691d68e
--- /dev/null
+++ b/scripts/export_parquet.py
@@ -0,0 +1,11 @@
+import pandas as pd
+from pathlib import Path
+p = Path('SWE-bench_Verified/test-00000-of-00001.parquet')
+df = pd.read_parquet(p, columns=['repo','instance_id','environment_setup_commit'])
+with open('rows.jsonl','w') as f:
+    for _, row in df.iterrows():
+        f.write({
+            'repo': row['repo'],
+            'instance_id': row['instance_id'],
+            'environment_setup_commit': row['environment_setup_commit']
+        }.__repr__().replace(", ') + n)