only hydrate the lfs files needed for ci

Author: thesofakillers

.github/workflows/paperbench_tests.yml

@@ -28,13 +28,13 @@ jobs:
     steps:
     - uses: actions/checkout@v4
       with:
-        lfs: true
+        lfs: false
         fetch-depth: 1
 
     - name: Hydrate PaperBench data
       run: |
-        git lfs fetch --include "project/paperbench/data/**" --exclude ""
-        git lfs checkout project/paperbench/data
+        git lfs fetch --include "project/paperbench/data/papers/lca-on-the-line/**" --exclude ""
+        git lfs checkout project/paperbench/data/papers/lca-on-the-line
 
     - name: Install uv
       uses: astral-sh/setup-uv@v5
@@ -48,11 +48,8 @@ jobs:
 
     - name: Set up Docker
       uses: docker/setup-buildx-action@v3
-
-    - uses: docker/setup-buildx-action@v3
       with:
         driver: docker
-        install: true
 
     - name: Build pb-env
       working-directory: ./project/paperbench

project/paperbench/paperbench/nano/eval.py

@@ -339,10 +339,22 @@ def check_for_docker(self) -> None:
             )
 
     def check_for_lfs(self) -> None:
-        # Check dataset has been pulled from git lfs
+        """
+        Ensure required papers for the selected split are hydrated from LFS.
+
+        We only validate papers that appear in the active ``paper_split`` to allow
+        lightweight CI runs that hydrate a minimal subset of the dataset.
+        """
+
         papers_dir = get_paperbench_data_dir() / "papers"
-        papers = list(papers_dir.glob("**/paper.md"))
+        split_path = get_experiments_dir() / "splits" / f"{self.paper_split}.txt"
+
+        paper_ids = [line.strip() for line in split_path.read_text().splitlines() if line.strip()]
 
-        for paper in papers:
-            with open(paper, "r") as f:
-                assert len(f.readlines()) > 5, f"Paper at {paper} should be pulled from git lfs"
+        for paper_id in paper_ids:
+            paper_path = papers_dir / paper_id / "paper.md"
+            assert paper_path.exists(), f"Paper at {paper_path} is missing; hydrate via git lfs."
+            with open(paper_path, "r") as f:
+                assert len(f.readlines()) > 5, (
+                    f"Paper at {paper_path} should be pulled from git lfs"
+                )