Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 6 additions & 2 deletions .github/workflows/build-swebenchmultimodal-images.yml
Original file line number Diff line number Diff line change
Expand Up @@ -161,8 +161,12 @@ jobs:
if: ${{ github.event_name == 'workflow_dispatch' && inputs.sdk-commit != '' }}
run: |
cd vendor/software-agent-sdk
git fetch origin ${{ inputs.sdk-commit }}
git checkout FETCH_HEAD
# Fetch from origin (handles both branch names and commit SHAs)
git fetch origin
# Try to fetch the specific ref (works for branches/tags)
git fetch origin ${{ inputs.sdk-commit }} 2>/dev/null || true
# Checkout the commit/ref (works for both branches and commit SHAs)
git checkout ${{ inputs.sdk-commit }}
SDK_SHA=$(git rev-parse HEAD)
cd ../..
# Stage the submodule reference update so make build uses it
Expand Down
17 changes: 14 additions & 3 deletions benchmarks/swebenchmultimodal/build_images.py
Original file line number Diff line number Diff line change
Expand Up @@ -57,9 +57,17 @@ def extract_custom_tag(base_image: str) -> str:
return name


def collect_unique_base_images(dataset, split, n_limit):
def collect_unique_base_images(
dataset,
split,
n_limit,
selected_instances_file: str | None = None,
):
df = get_dataset(
dataset_name=dataset, split=split, eval_limit=n_limit if n_limit else None
dataset_name=dataset,
split=split,
eval_limit=n_limit if n_limit else None,
selected_instances_file=selected_instances_file,
)
return sorted(
{get_official_docker_image(str(row["instance_id"])) for _, row in df.iterrows()}
Expand All @@ -71,7 +79,10 @@ def main(argv: list[str]) -> int:
args = parser.parse_args(argv)

base_images: list[str] = collect_unique_base_images(
args.dataset, args.split, args.n_limit
args.dataset,
args.split,
args.n_limit,
args.select,
)
build_dir = default_build_output_dir(args.dataset, args.split)
return build_all_images(
Expand Down