[test/fix]: isolate VLM MMMU eval output dirs to fix nightly-4-gpu cross-test pollution (#24623)

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
This commit is contained in:
Jimmy Shong
2026-05-08 15:01:53 -07:00
committed by GitHub
co-authored by gemini-code-assist[bot]
parent 5dc4c7bef1
commit fa8985486e
3 changed files with 35 additions and 433 deletions
-8
View File
@@ -198,11 +198,7 @@ class MMMUMixin:
# Run evaluation
self.run_mmmu_eval(self.model, output_path)
# Get the result file
# Search recursively for JSON result files (lmms-eval v0.4.1+ creates subdirectories)
result_files = glob.glob(f"{output_path}/**/*.json", recursive=True)
if not result_files:
result_files = glob.glob(f"{output_path}/*.json")
if not result_files:
raise FileNotFoundError(f"No JSON result files found in {output_path}")
@@ -389,11 +385,7 @@ class MMMUMultiModelTestBase(CustomTestCase):
# Run evaluation
self.run_mmmu_eval(model.model, output_path)
# Get the result file
# Search recursively for JSON result files (lmms-eval v0.4.1+ creates subdirectories)
result_files = glob.glob(f"{output_path}/**/*.json", recursive=True)
if not result_files:
result_files = glob.glob(f"{output_path}/*.json")
if not result_files:
raise FileNotFoundError(f"No JSON result files found in {output_path}")