[CI tiny fix] Enhance robustness of vision chunked prefill test with ROUGE-L metric (#13793)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
This commit is contained in:
co-authored by
github-actions[bot]
parent
7cc43bd453
commit
407cb3ce1e
@@ -19,6 +19,7 @@ from sglang.test.test_utils import (
|
|||||||
DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
|
DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
|
||||||
DEFAULT_URL_FOR_TEST,
|
DEFAULT_URL_FOR_TEST,
|
||||||
CustomTestCase,
|
CustomTestCase,
|
||||||
|
calculate_rouge_l,
|
||||||
popen_launch_server,
|
popen_launch_server,
|
||||||
)
|
)
|
||||||
|
|
||||||
@@ -178,7 +179,18 @@ class TestVisionChunkedPrefill(CustomTestCase):
|
|||||||
print(output_chunked)
|
print(output_chunked)
|
||||||
print("output without chunked prefill:")
|
print("output without chunked prefill:")
|
||||||
print(output_no_chunked)
|
print(output_no_chunked)
|
||||||
self.assertEqual(output_chunked, output_no_chunked)
|
self.assertEqual(len(output_chunked), len(output_no_chunked))
|
||||||
|
rouge_scores = calculate_rouge_l(output_chunked, output_no_chunked)
|
||||||
|
avg_score = sum(rouge_scores) / len(rouge_scores)
|
||||||
|
print(f"ROUGE-L scores: {rouge_scores}")
|
||||||
|
print(f"Average ROUGE-L score: {avg_score:.4f}")
|
||||||
|
# Allow for occasional divergence in one item while maintaining overall output quality
|
||||||
|
self.assertGreater(
|
||||||
|
avg_score,
|
||||||
|
0.90,
|
||||||
|
f"Average ROUGE-L score too low: {avg_score:.4f}. "
|
||||||
|
f"Individual scores: {rouge_scores}",
|
||||||
|
)
|
||||||
|
|
||||||
def test_chunked_prefill(self):
|
def test_chunked_prefill(self):
|
||||||
self._test_chunked_prefill(batches=[False, True], num_frames=[1, [2, 6, 8, 10]])
|
self._test_chunked_prefill(batches=[False, True], num_frames=[1, [2, 6, 8, 10]])
|
||||||
|
|||||||
Reference in New Issue
Block a user