A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 10, 12:45 UTC
[BugFix] Fix flakiness in test_eagle_dp for PyTorch 2.10 (#31915)
Signed-off-by: Richard Zou <zou3519@gmail.com>
Richard Zou committed
Jan 7, 2026 at 23:04 UTC
a79079feef63d7a8124782cf2eb1cd68cbde99ee
1 file changed
+3
-1
tests/v1/distributed/test_eagle_dp.py
+3
-1
@@ -47,7 +47,9 @@ async def test_run_eagle_dp(monkeypatch: pytest.MonkeyPatch):
47
)
48
49
prompt = "This is a test of data parallel with eagle"
50
- num_expected_tokens = 100
50
+ # This test might be flaky, see
51
+ # https://github.com/vllm-project/vllm/issues/31913
52
+ num_expected_tokens = 20
53
sampling_params = SamplingParams(
54
min_tokens=num_expected_tokens,
55
max_tokens=num_expected_tokens,