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 15, 02:15 UTC
Temporarily suspend google/gemma-3-1b-it. (#20722)
Signed-off-by: Qiliang Cui <derrhein@gmail.com>
QiliangCui committed
Jul 11, 2025 at 04:21 UTC
b4f0b5f9aaa8b93a335c12b7f412f44f17d704e0
1 file changed
+6
tests/entrypoints/llm/test_accuracy.py
+6
@@ -69,6 +69,12 @@ def test_lm_eval_accuracy_v1_engine(model, monkeypatch: pytest.MonkeyPatch):
69
more_args = None
70
if current_platform.is_tpu():
71
# Limit compilation time for TPU V1
72
+
73
+ if model == "google/gemma-3-1b-it":
74
+ pytest.skip(
75
+ "Temporarily disabled due to test failures"
76
+ "(timeout or accuracy mismatch). Re-enable once fixed.")
77
+
78
more_args = "max_model_len=2048,max_num_seqs=64"
79
80
# Add TP test (if provided)