[Model Runner V2] Migration to support quantized model by default [5/N] (#44446)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Wentao Ye committed
Jun 18, 2026 at 15:20 UTC
b53b1c7ffe7aebdafd0876350f30e51d1226c92a
2 files changed
+1
-4
tests/test_config.py
+1
-1
@@ -188,7 +188,7 @@ def test_v2_model_runner_env_tri_state(monkeypatch, env_value, expected):
188
is_moe=False,
189
is_quantized=True,
190
),
191
- False,
191
+ True,
192
),
193
(
194
SimpleNamespace(
vllm/config/vllm.py
-3
@@ -555,9 +555,6 @@ class VllmConfig:
555
if model_config.runner_type != "generate":
556
return False
557
558
- if model_config.is_quantized:
559
- return False
560
-
558
architectures = getattr(model_config, "architectures", [])
559
return any(
560
arch in DEFAULT_V2_MODEL_RUNNER_ARCHITECTURES for arch in architectures