From b53b1c7ffe7aebdafd0876350f30e51d1226c92a Mon Sep 17 00:00:00 2001 From: Wentao Ye <44945378+yewentao256@users.noreply.github.com> Date: Thu, 18 Jun 2026 15:20:44 -0400 Subject: [PATCH] [Model Runner V2] Migration to support quantized model by default [5/N] (#44446) Signed-off-by: yewentao256 --- tests/test_config.py | 2 +- vllm/config/vllm.py | 3 --- 2 files changed, 1 insertion(+), 4 deletions(-) diff --git a/tests/test_config.py b/tests/test_config.py index d992ac2969..eb9b11535b 100644 --- a/tests/test_config.py +++ b/tests/test_config.py @@ -188,7 +188,7 @@ def test_v2_model_runner_env_tri_state(monkeypatch, env_value, expected): is_moe=False, is_quantized=True, ), - False, + True, ), ( SimpleNamespace( diff --git a/vllm/config/vllm.py b/vllm/config/vllm.py index ba20d75fa1..ba7d26c93b 100644 --- a/vllm/config/vllm.py +++ b/vllm/config/vllm.py @@ -555,9 +555,6 @@ class VllmConfig: if model_config.runner_type != "generate": return False - if model_config.is_quantized: - return False - architectures = getattr(model_config, "architectures", []) return any( arch in DEFAULT_V2_MODEL_RUNNER_ARCHITECTURES for arch in architectures