From 36429096171ff8785645c40c662d859dddedd829 Mon Sep 17 00:00:00 2001 From: JartX Date: Sat, 20 Sep 2025 05:18:13 +0200 Subject: [PATCH] [BUGFIX] GPTQ quantization compatibility for Qwen3 Next MOE models (AutoGPTQ and AutoRound-GPTQ) (#25268) Signed-off-by: JartX --- vllm/model_executor/models/qwen3_next.py | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/vllm/model_executor/models/qwen3_next.py b/vllm/model_executor/models/qwen3_next.py index 98749c160ba4..ce917f92bd2e 100644 --- a/vllm/model_executor/models/qwen3_next.py +++ b/vllm/model_executor/models/qwen3_next.py @@ -148,9 +148,11 @@ class Qwen3NextSparseMoeBlock(nn.Module): def _maybe_ignore_quant_config(self, quant_config: QuantizationConfig): # GPTQ configs do not have a list of ignored modules, however AutoGPTQ - # seems to avoid gate quantization. - # See: https://huggingface.co/Qwen/Qwen3-30B-A3B-GPTQ-Int4 - if isinstance(quant_config, (GPTQConfig, GPTQMarlinConfig)): + # seems to avoid gate quantization while AutoRound does. + if isinstance( + quant_config, + (GPTQConfig, + GPTQMarlinConfig)) and not quant_config.autoround_version: return None return quant_config