From 3de3eadf5b1c271ccd7140526ffb3f850d6b0189 Mon Sep 17 00:00:00 2001 From: Wenhua Cheng Date: Fri, 30 May 2025 10:24:47 +0800 Subject: [PATCH] improve the robustness of parsing vlms config in AutoRound (#18894) Signed-off-by: wenhuach21 --- vllm/model_executor/layers/quantization/auto_round.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/vllm/model_executor/layers/quantization/auto_round.py b/vllm/model_executor/layers/quantization/auto_round.py index 2d9f5e52bd65a..eb8ffa37882cb 100644 --- a/vllm/model_executor/layers/quantization/auto_round.py +++ b/vllm/model_executor/layers/quantization/auto_round.py @@ -116,8 +116,9 @@ class AutoRoundConfig(QuantizationConfig): quantized = True if self.block_name_to_quantize: - quantized = any(name in layer_name - for name in self.block_name_to_quantize) + quantized = any( + layer_name.startswith(name) + for name in self.block_name_to_quantize) elif isinstance(layer, ParallelLMHead): quantized = False