improve the robustness of parsing vlms config in AutoRound (#18894)

Signed-off-by: wenhuach21 <wenhua.cheng@intel.com>

improve the robustness of parsing vlms config in AutoRound (#18894)
Signed-off-by: wenhuach21 <wenhua.cheng@intel.com>
3de3eadf · Wenhua Cheng · GitHub · 3132290a · 3de3eadf
Unverified Commit 3de3eadf authored May 30, 2025 by Wenhua Cheng Committed by GitHub May 29, 2025
Hide whitespace changes
Inline Side-by-side

Showing with 3 additions and 2 deletions

vllm/model_executor/layers/quantization/auto_round.py vllm/model_executor/layers/quantization/auto_round.py +3 -2

No files found.
--- 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
+            quantized = any(
-                            for name in self.block_name_to_quantize)
+                layer_name.startswith(name)
+                for name in self.block_name_to_quantize)
        elif isinstance(layer, ParallelLMHead):
            quantized = False