fix compressed tensors WNA16 imports (#8142)

2025-07-18 19:34:14 +01:00
parent 719b29f218
commit fd63b62eaa
1 changed files with 4 additions and 1 deletions
--- a/python/sglang/srt/layers/quantization/compressed_tensors/compressed_tensors.py
+++ b/python/sglang/srt/layers/quantization/compressed_tensors/compressed_tensors.py
@@ -40,7 +40,10 @@ from sglang.srt.layers.quantization.compressed_tensors.utils import (
 from sglang.srt.layers.quantization.unquant import UnquantizedLinearMethod

 try:
-    import vllm
+    from vllm.model_executor.layers.quantization.compressed_tensors.schemes.compressed_tensors_wNa16 import (
+        WNA16_SUPPORTED_BITS,
+        CompressedTensorsWNA16,
+    )

    VLLM_AVAILABLE = True
 except ImportError: