修复CompressedTensors的w8a16的支持

3d4721f2 · yangql1 · 3249f6cc · 3d4721f2
Commit 3d4721f2 authored Feb 11, 2026 by yangql1
Hide whitespace changes
Inline Side-by-side

Showing with 1 addition and 1 deletion

vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors.py ...ers/quantization/compressed_tensors/compressed_tensors.py +1 -1

No files found.
--- a/vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors.py
+++ b/vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors.py
@@ -730,7 +730,7 @@ class CompressedTensorsLinearMethod(LinearMethodBase):
        )
        if weights_scheme is not None:
            num_bits = weights_scheme.num_bits
-            if num_bits == 4:
+            if isinstance(layer.scheme, CompressedTensorsWNA16):
                return layer.scheme.process_weights_after_loading(layer)   

        n=layer.weight.shape[0]