[Bugfix] Fix w8a8_int8 import error on NPU (#8147)

2025-07-19 02:34:55 +08:00
parent fd63b62eaa
commit 6737671c82
1 changed files with 2 additions and 0 deletions
--- a/python/sglang/srt/layers/quantization/w8a8_int8.py
+++ b/python/sglang/srt/layers/quantization/w8a8_int8.py
@@ -754,6 +754,8 @@ class NPU_W8A8LinearMethod(LinearMethodBase):
        x: torch.Tensor,
        bias: Optional[torch.Tensor] = None,
    ) -> torch.Tensor:
+        from sglang.srt.layers.linear import RowParallelLinear
+
        if isinstance(layer, RowParallelLinear):
            tp_rank = get_tensor_model_parallel_rank()
            return self.quant_method.apply(layer, x, bias, tp_rank)