Add AWQ quantization support for NPU. (#10158)

Co-authored-by: Alisehen <814073252@qq.com>
Co-authored-by: Yaochen Han <48639761+Alisehen@users.noreply.github.com>
Co-authored-by: Zhengda Qin <zhengdqin@gmail.com>

python/sglang/srt/model_loader/loader.py

@@ -612,6 +612,8 @@ class DefaultModelLoader(BaseModelLoader):
                 # parameters onto device for processing and back off after.
                 with device_loading_context(module, target_device):
                     quant_method.process_weights_after_loading(module)
+                if _is_npu:
+                    torch.npu.empty_cache()
 
 
 class LayeredModelLoader(DefaultModelLoader):