xtcl support fp16 onnx

FlagOpen · Aug 25, 2023 · f39b1f4 · f39b1f4
1 parent 58f41ae
commit f39b1f4
Show file tree

Hide file tree

Showing 2 changed files with 1 addition and 1 deletion.
diff --git a/inference/benchmarks/resnet50/README.md b/inference/benchmarks/resnet50/README.md
@@ -37,7 +37,6 @@ find ./val -name "*JPEG" | wc -l
   * pytorch：torchvision.models.resnet50
 * 权重下载
   * pytorch：https://download.pytorch.org/models/resnet50-0676ba61.pth
-  * kunlunxin_onnx: https://baidu-kunlun-public.su.bcebos.com/XTCL/kunlunxin_flagperf_resnet50.tar.gz
 
 ### 2. 软硬件配置与运行信息参考
 

diff --git a/inference/inference_engine/kunlunxin/xtcl.py b/inference/inference_engine/kunlunxin/xtcl.py
@@ -27,6 +27,7 @@ def build_engine(self, config, onnx_path):
             input_name = input.name #'inputs:0'
             self.input_names.append(input_name)
             shape_dict[input_name] = input_shape
+
         mod, params = relay.frontend.from_onnx(onnx_model, shape_dict)
 
         target_host = f'llvm -acc=xpu{os.environ.get("XPUSIM_DEVICE_MODEL", "KUNLUN1")[-1]}'