【寒武纪硬件产品型号】：MLU270

【使用操作系统】：ubuntu18.04

【使用操作系统】：Cambricon PyTorch 适配的原生 PyTorch 版本为 v1.3.0

【当前已做了哪些信息确认】：

在生成离线模型过程中，打印部分信息如下：

max Op cannot run on MLU device, start running on CPU!
[WARNING][/pytorch/catch/torch_mlu/csrc/aten/operators/op_methods.cpp][line:1437][min][thread:140425527846720][process:37]:
min Op cannot run on MLU device, start running on CPU!
[WARNING][/pytorch/catch/torch_mlu/csrc/aten/operators/op_methods.cpp][line:1379][max][thread:140425527846720][process:37]:
max Op cannot run on MLU device, start running on CPU!
[WARNING][/pytorch/catch/torch_mlu/csrc/aten/operators/op_methods.cpp][line:1437][min][thread:140425527846720][process:37]:
min Op cannot run on MLU device, start running on CPU!
[WARNING][/pytorch/catch/torch_mlu/csrc/aten/operators/op_methods.cpp][line:1379][max][thread:140425527846720][process:37]:
max Op cannot run on MLU device, start running on CPU!
[WARNING][/pytorch/catch/torch_mlu/csrc/aten/operators/op_methods.cpp][line:1437][min][thread:140425527846720][process:37]:
min Op cannot run on MLU device, start running on CPU!
[WARNING][/pytorch/catch/torch_mlu/csrc/aten/operators/op_methods.cpp][line:1379][max][thread:140425527846720][process:37]:
max Op cannot run on MLU device, start running on CPU!
[WARNING][/pytorch/catch/torch_mlu/csrc/aten/operators/op_methods.cpp][line:1705][norm][thread:140425527846720][process:37]:
norm Op cannot run on MLU device, start running on CPU!
[WARNING][/pytorch/catch/torch_mlu/csrc/jit/passes/segment_graph.cpp][line:41][MLUSupport][thread:140425527846720][process:37]:
[Fusion Segment] Please check mlu_functions.yaml && Maybe MLU fusion does NOT supports op: norm
%norm : Tensor = aten::norm(%input, %925, %2935, %919), scope: Backbone # /torch/venv3/pytorch/lib/python3.6/site-packages/torch/functional.py:698:0

[WARNING][/pytorch/catch/torch_mlu/csrc/jit/passes/segment_graph.cpp][line:41][MLUSupport][thread:140425527846720][process:37]:
[Fusion Segment] Please check mlu_functions.yaml && Maybe MLU fusion does NOT supports op: norm
%norm : Tensor = aten::norm(%input, %925, %2935, %919), scope: Backbone # /torch/venv3/pytorch/lib/python3.6/site-packages/torch/functional.py:698:0

[WARNING][/pytorch/catch/torch_mlu/csrc/jit/passes/segment_graph.cpp][line:41][MLUSupport][thread:140425527846720][process:37]:
[Fusion Segment] Please check mlu_functions.yaml && Maybe MLU fusion does NOT supports op: norm
%norm : Tensor = aten::norm(%input, %925, %2935, %919), scope: Backbone # /torch/venv3/pytorch/lib/python3.6/site-packages/torch/functional.py:698:0

[WARNING][/pytorch/catch/torch_mlu/csrc/jit/passes/segment_graph.cpp][line:41][MLUSupport][thread:140425527846720][process:37]:
[Fusion Segment] Please check mlu_functions.yaml && Maybe MLU fusion does NOT supports op: norm
%norm : Tensor = aten::norm(%input, %925, %2935, %919), scope: Backbone # /torch/venv3/pytorch/lib/python3.6/site-packages/torch/functional.py:698:0

[WARNING][/pytorch/catch/torch_mlu/csrc/jit/passes/segment_graph.cpp][line:41][MLUSupport][thread:140425527846720][process:37]:
[Fusion Segment] Please check mlu_functions.yaml && Maybe MLU fusion does NOT supports op: norm
%norm : Tensor = aten::norm(%input, %925, %2935, %919), scope: Backbone # /torch/venv3/pytorch/lib/python3.6/site-packages/torch/functional.py:698:0

[WARNING][/pytorch/catch/torch_mlu/csrc/jit/passes/segment_graph.cpp][line:217][SegmentGraph][thread:140425527846720][process:37]:
Graph is segmented into 2 subgraphs.
Please check the above WARNING logs which include "[Fusion Segment]" to see whether there are MLU unsupported ops exists in pytorch model.

[WARNING][/pytorch/catch/torch_mlu/csrc/aten/operators/op_methods.cpp][line:1705][norm][thread:140425527846720][process:37]:
norm Op cannot run on MLU device, start running on CPU!
...

打开离线模型信息：

This is the Twins File of "test.cambricon".
Function number: 1
---------------kernel graph--------------------

Function #0 {
Kernel num:1
Cache mode:0
Name: subnet0
Input number: 2
    Input #0.
        Mask: 338036233
        Shape(dim): 1 1 1 512
        Name:
        Id: 3045
        Data type: CNRT_FLOAT32        Dim Order: CNRT_NHWC
        Quantize position: 0
        Quantize scale: 0.000000
    Input #1.
        Mask: 338036233
        Shape(dim): 1 1 1 1
        Name:
        Id: 3047
        Data type: CNRT_FLOAT32        Dim Order: CNRT_NHWC
        Quantize position: 32694
        Quantize scale: 0.000000
Output number: 1
    Output #0.
        Mask: 338036233
        Shape(dim): 1 1 1 512
        Name:
        Id: 3050
        Data type: CNRT_FLOAT32        Dim Order: CNRT_NHWC
        Quantize position: 32694
        Quantize scale: 0.000000
Kernel header #0:
    Name: subnet0762
    Core version:
    Model Parallelism: 16
    Core limit: 16
    Inst data split: true

可以看到离线模型被拆成了两个输入，正确的input输入应该是1，3，112，112

有查看了“mlu_functions.yaml

插件有norm,min,max等层

你好，请问norm层报错，需要怎么解决呢？