专栏算法工具链模型转换过程报错,看之前的帖子此算子应该是支持的,求解,使用的是2.4.2 版本

模型转换过程报错,看之前的帖子此算子应该是支持的,求解,使用的是2.4.2 版本

已解决
超人不会飞2023-03-22
109
6

2023-03-22 11:30:17,008 file: hb_mapper.py func: hb_mapper line No: 70 Start hb_mapper....

2023-03-22 11:30:17,009 file: hb_mapper.py func: hb_mapper line No: 71 log will be stored in /open_explorer/hb_mapper_checker.log

2023-03-22 11:30:17,010 file: hb_mapper.py func: hb_mapper line No: 72 hbdk version 3.27.1

2023-03-22 11:30:17,010 file: hb_mapper.py func: hb_mapper line No: 73 horizon_nn version 0.12.8

2023-03-22 11:30:17,011 file: hb_mapper.py func: hb_mapper line No: 74 hb_mapper version 1.5.7

2023-03-22 11:30:17,011 file: hb_mapper.py func: hb_mapper line No: 76 parameter [output] is deprecated

2023-03-22 11:30:17,039 file: helper.py func: helper line No: 124 Model input names: ['x']

2023-03-22 11:30:17,040 file: hb_mapper_checker.py func: hb_mapper_checker line No: 104 Model type: onnx

2023-03-22 11:30:17,041 file: hb_mapper_checker.py func: hb_mapper_checker line No: 105 march: bernoulli2

2023-03-22 11:30:17,042 file: hb_mapper_checker.py func: hb_mapper_checker line No: 110 input names []

2023-03-22 11:30:17,042 file: hb_mapper_checker.py func: hb_mapper_checker line No: 111 input shapes {}

2023-03-22 11:30:17,042 file: hb_mapper_checker.py func: hb_mapper_checker line No: 117 Begin model checking....

2023-03-22 11:30:17,045 file: build.py func: build line No: 36 [Wed Mar 22 11:30:17 2023] Start to Horizon NN Model Convert.

2023-03-22 11:30:17,045 file: dict_parser.py func: dict_parser line No: 26 The input parameter is not specified, convert with default parameters.

2023-03-22 11:30:17,046 file: dict_parser.py func: dict_parser line No: 491 The hbdk parameter is not specified, and the submodel will be compiled with the default parameter.

2023-03-22 11:30:17,046 file: build.py func: build line No: 143 HorizonNN version: 0.12.8

2023-03-22 11:30:17,047 file: build.py func: build line No: 147 HBDK version: 3.27.1

2023-03-22 11:30:17,048 file: build.py func: build line No: 36 [Wed Mar 22 11:30:17 2023] Start to parse the onnx model.

2023-03-22 11:30:17,120 file: onnx_parser.py func: onnx_parser line No: 146 ONNX model info:

ONNX IR version: 7

Opset version: 11

Input name: x, [1, 3, 48, 320]

2023-03-22 11:30:17,176 file: build.py func: build line No: 39 [Wed Mar 22 11:30:17 2023] End to parse the onnx model.

2023-03-22 11:30:17,177 file: build.py func: build line No: 266 Model input names: ['x']

2023-03-22 11:30:17,201 file: build.py func: build line No: 535 Saving the original float model: ./.hb_check/original_float_model.onnx.

2023-03-22 11:30:17,202 file: build.py func: build line No: 36 [Wed Mar 22 11:30:17 2023] Start to optimize the model.

2023-03-22 11:30:17,807 file: build.py func: build line No: 39 [Wed Mar 22 11:30:17 2023] End to optimize the model.

2023-03-22 11:30:17,831 file: build.py func: build line No: 546 Saving the optimized model: ./.hb_check/optimized_float_model.onnx.

2023-03-22 11:30:17,832 file: build.py func: build line No: 36 [Wed Mar 22 11:30:17 2023] Start to calibrate the model.

2023-03-22 11:30:17,973 file: calibration_data_set.py func: calibration_data_set line No: 66 There are 1 samples in the calibration data set.

2023-03-22 11:30:18,027 file: max_calibrater.py func: max_calibrater line No: 68 Run calibration model with max method.

2023-03-22 11:30:18,291 file: build.py func: build line No: 39 [Wed Mar 22 11:30:18 2023] End to calibrate the model.

2023-03-22 11:30:18,291 file: build.py func: build line No: 36 [Wed Mar 22 11:30:18 2023] Start to quantize the model.

2023-03-22 11:30:18,621 file: build.py func: build line No: 39 [Wed Mar 22 11:30:18 2023] End to quantize the model.

2023-03-22 11:30:18,703 file: build.py func: build line No: 560 Saving the quantized model: ./.hb_check/quantized_model.onnx.

2023-03-22 11:30:18,703 file: build.py func: build line No: 36 [Wed Mar 22 11:30:18 2023] Start to compile the model with march bernoulli2.

2023-03-22 11:30:18,704 file: dict_parser.py func: dict_parser line No: 496 Parsing the hbdk parameter:{'hbdk_pass_through_params': '--O0'}

2023-03-22 11:30:18,816 file: hybrid_build.py func: hybrid_build line No: 121 Compile submodel: Model from PaddlePaddle._subgraph_0

2023-03-22 11:30:18,862 file: hbdk_cc.py func: hbdk_cc line No: 119 hbdk-cc parameters:['--O0', '--input-layout', 'NHWC', '--output-layout', 'NHWC']

2023-03-22 11:30:18,983 file: hybrid_build.py func: hybrid_build line No: 121 Compile submodel: Model from PaddlePaddle._subgraph_1

2023-03-22 11:30:19,013 file: hbdk_cc.py func: hbdk_cc line No: 119 hbdk-cc parameters:['--O0', '--input-layout', 'NHWC', '--output-layout', 'NHWC']

2023-03-22 11:30:19,078 file: hybrid_build.py func: hybrid_build line No: 121 Compile submodel: Model from PaddlePaddle._subgraph_2

2023-03-22 11:30:19,124 file: hbdk_cc.py func: hbdk_cc line No: 119 hbdk-cc parameters:['--O0', '--input-layout', 'NHWC', '--output-layout', 'NHWC']

2023-03-22 11:30:19,206 file: hybrid_build.py func: hybrid_build line No: 121 Compile submodel: Model from PaddlePaddle._subgraph_3

2023-03-22 11:30:19,220 file: hbdk_cc.py func: hbdk_cc line No: 119 hbdk-cc parameters:['--O0', '--input-layout', 'NHWC', '--output-layout', 'NCHW']

2023-03-22 11:30:19,260 file: hybrid_build.py func: hybrid_build line No: 121 Compile submodel: Model from PaddlePaddle._subgraph_4

2023-03-22 11:30:19,273 file: hbdk_cc.py func: hbdk_cc line No: 119 hbdk-cc parameters:['--O0', '--input-layout', 'NHWC', '--output-layout', 'NCHW']

2023-03-22 11:30:19,313 file: hybrid_build.py func: hybrid_build line No: 121 Compile submodel: Model from PaddlePaddle._subgraph_5

2023-03-22 11:30:19,361 file: hbdk_cc.py func: hbdk_cc line No: 119 hbdk-cc parameters:['--O0', '--input-layout', 'NHWC', '--output-layout', 'NCHW']

2023-03-22 11:30:19,599 file: build.py func: build line No: 39 [Wed Mar 22 11:30:19 2023] End to compile the model with march bernoulli2.

2023-03-22 11:30:19,601 file: node_info.py func: node_info line No: 48 The converted model node information:

=====================================================================================

Node ON Subgraph Type

-------------------------------------------------------------------------------------

p2o.Conv.0 BPU id(0) HzSQuantizedConv

p2o.Mul.0 BPU id(0) HzLut

p2o.Conv.1 BPU id(0) HzSQuantizedConv

p2o.Mul.2 BPU id(0) HzLut

p2o.Conv.2 BPU id(0) HzSQuantizedConv

p2o.Mul.4 BPU id(0) HzLut

p2o.Conv.3 BPU id(0) HzSQuantizedConv

p2o.Mul.6 BPU id(0) HzLut

p2o.Conv.4 BPU id(0) HzSQuantizedConv

p2o.Mul.8 BPU id(0) HzLut

p2o.Conv.5 BPU id(0) HzSQuantizedConv

p2o.Mul.10 BPU id(0) HzLut

p2o.Conv.6 BPU id(0) HzSQuantizedConv

p2o.Mul.12 BPU id(0) HzLut

p2o.Conv.7 BPU id(0) HzSQuantizedConv

p2o.Mul.14 BPU id(0) HzLut

p2o.Conv.8 BPU id(0) HzSQuantizedConv

p2o.Mul.16 BPU id(0) HzLut

p2o.Conv.9 BPU id(0) HzSQuantizedConv

p2o.Mul.18 BPU id(0) HzLut

p2o.Conv.10 BPU id(0) HzSQuantizedConv

p2o.Mul.20 BPU id(0) HzLut

p2o.Conv.11 BPU id(0) HzSQuantizedConv

p2o.Mul.22 BPU id(0) HzLut

p2o.Conv.12 BPU id(0) HzSQuantizedConv

p2o.Mul.24 BPU id(0) HzLut

p2o.Conv.13 BPU id(0) HzSQuantizedConv

p2o.Mul.26 BPU id(0) HzLut

p2o.Conv.14 BPU id(0) HzSQuantizedConv

p2o.Mul.28 BPU id(0) HzLut

p2o.Conv.15 BPU id(0) HzSQuantizedConv

p2o.Mul.30 BPU id(0) HzLut

p2o.Conv.16 BPU id(0) HzSQuantizedConv

p2o.Mul.32 BPU id(0) HzLut

p2o.Conv.17 BPU id(0) HzSQuantizedConv

p2o.Mul.34 BPU id(0) HzLut

p2o.Conv.18 BPU id(0) HzSQuantizedConv

p2o.Mul.36 BPU id(0) HzLut

p2o.Conv.19 BPU id(0) HzSQuantizedConv

p2o.Mul.38 BPU id(0) HzLut

p2o.Conv.20 BPU id(0) HzSQuantizedConv

p2o.Mul.40 BPU id(0) HzLut

p2o.Conv.21 BPU id(0) HzSQuantizedConv

p2o.Mul.42 BPU id(0) HzLut

p2o.Conv.22 BPU id(0) HzSQuantizedConv

p2o.Mul.44 BPU id(0) HzLut

p2o.Conv.23 BPU id(0) HzSQuantizedConv

p2o.Mul.46 BPU id(0) HzLut

...GlobalAveragePool.0_SPLIT_WITH_DEPTHWISE_CONV_0 BPU id(0) HzSQuantizedConv

...GlobalAveragePool.0_SPLIT_WITH_DEPTHWISE_CONV_1 BPU id(0) HzSQuantizedConv

...GlobalAveragePool.0_SPLIT_WITH_DEPTHWISE_CONV_2 BPU id(0) HzSQuantizedConv

p2o.GlobalAveragePool.0 BPU id(0) HzSQuantizedConv

p2o.Conv.24 BPU id(0) HzSQuantizedConv

p2o.Conv.25 BPU id(0) HzSQuantizedConv

p2o.HardSigmoid.0 CPU -- HardSigmoid

p2o.Mul.48 BPU id(1) HzSQuantizedMul

p2o.Conv.26 BPU id(1) HzSQuantizedConv

p2o.Mul.50 BPU id(1) HzLut

p2o.Conv.27 BPU id(1) HzSQuantizedConv

p2o.Mul.52 BPU id(1) HzLut

...GlobalAveragePool.2_SPLIT_WITH_DEPTHWISE_CONV_0 BPU id(1) HzSQuantizedConv

...GlobalAveragePool.2_SPLIT_WITH_DEPTHWISE_CONV_1 BPU id(1) HzSQuantizedConv

p2o.GlobalAveragePool.2 BPU id(1) HzSQuantizedConv

p2o.Conv.28 BPU id(1) HzSQuantizedConv

p2o.Conv.29 BPU id(1) HzSQuantizedConv

p2o.HardSigmoid.1 CPU -- HardSigmoid

p2o.Mul.54 BPU id(2) HzSQuantizedMul

p2o.Conv.30 BPU id(2) HzSQuantizedConv

p2o.Mul.56 BPU id(2) HzLut

p2o.AveragePool.0 BPU id(2) HzSQuantizedConv

p2o.Conv.31 BPU id(2) HzSQuantizedConv

p2o.Mul.58 BPU id(2) HzSQuantizedConv

p2o.Sigmoid.0 BPU id(2) HzLut

p2o.Mul.60 BPU id(2) HzSQuantizedMul

p2o.Conv.32 BPU id(2) HzSQuantizedConv

p2o.Mul.61 BPU id(2) HzSQuantizedConv

p2o.Sigmoid.2 BPU id(2) HzLut

p2o.Mul.63 CPU -- Mul

p2o.Reshape.8 CPU -- Reshape

p2o.Transpose.0 CPU -- Transpose

p2o.ReduceMean.0 CPU -- ReduceMean

p2o.Sub.0 CPU -- Sub

p2o.Pow.0 CPU -- Pow

p2o.ReduceMean.2 CPU -- ReduceMean

p2o.Div.27_reciprocal CPU -- HzRsqrt

p2o.Div.27_mul CPU -- Mul

p2o.Mul.64 CPU -- Mul

p2o.Add.64 CPU -- Add

p2o.MatMul.0 CPU -- MatMul

p2o.Add.66 CPU -- Add

p2o.Reshape.11 CPU -- Reshape

p2o.Transpose.1 CPU -- Transpose

p2o.Slice.2 CPU -- Slice

p2o.Squeeze.0 CPU -- Reshape

p2o.Mul.66 BPU id(3) HzSQuantizedConv

p2o.Slice.4 CPU -- Slice

p2o.Squeeze.1 CPU -- Reshape

p2o.Slice.6 CPU -- Slice

p2o.Squeeze.2 CPU -- Reshape

p2o.Transpose.2 CPU -- Transpose

p2o.MatMul.2 CPU -- MatMul

p2o.Softmax.0 CPU -- Softmax

p2o.MatMul.4 CPU -- MatMul

p2o.Transpose.3 CPU -- Transpose

p2o.Reshape.12 CPU -- Reshape

p2o.MatMul.6 CPU -- MatMul

p2o.Add.68 CPU -- Add

p2o.Add.70 CPU -- Add

p2o.ReduceMean.4 CPU -- ReduceMean

p2o.Sub.2 CPU -- Sub

p2o.Pow.2 CPU -- Pow

p2o.ReduceMean.6 CPU -- ReduceMean

p2o.Div.29_reciprocal CPU -- HzRsqrt

p2o.Div.29_mul CPU -- Mul

p2o.Mul.68 CPU -- Mul

p2o.Add.74 CPU -- Add

p2o.MatMul.8 CPU -- MatMul

p2o.Add.76 CPU -- Add

p2o.Mul.70 CPU -- Mul

p2o.Sigmoid.4 CPU -- Sigmoid

p2o.Mul.72 CPU -- Mul

p2o.MatMul.10 CPU -- MatMul

p2o.Add.78 CPU -- Add

p2o.Add.80 CPU -- Add

p2o.ReduceMean.8 CPU -- ReduceMean

p2o.Sub.4 CPU -- Sub

p2o.Pow.4 CPU -- Pow

p2o.ReduceMean.10 CPU -- ReduceMean

p2o.Div.31_reciprocal CPU -- HzRsqrt

p2o.Div.31_mul CPU -- Mul

p2o.Mul.73 CPU -- Mul

p2o.Add.84 CPU -- Add

p2o.MatMul.12 CPU -- MatMul

p2o.Add.86 CPU -- Add

p2o.Reshape.17 CPU -- Reshape

p2o.Transpose.4 CPU -- Transpose

p2o.Slice.10 CPU -- Slice

p2o.Squeeze.3 CPU -- Reshape

p2o.Mul.75 BPU id(4) HzSQuantizedConv

p2o.Slice.12 CPU -- Slice

p2o.Squeeze.4 CPU -- Reshape

p2o.Slice.14 CPU -- Slice

p2o.Squeeze.5 CPU -- Reshape

p2o.Transpose.5 CPU -- Transpose

p2o.MatMul.14 CPU -- MatMul

p2o.Softmax.1 CPU -- Softmax

p2o.MatMul.16 CPU -- MatMul

p2o.Transpose.6 CPU -- Transpose

p2o.Reshape.18 CPU -- Reshape

p2o.MatMul.18 CPU -- MatMul

p2o.Add.88 CPU -- Add

p2o.Add.90 CPU -- Add

p2o.ReduceMean.12 CPU -- ReduceMean

p2o.Sub.6 CPU -- Sub

p2o.Pow.6 CPU -- Pow

p2o.ReduceMean.14 CPU -- ReduceMean

p2o.Div.33_reciprocal CPU -- HzRsqrt

p2o.Div.33_mul CPU -- Mul

p2o.Mul.77 CPU -- Mul

p2o.Add.94 CPU -- Add

p2o.MatMul.20 CPU -- MatMul

p2o.Add.96 CPU -- Add

p2o.Mul.79 CPU -- Mul

p2o.Sigmoid.6 CPU -- Sigmoid

p2o.Mul.81 CPU -- Mul

p2o.MatMul.22 CPU -- MatMul

p2o.Add.98 CPU -- Add

p2o.Add.100 CPU -- Add

p2o.ReduceMean.16 CPU -- ReduceMean

p2o.Sub.8 CPU -- Sub

p2o.Pow.8 CPU -- Pow

p2o.ReduceMean.18 CPU -- ReduceMean

p2o.Div.35_reciprocal CPU -- HzRsqrt

p2o.Div.35_mul CPU -- Mul

p2o.Mul.82 CPU -- Mul

p2o.Add.104 CPU -- Add

p2o.Reshape.23 CPU -- Reshape

p2o.Conv.33 BPU id(5) HzSQuantizedConv

p2o.Mul.84 BPU id(5) HzSQuantizedConv

p2o.Sigmoid.8 BPU id(5) HzLut

p2o.Mul.86 BPU id(5) HzSQuantizedMul

p2o.Concat.2 BPU id(5) Concat

p2o.Conv.34 BPU id(5) HzSQuantizedConv

p2o.Mul.87 BPU id(5) HzSQuantizedConv

p2o.Sigmoid.10 BPU id(5) HzLut

p2o.Mul.89 BPU id(5) HzSQuantizedMul

p2o.Conv.35 BPU id(5) HzSQuantizedConv

p2o.Mul.90 BPU id(5) HzSQuantizedConv

p2o.Sigmoid.12 BPU id(5) HzLut

p2o.Mul.92 CPU -- Mul

p2o.Squeeze.6 CPU -- Reshape

p2o.Transpose.8 CPU -- Transpose

p2o.MatMul.24 CPU -- MatMul

p2o.Add.106 CPU -- Add

p2o.Softmax.2 CPU -- Softmax

2023-03-22 11:30:19,603 file: build.py func: build line No: 39 [Wed Mar 22 11:30:19 2023] End to Horizon NN Model Convert.

2023-03-22 11:30:19,610 file: onnx2horizonrt.py func: onnx2horizonrt line No: 2534 ONNX model output num : 1

2023-03-22 11:30:19,613 file: tool_utils.py func: tool_utils line No: 116 exception in command: checker

2023-03-22 11:30:19,616 file: tool_utils.py func: tool_utils line No: 117 Traceback (most recent call last):

File "/usr/local/lib/python3.6/site-packages/horizon_tc_ui/utils/tool_utils.py", line 114, in __decorator

func(*args, **kargs)

File "/usr/local/lib/python3.6/site-packages/horizon_tc_ui/hb_mapper.py", line 84, in checker

input_shape).run(__version__)

File "/usr/local/lib/python3.6/site-packages/horizon_tc_ui/hb_mapper_checker.py", line 140, in run

{"hb_mapper_version": version})

File "/usr/local/lib/python3.6/site-packages/horizon_tc_ui/hbdtort/onnx2horizonrt.py", line 2549, in build_runtime_model_wrapper

make_nodes(onnx_graph, runtime_graph)

File "/usr/local/lib/python3.6/site-packages/horizon_tc_ui/hbdtort/onnx2horizonrt.py", line 2032, in make_nodes

" ".join(set(invalid_operators)))

ValueError: HorizonRT not support these cpu operators: HzRsqrt

2023-03-22 11:30:19,616 file: tool_utils.py func: tool_utils line No: 118 HorizonRT not support these cpu operators: HzRsqrt

算法工具链
评论2
0/1000
  • 颜值即正义
    Lv.2

    你好,麻烦提供一下模型,我这边来复现一下这个问题哈,模型可以通过网盘链接提供。

    2023-03-22
    0
    4
    • 超人不会飞回复颜值即正义:

      链接:https://pan.baidu.com/s/1Df6_vFCAFAOHLY_d6LE5tw?pwd=2lhe

      提取码:2lhe

      --来自百度网盘超级会员V4的分享

      你好,这是要转换的文件,帮忙看一下原因吧

      2023-03-22
      0
    • 颜值即正义回复超人不会飞:
      你好,这个问题已解决,为了尽快解决您的问题,通过一个临时patch解决,也会在下次正式发版中解决这个问题。
      压缩包:https://pan.horizon.ai/index.php/s/wS6fH3bfRHYi8Gi
      通过在docker中pip install xxx.whl来安装,另外一个是对应的板端工具
      2023-03-24
      2
    • 一朵小白云回复颜值即正义:

      这个包失效了,可以再分享一下吗,我也遇到了这个算子的问题

      2023-05-05
      0
    • 颜值即正义回复一朵小白云:

      直接更新至OE2.5.2即可~

      2023-05-05
      0
  • 颜值即正义
    Lv.2
    2023-04-24
    0
    0