horizon_nn.build_onnx报错

docker image: openexplorer/ai_toolchain_ubuntu_20_xj3_cpu:v2.6.2b

hb_mapper makertbin报错如何，其中Add_601是Conv的输出

WARNING: Sub_903 not supported by BPU

Layer ArgMax_907

When keep_score != keep_index, the number of channels of output is expected to be 1*group_num (group_num=1), but 2048 provided.

WARNING: ReduceSum_905 not supported by BPU

WARNING: ReduceMax_902 not supported by BPU

WARNING: Sub_903 not supported by BPU

Layer ArgMax_907

When keep_score != keep_index, the number of channels of output is expected to be 1*group_num (group_num=1), but 2048 provided.

WARNING: ReduceSum_905 not supported by BPU

WARNING: ReduceMax_902 not supported by BPU

2023-08-18 11:13:34,069 ERROR *** ERROR-OCCUR-DURING {horizon_nn.build_onnx} ***, error message: There is no input Add_38_onnx::Add_601_x_scale1 in the calibration data set.

2023-08-18 11:13:34,083 INFO [Fri Aug 18 11:13:34 2023] End to calibrate the model.

2023-08-18 11:13:34,086 INFO [Fri Aug 18 11:13:34 2023] End to Horizon NN Model Convert.