Error in Compiler steps

[whs111]

AssertionError Traceback (most recent call last) in () 11 xclbin=XCLBIN, 12 device='FPGA', ---> 13 **get_args(inputNode,outputNode ) 14 )

/opt/ml-suite/xfdnn/rt/xdnn_rt_tf.pyc in init(self, args, **kwargs) 48 class TFxdnnRT(_xdnnRT): 49 def init (self, args, **kwargs): ---> 50 super(TFxdnnRT, self).init(TFFrontend, args, **kwargs) 51 52 if not hasattr(self, 'graph_def'):

/opt/ml-suite/xfdnn/rt/xdnn_rt_base.pyc in init(self, compilerFunc, args, **kwargs) 217 ssize,
218 compilerJson
--> 219 = compiler.compile() 220 221 if compilerJson is None:

/opt/ml-suite/xfdnn/tools/compile/bin/xfdnn_compiler_tensorflow.pyc in compile(self)

/opt/ml-suite/xfdnn/tools/compile/network/tensor_tools.pyc in from_tfgraph_to_fpga_code(graph, schedulefilename, outputpng, args)

/opt/ml-suite/xfdnn/tools/compile/bin/xfdnn_compiler_base.pyc in GraphMassage(g, schedule, args)

/opt/ml-suite/xfdnn/tools/compile/graph/dagtools.pyc in convert_offline(graph, xdnn_version)

/opt/ml-suite/xfdnn/tools/compile/quantz/deephi_quant_info.pyc in set_deephi(self, name, num_output, deepy)

/opt/ml-suite/xfdnn/tools/compile/quantz/deephi_quant_info.pyc in set_from_scalars(self, name, num_output, inputs, outputs, weights, biases)

/opt/ml-suite/xfdnn/tools/compile/quantz/deephi_quant_info.pyc in set(self, name, num_output, inputs, outputs, weights, biases)

AssertionError: Mul

It seems that the compiler is not able to handle mul node in pb file.

[wilderfield]

What model were you trying to run? What version of Tensorflow was used to train the model? I suspect maybe your model is from old TF, and the name of the multiply layer has changed. Please note that Vitis-AI is now the official Xilinx AI software deliverable.