Forums - Invalid Pointer Error

1 post / 0 new
Invalid Pointer Error
jasonjeong
Join Date: 21 Aug 19
Posts: 1
Posted: Thu, 2019-08-22 00:51

Hi,

I'm having an error while using the snpe-tensorflow-to-dlc file to convert a .pb file to a dlc file.

Previously I had errors like this which I was able to resolve using the tf.split function instead of using python notation to do strided splice.

<snpe.converters.tensorflow.layers.eltwise.Descriptor object at 0x7ff4c4360090>
pred_sbbox/mul_2
<snpe.converters.tensorflow.layers.concat.Descriptor object at 0x7ff4c45cd850>
pred_sbbox/concat_1
E0820 17:33:40.243328 140690403510080 snpe-tensorflow-to-dlc:109] Encountered Error: error_code=1002; error_message=Layer paramter value is invalid. Layer pred_lbbox/strided_slice_2 input rank must be equal to the strides size.; error_component=Model Validation; line_no=3140; thread_id=140690403510080
2019-08-20 17:33:40,243 - 109 - ERROR - Encountered Error: error_code=1002; error_message=Layer paramter value is invalid. Layer pred_lbbox/strided_slice_2 input rank must be equal to the strides size.; error_component=Model Validation; line_no=3140; thread_id=140690403510080
Traceback (most recent call last):
  File "/home/jasonjeong/Downloads/snpe-1.25.1.310/bin/x86_64-linux-clang/snpe-tensorflow-to-dlc", line 103, in main
    converter.convert(args.dlc, args.copyright_file, args.model_version, converter_command)
  File "/home/jasonjeong/Downloads/snpe-1.25.1.310/lib/python/snpe/converters/tensorflow/converter.py", line 309, in convert
    self._convert_layers()
  File "/home/jasonjeong/Downloads/snpe-1.25.1.310/lib/python/snpe/converters/tensorflow/converter.py", line 361, in _convert_layers
    self._create_layers(descriptors)
  File "/home/jasonjeong/Downloads/snpe-1.25.1.310/lib/python/snpe/converters/tensorflow/converter.py", line 408, in _create_layers
    self._create_layer(layer_builder, descriptor)
  File "/home/jasonjeong/Downloads/snpe-1.25.1.310/lib/python/snpe/converters/tensorflow/converter.py", line 511, in _create_layer
    layer_builder.build_layer(self._context, descriptor, inputs, outputs)
  File "/home/jasonjeong/Downloads/snpe-1.25.1.310/lib/python/snpe/converters/tensorflow/layers/strided_slice.py", line 149, in build_layer
    shrink_axis_mask=descriptor.shrink_axis_mask)
RuntimeError: error_code=1002; error_message=Layer paramter value is invalid. Layer pred_lbbox/strided_slice_2 input rank must be equal to the strides size.; error_component=Model Validation; line_no=3140; thread_id=140690403510080
 

 However after running the snpe-to-tensorflow-dlc, I encountered a munmap_chunk(): invalid pointer error, which after looking around online is a result during free() portion of the code. 

<snpe.converters.tensorflow.layers.constant.Descriptor object at 0x7fa87c0e4090>
pred_lbbox/mul_1/y
<snpe.converters.tensorflow.layers.eltwise.Descriptor object at 0x7fa87c2bfd90>
pred_lbbox/mul_1
<snpe.converters.tensorflow.layers.eltwise.Descriptor object at 0x7fa87c2bfdd0>
pred_lbbox/mul_2
<snpe.converters.tensorflow.layers.concat.Descriptor object at 0x7fa86077fe10>
pred_lbbox/concat_1
munmap_chunk(): invalid pointer
Aborted (core dumped)
I added a print line in one of the create layers python file to see if the layers were being made correctly (that's why there's all these pred_lbbox/mul_1/y lines). 
For reference, I'm using a yolov3 tensorflow code from https://github.com/YunYang1994/tensorflow-yolov3
 
I'm wondering if anyone can help me with figuring out what would possibly cause the munmap_chunk(): invalid pointer error. One of my colleagues said maybe it could be due to the network having residual layers(?).
 
If someone could advise on how to manually debug snpe-to-tensorflow-dlc file, that would also help greatly
 
Thanks,
Jason
  • Up0
  • Down0

Opinions expressed in the content posted here are the personal opinions of the original authors, and do not necessarily reflect those of Qualcomm Incorporated or its subsidiaries (“Qualcomm”). The content is provided for informational purposes only and is not meant to be an endorsement or representation by Qualcomm or any other party. This site may also provide links or references to non-Qualcomm sites and resources. Qualcomm makes no representations, warranties, or other commitments whatsoever about any non-Qualcomm sites or third-party resources that may be referenced, accessible from, or linked to this site.