I use tensorflow-to-dlc script of snpe 1.8 to convert an tensorflow model with tf.add layer.I'm met with:
WARNING_TF_SCOPE_OP_NOT_CONSUMED:Add
But tensorflow-to-dlc script of snpe 1.2 is ok.
if I use tf.add in snpe-1.8，how to correct the error
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.
Hi,
How do you write add operation?
Do you mean the same code performs different on snpe 1.2 and 1.8?
yes .I add two convolution layers in tensorflow.It is elementwise layer operation.snpe 1.2 is ok with same code.
If you used tf.add for adding biases on convolution layer, you'd rather change it to tf.nn.bias_add
I should do that two convolution layer add，their dimensions are（1，15，15，128）and （1，15，15，128）.I convert the operation to （1，1，1，15*15*128）add （15*15*128）with tf.nn.bias_add，but I still meet the following error.
WARING -WRANG_TF_SCOPE_OP_NOT_CONSUMED：Operation（add）not consumed by converter：BiasAdd.
ERROR-Conversion failed：ERROR_TF_OPERATION_NOT_MAPPED_TO_LAYER：Some operations in the Tensorflow graph were not resolved to a layer.