Hemanth-Jonnala
Hemanth-Jonnala
@RobertChenKFC yes, it is an issue on our side. Unfortunately, as of now we don't have any plans to fix this issue. I have a filed a internal bug for...
@xsacha yes, you can map that op to the CPU using `-i flag` during the compilation. `! edgetpu_compiler -s -i "node(operation) name" model.tflite`
@tino926 As tflite does not support partial quantization yet, the only option is to split the mode into [two parts](https://stackoverflow.com/questions/43349489/how-to-split-a-model-into-two-seperate-models/43363915). But, I am not sure if it can be done...
> Do you think that if I can solve the inconsistent result by updating the tools do make it compatible with tf2.5? @tino926 may be, please try with tf2.5.
Hi, can you share the output of following commands: - uname -r - lsmod | grep apex
@taloot please check this comment if it helps https://github.com/google-coral/edgetpu/issues/419#issuecomment-889878751
@taloot does your issue got resolved with new compiler? If yes, please close the ticket.
@taloot can you please share the tflite model with size 640. Thanks!
can you please share the output of below snippet. ``` root@root ~# python3 Python 3.7.3 (default, Jan 22 2021, 20:04:44) [GCC 8.3.0] on linux Type "help", "copyright", "credits" or "license"...
can you please paste the output of below command: `python3 -c 'from pycoral.utils.edgetpu import get_runtime_version; print(get_runtime_version())'`