-
Notifications
You must be signed in to change notification settings - Fork 6.8k
Cannot continue with MXNet Tutorial: Real-time Object Detection with MXNet On The Raspberry Pi #9094
Comments
I'm experiencing the same problem with a similar config (the difference is I'm running MXNet in a virtual env). I have tried:
Neither fix the problem. The log message suggests that an attempt is being made to upgrade the symbol from v0.8.0 but fails for some reason. I'm new to MXNet and not sure what the differences might be in the symbol json file or API but I have also scoured the internet looking for way to manually edit the json file with no luck. Any suggestions would be appreciated. |
Hi, |
I have the same issue here. Using mxnet 1.0.1. Getting the actual error message. Do I have to reinstall mxnet using 0.8.0? |
Hi yimingliu0216 |
Thanks for the reply. I already downgrade to 1.0.0. It works fine now, but it wastes my date to reinstall the mxnet. |
Can you @wildhunter-66 tell me how to downgrade the mxnet version form 1.0.1 to v1.0.0 without having to reinstall from starting? I'm getting same error on v1.0.1 |
@yimingliu0216 Could you tell me how you downgraded version? |
Is this fixed with the wine detection tutorial improvements? |
If you have further issues, feel free to reopen or create a new one. I think we should close this for now. |
Description
Cannot continue in MXNet Raspberry Pi tutorial (https://mxnet.incubator.apache.org/tutorials/embedded/wine_detector.html) after running
import inception_predict
in Python interpreter. Error describes the following:src/nnvm/legacy_json_util.cc:190: Loading symbol saved by previous version v0.8.0. Attempting to upgrade ... Segmentation fault
Environment info (Required)
Package used (Python/R/Scala/Julia):
Python 2.7
Build info (Required if built from source)
I used make? Instructions found at #7852
MXNet commit hash:
41394a4
Build config:
Error Message:
Minimum reproducible example
Steps to reproduce
What have you tried to solve it?
Should I try to build an older version of MXNet and try it again?
The text was updated successfully, but these errors were encountered: