Application exited with code 128 (Unknown)

Help
#1

Question/Issue: I have tried the BYOM feature by training the model locally and then pushing it to the project. However, when I try to add the block as a training block and train/retrain the model I get a job failed message stating: “Application exited with code 128 (Unknown)”. What is the meaning of this message, I haven’t yet seen this error in the forum.

Project ID: 337888.

Context/Use case: Object detection.

BR, Jan

#2

Hi @janp,

If you are creating a custom learning block, you need to make sure that the block’s and model’s input and output dimensions are very specific shapes. Please see here for more information.

If you are training locally, then you do not want to use a Custom Learning Block. Instead, you should be using the BYOM feature, which is different from custom learning blocks.

1 Like
#3

Hi @shawn_edgeimpulse, thank you for your reply!

I followed the instructions in custom learning blocks and I used docker to build and run the container. As is stated under the subsection editing built-in-blocks, I edited the FOMO block from my project locally and then pushed back as I followed the instructions in the README file. As I tried to train in Edge Impulse with the pushed block I got the message: “Application exited with code 128 (Unknown)”. Does the message appear because of the specific model’s input and output dimension as you mentioned? How could I try to overcome that?

BR, Jan

#4

Hi @janp,

I apologize for the long delay. Have you had any luck with your custom model? If not, could you describe what you are trying to do? Do you want to train a model locally, create custom training code, modify a model, etc.?