I am currently working on a time-sensitive project and urgently need assistance with an issue I believe is related to a bug in the export feature.
My model trains and previews correctly without any issues. However, when I attempt to export it as a TensorFlow Lite model (both floating point and quantized), or Tensor flow itself I consistently receive the message: “Something went wrong while converting.” The downloaded file is not a valid .zip file and instead appears to be an HTML document.
I have tried multiple troubleshooting steps, including:
Reducing the number of classes and images
Re-uploading all training data
Duplicating the project and exporting from a copy
Creating entirely new models
Trying different browsers and sessions
Uploading and re-importing the project file
Despite all of this, the issue persists across all attempts.
Given that the model itself functions perfectly during preview and training, I believe this may be a backend conversion bug or glitch. This issue is preventing me from completing my project, which is due in 2 days, so I would greatly appreciate any immediate assistance or fix.
If there is any workaround or alternative export method you recommend, please let me know as soon as possible.
I am currently working on a time-sensitive project and urgently need assistance with an issue I believe is related to a bug in the export feature.
My model trains and previews correctly without any issues. However, when I attempt to export it as a TensorFlow Lite model (both floating point and quantized), or Tensor flow itself I consistently receive the message: “Something went wrong while converting.” The downloaded file is not a valid .zip file and instead appears to be an HTML document.
I have tried multiple troubleshooting steps, including:
Reducing the number of classes and images
Re-uploading all training data
Duplicating the project and exporting from a copy
Creating entirely new models
Trying different browsers and sessions
Uploading and re-importing the project file
Despite all of this, the issue persists across all attempts.
Given that the model itself functions perfectly during preview and training, I believe this may be a backend conversion bug or glitch. This issue is preventing me from completing my project, which is due in 2 days, so I would greatly appreciate any immediate assistance or fix.
If there is any workaround or alternative export method you recommend, please let me know as soon as possible.