I have two zip files of image classification files that are structurally identical. Approx 200 class folders.
The smaller one is around 1GB and uploads and trains fine.
The larger one is ~4GB and it crashes the Docker container running the Streamlit app but without any errors. It appears to upload the whole way but then nothing happens and in the terminal the app has died.
It says the limit is 10GB. Anyone else experienced this?
you could try running the app locally or on a CPU upgrade as it might need more RAM to process larger files.
this was running locally in the docker container. Is that what you mean or is there another way?
If I try to run it online with upgraded CPU I get this:
An error has occurred: Value out of range: 5154181263
Seems like a zipping issue. Does it decompress fine?
it unzips ok on OSX. Any potential issues zipping from that OS?
I just tested by creating an archive in OSX (although not 4gb in size) so, I dont see a reason why that wont work. Is a smaller zip working fine for you? If yes, I’ll try by creating a larger dataset on my end.
yes I have a subset which is 1GB and works fine
thanks! im taking a look!
Much appreciated. I just tried with a larger subset at 2.6GB and it also failed in the same way
I’m not sure if it’s related, but it seems like there is an error with the form leading to a erroneous submission
ERROR app has failed due to an exception:
ERROR Traceback (most recent call last):
File “/app/src/autotrain/utils.py”, line 120, in wrapper
return func(*args, **kwargs)
File “/app/src/autotrain/app.py”, line 464, in app
raise ValueError(“Please upload training images”)
ValueError: Please upload training images
It still seeeeeems like its working after this appears in the terminal but not 100% sure
did you discover anything? Did it work with a larger upload for you?