In a Linux virtual machine system, I tried to run a large model using Streamlit, but it failed to load the model. Is it because of a timeout or because the device itself cannot run the model? Is there any way to view the logs of Streamlit to help me find the reason for the failure?
Yes you can access this by clicking on the “Manage app” at the bottom right hand corner of the browser, then click “ Download logs”.
Streamlit is logging to the console. Your application is being killed while “Loading checkpoint shards”, whatever that is. It doesn’t look like the issue is related to Streamlit at all.
I have found the problem. The memory is full and the process has been killed.
In the case of memory being full and causing the app to crash, perhaps you could also look into profiling the app as well as several other ways in improving the app performance. Here are a few blogs to get you started:
Hope this helps!