Hi,
That message usually means that the rsession terminated unexpectedly. That project shows up in our logs as exceeding the container memory limit (1GB) and keras has a tendency to exceed that limit. So I think that's the root cause.
This is happening when you execute your code or when the project loads?
Sean