Logging is a bit more stable (i.e. no crash bugs when using a normal naked 'print' inside of the main training loop, which seems generally user-hostile). Logging is also now, I believe, more hackable/straightforward and does not require an external library (rich). Unfortunately, this comes at the expense of some long lines and a more spartan printout.
I also updated README.md to be a bit less verbose. That's something I surely love doing.
Logging is a bit more stable (i.e. no crash bugs when using a normal naked 'print' inside of the main training loop, which seems generally user-hostile). Logging is also now, I believe, more hackable/straightforward and does not require an external library (rich). Unfortunately, this comes at the expense of some long lines and a more spartan printout.
I also updated README.md to be a bit less verbose. That's something I surely love doing.