baseline

Update main.py with the version of the network that trains in under 1…

…0 seconds on most A100s

Update Readme.md with clarification

Previously it seem to imply we were aiming for the world record of <~12.31-12.38 seconds, when we currently have it.

Merge pull request tysam-code#1 from tysam-code/cleaner-logging

Updated the logging to better align with repo goals, README.md