We all have been there, tinkering around with models is fun but we rarely do it with research precision and issues are often subtle and hard to reproduce. There are a lot of benchmarks but running them isnt viable often. What I am looking for: A test that does not take too much time (30mins to 1h ma
