@mmitchell_ai @BayesForDays @haldaume3 as soon as…

@mmitchell_ai @BayesForDays @haldaume3 as soon as you have >1 timescale on which you’re simultaneously training, need to cope.