Don't know if this one has been posted yet, but NBC's election page is tracking early voting, broken down by age group, party affiliation, and gender.
I try to avoid drawing too many conclusions from early voting, so let's just call this speculation for the hell of it. Take it with a huge grain of salt.
For example, Pennsylvania - link
Comparing mail in ballot requests vs. combo mail-in returned and in-person early voting, the Dems request 63% of the ballots, but have 68% of MI+EV have been Dem. That seems like a good sign, as it could indicate a better Dem turnout on MI votes. Or it could just indicate that a higher percentage of Dems are doing EV, even relative the percent that requested MI. But overall, a good sign.
Looking at the gender breakdown, 56% women is a good sign. But it may just reflect more women voting early, rather than voting in total.
The biggest head scratcher to me is the total MI+EV compared to their 2016 totals. They are less than 36% of the total 2016 vote. Now, that is an increase of 1400% from MI+EV in 2016, so maybe it is just a massive pain in the ass to do so in Penn. But, (speculating) it may show a lack of effectiveness in the Dems pushing MI+EVs in the state, which is concerning for their effectiveness in total. And certainly mailing issues have been a problem.