The trend has been great. I hope she sustains this momentum and I think she's got the right team to do it. The variations are probably due to how the different pollsters apply handicaps to the data.
I looked at the methodology for Marist, though, and while I expected it to be in the neighborhood of +2 to +3 Republican respondents it was actually +3 Dems (Poll of 1,754 Pennsylvania Adults, among them 1663 registered voters and among them 1476 likely voters).
Of the Trump voters, 90% strongly support him and 1% might vote differently, while for Harris voters 91% strongly support her and 0% might vote differently.
Of the self-identified Democrats, 4% are voting Trump; of the self-identified Republicans, 3% are voting Harris.
Of the self-identified Independents, Trump leads 49-45 (in a 2020 PA exit poll, Biden won independents 52-44)
https://www.cnn.com/election/2020/exit-polls/president/pennsylvania
Of the respondents who believe the most important quality is change, Trump leads 60-38.
Harris leads in Philadelphia 83-15, +68
Harris leads in Philadelphia suburbs 59-40, +19 (that's a touch less than what Biden did in 2020)
Trump leads NEPA 51-47, +4 (Trump's margins there were much higher in 2020, although it's hard to know how Marist was defining NEPA)
Trump leads Central PA 62-35, +27 (Trump's Central PA margins in 2020 were generally +30)
Trump leads West PA 56-43, +13 (Hard to know how Marist is defining this, but his margins in rural West PA counties ranged from +18 to +52 and he lost Erie and Allegheny coutnies)
Trump leads with whites +3 (in 2020 he won whites +15)
Trump leads with men +10 (it was +11 in 2020)
Harris leads with women +12 (Biden had them +11 in 2020)
4% of Biden 2020 voters are supporting Trump 2024, while 3% of Trump 2020 voters are supporting Harris 2024.