Hi Aline,
So, I trained a classifier for this region just a few weeks ago using the SILVA 138-99 full length sequences (2020.8) using these primers: 27FYM & 519R, but do want to let you know that I had some issues with classifications for this region that we ultimately think stemmed from subpar DNA extraction/amplification methods.
You can find the post here if you would like more info. But you're welcome to try this classifier if it could work for your purposes while waiting for someone else to respond.
Here are the silva links used to train the classifier:
wget https://data.qiime2.org/2020.8/common/silva-138-99-seqs.qza
wget https://data.qiime2.org/2020.8/common/silva-138-99-tax.qza
Also, the qza file is too large to upload here. But if this would work for you, I can send you a google drive link to download it if you feel comfortable sending a direct message to me.
Wishing you all the best.