How is 12...Nxg3 a blunder?

4 passed pawns and the bishop pair is an automatic victory, yet the analysis calls it a blunder putting white at +1.32. I consider that knight an amazing sacrifice. I was able to win the game virtually blind folded from that point on.

 I don't think any computer could lose as Black in this position, nor do I think any computer could win as white against a GM or another computer from this position. 

5.....d5 wasn't a very bright move though was it ?

Perhaps the Stockfish8 analysis was set to quick analysis. If it was set to medium analysis or deep analysis it might evaluate the position differently. Perhaps on quick analysis it thinks you are opening files for your opponent towards where your king might want to castle to. That was a very bold sacrifice. Your opponent should have sent you a trophy that reflects just how bold and ambitious of a move you made there. 

You were already 2 pawns up so Nxd2 looks the most sensible, Kxd2 is forced or Be3 wins the rook and you still have the bishop pair.

No d5 wasn't. I purposely left it there because that wasn't evaluted as a blunder, yet it obivously was. But the first move in the series registers a blunder. I still won though lol.