Best attention mechanism for Enc-Dec LSTM model

I implemented Encoder-Decoder LSTM model in Pytorch. Now I want to integrate Attention mechanism in Decoder. Can anyone give the link to Attention model which gives the best result on NLP tasks?

There’s no such a model, no one gives the best results on NLP tasks.
Here’s a good paper about attention mechanisms: https://arxiv.org/pdf/1508.04025.pdf