how to use it for only 2 turns? Any ideas?

nouhadziri / THRED

The implementation of the paper "Augmenting Neural Response Generation with Context-Aware Topical Attention"

https://arxiv.org/abs/1811.01063

MIT License

111 stars 25 forks source link

how to use it for only 2 turns? Any ideas? #18

Closed Monica9502 closed 4 years ago

ehsk commented 4 years ago

The Vanilla Seq2Seq and TA-Seq2Seq models can be used for 2-turn dialogues. In the data file, each utterance should be separated by TAB. Note that THRED and HRED need context to generate a response and therefore, cannot be used for 2-turn dialogues.

Monica9502 commented 4 years ago

Got it, thanks.