# CloneBot: Personalized Dialogue-Response Predictions

arXiv: https://arxiv.org/abs/2103.16750 · HTML: https://mlsanity.com/paper/2103.16750/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 7 of its 11 references that are in the map.
- First posted on arXiv 31 Mar 2021.

## Builds on (most cited first, 7 of 7)

- [Dense Passage Retrieval for Open-Domain Question Answering](https://mlsanity.com/paper/2004.04906/) (2020)
- [A Call for Clarity in Reporting BLEU Scores](https://mlsanity.com/paper/1804.08771/) (2018)
- [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://mlsanity.com/paper/1911.00536/) (2019)
- [Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis](https://mlsanity.com/paper/1806.04558/) (2018)
- [TransferTransfo: A Transfer Learning Approach for Neural Network Based Conversational Agents](https://mlsanity.com/paper/1901.08149/) (2019)
- [Gmail Smart Compose: Real-Time Assisted Writing](https://mlsanity.com/paper/1906.00080/) (2019)
- [EmpTransfo: A Multi-head Transformer Architecture for Creating Empathetic Dialog Systems](https://mlsanity.com/paper/2003.02958/) (2020)

## Abstract (arXiv)

Our project task was to create a model that, given a speaker ID, chat history, and an utterance query, can predict the response utterance in a conversation. The model is personalized for each speaker. This task can be a useful tool for building speech bots that talk in a human-like manner in a live conversation. Further, we succeeded at using dense-vector encoding clustering to be able to retrieve relevant historical dialogue context, a useful strategy for overcoming the input limitations of neural-based models when predictions require longer-term references from the dialogue history. In this paper, we have implemented a state-of-the-art model using pre-training and fine-tuning techniques built on transformer architecture and multi-headed attention blocks for the Switchboard corpus. We also show how efficient vector clustering algorithms can be used for real-time utterance predictions that require no training and therefore work on offline and encrypted message histories.