nicolas-ivanov/neural_chatbot_papers

Description of state-of-the-art approaches for neural dialogs generation

neural_chatbot_papers

Description of state-of-the-art approaches for neural dialogs generation

Tokens encodings

Sentence level

• BERT
• ELMO

Tokens level

• FastText
• Word2Vec
• Glove

Sentence encodings

Transformers

• BERT
• OpenAI GPT

RNN

• Bidirectional LSTM

CNN

• ?

DAN

Generative models

Encoder

See "Sentence encodings"

Mapper

Decoder

• Softmax
• Von Mises-fisher Loss

Retrieval models

• see "Sentence encodings"
• cross-attention

Reranking models

Memory networks

Attention over external database.

---

Sources of inspiration:

• EMNLP'18
• Deep Chit-Chat: Deep Learning for ChatBots by Microsoft
• NeurIPS'18
• Neural Network Dialog System Papers by @snakeztc
• ML Writing Month: November 2018ML Writing Month: November 2018 by Cody Wild
• Dialog systems by @rodart