soheil-mp/nlp-project-conversational-based-agent

NLP Project Conversational Based Agent

In this project, students will build an end-to-end voice conversational agent, which can take a voice input audio line, and synthesize a response. The chatbot agent will be executed locally on a computer.

Environment setup

Create conda environment with python 3.7

conda create --name chatbot python=3.7

Activate the environment

conda activate chatbot

Install project requirements

pip install -r requirements.txt

Install the conda environment for use as a Jupyter kernel

python -m ipykernel install --user --name chatbot --display-name "chatbot"

nlp-project-conversational-based-agent's Issues

Feature request: Add seq2seq model evaluation

create train dev test split
add model evaluation (loss function, word error rate - NLTK edit distance )

Feature request: Add attention mechanism

perform error analysis for the seq2seq model

error analysis sections to answer

• where is the model making errors?
• how can we improve the model?
• do we need more data or better encoding / decoding?

Feature request: Add GloVe and word2vec embeddings

Feature request: wrap prediction in flask API

add additional sequence models

add experimentation sections, including hyperparameter tuning, architecture changes (bi-directional?), different layer types, optimizers