A ready-to-install jupyter extension which converts english queries into relevant code. Built as a proof of concept for the problem we personally face of forgetting less-used syntaxes of pandas and plotly libraries which is often used in Exploratory Data Analysis. This tool allows us to query in a generic language without having to remember the syntax.

Inspired by cool GPT-3 demos and not having the access to the API, here is an attempt by us to build a Text2Code extension using publicly available libraries. The approach isn't generative, but relies on identifying & matching from a set of predefined intents and generating the relevant code by extracting relevant entities and inserting them in a template. Adding new intents and extending the functionality is easy once the pipeline is in place.

Technologies used:

  • Universal Sentence Encoder
  • Spacy
  • Faiss
  • Jupyter extension

Don't forget to tag @dk-crazydiv , @deepak-deepklarity in your comment, otherwise they may not be notified.

Authors original post
Share this project
Similar projects
WExDA - Web-based Data Exploration Tool
WExDA is a web-based data exploration tool made with streamlit. It can be used for Exploratory Data Analysis.
New layout options for Streamlit
Introducing new layout primitives - including columns, containers and expanders!
Missingno: Missing data visualization module for Python.
Missingno provides a small toolset of flexible and easy-to-use missing data visualizations.
Beginner's Guide to Altair Visualization
Getting started with Visualization using Altair on Kaggle with this simple tutorial.
Top collections