For Feedbacks | Enquiries | Questions | Comments - Contact us @ innovationmerge@gmail.com
What?
- Advanced technologies are making the way of writing code easy.
- Natural Language Processing(NLP), Named Entity Recognition(NER) are base to understand developer queries and extract intent out of it.
Why?
- There will be normal queries which Developers use frequently during Exploratory data analysis.
- Instead of writing the syntax for each query, Developers can now use normal English similar to a comment and generate a Python code.
How?
- Found a tweet from Deepak Rawat
@kartik_godawat and I have built a ready-to-install @ProjectJupyter extension which converts English queries into relevant python code. Inspired by GPT-3, built using @TensorFlow Universal Sentence Encoder and @spacy_io NER. https://t.co/vwVveWuPII #DataScience #MachineLearning pic.twitter.com/hOYzkvfuAw
— Deepak (@deepak_s_rawat) September 7, 2020 - With the inspiration of OpenAI GPT-3, Kartik Godawat and Deepak Rawat have built a ready to install Jupyter extension named Text2Code, which converts English queries into relevant python code using Tensorflow Universal Sentence Encoder and Spacy NER.
Related Article:
Software’s Required:
- Python 3.6
- Browser
Network Requirements
- Internet to download packages
Implementation
- Blog post explains the implementation using open source libraries.
- Relevant python code is generated by extracting the entities from user input text, identifying, matching with predefined intents and filling the template.
- Text2Code Jupyter extension is packaged with required frontend which loads with Jupyter notebook for user interaction and created with below process
- Generating training data
- Intent Matching
- NER(Named Entity Recognition)
- Fill Template
Demo from DeepKlarity
