The active_learning_in_nlp from abinayam02

active_learning_in_nlp's Introduction

Vanakkam Makkale 🙏

I'm Abinaya Mahendiran, CTO, Nunnari Labs and Part-time Program Manager, IITM.

What do I do?

🔭 I build end-to-end NLP and MLOps products.
👯 I’m looking to collaborate on open-source applied/research projects focusing on NLP.
👉Some of my open source contributions can be found here:
📚 I like to teach so I write techincal articles at and give talks at many events.
💬 Ask me about NLP/NLU, NLG, MLOps, and Architecting ML systems.
😄 Pronouns: She/Her
⚡ Fun fact: Other than coding, I love cooking (secretly aspiring to be a chef someday!) and gardening.

How to contact me?

📫 How to reach me:
🧑‍🏫 Services provided: I do provide a variety of paid services. Do book a call on

Github Stats

Technologies

Python

Jupyter

TensorFlow

Pytorch

Scikit Learn

HuggingFace

FastAPI

Docker

Thank you for visiting my profile!

active_learning_in_nlp's People

Contributors

Stargazers

Watchers

active_learning_in_nlp's Issues

Decide NLP task

Two of us together can invest around 80 hours on this project, so scoping the task is important

Sentiment classification, or
NER classification

Should we go by the recommendation given the course project guideline, and take a dataset which is labelled, but throw away the label for around 75% of the data. Train model on remaining 25% of the data, and use the model to recommend items to be tagged. We will use english language data. We will mostly try to take data from hugging face.

Finalize baseline

Finalize metric

Look and feel of final app

Dash, or Streamlit
Display the result, while showing how things are working, or
Have it more like a product, where end user can choose the model, dataset and strategy for getting items to annotate ? Can be stretch goal

Recommend Projects