Ishan Kumar
  • About Me
  • Blogs
  • Projects
  • Experience
  • CV
  • Courses
Navigation bar avatar

Projects


...

Recent Projects

Mechanistic Interpretability of LLMs

Leveraging Sparse Autoencoders to understand the learning process of an LLM

2025-02-01

LabCast

A multi-agent system turning research papers into conversational podcasts

2025-02-01

MQA Adapt - Adapting MQA for efficient Inference

Layerwise Adaptive Multi-Query Attention for efficient inference

2024-12-01

Older Projects (from Undergraduate)

Open Source Contributions

A summary of the Open Source projects I have been involved with

2023-02-01

Improving Mini BERT’s predictions using Multilingual Knowledge Distillation

Improving mini BERT's performance on Hindi COVID Fake News pred without using Parallel Finetuning data

2021-02-01

Reddit Flair Detector

Model which predicts the flair of a reddit post from its contents

2021-02-01

Molecular VAE

PyTorch implementation of the paper "Automatic Chemical Design Using a Data-Driven Continuous Representation of Molecules"

2021-01-01

RL 2048

Implementation of Deep Q-network to play the game 2048 using Keras.

2020-02-01

Neural Question Generation

This is an attempt to compare and test various techniques used in seq2seq modelling, on a question generation task

2020-02-01

d2l-pytorch

Dive Into Deep Learning Book with PyTorch Implementation

2019-05-01

  • Email me
  • Facebook
  • GitHub
  • Twitter
  • LinkedIn
  • Google Scholar
  • Medium

Ishan Kumar  •  2025

Powered by Beautiful Jekyll