Anupam Nautiyal

LoomFlow

May 6, 2026 – Present

LoomFlow — GitHub repository

TransformerIsAllYouNeed

August 13, 2024 – January 12, 2025

This repository contains code for two key projects: Encoder-Decoder Architecture: Implements the Transformer architecture described in "Attention Is All You Need." Custom BERT Replica: A customized version of the BERT-base-uncased model that loads pre-trained weights from Hugging Face.

FineTuningLLMsWithDPO

February 16, 2024 – February 18, 2024

This project discusses the complete pipeline, of creating an end-to-end system of finetuning the LLMs, with PEFT STTF and DPO. This project explanation is also available in the medium article which can be accessed through this link https://medium.com/dev-genius/how-to-harness-peft-sftt-and-dpo-to-fine-tune-llms-394e9cd0b150.

DVC

December 11, 2022 – December 12, 2022

This repository show the very basic music recommendation system. For this task we have used DVC to store the heavy data into the remote location while keep the code files in github repository.

Sound-Reconstruction-Using-VAE

May 17, 2022 – May 20, 2022

In this task I tried to reconstruct the music using variational auto encoder. VAE consist of encoder and decoder network which provide a provide probabilistic manner for describing the latent space.

DQN-Reinforcement-Learning

March 22, 2021 – March 23, 2021

DQN is the first deep learning model to successfully learn control policies directly from high-dimensional sensory input using reinforcement learning. The model is a convolutional neural network, trained with a variant of Q-learning, whose input is raw pixels and whose output is a value function estimating future rewards. They apply DQN to seven Atari 2600 games from the Arcade Learning Environment, with no adjustment of the architecture or learning algorithm. DQN outperform all the previous approches.

Image-captioning

January 20, 2021 – November 18, 2021

Image Captioning is the process of generating textual description of an image. It uses both Natural Language Processing and Computer Vision to generate the captions. So in this work I try to implement neural network that are capable of generating text given an image.

Chinese-Word-Segmentation

January 6, 2021 – January 10, 2021

Chinese word segmentation is a necessary first step in Chinese language processing and there are many approach to solve this problem of CWS one of the approach is using neural network, In this paper I will discuss about implementation, Preprocessing, Network Architecture, Hyperparameter and result.

Traffic-Sign-Classification

December 28, 2020 – December 28, 2020

In this work I make a small model for classification of different signs on road for example stop sign, walking sign, no entry sign etc. This work is not based on any paper instead inspired by udacity nano degree program for self driving car. The blog post for this is available in Medium artical of udacity self driving car, you can check that out.

Cyclic-GAN

July 31, 2019 – December 22, 2020

Image-to-image translation is a class of vision and graphics problems where the goal is to learn the mapping between an input image and an output image using a training set of aligned image pairs. However, for many tasks, paired training data will not be available. Cycle-Consistent Adversarial Network is an approach for learning to translate an image from a source domain X to a target domain Y in the absence of paired examples.