Cross-backend differentiable Gaussian splatting library with precomputed projections and parallel rendering for integration into DL Models — no manual gradients required.

Explore Abhas Kumar Sinha's journey in Machine Learning Engineering through projects and insights on NLP, Computer Vision, and AI methodologies.

A Keras-based motion transfer model leveraging first-order Jacobian-based motion estimation to animate a static image using keypoints, inspired by Aliaksandr Siarohin's first-order motion model.

DA 623: Computing with Signals course project.

"Dispel the ignorance of the illusionary world; transcend the game engine!": presenting a Game Engine written on Python, extending support across multiple GPU devices with 3D rendering, PBR Rendere…

Corpus2GPT: A project enabling users to train their own GPT models on diverse datasets, including local languages and various corpus types, using Keras and compatible with TensorFlow, PyTorch, or J…

MinimalGPT is a concise, adaptable, and streamlined code framework that encompasses the essential components necessary for the construction, training, inference, and fine-tuning of the GPT model. T…

A Keras implementation of Blenderbot, a state-of-the-art open-domain chatbot model, supporting TensorFlow, PyTorch, and JAX backends seamlessly.

RLHF – Reinforcement Learning from Human Preferences. Models are fine tuned using RL from human feedback. This repository completes the implementation of ChatGPT from Transformer.

This repository presents a Python-based implementation of the Transformer architecture on Keras TensorFlow library, as proposed by Vaswani et al. in their 2017 paper "Attention is all you need."

A language translator based on a very simple NLP Transformer model, backed by encoder, decoder and a Bahdanau Attention Layer in between, implemented on TensorFlow.

An implementation of classical GRU (Cho, el at. 2014) along with Optimized versions (Dey, Rahul. 2017) on TensorFlow that outperforms Native tf.keras.layers.GRU(units) implementation of Keras.

A simple well-documented tutorial on implementing a 1D GAN on Keras using a Python Jupyter Notebook

A classic GAN that generates Images: A study on problems like Mode Collapse.

A very simple and plain DC GAN to generate Image Flower Pictures out of the dataset.

A word2vec embedding, called Continuous Skip Gram Model encodes each word to a Vector depending on how closely they are related or how often they are used along.

Used to extract all the data from twitter and print it to the Jupyter Notebook