vaibtan's Projects
š§āš« 60 Implementations/tutorials of deep learning papers with side-by-side notes š; including transformers (original, xl, switch, feedback, vit, ...), optimizers (adam, adabelief, sophia, ...), gans(cyclegan, stylegan2, ...), š® reinforcement learning (ppo, dqn), capsnet, distillation, ... š§
A subset of PyTorch's neural network modules, written in Python using OpenAI's Triton.
Code for the Assignment given in CSE515 Bayesian Machine Learning
RISCV RV32i Simulator and Assembler
Generate python ctypes classes from C headers. Requires LLVM clang
Material for cuda-mode lectures
DeepSpeed is a deep learning optimization library that makes distributed training easy, efficient, and effective.
Implemented a database storing string key-value pairs using Raft Consensus Algorithm, ensuring consistent data replication and fault recovery across the distributed network of nodes. Obtained Linearizable Reads without paying high latency penalties for each request by utilizing the leader lease mechanism.
A full stack website for an app which lets users order delicacies from various cuisines, with features including user authentication, registration, profile management; sort, filter, pagination for search results page; cart management, stripe payment checkout and real-time order status.
Simple and efficient pytorch-native transformer text generation in <1000 LOC of python.
An implementation of model parallel autoregressive transformers on GPUs, based on the DeepSpeed library.
My solutions to the problems sets of Haskell MOOC offered by the University of Helsinki
File compression library | Shennon | Fano | Huffman | Bigram Huffman | Adaptive Huffman | Arithmetic coding
Implementation of the Simplex Algorithm without degenerated case
LLM inference in C/C++
Inference Llama 2 in one file of pure C
An ecosystem of Rust libraries for working with large language models
LLM training in simple, raw C/CUDA
Implementing a ChatGPT-like LLM from scratch, step by step
A collection of lock-free data structures written in standard C++11
Modeling Stock Market Returns using Statistical, Classical Machine Learning and Deep Learning Models trained on CNNPred Dataset
Minimal, clean code for the Byte Pair Encoding (BPE) algorithm commonly used in LLM tokenization.
[ACL'19] [PyTorch] Multimodal Transformer
Open Source Computer Vision Library
3-D portfolio made using ThreeJS and React
ā” Competitive Programming Library