An implementation of model parallel autoregressive transformers on GPUs, based on the DeepSpeed library.

Training LLMs with QLoRA + FSDP

A Gym-like environment for Reinforcement Learning in Rocket League

Easily create large video dataset from video urls

library to read/write .npy and .npz files in C/C++

Android view for displaying PDFs rendered with PdfiumAndroid

A repo for distributed training of language models with Reinforcement Learning via Human Feedback (RLHF)

OSLO: Open Source for Large-scale Optimization

Phind API reverse-engineered implementation

Provides a minimal implementation to extract FLAN datasets for further processing

Download a large file from Google Drive (curl/wget fails because of the security notice). Change max downloadable files to 100

DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.

An eye-catching developer Portfolio, Built on NextJS.

Used for adaptive human in the loop evaluation of language and embedding models.

Data augmentation for NLP