ArcticInference: vLLM plugin for high-throughput, low-latency inference

🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.

ArcticTraining is a framework designed to simplify and accelerate the post-training process for large language models (LLMs)

Minimal example scripts of the Hugging Face Trainer, focused on staying under 150 lines

Machine Learning Engineering Open Book

🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.

MII makes low-latency and high-throughput inference possible, powered by DeepSpeed.

Pretrained language model with 100B parameters

Trainable, memory-efficient, and GPU-friendly PyTorch reproduction of AlphaFold 2

Code release for SLIP Self-supervision meets Language-Image Pre-training

Azure HPC/AI VM Images

Library for 8-bit optimizers and quantization routines.

Ongoing research training transformer language models at scale, including: BERT & GPT-2

Distribution transparent Machine Learning experiments on Apache Spark

An implementation of model parallel autoregressive transformers on GPUs, based on the Megatron and DeepSpeed libraries

Guide: Finetune GPT2-XL (1.5 Billion Parameters) and finetune GPT-NEO (2.7 B) on a single GPU with Huggingface Transformers using DeepSpeed

Accelerate your Neural Architecture Search (NAS) through fast, reproducible and modular research.

RDMA and SHARP plugins for nccl library

Example models using DeepSpeed

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

A minimal & modern LaTeX template for your (bachelor's | master's | doctoral) thesis

Find the smallest number of switches necessary to build topologies of a given number of hosts and bisection bandwidth for the EGFT, HyperX, and Jellyfish topologies.