justheuristic

prefetch_generator

Simple package that makes your generator work in background thread

as_a_service

A simple module that turns your batch-parallel function into a background service. Useful for RL/DL experiments.

chat.petals.ml

Chatbot web app + HTTP and Websocket endpoints for LLM inference with the Petals client

dbg3

justheuristic.github.io

bigscience

Codebase for the project

bitsandbytes

8-bit CUDA functions for PyTorch

bloom-demo

Early dev prototype for decentralized bloom. Not for public eyes yet.

CALM

discord-banners

A reference for the Official Discord banners

flexfork

WIP version for playing around with flexflow

gpt-neox

An implementation of model parallel GPT-3-like models on GPUs, based on the DeepSpeed library. Designed to be able to train models in the hundreds of billions of parameters or larger.

hivemind-lightning

junk

assorted stuff that i worked on and had to show to someone. nothing of interest. shoo!

Machine-Translation

Megatron-LM

Ongoing research training transformer models at scale

model_card

nirvana_cifar_10_example

peft

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

prodige

A supplementary code for Beyond Vector Spaces: Compact Data Representation as Differentiable Weighted Graphs.

py-multiaddr

multiaddr implementation in Python

python-autoclass

A python 3 library providing functions and decorators to automatically generate class code, such as constructor body or properties getters/setters along with optional support of validation contracts on the generated setters. Its objective is to reduce the amount of copy/paste code in your classes - and hence to help reducing human mistakes :).

recommonmark

A markdown parser for docutils

revlib

Simple and efficient RevNet-Library for PyTorch with XLA and DeepSpeed support and parameter offload

rpcudp

Python library for RPC over UDP

sahajbert

shields

Concise, consistent, and legible badges in SVG and raster format

SpQR

transformers

🤗Transformers: State-of-the-art Natural Language Processing for Pytorch, TensorFlow, and JAX.

xformers

Hackable and optimized Transformers building blocks, supporting a composable construction.