zigzagcai Goto Github PK

alpa

Training and serving large-scale neural networks with auto parallelization.

apex

A PyTorch Extension: Tools for easy mixed precision and distributed training in Pytorch

colossalai

Making large AI models cheaper, faster and more accessible

cutlass

CUDA Templates for Linear Algebra Subroutines

deepspeed

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

flash-attention

Fast and memory-efficient exact attention

gloo

Collective communications library with various primitives for multi-machine training.

internevo

jax

Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more

llm.c

LLM training in simple, raw C/CUDA

llvm-project

The LLVM Project is a collection of modular and reusable compiler and toolchain technologies.

mamba

megatron-lm

Ongoing research training transformer models at scale

nanogpt

The simplest, fastest repository for training/finetuning medium-sized GPTs.

nccl

Optimized primitives for collective multi-GPU communication

nemo

NeMo: a toolkit for conversational AI

nemo-megatron-launcher

NeMo Megatron launcher and tools

open-gpu-kernel-modules

NVIDIA Linux open GPU kernel module source

patchelf

A small utility to modify the dynamic linker and RPATH of ELF executables

rwkv-lm

RWKV is an RNN with transformer-level LLM performance. It can be directly trained like a GPT (parallelizable). So it's combining the best of RNN and transformer - great performance, fast inference, saves VRAM, fast training, "infinite" ctx_len, and free sentence embedding.

triton

Development repository for the Triton language and compiler

vllm

A high-throughput and memory-efficient inference and serving engine for LLMs

xla

A machine learning compiler for GPUs, CPUs, and ML accelerators