NVIDIA / nvidia-resiliency-ext
NVIDIA Resiliency Extension is a python package for framework developers and users to implement fault-tolerant features. It improves the effective training time by minimizing the downtime due to failures and interruptions.
☆136Updated last week
Alternatives and similar repositories for nvidia-resiliency-ext:
Users that are interested in nvidia-resiliency-ext are comparing it to the libraries listed below
- NCCL Fast Socket is a transport layer plugin to improve NCCL collective communication performance on Google Cloud.☆116Updated last year
- Applied AI experiments and examples for PyTorch☆256Updated 3 weeks ago
- A library to analyze PyTorch traces.☆366Updated this week
- NCCL Profiling Kit☆129Updated 9 months ago
- ☆64Updated 3 months ago
- ☆198Updated this week
- Perplexity GPU Kernels☆185Updated last week
- NVIDIA Inference Xfer Library (NIXL)☆242Updated this week
- 🚀 Collection of components for development, training, tuning, and inference of foundation models leveraging PyTorch native components.☆192Updated this week
- Microsoft Collective Communication Library☆342Updated last year
- A performant, memory-efficient checkpointing library for PyTorch applications, designed with large, complex distributed workloads in mind…☆156Updated 4 months ago
- ☆197Updated 9 months ago
- RDMA and SHARP plugins for nccl library☆187Updated this week
- Zero Bubble Pipeline Parallelism☆381Updated this week
- Synthesizer for optimal collective communication algorithms☆105Updated last year
- MSCCL++: A GPU-driven communication stack for scalable AI applications☆330Updated this week
- This repository contains the experimental PyTorch native float8 training UX☆222Updated 8 months ago
- extensible collectives library in triton☆84Updated last week
- Fast low-bit matmul kernels in Triton☆285Updated this week
- ☆76Updated 5 months ago
- A low-latency & high-throughput serving engine for LLMs☆337Updated 2 months ago
- Microsoft Collective Communication Library☆65Updated 4 months ago
- ☆95Updated last month
- Cataloging released Triton kernels.☆213Updated 3 months ago
- NVIDIA NCCL Tests for Distributed Training☆88Updated this week
- QuickReduce is a performant all-reduce library designed for AMD ROCm that supports inline compression.☆24Updated 3 weeks ago
- Dynamic Memory Management for Serving LLMs without PagedAttention☆345Updated 2 weeks ago
- PArametrized Recommendation and Ai Model benchmark is a repository for development of numerous uBenchmarks as well as end to end nets for…☆134Updated this week
- This is a plugin which lets EC2 developers use libfabric as network provider while running NCCL applications.☆167Updated this week
- ☆36Updated 4 months ago