facebookresearch/fairscale

PyTorch extensions for high performance and large scale training.

PythonOther
This is stars and forks stats for /facebookresearch/fairscale repository. As of 20 Apr, 2024 this repository has 2591 stars and 246 forks.

Description FairScale is a PyTorch extension library for high performance and large scale training. This library extends basic PyTorch capabilities while adding new SOTA scaling techniques. FairScale makes available the latest distributed training techniques in the form of composable modules and easy to use APIs. These APIs are a fundamental part of a researcher's toolbox as they attempt to scale models with limited resources. FairScale was designed with the following values in mind: Usability -...
Read on GithubGithub Stats Page
repotechsstarsweeklyforksweekly
buildkite/docsRubySCSSHTML360246+1
rust-nostr/nostrRustMakefileJavaScript2710580
evcxr/evcxrRustJupyter NotebookOther4.7k02050
Netflix/eddaScalaOther55701400
AlexWarnes/svelte-sequenceSvelteTypeScriptCSS28010
epfml/OptML_courseJupyter NotebookTeXPython89602760
z80dev/layerzero-vyper-examplesVyperPythonSolidity14000
NVlabs/VoxFormerPythonCudaC++8020620
UE4SS-RE/RE-UE4SSC++LuaOther251+562+1
NootInc/WhateverRedC++PythonShell1.1k01000