A Little Guide to Learning Distributed Algorithms for LLMS Training and Inference [D]

Distributed Training and Inference both involves having a fundamental understanding of how distributed systems work in general

• Distributed Parallelism
• Tensor Parallelism
• Pipeline Parallelism
• Model Parallelism

Reading and reading and reading or even worse, not knowing where to start ;(

That’s boring!

We want to read what’s just needed and quickly get started with applications and that’s what exactly what I have for you all today

Here’s the list of a few initial papers I have read for the past three months that is enough to understand

And a few basics too!

Read them Code them Play with them

I have implemented a few at basic level which you could use as a reference too (the repo is a bit all over the place but I actively trying to maintain and love your feedback too)

Link: https://alphaxiv.org/shared/folder/019de088-28f7-7f02-acd4-c22459fe153e

Gh repo: https://github.com/YuvrajSingh-mist/smolcluster