fattorib / zero-transformer Goto Github PK
View Code? Open in Web Editor NEWTwo implementations of ZeRO-1 optimizer sharding in JAX
License: MIT License
Two implementations of ZeRO-1 optimizer sharding in JAX
License: MIT License
Wandb log:
37129af4df4360d64b8518d4da9fe45da8c71df9
with pmap scriptWill run with smaller model to see if same behavior is occurring, possible sources of error:
Port over your code from Little-GPT and add extraction/conversion methods
Current attention implementation in src/layers.py
doesn't keep separate attention heads sharded during computation. I believe this is due to the attn_out = attn_out.reshape(B, T, C)
line.
Can probably look at HF, MTJ or another flax implementation for a better way to implement this
Similar timing code to benchmark_pmap.py
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.