Home Projects Sophia
Sophia
Python

Sophia

The official implementation of “Sophia: A Scalable Stochastic Second-order Optimizer for Language Model Pre-training”

by Liuhong99 · GitHub
Stars
Forks
License
Created
Last commit
Language
deep-learninglarge-language-modelsoptimizerMITPython
View on GitHub
In plain words

Train language models more efficiently using a special optimization method.

You maintain this project?

Claim its page: indexed whatever its rank, translated into six languages, and enriched with what you write yourself.

Claim this page →
Sophia — GitHub preview card
📈 Star history
1 0031 000
2026-07-202026-08-31
📈 Track Sophia

Get an email alert on its next release or when it starts trending — never miss the moment.

Free · no card · unsubscribe anytime
Get email alerts →
📄 About

The official implementation of “Sophia: A Scalable Stochastic Second-order Optimizer for Language Model Pre-training”

Sophia has 1k stars on GitHub. It has been forked 58 times. Sophia is written mainly in Python. It has been in active development since 2023. Sophia is available under the MIT license. Its main topics are deep-learning, large-language-models, optimizer.

Frequently asked questions

What is Sophia?

The official implementation of “Sophia: A Scalable Stochastic Second-order Optimizer for Language Model Pre-training”

Is Sophia open source?

Sophia is an open-source project. It is released under the MIT license.

Is Sophia free?

Yes. Sophia is free and open source — you can use, modify and self-host it.

What license does Sophia use?

Sophia is available under the MIT license.

What language is Sophia written in?

Sophia is written mainly in Python.

🏅 Maintainer of this project?
olud.ai badge — Sophia

Add this live badge to your README — your GitHub stars and directory rank, refreshed daily.

[![olud.ai](https://olud.ai/badge.php?tool=liuhong99-sophia)](https://olud.ai/project/liuhong99-sophia.html)
More badge options →
🧬 Shares DNA with🧬 View the DNA map →

Measured from GitHub topics shared by both projects, weighted by how rare each topic is.