Professional Documents
Culture Documents
GitHub AI
GitHub AI
huyenchip.com
13-16 minutos
Table of contents
Data
…. How to add missing repos
The New AI Stack
…. AI stack over time
…….. Applications
…….. AI engineering
…….. Model development
…….. Infrastructure
Open source AI developers
1 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
Data
2 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
Feel free to submit the repos with less than 500 stars. I’ll continue
tracking them and add them to the list when they reach 500 stars!
1. Infrastructure
2. Model development
3 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
can develop applications on top of them. This is the layer that has
seen the most actions in the last 2 years and is still rapidly
evolving. This layer is also known as AI engineering.
4. Applications
2. Most low-hanging fruits have been picked. What is left takes more
effort to build, hence fewer people can build them.
4 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
In 2023, the layers that saw the highest increases were the
applications and application development layers. The
infrastructure layer saw a little bit of growth, but it was far from the
level of growth seen in other layers.
Applications
5 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
AI engineering
6 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
• Bots via chat apps like Slack, Discord, WeChat, and WhatsApp.
AIE framework is a catch-all term for all platforms that help you
develop AI applications. Many of them are built around RAG, but
many also provide other toolings such as monitoring, evaluation,
etc.
7 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
Model development
8 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
Infrastructure
Open source software, like many things, follows the long tail
distribution. A handful of accounts control a large portion of the
repos.
845 repos are hosted on 594 unique GitHub accounts. There are
20 accounts with at least 4 repos. These top 20 accounts host 195
of the repos, or 23% of all the repos on the list. These 195 repos
have gained a total of 1,650,000 stars.
9 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
10 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
11 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
1 million commits
12 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
It’s been known for a long time that China’s AI ecosystem has
diverged from the US (I also mentioned that in a 2020 blog post).
At that time, I was under the impression that GitHub wasn’t widely
used in China, and my view back then was perhaps colored by
China’s 2013 ban on GitHub.
While in the US, many research labs have moved away from the
RNN architecture for language models, the RNN-based model
family RWKV is still popular.
13 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
One pattern that I saw last year is that many repos quickly gained
a massive amount of eyeballs, then quickly died down. Some of
my friends call this the “hype curve”. Out of these 845 repos with
at least 500 GitHub stars, 158 repos (18.8%) haven’t gained any
new stars in the last 24 hours, and 37 repos (4.5%) haven’t gained
any new stars in the last week.
14 de 15 03/18/2024, 9:17 a. m.
What I learned from looking at 900 most popular open source AI tools about:reader?url=https%3A%2F%2Fhuyenchip.com%2F2024%2F03...
• Seemingly niche tools that solve one problem really well, such as
einops and safetensors.
Conclusion
15 de 15 03/18/2024, 9:17 a. m.