open-source-ai

/open-source-ai77

currently most excited about:
- Deep RL / RL in general
- World Models & training policies in sim
- end-to-end learning
- sim2real
- training diffusion / AC / PPO policies for robotics purposes
I’ve been experimenting with ollama and LMStudio to run open source LLMs on my Mac locally.

any other recommendations for apps? any Obsidian extensions or integrations that you like?
Welcome @nintynick.eth !
Nick has been working on @hatsprotocol through an AI lens from day one. They are ai16z partners.
Arguably the best meeting transcription app, Granola, now has an open source fork.

https://warpcast.com/mikedemarais.eth/0x08456118

Has anyone tried and compared it yet?
Sharing this bounty here in case someone is interested in replying

https://warpcast.com/pirosb3/0xc7414d0d
2025 is the Year of Open Source AI
Request posting permissions below
current status installing phi4. A small modal that reportedly performs as well as chatgpt-4o
Just invited the following casters to the open source ai stage:

@shoni.eth @adam- @sdv.eth @noun839.eth @bias @july

Who am I missing?
My favorite AI safety researcher left OpenAI this week because he couldn’t do the research he wanted to do there.

So the prophecy is fulfilled.
“Only the free world can beat the firewall”
- @balajis.eth on the ongoing open vs closed AI debate

https://x.com/balajis/status/1793070814671290586?s=46
Open source alternative to ChatGPT chatbots. One major advantage of going this route over OpenAI's is that you can share them without the end user having to upgrade their account to use it.
https://github.com/AyaanZaveri/cognite
rem, an open source alternative to rewind dot ai. Currently in alpha, I can see this gaining traction with more awareness over time.
Recording things locally will always be prioritized by the security cautious, not to mention act as a moat against similar cloud based offerings.

https://github.com/jasonjmcghee/rem
Roon is trying to argue OSS will not figure out AGI.

https://twitter.com/tszzl/status/1737986906229768273

I don’t see why OSS cannot have “rapid innovation and research novelty and compute resources”
🚨📄

“PowerInfer: Fast Large Language Model Serving with a Consumer-grade GPU”

https://ipads.se.sjtu.edu.cn/_media/publications/powerinfer-20231219.pdf
This is a channel for actually open, collective AI and the advancements that will help us achieve it.