Ruoming Pang (@ruomingpang) 's Twitter Profile
Ruoming Pang

@ruomingpang

Apple Foundation Models

ID: 109550268

calendar_today29-01-2010 12:46:19

111 Tweet

2,2K Followers

1,1K Following

Haotian Sun (@haotiansun014) 's Twitter Profile Photo

✨ Introducing EC-DIT: a Mixture-of-Experts (MoE) model with adaptive computation, scaling diffusion transformers up to 97B parameters. It improves training, alignment, and generation quality, with a GenEval score of 71.68%, while keeping fast inference

✨ Introducing EC-DIT: a Mixture-of-Experts (MoE) model with adaptive computation, scaling diffusion transformers up to 97B parameters. It improves training, alignment, and generation quality, with a GenEval score of 71.68%, while keeping fast inference
Haoxuan You (@xyouh) 's Twitter Profile Photo

Looking for a 2025 summer research intern, in the Foundation Model Team at Apple AI/ML, with the focus of Multimodal LLM / Vision-Language. Phd preferred. Apply through jobs.apple.com/en-us/details/… Also email me your resume to [email protected]! 😊

Jing-Jing Li (@drjingjing2026) 's Twitter Profile Photo

1/3 Today, an anecdote shared by an invited speaker at #NeurIPS2024 left many Chinese scholars, myself included, feeling uncomfortable. As a community, I believe we should take a moment to reflect on why such remarks in public discourse can be offensive and harmful.

1/3 Today, an anecdote shared by an invited speaker at #NeurIPS2024 left many Chinese scholars, myself included, feeling uncomfortable. As a community, I believe we should take a moment to reflect on why such remarks in public discourse can be offensive and harmful.
Bairu Hou (@hou_bairu) 's Twitter Profile Photo

Can LLMs learn to activate only the most suited parameters based on task descriptions? 🤔 Introducing Instruction-Following Pruning – dynamically prune LLMs based on the user prompt. This is like having a smaller, input-specific LLM for every input! By activating 3B parameters,

Can LLMs learn to activate only the most suited parameters based on task descriptions? 🤔

Introducing Instruction-Following Pruning – dynamically prune LLMs based on the user prompt. This is like having a smaller, input-specific LLM for every input!

By activating 3B parameters,
Ruoming Pang (@ruomingpang) 's Twitter Profile Photo

With the rapid advancement of AI, I have been reflecting on which aspects of the human mind cannot be easily replaced by machines. I believe one answer is free will: the desires, decisions, and determination to pursue goals not out of necessity, but simply “because it’s there.”

With the rapid advancement of AI, I have been reflecting on which aspects of the human mind cannot be easily replaced by machines. I believe one answer is free will: the desires, decisions, and determination to pursue goals not out of necessity, but simply “because it’s there.”
Ruoming Pang (@ruomingpang) 's Twitter Profile Photo

In Ithaca NY this weekend. Please DM me if you happen to be here and would like to chat about Apple’s foundation model efforts.

In Ithaca NY this weekend. Please DM me if you happen to be here and would like to chat about Apple’s foundation model efforts.
Richard Wei (@rxwei) 's Twitter Profile Photo

On behalf of the whole team I'm so proud to introduce the Foundation Models framework, an API to access our on-device LLM! Check out the Platforms State of the Union for an introduction and 4 sessions later today! developer.apple.com/news/?id=us98z… #WWDC25

On behalf of the whole team I'm so proud to introduce the Foundation Models framework, an API to access our on-device LLM!  Check out the Platforms State of the Union for an introduction and 4 sessions later today!
developer.apple.com/news/?id=us98z…

 #WWDC25
Pallav Agarwal (@pallavmac) 's Twitter Profile Photo

I just built Apple Intelligence Chat 🚀. It uses the new Foundation Models Framework to let you chat offline with the Apple Intelligence model on iOS 26. Let me know if you'd like the code and i'll make it open source. And yes the UI uses Liquid Glass 🧊

I just built Apple Intelligence Chat 🚀. It uses the new Foundation Models Framework to let you chat offline with the Apple Intelligence model on iOS 26.

Let me know if you'd like the code and i'll make it open source. And yes the UI uses Liquid Glass 🧊