Haoran (Richard) Cai

Multimodal foundation model @LumaLabsAI. Previously, @AdobeResearch, @UWStat, @uwgenome.

[Haoran (Richard) Cai @Richard_Seattle]

So incredibly proud to be part of the team building Uni-1 @LumaLabsAI!
We built it to be intelligent, cultured, and incredibly directable (look at how perfectly it understands scale and physical interaction here!). But honestly? It is just super fun to play with!

[Luma @LumaLabsAI]

Uni-1 is here! A new kind of model that thinks and generates pixels simultaneously. Less artificial. More intelligent.

[William Shen @shenbokui]

🚀 UNI-1 debuts us as the best lab not named @OpenAI / @GeminiApp. Not bad for our first generation of unified image model!

Interestingly, with the current update, GPT Image 2’s ELO score is now 110 points lower than before. not sure what happened… cc @kenjihata @BoyuanChen0

Sharing a few genuinely GPT-Image-2-level generations from UNI-1 below 🤩

“Generate a news website page from the year 2036, featuring relevant news stories and ad blocks designed not for humans, but for AI agents who have evolved into distinct personalities. Both the website and all the advertisements featured on it should be in English.”

[William Shen @shenbokui]

Most image models are good at one thing.
Uni-1 has been good at everything we've thrown at it.

Our team generated thousands of images leading up to Uni-1 launch. We embedded them all into a single map where visual similarity determines proximity. The result speaks for itself.

[Luma @LumaLabsAI]

We are loving the energy around Uni-1!

Quick note since we’re seeing questions:

With Luma Agents, requests can route across models. If you want to make sure you’re using Uni-1, here’s how:

- Select Create Image → Uni-1
- Or, explicitly ask the agent to use Uni-1
- Check the model label on outputs to confirm

API access coming soon for more direct testing.

Keep the feedback coming, and keep on creating → https://t.co/zjJZMt8Dt6.

[Jiaming Song @baaadas]

Thanks for the amazing team that made this possible!

We saw a significant improvement on the model even in the 2 weeks between the announcement and the launch.

Now, to new heights!

[William Shen @shenbokui]

UNI-1 is intelligent, directable, cultured. Incredible range it can do.

Incredibly proud of the world-class team building a world-class model.
It’s a daunting task to go up against industry giants like Deepmind/OpenAI/Bytedance.

More to come! API, technical report, model card…

Come join us!

[Haoran (Richard) Cai @Richard_Seattle]

Incredible work by the @LumaLabsAI team on Uni-1! A massive part of how a lean team goes toe-to-toe with the giants is having relentless, world-class infrastructure. So hyped to have just joined @shenbokui's team to help build the data infra that powers what's next! 🚀

[William Shen @shenbokui]

Excited to introduce Uni-1, our new multimodal model that *unifies* understanding and generation.

TLDR: a team of ~15 researchers is going pound-for-pound with nano banana and gpt image 🧵

[Jiaming Song @baaadas]

Excited to introduce Uni-1, our new *unified* multimodal model that does both understanding and generation: https://t.co/VkgMNnYtZv

TLDR: I think Uni-1 @LumaLabsAI is > GPT Image 1.5 in many cases, and toe-to-toe with Nano Banana Pro/2. (showcase below)

[Haoran (Richard) Cai @Richard_Seattle]

@ailiefraser Bend looks so beautiful! Look forward to my trip there soon!

[Haoran (Richard) Cai @Richard_Seattle]

@ahmetb Our team @AdobeResearch is hiring k8s expert to work on ML infra powering AI-first video editing on the web! Please talk to me and @aseemaa for more details and check out our job post https://t.co/9KdsoHzgTI

[Haoran (Richard) Cai @Richard_Seattle]

I am particularly interested in Torch Eval library in this @PyTorch release lead by @ninginthecloud and @BobakHashemi! Torch Eval provides rich collection of performant model metrics and they are optimized for distributed training!

[PyTorch @PyTorch]

📚 We’re including new libraries in the release of PyTorch 1.13, including TorchMultimodal, Torch-TensorRT, Torch Eval and TorchSnapshot.

Learn more 👇

[Adobe Research @AdobeResearch]

Project Blink, a web-based AI-powered video editing tool developed by @AdobeResearch, was showcased at this year's @adobemax. It uses #AI to help users find and quickly pull out highlights from video content. You can now sign up for beta access! https://t.co/MXjZVKa0Jx

[Maneesh Agrawala @magrawala]

Adobe's Blink project on transcript-based video editing looks great!
https://t.co/x9VkZpRpZt

[Haoran (Richard) Cai @Richard_Seattle]

Checkout #ProjectBlink demo! | Adobe MAX Sneaks 2022 https://t.co/eU1dF3tY7r

[Haoran (Richard) Cai @Richard_Seattle]

#ProjectBlink was finally announced at @adobemax . I have been working on this for the past few years and I am so proud of the incredible #ProjectBlink team and collaborators at @AdobeResearch and @Adobe Sensei! Sign up for early access NOW!! https://t.co/ZTEz7XiTde

[Adobe @Adobe]

Video editing and trimming is as easy as highlighting text with #ProjectBlink.

[CNET News @CNETNews]

Adobe's Blink AI Tool Makes Video Editing Look Like Word Processing https://t.co/ON41ove0vm

[Haoran (Richard) Cai @Richard_Seattle]

Exciting research direction, impactful product and more importantly, great mentor!

[Ding Li @dingzeyuli]

I am hiring research interns for 2023! If you or you know any student with experience with large language models (CLIP or others) and video understanding for authoring, send them my way! Happy to talk to you/them to see if it's a good fit!

Please retweet to spread the word!

[Aseem Agarwala @aseemaa]

We are expanding our team and efforts in ML-assisted video workflows, and have openings to hire a senior product manager, a senior ML infrastructure engineer, and a senior front-end developer. (1/2)

[Haoran (Richard) Cai @Richard_Seattle]

@oliverwang81 @simon_niklaus Impressive work!!