←
Azalia Mirhoseini644 posts
Azalia Mirhoseini avatar✓

Azalia Mirhoseini ✓

@Azaliamirh

Founder @RicursiveAI, Asst. Prof. of CS at Stanford. Prev: DeepMind, Anthropic, Brain. Co-Creator of MoEs, AlphaChip, Test Time Scaling.

◷ Joined May 2013▧ 61 Photos and videos
Azalia Mirhoseini avatar
Azalia Mirhoseini@Azaliamirh·Nov 4

Very excited to launch Ripple-65B Preview, a new state of the art for U.S. open-weight model on agentic coding! You can run Ripple-65B locally and efficiently on a MacBook Pro or DGX Spark! Give it a try: HuggingFace: huggingface.co/ricursiveAI/R… Together AI: together.ai/models/ripple-65… GMI Cloud: console.gmicloud.ai/model/ll… OpenRouter: openrouter.ai/gmicloud/rippl…

Post attachment
1:14
Hangoo Kang avatar
⟳ Azalia Mirhoseini retweeted
Hangoo Kang@hangoo_kang·Nov 16

Had a great time at @RicursiveAI’s hackathon with @seilk_ and @TarunSures41845 ! Ripple was capable of building its own code IDE from scratch, achieving < 80 ms autocomplete and 300 tok/s code generation. So much potential here, excited to see what’s next from the team!

Krrish Agarwalla avatar
⟳ Azalia Mirhoseini retweeted
Krrish Agarwalla@Krrishagar·Nov 16

I really enjoyed the hackathon as well. There was an insane amount of crazy ideas flowing through. We built an autonomous loop where ripple could quantize itself when paired up with open evolve framework. We got the size of model from ~60GB ( NVFP4 ) to 30 GBs of model size without compromising much on model quality. Some details here : x.com/Krrishagar/status/2122…

Azalia Mirhoseini avatar
Azalia Mirhoseini@Azaliamirh·Nov 16

The first @RicursiveAI hackathon, centered on Ripple-65B Preview, was so much fun! Participants built impressive computer use demos, a lightning-fast IDE built and operated by Ripple, kernel designs that outperformed Luna, and more. We never directly optimized the model for some of these capabilities, like some cool visual demos.

Post attachment
Azalia Mirhoseini avatar
Azalia Mirhoseini@Azaliamirh·Nov 12

Introducing Speculative Self-Distillation (SSD), which improves accuracy and effiecieny and reduces forgetting in self distillation. It turns out only a few tokens in a rollout actually depend on the new knowledge. Letting the teacher generate just those tokens results in nearly matching on-policy self-distillation with 38–69% fewer training tokens, and the model forgets less. Great work co-led by @talaei_shayan, @chewing_a_gum and @ckodser, with @jonashubotter and Amin Saberi (NeurIPS 2026). Check out Shayan’s thread for more.

Post attachment
0:57
Anna Goldie avatar
⟳ Azalia Mirhoseini retweeted
Anna Goldie@annadgoldie·Nov 12

We'll be hosting our first hackathon this Friday! Come try out fun ideas with Ripple and hang out with our team - we'll be providing food, compute, and chips to take home for the winners! Space is limited, so apply in the Luma link below.

Amir Zeinali avatar
⟳ Azalia Mirhoseini retweeted
Amir Zeinali@amirzeinaalii·Nov 4

AI is running into energy limits, and people are asking whether AI investments will earn back what they cost. That’s exactly why this moment is so exciting. We need to focus on the sectors and workloads that create the most value, and build models and chips that do more with every unit of energy. Really enjoyed working with @JonSaadFalcon, @Avanika15, and the entire team on Joules-to-GDP! Thanks to everyone who helped along the way, especially @Jon_Hartley_, @Azaliamirh, and @HazyResearch.

0:24
Azalia Mirhoseini avatar
Azalia Mirhoseini@Azaliamirh·Nov 5

Introducing Joules-to-GDP: a new metric and framework for measuring how much economic value AI generates per unit of energy. Key insight: selecting the right model on the right accelerators can deliver several times more value per joule for the same task. We believe as AI serving scales across every part of the economy, optimizing for this metric will matter increasingly more. Great work co-led by @amirzeinaalii @JonSaadFaclon and @Avanika15, with an amazing team @Mersad_Abbasi, @Jon_Hartley_, John Hennessy, and @HazyResearch. Check out Jon's thread for several other interesting findings:

0:24