Azalia Mirhoseini
@Azaliamirh
Followers
15K
Following
2K
Media
37
Statuses
382
Asst. Prof. of CS at Stanford, Google DeepMind. Prev: Anthropic, Google Brain. Co-Creator of MoEs, AlphaChip, Test Time Scaling Laws.
Stanford, CA
Joined May 2013
Introducing intelligence per watt, and predicting a distribution shift from cloud to edge inference!
Data centers dominate AI, but they're hitting physical limits. What if the future of AI isn't just bigger data centers, but local intelligence in our hands? The viability of local AI depends on intelligence efficiency. To measure this, we propose intelligence per watt (IPW):
3
9
66
compute efficiency defined the transition from mainframes to pcs intelligence efficiency (intelligence per watt) will define the transition from cloud ai to local ai excited to release new work studying local ai efficiency with the 🐐's @jonsaadfalcon and @hazyresearch we find
Data centers dominate AI, but they're hitting physical limits. What if the future of AI isn't just bigger data centers, but local intelligence in our hands? The viability of local AI depends on intelligence efficiency. To measure this, we propose intelligence per watt (IPW):
3
4
17
1
0
5
Style built for the spotlight. Crafted for performance. Blake Snell wears JAXXON.
0
11
117
A shift from cloud to edge? We took a closer look at “Local LMs” (≤20B active parameters) and found that they are: - Surprisingly capable, with 3.1× improvement since 2023 - Increasingly efficient, with 5.3x improvement since 2023 This suggests a shift from mainframe inference
4
17
76
At the retreat, we’re hearing about the exciting work of a few of our current students: @JulieKallini, @JonSaadFalcon, @ShichengGLiu, @kothasuhas, …
2
10
74
Really awesome writeup by my lab mate Simon @simonguozirui on AI-generated GPU kernels! We've believe in this problem since I first met my advisor Azalia @Azaliamirh as I was applying to Stanford. Grateful to be working on such a cool problem, amazing to see how far the field
Wrote a 1-year retrospective with @a1zhang on KernelBench and the journey toward automated GPU/CUDA kernel generations! Since my labmates (@anneouyang, @simran_s_arora, @_williamhu) and I first started working towards this vision around last year’s @GPU_mode hackathon, we have
2
9
142
Was super fun to organize this workshop!! Thanks everyone: speakers, panelists, audience. https://t.co/ccZzIXFgTY
5
11
140
When Bitcoin moves 2% and you start planning early retirement.
63
62
117
I've spent years pushing the boundaries of pretraining—first as lead author on PaLM, then as a lead contributor on Gemini pre-training. Now I'm at Reflection, building open-weight agentic models at the frontier from the ground up. Today we're announcing our Series B to
reflection.ai
Reflection is a research and product company with the mission to build open superintelligence. We are hiring across all roles!
Today we're sharing the next phase of Reflection. We're building frontier open intelligence accessible to all. We've assembled an extraordinary AI team, built a frontier LLM training stack, and raised $2 billion. Why Open Intelligence Matters Technological and scientific
49
28
588
Cool to see works like AB-MCTS, The AI Scientist and the Darwin Gödel Machine being included in the new “Self-improving AI Agents” course at Stanford! https://t.co/gwgnNe47Hp
5
68
463
Very excited to teach CS329A: Self-Improving AI Agents, with @achowdhery for the second time in 2025! Course website:
Fully packed first class! Excited to TA Stanford CS329A, Self-Improving AI Agents, with @achowdhery @Azaliamirh @boson2photon. Here's to an amazing quarter!
22
123
1K
Congrats to my PhD student @anneouyang on @Standard_Kernel! Super excited to see the amazing things she will build!!
Excited to share what friends and I have been working on at @Standard_Kernel We've raised from General Catalyst (@generalcatalyst), Felicis (@felicis), and a group of exceptional angels. We have some great H100 BF16 kernels in pure CUDA+PTX, featuring: - Matmul 102%-105% perf
4
16
181
Who says crypto has to be complicated? Meet our cutest educators yet: Cross River kids! They're helping us explore crypto concepts with clarity, curiosity, and a lot of charm! Watch the video to learn from the experts!
13
16
60
More details in Jacky's post: https://t.co/BbL3AuergA
✨ Test-Time Scaling for Robotics ✨ Excited to release 🤖 RoboMonkey, which characterizes test-time scaling laws for Vision-Language-Action (VLA) models and introduces a framework that significantly improves the generalization and robustness of VLAs! 🧵(1 / N) 🌐 Website:
0
0
11
Happy to share RoboMonkey, a framework for synthetic data generation + scaling test time compute for VLAs: Turns out generation (via repeated sampling) and verification (via training a verifier on synthetic data) works well for robotics too! Training the verifier: we sample N
5
32
166
So happy to see the strong interest in KernelBench, our AI for AI acceleration benchmark! The team has released some updates today:
KernelBench v0.1 is out, featuring: - A guideline on analyzing the validity of results and ruling out physically impossible performance claims. - Support for randomized testing beyond normal distributions. - Fixed problem sizes and improved numerics
1
8
56
Very excited to share that an advanced version of Gemini Deep Think is the first to have achieved gold-medal level in the International Mathematical Olympiad! 🏆, solving five out of six problems perfectly, as verified by the IMO organizers! It’s been a wild run to lead this
Super thrilled to share that our AI has has now reached silver medalist level in Math at #imo2024 (1 point away from 🥇)! Since Jan, we now not only have a much stronger version of #AlphaGeometry, but also an entirely new system called #AlphaProof, capable of solving many more
79
231
2K
Excited to share that a scaled up version of Gemini DeepThink achieves gold-medal standard at the International Mathematical Olympiad. This result is official, and certified by the IMO organizers. Watch out this space, more to come soon! https://t.co/4KynCY6M6C
deepmind.google
The International Mathematical Olympiad (“IMO”) is the world’s most prestigious competition for young mathematicians, and has been held annually since 1959. Each country taking part is represented by…
9
54
708
On-chain DATs are coming to @Solana. Imagine if they could never fall below NAV. That’s Samsara. 📿 Which asset should go first? $BONK, $PENGU, $PUMP or something new? Comment below ↓
14
8
55
If you want to learn about the power (laws) of large language monkeys (and get a free banana 🍌), come to our poster at #ICML2025 !!
1
6
68
cant stop thinking about this one insanely elegant, seems insanely powerful
26
55
846