🤖 data/compute-maxxing
San Francisco, CA
Joined December 2025
- Tweets97
- Following227
- Followers284
- Likes348
Pinned Tweet
ICML 2026 Jeju Island Retreat 🏝️
With @DeltaCommunityX, we brought together nearly 50 Researchers and Founders on Jeju Island, a beautiful island off the coast of South Korea.
From a private yacht and Saryeoni forest to the Osulloc tea museum, swimming, and of course, Korean BBQ, it was an unforgettable retreat!
Huge thanks to our sponsors, @RicursiveAI, @khoslaventures, @BainCapVC, @AfterQuery, @Sky9Capital, @abundant_labs, and to everyone who joined us for the retreat!
Next up is NeurIPS 2026! 👀
Sukone Hong retweeted
These are all on @launchpoint by the way
Sukone Hong retweeted
the best startups are usually about 12 months too early. if you're not getting any traction, it's not obvious yet, but it will be soon. hang in there. Launchpoint wasn’t seeing much growth this year until May. since then we have grown more than I could have ever imagined.
Sukone Hong retweeted
unpopular opinion:
work from home is a cancer to startups.
get your team in one room. trust me.
October was our biggest month yet.
November will be record breaking.
Launchpoint is just getting started.
Sukone Hong retweeted
Great event!
Throwback to the SF Research Night last month
With @GVteam & @stanfordaiclub, we brought together nearly 70 researchers and founders together at the Ferry Building to share research, side projects, and startups!
Huge thanks to our speakers @AbecidAdam, @ericavaneee, @RylanSchaeffer, @charley3c for presenting their amazing work, from one-step video prediction to building verifiers for agentic markets, and much more!
Special thanks to @Rishimodi99 and the GV team for helping organize the event, and to everyone who joined us! 🩶
Sukone Hong retweeted
A lot of noisy events in the bay, but this was a great one!
Throwback to the SF Research Night last month
With @GVteam & @stanfordaiclub, we brought together nearly 70 researchers and founders together at the Ferry Building to share research, side projects, and startups!
Huge thanks to our speakers @AbecidAdam, @ericavaneee, @RylanSchaeffer, @charley3c for presenting their amazing work, from one-step video prediction to building verifiers for agentic markets, and much more!
Special thanks to @Rishimodi99 and the GV team for helping organize the event, and to everyone who joined us! 🩶
Throwback to the SF Research Night last month
With @GVteam & @stanfordaiclub, we brought together nearly 70 researchers and founders together at the Ferry Building to share research, side projects, and startups!
Huge thanks to our speakers @AbecidAdam, @ericavaneee, @RylanSchaeffer, @charley3c for presenting their amazing work, from one-step video prediction to building verifiers for agentic markets, and much more!
Special thanks to @Rishimodi99 and the GV team for helping organize the event, and to everyone who joined us! 🩶
Sukone Hong retweeted
Proud to be the cyber partner for @SemiAnalysis_ GDPx benchmark!
@arithmeticlabs1
Sukone Hong retweeted
Official apology to Google from me, they cooked with Carbon.
It's spikier in terms of domains and capabilities, but it's really good in a few areas. – at San Francisco, CA
Huge congrats to @arithmeticlabs1 @uri_rolls @danielpeters543 on the new benchmark! Incredibly excited about what’s ahead for cybersecurity!
Awesome to have contributed as the cyber partner for @SemiAnalysis_ GDPx benchmark! An incredibly important project to capture the current state of frontier AI's impact on our economy
Hackers are improving quickly, and cyberattacks are becoming more common. Now more than ever, we need to scale defensive capabilities!
@arithmeticlabs1
Sukone Hong retweeted
GDPx covers the following domains. Each was a collaboration with at least one specialized data partner. SemiAnalysis was responsible for final QA and heavily involved with task iteration. (2/7)
Sukone Hong retweeted
1/7 Introducing DeXplicit: making physical state explicit in dexterous manipulation world models.
Given an initial scene and a proposed hand trajectory, DeXplicit 3D forecasts how objects move, articulate and deform. The DeXplicit renderer turns that geometry into future video.
The intermediate 3D state makes the physical response inspectable and provides a control signal for visual generation. Joint-aware actions, coherent object motion and transported appearance connect the two stages.
Sukone Hong retweeted
Today, we introduce Priming Models, a new paradigm for training one-step generative models from scratch.
No diffusion teacher. No flow map. No CFG. No JVP.
We believe this work will challenge how we think about building one-step models.
Joint work with @sitanch, @du_yilun, @parkwbin03, Bohyung Han.
🔥 1.19 FID one-step on ImageNet 256×256 while using substantially less compute, e.g., 12X less than Drifting-B, 6X less than Drifting-L
🎨 Scales to text-to-image generation (GenEval 0.82)
🤖 World modeling on RT-1 / Bridge-V2
Priming Model is built on one core message: [Reconstruction pretraining --> Distributional post-training].
Sukone Hong retweeted
We made next-frame prediction in world models go brrrr
1/8 Introducing Beckmann World Models (BWM): adapting Beckmann transport into fast, one-step action-conditioned world models.
Predicting the next video state from recent observations, proposed actions in one denoising step function call.
Our final RGB model predicts four PushT frames in 24.7 ms on one A800 GPU.
Sukone Hong retweeted
1/8 Introducing Beckmann World Models (BWM): adapting Beckmann transport into fast, one-step action-conditioned world models.
Predicting the next video state from recent observations, proposed actions in one denoising step function call.
Our final RGB model predicts four PushT frames in 24.7 ms on one A800 GPU.
I asked people: "who has hosted great events in SF?"
2 votes each:
- Alicia (@sharedalbums)
- Ben Lang (@benln)
- Dave Fontenot (@davefontenot)
- Ankit Gupta (@AnkitGuptaAI)
- Harbor (@alexgshaw, @ryan_marten)
- Joanne Shang (@JoanneShang)
- Midjourney (@DavidSHolz, @aalleexx)
- Ricardo de Arruda (@notawizard)
- Sarah Chieng (@MilksandMatcha)
- Sukone Hong (@sukonehong)
- Umesh Khanna (@forwarddeploy)
- Vivian Shen (@vivianmshen)
40 people voted. People couldn't pick themselves. I didn't vote.
Continued:
14
706
Sukone Hong retweeted
How do you build the most diverse UMI dataset in robotics?
In the last 8 weeks, we scaled an international UMI data collection operation from 5 to 90 operators, built out a dataset of over 1M+ unique tasks, and continue to capture over 45k+ new tasks each day.
Robotic World Models require much higher diversity of tasks than imitation learning. Here's the why and how behind our operations. 🧵
Sukone Hong retweeted
nimble funds are killing it - have been supporting us the closest since day 1
@sukonehong 🔥
Sukone Hong retweeted
people are asking if its my actual dad in the launch video.
yes it is. he's been a huge inspiration for what I'm building.
and I'm willing to bet everyone knows someone who's about to retire and take decades of experience with them.
authenticity matters especially today.
today we launch @morphic_io: the sovereign AI platform around one core principle.
human knowledge is our most precious asset and needs to compound for the people contributing it.
we unlock our potential only when experience does not walk out the door every time someone leaves
Congrats on the launch of @morphic_io!
Grateful to back @joshsum_ on this journey as their first institutional check. I’m all in on Morphic’s vision to redefine the sovereign operating system, starting with its proactive revenue agent.
today we launch @morphic_io: the sovereign AI platform around one core principle.
human knowledge is our most precious asset and needs to compound for the people contributing it.
we unlock our potential only when experience does not walk out the door every time someone leaves