Florian Dubost, PhD

World Models and GenAI. Principal Machine Learning Engineer @Unity. Previously Staff MLE @Google. Formerly researcher @Stanford, @Harvard and @MGHNeurology.

[Florian Dubost, PhD @fpgDubost]

@AirbnbHelp: a Senior Manager took it over, called it urgent, then went silent a day and a half, no reply to my follow-ups. Family still displaced in a hotel ($860 out of pocket) after the host’s unit failed a safety check. I need a response today

[Florian Dubost, PhD @fpgDubost]

@AirbnbHelp Month-long stay, unit uninhabitable at check-in — broken gas stove control (safety) + not cleaned. Reported w/ photos, host acknowledged it. Waiting 50+ min for a promised supervisor in-app, family displaced tonight. Please escalate — case is in my app messages.

[Florian Dubost, PhD @fpgDubost]

@AirbnbHelp Over an hour with no response, family displaced from an uninhabitable unit with a gas stove safety defect the host acknowledged. Nowhere to sleep tonight. Booking a hotel myself now since Airbnb hasn’t rebooked or replied — will be seeking reimbursement + refund.

[Florian Dubost, PhD @fpgDubost]

@huizi_mao Got it, thanks!

[Florian Dubost, PhD @fpgDubost]

@huizi_mao Thanks! What would you say are drawbacks of NVFP4?

[Florian Dubost, PhD @fpgDubost]

@huizi_mao Hey Huizi, can you explain MXFP4, high level?

[Florian Dubost, PhD @fpgDubost]

@Ronald_vanLoon @FrRonconi @stratorob @iotforall @SAPAriba Not sure about the rest but the last clip was faked and debunked here: https://t.co/vAlhgt50jg

[Google @Google]

Introducing Android XR, our new platform for headsets and glasses built for the Gemini era

[Florian Dubost, PhD @fpgDubost]

@Stearns2Colton @AdamWHarley @mikacuy @fedassa @GordonWetzstein @GuibasLeonidas Amazing work @Stearns2Colton ! That is going to be a game changer for 3D content creation! Very thankful to be a part of the effort!

[Colton Stearns @Stearns2Colton]

Video to 4D is an incredibly challenging task! Excited to announce that Dynamic Gaussian Marbles was accepted to #SIGGRAPHAsia2024 with @AdamWHarley @mikacuy @fedassa @fpgDubost @GordonWetzstein @GuibasLeonidas. Check out our webpage and code!
https://t.co/rLz8cRfU1B

[Florian Dubost, PhD @fpgDubost]

@Stearns2Colton @GordonWetzstein @AdamWHarley @mikacuy @fedassa @GuibasLeonidas Great work!

[Florian Dubost, PhD @fpgDubost]

@donnelly_brent @JeremiahDJohns I have not double checked all of them. But the few I did were fake. Not surprising…

[Florian Dubost, PhD @fpgDubost]

@JeremiahDJohns This is super fake. Check it yourself (e.g. by running the prompt through Geminj) before believing the first thing you read. The first defense against false information is your own critical thinking.

[Qingqing Zhao @qingqing_zhao_]

How do we create realistic models of dressed humans directly from visual data?

We introduce PhysAvatar, a framework that estimates the shape, appearance, and physical parameters of dressed human avatars from multi-view videos.

Page: https://t.co/d87GBIsse2

(1/6)

[Gabriel Peyré @gabrielpeyre]

Oldies but goldies: A. Efros and T. Leung, Texture Synthesis by Non-parametric Sampling, 1999. The first copy-based texture synthesis method. https://t.co/G43DVZs4x3

[Jeff Dean @JeffDean]

I’m very excited to share our work on Gemini today! Gemini is a family of multimodal models that demonstrate really strong capabilities across the image, audio, video, and text domains. Our most-capable model, Gemini Ultra, advances the state of the art in 30 of 32 benchmarks, including 10 of 12 popular text and reasoning benchmarks, 9 of 9 image understanding benchmarks, 6 of 6 video understanding benchmarks, and 5 of 5 speech recognition and speech translation benchmarks. Gemini Ultra is the first model to achieve human-expert performance on MMLU across 57 subjects with a score above 90%. It also achieves a new state-of-the-art score of 62.4% on the new MMMU multimodal reasoning benchmark, outperforming the previous best model by more than 5 percentage points.

Gemini was built by an awesome team of people from @GoogleDeepMind, @GoogleResearch, and elsewhere at @Google, and is one of the largest science and engineering efforts we’ve ever undertaken. As one of the two overall technical leads of the Gemini effort, along with my colleague @OriolVinyalsML, I am incredibly proud of the whole team, and we’re so excited to be sharing our work with you today!

There’s quite a lot of different material about Gemini available, starting with:

Main blog post: https://t.co/NzSycJl7aE

60-page technical report authored by th Gemini Team: https://t.co/CEdMRyYSLo

In this thread, I’ll walk you through some of the highlights.

[Florian Dubost, PhD @fpgDubost]

@harshmadhusudan @muktabh According to the graph, Australia actually has the lowest sensitivity. On the contrary, India is very correlated to the Chinese market, it is just inversely correlated…

[Florian Dubost, PhD @fpgDubost]

@mvollmer1 Where is the 5th advice? Haha

[Florian Dubost, PhD @fpgDubost]

@Ronald_vanLoon @Tesla @sebbourguignon @wil_bielert Is that true? What’s the context?

[Florian Dubost, PhD @fpgDubost]

@peter_richtarik Just write exactly that in the review, see how the authors answer in the rebuttal and write what you think of their answer in the second set if reviews. Ultimately, the pc will make the decision. Your responsability is just to give advice. Don’t overthink it ;)