Efthymios Tzinis @ETzinis

Senior Research Scientist @GoogleAI | Ph.D. from @IllinoisCS | Formerly @merl_news, @RealityLabs | My opinions do not represent my employer etzinis.com Cambridge, MA, USA Joined November 2017

Tweets

171
Followers

552
Following

303
Likes

4K

Milton Friedman Quotes @MiltonFriedmanW

8 months ago

“A major source of objection to a free economy is precisely that it does this task so well: it gives people what they want instead of what a particular group thinks they ought to want.” — Milton Friedman

29 209 757 27K 55

View Details

Yuchen Jin @Yuchenj_UW

8 months ago

Larry Page & Sergey Brin had the PageRank paper (the algorithm behind Google Search) rejected. A reviewer called it “disjointed.” Geoffrey Hinton's Dropout was rejected for being “too simple.” I often feel the academic peer review is like a random process, especially when a paper is very innovative and changes the paradigm; it often looks "wrong" to reviewers in the old paradigm.

124 398 4K 326K 1K

View Details

unusual_whales @unusual_whales

8 months ago

Americans now believe that college is not worth the cost, per NBC:

198 209 2K 198K 152

View Details

Rohan Paul @rohanpaul_ai

8 months ago

And that's Google's ‘Ironwood,’ its most powerful 7-th gen TPU. - delivers 4,614 FP8 TFLOPS of performance and is equipped with 192 GB of HBM3E memory, offering a bandwidth of up to 7.37 TB/s. - Ironwood pods scale up to 9,216 AI accelerators, delivering a total of 42.5 FP8 ExaFLOPS for training and inference, which by far exceeds the FP8 capabilities of Nvidia's GB300 NVL72 system that stands at 0.36 ExaFLOPS. - The pod is interconnected using a proprietary 9.6 Tb/s Inter-Chip Interconnect network, and carries roughly 1.77 PB of HBM3E memory in total, once again exceeding what Nvidia's competing platform can offer. --- From `Google Cloud` YT channel

Rohan Paul @rohanpaul_ai

8 months ago

Google’s TPUs are on a serious winning streak and now they plan to produce over 5 million TPUs by 2027 to boost supply. A few days back SemiAnalysis reported that, for large buyers Google TPU can deliver roughly 20%–50% lower total cost per useful FLOP compared to Nvidia’s top

24 58 479 132K 171

18 59 389 65K 133

View Details

Turing Post @TheTuringPost

8 months ago

This Google paper presented at #NeurIPS2025 is a true gem. In their search for a better backbone for sequence models, they: • Reframe Transformers & RNNs as associative memory systems driven by attentional bias • Reinterpret "forgetting" as retention regularization, not as erasure • Combine these insights into Miras – a unified framework for designing next-gen sequence architectures From this perspective, they introduce 3 new models, Moneta, Yaad, and Memora, that: - Beat Transformers, Mamba2, DeltaNet, and hybrids across key benchmarks - Scale better to long contexts - Deliver state-of-the-art recall on needle-in-a-haystack tests Here are the details (really worth exploring):

14 85 507 35K 450

View Details

Pedro Domingos @pmddomingos

2 years ago

Unbelievable. China has a huge problem with scientific fraud, and when someone alludes to it the conversation is about how racist they are.

58 27 487 92K 54

View Details

Anastasis Stathopoulos @statho_

2 years ago

Check out our new work "Score-Guided Diffusion for 3D Human Recovery", a.k.a. ScoreHMR, with @ligongh and Dimitris Metaxas that will appear at #CVPR2024! Paper: arxiv.org/abs/2403.09623 Project Page: statho.github.io/ScoreHMR/ Code & models: github.com/statho/ScoreHMR

2 15 69 16K 35

View Details

Efthymios Tzinis @ETzinis

3 years ago

This is the proper usage of machine learning models, removing communication barriers and enabling ideas of freedom to reach people's minds. Exceptional speech from Milei stating the obvious truths.

Linus ✦ Ekenstam @LinusEkenstam

3 years ago

Milei’s talk at Davos This never happened. But with AI (HeyGen) his voice is cloned and turned to English (Elevenlabs) and his mouth is re-synched to match the English words. This is the true power of AI, anything can be consumed in any language

101 207 2K 319K 858

0 0 1 606 0

View Details

Yuma Koizumi @yuma_koizumi

3 years ago

Speech restoration method Miipher (used to generate LibriTTS-R) has been accepted to WASPAA!! It converts degraded speech to studio quality, and generates almost inexhaustible training data for speech generation. Demo: google.github.io/df-conformer/m… Paper: arxiv.org/abs/2303.01664

2 13 47 10K 8

View Details

Jonathan Le Roux @JonathanLeRoux

3 years ago

🥳🎉

IEEE ICASSP @ieeeICASSP

3 years ago

"Hyperbolic audio source separation" by Darius Petermann, Gordon Wichern, Aswin Shanmugam Subramanian, Jonathan LeRoux ieeexplore.ieee.org/document/10094…

1 2 12 4K 2

0 1 14 1K 0

View Details

Jonathan Le Roux @JonathanLeRoux

3 years ago

I will be presenting MERL's work on audio source separation and transcription during a seminar organized by #HiPARIScenter this Thursday, 6/22 @telecomparis and on Zoom. @merl_news @IP_Paris_ @HECParis Registration⬇️ hi-paris.fr/2023/03/09/hi-…

2 5 28 3K 2

View Details

Hao-Wen (Herman) Dong 董皓文 @hermanhwdong

3 years ago

Excited to share that I will be presenting our audio synthesis paper at the CVPR Sight & Sound Workshop! 🤩 In this paper, we proposed a new text-to-audio synthesis model that can be trained using videos and a pretrained language-vision model. #CVPR2023 youtu.be/GyHxVS6PZpk

2 6 41 3K 8

View Details

Efthymios Tzinis @ETzinis

3 years ago

@arouditchenko @Google @GooglePixelFC Hey Andrew, I am in Cambridge as well! Just text me at the internal, see ya there!

1 0 2 201 0

View Details

Efthymios Tzinis @ETzinis

3 years ago

I am super excited to share that I just started as a Research Scientist @Google this week!🤩 Looking forward to delving deeper into audio-visual perception and building great products with the @GooglePixelFC folks!😃

3 3 74 5K 0

View Details

Jonathan Le Roux @JonathanLeRoux

3 years ago

Darius's "Hyperbolic Audio Source Separation" paper has been recognized as one of the top 3% of all papers accepted at #ICASSP2023 @ieeeICASSP 👏🎉🍾 Paper: arxiv.org/abs/2212.05008 Code: github.com/merlresearch/h… Demo video: youtube.com/watch?v=RKsAMb…

Jonathan Le Roux @JonathanLeRoux

4 years ago

The word is out😊New paper w/ Darius Petermann, Gordon Wichern, and @S_Aswin19, "Hyperbolic Audio Source Separation," where we explore embedding time-frequency bins in hyperbolic space for hierarchical separation. Check out Darius's cool demo video on the project page below👇

1 0 24 0 2

0 10 58 17K 8

View Details

Efthymios Tzinis @ETzinis

3 years ago

@actuallyaswin Thanks so much Aswin!! Your self-supervised presentation was super as well!

0 0 1 35 0

View Details

Efthymios Tzinis @ETzinis

3 years ago

Feel free to stop by around 4pm to learn more about optimal condition training for sound separation! Higher upper bounds for conditional separation systems and state-of-the-art text-based separation performance! 🤩 @ieeeICASSP

Efthymios Tzinis @ETzinis

3 years ago

Our @ieeeICASSP23 optimal condition training presentation is online!youtu.be/n2i5kXwvZlM Read our paper with @JonathanLeRoux,Gordon,Paris to learn how to train robust multi-condition separation systems + how to get SOTA text-based separation results!🧐arxiv.org/pdf/2211.05927…

3 1 24 5K 2

1 2 15 3K 1

View Details

Θάνος Τζήμερος @ThanosTzimeros

3 years ago

Κάθε λαός έχει τους ηγέτες που του αξίζουν. Οι πολίτες δεν θέλουν εξάλειψη της διαφθοράς, θέλουν περισσότερες ευκαιρίες να συμμετέχουν σ΄ αυτή. Δεν θέλουν ευνομία, θέλουν ανοχή της δικής τους ανομίας. Πριν από λίγο υπέβαλα την παραίτησή μου από πρόεδρος της Δημιουργίας ξανά.