Michal Valko @misovalko Twitter profile

Last Seen Profiles

@JacobKing67

@wintersentroses

@shopreme

@MatcoNFT

@AsedDale

@rkivequotes

@BasketballsGoat

@Leoxx0811

@KollegerBoss

@MoriartySats

@GeorgeForeman

@bokeplokalmalam

@blazeringtrails

@kirawontmiss

@Top2Forever

@reiicatt

@KILLUNIMATTY

@jkahn_5

@Andyalmeyra

@22Goingviral_

@bokeplokalmalam

@miiraaacleee_

@AshokranwaRlp14

@vhay5

@lilkabby1

@eclecticbrix

@grossisergio45

@Therussian_12

@jujumuniz9

@Frankie070712

@moises_c_r

@cpacs

@2dearanda

@etoilefoudroyee

@ramsdensfish

@fmcgatepa

Michal Valko

@misovalko

9 months

Fast-forward ⏩ alignment research from @GoogleDeepMind ! Our latest results enhance alignment outcomes in Large Language Models (LLMs). Presenting NashLLM!

4

129

832

Michal Valko

@misovalko

6 months

Life update! 📢 Today is my first day as a Principal Llama Engineer of the @Meta 's GenAI in Paris! Massive thanks for a very personal involvement of Mano, @edunov , Mark, @ylecun , @jpineau1 , Naila, Laurens & Ricardo to bring me onboard to of @AIatMeta !

44

14

561

Michal Valko

@misovalko

4 months

Come, study, and get a Ph.D. with us at @AIatMeta ! Let’s make LLMs more useful together for everyone! Positions are 4-year fixed-term in London - jointly supervised by academics from @ai_ucl and researchers at FAIR. @Meta @ucl @MetaOpenSource

Error

Meta's mission is to give people the power to build community and bring the world closer together. Together, we can help people build stronger communities - join us.

www.metacareers.com

5

57

417

Michal Valko

@misovalko

10 months

🚀Alignment improved! 🌐 Our RL fine-tuning group at @GoogleDeepMind presents a new algorithm for LLM fine-tuning—say hello to IPO: Identity Preference Optimization! 🌟

Mohammad Azar

@Learnius

10 months

At the middle of this surreal #OpenAI saga some good AI news: checkout our latest work Identity Preference Optimization (IPO): When compared to RLHF used to train #ChatGPT IPO is more general (no reliance on reward), more stable and easier to implement!

3

13

105

3

44

237

Michal Valko

@misovalko

4 months

«Merci, cher Mark» 😏 It was a fabulous 4-hour nuit blanche 🌃 at Matignon w/Côte-Rôtie🍷sealing the deal «Valko 6 pack» celebrating Llama 3, 4️⃣alignment papers accepted to @icmlconf , chatting about NashLLMs, skills, Hejný, OSS, AI for good, safety & agents at @AIatMeta , @finkd !

7

8

211

Michal Valko

@misovalko

4 years

The list NeurIPS 2020 Accepted Papers is up!

0

44

149

Michal Valko

@misovalko

10 months

🤗 Amazing speed 💨 from friends from @huggingface ! Just 24 hours after I talked to them the code of our IPO for LLM Alignment is in their TRL codebase! Thank you @younesbelkada @lvwerra @_lewtun @krasul and other friends at @huggingface - @HugoLaurencon , @SanhEstPasMoi ,

younes

@younesbelkada

10 months

IPO algorithm, a new method from Google Deepmind: has been just added in Hugging Face TRL library ! Try it out now by installing TRL from source, simply pass `loss_type="ipo"` when initializing DPOTrainer:

5

72

428

2

20

140

Michal Valko

@misovalko

2 years

12 years ago, both of us met Remi Munos as a Ph.D. / postdoc advisor and this forever changed our lives! Today he took us to Casino La Nouvelle Orléans to test our multi-arm bandit algorithms in practice! Theory to practice! 🙂

3

2

132

Michal Valko

@misovalko

2 months

🚀 Exciting news! At @AIatMeta , we just kicked off a new AI accelerator program to empower European startups, in partnership with @huggingface , @Scaleway , @HECParis , and hosted at @joinstationf in Paris! 🌍✨

2

20

130

Michal Valko

@misovalko

6 months

📣 Ph.D. proposal - Fall 2024: 𝗧𝗿𝗲𝗲 𝗦𝗲𝗮𝗿𝗰𝗵 𝗔𝗹𝗴𝗼𝗿𝗶𝘁𝗵𝗺𝘀 - 𝗕𝗲𝘆𝗼𝗻𝗱 𝗠𝗼𝗻𝘁𝗲 𝗖𝗮𝗿𝗹𝗼 𝗧𝗿𝗲𝗲 𝗦𝗲𝗮𝗿𝗰𝗵! Come and work with the advising team of Amine, Pierre, and Michal - headed by excellent ✨Bruno Scherrer ✨!

2

27

126

Michal Valko

@misovalko

4 months

🎓At @AIatMeta even the Chief Al Scientist supervises Ph.D. students! Together with Jean Ponce, @ylecun gave birth this week to Dr. @AdrienBardes 🎓, crème de la crème in predictive non-contrastive learning, the topic I love💚 #byolforever , and brought us ViCReg, IMJEPA and more!

4

10

110

Michal Valko

@misovalko

3 months

Ça y est, today's the day ! Holubica mieru🕊️ and kamzík 🐐 meeting Coq Gaulois 🐓 in a battle of paperwork! 🙏 to my pays des Lumières☀️ for promptly organizing elections just for me to try droit de vote ! And 🙏 to all my friends 🤗 coming to the soirée de #naturalization !🥂🇫🇷

12

0

104

Michal Valko

@misovalko

5 months

MoD is out! 🔥🤩🌳

Aran Komatsuzaki

@arankomatsuzaki

5 months

Google presents Mixture-of-Depths Dynamically allocating compute in transformer-based language models Same performance w/ a fraction of the FLOPs per forward pass

6

90

609

1

8

99

Michal Valko

@misovalko

5 months

Principal Llama Engineer finally got to meet his boss Chief Llama Officer @osanseviero for the first time in person in Parisian Tigermilk bar and it was a blast! @huggingface @AIatMeta partnership and #llamalove 🦙💚

1

2

94

Michal Valko

@misovalko

5 months

Anyone else still does not feel how is our open-weights space is heating up rn? 🔥 The alignment team of @huggingface meets once a year... in Paris! 🇫🇷 And they have a few representatives of LLama 3🦙 alignment team over for a nice dinner! 🤗 @Meta @AIatMeta @MetaOpenSource

4

10

91

Michal Valko

@misovalko

1 year

His first Ph.D. work, his first conf #ICML2023 and already a paper award , what a start! 🔥 So proud of you, Côme! With 💚 your Pierre & @Tdash_Koz & Rémi & @VianneyPerchet & @misovalko - @GoogleDeepMind @ENSdeLyon @ENSAEparis @CriteoAILab @omron_sinicx

1

5

89

Michal Valko

@misovalko

4 years

Waking up at 4:30am to be able to see this beauty❤️

1

87

Michal Valko

@misovalko

4 years

The day has come! We release the BYOL source code! w/ @c_tallec @KloudStrife @Piotbilal @koraykv @CarlDoersch @obernardo @Learnius @misovalko Florent Altché, Florian Strub, Jean-Bastien Grill, Daniel Guo, Elena Buchatskaya & Rémi Munos! @DeepMind

2

26

85

Michal Valko

@misovalko

7 months

Let’s get closer to understanding the ➡️ scaling laws ⬅️ for human feedback in RHLF! 🚀 Our rising star student @dtiapkin has only started his Ph.D. with us 5 months ago and he has already cracked some open questions behind the LLM alignment in our latest #ICLR2024 paper! 💥

2

7

82

Michal Valko

@misovalko

2 months

@ylecun from @AIatMeta greets the public at the exclusive Open Source AI Day event at @joinstationf at panel with @adriennejan from @Scaleway , @Thom_Wolf from @huggingface and @ptrkprz from @kyutai_labs Join us as We will continue by great...

3

12

80

Michal Valko

@misovalko

4 months

In @GoogleDeepMind we used to say "If @laurentsifre is part of this project, he will make it work!." Watch out for Laurent, he will do many more great things!

Laurent Sifre

@laurentsifre

4 months

H

25

19

268

1

78

Michal Valko

@misovalko

4 months

I spent all my PhD working on with medical data, it’s super challenging. This work is amazing and great collaboration!

AI at Meta

@AIatMeta

4 months

Researchers at @ICepfl & @YaleMed teamed up to build Meditron, an LLM suite for low-resource medical settings. With Llama 3, their new model outperforms most open models in its parameter class on benchmarks like MedQA & MedMCQA. More details ➡️

11

110

532

0

2

74

Michal Valko

@misovalko

2 years

After 11 years of living in France, Mr. Macron shook my hand - in the the United States! La Nouvelle Orléans 🇫🇷. "Monsieur Macron, veuillez accepter ma demande de naturalisation?" I asked. And he just laughed. So French!

0

2

72

Michal Valko

@misovalko

5 months

💾 Llama 3 - download a sneak peak now!💾 I am way too lucky that I could enjoy the Llama 3 joyride during exactly 1 month from starting at Meta! Aligned with care! Aligned with friends! 🦙🦙🦙 @Meta @AIatMeta @MetaOpenSource

0

7

70

Michal Valko

@misovalko

3 months

He is no° 1! Congrats to already famous @ramealexandre for the best ML thesis in the very first edition SSFAM awards! @GoogleDeepMind is happy to have you and I'm grateful for the chance of learning from you! Check the 192 pages of ratatouille in !

1

3

68

Michal Valko

@misovalko

2 years

Wed at 11am at #NeurIPS2022 : PSRL or Thompson Sampling for RL dates to 1933 and was popularized by @IanOsband @DanielRuss0 and Ben Van Roy in 2013 - but is it optimal? In 2022, we answer the open questions raised by Agrawal and Jia (2017) in the episodic setting!

3

6

63

Michal Valko

@misovalko

6 months

📢 If you are an academic then for the next *one month* you can use this form to apply for *free* Gemma credits and get USD $5K per award! 📡 🔭 🧪 🔬 🧫 🧬 @GoogleDeepMind @Google

2

15

62

Michal Valko

@misovalko

4 months

LLM Hackathons are on 🔥 in Paris too! This one was organized by @doniagharbi4 and friends from @join_ef . Our LLama team 🦙🦙🦙 from AI at Meta EMEA was represented in jury by @_vassim , a @AIatMeta researcher from Paris! @Meta @MetaOpenSource @ThomasScialom @mialon_gregoire

2

10

58

Michal Valko

@misovalko

2 years

Visiting the foundational @MIT_CSAIL in Cambridge, USA! The home of my advisor and his advisor!

0

1

57

Michal Valko

@misovalko

1 year

📢 We made it! First word-class AI event is coming to Slovakia 🇸🇰 next week, @EEMLcommunity ! In past days we further enriched the program with more stellar speakers such as @StefanieJegelka and @fhuszar ! Stay tuned for more updates and press next week! -

0

5

56

Michal Valko

@misovalko

4 years

Getting up at 5am for a hike to Bystré Sedlo with my mom. Starting with a beautiful view right now!

2

0

56

Michal Valko

@misovalko

7 months

Aligned more openly ;) enjoy these checkpoints from @GoogleDeepMind !

Google

@Google

7 months

Gemma is a new family of open models that help developers and researchers build AI. Along with the lightweight models, we’re launching tooling that encourages collaboration and a guide to responsible use of these models. Learn more →

1K

973

4K

2

3

56

Michal Valko

@misovalko

4 years

Exploration unlimited! For reward-free exploration, we show that the bonuses that scale with 1/n as opposed to 1/sqrt(n) bring faster learning rates! Explore much more! Read ! pm @omardrwch aj ek @eleurent @misovalko @Inria @UPFBarcelona @CNRS @DeepMind

4

17

53

Michal Valko

@misovalko

4 months

A sneak peek photo from the beginning of Lllama 3 baby shower 🚿 at @AIatMeta premises in Paris! 🥖 Thanks for all for coming and thanks @ThomasScialom and @ #RicardoSilveiraCabral for organizing and sponsoring the event and to the team that made this possible! @Meta

0

52

Michal Valko

@misovalko

4 months

I started work on online kernel clustering back in 2008 at Intel with Brano Kveton for our online face recognizer. . 15 years later it became super handy for intrinsic motivation for deep RL: Check out our RECODE work today at @iclr_conf with a spotlight!

4

7

51

Michal Valko

@misovalko

4 months

The open-sourcing of IPO and NashMD and the coolest new algorithms that now align Llama 3 start with running 10km every week with my good friends @HugoLaurencon and @mishig25 , lots of amazing work done by @krasul from TRL! @AIatMeta @huggingface Join us!

Mishig Davaadorj

@mishig25

4 months

Anyone wanna join the Paris AI running club? Comment below Here with @HugoLaurencon from Idefics team & bug dawg @misovalko from Llama 3 team

7

2

32

2

4

51

Michal Valko

@misovalko

4 months

Hear, hear Paris! 🥖 A cool new opportunity to learn 📚from AI apostles⭐ like @BachFrancis , @gabrielpeyre , @olivier_cappe , @mblondel_ml @pierresenellart and many others and get a master degree! 🎓 30 more days to sign up at

1

12

47

Michal Valko

@misovalko

4 years

Seven years ago, with Daniele, Pierre, and @omardrwch we created "Graphs in Machine Learning" for , first of its kind. From now, the future is Daniele at ! All the past material: @UnivParisSaclay @DeepMind

1

12

47

Michal Valko

@misovalko

2 years

The amazing performance of these two on the stage! The up and coming young @Inria 's stars @mathusmassias and @tomamoral , inventors of the digital work present BenchOpt toolbox at pre-screening of #NeurIPS2022 - the French event of real human talks and human interactions ...

1

6

47

Michal Valko

@misovalko

2 years

Restelli Lab of th @ELLISUnitMilan receiving the award at #ICML2022 !

0

5

46

Michal Valko

@misovalko

2 years

And our baby Omar with Emilie and Pierre now becomes Dr. Darwiche Domingues ! Full of passion that transpired to the presentation, bravo doc 💪🤗 @omardrwch @Inria @Inria_Lille @CNRS @CNRS_HdF @OwkinScience @univ_lille

3

45

Michal Valko

@misovalko

1 year

Dr. @ahtouati , also know is Tunisian ML star from Sfax steals the show of RL researchat ICLR 2023 and passionately answers Does Zero-Shot Reinforcement Learning Exist? Check it out the demo at 🎉 @iclr_conf @MetaAI

1

4

44

Michal Valko

@misovalko

2 years

A coffee with now Frau Professorin Dr. @vernadec who has just is started this January her own research group in @uni_tue on theoretical aspects of reinforcement learning at @MPI_IS as @noether_emmy Group Leader! Ein tolles Geschenk für Deutschland! 🇩🇪

0

2

44

Michal Valko

@misovalko

9 months

Pleasure to work with @huggingface ! We gave a talk to them on our IPO () on Thursday & the official TRL release of today (0.7.6) fully supports it! Try it w/ loss_type="ipo" argument to the trainer! w/ @Learnius , mr, @Piotbilal , dg, dc, rm @GoogleDeepMind

A General Theoretical Paradigm to Understand Learning from Human...

The prevalent deployment of learning from human preferences through reinforcement learning (RLHF) relies on two important approximations: the first assumes that pairwise preferences can be...

arxiv.org

younes

@younesbelkada

9 months

In the recent TRL release (0.7.6) we are happy to ship many cool features to the community with respect to DPOTrainer! IPO, KTO, cDPO loss, support for pre-computed logits, and many more! Check out the thread below for more details 🧵

3

34

150

0

8

43

Michal Valko

@misovalko

6 months

Come and join our Paris office! @syhw and the CodeGen team are strong, cool, happy people! 🌳

Gabriel Synnaeve

@syhw

6 months

The CodeGen team at FAIR *in Paris* is recruiting junior and senior research engineers! Come work with us @jadecopet @b_roziere @qcar_ @FabianGloeckle @KunhaoZ et al., and folks in EMEA @jnsgehring @TacoCohen @adiyossLC @FelixKreuk et al.

1

27

105

1

43

Michal Valko

@misovalko

1 year

Little professor moment with @ArthurGretton with 'Professor' wine at @gkdziugaite place in Kigali to finish up @iclr_conf !

1

43

Michal Valko

@misovalko

6 months

@misovalko - Category: Al and the rise of large language models @GoogleDeepMind @Inria @ENS_ParisSaclay @ESETresearch @EEMLcommunity @ComeniusUni @universitypittsbu @PittTweet

Michal Valko - Forbes changemakeri 2024

Predstavujeme vám nový rebríček Forbes Changemakeri – 12 osobností, ktoré posúvajú vpred svoj odbor aj Slovensko ako krajinu.

www.forbes.sk

4

3

43

Michal Valko

@misovalko

2 years

Giving a sneak preview at @NeurIPSConf of our "BYOL Hindsight" work with Dan Jarrett, @c_tallec , Florent Altché, @mesnard_thomas , and Rémi Munos. It will be presented in Deep Reinforcement Learning Workshop on Fri 9 Dec 2022! @DeepMind Photo: @ahtouati

Google DeepMind

@GoogleDeepMind

2 years

How could AI become as curious as humans? At #NeurIPS2022 , we're presenting a simple and scalable generalization of curiosity-driven exploration to help agents tell the difference between “noise” and “novelty - and stay robust in uncertain environments:

9

67

392

0

6

42

Michal Valko

@misovalko

4 years

Three months of lockdown lead to our three months intense self-supervised learning :-). Thanks to all and helped us make it happen, the journey was as enjoyable as the finish! @BernadoPires bp @CarlDoersch @c_tallec dg @ebuchatskaya fa fs jbg @koraykv @Learnius @KloudStrife rm

Google DeepMind

@GoogleDeepMind

4 years

Moving away from negative pairs in self-supervised representation learning: our new SotA method, Bootstrap Your Own Latent (BYOL), narrows the gap between self-supervised & supervised methods simply by predicting previous versions of itself. See here:

9

222

795

2

12

42

Michal Valko

@misovalko

7 months

Aligned with care 🤗

Google

@Google

7 months

Today we're introducing Gemini 1.5, our next-generation AI model. It shows dramatically enhanced performance, including long-context understanding across modalities, which opens up new possibilities for people to create and build with AI → #GeminiAI

418

911

5K

2

1

42

Michal Valko

@misovalko

5 years

Announcing the #humAIn workshop "The Power of Graphs in Machine Learning and Sequential Decision Strategies" in London, June 3-4, 2019 - Sign up at . @Inria @Inria_Lille @ucl @cnrs @univ_lille

0

16

38

Michal Valko

@misovalko

2 years

Graph representation learning with BGRL! ICLR 2022 at @SkVedator with Shantanu Thakoor, @c_tallec , @Learnius , @mehdiazabou , Eva Dyer, Rémi Munos, and @PetarV_93 - @DeepMind (c) Stanislav Griguš @ComeniusUni

0

7

40

Michal Valko

@misovalko

4 years

This Tuesday at reinforcement learning theory seminar, @RLtheory , Pierre Ménard from @Inria_Lille will tell you all you wanted to know about 𝗿𝗲𝘄𝗮𝗿𝗱-𝗳𝗿𝗲𝗲 𝗲𝘅𝗽𝗹𝗼𝗿𝗮𝘁𝗶𝗼𝗻! Watch at: ! w/ @Inria @cnrs @UPFBarcelona @renault_fr @DeepMind

3

10

39

Michal Valko

@misovalko

2 years

Viorica opens up hybrid @EEMLcommunity Summer School in Vilnius, Lithuania with 150 in-person participants! Check it at ! Organized by Doina, @fhuszar , @rpascanu , Viorica, @DovydasCeilutka , @JevGamper , @gkdziugaite , @LinasTw , LinasP!

0

4

39

Michal Valko

@misovalko

6 years

#SequeL presents: Applications to Reinforcement Learning Summer #SCOOL in 1-12 July 2019, Lille, France justed opened! Check out the list of confirmed speakers at: @Inria @Inria_Lille @CNRS_HdF @CNRS_HdF @univ_lille #humAIn

0

21

36

Michal Valko

@misovalko

3 months

June 10, 2024 is the day when I welcome my dear friend Pierre Ménard to the Llama team. “Nous serions Gilles et moi très heureux que tu puisses faire ce rapport” wrote Auré in 2008 and 23 papers later, the journey has never stopped! Photo from Peru. Revealing a bit of #v6p too!

0

35

Michal Valko

@misovalko

4 years

Interested in @DeepMind ? Come and talk to me and other DeepMinders at Sponsor Hall at @NeurIPSConf ! We are starting in 1 hour. Welcome to #NeurIPS2020 ! DeepMind's stand:

1

2

34

Michal Valko

@misovalko

1 year

Happy anniversary 🎉 to @ml_collective 🎉 @savvyRL you deserve some cakes! 🎂

1

2

33

Michal Valko

@misovalko

4 months

Come in next hour to chat with @dtiapkin and Daniele at @iclr_conf in Vienna 🇦🇹 about first full fledged learning theory take on RLHF taking into account ALL pieces! @dtiapkin · D. Belomestny · D. Calandriello · @EricMoulines · A. Naumov · P. Perrault · @misovalko · P. Ménard

Michal Valko

@misovalko

7 months

Let’s get closer to understanding the ➡️ scaling laws ⬅️ for human feedback in RHLF! 🚀 Our rising star student @dtiapkin has only started his Ph.D. with us 5 months ago and he has already cracked some open questions behind the LLM alignment in our latest #ICLR2024 paper! 💥

2

7

82

2

4

33

Michal Valko

@misovalko

7 months

We made it! 🇪🇺 Google DeepMind

0

1

32

Michal Valko

@misovalko

2 years

Congrats to @jean_tarbou for becoming Dr. Tarbouriech that we have supervised together with @alelazaric ! Goal-conditioned reinforcement learning was at the core of his contributions over the past years, check them out at !

3

1

32

Michal Valko

@misovalko

1 year

Silicon Valley beers with all time inter galactic reinforcement learning heroes and finally also all my lovely coauthors and mentors from Hungary, France, Iran, America, Poland, and Slovakia 😍😂🍺

0

2

32

Michal Valko

@misovalko

1 year

Day 4 of @EEMLcommunity : We are blessed to have @fhuszar , @Cambridge_Uni that is telling us "Why Deep Learning Works"!

0

2

30

Michal Valko

@misovalko

5 months

Come and join us within @NailaMurray ’s org who leads the foundational and exploratory research in Europe! Apply at: @AIatMeta @AIatMeta @MetaOpenSource

0

6

29

Michal Valko

@misovalko

4 years

October 1st and and this is the day when we start full 1 month of onboarding because yes we love you so much! Welcome to @DeepMind Paris, @arthurmensch !

1

0

30

Michal Valko

@misovalko

2 years

In-person participants of @EEMLcommunity summer school in Lithuania 2022 exploring the city of Vilnius and posing in front of the Vilnius Cathedral! 🇱🇹 (c) @lithuania_ai

0

3

30

Michal Valko

@misovalko

2 years

After 11 years of living in France, Mr. @EmmanuelMacron shook my hand - in the the United States! La Nouvelle Orléans 🇫🇷. "Monsieur Macron, veuillez accepter ma demande de naturalisation?" I asked. And he just laughed. So French!

0

29

Michal Valko

@misovalko

3 years

Back to in-person PhmD. defense of the up and running graph transformer prodigy @mialon_gregoire with @Korba_Anna , @mmbronstein , @gabrielpeyre , @agramfort @pascal ( @julienmairal and Alex of course) in the jury. Let's enjoy the ride! 🤗

1

0

29

Michal Valko

@misovalko

3 months

Another cool event at @joinstationf by @join_ef and @hugolb05 on AI Safety! @AIatMeta will be represented by our in-house expert @pierrefdz that you know from StableSignature or AudioSeal in our long-term collaboration with @Inria ! Sign up for free at

1

3

29

Michal Valko

@misovalko

2 years

Opening the new @OwkinScience in Paris with @omardrwch @jeanphi_vert @agramfort @VGabillon @TClozel Gilles Wainrib

0

3

29

Michal Valko

@misovalko

4 years

Congrats Dr. Pierre Perrault for the enthusiastic :-) Ph.D. defense and giving the hottest 🔥 results on statistically AND computationally efficient algorithms for combinatorial semi-bandits! 🍾

0

28

Michal Valko

@misovalko

6 months

Wow this is just amazing and immesly a great feat withing academia, congrats @prfsanjeevarora and @PrincetonPLI !

Sanjeev Arora

@prfsanjeevarora

6 months

The GPUs have landed! (300 H100s on tap for AI research @PrincetonPLI )

5

27

390

2

0

28

Michal Valko

@misovalko

1 year

We started talking about it with Razvan 4 years ago, made a tour of all institutes in Slovakia 🇸🇰and today I am presenting you the équipe de choc who will bring @EEMLcommunity to my hometown @VisitKosice in 2 months! @ESET @DeepMind

0

1

28

Michal Valko

@misovalko

6 years

The very last poster session of NeurIPS 2018. That means that this was the last paper that SequeL will present EVER. 12 years of joy. King is dead long live the king. Get ready for the coronation of the new one next year! @Inria_Lille @Inria @univ_lille @DeepMindAI @CNRS

1

2

28

Michal Valko

@misovalko

2 years

Two years after BYOL, we present BYOL-Explore! BYOL-Explore performs representation learning and exploration together. BYOL-Explore is effective across many domains and scores well in the hard exploration suite DeepMind Hard 8. Check it out!

Google DeepMind

@GoogleDeepMind

2 years

Exploring in sparse-reward partially-observable tasks? BYOL-Explore harnesses the power of Self-Supervised Learning to solve such tasks only with prediction error as intrinsic reward, producing a conceptually simple yet powerful exploration agent: 1/

7

48

211

2

0

27

Michal Valko

@misovalko

4 years

This academic year we are extending the @DeepMind scholarship program to France as a result of our longterm collaboration with the #MVA master - mathématiques / vision / apprentissage - at @ENS_ParisSaclay ! Thanks to Agnes Desolneux and @NVayatis for their support!

Google DeepMind

@GoogleDeepMind

4 years

Our scholarship programme was established in an effort to help build a stronger and more inclusive AI community. This week, we welcomed our largest and most geographically diverse cohort yet. (1/3)

6

64

289

0

3

27

Michal Valko

@misovalko

5 months

Let the home style 🏡 cooking and stirring (alignment) begin! @Meta @AIatMeta

0

2

27

Michal Valko

@misovalko

6 months

Famous Palo Alto lemons!

0

1

26

Michal Valko

@misovalko

2 years

Happy birthday to Sir @jean_tarbou who has graduated hands down and now started a research scientist position in @DeepMind ! The new discovery is near!

0

1

25

Michal Valko

@misovalko

4 years

The SequeL life does not finish when on Friday night you put down the pen, you clean the whiteboard, and take the last metro home. Because SequeL life just does not finish period ❤️ Thank you Remi Munos for creating this family! This Saturday is a SequeL field trip to Dunkerque!

0

2

26

Michal Valko

@misovalko

3 years

Bright news of the day, the thesis defense of Y̶u̶n̶h̶a̶o̶ Dr. Tang! Overflowing with spectacular contributions! Congrats @robinphysics ! @Columbia w/ @ashipra , @DanielRuss0 , @ChrKroer , Yuri Faenza 🎉🎉🎉

1

0

26

Michal Valko

@misovalko

2 years

AI for Africa is full swing with lectures today and tomorrow! Organized by @EricMoulines and Mohammad and their teams! @MetaAI @DeepMind ECC UM6P @Polytechnique

2

5

26

Michal Valko

@misovalko

2 years

In-person poster presentation at #ICML2022 ICML where we got to present our work on MiniMeta and show another way how to scale up GPUCB / BBKB! @DeepMind @MetaAI @malga_center @lrntzrsc @alelazaric @luigicarratino Dani e Ale!

2

22

Michal Valko

@misovalko

4 years

MDP-GapE! Planning in Markov Decision Processes with Gap-Dependent Sample Complexity w/ Anders Jonsson, Emilie Kaufmann, Pierre Ménard, @omardrwch , @eleurent - @UPFBarcelona @CNRS / @CNRS_HdF @Inria / @Inria_Lille @renault_fr @DeepMind / #DeepMindParis #DELTA

DeepAI

@DeepAI

4 years

Planning in Markov Decision Processes with Gap-Dependent Sample Complexity by Anders Jonsson et al. including @misovalko #Probability #MarkovDecisionProcess

0

4

1

7

26

Michal Valko

@misovalko

5 years

Up and ready. Strong and stable. COLT 2019. @Inria_Lille @CNRS @Alliance_humAIn @univ_lille @facebookai @Inria @CNRS_HdF

0

3

25

Michal Valko

@misovalko

4 years

Congrats for the test-of-time award for GP-UCB! And if you are looking for an algorithm with same accuracy but a fraction of GP-UCB’s runtime, look no further! Check the test-of-RUNtime award candidate: BBKB! #ICML2020

Hal Daumé III

@haldaume3

4 years

We are very pleased to announce that the #icml2020 Test of Time award goes to Gaussian Process Optimization in the Bandit Setting: No Regret and Experimental Design by Niranjan Srinivas, Andreas Krause, Sham Kakade and Matthias Seeger >

3

90

467

1

6

25

Michal Valko

@misovalko

4 years

I'm just 4 months old but when I grow up... @DeepMind @Bratislava

0

25

Michal Valko

@misovalko

5 months

📢 Science x engineering at @ghostdayamlc ! @AIatMeta collaborates with @Inria to ✈️ today the 3 musketeers from Paris to Poznań headed by famous DINOsaur tamer @TimDarcet - as he was nurtured on macarons, he will be surrounded by two Slavic bodyguards @p_bojanowski & @misovalko !

1

4

25

Michal Valko

@misovalko

4 years

Congratulations to our @DeepMind scholars of #MVA 2020/2021 at @ENS_ParisSaclay : Imad Aouali, Léna Néhale Ezzine, and Margaux Törnqvist! Read more about the scholarship and the selection process at

0

2

25

Michal Valko

@misovalko

4 years

Sequel lunch, sequel coffee, sequel walk, sequel run, and now Sequel dinner and beer! All in 7 hours! They can't stop us! @Inria_Lille @Inria @CNRS_HdF @CNRS @univ_lille #humAIn @AbsolutSamuel @omardrwch @odalric_ambrym @llvllahsaa

0

1

25

Michal Valko

@misovalko

2 years

"Les choses élevées maintiennent la paix !" Under which @umutsimsekli , one of the strongest inventeur du monde numérique present one of many of his @NeurIPSConf results! You are going to more and more from this ML hero, the future is bright!🌞 @inria_paris @InstitutPrairie

0

23

Michal Valko

@misovalko

4 years

Post NeurIPS deadline hike of 25 bosses de Fontainebleau ! Thanks to European Galileo system and out GPS boss with OSM that leads us. Special thanks to the food delivery incarnated by Florian Strub :)

1

0

24

Michal Valko

@misovalko

5 months

83.33% #v6p

1

24

Michal Valko

@misovalko

1 year

The dome is ready! 😍 @iclr_conf

0

24

Michal Valko

@misovalko

1 year

Second day of @EEMLcommunity in Košice, Slovakia starts with the *local* ML star @peter_richtarik now at @KAUST_News giving the full account by now already 5 generations of *local* training!