Jenny Zhang @jennyzhangzt Twitter profile

Pinned Tweet

Jenny Zhang

1 month

OMNI-EPIC!!! Done together with @maxencefaldor , @CULLYAntoine , and @jeffclune 📄Paper: 🌐Website: 🤩Huge shoutout to @rachpradhan for the website template. And really grateful for people who gave feedback along the way

Coming soon

omni-epic.vercel.app

Jeff Clune

@jeffclune

1 month

I am thrilled to introduce OMNI-EPIC: Open-endedness via Models of human Notions of Interestingness with Environments Programmed in Code. Led by @maxencefaldor and @jennyzhangzt , with @CULLYAntoine and myself. 🧵👇

10

41

204

9

19

101

Last Seen Profiles

@xxMBD6x

@cherrysplashart

@pasutri_jakarta

@tmichael1520

@fujihara_syoken

@ibkgngr420

@ArizonaFBall

@alialotaibi_9

@domiboop1

@FarzinFarhady

@lil8br

@_holyfool

@Novel19055488

@ffathle

@junata178

@NusaybahD12512

@hitthespace

@Shawnifee

@rqlxny

@19950808_aug

@yousuke2471

@CWmobil

@jandakembangstw

@SFGiants

@NStorages

@pengen_stw

@ArkActonAcademy

@xinxingegeg

@Pac12Network

@naandytwi

@picsofsakura

@musas_244

@longformprofile

@EYnews

@doyiieme

@DaudSimon161301

Jenny Zhang

@jennyzhangzt

1 month

Am I the only one who didn't know that the entire internet is literally wired physically with undersea cables?

2K

600

8K

Jenny Zhang

@jennyzhangzt

1 month

@srhnylmz14 Thank you for sharing! It was a really good read, a good reminder of the many things we don't see or think about often. “We are one of the best-kept secrets in the world, because things just work.”

10

14

589

Jenny Zhang

@jennyzhangzt

1 year

Fascinated by the idea of enabling AI to learn new, interesting behaviors forever? Started an awesome list for open-ended AI: . Make a pull request to contribute relevant works! 😎

GitHub - jennyzzt/awesome-open-ended: Awesome Open-ended AI

Awesome Open-ended AI. Contribute to jennyzzt/awesome-open-ended development by creating an account on GitHub.

github.com

6

29

164

Jenny Zhang

@jennyzhangzt

7 months

The best gift is the gift of knowledge. Happy holidays! 🎁

4

11

138

Jenny Zhang

@jennyzhangzt

5 months

🦖3D printed this model generated by text-to-3D foundation model! Resin printer No other changes than upscaling and adding support (done automatically through an app) Result is pretty good!🤩 Conclusion: models generated by text-to-3D FMs are printable if smooth enough

3

25

134

Jenny Zhang

@jennyzhangzt

5 months

OMNI is accepted at #ICLR2024 ! Updated paper has exciting new results in an embodied robotics domain, where AI learns new interesting tasks endlessly in an INFINITE task space. Open-endedness through Models of Human Notions of Interestingness Website: 🧵👇

2

29

107

Jenny Zhang

@jennyzhangzt

7 months

Look mum, it's me!

Julian Togelius

@togelius

7 months

Good panel is good. (On open-endedness, at @aloeworkshop #NeurIPS2023 )

1

2

31

2

3

103

Jenny Zhang

@jennyzhangzt

23 days

ARC is a pretty interesting challenge and dataset! I briefly explored if debate between LLMs could improve the performance on the ARC dataset. 👉 LLM-debate paper: ARC prize:

ARC Prize

ARC Prize is a $1,000,000+ nonprofit, public competition to beat and open source a solution to the ARC-AGI benchmark.

arcprize.org

François Chollet

@fchollet

26 days

I'm partnering with @mikeknoop to launch ARC Prize: a $1,000,000 competition to create an AI that can adapt to novelty and solve simple reasoning problems. Let's get back on track towards AGI. Website: ARC Prize on @kaggle :

87

575

3K

5

15

93

Jenny Zhang

@jennyzhangzt

2 months

Thank you to all #ICLR2024 organizers! Had a blast catching up with old friends and meeting new ones 🌈 Also a full supporter of more colorful posters

3

4

89

Jenny Zhang

@jennyzhangzt

9 months

Update on OMNI (Open-endedness via Models of human Notions of Interestingness). We tried it on a new, off-the-shelf domain (BabyAI), and OMNI outperforms baselines there too! Website: 1/

2

13

78

Jenny Zhang

@jennyzhangzt

2 months

I will be in Vienna next week for #ICLR2024 , excited to discuss anything fire in AI right now (or in future) 🔥 Come drop by our posters too! 🕸️ Project websites OMNI: QDAIF:

1

9

79

Jenny Zhang

@jennyzhangzt

1 year

Thank you to all the organizers of @ieee_ras_icra ! My first robotics conference and had so much fun trying cool robots, learning what others are doing, and meeting old and new friends. Let's keep in touch! #ICRA2023

1

10

68

Jenny Zhang

@jennyzhangzt

7 months

Thank you to all #NeurIPS2023 organizers! I had so much fun talking science (and non-science) with everyone, looking forward to seeing more cool works🤩 Presented works👇 OMNI: QDHF: QDAIF:

0

4

66

Jenny Zhang

@jennyzhangzt

1 year

Very happy to share this new work! Done together with @joelbot3000 , @kenneth0stanley , and @jeffclune . And very grateful to everyone who gave me feedback and engaged in insightful discussions while doing this work 🙏

Jeff Clune

@jeffclune

1 year

Excited to share OMNI: Open-endedness via Models of human Notions of Interestingness. Lead: @jennyzhangzt Open-ended learning requires a vast space of possible tasks, but search thus can get lost. Agents must focus only on *interesting* tasks, but how?🧵1/

8

49

210

2

5

59

Jenny Zhang

@jennyzhangzt

7 months

I am in NOLA for #NeurIPS2023 ! Message me to chat about reinforcement learning, open-endedness, or the road to AGI 🚀 Will also be presenting OMNI: Open-endedness via Models of human Notions of Interestingness at the @aloeworkshop on Friday!

2

4

50

Jenny Zhang

@jennyzhangzt

10 months

No more to manual engineering efforts in selecting the most suitable states for most desired locomotion behaviors.

Identifying important sensory feedback for learning locomotion skills

Nature Machine Intelligence - Traditional feedback-state selection in robot learning is empirical and requires substantial engineering efforts. Yu et al. develop a quantitative and systematic...

www.nature.com

1

9

37

Jenny Zhang

@jennyzhangzt

1 year

I will be in London for #ICRA2023 @ieee_ras_icra . Excited to present my work on Good Time to Ask on 29 May at the Communicating Robot Learning across Human-Robot Interaction workshop!

0

3

33

Jenny Zhang

@jennyzhangzt

24 days

Just the start of what automating AI innovation might look like! 💥

Sakana AI

@SakanaAILabs

25 days

Can LLMs invent better ways to train LLMs? At Sakana AI, we’re pioneering AI-driven methods to automate AI research and discovery. We’re excited to release DiscoPOP: a new SOTA preference optimization algorithm that was discovered and written by an LLM!

19

254

1K

0

1

32

Jenny Zhang

@jennyzhangzt

1 year

Happy to share that our work Good Time to Ask was awarded the best paper award at Ubiquitous Robotics 2023! Thank you to Korea Robotics Society (KROS) for organizing and hope to keep in touch with everyone I met.

4

1

28

Jenny Zhang

@jennyzhangzt

8 months

In just a few months of starting the awesome open-ended AI list, I have come across many new notable papers. Make a pull request to contribute relevant works! 😎

GitHub - jennyzzt/awesome-open-ended: Awesome Open-ended AI

Awesome Open-ended AI. Contribute to jennyzzt/awesome-open-ended development by creating an account on GitHub.

github.com

0

10

28

Jenny Zhang

@jennyzhangzt

9 months

Enough of manually defining diversity, let's use human feedback! Led by Li Ding @li_ding_ , with Jeff Clune @jeffclune , Lee Spector, and Joel Lehman @joelbot3000

Li Ding

@li_ding_

9 months

🚀Excited to share Quality Diversity through Human Feedback (QDHF): . QDHF learns diversity metrics from human judgments of difference, and generates diverse and high-quality responses for tasks such as text-to-image generation. 1/

2

5

39

0

3

27

Jenny Zhang

@jennyzhangzt

2 months

another pretty awesome poster ✨

0

6

17

Jenny Zhang

@jennyzhangzt

5 months

Text-to-3D foundation model used: 🙏 Amazing work by team!

GitHub - 3DTopia/3DTopia: Text-to-3D Generation within 5 Minutes

Text-to-3D Generation within 5 Minutes. Contribute to 3DTopia/3DTopia development by creating an account on GitHub.

github.com

0

2

16

Jenny Zhang

@jennyzhangzt

9 months

Amazing work and written so well!! It doesn't just design better prompts, but also improves in designing better prompts. I wonder if the authors ever got confused discussing the meta parts. "mutated mutation prompts"

Chrisantha Fernando

@chrisantha_f

9 months

🌱 Introducing Promptbreeder: LLMs evolve their own prompts through self-referential self-improvement! Paper: #PromptEngineering #LLM #AI #ML #Promptbreeder

12

73

363

1

3

15

Jenny Zhang

@jennyzhangzt

2 months

🎉 Congratulations to @li_ding_ !!! 💥 The QDHF tutorial on Pyribs takes less than 1 hour to try!

Incorporating Human Feedback into Quality Diversity for Diversified Text-to-Image Generation

docs.pyribs.org

Li Ding

@li_ding_

2 months

Exciting news! Our QDHF paper has been accepted at #ICML2024 ! Huge thanks to incredible coauthors: @jennyzhangzt , @jeffclune , Lee Spector, and @joelbot3000 . Stay tuned for updates on arXiv with our latest findings! Project website: 👇

2

9

34

0

2

16

Jenny Zhang

@jennyzhangzt

16 days

Built something fun over the last few days ⚡AI wallpapers⚡ Available on iOS and Android

AiWallz - AI Wallpapers

Ai Wallpapers - Unparalleled quality. Simply Stunning. Wallpapers, redefined.

www.aiwallz.com

1

2

14

Jenny Zhang

@jennyzhangzt

7 months

@jeffclune @aloeworkshop @DrJimFan @robertarail @snikolaidis19 @MasterMilkX @_rockt Really honored to be on the panel 🙏 A tiny glimpse of how much we talk about open-endedness in our lab

0

11

Jenny Zhang

@jennyzhangzt

10 months

Reading the caption before seeing the image, it's not quite what I expected. Where are the thrones? Are the potatoes sure that there are multiple kings?

OpenAI

@OpenAI

10 months

"Tiny potato kings wearing majestic crowns, sitting on thrones, overseeing their vast potato kingdom filled with potato subjects and potato castles."

23

85

852

0

10

Jenny Zhang

@jennyzhangzt

2 years

A year ago, nobody in the lab I was working at then has heard of Quality-Diversity (QD) algorithms. But now, we see increasingly more concepts from QD algorithms used in open-ended research from big companies like DeepMind and OpenAI.

0

8

Jenny Zhang

@jennyzhangzt

1 year

Super excited about this new direction! Thank you @herbiebradley and @joelbot3000 for spearheading this

0

9

Jenny Zhang

@jennyzhangzt

5 months

Essential to training an agent capable of learning ANY task is the development of what we call a “Universal Reward Function”, which can evaluate if *any* task has been completed or not. As a stand-in, we have OMNI generate executable code to assess task completion 2/

1

9

Jenny Zhang

@jennyzhangzt

5 months

Looking forward to presenting OMNI in Vienna! OMNI was led by me, with Joel Lehman @joelbot3000 , Ken Stanley @kenneth0stanley , and Jeff Clune @jeffclune . Original OMNI tweet: 3/3

Jeff Clune

@jeffclune

1 year

Excited to share OMNI: Open-endedness via Models of human Notions of Interestingness. Lead: @jennyzhangzt Open-ended learning requires a vast space of possible tasks, but search thus can get lost. Agents must focus only on *interesting* tasks, but how?🧵1/

8

49

210

0

7

Jenny Zhang

@jennyzhangzt

10 months

This is so clever ⚽ : French ad showing how women's soccer is just as good as men's

Orange - la Compil des Bleues (English version)

null

www.youtube.com

0

3

8

Jenny Zhang

@jennyzhangzt

6 months

Favourite feature of Maven so far is the option for serendipity mode 👾

2

0

8

Jenny Zhang

@jennyzhangzt

1 year

Everytime I read numbers like this...

1

0

7

Jenny Zhang

@jennyzhangzt

1 year

Rewatching Doraemon for AI ideas

AK

@_akhaliq

1 year

Doraemon predicted all the AI stuff around 44 years ago

30

503

2K

0

8

Jenny Zhang

@jennyzhangzt

2 years

Started my PhD at @UBC_CS working with @jeffclune just a month ago. Planning for a research project is just as important as its execution or analysis. But it does make you feel itchy to start coding.

0

7

Jenny Zhang

@jennyzhangzt

1 year

Created my first web3 app using @Livepeer API over the weekend at @nwplusubc hackathon! It was such a great experience, and I learnt a lot about web3 services from @laxbrownie , @shanvasion , Qiang Han and more. Glad people liked my app enough to win the Livepeer category!😊

1

0

6

Jenny Zhang

@jennyzhangzt

1 year

@alec_helbling @manim_community Insertion Sort visualized using only the code shown on the right side

0

1

6

Jenny Zhang

@jennyzhangzt

6 months

Has anyone tried 3D printing models generated by these FMs? How did it go?

Ziwei Liu

@liuziwei7

6 months

🔥Text-to-3D Foundation Model🔥 We are excited to announce #3DTopia , a generalist 🧊text-to-3D🧊 foundation model, which produces ** high-quality 3D assets within 5 minutes ** - Code: - Video:

9

101

382

1

0

6

Jenny Zhang

@jennyzhangzt

1 year

I added a few works to start off, but forgive me if I missed out some/ many others

0

6

Jenny Zhang

@jennyzhangzt

8 months

I am not ready for a dystopian future, please make AI safe!

Jeff Clune

@jeffclune

9 months

I, with 3 Turing Award winners (including @geoffreyhinton , Bengio, Yao), Nobel laureate @kahneman_daniel , & other leaders in AI and AI governance, release a paper today recommending specific government & company actions on AI development to minimize risk

6

42

183

0

6

Jenny Zhang

@jennyzhangzt

9 months

OMNI was led by me, with Joel Lehman @joelbot3000 , Ken Stanley @kenneth0stanley , and Jeff Clune @jeffclune . Original OMNI tweet: 5/5

Jeff Clune

@jeffclune

1 year

Excited to share OMNI: Open-endedness via Models of human Notions of Interestingness. Lead: @jennyzhangzt Open-ended learning requires a vast space of possible tasks, but search thus can get lost. Agents must focus only on *interesting* tasks, but how?🧵1/

8

49

210

0

6

Jenny Zhang

@jennyzhangzt

4 months

@jeffclune RL agent slices EVERYTHING to "slice the apple". Goal achieved, but do we really want this behavior?

0

2

5

Jenny Zhang

@jennyzhangzt

9 months

We utilize large (language) models (LMs) as models of interestingness (MoI), because they *already* internalize human concepts of interestingness from training on vast amounts of human-generated data, where humans naturally write about what they find interesting. 3/

1

0

5

Jenny Zhang

@jennyzhangzt

1 year

photoshopping with AI

0

4

Jenny Zhang

@jennyzhangzt

1 year

Come join our lab!

Jeff Clune

@jeffclune

1 year

Seeking a postdoc to join my lab at UBC! Interested in combining deep RL & large language models to advance open-endedness? Candidates should have pubs in one of: RL, large models, open-endedness, or related areas. 1-2+-year positions in paradise! Apply:

4

39

115

0

4

Jenny Zhang

@jennyzhangzt

2 months

@hannah_teufel Thank you @hannah_teufel !!! 😊

0

2

Jenny Zhang

@jennyzhangzt

3 months

A good laugh to start off the day! Wildest reasons for resignation 😶‍🌫️

Rach

@rachpradhan

3 months

🚀 Need a good reason to resign? has got your back📝 🤖 Powered by @OpenAI 's GPT-3.5 Turbo, @nextjs 's AI streams and @vercel 's edge, it offers intelligent, personalized letter drafting. 🎨 Thanks to @tiptap_editor & @shadcn 's brilliance, editing is a

2

6

13

0

1

4

Jenny Zhang

@jennyzhangzt

9 months

@mayalen_etc @FlowersINRIA @drmichaellevin @pyoudeyer @Clement_MF_ This is awesome! 💯

0

3

Jenny Zhang

@jennyzhangzt

9 months

Open-ended learning requires a vast space of possible tasks, but search thus can get lost. A core challenge in open-endedness is the ability to quantify and thus prioritize *interesting* tasks. OMNI solves that by modeling human notions of what is interesting. 2/

1

0

4

Jenny Zhang

@jennyzhangzt

2 months

@shahdhruv_ @berkeley_ai @ajaysridhar0 @CatGlossop @svlevine Congratulations @shahdhruv_ !

0

1

Jenny Zhang

@jennyzhangzt

16 days

Why racoons?

1

0

3

Jenny Zhang

@jennyzhangzt

2 years

Just finished reading this over the holidays. Recommending it to everyone!

Why Greatness Cannot Be Planned

books.google.com

1

0

3

Jenny Zhang

@jennyzhangzt

9 months

That OMNI works in a new, off-the-shelf domain highlights its generality. We look forward to the future directions OMNI will inspire in open-endedness research, empowering AI to join the never-ending dance of creativity that has shaped our civilization. 4/

1

0

3

Jenny Zhang

@jennyzhangzt

2 months

@bryanlimwt @CULLYAntoine Congratulations Bryan!!! Excited to see what else you do Adding to your fun fact, Antoine went through C++ code with me line by line when I first joined

1

0

3

Jenny Zhang

@jennyzhangzt

2 years

Makes me think of the heavy emphasis on publications for grad school applications, and how many applicants chase this objective instead of actually honing their technical/ analytical skills or doing something they are really passionate about

0

3

Jenny Zhang

@jennyzhangzt

1 year

@jayelmnop @jeffclune @allisontam_ Thanks for sharing, will check them out.

0

3

Jenny Zhang

@jennyzhangzt

6 months

@thtrieu_ This is awesome, congratulations! How did you decide on the domain?

0

2

Jenny Zhang

@jennyzhangzt

11 days

@_rockt Yesss I really liked that part of OMNI-EPIC too, thank you @_rockt !

0

1

2

Jenny Zhang

@jennyzhangzt

9 months

@Jesse_Y_Zhang I enjoyed reading and thinking about this work! 🙌

0

1

Jenny Zhang

@jennyzhangzt

1 month

@robertarail @maxencefaldor @CULLYAntoine @jeffclune @rachpradhan Thank you @robertarail !!

0

2

Jenny Zhang

@jennyzhangzt

5 months

@imotarite99747 That's awesome! It kind of looks like godzilla. I actually didn't do much editing from the 3D model generated by 3DTopia

0

1

Jenny Zhang

@jennyzhangzt

1 year

@DavidSKrueger Thank you for sharing! AI and potentially AGI can be considered a great human civilization achievement. However, we are uncertain if it will bring more good/harm. The core problem doesn't seem like any specific technological advancement, but how we treat each other. Thoughts?

0

2

Jenny Zhang

@jennyzhangzt

7 months

Also up for a game of tennis if anyone is interested! 🎾

0

2

Jenny Zhang

@jennyzhangzt

6 months

@bhatia_mehar @GlobalBC @JenPalmaGlobal @VeredShwartz @Alex_Walls_ @UBCNews @UBC_NLP @UBC_CS @VectorInst @UBC Awesome!! 🤩

0

1

Jenny Zhang

@jennyzhangzt

1 month

@sharathraparthy @maxencefaldor @CULLYAntoine @jeffclune @rachpradhan Thank you!!

0

2

Jenny Zhang

@jennyzhangzt

8 months

@jparkerholder @li_ding_ Thank you!

0

2

Jenny Zhang

@jennyzhangzt

22 days

@tyleryzhu definitely a fan of self-play or evolutionary approaches! it might be more computationally inefficient, but there are libraries that can run such algorithms a lot faster now! evojax: qdax:

0

2

Jenny Zhang

@jennyzhangzt

1 year

@alec_helbling @manim_community Beautiful animations! Reminds me of one of my undergraduate projects using @manim_community - Automatic animation generation for Data Structures

GitHub - jennyzzt/AlgoManim

Contribute to jennyzzt/AlgoManim development by creating an account on GitHub.

github.com

1

2

Jenny Zhang

@jennyzhangzt

1 month

@tongzhou_mu @maxencefaldor @CULLYAntoine @jeffclune @rachpradhan Working hard to get it out asap!

0

1

Jenny Zhang

@jennyzhangzt

6 months

@alexUnder_sky It's a new social network platform where you follow your interests! It enables you to really engage with what you are interested in or start a conversation about what you care about and hear other's opinions.

Kenneth Stanley

@kenneth0stanley

6 months

Announcing Maven: We’ve created a new kind of social network - a serendipity network - that’s directly inspired by insights from open-endedness and Why Greatness Cannot Be Planned. iPhone: Android: It's very different.... More 👉