@csaba_botos

curriculum design @GoogleDeepMind https://xitter.cf/t.co/WzU6h3qhag

 Oxford, England
Joined May 2017
  • Tweets365
  • Following396
  • Followers472
75Media
Botos Csabi retweeted
Oliver Habryka
@ohabrykaAug 20
Applications for the first round of lightconecommons.com are closing this Sunday! We'll likely facilitate $35M of grants by end of October. The application process is minimal and extremely flexible. If you have an ambitious project of almost any scale, send in what you have.

Lightcone Commons — An end-to-end platform for ambitious philanthropy

Lightcone Commons is a platform for coordinating ambitious philanthropy. Leverage evaluators with strong track records. Split the bill fairly with other funders. Maintain full control over your...

lightconecommons.com
6
37
6
306
45,623
Botos Csabi retweeted
Fanfei Li@fanfei__liAug 14
We hope LittleLearner + LittleCurriculum will serve as a controlled sandbox for studying learning beyond pretraining: from continual learning and interpretability to curriculum design and RL-driven discovery. 📄 Paper: arxiv.org/abs/2608.13545

LittleLearner: Language Models Under Pedagogically Controlled...

Modern language models are trained on heterogeneous web-scale text corpora. Consequently, studying knowledge and skill acquisition is difficult, as prior exposure to related content is hard to...

arxiv.org
1
6
2
191
15,398
Botos Csabi retweeted
Noémi Éltető@EltetoNoemiJun 19
First paper since joining @GoogleDeepmind! We present 🌍ATLAS (Active Theory Learning for Automated Science), a pipeline that generates interpretable mechanistic models from data and optimizes experiments to test them. Thread below
4
27
4
208
29,153
Botos Csabi retweeted
Sreejan Kumar@sreejan_kumarMay 11
I'm excited to share my new work joint with @csaba_botos: how well do frontier LRMs like DeepSeek-V4 learn novel video games and predict brain signatures associated with game learning? We found that they achieve SOTA on both game performance and brain predictivity! Thread 1/7
Botos Csabi@csaba_botosMay 11
Jeee 🐦‍⬛ I am very proud of our joint effort with @sreejan_kumar Kumar on the project "Reason to Play" LRMs show human-like rule discovery, and their hidden states predict human brain activity during gameplay. Interactive demo + paper: botcs.github.io/reason-to-pl…
1
6
29
3,171
Botos Csabi retweeted
Ferenc Huszár
@fhuszar22 Nov 2024
Call For Tasks for IOAI 2025: ML olympiad for high schoolers. I know you all have strong opinions on the ML topics future computer scientists should engage with to build mastery. This is a high-impact opportunity to help us by proposing tasks. ioai-official.org/call-for-t…

Call for tasks

ioai-official.org
4
6
1,785
Botos Csabi retweeted
Ferenc Huszár
@fhuszar15 Aug 2024
Teams of Hungary took 5th and 6th place in the first International Olympiad in AI taking the top 2 spots for silver 🥈. Congrats to the participants who brought a lot of self-taught expertise, adaptibility to the event continuing to build our tradition of problem solving. 🇭🇺🇭🇺🇭🇺
6
4
36
4,221
Gen AI is poised to transform many fields, sparking major debates over its risks & calls for tighter regulation. ❗Over-regulation could be catastrophic to open-source Gen AI. 🚀 Our paper (arxiv.org/pdf/2405.08597) argues the benefits of open-source Gen AI outweigh its risks. 🧵
4
33
13
110
21,451
1/n 🧵Happy to be presenting our recent work on Semi-Private Learning at @satml_conf in #Toronto @UofT with @yaxi_hu @AmartyaSanyal @FannyYangETH. Work done at @ETH_AI_Center @MPI_IS 📜Paper: openreview.net/forum?id=Ps1I… 🧑‍💻Code: github.com/FrancescoPinto/PI…
1
9
2
22
3,717
My deep learning course @unige_en is available on-line. 1000+ slides, ~20h of screen-casts. Full of examples in @PyTorch. fleuret.org/dlc/ And my "Little Book of Deep Learning" is available as a phone-formatted pdf (400k downloads!) fleuret.org/lbdl/
62
706
41
3,622
585,251
🤩We have an exciting opportunity to join TVG as a PDRA in Machine Learning, this position is in collaboration with Microsoft! Apply here: rb.gy/w7aum3 Deadline: 4th April @oxengsci @microsoft @Windows @surface Applied Sciences Group
13
1
18
3,584
You find some training data sources were compromised, potentially causing #backdoors, #bias, #mislabels etc. Can you remove its influence from previously trained #ML models instead of stopping their use? New work with @AmyPrb @AmartyaSanyal @ponguru @OxfordTVG studies this🧵👇
1
11
2
40
5,010
Congratulations to @AleksPPetrov for winning the Entropic Paper award at the 2023 @ICBINBWorkshop @NeurIPSConf When Do Prompting and Prefix-Tuning Work? A Theory of Capabilities and Limitations arxiv.org/abs/2310.19698
2
5
11
1,925
Botos Csabi retweeted
Ferenc Huszár
@fhuszar18 Dec 2023
I went on a tour giving talks on AI to high-schoolers in Transylvania. In the school I was going to visit on Wednesday, Tamási Áron School, a roof collapsed today, killing a teenager, leaving another in coma. I started a fundraiser for families and peers gofund.me/f146d1a7
4
19
35
6,568
Happy to share my recent work on using ML to study topological quantities on interesting Saskian and G2 geometries! arxiv.org/abs/2310.03064 If for some reason my tweet reaches someone interested, join my talk at 4pm BST! cam-ac-uk.zoom.us/j/65908220…

Machine learning Sasakian and $G_2$ topology on contact Calabi-Yau...

We propose a machine learning approach to study topological quantities related to the Sasakian and $G_2$-geometries of contact Calabi-Yau $7$-manifolds. Specifically, we compute datasets for...

arxiv.org
3
16
522
BREAKING NEWS The 2023 #NobelPrize in Physiology or Medicine has been awarded to Katalin Karikó and Drew Weissman for their discoveries concerning nucleoside base modifications that enabled the development of effective mRNA vaccines against COVID-19.
3,535
25,017
5,745
70,941
19,577,106
📢Worried about📢 1⃣ Terrible accuracies OR high computational cost with Differential Privacy ? 2⃣ Loose privacy parameters ε > 1 ? 3⃣ Unfair pre-training on datasets similar to private dataset ? Presenting a new semi-private algorithm 🚀🚀PILLAR🚀🚀 arxiv.org/abs/2306.03962
1
10
1
34
8,299
GPT-4 "discovered" the same sorting algorithm as AlphaDev by removing "mov S P". No RL needed. Can I publish this on nature? here are the prompts I used chat.openai.com/share/95693d… (excuse my idiotic typos, but gpt4 doesn't mind anyways)
Jim Fan
@DrJimFan7 Jun 2023
Sorting algorithm underpins all critical softwares. DeepMind's AlphaDev speeds up sorting small sequences (3-5 items) by 70%. Key takeaways: * The main RL algorithm is based on AlphaZero that originally played Go, Chess & Shogi. Same idea applies to searching programs! * Instead of optimizing over C code, they optimize assembly code instead. It's a deliberate choice to go low-level to squeeze out every instruction saving. * The assembly code is then reverse-engineered by hand to C, and open-sourced in LLVM. * Even though the representation network uses transformer, it is NOT a foundation model. The whole pipeline only works on sorting, and has to be re-trained for other tasks like hashing.
87
402
110
2,570
1,767,579
Botos Csabi retweeted
Chris Lu
@_chris_lu_6 Apr 2023
1/ 🚀 Presenting PureJaxRL: A game-changing approach to Deep Reinforcement Learning! We achieve over 4000x training speedups in RL by vectorizing agent training on GPUs with concise, accessible code. Blog post: chrislu.page/blog/meta-disco… 🧵
17
166
24
847
165,667
Botos Csabi retweeted
Gergely Neu@neu_rips2 Mar 2023
Replying to @OmarRivasplata @antoine_mln
here's all i know: 16:00-17:00: OxCSML Seminar, Gergely Neu [Large Lecture Theatre]
1
1
2
395
Dare I say...computer vision with human feedback is on its way in? Super happy to have contributed to this important piece of work during my internship at Google Brain!
Alexander Kolesnikov@__kolesnikov__17 Feb 2023
Vision meets RL! We reveal that policy gradient can be used for tuning vision models to optimize complex metrics, such as mAP, PQ or “color diversity”, observing large performance boosts on tasks like object detection, panoptic segmentation, etc. arxiv.org/abs/2302.08242
18
2
193
38,631
Load more