XOutlier
Download .zipAdd to Chrome

Sebastien Bubeck

@SebastienBubeck93k followers

I work on AI at OpenAI. Former VP AI and Distinguished Scientist at Microsoft.

Open on X
Top outlier
549×
Best organic post
Peak views
7.5M
Single post
Bookmarks
24k
Across 87 organic posts
Last scan
Oct 16
Community cache
Posts 87 posts
#DatePost↓ OutlierViewsLikesRepostsBookmarks
1Aug 20, '25Claim: gpt-5-pro can prove new interesting mathematics. Proof: I took a convex optimization paper with a clean open problem in it and asked gpt-5-pro to work on it. It proved a bet549×7.5M8k1.2k3.3k
2Mar 23, '23At @MSFTResearch we had early access to the marvelous #GPT4 from @OpenAI for our work on @bing. We took this opportunity to document our experience. We're so excited to share our f150×2.1M2.8k6771.9k
3Sep 28, '25Yet more evidence that a pretty major shift is happening, this time by Scott Aaronson80.6×1.1M3.4k4261.8k
4Jun 21, '23New LLM in town: ***phi-1 achieves 51% on HumanEval w. only 1.3B parameters & 7B tokens training dataset*** Any other >50% HumanEval model is >1000x bigger (e.g., WizardC69.6×953k1.6k311837
5Oct 3, '25Well, this time it's by Terence Tao himself:62.2×852k3.3k3721k
6Jan 8, '25Enjoy!46.1×631k3.5k3861.1k
7Oct 20, '25My posts last week created a lot of unnecessary confusion*, so today I would like to do a deep dive on one example to explain why I was so excited. In short, it’s not about AIs dis45.1×618k1.7k215727
8Oct 12, '25gpt5-pro is superhuman at literature search: it just solved Erdos Problem #339 (listed as open in the official database ) by realizing that it had actually been solved 20 years ago38.4×526k1.5k119455
9Apr 23, '24phi-3 is here, and it's ... good :-). I made a quick short demo to give you a feel of what phi-3-mini (3.8B) can do. Stay tuned for the open weights release and more announcements 35.5×486k906173291
10Nov 28, '23We trained a small transformer (100M params) for basic arithmetic. W. the right training data it nails 12x12 digits multiplication w/o CoT (that's 10^24 possibilities, so no it's n33.5×458k1.9k239992
11Mar 10, '23The Chomsky et al. opinion piece in the @nytimes about ChatGPT is making the rounds. Rather than trying to deconstruct their argument, I asked @bing what it thinks of it. Now you c32.5×445k1.4k270200
12Sep 24, '25It's becoming increasingly clear that gpt5 can solve MINOR open math problems, those that would require a day/few days of a good PhD student. Ofc it's not a 100% guarantee, eg belo30.2×413k2.1k253644
13Jan 31, '25o3-mini is a remarkable model. Somehow it has *grokked arxiv* in a way that no other model on the planet has, turning it into a valuable research partner! Below is a deceitfully si26.2×359k1.2k107450
14Oct 18, '25I deleted the post, I didn't mean to mislead anyone obviously, I thought the phrasing was clear, sorry about that. Only solutions in the literature were found that's it, and I find23.4×320k1.1k33183
15Jan 6, '24Starting the year with a small update, phi-2 is now under MIT license, enjoy everyone!20.9×287k1.8k257357
16Nov 15, '23Microsoft💜Open Source + SLMs!!!!! We're so excited to announce our new *phi-2* model that was just revealed at #MSIgnite by @satyanadella! At 2.7B size, phi-2 is much more robust 20.7×283k46280180
17Dec 20, '24o3 and o3-mini are my favorite models ever. o3 essentially solves AIME (>90%), GPQA (~90%), ARC-AGI (~90%), and it gets 1/4th of the Frontier Maths. To understand how insane 25%20.6×283k1.5k158379
18Dec 13, '23Enjoy everyone! (And remember it's a base model so you might have to play around with your prompts; if you want it to follow instructions you can try the format "Instruct:... Ouput19.4×266k1.3k178481
19Jul 19, '25It’s hard to overstate the significance of this. It may end up looking like a “moon‑landing moment” for AI. Just to spell it out as clearly as possible: a next-word prediction mach19.1×261k1.4k154373
20Oct 22, '25Are you ready for another twist in the Erdos problems story? Today a $1000 prize open problem (#707 ) was solved by two humans who … realized that the problem was solved 30 years b18.3×250k69576211
21Aug 7, '25Here at @OpenAI we've cracked pretraining, then reasoning, and now we're experimenting with a new set of techniques that maximally leverage their interaction. GPT-5 is just the fir18.2×249k6095495
22Mar 25, '25View post on X18.2×249k4k205310
23Dec 12, '23Phi-2 numbers, finally! We're seeing a consistent ranking: phi-2 outperforms Mistral 7B & Gemini Nano 2* (*on their reported benchmarks) and is roughly comparable to Llama 2-7017.9×245k56071164
24Sep 12, '23How far does one billion parameters take you? As it turns out, pretty far!!! Today we're releasing phi-1.5, a 1.3B parameter LLM exhibiting emergent behaviors surprisingly close to17.8×244k785169325
25Aug 7, '25Awwww, working daily with you guys is the highlight of my career, and I have really high hopes that we have barely gotten started! 💜17.4×239k7022881
26Dec 13, '23Sorry I know it's a bit confusing: to download phi-2 go to Azure AI Studio, find the phi-2 page and click on the "artifacts" tab. See picture.14.3×196k47351295
27Apr 6, '23Last couple of weeks I gave a few talks on the Sparks paper, here is the MIT recording! The talk doesn't do justice to all the insights we have in the paper itself. Neither talk no13.7×187k549282304
28Nov 24, '23Join us on YouTube at 1pm PT/4pm ET today for the premiere of our "debate" with @bgreene @ylecun @tristanharris on whether a new kind of intelligence has emerged with GPT-4, and wh13.2×181k22541176
29Aug 31, '24This is the coolest "application" of LLMs I have seen in a while: instead of using an LLM to solve a problem set, flip the table and ask the student to TEACH the LLM how to solve t11.1×152k751127779
32Oct 5, '25I guess it's now every day until the end of time9.9×135k98658215
37Nov 13, '23My group is hiring a large cohort of interns for the summer of 2024 to work on the Foundations of Large Language Models! Come help us uncover the new physics of A.I. to improve the7.8×107k38448297
38May 2, '24She is coming back, soon ....7.6×104k3291830
39Oct 30, '24Just started at @OpenAI and I couldn't be more excited to join at this pivotal moment of safe AGI development! Met so many old friends already, talent density of this place is just7.6×104k8942765
40Dec 15, '23phi-2 is really a good base for further fine-tuning: we FT on 1M math exercises (similar to phi-1 w. CodeExercises) & test on recent French nation-wide math exam (published aft7.6×103k49764187
41Dec 20, '24I recommend to tune in!!!7.4×101k451827
42Dec 19, '23We're so pumped to see phi-2 at the top of trending models on @huggingface ! It's sibling phi-1.5 has already half a million downloads. Can't wait to see the mechanistic interpreta7.2×99k2926837
43Dec 13, '24Surprise #NeurIPS2024 drop for y'all: phi-4 available open weights and with amazing results!!! Tl;dr: phi-4 is in Llama 3.3-70B category (win some lose some) with 5x fewer paramete6.9×95k40568143
44Aug 11, '25I copy pasted an unpublished manuscript of mine in ChatGPT and asked it to improve it. I expected that the method we're using was pushed to its limit: gpt-5-pro actually proved it.6.2×85k62945171
45Jun 6, '24Every day I witness the AI revolution in action, and every day I see 1 or 2 questions that would deserve an entire PhD thesis to explore fully ... Honestly, how lucky we are to do 6.1×83k3053672
46May 21, '24Can't wait for your feedback on our largest and most capable Phi model yet!5.9×81k4435995
47Jan 15, '25sparks of asi when???5.6×77k4452049
49Dec 13, '23phi-2 is coming to Hugging Face, hold tight :-)5.2×71k5122824
50Nov 5, '25Looks like a pretty long CoT5.1×70k6732073
52Feb 27, '25What could that be? 🤔5.0×69k406911
53Jun 20, '24Just to state the obvious, AGI might not happen with current techniques. At a very basic level the argument for AGI is: if you compress sufficiently complex material, like the web,4.8×65k22826144
54Sep 21, '25Textbooks are all you need ;-)4.6×62k39610112
55May 2, '24I'm super excited by the new eval released by Scale AI! They developed an alternative 1k GSM8k-like examples that no model has ever seen. Here are the numbers with the alt format (4.4×60k2461987
56Apr 23, '24Game on!4.2×58k3384871
57Jun 10, '22Transformers are changing the world. But how do they learn? And what do they learn? Our 1st @MSFTResearch ML Foundations team paper proposes a synthetic task, LEGO, to investigate 4.1×0695107304
58Feb 27, '25Maybe pretraining still works after all??4.1×56k5935558
60Jun 18, '23Terence Tao reflecting on GPT-4 in the AI Anthology coordinated by @erichorvitz : "I expect, say, 2026-level AI, when used properly, will be a trustworthy co-author in mathematical4.1×56k2944685
61Aug 16, '25I really recommend trying gpt5-pro ... For research level questions it is in a league of its own imo3.7×51k3011349
62Jun 7, '23I cannot recommend this podcast episode strongly enough. It's simply THE MOST INSIGHTFUL 2 hours content that you can find on LLMs. And it's by none others than @EldanRonen and Yua3.7×51k29040382
63Dec 16, '22Why do neural networks generalize? IMO we still have no (good) idea. Recent emerging hypothesis: NN learning dynamics discovers *general-purpose circuits* (e.g., induction head in 3.7×50k30242149
64May 27, '24Updated phi-3 tech report with final numbers for 7B/14B and a new section on phi-3-V (e.g., MMMU at 40.4, in the ballpark of Claude 3-haiku and Gemini-1.0 pro) :3.6×50k23550114
65Mar 3, '25Uniformly number 1 with fully classical pre-training; there is no wall?!? (Actually, there is no wall for at least 10 different reasons, but that's a different topic.) In any case,3.5×47k3781649
66Feb 3, '25Deep Research on the query below comes back with these references, which are just **excellent** (basically what a grad student would come back with the next day after being given t3.5×47k2641168
67Jun 9, '21We may have found a solid hypothesis to explain why extreme overparametrization is so helpful in #DeepLearning, especially if one is concerned about adversarial robustness. 1/73.4×0636118185
70May 1, '25Btw3.2×43k40363113
71Jun 30, '22The video of my talk @EPFL_en today on Transformers and how to make sense of them is online!3.0×048365243
72Jan 7, '25This year I turn 40. Would be nice to celebrate half-life with AGI ...2.9×40k313727
73Aug 1, '25One thing folks outside of OpenAI don't necessarily realize, is how TRULY FORWARD LOOKING the leadership team is: they're happy to take short term hits if it means doing the right 2.9×40k2811052
74Aug 2, '25It's happening ...2.9×39k3832467
76Aug 5, '19For my 500th tweet I'm super excited to release five 1h videos covering the most important results presented in my monograph Convex Optimization: Algorithms and Complexity. This ti2.6×049698128
77Feb 10, '21(Nesterov) Acceleration in convex optimization is one of the most striking phenomenon in all of optimization, and now you can learn about all the different viewpoints on it from a 2.4×046791117
78Nov 3, '21Just watched an incredible talk by @AlexGDimakis at the Simons Institute, highly recommended. Their Iterative Layer Optimization technique to solve inverse problems with GANs make 2.4×043975144
79Sep 30, '25The unicorn challenge got upgraded2.3×31k3312444
80Dec 30, '19Time for some retrospective! I shared some 25 papers that I particularly enjoyed in the last decade. I would love for you to share some papers that are missing in this list (there 2.1×04069992
81Jul 10, '19A major open problem in ML is whether convex techniques (kernel methods in particular) can reproduce the striking successes of deep learning. In a guest post series (two parts) on 2.1×044211456
82Apr 20, '22We (@gauthier_gidel @velythyl @busycalibrating @vernadec & myself) would like to announce the accepted blog posts to @iclr_conf's 1st Blogpost Track. Experiment was a great suc1.8×03496479
83Jun 7, '25Just finished first triathlon (Olympic) of the season! I'm really happy with this one, the training was extremely minimal given all the exciting things happening at OAI that are ke1.8×25k26438
84Dec 2, '24This should be interesting!1.7×24k23129103
85Oct 26, '20The **Machine Learning Foundations** group at @MSFTResearch Redmond is hiring at all levels (including postdoc)! Come join @ZeyuanAllenZhu @suriyagnskr @jerryzli @ilyaraz2 @talw an1.7×03218058
86Jan 20, '22New video! Probably best described as "a motivational speech to study deep learning mathematically" :-). The ever so slightly more formal title is "Mathematical theory of deep lear1.6×026635134
87Dec 1, '21I'm really happy that the law of robustness got recognized as an important new insight with a NeurIPS outstanding paper award! The video below summarizes what the law is about, wha1.6×03473959
89Nov 25, '22Microsoft Research is hiring (in-person) interns! There are many different opportunities in all the labs. Here are some options in the Machine Learning research area in MSR@Redmond1.5×025149104
90Aug 24, '21The universal law of robustness is a tentative theoretical justification for *large* overparametrization in neural network learning. Here is a video explaining the law, in the cont1.5×02546180
91Nov 26, '20Karen Uhlenbeck concludes her Abel prize lecture with 5 minutes on #DeepLearning!!! She says about it: ``My conjecture is there is some interesting mathematics of some sort that I 1.4×03033953
92Dec 3, '18At #NeurIPS2018 where it was just announced that our paper on non-smooth distributed optimization with Kevin Scaman, @BachFrancis , Laurent Massoulie and Yin Tat Lee got a best pap1.3×03664710
93Nov 30, '20New video with a crash course on *tensors* (spoiler: no they aren't JUST multi-dimensional arrays!). Include discussion of cross norms & basic facts about rank. We then use it 1.3×02572582
94Nov 20, '18A fun way to describe Nesterov's momentum:1.2×03314719
95Jan 26, '21Interesting thread! To me the ``reason" for CLT is simply high-dim geometry. Consider unit ball in dim n+1 & slice it at distance x from the origin to get a dim n ball of radiu1.2×02372766
96Jul 17, '19Part II of @julienmairal guest post on CNN-inspired kernel methods: you will learn how to efficiently approximate those kernels, and even push the CNN analogy further by doing an e1.1×02415339
97Jul 15, '19Every Mon/Thu I will post a 1h lecture on the ``Five Miracles of Mirror Descent". We start with basic reminders of convexity, the classical analysis of gradient descent, and a disc1.0×02573919
98Jun 20, '22AI still has a long way to go .... to me this example is exactly what happened with the whole "sentient discussion": if you prompt with the seed of an answer, the transformer archi1.0×02502023
99Dec 8, '19Congratulations to our colleague Lin Xiao @MSFTResearch for the #NeurIPS2019 test of time award!!! Online convex optimization and mirror descent for the win!! (As always? :-).)1.0×0242437
100Mar 17, '21Congratulations to Laslo Lovasz and Avi Wigderson for winning the 2021 Abel Prize!!!!!!! What a fantastic recognition for theoretical computer science from the mathematics communit0.8×0228312

Scan another creator

@
@karpathy@levelsio@paulg@naval@marclou@illyism

Run this on any account.

XOutlier builds this table for every public profile you open on x.com, with live data and every post, not just the top 50.

Add to Chrome Download .zip