François Chollet ✓
@fchollet · United States · joined 25 Aug 2009
Co-founder @ndea. Co-founder @arcprize. Creator of Keras and ARC-AGI. Author of 'Deep Learning with Python'.
730 473Followers
829Following
25 719Posts total
4.4MViews on collected posts
Ultimi post
Cerchiamo nuovi post…
Inside you there are two wolves https://t.co/WqDFrc9KvO
42.5K views · 303 likes
· 13 reposts · 14 replies
Open on X →
How Keras 3 helped modernise Expedia's ranking stack
https://t.co/ReBzub7J9F
34.2K views · 43 likes
· 3 reposts · 20 replies
Open on X →
> One could even define intelligence as the efficiency with which one converts experience into competence; by this definition they lag very far behind us
Yes, one could define intelligence in this way, and one would be right. By this metric current AI is approximately 6 OOMs
64.1K views · 717 likes
· 66 reposts · 79 replies
Open on X →
You cannot separate "this person genuinely believes X, therefore they want Y" and "Y is in this person's interest". People will almost always align their beliefs and their interests.
You can be sincere *and* self-interested, and that's usually the default.
31.9K views · 345 likes
· 21 reposts · 38 replies
Open on X →
This is part of why the current frenzy is so frustrating: Virtually every single prominent entity speaking loudly about AI has a massive financial interest in things shaking out one way or another
126.7K views · 260 likes
· 33 reposts · 10 replies
Open on X →
One of the most worrying risks linked to frontier AI is extreme power concentration.
The only way to avoid extreme power concentration is to ensure we have multiple independent providers of frontier AI models, including open-source options.
69.9K views · 1.8K likes
· 262 reposts · 97 replies
Open on X →
In the near term (definitely not in the long term), more capable models should mean safer models (maybe paradoxically).
Current models are unsafe not because they're too smart, but because they take goals too literally or take nonsensical shortcuts to achieve these goals, i.e.
73.7K views · 1.3K likes
· 158 reposts · 155 replies
Open on X →
If there really is a high chance of AI leading to the extinction of humanity within years/decades, then the only rational stance towards safety monitoring and research pacing should be stringent, top-down government involvement and universally ratified international treaties.
191.9K views · 1.5K likes
· 181 reposts · 149 replies
Open on X →
About a year ago, before it was on anyone's radar, we began exploring the idea of a benchmark for open-ended invention. Since then, we've developed several promising directions that will serve as the foundation for ARC 4 and ARC 5.
We're incredibly excited to share what we've
63.1K views · 906 likes
· 61 reposts · 44 replies
Open on X →
I hope the proposals by frontier labs to pace AI research stem from a genuine concern for safety and a recognition of the potential risks posed by future models, rather than a strategic effort to consolidate power and permanently solidify the market dominance of a few top
90.3K views · 1.4K likes
· 235 reposts · 108 replies
Open on X →
A very common sentiment I'm hearing among math students: "I don't want to become a mathematician anymore." I've seen this before, I think the generation of digital artists who came of age around 2022 will be the last.
AI might not functionally replace mathematicians or artists,
170.2K views · 2.2K likes
· 150 reposts · 223 replies
Open on X →
Dreams are when your mind melts a little and then resolidifies by morning in a slightly different shape
28.1K views · 422 likes
· 25 reposts · 31 replies
Open on X →
Astra is making better memes than redditors, it's over
71.6K views · 1.1K likes
· 27 reposts · 70 replies
Open on X →
Terence Tao just posted this really insightful thread which seems to have been written in reaction to the Navier Stokes announcement. Please give it a read. https://t.co/SiuLLyp8jQ
184.9K views · 2.4K likes
· 273 reposts · 63 replies
Open on X →
The narrative around why building AGI was a good idea has always been about invention -- "it will cure cancer", "it will solve fusion energy", etc.
So I would not "declare AGI" until we have AI that is, at last, capable of invention -- conceptual breakthroughs, novel insights,
205.4K views · 2.8K likes
· 222 reposts · 213 replies
Open on X →
@fchollet Ineed it is, as I predicted, but there is a caveat, what is the compute/power rational to achieve it?
how about #NoGPU #NoLLM to solve ARCAGI3 this is what #Neuraxon is a real bioinspired brain architecture https://t.co/NQ4JiO9dK1
5.5K views · 190 likes
· 48 reposts · 2 replies
Open on X →
@fchollet how'd you determine these OOM differences?
471 views · 1 likes
· 0 reposts · 0 replies
Open on X →
When we released ARC 3, I got asked, "when do you think a frontier model will saturate it?", and I answered "in about a year, though it depends on how much it gets explicitly targeted"
That was 6 months ago, so the progress that Astra represents happened about 2x faster than I
36.4K views · 512 likes
· 38 reposts · 12 replies
Open on X →
Benchmarking AI systems is a continual process that co-evolves with the models. New benchmarks challenge AI capabilities with emerging questions to shape the directions and feedback signal of the research process. Then they adapt as models progress, targeting the residual between
36.3K views · 324 likes
· 10 reposts · 15 replies
Open on X →
Many of you will ask, "if it saturates ARC 3, is it AGI?"
We're not making this claim. All we know about the system so far are its benchmark scores.
When we launched ARC 3, and in every presentation we made about it, we were very insistent on one thing: solving it is not proof
57.2K views · 520 likes
· 28 reposts · 19 replies
Open on X →
GPT-6 Astra represents a step-function change in model capability for interactive reasoning problems. It scores 66% on ARC-AGI-3 using our standard harness, and nearly 100% with a continuous conversation harness and custom compaction, at a cost of roughly $360 per game.
In fact,
873.8K views · 6.8K likes
· 859 reposts · 177 replies
Open on X →
For a long time, the limits of AI deployment were only defined by technical capabilities. But as AI continues to make rapid progress, it needs to become a deliberate, collective choice about what kind of world we want to shape.
10.9K views · 115 likes
· 6 reposts · 9 replies
Open on X →
@fchollet I don’t think it’s possible to keep humans in the loop. It may be right now, but AI is advancing to rapidly. We’re going to get to a point very soon where we humans simply cannot keep up.
1.1K views · 18 likes
· 0 reposts · 4 replies
Open on X →
I believe we need to make a deliberate effort to keep humans in the loop in all critical processes across our economy and society, regardless of whether it is technically necessary.
Even if AI develops the *capability* for advanced autonomy, we should not make it highly
105.5K views · 2K likes
· 227 reposts · 206 replies
Open on X →
I wrote Deep Learning with Python to be the definitive guide to how deep learning works and how to best make use of it. Tens of thousands of people got their career start via this book. 120,000 copies sold, and downloaded by millions more.
And now it's free to read online:
780.5K views · 4K likes
· 623 reposts · 96 replies
Open on X →
@fchollet The meme thread is already up yo😉 https://t.co/L5AitaV1TA
7.3K views · 67 likes
· 1 reposts · 2 replies
Open on X →
Keras Core is a beta release. Please try it out, report issues, and we'll work hard to make the library the best it can be.
When ready, in Fall 2023, Keras Core will become Keras 3.0.
18.6K views · 149 likes
· 3 reposts · 1 replies
Open on X →
We're launching Keras Core, a new library that brings the Keras API to JAX and PyTorch in addition to TensorFlow.
It enables you to write cross-framework deep learning components and to benefit from the best that each framework has to offer.
Read more: https://t.co/xmmxBfSZgh h
966.3K views · 3.8K likes
· 760 reposts · 118 replies
Open on X →
You think I'm waiting for the official release?? Keras Core meme thread. Repost from here when appropriate #Keras3 https://t.co/PH6pdzCCRD
33.7K views · 211 likes
· 23 reposts · 8 replies
Open on X →
Rispetto ad account della stessa dimensione
24 post degli ultimi 90 giorni, accanto alla fascia di 100K–1M follower. arriva ben oltre il proprio pubblico, e quel pubblico reagisce.
Visualizzazioni mediane63 586questo account5 983mediana per 100K–1M
Copertura, %8.70%questo account1.49%mediana per 100K–1M
Interazione, %1.50%questo account0.93%mediana per 100K–1M
| Metrica | Questo account | Mediana per 100K–1M | Rapporto |
|---|---|---|---|
| Visualizzazioni mediane per post | 63 586 | 5 983 | 10.6× |
| Copertura (visualizzazioni ÷ follower) | 8.70% | 1.49% | 5.84× |
| Tasso di interazione | 1.50% | 0.93% | 1.62× |
Altri account di questa fascia → Confronta con un altro account → Come sono costruiti questi parametri →
Growth & engagement
How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.
Views per post
184.9K8 Sep
71.6K10 Sep
28.1K11 Sep
170.2K
90.3K12 Sep
63.1K
191.9K13 Sep
73.7K
69.9K
126.7K14 Sep
31.9K
64.1K
34.2K17 Sep
42.5K20 Sep
Last 14 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.
Engagement rate per post
1.49%8 Sep
1.72%10 Sep
1.70%11 Sep
1.52%
1.93%12 Sep
1.60%
0.95%13 Sep
2.22%
3.08%
0.24%14 Sep
1.28%
1.37%
0.20%17 Sep
0.78%20 Sep
Reactions — likes, reposts, replies and quotes — divided by views. Median for 100K–1M accounts is 0.93%.
What the audience does
Likes67.3%36 309 in total
Reposts8.1%4 356 in total
Replies3.7%1 983 in total
Quotes1.3%698 in total
Bookmarks19.6%10 597 in total
Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.
The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.
Account simili
Florcita ✓
@flor_vigna
1.5M followers
· 12.4K posts
サーティワン アイスクリーム
@BR31_Icecream
1.5M followers
· 656.4K posts
GURMEET CHOUDHARY ✓
@gurruchoudhary
1.5M followers
· 10.1K posts
natalia 🍒 ✓
@angelicfukdoll
1.5M followers
· 10.1K posts
諏訪部順一 Junichi Suwabe ✓
@MY_MURMUR
1.5M followers
· 38.8K posts
Cernovich ✓
@Cernovich
1.5M followers
· 330.7K posts
Gibran Rakabuming
@gibran_tweet
1.5M followers
· 6.7K posts
Lecrae ✓
@lecrae
1.5M followers
· 16.9K posts