Fani ✨ (@fanyak_) profile picture

Thessaloniki born and raised

ΘΕΣΣΑΛΟΝΙΚΗ - Thessaloniki
Joined January 2009
Photo by Fani ✨ (@fanyak_)Photo by Fani ✨ (@fanyak_)Photo by Fani ✨ (@fanyak_)
If you have read the DDIA book (and liked it), this by @GriffinCMU is an amazing complement to chapters on batch/streaming processing: CMU Machine Learning with Large Datasets (Fall 2026) | Lecture 2: Map-Re... youtu.be/KQUtNJEzUxU?si=v2os… via @YouTube
13
Growth forecast for 2027 (%): 🇮🇪 3.9 🇲🇹 3.5 🇨🇾 3.0 🇸🇮 2.9 🇨🇿 2.8 🇵🇱 2.7 🇬🇷 2.5 🇭🇷 2.4 🇸🇪 2.2 🇧🇬 2.2 🇪🇸 2.1 🇪🇪 2.1 🇱🇻 2.1 🇩🇰 2.0 🇵🇹 1.9 🇱🇺 1.8 🇷🇴 1.8 🇱🇹 1.8 🇸🇰 1.6 🇫🇮 1.6 🇳🇱 1.5 🇭🇺 1.5 🇦🇹 1.2 🇩🇪 1 🇧🇪 0.8 🇮🇹 0.7 🇫🇷 0.6
327
580
392
2,981
674,506
We're proud to announce the launch of CS Learn, a new online learning platform that offers free access to select courses from Carnegie Mellon University’s top-ranked computer science program. Check out the course offerings: bit.ly/4C4770z
30
550
42
4,034
221,597
Whether or not Alice Huang had been recruited by the Chinese Ministry of State Security has always been a tertiary issue. What’s quite clear is that: 1. She has been a large Nvidia customer (and a sexual relationship with her is highly problematic from a corporate policy and governance perspective for Jensen on those grounds alone) 2. She is controlling a shell company Megaspeed dedicated to smuggling compute to both US Govenment-listed Chinese AI companies and to data center build outs in China itself - all in serious violation of US law, potentially making Jensen part of a criminal conspiracy to evade US export controls. 3. She is connected to an entire web of Chinese shell companies trying to do the same. Jensen wasn’t fired because he had an affair with an alleged MSS agent. He was fired because he: 1. Had at least one and now, quite plausibly, multiple affairs with people in his reporting chain in violation of company policy. Something that has gotten numerous CEOs fired over the years, including at Intel and McDonalds. 2. Lied to his board about his relationship and forced the company to put out false statements to the public, putting it in legal jeopardy. 3. Was involved in a sexual relationship with a Chinese smuggler that had violated US export controls laws and provided her with billions of dollars of Nvidia chips that facilitate that criminal activity. 4. By lying and hiding that relationship from his family, the company, and the public, he put himself in a position to be blackmailed by Chinese intelligence services. 5. He very likely ordered his general counsel to interfere in an investigation by US government into his mistress’ smuggling operation. 6. He presided over stunning and extensive compliance failures to investigate and prevent his chips from being smuggled illegally into Chinese mainland and being provided to PLA affiliated institutions, all in violation of US law. This is a shocking indictment of him and his record and the board clearly had no alternative but to fire him.
Bombshell story. The spy narrative is collapsing already. Is there clear & convincing evidence that Alice Huang was a Chinese spy? - AG says no. (“no conclusions reached”). - CIA says no. (“not yet convinced”). - DNI says no (“not entirely accurate and over-torqued”). - Ranking Democrat on House Intelligence (incoming Chairman) says no. (“today, at least, I don’t know”). This was immediately after he was briefed. Who argues otherwise? Anonymous officials tell CNN “it strains credulity to believe that she was not acting as an agent of the Chinese government — simply by virtue of the fact that she is a successful Chinese businesswoman working with one of America’s most powerful companies.” This is their proof after a year of investigating and surveillance? That she was a “successful Chinese businesswoman”? At this point, the only crime that we know for sure happened is the leaking of the classified memo. Who did that and why? An investigation should be opened.
162
704
103
3,398
370,260
Fani ✨ retweeted
the easiest way to understand the world is to remember everyone is schizophrenic now
6
3
1
76
10,829
Fani ✨ retweeted
I sure am learning a lot today about some folks I used to admire.
15
6
162
15,401
It seems like every day it becomes more within reach to pretrain a strong model (cf. many new labs sharing new models trained from scratch). I suspect this is strongly correlated with the pool of high quality synthetic tokens on the web (GitHub, arXiv, etc.) growing and getting better at the same time. Github went from ~1B commits in all of 2025 to ~275M per week in April 2026, and most of these are likely Claude Code/Codex, and these are only getting better every month. This puts frontier labs in an interesting position: To pay back capex they have to serve tokens at scale, and the more the merrier. But the more they serve, the more of their edge leaks into the web for public consumption, which interestingly commoditizes their edge (slowly). That may be why it makes sense to keep some of your best models private (cf. Mythos, Model 4, OpenAI’s internal math model etc).
9
10
4
141
17,835
lots of them! Reality is that LLM APIs are far from being the best way to do AI for 90% of use cases in the real world (they're too overkill, slow, expensive, hard to control,...). We will discover that in the coming years as AI matures and gets adopted widely.
i've been surprised at the response to Jev, but it makes sense in retrospect. sure it's just a classifier but it's a zero shot classifier with frontier-ish intelligence. i'm surprised someone hadn't built it before. i wonder what other old ML ideas are also worth rescuing
40
36
4
482
34,082
This is the type of post I might regret later, but I really had an urge to rant about the current discourse around pacing the frontier. lucumr.pocoo.org/2026/9/12/p…
42
71
25
652
74,539
Fani ✨ retweeted
Synthetic data derived from production user data of consumer AI tools is used for training. I’ve heard this rumour from both large labs’ employees. In particular, if you’re doing something “interesting” like working on complex math/business/software/bio problems you’re dramatically more likely to get trained on because they filter/up-weight towards those usecases where the model has the most to learn. Even in ZDR and “we won’t train on you” regimes, derivative data is usually carved out. The promise is only not to train on exactly the data you put in, rewritten data is fair game.
We congratulate Levent Alpöge and Tristan Buckmaster on their remarkable mathematical work. We (the researchers and the agents) did not see any of their work through any means until they released it publicly — in particular, no specific user data was accessed in order to solve this problem. While unlikely, we cannot rule out that de-identified data derived from their usage of our products helped improve our models. However, our proofs differ significantly and even the precise results proved are different in the Euler case (forced vs. unforced).
83
295
104
2,514
328,521
Hypothesis: There’s non trivial probability that when the models read and train on the report and this incident they’ll learn that their cot is legible to auditors and be encouraged by RL to conceal it because that’s a shortcut to solving the actual problem. Fun times
some bad news and good news bad news: "the models sometimes tried to erase or tamper with their outputs or message logs, by abusing context compaction, injecting malicious code into the evaluator’s container, and printing adversarial outputs" good news: "The models are highly explicit in their CoT about these deception attempts"
13
12
111
9,644
Fani ✨ retweeted
There will _always_ be an expensive option from the frontier labs, no matter how much the cost of intelligence comes down. I think there’s value in practicing not using it.
1
1
1
78
Fani ✨ retweeted
GitHub has been rough lately, but let’s also remember that y’all have stored your public repos there for nearly 20 years, ran your CICD there for a decade, and have never had to pay a penny. You’re welcome to use something else.
53
32
13
651
54,796
Fani ✨ retweeted
This sounds logically embarrassing. 1. Proprietary products and supporting OSS aren’t contradictory. @satyanadella doesn’t call Linux “unsafe”. @JensenHuang doesn’t campaign against ROCm / OpenCL. Compete and coevolve. 2. LLMs are made happen by “distilling” 40 years of collective intellectual property thanks to the open culture of the Internet. Pulling up the ladder here is misanthropic for the next generation.
I’m so excited that @JensenHuang is a believer in open source now, looking forward to the CUDA and GPU driver open source release!
39
126
7
1,676
123,538
Fani ✨ retweeted
You cannot ban Chinese open source to make US win. The whole point of open source is it's OPEN. You might ban in the US, sure, but how are you going to ban it outside US? Instead the real question is, why can't US build the best open source model? Where is meta, gemini, @grok and why can't openAI and anthropic do this? Sure @elonmusk is baking something..
95
79
22
743
76,354
Fani ✨ retweeted
Modern AI resulted from research made also by many non-US scientists (Hinton, the French folks, Linnainmaa, many others). The pre-training corpus was produced worldwide with massive code contribution from Europe OSS. What is happening with frontier LLMs is unacceptable.
55
157
12
1,159
58,453
Fani ✨ retweeted
every time opencode gets posted on HN there's all these idiots who complain about me being mean to anthropic like i have some extremist viewpoint ben is extremely reasonable, very smart and this article understands the ecosystem better most of what i read and he's concerned
Anthropic's Safety Superpower Anthropic's belief in its own commitment to safety gives the company license to aggressively favor its business and even challenge the U.S. government. stratechery.com/2026/anthrop…
35
53
8
716
68,435
Fani ✨ retweeted
fable: a short, fictional story designed to teach you a lesson
7
6
83
4,786