Frédéric Nguyen (@HLFNguyen) profile picture

Intéressé par le marché des céréales, l'agriculture, et plus #FrAgTw de 💛!!

Vendée, Pays de la Loire
Joined June 2011
Photo by Frédéric Nguyen (@HLFNguyen)Photo by Frédéric Nguyen (@HLFNguyen)Photo by Frédéric Nguyen (@HLFNguyen)Photo by Frédéric Nguyen (@HLFNguyen)Photo by Frédéric Nguyen (@HLFNguyen)Photo by Frédéric Nguyen (@HLFNguyen)Photo by Frédéric Nguyen (@HLFNguyen)Photo by Frédéric Nguyen (@HLFNguyen)Photo by Frédéric Nguyen (@HLFNguyen)Photo by Frédéric Nguyen (@HLFNguyen)
Frédéric Nguyen retweeted
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 #ECForecast ➡️link.europa.eu/43hyXw
5
41
22
100
35,413
Frédéric Nguyen retweeted
❗️ [ 🇬🇧 ROYAUME-UNI | 🇺🇸 ÉTATS-UNIS | 🇷🇺 RUSSIE ] 🔸 Un militaire américain de 41 ans a été arrêté au Royaume-Uni dans le cadre d'une enquête antiterroriste liée à la Russie, selon la police britannique. L'homme a été interpellé mardi près de Gloucester, en vertu de la législation britannique sur la sécurité nationale. Les autorités précisent que les faits qui lui sont reprochés présentent un lien avec la Russie, sans en dévoiler davantage.
2
11
219
39,643
Frédéric Nguyen retweeted
🇨🇳 CHINA IMPORTARÍA 115 Mt DE SOJA EN 2026/27 🌱 Su consumo alcanzaría 136 Mt, mientras que la producción local apenas llegaría a 21 Mt. 📊 La brecha consolida su dependencia del mercado internacional. #Soja #China #BCR bcr.com.ar/es/mercados/inves…
2
11
1
83
4,518
Frédéric Nguyen retweeted
#Mondaymotivation une belle perturbation...enfin ! La situation est tellement critique, sur une surface si immense qu'elle en est difficile à appréhender 2026 est-il un ovni statistique ? Ou la préfiguration du climat de 2050 arrivé avant l'heure ? source: EFAS
3
3
10
922
Frédéric Nguyen retweeted
Hey all! Want to know what happend to #wheat markets, and a quick technical analysis outlook? And same for #soy, #oil and #corn! Check my weekly newsletters about #commodities here : macropulseresearch.substack.…
2
118
Frédéric Nguyen retweeted
Hey Twitter!! Here's my #Bitcoin weekly chart!! macropulseresearch.substack.…
1
31
Frédéric Nguyen retweeted
And here's my daily 1 chart, 1 contract 10 days ! This morning we'll take a look at the Nasdaq, and yea, it's quite bold to publish anything today, right? xDD macropulseresearch.substack.…
2
2
42
Frédéric Nguyen retweeted
1
1
986
Frédéric Nguyen retweeted
We @huggingface reproduced this paper!! It's built with ML Intern, an open-source harness to do AI research, train models, and more Model: huggingface.co/ML-Intern-lab… Demo: huggingface.co/spaces/ML-Int…
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math. rikiyat.github.io/blog/weight-sp… github.com/RikiyaT/lm_und… huggingface.co/Qwen/Qwen3.5-0… huggingface.co/RikiyaT/Qwen3.…
65
181
26
1,638
96,811
Frédéric Nguyen retweeted
that GPT-6 Astra weight-editing blog is wild, but read the Gemma part Qwen3.5-0.8B: it found the thinking-loop direction, subtracted it, MATH-500 went 7.9% to 48.1% Gemma 3 270M, which rarely loops: basically no change @grok so is this fixing bugs, not adding capability?
10
2
254
35,884
Frédéric Nguyen retweeted
Russia increases #wheat shipments from 2 mmt to 3 mmt/month. That’s an extra 1 mmt of wheat hitting the global market every month. Media reaction: 2 inquiries. Russia starts loading 3K mt of wheat onto one lonely 30K mt vessel in Russia’s Arctic port of Murmansk. A shipment that’s been discussed for months. Media reaction: 20 inquiries. This is how media works. 😉 #oatt
7
48
7,274
Frédéric Nguyen retweeted
Hello everyone!! This week-end I posted my weekly #grains and #oil newsletter, with analysis of the latest fundamentals and charts! Enjoy!!
1
2
1
72
Frédéric Nguyen retweeted
Yann LeCun was right & nobody in this industry wants to say it out loud. 109 million parameters beat 72 billion. He has spent years saying scale is not the thing that makes a model understand the world. Meta bet the other way. He left in November and started his own lab in Paris. Last Thursday he published the proof. The theory everyone has been running on since 2024 is that if you train models big enough, on text, on images, on sound, they all end up with the same picture of reality inside them. Size does it. Nobody ever proved that part. Randall Balestriero and LeCun proved it. And the proof says size was never doing the work. What does the work is which two things you show a model together. A photo and the caption of that photo are two views of one thing. Models trained on pairs like that get there fast. Two paragraphs pulled from the same document are not two views of one thing. Models trained on those get there slowly and expensively. Next word prediction is the second kind. That is what every chatbot in the world runs on. In their tests the first method reached the same understanding after 21.6 million examples. Next word prediction needed 900 million. And a 109 million parameter text model ended up closer to an image model's view of the world than a 72 billion parameter language model was. 660 times smaller.
267
918
161
6,866
811,473
A 109M outperform a 72B, just what ???
Yann LeCun and Randall Balestriero just Proved the Platonic Representation Hypothesis, and found a better way to achieve it. The Platonic Representation Hypothesis suggests that AI models trained on different modalities, like images and text, eventually learn similar representations of the world. But this paper shows that how models are trained matters more than simply scaling them up. So they mathematically show that, under certain conditions, two JEPAs trained completely separately can learn essentially the same underlying representation of the world. One learns from different images of the same scene, while the other learns from different descriptions of that scene, without ever being trained on image-text pairs. Their approach achieves the same image-text alignment with 42x fewer training samples than next-token prediction. A 109M-parameter text JEPA even outperforms LLMs up to 72B parameters on cross-modal alignment benchmarks. alphaxiv.org/abs/2610.39302
78
Frédéric Nguyen retweeted
CME Bitcoin futures: a failed rebound in a downtrend, 30 October 2026 #BTC #Trading #chart macropulseresearch.substack.…
1
1
54
🧵Three blacklisted supertankers holding 140,000 tons of Russian fuel are drifting in circles in the Atlantic because the White House begged US traders to buy Putin's diesel, and literally zero buyers answered the phone. 1/🧵
361
3,672
470
11,190
504,025
Hormuz and fertiliser relations not clear? Here's a 3 minutes read substack.com/@commodityconve…
52
Frédéric Nguyen retweeted
📉 Las inspecciones de exportación de maíz estadounidense sumaron apenas 630.000 toneladas semanales, frente a 1,24 Mt en igual período del año pasado. El acumulado de la campaña se ubica 800.000 toneladas por debajo del nivel de 2025.
1
438
Frédéric Nguyen retweeted
🇧🇷 En Brasil, la siembra del maíz de primera alcanzó el 47%, en línea con el año pasado. 🇦🇷 En Argentina, llegó al 40%, el ritmo más rápido para esta fecha en al menos seis años.
2
9
622