Dual Acies
314 posts

Dual Acies
@DualAcies
Prop trader. US equities intraday, structural theses long-term. 🇪🇪






@teortaxesTex They probably have a Mythos tier LLM for defense purposes. Their "intelligentised warfare" doctrine in the new FYP is heavy on AI and unmanned systems integration. Chinese Academy of Science definitely has the brains and resources to create something like Mythos.




Er is een akkoord bereikt met ENGIE om de voorwaarden te bepalen en de nodige studies op te starten voor een volledige overname van het Belgische nucleaire park. In afwachting daarvan worden alle ontmantelingsactiviteiten per direct stopgezet. Deze regering kiest voor zekere, betaalbare en duurzame energie. Met minder afhankelijkheid van fossiele import en meer controle over onze eigen voorziening. ——— Un accord a été conclu avec ENGIE afin de définir les conditions et de lancer les études nécessaires en vue d’une reprise complète du parc nucléaire belge. D'ici là, toutes les activités de démantèlement sont suspendues avec effet immédiat. Ce gouvernement fait le choix d'une énergie sûre, abordable et durable, avec moins de dépendance aux importations fossiles et plus de contrôle sur notre propre approvisionnement.


Q1 earnings are in: 2026 is off to a terrific start. Our AI investments and full stack approach are lighting up every part of the business: Search queries are at an all-time high with AI continuing to drive usage. Google Cloud revenue grew 63%, Gemini models have incredible momentum, and it was our strongest quarter ever for consumer AI subs, driven by @GeminiApp. Thanks to our partners + employees around the world. Much more to share on our earnings call in 20 minutes… and at Google I/O in 20 days!


Can LLMs simply tell us about unwanted behaviors they’ve picked up in training? We train a single Introspection Adapter (IA) that makes fine-tuned models describe their behaviors. It generalizes to detecting hidden misalignment, backdoors and safeguard removal.










DeepSeek V4: Concerns Outweigh Hype Key takeaways from in-depth tech analysis from Zhihu contributor 小盒子👇 ⏳Early Computing Moat • Reserved large-scale GPU clusters and advanced chips years before ChatGPT’s launch. • Built an irreplaceable hardware foundation ahead of global peers. ⚙️Unique Organizational Background • Rooted in a quantitative fund with steady cash flow for long-cycle AI research. • Free from big-tech KPIs, quarterly targets and short-term capital pressure. 💡Haven for Original Innovation • Flat management and flexible resources, no rigid approval for high-risk projects. • Open-source contributions prioritize industry progress over instant profits. 📉Hidden Risks • Core researchers poached by tech giants with premium salaries. • Upcoming financing brings valuation and profit pressure. • Open-source model weakens independent monetization. ⚠️Critical Crossroad • Faces the same dilemma as OpenAI: idealism vs. commercialization. • Its true value lies in tolerating uneconomical, frontier research (not just model parameters). 📜Core Motto (Self-Restraint) • Unswayed by praise, undaunted by criticism, uphold principles and stay steadfast. 🧭 Final Outlook • V4 is not a milestone, but a stress test for its innovative DNA. • The key: retain originality amid capital expansion and competition. 🔗A very worthwhile read — highly recommended. Read full analysis: zhuanlan.zhihu.com/p/203105441153… #DeepSeekV4 #GenAI #Tech









