Image created with Flux Pro Ultra. Image prompt: A Minecraft screenshot showing different biome types meeting at a center point with various styles of village buildings from different cultures, with “INTERNATIONAL” written in pixelated Minecraft font across the top
“I am so sorry that the U.S. is letting down our friends and allies. Broad tariffs, implemented not just against adversaries but also steadfast allies, will damage the livelihoods of billions of people, create inflation, make the world more fragmented, and leave the U.S. and the” / X https://x.com/AndrewYNg/status/1910388768487727535
Arabic Leaderboards: Introducing Arabic Instruction Following, Updating AraGen, and More https://huggingface.co/blog/leaderboard-3c3h-aragen-ifeval
“Imports rocketed up in Q1 cause of impending tariffs which made GDP look bad Now big inventory destocking in Q2, temprarily lowering imports with hope tariffs go away That means GDP goes up temporarily, which Trump is def gonna champion as tariffs growing the economy” / X https://x.com/dylan522p/status/1908994833999675783
“The tariffs are much more complicated than they seem Industry and the market are misunderstanding the ramifications We cover everything from fabs to packaging to servers to cables to optical modules to datacenter equipment The impacts differ greatly We also found a huge loophole” / X https://x.com/dylan522p/status/1910255795603963923
“Who knew Trump had been secretly an anti-capitalist radical degrowther all this time. Well done Greta, well done. *That’s* 5d chess.” / X https://x.com/francoisfleuret/status/1909839428773646797
“@BillAckman @realDonaldTrump How can you be so clueless to think it’s about other countrie’s tariffs, or that those even matter big picture? The trade deficits are not due to those tariffs. If all of the other countries dropped their tariffs to zero it would have almost no impact, the average tariff rates” / X https://x.com/wightmanr/status/1909007036869943525
“@ja3k_ In the AI-2027 scenario, GDP growth deviates from long running trends. (Though arguably matches even longer running trends going back 10k years.) It doesn’t suffice to just cite trends, you need some understanding of the situation. IMO the AI trend will break the growth trend.” / X https://x.com/RyanPGreenblatt/status/1909075830824968580
“@chamath i realized that if i treat all of your tweets as masteful 200iq bait then they become really funny. great account, im gonna give it a follow actually” / X https://x.com/nearcyan/status/1909757713200103492
“InternVL3 is out 💥 > 7 ckpts with various sizes (1B to 78B) > Built on InternViT encoder and Qwen2.5VL decoder, improves many points on Qwen2.5VL > Can do reasoning, document tasks, extending to tool use and agentic capabilities 🤖 > easily use with transformers 🤗 https://x.com/mervenoyann/status/1910687031505674706
“🤖📚 DeepSeek-R1 Guide Learn to build RAG applications with DeepSeek-R1, an open-source alternative to OpenAI. This guide demonstrates local and cloud implementations using LangChain’s document processing tools. Check out the full guide: https://x.com/LangChainAI/status/1909274972339454227
“Happy to share that weā€™ve just signed a ā‚¬100 million partnership with @cmacgm! @MistralAI is going to help CMA CGM adopt custom-designed AI solutions to support its shipping, logistics, and media activities. We will provide CMA CGM with our entire AI solutions portfolio and we https://x.com/sophiamyang/status/1909243949920768497
“Tencent’s new model ‘Hunyuan-T1’ looks incredible —This is world’s first Mamba-powered ultra-Large Model. 👏 Flexes ultra-large reasoning, Hybrid-Mamba-Transformer MoE, tight logic, & 60–80 tokens/sec speed. @TencentHunyuan 👏 🧵 1/n 📌 Key Highlights → Ultra-large-scale https://x.com/rohanpaul_ai/status/1908861892753322142
“💕 @ollama finally added supports for @MistralAI Small 3.1! Give it a try: `ollama run mistral-small3.1` https://x.com/sophiamyang/status/1909312680424251392
“New Algorithm from China’s ByteDance: VAPO Value-based Augmented Proximal Policy Optimization framework for reasoning models. VAPO refines value-based learning so long reasoning sequences finally become manageable. Tames lengthy outputs through smart advantage calibration for https://x.com/rohanpaul_ai/status/1909581398991946230
“Alibaba just released LAM on Hugging Face Large Avatar Model for One-shot Animatable Gaussian Head https://x.com/_akhaliq/status/1910259092972589432
“University of Hong Kong releases Dream 7B (Diffusion reasoning model). Incredibly performant open-source diffusion model. You can adjust the number of diffusion timesteps for speed vs accuracy. – Consistently outperforms existing diffusion language models by a large margin; – https://x.com/rohanpaul_ai/status/1909080035064946998
“On vibes, I don’t think the currently released Llama 4 models are a big enough increase in capability that they definitively take the lead in open weights models from the main Chinese models (and open weights continues to lag frontier closed). Meta is still training, though.” / X https://x.com/emollick/status/1908921025057653175
“I built a web app using @lovable_dev that leverages AI trained in Nigerian Tenancy Law to: ✅ Inform tenants & landlords about their rights ⚖️ Connect users with real lawyers for dispute resolution 💰 Enable rent payments & property management https://x.com/boysod_/status/1906760339158343764
“HiDream-I1-Dev is the new leading open-weights image generation model, overtaking FLUX1.1 [pro] in the Artificial Analysis Image Arena! HiDream, the Chinese company behind @vivago_ai, has just open-sourced their HiDream-I1 family of models under the MIT License. This impressive https://x.com/ArtificialAnlys/status/1909624716111045115
“American open-source has fallen. It’s all on Google and China bros now 🫠 https://x.com/scaling01/status/1909165768874336620
unsloth/DeepSeek-R1-GGUF at main https://huggingface.co/unsloth/DeepSeek-R1-GGUF/tree/main/DeepSeek-R1-UD-IQ1_S
“Looks interesting! The twitter thread forgot to mention the base model, which is deepseek-qwen — so many thanks to the qwen and deepseek teams too. :D” / X https://x.com/jeremyphoward/status/1909705022935646541
QVQ-Max: Think with Evidence | Qwen https://qwenlm.github.io/blog/qvq-max-preview/
“DeepSeek-R1 is underrated” / X https://x.com/scaling01/status/1909304510075318318
“The absolute largest reasoning coding dataset just dropped in @huggingface A dataset of 736,712 DeepSeek-R1 generated code solutions (in Python) with reasoning traces, Ready for commercial/non-commercial use. https://x.com/rohanpaul_ai/status/1909602083923485168
“Deepseek just announced Inference-Time Scaling for Generalist Reward Modeling on Hugging Face show that SPCT significantly improves the quality and scalability of GRMs, outperforming existing methods and models in various RM benchmarks without severe biases, and could achieve https://x.com/_akhaliq/status/1908167564057849903
“Bytedance releases Seed-Thinking-v1.5, a 200B reasoning model with 20B active parameters that beats DeepSeek R1 across domains 🤯 Read on for a detailed breakdown👇 https://x.com/casper_hansen_/status/1910373327832498242
“NEW: SoTA OCR model – Apache 2.0 licensed based on Qwen 2.5 VL 🔥 https://x.com/reach_vb/status/1908232634943365478
“Announcing DeepSeek V3 Company Researcher 🔦 Get structured company data from the web using Firecrawl’s new /extract endpoint. Open-source, powered by DeepSeek V3 & @firecrawl_dev. https://x.com/ericciarla/status/1910361910136258752
“@ryanbijoy It’s built on top of a good base model: Deepseek-R1-Distilled-Qwen-14B an open-source RL framework: ByteDance’s verl, and now they open source their extension verl-pipeline. exciting time for open source.” / X https://x.com/Yuchenj_UW/status/1910008307202548074
“進化したAIサイエンティスト、「The AI Scientist-v2」を公開 AIが完全自動で科学論文を生成するAIサイエンティストの改訂版、The AI Scientist-v2を公開しました。 論文： https://x.com/SakanaAILabs/status/1909527887482355754
AI continent – European Commission https://commission.europa.eu/topics/eu-competitiveness/ai-continent_en
“We @MistralAI are hiring lots in France, UK, USA, Germany and Singapore 🇫🇷 🇬🇧 🇺🇸 🇩🇪 🇸🇬 We need passionate innovators like you to make it happen! Join us! 🏗️ AI Solutions Architect: Empower our customers to unlock the full potential of AI. You’ll guide them through adopting our https://x.com/sophiamyang/status/1909289524959572460
