The race to build the Deepseek of Europe is on.
As the relationship between the US and its European allies shows signs of strain, Europe's push to become a self-sufficient AI superpower has become more urgent.
#Europe #DigitalSovereignty #AI #LargeLanguageModels #Deepseek #Business #MagicsTechnologies #SOOFI #OpenSource #Science #Research #Tech
vLLM large scale serving: DeepSeek 2.2k tok/s/h200 with wide-ep
https://blog.vllm.ai/2025/12/17/large-scale-serving.html
#HackerNews #vLLM #large #scale #serving #DeepSeek #tok/s #wide-ep #AI #technology
Reproducing DeepSeek's MHC: When Residual Connections Explode
https://taylorkolasinski.com/notes/mhc-reproduction/
#HackerNews #Reproducing #DeepSeek #MHC #Residual #Connections #Explode #Deep #Learning #AI #Research
Once a pioneer that helped the world lay the foundations of machine learning, Japan’s renewed push into artificial intelligence reflects a search for economic resilience as much as technological relevance. https://www.japantimes.co.jp/commentary/2026/01/06/japan/japans-return-to-ai/?utm_medium=Social&utm_source=mastodon #commentary #japan #artificialintelligence #ai #japaneseeconomy #chatgpt #deepseek #healthcare
DeepSeek uses banned Nvidia chips for AI model, report says
https://finance.yahoo.com/news/china-deepseek-uses-banned-nvidia-131207746.html
#HackerNews #DeepSeek #Nvidia #AI #bannedchips #technology #news
#China's #DeepSeek Releases New #OpenSource #AI Model Amid Google's Gemini 3 Roll Out - https://www.investors.com/news/technology/nvidia-stock-google-stock-deepseek-china-open-source-ai-models/ "Other Chinese companies building open source models include tech giants Baidu (BIDU) and Alibaba (BABA) as well as Zhipu AI, MiniMax and Moonshot AI."
DeepSeek-v3.2: Pushing the frontier of open large language models [pdf]
https://huggingface.co/deepseek-ai/DeepSeek-V3.2/resolve/main/assets/paper.pdf
#HackerNews #DeepSeek #Pushing #Frontier #OpenAI #LanguageModels #PDF
DeepSeek releases open-weights math model with IMO gold medal performance
https://huggingface.co/deepseek-ai/DeepSeek-Math-V2
#HackerNews #DeepSeek #OpenWeights #Math #Model #IMO #Gold #Medal #Performance #AI #Innovation #MachineLearning
#China's #DeepSeek Releases New #OpenSource #AI Model Amid Google's Gemini 3 Roll Out - https://www.investors.com/news/technology/nvidia-stock-google-stock-deepseek-china-open-source-ai-models/ "Other Chinese companies building open source models include tech giants Baidu (BIDU) and Alibaba (BABA) as well as Zhipu AI, MiniMax and Moonshot AI."
DeepSeek-R1 erzeugt unsicheren Code bei politisch sensiblen Begriffen
Die chinesische KI DeepSeek-R1 erzeugt schlechteren Code, wenn Begriffe wie Falun Gong oder Taiwan im Prompt stehen. Das fanden Sicherheitsforscher heraus.
#DeepSeek #IT #KünstlicheIntelligenz #Programmierung #Security #Sicherheitslücken #Zensur #news
China is looking at deploying robot dogs and drone swarms, among other artificial intelligence advancements, for military advantage. https://www.japantimes.co.jp/news/2025/10/28/world/science-health/robot-ai-drone-china-deepseek-war/?utm_medium=Social&utm_source=mastodon #worldnews #sciencehealth #ai #deepseek #defense #robots #china #semiconductors #pla #huawei
Despite all the enthusiasm for AI, the signal from bond markets suggests that investors are unconvinced that the technology will generate sustained or widespread growth. https://www.japantimes.co.jp/commentary/2025/10/27/world/what-the-markets-really-think-about-ai/?utm_medium=Social&utm_source=mastodon #commentary #worldnews #ai #artificialintelligence #openai #google #anthropic #deepseek #markets #globaleconomy
DeepSeek's novel approach renders text as an image to optically compress it, using far fewer vision tokens than text tokens to bypass LLM context limits. This method is highly effective, achieving ~97% precision when retrieving text at a 10x compression ratio. It could enable a "forgetting mechanism" for LLMs, where older chat history is rendered as blurrier, lower-token images, simulating a fading human-like memory for unlimited context.
AI CINESE. POCHI SOLDI, TANTA RESA
#Deepseek AI, il corso cinese dell' #intelligenzaartificiale, è addestrato con una frazione dei costi sostenuti dai giganti occidentali.
In uno studio pubblicato su #Nature, il team di ricerca ha rivelato che ha incentivato le capacità di ragionamento degli #LLM con l'apprendimento per rinforzo (RL) puro, ovvero un sistema di punteggio: alto per le risposte corrette e basso per le sbagliate.
AI CINESE. POCHI SOLDI, TANTA RESA
#Deepseek AI, il corso cinese dell' #intelligenzaartificiale, è addestrato con una frazione dei costi sostenuti dai giganti occidentali.
In uno studio pubblicato su #Nature, il team di ricerca ha rivelato che ha incentivato le capacità di ragionamento degli #LLM con l'apprendimento per rinforzo (RL) puro, ovvero un sistema di punteggio: alto per le risposte corrette e basso per le sbagliate.
While markets clearly reflect the belief that the United States is ahead in the AI arms race, early leadership does not mean victory — especially when it comes to innovation. https://www.japantimes.co.jp/commentary/2025/09/04/world/china-europe-climate-connection/?utm_medium=Social&utm_source=mastodon #commentary #worldnews #china #us #ai #artificialintelligence #tech #openai #chatgpt #deepseek
Heel interessante @tegenlicht met de Chinese econoom Andy Xie over de AI-bubbel.
Het gaat o.a. ook over de Chinese AI Deepseek, die in tegenstelling tot OpenAI wel open source is, hoe en waarom er in de VS zoveel geld rondgepompt wordt en over de enorme ongelijkheid daar.
Artikel en video:
https://tegenlicht.vpro.nl/artikelen/wanneer-barst-de-ai-bubbel
Heel interessante @tegenlicht met de Chinese econoom Andy Xie over de AI-bubbel.
Het gaat o.a. ook over de Chinese AI Deepseek, die in tegenstelling tot OpenAI wel open source is, hoe en waarom er in de VS zoveel geld rondgepompt wordt en over de enorme ongelijkheid daar.
Artikel en video:
https://tegenlicht.vpro.nl/artikelen/wanneer-barst-de-ai-bubbel
Boníssim! 😂 #Deepseek té sentit del humor 😃
