Laguna M.1 (Poolside): arquitetura que sacrifica conhecimento geral por especialização em código (contexto 262K).• Novo: Otimização nativa para tool calling e latência de ~150ms TTFT (Tempo até ao Primeiro Token).• Comparação: Bate Gemini/Claude no SWE-bench (Resolved)(48.2%) e BFCL. Troca a versatilidade multimodal por execução determinística no terminal.🔗 https://poolside.ai/blog/introducing-laguna-xs2-m1#MachineLearning #LLM #Laguna #SWEbench #CodingAgents #SOTA
Related
🎮 You may not have to wait for Prime Day for the best PC gaming dealsNewegg's Price Protection guarantee and Best Buy's ...
🎮 You may not have to wait for Prime Day for the best PC gaming dealsNewegg's Price Protection guarantee and Best Buy's Price Match mean you could shop today and maybe reap the ben...
Google leaked its next Pixel Drop with Screen Reactions and paid featuresGoogle is advertising some Gemini features you ...
Google leaked its next Pixel Drop with Screen Reactions and paid featuresGoogle is advertising some Gemini features you already paid for as something new.https://www.androidauthori...
RT by @EU_UNGeneva: How can #AI serve all of humanity? In just 3 weeks, the Global Dialogue on AI Governance will bring ...
RT by @EU_UNGeneva: How can #AI serve all of humanity? In just 3 weeks, the Global Dialogue on AI Governance will bring governments, tech leaders & civil society to Geneva. Join us...