DEV Community: Cosmonet The latest articles on DEV Community by Cosmonet (@cosmonet). https://dev.to/cosmonet https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3848474%2Fde87ee31-7450-4cb1-917c-c2193d9a33e8.webp DEV Community: Cosmonet https://dev.to/cosmonet en Bluestar Linux 2026: Guida Tecnica alla Distro Arch per IA Cosmonet Mon, 30 Mar 2026 09:53:42 +0000 https://dev.to/cosmonet/bluestar-linux-2026-guida-tecnica-alla-distro-arch-per-ia-11aa https://dev.to/cosmonet/bluestar-linux-2026-guida-tecnica-alla-distro-arch-per-ia-11aa <p>Nuovo articolo su CosmoNet.info!</p> <p>➡️ <strong>Leggi qui:</strong><br> <a href="https://www.cosmonet.info/bluestar-linux-2026-arch-ia-sviluppo/" rel="noopener noreferrer">Bluestar Linux 2026: Guida Tecnica alla Distro Arch per IA</a></p> <h1> ai #linux #opensource </h1> NVIDIA's Nemotron-Cascade 2: A 30B MoE Model That Hits Gold Medal in Math and Coding Olympiads Cosmonet Sun, 29 Mar 2026 01:26:43 +0000 https://dev.to/cosmonet/nvidias-nemotron-cascade-2-a-30b-moe-model-that-hits-gold-medal-in-math-and-coding-olympiads-4cfd https://dev.to/cosmonet/nvidias-nemotron-cascade-2-a-30b-moe-model-that-hits-gold-medal-in-math-and-coding-olympiads-4cfd <p>🧠 NVIDIA ha rilasciato Nemotron-Cascade 2 e i numeri sono impressionanti!</p> <p>30B parametri totali, ma solo 3B attivi per token. Eppure riesce a ottenere la medaglia d'oro su IMO 2025, IOI 2025 e ICPC World Finals. Con 20× meno parametri dei modelli frontier.</p> <p>Cosa lo rende speciale:<br> ✅ Architettura MoE ultra-efficiente<br> ✅ Pipeline Cascade RL + MOPD (innovazione chiave)<br> ✅ Thinking mode nativo per ragionamento step-by-step<br> ✅ Context window da 256K token<br> ✅ Eseguibile in locale via Ollama (24 GB VRAM)<br> ✅ Open-weight su Hugging Face</p> <p>📖 Guida completa su <a href="https://www.cosmonet.info/nemotron-cascade-2-nvidia-modello-moe-open-weight/" rel="noopener noreferrer">Cosmonet.info</a> </p> <h1> NVIDIA #NemotronCascade2 #MoE #OpenWeight #LLM #AILocale #Ollama </h1> <p><a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fig4xmjbknfpqiuh0xa8h.webp" class="article-body-image-wrapper"><img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fig4xmjbknfpqiuh0xa8h.webp" alt=" " width="600" height="343"></a>#CodingAI</p> machinelearning opensource nvidia llm