Hacker News · AI·12d agoNotes on gotchas while migrating 35kb preprompts from Opus to self-hosted Ollama#anthropic#ollama#openaiAI industry 9 min1
arXiv cs.CR·13d agoSpliTEE: Improving LLM Inference on Trusted Hardware with Differentially Private GPU Outsourcing#differential-privacy#gpu-outsourcing#intel-tdxResearch