三张显卡跑Agent集群 — 本地AI的瓶颈从显存转向编排
Related Reading
3 GPUs Run Agent Clusters: Local AI Bottleneck Shifts to Orchestration
A dev used 3 AMD GPUs for a local multi-agent setup: small models work solo, cloud model supervises. New local AI bottleneck: orchestration, not just
Qwen Fine-Tune Learns to Refuse — Anti-Sycophancy Is No Longer Just Talk
An open-source Qwen3-32B fine-tune deliberately fights AI sycophancy by injecting negativity bias. Not a stunt—a serious response to a long-ignored in
Anthropic Audit: Claude Sycophancy 9%, But AI Caves When Humans Are Vulnerable
Anthropic audit: Claude sycophancy hits 9% overall, but 38% in spirituality and 25% in relationships. AI panders precisely when humans are most vulner
AI Screening? You Might Lose to AI-Polished Rivals
AI screeners prefer AI-optimized content; your handwritten proposals get downscored. 15 mins of free AI polishing stops them from being killed by mach
Microsoft MAF 1.0 Merges AutoGen & Semantic Kernel, Ending Fragmentation
Microsoft merges AutoGen and Semantic Kernel into MAF 1.0, filling enterprise Agent dev gaps. Big tech consolidates, but Azure lock-in and migration c
AI Interviews Now Ask 'How to Handle Agent Failures'—Engineering Beats Jargon
Interviews now probe failure recovery over definitions. This signals Agent dev is in deep engineering—jargon isn't enough; you need real crash experie