// 主题信号

基础设施

训练框架、推理引擎、向量数据库、GPU 调度、可观测性等支撑技术。

GPU-Initiated Communication: Dissecting Down to the Bone — Arxiv
Arxiv

GPU-Initiated Communication: Dissecting Down to the Bone

GPU 发起的通信允许 GPU 线程直接将 RDMA 操作发布到网卡。它是 NVSHMEM、NCCL GIN 和 DeepEP 的基础,这些技术支撑着混合专家(MoE)模型中细粒度、低延迟的关键通信需求,但其性能特征和优化方法至今鲜有深入研究。

未知来源

在 Guides & References / AI 下添加"本地运行 LLM"子章节:本地部署的原因与时机、硬件配置(VRAM、KV 缓存、量化)、Ollama、llama.cpp、vLLM / SGLang、桌面端运行器(LM Studio、Jan、Docker Model Runner、LocalAI、MLX-LM)、NVIDIA NIM、开源权重模型系列与许可、LLM 网关(LiteLLM)、推理优化,以及将本地模型与 GitHub Copilot(BYOK)、Claude Code(LLM 网关 / Anthropic 兼容端点)及其他编程助手集成,包含 MathJax、参考文献与一页速查表 PDF

Add "Running LLMs Locally" sub-section under Guides & References / AI: why and when to run on premise, hardware sizing (VRAM, KV cache, quantization), Ollama, llama.cpp, vLLM / SGLang, desktop runners (LM Studio, Jan, Docker Model Runner, LocalAI, MLX-LM), NVIDIA NIM, open-weight model families and licences, LLM gateways (LiteLLM), inference optimisation, and integrating local models with GitHub Copilot (BYOK), Claude Code (LLM gateway / Anthropic-compatible endpoints) and other coding assistants, with MathJax, bibliography and a one-page cheat sheet PDF

未知来源

在 Guides & References 下添加"AI"章节及其首个子章节"LLM 基础与提示工程":AI 落地页、LLM 工作原理(token、transformer、训练、采样、上下文、推理、多模态、幻觉)、提示工程与上下文工程、结构化输出、模型选择(基准测试、成本/延迟、开源 vs 闭源),包含 MathJax、参考文献与一页速查表 PDF

Add "AI" section under Guides & References and its first sub-section "LLM Foundations & Prompting": AI landing page, how LLMs work (tokens, transformers, training, sampling, context, reasoning, multimodality, hallucination), prompt and context engineering, structured output, choosing models (benchmarks, cost/latency, open vs. closed), with MathJax, bibliography and a one-page cheat sheet PDF