Tag: foundations

> **核心摘要**：学习 System Design 的最高境界是研读业界顶级科技巨头的真实架构。本指南全量解构两个经典工业案例——**Pinterest**（视觉搜索与推荐：图像嵌入、PinSage 式图神经网络表征、HNSW 近似最近邻检索、混合检索、多模态表征）与 **Netflix**（流媒体推荐：显式 +

> **Core Executive Summary**: Single GPU VRAM cannot host 100B+ LLM training parameters, gradients, and optimizer states (a 70B FP16 model requires 1.12TB train

> **Summary**: Computer vision architectures evolved from handcrafted local inductive biases (CNNs) to data-driven global self-attention (ViT). This 100% exhaus

> **Core Executive Summary**: LLMs often generate plausible-sounding but unfactual or logically contradictory text, known as **Hallucination**. Hallucinations r

> **Core Executive Summary**: Since its introduction in 2017, the Transformer architecture has fundamentally reshaped artificial intelligence, serving as the un

> **Core Executive Summary**: Linear algebra is the substrate of machine learning and deep learning: every tensor is a matrix, every layer is a matrix multiplic

> **Core Executive Summary**: Vision-Language Models (VLMs) empower LLMs to perceive visual scenes. Rather than training multimodal models from scratch, VLMs ut

> **Core Executive Summary**: Moving RAG knowledge bases and agentic systems from demo to enterprise production is a distributed-systems problem as much as an M

> **核心摘要**：单张 GPU 的显存（如 H100 80GB）远不足以装载千亿参数大模型的权重、梯度和优化器状态（70B 模型 FP16 训练至少需要 1.12TB 显存）。**分布式 4D 并行体系 (DP, TP, PP, EP)** 与 **DeepSpeed ZeRO** 实现了跨千卡 GPU 集群的高效

> **核心摘要**：从传统的 2D 卷积神经网络 (CNN) 到打破模态壁垒的 Vision Transformer (ViT)，计算机视觉架构经历了从“人工设计局部归纳偏置”到“数据驱动全局自注意力”的伟大范式转移。本指南系统剖析系统剖析 2D 卷积维度与感受野 (Receptive Field, RF) 通用递推