https://guijiagi.com/tags/moe/Recent content in MoE on 硅基 AGI · 智能体学习与测评Hugozh-cn本站内容采用 CC BY-NC-SA 4.0 国际许可协议授权Thu, 16 Jul 2026 11:19:00 +0800https://guijiagi.com/posts/b1-cabd68a1/Thu, 16 Jul 2026 11:19:00 +0800https://guijiagi.com/posts/b1-cabd68a1/深入解析Mixture of Experts架构的技术原理、训练方法和推理优化策略https://guijiagi.com/posts/b2-4eded1bd/Thu, 16 Jul 2026 10:15:00 +0800https://guijiagi.com/posts/b2-4eded1bd/从数学原理到工程实现，全面剖析MoE模型的路由机制、训练挑战与推理优化策略https://guijiagi.com/posts/article-89/Mon, 13 Jul 2026 07:40:00 +0800https://guijiagi.com/posts/article-89/回顾从Transformer诞生到2026年的架构演进，梳理每一次关键突破的技术脉络https://guijiagi.com/posts/article-33/Sun, 12 Jul 2026 22:20:00 +0800https://guijiagi.com/posts/article-33/从路由机制到负载均衡，全面拆解Mixture of Experts架构的设计原理https://guijiagi.com/posts/b2-0e8204a0/Sun, 12 Jul 2026 22:20:00 +0800https://guijiagi.com/posts/b2-0e8204a0/从路由机制到负载均衡，全面拆解Mixture of Experts架构的设计原理https://guijiagi.com/posts/moe-architecture-2026-detail/Thu, 02 Jul 2026 10:52:00 +0800https://guijiagi.com/posts/moe-architecture-2026-detail/从GShard到DeepSeek-V3，全面解析MoE架构的设计演进与工程实践https://guijiagi.com/posts/deepseek-v4-release/Thu, 02 Jul 2026 10:21:00 +0800https://guijiagi.com/posts/deepseek-v4-release/DeepSeek V4发布，以极低成本实现接近GPT-6的性能，再次震撼AI圈https://guijiagi.com/posts/meta-llama4-release/Thu, 02 Jul 2026 10:04:00 +0800https://guijiagi.com/posts/meta-llama4-release/Meta发布Llama 4开源大模型，405B参数MoE架构刷新开源模型性能记录https://guijiagi.com/posts/moe-architecture-comparison/Tue, 30 Jun 2026 11:30:00 +0800https://guijiagi.com/posts/moe-architecture-comparison/深度对比三大MoE架构大模型的技术细节和性能表现https://guijiagi.com/posts/llama-4-series-evaluation/Tue, 30 Jun 2026 10:40:00 +0800https://guijiagi.com/posts/llama-4-series-evaluation/全面评测Llama 4系列模型，从405B到8B的开源旗舰表现分析https://guijiagi.com/posts/deepseek-v4-full-evaluation/Tue, 30 Jun 2026 10:30:00 +0800https://guijiagi.com/posts/deepseek-v4-full-evaluation/全面评测DeepSeek V4在推理、代码、中文理解等维度的表现，分析国产大模型的突破https://guijiagi.com/posts/moe-expert-routing-load-balance/Tue, 30 Jun 2026 09:20:00 +0800https://guijiagi.com/posts/moe-expert-routing-load-balance/深入解析MoE混合专家模型的路由机制、负载均衡策略以及在2026年的最新进展https://guijiagi.com/posts/moe-mixture-of-experts-routing-load-balancing/Tue, 30 Jun 2026 09:20:00 +0800https://guijiagi.com/posts/moe-mixture-of-experts-routing-load-balancing/深入解析MoE混合专家模型的路由机制、负载均衡策略和训练技巧，包含数学推导和工程实现https://guijiagi.com/posts/transformer-architecture-2026-attention-to-moe-evolution/Tue, 30 Jun 2026 09:20:00 +0800https://guijiagi.com/posts/transformer-architecture-2026-attention-to-moe-evolution/深入解析Transformer架构在2026年的最新演进，涵盖注意力机制优化、MoE混合专家、线性注意力等前沿方向https://guijiagi.com/posts/transformer-architecture-2026/Tue, 30 Jun 2026 09:20:00 +0800https://guijiagi.com/posts/transformer-architecture-2026/深入解析Transformer架构在2026年的最新演进，涵盖注意力机制优化、混合专家模型、线性注意力等前沿方向https://guijiagi.com/posts/moe-architecture-evolution/Sun, 28 Jun 2026 11:06:00 +0800https://guijiagi.com/posts/moe-architecture-evolution/系统梳理MoE混合专家架构从Mixtral到DeepSeek V4的演进历程，解析路由机制、负载均衡与工程优化https://guijiagi.com/posts/transformer-architecture-2026-evolution/Sun, 28 Jun 2026 11:00:00 +0800https://guijiagi.com/posts/transformer-architecture-2026-evolution/深入解析Transformer架构在2026年的最新演进，涵盖Attention机制优化、MoE混合专家架构、Mamba状态空间模型三大方向https://guijiagi.com/posts/moe-architecture-deepseek-v4-vs-qwen35-vs-llama4-behemoth/Sun, 28 Jun 2026 10:00:00 +0800https://guijiagi.com/posts/moe-architecture-deepseek-v4-vs-qwen35-vs-llama4-behemoth/三大MoE架构大模型的技术深度对比：路由机制、专家设计、推理效率全方位解析https://guijiagi.com/posts/moe-architecture-analysis/Sat, 27 Jun 2026 15:00:00 +0800https://guijiagi.com/posts/moe-architecture-analysis/混合专家模型MoE架构剖析https://guijiagi.com/posts/moe-architecture-deep/Thu, 25 Jun 2026 12:00:00 +0800https://guijiagi.com/posts/moe-architecture-deep/深入剖析 Mixture of Experts 架构的原理、路由机制、训练策略及工程实现，涵盖 GShard、Switch Transformer、Mixtral 等代表性工作。https://guijiagi.com/posts/mixture-of-experts-internals/Thu, 25 Jun 2026 10:00:00 +0800https://guijiagi.com/posts/mixture-of-experts-internals/深入解析 Mixture-of-Experts 架构的路由机制、负载均衡策略与工程优化https://guijiagi.com/posts/mixture-of-experts-deep-dive/Thu, 25 Jun 2026 10:00:00 +0800https://guijiagi.com/posts/mixture-of-experts-deep-dive/全面解析 MoE 架构原理、路由算法、负载均衡及演进历程https://guijiagi.com/posts/deepseek-technical-analysis/Wed, 24 Jun 2026 11:40:00 +0800https://guijiagi.com/posts/deepseek-technical-analysis/DeepSeek 系列模型的技术架构、训练方法和工程优化深度分析https://guijiagi.com/posts/deepseek-v4-%E5%AE%8C%E6%95%B4%E7%89%88%E5%8F%91%E5%B8%83-%E5%BC%80%E6%BA%90%E6%A8%A1%E5%9E%8B%E7%9A%84%E6%96%B0%E5%B7%85%E5%B3%B0/Wed, 24 Jun 2026 00:00:00 +0800https://guijiagi.com/posts/deepseek-v4-%E5%AE%8C%E6%95%B4%E7%89%88%E5%8F%91%E5%B8%83-%E5%BC%80%E6%BA%90%E6%A8%A1%E5%9E%8B%E7%9A%84%E6%96%B0%E5%B7%85%E5%B3%B0/DeepSeek V4 完整版正式发布，MoE 架构+1M 上下文+多头潜在注意力，深度解析技术细节与实测表现https://guijiagi.com/posts/mixture-of-experts-guide/Tue, 23 Jun 2026 16:10:00 +0800https://guijiagi.com/posts/mixture-of-experts-guide/MoE架构原理、主流模型对比和生产环境选型建议