Tag: LLM 2026 KDA - 从线性注意力到 Kimi Delta Attention Aug 01 2026 GLM-5: From Vibe Coding to Agentic Engineering Jun 30 2026 DeepSeek-V4: 1MM Token Context Jun 19 2026 DeepSeek-V3.2 Lightning Indexer Feb 18 2026 2025 MHA和MLA的KV cache比较 Aug 01 2025 DeepSeek Expert Parallelism Load Balancer (EPLB) Code Reading Apr 20 2025 DeepSeek V3 MoE Mar 20 2025 DeepSeek V3 Multi-head Latent Attention (MLA) Mar 20 2025 DeepSeek V3 learning notes Feb 23 2025 DeepSeek V3 Feb 16 2025 2024 Prediction in decoder and KV-Cache Apr 21 2024 2023 Image Generation 2: Latent Diffusion model / Stable Diffusion Oct 01 2023 Image Generation 1: Diffusion model Jul 04 2023 GPT-1, GPT-2, GPT-3, InstructGPT / ChatGPT and GPT-4 summary May 28 2023