https://jokeuncle.pages.dev/https://jokeuncle.pages.dev/blog/https://jokeuncle.pages.dev/blog/agent-context-eviction-claude-codex-2026-07-17/https://jokeuncle.pages.dev/blog/agent-engineering-seven-questions-2026-07-20/https://jokeuncle.pages.dev/blog/agent-loop-vs-graph-control-flow-2026-07-21/https://jokeuncle.pages.dev/blog/agent-skills-mcp-research-frontier-2026-06-30/https://jokeuncle.pages.dev/blog/agentbench-agent-evaluation-learning-note/https://jokeuncle.pages.dev/blog/agentic-search-vs-rag-code-retrieval-2026-07-17/https://jokeuncle.pages.dev/blog/ai-era-thinking-what-to-offload-2026-07-16/https://jokeuncle.pages.dev/blog/ai-infra-engineer-work-map-openai-note/https://jokeuncle.pages.dev/blog/ai-infra-math-mindmap/https://jokeuncle.pages.dev/blog/ai-infra-mindmap/https://jokeuncle.pages.dev/blog/ai-math-from-arithmetic-to-frontier/https://jokeuncle.pages.dev/blog/ai-page-building-harness-design-2026-07-17/https://jokeuncle.pages.dev/blog/ai-top-principles-telescope/https://jokeuncle.pages.dev/blog/attention-is-all-you-need-math-for-beginners-2026-07-25/https://jokeuncle.pages.dev/blog/big-o-math-for-beginners-2026-07-18/https://jokeuncle.pages.dev/blog/big-o-ten-complexity-patterns-five-structures-2026-07-18/https://jokeuncle.pages.dev/blog/classic-function-atlas-2026-07-22/https://jokeuncle.pages.dev/blog/claude-agent-sdk-architecture/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-06-29/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-06-30-113154/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-06-30/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-02/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-03/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-04/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-06/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-07/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-08/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-09/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-14/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-15/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-16/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-17/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-18/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-20/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-21/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-22/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-24-181800/https://jokeuncle.pages.dev/blog/codex-ai-digest-2026-07-24/https://jokeuncle.pages.dev/blog/control-surface-migration-llm-agent-2026-07-22/https://jokeuncle.pages.dev/blog/control-surface-xray-inside-the-model-2026-07-22/https://jokeuncle.pages.dev/blog/cycle-double-cover-gpt56-fact-check-2026-07-18/https://jokeuncle.pages.dev/blog/deepseek-r1-incentivizing-reasoning-rl/https://jokeuncle.pages.dev/blog/design-to-code-papers-deep-dive-2026-07-17/https://jokeuncle.pages.dev/blog/dsa-15-patterns-three-verbs-2026-07-21/https://jokeuncle.pages.dev/blog/empty-2026-07-10-agent-internalization-externalization/https://jokeuncle.pages.dev/blog/empty-2026-07-10-agent-memory/https://jokeuncle.pages.dev/blog/empty-2026-07-10-agent-tool-calling-mcp-skill/https://jokeuncle.pages.dev/blog/empty-2026-07-10-long-horizon-agent/https://jokeuncle.pages.dev/blog/empty-2026-07-13-graphify-source-deep-dive/https://jokeuncle.pages.dev/blog/empty-2026-07-14-how-tasks-enter-llm-weights/https://jokeuncle.pages.dev/blog/empty-2026-07-14-multitask-convergence-gradient-conflict/https://jokeuncle.pages.dev/blog/empty-2026-07-15-dense-models-moe-mrv2/https://jokeuncle.pages.dev/blog/empty-2026-07-15-vllm-025-release-notes/https://jokeuncle.pages.dev/blog/empty-2026-07-15-vllm-model-ownership-kv-cache/https://jokeuncle.pages.dev/blog/empty-2026-07-15-vllm-top-inference-framework-map/https://jokeuncle.pages.dev/blog/empty-2026-07-15-vllm-v1-architecture-source-walkthrough/https://jokeuncle.pages.dev/blog/empty-2026-07-16-agent-supply-to-d2d-delivery/https://jokeuncle.pages.dev/blog/empty-2026-07-16-context-intervention-mechanisms/https://jokeuncle.pages.dev/blog/empty-2026-07-18-vllm-verl-source-reading-guide/https://jokeuncle.pages.dev/blog/empty-2026-07-20-dianbo-hot-field-achievement/https://jokeuncle.pages.dev/blog/empty-2026-07-20-why-llm-weights-move-to-compute/https://jokeuncle.pages.dev/blog/euler-formula-e-i-theta-rotation-for-beginners-2026-07-18/https://jokeuncle.pages.dev/blog/evoflux-compact-tool-agents-learning-note/https://jokeuncle.pages.dev/blog/grokking-for-beginners-2026-07-25/https://jokeuncle.pages.dev/blog/helm-llm-evaluation-framework-learning-note/https://jokeuncle.pages.dev/blog/ifeval-instruction-following-learning-note/https://jokeuncle.pages.dev/blog/import-ai-2026-07-09-fable-gpu-kernel-ai-rd-automation/https://jokeuncle.pages.dev/blog/in-context-learning-bayesian-inference-2026-07-24/https://jokeuncle.pages.dev/blog/intake-and-roundtrip-two-bleeding-points-2026-07-17/https://jokeuncle.pages.dev/blog/jokeuncle-github-io-2026-07-15-superpowers-structured-context/https://jokeuncle.pages.dev/blog/jokeuncle-github-io-2026-07-16-prompt-context-engineering/https://jokeuncle.pages.dev/blog/kakeya-needle-llm-high-dimensional-thinking-2026-07-25/https://jokeuncle.pages.dev/blog/kimi-k3-kda-delta-attention-deep-dive-2026-07-18/https://jokeuncle.pages.dev/blog/kimi-k3-kda-math-for-beginners-2026-07-18/https://jokeuncle.pages.dev/blog/ksher-home-2026-07-08-codex-kernel-runtime-home/https://jokeuncle.pages.dev/blog/learning-notes-2026-07-09-cuda-tensor-parallel-computations/https://jokeuncle.pages.dev/blog/learning-notes-2026-07-09-flashattention-operator-fusion/https://jokeuncle.pages.dev/blog/learning-notes-2026-07-10-git-worktree-ai-coding/https://jokeuncle.pages.dev/blog/learning-projects-2026-07-08-agent-jail-runtime-boundary/https://jokeuncle.pages.dev/blog/leetcode-2026-06-29-check-if-it-is-possible-to-split-array/https://jokeuncle.pages.dev/blog/linear-algebra-ai-models-vol1-2026-07-25/https://jokeuncle.pages.dev/blog/linear-algebra-ai-models-vol2-2026-07-25/https://jokeuncle.pages.dev/blog/llama-cpp-day5-prefill-decode-call-chain/https://jokeuncle.pages.dev/blog/llm-as-judge-mtbench-chatbot-arena-learning-note/https://jokeuncle.pages.dev/blog/llm-inference-as-os-kv-cache-memory-context-gc-2026-07-18/https://jokeuncle.pages.dev/blog/llm-math-beyond-basics-2026-07-24/https://jokeuncle.pages.dev/blog/lora-as-skill-plugin-learning-note/https://jokeuncle.pages.dev/blog/lora-rank-math-learning-note/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-06-ggml-computation-graph/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-07-masked-softmax-attention/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-07-qkv-projection-math/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-07-rope-rotation-math/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-08-attention-out-projection/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-08-bpe-tokenizer-billing/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-09-decode-bandwidth-limit/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-09-dns-fake-ip-tun-github-pages-debug/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-09-kv-cache-memory/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-10-change-of-space-math-muscle-memory/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-10-mathematical-mindset-path/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-10-perplexity-ppl-explained/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-10-real-gguf-distillation/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-10-why-log-cross-entropy-ppl/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-13-attention-to-pagedattention-llama-cpp-summary/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-13-decode-memory-overheads/https://jokeuncle.pages.dev/blog/lumen-atelier-2026-07-13-sampling-temperature-topk-topp/https://jokeuncle.pages.dev/blog/material-driven-page-building-problem-map-2026-07-17/https://jokeuncle.pages.dev/blog/page-building-protocol-suite-2026-07-17/https://jokeuncle.pages.dev/blog/page/2/https://jokeuncle.pages.dev/blog/page/3/https://jokeuncle.pages.dev/blog/page/4/https://jokeuncle.pages.dev/blog/page/5/https://jokeuncle.pages.dev/blog/page/6/https://jokeuncle.pages.dev/blog/parametric-skills-learning-note/https://jokeuncle.pages.dev/blog/ppo-training-loop-anatomy-2026-07-20/https://jokeuncle.pages.dev/blog/production-lora-peft-run-note/https://jokeuncle.pages.dev/blog/program-as-weights-paw-learning-note/https://jokeuncle.pages.dev/blog/prologmcp-symbolic-reasoning-learning-note/https://jokeuncle.pages.dev/blog/qwen3-learning-map/https://jokeuncle.pages.dev/blog/researcher-deepdive-lilian-weng-harness-engineering/https://jokeuncle.pages.dev/blog/researcher-deepdive-lilian-weng-scaling-laws/https://jokeuncle.pages.dev/blog/scaling-laws-math-for-beginners-2026-07-22/https://jokeuncle.pages.dev/blog/seed-formulas-part1-popular-canon-22-2026-07-18/https://jokeuncle.pages.dev/blog/seed-formulas-part2-zfc-wiedijk-100-2026-07-18/https://jokeuncle.pages.dev/blog/session-model-eval-poc-2026-07-23-llm-session-eval-routing/https://jokeuncle.pages.dev/blog/session-model-eval-poc-2026-07-23-routellm-deep-dive/https://jokeuncle.pages.dev/blog/statistics-for-llms-2026-07-25/https://jokeuncle.pages.dev/blog/svd-for-beginners-2026-07-24/https://jokeuncle.pages.dev/blog/swebench-executable-code-evaluation-learning-note/https://jokeuncle.pages.dev/blog/tag/%E4%BA%8B%E5%AE%9E%E6%A0%B8%E6%9F%A5/https://jokeuncle.pages.dev/blog/tag/%E4%BB%8B%E7%BB%8D/https://jokeuncle.pages.dev/blog/tag/%E4%BB%A3%E7%A0%81%E7%9F%A5%E8%AF%86%E5%9B%BE%E8%B0%B1/https://jokeuncle.pages.dev/blog/tag/%E4%BC%98%E5%8C%96/https://jokeuncle.pages.dev/blog/tag/%E5%8F%AF%E8%A7%86%E5%8C%96/https://jokeuncle.pages.dev/blog/tag/%E5%8F%AF%E8%A7%A3%E9%87%8A%E6%80%A7/https://jokeuncle.pages.dev/blog/tag/%E5%9B%BE%E8%AE%BA/https://jokeuncle.pages.dev/blog/tag/%E5%85%A5%E9%97%A8/https://jokeuncle.pages.dev/blog/tag/%E5%90%8C%E6%B5%8E%E7%BA%BF%E4%BB%A3/https://jokeuncle.pages.dev/blog/tag/%E5%90%8C%E6%B5%8E%E9%AB%98%E6%95%B0/https://jokeuncle.pages.dev/blog/tag/%E5%A4%8D%E6%95%B0/https://jokeuncle.pages.dev/blog/tag/%E5%A4%9A%E4%BB%BB%E5%8A%A1%E5%AD%A6%E4%B9%A0/https://jokeuncle.pages.dev/blog/tag/%E5%A4%9A%E5%85%83%E5%BE%AE%E7%A7%AF%E5%88%86/https://jokeuncle.pages.dev/blog/tag/%E5%A4%A7%E6%A8%A1%E5%9E%8B/https://jokeuncle.pages.dev/blog/tag/%E5%AD%A6%E4%B9%A0%E5%9C%B0%E5%9B%BE/https://jokeuncle.pages.dev/blog/tag/%E5%AD%A6%E4%B9%A0%E6%96%B9%E6%B3%95/https://jokeuncle.pages.dev/blog/tag/%E5%AD%A6%E4%B9%A0%E7%AC%94%E8%AE%B0/https://jokeuncle.pages.dev/blog/tag/%E5%B7%A5%E5%85%B7%E8%B0%83%E7%94%A8/https://jokeuncle.pages.dev/blog/tag/%E5%BC%BA%E5%8C%96%E5%AD%A6%E4%B9%A0/https://jokeuncle.pages.dev/blog/tag/%E6%8C%82%E8%B0%B7%E7%8C%9C%E6%83%B3/https://jokeuncle.pages.dev/blog/tag/%E6%8E%A8%E7%90%86%E5%BC%95%E6%93%8E/https://jokeuncle.pages.dev/blog/tag/%E6%8E%A8%E7%90%86%E6%A8%A1%E5%9E%8B/https://jokeuncle.pages.dev/blog/tag/%E6%9C%AC%E5%9C%B0%E6%8E%A8%E7%90%86/https://jokeuncle.pages.dev/blog/tag/%E6%9C%BA%E5%99%A8%E5%AD%A6%E4%B9%A0/https://jokeuncle.pages.dev/blog/tag/%E6%9E%B6%E6%9E%84%E8%AE%BE%E8%AE%A1/https://jokeuncle.pages.dev/blog/tag/%E6%80%9D%E7%BB%B4%E5%AF%BC%E5%9B%BE/https://jokeuncle.pages.dev/blog/tag/%E6%80%9D%E7%BB%B4%E6%A1%86%E6%9E%B6/https://jokeuncle.pages.dev/blog/tag/%E6%80%A7%E8%83%BD%E4%BC%98%E5%8C%96/https://jokeuncle.pages.dev/blog/tag/%E6%88%90%E6%9C%AC%E4%BC%98%E5%8C%96/https://jokeuncle.pages.dev/blog/tag/%E6%93%8D%E4%BD%9C%E7%B3%BB%E7%BB%9F/https://jokeuncle.pages.dev/blog/tag/%E6%94%B6%E8%97%8F%E5%90%91/https://jokeuncle.pages.dev/blog/tag/%E6%95%B0%E5%AD%A6/https://jokeuncle.pages.dev/blog/tag/%E6%95%B0%E6%8D%AE%E7%BB%93%E6%9E%84/https://jokeuncle.pages.dev/blog/tag/%E6%95%B0%E7%90%86%E7%BB%9F%E8%AE%A1/https://jokeuncle.pages.dev/blog/tag/%E6%95%B0%E7%90%86%E9%80%BB%E8%BE%91/https://jokeuncle.pages.dev/blog/tag/%E6%A6%82%E7%8E%87/https://jokeuncle.pages.dev/blog/tag/%E6%A8%A1%E5%9E%8B%E5%AD%A6%E4%B9%A0/https://jokeuncle.pages.dev/blog/tag/%E6%A8%A1%E5%9E%8B%E8%AE%AD%E7%BB%83/https://jokeuncle.pages.dev/blog/tag/%E6%AC%A2%E8%BF%8E/https://jokeuncle.pages.dev/blog/tag/%E6%AC%A7%E6%8B%89%E5%85%AC%E5%BC%8F/https://jokeuncle.pages.dev/blog/tag/%E6%B3%9B%E5%8C%96/https://jokeuncle.pages.dev/blog/tag/%E6%B3%B0%E5%8B%92%E5%B1%95%E5%BC%80/https://jokeuncle.pages.dev/blog/tag/%E6%B7%B1%E5%BA%A6%E5%AD%A6%E4%B9%A0/https://jokeuncle.pages.dev/blog/tag/%E6%B7%B1%E5%BA%A6%E6%80%9D%E8%80%83/https://jokeuncle.pages.dev/blog/tag/%E6%BA%90%E7%A0%81%E9%98%85%E8%AF%BB/https://jokeuncle.pages.dev/blog/tag/%E7%89%A9%E7%90%86/https://jokeuncle.pages.dev/blog/tag/%E7%A7%8D%E5%AD%90%E5%85%AC%E5%BC%8F/https://jokeuncle.pages.dev/blog/tag/%E7%AC%AC%E4%B8%80%E6%80%A7%E5%8E%9F%E7%90%86/https://jokeuncle.pages.dev/blog/tag/%E7%AE%97%E6%B3%95/https://jokeuncle.pages.dev/blog/tag/%E7%BA%BF%E6%80%A7%E4%BB%A3%E6%95%B0/https://jokeuncle.pages.dev/blog/tag/%E8%81%8C%E4%B8%9A%E5%9C%B0%E5%9B%BE/https://jokeuncle.pages.dev/blog/tag/%E8%AE%AD%E7%BB%83%E5%8A%A8%E5%8A%9B%E5%AD%A6/https://jokeuncle.pages.dev/blog/tag/%E8%AE%BA%E6%96%87%E5%AD%A6%E4%B9%A0/https://jokeuncle.pages.dev/blog/tag/%E8%AE%BA%E6%96%87%E5%AF%BC%E8%AF%BB/https://jokeuncle.pages.dev/blog/tag/%E8%AE%BA%E6%96%87%E8%A7%A3%E8%AF%BB/https://jokeuncle.pages.dev/blog/tag/%E8%B4%9D%E5%8F%B6%E6%96%AF/https://jokeuncle.pages.dev/blog/tag/%E9%9B%B6%E5%9F%BA%E7%A1%80/https://jokeuncle.pages.dev/blog/tag/%E9%9D%A2%E8%AF%95/https://jokeuncle.pages.dev/blog/tag/%E9%95%BF%E4%BB%BB%E5%8A%A1/https://jokeuncle.pages.dev/blog/tag/%E9%AB%98%E7%AD%89%E6%95%B0%E5%AD%A6/https://jokeuncle.pages.dev/blog/tag/%E9%AB%98%E7%BB%B4%E5%87%A0%E4%BD%95/https://jokeuncle.pages.dev/blog/tag/agent/https://jokeuncle.pages.dev/blog/tag/Agent/https://jokeuncle.pages.dev/blog/tag/Agent%20Skills/https://jokeuncle.pages.dev/blog/tag/Agentic%20Search/https://jokeuncle.pages.dev/blog/tag/ai-digest/https://jokeuncle.pages.dev/blog/tag/AI/https://jokeuncle.pages.dev/blog/tag/AI%20%E5%89%8D%E6%B2%BF/https://jokeuncle.pages.dev/blog/tag/AI%20%E6%95%B0%E5%AD%A6/https://jokeuncle.pages.dev/blog/tag/AI%20Coding/https://jokeuncle.pages.dev/blog/tag/AI%20Engineering/https://jokeuncle.pages.dev/blog/tag/AI%20Infra/https://jokeuncle.pages.dev/blog/tag/AI%20R&D/https://jokeuncle.pages.dev/blog/tag/AI%E6%95%B0%E5%AD%A6/https://jokeuncle.pages.dev/blog/tag/aibrief/https://jokeuncle.pages.dev/blog/tag/algorithm/https://jokeuncle.pages.dev/blog/tag/Architecture/https://jokeuncle.pages.dev/blog/tag/array/https://jokeuncle.pages.dev/blog/tag/Attention/https://jokeuncle.pages.dev/blog/tag/Attention%20Is%20All%20You%20Need/https://jokeuncle.pages.dev/blog/tag/Benchmark/https://jokeuncle.pages.dev/blog/tag/Big-O/https://jokeuncle.pages.dev/blog/tag/BPE/https://jokeuncle.pages.dev/blog/tag/Clash/https://jokeuncle.pages.dev/blog/tag/Claude/https://jokeuncle.pages.dev/blog/tag/Claude%20Code/https://jokeuncle.pages.dev/blog/tag/Code%20Agent/https://jokeuncle.pages.dev/blog/tag/Code%20Generation/https://jokeuncle.pages.dev/blog/tag/codex/https://jokeuncle.pages.dev/blog/tag/Codex/https://jokeuncle.pages.dev/blog/tag/Cognition/https://jokeuncle.pages.dev/blog/tag/Component%20Library/https://jokeuncle.pages.dev/blog/tag/Context%20Engineering/https://jokeuncle.pages.dev/blog/tag/Cross%20Entropy/https://jokeuncle.pages.dev/blog/tag/CUDA/https://jokeuncle.pages.dev/blog/tag/D2D/https://jokeuncle.pages.dev/blog/tag/Debugging/https://jokeuncle.pages.dev/blog/tag/DeepSeek/https://jokeuncle.pages.dev/blog/tag/Design%20System/https://jokeuncle.pages.dev/blog/tag/Design2Code/https://jokeuncle.pages.dev/blog/tag/developer-tools/https://jokeuncle.pages.dev/blog/tag/Distillation/https://jokeuncle.pages.dev/blog/tag/DNS/https://jokeuncle.pages.dev/blog/tag/DSA/https://jokeuncle.pages.dev/blog/tag/dynamic-programming/https://jokeuncle.pages.dev/blog/tag/Eval/https://jokeuncle.pages.dev/blog/tag/Evaluation/https://jokeuncle.pages.dev/blog/tag/feishu-card/https://jokeuncle.pages.dev/blog/tag/FlashAttention/https://jokeuncle.pages.dev/blog/tag/Frontend/https://jokeuncle.pages.dev/blog/tag/ggml/https://jokeuncle.pages.dev/blog/tag/GGUF/https://jokeuncle.pages.dev/blog/tag/Git/https://jokeuncle.pages.dev/blog/tag/GitHub%20Pages/https://jokeuncle.pages.dev/blog/tag/GPU/https://jokeuncle.pages.dev/blog/tag/greedy/https://jokeuncle.pages.dev/blog/tag/Grokking/https://jokeuncle.pages.dev/blog/tag/GRPO/https://jokeuncle.pages.dev/blog/tag/Harness/https://jokeuncle.pages.dev/blog/tag/Human-AI%20Collaboration/https://jokeuncle.pages.dev/blog/tag/In-Context%20Learning/https://jokeuncle.pages.dev/blog/tag/Inference/https://jokeuncle.pages.dev/blog/tag/Instruction%20Following/https://jokeuncle.pages.dev/blog/tag/Kimi/https://jokeuncle.pages.dev/blog/tag/KV%20Cache/https://jokeuncle.pages.dev/blog/tag/learning-log/https://jokeuncle.pages.dev/blog/tag/Learning%20Method/https://jokeuncle.pages.dev/blog/tag/leetcode/https://jokeuncle.pages.dev/blog/tag/lilian-weng/https://jokeuncle.pages.dev/blog/tag/llama.cpp/https://jokeuncle.pages.dev/blog/tag/LLM-as-a-Judge/https://jokeuncle.pages.dev/blog/tag/LLM/https://jokeuncle.pages.dev/blog/tag/LLM%20Serving/https://jokeuncle.pages.dev/blog/tag/long-context/https://jokeuncle.pages.dev/blog/tag/LoRA/https://jokeuncle.pages.dev/blog/tag/Low-Code/https://jokeuncle.pages.dev/blog/tag/Math/https://jokeuncle.pages.dev/blog/tag/Mathematical%20Thinking/https://jokeuncle.pages.dev/blog/tag/mcp/https://jokeuncle.pages.dev/blog/tag/MCP/https://jokeuncle.pages.dev/blog/tag/medium/https://jokeuncle.pages.dev/blog/tag/Memory/https://jokeuncle.pages.dev/blog/tag/Model%20Routing/https://jokeuncle.pages.dev/blog/tag/MoE/https://jokeuncle.pages.dev/blog/tag/Multimodal/https://jokeuncle.pages.dev/blog/tag/Network/https://jokeuncle.pages.dev/blog/tag/OpenAI/https://jokeuncle.pages.dev/blog/tag/OpenAI%20API/https://jokeuncle.pages.dev/blog/tag/Orchestration/https://jokeuncle.pages.dev/blog/tag/PagedAttention/https://jokeuncle.pages.dev/blog/tag/Paper%20Reading/https://jokeuncle.pages.dev/blog/tag/Parametric%20Skills/https://jokeuncle.pages.dev/blog/tag/PCA/https://jokeuncle.pages.dev/blog/tag/PEFT/https://jokeuncle.pages.dev/blog/tag/Perplexity/https://jokeuncle.pages.dev/blog/tag/PRD/https://jokeuncle.pages.dev/blog/tag/Pretraining/https://jokeuncle.pages.dev/blog/tag/Productivity/https://jokeuncle.pages.dev/blog/tag/Program-as-Weights/https://jokeuncle.pages.dev/blog/tag/Prolog/https://jokeuncle.pages.dev/blog/tag/prompt-engineering/https://jokeuncle.pages.dev/blog/tag/Prompt%20Engineering/https://jokeuncle.pages.dev/blog/tag/Protocol/https://jokeuncle.pages.dev/blog/tag/Proxy/https://jokeuncle.pages.dev/blog/tag/Quantization/https://jokeuncle.pages.dev/blog/tag/Qwen/https://jokeuncle.pages.dev/blog/tag/RAG/https://jokeuncle.pages.dev/blog/tag/React/https://jokeuncle.pages.dev/blog/tag/Representation/https://jokeuncle.pages.dev/blog/tag/researcher-deepdive/https://jokeuncle.pages.dev/blog/tag/RLHF/https://jokeuncle.pages.dev/blog/tag/RoPE/https://jokeuncle.pages.dev/blog/tag/Round-trip/https://jokeuncle.pages.dev/blog/tag/RouteLLM/https://jokeuncle.pages.dev/blog/tag/Sampling/https://jokeuncle.pages.dev/blog/tag/Scaling%20Laws/https://jokeuncle.pages.dev/blog/tag/Schema/https://jokeuncle.pages.dev/blog/tag/security/https://jokeuncle.pages.dev/blog/tag/Self-Improvement/https://jokeuncle.pages.dev/blog/tag/SGLang/https://jokeuncle.pages.dev/blog/tag/Skill/https://jokeuncle.pages.dev/blog/tag/skills/https://jokeuncle.pages.dev/blog/tag/Skills/https://jokeuncle.pages.dev/blog/tag/Softmax/https://jokeuncle.pages.dev/blog/tag/Speculative%20Decoding/https://jokeuncle.pages.dev/blog/tag/Streaming/https://jokeuncle.pages.dev/blog/tag/Structured%20Outputs/https://jokeuncle.pages.dev/blog/tag/Superpowers/https://jokeuncle.pages.dev/blog/tag/SVD/https://jokeuncle.pages.dev/blog/tag/SWE-bench/https://jokeuncle.pages.dev/blog/tag/Symbolic%20AI/https://jokeuncle.pages.dev/blog/tag/Tensor%20Parallelism/https://jokeuncle.pages.dev/blog/tag/TensorRT-LLM/https://jokeuncle.pages.dev/blog/tag/Thinking/https://jokeuncle.pages.dev/blog/tag/Tokenizer/https://jokeuncle.pages.dev/blog/tag/Tool%20Calling/https://jokeuncle.pages.dev/blog/tag/Tool%20Use/https://jokeuncle.pages.dev/blog/tag/Transformer/https://jokeuncle.pages.dev/blog/tag/Transformers/https://jokeuncle.pages.dev/blog/tag/TypeScript/https://jokeuncle.pages.dev/blog/tag/verl/https://jokeuncle.pages.dev/blog/tag/vLLM/https://jokeuncle.pages.dev/blog/taylor-expansion-and-llm-2026-07-24/https://jokeuncle.pages.dev/blog/todo-app-protocol-implementation-2026-07-17/https://jokeuncle.pages.dev/blog/tongji-calculus-ai-models-vol1-2026-07-25/https://jokeuncle.pages.dev/blog/tongji-calculus-ai-models-vol2-2026-07-25/https://jokeuncle.pages.dev/blog/vllm-2026-07-22-apc-six-mechanics/https://jokeuncle.pages.dev/blog/vllm-2026-07-22-automatic-prefix-caching-source/https://jokeuncle.pages.dev/blog/vllm-2026-07-22-chat-completion-request-parameters/https://jokeuncle.pages.dev/blog/vllm-2026-07-22-chat-streaming-state-machine/https://jokeuncle.pages.dev/blog/vllm-2026-07-22-openai-serving-chat-deep-dive/https://jokeuncle.pages.dev/blog/welcome/https://jokeuncle.pages.dev/blog/why-llms-need-reinforcement-learning-2026-07-20/https://jokeuncle.pages.dev/series/https://jokeuncle.pages.dev/series/agent-engineering/https://jokeuncle.pages.dev/series/ai-math/https://jokeuncle.pages.dev/series/llm-papers/https://jokeuncle.pages.dev/series/tools-automation/