PKM vs RAG vs Wiki vs Memory Systems Explained Clearly
Compare PKM, RAG, wikis, and AI memory systems by structure, retrieval, ownership, evolution, and real-world use cases. PKM vs RAG vs Wiki vs Memory Systems Explained Clearly
Posts
OpenClaw: Examining a Self-Hosted AI Assistant as a Real System
A case-study exploration of OpenClaw — a self-hosted AI assistant system that integrates local LLMs, retrieval, memory, routing, and observability into a cohesive local infrastructure. OpenClaw: Examining a Self-Hosted AI Assistant as a Real System
LLM Performance in 2026: Benchmarks, Bottlenecks & Optimization
LLM performance is not just about having a powerful GPU. Inference speed, latency, and cost efficiency depend on constraints across the entire stack: Model size and quantization VRAM capacity and memory bandwidth Context length and prompt size Runtime scheduling and batching CPU core utilization System topology (PCIe lanes, NUMA, etc.) This hub organizes deep dives into how large language models behave under real workloads — and how to optimize them. What LLM Performance Really Means Performance is multi-dimensional. Throughput vs Latency Throughput = tokens per second across many requests Latency = time to first token + total response time Most real systems must balance both. The Constraint Order In practice, bottlenecks usually appear in this order: VRAM capacity Memory bandwidth Runtime scheduling Context window size CPU overhead Understanding which constraint you’re hitting is more important than “upgrading hardware”.
Curated List of Articles about Coding in Python:
Architecture and Design Patterns, Modern Package Management, Building Production-Ready APIs, AI, RAG and LLM Integration, Data Science and Analysis, Document Processing and Web Scraping, Testing, DevOps Deployment: Curated List of Articles about Coding in Python #python, #coding #dev #devops #datascience #ai #rag #llm #architecture #web #testing
Ollama GPU Acceleration: Ultimate CUDA & ROCm Guide:
Learn how to configure and optimize Ollama for production AI deployment using NVIDIA CUDA and AMD ROCm. This guide covers GPU acceleration setup, performance tuning, and best practices for scalable inference.: https://dasroot.net/posts/2026/01/ollama-gpu-acceleration-nvidia-cuda-amd-rocm-guide/ #Ollama #NVIDIA #CUDA #AMDROCm #GPU #AI
Open WebUI: Self-Hosted LLM Interface.
Complete guide to Open WebUI: a powerful self-hosted web interface for Ollama and OpenAI-compatible APIs with RAG, multi-user auth, and Docker deployment. https://www.glukhov.org/post/2026/01/open-webui-overview-quickstart-and-alternatives/ #AI #LLM #Ollama #Docker #SelfHosting #OpenSource #Python #K8S
Vibe Coding - meaning, description, origins, challenges and efficiency and risk aspects
Vibe Coding - meaning, description, origins, challenges and efficiency and risk aspects https://www.glukhov.org/post/2025/04/vibe-coding/ Vibe coding is an AI-driven programming approach where developers describe desired functionality in natural language, allowing AI tools to generate code automatically. vibe coding with microphone Coined by OpenAI co-founder Andrej Karpathy in February 2025, it emphasizes a conversational workflow where programmers focus on guiding AI outputs rather than manually writing code. AI-driven code generation Key aspects: Natural language prompts: Users articulate requirements in plain English (or other languages), and AI coding assistants like GitHub Copilot or Replit translate these into functional code. Iterative refinement: Adopts a “code first, refine later” mindset, prioritizing rapid prototyping over immediate optimization. Limited code oversight: Practitioners often accept AI-generated code without exhaustive review, relying on testing and i...
How to rerank documents with Embedding models
How to rerank documents with Embedding models & similarity calculation in RAG: https://www.glukhov.org/post/2024/09/reranking-with-embedding-models #LLM #AI #Ollama #RAG #Embedding
Move Ollama Models to different location
After installing ollama better to reconfigure ollama to store them in new place right away. So after we pull a new model, it doesn’t get downloaded to the old location. Ollama is a text-based frontend to LLM AI Models and an API that can host those too. Install Ollama Goto https://ollama.com/download To install Ollama on linux: curl -fsSL https://ollama.com/install.sh | sh Ollama on Windows is on page: https://ollama.com/download/windows Ollama for Mac is there too: https://ollama.com/download/macOllamaSetup.exe Download, List and Remove Ollama models To download some Ollama models: Go to Ollama Library (https://ollama.com/library) and find the model you need, there you can also find model tags and sizes. Then run: ollama pull gemma2:latest # Or get slightly smarter one still nicely fitting into 16GB VRAM: ollama pull gemma2:27b-instruct-q3_K_S # Or: ollama pull llama3.1:latest ollama pull llama3.1:8b-instruct-q8_0 ollama pull mistral-nemo:12b-instruct-2407-q6_K ollama ...