教程

Run DeepSeek locally the right way: R1 distills vs the full model vs V4, which Ollama tag fits 8GB to 48GB of VRAM, and four commands to get started.

LLM quantization explained: what Q4_K_M means, GGUF quant sizes and speeds side by side, and how to pick Q4, Q5 or Q8 for your VRAM, with Ollama commands.

A complete LM Studio tutorial: install, search and download models, pick a quantization, tune context, chat with documents and start the local API.

Want to run an LLM locally? Learn how much VRAM you need, which models fit 8GB to 48GB, and how to get your first model running with Ollama in five steps.

Install Ollama on Windows, macOS or Linux, verify it works, run your first model, move the model folder and fix common problems. Updated October 2026.

Ollama not using GPU? Confirm with ollama ps, then check NVIDIA drivers, AMD ROCm, Vulkan, Mac and Docker, with the env vars and log paths to use.

Open WebUI Ollama setup with Docker or pip: create the admin account, connect to Ollama, download models and fix ‘can’t reach Ollama’ errors.

How to run Qwen locally: Ollama file sizes for Qwen3, 3.5, 3.6, 3.8, Coder and VL, a VRAM matching table, one-line commands and thinking-mode control.

What is a local LLM? See how local, offline and on-device models differ, how they compare with cloud AI on privacy, cost and capability, and who it suits.

想在自己电脑上本地部署大模型？按显存和内存讲清需要什么配置、8GB 到 48GB 能跑多大的模型，附 Ollama 上手五步和常见误区。