# Ollama llma2 模型管理工具,在本地启动并运行大型语言模型。 ## Usage https://github.com/jmorganca/ollama 地址已迁移到 https://github.com/ollama/ollama ``` # 安装 ollama wget https://github.com/jmorganca/ollama/releases/download/v0.1.11/ollama-linux-amd64 chmod +x ollama-linux-amd64 # 或者 curl -fsSL https://ollama.com/install.sh | sh ollama pull llava # ollama push llava # 发布一个镜像 ollama list ollama cp ollama show ollama create # 通过 Modelfile 创建模型 ollama rm ollama run mistral ollama run codellama ollama run llama2 ollama run llama2-uncensored ollama run llama2:13b ollama run llama2:70b ollama run deepseek-r1:7b ollama run deepseek-r7b ollama run deepseek-r8b ollama run deepseek-r14b ollama run deepseek-r32b ollama run orca-mini ollama run vicuna ./ollama serve # 或 ollama serve,端口 8434 ``` 相关GUI工具: chatbox ``` chatbot-ollama export DEFAULT_MODEL="deepseek-r1:1.5b" && export OLLAMA_HOST="http://localhost:8434" && cd /opt/chatbot-ollama && npm run dev # web 应用的启动命令 ``` open-webui ``` export ENABLE_OPENAI_API=false && export OLLAMA_API_BASE_URL="http://localhost:8434" && open-webui serve # open-webui 的启动命令 ``` anythingllm-server ``` cd /opt/anythingllm-template && yarn setup && export OLLAMA_HOST="http://localhost:8434" && env LLM_PROVIDER='ollama' OLLAMA_BASE_PATH='http://127.0.0.1:8434' OLLAMA_MODEL_PREF='deepseek-r1:1.5b' OLLAMA_MODEL_TOKEN_LIMIT=4096 EMBEDDING_ENGINE='ollama' EMBEDDING_BASE_PATH='http://127.0.0.1:8434' EMBEDDING_MODEL_PREF='bge-m3:latest' yarn dev:server # anythingllm server 的启动命令 ``` ## Develop 安装基础依赖 git build-essential cmake go1.21 ``` go build ``` 或在容器中构建, Dockerfile.build