Ollama 本地 LLM 流式聊天界面｜AI 生成单文件 HTML 示例

Create a complete single-file HTML application that connects to a locally running Ollama API (default http://localhost:11434/api/generate) using JavaScript fetch with streaming (ReadableStream). The UI must include: a dark theme with a centered chat container, a header showing 'Ollama Chat', a scrollable message area with alternating user/assistant bubbles, a text input and send button at the bottom. For the input, the user enters a prompt and clicks Send; the app sends a POST request to /api/generate with model 'llama3.2:1b' and stream:true, then reads the stream chunk by chunk, building the assistant response and rendering it incrementally in the chat. Include a 'Clear' button to reset messages. Use CSS gradients and subtle animations for message appearance. Use vanilla JS only, no frameworks. Keep the design responsive and modern.