v1/chat/completions是为了兼容openai提出的协议,如果要使用完整的协议例如理解图片还是用ollama自己的rest api api/chat:
curl http://localhost:11434/api/chat -d '{
"model": "llava",
"messages": [
{ "role": "user", "content": "why is in the images?","images":['base64图片'] }
]
}' 网友回复


