Skip to content

Latest commit

 

History

History
168 lines (147 loc) · 6.62 KB

miaohua_task_submit.md

File metadata and controls

168 lines (147 loc) · 6.62 KB
Raw

融合接口

POST https://sensenova.sensetime.com/v1/fusion/chat-with-image

支持流式 该接口提供聊天和画图功能，通过 POST 方式创建 url 请求，注意请求的 http Header 中需要包含 Authorization 项，其值为你申请得到的sensechat的 API_SECRET_KEY。

请求参数

参数名称 类型 是否必须 默认值 含义
messages array 是 无 对话上下文，数组中的对象为聊天的上下文信息，格式如
[{"role": "user", "content": Say this is a test!"}], role取值user或assistant
temperature float 否 0.8 chat温度采样参数，取值(0,2]。大于1的值倾向于生成更加多样的回复，小于1倾向于生成更加稳定的回复
top_p float 否 0.7 chat核采样参数，取值(0,1]。解码生成token时，在概率和大于等于top_p的最小token集合中进行采样
max_new_tokens int 否 2048 chat的token生成的最大数量
repetition_penalty float 否 1 chat重复惩罚系数，1代表不惩罚，大于1倾向于生成不重复token，小于1倾向于生成重复token
stream bool 否 false chat是否使用流式传输，如果开启，数据将按照data-only server-sent events传输中间结果，并以data: [DONE]结束
user_id string 是 用户ID
output_img bool 否 true 是否输出秒画生成的图片。如果为false，则对画图请求只做意图判断，如果是画图请求，则输出"This is a drawing request*&^**#"，如果不是画图请求，则正常输出对话内容
mh_key string 是 无 miaohua的token
mh_add_prompt bool 否 false miaohua的提示词优化
mh_ddim_steps int 否 50 秒画的ddim_steps
mh_model_name string 是 Artist_V0.1.3 秒画模型名称
mh_output_size string 是 960x960 图片的输出尺寸，如："960x960"
mh_select_seed int 否 -1 随机数种子
mh_scale int 是 7 文本控制力度(1-20)
mh_controlnet_model string 否 "" 若为空则不启用controlnet,可选值为"openpose","canny","depth","fake_scribble","scribble","hed","hough","normal","seg"

请求示例

curl 示例 示例中的API_SECRET_KEY为sensechat的API_SECRET_KEY，MIAOHUA_TOKEN是秒画的token。

curl https://sensenova.sensetime.com/v1/fusion/chat-with-image \
  -H "Content-Type: application/json" \
  -H "Authorization: $API_SECRET_KEY" \
  -d '{
        "messages": [{"role": "user", "content": "Say this is a test!"}],
        "temperature": 0.8,
        "top_p": 0.7,
        "max_new_tokens": 2048,
        "repetition_penalty": 1,
        "stream": false,
        "user_id": "test",
        "mh_key": $MIAOHUA_TOKEN,
        "mh_model_name": "Artist_V0.1.3",
        "mh_scale": 7, 
        "mh_output_size": "640x640",
        "mh_select_seed": -1
  }'

python示例

import requests
api_secret_key = "xxxxxxxxxx"  # your api_secret_key

url = 'https://sensenova.sensetime.com/v1/fusion/chat-with-image'  
data = {
      "messages": [{"role": "user", "content": "Say this is a test!"}],
      "temperature": 0.8,
      "top_p": 0.7,
      "max_new_tokens": 2048,
      "repetition_penalty": 1,
      "stream": True,
      "user_id": "test",
      "mh_key": $MIAOHUA_TOKEN,
      "mh_model_name": "Artist_V0.1.3",
      "mh_scale": 7, 
      "mh_output_size": "640x640",
      "mh_select_seed": -1
}
headers = {
    'Content-Type': 'application/json',
    'Authorization': api_secret_key
}

response = requests.post(url, headers=headers, json=data)

print(response.status_code) 
print(response.text)

非流式返回示例

{
    "code":200,
    "msg":"ok",
    "data":{
        "id":"4b44cd86cd2c000"
        "choices":[
        	{ 
              "content": "this is a test!(or the url of the generated picture)",
              "finish_reason": "stop"
            }
        ],
        "usage": {
            "prompt_tokens": 6,
            "completion_tokens": 6,
            "total_tokens": 12,
            "picture": false, # 是否生成图片
        }
        "status": 0,
    }
}

流式返回示例

{"code": 0, "msg": "OK", "data": {"id": "53c96a73178f000", "choices": [{"finish_reason": "", "delta": " This"}], "usage": {"completion_tokens": 0, "total_tokens": 0, "picture": false}, "status": 0}}
{"code": 0, "msg": "OK", "data": {"id": "53c96a73178f000", "choices": [{"finish_reason": "", "delta": " is"}], "usage": {"completion_tokens": 0, "total_tokens": 0, "picture": false}, "status": 0}}
{"code": 0, "msg": "OK", "data": {"id": "53c96a73178f000", "choices": [{"finish_reason": "", "delta": " a"}], "usage": {"completion_tokens": 0, "total_tokens": 0, "picture": false}, "status": 0}}
{"code": 0, "msg": "OK", "data": {"id": "53c96a73178f000", "choices": [{"finish_reason": "", "delta": " test"}], "usage": {"completion_tokens": 0, "total_tokens": 0, "picture": false}, "status": 0}}
{"code": 0, "msg": "OK", "data": {"id": "53c96a73178f000", "choices": [{"finish_reason": "stop", "delta": ""}], "usage": {"completion_tokens": 5, "total_tokens": 56, "picture": false}, "status": 0}}

触发秒画敏感词时的返回示例

{
  "code": 0,
  "data": {
    "choices": [
      {
        "content": "我们检测到您的提示词中有敏感词语。请勿使用违反社区条约的提示词。",
        "finish_reason": "stop"
      }
    ],
    "id": "53c8dbea5160000",
    "status": 0,
    "usage": {
      "completion_tokens": 40,
      "picture": true,
      "total_tokens": 413
    }
  },
  "msg": "OK"
}

秒画模型指定错误的返回示例

{
  "code": 0,
  "data": {
    "choices": [
      {
        "content": "不支持的模型: Artist_V0.3.0 Beta",
        "finish_reason": "stop"
      }
    ],
    "id": "53c97969538f000",
    "status": 0,
    "usage": {
      "completion_tokens": 38,
      "picture": true,
      "total_tokens": 408
    }
  },
  "msg": "OK"
}