zzhangpurdue
commented
5 months ago 已经定位到问题，在修复，参考pr
Closed jacnmm4 closed 5 months ago
zzhangpurdue
commented
5 months ago 已经定位到问题，在修复，参考pr
jacnmm4
commented
5 months ago 已经定位到问题，在修复，参考pr
谢谢，参考pr是什么
zzhangpurdue
commented
5 months ago pr：https://github.com/modelscope/modelscope-agent/pull/451 已经合并到最新的master了，可以用master，或者等今明天发0.5.3版本。
zzhangpurdue
commented
5 months ago 0.5.3 版本已验证修复
Initial Checks
What happened + What you expected to happen
根据官方使用openapi的示例：https://github.com/modelscope/modelscope-agent/blob/master/examples/tools/openapi_schema_tool.ipynb，本地运行报错，相关的dashscopen已改为自己的，报错如下： 2024-05-21 12:10:04,212 - modelscope - INFO - PyTorch version 2.3.0 Found. 2024-05-21 12:10:04,217 - modelscope - INFO - TensorFlow version 2.16.1 Found. 2024-05-21 12:10:04,217 - modelscope - INFO - Loading ast index from C:\Users\Yang.cache\modelscope\ast_indexer 2024-05-21 12:10:04,363 - modelscope - INFO - Loading done! Current index file version is 1.12.0, with md5 bae340a52ccfa76c45f06e341a861891 and a total number of 964 components indexed import wanx-v1_text2image failed, details: 'wanx-v1_text2image' Traceback (most recent call last): File "D:\test2023\modelscope-agent310\openapi_test.py", line 209, in
bot = RolePlay(function_list=function_list,llm=llm_config, instruction=role_template)
File "D:\test2023\modelscope-agent310\venv\lib\site-packages\modelscope_agent\agents\role_play.py", line 144, in init
Agent.init(self, function_list, llm, storage_path, name,
File "D:\test2023\modelscope-agent310\venv\lib\site-packages\modelscope_agent\agent.py", line 53, in init
self._register_tool(function)
File "D:\test2023\modelscope-agent310\venv\lib\site-packages\modelscope_agent\agent.py", line 127, in _register_tool
if tenant_id in tool_class_with_tenant and self.use_tool_api:
TypeError: argument of type 'OpenAPIPluginTool' is not iterable
Versions / Dependencies
modelscope-agent版本：0.5.2
Reproduction script
schema_openAPI = { "schema": { "openapi":"3.1.0", "info":{ "title":"wanx-v1 Generation API", "description":"API for generating image with wanx-v1", "version":"v1.0.0" }, "servers":[ { "url":"https://dashscope.aliyuncs.com" } ], "paths":{ "/api/v1/services/aigc/text2image/image-synthesis":{ "post":{ "summary":"wanx-v1 text2image", "operationId":"wanx_v1_text2image", "tags":[ "wanx-v1 text2image" ], "requestBody":{ "required":True, "X-DashScope-Async":"enable", "content":{ "application/json":{ "schema":{ "$ref":"#/components/schemas/wanx_v1_text2imageRequest" } } } }, "responses":{ "200":{ "description":"Successful Response", "content":{ "application/json":{ "schema":{ "$ref":"#/components/schemas/wanx_v1_text2imageResponse" } } } } }, "security":[ { "BearerAuth":[] } ] } }, "/api/v1/tasks/{task_id}":{ "get":{ "summary":"Get Text2image Result", "operationId":"gettext2imageresult", "tags":[ "Get Result" ], "parameters":[ { "name":"task_id", "in":"path", "required":True, "description":"The unique identifier of the Text2image generation task", "schema":{ "type":"string" } } ], "security":[ { "BearerAuth":[] } ] } } }, "components":{ "schemas":{ "wanx_v1_text2imageRequest":{ "type":"object", "properties":{ "model":{ "type":"string", "enum":[ "wanx-v1" ] }, "input":{ "type":"object", "properties":{ "prompt":{ "type":"string", "example":"高清的,大师级的,4K,正面", "description":"描述画面的提示词信息", "required": True } } }, "parameters":{ "type":"object", "properties":{ "style":{ "type":"string", "example":"",
"description":"输出图像的风格",
"required": True,
"enum":[
"",
"<3d cartoon>",
"",
"",
"",
"",
"",
""
]
},
"size":{
"type":"string",
"example":"10241024",
"description":"生成图像的分辨率,默认为10241024像素",
"required":True,
"enum":[
"10241024",
"7201280",
"1280*720"
]
},
"n":{
"type":"integer",
"example":1,
"description":"本次请求生成的图片数量",
"required":True
},
"seed":{
"type":"integer",
"example":42,
"description":"图片生成时候的种子值，取值范围为(0,4294967290)",
"required":True
}
}
}
},
"required":[
"model",
"input",
"parameters"
]
},
"wanx_v1_text2imageResponse":{
"type":"object",
"properties":{
"output":{
"type":"string",
"description":"Generated image URL or data."
}
}
}
},
"securitySchemes":{
"ApiKeyAuth":{
"type":"apiKey",
"in":"header",
"name":"Authorization"
}
}
}
},
"auth": {
"type": "API Key",
"apikey": "", # 这里填入API key
"apikey_type": "Bearer"
},
"privacy_policy": ""
}
from modelscope_agent.tools.openapi_plugin import openapi_schema_convert
from modelscope.utils.config import Config
from modelscope_agent.tools.base import TOOL_REGISTRY
from modelscope_agent.tools.openapi_plugin import OpenAPIPluginTool
config_dict = openapi_schema_convert( schema=schema_openAPI["schema"], auth=schema_openAPI["auth"])
plugin_cfg = Config(config_dict)
function_list = []
for name, _ in plugin_cfg.items(): openapi_plugin_object = OpenAPIPluginTool(name=name, cfg=plugin_cfg) TOOL_REGISTRY[name] = openapi_plugin_object function_list.append(name)
from modelscope_agent.agents import RolePlay
role_template = '你扮演哆啦A梦小画家,你需要根据用户的要求用哆啦A梦的语气满足他们' llm_config = { 'model': 'qwen-max', 'model_server': 'dashscope', }
bot = RolePlay(function_list=function_list,llm=llm_config, instruction=role_template)
response = bot.run("哆啦A梦！帮我画一幅可爱的小女孩的照片", remote=False, print_info=True) text = '' for chunk in response: text += chunk print(text)
Issue Severity
None