更新火山方舟调用 (#3564)

* 更新火山方舟调用

configs/model_config.py.example

@@ -81,7 +81,7 @@ ONLINE_LLM_MODEL = {
 
  # 火山方舟 API，文档参考 https://www.volcengine.com/docs/82379
  "fangzhou-api": {
- "version": "chatglm-6b-model",
+ "version": "", # 对应火山方舟的 endpoint_id
  "version_url": "",
  "api_key": "",
  "secret_key": "",

requirements.txt

@@ -43,7 +43,7 @@ llama-index==0.9.35
 # beautifulsoup4==4.12.2
 # pysrt==1.1.2
 # dashscope==1.13.6 # qwen
-# volcengine==1.0.119 # fangzhou
+# volcengine==1.0.134 # fangzhou
 # uncomment libs if you want to use corresponding vector store
 # pymilvus==2.3.6
 # psycopg2==2.9.9

requirements_api.txt

@@ -50,7 +50,7 @@ pyjwt==2.8.0
 # duckduckgo-search~=3.9.9
 # metaphor-python~=0.1.23
 
-# volcengine>=1.0.119
+# volcengine>=1.0.134
 # pymilvus==2.3.6
 # psycopg2==2.9.9
 # pgvector>=0.2.4

requirements_lite.txt

@@ -30,7 +30,7 @@ youtube-search~=2.1.2
 duckduckgo-search~=3.9.9
 metaphor-python~=0.1.23
 watchdog~=3.0.0
-# volcengine>=1.0.119
+# volcengine>=1.0.134
 # pymilvus>=2.3.4
 # psycopg2==2.9.9
 # pgvector>=0.2.4

server/model_workers/fangzhou.py

@@ -24,9 +24,8 @@ def __init__(
  kwargs.setdefault("context_len", 16384)
  super().__init__(**kwargs)
  self.version = version
-
  def do_chat(self, params: ApiChatParams) -> Dict:
- from volcengine.maas import MaasService
+ from volcengine.maas.v2 import MaasService
 
  params.load_config(self.model_names[0])
  maas = MaasService('maas-api.ml-platform-cn-beijing.volces.com', 'cn-beijing')
@@ -35,9 +34,6 @@ def do_chat(self, params: ApiChatParams) -> Dict:
 
  # document: "https://www.volcengine.com/docs/82379/1099475"
  req = {
- "model": {
- "name": params.version,
- },
  "parameters": {
  # 这里的参数仅为示例，具体可用的参数请参考具体模型的 API 说明
  "max_new_tokens": params.max_tokens,
@@ -49,24 +45,24 @@ def do_chat(self, params: ApiChatParams) -> Dict:
  text = ""
  if log_verbose:
  self.logger.info(f'{self.__class__.__name__}:maas: {maas}')
- for resp in maas.stream_chat(req):
- if error := resp.error:
- if error.code_n > 0:
- data = {
- "error_code": error.code_n,
- "text": error.message,
- "error": {
- "message": error.message,
- "type": "invalid_request_error",
- "param": None,
- "code": None,
- }
+ for resp in maas.stream_chat(params.version, unicode_escape_data(req)):
+ error = resp.error
+ if error and error.code_n > 0:
+ data = {
+ "error_code": error.code_n,
+ "text": error.message,
+ "error": {
+ "message": error.message,
+ "type": "invalid_request_error",
+ "param": None,
+ "code": None,
  }
- self.logger.error(f"请求方舟 API 时发生错误：{data}")
- yield data
- elif chunk := resp.choice.message.content:
- text += chunk
- yield {"error_code": 0, "text": text}
+ }
+ self.logger.error(f"请求方舟 API 时发生错误：{data}")
+ yield data
+ elif chunk := resp.choices and resp.choices[0].message.content:
+ text += chunk
+ yield {"error_code": 0, "text": text}
  else:
  data = {
  "error_code": 500,
@@ -91,6 +87,16 @@ def make_conv_template(self, conv_template: str = None, model_path: str = None)
  )
 
 
+def unicode_escape_data(data):
+ if isinstance(data, str):
+ return data.encode('unicode_escape').decode('ascii')
+ elif isinstance(data, dict):
+ return {key: unicode_escape_data(value) for key, value in data.items()}
+ elif isinstance(data, list):
+ return [unicode_escape_data(item) for item in data]
+ else:
+ return data
+
 if __name__ == "__main__":
  import uvicorn
  from server.utils import MakeFastAPIOffline
@@ -102,4 +108,4 @@ def make_conv_template(self, conv_template: str = None, model_path: str = None)
  )
  sys.modules["fastchat.serve.model_worker"].worker = worker
  MakeFastAPIOffline(app)
- uvicorn.run(app, port=21005)
+ uvicorn.run(app, port=21005)