浏览代码

feat: add xinference llm context size (#2336)

tags/0.5.3
Yeuoly 1年前
父节点
当前提交
0c330fc020
没有帐户链接到提交者的电子邮件

+ 5
- 0
api/core/model_runtime/model_providers/xinference/llm/llm.py 查看文件

@@ -75,6 +75,9 @@ class XinferenceAILargeLanguageModel(LargeLanguageModel):
if extra_param.support_function_call:
credentials['support_function_call'] = True

if extra_param.context_length:
credentials['context_length'] = extra_param.context_length

except RuntimeError as e:
raise CredentialsValidateFailedError(f'Xinference credentials validate failed: {e}')
except KeyError as e:
@@ -296,6 +299,7 @@ class XinferenceAILargeLanguageModel(LargeLanguageModel):
raise ValueError(f'xinference model ability {extra_args.model_ability} is not supported')
support_function_call = credentials.get('support_function_call', False)
context_length = credentials.get('context_length', 2048)

entity = AIModelEntity(
model=model,
@@ -309,6 +313,7 @@ class XinferenceAILargeLanguageModel(LargeLanguageModel):
] if support_function_call else [],
model_properties={
ModelPropertyKey.MODE: completion_type,
ModelPropertyKey.CONTEXT_SIZE: context_length
},
parameter_rules=rules
)

+ 8
- 3
api/core/model_runtime/model_providers/xinference/xinference_helper.py 查看文件

@@ -14,15 +14,17 @@ class XinferenceModelExtraParameter(object):
model_handle_type: str
model_ability: List[str]
max_tokens: int = 512
context_length: int = 2048
support_function_call: bool = False

def __init__(self, model_format: str, model_handle_type: str, model_ability: List[str],
support_function_call: bool, max_tokens: int) -> None:
support_function_call: bool, max_tokens: int, context_length: int) -> None:
self.model_format = model_format
self.model_handle_type = model_handle_type
self.model_ability = model_ability
self.support_function_call = support_function_call
self.max_tokens = max_tokens
self.context_length = context_length

cache = {}
cache_lock = Lock()
@@ -57,7 +59,7 @@ class XinferenceHelper:

url = path.join(server_url, 'v1/models', model_uid)

# this methid is surrounded by a lock, and default requests may hang forever, so we just set a Adapter with max_retries=3
# this method is surrounded by a lock, and default requests may hang forever, so we just set a Adapter with max_retries=3
session = Session()
session.mount('http://', HTTPAdapter(max_retries=3))
session.mount('https://', HTTPAdapter(max_retries=3))
@@ -88,11 +90,14 @@ class XinferenceHelper:
support_function_call = 'tools' in model_ability
max_tokens = response_json.get('max_tokens', 512)

context_length = response_json.get('context_length', 2048)
return XinferenceModelExtraParameter(
model_format=model_format,
model_handle_type=model_handle_type,
model_ability=model_ability,
support_function_call=support_function_call,
max_tokens=max_tokens
max_tokens=max_tokens,
context_length=context_length
)

正在加载...
取消
保存