adai/MemoryBear
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge remote-tracking branch 'upstream/develop' into feat/quota-check-decorator

Browse Source
This commit is contained in:
wxy
2026-04-17 10:38:24 +08:00
parent f883c1469d fd90a4c2ad
commit a105df33ab
34 changed files with 405 additions and 156 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
api/app/services/app_chat_service.py
View File

@@ -120,6 +120,7 @@ class AppChatService:
tools=tools,
deep_thinking=model_parameters.get("deep_thinking", False),
thinking_budget_tokens=model_parameters.get("thinking_budget_tokens"),
json_output=model_parameters.get("json_output", False),
capability=api_key_obj.capability or [],
)
@@ -392,6 +393,7 @@ class AppChatService:
streaming=True,
deep_thinking=model_parameters.get("deep_thinking", False),
thinking_budget_tokens=model_parameters.get("thinking_budget_tokens"),
json_output=model_parameters.get("json_output", False),
capability=api_key_obj.capability or [],
)

2
api/app/services/conversation_service.py
View File

@@ -544,7 +544,7 @@ class ConversationService:
api_key=api_key,
base_url=api_base,
is_omni=is_omni,
support_thinking="thinking" in (capability or []),
capability=capability,
),
type=ModelType(model_type)
)

2
api/app/services/draft_run_service.py
View File

@@ -597,6 +597,7 @@ class AgentRunService:
tools=tools,
deep_thinking=effective_params.get("deep_thinking", False),
thinking_budget_tokens=effective_params.get("thinking_budget_tokens"),
json_output=effective_params.get("json_output", False),
capability=api_key_config.get("capability", []),
)
@@ -853,6 +854,7 @@ class AgentRunService:
streaming=True,
deep_thinking=effective_params.get("deep_thinking", False),
thinking_budget_tokens=effective_params.get("thinking_budget_tokens"),
json_output=effective_params.get("json_output", False),
capability=api_key_config.get("capability", []),
)

8
api/app/services/llm_router.py
View File

@@ -415,9 +415,11 @@ class LLMRouter:
api_key=api_key_config.api_key,
base_url=api_key_config.api_base,
is_omni=api_key_config.is_omni,
support_thinking="thinking" in (api_key_config.capability or []),
temperature=0.3,
max_tokens=500
capability=api_key_config.capability,
extra_params={
"temperature": 0.3,
"max_tokens": 500
}
)
logger.debug(f"创建 LLM 实例 - Provider: {api_key_config.provider}, Model: {api_key_config.model_name}")

2
api/app/services/master_agent_router.py
View File

@@ -393,7 +393,7 @@ class MasterAgentRouter:
api_key=api_key_config.api_key,
base_url=api_key_config.api_base,
is_omni=api_key_config.is_omni,
support_thinking="thinking" in (api_key_config.capability or []),
capability=api_key_config.capability,
extra_params = extra_params
)

2
api/app/services/memory_perceptual_service.py
View File

@@ -233,7 +233,7 @@ class MemoryPerceptualService:
api_key=model_config.api_key,
base_url=model_config.api_base,
is_omni=model_config.is_omni,
support_thinking="thinking" in (model_config.capability or []),
capability=model_config.capability,
)
)
return llm, model_config

3
api/app/services/model_parameter_merger.py
View File

@@ -47,7 +47,8 @@ class ModelParameterMerger:
"n": 1,
"stop": None,
"deep_thinking": False,
"thinking_budget_tokens": None
"thinking_budget_tokens": None,
"json_output": False
}
# 合并参数：默认值 -> 模型配置 -> Agent 配置

8
api/app/services/model_service.py
View File

@@ -125,9 +125,11 @@ class ModelConfigService:
api_key=api_key,
base_url=api_base,
is_omni=is_omni,
support_thinking="thinking" in (capability or []),
temperature=0.7,
max_tokens=100
capability=capability,
extra_params={
"temperature": 0.7,
"max_tokens": 100
}
)
# 根据模型类型选择不同的验证方式

18
api/app/services/multi_agent_orchestrator.py
View File

@@ -2616,9 +2616,11 @@ class MultiAgentOrchestrator:
api_key=api_key_config.api_key,
base_url=api_key_config.api_base,
is_omni=api_key_config.is_omni,
support_thinking="thinking" in (api_key_config.capability or []),
temperature=0.7, # 整合任务使用中等温度
max_tokens=2000
capability=api_key_config.capability,
extra_params={
"temperature": 0.7, # 整合任务使用中等温度
"max_tokens": 2000
}
)
# 创建 LLM 实例
@@ -2795,10 +2797,12 @@ class MultiAgentOrchestrator:
api_key=api_key_config.api_key,
base_url=api_key_config.api_base,
is_omni=api_key_config.is_omni,
support_thinking="thinking" in (api_key_config.capability or []),
temperature=0.7,
max_tokens=2000,
extra_params={"streaming": True} # 启用流式输出
capability=api_key_config.capability,
extra_params={
"temperature": 0.7,
"max_tokens": 2000,
"streaming": True # 启用流式输出
}
)
# 创建 LLM 实例

2
api/app/services/prompt_optimizer_service.py
View File

@@ -186,7 +186,7 @@ class PromptOptimizerService:
api_key=api_config.api_key,
base_url=api_config.api_base,
is_omni=api_config.is_omni,
support_thinking="thinking" in (api_config.capability or []),
capability=api_config.capability,
), type=ModelType(model_config.type))
try:
prompt_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'prompt')

4
api/app/services/shared_chat_service.py
View File

@@ -250,7 +250,8 @@ class SharedChatService:
tools=tools,
deep_thinking=model_parameters.get("deep_thinking", False),
thinking_budget_tokens=model_parameters.get("thinking_budget_tokens"),
capability=api_key_obj.capability or [],
json_output=model_parameters.get("json_output", False),
capability=api_key_obj.capability,
)
# 加载历史消息
@@ -455,6 +456,7 @@ class SharedChatService:
streaming=True,
deep_thinking=model_parameters.get("deep_thinking", False),
thinking_budget_tokens=model_parameters.get("thinking_budget_tokens"),
json_output=model_parameters.get("json_output", False),
capability=api_key_obj.capability or [],
)