aaron/stock-ai-agent
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

update

Browse Source
This commit is contained in:
aaron 2026-02-10 00:04:22 +08:00
parent 935877b1d1
commit 271938b823
8 changed files with 57 additions and 58 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
backend/app/agent/smart_agent.py
View File

@ -37,11 +37,14 @@ class SmartStockAgent:
# 注册技能
self._register_skills()
# 获取配置的模型
self.model_override = getattr(self.settings, 'smart_agent_model', None)
# 检查LLM是否可用
self.use_llm = bool(self.settings.zhipuai_api_key) and llm_service.client is not None
if self.use_llm:
logger.info("Smart Agent初始化完成（智能模式 + LLM深度集成 + Tushare Pro高级数据）")
logger.info(f"Smart Agent初始化完成（智能模式 + LLM深度集成 + Tushare Pro高级数据，模型: {self.model_override or '默认'}）")
else:
logger.warning("Smart Agent初始化完成（规则模式，建议配置LLM）")
@ -50,7 +53,7 @@ class SmartStockAgent:
loop = asyncio.get_event_loop()
return await loop.run_in_executor(
None,
lambda: llm_service.chat(messages, temperature, max_tokens)
lambda: llm_service.chat(messages, temperature, max_tokens, model_override=self.model_override)
)
def _register_skills(self):
@ -2088,7 +2091,8 @@ RSI：{technical.get('rsi', 0):.2f if technical.get('rsi') else '计算中'}
stream = llm_service.chat_stream(
messages=[{"role": "user", "content": prompt}],
temperature=0.7,
max_tokens=2500 # 增加 token 数量以容纳新闻分析
max_tokens=2500, # 增加 token 数量以容纳新闻分析
model_override=self.model_override
)
for chunk in stream:
@ -2163,7 +2167,8 @@ RSI：{technical.get('rsi', 0):.2f if technical.get('rsi') else '计算中'}
stream = llm_service.chat_stream(
messages=[{"role": "user", "content": prompt}],
temperature=0.7,
max_tokens=2500
max_tokens=2500,
model_override=self.model_override
)
for chunk in stream:
@ -2480,7 +2485,8 @@ RSI：{technical.get('rsi', 0):.2f if technical.get('rsi') else '计算中'}
stream = llm_service.chat_stream(
messages=[{"role": "user", "content": prompt}],
temperature=0.7,
max_tokens=2500
max_tokens=2500,
model_override=self.model_override
)
# 在线程中迭代同步生成器，避免阻塞事件循环
@ -2581,7 +2587,8 @@ MACD：{f"{technical.get('macd'):.4f}" if technical.get('macd') else '计算中'
stream = llm_service.chat_stream(
messages=[{"role": "user", "content": prompt}],
temperature=0.7,
max_tokens=2500 # 增加 token 数量以容纳新闻分析
max_tokens=2500, # 增加 token 数量以容纳新闻分析
model_override=self.model_override
)
# 在线程中迭代同步生成器，避免阻塞事件循环
@ -2984,7 +2991,8 @@ MACD：{f"{technical.get('macd'):.4f}" if technical.get('macd') else '计算中'
stream = llm_service.chat_stream(
messages=[{"role": "user", "content": prompt}],
temperature=0.7,
max_tokens=self._calculate_max_tokens(intent)
max_tokens=self._calculate_max_tokens(intent),
model_override=self.model_override
)
for chunk in stream:

4
backend/app/config.py
View File

@ -116,6 +116,10 @@ class Settings(BaseSettings):
paper_trading_position_b: float = 500 # B级信号仓位 (USDT)
paper_trading_position_c: float = 200 # C级信号仓位 (USDT)
# Agent 模型配置 (可选值: zhipu, deepseek)
smart_agent_model: str = "deepseek" # SmartAgent 使用的模型
crypto_agent_model: str = "deepseek" # CryptoAgent 使用的模型
class Config:
env_file = find_env_file()
case_sensitive = False

7
backend/app/crypto_agent/llm_signal_analyzer.py
View File

@ -143,8 +143,11 @@ class LLMSignalAnalyzer:
def __init__(self):
"""初始化分析器"""
from app.config import get_settings
self.news_service = get_news_service()
logger.info("LLM 信号分析器初始化完成（含新闻舆情）")
settings = get_settings()
self.model_override = getattr(settings, 'crypto_agent_model', None)
logger.info(f"LLM 信号分析器初始化完成（含新闻舆情，模型: {self.model_override or '默认'}）")
async def analyze(self, symbol: str, data: Dict[str, pd.DataFrame],
symbols: List[str] = None) -> Dict[str, Any]:
@ -170,7 +173,7 @@ class LLMSignalAnalyzer:
response = llm_service.chat([
{"role": "system", "content": self.SYSTEM_PROMPT},
{"role": "user", "content": data_prompt}
])
], model_override=self.model_override)
if not response:
logger.warning(f"{symbol} LLM 分析无响应")

16
backend/app/services/llm_service.py
View File

@ -19,16 +19,18 @@ class LLMService:
messages: List[Dict[str, str]],
model: str = None,
temperature: float = 0.7,
max_tokens: int = 2000
max_tokens: int = 2000,
model_override: str = None
) -> Optional[str]:
"""
调用LLM进行对话
Args:
messages: 消息列表
model: 模型名称（忽略，使用当前选择的模型）
model: 模型名称（已废弃，使用 model_override）
temperature: 温度参数
max_tokens: 最大token数
model_override: 指定使用的模型 (zhipu/deepseek)
Returns:
LLM响应文本
@ -36,7 +38,8 @@ class LLMService:
return self.multi_service.chat(
messages=messages,
temperature=temperature,
max_tokens=max_tokens
max_tokens=max_tokens,
model_override=model_override
)
def analyze_intent(self, user_message: str) -> Dict[str, Any]:
@ -47,7 +50,8 @@ class LLMService:
self,
messages: List[Dict[str, str]],
temperature: float = 0.7,
max_tokens: int = 2000
max_tokens: int = 2000,
model_override: str = None
):
"""
流式调用LLM进行对话
@ -56,6 +60,7 @@ class LLMService:
messages: 消息列表
temperature: 温度参数
max_tokens: 最大token数
model_override: 指定使用的模型 (zhipu/deepseek)
Yields:
LLM响应的文本片段
@ -63,7 +68,8 @@ class LLMService:
return self.multi_service.chat_stream(
messages=messages,
temperature=temperature,
max_tokens=max_tokens
max_tokens=max_tokens,
model_override=model_override
)

16
backend/app/services/multi_llm_service.py
View File

@ -66,8 +66,12 @@ class MultiLLMService:
except Exception as e:
logger.error(f"DeepSeek初始化失败: {e}")
# 设置默认模型（优先DeepSeek，因为更便宜）
if 'deepseek' in self.clients:
# 设置默认模型（优先使用配置文件中的设置）
preferred_model = getattr(settings, 'smart_agent_model', None)
if preferred_model and preferred_model in self.clients:
self.current_model = preferred_model
logger.info(f"使用配置的模型: {preferred_model}")
elif 'deepseek' in self.clients:
self.current_model = 'deepseek'
elif 'zhipu' in self.clients:
self.current_model = 'zhipu'
@ -138,10 +142,12 @@ class MultiLLMService:
if provider == 'zhipu':
# 智谱AI调用
# Zhipu对参数更严格，temperature范围是0.0-1.0
safe_temperature = max(0.0, min(1.0, temperature))
response = client.chat.completions.create(
model=model_id,
messages=messages,
temperature=temperature,
temperature=safe_temperature,
max_tokens=max_tokens
)
elif provider == 'deepseek':
@ -205,10 +211,12 @@ class MultiLLMService:
if provider == 'zhipu':
# 智谱AI流式调用
# Zhipu对参数更严格，temperature范围是0.0-1.0
safe_temperature = max(0.0, min(1.0, temperature))
response = client.chat.completions.create(
model=model_id,
messages=messages,
temperature=temperature,
temperature=safe_temperature,
max_tokens=max_tokens,
stream=True
)

6
frontend/css/style.css
View File

@ -135,6 +135,12 @@ html, body {
color: var(--text-primary);
}
.model-name {
color: var(--text-primary);
font-size: 13px;
font-family: inherit;
}
.status {
display: flex;
align-items: center;

8
frontend/index.html
View File

@ -29,17 +29,13 @@
<span>Tradus｜AI金融智能体</span>
</div>
<div class="header-right">
<!-- Model Selector -->
<!-- Model Display (只显示，不可切换) -->
<div class="model-selector" v-if="currentModel">
<svg width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2">
<circle cx="12" cy="12" r="3"/>
<path d="M12 1v6m0 6v6M5.64 5.64l4.24 4.24m4.24 4.24l4.24 4.24M1 12h6m6 0h6M5.64 18.36l4.24-4.24m4.24-4.24l4.24-4.24"/>
</svg>
<select v-model="selectedModel" @change="switchModel" class="model-select">
<option v-for="model in availableModels" :key="model.provider" :value="model.provider">
{{ model.name }}
</option>
</select>
<span class="model-name">{{ currentModel.name }}</span>
</div>
<!-- Logout Button -->
<button class="logout-btn" @click="logout" title="登出">

36
frontend/js/app.js
View File

@ -12,9 +12,7 @@ createApp({
showImageModal: false,
modalImageUrl: '',
showContactModal: false,
availableModels: [],
currentModel: null,
selectedModel: null
currentModel: null
};
},
mounted() {
@ -469,40 +467,10 @@ createApp({
const data = await response.json();
if (data.success) {
this.availableModels = data.models;
this.currentModel = data.current;
this.selectedModel = data.current ? data.current.provider : null;
}
} catch (error) {
}
},
async switchModel() {
if (!this.selectedModel) return;
try {
const response = await fetch('/api/llm/switch', {
method: 'POST',
headers: {
'Content-Type': 'application/json'
},
body: JSON.stringify({
provider: this.selectedModel
})
});
const data = await response.json();
if (data.success) {
this.currentModel = data.current;
this.showNotification(`已切换到 ${data.current.name}`);
} else {
this.showNotification('切换失败');
}
} catch (error) {
this.showNotification('切换失败');
// 忽略错误
}
}
},