Completely refactor the LLM integration method to easily replace the LLM used by each module and optimize the retransmission mechanism.

This commit is contained in:
666ghj
2025-10-09 13:45:39 +08:00
parent ce74f00137
commit 154b29c0d7
73 changed files with 942 additions and 51758 deletions
+4 -4
View File
@@ -230,10 +230,10 @@ def make_retryable_request(
# 预定义一些常用的重试配置
LLM_RETRY_CONFIG = RetryConfig(
max_retries=5, # 增加到5次重试
initial_delay=3.0, # 增加初始延迟到3秒
backoff_factor=1.8, # 调整退避因子
max_delay=45.0 # 增加最大延迟
max_retries=6, # 保持额外重试次数
initial_delay=60.0, # 首次等待至少 1 分钟
backoff_factor=2.0, # 继续使用指数退避
max_delay=600.0 # 单次等待最长 10 分钟
)
SEARCH_API_RETRY_CONFIG = RetryConfig(