gardel/Mofox-Core
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

孩子们，知识库的高并发太恐怖了

Browse Source
This commit is contained in:
tt-P607
2025-10-19 14:11:45 +08:00
parent 4f2cb56740
commit 88a4b01556
2 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
scripts/lpmm_learning_tool.py
View File

@@ -204,7 +204,7 @@ def extract_information(paragraphs_dict, model_set):
failed_hashes, open_ie_docs = [], [] failed_hashes, open_ie_docs = [], []
with ThreadPoolExecutor(max_workers=5) as executor: with ThreadPoolExecutor(max_workers=3) as executor:
f_to_hash = { f_to_hash = {
executor.submit(extract_info_sync, p_hash, p, model_set): p_hash executor.submit(extract_info_sync, p_hash, p, model_set): p_hash
for p_hash, p in paragraphs_dict.items() for p_hash, p in paragraphs_dict.items()

4
src/chat/knowledge/embedding_store.py
View File

@@ -30,8 +30,8 @@ from .utils.hash import get_sha256
install(extra_lines=3) install(extra_lines=3)
# 多线程embedding配置常量 # 多线程embedding配置常量
DEFAULT_MAX_WORKERS = 10 # 默认最大线程数 DEFAULT_MAX_WORKERS = 3 # 默认最大线程数
DEFAULT_CHUNK_SIZE = 10 # 默认每个线程处理的数据块大小 DEFAULT_CHUNK_SIZE = 5 # 默认每个线程处理的数据块大小
MIN_CHUNK_SIZE = 1 # 最小分块大小 MIN_CHUNK_SIZE = 1 # 最小分块大小
MAX_CHUNK_SIZE = 50 # 最大分块大小 MAX_CHUNK_SIZE = 50 # 最大分块大小
MIN_WORKERS = 1 # 最小线程数 MIN_WORKERS = 1 # 最小线程数