yunbiaolin
commented
5 months ago there is other minor issue that in the chat LLM mode, it show up message with <!assistant!> <!user!>, the prompt template should be changed.
Open yunbiaolin opened 5 months ago
yunbiaolin
commented
5 months ago there is other minor issue that in the chat LLM mode, it show up message with <!assistant!> <!user!>, the prompt template should be changed.
Oscilloscope98
commented
5 months ago Hi @yunbiaolin,
For the first issue, we have not reproduced it on our Windows A770. However, after some investigation on your machine, the problem is likely due to the stuck of loading for langchain document loader:
For the second issue, this may due to prompt tuning. We will fix it in further PRs and let you know the updates.
Oscilloscope98
commented
5 months ago there is other minor issue that in the chat LLM mode, it show up message with <!assistant!> <!user!>, the prompt template should be changed.
Hi @yunbiaolin,
Sorry for the late reply. This problem can be resolved by using the latest chatglm3-6b model :)
Followed the steps from https://github.com/intel-analytics/Langchain-Chatchat/blob/ipex-llm/INSTALL_win_arc.md# to install the the Langchain-chatchat on WIndows ARC770.
the installation itself was success, no any issue. but it is keeping running for a long time (> 1 hour) to add a txt file into the knowledge base, still not completed. the txt file has only 28k.
tested the warmup.py for embedding model only, it is success.
from the log, i can see below error, but not sure whether it matter or not.
INFO: 127.0.0.1:51572 - "POST /knowledge_base/recreate_vector_store HTTP/1.1" 200 OK 2024-04-10 17:06:12,399 - _client.py[line:1027] - INFO: HTTP Request: POST http://127.0.0.1:7861/knowledge_base/recreate_vector_store "HTTP/1.1 200 OK" 2024-04-10 17:06:12,403 - faiss_cache.py[line:92] - INFO: loading vector store in 'Intel/vector_store/bge-large-zh-v1.5' from disk. 2024-04-10 17:06:12,453 - faiss_cache.py[line:38] - INFO: 已将向量库 ('Intel', 'bge-large-zh-v1.5') 保存到磁盘 2024-04-10 17:06:21,455 - utils.py[line:295] - INFO: UnstructuredFileLoader used for C:\Users\SAS\Downloads\Langchain-Chatchat-ipex-llm\knowledge_base\Intel\content\ultra.txt 2024-04-10 17:10:46,059 - utils.py[line:95] - ERROR: ReadTimeout: error when post /knowledge_base/update_docs: timed out 2024-04-10 17:10:46,066 - utils.py[line:295] - INFO: UnstructuredFileLoader used for C:\Users\SAS\Downloads\Langchain-Chatchat-ipex-llm\knowledge_base\Intel\content\ultra.txt 2024-04-10 17:11:21,462 - utils.py[line:95] - ERROR: ReadTimeout: error when post /knowledge_base/upload_docs: timed out 2024-04-10 17:11:21,470 - utils.py[line:295] - INFO: UnstructuredFileLoader used for C:\Users\SAS\Downloads\Langchain-Chatchat-ipex-llm\knowledge_base\Intel\content\ultra.txt 2024-04-10 17:15:46,060 - utils.py[line:95] - ERROR: ReadTimeout: error when post /knowledge_base/update_docs: timed out 2024-04-10 17:15:46,079 - utils.py[line:295] - INFO: UnstructuredFileLoader used for C:\Users\SAS\Downloads\Langchain-Chatchat-ipex-llm\knowledge_base\Intel\content\ultra.txt 2024-04-10 17:16:21,468 - utils.py[line:95] - ERROR: ReadTimeout: error when post /knowledge_base/upload_docs: timed out 2024-04-10 17:16:21,474 - utils.py[line:295] - INFO: UnstructuredFileLoader used for C:\Users\SAS\Downloads\Langchain-Chatchat-ipex-llm\knowledge_base\Intel\content\ultra.txt 2024-04-10 17:20:46,079 - utils.py[line:95] - ERROR: ReadTimeout: error when post /knowledge_base/update_docs: timed out 2024-04-10 17:21:21,474 - utils.py[line:95] - ERROR: ReadTimeout: error when post /knowledge_base/upload_docs: timed out INFO: 127.0.0.1:51964 - "POST /knowledge_base/search_docs HTTP/1.1" 200 OK 2024-04-10 17:21:21,928 - _client.py[line:1027] - INFO: HTTP Request: POST http://127.0.0.1:7861/knowledge_base/search_docs "HTTP/1.1 200 OK" INFO: 127.0.0.1:52098 - "POST /knowledge_base/search_docs HTTP/1.1" 200 OK