search

h2oai / h2ogpt

Private chat with local GPT with document, images, video, etc. 100% private, Apache 2.0. Supports oLLaMa, Mixtral, llama.cpp, and more. Demo: https://gpt.h2o.ai/ https://gpt-docs.h2o.ai/
http://h2o.ai
Apache License 2.0
11.44k stars 1.25k forks source link

gradio_client==0.1.3 fails causes gradio app to fail with recursion error when using client_test.py #44

Open pseudotensor opened 1 year ago

pseudotensor commented 1 year ago 
python generate.py  --base_model=gpt2

causes gradio server to fail with recursion error:

clientout.log

Still fails with same requirements back on ebaedb74599aefb6cc0361dcbff92cee5bd6a29c that worked fine

Changing gradio version doesn't help, but changing gradio_client from 0.1.3 back to 0.0.8 leads to no issues on old or current hash: 3548454085fff8498d42f5ca1322bfe406cf08a0

pseudotensor commented 1 year ago

A functioning example:

Server:

python generate.py  --base_model=h2oai/h2ogpt-oig-oasst1-256-6.9b --use_auth_token=True --chat=False --stream_output=False

Client (streaming mode, set stream_output=False for just simple response):

pytest -s -v client_test.py

Result:

I am Open Assistant, a large language model trained by myself.
pseudotensor commented 1 year ago

https://github.com/gradio-app/gradio/issues/3906