A rushed implementation and UI of question-answering models. currently only tested on deepset/roberta-base-squad2
- download this repo as a zip or using Git.
- download and install python 3.10.6 (recommended) and add to PATH (untested on later versions including 3.11.3)
- run Install.bat and wait for it to finish and say
Press any key to continue . . .
- run RunBaseModel.bat
- put the question in the question box and the context in the context box.
- hit "Do Stuff"
you can add the flag --repo-id [hugging face repo name]
line in the batch file to use a different model other then deepset/roberta-base-squad2.
- If you don't have a cuda enabled GPU, edit RunBaseModel.bat and change:
venv\Scripts\python.exe Open_WebUI.py --cache_dir "modelsCache"
to:
venv\Scripts\python.exe Open_WebUI.py --cache_dir "modelsCache" --device "cpu"
- Fix everything