Add grants_tagger visualize #74
Conversation
| @app.command() | ||
| def visualize(): | ||
| st_app_path = os.path.join(os.path.dirname(__file__), "streamlit_visualize.py") | ||
| subprocess.Popen(["streamlit", "run", st_app_path]) |
There was a problem hiding this comment.
Wonder if there is a way to run Streamlit programatically, like Flasks' import app; app.run()
There was a problem hiding this comment.
I did not find a way, but will look again. Btw @jdu is this a good way if invoking a script? I remember hunter had mentioned that some python way was better than another
There was a problem hiding this comment.
There's a few different ways of spawning a sub-process across the different python versions. This is probably fine for what is just a labs component.
| text = st.text_area('Grant abstract', 'The cell is...', height=300) | ||
|
|
||
| models = { | ||
| "disease_mesh_cnn-2021.03.1": { |
There was a problem hiding this comment.
Nice model dispatcher, although in the future you can probably accept the model config file as parameter to your grants_tagger visualize.
There was a problem hiding this comment.
I plan to save all these in meta.json so that I only have to pass the model_path
|
|
||
| probabilities = st.sidebar.checkbox("Display probabilities") | ||
|
|
||
| with st.spinner('Calculating tags...'): |
Adds
grants_tagger vizualizecommand that runs a streamlit app. It allows the user to select a model, threshold and whether to display probabilities and then paste any text to get tags