TRI-ML/risk_biased_prediction

plotly version

Pin plotly==5.18.0; instrument figure data so we can see if traces are populated

Pass auth token to load_dataset for private/gated dataset access

debug: print dataset/model state to diagnose empty plot

fix auth issue

fix dependency

Switch to Docker SDK; pin setuptools 59.8.0 to fix mmcv 1.4.7 build

trust_remote_code=true when loading private dataset

Update requirements.txt

free numpy version again

downgrade numpy to 1.18

downgrade numpy to 1.19

update numpy.bool8 to numpy.bool_

update requirements

added arxiv link to model card

change plot dimensions and update markdown text

correction, sdk version, plot size

Change window centering

separate load data and load model, set debug False

try to get access with secret

Removed history to avoid any unverified information being released