Closed nmdefries closed 2 years ago
Manually ran the pipeline workflow off of this branch and it works as expected.
Yep, definitely. Pinning arrow
should fix this, but I'm planning to do a pipeline run to double-check that the output matches. That should be fairly quick.
Some models are being dropped during scoring or aren't included for the full date span they should be. This seems to be related to a newer docker image (even though it shouldn't have any changed code), so revert to previous version that was known to work.
Related to #215.