Changes made to the way that darwin-py refers to annotation files a few versions ago mean that the "stems" passed to the compute_distributions() function when training in-platform models now end with .json. This breaks this function, which so far as assumed that they don't
Solution
Allow the function to work with both files ending with .json and those that don't
Changelog
Improved tolerance of a darwin-py function used for in-platform model training
linear[bot]
commented
4 months ago
Problem
Changes made to the way that darwin-py refers to annotation files a few versions ago mean that the "stems" passed to the
compute_distributions()function when training in-platform models now end with.json. This breaks this function, which so far as assumed that they don'tSolution
Allow the function to work with both files ending with
.jsonand those that don'tChangelog
Improved tolerance of a darwin-py function used for in-platform model training