So far, we did not add upper cap or confidence filter but this can be easily done on top of GoldDensityWeigher.get_weights_per_idx:
weights = GoldDensityWeigher.get_weights_per_idx(table)
filtered_weights = {
idx for idx, weight in weights.items() if weight > 0.75
}
If you need, I can help you setting things up ;-)
By the way, for the purpose of curation, we are currently adding a novelty scorer: https://github.com/goldener-data/goldener/pull/419
The workflow would be: You review some examples per class and once validated you keep them as golden dataset. Then you use this dataset to check the remaining ones.
Finally, if you like library, feel free to give us a ā !