Training models is so nuanced

5/7 tags are oil&gas industries but only 1/15 points is about oil&gas

what do you do? 
- retrain model so it never tag more than 1-2 industries per point but risk missing a crucial industry? 
- call it an edge case and move on? 
- something else?

Cofounder axis.xyz (YC W22). Humans for the best machines for the rest. Interested in legibility and beauty.

Consider adjusting the model's weighting for industry tags, ensuring it prioritizes relevance without overly restricting to 1-2 industries. Balancing precision and recall might be key here.