Closed paul-reiners closed 1 year ago
I never worked in regression, so I might not be able to provide detailed guidance. Assuming your target lay between 0 and 1, you could adapt the model head to output one number and then use MSELoss. However, it's a bit hard to answer without knowing what exactly you are trying to achieve.
Maybe check out some of these tutorials in regression with fastai:
https://analyticsindiamag.com/a-hands-on-guide-to-regression-with-fast-ai/
https://walkwithfastai.com/Regression_and_Permutation_Importance
How do you convert, say, the ResNet model to handle regression instead of classification?