search

sberbank-ai-lab / LightAutoML

LAMA - automatic model creation framework
Apache License 2.0
894 stars 92 forks source link

Cannot use mae as loss in regression task #52

Closed Promisery closed 3 years ago

Promisery commented 3 years ago

I'm using LightAutoML 0.2.16. When I define a new task with task = Task('reg', loss='mae') I get sklearn doesn't support in general case mae and will not be used. Btw set metric='mae' works fine for me

alexmryzhkov commented 3 years ago

Hi @Promisery,

That's pretty normal story - it is a warning that means "for sklearn algos we can't use mae as loss, but for other algos it works fine". In case we can't use the loss for some algos they use default loss for training, others use the loss you set.

Alex