after a model is chosen through AutoTFT, does the ...
# neural-forecastn
nickeleres
04/18/2024, 2:58 PMafter a model is chosen through AutoTFT, does the chosen model get a full pass of the training data at the end?
m
Marco
04/18/2024, 5:32 PMHello!
Do you mean that the selected model gets retrained again on the training data?
n
nickeleres
04/18/2024, 5:33 PMcorrect
nickeleres
04/18/2024, 5:34 PMin general im having trouble fine tuning the selected TFT model because i have early stopping enabled in hyper parameter search but i dont want fine tuning to stop early, i want it to see 100% of the new data the model wasnt trained on...am I missing something here? @Marco
nickeleres
04/18/2024, 5:40 PMi want to use validation loss only for model selection, not for model training
m
Marco
04/18/2024, 5:57 PMI don't think it's possible at the moment. Early stop always checks the validation loss.
n
nickeleres
04/18/2024, 5:58 PMso what is the recommended way to fine tune a model that has early stop enabled? does fine tuning even work in the canonical way?