Open StephenTemp opened 2 years ago
Hello! I want to know if you have do something for this topic.I think these datas are very well.
Sorry for not noticing this thread earlier. I think you can use your own data set to retrain the model. It basically takes only a few hours to train our biggest model on our largest data set on a CPU 4 years ago. It won't take long. I agree that training and testing set should follow similar distributions.
Hello! I'm a student at UMass Amherst hoping to apply TIMME to my research work. It seems, however, a bit cryptic currently to apply the trained model to a new test set.
Any tips or thoughts?