automl / Auto-PyTorch

Automatic architecture search and hyperparameter optimization for PyTorch
Apache License 2.0
2.37k stars 287 forks source link

It only support Classification task?? #2

Closed Johnson-yue closed 5 years ago

Johnson-yue commented 5 years ago

hi,auto-pytorch is only support classification task?? If I have some special issue and some loss function, can I use this for find hyper-parameter like classification task?

mlindauer commented 5 years ago

@Johnson-yue Thank you for your interest in AutoPyTorch.

Unfortunately, our main developer is sick this week. So, we will come back to your request next week.

Cheers, Marius

urbanmatthias commented 5 years ago

Hi @Johnson-yue ,

Auto-PyTorch supports classification (binary/multiclass/multilabel) and regression with featurized data.

You can specify the loss function for training the neural networks and a metric (called train_metric, for example accuracy) used by the hyperparameter optimization tool BOHB to evaluate the networks.

Cheers,

Matthias

Johnson-yue commented 5 years ago

If I want to train a encoder-decoder model like auto-encoder,can I use it?

urbanmatthias commented 5 years ago

No, unforunately this is currently not supported and we won't be able to work on that anytime soon.

If you like to extend Auto-PyTorch, we will be happy to accept pull requests.

Cheers, Matthias

Johnson-yue commented 5 years ago

thank you very much!