Tranformer model with float inputs and outputs

[Arben Sabani]

Hi everyone

I think this problem is easy to solve but not sure which parameters in tensor2tensor to set in order to use float types as input. I guess it has something to do with Modalities and the Problem class or hyperparameters, but not sure.

So basically similar to translate problem I have input sequences of float: (batch_size, sequence_size, feature_size) and as output also a float sequences (sequence_size, 1)

Can anyone help hot to handle this case using tensor2tensor and transformer?

Thanks

[Arben Sabani]

I know that the word sequnce in translate problem is converted to tensors of float, but i want to pass over directly float like we do with sentence in case of translate problem.