LSTM: One to many


I’m trying to understand the One-to-Many LSTM setting. I created a toy problem, that yields good results, however, I’m not completely sure it is thanks to the lstm.

So, I’m creating polynomial curves, and using the parameters to predict all the trajectory. The LSTM part receives a tensor from a previous linear layer. However, I’m not sure it is the way to do one-to-many. Could anyone help ?

Thanks a lot !