Is there any Stateful LSTM in Knime?

Hi,
recently i’m working on some prediction(univariate) using Keras nodes and i’ve found out it is best for me to use Stateful LSTMs.I was wondering if there is any “Stateful LSTM” parameter in Keras nodes?What is the default state?is it stateful or stateless?

Also i have Keras 2.2.4 and KNIME 4.3.2

Thanks for any help.

2 Likes

Hi @Nishef -

Have you checked out the “Return State” option in the Keras LSTM Layer node? As stated in the description, this determines “Whether to return the hidden states in addition to the layer output. If selected the layer returns three tensors, the normal output and the two hidden states of the layer.”

Is this what you’re looking for, or I have I misunderstood your question?

Hi @ScottF
I appreciate your answer,but it’s not what i’m looking for.
according to this siteLSTM layer in python:

  • return_state: Boolean. Whether to return the last state in addition to the output. Default: False.
  • stateful: Boolean (default False). If True, the last state for each sample at index i in a batch will be used as initial state for the sample of index i in the following batch.

Also for more clarification on Stateful and Stateless:

1 Like

Hi @Nishef -

Sorry for the delayed reply here. I asked one of our experts internally, and she said that currently we don’t have a stateful option for the Keras LSTM Layer.

I have created a ticket this for this feature request (AP-16892). Thanks for your patience :slight_smile:

1 Like