I am a beginner in pytorch. I am trying to perform a text summarization task on gigawaord dataset.
Till now, I have implemented glove embeddings as a feature and tested it over the LSTM.
I have also separately extracted the contextual layer of every line in the dataset, which contains of subject
,object and predicate for every line in the dataset.
I want to use this contextual layer as an additional feature in my model along with glove embeddings.
How can I convert this contextual layer in a form which can be accepted by the model and is it wise to concatenate with the existing embeddings or should I take some other approach?
Any help is appreciated.
Thanks in advance.