Grokking LSTM and GRUs some questions (Week 1 and 2)

[Deepti_Prasad]

hi @talwrii

The idea of significance of lstm with similarity does come same as you mentioned for GRU.

The only difference here would be lstm uses activation function for this forget and update gate when it comes to an input.

The significance of GRU and LSTM holds significance when we have a very long sequence of data and the idea behind prediction of next word or translating a corpus(chunk) in a given long sequence of word can be addressed using GRU and LSTM.

I am sharing a link about your query on how lstm holds similarity with GRU,

W1 Quiz: Inconsistent grading logic applied to Question 9: Update Gate and Forget Gate

Hi @Jason_Kirst The question here mentions about update gate and forget related to LSTM, so the two gates uses ru and rf, I am sharing an image from the same week mentioned i.r.t. LSTM [image] The other question you are talking about mentions the update and forget date in the LSTM plays similar role in the GRU layer, there the answer ru and 1-ru holds correct, see the below image [image] The forget gate applies a sigmoid activation function to the input and previous hidden state, produc…

Feel free to ask or give any feedback.

Regards
DP