Resolved: Inputs and outputs are a part of the dataset for translation?
Hi! In this, it was explained as the inputs are for example in French and the outputs are for example in English. So the dataset will comprise of input sentences in French, output sentences in English and they will be given to the model separately to an encoder block (a French sentence) and a decoder block (and its corresponding English sentence)?
2 answers ( 1 marked as helpful)
Hi Tulasi
Yes, that's correct. The dataset consists of pairs of sentences: the input sentence in French and the corresponding output sentence in English. During training these sentence pairs are fed in to the model as follows
To Encoder: The French sentence is fed into the encoder.
To Decoder: The English sentence is fed into the decoder.
Thank you!