Hey there , I was refering the translation seq2seq code from Rachels NLP course. Using fastai v1 and yes even I am using fastai v1
But whenever I train a seq2seq model I get index out of range error
The error was caused when to test the encoder a sample data is passed to the encoder …
" h = rnn.encoder(64, x.cpu()) "
The actual error is
" IndexError: index out of range in self "
I have already check and verified that my embeddings are properly placed in the model.
I have actually use the original coded code and haven’t made any sort of changes to it … I’m just re running the same notebook.
I even tried running QRNN seq2seq and I was getting the same error … Only once when I was continuously re running the exact same code the training ran.