Trouble with nn.embedding in pytorch, expected scalar type Long, but got torch.cuda.FloatTensor (how to fix)?

Question

so I have an RNN encoder that is part of a larger language model, where the process is encode -> rnn -> decode.

As part of my __init__ for my rnn class I have the following:

self.encode_this = nn.Embedding(self.vocab_size, self.embedded_vocab_dim)

now I am trying to implement a forward class, which takes in batches and performs encoding then decoding,

def f_calc(self, batch):
    #Here, batch.shape[0] is the size of batch while batch.shape[1] is the sequence length

    hidden_states = (torch.zeros(self.num_layers, batch.shape[0], self.hidden_vocab_dim).to(device))
    embedded_states = (torch.zeros(batch.shape[0],batch.shape[1], self.embedded_vocab_dim).to(device))

    o1, h = self.encode_this(embedded_states)

however, my problem is always with the encoder which gives me the following error:

/usr/local/lib/python3.6/dist-packages/torch/nn/functional.py in embedding(input, weight, padding_idx, max_norm, norm_type, scale_grad_by_freq, sparse)
   1465         # remove once script supports set_grad_enabled
   1466         _no_grad_embedding_renorm_(weight, input, max_norm, norm_type)
-> 1467     return torch.embedding(weight, input, padding_idx, scale_grad_by_freq, sparse)
   1468 
   1469 

RuntimeError: Expected tensor for argument #1 'indices' to have scalar type Long; but got torch.cuda.FloatTensor instead (while checking arguments for embedding)

Anyone have any idea how to fix? I am completely new to pytorch so please excuse me if this is a stupid question. I know there is some form of type casting involved but I am not sure how to go about doing it...

much appreciated!

Answer 1

Embedding layer expects integers at the input.

import torch as t

emb = t.nn.Embedding(embedding_dim=3, num_embeddings=26)

emb(t.LongTensor([0,1,2]))

Add long() in your code:

embedded_states = (torch.zeros(batch.shape[0],batch.shape[1], self.embedded_vocab_dim).to(device)).long()