Can we use Tfidf output as the input to Conv1d ?
As I read some articles, most of them have used Embedding layer to get the input.
In one article
https://towardsdatascience.com/text-classification-with-cnns-in-pytorch-1113df31e79f
self.conv_1 = nn.Conv1d(self.seq_len, self.out_size, self.kernel_1, self.stride)
seq_len is the number of words in a sentence.
In another article
self.convs = nn.ModuleList([nn.Conv1d(in_channels=1,
out_channels=n_filters,
kernel_size=(fs, embed_size))
Since this is for text, they have said, in_channels is 1. So I am confused about this explanation. Can you please explain this to me.