Error in seq_len(nrow(x)) : argument must be coercible to non-negative integer about etm HOT 4 CLOSED

code seq_len(nrow(x)) is only used when splitting the data in a train/test set in https://github.com/bnosac/ETM/blob/master/R/ETM.R#L398 namely at https://github.com/bnosac/ETM/blob/master/R/ETM.R#L475
The error indicates your dtm argument has no data.

Did you check on what you were passing on to the function calls?

from etm.

code seq_len(nrow(x)) is only used when splitting the data in a train/test set in https://github.com/bnosac/ETM/blob/master/R/ETM.R#L398 namely at https://github.com/bnosac/ETM/blob/master/R/ETM.R#L475 The error indicates your dtm argument has no data.

Did you check on what you were passing on to the function calls?

Thank you, I will work on your query and input above.

Earlier, though, I changed my dataset to a little bit more data, which resulted to these dimensions:

dim(dtm)
[1] 190 31
dim(embeddings)
[1] 31 25

After entering this command: loss <- model$fit(data = dtm, optimizer = optimizer, epoch = 20, batch_size = 1000), the prior error did not come out. But I got this new error instead:

Error in Tensor_slice_put(tensor$ptr, environment(), value, mask = .d) :
rhs must be a torch_tensor or scalar value.

I will review the R code as well...

from etm.

• Check on your input data of dtm and embeddings. Make sure there are no NA values in embeddings due to mismatch between embedding matrix and document term matrix
• Think twice before applying this model on merely 190 text records which is just not what this model is built for

from etm.

• Check on your input data of dtm and embeddings. Make sure there are no NA values in embeddings due to mismatch between embedding matrix and document term matrix
• Think twice before applying this model on merely 190 text records which is just not what this model is built for

I tried the algorithm on the 20 newsgroups dataset, and it worked smoothly! (Just had a ggrepel warning saying "10 unlabeled data points (too many overlaps). Consider increasing max.overlaps.")

I will look for a larger dataset than the one I'm using. I will also look into the NA values in embeddings for the previous dataset. These may be the sources of the original error I had been encountering.

I will also try the other suggested plots in pythonrepo. Thank you very much for your great help! 👍 👍 👍

from etm.