Non-conformable arguments error when invoking lsa$get_word_vectors()

[Silverneo]

I was trying to fit my text data into LSA model and it works. However, when I tried get_word_vectors() method, I received the following error:

Error in diag(x = private$singular_values) %*% private$lsa_factor_matrix : non-conformable arguments

The function definition in model_LSA.R is as follows and I suspect the matrices multiplied do not match each other in dimensions.

    get_word_vectors = function() {
      diag(x = private$singular_values) %*% private$lsa_factor_matrix
    }

I did a test as follows:

svd_fit_d <- lsa$.__enclos_env__$private$singular_values
length(svd_fit_d) # 25
svd_fit_v <- lsa$.__enclos_env__$private$lsa_factor_matrix
dim(svd_fit_v)     # 3458 25

Can you help with this issue? Many thanks!

Chunmeng

[dselivanov]

Thanks for reporting, this is a bug, will fix today.

[dselivanov]

@Silverneo should be fixed now, check latest version from github.

[Silverneo]

@dselivanov Thanks!