kaknikhil commented on a change in pull request #432: MADLIB-1351 : Added
stopping criteria on perplexity to LDA
URL: https://github.com/apache/madlib/pull/432#discussion_r333279847
##########
File path: src/ports/postgres/modules/lda/test/lda.sql_in
##########
@@ -448,12 +456,15 @@ SELECT lda_train(
'lda_output_data', -- output_data_table
20, -- voc_size
5, -- topic_num
- 3, -- iter_num
+ 10, -- iter_num
10, -- alpha
0.01, -- beta
2, -- evaluate_every
- 10 -- perplexity_tol
+ 100 -- perplexity_tol
);
-select assert(abs(perplexity[2] - perplexity[1]) > .1, 'Perplexity toll is
less than the perplexity difference') from lda_model ;
\ No newline at end of file
+select assert(abs(perplexity[2] - perplexity[1]) <10, 'Perplexity tol is less
than the perplexity difference') from lda_model ;
Review comment:
I think can add another assert to all the dev-check tests `assert that all
the perplexity values are unique`. What do you think ?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
