Added malletmodel2ldamodel transformation function to mallet wrapper #766
Conversation
| ------- | ||
| model_gensim : LdaModel instance; copied gensim LdaModel | ||
| """ | ||
| model_gensim = LdaModel(id2word=mallet_model.id2word, num_topics=mallet_model.num_topics, alpha=mallet_model.alpha, iterations=100) |
There was a problem hiding this comment.
Just curious, is there no way to pick up the number of iterations from the mallet model? I notice you've kept it as the default 100 right now.
There was a problem hiding this comment.
Yes even I thought of doing that but I instead just followed what @piskvorky wrote in the mailing list. Both are giving similar results though. I guess I'll switch this to mallet_model.iterations for better resemblance to the mallet model.
devashishd12
changed the title
| tm2 = ldamallet.malletmodel2ldamodel(tm1) | ||
| for document in corpus: | ||
| self.assertAlmostEqual(tm1[document][0][1], tm2[document][0][1], places=1) | ||
| self.assertAlmostEqual(tm1[document][1][1], tm2[document][1][1]) |
There was a problem hiding this comment.
@piskvorky I'm not too sure about this test. Values sometimes are differing at the first place too.
There was a problem hiding this comment.
Try more accurate gamma_threshold and iterations. But I don't think the results can really be made identical -- the inference algorithms are completely different.
I think the best we can hope for is a sanity check -- topics in same order, on some (model, query_document) pair where it's clear what the topic order should be, and where any deviation in topic order is clearly an error.
|
@piskvorky I've addressed the comments. Could you please check? |
| model_gensim = LdaModel(id2word=mallet_model.id2word, num_topics=mallet_model.num_topics, | ||
| alpha=mallet_model.alpha, iterations=mallet_model.iterations) | ||
| model_gensim = LdaModel( | ||
| id2word=mallet_model.id2word, num_topics=mallet_model.num_topics, |
There was a problem hiding this comment.
PEP8: Hanging indent of 4 spaces.
|
@piskvorky I've made the changes. Could you please check? |
This is concerning this thread in the mailing list. @piskvorky @tmylk could you please check?