Train using an asymmetric prior?

[artichox]

My corpus has an asymmetric document-topic distribution. Gensim's Lda model allows you to specify alpha as either 'asymmetric', 'auto' (which derives alpha from the data), or as an array of your choice. Is there a way to specify an asymmetric alpha with GuidedLda?

Very helpful library by the way. Thanks for making this available!

[vi3k6i5]

Hey Lucy,

The model currently supports only a symmetric prior. I know how useful an asymmetric prior can be, sorry it's not there as of now :(

Vikash

[shassanin]

Hey Lucy,

The model currently supports only a symmetric prior. I know how useful an asymmetric prior can be, sorry it's not there as of now :(

Vikash

hi vikash!

thank you very much for your efforts and time invested in GuidedLDA. Is GuidedLDA still in development or did you stop developing it?

Greetings

[vi3k6i5]

Hi,

No this is not under development anymore.

Regards, Vikash

On Thu, May 23, 2019 at 5:48 PM Salah Hassanin [email protected] wrote:

Hey Lucy,

The model currently supports only a symmetric prior. I know how useful an asymmetric prior can be, sorry it's not there as of now :(

Vikash

hi vikash!

thank you very much for your efforts and time invested in GuidedLDA. Is GuidedLDA still in development or did you stop developing it?

Greetings

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/vi3k6i5/GuidedLDA/issues/12?email_source=notifications&email_token=AAXY3QTSQX2C3V6MWWVEMBDPW2DRTA5CNFSM4EQR7RV2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWCBDTQ#issuecomment-495194574, or mute the thread https://github.com/notifications/unsubscribe-auth/AAXY3QR45BWYAECOQVGODILPW2DRTANCNFSM4EQR7RVQ .

--

Vikash