update an existing model with new training data (may be using supervised-append parameter)?

[prachij94]

Hi,

I tried with the fasttext v1 as well as v2, but I am unable to run the supervised-append parameter in the process of adding new training data to my existing trained model.

It just shows the available parameters to run with the fasttext.exe file (which does not include the supervised-command). Any help on this would be great. Thanks.

[Celebio]

Hi @prachij94 , Could you please copy/paste here what you get when it shows the available parameters?

[prachij94]

Hi @Celebio ,

I get the following parameters as available for both v1 and v2:

prachi@DESKTOP-BDNGQSJ ~ $ cd fastText-0.2.0/

prachi@DESKTOP-BDNGQSJ ~/fastText-0.2.0 $ fasttext.exe usage: fasttext

The commands supported by fasttext are:

supervised train a supervised classifier quantize quantize a model to reduce the memory usage test evaluate a supervised classifier predict predict most likely labels predict-prob predict most likely labels with probabilities skipgram train a skipgram model cbow train a cbow model print-word-vectors print word vectors given a trained model print-sentence-vectors print sentence vectors given a trained model nn query for nearest neighbors analogies query for analogies

[prachij94]

Hello @Celebio , any update on the issue?

[Celebio]

Hi @prachij94 , There is no such thing as supervised-append. If you have new data you want to include in an existing model, you have to train everything from scratch.

We made some experiments to handle this case and got promising results, but there is no option in our public code to handle that for the moment.

Regards, Onur