Model quantization does not work anymore

[sworddish]

I tried model.quantize yesterday and if failed with the following error message:

Assertion failed: (counts.size() == osz_), function setTargetCounts, file src/model.cc, line 331.

Even if I comment the line 331 out, I can get the compressed model, but the P@1 dropped significantly.

@mwydmuch I hope you can take a look at this issue, I'm trying to fix it right now.

And by the way, there was a few issues which mentioned similar error message in original facebook/fastText forum facebookresearch#73

[mathias3]

same problem here. Code is based on fasttext 0.1.0 - is there a plan to update it to current 0.9.2 or any active /passive development. Or was it just exploratory repo as a POC added to published paper @mwydmuch ?

[mwydmuch]

Hi @sworddish, @mathias3, I plan to move extremeText to the new fastText code, but I will have time for at the end of June.