findMostFrequentLanguages gives incorrect results about cld3-asm

[hixus]

I also tried wit 3-5x longer text and the results where same.

[kwonoj]

Example sentence in above snippet is too short even if it's 3-5x longer. Recommended min length from cld3 is 150-200 char per length. Shorter text will greatly decrease accuracy. For accuracy and other feature of cld3 I suggest to file issue https://github.com/google/cld3 - this module doesn't do special handling.

[hixus]

hmm, tried with 400-500 char per lang text combinations and still results remained similar. Will look into cld3 repo you linked.