Update to FastText v0.9.1

[carschno]

With the release of FastText v0.9.1, JFastText needs to be updated.

Based on an issue reported on original JFastText repo.

[nirzohar]

Cannot compile 0.9.1 version. FastText using newer version of C++ which require std::sort_heap (in loss.cc line 71), so need to update the pom.xml.

[carschno]

@nirzohar just to clarify: are you referring to compiling FastText standalone, or within JFastText? Regarding the former, I have been able to compile v0.9.1 with c++ (GCC) 8.3.1 20190223, from 2018.

Within JFastText, I'm working on updating the wrapper to handle version 0.9.1.

[carschno]

@nirzohar The branch at https://github.com/carschno/JFastText/tree/fasttext_0.9.1 includes the latest FastText version and seems to work fine.
It depends on a fix on FastText whough which still needs to be merged (facebookresearch/fastText#936). However, the branch should already work now.

Note that this does not (yet) fix the issue that this still does not work for Windows and Mac OSX (#5).

[nirzohar]

Thanks for the update.

…

________________________________ From: Carsten Schnober <notifications@github.com> Sent: Friday, October 25, 2019 3:56:05 PM To: carschno/JFastText <JFastText@noreply.github.com> Cc: Nir Zohar <nirzohar@hotmail.com>; Mention <mention@noreply.github.com> Subject: Re: [carschno/JFastText] Update to FastText v0.9.1 (#4) @nirzohar<https://github.com/nirzohar> The branch at https://github.com/carschno/JFastText/tree/fasttext_0.9.1 includes the latest FastText version and seems to work fine. It depends on a fix on FastText whough which still needs to be merged (facebookresearch/fastText#936<facebookresearch/fastText#936>). However, the branch should already work now. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#4?email_source=notifications&email_token=ABLJBJ4F46UL77R7Y32FWKTQQLUGLA5CNFSM4IHM3442YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOECIIMXI#issuecomment-546342493>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABLJBJZBOENAIK4HXEO3KODQQLUGLANCNFSM4IHM344Q>.

[jingcangcang]

@carschno hi, I find the result in terminal through "./fasttext predict" have different result with java class (jft.predictProba(text)), have you verified the results of this two ways?

[nirzohar]

Hey Cannot remember, and I'm no longer into it. Sorry

…

________________________________ From: jingcangcang <notifications@github.com> Sent: Tuesday, August 11, 2020 3:32:40 PM To: carschno/JFastText <JFastText@noreply.github.com> Cc: Nir Zohar <nirzohar@hotmail.com>; Mention <mention@noreply.github.com> Subject: Re: [carschno/JFastText] Update to FastText v0.9.1 (#4) @carschno<https://github.com/carschno> hi, I find the result in terminal through "./fasttext predict" have different result with java class (jft.predictProba(text)), have you verified the results of this two ways? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#4 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABLJBJ2RPY7BN3RCDDNEWYDSAE26RANCNFSM4IHM344Q>.

[carschno]

@carschno hi, I find the result in terminal through "./fasttext predict" have different result with java class (jft.predictProba(text)), have you verified the results of this two ways?

Hi @jingcangcang ,
Thanks for reporting this.

Could you report some more details on this in a dedicated issue. It would be important to know as many details as possible, at least:

• What are the numbers you see in either approach?
• How can you reproduce this?