Moses-support Digest, Vol 113, Issue 16

Send Moses-support mailing list submissions to
moses-support@mit.edu

To subscribe or unsubscribe via the World Wide Web, visit
http://mailman.mit.edu/mailman/listinfo/moses-support
or, via email, send a message with subject or body 'help' to
moses-support-request@mit.edu

You can reach the person managing the list at
moses-support-owner@mit.edu

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Moses-support digest..."


Today's Topics:

1. RNNLM Integration? (Jake Ballinger)
2. Re: Training: ferility limit (Philipp Koehn)
3. Re: RNNLM Integration? (Philipp Koehn)
4. Re: RNNLM Integration? (Lane Schwartz)


----------------------------------------------------------------------

Message: 1
Date: Mon, 7 Mar 2016 15:46:56 -0500
From: Jake Ballinger <ballingerj@allegheny.edu>
Subject: [Moses-support] RNNLM Integration?
To: moses-support@mit.edu
Message-ID:
<CA+5d++gmQhspv8LuffLYosan=Okc9RbhNZ8xqB6QSaX=m58ewg@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

Hello everyone,

Has anyone used an RNNLM language model instead of one of the recommended
language models? I was specifically looking at the RNNLM toolkit provided
by Tomas Mikolov at http://rnnlm.org/.

Thank you!

--
Jake Ballinger
Major: Computer Science
Minors: Chinese, French, Spanish, & Math
443-974-6184
ballingerj@allegheny.edu
Box 582
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/mailman/private/moses-support/attachments/20160307/b26baa65/attachment-0001.html

------------------------------

Message: 2
Date: Mon, 7 Mar 2016 15:51:27 -0500
From: Philipp Koehn <phi@jhu.edu>
Subject: Re: [Moses-support] Training: ferility limit
To: Masa Taka <grantaka36@gmail.com>
Cc: "moses-support@mit.edu" <moses-support@mit.edu>
Message-ID:
<CAAFADDCqo9BBjtCD8jNeOexjx-Rn5_WEpx1NF1m2BNfj4umfCA@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

Hi,

it generally makes little sense to align a sentence with 1 word to a
sentence with 28 words.

It may be possible to change the hard-coded limit to 9 words, but for all
practical purposes it is better to filter out sentence pairs that violate
exceed the default ratio.

-phi

On Mon, Mar 7, 2016 at 4:43 AM, Masa Taka <grantaka36@gmail.com> wrote:

> > the maximum allowed limit for a source word fertility
> > source length = 1 target length = 28 ratio 28 ferility limit : 9
>
> With GIZA++ training, we have found some warning as above (in the attached
> .txt file.) This mailing list archive says the following and I know what
> and why.
>
> https://www.mail-archive.com/moses-support%40mit.edu/msg00603.html
>
> Q1) Can we change the preset value "9" to others?
> Q2) Why the initial value set to "9?"
> Q3) I think the message "ferility limit" should be "fertility limit," so
> how do I escalate or report this ? (Or, is a pull request to some
> repository available?)
>
> Regards,
> Masataka
>
>
> _______________________________________________
> Moses-support mailing list
> Moses-support@mit.edu
> http://mailman.mit.edu/mailman/listinfo/moses-support
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/mailman/private/moses-support/attachments/20160307/780ad228/attachment-0001.html

------------------------------

Message: 3
Date: Mon, 7 Mar 2016 16:19:33 -0500
From: Philipp Koehn <phi@jhu.edu>
Subject: Re: [Moses-support] RNNLM Integration?
To: Jake Ballinger <ballingerj@allegheny.edu>
Cc: "moses-support@mit.edu" <moses-support@mit.edu>
Message-ID:
<CAAFADDCEAnU6ERMABRgMsLSiy30Fi1L2a=w6pH_PF861eKGpgA@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

Hi,

integrating this into the decoder will break all hypothesis recombination,
so it may be better (and definitely easier) to use the RNNLM to rerank
n-best lists.

-phi

On Mon, Mar 7, 2016 at 3:46 PM, Jake Ballinger <ballingerj@allegheny.edu>
wrote:

> Hello everyone,
>
> Has anyone used an RNNLM language model instead of one of the recommended
> language models? I was specifically looking at the RNNLM toolkit provided
> by Tomas Mikolov at http://rnnlm.org/.
>
> Thank you!
>
> --
> Jake Ballinger
> Major: Computer Science
> Minors: Chinese, French, Spanish, & Math
> 443-974-6184
> ballingerj@allegheny.edu
> Box 582
>
> _______________________________________________
> Moses-support mailing list
> Moses-support@mit.edu
> http://mailman.mit.edu/mailman/listinfo/moses-support
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/mailman/private/moses-support/attachments/20160307/1928040d/attachment-0001.html

------------------------------

Message: 4
Date: Mon, 7 Mar 2016 15:41:21 -0600
From: Lane Schwartz <dowobeha@gmail.com>
Subject: Re: [Moses-support] RNNLM Integration?
To: Philipp Koehn <phi@jhu.edu>
Cc: "moses-support@mit.edu" <moses-support@mit.edu>
Message-ID:
<CABv3vZ=yU34+SxbUa7Mv72XzS0y9zb86UzoLoidoQRwtHU_-8Q@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

Philipp,

Are you aware of any published work examining the importance of hypothesis
recombination in terms of time/space/quality tradeoffs?

Lane


On Mon, Mar 7, 2016 at 3:19 PM, Philipp Koehn <phi@jhu.edu> wrote:

> Hi,
>
> integrating this into the decoder will break all hypothesis recombination,
> so it may be better (and definitely easier) to use the RNNLM to rerank
> n-best lists.
>
> -phi
>
> On Mon, Mar 7, 2016 at 3:46 PM, Jake Ballinger <ballingerj@allegheny.edu>
> wrote:
>
>> Hello everyone,
>>
>> Has anyone used an RNNLM language model instead of one of the recommended
>> language models? I was specifically looking at the RNNLM toolkit provided
>> by Tomas Mikolov at http://rnnlm.org/.
>>
>> Thank you!
>>
>> --
>> Jake Ballinger
>> Major: Computer Science
>> Minors: Chinese, French, Spanish, & Math
>> 443-974-6184
>> ballingerj@allegheny.edu
>> Box 582
>>
>> _______________________________________________
>> Moses-support mailing list
>> Moses-support@mit.edu
>> http://mailman.mit.edu/mailman/listinfo/moses-support
>>
>>
>
> _______________________________________________
> Moses-support mailing list
> Moses-support@mit.edu
> http://mailman.mit.edu/mailman/listinfo/moses-support
>
>


--
When a place gets crowded enough to require ID's, social collapse is not
far away. It is time to go elsewhere. The best thing about space travel
is that it made it possible to go elsewhere.
-- R.A. Heinlein, "Time Enough For Love"
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/mailman/private/moses-support/attachments/20160307/d3b309b0/attachment.html

------------------------------

_______________________________________________
Moses-support mailing list
Moses-support@mit.edu
http://mailman.mit.edu/mailman/listinfo/moses-support


End of Moses-support Digest, Vol 113, Issue 16
**********************************************

Tweet