Hi,
this is using "xml-input: exclusive", so none of the phrase table entries
should be used.
-phi
On Fri, Nov 24, 2017 at 11:04 AM, Mike Ladwig wrote:
> Hieu wrote:
> >at a guess, there is an entry in the phrase-table with an erroneous word
> >alignment
>
> >?? @numv@ .
Hieu wrote:
>at a guess, there is an entry in the phrase-table with an erroneous word
>alignment
>?? @numv@ . ||| @numv@ . ||| *0-0 0-1*
Are there any scripts/tools/etc to search for the bad entry? I'm not seeing
anything in just looking, and the training data seems clean as per
at a guess, there is an entry in the phrase-table with an erroneous word
alignment
@numv@ . ||| @numv@ . ||| *0-0 0-1*
On 21/11/17 12:24, Mike Ladwig wrote:
nslates fine as foobar f
--
Hieu Hoang
http://moses-smt.org/
___
Moses-support
I'm doing some large-scale testing and something odd has come up with
placeholders.
I boiled down the various issues and they seem to all be related to these
sentences (which differ only in the final token):
foobar foobar @punctv@ @numv@ foobar
and
foobar foobar @punctv@ @numv@ .
The first