I Love Natural Language Processing

[Moses-support] word lattice + multiple translation tables optimization problem

Dear all,

I am trying to optimize a system with multiple translation tables and a word lattice input. The first iteration of the optimization is done ok. However, after that, I get the following error:

“The decoder produced also some ‘I’ scores, but we do not know the ranges for them, no way to optimize them”

If I use only one translation table, I get exactly the same error, so it may be a problem related with word lattice and/or binary translation table.

thanks for your concern,
Marta

Re: [Moses-support] word lattice + multiple translation tables optimization problem

I think the issue here has to do with the MERT configuration. When you use lattice input, the weights on the lattice edges are included in the translation model as a feature that MERT can optimize. I’m not familiar with the MERT optimizers that are included with Moses, but it sounds like it is expecting that you configure a range of possible values you expect the feature to get. Hope this helps…
–Chris

NOTICE:This is digested from the Moses-support mailing list, which supports for the moses SMT decoder.

Related posts:

1. Moses Support Digest:ConfusionNet GetSubString error when using lattice with UTF8 input
2. Moses Support Digest: Any documentation about the Multiple Decoding Path functionality
3. Moses Support Digest: moses-irstlm memory racing with 5-gram lm
4. Moses Support Digest: Word Alignment – Moses
5. Moses Support Digest:Alignment information from binary phrase table
6. Moses Support Digest:mert extractor
7. Moses Support Digest:POS translation
8. Moses Support Digest:CreateBerkeleyPt and On-Disk Rule Table
9. Moses Support Digest: Moses seems to hang
10. Moses Support Digest: Issues with Score data