The files in this folder contain the BLEU scores obtained on different test datasets 
when translated using translation models trained on parallel corpora of different sizes. 

The format of the files is as follows: 
    1. The first column corresponds to the number of sentences in the parallel corpus
    used to train the translation models. 
    2. The second column corresponds to number of tokens in the source side of the parallel corpus.
    3. The rest of the columns (column 3/4/5/6 depending on the number of test sets 
    for a particular configuration) contain the normalized BLEU score obtained from the Baseline systems. 

Number of test sets available for a particular domain: 
    a.  Europarl - 4 [devtest2006, test2006, test2007, test2008] - All
    available from WMT series of workshops 
    b.  News     - 3 [newstest2009, newstest2010, newstest2011] - All
    available from WMT series of workshops
    b.  EMEA     - 4 [devtest, test1, test2, test3] - Manually partitioned
    from the entire corpus
    d.  KFTT     - 2 [devtest, test] - Available from http://www.phontron.com/kftt/
