diff --git a/training/domain_proportions.yml b/training/domain_proportions.yml index c979ead..865d7a7 100644 --- a/training/domain_proportions.yml +++ b/training/domain_proportions.yml @@ -1,16 +1,33 @@ -newspaper: 1 -book: 2.5 -technical: 2 -wiki: 3 -legal: 2 -misc: 0.1 -dialogue: 2 -parlementary: 0.05 -forum: 1 -math: 3 -code: 1 -aligned: 3 -legi_dialogue: 2 -legi_spoken: 2 -legi_written: 2 -programming: 1 +fr--newspaper: 13.27 +fr--book: 5.9 +fr--technical: 9.47 +fr--wiki: 1.31 +fr--legi_written: 0.36 +fr--dialogue: 0.01 +fr--legi_spoken: 0.09 +fr--legi_dialogue: 0.08 +en--technical: 16.22 +en--newspaper: 1.62 +en--legi_written: 1.88 +en--wiki: 2.69 +en--forum: 1.41 +en--book: 1.93 +en--math: 4.95 +en--dialogue: 0.26 +en--legi_dialogue: 0.02 +de--wiki: 1.0 +de--legi_written: 0.41 +de--book: 0.06 +de--legi_dialogue: 0.02 +es--wiki: 0.8 +es--legi_written: 0.3 +es--legi_dialogue: 0.01 +es--book: 0.04 +it--wiki: 0.77 +it--legi_written: 0.3 +it--book: 0.04 +es-en--aligned: 0.05 +it-en--aligned: 0.05 +de-fr--aligned: 0.05 +fr-en--aligned: 8.67 +code--programming: 25.96 \ No newline at end of file