Latvian Treebank (LVTB) is in development since 2010. Corpus is annotated manually according to SemTi-Kamols hybrid dependency-constituency grammar model. This version of the treebank contains data used for deriving LVTB version 2.5.