<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-0622">
  <Title>Semantic Role Labelling with Tree Conditional Random Fields</Title>
  <Section position="7" start_page="170" end_page="171" type="evalu">
    <SectionTitle>
5 Experimental Results
</SectionTitle>
    <Paragraph position="0"> The model was trained on the full training set after removing unparsable sentences, yielding 90,388 predicates and 1,971,985 binary features. A Gaussian prior was used to regularise the model, with variance s2 = 1. Training was performed on a 20 node PowerPC cluster, consuming a total of 62Gb of RAM and taking approximately 15 hours.</Paragraph>
    <Paragraph position="1"> Decoding required only 3Gb of RAM and about 5 minutes for the 3,228 predicates in the development set. Results are shown in Table 1.</Paragraph>
  </Section>
class="xml-element"></Paper>