AveragedPerceptronPython
AveragedPerceptronPython copied to clipboard
这里的`train`是更新条件遗漏了吗?
你好,我想请教一下200行Python代码实现感知机词性标注器中【搜索】哪一节中train函数的一个问题:
文中前端说的是,当guess != true_tag时就更新权重(犯错时),第一个train出现的时候。可是【搜索】这一段中却总是在更新,是遗漏了吗?
guess = self.tagdict.get(word)
if not guess:
feats = self._get_features(
i, word, context, prev, prev2)
guess = self.model.predict(feats)
self.model.update(tags[i], guess, feats)
# Set the history features from the guesses, not the
# true tags
update内部有判断
if truth == guess:
return None
@hankcs thanks