Fig. 3.
For each class in the site task, the maximum model output probability for all bigrams in the development training corpus is shown vs the number of training samples.
For each class in the site task, the maximum model output probability for all bigrams in the development training corpus is shown vs the number of training samples.