base_train_selfsupervised.txt

linking.similarity_threshold = 0.1
ner.min_name_len = 3
ner.upper_case_limit_len = 3
linking.disamb_length_limit = 3
linking.train_count_threshold = -2
linking.filters = {'cuis': set()}
linking.context_vector_sizes = {'xlong': 27, 'long': 18, 'medium': 9, 'short': 3}
linking.context_vector_weights = {'xlong': 0.1, 'long': 0.4, 'medium': 0.4, 'short': 0.1}
linking.weighted_average_function = lambda step: max(0.1, 1-(step**2*0.0004))
linking.similarity_threshold_type = 'static'