Oops

Author: AngledLuffa

stanza/models/tokenizer.py

@@ -59,8 +59,7 @@ def build_argparse():
     parser.add_argument('--conv_res', type=str, default=None, help="Convolutional residual layers for the RNN")
     parser.add_argument('--rnn_layers', type=int, default=1, help="Layers of RNN in the tokenizer")
     parser.add_argument('--use_dictionary', action='store_true', help="Use dictionary feature. The lexicon is created using the training data and external dict (if any) expected to be found under the same folder of training dataset, formatted as SHORTHAND-externaldict.txt where each line in this file is a word. For example, data/tokenize/zh_gsdsimp-externaldict.txt")
-    parser.add_argument('--feat_funcs', type=str, default=None, help="Comma-separated list of feature functions to use, overriding the default set (space_befor
-e,capitalized,numeric,end_of_para,start_of_para). Leave unset to keep the existing default behavior unchanged. See #1640 for the structural features (labeled_field, phone_id, date_pattern, currency) added to help the English tokenizer distinguish tabular/header text (datelines, signatures, salary tables) from narrative prose near digit-capital boundaries.")
+    parser.add_argument('--feat_funcs', type=str, default=None, help="Comma-separated list of feature functions to use, overriding the default set (space_before,capitalized,numeric,end_of_para,start_of_para). Leave unset to keep the existing default behavior unchanged. See #1640 for the structural features (labeled_field, phone_id, date_pattern, currency) added to help the English tokenizer distinguish tabular/header text (datelines, signatures, salary tables) from narrative prose near digit-capital boundaries.")
 
     parser.add_argument('--max_grad_norm', type=float, default=1.0, help="Maximum gradient norm to clip to")
     parser.add_argument('--anneal', type=float, default=.999, help="Anneal the learning rate by this amount when dev performance deteriorate")