From 8be231d49002aac8c4705bdbf8a915414070c706 Mon Sep 17 00:00:00 2001 From: Matthew Honnibal Date: Thu, 8 Sep 2016 13:00:24 +0200 Subject: [PATCH] Tmp. Working on NN NER. --- bin/parser/train.py | 21 ++++++++++++++------- 1 file changed, 14 insertions(+), 7 deletions(-) diff --git a/bin/parser/train.py b/bin/parser/train.py index d430d92b5..7cc26ead4 100755 --- a/bin/parser/train.py +++ b/bin/parser/train.py @@ -91,7 +91,7 @@ def train(Language, gold_tuples, model_dir, dev_loc, n_iter=15, feat_set=u'basic Config.write(dep_model_dir, 'config', feat_set=feat_set, seed=seed, labels=ArcEager.get_labels(gold_tuples), - rho=0.0, eta=1.0, mu=0.9, noise=0.0, + rho=1e-5, eta=1.0, mu=0.9, noise=0.0, beam_width=beam_width,projectivize=pseudoprojective) #feat_set, slots = get_templates('neural') #vector_widths = [10, 10, 10] @@ -112,15 +112,19 @@ def train(Language, gold_tuples, model_dir, dev_loc, n_iter=15, feat_set=u'basic Config.write(ner_model_dir, 'config', feat_set='ner', seed=seed, labels=BiluoPushDown.get_labels(gold_tuples), - beam_width=0, rho=0.0, eta=1.0, mu=0.9, noise=0.0) + beam_width=beam_width, rho=1e-8, eta=1.0, mu=0.9, noise=0.0) if n_sents > 0: gold_tuples = gold_tuples[:n_sents] micro_eval = gold_tuples[:50] nlp = Language(data_dir=model_dir, tagger=False, parser=False, entity=False) nlp.tagger = Tagger.blank(nlp.vocab, Tagger.default_templates()) - nlp.parser = Parser.from_dir(dep_model_dir, nlp.vocab.strings, ArcEager) - nlp.entity = Parser.from_dir(ner_model_dir, nlp.vocab.strings, BiluoPushDown) + if beam_width >= 2: + nlp.parser = Parser.from_dir(dep_model_dir, nlp.vocab.strings, ArcEager) + nlp.entity = BeamParser.from_dir(ner_model_dir, nlp.vocab.strings, BiluoPushDown) + else: + nlp.parser = Parser.from_dir(dep_model_dir, nlp.vocab.strings, ArcEager) + nlp.entity = Parser.from_dir(ner_model_dir, nlp.vocab.strings, BiluoPushDown) print(nlp.parser.model.widths) for raw_text, sents in gold_tuples: for annot_tuples, ctnt in sents: @@ -178,7 +182,7 @@ def _train_epoch(nlp, gold_tuples, eg_seen, itn, dev_loc, micro_eval, nlp.parser.model.nr_active_feat, nlp.entity.model.nr_active_feat)) loss = 0 - nlp.parser.model.learn_rate *= 0.99 + #nlp.parser.model.learn_rate *= 0.99 scorer = score_file(nlp, dev_loc) print('D:\t%d\t%.3f\t%.3f\t%.3f\t%.3f' % (loss, scorer.uas, scorer.ents_f, scorer.tags_acc, scorer.token_acc)) @@ -271,6 +275,7 @@ def write_parses(Language, dev_loc, model_dir, out_loc): train_loc=("Location of training file or directory"), dev_loc=("Location of development file or directory"), model_dir=("Location of output model directory",), + beam_width=("Parser and NER beam width", "option", "k", int), eval_only=("Skip training, and only evaluate", "flag", "e", bool), corruption_level=("Amount of noise to add to training data", "option", "c", float), gold_preproc=("Use gold-standard sentence boundaries in training?", "flag", "g", bool), @@ -282,7 +287,8 @@ def write_parses(Language, dev_loc, model_dir, out_loc): pseudoprojective=("Use pseudo-projective parsing", "flag", "p", bool), ) def main(language, train_loc, dev_loc, model_dir, n_sents=0, n_iter=15, out_loc="", verbose=False, - debug=False, corruption_level=0.0, gold_preproc=False, eval_only=False, pseudoprojective=False): + debug=False, corruption_level=0.0, beam_width=1, + gold_preproc=False, eval_only=False, pseudoprojective=False): lang = spacy.util.get_lang_class(language) if not eval_only: @@ -291,7 +297,8 @@ def main(language, train_loc, dev_loc, model_dir, n_sents=0, n_iter=15, out_loc= feat_set='basic', #'neural' if not debug else 'debug', gold_preproc=gold_preproc, n_sents=n_sents, corruption_level=corruption_level, n_iter=n_iter, - verbose=verbose,pseudoprojective=pseudoprojective) + verbose=verbose, pseudoprojective=pseudoprojective, + beam_width=beam_width) if out_loc: write_parses(lang, dev_loc, model_dir, out_loc) print(model_dir)