Commit Graph

308 Commits

Author SHA1 Message Date
Matthew Honnibal
f35632e2e5 * Remove SBD print statement in train, after SBD evaluation was removed from Scorer 2015-10-09 11:08:58 +02:00
Matthew Honnibal
6ea1601e93 * Add script to train models off the UD treebanks. Note that the UD data is restricted to research purposes only, and should only be used to train models for academic experiments. 2015-10-08 12:01:08 +11:00
Matthew Honnibal
c503654ec1 * Update bin/parser/train for printing output. 2015-10-06 10:35:22 +11:00
alvations
8caedba42a caught more codecs.open -> io.open 2015-09-30 20:20:09 +02:00
alvations
764bdc62e7 caught another codecs.open 2015-09-30 20:16:52 +02:00
Matthew Honnibal
1ae55cb63a * Copy tag_map.json in init_model 2015-09-12 05:54:02 +02:00
Matthew Honnibal
b2e82e55f6 * Create POS model dir in training script 2015-09-08 15:36:23 +02:00
Matthew Honnibal
5ad4527c42 * Rename Deutsch to German 2015-09-06 20:18:58 +02:00
Matthew Honnibal
d1eea2d865 * Update train.py for language-generic spaCy 2015-09-06 17:51:48 +02:00
Matthew Honnibal
950ce36660 * Update init model 2015-09-06 17:51:30 +02:00
Matthew Honnibal
b6b1e1aa12 * Add link for Finnish model 2015-08-27 10:26:02 +02:00
Matthew Honnibal
320ced276a * Add tagger training script 2015-08-27 09:15:41 +02:00
Matthew Honnibal
dc13edd7cb * Refactor init_model to accomodate other languages 2015-08-26 19:14:05 +02:00
Matthew Honnibal
bbf07ac253 * Cut down init_model to work on more languages 2015-08-24 01:05:20 +02:00
Matthew Honnibal
3ecacb9635 * Copy gazetteer file in init_model 2015-08-06 16:07:23 +02:00
Matthew Honnibal
ddc1a5cfe5 * Fix training under python3 2015-07-28 14:09:30 +02:00
Matthew Honnibal
174ed1ad20 * Tighten the frequency filter in init_model 2015-07-27 21:44:51 +02:00
Matthew Honnibal
6047f2aa35 * Fix path to freqs.txt 2015-07-27 02:22:35 +02:00
Matthew Honnibal
0368889d6c * Support gzipped frequencies in init_model 2015-07-26 22:39:22 +02:00
Matthew Honnibal
c4f20847da * Fix init_model for travis tests 2015-07-26 14:03:30 +02:00
Matthew Honnibal
09312b9353 * Fix init_model for travis tests 2015-07-26 13:55:47 +02:00
Matthew Honnibal
90ad717dc4 * Update default freq thresholds in init_model 2015-07-26 01:41:17 +02:00
Matthew Honnibal
6a5e035a48 * Ensure data files are copied for tokenizer in init_model 2015-07-26 01:36:19 +02:00
Matthew Honnibal
ab93898ac6 * Make heuristics more explicit in init_model 2015-07-26 00:22:19 +02:00
Matthew Honnibal
5c04dcd7c1 * Fix init_model 2015-07-25 23:33:02 +02:00
Matthew Honnibal
fd525f0675 * Pass OOV probability around 2015-07-25 23:29:51 +02:00
Matthew Honnibal
5b6bf4d4a6 * Remove probability cap on lexicon 2015-07-25 23:05:51 +02:00
Matthew Honnibal
c62eb110c0 * Fix merge conflict in init_model 2015-07-25 23:04:30 +02:00
Matthew Honnibal
0301472d15 * Fix init_model 2015-07-25 22:56:35 +02:00
Matthew Honnibal
8e800adfbc * Fix init_model 2015-07-25 22:54:08 +02:00
Matthew Honnibal
5f183098e4 Merge branch 'master' of ssh://github.com/honnibal/spaCy 2015-07-25 22:37:04 +02:00
Matthew Honnibal
6076213c16 * Fix init_model script 2015-07-25 22:35:52 +02:00
Matthew Honnibal
1a99eb69da Merge branch 'master' of https://github.com/honnibal/spaCy 2015-07-25 22:19:48 +02:00
Matthew Honnibal
ef448649b3 * Add read_freqs function in init_model 2015-07-25 22:16:36 +02:00
Matthew Honnibal
2e6a60eaec Merge branch 'master' of https://github.com/honnibal/spaCy 2015-07-25 21:14:07 +02:00
Matthew Honnibal
105305b4aa * Upd get_freqs script 2015-07-25 21:13:41 +02:00
Matthew Honnibal
616445e027 * Add simple script to collate frequencies from sorted file 2015-07-25 21:12:45 +02:00
Matthew Honnibal
c52179f5fa * Use print function in train.py, for py 2/3 compatibility 2015-07-24 04:52:35 +02:00
Matthew Honnibal
6be3ee311c Py3 compatibility tweak 2015-07-23 13:13:15 +02:00
Matthew Honnibal
d4407d8e2f Py3 compatibility tweak 2015-07-23 09:45:15 +02:00
Matthew Honnibal
da4821fc14 * Add cluster words to probs in init_model 2015-07-23 09:27:07 +02:00
Matthew Honnibal
4af2595d99 * Fix structure of wordnet directory for init_model 2015-07-23 06:35:38 +02:00
Matthew Honnibal
83c0f0da22 * Remove lemmatizer from init_model 2015-07-23 02:32:34 +02:00
Matthew Honnibal
4729200dfc * Whitespace 2015-07-23 01:19:26 +02:00
Matthew Honnibal
2b7bd46508 * Update get_freqs script 2015-07-22 15:43:06 +02:00
Matthew Honnibal
386246db5b * Update init_model, making language resources optional 2015-07-22 00:25:14 +02:00
Matthew Honnibal
317cbbc015 * Serialization round trip now working with decent API, but with rough spots in the organisation and requiring vocabulary to be fixed ahead of time. 2015-07-19 15:18:17 +02:00
Matthew Honnibal
a6ff7e6ca4 * Fix redundant options in train.py 2015-07-17 22:38:05 +02:00
Matthew Honnibal
6cfa83157e Merge branch 'refactor' of ssh://github.com/honnibal/spaCy into refactor 2015-07-17 21:38:04 +02:00
Matthew Honnibal
38ca0c33f5 Merge branch 'neuralnet' into refactor
Mostly refactors parser, to use new thinc3.2 Example class.
Aim is to remove use of shared memory, so that we can parallelize
over documents easily.

Conflicts:
	setup.py
	spacy/syntax/parser.pxd
	spacy/syntax/parser.pyx
	spacy/syntax/stateclass.pyx
2015-07-14 14:13:47 +02:00
Matthew Honnibal
af54d05d60 * Remove sense stuff from init_model 2015-07-14 10:56:17 +02:00
Matthew Honnibal
3de1b3ef1d * Change get_freqs to take a list of files 2015-07-14 10:55:56 +02:00
Matthew Honnibal
39c93116eb * Add get_freqs script 2015-07-14 02:31:32 +02:00
Matthew Honnibal
62cfcd76fe * Add supersense sets to lexemes, from WordNet. Look-up via lemmatization. 2015-07-01 18:48:59 +02:00
Matthew Honnibal
31b5e58aeb * Begin reorganizing neuralnet work 2015-06-30 14:26:53 +02:00
Matthew Honnibal
1135cfe50a * Tidy nn_train a bit 2015-06-29 16:45:14 +02:00
Matthew Honnibal
df8179ca4f * Add separate Param and AdadeltaParam classes. AdadeltaParam seems broken. 2015-06-29 16:39:16 +02:00
Matthew Honnibal
1dff04acb5 * Apply regularization to the softmax, not the bias 2015-06-29 11:45:38 +02:00
Matthew Honnibal
ca30fe1582 * Use He initialization trick 2015-06-29 10:56:02 +02:00
Matthew Honnibal
fc34e1b6e4 * Move Theano functions into nn_train.py script 2015-06-29 07:09:16 +02:00
Matthew Honnibal
fe7b24ecef * whitespace 2015-06-28 11:37:17 +02:00
Matthew Honnibal
7b8275fcc4 * Wire hyperparameters to script interface 2015-06-28 11:37:17 +02:00
Matthew Honnibal
897dd0dd0b * Merge changes, and adjust Example to use memoryview 2015-06-28 11:36:11 +02:00
Matthew Honnibal
ef97b90833 * Fix token scoring 2015-06-28 06:22:18 +02:00
Matthew Honnibal
34c0ef2ee8 * Don't compile the orig_arc_eager and tree_arc_eager modules used for the EMNLP paper 2015-06-23 05:38:17 +02:00
Matthew Honnibal
59e9f9153c * Remove projectivity constraint in train.py, but raise Exception if non-projective sentence is encountered, since we've told GoldParse to projectivize 2015-06-23 05:04:46 +02:00
Matthew Honnibal
839e5038b7 * Raise exception on non-projective input 2015-06-23 00:01:55 +02:00
Matthew Honnibal
4dad4058c3 * Uncomment NER training 2015-06-16 23:36:54 +02:00
Matthew Honnibal
5699585278 * Use tree_arc_eager system as baseline in experiments 2015-06-15 08:23:43 +02:00
Matthew Honnibal
4841f8ad5e * Set transition system early 2015-06-15 02:54:12 +02:00
Matthew Honnibal
bcfdf126a4 * Add toggle for OrigArcEager system 2015-06-14 20:28:14 +02:00
Matthew Honnibal
c500d72dc2 * Temporarily disable NER, and wire up the verbose flag during training 2015-06-14 17:45:31 +02:00
Matthew Honnibal
ac422492cf * Fix write_parses mode of bin/parser/train.py 2015-06-07 19:08:48 +02:00
Matthew Honnibal
4073533e28 * Upd munge_ewtb for the new json format 2015-06-06 02:10:33 +02:00
Matthew Honnibal
6a1341b29e * Add tb pre-process script 2015-06-06 01:59:44 +02:00
Matthew Honnibal
1736fc5a67 * Add more options to bin/parser/train 2015-06-05 23:49:26 +02:00
Matthew Honnibal
362f87dc3a * Update input corruption method to work with lists as well as trings 2015-06-05 19:33:32 +02:00
Matthew Honnibal
0aed9c9a33 * Fix train.py 2015-06-05 15:50:24 +02:00
Matthew Honnibal
8466600add * Clean up train.py, removing unused tag jackknifing code 2015-06-05 15:01:28 +02:00
Matthew Honnibal
e772b48dcd * Skip sentences of length 1 in training 2015-06-05 02:29:03 +02:00
Matthew Honnibal
e822df0867 * Fix bugs in new greedy/beam parser 2015-06-02 02:01:33 +02:00
Matthew Honnibal
70a7ad89ca * Removed unused imports from train.py 2015-06-02 00:59:09 +02:00
Matthew Honnibal
a3de20118e * Wire up beam-width command line argument 2015-06-02 00:54:12 +02:00
Matthew Honnibal
08044ea70c * Remove try/except around parser.train 2015-05-31 15:21:56 +02:00
Matthew Honnibal
c8a553fe91 * Fix cluster initialization 2015-05-31 15:21:28 +02:00
Matthew Honnibal
d7cc2338e7 * Fix bug in train.py 2015-05-31 06:49:06 +02:00
Matthew Honnibal
c037f80638 * Add case expansion to Brown clusters 2015-05-31 05:50:50 +02:00
Matthew Honnibal
5ab0f233a1 * Ensure words in Brown clusters make it into the vocab, even if they're not in our probs list 2015-05-31 05:46:16 +02:00
Matthew Honnibal
d42dda0372 * Shuffle docs before doing jackknife partition --- otherwise we'll not get the right genre mixes... 2015-05-31 01:25:02 +02:00
Matthew Honnibal
4d8d490547 * Exclude empty sentences in prepare_treebank 2015-05-31 01:12:46 +02:00
Matthew Honnibal
d512d20d81 * Allow parser to jackknife POS tags before training. 2015-05-31 01:11:11 +02:00
Matthew Honnibal
6bbdcc5db5 * Fix gold_preproc flag in train.py 2015-05-30 05:23:02 +02:00
Matthew Honnibal
76300bbb1b * Use updated JSON format, with sentences below paragraphs. Allows use of gold preprocessing flag. 2015-05-30 01:25:46 +02:00
Matthew Honnibal
2d11739f28 * Change data format of JSON corpus, putting sentences into lists with the paragraph 2015-05-30 01:25:00 +02:00
Matthew Honnibal
784e577f45 * Check NER length matches conll length in prepare_treebank 2015-05-29 03:54:06 +02:00
Matthew Honnibal
b76bbbd12c * Read json files recursively from a directory, instead of requiring a single .json file 2015-05-29 03:52:55 +02:00
Matthew Honnibal
ef67ef7a4c * Recomment in training in train.py 2015-05-28 22:40:26 +02:00
Matthew Honnibal
5eb64eeb11 * Print json treebank by genre, instead of by large file 2015-05-28 22:40:01 +02:00
Matthew Honnibal
f42dc1f7d8 * Fix evaluate method in train.py, to use sentences which don't have raw text 2015-05-28 16:30:23 +02:00
Matthew Honnibal
a7cee46fe9 * Update train.py, to support paragraphs where there's no raw_text 2015-05-27 19:14:02 +02:00
Matthew Honnibal
ef1333cf89 * Have prepare_treebank read train/dev/test IDs. 2015-05-27 17:35:05 +02:00
Matthew Honnibal
e140e03516 * Read in OntoNotes. Doesn't support train/test/dev split yet 2015-05-27 17:04:29 +02:00
Matthew Honnibal
895060e774 * Ensure tagger and NER are trained, even if non-projective problem 2015-05-27 03:16:21 +02:00
Matthew Honnibal
32ae2cdabe * In prepare_treebank, move ner into the token descriptions 2015-05-26 19:52:39 +02:00
Matthew Honnibal
61885aee76 * Work on prepare_treebank script, adding NER to it 2015-05-26 19:28:29 +02:00
Matthew Honnibal
15bbbf4901 * Remove cruft from train.py 2015-05-25 07:54:10 +02:00
Matthew Honnibal
fc75210941 * Move spacy.syntax.conll to spacy.gold 2015-05-24 21:35:02 +02:00
Matthew Honnibal
541c62c126 * Remove import of removed read_docparse_file function 2015-05-24 20:05:13 +02:00
Matthew Honnibal
bfeb29ebd1 * Tmp commit 2015-05-24 02:50:14 +02:00
Matthew Honnibal
983d954ef4 * Tmp commit, while switch to new format that assumes alignment happens during training 2015-05-23 17:39:04 +02:00
Matthew Honnibal
f35503018e * Tmp commit of train, while I move to better alignment in gold standard 2015-05-23 17:21:25 +02:00
Matthew Honnibal
3d6b3fc6fb * Restore shuffling, and remove print statements from train.py 2015-05-12 20:27:56 +02:00
Matthew Honnibal
e167355505 * Use JSON docs for training and evaluation. Currently a bug that is costing 0.6 acc 2015-05-12 20:27:56 +02:00
Matthew Honnibal
e0ef6b6992 * Fix alignment in prepare_treebank 2015-05-12 20:27:56 +02:00
Matthew Honnibal
0ad72a77ce * Write JSON files, with both dependency and PSG parses 2015-05-12 20:27:55 +02:00
Matthew Honnibal
5078a32213 * Work on script to format training data as a JSON file. 2015-05-12 20:27:55 +02:00
Matthew Honnibal
fb8d50b3d5 Merge branch 'master' of ssh://github.com/honnibal/spaCy 2015-04-30 12:45:15 +02:00
Matthew Honnibal
4489d87550 * Add cluster=0 by default in init_model 2015-04-29 14:23:13 +02:00
Jordan Suchow
3a8d9b37a6 Remove trailing whitespace 2015-04-19 13:01:38 -07:00
Jordan Suchow
5f0f940a1f Remove unused imports 2015-04-19 01:05:22 -07:00
Matthew Honnibal
693c5a1558 * Exclude clusterings for words only seen 1 or 2 times, as their clusters are unreliable 2015-04-17 04:44:52 +02:00
Matthew Honnibal
adcad4f353 * Clean up train.py 2015-04-15 06:02:04 +02:00
Matthew Honnibal
1629b33082 * Fix copying of tokenizer data in init_model 2015-04-12 04:45:31 +02:00
Matthew Honnibal
a6ac92f077 * Respect the model_dir input parameter to train.py 2015-04-08 22:48:26 +02:00
Matthew Honnibal
baff0f8ad8 * Add docstring explaining script a bit, and add handling of word vectors 2015-04-08 08:20:15 +02:00
Matthew Honnibal
156b70ed82 * Add new script to replace make_lexicon, that does full setup of data 2015-04-08 07:46:53 +02:00
Matthew Honnibal
e775e05313 * Use merge_mwe=False in evaluation in train.py 2015-04-08 00:35:19 +02:00
Matthew Honnibal
f26f381b0e * Add simple ner_tag script 2015-04-03 17:26:58 +02:00
Matthew Honnibal
bb27979352 * Add prepare_vecs script 2015-04-02 06:19:39 +02:00
Matthew Honnibal
cd054c6c9f * Remove stray print statement 2015-03-27 15:20:42 +01:00
Matthew Honnibal
ee385b439a * Ensure StringStore is dumped during training 2015-03-26 16:44:47 +01:00
Matthew Honnibal
e854ba0a13 * Remove support for force_gold flag from GreedyParser, since it's not so useful, and it's clutter 2015-03-26 16:44:47 +01:00
Matthew Honnibal
221f43c370 * Ensure better separation between score printing and training in train.py 2015-03-26 16:44:46 +01:00
Matthew Honnibal
6d49f8717b * Move scoring away from training. Does not support scoring on gold preproc. 2015-03-26 16:44:46 +01:00
Matthew Honnibal
52429625f0 * Add write_parses function 2015-03-26 16:44:46 +01:00
Matthew Honnibal
0c91dd9e15 * Re-enable entity training 2015-03-26 16:44:46 +01:00
Matthew Honnibal
9061bbaf61 * Move to fixing up ent_strings and dep_strings passing 2015-03-26 16:44:45 +01:00
Matthew Honnibal
8f7eeb1c2d * Add verbose flag for Scorer, for debugging, and fix ent_strings bug 2015-03-26 16:44:45 +01:00
Matthew Honnibal
8057a95f20 * NER seems to be working, scoring 69 F. Need to add decision-history features --- currently only use current word, 2 words context. Need refactoring. 2015-03-26 16:44:44 +01:00
Matthew Honnibal
ae235e07b9 * Refactoring working for parser, but now need to rig up features for NER, and then debug oracle etc. 2015-03-26 16:44:44 +01:00
Matthew Honnibal
4539c70542 * Work on updating train script for named entity recognition 2015-03-26 16:44:44 +01:00
Matthew Honnibal
7a1a333f04 * Allow gold tokenization training, for debugging 2015-03-26 16:44:43 +01:00
Matthew Honnibal
f5f15a1ef2 * Tmp commit 2015-03-26 16:44:43 +01:00
Matthew Honnibal
34215de61b * Upd train script, moving lots of functionality to new GoldParse class 2015-03-26 16:44:42 +01:00
Matthew Honnibal
8715101239 * Merge changes from stash 2015-03-26 16:44:42 +01:00
Matthew Honnibal
3a302ae6f2 * Merge train.py 2015-03-26 16:44:41 +01:00
Matthew Honnibal
053814ffc8 * Report LAS in train script 2015-03-03 04:35:11 -05:00
Matthew Honnibal
1019939c7a * Whitespace 2015-02-24 23:03:02 -05:00
Matthew Honnibal
ee33be31dd * Fix parser training script 2015-02-09 03:57:56 -05:00
Matthew Honnibal
27986d7f5c * Fix standard conll file reading. Script needs refactoring. 2015-02-02 23:02:48 +11:00
Matthew Honnibal
543fe5a280 * Comment training function back in 2015-01-31 16:37:57 +11:00
Matthew Honnibal
df749f257d * Upd train script 2015-01-31 13:44:37 +11:00
Matthew Honnibal
ca7577d8a9 * Allow parsers and taggers to be trained on text without gold pre-processing. 2015-01-30 16:36:24 +11:00
Matthew Honnibal
b38093237e * More debug prints 2015-01-30 11:15:54 +11:00
Matthew Honnibal
11ed65b93c * Work on alignment, for evaluation with non-gold preprocessing 2015-01-30 10:31:03 +11:00
Matthew Honnibal
b4348ce1c3 * Messily use unsegmented sentences to train the parser 2015-01-29 04:21:13 +11:00
Matthew Honnibal
72ff9c5082 * Update parser training script for tweaked parser API 2015-01-25 02:20:49 +11:00
Matthew Honnibal
61904e590f * Add parser training script 2015-01-10 04:53:26 +11:00