Commit Graph

  • 64fafa98be * Add senses.pyx and senses.pxd Matthew Honnibal 2015-07-01 18:49:44 +0200
  • 1e8dd0e2c5 * Comple senses.pyx Matthew Honnibal 2015-07-01 18:49:15 +0200
  • 62cfcd76fe * Add supersense sets to lexemes, from WordNet. Look-up via lemmatization. Matthew Honnibal 2015-07-01 18:48:59 +0200
  • aa3d06857e * Upd license page Matthew Honnibal 2015-07-01 16:05:57 +0200
  • be3e935bd5 * Upd readme Matthew Honnibal 2015-07-01 15:39:38 +0200
  • a96f1c26a3 * Upd readme Matthew Honnibal 2015-07-01 15:38:55 +0200
  • 2f46015c34 Merge branch 'master' of ssh://github.com/honnibal/spaCy Matthew Honnibal 2015-07-01 15:37:11 +0200
  • 17792bb4e3 * Announce v0.87 in docs Matthew Honnibal 2015-07-01 15:36:41 +0200
  • d8ef2d6b61 * Upd README.md Matthew Honnibal 2015-07-01 15:27:37 +0200
  • fc0780600e fixed pizza example Kyle McDonald 2015-06-30 21:39:04 -0400
  • 90e2059200 * Include spacy.munge in the built library Matthew Honnibal 2015-06-30 18:35:39 +0200
  • 94dab94e5f uerge branch 'master' of https://github.com/honnibal/spaCy Matthew Honnibal 2015-06-30 18:16:26 +0200
  • 9af86b0b0b * Fix attrs.pxd Matthew Honnibal 2015-06-30 18:16:30 +0200
  • af9c82f7a6 Merge branch 'master' of https://github.com/honnibal/spaCy Matthew Honnibal 2015-06-30 18:11:37 +0200
  • 5d595b5a8c * Inc versions Matthew Honnibal 2015-06-30 18:10:37 +0200
  • d2eeba6667 * Start wiring up color and emotion lexicons. Hopefully we get to use them. Matthew Honnibal 2015-06-30 16:19:43 +0200
  • 341cd0c99f * Require thinc==3.2 Matthew Honnibal 2015-06-30 14:27:11 +0200
  • 31b5e58aeb * Begin reorganizing neuralnet work Matthew Honnibal 2015-06-30 14:26:53 +0200
  • e20106fdff * Begin reorganizing neuralnet work Matthew Honnibal 2015-06-30 14:26:32 +0200
  • 1135cfe50a * Tidy nn_train a bit Matthew Honnibal 2015-06-29 16:45:14 +0200
  • 5cd3ed42d4 * Reenable averaging Matthew Honnibal 2015-06-29 16:44:42 +0200
  • df8179ca4f * Add separate Param and AdadeltaParam classes. AdadeltaParam seems broken. Matthew Honnibal 2015-06-29 16:39:16 +0200
  • 1dff04acb5 * Apply regularization to the softmax, not the bias Matthew Honnibal 2015-06-29 11:45:38 +0200
  • ca30fe1582 * Use He initialization trick Matthew Honnibal 2015-06-29 10:56:02 +0200
  • 894cbef8ba * Wire eta and mu parameters up for neural net Matthew Honnibal 2015-06-29 07:10:33 +0200
  • fc34e1b6e4 * Move Theano functions into nn_train.py script Matthew Honnibal 2015-06-29 07:09:16 +0200
  • 8e7ffd2cdd * Use thinc 3.1 Matthew Honnibal 2015-06-29 02:13:23 +0200
  • 3bb5876c5a * Inline methods in StateClass Matthew Honnibal 2015-06-29 01:06:28 +0200
  • 313a7f87b3 * Inline methods in StateClass Matthew Honnibal 2015-06-29 01:06:28 +0200
  • a02fd3af5d * Check valency in L and R feature methods, to make feaure calculation faster Matthew Honnibal 2015-06-29 00:17:29 +0200
  • 5d870720bc * Check valency in L and R feature methods, to make feaure calculation faster Matthew Honnibal 2015-06-29 00:17:29 +0200
  • f4986d5d3c * Use new Example class Matthew Honnibal 2015-06-28 22:36:03 +0200
  • 735f1af91f * Fix neural net stuff Matthew Honnibal 2015-06-28 11:44:58 +0200
  • fe7b24ecef * whitespace Matthew Honnibal 2015-06-28 01:29:37 +0200
  • e7003f1cf3 * Remove hard-coding of vector lengths Matthew Honnibal 2015-06-27 04:18:47 +0200
  • 7b8275fcc4 * Wire hyperparameters to script interface Matthew Honnibal 2015-06-27 04:18:01 +0200
  • 897dd0dd0b * Merge changes, and adjust Example to use memoryview Matthew Honnibal 2015-06-28 11:36:11 +0200
  • 9282a8e72c * Prepare for new models to be plugged in by using Example class Matthew Honnibal 2015-06-26 13:51:39 +0200
  • 75aeccc064 * Rejig parser interface to use new thinc.api.Example class, in prep of theano model. Comment out beam search Matthew Honnibal 2015-06-26 06:25:36 +0200
  • bf33598b34 * Work on a theano-driven model for the parser Matthew Honnibal 2015-06-23 22:55:58 +0200
  • bbef71f213 * Fix min function in fill_context Matthew Honnibal 2015-06-28 10:46:39 +0200
  • 142b6f9510 * Revert last changes Matthew Honnibal 2015-06-28 10:44:28 +0200
  • b06962f18b * Pad buffers in state Matthew Honnibal 2015-06-28 10:36:14 +0200
  • 53be72387c * Hack at fill_context to investigate performance loss Matthew Honnibal 2015-06-28 10:34:28 +0200
  • b05854bc13 Merge branch 'master' of ssh://github.com/honnibal/spaCy Matthew Honnibal 2015-06-28 09:43:11 +0200
  • 71a4e876a9 * Fix parse features Matthew Honnibal 2015-06-28 09:27:33 +0200
  • aba0257894 * Add lemma rule for better and best in morphs.json Matthew Honnibal 2015-06-28 09:26:25 +0200
  • ef97b90833 * Fix token scoring Matthew Honnibal 2015-06-28 06:22:18 +0200
  • 0c4b5a2bb0 * Start scoring tokens Matthew Honnibal 2015-06-28 06:21:38 +0200
  • 4944d3ba20 * Update requirement to thinc 3.0 Matthew Honnibal 2015-06-28 06:21:20 +0200
  • 5af500909c * Remove unused directve from parser.pyx Matthew Honnibal 2015-06-28 06:20:21 +0200
  • d5b4090705 * Add profile directive Matthew Honnibal 2015-06-28 06:19:33 +0200
  • b5b869366b * Adjust hyphenation rule in tokenizer Matthew Honnibal 2015-06-28 06:18:58 +0200
  • 2b5421e60c * Add profile directive Matthew Honnibal 2015-06-28 06:07:04 +0200
  • 8b5de4a411 * Add word / tag / label sets, for use in neural net Matthew Honnibal 2015-06-28 05:46:53 +0200
  • cfcbd8d256 * Fix punctuation eval in scorer.py Matthew Honnibal 2015-06-28 01:31:39 +0200
  • 65ac389191 * whitespace Matthew Honnibal 2015-06-28 01:29:37 +0200
  • ed40a8380e * Remove hard-coding of vector lengths Matthew Honnibal 2015-06-27 04:18:47 +0200
  • da793073d0 * Wire hyperparameters to script interface Matthew Honnibal 2015-06-27 04:18:01 +0200
  • ebe630cc8d * Enable more features for NN Matthew Honnibal 2015-06-27 04:17:29 +0200
  • f8bb43475e * Bridge to Theano working. Very disorganised. Using thinc adb60aba966ed2 Matthew Honnibal 2015-06-27 02:38:51 +0200
  • 52412867a9 Merge pull request #34 from viksit/patch-1 Matthew Honnibal 2015-06-27 00:14:10 +1000
  • 2fe98b8a9a * Prepare for new models to be plugged in by using Example class Matthew Honnibal 2015-06-26 13:51:39 +0200
  • 6896455884 * Rejig parser interface to use new thinc.api.Example class, in prep of theano model. Comment out beam search Matthew Honnibal 2015-06-26 06:25:36 +0200
  • 335ad66421 * Adjust speed claim in index.rst Matthew Honnibal 2015-06-26 04:43:52 +0200
  • c21ffc84d2 * Upd index.rst Matthew Honnibal 2015-06-26 04:39:48 +0200
  • 91c97009e2 * Declare new release in docs Matthew Honnibal 2015-06-24 05:05:33 +0200
  • b266a63f2c * Inc version of downloadble data Matthew Honnibal 2015-06-24 04:53:08 +0200
  • dc10aa2518 * Increment version Matthew Honnibal 2015-06-24 04:51:38 +0200
  • 886100e1a2 * Increment version Matthew Honnibal 2015-06-24 04:51:38 +0200
  • 02b171ee67 * Bug fixes to edge calculation Matthew Honnibal 2015-06-24 04:28:02 +0200
  • a4e9bdf4c1 * Work on a theano-driven model for the parser Matthew Honnibal 2015-06-23 22:55:58 +0200
  • 7f9384f53c * Remove deprecated _state module Matthew Honnibal 2015-06-23 17:28:24 +0200
  • 6dbe182491 * Fix merge conflicts Matthew Honnibal 2015-06-23 17:28:00 +0200
  • 579735a095 * Remove import of _state module Matthew Honnibal 2015-06-23 17:25:08 +0200
  • 88f55d136b * Remove deprecated _state module Matthew Honnibal 2015-06-23 17:19:51 +0200
  • 9ab9dd2bf7 * Clean up unused orig_arc_eager and tree_arc_eager modules, which were only added for EMNLP experiments Matthew Honnibal 2015-06-23 17:17:33 +0200
  • 7ebfe4b983 * Fixes to edge features Matthew Honnibal 2015-06-23 16:32:54 +0200
  • 7b125f5a86 * Fixes to edge features Matthew Honnibal 2015-06-23 16:31:01 +0200
  • 8d4bbacfc5 * Fix edge navigation in Token objects Matthew Honnibal 2015-06-23 16:07:34 +0200
  • 35c290bee4 * Fix edge features Matthew Honnibal 2015-06-23 15:50:56 +0200
  • 221e2e485f * Assign 'ROOT' as label, not 'root' Matthew Honnibal 2015-06-23 15:08:36 +0200
  • a7bf7b0626 * Rename sent_start to sent_end, to reflect its new usage in the Break transition Matthew Honnibal 2015-06-23 05:39:23 +0200
  • 34c0ef2ee8 * Don't compile the orig_arc_eager and tree_arc_eager modules used for the EMNLP paper Matthew Honnibal 2015-06-23 05:38:17 +0200
  • 59e9f9153c * Remove projectivity constraint in train.py, but raise Exception if non-projective sentence is encountered, since we've told GoldParse to projectivize Matthew Honnibal 2015-06-23 05:04:46 +0200
  • ee3e56f27b * Fix bounds checking on entities Matthew Honnibal 2015-06-23 04:35:08 +0200
  • 43ef5ddea5 * Ensure root albel is spelled ROOT, for backwards compatibility Matthew Honnibal 2015-06-23 04:14:03 +0200
  • 065c2e1d2d * Add some bounds checking around state arrays Matthew Honnibal 2015-06-23 04:13:09 +0200
  • 89ae218b75 * Add import to tokens.pyx from weird Cython compiler issue with casting from memory views Matthew Honnibal 2015-06-23 03:04:34 +0200
  • f01b3d043e * Add padding to arrays in stateclass. May be papering over a deeper bug. Matthew Honnibal 2015-06-23 03:03:22 +0200
  • 5e94b5d581 * Have Tokens return proper numpy arrays, not Cython views. Matthew Honnibal 2015-06-23 00:07:06 +0200
  • 69507bc729 * Re-enable Break transition in arc_eager.pyx Matthew Honnibal 2015-06-23 00:03:30 +0200
  • cc579ed429 * Add __len__ function to StringStore Matthew Honnibal 2015-06-23 00:02:50 +0200
  • 46fb24e9fd * Add cycle-checking code in gold.pyx Matthew Honnibal 2015-06-23 00:02:22 +0200
  • 839e5038b7 * Raise exception on non-projective input Matthew Honnibal 2015-06-23 00:01:55 +0200
  • fe9118a528 * Add test for strip_bad_periods reading in read_conll.parse Matthew Honnibal 2015-06-18 16:36:04 +0200
  • 60d26243e3 * Fix head alignment in read_conll.parse, which was causing corrupt parses when strip_bad_periods=True. A similar problem may apply to other data readers. Matthew Honnibal 2015-06-18 16:35:27 +0200
  • f868175e43 * Whitespace Matthew Honnibal 2015-06-16 23:37:46 +0200
  • ab110be125 * Remove debugging in parser.pyx Matthew Honnibal 2015-06-16 23:37:25 +0200
  • 4dad4058c3 * Uncomment NER training Matthew Honnibal 2015-06-16 23:36:54 +0200