Rooba
/
spaCy
mirror of https://github.com/explosion/spaCy.git
1
1
Fork 0
Code Issues Packages Projects Releases Wiki Activity
8,488 Commits 70 Branches 177 Tags 298 MiB
Commit Graph

364 Commits

Author SHA1 Message Date
Matthew Honnibal c059fcb0ba Update thinc requirement 2018-03-25 19:29:36 +02:00
Matthew Honnibal bede11b67c
Improve label management in parser and NER (#2108) 
This patch does a few smallish things that tighten up the training workflow a little, and allow memory use during training to be reduced by letting the GoldCorpus stream data properly.

Previously, the parser and entity recognizer read and saved labels as lists, with extra labels noted separately. Lists were used becaue ordering is very important, to ensure that the label-to-class mapping is stable.

We now manage labels as nested dictionaries, first keyed by the action, and then keyed by the label. Values are frequencies. The trick is, how do we save new labels? We need to make sure we iterate over these in the same order they're added. Otherwise, we'll get different class IDs, and the model's predictions won't make sense.

To allow stable sorting, we map the new labels to negative values. If we have two new labels, they'll be noted as having "frequency" -1 and -2. The next new label will then have "frequency" -3. When we sort by (frequency, label), we then get a stable sort.

Storing frequencies then allows us to make the next nice improvement. Previously we had to iterate over the whole training set, to pre-process it for the deprojectivisation. This led to storing the whole training set in memory. This was most of the required memory during training.

To prevent this, we now store the frequencies as we stream in the data, and deprojectivize as we go. Once we've built the frequencies, we can then apply a frequency cut-off when we decide how many classes to make.

Finally, to allow proper data streaming, we also have to have some way of shuffling the iterator. This is awkward if the training files have multiple documents in them. To solve this, the GoldCorpus class now writes the training data to disk in msgpack files, one per document. We can then shuffle the data by shuffling the paths.

This is a squash merge, as I made a lot of very small commits. Individual commit messages below.

* Simplify label management for TransitionSystem and its subclasses

* Fix serialization for new label handling format in parser

* Simplify and improve GoldCorpus class. Reduce memory use, write to temp dir

* Set actions in transition system

* Require thinc 6.11.1.dev4

* Fix error in parser init

* Add unicode declaration

* Fix unicode declaration

* Update textcat test

* Try to get model training on less memory

* Print json loc for now

* Try rapidjson to reduce memory use

* Remove rapidjson requirement

* Try rapidjson for reduced mem usage

* Handle None heads when projectivising

* Stream json docs

* Fix train script

* Handle projectivity in GoldParse

* Fix projectivity handling

* Add minibatch_by_words util from ud_train

* Minibatch by number of words in spacy.cli.train

* Move minibatch_by_words util to spacy.util

* Fix label handling

* More hacking at label management in parser

* Fix encoding in msgpack serialization in GoldParse

* Adjust batch sizes in parser training

* Fix minibatch_by_words

* Add merge_subtokens function to pipeline.pyx

* Register merge_subtokens factory

* Restore use of msgpack tmp directory

* Use minibatch-by-words in train

* Handle retokenization in scorer

* Change back-off approach for missing labels. Use 'dep' label

* Update NER for new label management

* Set NER tags for over-segmented words

* Fix label alignment in gold

* Fix label back-off for infrequent labels

* Fix int type in labels dict key

* Fix int type in labels dict key

* Update feature definition for 8 feature set

* Update ud-train script for new label stuff

* Fix json streamer

* Print the line number if conll eval fails

* Update children and sentence boundaries after deprojectivisation

* Export set_children_from_heads from doc.pxd

* Render parses during UD training

* Remove print statement

* Require thinc 6.11.1.dev6. Try adding wheel as install_requires

* Set different dev version, to flush pip cache

* Update thinc version

* Update GoldCorpus docs

* Remove print statements

* Fix formatting and links [ci skip]
 2018-03-19 02:58:08 +01:00
Matthew Honnibal 318c23d318 Increment thinc 2018-03-16 13:12:53 +01:00
Matthew Honnibal 39c50225e8 Update thinc 2018-03-16 03:57:47 +01:00
Matthew Honnibal 7be561c8be Fix thinc requirement 2018-03-16 03:34:12 +01:00
Matthew Honnibal 53df6d867b Require new thinc 2018-03-16 03:20:01 +01:00
Matthew Honnibal f2fa8481c4 Require thinc v6.11 2018-03-13 13:59:35 +01:00
ines 9c8a0f6eba Version-lock msgpack-python (see #2015) 2018-02-22 19:42:03 +01:00
Matthew Honnibal f46bf2a7e9 Build _align.pyx 2018-02-20 17:32:13 +01:00
ines 6bba1db4cc Drop six and related hacks as a dependency 2018-02-18 13:29:56 +01:00
ines 002ee80ddf Add html5lib to setup.py to fix six error (see #1924) 2018-02-02 20:32:08 +01:00
Matthew Honnibal 2e449c1fbf Fix compiler flags, addressing #1591 2018-01-14 14:34:36 +01:00
Matthew Honnibal 04a92bd75e Pin msgpack-numpy requirement 2017-12-06 03:24:24 +01:00
Hugo aa898ab4e4 Drop support for EOL Python 2.6 and 3.3 2017-11-26 19:46:24 +02:00
Matthew Honnibal 716ccbb71e Require thinc 6.10.1 2017-11-15 14:59:34 +01:00
Matthew Honnibal 314f5b9cdb Require thinc 6.10.0 2017-10-28 18:20:10 +00:00
Matthew Honnibal 64e4ff7c4b Merge 'tidy-up' changes into branch. Resolve conflicts 2017-10-28 13:16:06 +02:00
ines 7946464742 Remove spacy.tagger (now in pipeline) 2017-10-27 19:45:04 +02:00
Matthew Honnibal 531142a933 Merge remote-tracking branch 'origin/develop' into feature/better-parser 2017-10-27 12:34:48 +00:00
Matthew Honnibal 642eb28c16 Don't compile with OpenMP by default 2017-10-27 10:16:58 +00:00
Matthew Honnibal 90d1d9b230 Remove obsolete parser code 2017-10-26 13:22:45 +02:00
Matthew Honnibal 79fcf8576a Compile with march=native 2017-10-18 21:46:34 +02:00
Matthew Honnibal 2eb0fe4957 Fix setup.py 2017-10-03 21:40:04 +02:00
Matthew Honnibal b49cc8153a Require correct thinc 2017-09-26 10:00:18 -05:00
ines 68f66aebf8 Use pkg_resources instead of pip for is_package (resolves #1293) 2017-09-16 20:27:59 +02:00
Matthew Honnibal 07cdbd1219 Require thinc 6.8.1, for Windows 2017-09-15 22:47:53 +02:00
Matthew Honnibal 96a4a9070b Compile _beam_utils 2017-08-18 21:56:19 +02:00
Matthew Honnibal f9ae86b01c Fix requirement 2017-08-18 20:56:53 +02:00
Matthew Honnibal 69bcacdc09 Merge branch 'develop' of https://github.com/explosion/spaCy into develop 2017-08-18 20:47:13 +02:00
Matthew Honnibal de7f3509d2 Compile CFile, for vector loading 2017-08-18 20:46:41 +02:00
Matthew Honnibal 426f84937f Resolve conflicts when merging new beam parsing stuff 2017-08-18 13:38:32 -05:00
Matthew Honnibal 60d8111245 Require thinc 6.8.1 2017-08-15 03:12:26 -05:00
Matthew Honnibal 52c180ecf5 Revert "Merge branch 'develop' of https://github.com/explosion/spaCy into develop" 
This reverts commit ea8de11ad5, reversing
changes made to 08e443e083.
 2017-08-14 13:00:23 +02:00
Matthew Honnibal b353e4d843 Work on parser beam training 2017-08-12 14:47:45 -05:00
ines 495e042429 Add entry point-style auto alias for "spacy" 
Simplest way to run commands as spacy xxx instead of python -m spacy
xxx, while avoiding environment conflicts
 2017-08-09 12:17:30 +02:00
Matthew Honnibal ff7418b0d9 Update requirements 2017-07-25 18:58:15 +02:00
Matthew Honnibal b4cdd05466 Add vectors.pyx in setup 2017-06-05 12:45:29 +02:00
Matthew Honnibal c811790095 Register vectors.pyx in setup 2017-06-05 12:32:22 +02:00
ines 152dc018a6 Remove syntax iterators from setup.py 2017-06-05 12:30:22 +02:00
Matthew Honnibal a4dcc96c54 Require thinc bugfix 2017-06-05 04:02:52 -05:00
ines 71954d5fe7 Update Thinc version 2017-06-03 10:32:53 +02:00
ines f45cd174bf Update Thinc version 2017-06-02 18:48:16 +02:00
Matthew Honnibal ae8010b526 Move weight serialization to Thinc 2017-06-01 02:56:12 -05:00
Matthew Honnibal 2e364f7ecd Require msgpack 2017-05-29 13:47:29 +02:00
ines 3cc6fe1484 Add pip to requirements.txt and setup.py 2017-05-17 12:04:03 +02:00
Matthew Honnibal 48de4ed49f Require thinc 6.6, and compile the nn_parser module 2017-05-14 01:20:28 +02:00
Matthew Honnibal 825c6403d8 Remove serializer 2017-05-09 17:28:30 +02:00
ines 564939391a Remove spacy.orth 2017-05-09 01:21:47 +02:00
ines 229b8c3974 Tidy up 2017-05-07 18:36:35 +02:00
ines a793174ae9 Use setuptools.find_packages() 2017-05-03 20:11:02 +02:00