Commit Graph

6921 Commits

Author SHA1 Message Date
Ines Montani
24f72c669c Merge branch 'develop' into master-tmp 2020-05-21 18:39:06 +02:00
Ines Montani
c6ec19c844 Add missing declaration 2020-05-21 17:30:05 +02:00
Matthew Honnibal
884d9b060d
Merge pull request #5466 from adrianeboyd/feature/omit-extra-lexeme-info
Add option to omit extra lexeme tables in CLI
2020-05-21 16:40:02 +02:00
Matthew Honnibal
e6c4c1a507
Merge pull request #5468 from adrianeboyd/feature/cli-conllu-misc-ner
Improve handling of NER in CoNLL-U MISC
2020-05-21 16:39:46 +02:00
Matthew Honnibal
26cd6a0229
Merge pull request #5462 from adrianeboyd/feature/lemmatizer-all-upos
Extend lemmatizer rules for all UPOS tags
2020-05-21 16:05:31 +02:00
Matthew Honnibal
cad9b290a2
Merge branch 'master' into feature/omit-extra-lexeme-info 2020-05-21 16:04:24 +02:00
Matthew Honnibal
1f572ce89b
Merge pull request #5473 from explosion/fix/travis-tests
Fix Python 2.7 compat
2020-05-21 15:56:16 +02:00
Ines Montani
a9cb2882cb
Rename argument: doc_or_span/obj -> doclike (#5463)
* doc_or_span -> obj

* Revert "doc_or_span -> obj"

This reverts commit 78bb9ff5e0.

* obj -> doclike

* Refer to correct object
2020-05-21 15:17:39 +02:00
Ines Montani
bea863acd2 Fix naming conflict and formatting 2020-05-21 14:24:38 +02:00
Ines Montani
bd6353715a Merge branch 'master' into fix/travis-tests 2020-05-21 14:23:04 +02:00
Ines Montani
d8f3190c0a Tidy up and auto-format 2020-05-21 14:14:01 +02:00
Ines Montani
56de520afd Try to fix tests on Travis (2.7) 2020-05-21 14:04:57 +02:00
adrianeboyd
d45602bc11
Merge branch 'master' into feature/omit-extra-lexeme-info 2020-05-21 10:26:01 +02:00
adrianeboyd
49ef06d793
Add option for base model in init-model CLI (#5467)
Intended for languages like Chinese with a custom tokenizer.
2020-05-20 18:49:11 +02:00
Adriane Boyd
4b229bfc22 Improve handling of NER in CoNLL-U MISC 2020-05-20 18:48:51 +02:00
Matthew Honnibal
609c0ba557
Fix accidentally quadratic runtime in Example.split_sents (#5464)
* Tidy up train-from-config a bit

* Fix accidentally quadratic perf in TokenAnnotation.brackets

When we're reading in the gold data, we had a nested loop where
we looped over the brackets for each token, looking for brackets
that start on that word. This is accidentally quadratic, because
we have one bracket per word (for the POS tags). So we had
an O(N**2) behaviour here that ended up being pretty slow.

To solve this I'm indexing the brackets by their starting word
on the TokenAnnotations object, and having a property to provide
the previous view.

* Fixes
2020-05-20 18:48:18 +02:00
Adriane Boyd
daaa7bf451 Add option to omit extra lexeme tables in CLI 2020-05-20 15:51:44 +02:00
Adriane Boyd
8cba0e41d8 Return lowercase form as default except for PROPN 2020-05-20 15:35:08 +02:00
adrianeboyd
9393253b66
Remove peeking from Parser.begin_training (#5456)
Inspect all instances in `Parser.begin_training` rather than only the
first 1000.
2020-05-20 15:18:06 +02:00
Matthw Honnibal
fda7355508 Fix train-from-config 2020-05-20 12:30:21 +02:00
Matthw Honnibal
24efd54a42 Merge from develop 2020-05-20 12:27:31 +02:00
Sofie Van Landeghem
7f5715a081
Various fixes to NEL functionality, Example class etc (#5460)
* setting KB in the EL constructor, similar to how the model is passed on

* removing wikipedia example files - moved to projects

* throw an error when nlp.update is called with 2 positional arguments

* rewriting the config logic in create pipe to accomodate for other objects (e.g. KB) in the config

* update config files with new parameters

* avoid training pipeline components that don't have a model (like sentencizer)

* various small fixes + UX improvements

* small fixes

* set thinc to 8.0.0a9 everywhere

* remove outdated comment
2020-05-20 11:41:12 +02:00
Adriane Boyd
4fa9670537 Extend lemmatizer rules for all UPOS tags 2020-05-20 10:15:43 +02:00
Matthew Honnibal
664a3603b0 Set version to v3.0.0.dev8 2020-05-19 17:15:39 +02:00
adrianeboyd
40e65d6f63
Fix most_similar for vectors with unused rows (#5348)
* Fix most_similar for vectors with unused rows

Address issues related to the unused rows in the vector table and
`most_similar`:

* Update `most_similar()` to search only through rows that are in use
according to `key2row`.

* Raise an error when `most_similar(n=n)` is larger than the number of
vectors in the table.

* Set and restore `_unset` correctly when vectors are added or
deserialized so that new vectors are added in the correct row.

* Set data and keys to the same length in `Vocab.prune_vectors()` to
avoid spurious entries in `key2row`.

* Fix regression test using `most_similar`

Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-05-19 16:41:26 +02:00
Sofie Van Landeghem
f00de445dd
default models defined in component decorator (#5452)
* move defaults to pipeline and use in component decorator

* black formatting

* relative import
2020-05-19 16:20:03 +02:00
adrianeboyd
70da1fd2d6
Add warning for misaligned character offset spans (#5007)
* Add warning for misaligned character offset spans

* Resolve conflict

* Filter warnings in example scripts

Filter warnings in example scripts to show warnings once, in particular
warnings about misaligned entities.

Co-authored-by: Ines Montani <ines@ines.io>
2020-05-19 16:01:18 +02:00
adrianeboyd
0061992d95
Update Polish tokenizer for UD_Polish-PDB (#5432)
Update Polish tokenizer for UD_Polish-PDB, which is a relatively major
change from the existing tokenizer. Unused exceptions files and
conflicting test cases removed.

Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-05-19 15:59:55 +02:00
adrianeboyd
a5cd203284
Reduce stored lexemes data, move feats to lookups (#5238)
* Reduce stored lexemes data, move feats to lookups

* Move non-derivable lexemes features (`norm / cluster / prob`) to
`spacy-lookups-data` as lookups
  * Get/set `norm` in both lookups and `LexemeC`, serialize in lookups
  * Remove `cluster` and `prob` from `LexemesC`, get/set/serialize in
    lookups only
* Remove serialization of lexemes data as `vocab/lexemes.bin`
  * Remove `SerializedLexemeC`
  * Remove `Lexeme.to_bytes/from_bytes`
* Modify normalization exception loading:
  * Always create `Vocab.lookups` table `lexeme_norm` for
    normalization exceptions
  * Load base exceptions from `lang.norm_exceptions`, but load
    language-specific exceptions from lookups
  * Set `lex_attr_getter[NORM]` including new lookups table in
    `BaseDefaults.create_vocab()` and when deserializing `Vocab`
* Remove all cached lexemes when deserializing vocab to override
  existing normalizations with the new normalizations (as a replacement
  for the previous step that replaced all lexemes data with the
  deserialized data)

* Skip English normalization test

Skip English normalization test because the data is now in
`spacy-lookups-data`.

* Remove norm exceptions

Moved to spacy-lookups-data.

* Move norm exceptions test to spacy-lookups-data

* Load extra lookups from spacy-lookups-data lazily

Load extra lookups (currently for cluster and prob) lazily from the
entry point `lg_extra` as `Vocab.lookups_extra`.

* Skip creating lexeme cache on load

To improve model loading times, do not create the full lexeme cache when
loading. The lexemes will be created on demand when processing.

* Identify numeric values in Lexeme.set_attrs()

With the removal of a special case for `PROB`, also identify `float` to
avoid trying to convert it with the `StringStore`.

* Skip lexeme cache init in from_bytes

* Unskip and update lookups tests for python3.6+

* Update vocab pickle to include lookups_extra

* Update vocab serialization tests

Check strings rather than lexemes since lexemes aren't initialized
automatically, account for addition of "_SP".

* Re-skip lookups test because of python3.5

* Skip PROB/float values in Lexeme.set_attrs

* Convert is_oov from lexeme flag to lex in vectors

Instead of storing `is_oov` as a lexeme flag, `is_oov` reports whether
the lexeme has a vector.

Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-05-19 15:59:14 +02:00
Sofie Van Landeghem
0d94737857
Feature toggle_pipes (#5378)
* make disable_pipes deprecated in favour of the new toggle_pipes

* rewrite disable_pipes statements

* update documentation

* remove bin/wiki_entity_linking folder

* one more fix

* remove deprecated link to documentation

* few more doc fixes

* add note about name change to the docs

* restore original disable_pipes

* small fixes

* fix typo

* fix error number to W096

* rename to select_pipes

* also make changes to the documentation

Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-05-18 22:27:10 +02:00
Matthew Honnibal
333b1a308b
Adapt parser and NER for transformers (#5449)
* Draft layer for BILUO actions

* Fixes to biluo layer

* WIP on BILUO layer

* Add tests for BILUO layer

* Format

* Fix transitions

* Update test

* Link in the simple_ner

* Update BILUO tagger

* Update __init__

* Import simple_ner

* Update test

* Import

* Add files

* Add config

* Fix label passing for BILUO and tagger

* Fix label handling for simple_ner component

* Update simple NER test

* Update config

* Hack train script

* Update BILUO layer

* Fix SimpleNER component

* Update train_from_config

* Add biluo_to_iob helper

* Add IOB layer

* Add IOBTagger model

* Update biluo layer

* Update SimpleNER tagger

* Update BILUO

* Read random seed in train-from-config

* Update use of normal_init

* Fix normalization of gradient in SimpleNER

* Update IOBTagger

* Remove print

* Tweak masking in BILUO

* Add dropout in SimpleNER

* Update thinc

* Tidy up simple_ner

* Fix biluo model

* Unhack train-from-config

* Update setup.cfg and requirements

* Add tb_framework.py for parser model

* Try to avoid memory leak in BILUO

* Move ParserModel into spacy.ml, avoid need for subclass.

* Use updated parser model

* Remove incorrect call to model.initializre in PrecomputableAffine

* Update parser model

* Avoid divide by zero in tagger

* Add extra dropout layer in tagger

* Refine minibatch_by_words function to avoid oom

* Fix parser model after refactor

* Try to avoid div-by-zero in SimpleNER

* Fix infinite loop in minibatch_by_words

* Use SequenceCategoricalCrossentropy in Tagger

* Fix parser model when hidden layer

* Remove extra dropout from tagger

* Add extra nan check in tagger

* Fix thinc version

* Update tests and imports

* Fix test

* Update test

* Update tests

* Fix tests

* Fix test

Co-authored-by: Ines Montani <ines@ines.io>
2020-05-18 22:23:33 +02:00
Ines Montani
a41e28ceba
Merge pull request #5436 from ilivans/fix_errors_with_codes 2020-05-18 10:45:56 +02:00
Ilkyu Ju
72a25c9cef
Very minor issues in Korean example sentences (#5446)
* Add contributor agreement

* Improve ko translation of example sentences

I fixed unnatural translations and word spacing errors.

* Update osori.md
2020-05-17 13:43:34 +02:00
svlandeg
6fb6a8518c bump to 3.0.0.dev7 and thinc to 8.0.0a8 2020-05-15 13:25:54 +02:00
svlandeg
047f3d7d94 remove ops argument for Adam 2020-05-15 13:25:00 +02:00
svlandeg
e0fda2bd81 throw warning when model_cfg is None 2020-05-15 11:02:10 +02:00
adrianeboyd
908dea3939
Skip duplicate lexeme rank setting (#5401)
Skip duplicate lexeme rank setting within
`_fix_pretrained_vectors_name()`.
2020-05-14 18:26:12 +02:00
adrianeboyd
f49e2810e6
Add Polish lemmatizer (#5413)
* Add Polish lemmatizer

Contributed by @ryszardtuora

* Add missing import
2020-05-14 18:23:19 +02:00
adrianeboyd
e63880e081
Use Token.sent_start for Span.sent (#5439)
Use `Token.sent_start` for sentence boundaries in `Span.sent` so that
`Doc.sents` and `Span.sent` return the same sentence boundaries.
2020-05-14 18:22:51 +02:00
adrianeboyd
780b869345
Fix syntax iterators for Persian (#5437) 2020-05-14 16:51:03 +02:00
Ilia Ivanov
712d9d4820 fixup! Fix ErrorsWithCodes().__class__ return value 2020-05-14 15:45:58 +02:00
Ilia Ivanov
a987e9e45d Fix ErrorsWithCodes().__class__ return value 2020-05-14 14:14:15 +02:00
Vishnu Priya VR
9ce059dd06
Limiting noun_chunks for specific languages (#5396)
* Limiting noun_chunks for specific langauges

* Limiting noun_chunks for specific languages

Contributor Agreement

* Addressing review comments

* Removed unused fixtures and imports

* Add fa_tokenizer in test suite

* Use fa_tokenizer in test

* Undo extraneous reformatting

Co-authored-by: adrianeboyd <adrianeboyd@gmail.com>
2020-05-14 12:58:06 +02:00
Sofie Van Landeghem
b04738903e
prevent None in gold fields (#5425)
* set gold fields to empty list instead of keeping them as None

* add unit test
2020-05-13 22:08:50 +02:00
adrianeboyd
113e7981d0
Check that row is within bounds when adding vector (#5430)
Check that row is within bounds for the vector data array when adding a
vector.

Don't add vectors with rank OOV_RANK in `init-model` (change is due to
shift from OOV as 0 to OOV as OOV_RANK).
2020-05-13 22:08:28 +02:00
adrianeboyd
07639dd6ac
Remove TAG from da/sv tokenizer exceptions (#5428)
Remove `TAG` value from Danish and Swedish tokenizer exceptions because
it may not be included in a tag map (and these settings are problematic
as tokenizer exceptions anyway).
2020-05-13 10:25:54 +02:00
adrianeboyd
24e7108f80
Modify array type to accommodate OOV_RANK (#5429)
Modify indices array type in `Vocab.prune_vectors` to accommodate
OOV_RANK index as max(uint64).
2020-05-13 10:25:05 +02:00
svlandeg
102c8c7e2f fix fan_in renaming 2020-05-12 13:56:10 +02:00
adrianeboyd
440b81bddc
Improve exceptions for 'd (would/had) in English (#5379)
Instead of treating `'d` in contractions like `I'd` as `would` in all
cases in the tokenizer exceptions, leave the tagging and lemmatization
up to later components.
2020-05-08 15:10:57 +02:00
adrianeboyd
c963e269ba
Add method to update / reset pkuseg user dict (#5404) 2020-05-08 11:21:46 +02:00
Samuel Rodríguez Medina
5e55bfa821
Fixed tests for Swedish that were written in Danish. (#5395) 2020-05-05 14:06:27 +02:00
Matthew Honnibal
eb117e2fce Add load_config_from_str helper 2020-05-02 14:09:21 +02:00
adrianeboyd
c045a9c7f6
Fix logic in train CLI timing eval on GPU (#5387)
Run CPU timing in first iteration only
2020-05-01 12:05:33 +02:00
Samuel Rodríguez Medina
148b036e0c
Spanish like num improvement (#5381)
* Add tests for Spanish like_num.

* Add missing numbers in Spanish lexical attributes for like_num.

* Modify Spanish test function name.

* Add contributor agreement.
2020-04-30 11:13:23 +02:00
Samuel Rodríguez Medina
8602daba85
Swedish like_num (#5371)
* Sign contributor agreement.

* Add like_num functionality to Swedish.

* Update spacy/tests/lang/sv/test_lex_attrs.py

Co-Authored-By: Sofie Van Landeghem <svlandeg@users.noreply.github.com>

* Update contributor agreement

Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
2020-04-29 21:25:22 +02:00
adrianeboyd
74da669326
Fix problems with lower and whitespace in variants (#5361)
* Initialize lower flag explicitly

* Handle whitespace words from GoldParse correctly when creating raw
text with orth variants

* Return the text with original casing if anything goes wrong
2020-04-29 13:01:25 +02:00
adrianeboyd
3f43c73d37
Normalize TokenC.sent_start values for Matcher (#5346)
Normalize TokenC.sent_start values to booleans for the `Matcher`.
2020-04-29 12:57:30 +02:00
adrianeboyd
bdff76dede
Various updates/additions to CLI scripts (#5362)
* `debug-data`: determine coverage of provided vectors

* `evaluate`: support `blank:lg` model to make it possible to just evaluate
tokenization

* `init-model`: add option to truncate vectors to N most frequent vectors
from word2vec file

* `train`:

  * if training on GPU, only run evaluation/timing on CPU in the first
    iteration

  * if training is aborted, exit with a non-0 exit status
2020-04-29 12:56:46 +02:00
Sofie Van Landeghem
cfdaf99b80
Fix passing of component configuration (#5374)
* add kwargs to to_disk methods in docs - otherwise crashes on 'exclude' argument

* add fix and test for Issue 5137
2020-04-29 12:56:17 +02:00
Ines Montani
efec28ce70
Merge pull request #5367 from adrianeboyd/feature/simplify-warnings-v2 2020-04-29 12:55:37 +02:00
Sofie Van Landeghem
f67343295d
Update NEL examples and documentation (#5370)
* simplify creation of KB by skipping dim reduction

* small fixes to train EL example script

* add KB creation and NEL training example scripts to example section

* update descriptions of example scripts in the documentation

* moving wiki_entity_linking folder from bin to projects

* remove test for wiki NEL functionality that is being moved
2020-04-29 12:53:53 +02:00
adrianeboyd
a6e521cd79
Add is_sent_end token property (#5375)
Reconstruction of the original PR #4697 by @MiniLau.

Removes unused `SENT_END` symbol and `IS_SENT_END` from `Matcher` schema
because the Matcher is only going to be able to support `IS_SENT_START`.
2020-04-29 12:53:16 +02:00
Ines Montani
eac47971f1
Merge pull request #5258 from mirfan899/master 2020-04-29 12:51:55 +02:00
adrianeboyd
d5f18f8307
Add missing import 2020-04-28 14:01:29 +02:00
adrianeboyd
ac40a8f7a5
Add missing import 2020-04-28 14:00:11 +02:00
Adriane Boyd
3a045572ed Add missing import 2020-04-28 13:48:37 +02:00
Adriane Boyd
bc39f97e11 Simplify warnings 2020-04-28 13:37:37 +02:00
adrianeboyd
f8ac5b9f56
bugfix in span similarity (#5155) (#5358)
* bugfix in span similarity

* also rewrite doc.pyx for clarity

* formatting

Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
2020-04-27 16:51:27 +02:00
Sofie Van Landeghem
9203d821ae
Add 2 ini files in tests/lang (#5359) 2020-04-27 13:01:54 +02:00
Punitvara
b2b7e1f37a
This PR adds Gujarati Language class along with (#5355)
* This PR adds Gujarati Language class along with
- stop words

* Add test for gu tokenizer
2020-04-27 11:07:37 +02:00
sabiqueqb
fc91660aa2
Gh 5339 language class for malayalam (#5342)
* Initialize Malayalam Language class

* Add lex_attrs and examples for Malayalam

* Add spaCy Contributor Agreement

* Add test for ml tokenizer
2020-04-27 09:45:08 +02:00
adrianeboyd
84e06f9fb7
Improve GoldParse NER alignment (#5335)
Improve GoldParse NER alignment by including all cases where the start
and end of the NER span can be aligned, regardless of internal
tokenization differences.

To do this, convert BILUO tags to character offsets, check start/end
alignment with `doc.char_span()`, and assign the BILUO tags for the
aligned spans. Alignment for `O/-` tags is handled through the
one-to-one and multi alignments.
2020-04-23 16:58:23 +02:00
adrianeboyd
521f361052
Switch to new gold.align method (#5334)
* Switch from original `_align` to new simpler alignment algorithm from
  #4526

* Remove alignment normalizations beyond whitespace and lowercasing
2020-04-21 19:31:03 +02:00
Matthew Honnibal
b2ef6100af
Only run backprop once when shared tok2vec weights (#5331)
Previously, pipelines with shared tok2vec weights would call the
tok2vec backprop callback multiple times, once for each pipeline
component. This caused errors for PyTorch, and was inefficient.

Instead, accumulate the gradient for all but one component, and just
call the callback once.
2020-04-21 19:30:41 +02:00
adrianeboyd
bf5c13d170
Modify jieba install message (#5328)
Modify jieba install message to instruct the user to use
`ChineseDefaults.use_jieba = False` so that it's possible to load
pkuseg-only models without jieba installed.
2020-04-20 22:06:53 +02:00
Matthew Honnibal
6918d99b6c
Improve GPU usage for train-with-config (#5330)
* Adjust for no ops in Optimizer

* Fix gpu in train-from-config

* Update train-from-config script

* Fix parser

* Fix GPU efficiency of padding backprop
2020-04-20 22:06:28 +02:00
adrianeboyd
f7471abd82
Add pkuseg and serialization support for Chinese (#5308)
* Add pkuseg and serialization support for Chinese

Add support for pkuseg alongside jieba

* Specify model through `Language` meta:

  * split on characters (if no word segmentation packages are installed)

```
Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}})
```

  * jieba (remains the default tokenizer if installed)

```
Chinese()
Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit
```

  * pkuseg

```
Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}})
```

* The new tokenizer setting `require_pkuseg` is used to override
`use_jieba` default, which is intended for models that provide a pkuseg
model:

```
nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}})
nlp = Chinese() # has `use_jieba` as `True` by default
nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer
```

Add support for serialization of tokenizer settings and pkuseg model, if
loaded

* Add sorting for `Language.to_bytes()` serialization of `Language.meta`
so that the (emptied, but still present) tokenizer metadata is in a
consistent position in the serialized data

Extend tests to cover all three tokenizer configurations and
serialization

* Fix from_disk and tests without jieba or pkuseg

* Load cfg first and only show error if `use_pkuseg`
* Fix blank/default initialization in serialization tests

* Explicitly initialize jieba's cache on init

* Add serialization for pkuseg pre/postprocessors

* Reformat pkuseg install message
2020-04-18 17:01:53 +02:00
Jakob Jul Elben
663333c3b2
Fixes #5413 (#5315)
* Fix 5314

* Add contributor

* Resolve requested changes

Co-authored-by: Jakob Jul Elben <jakob@datamaga.com>
2020-04-16 13:29:02 +02:00
Paolo Arduin
1ca32d8f9c
Matcher support for Span as well as Doc (#5113)
* Matcher support for Span, as well as Doc #5056

* Removes an import unused

* Signed contributors agreement

* Code optimization and better test

* Add error message for bad Matcher call argument

* Fix merging
2020-04-15 13:51:33 +02:00
adrianeboyd
98c59027ed
Use max(uint64) for OOV lexeme rank (#5303)
* Use max(uint64) for OOV lexeme rank

* Add test for default OOV rank

* Revert back to thinc==7.4.0

Requiring the updated version of thinc was unnecessary.

* Define OOV_RANK in one place

Define OOV_RANK in one place in `util`.

* Fix formatting [ci skip]

* Switch to external definitions of max(uint64)

Switch to external defintions of max(uint64) and confirm that they are
equal.
2020-04-15 13:49:47 +02:00
adrianeboyd
3d2c308906
Add Doc init from list of words and text (#5251)
* Add Doc init from list of words and text

Add an option to initialize a `Doc` from a text and list of words where
the words may or may not include all whitespace tokens. If the text and
words are mismatched, raise an error.

* Fix error code

* Remove all whitespace before aligning words/text

* Move words/text init to util function

* Update error message

* Rename to get_words_and_spaces

* Fix formatting
2020-04-14 19:15:52 +02:00
Paolo Arduin
8ce408d2e1
Comparison predicate handling for != (#5282)
* Fix #5281

* Optim test
2020-04-14 19:14:15 +02:00
Umar Butler
8952effcc4
Fixed Typo in Warning (#5284)
* Fixed typo in cli warning

Fixed a typo in the warning for the provision of exactly two labels, which have not been designated as binary, to textcat.

* Create and signed contributor form
2020-04-09 15:46:15 +02:00
Sofie Van Landeghem
42364dcd9f
Remove "pala" tokenizer exception for Spanish (#5265) 2020-04-09 10:21:20 +02:00
adrianeboyd
cf579a398d
Add __init__.py to eu and hy tests (#5278) 2020-04-08 20:03:06 +02:00
adrianeboyd
ae4af52ce7
Add ideographic stops to sentencizer (#5263)
Add ideographic half- and fullwidth full stops to default sentencizer
punctuation.
2020-04-08 12:58:39 +02:00
adrianeboyd
fa760010a5
Set rank for new vector in Vocab.set_vector (#5266)
Set `Lexeme.rank` for vectors added with `Vocab.set_vector` so that the
lexeme `ID` accessed by a model points the right row for the new vector.
2020-04-07 12:04:51 +02:00
adrianeboyd
c981aa6684
Use inline flags in token_match patterns (#5257)
* Use inline flags in token_match patterns

Use inline flags in `token_match` patterns so that serializing does not
lose the flag information.

* Modify inline flag

* Modify inline flag
2020-04-06 13:19:04 +02:00
adrianeboyd
e8be15e9b7
Improve tokenization for UD Spanish AnCora (#5253) 2020-04-06 13:18:23 +02:00
adrianeboyd
f4ef64a526
Improve tokenization for UD Dutch corpora (#5259)
* Improve tokenization for UD Dutch corpora

Improve tokenization for UD Dutch Alpino and LassySmall.

* Format Dutch tokenizer exceptions
2020-04-06 13:18:07 +02:00
Muhammad Irfan
406d5748b3 add missing Urdu tags 2020-04-05 20:55:38 +05:00
Sofie Van Landeghem
b2e93be867
Optimizer defaults (#5244)
* set optimizer defaults to mimic thinc 7 + bump to dev6

* larger error range for senter overfitting test
2020-04-03 13:02:46 +02:00
YohannesDatasci
beef184e53
Armenian language support (#5246)
* add Armenian language and test cases

* agreement submission
2020-04-03 13:02:18 +02:00
Michael Leichtfried
2b14997b68
Remove duplicated branch in if/else-if statement (#5234)
* Remove duplicated branch in if-elif-statement

* Add contributor agreement for leicmi
2020-04-02 14:47:42 +02:00
adrianeboyd
b71a11ff6d
Update morphologizer (#5108)
* Add pos and morph scoring to Scorer

Add pos, morph, and morph_per_type to `Scorer`. Report pos and morph
accuracy in `spacy evaluate`.

* Update morphologizer for v3

* switch to tagger-based morphologizer
* use `spacy.HashCharEmbedCNN` for morphologizer defaults
* add `Doc.is_morphed` flag

* Add morphologizer to train CLI

* Add basic morphologizer pipeline tests

* Add simple morphologizer training example

* Remove subword_features from CharEmbed models

Remove `subword_features` argument from `spacy.HashCharEmbedCNN.v1` and
`spacy.HashCharEmbedBiLSTM.v1` since in these cases `subword_features`
is always `False`.

* Rename setting in morphologizer example

Use `with_pos_tags` instead of `without_pos_tags`.

* Fix kwargs for spacy.HashCharEmbedBiLSTM.v1

* Remove defaults for spacy.HashCharEmbedBiLSTM.v1

Remove default `nM/nC` for `spacy.HashCharEmbedBiLSTM.v1`.

* Set random seed for textcat overfitting test
2020-04-02 14:46:32 +02:00
adrianeboyd
d107afcffb
Raise error for inplace resize with new vector dim (#5228)
Raise an error if there is an attempt to resize the vectors in place with
a different vector dimension.
2020-04-02 10:43:13 +02:00
Jacob Lauritzen
0b76212831
Extend and fix Danish examples (#5227)
* Extend and fix Danish examples

This PR fixes two examples, adds additional examples translated from the english version, and adds punctuation.

The two changed examples are:
* "fortov" changed to "fortovet", which is more [used](https://www.google.com/search?client=firefox-b-d&sxsrf=ALeKk0143gEuPe4IbIUpzBBt-oU10OMVqA%3A1585549036477&ei=7I6BXuvJHMGOrwSqi46oCQ&q=l%C3%B8behjul+p%C3%A5+fortov&oq=l%C3%B8behjul+p%C3%A5+fortov&gs_lcp=CgZwc3ktYWIQAzIECAAQRzIECAAQRzIECAAQRzIECAAQRzIECAAQRzIECAAQRzIECAAQRzIECAAQR1DT8xZY0_MWYK_0FmgAcAZ4AIABAIgBAJIBAJgBAKABAaoBB2d3cy13aXo&sclient=psy-ab&ved=0ahUKEwjr7964xsHoAhVBx4sKHaqFA5UQ4dUDCAo&uact=5) and more natural. The Swedish and Norwegian examples also use this version of the word.
* "stor by" changed to "storby". In Danish we have a specific noun to describe a large, metropolitan city which is different from just describing a city as "large". In this sentence it would be much more natural to describe London as a "storby". Google even correct as search for "London stor by" to "London storby".

* Sign contrib agreement
2020-04-02 10:42:35 +02:00
Sofie Van Landeghem
ab59f3124e
fix NEL overfitting test for GPU (#5236) 2020-04-02 10:32:52 +02:00
Sofie Van Landeghem
311133e579
Train textcat with config (#5143)
* bring back default build_text_classifier method

* remove _set_dims_ hack in favor of proper dim inference

* add tok2vec initialize to unit test

* small fixes

* add unit test for various textcat config settings

* logistic output layer does not have nO

* fix window_size setting

* proper fix

* fix W initialization

* Update textcat training example

* Use ml_datasets
* Convert training data to `Example` format
* Use `n_texts` to set proportionate dev size

* fix _init renaming on latest thinc

* avoid setting a non-existing dim

* update to thinc==8.0.0a2

* add BOW and CNN defaults for easy testing

* various experiments with train_textcat script, fix softmax activation in textcat bow

* allow textcat train script to work on other datasets as well

* have dataset as a parameter

* train textcat from config, with example config

* add config for training textcat

* formatting

* fix exclusive_classes

* fixing BOW for GPU

* bump thinc to 8.0.0a3 (not published yet so CI will fail)

* add in link_vectors_to_models which got deleted

Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
2020-03-29 19:40:36 +02:00
adrianeboyd
ce0e538068
Check whether doc is instantiated in Example.get_gold_parses() (#5167)
* Check whether doc is instantiated

When creating docs to pair with gold parses, modify test to check
whether a doc is unset rather than whether it contains tokens.

* Restore test of evaluate on an empty doc

* Set a minimal gold.orig for the scorer

Without a minimal gold.orig the scorer can't evaluate empty docs. This
is the v3 equivalent of #4925.
2020-03-29 13:57:00 +02:00