Commit Graph

13978 Commits

Author SHA1 Message Date
Ines Montani
6c7a930ee8 Fix variable 2020-12-08 20:44:59 +11:00
Ines Montani
94a5a9814f Update argument handling and documentation 2020-12-08 20:41:18 +11:00
Adriane Boyd
6c221d4841 Fix subsequent pipe detection in EntityRuler
Fix subsequent pipe detection to detect the position of the current
object by comparing the component itself rather than from the factory
name.
2020-12-08 10:01:30 +01:00
Ines Montani
b87793a89a
Merge pull request #6523 from adrianeboyd/bugfix/remove-use-chars
Remove non-working --use-chars from train CLI
2020-12-08 09:30:48 +01:00
Adriane Boyd
5ceac425ee Remove non-working --use-chars from train CLI
Remove the non-working `--use-chars` option from the train CLI. The
implementation of the option across component types and the CLI settings
could be fixed, but the `CharacterEmbed` model does not work on GPU in
v2 so it's better to remove it.
2020-12-08 08:30:00 +01:00
Ines Montani
ef59ce783b Adjust install instructions [ci skip] 2020-12-08 18:06:50 +11:00
Ines Montani
d25b1606d6 Allow reading config from sdtin in spacy train 2020-12-08 18:01:40 +11:00
Ines Montani
6cfa66ed1c
Make training.loop return nlp object and path (#6520) 2020-12-08 14:55:55 +08:00
Sofie Van Landeghem
2c27093c5f
require_cpu functionality (#6336)
* add require_cpu from Thinc 8.0.0rc2

* add docs

* fix test if cupy is not installed
2020-12-08 14:42:40 +08:00
Ines Montani
d8e01ca931
Merge pull request #6391 from adrianeboyd/docs/install-guide 2020-12-08 07:42:16 +01:00
Sofie Van Landeghem
f98a04434a
pretrain architectures (#6451)
* define new architectures for the pretraining objective

* add loss function as attr of the omdel

* cleanup

* cleanup

* shorten name

* fix typo

* remove unused error
2020-12-08 14:41:03 +08:00
Adriane Boyd
dcecc75270
Improve blis and numpy build dependencies (#6455)
* Fix blis build dependencies

* Add blis with python_version constraints to pyproject.toml
* Add blis to setup_requires

* Remove --only-binary from CI

* Reduce number of builds to speed up CI

* Add hack to install wheel for python 3.5 in linux

* Remove os spec from CI

* Remove detailed numpy build constraints

* Remove detailed numpy build constraints from `pyproject.toml` because
  it is too difficult to maintain for many architectures
  * These constraints are more a reflection of what is available on
    pypi as binary wheels rather than any real build requirements that
    it is necessary for users to follow when building from source
  * Users building their own binary packages will need to enforce the
    constraints that make sense in their environments, e.g., the `conda`
    compatible numpy pins

* Keep the build constraints in `build-constraints.txt` for use with our
  builds
  * Our builds with wheelwright are built against the earliest
    compatible binary versions of numpy on pypi
  * These constraints are documented within the distribution

* Revert "Remove os spec from CI"

This reverts commit 7489476688.
2020-12-08 14:29:34 +08:00
Adriane Boyd
29b058ebdc
Fix spacy when retokenizing cases with affixes (#6475)
Preserve `token.spacy` corresponding to the span end token in the
original doc rather than adjusting for the current offset.

* If not modifying in place, this checks in the original document
(`doc.c` rather than `tokens`).
* If modifying in place, the document has not been modified past the
current span start position so the value at the current span end
position is valid.
2020-12-08 14:25:56 +08:00
Adriane Boyd
4448680750
Fix alignment for 1-to-1 tokens and lowercasing (#6476)
* When checking for token alignments, check not only that the tokens are
identical but that the character positions are both at the start of a
token.

  It's possible for the tokens to be identical even though the two
tokens aren't aligned one-to-one in a case like `["a'", "''"]` vs.
`["a", "''", "'"]`, where the middle tokens are identical but should not
be aligned on the token level at character position 2 since it's the
start of one token but the middle of another.

* Use the lowercased version of the token texts to create the
character-to-token alignment because lowercasing can change the string
length (e.g., for `İ`, see the not-a-bug bug report:
https://bugs.python.org/issue34723)
2020-12-08 14:25:16 +08:00
Adriane Boyd
e931d3f72b
Move max_length to nlp.make_doc() (#6512)
Move max_length check to `nlp.make_doc()` so that's it's also checked
for `nlp.pipe()`.
2020-12-08 14:24:02 +08:00
Ines Montani
ee2ec52f48
Merge pull request #6409 from svlandeg/feature/trf-docs 2020-12-08 06:32:10 +01:00
Ines Montani
c2b196c2c1
Merge pull request #6419 from svlandeg/feature/rel-docs 2020-12-08 06:30:41 +01:00
Ines Montani
82e88f0e3b
Merge pull request #6379 from svlandeg/fix/labels-constructor 2020-12-08 06:29:56 +01:00
Sofie Van Landeghem
52fa46dd58
tested EL scripts with 2.3.4 (#6517) 2020-12-07 20:46:38 +01:00
Adriane Boyd
d70950605c Warn on empty POS for the rule-based lemmatizer
Add a warning to the rule-based lemmatizer for any tokens without POS
annotation.
2020-12-04 11:46:15 +01:00
Adriane Boyd
78085fab1f
Check for spacy-nightly package in download (#6502)
Also check for spacy-nightly in download so that `--no-deps` isn't set
for normal nightly installs.
2020-12-04 09:40:03 +01:00
Ines Montani
63f83e7034
Merge pull request #6470 from adrianeboyd/feature/license-in-package 2020-12-04 03:55:54 +01:00
Sofie Van Landeghem
d6c616a125
Fixes in test suite (#6457)
* fix slow test for textcat readers

* cleanup test_issue5551

* add explicit score weight

* cleanup
2020-12-02 12:57:08 +01:00
Adriane Boyd
31ec9a906e
Clean up 3rd party license info (#6478)
Move scikit-learn license from `Scorer` to
`licenses/3rd_party_licenses.txt`.
2020-12-02 10:15:23 +01:00
Adriane Boyd
591cd48aa8 Remove config.cfg from MANIFEST 2020-12-01 12:58:02 +01:00
Adriane Boyd
b0dd13e0ba Support LICENSE in spacy package
If present, include the file `input_dir/LICENSE` at the top level of the
packaged model.
2020-11-30 13:43:58 +01:00
Adriane Boyd
1442d2f213
Improve simple training example in v3 migration (#6438)
* Create the examples once
* Use the examples in the initialization
* Provide the batch size
* Fix `begin_training` migration example
2020-11-30 09:39:45 +08:00
Adriane Boyd
53c0fb7431
Only set NORM on Token in retokenizer (#6464)
* Only set NORM on Token in retokenizer

Instead of setting `NORM` on both the token and lexeme, set `NORM` only
on the token.

The retokenizer tries to set all possible attributes with
`Token/Lexeme.set_struct_attr` so that it doesn't have to enumerate
which attributes are available for each. `NORM` is the only attribute
that's stored on both and for most cases it doesn't make sense to set
the global norms based on a individual retokenization. For lexeme-only
attributes like `IS_STOP` there's no way to avoid the global side
effects, but I think that `NORM` would be better only on the token.

* Fix test
2020-11-30 09:35:42 +08:00
Adriane Boyd
03ae77e603
Add SPACY as a Matcher attribute (#6463) 2020-11-30 09:34:50 +08:00
Sofie Van Landeghem
079f6ea474
avoid resolving the full config (#6465) 2020-11-30 09:34:29 +08:00
Ines Montani
9beba7164f Make jinja2 top-level import
No problem anymore since it's now an official dependency
2020-11-27 15:17:14 +08:00
Ines Montani
d21d2c2e59 Don't multiply accuracy by 100 2020-11-27 15:15:51 +08:00
Adriane Boyd
26296ab223
Add error message if DocBin zlib decompress fails (#6394)
Add a better error message if DocBin zlib decompress fails, indicating
that the data is not in `DocBin` format.
2020-11-27 14:39:49 +08:00
Adriane Boyd
3a5cc5f8b4 Set version to v2.3.4 2020-11-26 08:48:52 +01:00
Adriane Boyd
e0f5646a4a
Restore cleanup_beam method (#6446) 2020-11-25 13:21:48 +01:00
Adriane Boyd
40c583a41b Remove --prefer-binary and --only-binary from CI 2020-11-25 12:24:11 +01:00
Adriane Boyd
cf693f0eae Fix token_match in tokenizer 2020-11-25 11:49:34 +01:00
Adriane Boyd
724831b066 Merge remote-tracking branch 'upstream/master' into chore/update-develop-from-master
* Update Macedonian for v3
* Update Turkish for v3
2020-11-25 11:49:34 +01:00
Jacob Bortell
fe9009911a Update rule-based-matching.md (#6421)
* Update rule-based-matching.md

Clarified case-sensititivy of dictionary-referencing attributes (POS/TAG/DEP/etc).

Clarified "Type" column header to "Value Type"

* Update rule-based-matching.md

Improved clarity of wording
2020-11-24 16:20:19 +01:00
Jacob Bortell
992723dfac
Add jabortell to the contributors (#6422)
* Add jabortell to the contributors

* Update jabortell.md

Added tick to applicable statement
2020-11-24 16:15:31 +01:00
Adriane Boyd
6f133877aa Update source install instructions
* Don't recommend an editable install in the default source
instructions.
* Use `pip install --no-build-isolation` for editable installs.
* Remove reference to `virtualenv`.
2020-11-24 14:44:13 +01:00
Adriane Boyd
afd744bc05
Update Travis CI pip install steps (#6440) 2020-11-24 14:10:16 +01:00
Adriane Boyd
573f5c863f
Fix tag map clobbering in spacy train (#6437)
Fix bug from #5768 where the tag map is clobbered if a custom tag map
isn't provided.
2020-11-24 13:13:16 +01:00
Adriane Boyd
ce18fc6588 Set version to v2.3.3 2020-11-24 10:03:45 +01:00
Adriane Boyd
cd61d264ef Set version to v2.3.3.dev0 2020-11-23 13:51:59 +01:00
Sofie Van Landeghem
2af31a8c8d
Bugfix textcat reproducibility on GPU (#6411)
* add seed argument to ParametricAttention layer

* bump thinc to 7.4.3

* set thinc version range

Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
2020-11-23 12:29:35 +01:00
Adriane Boyd
cdca44ac11
Dynamically include numpy headers (#6418)
* Dynamically include numpy headers

* Add `build-constraints.txt` with numpy version pins for building wheels with `pip` and `wheelwright`
* Update `setup.py` to add current numpy include directory
* Assume `cython` and `numpy` are installed for `setup.py`
* Remove included numpy headers

* Fix typo in requirements.txt

* Use script in CI
2020-11-23 11:15:11 +01:00
Adriane Boyd
3f61f5eb54
Use int8_t instead of char in Matcher (#6413)
* Use signed char instead of char in Matcher

Remove unused char* utf8_t typedef

* Use int8_t instead of signed char
2020-11-23 10:26:47 +01:00
Adriane Boyd
4284605683
Remove Beam cleanup (#6414)
Beam cleanup is handled through the Beam finalization method.
2020-11-23 10:01:46 +01:00
Adriane Boyd
a8c2dad466
Add all vectors to vocab before pruning (#6408)
Add all vectors to the vocab before pruning to correct the selection of
vectors to prioritize.
2020-11-23 10:00:59 +01:00