Ines Montani
c41a4332e4
Add test for custom data augmentation
2020-10-02 11:37:56 +02:00
Ines Montani
6d8df081bd
Merge pull request #6180 from adrianeboyd/docs/minor-v3-2 [ci skip]
2020-10-02 11:37:25 +02:00
svlandeg
acc391c2a8
remove redundant str() call
2020-10-02 11:05:59 +02:00
Ines Montani
3856048437
Merge pull request #6178 from explosion/feature/file-readers
...
Integrate file readers via srsly, update orth_variants loading
2020-10-02 10:26:09 +02:00
Adriane Boyd
f83dfe62da
Fix test
2020-10-02 10:17:26 +02:00
Adriane Boyd
351f352cdc
Update Japanese docs and pin for sudachipy
2020-10-02 10:12:44 +02:00
Adriane Boyd
7670df04dd
Update Chinese usage docs
2020-10-02 10:09:03 +02:00
Adriane Boyd
3908fff899
Remove tag map sidebar
2020-10-02 09:07:55 +02:00
Adriane Boyd
fd09e6b140
Update docs for Token.morph / Token.set_morph
2020-10-02 09:05:15 +02:00
Adriane Boyd
65dfaa4f4b
Also accept MorphAnalysis in set_morph
2020-10-02 08:33:43 +02:00
Adriane Boyd
77e08c398f
Switch reset value for set_morph to None
2020-10-02 08:25:15 +02:00
Ines Montani
568768643e
Increment version [ci skip]
2020-10-02 01:50:13 +02:00
Ines Montani
01c1538c72
Integrate file readers
2020-10-02 01:36:06 +02:00
Ines Montani
af282ae732
Fix import
2020-10-02 01:12:34 +02:00
Ines Montani
e59ecb12c0
Auto-format
2020-10-02 01:12:30 +02:00
Ines Montani
6b94cee468
Fix docs [ci skip]
2020-10-02 01:11:19 +02:00
Matthew Honnibal
75a1569908
Merge
2020-10-01 23:07:53 +02:00
Matthew Honnibal
300e5a9928
Avoid relying on NORM in default v3 models ( #6176 )
...
* Allow CharacterEmbed to specify feature
* Default to LOWER in character embed
* Update tok2vec
* Use LOWER, not NORM
2020-10-01 23:05:55 +02:00
Ines Montani
50162b8726
Try to work around Sharp build issue [ci skip]
2020-10-01 22:27:45 +02:00
Ines Montani
5762876dcc
Update default config [ci skip]
2020-10-01 22:27:37 +02:00
Adriane Boyd
86c3ec9c2b
Refactor Token morph setting ( #6175 )
...
* Refactor Token morph setting
* Remove `Token.morph_`
* Add `Token.set_morph()`
* `0` resets `token.c.morph` to unset
* Any other values are passed to `Morphology.add`
* Add token.morph setter to set from MorphAnalysis
2020-10-01 22:21:46 +02:00
Matthew Honnibal
b854bca15c
Default to LOWER in character embed
2020-10-01 22:17:58 +02:00
Matthew Honnibal
684a77870b
Allow CharacterEmbed to specify feature
2020-10-01 22:17:26 +02:00
Ines Montani
da30701cd1
Increment version [ci skip]
2020-10-01 21:58:11 +02:00
Ines Montani
d48ddd6c9a
Remove default initialize lookups
2020-10-01 21:54:33 +02:00
Ines Montani
1700c8541e
Increment version [ci skip]
2020-10-01 17:57:16 +02:00
Ines Montani
b6b73a3ca8
Update docs [ci skip]
2020-10-01 17:45:29 +02:00
Ines Montani
f2627157c8
Update docs [ci skip]
2020-10-01 17:38:17 +02:00
svlandeg
1328c9fd14
consistently use --code instead of --code-path
2020-10-01 16:59:22 +02:00
Ines Montani
7f68f4bd92
Hide jsonl_loc on init vectors and tidy up [ci skip]
2020-10-01 16:44:17 +02:00
Adriane Boyd
27cbffff1b
Minor edit to CoNLL-U converter ( #6172 )
...
This doesn't make a difference given how the `merged_morph` values
override the `morph` values for all the final docs, but could have led
to unexpected bugs in the future if the converter is modified.
2020-10-01 16:23:42 +02:00
Sofie Van Landeghem
a22215f427
Add FeatureExtractor from Thinc ( #6170 )
...
* move featureextractor from Thinc
* Update website/docs/api/architectures.md
Co-authored-by: Ines Montani <ines@ines.io>
* Update website/docs/api/architectures.md
Co-authored-by: Ines Montani <ines@ines.io>
Co-authored-by: Ines Montani <ines@ines.io>
2020-10-01 16:22:48 +02:00
Adriane Boyd
73538782a0
Switch Doc.__init__(ents=) to IOB tags ( #6173 )
...
* Switch Doc.__init__(ents=) to IOB tags
* Fix check for "-"
* Allow "" or None as missing IOB tag
2020-10-01 16:22:18 +02:00
Adriane Boyd
df98d3ef9f
Update import from collections.abc ( #6174 )
2020-10-01 16:21:49 +02:00
Yohei Tamura
3243ddac8f
Fix/span.sent ( #6083 )
...
* add fail test
* fix test
* fix span.sent
* Remove incorrect implicit check
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
2020-10-01 14:01:52 +02:00
Ines Montani
0a8a124a6e
Update docs [ci skip]
2020-10-01 12:15:53 +02:00
Ines Montani
44160cd52f
Tidy up [ci skip]
2020-10-01 10:41:19 +02:00
Ines Montani
381258b75b
Merge pull request #6165 from explosion/feature/update-tokenizers-initialize
2020-10-01 09:49:47 +02:00
svlandeg
6787e56315
print debugging warning before raising error if model not properly initialized
2020-10-01 09:21:00 +02:00
svlandeg
5121972930
add types of Tok2Vec embedding layers
2020-10-01 09:20:09 +02:00
Ines Montani
4b6afd3611
Remove English [initialize] default block for now to get tests to pass
2020-09-30 23:49:29 +02:00
Ines Montani
6f29f68f69
Update errors and make Tokenizer.initialize args less strict
2020-09-30 23:48:47 +02:00
Ines Montani
a103ab5f1a
Update augmenter lookups and docs
2020-09-30 23:03:47 +02:00
Matthew Honnibal
5128298964
Add missing augmenter
2020-09-30 20:18:45 +02:00
Matthew Honnibal
59294e91aa
Restore the 'jsonl' arg for init vectors
...
The lexemes.jsonl file is still used in our English vectors, and it may
be required by users as well. I think it's worth supporting the option.
2020-09-30 19:06:50 +02:00
Matthew Honnibal
c379a4274a
Merge branch 'develop' of https://github.com/explosion/spaCy into develop
2020-09-30 16:52:42 +02:00
Matthew Honnibal
e58dca3028
Add read_labels
2020-09-30 16:52:27 +02:00
Ines Montani
115481aca7
Update docs [ci skip]
2020-09-30 15:16:00 +02:00
Ines Montani
23c63eefaf
Tidy up env vars [ci skip]
2020-09-30 15:15:11 +02:00
Elijah Rippeth
4cbb954281
reorder so tagmap is replaced only if a custom file is provided. ( #6164 )
...
* reorder so tagmap is replaced only if a custom file is provided.
* Remove unneeded variable initialization
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
2020-09-30 13:26:06 +02:00
